diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
--- "a/last-checkpoint/trainer_state.json"
+++ "b/last-checkpoint/trainer_state.json"
@@ -1,339207 +1,3 @@
-{
-  "best_metric": null,
-  "best_model_checkpoint": null,
-  "epoch": 3.7043408452319513,
-  "eval_steps": 500,
-  "global_step": 48450,
-  "is_hyper_param_search": false,
-  "is_local_process_zero": true,
-  "is_world_process_zero": true,
-  "log_history": [
-    {
-      "epoch": 7.645698338972036e-05,
-      "grad_norm": 1.792826697055716e-05,
-      "learning_rate": 2e-05,
-      "loss": 46.0,
-      "step": 1
-    },
-    {
-      "epoch": 0.00015291396677944073,
-      "grad_norm": 7.698028639424592e-06,
-      "learning_rate": 4e-05,
-      "loss": 46.0,
-      "step": 2
-    },
-    {
-      "epoch": 0.00022937095016916108,
-      "grad_norm": 1.7271497199544683e-05,
-      "learning_rate": 6e-05,
-      "loss": 46.0,
-      "step": 3
-    },
-    {
-      "epoch": 0.00030582793355888146,
-      "grad_norm": 7.746815754217096e-06,
-      "learning_rate": 8e-05,
-      "loss": 46.0,
-      "step": 4
-    },
-    {
-      "epoch": 0.0003822849169486018,
-      "grad_norm": 3.2548108720220625e-05,
-      "learning_rate": 0.0001,
-      "loss": 46.0,
-      "step": 5
-    },
-    {
-      "epoch": 0.00045874190033832216,
-      "grad_norm": 2.1352847397793084e-05,
-      "learning_rate": 0.00012,
-      "loss": 46.0,
-      "step": 6
-    },
-    {
-      "epoch": 0.0005351988837280425,
-      "grad_norm": 1.3152564861229621e-05,
-      "learning_rate": 0.00014,
-      "loss": 46.0,
-      "step": 7
-    },
-    {
-      "epoch": 0.0006116558671177629,
-      "grad_norm": 1.4066746189200785e-05,
-      "learning_rate": 0.00016,
-      "loss": 46.0,
-      "step": 8
-    },
-    {
-      "epoch": 0.0006881128505074832,
-      "grad_norm": 2.2254904251894914e-05,
-      "learning_rate": 0.00018,
-      "loss": 46.0,
-      "step": 9
-    },
-    {
-      "epoch": 0.0007645698338972036,
-      "grad_norm": 4.437996176420711e-05,
-      "learning_rate": 0.0002,
-      "loss": 46.0,
-      "step": 10
-    },
-    {
-      "epoch": 0.0008410268172869239,
-      "grad_norm": 2.1450669009936973e-05,
-      "learning_rate": 0.00019999999999999714,
-      "loss": 46.0,
-      "step": 11
-    },
-    {
-      "epoch": 0.0009174838006766443,
-      "grad_norm": 3.5947781725553796e-05,
-      "learning_rate": 0.00019999999999998846,
-      "loss": 46.0,
-      "step": 12
-    },
-    {
-      "epoch": 0.0009939407840663646,
-      "grad_norm": 1.1451102182036266e-05,
-      "learning_rate": 0.00019999999999997404,
-      "loss": 46.0,
-      "step": 13
-    },
-    {
-      "epoch": 0.001070397767456085,
-      "grad_norm": 1.5752761100884527e-05,
-      "learning_rate": 0.00019999999999995388,
-      "loss": 46.0,
-      "step": 14
-    },
-    {
-      "epoch": 0.0011468547508458053,
-      "grad_norm": 1.1428691323089879e-05,
-      "learning_rate": 0.00019999999999992788,
-      "loss": 46.0,
-      "step": 15
-    },
-    {
-      "epoch": 0.0012233117342355258,
-      "grad_norm": 4.6794833906460553e-05,
-      "learning_rate": 0.00019999999999989617,
-      "loss": 46.0,
-      "step": 16
-    },
-    {
-      "epoch": 0.001299768717625246,
-      "grad_norm": 2.1383812054409645e-05,
-      "learning_rate": 0.00019999999999985866,
-      "loss": 46.0,
-      "step": 17
-    },
-    {
-      "epoch": 0.0013762257010149664,
-      "grad_norm": 2.292082535859663e-05,
-      "learning_rate": 0.0001999999999998154,
-      "loss": 46.0,
-      "step": 18
-    },
-    {
-      "epoch": 0.0014526826844046868,
-      "grad_norm": 1.3927367945143487e-05,
-      "learning_rate": 0.00019999999999976636,
-      "loss": 46.0,
-      "step": 19
-    },
-    {
-      "epoch": 0.0015291396677944071,
-      "grad_norm": 7.859431934775785e-05,
-      "learning_rate": 0.00019999999999971153,
-      "loss": 46.0,
-      "step": 20
-    },
-    {
-      "epoch": 0.0016055966511841276,
-      "grad_norm": 2.4063776436378248e-05,
-      "learning_rate": 0.00019999999999965095,
-      "loss": 46.0,
-      "step": 21
-    },
-    {
-      "epoch": 0.0016820536345738479,
-      "grad_norm": 2.7152964321430773e-05,
-      "learning_rate": 0.0001999999999995846,
-      "loss": 46.0,
-      "step": 22
-    },
-    {
-      "epoch": 0.0017585106179635681,
-      "grad_norm": 1.9019813407794572e-05,
-      "learning_rate": 0.0001999999999995125,
-      "loss": 46.0,
-      "step": 23
-    },
-    {
-      "epoch": 0.0018349676013532886,
-      "grad_norm": 1.2315812455199193e-05,
-      "learning_rate": 0.0001999999999994346,
-      "loss": 46.0,
-      "step": 24
-    },
-    {
-      "epoch": 0.001911424584743009,
-      "grad_norm": 2.5389064830960706e-05,
-      "learning_rate": 0.00019999999999935095,
-      "loss": 46.0,
-      "step": 25
-    },
-    {
-      "epoch": 0.001987881568132729,
-      "grad_norm": 1.290810450882418e-05,
-      "learning_rate": 0.00019999999999926153,
-      "loss": 46.0,
-      "step": 26
-    },
-    {
-      "epoch": 0.00206433855152245,
-      "grad_norm": 9.934929948940407e-06,
-      "learning_rate": 0.0001999999999991663,
-      "loss": 46.0,
-      "step": 27
-    },
-    {
-      "epoch": 0.00214079553491217,
-      "grad_norm": 1.172078827949008e-05,
-      "learning_rate": 0.00019999999999906535,
-      "loss": 46.0,
-      "step": 28
-    },
-    {
-      "epoch": 0.0022172525183018904,
-      "grad_norm": 3.653451130958274e-05,
-      "learning_rate": 0.00019999999999895863,
-      "loss": 46.0,
-      "step": 29
-    },
-    {
-      "epoch": 0.0022937095016916107,
-      "grad_norm": 2.3206204787129536e-05,
-      "learning_rate": 0.00019999999999884612,
-      "loss": 46.0,
-      "step": 30
-    },
-    {
-      "epoch": 0.002370166485081331,
-      "grad_norm": 1.5248630916175898e-05,
-      "learning_rate": 0.00019999999999872786,
-      "loss": 46.0,
-      "step": 31
-    },
-    {
-      "epoch": 0.0024466234684710516,
-      "grad_norm": 2.0085321011720225e-05,
-      "learning_rate": 0.0001999999999986038,
-      "loss": 46.0,
-      "step": 32
-    },
-    {
-      "epoch": 0.002523080451860772,
-      "grad_norm": 8.577127118769567e-06,
-      "learning_rate": 0.000199999999998474,
-      "loss": 46.0,
-      "step": 33
-    },
-    {
-      "epoch": 0.002599537435250492,
-      "grad_norm": 2.6271172828273848e-05,
-      "learning_rate": 0.00019999999999833842,
-      "loss": 46.0,
-      "step": 34
-    },
-    {
-      "epoch": 0.0026759944186402125,
-      "grad_norm": 1.3005247637920547e-05,
-      "learning_rate": 0.00019999999999819706,
-      "loss": 46.0,
-      "step": 35
-    },
-    {
-      "epoch": 0.0027524514020299327,
-      "grad_norm": 1.2495203009166289e-05,
-      "learning_rate": 0.00019999999999804994,
-      "loss": 46.0,
-      "step": 36
-    },
-    {
-      "epoch": 0.0028289083854196534,
-      "grad_norm": 2.2880592950969003e-05,
-      "learning_rate": 0.00019999999999789704,
-      "loss": 46.0,
-      "step": 37
-    },
-    {
-      "epoch": 0.0029053653688093737,
-      "grad_norm": 5.68693176319357e-05,
-      "learning_rate": 0.0001999999999977384,
-      "loss": 46.0,
-      "step": 38
-    },
-    {
-      "epoch": 0.002981822352199094,
-      "grad_norm": 2.655573553056456e-05,
-      "learning_rate": 0.00019999999999757394,
-      "loss": 46.0,
-      "step": 39
-    },
-    {
-      "epoch": 0.0030582793355888142,
-      "grad_norm": 2.349610258534085e-05,
-      "learning_rate": 0.00019999999999740375,
-      "loss": 46.0,
-      "step": 40
-    },
-    {
-      "epoch": 0.0031347363189785345,
-      "grad_norm": 2.5527391699142754e-05,
-      "learning_rate": 0.00019999999999722779,
-      "loss": 46.0,
-      "step": 41
-    },
-    {
-      "epoch": 0.003211193302368255,
-      "grad_norm": 1.3046727872279007e-05,
-      "learning_rate": 0.00019999999999704607,
-      "loss": 46.0,
-      "step": 42
-    },
-    {
-      "epoch": 0.0032876502857579755,
-      "grad_norm": 1.6746318578952923e-05,
-      "learning_rate": 0.00019999999999685853,
-      "loss": 46.0,
-      "step": 43
-    },
-    {
-      "epoch": 0.0033641072691476957,
-      "grad_norm": 2.166387639590539e-05,
-      "learning_rate": 0.00019999999999666525,
-      "loss": 46.0,
-      "step": 44
-    },
-    {
-      "epoch": 0.003440564252537416,
-      "grad_norm": 1.7299164028372616e-05,
-      "learning_rate": 0.00019999999999646622,
-      "loss": 46.0,
-      "step": 45
-    },
-    {
-      "epoch": 0.0035170212359271363,
-      "grad_norm": 3.4763092116918415e-05,
-      "learning_rate": 0.0001999999999962614,
-      "loss": 46.0,
-      "step": 46
-    },
-    {
-      "epoch": 0.003593478219316857,
-      "grad_norm": 2.022327134909574e-05,
-      "learning_rate": 0.0001999999999960508,
-      "loss": 46.0,
-      "step": 47
-    },
-    {
-      "epoch": 0.0036699352027065773,
-      "grad_norm": 1.1685785466397647e-05,
-      "learning_rate": 0.00019999999999583445,
-      "loss": 46.0,
-      "step": 48
-    },
-    {
-      "epoch": 0.0037463921860962975,
-      "grad_norm": 2.883524211938493e-05,
-      "learning_rate": 0.00019999999999561235,
-      "loss": 46.0,
-      "step": 49
-    },
-    {
-      "epoch": 0.003822849169486018,
-      "grad_norm": 1.3219233551353682e-05,
-      "learning_rate": 0.00019999999999538445,
-      "loss": 46.0,
-      "step": 50
-    },
-    {
-      "epoch": 0.0038993061528757385,
-      "grad_norm": 1.3463150935422163e-05,
-      "learning_rate": 0.00019999999999515078,
-      "loss": 46.0,
-      "step": 51
-    },
-    {
-      "epoch": 0.003975763136265458,
-      "grad_norm": 1.7962094716494903e-05,
-      "learning_rate": 0.00019999999999491133,
-      "loss": 46.0,
-      "step": 52
-    },
-    {
-      "epoch": 0.004052220119655179,
-      "grad_norm": 1.865141348389443e-05,
-      "learning_rate": 0.00019999999999466614,
-      "loss": 46.0,
-      "step": 53
-    },
-    {
-      "epoch": 0.0041286771030449,
-      "grad_norm": 3.395900421310216e-05,
-      "learning_rate": 0.00019999999999441518,
-      "loss": 46.0,
-      "step": 54
-    },
-    {
-      "epoch": 0.00420513408643462,
-      "grad_norm": 5.2411214710446075e-06,
-      "learning_rate": 0.00019999999999415844,
-      "loss": 46.0,
-      "step": 55
-    },
-    {
-      "epoch": 0.00428159106982434,
-      "grad_norm": 2.3695289200986736e-05,
-      "learning_rate": 0.00019999999999389592,
-      "loss": 46.0,
-      "step": 56
-    },
-    {
-      "epoch": 0.0043580480532140605,
-      "grad_norm": 4.788398291566409e-05,
-      "learning_rate": 0.00019999999999362764,
-      "loss": 46.0,
-      "step": 57
-    },
-    {
-      "epoch": 0.004434505036603781,
-      "grad_norm": 1.4658174222859088e-05,
-      "learning_rate": 0.0001999999999933536,
-      "loss": 46.0,
-      "step": 58
-    },
-    {
-      "epoch": 0.004510962019993501,
-      "grad_norm": 3.2516836654394865e-05,
-      "learning_rate": 0.00019999999999307377,
-      "loss": 46.0,
-      "step": 59
-    },
-    {
-      "epoch": 0.004587419003383221,
-      "grad_norm": 1.7249649317818694e-05,
-      "learning_rate": 0.00019999999999278817,
-      "loss": 46.0,
-      "step": 60
-    },
-    {
-      "epoch": 0.004663875986772942,
-      "grad_norm": 3.424560418352485e-05,
-      "learning_rate": 0.00019999999999249684,
-      "loss": 46.0,
-      "step": 61
-    },
-    {
-      "epoch": 0.004740332970162662,
-      "grad_norm": 1.6180661987164058e-05,
-      "learning_rate": 0.00019999999999219972,
-      "loss": 46.0,
-      "step": 62
-    },
-    {
-      "epoch": 0.004816789953552382,
-      "grad_norm": 2.7245410819887184e-05,
-      "learning_rate": 0.0001999999999918968,
-      "loss": 46.0,
-      "step": 63
-    },
-    {
-      "epoch": 0.004893246936942103,
-      "grad_norm": 1.8434240701026283e-05,
-      "learning_rate": 0.00019999999999158817,
-      "loss": 46.0,
-      "step": 64
-    },
-    {
-      "epoch": 0.0049697039203318236,
-      "grad_norm": 1.1324344995955471e-05,
-      "learning_rate": 0.0001999999999912737,
-      "loss": 46.0,
-      "step": 65
-    },
-    {
-      "epoch": 0.005046160903721544,
-      "grad_norm": 1.2631042409338988e-05,
-      "learning_rate": 0.0001999999999909535,
-      "loss": 46.0,
-      "step": 66
-    },
-    {
-      "epoch": 0.005122617887111264,
-      "grad_norm": 1.1797097613452934e-05,
-      "learning_rate": 0.00019999999999062752,
-      "loss": 46.0,
-      "step": 67
-    },
-    {
-      "epoch": 0.005199074870500984,
-      "grad_norm": 0.0001484593958593905,
-      "learning_rate": 0.00019999999999029578,
-      "loss": 46.0,
-      "step": 68
-    },
-    {
-      "epoch": 0.005275531853890705,
-      "grad_norm": 7.72525709180627e-06,
-      "learning_rate": 0.00019999999998995826,
-      "loss": 46.0,
-      "step": 69
-    },
-    {
-      "epoch": 0.005351988837280425,
-      "grad_norm": 2.2868665837449953e-05,
-      "learning_rate": 0.00019999999998961498,
-      "loss": 46.0,
-      "step": 70
-    },
-    {
-      "epoch": 0.005428445820670145,
-      "grad_norm": 2.722097997320816e-05,
-      "learning_rate": 0.00019999999998926592,
-      "loss": 46.0,
-      "step": 71
-    },
-    {
-      "epoch": 0.0055049028040598654,
-      "grad_norm": 2.157407652703114e-05,
-      "learning_rate": 0.00019999999998891112,
-      "loss": 46.0,
-      "step": 72
-    },
-    {
-      "epoch": 0.005581359787449587,
-      "grad_norm": 5.3770130762131885e-05,
-      "learning_rate": 0.00019999999998855054,
-      "loss": 46.0,
-      "step": 73
-    },
-    {
-      "epoch": 0.005657816770839307,
-      "grad_norm": 3.084981653955765e-05,
-      "learning_rate": 0.00019999999998818418,
-      "loss": 46.0,
-      "step": 74
-    },
-    {
-      "epoch": 0.005734273754229027,
-      "grad_norm": 2.2627700673183426e-05,
-      "learning_rate": 0.00019999999998781203,
-      "loss": 46.0,
-      "step": 75
-    },
-    {
-      "epoch": 0.005810730737618747,
-      "grad_norm": 4.124066617805511e-05,
-      "learning_rate": 0.00019999999998743416,
-      "loss": 46.0,
-      "step": 76
-    },
-    {
-      "epoch": 0.005887187721008468,
-      "grad_norm": 2.420853343210183e-05,
-      "learning_rate": 0.00019999999998705046,
-      "loss": 46.0,
-      "step": 77
-    },
-    {
-      "epoch": 0.005963644704398188,
-      "grad_norm": 4.235423693899065e-05,
-      "learning_rate": 0.00019999999998666104,
-      "loss": 46.0,
-      "step": 78
-    },
-    {
-      "epoch": 0.006040101687787908,
-      "grad_norm": 5.554194649448618e-05,
-      "learning_rate": 0.00019999999998626585,
-      "loss": 46.0,
-      "step": 79
-    },
-    {
-      "epoch": 0.0061165586711776285,
-      "grad_norm": 1.315249301114818e-05,
-      "learning_rate": 0.00019999999998586486,
-      "loss": 46.0,
-      "step": 80
-    },
-    {
-      "epoch": 0.006193015654567349,
-      "grad_norm": 1.062934552464867e-05,
-      "learning_rate": 0.0001999999999854581,
-      "loss": 46.0,
-      "step": 81
-    },
-    {
-      "epoch": 0.006269472637957069,
-      "grad_norm": 2.8291991839068942e-05,
-      "learning_rate": 0.00019999999998504558,
-      "loss": 46.0,
-      "step": 82
-    },
-    {
-      "epoch": 0.00634592962134679,
-      "grad_norm": 9.785777365323156e-06,
-      "learning_rate": 0.0001999999999846273,
-      "loss": 46.0,
-      "step": 83
-    },
-    {
-      "epoch": 0.00642238660473651,
-      "grad_norm": 3.804332664003596e-05,
-      "learning_rate": 0.00019999999998420324,
-      "loss": 46.0,
-      "step": 84
-    },
-    {
-      "epoch": 0.006498843588126231,
-      "grad_norm": 3.185340028721839e-05,
-      "learning_rate": 0.00019999999998377343,
-      "loss": 46.0,
-      "step": 85
-    },
-    {
-      "epoch": 0.006575300571515951,
-      "grad_norm": 2.0756384401465766e-05,
-      "learning_rate": 0.00019999999998333783,
-      "loss": 46.0,
-      "step": 86
-    },
-    {
-      "epoch": 0.006651757554905671,
-      "grad_norm": 1.887604594230652e-05,
-      "learning_rate": 0.00019999999998289645,
-      "loss": 46.0,
-      "step": 87
-    },
-    {
-      "epoch": 0.0067282145382953915,
-      "grad_norm": 1.6534835594939068e-05,
-      "learning_rate": 0.00019999999998244932,
-      "loss": 46.0,
-      "step": 88
-    },
-    {
-      "epoch": 0.006804671521685112,
-      "grad_norm": 1.6333686289726757e-05,
-      "learning_rate": 0.00019999999998199643,
-      "loss": 46.0,
-      "step": 89
-    },
-    {
-      "epoch": 0.006881128505074832,
-      "grad_norm": 3.246934647904709e-05,
-      "learning_rate": 0.00019999999998153775,
-      "loss": 46.0,
-      "step": 90
-    },
-    {
-      "epoch": 0.006957585488464552,
-      "grad_norm": 1.181766674562823e-05,
-      "learning_rate": 0.00019999999998107334,
-      "loss": 46.0,
-      "step": 91
-    },
-    {
-      "epoch": 0.007034042471854273,
-      "grad_norm": 1.2396643796819262e-05,
-      "learning_rate": 0.0001999999999806031,
-      "loss": 46.0,
-      "step": 92
-    },
-    {
-      "epoch": 0.007110499455243994,
-      "grad_norm": 4.238028486724943e-05,
-      "learning_rate": 0.00019999999998012713,
-      "loss": 46.0,
-      "step": 93
-    },
-    {
-      "epoch": 0.007186956438633714,
-      "grad_norm": 2.364716783631593e-05,
-      "learning_rate": 0.0001999999999796454,
-      "loss": 46.0,
-      "step": 94
-    },
-    {
-      "epoch": 0.007263413422023434,
-      "grad_norm": 4.005626396974549e-05,
-      "learning_rate": 0.00019999999997915785,
-      "loss": 46.0,
-      "step": 95
-    },
-    {
-      "epoch": 0.0073398704054131545,
-      "grad_norm": 2.0414901882759295e-05,
-      "learning_rate": 0.00019999999997866456,
-      "loss": 46.0,
-      "step": 96
-    },
-    {
-      "epoch": 0.007416327388802875,
-      "grad_norm": 2.954376395791769e-05,
-      "learning_rate": 0.00019999999997816553,
-      "loss": 46.0,
-      "step": 97
-    },
-    {
-      "epoch": 0.007492784372192595,
-      "grad_norm": 4.945532054989599e-05,
-      "learning_rate": 0.00019999999997766067,
-      "loss": 46.0,
-      "step": 98
-    },
-    {
-      "epoch": 0.007569241355582315,
-      "grad_norm": 1.6079820852610283e-05,
-      "learning_rate": 0.0001999999999771501,
-      "loss": 46.0,
-      "step": 99
-    },
-    {
-      "epoch": 0.007645698338972036,
-      "grad_norm": 1.7326456145383418e-05,
-      "learning_rate": 0.00019999999997663372,
-      "loss": 46.0,
-      "step": 100
-    },
-    {
-      "epoch": 0.007722155322361756,
-      "grad_norm": 2.4560122255934402e-05,
-      "learning_rate": 0.0001999999999761116,
-      "loss": 46.0,
-      "step": 101
-    },
-    {
-      "epoch": 0.007798612305751477,
-      "grad_norm": 2.7380283427191898e-05,
-      "learning_rate": 0.00019999999997558367,
-      "loss": 46.0,
-      "step": 102
-    },
-    {
-      "epoch": 0.007875069289141197,
-      "grad_norm": 4.690876085078344e-05,
-      "learning_rate": 0.00019999999997505002,
-      "loss": 46.0,
-      "step": 103
-    },
-    {
-      "epoch": 0.007951526272530917,
-      "grad_norm": 4.03676567657385e-05,
-      "learning_rate": 0.00019999999997451058,
-      "loss": 46.0,
-      "step": 104
-    },
-    {
-      "epoch": 0.008027983255920638,
-      "grad_norm": 3.087197183049284e-05,
-      "learning_rate": 0.00019999999997396533,
-      "loss": 46.0,
-      "step": 105
-    },
-    {
-      "epoch": 0.008104440239310357,
-      "grad_norm": 3.537006341503002e-05,
-      "learning_rate": 0.00019999999997341437,
-      "loss": 46.0,
-      "step": 106
-    },
-    {
-      "epoch": 0.008180897222700078,
-      "grad_norm": 2.691395093279425e-05,
-      "learning_rate": 0.0001999999999728576,
-      "loss": 46.0,
-      "step": 107
-    },
-    {
-      "epoch": 0.0082573542060898,
-      "grad_norm": 1.6602109099039808e-05,
-      "learning_rate": 0.0001999999999722951,
-      "loss": 46.0,
-      "step": 108
-    },
-    {
-      "epoch": 0.008333811189479519,
-      "grad_norm": 4.141966564930044e-05,
-      "learning_rate": 0.0001999999999717268,
-      "loss": 46.0,
-      "step": 109
-    },
-    {
-      "epoch": 0.00841026817286924,
-      "grad_norm": 5.193466495256871e-05,
-      "learning_rate": 0.00019999999997115272,
-      "loss": 46.0,
-      "step": 110
-    },
-    {
-      "epoch": 0.00848672515625896,
-      "grad_norm": 2.273494283144828e-05,
-      "learning_rate": 0.0001999999999705729,
-      "loss": 46.0,
-      "step": 111
-    },
-    {
-      "epoch": 0.00856318213964868,
-      "grad_norm": 1.8635138985700905e-05,
-      "learning_rate": 0.00019999999996998731,
-      "loss": 46.0,
-      "step": 112
-    },
-    {
-      "epoch": 0.0086396391230384,
-      "grad_norm": 5.76334678044077e-05,
-      "learning_rate": 0.00019999999996939594,
-      "loss": 46.0,
-      "step": 113
-    },
-    {
-      "epoch": 0.008716096106428121,
-      "grad_norm": 1.4108022696746048e-05,
-      "learning_rate": 0.00019999999996879879,
-      "loss": 46.0,
-      "step": 114
-    },
-    {
-      "epoch": 0.00879255308981784,
-      "grad_norm": 5.847453940077685e-05,
-      "learning_rate": 0.0001999999999681959,
-      "loss": 46.0,
-      "step": 115
-    },
-    {
-      "epoch": 0.008869010073207562,
-      "grad_norm": 2.4478789782733656e-05,
-      "learning_rate": 0.00019999999996758722,
-      "loss": 46.0,
-      "step": 116
-    },
-    {
-      "epoch": 0.008945467056597283,
-      "grad_norm": 1.6160238374141045e-05,
-      "learning_rate": 0.00019999999996697277,
-      "loss": 46.0,
-      "step": 117
-    },
-    {
-      "epoch": 0.009021924039987002,
-      "grad_norm": 1.1715706023096573e-05,
-      "learning_rate": 0.00019999999996635258,
-      "loss": 46.0,
-      "step": 118
-    },
-    {
-      "epoch": 0.009098381023376723,
-      "grad_norm": 2.351373404962942e-05,
-      "learning_rate": 0.0001999999999657266,
-      "loss": 46.0,
-      "step": 119
-    },
-    {
-      "epoch": 0.009174838006766443,
-      "grad_norm": 1.4249350897443946e-05,
-      "learning_rate": 0.00019999999996509483,
-      "loss": 46.0,
-      "step": 120
-    },
-    {
-      "epoch": 0.009251294990156164,
-      "grad_norm": 1.539502409286797e-05,
-      "learning_rate": 0.0001999999999644573,
-      "loss": 46.0,
-      "step": 121
-    },
-    {
-      "epoch": 0.009327751973545883,
-      "grad_norm": 1.3343144928512629e-05,
-      "learning_rate": 0.00019999999996381398,
-      "loss": 46.0,
-      "step": 122
-    },
-    {
-      "epoch": 0.009404208956935604,
-      "grad_norm": 2.421860881440807e-05,
-      "learning_rate": 0.00019999999996316495,
-      "loss": 46.0,
-      "step": 123
-    },
-    {
-      "epoch": 0.009480665940325324,
-      "grad_norm": 1.230876750923926e-05,
-      "learning_rate": 0.0001999999999625101,
-      "loss": 46.0,
-      "step": 124
-    },
-    {
-      "epoch": 0.009557122923715045,
-      "grad_norm": 1.5687432096456178e-05,
-      "learning_rate": 0.0001999999999618495,
-      "loss": 46.0,
-      "step": 125
-    },
-    {
-      "epoch": 0.009633579907104764,
-      "grad_norm": 2.405607665423304e-05,
-      "learning_rate": 0.00019999999996118313,
-      "loss": 46.0,
-      "step": 126
-    },
-    {
-      "epoch": 0.009710036890494485,
-      "grad_norm": 1.0055097845906857e-05,
-      "learning_rate": 0.00019999999996051098,
-      "loss": 46.0,
-      "step": 127
-    },
-    {
-      "epoch": 0.009786493873884207,
-      "grad_norm": 1.994884769374039e-05,
-      "learning_rate": 0.00019999999995983308,
-      "loss": 46.0,
-      "step": 128
-    },
-    {
-      "epoch": 0.009862950857273926,
-      "grad_norm": 2.4485007088514976e-05,
-      "learning_rate": 0.00019999999995914939,
-      "loss": 46.0,
-      "step": 129
-    },
-    {
-      "epoch": 0.009939407840663647,
-      "grad_norm": 3.35947479470633e-05,
-      "learning_rate": 0.00019999999995845994,
-      "loss": 46.0,
-      "step": 130
-    },
-    {
-      "epoch": 0.010015864824053367,
-      "grad_norm": 1.7626496628508903e-05,
-      "learning_rate": 0.00019999999995776472,
-      "loss": 46.0,
-      "step": 131
-    },
-    {
-      "epoch": 0.010092321807443088,
-      "grad_norm": 1.8997772713191807e-05,
-      "learning_rate": 0.00019999999995706373,
-      "loss": 46.0,
-      "step": 132
-    },
-    {
-      "epoch": 0.010168778790832807,
-      "grad_norm": 1.9099250494036824e-05,
-      "learning_rate": 0.00019999999995635697,
-      "loss": 46.0,
-      "step": 133
-    },
-    {
-      "epoch": 0.010245235774222528,
-      "grad_norm": 2.0239402147126384e-05,
-      "learning_rate": 0.00019999999995564446,
-      "loss": 46.0,
-      "step": 134
-    },
-    {
-      "epoch": 0.010321692757612248,
-      "grad_norm": 1.774134398146998e-05,
-      "learning_rate": 0.00019999999995492615,
-      "loss": 46.0,
-      "step": 135
-    },
-    {
-      "epoch": 0.010398149741001969,
-      "grad_norm": 1.6570445950492285e-05,
-      "learning_rate": 0.00019999999995420206,
-      "loss": 46.0,
-      "step": 136
-    },
-    {
-      "epoch": 0.01047460672439169,
-      "grad_norm": 1.1938095667574089e-05,
-      "learning_rate": 0.00019999999995347226,
-      "loss": 46.0,
-      "step": 137
-    },
-    {
-      "epoch": 0.01055106370778141,
-      "grad_norm": 1.392544527334394e-05,
-      "learning_rate": 0.00019999999995273663,
-      "loss": 46.0,
-      "step": 138
-    },
-    {
-      "epoch": 0.01062752069117113,
-      "grad_norm": 7.934186214697547e-06,
-      "learning_rate": 0.00019999999995199525,
-      "loss": 46.0,
-      "step": 139
-    },
-    {
-      "epoch": 0.01070397767456085,
-      "grad_norm": 3.369540718267672e-05,
-      "learning_rate": 0.00019999999995124813,
-      "loss": 46.0,
-      "step": 140
-    },
-    {
-      "epoch": 0.010780434657950571,
-      "grad_norm": 1.83624815690564e-05,
-      "learning_rate": 0.0001999999999504952,
-      "loss": 46.0,
-      "step": 141
-    },
-    {
-      "epoch": 0.01085689164134029,
-      "grad_norm": 4.4597763917408884e-05,
-      "learning_rate": 0.00019999999994973653,
-      "loss": 46.0,
-      "step": 142
-    },
-    {
-      "epoch": 0.010933348624730011,
-      "grad_norm": 1.9854931451845914e-05,
-      "learning_rate": 0.00019999999994897206,
-      "loss": 46.0,
-      "step": 143
-    },
-    {
-      "epoch": 0.011009805608119731,
-      "grad_norm": 1.549391890875995e-05,
-      "learning_rate": 0.00019999999994820184,
-      "loss": 46.0,
-      "step": 144
-    },
-    {
-      "epoch": 0.011086262591509452,
-      "grad_norm": 1.8510312656871974e-05,
-      "learning_rate": 0.00019999999994742585,
-      "loss": 46.0,
-      "step": 145
-    },
-    {
-      "epoch": 0.011162719574899173,
-      "grad_norm": 7.089629070833325e-05,
-      "learning_rate": 0.00019999999994664412,
-      "loss": 46.0,
-      "step": 146
-    },
-    {
-      "epoch": 0.011239176558288893,
-      "grad_norm": 1.4443186955759302e-05,
-      "learning_rate": 0.00019999999994585655,
-      "loss": 46.0,
-      "step": 147
-    },
-    {
-      "epoch": 0.011315633541678614,
-      "grad_norm": 1.778230034688022e-05,
-      "learning_rate": 0.00019999999994506327,
-      "loss": 46.0,
-      "step": 148
-    },
-    {
-      "epoch": 0.011392090525068333,
-      "grad_norm": 1.8103053662343882e-05,
-      "learning_rate": 0.00019999999994426418,
-      "loss": 46.0,
-      "step": 149
-    },
-    {
-      "epoch": 0.011468547508458054,
-      "grad_norm": 1.6613617844996043e-05,
-      "learning_rate": 0.00019999999994345938,
-      "loss": 46.0,
-      "step": 150
-    },
-    {
-      "epoch": 0.011545004491847774,
-      "grad_norm": 9.72727775661042e-06,
-      "learning_rate": 0.00019999999994264875,
-      "loss": 46.0,
-      "step": 151
-    },
-    {
-      "epoch": 0.011621461475237495,
-      "grad_norm": 4.356676072347909e-05,
-      "learning_rate": 0.00019999999994183237,
-      "loss": 46.0,
-      "step": 152
-    },
-    {
-      "epoch": 0.011697918458627214,
-      "grad_norm": 1.8201697457698174e-05,
-      "learning_rate": 0.00019999999994101022,
-      "loss": 46.0,
-      "step": 153
-    },
-    {
-      "epoch": 0.011774375442016935,
-      "grad_norm": 1.672066173341591e-05,
-      "learning_rate": 0.0001999999999401823,
-      "loss": 46.0,
-      "step": 154
-    },
-    {
-      "epoch": 0.011850832425406655,
-      "grad_norm": 1.5591931514791213e-05,
-      "learning_rate": 0.0001999999999393486,
-      "loss": 46.0,
-      "step": 155
-    },
-    {
-      "epoch": 0.011927289408796376,
-      "grad_norm": 1.747531132423319e-05,
-      "learning_rate": 0.00019999999993850916,
-      "loss": 46.0,
-      "step": 156
-    },
-    {
-      "epoch": 0.012003746392186097,
-      "grad_norm": 2.5923198336386122e-05,
-      "learning_rate": 0.00019999999993766394,
-      "loss": 46.0,
-      "step": 157
-    },
-    {
-      "epoch": 0.012080203375575816,
-      "grad_norm": 1.825248727982398e-05,
-      "learning_rate": 0.00019999999993681295,
-      "loss": 46.0,
-      "step": 158
-    },
-    {
-      "epoch": 0.012156660358965538,
-      "grad_norm": 1.7134279914898798e-05,
-      "learning_rate": 0.00019999999993595616,
-      "loss": 46.0,
-      "step": 159
-    },
-    {
-      "epoch": 0.012233117342355257,
-      "grad_norm": 1.23955860544811e-05,
-      "learning_rate": 0.00019999999993509365,
-      "loss": 46.0,
-      "step": 160
-    },
-    {
-      "epoch": 0.012309574325744978,
-      "grad_norm": 1.9497911125654355e-05,
-      "learning_rate": 0.00019999999993422534,
-      "loss": 46.0,
-      "step": 161
-    },
-    {
-      "epoch": 0.012386031309134697,
-      "grad_norm": 1.4672481484012678e-05,
-      "learning_rate": 0.00019999999993335128,
-      "loss": 46.0,
-      "step": 162
-    },
-    {
-      "epoch": 0.012462488292524419,
-      "grad_norm": 1.6997120837913826e-05,
-      "learning_rate": 0.00019999999993247145,
-      "loss": 46.0,
-      "step": 163
-    },
-    {
-      "epoch": 0.012538945275914138,
-      "grad_norm": 1.876752867246978e-05,
-      "learning_rate": 0.00019999999993158582,
-      "loss": 46.0,
-      "step": 164
-    },
-    {
-      "epoch": 0.01261540225930386,
-      "grad_norm": 2.1698650016332977e-05,
-      "learning_rate": 0.00019999999993069442,
-      "loss": 46.0,
-      "step": 165
-    },
-    {
-      "epoch": 0.01269185924269358,
-      "grad_norm": 7.307679334189743e-05,
-      "learning_rate": 0.00019999999992979727,
-      "loss": 46.0,
-      "step": 166
-    },
-    {
-      "epoch": 0.0127683162260833,
-      "grad_norm": 9.80331878963625e-06,
-      "learning_rate": 0.00019999999992889437,
-      "loss": 46.0,
-      "step": 167
-    },
-    {
-      "epoch": 0.01284477320947302,
-      "grad_norm": 1.5055651601869613e-05,
-      "learning_rate": 0.00019999999992798567,
-      "loss": 46.0,
-      "step": 168
-    },
-    {
-      "epoch": 0.01292123019286274,
-      "grad_norm": 3.526509317453019e-05,
-      "learning_rate": 0.00019999999992707123,
-      "loss": 46.0,
-      "step": 169
-    },
-    {
-      "epoch": 0.012997687176252461,
-      "grad_norm": 5.940972550888546e-05,
-      "learning_rate": 0.000199999999926151,
-      "loss": 46.0,
-      "step": 170
-    },
-    {
-      "epoch": 0.01307414415964218,
-      "grad_norm": 1.5164956494118087e-05,
-      "learning_rate": 0.000199999999925225,
-      "loss": 46.0,
-      "step": 171
-    },
-    {
-      "epoch": 0.013150601143031902,
-      "grad_norm": 8.361901564057916e-06,
-      "learning_rate": 0.00019999999992429323,
-      "loss": 46.0,
-      "step": 172
-    },
-    {
-      "epoch": 0.013227058126421621,
-      "grad_norm": 1.1857358913403004e-05,
-      "learning_rate": 0.0001999999999233557,
-      "loss": 46.0,
-      "step": 173
-    },
-    {
-      "epoch": 0.013303515109811342,
-      "grad_norm": 2.8307274988037534e-05,
-      "learning_rate": 0.0001999999999224124,
-      "loss": 46.0,
-      "step": 174
-    },
-    {
-      "epoch": 0.013379972093201064,
-      "grad_norm": 1.7267728253500536e-05,
-      "learning_rate": 0.00019999999992146333,
-      "loss": 46.0,
-      "step": 175
-    },
-    {
-      "epoch": 0.013456429076590783,
-      "grad_norm": 2.200157723564189e-05,
-      "learning_rate": 0.00019999999992050847,
-      "loss": 46.0,
-      "step": 176
-    },
-    {
-      "epoch": 0.013532886059980504,
-      "grad_norm": 2.0857438357779756e-05,
-      "learning_rate": 0.00019999999991954784,
-      "loss": 46.0,
-      "step": 177
-    },
-    {
-      "epoch": 0.013609343043370224,
-      "grad_norm": 3.0190241886884905e-05,
-      "learning_rate": 0.00019999999991858147,
-      "loss": 46.0,
-      "step": 178
-    },
-    {
-      "epoch": 0.013685800026759945,
-      "grad_norm": 1.7986943930736743e-05,
-      "learning_rate": 0.00019999999991760932,
-      "loss": 46.0,
-      "step": 179
-    },
-    {
-      "epoch": 0.013762257010149664,
-      "grad_norm": 1.136941045842832e-05,
-      "learning_rate": 0.00019999999991663142,
-      "loss": 46.0,
-      "step": 180
-    },
-    {
-      "epoch": 0.013838713993539385,
-      "grad_norm": 2.2450272808782756e-05,
-      "learning_rate": 0.0001999999999156477,
-      "loss": 46.0,
-      "step": 181
-    },
-    {
-      "epoch": 0.013915170976929105,
-      "grad_norm": 2.47952248173533e-05,
-      "learning_rate": 0.00019999999991465823,
-      "loss": 46.0,
-      "step": 182
-    },
-    {
-      "epoch": 0.013991627960318826,
-      "grad_norm": 2.660143763932865e-05,
-      "learning_rate": 0.00019999999991366304,
-      "loss": 46.0,
-      "step": 183
-    },
-    {
-      "epoch": 0.014068084943708545,
-      "grad_norm": 4.1635230445535854e-05,
-      "learning_rate": 0.000199999999912662,
-      "loss": 46.0,
-      "step": 184
-    },
-    {
-      "epoch": 0.014144541927098266,
-      "grad_norm": 2.161614793294575e-05,
-      "learning_rate": 0.00019999999991165526,
-      "loss": 46.0,
-      "step": 185
-    },
-    {
-      "epoch": 0.014220998910487987,
-      "grad_norm": 1.880426316347439e-05,
-      "learning_rate": 0.0001999999999106427,
-      "loss": 46.0,
-      "step": 186
-    },
-    {
-      "epoch": 0.014297455893877707,
-      "grad_norm": 2.4131679310812615e-05,
-      "learning_rate": 0.00019999999990962439,
-      "loss": 46.0,
-      "step": 187
-    },
-    {
-      "epoch": 0.014373912877267428,
-      "grad_norm": 3.642721276264638e-05,
-      "learning_rate": 0.00019999999990860033,
-      "loss": 46.0,
-      "step": 188
-    },
-    {
-      "epoch": 0.014450369860657147,
-      "grad_norm": 1.3916353054810315e-05,
-      "learning_rate": 0.00019999999990757047,
-      "loss": 46.0,
-      "step": 189
-    },
-    {
-      "epoch": 0.014526826844046868,
-      "grad_norm": 1.937287743203342e-05,
-      "learning_rate": 0.00019999999990653487,
-      "loss": 46.0,
-      "step": 190
-    },
-    {
-      "epoch": 0.014603283827436588,
-      "grad_norm": 2.7034817321691662e-05,
-      "learning_rate": 0.00019999999990549347,
-      "loss": 46.0,
-      "step": 191
-    },
-    {
-      "epoch": 0.014679740810826309,
-      "grad_norm": 3.415107858018018e-05,
-      "learning_rate": 0.00019999999990444632,
-      "loss": 46.0,
-      "step": 192
-    },
-    {
-      "epoch": 0.014756197794216028,
-      "grad_norm": 1.394677019561641e-05,
-      "learning_rate": 0.00019999999990339337,
-      "loss": 46.0,
-      "step": 193
-    },
-    {
-      "epoch": 0.01483265477760575,
-      "grad_norm": 1.77701331267599e-05,
-      "learning_rate": 0.0001999999999023347,
-      "loss": 46.0,
-      "step": 194
-    },
-    {
-      "epoch": 0.01490911176099547,
-      "grad_norm": 4.800179522135295e-05,
-      "learning_rate": 0.00019999999990127023,
-      "loss": 46.0,
-      "step": 195
-    },
-    {
-      "epoch": 0.01498556874438519,
-      "grad_norm": 3.9848117012297735e-05,
-      "learning_rate": 0.00019999999990019999,
-      "loss": 46.0,
-      "step": 196
-    },
-    {
-      "epoch": 0.015062025727774911,
-      "grad_norm": 4.420596087584272e-05,
-      "learning_rate": 0.000199999999899124,
-      "loss": 46.0,
-      "step": 197
-    },
-    {
-      "epoch": 0.01513848271116463,
-      "grad_norm": 3.029867002624087e-05,
-      "learning_rate": 0.0001999999998980422,
-      "loss": 46.0,
-      "step": 198
-    },
-    {
-      "epoch": 0.015214939694554352,
-      "grad_norm": 0.00012269854778423905,
-      "learning_rate": 0.00019999999989695467,
-      "loss": 46.0,
-      "step": 199
-    },
-    {
-      "epoch": 0.015291396677944071,
-      "grad_norm": 6.0171842051204294e-05,
-      "learning_rate": 0.00019999999989586136,
-      "loss": 46.0,
-      "step": 200
-    },
-    {
-      "epoch": 0.015367853661333792,
-      "grad_norm": 3.885979822371155e-05,
-      "learning_rate": 0.00019999999989476228,
-      "loss": 46.0,
-      "step": 201
-    },
-    {
-      "epoch": 0.015444310644723512,
-      "grad_norm": 6.002607915434055e-05,
-      "learning_rate": 0.00019999999989365742,
-      "loss": 46.0,
-      "step": 202
-    },
-    {
-      "epoch": 0.015520767628113233,
-      "grad_norm": 3.7406844057841226e-05,
-      "learning_rate": 0.00019999999989254682,
-      "loss": 46.0,
-      "step": 203
-    },
-    {
-      "epoch": 0.015597224611502954,
-      "grad_norm": 4.4123895349912345e-05,
-      "learning_rate": 0.00019999999989143042,
-      "loss": 46.0,
-      "step": 204
-    },
-    {
-      "epoch": 0.015673681594892675,
-      "grad_norm": 2.500749542377889e-05,
-      "learning_rate": 0.00019999999989030827,
-      "loss": 46.0,
-      "step": 205
-    },
-    {
-      "epoch": 0.015750138578282395,
-      "grad_norm": 4.2419942474225536e-05,
-      "learning_rate": 0.00019999999988918034,
-      "loss": 46.0,
-      "step": 206
-    },
-    {
-      "epoch": 0.015826595561672114,
-      "grad_norm": 2.4852703063515946e-05,
-      "learning_rate": 0.00019999999988804662,
-      "loss": 46.0,
-      "step": 207
-    },
-    {
-      "epoch": 0.015903052545061833,
-      "grad_norm": 1.3398740520642605e-05,
-      "learning_rate": 0.00019999999988690718,
-      "loss": 46.0,
-      "step": 208
-    },
-    {
-      "epoch": 0.015979509528451556,
-      "grad_norm": 3.970724719692953e-05,
-      "learning_rate": 0.00019999999988576194,
-      "loss": 46.0,
-      "step": 209
-    },
-    {
-      "epoch": 0.016055966511841276,
-      "grad_norm": 3.1210820452542976e-05,
-      "learning_rate": 0.00019999999988461092,
-      "loss": 46.0,
-      "step": 210
-    },
-    {
-      "epoch": 0.016132423495230995,
-      "grad_norm": 4.0842809539753944e-05,
-      "learning_rate": 0.00019999999988345416,
-      "loss": 46.0,
-      "step": 211
-    },
-    {
-      "epoch": 0.016208880478620714,
-      "grad_norm": 1.1598122910072561e-05,
-      "learning_rate": 0.00019999999988229162,
-      "loss": 46.0,
-      "step": 212
-    },
-    {
-      "epoch": 0.016285337462010437,
-      "grad_norm": 1.5735586202936247e-05,
-      "learning_rate": 0.00019999999988112329,
-      "loss": 46.0,
-      "step": 213
-    },
-    {
-      "epoch": 0.016361794445400157,
-      "grad_norm": 1.9531549696694128e-05,
-      "learning_rate": 0.0001999999998799492,
-      "loss": 46.0,
-      "step": 214
-    },
-    {
-      "epoch": 0.016438251428789876,
-      "grad_norm": 1.3231883713160641e-05,
-      "learning_rate": 0.00019999999987876935,
-      "loss": 46.0,
-      "step": 215
-    },
-    {
-      "epoch": 0.0165147084121796,
-      "grad_norm": 7.973035098984838e-05,
-      "learning_rate": 0.00019999999987758375,
-      "loss": 46.0,
-      "step": 216
-    },
-    {
-      "epoch": 0.01659116539556932,
-      "grad_norm": 3.848735286737792e-05,
-      "learning_rate": 0.00019999999987639234,
-      "loss": 46.0,
-      "step": 217
-    },
-    {
-      "epoch": 0.016667622378959038,
-      "grad_norm": 4.15618451370392e-05,
-      "learning_rate": 0.00019999999987519517,
-      "loss": 46.0,
-      "step": 218
-    },
-    {
-      "epoch": 0.016744079362348757,
-      "grad_norm": 3.883221870637499e-05,
-      "learning_rate": 0.00019999999987399224,
-      "loss": 46.0,
-      "step": 219
-    },
-    {
-      "epoch": 0.01682053634573848,
-      "grad_norm": 2.038398633885663e-05,
-      "learning_rate": 0.00019999999987278355,
-      "loss": 46.0,
-      "step": 220
-    },
-    {
-      "epoch": 0.0168969933291282,
-      "grad_norm": 1.29841910165851e-05,
-      "learning_rate": 0.00019999999987156908,
-      "loss": 46.0,
-      "step": 221
-    },
-    {
-      "epoch": 0.01697345031251792,
-      "grad_norm": 3.4105782106053084e-05,
-      "learning_rate": 0.00019999999987034884,
-      "loss": 46.0,
-      "step": 222
-    },
-    {
-      "epoch": 0.017049907295907638,
-      "grad_norm": 6.171867426019162e-05,
-      "learning_rate": 0.00019999999986912282,
-      "loss": 46.0,
-      "step": 223
-    },
-    {
-      "epoch": 0.01712636427929736,
-      "grad_norm": 1.5947302017593756e-05,
-      "learning_rate": 0.00019999999986789106,
-      "loss": 46.0,
-      "step": 224
-    },
-    {
-      "epoch": 0.01720282126268708,
-      "grad_norm": 1.7624932297621854e-05,
-      "learning_rate": 0.0001999999998666535,
-      "loss": 46.0,
-      "step": 225
-    },
-    {
-      "epoch": 0.0172792782460768,
-      "grad_norm": 1.3201419278630055e-05,
-      "learning_rate": 0.0001999999998654102,
-      "loss": 46.0,
-      "step": 226
-    },
-    {
-      "epoch": 0.017355735229466523,
-      "grad_norm": 2.5239411115762778e-05,
-      "learning_rate": 0.00019999999986416108,
-      "loss": 46.0,
-      "step": 227
-    },
-    {
-      "epoch": 0.017432192212856242,
-      "grad_norm": 3.751131225726567e-05,
-      "learning_rate": 0.00019999999986290625,
-      "loss": 46.0,
-      "step": 228
-    },
-    {
-      "epoch": 0.01750864919624596,
-      "grad_norm": 5.078049434814602e-05,
-      "learning_rate": 0.00019999999986164562,
-      "loss": 46.0,
-      "step": 229
-    },
-    {
-      "epoch": 0.01758510617963568,
-      "grad_norm": 1.581860487931408e-05,
-      "learning_rate": 0.00019999999986037922,
-      "loss": 46.0,
-      "step": 230
-    },
-    {
-      "epoch": 0.017661563163025404,
-      "grad_norm": 1.7339454643661156e-05,
-      "learning_rate": 0.00019999999985910707,
-      "loss": 46.0,
-      "step": 231
-    },
-    {
-      "epoch": 0.017738020146415123,
-      "grad_norm": 2.3564203729620203e-05,
-      "learning_rate": 0.00019999999985782912,
-      "loss": 46.0,
-      "step": 232
-    },
-    {
-      "epoch": 0.017814477129804843,
-      "grad_norm": 3.812616705545224e-05,
-      "learning_rate": 0.00019999999985654543,
-      "loss": 46.0,
-      "step": 233
-    },
-    {
-      "epoch": 0.017890934113194566,
-      "grad_norm": 6.810265040257946e-05,
-      "learning_rate": 0.00019999999985525596,
-      "loss": 46.0,
-      "step": 234
-    },
-    {
-      "epoch": 0.017967391096584285,
-      "grad_norm": 2.8255031793378294e-05,
-      "learning_rate": 0.0001999999998539607,
-      "loss": 46.0,
-      "step": 235
-    },
-    {
-      "epoch": 0.018043848079974004,
-      "grad_norm": 1.70758521562675e-05,
-      "learning_rate": 0.0001999999998526597,
-      "loss": 46.0,
-      "step": 236
-    },
-    {
-      "epoch": 0.018120305063363724,
-      "grad_norm": 3.266323619754985e-05,
-      "learning_rate": 0.0001999999998513529,
-      "loss": 46.0,
-      "step": 237
-    },
-    {
-      "epoch": 0.018196762046753447,
-      "grad_norm": 2.1402192942332476e-05,
-      "learning_rate": 0.00019999999985004038,
-      "loss": 46.0,
-      "step": 238
-    },
-    {
-      "epoch": 0.018273219030143166,
-      "grad_norm": 1.3219912034401204e-05,
-      "learning_rate": 0.00019999999984872204,
-      "loss": 46.0,
-      "step": 239
-    },
-    {
-      "epoch": 0.018349676013532885,
-      "grad_norm": 1.8829701730282977e-05,
-      "learning_rate": 0.00019999999984739796,
-      "loss": 46.0,
-      "step": 240
-    },
-    {
-      "epoch": 0.018426132996922605,
-      "grad_norm": 3.477720747468993e-05,
-      "learning_rate": 0.00019999999984606808,
-      "loss": 46.0,
-      "step": 241
-    },
-    {
-      "epoch": 0.018502589980312328,
-      "grad_norm": 2.4586177460150793e-05,
-      "learning_rate": 0.00019999999984473245,
-      "loss": 46.0,
-      "step": 242
-    },
-    {
-      "epoch": 0.018579046963702047,
-      "grad_norm": 2.229872734460514e-05,
-      "learning_rate": 0.00019999999984339107,
-      "loss": 46.0,
-      "step": 243
-    },
-    {
-      "epoch": 0.018655503947091766,
-      "grad_norm": 5.301136116031557e-05,
-      "learning_rate": 0.0001999999998420439,
-      "loss": 46.0,
-      "step": 244
-    },
-    {
-      "epoch": 0.01873196093048149,
-      "grad_norm": 6.441616278607398e-05,
-      "learning_rate": 0.00019999999984069095,
-      "loss": 46.0,
-      "step": 245
-    },
-    {
-      "epoch": 0.01880841791387121,
-      "grad_norm": 6.231950828805566e-05,
-      "learning_rate": 0.00019999999983933225,
-      "loss": 46.0,
-      "step": 246
-    },
-    {
-      "epoch": 0.018884874897260928,
-      "grad_norm": 2.3630893338122405e-05,
-      "learning_rate": 0.00019999999983796778,
-      "loss": 46.0,
-      "step": 247
-    },
-    {
-      "epoch": 0.018961331880650648,
-      "grad_norm": 4.4150590838398784e-05,
-      "learning_rate": 0.00019999999983659751,
-      "loss": 46.0,
-      "step": 248
-    },
-    {
-      "epoch": 0.01903778886404037,
-      "grad_norm": 4.186800288152881e-05,
-      "learning_rate": 0.0001999999998352215,
-      "loss": 46.0,
-      "step": 249
-    },
-    {
-      "epoch": 0.01911424584743009,
-      "grad_norm": 1.8623808500706218e-05,
-      "learning_rate": 0.0001999999998338397,
-      "loss": 46.0,
-      "step": 250
-    },
-    {
-      "epoch": 0.01919070283081981,
-      "grad_norm": 7.354863191721961e-05,
-      "learning_rate": 0.00019999999983245218,
-      "loss": 46.0,
-      "step": 251
-    },
-    {
-      "epoch": 0.01926715981420953,
-      "grad_norm": 2.2413632905227132e-05,
-      "learning_rate": 0.00019999999983105884,
-      "loss": 46.0,
-      "step": 252
-    },
-    {
-      "epoch": 0.01934361679759925,
-      "grad_norm": 3.723469126271084e-05,
-      "learning_rate": 0.00019999999982965976,
-      "loss": 46.0,
-      "step": 253
-    },
-    {
-      "epoch": 0.01942007378098897,
-      "grad_norm": 1.9338929632795043e-05,
-      "learning_rate": 0.0001999999998282549,
-      "loss": 46.0,
-      "step": 254
-    },
-    {
-      "epoch": 0.01949653076437869,
-      "grad_norm": 2.176074485760182e-05,
-      "learning_rate": 0.00019999999982684425,
-      "loss": 46.0,
-      "step": 255
-    },
-    {
-      "epoch": 0.019572987747768413,
-      "grad_norm": 5.1866609283024445e-05,
-      "learning_rate": 0.00019999999982542787,
-      "loss": 46.0,
-      "step": 256
-    },
-    {
-      "epoch": 0.019649444731158133,
-      "grad_norm": 4.26997612521518e-05,
-      "learning_rate": 0.00019999999982400567,
-      "loss": 46.0,
-      "step": 257
-    },
-    {
-      "epoch": 0.019725901714547852,
-      "grad_norm": 2.0059625967405736e-05,
-      "learning_rate": 0.00019999999982257775,
-      "loss": 46.0,
-      "step": 258
-    },
-    {
-      "epoch": 0.01980235869793757,
-      "grad_norm": 3.8806952943559736e-05,
-      "learning_rate": 0.00019999999982114406,
-      "loss": 46.0,
-      "step": 259
-    },
-    {
-      "epoch": 0.019878815681327294,
-      "grad_norm": 3.745260619325563e-05,
-      "learning_rate": 0.00019999999981970456,
-      "loss": 46.0,
-      "step": 260
-    },
-    {
-      "epoch": 0.019955272664717014,
-      "grad_norm": 4.402872218634002e-05,
-      "learning_rate": 0.00019999999981825932,
-      "loss": 46.0,
-      "step": 261
-    },
-    {
-      "epoch": 0.020031729648106733,
-      "grad_norm": 1.1846618690469768e-05,
-      "learning_rate": 0.0001999999998168083,
-      "loss": 46.0,
-      "step": 262
-    },
-    {
-      "epoch": 0.020108186631496456,
-      "grad_norm": 1.715353391773533e-05,
-      "learning_rate": 0.00019999999981535152,
-      "loss": 46.0,
-      "step": 263
-    },
-    {
-      "epoch": 0.020184643614886175,
-      "grad_norm": 3.253341856179759e-05,
-      "learning_rate": 0.00019999999981388898,
-      "loss": 46.0,
-      "step": 264
-    },
-    {
-      "epoch": 0.020261100598275895,
-      "grad_norm": 3.172983633703552e-05,
-      "learning_rate": 0.00019999999981242065,
-      "loss": 46.0,
-      "step": 265
-    },
-    {
-      "epoch": 0.020337557581665614,
-      "grad_norm": 5.8371944760438055e-05,
-      "learning_rate": 0.00019999999981094654,
-      "loss": 46.0,
-      "step": 266
-    },
-    {
-      "epoch": 0.020414014565055337,
-      "grad_norm": 2.696877345442772e-05,
-      "learning_rate": 0.00019999999980946668,
-      "loss": 46.0,
-      "step": 267
-    },
-    {
-      "epoch": 0.020490471548445056,
-      "grad_norm": 3.4904631320387125e-05,
-      "learning_rate": 0.00019999999980798103,
-      "loss": 46.0,
-      "step": 268
-    },
-    {
-      "epoch": 0.020566928531834776,
-      "grad_norm": 4.222767529427074e-05,
-      "learning_rate": 0.00019999999980648965,
-      "loss": 46.0,
-      "step": 269
-    },
-    {
-      "epoch": 0.020643385515224495,
-      "grad_norm": 2.7723754101316445e-05,
-      "learning_rate": 0.00019999999980499245,
-      "loss": 46.0,
-      "step": 270
-    },
-    {
-      "epoch": 0.020719842498614218,
-      "grad_norm": 7.65265867812559e-05,
-      "learning_rate": 0.0001999999998034895,
-      "loss": 46.0,
-      "step": 271
-    },
-    {
-      "epoch": 0.020796299482003937,
-      "grad_norm": 6.078218575567007e-05,
-      "learning_rate": 0.0001999999998019808,
-      "loss": 46.0,
-      "step": 272
-    },
-    {
-      "epoch": 0.020872756465393657,
-      "grad_norm": 2.5883078706101514e-05,
-      "learning_rate": 0.00019999999980046634,
-      "loss": 46.0,
-      "step": 273
-    },
-    {
-      "epoch": 0.02094921344878338,
-      "grad_norm": 3.141328488709405e-05,
-      "learning_rate": 0.00019999999979894607,
-      "loss": 46.0,
-      "step": 274
-    },
-    {
-      "epoch": 0.0210256704321731,
-      "grad_norm": 4.7406323574250564e-05,
-      "learning_rate": 0.00019999999979742006,
-      "loss": 46.0,
-      "step": 275
-    },
-    {
-      "epoch": 0.02110212741556282,
-      "grad_norm": 3.114222636213526e-05,
-      "learning_rate": 0.00019999999979588827,
-      "loss": 46.0,
-      "step": 276
-    },
-    {
-      "epoch": 0.021178584398952538,
-      "grad_norm": 3.695775012602098e-05,
-      "learning_rate": 0.00019999999979435068,
-      "loss": 46.0,
-      "step": 277
-    },
-    {
-      "epoch": 0.02125504138234226,
-      "grad_norm": 4.6277407818706706e-05,
-      "learning_rate": 0.00019999999979280738,
-      "loss": 46.0,
-      "step": 278
-    },
-    {
-      "epoch": 0.02133149836573198,
-      "grad_norm": 3.713286423590034e-05,
-      "learning_rate": 0.00019999999979125827,
-      "loss": 46.0,
-      "step": 279
-    },
-    {
-      "epoch": 0.0214079553491217,
-      "grad_norm": 5.980370769975707e-05,
-      "learning_rate": 0.0001999999997897034,
-      "loss": 46.0,
-      "step": 280
-    },
-    {
-      "epoch": 0.02148441233251142,
-      "grad_norm": 2.9063308829790913e-05,
-      "learning_rate": 0.00019999999978814276,
-      "loss": 46.0,
-      "step": 281
-    },
-    {
-      "epoch": 0.021560869315901142,
-      "grad_norm": 2.6975656510330737e-05,
-      "learning_rate": 0.00019999999978657638,
-      "loss": 46.0,
-      "step": 282
-    },
-    {
-      "epoch": 0.02163732629929086,
-      "grad_norm": 7.82805509516038e-05,
-      "learning_rate": 0.00019999999978500418,
-      "loss": 46.0,
-      "step": 283
-    },
-    {
-      "epoch": 0.02171378328268058,
-      "grad_norm": 3.2012809242587537e-05,
-      "learning_rate": 0.00019999999978342623,
-      "loss": 46.0,
-      "step": 284
-    },
-    {
-      "epoch": 0.021790240266070304,
-      "grad_norm": 8.559294656151906e-05,
-      "learning_rate": 0.0001999999997818425,
-      "loss": 46.0,
-      "step": 285
-    },
-    {
-      "epoch": 0.021866697249460023,
-      "grad_norm": 2.8340638891677372e-05,
-      "learning_rate": 0.00019999999978025305,
-      "loss": 46.0,
-      "step": 286
-    },
-    {
-      "epoch": 0.021943154232849742,
-      "grad_norm": 2.6537731173448265e-05,
-      "learning_rate": 0.00019999999977865778,
-      "loss": 46.0,
-      "step": 287
-    },
-    {
-      "epoch": 0.022019611216239462,
-      "grad_norm": 7.063262455631047e-05,
-      "learning_rate": 0.00019999999977705674,
-      "loss": 46.0,
-      "step": 288
-    },
-    {
-      "epoch": 0.022096068199629185,
-      "grad_norm": 0.00013856767327524722,
-      "learning_rate": 0.00019999999977544998,
-      "loss": 46.0,
-      "step": 289
-    },
-    {
-      "epoch": 0.022172525183018904,
-      "grad_norm": 4.319584695622325e-05,
-      "learning_rate": 0.00019999999977383742,
-      "loss": 46.0,
-      "step": 290
-    },
-    {
-      "epoch": 0.022248982166408623,
-      "grad_norm": 6.633839075220749e-05,
-      "learning_rate": 0.00019999999977221905,
-      "loss": 46.0,
-      "step": 291
-    },
-    {
-      "epoch": 0.022325439149798346,
-      "grad_norm": 2.470920117048081e-05,
-      "learning_rate": 0.00019999999977059497,
-      "loss": 46.0,
-      "step": 292
-    },
-    {
-      "epoch": 0.022401896133188066,
-      "grad_norm": 2.4265709726023488e-05,
-      "learning_rate": 0.00019999999976896512,
-      "loss": 46.0,
-      "step": 293
-    },
-    {
-      "epoch": 0.022478353116577785,
-      "grad_norm": 0.00011469124729046598,
-      "learning_rate": 0.00019999999976732946,
-      "loss": 46.0,
-      "step": 294
-    },
-    {
-      "epoch": 0.022554810099967505,
-      "grad_norm": 3.213988748029806e-05,
-      "learning_rate": 0.00019999999976568806,
-      "loss": 46.0,
-      "step": 295
-    },
-    {
-      "epoch": 0.022631267083357227,
-      "grad_norm": 1.801499820430763e-05,
-      "learning_rate": 0.00019999999976404086,
-      "loss": 46.0,
-      "step": 296
-    },
-    {
-      "epoch": 0.022707724066746947,
-      "grad_norm": 3.2129533792613074e-05,
-      "learning_rate": 0.00019999999976238791,
-      "loss": 46.0,
-      "step": 297
-    },
-    {
-      "epoch": 0.022784181050136666,
-      "grad_norm": 5.563296508626081e-05,
-      "learning_rate": 0.0001999999997607292,
-      "loss": 46.0,
-      "step": 298
-    },
-    {
-      "epoch": 0.022860638033526386,
-      "grad_norm": 5.471260010381229e-05,
-      "learning_rate": 0.00019999999975906473,
-      "loss": 46.0,
-      "step": 299
-    },
-    {
-      "epoch": 0.02293709501691611,
-      "grad_norm": 5.7839060900732875e-05,
-      "learning_rate": 0.00019999999975739449,
-      "loss": 46.0,
-      "step": 300
-    },
-    {
-      "epoch": 0.023013552000305828,
-      "grad_norm": 3.9000693504931405e-05,
-      "learning_rate": 0.00019999999975571845,
-      "loss": 46.0,
-      "step": 301
-    },
-    {
-      "epoch": 0.023090008983695547,
-      "grad_norm": 2.7543690521270037e-05,
-      "learning_rate": 0.00019999999975403663,
-      "loss": 46.0,
-      "step": 302
-    },
-    {
-      "epoch": 0.02316646596708527,
-      "grad_norm": 3.6833534977631643e-05,
-      "learning_rate": 0.0001999999997523491,
-      "loss": 46.0,
-      "step": 303
-    },
-    {
-      "epoch": 0.02324292295047499,
-      "grad_norm": 2.8821881642215885e-05,
-      "learning_rate": 0.00019999999975065574,
-      "loss": 46.0,
-      "step": 304
-    },
-    {
-      "epoch": 0.02331937993386471,
-      "grad_norm": 2.8563161322381347e-05,
-      "learning_rate": 0.00019999999974895666,
-      "loss": 46.0,
-      "step": 305
-    },
-    {
-      "epoch": 0.02339583691725443,
-      "grad_norm": 3.7597033951897174e-05,
-      "learning_rate": 0.00019999999974725177,
-      "loss": 46.0,
-      "step": 306
-    },
-    {
-      "epoch": 0.02347229390064415,
-      "grad_norm": 8.671709656482562e-05,
-      "learning_rate": 0.00019999999974554112,
-      "loss": 46.0,
-      "step": 307
-    },
-    {
-      "epoch": 0.02354875088403387,
-      "grad_norm": 3.126203591818921e-05,
-      "learning_rate": 0.0001999999997438247,
-      "loss": 46.0,
-      "step": 308
-    },
-    {
-      "epoch": 0.02362520786742359,
-      "grad_norm": 7.032913708826527e-05,
-      "learning_rate": 0.00019999999974210252,
-      "loss": 46.0,
-      "step": 309
-    },
-    {
-      "epoch": 0.02370166485081331,
-      "grad_norm": 3.444773028604686e-05,
-      "learning_rate": 0.00019999999974037457,
-      "loss": 46.0,
-      "step": 310
-    },
-    {
-      "epoch": 0.023778121834203032,
-      "grad_norm": 2.260547989862971e-05,
-      "learning_rate": 0.00019999999973864085,
-      "loss": 46.0,
-      "step": 311
-    },
-    {
-      "epoch": 0.02385457881759275,
-      "grad_norm": 4.152274050284177e-05,
-      "learning_rate": 0.00019999999973690136,
-      "loss": 46.0,
-      "step": 312
-    },
-    {
-      "epoch": 0.02393103580098247,
-      "grad_norm": 3.422985173529014e-05,
-      "learning_rate": 0.0001999999997351561,
-      "loss": 46.0,
-      "step": 313
-    },
-    {
-      "epoch": 0.024007492784372194,
-      "grad_norm": 5.12860351591371e-05,
-      "learning_rate": 0.00019999999973340508,
-      "loss": 46.0,
-      "step": 314
-    },
-    {
-      "epoch": 0.024083949767761913,
-      "grad_norm": 2.766688157862518e-05,
-      "learning_rate": 0.00019999999973164826,
-      "loss": 46.0,
-      "step": 315
-    },
-    {
-      "epoch": 0.024160406751151633,
-      "grad_norm": 3.869930515065789e-05,
-      "learning_rate": 0.0001999999997298857,
-      "loss": 46.0,
-      "step": 316
-    },
-    {
-      "epoch": 0.024236863734541352,
-      "grad_norm": 4.1635878005763516e-05,
-      "learning_rate": 0.00019999999972811737,
-      "loss": 46.0,
-      "step": 317
-    },
-    {
-      "epoch": 0.024313320717931075,
-      "grad_norm": 2.8059675969416276e-05,
-      "learning_rate": 0.00019999999972634326,
-      "loss": 46.0,
-      "step": 318
-    },
-    {
-      "epoch": 0.024389777701320794,
-      "grad_norm": 2.6571789931040257e-05,
-      "learning_rate": 0.00019999999972456338,
-      "loss": 46.0,
-      "step": 319
-    },
-    {
-      "epoch": 0.024466234684710514,
-      "grad_norm": 0.00010475471208337694,
-      "learning_rate": 0.00019999999972277775,
-      "loss": 46.0,
-      "step": 320
-    },
-    {
-      "epoch": 0.024542691668100237,
-      "grad_norm": 7.082799129420891e-05,
-      "learning_rate": 0.0001999999997209863,
-      "loss": 46.0,
-      "step": 321
-    },
-    {
-      "epoch": 0.024619148651489956,
-      "grad_norm": 3.336452573421411e-05,
-      "learning_rate": 0.00019999999971918913,
-      "loss": 46.0,
-      "step": 322
-    },
-    {
-      "epoch": 0.024695605634879676,
-      "grad_norm": 4.1005736420629546e-05,
-      "learning_rate": 0.00019999999971738618,
-      "loss": 46.0,
-      "step": 323
-    },
-    {
-      "epoch": 0.024772062618269395,
-      "grad_norm": 3.6239729524822906e-05,
-      "learning_rate": 0.00019999999971557746,
-      "loss": 46.0,
-      "step": 324
-    },
-    {
-      "epoch": 0.024848519601659118,
-      "grad_norm": 5.612520908471197e-05,
-      "learning_rate": 0.00019999999971376297,
-      "loss": 46.0,
-      "step": 325
-    },
-    {
-      "epoch": 0.024924976585048837,
-      "grad_norm": 4.01418365072459e-05,
-      "learning_rate": 0.0001999999997119427,
-      "loss": 46.0,
-      "step": 326
-    },
-    {
-      "epoch": 0.025001433568438557,
-      "grad_norm": 7.070326682878658e-05,
-      "learning_rate": 0.00019999999971011666,
-      "loss": 46.0,
-      "step": 327
-    },
-    {
-      "epoch": 0.025077890551828276,
-      "grad_norm": 2.9590713893412612e-05,
-      "learning_rate": 0.00019999999970828487,
-      "loss": 46.0,
-      "step": 328
-    },
-    {
-      "epoch": 0.025154347535218,
-      "grad_norm": 4.511591396294534e-05,
-      "learning_rate": 0.00019999999970644728,
-      "loss": 46.0,
-      "step": 329
-    },
-    {
-      "epoch": 0.02523080451860772,
-      "grad_norm": 5.778960985480808e-05,
-      "learning_rate": 0.00019999999970460395,
-      "loss": 46.0,
-      "step": 330
-    },
-    {
-      "epoch": 0.025307261501997438,
-      "grad_norm": 5.33730853931047e-05,
-      "learning_rate": 0.00019999999970275484,
-      "loss": 46.0,
-      "step": 331
-    },
-    {
-      "epoch": 0.02538371848538716,
-      "grad_norm": 5.5860473366919905e-05,
-      "learning_rate": 0.00019999999970089996,
-      "loss": 46.0,
-      "step": 332
-    },
-    {
-      "epoch": 0.02546017546877688,
-      "grad_norm": 3.790226764976978e-05,
-      "learning_rate": 0.00019999999969903934,
-      "loss": 46.0,
-      "step": 333
-    },
-    {
-      "epoch": 0.0255366324521666,
-      "grad_norm": 2.741688149399124e-05,
-      "learning_rate": 0.0001999999996971729,
-      "loss": 46.0,
-      "step": 334
-    },
-    {
-      "epoch": 0.02561308943555632,
-      "grad_norm": 5.0812501285690814e-05,
-      "learning_rate": 0.0001999999996953007,
-      "loss": 46.0,
-      "step": 335
-    },
-    {
-      "epoch": 0.02568954641894604,
-      "grad_norm": 0.000129085558000952,
-      "learning_rate": 0.00019999999969342276,
-      "loss": 46.0,
-      "step": 336
-    },
-    {
-      "epoch": 0.02576600340233576,
-      "grad_norm": 8.084265573415905e-05,
-      "learning_rate": 0.00019999999969153902,
-      "loss": 46.0,
-      "step": 337
-    },
-    {
-      "epoch": 0.02584246038572548,
-      "grad_norm": 3.692197424243204e-05,
-      "learning_rate": 0.00019999999968964952,
-      "loss": 46.0,
-      "step": 338
-    },
-    {
-      "epoch": 0.0259189173691152,
-      "grad_norm": 4.907184120384045e-05,
-      "learning_rate": 0.00019999999968775426,
-      "loss": 46.0,
-      "step": 339
-    },
-    {
-      "epoch": 0.025995374352504923,
-      "grad_norm": 8.835240441840142e-05,
-      "learning_rate": 0.00019999999968585324,
-      "loss": 46.0,
-      "step": 340
-    },
-    {
-      "epoch": 0.026071831335894642,
-      "grad_norm": 3.8036225305404514e-05,
-      "learning_rate": 0.00019999999968394643,
-      "loss": 46.0,
-      "step": 341
-    },
-    {
-      "epoch": 0.02614828831928436,
-      "grad_norm": 2.0803265215363353e-05,
-      "learning_rate": 0.00019999999968203387,
-      "loss": 46.0,
-      "step": 342
-    },
-    {
-      "epoch": 0.026224745302674084,
-      "grad_norm": 5.061951742391102e-05,
-      "learning_rate": 0.0001999999996801155,
-      "loss": 46.0,
-      "step": 343
-    },
-    {
-      "epoch": 0.026301202286063804,
-      "grad_norm": 2.6827849069377407e-05,
-      "learning_rate": 0.00019999999967819138,
-      "loss": 46.0,
-      "step": 344
-    },
-    {
-      "epoch": 0.026377659269453523,
-      "grad_norm": 0.00010376806312706321,
-      "learning_rate": 0.0001999999996762615,
-      "loss": 46.0,
-      "step": 345
-    },
-    {
-      "epoch": 0.026454116252843243,
-      "grad_norm": 3.130981349386275e-05,
-      "learning_rate": 0.00019999999967432584,
-      "loss": 46.0,
-      "step": 346
-    },
-    {
-      "epoch": 0.026530573236232965,
-      "grad_norm": 6.62563179503195e-05,
-      "learning_rate": 0.00019999999967238444,
-      "loss": 46.0,
-      "step": 347
-    },
-    {
-      "epoch": 0.026607030219622685,
-      "grad_norm": 5.259773388388567e-05,
-      "learning_rate": 0.00019999999967043724,
-      "loss": 46.0,
-      "step": 348
-    },
-    {
-      "epoch": 0.026683487203012404,
-      "grad_norm": 2.115760071319528e-05,
-      "learning_rate": 0.00019999999966848427,
-      "loss": 46.0,
-      "step": 349
-    },
-    {
-      "epoch": 0.026759944186402127,
-      "grad_norm": 8.141921716742218e-05,
-      "learning_rate": 0.00019999999966652555,
-      "loss": 46.0,
-      "step": 350
-    },
-    {
-      "epoch": 0.026836401169791847,
-      "grad_norm": 5.7557695981813595e-05,
-      "learning_rate": 0.00019999999966456106,
-      "loss": 46.0,
-      "step": 351
-    },
-    {
-      "epoch": 0.026912858153181566,
-      "grad_norm": 3.140844273730181e-05,
-      "learning_rate": 0.0001999999996625908,
-      "loss": 46.0,
-      "step": 352
-    },
-    {
-      "epoch": 0.026989315136571285,
-      "grad_norm": 0.0001673676015343517,
-      "learning_rate": 0.00019999999966061475,
-      "loss": 46.0,
-      "step": 353
-    },
-    {
-      "epoch": 0.027065772119961008,
-      "grad_norm": 2.577383020252455e-05,
-      "learning_rate": 0.00019999999965863294,
-      "loss": 46.0,
-      "step": 354
-    },
-    {
-      "epoch": 0.027142229103350728,
-      "grad_norm": 4.645596709451638e-05,
-      "learning_rate": 0.00019999999965664538,
-      "loss": 46.0,
-      "step": 355
-    },
-    {
-      "epoch": 0.027218686086740447,
-      "grad_norm": 4.3698026274796575e-05,
-      "learning_rate": 0.00019999999965465205,
-      "loss": 46.0,
-      "step": 356
-    },
-    {
-      "epoch": 0.027295143070130166,
-      "grad_norm": 5.712471102015115e-05,
-      "learning_rate": 0.0001999999996526529,
-      "loss": 46.0,
-      "step": 357
-    },
-    {
-      "epoch": 0.02737160005351989,
-      "grad_norm": 4.390337198856287e-05,
-      "learning_rate": 0.000199999999650648,
-      "loss": 46.0,
-      "step": 358
-    },
-    {
-      "epoch": 0.02744805703690961,
-      "grad_norm": 7.158845255617052e-05,
-      "learning_rate": 0.00019999999964863738,
-      "loss": 46.0,
-      "step": 359
-    },
-    {
-      "epoch": 0.027524514020299328,
-      "grad_norm": 4.132555841351859e-05,
-      "learning_rate": 0.00019999999964662095,
-      "loss": 46.0,
-      "step": 360
-    },
-    {
-      "epoch": 0.02760097100368905,
-      "grad_norm": 5.734768637921661e-05,
-      "learning_rate": 0.00019999999964459875,
-      "loss": 46.0,
-      "step": 361
-    },
-    {
-      "epoch": 0.02767742798707877,
-      "grad_norm": 5.01650501973927e-05,
-      "learning_rate": 0.00019999999964257078,
-      "loss": 46.0,
-      "step": 362
-    },
-    {
-      "epoch": 0.02775388497046849,
-      "grad_norm": 6.403920997399837e-05,
-      "learning_rate": 0.00019999999964053706,
-      "loss": 46.0,
-      "step": 363
-    },
-    {
-      "epoch": 0.02783034195385821,
-      "grad_norm": 9.236169717041776e-05,
-      "learning_rate": 0.00019999999963849754,
-      "loss": 46.0,
-      "step": 364
-    },
-    {
-      "epoch": 0.027906798937247932,
-      "grad_norm": 0.00011183096648892388,
-      "learning_rate": 0.00019999999963645228,
-      "loss": 46.0,
-      "step": 365
-    },
-    {
-      "epoch": 0.02798325592063765,
-      "grad_norm": 2.7487831175676547e-05,
-      "learning_rate": 0.00019999999963440124,
-      "loss": 46.0,
-      "step": 366
-    },
-    {
-      "epoch": 0.02805971290402737,
-      "grad_norm": 9.670599683886394e-05,
-      "learning_rate": 0.00019999999963234442,
-      "loss": 46.0,
-      "step": 367
-    },
-    {
-      "epoch": 0.02813616988741709,
-      "grad_norm": 9.182581561617553e-05,
-      "learning_rate": 0.00019999999963028184,
-      "loss": 46.0,
-      "step": 368
-    },
-    {
-      "epoch": 0.028212626870806813,
-      "grad_norm": 3.484496846795082e-05,
-      "learning_rate": 0.0001999999996282135,
-      "loss": 46.0,
-      "step": 369
-    },
-    {
-      "epoch": 0.028289083854196533,
-      "grad_norm": 2.1098958313814364e-05,
-      "learning_rate": 0.00019999999962613937,
-      "loss": 46.0,
-      "step": 370
-    },
-    {
-      "epoch": 0.028365540837586252,
-      "grad_norm": 3.1412764656124637e-05,
-      "learning_rate": 0.0001999999996240595,
-      "loss": 46.0,
-      "step": 371
-    },
-    {
-      "epoch": 0.028441997820975975,
-      "grad_norm": 7.061860378598794e-05,
-      "learning_rate": 0.0001999999996219738,
-      "loss": 46.0,
-      "step": 372
-    },
-    {
-      "epoch": 0.028518454804365694,
-      "grad_norm": 0.00010549635771894827,
-      "learning_rate": 0.00019999999961988241,
-      "loss": 46.0,
-      "step": 373
-    },
-    {
-      "epoch": 0.028594911787755414,
-      "grad_norm": 3.301572360214777e-05,
-      "learning_rate": 0.00019999999961778521,
-      "loss": 46.0,
-      "step": 374
-    },
-    {
-      "epoch": 0.028671368771145133,
-      "grad_norm": 3.148805626551621e-05,
-      "learning_rate": 0.00019999999961568224,
-      "loss": 46.0,
-      "step": 375
-    },
-    {
-      "epoch": 0.028747825754534856,
-      "grad_norm": 0.00010135354386875406,
-      "learning_rate": 0.00019999999961357352,
-      "loss": 46.0,
-      "step": 376
-    },
-    {
-      "epoch": 0.028824282737924575,
-      "grad_norm": 2.105231942550745e-05,
-      "learning_rate": 0.00019999999961145898,
-      "loss": 46.0,
-      "step": 377
-    },
-    {
-      "epoch": 0.028900739721314295,
-      "grad_norm": 3.620113420765847e-05,
-      "learning_rate": 0.00019999999960933874,
-      "loss": 46.0,
-      "step": 378
-    },
-    {
-      "epoch": 0.028977196704704018,
-      "grad_norm": 5.260032776277512e-05,
-      "learning_rate": 0.0001999999996072127,
-      "loss": 46.0,
-      "step": 379
-    },
-    {
-      "epoch": 0.029053653688093737,
-      "grad_norm": 9.694312029751018e-05,
-      "learning_rate": 0.00019999999960508086,
-      "loss": 46.0,
-      "step": 380
-    },
-    {
-      "epoch": 0.029130110671483456,
-      "grad_norm": 6.499604205600917e-05,
-      "learning_rate": 0.0001999999996029433,
-      "loss": 46.0,
-      "step": 381
-    },
-    {
-      "epoch": 0.029206567654873176,
-      "grad_norm": 2.0910307284793817e-05,
-      "learning_rate": 0.00019999999960079994,
-      "loss": 46.0,
-      "step": 382
-    },
-    {
-      "epoch": 0.0292830246382629,
-      "grad_norm": 7.609938620589674e-05,
-      "learning_rate": 0.0001999999995986508,
-      "loss": 46.0,
-      "step": 383
-    },
-    {
-      "epoch": 0.029359481621652618,
-      "grad_norm": 6.608160038013011e-05,
-      "learning_rate": 0.00019999999959649593,
-      "loss": 46.0,
-      "step": 384
-    },
-    {
-      "epoch": 0.029435938605042337,
-      "grad_norm": 4.28716630267445e-05,
-      "learning_rate": 0.00019999999959433525,
-      "loss": 46.0,
-      "step": 385
-    },
-    {
-      "epoch": 0.029512395588432057,
-      "grad_norm": 3.7647692806785926e-05,
-      "learning_rate": 0.00019999999959216882,
-      "loss": 46.0,
-      "step": 386
-    },
-    {
-      "epoch": 0.02958885257182178,
-      "grad_norm": 3.8156184018589556e-05,
-      "learning_rate": 0.00019999999958999665,
-      "loss": 46.0,
-      "step": 387
-    },
-    {
-      "epoch": 0.0296653095552115,
-      "grad_norm": 3.738639134098776e-05,
-      "learning_rate": 0.00019999999958781865,
-      "loss": 46.0,
-      "step": 388
-    },
-    {
-      "epoch": 0.02974176653860122,
-      "grad_norm": 6.814857624704018e-05,
-      "learning_rate": 0.00019999999958563493,
-      "loss": 46.0,
-      "step": 389
-    },
-    {
-      "epoch": 0.02981822352199094,
-      "grad_norm": 3.3049222111003473e-05,
-      "learning_rate": 0.00019999999958344541,
-      "loss": 46.0,
-      "step": 390
-    },
-    {
-      "epoch": 0.02989468050538066,
-      "grad_norm": 6.526375364046544e-05,
-      "learning_rate": 0.00019999999958125012,
-      "loss": 46.0,
-      "step": 391
-    },
-    {
-      "epoch": 0.02997113748877038,
-      "grad_norm": 5.8769906900124624e-05,
-      "learning_rate": 0.00019999999957904908,
-      "loss": 46.0,
-      "step": 392
-    },
-    {
-      "epoch": 0.0300475944721601,
-      "grad_norm": 0.00011813946912297979,
-      "learning_rate": 0.00019999999957684227,
-      "loss": 46.0,
-      "step": 393
-    },
-    {
-      "epoch": 0.030124051455549822,
-      "grad_norm": 5.27090705872979e-05,
-      "learning_rate": 0.00019999999957462969,
-      "loss": 46.0,
-      "step": 394
-    },
-    {
-      "epoch": 0.030200508438939542,
-      "grad_norm": 7.135706255212426e-05,
-      "learning_rate": 0.00019999999957241133,
-      "loss": 46.0,
-      "step": 395
-    },
-    {
-      "epoch": 0.03027696542232926,
-      "grad_norm": 0.00012790884647984058,
-      "learning_rate": 0.0001999999995701872,
-      "loss": 46.0,
-      "step": 396
-    },
-    {
-      "epoch": 0.03035342240571898,
-      "grad_norm": 0.00010968629794660956,
-      "learning_rate": 0.00019999999956795732,
-      "loss": 46.0,
-      "step": 397
-    },
-    {
-      "epoch": 0.030429879389108704,
-      "grad_norm": 7.167622970882803e-05,
-      "learning_rate": 0.00019999999956572166,
-      "loss": 46.0,
-      "step": 398
-    },
-    {
-      "epoch": 0.030506336372498423,
-      "grad_norm": 0.0002279918553540483,
-      "learning_rate": 0.0001999999995634802,
-      "loss": 46.0,
-      "step": 399
-    },
-    {
-      "epoch": 0.030582793355888142,
-      "grad_norm": 4.593819903675467e-05,
-      "learning_rate": 0.00019999999956123304,
-      "loss": 46.0,
-      "step": 400
-    },
-    {
-      "epoch": 0.030659250339277865,
-      "grad_norm": 5.459985914058052e-05,
-      "learning_rate": 0.00019999999955898004,
-      "loss": 46.0,
-      "step": 401
-    },
-    {
-      "epoch": 0.030735707322667585,
-      "grad_norm": 8.810993313090876e-05,
-      "learning_rate": 0.0001999999995567213,
-      "loss": 46.0,
-      "step": 402
-    },
-    {
-      "epoch": 0.030812164306057304,
-      "grad_norm": 4.9673231842461973e-05,
-      "learning_rate": 0.0001999999995544568,
-      "loss": 46.0,
-      "step": 403
-    },
-    {
-      "epoch": 0.030888621289447023,
-      "grad_norm": 0.000129247346194461,
-      "learning_rate": 0.0001999999995521865,
-      "loss": 46.0,
-      "step": 404
-    },
-    {
-      "epoch": 0.030965078272836746,
-      "grad_norm": 5.8459922001929954e-05,
-      "learning_rate": 0.00019999999954991047,
-      "loss": 46.0,
-      "step": 405
-    },
-    {
-      "epoch": 0.031041535256226466,
-      "grad_norm": 5.5456195696024224e-05,
-      "learning_rate": 0.00019999999954762863,
-      "loss": 46.0,
-      "step": 406
-    },
-    {
-      "epoch": 0.031117992239616185,
-      "grad_norm": 9.113064152188599e-05,
-      "learning_rate": 0.00019999999954534105,
-      "loss": 46.0,
-      "step": 407
-    },
-    {
-      "epoch": 0.031194449223005908,
-      "grad_norm": 9.12557152332738e-05,
-      "learning_rate": 0.00019999999954304772,
-      "loss": 46.0,
-      "step": 408
-    },
-    {
-      "epoch": 0.031270906206395624,
-      "grad_norm": 3.194018790964037e-05,
-      "learning_rate": 0.00019999999954074858,
-      "loss": 46.0,
-      "step": 409
-    },
-    {
-      "epoch": 0.03134736318978535,
-      "grad_norm": 5.113500446896069e-05,
-      "learning_rate": 0.00019999999953844365,
-      "loss": 46.0,
-      "step": 410
-    },
-    {
-      "epoch": 0.03142382017317507,
-      "grad_norm": 6.490101804956794e-05,
-      "learning_rate": 0.00019999999953613303,
-      "loss": 46.0,
-      "step": 411
-    },
-    {
-      "epoch": 0.03150027715656479,
-      "grad_norm": 8.563043957110494e-05,
-      "learning_rate": 0.00019999999953381658,
-      "loss": 46.0,
-      "step": 412
-    },
-    {
-      "epoch": 0.03157673413995451,
-      "grad_norm": 6.562795169884339e-05,
-      "learning_rate": 0.00019999999953149438,
-      "loss": 46.0,
-      "step": 413
-    },
-    {
-      "epoch": 0.03165319112334423,
-      "grad_norm": 4.411544796312228e-05,
-      "learning_rate": 0.0001999999995291664,
-      "loss": 46.0,
-      "step": 414
-    },
-    {
-      "epoch": 0.03172964810673395,
-      "grad_norm": 8.868756412994117e-05,
-      "learning_rate": 0.00019999999952683263,
-      "loss": 46.0,
-      "step": 415
-    },
-    {
-      "epoch": 0.03180610509012367,
-      "grad_norm": 0.0001218832258018665,
-      "learning_rate": 0.00019999999952449314,
-      "loss": 46.0,
-      "step": 416
-    },
-    {
-      "epoch": 0.03188256207351339,
-      "grad_norm": 8.545455057173967e-05,
-      "learning_rate": 0.00019999999952214785,
-      "loss": 46.0,
-      "step": 417
-    },
-    {
-      "epoch": 0.03195901905690311,
-      "grad_norm": 5.497948586707935e-05,
-      "learning_rate": 0.0001999999995197968,
-      "loss": 46.0,
-      "step": 418
-    },
-    {
-      "epoch": 0.03203547604029283,
-      "grad_norm": 5.394974505179562e-05,
-      "learning_rate": 0.00019999999951743997,
-      "loss": 46.0,
-      "step": 419
-    },
-    {
-      "epoch": 0.03211193302368255,
-      "grad_norm": 5.120428977534175e-05,
-      "learning_rate": 0.00019999999951507736,
-      "loss": 46.0,
-      "step": 420
-    },
-    {
-      "epoch": 0.03218839000707227,
-      "grad_norm": 4.472916043596342e-05,
-      "learning_rate": 0.00019999999951270903,
-      "loss": 46.0,
-      "step": 421
-    },
-    {
-      "epoch": 0.03226484699046199,
-      "grad_norm": 0.00014817029295954853,
-      "learning_rate": 0.0001999999995103349,
-      "loss": 46.0,
-      "step": 422
-    },
-    {
-      "epoch": 0.03234130397385171,
-      "grad_norm": 7.525101682404056e-05,
-      "learning_rate": 0.000199999999507955,
-      "loss": 46.0,
-      "step": 423
-    },
-    {
-      "epoch": 0.03241776095724143,
-      "grad_norm": 0.0001744249602779746,
-      "learning_rate": 0.00019999999950556934,
-      "loss": 46.0,
-      "step": 424
-    },
-    {
-      "epoch": 0.032494217940631155,
-      "grad_norm": 5.363115997170098e-05,
-      "learning_rate": 0.0001999999995031779,
-      "loss": 46.0,
-      "step": 425
-    },
-    {
-      "epoch": 0.032570674924020875,
-      "grad_norm": 9.20443344512023e-05,
-      "learning_rate": 0.00019999999950078066,
-      "loss": 46.0,
-      "step": 426
-    },
-    {
-      "epoch": 0.032647131907410594,
-      "grad_norm": 6.859291897853836e-05,
-      "learning_rate": 0.00019999999949837772,
-      "loss": 46.0,
-      "step": 427
-    },
-    {
-      "epoch": 0.03272358889080031,
-      "grad_norm": 3.25442997564096e-05,
-      "learning_rate": 0.00019999999949596898,
-      "loss": 46.0,
-      "step": 428
-    },
-    {
-      "epoch": 0.03280004587419003,
-      "grad_norm": 0.00013730718637816608,
-      "learning_rate": 0.00019999999949355443,
-      "loss": 46.0,
-      "step": 429
-    },
-    {
-      "epoch": 0.03287650285757975,
-      "grad_norm": 0.00010657977691153064,
-      "learning_rate": 0.00019999999949113414,
-      "loss": 46.0,
-      "step": 430
-    },
-    {
-      "epoch": 0.03295295984096947,
-      "grad_norm": 8.789327694103122e-05,
-      "learning_rate": 0.0001999999994887081,
-      "loss": 46.0,
-      "step": 431
-    },
-    {
-      "epoch": 0.0330294168243592,
-      "grad_norm": 3.960489993914962e-05,
-      "learning_rate": 0.0001999999994862763,
-      "loss": 46.0,
-      "step": 432
-    },
-    {
-      "epoch": 0.03310587380774892,
-      "grad_norm": 0.00012432123185135424,
-      "learning_rate": 0.00019999999948383868,
-      "loss": 46.0,
-      "step": 433
-    },
-    {
-      "epoch": 0.03318233079113864,
-      "grad_norm": 7.518427446484566e-05,
-      "learning_rate": 0.00019999999948139532,
-      "loss": 46.0,
-      "step": 434
-    },
-    {
-      "epoch": 0.033258787774528356,
-      "grad_norm": 5.9240293921902776e-05,
-      "learning_rate": 0.0001999999994789462,
-      "loss": 46.0,
-      "step": 435
-    },
-    {
-      "epoch": 0.033335244757918076,
-      "grad_norm": 7.50075196265243e-05,
-      "learning_rate": 0.00019999999947649126,
-      "loss": 46.0,
-      "step": 436
-    },
-    {
-      "epoch": 0.033411701741307795,
-      "grad_norm": 9.26538705243729e-05,
-      "learning_rate": 0.0001999999994740306,
-      "loss": 46.0,
-      "step": 437
-    },
-    {
-      "epoch": 0.033488158724697514,
-      "grad_norm": 8.948218601290137e-05,
-      "learning_rate": 0.00019999999947156415,
-      "loss": 46.0,
-      "step": 438
-    },
-    {
-      "epoch": 0.03356461570808724,
-      "grad_norm": 7.06083737895824e-05,
-      "learning_rate": 0.00019999999946909196,
-      "loss": 46.0,
-      "step": 439
-    },
-    {
-      "epoch": 0.03364107269147696,
-      "grad_norm": 6.116050644777715e-05,
-      "learning_rate": 0.00019999999946661396,
-      "loss": 46.0,
-      "step": 440
-    },
-    {
-      "epoch": 0.03371752967486668,
-      "grad_norm": 8.292579150293022e-05,
-      "learning_rate": 0.0001999999994641302,
-      "loss": 46.0,
-      "step": 441
-    },
-    {
-      "epoch": 0.0337939866582564,
-      "grad_norm": 0.0001200176848215051,
-      "learning_rate": 0.0001999999994616407,
-      "loss": 46.0,
-      "step": 442
-    },
-    {
-      "epoch": 0.03387044364164612,
-      "grad_norm": 6.51444643153809e-05,
-      "learning_rate": 0.0001999999994591454,
-      "loss": 46.0,
-      "step": 443
-    },
-    {
-      "epoch": 0.03394690062503584,
-      "grad_norm": 6.607238901779056e-05,
-      "learning_rate": 0.00019999999945664437,
-      "loss": 46.0,
-      "step": 444
-    },
-    {
-      "epoch": 0.03402335760842556,
-      "grad_norm": 7.727661431999877e-05,
-      "learning_rate": 0.00019999999945413755,
-      "loss": 46.0,
-      "step": 445
-    },
-    {
-      "epoch": 0.034099814591815276,
-      "grad_norm": 0.00010184628627030179,
-      "learning_rate": 0.00019999999945162494,
-      "loss": 46.0,
-      "step": 446
-    },
-    {
-      "epoch": 0.034176271575205,
-      "grad_norm": 7.788404764141887e-05,
-      "learning_rate": 0.00019999999944910656,
-      "loss": 46.0,
-      "step": 447
-    },
-    {
-      "epoch": 0.03425272855859472,
-      "grad_norm": 0.0001848977553891018,
-      "learning_rate": 0.00019999999944658243,
-      "loss": 46.0,
-      "step": 448
-    },
-    {
-      "epoch": 0.03432918554198444,
-      "grad_norm": 4.975163392373361e-05,
-      "learning_rate": 0.00019999999944405252,
-      "loss": 46.0,
-      "step": 449
-    },
-    {
-      "epoch": 0.03440564252537416,
-      "grad_norm": 6.76798154017888e-05,
-      "learning_rate": 0.00019999999944151685,
-      "loss": 46.0,
-      "step": 450
-    },
-    {
-      "epoch": 0.03448209950876388,
-      "grad_norm": 6.532653787871823e-05,
-      "learning_rate": 0.0001999999994389754,
-      "loss": 46.0,
-      "step": 451
-    },
-    {
-      "epoch": 0.0345585564921536,
-      "grad_norm": 6.454264075728133e-05,
-      "learning_rate": 0.0001999999994364282,
-      "loss": 46.0,
-      "step": 452
-    },
-    {
-      "epoch": 0.03463501347554332,
-      "grad_norm": 0.00014598850975744426,
-      "learning_rate": 0.00019999999943387523,
-      "loss": 46.0,
-      "step": 453
-    },
-    {
-      "epoch": 0.034711470458933046,
-      "grad_norm": 0.00027584374765865505,
-      "learning_rate": 0.00019999999943131645,
-      "loss": 46.0,
-      "step": 454
-    },
-    {
-      "epoch": 0.034787927442322765,
-      "grad_norm": 9.978777234209701e-05,
-      "learning_rate": 0.00019999999942875194,
-      "loss": 46.0,
-      "step": 455
-    },
-    {
-      "epoch": 0.034864384425712484,
-      "grad_norm": 7.483481749659404e-05,
-      "learning_rate": 0.00019999999942618165,
-      "loss": 46.0,
-      "step": 456
-    },
-    {
-      "epoch": 0.034940841409102204,
-      "grad_norm": 4.1194238292519e-05,
-      "learning_rate": 0.00019999999942360555,
-      "loss": 46.0,
-      "step": 457
-    },
-    {
-      "epoch": 0.03501729839249192,
-      "grad_norm": 0.00022903215722180903,
-      "learning_rate": 0.00019999999942102374,
-      "loss": 46.0,
-      "step": 458
-    },
-    {
-      "epoch": 0.03509375537588164,
-      "grad_norm": 0.0002285886148456484,
-      "learning_rate": 0.00019999999941843613,
-      "loss": 46.0,
-      "step": 459
-    },
-    {
-      "epoch": 0.03517021235927136,
-      "grad_norm": 9.704591502668336e-05,
-      "learning_rate": 0.00019999999941584278,
-      "loss": 46.0,
-      "step": 460
-    },
-    {
-      "epoch": 0.03524666934266109,
-      "grad_norm": 4.935009201290086e-05,
-      "learning_rate": 0.00019999999941324364,
-      "loss": 46.0,
-      "step": 461
-    },
-    {
-      "epoch": 0.03532312632605081,
-      "grad_norm": 6.34451862424612e-05,
-      "learning_rate": 0.00019999999941063874,
-      "loss": 46.0,
-      "step": 462
-    },
-    {
-      "epoch": 0.03539958330944053,
-      "grad_norm": 6.689593283226714e-05,
-      "learning_rate": 0.00019999999940802804,
-      "loss": 46.0,
-      "step": 463
-    },
-    {
-      "epoch": 0.035476040292830247,
-      "grad_norm": 0.00016250558837782592,
-      "learning_rate": 0.00019999999940541158,
-      "loss": 46.0,
-      "step": 464
-    },
-    {
-      "epoch": 0.035552497276219966,
-      "grad_norm": 7.079118222463876e-05,
-      "learning_rate": 0.00019999999940278936,
-      "loss": 46.0,
-      "step": 465
-    },
-    {
-      "epoch": 0.035628954259609685,
-      "grad_norm": 7.57503803470172e-05,
-      "learning_rate": 0.0001999999994001614,
-      "loss": 46.0,
-      "step": 466
-    },
-    {
-      "epoch": 0.035705411242999405,
-      "grad_norm": 9.18152800295502e-05,
-      "learning_rate": 0.00019999999939752765,
-      "loss": 46.0,
-      "step": 467
-    },
-    {
-      "epoch": 0.03578186822638913,
-      "grad_norm": 8.724846702534705e-05,
-      "learning_rate": 0.00019999999939488813,
-      "loss": 46.0,
-      "step": 468
-    },
-    {
-      "epoch": 0.03585832520977885,
-      "grad_norm": 0.00019557027553673834,
-      "learning_rate": 0.00019999999939224284,
-      "loss": 46.0,
-      "step": 469
-    },
-    {
-      "epoch": 0.03593478219316857,
-      "grad_norm": 0.00016852102999109775,
-      "learning_rate": 0.00019999999938959175,
-      "loss": 46.0,
-      "step": 470
-    },
-    {
-      "epoch": 0.03601123917655829,
-      "grad_norm": 9.384356235386804e-05,
-      "learning_rate": 0.0001999999993869349,
-      "loss": 46.0,
-      "step": 471
-    },
-    {
-      "epoch": 0.03608769615994801,
-      "grad_norm": 4.3559190089581534e-05,
-      "learning_rate": 0.00019999999938427233,
-      "loss": 46.0,
-      "step": 472
-    },
-    {
-      "epoch": 0.03616415314333773,
-      "grad_norm": 9.730083547765389e-05,
-      "learning_rate": 0.00019999999938160397,
-      "loss": 46.0,
-      "step": 473
-    },
-    {
-      "epoch": 0.03624061012672745,
-      "grad_norm": 9.731669706525281e-05,
-      "learning_rate": 0.00019999999937892978,
-      "loss": 46.0,
-      "step": 474
-    },
-    {
-      "epoch": 0.03631706711011717,
-      "grad_norm": 8.360798528883606e-05,
-      "learning_rate": 0.00019999999937624988,
-      "loss": 46.0,
-      "step": 475
-    },
-    {
-      "epoch": 0.03639352409350689,
-      "grad_norm": 8.222902397392318e-05,
-      "learning_rate": 0.00019999999937356423,
-      "loss": 46.0,
-      "step": 476
-    },
-    {
-      "epoch": 0.03646998107689661,
-      "grad_norm": 0.00017370429122820497,
-      "learning_rate": 0.00019999999937087278,
-      "loss": 46.0,
-      "step": 477
-    },
-    {
-      "epoch": 0.03654643806028633,
-      "grad_norm": 0.00026396807515993714,
-      "learning_rate": 0.00019999999936817556,
-      "loss": 46.0,
-      "step": 478
-    },
-    {
-      "epoch": 0.03662289504367605,
-      "grad_norm": 7.789378287270665e-05,
-      "learning_rate": 0.00019999999936547256,
-      "loss": 46.0,
-      "step": 479
-    },
-    {
-      "epoch": 0.03669935202706577,
-      "grad_norm": 0.00016577373025938869,
-      "learning_rate": 0.0001999999993627638,
-      "loss": 46.0,
-      "step": 480
-    },
-    {
-      "epoch": 0.03677580901045549,
-      "grad_norm": 0.00011259245366090909,
-      "learning_rate": 0.00019999999936004927,
-      "loss": 46.0,
-      "step": 481
-    },
-    {
-      "epoch": 0.03685226599384521,
-      "grad_norm": 0.0003629502316471189,
-      "learning_rate": 0.00019999999935732898,
-      "loss": 46.0,
-      "step": 482
-    },
-    {
-      "epoch": 0.036928722977234936,
-      "grad_norm": 0.00021626398665830493,
-      "learning_rate": 0.0001999999993546029,
-      "loss": 46.0,
-      "step": 483
-    },
-    {
-      "epoch": 0.037005179960624655,
-      "grad_norm": 0.00010009731340687722,
-      "learning_rate": 0.00019999999935187108,
-      "loss": 46.0,
-      "step": 484
-    },
-    {
-      "epoch": 0.037081636944014375,
-      "grad_norm": 8.321260975208133e-05,
-      "learning_rate": 0.00019999999934913344,
-      "loss": 46.0,
-      "step": 485
-    },
-    {
-      "epoch": 0.037158093927404094,
-      "grad_norm": 7.356579590123147e-05,
-      "learning_rate": 0.00019999999934639009,
-      "loss": 46.0,
-      "step": 486
-    },
-    {
-      "epoch": 0.037234550910793814,
-      "grad_norm": 9.992928971769288e-05,
-      "learning_rate": 0.00019999999934364096,
-      "loss": 46.0,
-      "step": 487
-    },
-    {
-      "epoch": 0.03731100789418353,
-      "grad_norm": 0.00011778913903981447,
-      "learning_rate": 0.00019999999934088602,
-      "loss": 46.0,
-      "step": 488
-    },
-    {
-      "epoch": 0.03738746487757325,
-      "grad_norm": 0.00021629378898069263,
-      "learning_rate": 0.00019999999933812535,
-      "loss": 46.0,
-      "step": 489
-    },
-    {
-      "epoch": 0.03746392186096298,
-      "grad_norm": 0.00011395724868634716,
-      "learning_rate": 0.0001999999993353589,
-      "loss": 46.0,
-      "step": 490
-    },
-    {
-      "epoch": 0.0375403788443527,
-      "grad_norm": 0.0001429607073077932,
-      "learning_rate": 0.00019999999933258667,
-      "loss": 46.0,
-      "step": 491
-    },
-    {
-      "epoch": 0.03761683582774242,
-      "grad_norm": 9.531518298899755e-05,
-      "learning_rate": 0.00019999999932980868,
-      "loss": 46.0,
-      "step": 492
-    },
-    {
-      "epoch": 0.03769329281113214,
-      "grad_norm": 7.5924486736767e-05,
-      "learning_rate": 0.0001999999993270249,
-      "loss": 46.0,
-      "step": 493
-    },
-    {
-      "epoch": 0.037769749794521856,
-      "grad_norm": 0.00014676700811833143,
-      "learning_rate": 0.0001999999993242354,
-      "loss": 46.0,
-      "step": 494
-    },
-    {
-      "epoch": 0.037846206777911576,
-      "grad_norm": 0.00011669647938106209,
-      "learning_rate": 0.0001999999993214401,
-      "loss": 46.0,
-      "step": 495
-    },
-    {
-      "epoch": 0.037922663761301295,
-      "grad_norm": 0.0001441106724087149,
-      "learning_rate": 0.000199999999318639,
-      "loss": 46.0,
-      "step": 496
-    },
-    {
-      "epoch": 0.03799912074469102,
-      "grad_norm": 8.739835175219923e-05,
-      "learning_rate": 0.00019999999931583217,
-      "loss": 46.0,
-      "step": 497
-    },
-    {
-      "epoch": 0.03807557772808074,
-      "grad_norm": 0.00025402920437045395,
-      "learning_rate": 0.00019999999931301956,
-      "loss": 46.0,
-      "step": 498
-    },
-    {
-      "epoch": 0.03815203471147046,
-      "grad_norm": 0.0001071511724148877,
-      "learning_rate": 0.00019999999931020118,
-      "loss": 46.0,
-      "step": 499
-    },
-    {
-      "epoch": 0.03822849169486018,
-      "grad_norm": 0.00023775412410032004,
-      "learning_rate": 0.00019999999930737702,
-      "loss": 46.0,
-      "step": 500
-    },
-    {
-      "epoch": 0.0383049486782499,
-      "grad_norm": 5.465760477818549e-05,
-      "learning_rate": 0.00019999999930454712,
-      "loss": 46.0,
-      "step": 501
-    },
-    {
-      "epoch": 0.03838140566163962,
-      "grad_norm": 0.00017646980995777994,
-      "learning_rate": 0.00019999999930171141,
-      "loss": 46.0,
-      "step": 502
-    },
-    {
-      "epoch": 0.03845786264502934,
-      "grad_norm": 7.21248215995729e-05,
-      "learning_rate": 0.00019999999929886996,
-      "loss": 46.0,
-      "step": 503
-    },
-    {
-      "epoch": 0.03853431962841906,
-      "grad_norm": 0.0001972862664842978,
-      "learning_rate": 0.00019999999929602277,
-      "loss": 46.0,
-      "step": 504
-    },
-    {
-      "epoch": 0.038610776611808784,
-      "grad_norm": 0.0003103954659309238,
-      "learning_rate": 0.00019999999929316977,
-      "loss": 46.0,
-      "step": 505
-    },
-    {
-      "epoch": 0.0386872335951985,
-      "grad_norm": 0.00019607577996794134,
-      "learning_rate": 0.000199999999290311,
-      "loss": 46.0,
-      "step": 506
-    },
-    {
-      "epoch": 0.03876369057858822,
-      "grad_norm": 0.00010769756772788242,
-      "learning_rate": 0.00019999999928744646,
-      "loss": 46.0,
-      "step": 507
-    },
-    {
-      "epoch": 0.03884014756197794,
-      "grad_norm": 0.0003284113190602511,
-      "learning_rate": 0.00019999999928457617,
-      "loss": 46.0,
-      "step": 508
-    },
-    {
-      "epoch": 0.03891660454536766,
-      "grad_norm": 0.0001277989213122055,
-      "learning_rate": 0.00019999999928170008,
-      "loss": 46.0,
-      "step": 509
-    },
-    {
-      "epoch": 0.03899306152875738,
-      "grad_norm": 0.00020971034246031195,
-      "learning_rate": 0.00019999999927881824,
-      "loss": 46.0,
-      "step": 510
-    },
-    {
-      "epoch": 0.0390695185121471,
-      "grad_norm": 6.318332452792674e-05,
-      "learning_rate": 0.00019999999927593063,
-      "loss": 46.0,
-      "step": 511
-    },
-    {
-      "epoch": 0.039145975495536826,
-      "grad_norm": 8.429765148321167e-05,
-      "learning_rate": 0.00019999999927303725,
-      "loss": 46.0,
-      "step": 512
-    },
-    {
-      "epoch": 0.039222432478926546,
-      "grad_norm": 0.0002804486430250108,
-      "learning_rate": 0.0001999999992701381,
-      "loss": 46.0,
-      "step": 513
-    },
-    {
-      "epoch": 0.039298889462316265,
-      "grad_norm": 9.464767936151475e-05,
-      "learning_rate": 0.0001999999992672332,
-      "loss": 46.0,
-      "step": 514
-    },
-    {
-      "epoch": 0.039375346445705985,
-      "grad_norm": 0.0003463878238108009,
-      "learning_rate": 0.00019999999926432249,
-      "loss": 46.0,
-      "step": 515
-    },
-    {
-      "epoch": 0.039451803429095704,
-      "grad_norm": 3.753870623768307e-05,
-      "learning_rate": 0.00019999999926140604,
-      "loss": 46.0,
-      "step": 516
-    },
-    {
-      "epoch": 0.03952826041248542,
-      "grad_norm": 0.00020038125512655824,
-      "learning_rate": 0.00019999999925848379,
-      "loss": 46.0,
-      "step": 517
-    },
-    {
-      "epoch": 0.03960471739587514,
-      "grad_norm": 0.00027526941266842186,
-      "learning_rate": 0.00019999999925555582,
-      "loss": 46.0,
-      "step": 518
-    },
-    {
-      "epoch": 0.03968117437926487,
-      "grad_norm": 0.000571784854400903,
-      "learning_rate": 0.00019999999925262202,
-      "loss": 46.0,
-      "step": 519
-    },
-    {
-      "epoch": 0.03975763136265459,
-      "grad_norm": 0.00020035584748256952,
-      "learning_rate": 0.0001999999992496825,
-      "loss": 46.0,
-      "step": 520
-    },
-    {
-      "epoch": 0.03983408834604431,
-      "grad_norm": 0.00019171522581018507,
-      "learning_rate": 0.0001999999992467372,
-      "loss": 46.0,
-      "step": 521
-    },
-    {
-      "epoch": 0.03991054532943403,
-      "grad_norm": 0.00016133245662786067,
-      "learning_rate": 0.00019999999924378612,
-      "loss": 46.0,
-      "step": 522
-    },
-    {
-      "epoch": 0.03998700231282375,
-      "grad_norm": 0.0001265967875951901,
-      "learning_rate": 0.0001999999992408293,
-      "loss": 46.0,
-      "step": 523
-    },
-    {
-      "epoch": 0.040063459296213466,
-      "grad_norm": 0.00018212999566458166,
-      "learning_rate": 0.00019999999923786665,
-      "loss": 46.0,
-      "step": 524
-    },
-    {
-      "epoch": 0.040139916279603186,
-      "grad_norm": 0.00015810677723493427,
-      "learning_rate": 0.00019999999923489827,
-      "loss": 46.0,
-      "step": 525
-    },
-    {
-      "epoch": 0.04021637326299291,
-      "grad_norm": 0.0003167934773955494,
-      "learning_rate": 0.0001999999992319241,
-      "loss": 46.0,
-      "step": 526
-    },
-    {
-      "epoch": 0.04029283024638263,
-      "grad_norm": 0.0001548193395137787,
-      "learning_rate": 0.00019999999922894418,
-      "loss": 46.0,
-      "step": 527
-    },
-    {
-      "epoch": 0.04036928722977235,
-      "grad_norm": 0.0002268975949846208,
-      "learning_rate": 0.0001999999992259585,
-      "loss": 46.0,
-      "step": 528
-    },
-    {
-      "epoch": 0.04044574421316207,
-      "grad_norm": 0.0004839841276407242,
-      "learning_rate": 0.00019999999922296703,
-      "loss": 46.0,
-      "step": 529
-    },
-    {
-      "epoch": 0.04052220119655179,
-      "grad_norm": 0.00027105200570076704,
-      "learning_rate": 0.00019999999921996978,
-      "loss": 46.0,
-      "step": 530
-    },
-    {
-      "epoch": 0.04059865817994151,
-      "grad_norm": 0.0003971451660618186,
-      "learning_rate": 0.0001999999992169668,
-      "loss": 46.0,
-      "step": 531
-    },
-    {
-      "epoch": 0.04067511516333123,
-      "grad_norm": 0.00019004048954229802,
-      "learning_rate": 0.00019999999921395802,
-      "loss": 46.0,
-      "step": 532
-    },
-    {
-      "epoch": 0.04075157214672095,
-      "grad_norm": 0.00024154713901225477,
-      "learning_rate": 0.0001999999992109435,
-      "loss": 46.0,
-      "step": 533
-    },
-    {
-      "epoch": 0.040828029130110674,
-      "grad_norm": 0.0004077182966284454,
-      "learning_rate": 0.00019999999920792318,
-      "loss": 46.0,
-      "step": 534
-    },
-    {
-      "epoch": 0.04090448611350039,
-      "grad_norm": 0.00010285509779350832,
-      "learning_rate": 0.0001999999992048971,
-      "loss": 46.0,
-      "step": 535
-    },
-    {
-      "epoch": 0.04098094309689011,
-      "grad_norm": 0.00016635113570373505,
-      "learning_rate": 0.00019999999920186526,
-      "loss": 46.0,
-      "step": 536
-    },
-    {
-      "epoch": 0.04105740008027983,
-      "grad_norm": 0.0001531173475086689,
-      "learning_rate": 0.00019999999919882765,
-      "loss": 46.0,
-      "step": 537
-    },
-    {
-      "epoch": 0.04113385706366955,
-      "grad_norm": 0.00016305070312228054,
-      "learning_rate": 0.00019999999919578424,
-      "loss": 46.0,
-      "step": 538
-    },
-    {
-      "epoch": 0.04121031404705927,
-      "grad_norm": 0.0001529300061520189,
-      "learning_rate": 0.00019999999919273508,
-      "loss": 46.0,
-      "step": 539
-    },
-    {
-      "epoch": 0.04128677103044899,
-      "grad_norm": 0.00027484921156428754,
-      "learning_rate": 0.00019999999918968018,
-      "loss": 46.0,
-      "step": 540
-    },
-    {
-      "epoch": 0.04136322801383872,
-      "grad_norm": 0.00030769151635468006,
-      "learning_rate": 0.00019999999918661948,
-      "loss": 46.0,
-      "step": 541
-    },
-    {
-      "epoch": 0.041439684997228436,
-      "grad_norm": 0.0004475169407669455,
-      "learning_rate": 0.000199999999183553,
-      "loss": 46.0,
-      "step": 542
-    },
-    {
-      "epoch": 0.041516141980618156,
-      "grad_norm": 0.0002054843498626724,
-      "learning_rate": 0.00019999999918048078,
-      "loss": 46.0,
-      "step": 543
-    },
-    {
-      "epoch": 0.041592598964007875,
-      "grad_norm": 0.00040626409463584423,
-      "learning_rate": 0.00019999999917740278,
-      "loss": 46.0,
-      "step": 544
-    },
-    {
-      "epoch": 0.041669055947397594,
-      "grad_norm": 0.0003186376125086099,
-      "learning_rate": 0.000199999999174319,
-      "loss": 46.0,
-      "step": 545
-    },
-    {
-      "epoch": 0.041745512930787314,
-      "grad_norm": 0.00023347382375504822,
-      "learning_rate": 0.00019999999917122945,
-      "loss": 46.0,
-      "step": 546
-    },
-    {
-      "epoch": 0.04182196991417703,
-      "grad_norm": 7.283046579686925e-05,
-      "learning_rate": 0.00019999999916813416,
-      "loss": 46.0,
-      "step": 547
-    },
-    {
-      "epoch": 0.04189842689756676,
-      "grad_norm": 0.0001351140672340989,
-      "learning_rate": 0.00019999999916503307,
-      "loss": 46.0,
-      "step": 548
-    },
-    {
-      "epoch": 0.04197488388095648,
-      "grad_norm": 0.00022846198407933116,
-      "learning_rate": 0.00019999999916192623,
-      "loss": 46.0,
-      "step": 549
-    },
-    {
-      "epoch": 0.0420513408643462,
-      "grad_norm": 0.00027839007088914514,
-      "learning_rate": 0.0001999999991588136,
-      "loss": 46.0,
-      "step": 550
-    },
-    {
-      "epoch": 0.04212779784773592,
-      "grad_norm": 0.0003140224434901029,
-      "learning_rate": 0.0001999999991556952,
-      "loss": 46.0,
-      "step": 551
-    },
-    {
-      "epoch": 0.04220425483112564,
-      "grad_norm": 0.0001425761729478836,
-      "learning_rate": 0.00019999999915257103,
-      "loss": 46.0,
-      "step": 552
-    },
-    {
-      "epoch": 0.04228071181451536,
-      "grad_norm": 8.557864202884957e-05,
-      "learning_rate": 0.00019999999914944113,
-      "loss": 46.0,
-      "step": 553
-    },
-    {
-      "epoch": 0.042357168797905076,
-      "grad_norm": 0.00045285187661647797,
-      "learning_rate": 0.00019999999914630543,
-      "loss": 46.0,
-      "step": 554
-    },
-    {
-      "epoch": 0.0424336257812948,
-      "grad_norm": 0.0001562376564834267,
-      "learning_rate": 0.00019999999914316395,
-      "loss": 46.0,
-      "step": 555
-    },
-    {
-      "epoch": 0.04251008276468452,
-      "grad_norm": 0.0002117129333782941,
-      "learning_rate": 0.00019999999914001673,
-      "loss": 46.0,
-      "step": 556
-    },
-    {
-      "epoch": 0.04258653974807424,
-      "grad_norm": 0.00028741630376316607,
-      "learning_rate": 0.00019999999913686368,
-      "loss": 46.0,
-      "step": 557
-    },
-    {
-      "epoch": 0.04266299673146396,
-      "grad_norm": 0.00020813473383896053,
-      "learning_rate": 0.00019999999913370494,
-      "loss": 46.0,
-      "step": 558
-    },
-    {
-      "epoch": 0.04273945371485368,
-      "grad_norm": 0.00019847205840051174,
-      "learning_rate": 0.0001999999991305404,
-      "loss": 46.0,
-      "step": 559
-    },
-    {
-      "epoch": 0.0428159106982434,
-      "grad_norm": 0.0001927387056639418,
-      "learning_rate": 0.00019999999912737005,
-      "loss": 46.0,
-      "step": 560
-    },
-    {
-      "epoch": 0.04289236768163312,
-      "grad_norm": 0.00036941171856597066,
-      "learning_rate": 0.000199999999124194,
-      "loss": 46.0,
-      "step": 561
-    },
-    {
-      "epoch": 0.04296882466502284,
-      "grad_norm": 0.0006141941412352026,
-      "learning_rate": 0.00019999999912101213,
-      "loss": 46.0,
-      "step": 562
-    },
-    {
-      "epoch": 0.043045281648412564,
-      "grad_norm": 0.0004593846097122878,
-      "learning_rate": 0.00019999999911782452,
-      "loss": 46.0,
-      "step": 563
-    },
-    {
-      "epoch": 0.043121738631802284,
-      "grad_norm": 0.0013217341620475054,
-      "learning_rate": 0.0001999999991146311,
-      "loss": 46.0,
-      "step": 564
-    },
-    {
-      "epoch": 0.043198195615192,
-      "grad_norm": 0.00018051821098197252,
-      "learning_rate": 0.00019999999911143195,
-      "loss": 46.0,
-      "step": 565
-    },
-    {
-      "epoch": 0.04327465259858172,
-      "grad_norm": 0.00019405983039177954,
-      "learning_rate": 0.00019999999910822703,
-      "loss": 46.0,
-      "step": 566
-    },
-    {
-      "epoch": 0.04335110958197144,
-      "grad_norm": 0.00010190376633545384,
-      "learning_rate": 0.00019999999910501632,
-      "loss": 46.0,
-      "step": 567
-    },
-    {
-      "epoch": 0.04342756656536116,
-      "grad_norm": 0.00011113641085103154,
-      "learning_rate": 0.00019999999910179985,
-      "loss": 46.0,
-      "step": 568
-    },
-    {
-      "epoch": 0.04350402354875088,
-      "grad_norm": 0.00010373206896474585,
-      "learning_rate": 0.0001999999990985776,
-      "loss": 46.0,
-      "step": 569
-    },
-    {
-      "epoch": 0.04358048053214061,
-      "grad_norm": 0.00019257872190792114,
-      "learning_rate": 0.0001999999990953496,
-      "loss": 46.0,
-      "step": 570
-    },
-    {
-      "epoch": 0.04365693751553033,
-      "grad_norm": 0.0003890043299179524,
-      "learning_rate": 0.0001999999990921158,
-      "loss": 46.0,
-      "step": 571
-    },
-    {
-      "epoch": 0.043733394498920046,
-      "grad_norm": 0.0002035276556853205,
-      "learning_rate": 0.00019999999908887627,
-      "loss": 46.0,
-      "step": 572
-    },
-    {
-      "epoch": 0.043809851482309765,
-      "grad_norm": 0.00011801308573922142,
-      "learning_rate": 0.00019999999908563095,
-      "loss": 46.0,
-      "step": 573
-    },
-    {
-      "epoch": 0.043886308465699485,
-      "grad_norm": 0.00014850600564386696,
-      "learning_rate": 0.00019999999908237986,
-      "loss": 46.0,
-      "step": 574
-    },
-    {
-      "epoch": 0.043962765449089204,
-      "grad_norm": 0.0006481784512288868,
-      "learning_rate": 0.000199999999079123,
-      "loss": 46.0,
-      "step": 575
-    },
-    {
-      "epoch": 0.044039222432478924,
-      "grad_norm": 0.0001285257749259472,
-      "learning_rate": 0.00019999999907586037,
-      "loss": 46.0,
-      "step": 576
-    },
-    {
-      "epoch": 0.04411567941586865,
-      "grad_norm": 0.0001004383884719573,
-      "learning_rate": 0.00019999999907259198,
-      "loss": 46.0,
-      "step": 577
-    },
-    {
-      "epoch": 0.04419213639925837,
-      "grad_norm": 0.00016313082596752793,
-      "learning_rate": 0.0001999999990693178,
-      "loss": 46.0,
-      "step": 578
-    },
-    {
-      "epoch": 0.04426859338264809,
-      "grad_norm": 0.00016484205843880773,
-      "learning_rate": 0.0001999999990660379,
-      "loss": 46.0,
-      "step": 579
-    },
-    {
-      "epoch": 0.04434505036603781,
-      "grad_norm": 0.00018597515008877963,
-      "learning_rate": 0.00019999999906275217,
-      "loss": 46.0,
-      "step": 580
-    },
-    {
-      "epoch": 0.04442150734942753,
-      "grad_norm": 0.00018805245053954422,
-      "learning_rate": 0.00019999999905946072,
-      "loss": 46.0,
-      "step": 581
-    },
-    {
-      "epoch": 0.04449796433281725,
-      "grad_norm": 0.0006113880081102252,
-      "learning_rate": 0.00019999999905616348,
-      "loss": 46.0,
-      "step": 582
-    },
-    {
-      "epoch": 0.044574421316206966,
-      "grad_norm": 0.00032206092146225274,
-      "learning_rate": 0.00019999999905286045,
-      "loss": 46.0,
-      "step": 583
-    },
-    {
-      "epoch": 0.04465087829959669,
-      "grad_norm": 0.00019632400653790683,
-      "learning_rate": 0.00019999999904955166,
-      "loss": 46.0,
-      "step": 584
-    },
-    {
-      "epoch": 0.04472733528298641,
-      "grad_norm": 0.00037773430813103914,
-      "learning_rate": 0.00019999999904623712,
-      "loss": 46.0,
-      "step": 585
-    },
-    {
-      "epoch": 0.04480379226637613,
-      "grad_norm": 0.00031085708178579807,
-      "learning_rate": 0.0001999999990429168,
-      "loss": 46.0,
-      "step": 586
-    },
-    {
-      "epoch": 0.04488024924976585,
-      "grad_norm": 0.00029227888444438577,
-      "learning_rate": 0.0001999999990395907,
-      "loss": 46.0,
-      "step": 587
-    },
-    {
-      "epoch": 0.04495670623315557,
-      "grad_norm": 0.0002536651154514402,
-      "learning_rate": 0.00019999999903625882,
-      "loss": 46.0,
-      "step": 588
-    },
-    {
-      "epoch": 0.04503316321654529,
-      "grad_norm": 0.00011346046085236594,
-      "learning_rate": 0.00019999999903292122,
-      "loss": 46.0,
-      "step": 589
-    },
-    {
-      "epoch": 0.04510962019993501,
-      "grad_norm": 0.000156095513375476,
-      "learning_rate": 0.00019999999902957783,
-      "loss": 46.0,
-      "step": 590
-    },
-    {
-      "epoch": 0.04518607718332473,
-      "grad_norm": 0.0003040801966562867,
-      "learning_rate": 0.00019999999902622865,
-      "loss": 46.0,
-      "step": 591
-    },
-    {
-      "epoch": 0.045262534166714455,
-      "grad_norm": 0.00018165202345699072,
-      "learning_rate": 0.00019999999902287372,
-      "loss": 46.0,
-      "step": 592
-    },
-    {
-      "epoch": 0.045338991150104174,
-      "grad_norm": 0.00017982885765377432,
-      "learning_rate": 0.000199999999019513,
-      "loss": 46.0,
-      "step": 593
-    },
-    {
-      "epoch": 0.045415448133493894,
-      "grad_norm": 0.00012296812201384455,
-      "learning_rate": 0.00019999999901614652,
-      "loss": 46.0,
-      "step": 594
-    },
-    {
-      "epoch": 0.04549190511688361,
-      "grad_norm": 0.00031843906617723405,
-      "learning_rate": 0.00019999999901277427,
-      "loss": 46.0,
-      "step": 595
-    },
-    {
-      "epoch": 0.04556836210027333,
-      "grad_norm": 0.00010394735727459192,
-      "learning_rate": 0.00019999999900939628,
-      "loss": 46.0,
-      "step": 596
-    },
-    {
-      "epoch": 0.04564481908366305,
-      "grad_norm": 0.00028753752121701837,
-      "learning_rate": 0.00019999999900601248,
-      "loss": 46.0,
-      "step": 597
-    },
-    {
-      "epoch": 0.04572127606705277,
-      "grad_norm": 0.0002125209430232644,
-      "learning_rate": 0.00019999999900262294,
-      "loss": 46.0,
-      "step": 598
-    },
-    {
-      "epoch": 0.0457977330504425,
-      "grad_norm": 0.0004885608796030283,
-      "learning_rate": 0.0001999999989992276,
-      "loss": 46.0,
-      "step": 599
-    },
-    {
-      "epoch": 0.04587419003383222,
-      "grad_norm": 0.0002425906277494505,
-      "learning_rate": 0.0001999999989958265,
-      "loss": 46.0,
-      "step": 600
-    },
-    {
-      "epoch": 0.045950647017221936,
-      "grad_norm": 0.0008075024816207588,
-      "learning_rate": 0.00019999999899241962,
-      "loss": 46.0,
-      "step": 601
-    },
-    {
-      "epoch": 0.046027104000611656,
-      "grad_norm": 0.0002986732288263738,
-      "learning_rate": 0.00019999999898900704,
-      "loss": 46.0,
-      "step": 602
-    },
-    {
-      "epoch": 0.046103560984001375,
-      "grad_norm": 0.0004090134461876005,
-      "learning_rate": 0.0001999999989855886,
-      "loss": 46.0,
-      "step": 603
-    },
-    {
-      "epoch": 0.046180017967391095,
-      "grad_norm": 0.0003536487347446382,
-      "learning_rate": 0.00019999999898216445,
-      "loss": 46.0,
-      "step": 604
-    },
-    {
-      "epoch": 0.046256474950780814,
-      "grad_norm": 0.00017800123896449804,
-      "learning_rate": 0.0001999999989787345,
-      "loss": 46.0,
-      "step": 605
-    },
-    {
-      "epoch": 0.04633293193417054,
-      "grad_norm": 0.0003169942938257009,
-      "learning_rate": 0.0001999999989752988,
-      "loss": 46.0,
-      "step": 606
-    },
-    {
-      "epoch": 0.04640938891756026,
-      "grad_norm": 0.0005397346685640514,
-      "learning_rate": 0.00019999999897185732,
-      "loss": 46.0,
-      "step": 607
-    },
-    {
-      "epoch": 0.04648584590094998,
-      "grad_norm": 0.00015565697685815394,
-      "learning_rate": 0.00019999999896841008,
-      "loss": 46.0,
-      "step": 608
-    },
-    {
-      "epoch": 0.0465623028843397,
-      "grad_norm": 0.00035490491427481174,
-      "learning_rate": 0.00019999999896495705,
-      "loss": 46.0,
-      "step": 609
-    },
-    {
-      "epoch": 0.04663875986772942,
-      "grad_norm": 0.00017223434406332672,
-      "learning_rate": 0.00019999999896149826,
-      "loss": 46.0,
-      "step": 610
-    },
-    {
-      "epoch": 0.04671521685111914,
-      "grad_norm": 0.00021444361482281238,
-      "learning_rate": 0.00019999999895803372,
-      "loss": 46.0,
-      "step": 611
-    },
-    {
-      "epoch": 0.04679167383450886,
-      "grad_norm": 0.00018947887292597443,
-      "learning_rate": 0.00019999999895456338,
-      "loss": 46.0,
-      "step": 612
-    },
-    {
-      "epoch": 0.04686813081789858,
-      "grad_norm": 0.0005515294033102691,
-      "learning_rate": 0.00019999999895108726,
-      "loss": 46.0,
-      "step": 613
-    },
-    {
-      "epoch": 0.0469445878012883,
-      "grad_norm": 0.0005505699082277715,
-      "learning_rate": 0.00019999999894760543,
-      "loss": 46.0,
-      "step": 614
-    },
-    {
-      "epoch": 0.04702104478467802,
-      "grad_norm": 0.00024242745712399483,
-      "learning_rate": 0.0001999999989441178,
-      "loss": 46.0,
-      "step": 615
-    },
-    {
-      "epoch": 0.04709750176806774,
-      "grad_norm": 0.0003676850174088031,
-      "learning_rate": 0.00019999999894062439,
-      "loss": 46.0,
-      "step": 616
-    },
-    {
-      "epoch": 0.04717395875145746,
-      "grad_norm": 0.00020813284208998084,
-      "learning_rate": 0.0001999999989371252,
-      "loss": 46.0,
-      "step": 617
-    },
-    {
-      "epoch": 0.04725041573484718,
-      "grad_norm": 0.0001982890535145998,
-      "learning_rate": 0.00019999999893362028,
-      "loss": 46.0,
-      "step": 618
-    },
-    {
-      "epoch": 0.0473268727182369,
-      "grad_norm": 0.0007860666955821216,
-      "learning_rate": 0.00019999999893010955,
-      "loss": 46.0,
-      "step": 619
-    },
-    {
-      "epoch": 0.04740332970162662,
-      "grad_norm": 0.00025393030955456197,
-      "learning_rate": 0.00019999999892659308,
-      "loss": 46.0,
-      "step": 620
-    },
-    {
-      "epoch": 0.047479786685016345,
-      "grad_norm": 0.00033899443224072456,
-      "learning_rate": 0.0001999999989230708,
-      "loss": 46.0,
-      "step": 621
-    },
-    {
-      "epoch": 0.047556243668406065,
-      "grad_norm": 0.00021985615603625774,
-      "learning_rate": 0.0001999999989195428,
-      "loss": 46.0,
-      "step": 622
-    },
-    {
-      "epoch": 0.047632700651795784,
-      "grad_norm": 0.00035814798320643604,
-      "learning_rate": 0.000199999998916009,
-      "loss": 46.0,
-      "step": 623
-    },
-    {
-      "epoch": 0.0477091576351855,
-      "grad_norm": 0.0002714464208111167,
-      "learning_rate": 0.00019999999891246944,
-      "loss": 46.0,
-      "step": 624
-    },
-    {
-      "epoch": 0.04778561461857522,
-      "grad_norm": 0.0002546482137404382,
-      "learning_rate": 0.0001999999989089241,
-      "loss": 46.0,
-      "step": 625
-    },
-    {
-      "epoch": 0.04786207160196494,
-      "grad_norm": 0.00033310535945929587,
-      "learning_rate": 0.000199999998905373,
-      "loss": 46.0,
-      "step": 626
-    },
-    {
-      "epoch": 0.04793852858535466,
-      "grad_norm": 0.0009725523414090276,
-      "learning_rate": 0.00019999999890181616,
-      "loss": 46.0,
-      "step": 627
-    },
-    {
-      "epoch": 0.04801498556874439,
-      "grad_norm": 0.00013420562027022243,
-      "learning_rate": 0.0001999999988982535,
-      "loss": 46.0,
-      "step": 628
-    },
-    {
-      "epoch": 0.04809144255213411,
-      "grad_norm": 0.0005744160735048354,
-      "learning_rate": 0.00019999999889468511,
-      "loss": 46.0,
-      "step": 629
-    },
-    {
-      "epoch": 0.04816789953552383,
-      "grad_norm": 0.00042676873272284865,
-      "learning_rate": 0.0001999999988911109,
-      "loss": 46.0,
-      "step": 630
-    },
-    {
-      "epoch": 0.048244356518913546,
-      "grad_norm": 0.0004222260322421789,
-      "learning_rate": 0.00019999999888753095,
-      "loss": 46.0,
-      "step": 631
-    },
-    {
-      "epoch": 0.048320813502303266,
-      "grad_norm": 0.00023509195307269692,
-      "learning_rate": 0.00019999999888394525,
-      "loss": 46.0,
-      "step": 632
-    },
-    {
-      "epoch": 0.048397270485692985,
-      "grad_norm": 0.0004956149496138096,
-      "learning_rate": 0.00019999999888035378,
-      "loss": 46.0,
-      "step": 633
-    },
-    {
-      "epoch": 0.048473727469082704,
-      "grad_norm": 0.000332842260831967,
-      "learning_rate": 0.0001999999988767565,
-      "loss": 46.0,
-      "step": 634
-    },
-    {
-      "epoch": 0.04855018445247243,
-      "grad_norm": 0.00015729085134807974,
-      "learning_rate": 0.00019999999887315348,
-      "loss": 46.0,
-      "step": 635
-    },
-    {
-      "epoch": 0.04862664143586215,
-      "grad_norm": 0.00039549326174892485,
-      "learning_rate": 0.00019999999886954472,
-      "loss": 46.0,
-      "step": 636
-    },
-    {
-      "epoch": 0.04870309841925187,
-      "grad_norm": 0.0005816948250867426,
-      "learning_rate": 0.00019999999886593015,
-      "loss": 46.0,
-      "step": 637
-    },
-    {
-      "epoch": 0.04877955540264159,
-      "grad_norm": 0.0006450708606280386,
-      "learning_rate": 0.0001999999988623098,
-      "loss": 46.0,
-      "step": 638
-    },
-    {
-      "epoch": 0.04885601238603131,
-      "grad_norm": 0.0005489220493473113,
-      "learning_rate": 0.0001999999988586837,
-      "loss": 46.0,
-      "step": 639
-    },
-    {
-      "epoch": 0.04893246936942103,
-      "grad_norm": 0.0005445321439765394,
-      "learning_rate": 0.00019999999885505183,
-      "loss": 46.0,
-      "step": 640
-    },
-    {
-      "epoch": 0.04900892635281075,
-      "grad_norm": 0.00037244780105538666,
-      "learning_rate": 0.0001999999988514142,
-      "loss": 46.0,
-      "step": 641
-    },
-    {
-      "epoch": 0.049085383336200474,
-      "grad_norm": 0.0006239353679120541,
-      "learning_rate": 0.0001999999988477708,
-      "loss": 46.0,
-      "step": 642
-    },
-    {
-      "epoch": 0.04916184031959019,
-      "grad_norm": 0.0004768453654833138,
-      "learning_rate": 0.00019999999884412158,
-      "loss": 46.0,
-      "step": 643
-    },
-    {
-      "epoch": 0.04923829730297991,
-      "grad_norm": 0.00021584770001936704,
-      "learning_rate": 0.00019999999884046666,
-      "loss": 46.0,
-      "step": 644
-    },
-    {
-      "epoch": 0.04931475428636963,
-      "grad_norm": 0.0004969947622157633,
-      "learning_rate": 0.00019999999883680593,
-      "loss": 46.0,
-      "step": 645
-    },
-    {
-      "epoch": 0.04939121126975935,
-      "grad_norm": 0.0006911872769705951,
-      "learning_rate": 0.00019999999883313946,
-      "loss": 46.0,
-      "step": 646
-    },
-    {
-      "epoch": 0.04946766825314907,
-      "grad_norm": 0.0009078672155737877,
-      "learning_rate": 0.00019999999882946718,
-      "loss": 46.0,
-      "step": 647
-    },
-    {
-      "epoch": 0.04954412523653879,
-      "grad_norm": 0.0010897921165451407,
-      "learning_rate": 0.00019999999882578913,
-      "loss": 46.0,
-      "step": 648
-    },
-    {
-      "epoch": 0.04962058221992851,
-      "grad_norm": 0.0005299266777001321,
-      "learning_rate": 0.00019999999882210537,
-      "loss": 46.0,
-      "step": 649
-    },
-    {
-      "epoch": 0.049697039203318236,
-      "grad_norm": 0.00026388245169073343,
-      "learning_rate": 0.0001999999988184158,
-      "loss": 46.0,
-      "step": 650
-    },
-    {
-      "epoch": 0.049773496186707955,
-      "grad_norm": 0.0002868471492547542,
-      "learning_rate": 0.00019999999881472046,
-      "loss": 46.0,
-      "step": 651
-    },
-    {
-      "epoch": 0.049849953170097674,
-      "grad_norm": 0.00046446488704532385,
-      "learning_rate": 0.00019999999881101937,
-      "loss": 46.0,
-      "step": 652
-    },
-    {
-      "epoch": 0.049926410153487394,
-      "grad_norm": 0.0015076535055413842,
-      "learning_rate": 0.00019999999880731249,
-      "loss": 46.0,
-      "step": 653
-    },
-    {
-      "epoch": 0.05000286713687711,
-      "grad_norm": 0.0005419623921625316,
-      "learning_rate": 0.00019999999880359983,
-      "loss": 46.0,
-      "step": 654
-    },
-    {
-      "epoch": 0.05007932412026683,
-      "grad_norm": 0.000653263065032661,
-      "learning_rate": 0.00019999999879988142,
-      "loss": 46.0,
-      "step": 655
-    },
-    {
-      "epoch": 0.05015578110365655,
-      "grad_norm": 0.00019291364878881723,
-      "learning_rate": 0.00019999999879615724,
-      "loss": 46.0,
-      "step": 656
-    },
-    {
-      "epoch": 0.05023223808704628,
-      "grad_norm": 0.0004329807707108557,
-      "learning_rate": 0.0001999999987924273,
-      "loss": 46.0,
-      "step": 657
-    },
-    {
-      "epoch": 0.050308695070436,
-      "grad_norm": 0.00031730011687614024,
-      "learning_rate": 0.00019999999878869159,
-      "loss": 46.0,
-      "step": 658
-    },
-    {
-      "epoch": 0.05038515205382572,
-      "grad_norm": 0.0005122473812662065,
-      "learning_rate": 0.00019999999878495008,
-      "loss": 46.0,
-      "step": 659
-    },
-    {
-      "epoch": 0.05046160903721544,
-      "grad_norm": 0.0005939305992797017,
-      "learning_rate": 0.0001999999987812028,
-      "loss": 46.0,
-      "step": 660
-    },
-    {
-      "epoch": 0.050538066020605156,
-      "grad_norm": 0.000692489673383534,
-      "learning_rate": 0.0001999999987774498,
-      "loss": 46.0,
-      "step": 661
-    },
-    {
-      "epoch": 0.050614523003994875,
-      "grad_norm": 0.00040229037404060364,
-      "learning_rate": 0.000199999998773691,
-      "loss": 46.0,
-      "step": 662
-    },
-    {
-      "epoch": 0.050690979987384595,
-      "grad_norm": 0.0014655481791123748,
-      "learning_rate": 0.0001999999987699264,
-      "loss": 46.0,
-      "step": 663
-    },
-    {
-      "epoch": 0.05076743697077432,
-      "grad_norm": 0.00015594468277413398,
-      "learning_rate": 0.0001999999987661561,
-      "loss": 46.0,
-      "step": 664
-    },
-    {
-      "epoch": 0.05084389395416404,
-      "grad_norm": 0.0004371341201476753,
-      "learning_rate": 0.00019999999876237998,
-      "loss": 46.0,
-      "step": 665
-    },
-    {
-      "epoch": 0.05092035093755376,
-      "grad_norm": 0.00044017413165420294,
-      "learning_rate": 0.0001999999987585981,
-      "loss": 46.0,
-      "step": 666
-    },
-    {
-      "epoch": 0.05099680792094348,
-      "grad_norm": 0.0006631204159930348,
-      "learning_rate": 0.00019999999875481046,
-      "loss": 46.0,
-      "step": 667
-    },
-    {
-      "epoch": 0.0510732649043332,
-      "grad_norm": 0.0005109034827910364,
-      "learning_rate": 0.00019999999875101705,
-      "loss": 46.0,
-      "step": 668
-    },
-    {
-      "epoch": 0.05114972188772292,
-      "grad_norm": 0.0002045494766207412,
-      "learning_rate": 0.00019999999874721787,
-      "loss": 46.0,
-      "step": 669
-    },
-    {
-      "epoch": 0.05122617887111264,
-      "grad_norm": 0.00024516473058611155,
-      "learning_rate": 0.00019999999874341292,
-      "loss": 46.0,
-      "step": 670
-    },
-    {
-      "epoch": 0.051302635854502364,
-      "grad_norm": 0.0008533840300515294,
-      "learning_rate": 0.0001999999987396022,
-      "loss": 46.0,
-      "step": 671
-    },
-    {
-      "epoch": 0.05137909283789208,
-      "grad_norm": 0.000275343976682052,
-      "learning_rate": 0.00019999999873578566,
-      "loss": 46.0,
-      "step": 672
-    },
-    {
-      "epoch": 0.0514555498212818,
-      "grad_norm": 0.001113425474613905,
-      "learning_rate": 0.0001999999987319634,
-      "loss": 46.0,
-      "step": 673
-    },
-    {
-      "epoch": 0.05153200680467152,
-      "grad_norm": 0.0005910403560847044,
-      "learning_rate": 0.0001999999987281354,
-      "loss": 46.0,
-      "step": 674
-    },
-    {
-      "epoch": 0.05160846378806124,
-      "grad_norm": 0.000977934803813696,
-      "learning_rate": 0.00019999999872430158,
-      "loss": 46.0,
-      "step": 675
-    },
-    {
-      "epoch": 0.05168492077145096,
-      "grad_norm": 0.0004100938094779849,
-      "learning_rate": 0.000199999998720462,
-      "loss": 46.0,
-      "step": 676
-    },
-    {
-      "epoch": 0.05176137775484068,
-      "grad_norm": 0.0031233434565365314,
-      "learning_rate": 0.00019999999871661664,
-      "loss": 46.0,
-      "step": 677
-    },
-    {
-      "epoch": 0.0518378347382304,
-      "grad_norm": 0.0007034446462057531,
-      "learning_rate": 0.00019999999871276556,
-      "loss": 46.0,
-      "step": 678
-    },
-    {
-      "epoch": 0.051914291721620126,
-      "grad_norm": 0.0006807809113524854,
-      "learning_rate": 0.0001999999987089087,
-      "loss": 46.0,
-      "step": 679
-    },
-    {
-      "epoch": 0.051990748705009845,
-      "grad_norm": 0.00047518391511403024,
-      "learning_rate": 0.000199999998705046,
-      "loss": 46.0,
-      "step": 680
-    },
-    {
-      "epoch": 0.052067205688399565,
-      "grad_norm": 0.0010379269951954484,
-      "learning_rate": 0.0001999999987011776,
-      "loss": 46.0,
-      "step": 681
-    },
-    {
-      "epoch": 0.052143662671789284,
-      "grad_norm": 0.0004005995288025588,
-      "learning_rate": 0.00019999999869730343,
-      "loss": 46.0,
-      "step": 682
-    },
-    {
-      "epoch": 0.052220119655179004,
-      "grad_norm": 0.0038646352477371693,
-      "learning_rate": 0.00019999999869342347,
-      "loss": 46.0,
-      "step": 683
-    },
-    {
-      "epoch": 0.05229657663856872,
-      "grad_norm": 0.0006725925486534834,
-      "learning_rate": 0.00019999999868953775,
-      "loss": 46.0,
-      "step": 684
-    },
-    {
-      "epoch": 0.05237303362195844,
-      "grad_norm": 0.0007942290976643562,
-      "learning_rate": 0.00019999999868564622,
-      "loss": 46.0,
-      "step": 685
-    },
-    {
-      "epoch": 0.05244949060534817,
-      "grad_norm": 0.0013885163934901357,
-      "learning_rate": 0.00019999999868174897,
-      "loss": 46.0,
-      "step": 686
-    },
-    {
-      "epoch": 0.05252594758873789,
-      "grad_norm": 0.0005410252488218248,
-      "learning_rate": 0.00019999999867784593,
-      "loss": 46.0,
-      "step": 687
-    },
-    {
-      "epoch": 0.05260240457212761,
-      "grad_norm": 0.0006587334210053086,
-      "learning_rate": 0.00019999999867393713,
-      "loss": 46.0,
-      "step": 688
-    },
-    {
-      "epoch": 0.05267886155551733,
-      "grad_norm": 0.0006875649560242891,
-      "learning_rate": 0.00019999999867002254,
-      "loss": 46.0,
-      "step": 689
-    },
-    {
-      "epoch": 0.052755318538907046,
-      "grad_norm": 0.0004901383072137833,
-      "learning_rate": 0.00019999999866610223,
-      "loss": 46.0,
-      "step": 690
-    },
-    {
-      "epoch": 0.052831775522296766,
-      "grad_norm": 0.0004494905879255384,
-      "learning_rate": 0.0001999999986621761,
-      "loss": 46.0,
-      "step": 691
-    },
-    {
-      "epoch": 0.052908232505686485,
-      "grad_norm": 0.001209255657158792,
-      "learning_rate": 0.0001999999986582442,
-      "loss": 46.0,
-      "step": 692
-    },
-    {
-      "epoch": 0.05298468948907621,
-      "grad_norm": 0.0002568756463006139,
-      "learning_rate": 0.00019999999865430657,
-      "loss": 46.0,
-      "step": 693
-    },
-    {
-      "epoch": 0.05306114647246593,
-      "grad_norm": 0.0006126717198640108,
-      "learning_rate": 0.00019999999865036314,
-      "loss": 46.0,
-      "step": 694
-    },
-    {
-      "epoch": 0.05313760345585565,
-      "grad_norm": 0.0024391694460064173,
-      "learning_rate": 0.00019999999864641396,
-      "loss": 46.0,
-      "step": 695
-    },
-    {
-      "epoch": 0.05321406043924537,
-      "grad_norm": 0.0005382664385251701,
-      "learning_rate": 0.000199999998642459,
-      "loss": 46.0,
-      "step": 696
-    },
-    {
-      "epoch": 0.05329051742263509,
-      "grad_norm": 0.0008495989604853094,
-      "learning_rate": 0.00019999999863849825,
-      "loss": 46.0,
-      "step": 697
-    },
-    {
-      "epoch": 0.05336697440602481,
-      "grad_norm": 0.0006599112530238926,
-      "learning_rate": 0.00019999999863453175,
-      "loss": 46.0,
-      "step": 698
-    },
-    {
-      "epoch": 0.05344343138941453,
-      "grad_norm": 0.0015425707679241896,
-      "learning_rate": 0.0001999999986305595,
-      "loss": 46.0,
-      "step": 699
-    },
-    {
-      "epoch": 0.053519888372804254,
-      "grad_norm": 0.0006688204011879861,
-      "learning_rate": 0.00019999999862658146,
-      "loss": 46.0,
-      "step": 700
-    },
-    {
-      "epoch": 0.053596345356193974,
-      "grad_norm": 0.0012332834303379059,
-      "learning_rate": 0.00019999999862259764,
-      "loss": 46.0,
-      "step": 701
-    },
-    {
-      "epoch": 0.05367280233958369,
-      "grad_norm": 0.00041103101102635264,
-      "learning_rate": 0.00019999999861860807,
-      "loss": 46.0,
-      "step": 702
-    },
-    {
-      "epoch": 0.05374925932297341,
-      "grad_norm": 0.0012126514920964837,
-      "learning_rate": 0.0001999999986146127,
-      "loss": 46.0,
-      "step": 703
-    },
-    {
-      "epoch": 0.05382571630636313,
-      "grad_norm": 0.0007691223290748894,
-      "learning_rate": 0.0001999999986106116,
-      "loss": 46.0,
-      "step": 704
-    },
-    {
-      "epoch": 0.05390217328975285,
-      "grad_norm": 0.0015347336884588003,
-      "learning_rate": 0.0001999999986066047,
-      "loss": 46.0,
-      "step": 705
-    },
-    {
-      "epoch": 0.05397863027314257,
-      "grad_norm": 0.00041275881812907755,
-      "learning_rate": 0.00019999999860259205,
-      "loss": 46.0,
-      "step": 706
-    },
-    {
-      "epoch": 0.05405508725653229,
-      "grad_norm": 0.0009639740455895662,
-      "learning_rate": 0.00019999999859857364,
-      "loss": 46.0,
-      "step": 707
-    },
-    {
-      "epoch": 0.054131544239922016,
-      "grad_norm": 0.0007914204034022987,
-      "learning_rate": 0.00019999999859454947,
-      "loss": 46.0,
-      "step": 708
-    },
-    {
-      "epoch": 0.054208001223311736,
-      "grad_norm": 0.0005744309164583683,
-      "learning_rate": 0.00019999999859051949,
-      "loss": 46.0,
-      "step": 709
-    },
-    {
-      "epoch": 0.054284458206701455,
-      "grad_norm": 0.0006608443800359964,
-      "learning_rate": 0.00019999999858648373,
-      "loss": 46.0,
-      "step": 710
-    },
-    {
-      "epoch": 0.054360915190091175,
-      "grad_norm": 0.00045458797831088305,
-      "learning_rate": 0.00019999999858244226,
-      "loss": 46.0,
-      "step": 711
-    },
-    {
-      "epoch": 0.054437372173480894,
-      "grad_norm": 0.00039117992855608463,
-      "learning_rate": 0.00019999999857839496,
-      "loss": 46.0,
-      "step": 712
-    },
-    {
-      "epoch": 0.054513829156870613,
-      "grad_norm": 0.0006807833560742438,
-      "learning_rate": 0.00019999999857434194,
-      "loss": 46.0,
-      "step": 713
-    },
-    {
-      "epoch": 0.05459028614026033,
-      "grad_norm": 0.0005720146582461894,
-      "learning_rate": 0.00019999999857028312,
-      "loss": 46.0,
-      "step": 714
-    },
-    {
-      "epoch": 0.05466674312365006,
-      "grad_norm": 0.00039807340363040566,
-      "learning_rate": 0.00019999999856621853,
-      "loss": 46.0,
-      "step": 715
-    },
-    {
-      "epoch": 0.05474320010703978,
-      "grad_norm": 0.00148311210796237,
-      "learning_rate": 0.00019999999856214816,
-      "loss": 46.0,
-      "step": 716
-    },
-    {
-      "epoch": 0.0548196570904295,
-      "grad_norm": 0.0012500175507739186,
-      "learning_rate": 0.00019999999855807205,
-      "loss": 46.0,
-      "step": 717
-    },
-    {
-      "epoch": 0.05489611407381922,
-      "grad_norm": 0.0014031262835487723,
-      "learning_rate": 0.00019999999855399017,
-      "loss": 46.0,
-      "step": 718
-    },
-    {
-      "epoch": 0.05497257105720894,
-      "grad_norm": 0.0023746713995933533,
-      "learning_rate": 0.00019999999854990253,
-      "loss": 46.0,
-      "step": 719
-    },
-    {
-      "epoch": 0.055049028040598656,
-      "grad_norm": 0.0006557343294844031,
-      "learning_rate": 0.00019999999854580908,
-      "loss": 46.0,
-      "step": 720
-    },
-    {
-      "epoch": 0.055125485023988376,
-      "grad_norm": 0.0008830199367366731,
-      "learning_rate": 0.0001999999985417099,
-      "loss": 46.0,
-      "step": 721
-    },
-    {
-      "epoch": 0.0552019420073781,
-      "grad_norm": 0.0013945880346000195,
-      "learning_rate": 0.00019999999853760492,
-      "loss": 46.0,
-      "step": 722
-    },
-    {
-      "epoch": 0.05527839899076782,
-      "grad_norm": 0.0006048845243640244,
-      "learning_rate": 0.0001999999985334942,
-      "loss": 46.0,
-      "step": 723
-    },
-    {
-      "epoch": 0.05535485597415754,
-      "grad_norm": 0.00040394082316197455,
-      "learning_rate": 0.00019999999852937767,
-      "loss": 46.0,
-      "step": 724
-    },
-    {
-      "epoch": 0.05543131295754726,
-      "grad_norm": 0.001725990092381835,
-      "learning_rate": 0.00019999999852525542,
-      "loss": 46.0,
-      "step": 725
-    },
-    {
-      "epoch": 0.05550776994093698,
-      "grad_norm": 0.00040107901440933347,
-      "learning_rate": 0.00019999999852112738,
-      "loss": 46.0,
-      "step": 726
-    },
-    {
-      "epoch": 0.0555842269243267,
-      "grad_norm": 0.0016303808661177754,
-      "learning_rate": 0.00019999999851699353,
-      "loss": 46.0,
-      "step": 727
-    },
-    {
-      "epoch": 0.05566068390771642,
-      "grad_norm": 0.001971200807020068,
-      "learning_rate": 0.00019999999851285396,
-      "loss": 46.0,
-      "step": 728
-    },
-    {
-      "epoch": 0.055737140891106145,
-      "grad_norm": 0.0008331132121384144,
-      "learning_rate": 0.0001999999985087086,
-      "loss": 46.0,
-      "step": 729
-    },
-    {
-      "epoch": 0.055813597874495864,
-      "grad_norm": 0.0012251917505636811,
-      "learning_rate": 0.0001999999985045575,
-      "loss": 46.0,
-      "step": 730
-    },
-    {
-      "epoch": 0.055890054857885584,
-      "grad_norm": 0.0004360883904155344,
-      "learning_rate": 0.00019999999850040057,
-      "loss": 46.0,
-      "step": 731
-    },
-    {
-      "epoch": 0.0559665118412753,
-      "grad_norm": 0.00035662445588968694,
-      "learning_rate": 0.00019999999849623794,
-      "loss": 46.0,
-      "step": 732
-    },
-    {
-      "epoch": 0.05604296882466502,
-      "grad_norm": 0.0008678619633428752,
-      "learning_rate": 0.0001999999984920695,
-      "loss": 46.0,
-      "step": 733
-    },
-    {
-      "epoch": 0.05611942580805474,
-      "grad_norm": 0.0005197175778448582,
-      "learning_rate": 0.0001999999984878953,
-      "loss": 46.0,
-      "step": 734
-    },
-    {
-      "epoch": 0.05619588279144446,
-      "grad_norm": 0.003531946800649166,
-      "learning_rate": 0.00019999999848371535,
-      "loss": 46.0,
-      "step": 735
-    },
-    {
-      "epoch": 0.05627233977483418,
-      "grad_norm": 0.0005907154991291463,
-      "learning_rate": 0.0001999999984795296,
-      "loss": 46.0,
-      "step": 736
-    },
-    {
-      "epoch": 0.05634879675822391,
-      "grad_norm": 0.0016546265687793493,
-      "learning_rate": 0.0001999999984753381,
-      "loss": 46.0,
-      "step": 737
-    },
-    {
-      "epoch": 0.056425253741613626,
-      "grad_norm": 0.0007478833431378007,
-      "learning_rate": 0.0001999999984711408,
-      "loss": 46.0,
-      "step": 738
-    },
-    {
-      "epoch": 0.056501710725003346,
-      "grad_norm": 0.0009737526997923851,
-      "learning_rate": 0.00019999999846693776,
-      "loss": 46.0,
-      "step": 739
-    },
-    {
-      "epoch": 0.056578167708393065,
-      "grad_norm": 0.0009089783998206258,
-      "learning_rate": 0.00019999999846272894,
-      "loss": 46.0,
-      "step": 740
-    },
-    {
-      "epoch": 0.056654624691782784,
-      "grad_norm": 0.0008082463755272329,
-      "learning_rate": 0.00019999999845851435,
-      "loss": 46.0,
-      "step": 741
-    },
-    {
-      "epoch": 0.056731081675172504,
-      "grad_norm": 0.00042460591066628695,
-      "learning_rate": 0.000199999998454294,
-      "loss": 46.0,
-      "step": 742
-    },
-    {
-      "epoch": 0.05680753865856222,
-      "grad_norm": 0.0007730139186605811,
-      "learning_rate": 0.00019999999845006787,
-      "loss": 46.0,
-      "step": 743
-    },
-    {
-      "epoch": 0.05688399564195195,
-      "grad_norm": 0.000482582749100402,
-      "learning_rate": 0.00019999999844583599,
-      "loss": 46.0,
-      "step": 744
-    },
-    {
-      "epoch": 0.05696045262534167,
-      "grad_norm": 0.00027285594842396677,
-      "learning_rate": 0.00019999999844159833,
-      "loss": 46.0,
-      "step": 745
-    },
-    {
-      "epoch": 0.05703690960873139,
-      "grad_norm": 0.00045095005771145225,
-      "learning_rate": 0.0001999999984373549,
-      "loss": 46.0,
-      "step": 746
-    },
-    {
-      "epoch": 0.05711336659212111,
-      "grad_norm": 0.00031077791936695576,
-      "learning_rate": 0.0001999999984331057,
-      "loss": 46.0,
-      "step": 747
-    },
-    {
-      "epoch": 0.05718982357551083,
-      "grad_norm": 0.000746348116081208,
-      "learning_rate": 0.0001999999984288507,
-      "loss": 46.0,
-      "step": 748
-    },
-    {
-      "epoch": 0.05726628055890055,
-      "grad_norm": 0.0015635429881513119,
-      "learning_rate": 0.00019999999842459,
-      "loss": 46.0,
-      "step": 749
-    },
-    {
-      "epoch": 0.057342737542290266,
-      "grad_norm": 0.0019601222593337297,
-      "learning_rate": 0.00019999999842032346,
-      "loss": 46.0,
-      "step": 750
-    },
-    {
-      "epoch": 0.05741919452567999,
-      "grad_norm": 0.001063704607076943,
-      "learning_rate": 0.00019999999841605116,
-      "loss": 46.0,
-      "step": 751
-    },
-    {
-      "epoch": 0.05749565150906971,
-      "grad_norm": 0.00038144036079756916,
-      "learning_rate": 0.00019999999841177312,
-      "loss": 46.0,
-      "step": 752
-    },
-    {
-      "epoch": 0.05757210849245943,
-      "grad_norm": 0.0005947853205725551,
-      "learning_rate": 0.0001999999984074893,
-      "loss": 46.0,
-      "step": 753
-    },
-    {
-      "epoch": 0.05764856547584915,
-      "grad_norm": 0.0008315677405335009,
-      "learning_rate": 0.0001999999984031997,
-      "loss": 46.0,
-      "step": 754
-    },
-    {
-      "epoch": 0.05772502245923887,
-      "grad_norm": 0.0005147226038388908,
-      "learning_rate": 0.00019999999839890434,
-      "loss": 46.0,
-      "step": 755
-    },
-    {
-      "epoch": 0.05780147944262859,
-      "grad_norm": 0.00045732385478913784,
-      "learning_rate": 0.00019999999839460323,
-      "loss": 46.0,
-      "step": 756
-    },
-    {
-      "epoch": 0.05787793642601831,
-      "grad_norm": 0.0015032520750537515,
-      "learning_rate": 0.00019999999839029635,
-      "loss": 46.0,
-      "step": 757
-    },
-    {
-      "epoch": 0.057954393409408035,
-      "grad_norm": 0.0011783275986090302,
-      "learning_rate": 0.00019999999838598366,
-      "loss": 46.0,
-      "step": 758
-    },
-    {
-      "epoch": 0.058030850392797755,
-      "grad_norm": 0.0007772682001814246,
-      "learning_rate": 0.00019999999838166523,
-      "loss": 46.0,
-      "step": 759
-    },
-    {
-      "epoch": 0.058107307376187474,
-      "grad_norm": 0.004088142886757851,
-      "learning_rate": 0.00019999999837734103,
-      "loss": 46.0,
-      "step": 760
-    },
-    {
-      "epoch": 0.05818376435957719,
-      "grad_norm": 0.0005342710064724088,
-      "learning_rate": 0.00019999999837301107,
-      "loss": 46.0,
-      "step": 761
-    },
-    {
-      "epoch": 0.05826022134296691,
-      "grad_norm": 0.0007272792863659561,
-      "learning_rate": 0.00019999999836867532,
-      "loss": 46.0,
-      "step": 762
-    },
-    {
-      "epoch": 0.05833667832635663,
-      "grad_norm": 0.0035962089896202087,
-      "learning_rate": 0.0001999999983643338,
-      "loss": 46.0,
-      "step": 763
-    },
-    {
-      "epoch": 0.05841313530974635,
-      "grad_norm": 0.0009250107686966658,
-      "learning_rate": 0.0001999999983599865,
-      "loss": 46.0,
-      "step": 764
-    },
-    {
-      "epoch": 0.05848959229313607,
-      "grad_norm": 0.0007521944935433567,
-      "learning_rate": 0.00019999999835563345,
-      "loss": 46.0,
-      "step": 765
-    },
-    {
-      "epoch": 0.0585660492765258,
-      "grad_norm": 0.0010559684596955776,
-      "learning_rate": 0.00019999999835127464,
-      "loss": 46.0,
-      "step": 766
-    },
-    {
-      "epoch": 0.05864250625991552,
-      "grad_norm": 0.0008711419068276882,
-      "learning_rate": 0.00019999999834691005,
-      "loss": 46.0,
-      "step": 767
-    },
-    {
-      "epoch": 0.058718963243305236,
-      "grad_norm": 0.000753355969209224,
-      "learning_rate": 0.00019999999834253968,
-      "loss": 46.0,
-      "step": 768
-    },
-    {
-      "epoch": 0.058795420226694955,
-      "grad_norm": 0.0006714974297210574,
-      "learning_rate": 0.00019999999833816357,
-      "loss": 46.0,
-      "step": 769
-    },
-    {
-      "epoch": 0.058871877210084675,
-      "grad_norm": 0.000910244882106781,
-      "learning_rate": 0.00019999999833378166,
-      "loss": 46.0,
-      "step": 770
-    },
-    {
-      "epoch": 0.058948334193474394,
-      "grad_norm": 0.0008632204262539744,
-      "learning_rate": 0.00019999999832939397,
-      "loss": 46.0,
-      "step": 771
-    },
-    {
-      "epoch": 0.059024791176864114,
-      "grad_norm": 0.0012319379020482302,
-      "learning_rate": 0.00019999999832500054,
-      "loss": 46.0,
-      "step": 772
-    },
-    {
-      "epoch": 0.05910124816025384,
-      "grad_norm": 0.0009262815001420677,
-      "learning_rate": 0.00019999999832060134,
-      "loss": 46.0,
-      "step": 773
-    },
-    {
-      "epoch": 0.05917770514364356,
-      "grad_norm": 0.0003459417202975601,
-      "learning_rate": 0.00019999999831619636,
-      "loss": 46.0,
-      "step": 774
-    },
-    {
-      "epoch": 0.05925416212703328,
-      "grad_norm": 0.0009886884363368154,
-      "learning_rate": 0.0001999999983117856,
-      "loss": 46.0,
-      "step": 775
-    },
-    {
-      "epoch": 0.059330619110423,
-      "grad_norm": 0.0003174004959873855,
-      "learning_rate": 0.00019999999830736908,
-      "loss": 46.0,
-      "step": 776
-    },
-    {
-      "epoch": 0.05940707609381272,
-      "grad_norm": 0.0006498537841252983,
-      "learning_rate": 0.0001999999983029468,
-      "loss": 46.0,
-      "step": 777
-    },
-    {
-      "epoch": 0.05948353307720244,
-      "grad_norm": 0.0006045684567652643,
-      "learning_rate": 0.00019999999829851874,
-      "loss": 46.0,
-      "step": 778
-    },
-    {
-      "epoch": 0.059559990060592156,
-      "grad_norm": 0.0008451056783087552,
-      "learning_rate": 0.00019999999829408492,
-      "loss": 46.0,
-      "step": 779
-    },
-    {
-      "epoch": 0.05963644704398188,
-      "grad_norm": 0.0006718471995554864,
-      "learning_rate": 0.00019999999828964533,
-      "loss": 46.0,
-      "step": 780
-    },
-    {
-      "epoch": 0.0597129040273716,
-      "grad_norm": 0.0004686721949838102,
-      "learning_rate": 0.00019999999828519997,
-      "loss": 46.0,
-      "step": 781
-    },
-    {
-      "epoch": 0.05978936101076132,
-      "grad_norm": 0.0010273698717355728,
-      "learning_rate": 0.00019999999828074883,
-      "loss": 46.0,
-      "step": 782
-    },
-    {
-      "epoch": 0.05986581799415104,
-      "grad_norm": 0.0007598138763569295,
-      "learning_rate": 0.00019999999827629192,
-      "loss": 46.0,
-      "step": 783
-    },
-    {
-      "epoch": 0.05994227497754076,
-      "grad_norm": 0.0017071543261408806,
-      "learning_rate": 0.00019999999827182923,
-      "loss": 46.0,
-      "step": 784
-    },
-    {
-      "epoch": 0.06001873196093048,
-      "grad_norm": 0.0026008524000644684,
-      "learning_rate": 0.00019999999826736083,
-      "loss": 46.0,
-      "step": 785
-    },
-    {
-      "epoch": 0.0600951889443202,
-      "grad_norm": 0.0023794500157237053,
-      "learning_rate": 0.0001999999982628866,
-      "loss": 46.0,
-      "step": 786
-    },
-    {
-      "epoch": 0.060171645927709926,
-      "grad_norm": 0.0008658910519443452,
-      "learning_rate": 0.0001999999982584066,
-      "loss": 46.0,
-      "step": 787
-    },
-    {
-      "epoch": 0.060248102911099645,
-      "grad_norm": 0.000469990394776687,
-      "learning_rate": 0.00019999999825392087,
-      "loss": 46.0,
-      "step": 788
-    },
-    {
-      "epoch": 0.060324559894489364,
-      "grad_norm": 0.0009368501487188041,
-      "learning_rate": 0.00019999999824942935,
-      "loss": 46.0,
-      "step": 789
-    },
-    {
-      "epoch": 0.060401016877879084,
-      "grad_norm": 0.0008547449833713472,
-      "learning_rate": 0.00019999999824493205,
-      "loss": 46.0,
-      "step": 790
-    },
-    {
-      "epoch": 0.0604774738612688,
-      "grad_norm": 0.0006957253790460527,
-      "learning_rate": 0.00019999999824042898,
-      "loss": 46.0,
-      "step": 791
-    },
-    {
-      "epoch": 0.06055393084465852,
-      "grad_norm": 0.0008682442712597549,
-      "learning_rate": 0.00019999999823592016,
-      "loss": 46.0,
-      "step": 792
-    },
-    {
-      "epoch": 0.06063038782804824,
-      "grad_norm": 0.0010879782494157553,
-      "learning_rate": 0.00019999999823140557,
-      "loss": 46.0,
-      "step": 793
-    },
-    {
-      "epoch": 0.06070684481143796,
-      "grad_norm": 0.0005859620287083089,
-      "learning_rate": 0.0001999999982268852,
-      "loss": 46.0,
-      "step": 794
-    },
-    {
-      "epoch": 0.06078330179482769,
-      "grad_norm": 0.0009435936226509511,
-      "learning_rate": 0.00019999999822235907,
-      "loss": 46.0,
-      "step": 795
-    },
-    {
-      "epoch": 0.06085975877821741,
-      "grad_norm": 0.0006902844179421663,
-      "learning_rate": 0.00019999999821782716,
-      "loss": 46.0,
-      "step": 796
-    },
-    {
-      "epoch": 0.060936215761607126,
-      "grad_norm": 0.0004523476236499846,
-      "learning_rate": 0.0001999999982132895,
-      "loss": 46.0,
-      "step": 797
-    },
-    {
-      "epoch": 0.061012672744996846,
-      "grad_norm": 0.0014437553472816944,
-      "learning_rate": 0.00019999999820874604,
-      "loss": 46.0,
-      "step": 798
-    },
-    {
-      "epoch": 0.061089129728386565,
-      "grad_norm": 0.000748743477743119,
-      "learning_rate": 0.0001999999982041968,
-      "loss": 46.0,
-      "step": 799
-    },
-    {
-      "epoch": 0.061165586711776285,
-      "grad_norm": 0.001465115463361144,
-      "learning_rate": 0.00019999999819964183,
-      "loss": 46.0,
-      "step": 800
-    },
-    {
-      "epoch": 0.061242043695166004,
-      "grad_norm": 0.00201000296510756,
-      "learning_rate": 0.00019999999819508108,
-      "loss": 46.0,
-      "step": 801
-    },
-    {
-      "epoch": 0.06131850067855573,
-      "grad_norm": 0.0008797416812740266,
-      "learning_rate": 0.00019999999819051456,
-      "loss": 46.0,
-      "step": 802
-    },
-    {
-      "epoch": 0.06139495766194545,
-      "grad_norm": 0.0008056176011450589,
-      "learning_rate": 0.00019999999818594226,
-      "loss": 46.0,
-      "step": 803
-    },
-    {
-      "epoch": 0.06147141464533517,
-      "grad_norm": 0.0008374870521947742,
-      "learning_rate": 0.00019999999818136422,
-      "loss": 46.0,
-      "step": 804
-    },
-    {
-      "epoch": 0.06154787162872489,
-      "grad_norm": 0.0007310323417186737,
-      "learning_rate": 0.0001999999981767804,
-      "loss": 46.0,
-      "step": 805
-    },
-    {
-      "epoch": 0.06162432861211461,
-      "grad_norm": 0.0005377045599743724,
-      "learning_rate": 0.00019999999817219078,
-      "loss": 46.0,
-      "step": 806
-    },
-    {
-      "epoch": 0.06170078559550433,
-      "grad_norm": 0.0023042112588882446,
-      "learning_rate": 0.00019999999816759542,
-      "loss": 46.0,
-      "step": 807
-    },
-    {
-      "epoch": 0.06177724257889405,
-      "grad_norm": 0.0018868353217840195,
-      "learning_rate": 0.00019999999816299426,
-      "loss": 46.0,
-      "step": 808
-    },
-    {
-      "epoch": 0.06185369956228377,
-      "grad_norm": 0.0009269246947951615,
-      "learning_rate": 0.00019999999815838737,
-      "loss": 46.0,
-      "step": 809
-    },
-    {
-      "epoch": 0.06193015654567349,
-      "grad_norm": 0.0008533610962331295,
-      "learning_rate": 0.0001999999981537747,
-      "loss": 46.0,
-      "step": 810
-    },
-    {
-      "epoch": 0.06200661352906321,
-      "grad_norm": 0.0012293921317905188,
-      "learning_rate": 0.00019999999814915623,
-      "loss": 46.0,
-      "step": 811
-    },
-    {
-      "epoch": 0.06208307051245293,
-      "grad_norm": 0.0009388715261593461,
-      "learning_rate": 0.000199999998144532,
-      "loss": 46.0,
-      "step": 812
-    },
-    {
-      "epoch": 0.06215952749584265,
-      "grad_norm": 0.0009311331086792052,
-      "learning_rate": 0.00019999999813990202,
-      "loss": 46.0,
-      "step": 813
-    },
-    {
-      "epoch": 0.06223598447923237,
-      "grad_norm": 0.0012165793450549245,
-      "learning_rate": 0.00019999999813526627,
-      "loss": 46.0,
-      "step": 814
-    },
-    {
-      "epoch": 0.06231244146262209,
-      "grad_norm": 0.0009007969638332725,
-      "learning_rate": 0.00019999999813062472,
-      "loss": 46.0,
-      "step": 815
-    },
-    {
-      "epoch": 0.062388898446011816,
-      "grad_norm": 0.0006988821551203728,
-      "learning_rate": 0.00019999999812597745,
-      "loss": 46.0,
-      "step": 816
-    },
-    {
-      "epoch": 0.062465355429401535,
-      "grad_norm": 0.001399722881615162,
-      "learning_rate": 0.00019999999812132438,
-      "loss": 46.0,
-      "step": 817
-    },
-    {
-      "epoch": 0.06254181241279125,
-      "grad_norm": 0.0003073534753639251,
-      "learning_rate": 0.00019999999811666554,
-      "loss": 46.0,
-      "step": 818
-    },
-    {
-      "epoch": 0.06261826939618097,
-      "grad_norm": 0.00043782329885289073,
-      "learning_rate": 0.00019999999811200098,
-      "loss": 46.0,
-      "step": 819
-    },
-    {
-      "epoch": 0.0626947263795707,
-      "grad_norm": 0.0018148773815482855,
-      "learning_rate": 0.00019999999810733056,
-      "loss": 46.0,
-      "step": 820
-    },
-    {
-      "epoch": 0.06277118336296042,
-      "grad_norm": 0.0006111579714342952,
-      "learning_rate": 0.00019999999810265445,
-      "loss": 46.0,
-      "step": 821
-    },
-    {
-      "epoch": 0.06284764034635014,
-      "grad_norm": 0.0015046574408188462,
-      "learning_rate": 0.0001999999980979725,
-      "loss": 46.0,
-      "step": 822
-    },
-    {
-      "epoch": 0.06292409732973986,
-      "grad_norm": 0.000985735678113997,
-      "learning_rate": 0.00019999999809328485,
-      "loss": 46.0,
-      "step": 823
-    },
-    {
-      "epoch": 0.06300055431312958,
-      "grad_norm": 0.00037988051190041006,
-      "learning_rate": 0.00019999999808859137,
-      "loss": 46.0,
-      "step": 824
-    },
-    {
-      "epoch": 0.0630770112965193,
-      "grad_norm": 0.0026610256172716618,
-      "learning_rate": 0.00019999999808389217,
-      "loss": 46.0,
-      "step": 825
-    },
-    {
-      "epoch": 0.06315346827990902,
-      "grad_norm": 0.0013666359009221196,
-      "learning_rate": 0.00019999999807918716,
-      "loss": 46.0,
-      "step": 826
-    },
-    {
-      "epoch": 0.06322992526329874,
-      "grad_norm": 0.0012934188125655055,
-      "learning_rate": 0.0001999999980744764,
-      "loss": 46.0,
-      "step": 827
-    },
-    {
-      "epoch": 0.06330638224668846,
-      "grad_norm": 0.001372310915030539,
-      "learning_rate": 0.00019999999806975992,
-      "loss": 46.0,
-      "step": 828
-    },
-    {
-      "epoch": 0.06338283923007818,
-      "grad_norm": 0.0019590971060097218,
-      "learning_rate": 0.0001999999980650376,
-      "loss": 46.0,
-      "step": 829
-    },
-    {
-      "epoch": 0.0634592962134679,
-      "grad_norm": 0.0009822536958381534,
-      "learning_rate": 0.00019999999806030952,
-      "loss": 46.0,
-      "step": 830
-    },
-    {
-      "epoch": 0.06353575319685761,
-      "grad_norm": 0.0006497660069726408,
-      "learning_rate": 0.0001999999980555757,
-      "loss": 46.0,
-      "step": 831
-    },
-    {
-      "epoch": 0.06361221018024733,
-      "grad_norm": 0.0016641374677419662,
-      "learning_rate": 0.00019999999805083606,
-      "loss": 46.0,
-      "step": 832
-    },
-    {
-      "epoch": 0.06368866716363705,
-      "grad_norm": 0.00133910879958421,
-      "learning_rate": 0.0001999999980460907,
-      "loss": 46.0,
-      "step": 833
-    },
-    {
-      "epoch": 0.06376512414702679,
-      "grad_norm": 0.0017276826547458768,
-      "learning_rate": 0.00019999999804133956,
-      "loss": 46.0,
-      "step": 834
-    },
-    {
-      "epoch": 0.0638415811304165,
-      "grad_norm": 0.0004873050202149898,
-      "learning_rate": 0.00019999999803658265,
-      "loss": 46.0,
-      "step": 835
-    },
-    {
-      "epoch": 0.06391803811380622,
-      "grad_norm": 0.0010182983241975307,
-      "learning_rate": 0.00019999999803181994,
-      "loss": 46.0,
-      "step": 836
-    },
-    {
-      "epoch": 0.06399449509719594,
-      "grad_norm": 0.0010844813659787178,
-      "learning_rate": 0.00019999999802705149,
-      "loss": 46.0,
-      "step": 837
-    },
-    {
-      "epoch": 0.06407095208058566,
-      "grad_norm": 0.0010255974484607577,
-      "learning_rate": 0.00019999999802227726,
-      "loss": 46.0,
-      "step": 838
-    },
-    {
-      "epoch": 0.06414740906397538,
-      "grad_norm": 0.001574929105117917,
-      "learning_rate": 0.00019999999801749728,
-      "loss": 46.0,
-      "step": 839
-    },
-    {
-      "epoch": 0.0642238660473651,
-      "grad_norm": 0.0017449459992349148,
-      "learning_rate": 0.00019999999801271153,
-      "loss": 46.0,
-      "step": 840
-    },
-    {
-      "epoch": 0.06430032303075482,
-      "grad_norm": 0.001526626176200807,
-      "learning_rate": 0.00019999999800792,
-      "loss": 46.0,
-      "step": 841
-    },
-    {
-      "epoch": 0.06437678001414454,
-      "grad_norm": 0.001373098697513342,
-      "learning_rate": 0.0001999999980031227,
-      "loss": 46.0,
-      "step": 842
-    },
-    {
-      "epoch": 0.06445323699753426,
-      "grad_norm": 0.0008916949154809117,
-      "learning_rate": 0.00019999999799831964,
-      "loss": 46.0,
-      "step": 843
-    },
-    {
-      "epoch": 0.06452969398092398,
-      "grad_norm": 0.0014329302357509732,
-      "learning_rate": 0.0001999999979935108,
-      "loss": 46.0,
-      "step": 844
-    },
-    {
-      "epoch": 0.0646061509643137,
-      "grad_norm": 0.00162373052444309,
-      "learning_rate": 0.00019999999798869618,
-      "loss": 46.0,
-      "step": 845
-    },
-    {
-      "epoch": 0.06468260794770342,
-      "grad_norm": 0.0010249036131426692,
-      "learning_rate": 0.00019999999798387582,
-      "loss": 46.0,
-      "step": 846
-    },
-    {
-      "epoch": 0.06475906493109314,
-      "grad_norm": 0.0014899412635713816,
-      "learning_rate": 0.00019999999797904965,
-      "loss": 46.0,
-      "step": 847
-    },
-    {
-      "epoch": 0.06483552191448286,
-      "grad_norm": 0.0010711546055972576,
-      "learning_rate": 0.00019999999797421774,
-      "loss": 46.0,
-      "step": 848
-    },
-    {
-      "epoch": 0.06491197889787259,
-      "grad_norm": 0.0004599509993568063,
-      "learning_rate": 0.00019999999796938006,
-      "loss": 46.0,
-      "step": 849
-    },
-    {
-      "epoch": 0.06498843588126231,
-      "grad_norm": 0.002075366908684373,
-      "learning_rate": 0.00019999999796453658,
-      "loss": 46.0,
-      "step": 850
-    },
-    {
-      "epoch": 0.06506489286465203,
-      "grad_norm": 0.0010055272141471505,
-      "learning_rate": 0.00019999999795968735,
-      "loss": 46.0,
-      "step": 851
-    },
-    {
-      "epoch": 0.06514134984804175,
-      "grad_norm": 0.0009296773350797594,
-      "learning_rate": 0.00019999999795483237,
-      "loss": 46.0,
-      "step": 852
-    },
-    {
-      "epoch": 0.06521780683143147,
-      "grad_norm": 0.0006457075360231102,
-      "learning_rate": 0.0001999999979499716,
-      "loss": 46.0,
-      "step": 853
-    },
-    {
-      "epoch": 0.06529426381482119,
-      "grad_norm": 0.0009924647165462375,
-      "learning_rate": 0.00019999999794510507,
-      "loss": 46.0,
-      "step": 854
-    },
-    {
-      "epoch": 0.06537072079821091,
-      "grad_norm": 0.0007508395938202739,
-      "learning_rate": 0.00019999999794023278,
-      "loss": 46.0,
-      "step": 855
-    },
-    {
-      "epoch": 0.06544717778160063,
-      "grad_norm": 0.0025950735434889793,
-      "learning_rate": 0.00019999999793535468,
-      "loss": 46.0,
-      "step": 856
-    },
-    {
-      "epoch": 0.06552363476499035,
-      "grad_norm": 0.0009169146069325507,
-      "learning_rate": 0.00019999999793047084,
-      "loss": 46.0,
-      "step": 857
-    },
-    {
-      "epoch": 0.06560009174838007,
-      "grad_norm": 0.0007553419563919306,
-      "learning_rate": 0.00019999999792558125,
-      "loss": 46.0,
-      "step": 858
-    },
-    {
-      "epoch": 0.06567654873176978,
-      "grad_norm": 0.0003439235151745379,
-      "learning_rate": 0.00019999999792068586,
-      "loss": 46.0,
-      "step": 859
-    },
-    {
-      "epoch": 0.0657530057151595,
-      "grad_norm": 0.0004695653624366969,
-      "learning_rate": 0.0001999999979157847,
-      "loss": 46.0,
-      "step": 860
-    },
-    {
-      "epoch": 0.06582946269854922,
-      "grad_norm": 0.0007054957677610219,
-      "learning_rate": 0.00019999999791087779,
-      "loss": 46.0,
-      "step": 861
-    },
-    {
-      "epoch": 0.06590591968193894,
-      "grad_norm": 0.0005009636515751481,
-      "learning_rate": 0.00019999999790596508,
-      "loss": 46.0,
-      "step": 862
-    },
-    {
-      "epoch": 0.06598237666532866,
-      "grad_norm": 0.0007160931127145886,
-      "learning_rate": 0.00019999999790104665,
-      "loss": 46.0,
-      "step": 863
-    },
-    {
-      "epoch": 0.0660588336487184,
-      "grad_norm": 0.0018645310774445534,
-      "learning_rate": 0.0001999999978961224,
-      "loss": 46.0,
-      "step": 864
-    },
-    {
-      "epoch": 0.06613529063210812,
-      "grad_norm": 0.0009724583942443132,
-      "learning_rate": 0.0001999999978911924,
-      "loss": 46.0,
-      "step": 865
-    },
-    {
-      "epoch": 0.06621174761549783,
-      "grad_norm": 0.001802563201636076,
-      "learning_rate": 0.00019999999788625664,
-      "loss": 46.0,
-      "step": 866
-    },
-    {
-      "epoch": 0.06628820459888755,
-      "grad_norm": 0.0010456809541210532,
-      "learning_rate": 0.0001999999978813151,
-      "loss": 46.0,
-      "step": 867
-    },
-    {
-      "epoch": 0.06636466158227727,
-      "grad_norm": 0.0007007730309851468,
-      "learning_rate": 0.0001999999978763678,
-      "loss": 46.0,
-      "step": 868
-    },
-    {
-      "epoch": 0.06644111856566699,
-      "grad_norm": 0.0006793321226723492,
-      "learning_rate": 0.0001999999978714147,
-      "loss": 46.0,
-      "step": 869
-    },
-    {
-      "epoch": 0.06651757554905671,
-      "grad_norm": 0.0010249931365251541,
-      "learning_rate": 0.00019999999786645588,
-      "loss": 46.0,
-      "step": 870
-    },
-    {
-      "epoch": 0.06659403253244643,
-      "grad_norm": 0.0005554365343414247,
-      "learning_rate": 0.00019999999786149127,
-      "loss": 46.0,
-      "step": 871
-    },
-    {
-      "epoch": 0.06667048951583615,
-      "grad_norm": 0.0017067046137526631,
-      "learning_rate": 0.00019999999785652088,
-      "loss": 46.0,
-      "step": 872
-    },
-    {
-      "epoch": 0.06674694649922587,
-      "grad_norm": 0.0014227716019377112,
-      "learning_rate": 0.00019999999785154474,
-      "loss": 46.0,
-      "step": 873
-    },
-    {
-      "epoch": 0.06682340348261559,
-      "grad_norm": 0.0004888643161393702,
-      "learning_rate": 0.0001999999978465628,
-      "loss": 46.0,
-      "step": 874
-    },
-    {
-      "epoch": 0.06689986046600531,
-      "grad_norm": 0.0007449073018506169,
-      "learning_rate": 0.0001999999978415751,
-      "loss": 46.0,
-      "step": 875
-    },
-    {
-      "epoch": 0.06697631744939503,
-      "grad_norm": 0.0009143061470240355,
-      "learning_rate": 0.00019999999783658164,
-      "loss": 46.0,
-      "step": 876
-    },
-    {
-      "epoch": 0.06705277443278475,
-      "grad_norm": 0.003081568516790867,
-      "learning_rate": 0.00019999999783158238,
-      "loss": 46.0,
-      "step": 877
-    },
-    {
-      "epoch": 0.06712923141617448,
-      "grad_norm": 0.0012243265518918633,
-      "learning_rate": 0.0001999999978265774,
-      "loss": 46.0,
-      "step": 878
-    },
-    {
-      "epoch": 0.0672056883995642,
-      "grad_norm": 0.0009646289981901646,
-      "learning_rate": 0.00019999999782156663,
-      "loss": 46.0,
-      "step": 879
-    },
-    {
-      "epoch": 0.06728214538295392,
-      "grad_norm": 0.0009335740469396114,
-      "learning_rate": 0.0001999999978165501,
-      "loss": 46.0,
-      "step": 880
-    },
-    {
-      "epoch": 0.06735860236634364,
-      "grad_norm": 0.0005281473277136683,
-      "learning_rate": 0.0001999999978115278,
-      "loss": 46.0,
-      "step": 881
-    },
-    {
-      "epoch": 0.06743505934973336,
-      "grad_norm": 0.0003626249963417649,
-      "learning_rate": 0.0001999999978064997,
-      "loss": 46.0,
-      "step": 882
-    },
-    {
-      "epoch": 0.06751151633312308,
-      "grad_norm": 0.0009607441024854779,
-      "learning_rate": 0.00019999999780146585,
-      "loss": 46.0,
-      "step": 883
-    },
-    {
-      "epoch": 0.0675879733165128,
-      "grad_norm": 0.0008950056508183479,
-      "learning_rate": 0.00019999999779642623,
-      "loss": 46.0,
-      "step": 884
-    },
-    {
-      "epoch": 0.06766443029990252,
-      "grad_norm": 0.0006860795547254384,
-      "learning_rate": 0.00019999999779138084,
-      "loss": 46.0,
-      "step": 885
-    },
-    {
-      "epoch": 0.06774088728329224,
-      "grad_norm": 0.0009584408835507929,
-      "learning_rate": 0.0001999999977863297,
-      "loss": 46.0,
-      "step": 886
-    },
-    {
-      "epoch": 0.06781734426668196,
-      "grad_norm": 0.0011941768461838365,
-      "learning_rate": 0.00019999999778127277,
-      "loss": 46.0,
-      "step": 887
-    },
-    {
-      "epoch": 0.06789380125007168,
-      "grad_norm": 0.0008309573750011623,
-      "learning_rate": 0.00019999999777621006,
-      "loss": 46.0,
-      "step": 888
-    },
-    {
-      "epoch": 0.0679702582334614,
-      "grad_norm": 0.0009365335572510958,
-      "learning_rate": 0.0001999999977711416,
-      "loss": 46.0,
-      "step": 889
-    },
-    {
-      "epoch": 0.06804671521685111,
-      "grad_norm": 0.0010838804300874472,
-      "learning_rate": 0.00019999999776606738,
-      "loss": 46.0,
-      "step": 890
-    },
-    {
-      "epoch": 0.06812317220024083,
-      "grad_norm": 0.0005349599523469806,
-      "learning_rate": 0.00019999999776098735,
-      "loss": 46.0,
-      "step": 891
-    },
-    {
-      "epoch": 0.06819962918363055,
-      "grad_norm": 0.0038097223732620478,
-      "learning_rate": 0.0001999999977559016,
-      "loss": 46.0,
-      "step": 892
-    },
-    {
-      "epoch": 0.06827608616702029,
-      "grad_norm": 0.0010471109999343753,
-      "learning_rate": 0.00019999999775081005,
-      "loss": 46.0,
-      "step": 893
-    },
-    {
-      "epoch": 0.06835254315041,
-      "grad_norm": 0.0009728758595883846,
-      "learning_rate": 0.00019999999774571273,
-      "loss": 46.0,
-      "step": 894
-    },
-    {
-      "epoch": 0.06842900013379973,
-      "grad_norm": 0.000665025319904089,
-      "learning_rate": 0.00019999999774060964,
-      "loss": 46.0,
-      "step": 895
-    },
-    {
-      "epoch": 0.06850545711718944,
-      "grad_norm": 0.0013135699555277824,
-      "learning_rate": 0.0001999999977355008,
-      "loss": 46.0,
-      "step": 896
-    },
-    {
-      "epoch": 0.06858191410057916,
-      "grad_norm": 0.0027354967314749956,
-      "learning_rate": 0.00019999999773038618,
-      "loss": 46.0,
-      "step": 897
-    },
-    {
-      "epoch": 0.06865837108396888,
-      "grad_norm": 0.0038396650925278664,
-      "learning_rate": 0.0001999999977252658,
-      "loss": 46.0,
-      "step": 898
-    },
-    {
-      "epoch": 0.0687348280673586,
-      "grad_norm": 0.001014421577565372,
-      "learning_rate": 0.00019999999772013963,
-      "loss": 46.0,
-      "step": 899
-    },
-    {
-      "epoch": 0.06881128505074832,
-      "grad_norm": 0.0014678931329399347,
-      "learning_rate": 0.0001999999977150077,
-      "loss": 46.0,
-      "step": 900
-    },
-    {
-      "epoch": 0.06888774203413804,
-      "grad_norm": 0.0017121133860200644,
-      "learning_rate": 0.00019999999770987,
-      "loss": 46.0,
-      "step": 901
-    },
-    {
-      "epoch": 0.06896419901752776,
-      "grad_norm": 0.0008576174732297659,
-      "learning_rate": 0.00019999999770472653,
-      "loss": 46.0,
-      "step": 902
-    },
-    {
-      "epoch": 0.06904065600091748,
-      "grad_norm": 0.003078488865867257,
-      "learning_rate": 0.0001999999976995773,
-      "loss": 46.0,
-      "step": 903
-    },
-    {
-      "epoch": 0.0691171129843072,
-      "grad_norm": 0.0006379986298270524,
-      "learning_rate": 0.00019999999769442227,
-      "loss": 46.0,
-      "step": 904
-    },
-    {
-      "epoch": 0.06919356996769692,
-      "grad_norm": 0.00044800524483434856,
-      "learning_rate": 0.00019999999768926152,
-      "loss": 46.0,
-      "step": 905
-    },
-    {
-      "epoch": 0.06927002695108664,
-      "grad_norm": 0.0008939265972003341,
-      "learning_rate": 0.00019999999768409497,
-      "loss": 46.0,
-      "step": 906
-    },
-    {
-      "epoch": 0.06934648393447637,
-      "grad_norm": 0.0008266211370937526,
-      "learning_rate": 0.00019999999767892262,
-      "loss": 46.0,
-      "step": 907
-    },
-    {
-      "epoch": 0.06942294091786609,
-      "grad_norm": 0.0026271925307810307,
-      "learning_rate": 0.00019999999767374456,
-      "loss": 46.0,
-      "step": 908
-    },
-    {
-      "epoch": 0.06949939790125581,
-      "grad_norm": 0.0006726975552737713,
-      "learning_rate": 0.00019999999766856071,
-      "loss": 46.0,
-      "step": 909
-    },
-    {
-      "epoch": 0.06957585488464553,
-      "grad_norm": 0.001108812284655869,
-      "learning_rate": 0.00019999999766337107,
-      "loss": 46.0,
-      "step": 910
-    },
-    {
-      "epoch": 0.06965231186803525,
-      "grad_norm": 0.0012064531911164522,
-      "learning_rate": 0.00019999999765817568,
-      "loss": 46.0,
-      "step": 911
-    },
-    {
-      "epoch": 0.06972876885142497,
-      "grad_norm": 0.0011297069722786546,
-      "learning_rate": 0.00019999999765297452,
-      "loss": 46.0,
-      "step": 912
-    },
-    {
-      "epoch": 0.06980522583481469,
-      "grad_norm": 0.0018110997043550014,
-      "learning_rate": 0.00019999999764776761,
-      "loss": 46.0,
-      "step": 913
-    },
-    {
-      "epoch": 0.06988168281820441,
-      "grad_norm": 0.0003823382721748203,
-      "learning_rate": 0.00019999999764255488,
-      "loss": 46.0,
-      "step": 914
-    },
-    {
-      "epoch": 0.06995813980159413,
-      "grad_norm": 0.0005454560159705579,
-      "learning_rate": 0.00019999999763733642,
-      "loss": 46.0,
-      "step": 915
-    },
-    {
-      "epoch": 0.07003459678498385,
-      "grad_norm": 0.002416105940937996,
-      "learning_rate": 0.00019999999763211217,
-      "loss": 46.0,
-      "step": 916
-    },
-    {
-      "epoch": 0.07011105376837357,
-      "grad_norm": 0.0006399700650945306,
-      "learning_rate": 0.00019999999762688217,
-      "loss": 46.0,
-      "step": 917
-    },
-    {
-      "epoch": 0.07018751075176329,
-      "grad_norm": 0.0007986078271642327,
-      "learning_rate": 0.00019999999762164642,
-      "loss": 46.0,
-      "step": 918
-    },
-    {
-      "epoch": 0.070263967735153,
-      "grad_norm": 0.0013469344703480601,
-      "learning_rate": 0.00019999999761640484,
-      "loss": 46.0,
-      "step": 919
-    },
-    {
-      "epoch": 0.07034042471854272,
-      "grad_norm": 0.00181288190651685,
-      "learning_rate": 0.0001999999976111575,
-      "loss": 46.0,
-      "step": 920
-    },
-    {
-      "epoch": 0.07041688170193244,
-      "grad_norm": 0.0010183494305238128,
-      "learning_rate": 0.00019999999760590443,
-      "loss": 46.0,
-      "step": 921
-    },
-    {
-      "epoch": 0.07049333868532218,
-      "grad_norm": 0.00047780590830370784,
-      "learning_rate": 0.00019999999760064556,
-      "loss": 46.0,
-      "step": 922
-    },
-    {
-      "epoch": 0.0705697956687119,
-      "grad_norm": 0.0014317670138552785,
-      "learning_rate": 0.00019999999759538094,
-      "loss": 46.0,
-      "step": 923
-    },
-    {
-      "epoch": 0.07064625265210162,
-      "grad_norm": 0.0009017597767524421,
-      "learning_rate": 0.00019999999759011056,
-      "loss": 46.0,
-      "step": 924
-    },
-    {
-      "epoch": 0.07072270963549133,
-      "grad_norm": 0.0011452081380411983,
-      "learning_rate": 0.0001999999975848344,
-      "loss": 46.0,
-      "step": 925
-    },
-    {
-      "epoch": 0.07079916661888105,
-      "grad_norm": 0.001637183828279376,
-      "learning_rate": 0.00019999999757955246,
-      "loss": 46.0,
-      "step": 926
-    },
-    {
-      "epoch": 0.07087562360227077,
-      "grad_norm": 0.006138650234788656,
-      "learning_rate": 0.00019999999757426475,
-      "loss": 46.0,
-      "step": 927
-    },
-    {
-      "epoch": 0.07095208058566049,
-      "grad_norm": 0.0009082559263333678,
-      "learning_rate": 0.00019999999756897127,
-      "loss": 46.0,
-      "step": 928
-    },
-    {
-      "epoch": 0.07102853756905021,
-      "grad_norm": 0.001173483207821846,
-      "learning_rate": 0.00019999999756367201,
-      "loss": 46.0,
-      "step": 929
-    },
-    {
-      "epoch": 0.07110499455243993,
-      "grad_norm": 0.000761194562073797,
-      "learning_rate": 0.000199999997558367,
-      "loss": 46.0,
-      "step": 930
-    },
-    {
-      "epoch": 0.07118145153582965,
-      "grad_norm": 0.0019090657588094473,
-      "learning_rate": 0.00019999999755305624,
-      "loss": 46.0,
-      "step": 931
-    },
-    {
-      "epoch": 0.07125790851921937,
-      "grad_norm": 0.000802334223408252,
-      "learning_rate": 0.0001999999975477397,
-      "loss": 46.0,
-      "step": 932
-    },
-    {
-      "epoch": 0.07133436550260909,
-      "grad_norm": 0.0006604130030609667,
-      "learning_rate": 0.00019999999754241737,
-      "loss": 46.0,
-      "step": 933
-    },
-    {
-      "epoch": 0.07141082248599881,
-      "grad_norm": 0.00108730373904109,
-      "learning_rate": 0.00019999999753708927,
-      "loss": 46.0,
-      "step": 934
-    },
-    {
-      "epoch": 0.07148727946938853,
-      "grad_norm": 0.0008749807602725923,
-      "learning_rate": 0.00019999999753175543,
-      "loss": 46.0,
-      "step": 935
-    },
-    {
-      "epoch": 0.07156373645277826,
-      "grad_norm": 0.000738863309379667,
-      "learning_rate": 0.0001999999975264158,
-      "loss": 46.0,
-      "step": 936
-    },
-    {
-      "epoch": 0.07164019343616798,
-      "grad_norm": 0.0009789742762222886,
-      "learning_rate": 0.00019999999752107038,
-      "loss": 46.0,
-      "step": 937
-    },
-    {
-      "epoch": 0.0717166504195577,
-      "grad_norm": 0.00181388680357486,
-      "learning_rate": 0.00019999999751571922,
-      "loss": 46.0,
-      "step": 938
-    },
-    {
-      "epoch": 0.07179310740294742,
-      "grad_norm": 0.0017250460805371404,
-      "learning_rate": 0.00019999999751036228,
-      "loss": 46.0,
-      "step": 939
-    },
-    {
-      "epoch": 0.07186956438633714,
-      "grad_norm": 0.0009100417955778539,
-      "learning_rate": 0.00019999999750499957,
-      "loss": 46.0,
-      "step": 940
-    },
-    {
-      "epoch": 0.07194602136972686,
-      "grad_norm": 0.0007340360316447914,
-      "learning_rate": 0.00019999999749963107,
-      "loss": 46.0,
-      "step": 941
-    },
-    {
-      "epoch": 0.07202247835311658,
-      "grad_norm": 0.001073396997526288,
-      "learning_rate": 0.00019999999749425684,
-      "loss": 46.0,
-      "step": 942
-    },
-    {
-      "epoch": 0.0720989353365063,
-      "grad_norm": 0.00076859793625772,
-      "learning_rate": 0.00019999999748887684,
-      "loss": 46.0,
-      "step": 943
-    },
-    {
-      "epoch": 0.07217539231989602,
-      "grad_norm": 0.0008783076191321015,
-      "learning_rate": 0.00019999999748349106,
-      "loss": 46.0,
-      "step": 944
-    },
-    {
-      "epoch": 0.07225184930328574,
-      "grad_norm": 0.0005738537875004113,
-      "learning_rate": 0.0001999999974780995,
-      "loss": 46.0,
-      "step": 945
-    },
-    {
-      "epoch": 0.07232830628667546,
-      "grad_norm": 0.0015352055197581649,
-      "learning_rate": 0.00019999999747270217,
-      "loss": 46.0,
-      "step": 946
-    },
-    {
-      "epoch": 0.07240476327006518,
-      "grad_norm": 0.0012679146602749825,
-      "learning_rate": 0.00019999999746729907,
-      "loss": 46.0,
-      "step": 947
-    },
-    {
-      "epoch": 0.0724812202534549,
-      "grad_norm": 0.0005767527618445456,
-      "learning_rate": 0.0001999999974618902,
-      "loss": 46.0,
-      "step": 948
-    },
-    {
-      "epoch": 0.07255767723684461,
-      "grad_norm": 0.0008155430550687015,
-      "learning_rate": 0.00019999999745647556,
-      "loss": 46.0,
-      "step": 949
-    },
-    {
-      "epoch": 0.07263413422023433,
-      "grad_norm": 0.0011201055021956563,
-      "learning_rate": 0.00019999999745105518,
-      "loss": 46.0,
-      "step": 950
-    },
-    {
-      "epoch": 0.07271059120362407,
-      "grad_norm": 0.0010096292244270444,
-      "learning_rate": 0.000199999997445629,
-      "loss": 46.0,
-      "step": 951
-    },
-    {
-      "epoch": 0.07278704818701379,
-      "grad_norm": 0.001026639831252396,
-      "learning_rate": 0.00019999999744019705,
-      "loss": 46.0,
-      "step": 952
-    },
-    {
-      "epoch": 0.0728635051704035,
-      "grad_norm": 0.0006314333877526224,
-      "learning_rate": 0.00019999999743475935,
-      "loss": 46.0,
-      "step": 953
-    },
-    {
-      "epoch": 0.07293996215379323,
-      "grad_norm": 0.0014932324411347508,
-      "learning_rate": 0.00019999999742931587,
-      "loss": 46.0,
-      "step": 954
-    },
-    {
-      "epoch": 0.07301641913718294,
-      "grad_norm": 0.0007988548022694886,
-      "learning_rate": 0.0001999999974238666,
-      "loss": 46.0,
-      "step": 955
-    },
-    {
-      "epoch": 0.07309287612057266,
-      "grad_norm": 0.0011979959672316909,
-      "learning_rate": 0.0001999999974184116,
-      "loss": 46.0,
-      "step": 956
-    },
-    {
-      "epoch": 0.07316933310396238,
-      "grad_norm": 0.0006841117865405977,
-      "learning_rate": 0.0001999999974129508,
-      "loss": 46.0,
-      "step": 957
-    },
-    {
-      "epoch": 0.0732457900873521,
-      "grad_norm": 0.0005685442010872066,
-      "learning_rate": 0.00019999999740748426,
-      "loss": 46.0,
-      "step": 958
-    },
-    {
-      "epoch": 0.07332224707074182,
-      "grad_norm": 0.0009426722535863519,
-      "learning_rate": 0.00019999999740201191,
-      "loss": 46.0,
-      "step": 959
-    },
-    {
-      "epoch": 0.07339870405413154,
-      "grad_norm": 0.0014698036247864366,
-      "learning_rate": 0.00019999999739653382,
-      "loss": 46.0,
-      "step": 960
-    },
-    {
-      "epoch": 0.07347516103752126,
-      "grad_norm": 0.0009936311980709434,
-      "learning_rate": 0.00019999999739104998,
-      "loss": 46.0,
-      "step": 961
-    },
-    {
-      "epoch": 0.07355161802091098,
-      "grad_norm": 0.001483134226873517,
-      "learning_rate": 0.0001999999973855603,
-      "loss": 46.0,
-      "step": 962
-    },
-    {
-      "epoch": 0.0736280750043007,
-      "grad_norm": 0.0008906892035156488,
-      "learning_rate": 0.00019999999738006492,
-      "loss": 46.0,
-      "step": 963
-    },
-    {
-      "epoch": 0.07370453198769042,
-      "grad_norm": 0.001091508544050157,
-      "learning_rate": 0.00019999999737456374,
-      "loss": 46.0,
-      "step": 964
-    },
-    {
-      "epoch": 0.07378098897108015,
-      "grad_norm": 0.0006207852275110781,
-      "learning_rate": 0.0001999999973690568,
-      "loss": 46.0,
-      "step": 965
-    },
-    {
-      "epoch": 0.07385744595446987,
-      "grad_norm": 0.0018408653559163213,
-      "learning_rate": 0.0001999999973635441,
-      "loss": 46.0,
-      "step": 966
-    },
-    {
-      "epoch": 0.07393390293785959,
-      "grad_norm": 0.002001339802518487,
-      "learning_rate": 0.0001999999973580256,
-      "loss": 46.0,
-      "step": 967
-    },
-    {
-      "epoch": 0.07401035992124931,
-      "grad_norm": 0.0011400439543649554,
-      "learning_rate": 0.00019999999735250133,
-      "loss": 46.0,
-      "step": 968
-    },
-    {
-      "epoch": 0.07408681690463903,
-      "grad_norm": 0.0028275521472096443,
-      "learning_rate": 0.00019999999734697133,
-      "loss": 46.0,
-      "step": 969
-    },
-    {
-      "epoch": 0.07416327388802875,
-      "grad_norm": 0.0011981431161984801,
-      "learning_rate": 0.00019999999734143553,
-      "loss": 46.0,
-      "step": 970
-    },
-    {
-      "epoch": 0.07423973087141847,
-      "grad_norm": 0.0008723320788703859,
-      "learning_rate": 0.00019999999733589399,
-      "loss": 46.0,
-      "step": 971
-    },
-    {
-      "epoch": 0.07431618785480819,
-      "grad_norm": 0.0011318976758047938,
-      "learning_rate": 0.00019999999733034667,
-      "loss": 46.0,
-      "step": 972
-    },
-    {
-      "epoch": 0.07439264483819791,
-      "grad_norm": 0.0007382151670753956,
-      "learning_rate": 0.00019999999732479354,
-      "loss": 46.0,
-      "step": 973
-    },
-    {
-      "epoch": 0.07446910182158763,
-      "grad_norm": 0.0004299880238249898,
-      "learning_rate": 0.00019999999731923468,
-      "loss": 46.0,
-      "step": 974
-    },
-    {
-      "epoch": 0.07454555880497735,
-      "grad_norm": 0.000987416016869247,
-      "learning_rate": 0.00019999999731367006,
-      "loss": 46.0,
-      "step": 975
-    },
-    {
-      "epoch": 0.07462201578836707,
-      "grad_norm": 0.0010482885409146547,
-      "learning_rate": 0.00019999999730809965,
-      "loss": 46.0,
-      "step": 976
-    },
-    {
-      "epoch": 0.07469847277175679,
-      "grad_norm": 0.0005643464392051101,
-      "learning_rate": 0.00019999999730252346,
-      "loss": 46.0,
-      "step": 977
-    },
-    {
-      "epoch": 0.0747749297551465,
-      "grad_norm": 0.0006474208785220981,
-      "learning_rate": 0.00019999999729694153,
-      "loss": 46.0,
-      "step": 978
-    },
-    {
-      "epoch": 0.07485138673853622,
-      "grad_norm": 0.001107594114728272,
-      "learning_rate": 0.0001999999972913538,
-      "loss": 46.0,
-      "step": 979
-    },
-    {
-      "epoch": 0.07492784372192596,
-      "grad_norm": 0.0016678207321092486,
-      "learning_rate": 0.00019999999728576032,
-      "loss": 46.0,
-      "step": 980
-    },
-    {
-      "epoch": 0.07500430070531568,
-      "grad_norm": 0.000690351938828826,
-      "learning_rate": 0.00019999999728016106,
-      "loss": 46.0,
-      "step": 981
-    },
-    {
-      "epoch": 0.0750807576887054,
-      "grad_norm": 0.0011274040443822742,
-      "learning_rate": 0.00019999999727455603,
-      "loss": 46.0,
-      "step": 982
-    },
-    {
-      "epoch": 0.07515721467209512,
-      "grad_norm": 0.0009322416153736413,
-      "learning_rate": 0.00019999999726894526,
-      "loss": 46.0,
-      "step": 983
-    },
-    {
-      "epoch": 0.07523367165548484,
-      "grad_norm": 0.0013927072286605835,
-      "learning_rate": 0.0001999999972633287,
-      "loss": 46.0,
-      "step": 984
-    },
-    {
-      "epoch": 0.07531012863887455,
-      "grad_norm": 0.0009432007791474462,
-      "learning_rate": 0.00019999999725770634,
-      "loss": 46.0,
-      "step": 985
-    },
-    {
-      "epoch": 0.07538658562226427,
-      "grad_norm": 0.00047698491835035384,
-      "learning_rate": 0.00019999999725207825,
-      "loss": 46.0,
-      "step": 986
-    },
-    {
-      "epoch": 0.075463042605654,
-      "grad_norm": 0.0006307463045231998,
-      "learning_rate": 0.00019999999724644435,
-      "loss": 46.0,
-      "step": 987
-    },
-    {
-      "epoch": 0.07553949958904371,
-      "grad_norm": 0.0013307937188073993,
-      "learning_rate": 0.00019999999724080474,
-      "loss": 46.0,
-      "step": 988
-    },
-    {
-      "epoch": 0.07561595657243343,
-      "grad_norm": 0.0009763085399754345,
-      "learning_rate": 0.0001999999972351593,
-      "loss": 46.0,
-      "step": 989
-    },
-    {
-      "epoch": 0.07569241355582315,
-      "grad_norm": 0.0010025585070252419,
-      "learning_rate": 0.00019999999722950814,
-      "loss": 46.0,
-      "step": 990
-    },
-    {
-      "epoch": 0.07576887053921287,
-      "grad_norm": 0.0009095129207707942,
-      "learning_rate": 0.00019999999722385118,
-      "loss": 46.0,
-      "step": 991
-    },
-    {
-      "epoch": 0.07584532752260259,
-      "grad_norm": 0.0009452021331526339,
-      "learning_rate": 0.00019999999721818847,
-      "loss": 46.0,
-      "step": 992
-    },
-    {
-      "epoch": 0.07592178450599231,
-      "grad_norm": 0.0007823688210919499,
-      "learning_rate": 0.00019999999721252,
-      "loss": 46.0,
-      "step": 993
-    },
-    {
-      "epoch": 0.07599824148938204,
-      "grad_norm": 0.0006104687927290797,
-      "learning_rate": 0.0001999999972068457,
-      "loss": 46.0,
-      "step": 994
-    },
-    {
-      "epoch": 0.07607469847277176,
-      "grad_norm": 0.0005413282196968794,
-      "learning_rate": 0.00019999999720116568,
-      "loss": 46.0,
-      "step": 995
-    },
-    {
-      "epoch": 0.07615115545616148,
-      "grad_norm": 0.0019189310260117054,
-      "learning_rate": 0.00019999999719547988,
-      "loss": 46.0,
-      "step": 996
-    },
-    {
-      "epoch": 0.0762276124395512,
-      "grad_norm": 0.0006479976000264287,
-      "learning_rate": 0.0001999999971897883,
-      "loss": 46.0,
-      "step": 997
-    },
-    {
-      "epoch": 0.07630406942294092,
-      "grad_norm": 0.0024212358985096216,
-      "learning_rate": 0.000199999997184091,
-      "loss": 46.0,
-      "step": 998
-    },
-    {
-      "epoch": 0.07638052640633064,
-      "grad_norm": 0.001033797743730247,
-      "learning_rate": 0.0001999999971783879,
-      "loss": 46.0,
-      "step": 999
-    },
-    {
-      "epoch": 0.07645698338972036,
-      "grad_norm": 0.0011149945203214884,
-      "learning_rate": 0.000199999997172679,
-      "loss": 46.0,
-      "step": 1000
-    },
-    {
-      "epoch": 0.07653344037311008,
-      "grad_norm": 0.0009869968052953482,
-      "learning_rate": 0.0001999999971669644,
-      "loss": 46.0,
-      "step": 1001
-    },
-    {
-      "epoch": 0.0766098973564998,
-      "grad_norm": 0.0008997033583000302,
-      "learning_rate": 0.00019999999716124395,
-      "loss": 46.0,
-      "step": 1002
-    },
-    {
-      "epoch": 0.07668635433988952,
-      "grad_norm": 0.0005471960757859051,
-      "learning_rate": 0.00019999999715551776,
-      "loss": 46.0,
-      "step": 1003
-    },
-    {
-      "epoch": 0.07676281132327924,
-      "grad_norm": 0.002586745424196124,
-      "learning_rate": 0.0001999999971497858,
-      "loss": 46.0,
-      "step": 1004
-    },
-    {
-      "epoch": 0.07683926830666896,
-      "grad_norm": 0.0005815322510898113,
-      "learning_rate": 0.0001999999971440481,
-      "loss": 46.0,
-      "step": 1005
-    },
-    {
-      "epoch": 0.07691572529005868,
-      "grad_norm": 0.001637510024011135,
-      "learning_rate": 0.0001999999971383046,
-      "loss": 46.0,
-      "step": 1006
-    },
-    {
-      "epoch": 0.0769921822734484,
-      "grad_norm": 0.0006479872390627861,
-      "learning_rate": 0.00019999999713255534,
-      "loss": 46.0,
-      "step": 1007
-    },
-    {
-      "epoch": 0.07706863925683811,
-      "grad_norm": 0.0017759514739736915,
-      "learning_rate": 0.0001999999971268003,
-      "loss": 46.0,
-      "step": 1008
-    },
-    {
-      "epoch": 0.07714509624022785,
-      "grad_norm": 0.0005267034866847098,
-      "learning_rate": 0.0001999999971210395,
-      "loss": 46.0,
-      "step": 1009
-    },
-    {
-      "epoch": 0.07722155322361757,
-      "grad_norm": 0.0009080252493731678,
-      "learning_rate": 0.00019999999711527294,
-      "loss": 46.0,
-      "step": 1010
-    },
-    {
-      "epoch": 0.07729801020700729,
-      "grad_norm": 0.0006503700278699398,
-      "learning_rate": 0.00019999999710950062,
-      "loss": 46.0,
-      "step": 1011
-    },
-    {
-      "epoch": 0.077374467190397,
-      "grad_norm": 0.0009566502994857728,
-      "learning_rate": 0.00019999999710372247,
-      "loss": 46.0,
-      "step": 1012
-    },
-    {
-      "epoch": 0.07745092417378673,
-      "grad_norm": 0.0010661629494279623,
-      "learning_rate": 0.0001999999970979386,
-      "loss": 46.0,
-      "step": 1013
-    },
-    {
-      "epoch": 0.07752738115717644,
-      "grad_norm": 0.0011315967421978712,
-      "learning_rate": 0.000199999997092149,
-      "loss": 46.0,
-      "step": 1014
-    },
-    {
-      "epoch": 0.07760383814056616,
-      "grad_norm": 0.0012493714457377791,
-      "learning_rate": 0.00019999999708635355,
-      "loss": 46.0,
-      "step": 1015
-    },
-    {
-      "epoch": 0.07768029512395588,
-      "grad_norm": 0.0011820078361779451,
-      "learning_rate": 0.00019999999708055237,
-      "loss": 46.0,
-      "step": 1016
-    },
-    {
-      "epoch": 0.0777567521073456,
-      "grad_norm": 0.0015900529688224196,
-      "learning_rate": 0.0001999999970747454,
-      "loss": 46.0,
-      "step": 1017
-    },
-    {
-      "epoch": 0.07783320909073532,
-      "grad_norm": 0.0008089744369499385,
-      "learning_rate": 0.00019999999706893267,
-      "loss": 46.0,
-      "step": 1018
-    },
-    {
-      "epoch": 0.07790966607412504,
-      "grad_norm": 0.0026222998276352882,
-      "learning_rate": 0.00019999999706311422,
-      "loss": 46.0,
-      "step": 1019
-    },
-    {
-      "epoch": 0.07798612305751476,
-      "grad_norm": 0.0018962507601827383,
-      "learning_rate": 0.00019999999705728994,
-      "loss": 46.0,
-      "step": 1020
-    },
-    {
-      "epoch": 0.07806258004090448,
-      "grad_norm": 0.00139760528691113,
-      "learning_rate": 0.0001999999970514599,
-      "loss": 46.0,
-      "step": 1021
-    },
-    {
-      "epoch": 0.0781390370242942,
-      "grad_norm": 0.003917804919183254,
-      "learning_rate": 0.0001999999970456241,
-      "loss": 46.0,
-      "step": 1022
-    },
-    {
-      "epoch": 0.07821549400768393,
-      "grad_norm": 0.0006986702792346478,
-      "learning_rate": 0.00019999999703978252,
-      "loss": 46.0,
-      "step": 1023
-    },
-    {
-      "epoch": 0.07829195099107365,
-      "grad_norm": 0.004320481792092323,
-      "learning_rate": 0.0001999999970339352,
-      "loss": 46.0,
-      "step": 1024
-    },
-    {
-      "epoch": 0.07836840797446337,
-      "grad_norm": 0.0010345918126404285,
-      "learning_rate": 0.00019999999702808205,
-      "loss": 46.0,
-      "step": 1025
-    },
-    {
-      "epoch": 0.07844486495785309,
-      "grad_norm": 0.0016160261584445834,
-      "learning_rate": 0.00019999999702222318,
-      "loss": 46.0,
-      "step": 1026
-    },
-    {
-      "epoch": 0.07852132194124281,
-      "grad_norm": 0.0013722765725106,
-      "learning_rate": 0.00019999999701635855,
-      "loss": 46.0,
-      "step": 1027
-    },
-    {
-      "epoch": 0.07859777892463253,
-      "grad_norm": 0.0010916527826339006,
-      "learning_rate": 0.00019999999701048813,
-      "loss": 46.0,
-      "step": 1028
-    },
-    {
-      "epoch": 0.07867423590802225,
-      "grad_norm": 0.0010964693501591682,
-      "learning_rate": 0.00019999999700461195,
-      "loss": 46.0,
-      "step": 1029
-    },
-    {
-      "epoch": 0.07875069289141197,
-      "grad_norm": 0.0010036800522357225,
-      "learning_rate": 0.00019999999699873,
-      "loss": 46.0,
-      "step": 1030
-    },
-    {
-      "epoch": 0.07882714987480169,
-      "grad_norm": 0.0009888997301459312,
-      "learning_rate": 0.00019999999699284225,
-      "loss": 46.0,
-      "step": 1031
-    },
-    {
-      "epoch": 0.07890360685819141,
-      "grad_norm": 0.001753503573127091,
-      "learning_rate": 0.00019999999698694872,
-      "loss": 46.0,
-      "step": 1032
-    },
-    {
-      "epoch": 0.07898006384158113,
-      "grad_norm": 0.0009753037593327463,
-      "learning_rate": 0.00019999999698104947,
-      "loss": 46.0,
-      "step": 1033
-    },
-    {
-      "epoch": 0.07905652082497085,
-      "grad_norm": 0.0011483731213957071,
-      "learning_rate": 0.00019999999697514444,
-      "loss": 46.0,
-      "step": 1034
-    },
-    {
-      "epoch": 0.07913297780836057,
-      "grad_norm": 0.0008962884312495589,
-      "learning_rate": 0.00019999999696923364,
-      "loss": 46.0,
-      "step": 1035
-    },
-    {
-      "epoch": 0.07920943479175029,
-      "grad_norm": 0.000925900531001389,
-      "learning_rate": 0.00019999999696331707,
-      "loss": 46.0,
-      "step": 1036
-    },
-    {
-      "epoch": 0.07928589177514,
-      "grad_norm": 0.0007393479463644326,
-      "learning_rate": 0.00019999999695739473,
-      "loss": 46.0,
-      "step": 1037
-    },
-    {
-      "epoch": 0.07936234875852974,
-      "grad_norm": 0.0013094200985506177,
-      "learning_rate": 0.0001999999969514666,
-      "loss": 46.0,
-      "step": 1038
-    },
-    {
-      "epoch": 0.07943880574191946,
-      "grad_norm": 0.002036183374002576,
-      "learning_rate": 0.0001999999969455327,
-      "loss": 46.0,
-      "step": 1039
-    },
-    {
-      "epoch": 0.07951526272530918,
-      "grad_norm": 0.0010627619922161102,
-      "learning_rate": 0.00019999999693959305,
-      "loss": 46.0,
-      "step": 1040
-    },
-    {
-      "epoch": 0.0795917197086989,
-      "grad_norm": 0.0008229623781517148,
-      "learning_rate": 0.00019999999693364763,
-      "loss": 46.0,
-      "step": 1041
-    },
-    {
-      "epoch": 0.07966817669208862,
-      "grad_norm": 0.0008170257788151503,
-      "learning_rate": 0.00019999999692769645,
-      "loss": 46.0,
-      "step": 1042
-    },
-    {
-      "epoch": 0.07974463367547834,
-      "grad_norm": 0.0005138233536854386,
-      "learning_rate": 0.00019999999692173946,
-      "loss": 46.0,
-      "step": 1043
-    },
-    {
-      "epoch": 0.07982109065886805,
-      "grad_norm": 0.0014529372565448284,
-      "learning_rate": 0.00019999999691577676,
-      "loss": 46.0,
-      "step": 1044
-    },
-    {
-      "epoch": 0.07989754764225777,
-      "grad_norm": 0.0009753629565238953,
-      "learning_rate": 0.00019999999690980825,
-      "loss": 46.0,
-      "step": 1045
-    },
-    {
-      "epoch": 0.0799740046256475,
-      "grad_norm": 0.001065521384589374,
-      "learning_rate": 0.00019999999690383397,
-      "loss": 46.0,
-      "step": 1046
-    },
-    {
-      "epoch": 0.08005046160903721,
-      "grad_norm": 0.0012711328454315662,
-      "learning_rate": 0.00019999999689785395,
-      "loss": 46.0,
-      "step": 1047
-    },
-    {
-      "epoch": 0.08012691859242693,
-      "grad_norm": 0.0011729527032002807,
-      "learning_rate": 0.00019999999689186812,
-      "loss": 46.0,
-      "step": 1048
-    },
-    {
-      "epoch": 0.08020337557581665,
-      "grad_norm": 0.000720737676601857,
-      "learning_rate": 0.00019999999688587655,
-      "loss": 46.0,
-      "step": 1049
-    },
-    {
-      "epoch": 0.08027983255920637,
-      "grad_norm": 0.0017695153364911675,
-      "learning_rate": 0.00019999999687987923,
-      "loss": 46.0,
-      "step": 1050
-    },
-    {
-      "epoch": 0.08035628954259609,
-      "grad_norm": 0.0010254306253045797,
-      "learning_rate": 0.00019999999687387608,
-      "loss": 46.0,
-      "step": 1051
-    },
-    {
-      "epoch": 0.08043274652598582,
-      "grad_norm": 0.0009627710096538067,
-      "learning_rate": 0.00019999999686786722,
-      "loss": 46.0,
-      "step": 1052
-    },
-    {
-      "epoch": 0.08050920350937554,
-      "grad_norm": 0.000522421847563237,
-      "learning_rate": 0.00019999999686185255,
-      "loss": 46.0,
-      "step": 1053
-    },
-    {
-      "epoch": 0.08058566049276526,
-      "grad_norm": 0.0007806363864801824,
-      "learning_rate": 0.00019999999685583214,
-      "loss": 46.0,
-      "step": 1054
-    },
-    {
-      "epoch": 0.08066211747615498,
-      "grad_norm": 0.0035145070869475603,
-      "learning_rate": 0.00019999999684980596,
-      "loss": 46.0,
-      "step": 1055
-    },
-    {
-      "epoch": 0.0807385744595447,
-      "grad_norm": 0.0007616059156134725,
-      "learning_rate": 0.00019999999684377397,
-      "loss": 46.0,
-      "step": 1056
-    },
-    {
-      "epoch": 0.08081503144293442,
-      "grad_norm": 0.001607340993359685,
-      "learning_rate": 0.00019999999683773624,
-      "loss": 46.0,
-      "step": 1057
-    },
-    {
-      "epoch": 0.08089148842632414,
-      "grad_norm": 0.0024056860711425543,
-      "learning_rate": 0.00019999999683169276,
-      "loss": 46.0,
-      "step": 1058
-    },
-    {
-      "epoch": 0.08096794540971386,
-      "grad_norm": 0.0007960721268318594,
-      "learning_rate": 0.00019999999682564345,
-      "loss": 46.0,
-      "step": 1059
-    },
-    {
-      "epoch": 0.08104440239310358,
-      "grad_norm": 0.0010764168109744787,
-      "learning_rate": 0.00019999999681958843,
-      "loss": 46.0,
-      "step": 1060
-    },
-    {
-      "epoch": 0.0811208593764933,
-      "grad_norm": 0.001000276650302112,
-      "learning_rate": 0.0001999999968135276,
-      "loss": 46.0,
-      "step": 1061
-    },
-    {
-      "epoch": 0.08119731635988302,
-      "grad_norm": 0.0010639075189828873,
-      "learning_rate": 0.00019999999680746103,
-      "loss": 46.0,
-      "step": 1062
-    },
-    {
-      "epoch": 0.08127377334327274,
-      "grad_norm": 0.0025826103519648314,
-      "learning_rate": 0.00019999999680138868,
-      "loss": 46.0,
-      "step": 1063
-    },
-    {
-      "epoch": 0.08135023032666246,
-      "grad_norm": 0.0008312023128382862,
-      "learning_rate": 0.00019999999679531057,
-      "loss": 46.0,
-      "step": 1064
-    },
-    {
-      "epoch": 0.08142668731005218,
-      "grad_norm": 0.0008342218352481723,
-      "learning_rate": 0.00019999999678922667,
-      "loss": 46.0,
-      "step": 1065
-    },
-    {
-      "epoch": 0.0815031442934419,
-      "grad_norm": 0.0007059397175908089,
-      "learning_rate": 0.000199999996783137,
-      "loss": 46.0,
-      "step": 1066
-    },
-    {
-      "epoch": 0.08157960127683163,
-      "grad_norm": 0.0006852514925412834,
-      "learning_rate": 0.00019999999677704157,
-      "loss": 46.0,
-      "step": 1067
-    },
-    {
-      "epoch": 0.08165605826022135,
-      "grad_norm": 0.00293467054143548,
-      "learning_rate": 0.00019999999677094036,
-      "loss": 46.0,
-      "step": 1068
-    },
-    {
-      "epoch": 0.08173251524361107,
-      "grad_norm": 0.0012233840534463525,
-      "learning_rate": 0.00019999999676483343,
-      "loss": 46.0,
-      "step": 1069
-    },
-    {
-      "epoch": 0.08180897222700079,
-      "grad_norm": 0.0007880316697992384,
-      "learning_rate": 0.00019999999675872067,
-      "loss": 46.0,
-      "step": 1070
-    },
-    {
-      "epoch": 0.0818854292103905,
-      "grad_norm": 0.0005319473566487432,
-      "learning_rate": 0.00019999999675260216,
-      "loss": 46.0,
-      "step": 1071
-    },
-    {
-      "epoch": 0.08196188619378023,
-      "grad_norm": 0.0007042661309242249,
-      "learning_rate": 0.00019999999674647788,
-      "loss": 46.0,
-      "step": 1072
-    },
-    {
-      "epoch": 0.08203834317716995,
-      "grad_norm": 0.0010053367586806417,
-      "learning_rate": 0.00019999999674034786,
-      "loss": 46.0,
-      "step": 1073
-    },
-    {
-      "epoch": 0.08211480016055966,
-      "grad_norm": 0.004070596303790808,
-      "learning_rate": 0.00019999999673421203,
-      "loss": 46.0,
-      "step": 1074
-    },
-    {
-      "epoch": 0.08219125714394938,
-      "grad_norm": 0.002002035966143012,
-      "learning_rate": 0.00019999999672807043,
-      "loss": 46.0,
-      "step": 1075
-    },
-    {
-      "epoch": 0.0822677141273391,
-      "grad_norm": 0.0011843186803162098,
-      "learning_rate": 0.0001999999967219231,
-      "loss": 46.0,
-      "step": 1076
-    },
-    {
-      "epoch": 0.08234417111072882,
-      "grad_norm": 0.0007600758108310401,
-      "learning_rate": 0.00019999999671576997,
-      "loss": 46.0,
-      "step": 1077
-    },
-    {
-      "epoch": 0.08242062809411854,
-      "grad_norm": 0.001096943742595613,
-      "learning_rate": 0.00019999999670961108,
-      "loss": 46.0,
-      "step": 1078
-    },
-    {
-      "epoch": 0.08249708507750826,
-      "grad_norm": 0.000551481731235981,
-      "learning_rate": 0.00019999999670344641,
-      "loss": 46.0,
-      "step": 1079
-    },
-    {
-      "epoch": 0.08257354206089798,
-      "grad_norm": 0.002418719232082367,
-      "learning_rate": 0.000199999996697276,
-      "loss": 46.0,
-      "step": 1080
-    },
-    {
-      "epoch": 0.08264999904428771,
-      "grad_norm": 0.0008597131236456335,
-      "learning_rate": 0.0001999999966910998,
-      "loss": 46.0,
-      "step": 1081
-    },
-    {
-      "epoch": 0.08272645602767743,
-      "grad_norm": 0.0015707275597378612,
-      "learning_rate": 0.00019999999668491783,
-      "loss": 46.0,
-      "step": 1082
-    },
-    {
-      "epoch": 0.08280291301106715,
-      "grad_norm": 0.003683779388666153,
-      "learning_rate": 0.00019999999667873007,
-      "loss": 46.0,
-      "step": 1083
-    },
-    {
-      "epoch": 0.08287936999445687,
-      "grad_norm": 0.0003693265316542238,
-      "learning_rate": 0.00019999999667253657,
-      "loss": 46.0,
-      "step": 1084
-    },
-    {
-      "epoch": 0.08295582697784659,
-      "grad_norm": 0.0010062635410577059,
-      "learning_rate": 0.0001999999966663373,
-      "loss": 46.0,
-      "step": 1085
-    },
-    {
-      "epoch": 0.08303228396123631,
-      "grad_norm": 0.0003948551893699914,
-      "learning_rate": 0.00019999999666013224,
-      "loss": 46.0,
-      "step": 1086
-    },
-    {
-      "epoch": 0.08310874094462603,
-      "grad_norm": 0.0009817811660468578,
-      "learning_rate": 0.00019999999665392144,
-      "loss": 46.0,
-      "step": 1087
-    },
-    {
-      "epoch": 0.08318519792801575,
-      "grad_norm": 0.0008574684034101665,
-      "learning_rate": 0.00019999999664770487,
-      "loss": 46.0,
-      "step": 1088
-    },
-    {
-      "epoch": 0.08326165491140547,
-      "grad_norm": 0.0024348909500986338,
-      "learning_rate": 0.0001999999966414825,
-      "loss": 46.0,
-      "step": 1089
-    },
-    {
-      "epoch": 0.08333811189479519,
-      "grad_norm": 0.004272660706192255,
-      "learning_rate": 0.00019999999663525438,
-      "loss": 46.0,
-      "step": 1090
-    },
-    {
-      "epoch": 0.08341456887818491,
-      "grad_norm": 0.0010652975179255009,
-      "learning_rate": 0.00019999999662902046,
-      "loss": 46.0,
-      "step": 1091
-    },
-    {
-      "epoch": 0.08349102586157463,
-      "grad_norm": 0.0014635265106335282,
-      "learning_rate": 0.0001999999966227808,
-      "loss": 46.0,
-      "step": 1092
-    },
-    {
-      "epoch": 0.08356748284496435,
-      "grad_norm": 0.002833076985552907,
-      "learning_rate": 0.00019999999661653536,
-      "loss": 46.0,
-      "step": 1093
-    },
-    {
-      "epoch": 0.08364393982835407,
-      "grad_norm": 0.001694944454357028,
-      "learning_rate": 0.00019999999661028417,
-      "loss": 46.0,
-      "step": 1094
-    },
-    {
-      "epoch": 0.08372039681174379,
-      "grad_norm": 0.0007754330872558057,
-      "learning_rate": 0.0001999999966040272,
-      "loss": 46.0,
-      "step": 1095
-    },
-    {
-      "epoch": 0.08379685379513352,
-      "grad_norm": 0.0017635487020015717,
-      "learning_rate": 0.00019999999659776446,
-      "loss": 46.0,
-      "step": 1096
-    },
-    {
-      "epoch": 0.08387331077852324,
-      "grad_norm": 0.0010679919505491853,
-      "learning_rate": 0.00019999999659149593,
-      "loss": 46.0,
-      "step": 1097
-    },
-    {
-      "epoch": 0.08394976776191296,
-      "grad_norm": 0.0011753238504752517,
-      "learning_rate": 0.00019999999658522165,
-      "loss": 46.0,
-      "step": 1098
-    },
-    {
-      "epoch": 0.08402622474530268,
-      "grad_norm": 0.001619552611373365,
-      "learning_rate": 0.0001999999965789416,
-      "loss": 46.0,
-      "step": 1099
-    },
-    {
-      "epoch": 0.0841026817286924,
-      "grad_norm": 0.0029004209209233522,
-      "learning_rate": 0.00019999999657265577,
-      "loss": 46.0,
-      "step": 1100
-    },
-    {
-      "epoch": 0.08417913871208212,
-      "grad_norm": 0.0013809700030833483,
-      "learning_rate": 0.0001999999965663642,
-      "loss": 46.0,
-      "step": 1101
-    },
-    {
-      "epoch": 0.08425559569547184,
-      "grad_norm": 0.00043901524622924626,
-      "learning_rate": 0.00019999999656006683,
-      "loss": 46.0,
-      "step": 1102
-    },
-    {
-      "epoch": 0.08433205267886155,
-      "grad_norm": 0.0012373154750093818,
-      "learning_rate": 0.00019999999655376368,
-      "loss": 46.0,
-      "step": 1103
-    },
-    {
-      "epoch": 0.08440850966225127,
-      "grad_norm": 0.0013792196987196803,
-      "learning_rate": 0.0001999999965474548,
-      "loss": 46.0,
-      "step": 1104
-    },
-    {
-      "epoch": 0.084484966645641,
-      "grad_norm": 0.0009669942664913833,
-      "learning_rate": 0.00019999999654114016,
-      "loss": 46.0,
-      "step": 1105
-    },
-    {
-      "epoch": 0.08456142362903071,
-      "grad_norm": 0.0008969721966423094,
-      "learning_rate": 0.0001999999965348197,
-      "loss": 46.0,
-      "step": 1106
-    },
-    {
-      "epoch": 0.08463788061242043,
-      "grad_norm": 0.0032094954513013363,
-      "learning_rate": 0.0001999999965284935,
-      "loss": 46.0,
-      "step": 1107
-    },
-    {
-      "epoch": 0.08471433759581015,
-      "grad_norm": 0.0007989536970853806,
-      "learning_rate": 0.00019999999652216152,
-      "loss": 46.0,
-      "step": 1108
-    },
-    {
-      "epoch": 0.08479079457919987,
-      "grad_norm": 0.0007923963130451739,
-      "learning_rate": 0.00019999999651582376,
-      "loss": 46.0,
-      "step": 1109
-    },
-    {
-      "epoch": 0.0848672515625896,
-      "grad_norm": 0.002309756353497505,
-      "learning_rate": 0.00019999999650948026,
-      "loss": 46.0,
-      "step": 1110
-    },
-    {
-      "epoch": 0.08494370854597932,
-      "grad_norm": 0.0006696914788335562,
-      "learning_rate": 0.00019999999650313098,
-      "loss": 46.0,
-      "step": 1111
-    },
-    {
-      "epoch": 0.08502016552936904,
-      "grad_norm": 0.0010415037395432591,
-      "learning_rate": 0.00019999999649677593,
-      "loss": 46.0,
-      "step": 1112
-    },
-    {
-      "epoch": 0.08509662251275876,
-      "grad_norm": 0.00045059010153636336,
-      "learning_rate": 0.0001999999964904151,
-      "loss": 46.0,
-      "step": 1113
-    },
-    {
-      "epoch": 0.08517307949614848,
-      "grad_norm": 0.001029227045364678,
-      "learning_rate": 0.00019999999648404854,
-      "loss": 46.0,
-      "step": 1114
-    },
-    {
-      "epoch": 0.0852495364795382,
-      "grad_norm": 0.0010876213200390339,
-      "learning_rate": 0.00019999999647767617,
-      "loss": 46.0,
-      "step": 1115
-    },
-    {
-      "epoch": 0.08532599346292792,
-      "grad_norm": 0.0008422642713412642,
-      "learning_rate": 0.00019999999647129802,
-      "loss": 46.0,
-      "step": 1116
-    },
-    {
-      "epoch": 0.08540245044631764,
-      "grad_norm": 0.0008873858023434877,
-      "learning_rate": 0.0001999999964649141,
-      "loss": 46.0,
-      "step": 1117
-    },
-    {
-      "epoch": 0.08547890742970736,
-      "grad_norm": 0.0006313907215371728,
-      "learning_rate": 0.00019999999645852447,
-      "loss": 46.0,
-      "step": 1118
-    },
-    {
-      "epoch": 0.08555536441309708,
-      "grad_norm": 0.0017866980051621795,
-      "learning_rate": 0.000199999996452129,
-      "loss": 46.0,
-      "step": 1119
-    },
-    {
-      "epoch": 0.0856318213964868,
-      "grad_norm": 0.0027931237127631903,
-      "learning_rate": 0.0001999999964457278,
-      "loss": 46.0,
-      "step": 1120
-    },
-    {
-      "epoch": 0.08570827837987652,
-      "grad_norm": 0.0002896841906476766,
-      "learning_rate": 0.0001999999964393208,
-      "loss": 46.0,
-      "step": 1121
-    },
-    {
-      "epoch": 0.08578473536326624,
-      "grad_norm": 0.0004619108512997627,
-      "learning_rate": 0.00019999999643290805,
-      "loss": 46.0,
-      "step": 1122
-    },
-    {
-      "epoch": 0.08586119234665596,
-      "grad_norm": 0.0009961490286514163,
-      "learning_rate": 0.00019999999642648954,
-      "loss": 46.0,
-      "step": 1123
-    },
-    {
-      "epoch": 0.08593764933004568,
-      "grad_norm": 0.00150589132681489,
-      "learning_rate": 0.00019999999642006527,
-      "loss": 46.0,
-      "step": 1124
-    },
-    {
-      "epoch": 0.08601410631343541,
-      "grad_norm": 0.0011365427635610104,
-      "learning_rate": 0.00019999999641363522,
-      "loss": 46.0,
-      "step": 1125
-    },
-    {
-      "epoch": 0.08609056329682513,
-      "grad_norm": 0.0011451183818280697,
-      "learning_rate": 0.00019999999640719936,
-      "loss": 46.0,
-      "step": 1126
-    },
-    {
-      "epoch": 0.08616702028021485,
-      "grad_norm": 0.0008174086106009781,
-      "learning_rate": 0.00019999999640075777,
-      "loss": 46.0,
-      "step": 1127
-    },
-    {
-      "epoch": 0.08624347726360457,
-      "grad_norm": 0.001513507915660739,
-      "learning_rate": 0.00019999999639431042,
-      "loss": 46.0,
-      "step": 1128
-    },
-    {
-      "epoch": 0.08631993424699429,
-      "grad_norm": 0.0011059994576498866,
-      "learning_rate": 0.00019999999638785728,
-      "loss": 46.0,
-      "step": 1129
-    },
-    {
-      "epoch": 0.086396391230384,
-      "grad_norm": 0.0010196544462814927,
-      "learning_rate": 0.00019999999638139836,
-      "loss": 46.0,
-      "step": 1130
-    },
-    {
-      "epoch": 0.08647284821377373,
-      "grad_norm": 0.0021436179522424936,
-      "learning_rate": 0.00019999999637493373,
-      "loss": 46.0,
-      "step": 1131
-    },
-    {
-      "epoch": 0.08654930519716345,
-      "grad_norm": 0.0025370994117110968,
-      "learning_rate": 0.00019999999636846326,
-      "loss": 46.0,
-      "step": 1132
-    },
-    {
-      "epoch": 0.08662576218055316,
-      "grad_norm": 0.003925876226276159,
-      "learning_rate": 0.00019999999636198705,
-      "loss": 46.0,
-      "step": 1133
-    },
-    {
-      "epoch": 0.08670221916394288,
-      "grad_norm": 0.0011787598486989737,
-      "learning_rate": 0.0001999999963555051,
-      "loss": 46.0,
-      "step": 1134
-    },
-    {
-      "epoch": 0.0867786761473326,
-      "grad_norm": 0.0005098643596284091,
-      "learning_rate": 0.0001999999963490173,
-      "loss": 46.0,
-      "step": 1135
-    },
-    {
-      "epoch": 0.08685513313072232,
-      "grad_norm": 0.0009188891854137182,
-      "learning_rate": 0.0001999999963425238,
-      "loss": 46.0,
-      "step": 1136
-    },
-    {
-      "epoch": 0.08693159011411204,
-      "grad_norm": 0.0027898759581148624,
-      "learning_rate": 0.0001999999963360245,
-      "loss": 46.0,
-      "step": 1137
-    },
-    {
-      "epoch": 0.08700804709750176,
-      "grad_norm": 0.0013244698056951165,
-      "learning_rate": 0.00019999999632951945,
-      "loss": 46.0,
-      "step": 1138
-    },
-    {
-      "epoch": 0.0870845040808915,
-      "grad_norm": 0.0010905115632340312,
-      "learning_rate": 0.00019999999632300863,
-      "loss": 46.0,
-      "step": 1139
-    },
-    {
-      "epoch": 0.08716096106428121,
-      "grad_norm": 0.0010143949184566736,
-      "learning_rate": 0.00019999999631649203,
-      "loss": 46.0,
-      "step": 1140
-    },
-    {
-      "epoch": 0.08723741804767093,
-      "grad_norm": 0.0008232492255046964,
-      "learning_rate": 0.00019999999630996963,
-      "loss": 46.0,
-      "step": 1141
-    },
-    {
-      "epoch": 0.08731387503106065,
-      "grad_norm": 0.0018145126523450017,
-      "learning_rate": 0.00019999999630344152,
-      "loss": 46.0,
-      "step": 1142
-    },
-    {
-      "epoch": 0.08739033201445037,
-      "grad_norm": 0.0008560459245927632,
-      "learning_rate": 0.0001999999962969076,
-      "loss": 46.0,
-      "step": 1143
-    },
-    {
-      "epoch": 0.08746678899784009,
-      "grad_norm": 0.0014575072564184666,
-      "learning_rate": 0.00019999999629036793,
-      "loss": 46.0,
-      "step": 1144
-    },
-    {
-      "epoch": 0.08754324598122981,
-      "grad_norm": 0.0031108111143112183,
-      "learning_rate": 0.0001999999962838225,
-      "loss": 46.0,
-      "step": 1145
-    },
-    {
-      "epoch": 0.08761970296461953,
-      "grad_norm": 0.00047231800272129476,
-      "learning_rate": 0.00019999999627727126,
-      "loss": 46.0,
-      "step": 1146
-    },
-    {
-      "epoch": 0.08769615994800925,
-      "grad_norm": 0.0017584000015631318,
-      "learning_rate": 0.00019999999627071428,
-      "loss": 46.0,
-      "step": 1147
-    },
-    {
-      "epoch": 0.08777261693139897,
-      "grad_norm": 0.001495389617048204,
-      "learning_rate": 0.00019999999626415155,
-      "loss": 46.0,
-      "step": 1148
-    },
-    {
-      "epoch": 0.08784907391478869,
-      "grad_norm": 0.005392612423747778,
-      "learning_rate": 0.00019999999625758302,
-      "loss": 46.0,
-      "step": 1149
-    },
-    {
-      "epoch": 0.08792553089817841,
-      "grad_norm": 0.001163176610134542,
-      "learning_rate": 0.0001999999962510087,
-      "loss": 46.0,
-      "step": 1150
-    },
-    {
-      "epoch": 0.08800198788156813,
-      "grad_norm": 0.0011836659396067262,
-      "learning_rate": 0.00019999999624442863,
-      "loss": 46.0,
-      "step": 1151
-    },
-    {
-      "epoch": 0.08807844486495785,
-      "grad_norm": 0.0006165836821310222,
-      "learning_rate": 0.0001999999962378428,
-      "loss": 46.0,
-      "step": 1152
-    },
-    {
-      "epoch": 0.08815490184834757,
-      "grad_norm": 0.0011477165389806032,
-      "learning_rate": 0.00019999999623125124,
-      "loss": 46.0,
-      "step": 1153
-    },
-    {
-      "epoch": 0.0882313588317373,
-      "grad_norm": 0.001473361742682755,
-      "learning_rate": 0.00019999999622465384,
-      "loss": 46.0,
-      "step": 1154
-    },
-    {
-      "epoch": 0.08830781581512702,
-      "grad_norm": 0.0011002605315297842,
-      "learning_rate": 0.00019999999621805073,
-      "loss": 46.0,
-      "step": 1155
-    },
-    {
-      "epoch": 0.08838427279851674,
-      "grad_norm": 0.0009919332806020975,
-      "learning_rate": 0.00019999999621144178,
-      "loss": 46.0,
-      "step": 1156
-    },
-    {
-      "epoch": 0.08846072978190646,
-      "grad_norm": 0.0008903579437173903,
-      "learning_rate": 0.00019999999620482712,
-      "loss": 46.0,
-      "step": 1157
-    },
-    {
-      "epoch": 0.08853718676529618,
-      "grad_norm": 0.0009217155165970325,
-      "learning_rate": 0.00019999999619820668,
-      "loss": 46.0,
-      "step": 1158
-    },
-    {
-      "epoch": 0.0886136437486859,
-      "grad_norm": 0.0007448533433489501,
-      "learning_rate": 0.00019999999619158045,
-      "loss": 46.0,
-      "step": 1159
-    },
-    {
-      "epoch": 0.08869010073207562,
-      "grad_norm": 0.0004429701657500118,
-      "learning_rate": 0.00019999999618494846,
-      "loss": 46.0,
-      "step": 1160
-    },
-    {
-      "epoch": 0.08876655771546534,
-      "grad_norm": 0.0014019494410604239,
-      "learning_rate": 0.0001999999961783107,
-      "loss": 46.0,
-      "step": 1161
-    },
-    {
-      "epoch": 0.08884301469885506,
-      "grad_norm": 0.0009256371413357556,
-      "learning_rate": 0.00019999999617166718,
-      "loss": 46.0,
-      "step": 1162
-    },
-    {
-      "epoch": 0.08891947168224477,
-      "grad_norm": 0.0029703632462769747,
-      "learning_rate": 0.0001999999961650179,
-      "loss": 46.0,
-      "step": 1163
-    },
-    {
-      "epoch": 0.0889959286656345,
-      "grad_norm": 0.0011461441172286868,
-      "learning_rate": 0.00019999999615836282,
-      "loss": 46.0,
-      "step": 1164
-    },
-    {
-      "epoch": 0.08907238564902421,
-      "grad_norm": 0.0007109674625098705,
-      "learning_rate": 0.00019999999615170197,
-      "loss": 46.0,
-      "step": 1165
-    },
-    {
-      "epoch": 0.08914884263241393,
-      "grad_norm": 0.0009964013006538153,
-      "learning_rate": 0.0001999999961450354,
-      "loss": 46.0,
-      "step": 1166
-    },
-    {
-      "epoch": 0.08922529961580365,
-      "grad_norm": 0.0015873070806264877,
-      "learning_rate": 0.000199999996138363,
-      "loss": 46.0,
-      "step": 1167
-    },
-    {
-      "epoch": 0.08930175659919339,
-      "grad_norm": 0.003190083894878626,
-      "learning_rate": 0.00019999999613168486,
-      "loss": 46.0,
-      "step": 1168
-    },
-    {
-      "epoch": 0.0893782135825831,
-      "grad_norm": 0.0008296355954371393,
-      "learning_rate": 0.00019999999612500095,
-      "loss": 46.0,
-      "step": 1169
-    },
-    {
-      "epoch": 0.08945467056597282,
-      "grad_norm": 0.000597225094679743,
-      "learning_rate": 0.00019999999611831126,
-      "loss": 46.0,
-      "step": 1170
-    },
-    {
-      "epoch": 0.08953112754936254,
-      "grad_norm": 0.000980622717179358,
-      "learning_rate": 0.00019999999611161582,
-      "loss": 46.0,
-      "step": 1171
-    },
-    {
-      "epoch": 0.08960758453275226,
-      "grad_norm": 0.0015572402626276016,
-      "learning_rate": 0.0001999999961049146,
-      "loss": 46.0,
-      "step": 1172
-    },
-    {
-      "epoch": 0.08968404151614198,
-      "grad_norm": 0.002365921624004841,
-      "learning_rate": 0.0001999999960982076,
-      "loss": 46.0,
-      "step": 1173
-    },
-    {
-      "epoch": 0.0897604984995317,
-      "grad_norm": 0.0024859767872840166,
-      "learning_rate": 0.00019999999609149484,
-      "loss": 46.0,
-      "step": 1174
-    },
-    {
-      "epoch": 0.08983695548292142,
-      "grad_norm": 0.0008226605132222176,
-      "learning_rate": 0.00019999999608477631,
-      "loss": 46.0,
-      "step": 1175
-    },
-    {
-      "epoch": 0.08991341246631114,
-      "grad_norm": 0.0015789471799507737,
-      "learning_rate": 0.000199999996078052,
-      "loss": 46.0,
-      "step": 1176
-    },
-    {
-      "epoch": 0.08998986944970086,
-      "grad_norm": 0.0004363392072264105,
-      "learning_rate": 0.00019999999607132196,
-      "loss": 46.0,
-      "step": 1177
-    },
-    {
-      "epoch": 0.09006632643309058,
-      "grad_norm": 0.0006765234284102917,
-      "learning_rate": 0.0001999999960645861,
-      "loss": 46.0,
-      "step": 1178
-    },
-    {
-      "epoch": 0.0901427834164803,
-      "grad_norm": 0.0006881494191475213,
-      "learning_rate": 0.00019999999605784452,
-      "loss": 46.0,
-      "step": 1179
-    },
-    {
-      "epoch": 0.09021924039987002,
-      "grad_norm": 0.010233285836875439,
-      "learning_rate": 0.00019999999605109715,
-      "loss": 46.0,
-      "step": 1180
-    },
-    {
-      "epoch": 0.09029569738325974,
-      "grad_norm": 0.002100891899317503,
-      "learning_rate": 0.00019999999604434398,
-      "loss": 46.0,
-      "step": 1181
-    },
-    {
-      "epoch": 0.09037215436664946,
-      "grad_norm": 0.000936768192332238,
-      "learning_rate": 0.0001999999960375851,
-      "loss": 46.0,
-      "step": 1182
-    },
-    {
-      "epoch": 0.09044861135003919,
-      "grad_norm": 0.0007137735956348479,
-      "learning_rate": 0.0001999999960308204,
-      "loss": 46.0,
-      "step": 1183
-    },
-    {
-      "epoch": 0.09052506833342891,
-      "grad_norm": 0.0013765707844868302,
-      "learning_rate": 0.00019999999602404993,
-      "loss": 46.0,
-      "step": 1184
-    },
-    {
-      "epoch": 0.09060152531681863,
-      "grad_norm": 0.0021736095659434795,
-      "learning_rate": 0.0001999999960172737,
-      "loss": 46.0,
-      "step": 1185
-    },
-    {
-      "epoch": 0.09067798230020835,
-      "grad_norm": 0.0019218625966459513,
-      "learning_rate": 0.00019999999601049174,
-      "loss": 46.0,
-      "step": 1186
-    },
-    {
-      "epoch": 0.09075443928359807,
-      "grad_norm": 0.0011082632699981332,
-      "learning_rate": 0.00019999999600370396,
-      "loss": 46.0,
-      "step": 1187
-    },
-    {
-      "epoch": 0.09083089626698779,
-      "grad_norm": 0.000510059529915452,
-      "learning_rate": 0.00019999999599691043,
-      "loss": 46.0,
-      "step": 1188
-    },
-    {
-      "epoch": 0.0909073532503775,
-      "grad_norm": 0.00044670479837805033,
-      "learning_rate": 0.00019999999599011113,
-      "loss": 46.0,
-      "step": 1189
-    },
-    {
-      "epoch": 0.09098381023376723,
-      "grad_norm": 0.0015187797835096717,
-      "learning_rate": 0.00019999999598330605,
-      "loss": 46.0,
-      "step": 1190
-    },
-    {
-      "epoch": 0.09106026721715695,
-      "grad_norm": 0.0005870451568625867,
-      "learning_rate": 0.0001999999959764952,
-      "loss": 46.0,
-      "step": 1191
-    },
-    {
-      "epoch": 0.09113672420054666,
-      "grad_norm": 0.0006216135807335377,
-      "learning_rate": 0.0001999999959696786,
-      "loss": 46.0,
-      "step": 1192
-    },
-    {
-      "epoch": 0.09121318118393638,
-      "grad_norm": 0.0009201872744597495,
-      "learning_rate": 0.0001999999959628562,
-      "loss": 46.0,
-      "step": 1193
-    },
-    {
-      "epoch": 0.0912896381673261,
-      "grad_norm": 0.0006811945931985974,
-      "learning_rate": 0.00019999999595602807,
-      "loss": 46.0,
-      "step": 1194
-    },
-    {
-      "epoch": 0.09136609515071582,
-      "grad_norm": 0.001340296003036201,
-      "learning_rate": 0.00019999999594919415,
-      "loss": 46.0,
-      "step": 1195
-    },
-    {
-      "epoch": 0.09144255213410554,
-      "grad_norm": 0.0010067133698612452,
-      "learning_rate": 0.00019999999594235446,
-      "loss": 46.0,
-      "step": 1196
-    },
-    {
-      "epoch": 0.09151900911749528,
-      "grad_norm": 0.0012444059830158949,
-      "learning_rate": 0.000199999995935509,
-      "loss": 46.0,
-      "step": 1197
-    },
-    {
-      "epoch": 0.091595466100885,
-      "grad_norm": 0.00182445312384516,
-      "learning_rate": 0.00019999999592865776,
-      "loss": 46.0,
-      "step": 1198
-    },
-    {
-      "epoch": 0.09167192308427471,
-      "grad_norm": 0.000848781899549067,
-      "learning_rate": 0.00019999999592180078,
-      "loss": 46.0,
-      "step": 1199
-    },
-    {
-      "epoch": 0.09174838006766443,
-      "grad_norm": 0.0014819852076470852,
-      "learning_rate": 0.00019999999591493802,
-      "loss": 46.0,
-      "step": 1200
-    },
-    {
-      "epoch": 0.09182483705105415,
-      "grad_norm": 0.0027086471673101187,
-      "learning_rate": 0.0001999999959080695,
-      "loss": 46.0,
-      "step": 1201
-    },
-    {
-      "epoch": 0.09190129403444387,
-      "grad_norm": 0.0004966605338267982,
-      "learning_rate": 0.0001999999959011952,
-      "loss": 46.0,
-      "step": 1202
-    },
-    {
-      "epoch": 0.09197775101783359,
-      "grad_norm": 0.0004992112517356873,
-      "learning_rate": 0.00019999999589431511,
-      "loss": 46.0,
-      "step": 1203
-    },
-    {
-      "epoch": 0.09205420800122331,
-      "grad_norm": 0.0009974294807761908,
-      "learning_rate": 0.00019999999588742927,
-      "loss": 46.0,
-      "step": 1204
-    },
-    {
-      "epoch": 0.09213066498461303,
-      "grad_norm": 0.0006977037410251796,
-      "learning_rate": 0.00019999999588053764,
-      "loss": 46.0,
-      "step": 1205
-    },
-    {
-      "epoch": 0.09220712196800275,
-      "grad_norm": 0.00046226897393353283,
-      "learning_rate": 0.00019999999587364027,
-      "loss": 46.0,
-      "step": 1206
-    },
-    {
-      "epoch": 0.09228357895139247,
-      "grad_norm": 0.001052415114827454,
-      "learning_rate": 0.00019999999586673713,
-      "loss": 46.0,
-      "step": 1207
-    },
-    {
-      "epoch": 0.09236003593478219,
-      "grad_norm": 0.0013654630165547132,
-      "learning_rate": 0.00019999999585982822,
-      "loss": 46.0,
-      "step": 1208
-    },
-    {
-      "epoch": 0.09243649291817191,
-      "grad_norm": 0.0013527219416573644,
-      "learning_rate": 0.0001999999958529135,
-      "loss": 46.0,
-      "step": 1209
-    },
-    {
-      "epoch": 0.09251294990156163,
-      "grad_norm": 0.0006191693828441203,
-      "learning_rate": 0.00019999999584599304,
-      "loss": 46.0,
-      "step": 1210
-    },
-    {
-      "epoch": 0.09258940688495135,
-      "grad_norm": 0.0010233297944068909,
-      "learning_rate": 0.0001999999958390668,
-      "loss": 46.0,
-      "step": 1211
-    },
-    {
-      "epoch": 0.09266586386834108,
-      "grad_norm": 0.0009534502169117332,
-      "learning_rate": 0.0001999999958321348,
-      "loss": 46.0,
-      "step": 1212
-    },
-    {
-      "epoch": 0.0927423208517308,
-      "grad_norm": 0.001106227864511311,
-      "learning_rate": 0.00019999999582519704,
-      "loss": 46.0,
-      "step": 1213
-    },
-    {
-      "epoch": 0.09281877783512052,
-      "grad_norm": 0.0005697849555872381,
-      "learning_rate": 0.0001999999958182535,
-      "loss": 46.0,
-      "step": 1214
-    },
-    {
-      "epoch": 0.09289523481851024,
-      "grad_norm": 0.0009686643606983125,
-      "learning_rate": 0.0001999999958113042,
-      "loss": 46.0,
-      "step": 1215
-    },
-    {
-      "epoch": 0.09297169180189996,
-      "grad_norm": 0.0023167189210653305,
-      "learning_rate": 0.0001999999958043491,
-      "loss": 46.0,
-      "step": 1216
-    },
-    {
-      "epoch": 0.09304814878528968,
-      "grad_norm": 0.0004048386763315648,
-      "learning_rate": 0.00019999999579738828,
-      "loss": 46.0,
-      "step": 1217
-    },
-    {
-      "epoch": 0.0931246057686794,
-      "grad_norm": 0.0010416668374091387,
-      "learning_rate": 0.00019999999579042166,
-      "loss": 46.0,
-      "step": 1218
-    },
-    {
-      "epoch": 0.09320106275206912,
-      "grad_norm": 0.0005037899245508015,
-      "learning_rate": 0.0001999999957834493,
-      "loss": 46.0,
-      "step": 1219
-    },
-    {
-      "epoch": 0.09327751973545884,
-      "grad_norm": 0.0008396848570555449,
-      "learning_rate": 0.0001999999957764711,
-      "loss": 46.0,
-      "step": 1220
-    },
-    {
-      "epoch": 0.09335397671884856,
-      "grad_norm": 0.0013662183191627264,
-      "learning_rate": 0.00019999999576948718,
-      "loss": 46.0,
-      "step": 1221
-    },
-    {
-      "epoch": 0.09343043370223827,
-      "grad_norm": 0.0013979448704048991,
-      "learning_rate": 0.00019999999576249752,
-      "loss": 46.0,
-      "step": 1222
-    },
-    {
-      "epoch": 0.093506890685628,
-      "grad_norm": 0.002754638670012355,
-      "learning_rate": 0.00019999999575550203,
-      "loss": 46.0,
-      "step": 1223
-    },
-    {
-      "epoch": 0.09358334766901771,
-      "grad_norm": 0.000634212396107614,
-      "learning_rate": 0.00019999999574850082,
-      "loss": 46.0,
-      "step": 1224
-    },
-    {
-      "epoch": 0.09365980465240743,
-      "grad_norm": 0.0014306706143543124,
-      "learning_rate": 0.00019999999574149378,
-      "loss": 46.0,
-      "step": 1225
-    },
-    {
-      "epoch": 0.09373626163579717,
-      "grad_norm": 0.004412438254803419,
-      "learning_rate": 0.00019999999573448103,
-      "loss": 46.0,
-      "step": 1226
-    },
-    {
-      "epoch": 0.09381271861918689,
-      "grad_norm": 0.0008019739761948586,
-      "learning_rate": 0.0001999999957274625,
-      "loss": 46.0,
-      "step": 1227
-    },
-    {
-      "epoch": 0.0938891756025766,
-      "grad_norm": 0.0008827553829178214,
-      "learning_rate": 0.0001999999957204382,
-      "loss": 46.0,
-      "step": 1228
-    },
-    {
-      "epoch": 0.09396563258596632,
-      "grad_norm": 0.0012745364801958203,
-      "learning_rate": 0.0001999999957134081,
-      "loss": 46.0,
-      "step": 1229
-    },
-    {
-      "epoch": 0.09404208956935604,
-      "grad_norm": 0.0017059399979189038,
-      "learning_rate": 0.00019999999570637225,
-      "loss": 46.0,
-      "step": 1230
-    },
-    {
-      "epoch": 0.09411854655274576,
-      "grad_norm": 0.0014126162277534604,
-      "learning_rate": 0.00019999999569933062,
-      "loss": 46.0,
-      "step": 1231
-    },
-    {
-      "epoch": 0.09419500353613548,
-      "grad_norm": 0.000933696108404547,
-      "learning_rate": 0.00019999999569228326,
-      "loss": 46.0,
-      "step": 1232
-    },
-    {
-      "epoch": 0.0942714605195252,
-      "grad_norm": 0.0009252531453967094,
-      "learning_rate": 0.0001999999956852301,
-      "loss": 46.0,
-      "step": 1233
-    },
-    {
-      "epoch": 0.09434791750291492,
-      "grad_norm": 0.0006862206500954926,
-      "learning_rate": 0.00019999999567817117,
-      "loss": 46.0,
-      "step": 1234
-    },
-    {
-      "epoch": 0.09442437448630464,
-      "grad_norm": 0.0011284563224762678,
-      "learning_rate": 0.00019999999567110646,
-      "loss": 46.0,
-      "step": 1235
-    },
-    {
-      "epoch": 0.09450083146969436,
-      "grad_norm": 0.0008051855838857591,
-      "learning_rate": 0.000199999995664036,
-      "loss": 46.0,
-      "step": 1236
-    },
-    {
-      "epoch": 0.09457728845308408,
-      "grad_norm": 0.0008793355664238334,
-      "learning_rate": 0.00019999999565695976,
-      "loss": 46.0,
-      "step": 1237
-    },
-    {
-      "epoch": 0.0946537454364738,
-      "grad_norm": 0.0009020173456519842,
-      "learning_rate": 0.00019999999564987775,
-      "loss": 46.0,
-      "step": 1238
-    },
-    {
-      "epoch": 0.09473020241986352,
-      "grad_norm": 0.0006941471947357059,
-      "learning_rate": 0.00019999999564279,
-      "loss": 46.0,
-      "step": 1239
-    },
-    {
-      "epoch": 0.09480665940325324,
-      "grad_norm": 0.0010112235322594643,
-      "learning_rate": 0.0001999999956356964,
-      "loss": 46.0,
-      "step": 1240
-    },
-    {
-      "epoch": 0.09488311638664297,
-      "grad_norm": 0.0003985882503911853,
-      "learning_rate": 0.00019999999562859714,
-      "loss": 46.0,
-      "step": 1241
-    },
-    {
-      "epoch": 0.09495957337003269,
-      "grad_norm": 0.0010461274068802595,
-      "learning_rate": 0.00019999999562149206,
-      "loss": 46.0,
-      "step": 1242
-    },
-    {
-      "epoch": 0.09503603035342241,
-      "grad_norm": 0.0012383535504341125,
-      "learning_rate": 0.0001999999956143812,
-      "loss": 46.0,
-      "step": 1243
-    },
-    {
-      "epoch": 0.09511248733681213,
-      "grad_norm": 0.0006342925480566919,
-      "learning_rate": 0.00019999999560726457,
-      "loss": 46.0,
-      "step": 1244
-    },
-    {
-      "epoch": 0.09518894432020185,
-      "grad_norm": 0.0017313059652224183,
-      "learning_rate": 0.00019999999560014217,
-      "loss": 46.0,
-      "step": 1245
-    },
-    {
-      "epoch": 0.09526540130359157,
-      "grad_norm": 0.0011314081493765116,
-      "learning_rate": 0.00019999999559301403,
-      "loss": 46.0,
-      "step": 1246
-    },
-    {
-      "epoch": 0.09534185828698129,
-      "grad_norm": 0.0005559470737352967,
-      "learning_rate": 0.0001999999955858801,
-      "loss": 46.0,
-      "step": 1247
-    },
-    {
-      "epoch": 0.095418315270371,
-      "grad_norm": 0.00640692375600338,
-      "learning_rate": 0.00019999999557874037,
-      "loss": 46.0,
-      "step": 1248
-    },
-    {
-      "epoch": 0.09549477225376073,
-      "grad_norm": 0.004177349153906107,
-      "learning_rate": 0.0001999999955715949,
-      "loss": 46.0,
-      "step": 1249
-    },
-    {
-      "epoch": 0.09557122923715045,
-      "grad_norm": 0.0006592542049475014,
-      "learning_rate": 0.00019999999556444368,
-      "loss": 46.0,
-      "step": 1250
-    },
-    {
-      "epoch": 0.09564768622054017,
-      "grad_norm": 0.0027955283876508474,
-      "learning_rate": 0.00019999999555728667,
-      "loss": 46.0,
-      "step": 1251
-    },
-    {
-      "epoch": 0.09572414320392988,
-      "grad_norm": 0.0012236223556101322,
-      "learning_rate": 0.0001999999955501239,
-      "loss": 46.0,
-      "step": 1252
-    },
-    {
-      "epoch": 0.0958006001873196,
-      "grad_norm": 0.0017678553704172373,
-      "learning_rate": 0.00019999999554295536,
-      "loss": 46.0,
-      "step": 1253
-    },
-    {
-      "epoch": 0.09587705717070932,
-      "grad_norm": 0.0006399803096428514,
-      "learning_rate": 0.00019999999553578103,
-      "loss": 46.0,
-      "step": 1254
-    },
-    {
-      "epoch": 0.09595351415409906,
-      "grad_norm": 0.0006326381699182093,
-      "learning_rate": 0.00019999999552860093,
-      "loss": 46.0,
-      "step": 1255
-    },
-    {
-      "epoch": 0.09602997113748878,
-      "grad_norm": 0.0008295346051454544,
-      "learning_rate": 0.00019999999552141508,
-      "loss": 46.0,
-      "step": 1256
-    },
-    {
-      "epoch": 0.0961064281208785,
-      "grad_norm": 0.0018643852090463042,
-      "learning_rate": 0.00019999999551422346,
-      "loss": 46.0,
-      "step": 1257
-    },
-    {
-      "epoch": 0.09618288510426821,
-      "grad_norm": 0.0015572465490549803,
-      "learning_rate": 0.00019999999550702606,
-      "loss": 46.0,
-      "step": 1258
-    },
-    {
-      "epoch": 0.09625934208765793,
-      "grad_norm": 0.005144514609128237,
-      "learning_rate": 0.00019999999549982292,
-      "loss": 46.0,
-      "step": 1259
-    },
-    {
-      "epoch": 0.09633579907104765,
-      "grad_norm": 0.0018445689929649234,
-      "learning_rate": 0.00019999999549261398,
-      "loss": 46.0,
-      "step": 1260
-    },
-    {
-      "epoch": 0.09641225605443737,
-      "grad_norm": 0.003999857697635889,
-      "learning_rate": 0.00019999999548539927,
-      "loss": 46.0,
-      "step": 1261
-    },
-    {
-      "epoch": 0.09648871303782709,
-      "grad_norm": 0.0009839792037382722,
-      "learning_rate": 0.0001999999954781788,
-      "loss": 46.0,
-      "step": 1262
-    },
-    {
-      "epoch": 0.09656517002121681,
-      "grad_norm": 0.0015999675961211324,
-      "learning_rate": 0.00019999999547095257,
-      "loss": 46.0,
-      "step": 1263
-    },
-    {
-      "epoch": 0.09664162700460653,
-      "grad_norm": 0.0011857443023473024,
-      "learning_rate": 0.00019999999546372056,
-      "loss": 46.0,
-      "step": 1264
-    },
-    {
-      "epoch": 0.09671808398799625,
-      "grad_norm": 0.0004027715476695448,
-      "learning_rate": 0.00019999999545648275,
-      "loss": 46.0,
-      "step": 1265
-    },
-    {
-      "epoch": 0.09679454097138597,
-      "grad_norm": 0.001518356497399509,
-      "learning_rate": 0.00019999999544923923,
-      "loss": 46.0,
-      "step": 1266
-    },
-    {
-      "epoch": 0.09687099795477569,
-      "grad_norm": 0.0010775034315884113,
-      "learning_rate": 0.0001999999954419899,
-      "loss": 46.0,
-      "step": 1267
-    },
-    {
-      "epoch": 0.09694745493816541,
-      "grad_norm": 0.0009000435820780694,
-      "learning_rate": 0.00019999999543473482,
-      "loss": 46.0,
-      "step": 1268
-    },
-    {
-      "epoch": 0.09702391192155513,
-      "grad_norm": 0.0011318640317767859,
-      "learning_rate": 0.00019999999542747395,
-      "loss": 46.0,
-      "step": 1269
-    },
-    {
-      "epoch": 0.09710036890494486,
-      "grad_norm": 0.0022418838925659657,
-      "learning_rate": 0.00019999999542020733,
-      "loss": 46.0,
-      "step": 1270
-    },
-    {
-      "epoch": 0.09717682588833458,
-      "grad_norm": 0.0009755489300005138,
-      "learning_rate": 0.00019999999541293493,
-      "loss": 46.0,
-      "step": 1271
-    },
-    {
-      "epoch": 0.0972532828717243,
-      "grad_norm": 0.0011484373826533556,
-      "learning_rate": 0.00019999999540565677,
-      "loss": 46.0,
-      "step": 1272
-    },
-    {
-      "epoch": 0.09732973985511402,
-      "grad_norm": 0.0031784288585186005,
-      "learning_rate": 0.00019999999539837285,
-      "loss": 46.0,
-      "step": 1273
-    },
-    {
-      "epoch": 0.09740619683850374,
-      "grad_norm": 0.010791708715260029,
-      "learning_rate": 0.0001999999953910831,
-      "loss": 46.0,
-      "step": 1274
-    },
-    {
-      "epoch": 0.09748265382189346,
-      "grad_norm": 0.001676803221926093,
-      "learning_rate": 0.00019999999538378768,
-      "loss": 46.0,
-      "step": 1275
-    },
-    {
-      "epoch": 0.09755911080528318,
-      "grad_norm": 0.00092638551723212,
-      "learning_rate": 0.0001999999953764864,
-      "loss": 46.0,
-      "step": 1276
-    },
-    {
-      "epoch": 0.0976355677886729,
-      "grad_norm": 0.0006048434879630804,
-      "learning_rate": 0.00019999999536917938,
-      "loss": 46.0,
-      "step": 1277
-    },
-    {
-      "epoch": 0.09771202477206262,
-      "grad_norm": 0.0007846732623875141,
-      "learning_rate": 0.00019999999536186663,
-      "loss": 46.0,
-      "step": 1278
-    },
-    {
-      "epoch": 0.09778848175545234,
-      "grad_norm": 0.0016191190807148814,
-      "learning_rate": 0.00019999999535454804,
-      "loss": 46.0,
-      "step": 1279
-    },
-    {
-      "epoch": 0.09786493873884206,
-      "grad_norm": 0.0004936332697980106,
-      "learning_rate": 0.00019999999534722374,
-      "loss": 46.0,
-      "step": 1280
-    },
-    {
-      "epoch": 0.09794139572223177,
-      "grad_norm": 0.002149005653336644,
-      "learning_rate": 0.00019999999533989364,
-      "loss": 46.0,
-      "step": 1281
-    },
-    {
-      "epoch": 0.0980178527056215,
-      "grad_norm": 0.0007897565956227481,
-      "learning_rate": 0.0001999999953325578,
-      "loss": 46.0,
-      "step": 1282
-    },
-    {
-      "epoch": 0.09809430968901121,
-      "grad_norm": 0.0018300628289580345,
-      "learning_rate": 0.00019999999532521615,
-      "loss": 46.0,
-      "step": 1283
-    },
-    {
-      "epoch": 0.09817076667240095,
-      "grad_norm": 0.0013459909241646528,
-      "learning_rate": 0.00019999999531786875,
-      "loss": 46.0,
-      "step": 1284
-    },
-    {
-      "epoch": 0.09824722365579067,
-      "grad_norm": 0.0019359695725142956,
-      "learning_rate": 0.00019999999531051559,
-      "loss": 46.0,
-      "step": 1285
-    },
-    {
-      "epoch": 0.09832368063918039,
-      "grad_norm": 0.0015307387802749872,
-      "learning_rate": 0.00019999999530315665,
-      "loss": 46.0,
-      "step": 1286
-    },
-    {
-      "epoch": 0.0984001376225701,
-      "grad_norm": 0.0029983101412653923,
-      "learning_rate": 0.00019999999529579193,
-      "loss": 46.0,
-      "step": 1287
-    },
-    {
-      "epoch": 0.09847659460595982,
-      "grad_norm": 0.0018178214086219668,
-      "learning_rate": 0.00019999999528842147,
-      "loss": 46.0,
-      "step": 1288
-    },
-    {
-      "epoch": 0.09855305158934954,
-      "grad_norm": 0.0004463070072233677,
-      "learning_rate": 0.00019999999528104524,
-      "loss": 46.0,
-      "step": 1289
-    },
-    {
-      "epoch": 0.09862950857273926,
-      "grad_norm": 0.0013351240195333958,
-      "learning_rate": 0.0001999999952736632,
-      "loss": 46.0,
-      "step": 1290
-    },
-    {
-      "epoch": 0.09870596555612898,
-      "grad_norm": 0.0009454942191950977,
-      "learning_rate": 0.0001999999952662754,
-      "loss": 46.0,
-      "step": 1291
-    },
-    {
-      "epoch": 0.0987824225395187,
-      "grad_norm": 0.0004892030847258866,
-      "learning_rate": 0.00019999999525888184,
-      "loss": 46.0,
-      "step": 1292
-    },
-    {
-      "epoch": 0.09885887952290842,
-      "grad_norm": 0.000888179347384721,
-      "learning_rate": 0.0001999999952514825,
-      "loss": 46.0,
-      "step": 1293
-    },
-    {
-      "epoch": 0.09893533650629814,
-      "grad_norm": 0.0007274766685441136,
-      "learning_rate": 0.0001999999952440774,
-      "loss": 46.0,
-      "step": 1294
-    },
-    {
-      "epoch": 0.09901179348968786,
-      "grad_norm": 0.0005824375548399985,
-      "learning_rate": 0.00019999999523666656,
-      "loss": 46.0,
-      "step": 1295
-    },
-    {
-      "epoch": 0.09908825047307758,
-      "grad_norm": 0.001742347958497703,
-      "learning_rate": 0.00019999999522924994,
-      "loss": 46.0,
-      "step": 1296
-    },
-    {
-      "epoch": 0.0991647074564673,
-      "grad_norm": 0.0010265177115797997,
-      "learning_rate": 0.00019999999522182752,
-      "loss": 46.0,
-      "step": 1297
-    },
-    {
-      "epoch": 0.09924116443985702,
-      "grad_norm": 0.0022620977833867073,
-      "learning_rate": 0.00019999999521439936,
-      "loss": 46.0,
-      "step": 1298
-    },
-    {
-      "epoch": 0.09931762142324675,
-      "grad_norm": 0.001948086079210043,
-      "learning_rate": 0.00019999999520696542,
-      "loss": 46.0,
-      "step": 1299
-    },
-    {
-      "epoch": 0.09939407840663647,
-      "grad_norm": 0.003251249436289072,
-      "learning_rate": 0.00019999999519952573,
-      "loss": 46.0,
-      "step": 1300
-    },
-    {
-      "epoch": 0.09947053539002619,
-      "grad_norm": 0.002538104774430394,
-      "learning_rate": 0.00019999999519208024,
-      "loss": 46.0,
-      "step": 1301
-    },
-    {
-      "epoch": 0.09954699237341591,
-      "grad_norm": 0.0008535406668670475,
-      "learning_rate": 0.00019999999518462898,
-      "loss": 46.0,
-      "step": 1302
-    },
-    {
-      "epoch": 0.09962344935680563,
-      "grad_norm": 0.0009606928797438741,
-      "learning_rate": 0.00019999999517717197,
-      "loss": 46.0,
-      "step": 1303
-    },
-    {
-      "epoch": 0.09969990634019535,
-      "grad_norm": 0.0006648574490100145,
-      "learning_rate": 0.00019999999516970917,
-      "loss": 46.0,
-      "step": 1304
-    },
-    {
-      "epoch": 0.09977636332358507,
-      "grad_norm": 0.001308361068367958,
-      "learning_rate": 0.0001999999951622406,
-      "loss": 46.0,
-      "step": 1305
-    },
-    {
-      "epoch": 0.09985282030697479,
-      "grad_norm": 0.0016028402606025338,
-      "learning_rate": 0.00019999999515476628,
-      "loss": 46.0,
-      "step": 1306
-    },
-    {
-      "epoch": 0.09992927729036451,
-      "grad_norm": 0.0020908236037939787,
-      "learning_rate": 0.00019999999514728618,
-      "loss": 46.0,
-      "step": 1307
-    },
-    {
-      "epoch": 0.10000573427375423,
-      "grad_norm": 0.0009625316015444696,
-      "learning_rate": 0.0001999999951398003,
-      "loss": 46.0,
-      "step": 1308
-    },
-    {
-      "epoch": 0.10008219125714395,
-      "grad_norm": 0.0007036154274828732,
-      "learning_rate": 0.0001999999951323087,
-      "loss": 46.0,
-      "step": 1309
-    },
-    {
-      "epoch": 0.10015864824053367,
-      "grad_norm": 0.00224491860717535,
-      "learning_rate": 0.00019999999512481127,
-      "loss": 46.0,
-      "step": 1310
-    },
-    {
-      "epoch": 0.10023510522392338,
-      "grad_norm": 0.0017687119543552399,
-      "learning_rate": 0.0001999999951173081,
-      "loss": 46.0,
-      "step": 1311
-    },
-    {
-      "epoch": 0.1003115622073131,
-      "grad_norm": 0.000878228573128581,
-      "learning_rate": 0.00019999999510979916,
-      "loss": 46.0,
-      "step": 1312
-    },
-    {
-      "epoch": 0.10038801919070284,
-      "grad_norm": 0.0011639725416898727,
-      "learning_rate": 0.00019999999510228445,
-      "loss": 46.0,
-      "step": 1313
-    },
-    {
-      "epoch": 0.10046447617409256,
-      "grad_norm": 0.0017363595543429255,
-      "learning_rate": 0.00019999999509476396,
-      "loss": 46.0,
-      "step": 1314
-    },
-    {
-      "epoch": 0.10054093315748228,
-      "grad_norm": 0.0003351522027514875,
-      "learning_rate": 0.00019999999508723773,
-      "loss": 46.0,
-      "step": 1315
-    },
-    {
-      "epoch": 0.100617390140872,
-      "grad_norm": 0.0007463704678229988,
-      "learning_rate": 0.0001999999950797057,
-      "loss": 46.0,
-      "step": 1316
-    },
-    {
-      "epoch": 0.10069384712426172,
-      "grad_norm": 0.0011413536267355084,
-      "learning_rate": 0.00019999999507216791,
-      "loss": 46.0,
-      "step": 1317
-    },
-    {
-      "epoch": 0.10077030410765143,
-      "grad_norm": 0.001504633342847228,
-      "learning_rate": 0.00019999999506462433,
-      "loss": 46.0,
-      "step": 1318
-    },
-    {
-      "epoch": 0.10084676109104115,
-      "grad_norm": 0.0015678256750106812,
-      "learning_rate": 0.000199999995057075,
-      "loss": 46.0,
-      "step": 1319
-    },
-    {
-      "epoch": 0.10092321807443087,
-      "grad_norm": 0.0018599717877805233,
-      "learning_rate": 0.0001999999950495199,
-      "loss": 46.0,
-      "step": 1320
-    },
-    {
-      "epoch": 0.10099967505782059,
-      "grad_norm": 0.0007657839450985193,
-      "learning_rate": 0.00019999999504195903,
-      "loss": 46.0,
-      "step": 1321
-    },
-    {
-      "epoch": 0.10107613204121031,
-      "grad_norm": 0.00315374624915421,
-      "learning_rate": 0.00019999999503439239,
-      "loss": 46.0,
-      "step": 1322
-    },
-    {
-      "epoch": 0.10115258902460003,
-      "grad_norm": 0.0013712298823520541,
-      "learning_rate": 0.00019999999502682,
-      "loss": 46.0,
-      "step": 1323
-    },
-    {
-      "epoch": 0.10122904600798975,
-      "grad_norm": 0.0013697234680876136,
-      "learning_rate": 0.0001999999950192418,
-      "loss": 46.0,
-      "step": 1324
-    },
-    {
-      "epoch": 0.10130550299137947,
-      "grad_norm": 0.00043646697304211557,
-      "learning_rate": 0.00019999999501165786,
-      "loss": 46.0,
-      "step": 1325
-    },
-    {
-      "epoch": 0.10138195997476919,
-      "grad_norm": 0.0012329216115176678,
-      "learning_rate": 0.00019999999500406815,
-      "loss": 46.0,
-      "step": 1326
-    },
-    {
-      "epoch": 0.10145841695815891,
-      "grad_norm": 0.001268675783649087,
-      "learning_rate": 0.00019999999499647266,
-      "loss": 46.0,
-      "step": 1327
-    },
-    {
-      "epoch": 0.10153487394154864,
-      "grad_norm": 0.0014134745579212904,
-      "learning_rate": 0.0001999999949888714,
-      "loss": 46.0,
-      "step": 1328
-    },
-    {
-      "epoch": 0.10161133092493836,
-      "grad_norm": 0.0012230125721544027,
-      "learning_rate": 0.0001999999949812644,
-      "loss": 46.0,
-      "step": 1329
-    },
-    {
-      "epoch": 0.10168778790832808,
-      "grad_norm": 0.0012176332529634237,
-      "learning_rate": 0.00019999999497365159,
-      "loss": 46.0,
-      "step": 1330
-    },
-    {
-      "epoch": 0.1017642448917178,
-      "grad_norm": 0.0008769358391873538,
-      "learning_rate": 0.00019999999496603303,
-      "loss": 46.0,
-      "step": 1331
-    },
-    {
-      "epoch": 0.10184070187510752,
-      "grad_norm": 0.000845807371661067,
-      "learning_rate": 0.00019999999495840868,
-      "loss": 46.0,
-      "step": 1332
-    },
-    {
-      "epoch": 0.10191715885849724,
-      "grad_norm": 0.0010382183827459812,
-      "learning_rate": 0.00019999999495077858,
-      "loss": 46.0,
-      "step": 1333
-    },
-    {
-      "epoch": 0.10199361584188696,
-      "grad_norm": 0.002728296210989356,
-      "learning_rate": 0.00019999999494314273,
-      "loss": 46.0,
-      "step": 1334
-    },
-    {
-      "epoch": 0.10207007282527668,
-      "grad_norm": 0.005204348359256983,
-      "learning_rate": 0.00019999999493550106,
-      "loss": 46.0,
-      "step": 1335
-    },
-    {
-      "epoch": 0.1021465298086664,
-      "grad_norm": 0.0008558625122532248,
-      "learning_rate": 0.00019999999492785367,
-      "loss": 46.0,
-      "step": 1336
-    },
-    {
-      "epoch": 0.10222298679205612,
-      "grad_norm": 0.0006963508203625679,
-      "learning_rate": 0.00019999999492020047,
-      "loss": 46.0,
-      "step": 1337
-    },
-    {
-      "epoch": 0.10229944377544584,
-      "grad_norm": 0.0006648279377259314,
-      "learning_rate": 0.0001999999949125415,
-      "loss": 46.0,
-      "step": 1338
-    },
-    {
-      "epoch": 0.10237590075883556,
-      "grad_norm": 0.0010026381351053715,
-      "learning_rate": 0.0001999999949048768,
-      "loss": 46.0,
-      "step": 1339
-    },
-    {
-      "epoch": 0.10245235774222528,
-      "grad_norm": 0.00200001266784966,
-      "learning_rate": 0.0001999999948972063,
-      "loss": 46.0,
-      "step": 1340
-    },
-    {
-      "epoch": 0.102528814725615,
-      "grad_norm": 0.0011981796706095338,
-      "learning_rate": 0.00019999999488953005,
-      "loss": 46.0,
-      "step": 1341
-    },
-    {
-      "epoch": 0.10260527170900473,
-      "grad_norm": 0.0011386389378458261,
-      "learning_rate": 0.00019999999488184801,
-      "loss": 46.0,
-      "step": 1342
-    },
-    {
-      "epoch": 0.10268172869239445,
-      "grad_norm": 0.0009121635230258107,
-      "learning_rate": 0.00019999999487416023,
-      "loss": 46.0,
-      "step": 1343
-    },
-    {
-      "epoch": 0.10275818567578417,
-      "grad_norm": 0.0010379153536632657,
-      "learning_rate": 0.00019999999486646668,
-      "loss": 46.0,
-      "step": 1344
-    },
-    {
-      "epoch": 0.10283464265917389,
-      "grad_norm": 0.0006629675044678152,
-      "learning_rate": 0.00019999999485876733,
-      "loss": 46.0,
-      "step": 1345
-    },
-    {
-      "epoch": 0.1029110996425636,
-      "grad_norm": 0.0009854877134785056,
-      "learning_rate": 0.00019999999485106223,
-      "loss": 46.0,
-      "step": 1346
-    },
-    {
-      "epoch": 0.10298755662595332,
-      "grad_norm": 0.00296715646982193,
-      "learning_rate": 0.00019999999484335136,
-      "loss": 46.0,
-      "step": 1347
-    },
-    {
-      "epoch": 0.10306401360934304,
-      "grad_norm": 0.0007912347791716456,
-      "learning_rate": 0.0001999999948356347,
-      "loss": 46.0,
-      "step": 1348
-    },
-    {
-      "epoch": 0.10314047059273276,
-      "grad_norm": 0.0035866890102624893,
-      "learning_rate": 0.00019999999482791226,
-      "loss": 46.0,
-      "step": 1349
-    },
-    {
-      "epoch": 0.10321692757612248,
-      "grad_norm": 0.001526356441900134,
-      "learning_rate": 0.0001999999948201841,
-      "loss": 46.0,
-      "step": 1350
-    },
-    {
-      "epoch": 0.1032933845595122,
-      "grad_norm": 0.0008646687492728233,
-      "learning_rate": 0.00019999999481245016,
-      "loss": 46.0,
-      "step": 1351
-    },
-    {
-      "epoch": 0.10336984154290192,
-      "grad_norm": 0.0016209310851991177,
-      "learning_rate": 0.00019999999480471044,
-      "loss": 46.0,
-      "step": 1352
-    },
-    {
-      "epoch": 0.10344629852629164,
-      "grad_norm": 0.0020330268889665604,
-      "learning_rate": 0.00019999999479696493,
-      "loss": 46.0,
-      "step": 1353
-    },
-    {
-      "epoch": 0.10352275550968136,
-      "grad_norm": 0.0007184363785199821,
-      "learning_rate": 0.00019999999478921367,
-      "loss": 46.0,
-      "step": 1354
-    },
-    {
-      "epoch": 0.10359921249307108,
-      "grad_norm": 0.0011526669841259718,
-      "learning_rate": 0.00019999999478145664,
-      "loss": 46.0,
-      "step": 1355
-    },
-    {
-      "epoch": 0.1036756694764608,
-      "grad_norm": 0.000992448185570538,
-      "learning_rate": 0.00019999999477369383,
-      "loss": 46.0,
-      "step": 1356
-    },
-    {
-      "epoch": 0.10375212645985053,
-      "grad_norm": 0.0019610212184488773,
-      "learning_rate": 0.00019999999476592528,
-      "loss": 46.0,
-      "step": 1357
-    },
-    {
-      "epoch": 0.10382858344324025,
-      "grad_norm": 0.008228207938373089,
-      "learning_rate": 0.00019999999475815093,
-      "loss": 46.0,
-      "step": 1358
-    },
-    {
-      "epoch": 0.10390504042662997,
-      "grad_norm": 0.0012455694377422333,
-      "learning_rate": 0.00019999999475037083,
-      "loss": 46.0,
-      "step": 1359
-    },
-    {
-      "epoch": 0.10398149741001969,
-      "grad_norm": 0.0009230130817741156,
-      "learning_rate": 0.00019999999474258496,
-      "loss": 46.0,
-      "step": 1360
-    },
-    {
-      "epoch": 0.10405795439340941,
-      "grad_norm": 0.0013253414072096348,
-      "learning_rate": 0.00019999999473479328,
-      "loss": 46.0,
-      "step": 1361
-    },
-    {
-      "epoch": 0.10413441137679913,
-      "grad_norm": 0.0012087085051462054,
-      "learning_rate": 0.0001999999947269959,
-      "loss": 46.0,
-      "step": 1362
-    },
-    {
-      "epoch": 0.10421086836018885,
-      "grad_norm": 0.00046159615158103406,
-      "learning_rate": 0.0001999999947191927,
-      "loss": 46.0,
-      "step": 1363
-    },
-    {
-      "epoch": 0.10428732534357857,
-      "grad_norm": 0.0009205068345181644,
-      "learning_rate": 0.00019999999471138376,
-      "loss": 46.0,
-      "step": 1364
-    },
-    {
-      "epoch": 0.10436378232696829,
-      "grad_norm": 0.0014563918812200427,
-      "learning_rate": 0.00019999999470356902,
-      "loss": 46.0,
-      "step": 1365
-    },
-    {
-      "epoch": 0.10444023931035801,
-      "grad_norm": 0.0005110208876430988,
-      "learning_rate": 0.00019999999469574853,
-      "loss": 46.0,
-      "step": 1366
-    },
-    {
-      "epoch": 0.10451669629374773,
-      "grad_norm": 0.0013468445977196097,
-      "learning_rate": 0.00019999999468792225,
-      "loss": 46.0,
-      "step": 1367
-    },
-    {
-      "epoch": 0.10459315327713745,
-      "grad_norm": 0.0008579662535339594,
-      "learning_rate": 0.00019999999468009021,
-      "loss": 46.0,
-      "step": 1368
-    },
-    {
-      "epoch": 0.10466961026052717,
-      "grad_norm": 0.0013018710305914283,
-      "learning_rate": 0.00019999999467225243,
-      "loss": 46.0,
-      "step": 1369
-    },
-    {
-      "epoch": 0.10474606724391688,
-      "grad_norm": 0.0007715172250755131,
-      "learning_rate": 0.00019999999466440883,
-      "loss": 46.0,
-      "step": 1370
-    },
-    {
-      "epoch": 0.10482252422730662,
-      "grad_norm": 0.0006521688774228096,
-      "learning_rate": 0.00019999999465655953,
-      "loss": 46.0,
-      "step": 1371
-    },
-    {
-      "epoch": 0.10489898121069634,
-      "grad_norm": 0.002342222724109888,
-      "learning_rate": 0.00019999999464870438,
-      "loss": 46.0,
-      "step": 1372
-    },
-    {
-      "epoch": 0.10497543819408606,
-      "grad_norm": 0.0009031341760419309,
-      "learning_rate": 0.0001999999946408435,
-      "loss": 46.0,
-      "step": 1373
-    },
-    {
-      "epoch": 0.10505189517747578,
-      "grad_norm": 0.0006895132246427238,
-      "learning_rate": 0.00019999999463297689,
-      "loss": 46.0,
-      "step": 1374
-    },
-    {
-      "epoch": 0.1051283521608655,
-      "grad_norm": 0.0007181079126894474,
-      "learning_rate": 0.00019999999462510444,
-      "loss": 46.0,
-      "step": 1375
-    },
-    {
-      "epoch": 0.10520480914425522,
-      "grad_norm": 0.0008163030724972486,
-      "learning_rate": 0.00019999999461722625,
-      "loss": 46.0,
-      "step": 1376
-    },
-    {
-      "epoch": 0.10528126612764493,
-      "grad_norm": 0.0004917309852316976,
-      "learning_rate": 0.00019999999460934228,
-      "loss": 46.0,
-      "step": 1377
-    },
-    {
-      "epoch": 0.10535772311103465,
-      "grad_norm": 0.001014159875921905,
-      "learning_rate": 0.00019999999460145257,
-      "loss": 46.0,
-      "step": 1378
-    },
-    {
-      "epoch": 0.10543418009442437,
-      "grad_norm": 0.0010242253774777055,
-      "learning_rate": 0.00019999999459355708,
-      "loss": 46.0,
-      "step": 1379
-    },
-    {
-      "epoch": 0.10551063707781409,
-      "grad_norm": 0.0006272322498261929,
-      "learning_rate": 0.0001999999945856558,
-      "loss": 46.0,
-      "step": 1380
-    },
-    {
-      "epoch": 0.10558709406120381,
-      "grad_norm": 0.0014750051777809858,
-      "learning_rate": 0.00019999999457774877,
-      "loss": 46.0,
-      "step": 1381
-    },
-    {
-      "epoch": 0.10566355104459353,
-      "grad_norm": 0.0006043561734259129,
-      "learning_rate": 0.00019999999456983596,
-      "loss": 46.0,
-      "step": 1382
-    },
-    {
-      "epoch": 0.10574000802798325,
-      "grad_norm": 0.0010625538416206837,
-      "learning_rate": 0.00019999999456191738,
-      "loss": 46.0,
-      "step": 1383
-    },
-    {
-      "epoch": 0.10581646501137297,
-      "grad_norm": 0.004903338849544525,
-      "learning_rate": 0.00019999999455399303,
-      "loss": 46.0,
-      "step": 1384
-    },
-    {
-      "epoch": 0.10589292199476269,
-      "grad_norm": 0.0020839418284595013,
-      "learning_rate": 0.00019999999454606293,
-      "loss": 46.0,
-      "step": 1385
-    },
-    {
-      "epoch": 0.10596937897815242,
-      "grad_norm": 0.0008051164913922548,
-      "learning_rate": 0.00019999999453812703,
-      "loss": 46.0,
-      "step": 1386
-    },
-    {
-      "epoch": 0.10604583596154214,
-      "grad_norm": 0.0009010107605718076,
-      "learning_rate": 0.00019999999453018539,
-      "loss": 46.0,
-      "step": 1387
-    },
-    {
-      "epoch": 0.10612229294493186,
-      "grad_norm": 0.0011650670785456896,
-      "learning_rate": 0.00019999999452223794,
-      "loss": 46.0,
-      "step": 1388
-    },
-    {
-      "epoch": 0.10619874992832158,
-      "grad_norm": 0.0010050187120214105,
-      "learning_rate": 0.00019999999451428477,
-      "loss": 46.0,
-      "step": 1389
-    },
-    {
-      "epoch": 0.1062752069117113,
-      "grad_norm": 0.000857661827467382,
-      "learning_rate": 0.0001999999945063258,
-      "loss": 46.0,
-      "step": 1390
-    },
-    {
-      "epoch": 0.10635166389510102,
-      "grad_norm": 0.0009748273878358305,
-      "learning_rate": 0.0001999999944983611,
-      "loss": 46.0,
-      "step": 1391
-    },
-    {
-      "epoch": 0.10642812087849074,
-      "grad_norm": 0.0028728360775858164,
-      "learning_rate": 0.00019999999449039059,
-      "loss": 46.0,
-      "step": 1392
-    },
-    {
-      "epoch": 0.10650457786188046,
-      "grad_norm": 0.0011886555003002286,
-      "learning_rate": 0.0001999999944824143,
-      "loss": 46.0,
-      "step": 1393
-    },
-    {
-      "epoch": 0.10658103484527018,
-      "grad_norm": 0.0017759384354576468,
-      "learning_rate": 0.00019999999447443227,
-      "loss": 46.0,
-      "step": 1394
-    },
-    {
-      "epoch": 0.1066574918286599,
-      "grad_norm": 0.000757109432015568,
-      "learning_rate": 0.00019999999446644444,
-      "loss": 46.0,
-      "step": 1395
-    },
-    {
-      "epoch": 0.10673394881204962,
-      "grad_norm": 0.00039343966636806726,
-      "learning_rate": 0.00019999999445845088,
-      "loss": 46.0,
-      "step": 1396
-    },
-    {
-      "epoch": 0.10681040579543934,
-      "grad_norm": 0.0009081176249310374,
-      "learning_rate": 0.00019999999445045153,
-      "loss": 46.0,
-      "step": 1397
-    },
-    {
-      "epoch": 0.10688686277882906,
-      "grad_norm": 0.0014924656134098768,
-      "learning_rate": 0.0001999999944424464,
-      "loss": 46.0,
-      "step": 1398
-    },
-    {
-      "epoch": 0.10696331976221878,
-      "grad_norm": 0.0026354840956628323,
-      "learning_rate": 0.0001999999944344355,
-      "loss": 46.0,
-      "step": 1399
-    },
-    {
-      "epoch": 0.10703977674560851,
-      "grad_norm": 0.00043676828499883413,
-      "learning_rate": 0.00019999999442641886,
-      "loss": 46.0,
-      "step": 1400
-    },
-    {
-      "epoch": 0.10711623372899823,
-      "grad_norm": 0.0005772305885329843,
-      "learning_rate": 0.00019999999441839644,
-      "loss": 46.0,
-      "step": 1401
-    },
-    {
-      "epoch": 0.10719269071238795,
-      "grad_norm": 0.0016556215705350041,
-      "learning_rate": 0.00019999999441036822,
-      "loss": 46.0,
-      "step": 1402
-    },
-    {
-      "epoch": 0.10726914769577767,
-      "grad_norm": 0.0012777734082192183,
-      "learning_rate": 0.0001999999944023343,
-      "loss": 46.0,
-      "step": 1403
-    },
-    {
-      "epoch": 0.10734560467916739,
-      "grad_norm": 0.0009159351466223598,
-      "learning_rate": 0.00019999999439429455,
-      "loss": 46.0,
-      "step": 1404
-    },
-    {
-      "epoch": 0.1074220616625571,
-      "grad_norm": 0.0022170941811054945,
-      "learning_rate": 0.00019999999438624904,
-      "loss": 46.0,
-      "step": 1405
-    },
-    {
-      "epoch": 0.10749851864594683,
-      "grad_norm": 0.0008294574217870831,
-      "learning_rate": 0.00019999999437819778,
-      "loss": 46.0,
-      "step": 1406
-    },
-    {
-      "epoch": 0.10757497562933654,
-      "grad_norm": 0.0010854746215045452,
-      "learning_rate": 0.00019999999437014072,
-      "loss": 46.0,
-      "step": 1407
-    },
-    {
-      "epoch": 0.10765143261272626,
-      "grad_norm": 0.0009142841445282102,
-      "learning_rate": 0.00019999999436207791,
-      "loss": 46.0,
-      "step": 1408
-    },
-    {
-      "epoch": 0.10772788959611598,
-      "grad_norm": 0.0007848692475818098,
-      "learning_rate": 0.00019999999435400934,
-      "loss": 46.0,
-      "step": 1409
-    },
-    {
-      "epoch": 0.1078043465795057,
-      "grad_norm": 0.0013571399031206965,
-      "learning_rate": 0.00019999999434593498,
-      "loss": 46.0,
-      "step": 1410
-    },
-    {
-      "epoch": 0.10788080356289542,
-      "grad_norm": 0.0006752159097231925,
-      "learning_rate": 0.00019999999433785486,
-      "loss": 46.0,
-      "step": 1411
-    },
-    {
-      "epoch": 0.10795726054628514,
-      "grad_norm": 0.0006309831514954567,
-      "learning_rate": 0.00019999999432976896,
-      "loss": 46.0,
-      "step": 1412
-    },
-    {
-      "epoch": 0.10803371752967486,
-      "grad_norm": 0.0014313762076199055,
-      "learning_rate": 0.00019999999432167732,
-      "loss": 46.0,
-      "step": 1413
-    },
-    {
-      "epoch": 0.10811017451306458,
-      "grad_norm": 0.0020970685873180628,
-      "learning_rate": 0.00019999999431357987,
-      "loss": 46.0,
-      "step": 1414
-    },
-    {
-      "epoch": 0.10818663149645431,
-      "grad_norm": 0.001207476481795311,
-      "learning_rate": 0.00019999999430547668,
-      "loss": 46.0,
-      "step": 1415
-    },
-    {
-      "epoch": 0.10826308847984403,
-      "grad_norm": 0.0011039258679375052,
-      "learning_rate": 0.00019999999429736774,
-      "loss": 46.0,
-      "step": 1416
-    },
-    {
-      "epoch": 0.10833954546323375,
-      "grad_norm": 0.0011299074394628406,
-      "learning_rate": 0.00019999999428925298,
-      "loss": 46.0,
-      "step": 1417
-    },
-    {
-      "epoch": 0.10841600244662347,
-      "grad_norm": 0.003093366278335452,
-      "learning_rate": 0.00019999999428113246,
-      "loss": 46.0,
-      "step": 1418
-    },
-    {
-      "epoch": 0.10849245943001319,
-      "grad_norm": 0.0013180439127609134,
-      "learning_rate": 0.0001999999942730062,
-      "loss": 46.0,
-      "step": 1419
-    },
-    {
-      "epoch": 0.10856891641340291,
-      "grad_norm": 0.002898380160331726,
-      "learning_rate": 0.00019999999426487418,
-      "loss": 46.0,
-      "step": 1420
-    },
-    {
-      "epoch": 0.10864537339679263,
-      "grad_norm": 0.001105992472730577,
-      "learning_rate": 0.00019999999425673632,
-      "loss": 46.0,
-      "step": 1421
-    },
-    {
-      "epoch": 0.10872183038018235,
-      "grad_norm": 0.0031519923359155655,
-      "learning_rate": 0.00019999999424859274,
-      "loss": 46.0,
-      "step": 1422
-    },
-    {
-      "epoch": 0.10879828736357207,
-      "grad_norm": 0.0011741496855393052,
-      "learning_rate": 0.0001999999942404434,
-      "loss": 46.0,
-      "step": 1423
-    },
-    {
-      "epoch": 0.10887474434696179,
-      "grad_norm": 0.001801345613785088,
-      "learning_rate": 0.0001999999942322883,
-      "loss": 46.0,
-      "step": 1424
-    },
-    {
-      "epoch": 0.10895120133035151,
-      "grad_norm": 0.0026288768276572227,
-      "learning_rate": 0.0001999999942241274,
-      "loss": 46.0,
-      "step": 1425
-    },
-    {
-      "epoch": 0.10902765831374123,
-      "grad_norm": 0.0019972689915448427,
-      "learning_rate": 0.00019999999421596072,
-      "loss": 46.0,
-      "step": 1426
-    },
-    {
-      "epoch": 0.10910411529713095,
-      "grad_norm": 0.001267926418222487,
-      "learning_rate": 0.00019999999420778827,
-      "loss": 46.0,
-      "step": 1427
-    },
-    {
-      "epoch": 0.10918057228052067,
-      "grad_norm": 0.0011345412349328399,
-      "learning_rate": 0.00019999999419961008,
-      "loss": 46.0,
-      "step": 1428
-    },
-    {
-      "epoch": 0.1092570292639104,
-      "grad_norm": 0.000510324549395591,
-      "learning_rate": 0.00019999999419142612,
-      "loss": 46.0,
-      "step": 1429
-    },
-    {
-      "epoch": 0.10933348624730012,
-      "grad_norm": 0.0005570969078689814,
-      "learning_rate": 0.00019999999418323638,
-      "loss": 46.0,
-      "step": 1430
-    },
-    {
-      "epoch": 0.10940994323068984,
-      "grad_norm": 0.0011050528846681118,
-      "learning_rate": 0.00019999999417504087,
-      "loss": 46.0,
-      "step": 1431
-    },
-    {
-      "epoch": 0.10948640021407956,
-      "grad_norm": 0.0005611145752482116,
-      "learning_rate": 0.00019999999416683958,
-      "loss": 46.0,
-      "step": 1432
-    },
-    {
-      "epoch": 0.10956285719746928,
-      "grad_norm": 0.0015184417134150863,
-      "learning_rate": 0.00019999999415863253,
-      "loss": 46.0,
-      "step": 1433
-    },
-    {
-      "epoch": 0.109639314180859,
-      "grad_norm": 0.0009100169991143048,
-      "learning_rate": 0.00019999999415041972,
-      "loss": 46.0,
-      "step": 1434
-    },
-    {
-      "epoch": 0.10971577116424872,
-      "grad_norm": 0.007965796627104282,
-      "learning_rate": 0.00019999999414220114,
-      "loss": 46.0,
-      "step": 1435
-    },
-    {
-      "epoch": 0.10979222814763843,
-      "grad_norm": 0.0007768543437123299,
-      "learning_rate": 0.0001999999941339768,
-      "loss": 46.0,
-      "step": 1436
-    },
-    {
-      "epoch": 0.10986868513102815,
-      "grad_norm": 0.0009607757674530149,
-      "learning_rate": 0.00019999999412574667,
-      "loss": 46.0,
-      "step": 1437
-    },
-    {
-      "epoch": 0.10994514211441787,
-      "grad_norm": 0.0006826695171184838,
-      "learning_rate": 0.00019999999411751077,
-      "loss": 46.0,
-      "step": 1438
-    },
-    {
-      "epoch": 0.11002159909780759,
-      "grad_norm": 0.0016457445453852415,
-      "learning_rate": 0.00019999999410926907,
-      "loss": 46.0,
-      "step": 1439
-    },
-    {
-      "epoch": 0.11009805608119731,
-      "grad_norm": 0.0008279511239379644,
-      "learning_rate": 0.00019999999410102166,
-      "loss": 46.0,
-      "step": 1440
-    },
-    {
-      "epoch": 0.11017451306458703,
-      "grad_norm": 0.001849550404585898,
-      "learning_rate": 0.00019999999409276846,
-      "loss": 46.0,
-      "step": 1441
-    },
-    {
-      "epoch": 0.11025097004797675,
-      "grad_norm": 0.0016342638991773129,
-      "learning_rate": 0.00019999999408450947,
-      "loss": 46.0,
-      "step": 1442
-    },
-    {
-      "epoch": 0.11032742703136647,
-      "grad_norm": 0.0023353334981948137,
-      "learning_rate": 0.00019999999407624474,
-      "loss": 46.0,
-      "step": 1443
-    },
-    {
-      "epoch": 0.1104038840147562,
-      "grad_norm": 0.0011888748267665505,
-      "learning_rate": 0.00019999999406797422,
-      "loss": 46.0,
-      "step": 1444
-    },
-    {
-      "epoch": 0.11048034099814592,
-      "grad_norm": 0.0009299297234974802,
-      "learning_rate": 0.00019999999405969794,
-      "loss": 46.0,
-      "step": 1445
-    },
-    {
-      "epoch": 0.11055679798153564,
-      "grad_norm": 0.0008306152885779738,
-      "learning_rate": 0.00019999999405141588,
-      "loss": 46.0,
-      "step": 1446
-    },
-    {
-      "epoch": 0.11063325496492536,
-      "grad_norm": 0.0020928976591676474,
-      "learning_rate": 0.00019999999404312808,
-      "loss": 46.0,
-      "step": 1447
-    },
-    {
-      "epoch": 0.11070971194831508,
-      "grad_norm": 0.0013051697751507163,
-      "learning_rate": 0.0001999999940348345,
-      "loss": 46.0,
-      "step": 1448
-    },
-    {
-      "epoch": 0.1107861689317048,
-      "grad_norm": 0.002895282581448555,
-      "learning_rate": 0.00019999999402653512,
-      "loss": 46.0,
-      "step": 1449
-    },
-    {
-      "epoch": 0.11086262591509452,
-      "grad_norm": 0.0020371994469314814,
-      "learning_rate": 0.00019999999401823,
-      "loss": 46.0,
-      "step": 1450
-    },
-    {
-      "epoch": 0.11093908289848424,
-      "grad_norm": 0.000690752174705267,
-      "learning_rate": 0.0001999999940099191,
-      "loss": 46.0,
-      "step": 1451
-    },
-    {
-      "epoch": 0.11101553988187396,
-      "grad_norm": 0.0012538381852209568,
-      "learning_rate": 0.00019999999400160243,
-      "loss": 46.0,
-      "step": 1452
-    },
-    {
-      "epoch": 0.11109199686526368,
-      "grad_norm": 0.0018687015399336815,
-      "learning_rate": 0.00019999999399327999,
-      "loss": 46.0,
-      "step": 1453
-    },
-    {
-      "epoch": 0.1111684538486534,
-      "grad_norm": 0.0005745233502238989,
-      "learning_rate": 0.00019999999398495177,
-      "loss": 46.0,
-      "step": 1454
-    },
-    {
-      "epoch": 0.11124491083204312,
-      "grad_norm": 0.0011587274493649602,
-      "learning_rate": 0.0001999999939766178,
-      "loss": 46.0,
-      "step": 1455
-    },
-    {
-      "epoch": 0.11132136781543284,
-      "grad_norm": 0.0012289321748539805,
-      "learning_rate": 0.00019999999396827804,
-      "loss": 46.0,
-      "step": 1456
-    },
-    {
-      "epoch": 0.11139782479882256,
-      "grad_norm": 0.0006031632656231523,
-      "learning_rate": 0.00019999999395993253,
-      "loss": 46.0,
-      "step": 1457
-    },
-    {
-      "epoch": 0.11147428178221229,
-      "grad_norm": 0.0010334086837247014,
-      "learning_rate": 0.00019999999395158127,
-      "loss": 46.0,
-      "step": 1458
-    },
-    {
-      "epoch": 0.11155073876560201,
-      "grad_norm": 0.0022469027899205685,
-      "learning_rate": 0.00019999999394322421,
-      "loss": 46.0,
-      "step": 1459
-    },
-    {
-      "epoch": 0.11162719574899173,
-      "grad_norm": 0.0007784420158714056,
-      "learning_rate": 0.00019999999393486138,
-      "loss": 46.0,
-      "step": 1460
-    },
-    {
-      "epoch": 0.11170365273238145,
-      "grad_norm": 0.0029464373365044594,
-      "learning_rate": 0.0001999999939264928,
-      "loss": 46.0,
-      "step": 1461
-    },
-    {
-      "epoch": 0.11178010971577117,
-      "grad_norm": 0.0010416472796350718,
-      "learning_rate": 0.00019999999391811843,
-      "loss": 46.0,
-      "step": 1462
-    },
-    {
-      "epoch": 0.11185656669916089,
-      "grad_norm": 0.0014502541162073612,
-      "learning_rate": 0.0001999999939097383,
-      "loss": 46.0,
-      "step": 1463
-    },
-    {
-      "epoch": 0.1119330236825506,
-      "grad_norm": 0.0013804328627884388,
-      "learning_rate": 0.00019999999390135238,
-      "loss": 46.0,
-      "step": 1464
-    },
-    {
-      "epoch": 0.11200948066594033,
-      "grad_norm": 0.0004934798344038427,
-      "learning_rate": 0.00019999999389296074,
-      "loss": 46.0,
-      "step": 1465
-    },
-    {
-      "epoch": 0.11208593764933004,
-      "grad_norm": 0.0012159311445429921,
-      "learning_rate": 0.00019999999388456327,
-      "loss": 46.0,
-      "step": 1466
-    },
-    {
-      "epoch": 0.11216239463271976,
-      "grad_norm": 0.0015492700040340424,
-      "learning_rate": 0.00019999999387616008,
-      "loss": 46.0,
-      "step": 1467
-    },
-    {
-      "epoch": 0.11223885161610948,
-      "grad_norm": 0.0008706760709173977,
-      "learning_rate": 0.00019999999386775109,
-      "loss": 46.0,
-      "step": 1468
-    },
-    {
-      "epoch": 0.1123153085994992,
-      "grad_norm": 0.0009424204472452402,
-      "learning_rate": 0.00019999999385933635,
-      "loss": 46.0,
-      "step": 1469
-    },
-    {
-      "epoch": 0.11239176558288892,
-      "grad_norm": 0.0019854246638715267,
-      "learning_rate": 0.00019999999385091584,
-      "loss": 46.0,
-      "step": 1470
-    },
-    {
-      "epoch": 0.11246822256627864,
-      "grad_norm": 0.0011662030592560768,
-      "learning_rate": 0.00019999999384248955,
-      "loss": 46.0,
-      "step": 1471
-    },
-    {
-      "epoch": 0.11254467954966836,
-      "grad_norm": 0.0038098893128335476,
-      "learning_rate": 0.00019999999383405747,
-      "loss": 46.0,
-      "step": 1472
-    },
-    {
-      "epoch": 0.1126211365330581,
-      "grad_norm": 0.0011925911530852318,
-      "learning_rate": 0.00019999999382561967,
-      "loss": 46.0,
-      "step": 1473
-    },
-    {
-      "epoch": 0.11269759351644781,
-      "grad_norm": 0.0010681368876248598,
-      "learning_rate": 0.00019999999381717606,
-      "loss": 46.0,
-      "step": 1474
-    },
-    {
-      "epoch": 0.11277405049983753,
-      "grad_norm": 0.002948192646726966,
-      "learning_rate": 0.00019999999380872669,
-      "loss": 46.0,
-      "step": 1475
-    },
-    {
-      "epoch": 0.11285050748322725,
-      "grad_norm": 0.001269427128136158,
-      "learning_rate": 0.00019999999380027153,
-      "loss": 46.0,
-      "step": 1476
-    },
-    {
-      "epoch": 0.11292696446661697,
-      "grad_norm": 0.001073826220817864,
-      "learning_rate": 0.00019999999379181067,
-      "loss": 46.0,
-      "step": 1477
-    },
-    {
-      "epoch": 0.11300342145000669,
-      "grad_norm": 0.0010608448646962643,
-      "learning_rate": 0.00019999999378334397,
-      "loss": 46.0,
-      "step": 1478
-    },
-    {
-      "epoch": 0.11307987843339641,
-      "grad_norm": 0.0005864224513061345,
-      "learning_rate": 0.00019999999377487155,
-      "loss": 46.0,
-      "step": 1479
-    },
-    {
-      "epoch": 0.11315633541678613,
-      "grad_norm": 0.0008025203133001924,
-      "learning_rate": 0.0001999999937663933,
-      "loss": 46.0,
-      "step": 1480
-    },
-    {
-      "epoch": 0.11323279240017585,
-      "grad_norm": 0.0012322886614128947,
-      "learning_rate": 0.00019999999375790934,
-      "loss": 46.0,
-      "step": 1481
-    },
-    {
-      "epoch": 0.11330924938356557,
-      "grad_norm": 0.0012553668348118663,
-      "learning_rate": 0.0001999999937494196,
-      "loss": 46.0,
-      "step": 1482
-    },
-    {
-      "epoch": 0.11338570636695529,
-      "grad_norm": 0.0005710609257221222,
-      "learning_rate": 0.0001999999937409241,
-      "loss": 46.0,
-      "step": 1483
-    },
-    {
-      "epoch": 0.11346216335034501,
-      "grad_norm": 0.002128420863300562,
-      "learning_rate": 0.00019999999373242279,
-      "loss": 46.0,
-      "step": 1484
-    },
-    {
-      "epoch": 0.11353862033373473,
-      "grad_norm": 0.001062526716850698,
-      "learning_rate": 0.0001999999937239157,
-      "loss": 46.0,
-      "step": 1485
-    },
-    {
-      "epoch": 0.11361507731712445,
-      "grad_norm": 0.0016776778502389789,
-      "learning_rate": 0.0001999999937154029,
-      "loss": 46.0,
-      "step": 1486
-    },
-    {
-      "epoch": 0.11369153430051418,
-      "grad_norm": 0.0010068764677271247,
-      "learning_rate": 0.0001999999937068843,
-      "loss": 46.0,
-      "step": 1487
-    },
-    {
-      "epoch": 0.1137679912839039,
-      "grad_norm": 0.0009164699586108327,
-      "learning_rate": 0.00019999999369835992,
-      "loss": 46.0,
-      "step": 1488
-    },
-    {
-      "epoch": 0.11384444826729362,
-      "grad_norm": 0.0009315891074948013,
-      "learning_rate": 0.00019999999368982977,
-      "loss": 46.0,
-      "step": 1489
-    },
-    {
-      "epoch": 0.11392090525068334,
-      "grad_norm": 0.004191101063042879,
-      "learning_rate": 0.00019999999368129387,
-      "loss": 46.0,
-      "step": 1490
-    },
-    {
-      "epoch": 0.11399736223407306,
-      "grad_norm": 0.001186610315926373,
-      "learning_rate": 0.0001999999936727522,
-      "loss": 46.0,
-      "step": 1491
-    },
-    {
-      "epoch": 0.11407381921746278,
-      "grad_norm": 0.0006289630546234548,
-      "learning_rate": 0.00019999999366420473,
-      "loss": 46.0,
-      "step": 1492
-    },
-    {
-      "epoch": 0.1141502762008525,
-      "grad_norm": 0.0005648771184496582,
-      "learning_rate": 0.00019999999365565154,
-      "loss": 46.0,
-      "step": 1493
-    },
-    {
-      "epoch": 0.11422673318424222,
-      "grad_norm": 0.0008261979091912508,
-      "learning_rate": 0.00019999999364709255,
-      "loss": 46.0,
-      "step": 1494
-    },
-    {
-      "epoch": 0.11430319016763194,
-      "grad_norm": 0.0027124767657369375,
-      "learning_rate": 0.0001999999936385278,
-      "loss": 46.0,
-      "step": 1495
-    },
-    {
-      "epoch": 0.11437964715102165,
-      "grad_norm": 0.00077613111352548,
-      "learning_rate": 0.00019999999362995728,
-      "loss": 46.0,
-      "step": 1496
-    },
-    {
-      "epoch": 0.11445610413441137,
-      "grad_norm": 0.0007746916962787509,
-      "learning_rate": 0.00019999999362138097,
-      "loss": 46.0,
-      "step": 1497
-    },
-    {
-      "epoch": 0.1145325611178011,
-      "grad_norm": 0.002079334808513522,
-      "learning_rate": 0.00019999999361279894,
-      "loss": 46.0,
-      "step": 1498
-    },
-    {
-      "epoch": 0.11460901810119081,
-      "grad_norm": 0.00495891971513629,
-      "learning_rate": 0.00019999999360421108,
-      "loss": 46.0,
-      "step": 1499
-    },
-    {
-      "epoch": 0.11468547508458053,
-      "grad_norm": 0.0011691523250192404,
-      "learning_rate": 0.00019999999359561748,
-      "loss": 46.0,
-      "step": 1500
-    },
-    {
-      "epoch": 0.11476193206797025,
-      "grad_norm": 0.0010779015719890594,
-      "learning_rate": 0.0001999999935870181,
-      "loss": 46.0,
-      "step": 1501
-    },
-    {
-      "epoch": 0.11483838905135998,
-      "grad_norm": 0.0004119532823096961,
-      "learning_rate": 0.00019999999357841298,
-      "loss": 46.0,
-      "step": 1502
-    },
-    {
-      "epoch": 0.1149148460347497,
-      "grad_norm": 0.0021948053035885096,
-      "learning_rate": 0.00019999999356980206,
-      "loss": 46.0,
-      "step": 1503
-    },
-    {
-      "epoch": 0.11499130301813942,
-      "grad_norm": 0.0027411372866481543,
-      "learning_rate": 0.00019999999356118539,
-      "loss": 46.0,
-      "step": 1504
-    },
-    {
-      "epoch": 0.11506776000152914,
-      "grad_norm": 0.003387674456462264,
-      "learning_rate": 0.00019999999355256294,
-      "loss": 46.0,
-      "step": 1505
-    },
-    {
-      "epoch": 0.11514421698491886,
-      "grad_norm": 0.0023454974871128798,
-      "learning_rate": 0.00019999999354393473,
-      "loss": 46.0,
-      "step": 1506
-    },
-    {
-      "epoch": 0.11522067396830858,
-      "grad_norm": 0.0013973686145618558,
-      "learning_rate": 0.0001999999935353007,
-      "loss": 46.0,
-      "step": 1507
-    },
-    {
-      "epoch": 0.1152971309516983,
-      "grad_norm": 0.00044357802835293114,
-      "learning_rate": 0.00019999999352666097,
-      "loss": 46.0,
-      "step": 1508
-    },
-    {
-      "epoch": 0.11537358793508802,
-      "grad_norm": 0.0015124761266633868,
-      "learning_rate": 0.00019999999351801544,
-      "loss": 46.0,
-      "step": 1509
-    },
-    {
-      "epoch": 0.11545004491847774,
-      "grad_norm": 0.0008096142555586994,
-      "learning_rate": 0.00019999999350936415,
-      "loss": 46.0,
-      "step": 1510
-    },
-    {
-      "epoch": 0.11552650190186746,
-      "grad_norm": 0.001232494949363172,
-      "learning_rate": 0.00019999999350070707,
-      "loss": 46.0,
-      "step": 1511
-    },
-    {
-      "epoch": 0.11560295888525718,
-      "grad_norm": 0.0010967283742502332,
-      "learning_rate": 0.00019999999349204424,
-      "loss": 46.0,
-      "step": 1512
-    },
-    {
-      "epoch": 0.1156794158686469,
-      "grad_norm": 0.0011971114436164498,
-      "learning_rate": 0.00019999999348337564,
-      "loss": 46.0,
-      "step": 1513
-    },
-    {
-      "epoch": 0.11575587285203662,
-      "grad_norm": 0.0020442737732082605,
-      "learning_rate": 0.00019999999347470126,
-      "loss": 46.0,
-      "step": 1514
-    },
-    {
-      "epoch": 0.11583232983542634,
-      "grad_norm": 0.0030483112204819918,
-      "learning_rate": 0.0001999999934660211,
-      "loss": 46.0,
-      "step": 1515
-    },
-    {
-      "epoch": 0.11590878681881607,
-      "grad_norm": 0.0011137648252770305,
-      "learning_rate": 0.0001999999934573352,
-      "loss": 46.0,
-      "step": 1516
-    },
-    {
-      "epoch": 0.11598524380220579,
-      "grad_norm": 0.0008476824150420725,
-      "learning_rate": 0.00019999999344864352,
-      "loss": 46.0,
-      "step": 1517
-    },
-    {
-      "epoch": 0.11606170078559551,
-      "grad_norm": 0.000687445281073451,
-      "learning_rate": 0.00019999999343994608,
-      "loss": 46.0,
-      "step": 1518
-    },
-    {
-      "epoch": 0.11613815776898523,
-      "grad_norm": 0.00230810372158885,
-      "learning_rate": 0.00019999999343124286,
-      "loss": 46.0,
-      "step": 1519
-    },
-    {
-      "epoch": 0.11621461475237495,
-      "grad_norm": 0.0007136607891879976,
-      "learning_rate": 0.00019999999342253387,
-      "loss": 46.0,
-      "step": 1520
-    },
-    {
-      "epoch": 0.11629107173576467,
-      "grad_norm": 0.0013810483505949378,
-      "learning_rate": 0.0001999999934138191,
-      "loss": 46.0,
-      "step": 1521
-    },
-    {
-      "epoch": 0.11636752871915439,
-      "grad_norm": 0.0020662571769207716,
-      "learning_rate": 0.00019999999340509857,
-      "loss": 46.0,
-      "step": 1522
-    },
-    {
-      "epoch": 0.1164439857025441,
-      "grad_norm": 0.0006200867937877774,
-      "learning_rate": 0.00019999999339637226,
-      "loss": 46.0,
-      "step": 1523
-    },
-    {
-      "epoch": 0.11652044268593383,
-      "grad_norm": 0.001146310125477612,
-      "learning_rate": 0.0001999999933876402,
-      "loss": 46.0,
-      "step": 1524
-    },
-    {
-      "epoch": 0.11659689966932354,
-      "grad_norm": 0.001682683709077537,
-      "learning_rate": 0.00019999999337890235,
-      "loss": 46.0,
-      "step": 1525
-    },
-    {
-      "epoch": 0.11667335665271326,
-      "grad_norm": 0.0017838873900473118,
-      "learning_rate": 0.00019999999337015875,
-      "loss": 46.0,
-      "step": 1526
-    },
-    {
-      "epoch": 0.11674981363610298,
-      "grad_norm": 0.0004894313751719892,
-      "learning_rate": 0.00019999999336140937,
-      "loss": 46.0,
-      "step": 1527
-    },
-    {
-      "epoch": 0.1168262706194927,
-      "grad_norm": 0.002732678782194853,
-      "learning_rate": 0.00019999999335265422,
-      "loss": 46.0,
-      "step": 1528
-    },
-    {
-      "epoch": 0.11690272760288242,
-      "grad_norm": 0.0005846915300935507,
-      "learning_rate": 0.00019999999334389333,
-      "loss": 46.0,
-      "step": 1529
-    },
-    {
-      "epoch": 0.11697918458627214,
-      "grad_norm": 0.0008995202369987965,
-      "learning_rate": 0.00019999999333512666,
-      "loss": 46.0,
-      "step": 1530
-    },
-    {
-      "epoch": 0.11705564156966188,
-      "grad_norm": 0.0016604404663667083,
-      "learning_rate": 0.0001999999933263542,
-      "loss": 46.0,
-      "step": 1531
-    },
-    {
-      "epoch": 0.1171320985530516,
-      "grad_norm": 0.001645913696847856,
-      "learning_rate": 0.00019999999331757595,
-      "loss": 46.0,
-      "step": 1532
-    },
-    {
-      "epoch": 0.11720855553644131,
-      "grad_norm": 0.006875058636069298,
-      "learning_rate": 0.00019999999330879198,
-      "loss": 46.0,
-      "step": 1533
-    },
-    {
-      "epoch": 0.11728501251983103,
-      "grad_norm": 0.0011267911177128553,
-      "learning_rate": 0.00019999999330000222,
-      "loss": 46.0,
-      "step": 1534
-    },
-    {
-      "epoch": 0.11736146950322075,
-      "grad_norm": 0.0006858114502392709,
-      "learning_rate": 0.00019999999329120666,
-      "loss": 46.0,
-      "step": 1535
-    },
-    {
-      "epoch": 0.11743792648661047,
-      "grad_norm": 0.0014258320443332195,
-      "learning_rate": 0.00019999999328240535,
-      "loss": 46.0,
-      "step": 1536
-    },
-    {
-      "epoch": 0.11751438347000019,
-      "grad_norm": 0.0008733182330615819,
-      "learning_rate": 0.00019999999327359832,
-      "loss": 46.0,
-      "step": 1537
-    },
-    {
-      "epoch": 0.11759084045338991,
-      "grad_norm": 0.0006535431020893157,
-      "learning_rate": 0.00019999999326478544,
-      "loss": 46.0,
-      "step": 1538
-    },
-    {
-      "epoch": 0.11766729743677963,
-      "grad_norm": 0.0006404178566299379,
-      "learning_rate": 0.00019999999325596686,
-      "loss": 46.0,
-      "step": 1539
-    },
-    {
-      "epoch": 0.11774375442016935,
-      "grad_norm": 0.0011238687438890338,
-      "learning_rate": 0.00019999999324714246,
-      "loss": 46.0,
-      "step": 1540
-    },
-    {
-      "epoch": 0.11782021140355907,
-      "grad_norm": 0.0014135445235297084,
-      "learning_rate": 0.0001999999932383123,
-      "loss": 46.0,
-      "step": 1541
-    },
-    {
-      "epoch": 0.11789666838694879,
-      "grad_norm": 0.0005383029347285628,
-      "learning_rate": 0.00019999999322947641,
-      "loss": 46.0,
-      "step": 1542
-    },
-    {
-      "epoch": 0.11797312537033851,
-      "grad_norm": 0.0007065276149660349,
-      "learning_rate": 0.00019999999322063472,
-      "loss": 46.0,
-      "step": 1543
-    },
-    {
-      "epoch": 0.11804958235372823,
-      "grad_norm": 0.0010419221362099051,
-      "learning_rate": 0.00019999999321178725,
-      "loss": 46.0,
-      "step": 1544
-    },
-    {
-      "epoch": 0.11812603933711796,
-      "grad_norm": 0.0015286566922441125,
-      "learning_rate": 0.00019999999320293403,
-      "loss": 46.0,
-      "step": 1545
-    },
-    {
-      "epoch": 0.11820249632050768,
-      "grad_norm": 0.0021061627194285393,
-      "learning_rate": 0.00019999999319407502,
-      "loss": 46.0,
-      "step": 1546
-    },
-    {
-      "epoch": 0.1182789533038974,
-      "grad_norm": 0.005479545332491398,
-      "learning_rate": 0.00019999999318521026,
-      "loss": 46.0,
-      "step": 1547
-    },
-    {
-      "epoch": 0.11835541028728712,
-      "grad_norm": 0.0011062328703701496,
-      "learning_rate": 0.0001999999931763397,
-      "loss": 46.0,
-      "step": 1548
-    },
-    {
-      "epoch": 0.11843186727067684,
-      "grad_norm": 0.0010945976246148348,
-      "learning_rate": 0.00019999999316746344,
-      "loss": 46.0,
-      "step": 1549
-    },
-    {
-      "epoch": 0.11850832425406656,
-      "grad_norm": 0.0019600046798586845,
-      "learning_rate": 0.00019999999315858133,
-      "loss": 46.0,
-      "step": 1550
-    },
-    {
-      "epoch": 0.11858478123745628,
-      "grad_norm": 0.0015371062327176332,
-      "learning_rate": 0.0001999999931496935,
-      "loss": 46.0,
-      "step": 1551
-    },
-    {
-      "epoch": 0.118661238220846,
-      "grad_norm": 0.0026392927393317223,
-      "learning_rate": 0.00019999999314079987,
-      "loss": 46.0,
-      "step": 1552
-    },
-    {
-      "epoch": 0.11873769520423572,
-      "grad_norm": 0.0008766052778810263,
-      "learning_rate": 0.0001999999931319005,
-      "loss": 46.0,
-      "step": 1553
-    },
-    {
-      "epoch": 0.11881415218762544,
-      "grad_norm": 0.0020937644876539707,
-      "learning_rate": 0.00019999999312299535,
-      "loss": 46.0,
-      "step": 1554
-    },
-    {
-      "epoch": 0.11889060917101515,
-      "grad_norm": 0.0025007661897689104,
-      "learning_rate": 0.00019999999311408443,
-      "loss": 46.0,
-      "step": 1555
-    },
-    {
-      "epoch": 0.11896706615440487,
-      "grad_norm": 0.0004965580883435905,
-      "learning_rate": 0.00019999999310516776,
-      "loss": 46.0,
-      "step": 1556
-    },
-    {
-      "epoch": 0.1190435231377946,
-      "grad_norm": 0.001314135268330574,
-      "learning_rate": 0.0001999999930962453,
-      "loss": 46.0,
-      "step": 1557
-    },
-    {
-      "epoch": 0.11911998012118431,
-      "grad_norm": 0.0008292917627841234,
-      "learning_rate": 0.00019999999308731705,
-      "loss": 46.0,
-      "step": 1558
-    },
-    {
-      "epoch": 0.11919643710457403,
-      "grad_norm": 0.0013552983291447163,
-      "learning_rate": 0.00019999999307838303,
-      "loss": 46.0,
-      "step": 1559
-    },
-    {
-      "epoch": 0.11927289408796377,
-      "grad_norm": 0.0005517539102584124,
-      "learning_rate": 0.00019999999306944327,
-      "loss": 46.0,
-      "step": 1560
-    },
-    {
-      "epoch": 0.11934935107135348,
-      "grad_norm": 0.0017526630545035005,
-      "learning_rate": 0.00019999999306049776,
-      "loss": 46.0,
-      "step": 1561
-    },
-    {
-      "epoch": 0.1194258080547432,
-      "grad_norm": 0.0008628126233816147,
-      "learning_rate": 0.00019999999305154642,
-      "loss": 46.0,
-      "step": 1562
-    },
-    {
-      "epoch": 0.11950226503813292,
-      "grad_norm": 0.0014163063606247306,
-      "learning_rate": 0.00019999999304258937,
-      "loss": 46.0,
-      "step": 1563
-    },
-    {
-      "epoch": 0.11957872202152264,
-      "grad_norm": 0.001431107404641807,
-      "learning_rate": 0.00019999999303362651,
-      "loss": 46.0,
-      "step": 1564
-    },
-    {
-      "epoch": 0.11965517900491236,
-      "grad_norm": 0.004410839639604092,
-      "learning_rate": 0.00019999999302465788,
-      "loss": 46.0,
-      "step": 1565
-    },
-    {
-      "epoch": 0.11973163598830208,
-      "grad_norm": 0.0005859207012690604,
-      "learning_rate": 0.0001999999930156835,
-      "loss": 46.0,
-      "step": 1566
-    },
-    {
-      "epoch": 0.1198080929716918,
-      "grad_norm": 0.0008173504029400647,
-      "learning_rate": 0.00019999999300670336,
-      "loss": 46.0,
-      "step": 1567
-    },
-    {
-      "epoch": 0.11988454995508152,
-      "grad_norm": 0.003606301499530673,
-      "learning_rate": 0.00019999999299771744,
-      "loss": 46.0,
-      "step": 1568
-    },
-    {
-      "epoch": 0.11996100693847124,
-      "grad_norm": 0.0006822749273851514,
-      "learning_rate": 0.00019999999298872574,
-      "loss": 46.0,
-      "step": 1569
-    },
-    {
-      "epoch": 0.12003746392186096,
-      "grad_norm": 0.0013308299239724874,
-      "learning_rate": 0.00019999999297972828,
-      "loss": 46.0,
-      "step": 1570
-    },
-    {
-      "epoch": 0.12011392090525068,
-      "grad_norm": 0.0026975388173013926,
-      "learning_rate": 0.00019999999297072503,
-      "loss": 46.0,
-      "step": 1571
-    },
-    {
-      "epoch": 0.1201903778886404,
-      "grad_norm": 0.0006454059621319175,
-      "learning_rate": 0.00019999999296171602,
-      "loss": 46.0,
-      "step": 1572
-    },
-    {
-      "epoch": 0.12026683487203012,
-      "grad_norm": 0.0020386148244142532,
-      "learning_rate": 0.00019999999295270126,
-      "loss": 46.0,
-      "step": 1573
-    },
-    {
-      "epoch": 0.12034329185541985,
-      "grad_norm": 0.0009303594706580043,
-      "learning_rate": 0.0001999999929436807,
-      "loss": 46.0,
-      "step": 1574
-    },
-    {
-      "epoch": 0.12041974883880957,
-      "grad_norm": 0.001186365494504571,
-      "learning_rate": 0.00019999999293465441,
-      "loss": 46.0,
-      "step": 1575
-    },
-    {
-      "epoch": 0.12049620582219929,
-      "grad_norm": 0.0010818771552294493,
-      "learning_rate": 0.00019999999292562233,
-      "loss": 46.0,
-      "step": 1576
-    },
-    {
-      "epoch": 0.12057266280558901,
-      "grad_norm": 0.001366869080811739,
-      "learning_rate": 0.00019999999291658448,
-      "loss": 46.0,
-      "step": 1577
-    },
-    {
-      "epoch": 0.12064911978897873,
-      "grad_norm": 0.0028633789625018835,
-      "learning_rate": 0.00019999999290754085,
-      "loss": 46.0,
-      "step": 1578
-    },
-    {
-      "epoch": 0.12072557677236845,
-      "grad_norm": 0.0005275193834677339,
-      "learning_rate": 0.00019999999289849147,
-      "loss": 46.0,
-      "step": 1579
-    },
-    {
-      "epoch": 0.12080203375575817,
-      "grad_norm": 0.0007849180838093162,
-      "learning_rate": 0.00019999999288943633,
-      "loss": 46.0,
-      "step": 1580
-    },
-    {
-      "epoch": 0.12087849073914789,
-      "grad_norm": 0.0007656039670109749,
-      "learning_rate": 0.00019999999288037538,
-      "loss": 46.0,
-      "step": 1581
-    },
-    {
-      "epoch": 0.1209549477225376,
-      "grad_norm": 0.0012410631170496345,
-      "learning_rate": 0.00019999999287130868,
-      "loss": 46.0,
-      "step": 1582
-    },
-    {
-      "epoch": 0.12103140470592733,
-      "grad_norm": 0.002462558913975954,
-      "learning_rate": 0.00019999999286223621,
-      "loss": 46.0,
-      "step": 1583
-    },
-    {
-      "epoch": 0.12110786168931705,
-      "grad_norm": 0.0009302882244810462,
-      "learning_rate": 0.00019999999285315797,
-      "loss": 46.0,
-      "step": 1584
-    },
-    {
-      "epoch": 0.12118431867270676,
-      "grad_norm": 0.0015897693810984492,
-      "learning_rate": 0.00019999999284407398,
-      "loss": 46.0,
-      "step": 1585
-    },
-    {
-      "epoch": 0.12126077565609648,
-      "grad_norm": 0.001534050446934998,
-      "learning_rate": 0.0001999999928349842,
-      "loss": 46.0,
-      "step": 1586
-    },
-    {
-      "epoch": 0.1213372326394862,
-      "grad_norm": 0.0007494865567423403,
-      "learning_rate": 0.00019999999282588866,
-      "loss": 46.0,
-      "step": 1587
-    },
-    {
-      "epoch": 0.12141368962287592,
-      "grad_norm": 0.0010759225115180016,
-      "learning_rate": 0.00019999999281678735,
-      "loss": 46.0,
-      "step": 1588
-    },
-    {
-      "epoch": 0.12149014660626566,
-      "grad_norm": 0.0006454409449361265,
-      "learning_rate": 0.00019999999280768024,
-      "loss": 46.0,
-      "step": 1589
-    },
-    {
-      "epoch": 0.12156660358965538,
-      "grad_norm": 0.0008099248516373336,
-      "learning_rate": 0.00019999999279856742,
-      "loss": 46.0,
-      "step": 1590
-    },
-    {
-      "epoch": 0.1216430605730451,
-      "grad_norm": 0.0014339295448735356,
-      "learning_rate": 0.0001999999927894488,
-      "loss": 46.0,
-      "step": 1591
-    },
-    {
-      "epoch": 0.12171951755643481,
-      "grad_norm": 0.0006267589633353055,
-      "learning_rate": 0.0001999999927803244,
-      "loss": 46.0,
-      "step": 1592
-    },
-    {
-      "epoch": 0.12179597453982453,
-      "grad_norm": 0.000685067439917475,
-      "learning_rate": 0.00019999999277119424,
-      "loss": 46.0,
-      "step": 1593
-    },
-    {
-      "epoch": 0.12187243152321425,
-      "grad_norm": 0.001076327171176672,
-      "learning_rate": 0.0001999999927620583,
-      "loss": 46.0,
-      "step": 1594
-    },
-    {
-      "epoch": 0.12194888850660397,
-      "grad_norm": 0.00198566191829741,
-      "learning_rate": 0.0001999999927529166,
-      "loss": 46.0,
-      "step": 1595
-    },
-    {
-      "epoch": 0.12202534548999369,
-      "grad_norm": 0.00043639555224217474,
-      "learning_rate": 0.00019999999274376916,
-      "loss": 46.0,
-      "step": 1596
-    },
-    {
-      "epoch": 0.12210180247338341,
-      "grad_norm": 0.0013530422002077103,
-      "learning_rate": 0.0001999999927346159,
-      "loss": 46.0,
-      "step": 1597
-    },
-    {
-      "epoch": 0.12217825945677313,
-      "grad_norm": 0.0025155062321573496,
-      "learning_rate": 0.00019999999272545687,
-      "loss": 46.0,
-      "step": 1598
-    },
-    {
-      "epoch": 0.12225471644016285,
-      "grad_norm": 0.0008100735140033066,
-      "learning_rate": 0.0001999999927162921,
-      "loss": 46.0,
-      "step": 1599
-    },
-    {
-      "epoch": 0.12233117342355257,
-      "grad_norm": 0.0015133937122300267,
-      "learning_rate": 0.00019999999270712158,
-      "loss": 46.0,
-      "step": 1600
-    },
-    {
-      "epoch": 0.12240763040694229,
-      "grad_norm": 0.0012205112725496292,
-      "learning_rate": 0.00019999999269794524,
-      "loss": 46.0,
-      "step": 1601
-    },
-    {
-      "epoch": 0.12248408739033201,
-      "grad_norm": 0.0005722329951822758,
-      "learning_rate": 0.00019999999268876316,
-      "loss": 46.0,
-      "step": 1602
-    },
-    {
-      "epoch": 0.12256054437372174,
-      "grad_norm": 0.0007153971237130463,
-      "learning_rate": 0.0001999999926795753,
-      "loss": 46.0,
-      "step": 1603
-    },
-    {
-      "epoch": 0.12263700135711146,
-      "grad_norm": 0.0010634405771270394,
-      "learning_rate": 0.00019999999267038168,
-      "loss": 46.0,
-      "step": 1604
-    },
-    {
-      "epoch": 0.12271345834050118,
-      "grad_norm": 0.0007890252163633704,
-      "learning_rate": 0.00019999999266118228,
-      "loss": 46.0,
-      "step": 1605
-    },
-    {
-      "epoch": 0.1227899153238909,
-      "grad_norm": 0.0008101451676338911,
-      "learning_rate": 0.0001999999926519771,
-      "loss": 46.0,
-      "step": 1606
-    },
-    {
-      "epoch": 0.12286637230728062,
-      "grad_norm": 0.0022999884095042944,
-      "learning_rate": 0.00019999999264276618,
-      "loss": 46.0,
-      "step": 1607
-    },
-    {
-      "epoch": 0.12294282929067034,
-      "grad_norm": 0.001534348470158875,
-      "learning_rate": 0.0001999999926335495,
-      "loss": 46.0,
-      "step": 1608
-    },
-    {
-      "epoch": 0.12301928627406006,
-      "grad_norm": 0.0007720842259004712,
-      "learning_rate": 0.000199999992624327,
-      "loss": 46.0,
-      "step": 1609
-    },
-    {
-      "epoch": 0.12309574325744978,
-      "grad_norm": 0.0006925991037860513,
-      "learning_rate": 0.00019999999261509878,
-      "loss": 46.0,
-      "step": 1610
-    },
-    {
-      "epoch": 0.1231722002408395,
-      "grad_norm": 0.0008491818443872035,
-      "learning_rate": 0.00019999999260586474,
-      "loss": 46.0,
-      "step": 1611
-    },
-    {
-      "epoch": 0.12324865722422922,
-      "grad_norm": 0.0011237066937610507,
-      "learning_rate": 0.00019999999259662498,
-      "loss": 46.0,
-      "step": 1612
-    },
-    {
-      "epoch": 0.12332511420761894,
-      "grad_norm": 0.0008256579167209566,
-      "learning_rate": 0.00019999999258737942,
-      "loss": 46.0,
-      "step": 1613
-    },
-    {
-      "epoch": 0.12340157119100865,
-      "grad_norm": 0.00036529547651298344,
-      "learning_rate": 0.0001999999925781281,
-      "loss": 46.0,
-      "step": 1614
-    },
-    {
-      "epoch": 0.12347802817439837,
-      "grad_norm": 0.0004993222537450492,
-      "learning_rate": 0.000199999992568871,
-      "loss": 46.0,
-      "step": 1615
-    },
-    {
-      "epoch": 0.1235544851577881,
-      "grad_norm": 0.00123459508176893,
-      "learning_rate": 0.00019999999255960815,
-      "loss": 46.0,
-      "step": 1616
-    },
-    {
-      "epoch": 0.12363094214117781,
-      "grad_norm": 0.0006311935721896589,
-      "learning_rate": 0.00019999999255033952,
-      "loss": 46.0,
-      "step": 1617
-    },
-    {
-      "epoch": 0.12370739912456755,
-      "grad_norm": 0.001889835111796856,
-      "learning_rate": 0.00019999999254106515,
-      "loss": 46.0,
-      "step": 1618
-    },
-    {
-      "epoch": 0.12378385610795727,
-      "grad_norm": 0.0004617491504177451,
-      "learning_rate": 0.000199999992531785,
-      "loss": 46.0,
-      "step": 1619
-    },
-    {
-      "epoch": 0.12386031309134699,
-      "grad_norm": 0.000762148411013186,
-      "learning_rate": 0.00019999999252249902,
-      "loss": 46.0,
-      "step": 1620
-    },
-    {
-      "epoch": 0.1239367700747367,
-      "grad_norm": 0.0007758060819469392,
-      "learning_rate": 0.0001999999925132073,
-      "loss": 46.0,
-      "step": 1621
-    },
-    {
-      "epoch": 0.12401322705812642,
-      "grad_norm": 0.0009397360845468938,
-      "learning_rate": 0.00019999999250390986,
-      "loss": 46.0,
-      "step": 1622
-    },
-    {
-      "epoch": 0.12408968404151614,
-      "grad_norm": 0.001772842719219625,
-      "learning_rate": 0.00019999999249460662,
-      "loss": 46.0,
-      "step": 1623
-    },
-    {
-      "epoch": 0.12416614102490586,
-      "grad_norm": 0.0011918810196220875,
-      "learning_rate": 0.00019999999248529758,
-      "loss": 46.0,
-      "step": 1624
-    },
-    {
-      "epoch": 0.12424259800829558,
-      "grad_norm": 0.0007972493185661733,
-      "learning_rate": 0.00019999999247598282,
-      "loss": 46.0,
-      "step": 1625
-    },
-    {
-      "epoch": 0.1243190549916853,
-      "grad_norm": 0.006198294460773468,
-      "learning_rate": 0.00019999999246666226,
-      "loss": 46.0,
-      "step": 1626
-    },
-    {
-      "epoch": 0.12439551197507502,
-      "grad_norm": 0.0014145338209345937,
-      "learning_rate": 0.00019999999245733593,
-      "loss": 46.0,
-      "step": 1627
-    },
-    {
-      "epoch": 0.12447196895846474,
-      "grad_norm": 0.0006201372598297894,
-      "learning_rate": 0.00019999999244800384,
-      "loss": 46.0,
-      "step": 1628
-    },
-    {
-      "epoch": 0.12454842594185446,
-      "grad_norm": 0.0004219146794639528,
-      "learning_rate": 0.000199999992438666,
-      "loss": 46.0,
-      "step": 1629
-    },
-    {
-      "epoch": 0.12462488292524418,
-      "grad_norm": 0.0005916628870181739,
-      "learning_rate": 0.00019999999242932234,
-      "loss": 46.0,
-      "step": 1630
-    },
-    {
-      "epoch": 0.1247013399086339,
-      "grad_norm": 0.003980670124292374,
-      "learning_rate": 0.00019999999241997296,
-      "loss": 46.0,
-      "step": 1631
-    },
-    {
-      "epoch": 0.12477779689202363,
-      "grad_norm": 0.002719270996749401,
-      "learning_rate": 0.0001999999924106178,
-      "loss": 46.0,
-      "step": 1632
-    },
-    {
-      "epoch": 0.12485425387541335,
-      "grad_norm": 0.0005816468619741499,
-      "learning_rate": 0.00019999999240125684,
-      "loss": 46.0,
-      "step": 1633
-    },
-    {
-      "epoch": 0.12493071085880307,
-      "grad_norm": 0.0010520482901483774,
-      "learning_rate": 0.00019999999239189012,
-      "loss": 46.0,
-      "step": 1634
-    },
-    {
-      "epoch": 0.1250071678421928,
-      "grad_norm": 0.0015013177180662751,
-      "learning_rate": 0.00019999999238251765,
-      "loss": 46.0,
-      "step": 1635
-    },
-    {
-      "epoch": 0.1250836248255825,
-      "grad_norm": 0.0013738133711740375,
-      "learning_rate": 0.0001999999923731394,
-      "loss": 46.0,
-      "step": 1636
-    },
-    {
-      "epoch": 0.12516008180897223,
-      "grad_norm": 0.0010658979881554842,
-      "learning_rate": 0.0001999999923637554,
-      "loss": 46.0,
-      "step": 1637
-    },
-    {
-      "epoch": 0.12523653879236193,
-      "grad_norm": 0.0010128950234502554,
-      "learning_rate": 0.0001999999923543656,
-      "loss": 46.0,
-      "step": 1638
-    },
-    {
-      "epoch": 0.12531299577575167,
-      "grad_norm": 0.0007655302179045975,
-      "learning_rate": 0.00019999999234497005,
-      "loss": 46.0,
-      "step": 1639
-    },
-    {
-      "epoch": 0.1253894527591414,
-      "grad_norm": 0.0012335422215983272,
-      "learning_rate": 0.00019999999233556872,
-      "loss": 46.0,
-      "step": 1640
-    },
-    {
-      "epoch": 0.1254659097425311,
-      "grad_norm": 0.0015967022627592087,
-      "learning_rate": 0.00019999999232616164,
-      "loss": 46.0,
-      "step": 1641
-    },
-    {
-      "epoch": 0.12554236672592084,
-      "grad_norm": 0.0010169546585530043,
-      "learning_rate": 0.00019999999231674878,
-      "loss": 46.0,
-      "step": 1642
-    },
-    {
-      "epoch": 0.12561882370931055,
-      "grad_norm": 0.0006275867926888168,
-      "learning_rate": 0.00019999999230733013,
-      "loss": 46.0,
-      "step": 1643
-    },
-    {
-      "epoch": 0.12569528069270028,
-      "grad_norm": 0.0013624482089653611,
-      "learning_rate": 0.0001999999922979057,
-      "loss": 46.0,
-      "step": 1644
-    },
-    {
-      "epoch": 0.12577173767608998,
-      "grad_norm": 0.00048296854947693646,
-      "learning_rate": 0.00019999999228847556,
-      "loss": 46.0,
-      "step": 1645
-    },
-    {
-      "epoch": 0.12584819465947972,
-      "grad_norm": 0.0007546735578216612,
-      "learning_rate": 0.0001999999922790396,
-      "loss": 46.0,
-      "step": 1646
-    },
-    {
-      "epoch": 0.12592465164286942,
-      "grad_norm": 0.0007814841228537261,
-      "learning_rate": 0.00019999999226959792,
-      "loss": 46.0,
-      "step": 1647
-    },
-    {
-      "epoch": 0.12600110862625916,
-      "grad_norm": 0.003359179710969329,
-      "learning_rate": 0.00019999999226015042,
-      "loss": 46.0,
-      "step": 1648
-    },
-    {
-      "epoch": 0.12607756560964886,
-      "grad_norm": 0.0009251692681573331,
-      "learning_rate": 0.00019999999225069716,
-      "loss": 46.0,
-      "step": 1649
-    },
-    {
-      "epoch": 0.1261540225930386,
-      "grad_norm": 0.0019433617126196623,
-      "learning_rate": 0.00019999999224123814,
-      "loss": 46.0,
-      "step": 1650
-    },
-    {
-      "epoch": 0.1262304795764283,
-      "grad_norm": 0.0014663782203570008,
-      "learning_rate": 0.00019999999223177338,
-      "loss": 46.0,
-      "step": 1651
-    },
-    {
-      "epoch": 0.12630693655981803,
-      "grad_norm": 0.0020708160009235144,
-      "learning_rate": 0.00019999999222230282,
-      "loss": 46.0,
-      "step": 1652
-    },
-    {
-      "epoch": 0.12638339354320774,
-      "grad_norm": 0.007735388353466988,
-      "learning_rate": 0.0001999999922128265,
-      "loss": 46.0,
-      "step": 1653
-    },
-    {
-      "epoch": 0.12645985052659747,
-      "grad_norm": 0.002494325628504157,
-      "learning_rate": 0.00019999999220334438,
-      "loss": 46.0,
-      "step": 1654
-    },
-    {
-      "epoch": 0.1265363075099872,
-      "grad_norm": 0.0014957452658563852,
-      "learning_rate": 0.0001999999921938565,
-      "loss": 46.0,
-      "step": 1655
-    },
-    {
-      "epoch": 0.1266127644933769,
-      "grad_norm": 0.0009274912299588323,
-      "learning_rate": 0.00019999999218436287,
-      "loss": 46.0,
-      "step": 1656
-    },
-    {
-      "epoch": 0.12668922147676664,
-      "grad_norm": 0.0009796343510970473,
-      "learning_rate": 0.00019999999217486347,
-      "loss": 46.0,
-      "step": 1657
-    },
-    {
-      "epoch": 0.12676567846015635,
-      "grad_norm": 0.001824269420467317,
-      "learning_rate": 0.0001999999921653583,
-      "loss": 46.0,
-      "step": 1658
-    },
-    {
-      "epoch": 0.12684213544354608,
-      "grad_norm": 0.0014029761077836156,
-      "learning_rate": 0.00019999999215584735,
-      "loss": 46.0,
-      "step": 1659
-    },
-    {
-      "epoch": 0.1269185924269358,
-      "grad_norm": 0.0010570652084425092,
-      "learning_rate": 0.00019999999214633063,
-      "loss": 46.0,
-      "step": 1660
-    },
-    {
-      "epoch": 0.12699504941032552,
-      "grad_norm": 0.003979287575930357,
-      "learning_rate": 0.00019999999213680814,
-      "loss": 46.0,
-      "step": 1661
-    },
-    {
-      "epoch": 0.12707150639371523,
-      "grad_norm": 0.0005274254945106804,
-      "learning_rate": 0.0001999999921272799,
-      "loss": 46.0,
-      "step": 1662
-    },
-    {
-      "epoch": 0.12714796337710496,
-      "grad_norm": 0.0013867878587916493,
-      "learning_rate": 0.00019999999211774586,
-      "loss": 46.0,
-      "step": 1663
-    },
-    {
-      "epoch": 0.12722442036049467,
-      "grad_norm": 0.001509237103164196,
-      "learning_rate": 0.00019999999210820607,
-      "loss": 46.0,
-      "step": 1664
-    },
-    {
-      "epoch": 0.1273008773438844,
-      "grad_norm": 0.0016890116967260838,
-      "learning_rate": 0.00019999999209866051,
-      "loss": 46.0,
-      "step": 1665
-    },
-    {
-      "epoch": 0.1273773343272741,
-      "grad_norm": 0.0016593248583376408,
-      "learning_rate": 0.00019999999208910918,
-      "loss": 46.0,
-      "step": 1666
-    },
-    {
-      "epoch": 0.12745379131066384,
-      "grad_norm": 0.0013300592545419931,
-      "learning_rate": 0.0001999999920795521,
-      "loss": 46.0,
-      "step": 1667
-    },
-    {
-      "epoch": 0.12753024829405357,
-      "grad_norm": 0.00036369453300721943,
-      "learning_rate": 0.00019999999206998922,
-      "loss": 46.0,
-      "step": 1668
-    },
-    {
-      "epoch": 0.12760670527744328,
-      "grad_norm": 0.0005816552438773215,
-      "learning_rate": 0.00019999999206042057,
-      "loss": 46.0,
-      "step": 1669
-    },
-    {
-      "epoch": 0.127683162260833,
-      "grad_norm": 0.0015860212733969092,
-      "learning_rate": 0.00019999999205084617,
-      "loss": 46.0,
-      "step": 1670
-    },
-    {
-      "epoch": 0.12775961924422272,
-      "grad_norm": 0.0007984215044416487,
-      "learning_rate": 0.00019999999204126597,
-      "loss": 46.0,
-      "step": 1671
-    },
-    {
-      "epoch": 0.12783607622761245,
-      "grad_norm": 0.0011627302737906575,
-      "learning_rate": 0.00019999999203168005,
-      "loss": 46.0,
-      "step": 1672
-    },
-    {
-      "epoch": 0.12791253321100216,
-      "grad_norm": 0.0016763482708483934,
-      "learning_rate": 0.00019999999202208833,
-      "loss": 46.0,
-      "step": 1673
-    },
-    {
-      "epoch": 0.1279889901943919,
-      "grad_norm": 0.008085369132459164,
-      "learning_rate": 0.0001999999920124908,
-      "loss": 46.0,
-      "step": 1674
-    },
-    {
-      "epoch": 0.1280654471777816,
-      "grad_norm": 0.0028306087478995323,
-      "learning_rate": 0.00019999999200288757,
-      "loss": 46.0,
-      "step": 1675
-    },
-    {
-      "epoch": 0.12814190416117133,
-      "grad_norm": 0.0013863978674635291,
-      "learning_rate": 0.00019999999199327856,
-      "loss": 46.0,
-      "step": 1676
-    },
-    {
-      "epoch": 0.12821836114456103,
-      "grad_norm": 0.000501670059747994,
-      "learning_rate": 0.00019999999198366374,
-      "loss": 46.0,
-      "step": 1677
-    },
-    {
-      "epoch": 0.12829481812795077,
-      "grad_norm": 0.0011134223314002156,
-      "learning_rate": 0.0001999999919740432,
-      "loss": 46.0,
-      "step": 1678
-    },
-    {
-      "epoch": 0.12837127511134047,
-      "grad_norm": 0.000379667297238484,
-      "learning_rate": 0.00019999999196441688,
-      "loss": 46.0,
-      "step": 1679
-    },
-    {
-      "epoch": 0.1284477320947302,
-      "grad_norm": 0.0010382885811850429,
-      "learning_rate": 0.00019999999195478477,
-      "loss": 46.0,
-      "step": 1680
-    },
-    {
-      "epoch": 0.1285241890781199,
-      "grad_norm": 0.0009940988384187222,
-      "learning_rate": 0.0001999999919451469,
-      "loss": 46.0,
-      "step": 1681
-    },
-    {
-      "epoch": 0.12860064606150964,
-      "grad_norm": 0.001528052263893187,
-      "learning_rate": 0.00019999999193550324,
-      "loss": 46.0,
-      "step": 1682
-    },
-    {
-      "epoch": 0.12867710304489938,
-      "grad_norm": 0.0008618487627245486,
-      "learning_rate": 0.00019999999192585384,
-      "loss": 46.0,
-      "step": 1683
-    },
-    {
-      "epoch": 0.12875356002828908,
-      "grad_norm": 0.0008057771483436227,
-      "learning_rate": 0.00019999999191619864,
-      "loss": 46.0,
-      "step": 1684
-    },
-    {
-      "epoch": 0.12883001701167882,
-      "grad_norm": 0.0008462500409223139,
-      "learning_rate": 0.00019999999190653772,
-      "loss": 46.0,
-      "step": 1685
-    },
-    {
-      "epoch": 0.12890647399506852,
-      "grad_norm": 0.0009672439191490412,
-      "learning_rate": 0.00019999999189687097,
-      "loss": 46.0,
-      "step": 1686
-    },
-    {
-      "epoch": 0.12898293097845825,
-      "grad_norm": 0.0020651649683713913,
-      "learning_rate": 0.00019999999188719848,
-      "loss": 46.0,
-      "step": 1687
-    },
-    {
-      "epoch": 0.12905938796184796,
-      "grad_norm": 0.0013438333990052342,
-      "learning_rate": 0.00019999999187752024,
-      "loss": 46.0,
-      "step": 1688
-    },
-    {
-      "epoch": 0.1291358449452377,
-      "grad_norm": 0.0018028083723038435,
-      "learning_rate": 0.0001999999918678362,
-      "loss": 46.0,
-      "step": 1689
-    },
-    {
-      "epoch": 0.1292123019286274,
-      "grad_norm": 0.0005871797911822796,
-      "learning_rate": 0.00019999999185814642,
-      "loss": 46.0,
-      "step": 1690
-    },
-    {
-      "epoch": 0.12928875891201713,
-      "grad_norm": 0.0031369503121823072,
-      "learning_rate": 0.00019999999184845083,
-      "loss": 46.0,
-      "step": 1691
-    },
-    {
-      "epoch": 0.12936521589540684,
-      "grad_norm": 0.0004994928021915257,
-      "learning_rate": 0.0001999999918387495,
-      "loss": 46.0,
-      "step": 1692
-    },
-    {
-      "epoch": 0.12944167287879657,
-      "grad_norm": 0.0007263327715918422,
-      "learning_rate": 0.00019999999182904242,
-      "loss": 46.0,
-      "step": 1693
-    },
-    {
-      "epoch": 0.12951812986218628,
-      "grad_norm": 0.0016645743744447827,
-      "learning_rate": 0.0001999999918193295,
-      "loss": 46.0,
-      "step": 1694
-    },
-    {
-      "epoch": 0.129594586845576,
-      "grad_norm": 0.0006018921849317849,
-      "learning_rate": 0.0001999999918096109,
-      "loss": 46.0,
-      "step": 1695
-    },
-    {
-      "epoch": 0.12967104382896572,
-      "grad_norm": 0.001130896620452404,
-      "learning_rate": 0.00019999999179988643,
-      "loss": 46.0,
-      "step": 1696
-    },
-    {
-      "epoch": 0.12974750081235545,
-      "grad_norm": 0.0007095420151017606,
-      "learning_rate": 0.00019999999179015626,
-      "loss": 46.0,
-      "step": 1697
-    },
-    {
-      "epoch": 0.12982395779574518,
-      "grad_norm": 0.0006771529442630708,
-      "learning_rate": 0.00019999999178042032,
-      "loss": 46.0,
-      "step": 1698
-    },
-    {
-      "epoch": 0.1299004147791349,
-      "grad_norm": 0.0011925175786018372,
-      "learning_rate": 0.0001999999917706786,
-      "loss": 46.0,
-      "step": 1699
-    },
-    {
-      "epoch": 0.12997687176252462,
-      "grad_norm": 0.0011560809798538685,
-      "learning_rate": 0.0001999999917609311,
-      "loss": 46.0,
-      "step": 1700
-    },
-    {
-      "epoch": 0.13005332874591433,
-      "grad_norm": 0.0009192679426632822,
-      "learning_rate": 0.00019999999175117784,
-      "loss": 46.0,
-      "step": 1701
-    },
-    {
-      "epoch": 0.13012978572930406,
-      "grad_norm": 0.001408636337146163,
-      "learning_rate": 0.00019999999174141883,
-      "loss": 46.0,
-      "step": 1702
-    },
-    {
-      "epoch": 0.13020624271269376,
-      "grad_norm": 0.0012174068251624703,
-      "learning_rate": 0.00019999999173165401,
-      "loss": 46.0,
-      "step": 1703
-    },
-    {
-      "epoch": 0.1302826996960835,
-      "grad_norm": 0.0016217316733673215,
-      "learning_rate": 0.00019999999172188343,
-      "loss": 46.0,
-      "step": 1704
-    },
-    {
-      "epoch": 0.1303591566794732,
-      "grad_norm": 0.0007246576133184135,
-      "learning_rate": 0.0001999999917121071,
-      "loss": 46.0,
-      "step": 1705
-    },
-    {
-      "epoch": 0.13043561366286294,
-      "grad_norm": 0.0011429123114794493,
-      "learning_rate": 0.000199999991702325,
-      "loss": 46.0,
-      "step": 1706
-    },
-    {
-      "epoch": 0.13051207064625264,
-      "grad_norm": 0.0015667779371142387,
-      "learning_rate": 0.0001999999916925371,
-      "loss": 46.0,
-      "step": 1707
-    },
-    {
-      "epoch": 0.13058852762964238,
-      "grad_norm": 0.0011551033239811659,
-      "learning_rate": 0.00019999999168274346,
-      "loss": 46.0,
-      "step": 1708
-    },
-    {
-      "epoch": 0.13066498461303208,
-      "grad_norm": 0.0006915386184118688,
-      "learning_rate": 0.00019999999167294403,
-      "loss": 46.0,
-      "step": 1709
-    },
-    {
-      "epoch": 0.13074144159642181,
-      "grad_norm": 0.0023293134290724993,
-      "learning_rate": 0.00019999999166313886,
-      "loss": 46.0,
-      "step": 1710
-    },
-    {
-      "epoch": 0.13081789857981152,
-      "grad_norm": 0.0008786504622548819,
-      "learning_rate": 0.0001999999916533279,
-      "loss": 46.0,
-      "step": 1711
-    },
-    {
-      "epoch": 0.13089435556320125,
-      "grad_norm": 0.0006594064179807901,
-      "learning_rate": 0.00019999999164351117,
-      "loss": 46.0,
-      "step": 1712
-    },
-    {
-      "epoch": 0.130970812546591,
-      "grad_norm": 0.0014750296249985695,
-      "learning_rate": 0.00019999999163368868,
-      "loss": 46.0,
-      "step": 1713
-    },
-    {
-      "epoch": 0.1310472695299807,
-      "grad_norm": 0.0009271716699004173,
-      "learning_rate": 0.0001999999916238604,
-      "loss": 46.0,
-      "step": 1714
-    },
-    {
-      "epoch": 0.13112372651337043,
-      "grad_norm": 0.0009438525885343552,
-      "learning_rate": 0.00019999999161402637,
-      "loss": 46.0,
-      "step": 1715
-    },
-    {
-      "epoch": 0.13120018349676013,
-      "grad_norm": 0.0027726551052182913,
-      "learning_rate": 0.0001999999916041866,
-      "loss": 46.0,
-      "step": 1716
-    },
-    {
-      "epoch": 0.13127664048014986,
-      "grad_norm": 0.0009162983624264598,
-      "learning_rate": 0.000199999991594341,
-      "loss": 46.0,
-      "step": 1717
-    },
-    {
-      "epoch": 0.13135309746353957,
-      "grad_norm": 0.002412374597042799,
-      "learning_rate": 0.00019999999158448967,
-      "loss": 46.0,
-      "step": 1718
-    },
-    {
-      "epoch": 0.1314295544469293,
-      "grad_norm": 0.0010640060063451529,
-      "learning_rate": 0.00019999999157463254,
-      "loss": 46.0,
-      "step": 1719
-    },
-    {
-      "epoch": 0.131506011430319,
-      "grad_norm": 0.0011001689126715064,
-      "learning_rate": 0.00019999999156476969,
-      "loss": 46.0,
-      "step": 1720
-    },
-    {
-      "epoch": 0.13158246841370874,
-      "grad_norm": 0.0011513471836224198,
-      "learning_rate": 0.00019999999155490103,
-      "loss": 46.0,
-      "step": 1721
-    },
-    {
-      "epoch": 0.13165892539709845,
-      "grad_norm": 0.00048143602907657623,
-      "learning_rate": 0.0001999999915450266,
-      "loss": 46.0,
-      "step": 1722
-    },
-    {
-      "epoch": 0.13173538238048818,
-      "grad_norm": 0.0010478375479578972,
-      "learning_rate": 0.00019999999153514638,
-      "loss": 46.0,
-      "step": 1723
-    },
-    {
-      "epoch": 0.1318118393638779,
-      "grad_norm": 0.001411057193763554,
-      "learning_rate": 0.00019999999152526044,
-      "loss": 46.0,
-      "step": 1724
-    },
-    {
-      "epoch": 0.13188829634726762,
-      "grad_norm": 0.0012279852526262403,
-      "learning_rate": 0.00019999999151536872,
-      "loss": 46.0,
-      "step": 1725
-    },
-    {
-      "epoch": 0.13196475333065733,
-      "grad_norm": 0.0014598125126212835,
-      "learning_rate": 0.00019999999150547123,
-      "loss": 46.0,
-      "step": 1726
-    },
-    {
-      "epoch": 0.13204121031404706,
-      "grad_norm": 0.0014014207990840077,
-      "learning_rate": 0.00019999999149556796,
-      "loss": 46.0,
-      "step": 1727
-    },
-    {
-      "epoch": 0.1321176672974368,
-      "grad_norm": 0.000739993411116302,
-      "learning_rate": 0.00019999999148565892,
-      "loss": 46.0,
-      "step": 1728
-    },
-    {
-      "epoch": 0.1321941242808265,
-      "grad_norm": 0.001423330744728446,
-      "learning_rate": 0.0001999999914757441,
-      "loss": 46.0,
-      "step": 1729
-    },
-    {
-      "epoch": 0.13227058126421623,
-      "grad_norm": 0.0021808992605656385,
-      "learning_rate": 0.00019999999146582352,
-      "loss": 46.0,
-      "step": 1730
-    },
-    {
-      "epoch": 0.13234703824760594,
-      "grad_norm": 0.000990840489976108,
-      "learning_rate": 0.0001999999914558972,
-      "loss": 46.0,
-      "step": 1731
-    },
-    {
-      "epoch": 0.13242349523099567,
-      "grad_norm": 0.0009367921738885343,
-      "learning_rate": 0.00019999999144596506,
-      "loss": 46.0,
-      "step": 1732
-    },
-    {
-      "epoch": 0.13249995221438537,
-      "grad_norm": 0.001360927359201014,
-      "learning_rate": 0.00019999999143602718,
-      "loss": 46.0,
-      "step": 1733
-    },
-    {
-      "epoch": 0.1325764091977751,
-      "grad_norm": 0.0020556238014250994,
-      "learning_rate": 0.00019999999142608353,
-      "loss": 46.0,
-      "step": 1734
-    },
-    {
-      "epoch": 0.1326528661811648,
-      "grad_norm": 0.0017063411651179194,
-      "learning_rate": 0.00019999999141613413,
-      "loss": 46.0,
-      "step": 1735
-    },
-    {
-      "epoch": 0.13272932316455455,
-      "grad_norm": 0.0010955114848911762,
-      "learning_rate": 0.00019999999140617893,
-      "loss": 46.0,
-      "step": 1736
-    },
-    {
-      "epoch": 0.13280578014794425,
-      "grad_norm": 0.0020986285526305437,
-      "learning_rate": 0.00019999999139621796,
-      "loss": 46.0,
-      "step": 1737
-    },
-    {
-      "epoch": 0.13288223713133399,
-      "grad_norm": 0.0014130845665931702,
-      "learning_rate": 0.00019999999138625122,
-      "loss": 46.0,
-      "step": 1738
-    },
-    {
-      "epoch": 0.1329586941147237,
-      "grad_norm": 0.0009628068073652685,
-      "learning_rate": 0.00019999999137627872,
-      "loss": 46.0,
-      "step": 1739
-    },
-    {
-      "epoch": 0.13303515109811342,
-      "grad_norm": 0.0011783394729718566,
-      "learning_rate": 0.00019999999136630046,
-      "loss": 46.0,
-      "step": 1740
-    },
-    {
-      "epoch": 0.13311160808150316,
-      "grad_norm": 0.001116079045459628,
-      "learning_rate": 0.00019999999135631642,
-      "loss": 46.0,
-      "step": 1741
-    },
-    {
-      "epoch": 0.13318806506489286,
-      "grad_norm": 0.0008289802353829145,
-      "learning_rate": 0.00019999999134632658,
-      "loss": 46.0,
-      "step": 1742
-    },
-    {
-      "epoch": 0.1332645220482826,
-      "grad_norm": 0.0019486815435811877,
-      "learning_rate": 0.00019999999133633102,
-      "loss": 46.0,
-      "step": 1743
-    },
-    {
-      "epoch": 0.1333409790316723,
-      "grad_norm": 0.0013661953853443265,
-      "learning_rate": 0.00019999999132632967,
-      "loss": 46.0,
-      "step": 1744
-    },
-    {
-      "epoch": 0.13341743601506204,
-      "grad_norm": 0.0009323824779130518,
-      "learning_rate": 0.00019999999131632253,
-      "loss": 46.0,
-      "step": 1745
-    },
-    {
-      "epoch": 0.13349389299845174,
-      "grad_norm": 0.0030559145379811525,
-      "learning_rate": 0.00019999999130630966,
-      "loss": 46.0,
-      "step": 1746
-    },
-    {
-      "epoch": 0.13357034998184147,
-      "grad_norm": 0.0009545132634229958,
-      "learning_rate": 0.000199999991296291,
-      "loss": 46.0,
-      "step": 1747
-    },
-    {
-      "epoch": 0.13364680696523118,
-      "grad_norm": 0.002411074936389923,
-      "learning_rate": 0.00019999999128626658,
-      "loss": 46.0,
-      "step": 1748
-    },
-    {
-      "epoch": 0.1337232639486209,
-      "grad_norm": 0.0011840452207252383,
-      "learning_rate": 0.00019999999127623638,
-      "loss": 46.0,
-      "step": 1749
-    },
-    {
-      "epoch": 0.13379972093201062,
-      "grad_norm": 0.0012516540009528399,
-      "learning_rate": 0.0001999999912662004,
-      "loss": 46.0,
-      "step": 1750
-    },
-    {
-      "epoch": 0.13387617791540035,
-      "grad_norm": 0.0007451776182278991,
-      "learning_rate": 0.0001999999912561587,
-      "loss": 46.0,
-      "step": 1751
-    },
-    {
-      "epoch": 0.13395263489879006,
-      "grad_norm": 0.0010144745465368032,
-      "learning_rate": 0.0001999999912461112,
-      "loss": 46.0,
-      "step": 1752
-    },
-    {
-      "epoch": 0.1340290918821798,
-      "grad_norm": 0.0008639212464913726,
-      "learning_rate": 0.0001999999912360579,
-      "loss": 46.0,
-      "step": 1753
-    },
-    {
-      "epoch": 0.1341055488655695,
-      "grad_norm": 0.005747777875512838,
-      "learning_rate": 0.00019999999122599887,
-      "loss": 46.0,
-      "step": 1754
-    },
-    {
-      "epoch": 0.13418200584895923,
-      "grad_norm": 0.0015556936850771308,
-      "learning_rate": 0.00019999999121593403,
-      "loss": 46.0,
-      "step": 1755
-    },
-    {
-      "epoch": 0.13425846283234896,
-      "grad_norm": 0.0008981954888440669,
-      "learning_rate": 0.00019999999120586348,
-      "loss": 46.0,
-      "step": 1756
-    },
-    {
-      "epoch": 0.13433491981573867,
-      "grad_norm": 0.0012614093720912933,
-      "learning_rate": 0.00019999999119578714,
-      "loss": 46.0,
-      "step": 1757
-    },
-    {
-      "epoch": 0.1344113767991284,
-      "grad_norm": 0.0021436798851937056,
-      "learning_rate": 0.000199999991185705,
-      "loss": 46.0,
-      "step": 1758
-    },
-    {
-      "epoch": 0.1344878337825181,
-      "grad_norm": 0.0009233438177034259,
-      "learning_rate": 0.0001999999911756171,
-      "loss": 46.0,
-      "step": 1759
-    },
-    {
-      "epoch": 0.13456429076590784,
-      "grad_norm": 0.0005031541222706437,
-      "learning_rate": 0.00019999999116552343,
-      "loss": 46.0,
-      "step": 1760
-    },
-    {
-      "epoch": 0.13464074774929755,
-      "grad_norm": 0.0016876172740012407,
-      "learning_rate": 0.00019999999115542403,
-      "loss": 46.0,
-      "step": 1761
-    },
-    {
-      "epoch": 0.13471720473268728,
-      "grad_norm": 0.0007119384245015681,
-      "learning_rate": 0.00019999999114531883,
-      "loss": 46.0,
-      "step": 1762
-    },
-    {
-      "epoch": 0.13479366171607698,
-      "grad_norm": 0.001981558511033654,
-      "learning_rate": 0.00019999999113520786,
-      "loss": 46.0,
-      "step": 1763
-    },
-    {
-      "epoch": 0.13487011869946672,
-      "grad_norm": 0.00034314021468162537,
-      "learning_rate": 0.00019999999112509112,
-      "loss": 46.0,
-      "step": 1764
-    },
-    {
-      "epoch": 0.13494657568285642,
-      "grad_norm": 0.0012478202115744352,
-      "learning_rate": 0.0001999999911149686,
-      "loss": 46.0,
-      "step": 1765
-    },
-    {
-      "epoch": 0.13502303266624616,
-      "grad_norm": 0.0010478689800947905,
-      "learning_rate": 0.00019999999110484034,
-      "loss": 46.0,
-      "step": 1766
-    },
-    {
-      "epoch": 0.13509948964963586,
-      "grad_norm": 0.0008405048865824938,
-      "learning_rate": 0.0001999999910947063,
-      "loss": 46.0,
-      "step": 1767
-    },
-    {
-      "epoch": 0.1351759466330256,
-      "grad_norm": 0.0012383825378492475,
-      "learning_rate": 0.00019999999108456646,
-      "loss": 46.0,
-      "step": 1768
-    },
-    {
-      "epoch": 0.1352524036164153,
-      "grad_norm": 0.0007070383871905506,
-      "learning_rate": 0.0001999999910744209,
-      "loss": 46.0,
-      "step": 1769
-    },
-    {
-      "epoch": 0.13532886059980503,
-      "grad_norm": 0.0011966886231675744,
-      "learning_rate": 0.00019999999106426954,
-      "loss": 46.0,
-      "step": 1770
-    },
-    {
-      "epoch": 0.13540531758319477,
-      "grad_norm": 0.0036497600376605988,
-      "learning_rate": 0.0001999999910541124,
-      "loss": 46.0,
-      "step": 1771
-    },
-    {
-      "epoch": 0.13548177456658447,
-      "grad_norm": 0.0010715022217482328,
-      "learning_rate": 0.00019999999104394954,
-      "loss": 46.0,
-      "step": 1772
-    },
-    {
-      "epoch": 0.1355582315499742,
-      "grad_norm": 0.0021324383560568094,
-      "learning_rate": 0.00019999999103378086,
-      "loss": 46.0,
-      "step": 1773
-    },
-    {
-      "epoch": 0.1356346885333639,
-      "grad_norm": 0.0011359077179804444,
-      "learning_rate": 0.00019999999102360643,
-      "loss": 46.0,
-      "step": 1774
-    },
-    {
-      "epoch": 0.13571114551675365,
-      "grad_norm": 0.000986409606412053,
-      "learning_rate": 0.0001999999910134262,
-      "loss": 46.0,
-      "step": 1775
-    },
-    {
-      "epoch": 0.13578760250014335,
-      "grad_norm": 0.0005540827987715602,
-      "learning_rate": 0.00019999999100324024,
-      "loss": 46.0,
-      "step": 1776
-    },
-    {
-      "epoch": 0.13586405948353308,
-      "grad_norm": 0.000653906783554703,
-      "learning_rate": 0.00019999999099304852,
-      "loss": 46.0,
-      "step": 1777
-    },
-    {
-      "epoch": 0.1359405164669228,
-      "grad_norm": 0.0011749338591471314,
-      "learning_rate": 0.000199999990982851,
-      "loss": 46.0,
-      "step": 1778
-    },
-    {
-      "epoch": 0.13601697345031252,
-      "grad_norm": 0.0011938236420974135,
-      "learning_rate": 0.0001999999909726477,
-      "loss": 46.0,
-      "step": 1779
-    },
-    {
-      "epoch": 0.13609343043370223,
-      "grad_norm": 0.0019144678954035044,
-      "learning_rate": 0.0001999999909624387,
-      "loss": 46.0,
-      "step": 1780
-    },
-    {
-      "epoch": 0.13616988741709196,
-      "grad_norm": 0.0009658513008616865,
-      "learning_rate": 0.00019999999095222384,
-      "loss": 46.0,
-      "step": 1781
-    },
-    {
-      "epoch": 0.13624634440048167,
-      "grad_norm": 0.0012276963097974658,
-      "learning_rate": 0.00019999999094200325,
-      "loss": 46.0,
-      "step": 1782
-    },
-    {
-      "epoch": 0.1363228013838714,
-      "grad_norm": 0.0007017145399004221,
-      "learning_rate": 0.0001999999909317769,
-      "loss": 46.0,
-      "step": 1783
-    },
-    {
-      "epoch": 0.1363992583672611,
-      "grad_norm": 0.002202402101829648,
-      "learning_rate": 0.0001999999909215448,
-      "loss": 46.0,
-      "step": 1784
-    },
-    {
-      "epoch": 0.13647571535065084,
-      "grad_norm": 0.003947373945266008,
-      "learning_rate": 0.0001999999909113069,
-      "loss": 46.0,
-      "step": 1785
-    },
-    {
-      "epoch": 0.13655217233404057,
-      "grad_norm": 0.0015952371759340167,
-      "learning_rate": 0.0001999999909010632,
-      "loss": 46.0,
-      "step": 1786
-    },
-    {
-      "epoch": 0.13662862931743028,
-      "grad_norm": 0.001507932785898447,
-      "learning_rate": 0.00019999999089081379,
-      "loss": 46.0,
-      "step": 1787
-    },
-    {
-      "epoch": 0.13670508630082,
-      "grad_norm": 0.0004487130790948868,
-      "learning_rate": 0.0001999999908805586,
-      "loss": 46.0,
-      "step": 1788
-    },
-    {
-      "epoch": 0.13678154328420972,
-      "grad_norm": 0.0010462729260325432,
-      "learning_rate": 0.0001999999908702976,
-      "loss": 46.0,
-      "step": 1789
-    },
-    {
-      "epoch": 0.13685800026759945,
-      "grad_norm": 0.0009094856213778257,
-      "learning_rate": 0.00019999999086003087,
-      "loss": 46.0,
-      "step": 1790
-    },
-    {
-      "epoch": 0.13693445725098916,
-      "grad_norm": 0.0011709691025316715,
-      "learning_rate": 0.00019999999084975833,
-      "loss": 46.0,
-      "step": 1791
-    },
-    {
-      "epoch": 0.1370109142343789,
-      "grad_norm": 0.002708164742216468,
-      "learning_rate": 0.00019999999083948006,
-      "loss": 46.0,
-      "step": 1792
-    },
-    {
-      "epoch": 0.1370873712177686,
-      "grad_norm": 0.001033182255923748,
-      "learning_rate": 0.00019999999082919603,
-      "loss": 46.0,
-      "step": 1793
-    },
-    {
-      "epoch": 0.13716382820115833,
-      "grad_norm": 0.0006695556221529841,
-      "learning_rate": 0.0001999999908189062,
-      "loss": 46.0,
-      "step": 1794
-    },
-    {
-      "epoch": 0.13724028518454803,
-      "grad_norm": 0.002639173064380884,
-      "learning_rate": 0.0001999999908086106,
-      "loss": 46.0,
-      "step": 1795
-    },
-    {
-      "epoch": 0.13731674216793777,
-      "grad_norm": 0.0016310201026499271,
-      "learning_rate": 0.00019999999079830925,
-      "loss": 46.0,
-      "step": 1796
-    },
-    {
-      "epoch": 0.13739319915132747,
-      "grad_norm": 0.00037465078639797866,
-      "learning_rate": 0.00019999999078800212,
-      "loss": 46.0,
-      "step": 1797
-    },
-    {
-      "epoch": 0.1374696561347172,
-      "grad_norm": 0.00067168939858675,
-      "learning_rate": 0.00019999999077768921,
-      "loss": 46.0,
-      "step": 1798
-    },
-    {
-      "epoch": 0.13754611311810694,
-      "grad_norm": 0.00045956438407301903,
-      "learning_rate": 0.00019999999076737054,
-      "loss": 46.0,
-      "step": 1799
-    },
-    {
-      "epoch": 0.13762257010149664,
-      "grad_norm": 0.00042533682426437736,
-      "learning_rate": 0.0001999999907570461,
-      "loss": 46.0,
-      "step": 1800
-    },
-    {
-      "epoch": 0.13769902708488638,
-      "grad_norm": 0.004772007465362549,
-      "learning_rate": 0.00019999999074671592,
-      "loss": 46.0,
-      "step": 1801
-    },
-    {
-      "epoch": 0.13777548406827608,
-      "grad_norm": 0.0006424225866794586,
-      "learning_rate": 0.00019999999073637992,
-      "loss": 46.0,
-      "step": 1802
-    },
-    {
-      "epoch": 0.13785194105166582,
-      "grad_norm": 0.0012661645887419581,
-      "learning_rate": 0.00019999999072603818,
-      "loss": 46.0,
-      "step": 1803
-    },
-    {
-      "epoch": 0.13792839803505552,
-      "grad_norm": 0.0010105270193889737,
-      "learning_rate": 0.00019999999071569066,
-      "loss": 46.0,
-      "step": 1804
-    },
-    {
-      "epoch": 0.13800485501844525,
-      "grad_norm": 0.001936870045028627,
-      "learning_rate": 0.00019999999070533737,
-      "loss": 46.0,
-      "step": 1805
-    },
-    {
-      "epoch": 0.13808131200183496,
-      "grad_norm": 0.0015983741031959653,
-      "learning_rate": 0.00019999999069497833,
-      "loss": 46.0,
-      "step": 1806
-    },
-    {
-      "epoch": 0.1381577689852247,
-      "grad_norm": 0.0010567340068519115,
-      "learning_rate": 0.0001999999906846135,
-      "loss": 46.0,
-      "step": 1807
-    },
-    {
-      "epoch": 0.1382342259686144,
-      "grad_norm": 0.004460916388779879,
-      "learning_rate": 0.0001999999906742429,
-      "loss": 46.0,
-      "step": 1808
-    },
-    {
-      "epoch": 0.13831068295200413,
-      "grad_norm": 0.0007848286768421531,
-      "learning_rate": 0.00019999999066386655,
-      "loss": 46.0,
-      "step": 1809
-    },
-    {
-      "epoch": 0.13838713993539384,
-      "grad_norm": 0.0028831337112933397,
-      "learning_rate": 0.0001999999906534844,
-      "loss": 46.0,
-      "step": 1810
-    },
-    {
-      "epoch": 0.13846359691878357,
-      "grad_norm": 0.002073341980576515,
-      "learning_rate": 0.00019999999064309652,
-      "loss": 46.0,
-      "step": 1811
-    },
-    {
-      "epoch": 0.13854005390217328,
-      "grad_norm": 0.0007669876213185489,
-      "learning_rate": 0.00019999999063270284,
-      "loss": 46.0,
-      "step": 1812
-    },
-    {
-      "epoch": 0.138616510885563,
-      "grad_norm": 0.0014372995356097817,
-      "learning_rate": 0.0001999999906223034,
-      "loss": 46.0,
-      "step": 1813
-    },
-    {
-      "epoch": 0.13869296786895274,
-      "grad_norm": 0.0006090968963690102,
-      "learning_rate": 0.0001999999906118982,
-      "loss": 46.0,
-      "step": 1814
-    },
-    {
-      "epoch": 0.13876942485234245,
-      "grad_norm": 0.0007370939711108804,
-      "learning_rate": 0.0001999999906014872,
-      "loss": 46.0,
-      "step": 1815
-    },
-    {
-      "epoch": 0.13884588183573218,
-      "grad_norm": 0.0013098521158099174,
-      "learning_rate": 0.00019999999059107046,
-      "loss": 46.0,
-      "step": 1816
-    },
-    {
-      "epoch": 0.1389223388191219,
-      "grad_norm": 0.0012884825700893998,
-      "learning_rate": 0.00019999999058064794,
-      "loss": 46.0,
-      "step": 1817
-    },
-    {
-      "epoch": 0.13899879580251162,
-      "grad_norm": 0.0009428421035408974,
-      "learning_rate": 0.00019999999057021965,
-      "loss": 46.0,
-      "step": 1818
-    },
-    {
-      "epoch": 0.13907525278590133,
-      "grad_norm": 0.001111352350562811,
-      "learning_rate": 0.00019999999055978558,
-      "loss": 46.0,
-      "step": 1819
-    },
-    {
-      "epoch": 0.13915170976929106,
-      "grad_norm": 0.0011604103492572904,
-      "learning_rate": 0.00019999999054934575,
-      "loss": 46.0,
-      "step": 1820
-    },
-    {
-      "epoch": 0.13922816675268077,
-      "grad_norm": 0.0011482506524771452,
-      "learning_rate": 0.00019999999053890016,
-      "loss": 46.0,
-      "step": 1821
-    },
-    {
-      "epoch": 0.1393046237360705,
-      "grad_norm": 0.002155241323634982,
-      "learning_rate": 0.0001999999905284488,
-      "loss": 46.0,
-      "step": 1822
-    },
-    {
-      "epoch": 0.1393810807194602,
-      "grad_norm": 0.0007282591541297734,
-      "learning_rate": 0.00019999999051799168,
-      "loss": 46.0,
-      "step": 1823
-    },
-    {
-      "epoch": 0.13945753770284994,
-      "grad_norm": 0.0009172306163236499,
-      "learning_rate": 0.00019999999050752875,
-      "loss": 46.0,
-      "step": 1824
-    },
-    {
-      "epoch": 0.13953399468623964,
-      "grad_norm": 0.0009994545252993703,
-      "learning_rate": 0.0001999999904970601,
-      "loss": 46.0,
-      "step": 1825
-    },
-    {
-      "epoch": 0.13961045166962938,
-      "grad_norm": 0.000805270450655371,
-      "learning_rate": 0.00019999999048658565,
-      "loss": 46.0,
-      "step": 1826
-    },
-    {
-      "epoch": 0.13968690865301908,
-      "grad_norm": 0.0014365249080583453,
-      "learning_rate": 0.00019999999047610545,
-      "loss": 46.0,
-      "step": 1827
-    },
-    {
-      "epoch": 0.13976336563640882,
-      "grad_norm": 0.001337170717306435,
-      "learning_rate": 0.00019999999046561943,
-      "loss": 46.0,
-      "step": 1828
-    },
-    {
-      "epoch": 0.13983982261979855,
-      "grad_norm": 0.0025257414672523737,
-      "learning_rate": 0.00019999999045512768,
-      "loss": 46.0,
-      "step": 1829
-    },
-    {
-      "epoch": 0.13991627960318825,
-      "grad_norm": 0.001134646707214415,
-      "learning_rate": 0.00019999999044463017,
-      "loss": 46.0,
-      "step": 1830
-    },
-    {
-      "epoch": 0.139992736586578,
-      "grad_norm": 0.0008038088562898338,
-      "learning_rate": 0.0001999999904341269,
-      "loss": 46.0,
-      "step": 1831
-    },
-    {
-      "epoch": 0.1400691935699677,
-      "grad_norm": 0.003690134035423398,
-      "learning_rate": 0.00019999999042361784,
-      "loss": 46.0,
-      "step": 1832
-    },
-    {
-      "epoch": 0.14014565055335743,
-      "grad_norm": 0.0011938436655327678,
-      "learning_rate": 0.000199999990413103,
-      "loss": 46.0,
-      "step": 1833
-    },
-    {
-      "epoch": 0.14022210753674713,
-      "grad_norm": 0.0008435413474217057,
-      "learning_rate": 0.0001999999904025824,
-      "loss": 46.0,
-      "step": 1834
-    },
-    {
-      "epoch": 0.14029856452013686,
-      "grad_norm": 0.0007511136354878545,
-      "learning_rate": 0.000199999990392056,
-      "loss": 46.0,
-      "step": 1835
-    },
-    {
-      "epoch": 0.14037502150352657,
-      "grad_norm": 0.0016807162901386619,
-      "learning_rate": 0.0001999999903815239,
-      "loss": 46.0,
-      "step": 1836
-    },
-    {
-      "epoch": 0.1404514784869163,
-      "grad_norm": 0.001615367946214974,
-      "learning_rate": 0.00019999999037098598,
-      "loss": 46.0,
-      "step": 1837
-    },
-    {
-      "epoch": 0.140527935470306,
-      "grad_norm": 0.000741770607419312,
-      "learning_rate": 0.0001999999903604423,
-      "loss": 46.0,
-      "step": 1838
-    },
-    {
-      "epoch": 0.14060439245369574,
-      "grad_norm": 0.0012382975546643138,
-      "learning_rate": 0.00019999999034989285,
-      "loss": 46.0,
-      "step": 1839
-    },
-    {
-      "epoch": 0.14068084943708545,
-      "grad_norm": 0.00300770765170455,
-      "learning_rate": 0.00019999999033933766,
-      "loss": 46.0,
-      "step": 1840
-    },
-    {
-      "epoch": 0.14075730642047518,
-      "grad_norm": 0.0013860536273568869,
-      "learning_rate": 0.00019999999032877666,
-      "loss": 46.0,
-      "step": 1841
-    },
-    {
-      "epoch": 0.1408337634038649,
-      "grad_norm": 0.001102592796087265,
-      "learning_rate": 0.0001999999903182099,
-      "loss": 46.0,
-      "step": 1842
-    },
-    {
-      "epoch": 0.14091022038725462,
-      "grad_norm": 0.0020995796658098698,
-      "learning_rate": 0.00019999999030763737,
-      "loss": 46.0,
-      "step": 1843
-    },
-    {
-      "epoch": 0.14098667737064435,
-      "grad_norm": 0.001294348039664328,
-      "learning_rate": 0.00019999999029705906,
-      "loss": 46.0,
-      "step": 1844
-    },
-    {
-      "epoch": 0.14106313435403406,
-      "grad_norm": 0.0011832526652142406,
-      "learning_rate": 0.000199999990286475,
-      "loss": 46.0,
-      "step": 1845
-    },
-    {
-      "epoch": 0.1411395913374238,
-      "grad_norm": 0.0004684127343352884,
-      "learning_rate": 0.00019999999027588519,
-      "loss": 46.0,
-      "step": 1846
-    },
-    {
-      "epoch": 0.1412160483208135,
-      "grad_norm": 0.0010823906632140279,
-      "learning_rate": 0.0001999999902652896,
-      "loss": 46.0,
-      "step": 1847
-    },
-    {
-      "epoch": 0.14129250530420323,
-      "grad_norm": 0.001460356405004859,
-      "learning_rate": 0.00019999999025468822,
-      "loss": 46.0,
-      "step": 1848
-    },
-    {
-      "epoch": 0.14136896228759294,
-      "grad_norm": 0.010155671276152134,
-      "learning_rate": 0.00019999999024408106,
-      "loss": 46.0,
-      "step": 1849
-    },
-    {
-      "epoch": 0.14144541927098267,
-      "grad_norm": 0.001174120930954814,
-      "learning_rate": 0.00019999999023346816,
-      "loss": 46.0,
-      "step": 1850
-    },
-    {
-      "epoch": 0.14152187625437238,
-      "grad_norm": 0.002394245006144047,
-      "learning_rate": 0.00019999999022284948,
-      "loss": 46.0,
-      "step": 1851
-    },
-    {
-      "epoch": 0.1415983332377621,
-      "grad_norm": 0.0011494889622554183,
-      "learning_rate": 0.000199999990212225,
-      "loss": 46.0,
-      "step": 1852
-    },
-    {
-      "epoch": 0.14167479022115181,
-      "grad_norm": 0.0013915582094341516,
-      "learning_rate": 0.0001999999902015948,
-      "loss": 46.0,
-      "step": 1853
-    },
-    {
-      "epoch": 0.14175124720454155,
-      "grad_norm": 0.0006613280274905264,
-      "learning_rate": 0.00019999999019095884,
-      "loss": 46.0,
-      "step": 1854
-    },
-    {
-      "epoch": 0.14182770418793125,
-      "grad_norm": 0.0019673940259963274,
-      "learning_rate": 0.00019999999018031707,
-      "loss": 46.0,
-      "step": 1855
-    },
-    {
-      "epoch": 0.14190416117132099,
-      "grad_norm": 0.0023109884932637215,
-      "learning_rate": 0.00019999999016966956,
-      "loss": 46.0,
-      "step": 1856
-    },
-    {
-      "epoch": 0.14198061815471072,
-      "grad_norm": 0.0017937744269147515,
-      "learning_rate": 0.00019999999015901624,
-      "loss": 46.0,
-      "step": 1857
-    },
-    {
-      "epoch": 0.14205707513810042,
-      "grad_norm": 0.0016560000367462635,
-      "learning_rate": 0.00019999999014835718,
-      "loss": 46.0,
-      "step": 1858
-    },
-    {
-      "epoch": 0.14213353212149016,
-      "grad_norm": 0.0010846997611224651,
-      "learning_rate": 0.00019999999013769232,
-      "loss": 46.0,
-      "step": 1859
-    },
-    {
-      "epoch": 0.14220998910487986,
-      "grad_norm": 0.004100044257938862,
-      "learning_rate": 0.00019999999012702173,
-      "loss": 46.0,
-      "step": 1860
-    },
-    {
-      "epoch": 0.1422864460882696,
-      "grad_norm": 0.0015313801122829318,
-      "learning_rate": 0.00019999999011634535,
-      "loss": 46.0,
-      "step": 1861
-    },
-    {
-      "epoch": 0.1423629030716593,
-      "grad_norm": 0.0009085564524866641,
-      "learning_rate": 0.00019999999010566322,
-      "loss": 46.0,
-      "step": 1862
-    },
-    {
-      "epoch": 0.14243936005504904,
-      "grad_norm": 0.0016520225908607244,
-      "learning_rate": 0.0001999999900949753,
-      "loss": 46.0,
-      "step": 1863
-    },
-    {
-      "epoch": 0.14251581703843874,
-      "grad_norm": 0.0022325743921101093,
-      "learning_rate": 0.00019999999008428162,
-      "loss": 46.0,
-      "step": 1864
-    },
-    {
-      "epoch": 0.14259227402182847,
-      "grad_norm": 0.00046500880853272974,
-      "learning_rate": 0.00019999999007358217,
-      "loss": 46.0,
-      "step": 1865
-    },
-    {
-      "epoch": 0.14266873100521818,
-      "grad_norm": 0.0008897559018805623,
-      "learning_rate": 0.00019999999006287692,
-      "loss": 46.0,
-      "step": 1866
-    },
-    {
-      "epoch": 0.1427451879886079,
-      "grad_norm": 0.000936188967898488,
-      "learning_rate": 0.00019999999005216595,
-      "loss": 46.0,
-      "step": 1867
-    },
-    {
-      "epoch": 0.14282164497199762,
-      "grad_norm": 0.002499505877494812,
-      "learning_rate": 0.00019999999004144918,
-      "loss": 46.0,
-      "step": 1868
-    },
-    {
-      "epoch": 0.14289810195538735,
-      "grad_norm": 0.0015446791658177972,
-      "learning_rate": 0.00019999999003072664,
-      "loss": 46.0,
-      "step": 1869
-    },
-    {
-      "epoch": 0.14297455893877706,
-      "grad_norm": 0.0025301750283688307,
-      "learning_rate": 0.00019999999001999838,
-      "loss": 46.0,
-      "step": 1870
-    },
-    {
-      "epoch": 0.1430510159221668,
-      "grad_norm": 0.0008142556180246174,
-      "learning_rate": 0.0001999999900092643,
-      "loss": 46.0,
-      "step": 1871
-    },
-    {
-      "epoch": 0.14312747290555652,
-      "grad_norm": 0.0009948518127202988,
-      "learning_rate": 0.00019999998999852445,
-      "loss": 46.0,
-      "step": 1872
-    },
-    {
-      "epoch": 0.14320392988894623,
-      "grad_norm": 0.0011949791805818677,
-      "learning_rate": 0.00019999998998777884,
-      "loss": 46.0,
-      "step": 1873
-    },
-    {
-      "epoch": 0.14328038687233596,
-      "grad_norm": 0.0004118412034586072,
-      "learning_rate": 0.00019999998997702746,
-      "loss": 46.0,
-      "step": 1874
-    },
-    {
-      "epoch": 0.14335684385572567,
-      "grad_norm": 0.002504717791453004,
-      "learning_rate": 0.00019999998996627033,
-      "loss": 46.0,
-      "step": 1875
-    },
-    {
-      "epoch": 0.1434333008391154,
-      "grad_norm": 0.0026481417007744312,
-      "learning_rate": 0.0001999999899555074,
-      "loss": 46.0,
-      "step": 1876
-    },
-    {
-      "epoch": 0.1435097578225051,
-      "grad_norm": 0.003533893497660756,
-      "learning_rate": 0.00019999998994473873,
-      "loss": 46.0,
-      "step": 1877
-    },
-    {
-      "epoch": 0.14358621480589484,
-      "grad_norm": 0.008359206840395927,
-      "learning_rate": 0.00019999998993396428,
-      "loss": 46.0,
-      "step": 1878
-    },
-    {
-      "epoch": 0.14366267178928455,
-      "grad_norm": 0.0005306825041770935,
-      "learning_rate": 0.00019999998992318405,
-      "loss": 46.0,
-      "step": 1879
-    },
-    {
-      "epoch": 0.14373912877267428,
-      "grad_norm": 0.0011779069900512695,
-      "learning_rate": 0.00019999998991239806,
-      "loss": 46.0,
-      "step": 1880
-    },
-    {
-      "epoch": 0.14381558575606398,
-      "grad_norm": 0.0020184062886983156,
-      "learning_rate": 0.0001999999899016063,
-      "loss": 46.0,
-      "step": 1881
-    },
-    {
-      "epoch": 0.14389204273945372,
-      "grad_norm": 0.0015682012308388948,
-      "learning_rate": 0.00019999998989080875,
-      "loss": 46.0,
-      "step": 1882
-    },
-    {
-      "epoch": 0.14396849972284342,
-      "grad_norm": 0.0009336181683465838,
-      "learning_rate": 0.00019999998988000543,
-      "loss": 46.0,
-      "step": 1883
-    },
-    {
-      "epoch": 0.14404495670623316,
-      "grad_norm": 0.0010547672864049673,
-      "learning_rate": 0.0001999999898691964,
-      "loss": 46.0,
-      "step": 1884
-    },
-    {
-      "epoch": 0.14412141368962286,
-      "grad_norm": 0.0007349682855419815,
-      "learning_rate": 0.00019999998985838153,
-      "loss": 46.0,
-      "step": 1885
-    },
-    {
-      "epoch": 0.1441978706730126,
-      "grad_norm": 0.0011008752044290304,
-      "learning_rate": 0.00019999998984756093,
-      "loss": 46.0,
-      "step": 1886
-    },
-    {
-      "epoch": 0.14427432765640233,
-      "grad_norm": 0.0026120059192180634,
-      "learning_rate": 0.00019999998983673454,
-      "loss": 46.0,
-      "step": 1887
-    },
-    {
-      "epoch": 0.14435078463979203,
-      "grad_norm": 0.001931669656187296,
-      "learning_rate": 0.0001999999898259024,
-      "loss": 46.0,
-      "step": 1888
-    },
-    {
-      "epoch": 0.14442724162318177,
-      "grad_norm": 0.0013773287646472454,
-      "learning_rate": 0.00019999998981506446,
-      "loss": 46.0,
-      "step": 1889
-    },
-    {
-      "epoch": 0.14450369860657147,
-      "grad_norm": 0.0021041342988610268,
-      "learning_rate": 0.00019999998980422078,
-      "loss": 46.0,
-      "step": 1890
-    },
-    {
-      "epoch": 0.1445801555899612,
-      "grad_norm": 0.0014199084835126996,
-      "learning_rate": 0.0001999999897933713,
-      "loss": 46.0,
-      "step": 1891
-    },
-    {
-      "epoch": 0.1446566125733509,
-      "grad_norm": 0.0007218638784252107,
-      "learning_rate": 0.0001999999897825161,
-      "loss": 46.0,
-      "step": 1892
-    },
-    {
-      "epoch": 0.14473306955674065,
-      "grad_norm": 0.0005682877381332219,
-      "learning_rate": 0.00019999998977165512,
-      "loss": 46.0,
-      "step": 1893
-    },
-    {
-      "epoch": 0.14480952654013035,
-      "grad_norm": 0.0018532413523644209,
-      "learning_rate": 0.00019999998976078832,
-      "loss": 46.0,
-      "step": 1894
-    },
-    {
-      "epoch": 0.14488598352352008,
-      "grad_norm": 0.0009956990834325552,
-      "learning_rate": 0.0001999999897499158,
-      "loss": 46.0,
-      "step": 1895
-    },
-    {
-      "epoch": 0.1449624405069098,
-      "grad_norm": 0.0023877564817667007,
-      "learning_rate": 0.0001999999897390375,
-      "loss": 46.0,
-      "step": 1896
-    },
-    {
-      "epoch": 0.14503889749029952,
-      "grad_norm": 0.0008169774082489312,
-      "learning_rate": 0.00019999998972815343,
-      "loss": 46.0,
-      "step": 1897
-    },
-    {
-      "epoch": 0.14511535447368923,
-      "grad_norm": 0.0012659343192353845,
-      "learning_rate": 0.00019999998971726357,
-      "loss": 46.0,
-      "step": 1898
-    },
-    {
-      "epoch": 0.14519181145707896,
-      "grad_norm": 0.0036854087375104427,
-      "learning_rate": 0.00019999998970636796,
-      "loss": 46.0,
-      "step": 1899
-    },
-    {
-      "epoch": 0.14526826844046867,
-      "grad_norm": 0.001317733433097601,
-      "learning_rate": 0.00019999998969546658,
-      "loss": 46.0,
-      "step": 1900
-    },
-    {
-      "epoch": 0.1453447254238584,
-      "grad_norm": 0.0017411282751709223,
-      "learning_rate": 0.00019999998968455942,
-      "loss": 46.0,
-      "step": 1901
-    },
-    {
-      "epoch": 0.14542118240724813,
-      "grad_norm": 0.0010978442151099443,
-      "learning_rate": 0.0001999999896736465,
-      "loss": 46.0,
-      "step": 1902
-    },
-    {
-      "epoch": 0.14549763939063784,
-      "grad_norm": 0.0007700786809436977,
-      "learning_rate": 0.00019999998966272782,
-      "loss": 46.0,
-      "step": 1903
-    },
-    {
-      "epoch": 0.14557409637402757,
-      "grad_norm": 0.001506231026723981,
-      "learning_rate": 0.00019999998965180334,
-      "loss": 46.0,
-      "step": 1904
-    },
-    {
-      "epoch": 0.14565055335741728,
-      "grad_norm": 0.0007330658263526857,
-      "learning_rate": 0.00019999998964087312,
-      "loss": 46.0,
-      "step": 1905
-    },
-    {
-      "epoch": 0.145727010340807,
-      "grad_norm": 0.001693650265224278,
-      "learning_rate": 0.0001999999896299371,
-      "loss": 46.0,
-      "step": 1906
-    },
-    {
-      "epoch": 0.14580346732419672,
-      "grad_norm": 0.002958431839942932,
-      "learning_rate": 0.00019999998961899536,
-      "loss": 46.0,
-      "step": 1907
-    },
-    {
-      "epoch": 0.14587992430758645,
-      "grad_norm": 0.0007097668130882084,
-      "learning_rate": 0.00019999998960804782,
-      "loss": 46.0,
-      "step": 1908
-    },
-    {
-      "epoch": 0.14595638129097616,
-      "grad_norm": 0.0016484018415212631,
-      "learning_rate": 0.0001999999895970945,
-      "loss": 46.0,
-      "step": 1909
-    },
-    {
-      "epoch": 0.1460328382743659,
-      "grad_norm": 0.0011686907382681966,
-      "learning_rate": 0.00019999998958613544,
-      "loss": 46.0,
-      "step": 1910
-    },
-    {
-      "epoch": 0.1461092952577556,
-      "grad_norm": 0.0017345441738143563,
-      "learning_rate": 0.00019999998957517055,
-      "loss": 46.0,
-      "step": 1911
-    },
-    {
-      "epoch": 0.14618575224114533,
-      "grad_norm": 0.003488808637484908,
-      "learning_rate": 0.00019999998956419997,
-      "loss": 46.0,
-      "step": 1912
-    },
-    {
-      "epoch": 0.14626220922453503,
-      "grad_norm": 0.0011651850072667003,
-      "learning_rate": 0.0001999999895532236,
-      "loss": 46.0,
-      "step": 1913
-    },
-    {
-      "epoch": 0.14633866620792477,
-      "grad_norm": 0.002016558311879635,
-      "learning_rate": 0.0001999999895422414,
-      "loss": 46.0,
-      "step": 1914
-    },
-    {
-      "epoch": 0.1464151231913145,
-      "grad_norm": 0.0005485732108354568,
-      "learning_rate": 0.0001999999895312535,
-      "loss": 46.0,
-      "step": 1915
-    },
-    {
-      "epoch": 0.1464915801747042,
-      "grad_norm": 0.0007180381799116731,
-      "learning_rate": 0.0001999999895202598,
-      "loss": 46.0,
-      "step": 1916
-    },
-    {
-      "epoch": 0.14656803715809394,
-      "grad_norm": 0.00216016941703856,
-      "learning_rate": 0.00019999998950926033,
-      "loss": 46.0,
-      "step": 1917
-    },
-    {
-      "epoch": 0.14664449414148364,
-      "grad_norm": 0.00147823931183666,
-      "learning_rate": 0.0001999999894982551,
-      "loss": 46.0,
-      "step": 1918
-    },
-    {
-      "epoch": 0.14672095112487338,
-      "grad_norm": 0.0007160396198742092,
-      "learning_rate": 0.0001999999894872441,
-      "loss": 46.0,
-      "step": 1919
-    },
-    {
-      "epoch": 0.14679740810826308,
-      "grad_norm": 0.0011503099231049418,
-      "learning_rate": 0.00019999998947622732,
-      "loss": 46.0,
-      "step": 1920
-    },
-    {
-      "epoch": 0.14687386509165282,
-      "grad_norm": 0.000507137447129935,
-      "learning_rate": 0.0001999999894652048,
-      "loss": 46.0,
-      "step": 1921
-    },
-    {
-      "epoch": 0.14695032207504252,
-      "grad_norm": 0.001573152025230229,
-      "learning_rate": 0.00019999998945417646,
-      "loss": 46.0,
-      "step": 1922
-    },
-    {
-      "epoch": 0.14702677905843226,
-      "grad_norm": 0.0007994626648724079,
-      "learning_rate": 0.0001999999894431424,
-      "loss": 46.0,
-      "step": 1923
-    },
-    {
-      "epoch": 0.14710323604182196,
-      "grad_norm": 0.0010093190940096974,
-      "learning_rate": 0.00019999998943210254,
-      "loss": 46.0,
-      "step": 1924
-    },
-    {
-      "epoch": 0.1471796930252117,
-      "grad_norm": 0.0007949511636979878,
-      "learning_rate": 0.0001999999894210569,
-      "loss": 46.0,
-      "step": 1925
-    },
-    {
-      "epoch": 0.1472561500086014,
-      "grad_norm": 0.0013361780438572168,
-      "learning_rate": 0.00019999998941000552,
-      "loss": 46.0,
-      "step": 1926
-    },
-    {
-      "epoch": 0.14733260699199113,
-      "grad_norm": 0.004195475019514561,
-      "learning_rate": 0.00019999998939894837,
-      "loss": 46.0,
-      "step": 1927
-    },
-    {
-      "epoch": 0.14740906397538084,
-      "grad_norm": 0.0006216627079993486,
-      "learning_rate": 0.00019999998938788544,
-      "loss": 46.0,
-      "step": 1928
-    },
-    {
-      "epoch": 0.14748552095877057,
-      "grad_norm": 0.0016950040590018034,
-      "learning_rate": 0.00019999998937681674,
-      "loss": 46.0,
-      "step": 1929
-    },
-    {
-      "epoch": 0.1475619779421603,
-      "grad_norm": 0.001891896827146411,
-      "learning_rate": 0.00019999998936574227,
-      "loss": 46.0,
-      "step": 1930
-    },
-    {
-      "epoch": 0.14763843492555,
-      "grad_norm": 0.00175075966399163,
-      "learning_rate": 0.00019999998935466205,
-      "loss": 46.0,
-      "step": 1931
-    },
-    {
-      "epoch": 0.14771489190893974,
-      "grad_norm": 0.0012245957041159272,
-      "learning_rate": 0.00019999998934357605,
-      "loss": 46.0,
-      "step": 1932
-    },
-    {
-      "epoch": 0.14779134889232945,
-      "grad_norm": 0.006115789990872145,
-      "learning_rate": 0.00019999998933248428,
-      "loss": 46.0,
-      "step": 1933
-    },
-    {
-      "epoch": 0.14786780587571918,
-      "grad_norm": 0.0010071592405438423,
-      "learning_rate": 0.00019999998932138672,
-      "loss": 46.0,
-      "step": 1934
-    },
-    {
-      "epoch": 0.1479442628591089,
-      "grad_norm": 0.0008599729044362903,
-      "learning_rate": 0.0001999999893102834,
-      "loss": 46.0,
-      "step": 1935
-    },
-    {
-      "epoch": 0.14802071984249862,
-      "grad_norm": 0.0007196691003628075,
-      "learning_rate": 0.0001999999892991743,
-      "loss": 46.0,
-      "step": 1936
-    },
-    {
-      "epoch": 0.14809717682588833,
-      "grad_norm": 0.0008595269755460322,
-      "learning_rate": 0.00019999998928805948,
-      "loss": 46.0,
-      "step": 1937
-    },
-    {
-      "epoch": 0.14817363380927806,
-      "grad_norm": 0.0012517975410446525,
-      "learning_rate": 0.00019999998927693887,
-      "loss": 46.0,
-      "step": 1938
-    },
-    {
-      "epoch": 0.14825009079266777,
-      "grad_norm": 0.0008255209540948272,
-      "learning_rate": 0.00019999998926581246,
-      "loss": 46.0,
-      "step": 1939
-    },
-    {
-      "epoch": 0.1483265477760575,
-      "grad_norm": 0.0016731730429455638,
-      "learning_rate": 0.0001999999892546803,
-      "loss": 46.0,
-      "step": 1940
-    },
-    {
-      "epoch": 0.1484030047594472,
-      "grad_norm": 0.0007372979307547212,
-      "learning_rate": 0.00019999998924354238,
-      "loss": 46.0,
-      "step": 1941
-    },
-    {
-      "epoch": 0.14847946174283694,
-      "grad_norm": 0.00036389255546964705,
-      "learning_rate": 0.00019999998923239865,
-      "loss": 46.0,
-      "step": 1942
-    },
-    {
-      "epoch": 0.14855591872622664,
-      "grad_norm": 0.0007508700946345925,
-      "learning_rate": 0.00019999998922124918,
-      "loss": 46.0,
-      "step": 1943
-    },
-    {
-      "epoch": 0.14863237570961638,
-      "grad_norm": 0.001239851932041347,
-      "learning_rate": 0.00019999998921009396,
-      "loss": 46.0,
-      "step": 1944
-    },
-    {
-      "epoch": 0.1487088326930061,
-      "grad_norm": 0.0017710485262796283,
-      "learning_rate": 0.00019999998919893294,
-      "loss": 46.0,
-      "step": 1945
-    },
-    {
-      "epoch": 0.14878528967639582,
-      "grad_norm": 0.0015017925761640072,
-      "learning_rate": 0.00019999998918776617,
-      "loss": 46.0,
-      "step": 1946
-    },
-    {
-      "epoch": 0.14886174665978555,
-      "grad_norm": 0.0018905504839494824,
-      "learning_rate": 0.00019999998917659363,
-      "loss": 46.0,
-      "step": 1947
-    },
-    {
-      "epoch": 0.14893820364317525,
-      "grad_norm": 0.0011791565921157598,
-      "learning_rate": 0.00019999998916541532,
-      "loss": 46.0,
-      "step": 1948
-    },
-    {
-      "epoch": 0.149014660626565,
-      "grad_norm": 0.0034085509832948446,
-      "learning_rate": 0.00019999998915423123,
-      "loss": 46.0,
-      "step": 1949
-    },
-    {
-      "epoch": 0.1490911176099547,
-      "grad_norm": 0.0008498480310663581,
-      "learning_rate": 0.00019999998914304137,
-      "loss": 46.0,
-      "step": 1950
-    },
-    {
-      "epoch": 0.14916757459334443,
-      "grad_norm": 0.0006049004732631147,
-      "learning_rate": 0.00019999998913184573,
-      "loss": 46.0,
-      "step": 1951
-    },
-    {
-      "epoch": 0.14924403157673413,
-      "grad_norm": 0.0005667923833243549,
-      "learning_rate": 0.00019999998912064433,
-      "loss": 46.0,
-      "step": 1952
-    },
-    {
-      "epoch": 0.14932048856012387,
-      "grad_norm": 0.002274232218042016,
-      "learning_rate": 0.00019999998910943717,
-      "loss": 46.0,
-      "step": 1953
-    },
-    {
-      "epoch": 0.14939694554351357,
-      "grad_norm": 0.0009235851466655731,
-      "learning_rate": 0.00019999998909822424,
-      "loss": 46.0,
-      "step": 1954
-    },
-    {
-      "epoch": 0.1494734025269033,
-      "grad_norm": 0.0006369143957272172,
-      "learning_rate": 0.00019999998908700554,
-      "loss": 46.0,
-      "step": 1955
-    },
-    {
-      "epoch": 0.149549859510293,
-      "grad_norm": 0.001315504894591868,
-      "learning_rate": 0.00019999998907578107,
-      "loss": 46.0,
-      "step": 1956
-    },
-    {
-      "epoch": 0.14962631649368274,
-      "grad_norm": 0.0012645369861274958,
-      "learning_rate": 0.00019999998906455085,
-      "loss": 46.0,
-      "step": 1957
-    },
-    {
-      "epoch": 0.14970277347707245,
-      "grad_norm": 0.002732011955231428,
-      "learning_rate": 0.0001999999890533148,
-      "loss": 46.0,
-      "step": 1958
-    },
-    {
-      "epoch": 0.14977923046046218,
-      "grad_norm": 0.0005908128223381937,
-      "learning_rate": 0.00019999998904207303,
-      "loss": 46.0,
-      "step": 1959
-    },
-    {
-      "epoch": 0.14985568744385191,
-      "grad_norm": 0.001558394986204803,
-      "learning_rate": 0.0001999999890308255,
-      "loss": 46.0,
-      "step": 1960
-    },
-    {
-      "epoch": 0.14993214442724162,
-      "grad_norm": 0.0009301886311732233,
-      "learning_rate": 0.00019999998901957215,
-      "loss": 46.0,
-      "step": 1961
-    },
-    {
-      "epoch": 0.15000860141063135,
-      "grad_norm": 0.0005543414736166596,
-      "learning_rate": 0.00019999998900831307,
-      "loss": 46.0,
-      "step": 1962
-    },
-    {
-      "epoch": 0.15008505839402106,
-      "grad_norm": 0.0023252021055668592,
-      "learning_rate": 0.0001999999889970482,
-      "loss": 46.0,
-      "step": 1963
-    },
-    {
-      "epoch": 0.1501615153774108,
-      "grad_norm": 0.0009747722069732845,
-      "learning_rate": 0.0001999999889857776,
-      "loss": 46.0,
-      "step": 1964
-    },
-    {
-      "epoch": 0.1502379723608005,
-      "grad_norm": 0.0006042062886990607,
-      "learning_rate": 0.0001999999889745012,
-      "loss": 46.0,
-      "step": 1965
-    },
-    {
-      "epoch": 0.15031442934419023,
-      "grad_norm": 0.0009974264539778233,
-      "learning_rate": 0.000199999988963219,
-      "loss": 46.0,
-      "step": 1966
-    },
-    {
-      "epoch": 0.15039088632757994,
-      "grad_norm": 0.0011752208229154348,
-      "learning_rate": 0.00019999998895193108,
-      "loss": 46.0,
-      "step": 1967
-    },
-    {
-      "epoch": 0.15046734331096967,
-      "grad_norm": 0.0004962339298799634,
-      "learning_rate": 0.00019999998894063738,
-      "loss": 46.0,
-      "step": 1968
-    },
-    {
-      "epoch": 0.15054380029435938,
-      "grad_norm": 0.0013381093740463257,
-      "learning_rate": 0.00019999998892933788,
-      "loss": 46.0,
-      "step": 1969
-    },
-    {
-      "epoch": 0.1506202572777491,
-      "grad_norm": 0.0005600806907750666,
-      "learning_rate": 0.00019999998891803266,
-      "loss": 46.0,
-      "step": 1970
-    },
-    {
-      "epoch": 0.15069671426113881,
-      "grad_norm": 0.0025287673342972994,
-      "learning_rate": 0.00019999998890672164,
-      "loss": 46.0,
-      "step": 1971
-    },
-    {
-      "epoch": 0.15077317124452855,
-      "grad_norm": 0.006942170672118664,
-      "learning_rate": 0.00019999998889540485,
-      "loss": 46.0,
-      "step": 1972
-    },
-    {
-      "epoch": 0.15084962822791828,
-      "grad_norm": 0.0009068876970559359,
-      "learning_rate": 0.0001999999888840823,
-      "loss": 46.0,
-      "step": 1973
-    },
-    {
-      "epoch": 0.150926085211308,
-      "grad_norm": 0.0008322437643073499,
-      "learning_rate": 0.000199999988872754,
-      "loss": 46.0,
-      "step": 1974
-    },
-    {
-      "epoch": 0.15100254219469772,
-      "grad_norm": 0.0015376051887869835,
-      "learning_rate": 0.0001999999888614199,
-      "loss": 46.0,
-      "step": 1975
-    },
-    {
-      "epoch": 0.15107899917808743,
-      "grad_norm": 0.0015460910508409142,
-      "learning_rate": 0.00019999998885008005,
-      "loss": 46.0,
-      "step": 1976
-    },
-    {
-      "epoch": 0.15115545616147716,
-      "grad_norm": 0.0014701655600219965,
-      "learning_rate": 0.0001999999888387344,
-      "loss": 46.0,
-      "step": 1977
-    },
-    {
-      "epoch": 0.15123191314486686,
-      "grad_norm": 0.0007002081838436425,
-      "learning_rate": 0.000199999988827383,
-      "loss": 46.0,
-      "step": 1978
-    },
-    {
-      "epoch": 0.1513083701282566,
-      "grad_norm": 0.0016067997785285115,
-      "learning_rate": 0.00019999998881602583,
-      "loss": 46.0,
-      "step": 1979
-    },
-    {
-      "epoch": 0.1513848271116463,
-      "grad_norm": 0.0018214803421869874,
-      "learning_rate": 0.0001999999888046629,
-      "loss": 46.0,
-      "step": 1980
-    },
-    {
-      "epoch": 0.15146128409503604,
-      "grad_norm": 0.0013334158575162292,
-      "learning_rate": 0.00019999998879329417,
-      "loss": 46.0,
-      "step": 1981
-    },
-    {
-      "epoch": 0.15153774107842574,
-      "grad_norm": 0.0008082379354164004,
-      "learning_rate": 0.0001999999887819197,
-      "loss": 46.0,
-      "step": 1982
-    },
-    {
-      "epoch": 0.15161419806181547,
-      "grad_norm": 0.0010814432753250003,
-      "learning_rate": 0.00019999998877053945,
-      "loss": 46.0,
-      "step": 1983
-    },
-    {
-      "epoch": 0.15169065504520518,
-      "grad_norm": 0.0008795211324468255,
-      "learning_rate": 0.00019999998875915343,
-      "loss": 46.0,
-      "step": 1984
-    },
-    {
-      "epoch": 0.1517671120285949,
-      "grad_norm": 0.0008336423779837787,
-      "learning_rate": 0.00019999998874776164,
-      "loss": 46.0,
-      "step": 1985
-    },
-    {
-      "epoch": 0.15184356901198462,
-      "grad_norm": 0.0014741772320121527,
-      "learning_rate": 0.0001999999887363641,
-      "loss": 46.0,
-      "step": 1986
-    },
-    {
-      "epoch": 0.15192002599537435,
-      "grad_norm": 0.0005251533002592623,
-      "learning_rate": 0.00019999998872496076,
-      "loss": 46.0,
-      "step": 1987
-    },
-    {
-      "epoch": 0.15199648297876409,
-      "grad_norm": 0.00796753540635109,
-      "learning_rate": 0.00019999998871355164,
-      "loss": 46.0,
-      "step": 1988
-    },
-    {
-      "epoch": 0.1520729399621538,
-      "grad_norm": 0.002017762279137969,
-      "learning_rate": 0.0001999999887021368,
-      "loss": 46.0,
-      "step": 1989
-    },
-    {
-      "epoch": 0.15214939694554352,
-      "grad_norm": 0.0010771476663649082,
-      "learning_rate": 0.00019999998869071618,
-      "loss": 46.0,
-      "step": 1990
-    },
-    {
-      "epoch": 0.15222585392893323,
-      "grad_norm": 0.0007402400951832533,
-      "learning_rate": 0.00019999998867928977,
-      "loss": 46.0,
-      "step": 1991
-    },
-    {
-      "epoch": 0.15230231091232296,
-      "grad_norm": 0.0027147787623107433,
-      "learning_rate": 0.0001999999886678576,
-      "loss": 46.0,
-      "step": 1992
-    },
-    {
-      "epoch": 0.15237876789571267,
-      "grad_norm": 0.001034329878166318,
-      "learning_rate": 0.00019999998865641966,
-      "loss": 46.0,
-      "step": 1993
-    },
-    {
-      "epoch": 0.1524552248791024,
-      "grad_norm": 0.002314161043614149,
-      "learning_rate": 0.00019999998864497594,
-      "loss": 46.0,
-      "step": 1994
-    },
-    {
-      "epoch": 0.1525316818624921,
-      "grad_norm": 0.0010412337724119425,
-      "learning_rate": 0.00019999998863352646,
-      "loss": 46.0,
-      "step": 1995
-    },
-    {
-      "epoch": 0.15260813884588184,
-      "grad_norm": 0.0012637953041121364,
-      "learning_rate": 0.00019999998862207122,
-      "loss": 46.0,
-      "step": 1996
-    },
-    {
-      "epoch": 0.15268459582927155,
-      "grad_norm": 0.004517598543316126,
-      "learning_rate": 0.00019999998861061017,
-      "loss": 46.0,
-      "step": 1997
-    },
-    {
-      "epoch": 0.15276105281266128,
-      "grad_norm": 0.0006794864893890917,
-      "learning_rate": 0.0001999999885991434,
-      "loss": 46.0,
-      "step": 1998
-    },
-    {
-      "epoch": 0.15283750979605099,
-      "grad_norm": 0.0006668080459348857,
-      "learning_rate": 0.00019999998858767084,
-      "loss": 46.0,
-      "step": 1999
-    },
-    {
-      "epoch": 0.15291396677944072,
-      "grad_norm": 0.00331284711137414,
-      "learning_rate": 0.0001999999885761925,
-      "loss": 46.0,
-      "step": 2000
-    },
-    {
-      "epoch": 0.15299042376283042,
-      "grad_norm": 0.0020922617986798286,
-      "learning_rate": 0.0001999999885647084,
-      "loss": 46.0,
-      "step": 2001
-    },
-    {
-      "epoch": 0.15306688074622016,
-      "grad_norm": 0.0015879099955782294,
-      "learning_rate": 0.00019999998855321853,
-      "loss": 46.0,
-      "step": 2002
-    },
-    {
-      "epoch": 0.1531433377296099,
-      "grad_norm": 0.0008255551219917834,
-      "learning_rate": 0.0001999999885417229,
-      "loss": 46.0,
-      "step": 2003
-    },
-    {
-      "epoch": 0.1532197947129996,
-      "grad_norm": 0.0009046594495885074,
-      "learning_rate": 0.0001999999885302215,
-      "loss": 46.0,
-      "step": 2004
-    },
-    {
-      "epoch": 0.15329625169638933,
-      "grad_norm": 0.0006455119582824409,
-      "learning_rate": 0.0001999999885187143,
-      "loss": 46.0,
-      "step": 2005
-    },
-    {
-      "epoch": 0.15337270867977904,
-      "grad_norm": 0.0012476410483941436,
-      "learning_rate": 0.00019999998850720138,
-      "loss": 46.0,
-      "step": 2006
-    },
-    {
-      "epoch": 0.15344916566316877,
-      "grad_norm": 0.001307318452745676,
-      "learning_rate": 0.00019999998849568265,
-      "loss": 46.0,
-      "step": 2007
-    },
-    {
-      "epoch": 0.15352562264655847,
-      "grad_norm": 0.001295082620345056,
-      "learning_rate": 0.00019999998848415818,
-      "loss": 46.0,
-      "step": 2008
-    },
-    {
-      "epoch": 0.1536020796299482,
-      "grad_norm": 0.002381877275183797,
-      "learning_rate": 0.00019999998847262794,
-      "loss": 46.0,
-      "step": 2009
-    },
-    {
-      "epoch": 0.1536785366133379,
-      "grad_norm": 0.003229626687243581,
-      "learning_rate": 0.00019999998846109192,
-      "loss": 46.0,
-      "step": 2010
-    },
-    {
-      "epoch": 0.15375499359672765,
-      "grad_norm": 0.0010005870135501027,
-      "learning_rate": 0.00019999998844955012,
-      "loss": 46.0,
-      "step": 2011
-    },
-    {
-      "epoch": 0.15383145058011735,
-      "grad_norm": 0.0009752497426234186,
-      "learning_rate": 0.00019999998843800256,
-      "loss": 46.0,
-      "step": 2012
-    },
-    {
-      "epoch": 0.15390790756350708,
-      "grad_norm": 0.00232255132868886,
-      "learning_rate": 0.00019999998842644922,
-      "loss": 46.0,
-      "step": 2013
-    },
-    {
-      "epoch": 0.1539843645468968,
-      "grad_norm": 0.001352866762317717,
-      "learning_rate": 0.0001999999884148901,
-      "loss": 46.0,
-      "step": 2014
-    },
-    {
-      "epoch": 0.15406082153028652,
-      "grad_norm": 0.0010676317615434527,
-      "learning_rate": 0.00019999998840332524,
-      "loss": 46.0,
-      "step": 2015
-    },
-    {
-      "epoch": 0.15413727851367623,
-      "grad_norm": 0.0017763529904186726,
-      "learning_rate": 0.0001999999883917546,
-      "loss": 46.0,
-      "step": 2016
-    },
-    {
-      "epoch": 0.15421373549706596,
-      "grad_norm": 0.001128602772951126,
-      "learning_rate": 0.0001999999883801782,
-      "loss": 46.0,
-      "step": 2017
-    },
-    {
-      "epoch": 0.1542901924804557,
-      "grad_norm": 0.0012293212348595262,
-      "learning_rate": 0.00019999998836859603,
-      "loss": 46.0,
-      "step": 2018
-    },
-    {
-      "epoch": 0.1543666494638454,
-      "grad_norm": 0.004482009913772345,
-      "learning_rate": 0.00019999998835700807,
-      "loss": 46.0,
-      "step": 2019
-    },
-    {
-      "epoch": 0.15444310644723513,
-      "grad_norm": 0.0013166291173547506,
-      "learning_rate": 0.00019999998834541435,
-      "loss": 46.0,
-      "step": 2020
-    },
-    {
-      "epoch": 0.15451956343062484,
-      "grad_norm": 0.0013117279158905149,
-      "learning_rate": 0.00019999998833381487,
-      "loss": 46.0,
-      "step": 2021
-    },
-    {
-      "epoch": 0.15459602041401457,
-      "grad_norm": 0.0010317321866750717,
-      "learning_rate": 0.0001999999883222096,
-      "loss": 46.0,
-      "step": 2022
-    },
-    {
-      "epoch": 0.15467247739740428,
-      "grad_norm": 0.0026497314684093,
-      "learning_rate": 0.00019999998831059858,
-      "loss": 46.0,
-      "step": 2023
-    },
-    {
-      "epoch": 0.154748934380794,
-      "grad_norm": 0.0003948312078136951,
-      "learning_rate": 0.0001999999882989818,
-      "loss": 46.0,
-      "step": 2024
-    },
-    {
-      "epoch": 0.15482539136418372,
-      "grad_norm": 0.001416828134097159,
-      "learning_rate": 0.00019999998828735922,
-      "loss": 46.0,
-      "step": 2025
-    },
-    {
-      "epoch": 0.15490184834757345,
-      "grad_norm": 0.0013770645018666983,
-      "learning_rate": 0.0001999999882757309,
-      "loss": 46.0,
-      "step": 2026
-    },
-    {
-      "epoch": 0.15497830533096316,
-      "grad_norm": 0.0012552356347441673,
-      "learning_rate": 0.00019999998826409677,
-      "loss": 46.0,
-      "step": 2027
-    },
-    {
-      "epoch": 0.1550547623143529,
-      "grad_norm": 0.00109623814933002,
-      "learning_rate": 0.0001999999882524569,
-      "loss": 46.0,
-      "step": 2028
-    },
-    {
-      "epoch": 0.1551312192977426,
-      "grad_norm": 0.0020386953838169575,
-      "learning_rate": 0.00019999998824081128,
-      "loss": 46.0,
-      "step": 2029
-    },
-    {
-      "epoch": 0.15520767628113233,
-      "grad_norm": 0.0008147793123498559,
-      "learning_rate": 0.00019999998822915985,
-      "loss": 46.0,
-      "step": 2030
-    },
-    {
-      "epoch": 0.15528413326452206,
-      "grad_norm": 0.0021701687946915627,
-      "learning_rate": 0.00019999998821750267,
-      "loss": 46.0,
-      "step": 2031
-    },
-    {
-      "epoch": 0.15536059024791177,
-      "grad_norm": 0.004709912929683924,
-      "learning_rate": 0.00019999998820583974,
-      "loss": 46.0,
-      "step": 2032
-    },
-    {
-      "epoch": 0.1554370472313015,
-      "grad_norm": 0.0016737825935706496,
-      "learning_rate": 0.00019999998819417102,
-      "loss": 46.0,
-      "step": 2033
-    },
-    {
-      "epoch": 0.1555135042146912,
-      "grad_norm": 0.0007028424297459424,
-      "learning_rate": 0.0001999999881824965,
-      "loss": 46.0,
-      "step": 2034
-    },
-    {
-      "epoch": 0.15558996119808094,
-      "grad_norm": 0.0011884956620633602,
-      "learning_rate": 0.00019999998817081625,
-      "loss": 46.0,
-      "step": 2035
-    },
-    {
-      "epoch": 0.15566641818147064,
-      "grad_norm": 0.0010022360365837812,
-      "learning_rate": 0.00019999998815913023,
-      "loss": 46.0,
-      "step": 2036
-    },
-    {
-      "epoch": 0.15574287516486038,
-      "grad_norm": 0.0005076303496025503,
-      "learning_rate": 0.0001999999881474384,
-      "loss": 46.0,
-      "step": 2037
-    },
-    {
-      "epoch": 0.15581933214825008,
-      "grad_norm": 0.0014753614086657763,
-      "learning_rate": 0.00019999998813574087,
-      "loss": 46.0,
-      "step": 2038
-    },
-    {
-      "epoch": 0.15589578913163982,
-      "grad_norm": 0.0006860756548121572,
-      "learning_rate": 0.00019999998812403753,
-      "loss": 46.0,
-      "step": 2039
-    },
-    {
-      "epoch": 0.15597224611502952,
-      "grad_norm": 0.0010463307844474912,
-      "learning_rate": 0.00019999998811232842,
-      "loss": 46.0,
-      "step": 2040
-    },
-    {
-      "epoch": 0.15604870309841926,
-      "grad_norm": 0.003599617863073945,
-      "learning_rate": 0.00019999998810061353,
-      "loss": 46.0,
-      "step": 2041
-    },
-    {
-      "epoch": 0.15612516008180896,
-      "grad_norm": 0.0015982154291123152,
-      "learning_rate": 0.0001999999880888929,
-      "loss": 46.0,
-      "step": 2042
-    },
-    {
-      "epoch": 0.1562016170651987,
-      "grad_norm": 0.0025439190212637186,
-      "learning_rate": 0.0001999999880771665,
-      "loss": 46.0,
-      "step": 2043
-    },
-    {
-      "epoch": 0.1562780740485884,
-      "grad_norm": 0.0015923408791422844,
-      "learning_rate": 0.0001999999880654343,
-      "loss": 46.0,
-      "step": 2044
-    },
-    {
-      "epoch": 0.15635453103197813,
-      "grad_norm": 0.0007652989006601274,
-      "learning_rate": 0.00019999998805369634,
-      "loss": 46.0,
-      "step": 2045
-    },
-    {
-      "epoch": 0.15643098801536787,
-      "grad_norm": 0.001005074242129922,
-      "learning_rate": 0.0001999999880419526,
-      "loss": 46.0,
-      "step": 2046
-    },
-    {
-      "epoch": 0.15650744499875757,
-      "grad_norm": 0.011235262267291546,
-      "learning_rate": 0.0001999999880302031,
-      "loss": 46.0,
-      "step": 2047
-    },
-    {
-      "epoch": 0.1565839019821473,
-      "grad_norm": 0.0014421091182157397,
-      "learning_rate": 0.00019999998801844787,
-      "loss": 46.0,
-      "step": 2048
-    },
-    {
-      "epoch": 0.156660358965537,
-      "grad_norm": 0.0009523187763988972,
-      "learning_rate": 0.00019999998800668682,
-      "loss": 46.0,
-      "step": 2049
-    },
-    {
-      "epoch": 0.15673681594892674,
-      "grad_norm": 0.0015538878506049514,
-      "learning_rate": 0.00019999998799492,
-      "loss": 46.0,
-      "step": 2050
-    },
-    {
-      "epoch": 0.15681327293231645,
-      "grad_norm": 0.0025210848543792963,
-      "learning_rate": 0.00019999998798314746,
-      "loss": 46.0,
-      "step": 2051
-    },
-    {
-      "epoch": 0.15688972991570618,
-      "grad_norm": 0.0003893290995620191,
-      "learning_rate": 0.00019999998797136912,
-      "loss": 46.0,
-      "step": 2052
-    },
-    {
-      "epoch": 0.1569661868990959,
-      "grad_norm": 0.0023423300590366125,
-      "learning_rate": 0.000199999987959585,
-      "loss": 46.0,
-      "step": 2053
-    },
-    {
-      "epoch": 0.15704264388248562,
-      "grad_norm": 0.00150935014244169,
-      "learning_rate": 0.00019999998794779513,
-      "loss": 46.0,
-      "step": 2054
-    },
-    {
-      "epoch": 0.15711910086587533,
-      "grad_norm": 0.0014054529601708055,
-      "learning_rate": 0.00019999998793599947,
-      "loss": 46.0,
-      "step": 2055
-    },
-    {
-      "epoch": 0.15719555784926506,
-      "grad_norm": 0.001149459625594318,
-      "learning_rate": 0.00019999998792419806,
-      "loss": 46.0,
-      "step": 2056
-    },
-    {
-      "epoch": 0.15727201483265477,
-      "grad_norm": 0.005198684521019459,
-      "learning_rate": 0.00019999998791239088,
-      "loss": 46.0,
-      "step": 2057
-    },
-    {
-      "epoch": 0.1573484718160445,
-      "grad_norm": 0.004143446683883667,
-      "learning_rate": 0.00019999998790057793,
-      "loss": 46.0,
-      "step": 2058
-    },
-    {
-      "epoch": 0.1574249287994342,
-      "grad_norm": 0.0008370177820324898,
-      "learning_rate": 0.0001999999878887592,
-      "loss": 46.0,
-      "step": 2059
-    },
-    {
-      "epoch": 0.15750138578282394,
-      "grad_norm": 0.0007398803136311471,
-      "learning_rate": 0.0001999999878769347,
-      "loss": 46.0,
-      "step": 2060
-    },
-    {
-      "epoch": 0.15757784276621367,
-      "grad_norm": 0.001524512772448361,
-      "learning_rate": 0.00019999998786510444,
-      "loss": 46.0,
-      "step": 2061
-    },
-    {
-      "epoch": 0.15765429974960338,
-      "grad_norm": 0.002082538092508912,
-      "learning_rate": 0.0001999999878532684,
-      "loss": 46.0,
-      "step": 2062
-    },
-    {
-      "epoch": 0.1577307567329931,
-      "grad_norm": 0.0008017992950044572,
-      "learning_rate": 0.0001999999878414266,
-      "loss": 46.0,
-      "step": 2063
-    },
-    {
-      "epoch": 0.15780721371638282,
-      "grad_norm": 0.000616123725194484,
-      "learning_rate": 0.00019999998782957903,
-      "loss": 46.0,
-      "step": 2064
-    },
-    {
-      "epoch": 0.15788367069977255,
-      "grad_norm": 0.0007139994413591921,
-      "learning_rate": 0.00019999998781772567,
-      "loss": 46.0,
-      "step": 2065
-    },
-    {
-      "epoch": 0.15796012768316225,
-      "grad_norm": 0.001166562782600522,
-      "learning_rate": 0.00019999998780586658,
-      "loss": 46.0,
-      "step": 2066
-    },
-    {
-      "epoch": 0.158036584666552,
-      "grad_norm": 0.0005792691372334957,
-      "learning_rate": 0.00019999998779400167,
-      "loss": 46.0,
-      "step": 2067
-    },
-    {
-      "epoch": 0.1581130416499417,
-      "grad_norm": 0.0011840608203783631,
-      "learning_rate": 0.000199999987782131,
-      "loss": 46.0,
-      "step": 2068
-    },
-    {
-      "epoch": 0.15818949863333143,
-      "grad_norm": 0.0016025849618017673,
-      "learning_rate": 0.0001999999877702546,
-      "loss": 46.0,
-      "step": 2069
-    },
-    {
-      "epoch": 0.15826595561672113,
-      "grad_norm": 0.0004976553027518094,
-      "learning_rate": 0.00019999998775837243,
-      "loss": 46.0,
-      "step": 2070
-    },
-    {
-      "epoch": 0.15834241260011087,
-      "grad_norm": 0.001387871685437858,
-      "learning_rate": 0.00019999998774648445,
-      "loss": 46.0,
-      "step": 2071
-    },
-    {
-      "epoch": 0.15841886958350057,
-      "grad_norm": 0.001114772749133408,
-      "learning_rate": 0.00019999998773459073,
-      "loss": 46.0,
-      "step": 2072
-    },
-    {
-      "epoch": 0.1584953265668903,
-      "grad_norm": 0.0009465409675613046,
-      "learning_rate": 0.00019999998772269123,
-      "loss": 46.0,
-      "step": 2073
-    },
-    {
-      "epoch": 0.15857178355028,
-      "grad_norm": 0.002183204982429743,
-      "learning_rate": 0.00019999998771078596,
-      "loss": 46.0,
-      "step": 2074
-    },
-    {
-      "epoch": 0.15864824053366974,
-      "grad_norm": 0.0007752188830636442,
-      "learning_rate": 0.0001999999876988749,
-      "loss": 46.0,
-      "step": 2075
-    },
-    {
-      "epoch": 0.15872469751705948,
-      "grad_norm": 0.0007461218046955764,
-      "learning_rate": 0.00019999998768695812,
-      "loss": 46.0,
-      "step": 2076
-    },
-    {
-      "epoch": 0.15880115450044918,
-      "grad_norm": 0.0009555542492307723,
-      "learning_rate": 0.00019999998767503553,
-      "loss": 46.0,
-      "step": 2077
-    },
-    {
-      "epoch": 0.15887761148383892,
-      "grad_norm": 0.0009751114994287491,
-      "learning_rate": 0.0001999999876631072,
-      "loss": 46.0,
-      "step": 2078
-    },
-    {
-      "epoch": 0.15895406846722862,
-      "grad_norm": 0.0006597651517949998,
-      "learning_rate": 0.00019999998765117308,
-      "loss": 46.0,
-      "step": 2079
-    },
-    {
-      "epoch": 0.15903052545061835,
-      "grad_norm": 0.0008710972033441067,
-      "learning_rate": 0.0001999999876392332,
-      "loss": 46.0,
-      "step": 2080
-    },
-    {
-      "epoch": 0.15910698243400806,
-      "grad_norm": 0.0006096456781961024,
-      "learning_rate": 0.00019999998762728754,
-      "loss": 46.0,
-      "step": 2081
-    },
-    {
-      "epoch": 0.1591834394173978,
-      "grad_norm": 0.004188179969787598,
-      "learning_rate": 0.0001999999876153361,
-      "loss": 46.0,
-      "step": 2082
-    },
-    {
-      "epoch": 0.1592598964007875,
-      "grad_norm": 0.0004476464237086475,
-      "learning_rate": 0.00019999998760337893,
-      "loss": 46.0,
-      "step": 2083
-    },
-    {
-      "epoch": 0.15933635338417723,
-      "grad_norm": 0.0009876301046460867,
-      "learning_rate": 0.00019999998759141595,
-      "loss": 46.0,
-      "step": 2084
-    },
-    {
-      "epoch": 0.15941281036756694,
-      "grad_norm": 0.0017412264132872224,
-      "learning_rate": 0.00019999998757944722,
-      "loss": 46.0,
-      "step": 2085
-    },
-    {
-      "epoch": 0.15948926735095667,
-      "grad_norm": 0.0010906915413215756,
-      "learning_rate": 0.00019999998756747273,
-      "loss": 46.0,
-      "step": 2086
-    },
-    {
-      "epoch": 0.15956572433434638,
-      "grad_norm": 0.0009544620988890529,
-      "learning_rate": 0.00019999998755549245,
-      "loss": 46.0,
-      "step": 2087
-    },
-    {
-      "epoch": 0.1596421813177361,
-      "grad_norm": 0.0013774504186585546,
-      "learning_rate": 0.0001999999875435064,
-      "loss": 46.0,
-      "step": 2088
-    },
-    {
-      "epoch": 0.15971863830112584,
-      "grad_norm": 0.00046507621300406754,
-      "learning_rate": 0.00019999998753151462,
-      "loss": 46.0,
-      "step": 2089
-    },
-    {
-      "epoch": 0.15979509528451555,
-      "grad_norm": 0.0033230867702513933,
-      "learning_rate": 0.000199999987519517,
-      "loss": 46.0,
-      "step": 2090
-    },
-    {
-      "epoch": 0.15987155226790528,
-      "grad_norm": 0.0015882361913099885,
-      "learning_rate": 0.00019999998750751366,
-      "loss": 46.0,
-      "step": 2091
-    },
-    {
-      "epoch": 0.159948009251295,
-      "grad_norm": 0.0010441900230944157,
-      "learning_rate": 0.00019999998749550458,
-      "loss": 46.0,
-      "step": 2092
-    },
-    {
-      "epoch": 0.16002446623468472,
-      "grad_norm": 0.0013350731460377574,
-      "learning_rate": 0.00019999998748348967,
-      "loss": 46.0,
-      "step": 2093
-    },
-    {
-      "epoch": 0.16010092321807443,
-      "grad_norm": 0.0006891741650179029,
-      "learning_rate": 0.000199999987471469,
-      "loss": 46.0,
-      "step": 2094
-    },
-    {
-      "epoch": 0.16017738020146416,
-      "grad_norm": 0.0014986591413617134,
-      "learning_rate": 0.00019999998745944258,
-      "loss": 46.0,
-      "step": 2095
-    },
-    {
-      "epoch": 0.16025383718485386,
-      "grad_norm": 0.0008050437900237739,
-      "learning_rate": 0.00019999998744741037,
-      "loss": 46.0,
-      "step": 2096
-    },
-    {
-      "epoch": 0.1603302941682436,
-      "grad_norm": 0.0007129543810151517,
-      "learning_rate": 0.00019999998743537242,
-      "loss": 46.0,
-      "step": 2097
-    },
-    {
-      "epoch": 0.1604067511516333,
-      "grad_norm": 0.0016608787700533867,
-      "learning_rate": 0.00019999998742332867,
-      "loss": 46.0,
-      "step": 2098
-    },
-    {
-      "epoch": 0.16048320813502304,
-      "grad_norm": 0.0016671482007950544,
-      "learning_rate": 0.0001999999874112792,
-      "loss": 46.0,
-      "step": 2099
-    },
-    {
-      "epoch": 0.16055966511841274,
-      "grad_norm": 0.0006881877197884023,
-      "learning_rate": 0.0001999999873992239,
-      "loss": 46.0,
-      "step": 2100
-    },
-    {
-      "epoch": 0.16063612210180248,
-      "grad_norm": 0.0008800942450761795,
-      "learning_rate": 0.00019999998738716288,
-      "loss": 46.0,
-      "step": 2101
-    },
-    {
-      "epoch": 0.16071257908519218,
-      "grad_norm": 0.0007842633640393615,
-      "learning_rate": 0.00019999998737509604,
-      "loss": 46.0,
-      "step": 2102
-    },
-    {
-      "epoch": 0.16078903606858191,
-      "grad_norm": 0.0011482764966785908,
-      "learning_rate": 0.00019999998736302347,
-      "loss": 46.0,
-      "step": 2103
-    },
-    {
-      "epoch": 0.16086549305197165,
-      "grad_norm": 0.0014609784120693803,
-      "learning_rate": 0.0001999999873509451,
-      "loss": 46.0,
-      "step": 2104
-    },
-    {
-      "epoch": 0.16094195003536135,
-      "grad_norm": 0.002479089191183448,
-      "learning_rate": 0.00019999998733886097,
-      "loss": 46.0,
-      "step": 2105
-    },
-    {
-      "epoch": 0.1610184070187511,
-      "grad_norm": 0.0009622796787880361,
-      "learning_rate": 0.0001999999873267711,
-      "loss": 46.0,
-      "step": 2106
-    },
-    {
-      "epoch": 0.1610948640021408,
-      "grad_norm": 0.000658390112221241,
-      "learning_rate": 0.00019999998731467543,
-      "loss": 46.0,
-      "step": 2107
-    },
-    {
-      "epoch": 0.16117132098553053,
-      "grad_norm": 0.0032200063578784466,
-      "learning_rate": 0.000199999987302574,
-      "loss": 46.0,
-      "step": 2108
-    },
-    {
-      "epoch": 0.16124777796892023,
-      "grad_norm": 0.001384242088533938,
-      "learning_rate": 0.0001999999872904668,
-      "loss": 46.0,
-      "step": 2109
-    },
-    {
-      "epoch": 0.16132423495230996,
-      "grad_norm": 0.0031077503226697445,
-      "learning_rate": 0.00019999998727835384,
-      "loss": 46.0,
-      "step": 2110
-    },
-    {
-      "epoch": 0.16140069193569967,
-      "grad_norm": 0.0007535358308814466,
-      "learning_rate": 0.00019999998726623512,
-      "loss": 46.0,
-      "step": 2111
-    },
-    {
-      "epoch": 0.1614771489190894,
-      "grad_norm": 0.0013495383318513632,
-      "learning_rate": 0.0001999999872541106,
-      "loss": 46.0,
-      "step": 2112
-    },
-    {
-      "epoch": 0.1615536059024791,
-      "grad_norm": 0.0016731504583731294,
-      "learning_rate": 0.00019999998724198033,
-      "loss": 46.0,
-      "step": 2113
-    },
-    {
-      "epoch": 0.16163006288586884,
-      "grad_norm": 0.0009444876341149211,
-      "learning_rate": 0.00019999998722984428,
-      "loss": 46.0,
-      "step": 2114
-    },
-    {
-      "epoch": 0.16170651986925855,
-      "grad_norm": 0.005001413635909557,
-      "learning_rate": 0.0001999999872177025,
-      "loss": 46.0,
-      "step": 2115
-    },
-    {
-      "epoch": 0.16178297685264828,
-      "grad_norm": 0.0032453536987304688,
-      "learning_rate": 0.00019999998720555487,
-      "loss": 46.0,
-      "step": 2116
-    },
-    {
-      "epoch": 0.16185943383603799,
-      "grad_norm": 0.0007565810228697956,
-      "learning_rate": 0.0001999999871934015,
-      "loss": 46.0,
-      "step": 2117
-    },
-    {
-      "epoch": 0.16193589081942772,
-      "grad_norm": 0.0006535269203595817,
-      "learning_rate": 0.0001999999871812424,
-      "loss": 46.0,
-      "step": 2118
-    },
-    {
-      "epoch": 0.16201234780281745,
-      "grad_norm": 0.0010752216912806034,
-      "learning_rate": 0.00019999998716907751,
-      "loss": 46.0,
-      "step": 2119
-    },
-    {
-      "epoch": 0.16208880478620716,
-      "grad_norm": 0.001103323302231729,
-      "learning_rate": 0.00019999998715690683,
-      "loss": 46.0,
-      "step": 2120
-    },
-    {
-      "epoch": 0.1621652617695969,
-      "grad_norm": 0.0011689539533108473,
-      "learning_rate": 0.0001999999871447304,
-      "loss": 46.0,
-      "step": 2121
-    },
-    {
-      "epoch": 0.1622417187529866,
-      "grad_norm": 0.00039576523704454303,
-      "learning_rate": 0.0001999999871325482,
-      "loss": 46.0,
-      "step": 2122
-    },
-    {
-      "epoch": 0.16231817573637633,
-      "grad_norm": 0.004719104617834091,
-      "learning_rate": 0.00019999998712036024,
-      "loss": 46.0,
-      "step": 2123
-    },
-    {
-      "epoch": 0.16239463271976604,
-      "grad_norm": 0.003539037425071001,
-      "learning_rate": 0.0001999999871081665,
-      "loss": 46.0,
-      "step": 2124
-    },
-    {
-      "epoch": 0.16247108970315577,
-      "grad_norm": 0.0017531836638227105,
-      "learning_rate": 0.000199999987095967,
-      "loss": 46.0,
-      "step": 2125
-    },
-    {
-      "epoch": 0.16254754668654547,
-      "grad_norm": 0.0008034551865421236,
-      "learning_rate": 0.0001999999870837617,
-      "loss": 46.0,
-      "step": 2126
-    },
-    {
-      "epoch": 0.1626240036699352,
-      "grad_norm": 0.0015628214459866285,
-      "learning_rate": 0.00019999998707155066,
-      "loss": 46.0,
-      "step": 2127
-    },
-    {
-      "epoch": 0.1627004606533249,
-      "grad_norm": 0.001498895580880344,
-      "learning_rate": 0.00019999998705933384,
-      "loss": 46.0,
-      "step": 2128
-    },
-    {
-      "epoch": 0.16277691763671465,
-      "grad_norm": 0.0008606580668129027,
-      "learning_rate": 0.00019999998704711125,
-      "loss": 46.0,
-      "step": 2129
-    },
-    {
-      "epoch": 0.16285337462010435,
-      "grad_norm": 0.003488902933895588,
-      "learning_rate": 0.00019999998703488288,
-      "loss": 46.0,
-      "step": 2130
-    },
-    {
-      "epoch": 0.16292983160349409,
-      "grad_norm": 0.0008501252741552889,
-      "learning_rate": 0.00019999998702264877,
-      "loss": 46.0,
-      "step": 2131
-    },
-    {
-      "epoch": 0.1630062885868838,
-      "grad_norm": 0.0006707796710543334,
-      "learning_rate": 0.00019999998701040886,
-      "loss": 46.0,
-      "step": 2132
-    },
-    {
-      "epoch": 0.16308274557027352,
-      "grad_norm": 0.0011532234493643045,
-      "learning_rate": 0.0001999999869981632,
-      "loss": 46.0,
-      "step": 2133
-    },
-    {
-      "epoch": 0.16315920255366326,
-      "grad_norm": 0.001927627483382821,
-      "learning_rate": 0.00019999998698591175,
-      "loss": 46.0,
-      "step": 2134
-    },
-    {
-      "epoch": 0.16323565953705296,
-      "grad_norm": 0.004618369974195957,
-      "learning_rate": 0.00019999998697365457,
-      "loss": 46.0,
-      "step": 2135
-    },
-    {
-      "epoch": 0.1633121165204427,
-      "grad_norm": 0.001824512262828648,
-      "learning_rate": 0.0001999999869613916,
-      "loss": 46.0,
-      "step": 2136
-    },
-    {
-      "epoch": 0.1633885735038324,
-      "grad_norm": 0.002617462072521448,
-      "learning_rate": 0.00019999998694912282,
-      "loss": 46.0,
-      "step": 2137
-    },
-    {
-      "epoch": 0.16346503048722213,
-      "grad_norm": 0.012715129181742668,
-      "learning_rate": 0.00019999998693684835,
-      "loss": 46.0,
-      "step": 2138
-    },
-    {
-      "epoch": 0.16354148747061184,
-      "grad_norm": 0.0006668414571322501,
-      "learning_rate": 0.00019999998692456805,
-      "loss": 46.0,
-      "step": 2139
-    },
-    {
-      "epoch": 0.16361794445400157,
-      "grad_norm": 0.0008075271034613252,
-      "learning_rate": 0.00019999998691228198,
-      "loss": 46.0,
-      "step": 2140
-    },
-    {
-      "epoch": 0.16369440143739128,
-      "grad_norm": 0.0009113276028074324,
-      "learning_rate": 0.0001999999868999902,
-      "loss": 46.0,
-      "step": 2141
-    },
-    {
-      "epoch": 0.163770858420781,
-      "grad_norm": 0.005092452745884657,
-      "learning_rate": 0.0001999999868876926,
-      "loss": 46.0,
-      "step": 2142
-    },
-    {
-      "epoch": 0.16384731540417072,
-      "grad_norm": 0.0009798347018659115,
-      "learning_rate": 0.00019999998687538924,
-      "loss": 46.0,
-      "step": 2143
-    },
-    {
-      "epoch": 0.16392377238756045,
-      "grad_norm": 0.0004848692915402353,
-      "learning_rate": 0.0001999999868630801,
-      "loss": 46.0,
-      "step": 2144
-    },
-    {
-      "epoch": 0.16400022937095016,
-      "grad_norm": 0.0009060886804945767,
-      "learning_rate": 0.0001999999868507652,
-      "loss": 46.0,
-      "step": 2145
-    },
-    {
-      "epoch": 0.1640766863543399,
-      "grad_norm": 0.004894052632153034,
-      "learning_rate": 0.00019999998683844453,
-      "loss": 46.0,
-      "step": 2146
-    },
-    {
-      "epoch": 0.16415314333772962,
-      "grad_norm": 0.0008194067631848156,
-      "learning_rate": 0.0001999999868261181,
-      "loss": 46.0,
-      "step": 2147
-    },
-    {
-      "epoch": 0.16422960032111933,
-      "grad_norm": 0.0006418468547053635,
-      "learning_rate": 0.0001999999868137859,
-      "loss": 46.0,
-      "step": 2148
-    },
-    {
-      "epoch": 0.16430605730450906,
-      "grad_norm": 0.0005521623534150422,
-      "learning_rate": 0.0001999999868014479,
-      "loss": 46.0,
-      "step": 2149
-    },
-    {
-      "epoch": 0.16438251428789877,
-      "grad_norm": 0.004384755622595549,
-      "learning_rate": 0.00019999998678910417,
-      "loss": 46.0,
-      "step": 2150
-    },
-    {
-      "epoch": 0.1644589712712885,
-      "grad_norm": 0.001799800549633801,
-      "learning_rate": 0.00019999998677675465,
-      "loss": 46.0,
-      "step": 2151
-    },
-    {
-      "epoch": 0.1645354282546782,
-      "grad_norm": 0.0008425433188676834,
-      "learning_rate": 0.00019999998676439935,
-      "loss": 46.0,
-      "step": 2152
-    },
-    {
-      "epoch": 0.16461188523806794,
-      "grad_norm": 0.0023057525977492332,
-      "learning_rate": 0.0001999999867520383,
-      "loss": 46.0,
-      "step": 2153
-    },
-    {
-      "epoch": 0.16468834222145765,
-      "grad_norm": 0.0004937535850331187,
-      "learning_rate": 0.0001999999867396715,
-      "loss": 46.0,
-      "step": 2154
-    },
-    {
-      "epoch": 0.16476479920484738,
-      "grad_norm": 0.001624703872948885,
-      "learning_rate": 0.0001999999867272989,
-      "loss": 46.0,
-      "step": 2155
-    },
-    {
-      "epoch": 0.16484125618823708,
-      "grad_norm": 0.0017441855743527412,
-      "learning_rate": 0.00019999998671492054,
-      "loss": 46.0,
-      "step": 2156
-    },
-    {
-      "epoch": 0.16491771317162682,
-      "grad_norm": 0.002246905816718936,
-      "learning_rate": 0.0001999999867025364,
-      "loss": 46.0,
-      "step": 2157
-    },
-    {
-      "epoch": 0.16499417015501652,
-      "grad_norm": 0.0006978364544920623,
-      "learning_rate": 0.0001999999866901465,
-      "loss": 46.0,
-      "step": 2158
-    },
-    {
-      "epoch": 0.16507062713840626,
-      "grad_norm": 0.0007684300653636456,
-      "learning_rate": 0.0001999999866777508,
-      "loss": 46.0,
-      "step": 2159
-    },
-    {
-      "epoch": 0.16514708412179596,
-      "grad_norm": 0.0004500261857174337,
-      "learning_rate": 0.00019999998666534938,
-      "loss": 46.0,
-      "step": 2160
-    },
-    {
-      "epoch": 0.1652235411051857,
-      "grad_norm": 0.0007735901745036244,
-      "learning_rate": 0.00019999998665294215,
-      "loss": 46.0,
-      "step": 2161
-    },
-    {
-      "epoch": 0.16529999808857543,
-      "grad_norm": 0.0009222542867064476,
-      "learning_rate": 0.0001999999866405292,
-      "loss": 46.0,
-      "step": 2162
-    },
-    {
-      "epoch": 0.16537645507196513,
-      "grad_norm": 0.0011216216953471303,
-      "learning_rate": 0.00019999998662811045,
-      "loss": 46.0,
-      "step": 2163
-    },
-    {
-      "epoch": 0.16545291205535487,
-      "grad_norm": 0.009975769557058811,
-      "learning_rate": 0.0001999999866156859,
-      "loss": 46.0,
-      "step": 2164
-    },
-    {
-      "epoch": 0.16552936903874457,
-      "grad_norm": 0.0022531873546540737,
-      "learning_rate": 0.00019999998660325563,
-      "loss": 46.0,
-      "step": 2165
-    },
-    {
-      "epoch": 0.1656058260221343,
-      "grad_norm": 0.0005565266474150121,
-      "learning_rate": 0.0001999999865908196,
-      "loss": 46.0,
-      "step": 2166
-    },
-    {
-      "epoch": 0.165682283005524,
-      "grad_norm": 0.0012787970481440425,
-      "learning_rate": 0.00019999998657837774,
-      "loss": 46.0,
-      "step": 2167
-    },
-    {
-      "epoch": 0.16575873998891374,
-      "grad_norm": 0.0013191215693950653,
-      "learning_rate": 0.00019999998656593015,
-      "loss": 46.0,
-      "step": 2168
-    },
-    {
-      "epoch": 0.16583519697230345,
-      "grad_norm": 0.0009020023280754685,
-      "learning_rate": 0.0001999999865534768,
-      "loss": 46.0,
-      "step": 2169
-    },
-    {
-      "epoch": 0.16591165395569318,
-      "grad_norm": 0.0016326627228409052,
-      "learning_rate": 0.00019999998654101763,
-      "loss": 46.0,
-      "step": 2170
-    },
-    {
-      "epoch": 0.1659881109390829,
-      "grad_norm": 0.0009855673415586352,
-      "learning_rate": 0.00019999998652855272,
-      "loss": 46.0,
-      "step": 2171
-    },
-    {
-      "epoch": 0.16606456792247262,
-      "grad_norm": 0.0007167229196056724,
-      "learning_rate": 0.00019999998651608206,
-      "loss": 46.0,
-      "step": 2172
-    },
-    {
-      "epoch": 0.16614102490586233,
-      "grad_norm": 0.0018436299869790673,
-      "learning_rate": 0.0001999999865036056,
-      "loss": 46.0,
-      "step": 2173
-    },
-    {
-      "epoch": 0.16621748188925206,
-      "grad_norm": 0.0032357568852603436,
-      "learning_rate": 0.0001999999864911234,
-      "loss": 46.0,
-      "step": 2174
-    },
-    {
-      "epoch": 0.16629393887264177,
-      "grad_norm": 0.003142263740301132,
-      "learning_rate": 0.0001999999864786354,
-      "loss": 46.0,
-      "step": 2175
-    },
-    {
-      "epoch": 0.1663703958560315,
-      "grad_norm": 0.001277631032280624,
-      "learning_rate": 0.00019999998646614168,
-      "loss": 46.0,
-      "step": 2176
-    },
-    {
-      "epoch": 0.16644685283942123,
-      "grad_norm": 0.003091112244874239,
-      "learning_rate": 0.00019999998645364215,
-      "loss": 46.0,
-      "step": 2177
-    },
-    {
-      "epoch": 0.16652330982281094,
-      "grad_norm": 0.0004034547018818557,
-      "learning_rate": 0.00019999998644113686,
-      "loss": 46.0,
-      "step": 2178
-    },
-    {
-      "epoch": 0.16659976680620067,
-      "grad_norm": 0.0009671281441114843,
-      "learning_rate": 0.00019999998642862581,
-      "loss": 46.0,
-      "step": 2179
-    },
-    {
-      "epoch": 0.16667622378959038,
-      "grad_norm": 0.0009685232653282583,
-      "learning_rate": 0.00019999998641610897,
-      "loss": 46.0,
-      "step": 2180
-    },
-    {
-      "epoch": 0.1667526807729801,
-      "grad_norm": 0.0009498662548139691,
-      "learning_rate": 0.00019999998640358638,
-      "loss": 46.0,
-      "step": 2181
-    },
-    {
-      "epoch": 0.16682913775636982,
-      "grad_norm": 0.0013346386840566993,
-      "learning_rate": 0.000199999986391058,
-      "loss": 46.0,
-      "step": 2182
-    },
-    {
-      "epoch": 0.16690559473975955,
-      "grad_norm": 0.0021585833746939898,
-      "learning_rate": 0.00019999998637852383,
-      "loss": 46.0,
-      "step": 2183
-    },
-    {
-      "epoch": 0.16698205172314926,
-      "grad_norm": 0.0008964298758655787,
-      "learning_rate": 0.00019999998636598395,
-      "loss": 46.0,
-      "step": 2184
-    },
-    {
-      "epoch": 0.167058508706539,
-      "grad_norm": 0.001154283294454217,
-      "learning_rate": 0.00019999998635343826,
-      "loss": 46.0,
-      "step": 2185
-    },
-    {
-      "epoch": 0.1671349656899287,
-      "grad_norm": 0.001135069178417325,
-      "learning_rate": 0.00019999998634088684,
-      "loss": 46.0,
-      "step": 2186
-    },
-    {
-      "epoch": 0.16721142267331843,
-      "grad_norm": 0.002007537754252553,
-      "learning_rate": 0.00019999998632832958,
-      "loss": 46.0,
-      "step": 2187
-    },
-    {
-      "epoch": 0.16728787965670813,
-      "grad_norm": 0.0010732619557529688,
-      "learning_rate": 0.00019999998631576663,
-      "loss": 46.0,
-      "step": 2188
-    },
-    {
-      "epoch": 0.16736433664009787,
-      "grad_norm": 0.002044120104983449,
-      "learning_rate": 0.00019999998630319788,
-      "loss": 46.0,
-      "step": 2189
-    },
-    {
-      "epoch": 0.16744079362348757,
-      "grad_norm": 0.0047595989890396595,
-      "learning_rate": 0.00019999998629062336,
-      "loss": 46.0,
-      "step": 2190
-    },
-    {
-      "epoch": 0.1675172506068773,
-      "grad_norm": 0.0016334770480170846,
-      "learning_rate": 0.00019999998627804306,
-      "loss": 46.0,
-      "step": 2191
-    },
-    {
-      "epoch": 0.16759370759026704,
-      "grad_norm": 0.0022198061924427748,
-      "learning_rate": 0.000199999986265457,
-      "loss": 46.0,
-      "step": 2192
-    },
-    {
-      "epoch": 0.16767016457365674,
-      "grad_norm": 0.001590372179634869,
-      "learning_rate": 0.00019999998625286515,
-      "loss": 46.0,
-      "step": 2193
-    },
-    {
-      "epoch": 0.16774662155704648,
-      "grad_norm": 0.002364665037021041,
-      "learning_rate": 0.00019999998624026753,
-      "loss": 46.0,
-      "step": 2194
-    },
-    {
-      "epoch": 0.16782307854043618,
-      "grad_norm": 0.0008185718324966729,
-      "learning_rate": 0.00019999998622766417,
-      "loss": 46.0,
-      "step": 2195
-    },
-    {
-      "epoch": 0.16789953552382592,
-      "grad_norm": 0.0012419706908985972,
-      "learning_rate": 0.00019999998621505504,
-      "loss": 46.0,
-      "step": 2196
-    },
-    {
-      "epoch": 0.16797599250721562,
-      "grad_norm": 0.0007304903701879084,
-      "learning_rate": 0.00019999998620244013,
-      "loss": 46.0,
-      "step": 2197
-    },
-    {
-      "epoch": 0.16805244949060535,
-      "grad_norm": 0.0011022234102711082,
-      "learning_rate": 0.00019999998618981944,
-      "loss": 46.0,
-      "step": 2198
-    },
-    {
-      "epoch": 0.16812890647399506,
-      "grad_norm": 0.0006423944141715765,
-      "learning_rate": 0.000199999986177193,
-      "loss": 46.0,
-      "step": 2199
-    },
-    {
-      "epoch": 0.1682053634573848,
-      "grad_norm": 0.0011182207381352782,
-      "learning_rate": 0.00019999998616456076,
-      "loss": 46.0,
-      "step": 2200
-    },
-    {
-      "epoch": 0.1682818204407745,
-      "grad_norm": 0.0011730561964213848,
-      "learning_rate": 0.00019999998615192278,
-      "loss": 46.0,
-      "step": 2201
-    },
-    {
-      "epoch": 0.16835827742416423,
-      "grad_norm": 0.0008415406919084489,
-      "learning_rate": 0.00019999998613927903,
-      "loss": 46.0,
-      "step": 2202
-    },
-    {
-      "epoch": 0.16843473440755394,
-      "grad_norm": 0.0019437976879999042,
-      "learning_rate": 0.0001999999861266295,
-      "loss": 46.0,
-      "step": 2203
-    },
-    {
-      "epoch": 0.16851119139094367,
-      "grad_norm": 0.0013399278977885842,
-      "learning_rate": 0.00019999998611397422,
-      "loss": 46.0,
-      "step": 2204
-    },
-    {
-      "epoch": 0.1685876483743334,
-      "grad_norm": 0.0005270153633318841,
-      "learning_rate": 0.00019999998610131315,
-      "loss": 46.0,
-      "step": 2205
-    },
-    {
-      "epoch": 0.1686641053577231,
-      "grad_norm": 0.0006866326439194381,
-      "learning_rate": 0.0001999999860886463,
-      "loss": 46.0,
-      "step": 2206
-    },
-    {
-      "epoch": 0.16874056234111284,
-      "grad_norm": 0.0012157591991126537,
-      "learning_rate": 0.0001999999860759737,
-      "loss": 46.0,
-      "step": 2207
-    },
-    {
-      "epoch": 0.16881701932450255,
-      "grad_norm": 0.0005165923503227532,
-      "learning_rate": 0.00019999998606329533,
-      "loss": 46.0,
-      "step": 2208
-    },
-    {
-      "epoch": 0.16889347630789228,
-      "grad_norm": 0.008756878785789013,
-      "learning_rate": 0.00019999998605061117,
-      "loss": 46.0,
-      "step": 2209
-    },
-    {
-      "epoch": 0.168969933291282,
-      "grad_norm": 0.0021029466297477484,
-      "learning_rate": 0.00019999998603792128,
-      "loss": 46.0,
-      "step": 2210
-    },
-    {
-      "epoch": 0.16904639027467172,
-      "grad_norm": 0.0008491664775647223,
-      "learning_rate": 0.00019999998602522557,
-      "loss": 46.0,
-      "step": 2211
-    },
-    {
-      "epoch": 0.16912284725806143,
-      "grad_norm": 0.0011756946332752705,
-      "learning_rate": 0.00019999998601252412,
-      "loss": 46.0,
-      "step": 2212
-    },
-    {
-      "epoch": 0.16919930424145116,
-      "grad_norm": 0.0006581502384506166,
-      "learning_rate": 0.00019999998599981692,
-      "loss": 46.0,
-      "step": 2213
-    },
-    {
-      "epoch": 0.16927576122484086,
-      "grad_norm": 0.0003999524633400142,
-      "learning_rate": 0.00019999998598710391,
-      "loss": 46.0,
-      "step": 2214
-    },
-    {
-      "epoch": 0.1693522182082306,
-      "grad_norm": 0.0012821991695091128,
-      "learning_rate": 0.00019999998597438517,
-      "loss": 46.0,
-      "step": 2215
-    },
-    {
-      "epoch": 0.1694286751916203,
-      "grad_norm": 0.0007446478120982647,
-      "learning_rate": 0.00019999998596166062,
-      "loss": 46.0,
-      "step": 2216
-    },
-    {
-      "epoch": 0.16950513217501004,
-      "grad_norm": 0.0012399741681292653,
-      "learning_rate": 0.00019999998594893032,
-      "loss": 46.0,
-      "step": 2217
-    },
-    {
-      "epoch": 0.16958158915839974,
-      "grad_norm": 0.00047427957179024816,
-      "learning_rate": 0.00019999998593619425,
-      "loss": 46.0,
-      "step": 2218
-    },
-    {
-      "epoch": 0.16965804614178948,
-      "grad_norm": 0.0009618039475753903,
-      "learning_rate": 0.00019999998592345238,
-      "loss": 46.0,
-      "step": 2219
-    },
-    {
-      "epoch": 0.1697345031251792,
-      "grad_norm": 0.0028572191949933767,
-      "learning_rate": 0.00019999998591070482,
-      "loss": 46.0,
-      "step": 2220
-    },
-    {
-      "epoch": 0.16981096010856891,
-      "grad_norm": 0.0007253816584125161,
-      "learning_rate": 0.0001999999858979514,
-      "loss": 46.0,
-      "step": 2221
-    },
-    {
-      "epoch": 0.16988741709195865,
-      "grad_norm": 0.0007867772947065532,
-      "learning_rate": 0.0001999999858851923,
-      "loss": 46.0,
-      "step": 2222
-    },
-    {
-      "epoch": 0.16996387407534835,
-      "grad_norm": 0.0018229043344035745,
-      "learning_rate": 0.00019999998587242736,
-      "loss": 46.0,
-      "step": 2223
-    },
-    {
-      "epoch": 0.1700403310587381,
-      "grad_norm": 0.0021692633163183928,
-      "learning_rate": 0.00019999998585965668,
-      "loss": 46.0,
-      "step": 2224
-    },
-    {
-      "epoch": 0.1701167880421278,
-      "grad_norm": 0.001353766187094152,
-      "learning_rate": 0.00019999998584688023,
-      "loss": 46.0,
-      "step": 2225
-    },
-    {
-      "epoch": 0.17019324502551753,
-      "grad_norm": 0.00323905935510993,
-      "learning_rate": 0.000199999985834098,
-      "loss": 46.0,
-      "step": 2226
-    },
-    {
-      "epoch": 0.17026970200890723,
-      "grad_norm": 0.0014555241214111447,
-      "learning_rate": 0.00019999998582131002,
-      "loss": 46.0,
-      "step": 2227
-    },
-    {
-      "epoch": 0.17034615899229696,
-      "grad_norm": 0.0011925987200811505,
-      "learning_rate": 0.00019999998580851625,
-      "loss": 46.0,
-      "step": 2228
-    },
-    {
-      "epoch": 0.17042261597568667,
-      "grad_norm": 0.00040810563950799406,
-      "learning_rate": 0.0001999999857957167,
-      "loss": 46.0,
-      "step": 2229
-    },
-    {
-      "epoch": 0.1704990729590764,
-      "grad_norm": 0.0016967917326837778,
-      "learning_rate": 0.0001999999857829114,
-      "loss": 46.0,
-      "step": 2230
-    },
-    {
-      "epoch": 0.1705755299424661,
-      "grad_norm": 0.0008670039242133498,
-      "learning_rate": 0.00019999998577010033,
-      "loss": 46.0,
-      "step": 2231
-    },
-    {
-      "epoch": 0.17065198692585584,
-      "grad_norm": 0.0011233967961743474,
-      "learning_rate": 0.0001999999857572835,
-      "loss": 46.0,
-      "step": 2232
-    },
-    {
-      "epoch": 0.17072844390924555,
-      "grad_norm": 0.0011340989731252193,
-      "learning_rate": 0.00019999998574446085,
-      "loss": 46.0,
-      "step": 2233
-    },
-    {
-      "epoch": 0.17080490089263528,
-      "grad_norm": 0.0021429695188999176,
-      "learning_rate": 0.0001999999857316325,
-      "loss": 46.0,
-      "step": 2234
-    },
-    {
-      "epoch": 0.17088135787602501,
-      "grad_norm": 0.005632530432194471,
-      "learning_rate": 0.00019999998571879835,
-      "loss": 46.0,
-      "step": 2235
-    },
-    {
-      "epoch": 0.17095781485941472,
-      "grad_norm": 0.0027736122719943523,
-      "learning_rate": 0.00019999998570595842,
-      "loss": 46.0,
-      "step": 2236
-    },
-    {
-      "epoch": 0.17103427184280445,
-      "grad_norm": 0.0009647748083807528,
-      "learning_rate": 0.00019999998569311274,
-      "loss": 46.0,
-      "step": 2237
-    },
-    {
-      "epoch": 0.17111072882619416,
-      "grad_norm": 0.0012944993795827031,
-      "learning_rate": 0.00019999998568026128,
-      "loss": 46.0,
-      "step": 2238
-    },
-    {
-      "epoch": 0.1711871858095839,
-      "grad_norm": 0.0036405797582119703,
-      "learning_rate": 0.00019999998566740405,
-      "loss": 46.0,
-      "step": 2239
-    },
-    {
-      "epoch": 0.1712636427929736,
-      "grad_norm": 0.0016364466864615679,
-      "learning_rate": 0.00019999998565454105,
-      "loss": 46.0,
-      "step": 2240
-    },
-    {
-      "epoch": 0.17134009977636333,
-      "grad_norm": 0.001100799418054521,
-      "learning_rate": 0.00019999998564167228,
-      "loss": 46.0,
-      "step": 2241
-    },
-    {
-      "epoch": 0.17141655675975304,
-      "grad_norm": 0.000934005540329963,
-      "learning_rate": 0.00019999998562879773,
-      "loss": 46.0,
-      "step": 2242
-    },
-    {
-      "epoch": 0.17149301374314277,
-      "grad_norm": 0.0016864098142832518,
-      "learning_rate": 0.00019999998561591743,
-      "loss": 46.0,
-      "step": 2243
-    },
-    {
-      "epoch": 0.17156947072653247,
-      "grad_norm": 0.0006700264639221132,
-      "learning_rate": 0.00019999998560303137,
-      "loss": 46.0,
-      "step": 2244
-    },
-    {
-      "epoch": 0.1716459277099222,
-      "grad_norm": 0.0008497000089846551,
-      "learning_rate": 0.0001999999855901395,
-      "loss": 46.0,
-      "step": 2245
-    },
-    {
-      "epoch": 0.1717223846933119,
-      "grad_norm": 0.001969871111214161,
-      "learning_rate": 0.0001999999855772419,
-      "loss": 46.0,
-      "step": 2246
-    },
-    {
-      "epoch": 0.17179884167670165,
-      "grad_norm": 0.0012944393092766404,
-      "learning_rate": 0.00019999998556433852,
-      "loss": 46.0,
-      "step": 2247
-    },
-    {
-      "epoch": 0.17187529866009135,
-      "grad_norm": 0.0018585756188258529,
-      "learning_rate": 0.00019999998555142936,
-      "loss": 46.0,
-      "step": 2248
-    },
-    {
-      "epoch": 0.17195175564348109,
-      "grad_norm": 0.0005201755557209253,
-      "learning_rate": 0.00019999998553851445,
-      "loss": 46.0,
-      "step": 2249
-    },
-    {
-      "epoch": 0.17202821262687082,
-      "grad_norm": 0.0038005460519343615,
-      "learning_rate": 0.00019999998552559377,
-      "loss": 46.0,
-      "step": 2250
-    },
-    {
-      "epoch": 0.17210466961026052,
-      "grad_norm": 0.00512295588850975,
-      "learning_rate": 0.0001999999855126673,
-      "loss": 46.0,
-      "step": 2251
-    },
-    {
-      "epoch": 0.17218112659365026,
-      "grad_norm": 0.0014348982367664576,
-      "learning_rate": 0.00019999998549973506,
-      "loss": 46.0,
-      "step": 2252
-    },
-    {
-      "epoch": 0.17225758357703996,
-      "grad_norm": 0.0007697799592278898,
-      "learning_rate": 0.00019999998548679706,
-      "loss": 46.0,
-      "step": 2253
-    },
-    {
-      "epoch": 0.1723340405604297,
-      "grad_norm": 0.0014597177505493164,
-      "learning_rate": 0.00019999998547385329,
-      "loss": 46.0,
-      "step": 2254
-    },
-    {
-      "epoch": 0.1724104975438194,
-      "grad_norm": 0.0006445308681577444,
-      "learning_rate": 0.00019999998546090374,
-      "loss": 46.0,
-      "step": 2255
-    },
-    {
-      "epoch": 0.17248695452720914,
-      "grad_norm": 0.0006086422363296151,
-      "learning_rate": 0.00019999998544794844,
-      "loss": 46.0,
-      "step": 2256
-    },
-    {
-      "epoch": 0.17256341151059884,
-      "grad_norm": 0.0010298449778929353,
-      "learning_rate": 0.00019999998543498738,
-      "loss": 46.0,
-      "step": 2257
-    },
-    {
-      "epoch": 0.17263986849398857,
-      "grad_norm": 0.0030916971154510975,
-      "learning_rate": 0.0001999999854220205,
-      "loss": 46.0,
-      "step": 2258
-    },
-    {
-      "epoch": 0.17271632547737828,
-      "grad_norm": 0.0019768341444432735,
-      "learning_rate": 0.0001999999854090479,
-      "loss": 46.0,
-      "step": 2259
-    },
-    {
-      "epoch": 0.172792782460768,
-      "grad_norm": 0.05438971891999245,
-      "learning_rate": 0.0001999999853960695,
-      "loss": 46.0,
-      "step": 2260
-    },
-    {
-      "epoch": 0.17286923944415772,
-      "grad_norm": 0.0028179828077554703,
-      "learning_rate": 0.00019999998538308534,
-      "loss": 46.0,
-      "step": 2261
-    },
-    {
-      "epoch": 0.17294569642754745,
-      "grad_norm": 0.0009732816251926124,
-      "learning_rate": 0.00019999998537009544,
-      "loss": 46.0,
-      "step": 2262
-    },
-    {
-      "epoch": 0.17302215341093719,
-      "grad_norm": 0.0005932300700806081,
-      "learning_rate": 0.00019999998535709973,
-      "loss": 46.0,
-      "step": 2263
-    },
-    {
-      "epoch": 0.1730986103943269,
-      "grad_norm": 0.0023620782885700464,
-      "learning_rate": 0.00019999998534409827,
-      "loss": 46.0,
-      "step": 2264
-    },
-    {
-      "epoch": 0.17317506737771662,
-      "grad_norm": 0.0008549176855012774,
-      "learning_rate": 0.00019999998533109105,
-      "loss": 46.0,
-      "step": 2265
-    },
-    {
-      "epoch": 0.17325152436110633,
-      "grad_norm": 0.0017323002684861422,
-      "learning_rate": 0.00019999998531807802,
-      "loss": 46.0,
-      "step": 2266
-    },
-    {
-      "epoch": 0.17332798134449606,
-      "grad_norm": 0.0014616270782426,
-      "learning_rate": 0.00019999998530505927,
-      "loss": 46.0,
-      "step": 2267
-    },
-    {
-      "epoch": 0.17340443832788577,
-      "grad_norm": 0.002720715943723917,
-      "learning_rate": 0.00019999998529203472,
-      "loss": 46.0,
-      "step": 2268
-    },
-    {
-      "epoch": 0.1734808953112755,
-      "grad_norm": 0.0011732501443475485,
-      "learning_rate": 0.0001999999852790044,
-      "loss": 46.0,
-      "step": 2269
-    },
-    {
-      "epoch": 0.1735573522946652,
-      "grad_norm": 0.001845426158979535,
-      "learning_rate": 0.00019999998526596833,
-      "loss": 46.0,
-      "step": 2270
-    },
-    {
-      "epoch": 0.17363380927805494,
-      "grad_norm": 0.000726019439753145,
-      "learning_rate": 0.0001999999852529265,
-      "loss": 46.0,
-      "step": 2271
-    },
-    {
-      "epoch": 0.17371026626144465,
-      "grad_norm": 0.0007406814838759601,
-      "learning_rate": 0.00019999998523987885,
-      "loss": 46.0,
-      "step": 2272
-    },
-    {
-      "epoch": 0.17378672324483438,
-      "grad_norm": 0.0009963270276784897,
-      "learning_rate": 0.00019999998522682546,
-      "loss": 46.0,
-      "step": 2273
-    },
-    {
-      "epoch": 0.17386318022822408,
-      "grad_norm": 0.0013505546376109123,
-      "learning_rate": 0.0001999999852137663,
-      "loss": 46.0,
-      "step": 2274
-    },
-    {
-      "epoch": 0.17393963721161382,
-      "grad_norm": 0.0020522945560514927,
-      "learning_rate": 0.00019999998520070137,
-      "loss": 46.0,
-      "step": 2275
-    },
-    {
-      "epoch": 0.17401609419500352,
-      "grad_norm": 0.0008776316535659134,
-      "learning_rate": 0.0001999999851876307,
-      "loss": 46.0,
-      "step": 2276
-    },
-    {
-      "epoch": 0.17409255117839326,
-      "grad_norm": 0.002591426018625498,
-      "learning_rate": 0.0001999999851745542,
-      "loss": 46.0,
-      "step": 2277
-    },
-    {
-      "epoch": 0.174169008161783,
-      "grad_norm": 0.001668052631430328,
-      "learning_rate": 0.00019999998516147198,
-      "loss": 46.0,
-      "step": 2278
-    },
-    {
-      "epoch": 0.1742454651451727,
-      "grad_norm": 0.0009902473539113998,
-      "learning_rate": 0.00019999998514838398,
-      "loss": 46.0,
-      "step": 2279
-    },
-    {
-      "epoch": 0.17432192212856243,
-      "grad_norm": 0.0013911043060943484,
-      "learning_rate": 0.00019999998513529018,
-      "loss": 46.0,
-      "step": 2280
-    },
-    {
-      "epoch": 0.17439837911195213,
-      "grad_norm": 0.0019023942295461893,
-      "learning_rate": 0.00019999998512219066,
-      "loss": 46.0,
-      "step": 2281
-    },
-    {
-      "epoch": 0.17447483609534187,
-      "grad_norm": 0.0009533657575957477,
-      "learning_rate": 0.00019999998510908534,
-      "loss": 46.0,
-      "step": 2282
-    },
-    {
-      "epoch": 0.17455129307873157,
-      "grad_norm": 0.000650926842354238,
-      "learning_rate": 0.00019999998509597427,
-      "loss": 46.0,
-      "step": 2283
-    },
-    {
-      "epoch": 0.1746277500621213,
-      "grad_norm": 0.0038995062932372093,
-      "learning_rate": 0.0001999999850828574,
-      "loss": 46.0,
-      "step": 2284
-    },
-    {
-      "epoch": 0.174704207045511,
-      "grad_norm": 0.0018009381601586938,
-      "learning_rate": 0.00019999998506973476,
-      "loss": 46.0,
-      "step": 2285
-    },
-    {
-      "epoch": 0.17478066402890075,
-      "grad_norm": 0.0023535252548754215,
-      "learning_rate": 0.00019999998505660637,
-      "loss": 46.0,
-      "step": 2286
-    },
-    {
-      "epoch": 0.17485712101229045,
-      "grad_norm": 0.0011501583503559232,
-      "learning_rate": 0.0001999999850434722,
-      "loss": 46.0,
-      "step": 2287
-    },
-    {
-      "epoch": 0.17493357799568018,
-      "grad_norm": 0.0007303390302695334,
-      "learning_rate": 0.00019999998503033228,
-      "loss": 46.0,
-      "step": 2288
-    },
-    {
-      "epoch": 0.1750100349790699,
-      "grad_norm": 0.0006898452411405742,
-      "learning_rate": 0.0001999999850171866,
-      "loss": 46.0,
-      "step": 2289
-    },
-    {
-      "epoch": 0.17508649196245962,
-      "grad_norm": 0.000676496943924576,
-      "learning_rate": 0.00019999998500403514,
-      "loss": 46.0,
-      "step": 2290
-    },
-    {
-      "epoch": 0.17516294894584933,
-      "grad_norm": 0.0024054618552327156,
-      "learning_rate": 0.0001999999849908779,
-      "loss": 46.0,
-      "step": 2291
-    },
-    {
-      "epoch": 0.17523940592923906,
-      "grad_norm": 0.0006575587904080749,
-      "learning_rate": 0.00019999998497771486,
-      "loss": 46.0,
-      "step": 2292
-    },
-    {
-      "epoch": 0.1753158629126288,
-      "grad_norm": 0.001693792175501585,
-      "learning_rate": 0.00019999998496454612,
-      "loss": 46.0,
-      "step": 2293
-    },
-    {
-      "epoch": 0.1753923198960185,
-      "grad_norm": 0.0007719940040260553,
-      "learning_rate": 0.00019999998495137154,
-      "loss": 46.0,
-      "step": 2294
-    },
-    {
-      "epoch": 0.17546877687940823,
-      "grad_norm": 0.0008060596883296967,
-      "learning_rate": 0.00019999998493819122,
-      "loss": 46.0,
-      "step": 2295
-    },
-    {
-      "epoch": 0.17554523386279794,
-      "grad_norm": 0.0019577713683247566,
-      "learning_rate": 0.00019999998492500515,
-      "loss": 46.0,
-      "step": 2296
-    },
-    {
-      "epoch": 0.17562169084618767,
-      "grad_norm": 0.0019999847281724215,
-      "learning_rate": 0.0001999999849118133,
-      "loss": 46.0,
-      "step": 2297
-    },
-    {
-      "epoch": 0.17569814782957738,
-      "grad_norm": 0.009767111390829086,
-      "learning_rate": 0.00019999998489861567,
-      "loss": 46.0,
-      "step": 2298
-    },
-    {
-      "epoch": 0.1757746048129671,
-      "grad_norm": 0.0017186939949169755,
-      "learning_rate": 0.00019999998488541226,
-      "loss": 46.0,
-      "step": 2299
-    },
-    {
-      "epoch": 0.17585106179635682,
-      "grad_norm": 0.0011332217836752534,
-      "learning_rate": 0.0001999999848722031,
-      "loss": 46.0,
-      "step": 2300
-    },
-    {
-      "epoch": 0.17592751877974655,
-      "grad_norm": 0.006631455849856138,
-      "learning_rate": 0.00019999998485898817,
-      "loss": 46.0,
-      "step": 2301
-    },
-    {
-      "epoch": 0.17600397576313626,
-      "grad_norm": 0.0012229913845658302,
-      "learning_rate": 0.00019999998484576749,
-      "loss": 46.0,
-      "step": 2302
-    },
-    {
-      "epoch": 0.176080432746526,
-      "grad_norm": 0.0012728262227028608,
-      "learning_rate": 0.00019999998483254098,
-      "loss": 46.0,
-      "step": 2303
-    },
-    {
-      "epoch": 0.1761568897299157,
-      "grad_norm": 0.004363963846117258,
-      "learning_rate": 0.00019999998481930875,
-      "loss": 46.0,
-      "step": 2304
-    },
-    {
-      "epoch": 0.17623334671330543,
-      "grad_norm": 0.0035214009694755077,
-      "learning_rate": 0.00019999998480607075,
-      "loss": 46.0,
-      "step": 2305
-    },
-    {
-      "epoch": 0.17630980369669513,
-      "grad_norm": 0.0012037267442792654,
-      "learning_rate": 0.00019999998479282695,
-      "loss": 46.0,
-      "step": 2306
-    },
-    {
-      "epoch": 0.17638626068008487,
-      "grad_norm": 0.0009421817376278341,
-      "learning_rate": 0.0001999999847795774,
-      "loss": 46.0,
-      "step": 2307
-    },
-    {
-      "epoch": 0.1764627176634746,
-      "grad_norm": 0.0012976977741345763,
-      "learning_rate": 0.00019999998476632208,
-      "loss": 46.0,
-      "step": 2308
-    },
-    {
-      "epoch": 0.1765391746468643,
-      "grad_norm": 0.00046510202810168266,
-      "learning_rate": 0.00019999998475306102,
-      "loss": 46.0,
-      "step": 2309
-    },
-    {
-      "epoch": 0.17661563163025404,
-      "grad_norm": 0.003901679767295718,
-      "learning_rate": 0.00019999998473979412,
-      "loss": 46.0,
-      "step": 2310
-    },
-    {
-      "epoch": 0.17669208861364374,
-      "grad_norm": 0.0007329244399443269,
-      "learning_rate": 0.0001999999847265215,
-      "loss": 46.0,
-      "step": 2311
-    },
-    {
-      "epoch": 0.17676854559703348,
-      "grad_norm": 0.0011444626143202186,
-      "learning_rate": 0.0001999999847132431,
-      "loss": 46.0,
-      "step": 2312
-    },
-    {
-      "epoch": 0.17684500258042318,
-      "grad_norm": 0.000476592977065593,
-      "learning_rate": 0.00019999998469995896,
-      "loss": 46.0,
-      "step": 2313
-    },
-    {
-      "epoch": 0.17692145956381292,
-      "grad_norm": 0.0006228131242096424,
-      "learning_rate": 0.000199999984686669,
-      "loss": 46.0,
-      "step": 2314
-    },
-    {
-      "epoch": 0.17699791654720262,
-      "grad_norm": 0.001816373085603118,
-      "learning_rate": 0.0001999999846733733,
-      "loss": 46.0,
-      "step": 2315
-    },
-    {
-      "epoch": 0.17707437353059235,
-      "grad_norm": 0.0008470119209960103,
-      "learning_rate": 0.00019999998466007184,
-      "loss": 46.0,
-      "step": 2316
-    },
-    {
-      "epoch": 0.17715083051398206,
-      "grad_norm": 0.0010827983496710658,
-      "learning_rate": 0.00019999998464676456,
-      "loss": 46.0,
-      "step": 2317
-    },
-    {
-      "epoch": 0.1772272874973718,
-      "grad_norm": 0.0013238334795460105,
-      "learning_rate": 0.00019999998463345156,
-      "loss": 46.0,
-      "step": 2318
-    },
-    {
-      "epoch": 0.1773037444807615,
-      "grad_norm": 0.0010916294995695353,
-      "learning_rate": 0.00019999998462013276,
-      "loss": 46.0,
-      "step": 2319
-    },
-    {
-      "epoch": 0.17738020146415123,
-      "grad_norm": 0.0005368746933527291,
-      "learning_rate": 0.00019999998460680822,
-      "loss": 46.0,
-      "step": 2320
-    },
-    {
-      "epoch": 0.17745665844754097,
-      "grad_norm": 0.0010060060303658247,
-      "learning_rate": 0.0001999999845934779,
-      "loss": 46.0,
-      "step": 2321
-    },
-    {
-      "epoch": 0.17753311543093067,
-      "grad_norm": 0.003440416418015957,
-      "learning_rate": 0.00019999998458014183,
-      "loss": 46.0,
-      "step": 2322
-    },
-    {
-      "epoch": 0.1776095724143204,
-      "grad_norm": 0.0006268015713430941,
-      "learning_rate": 0.00019999998456679996,
-      "loss": 46.0,
-      "step": 2323
-    },
-    {
-      "epoch": 0.1776860293977101,
-      "grad_norm": 0.0026197130791842937,
-      "learning_rate": 0.00019999998455345232,
-      "loss": 46.0,
-      "step": 2324
-    },
-    {
-      "epoch": 0.17776248638109984,
-      "grad_norm": 0.0009889118373394012,
-      "learning_rate": 0.0001999999845400989,
-      "loss": 46.0,
-      "step": 2325
-    },
-    {
-      "epoch": 0.17783894336448955,
-      "grad_norm": 0.0011897907825186849,
-      "learning_rate": 0.00019999998452673975,
-      "loss": 46.0,
-      "step": 2326
-    },
-    {
-      "epoch": 0.17791540034787928,
-      "grad_norm": 0.0011173890670761466,
-      "learning_rate": 0.00019999998451337482,
-      "loss": 46.0,
-      "step": 2327
-    },
-    {
-      "epoch": 0.177991857331269,
-      "grad_norm": 0.0017595734680071473,
-      "learning_rate": 0.0001999999845000041,
-      "loss": 46.0,
-      "step": 2328
-    },
-    {
-      "epoch": 0.17806831431465872,
-      "grad_norm": 0.0009243313106708229,
-      "learning_rate": 0.0001999999844866276,
-      "loss": 46.0,
-      "step": 2329
-    },
-    {
-      "epoch": 0.17814477129804843,
-      "grad_norm": 0.001011726912111044,
-      "learning_rate": 0.00019999998447324538,
-      "loss": 46.0,
-      "step": 2330
-    },
-    {
-      "epoch": 0.17822122828143816,
-      "grad_norm": 0.0018673008307814598,
-      "learning_rate": 0.00019999998445985735,
-      "loss": 46.0,
-      "step": 2331
-    },
-    {
-      "epoch": 0.17829768526482787,
-      "grad_norm": 0.0012218868359923363,
-      "learning_rate": 0.00019999998444646355,
-      "loss": 46.0,
-      "step": 2332
-    },
-    {
-      "epoch": 0.1783741422482176,
-      "grad_norm": 0.0010105993133038282,
-      "learning_rate": 0.000199999984433064,
-      "loss": 46.0,
-      "step": 2333
-    },
-    {
-      "epoch": 0.1784505992316073,
-      "grad_norm": 0.000628974346909672,
-      "learning_rate": 0.00019999998441965868,
-      "loss": 46.0,
-      "step": 2334
-    },
-    {
-      "epoch": 0.17852705621499704,
-      "grad_norm": 0.001163146342150867,
-      "learning_rate": 0.0001999999844062476,
-      "loss": 46.0,
-      "step": 2335
-    },
-    {
-      "epoch": 0.17860351319838677,
-      "grad_norm": 0.0012094784760847688,
-      "learning_rate": 0.00019999998439283073,
-      "loss": 46.0,
-      "step": 2336
-    },
-    {
-      "epoch": 0.17867997018177648,
-      "grad_norm": 0.0059892479330301285,
-      "learning_rate": 0.00019999998437940809,
-      "loss": 46.0,
-      "step": 2337
-    },
-    {
-      "epoch": 0.1787564271651662,
-      "grad_norm": 0.0007704837480559945,
-      "learning_rate": 0.00019999998436597967,
-      "loss": 46.0,
-      "step": 2338
-    },
-    {
-      "epoch": 0.17883288414855592,
-      "grad_norm": 0.002447045873850584,
-      "learning_rate": 0.0001999999843525455,
-      "loss": 46.0,
-      "step": 2339
-    },
-    {
-      "epoch": 0.17890934113194565,
-      "grad_norm": 0.003908355720341206,
-      "learning_rate": 0.00019999998433910555,
-      "loss": 46.0,
-      "step": 2340
-    },
-    {
-      "epoch": 0.17898579811533535,
-      "grad_norm": 0.0011193696409463882,
-      "learning_rate": 0.00019999998432565985,
-      "loss": 46.0,
-      "step": 2341
-    },
-    {
-      "epoch": 0.1790622550987251,
-      "grad_norm": 0.0011555581586435437,
-      "learning_rate": 0.00019999998431220837,
-      "loss": 46.0,
-      "step": 2342
-    },
-    {
-      "epoch": 0.1791387120821148,
-      "grad_norm": 0.002081210259348154,
-      "learning_rate": 0.00019999998429875114,
-      "loss": 46.0,
-      "step": 2343
-    },
-    {
-      "epoch": 0.17921516906550453,
-      "grad_norm": 0.0015744606498628855,
-      "learning_rate": 0.0001999999842852881,
-      "loss": 46.0,
-      "step": 2344
-    },
-    {
-      "epoch": 0.17929162604889423,
-      "grad_norm": 0.0026379143819212914,
-      "learning_rate": 0.0001999999842718193,
-      "loss": 46.0,
-      "step": 2345
-    },
-    {
-      "epoch": 0.17936808303228396,
-      "grad_norm": 0.000804356241133064,
-      "learning_rate": 0.00019999998425834474,
-      "loss": 46.0,
-      "step": 2346
-    },
-    {
-      "epoch": 0.17944454001567367,
-      "grad_norm": 0.0028589973226189613,
-      "learning_rate": 0.00019999998424486442,
-      "loss": 46.0,
-      "step": 2347
-    },
-    {
-      "epoch": 0.1795209969990634,
-      "grad_norm": 0.0017277380684390664,
-      "learning_rate": 0.0001999999842313783,
-      "loss": 46.0,
-      "step": 2348
-    },
-    {
-      "epoch": 0.1795974539824531,
-      "grad_norm": 0.0008444335544481874,
-      "learning_rate": 0.00019999998421788644,
-      "loss": 46.0,
-      "step": 2349
-    },
-    {
-      "epoch": 0.17967391096584284,
-      "grad_norm": 0.0007687475881539285,
-      "learning_rate": 0.00019999998420438883,
-      "loss": 46.0,
-      "step": 2350
-    },
-    {
-      "epoch": 0.17975036794923258,
-      "grad_norm": 0.0011582612060010433,
-      "learning_rate": 0.0001999999841908854,
-      "loss": 46.0,
-      "step": 2351
-    },
-    {
-      "epoch": 0.17982682493262228,
-      "grad_norm": 0.0004720762080978602,
-      "learning_rate": 0.00019999998417737623,
-      "loss": 46.0,
-      "step": 2352
-    },
-    {
-      "epoch": 0.17990328191601201,
-      "grad_norm": 0.0012853079242631793,
-      "learning_rate": 0.0001999999841638613,
-      "loss": 46.0,
-      "step": 2353
-    },
-    {
-      "epoch": 0.17997973889940172,
-      "grad_norm": 0.0006842748844064772,
-      "learning_rate": 0.00019999998415034056,
-      "loss": 46.0,
-      "step": 2354
-    },
-    {
-      "epoch": 0.18005619588279145,
-      "grad_norm": 0.0013378605945035815,
-      "learning_rate": 0.00019999998413681408,
-      "loss": 46.0,
-      "step": 2355
-    },
-    {
-      "epoch": 0.18013265286618116,
-      "grad_norm": 0.0008957096724770963,
-      "learning_rate": 0.00019999998412328186,
-      "loss": 46.0,
-      "step": 2356
-    },
-    {
-      "epoch": 0.1802091098495709,
-      "grad_norm": 0.0009715714841149747,
-      "learning_rate": 0.0001999999841097438,
-      "loss": 46.0,
-      "step": 2357
-    },
-    {
-      "epoch": 0.1802855668329606,
-      "grad_norm": 0.0034031725954264402,
-      "learning_rate": 0.00019999998409620003,
-      "loss": 46.0,
-      "step": 2358
-    },
-    {
-      "epoch": 0.18036202381635033,
-      "grad_norm": 0.0007429048418998718,
-      "learning_rate": 0.00019999998408265046,
-      "loss": 46.0,
-      "step": 2359
-    },
-    {
-      "epoch": 0.18043848079974004,
-      "grad_norm": 0.0018524250481277704,
-      "learning_rate": 0.0001999999840690951,
-      "loss": 46.0,
-      "step": 2360
-    },
-    {
-      "epoch": 0.18051493778312977,
-      "grad_norm": 0.0015551699325442314,
-      "learning_rate": 0.00019999998405553402,
-      "loss": 46.0,
-      "step": 2361
-    },
-    {
-      "epoch": 0.18059139476651948,
-      "grad_norm": 0.000425168895162642,
-      "learning_rate": 0.00019999998404196715,
-      "loss": 46.0,
-      "step": 2362
-    },
-    {
-      "epoch": 0.1806678517499092,
-      "grad_norm": 0.001114976592361927,
-      "learning_rate": 0.00019999998402839452,
-      "loss": 46.0,
-      "step": 2363
-    },
-    {
-      "epoch": 0.18074430873329891,
-      "grad_norm": 0.0019761016592383385,
-      "learning_rate": 0.0001999999840148161,
-      "loss": 46.0,
-      "step": 2364
-    },
-    {
-      "epoch": 0.18082076571668865,
-      "grad_norm": 0.002594160847365856,
-      "learning_rate": 0.00019999998400123192,
-      "loss": 46.0,
-      "step": 2365
-    },
-    {
-      "epoch": 0.18089722270007838,
-      "grad_norm": 0.0005472972407005727,
-      "learning_rate": 0.00019999998398764196,
-      "loss": 46.0,
-      "step": 2366
-    },
-    {
-      "epoch": 0.18097367968346809,
-      "grad_norm": 0.0027455140370875597,
-      "learning_rate": 0.00019999998397404625,
-      "loss": 46.0,
-      "step": 2367
-    },
-    {
-      "epoch": 0.18105013666685782,
-      "grad_norm": 0.0005896030925214291,
-      "learning_rate": 0.00019999998396044477,
-      "loss": 46.0,
-      "step": 2368
-    },
-    {
-      "epoch": 0.18112659365024752,
-      "grad_norm": 0.0009163576178252697,
-      "learning_rate": 0.00019999998394683752,
-      "loss": 46.0,
-      "step": 2369
-    },
-    {
-      "epoch": 0.18120305063363726,
-      "grad_norm": 0.002880948130041361,
-      "learning_rate": 0.0001999999839332245,
-      "loss": 46.0,
-      "step": 2370
-    },
-    {
-      "epoch": 0.18127950761702696,
-      "grad_norm": 0.002056585159152746,
-      "learning_rate": 0.0001999999839196057,
-      "loss": 46.0,
-      "step": 2371
-    },
-    {
-      "epoch": 0.1813559646004167,
-      "grad_norm": 0.0008534000953659415,
-      "learning_rate": 0.00019999998390598112,
-      "loss": 46.0,
-      "step": 2372
-    },
-    {
-      "epoch": 0.1814324215838064,
-      "grad_norm": 0.0006491973763331771,
-      "learning_rate": 0.0001999999838923508,
-      "loss": 46.0,
-      "step": 2373
-    },
-    {
-      "epoch": 0.18150887856719614,
-      "grad_norm": 0.0004762164026033133,
-      "learning_rate": 0.00019999998387871469,
-      "loss": 46.0,
-      "step": 2374
-    },
-    {
-      "epoch": 0.18158533555058584,
-      "grad_norm": 0.0010661830892786384,
-      "learning_rate": 0.00019999998386507282,
-      "loss": 46.0,
-      "step": 2375
-    },
-    {
-      "epoch": 0.18166179253397557,
-      "grad_norm": 0.0011364857200533152,
-      "learning_rate": 0.00019999998385142518,
-      "loss": 46.0,
-      "step": 2376
-    },
-    {
-      "epoch": 0.18173824951736528,
-      "grad_norm": 0.0012326686410233378,
-      "learning_rate": 0.00019999998383777174,
-      "loss": 46.0,
-      "step": 2377
-    },
-    {
-      "epoch": 0.181814706500755,
-      "grad_norm": 0.000846969080157578,
-      "learning_rate": 0.00019999998382411258,
-      "loss": 46.0,
-      "step": 2378
-    },
-    {
-      "epoch": 0.18189116348414475,
-      "grad_norm": 0.0004108122084289789,
-      "learning_rate": 0.00019999998381044763,
-      "loss": 46.0,
-      "step": 2379
-    },
-    {
-      "epoch": 0.18196762046753445,
-      "grad_norm": 0.001905068987980485,
-      "learning_rate": 0.0001999999837967769,
-      "loss": 46.0,
-      "step": 2380
-    },
-    {
-      "epoch": 0.18204407745092419,
-      "grad_norm": 0.002693180926144123,
-      "learning_rate": 0.00019999998378310041,
-      "loss": 46.0,
-      "step": 2381
-    },
-    {
-      "epoch": 0.1821205344343139,
-      "grad_norm": 0.0005059579852968454,
-      "learning_rate": 0.00019999998376941816,
-      "loss": 46.0,
-      "step": 2382
-    },
-    {
-      "epoch": 0.18219699141770362,
-      "grad_norm": 0.0033266465179622173,
-      "learning_rate": 0.00019999998375573014,
-      "loss": 46.0,
-      "step": 2383
-    },
-    {
-      "epoch": 0.18227344840109333,
-      "grad_norm": 0.001252981717698276,
-      "learning_rate": 0.0001999999837420363,
-      "loss": 46.0,
-      "step": 2384
-    },
-    {
-      "epoch": 0.18234990538448306,
-      "grad_norm": 0.0008392962045036256,
-      "learning_rate": 0.00019999998372833674,
-      "loss": 46.0,
-      "step": 2385
-    },
-    {
-      "epoch": 0.18242636236787277,
-      "grad_norm": 0.0014696762664243579,
-      "learning_rate": 0.00019999998371463142,
-      "loss": 46.0,
-      "step": 2386
-    },
-    {
-      "epoch": 0.1825028193512625,
-      "grad_norm": 0.001209444017149508,
-      "learning_rate": 0.00019999998370092033,
-      "loss": 46.0,
-      "step": 2387
-    },
-    {
-      "epoch": 0.1825792763346522,
-      "grad_norm": 0.0018046700861304998,
-      "learning_rate": 0.00019999998368720344,
-      "loss": 46.0,
-      "step": 2388
-    },
-    {
-      "epoch": 0.18265573331804194,
-      "grad_norm": 0.0009530246024951339,
-      "learning_rate": 0.00019999998367348077,
-      "loss": 46.0,
-      "step": 2389
-    },
-    {
-      "epoch": 0.18273219030143165,
-      "grad_norm": 0.0012285119155421853,
-      "learning_rate": 0.00019999998365975236,
-      "loss": 46.0,
-      "step": 2390
-    },
-    {
-      "epoch": 0.18280864728482138,
-      "grad_norm": 0.0006943094194866717,
-      "learning_rate": 0.0001999999836460182,
-      "loss": 46.0,
-      "step": 2391
-    },
-    {
-      "epoch": 0.18288510426821109,
-      "grad_norm": 0.000586421403568238,
-      "learning_rate": 0.00019999998363227824,
-      "loss": 46.0,
-      "step": 2392
-    },
-    {
-      "epoch": 0.18296156125160082,
-      "grad_norm": 0.0037157570477575064,
-      "learning_rate": 0.00019999998361853248,
-      "loss": 46.0,
-      "step": 2393
-    },
-    {
-      "epoch": 0.18303801823499055,
-      "grad_norm": 0.0012294065672904253,
-      "learning_rate": 0.000199999983604781,
-      "loss": 46.0,
-      "step": 2394
-    },
-    {
-      "epoch": 0.18311447521838026,
-      "grad_norm": 0.0027290915604680777,
-      "learning_rate": 0.00019999998359102375,
-      "loss": 46.0,
-      "step": 2395
-    },
-    {
-      "epoch": 0.18319093220177,
-      "grad_norm": 0.0008222172618843615,
-      "learning_rate": 0.0001999999835772607,
-      "loss": 46.0,
-      "step": 2396
-    },
-    {
-      "epoch": 0.1832673891851597,
-      "grad_norm": 0.0007116239285096526,
-      "learning_rate": 0.00019999998356349193,
-      "loss": 46.0,
-      "step": 2397
-    },
-    {
-      "epoch": 0.18334384616854943,
-      "grad_norm": 0.001192706055007875,
-      "learning_rate": 0.00019999998354971736,
-      "loss": 46.0,
-      "step": 2398
-    },
-    {
-      "epoch": 0.18342030315193913,
-      "grad_norm": 0.0014561553252860904,
-      "learning_rate": 0.00019999998353593701,
-      "loss": 46.0,
-      "step": 2399
-    },
-    {
-      "epoch": 0.18349676013532887,
-      "grad_norm": 0.0017072921618819237,
-      "learning_rate": 0.0001999999835221509,
-      "loss": 46.0,
-      "step": 2400
-    },
-    {
-      "epoch": 0.18357321711871857,
-      "grad_norm": 0.002692918060347438,
-      "learning_rate": 0.00019999998350835903,
-      "loss": 46.0,
-      "step": 2401
-    },
-    {
-      "epoch": 0.1836496741021083,
-      "grad_norm": 0.002356676384806633,
-      "learning_rate": 0.00019999998349456137,
-      "loss": 46.0,
-      "step": 2402
-    },
-    {
-      "epoch": 0.183726131085498,
-      "grad_norm": 0.0011876497883349657,
-      "learning_rate": 0.00019999998348075795,
-      "loss": 46.0,
-      "step": 2403
-    },
-    {
-      "epoch": 0.18380258806888775,
-      "grad_norm": 0.0011347861727699637,
-      "learning_rate": 0.00019999998346694877,
-      "loss": 46.0,
-      "step": 2404
-    },
-    {
-      "epoch": 0.18387904505227745,
-      "grad_norm": 0.0009794689249247313,
-      "learning_rate": 0.0001999999834531338,
-      "loss": 46.0,
-      "step": 2405
-    },
-    {
-      "epoch": 0.18395550203566718,
-      "grad_norm": 0.0011760820634663105,
-      "learning_rate": 0.00019999998343931308,
-      "loss": 46.0,
-      "step": 2406
-    },
-    {
-      "epoch": 0.1840319590190569,
-      "grad_norm": 0.00043110037222504616,
-      "learning_rate": 0.00019999998342548657,
-      "loss": 46.0,
-      "step": 2407
-    },
-    {
-      "epoch": 0.18410841600244662,
-      "grad_norm": 0.0011725655058398843,
-      "learning_rate": 0.00019999998341165432,
-      "loss": 46.0,
-      "step": 2408
-    },
-    {
-      "epoch": 0.18418487298583636,
-      "grad_norm": 0.0022977543994784355,
-      "learning_rate": 0.0001999999833978163,
-      "loss": 46.0,
-      "step": 2409
-    },
-    {
-      "epoch": 0.18426132996922606,
-      "grad_norm": 0.001277922885492444,
-      "learning_rate": 0.00019999998338397247,
-      "loss": 46.0,
-      "step": 2410
-    },
-    {
-      "epoch": 0.1843377869526158,
-      "grad_norm": 0.0029628209304064512,
-      "learning_rate": 0.0001999999833701229,
-      "loss": 46.0,
-      "step": 2411
-    },
-    {
-      "epoch": 0.1844142439360055,
-      "grad_norm": 0.00076859857654199,
-      "learning_rate": 0.00019999998335626758,
-      "loss": 46.0,
-      "step": 2412
-    },
-    {
-      "epoch": 0.18449070091939523,
-      "grad_norm": 0.0006292408215813339,
-      "learning_rate": 0.00019999998334240644,
-      "loss": 46.0,
-      "step": 2413
-    },
-    {
-      "epoch": 0.18456715790278494,
-      "grad_norm": 0.0025320679415017366,
-      "learning_rate": 0.00019999998332853957,
-      "loss": 46.0,
-      "step": 2414
-    },
-    {
-      "epoch": 0.18464361488617467,
-      "grad_norm": 0.001478655613027513,
-      "learning_rate": 0.0001999999833146669,
-      "loss": 46.0,
-      "step": 2415
-    },
-    {
-      "epoch": 0.18472007186956438,
-      "grad_norm": 0.0017672876128926873,
-      "learning_rate": 0.0001999999833007885,
-      "loss": 46.0,
-      "step": 2416
-    },
-    {
-      "epoch": 0.1847965288529541,
-      "grad_norm": 0.0011046217987313867,
-      "learning_rate": 0.0001999999832869043,
-      "loss": 46.0,
-      "step": 2417
-    },
-    {
-      "epoch": 0.18487298583634382,
-      "grad_norm": 0.0016992365708574653,
-      "learning_rate": 0.00019999998327301435,
-      "loss": 46.0,
-      "step": 2418
-    },
-    {
-      "epoch": 0.18494944281973355,
-      "grad_norm": 0.001854369300417602,
-      "learning_rate": 0.00019999998325911862,
-      "loss": 46.0,
-      "step": 2419
-    },
-    {
-      "epoch": 0.18502589980312326,
-      "grad_norm": 0.0010395062854513526,
-      "learning_rate": 0.00019999998324521712,
-      "loss": 46.0,
-      "step": 2420
-    },
-    {
-      "epoch": 0.185102356786513,
-      "grad_norm": 0.0019571902230381966,
-      "learning_rate": 0.00019999998323130984,
-      "loss": 46.0,
-      "step": 2421
-    },
-    {
-      "epoch": 0.1851788137699027,
-      "grad_norm": 0.0010250064078718424,
-      "learning_rate": 0.00019999998321739682,
-      "loss": 46.0,
-      "step": 2422
-    },
-    {
-      "epoch": 0.18525527075329243,
-      "grad_norm": 0.00210104463621974,
-      "learning_rate": 0.00019999998320347802,
-      "loss": 46.0,
-      "step": 2423
-    },
-    {
-      "epoch": 0.18533172773668216,
-      "grad_norm": 0.0009446580079384148,
-      "learning_rate": 0.00019999998318955342,
-      "loss": 46.0,
-      "step": 2424
-    },
-    {
-      "epoch": 0.18540818472007187,
-      "grad_norm": 0.0008213113178499043,
-      "learning_rate": 0.00019999998317562308,
-      "loss": 46.0,
-      "step": 2425
-    },
-    {
-      "epoch": 0.1854846417034616,
-      "grad_norm": 0.0018267889972776175,
-      "learning_rate": 0.00019999998316168696,
-      "loss": 46.0,
-      "step": 2426
-    },
-    {
-      "epoch": 0.1855610986868513,
-      "grad_norm": 0.0005351110594347119,
-      "learning_rate": 0.0001999999831477451,
-      "loss": 46.0,
-      "step": 2427
-    },
-    {
-      "epoch": 0.18563755567024104,
-      "grad_norm": 0.0006839693523943424,
-      "learning_rate": 0.00019999998313379743,
-      "loss": 46.0,
-      "step": 2428
-    },
-    {
-      "epoch": 0.18571401265363074,
-      "grad_norm": 0.0007742916932329535,
-      "learning_rate": 0.000199999983119844,
-      "loss": 46.0,
-      "step": 2429
-    },
-    {
-      "epoch": 0.18579046963702048,
-      "grad_norm": 0.0012417191173881292,
-      "learning_rate": 0.00019999998310588478,
-      "loss": 46.0,
-      "step": 2430
-    },
-    {
-      "epoch": 0.18586692662041018,
-      "grad_norm": 0.0009488030918873847,
-      "learning_rate": 0.00019999998309191982,
-      "loss": 46.0,
-      "step": 2431
-    },
-    {
-      "epoch": 0.18594338360379992,
-      "grad_norm": 0.0016189453890547156,
-      "learning_rate": 0.0001999999830779491,
-      "loss": 46.0,
-      "step": 2432
-    },
-    {
-      "epoch": 0.18601984058718962,
-      "grad_norm": 0.0009394802036695182,
-      "learning_rate": 0.0001999999830639726,
-      "loss": 46.0,
-      "step": 2433
-    },
-    {
-      "epoch": 0.18609629757057936,
-      "grad_norm": 0.0008046234142966568,
-      "learning_rate": 0.00019999998304999034,
-      "loss": 46.0,
-      "step": 2434
-    },
-    {
-      "epoch": 0.18617275455396906,
-      "grad_norm": 0.0007821296458132565,
-      "learning_rate": 0.00019999998303600228,
-      "loss": 46.0,
-      "step": 2435
-    },
-    {
-      "epoch": 0.1862492115373588,
-      "grad_norm": 0.0015971878310665488,
-      "learning_rate": 0.0001999999830220085,
-      "loss": 46.0,
-      "step": 2436
-    },
-    {
-      "epoch": 0.18632566852074853,
-      "grad_norm": 0.0010250088525936007,
-      "learning_rate": 0.00019999998300800892,
-      "loss": 46.0,
-      "step": 2437
-    },
-    {
-      "epoch": 0.18640212550413823,
-      "grad_norm": 0.0006358137470670044,
-      "learning_rate": 0.00019999998299400355,
-      "loss": 46.0,
-      "step": 2438
-    },
-    {
-      "epoch": 0.18647858248752797,
-      "grad_norm": 0.0006607634713873267,
-      "learning_rate": 0.00019999998297999243,
-      "loss": 46.0,
-      "step": 2439
-    },
-    {
-      "epoch": 0.18655503947091767,
-      "grad_norm": 0.0020706020295619965,
-      "learning_rate": 0.00019999998296597556,
-      "loss": 46.0,
-      "step": 2440
-    },
-    {
-      "epoch": 0.1866314964543074,
-      "grad_norm": 0.0005927235470153391,
-      "learning_rate": 0.0001999999829519529,
-      "loss": 46.0,
-      "step": 2441
-    },
-    {
-      "epoch": 0.1867079534376971,
-      "grad_norm": 0.0008116699755191803,
-      "learning_rate": 0.0001999999829379245,
-      "loss": 46.0,
-      "step": 2442
-    },
-    {
-      "epoch": 0.18678441042108684,
-      "grad_norm": 0.0009640644420869648,
-      "learning_rate": 0.00019999998292389028,
-      "loss": 46.0,
-      "step": 2443
-    },
-    {
-      "epoch": 0.18686086740447655,
-      "grad_norm": 0.0011444780975580215,
-      "learning_rate": 0.00019999998290985032,
-      "loss": 46.0,
-      "step": 2444
-    },
-    {
-      "epoch": 0.18693732438786628,
-      "grad_norm": 0.0011077811941504478,
-      "learning_rate": 0.00019999998289580456,
-      "loss": 46.0,
-      "step": 2445
-    },
-    {
-      "epoch": 0.187013781371256,
-      "grad_norm": 0.0008246845682151616,
-      "learning_rate": 0.00019999998288175306,
-      "loss": 46.0,
-      "step": 2446
-    },
-    {
-      "epoch": 0.18709023835464572,
-      "grad_norm": 0.0013810722157359123,
-      "learning_rate": 0.00019999998286769578,
-      "loss": 46.0,
-      "step": 2447
-    },
-    {
-      "epoch": 0.18716669533803543,
-      "grad_norm": 0.002938495483249426,
-      "learning_rate": 0.00019999998285363276,
-      "loss": 46.0,
-      "step": 2448
-    },
-    {
-      "epoch": 0.18724315232142516,
-      "grad_norm": 0.0007734111859463155,
-      "learning_rate": 0.0001999999828395639,
-      "loss": 46.0,
-      "step": 2449
-    },
-    {
-      "epoch": 0.18731960930481487,
-      "grad_norm": 0.00296254875138402,
-      "learning_rate": 0.00019999998282548936,
-      "loss": 46.0,
-      "step": 2450
-    },
-    {
-      "epoch": 0.1873960662882046,
-      "grad_norm": 0.001780922175385058,
-      "learning_rate": 0.000199999982811409,
-      "loss": 46.0,
-      "step": 2451
-    },
-    {
-      "epoch": 0.18747252327159433,
-      "grad_norm": 0.0027095621917396784,
-      "learning_rate": 0.00019999998279732285,
-      "loss": 46.0,
-      "step": 2452
-    },
-    {
-      "epoch": 0.18754898025498404,
-      "grad_norm": 0.0012391556520015001,
-      "learning_rate": 0.00019999998278323098,
-      "loss": 46.0,
-      "step": 2453
-    },
-    {
-      "epoch": 0.18762543723837377,
-      "grad_norm": 0.0022775058168917894,
-      "learning_rate": 0.00019999998276913332,
-      "loss": 46.0,
-      "step": 2454
-    },
-    {
-      "epoch": 0.18770189422176348,
-      "grad_norm": 0.0006520353490486741,
-      "learning_rate": 0.00019999998275502988,
-      "loss": 46.0,
-      "step": 2455
-    },
-    {
-      "epoch": 0.1877783512051532,
-      "grad_norm": 0.001414004829712212,
-      "learning_rate": 0.0001999999827409207,
-      "loss": 46.0,
-      "step": 2456
-    },
-    {
-      "epoch": 0.18785480818854292,
-      "grad_norm": 0.0009541686740703881,
-      "learning_rate": 0.00019999998272680571,
-      "loss": 46.0,
-      "step": 2457
-    },
-    {
-      "epoch": 0.18793126517193265,
-      "grad_norm": 0.0011238297447562218,
-      "learning_rate": 0.00019999998271268498,
-      "loss": 46.0,
-      "step": 2458
-    },
-    {
-      "epoch": 0.18800772215532235,
-      "grad_norm": 0.0057695223949849606,
-      "learning_rate": 0.00019999998269855848,
-      "loss": 46.0,
-      "step": 2459
-    },
-    {
-      "epoch": 0.1880841791387121,
-      "grad_norm": 0.0013992065796628594,
-      "learning_rate": 0.0001999999826844262,
-      "loss": 46.0,
-      "step": 2460
-    },
-    {
-      "epoch": 0.1881606361221018,
-      "grad_norm": 0.0009604279766790569,
-      "learning_rate": 0.00019999998267028818,
-      "loss": 46.0,
-      "step": 2461
-    },
-    {
-      "epoch": 0.18823709310549153,
-      "grad_norm": 0.0005514805670827627,
-      "learning_rate": 0.00019999998265614433,
-      "loss": 46.0,
-      "step": 2462
-    },
-    {
-      "epoch": 0.18831355008888123,
-      "grad_norm": 0.0015055537223815918,
-      "learning_rate": 0.00019999998264199476,
-      "loss": 46.0,
-      "step": 2463
-    },
-    {
-      "epoch": 0.18839000707227097,
-      "grad_norm": 0.0018673110753297806,
-      "learning_rate": 0.00019999998262783941,
-      "loss": 46.0,
-      "step": 2464
-    },
-    {
-      "epoch": 0.18846646405566067,
-      "grad_norm": 0.005999795626848936,
-      "learning_rate": 0.00019999998261367827,
-      "loss": 46.0,
-      "step": 2465
-    },
-    {
-      "epoch": 0.1885429210390504,
-      "grad_norm": 0.0008092976058833301,
-      "learning_rate": 0.00019999998259951138,
-      "loss": 46.0,
-      "step": 2466
-    },
-    {
-      "epoch": 0.18861937802244014,
-      "grad_norm": 0.0011135678505524993,
-      "learning_rate": 0.00019999998258533872,
-      "loss": 46.0,
-      "step": 2467
-    },
-    {
-      "epoch": 0.18869583500582984,
-      "grad_norm": 0.0010720588034018874,
-      "learning_rate": 0.00019999998257116028,
-      "loss": 46.0,
-      "step": 2468
-    },
-    {
-      "epoch": 0.18877229198921958,
-      "grad_norm": 0.0009085648343898356,
-      "learning_rate": 0.0001999999825569761,
-      "loss": 46.0,
-      "step": 2469
-    },
-    {
-      "epoch": 0.18884874897260928,
-      "grad_norm": 0.000594708020798862,
-      "learning_rate": 0.0001999999825427861,
-      "loss": 46.0,
-      "step": 2470
-    },
-    {
-      "epoch": 0.18892520595599901,
-      "grad_norm": 0.0012979174498468637,
-      "learning_rate": 0.00019999998252859036,
-      "loss": 46.0,
-      "step": 2471
-    },
-    {
-      "epoch": 0.18900166293938872,
-      "grad_norm": 0.0007752353558316827,
-      "learning_rate": 0.00019999998251438888,
-      "loss": 46.0,
-      "step": 2472
-    },
-    {
-      "epoch": 0.18907811992277845,
-      "grad_norm": 0.0005088779143989086,
-      "learning_rate": 0.00019999998250018157,
-      "loss": 46.0,
-      "step": 2473
-    },
-    {
-      "epoch": 0.18915457690616816,
-      "grad_norm": 0.001224258914589882,
-      "learning_rate": 0.00019999998248596855,
-      "loss": 46.0,
-      "step": 2474
-    },
-    {
-      "epoch": 0.1892310338895579,
-      "grad_norm": 0.0009395096567459404,
-      "learning_rate": 0.0001999999824717497,
-      "loss": 46.0,
-      "step": 2475
-    },
-    {
-      "epoch": 0.1893074908729476,
-      "grad_norm": 0.0018619762267917395,
-      "learning_rate": 0.0001999999824575251,
-      "loss": 46.0,
-      "step": 2476
-    },
-    {
-      "epoch": 0.18938394785633733,
-      "grad_norm": 0.0009954768465831876,
-      "learning_rate": 0.00019999998244329476,
-      "loss": 46.0,
-      "step": 2477
-    },
-    {
-      "epoch": 0.18946040483972704,
-      "grad_norm": 0.002437873976305127,
-      "learning_rate": 0.00019999998242905864,
-      "loss": 46.0,
-      "step": 2478
-    },
-    {
-      "epoch": 0.18953686182311677,
-      "grad_norm": 0.00048663877532817423,
-      "learning_rate": 0.00019999998241481675,
-      "loss": 46.0,
-      "step": 2479
-    },
-    {
-      "epoch": 0.18961331880650648,
-      "grad_norm": 0.0005779552739113569,
-      "learning_rate": 0.00019999998240056906,
-      "loss": 46.0,
-      "step": 2480
-    },
-    {
-      "epoch": 0.1896897757898962,
-      "grad_norm": 0.0005875512724742293,
-      "learning_rate": 0.00019999998238631563,
-      "loss": 46.0,
-      "step": 2481
-    },
-    {
-      "epoch": 0.18976623277328594,
-      "grad_norm": 0.0017303097993135452,
-      "learning_rate": 0.00019999998237205644,
-      "loss": 46.0,
-      "step": 2482
-    },
-    {
-      "epoch": 0.18984268975667565,
-      "grad_norm": 0.0005524460575543344,
-      "learning_rate": 0.00019999998235779146,
-      "loss": 46.0,
-      "step": 2483
-    },
-    {
-      "epoch": 0.18991914674006538,
-      "grad_norm": 0.001414101687259972,
-      "learning_rate": 0.0001999999823435207,
-      "loss": 46.0,
-      "step": 2484
-    },
-    {
-      "epoch": 0.1899956037234551,
-      "grad_norm": 0.0005345565732568502,
-      "learning_rate": 0.0001999999823292442,
-      "loss": 46.0,
-      "step": 2485
-    },
-    {
-      "epoch": 0.19007206070684482,
-      "grad_norm": 0.0027988983783870935,
-      "learning_rate": 0.0001999999823149619,
-      "loss": 46.0,
-      "step": 2486
-    },
-    {
-      "epoch": 0.19014851769023453,
-      "grad_norm": 0.0008862941176630557,
-      "learning_rate": 0.00019999998230067387,
-      "loss": 46.0,
-      "step": 2487
-    },
-    {
-      "epoch": 0.19022497467362426,
-      "grad_norm": 0.0006016691331751645,
-      "learning_rate": 0.00019999998228638005,
-      "loss": 46.0,
-      "step": 2488
-    },
-    {
-      "epoch": 0.19030143165701396,
-      "grad_norm": 0.0033350768499076366,
-      "learning_rate": 0.00019999998227208045,
-      "loss": 46.0,
-      "step": 2489
-    },
-    {
-      "epoch": 0.1903778886404037,
-      "grad_norm": 0.0009862019214779139,
-      "learning_rate": 0.00019999998225777508,
-      "loss": 46.0,
-      "step": 2490
-    },
-    {
-      "epoch": 0.1904543456237934,
-      "grad_norm": 0.0016853954875841737,
-      "learning_rate": 0.00019999998224346394,
-      "loss": 46.0,
-      "step": 2491
-    },
-    {
-      "epoch": 0.19053080260718314,
-      "grad_norm": 0.0005088983452878892,
-      "learning_rate": 0.00019999998222914708,
-      "loss": 46.0,
-      "step": 2492
-    },
-    {
-      "epoch": 0.19060725959057284,
-      "grad_norm": 0.0015099769225344062,
-      "learning_rate": 0.0001999999822148244,
-      "loss": 46.0,
-      "step": 2493
-    },
-    {
-      "epoch": 0.19068371657396258,
-      "grad_norm": 0.012772087007761002,
-      "learning_rate": 0.00019999998220049595,
-      "loss": 46.0,
-      "step": 2494
-    },
-    {
-      "epoch": 0.1907601735573523,
-      "grad_norm": 0.0011454602936282754,
-      "learning_rate": 0.00019999998218616174,
-      "loss": 46.0,
-      "step": 2495
-    },
-    {
-      "epoch": 0.190836630540742,
-      "grad_norm": 0.0007046663085930049,
-      "learning_rate": 0.00019999998217182176,
-      "loss": 46.0,
-      "step": 2496
-    },
-    {
-      "epoch": 0.19091308752413175,
-      "grad_norm": 0.0014262953773140907,
-      "learning_rate": 0.00019999998215747603,
-      "loss": 46.0,
-      "step": 2497
-    },
-    {
-      "epoch": 0.19098954450752145,
-      "grad_norm": 0.001202348037622869,
-      "learning_rate": 0.00019999998214312453,
-      "loss": 46.0,
-      "step": 2498
-    },
-    {
-      "epoch": 0.19106600149091119,
-      "grad_norm": 0.0009572213748469949,
-      "learning_rate": 0.0001999999821287672,
-      "loss": 46.0,
-      "step": 2499
-    },
-    {
-      "epoch": 0.1911424584743009,
-      "grad_norm": 0.001123833004385233,
-      "learning_rate": 0.00019999998211440417,
-      "loss": 46.0,
-      "step": 2500
-    },
-    {
-      "epoch": 0.19121891545769062,
-      "grad_norm": 0.0004405451472848654,
-      "learning_rate": 0.00019999998210003535,
-      "loss": 46.0,
-      "step": 2501
-    },
-    {
-      "epoch": 0.19129537244108033,
-      "grad_norm": 0.0008608513162471354,
-      "learning_rate": 0.00019999998208566075,
-      "loss": 46.0,
-      "step": 2502
-    },
-    {
-      "epoch": 0.19137182942447006,
-      "grad_norm": 0.0019066163804382086,
-      "learning_rate": 0.00019999998207128038,
-      "loss": 46.0,
-      "step": 2503
-    },
-    {
-      "epoch": 0.19144828640785977,
-      "grad_norm": 0.0017126858001574874,
-      "learning_rate": 0.00019999998205689424,
-      "loss": 46.0,
-      "step": 2504
-    },
-    {
-      "epoch": 0.1915247433912495,
-      "grad_norm": 0.0008729486726224422,
-      "learning_rate": 0.00019999998204250235,
-      "loss": 46.0,
-      "step": 2505
-    },
-    {
-      "epoch": 0.1916012003746392,
-      "grad_norm": 0.0019524359377101064,
-      "learning_rate": 0.00019999998202810466,
-      "loss": 46.0,
-      "step": 2506
-    },
-    {
-      "epoch": 0.19167765735802894,
-      "grad_norm": 0.0014123177388682961,
-      "learning_rate": 0.00019999998201370125,
-      "loss": 46.0,
-      "step": 2507
-    },
-    {
-      "epoch": 0.19175411434141865,
-      "grad_norm": 0.0007191233453340828,
-      "learning_rate": 0.00019999998199929202,
-      "loss": 46.0,
-      "step": 2508
-    },
-    {
-      "epoch": 0.19183057132480838,
-      "grad_norm": 0.000843721441924572,
-      "learning_rate": 0.00019999998198487703,
-      "loss": 46.0,
-      "step": 2509
-    },
-    {
-      "epoch": 0.1919070283081981,
-      "grad_norm": 0.0014823595993220806,
-      "learning_rate": 0.00019999998197045628,
-      "loss": 46.0,
-      "step": 2510
-    },
-    {
-      "epoch": 0.19198348529158782,
-      "grad_norm": 0.0014458306832239032,
-      "learning_rate": 0.00019999998195602977,
-      "loss": 46.0,
-      "step": 2511
-    },
-    {
-      "epoch": 0.19205994227497755,
-      "grad_norm": 0.001128409756347537,
-      "learning_rate": 0.00019999998194159747,
-      "loss": 46.0,
-      "step": 2512
-    },
-    {
-      "epoch": 0.19213639925836726,
-      "grad_norm": 0.0012496395502239466,
-      "learning_rate": 0.00019999998192715942,
-      "loss": 46.0,
-      "step": 2513
-    },
-    {
-      "epoch": 0.192212856241757,
-      "grad_norm": 0.0013120636576786637,
-      "learning_rate": 0.00019999998191271557,
-      "loss": 46.0,
-      "step": 2514
-    },
-    {
-      "epoch": 0.1922893132251467,
-      "grad_norm": 0.0006523728370666504,
-      "learning_rate": 0.00019999998189826598,
-      "loss": 46.0,
-      "step": 2515
-    },
-    {
-      "epoch": 0.19236577020853643,
-      "grad_norm": 0.0007474781014025211,
-      "learning_rate": 0.00019999998188381063,
-      "loss": 46.0,
-      "step": 2516
-    },
-    {
-      "epoch": 0.19244222719192614,
-      "grad_norm": 0.0016185181448236108,
-      "learning_rate": 0.0001999999818693495,
-      "loss": 46.0,
-      "step": 2517
-    },
-    {
-      "epoch": 0.19251868417531587,
-      "grad_norm": 0.002002629218623042,
-      "learning_rate": 0.0001999999818548826,
-      "loss": 46.0,
-      "step": 2518
-    },
-    {
-      "epoch": 0.19259514115870557,
-      "grad_norm": 0.0005933846114203334,
-      "learning_rate": 0.0001999999818404099,
-      "loss": 46.0,
-      "step": 2519
-    },
-    {
-      "epoch": 0.1926715981420953,
-      "grad_norm": 0.0009361347765661776,
-      "learning_rate": 0.00019999998182593148,
-      "loss": 46.0,
-      "step": 2520
-    },
-    {
-      "epoch": 0.192748055125485,
-      "grad_norm": 0.0010292287915945053,
-      "learning_rate": 0.00019999998181144727,
-      "loss": 46.0,
-      "step": 2521
-    },
-    {
-      "epoch": 0.19282451210887475,
-      "grad_norm": 0.0007696306565776467,
-      "learning_rate": 0.00019999998179695729,
-      "loss": 46.0,
-      "step": 2522
-    },
-    {
-      "epoch": 0.19290096909226445,
-      "grad_norm": 0.0012693873140960932,
-      "learning_rate": 0.0001999999817824615,
-      "loss": 46.0,
-      "step": 2523
-    },
-    {
-      "epoch": 0.19297742607565418,
-      "grad_norm": 0.001217762823216617,
-      "learning_rate": 0.00019999998176796,
-      "loss": 46.0,
-      "step": 2524
-    },
-    {
-      "epoch": 0.19305388305904392,
-      "grad_norm": 0.001914670574478805,
-      "learning_rate": 0.00019999998175345272,
-      "loss": 46.0,
-      "step": 2525
-    },
-    {
-      "epoch": 0.19313034004243362,
-      "grad_norm": 0.0015102314064279199,
-      "learning_rate": 0.00019999998173893965,
-      "loss": 46.0,
-      "step": 2526
-    },
-    {
-      "epoch": 0.19320679702582336,
-      "grad_norm": 0.001179531100206077,
-      "learning_rate": 0.00019999998172442083,
-      "loss": 46.0,
-      "step": 2527
-    },
-    {
-      "epoch": 0.19328325400921306,
-      "grad_norm": 0.001162834814749658,
-      "learning_rate": 0.00019999998170989623,
-      "loss": 46.0,
-      "step": 2528
-    },
-    {
-      "epoch": 0.1933597109926028,
-      "grad_norm": 0.0016187006840482354,
-      "learning_rate": 0.00019999998169536583,
-      "loss": 46.0,
-      "step": 2529
-    },
-    {
-      "epoch": 0.1934361679759925,
-      "grad_norm": 0.00241902656853199,
-      "learning_rate": 0.0001999999816808297,
-      "loss": 46.0,
-      "step": 2530
-    },
-    {
-      "epoch": 0.19351262495938223,
-      "grad_norm": 0.0016595746856182814,
-      "learning_rate": 0.00019999998166628783,
-      "loss": 46.0,
-      "step": 2531
-    },
-    {
-      "epoch": 0.19358908194277194,
-      "grad_norm": 0.0004893513396382332,
-      "learning_rate": 0.00019999998165174011,
-      "loss": 46.0,
-      "step": 2532
-    },
-    {
-      "epoch": 0.19366553892616167,
-      "grad_norm": 0.001225075451657176,
-      "learning_rate": 0.00019999998163718668,
-      "loss": 46.0,
-      "step": 2533
-    },
-    {
-      "epoch": 0.19374199590955138,
-      "grad_norm": 0.0011962822172790766,
-      "learning_rate": 0.00019999998162262744,
-      "loss": 46.0,
-      "step": 2534
-    },
-    {
-      "epoch": 0.1938184528929411,
-      "grad_norm": 0.0006792786298319697,
-      "learning_rate": 0.0001999999816080625,
-      "loss": 46.0,
-      "step": 2535
-    },
-    {
-      "epoch": 0.19389490987633082,
-      "grad_norm": 0.002364215673878789,
-      "learning_rate": 0.0001999999815934917,
-      "loss": 46.0,
-      "step": 2536
-    },
-    {
-      "epoch": 0.19397136685972055,
-      "grad_norm": 0.0007861287449486554,
-      "learning_rate": 0.0001999999815789152,
-      "loss": 46.0,
-      "step": 2537
-    },
-    {
-      "epoch": 0.19404782384311026,
-      "grad_norm": 0.0011116234818473458,
-      "learning_rate": 0.00019999998156433287,
-      "loss": 46.0,
-      "step": 2538
-    },
-    {
-      "epoch": 0.1941242808265,
-      "grad_norm": 0.0026597920805215836,
-      "learning_rate": 0.00019999998154974482,
-      "loss": 46.0,
-      "step": 2539
-    },
-    {
-      "epoch": 0.19420073780988972,
-      "grad_norm": 0.0011285573709756136,
-      "learning_rate": 0.000199999981535151,
-      "loss": 46.0,
-      "step": 2540
-    },
-    {
-      "epoch": 0.19427719479327943,
-      "grad_norm": 0.0006225054967217147,
-      "learning_rate": 0.0001999999815205514,
-      "loss": 46.0,
-      "step": 2541
-    },
-    {
-      "epoch": 0.19435365177666916,
-      "grad_norm": 0.0020340282935649157,
-      "learning_rate": 0.00019999998150594604,
-      "loss": 46.0,
-      "step": 2542
-    },
-    {
-      "epoch": 0.19443010876005887,
-      "grad_norm": 0.0014335649320855737,
-      "learning_rate": 0.00019999998149133487,
-      "loss": 46.0,
-      "step": 2543
-    },
-    {
-      "epoch": 0.1945065657434486,
-      "grad_norm": 0.004374792333692312,
-      "learning_rate": 0.00019999998147671798,
-      "loss": 46.0,
-      "step": 2544
-    },
-    {
-      "epoch": 0.1945830227268383,
-      "grad_norm": 0.001119488850235939,
-      "learning_rate": 0.0001999999814620953,
-      "loss": 46.0,
-      "step": 2545
-    },
-    {
-      "epoch": 0.19465947971022804,
-      "grad_norm": 0.0023181482683867216,
-      "learning_rate": 0.00019999998144746686,
-      "loss": 46.0,
-      "step": 2546
-    },
-    {
-      "epoch": 0.19473593669361774,
-      "grad_norm": 0.001117787091061473,
-      "learning_rate": 0.0001999999814328326,
-      "loss": 46.0,
-      "step": 2547
-    },
-    {
-      "epoch": 0.19481239367700748,
-      "grad_norm": 0.0004660346021410078,
-      "learning_rate": 0.00019999998141819264,
-      "loss": 46.0,
-      "step": 2548
-    },
-    {
-      "epoch": 0.19488885066039718,
-      "grad_norm": 0.0014202577294781804,
-      "learning_rate": 0.00019999998140354688,
-      "loss": 46.0,
-      "step": 2549
-    },
-    {
-      "epoch": 0.19496530764378692,
-      "grad_norm": 0.0025082866195589304,
-      "learning_rate": 0.00019999998138889535,
-      "loss": 46.0,
-      "step": 2550
-    },
-    {
-      "epoch": 0.19504176462717662,
-      "grad_norm": 0.0015981708420440555,
-      "learning_rate": 0.00019999998137423805,
-      "loss": 46.0,
-      "step": 2551
-    },
-    {
-      "epoch": 0.19511822161056636,
-      "grad_norm": 0.000774684885982424,
-      "learning_rate": 0.00019999998135957498,
-      "loss": 46.0,
-      "step": 2552
-    },
-    {
-      "epoch": 0.1951946785939561,
-      "grad_norm": 0.001830187626183033,
-      "learning_rate": 0.00019999998134490613,
-      "loss": 46.0,
-      "step": 2553
-    },
-    {
-      "epoch": 0.1952711355773458,
-      "grad_norm": 0.0020728816743940115,
-      "learning_rate": 0.00019999998133023153,
-      "loss": 46.0,
-      "step": 2554
-    },
-    {
-      "epoch": 0.19534759256073553,
-      "grad_norm": 0.0007425218354910612,
-      "learning_rate": 0.00019999998131555116,
-      "loss": 46.0,
-      "step": 2555
-    },
-    {
-      "epoch": 0.19542404954412523,
-      "grad_norm": 0.0014173837844282389,
-      "learning_rate": 0.00019999998130086502,
-      "loss": 46.0,
-      "step": 2556
-    },
-    {
-      "epoch": 0.19550050652751497,
-      "grad_norm": 0.0005700261681340635,
-      "learning_rate": 0.00019999998128617313,
-      "loss": 46.0,
-      "step": 2557
-    },
-    {
-      "epoch": 0.19557696351090467,
-      "grad_norm": 0.005316985305398703,
-      "learning_rate": 0.00019999998127147542,
-      "loss": 46.0,
-      "step": 2558
-    },
-    {
-      "epoch": 0.1956534204942944,
-      "grad_norm": 0.001112484373152256,
-      "learning_rate": 0.00019999998125677198,
-      "loss": 46.0,
-      "step": 2559
-    },
-    {
-      "epoch": 0.1957298774776841,
-      "grad_norm": 0.001547213876619935,
-      "learning_rate": 0.00019999998124206275,
-      "loss": 46.0,
-      "step": 2560
-    },
-    {
-      "epoch": 0.19580633446107384,
-      "grad_norm": 0.0007029794505797327,
-      "learning_rate": 0.00019999998122734777,
-      "loss": 46.0,
-      "step": 2561
-    },
-    {
-      "epoch": 0.19588279144446355,
-      "grad_norm": 0.0005506088491529226,
-      "learning_rate": 0.00019999998121262699,
-      "loss": 46.0,
-      "step": 2562
-    },
-    {
-      "epoch": 0.19595924842785328,
-      "grad_norm": 0.0009937337599694729,
-      "learning_rate": 0.00019999998119790046,
-      "loss": 46.0,
-      "step": 2563
-    },
-    {
-      "epoch": 0.196035705411243,
-      "grad_norm": 0.0013787487987428904,
-      "learning_rate": 0.00019999998118316815,
-      "loss": 46.0,
-      "step": 2564
-    },
-    {
-      "epoch": 0.19611216239463272,
-      "grad_norm": 0.003115082159638405,
-      "learning_rate": 0.0001999999811684301,
-      "loss": 46.0,
-      "step": 2565
-    },
-    {
-      "epoch": 0.19618861937802243,
-      "grad_norm": 0.0009442034061066806,
-      "learning_rate": 0.00019999998115368626,
-      "loss": 46.0,
-      "step": 2566
-    },
-    {
-      "epoch": 0.19626507636141216,
-      "grad_norm": 0.0014289517421275377,
-      "learning_rate": 0.00019999998113893664,
-      "loss": 46.0,
-      "step": 2567
-    },
-    {
-      "epoch": 0.1963415333448019,
-      "grad_norm": 0.0032986123114824295,
-      "learning_rate": 0.00019999998112418127,
-      "loss": 46.0,
-      "step": 2568
-    },
-    {
-      "epoch": 0.1964179903281916,
-      "grad_norm": 0.001238914323039353,
-      "learning_rate": 0.0001999999811094201,
-      "loss": 46.0,
-      "step": 2569
-    },
-    {
-      "epoch": 0.19649444731158133,
-      "grad_norm": 0.0007793986005708575,
-      "learning_rate": 0.0001999999810946532,
-      "loss": 46.0,
-      "step": 2570
-    },
-    {
-      "epoch": 0.19657090429497104,
-      "grad_norm": 0.0018220782512798905,
-      "learning_rate": 0.00019999998107988052,
-      "loss": 46.0,
-      "step": 2571
-    },
-    {
-      "epoch": 0.19664736127836077,
-      "grad_norm": 0.0010378597071394324,
-      "learning_rate": 0.00019999998106510206,
-      "loss": 46.0,
-      "step": 2572
-    },
-    {
-      "epoch": 0.19672381826175048,
-      "grad_norm": 0.0005865950952284038,
-      "learning_rate": 0.00019999998105031783,
-      "loss": 46.0,
-      "step": 2573
-    },
-    {
-      "epoch": 0.1968002752451402,
-      "grad_norm": 0.0006227686535567045,
-      "learning_rate": 0.00019999998103552785,
-      "loss": 46.0,
-      "step": 2574
-    },
-    {
-      "epoch": 0.19687673222852992,
-      "grad_norm": 0.0010060693603008986,
-      "learning_rate": 0.00019999998102073206,
-      "loss": 46.0,
-      "step": 2575
-    },
-    {
-      "epoch": 0.19695318921191965,
-      "grad_norm": 0.0018384178401902318,
-      "learning_rate": 0.00019999998100593054,
-      "loss": 46.0,
-      "step": 2576
-    },
-    {
-      "epoch": 0.19702964619530935,
-      "grad_norm": 0.0007536253542639315,
-      "learning_rate": 0.00019999998099112323,
-      "loss": 46.0,
-      "step": 2577
-    },
-    {
-      "epoch": 0.1971061031786991,
-      "grad_norm": 0.0010045995004475117,
-      "learning_rate": 0.00019999998097631016,
-      "loss": 46.0,
-      "step": 2578
-    },
-    {
-      "epoch": 0.1971825601620888,
-      "grad_norm": 0.002119707176461816,
-      "learning_rate": 0.00019999998096149134,
-      "loss": 46.0,
-      "step": 2579
-    },
-    {
-      "epoch": 0.19725901714547853,
-      "grad_norm": 0.0019034318393096328,
-      "learning_rate": 0.0001999999809466667,
-      "loss": 46.0,
-      "step": 2580
-    },
-    {
-      "epoch": 0.19733547412886823,
-      "grad_norm": 0.002396214520558715,
-      "learning_rate": 0.00019999998093183635,
-      "loss": 46.0,
-      "step": 2581
-    },
-    {
-      "epoch": 0.19741193111225797,
-      "grad_norm": 0.0009230375289916992,
-      "learning_rate": 0.00019999998091700018,
-      "loss": 46.0,
-      "step": 2582
-    },
-    {
-      "epoch": 0.1974883880956477,
-      "grad_norm": 0.0025573682505637407,
-      "learning_rate": 0.00019999998090215827,
-      "loss": 46.0,
-      "step": 2583
-    },
-    {
-      "epoch": 0.1975648450790374,
-      "grad_norm": 0.001235230010934174,
-      "learning_rate": 0.00019999998088731058,
-      "loss": 46.0,
-      "step": 2584
-    },
-    {
-      "epoch": 0.19764130206242714,
-      "grad_norm": 0.0012933886609971523,
-      "learning_rate": 0.00019999998087245712,
-      "loss": 46.0,
-      "step": 2585
-    },
-    {
-      "epoch": 0.19771775904581684,
-      "grad_norm": 0.0014868311118334532,
-      "learning_rate": 0.0001999999808575979,
-      "loss": 46.0,
-      "step": 2586
-    },
-    {
-      "epoch": 0.19779421602920658,
-      "grad_norm": 0.0010652708588168025,
-      "learning_rate": 0.0001999999808427329,
-      "loss": 46.0,
-      "step": 2587
-    },
-    {
-      "epoch": 0.19787067301259628,
-      "grad_norm": 0.0012538136215880513,
-      "learning_rate": 0.00019999998082786212,
-      "loss": 46.0,
-      "step": 2588
-    },
-    {
-      "epoch": 0.19794712999598602,
-      "grad_norm": 0.0007710539503023028,
-      "learning_rate": 0.0001999999808129856,
-      "loss": 46.0,
-      "step": 2589
-    },
-    {
-      "epoch": 0.19802358697937572,
-      "grad_norm": 0.0004720716387964785,
-      "learning_rate": 0.00019999998079810326,
-      "loss": 46.0,
-      "step": 2590
-    },
-    {
-      "epoch": 0.19810004396276545,
-      "grad_norm": 0.0012821248965337873,
-      "learning_rate": 0.0001999999807832152,
-      "loss": 46.0,
-      "step": 2591
-    },
-    {
-      "epoch": 0.19817650094615516,
-      "grad_norm": 0.0036623727064579725,
-      "learning_rate": 0.00019999998076832137,
-      "loss": 46.0,
-      "step": 2592
-    },
-    {
-      "epoch": 0.1982529579295449,
-      "grad_norm": 0.001504046143963933,
-      "learning_rate": 0.00019999998075342175,
-      "loss": 46.0,
-      "step": 2593
-    },
-    {
-      "epoch": 0.1983294149129346,
-      "grad_norm": 0.0006991566042415798,
-      "learning_rate": 0.00019999998073851635,
-      "loss": 46.0,
-      "step": 2594
-    },
-    {
-      "epoch": 0.19840587189632433,
-      "grad_norm": 0.0013940909411758184,
-      "learning_rate": 0.0001999999807236052,
-      "loss": 46.0,
-      "step": 2595
-    },
-    {
-      "epoch": 0.19848232887971404,
-      "grad_norm": 0.003455542493611574,
-      "learning_rate": 0.00019999998070868827,
-      "loss": 46.0,
-      "step": 2596
-    },
-    {
-      "epoch": 0.19855878586310377,
-      "grad_norm": 0.0006003906019032001,
-      "learning_rate": 0.00019999998069376558,
-      "loss": 46.0,
-      "step": 2597
-    },
-    {
-      "epoch": 0.1986352428464935,
-      "grad_norm": 0.002141443779692054,
-      "learning_rate": 0.00019999998067883712,
-      "loss": 46.0,
-      "step": 2598
-    },
-    {
-      "epoch": 0.1987116998298832,
-      "grad_norm": 0.004281296860426664,
-      "learning_rate": 0.0001999999806639029,
-      "loss": 46.0,
-      "step": 2599
-    },
-    {
-      "epoch": 0.19878815681327294,
-      "grad_norm": 0.0006730580935254693,
-      "learning_rate": 0.00019999998064896288,
-      "loss": 46.0,
-      "step": 2600
-    },
-    {
-      "epoch": 0.19886461379666265,
-      "grad_norm": 0.0012281349627301097,
-      "learning_rate": 0.00019999998063401713,
-      "loss": 46.0,
-      "step": 2601
-    },
-    {
-      "epoch": 0.19894107078005238,
-      "grad_norm": 0.000990453059785068,
-      "learning_rate": 0.0001999999806190656,
-      "loss": 46.0,
-      "step": 2602
-    },
-    {
-      "epoch": 0.1990175277634421,
-      "grad_norm": 0.0018406666349619627,
-      "learning_rate": 0.00019999998060410827,
-      "loss": 46.0,
-      "step": 2603
-    },
-    {
-      "epoch": 0.19909398474683182,
-      "grad_norm": 0.0007173509802669287,
-      "learning_rate": 0.0001999999805891452,
-      "loss": 46.0,
-      "step": 2604
-    },
-    {
-      "epoch": 0.19917044173022153,
-      "grad_norm": 0.0012165848165750504,
-      "learning_rate": 0.00019999998057417635,
-      "loss": 46.0,
-      "step": 2605
-    },
-    {
-      "epoch": 0.19924689871361126,
-      "grad_norm": 0.0017083440907299519,
-      "learning_rate": 0.00019999998055920175,
-      "loss": 46.0,
-      "step": 2606
-    },
-    {
-      "epoch": 0.19932335569700096,
-      "grad_norm": 0.0018412291537970304,
-      "learning_rate": 0.00019999998054422133,
-      "loss": 46.0,
-      "step": 2607
-    },
-    {
-      "epoch": 0.1993998126803907,
-      "grad_norm": 0.0015125240897759795,
-      "learning_rate": 0.0001999999805292352,
-      "loss": 46.0,
-      "step": 2608
-    },
-    {
-      "epoch": 0.1994762696637804,
-      "grad_norm": 0.0023048233706504107,
-      "learning_rate": 0.00019999998051424328,
-      "loss": 46.0,
-      "step": 2609
-    },
-    {
-      "epoch": 0.19955272664717014,
-      "grad_norm": 0.002014223486185074,
-      "learning_rate": 0.00019999998049924556,
-      "loss": 46.0,
-      "step": 2610
-    },
-    {
-      "epoch": 0.19962918363055987,
-      "grad_norm": 0.0011433360632508993,
-      "learning_rate": 0.0001999999804842421,
-      "loss": 46.0,
-      "step": 2611
-    },
-    {
-      "epoch": 0.19970564061394958,
-      "grad_norm": 0.0006147747044451535,
-      "learning_rate": 0.00019999998046923287,
-      "loss": 46.0,
-      "step": 2612
-    },
-    {
-      "epoch": 0.1997820975973393,
-      "grad_norm": 0.0009227930568158627,
-      "learning_rate": 0.00019999998045421786,
-      "loss": 46.0,
-      "step": 2613
-    },
-    {
-      "epoch": 0.19985855458072901,
-      "grad_norm": 0.0007886108360253274,
-      "learning_rate": 0.0001999999804391971,
-      "loss": 46.0,
-      "step": 2614
-    },
-    {
-      "epoch": 0.19993501156411875,
-      "grad_norm": 0.004000439308583736,
-      "learning_rate": 0.00019999998042417055,
-      "loss": 46.0,
-      "step": 2615
-    },
-    {
-      "epoch": 0.20001146854750845,
-      "grad_norm": 0.0003597511677071452,
-      "learning_rate": 0.00019999998040913823,
-      "loss": 46.0,
-      "step": 2616
-    },
-    {
-      "epoch": 0.2000879255308982,
-      "grad_norm": 0.001686968607828021,
-      "learning_rate": 0.00019999998039410018,
-      "loss": 46.0,
-      "step": 2617
-    },
-    {
-      "epoch": 0.2001643825142879,
-      "grad_norm": 0.0009858241537585855,
-      "learning_rate": 0.0001999999803790563,
-      "loss": 46.0,
-      "step": 2618
-    },
-    {
-      "epoch": 0.20024083949767763,
-      "grad_norm": 0.0008331689750775695,
-      "learning_rate": 0.00019999998036400668,
-      "loss": 46.0,
-      "step": 2619
-    },
-    {
-      "epoch": 0.20031729648106733,
-      "grad_norm": 0.0013714659726247191,
-      "learning_rate": 0.00019999998034895132,
-      "loss": 46.0,
-      "step": 2620
-    },
-    {
-      "epoch": 0.20039375346445706,
-      "grad_norm": 0.0023798574693500996,
-      "learning_rate": 0.00019999998033389015,
-      "loss": 46.0,
-      "step": 2621
-    },
-    {
-      "epoch": 0.20047021044784677,
-      "grad_norm": 0.0006794448709115386,
-      "learning_rate": 0.0001999999803188232,
-      "loss": 46.0,
-      "step": 2622
-    },
-    {
-      "epoch": 0.2005466674312365,
-      "grad_norm": 0.0012324457056820393,
-      "learning_rate": 0.00019999998030375052,
-      "loss": 46.0,
-      "step": 2623
-    },
-    {
-      "epoch": 0.2006231244146262,
-      "grad_norm": 0.0015901293372735381,
-      "learning_rate": 0.00019999998028867206,
-      "loss": 46.0,
-      "step": 2624
-    },
-    {
-      "epoch": 0.20069958139801594,
-      "grad_norm": 0.000446124846348539,
-      "learning_rate": 0.00019999998027358783,
-      "loss": 46.0,
-      "step": 2625
-    },
-    {
-      "epoch": 0.20077603838140567,
-      "grad_norm": 0.0008350448915734887,
-      "learning_rate": 0.0001999999802584978,
-      "loss": 46.0,
-      "step": 2626
-    },
-    {
-      "epoch": 0.20085249536479538,
-      "grad_norm": 0.002167291473597288,
-      "learning_rate": 0.00019999998024340204,
-      "loss": 46.0,
-      "step": 2627
-    },
-    {
-      "epoch": 0.2009289523481851,
-      "grad_norm": 0.0013825835194438696,
-      "learning_rate": 0.0001999999802283005,
-      "loss": 46.0,
-      "step": 2628
-    },
-    {
-      "epoch": 0.20100540933157482,
-      "grad_norm": 0.0007878118776716292,
-      "learning_rate": 0.00019999998021319316,
-      "loss": 46.0,
-      "step": 2629
-    },
-    {
-      "epoch": 0.20108186631496455,
-      "grad_norm": 0.0033684580121189356,
-      "learning_rate": 0.00019999998019808009,
-      "loss": 46.0,
-      "step": 2630
-    },
-    {
-      "epoch": 0.20115832329835426,
-      "grad_norm": 0.00131437205709517,
-      "learning_rate": 0.00019999998018296124,
-      "loss": 46.0,
-      "step": 2631
-    },
-    {
-      "epoch": 0.201234780281744,
-      "grad_norm": 0.0019817743450403214,
-      "learning_rate": 0.00019999998016783662,
-      "loss": 46.0,
-      "step": 2632
-    },
-    {
-      "epoch": 0.2013112372651337,
-      "grad_norm": 0.000990221044048667,
-      "learning_rate": 0.0001999999801527062,
-      "loss": 46.0,
-      "step": 2633
-    },
-    {
-      "epoch": 0.20138769424852343,
-      "grad_norm": 0.0023654510732740164,
-      "learning_rate": 0.00019999998013757003,
-      "loss": 46.0,
-      "step": 2634
-    },
-    {
-      "epoch": 0.20146415123191314,
-      "grad_norm": 0.004006518982350826,
-      "learning_rate": 0.00019999998012242812,
-      "loss": 46.0,
-      "step": 2635
-    },
-    {
-      "epoch": 0.20154060821530287,
-      "grad_norm": 0.0007428004173561931,
-      "learning_rate": 0.0001999999801072804,
-      "loss": 46.0,
-      "step": 2636
-    },
-    {
-      "epoch": 0.20161706519869257,
-      "grad_norm": 0.0007062962977215648,
-      "learning_rate": 0.00019999998009212694,
-      "loss": 46.0,
-      "step": 2637
-    },
-    {
-      "epoch": 0.2016935221820823,
-      "grad_norm": 0.0013279336271807551,
-      "learning_rate": 0.0001999999800769677,
-      "loss": 46.0,
-      "step": 2638
-    },
-    {
-      "epoch": 0.201769979165472,
-      "grad_norm": 0.0021120128221809864,
-      "learning_rate": 0.0001999999800618027,
-      "loss": 46.0,
-      "step": 2639
-    },
-    {
-      "epoch": 0.20184643614886175,
-      "grad_norm": 0.0010096341138705611,
-      "learning_rate": 0.00019999998004663192,
-      "loss": 46.0,
-      "step": 2640
-    },
-    {
-      "epoch": 0.20192289313225148,
-      "grad_norm": 0.001144168432801962,
-      "learning_rate": 0.00019999998003145537,
-      "loss": 46.0,
-      "step": 2641
-    },
-    {
-      "epoch": 0.20199935011564119,
-      "grad_norm": 0.0014059501700103283,
-      "learning_rate": 0.00019999998001627307,
-      "loss": 46.0,
-      "step": 2642
-    },
-    {
-      "epoch": 0.20207580709903092,
-      "grad_norm": 0.0007812492549419403,
-      "learning_rate": 0.00019999998000108497,
-      "loss": 46.0,
-      "step": 2643
-    },
-    {
-      "epoch": 0.20215226408242062,
-      "grad_norm": 0.0015959207667037845,
-      "learning_rate": 0.0001999999799858911,
-      "loss": 46.0,
-      "step": 2644
-    },
-    {
-      "epoch": 0.20222872106581036,
-      "grad_norm": 0.0012493618996813893,
-      "learning_rate": 0.00019999997997069148,
-      "loss": 46.0,
-      "step": 2645
-    },
-    {
-      "epoch": 0.20230517804920006,
-      "grad_norm": 0.0005881648976355791,
-      "learning_rate": 0.00019999997995548608,
-      "loss": 46.0,
-      "step": 2646
-    },
-    {
-      "epoch": 0.2023816350325898,
-      "grad_norm": 0.001119842054322362,
-      "learning_rate": 0.00019999997994027492,
-      "loss": 46.0,
-      "step": 2647
-    },
-    {
-      "epoch": 0.2024580920159795,
-      "grad_norm": 0.0007371313986368477,
-      "learning_rate": 0.000199999979925058,
-      "loss": 46.0,
-      "step": 2648
-    },
-    {
-      "epoch": 0.20253454899936923,
-      "grad_norm": 0.0008875987841747701,
-      "learning_rate": 0.0001999999799098353,
-      "loss": 46.0,
-      "step": 2649
-    },
-    {
-      "epoch": 0.20261100598275894,
-      "grad_norm": 0.0006791872438043356,
-      "learning_rate": 0.0001999999798946068,
-      "loss": 46.0,
-      "step": 2650
-    },
-    {
-      "epoch": 0.20268746296614867,
-      "grad_norm": 0.0010499234776943922,
-      "learning_rate": 0.00019999997987937257,
-      "loss": 46.0,
-      "step": 2651
-    },
-    {
-      "epoch": 0.20276391994953838,
-      "grad_norm": 0.0023816260509192944,
-      "learning_rate": 0.00019999997986413254,
-      "loss": 46.0,
-      "step": 2652
-    },
-    {
-      "epoch": 0.2028403769329281,
-      "grad_norm": 0.0023006810806691647,
-      "learning_rate": 0.00019999997984888678,
-      "loss": 46.0,
-      "step": 2653
-    },
-    {
-      "epoch": 0.20291683391631782,
-      "grad_norm": 0.0008985421154648066,
-      "learning_rate": 0.00019999997983363523,
-      "loss": 46.0,
-      "step": 2654
-    },
-    {
-      "epoch": 0.20299329089970755,
-      "grad_norm": 0.002558620646595955,
-      "learning_rate": 0.0001999999798183779,
-      "loss": 46.0,
-      "step": 2655
-    },
-    {
-      "epoch": 0.20306974788309728,
-      "grad_norm": 0.001698905834928155,
-      "learning_rate": 0.00019999997980311483,
-      "loss": 46.0,
-      "step": 2656
-    },
-    {
-      "epoch": 0.203146204866487,
-      "grad_norm": 0.0009687116253189743,
-      "learning_rate": 0.00019999997978784598,
-      "loss": 46.0,
-      "step": 2657
-    },
-    {
-      "epoch": 0.20322266184987672,
-      "grad_norm": 0.0012422003783285618,
-      "learning_rate": 0.00019999997977257134,
-      "loss": 46.0,
-      "step": 2658
-    },
-    {
-      "epoch": 0.20329911883326643,
-      "grad_norm": 0.0019307717448100448,
-      "learning_rate": 0.00019999997975729094,
-      "loss": 46.0,
-      "step": 2659
-    },
-    {
-      "epoch": 0.20337557581665616,
-      "grad_norm": 0.00837969034910202,
-      "learning_rate": 0.00019999997974200478,
-      "loss": 46.0,
-      "step": 2660
-    },
-    {
-      "epoch": 0.20345203280004587,
-      "grad_norm": 0.0011417909990996122,
-      "learning_rate": 0.00019999997972671284,
-      "loss": 46.0,
-      "step": 2661
-    },
-    {
-      "epoch": 0.2035284897834356,
-      "grad_norm": 0.0011604258324950933,
-      "learning_rate": 0.0001999999797114151,
-      "loss": 46.0,
-      "step": 2662
-    },
-    {
-      "epoch": 0.2036049467668253,
-      "grad_norm": 0.0008633575052954257,
-      "learning_rate": 0.00019999997969611164,
-      "loss": 46.0,
-      "step": 2663
-    },
-    {
-      "epoch": 0.20368140375021504,
-      "grad_norm": 0.0008733525755815208,
-      "learning_rate": 0.0001999999796808024,
-      "loss": 46.0,
-      "step": 2664
-    },
-    {
-      "epoch": 0.20375786073360475,
-      "grad_norm": 0.0012839054688811302,
-      "learning_rate": 0.00019999997966548737,
-      "loss": 46.0,
-      "step": 2665
-    },
-    {
-      "epoch": 0.20383431771699448,
-      "grad_norm": 0.0004955227486789227,
-      "learning_rate": 0.0001999999796501666,
-      "loss": 46.0,
-      "step": 2666
-    },
-    {
-      "epoch": 0.20391077470038418,
-      "grad_norm": 0.0011039386736229062,
-      "learning_rate": 0.00019999997963484004,
-      "loss": 46.0,
-      "step": 2667
-    },
-    {
-      "epoch": 0.20398723168377392,
-      "grad_norm": 0.0022267145104706287,
-      "learning_rate": 0.0001999999796195077,
-      "loss": 46.0,
-      "step": 2668
-    },
-    {
-      "epoch": 0.20406368866716365,
-      "grad_norm": 0.0029530690517276525,
-      "learning_rate": 0.00019999997960416964,
-      "loss": 46.0,
-      "step": 2669
-    },
-    {
-      "epoch": 0.20414014565055336,
-      "grad_norm": 0.0025334572419524193,
-      "learning_rate": 0.00019999997958882577,
-      "loss": 46.0,
-      "step": 2670
-    },
-    {
-      "epoch": 0.2042166026339431,
-      "grad_norm": 0.0008771863067522645,
-      "learning_rate": 0.00019999997957347615,
-      "loss": 46.0,
-      "step": 2671
-    },
-    {
-      "epoch": 0.2042930596173328,
-      "grad_norm": 0.0009226668043993413,
-      "learning_rate": 0.00019999997955812073,
-      "loss": 46.0,
-      "step": 2672
-    },
-    {
-      "epoch": 0.20436951660072253,
-      "grad_norm": 0.0010280348360538483,
-      "learning_rate": 0.00019999997954275956,
-      "loss": 46.0,
-      "step": 2673
-    },
-    {
-      "epoch": 0.20444597358411223,
-      "grad_norm": 0.004375841002911329,
-      "learning_rate": 0.00019999997952739262,
-      "loss": 46.0,
-      "step": 2674
-    },
-    {
-      "epoch": 0.20452243056750197,
-      "grad_norm": 0.0006381785497069359,
-      "learning_rate": 0.0001999999795120199,
-      "loss": 46.0,
-      "step": 2675
-    },
-    {
-      "epoch": 0.20459888755089167,
-      "grad_norm": 0.00226203934289515,
-      "learning_rate": 0.00019999997949664145,
-      "loss": 46.0,
-      "step": 2676
-    },
-    {
-      "epoch": 0.2046753445342814,
-      "grad_norm": 0.0019268908072263002,
-      "learning_rate": 0.00019999997948125716,
-      "loss": 46.0,
-      "step": 2677
-    },
-    {
-      "epoch": 0.2047518015176711,
-      "grad_norm": 0.0012107514776289463,
-      "learning_rate": 0.00019999997946586718,
-      "loss": 46.0,
-      "step": 2678
-    },
-    {
-      "epoch": 0.20482825850106084,
-      "grad_norm": 0.0006391117931343615,
-      "learning_rate": 0.00019999997945047138,
-      "loss": 46.0,
-      "step": 2679
-    },
-    {
-      "epoch": 0.20490471548445055,
-      "grad_norm": 0.0015201630303636193,
-      "learning_rate": 0.0001999999794350698,
-      "loss": 46.0,
-      "step": 2680
-    },
-    {
-      "epoch": 0.20498117246784028,
-      "grad_norm": 0.001089393044821918,
-      "learning_rate": 0.0001999999794196625,
-      "loss": 46.0,
-      "step": 2681
-    },
-    {
-      "epoch": 0.20505762945123,
-      "grad_norm": 0.0020287600345909595,
-      "learning_rate": 0.0001999999794042494,
-      "loss": 46.0,
-      "step": 2682
-    },
-    {
-      "epoch": 0.20513408643461972,
-      "grad_norm": 0.001776363467797637,
-      "learning_rate": 0.00019999997938883053,
-      "loss": 46.0,
-      "step": 2683
-    },
-    {
-      "epoch": 0.20521054341800946,
-      "grad_norm": 0.0014043195405974984,
-      "learning_rate": 0.00019999997937340588,
-      "loss": 46.0,
-      "step": 2684
-    },
-    {
-      "epoch": 0.20528700040139916,
-      "grad_norm": 0.0030978552531450987,
-      "learning_rate": 0.0001999999793579755,
-      "loss": 46.0,
-      "step": 2685
-    },
-    {
-      "epoch": 0.2053634573847889,
-      "grad_norm": 0.0008114740485325456,
-      "learning_rate": 0.00019999997934253932,
-      "loss": 46.0,
-      "step": 2686
-    },
-    {
-      "epoch": 0.2054399143681786,
-      "grad_norm": 0.0008145145257003605,
-      "learning_rate": 0.00019999997932709738,
-      "loss": 46.0,
-      "step": 2687
-    },
-    {
-      "epoch": 0.20551637135156833,
-      "grad_norm": 0.001314866472966969,
-      "learning_rate": 0.00019999997931164967,
-      "loss": 46.0,
-      "step": 2688
-    },
-    {
-      "epoch": 0.20559282833495804,
-      "grad_norm": 0.002413044683635235,
-      "learning_rate": 0.00019999997929619619,
-      "loss": 46.0,
-      "step": 2689
-    },
-    {
-      "epoch": 0.20566928531834777,
-      "grad_norm": 0.00035419585765339434,
-      "learning_rate": 0.00019999997928073695,
-      "loss": 46.0,
-      "step": 2690
-    },
-    {
-      "epoch": 0.20574574230173748,
-      "grad_norm": 0.0013094025198370218,
-      "learning_rate": 0.0001999999792652719,
-      "loss": 46.0,
-      "step": 2691
-    },
-    {
-      "epoch": 0.2058221992851272,
-      "grad_norm": 0.002120288321748376,
-      "learning_rate": 0.00019999997924980111,
-      "loss": 46.0,
-      "step": 2692
-    },
-    {
-      "epoch": 0.20589865626851692,
-      "grad_norm": 0.0012566301738843322,
-      "learning_rate": 0.00019999997923432456,
-      "loss": 46.0,
-      "step": 2693
-    },
-    {
-      "epoch": 0.20597511325190665,
-      "grad_norm": 0.007534113246947527,
-      "learning_rate": 0.00019999997921884224,
-      "loss": 46.0,
-      "step": 2694
-    },
-    {
-      "epoch": 0.20605157023529636,
-      "grad_norm": 0.003485876601189375,
-      "learning_rate": 0.00019999997920335414,
-      "loss": 46.0,
-      "step": 2695
-    },
-    {
-      "epoch": 0.2061280272186861,
-      "grad_norm": 0.0016508623957633972,
-      "learning_rate": 0.00019999997918786026,
-      "loss": 46.0,
-      "step": 2696
-    },
-    {
-      "epoch": 0.2062044842020758,
-      "grad_norm": 0.0007939081406220794,
-      "learning_rate": 0.00019999997917236062,
-      "loss": 46.0,
-      "step": 2697
-    },
-    {
-      "epoch": 0.20628094118546553,
-      "grad_norm": 0.0005702505586668849,
-      "learning_rate": 0.00019999997915685523,
-      "loss": 46.0,
-      "step": 2698
-    },
-    {
-      "epoch": 0.20635739816885526,
-      "grad_norm": 0.0021604765206575394,
-      "learning_rate": 0.00019999997914134406,
-      "loss": 46.0,
-      "step": 2699
-    },
-    {
-      "epoch": 0.20643385515224497,
-      "grad_norm": 0.0031259257812052965,
-      "learning_rate": 0.0001999999791258271,
-      "loss": 46.0,
-      "step": 2700
-    },
-    {
-      "epoch": 0.2065103121356347,
-      "grad_norm": 0.0024204538203775883,
-      "learning_rate": 0.00019999997911030438,
-      "loss": 46.0,
-      "step": 2701
-    },
-    {
-      "epoch": 0.2065867691190244,
-      "grad_norm": 0.0021650376729667187,
-      "learning_rate": 0.0001999999790947759,
-      "loss": 46.0,
-      "step": 2702
-    },
-    {
-      "epoch": 0.20666322610241414,
-      "grad_norm": 0.0031147070694714785,
-      "learning_rate": 0.00019999997907924164,
-      "loss": 46.0,
-      "step": 2703
-    },
-    {
-      "epoch": 0.20673968308580384,
-      "grad_norm": 0.0015438698465004563,
-      "learning_rate": 0.00019999997906370163,
-      "loss": 46.0,
-      "step": 2704
-    },
-    {
-      "epoch": 0.20681614006919358,
-      "grad_norm": 0.0012661865912377834,
-      "learning_rate": 0.00019999997904815583,
-      "loss": 46.0,
-      "step": 2705
-    },
-    {
-      "epoch": 0.20689259705258328,
-      "grad_norm": 0.0006658679340034723,
-      "learning_rate": 0.00019999997903260428,
-      "loss": 46.0,
-      "step": 2706
-    },
-    {
-      "epoch": 0.20696905403597302,
-      "grad_norm": 0.002733797300606966,
-      "learning_rate": 0.00019999997901704695,
-      "loss": 46.0,
-      "step": 2707
-    },
-    {
-      "epoch": 0.20704551101936272,
-      "grad_norm": 0.0009632634464651346,
-      "learning_rate": 0.00019999997900148385,
-      "loss": 46.0,
-      "step": 2708
-    },
-    {
-      "epoch": 0.20712196800275245,
-      "grad_norm": 0.0006991362315602601,
-      "learning_rate": 0.00019999997898591495,
-      "loss": 46.0,
-      "step": 2709
-    },
-    {
-      "epoch": 0.20719842498614216,
-      "grad_norm": 0.0010518896160647273,
-      "learning_rate": 0.0001999999789703403,
-      "loss": 46.0,
-      "step": 2710
-    },
-    {
-      "epoch": 0.2072748819695319,
-      "grad_norm": 0.0029758852906525135,
-      "learning_rate": 0.00019999997895475991,
-      "loss": 46.0,
-      "step": 2711
-    },
-    {
-      "epoch": 0.2073513389529216,
-      "grad_norm": 0.0013607953442260623,
-      "learning_rate": 0.00019999997893917372,
-      "loss": 46.0,
-      "step": 2712
-    },
-    {
-      "epoch": 0.20742779593631133,
-      "grad_norm": 0.000644504267256707,
-      "learning_rate": 0.0001999999789235818,
-      "loss": 46.0,
-      "step": 2713
-    },
-    {
-      "epoch": 0.20750425291970107,
-      "grad_norm": 0.0007269313791766763,
-      "learning_rate": 0.00019999997890798404,
-      "loss": 46.0,
-      "step": 2714
-    },
-    {
-      "epoch": 0.20758070990309077,
-      "grad_norm": 0.0016563581302762032,
-      "learning_rate": 0.0001999999788923806,
-      "loss": 46.0,
-      "step": 2715
-    },
-    {
-      "epoch": 0.2076571668864805,
-      "grad_norm": 0.0008605378097854555,
-      "learning_rate": 0.00019999997887677133,
-      "loss": 46.0,
-      "step": 2716
-    },
-    {
-      "epoch": 0.2077336238698702,
-      "grad_norm": 0.0016932736616581678,
-      "learning_rate": 0.0001999999788611563,
-      "loss": 46.0,
-      "step": 2717
-    },
-    {
-      "epoch": 0.20781008085325994,
-      "grad_norm": 0.0003803808067459613,
-      "learning_rate": 0.0001999999788455355,
-      "loss": 46.0,
-      "step": 2718
-    },
-    {
-      "epoch": 0.20788653783664965,
-      "grad_norm": 0.0005976628162898123,
-      "learning_rate": 0.00019999997882990894,
-      "loss": 46.0,
-      "step": 2719
-    },
-    {
-      "epoch": 0.20796299482003938,
-      "grad_norm": 0.0013230486074462533,
-      "learning_rate": 0.00019999997881427661,
-      "loss": 46.0,
-      "step": 2720
-    },
-    {
-      "epoch": 0.2080394518034291,
-      "grad_norm": 0.00038978815427981317,
-      "learning_rate": 0.0001999999787986385,
-      "loss": 46.0,
-      "step": 2721
-    },
-    {
-      "epoch": 0.20811590878681882,
-      "grad_norm": 0.001036260393448174,
-      "learning_rate": 0.00019999997878299462,
-      "loss": 46.0,
-      "step": 2722
-    },
-    {
-      "epoch": 0.20819236577020853,
-      "grad_norm": 0.0026271517854183912,
-      "learning_rate": 0.00019999997876734497,
-      "loss": 46.0,
-      "step": 2723
-    },
-    {
-      "epoch": 0.20826882275359826,
-      "grad_norm": 0.0026575273368507624,
-      "learning_rate": 0.00019999997875168955,
-      "loss": 46.0,
-      "step": 2724
-    },
-    {
-      "epoch": 0.20834527973698796,
-      "grad_norm": 0.0007150326273404062,
-      "learning_rate": 0.0001999999787360284,
-      "loss": 46.0,
-      "step": 2725
-    },
-    {
-      "epoch": 0.2084217367203777,
-      "grad_norm": 0.0005619221483357251,
-      "learning_rate": 0.00019999997872036142,
-      "loss": 46.0,
-      "step": 2726
-    },
-    {
-      "epoch": 0.20849819370376743,
-      "grad_norm": 0.0007964373799040914,
-      "learning_rate": 0.0001999999787046887,
-      "loss": 46.0,
-      "step": 2727
-    },
-    {
-      "epoch": 0.20857465068715714,
-      "grad_norm": 0.0012551010586321354,
-      "learning_rate": 0.00019999997868901023,
-      "loss": 46.0,
-      "step": 2728
-    },
-    {
-      "epoch": 0.20865110767054687,
-      "grad_norm": 0.0014274526620283723,
-      "learning_rate": 0.00019999997867332597,
-      "loss": 46.0,
-      "step": 2729
-    },
-    {
-      "epoch": 0.20872756465393658,
-      "grad_norm": 0.0013953399611636996,
-      "learning_rate": 0.00019999997865763594,
-      "loss": 46.0,
-      "step": 2730
-    },
-    {
-      "epoch": 0.2088040216373263,
-      "grad_norm": 0.001514603616669774,
-      "learning_rate": 0.00019999997864194013,
-      "loss": 46.0,
-      "step": 2731
-    },
-    {
-      "epoch": 0.20888047862071601,
-      "grad_norm": 0.001486646244302392,
-      "learning_rate": 0.00019999997862623855,
-      "loss": 46.0,
-      "step": 2732
-    },
-    {
-      "epoch": 0.20895693560410575,
-      "grad_norm": 0.0008643695036880672,
-      "learning_rate": 0.00019999997861053123,
-      "loss": 46.0,
-      "step": 2733
-    },
-    {
-      "epoch": 0.20903339258749545,
-      "grad_norm": 0.0012586575467139482,
-      "learning_rate": 0.00019999997859481813,
-      "loss": 46.0,
-      "step": 2734
-    },
-    {
-      "epoch": 0.2091098495708852,
-      "grad_norm": 0.003097428707405925,
-      "learning_rate": 0.00019999997857909923,
-      "loss": 46.0,
-      "step": 2735
-    },
-    {
-      "epoch": 0.2091863065542749,
-      "grad_norm": 0.002444636542350054,
-      "learning_rate": 0.0001999999785633746,
-      "loss": 46.0,
-      "step": 2736
-    },
-    {
-      "epoch": 0.20926276353766463,
-      "grad_norm": 0.0005594029207713902,
-      "learning_rate": 0.00019999997854764417,
-      "loss": 46.0,
-      "step": 2737
-    },
-    {
-      "epoch": 0.20933922052105433,
-      "grad_norm": 0.0014575031818822026,
-      "learning_rate": 0.000199999978531908,
-      "loss": 46.0,
-      "step": 2738
-    },
-    {
-      "epoch": 0.20941567750444406,
-      "grad_norm": 0.0028861218597739935,
-      "learning_rate": 0.00019999997851616606,
-      "loss": 46.0,
-      "step": 2739
-    },
-    {
-      "epoch": 0.20949213448783377,
-      "grad_norm": 0.0006614363519474864,
-      "learning_rate": 0.00019999997850041833,
-      "loss": 46.0,
-      "step": 2740
-    },
-    {
-      "epoch": 0.2095685914712235,
-      "grad_norm": 0.0009695856715552509,
-      "learning_rate": 0.00019999997848466481,
-      "loss": 46.0,
-      "step": 2741
-    },
-    {
-      "epoch": 0.20964504845461324,
-      "grad_norm": 0.001881250413134694,
-      "learning_rate": 0.00019999997846890556,
-      "loss": 46.0,
-      "step": 2742
-    },
-    {
-      "epoch": 0.20972150543800294,
-      "grad_norm": 0.0012821612181141973,
-      "learning_rate": 0.00019999997845314053,
-      "loss": 46.0,
-      "step": 2743
-    },
-    {
-      "epoch": 0.20979796242139268,
-      "grad_norm": 0.0011615402763709426,
-      "learning_rate": 0.00019999997843736972,
-      "loss": 46.0,
-      "step": 2744
-    },
-    {
-      "epoch": 0.20987441940478238,
-      "grad_norm": 0.0010495653841644526,
-      "learning_rate": 0.00019999997842159314,
-      "loss": 46.0,
-      "step": 2745
-    },
-    {
-      "epoch": 0.20995087638817211,
-      "grad_norm": 0.0009658645722083747,
-      "learning_rate": 0.00019999997840581082,
-      "loss": 46.0,
-      "step": 2746
-    },
-    {
-      "epoch": 0.21002733337156182,
-      "grad_norm": 0.0016275092493742704,
-      "learning_rate": 0.0001999999783900227,
-      "loss": 46.0,
-      "step": 2747
-    },
-    {
-      "epoch": 0.21010379035495155,
-      "grad_norm": 0.001446898328140378,
-      "learning_rate": 0.00019999997837422885,
-      "loss": 46.0,
-      "step": 2748
-    },
-    {
-      "epoch": 0.21018024733834126,
-      "grad_norm": 0.0021533663384616375,
-      "learning_rate": 0.00019999997835842915,
-      "loss": 46.0,
-      "step": 2749
-    },
-    {
-      "epoch": 0.210256704321731,
-      "grad_norm": 0.00771222123876214,
-      "learning_rate": 0.00019999997834262376,
-      "loss": 46.0,
-      "step": 2750
-    },
-    {
-      "epoch": 0.2103331613051207,
-      "grad_norm": 0.0015932237729430199,
-      "learning_rate": 0.00019999997832681257,
-      "loss": 46.0,
-      "step": 2751
-    },
-    {
-      "epoch": 0.21040961828851043,
-      "grad_norm": 0.0008433216717094183,
-      "learning_rate": 0.0001999999783109956,
-      "loss": 46.0,
-      "step": 2752
-    },
-    {
-      "epoch": 0.21048607527190014,
-      "grad_norm": 0.0011260309256613255,
-      "learning_rate": 0.00019999997829517287,
-      "loss": 46.0,
-      "step": 2753
-    },
-    {
-      "epoch": 0.21056253225528987,
-      "grad_norm": 0.0010261379648000002,
-      "learning_rate": 0.00019999997827934438,
-      "loss": 46.0,
-      "step": 2754
-    },
-    {
-      "epoch": 0.21063898923867957,
-      "grad_norm": 0.0014874220360070467,
-      "learning_rate": 0.00019999997826351012,
-      "loss": 46.0,
-      "step": 2755
-    },
-    {
-      "epoch": 0.2107154462220693,
-      "grad_norm": 0.0020668439101427794,
-      "learning_rate": 0.00019999997824767006,
-      "loss": 46.0,
-      "step": 2756
-    },
-    {
-      "epoch": 0.21079190320545904,
-      "grad_norm": 0.0011867792345583439,
-      "learning_rate": 0.0001999999782318243,
-      "loss": 46.0,
-      "step": 2757
-    },
-    {
-      "epoch": 0.21086836018884875,
-      "grad_norm": 0.0031772837974131107,
-      "learning_rate": 0.0001999999782159727,
-      "loss": 46.0,
-      "step": 2758
-    },
-    {
-      "epoch": 0.21094481717223848,
-      "grad_norm": 0.0012536614667624235,
-      "learning_rate": 0.00019999997820011536,
-      "loss": 46.0,
-      "step": 2759
-    },
-    {
-      "epoch": 0.21102127415562819,
-      "grad_norm": 0.0014362270012497902,
-      "learning_rate": 0.00019999997818425223,
-      "loss": 46.0,
-      "step": 2760
-    },
-    {
-      "epoch": 0.21109773113901792,
-      "grad_norm": 0.00485815666615963,
-      "learning_rate": 0.00019999997816838336,
-      "loss": 46.0,
-      "step": 2761
-    },
-    {
-      "epoch": 0.21117418812240762,
-      "grad_norm": 0.0010709596099331975,
-      "learning_rate": 0.00019999997815250872,
-      "loss": 46.0,
-      "step": 2762
-    },
-    {
-      "epoch": 0.21125064510579736,
-      "grad_norm": 0.0027423000428825617,
-      "learning_rate": 0.0001999999781366283,
-      "loss": 46.0,
-      "step": 2763
-    },
-    {
-      "epoch": 0.21132710208918706,
-      "grad_norm": 0.0004806232755072415,
-      "learning_rate": 0.00019999997812074208,
-      "loss": 46.0,
-      "step": 2764
-    },
-    {
-      "epoch": 0.2114035590725768,
-      "grad_norm": 0.0008911628974601626,
-      "learning_rate": 0.00019999997810485015,
-      "loss": 46.0,
-      "step": 2765
-    },
-    {
-      "epoch": 0.2114800160559665,
-      "grad_norm": 0.0007796669378876686,
-      "learning_rate": 0.0001999999780889524,
-      "loss": 46.0,
-      "step": 2766
-    },
-    {
-      "epoch": 0.21155647303935624,
-      "grad_norm": 0.0013641685945913196,
-      "learning_rate": 0.00019999997807304892,
-      "loss": 46.0,
-      "step": 2767
-    },
-    {
-      "epoch": 0.21163293002274594,
-      "grad_norm": 0.0009681656956672668,
-      "learning_rate": 0.00019999997805713964,
-      "loss": 46.0,
-      "step": 2768
-    },
-    {
-      "epoch": 0.21170938700613567,
-      "grad_norm": 0.004737140145152807,
-      "learning_rate": 0.00019999997804122458,
-      "loss": 46.0,
-      "step": 2769
-    },
-    {
-      "epoch": 0.21178584398952538,
-      "grad_norm": 0.0018367001321166754,
-      "learning_rate": 0.0001999999780253038,
-      "loss": 46.0,
-      "step": 2770
-    },
-    {
-      "epoch": 0.2118623009729151,
-      "grad_norm": 0.002287733368575573,
-      "learning_rate": 0.00019999997800937723,
-      "loss": 46.0,
-      "step": 2771
-    },
-    {
-      "epoch": 0.21193875795630485,
-      "grad_norm": 0.0016572296153753996,
-      "learning_rate": 0.00019999997799344488,
-      "loss": 46.0,
-      "step": 2772
-    },
-    {
-      "epoch": 0.21201521493969455,
-      "grad_norm": 0.0010927787516266108,
-      "learning_rate": 0.00019999997797750675,
-      "loss": 46.0,
-      "step": 2773
-    },
-    {
-      "epoch": 0.21209167192308429,
-      "grad_norm": 0.0023603811860084534,
-      "learning_rate": 0.00019999997796156288,
-      "loss": 46.0,
-      "step": 2774
-    },
-    {
-      "epoch": 0.212168128906474,
-      "grad_norm": 0.0039044448640197515,
-      "learning_rate": 0.0001999999779456132,
-      "loss": 46.0,
-      "step": 2775
-    },
-    {
-      "epoch": 0.21224458588986372,
-      "grad_norm": 0.001130746561102569,
-      "learning_rate": 0.0001999999779296578,
-      "loss": 46.0,
-      "step": 2776
-    },
-    {
-      "epoch": 0.21232104287325343,
-      "grad_norm": 0.001227583852596581,
-      "learning_rate": 0.0001999999779136966,
-      "loss": 46.0,
-      "step": 2777
-    },
-    {
-      "epoch": 0.21239749985664316,
-      "grad_norm": 0.00040609861025586724,
-      "learning_rate": 0.00019999997789772964,
-      "loss": 46.0,
-      "step": 2778
-    },
-    {
-      "epoch": 0.21247395684003287,
-      "grad_norm": 0.0012875209795311093,
-      "learning_rate": 0.0001999999778817569,
-      "loss": 46.0,
-      "step": 2779
-    },
-    {
-      "epoch": 0.2125504138234226,
-      "grad_norm": 0.001201312756165862,
-      "learning_rate": 0.0001999999778657784,
-      "loss": 46.0,
-      "step": 2780
-    },
-    {
-      "epoch": 0.2126268708068123,
-      "grad_norm": 0.0006764173740521073,
-      "learning_rate": 0.0001999999778497941,
-      "loss": 46.0,
-      "step": 2781
-    },
-    {
-      "epoch": 0.21270332779020204,
-      "grad_norm": 0.0017223118338733912,
-      "learning_rate": 0.00019999997783380407,
-      "loss": 46.0,
-      "step": 2782
-    },
-    {
-      "epoch": 0.21277978477359175,
-      "grad_norm": 0.0002848854346666485,
-      "learning_rate": 0.00019999997781780827,
-      "loss": 46.0,
-      "step": 2783
-    },
-    {
-      "epoch": 0.21285624175698148,
-      "grad_norm": 0.0012034042738378048,
-      "learning_rate": 0.0001999999778018067,
-      "loss": 46.0,
-      "step": 2784
-    },
-    {
-      "epoch": 0.2129326987403712,
-      "grad_norm": 0.001390379387885332,
-      "learning_rate": 0.00019999997778579934,
-      "loss": 46.0,
-      "step": 2785
-    },
-    {
-      "epoch": 0.21300915572376092,
-      "grad_norm": 0.002384237013757229,
-      "learning_rate": 0.00019999997776978622,
-      "loss": 46.0,
-      "step": 2786
-    },
-    {
-      "epoch": 0.21308561270715065,
-      "grad_norm": 0.00044011662248522043,
-      "learning_rate": 0.00019999997775376732,
-      "loss": 46.0,
-      "step": 2787
-    },
-    {
-      "epoch": 0.21316206969054036,
-      "grad_norm": 0.0022945080418139696,
-      "learning_rate": 0.00019999997773774268,
-      "loss": 46.0,
-      "step": 2788
-    },
-    {
-      "epoch": 0.2132385266739301,
-      "grad_norm": 0.0004650302289519459,
-      "learning_rate": 0.00019999997772171226,
-      "loss": 46.0,
-      "step": 2789
-    },
-    {
-      "epoch": 0.2133149836573198,
-      "grad_norm": 0.0004878402396570891,
-      "learning_rate": 0.00019999997770567607,
-      "loss": 46.0,
-      "step": 2790
-    },
-    {
-      "epoch": 0.21339144064070953,
-      "grad_norm": 0.0007357585127465427,
-      "learning_rate": 0.0001999999776896341,
-      "loss": 46.0,
-      "step": 2791
-    },
-    {
-      "epoch": 0.21346789762409923,
-      "grad_norm": 0.0009692542953416705,
-      "learning_rate": 0.00019999997767358637,
-      "loss": 46.0,
-      "step": 2792
-    },
-    {
-      "epoch": 0.21354435460748897,
-      "grad_norm": 0.0007882479694671929,
-      "learning_rate": 0.00019999997765753283,
-      "loss": 46.0,
-      "step": 2793
-    },
-    {
-      "epoch": 0.21362081159087867,
-      "grad_norm": 0.0006892674136906862,
-      "learning_rate": 0.00019999997764147358,
-      "loss": 46.0,
-      "step": 2794
-    },
-    {
-      "epoch": 0.2136972685742684,
-      "grad_norm": 0.001179009792394936,
-      "learning_rate": 0.00019999997762540855,
-      "loss": 46.0,
-      "step": 2795
-    },
-    {
-      "epoch": 0.2137737255576581,
-      "grad_norm": 0.0012044119648635387,
-      "learning_rate": 0.00019999997760933771,
-      "loss": 46.0,
-      "step": 2796
-    },
-    {
-      "epoch": 0.21385018254104785,
-      "grad_norm": 0.0009840520797297359,
-      "learning_rate": 0.00019999997759326114,
-      "loss": 46.0,
-      "step": 2797
-    },
-    {
-      "epoch": 0.21392663952443755,
-      "grad_norm": 0.0010541044175624847,
-      "learning_rate": 0.0001999999775771788,
-      "loss": 46.0,
-      "step": 2798
-    },
-    {
-      "epoch": 0.21400309650782728,
-      "grad_norm": 0.0011699431343004107,
-      "learning_rate": 0.00019999997756109066,
-      "loss": 46.0,
-      "step": 2799
-    },
-    {
-      "epoch": 0.21407955349121702,
-      "grad_norm": 0.003357216250151396,
-      "learning_rate": 0.00019999997754499677,
-      "loss": 46.0,
-      "step": 2800
-    },
-    {
-      "epoch": 0.21415601047460672,
-      "grad_norm": 0.0013296111719682813,
-      "learning_rate": 0.00019999997752889712,
-      "loss": 46.0,
-      "step": 2801
-    },
-    {
-      "epoch": 0.21423246745799646,
-      "grad_norm": 0.0030206083320081234,
-      "learning_rate": 0.00019999997751279168,
-      "loss": 46.0,
-      "step": 2802
-    },
-    {
-      "epoch": 0.21430892444138616,
-      "grad_norm": 0.001299820956774056,
-      "learning_rate": 0.0001999999774966805,
-      "loss": 46.0,
-      "step": 2803
-    },
-    {
-      "epoch": 0.2143853814247759,
-      "grad_norm": 0.0005900601972825825,
-      "learning_rate": 0.00019999997748056353,
-      "loss": 46.0,
-      "step": 2804
-    },
-    {
-      "epoch": 0.2144618384081656,
-      "grad_norm": 0.0011733282590284944,
-      "learning_rate": 0.00019999997746444076,
-      "loss": 46.0,
-      "step": 2805
-    },
-    {
-      "epoch": 0.21453829539155533,
-      "grad_norm": 0.004266679286956787,
-      "learning_rate": 0.00019999997744831228,
-      "loss": 46.0,
-      "step": 2806
-    },
-    {
-      "epoch": 0.21461475237494504,
-      "grad_norm": 0.0005597809795290232,
-      "learning_rate": 0.000199999977432178,
-      "loss": 46.0,
-      "step": 2807
-    },
-    {
-      "epoch": 0.21469120935833477,
-      "grad_norm": 0.0010960650397464633,
-      "learning_rate": 0.00019999997741603794,
-      "loss": 46.0,
-      "step": 2808
-    },
-    {
-      "epoch": 0.21476766634172448,
-      "grad_norm": 0.0012118702288717031,
-      "learning_rate": 0.00019999997739989214,
-      "loss": 46.0,
-      "step": 2809
-    },
-    {
-      "epoch": 0.2148441233251142,
-      "grad_norm": 0.0007235713419504464,
-      "learning_rate": 0.00019999997738374053,
-      "loss": 46.0,
-      "step": 2810
-    },
-    {
-      "epoch": 0.21492058030850392,
-      "grad_norm": 0.02052875980734825,
-      "learning_rate": 0.00019999997736758318,
-      "loss": 46.0,
-      "step": 2811
-    },
-    {
-      "epoch": 0.21499703729189365,
-      "grad_norm": 0.0031311828643083572,
-      "learning_rate": 0.00019999997735142006,
-      "loss": 46.0,
-      "step": 2812
-    },
-    {
-      "epoch": 0.21507349427528336,
-      "grad_norm": 0.0007861799676902592,
-      "learning_rate": 0.0001999999773352512,
-      "loss": 46.0,
-      "step": 2813
-    },
-    {
-      "epoch": 0.2151499512586731,
-      "grad_norm": 0.0007042209035716951,
-      "learning_rate": 0.00019999997731907652,
-      "loss": 46.0,
-      "step": 2814
-    },
-    {
-      "epoch": 0.21522640824206282,
-      "grad_norm": 0.0066701313480734825,
-      "learning_rate": 0.00019999997730289607,
-      "loss": 46.0,
-      "step": 2815
-    },
-    {
-      "epoch": 0.21530286522545253,
-      "grad_norm": 0.0010436145821586251,
-      "learning_rate": 0.00019999997728670989,
-      "loss": 46.0,
-      "step": 2816
-    },
-    {
-      "epoch": 0.21537932220884226,
-      "grad_norm": 0.0008254246204160154,
-      "learning_rate": 0.0001999999772705179,
-      "loss": 46.0,
-      "step": 2817
-    },
-    {
-      "epoch": 0.21545577919223197,
-      "grad_norm": 0.0012020569993183017,
-      "learning_rate": 0.00019999997725432016,
-      "loss": 46.0,
-      "step": 2818
-    },
-    {
-      "epoch": 0.2155322361756217,
-      "grad_norm": 0.0012380763655528426,
-      "learning_rate": 0.00019999997723811665,
-      "loss": 46.0,
-      "step": 2819
-    },
-    {
-      "epoch": 0.2156086931590114,
-      "grad_norm": 0.0006606707465834916,
-      "learning_rate": 0.00019999997722190737,
-      "loss": 46.0,
-      "step": 2820
-    },
-    {
-      "epoch": 0.21568515014240114,
-      "grad_norm": 0.001367033808492124,
-      "learning_rate": 0.00019999997720569234,
-      "loss": 46.0,
-      "step": 2821
-    },
-    {
-      "epoch": 0.21576160712579084,
-      "grad_norm": 0.0010483409278094769,
-      "learning_rate": 0.0001999999771894715,
-      "loss": 46.0,
-      "step": 2822
-    },
-    {
-      "epoch": 0.21583806410918058,
-      "grad_norm": 0.0014846177073195577,
-      "learning_rate": 0.0001999999771732449,
-      "loss": 46.0,
-      "step": 2823
-    },
-    {
-      "epoch": 0.21591452109257028,
-      "grad_norm": 0.0007387480582110584,
-      "learning_rate": 0.00019999997715701255,
-      "loss": 46.0,
-      "step": 2824
-    },
-    {
-      "epoch": 0.21599097807596002,
-      "grad_norm": 0.001421895227394998,
-      "learning_rate": 0.00019999997714077443,
-      "loss": 46.0,
-      "step": 2825
-    },
-    {
-      "epoch": 0.21606743505934972,
-      "grad_norm": 0.0008972487994469702,
-      "learning_rate": 0.00019999997712453053,
-      "loss": 46.0,
-      "step": 2826
-    },
-    {
-      "epoch": 0.21614389204273945,
-      "grad_norm": 0.0010549025610089302,
-      "learning_rate": 0.00019999997710828086,
-      "loss": 46.0,
-      "step": 2827
-    },
-    {
-      "epoch": 0.21622034902612916,
-      "grad_norm": 0.0015169063117355108,
-      "learning_rate": 0.00019999997709202542,
-      "loss": 46.0,
-      "step": 2828
-    },
-    {
-      "epoch": 0.2162968060095189,
-      "grad_norm": 0.0009279120131395757,
-      "learning_rate": 0.0001999999770757642,
-      "loss": 46.0,
-      "step": 2829
-    },
-    {
-      "epoch": 0.21637326299290863,
-      "grad_norm": 0.0007443832000717521,
-      "learning_rate": 0.00019999997705949724,
-      "loss": 46.0,
-      "step": 2830
-    },
-    {
-      "epoch": 0.21644971997629833,
-      "grad_norm": 0.0011436827480793,
-      "learning_rate": 0.0001999999770432245,
-      "loss": 46.0,
-      "step": 2831
-    },
-    {
-      "epoch": 0.21652617695968807,
-      "grad_norm": 0.0015338954981416464,
-      "learning_rate": 0.00019999997702694597,
-      "loss": 46.0,
-      "step": 2832
-    },
-    {
-      "epoch": 0.21660263394307777,
-      "grad_norm": 0.000506790354847908,
-      "learning_rate": 0.0001999999770106617,
-      "loss": 46.0,
-      "step": 2833
-    },
-    {
-      "epoch": 0.2166790909264675,
-      "grad_norm": 0.0006653231102973223,
-      "learning_rate": 0.00019999997699437166,
-      "loss": 46.0,
-      "step": 2834
-    },
-    {
-      "epoch": 0.2167555479098572,
-      "grad_norm": 0.0011177490232512355,
-      "learning_rate": 0.00019999997697807583,
-      "loss": 46.0,
-      "step": 2835
-    },
-    {
-      "epoch": 0.21683200489324694,
-      "grad_norm": 0.0007280634599737823,
-      "learning_rate": 0.00019999997696177425,
-      "loss": 46.0,
-      "step": 2836
-    },
-    {
-      "epoch": 0.21690846187663665,
-      "grad_norm": 0.007233390584588051,
-      "learning_rate": 0.00019999997694546687,
-      "loss": 46.0,
-      "step": 2837
-    },
-    {
-      "epoch": 0.21698491886002638,
-      "grad_norm": 0.0009809202747419477,
-      "learning_rate": 0.00019999997692915375,
-      "loss": 46.0,
-      "step": 2838
-    },
-    {
-      "epoch": 0.2170613758434161,
-      "grad_norm": 0.0008100091945379972,
-      "learning_rate": 0.00019999997691283485,
-      "loss": 46.0,
-      "step": 2839
-    },
-    {
-      "epoch": 0.21713783282680582,
-      "grad_norm": 0.00035515156923793256,
-      "learning_rate": 0.00019999997689651019,
-      "loss": 46.0,
-      "step": 2840
-    },
-    {
-      "epoch": 0.21721428981019553,
-      "grad_norm": 0.0005504949949681759,
-      "learning_rate": 0.00019999997688017974,
-      "loss": 46.0,
-      "step": 2841
-    },
-    {
-      "epoch": 0.21729074679358526,
-      "grad_norm": 0.0009517159196548164,
-      "learning_rate": 0.00019999997686384353,
-      "loss": 46.0,
-      "step": 2842
-    },
-    {
-      "epoch": 0.217367203776975,
-      "grad_norm": 0.0021004853770136833,
-      "learning_rate": 0.00019999997684750156,
-      "loss": 46.0,
-      "step": 2843
-    },
-    {
-      "epoch": 0.2174436607603647,
-      "grad_norm": 0.003291824832558632,
-      "learning_rate": 0.0001999999768311538,
-      "loss": 46.0,
-      "step": 2844
-    },
-    {
-      "epoch": 0.21752011774375443,
-      "grad_norm": 0.002038466976955533,
-      "learning_rate": 0.0001999999768148003,
-      "loss": 46.0,
-      "step": 2845
-    },
-    {
-      "epoch": 0.21759657472714414,
-      "grad_norm": 0.00028456581640057266,
-      "learning_rate": 0.000199999976798441,
-      "loss": 46.0,
-      "step": 2846
-    },
-    {
-      "epoch": 0.21767303171053387,
-      "grad_norm": 0.0016658210661262274,
-      "learning_rate": 0.00019999997678207593,
-      "loss": 46.0,
-      "step": 2847
-    },
-    {
-      "epoch": 0.21774948869392358,
-      "grad_norm": 0.00036686030216515064,
-      "learning_rate": 0.00019999997676570512,
-      "loss": 46.0,
-      "step": 2848
-    },
-    {
-      "epoch": 0.2178259456773133,
-      "grad_norm": 0.0020448833238333464,
-      "learning_rate": 0.00019999997674932852,
-      "loss": 46.0,
-      "step": 2849
-    },
-    {
-      "epoch": 0.21790240266070302,
-      "grad_norm": 0.001440804684534669,
-      "learning_rate": 0.00019999997673294617,
-      "loss": 46.0,
-      "step": 2850
-    },
-    {
-      "epoch": 0.21797885964409275,
-      "grad_norm": 0.0030311725568026304,
-      "learning_rate": 0.00019999997671655802,
-      "loss": 46.0,
-      "step": 2851
-    },
-    {
-      "epoch": 0.21805531662748245,
-      "grad_norm": 0.0007056264439597726,
-      "learning_rate": 0.00019999997670016413,
-      "loss": 46.0,
-      "step": 2852
-    },
-    {
-      "epoch": 0.2181317736108722,
-      "grad_norm": 0.002115414012223482,
-      "learning_rate": 0.00019999997668376446,
-      "loss": 46.0,
-      "step": 2853
-    },
-    {
-      "epoch": 0.2182082305942619,
-      "grad_norm": 0.000498169451020658,
-      "learning_rate": 0.00019999997666735901,
-      "loss": 46.0,
-      "step": 2854
-    },
-    {
-      "epoch": 0.21828468757765163,
-      "grad_norm": 0.002279487904161215,
-      "learning_rate": 0.0001999999766509478,
-      "loss": 46.0,
-      "step": 2855
-    },
-    {
-      "epoch": 0.21836114456104133,
-      "grad_norm": 0.0026159349363297224,
-      "learning_rate": 0.0001999999766345308,
-      "loss": 46.0,
-      "step": 2856
-    },
-    {
-      "epoch": 0.21843760154443106,
-      "grad_norm": 0.0016177184879779816,
-      "learning_rate": 0.00019999997661810807,
-      "loss": 46.0,
-      "step": 2857
-    },
-    {
-      "epoch": 0.2185140585278208,
-      "grad_norm": 0.0008091639028862119,
-      "learning_rate": 0.00019999997660167954,
-      "loss": 46.0,
-      "step": 2858
-    },
-    {
-      "epoch": 0.2185905155112105,
-      "grad_norm": 0.0008049023454077542,
-      "learning_rate": 0.00019999997658524526,
-      "loss": 46.0,
-      "step": 2859
-    },
-    {
-      "epoch": 0.21866697249460024,
-      "grad_norm": 0.002862535184249282,
-      "learning_rate": 0.0001999999765688052,
-      "loss": 46.0,
-      "step": 2860
-    },
-    {
-      "epoch": 0.21874342947798994,
-      "grad_norm": 0.0019411144312471151,
-      "learning_rate": 0.00019999997655235937,
-      "loss": 46.0,
-      "step": 2861
-    },
-    {
-      "epoch": 0.21881988646137968,
-      "grad_norm": 0.0014282979536801577,
-      "learning_rate": 0.0001999999765359078,
-      "loss": 46.0,
-      "step": 2862
-    },
-    {
-      "epoch": 0.21889634344476938,
-      "grad_norm": 0.0007232266943901777,
-      "learning_rate": 0.00019999997651945042,
-      "loss": 46.0,
-      "step": 2863
-    },
-    {
-      "epoch": 0.21897280042815911,
-      "grad_norm": 0.001601203577592969,
-      "learning_rate": 0.00019999997650298727,
-      "loss": 46.0,
-      "step": 2864
-    },
-    {
-      "epoch": 0.21904925741154882,
-      "grad_norm": 0.0010538886999711394,
-      "learning_rate": 0.00019999997648651835,
-      "loss": 46.0,
-      "step": 2865
-    },
-    {
-      "epoch": 0.21912571439493855,
-      "grad_norm": 0.004148329608142376,
-      "learning_rate": 0.0001999999764700437,
-      "loss": 46.0,
-      "step": 2866
-    },
-    {
-      "epoch": 0.21920217137832826,
-      "grad_norm": 0.0009113341220654547,
-      "learning_rate": 0.00019999997645356326,
-      "loss": 46.0,
-      "step": 2867
-    },
-    {
-      "epoch": 0.219278628361718,
-      "grad_norm": 0.0017885052366182208,
-      "learning_rate": 0.00019999997643707705,
-      "loss": 46.0,
-      "step": 2868
-    },
-    {
-      "epoch": 0.2193550853451077,
-      "grad_norm": 0.0013525685062631965,
-      "learning_rate": 0.00019999997642058506,
-      "loss": 46.0,
-      "step": 2869
-    },
-    {
-      "epoch": 0.21943154232849743,
-      "grad_norm": 0.002189547521993518,
-      "learning_rate": 0.0001999999764040873,
-      "loss": 46.0,
-      "step": 2870
-    },
-    {
-      "epoch": 0.21950799931188714,
-      "grad_norm": 0.0014040491078048944,
-      "learning_rate": 0.00019999997638758376,
-      "loss": 46.0,
-      "step": 2871
-    },
-    {
-      "epoch": 0.21958445629527687,
-      "grad_norm": 0.0012876776745542884,
-      "learning_rate": 0.00019999997637107448,
-      "loss": 46.0,
-      "step": 2872
-    },
-    {
-      "epoch": 0.2196609132786666,
-      "grad_norm": 0.0038134800270199776,
-      "learning_rate": 0.00019999997635455942,
-      "loss": 46.0,
-      "step": 2873
-    },
-    {
-      "epoch": 0.2197373702620563,
-      "grad_norm": 0.00109303358476609,
-      "learning_rate": 0.0001999999763380386,
-      "loss": 46.0,
-      "step": 2874
-    },
-    {
-      "epoch": 0.21981382724544604,
-      "grad_norm": 0.001133384765125811,
-      "learning_rate": 0.000199999976321512,
-      "loss": 46.0,
-      "step": 2875
-    },
-    {
-      "epoch": 0.21989028422883575,
-      "grad_norm": 0.0010685939341783524,
-      "learning_rate": 0.00019999997630497962,
-      "loss": 46.0,
-      "step": 2876
-    },
-    {
-      "epoch": 0.21996674121222548,
-      "grad_norm": 0.0007618262316100299,
-      "learning_rate": 0.00019999997628844147,
-      "loss": 46.0,
-      "step": 2877
-    },
-    {
-      "epoch": 0.22004319819561519,
-      "grad_norm": 0.0007574647897854447,
-      "learning_rate": 0.0001999999762718976,
-      "loss": 46.0,
-      "step": 2878
-    },
-    {
-      "epoch": 0.22011965517900492,
-      "grad_norm": 0.0008021695539355278,
-      "learning_rate": 0.0001999999762553479,
-      "loss": 46.0,
-      "step": 2879
-    },
-    {
-      "epoch": 0.22019611216239462,
-      "grad_norm": 0.007086888886988163,
-      "learning_rate": 0.00019999997623879246,
-      "loss": 46.0,
-      "step": 2880
-    },
-    {
-      "epoch": 0.22027256914578436,
-      "grad_norm": 0.0032493509352207184,
-      "learning_rate": 0.00019999997622223122,
-      "loss": 46.0,
-      "step": 2881
-    },
-    {
-      "epoch": 0.22034902612917406,
-      "grad_norm": 0.0014883350813761353,
-      "learning_rate": 0.00019999997620566426,
-      "loss": 46.0,
-      "step": 2882
-    },
-    {
-      "epoch": 0.2204254831125638,
-      "grad_norm": 0.0011161850998178124,
-      "learning_rate": 0.0001999999761890915,
-      "loss": 46.0,
-      "step": 2883
-    },
-    {
-      "epoch": 0.2205019400959535,
-      "grad_norm": 0.001158189377747476,
-      "learning_rate": 0.00019999997617251296,
-      "loss": 46.0,
-      "step": 2884
-    },
-    {
-      "epoch": 0.22057839707934324,
-      "grad_norm": 0.003696287516504526,
-      "learning_rate": 0.00019999997615592868,
-      "loss": 46.0,
-      "step": 2885
-    },
-    {
-      "epoch": 0.22065485406273294,
-      "grad_norm": 0.002970096655189991,
-      "learning_rate": 0.0001999999761393386,
-      "loss": 46.0,
-      "step": 2886
-    },
-    {
-      "epoch": 0.22073131104612267,
-      "grad_norm": 0.0013597555225715041,
-      "learning_rate": 0.00019999997612274277,
-      "loss": 46.0,
-      "step": 2887
-    },
-    {
-      "epoch": 0.2208077680295124,
-      "grad_norm": 0.0021887505427002907,
-      "learning_rate": 0.00019999997610614117,
-      "loss": 46.0,
-      "step": 2888
-    },
-    {
-      "epoch": 0.2208842250129021,
-      "grad_norm": 0.0020471117459237576,
-      "learning_rate": 0.0001999999760895338,
-      "loss": 46.0,
-      "step": 2889
-    },
-    {
-      "epoch": 0.22096068199629185,
-      "grad_norm": 0.003169095842167735,
-      "learning_rate": 0.00019999997607292067,
-      "loss": 46.0,
-      "step": 2890
-    },
-    {
-      "epoch": 0.22103713897968155,
-      "grad_norm": 0.0011828108690679073,
-      "learning_rate": 0.00019999997605630175,
-      "loss": 46.0,
-      "step": 2891
-    },
-    {
-      "epoch": 0.22111359596307129,
-      "grad_norm": 0.0012258938513696194,
-      "learning_rate": 0.00019999997603967705,
-      "loss": 46.0,
-      "step": 2892
-    },
-    {
-      "epoch": 0.221190052946461,
-      "grad_norm": 0.0005020485841669142,
-      "learning_rate": 0.0001999999760230466,
-      "loss": 46.0,
-      "step": 2893
-    },
-    {
-      "epoch": 0.22126650992985072,
-      "grad_norm": 0.002062303712591529,
-      "learning_rate": 0.0001999999760064104,
-      "loss": 46.0,
-      "step": 2894
-    },
-    {
-      "epoch": 0.22134296691324043,
-      "grad_norm": 0.001256912131793797,
-      "learning_rate": 0.0001999999759897684,
-      "loss": 46.0,
-      "step": 2895
-    },
-    {
-      "epoch": 0.22141942389663016,
-      "grad_norm": 0.0018735273042693734,
-      "learning_rate": 0.00019999997597312065,
-      "loss": 46.0,
-      "step": 2896
-    },
-    {
-      "epoch": 0.22149588088001987,
-      "grad_norm": 0.0005598751595243812,
-      "learning_rate": 0.0001999999759564671,
-      "loss": 46.0,
-      "step": 2897
-    },
-    {
-      "epoch": 0.2215723378634096,
-      "grad_norm": 0.0008617884013801813,
-      "learning_rate": 0.0001999999759398078,
-      "loss": 46.0,
-      "step": 2898
-    },
-    {
-      "epoch": 0.2216487948467993,
-      "grad_norm": 0.0007058823830448091,
-      "learning_rate": 0.00019999997592314275,
-      "loss": 46.0,
-      "step": 2899
-    },
-    {
-      "epoch": 0.22172525183018904,
-      "grad_norm": 0.0025898173917084932,
-      "learning_rate": 0.0001999999759064719,
-      "loss": 46.0,
-      "step": 2900
-    },
-    {
-      "epoch": 0.22180170881357877,
-      "grad_norm": 0.000993963680230081,
-      "learning_rate": 0.00019999997588979532,
-      "loss": 46.0,
-      "step": 2901
-    },
-    {
-      "epoch": 0.22187816579696848,
-      "grad_norm": 0.001138947787694633,
-      "learning_rate": 0.00019999997587311294,
-      "loss": 46.0,
-      "step": 2902
-    },
-    {
-      "epoch": 0.2219546227803582,
-      "grad_norm": 0.0008392338640987873,
-      "learning_rate": 0.00019999997585642477,
-      "loss": 46.0,
-      "step": 2903
-    },
-    {
-      "epoch": 0.22203107976374792,
-      "grad_norm": 0.0015987890074029565,
-      "learning_rate": 0.0001999999758397309,
-      "loss": 46.0,
-      "step": 2904
-    },
-    {
-      "epoch": 0.22210753674713765,
-      "grad_norm": 0.002902098698541522,
-      "learning_rate": 0.00019999997582303118,
-      "loss": 46.0,
-      "step": 2905
-    },
-    {
-      "epoch": 0.22218399373052736,
-      "grad_norm": 0.0004394171410240233,
-      "learning_rate": 0.00019999997580632574,
-      "loss": 46.0,
-      "step": 2906
-    },
-    {
-      "epoch": 0.2222604507139171,
-      "grad_norm": 0.0012952984543517232,
-      "learning_rate": 0.00019999997578961452,
-      "loss": 46.0,
-      "step": 2907
-    },
-    {
-      "epoch": 0.2223369076973068,
-      "grad_norm": 0.0004091626906301826,
-      "learning_rate": 0.00019999997577289753,
-      "loss": 46.0,
-      "step": 2908
-    },
-    {
-      "epoch": 0.22241336468069653,
-      "grad_norm": 0.0013748976634815335,
-      "learning_rate": 0.00019999997575617475,
-      "loss": 46.0,
-      "step": 2909
-    },
-    {
-      "epoch": 0.22248982166408623,
-      "grad_norm": 0.0016329112695530057,
-      "learning_rate": 0.0001999999757394462,
-      "loss": 46.0,
-      "step": 2910
-    },
-    {
-      "epoch": 0.22256627864747597,
-      "grad_norm": 0.0006410747300833464,
-      "learning_rate": 0.00019999997572271193,
-      "loss": 46.0,
-      "step": 2911
-    },
-    {
-      "epoch": 0.22264273563086567,
-      "grad_norm": 0.001577257877215743,
-      "learning_rate": 0.00019999997570597188,
-      "loss": 46.0,
-      "step": 2912
-    },
-    {
-      "epoch": 0.2227191926142554,
-      "grad_norm": 0.0024543842300772667,
-      "learning_rate": 0.00019999997568922605,
-      "loss": 46.0,
-      "step": 2913
-    },
-    {
-      "epoch": 0.2227956495976451,
-      "grad_norm": 0.0006878508138470352,
-      "learning_rate": 0.00019999997567247442,
-      "loss": 46.0,
-      "step": 2914
-    },
-    {
-      "epoch": 0.22287210658103485,
-      "grad_norm": 0.0017850800650194287,
-      "learning_rate": 0.00019999997565571705,
-      "loss": 46.0,
-      "step": 2915
-    },
-    {
-      "epoch": 0.22294856356442458,
-      "grad_norm": 0.0016517764888703823,
-      "learning_rate": 0.0001999999756389539,
-      "loss": 46.0,
-      "step": 2916
-    },
-    {
-      "epoch": 0.22302502054781428,
-      "grad_norm": 0.0010513722663745284,
-      "learning_rate": 0.00019999997562218498,
-      "loss": 46.0,
-      "step": 2917
-    },
-    {
-      "epoch": 0.22310147753120402,
-      "grad_norm": 0.0009467603522352874,
-      "learning_rate": 0.00019999997560541028,
-      "loss": 46.0,
-      "step": 2918
-    },
-    {
-      "epoch": 0.22317793451459372,
-      "grad_norm": 0.0006628857227042317,
-      "learning_rate": 0.00019999997558862984,
-      "loss": 46.0,
-      "step": 2919
-    },
-    {
-      "epoch": 0.22325439149798346,
-      "grad_norm": 0.0026683774776756763,
-      "learning_rate": 0.0001999999755718436,
-      "loss": 46.0,
-      "step": 2920
-    },
-    {
-      "epoch": 0.22333084848137316,
-      "grad_norm": 0.0006932035321369767,
-      "learning_rate": 0.0001999999755550516,
-      "loss": 46.0,
-      "step": 2921
-    },
-    {
-      "epoch": 0.2234073054647629,
-      "grad_norm": 0.0029748145025223494,
-      "learning_rate": 0.00019999997553825385,
-      "loss": 46.0,
-      "step": 2922
-    },
-    {
-      "epoch": 0.2234837624481526,
-      "grad_norm": 0.0007320643635466695,
-      "learning_rate": 0.00019999997552145032,
-      "loss": 46.0,
-      "step": 2923
-    },
-    {
-      "epoch": 0.22356021943154233,
-      "grad_norm": 0.0020910215098410845,
-      "learning_rate": 0.00019999997550464103,
-      "loss": 46.0,
-      "step": 2924
-    },
-    {
-      "epoch": 0.22363667641493204,
-      "grad_norm": 0.0012187798274680972,
-      "learning_rate": 0.00019999997548782595,
-      "loss": 46.0,
-      "step": 2925
-    },
-    {
-      "epoch": 0.22371313339832177,
-      "grad_norm": 0.000737123133148998,
-      "learning_rate": 0.00019999997547100507,
-      "loss": 46.0,
-      "step": 2926
-    },
-    {
-      "epoch": 0.22378959038171148,
-      "grad_norm": 0.0008600133005529642,
-      "learning_rate": 0.00019999997545417847,
-      "loss": 46.0,
-      "step": 2927
-    },
-    {
-      "epoch": 0.2238660473651012,
-      "grad_norm": 0.0009574157302267849,
-      "learning_rate": 0.0001999999754373461,
-      "loss": 46.0,
-      "step": 2928
-    },
-    {
-      "epoch": 0.22394250434849092,
-      "grad_norm": 0.0007718966226093471,
-      "learning_rate": 0.00019999997542050795,
-      "loss": 46.0,
-      "step": 2929
-    },
-    {
-      "epoch": 0.22401896133188065,
-      "grad_norm": 0.000957387441303581,
-      "learning_rate": 0.00019999997540366403,
-      "loss": 46.0,
-      "step": 2930
-    },
-    {
-      "epoch": 0.22409541831527038,
-      "grad_norm": 0.0014276178553700447,
-      "learning_rate": 0.00019999997538681433,
-      "loss": 46.0,
-      "step": 2931
-    },
-    {
-      "epoch": 0.2241718752986601,
-      "grad_norm": 0.0013034959556534886,
-      "learning_rate": 0.0001999999753699589,
-      "loss": 46.0,
-      "step": 2932
-    },
-    {
-      "epoch": 0.22424833228204982,
-      "grad_norm": 0.0037667297292500734,
-      "learning_rate": 0.00019999997535309768,
-      "loss": 46.0,
-      "step": 2933
-    },
-    {
-      "epoch": 0.22432478926543953,
-      "grad_norm": 0.00153718376532197,
-      "learning_rate": 0.00019999997533623064,
-      "loss": 46.0,
-      "step": 2934
-    },
-    {
-      "epoch": 0.22440124624882926,
-      "grad_norm": 0.0014878826914355159,
-      "learning_rate": 0.0001999999753193579,
-      "loss": 46.0,
-      "step": 2935
-    },
-    {
-      "epoch": 0.22447770323221897,
-      "grad_norm": 0.005342686548829079,
-      "learning_rate": 0.00019999997530247937,
-      "loss": 46.0,
-      "step": 2936
-    },
-    {
-      "epoch": 0.2245541602156087,
-      "grad_norm": 0.0007300264551304281,
-      "learning_rate": 0.00019999997528559504,
-      "loss": 46.0,
-      "step": 2937
-    },
-    {
-      "epoch": 0.2246306171989984,
-      "grad_norm": 0.006637099198997021,
-      "learning_rate": 0.00019999997526870495,
-      "loss": 46.0,
-      "step": 2938
-    },
-    {
-      "epoch": 0.22470707418238814,
-      "grad_norm": 0.001090914011001587,
-      "learning_rate": 0.00019999997525180913,
-      "loss": 46.0,
-      "step": 2939
-    },
-    {
-      "epoch": 0.22478353116577784,
-      "grad_norm": 0.0008396320044994354,
-      "learning_rate": 0.0001999999752349075,
-      "loss": 46.0,
-      "step": 2940
-    },
-    {
-      "epoch": 0.22485998814916758,
-      "grad_norm": 0.0006463932804763317,
-      "learning_rate": 0.00019999997521800013,
-      "loss": 46.0,
-      "step": 2941
-    },
-    {
-      "epoch": 0.22493644513255728,
-      "grad_norm": 0.0005670917453244328,
-      "learning_rate": 0.00019999997520108698,
-      "loss": 46.0,
-      "step": 2942
-    },
-    {
-      "epoch": 0.22501290211594702,
-      "grad_norm": 0.0013487263349816203,
-      "learning_rate": 0.00019999997518416803,
-      "loss": 46.0,
-      "step": 2943
-    },
-    {
-      "epoch": 0.22508935909933672,
-      "grad_norm": 0.0013837008737027645,
-      "learning_rate": 0.00019999997516724336,
-      "loss": 46.0,
-      "step": 2944
-    },
-    {
-      "epoch": 0.22516581608272646,
-      "grad_norm": 0.003335876390337944,
-      "learning_rate": 0.0001999999751503129,
-      "loss": 46.0,
-      "step": 2945
-    },
-    {
-      "epoch": 0.2252422730661162,
-      "grad_norm": 0.0010610654717311263,
-      "learning_rate": 0.00019999997513337668,
-      "loss": 46.0,
-      "step": 2946
-    },
-    {
-      "epoch": 0.2253187300495059,
-      "grad_norm": 0.0017328435787931085,
-      "learning_rate": 0.00019999997511643467,
-      "loss": 46.0,
-      "step": 2947
-    },
-    {
-      "epoch": 0.22539518703289563,
-      "grad_norm": 0.0016389141092076898,
-      "learning_rate": 0.00019999997509948688,
-      "loss": 46.0,
-      "step": 2948
-    },
-    {
-      "epoch": 0.22547164401628533,
-      "grad_norm": 0.001042242394760251,
-      "learning_rate": 0.00019999997508253335,
-      "loss": 46.0,
-      "step": 2949
-    },
-    {
-      "epoch": 0.22554810099967507,
-      "grad_norm": 0.0008065119036473334,
-      "learning_rate": 0.00019999997506557404,
-      "loss": 46.0,
-      "step": 2950
-    },
-    {
-      "epoch": 0.22562455798306477,
-      "grad_norm": 0.004538060165941715,
-      "learning_rate": 0.00019999997504860896,
-      "loss": 46.0,
-      "step": 2951
-    },
-    {
-      "epoch": 0.2257010149664545,
-      "grad_norm": 0.0008074028301052749,
-      "learning_rate": 0.00019999997503163813,
-      "loss": 46.0,
-      "step": 2952
-    },
-    {
-      "epoch": 0.2257774719498442,
-      "grad_norm": 0.0028297309763729572,
-      "learning_rate": 0.0001999999750146615,
-      "loss": 46.0,
-      "step": 2953
-    },
-    {
-      "epoch": 0.22585392893323394,
-      "grad_norm": 0.009999656118452549,
-      "learning_rate": 0.0001999999749976791,
-      "loss": 46.0,
-      "step": 2954
-    },
-    {
-      "epoch": 0.22593038591662365,
-      "grad_norm": 0.0006880185683257878,
-      "learning_rate": 0.00019999997498069093,
-      "loss": 46.0,
-      "step": 2955
-    },
-    {
-      "epoch": 0.22600684290001338,
-      "grad_norm": 0.0012339602690190077,
-      "learning_rate": 0.00019999997496369704,
-      "loss": 46.0,
-      "step": 2956
-    },
-    {
-      "epoch": 0.2260832998834031,
-      "grad_norm": 0.0008526530582457781,
-      "learning_rate": 0.00019999997494669734,
-      "loss": 46.0,
-      "step": 2957
-    },
-    {
-      "epoch": 0.22615975686679282,
-      "grad_norm": 0.0011052998015657067,
-      "learning_rate": 0.00019999997492969188,
-      "loss": 46.0,
-      "step": 2958
-    },
-    {
-      "epoch": 0.22623621385018255,
-      "grad_norm": 0.0033115132246166468,
-      "learning_rate": 0.00019999997491268063,
-      "loss": 46.0,
-      "step": 2959
-    },
-    {
-      "epoch": 0.22631267083357226,
-      "grad_norm": 0.00083912443369627,
-      "learning_rate": 0.00019999997489566362,
-      "loss": 46.0,
-      "step": 2960
-    },
-    {
-      "epoch": 0.226389127816962,
-      "grad_norm": 0.000600250088609755,
-      "learning_rate": 0.00019999997487864086,
-      "loss": 46.0,
-      "step": 2961
-    },
-    {
-      "epoch": 0.2264655848003517,
-      "grad_norm": 0.0012474660761654377,
-      "learning_rate": 0.00019999997486161233,
-      "loss": 46.0,
-      "step": 2962
-    },
-    {
-      "epoch": 0.22654204178374143,
-      "grad_norm": 0.000983542762696743,
-      "learning_rate": 0.000199999974844578,
-      "loss": 46.0,
-      "step": 2963
-    },
-    {
-      "epoch": 0.22661849876713114,
-      "grad_norm": 0.0013953946763649583,
-      "learning_rate": 0.00019999997482753794,
-      "loss": 46.0,
-      "step": 2964
-    },
-    {
-      "epoch": 0.22669495575052087,
-      "grad_norm": 0.0009575795847922564,
-      "learning_rate": 0.00019999997481049206,
-      "loss": 46.0,
-      "step": 2965
-    },
-    {
-      "epoch": 0.22677141273391058,
-      "grad_norm": 0.0009301762329414487,
-      "learning_rate": 0.00019999997479344046,
-      "loss": 46.0,
-      "step": 2966
-    },
-    {
-      "epoch": 0.2268478697173003,
-      "grad_norm": 0.0007534794858656824,
-      "learning_rate": 0.00019999997477638306,
-      "loss": 46.0,
-      "step": 2967
-    },
-    {
-      "epoch": 0.22692432670069002,
-      "grad_norm": 0.0005373649764806032,
-      "learning_rate": 0.0001999999747593199,
-      "loss": 46.0,
-      "step": 2968
-    },
-    {
-      "epoch": 0.22700078368407975,
-      "grad_norm": 0.0007423251518048346,
-      "learning_rate": 0.00019999997474225097,
-      "loss": 46.0,
-      "step": 2969
-    },
-    {
-      "epoch": 0.22707724066746945,
-      "grad_norm": 0.0018095847917720675,
-      "learning_rate": 0.00019999997472517627,
-      "loss": 46.0,
-      "step": 2970
-    },
-    {
-      "epoch": 0.2271536976508592,
-      "grad_norm": 0.0008475442882627249,
-      "learning_rate": 0.0001999999747080958,
-      "loss": 46.0,
-      "step": 2971
-    },
-    {
-      "epoch": 0.2272301546342489,
-      "grad_norm": 0.0014801191864535213,
-      "learning_rate": 0.00019999997469100956,
-      "loss": 46.0,
-      "step": 2972
-    },
-    {
-      "epoch": 0.22730661161763863,
-      "grad_norm": 0.013205279596149921,
-      "learning_rate": 0.00019999997467391758,
-      "loss": 46.0,
-      "step": 2973
-    },
-    {
-      "epoch": 0.22738306860102836,
-      "grad_norm": 0.0015982815530151129,
-      "learning_rate": 0.0001999999746568198,
-      "loss": 46.0,
-      "step": 2974
-    },
-    {
-      "epoch": 0.22745952558441807,
-      "grad_norm": 0.002632777439430356,
-      "learning_rate": 0.00019999997463971626,
-      "loss": 46.0,
-      "step": 2975
-    },
-    {
-      "epoch": 0.2275359825678078,
-      "grad_norm": 0.0010471882997080684,
-      "learning_rate": 0.00019999997462260692,
-      "loss": 46.0,
-      "step": 2976
-    },
-    {
-      "epoch": 0.2276124395511975,
-      "grad_norm": 0.0012914057588204741,
-      "learning_rate": 0.00019999997460549184,
-      "loss": 46.0,
-      "step": 2977
-    },
-    {
-      "epoch": 0.22768889653458724,
-      "grad_norm": 0.0010279935086145997,
-      "learning_rate": 0.000199999974588371,
-      "loss": 46.0,
-      "step": 2978
-    },
-    {
-      "epoch": 0.22776535351797694,
-      "grad_norm": 0.000953810871578753,
-      "learning_rate": 0.00019999997457124436,
-      "loss": 46.0,
-      "step": 2979
-    },
-    {
-      "epoch": 0.22784181050136668,
-      "grad_norm": 0.0016735795652493834,
-      "learning_rate": 0.000199999974554112,
-      "loss": 46.0,
-      "step": 2980
-    },
-    {
-      "epoch": 0.22791826748475638,
-      "grad_norm": 0.0007070382707752287,
-      "learning_rate": 0.00019999997453697382,
-      "loss": 46.0,
-      "step": 2981
-    },
-    {
-      "epoch": 0.22799472446814611,
-      "grad_norm": 0.0009227624395862222,
-      "learning_rate": 0.00019999997451982987,
-      "loss": 46.0,
-      "step": 2982
-    },
-    {
-      "epoch": 0.22807118145153582,
-      "grad_norm": 0.0018291121814399958,
-      "learning_rate": 0.00019999997450268018,
-      "loss": 46.0,
-      "step": 2983
-    },
-    {
-      "epoch": 0.22814763843492555,
-      "grad_norm": 0.0007183416746556759,
-      "learning_rate": 0.0001999999744855247,
-      "loss": 46.0,
-      "step": 2984
-    },
-    {
-      "epoch": 0.22822409541831526,
-      "grad_norm": 0.0008718360331840813,
-      "learning_rate": 0.00019999997446836347,
-      "loss": 46.0,
-      "step": 2985
-    },
-    {
-      "epoch": 0.228300552401705,
-      "grad_norm": 0.0011529215844348073,
-      "learning_rate": 0.00019999997445119646,
-      "loss": 46.0,
-      "step": 2986
-    },
-    {
-      "epoch": 0.2283770093850947,
-      "grad_norm": 0.001746269059367478,
-      "learning_rate": 0.0001999999744340237,
-      "loss": 46.0,
-      "step": 2987
-    },
-    {
-      "epoch": 0.22845346636848443,
-      "grad_norm": 0.002684320555999875,
-      "learning_rate": 0.00019999997441684514,
-      "loss": 46.0,
-      "step": 2988
-    },
-    {
-      "epoch": 0.22852992335187416,
-      "grad_norm": 0.0008634284022264183,
-      "learning_rate": 0.00019999997439966083,
-      "loss": 46.0,
-      "step": 2989
-    },
-    {
-      "epoch": 0.22860638033526387,
-      "grad_norm": 0.0011859749210998416,
-      "learning_rate": 0.0001999999743824707,
-      "loss": 46.0,
-      "step": 2990
-    },
-    {
-      "epoch": 0.2286828373186536,
-      "grad_norm": 0.00141819822601974,
-      "learning_rate": 0.00019999997436527487,
-      "loss": 46.0,
-      "step": 2991
-    },
-    {
-      "epoch": 0.2287592943020433,
-      "grad_norm": 0.000725681777112186,
-      "learning_rate": 0.00019999997434807325,
-      "loss": 46.0,
-      "step": 2992
-    },
-    {
-      "epoch": 0.22883575128543304,
-      "grad_norm": 0.00043696275679394603,
-      "learning_rate": 0.00019999997433086585,
-      "loss": 46.0,
-      "step": 2993
-    },
-    {
-      "epoch": 0.22891220826882275,
-      "grad_norm": 0.002142752753570676,
-      "learning_rate": 0.0001999999743136527,
-      "loss": 46.0,
-      "step": 2994
-    },
-    {
-      "epoch": 0.22898866525221248,
-      "grad_norm": 0.00040243452531285584,
-      "learning_rate": 0.00019999997429643375,
-      "loss": 46.0,
-      "step": 2995
-    },
-    {
-      "epoch": 0.2290651222356022,
-      "grad_norm": 0.0006108229281380773,
-      "learning_rate": 0.00019999997427920903,
-      "loss": 46.0,
-      "step": 2996
-    },
-    {
-      "epoch": 0.22914157921899192,
-      "grad_norm": 0.0014221464516595006,
-      "learning_rate": 0.00019999997426197857,
-      "loss": 46.0,
-      "step": 2997
-    },
-    {
-      "epoch": 0.22921803620238163,
-      "grad_norm": 0.0011005630949512124,
-      "learning_rate": 0.00019999997424474233,
-      "loss": 46.0,
-      "step": 2998
-    },
-    {
-      "epoch": 0.22929449318577136,
-      "grad_norm": 0.0017183477757498622,
-      "learning_rate": 0.00019999997422750032,
-      "loss": 46.0,
-      "step": 2999
-    },
-    {
-      "epoch": 0.22937095016916106,
-      "grad_norm": 0.0008607763447798789,
-      "learning_rate": 0.00019999997421025253,
-      "loss": 46.0,
-      "step": 3000
-    },
-    {
-      "epoch": 0.2294474071525508,
-      "grad_norm": 0.0015489585930481553,
-      "learning_rate": 0.000199999974192999,
-      "loss": 46.0,
-      "step": 3001
-    },
-    {
-      "epoch": 0.2295238641359405,
-      "grad_norm": 0.0014900440583005548,
-      "learning_rate": 0.0001999999741757397,
-      "loss": 46.0,
-      "step": 3002
-    },
-    {
-      "epoch": 0.22960032111933024,
-      "grad_norm": 0.0009447936317883432,
-      "learning_rate": 0.00019999997415847458,
-      "loss": 46.0,
-      "step": 3003
-    },
-    {
-      "epoch": 0.22967677810271997,
-      "grad_norm": 0.0021479162387549877,
-      "learning_rate": 0.0001999999741412037,
-      "loss": 46.0,
-      "step": 3004
-    },
-    {
-      "epoch": 0.22975323508610968,
-      "grad_norm": 0.0021889146883040667,
-      "learning_rate": 0.00019999997412392708,
-      "loss": 46.0,
-      "step": 3005
-    },
-    {
-      "epoch": 0.2298296920694994,
-      "grad_norm": 0.0009473631507717073,
-      "learning_rate": 0.00019999997410664468,
-      "loss": 46.0,
-      "step": 3006
-    },
-    {
-      "epoch": 0.2299061490528891,
-      "grad_norm": 0.002369271358475089,
-      "learning_rate": 0.0001999999740893565,
-      "loss": 46.0,
-      "step": 3007
-    },
-    {
-      "epoch": 0.22998260603627885,
-      "grad_norm": 0.001820671372115612,
-      "learning_rate": 0.00019999997407206256,
-      "loss": 46.0,
-      "step": 3008
-    },
-    {
-      "epoch": 0.23005906301966855,
-      "grad_norm": 0.0010885591618716717,
-      "learning_rate": 0.00019999997405476287,
-      "loss": 46.0,
-      "step": 3009
-    },
-    {
-      "epoch": 0.23013552000305829,
-      "grad_norm": 0.0003443363239057362,
-      "learning_rate": 0.0001999999740374574,
-      "loss": 46.0,
-      "step": 3010
-    },
-    {
-      "epoch": 0.230211976986448,
-      "grad_norm": 0.0011062311241403222,
-      "learning_rate": 0.00019999997402014613,
-      "loss": 46.0,
-      "step": 3011
-    },
-    {
-      "epoch": 0.23028843396983772,
-      "grad_norm": 0.0027726131957024336,
-      "learning_rate": 0.00019999997400282915,
-      "loss": 46.0,
-      "step": 3012
-    },
-    {
-      "epoch": 0.23036489095322743,
-      "grad_norm": 0.001121298409998417,
-      "learning_rate": 0.00019999997398550634,
-      "loss": 46.0,
-      "step": 3013
-    },
-    {
-      "epoch": 0.23044134793661716,
-      "grad_norm": 0.0020850275177508593,
-      "learning_rate": 0.0001999999739681778,
-      "loss": 46.0,
-      "step": 3014
-    },
-    {
-      "epoch": 0.23051780492000687,
-      "grad_norm": 0.00258211069740355,
-      "learning_rate": 0.0001999999739508435,
-      "loss": 46.0,
-      "step": 3015
-    },
-    {
-      "epoch": 0.2305942619033966,
-      "grad_norm": 0.0010522013762965798,
-      "learning_rate": 0.0001999999739335034,
-      "loss": 46.0,
-      "step": 3016
-    },
-    {
-      "epoch": 0.23067071888678634,
-      "grad_norm": 0.0008258194429799914,
-      "learning_rate": 0.0001999999739161575,
-      "loss": 46.0,
-      "step": 3017
-    },
-    {
-      "epoch": 0.23074717587017604,
-      "grad_norm": 0.0010189589811488986,
-      "learning_rate": 0.0001999999738988059,
-      "loss": 46.0,
-      "step": 3018
-    },
-    {
-      "epoch": 0.23082363285356577,
-      "grad_norm": 0.0010131691815331578,
-      "learning_rate": 0.0001999999738814485,
-      "loss": 46.0,
-      "step": 3019
-    },
-    {
-      "epoch": 0.23090008983695548,
-      "grad_norm": 0.0065146577544510365,
-      "learning_rate": 0.00019999997386408532,
-      "loss": 46.0,
-      "step": 3020
-    },
-    {
-      "epoch": 0.2309765468203452,
-      "grad_norm": 0.0011046944418922067,
-      "learning_rate": 0.00019999997384671637,
-      "loss": 46.0,
-      "step": 3021
-    },
-    {
-      "epoch": 0.23105300380373492,
-      "grad_norm": 0.000710965774487704,
-      "learning_rate": 0.00019999997382934165,
-      "loss": 46.0,
-      "step": 3022
-    },
-    {
-      "epoch": 0.23112946078712465,
-      "grad_norm": 0.0033271685242652893,
-      "learning_rate": 0.00019999997381196118,
-      "loss": 46.0,
-      "step": 3023
-    },
-    {
-      "epoch": 0.23120591777051436,
-      "grad_norm": 0.000987824983894825,
-      "learning_rate": 0.00019999997379457495,
-      "loss": 46.0,
-      "step": 3024
-    },
-    {
-      "epoch": 0.2312823747539041,
-      "grad_norm": 0.0006469162763096392,
-      "learning_rate": 0.00019999997377718293,
-      "loss": 46.0,
-      "step": 3025
-    },
-    {
-      "epoch": 0.2313588317372938,
-      "grad_norm": 0.0033174417912960052,
-      "learning_rate": 0.00019999997375978512,
-      "loss": 46.0,
-      "step": 3026
-    },
-    {
-      "epoch": 0.23143528872068353,
-      "grad_norm": 0.0010534895118325949,
-      "learning_rate": 0.00019999997374238156,
-      "loss": 46.0,
-      "step": 3027
-    },
-    {
-      "epoch": 0.23151174570407324,
-      "grad_norm": 0.0005828223074786365,
-      "learning_rate": 0.00019999997372497226,
-      "loss": 46.0,
-      "step": 3028
-    },
-    {
-      "epoch": 0.23158820268746297,
-      "grad_norm": 0.0008897861116565764,
-      "learning_rate": 0.00019999997370755718,
-      "loss": 46.0,
-      "step": 3029
-    },
-    {
-      "epoch": 0.23166465967085267,
-      "grad_norm": 0.0008952495409175754,
-      "learning_rate": 0.0001999999736901363,
-      "loss": 46.0,
-      "step": 3030
-    },
-    {
-      "epoch": 0.2317411166542424,
-      "grad_norm": 0.0005366086261346936,
-      "learning_rate": 0.00019999997367270967,
-      "loss": 46.0,
-      "step": 3031
-    },
-    {
-      "epoch": 0.23181757363763214,
-      "grad_norm": 0.0013184205163270235,
-      "learning_rate": 0.00019999997365527725,
-      "loss": 46.0,
-      "step": 3032
-    },
-    {
-      "epoch": 0.23189403062102185,
-      "grad_norm": 0.0017260562162846327,
-      "learning_rate": 0.00019999997363783907,
-      "loss": 46.0,
-      "step": 3033
-    },
-    {
-      "epoch": 0.23197048760441158,
-      "grad_norm": 0.002803966635838151,
-      "learning_rate": 0.00019999997362039513,
-      "loss": 46.0,
-      "step": 3034
-    },
-    {
-      "epoch": 0.23204694458780128,
-      "grad_norm": 0.001317357993684709,
-      "learning_rate": 0.0001999999736029454,
-      "loss": 46.0,
-      "step": 3035
-    },
-    {
-      "epoch": 0.23212340157119102,
-      "grad_norm": 0.00045207326184026897,
-      "learning_rate": 0.00019999997358548994,
-      "loss": 46.0,
-      "step": 3036
-    },
-    {
-      "epoch": 0.23219985855458072,
-      "grad_norm": 0.0015102846082299948,
-      "learning_rate": 0.00019999997356802868,
-      "loss": 46.0,
-      "step": 3037
-    },
-    {
-      "epoch": 0.23227631553797046,
-      "grad_norm": 0.0023084136191755533,
-      "learning_rate": 0.0001999999735505617,
-      "loss": 46.0,
-      "step": 3038
-    },
-    {
-      "epoch": 0.23235277252136016,
-      "grad_norm": 0.0022218015510588884,
-      "learning_rate": 0.00019999997353308888,
-      "loss": 46.0,
-      "step": 3039
-    },
-    {
-      "epoch": 0.2324292295047499,
-      "grad_norm": 0.0018917821580544114,
-      "learning_rate": 0.0001999999735156103,
-      "loss": 46.0,
-      "step": 3040
-    },
-    {
-      "epoch": 0.2325056864881396,
-      "grad_norm": 0.0011300478363409638,
-      "learning_rate": 0.000199999973498126,
-      "loss": 46.0,
-      "step": 3041
-    },
-    {
-      "epoch": 0.23258214347152933,
-      "grad_norm": 0.0010690023191273212,
-      "learning_rate": 0.00019999997348063589,
-      "loss": 46.0,
-      "step": 3042
-    },
-    {
-      "epoch": 0.23265860045491904,
-      "grad_norm": 0.0008099564583972096,
-      "learning_rate": 0.00019999997346314,
-      "loss": 46.0,
-      "step": 3043
-    },
-    {
-      "epoch": 0.23273505743830877,
-      "grad_norm": 0.0010447854874655604,
-      "learning_rate": 0.00019999997344563835,
-      "loss": 46.0,
-      "step": 3044
-    },
-    {
-      "epoch": 0.23281151442169848,
-      "grad_norm": 0.002069528680294752,
-      "learning_rate": 0.00019999997342813098,
-      "loss": 46.0,
-      "step": 3045
-    },
-    {
-      "epoch": 0.2328879714050882,
-      "grad_norm": 0.0036557926796376705,
-      "learning_rate": 0.0001999999734106178,
-      "loss": 46.0,
-      "step": 3046
-    },
-    {
-      "epoch": 0.23296442838847795,
-      "grad_norm": 0.0016329380450770259,
-      "learning_rate": 0.00019999997339309884,
-      "loss": 46.0,
-      "step": 3047
-    },
-    {
-      "epoch": 0.23304088537186765,
-      "grad_norm": 0.0015899931313470006,
-      "learning_rate": 0.00019999997337557415,
-      "loss": 46.0,
-      "step": 3048
-    },
-    {
-      "epoch": 0.23311734235525738,
-      "grad_norm": 0.0011713940184563398,
-      "learning_rate": 0.00019999997335804363,
-      "loss": 46.0,
-      "step": 3049
-    },
-    {
-      "epoch": 0.2331937993386471,
-      "grad_norm": 0.0031643749680370092,
-      "learning_rate": 0.0001999999733405074,
-      "loss": 46.0,
-      "step": 3050
-    },
-    {
-      "epoch": 0.23327025632203682,
-      "grad_norm": 0.0034488870296627283,
-      "learning_rate": 0.00019999997332296538,
-      "loss": 46.0,
-      "step": 3051
-    },
-    {
-      "epoch": 0.23334671330542653,
-      "grad_norm": 0.0013652169145643711,
-      "learning_rate": 0.00019999997330541757,
-      "loss": 46.0,
-      "step": 3052
-    },
-    {
-      "epoch": 0.23342317028881626,
-      "grad_norm": 0.0006173512665554881,
-      "learning_rate": 0.000199999973287864,
-      "loss": 46.0,
-      "step": 3053
-    },
-    {
-      "epoch": 0.23349962727220597,
-      "grad_norm": 0.0008447043946944177,
-      "learning_rate": 0.00019999997327030468,
-      "loss": 46.0,
-      "step": 3054
-    },
-    {
-      "epoch": 0.2335760842555957,
-      "grad_norm": 0.0014501862460747361,
-      "learning_rate": 0.00019999997325273957,
-      "loss": 46.0,
-      "step": 3055
-    },
-    {
-      "epoch": 0.2336525412389854,
-      "grad_norm": 0.0009118857560679317,
-      "learning_rate": 0.00019999997323516872,
-      "loss": 46.0,
-      "step": 3056
-    },
-    {
-      "epoch": 0.23372899822237514,
-      "grad_norm": 0.001235531410202384,
-      "learning_rate": 0.00019999997321759207,
-      "loss": 46.0,
-      "step": 3057
-    },
-    {
-      "epoch": 0.23380545520576484,
-      "grad_norm": 0.0018249572021886706,
-      "learning_rate": 0.00019999997320000964,
-      "loss": 46.0,
-      "step": 3058
-    },
-    {
-      "epoch": 0.23388191218915458,
-      "grad_norm": 0.0022003059275448322,
-      "learning_rate": 0.0001999999731824215,
-      "loss": 46.0,
-      "step": 3059
-    },
-    {
-      "epoch": 0.23395836917254428,
-      "grad_norm": 0.005554493051022291,
-      "learning_rate": 0.00019999997316482752,
-      "loss": 46.0,
-      "step": 3060
-    },
-    {
-      "epoch": 0.23403482615593402,
-      "grad_norm": 0.003695337101817131,
-      "learning_rate": 0.00019999997314722783,
-      "loss": 46.0,
-      "step": 3061
-    },
-    {
-      "epoch": 0.23411128313932375,
-      "grad_norm": 0.0036623438354581594,
-      "learning_rate": 0.00019999997312962234,
-      "loss": 46.0,
-      "step": 3062
-    },
-    {
-      "epoch": 0.23418774012271346,
-      "grad_norm": 0.0017200737493112683,
-      "learning_rate": 0.00019999997311201105,
-      "loss": 46.0,
-      "step": 3063
-    },
-    {
-      "epoch": 0.2342641971061032,
-      "grad_norm": 0.0003641015209723264,
-      "learning_rate": 0.00019999997309439404,
-      "loss": 46.0,
-      "step": 3064
-    },
-    {
-      "epoch": 0.2343406540894929,
-      "grad_norm": 0.0006425449973903596,
-      "learning_rate": 0.00019999997307677125,
-      "loss": 46.0,
-      "step": 3065
-    },
-    {
-      "epoch": 0.23441711107288263,
-      "grad_norm": 0.001516220043413341,
-      "learning_rate": 0.0001999999730591427,
-      "loss": 46.0,
-      "step": 3066
-    },
-    {
-      "epoch": 0.23449356805627233,
-      "grad_norm": 0.0014770985580980778,
-      "learning_rate": 0.00019999997304150834,
-      "loss": 46.0,
-      "step": 3067
-    },
-    {
-      "epoch": 0.23457002503966207,
-      "grad_norm": 0.0016748914495110512,
-      "learning_rate": 0.00019999997302386823,
-      "loss": 46.0,
-      "step": 3068
-    },
-    {
-      "epoch": 0.23464648202305177,
-      "grad_norm": 0.0016839439049363136,
-      "learning_rate": 0.00019999997300622235,
-      "loss": 46.0,
-      "step": 3069
-    },
-    {
-      "epoch": 0.2347229390064415,
-      "grad_norm": 0.00153891381341964,
-      "learning_rate": 0.00019999997298857073,
-      "loss": 46.0,
-      "step": 3070
-    },
-    {
-      "epoch": 0.2347993959898312,
-      "grad_norm": 0.0013383397599682212,
-      "learning_rate": 0.0001999999729709133,
-      "loss": 46.0,
-      "step": 3071
-    },
-    {
-      "epoch": 0.23487585297322094,
-      "grad_norm": 0.0011115333763882518,
-      "learning_rate": 0.00019999997295325013,
-      "loss": 46.0,
-      "step": 3072
-    },
-    {
-      "epoch": 0.23495230995661065,
-      "grad_norm": 0.0012283881660550833,
-      "learning_rate": 0.00019999997293558116,
-      "loss": 46.0,
-      "step": 3073
-    },
-    {
-      "epoch": 0.23502876694000038,
-      "grad_norm": 0.002778014400973916,
-      "learning_rate": 0.00019999997291790647,
-      "loss": 46.0,
-      "step": 3074
-    },
-    {
-      "epoch": 0.23510522392339012,
-      "grad_norm": 0.0013519068015739322,
-      "learning_rate": 0.00019999997290022595,
-      "loss": 46.0,
-      "step": 3075
-    },
-    {
-      "epoch": 0.23518168090677982,
-      "grad_norm": 0.0013388131046667695,
-      "learning_rate": 0.00019999997288253971,
-      "loss": 46.0,
-      "step": 3076
-    },
-    {
-      "epoch": 0.23525813789016956,
-      "grad_norm": 0.0007907307590357959,
-      "learning_rate": 0.00019999997286484765,
-      "loss": 46.0,
-      "step": 3077
-    },
-    {
-      "epoch": 0.23533459487355926,
-      "grad_norm": 0.0009382362477481365,
-      "learning_rate": 0.00019999997284714986,
-      "loss": 46.0,
-      "step": 3078
-    },
-    {
-      "epoch": 0.235411051856949,
-      "grad_norm": 0.014919286593794823,
-      "learning_rate": 0.00019999997282944628,
-      "loss": 46.0,
-      "step": 3079
-    },
-    {
-      "epoch": 0.2354875088403387,
-      "grad_norm": 0.0009769061580300331,
-      "learning_rate": 0.00019999997281173698,
-      "loss": 46.0,
-      "step": 3080
-    },
-    {
-      "epoch": 0.23556396582372843,
-      "grad_norm": 0.008831021375954151,
-      "learning_rate": 0.00019999997279402187,
-      "loss": 46.0,
-      "step": 3081
-    },
-    {
-      "epoch": 0.23564042280711814,
-      "grad_norm": 0.0016875851433724165,
-      "learning_rate": 0.000199999972776301,
-      "loss": 46.0,
-      "step": 3082
-    },
-    {
-      "epoch": 0.23571687979050787,
-      "grad_norm": 0.0007098687347024679,
-      "learning_rate": 0.00019999997275857434,
-      "loss": 46.0,
-      "step": 3083
-    },
-    {
-      "epoch": 0.23579333677389758,
-      "grad_norm": 0.0028730009216815233,
-      "learning_rate": 0.00019999997274084192,
-      "loss": 46.0,
-      "step": 3084
-    },
-    {
-      "epoch": 0.2358697937572873,
-      "grad_norm": 0.00518947746604681,
-      "learning_rate": 0.00019999997272310375,
-      "loss": 46.0,
-      "step": 3085
-    },
-    {
-      "epoch": 0.23594625074067702,
-      "grad_norm": 0.00037465529749169946,
-      "learning_rate": 0.00019999997270535977,
-      "loss": 46.0,
-      "step": 3086
-    },
-    {
-      "epoch": 0.23602270772406675,
-      "grad_norm": 0.002074565039947629,
-      "learning_rate": 0.00019999997268761006,
-      "loss": 46.0,
-      "step": 3087
-    },
-    {
-      "epoch": 0.23609916470745645,
-      "grad_norm": 0.00031881104223430157,
-      "learning_rate": 0.00019999997266985457,
-      "loss": 46.0,
-      "step": 3088
-    },
-    {
-      "epoch": 0.2361756216908462,
-      "grad_norm": 0.0008691286784596741,
-      "learning_rate": 0.0001999999726520933,
-      "loss": 46.0,
-      "step": 3089
-    },
-    {
-      "epoch": 0.23625207867423592,
-      "grad_norm": 0.0007430244004353881,
-      "learning_rate": 0.0001999999726343263,
-      "loss": 46.0,
-      "step": 3090
-    },
-    {
-      "epoch": 0.23632853565762563,
-      "grad_norm": 0.004080537706613541,
-      "learning_rate": 0.00019999997261655348,
-      "loss": 46.0,
-      "step": 3091
-    },
-    {
-      "epoch": 0.23640499264101536,
-      "grad_norm": 0.0008010346791706979,
-      "learning_rate": 0.0001999999725987749,
-      "loss": 46.0,
-      "step": 3092
-    },
-    {
-      "epoch": 0.23648144962440507,
-      "grad_norm": 0.0005251724505797029,
-      "learning_rate": 0.00019999997258099054,
-      "loss": 46.0,
-      "step": 3093
-    },
-    {
-      "epoch": 0.2365579066077948,
-      "grad_norm": 0.0018075796542689204,
-      "learning_rate": 0.00019999997256320043,
-      "loss": 46.0,
-      "step": 3094
-    },
-    {
-      "epoch": 0.2366343635911845,
-      "grad_norm": 0.0024680241476744413,
-      "learning_rate": 0.00019999997254540455,
-      "loss": 46.0,
-      "step": 3095
-    },
-    {
-      "epoch": 0.23671082057457424,
-      "grad_norm": 0.0018383264541625977,
-      "learning_rate": 0.00019999997252760293,
-      "loss": 46.0,
-      "step": 3096
-    },
-    {
-      "epoch": 0.23678727755796394,
-      "grad_norm": 0.001028991537168622,
-      "learning_rate": 0.0001999999725097955,
-      "loss": 46.0,
-      "step": 3097
-    },
-    {
-      "epoch": 0.23686373454135368,
-      "grad_norm": 0.0008818877977319062,
-      "learning_rate": 0.00019999997249198234,
-      "loss": 46.0,
-      "step": 3098
-    },
-    {
-      "epoch": 0.23694019152474338,
-      "grad_norm": 0.0019731391221284866,
-      "learning_rate": 0.00019999997247416336,
-      "loss": 46.0,
-      "step": 3099
-    },
-    {
-      "epoch": 0.23701664850813312,
-      "grad_norm": 0.0007807589136064053,
-      "learning_rate": 0.00019999997245633862,
-      "loss": 46.0,
-      "step": 3100
-    },
-    {
-      "epoch": 0.23709310549152282,
-      "grad_norm": 0.0025067089591175318,
-      "learning_rate": 0.00019999997243850813,
-      "loss": 46.0,
-      "step": 3101
-    },
-    {
-      "epoch": 0.23716956247491255,
-      "grad_norm": 0.003142465138807893,
-      "learning_rate": 0.00019999997242067187,
-      "loss": 46.0,
-      "step": 3102
-    },
-    {
-      "epoch": 0.23724601945830226,
-      "grad_norm": 0.0017502025002613664,
-      "learning_rate": 0.00019999997240282983,
-      "loss": 46.0,
-      "step": 3103
-    },
-    {
-      "epoch": 0.237322476441692,
-      "grad_norm": 0.0007660277769900858,
-      "learning_rate": 0.00019999997238498202,
-      "loss": 46.0,
-      "step": 3104
-    },
-    {
-      "epoch": 0.23739893342508173,
-      "grad_norm": 0.0021017694380134344,
-      "learning_rate": 0.00019999997236712846,
-      "loss": 46.0,
-      "step": 3105
-    },
-    {
-      "epoch": 0.23747539040847143,
-      "grad_norm": 0.0015855930978432298,
-      "learning_rate": 0.0001999999723492691,
-      "loss": 46.0,
-      "step": 3106
-    },
-    {
-      "epoch": 0.23755184739186117,
-      "grad_norm": 0.0011727524688467383,
-      "learning_rate": 0.000199999972331404,
-      "loss": 46.0,
-      "step": 3107
-    },
-    {
-      "epoch": 0.23762830437525087,
-      "grad_norm": 0.0010773026151582599,
-      "learning_rate": 0.00019999997231353312,
-      "loss": 46.0,
-      "step": 3108
-    },
-    {
-      "epoch": 0.2377047613586406,
-      "grad_norm": 0.0005457886145450175,
-      "learning_rate": 0.00019999997229565647,
-      "loss": 46.0,
-      "step": 3109
-    },
-    {
-      "epoch": 0.2377812183420303,
-      "grad_norm": 0.001725172158330679,
-      "learning_rate": 0.00019999997227777402,
-      "loss": 46.0,
-      "step": 3110
-    },
-    {
-      "epoch": 0.23785767532542004,
-      "grad_norm": 0.0009092254331335425,
-      "learning_rate": 0.00019999997225988585,
-      "loss": 46.0,
-      "step": 3111
-    },
-    {
-      "epoch": 0.23793413230880975,
-      "grad_norm": 0.00058984593488276,
-      "learning_rate": 0.0001999999722419919,
-      "loss": 46.0,
-      "step": 3112
-    },
-    {
-      "epoch": 0.23801058929219948,
-      "grad_norm": 0.0009242399828508496,
-      "learning_rate": 0.00019999997222409216,
-      "loss": 46.0,
-      "step": 3113
-    },
-    {
-      "epoch": 0.2380870462755892,
-      "grad_norm": 0.0006454035174101591,
-      "learning_rate": 0.00019999997220618667,
-      "loss": 46.0,
-      "step": 3114
-    },
-    {
-      "epoch": 0.23816350325897892,
-      "grad_norm": 0.002925480017438531,
-      "learning_rate": 0.0001999999721882754,
-      "loss": 46.0,
-      "step": 3115
-    },
-    {
-      "epoch": 0.23823996024236863,
-      "grad_norm": 0.0006803838186897337,
-      "learning_rate": 0.00019999997217035834,
-      "loss": 46.0,
-      "step": 3116
-    },
-    {
-      "epoch": 0.23831641722575836,
-      "grad_norm": 0.0006977599114179611,
-      "learning_rate": 0.00019999997215243553,
-      "loss": 46.0,
-      "step": 3117
-    },
-    {
-      "epoch": 0.23839287420914806,
-      "grad_norm": 0.00117528869304806,
-      "learning_rate": 0.00019999997213450697,
-      "loss": 46.0,
-      "step": 3118
-    },
-    {
-      "epoch": 0.2384693311925378,
-      "grad_norm": 0.0012723889667540789,
-      "learning_rate": 0.00019999997211657262,
-      "loss": 46.0,
-      "step": 3119
-    },
-    {
-      "epoch": 0.23854578817592753,
-      "grad_norm": 0.0007753327372483909,
-      "learning_rate": 0.00019999997209863251,
-      "loss": 46.0,
-      "step": 3120
-    },
-    {
-      "epoch": 0.23862224515931724,
-      "grad_norm": 0.003216144861653447,
-      "learning_rate": 0.0001999999720806866,
-      "loss": 46.0,
-      "step": 3121
-    },
-    {
-      "epoch": 0.23869870214270697,
-      "grad_norm": 0.0016340861329808831,
-      "learning_rate": 0.000199999972062735,
-      "loss": 46.0,
-      "step": 3122
-    },
-    {
-      "epoch": 0.23877515912609668,
-      "grad_norm": 0.001154883299022913,
-      "learning_rate": 0.00019999997204477754,
-      "loss": 46.0,
-      "step": 3123
-    },
-    {
-      "epoch": 0.2388516161094864,
-      "grad_norm": 0.0010707852197811007,
-      "learning_rate": 0.00019999997202681434,
-      "loss": 46.0,
-      "step": 3124
-    },
-    {
-      "epoch": 0.23892807309287611,
-      "grad_norm": 0.0013649303000420332,
-      "learning_rate": 0.00019999997200884537,
-      "loss": 46.0,
-      "step": 3125
-    },
-    {
-      "epoch": 0.23900453007626585,
-      "grad_norm": 0.0008350344724021852,
-      "learning_rate": 0.00019999997199087065,
-      "loss": 46.0,
-      "step": 3126
-    },
-    {
-      "epoch": 0.23908098705965555,
-      "grad_norm": 0.003300977637991309,
-      "learning_rate": 0.00019999997197289013,
-      "loss": 46.0,
-      "step": 3127
-    },
-    {
-      "epoch": 0.2391574440430453,
-      "grad_norm": 0.0006466025370173156,
-      "learning_rate": 0.00019999997195490387,
-      "loss": 46.0,
-      "step": 3128
-    },
-    {
-      "epoch": 0.239233901026435,
-      "grad_norm": 0.0007208252209238708,
-      "learning_rate": 0.00019999997193691183,
-      "loss": 46.0,
-      "step": 3129
-    },
-    {
-      "epoch": 0.23931035800982473,
-      "grad_norm": 0.000784624891821295,
-      "learning_rate": 0.00019999997191891402,
-      "loss": 46.0,
-      "step": 3130
-    },
-    {
-      "epoch": 0.23938681499321443,
-      "grad_norm": 0.001098365057259798,
-      "learning_rate": 0.00019999997190091044,
-      "loss": 46.0,
-      "step": 3131
-    },
-    {
-      "epoch": 0.23946327197660416,
-      "grad_norm": 0.0013156288769096136,
-      "learning_rate": 0.00019999997188290108,
-      "loss": 46.0,
-      "step": 3132
-    },
-    {
-      "epoch": 0.2395397289599939,
-      "grad_norm": 0.005401108413934708,
-      "learning_rate": 0.00019999997186488598,
-      "loss": 46.0,
-      "step": 3133
-    },
-    {
-      "epoch": 0.2396161859433836,
-      "grad_norm": 0.0022071991115808487,
-      "learning_rate": 0.00019999997184686508,
-      "loss": 46.0,
-      "step": 3134
-    },
-    {
-      "epoch": 0.23969264292677334,
-      "grad_norm": 0.0017125934828072786,
-      "learning_rate": 0.00019999997182883843,
-      "loss": 46.0,
-      "step": 3135
-    },
-    {
-      "epoch": 0.23976909991016304,
-      "grad_norm": 0.0014313082210719585,
-      "learning_rate": 0.000199999971810806,
-      "loss": 46.0,
-      "step": 3136
-    },
-    {
-      "epoch": 0.23984555689355277,
-      "grad_norm": 0.001316345063969493,
-      "learning_rate": 0.0001999999717927678,
-      "loss": 46.0,
-      "step": 3137
-    },
-    {
-      "epoch": 0.23992201387694248,
-      "grad_norm": 0.0031909095123410225,
-      "learning_rate": 0.00019999997177472386,
-      "loss": 46.0,
-      "step": 3138
-    },
-    {
-      "epoch": 0.2399984708603322,
-      "grad_norm": 0.0016524961683899164,
-      "learning_rate": 0.00019999997175667412,
-      "loss": 46.0,
-      "step": 3139
-    },
-    {
-      "epoch": 0.24007492784372192,
-      "grad_norm": 0.0006484754267148674,
-      "learning_rate": 0.0001999999717386186,
-      "loss": 46.0,
-      "step": 3140
-    },
-    {
-      "epoch": 0.24015138482711165,
-      "grad_norm": 0.0012863075826317072,
-      "learning_rate": 0.0001999999717205573,
-      "loss": 46.0,
-      "step": 3141
-    },
-    {
-      "epoch": 0.24022784181050136,
-      "grad_norm": 0.0014462950639426708,
-      "learning_rate": 0.00019999997170249027,
-      "loss": 46.0,
-      "step": 3142
-    },
-    {
-      "epoch": 0.2403042987938911,
-      "grad_norm": 0.00083252222975716,
-      "learning_rate": 0.00019999997168441746,
-      "loss": 46.0,
-      "step": 3143
-    },
-    {
-      "epoch": 0.2403807557772808,
-      "grad_norm": 0.0022779309656471014,
-      "learning_rate": 0.00019999997166633888,
-      "loss": 46.0,
-      "step": 3144
-    },
-    {
-      "epoch": 0.24045721276067053,
-      "grad_norm": 0.0013867767993360758,
-      "learning_rate": 0.00019999997164825452,
-      "loss": 46.0,
-      "step": 3145
-    },
-    {
-      "epoch": 0.24053366974406024,
-      "grad_norm": 0.002565558534115553,
-      "learning_rate": 0.00019999997163016442,
-      "loss": 46.0,
-      "step": 3146
-    },
-    {
-      "epoch": 0.24061012672744997,
-      "grad_norm": 0.0019565375987440348,
-      "learning_rate": 0.00019999997161206852,
-      "loss": 46.0,
-      "step": 3147
-    },
-    {
-      "epoch": 0.2406865837108397,
-      "grad_norm": 0.0009911081288009882,
-      "learning_rate": 0.00019999997159396684,
-      "loss": 46.0,
-      "step": 3148
-    },
-    {
-      "epoch": 0.2407630406942294,
-      "grad_norm": 0.001760189770720899,
-      "learning_rate": 0.00019999997157585942,
-      "loss": 46.0,
-      "step": 3149
-    },
-    {
-      "epoch": 0.24083949767761914,
-      "grad_norm": 0.0032243893947452307,
-      "learning_rate": 0.00019999997155774622,
-      "loss": 46.0,
-      "step": 3150
-    },
-    {
-      "epoch": 0.24091595466100885,
-      "grad_norm": 0.004657264798879623,
-      "learning_rate": 0.00019999997153962725,
-      "loss": 46.0,
-      "step": 3151
-    },
-    {
-      "epoch": 0.24099241164439858,
-      "grad_norm": 0.0016785510815680027,
-      "learning_rate": 0.0001999999715215025,
-      "loss": 46.0,
-      "step": 3152
-    },
-    {
-      "epoch": 0.24106886862778829,
-      "grad_norm": 0.0014704017667099833,
-      "learning_rate": 0.00019999997150337202,
-      "loss": 46.0,
-      "step": 3153
-    },
-    {
-      "epoch": 0.24114532561117802,
-      "grad_norm": 0.0009119369206018746,
-      "learning_rate": 0.00019999997148523573,
-      "loss": 46.0,
-      "step": 3154
-    },
-    {
-      "epoch": 0.24122178259456772,
-      "grad_norm": 0.006077085621654987,
-      "learning_rate": 0.0001999999714670937,
-      "loss": 46.0,
-      "step": 3155
-    },
-    {
-      "epoch": 0.24129823957795746,
-      "grad_norm": 0.010737337172031403,
-      "learning_rate": 0.00019999997144894586,
-      "loss": 46.0,
-      "step": 3156
-    },
-    {
-      "epoch": 0.24137469656134716,
-      "grad_norm": 0.0013041161000728607,
-      "learning_rate": 0.00019999997143079227,
-      "loss": 46.0,
-      "step": 3157
-    },
-    {
-      "epoch": 0.2414511535447369,
-      "grad_norm": 0.0010203189449384809,
-      "learning_rate": 0.00019999997141263292,
-      "loss": 46.0,
-      "step": 3158
-    },
-    {
-      "epoch": 0.2415276105281266,
-      "grad_norm": 0.007857026532292366,
-      "learning_rate": 0.0001999999713944678,
-      "loss": 46.0,
-      "step": 3159
-    },
-    {
-      "epoch": 0.24160406751151633,
-      "grad_norm": 0.0016786643536761403,
-      "learning_rate": 0.0001999999713762969,
-      "loss": 46.0,
-      "step": 3160
-    },
-    {
-      "epoch": 0.24168052449490604,
-      "grad_norm": 0.0016964125679805875,
-      "learning_rate": 0.00019999997135812026,
-      "loss": 46.0,
-      "step": 3161
-    },
-    {
-      "epoch": 0.24175698147829577,
-      "grad_norm": 0.0016391290118917823,
-      "learning_rate": 0.0001999999713399378,
-      "loss": 46.0,
-      "step": 3162
-    },
-    {
-      "epoch": 0.2418334384616855,
-      "grad_norm": 0.0015250424621626735,
-      "learning_rate": 0.00019999997132174962,
-      "loss": 46.0,
-      "step": 3163
-    },
-    {
-      "epoch": 0.2419098954450752,
-      "grad_norm": 0.0009343001875095069,
-      "learning_rate": 0.00019999997130355565,
-      "loss": 46.0,
-      "step": 3164
-    },
-    {
-      "epoch": 0.24198635242846495,
-      "grad_norm": 0.0010346361668780446,
-      "learning_rate": 0.00019999997128535588,
-      "loss": 46.0,
-      "step": 3165
-    },
-    {
-      "epoch": 0.24206280941185465,
-      "grad_norm": 0.0010919218184426427,
-      "learning_rate": 0.0001999999712671504,
-      "loss": 46.0,
-      "step": 3166
-    },
-    {
-      "epoch": 0.24213926639524438,
-      "grad_norm": 0.0014733336865901947,
-      "learning_rate": 0.0001999999712489391,
-      "loss": 46.0,
-      "step": 3167
-    },
-    {
-      "epoch": 0.2422157233786341,
-      "grad_norm": 0.004036740865558386,
-      "learning_rate": 0.00019999997123072206,
-      "loss": 46.0,
-      "step": 3168
-    },
-    {
-      "epoch": 0.24229218036202382,
-      "grad_norm": 0.0006784771103411913,
-      "learning_rate": 0.00019999997121249922,
-      "loss": 46.0,
-      "step": 3169
-    },
-    {
-      "epoch": 0.24236863734541353,
-      "grad_norm": 0.0014110878109931946,
-      "learning_rate": 0.00019999997119427064,
-      "loss": 46.0,
-      "step": 3170
-    },
-    {
-      "epoch": 0.24244509432880326,
-      "grad_norm": 0.001428284216672182,
-      "learning_rate": 0.00019999997117603629,
-      "loss": 46.0,
-      "step": 3171
-    },
-    {
-      "epoch": 0.24252155131219297,
-      "grad_norm": 0.0006259740912355483,
-      "learning_rate": 0.00019999997115779616,
-      "loss": 46.0,
-      "step": 3172
-    },
-    {
-      "epoch": 0.2425980082955827,
-      "grad_norm": 0.0008021716494113207,
-      "learning_rate": 0.00019999997113955028,
-      "loss": 46.0,
-      "step": 3173
-    },
-    {
-      "epoch": 0.2426744652789724,
-      "grad_norm": 0.0010673960205167532,
-      "learning_rate": 0.0001999999711212986,
-      "loss": 46.0,
-      "step": 3174
-    },
-    {
-      "epoch": 0.24275092226236214,
-      "grad_norm": 0.0027660317718982697,
-      "learning_rate": 0.00019999997110304115,
-      "loss": 46.0,
-      "step": 3175
-    },
-    {
-      "epoch": 0.24282737924575185,
-      "grad_norm": 0.0016346651827916503,
-      "learning_rate": 0.00019999997108477796,
-      "loss": 46.0,
-      "step": 3176
-    },
-    {
-      "epoch": 0.24290383622914158,
-      "grad_norm": 0.001536539988592267,
-      "learning_rate": 0.000199999971066509,
-      "loss": 46.0,
-      "step": 3177
-    },
-    {
-      "epoch": 0.2429802932125313,
-      "grad_norm": 0.0007707267068326473,
-      "learning_rate": 0.00019999997104823422,
-      "loss": 46.0,
-      "step": 3178
-    },
-    {
-      "epoch": 0.24305675019592102,
-      "grad_norm": 0.001259678159840405,
-      "learning_rate": 0.00019999997102995373,
-      "loss": 46.0,
-      "step": 3179
-    },
-    {
-      "epoch": 0.24313320717931075,
-      "grad_norm": 0.0007284677121788263,
-      "learning_rate": 0.00019999997101166744,
-      "loss": 46.0,
-      "step": 3180
-    },
-    {
-      "epoch": 0.24320966416270046,
-      "grad_norm": 0.0007023262442089617,
-      "learning_rate": 0.0001999999709933754,
-      "loss": 46.0,
-      "step": 3181
-    },
-    {
-      "epoch": 0.2432861211460902,
-      "grad_norm": 0.0027272161096334457,
-      "learning_rate": 0.00019999997097507757,
-      "loss": 46.0,
-      "step": 3182
-    },
-    {
-      "epoch": 0.2433625781294799,
-      "grad_norm": 0.0008949391194619238,
-      "learning_rate": 0.00019999997095677396,
-      "loss": 46.0,
-      "step": 3183
-    },
-    {
-      "epoch": 0.24343903511286963,
-      "grad_norm": 0.001690986449830234,
-      "learning_rate": 0.0001999999709384646,
-      "loss": 46.0,
-      "step": 3184
-    },
-    {
-      "epoch": 0.24351549209625933,
-      "grad_norm": 0.0010557736968621612,
-      "learning_rate": 0.00019999997092014948,
-      "loss": 46.0,
-      "step": 3185
-    },
-    {
-      "epoch": 0.24359194907964907,
-      "grad_norm": 0.0006213740562088788,
-      "learning_rate": 0.00019999997090182857,
-      "loss": 46.0,
-      "step": 3186
-    },
-    {
-      "epoch": 0.24366840606303877,
-      "grad_norm": 0.001995612168684602,
-      "learning_rate": 0.0001999999708835019,
-      "loss": 46.0,
-      "step": 3187
-    },
-    {
-      "epoch": 0.2437448630464285,
-      "grad_norm": 0.0026208574417978525,
-      "learning_rate": 0.00019999997086516945,
-      "loss": 46.0,
-      "step": 3188
-    },
-    {
-      "epoch": 0.2438213200298182,
-      "grad_norm": 0.0008315776358358562,
-      "learning_rate": 0.00019999997084683125,
-      "loss": 46.0,
-      "step": 3189
-    },
-    {
-      "epoch": 0.24389777701320794,
-      "grad_norm": 0.0011688253143802285,
-      "learning_rate": 0.00019999997082848728,
-      "loss": 46.0,
-      "step": 3190
-    },
-    {
-      "epoch": 0.24397423399659768,
-      "grad_norm": 0.0007019365439191461,
-      "learning_rate": 0.00019999997081013751,
-      "loss": 46.0,
-      "step": 3191
-    },
-    {
-      "epoch": 0.24405069097998738,
-      "grad_norm": 0.0010562252718955278,
-      "learning_rate": 0.00019999997079178203,
-      "loss": 46.0,
-      "step": 3192
-    },
-    {
-      "epoch": 0.24412714796337712,
-      "grad_norm": 0.0009864402236416936,
-      "learning_rate": 0.00019999997077342074,
-      "loss": 46.0,
-      "step": 3193
-    },
-    {
-      "epoch": 0.24420360494676682,
-      "grad_norm": 0.0009028976201079786,
-      "learning_rate": 0.00019999997075505367,
-      "loss": 46.0,
-      "step": 3194
-    },
-    {
-      "epoch": 0.24428006193015656,
-      "grad_norm": 0.0013834924902766943,
-      "learning_rate": 0.00019999997073668087,
-      "loss": 46.0,
-      "step": 3195
-    },
-    {
-      "epoch": 0.24435651891354626,
-      "grad_norm": 0.002960958518087864,
-      "learning_rate": 0.00019999997071830228,
-      "loss": 46.0,
-      "step": 3196
-    },
-    {
-      "epoch": 0.244432975896936,
-      "grad_norm": 0.0032917845528572798,
-      "learning_rate": 0.0001999999706999179,
-      "loss": 46.0,
-      "step": 3197
-    },
-    {
-      "epoch": 0.2445094328803257,
-      "grad_norm": 0.002530928235501051,
-      "learning_rate": 0.00019999997068152777,
-      "loss": 46.0,
-      "step": 3198
-    },
-    {
-      "epoch": 0.24458588986371543,
-      "grad_norm": 0.000605881039518863,
-      "learning_rate": 0.00019999997066313187,
-      "loss": 46.0,
-      "step": 3199
-    },
-    {
-      "epoch": 0.24466234684710514,
-      "grad_norm": 0.0012429484631866217,
-      "learning_rate": 0.0001999999706447302,
-      "loss": 46.0,
-      "step": 3200
-    },
-    {
-      "epoch": 0.24473880383049487,
-      "grad_norm": 0.005064371973276138,
-      "learning_rate": 0.00019999997062632275,
-      "loss": 46.0,
-      "step": 3201
-    },
-    {
-      "epoch": 0.24481526081388458,
-      "grad_norm": 0.0011240814346820116,
-      "learning_rate": 0.00019999997060790955,
-      "loss": 46.0,
-      "step": 3202
-    },
-    {
-      "epoch": 0.2448917177972743,
-      "grad_norm": 0.00804038904607296,
-      "learning_rate": 0.00019999997058949055,
-      "loss": 46.0,
-      "step": 3203
-    },
-    {
-      "epoch": 0.24496817478066402,
-      "grad_norm": 0.0004708790220320225,
-      "learning_rate": 0.0001999999705710658,
-      "loss": 46.0,
-      "step": 3204
-    },
-    {
-      "epoch": 0.24504463176405375,
-      "grad_norm": 0.0010552947642281651,
-      "learning_rate": 0.0001999999705526353,
-      "loss": 46.0,
-      "step": 3205
-    },
-    {
-      "epoch": 0.24512108874744348,
-      "grad_norm": 0.000468751008156687,
-      "learning_rate": 0.000199999970534199,
-      "loss": 46.0,
-      "step": 3206
-    },
-    {
-      "epoch": 0.2451975457308332,
-      "grad_norm": 0.0004748433129861951,
-      "learning_rate": 0.00019999997051575695,
-      "loss": 46.0,
-      "step": 3207
-    },
-    {
-      "epoch": 0.24527400271422292,
-      "grad_norm": 0.0018256522016599774,
-      "learning_rate": 0.0001999999704973091,
-      "loss": 46.0,
-      "step": 3208
-    },
-    {
-      "epoch": 0.24535045969761263,
-      "grad_norm": 0.0006603817455470562,
-      "learning_rate": 0.0001999999704788555,
-      "loss": 46.0,
-      "step": 3209
-    },
-    {
-      "epoch": 0.24542691668100236,
-      "grad_norm": 0.0004485788522288203,
-      "learning_rate": 0.00019999997046039615,
-      "loss": 46.0,
-      "step": 3210
-    },
-    {
-      "epoch": 0.24550337366439207,
-      "grad_norm": 0.001897329930216074,
-      "learning_rate": 0.00019999997044193102,
-      "loss": 46.0,
-      "step": 3211
-    },
-    {
-      "epoch": 0.2455798306477818,
-      "grad_norm": 0.0006241471855901182,
-      "learning_rate": 0.00019999997042346012,
-      "loss": 46.0,
-      "step": 3212
-    },
-    {
-      "epoch": 0.2456562876311715,
-      "grad_norm": 0.0008216181886382401,
-      "learning_rate": 0.00019999997040498344,
-      "loss": 46.0,
-      "step": 3213
-    },
-    {
-      "epoch": 0.24573274461456124,
-      "grad_norm": 0.0010022129863500595,
-      "learning_rate": 0.000199999970386501,
-      "loss": 46.0,
-      "step": 3214
-    },
-    {
-      "epoch": 0.24580920159795094,
-      "grad_norm": 0.0019924382213503122,
-      "learning_rate": 0.00019999997036801277,
-      "loss": 46.0,
-      "step": 3215
-    },
-    {
-      "epoch": 0.24588565858134068,
-      "grad_norm": 0.0009271148010157049,
-      "learning_rate": 0.00019999997034951877,
-      "loss": 46.0,
-      "step": 3216
-    },
-    {
-      "epoch": 0.24596211556473038,
-      "grad_norm": 0.0008573323721066117,
-      "learning_rate": 0.00019999997033101906,
-      "loss": 46.0,
-      "step": 3217
-    },
-    {
-      "epoch": 0.24603857254812012,
-      "grad_norm": 0.001418428379110992,
-      "learning_rate": 0.00019999997031251352,
-      "loss": 46.0,
-      "step": 3218
-    },
-    {
-      "epoch": 0.24611502953150982,
-      "grad_norm": 0.001948562334291637,
-      "learning_rate": 0.0001999999702940022,
-      "loss": 46.0,
-      "step": 3219
-    },
-    {
-      "epoch": 0.24619148651489955,
-      "grad_norm": 0.0020987882744520903,
-      "learning_rate": 0.00019999997027548517,
-      "loss": 46.0,
-      "step": 3220
-    },
-    {
-      "epoch": 0.2462679434982893,
-      "grad_norm": 0.0036669347900897264,
-      "learning_rate": 0.00019999997025696233,
-      "loss": 46.0,
-      "step": 3221
-    },
-    {
-      "epoch": 0.246344400481679,
-      "grad_norm": 0.0014769320841878653,
-      "learning_rate": 0.00019999997023843372,
-      "loss": 46.0,
-      "step": 3222
-    },
-    {
-      "epoch": 0.24642085746506873,
-      "grad_norm": 0.003571388777345419,
-      "learning_rate": 0.00019999997021989937,
-      "loss": 46.0,
-      "step": 3223
-    },
-    {
-      "epoch": 0.24649731444845843,
-      "grad_norm": 0.0016468921676278114,
-      "learning_rate": 0.00019999997020135921,
-      "loss": 46.0,
-      "step": 3224
-    },
-    {
-      "epoch": 0.24657377143184817,
-      "grad_norm": 0.0011206233175471425,
-      "learning_rate": 0.0001999999701828133,
-      "loss": 46.0,
-      "step": 3225
-    },
-    {
-      "epoch": 0.24665022841523787,
-      "grad_norm": 0.0006686090491712093,
-      "learning_rate": 0.00019999997016426164,
-      "loss": 46.0,
-      "step": 3226
-    },
-    {
-      "epoch": 0.2467266853986276,
-      "grad_norm": 0.0012624136870726943,
-      "learning_rate": 0.00019999997014570416,
-      "loss": 46.0,
-      "step": 3227
-    },
-    {
-      "epoch": 0.2468031423820173,
-      "grad_norm": 0.0017163109732791781,
-      "learning_rate": 0.00019999997012714097,
-      "loss": 46.0,
-      "step": 3228
-    },
-    {
-      "epoch": 0.24687959936540704,
-      "grad_norm": 0.0011483390117064118,
-      "learning_rate": 0.000199999970108572,
-      "loss": 46.0,
-      "step": 3229
-    },
-    {
-      "epoch": 0.24695605634879675,
-      "grad_norm": 0.0005547954351641238,
-      "learning_rate": 0.00019999997008999723,
-      "loss": 46.0,
-      "step": 3230
-    },
-    {
-      "epoch": 0.24703251333218648,
-      "grad_norm": 0.001476628822274506,
-      "learning_rate": 0.00019999997007141672,
-      "loss": 46.0,
-      "step": 3231
-    },
-    {
-      "epoch": 0.2471089703155762,
-      "grad_norm": 0.0017434757901355624,
-      "learning_rate": 0.00019999997005283043,
-      "loss": 46.0,
-      "step": 3232
-    },
-    {
-      "epoch": 0.24718542729896592,
-      "grad_norm": 0.0012250370346009731,
-      "learning_rate": 0.00019999997003423837,
-      "loss": 46.0,
-      "step": 3233
-    },
-    {
-      "epoch": 0.24726188428235563,
-      "grad_norm": 0.0019066028762608767,
-      "learning_rate": 0.00019999997001564053,
-      "loss": 46.0,
-      "step": 3234
-    },
-    {
-      "epoch": 0.24733834126574536,
-      "grad_norm": 0.0013979371869936585,
-      "learning_rate": 0.00019999996999703692,
-      "loss": 46.0,
-      "step": 3235
-    },
-    {
-      "epoch": 0.2474147982491351,
-      "grad_norm": 0.0038987488951534033,
-      "learning_rate": 0.00019999996997842754,
-      "loss": 46.0,
-      "step": 3236
-    },
-    {
-      "epoch": 0.2474912552325248,
-      "grad_norm": 0.000744482793379575,
-      "learning_rate": 0.0001999999699598124,
-      "loss": 46.0,
-      "step": 3237
-    },
-    {
-      "epoch": 0.24756771221591453,
-      "grad_norm": 0.0019050112459808588,
-      "learning_rate": 0.00019999996994119149,
-      "loss": 46.0,
-      "step": 3238
-    },
-    {
-      "epoch": 0.24764416919930424,
-      "grad_norm": 0.0011425131233409047,
-      "learning_rate": 0.0001999999699225648,
-      "loss": 46.0,
-      "step": 3239
-    },
-    {
-      "epoch": 0.24772062618269397,
-      "grad_norm": 0.000778753194026649,
-      "learning_rate": 0.00019999996990393234,
-      "loss": 46.0,
-      "step": 3240
-    },
-    {
-      "epoch": 0.24779708316608368,
-      "grad_norm": 0.00144282600376755,
-      "learning_rate": 0.00019999996988529411,
-      "loss": 46.0,
-      "step": 3241
-    },
-    {
-      "epoch": 0.2478735401494734,
-      "grad_norm": 0.0014188982313498855,
-      "learning_rate": 0.00019999996986665015,
-      "loss": 46.0,
-      "step": 3242
-    },
-    {
-      "epoch": 0.24794999713286311,
-      "grad_norm": 0.0012028975179418921,
-      "learning_rate": 0.00019999996984800038,
-      "loss": 46.0,
-      "step": 3243
-    },
-    {
-      "epoch": 0.24802645411625285,
-      "grad_norm": 0.0005816504708491266,
-      "learning_rate": 0.00019999996982934486,
-      "loss": 46.0,
-      "step": 3244
-    },
-    {
-      "epoch": 0.24810291109964255,
-      "grad_norm": 0.0005952494102530181,
-      "learning_rate": 0.00019999996981068355,
-      "loss": 46.0,
-      "step": 3245
-    },
-    {
-      "epoch": 0.2481793680830323,
-      "grad_norm": 0.0005283544887788594,
-      "learning_rate": 0.00019999996979201646,
-      "loss": 46.0,
-      "step": 3246
-    },
-    {
-      "epoch": 0.248255825066422,
-      "grad_norm": 0.00039464057772420347,
-      "learning_rate": 0.00019999996977334365,
-      "loss": 46.0,
-      "step": 3247
-    },
-    {
-      "epoch": 0.24833228204981173,
-      "grad_norm": 0.0015188711695373058,
-      "learning_rate": 0.00019999996975466507,
-      "loss": 46.0,
-      "step": 3248
-    },
-    {
-      "epoch": 0.24840873903320146,
-      "grad_norm": 0.0006703032995574176,
-      "learning_rate": 0.00019999996973598066,
-      "loss": 46.0,
-      "step": 3249
-    },
-    {
-      "epoch": 0.24848519601659116,
-      "grad_norm": 0.0031718409154564142,
-      "learning_rate": 0.0001999999697172905,
-      "loss": 46.0,
-      "step": 3250
-    },
-    {
-      "epoch": 0.2485616529999809,
-      "grad_norm": 0.0019810611847788095,
-      "learning_rate": 0.00019999996969859464,
-      "loss": 46.0,
-      "step": 3251
-    },
-    {
-      "epoch": 0.2486381099833706,
-      "grad_norm": 0.004819038324058056,
-      "learning_rate": 0.0001999999696798929,
-      "loss": 46.0,
-      "step": 3252
-    },
-    {
-      "epoch": 0.24871456696676034,
-      "grad_norm": 0.001066030701622367,
-      "learning_rate": 0.0001999999696611855,
-      "loss": 46.0,
-      "step": 3253
-    },
-    {
-      "epoch": 0.24879102395015004,
-      "grad_norm": 0.002298054052516818,
-      "learning_rate": 0.00019999996964247224,
-      "loss": 46.0,
-      "step": 3254
-    },
-    {
-      "epoch": 0.24886748093353978,
-      "grad_norm": 0.0033549789804965258,
-      "learning_rate": 0.00019999996962375324,
-      "loss": 46.0,
-      "step": 3255
-    },
-    {
-      "epoch": 0.24894393791692948,
-      "grad_norm": 0.005768239963799715,
-      "learning_rate": 0.0001999999696050285,
-      "loss": 46.0,
-      "step": 3256
-    },
-    {
-      "epoch": 0.24902039490031921,
-      "grad_norm": 0.0016898488393053412,
-      "learning_rate": 0.00019999996958629796,
-      "loss": 46.0,
-      "step": 3257
-    },
-    {
-      "epoch": 0.24909685188370892,
-      "grad_norm": 0.0008953118231147528,
-      "learning_rate": 0.00019999996956756167,
-      "loss": 46.0,
-      "step": 3258
-    },
-    {
-      "epoch": 0.24917330886709865,
-      "grad_norm": 0.004425527527928352,
-      "learning_rate": 0.0001999999695488196,
-      "loss": 46.0,
-      "step": 3259
-    },
-    {
-      "epoch": 0.24924976585048836,
-      "grad_norm": 0.007059887982904911,
-      "learning_rate": 0.00019999996953007175,
-      "loss": 46.0,
-      "step": 3260
-    },
-    {
-      "epoch": 0.2493262228338781,
-      "grad_norm": 0.0011865132255479693,
-      "learning_rate": 0.00019999996951131817,
-      "loss": 46.0,
-      "step": 3261
-    },
-    {
-      "epoch": 0.2494026798172678,
-      "grad_norm": 0.009505679830908775,
-      "learning_rate": 0.00019999996949255876,
-      "loss": 46.0,
-      "step": 3262
-    },
-    {
-      "epoch": 0.24947913680065753,
-      "grad_norm": 0.004992562811821699,
-      "learning_rate": 0.00019999996947379364,
-      "loss": 46.0,
-      "step": 3263
-    },
-    {
-      "epoch": 0.24955559378404726,
-      "grad_norm": 0.0014782652724534273,
-      "learning_rate": 0.0001999999694550227,
-      "loss": 46.0,
-      "step": 3264
-    },
-    {
-      "epoch": 0.24963205076743697,
-      "grad_norm": 0.0010925416136160493,
-      "learning_rate": 0.000199999969436246,
-      "loss": 46.0,
-      "step": 3265
-    },
-    {
-      "epoch": 0.2497085077508267,
-      "grad_norm": 0.0027794234920293093,
-      "learning_rate": 0.00019999996941746356,
-      "loss": 46.0,
-      "step": 3266
-    },
-    {
-      "epoch": 0.2497849647342164,
-      "grad_norm": 0.0015363667625933886,
-      "learning_rate": 0.00019999996939867534,
-      "loss": 46.0,
-      "step": 3267
-    },
-    {
-      "epoch": 0.24986142171760614,
-      "grad_norm": 0.0014524442376568913,
-      "learning_rate": 0.00019999996937988135,
-      "loss": 46.0,
-      "step": 3268
-    },
-    {
-      "epoch": 0.24993787870099585,
-      "grad_norm": 0.000899262900929898,
-      "learning_rate": 0.00019999996936108158,
-      "loss": 46.0,
-      "step": 3269
-    },
-    {
-      "epoch": 0.2500143356843856,
-      "grad_norm": 0.0004008629184681922,
-      "learning_rate": 0.00019999996934227604,
-      "loss": 46.0,
-      "step": 3270
-    },
-    {
-      "epoch": 0.2500907926677753,
-      "grad_norm": 0.000693173089530319,
-      "learning_rate": 0.00019999996932346475,
-      "loss": 46.0,
-      "step": 3271
-    },
-    {
-      "epoch": 0.250167249651165,
-      "grad_norm": 0.0015817157691344619,
-      "learning_rate": 0.00019999996930464766,
-      "loss": 46.0,
-      "step": 3272
-    },
-    {
-      "epoch": 0.25024370663455475,
-      "grad_norm": 0.0015744644915685058,
-      "learning_rate": 0.00019999996928582483,
-      "loss": 46.0,
-      "step": 3273
-    },
-    {
-      "epoch": 0.25032016361794446,
-      "grad_norm": 0.000943234481383115,
-      "learning_rate": 0.00019999996926699622,
-      "loss": 46.0,
-      "step": 3274
-    },
-    {
-      "epoch": 0.25039662060133416,
-      "grad_norm": 0.00100785365793854,
-      "learning_rate": 0.0001999999692481618,
-      "loss": 46.0,
-      "step": 3275
-    },
-    {
-      "epoch": 0.25047307758472387,
-      "grad_norm": 0.0021864501759409904,
-      "learning_rate": 0.0001999999692293217,
-      "loss": 46.0,
-      "step": 3276
-    },
-    {
-      "epoch": 0.25054953456811363,
-      "grad_norm": 0.0008699607569724321,
-      "learning_rate": 0.00019999996921047576,
-      "loss": 46.0,
-      "step": 3277
-    },
-    {
-      "epoch": 0.25062599155150334,
-      "grad_norm": 0.0004331499512773007,
-      "learning_rate": 0.00019999996919162406,
-      "loss": 46.0,
-      "step": 3278
-    },
-    {
-      "epoch": 0.25070244853489304,
-      "grad_norm": 0.002732151886448264,
-      "learning_rate": 0.0001999999691727666,
-      "loss": 46.0,
-      "step": 3279
-    },
-    {
-      "epoch": 0.2507789055182828,
-      "grad_norm": 0.004899463150650263,
-      "learning_rate": 0.0001999999691539034,
-      "loss": 46.0,
-      "step": 3280
-    },
-    {
-      "epoch": 0.2508553625016725,
-      "grad_norm": 0.0009712368482723832,
-      "learning_rate": 0.00019999996913503437,
-      "loss": 46.0,
-      "step": 3281
-    },
-    {
-      "epoch": 0.2509318194850622,
-      "grad_norm": 0.004395731259137392,
-      "learning_rate": 0.00019999996911615963,
-      "loss": 46.0,
-      "step": 3282
-    },
-    {
-      "epoch": 0.2510082764684519,
-      "grad_norm": 0.0009943393524736166,
-      "learning_rate": 0.0001999999690972791,
-      "loss": 46.0,
-      "step": 3283
-    },
-    {
-      "epoch": 0.2510847334518417,
-      "grad_norm": 0.0015691715525463223,
-      "learning_rate": 0.00019999996907839278,
-      "loss": 46.0,
-      "step": 3284
-    },
-    {
-      "epoch": 0.2511611904352314,
-      "grad_norm": 0.0011420054361224174,
-      "learning_rate": 0.0001999999690595007,
-      "loss": 46.0,
-      "step": 3285
-    },
-    {
-      "epoch": 0.2512376474186211,
-      "grad_norm": 0.0010686635505408049,
-      "learning_rate": 0.00019999996904060285,
-      "loss": 46.0,
-      "step": 3286
-    },
-    {
-      "epoch": 0.2513141044020108,
-      "grad_norm": 0.0005065586883574724,
-      "learning_rate": 0.00019999996902169925,
-      "loss": 46.0,
-      "step": 3287
-    },
-    {
-      "epoch": 0.25139056138540056,
-      "grad_norm": 0.0025305699091404676,
-      "learning_rate": 0.00019999996900278987,
-      "loss": 46.0,
-      "step": 3288
-    },
-    {
-      "epoch": 0.25146701836879026,
-      "grad_norm": 0.0015879916027188301,
-      "learning_rate": 0.00019999996898387469,
-      "loss": 46.0,
-      "step": 3289
-    },
-    {
-      "epoch": 0.25154347535217997,
-      "grad_norm": 0.0010090366704389453,
-      "learning_rate": 0.00019999996896495379,
-      "loss": 46.0,
-      "step": 3290
-    },
-    {
-      "epoch": 0.2516199323355697,
-      "grad_norm": 0.00225488911382854,
-      "learning_rate": 0.00019999996894602709,
-      "loss": 46.0,
-      "step": 3291
-    },
-    {
-      "epoch": 0.25169638931895943,
-      "grad_norm": 0.0006276239291764796,
-      "learning_rate": 0.00019999996892709464,
-      "loss": 46.0,
-      "step": 3292
-    },
-    {
-      "epoch": 0.25177284630234914,
-      "grad_norm": 0.010474435985088348,
-      "learning_rate": 0.0001999999689081564,
-      "loss": 46.0,
-      "step": 3293
-    },
-    {
-      "epoch": 0.25184930328573885,
-      "grad_norm": 0.000672626425512135,
-      "learning_rate": 0.0001999999688892124,
-      "loss": 46.0,
-      "step": 3294
-    },
-    {
-      "epoch": 0.2519257602691286,
-      "grad_norm": 0.004152297042310238,
-      "learning_rate": 0.00019999996887026263,
-      "loss": 46.0,
-      "step": 3295
-    },
-    {
-      "epoch": 0.2520022172525183,
-      "grad_norm": 0.002188890939578414,
-      "learning_rate": 0.0001999999688513071,
-      "loss": 46.0,
-      "step": 3296
-    },
-    {
-      "epoch": 0.252078674235908,
-      "grad_norm": 0.0010546904522925615,
-      "learning_rate": 0.00019999996883234578,
-      "loss": 46.0,
-      "step": 3297
-    },
-    {
-      "epoch": 0.2521551312192977,
-      "grad_norm": 0.0012328902957960963,
-      "learning_rate": 0.0001999999688133787,
-      "loss": 46.0,
-      "step": 3298
-    },
-    {
-      "epoch": 0.2522315882026875,
-      "grad_norm": 0.000764339929446578,
-      "learning_rate": 0.00019999996879440586,
-      "loss": 46.0,
-      "step": 3299
-    },
-    {
-      "epoch": 0.2523080451860772,
-      "grad_norm": 0.0014056528452783823,
-      "learning_rate": 0.00019999996877542722,
-      "loss": 46.0,
-      "step": 3300
-    },
-    {
-      "epoch": 0.2523845021694669,
-      "grad_norm": 0.0009448706987313926,
-      "learning_rate": 0.00019999996875644284,
-      "loss": 46.0,
-      "step": 3301
-    },
-    {
-      "epoch": 0.2524609591528566,
-      "grad_norm": 0.001086943899281323,
-      "learning_rate": 0.0001999999687374527,
-      "loss": 46.0,
-      "step": 3302
-    },
-    {
-      "epoch": 0.25253741613624636,
-      "grad_norm": 0.0021563644986599684,
-      "learning_rate": 0.00019999996871845676,
-      "loss": 46.0,
-      "step": 3303
-    },
-    {
-      "epoch": 0.25261387311963607,
-      "grad_norm": 0.0020023297984153032,
-      "learning_rate": 0.00019999996869945506,
-      "loss": 46.0,
-      "step": 3304
-    },
-    {
-      "epoch": 0.2526903301030258,
-      "grad_norm": 0.00150933931581676,
-      "learning_rate": 0.0001999999686804476,
-      "loss": 46.0,
-      "step": 3305
-    },
-    {
-      "epoch": 0.2527667870864155,
-      "grad_norm": 0.003986313473433256,
-      "learning_rate": 0.00019999996866143437,
-      "loss": 46.0,
-      "step": 3306
-    },
-    {
-      "epoch": 0.25284324406980524,
-      "grad_norm": 0.0003875916881952435,
-      "learning_rate": 0.00019999996864241535,
-      "loss": 46.0,
-      "step": 3307
-    },
-    {
-      "epoch": 0.25291970105319495,
-      "grad_norm": 0.000934297451749444,
-      "learning_rate": 0.00019999996862339058,
-      "loss": 46.0,
-      "step": 3308
-    },
-    {
-      "epoch": 0.25299615803658465,
-      "grad_norm": 0.0016030014958232641,
-      "learning_rate": 0.00019999996860436004,
-      "loss": 46.0,
-      "step": 3309
-    },
-    {
-      "epoch": 0.2530726150199744,
-      "grad_norm": 0.0007313450332731009,
-      "learning_rate": 0.00019999996858532373,
-      "loss": 46.0,
-      "step": 3310
-    },
-    {
-      "epoch": 0.2531490720033641,
-      "grad_norm": 0.0011905212886631489,
-      "learning_rate": 0.00019999996856628167,
-      "loss": 46.0,
-      "step": 3311
-    },
-    {
-      "epoch": 0.2532255289867538,
-      "grad_norm": 0.0008470544125884771,
-      "learning_rate": 0.00019999996854723378,
-      "loss": 46.0,
-      "step": 3312
-    },
-    {
-      "epoch": 0.25330198597014353,
-      "grad_norm": 0.0006716848001815379,
-      "learning_rate": 0.0001999999685281802,
-      "loss": 46.0,
-      "step": 3313
-    },
-    {
-      "epoch": 0.2533784429535333,
-      "grad_norm": 0.001264550955966115,
-      "learning_rate": 0.0001999999685091208,
-      "loss": 46.0,
-      "step": 3314
-    },
-    {
-      "epoch": 0.253454899936923,
-      "grad_norm": 0.0007389072561636567,
-      "learning_rate": 0.00019999996849005562,
-      "loss": 46.0,
-      "step": 3315
-    },
-    {
-      "epoch": 0.2535313569203127,
-      "grad_norm": 0.000758472247980535,
-      "learning_rate": 0.0001999999684709847,
-      "loss": 46.0,
-      "step": 3316
-    },
-    {
-      "epoch": 0.2536078139037024,
-      "grad_norm": 0.0005641333991661668,
-      "learning_rate": 0.00019999996845190802,
-      "loss": 46.0,
-      "step": 3317
-    },
-    {
-      "epoch": 0.25368427088709217,
-      "grad_norm": 0.0012269377475604415,
-      "learning_rate": 0.00019999996843282554,
-      "loss": 46.0,
-      "step": 3318
-    },
-    {
-      "epoch": 0.2537607278704819,
-      "grad_norm": 0.0006507251528091729,
-      "learning_rate": 0.0001999999684137373,
-      "loss": 46.0,
-      "step": 3319
-    },
-    {
-      "epoch": 0.2538371848538716,
-      "grad_norm": 0.0008744804072193801,
-      "learning_rate": 0.0001999999683946433,
-      "loss": 46.0,
-      "step": 3320
-    },
-    {
-      "epoch": 0.2539136418372613,
-      "grad_norm": 0.0023502083495259285,
-      "learning_rate": 0.00019999996837554354,
-      "loss": 46.0,
-      "step": 3321
-    },
-    {
-      "epoch": 0.25399009882065104,
-      "grad_norm": 0.0025126736145466566,
-      "learning_rate": 0.00019999996835643797,
-      "loss": 46.0,
-      "step": 3322
-    },
-    {
-      "epoch": 0.25406655580404075,
-      "grad_norm": 0.0009097993024624884,
-      "learning_rate": 0.00019999996833732666,
-      "loss": 46.0,
-      "step": 3323
-    },
-    {
-      "epoch": 0.25414301278743046,
-      "grad_norm": 0.0009338732343167067,
-      "learning_rate": 0.00019999996831820957,
-      "loss": 46.0,
-      "step": 3324
-    },
-    {
-      "epoch": 0.2542194697708202,
-      "grad_norm": 0.0004946420667693019,
-      "learning_rate": 0.00019999996829908674,
-      "loss": 46.0,
-      "step": 3325
-    },
-    {
-      "epoch": 0.2542959267542099,
-      "grad_norm": 0.0016363008180633187,
-      "learning_rate": 0.00019999996827995808,
-      "loss": 46.0,
-      "step": 3326
-    },
-    {
-      "epoch": 0.2543723837375996,
-      "grad_norm": 0.001478428253903985,
-      "learning_rate": 0.0001999999682608237,
-      "loss": 46.0,
-      "step": 3327
-    },
-    {
-      "epoch": 0.25444884072098933,
-      "grad_norm": 0.0009009524364955723,
-      "learning_rate": 0.00019999996824168355,
-      "loss": 46.0,
-      "step": 3328
-    },
-    {
-      "epoch": 0.2545252977043791,
-      "grad_norm": 0.0015700636431574821,
-      "learning_rate": 0.00019999996822253762,
-      "loss": 46.0,
-      "step": 3329
-    },
-    {
-      "epoch": 0.2546017546877688,
-      "grad_norm": 0.0005721982452087104,
-      "learning_rate": 0.00019999996820338592,
-      "loss": 46.0,
-      "step": 3330
-    },
-    {
-      "epoch": 0.2546782116711585,
-      "grad_norm": 0.0010893730213865638,
-      "learning_rate": 0.00019999996818422845,
-      "loss": 46.0,
-      "step": 3331
-    },
-    {
-      "epoch": 0.2547546686545482,
-      "grad_norm": 0.0008342974469996989,
-      "learning_rate": 0.00019999996816506517,
-      "loss": 46.0,
-      "step": 3332
-    },
-    {
-      "epoch": 0.25483112563793797,
-      "grad_norm": 0.001518531353212893,
-      "learning_rate": 0.00019999996814589618,
-      "loss": 46.0,
-      "step": 3333
-    },
-    {
-      "epoch": 0.2549075826213277,
-      "grad_norm": 0.0014199293218553066,
-      "learning_rate": 0.00019999996812672142,
-      "loss": 46.0,
-      "step": 3334
-    },
-    {
-      "epoch": 0.2549840396047174,
-      "grad_norm": 0.002698963973671198,
-      "learning_rate": 0.00019999996810754085,
-      "loss": 46.0,
-      "step": 3335
-    },
-    {
-      "epoch": 0.25506049658810714,
-      "grad_norm": 0.0032409175764769316,
-      "learning_rate": 0.00019999996808835454,
-      "loss": 46.0,
-      "step": 3336
-    },
-    {
-      "epoch": 0.25513695357149685,
-      "grad_norm": 0.0016743586165830493,
-      "learning_rate": 0.00019999996806916245,
-      "loss": 46.0,
-      "step": 3337
-    },
-    {
-      "epoch": 0.25521341055488656,
-      "grad_norm": 0.0012604790972545743,
-      "learning_rate": 0.0001999999680499646,
-      "loss": 46.0,
-      "step": 3338
-    },
-    {
-      "epoch": 0.25528986753827626,
-      "grad_norm": 0.0008487821905873716,
-      "learning_rate": 0.00019999996803076099,
-      "loss": 46.0,
-      "step": 3339
-    },
-    {
-      "epoch": 0.255366324521666,
-      "grad_norm": 0.0008831185987219214,
-      "learning_rate": 0.00019999996801155158,
-      "loss": 46.0,
-      "step": 3340
-    },
-    {
-      "epoch": 0.2554427815050557,
-      "grad_norm": 0.001443361397832632,
-      "learning_rate": 0.00019999996799233643,
-      "loss": 46.0,
-      "step": 3341
-    },
-    {
-      "epoch": 0.25551923848844543,
-      "grad_norm": 0.0013351291418075562,
-      "learning_rate": 0.00019999996797311547,
-      "loss": 46.0,
-      "step": 3342
-    },
-    {
-      "epoch": 0.25559569547183514,
-      "grad_norm": 0.0008088031318038702,
-      "learning_rate": 0.00019999996795388878,
-      "loss": 46.0,
-      "step": 3343
-    },
-    {
-      "epoch": 0.2556721524552249,
-      "grad_norm": 0.0010591468308120966,
-      "learning_rate": 0.0001999999679346563,
-      "loss": 46.0,
-      "step": 3344
-    },
-    {
-      "epoch": 0.2557486094386146,
-      "grad_norm": 0.0008869795128703117,
-      "learning_rate": 0.00019999996791541806,
-      "loss": 46.0,
-      "step": 3345
-    },
-    {
-      "epoch": 0.2558250664220043,
-      "grad_norm": 0.0041189962066709995,
-      "learning_rate": 0.00019999996789617406,
-      "loss": 46.0,
-      "step": 3346
-    },
-    {
-      "epoch": 0.255901523405394,
-      "grad_norm": 0.0007569058216176927,
-      "learning_rate": 0.00019999996787692424,
-      "loss": 46.0,
-      "step": 3347
-    },
-    {
-      "epoch": 0.2559779803887838,
-      "grad_norm": 0.0012949596857652068,
-      "learning_rate": 0.00019999996785766873,
-      "loss": 46.0,
-      "step": 3348
-    },
-    {
-      "epoch": 0.2560544373721735,
-      "grad_norm": 0.0011699702590703964,
-      "learning_rate": 0.0001999999678384074,
-      "loss": 46.0,
-      "step": 3349
-    },
-    {
-      "epoch": 0.2561308943555632,
-      "grad_norm": 0.0021525020711123943,
-      "learning_rate": 0.0001999999678191403,
-      "loss": 46.0,
-      "step": 3350
-    },
-    {
-      "epoch": 0.25620735133895295,
-      "grad_norm": 0.0006216112524271011,
-      "learning_rate": 0.00019999996779986745,
-      "loss": 46.0,
-      "step": 3351
-    },
-    {
-      "epoch": 0.25628380832234265,
-      "grad_norm": 0.002517369342967868,
-      "learning_rate": 0.00019999996778058882,
-      "loss": 46.0,
-      "step": 3352
-    },
-    {
-      "epoch": 0.25636026530573236,
-      "grad_norm": 0.0015303826658055186,
-      "learning_rate": 0.00019999996776130444,
-      "loss": 46.0,
-      "step": 3353
-    },
-    {
-      "epoch": 0.25643672228912207,
-      "grad_norm": 0.0005715960287488997,
-      "learning_rate": 0.00019999996774201426,
-      "loss": 46.0,
-      "step": 3354
-    },
-    {
-      "epoch": 0.2565131792725118,
-      "grad_norm": 0.0013456068700179458,
-      "learning_rate": 0.0001999999677227183,
-      "loss": 46.0,
-      "step": 3355
-    },
-    {
-      "epoch": 0.25658963625590153,
-      "grad_norm": 0.001387232099659741,
-      "learning_rate": 0.0001999999677034166,
-      "loss": 46.0,
-      "step": 3356
-    },
-    {
-      "epoch": 0.25666609323929124,
-      "grad_norm": 0.0013406374491751194,
-      "learning_rate": 0.0001999999676841091,
-      "loss": 46.0,
-      "step": 3357
-    },
-    {
-      "epoch": 0.25674255022268094,
-      "grad_norm": 0.00104979716707021,
-      "learning_rate": 0.0001999999676647959,
-      "loss": 46.0,
-      "step": 3358
-    },
-    {
-      "epoch": 0.2568190072060707,
-      "grad_norm": 0.00042975760879926383,
-      "learning_rate": 0.00019999996764547687,
-      "loss": 46.0,
-      "step": 3359
-    },
-    {
-      "epoch": 0.2568954641894604,
-      "grad_norm": 0.0016505069797858596,
-      "learning_rate": 0.0001999999676261521,
-      "loss": 46.0,
-      "step": 3360
-    },
-    {
-      "epoch": 0.2569719211728501,
-      "grad_norm": 0.001937379944138229,
-      "learning_rate": 0.0001999999676068215,
-      "loss": 46.0,
-      "step": 3361
-    },
-    {
-      "epoch": 0.2570483781562398,
-      "grad_norm": 0.0012581613846123219,
-      "learning_rate": 0.0001999999675874852,
-      "loss": 46.0,
-      "step": 3362
-    },
-    {
-      "epoch": 0.2571248351396296,
-      "grad_norm": 0.0006858674460090697,
-      "learning_rate": 0.0001999999675681431,
-      "loss": 46.0,
-      "step": 3363
-    },
-    {
-      "epoch": 0.2572012921230193,
-      "grad_norm": 0.0019825738854706287,
-      "learning_rate": 0.00019999996754879525,
-      "loss": 46.0,
-      "step": 3364
-    },
-    {
-      "epoch": 0.257277749106409,
-      "grad_norm": 0.0006939525483176112,
-      "learning_rate": 0.00019999996752944162,
-      "loss": 46.0,
-      "step": 3365
-    },
-    {
-      "epoch": 0.25735420608979875,
-      "grad_norm": 0.0012705805711448193,
-      "learning_rate": 0.00019999996751008224,
-      "loss": 46.0,
-      "step": 3366
-    },
-    {
-      "epoch": 0.25743066307318846,
-      "grad_norm": 0.0012182053178548813,
-      "learning_rate": 0.00019999996749071706,
-      "loss": 46.0,
-      "step": 3367
-    },
-    {
-      "epoch": 0.25750712005657816,
-      "grad_norm": 0.0008177760173566639,
-      "learning_rate": 0.0001999999674713461,
-      "loss": 46.0,
-      "step": 3368
-    },
-    {
-      "epoch": 0.25758357703996787,
-      "grad_norm": 0.0009514178382232785,
-      "learning_rate": 0.0001999999674519694,
-      "loss": 46.0,
-      "step": 3369
-    },
-    {
-      "epoch": 0.25766003402335763,
-      "grad_norm": 0.0012822678545489907,
-      "learning_rate": 0.00019999996743258694,
-      "loss": 46.0,
-      "step": 3370
-    },
-    {
-      "epoch": 0.25773649100674734,
-      "grad_norm": 0.0005474091740325093,
-      "learning_rate": 0.00019999996741319867,
-      "loss": 46.0,
-      "step": 3371
-    },
-    {
-      "epoch": 0.25781294799013704,
-      "grad_norm": 0.0008514283108524978,
-      "learning_rate": 0.00019999996739380468,
-      "loss": 46.0,
-      "step": 3372
-    },
-    {
-      "epoch": 0.25788940497352675,
-      "grad_norm": 0.002983875572681427,
-      "learning_rate": 0.00019999996737440486,
-      "loss": 46.0,
-      "step": 3373
-    },
-    {
-      "epoch": 0.2579658619569165,
-      "grad_norm": 0.001140502979978919,
-      "learning_rate": 0.00019999996735499932,
-      "loss": 46.0,
-      "step": 3374
-    },
-    {
-      "epoch": 0.2580423189403062,
-      "grad_norm": 0.0012950376840308309,
-      "learning_rate": 0.00019999996733558798,
-      "loss": 46.0,
-      "step": 3375
-    },
-    {
-      "epoch": 0.2581187759236959,
-      "grad_norm": 0.0007958801579661667,
-      "learning_rate": 0.0001999999673161709,
-      "loss": 46.0,
-      "step": 3376
-    },
-    {
-      "epoch": 0.2581952329070856,
-      "grad_norm": 0.0012902435846626759,
-      "learning_rate": 0.000199999967296748,
-      "loss": 46.0,
-      "step": 3377
-    },
-    {
-      "epoch": 0.2582716898904754,
-      "grad_norm": 0.002661551581695676,
-      "learning_rate": 0.00019999996727731938,
-      "loss": 46.0,
-      "step": 3378
-    },
-    {
-      "epoch": 0.2583481468738651,
-      "grad_norm": 0.003200280712917447,
-      "learning_rate": 0.000199999967257885,
-      "loss": 46.0,
-      "step": 3379
-    },
-    {
-      "epoch": 0.2584246038572548,
-      "grad_norm": 0.0022828904911875725,
-      "learning_rate": 0.0001999999672384448,
-      "loss": 46.0,
-      "step": 3380
-    },
-    {
-      "epoch": 0.25850106084064456,
-      "grad_norm": 0.0009272780735045671,
-      "learning_rate": 0.00019999996721899884,
-      "loss": 46.0,
-      "step": 3381
-    },
-    {
-      "epoch": 0.25857751782403426,
-      "grad_norm": 0.0005332891596481204,
-      "learning_rate": 0.00019999996719954714,
-      "loss": 46.0,
-      "step": 3382
-    },
-    {
-      "epoch": 0.25865397480742397,
-      "grad_norm": 0.0013827787479385734,
-      "learning_rate": 0.00019999996718008967,
-      "loss": 46.0,
-      "step": 3383
-    },
-    {
-      "epoch": 0.2587304317908137,
-      "grad_norm": 0.000758643785957247,
-      "learning_rate": 0.0001999999671606264,
-      "loss": 46.0,
-      "step": 3384
-    },
-    {
-      "epoch": 0.25880688877420344,
-      "grad_norm": 0.002129457425326109,
-      "learning_rate": 0.00019999996714115738,
-      "loss": 46.0,
-      "step": 3385
-    },
-    {
-      "epoch": 0.25888334575759314,
-      "grad_norm": 0.0003911287640221417,
-      "learning_rate": 0.00019999996712168262,
-      "loss": 46.0,
-      "step": 3386
-    },
-    {
-      "epoch": 0.25895980274098285,
-      "grad_norm": 0.0008566020987927914,
-      "learning_rate": 0.00019999996710220205,
-      "loss": 46.0,
-      "step": 3387
-    },
-    {
-      "epoch": 0.25903625972437255,
-      "grad_norm": 0.001327891950495541,
-      "learning_rate": 0.00019999996708271571,
-      "loss": 46.0,
-      "step": 3388
-    },
-    {
-      "epoch": 0.2591127167077623,
-      "grad_norm": 0.0014818107010796666,
-      "learning_rate": 0.0001999999670632236,
-      "loss": 46.0,
-      "step": 3389
-    },
-    {
-      "epoch": 0.259189173691152,
-      "grad_norm": 0.001240147859789431,
-      "learning_rate": 0.00019999996704372574,
-      "loss": 46.0,
-      "step": 3390
-    },
-    {
-      "epoch": 0.2592656306745417,
-      "grad_norm": 0.0015240032225847244,
-      "learning_rate": 0.0001999999670242221,
-      "loss": 46.0,
-      "step": 3391
-    },
-    {
-      "epoch": 0.25934208765793143,
-      "grad_norm": 0.0010145625565201044,
-      "learning_rate": 0.0001999999670047127,
-      "loss": 46.0,
-      "step": 3392
-    },
-    {
-      "epoch": 0.2594185446413212,
-      "grad_norm": 0.005682583898305893,
-      "learning_rate": 0.00019999996698519753,
-      "loss": 46.0,
-      "step": 3393
-    },
-    {
-      "epoch": 0.2594950016247109,
-      "grad_norm": 0.0016105679096654058,
-      "learning_rate": 0.00019999996696567658,
-      "loss": 46.0,
-      "step": 3394
-    },
-    {
-      "epoch": 0.2595714586081006,
-      "grad_norm": 0.00141436536796391,
-      "learning_rate": 0.00019999996694614985,
-      "loss": 46.0,
-      "step": 3395
-    },
-    {
-      "epoch": 0.25964791559149036,
-      "grad_norm": 0.0061468142084777355,
-      "learning_rate": 0.00019999996692661738,
-      "loss": 46.0,
-      "step": 3396
-    },
-    {
-      "epoch": 0.25972437257488007,
-      "grad_norm": 0.0004221137205604464,
-      "learning_rate": 0.0001999999669070791,
-      "loss": 46.0,
-      "step": 3397
-    },
-    {
-      "epoch": 0.2598008295582698,
-      "grad_norm": 0.0006950510432943702,
-      "learning_rate": 0.00019999996688753512,
-      "loss": 46.0,
-      "step": 3398
-    },
-    {
-      "epoch": 0.2598772865416595,
-      "grad_norm": 0.0022589629516005516,
-      "learning_rate": 0.00019999996686798533,
-      "loss": 46.0,
-      "step": 3399
-    },
-    {
-      "epoch": 0.25995374352504924,
-      "grad_norm": 0.0009185707895085216,
-      "learning_rate": 0.00019999996684842973,
-      "loss": 46.0,
-      "step": 3400
-    },
-    {
-      "epoch": 0.26003020050843895,
-      "grad_norm": 0.0014534504152834415,
-      "learning_rate": 0.00019999996682886842,
-      "loss": 46.0,
-      "step": 3401
-    },
-    {
-      "epoch": 0.26010665749182865,
-      "grad_norm": 0.0030209224205464125,
-      "learning_rate": 0.0001999999668093013,
-      "loss": 46.0,
-      "step": 3402
-    },
-    {
-      "epoch": 0.26018311447521836,
-      "grad_norm": 0.0006988857639953494,
-      "learning_rate": 0.00019999996678972845,
-      "loss": 46.0,
-      "step": 3403
-    },
-    {
-      "epoch": 0.2602595714586081,
-      "grad_norm": 0.0010906365932896733,
-      "learning_rate": 0.0001999999667701498,
-      "loss": 46.0,
-      "step": 3404
-    },
-    {
-      "epoch": 0.2603360284419978,
-      "grad_norm": 0.0008640087908133864,
-      "learning_rate": 0.0001999999667505654,
-      "loss": 46.0,
-      "step": 3405
-    },
-    {
-      "epoch": 0.26041248542538753,
-      "grad_norm": 0.0005684989737346768,
-      "learning_rate": 0.0001999999667309752,
-      "loss": 46.0,
-      "step": 3406
-    },
-    {
-      "epoch": 0.26048894240877724,
-      "grad_norm": 0.0006818450638093054,
-      "learning_rate": 0.00019999996671137926,
-      "loss": 46.0,
-      "step": 3407
-    },
-    {
-      "epoch": 0.260565399392167,
-      "grad_norm": 0.0009807914029806852,
-      "learning_rate": 0.00019999996669177754,
-      "loss": 46.0,
-      "step": 3408
-    },
-    {
-      "epoch": 0.2606418563755567,
-      "grad_norm": 0.0005603322060778737,
-      "learning_rate": 0.00019999996667217004,
-      "loss": 46.0,
-      "step": 3409
-    },
-    {
-      "epoch": 0.2607183133589464,
-      "grad_norm": 0.0016643533017486334,
-      "learning_rate": 0.0001999999666525568,
-      "loss": 46.0,
-      "step": 3410
-    },
-    {
-      "epoch": 0.26079477034233617,
-      "grad_norm": 0.001112714409828186,
-      "learning_rate": 0.00019999996663293778,
-      "loss": 46.0,
-      "step": 3411
-    },
-    {
-      "epoch": 0.2608712273257259,
-      "grad_norm": 0.0004928434500470757,
-      "learning_rate": 0.00019999996661331299,
-      "loss": 46.0,
-      "step": 3412
-    },
-    {
-      "epoch": 0.2609476843091156,
-      "grad_norm": 0.0012750652385875583,
-      "learning_rate": 0.00019999996659368242,
-      "loss": 46.0,
-      "step": 3413
-    },
-    {
-      "epoch": 0.2610241412925053,
-      "grad_norm": 0.0012590659316629171,
-      "learning_rate": 0.00019999996657404608,
-      "loss": 46.0,
-      "step": 3414
-    },
-    {
-      "epoch": 0.26110059827589505,
-      "grad_norm": 0.001027178717777133,
-      "learning_rate": 0.00019999996655440397,
-      "loss": 46.0,
-      "step": 3415
-    },
-    {
-      "epoch": 0.26117705525928475,
-      "grad_norm": 0.0008961145649664104,
-      "learning_rate": 0.0001999999665347561,
-      "loss": 46.0,
-      "step": 3416
-    },
-    {
-      "epoch": 0.26125351224267446,
-      "grad_norm": 0.0015716425841674209,
-      "learning_rate": 0.00019999996651510246,
-      "loss": 46.0,
-      "step": 3417
-    },
-    {
-      "epoch": 0.26132996922606416,
-      "grad_norm": 0.002116632182151079,
-      "learning_rate": 0.00019999996649544302,
-      "loss": 46.0,
-      "step": 3418
-    },
-    {
-      "epoch": 0.2614064262094539,
-      "grad_norm": 0.001458898652344942,
-      "learning_rate": 0.00019999996647577785,
-      "loss": 46.0,
-      "step": 3419
-    },
-    {
-      "epoch": 0.26148288319284363,
-      "grad_norm": 0.0010803916957229376,
-      "learning_rate": 0.0001999999664561069,
-      "loss": 46.0,
-      "step": 3420
-    },
-    {
-      "epoch": 0.26155934017623333,
-      "grad_norm": 0.000599379651248455,
-      "learning_rate": 0.00019999996643643017,
-      "loss": 46.0,
-      "step": 3421
-    },
-    {
-      "epoch": 0.26163579715962304,
-      "grad_norm": 0.001198024838231504,
-      "learning_rate": 0.0001999999664167477,
-      "loss": 46.0,
-      "step": 3422
-    },
-    {
-      "epoch": 0.2617122541430128,
-      "grad_norm": 0.0007421146729029715,
-      "learning_rate": 0.00019999996639705946,
-      "loss": 46.0,
-      "step": 3423
-    },
-    {
-      "epoch": 0.2617887111264025,
-      "grad_norm": 0.0006218171329237521,
-      "learning_rate": 0.0001999999663773654,
-      "loss": 46.0,
-      "step": 3424
-    },
-    {
-      "epoch": 0.2618651681097922,
-      "grad_norm": 0.0023230707738548517,
-      "learning_rate": 0.0001999999663576656,
-      "loss": 46.0,
-      "step": 3425
-    },
-    {
-      "epoch": 0.261941625093182,
-      "grad_norm": 0.0007635362562723458,
-      "learning_rate": 0.00019999996633796006,
-      "loss": 46.0,
-      "step": 3426
-    },
-    {
-      "epoch": 0.2620180820765717,
-      "grad_norm": 0.002543106907978654,
-      "learning_rate": 0.00019999996631824872,
-      "loss": 46.0,
-      "step": 3427
-    },
-    {
-      "epoch": 0.2620945390599614,
-      "grad_norm": 0.0026661870069801807,
-      "learning_rate": 0.00019999996629853158,
-      "loss": 46.0,
-      "step": 3428
-    },
-    {
-      "epoch": 0.2621709960433511,
-      "grad_norm": 0.0017379287164658308,
-      "learning_rate": 0.0001999999662788087,
-      "loss": 46.0,
-      "step": 3429
-    },
-    {
-      "epoch": 0.26224745302674085,
-      "grad_norm": 0.0009609837434254587,
-      "learning_rate": 0.0001999999662590801,
-      "loss": 46.0,
-      "step": 3430
-    },
-    {
-      "epoch": 0.26232391001013056,
-      "grad_norm": 0.0023372580762952566,
-      "learning_rate": 0.00019999996623934564,
-      "loss": 46.0,
-      "step": 3431
-    },
-    {
-      "epoch": 0.26240036699352026,
-      "grad_norm": 0.0008000805974006653,
-      "learning_rate": 0.00019999996621960549,
-      "loss": 46.0,
-      "step": 3432
-    },
-    {
-      "epoch": 0.26247682397690997,
-      "grad_norm": 0.0016166317509487271,
-      "learning_rate": 0.0001999999661998595,
-      "loss": 46.0,
-      "step": 3433
-    },
-    {
-      "epoch": 0.26255328096029973,
-      "grad_norm": 0.002178431721404195,
-      "learning_rate": 0.00019999996618010778,
-      "loss": 46.0,
-      "step": 3434
-    },
-    {
-      "epoch": 0.26262973794368943,
-      "grad_norm": 0.0009274385520257056,
-      "learning_rate": 0.0001999999661603503,
-      "loss": 46.0,
-      "step": 3435
-    },
-    {
-      "epoch": 0.26270619492707914,
-      "grad_norm": 0.0010919689666479826,
-      "learning_rate": 0.00019999996614058704,
-      "loss": 46.0,
-      "step": 3436
-    },
-    {
-      "epoch": 0.26278265191046885,
-      "grad_norm": 0.0012274044565856457,
-      "learning_rate": 0.00019999996612081802,
-      "loss": 46.0,
-      "step": 3437
-    },
-    {
-      "epoch": 0.2628591088938586,
-      "grad_norm": 0.0015083723701536655,
-      "learning_rate": 0.0001999999661010432,
-      "loss": 46.0,
-      "step": 3438
-    },
-    {
-      "epoch": 0.2629355658772483,
-      "grad_norm": 0.0014632773818448186,
-      "learning_rate": 0.00019999996608126264,
-      "loss": 46.0,
-      "step": 3439
-    },
-    {
-      "epoch": 0.263012022860638,
-      "grad_norm": 0.0009205866372212768,
-      "learning_rate": 0.00019999996606147628,
-      "loss": 46.0,
-      "step": 3440
-    },
-    {
-      "epoch": 0.2630884798440278,
-      "grad_norm": 0.001350146485492587,
-      "learning_rate": 0.00019999996604168417,
-      "loss": 46.0,
-      "step": 3441
-    },
-    {
-      "epoch": 0.2631649368274175,
-      "grad_norm": 0.0011877267388626933,
-      "learning_rate": 0.0001999999660218863,
-      "loss": 46.0,
-      "step": 3442
-    },
-    {
-      "epoch": 0.2632413938108072,
-      "grad_norm": 0.0008765162201598287,
-      "learning_rate": 0.00019999996600208263,
-      "loss": 46.0,
-      "step": 3443
-    },
-    {
-      "epoch": 0.2633178507941969,
-      "grad_norm": 0.026009637862443924,
-      "learning_rate": 0.00019999996598227325,
-      "loss": 46.0,
-      "step": 3444
-    },
-    {
-      "epoch": 0.26339430777758666,
-      "grad_norm": 0.0007201490807347,
-      "learning_rate": 0.00019999996596245802,
-      "loss": 46.0,
-      "step": 3445
-    },
-    {
-      "epoch": 0.26347076476097636,
-      "grad_norm": 0.0004314627149142325,
-      "learning_rate": 0.00019999996594263707,
-      "loss": 46.0,
-      "step": 3446
-    },
-    {
-      "epoch": 0.26354722174436607,
-      "grad_norm": 0.0009992282139137387,
-      "learning_rate": 0.00019999996592281037,
-      "loss": 46.0,
-      "step": 3447
-    },
-    {
-      "epoch": 0.2636236787277558,
-      "grad_norm": 0.0012712192256003618,
-      "learning_rate": 0.00019999996590297787,
-      "loss": 46.0,
-      "step": 3448
-    },
-    {
-      "epoch": 0.26370013571114553,
-      "grad_norm": 0.0021229113917797804,
-      "learning_rate": 0.0001999999658831396,
-      "loss": 46.0,
-      "step": 3449
-    },
-    {
-      "epoch": 0.26377659269453524,
-      "grad_norm": 0.0008281507762148976,
-      "learning_rate": 0.00019999996586329556,
-      "loss": 46.0,
-      "step": 3450
-    },
-    {
-      "epoch": 0.26385304967792494,
-      "grad_norm": 0.0014178858837112784,
-      "learning_rate": 0.00019999996584344577,
-      "loss": 46.0,
-      "step": 3451
-    },
-    {
-      "epoch": 0.26392950666131465,
-      "grad_norm": 0.003944141324609518,
-      "learning_rate": 0.00019999996582359018,
-      "loss": 46.0,
-      "step": 3452
-    },
-    {
-      "epoch": 0.2640059636447044,
-      "grad_norm": 0.0011372524313628674,
-      "learning_rate": 0.00019999996580372884,
-      "loss": 46.0,
-      "step": 3453
-    },
-    {
-      "epoch": 0.2640824206280941,
-      "grad_norm": 0.0032555984798818827,
-      "learning_rate": 0.00019999996578386173,
-      "loss": 46.0,
-      "step": 3454
-    },
-    {
-      "epoch": 0.2641588776114838,
-      "grad_norm": 0.0027557043358683586,
-      "learning_rate": 0.00019999996576398882,
-      "loss": 46.0,
-      "step": 3455
-    },
-    {
-      "epoch": 0.2642353345948736,
-      "grad_norm": 0.0008548556361347437,
-      "learning_rate": 0.0001999999657441102,
-      "loss": 46.0,
-      "step": 3456
-    },
-    {
-      "epoch": 0.2643117915782633,
-      "grad_norm": 0.0005990201607346535,
-      "learning_rate": 0.00019999996572422576,
-      "loss": 46.0,
-      "step": 3457
-    },
-    {
-      "epoch": 0.264388248561653,
-      "grad_norm": 0.0007861117483116686,
-      "learning_rate": 0.00019999996570433556,
-      "loss": 46.0,
-      "step": 3458
-    },
-    {
-      "epoch": 0.2644647055450427,
-      "grad_norm": 0.0010544734541326761,
-      "learning_rate": 0.0001999999656844396,
-      "loss": 46.0,
-      "step": 3459
-    },
-    {
-      "epoch": 0.26454116252843246,
-      "grad_norm": 0.004305166658014059,
-      "learning_rate": 0.00019999996566453788,
-      "loss": 46.0,
-      "step": 3460
-    },
-    {
-      "epoch": 0.26461761951182217,
-      "grad_norm": 0.007666804827749729,
-      "learning_rate": 0.00019999996564463039,
-      "loss": 46.0,
-      "step": 3461
-    },
-    {
-      "epoch": 0.26469407649521187,
-      "grad_norm": 0.0017637485871091485,
-      "learning_rate": 0.00019999996562471712,
-      "loss": 46.0,
-      "step": 3462
-    },
-    {
-      "epoch": 0.2647705334786016,
-      "grad_norm": 0.000689836626406759,
-      "learning_rate": 0.00019999996560479807,
-      "loss": 46.0,
-      "step": 3463
-    },
-    {
-      "epoch": 0.26484699046199134,
-      "grad_norm": 0.0011632087407633662,
-      "learning_rate": 0.00019999996558487326,
-      "loss": 46.0,
-      "step": 3464
-    },
-    {
-      "epoch": 0.26492344744538104,
-      "grad_norm": 0.001704981317743659,
-      "learning_rate": 0.0001999999655649427,
-      "loss": 46.0,
-      "step": 3465
-    },
-    {
-      "epoch": 0.26499990442877075,
-      "grad_norm": 0.0023453945759683847,
-      "learning_rate": 0.00019999996554500636,
-      "loss": 46.0,
-      "step": 3466
-    },
-    {
-      "epoch": 0.2650763614121605,
-      "grad_norm": 0.00347949774004519,
-      "learning_rate": 0.00019999996552506425,
-      "loss": 46.0,
-      "step": 3467
-    },
-    {
-      "epoch": 0.2651528183955502,
-      "grad_norm": 0.0012835649540647864,
-      "learning_rate": 0.00019999996550511636,
-      "loss": 46.0,
-      "step": 3468
-    },
-    {
-      "epoch": 0.2652292753789399,
-      "grad_norm": 0.001120152068324387,
-      "learning_rate": 0.00019999996548516268,
-      "loss": 46.0,
-      "step": 3469
-    },
-    {
-      "epoch": 0.2653057323623296,
-      "grad_norm": 0.0005451901233755052,
-      "learning_rate": 0.00019999996546520325,
-      "loss": 46.0,
-      "step": 3470
-    },
-    {
-      "epoch": 0.2653821893457194,
-      "grad_norm": 0.00048714099102653563,
-      "learning_rate": 0.00019999996544523807,
-      "loss": 46.0,
-      "step": 3471
-    },
-    {
-      "epoch": 0.2654586463291091,
-      "grad_norm": 0.0031618874054402113,
-      "learning_rate": 0.00019999996542526712,
-      "loss": 46.0,
-      "step": 3472
-    },
-    {
-      "epoch": 0.2655351033124988,
-      "grad_norm": 0.00133831228595227,
-      "learning_rate": 0.00019999996540529037,
-      "loss": 46.0,
-      "step": 3473
-    },
-    {
-      "epoch": 0.2656115602958885,
-      "grad_norm": 0.002463582204654813,
-      "learning_rate": 0.00019999996538530787,
-      "loss": 46.0,
-      "step": 3474
-    },
-    {
-      "epoch": 0.26568801727927827,
-      "grad_norm": 0.0008574210805818439,
-      "learning_rate": 0.0001999999653653196,
-      "loss": 46.0,
-      "step": 3475
-    },
-    {
-      "epoch": 0.26576447426266797,
-      "grad_norm": 0.0006142286583781242,
-      "learning_rate": 0.00019999996534532553,
-      "loss": 46.0,
-      "step": 3476
-    },
-    {
-      "epoch": 0.2658409312460577,
-      "grad_norm": 0.002048185793682933,
-      "learning_rate": 0.00019999996532532577,
-      "loss": 46.0,
-      "step": 3477
-    },
-    {
-      "epoch": 0.2659173882294474,
-      "grad_norm": 0.0017892924370244145,
-      "learning_rate": 0.00019999996530532018,
-      "loss": 46.0,
-      "step": 3478
-    },
-    {
-      "epoch": 0.26599384521283714,
-      "grad_norm": 0.0006220433278940618,
-      "learning_rate": 0.00019999996528530882,
-      "loss": 46.0,
-      "step": 3479
-    },
-    {
-      "epoch": 0.26607030219622685,
-      "grad_norm": 0.001623353105969727,
-      "learning_rate": 0.0001999999652652917,
-      "loss": 46.0,
-      "step": 3480
-    },
-    {
-      "epoch": 0.26614675917961655,
-      "grad_norm": 0.00040237829671241343,
-      "learning_rate": 0.00019999996524526883,
-      "loss": 46.0,
-      "step": 3481
-    },
-    {
-      "epoch": 0.2662232161630063,
-      "grad_norm": 0.0005781952640973032,
-      "learning_rate": 0.00019999996522524017,
-      "loss": 46.0,
-      "step": 3482
-    },
-    {
-      "epoch": 0.266299673146396,
-      "grad_norm": 0.003110188990831375,
-      "learning_rate": 0.0001999999652052057,
-      "loss": 46.0,
-      "step": 3483
-    },
-    {
-      "epoch": 0.2663761301297857,
-      "grad_norm": 0.0012962182518094778,
-      "learning_rate": 0.00019999996518516554,
-      "loss": 46.0,
-      "step": 3484
-    },
-    {
-      "epoch": 0.26645258711317543,
-      "grad_norm": 0.0006347584421746433,
-      "learning_rate": 0.00019999996516511956,
-      "loss": 46.0,
-      "step": 3485
-    },
-    {
-      "epoch": 0.2665290440965652,
-      "grad_norm": 0.0017043203115463257,
-      "learning_rate": 0.00019999996514506784,
-      "loss": 46.0,
-      "step": 3486
-    },
-    {
-      "epoch": 0.2666055010799549,
-      "grad_norm": 0.005442338529974222,
-      "learning_rate": 0.0001999999651250103,
-      "loss": 46.0,
-      "step": 3487
-    },
-    {
-      "epoch": 0.2666819580633446,
-      "grad_norm": 0.0015704495599493384,
-      "learning_rate": 0.00019999996510494704,
-      "loss": 46.0,
-      "step": 3488
-    },
-    {
-      "epoch": 0.2667584150467343,
-      "grad_norm": 0.0008896812214516103,
-      "learning_rate": 0.00019999996508487803,
-      "loss": 46.0,
-      "step": 3489
-    },
-    {
-      "epoch": 0.26683487203012407,
-      "grad_norm": 0.000847585208248347,
-      "learning_rate": 0.00019999996506480318,
-      "loss": 46.0,
-      "step": 3490
-    },
-    {
-      "epoch": 0.2669113290135138,
-      "grad_norm": 0.0005499208928085864,
-      "learning_rate": 0.00019999996504472262,
-      "loss": 46.0,
-      "step": 3491
-    },
-    {
-      "epoch": 0.2669877859969035,
-      "grad_norm": 0.0010366961359977722,
-      "learning_rate": 0.00019999996502463626,
-      "loss": 46.0,
-      "step": 3492
-    },
-    {
-      "epoch": 0.2670642429802932,
-      "grad_norm": 0.0011850334703922272,
-      "learning_rate": 0.00019999996500454415,
-      "loss": 46.0,
-      "step": 3493
-    },
-    {
-      "epoch": 0.26714069996368295,
-      "grad_norm": 0.0008601018344052136,
-      "learning_rate": 0.00019999996498444624,
-      "loss": 46.0,
-      "step": 3494
-    },
-    {
-      "epoch": 0.26721715694707265,
-      "grad_norm": 0.0005690960679203272,
-      "learning_rate": 0.00019999996496434258,
-      "loss": 46.0,
-      "step": 3495
-    },
-    {
-      "epoch": 0.26729361393046236,
-      "grad_norm": 0.0008795912726782262,
-      "learning_rate": 0.00019999996494423315,
-      "loss": 46.0,
-      "step": 3496
-    },
-    {
-      "epoch": 0.2673700709138521,
-      "grad_norm": 0.0015382609562948346,
-      "learning_rate": 0.00019999996492411795,
-      "loss": 46.0,
-      "step": 3497
-    },
-    {
-      "epoch": 0.2674465278972418,
-      "grad_norm": 0.0005444943089969456,
-      "learning_rate": 0.00019999996490399697,
-      "loss": 46.0,
-      "step": 3498
-    },
-    {
-      "epoch": 0.26752298488063153,
-      "grad_norm": 0.008633997291326523,
-      "learning_rate": 0.00019999996488387025,
-      "loss": 46.0,
-      "step": 3499
-    },
-    {
-      "epoch": 0.26759944186402124,
-      "grad_norm": 0.0038967877626419067,
-      "learning_rate": 0.00019999996486373776,
-      "loss": 46.0,
-      "step": 3500
-    },
-    {
-      "epoch": 0.267675898847411,
-      "grad_norm": 0.0012343310518190265,
-      "learning_rate": 0.00019999996484359946,
-      "loss": 46.0,
-      "step": 3501
-    },
-    {
-      "epoch": 0.2677523558308007,
-      "grad_norm": 0.007553554605692625,
-      "learning_rate": 0.00019999996482345542,
-      "loss": 46.0,
-      "step": 3502
-    },
-    {
-      "epoch": 0.2678288128141904,
-      "grad_norm": 0.0018467012559995055,
-      "learning_rate": 0.0001999999648033056,
-      "loss": 46.0,
-      "step": 3503
-    },
-    {
-      "epoch": 0.2679052697975801,
-      "grad_norm": 0.000622952647972852,
-      "learning_rate": 0.00019999996478315,
-      "loss": 46.0,
-      "step": 3504
-    },
-    {
-      "epoch": 0.2679817267809699,
-      "grad_norm": 0.0019059015903621912,
-      "learning_rate": 0.00019999996476298865,
-      "loss": 46.0,
-      "step": 3505
-    },
-    {
-      "epoch": 0.2680581837643596,
-      "grad_norm": 0.000873995537403971,
-      "learning_rate": 0.00019999996474282154,
-      "loss": 46.0,
-      "step": 3506
-    },
-    {
-      "epoch": 0.2681346407477493,
-      "grad_norm": 0.004310471937060356,
-      "learning_rate": 0.00019999996472264863,
-      "loss": 46.0,
-      "step": 3507
-    },
-    {
-      "epoch": 0.268211097731139,
-      "grad_norm": 0.0005282750935293734,
-      "learning_rate": 0.00019999996470246997,
-      "loss": 46.0,
-      "step": 3508
-    },
-    {
-      "epoch": 0.26828755471452875,
-      "grad_norm": 0.0006494536646641791,
-      "learning_rate": 0.00019999996468228554,
-      "loss": 46.0,
-      "step": 3509
-    },
-    {
-      "epoch": 0.26836401169791846,
-      "grad_norm": 0.001211096765473485,
-      "learning_rate": 0.00019999996466209534,
-      "loss": 46.0,
-      "step": 3510
-    },
-    {
-      "epoch": 0.26844046868130816,
-      "grad_norm": 0.0005752411670982838,
-      "learning_rate": 0.0001999999646418994,
-      "loss": 46.0,
-      "step": 3511
-    },
-    {
-      "epoch": 0.2685169256646979,
-      "grad_norm": 0.0015640348428860307,
-      "learning_rate": 0.00019999996462169764,
-      "loss": 46.0,
-      "step": 3512
-    },
-    {
-      "epoch": 0.26859338264808763,
-      "grad_norm": 0.0029215794056653976,
-      "learning_rate": 0.00019999996460149012,
-      "loss": 46.0,
-      "step": 3513
-    },
-    {
-      "epoch": 0.26866983963147734,
-      "grad_norm": 0.0006118442979641259,
-      "learning_rate": 0.00019999996458127683,
-      "loss": 46.0,
-      "step": 3514
-    },
-    {
-      "epoch": 0.26874629661486704,
-      "grad_norm": 0.001395733910612762,
-      "learning_rate": 0.00019999996456105778,
-      "loss": 46.0,
-      "step": 3515
-    },
-    {
-      "epoch": 0.2688227535982568,
-      "grad_norm": 0.0012524726334959269,
-      "learning_rate": 0.00019999996454083297,
-      "loss": 46.0,
-      "step": 3516
-    },
-    {
-      "epoch": 0.2688992105816465,
-      "grad_norm": 0.001530499430373311,
-      "learning_rate": 0.00019999996452060238,
-      "loss": 46.0,
-      "step": 3517
-    },
-    {
-      "epoch": 0.2689756675650362,
-      "grad_norm": 0.001576151349581778,
-      "learning_rate": 0.00019999996450036602,
-      "loss": 46.0,
-      "step": 3518
-    },
-    {
-      "epoch": 0.2690521245484259,
-      "grad_norm": 0.0015910296933725476,
-      "learning_rate": 0.0001999999644801239,
-      "loss": 46.0,
-      "step": 3519
-    },
-    {
-      "epoch": 0.2691285815318157,
-      "grad_norm": 0.001185716944746673,
-      "learning_rate": 0.000199999964459876,
-      "loss": 46.0,
-      "step": 3520
-    },
-    {
-      "epoch": 0.2692050385152054,
-      "grad_norm": 0.0006350938929244876,
-      "learning_rate": 0.00019999996443962234,
-      "loss": 46.0,
-      "step": 3521
-    },
-    {
-      "epoch": 0.2692814954985951,
-      "grad_norm": 0.0006047820206731558,
-      "learning_rate": 0.0001999999644193629,
-      "loss": 46.0,
-      "step": 3522
-    },
-    {
-      "epoch": 0.2693579524819848,
-      "grad_norm": 0.0033772122114896774,
-      "learning_rate": 0.00019999996439909768,
-      "loss": 46.0,
-      "step": 3523
-    },
-    {
-      "epoch": 0.26943440946537456,
-      "grad_norm": 0.0004474870511330664,
-      "learning_rate": 0.0001999999643788267,
-      "loss": 46.0,
-      "step": 3524
-    },
-    {
-      "epoch": 0.26951086644876426,
-      "grad_norm": 0.0006406800821423531,
-      "learning_rate": 0.00019999996435854999,
-      "loss": 46.0,
-      "step": 3525
-    },
-    {
-      "epoch": 0.26958732343215397,
-      "grad_norm": 0.0016108149429783225,
-      "learning_rate": 0.00019999996433826746,
-      "loss": 46.0,
-      "step": 3526
-    },
-    {
-      "epoch": 0.26966378041554373,
-      "grad_norm": 0.0013555943733081222,
-      "learning_rate": 0.00019999996431797917,
-      "loss": 46.0,
-      "step": 3527
-    },
-    {
-      "epoch": 0.26974023739893344,
-      "grad_norm": 0.001304943347349763,
-      "learning_rate": 0.00019999996429768513,
-      "loss": 46.0,
-      "step": 3528
-    },
-    {
-      "epoch": 0.26981669438232314,
-      "grad_norm": 0.0015969887608662248,
-      "learning_rate": 0.0001999999642773853,
-      "loss": 46.0,
-      "step": 3529
-    },
-    {
-      "epoch": 0.26989315136571285,
-      "grad_norm": 0.0004311311640776694,
-      "learning_rate": 0.0001999999642570797,
-      "loss": 46.0,
-      "step": 3530
-    },
-    {
-      "epoch": 0.2699696083491026,
-      "grad_norm": 0.0009083780460059643,
-      "learning_rate": 0.00019999996423676833,
-      "loss": 46.0,
-      "step": 3531
-    },
-    {
-      "epoch": 0.2700460653324923,
-      "grad_norm": 0.0010997372446581721,
-      "learning_rate": 0.00019999996421645122,
-      "loss": 46.0,
-      "step": 3532
-    },
-    {
-      "epoch": 0.270122522315882,
-      "grad_norm": 0.0009625137900002301,
-      "learning_rate": 0.00019999996419612831,
-      "loss": 46.0,
-      "step": 3533
-    },
-    {
-      "epoch": 0.2701989792992717,
-      "grad_norm": 0.000788743665907532,
-      "learning_rate": 0.00019999996417579966,
-      "loss": 46.0,
-      "step": 3534
-    },
-    {
-      "epoch": 0.2702754362826615,
-      "grad_norm": 0.0027825627475976944,
-      "learning_rate": 0.0001999999641554652,
-      "loss": 46.0,
-      "step": 3535
-    },
-    {
-      "epoch": 0.2703518932660512,
-      "grad_norm": 0.001018956070765853,
-      "learning_rate": 0.00019999996413512497,
-      "loss": 46.0,
-      "step": 3536
-    },
-    {
-      "epoch": 0.2704283502494409,
-      "grad_norm": 0.012079456821084023,
-      "learning_rate": 0.00019999996411477903,
-      "loss": 46.0,
-      "step": 3537
-    },
-    {
-      "epoch": 0.2705048072328306,
-      "grad_norm": 0.0037696314975619316,
-      "learning_rate": 0.00019999996409442728,
-      "loss": 46.0,
-      "step": 3538
-    },
-    {
-      "epoch": 0.27058126421622036,
-      "grad_norm": 0.001351679558865726,
-      "learning_rate": 0.00019999996407406976,
-      "loss": 46.0,
-      "step": 3539
-    },
-    {
-      "epoch": 0.27065772119961007,
-      "grad_norm": 0.0011624229373410344,
-      "learning_rate": 0.00019999996405370646,
-      "loss": 46.0,
-      "step": 3540
-    },
-    {
-      "epoch": 0.2707341781829998,
-      "grad_norm": 0.0011623959289863706,
-      "learning_rate": 0.00019999996403333742,
-      "loss": 46.0,
-      "step": 3541
-    },
-    {
-      "epoch": 0.27081063516638954,
-      "grad_norm": 0.0009441355941817164,
-      "learning_rate": 0.0001999999640129626,
-      "loss": 46.0,
-      "step": 3542
-    },
-    {
-      "epoch": 0.27088709214977924,
-      "grad_norm": 0.001355852116830647,
-      "learning_rate": 0.000199999963992582,
-      "loss": 46.0,
-      "step": 3543
-    },
-    {
-      "epoch": 0.27096354913316895,
-      "grad_norm": 0.0017991584027186036,
-      "learning_rate": 0.00019999996397219563,
-      "loss": 46.0,
-      "step": 3544
-    },
-    {
-      "epoch": 0.27104000611655865,
-      "grad_norm": 0.001899454277008772,
-      "learning_rate": 0.0001999999639518035,
-      "loss": 46.0,
-      "step": 3545
-    },
-    {
-      "epoch": 0.2711164630999484,
-      "grad_norm": 0.0022068587131798267,
-      "learning_rate": 0.00019999996393140558,
-      "loss": 46.0,
-      "step": 3546
-    },
-    {
-      "epoch": 0.2711929200833381,
-      "grad_norm": 0.0008864528499543667,
-      "learning_rate": 0.0001999999639110019,
-      "loss": 46.0,
-      "step": 3547
-    },
-    {
-      "epoch": 0.2712693770667278,
-      "grad_norm": 0.000995575450360775,
-      "learning_rate": 0.00019999996389059247,
-      "loss": 46.0,
-      "step": 3548
-    },
-    {
-      "epoch": 0.27134583405011753,
-      "grad_norm": 0.0009679142385721207,
-      "learning_rate": 0.00019999996387017727,
-      "loss": 46.0,
-      "step": 3549
-    },
-    {
-      "epoch": 0.2714222910335073,
-      "grad_norm": 0.0004857187741436064,
-      "learning_rate": 0.00019999996384975627,
-      "loss": 46.0,
-      "step": 3550
-    },
-    {
-      "epoch": 0.271498748016897,
-      "grad_norm": 0.0008006109273992479,
-      "learning_rate": 0.00019999996382932955,
-      "loss": 46.0,
-      "step": 3551
-    },
-    {
-      "epoch": 0.2715752050002867,
-      "grad_norm": 0.0013027467066422105,
-      "learning_rate": 0.000199999963808897,
-      "loss": 46.0,
-      "step": 3552
-    },
-    {
-      "epoch": 0.2716516619836764,
-      "grad_norm": 0.0014260867610573769,
-      "learning_rate": 0.00019999996378845873,
-      "loss": 46.0,
-      "step": 3553
-    },
-    {
-      "epoch": 0.27172811896706617,
-      "grad_norm": 0.0011759180342778563,
-      "learning_rate": 0.0001999999637680147,
-      "loss": 46.0,
-      "step": 3554
-    },
-    {
-      "epoch": 0.2718045759504559,
-      "grad_norm": 0.0006172429420985281,
-      "learning_rate": 0.00019999996374756485,
-      "loss": 46.0,
-      "step": 3555
-    },
-    {
-      "epoch": 0.2718810329338456,
-      "grad_norm": 0.001046796445734799,
-      "learning_rate": 0.00019999996372710926,
-      "loss": 46.0,
-      "step": 3556
-    },
-    {
-      "epoch": 0.27195748991723534,
-      "grad_norm": 0.001194289536215365,
-      "learning_rate": 0.0001999999637066479,
-      "loss": 46.0,
-      "step": 3557
-    },
-    {
-      "epoch": 0.27203394690062505,
-      "grad_norm": 0.0023409014102071524,
-      "learning_rate": 0.00019999996368618073,
-      "loss": 46.0,
-      "step": 3558
-    },
-    {
-      "epoch": 0.27211040388401475,
-      "grad_norm": 0.004527931567281485,
-      "learning_rate": 0.00019999996366570785,
-      "loss": 46.0,
-      "step": 3559
-    },
-    {
-      "epoch": 0.27218686086740446,
-      "grad_norm": 0.0012273762840777636,
-      "learning_rate": 0.00019999996364522917,
-      "loss": 46.0,
-      "step": 3560
-    },
-    {
-      "epoch": 0.2722633178507942,
-      "grad_norm": 0.0007445935043506324,
-      "learning_rate": 0.00019999996362474472,
-      "loss": 46.0,
-      "step": 3561
-    },
-    {
-      "epoch": 0.2723397748341839,
-      "grad_norm": 0.0007178076775744557,
-      "learning_rate": 0.00019999996360425452,
-      "loss": 46.0,
-      "step": 3562
-    },
-    {
-      "epoch": 0.27241623181757363,
-      "grad_norm": 0.0018861166900023818,
-      "learning_rate": 0.00019999996358375851,
-      "loss": 46.0,
-      "step": 3563
-    },
-    {
-      "epoch": 0.27249268880096333,
-      "grad_norm": 0.0005265391082502902,
-      "learning_rate": 0.0001999999635632568,
-      "loss": 46.0,
-      "step": 3564
-    },
-    {
-      "epoch": 0.2725691457843531,
-      "grad_norm": 0.0013579848455265164,
-      "learning_rate": 0.00019999996354274924,
-      "loss": 46.0,
-      "step": 3565
-    },
-    {
-      "epoch": 0.2726456027677428,
-      "grad_norm": 0.0013015683507546782,
-      "learning_rate": 0.00019999996352223595,
-      "loss": 46.0,
-      "step": 3566
-    },
-    {
-      "epoch": 0.2727220597511325,
-      "grad_norm": 0.0014005533885210752,
-      "learning_rate": 0.0001999999635017169,
-      "loss": 46.0,
-      "step": 3567
-    },
-    {
-      "epoch": 0.2727985167345222,
-      "grad_norm": 0.0016418549930676818,
-      "learning_rate": 0.00019999996348119207,
-      "loss": 46.0,
-      "step": 3568
-    },
-    {
-      "epoch": 0.272874973717912,
-      "grad_norm": 0.0022029762621968985,
-      "learning_rate": 0.00019999996346066148,
-      "loss": 46.0,
-      "step": 3569
-    },
-    {
-      "epoch": 0.2729514307013017,
-      "grad_norm": 0.0007229609764181077,
-      "learning_rate": 0.00019999996344012512,
-      "loss": 46.0,
-      "step": 3570
-    },
-    {
-      "epoch": 0.2730278876846914,
-      "grad_norm": 0.0007221169653348625,
-      "learning_rate": 0.00019999996341958296,
-      "loss": 46.0,
-      "step": 3571
-    },
-    {
-      "epoch": 0.27310434466808114,
-      "grad_norm": 0.003370679449290037,
-      "learning_rate": 0.00019999996339903505,
-      "loss": 46.0,
-      "step": 3572
-    },
-    {
-      "epoch": 0.27318080165147085,
-      "grad_norm": 0.001101373927667737,
-      "learning_rate": 0.00019999996337848137,
-      "loss": 46.0,
-      "step": 3573
-    },
-    {
-      "epoch": 0.27325725863486056,
-      "grad_norm": 0.0007499827188439667,
-      "learning_rate": 0.00019999996335792194,
-      "loss": 46.0,
-      "step": 3574
-    },
-    {
-      "epoch": 0.27333371561825026,
-      "grad_norm": 0.0012965545756742358,
-      "learning_rate": 0.0001999999633373567,
-      "loss": 46.0,
-      "step": 3575
-    },
-    {
-      "epoch": 0.27341017260164,
-      "grad_norm": 0.0003533526905812323,
-      "learning_rate": 0.0001999999633167857,
-      "loss": 46.0,
-      "step": 3576
-    },
-    {
-      "epoch": 0.27348662958502973,
-      "grad_norm": 0.001247638720087707,
-      "learning_rate": 0.00019999996329620896,
-      "loss": 46.0,
-      "step": 3577
-    },
-    {
-      "epoch": 0.27356308656841943,
-      "grad_norm": 0.001336206216365099,
-      "learning_rate": 0.0001999999632756264,
-      "loss": 46.0,
-      "step": 3578
-    },
-    {
-      "epoch": 0.27363954355180914,
-      "grad_norm": 0.0005127611220814288,
-      "learning_rate": 0.00019999996325503815,
-      "loss": 46.0,
-      "step": 3579
-    },
-    {
-      "epoch": 0.2737160005351989,
-      "grad_norm": 0.0018877009861171246,
-      "learning_rate": 0.00019999996323444408,
-      "loss": 46.0,
-      "step": 3580
-    },
-    {
-      "epoch": 0.2737924575185886,
-      "grad_norm": 0.0008929924224503338,
-      "learning_rate": 0.00019999996321384424,
-      "loss": 46.0,
-      "step": 3581
-    },
-    {
-      "epoch": 0.2738689145019783,
-      "grad_norm": 0.0012983556371182203,
-      "learning_rate": 0.00019999996319323865,
-      "loss": 46.0,
-      "step": 3582
-    },
-    {
-      "epoch": 0.27394537148536807,
-      "grad_norm": 0.0039312588050961494,
-      "learning_rate": 0.00019999996317262726,
-      "loss": 46.0,
-      "step": 3583
-    },
-    {
-      "epoch": 0.2740218284687578,
-      "grad_norm": 0.002769493730738759,
-      "learning_rate": 0.00019999996315201013,
-      "loss": 46.0,
-      "step": 3584
-    },
-    {
-      "epoch": 0.2740982854521475,
-      "grad_norm": 0.0006664571701548994,
-      "learning_rate": 0.00019999996313138722,
-      "loss": 46.0,
-      "step": 3585
-    },
-    {
-      "epoch": 0.2741747424355372,
-      "grad_norm": 0.0011385715333744884,
-      "learning_rate": 0.00019999996311075854,
-      "loss": 46.0,
-      "step": 3586
-    },
-    {
-      "epoch": 0.27425119941892695,
-      "grad_norm": 0.002503304975107312,
-      "learning_rate": 0.0001999999630901241,
-      "loss": 46.0,
-      "step": 3587
-    },
-    {
-      "epoch": 0.27432765640231666,
-      "grad_norm": 0.00046438383287750185,
-      "learning_rate": 0.00019999996306948386,
-      "loss": 46.0,
-      "step": 3588
-    },
-    {
-      "epoch": 0.27440411338570636,
-      "grad_norm": 0.0008705140207894146,
-      "learning_rate": 0.00019999996304883788,
-      "loss": 46.0,
-      "step": 3589
-    },
-    {
-      "epoch": 0.27448057036909607,
-      "grad_norm": 0.0026737467851489782,
-      "learning_rate": 0.00019999996302818613,
-      "loss": 46.0,
-      "step": 3590
-    },
-    {
-      "epoch": 0.2745570273524858,
-      "grad_norm": 0.002014830010011792,
-      "learning_rate": 0.00019999996300752859,
-      "loss": 46.0,
-      "step": 3591
-    },
-    {
-      "epoch": 0.27463348433587553,
-      "grad_norm": 0.0014765787636861205,
-      "learning_rate": 0.0001999999629868653,
-      "loss": 46.0,
-      "step": 3592
-    },
-    {
-      "epoch": 0.27470994131926524,
-      "grad_norm": 0.0020565094891935587,
-      "learning_rate": 0.00019999996296619622,
-      "loss": 46.0,
-      "step": 3593
-    },
-    {
-      "epoch": 0.27478639830265494,
-      "grad_norm": 0.0014105496229603887,
-      "learning_rate": 0.00019999996294552138,
-      "loss": 46.0,
-      "step": 3594
-    },
-    {
-      "epoch": 0.2748628552860447,
-      "grad_norm": 0.0010734430979937315,
-      "learning_rate": 0.0001999999629248408,
-      "loss": 46.0,
-      "step": 3595
-    },
-    {
-      "epoch": 0.2749393122694344,
-      "grad_norm": 0.0009878509445115924,
-      "learning_rate": 0.0001999999629041544,
-      "loss": 46.0,
-      "step": 3596
-    },
-    {
-      "epoch": 0.2750157692528241,
-      "grad_norm": 0.002685703570023179,
-      "learning_rate": 0.00019999996288346225,
-      "loss": 46.0,
-      "step": 3597
-    },
-    {
-      "epoch": 0.2750922262362139,
-      "grad_norm": 0.0020470006857067347,
-      "learning_rate": 0.00019999996286276434,
-      "loss": 46.0,
-      "step": 3598
-    },
-    {
-      "epoch": 0.2751686832196036,
-      "grad_norm": 0.0010294730309396982,
-      "learning_rate": 0.00019999996284206066,
-      "loss": 46.0,
-      "step": 3599
-    },
-    {
-      "epoch": 0.2752451402029933,
-      "grad_norm": 0.0009523346088826656,
-      "learning_rate": 0.00019999996282135123,
-      "loss": 46.0,
-      "step": 3600
-    },
-    {
-      "epoch": 0.275321597186383,
-      "grad_norm": 0.0004560293455142528,
-      "learning_rate": 0.00019999996280063598,
-      "loss": 46.0,
-      "step": 3601
-    },
-    {
-      "epoch": 0.27539805416977275,
-      "grad_norm": 0.005681591108441353,
-      "learning_rate": 0.000199999962779915,
-      "loss": 46.0,
-      "step": 3602
-    },
-    {
-      "epoch": 0.27547451115316246,
-      "grad_norm": 0.0024480780120939016,
-      "learning_rate": 0.00019999996275918823,
-      "loss": 46.0,
-      "step": 3603
-    },
-    {
-      "epoch": 0.27555096813655217,
-      "grad_norm": 0.0016501410864293575,
-      "learning_rate": 0.0001999999627384557,
-      "loss": 46.0,
-      "step": 3604
-    },
-    {
-      "epoch": 0.27562742511994187,
-      "grad_norm": 0.001211164053529501,
-      "learning_rate": 0.00019999996271771742,
-      "loss": 46.0,
-      "step": 3605
-    },
-    {
-      "epoch": 0.27570388210333163,
-      "grad_norm": 0.000644101994112134,
-      "learning_rate": 0.00019999996269697335,
-      "loss": 46.0,
-      "step": 3606
-    },
-    {
-      "epoch": 0.27578033908672134,
-      "grad_norm": 0.0010957722552120686,
-      "learning_rate": 0.0001999999626762235,
-      "loss": 46.0,
-      "step": 3607
-    },
-    {
-      "epoch": 0.27585679607011104,
-      "grad_norm": 0.0007161587127484381,
-      "learning_rate": 0.00019999996265546787,
-      "loss": 46.0,
-      "step": 3608
-    },
-    {
-      "epoch": 0.27593325305350075,
-      "grad_norm": 0.003564264392480254,
-      "learning_rate": 0.0001999999626347065,
-      "loss": 46.0,
-      "step": 3609
-    },
-    {
-      "epoch": 0.2760097100368905,
-      "grad_norm": 0.0006503176991827786,
-      "learning_rate": 0.00019999996261393937,
-      "loss": 46.0,
-      "step": 3610
-    },
-    {
-      "epoch": 0.2760861670202802,
-      "grad_norm": 0.001754163415171206,
-      "learning_rate": 0.00019999996259316644,
-      "loss": 46.0,
-      "step": 3611
-    },
-    {
-      "epoch": 0.2761626240036699,
-      "grad_norm": 0.001982693327590823,
-      "learning_rate": 0.00019999996257238776,
-      "loss": 46.0,
-      "step": 3612
-    },
-    {
-      "epoch": 0.2762390809870597,
-      "grad_norm": 0.0016400435706600547,
-      "learning_rate": 0.0001999999625516033,
-      "loss": 46.0,
-      "step": 3613
-    },
-    {
-      "epoch": 0.2763155379704494,
-      "grad_norm": 0.00039631876279599965,
-      "learning_rate": 0.00019999996253081308,
-      "loss": 46.0,
-      "step": 3614
-    },
-    {
-      "epoch": 0.2763919949538391,
-      "grad_norm": 0.001241436111740768,
-      "learning_rate": 0.00019999996251001708,
-      "loss": 46.0,
-      "step": 3615
-    },
-    {
-      "epoch": 0.2764684519372288,
-      "grad_norm": 0.0005651030223816633,
-      "learning_rate": 0.0001999999624892153,
-      "loss": 46.0,
-      "step": 3616
-    },
-    {
-      "epoch": 0.27654490892061856,
-      "grad_norm": 0.001037145615555346,
-      "learning_rate": 0.00019999996246840778,
-      "loss": 46.0,
-      "step": 3617
-    },
-    {
-      "epoch": 0.27662136590400827,
-      "grad_norm": 0.0009783509885892272,
-      "learning_rate": 0.00019999996244759446,
-      "loss": 46.0,
-      "step": 3618
-    },
-    {
-      "epoch": 0.27669782288739797,
-      "grad_norm": 0.0007942967931739986,
-      "learning_rate": 0.00019999996242677542,
-      "loss": 46.0,
-      "step": 3619
-    },
-    {
-      "epoch": 0.2767742798707877,
-      "grad_norm": 0.0028216196224093437,
-      "learning_rate": 0.00019999996240595058,
-      "loss": 46.0,
-      "step": 3620
-    },
-    {
-      "epoch": 0.27685073685417744,
-      "grad_norm": 0.00045819449587725103,
-      "learning_rate": 0.00019999996238511997,
-      "loss": 46.0,
-      "step": 3621
-    },
-    {
-      "epoch": 0.27692719383756714,
-      "grad_norm": 0.0014861531089991331,
-      "learning_rate": 0.00019999996236428358,
-      "loss": 46.0,
-      "step": 3622
-    },
-    {
-      "epoch": 0.27700365082095685,
-      "grad_norm": 0.0012091546086594462,
-      "learning_rate": 0.00019999996234344142,
-      "loss": 46.0,
-      "step": 3623
-    },
-    {
-      "epoch": 0.27708010780434655,
-      "grad_norm": 0.0023589839693158865,
-      "learning_rate": 0.0001999999623225935,
-      "loss": 46.0,
-      "step": 3624
-    },
-    {
-      "epoch": 0.2771565647877363,
-      "grad_norm": 0.0011700313771143556,
-      "learning_rate": 0.0001999999623017398,
-      "loss": 46.0,
-      "step": 3625
-    },
-    {
-      "epoch": 0.277233021771126,
-      "grad_norm": 0.0021366956643760204,
-      "learning_rate": 0.00019999996228088035,
-      "loss": 46.0,
-      "step": 3626
-    },
-    {
-      "epoch": 0.2773094787545157,
-      "grad_norm": 0.0006147163221612573,
-      "learning_rate": 0.00019999996226001513,
-      "loss": 46.0,
-      "step": 3627
-    },
-    {
-      "epoch": 0.2773859357379055,
-      "grad_norm": 0.0011000499362125993,
-      "learning_rate": 0.00019999996223914413,
-      "loss": 46.0,
-      "step": 3628
-    },
-    {
-      "epoch": 0.2774623927212952,
-      "grad_norm": 0.0024103980977088213,
-      "learning_rate": 0.00019999996221826735,
-      "loss": 46.0,
-      "step": 3629
-    },
-    {
-      "epoch": 0.2775388497046849,
-      "grad_norm": 0.0010705257300287485,
-      "learning_rate": 0.00019999996219738483,
-      "loss": 46.0,
-      "step": 3630
-    },
-    {
-      "epoch": 0.2776153066880746,
-      "grad_norm": 0.005650050472468138,
-      "learning_rate": 0.0001999999621764965,
-      "loss": 46.0,
-      "step": 3631
-    },
-    {
-      "epoch": 0.27769176367146436,
-      "grad_norm": 0.00201904377900064,
-      "learning_rate": 0.00019999996215560244,
-      "loss": 46.0,
-      "step": 3632
-    },
-    {
-      "epoch": 0.27776822065485407,
-      "grad_norm": 0.0003853754315059632,
-      "learning_rate": 0.0001999999621347026,
-      "loss": 46.0,
-      "step": 3633
-    },
-    {
-      "epoch": 0.2778446776382438,
-      "grad_norm": 0.004631387535482645,
-      "learning_rate": 0.000199999962113797,
-      "loss": 46.0,
-      "step": 3634
-    },
-    {
-      "epoch": 0.2779211346216335,
-      "grad_norm": 0.000928018765989691,
-      "learning_rate": 0.00019999996209288558,
-      "loss": 46.0,
-      "step": 3635
-    },
-    {
-      "epoch": 0.27799759160502324,
-      "grad_norm": 0.0008019008091650903,
-      "learning_rate": 0.00019999996207196844,
-      "loss": 46.0,
-      "step": 3636
-    },
-    {
-      "epoch": 0.27807404858841295,
-      "grad_norm": 0.0010388718219473958,
-      "learning_rate": 0.0001999999620510455,
-      "loss": 46.0,
-      "step": 3637
-    },
-    {
-      "epoch": 0.27815050557180265,
-      "grad_norm": 0.0015405297745019197,
-      "learning_rate": 0.0001999999620301168,
-      "loss": 46.0,
-      "step": 3638
-    },
-    {
-      "epoch": 0.27822696255519236,
-      "grad_norm": 0.004980092402547598,
-      "learning_rate": 0.00019999996200918235,
-      "loss": 46.0,
-      "step": 3639
-    },
-    {
-      "epoch": 0.2783034195385821,
-      "grad_norm": 0.0022869580425322056,
-      "learning_rate": 0.00019999996198824212,
-      "loss": 46.0,
-      "step": 3640
-    },
-    {
-      "epoch": 0.2783798765219718,
-      "grad_norm": 0.0013629082823172212,
-      "learning_rate": 0.00019999996196729612,
-      "loss": 46.0,
-      "step": 3641
-    },
-    {
-      "epoch": 0.27845633350536153,
-      "grad_norm": 0.0009925602935254574,
-      "learning_rate": 0.00019999996194634435,
-      "loss": 46.0,
-      "step": 3642
-    },
-    {
-      "epoch": 0.2785327904887513,
-      "grad_norm": 0.001760657411068678,
-      "learning_rate": 0.0001999999619253868,
-      "loss": 46.0,
-      "step": 3643
-    },
-    {
-      "epoch": 0.278609247472141,
-      "grad_norm": 0.0014615763211622834,
-      "learning_rate": 0.0001999999619044235,
-      "loss": 46.0,
-      "step": 3644
-    },
-    {
-      "epoch": 0.2786857044555307,
-      "grad_norm": 0.00123655388597399,
-      "learning_rate": 0.00019999996188345445,
-      "loss": 46.0,
-      "step": 3645
-    },
-    {
-      "epoch": 0.2787621614389204,
-      "grad_norm": 0.0007111479062587023,
-      "learning_rate": 0.00019999996186247958,
-      "loss": 46.0,
-      "step": 3646
-    },
-    {
-      "epoch": 0.27883861842231017,
-      "grad_norm": 0.0012057044077664614,
-      "learning_rate": 0.000199999961841499,
-      "loss": 46.0,
-      "step": 3647
-    },
-    {
-      "epoch": 0.2789150754056999,
-      "grad_norm": 0.0017454746412113309,
-      "learning_rate": 0.00019999996182051258,
-      "loss": 46.0,
-      "step": 3648
-    },
-    {
-      "epoch": 0.2789915323890896,
-      "grad_norm": 0.0014446472050622106,
-      "learning_rate": 0.00019999996179952042,
-      "loss": 46.0,
-      "step": 3649
-    },
-    {
-      "epoch": 0.2790679893724793,
-      "grad_norm": 0.0013323846505954862,
-      "learning_rate": 0.0001999999617785225,
-      "loss": 46.0,
-      "step": 3650
-    },
-    {
-      "epoch": 0.27914444635586905,
-      "grad_norm": 0.0004912894801236689,
-      "learning_rate": 0.0001999999617575188,
-      "loss": 46.0,
-      "step": 3651
-    },
-    {
-      "epoch": 0.27922090333925875,
-      "grad_norm": 0.0013517895713448524,
-      "learning_rate": 0.00019999996173650935,
-      "loss": 46.0,
-      "step": 3652
-    },
-    {
-      "epoch": 0.27929736032264846,
-      "grad_norm": 0.0011540567502379417,
-      "learning_rate": 0.0001999999617154941,
-      "loss": 46.0,
-      "step": 3653
-    },
-    {
-      "epoch": 0.27937381730603816,
-      "grad_norm": 0.001616124645806849,
-      "learning_rate": 0.0001999999616944731,
-      "loss": 46.0,
-      "step": 3654
-    },
-    {
-      "epoch": 0.2794502742894279,
-      "grad_norm": 0.001074964297004044,
-      "learning_rate": 0.00019999996167344635,
-      "loss": 46.0,
-      "step": 3655
-    },
-    {
-      "epoch": 0.27952673127281763,
-      "grad_norm": 0.0008659610757604241,
-      "learning_rate": 0.00019999996165241378,
-      "loss": 46.0,
-      "step": 3656
-    },
-    {
-      "epoch": 0.27960318825620734,
-      "grad_norm": 0.008615114726126194,
-      "learning_rate": 0.00019999996163137546,
-      "loss": 46.0,
-      "step": 3657
-    },
-    {
-      "epoch": 0.2796796452395971,
-      "grad_norm": 0.0010578272631391883,
-      "learning_rate": 0.00019999996161033142,
-      "loss": 46.0,
-      "step": 3658
-    },
-    {
-      "epoch": 0.2797561022229868,
-      "grad_norm": 0.0016543472884222865,
-      "learning_rate": 0.00019999996158928153,
-      "loss": 46.0,
-      "step": 3659
-    },
-    {
-      "epoch": 0.2798325592063765,
-      "grad_norm": 0.0014690675307065248,
-      "learning_rate": 0.00019999996156822594,
-      "loss": 46.0,
-      "step": 3660
-    },
-    {
-      "epoch": 0.2799090161897662,
-      "grad_norm": 0.002100287936627865,
-      "learning_rate": 0.00019999996154716453,
-      "loss": 46.0,
-      "step": 3661
-    },
-    {
-      "epoch": 0.279985473173156,
-      "grad_norm": 0.0028917405288666487,
-      "learning_rate": 0.00019999996152609737,
-      "loss": 46.0,
-      "step": 3662
-    },
-    {
-      "epoch": 0.2800619301565457,
-      "grad_norm": 0.001529567874968052,
-      "learning_rate": 0.00019999996150502444,
-      "loss": 46.0,
-      "step": 3663
-    },
-    {
-      "epoch": 0.2801383871399354,
-      "grad_norm": 0.0008070769836194813,
-      "learning_rate": 0.00019999996148394576,
-      "loss": 46.0,
-      "step": 3664
-    },
-    {
-      "epoch": 0.2802148441233251,
-      "grad_norm": 0.0013607712462544441,
-      "learning_rate": 0.00019999996146286128,
-      "loss": 46.0,
-      "step": 3665
-    },
-    {
-      "epoch": 0.28029130110671485,
-      "grad_norm": 0.000600371859036386,
-      "learning_rate": 0.00019999996144177105,
-      "loss": 46.0,
-      "step": 3666
-    },
-    {
-      "epoch": 0.28036775809010456,
-      "grad_norm": 0.0009783091954886913,
-      "learning_rate": 0.00019999996142067503,
-      "loss": 46.0,
-      "step": 3667
-    },
-    {
-      "epoch": 0.28044421507349426,
-      "grad_norm": 0.0015447192126885056,
-      "learning_rate": 0.00019999996139957325,
-      "loss": 46.0,
-      "step": 3668
-    },
-    {
-      "epoch": 0.28052067205688397,
-      "grad_norm": 0.0018628071993589401,
-      "learning_rate": 0.00019999996137846573,
-      "loss": 46.0,
-      "step": 3669
-    },
-    {
-      "epoch": 0.28059712904027373,
-      "grad_norm": 0.0016895532608032227,
-      "learning_rate": 0.0001999999613573524,
-      "loss": 46.0,
-      "step": 3670
-    },
-    {
-      "epoch": 0.28067358602366344,
-      "grad_norm": 0.0005364527460187674,
-      "learning_rate": 0.00019999996133623332,
-      "loss": 46.0,
-      "step": 3671
-    },
-    {
-      "epoch": 0.28075004300705314,
-      "grad_norm": 0.001170719158835709,
-      "learning_rate": 0.00019999996131510846,
-      "loss": 46.0,
-      "step": 3672
-    },
-    {
-      "epoch": 0.2808264999904429,
-      "grad_norm": 0.0016270236810669303,
-      "learning_rate": 0.00019999996129397784,
-      "loss": 46.0,
-      "step": 3673
-    },
-    {
-      "epoch": 0.2809029569738326,
-      "grad_norm": 0.005730601958930492,
-      "learning_rate": 0.00019999996127284146,
-      "loss": 46.0,
-      "step": 3674
-    },
-    {
-      "epoch": 0.2809794139572223,
-      "grad_norm": 0.0019150610314682126,
-      "learning_rate": 0.0001999999612516993,
-      "loss": 46.0,
-      "step": 3675
-    },
-    {
-      "epoch": 0.281055870940612,
-      "grad_norm": 0.0022221652325242758,
-      "learning_rate": 0.00019999996123055137,
-      "loss": 46.0,
-      "step": 3676
-    },
-    {
-      "epoch": 0.2811323279240018,
-      "grad_norm": 0.0014204467879608274,
-      "learning_rate": 0.00019999996120939766,
-      "loss": 46.0,
-      "step": 3677
-    },
-    {
-      "epoch": 0.2812087849073915,
-      "grad_norm": 0.0008353561861440539,
-      "learning_rate": 0.00019999996118823818,
-      "loss": 46.0,
-      "step": 3678
-    },
-    {
-      "epoch": 0.2812852418907812,
-      "grad_norm": 0.0007339682779274881,
-      "learning_rate": 0.00019999996116707296,
-      "loss": 46.0,
-      "step": 3679
-    },
-    {
-      "epoch": 0.2813616988741709,
-      "grad_norm": 0.003266470041126013,
-      "learning_rate": 0.00019999996114590193,
-      "loss": 46.0,
-      "step": 3680
-    },
-    {
-      "epoch": 0.28143815585756066,
-      "grad_norm": 0.0012170193949714303,
-      "learning_rate": 0.00019999996112472516,
-      "loss": 46.0,
-      "step": 3681
-    },
-    {
-      "epoch": 0.28151461284095036,
-      "grad_norm": 0.0014374416787177324,
-      "learning_rate": 0.0001999999611035426,
-      "loss": 46.0,
-      "step": 3682
-    },
-    {
-      "epoch": 0.28159106982434007,
-      "grad_norm": 0.0007415685686282814,
-      "learning_rate": 0.0001999999610823543,
-      "loss": 46.0,
-      "step": 3683
-    },
-    {
-      "epoch": 0.2816675268077298,
-      "grad_norm": 0.001207210822030902,
-      "learning_rate": 0.0001999999610611602,
-      "loss": 46.0,
-      "step": 3684
-    },
-    {
-      "epoch": 0.28174398379111953,
-      "grad_norm": 0.0009158743196167052,
-      "learning_rate": 0.00019999996103996036,
-      "loss": 46.0,
-      "step": 3685
-    },
-    {
-      "epoch": 0.28182044077450924,
-      "grad_norm": 0.003506585257127881,
-      "learning_rate": 0.00019999996101875472,
-      "loss": 46.0,
-      "step": 3686
-    },
-    {
-      "epoch": 0.28189689775789895,
-      "grad_norm": 0.0006057792343199253,
-      "learning_rate": 0.00019999996099754334,
-      "loss": 46.0,
-      "step": 3687
-    },
-    {
-      "epoch": 0.2819733547412887,
-      "grad_norm": 0.0011886082356795669,
-      "learning_rate": 0.00019999996097632615,
-      "loss": 46.0,
-      "step": 3688
-    },
-    {
-      "epoch": 0.2820498117246784,
-      "grad_norm": 0.001971777528524399,
-      "learning_rate": 0.00019999996095510322,
-      "loss": 46.0,
-      "step": 3689
-    },
-    {
-      "epoch": 0.2821262687080681,
-      "grad_norm": 0.0007664765580557287,
-      "learning_rate": 0.0001999999609338745,
-      "loss": 46.0,
-      "step": 3690
-    },
-    {
-      "epoch": 0.2822027256914578,
-      "grad_norm": 0.005547497887164354,
-      "learning_rate": 0.00019999996091264003,
-      "loss": 46.0,
-      "step": 3691
-    },
-    {
-      "epoch": 0.2822791826748476,
-      "grad_norm": 0.000978269032202661,
-      "learning_rate": 0.0001999999608913998,
-      "loss": 46.0,
-      "step": 3692
-    },
-    {
-      "epoch": 0.2823556396582373,
-      "grad_norm": 0.0008185630431398749,
-      "learning_rate": 0.0001999999608701538,
-      "loss": 46.0,
-      "step": 3693
-    },
-    {
-      "epoch": 0.282432096641627,
-      "grad_norm": 0.00230449833907187,
-      "learning_rate": 0.000199999960848902,
-      "loss": 46.0,
-      "step": 3694
-    },
-    {
-      "epoch": 0.2825085536250167,
-      "grad_norm": 0.0020101654808968306,
-      "learning_rate": 0.00019999996082764446,
-      "loss": 46.0,
-      "step": 3695
-    },
-    {
-      "epoch": 0.28258501060840646,
-      "grad_norm": 0.0027369833551347256,
-      "learning_rate": 0.00019999996080638112,
-      "loss": 46.0,
-      "step": 3696
-    },
-    {
-      "epoch": 0.28266146759179617,
-      "grad_norm": 0.0007445720839314163,
-      "learning_rate": 0.00019999996078511205,
-      "loss": 46.0,
-      "step": 3697
-    },
-    {
-      "epoch": 0.2827379245751859,
-      "grad_norm": 0.0030912559013813734,
-      "learning_rate": 0.00019999996076383722,
-      "loss": 46.0,
-      "step": 3698
-    },
-    {
-      "epoch": 0.28281438155857563,
-      "grad_norm": 0.0008372279116883874,
-      "learning_rate": 0.00019999996074255655,
-      "loss": 46.0,
-      "step": 3699
-    },
-    {
-      "epoch": 0.28289083854196534,
-      "grad_norm": 0.002350414404645562,
-      "learning_rate": 0.00019999996072127016,
-      "loss": 46.0,
-      "step": 3700
-    },
-    {
-      "epoch": 0.28296729552535504,
-      "grad_norm": 0.0030697931069880724,
-      "learning_rate": 0.000199999960699978,
-      "loss": 46.0,
-      "step": 3701
-    },
-    {
-      "epoch": 0.28304375250874475,
-      "grad_norm": 0.0010113042080774903,
-      "learning_rate": 0.00019999996067868007,
-      "loss": 46.0,
-      "step": 3702
-    },
-    {
-      "epoch": 0.2831202094921345,
-      "grad_norm": 0.0004754991678055376,
-      "learning_rate": 0.00019999996065737637,
-      "loss": 46.0,
-      "step": 3703
-    },
-    {
-      "epoch": 0.2831966664755242,
-      "grad_norm": 0.003134047845378518,
-      "learning_rate": 0.00019999996063606686,
-      "loss": 46.0,
-      "step": 3704
-    },
-    {
-      "epoch": 0.2832731234589139,
-      "grad_norm": 0.0007566006970591843,
-      "learning_rate": 0.00019999996061475164,
-      "loss": 46.0,
-      "step": 3705
-    },
-    {
-      "epoch": 0.28334958044230363,
-      "grad_norm": 0.0005881906836293638,
-      "learning_rate": 0.0001999999605934306,
-      "loss": 46.0,
-      "step": 3706
-    },
-    {
-      "epoch": 0.2834260374256934,
-      "grad_norm": 0.0016896775923669338,
-      "learning_rate": 0.0001999999605721038,
-      "loss": 46.0,
-      "step": 3707
-    },
-    {
-      "epoch": 0.2835024944090831,
-      "grad_norm": 0.0010982779785990715,
-      "learning_rate": 0.00019999996055077127,
-      "loss": 46.0,
-      "step": 3708
-    },
-    {
-      "epoch": 0.2835789513924728,
-      "grad_norm": 0.0012666484108194709,
-      "learning_rate": 0.00019999996052943295,
-      "loss": 46.0,
-      "step": 3709
-    },
-    {
-      "epoch": 0.2836554083758625,
-      "grad_norm": 0.0011422195238992572,
-      "learning_rate": 0.00019999996050808886,
-      "loss": 46.0,
-      "step": 3710
-    },
-    {
-      "epoch": 0.28373186535925227,
-      "grad_norm": 0.0021781160030514,
-      "learning_rate": 0.000199999960486739,
-      "loss": 46.0,
-      "step": 3711
-    },
-    {
-      "epoch": 0.28380832234264197,
-      "grad_norm": 0.0008051304612308741,
-      "learning_rate": 0.00019999996046538335,
-      "loss": 46.0,
-      "step": 3712
-    },
-    {
-      "epoch": 0.2838847793260317,
-      "grad_norm": 0.0008812343585304916,
-      "learning_rate": 0.00019999996044402197,
-      "loss": 46.0,
-      "step": 3713
-    },
-    {
-      "epoch": 0.28396123630942144,
-      "grad_norm": 0.0035376292653381824,
-      "learning_rate": 0.00019999996042265478,
-      "loss": 46.0,
-      "step": 3714
-    },
-    {
-      "epoch": 0.28403769329281114,
-      "grad_norm": 0.0013234912184998393,
-      "learning_rate": 0.00019999996040128185,
-      "loss": 46.0,
-      "step": 3715
-    },
-    {
-      "epoch": 0.28411415027620085,
-      "grad_norm": 0.0006036676932126284,
-      "learning_rate": 0.00019999996037990317,
-      "loss": 46.0,
-      "step": 3716
-    },
-    {
-      "epoch": 0.28419060725959056,
-      "grad_norm": 0.0009874098468571901,
-      "learning_rate": 0.00019999996035851867,
-      "loss": 46.0,
-      "step": 3717
-    },
-    {
-      "epoch": 0.2842670642429803,
-      "grad_norm": 0.0012311360333114862,
-      "learning_rate": 0.00019999996033712842,
-      "loss": 46.0,
-      "step": 3718
-    },
-    {
-      "epoch": 0.28434352122637,
-      "grad_norm": 0.0006533898995257914,
-      "learning_rate": 0.0001999999603157324,
-      "loss": 46.0,
-      "step": 3719
-    },
-    {
-      "epoch": 0.2844199782097597,
-      "grad_norm": 0.0005601924494840205,
-      "learning_rate": 0.00019999996029433062,
-      "loss": 46.0,
-      "step": 3720
-    },
-    {
-      "epoch": 0.28449643519314943,
-      "grad_norm": 0.0023518491070717573,
-      "learning_rate": 0.00019999996027292305,
-      "loss": 46.0,
-      "step": 3721
-    },
-    {
-      "epoch": 0.2845728921765392,
-      "grad_norm": 0.001143717672675848,
-      "learning_rate": 0.00019999996025150973,
-      "loss": 46.0,
-      "step": 3722
-    },
-    {
-      "epoch": 0.2846493491599289,
-      "grad_norm": 0.001057030283845961,
-      "learning_rate": 0.00019999996023009064,
-      "loss": 46.0,
-      "step": 3723
-    },
-    {
-      "epoch": 0.2847258061433186,
-      "grad_norm": 0.002260795794427395,
-      "learning_rate": 0.00019999996020866577,
-      "loss": 46.0,
-      "step": 3724
-    },
-    {
-      "epoch": 0.2848022631267083,
-      "grad_norm": 0.002682718215510249,
-      "learning_rate": 0.00019999996018723514,
-      "loss": 46.0,
-      "step": 3725
-    },
-    {
-      "epoch": 0.28487872011009807,
-      "grad_norm": 0.000765470031183213,
-      "learning_rate": 0.00019999996016579875,
-      "loss": 46.0,
-      "step": 3726
-    },
-    {
-      "epoch": 0.2849551770934878,
-      "grad_norm": 0.001148425741121173,
-      "learning_rate": 0.00019999996014435657,
-      "loss": 46.0,
-      "step": 3727
-    },
-    {
-      "epoch": 0.2850316340768775,
-      "grad_norm": 0.0007466459064744413,
-      "learning_rate": 0.0001999999601229086,
-      "loss": 46.0,
-      "step": 3728
-    },
-    {
-      "epoch": 0.28510809106026724,
-      "grad_norm": 0.0030335683841258287,
-      "learning_rate": 0.0001999999601014549,
-      "loss": 46.0,
-      "step": 3729
-    },
-    {
-      "epoch": 0.28518454804365695,
-      "grad_norm": 0.0022319993004202843,
-      "learning_rate": 0.00019999996007999542,
-      "loss": 46.0,
-      "step": 3730
-    },
-    {
-      "epoch": 0.28526100502704665,
-      "grad_norm": 0.0024950969964265823,
-      "learning_rate": 0.00019999996005853017,
-      "loss": 46.0,
-      "step": 3731
-    },
-    {
-      "epoch": 0.28533746201043636,
-      "grad_norm": 0.0009043251629918814,
-      "learning_rate": 0.00019999996003705915,
-      "loss": 46.0,
-      "step": 3732
-    },
-    {
-      "epoch": 0.2854139189938261,
-      "grad_norm": 0.0013740276917815208,
-      "learning_rate": 0.00019999996001558238,
-      "loss": 46.0,
-      "step": 3733
-    },
-    {
-      "epoch": 0.2854903759772158,
-      "grad_norm": 0.0011639823205769062,
-      "learning_rate": 0.0001999999599940998,
-      "loss": 46.0,
-      "step": 3734
-    },
-    {
-      "epoch": 0.28556683296060553,
-      "grad_norm": 0.012872766703367233,
-      "learning_rate": 0.00019999995997261146,
-      "loss": 46.0,
-      "step": 3735
-    },
-    {
-      "epoch": 0.28564328994399524,
-      "grad_norm": 0.0012254433240741491,
-      "learning_rate": 0.00019999995995111734,
-      "loss": 46.0,
-      "step": 3736
-    },
-    {
-      "epoch": 0.285719746927385,
-      "grad_norm": 0.0017625355394557118,
-      "learning_rate": 0.00019999995992961748,
-      "loss": 46.0,
-      "step": 3737
-    },
-    {
-      "epoch": 0.2857962039107747,
-      "grad_norm": 0.0016765417531132698,
-      "learning_rate": 0.00019999995990811187,
-      "loss": 46.0,
-      "step": 3738
-    },
-    {
-      "epoch": 0.2858726608941644,
-      "grad_norm": 0.0004553665639832616,
-      "learning_rate": 0.00019999995988660046,
-      "loss": 46.0,
-      "step": 3739
-    },
-    {
-      "epoch": 0.2859491178775541,
-      "grad_norm": 0.0008050231263041496,
-      "learning_rate": 0.0001999999598650833,
-      "loss": 46.0,
-      "step": 3740
-    },
-    {
-      "epoch": 0.2860255748609439,
-      "grad_norm": 0.0011766463285312057,
-      "learning_rate": 0.0001999999598435603,
-      "loss": 46.0,
-      "step": 3741
-    },
-    {
-      "epoch": 0.2861020318443336,
-      "grad_norm": 0.000840594177134335,
-      "learning_rate": 0.00019999995982203164,
-      "loss": 46.0,
-      "step": 3742
-    },
-    {
-      "epoch": 0.2861784888277233,
-      "grad_norm": 0.0015770596219226718,
-      "learning_rate": 0.00019999995980049713,
-      "loss": 46.0,
-      "step": 3743
-    },
-    {
-      "epoch": 0.28625494581111305,
-      "grad_norm": 0.001588560058735311,
-      "learning_rate": 0.00019999995977895688,
-      "loss": 46.0,
-      "step": 3744
-    },
-    {
-      "epoch": 0.28633140279450275,
-      "grad_norm": 0.0010057406034320593,
-      "learning_rate": 0.00019999995975741085,
-      "loss": 46.0,
-      "step": 3745
-    },
-    {
-      "epoch": 0.28640785977789246,
-      "grad_norm": 0.0015824324218556285,
-      "learning_rate": 0.00019999995973585906,
-      "loss": 46.0,
-      "step": 3746
-    },
-    {
-      "epoch": 0.28648431676128217,
-      "grad_norm": 0.0011771401623263955,
-      "learning_rate": 0.00019999995971430149,
-      "loss": 46.0,
-      "step": 3747
-    },
-    {
-      "epoch": 0.2865607737446719,
-      "grad_norm": 0.0009981276234611869,
-      "learning_rate": 0.00019999995969273817,
-      "loss": 46.0,
-      "step": 3748
-    },
-    {
-      "epoch": 0.28663723072806163,
-      "grad_norm": 0.0010086316615343094,
-      "learning_rate": 0.00019999995967116908,
-      "loss": 46.0,
-      "step": 3749
-    },
-    {
-      "epoch": 0.28671368771145134,
-      "grad_norm": 0.0007743178866803646,
-      "learning_rate": 0.00019999995964959419,
-      "loss": 46.0,
-      "step": 3750
-    },
-    {
-      "epoch": 0.28679014469484104,
-      "grad_norm": 0.0016441509360447526,
-      "learning_rate": 0.00019999995962801355,
-      "loss": 46.0,
-      "step": 3751
-    },
-    {
-      "epoch": 0.2868666016782308,
-      "grad_norm": 0.0009750126046128571,
-      "learning_rate": 0.00019999995960642714,
-      "loss": 46.0,
-      "step": 3752
-    },
-    {
-      "epoch": 0.2869430586616205,
-      "grad_norm": 0.001294981804676354,
-      "learning_rate": 0.00019999995958483498,
-      "loss": 46.0,
-      "step": 3753
-    },
-    {
-      "epoch": 0.2870195156450102,
-      "grad_norm": 0.001622818293981254,
-      "learning_rate": 0.000199999959563237,
-      "loss": 46.0,
-      "step": 3754
-    },
-    {
-      "epoch": 0.2870959726283999,
-      "grad_norm": 0.0009695104090496898,
-      "learning_rate": 0.00019999995954163332,
-      "loss": 46.0,
-      "step": 3755
-    },
-    {
-      "epoch": 0.2871724296117897,
-      "grad_norm": 0.0012397788232192397,
-      "learning_rate": 0.0001999999595200238,
-      "loss": 46.0,
-      "step": 3756
-    },
-    {
-      "epoch": 0.2872488865951794,
-      "grad_norm": 0.0029442079830914736,
-      "learning_rate": 0.00019999995949840856,
-      "loss": 46.0,
-      "step": 3757
-    },
-    {
-      "epoch": 0.2873253435785691,
-      "grad_norm": 0.0020591500215232372,
-      "learning_rate": 0.00019999995947678754,
-      "loss": 46.0,
-      "step": 3758
-    },
-    {
-      "epoch": 0.28740180056195885,
-      "grad_norm": 0.0008922885172069073,
-      "learning_rate": 0.00019999995945516074,
-      "loss": 46.0,
-      "step": 3759
-    },
-    {
-      "epoch": 0.28747825754534856,
-      "grad_norm": 0.0014189706416800618,
-      "learning_rate": 0.00019999995943352817,
-      "loss": 46.0,
-      "step": 3760
-    },
-    {
-      "epoch": 0.28755471452873826,
-      "grad_norm": 0.0033019159454852343,
-      "learning_rate": 0.00019999995941188985,
-      "loss": 46.0,
-      "step": 3761
-    },
-    {
-      "epoch": 0.28763117151212797,
-      "grad_norm": 0.0008956018718890846,
-      "learning_rate": 0.00019999995939024573,
-      "loss": 46.0,
-      "step": 3762
-    },
-    {
-      "epoch": 0.28770762849551773,
-      "grad_norm": 0.0012777860974892974,
-      "learning_rate": 0.00019999995936859584,
-      "loss": 46.0,
-      "step": 3763
-    },
-    {
-      "epoch": 0.28778408547890744,
-      "grad_norm": 0.0033158992882817984,
-      "learning_rate": 0.0001999999593469402,
-      "loss": 46.0,
-      "step": 3764
-    },
-    {
-      "epoch": 0.28786054246229714,
-      "grad_norm": 0.000867457187268883,
-      "learning_rate": 0.0001999999593252788,
-      "loss": 46.0,
-      "step": 3765
-    },
-    {
-      "epoch": 0.28793699944568685,
-      "grad_norm": 0.0017554003279656172,
-      "learning_rate": 0.0001999999593036116,
-      "loss": 46.0,
-      "step": 3766
-    },
-    {
-      "epoch": 0.2880134564290766,
-      "grad_norm": 0.0006303522386588156,
-      "learning_rate": 0.00019999995928193865,
-      "loss": 46.0,
-      "step": 3767
-    },
-    {
-      "epoch": 0.2880899134124663,
-      "grad_norm": 0.0005204095505177975,
-      "learning_rate": 0.00019999995926025995,
-      "loss": 46.0,
-      "step": 3768
-    },
-    {
-      "epoch": 0.288166370395856,
-      "grad_norm": 0.0023115079384297132,
-      "learning_rate": 0.00019999995923857547,
-      "loss": 46.0,
-      "step": 3769
-    },
-    {
-      "epoch": 0.2882428273792457,
-      "grad_norm": 0.0011257268488407135,
-      "learning_rate": 0.0001999999592168852,
-      "loss": 46.0,
-      "step": 3770
-    },
-    {
-      "epoch": 0.2883192843626355,
-      "grad_norm": 0.0020739634055644274,
-      "learning_rate": 0.00019999995919518917,
-      "loss": 46.0,
-      "step": 3771
-    },
-    {
-      "epoch": 0.2883957413460252,
-      "grad_norm": 0.0014096315717324615,
-      "learning_rate": 0.00019999995917348735,
-      "loss": 46.0,
-      "step": 3772
-    },
-    {
-      "epoch": 0.2884721983294149,
-      "grad_norm": 0.0005879343370907009,
-      "learning_rate": 0.00019999995915177978,
-      "loss": 46.0,
-      "step": 3773
-    },
-    {
-      "epoch": 0.28854865531280466,
-      "grad_norm": 0.000697542738635093,
-      "learning_rate": 0.00019999995913006646,
-      "loss": 46.0,
-      "step": 3774
-    },
-    {
-      "epoch": 0.28862511229619436,
-      "grad_norm": 0.0005846460117027164,
-      "learning_rate": 0.00019999995910834734,
-      "loss": 46.0,
-      "step": 3775
-    },
-    {
-      "epoch": 0.28870156927958407,
-      "grad_norm": 0.0014586938777938485,
-      "learning_rate": 0.00019999995908662248,
-      "loss": 46.0,
-      "step": 3776
-    },
-    {
-      "epoch": 0.2887780262629738,
-      "grad_norm": 0.0016310411738231778,
-      "learning_rate": 0.00019999995906489181,
-      "loss": 46.0,
-      "step": 3777
-    },
-    {
-      "epoch": 0.28885448324636354,
-      "grad_norm": 0.0033955464605242014,
-      "learning_rate": 0.00019999995904315543,
-      "loss": 46.0,
-      "step": 3778
-    },
-    {
-      "epoch": 0.28893094022975324,
-      "grad_norm": 0.0020739377941936255,
-      "learning_rate": 0.00019999995902141322,
-      "loss": 46.0,
-      "step": 3779
-    },
-    {
-      "epoch": 0.28900739721314295,
-      "grad_norm": 0.003352381521835923,
-      "learning_rate": 0.0001999999589996653,
-      "loss": 46.0,
-      "step": 3780
-    },
-    {
-      "epoch": 0.28908385419653265,
-      "grad_norm": 0.002708357758820057,
-      "learning_rate": 0.00019999995897791156,
-      "loss": 46.0,
-      "step": 3781
-    },
-    {
-      "epoch": 0.2891603111799224,
-      "grad_norm": 0.0062057143077254295,
-      "learning_rate": 0.00019999995895615206,
-      "loss": 46.0,
-      "step": 3782
-    },
-    {
-      "epoch": 0.2892367681633121,
-      "grad_norm": 0.0014678441220894456,
-      "learning_rate": 0.00019999995893438678,
-      "loss": 46.0,
-      "step": 3783
-    },
-    {
-      "epoch": 0.2893132251467018,
-      "grad_norm": 0.0005883380654267967,
-      "learning_rate": 0.00019999995891261576,
-      "loss": 46.0,
-      "step": 3784
-    },
-    {
-      "epoch": 0.28938968213009153,
-      "grad_norm": 0.0006215718458406627,
-      "learning_rate": 0.00019999995889083896,
-      "loss": 46.0,
-      "step": 3785
-    },
-    {
-      "epoch": 0.2894661391134813,
-      "grad_norm": 0.0015106117352843285,
-      "learning_rate": 0.0001999999588690564,
-      "loss": 46.0,
-      "step": 3786
-    },
-    {
-      "epoch": 0.289542596096871,
-      "grad_norm": 0.0013974400935694575,
-      "learning_rate": 0.00019999995884726804,
-      "loss": 46.0,
-      "step": 3787
-    },
-    {
-      "epoch": 0.2896190530802607,
-      "grad_norm": 0.000646473781671375,
-      "learning_rate": 0.00019999995882547393,
-      "loss": 46.0,
-      "step": 3788
-    },
-    {
-      "epoch": 0.28969551006365046,
-      "grad_norm": 0.0031547234393656254,
-      "learning_rate": 0.00019999995880367406,
-      "loss": 46.0,
-      "step": 3789
-    },
-    {
-      "epoch": 0.28977196704704017,
-      "grad_norm": 0.0005789370043203235,
-      "learning_rate": 0.0001999999587818684,
-      "loss": 46.0,
-      "step": 3790
-    },
-    {
-      "epoch": 0.2898484240304299,
-      "grad_norm": 0.003012710716575384,
-      "learning_rate": 0.000199999958760057,
-      "loss": 46.0,
-      "step": 3791
-    },
-    {
-      "epoch": 0.2899248810138196,
-      "grad_norm": 0.0006698978831991553,
-      "learning_rate": 0.0001999999587382398,
-      "loss": 46.0,
-      "step": 3792
-    },
-    {
-      "epoch": 0.29000133799720934,
-      "grad_norm": 0.0009188276017084718,
-      "learning_rate": 0.00019999995871641682,
-      "loss": 46.0,
-      "step": 3793
-    },
-    {
-      "epoch": 0.29007779498059905,
-      "grad_norm": 0.0014205612242221832,
-      "learning_rate": 0.00019999995869458812,
-      "loss": 46.0,
-      "step": 3794
-    },
-    {
-      "epoch": 0.29015425196398875,
-      "grad_norm": 0.0007790869567543268,
-      "learning_rate": 0.00019999995867275362,
-      "loss": 46.0,
-      "step": 3795
-    },
-    {
-      "epoch": 0.29023070894737846,
-      "grad_norm": 0.00046087370719760656,
-      "learning_rate": 0.00019999995865091337,
-      "loss": 46.0,
-      "step": 3796
-    },
-    {
-      "epoch": 0.2903071659307682,
-      "grad_norm": 0.0009427714976482093,
-      "learning_rate": 0.00019999995862906734,
-      "loss": 46.0,
-      "step": 3797
-    },
-    {
-      "epoch": 0.2903836229141579,
-      "grad_norm": 0.0006847182521596551,
-      "learning_rate": 0.00019999995860721552,
-      "loss": 46.0,
-      "step": 3798
-    },
-    {
-      "epoch": 0.29046007989754763,
-      "grad_norm": 0.0008071879274211824,
-      "learning_rate": 0.00019999995858535795,
-      "loss": 46.0,
-      "step": 3799
-    },
-    {
-      "epoch": 0.29053653688093733,
-      "grad_norm": 0.0011825787369161844,
-      "learning_rate": 0.00019999995856349458,
-      "loss": 46.0,
-      "step": 3800
-    },
-    {
-      "epoch": 0.2906129938643271,
-      "grad_norm": 0.0016411119140684605,
-      "learning_rate": 0.0001999999585416255,
-      "loss": 46.0,
-      "step": 3801
-    },
-    {
-      "epoch": 0.2906894508477168,
-      "grad_norm": 0.0011062535922974348,
-      "learning_rate": 0.0001999999585197506,
-      "loss": 46.0,
-      "step": 3802
-    },
-    {
-      "epoch": 0.2907659078311065,
-      "grad_norm": 0.0021068945061415434,
-      "learning_rate": 0.00019999995849786996,
-      "loss": 46.0,
-      "step": 3803
-    },
-    {
-      "epoch": 0.29084236481449627,
-      "grad_norm": 0.004064976237714291,
-      "learning_rate": 0.00019999995847598353,
-      "loss": 46.0,
-      "step": 3804
-    },
-    {
-      "epoch": 0.290918821797886,
-      "grad_norm": 0.0008881333633325994,
-      "learning_rate": 0.00019999995845409134,
-      "loss": 46.0,
-      "step": 3805
-    },
-    {
-      "epoch": 0.2909952787812757,
-      "grad_norm": 0.0008308746037073433,
-      "learning_rate": 0.00019999995843219339,
-      "loss": 46.0,
-      "step": 3806
-    },
-    {
-      "epoch": 0.2910717357646654,
-      "grad_norm": 0.0015752946492284536,
-      "learning_rate": 0.00019999995841028966,
-      "loss": 46.0,
-      "step": 3807
-    },
-    {
-      "epoch": 0.29114819274805515,
-      "grad_norm": 0.0019235041690990329,
-      "learning_rate": 0.00019999995838838015,
-      "loss": 46.0,
-      "step": 3808
-    },
-    {
-      "epoch": 0.29122464973144485,
-      "grad_norm": 0.0009915533009916544,
-      "learning_rate": 0.00019999995836646488,
-      "loss": 46.0,
-      "step": 3809
-    },
-    {
-      "epoch": 0.29130110671483456,
-      "grad_norm": 0.004070097114890814,
-      "learning_rate": 0.00019999995834454385,
-      "loss": 46.0,
-      "step": 3810
-    },
-    {
-      "epoch": 0.29137756369822426,
-      "grad_norm": 0.0016343813622370362,
-      "learning_rate": 0.00019999995832261703,
-      "loss": 46.0,
-      "step": 3811
-    },
-    {
-      "epoch": 0.291454020681614,
-      "grad_norm": 0.0010043747024610639,
-      "learning_rate": 0.00019999995830068446,
-      "loss": 46.0,
-      "step": 3812
-    },
-    {
-      "epoch": 0.29153047766500373,
-      "grad_norm": 0.0012621752684935927,
-      "learning_rate": 0.00019999995827874612,
-      "loss": 46.0,
-      "step": 3813
-    },
-    {
-      "epoch": 0.29160693464839343,
-      "grad_norm": 0.00109544035512954,
-      "learning_rate": 0.000199999958256802,
-      "loss": 46.0,
-      "step": 3814
-    },
-    {
-      "epoch": 0.2916833916317832,
-      "grad_norm": 0.0036395767237991095,
-      "learning_rate": 0.0001999999582348521,
-      "loss": 46.0,
-      "step": 3815
-    },
-    {
-      "epoch": 0.2917598486151729,
-      "grad_norm": 0.001107160234823823,
-      "learning_rate": 0.00019999995821289645,
-      "loss": 46.0,
-      "step": 3816
-    },
-    {
-      "epoch": 0.2918363055985626,
-      "grad_norm": 0.0004128957516513765,
-      "learning_rate": 0.00019999995819093504,
-      "loss": 46.0,
-      "step": 3817
-    },
-    {
-      "epoch": 0.2919127625819523,
-      "grad_norm": 0.011029314249753952,
-      "learning_rate": 0.00019999995816896783,
-      "loss": 46.0,
-      "step": 3818
-    },
-    {
-      "epoch": 0.2919892195653421,
-      "grad_norm": 0.0036829912569373846,
-      "learning_rate": 0.00019999995814699487,
-      "loss": 46.0,
-      "step": 3819
-    },
-    {
-      "epoch": 0.2920656765487318,
-      "grad_norm": 0.0009113659616559744,
-      "learning_rate": 0.00019999995812501614,
-      "loss": 46.0,
-      "step": 3820
-    },
-    {
-      "epoch": 0.2921421335321215,
-      "grad_norm": 0.00145654974039644,
-      "learning_rate": 0.00019999995810303164,
-      "loss": 46.0,
-      "step": 3821
-    },
-    {
-      "epoch": 0.2922185905155112,
-      "grad_norm": 0.0010323075111955404,
-      "learning_rate": 0.00019999995808104136,
-      "loss": 46.0,
-      "step": 3822
-    },
-    {
-      "epoch": 0.29229504749890095,
-      "grad_norm": 0.0008147822809405625,
-      "learning_rate": 0.0001999999580590453,
-      "loss": 46.0,
-      "step": 3823
-    },
-    {
-      "epoch": 0.29237150448229066,
-      "grad_norm": 0.000815986655652523,
-      "learning_rate": 0.00019999995803704352,
-      "loss": 46.0,
-      "step": 3824
-    },
-    {
-      "epoch": 0.29244796146568036,
-      "grad_norm": 0.0014565408928319812,
-      "learning_rate": 0.00019999995801503592,
-      "loss": 46.0,
-      "step": 3825
-    },
-    {
-      "epoch": 0.29252441844907007,
-      "grad_norm": 0.0014035555068403482,
-      "learning_rate": 0.0001999999579930226,
-      "loss": 46.0,
-      "step": 3826
-    },
-    {
-      "epoch": 0.29260087543245983,
-      "grad_norm": 0.0025402100291103125,
-      "learning_rate": 0.00019999995797100346,
-      "loss": 46.0,
-      "step": 3827
-    },
-    {
-      "epoch": 0.29267733241584953,
-      "grad_norm": 0.0037015005946159363,
-      "learning_rate": 0.0001999999579489786,
-      "loss": 46.0,
-      "step": 3828
-    },
-    {
-      "epoch": 0.29275378939923924,
-      "grad_norm": 0.002406664425507188,
-      "learning_rate": 0.00019999995792694794,
-      "loss": 46.0,
-      "step": 3829
-    },
-    {
-      "epoch": 0.292830246382629,
-      "grad_norm": 0.0012523853220045567,
-      "learning_rate": 0.00019999995790491147,
-      "loss": 46.0,
-      "step": 3830
-    },
-    {
-      "epoch": 0.2929067033660187,
-      "grad_norm": 0.002196969697251916,
-      "learning_rate": 0.0001999999578828693,
-      "loss": 46.0,
-      "step": 3831
-    },
-    {
-      "epoch": 0.2929831603494084,
-      "grad_norm": 0.004046255256980658,
-      "learning_rate": 0.00019999995786082133,
-      "loss": 46.0,
-      "step": 3832
-    },
-    {
-      "epoch": 0.2930596173327981,
-      "grad_norm": 0.0005504401633515954,
-      "learning_rate": 0.0001999999578387676,
-      "loss": 46.0,
-      "step": 3833
-    },
-    {
-      "epoch": 0.2931360743161879,
-      "grad_norm": 0.001778360572643578,
-      "learning_rate": 0.00019999995781670808,
-      "loss": 46.0,
-      "step": 3834
-    },
-    {
-      "epoch": 0.2932125312995776,
-      "grad_norm": 0.0012232806766405702,
-      "learning_rate": 0.00019999995779464283,
-      "loss": 46.0,
-      "step": 3835
-    },
-    {
-      "epoch": 0.2932889882829673,
-      "grad_norm": 0.002202985342592001,
-      "learning_rate": 0.00019999995777257178,
-      "loss": 46.0,
-      "step": 3836
-    },
-    {
-      "epoch": 0.293365445266357,
-      "grad_norm": 0.0006656167097389698,
-      "learning_rate": 0.00019999995775049495,
-      "loss": 46.0,
-      "step": 3837
-    },
-    {
-      "epoch": 0.29344190224974676,
-      "grad_norm": 0.0006763612036593258,
-      "learning_rate": 0.00019999995772841239,
-      "loss": 46.0,
-      "step": 3838
-    },
-    {
-      "epoch": 0.29351835923313646,
-      "grad_norm": 0.0004571889585349709,
-      "learning_rate": 0.00019999995770632402,
-      "loss": 46.0,
-      "step": 3839
-    },
-    {
-      "epoch": 0.29359481621652617,
-      "grad_norm": 0.0016356800915673375,
-      "learning_rate": 0.00019999995768422993,
-      "loss": 46.0,
-      "step": 3840
-    },
-    {
-      "epoch": 0.29367127319991587,
-      "grad_norm": 0.0015060099540278316,
-      "learning_rate": 0.00019999995766213,
-      "loss": 46.0,
-      "step": 3841
-    },
-    {
-      "epoch": 0.29374773018330563,
-      "grad_norm": 0.003492431016638875,
-      "learning_rate": 0.00019999995764002435,
-      "loss": 46.0,
-      "step": 3842
-    },
-    {
-      "epoch": 0.29382418716669534,
-      "grad_norm": 0.0019442737102508545,
-      "learning_rate": 0.0001999999576179129,
-      "loss": 46.0,
-      "step": 3843
-    },
-    {
-      "epoch": 0.29390064415008504,
-      "grad_norm": 0.0009575521107763052,
-      "learning_rate": 0.00019999995759579573,
-      "loss": 46.0,
-      "step": 3844
-    },
-    {
-      "epoch": 0.2939771011334748,
-      "grad_norm": 0.00020072500046808273,
-      "learning_rate": 0.00019999995757367275,
-      "loss": 46.0,
-      "step": 3845
-    },
-    {
-      "epoch": 0.2940535581168645,
-      "grad_norm": 0.0019815845880657434,
-      "learning_rate": 0.00019999995755154402,
-      "loss": 46.0,
-      "step": 3846
-    },
-    {
-      "epoch": 0.2941300151002542,
-      "grad_norm": 0.0010517174378037453,
-      "learning_rate": 0.0001999999575294095,
-      "loss": 46.0,
-      "step": 3847
-    },
-    {
-      "epoch": 0.2942064720836439,
-      "grad_norm": 0.002135148737579584,
-      "learning_rate": 0.00019999995750726924,
-      "loss": 46.0,
-      "step": 3848
-    },
-    {
-      "epoch": 0.2942829290670337,
-      "grad_norm": 0.001201451406814158,
-      "learning_rate": 0.0001999999574851232,
-      "loss": 46.0,
-      "step": 3849
-    },
-    {
-      "epoch": 0.2943593860504234,
-      "grad_norm": 0.001454431563615799,
-      "learning_rate": 0.00019999995746297134,
-      "loss": 46.0,
-      "step": 3850
-    },
-    {
-      "epoch": 0.2944358430338131,
-      "grad_norm": 0.022603917866945267,
-      "learning_rate": 0.0001999999574408138,
-      "loss": 46.0,
-      "step": 3851
-    },
-    {
-      "epoch": 0.2945123000172028,
-      "grad_norm": 0.000887357338797301,
-      "learning_rate": 0.00019999995741865043,
-      "loss": 46.0,
-      "step": 3852
-    },
-    {
-      "epoch": 0.29458875700059256,
-      "grad_norm": 0.0008037868537940085,
-      "learning_rate": 0.0001999999573964813,
-      "loss": 46.0,
-      "step": 3853
-    },
-    {
-      "epoch": 0.29466521398398227,
-      "grad_norm": 0.002315834164619446,
-      "learning_rate": 0.00019999995737430643,
-      "loss": 46.0,
-      "step": 3854
-    },
-    {
-      "epoch": 0.29474167096737197,
-      "grad_norm": 0.0032458745408803225,
-      "learning_rate": 0.00019999995735212576,
-      "loss": 46.0,
-      "step": 3855
-    },
-    {
-      "epoch": 0.2948181279507617,
-      "grad_norm": 0.0006478840950876474,
-      "learning_rate": 0.00019999995732993933,
-      "loss": 46.0,
-      "step": 3856
-    },
-    {
-      "epoch": 0.29489458493415144,
-      "grad_norm": 0.012460929341614246,
-      "learning_rate": 0.00019999995730774712,
-      "loss": 46.0,
-      "step": 3857
-    },
-    {
-      "epoch": 0.29497104191754114,
-      "grad_norm": 0.0012602087808772922,
-      "learning_rate": 0.00019999995728554914,
-      "loss": 46.0,
-      "step": 3858
-    },
-    {
-      "epoch": 0.29504749890093085,
-      "grad_norm": 0.008179609663784504,
-      "learning_rate": 0.00019999995726334538,
-      "loss": 46.0,
-      "step": 3859
-    },
-    {
-      "epoch": 0.2951239558843206,
-      "grad_norm": 0.0030665250960737467,
-      "learning_rate": 0.0001999999572411359,
-      "loss": 46.0,
-      "step": 3860
-    },
-    {
-      "epoch": 0.2952004128677103,
-      "grad_norm": 0.003528802189975977,
-      "learning_rate": 0.00019999995721892063,
-      "loss": 46.0,
-      "step": 3861
-    },
-    {
-      "epoch": 0.2952768698511,
-      "grad_norm": 0.0010181349935010076,
-      "learning_rate": 0.00019999995719669955,
-      "loss": 46.0,
-      "step": 3862
-    },
-    {
-      "epoch": 0.2953533268344897,
-      "grad_norm": 0.0004096160701010376,
-      "learning_rate": 0.00019999995717447276,
-      "loss": 46.0,
-      "step": 3863
-    },
-    {
-      "epoch": 0.2954297838178795,
-      "grad_norm": 0.0006414162344299257,
-      "learning_rate": 0.00019999995715224014,
-      "loss": 46.0,
-      "step": 3864
-    },
-    {
-      "epoch": 0.2955062408012692,
-      "grad_norm": 0.0021652875002473593,
-      "learning_rate": 0.0001999999571300018,
-      "loss": 46.0,
-      "step": 3865
-    },
-    {
-      "epoch": 0.2955826977846589,
-      "grad_norm": 0.00095454475376755,
-      "learning_rate": 0.00019999995710775768,
-      "loss": 46.0,
-      "step": 3866
-    },
-    {
-      "epoch": 0.2956591547680486,
-      "grad_norm": 0.002390076406300068,
-      "learning_rate": 0.0001999999570855078,
-      "loss": 46.0,
-      "step": 3867
-    },
-    {
-      "epoch": 0.29573561175143837,
-      "grad_norm": 0.0016897311434149742,
-      "learning_rate": 0.00019999995706325213,
-      "loss": 46.0,
-      "step": 3868
-    },
-    {
-      "epoch": 0.29581206873482807,
-      "grad_norm": 0.001022031414322555,
-      "learning_rate": 0.0001999999570409907,
-      "loss": 46.0,
-      "step": 3869
-    },
-    {
-      "epoch": 0.2958885257182178,
-      "grad_norm": 0.0006169632542878389,
-      "learning_rate": 0.00019999995701872346,
-      "loss": 46.0,
-      "step": 3870
-    },
-    {
-      "epoch": 0.2959649827016075,
-      "grad_norm": 0.001120051252655685,
-      "learning_rate": 0.00019999995699645048,
-      "loss": 46.0,
-      "step": 3871
-    },
-    {
-      "epoch": 0.29604143968499724,
-      "grad_norm": 0.0021365159191191196,
-      "learning_rate": 0.00019999995697417172,
-      "loss": 46.0,
-      "step": 3872
-    },
-    {
-      "epoch": 0.29611789666838695,
-      "grad_norm": 0.001412899000570178,
-      "learning_rate": 0.00019999995695188725,
-      "loss": 46.0,
-      "step": 3873
-    },
-    {
-      "epoch": 0.29619435365177665,
-      "grad_norm": 0.0010116349440068007,
-      "learning_rate": 0.00019999995692959694,
-      "loss": 46.0,
-      "step": 3874
-    },
-    {
-      "epoch": 0.2962708106351664,
-      "grad_norm": 0.0009577778982929885,
-      "learning_rate": 0.00019999995690730092,
-      "loss": 46.0,
-      "step": 3875
-    },
-    {
-      "epoch": 0.2963472676185561,
-      "grad_norm": 0.0007074445602484047,
-      "learning_rate": 0.00019999995688499907,
-      "loss": 46.0,
-      "step": 3876
-    },
-    {
-      "epoch": 0.2964237246019458,
-      "grad_norm": 0.0004960338701494038,
-      "learning_rate": 0.00019999995686269148,
-      "loss": 46.0,
-      "step": 3877
-    },
-    {
-      "epoch": 0.29650018158533553,
-      "grad_norm": 0.000575447513256222,
-      "learning_rate": 0.00019999995684037814,
-      "loss": 46.0,
-      "step": 3878
-    },
-    {
-      "epoch": 0.2965766385687253,
-      "grad_norm": 0.0008145607425831258,
-      "learning_rate": 0.000199999956818059,
-      "loss": 46.0,
-      "step": 3879
-    },
-    {
-      "epoch": 0.296653095552115,
-      "grad_norm": 0.0005860779201611876,
-      "learning_rate": 0.00019999995679573408,
-      "loss": 46.0,
-      "step": 3880
-    },
-    {
-      "epoch": 0.2967295525355047,
-      "grad_norm": 0.0004820851900149137,
-      "learning_rate": 0.00019999995677340342,
-      "loss": 46.0,
-      "step": 3881
-    },
-    {
-      "epoch": 0.2968060095188944,
-      "grad_norm": 0.0011859680525958538,
-      "learning_rate": 0.000199999956751067,
-      "loss": 46.0,
-      "step": 3882
-    },
-    {
-      "epoch": 0.29688246650228417,
-      "grad_norm": 0.003930750302970409,
-      "learning_rate": 0.00019999995672872477,
-      "loss": 46.0,
-      "step": 3883
-    },
-    {
-      "epoch": 0.2969589234856739,
-      "grad_norm": 0.0008929710602387786,
-      "learning_rate": 0.0001999999567063768,
-      "loss": 46.0,
-      "step": 3884
-    },
-    {
-      "epoch": 0.2970353804690636,
-      "grad_norm": 0.0005970599595457315,
-      "learning_rate": 0.00019999995668402304,
-      "loss": 46.0,
-      "step": 3885
-    },
-    {
-      "epoch": 0.2971118374524533,
-      "grad_norm": 0.0006033756071701646,
-      "learning_rate": 0.00019999995666166354,
-      "loss": 46.0,
-      "step": 3886
-    },
-    {
-      "epoch": 0.29718829443584305,
-      "grad_norm": 0.0019969872664660215,
-      "learning_rate": 0.00019999995663929823,
-      "loss": 46.0,
-      "step": 3887
-    },
-    {
-      "epoch": 0.29726475141923275,
-      "grad_norm": 0.000959285069257021,
-      "learning_rate": 0.00019999995661692719,
-      "loss": 46.0,
-      "step": 3888
-    },
-    {
-      "epoch": 0.29734120840262246,
-      "grad_norm": 0.0007328496431000531,
-      "learning_rate": 0.00019999995659455036,
-      "loss": 46.0,
-      "step": 3889
-    },
-    {
-      "epoch": 0.2974176653860122,
-      "grad_norm": 0.0007862882921472192,
-      "learning_rate": 0.0001999999565721678,
-      "loss": 46.0,
-      "step": 3890
-    },
-    {
-      "epoch": 0.2974941223694019,
-      "grad_norm": 0.0022088452242314816,
-      "learning_rate": 0.0001999999565497794,
-      "loss": 46.0,
-      "step": 3891
-    },
-    {
-      "epoch": 0.29757057935279163,
-      "grad_norm": 0.0008607911877334118,
-      "learning_rate": 0.0001999999565273853,
-      "loss": 46.0,
-      "step": 3892
-    },
-    {
-      "epoch": 0.29764703633618134,
-      "grad_norm": 0.0006285980343818665,
-      "learning_rate": 0.00019999995650498537,
-      "loss": 46.0,
-      "step": 3893
-    },
-    {
-      "epoch": 0.2977234933195711,
-      "grad_norm": 0.0027029935736209154,
-      "learning_rate": 0.0001999999564825797,
-      "loss": 46.0,
-      "step": 3894
-    },
-    {
-      "epoch": 0.2977999503029608,
-      "grad_norm": 0.002828470664098859,
-      "learning_rate": 0.00019999995646016828,
-      "loss": 46.0,
-      "step": 3895
-    },
-    {
-      "epoch": 0.2978764072863505,
-      "grad_norm": 0.0025153513997793198,
-      "learning_rate": 0.00019999995643775104,
-      "loss": 46.0,
-      "step": 3896
-    },
-    {
-      "epoch": 0.2979528642697402,
-      "grad_norm": 0.002997411647811532,
-      "learning_rate": 0.00019999995641532806,
-      "loss": 46.0,
-      "step": 3897
-    },
-    {
-      "epoch": 0.29802932125313,
-      "grad_norm": 0.006343774497509003,
-      "learning_rate": 0.00019999995639289933,
-      "loss": 46.0,
-      "step": 3898
-    },
-    {
-      "epoch": 0.2981057782365197,
-      "grad_norm": 0.0021542010363191366,
-      "learning_rate": 0.0001999999563704648,
-      "loss": 46.0,
-      "step": 3899
-    },
-    {
-      "epoch": 0.2981822352199094,
-      "grad_norm": 0.0007479273481294513,
-      "learning_rate": 0.0001999999563480245,
-      "loss": 46.0,
-      "step": 3900
-    },
-    {
-      "epoch": 0.2982586922032991,
-      "grad_norm": 0.0007215498480945826,
-      "learning_rate": 0.00019999995632557845,
-      "loss": 46.0,
-      "step": 3901
-    },
-    {
-      "epoch": 0.29833514918668885,
-      "grad_norm": 0.0029518231749534607,
-      "learning_rate": 0.00019999995630312663,
-      "loss": 46.0,
-      "step": 3902
-    },
-    {
-      "epoch": 0.29841160617007856,
-      "grad_norm": 0.0011331290006637573,
-      "learning_rate": 0.00019999995628066904,
-      "loss": 46.0,
-      "step": 3903
-    },
-    {
-      "epoch": 0.29848806315346826,
-      "grad_norm": 0.0010614742059260607,
-      "learning_rate": 0.00019999995625820567,
-      "loss": 46.0,
-      "step": 3904
-    },
-    {
-      "epoch": 0.298564520136858,
-      "grad_norm": 0.0008829003199934959,
-      "learning_rate": 0.00019999995623573653,
-      "loss": 46.0,
-      "step": 3905
-    },
-    {
-      "epoch": 0.29864097712024773,
-      "grad_norm": 0.0013663776917383075,
-      "learning_rate": 0.00019999995621326161,
-      "loss": 46.0,
-      "step": 3906
-    },
-    {
-      "epoch": 0.29871743410363744,
-      "grad_norm": 0.007047569379210472,
-      "learning_rate": 0.00019999995619078095,
-      "loss": 46.0,
-      "step": 3907
-    },
-    {
-      "epoch": 0.29879389108702714,
-      "grad_norm": 0.002720632590353489,
-      "learning_rate": 0.0001999999561682945,
-      "loss": 46.0,
-      "step": 3908
-    },
-    {
-      "epoch": 0.2988703480704169,
-      "grad_norm": 0.0009586411179043353,
-      "learning_rate": 0.00019999995614580228,
-      "loss": 46.0,
-      "step": 3909
-    },
-    {
-      "epoch": 0.2989468050538066,
-      "grad_norm": 0.0021984048653393984,
-      "learning_rate": 0.0001999999561233043,
-      "loss": 46.0,
-      "step": 3910
-    },
-    {
-      "epoch": 0.2990232620371963,
-      "grad_norm": 0.0009016099502332509,
-      "learning_rate": 0.00019999995610080055,
-      "loss": 46.0,
-      "step": 3911
-    },
-    {
-      "epoch": 0.299099719020586,
-      "grad_norm": 0.0007355074631050229,
-      "learning_rate": 0.00019999995607829102,
-      "loss": 46.0,
-      "step": 3912
-    },
-    {
-      "epoch": 0.2991761760039758,
-      "grad_norm": 0.007912776432931423,
-      "learning_rate": 0.00019999995605577575,
-      "loss": 46.0,
-      "step": 3913
-    },
-    {
-      "epoch": 0.2992526329873655,
-      "grad_norm": 0.001287880353629589,
-      "learning_rate": 0.00019999995603325467,
-      "loss": 46.0,
-      "step": 3914
-    },
-    {
-      "epoch": 0.2993290899707552,
-      "grad_norm": 0.001875565038062632,
-      "learning_rate": 0.00019999995601072785,
-      "loss": 46.0,
-      "step": 3915
-    },
-    {
-      "epoch": 0.2994055469541449,
-      "grad_norm": 0.0014386377297341824,
-      "learning_rate": 0.00019999995598819526,
-      "loss": 46.0,
-      "step": 3916
-    },
-    {
-      "epoch": 0.29948200393753466,
-      "grad_norm": 0.0010124879190698266,
-      "learning_rate": 0.0001999999559656569,
-      "loss": 46.0,
-      "step": 3917
-    },
-    {
-      "epoch": 0.29955846092092436,
-      "grad_norm": 0.0017455739434808493,
-      "learning_rate": 0.00019999995594311275,
-      "loss": 46.0,
-      "step": 3918
-    },
-    {
-      "epoch": 0.29963491790431407,
-      "grad_norm": 0.001225010259076953,
-      "learning_rate": 0.00019999995592056283,
-      "loss": 46.0,
-      "step": 3919
-    },
-    {
-      "epoch": 0.29971137488770383,
-      "grad_norm": 0.0006972053088247776,
-      "learning_rate": 0.00019999995589800715,
-      "loss": 46.0,
-      "step": 3920
-    },
-    {
-      "epoch": 0.29978783187109354,
-      "grad_norm": 0.002167454455047846,
-      "learning_rate": 0.0001999999558754457,
-      "loss": 46.0,
-      "step": 3921
-    },
-    {
-      "epoch": 0.29986428885448324,
-      "grad_norm": 0.0012359477113932371,
-      "learning_rate": 0.0001999999558528785,
-      "loss": 46.0,
-      "step": 3922
-    },
-    {
-      "epoch": 0.29994074583787295,
-      "grad_norm": 0.000918172940146178,
-      "learning_rate": 0.0001999999558303055,
-      "loss": 46.0,
-      "step": 3923
-    },
-    {
-      "epoch": 0.3000172028212627,
-      "grad_norm": 0.00037611983134411275,
-      "learning_rate": 0.00019999995580772674,
-      "loss": 46.0,
-      "step": 3924
-    },
-    {
-      "epoch": 0.3000936598046524,
-      "grad_norm": 0.0004843119240831584,
-      "learning_rate": 0.00019999995578514224,
-      "loss": 46.0,
-      "step": 3925
-    },
-    {
-      "epoch": 0.3001701167880421,
-      "grad_norm": 0.0017777679022401571,
-      "learning_rate": 0.00019999995576255191,
-      "loss": 46.0,
-      "step": 3926
-    },
-    {
-      "epoch": 0.3002465737714318,
-      "grad_norm": 0.0005172256496734917,
-      "learning_rate": 0.00019999995573995587,
-      "loss": 46.0,
-      "step": 3927
-    },
-    {
-      "epoch": 0.3003230307548216,
-      "grad_norm": 0.0013011523988097906,
-      "learning_rate": 0.00019999995571735405,
-      "loss": 46.0,
-      "step": 3928
-    },
-    {
-      "epoch": 0.3003994877382113,
-      "grad_norm": 0.0007835791329853237,
-      "learning_rate": 0.00019999995569474642,
-      "loss": 46.0,
-      "step": 3929
-    },
-    {
-      "epoch": 0.300475944721601,
-      "grad_norm": 0.0007522383239120245,
-      "learning_rate": 0.00019999995567213306,
-      "loss": 46.0,
-      "step": 3930
-    },
-    {
-      "epoch": 0.30055240170499076,
-      "grad_norm": 0.004805904813110828,
-      "learning_rate": 0.00019999995564951392,
-      "loss": 46.0,
-      "step": 3931
-    },
-    {
-      "epoch": 0.30062885868838046,
-      "grad_norm": 0.002395146293565631,
-      "learning_rate": 0.000199999955626889,
-      "loss": 46.0,
-      "step": 3932
-    },
-    {
-      "epoch": 0.30070531567177017,
-      "grad_norm": 0.004002503585070372,
-      "learning_rate": 0.00019999995560425832,
-      "loss": 46.0,
-      "step": 3933
-    },
-    {
-      "epoch": 0.3007817726551599,
-      "grad_norm": 0.0005741655477322638,
-      "learning_rate": 0.00019999995558162188,
-      "loss": 46.0,
-      "step": 3934
-    },
-    {
-      "epoch": 0.30085822963854963,
-      "grad_norm": 0.003098508110269904,
-      "learning_rate": 0.00019999995555897965,
-      "loss": 46.0,
-      "step": 3935
-    },
-    {
-      "epoch": 0.30093468662193934,
-      "grad_norm": 0.00042936380486935377,
-      "learning_rate": 0.00019999995553633167,
-      "loss": 46.0,
-      "step": 3936
-    },
-    {
-      "epoch": 0.30101114360532905,
-      "grad_norm": 0.0006006730254739523,
-      "learning_rate": 0.0001999999555136779,
-      "loss": 46.0,
-      "step": 3937
-    },
-    {
-      "epoch": 0.30108760058871875,
-      "grad_norm": 0.0011409219587221742,
-      "learning_rate": 0.00019999995549101839,
-      "loss": 46.0,
-      "step": 3938
-    },
-    {
-      "epoch": 0.3011640575721085,
-      "grad_norm": 0.002090958645567298,
-      "learning_rate": 0.00019999995546835308,
-      "loss": 46.0,
-      "step": 3939
-    },
-    {
-      "epoch": 0.3012405145554982,
-      "grad_norm": 0.0014344867086037993,
-      "learning_rate": 0.000199999955445682,
-      "loss": 46.0,
-      "step": 3940
-    },
-    {
-      "epoch": 0.3013169715388879,
-      "grad_norm": 0.0014389441348612309,
-      "learning_rate": 0.0001999999554230052,
-      "loss": 46.0,
-      "step": 3941
-    },
-    {
-      "epoch": 0.30139342852227763,
-      "grad_norm": 0.0014627301134169102,
-      "learning_rate": 0.00019999995540032257,
-      "loss": 46.0,
-      "step": 3942
-    },
-    {
-      "epoch": 0.3014698855056674,
-      "grad_norm": 0.0008876899955794215,
-      "learning_rate": 0.0001999999553776342,
-      "loss": 46.0,
-      "step": 3943
-    },
-    {
-      "epoch": 0.3015463424890571,
-      "grad_norm": 0.0006306195282377303,
-      "learning_rate": 0.00019999995535494006,
-      "loss": 46.0,
-      "step": 3944
-    },
-    {
-      "epoch": 0.3016227994724468,
-      "grad_norm": 0.002020200015977025,
-      "learning_rate": 0.00019999995533224015,
-      "loss": 46.0,
-      "step": 3945
-    },
-    {
-      "epoch": 0.30169925645583656,
-      "grad_norm": 0.0005920823314227164,
-      "learning_rate": 0.00019999995530953446,
-      "loss": 46.0,
-      "step": 3946
-    },
-    {
-      "epoch": 0.30177571343922627,
-      "grad_norm": 0.000938529847189784,
-      "learning_rate": 0.000199999955286823,
-      "loss": 46.0,
-      "step": 3947
-    },
-    {
-      "epoch": 0.301852170422616,
-      "grad_norm": 0.0007786461501382291,
-      "learning_rate": 0.00019999995526410577,
-      "loss": 46.0,
-      "step": 3948
-    },
-    {
-      "epoch": 0.3019286274060057,
-      "grad_norm": 0.0006874241516925395,
-      "learning_rate": 0.0001999999552413828,
-      "loss": 46.0,
-      "step": 3949
-    },
-    {
-      "epoch": 0.30200508438939544,
-      "grad_norm": 0.004485182464122772,
-      "learning_rate": 0.00019999995521865403,
-      "loss": 46.0,
-      "step": 3950
-    },
-    {
-      "epoch": 0.30208154137278515,
-      "grad_norm": 0.008694938383996487,
-      "learning_rate": 0.0001999999551959195,
-      "loss": 46.0,
-      "step": 3951
-    },
-    {
-      "epoch": 0.30215799835617485,
-      "grad_norm": 0.0013990922598168254,
-      "learning_rate": 0.0001999999551731792,
-      "loss": 46.0,
-      "step": 3952
-    },
-    {
-      "epoch": 0.30223445533956456,
-      "grad_norm": 0.0009009944624267519,
-      "learning_rate": 0.00019999995515043313,
-      "loss": 46.0,
-      "step": 3953
-    },
-    {
-      "epoch": 0.3023109123229543,
-      "grad_norm": 0.0007439752225764096,
-      "learning_rate": 0.0001999999551276813,
-      "loss": 46.0,
-      "step": 3954
-    },
-    {
-      "epoch": 0.302387369306344,
-      "grad_norm": 0.0009529454400762916,
-      "learning_rate": 0.0001999999551049237,
-      "loss": 46.0,
-      "step": 3955
-    },
-    {
-      "epoch": 0.30246382628973373,
-      "grad_norm": 0.003594091860577464,
-      "learning_rate": 0.0001999999550821603,
-      "loss": 46.0,
-      "step": 3956
-    },
-    {
-      "epoch": 0.30254028327312343,
-      "grad_norm": 0.0011254452401772141,
-      "learning_rate": 0.00019999995505939116,
-      "loss": 46.0,
-      "step": 3957
-    },
-    {
-      "epoch": 0.3026167402565132,
-      "grad_norm": 0.0016184871783480048,
-      "learning_rate": 0.00019999995503661624,
-      "loss": 46.0,
-      "step": 3958
-    },
-    {
-      "epoch": 0.3026931972399029,
-      "grad_norm": 0.0006242019589990377,
-      "learning_rate": 0.00019999995501383556,
-      "loss": 46.0,
-      "step": 3959
-    },
-    {
-      "epoch": 0.3027696542232926,
-      "grad_norm": 0.0006411741487681866,
-      "learning_rate": 0.0001999999549910491,
-      "loss": 46.0,
-      "step": 3960
-    },
-    {
-      "epoch": 0.30284611120668237,
-      "grad_norm": 0.0021180137991905212,
-      "learning_rate": 0.00019999995496825686,
-      "loss": 46.0,
-      "step": 3961
-    },
-    {
-      "epoch": 0.3029225681900721,
-      "grad_norm": 0.00109868252184242,
-      "learning_rate": 0.00019999995494545888,
-      "loss": 46.0,
-      "step": 3962
-    },
-    {
-      "epoch": 0.3029990251734618,
-      "grad_norm": 0.00096026441315189,
-      "learning_rate": 0.0001999999549226551,
-      "loss": 46.0,
-      "step": 3963
-    },
-    {
-      "epoch": 0.3030754821568515,
-      "grad_norm": 0.008483518846333027,
-      "learning_rate": 0.00019999995489984558,
-      "loss": 46.0,
-      "step": 3964
-    },
-    {
-      "epoch": 0.30315193914024124,
-      "grad_norm": 0.0017056922661140561,
-      "learning_rate": 0.00019999995487703028,
-      "loss": 46.0,
-      "step": 3965
-    },
-    {
-      "epoch": 0.30322839612363095,
-      "grad_norm": 0.0007459048647433519,
-      "learning_rate": 0.0001999999548542092,
-      "loss": 46.0,
-      "step": 3966
-    },
-    {
-      "epoch": 0.30330485310702066,
-      "grad_norm": 0.001148242736235261,
-      "learning_rate": 0.00019999995483138236,
-      "loss": 46.0,
-      "step": 3967
-    },
-    {
-      "epoch": 0.30338131009041036,
-      "grad_norm": 0.0007320080767385662,
-      "learning_rate": 0.00019999995480854974,
-      "loss": 46.0,
-      "step": 3968
-    },
-    {
-      "epoch": 0.3034577670738001,
-      "grad_norm": 0.0035365570802241564,
-      "learning_rate": 0.00019999995478571137,
-      "loss": 46.0,
-      "step": 3969
-    },
-    {
-      "epoch": 0.3035342240571898,
-      "grad_norm": 0.0072437673807144165,
-      "learning_rate": 0.0001999999547628672,
-      "loss": 46.0,
-      "step": 3970
-    },
-    {
-      "epoch": 0.30361068104057953,
-      "grad_norm": 0.0005270031979307532,
-      "learning_rate": 0.0001999999547400173,
-      "loss": 46.0,
-      "step": 3971
-    },
-    {
-      "epoch": 0.30368713802396924,
-      "grad_norm": 0.0015035016695037484,
-      "learning_rate": 0.0001999999547171616,
-      "loss": 46.0,
-      "step": 3972
-    },
-    {
-      "epoch": 0.303763595007359,
-      "grad_norm": 0.009577272459864616,
-      "learning_rate": 0.00019999995469430014,
-      "loss": 46.0,
-      "step": 3973
-    },
-    {
-      "epoch": 0.3038400519907487,
-      "grad_norm": 0.0032476128544658422,
-      "learning_rate": 0.00019999995467143294,
-      "loss": 46.0,
-      "step": 3974
-    },
-    {
-      "epoch": 0.3039165089741384,
-      "grad_norm": 0.0009590466506779194,
-      "learning_rate": 0.00019999995464855993,
-      "loss": 46.0,
-      "step": 3975
-    },
-    {
-      "epoch": 0.30399296595752817,
-      "grad_norm": 0.0018426980823278427,
-      "learning_rate": 0.00019999995462568115,
-      "loss": 46.0,
-      "step": 3976
-    },
-    {
-      "epoch": 0.3040694229409179,
-      "grad_norm": 0.0026536802761256695,
-      "learning_rate": 0.00019999995460279662,
-      "loss": 46.0,
-      "step": 3977
-    },
-    {
-      "epoch": 0.3041458799243076,
-      "grad_norm": 0.0021176941227167845,
-      "learning_rate": 0.00019999995457990633,
-      "loss": 46.0,
-      "step": 3978
-    },
-    {
-      "epoch": 0.3042223369076973,
-      "grad_norm": 0.0017880977829918265,
-      "learning_rate": 0.00019999995455701023,
-      "loss": 46.0,
-      "step": 3979
-    },
-    {
-      "epoch": 0.30429879389108705,
-      "grad_norm": 0.0007800829480402172,
-      "learning_rate": 0.00019999995453410838,
-      "loss": 46.0,
-      "step": 3980
-    },
-    {
-      "epoch": 0.30437525087447675,
-      "grad_norm": 0.0011141517898067832,
-      "learning_rate": 0.00019999995451120079,
-      "loss": 46.0,
-      "step": 3981
-    },
-    {
-      "epoch": 0.30445170785786646,
-      "grad_norm": 0.0017600301653146744,
-      "learning_rate": 0.0001999999544882874,
-      "loss": 46.0,
-      "step": 3982
-    },
-    {
-      "epoch": 0.30452816484125617,
-      "grad_norm": 0.0010085272369906306,
-      "learning_rate": 0.00019999995446536825,
-      "loss": 46.0,
-      "step": 3983
-    },
-    {
-      "epoch": 0.3046046218246459,
-      "grad_norm": 0.0015926266787573695,
-      "learning_rate": 0.0001999999544424433,
-      "loss": 46.0,
-      "step": 3984
-    },
-    {
-      "epoch": 0.30468107880803563,
-      "grad_norm": 0.0007591758039779961,
-      "learning_rate": 0.00019999995441951263,
-      "loss": 46.0,
-      "step": 3985
-    },
-    {
-      "epoch": 0.30475753579142534,
-      "grad_norm": 0.000749889062717557,
-      "learning_rate": 0.00019999995439657617,
-      "loss": 46.0,
-      "step": 3986
-    },
-    {
-      "epoch": 0.30483399277481504,
-      "grad_norm": 0.0007087599369697273,
-      "learning_rate": 0.00019999995437363393,
-      "loss": 46.0,
-      "step": 3987
-    },
-    {
-      "epoch": 0.3049104497582048,
-      "grad_norm": 0.001025100820697844,
-      "learning_rate": 0.00019999995435068596,
-      "loss": 46.0,
-      "step": 3988
-    },
-    {
-      "epoch": 0.3049869067415945,
-      "grad_norm": 0.002785252872854471,
-      "learning_rate": 0.00019999995432773218,
-      "loss": 46.0,
-      "step": 3989
-    },
-    {
-      "epoch": 0.3050633637249842,
-      "grad_norm": 0.0014751151902601123,
-      "learning_rate": 0.00019999995430477262,
-      "loss": 46.0,
-      "step": 3990
-    },
-    {
-      "epoch": 0.305139820708374,
-      "grad_norm": 0.0025423127226531506,
-      "learning_rate": 0.00019999995428180732,
-      "loss": 46.0,
-      "step": 3991
-    },
-    {
-      "epoch": 0.3052162776917637,
-      "grad_norm": 0.0009441522997803986,
-      "learning_rate": 0.00019999995425883625,
-      "loss": 46.0,
-      "step": 3992
-    },
-    {
-      "epoch": 0.3052927346751534,
-      "grad_norm": 0.0018807377200573683,
-      "learning_rate": 0.0001999999542358594,
-      "loss": 46.0,
-      "step": 3993
-    },
-    {
-      "epoch": 0.3053691916585431,
-      "grad_norm": 0.002795872278511524,
-      "learning_rate": 0.00019999995421287676,
-      "loss": 46.0,
-      "step": 3994
-    },
-    {
-      "epoch": 0.30544564864193285,
-      "grad_norm": 0.00199322821572423,
-      "learning_rate": 0.0001999999541898884,
-      "loss": 46.0,
-      "step": 3995
-    },
-    {
-      "epoch": 0.30552210562532256,
-      "grad_norm": 0.0005783766973763704,
-      "learning_rate": 0.00019999995416689423,
-      "loss": 46.0,
-      "step": 3996
-    },
-    {
-      "epoch": 0.30559856260871227,
-      "grad_norm": 0.0011058659292757511,
-      "learning_rate": 0.00019999995414389429,
-      "loss": 46.0,
-      "step": 3997
-    },
-    {
-      "epoch": 0.30567501959210197,
-      "grad_norm": 0.0016042126808315516,
-      "learning_rate": 0.0001999999541208886,
-      "loss": 46.0,
-      "step": 3998
-    },
-    {
-      "epoch": 0.30575147657549173,
-      "grad_norm": 0.003017428098246455,
-      "learning_rate": 0.00019999995409787714,
-      "loss": 46.0,
-      "step": 3999
-    },
-    {
-      "epoch": 0.30582793355888144,
-      "grad_norm": 0.0010716089745983481,
-      "learning_rate": 0.0001999999540748599,
-      "loss": 46.0,
-      "step": 4000
-    },
-    {
-      "epoch": 0.30590439054227114,
-      "grad_norm": 0.0009396169916726649,
-      "learning_rate": 0.0001999999540518369,
-      "loss": 46.0,
-      "step": 4001
-    },
-    {
-      "epoch": 0.30598084752566085,
-      "grad_norm": 0.002436660695821047,
-      "learning_rate": 0.00019999995402880812,
-      "loss": 46.0,
-      "step": 4002
-    },
-    {
-      "epoch": 0.3060573045090506,
-      "grad_norm": 0.0020608168561011553,
-      "learning_rate": 0.00019999995400577357,
-      "loss": 46.0,
-      "step": 4003
-    },
-    {
-      "epoch": 0.3061337614924403,
-      "grad_norm": 0.001335165579803288,
-      "learning_rate": 0.00019999995398273327,
-      "loss": 46.0,
-      "step": 4004
-    },
-    {
-      "epoch": 0.30621021847583,
-      "grad_norm": 0.0012055481784045696,
-      "learning_rate": 0.0001999999539596872,
-      "loss": 46.0,
-      "step": 4005
-    },
-    {
-      "epoch": 0.3062866754592198,
-      "grad_norm": 0.001344876131042838,
-      "learning_rate": 0.00019999995393663535,
-      "loss": 46.0,
-      "step": 4006
-    },
-    {
-      "epoch": 0.3063631324426095,
-      "grad_norm": 0.0011385285761207342,
-      "learning_rate": 0.00019999995391357773,
-      "loss": 46.0,
-      "step": 4007
-    },
-    {
-      "epoch": 0.3064395894259992,
-      "grad_norm": 0.001564131467603147,
-      "learning_rate": 0.00019999995389051437,
-      "loss": 46.0,
-      "step": 4008
-    },
-    {
-      "epoch": 0.3065160464093889,
-      "grad_norm": 0.0036503213923424482,
-      "learning_rate": 0.00019999995386744517,
-      "loss": 46.0,
-      "step": 4009
-    },
-    {
-      "epoch": 0.30659250339277866,
-      "grad_norm": 0.003316072979941964,
-      "learning_rate": 0.00019999995384437024,
-      "loss": 46.0,
-      "step": 4010
-    },
-    {
-      "epoch": 0.30666896037616836,
-      "grad_norm": 0.007509085815399885,
-      "learning_rate": 0.00019999995382128955,
-      "loss": 46.0,
-      "step": 4011
-    },
-    {
-      "epoch": 0.30674541735955807,
-      "grad_norm": 0.0006065723719075322,
-      "learning_rate": 0.0001999999537982031,
-      "loss": 46.0,
-      "step": 4012
-    },
-    {
-      "epoch": 0.3068218743429478,
-      "grad_norm": 0.0011315278243273497,
-      "learning_rate": 0.00019999995377511086,
-      "loss": 46.0,
-      "step": 4013
-    },
-    {
-      "epoch": 0.30689833132633754,
-      "grad_norm": 0.0018448256887495518,
-      "learning_rate": 0.00019999995375201283,
-      "loss": 46.0,
-      "step": 4014
-    },
-    {
-      "epoch": 0.30697478830972724,
-      "grad_norm": 0.0007080396171659231,
-      "learning_rate": 0.00019999995372890907,
-      "loss": 46.0,
-      "step": 4015
-    },
-    {
-      "epoch": 0.30705124529311695,
-      "grad_norm": 0.0004142904654145241,
-      "learning_rate": 0.00019999995370579955,
-      "loss": 46.0,
-      "step": 4016
-    },
-    {
-      "epoch": 0.30712770227650665,
-      "grad_norm": 0.0011961066629737616,
-      "learning_rate": 0.0001999999536826842,
-      "loss": 46.0,
-      "step": 4017
-    },
-    {
-      "epoch": 0.3072041592598964,
-      "grad_norm": 0.000924957450479269,
-      "learning_rate": 0.00019999995365956312,
-      "loss": 46.0,
-      "step": 4018
-    },
-    {
-      "epoch": 0.3072806162432861,
-      "grad_norm": 0.0007961697410792112,
-      "learning_rate": 0.00019999995363643628,
-      "loss": 46.0,
-      "step": 4019
-    },
-    {
-      "epoch": 0.3073570732266758,
-      "grad_norm": 0.0012061174493283033,
-      "learning_rate": 0.00019999995361330366,
-      "loss": 46.0,
-      "step": 4020
-    },
-    {
-      "epoch": 0.3074335302100656,
-      "grad_norm": 0.0011080013355240226,
-      "learning_rate": 0.00019999995359016527,
-      "loss": 46.0,
-      "step": 4021
-    },
-    {
-      "epoch": 0.3075099871934553,
-      "grad_norm": 0.0007111974409781396,
-      "learning_rate": 0.00019999995356702107,
-      "loss": 46.0,
-      "step": 4022
-    },
-    {
-      "epoch": 0.307586444176845,
-      "grad_norm": 0.007970698177814484,
-      "learning_rate": 0.00019999995354387116,
-      "loss": 46.0,
-      "step": 4023
-    },
-    {
-      "epoch": 0.3076629011602347,
-      "grad_norm": 0.0012143987696617842,
-      "learning_rate": 0.00019999995352071545,
-      "loss": 46.0,
-      "step": 4024
-    },
-    {
-      "epoch": 0.30773935814362446,
-      "grad_norm": 0.0029039548244327307,
-      "learning_rate": 0.000199999953497554,
-      "loss": 46.0,
-      "step": 4025
-    },
-    {
-      "epoch": 0.30781581512701417,
-      "grad_norm": 0.0006623852532356977,
-      "learning_rate": 0.00019999995347438676,
-      "loss": 46.0,
-      "step": 4026
-    },
-    {
-      "epoch": 0.3078922721104039,
-      "grad_norm": 0.0012280797818675637,
-      "learning_rate": 0.00019999995345121375,
-      "loss": 46.0,
-      "step": 4027
-    },
-    {
-      "epoch": 0.3079687290937936,
-      "grad_norm": 0.0025175693444907665,
-      "learning_rate": 0.00019999995342803498,
-      "loss": 46.0,
-      "step": 4028
-    },
-    {
-      "epoch": 0.30804518607718334,
-      "grad_norm": 0.0009231767617166042,
-      "learning_rate": 0.0001999999534048504,
-      "loss": 46.0,
-      "step": 4029
-    },
-    {
-      "epoch": 0.30812164306057305,
-      "grad_norm": 0.002071147784590721,
-      "learning_rate": 0.0001999999533816601,
-      "loss": 46.0,
-      "step": 4030
-    },
-    {
-      "epoch": 0.30819810004396275,
-      "grad_norm": 0.00136423425283283,
-      "learning_rate": 0.000199999953358464,
-      "loss": 46.0,
-      "step": 4031
-    },
-    {
-      "epoch": 0.30827455702735246,
-      "grad_norm": 0.001215483178384602,
-      "learning_rate": 0.00019999995333526215,
-      "loss": 46.0,
-      "step": 4032
-    },
-    {
-      "epoch": 0.3083510140107422,
-      "grad_norm": 0.005483425222337246,
-      "learning_rate": 0.00019999995331205453,
-      "loss": 46.0,
-      "step": 4033
-    },
-    {
-      "epoch": 0.3084274709941319,
-      "grad_norm": 0.002995036542415619,
-      "learning_rate": 0.00019999995328884114,
-      "loss": 46.0,
-      "step": 4034
-    },
-    {
-      "epoch": 0.30850392797752163,
-      "grad_norm": 0.001227505155839026,
-      "learning_rate": 0.00019999995326562198,
-      "loss": 46.0,
-      "step": 4035
-    },
-    {
-      "epoch": 0.3085803849609114,
-      "grad_norm": 0.0007473343284800649,
-      "learning_rate": 0.00019999995324239704,
-      "loss": 46.0,
-      "step": 4036
-    },
-    {
-      "epoch": 0.3086568419443011,
-      "grad_norm": 0.0005458722007460892,
-      "learning_rate": 0.00019999995321916633,
-      "loss": 46.0,
-      "step": 4037
-    },
-    {
-      "epoch": 0.3087332989276908,
-      "grad_norm": 0.0003971096593886614,
-      "learning_rate": 0.00019999995319592987,
-      "loss": 46.0,
-      "step": 4038
-    },
-    {
-      "epoch": 0.3088097559110805,
-      "grad_norm": 0.0005233470583334565,
-      "learning_rate": 0.00019999995317268764,
-      "loss": 46.0,
-      "step": 4039
-    },
-    {
-      "epoch": 0.30888621289447027,
-      "grad_norm": 0.0018375712679699063,
-      "learning_rate": 0.0001999999531494396,
-      "loss": 46.0,
-      "step": 4040
-    },
-    {
-      "epoch": 0.30896266987786,
-      "grad_norm": 0.0036186447832733393,
-      "learning_rate": 0.00019999995312618583,
-      "loss": 46.0,
-      "step": 4041
-    },
-    {
-      "epoch": 0.3090391268612497,
-      "grad_norm": 0.004774878732860088,
-      "learning_rate": 0.00019999995310292627,
-      "loss": 46.0,
-      "step": 4042
-    },
-    {
-      "epoch": 0.3091155838446394,
-      "grad_norm": 0.0038502702955156565,
-      "learning_rate": 0.00019999995307966095,
-      "loss": 46.0,
-      "step": 4043
-    },
-    {
-      "epoch": 0.30919204082802915,
-      "grad_norm": 0.0012251483276486397,
-      "learning_rate": 0.00019999995305638985,
-      "loss": 46.0,
-      "step": 4044
-    },
-    {
-      "epoch": 0.30926849781141885,
-      "grad_norm": 0.004108453635126352,
-      "learning_rate": 0.000199999953033113,
-      "loss": 46.0,
-      "step": 4045
-    },
-    {
-      "epoch": 0.30934495479480856,
-      "grad_norm": 0.000770887709222734,
-      "learning_rate": 0.0001999999530098304,
-      "loss": 46.0,
-      "step": 4046
-    },
-    {
-      "epoch": 0.3094214117781983,
-      "grad_norm": 0.0006265907431952655,
-      "learning_rate": 0.000199999952986542,
-      "loss": 46.0,
-      "step": 4047
-    },
-    {
-      "epoch": 0.309497868761588,
-      "grad_norm": 0.0009616278694011271,
-      "learning_rate": 0.00019999995296324783,
-      "loss": 46.0,
-      "step": 4048
-    },
-    {
-      "epoch": 0.30957432574497773,
-      "grad_norm": 0.0012052081292495131,
-      "learning_rate": 0.0001999999529399479,
-      "loss": 46.0,
-      "step": 4049
-    },
-    {
-      "epoch": 0.30965078272836744,
-      "grad_norm": 0.0014100506668910384,
-      "learning_rate": 0.00019999995291664218,
-      "loss": 46.0,
-      "step": 4050
-    },
-    {
-      "epoch": 0.3097272397117572,
-      "grad_norm": 0.0006808596663177013,
-      "learning_rate": 0.0001999999528933307,
-      "loss": 46.0,
-      "step": 4051
-    },
-    {
-      "epoch": 0.3098036966951469,
-      "grad_norm": 0.0007321045268326998,
-      "learning_rate": 0.00019999995287001346,
-      "loss": 46.0,
-      "step": 4052
-    },
-    {
-      "epoch": 0.3098801536785366,
-      "grad_norm": 0.0010772771202027798,
-      "learning_rate": 0.00019999995284669043,
-      "loss": 46.0,
-      "step": 4053
-    },
-    {
-      "epoch": 0.3099566106619263,
-      "grad_norm": 0.0014033811166882515,
-      "learning_rate": 0.00019999995282336163,
-      "loss": 46.0,
-      "step": 4054
-    },
-    {
-      "epoch": 0.3100330676453161,
-      "grad_norm": 0.0012709693983197212,
-      "learning_rate": 0.0001999999528000271,
-      "loss": 46.0,
-      "step": 4055
-    },
-    {
-      "epoch": 0.3101095246287058,
-      "grad_norm": 0.0014826421393081546,
-      "learning_rate": 0.00019999995277668678,
-      "loss": 46.0,
-      "step": 4056
-    },
-    {
-      "epoch": 0.3101859816120955,
-      "grad_norm": 0.002636758843436837,
-      "learning_rate": 0.00019999995275334068,
-      "loss": 46.0,
-      "step": 4057
-    },
-    {
-      "epoch": 0.3102624385954852,
-      "grad_norm": 0.0008221042808145285,
-      "learning_rate": 0.0001999999527299888,
-      "loss": 46.0,
-      "step": 4058
-    },
-    {
-      "epoch": 0.31033889557887495,
-      "grad_norm": 0.0010345177724957466,
-      "learning_rate": 0.0001999999527066312,
-      "loss": 46.0,
-      "step": 4059
-    },
-    {
-      "epoch": 0.31041535256226466,
-      "grad_norm": 0.0009395965025760233,
-      "learning_rate": 0.0001999999526832678,
-      "loss": 46.0,
-      "step": 4060
-    },
-    {
-      "epoch": 0.31049180954565436,
-      "grad_norm": 0.0035355750005692244,
-      "learning_rate": 0.0001999999526598986,
-      "loss": 46.0,
-      "step": 4061
-    },
-    {
-      "epoch": 0.3105682665290441,
-      "grad_norm": 0.0007544115069322288,
-      "learning_rate": 0.00019999995263652367,
-      "loss": 46.0,
-      "step": 4062
-    },
-    {
-      "epoch": 0.31064472351243383,
-      "grad_norm": 0.0017829195130616426,
-      "learning_rate": 0.00019999995261314296,
-      "loss": 46.0,
-      "step": 4063
-    },
-    {
-      "epoch": 0.31072118049582353,
-      "grad_norm": 0.0014873375184834003,
-      "learning_rate": 0.0001999999525897565,
-      "loss": 46.0,
-      "step": 4064
-    },
-    {
-      "epoch": 0.31079763747921324,
-      "grad_norm": 0.0007083939271979034,
-      "learning_rate": 0.00019999995256636424,
-      "loss": 46.0,
-      "step": 4065
-    },
-    {
-      "epoch": 0.310874094462603,
-      "grad_norm": 0.0007060458301566541,
-      "learning_rate": 0.0001999999525429662,
-      "loss": 46.0,
-      "step": 4066
-    },
-    {
-      "epoch": 0.3109505514459927,
-      "grad_norm": 0.001243519363924861,
-      "learning_rate": 0.00019999995251956243,
-      "loss": 46.0,
-      "step": 4067
-    },
-    {
-      "epoch": 0.3110270084293824,
-      "grad_norm": 0.0007302002632059157,
-      "learning_rate": 0.00019999995249615288,
-      "loss": 46.0,
-      "step": 4068
-    },
-    {
-      "epoch": 0.3111034654127721,
-      "grad_norm": 0.0011072546476498246,
-      "learning_rate": 0.00019999995247273756,
-      "loss": 46.0,
-      "step": 4069
-    },
-    {
-      "epoch": 0.3111799223961619,
-      "grad_norm": 0.0012059843866154552,
-      "learning_rate": 0.00019999995244931643,
-      "loss": 46.0,
-      "step": 4070
-    },
-    {
-      "epoch": 0.3112563793795516,
-      "grad_norm": 0.0009788620518520474,
-      "learning_rate": 0.0001999999524258896,
-      "loss": 46.0,
-      "step": 4071
-    },
-    {
-      "epoch": 0.3113328363629413,
-      "grad_norm": 0.0007899851771071553,
-      "learning_rate": 0.00019999995240245694,
-      "loss": 46.0,
-      "step": 4072
-    },
-    {
-      "epoch": 0.311409293346331,
-      "grad_norm": 0.0016155410557985306,
-      "learning_rate": 0.00019999995237901855,
-      "loss": 46.0,
-      "step": 4073
-    },
-    {
-      "epoch": 0.31148575032972076,
-      "grad_norm": 0.001013038563542068,
-      "learning_rate": 0.0001999999523555744,
-      "loss": 46.0,
-      "step": 4074
-    },
-    {
-      "epoch": 0.31156220731311046,
-      "grad_norm": 0.0014393455348908901,
-      "learning_rate": 0.00019999995233212445,
-      "loss": 46.0,
-      "step": 4075
-    },
-    {
-      "epoch": 0.31163866429650017,
-      "grad_norm": 0.0028291363269090652,
-      "learning_rate": 0.00019999995230866874,
-      "loss": 46.0,
-      "step": 4076
-    },
-    {
-      "epoch": 0.31171512127988993,
-      "grad_norm": 0.001994832418859005,
-      "learning_rate": 0.00019999995228520723,
-      "loss": 46.0,
-      "step": 4077
-    },
-    {
-      "epoch": 0.31179157826327963,
-      "grad_norm": 0.001404017093591392,
-      "learning_rate": 0.00019999995226174,
-      "loss": 46.0,
-      "step": 4078
-    },
-    {
-      "epoch": 0.31186803524666934,
-      "grad_norm": 0.005517485085874796,
-      "learning_rate": 0.00019999995223826696,
-      "loss": 46.0,
-      "step": 4079
-    },
-    {
-      "epoch": 0.31194449223005905,
-      "grad_norm": 0.0009098463342525065,
-      "learning_rate": 0.0001999999522147882,
-      "loss": 46.0,
-      "step": 4080
-    },
-    {
-      "epoch": 0.3120209492134488,
-      "grad_norm": 0.0011220687301829457,
-      "learning_rate": 0.00019999995219130364,
-      "loss": 46.0,
-      "step": 4081
-    },
-    {
-      "epoch": 0.3120974061968385,
-      "grad_norm": 0.0005901592667214572,
-      "learning_rate": 0.00019999995216781328,
-      "loss": 46.0,
-      "step": 4082
-    },
-    {
-      "epoch": 0.3121738631802282,
-      "grad_norm": 0.0008901542751118541,
-      "learning_rate": 0.0001999999521443172,
-      "loss": 46.0,
-      "step": 4083
-    },
-    {
-      "epoch": 0.3122503201636179,
-      "grad_norm": 0.003306616796180606,
-      "learning_rate": 0.00019999995212081532,
-      "loss": 46.0,
-      "step": 4084
-    },
-    {
-      "epoch": 0.3123267771470077,
-      "grad_norm": 0.0007005054503679276,
-      "learning_rate": 0.0001999999520973077,
-      "loss": 46.0,
-      "step": 4085
-    },
-    {
-      "epoch": 0.3124032341303974,
-      "grad_norm": 0.0021608273964375257,
-      "learning_rate": 0.00019999995207379428,
-      "loss": 46.0,
-      "step": 4086
-    },
-    {
-      "epoch": 0.3124796911137871,
-      "grad_norm": 0.0030763857066631317,
-      "learning_rate": 0.00019999995205027512,
-      "loss": 46.0,
-      "step": 4087
-    },
-    {
-      "epoch": 0.3125561480971768,
-      "grad_norm": 0.0011079679243266582,
-      "learning_rate": 0.00019999995202675015,
-      "loss": 46.0,
-      "step": 4088
-    },
-    {
-      "epoch": 0.31263260508056656,
-      "grad_norm": 0.0017841230146586895,
-      "learning_rate": 0.00019999995200321947,
-      "loss": 46.0,
-      "step": 4089
-    },
-    {
-      "epoch": 0.31270906206395627,
-      "grad_norm": 0.0027018177788704634,
-      "learning_rate": 0.00019999995197968296,
-      "loss": 46.0,
-      "step": 4090
-    },
-    {
-      "epoch": 0.312785519047346,
-      "grad_norm": 0.000514084124006331,
-      "learning_rate": 0.00019999995195614073,
-      "loss": 46.0,
-      "step": 4091
-    },
-    {
-      "epoch": 0.31286197603073573,
-      "grad_norm": 0.0005712288548238575,
-      "learning_rate": 0.0001999999519325927,
-      "loss": 46.0,
-      "step": 4092
-    },
-    {
-      "epoch": 0.31293843301412544,
-      "grad_norm": 0.0017212866805493832,
-      "learning_rate": 0.0001999999519090389,
-      "loss": 46.0,
-      "step": 4093
-    },
-    {
-      "epoch": 0.31301488999751514,
-      "grad_norm": 0.0006872282247059047,
-      "learning_rate": 0.00019999995188547934,
-      "loss": 46.0,
-      "step": 4094
-    },
-    {
-      "epoch": 0.31309134698090485,
-      "grad_norm": 0.001309835584834218,
-      "learning_rate": 0.000199999951861914,
-      "loss": 46.0,
-      "step": 4095
-    },
-    {
-      "epoch": 0.3131678039642946,
-      "grad_norm": 0.0021782179828733206,
-      "learning_rate": 0.0001999999518383429,
-      "loss": 46.0,
-      "step": 4096
-    },
-    {
-      "epoch": 0.3132442609476843,
-      "grad_norm": 0.000896404788363725,
-      "learning_rate": 0.00019999995181476602,
-      "loss": 46.0,
-      "step": 4097
-    },
-    {
-      "epoch": 0.313320717931074,
-      "grad_norm": 0.0019859285093843937,
-      "learning_rate": 0.00019999995179118338,
-      "loss": 46.0,
-      "step": 4098
-    },
-    {
-      "epoch": 0.3133971749144637,
-      "grad_norm": 0.0020705722272396088,
-      "learning_rate": 0.00019999995176759496,
-      "loss": 46.0,
-      "step": 4099
-    },
-    {
-      "epoch": 0.3134736318978535,
-      "grad_norm": 0.000561065913643688,
-      "learning_rate": 0.00019999995174400082,
-      "loss": 46.0,
-      "step": 4100
-    },
-    {
-      "epoch": 0.3135500888812432,
-      "grad_norm": 0.0021201521158218384,
-      "learning_rate": 0.00019999995172040086,
-      "loss": 46.0,
-      "step": 4101
-    },
-    {
-      "epoch": 0.3136265458646329,
-      "grad_norm": 0.008403378538787365,
-      "learning_rate": 0.00019999995169679512,
-      "loss": 46.0,
-      "step": 4102
-    },
-    {
-      "epoch": 0.3137030028480226,
-      "grad_norm": 0.0005731437704525888,
-      "learning_rate": 0.00019999995167318364,
-      "loss": 46.0,
-      "step": 4103
-    },
-    {
-      "epoch": 0.31377945983141237,
-      "grad_norm": 0.002495912602171302,
-      "learning_rate": 0.0001999999516495664,
-      "loss": 46.0,
-      "step": 4104
-    },
-    {
-      "epoch": 0.31385591681480207,
-      "grad_norm": 0.0014699643943458796,
-      "learning_rate": 0.00019999995162594335,
-      "loss": 46.0,
-      "step": 4105
-    },
-    {
-      "epoch": 0.3139323737981918,
-      "grad_norm": 0.0006395775708369911,
-      "learning_rate": 0.00019999995160231455,
-      "loss": 46.0,
-      "step": 4106
-    },
-    {
-      "epoch": 0.31400883078158154,
-      "grad_norm": 0.0007599610253237188,
-      "learning_rate": 0.00019999995157868,
-      "loss": 46.0,
-      "step": 4107
-    },
-    {
-      "epoch": 0.31408528776497124,
-      "grad_norm": 0.0029232690576463938,
-      "learning_rate": 0.00019999995155503965,
-      "loss": 46.0,
-      "step": 4108
-    },
-    {
-      "epoch": 0.31416174474836095,
-      "grad_norm": 0.0023315108846873045,
-      "learning_rate": 0.00019999995153139355,
-      "loss": 46.0,
-      "step": 4109
-    },
-    {
-      "epoch": 0.31423820173175065,
-      "grad_norm": 0.0013846426736563444,
-      "learning_rate": 0.00019999995150774168,
-      "loss": 46.0,
-      "step": 4110
-    },
-    {
-      "epoch": 0.3143146587151404,
-      "grad_norm": 0.0009937274735420942,
-      "learning_rate": 0.00019999995148408406,
-      "loss": 46.0,
-      "step": 4111
-    },
-    {
-      "epoch": 0.3143911156985301,
-      "grad_norm": 0.0048923855647444725,
-      "learning_rate": 0.00019999995146042064,
-      "loss": 46.0,
-      "step": 4112
-    },
-    {
-      "epoch": 0.3144675726819198,
-      "grad_norm": 0.0037990878336131573,
-      "learning_rate": 0.00019999995143675145,
-      "loss": 46.0,
-      "step": 4113
-    },
-    {
-      "epoch": 0.31454402966530953,
-      "grad_norm": 0.0026742154732346535,
-      "learning_rate": 0.0001999999514130765,
-      "loss": 46.0,
-      "step": 4114
-    },
-    {
-      "epoch": 0.3146204866486993,
-      "grad_norm": 0.0010428677778691053,
-      "learning_rate": 0.00019999995138939578,
-      "loss": 46.0,
-      "step": 4115
-    },
-    {
-      "epoch": 0.314696943632089,
-      "grad_norm": 0.0006835341919213533,
-      "learning_rate": 0.00019999995136570927,
-      "loss": 46.0,
-      "step": 4116
-    },
-    {
-      "epoch": 0.3147734006154787,
-      "grad_norm": 0.001079073059372604,
-      "learning_rate": 0.00019999995134201704,
-      "loss": 46.0,
-      "step": 4117
-    },
-    {
-      "epoch": 0.3148498575988684,
-      "grad_norm": 0.0013314823154360056,
-      "learning_rate": 0.000199999951318319,
-      "loss": 46.0,
-      "step": 4118
-    },
-    {
-      "epoch": 0.31492631458225817,
-      "grad_norm": 0.0027630191762000322,
-      "learning_rate": 0.0001999999512946152,
-      "loss": 46.0,
-      "step": 4119
-    },
-    {
-      "epoch": 0.3150027715656479,
-      "grad_norm": 0.0007205162546597421,
-      "learning_rate": 0.00019999995127090563,
-      "loss": 46.0,
-      "step": 4120
-    },
-    {
-      "epoch": 0.3150792285490376,
-      "grad_norm": 0.0016001613112166524,
-      "learning_rate": 0.0001999999512471903,
-      "loss": 46.0,
-      "step": 4121
-    },
-    {
-      "epoch": 0.31515568553242734,
-      "grad_norm": 0.0015134575078263879,
-      "learning_rate": 0.0001999999512234692,
-      "loss": 46.0,
-      "step": 4122
-    },
-    {
-      "epoch": 0.31523214251581705,
-      "grad_norm": 0.0014422638341784477,
-      "learning_rate": 0.0001999999511997423,
-      "loss": 46.0,
-      "step": 4123
-    },
-    {
-      "epoch": 0.31530859949920675,
-      "grad_norm": 0.0025649317540228367,
-      "learning_rate": 0.00019999995117600967,
-      "loss": 46.0,
-      "step": 4124
-    },
-    {
-      "epoch": 0.31538505648259646,
-      "grad_norm": 0.001873421249911189,
-      "learning_rate": 0.00019999995115227125,
-      "loss": 46.0,
-      "step": 4125
-    },
-    {
-      "epoch": 0.3154615134659862,
-      "grad_norm": 0.0014939858810976148,
-      "learning_rate": 0.00019999995112852706,
-      "loss": 46.0,
-      "step": 4126
-    },
-    {
-      "epoch": 0.3155379704493759,
-      "grad_norm": 0.0008123088045977056,
-      "learning_rate": 0.00019999995110477712,
-      "loss": 46.0,
-      "step": 4127
-    },
-    {
-      "epoch": 0.31561442743276563,
-      "grad_norm": 0.0013626243453472853,
-      "learning_rate": 0.0001999999510810214,
-      "loss": 46.0,
-      "step": 4128
-    },
-    {
-      "epoch": 0.31569088441615534,
-      "grad_norm": 0.0014732293784618378,
-      "learning_rate": 0.00019999995105725988,
-      "loss": 46.0,
-      "step": 4129
-    },
-    {
-      "epoch": 0.3157673413995451,
-      "grad_norm": 0.0017963205464184284,
-      "learning_rate": 0.00019999995103349262,
-      "loss": 46.0,
-      "step": 4130
-    },
-    {
-      "epoch": 0.3158437983829348,
-      "grad_norm": 0.0015620823251083493,
-      "learning_rate": 0.0001999999510097196,
-      "loss": 46.0,
-      "step": 4131
-    },
-    {
-      "epoch": 0.3159202553663245,
-      "grad_norm": 0.0007996860658749938,
-      "learning_rate": 0.0001999999509859408,
-      "loss": 46.0,
-      "step": 4132
-    },
-    {
-      "epoch": 0.3159967123497142,
-      "grad_norm": 0.01104381587356329,
-      "learning_rate": 0.00019999995096215624,
-      "loss": 46.0,
-      "step": 4133
-    },
-    {
-      "epoch": 0.316073169333104,
-      "grad_norm": 0.0022612884640693665,
-      "learning_rate": 0.0001999999509383659,
-      "loss": 46.0,
-      "step": 4134
-    },
-    {
-      "epoch": 0.3161496263164937,
-      "grad_norm": 0.0031405312474817038,
-      "learning_rate": 0.0001999999509145698,
-      "loss": 46.0,
-      "step": 4135
-    },
-    {
-      "epoch": 0.3162260832998834,
-      "grad_norm": 0.000809003715403378,
-      "learning_rate": 0.0001999999508907679,
-      "loss": 46.0,
-      "step": 4136
-    },
-    {
-      "epoch": 0.31630254028327315,
-      "grad_norm": 0.0009952536784112453,
-      "learning_rate": 0.00019999995086696025,
-      "loss": 46.0,
-      "step": 4137
-    },
-    {
-      "epoch": 0.31637899726666285,
-      "grad_norm": 0.007186158560216427,
-      "learning_rate": 0.00019999995084314683,
-      "loss": 46.0,
-      "step": 4138
-    },
-    {
-      "epoch": 0.31645545425005256,
-      "grad_norm": 0.0030693570151925087,
-      "learning_rate": 0.00019999995081932764,
-      "loss": 46.0,
-      "step": 4139
-    },
-    {
-      "epoch": 0.31653191123344226,
-      "grad_norm": 0.0008522903081029654,
-      "learning_rate": 0.00019999995079550268,
-      "loss": 46.0,
-      "step": 4140
-    },
-    {
-      "epoch": 0.316608368216832,
-      "grad_norm": 0.0013415059074759483,
-      "learning_rate": 0.00019999995077167197,
-      "loss": 46.0,
-      "step": 4141
-    },
-    {
-      "epoch": 0.31668482520022173,
-      "grad_norm": 0.0007622973062098026,
-      "learning_rate": 0.00019999995074783546,
-      "loss": 46.0,
-      "step": 4142
-    },
-    {
-      "epoch": 0.31676128218361144,
-      "grad_norm": 0.005981494206935167,
-      "learning_rate": 0.0001999999507239932,
-      "loss": 46.0,
-      "step": 4143
-    },
-    {
-      "epoch": 0.31683773916700114,
-      "grad_norm": 0.013176447711884975,
-      "learning_rate": 0.00019999995070014518,
-      "loss": 46.0,
-      "step": 4144
-    },
-    {
-      "epoch": 0.3169141961503909,
-      "grad_norm": 0.0005784628447145224,
-      "learning_rate": 0.00019999995067629137,
-      "loss": 46.0,
-      "step": 4145
-    },
-    {
-      "epoch": 0.3169906531337806,
-      "grad_norm": 0.0030265958048403263,
-      "learning_rate": 0.0001999999506524318,
-      "loss": 46.0,
-      "step": 4146
-    },
-    {
-      "epoch": 0.3170671101171703,
-      "grad_norm": 0.0008855731575749815,
-      "learning_rate": 0.00019999995062856645,
-      "loss": 46.0,
-      "step": 4147
-    },
-    {
-      "epoch": 0.31714356710056,
-      "grad_norm": 0.0009584020590409636,
-      "learning_rate": 0.00019999995060469532,
-      "loss": 46.0,
-      "step": 4148
-    },
-    {
-      "epoch": 0.3172200240839498,
-      "grad_norm": 0.000916891498491168,
-      "learning_rate": 0.00019999995058081845,
-      "loss": 46.0,
-      "step": 4149
-    },
-    {
-      "epoch": 0.3172964810673395,
-      "grad_norm": 0.0007251052302308381,
-      "learning_rate": 0.00019999995055693578,
-      "loss": 46.0,
-      "step": 4150
-    },
-    {
-      "epoch": 0.3173729380507292,
-      "grad_norm": 0.0009630761342123151,
-      "learning_rate": 0.0001999999505330474,
-      "loss": 46.0,
-      "step": 4151
-    },
-    {
-      "epoch": 0.31744939503411895,
-      "grad_norm": 0.0010503242956474423,
-      "learning_rate": 0.00019999995050915318,
-      "loss": 46.0,
-      "step": 4152
-    },
-    {
-      "epoch": 0.31752585201750866,
-      "grad_norm": 0.001143371919170022,
-      "learning_rate": 0.00019999995048525321,
-      "loss": 46.0,
-      "step": 4153
-    },
-    {
-      "epoch": 0.31760230900089836,
-      "grad_norm": 0.0011370426509529352,
-      "learning_rate": 0.0001999999504613475,
-      "loss": 46.0,
-      "step": 4154
-    },
-    {
-      "epoch": 0.31767876598428807,
-      "grad_norm": 0.00044460356002673507,
-      "learning_rate": 0.000199999950437436,
-      "loss": 46.0,
-      "step": 4155
-    },
-    {
-      "epoch": 0.31775522296767783,
-      "grad_norm": 0.0006213995511643589,
-      "learning_rate": 0.00019999995041351874,
-      "loss": 46.0,
-      "step": 4156
-    },
-    {
-      "epoch": 0.31783167995106754,
-      "grad_norm": 0.0013278733240440488,
-      "learning_rate": 0.00019999995038959568,
-      "loss": 46.0,
-      "step": 4157
-    },
-    {
-      "epoch": 0.31790813693445724,
-      "grad_norm": 0.0019477899186313152,
-      "learning_rate": 0.00019999995036566688,
-      "loss": 46.0,
-      "step": 4158
-    },
-    {
-      "epoch": 0.31798459391784695,
-      "grad_norm": 0.0030430464539676905,
-      "learning_rate": 0.0001999999503417323,
-      "loss": 46.0,
-      "step": 4159
-    },
-    {
-      "epoch": 0.3180610509012367,
-      "grad_norm": 0.0008460903191007674,
-      "learning_rate": 0.00019999995031779198,
-      "loss": 46.0,
-      "step": 4160
-    },
-    {
-      "epoch": 0.3181375078846264,
-      "grad_norm": 0.0034120797645300627,
-      "learning_rate": 0.00019999995029384583,
-      "loss": 46.0,
-      "step": 4161
-    },
-    {
-      "epoch": 0.3182139648680161,
-      "grad_norm": 0.0005994453676976264,
-      "learning_rate": 0.00019999995026989396,
-      "loss": 46.0,
-      "step": 4162
-    },
-    {
-      "epoch": 0.3182904218514059,
-      "grad_norm": 0.00183070357888937,
-      "learning_rate": 0.0001999999502459363,
-      "loss": 46.0,
-      "step": 4163
-    },
-    {
-      "epoch": 0.3183668788347956,
-      "grad_norm": 0.0013423011405393481,
-      "learning_rate": 0.00019999995022197288,
-      "loss": 46.0,
-      "step": 4164
-    },
-    {
-      "epoch": 0.3184433358181853,
-      "grad_norm": 0.001122456043958664,
-      "learning_rate": 0.0001999999501980037,
-      "loss": 46.0,
-      "step": 4165
-    },
-    {
-      "epoch": 0.318519792801575,
-      "grad_norm": 0.003014274640008807,
-      "learning_rate": 0.00019999995017402873,
-      "loss": 46.0,
-      "step": 4166
-    },
-    {
-      "epoch": 0.31859624978496476,
-      "grad_norm": 0.0008784834644757211,
-      "learning_rate": 0.000199999950150048,
-      "loss": 46.0,
-      "step": 4167
-    },
-    {
-      "epoch": 0.31867270676835446,
-      "grad_norm": 0.008919299580156803,
-      "learning_rate": 0.00019999995012606148,
-      "loss": 46.0,
-      "step": 4168
-    },
-    {
-      "epoch": 0.31874916375174417,
-      "grad_norm": 0.0008249454549513757,
-      "learning_rate": 0.00019999995010206922,
-      "loss": 46.0,
-      "step": 4169
-    },
-    {
-      "epoch": 0.3188256207351339,
-      "grad_norm": 0.000595959194470197,
-      "learning_rate": 0.00019999995007807117,
-      "loss": 46.0,
-      "step": 4170
-    },
-    {
-      "epoch": 0.31890207771852364,
-      "grad_norm": 0.0012690945295616984,
-      "learning_rate": 0.00019999995005406737,
-      "loss": 46.0,
-      "step": 4171
-    },
-    {
-      "epoch": 0.31897853470191334,
-      "grad_norm": 0.005566779058426619,
-      "learning_rate": 0.0001999999500300578,
-      "loss": 46.0,
-      "step": 4172
-    },
-    {
-      "epoch": 0.31905499168530305,
-      "grad_norm": 0.0008208694634959102,
-      "learning_rate": 0.00019999995000604244,
-      "loss": 46.0,
-      "step": 4173
-    },
-    {
-      "epoch": 0.31913144866869275,
-      "grad_norm": 0.003568213665857911,
-      "learning_rate": 0.00019999994998202132,
-      "loss": 46.0,
-      "step": 4174
-    },
-    {
-      "epoch": 0.3192079056520825,
-      "grad_norm": 0.0006343183922581375,
-      "learning_rate": 0.00019999994995799445,
-      "loss": 46.0,
-      "step": 4175
-    },
-    {
-      "epoch": 0.3192843626354722,
-      "grad_norm": 0.0040811472572386265,
-      "learning_rate": 0.00019999994993396175,
-      "loss": 46.0,
-      "step": 4176
-    },
-    {
-      "epoch": 0.3193608196188619,
-      "grad_norm": 0.002137728501111269,
-      "learning_rate": 0.00019999994990992334,
-      "loss": 46.0,
-      "step": 4177
-    },
-    {
-      "epoch": 0.3194372766022517,
-      "grad_norm": 0.0026833750307559967,
-      "learning_rate": 0.00019999994988587915,
-      "loss": 46.0,
-      "step": 4178
-    },
-    {
-      "epoch": 0.3195137335856414,
-      "grad_norm": 0.0005680866306647658,
-      "learning_rate": 0.00019999994986182919,
-      "loss": 46.0,
-      "step": 4179
-    },
-    {
-      "epoch": 0.3195901905690311,
-      "grad_norm": 0.0033166774082928896,
-      "learning_rate": 0.00019999994983777345,
-      "loss": 46.0,
-      "step": 4180
-    },
-    {
-      "epoch": 0.3196666475524208,
-      "grad_norm": 0.0029382030479609966,
-      "learning_rate": 0.00019999994981371194,
-      "loss": 46.0,
-      "step": 4181
-    },
-    {
-      "epoch": 0.31974310453581056,
-      "grad_norm": 0.0033811938483268023,
-      "learning_rate": 0.00019999994978964466,
-      "loss": 46.0,
-      "step": 4182
-    },
-    {
-      "epoch": 0.31981956151920027,
-      "grad_norm": 0.0008562508155591786,
-      "learning_rate": 0.00019999994976557163,
-      "loss": 46.0,
-      "step": 4183
-    },
-    {
-      "epoch": 0.31989601850259,
-      "grad_norm": 0.0015214183367788792,
-      "learning_rate": 0.00019999994974149283,
-      "loss": 46.0,
-      "step": 4184
-    },
-    {
-      "epoch": 0.3199724754859797,
-      "grad_norm": 0.0022460981272161007,
-      "learning_rate": 0.00019999994971740823,
-      "loss": 46.0,
-      "step": 4185
-    },
-    {
-      "epoch": 0.32004893246936944,
-      "grad_norm": 0.0012656330363824964,
-      "learning_rate": 0.0001999999496933179,
-      "loss": 46.0,
-      "step": 4186
-    },
-    {
-      "epoch": 0.32012538945275915,
-      "grad_norm": 0.0010185488499701023,
-      "learning_rate": 0.00019999994966922175,
-      "loss": 46.0,
-      "step": 4187
-    },
-    {
-      "epoch": 0.32020184643614885,
-      "grad_norm": 0.0041980748064816,
-      "learning_rate": 0.00019999994964511989,
-      "loss": 46.0,
-      "step": 4188
-    },
-    {
-      "epoch": 0.32027830341953856,
-      "grad_norm": 0.001467040623538196,
-      "learning_rate": 0.00019999994962101222,
-      "loss": 46.0,
-      "step": 4189
-    },
-    {
-      "epoch": 0.3203547604029283,
-      "grad_norm": 0.0009076852584257722,
-      "learning_rate": 0.00019999994959689877,
-      "loss": 46.0,
-      "step": 4190
-    },
-    {
-      "epoch": 0.320431217386318,
-      "grad_norm": 0.0034510979894548655,
-      "learning_rate": 0.00019999994957277959,
-      "loss": 46.0,
-      "step": 4191
-    },
-    {
-      "epoch": 0.32050767436970773,
-      "grad_norm": 0.0005649134982377291,
-      "learning_rate": 0.00019999994954865462,
-      "loss": 46.0,
-      "step": 4192
-    },
-    {
-      "epoch": 0.3205841313530975,
-      "grad_norm": 0.0012544525088742375,
-      "learning_rate": 0.0001999999495245239,
-      "loss": 46.0,
-      "step": 4193
-    },
-    {
-      "epoch": 0.3206605883364872,
-      "grad_norm": 0.0010324690956622362,
-      "learning_rate": 0.00019999994950038738,
-      "loss": 46.0,
-      "step": 4194
-    },
-    {
-      "epoch": 0.3207370453198769,
-      "grad_norm": 0.0019627446308732033,
-      "learning_rate": 0.0001999999494762451,
-      "loss": 46.0,
-      "step": 4195
-    },
-    {
-      "epoch": 0.3208135023032666,
-      "grad_norm": 0.0030409370083361864,
-      "learning_rate": 0.00019999994945209704,
-      "loss": 46.0,
-      "step": 4196
-    },
-    {
-      "epoch": 0.32088995928665637,
-      "grad_norm": 0.0008453971240669489,
-      "learning_rate": 0.00019999994942794324,
-      "loss": 46.0,
-      "step": 4197
-    },
-    {
-      "epoch": 0.3209664162700461,
-      "grad_norm": 0.0005371761508285999,
-      "learning_rate": 0.00019999994940378364,
-      "loss": 46.0,
-      "step": 4198
-    },
-    {
-      "epoch": 0.3210428732534358,
-      "grad_norm": 0.013904278166592121,
-      "learning_rate": 0.0001999999493796183,
-      "loss": 46.0,
-      "step": 4199
-    },
-    {
-      "epoch": 0.3211193302368255,
-      "grad_norm": 0.002591705182567239,
-      "learning_rate": 0.00019999994935544717,
-      "loss": 46.0,
-      "step": 4200
-    },
-    {
-      "epoch": 0.32119578722021525,
-      "grad_norm": 0.0017301174812018871,
-      "learning_rate": 0.00019999994933127027,
-      "loss": 46.0,
-      "step": 4201
-    },
-    {
-      "epoch": 0.32127224420360495,
-      "grad_norm": 0.0035617249086499214,
-      "learning_rate": 0.0001999999493070876,
-      "loss": 46.0,
-      "step": 4202
-    },
-    {
-      "epoch": 0.32134870118699466,
-      "grad_norm": 0.01069709099829197,
-      "learning_rate": 0.0001999999492828992,
-      "loss": 46.0,
-      "step": 4203
-    },
-    {
-      "epoch": 0.32142515817038436,
-      "grad_norm": 0.0016002259217202663,
-      "learning_rate": 0.00019999994925870497,
-      "loss": 46.0,
-      "step": 4204
-    },
-    {
-      "epoch": 0.3215016151537741,
-      "grad_norm": 0.004124329425394535,
-      "learning_rate": 0.000199999949234505,
-      "loss": 46.0,
-      "step": 4205
-    },
-    {
-      "epoch": 0.32157807213716383,
-      "grad_norm": 0.0010847809026017785,
-      "learning_rate": 0.00019999994921029925,
-      "loss": 46.0,
-      "step": 4206
-    },
-    {
-      "epoch": 0.32165452912055353,
-      "grad_norm": 0.0018952528480440378,
-      "learning_rate": 0.00019999994918608774,
-      "loss": 46.0,
-      "step": 4207
-    },
-    {
-      "epoch": 0.3217309861039433,
-      "grad_norm": 0.0008499515242874622,
-      "learning_rate": 0.00019999994916187049,
-      "loss": 46.0,
-      "step": 4208
-    },
-    {
-      "epoch": 0.321807443087333,
-      "grad_norm": 0.001965031260624528,
-      "learning_rate": 0.0001999999491376474,
-      "loss": 46.0,
-      "step": 4209
-    },
-    {
-      "epoch": 0.3218839000707227,
-      "grad_norm": 0.0012979368912056088,
-      "learning_rate": 0.00019999994911341863,
-      "loss": 46.0,
-      "step": 4210
-    },
-    {
-      "epoch": 0.3219603570541124,
-      "grad_norm": 0.0005783950909972191,
-      "learning_rate": 0.000199999949089184,
-      "loss": 46.0,
-      "step": 4211
-    },
-    {
-      "epoch": 0.3220368140375022,
-      "grad_norm": 0.0014094688231125474,
-      "learning_rate": 0.00019999994906494365,
-      "loss": 46.0,
-      "step": 4212
-    },
-    {
-      "epoch": 0.3221132710208919,
-      "grad_norm": 0.0004982354585081339,
-      "learning_rate": 0.00019999994904069753,
-      "loss": 46.0,
-      "step": 4213
-    },
-    {
-      "epoch": 0.3221897280042816,
-      "grad_norm": 0.0024239635095000267,
-      "learning_rate": 0.00019999994901644564,
-      "loss": 46.0,
-      "step": 4214
-    },
-    {
-      "epoch": 0.3222661849876713,
-      "grad_norm": 0.0007969841826707125,
-      "learning_rate": 0.00019999994899218797,
-      "loss": 46.0,
-      "step": 4215
-    },
-    {
-      "epoch": 0.32234264197106105,
-      "grad_norm": 0.001183964079245925,
-      "learning_rate": 0.00019999994896792453,
-      "loss": 46.0,
-      "step": 4216
-    },
-    {
-      "epoch": 0.32241909895445076,
-      "grad_norm": 0.008951770141720772,
-      "learning_rate": 0.00019999994894365534,
-      "loss": 46.0,
-      "step": 4217
-    },
-    {
-      "epoch": 0.32249555593784046,
-      "grad_norm": 0.000656113785225898,
-      "learning_rate": 0.00019999994891938038,
-      "loss": 46.0,
-      "step": 4218
-    },
-    {
-      "epoch": 0.32257201292123017,
-      "grad_norm": 0.001815742114558816,
-      "learning_rate": 0.00019999994889509961,
-      "loss": 46.0,
-      "step": 4219
-    },
-    {
-      "epoch": 0.32264846990461993,
-      "grad_norm": 0.0017433996545150876,
-      "learning_rate": 0.0001999999488708131,
-      "loss": 46.0,
-      "step": 4220
-    },
-    {
-      "epoch": 0.32272492688800963,
-      "grad_norm": 0.0014687128132209182,
-      "learning_rate": 0.00019999994884652082,
-      "loss": 46.0,
-      "step": 4221
-    },
-    {
-      "epoch": 0.32280138387139934,
-      "grad_norm": 0.003201437648385763,
-      "learning_rate": 0.00019999994882222274,
-      "loss": 46.0,
-      "step": 4222
-    },
-    {
-      "epoch": 0.3228778408547891,
-      "grad_norm": 0.000897066667675972,
-      "learning_rate": 0.00019999994879791894,
-      "loss": 46.0,
-      "step": 4223
-    },
-    {
-      "epoch": 0.3229542978381788,
-      "grad_norm": 0.001856208429671824,
-      "learning_rate": 0.00019999994877360934,
-      "loss": 46.0,
-      "step": 4224
-    },
-    {
-      "epoch": 0.3230307548215685,
-      "grad_norm": 0.0012774377828463912,
-      "learning_rate": 0.00019999994874929397,
-      "loss": 46.0,
-      "step": 4225
-    },
-    {
-      "epoch": 0.3231072118049582,
-      "grad_norm": 0.0008346923277713358,
-      "learning_rate": 0.00019999994872497284,
-      "loss": 46.0,
-      "step": 4226
-    },
-    {
-      "epoch": 0.323183668788348,
-      "grad_norm": 0.0016784276813268661,
-      "learning_rate": 0.00019999994870064595,
-      "loss": 46.0,
-      "step": 4227
-    },
-    {
-      "epoch": 0.3232601257717377,
-      "grad_norm": 0.0009191537392325699,
-      "learning_rate": 0.00019999994867631328,
-      "loss": 46.0,
-      "step": 4228
-    },
-    {
-      "epoch": 0.3233365827551274,
-      "grad_norm": 0.0023871574085205793,
-      "learning_rate": 0.00019999994865197484,
-      "loss": 46.0,
-      "step": 4229
-    },
-    {
-      "epoch": 0.3234130397385171,
-      "grad_norm": 0.001587114529684186,
-      "learning_rate": 0.00019999994862763065,
-      "loss": 46.0,
-      "step": 4230
-    },
-    {
-      "epoch": 0.32348949672190686,
-      "grad_norm": 0.00043832772644236684,
-      "learning_rate": 0.0001999999486032807,
-      "loss": 46.0,
-      "step": 4231
-    },
-    {
-      "epoch": 0.32356595370529656,
-      "grad_norm": 0.0011136778630316257,
-      "learning_rate": 0.00019999994857892493,
-      "loss": 46.0,
-      "step": 4232
-    },
-    {
-      "epoch": 0.32364241068868627,
-      "grad_norm": 0.0014798268675804138,
-      "learning_rate": 0.00019999994855456342,
-      "loss": 46.0,
-      "step": 4233
-    },
-    {
-      "epoch": 0.32371886767207597,
-      "grad_norm": 0.0012873532250523567,
-      "learning_rate": 0.0001999999485301961,
-      "loss": 46.0,
-      "step": 4234
-    },
-    {
-      "epoch": 0.32379532465546573,
-      "grad_norm": 0.003759076353162527,
-      "learning_rate": 0.00019999994850582306,
-      "loss": 46.0,
-      "step": 4235
-    },
-    {
-      "epoch": 0.32387178163885544,
-      "grad_norm": 0.0006014717509970069,
-      "learning_rate": 0.00019999994848144423,
-      "loss": 46.0,
-      "step": 4236
-    },
-    {
-      "epoch": 0.32394823862224514,
-      "grad_norm": 0.0006482438766397536,
-      "learning_rate": 0.00019999994845705966,
-      "loss": 46.0,
-      "step": 4237
-    },
-    {
-      "epoch": 0.3240246956056349,
-      "grad_norm": 0.0010961649240925908,
-      "learning_rate": 0.00019999994843266928,
-      "loss": 46.0,
-      "step": 4238
-    },
-    {
-      "epoch": 0.3241011525890246,
-      "grad_norm": 0.001960871508345008,
-      "learning_rate": 0.00019999994840827316,
-      "loss": 46.0,
-      "step": 4239
-    },
-    {
-      "epoch": 0.3241776095724143,
-      "grad_norm": 0.0005564126768149436,
-      "learning_rate": 0.00019999994838387127,
-      "loss": 46.0,
-      "step": 4240
-    },
-    {
-      "epoch": 0.324254066555804,
-      "grad_norm": 0.00242434861138463,
-      "learning_rate": 0.00019999994835946357,
-      "loss": 46.0,
-      "step": 4241
-    },
-    {
-      "epoch": 0.3243305235391938,
-      "grad_norm": 0.0006800544797442853,
-      "learning_rate": 0.00019999994833505013,
-      "loss": 46.0,
-      "step": 4242
-    },
-    {
-      "epoch": 0.3244069805225835,
-      "grad_norm": 0.00216276990249753,
-      "learning_rate": 0.00019999994831063092,
-      "loss": 46.0,
-      "step": 4243
-    },
-    {
-      "epoch": 0.3244834375059732,
-      "grad_norm": 0.0018655310850590467,
-      "learning_rate": 0.00019999994828620596,
-      "loss": 46.0,
-      "step": 4244
-    },
-    {
-      "epoch": 0.3245598944893629,
-      "grad_norm": 0.0009296726784668863,
-      "learning_rate": 0.0001999999482617752,
-      "loss": 46.0,
-      "step": 4245
-    },
-    {
-      "epoch": 0.32463635147275266,
-      "grad_norm": 0.00227538519538939,
-      "learning_rate": 0.0001999999482373387,
-      "loss": 46.0,
-      "step": 4246
-    },
-    {
-      "epoch": 0.32471280845614237,
-      "grad_norm": 0.001492717768996954,
-      "learning_rate": 0.00019999994821289638,
-      "loss": 46.0,
-      "step": 4247
-    },
-    {
-      "epoch": 0.32478926543953207,
-      "grad_norm": 0.0005910330219194293,
-      "learning_rate": 0.00019999994818844835,
-      "loss": 46.0,
-      "step": 4248
-    },
-    {
-      "epoch": 0.3248657224229218,
-      "grad_norm": 0.0014322446659207344,
-      "learning_rate": 0.00019999994816399452,
-      "loss": 46.0,
-      "step": 4249
-    },
-    {
-      "epoch": 0.32494217940631154,
-      "grad_norm": 0.0016522066434845328,
-      "learning_rate": 0.00019999994813953492,
-      "loss": 46.0,
-      "step": 4250
-    },
-    {
-      "epoch": 0.32501863638970124,
-      "grad_norm": 0.0006958871963433921,
-      "learning_rate": 0.00019999994811506955,
-      "loss": 46.0,
-      "step": 4251
-    },
-    {
-      "epoch": 0.32509509337309095,
-      "grad_norm": 0.0007227236637845635,
-      "learning_rate": 0.0001999999480905984,
-      "loss": 46.0,
-      "step": 4252
-    },
-    {
-      "epoch": 0.3251715503564807,
-      "grad_norm": 0.001814017305150628,
-      "learning_rate": 0.0001999999480661215,
-      "loss": 46.0,
-      "step": 4253
-    },
-    {
-      "epoch": 0.3252480073398704,
-      "grad_norm": 0.0008652217220515013,
-      "learning_rate": 0.00019999994804163884,
-      "loss": 46.0,
-      "step": 4254
-    },
-    {
-      "epoch": 0.3253244643232601,
-      "grad_norm": 0.000539875531103462,
-      "learning_rate": 0.00019999994801715037,
-      "loss": 46.0,
-      "step": 4255
-    },
-    {
-      "epoch": 0.3254009213066498,
-      "grad_norm": 0.0008699905010871589,
-      "learning_rate": 0.00019999994799265616,
-      "loss": 46.0,
-      "step": 4256
-    },
-    {
-      "epoch": 0.3254773782900396,
-      "grad_norm": 0.00214879447594285,
-      "learning_rate": 0.0001999999479681562,
-      "loss": 46.0,
-      "step": 4257
-    },
-    {
-      "epoch": 0.3255538352734293,
-      "grad_norm": 0.002308016875758767,
-      "learning_rate": 0.00019999994794365044,
-      "loss": 46.0,
-      "step": 4258
-    },
-    {
-      "epoch": 0.325630292256819,
-      "grad_norm": 0.0005390993901528418,
-      "learning_rate": 0.00019999994791913893,
-      "loss": 46.0,
-      "step": 4259
-    },
-    {
-      "epoch": 0.3257067492402087,
-      "grad_norm": 0.0006740989629179239,
-      "learning_rate": 0.00019999994789462162,
-      "loss": 46.0,
-      "step": 4260
-    },
-    {
-      "epoch": 0.32578320622359846,
-      "grad_norm": 0.0008012195467017591,
-      "learning_rate": 0.00019999994787009854,
-      "loss": 46.0,
-      "step": 4261
-    },
-    {
-      "epoch": 0.32585966320698817,
-      "grad_norm": 0.0010206456063315272,
-      "learning_rate": 0.0001999999478455697,
-      "loss": 46.0,
-      "step": 4262
-    },
-    {
-      "epoch": 0.3259361201903779,
-      "grad_norm": 0.0007958015194162726,
-      "learning_rate": 0.00019999994782103514,
-      "loss": 46.0,
-      "step": 4263
-    },
-    {
-      "epoch": 0.3260125771737676,
-      "grad_norm": 0.003606856567785144,
-      "learning_rate": 0.00019999994779649476,
-      "loss": 46.0,
-      "step": 4264
-    },
-    {
-      "epoch": 0.32608903415715734,
-      "grad_norm": 0.0033516171388328075,
-      "learning_rate": 0.00019999994777194862,
-      "loss": 46.0,
-      "step": 4265
-    },
-    {
-      "epoch": 0.32616549114054705,
-      "grad_norm": 0.0005686180666089058,
-      "learning_rate": 0.00019999994774739672,
-      "loss": 46.0,
-      "step": 4266
-    },
-    {
-      "epoch": 0.32624194812393675,
-      "grad_norm": 0.0008491341141052544,
-      "learning_rate": 0.00019999994772283906,
-      "loss": 46.0,
-      "step": 4267
-    },
-    {
-      "epoch": 0.3263184051073265,
-      "grad_norm": 0.0008566360920667648,
-      "learning_rate": 0.0001999999476982756,
-      "loss": 46.0,
-      "step": 4268
-    },
-    {
-      "epoch": 0.3263948620907162,
-      "grad_norm": 0.0021916660480201244,
-      "learning_rate": 0.00019999994767370638,
-      "loss": 46.0,
-      "step": 4269
-    },
-    {
-      "epoch": 0.3264713190741059,
-      "grad_norm": 0.0007130063604563475,
-      "learning_rate": 0.0001999999476491314,
-      "loss": 46.0,
-      "step": 4270
-    },
-    {
-      "epoch": 0.32654777605749563,
-      "grad_norm": 0.0007190363830886781,
-      "learning_rate": 0.00019999994762455063,
-      "loss": 46.0,
-      "step": 4271
-    },
-    {
-      "epoch": 0.3266242330408854,
-      "grad_norm": 0.0003927747602574527,
-      "learning_rate": 0.0001999999475999641,
-      "loss": 46.0,
-      "step": 4272
-    },
-    {
-      "epoch": 0.3267006900242751,
-      "grad_norm": 0.0013259283732622862,
-      "learning_rate": 0.0001999999475753718,
-      "loss": 46.0,
-      "step": 4273
-    },
-    {
-      "epoch": 0.3267771470076648,
-      "grad_norm": 0.00265662744641304,
-      "learning_rate": 0.00019999994755077376,
-      "loss": 46.0,
-      "step": 4274
-    },
-    {
-      "epoch": 0.3268536039910545,
-      "grad_norm": 0.0015432748477905989,
-      "learning_rate": 0.00019999994752616993,
-      "loss": 46.0,
-      "step": 4275
-    },
-    {
-      "epoch": 0.32693006097444427,
-      "grad_norm": 0.001415196806192398,
-      "learning_rate": 0.00019999994750156033,
-      "loss": 46.0,
-      "step": 4276
-    },
-    {
-      "epoch": 0.327006517957834,
-      "grad_norm": 0.0011422167299315333,
-      "learning_rate": 0.00019999994747694496,
-      "loss": 46.0,
-      "step": 4277
-    },
-    {
-      "epoch": 0.3270829749412237,
-      "grad_norm": 0.0008736549643799663,
-      "learning_rate": 0.0001999999474523238,
-      "loss": 46.0,
-      "step": 4278
-    },
-    {
-      "epoch": 0.32715943192461344,
-      "grad_norm": 0.0008260179311037064,
-      "learning_rate": 0.0001999999474276969,
-      "loss": 46.0,
-      "step": 4279
-    },
-    {
-      "epoch": 0.32723588890800315,
-      "grad_norm": 0.0013766777701675892,
-      "learning_rate": 0.00019999994740306422,
-      "loss": 46.0,
-      "step": 4280
-    },
-    {
-      "epoch": 0.32731234589139285,
-      "grad_norm": 0.0009538207086734474,
-      "learning_rate": 0.00019999994737842576,
-      "loss": 46.0,
-      "step": 4281
-    },
-    {
-      "epoch": 0.32738880287478256,
-      "grad_norm": 0.001052172388881445,
-      "learning_rate": 0.00019999994735378157,
-      "loss": 46.0,
-      "step": 4282
-    },
-    {
-      "epoch": 0.3274652598581723,
-      "grad_norm": 0.0014353011501953006,
-      "learning_rate": 0.00019999994732913156,
-      "loss": 46.0,
-      "step": 4283
-    },
-    {
-      "epoch": 0.327541716841562,
-      "grad_norm": 0.001284641562961042,
-      "learning_rate": 0.0001999999473044758,
-      "loss": 46.0,
-      "step": 4284
-    },
-    {
-      "epoch": 0.32761817382495173,
-      "grad_norm": 0.0018336954526603222,
-      "learning_rate": 0.0001999999472798143,
-      "loss": 46.0,
-      "step": 4285
-    },
-    {
-      "epoch": 0.32769463080834144,
-      "grad_norm": 0.0022756552789360285,
-      "learning_rate": 0.00019999994725514698,
-      "loss": 46.0,
-      "step": 4286
-    },
-    {
-      "epoch": 0.3277710877917312,
-      "grad_norm": 0.002937385579571128,
-      "learning_rate": 0.0001999999472304739,
-      "loss": 46.0,
-      "step": 4287
-    },
-    {
-      "epoch": 0.3278475447751209,
-      "grad_norm": 0.003305448219180107,
-      "learning_rate": 0.00019999994720579508,
-      "loss": 46.0,
-      "step": 4288
-    },
-    {
-      "epoch": 0.3279240017585106,
-      "grad_norm": 0.004905517678707838,
-      "learning_rate": 0.00019999994718111048,
-      "loss": 46.0,
-      "step": 4289
-    },
-    {
-      "epoch": 0.3280004587419003,
-      "grad_norm": 0.0015234446618705988,
-      "learning_rate": 0.0001999999471564201,
-      "loss": 46.0,
-      "step": 4290
-    },
-    {
-      "epoch": 0.3280769157252901,
-      "grad_norm": 0.000986318220384419,
-      "learning_rate": 0.00019999994713172395,
-      "loss": 46.0,
-      "step": 4291
-    },
-    {
-      "epoch": 0.3281533727086798,
-      "grad_norm": 0.00145861750934273,
-      "learning_rate": 0.00019999994710702203,
-      "loss": 46.0,
-      "step": 4292
-    },
-    {
-      "epoch": 0.3282298296920695,
-      "grad_norm": 0.004262291826307774,
-      "learning_rate": 0.00019999994708231434,
-      "loss": 46.0,
-      "step": 4293
-    },
-    {
-      "epoch": 0.32830628667545925,
-      "grad_norm": 0.001263973070308566,
-      "learning_rate": 0.0001999999470576009,
-      "loss": 46.0,
-      "step": 4294
-    },
-    {
-      "epoch": 0.32838274365884895,
-      "grad_norm": 0.008075137622654438,
-      "learning_rate": 0.00019999994703288166,
-      "loss": 46.0,
-      "step": 4295
-    },
-    {
-      "epoch": 0.32845920064223866,
-      "grad_norm": 0.002123615937307477,
-      "learning_rate": 0.0001999999470081567,
-      "loss": 46.0,
-      "step": 4296
-    },
-    {
-      "epoch": 0.32853565762562836,
-      "grad_norm": 0.000710161926690489,
-      "learning_rate": 0.00019999994698342591,
-      "loss": 46.0,
-      "step": 4297
-    },
-    {
-      "epoch": 0.3286121146090181,
-      "grad_norm": 0.0008752677822485566,
-      "learning_rate": 0.00019999994695868938,
-      "loss": 46.0,
-      "step": 4298
-    },
-    {
-      "epoch": 0.32868857159240783,
-      "grad_norm": 0.0013771122321486473,
-      "learning_rate": 0.00019999994693394707,
-      "loss": 46.0,
-      "step": 4299
-    },
-    {
-      "epoch": 0.32876502857579754,
-      "grad_norm": 0.0017129092011600733,
-      "learning_rate": 0.00019999994690919902,
-      "loss": 46.0,
-      "step": 4300
-    },
-    {
-      "epoch": 0.32884148555918724,
-      "grad_norm": 0.0011252955300733447,
-      "learning_rate": 0.0001999999468844452,
-      "loss": 46.0,
-      "step": 4301
-    },
-    {
-      "epoch": 0.328917942542577,
-      "grad_norm": 0.001725336886011064,
-      "learning_rate": 0.0001999999468596856,
-      "loss": 46.0,
-      "step": 4302
-    },
-    {
-      "epoch": 0.3289943995259667,
-      "grad_norm": 0.0009346289443783462,
-      "learning_rate": 0.0001999999468349202,
-      "loss": 46.0,
-      "step": 4303
-    },
-    {
-      "epoch": 0.3290708565093564,
-      "grad_norm": 0.0009860260179266334,
-      "learning_rate": 0.00019999994681014905,
-      "loss": 46.0,
-      "step": 4304
-    },
-    {
-      "epoch": 0.3291473134927461,
-      "grad_norm": 0.0021675415337085724,
-      "learning_rate": 0.00019999994678537216,
-      "loss": 46.0,
-      "step": 4305
-    },
-    {
-      "epoch": 0.3292237704761359,
-      "grad_norm": 0.001486804336309433,
-      "learning_rate": 0.00019999994676058944,
-      "loss": 46.0,
-      "step": 4306
-    },
-    {
-      "epoch": 0.3293002274595256,
-      "grad_norm": 0.0037671176251024008,
-      "learning_rate": 0.000199999946735801,
-      "loss": 46.0,
-      "step": 4307
-    },
-    {
-      "epoch": 0.3293766844429153,
-      "grad_norm": 0.000998138333670795,
-      "learning_rate": 0.00019999994671100678,
-      "loss": 46.0,
-      "step": 4308
-    },
-    {
-      "epoch": 0.32945314142630505,
-      "grad_norm": 0.0028031668625772,
-      "learning_rate": 0.00019999994668620677,
-      "loss": 46.0,
-      "step": 4309
-    },
-    {
-      "epoch": 0.32952959840969476,
-      "grad_norm": 0.0010716308606788516,
-      "learning_rate": 0.000199999946661401,
-      "loss": 46.0,
-      "step": 4310
-    },
-    {
-      "epoch": 0.32960605539308446,
-      "grad_norm": 0.0014670881209895015,
-      "learning_rate": 0.00019999994663658948,
-      "loss": 46.0,
-      "step": 4311
-    },
-    {
-      "epoch": 0.32968251237647417,
-      "grad_norm": 0.00852916482836008,
-      "learning_rate": 0.00019999994661177217,
-      "loss": 46.0,
-      "step": 4312
-    },
-    {
-      "epoch": 0.32975896935986393,
-      "grad_norm": 0.0010498393094167113,
-      "learning_rate": 0.0001999999465869491,
-      "loss": 46.0,
-      "step": 4313
-    },
-    {
-      "epoch": 0.32983542634325363,
-      "grad_norm": 0.0005530392518267035,
-      "learning_rate": 0.00019999994656212026,
-      "loss": 46.0,
-      "step": 4314
-    },
-    {
-      "epoch": 0.32991188332664334,
-      "grad_norm": 0.0014768405817449093,
-      "learning_rate": 0.00019999994653728567,
-      "loss": 46.0,
-      "step": 4315
-    },
-    {
-      "epoch": 0.32998834031003305,
-      "grad_norm": 0.0005336601752787828,
-      "learning_rate": 0.00019999994651244527,
-      "loss": 46.0,
-      "step": 4316
-    },
-    {
-      "epoch": 0.3300647972934228,
-      "grad_norm": 0.0010253757936879992,
-      "learning_rate": 0.00019999994648759912,
-      "loss": 46.0,
-      "step": 4317
-    },
-    {
-      "epoch": 0.3301412542768125,
-      "grad_norm": 0.0025468647945672274,
-      "learning_rate": 0.0001999999464627472,
-      "loss": 46.0,
-      "step": 4318
-    },
-    {
-      "epoch": 0.3302177112602022,
-      "grad_norm": 0.0008193323737941682,
-      "learning_rate": 0.00019999994643788953,
-      "loss": 46.0,
-      "step": 4319
-    },
-    {
-      "epoch": 0.3302941682435919,
-      "grad_norm": 0.0011719852918758988,
-      "learning_rate": 0.00019999994641302604,
-      "loss": 46.0,
-      "step": 4320
-    },
-    {
-      "epoch": 0.3303706252269817,
-      "grad_norm": 0.0031661740504205227,
-      "learning_rate": 0.00019999994638815683,
-      "loss": 46.0,
-      "step": 4321
-    },
-    {
-      "epoch": 0.3304470822103714,
-      "grad_norm": 0.0006985975196585059,
-      "learning_rate": 0.00019999994636328184,
-      "loss": 46.0,
-      "step": 4322
-    },
-    {
-      "epoch": 0.3305235391937611,
-      "grad_norm": 0.000613821845036,
-      "learning_rate": 0.00019999994633840106,
-      "loss": 46.0,
-      "step": 4323
-    },
-    {
-      "epoch": 0.33059999617715086,
-      "grad_norm": 0.007687657605856657,
-      "learning_rate": 0.00019999994631351452,
-      "loss": 46.0,
-      "step": 4324
-    },
-    {
-      "epoch": 0.33067645316054056,
-      "grad_norm": 0.001659299829043448,
-      "learning_rate": 0.00019999994628862222,
-      "loss": 46.0,
-      "step": 4325
-    },
-    {
-      "epoch": 0.33075291014393027,
-      "grad_norm": 0.001331859384663403,
-      "learning_rate": 0.00019999994626372414,
-      "loss": 46.0,
-      "step": 4326
-    },
-    {
-      "epoch": 0.33082936712732,
-      "grad_norm": 0.0009261846425943077,
-      "learning_rate": 0.0001999999462388203,
-      "loss": 46.0,
-      "step": 4327
-    },
-    {
-      "epoch": 0.33090582411070973,
-      "grad_norm": 0.001558349234983325,
-      "learning_rate": 0.00019999994621391068,
-      "loss": 46.0,
-      "step": 4328
-    },
-    {
-      "epoch": 0.33098228109409944,
-      "grad_norm": 0.0011646501952782273,
-      "learning_rate": 0.00019999994618899529,
-      "loss": 46.0,
-      "step": 4329
-    },
-    {
-      "epoch": 0.33105873807748915,
-      "grad_norm": 0.0010463042417541146,
-      "learning_rate": 0.00019999994616407417,
-      "loss": 46.0,
-      "step": 4330
-    },
-    {
-      "epoch": 0.33113519506087885,
-      "grad_norm": 0.0007130959420464933,
-      "learning_rate": 0.00019999994613914722,
-      "loss": 46.0,
-      "step": 4331
-    },
-    {
-      "epoch": 0.3312116520442686,
-      "grad_norm": 0.0005763443186879158,
-      "learning_rate": 0.00019999994611421453,
-      "loss": 46.0,
-      "step": 4332
-    },
-    {
-      "epoch": 0.3312881090276583,
-      "grad_norm": 0.0007845947402529418,
-      "learning_rate": 0.0001999999460892761,
-      "loss": 46.0,
-      "step": 4333
-    },
-    {
-      "epoch": 0.331364566011048,
-      "grad_norm": 0.0008663633489049971,
-      "learning_rate": 0.00019999994606433185,
-      "loss": 46.0,
-      "step": 4334
-    },
-    {
-      "epoch": 0.33144102299443773,
-      "grad_norm": 0.0008883571135811508,
-      "learning_rate": 0.00019999994603938186,
-      "loss": 46.0,
-      "step": 4335
-    },
-    {
-      "epoch": 0.3315174799778275,
-      "grad_norm": 0.0020621302537620068,
-      "learning_rate": 0.00019999994601442608,
-      "loss": 46.0,
-      "step": 4336
-    },
-    {
-      "epoch": 0.3315939369612172,
-      "grad_norm": 0.0015221703797578812,
-      "learning_rate": 0.00019999994598946454,
-      "loss": 46.0,
-      "step": 4337
-    },
-    {
-      "epoch": 0.3316703939446069,
-      "grad_norm": 0.0024182454217225313,
-      "learning_rate": 0.00019999994596449724,
-      "loss": 46.0,
-      "step": 4338
-    },
-    {
-      "epoch": 0.33174685092799666,
-      "grad_norm": 0.0007478754268959165,
-      "learning_rate": 0.00019999994593952413,
-      "loss": 46.0,
-      "step": 4339
-    },
-    {
-      "epoch": 0.33182330791138637,
-      "grad_norm": 0.0014873365871608257,
-      "learning_rate": 0.0001999999459145453,
-      "loss": 46.0,
-      "step": 4340
-    },
-    {
-      "epoch": 0.3318997648947761,
-      "grad_norm": 0.003358301008120179,
-      "learning_rate": 0.00019999994588956068,
-      "loss": 46.0,
-      "step": 4341
-    },
-    {
-      "epoch": 0.3319762218781658,
-      "grad_norm": 0.0025808545760810375,
-      "learning_rate": 0.0001999999458645703,
-      "loss": 46.0,
-      "step": 4342
-    },
-    {
-      "epoch": 0.33205267886155554,
-      "grad_norm": 0.0005722782807424664,
-      "learning_rate": 0.00019999994583957414,
-      "loss": 46.0,
-      "step": 4343
-    },
-    {
-      "epoch": 0.33212913584494524,
-      "grad_norm": 0.0008264755015261471,
-      "learning_rate": 0.00019999994581457222,
-      "loss": 46.0,
-      "step": 4344
-    },
-    {
-      "epoch": 0.33220559282833495,
-      "grad_norm": 0.0036591014359146357,
-      "learning_rate": 0.00019999994578956453,
-      "loss": 46.0,
-      "step": 4345
-    },
-    {
-      "epoch": 0.33228204981172466,
-      "grad_norm": 0.002337259706109762,
-      "learning_rate": 0.0001999999457645511,
-      "loss": 46.0,
-      "step": 4346
-    },
-    {
-      "epoch": 0.3323585067951144,
-      "grad_norm": 0.0008739191689528525,
-      "learning_rate": 0.00019999994573953182,
-      "loss": 46.0,
-      "step": 4347
-    },
-    {
-      "epoch": 0.3324349637785041,
-      "grad_norm": 0.0007346050697378814,
-      "learning_rate": 0.00019999994571450684,
-      "loss": 46.0,
-      "step": 4348
-    },
-    {
-      "epoch": 0.33251142076189383,
-      "grad_norm": 0.0006359497201628983,
-      "learning_rate": 0.00019999994568947608,
-      "loss": 46.0,
-      "step": 4349
-    },
-    {
-      "epoch": 0.33258787774528353,
-      "grad_norm": 0.0005062447744421661,
-      "learning_rate": 0.00019999994566443952,
-      "loss": 46.0,
-      "step": 4350
-    },
-    {
-      "epoch": 0.3326643347286733,
-      "grad_norm": 0.00189758010674268,
-      "learning_rate": 0.00019999994563939718,
-      "loss": 46.0,
-      "step": 4351
-    },
-    {
-      "epoch": 0.332740791712063,
-      "grad_norm": 0.0012883191229775548,
-      "learning_rate": 0.0001999999456143491,
-      "loss": 46.0,
-      "step": 4352
-    },
-    {
-      "epoch": 0.3328172486954527,
-      "grad_norm": 0.000746228382922709,
-      "learning_rate": 0.00019999994558929528,
-      "loss": 46.0,
-      "step": 4353
-    },
-    {
-      "epoch": 0.33289370567884247,
-      "grad_norm": 0.0010014697909355164,
-      "learning_rate": 0.00019999994556423566,
-      "loss": 46.0,
-      "step": 4354
-    },
-    {
-      "epoch": 0.33297016266223217,
-      "grad_norm": 0.000663650338537991,
-      "learning_rate": 0.00019999994553917026,
-      "loss": 46.0,
-      "step": 4355
-    },
-    {
-      "epoch": 0.3330466196456219,
-      "grad_norm": 0.0008152858936227858,
-      "learning_rate": 0.0001999999455140991,
-      "loss": 46.0,
-      "step": 4356
-    },
-    {
-      "epoch": 0.3331230766290116,
-      "grad_norm": 0.0025598809588700533,
-      "learning_rate": 0.0001999999454890222,
-      "loss": 46.0,
-      "step": 4357
-    },
-    {
-      "epoch": 0.33319953361240134,
-      "grad_norm": 0.0004814782878383994,
-      "learning_rate": 0.00019999994546393947,
-      "loss": 46.0,
-      "step": 4358
-    },
-    {
-      "epoch": 0.33327599059579105,
-      "grad_norm": 0.00047982606338337064,
-      "learning_rate": 0.000199999945438851,
-      "loss": 46.0,
-      "step": 4359
-    },
-    {
-      "epoch": 0.33335244757918076,
-      "grad_norm": 0.0028309291228652,
-      "learning_rate": 0.00019999994541375677,
-      "loss": 46.0,
-      "step": 4360
-    },
-    {
-      "epoch": 0.33342890456257046,
-      "grad_norm": 0.0008123546140268445,
-      "learning_rate": 0.00019999994538865678,
-      "loss": 46.0,
-      "step": 4361
-    },
-    {
-      "epoch": 0.3335053615459602,
-      "grad_norm": 0.0009709223522804677,
-      "learning_rate": 0.000199999945363551,
-      "loss": 46.0,
-      "step": 4362
-    },
-    {
-      "epoch": 0.3335818185293499,
-      "grad_norm": 0.0009069971274584532,
-      "learning_rate": 0.00019999994533843944,
-      "loss": 46.0,
-      "step": 4363
-    },
-    {
-      "epoch": 0.33365827551273963,
-      "grad_norm": 0.001450235489755869,
-      "learning_rate": 0.00019999994531332213,
-      "loss": 46.0,
-      "step": 4364
-    },
-    {
-      "epoch": 0.33373473249612934,
-      "grad_norm": 0.0010049097472801805,
-      "learning_rate": 0.00019999994528819906,
-      "loss": 46.0,
-      "step": 4365
-    },
-    {
-      "epoch": 0.3338111894795191,
-      "grad_norm": 0.0011636137496680021,
-      "learning_rate": 0.0001999999452630702,
-      "loss": 46.0,
-      "step": 4366
-    },
-    {
-      "epoch": 0.3338876464629088,
-      "grad_norm": 0.0032694179099053144,
-      "learning_rate": 0.00019999994523793558,
-      "loss": 46.0,
-      "step": 4367
-    },
-    {
-      "epoch": 0.3339641034462985,
-      "grad_norm": 0.00974624790251255,
-      "learning_rate": 0.00019999994521279518,
-      "loss": 46.0,
-      "step": 4368
-    },
-    {
-      "epoch": 0.33404056042968827,
-      "grad_norm": 0.0007959165377542377,
-      "learning_rate": 0.000199999945187649,
-      "loss": 46.0,
-      "step": 4369
-    },
-    {
-      "epoch": 0.334117017413078,
-      "grad_norm": 0.00528798857703805,
-      "learning_rate": 0.0001999999451624971,
-      "loss": 46.0,
-      "step": 4370
-    },
-    {
-      "epoch": 0.3341934743964677,
-      "grad_norm": 0.001440275926142931,
-      "learning_rate": 0.0001999999451373394,
-      "loss": 46.0,
-      "step": 4371
-    },
-    {
-      "epoch": 0.3342699313798574,
-      "grad_norm": 0.0020948785822838545,
-      "learning_rate": 0.00019999994511217593,
-      "loss": 46.0,
-      "step": 4372
-    },
-    {
-      "epoch": 0.33434638836324715,
-      "grad_norm": 0.0010314505780115724,
-      "learning_rate": 0.00019999994508700667,
-      "loss": 46.0,
-      "step": 4373
-    },
-    {
-      "epoch": 0.33442284534663685,
-      "grad_norm": 0.0012385656591504812,
-      "learning_rate": 0.00019999994506183168,
-      "loss": 46.0,
-      "step": 4374
-    },
-    {
-      "epoch": 0.33449930233002656,
-      "grad_norm": 0.00176147127058357,
-      "learning_rate": 0.0001999999450366509,
-      "loss": 46.0,
-      "step": 4375
-    },
-    {
-      "epoch": 0.33457575931341627,
-      "grad_norm": 0.0009488561772741377,
-      "learning_rate": 0.00019999994501146434,
-      "loss": 46.0,
-      "step": 4376
-    },
-    {
-      "epoch": 0.334652216296806,
-      "grad_norm": 0.0010161340469494462,
-      "learning_rate": 0.00019999994498627203,
-      "loss": 46.0,
-      "step": 4377
-    },
-    {
-      "epoch": 0.33472867328019573,
-      "grad_norm": 0.0014125718735158443,
-      "learning_rate": 0.00019999994496107396,
-      "loss": 46.0,
-      "step": 4378
-    },
-    {
-      "epoch": 0.33480513026358544,
-      "grad_norm": 0.0003296210488770157,
-      "learning_rate": 0.00019999994493587008,
-      "loss": 46.0,
-      "step": 4379
-    },
-    {
-      "epoch": 0.33488158724697514,
-      "grad_norm": 0.0012108338996767998,
-      "learning_rate": 0.00019999994491066048,
-      "loss": 46.0,
-      "step": 4380
-    },
-    {
-      "epoch": 0.3349580442303649,
-      "grad_norm": 0.000855764839798212,
-      "learning_rate": 0.00019999994488544505,
-      "loss": 46.0,
-      "step": 4381
-    },
-    {
-      "epoch": 0.3350345012137546,
-      "grad_norm": 0.003351888619363308,
-      "learning_rate": 0.0001999999448602239,
-      "loss": 46.0,
-      "step": 4382
-    },
-    {
-      "epoch": 0.3351109581971443,
-      "grad_norm": 0.004446738865226507,
-      "learning_rate": 0.00019999994483499696,
-      "loss": 46.0,
-      "step": 4383
-    },
-    {
-      "epoch": 0.3351874151805341,
-      "grad_norm": 0.0009897021809592843,
-      "learning_rate": 0.00019999994480976425,
-      "loss": 46.0,
-      "step": 4384
-    },
-    {
-      "epoch": 0.3352638721639238,
-      "grad_norm": 0.001084261341020465,
-      "learning_rate": 0.00019999994478452578,
-      "loss": 46.0,
-      "step": 4385
-    },
-    {
-      "epoch": 0.3353403291473135,
-      "grad_norm": 0.0018345518037676811,
-      "learning_rate": 0.00019999994475928157,
-      "loss": 46.0,
-      "step": 4386
-    },
-    {
-      "epoch": 0.3354167861307032,
-      "grad_norm": 0.000890737515874207,
-      "learning_rate": 0.00019999994473403153,
-      "loss": 46.0,
-      "step": 4387
-    },
-    {
-      "epoch": 0.33549324311409295,
-      "grad_norm": 0.0012093266705051064,
-      "learning_rate": 0.00019999994470877575,
-      "loss": 46.0,
-      "step": 4388
-    },
-    {
-      "epoch": 0.33556970009748266,
-      "grad_norm": 0.0005385908880271018,
-      "learning_rate": 0.00019999994468351422,
-      "loss": 46.0,
-      "step": 4389
-    },
-    {
-      "epoch": 0.33564615708087236,
-      "grad_norm": 0.0007514269673265517,
-      "learning_rate": 0.00019999994465824688,
-      "loss": 46.0,
-      "step": 4390
-    },
-    {
-      "epoch": 0.33572261406426207,
-      "grad_norm": 0.012829259969294071,
-      "learning_rate": 0.0001999999446329738,
-      "loss": 46.0,
-      "step": 4391
-    },
-    {
-      "epoch": 0.33579907104765183,
-      "grad_norm": 0.0026637602131813765,
-      "learning_rate": 0.00019999994460769495,
-      "loss": 46.0,
-      "step": 4392
-    },
-    {
-      "epoch": 0.33587552803104154,
-      "grad_norm": 0.003662163158878684,
-      "learning_rate": 0.0001999999445824103,
-      "loss": 46.0,
-      "step": 4393
-    },
-    {
-      "epoch": 0.33595198501443124,
-      "grad_norm": 0.0014186989283189178,
-      "learning_rate": 0.0001999999445571199,
-      "loss": 46.0,
-      "step": 4394
-    },
-    {
-      "epoch": 0.336028441997821,
-      "grad_norm": 0.0011301335180178285,
-      "learning_rate": 0.00019999994453182373,
-      "loss": 46.0,
-      "step": 4395
-    },
-    {
-      "epoch": 0.3361048989812107,
-      "grad_norm": 0.00131536356639117,
-      "learning_rate": 0.0001999999445065218,
-      "loss": 46.0,
-      "step": 4396
-    },
-    {
-      "epoch": 0.3361813559646004,
-      "grad_norm": 0.0040961154736578465,
-      "learning_rate": 0.0001999999444812141,
-      "loss": 46.0,
-      "step": 4397
-    },
-    {
-      "epoch": 0.3362578129479901,
-      "grad_norm": 0.0013224639697000384,
-      "learning_rate": 0.00019999994445590063,
-      "loss": 46.0,
-      "step": 4398
-    },
-    {
-      "epoch": 0.3363342699313799,
-      "grad_norm": 0.0012360153486952186,
-      "learning_rate": 0.0001999999444305814,
-      "loss": 46.0,
-      "step": 4399
-    },
-    {
-      "epoch": 0.3364107269147696,
-      "grad_norm": 0.0014156968099996448,
-      "learning_rate": 0.00019999994440525636,
-      "loss": 46.0,
-      "step": 4400
-    },
-    {
-      "epoch": 0.3364871838981593,
-      "grad_norm": 0.0012368501629680395,
-      "learning_rate": 0.0001999999443799256,
-      "loss": 46.0,
-      "step": 4401
-    },
-    {
-      "epoch": 0.336563640881549,
-      "grad_norm": 0.0013104111421853304,
-      "learning_rate": 0.00019999994435458901,
-      "loss": 46.0,
-      "step": 4402
-    },
-    {
-      "epoch": 0.33664009786493876,
-      "grad_norm": 0.0010781441815197468,
-      "learning_rate": 0.0001999999443292467,
-      "loss": 46.0,
-      "step": 4403
-    },
-    {
-      "epoch": 0.33671655484832846,
-      "grad_norm": 0.001098192878998816,
-      "learning_rate": 0.00019999994430389863,
-      "loss": 46.0,
-      "step": 4404
-    },
-    {
-      "epoch": 0.33679301183171817,
-      "grad_norm": 0.0014286516234278679,
-      "learning_rate": 0.00019999994427854475,
-      "loss": 46.0,
-      "step": 4405
-    },
-    {
-      "epoch": 0.3368694688151079,
-      "grad_norm": 0.0013937565963715315,
-      "learning_rate": 0.0001999999442531851,
-      "loss": 46.0,
-      "step": 4406
-    },
-    {
-      "epoch": 0.33694592579849764,
-      "grad_norm": 0.0026233461685478687,
-      "learning_rate": 0.0001999999442278197,
-      "loss": 46.0,
-      "step": 4407
-    },
-    {
-      "epoch": 0.33702238278188734,
-      "grad_norm": 0.002443751785904169,
-      "learning_rate": 0.00019999994420244853,
-      "loss": 46.0,
-      "step": 4408
-    },
-    {
-      "epoch": 0.33709883976527705,
-      "grad_norm": 0.0006538288434967399,
-      "learning_rate": 0.00019999994417707162,
-      "loss": 46.0,
-      "step": 4409
-    },
-    {
-      "epoch": 0.3371752967486668,
-      "grad_norm": 0.0006216029287315905,
-      "learning_rate": 0.00019999994415168893,
-      "loss": 46.0,
-      "step": 4410
-    },
-    {
-      "epoch": 0.3372517537320565,
-      "grad_norm": 0.009112684987485409,
-      "learning_rate": 0.00019999994412630044,
-      "loss": 46.0,
-      "step": 4411
-    },
-    {
-      "epoch": 0.3373282107154462,
-      "grad_norm": 0.0023524202406406403,
-      "learning_rate": 0.00019999994410090617,
-      "loss": 46.0,
-      "step": 4412
-    },
-    {
-      "epoch": 0.3374046676988359,
-      "grad_norm": 0.0005722793866880238,
-      "learning_rate": 0.0001999999440755062,
-      "loss": 46.0,
-      "step": 4413
-    },
-    {
-      "epoch": 0.3374811246822257,
-      "grad_norm": 0.0006286685238592327,
-      "learning_rate": 0.00019999994405010038,
-      "loss": 46.0,
-      "step": 4414
-    },
-    {
-      "epoch": 0.3375575816656154,
-      "grad_norm": 0.0016417823499068618,
-      "learning_rate": 0.00019999994402468882,
-      "loss": 46.0,
-      "step": 4415
-    },
-    {
-      "epoch": 0.3376340386490051,
-      "grad_norm": 0.0013658159878104925,
-      "learning_rate": 0.00019999994399927151,
-      "loss": 46.0,
-      "step": 4416
-    },
-    {
-      "epoch": 0.3377104956323948,
-      "grad_norm": 0.0008710395777598023,
-      "learning_rate": 0.0001999999439738484,
-      "loss": 46.0,
-      "step": 4417
-    },
-    {
-      "epoch": 0.33778695261578456,
-      "grad_norm": 0.0011983991134911776,
-      "learning_rate": 0.00019999994394841953,
-      "loss": 46.0,
-      "step": 4418
-    },
-    {
-      "epoch": 0.33786340959917427,
-      "grad_norm": 0.0006315264035947621,
-      "learning_rate": 0.00019999994392298488,
-      "loss": 46.0,
-      "step": 4419
-    },
-    {
-      "epoch": 0.337939866582564,
-      "grad_norm": 0.0011137855472043157,
-      "learning_rate": 0.00019999994389754448,
-      "loss": 46.0,
-      "step": 4420
-    },
-    {
-      "epoch": 0.3380163235659537,
-      "grad_norm": 0.0010404310887679458,
-      "learning_rate": 0.00019999994387209834,
-      "loss": 46.0,
-      "step": 4421
-    },
-    {
-      "epoch": 0.33809278054934344,
-      "grad_norm": 0.0018426774768158793,
-      "learning_rate": 0.0001999999438466464,
-      "loss": 46.0,
-      "step": 4422
-    },
-    {
-      "epoch": 0.33816923753273315,
-      "grad_norm": 0.0012585781514644623,
-      "learning_rate": 0.00019999994382118868,
-      "loss": 46.0,
-      "step": 4423
-    },
-    {
-      "epoch": 0.33824569451612285,
-      "grad_norm": 0.002455796580761671,
-      "learning_rate": 0.0001999999437957252,
-      "loss": 46.0,
-      "step": 4424
-    },
-    {
-      "epoch": 0.3383221514995126,
-      "grad_norm": 0.0013544521061703563,
-      "learning_rate": 0.00019999994377025595,
-      "loss": 46.0,
-      "step": 4425
-    },
-    {
-      "epoch": 0.3383986084829023,
-      "grad_norm": 0.0008386909612454474,
-      "learning_rate": 0.0001999999437447809,
-      "loss": 46.0,
-      "step": 4426
-    },
-    {
-      "epoch": 0.338475065466292,
-      "grad_norm": 0.0021937384735792875,
-      "learning_rate": 0.00019999994371930013,
-      "loss": 46.0,
-      "step": 4427
-    },
-    {
-      "epoch": 0.33855152244968173,
-      "grad_norm": 0.0005258836899884045,
-      "learning_rate": 0.0001999999436938136,
-      "loss": 46.0,
-      "step": 4428
-    },
-    {
-      "epoch": 0.3386279794330715,
-      "grad_norm": 0.004467345774173737,
-      "learning_rate": 0.00019999994366832124,
-      "loss": 46.0,
-      "step": 4429
-    },
-    {
-      "epoch": 0.3387044364164612,
-      "grad_norm": 0.0008250975515693426,
-      "learning_rate": 0.00019999994364282316,
-      "loss": 46.0,
-      "step": 4430
-    },
-    {
-      "epoch": 0.3387808933998509,
-      "grad_norm": 0.0029229619540274143,
-      "learning_rate": 0.00019999994361731926,
-      "loss": 46.0,
-      "step": 4431
-    },
-    {
-      "epoch": 0.3388573503832406,
-      "grad_norm": 0.0012704990804195404,
-      "learning_rate": 0.00019999994359180964,
-      "loss": 46.0,
-      "step": 4432
-    },
-    {
-      "epoch": 0.33893380736663037,
-      "grad_norm": 0.0006872370140627027,
-      "learning_rate": 0.00019999994356629424,
-      "loss": 46.0,
-      "step": 4433
-    },
-    {
-      "epoch": 0.3390102643500201,
-      "grad_norm": 0.00018776826618704945,
-      "learning_rate": 0.00019999994354077307,
-      "loss": 46.0,
-      "step": 4434
-    },
-    {
-      "epoch": 0.3390867213334098,
-      "grad_norm": 0.0013287997571751475,
-      "learning_rate": 0.00019999994351524612,
-      "loss": 46.0,
-      "step": 4435
-    },
-    {
-      "epoch": 0.3391631783167995,
-      "grad_norm": 0.0006739794625900686,
-      "learning_rate": 0.0001999999434897134,
-      "loss": 46.0,
-      "step": 4436
-    },
-    {
-      "epoch": 0.33923963530018925,
-      "grad_norm": 0.001005502650514245,
-      "learning_rate": 0.00019999994346417492,
-      "loss": 46.0,
-      "step": 4437
-    },
-    {
-      "epoch": 0.33931609228357895,
-      "grad_norm": 0.0013392752734944224,
-      "learning_rate": 0.00019999994343863068,
-      "loss": 46.0,
-      "step": 4438
-    },
-    {
-      "epoch": 0.33939254926696866,
-      "grad_norm": 0.0011017059441655874,
-      "learning_rate": 0.00019999994341308064,
-      "loss": 46.0,
-      "step": 4439
-    },
-    {
-      "epoch": 0.3394690062503584,
-      "grad_norm": 0.002742550801485777,
-      "learning_rate": 0.00019999994338752486,
-      "loss": 46.0,
-      "step": 4440
-    },
-    {
-      "epoch": 0.3395454632337481,
-      "grad_norm": 0.0007794153643772006,
-      "learning_rate": 0.00019999994336196327,
-      "loss": 46.0,
-      "step": 4441
-    },
-    {
-      "epoch": 0.33962192021713783,
-      "grad_norm": 0.0006616926984861493,
-      "learning_rate": 0.00019999994333639594,
-      "loss": 46.0,
-      "step": 4442
-    },
-    {
-      "epoch": 0.33969837720052753,
-      "grad_norm": 0.0005805851542390883,
-      "learning_rate": 0.00019999994331082287,
-      "loss": 46.0,
-      "step": 4443
-    },
-    {
-      "epoch": 0.3397748341839173,
-      "grad_norm": 0.0007528080604970455,
-      "learning_rate": 0.000199999943285244,
-      "loss": 46.0,
-      "step": 4444
-    },
-    {
-      "epoch": 0.339851291167307,
-      "grad_norm": 0.0010019076289609075,
-      "learning_rate": 0.00019999994325965934,
-      "loss": 46.0,
-      "step": 4445
-    },
-    {
-      "epoch": 0.3399277481506967,
-      "grad_norm": 0.002531372243538499,
-      "learning_rate": 0.00019999994323406894,
-      "loss": 46.0,
-      "step": 4446
-    },
-    {
-      "epoch": 0.3400042051340864,
-      "grad_norm": 0.0014815061585977674,
-      "learning_rate": 0.00019999994320847275,
-      "loss": 46.0,
-      "step": 4447
-    },
-    {
-      "epoch": 0.3400806621174762,
-      "grad_norm": 0.0005497461534105241,
-      "learning_rate": 0.0001999999431828708,
-      "loss": 46.0,
-      "step": 4448
-    },
-    {
-      "epoch": 0.3401571191008659,
-      "grad_norm": 0.0017356178723275661,
-      "learning_rate": 0.00019999994315726308,
-      "loss": 46.0,
-      "step": 4449
-    },
-    {
-      "epoch": 0.3402335760842556,
-      "grad_norm": 0.004692509770393372,
-      "learning_rate": 0.0001999999431316496,
-      "loss": 46.0,
-      "step": 4450
-    },
-    {
-      "epoch": 0.3403100330676453,
-      "grad_norm": 0.0018580228788778186,
-      "learning_rate": 0.00019999994310603033,
-      "loss": 46.0,
-      "step": 4451
-    },
-    {
-      "epoch": 0.34038649005103505,
-      "grad_norm": 0.0013220901601016521,
-      "learning_rate": 0.00019999994308040532,
-      "loss": 46.0,
-      "step": 4452
-    },
-    {
-      "epoch": 0.34046294703442476,
-      "grad_norm": 0.00034104715450666845,
-      "learning_rate": 0.00019999994305477454,
-      "loss": 46.0,
-      "step": 4453
-    },
-    {
-      "epoch": 0.34053940401781446,
-      "grad_norm": 0.0004975063493475318,
-      "learning_rate": 0.00019999994302913795,
-      "loss": 46.0,
-      "step": 4454
-    },
-    {
-      "epoch": 0.3406158610012042,
-      "grad_norm": 0.005003293976187706,
-      "learning_rate": 0.00019999994300349562,
-      "loss": 46.0,
-      "step": 4455
-    },
-    {
-      "epoch": 0.34069231798459393,
-      "grad_norm": 0.0006284716073423624,
-      "learning_rate": 0.00019999994297784752,
-      "loss": 46.0,
-      "step": 4456
-    },
-    {
-      "epoch": 0.34076877496798363,
-      "grad_norm": 0.0011805167887359858,
-      "learning_rate": 0.00019999994295219367,
-      "loss": 46.0,
-      "step": 4457
-    },
-    {
-      "epoch": 0.34084523195137334,
-      "grad_norm": 0.0014241504250094295,
-      "learning_rate": 0.00019999994292653402,
-      "loss": 46.0,
-      "step": 4458
-    },
-    {
-      "epoch": 0.3409216889347631,
-      "grad_norm": 0.0010949663119390607,
-      "learning_rate": 0.0001999999429008686,
-      "loss": 46.0,
-      "step": 4459
-    },
-    {
-      "epoch": 0.3409981459181528,
-      "grad_norm": 0.0015556616708636284,
-      "learning_rate": 0.0001999999428751974,
-      "loss": 46.0,
-      "step": 4460
-    },
-    {
-      "epoch": 0.3410746029015425,
-      "grad_norm": 0.0016401479952037334,
-      "learning_rate": 0.00019999994284952046,
-      "loss": 46.0,
-      "step": 4461
-    },
-    {
-      "epoch": 0.3411510598849322,
-      "grad_norm": 0.0011257074074819684,
-      "learning_rate": 0.00019999994282383774,
-      "loss": 46.0,
-      "step": 4462
-    },
-    {
-      "epoch": 0.341227516868322,
-      "grad_norm": 0.0009559948230162263,
-      "learning_rate": 0.00019999994279814925,
-      "loss": 46.0,
-      "step": 4463
-    },
-    {
-      "epoch": 0.3413039738517117,
-      "grad_norm": 0.0007614089990966022,
-      "learning_rate": 0.000199999942772455,
-      "loss": 46.0,
-      "step": 4464
-    },
-    {
-      "epoch": 0.3413804308351014,
-      "grad_norm": 0.002122860634699464,
-      "learning_rate": 0.00019999994274675498,
-      "loss": 46.0,
-      "step": 4465
-    },
-    {
-      "epoch": 0.3414568878184911,
-      "grad_norm": 0.0006695298361591995,
-      "learning_rate": 0.0001999999427210492,
-      "loss": 46.0,
-      "step": 4466
-    },
-    {
-      "epoch": 0.34153334480188086,
-      "grad_norm": 0.006944767199456692,
-      "learning_rate": 0.0001999999426953376,
-      "loss": 46.0,
-      "step": 4467
-    },
-    {
-      "epoch": 0.34160980178527056,
-      "grad_norm": 0.0006343826535157859,
-      "learning_rate": 0.00019999994266962028,
-      "loss": 46.0,
-      "step": 4468
-    },
-    {
-      "epoch": 0.34168625876866027,
-      "grad_norm": 0.0006700207013636827,
-      "learning_rate": 0.00019999994264389718,
-      "loss": 46.0,
-      "step": 4469
-    },
-    {
-      "epoch": 0.34176271575205003,
-      "grad_norm": 0.000629562942776829,
-      "learning_rate": 0.00019999994261816827,
-      "loss": 46.0,
-      "step": 4470
-    },
-    {
-      "epoch": 0.34183917273543973,
-      "grad_norm": 0.0008044940186664462,
-      "learning_rate": 0.00019999994259243362,
-      "loss": 46.0,
-      "step": 4471
-    },
-    {
-      "epoch": 0.34191562971882944,
-      "grad_norm": 0.000491089653223753,
-      "learning_rate": 0.00019999994256669323,
-      "loss": 46.0,
-      "step": 4472
-    },
-    {
-      "epoch": 0.34199208670221914,
-      "grad_norm": 0.0008835463668219745,
-      "learning_rate": 0.00019999994254094703,
-      "loss": 46.0,
-      "step": 4473
-    },
-    {
-      "epoch": 0.3420685436856089,
-      "grad_norm": 0.0011827208800241351,
-      "learning_rate": 0.0001999999425151951,
-      "loss": 46.0,
-      "step": 4474
-    },
-    {
-      "epoch": 0.3421450006689986,
-      "grad_norm": 0.0028299165423959494,
-      "learning_rate": 0.00019999994248943737,
-      "loss": 46.0,
-      "step": 4475
-    },
-    {
-      "epoch": 0.3422214576523883,
-      "grad_norm": 0.003202289342880249,
-      "learning_rate": 0.00019999994246367386,
-      "loss": 46.0,
-      "step": 4476
-    },
-    {
-      "epoch": 0.342297914635778,
-      "grad_norm": 0.004717707633972168,
-      "learning_rate": 0.00019999994243790462,
-      "loss": 46.0,
-      "step": 4477
-    },
-    {
-      "epoch": 0.3423743716191678,
-      "grad_norm": 0.0030418529640883207,
-      "learning_rate": 0.00019999994241212958,
-      "loss": 46.0,
-      "step": 4478
-    },
-    {
-      "epoch": 0.3424508286025575,
-      "grad_norm": 0.0009572543203830719,
-      "learning_rate": 0.00019999994238634877,
-      "loss": 46.0,
-      "step": 4479
-    },
-    {
-      "epoch": 0.3425272855859472,
-      "grad_norm": 0.0011142563307657838,
-      "learning_rate": 0.00019999994236056222,
-      "loss": 46.0,
-      "step": 4480
-    },
-    {
-      "epoch": 0.3426037425693369,
-      "grad_norm": 0.0005846779677085578,
-      "learning_rate": 0.00019999994233476986,
-      "loss": 46.0,
-      "step": 4481
-    },
-    {
-      "epoch": 0.34268019955272666,
-      "grad_norm": 0.0010721880244091153,
-      "learning_rate": 0.00019999994230897173,
-      "loss": 46.0,
-      "step": 4482
-    },
-    {
-      "epoch": 0.34275665653611637,
-      "grad_norm": 0.004597486928105354,
-      "learning_rate": 0.00019999994228316788,
-      "loss": 46.0,
-      "step": 4483
-    },
-    {
-      "epoch": 0.34283311351950607,
-      "grad_norm": 0.0008741975179873407,
-      "learning_rate": 0.00019999994225735823,
-      "loss": 46.0,
-      "step": 4484
-    },
-    {
-      "epoch": 0.34290957050289583,
-      "grad_norm": 0.0020603176672011614,
-      "learning_rate": 0.00019999994223154278,
-      "loss": 46.0,
-      "step": 4485
-    },
-    {
-      "epoch": 0.34298602748628554,
-      "grad_norm": 0.0020062404219061136,
-      "learning_rate": 0.0001999999422057216,
-      "loss": 46.0,
-      "step": 4486
-    },
-    {
-      "epoch": 0.34306248446967524,
-      "grad_norm": 0.0007867870735935867,
-      "learning_rate": 0.00019999994217989467,
-      "loss": 46.0,
-      "step": 4487
-    },
-    {
-      "epoch": 0.34313894145306495,
-      "grad_norm": 0.0017731457483023405,
-      "learning_rate": 0.00019999994215406193,
-      "loss": 46.0,
-      "step": 4488
-    },
-    {
-      "epoch": 0.3432153984364547,
-      "grad_norm": 0.0017227425705641508,
-      "learning_rate": 0.00019999994212822344,
-      "loss": 46.0,
-      "step": 4489
-    },
-    {
-      "epoch": 0.3432918554198444,
-      "grad_norm": 0.0044555384665727615,
-      "learning_rate": 0.00019999994210237918,
-      "loss": 46.0,
-      "step": 4490
-    },
-    {
-      "epoch": 0.3433683124032341,
-      "grad_norm": 0.0008505648002028465,
-      "learning_rate": 0.00019999994207652914,
-      "loss": 46.0,
-      "step": 4491
-    },
-    {
-      "epoch": 0.3434447693866238,
-      "grad_norm": 0.0019806567579507828,
-      "learning_rate": 0.00019999994205067333,
-      "loss": 46.0,
-      "step": 4492
-    },
-    {
-      "epoch": 0.3435212263700136,
-      "grad_norm": 0.0010835396824404597,
-      "learning_rate": 0.00019999994202481175,
-      "loss": 46.0,
-      "step": 4493
-    },
-    {
-      "epoch": 0.3435976833534033,
-      "grad_norm": 0.00045218656305223703,
-      "learning_rate": 0.00019999994199894442,
-      "loss": 46.0,
-      "step": 4494
-    },
-    {
-      "epoch": 0.343674140336793,
-      "grad_norm": 0.0007355757406912744,
-      "learning_rate": 0.00019999994197307132,
-      "loss": 46.0,
-      "step": 4495
-    },
-    {
-      "epoch": 0.3437505973201827,
-      "grad_norm": 0.001620026072487235,
-      "learning_rate": 0.00019999994194719244,
-      "loss": 46.0,
-      "step": 4496
-    },
-    {
-      "epoch": 0.34382705430357247,
-      "grad_norm": 0.002306954935193062,
-      "learning_rate": 0.00019999994192130776,
-      "loss": 46.0,
-      "step": 4497
-    },
-    {
-      "epoch": 0.34390351128696217,
-      "grad_norm": 0.002012313809245825,
-      "learning_rate": 0.00019999994189541734,
-      "loss": 46.0,
-      "step": 4498
-    },
-    {
-      "epoch": 0.3439799682703519,
-      "grad_norm": 0.0005791075527667999,
-      "learning_rate": 0.00019999994186952117,
-      "loss": 46.0,
-      "step": 4499
-    },
-    {
-      "epoch": 0.34405642525374164,
-      "grad_norm": 0.0008548152400180697,
-      "learning_rate": 0.0001999999418436192,
-      "loss": 46.0,
-      "step": 4500
-    },
-    {
-      "epoch": 0.34413288223713134,
-      "grad_norm": 0.0043380907736718655,
-      "learning_rate": 0.00019999994181771146,
-      "loss": 46.0,
-      "step": 4501
-    },
-    {
-      "epoch": 0.34420933922052105,
-      "grad_norm": 0.003028023988008499,
-      "learning_rate": 0.00019999994179179797,
-      "loss": 46.0,
-      "step": 4502
-    },
-    {
-      "epoch": 0.34428579620391075,
-      "grad_norm": 0.0018402260029688478,
-      "learning_rate": 0.0001999999417658787,
-      "loss": 46.0,
-      "step": 4503
-    },
-    {
-      "epoch": 0.3443622531873005,
-      "grad_norm": 0.0015750774182379246,
-      "learning_rate": 0.00019999994173995367,
-      "loss": 46.0,
-      "step": 4504
-    },
-    {
-      "epoch": 0.3444387101706902,
-      "grad_norm": 0.001570810447447002,
-      "learning_rate": 0.00019999994171402286,
-      "loss": 46.0,
-      "step": 4505
-    },
-    {
-      "epoch": 0.3445151671540799,
-      "grad_norm": 0.0012720414670184255,
-      "learning_rate": 0.00019999994168808628,
-      "loss": 46.0,
-      "step": 4506
-    },
-    {
-      "epoch": 0.34459162413746963,
-      "grad_norm": 0.002900895895436406,
-      "learning_rate": 0.00019999994166214395,
-      "loss": 46.0,
-      "step": 4507
-    },
-    {
-      "epoch": 0.3446680811208594,
-      "grad_norm": 0.0009334111819043756,
-      "learning_rate": 0.00019999994163619582,
-      "loss": 46.0,
-      "step": 4508
-    },
-    {
-      "epoch": 0.3447445381042491,
-      "grad_norm": 0.0019853694830089808,
-      "learning_rate": 0.00019999994161024195,
-      "loss": 46.0,
-      "step": 4509
-    },
-    {
-      "epoch": 0.3448209950876388,
-      "grad_norm": 0.0009553749114274979,
-      "learning_rate": 0.00019999994158428227,
-      "loss": 46.0,
-      "step": 4510
-    },
-    {
-      "epoch": 0.34489745207102857,
-      "grad_norm": 0.0013566161505877972,
-      "learning_rate": 0.00019999994155831688,
-      "loss": 46.0,
-      "step": 4511
-    },
-    {
-      "epoch": 0.34497390905441827,
-      "grad_norm": 0.004217440728098154,
-      "learning_rate": 0.00019999994153234568,
-      "loss": 46.0,
-      "step": 4512
-    },
-    {
-      "epoch": 0.345050366037808,
-      "grad_norm": 0.001521850354038179,
-      "learning_rate": 0.0001999999415063687,
-      "loss": 46.0,
-      "step": 4513
-    },
-    {
-      "epoch": 0.3451268230211977,
-      "grad_norm": 0.0019429214298725128,
-      "learning_rate": 0.000199999941480386,
-      "loss": 46.0,
-      "step": 4514
-    },
-    {
-      "epoch": 0.34520328000458744,
-      "grad_norm": 0.0013594161719083786,
-      "learning_rate": 0.00019999994145439745,
-      "loss": 46.0,
-      "step": 4515
-    },
-    {
-      "epoch": 0.34527973698797715,
-      "grad_norm": 0.0015067590866237879,
-      "learning_rate": 0.00019999994142840322,
-      "loss": 46.0,
-      "step": 4516
-    },
-    {
-      "epoch": 0.34535619397136685,
-      "grad_norm": 0.0009392031352035701,
-      "learning_rate": 0.00019999994140240316,
-      "loss": 46.0,
-      "step": 4517
-    },
-    {
-      "epoch": 0.34543265095475656,
-      "grad_norm": 0.0007936739712022245,
-      "learning_rate": 0.00019999994137639737,
-      "loss": 46.0,
-      "step": 4518
-    },
-    {
-      "epoch": 0.3455091079381463,
-      "grad_norm": 0.0016129830619320273,
-      "learning_rate": 0.00019999994135038576,
-      "loss": 46.0,
-      "step": 4519
-    },
-    {
-      "epoch": 0.345585564921536,
-      "grad_norm": 0.0016986047849059105,
-      "learning_rate": 0.00019999994132436844,
-      "loss": 46.0,
-      "step": 4520
-    },
-    {
-      "epoch": 0.34566202190492573,
-      "grad_norm": 0.003739092266187072,
-      "learning_rate": 0.0001999999412983453,
-      "loss": 46.0,
-      "step": 4521
-    },
-    {
-      "epoch": 0.34573847888831544,
-      "grad_norm": 0.002710785483941436,
-      "learning_rate": 0.00019999994127231643,
-      "loss": 46.0,
-      "step": 4522
-    },
-    {
-      "epoch": 0.3458149358717052,
-      "grad_norm": 0.0009404800366610289,
-      "learning_rate": 0.00019999994124628176,
-      "loss": 46.0,
-      "step": 4523
-    },
-    {
-      "epoch": 0.3458913928550949,
-      "grad_norm": 0.0013227352173998952,
-      "learning_rate": 0.00019999994122024136,
-      "loss": 46.0,
-      "step": 4524
-    },
-    {
-      "epoch": 0.3459678498384846,
-      "grad_norm": 0.0023588594049215317,
-      "learning_rate": 0.00019999994119419517,
-      "loss": 46.0,
-      "step": 4525
-    },
-    {
-      "epoch": 0.34604430682187437,
-      "grad_norm": 0.0022572921589016914,
-      "learning_rate": 0.0001999999411681432,
-      "loss": 46.0,
-      "step": 4526
-    },
-    {
-      "epoch": 0.3461207638052641,
-      "grad_norm": 0.001386230462230742,
-      "learning_rate": 0.00019999994114208546,
-      "loss": 46.0,
-      "step": 4527
-    },
-    {
-      "epoch": 0.3461972207886538,
-      "grad_norm": 0.00137048470787704,
-      "learning_rate": 0.00019999994111602197,
-      "loss": 46.0,
-      "step": 4528
-    },
-    {
-      "epoch": 0.3462736777720435,
-      "grad_norm": 0.0008595773251727223,
-      "learning_rate": 0.00019999994108995268,
-      "loss": 46.0,
-      "step": 4529
-    },
-    {
-      "epoch": 0.34635013475543325,
-      "grad_norm": 0.0019864393398165703,
-      "learning_rate": 0.00019999994106387762,
-      "loss": 46.0,
-      "step": 4530
-    },
-    {
-      "epoch": 0.34642659173882295,
-      "grad_norm": 0.0019187870202586055,
-      "learning_rate": 0.00019999994103779683,
-      "loss": 46.0,
-      "step": 4531
-    },
-    {
-      "epoch": 0.34650304872221266,
-      "grad_norm": 0.000958569988142699,
-      "learning_rate": 0.00019999994101171025,
-      "loss": 46.0,
-      "step": 4532
-    },
-    {
-      "epoch": 0.34657950570560236,
-      "grad_norm": 0.004497377201914787,
-      "learning_rate": 0.00019999994098561787,
-      "loss": 46.0,
-      "step": 4533
-    },
-    {
-      "epoch": 0.3466559626889921,
-      "grad_norm": 0.001482845633290708,
-      "learning_rate": 0.00019999994095951977,
-      "loss": 46.0,
-      "step": 4534
-    },
-    {
-      "epoch": 0.34673241967238183,
-      "grad_norm": 0.00035967418807558715,
-      "learning_rate": 0.00019999994093341587,
-      "loss": 46.0,
-      "step": 4535
-    },
-    {
-      "epoch": 0.34680887665577154,
-      "grad_norm": 0.0036673317663371563,
-      "learning_rate": 0.00019999994090730622,
-      "loss": 46.0,
-      "step": 4536
-    },
-    {
-      "epoch": 0.34688533363916124,
-      "grad_norm": 0.0008209376828745008,
-      "learning_rate": 0.0001999999408811908,
-      "loss": 46.0,
-      "step": 4537
-    },
-    {
-      "epoch": 0.346961790622551,
-      "grad_norm": 0.0004369386297184974,
-      "learning_rate": 0.0001999999408550696,
-      "loss": 46.0,
-      "step": 4538
-    },
-    {
-      "epoch": 0.3470382476059407,
-      "grad_norm": 0.0014164528110995889,
-      "learning_rate": 0.00019999994082894263,
-      "loss": 46.0,
-      "step": 4539
-    },
-    {
-      "epoch": 0.3471147045893304,
-      "grad_norm": 0.0005142363952472806,
-      "learning_rate": 0.0001999999408028099,
-      "loss": 46.0,
-      "step": 4540
-    },
-    {
-      "epoch": 0.3471911615727202,
-      "grad_norm": 0.0010802133474498987,
-      "learning_rate": 0.00019999994077667138,
-      "loss": 46.0,
-      "step": 4541
-    },
-    {
-      "epoch": 0.3472676185561099,
-      "grad_norm": 0.0017264769412577152,
-      "learning_rate": 0.0001999999407505271,
-      "loss": 46.0,
-      "step": 4542
-    },
-    {
-      "epoch": 0.3473440755394996,
-      "grad_norm": 0.0007748965290375054,
-      "learning_rate": 0.00019999994072437708,
-      "loss": 46.0,
-      "step": 4543
-    },
-    {
-      "epoch": 0.3474205325228893,
-      "grad_norm": 0.0029790163971483707,
-      "learning_rate": 0.00019999994069822124,
-      "loss": 46.0,
-      "step": 4544
-    },
-    {
-      "epoch": 0.34749698950627905,
-      "grad_norm": 0.0010709143243730068,
-      "learning_rate": 0.00019999994067205966,
-      "loss": 46.0,
-      "step": 4545
-    },
-    {
-      "epoch": 0.34757344648966876,
-      "grad_norm": 0.002072112401947379,
-      "learning_rate": 0.00019999994064589228,
-      "loss": 46.0,
-      "step": 4546
-    },
-    {
-      "epoch": 0.34764990347305846,
-      "grad_norm": 0.0006146198138594627,
-      "learning_rate": 0.00019999994061971918,
-      "loss": 46.0,
-      "step": 4547
-    },
-    {
-      "epoch": 0.34772636045644817,
-      "grad_norm": 0.0006449267384596169,
-      "learning_rate": 0.00019999994059354028,
-      "loss": 46.0,
-      "step": 4548
-    },
-    {
-      "epoch": 0.34780281743983793,
-      "grad_norm": 0.0012200400233268738,
-      "learning_rate": 0.0001999999405673556,
-      "loss": 46.0,
-      "step": 4549
-    },
-    {
-      "epoch": 0.34787927442322764,
-      "grad_norm": 0.0011937841773033142,
-      "learning_rate": 0.00019999994054116518,
-      "loss": 46.0,
-      "step": 4550
-    },
-    {
-      "epoch": 0.34795573140661734,
-      "grad_norm": 0.004128133412450552,
-      "learning_rate": 0.000199999940514969,
-      "loss": 46.0,
-      "step": 4551
-    },
-    {
-      "epoch": 0.34803218839000705,
-      "grad_norm": 0.0008886297000572085,
-      "learning_rate": 0.000199999940488767,
-      "loss": 46.0,
-      "step": 4552
-    },
-    {
-      "epoch": 0.3481086453733968,
-      "grad_norm": 0.0006153713911771774,
-      "learning_rate": 0.00019999994046255928,
-      "loss": 46.0,
-      "step": 4553
-    },
-    {
-      "epoch": 0.3481851023567865,
-      "grad_norm": 0.0006943295011296868,
-      "learning_rate": 0.0001999999404363458,
-      "loss": 46.0,
-      "step": 4554
-    },
-    {
-      "epoch": 0.3482615593401762,
-      "grad_norm": 0.0017371104331687093,
-      "learning_rate": 0.00019999994041012647,
-      "loss": 46.0,
-      "step": 4555
-    },
-    {
-      "epoch": 0.348338016323566,
-      "grad_norm": 0.0005236382712610066,
-      "learning_rate": 0.00019999994038390144,
-      "loss": 46.0,
-      "step": 4556
-    },
-    {
-      "epoch": 0.3484144733069557,
-      "grad_norm": 0.0008119958220049739,
-      "learning_rate": 0.00019999994035767063,
-      "loss": 46.0,
-      "step": 4557
-    },
-    {
-      "epoch": 0.3484909302903454,
-      "grad_norm": 0.0017897101351991296,
-      "learning_rate": 0.00019999994033143402,
-      "loss": 46.0,
-      "step": 4558
-    },
-    {
-      "epoch": 0.3485673872737351,
-      "grad_norm": 0.0007238021353259683,
-      "learning_rate": 0.00019999994030519167,
-      "loss": 46.0,
-      "step": 4559
-    },
-    {
-      "epoch": 0.34864384425712486,
-      "grad_norm": 0.007205334957689047,
-      "learning_rate": 0.00019999994027894357,
-      "loss": 46.0,
-      "step": 4560
-    },
-    {
-      "epoch": 0.34872030124051456,
-      "grad_norm": 0.0003773700154852122,
-      "learning_rate": 0.00019999994025268967,
-      "loss": 46.0,
-      "step": 4561
-    },
-    {
-      "epoch": 0.34879675822390427,
-      "grad_norm": 0.001063156989403069,
-      "learning_rate": 0.00019999994022643,
-      "loss": 46.0,
-      "step": 4562
-    },
-    {
-      "epoch": 0.348873215207294,
-      "grad_norm": 0.0013440484181046486,
-      "learning_rate": 0.00019999994020016454,
-      "loss": 46.0,
-      "step": 4563
-    },
-    {
-      "epoch": 0.34894967219068374,
-      "grad_norm": 0.0021393182687461376,
-      "learning_rate": 0.00019999994017389335,
-      "loss": 46.0,
-      "step": 4564
-    },
-    {
-      "epoch": 0.34902612917407344,
-      "grad_norm": 0.000978855649009347,
-      "learning_rate": 0.00019999994014761638,
-      "loss": 46.0,
-      "step": 4565
-    },
-    {
-      "epoch": 0.34910258615746315,
-      "grad_norm": 0.0008459172677248716,
-      "learning_rate": 0.00019999994012133364,
-      "loss": 46.0,
-      "step": 4566
-    },
-    {
-      "epoch": 0.34917904314085285,
-      "grad_norm": 0.0005167752970010042,
-      "learning_rate": 0.0001999999400950451,
-      "loss": 46.0,
-      "step": 4567
-    },
-    {
-      "epoch": 0.3492555001242426,
-      "grad_norm": 0.0026665236800909042,
-      "learning_rate": 0.00019999994006875084,
-      "loss": 46.0,
-      "step": 4568
-    },
-    {
-      "epoch": 0.3493319571076323,
-      "grad_norm": 0.0009810024639591575,
-      "learning_rate": 0.00019999994004245078,
-      "loss": 46.0,
-      "step": 4569
-    },
-    {
-      "epoch": 0.349408414091022,
-      "grad_norm": 0.0012815716909244657,
-      "learning_rate": 0.00019999994001614497,
-      "loss": 46.0,
-      "step": 4570
-    },
-    {
-      "epoch": 0.3494848710744118,
-      "grad_norm": 0.003264606697484851,
-      "learning_rate": 0.00019999993998983336,
-      "loss": 46.0,
-      "step": 4571
-    },
-    {
-      "epoch": 0.3495613280578015,
-      "grad_norm": 0.003079640679061413,
-      "learning_rate": 0.000199999939963516,
-      "loss": 46.0,
-      "step": 4572
-    },
-    {
-      "epoch": 0.3496377850411912,
-      "grad_norm": 0.002139975316822529,
-      "learning_rate": 0.00019999993993719288,
-      "loss": 46.0,
-      "step": 4573
-    },
-    {
-      "epoch": 0.3497142420245809,
-      "grad_norm": 0.0009452769882045686,
-      "learning_rate": 0.00019999993991086398,
-      "loss": 46.0,
-      "step": 4574
-    },
-    {
-      "epoch": 0.34979069900797066,
-      "grad_norm": 0.0007609169115312397,
-      "learning_rate": 0.00019999993988452933,
-      "loss": 46.0,
-      "step": 4575
-    },
-    {
-      "epoch": 0.34986715599136037,
-      "grad_norm": 0.0016770473448559642,
-      "learning_rate": 0.00019999993985818888,
-      "loss": 46.0,
-      "step": 4576
-    },
-    {
-      "epoch": 0.3499436129747501,
-      "grad_norm": 0.0014944222057238221,
-      "learning_rate": 0.00019999993983184266,
-      "loss": 46.0,
-      "step": 4577
-    },
-    {
-      "epoch": 0.3500200699581398,
-      "grad_norm": 0.0004888473195023835,
-      "learning_rate": 0.00019999993980549067,
-      "loss": 46.0,
-      "step": 4578
-    },
-    {
-      "epoch": 0.35009652694152954,
-      "grad_norm": 0.0009985408978536725,
-      "learning_rate": 0.00019999993977913295,
-      "loss": 46.0,
-      "step": 4579
-    },
-    {
-      "epoch": 0.35017298392491925,
-      "grad_norm": 0.0012293473118916154,
-      "learning_rate": 0.00019999993975276944,
-      "loss": 46.0,
-      "step": 4580
-    },
-    {
-      "epoch": 0.35024944090830895,
-      "grad_norm": 0.0007471921271644533,
-      "learning_rate": 0.00019999993972640015,
-      "loss": 46.0,
-      "step": 4581
-    },
-    {
-      "epoch": 0.35032589789169866,
-      "grad_norm": 0.0026485580019652843,
-      "learning_rate": 0.0001999999397000251,
-      "loss": 46.0,
-      "step": 4582
-    },
-    {
-      "epoch": 0.3504023548750884,
-      "grad_norm": 0.0007579503580927849,
-      "learning_rate": 0.00019999993967364428,
-      "loss": 46.0,
-      "step": 4583
-    },
-    {
-      "epoch": 0.3504788118584781,
-      "grad_norm": 0.0006152355344966054,
-      "learning_rate": 0.00019999993964725767,
-      "loss": 46.0,
-      "step": 4584
-    },
-    {
-      "epoch": 0.35055526884186783,
-      "grad_norm": 0.0010391559917479753,
-      "learning_rate": 0.00019999993962086532,
-      "loss": 46.0,
-      "step": 4585
-    },
-    {
-      "epoch": 0.3506317258252576,
-      "grad_norm": 0.0006845281459391117,
-      "learning_rate": 0.00019999993959446717,
-      "loss": 46.0,
-      "step": 4586
-    },
-    {
-      "epoch": 0.3507081828086473,
-      "grad_norm": 0.001958251930773258,
-      "learning_rate": 0.00019999993956806327,
-      "loss": 46.0,
-      "step": 4587
-    },
-    {
-      "epoch": 0.350784639792037,
-      "grad_norm": 0.0031576522160321474,
-      "learning_rate": 0.0001999999395416536,
-      "loss": 46.0,
-      "step": 4588
-    },
-    {
-      "epoch": 0.3508610967754267,
-      "grad_norm": 0.001309097628109157,
-      "learning_rate": 0.00019999993951523817,
-      "loss": 46.0,
-      "step": 4589
-    },
-    {
-      "epoch": 0.35093755375881647,
-      "grad_norm": 0.001636011991649866,
-      "learning_rate": 0.00019999993948881698,
-      "loss": 46.0,
-      "step": 4590
-    },
-    {
-      "epoch": 0.3510140107422062,
-      "grad_norm": 0.0006237008492462337,
-      "learning_rate": 0.00019999993946238998,
-      "loss": 46.0,
-      "step": 4591
-    },
-    {
-      "epoch": 0.3510904677255959,
-      "grad_norm": 0.0030939914286136627,
-      "learning_rate": 0.00019999993943595724,
-      "loss": 46.0,
-      "step": 4592
-    },
-    {
-      "epoch": 0.3511669247089856,
-      "grad_norm": 0.0012296033091843128,
-      "learning_rate": 0.0001999999394095187,
-      "loss": 46.0,
-      "step": 4593
-    },
-    {
-      "epoch": 0.35124338169237534,
-      "grad_norm": 0.0014240751042962074,
-      "learning_rate": 0.00019999993938307441,
-      "loss": 46.0,
-      "step": 4594
-    },
-    {
-      "epoch": 0.35131983867576505,
-      "grad_norm": 0.007237753365188837,
-      "learning_rate": 0.00019999993935662435,
-      "loss": 46.0,
-      "step": 4595
-    },
-    {
-      "epoch": 0.35139629565915476,
-      "grad_norm": 0.0012333758641034365,
-      "learning_rate": 0.00019999993933016855,
-      "loss": 46.0,
-      "step": 4596
-    },
-    {
-      "epoch": 0.35147275264254446,
-      "grad_norm": 0.0021177921444177628,
-      "learning_rate": 0.00019999993930370694,
-      "loss": 46.0,
-      "step": 4597
-    },
-    {
-      "epoch": 0.3515492096259342,
-      "grad_norm": 0.0011861567618325353,
-      "learning_rate": 0.00019999993927723959,
-      "loss": 46.0,
-      "step": 4598
-    },
-    {
-      "epoch": 0.35162566660932393,
-      "grad_norm": 0.0008451088797301054,
-      "learning_rate": 0.00019999993925076646,
-      "loss": 46.0,
-      "step": 4599
-    },
-    {
-      "epoch": 0.35170212359271363,
-      "grad_norm": 0.00794740580022335,
-      "learning_rate": 0.00019999993922428756,
-      "loss": 46.0,
-      "step": 4600
-    },
-    {
-      "epoch": 0.3517785805761034,
-      "grad_norm": 0.0013230512849986553,
-      "learning_rate": 0.00019999993919780286,
-      "loss": 46.0,
-      "step": 4601
-    },
-    {
-      "epoch": 0.3518550375594931,
-      "grad_norm": 0.000470456900075078,
-      "learning_rate": 0.0001999999391713124,
-      "loss": 46.0,
-      "step": 4602
-    },
-    {
-      "epoch": 0.3519314945428828,
-      "grad_norm": 0.0005277706077322364,
-      "learning_rate": 0.00019999993914481622,
-      "loss": 46.0,
-      "step": 4603
-    },
-    {
-      "epoch": 0.3520079515262725,
-      "grad_norm": 0.0024399952962994576,
-      "learning_rate": 0.00019999993911831422,
-      "loss": 46.0,
-      "step": 4604
-    },
-    {
-      "epoch": 0.35208440850966227,
-      "grad_norm": 0.005916515365242958,
-      "learning_rate": 0.00019999993909180648,
-      "loss": 46.0,
-      "step": 4605
-    },
-    {
-      "epoch": 0.352160865493052,
-      "grad_norm": 0.0006022404413670301,
-      "learning_rate": 0.00019999993906529294,
-      "loss": 46.0,
-      "step": 4606
-    },
-    {
-      "epoch": 0.3522373224764417,
-      "grad_norm": 0.0038669996429234743,
-      "learning_rate": 0.00019999993903877368,
-      "loss": 46.0,
-      "step": 4607
-    },
-    {
-      "epoch": 0.3523137794598314,
-      "grad_norm": 0.0008730207337066531,
-      "learning_rate": 0.0001999999390122486,
-      "loss": 46.0,
-      "step": 4608
-    },
-    {
-      "epoch": 0.35239023644322115,
-      "grad_norm": 0.0013347804779186845,
-      "learning_rate": 0.00019999993898571776,
-      "loss": 46.0,
-      "step": 4609
-    },
-    {
-      "epoch": 0.35246669342661086,
-      "grad_norm": 0.0017347969114780426,
-      "learning_rate": 0.00019999993895918115,
-      "loss": 46.0,
-      "step": 4610
-    },
-    {
-      "epoch": 0.35254315041000056,
-      "grad_norm": 0.0007171679753810167,
-      "learning_rate": 0.0001999999389326388,
-      "loss": 46.0,
-      "step": 4611
-    },
-    {
-      "epoch": 0.35261960739339027,
-      "grad_norm": 0.0026676938869059086,
-      "learning_rate": 0.00019999993890609065,
-      "loss": 46.0,
-      "step": 4612
-    },
-    {
-      "epoch": 0.35269606437678,
-      "grad_norm": 0.0006391755887307227,
-      "learning_rate": 0.00019999993887953677,
-      "loss": 46.0,
-      "step": 4613
-    },
-    {
-      "epoch": 0.35277252136016973,
-      "grad_norm": 0.0006686265696771443,
-      "learning_rate": 0.0001999999388529771,
-      "loss": 46.0,
-      "step": 4614
-    },
-    {
-      "epoch": 0.35284897834355944,
-      "grad_norm": 0.0017134080408141017,
-      "learning_rate": 0.00019999993882641163,
-      "loss": 46.0,
-      "step": 4615
-    },
-    {
-      "epoch": 0.3529254353269492,
-      "grad_norm": 0.0013954247115179896,
-      "learning_rate": 0.00019999993879984043,
-      "loss": 46.0,
-      "step": 4616
-    },
-    {
-      "epoch": 0.3530018923103389,
-      "grad_norm": 0.0005169216310605407,
-      "learning_rate": 0.00019999993877326344,
-      "loss": 46.0,
-      "step": 4617
-    },
-    {
-      "epoch": 0.3530783492937286,
-      "grad_norm": 0.0005864306003786623,
-      "learning_rate": 0.00019999993874668067,
-      "loss": 46.0,
-      "step": 4618
-    },
-    {
-      "epoch": 0.3531548062771183,
-      "grad_norm": 0.0035213076043874025,
-      "learning_rate": 0.00019999993872009216,
-      "loss": 46.0,
-      "step": 4619
-    },
-    {
-      "epoch": 0.3532312632605081,
-      "grad_norm": 0.0015282656531780958,
-      "learning_rate": 0.00019999993869349787,
-      "loss": 46.0,
-      "step": 4620
-    },
-    {
-      "epoch": 0.3533077202438978,
-      "grad_norm": 0.007982725277543068,
-      "learning_rate": 0.0001999999386668978,
-      "loss": 46.0,
-      "step": 4621
-    },
-    {
-      "epoch": 0.3533841772272875,
-      "grad_norm": 0.0014074406353756785,
-      "learning_rate": 0.00019999993864029195,
-      "loss": 46.0,
-      "step": 4622
-    },
-    {
-      "epoch": 0.3534606342106772,
-      "grad_norm": 0.0016917784232646227,
-      "learning_rate": 0.00019999993861368037,
-      "loss": 46.0,
-      "step": 4623
-    },
-    {
-      "epoch": 0.35353709119406695,
-      "grad_norm": 0.001214608084410429,
-      "learning_rate": 0.00019999993858706302,
-      "loss": 46.0,
-      "step": 4624
-    },
-    {
-      "epoch": 0.35361354817745666,
-      "grad_norm": 0.0033316679764539003,
-      "learning_rate": 0.00019999993856043986,
-      "loss": 46.0,
-      "step": 4625
-    },
-    {
-      "epoch": 0.35369000516084637,
-      "grad_norm": 0.0008969350601546466,
-      "learning_rate": 0.00019999993853381094,
-      "loss": 46.0,
-      "step": 4626
-    },
-    {
-      "epoch": 0.3537664621442361,
-      "grad_norm": 0.002167478669434786,
-      "learning_rate": 0.00019999993850717626,
-      "loss": 46.0,
-      "step": 4627
-    },
-    {
-      "epoch": 0.35384291912762583,
-      "grad_norm": 0.0012287312420085073,
-      "learning_rate": 0.00019999993848053582,
-      "loss": 46.0,
-      "step": 4628
-    },
-    {
-      "epoch": 0.35391937611101554,
-      "grad_norm": 0.0006428689230233431,
-      "learning_rate": 0.00019999993845388957,
-      "loss": 46.0,
-      "step": 4629
-    },
-    {
-      "epoch": 0.35399583309440524,
-      "grad_norm": 0.0013685186859220266,
-      "learning_rate": 0.0001999999384272376,
-      "loss": 46.0,
-      "step": 4630
-    },
-    {
-      "epoch": 0.354072290077795,
-      "grad_norm": 0.0007062467630021274,
-      "learning_rate": 0.00019999993840057984,
-      "loss": 46.0,
-      "step": 4631
-    },
-    {
-      "epoch": 0.3541487470611847,
-      "grad_norm": 0.0008716791053302586,
-      "learning_rate": 0.0001999999383739163,
-      "loss": 46.0,
-      "step": 4632
-    },
-    {
-      "epoch": 0.3542252040445744,
-      "grad_norm": 0.0008582550217397511,
-      "learning_rate": 0.000199999938347247,
-      "loss": 46.0,
-      "step": 4633
-    },
-    {
-      "epoch": 0.3543016610279641,
-      "grad_norm": 0.001223880797624588,
-      "learning_rate": 0.00019999993832057195,
-      "loss": 46.0,
-      "step": 4634
-    },
-    {
-      "epoch": 0.3543781180113539,
-      "grad_norm": 0.000669918255880475,
-      "learning_rate": 0.00019999993829389112,
-      "loss": 46.0,
-      "step": 4635
-    },
-    {
-      "epoch": 0.3544545749947436,
-      "grad_norm": 0.005380552262067795,
-      "learning_rate": 0.0001999999382672045,
-      "loss": 46.0,
-      "step": 4636
-    },
-    {
-      "epoch": 0.3545310319781333,
-      "grad_norm": 0.002971152774989605,
-      "learning_rate": 0.00019999993824051213,
-      "loss": 46.0,
-      "step": 4637
-    },
-    {
-      "epoch": 0.354607488961523,
-      "grad_norm": 0.0016036460874602199,
-      "learning_rate": 0.000199999938213814,
-      "loss": 46.0,
-      "step": 4638
-    },
-    {
-      "epoch": 0.35468394594491276,
-      "grad_norm": 0.004605886060744524,
-      "learning_rate": 0.00019999993818711008,
-      "loss": 46.0,
-      "step": 4639
-    },
-    {
-      "epoch": 0.35476040292830247,
-      "grad_norm": 0.0025853635743260384,
-      "learning_rate": 0.00019999993816040038,
-      "loss": 46.0,
-      "step": 4640
-    },
-    {
-      "epoch": 0.35483685991169217,
-      "grad_norm": 0.0007215982768684626,
-      "learning_rate": 0.00019999993813368496,
-      "loss": 46.0,
-      "step": 4641
-    },
-    {
-      "epoch": 0.35491331689508193,
-      "grad_norm": 0.0013181599788367748,
-      "learning_rate": 0.00019999993810696374,
-      "loss": 46.0,
-      "step": 4642
-    },
-    {
-      "epoch": 0.35498977387847164,
-      "grad_norm": 0.0034462118055671453,
-      "learning_rate": 0.00019999993808023675,
-      "loss": 46.0,
-      "step": 4643
-    },
-    {
-      "epoch": 0.35506623086186134,
-      "grad_norm": 0.0016977038467302918,
-      "learning_rate": 0.00019999993805350398,
-      "loss": 46.0,
-      "step": 4644
-    },
-    {
-      "epoch": 0.35514268784525105,
-      "grad_norm": 0.0007073847227729857,
-      "learning_rate": 0.00019999993802676547,
-      "loss": 46.0,
-      "step": 4645
-    },
-    {
-      "epoch": 0.3552191448286408,
-      "grad_norm": 0.0018078909488394856,
-      "learning_rate": 0.00019999993800002118,
-      "loss": 46.0,
-      "step": 4646
-    },
-    {
-      "epoch": 0.3552956018120305,
-      "grad_norm": 0.0008559692651033401,
-      "learning_rate": 0.0001999999379732711,
-      "loss": 46.0,
-      "step": 4647
-    },
-    {
-      "epoch": 0.3553720587954202,
-      "grad_norm": 0.0005841772654093802,
-      "learning_rate": 0.00019999993794651526,
-      "loss": 46.0,
-      "step": 4648
-    },
-    {
-      "epoch": 0.3554485157788099,
-      "grad_norm": 0.0023644184693694115,
-      "learning_rate": 0.00019999993791975366,
-      "loss": 46.0,
-      "step": 4649
-    },
-    {
-      "epoch": 0.3555249727621997,
-      "grad_norm": 0.0011216261191293597,
-      "learning_rate": 0.00019999993789298625,
-      "loss": 46.0,
-      "step": 4650
-    },
-    {
-      "epoch": 0.3556014297455894,
-      "grad_norm": 0.0022794923279434443,
-      "learning_rate": 0.00019999993786621312,
-      "loss": 46.0,
-      "step": 4651
-    },
-    {
-      "epoch": 0.3556778867289791,
-      "grad_norm": 0.0017540324479341507,
-      "learning_rate": 0.0001999999378394342,
-      "loss": 46.0,
-      "step": 4652
-    },
-    {
-      "epoch": 0.3557543437123688,
-      "grad_norm": 0.0008257882436737418,
-      "learning_rate": 0.00019999993781264953,
-      "loss": 46.0,
-      "step": 4653
-    },
-    {
-      "epoch": 0.35583080069575856,
-      "grad_norm": 0.0005509591428562999,
-      "learning_rate": 0.00019999993778585905,
-      "loss": 46.0,
-      "step": 4654
-    },
-    {
-      "epoch": 0.35590725767914827,
-      "grad_norm": 0.0015455081593245268,
-      "learning_rate": 0.00019999993775906286,
-      "loss": 46.0,
-      "step": 4655
-    },
-    {
-      "epoch": 0.355983714662538,
-      "grad_norm": 0.0015083843609318137,
-      "learning_rate": 0.00019999993773226084,
-      "loss": 46.0,
-      "step": 4656
-    },
-    {
-      "epoch": 0.35606017164592774,
-      "grad_norm": 0.004210244864225388,
-      "learning_rate": 0.00019999993770545307,
-      "loss": 46.0,
-      "step": 4657
-    },
-    {
-      "epoch": 0.35613662862931744,
-      "grad_norm": 0.001926167169585824,
-      "learning_rate": 0.00019999993767863954,
-      "loss": 46.0,
-      "step": 4658
-    },
-    {
-      "epoch": 0.35621308561270715,
-      "grad_norm": 0.0009868277702480555,
-      "learning_rate": 0.00019999993765182022,
-      "loss": 46.0,
-      "step": 4659
-    },
-    {
-      "epoch": 0.35628954259609685,
-      "grad_norm": 0.00519395899027586,
-      "learning_rate": 0.00019999993762499516,
-      "loss": 46.0,
-      "step": 4660
-    },
-    {
-      "epoch": 0.3563659995794866,
-      "grad_norm": 0.001189759117551148,
-      "learning_rate": 0.00019999993759816433,
-      "loss": 46.0,
-      "step": 4661
-    },
-    {
-      "epoch": 0.3564424565628763,
-      "grad_norm": 0.0008527018944732845,
-      "learning_rate": 0.00019999993757132772,
-      "loss": 46.0,
-      "step": 4662
-    },
-    {
-      "epoch": 0.356518913546266,
-      "grad_norm": 0.000508213764987886,
-      "learning_rate": 0.00019999993754448535,
-      "loss": 46.0,
-      "step": 4663
-    },
-    {
-      "epoch": 0.35659537052965573,
-      "grad_norm": 0.0018172443378716707,
-      "learning_rate": 0.0001999999375176372,
-      "loss": 46.0,
-      "step": 4664
-    },
-    {
-      "epoch": 0.3566718275130455,
-      "grad_norm": 0.003778244135901332,
-      "learning_rate": 0.00019999993749078327,
-      "loss": 46.0,
-      "step": 4665
-    },
-    {
-      "epoch": 0.3567482844964352,
-      "grad_norm": 0.0016574617475271225,
-      "learning_rate": 0.0001999999374639236,
-      "loss": 46.0,
-      "step": 4666
-    },
-    {
-      "epoch": 0.3568247414798249,
-      "grad_norm": 0.001543763792142272,
-      "learning_rate": 0.00019999993743705812,
-      "loss": 46.0,
-      "step": 4667
-    },
-    {
-      "epoch": 0.3569011984632146,
-      "grad_norm": 0.0007495081517845392,
-      "learning_rate": 0.0001999999374101869,
-      "loss": 46.0,
-      "step": 4668
-    },
-    {
-      "epoch": 0.35697765544660437,
-      "grad_norm": 0.0035756188444793224,
-      "learning_rate": 0.0001999999373833099,
-      "loss": 46.0,
-      "step": 4669
-    },
-    {
-      "epoch": 0.3570541124299941,
-      "grad_norm": 0.0013363715261220932,
-      "learning_rate": 0.00019999993735642714,
-      "loss": 46.0,
-      "step": 4670
-    },
-    {
-      "epoch": 0.3571305694133838,
-      "grad_norm": 0.0007094431784935296,
-      "learning_rate": 0.0001999999373295386,
-      "loss": 46.0,
-      "step": 4671
-    },
-    {
-      "epoch": 0.35720702639677354,
-      "grad_norm": 0.0006934573757462204,
-      "learning_rate": 0.0001999999373026443,
-      "loss": 46.0,
-      "step": 4672
-    },
-    {
-      "epoch": 0.35728348338016325,
-      "grad_norm": 0.0009954001288861036,
-      "learning_rate": 0.0001999999372757442,
-      "loss": 46.0,
-      "step": 4673
-    },
-    {
-      "epoch": 0.35735994036355295,
-      "grad_norm": 0.001204869826324284,
-      "learning_rate": 0.00019999993724883838,
-      "loss": 46.0,
-      "step": 4674
-    },
-    {
-      "epoch": 0.35743639734694266,
-      "grad_norm": 0.0008512535714544356,
-      "learning_rate": 0.00019999993722192677,
-      "loss": 46.0,
-      "step": 4675
-    },
-    {
-      "epoch": 0.3575128543303324,
-      "grad_norm": 0.0007278133416548371,
-      "learning_rate": 0.0001999999371950094,
-      "loss": 46.0,
-      "step": 4676
-    },
-    {
-      "epoch": 0.3575893113137221,
-      "grad_norm": 0.0015090714441612363,
-      "learning_rate": 0.00019999993716808624,
-      "loss": 46.0,
-      "step": 4677
-    },
-    {
-      "epoch": 0.35766576829711183,
-      "grad_norm": 0.0011691406834870577,
-      "learning_rate": 0.0001999999371411573,
-      "loss": 46.0,
-      "step": 4678
-    },
-    {
-      "epoch": 0.35774222528050154,
-      "grad_norm": 0.002763924887403846,
-      "learning_rate": 0.00019999993711422261,
-      "loss": 46.0,
-      "step": 4679
-    },
-    {
-      "epoch": 0.3578186822638913,
-      "grad_norm": 0.001961057772859931,
-      "learning_rate": 0.00019999993708728214,
-      "loss": 46.0,
-      "step": 4680
-    },
-    {
-      "epoch": 0.357895139247281,
-      "grad_norm": 0.0006002667360007763,
-      "learning_rate": 0.00019999993706033592,
-      "loss": 46.0,
-      "step": 4681
-    },
-    {
-      "epoch": 0.3579715962306707,
-      "grad_norm": 0.0030982759781181812,
-      "learning_rate": 0.0001999999370333839,
-      "loss": 46.0,
-      "step": 4682
-    },
-    {
-      "epoch": 0.3580480532140604,
-      "grad_norm": 0.0033666484523564577,
-      "learning_rate": 0.00019999993700642614,
-      "loss": 46.0,
-      "step": 4683
-    },
-    {
-      "epoch": 0.3581245101974502,
-      "grad_norm": 0.00808589905500412,
-      "learning_rate": 0.00019999993697946263,
-      "loss": 46.0,
-      "step": 4684
-    },
-    {
-      "epoch": 0.3582009671808399,
-      "grad_norm": 0.002203570678830147,
-      "learning_rate": 0.00019999993695249334,
-      "loss": 46.0,
-      "step": 4685
-    },
-    {
-      "epoch": 0.3582774241642296,
-      "grad_norm": 0.0013378560543060303,
-      "learning_rate": 0.00019999993692551823,
-      "loss": 46.0,
-      "step": 4686
-    },
-    {
-      "epoch": 0.35835388114761935,
-      "grad_norm": 0.013525111600756645,
-      "learning_rate": 0.00019999993689853737,
-      "loss": 46.0,
-      "step": 4687
-    },
-    {
-      "epoch": 0.35843033813100905,
-      "grad_norm": 0.004850596655160189,
-      "learning_rate": 0.0001999999368715508,
-      "loss": 46.0,
-      "step": 4688
-    },
-    {
-      "epoch": 0.35850679511439876,
-      "grad_norm": 0.0011082233395427465,
-      "learning_rate": 0.00019999993684455839,
-      "loss": 46.0,
-      "step": 4689
-    },
-    {
-      "epoch": 0.35858325209778846,
-      "grad_norm": 0.0007109181024134159,
-      "learning_rate": 0.00019999993681756023,
-      "loss": 46.0,
-      "step": 4690
-    },
-    {
-      "epoch": 0.3586597090811782,
-      "grad_norm": 0.0014756397577002645,
-      "learning_rate": 0.0001999999367905563,
-      "loss": 46.0,
-      "step": 4691
-    },
-    {
-      "epoch": 0.35873616606456793,
-      "grad_norm": 0.0010675217490643263,
-      "learning_rate": 0.0001999999367635466,
-      "loss": 46.0,
-      "step": 4692
-    },
-    {
-      "epoch": 0.35881262304795764,
-      "grad_norm": 0.0005786485271528363,
-      "learning_rate": 0.00019999993673653114,
-      "loss": 46.0,
-      "step": 4693
-    },
-    {
-      "epoch": 0.35888908003134734,
-      "grad_norm": 0.0021037685219198465,
-      "learning_rate": 0.00019999993670950992,
-      "loss": 46.0,
-      "step": 4694
-    },
-    {
-      "epoch": 0.3589655370147371,
-      "grad_norm": 0.0035053843166679144,
-      "learning_rate": 0.0001999999366824829,
-      "loss": 46.0,
-      "step": 4695
-    },
-    {
-      "epoch": 0.3590419939981268,
-      "grad_norm": 0.0022866849321871996,
-      "learning_rate": 0.00019999993665545014,
-      "loss": 46.0,
-      "step": 4696
-    },
-    {
-      "epoch": 0.3591184509815165,
-      "grad_norm": 0.0010928100673481822,
-      "learning_rate": 0.0001999999366284116,
-      "loss": 46.0,
-      "step": 4697
-    },
-    {
-      "epoch": 0.3591949079649062,
-      "grad_norm": 0.0028621566016227007,
-      "learning_rate": 0.00019999993660136728,
-      "loss": 46.0,
-      "step": 4698
-    },
-    {
-      "epoch": 0.359271364948296,
-      "grad_norm": 0.0014850847655907273,
-      "learning_rate": 0.00019999993657431723,
-      "loss": 46.0,
-      "step": 4699
-    },
-    {
-      "epoch": 0.3593478219316857,
-      "grad_norm": 0.0007642853306606412,
-      "learning_rate": 0.00019999993654726134,
-      "loss": 46.0,
-      "step": 4700
-    },
-    {
-      "epoch": 0.3594242789150754,
-      "grad_norm": 0.002495459746569395,
-      "learning_rate": 0.00019999993652019974,
-      "loss": 46.0,
-      "step": 4701
-    },
-    {
-      "epoch": 0.35950073589846515,
-      "grad_norm": 0.0009976213332265615,
-      "learning_rate": 0.00019999993649313236,
-      "loss": 46.0,
-      "step": 4702
-    },
-    {
-      "epoch": 0.35957719288185486,
-      "grad_norm": 0.001287257531657815,
-      "learning_rate": 0.00019999993646605918,
-      "loss": 46.0,
-      "step": 4703
-    },
-    {
-      "epoch": 0.35965364986524456,
-      "grad_norm": 0.0011877205688506365,
-      "learning_rate": 0.00019999993643898028,
-      "loss": 46.0,
-      "step": 4704
-    },
-    {
-      "epoch": 0.35973010684863427,
-      "grad_norm": 0.001366660464555025,
-      "learning_rate": 0.00019999993641189558,
-      "loss": 46.0,
-      "step": 4705
-    },
-    {
-      "epoch": 0.35980656383202403,
-      "grad_norm": 0.0014981237472966313,
-      "learning_rate": 0.0001999999363848051,
-      "loss": 46.0,
-      "step": 4706
-    },
-    {
-      "epoch": 0.35988302081541373,
-      "grad_norm": 0.0028690313920378685,
-      "learning_rate": 0.0001999999363577089,
-      "loss": 46.0,
-      "step": 4707
-    },
-    {
-      "epoch": 0.35995947779880344,
-      "grad_norm": 0.0020689209923148155,
-      "learning_rate": 0.00019999993633060687,
-      "loss": 46.0,
-      "step": 4708
-    },
-    {
-      "epoch": 0.36003593478219315,
-      "grad_norm": 0.0010616728104650974,
-      "learning_rate": 0.00019999993630349908,
-      "loss": 46.0,
-      "step": 4709
-    },
-    {
-      "epoch": 0.3601123917655829,
-      "grad_norm": 0.0011188882635906339,
-      "learning_rate": 0.00019999993627638554,
-      "loss": 46.0,
-      "step": 4710
-    },
-    {
-      "epoch": 0.3601888487489726,
-      "grad_norm": 0.0011518624378368258,
-      "learning_rate": 0.00019999993624926623,
-      "loss": 46.0,
-      "step": 4711
-    },
-    {
-      "epoch": 0.3602653057323623,
-      "grad_norm": 0.001189077040180564,
-      "learning_rate": 0.00019999993622214115,
-      "loss": 46.0,
-      "step": 4712
-    },
-    {
-      "epoch": 0.360341762715752,
-      "grad_norm": 0.0008661917527206242,
-      "learning_rate": 0.0001999999361950103,
-      "loss": 46.0,
-      "step": 4713
-    },
-    {
-      "epoch": 0.3604182196991418,
-      "grad_norm": 0.0006917808786965907,
-      "learning_rate": 0.00019999993616787366,
-      "loss": 46.0,
-      "step": 4714
-    },
-    {
-      "epoch": 0.3604946766825315,
-      "grad_norm": 0.0020765475928783417,
-      "learning_rate": 0.00019999993614073128,
-      "loss": 46.0,
-      "step": 4715
-    },
-    {
-      "epoch": 0.3605711336659212,
-      "grad_norm": 0.0014501598197966814,
-      "learning_rate": 0.00019999993611358313,
-      "loss": 46.0,
-      "step": 4716
-    },
-    {
-      "epoch": 0.36064759064931096,
-      "grad_norm": 0.0012613646686077118,
-      "learning_rate": 0.0001999999360864292,
-      "loss": 46.0,
-      "step": 4717
-    },
-    {
-      "epoch": 0.36072404763270066,
-      "grad_norm": 0.0005951666971668601,
-      "learning_rate": 0.00019999993605926948,
-      "loss": 46.0,
-      "step": 4718
-    },
-    {
-      "epoch": 0.36080050461609037,
-      "grad_norm": 0.0025049233809113503,
-      "learning_rate": 0.000199999936032104,
-      "loss": 46.0,
-      "step": 4719
-    },
-    {
-      "epoch": 0.3608769615994801,
-      "grad_norm": 0.0011853475589305162,
-      "learning_rate": 0.0001999999360049328,
-      "loss": 46.0,
-      "step": 4720
-    },
-    {
-      "epoch": 0.36095341858286983,
-      "grad_norm": 0.0007207774324342608,
-      "learning_rate": 0.00019999993597775577,
-      "loss": 46.0,
-      "step": 4721
-    },
-    {
-      "epoch": 0.36102987556625954,
-      "grad_norm": 0.0006562497583217919,
-      "learning_rate": 0.000199999935950573,
-      "loss": 46.0,
-      "step": 4722
-    },
-    {
-      "epoch": 0.36110633254964924,
-      "grad_norm": 0.0017990752821788192,
-      "learning_rate": 0.00019999993592338444,
-      "loss": 46.0,
-      "step": 4723
-    },
-    {
-      "epoch": 0.36118278953303895,
-      "grad_norm": 0.001188163529150188,
-      "learning_rate": 0.00019999993589619016,
-      "loss": 46.0,
-      "step": 4724
-    },
-    {
-      "epoch": 0.3612592465164287,
-      "grad_norm": 0.0007358421571552753,
-      "learning_rate": 0.00019999993586899005,
-      "loss": 46.0,
-      "step": 4725
-    },
-    {
-      "epoch": 0.3613357034998184,
-      "grad_norm": 0.0007657320820726454,
-      "learning_rate": 0.00019999993584178421,
-      "loss": 46.0,
-      "step": 4726
-    },
-    {
-      "epoch": 0.3614121604832081,
-      "grad_norm": 0.0008777249022386968,
-      "learning_rate": 0.00019999993581457258,
-      "loss": 46.0,
-      "step": 4727
-    },
-    {
-      "epoch": 0.36148861746659783,
-      "grad_norm": 0.0034008976072072983,
-      "learning_rate": 0.00019999993578735515,
-      "loss": 46.0,
-      "step": 4728
-    },
-    {
-      "epoch": 0.3615650744499876,
-      "grad_norm": 0.0008502238779328763,
-      "learning_rate": 0.000199999935760132,
-      "loss": 46.0,
-      "step": 4729
-    },
-    {
-      "epoch": 0.3616415314333773,
-      "grad_norm": 0.0009893195237964392,
-      "learning_rate": 0.00019999993573290308,
-      "loss": 46.0,
-      "step": 4730
-    },
-    {
-      "epoch": 0.361717988416767,
-      "grad_norm": 0.0013479336630553007,
-      "learning_rate": 0.00019999993570566838,
-      "loss": 46.0,
-      "step": 4731
-    },
-    {
-      "epoch": 0.36179444540015676,
-      "grad_norm": 0.005582977551966906,
-      "learning_rate": 0.0001999999356784279,
-      "loss": 46.0,
-      "step": 4732
-    },
-    {
-      "epoch": 0.36187090238354647,
-      "grad_norm": 0.0009295704076066613,
-      "learning_rate": 0.00019999993565118164,
-      "loss": 46.0,
-      "step": 4733
-    },
-    {
-      "epoch": 0.36194735936693617,
-      "grad_norm": 0.0017829765565693378,
-      "learning_rate": 0.00019999993562392964,
-      "loss": 46.0,
-      "step": 4734
-    },
-    {
-      "epoch": 0.3620238163503259,
-      "grad_norm": 0.003539358964189887,
-      "learning_rate": 0.00019999993559667188,
-      "loss": 46.0,
-      "step": 4735
-    },
-    {
-      "epoch": 0.36210027333371564,
-      "grad_norm": 0.0009421908762305975,
-      "learning_rate": 0.00019999993556940832,
-      "loss": 46.0,
-      "step": 4736
-    },
-    {
-      "epoch": 0.36217673031710534,
-      "grad_norm": 0.0009855086682364345,
-      "learning_rate": 0.000199999935542139,
-      "loss": 46.0,
-      "step": 4737
-    },
-    {
-      "epoch": 0.36225318730049505,
-      "grad_norm": 0.0008309340919367969,
-      "learning_rate": 0.00019999993551486392,
-      "loss": 46.0,
-      "step": 4738
-    },
-    {
-      "epoch": 0.36232964428388476,
-      "grad_norm": 0.002351518953219056,
-      "learning_rate": 0.00019999993548758304,
-      "loss": 46.0,
-      "step": 4739
-    },
-    {
-      "epoch": 0.3624061012672745,
-      "grad_norm": 0.0032715625129640102,
-      "learning_rate": 0.00019999993546029643,
-      "loss": 46.0,
-      "step": 4740
-    },
-    {
-      "epoch": 0.3624825582506642,
-      "grad_norm": 0.05072646960616112,
-      "learning_rate": 0.00019999993543300403,
-      "loss": 46.0,
-      "step": 4741
-    },
-    {
-      "epoch": 0.3625590152340539,
-      "grad_norm": 0.002393092028796673,
-      "learning_rate": 0.00019999993540570588,
-      "loss": 46.0,
-      "step": 4742
-    },
-    {
-      "epoch": 0.3626354722174437,
-      "grad_norm": 0.0005407915450632572,
-      "learning_rate": 0.00019999993537840193,
-      "loss": 46.0,
-      "step": 4743
-    },
-    {
-      "epoch": 0.3627119292008334,
-      "grad_norm": 0.0010096813784912229,
-      "learning_rate": 0.00019999993535109223,
-      "loss": 46.0,
-      "step": 4744
-    },
-    {
-      "epoch": 0.3627883861842231,
-      "grad_norm": 0.001218139659613371,
-      "learning_rate": 0.00019999993532377676,
-      "loss": 46.0,
-      "step": 4745
-    },
-    {
-      "epoch": 0.3628648431676128,
-      "grad_norm": 0.0028415655251592398,
-      "learning_rate": 0.00019999993529645549,
-      "loss": 46.0,
-      "step": 4746
-    },
-    {
-      "epoch": 0.36294130015100257,
-      "grad_norm": 0.0008066268055699766,
-      "learning_rate": 0.0001999999352691285,
-      "loss": 46.0,
-      "step": 4747
-    },
-    {
-      "epoch": 0.36301775713439227,
-      "grad_norm": 0.0013411416439339519,
-      "learning_rate": 0.0001999999352417957,
-      "loss": 46.0,
-      "step": 4748
-    },
-    {
-      "epoch": 0.363094214117782,
-      "grad_norm": 0.0014214726397767663,
-      "learning_rate": 0.00019999993521445717,
-      "loss": 46.0,
-      "step": 4749
-    },
-    {
-      "epoch": 0.3631706711011717,
-      "grad_norm": 0.001000717980787158,
-      "learning_rate": 0.00019999993518711283,
-      "loss": 46.0,
-      "step": 4750
-    },
-    {
-      "epoch": 0.36324712808456144,
-      "grad_norm": 0.0011393342865630984,
-      "learning_rate": 0.00019999993515976274,
-      "loss": 46.0,
-      "step": 4751
-    },
-    {
-      "epoch": 0.36332358506795115,
-      "grad_norm": 0.0014566500904038548,
-      "learning_rate": 0.0001999999351324069,
-      "loss": 46.0,
-      "step": 4752
-    },
-    {
-      "epoch": 0.36340004205134085,
-      "grad_norm": 0.0006267392891459167,
-      "learning_rate": 0.00019999993510504526,
-      "loss": 46.0,
-      "step": 4753
-    },
-    {
-      "epoch": 0.36347649903473056,
-      "grad_norm": 0.0020487220026552677,
-      "learning_rate": 0.00019999993507767785,
-      "loss": 46.0,
-      "step": 4754
-    },
-    {
-      "epoch": 0.3635529560181203,
-      "grad_norm": 0.004392855800688267,
-      "learning_rate": 0.00019999993505030468,
-      "loss": 46.0,
-      "step": 4755
-    },
-    {
-      "epoch": 0.36362941300151,
-      "grad_norm": 0.0009140231995843351,
-      "learning_rate": 0.00019999993502292575,
-      "loss": 46.0,
-      "step": 4756
-    },
-    {
-      "epoch": 0.36370586998489973,
-      "grad_norm": 0.004033043514937162,
-      "learning_rate": 0.00019999993499554103,
-      "loss": 46.0,
-      "step": 4757
-    },
-    {
-      "epoch": 0.3637823269682895,
-      "grad_norm": 0.004864738322794437,
-      "learning_rate": 0.00019999993496815056,
-      "loss": 46.0,
-      "step": 4758
-    },
-    {
-      "epoch": 0.3638587839516792,
-      "grad_norm": 0.006420567631721497,
-      "learning_rate": 0.0001999999349407543,
-      "loss": 46.0,
-      "step": 4759
-    },
-    {
-      "epoch": 0.3639352409350689,
-      "grad_norm": 0.0005323308287188411,
-      "learning_rate": 0.0001999999349133523,
-      "loss": 46.0,
-      "step": 4760
-    },
-    {
-      "epoch": 0.3640116979184586,
-      "grad_norm": 0.0009971384424716234,
-      "learning_rate": 0.0001999999348859445,
-      "loss": 46.0,
-      "step": 4761
-    },
-    {
-      "epoch": 0.36408815490184837,
-      "grad_norm": 0.0005832620081491768,
-      "learning_rate": 0.00019999993485853094,
-      "loss": 46.0,
-      "step": 4762
-    },
-    {
-      "epoch": 0.3641646118852381,
-      "grad_norm": 0.002891265321522951,
-      "learning_rate": 0.00019999993483111163,
-      "loss": 46.0,
-      "step": 4763
-    },
-    {
-      "epoch": 0.3642410688686278,
-      "grad_norm": 0.005334098357707262,
-      "learning_rate": 0.00019999993480368652,
-      "loss": 46.0,
-      "step": 4764
-    },
-    {
-      "epoch": 0.3643175258520175,
-      "grad_norm": 0.0009453212842345238,
-      "learning_rate": 0.00019999993477625566,
-      "loss": 46.0,
-      "step": 4765
-    },
-    {
-      "epoch": 0.36439398283540725,
-      "grad_norm": 0.0016841174801811576,
-      "learning_rate": 0.000199999934748819,
-      "loss": 46.0,
-      "step": 4766
-    },
-    {
-      "epoch": 0.36447043981879695,
-      "grad_norm": 0.0011450157035142183,
-      "learning_rate": 0.0001999999347213766,
-      "loss": 46.0,
-      "step": 4767
-    },
-    {
-      "epoch": 0.36454689680218666,
-      "grad_norm": 0.0017618951387703419,
-      "learning_rate": 0.00019999993469392845,
-      "loss": 46.0,
-      "step": 4768
-    },
-    {
-      "epoch": 0.36462335378557637,
-      "grad_norm": 0.0007831259281374514,
-      "learning_rate": 0.00019999993466647453,
-      "loss": 46.0,
-      "step": 4769
-    },
-    {
-      "epoch": 0.3646998107689661,
-      "grad_norm": 0.005581744946539402,
-      "learning_rate": 0.0001999999346390148,
-      "loss": 46.0,
-      "step": 4770
-    },
-    {
-      "epoch": 0.36477626775235583,
-      "grad_norm": 0.0011554569937288761,
-      "learning_rate": 0.0001999999346115493,
-      "loss": 46.0,
-      "step": 4771
-    },
-    {
-      "epoch": 0.36485272473574554,
-      "grad_norm": 0.0028830962255597115,
-      "learning_rate": 0.0001999999345840781,
-      "loss": 46.0,
-      "step": 4772
-    },
-    {
-      "epoch": 0.3649291817191353,
-      "grad_norm": 0.0004013258730992675,
-      "learning_rate": 0.00019999993455660105,
-      "loss": 46.0,
-      "step": 4773
-    },
-    {
-      "epoch": 0.365005638702525,
-      "grad_norm": 0.0014745995867997408,
-      "learning_rate": 0.00019999993452911828,
-      "loss": 46.0,
-      "step": 4774
-    },
-    {
-      "epoch": 0.3650820956859147,
-      "grad_norm": 0.0008328726980835199,
-      "learning_rate": 0.0001999999345016297,
-      "loss": 46.0,
-      "step": 4775
-    },
-    {
-      "epoch": 0.3651585526693044,
-      "grad_norm": 0.002912326017394662,
-      "learning_rate": 0.00019999993447413535,
-      "loss": 46.0,
-      "step": 4776
-    },
-    {
-      "epoch": 0.3652350096526942,
-      "grad_norm": 0.0015257705235853791,
-      "learning_rate": 0.00019999993444663527,
-      "loss": 46.0,
-      "step": 4777
-    },
-    {
-      "epoch": 0.3653114666360839,
-      "grad_norm": 0.0015373674686998129,
-      "learning_rate": 0.00019999993441912941,
-      "loss": 46.0,
-      "step": 4778
-    },
-    {
-      "epoch": 0.3653879236194736,
-      "grad_norm": 0.0037738927640020847,
-      "learning_rate": 0.00019999993439161778,
-      "loss": 46.0,
-      "step": 4779
-    },
-    {
-      "epoch": 0.3654643806028633,
-      "grad_norm": 0.0019009530078619719,
-      "learning_rate": 0.00019999993436410035,
-      "loss": 46.0,
-      "step": 4780
-    },
-    {
-      "epoch": 0.36554083758625305,
-      "grad_norm": 0.0010089471470564604,
-      "learning_rate": 0.0001999999343365772,
-      "loss": 46.0,
-      "step": 4781
-    },
-    {
-      "epoch": 0.36561729456964276,
-      "grad_norm": 0.0012007777113467455,
-      "learning_rate": 0.00019999993430904825,
-      "loss": 46.0,
-      "step": 4782
-    },
-    {
-      "epoch": 0.36569375155303246,
-      "grad_norm": 0.0020197604317218065,
-      "learning_rate": 0.00019999993428151353,
-      "loss": 46.0,
-      "step": 4783
-    },
-    {
-      "epoch": 0.36577020853642217,
-      "grad_norm": 0.0022159023210406303,
-      "learning_rate": 0.00019999993425397306,
-      "loss": 46.0,
-      "step": 4784
-    },
-    {
-      "epoch": 0.36584666551981193,
-      "grad_norm": 0.005160761531442404,
-      "learning_rate": 0.0001999999342264268,
-      "loss": 46.0,
-      "step": 4785
-    },
-    {
-      "epoch": 0.36592312250320164,
-      "grad_norm": 0.0014021808747202158,
-      "learning_rate": 0.0001999999341988748,
-      "loss": 46.0,
-      "step": 4786
-    },
-    {
-      "epoch": 0.36599957948659134,
-      "grad_norm": 0.0013583634281530976,
-      "learning_rate": 0.00019999993417131698,
-      "loss": 46.0,
-      "step": 4787
-    },
-    {
-      "epoch": 0.3660760364699811,
-      "grad_norm": 0.014794185757637024,
-      "learning_rate": 0.00019999993414375342,
-      "loss": 46.0,
-      "step": 4788
-    },
-    {
-      "epoch": 0.3661524934533708,
-      "grad_norm": 0.0015073573449626565,
-      "learning_rate": 0.00019999993411618408,
-      "loss": 46.0,
-      "step": 4789
-    },
-    {
-      "epoch": 0.3662289504367605,
-      "grad_norm": 0.0014534660149365664,
-      "learning_rate": 0.000199999934088609,
-      "loss": 46.0,
-      "step": 4790
-    },
-    {
-      "epoch": 0.3663054074201502,
-      "grad_norm": 0.0008017596555873752,
-      "learning_rate": 0.00019999993406102812,
-      "loss": 46.0,
-      "step": 4791
-    },
-    {
-      "epoch": 0.36638186440354,
-      "grad_norm": 0.0021246725227683783,
-      "learning_rate": 0.00019999993403344149,
-      "loss": 46.0,
-      "step": 4792
-    },
-    {
-      "epoch": 0.3664583213869297,
-      "grad_norm": 0.0013715496752411127,
-      "learning_rate": 0.00019999993400584906,
-      "loss": 46.0,
-      "step": 4793
-    },
-    {
-      "epoch": 0.3665347783703194,
-      "grad_norm": 0.00122716068290174,
-      "learning_rate": 0.0001999999339782509,
-      "loss": 46.0,
-      "step": 4794
-    },
-    {
-      "epoch": 0.3666112353537091,
-      "grad_norm": 0.0005874124472029507,
-      "learning_rate": 0.00019999993395064696,
-      "loss": 46.0,
-      "step": 4795
-    },
-    {
-      "epoch": 0.36668769233709886,
-      "grad_norm": 0.0008420470403507352,
-      "learning_rate": 0.00019999993392303723,
-      "loss": 46.0,
-      "step": 4796
-    },
-    {
-      "epoch": 0.36676414932048856,
-      "grad_norm": 0.0027939307037740946,
-      "learning_rate": 0.00019999993389542176,
-      "loss": 46.0,
-      "step": 4797
-    },
-    {
-      "epoch": 0.36684060630387827,
-      "grad_norm": 0.0007634905050508678,
-      "learning_rate": 0.0001999999338678005,
-      "loss": 46.0,
-      "step": 4798
-    },
-    {
-      "epoch": 0.366917063287268,
-      "grad_norm": 0.00122022011782974,
-      "learning_rate": 0.00019999993384017345,
-      "loss": 46.0,
-      "step": 4799
-    },
-    {
-      "epoch": 0.36699352027065774,
-      "grad_norm": 0.0007961088558658957,
-      "learning_rate": 0.00019999993381254066,
-      "loss": 46.0,
-      "step": 4800
-    },
-    {
-      "epoch": 0.36706997725404744,
-      "grad_norm": 0.004239731468260288,
-      "learning_rate": 0.00019999993378490212,
-      "loss": 46.0,
-      "step": 4801
-    },
-    {
-      "epoch": 0.36714643423743715,
-      "grad_norm": 0.0014644551556557417,
-      "learning_rate": 0.0001999999337572578,
-      "loss": 46.0,
-      "step": 4802
-    },
-    {
-      "epoch": 0.3672228912208269,
-      "grad_norm": 0.0019061948405578732,
-      "learning_rate": 0.00019999993372960768,
-      "loss": 46.0,
-      "step": 4803
-    },
-    {
-      "epoch": 0.3672993482042166,
-      "grad_norm": 0.0006932874675840139,
-      "learning_rate": 0.0001999999337019518,
-      "loss": 46.0,
-      "step": 4804
-    },
-    {
-      "epoch": 0.3673758051876063,
-      "grad_norm": 0.0015004683518782258,
-      "learning_rate": 0.00019999993367429014,
-      "loss": 46.0,
-      "step": 4805
-    },
-    {
-      "epoch": 0.367452262170996,
-      "grad_norm": 0.0003630152787081897,
-      "learning_rate": 0.00019999993364662276,
-      "loss": 46.0,
-      "step": 4806
-    },
-    {
-      "epoch": 0.3675287191543858,
-      "grad_norm": 0.0006373982760123909,
-      "learning_rate": 0.00019999993361894959,
-      "loss": 46.0,
-      "step": 4807
-    },
-    {
-      "epoch": 0.3676051761377755,
-      "grad_norm": 0.0002189316292060539,
-      "learning_rate": 0.0001999999335912706,
-      "loss": 46.0,
-      "step": 4808
-    },
-    {
-      "epoch": 0.3676816331211652,
-      "grad_norm": 0.0023921134416013956,
-      "learning_rate": 0.00019999993356358591,
-      "loss": 46.0,
-      "step": 4809
-    },
-    {
-      "epoch": 0.3677580901045549,
-      "grad_norm": 0.0029419907368719578,
-      "learning_rate": 0.00019999993353589542,
-      "loss": 46.0,
-      "step": 4810
-    },
-    {
-      "epoch": 0.36783454708794466,
-      "grad_norm": 0.0006955838762223721,
-      "learning_rate": 0.00019999993350819915,
-      "loss": 46.0,
-      "step": 4811
-    },
-    {
-      "epoch": 0.36791100407133437,
-      "grad_norm": 0.0007645781734026968,
-      "learning_rate": 0.00019999993348049713,
-      "loss": 46.0,
-      "step": 4812
-    },
-    {
-      "epoch": 0.3679874610547241,
-      "grad_norm": 0.001570119522511959,
-      "learning_rate": 0.00019999993345278934,
-      "loss": 46.0,
-      "step": 4813
-    },
-    {
-      "epoch": 0.3680639180381138,
-      "grad_norm": 0.00292653264477849,
-      "learning_rate": 0.00019999993342507578,
-      "loss": 46.0,
-      "step": 4814
-    },
-    {
-      "epoch": 0.36814037502150354,
-      "grad_norm": 0.0005215071723796427,
-      "learning_rate": 0.00019999993339735644,
-      "loss": 46.0,
-      "step": 4815
-    },
-    {
-      "epoch": 0.36821683200489325,
-      "grad_norm": 0.000847869087010622,
-      "learning_rate": 0.00019999993336963133,
-      "loss": 46.0,
-      "step": 4816
-    },
-    {
-      "epoch": 0.36829328898828295,
-      "grad_norm": 0.004098173696547747,
-      "learning_rate": 0.00019999993334190042,
-      "loss": 46.0,
-      "step": 4817
-    },
-    {
-      "epoch": 0.3683697459716727,
-      "grad_norm": 0.0008783571538515389,
-      "learning_rate": 0.0001999999333141638,
-      "loss": 46.0,
-      "step": 4818
-    },
-    {
-      "epoch": 0.3684462029550624,
-      "grad_norm": 0.0007490054122172296,
-      "learning_rate": 0.0001999999332864214,
-      "loss": 46.0,
-      "step": 4819
-    },
-    {
-      "epoch": 0.3685226599384521,
-      "grad_norm": 0.0007143210968934,
-      "learning_rate": 0.0001999999332586732,
-      "loss": 46.0,
-      "step": 4820
-    },
-    {
-      "epoch": 0.36859911692184183,
-      "grad_norm": 0.003364644246175885,
-      "learning_rate": 0.00019999993323091927,
-      "loss": 46.0,
-      "step": 4821
-    },
-    {
-      "epoch": 0.3686755739052316,
-      "grad_norm": 0.001407845295034349,
-      "learning_rate": 0.00019999993320315952,
-      "loss": 46.0,
-      "step": 4822
-    },
-    {
-      "epoch": 0.3687520308886213,
-      "grad_norm": 0.0016733992379158735,
-      "learning_rate": 0.00019999993317539402,
-      "loss": 46.0,
-      "step": 4823
-    },
-    {
-      "epoch": 0.368828487872011,
-      "grad_norm": 0.00022612736211158335,
-      "learning_rate": 0.00019999993314762278,
-      "loss": 46.0,
-      "step": 4824
-    },
-    {
-      "epoch": 0.3689049448554007,
-      "grad_norm": 0.003249608911573887,
-      "learning_rate": 0.00019999993311984576,
-      "loss": 46.0,
-      "step": 4825
-    },
-    {
-      "epoch": 0.36898140183879047,
-      "grad_norm": 0.0008073369390331209,
-      "learning_rate": 0.00019999993309206295,
-      "loss": 46.0,
-      "step": 4826
-    },
-    {
-      "epoch": 0.3690578588221802,
-      "grad_norm": 0.0005007277941331267,
-      "learning_rate": 0.00019999993306427438,
-      "loss": 46.0,
-      "step": 4827
-    },
-    {
-      "epoch": 0.3691343158055699,
-      "grad_norm": 0.0014423951506614685,
-      "learning_rate": 0.00019999993303648005,
-      "loss": 46.0,
-      "step": 4828
-    },
-    {
-      "epoch": 0.3692107727889596,
-      "grad_norm": 0.0008734010043554008,
-      "learning_rate": 0.0001999999330086799,
-      "loss": 46.0,
-      "step": 4829
-    },
-    {
-      "epoch": 0.36928722977234935,
-      "grad_norm": 0.0010049734264612198,
-      "learning_rate": 0.00019999993298087406,
-      "loss": 46.0,
-      "step": 4830
-    },
-    {
-      "epoch": 0.36936368675573905,
-      "grad_norm": 0.00026240863371640444,
-      "learning_rate": 0.0001999999329530624,
-      "loss": 46.0,
-      "step": 4831
-    },
-    {
-      "epoch": 0.36944014373912876,
-      "grad_norm": 0.002178644761443138,
-      "learning_rate": 0.00019999993292524497,
-      "loss": 46.0,
-      "step": 4832
-    },
-    {
-      "epoch": 0.3695166007225185,
-      "grad_norm": 0.0006505553028546274,
-      "learning_rate": 0.0001999999328974218,
-      "loss": 46.0,
-      "step": 4833
-    },
-    {
-      "epoch": 0.3695930577059082,
-      "grad_norm": 0.0010698490077629685,
-      "learning_rate": 0.00019999993286959282,
-      "loss": 46.0,
-      "step": 4834
-    },
-    {
-      "epoch": 0.36966951468929793,
-      "grad_norm": 0.0014758114702999592,
-      "learning_rate": 0.0001999999328417581,
-      "loss": 46.0,
-      "step": 4835
-    },
-    {
-      "epoch": 0.36974597167268763,
-      "grad_norm": 0.0010098819620907307,
-      "learning_rate": 0.00019999993281391763,
-      "loss": 46.0,
-      "step": 4836
-    },
-    {
-      "epoch": 0.3698224286560774,
-      "grad_norm": 0.001681495807133615,
-      "learning_rate": 0.00019999993278607136,
-      "loss": 46.0,
-      "step": 4837
-    },
-    {
-      "epoch": 0.3698988856394671,
-      "grad_norm": 0.003068075980991125,
-      "learning_rate": 0.00019999993275821934,
-      "loss": 46.0,
-      "step": 4838
-    },
-    {
-      "epoch": 0.3699753426228568,
-      "grad_norm": 0.000599504099227488,
-      "learning_rate": 0.00019999993273036153,
-      "loss": 46.0,
-      "step": 4839
-    },
-    {
-      "epoch": 0.3700517996062465,
-      "grad_norm": 0.000580562453251332,
-      "learning_rate": 0.00019999993270249794,
-      "loss": 46.0,
-      "step": 4840
-    },
-    {
-      "epoch": 0.3701282565896363,
-      "grad_norm": 0.001212918316014111,
-      "learning_rate": 0.0001999999326746286,
-      "loss": 46.0,
-      "step": 4841
-    },
-    {
-      "epoch": 0.370204713573026,
-      "grad_norm": 0.0013744347961619496,
-      "learning_rate": 0.00019999993264675347,
-      "loss": 46.0,
-      "step": 4842
-    },
-    {
-      "epoch": 0.3702811705564157,
-      "grad_norm": 0.0016104707028716803,
-      "learning_rate": 0.0001999999326188726,
-      "loss": 46.0,
-      "step": 4843
-    },
-    {
-      "epoch": 0.3703576275398054,
-      "grad_norm": 0.0051866937428712845,
-      "learning_rate": 0.00019999993259098596,
-      "loss": 46.0,
-      "step": 4844
-    },
-    {
-      "epoch": 0.37043408452319515,
-      "grad_norm": 0.000993227818980813,
-      "learning_rate": 0.00019999993256309355,
-      "loss": 46.0,
-      "step": 4845
-    },
-    {
-      "epoch": 0.37051054150658486,
-      "grad_norm": 0.0013105841353535652,
-      "learning_rate": 0.00019999993253519532,
-      "loss": 46.0,
-      "step": 4846
-    },
-    {
-      "epoch": 0.37058699848997456,
-      "grad_norm": 0.0008735943119972944,
-      "learning_rate": 0.00019999993250729138,
-      "loss": 46.0,
-      "step": 4847
-    },
-    {
-      "epoch": 0.3706634554733643,
-      "grad_norm": 0.0008797018090263009,
-      "learning_rate": 0.00019999993247938168,
-      "loss": 46.0,
-      "step": 4848
-    },
-    {
-      "epoch": 0.37073991245675403,
-      "grad_norm": 0.0022008928935974836,
-      "learning_rate": 0.00019999993245146619,
-      "loss": 46.0,
-      "step": 4849
-    },
-    {
-      "epoch": 0.37081636944014373,
-      "grad_norm": 0.00046380568528547883,
-      "learning_rate": 0.00019999993242354492,
-      "loss": 46.0,
-      "step": 4850
-    },
-    {
-      "epoch": 0.37089282642353344,
-      "grad_norm": 0.002696666633710265,
-      "learning_rate": 0.00019999993239561787,
-      "loss": 46.0,
-      "step": 4851
-    },
-    {
-      "epoch": 0.3709692834069232,
-      "grad_norm": 0.002761688083410263,
-      "learning_rate": 0.00019999993236768506,
-      "loss": 46.0,
-      "step": 4852
-    },
-    {
-      "epoch": 0.3710457403903129,
-      "grad_norm": 0.0005381557857617736,
-      "learning_rate": 0.00019999993233974647,
-      "loss": 46.0,
-      "step": 4853
-    },
-    {
-      "epoch": 0.3711221973737026,
-      "grad_norm": 0.0019500568741932511,
-      "learning_rate": 0.0001999999323118021,
-      "loss": 46.0,
-      "step": 4854
-    },
-    {
-      "epoch": 0.3711986543570923,
-      "grad_norm": 0.012805874459445477,
-      "learning_rate": 0.00019999993228385203,
-      "loss": 46.0,
-      "step": 4855
-    },
-    {
-      "epoch": 0.3712751113404821,
-      "grad_norm": 0.0021064174361526966,
-      "learning_rate": 0.00019999993225589612,
-      "loss": 46.0,
-      "step": 4856
-    },
-    {
-      "epoch": 0.3713515683238718,
-      "grad_norm": 0.0019460027106106281,
-      "learning_rate": 0.0001999999322279345,
-      "loss": 46.0,
-      "step": 4857
-    },
-    {
-      "epoch": 0.3714280253072615,
-      "grad_norm": 0.0011560218408703804,
-      "learning_rate": 0.00019999993219996703,
-      "loss": 46.0,
-      "step": 4858
-    },
-    {
-      "epoch": 0.37150448229065125,
-      "grad_norm": 0.000566302624065429,
-      "learning_rate": 0.00019999993217199386,
-      "loss": 46.0,
-      "step": 4859
-    },
-    {
-      "epoch": 0.37158093927404096,
-      "grad_norm": 0.0006808766629546881,
-      "learning_rate": 0.0001999999321440149,
-      "loss": 46.0,
-      "step": 4860
-    },
-    {
-      "epoch": 0.37165739625743066,
-      "grad_norm": 0.0037500590551644564,
-      "learning_rate": 0.0001999999321160302,
-      "loss": 46.0,
-      "step": 4861
-    },
-    {
-      "epoch": 0.37173385324082037,
-      "grad_norm": 0.001662333612330258,
-      "learning_rate": 0.0001999999320880397,
-      "loss": 46.0,
-      "step": 4862
-    },
-    {
-      "epoch": 0.3718103102242101,
-      "grad_norm": 0.001028480939567089,
-      "learning_rate": 0.00019999993206004342,
-      "loss": 46.0,
-      "step": 4863
-    },
-    {
-      "epoch": 0.37188676720759983,
-      "grad_norm": 0.0006017939304001629,
-      "learning_rate": 0.00019999993203204138,
-      "loss": 46.0,
-      "step": 4864
-    },
-    {
-      "epoch": 0.37196322419098954,
-      "grad_norm": 0.0038125556893646717,
-      "learning_rate": 0.00019999993200403354,
-      "loss": 46.0,
-      "step": 4865
-    },
-    {
-      "epoch": 0.37203968117437924,
-      "grad_norm": 0.0035511080641299486,
-      "learning_rate": 0.00019999993197601998,
-      "loss": 46.0,
-      "step": 4866
-    },
-    {
-      "epoch": 0.372116138157769,
-      "grad_norm": 0.002554011531174183,
-      "learning_rate": 0.00019999993194800064,
-      "loss": 46.0,
-      "step": 4867
-    },
-    {
-      "epoch": 0.3721925951411587,
-      "grad_norm": 0.0005075350636616349,
-      "learning_rate": 0.0001999999319199755,
-      "loss": 46.0,
-      "step": 4868
-    },
-    {
-      "epoch": 0.3722690521245484,
-      "grad_norm": 0.001362888957373798,
-      "learning_rate": 0.00019999993189194463,
-      "loss": 46.0,
-      "step": 4869
-    },
-    {
-      "epoch": 0.3723455091079381,
-      "grad_norm": 0.005210305564105511,
-      "learning_rate": 0.00019999993186390797,
-      "loss": 46.0,
-      "step": 4870
-    },
-    {
-      "epoch": 0.3724219660913279,
-      "grad_norm": 0.003195004304870963,
-      "learning_rate": 0.00019999993183586554,
-      "loss": 46.0,
-      "step": 4871
-    },
-    {
-      "epoch": 0.3724984230747176,
-      "grad_norm": 0.0004451118584256619,
-      "learning_rate": 0.00019999993180781734,
-      "loss": 46.0,
-      "step": 4872
-    },
-    {
-      "epoch": 0.3725748800581073,
-      "grad_norm": 0.0003656750777736306,
-      "learning_rate": 0.0001999999317797634,
-      "loss": 46.0,
-      "step": 4873
-    },
-    {
-      "epoch": 0.37265133704149705,
-      "grad_norm": 0.0005771152791567147,
-      "learning_rate": 0.00019999993175170367,
-      "loss": 46.0,
-      "step": 4874
-    },
-    {
-      "epoch": 0.37272779402488676,
-      "grad_norm": 0.001187969814054668,
-      "learning_rate": 0.00019999993172363815,
-      "loss": 46.0,
-      "step": 4875
-    },
-    {
-      "epoch": 0.37280425100827647,
-      "grad_norm": 0.0004604406713042408,
-      "learning_rate": 0.00019999993169556688,
-      "loss": 46.0,
-      "step": 4876
-    },
-    {
-      "epoch": 0.37288070799166617,
-      "grad_norm": 0.00129945226944983,
-      "learning_rate": 0.00019999993166748981,
-      "loss": 46.0,
-      "step": 4877
-    },
-    {
-      "epoch": 0.37295716497505593,
-      "grad_norm": 0.0007609662134200335,
-      "learning_rate": 0.00019999993163940703,
-      "loss": 46.0,
-      "step": 4878
-    },
-    {
-      "epoch": 0.37303362195844564,
-      "grad_norm": 0.0035767932422459126,
-      "learning_rate": 0.00019999993161131844,
-      "loss": 46.0,
-      "step": 4879
-    },
-    {
-      "epoch": 0.37311007894183534,
-      "grad_norm": 0.0013194376369938254,
-      "learning_rate": 0.00019999993158322408,
-      "loss": 46.0,
-      "step": 4880
-    },
-    {
-      "epoch": 0.37318653592522505,
-      "grad_norm": 0.0028642769902944565,
-      "learning_rate": 0.00019999993155512397,
-      "loss": 46.0,
-      "step": 4881
-    },
-    {
-      "epoch": 0.3732629929086148,
-      "grad_norm": 0.0012514109257608652,
-      "learning_rate": 0.00019999993152701809,
-      "loss": 46.0,
-      "step": 4882
-    },
-    {
-      "epoch": 0.3733394498920045,
-      "grad_norm": 0.0011644541518762708,
-      "learning_rate": 0.0001999999314989064,
-      "loss": 46.0,
-      "step": 4883
-    },
-    {
-      "epoch": 0.3734159068753942,
-      "grad_norm": 0.00152407621499151,
-      "learning_rate": 0.000199999931470789,
-      "loss": 46.0,
-      "step": 4884
-    },
-    {
-      "epoch": 0.3734923638587839,
-      "grad_norm": 0.003909895662218332,
-      "learning_rate": 0.0001999999314426658,
-      "loss": 46.0,
-      "step": 4885
-    },
-    {
-      "epoch": 0.3735688208421737,
-      "grad_norm": 0.0018088618526235223,
-      "learning_rate": 0.00019999993141453683,
-      "loss": 46.0,
-      "step": 4886
-    },
-    {
-      "epoch": 0.3736452778255634,
-      "grad_norm": 0.003985913470387459,
-      "learning_rate": 0.0001999999313864021,
-      "loss": 46.0,
-      "step": 4887
-    },
-    {
-      "epoch": 0.3737217348089531,
-      "grad_norm": 0.01101826224476099,
-      "learning_rate": 0.00019999993135826158,
-      "loss": 46.0,
-      "step": 4888
-    },
-    {
-      "epoch": 0.37379819179234286,
-      "grad_norm": 0.0017482512630522251,
-      "learning_rate": 0.0001999999313301153,
-      "loss": 46.0,
-      "step": 4889
-    },
-    {
-      "epoch": 0.37387464877573257,
-      "grad_norm": 0.0036681180354207754,
-      "learning_rate": 0.00019999993130196325,
-      "loss": 46.0,
-      "step": 4890
-    },
-    {
-      "epoch": 0.37395110575912227,
-      "grad_norm": 0.0023561078123748302,
-      "learning_rate": 0.00019999993127380546,
-      "loss": 46.0,
-      "step": 4891
-    },
-    {
-      "epoch": 0.374027562742512,
-      "grad_norm": 0.0033678405452519655,
-      "learning_rate": 0.00019999993124564187,
-      "loss": 46.0,
-      "step": 4892
-    },
-    {
-      "epoch": 0.37410401972590174,
-      "grad_norm": 0.0008517990354448557,
-      "learning_rate": 0.0001999999312174725,
-      "loss": 46.0,
-      "step": 4893
-    },
-    {
-      "epoch": 0.37418047670929144,
-      "grad_norm": 0.0006096003344282508,
-      "learning_rate": 0.00019999993118929738,
-      "loss": 46.0,
-      "step": 4894
-    },
-    {
-      "epoch": 0.37425693369268115,
-      "grad_norm": 0.0004946135450154543,
-      "learning_rate": 0.0001999999311611165,
-      "loss": 46.0,
-      "step": 4895
-    },
-    {
-      "epoch": 0.37433339067607085,
-      "grad_norm": 0.001580708078108728,
-      "learning_rate": 0.00019999993113292984,
-      "loss": 46.0,
-      "step": 4896
-    },
-    {
-      "epoch": 0.3744098476594606,
-      "grad_norm": 0.0027992581017315388,
-      "learning_rate": 0.00019999993110473741,
-      "loss": 46.0,
-      "step": 4897
-    },
-    {
-      "epoch": 0.3744863046428503,
-      "grad_norm": 0.0019262167625129223,
-      "learning_rate": 0.0001999999310765392,
-      "loss": 46.0,
-      "step": 4898
-    },
-    {
-      "epoch": 0.37456276162624,
-      "grad_norm": 0.001027535181492567,
-      "learning_rate": 0.00019999993104833524,
-      "loss": 46.0,
-      "step": 4899
-    },
-    {
-      "epoch": 0.37463921860962973,
-      "grad_norm": 0.0014404486864805222,
-      "learning_rate": 0.0001999999310201255,
-      "loss": 46.0,
-      "step": 4900
-    },
-    {
-      "epoch": 0.3747156755930195,
-      "grad_norm": 0.003168889321386814,
-      "learning_rate": 0.00019999993099190997,
-      "loss": 46.0,
-      "step": 4901
-    },
-    {
-      "epoch": 0.3747921325764092,
-      "grad_norm": 0.0012133385753259063,
-      "learning_rate": 0.00019999993096368873,
-      "loss": 46.0,
-      "step": 4902
-    },
-    {
-      "epoch": 0.3748685895597989,
-      "grad_norm": 0.0016722440486773849,
-      "learning_rate": 0.00019999993093546166,
-      "loss": 46.0,
-      "step": 4903
-    },
-    {
-      "epoch": 0.37494504654318866,
-      "grad_norm": 0.0020575562957674265,
-      "learning_rate": 0.00019999993090722885,
-      "loss": 46.0,
-      "step": 4904
-    },
-    {
-      "epoch": 0.37502150352657837,
-      "grad_norm": 0.0008931782795116305,
-      "learning_rate": 0.00019999993087899026,
-      "loss": 46.0,
-      "step": 4905
-    },
-    {
-      "epoch": 0.3750979605099681,
-      "grad_norm": 0.0012393470387905836,
-      "learning_rate": 0.0001999999308507459,
-      "loss": 46.0,
-      "step": 4906
-    },
-    {
-      "epoch": 0.3751744174933578,
-      "grad_norm": 0.0017042583785951138,
-      "learning_rate": 0.00019999993082249576,
-      "loss": 46.0,
-      "step": 4907
-    },
-    {
-      "epoch": 0.37525087447674754,
-      "grad_norm": 0.004278429783880711,
-      "learning_rate": 0.00019999993079423985,
-      "loss": 46.0,
-      "step": 4908
-    },
-    {
-      "epoch": 0.37532733146013725,
-      "grad_norm": 0.008249441161751747,
-      "learning_rate": 0.0001999999307659782,
-      "loss": 46.0,
-      "step": 4909
-    },
-    {
-      "epoch": 0.37540378844352695,
-      "grad_norm": 0.002022908767685294,
-      "learning_rate": 0.00019999993073771077,
-      "loss": 46.0,
-      "step": 4910
-    },
-    {
-      "epoch": 0.37548024542691666,
-      "grad_norm": 0.0018369735917076468,
-      "learning_rate": 0.00019999993070943757,
-      "loss": 46.0,
-      "step": 4911
-    },
-    {
-      "epoch": 0.3755567024103064,
-      "grad_norm": 0.0015796294901520014,
-      "learning_rate": 0.0001999999306811586,
-      "loss": 46.0,
-      "step": 4912
-    },
-    {
-      "epoch": 0.3756331593936961,
-      "grad_norm": 0.0009233673335984349,
-      "learning_rate": 0.00019999993065287385,
-      "loss": 46.0,
-      "step": 4913
-    },
-    {
-      "epoch": 0.37570961637708583,
-      "grad_norm": 0.001429198426194489,
-      "learning_rate": 0.00019999993062458336,
-      "loss": 46.0,
-      "step": 4914
-    },
-    {
-      "epoch": 0.37578607336047554,
-      "grad_norm": 0.002765101147815585,
-      "learning_rate": 0.00019999993059628706,
-      "loss": 46.0,
-      "step": 4915
-    },
-    {
-      "epoch": 0.3758625303438653,
-      "grad_norm": 0.0012760492973029613,
-      "learning_rate": 0.00019999993056798502,
-      "loss": 46.0,
-      "step": 4916
-    },
-    {
-      "epoch": 0.375938987327255,
-      "grad_norm": 0.0011727642267942429,
-      "learning_rate": 0.00019999993053967718,
-      "loss": 46.0,
-      "step": 4917
-    },
-    {
-      "epoch": 0.3760154443106447,
-      "grad_norm": 0.0014468106674030423,
-      "learning_rate": 0.00019999993051136362,
-      "loss": 46.0,
-      "step": 4918
-    },
-    {
-      "epoch": 0.37609190129403447,
-      "grad_norm": 0.0010393660049885511,
-      "learning_rate": 0.00019999993048304423,
-      "loss": 46.0,
-      "step": 4919
-    },
-    {
-      "epoch": 0.3761683582774242,
-      "grad_norm": 0.0011485849972814322,
-      "learning_rate": 0.0001999999304547191,
-      "loss": 46.0,
-      "step": 4920
-    },
-    {
-      "epoch": 0.3762448152608139,
-      "grad_norm": 0.002423313446342945,
-      "learning_rate": 0.0001999999304263882,
-      "loss": 46.0,
-      "step": 4921
-    },
-    {
-      "epoch": 0.3763212722442036,
-      "grad_norm": 0.004466170445084572,
-      "learning_rate": 0.00019999993039805154,
-      "loss": 46.0,
-      "step": 4922
-    },
-    {
-      "epoch": 0.37639772922759335,
-      "grad_norm": 0.000734652450773865,
-      "learning_rate": 0.0001999999303697091,
-      "loss": 46.0,
-      "step": 4923
-    },
-    {
-      "epoch": 0.37647418621098305,
-      "grad_norm": 0.000604241737164557,
-      "learning_rate": 0.0001999999303413609,
-      "loss": 46.0,
-      "step": 4924
-    },
-    {
-      "epoch": 0.37655064319437276,
-      "grad_norm": 0.035608306527137756,
-      "learning_rate": 0.00019999993031300694,
-      "loss": 46.0,
-      "step": 4925
-    },
-    {
-      "epoch": 0.37662710017776246,
-      "grad_norm": 0.009268702939152718,
-      "learning_rate": 0.0001999999302846472,
-      "loss": 46.0,
-      "step": 4926
-    },
-    {
-      "epoch": 0.3767035571611522,
-      "grad_norm": 0.0009004255407489836,
-      "learning_rate": 0.00019999993025628164,
-      "loss": 46.0,
-      "step": 4927
-    },
-    {
-      "epoch": 0.37678001414454193,
-      "grad_norm": 0.0007173077901825309,
-      "learning_rate": 0.0001999999302279104,
-      "loss": 46.0,
-      "step": 4928
-    },
-    {
-      "epoch": 0.37685647112793164,
-      "grad_norm": 0.005671788938343525,
-      "learning_rate": 0.00019999993019953333,
-      "loss": 46.0,
-      "step": 4929
-    },
-    {
-      "epoch": 0.37693292811132134,
-      "grad_norm": 0.0031418928410857916,
-      "learning_rate": 0.00019999993017115052,
-      "loss": 46.0,
-      "step": 4930
-    },
-    {
-      "epoch": 0.3770093850947111,
-      "grad_norm": 0.0012169423280283809,
-      "learning_rate": 0.00019999993014276193,
-      "loss": 46.0,
-      "step": 4931
-    },
-    {
-      "epoch": 0.3770858420781008,
-      "grad_norm": 0.0004836243751924485,
-      "learning_rate": 0.00019999993011436755,
-      "loss": 46.0,
-      "step": 4932
-    },
-    {
-      "epoch": 0.3771622990614905,
-      "grad_norm": 0.0009299837402068079,
-      "learning_rate": 0.0001999999300859674,
-      "loss": 46.0,
-      "step": 4933
-    },
-    {
-      "epoch": 0.3772387560448803,
-      "grad_norm": 0.0011206520721316338,
-      "learning_rate": 0.00019999993005756153,
-      "loss": 46.0,
-      "step": 4934
-    },
-    {
-      "epoch": 0.37731521302827,
-      "grad_norm": 0.0023651737719774246,
-      "learning_rate": 0.00019999993002914985,
-      "loss": 46.0,
-      "step": 4935
-    },
-    {
-      "epoch": 0.3773916700116597,
-      "grad_norm": 0.004383682273328304,
-      "learning_rate": 0.0001999999300007324,
-      "loss": 46.0,
-      "step": 4936
-    },
-    {
-      "epoch": 0.3774681269950494,
-      "grad_norm": 0.0016093684826046228,
-      "learning_rate": 0.0001999999299723092,
-      "loss": 46.0,
-      "step": 4937
-    },
-    {
-      "epoch": 0.37754458397843915,
-      "grad_norm": 0.001072571729309857,
-      "learning_rate": 0.00019999992994388022,
-      "loss": 46.0,
-      "step": 4938
-    },
-    {
-      "epoch": 0.37762104096182886,
-      "grad_norm": 0.0010431492701172829,
-      "learning_rate": 0.00019999992991544548,
-      "loss": 46.0,
-      "step": 4939
-    },
-    {
-      "epoch": 0.37769749794521856,
-      "grad_norm": 0.0010073445737361908,
-      "learning_rate": 0.00019999992988700496,
-      "loss": 46.0,
-      "step": 4940
-    },
-    {
-      "epoch": 0.37777395492860827,
-      "grad_norm": 0.0012456390541046858,
-      "learning_rate": 0.00019999992985855866,
-      "loss": 46.0,
-      "step": 4941
-    },
-    {
-      "epoch": 0.37785041191199803,
-      "grad_norm": 0.0005301701021380723,
-      "learning_rate": 0.0001999999298301066,
-      "loss": 46.0,
-      "step": 4942
-    },
-    {
-      "epoch": 0.37792686889538774,
-      "grad_norm": 0.0023952494375407696,
-      "learning_rate": 0.00019999992980164878,
-      "loss": 46.0,
-      "step": 4943
-    },
-    {
-      "epoch": 0.37800332587877744,
-      "grad_norm": 0.0035832435823976994,
-      "learning_rate": 0.0001999999297731852,
-      "loss": 46.0,
-      "step": 4944
-    },
-    {
-      "epoch": 0.37807978286216715,
-      "grad_norm": 0.0015766293508931994,
-      "learning_rate": 0.00019999992974471583,
-      "loss": 46.0,
-      "step": 4945
-    },
-    {
-      "epoch": 0.3781562398455569,
-      "grad_norm": 0.0011472138576209545,
-      "learning_rate": 0.0001999999297162407,
-      "loss": 46.0,
-      "step": 4946
-    },
-    {
-      "epoch": 0.3782326968289466,
-      "grad_norm": 0.003403461305424571,
-      "learning_rate": 0.0001999999296877598,
-      "loss": 46.0,
-      "step": 4947
-    },
-    {
-      "epoch": 0.3783091538123363,
-      "grad_norm": 0.003906186204403639,
-      "learning_rate": 0.00019999992965927311,
-      "loss": 46.0,
-      "step": 4948
-    },
-    {
-      "epoch": 0.3783856107957261,
-      "grad_norm": 0.0017572161741554737,
-      "learning_rate": 0.00019999992963078066,
-      "loss": 46.0,
-      "step": 4949
-    },
-    {
-      "epoch": 0.3784620677791158,
-      "grad_norm": 0.0020414418540894985,
-      "learning_rate": 0.00019999992960228246,
-      "loss": 46.0,
-      "step": 4950
-    },
-    {
-      "epoch": 0.3785385247625055,
-      "grad_norm": 0.001186339184641838,
-      "learning_rate": 0.0001999999295737785,
-      "loss": 46.0,
-      "step": 4951
-    },
-    {
-      "epoch": 0.3786149817458952,
-      "grad_norm": 0.007186608854681253,
-      "learning_rate": 0.0001999999295452687,
-      "loss": 46.0,
-      "step": 4952
-    },
-    {
-      "epoch": 0.37869143872928496,
-      "grad_norm": 0.0008579810382798314,
-      "learning_rate": 0.00019999992951675322,
-      "loss": 46.0,
-      "step": 4953
-    },
-    {
-      "epoch": 0.37876789571267466,
-      "grad_norm": 0.0020101191475987434,
-      "learning_rate": 0.00019999992948823193,
-      "loss": 46.0,
-      "step": 4954
-    },
-    {
-      "epoch": 0.37884435269606437,
-      "grad_norm": 0.0015061977319419384,
-      "learning_rate": 0.00019999992945970486,
-      "loss": 46.0,
-      "step": 4955
-    },
-    {
-      "epoch": 0.3789208096794541,
-      "grad_norm": 0.0013935064198449254,
-      "learning_rate": 0.00019999992943117202,
-      "loss": 46.0,
-      "step": 4956
-    },
-    {
-      "epoch": 0.37899726666284383,
-      "grad_norm": 0.0010541134979575872,
-      "learning_rate": 0.00019999992940263343,
-      "loss": 46.0,
-      "step": 4957
-    },
-    {
-      "epoch": 0.37907372364623354,
-      "grad_norm": 0.0024583195336163044,
-      "learning_rate": 0.00019999992937408907,
-      "loss": 46.0,
-      "step": 4958
-    },
-    {
-      "epoch": 0.37915018062962325,
-      "grad_norm": 0.000910849601496011,
-      "learning_rate": 0.00019999992934553894,
-      "loss": 46.0,
-      "step": 4959
-    },
-    {
-      "epoch": 0.37922663761301295,
-      "grad_norm": 0.0007606238941662014,
-      "learning_rate": 0.000199999929316983,
-      "loss": 46.0,
-      "step": 4960
-    },
-    {
-      "epoch": 0.3793030945964027,
-      "grad_norm": 0.00043626653496176004,
-      "learning_rate": 0.00019999992928842133,
-      "loss": 46.0,
-      "step": 4961
-    },
-    {
-      "epoch": 0.3793795515797924,
-      "grad_norm": 0.0008070764597505331,
-      "learning_rate": 0.0001999999292598539,
-      "loss": 46.0,
-      "step": 4962
-    },
-    {
-      "epoch": 0.3794560085631821,
-      "grad_norm": 0.0007010205881670117,
-      "learning_rate": 0.00019999992923128067,
-      "loss": 46.0,
-      "step": 4963
-    },
-    {
-      "epoch": 0.3795324655465719,
-      "grad_norm": 0.002104604383930564,
-      "learning_rate": 0.0001999999292027017,
-      "loss": 46.0,
-      "step": 4964
-    },
-    {
-      "epoch": 0.3796089225299616,
-      "grad_norm": 0.0011542040156200528,
-      "learning_rate": 0.00019999992917411693,
-      "loss": 46.0,
-      "step": 4965
-    },
-    {
-      "epoch": 0.3796853795133513,
-      "grad_norm": 0.00029347772942855954,
-      "learning_rate": 0.00019999992914552644,
-      "loss": 46.0,
-      "step": 4966
-    },
-    {
-      "epoch": 0.379761836496741,
-      "grad_norm": 0.000831316749099642,
-      "learning_rate": 0.00019999992911693014,
-      "loss": 46.0,
-      "step": 4967
-    },
-    {
-      "epoch": 0.37983829348013076,
-      "grad_norm": 0.002270141616463661,
-      "learning_rate": 0.00019999992908832805,
-      "loss": 46.0,
-      "step": 4968
-    },
-    {
-      "epoch": 0.37991475046352047,
-      "grad_norm": 0.007392963860183954,
-      "learning_rate": 0.00019999992905972024,
-      "loss": 46.0,
-      "step": 4969
-    },
-    {
-      "epoch": 0.3799912074469102,
-      "grad_norm": 0.0010177148506045341,
-      "learning_rate": 0.00019999992903110662,
-      "loss": 46.0,
-      "step": 4970
-    },
-    {
-      "epoch": 0.3800676644302999,
-      "grad_norm": 0.0010335877304896712,
-      "learning_rate": 0.00019999992900248726,
-      "loss": 46.0,
-      "step": 4971
-    },
-    {
-      "epoch": 0.38014412141368964,
-      "grad_norm": 0.0015258160419762135,
-      "learning_rate": 0.00019999992897386213,
-      "loss": 46.0,
-      "step": 4972
-    },
-    {
-      "epoch": 0.38022057839707935,
-      "grad_norm": 0.0016457870369777083,
-      "learning_rate": 0.0001999999289452312,
-      "loss": 46.0,
-      "step": 4973
-    },
-    {
-      "epoch": 0.38029703538046905,
-      "grad_norm": 0.002090898808091879,
-      "learning_rate": 0.00019999992891659454,
-      "loss": 46.0,
-      "step": 4974
-    },
-    {
-      "epoch": 0.3803734923638588,
-      "grad_norm": 0.006494347937405109,
-      "learning_rate": 0.00019999992888795206,
-      "loss": 46.0,
-      "step": 4975
-    },
-    {
-      "epoch": 0.3804499493472485,
-      "grad_norm": 0.001054001972079277,
-      "learning_rate": 0.00019999992885930387,
-      "loss": 46.0,
-      "step": 4976
-    },
-    {
-      "epoch": 0.3805264063306382,
-      "grad_norm": 0.002643430605530739,
-      "learning_rate": 0.0001999999288306499,
-      "loss": 46.0,
-      "step": 4977
-    },
-    {
-      "epoch": 0.38060286331402793,
-      "grad_norm": 0.0019271423807367682,
-      "learning_rate": 0.00019999992880199012,
-      "loss": 46.0,
-      "step": 4978
-    },
-    {
-      "epoch": 0.3806793202974177,
-      "grad_norm": 0.0014187066117301583,
-      "learning_rate": 0.0001999999287733246,
-      "loss": 46.0,
-      "step": 4979
-    },
-    {
-      "epoch": 0.3807557772808074,
-      "grad_norm": 0.000565417343750596,
-      "learning_rate": 0.0001999999287446533,
-      "loss": 46.0,
-      "step": 4980
-    },
-    {
-      "epoch": 0.3808322342641971,
-      "grad_norm": 0.0012101131724193692,
-      "learning_rate": 0.00019999992871597624,
-      "loss": 46.0,
-      "step": 4981
-    },
-    {
-      "epoch": 0.3809086912475868,
-      "grad_norm": 0.0024158800952136517,
-      "learning_rate": 0.0001999999286872934,
-      "loss": 46.0,
-      "step": 4982
-    },
-    {
-      "epoch": 0.38098514823097657,
-      "grad_norm": 0.00048457173397764564,
-      "learning_rate": 0.0001999999286586048,
-      "loss": 46.0,
-      "step": 4983
-    },
-    {
-      "epoch": 0.3810616052143663,
-      "grad_norm": 0.003602725686505437,
-      "learning_rate": 0.00019999992862991043,
-      "loss": 46.0,
-      "step": 4984
-    },
-    {
-      "epoch": 0.381138062197756,
-      "grad_norm": 0.0021635626908391714,
-      "learning_rate": 0.0001999999286012103,
-      "loss": 46.0,
-      "step": 4985
-    },
-    {
-      "epoch": 0.3812145191811457,
-      "grad_norm": 0.002915341407060623,
-      "learning_rate": 0.00019999992857250436,
-      "loss": 46.0,
-      "step": 4986
-    },
-    {
-      "epoch": 0.38129097616453544,
-      "grad_norm": 0.004720613826066256,
-      "learning_rate": 0.0001999999285437927,
-      "loss": 46.0,
-      "step": 4987
-    },
-    {
-      "epoch": 0.38136743314792515,
-      "grad_norm": 0.0014432631433010101,
-      "learning_rate": 0.00019999992851507523,
-      "loss": 46.0,
-      "step": 4988
-    },
-    {
-      "epoch": 0.38144389013131486,
-      "grad_norm": 0.0010512990411370993,
-      "learning_rate": 0.00019999992848635203,
-      "loss": 46.0,
-      "step": 4989
-    },
-    {
-      "epoch": 0.3815203471147046,
-      "grad_norm": 0.0019252338679507375,
-      "learning_rate": 0.000199999928457623,
-      "loss": 46.0,
-      "step": 4990
-    },
-    {
-      "epoch": 0.3815968040980943,
-      "grad_norm": 0.01550122071057558,
-      "learning_rate": 0.00019999992842888826,
-      "loss": 46.0,
-      "step": 4991
-    },
-    {
-      "epoch": 0.381673261081484,
-      "grad_norm": 0.0017938502132892609,
-      "learning_rate": 0.00019999992840014774,
-      "loss": 46.0,
-      "step": 4992
-    },
-    {
-      "epoch": 0.38174971806487373,
-      "grad_norm": 0.000750053150113672,
-      "learning_rate": 0.00019999992837140145,
-      "loss": 46.0,
-      "step": 4993
-    },
-    {
-      "epoch": 0.3818261750482635,
-      "grad_norm": 0.007179294712841511,
-      "learning_rate": 0.00019999992834264938,
-      "loss": 46.0,
-      "step": 4994
-    },
-    {
-      "epoch": 0.3819026320316532,
-      "grad_norm": 0.001865683007054031,
-      "learning_rate": 0.00019999992831389154,
-      "loss": 46.0,
-      "step": 4995
-    },
-    {
-      "epoch": 0.3819790890150429,
-      "grad_norm": 0.003025611164048314,
-      "learning_rate": 0.00019999992828512793,
-      "loss": 46.0,
-      "step": 4996
-    },
-    {
-      "epoch": 0.3820555459984326,
-      "grad_norm": 0.0005240002064965665,
-      "learning_rate": 0.00019999992825635855,
-      "loss": 46.0,
-      "step": 4997
-    },
-    {
-      "epoch": 0.38213200298182237,
-      "grad_norm": 0.001084992429241538,
-      "learning_rate": 0.00019999992822758339,
-      "loss": 46.0,
-      "step": 4998
-    },
-    {
-      "epoch": 0.3822084599652121,
-      "grad_norm": 0.0010686705354601145,
-      "learning_rate": 0.00019999992819880248,
-      "loss": 46.0,
-      "step": 4999
-    },
-    {
-      "epoch": 0.3822849169486018,
-      "grad_norm": 0.0018099829321727157,
-      "learning_rate": 0.0001999999281700158,
-      "loss": 46.0,
-      "step": 5000
-    },
-    {
-      "epoch": 0.3823613739319915,
-      "grad_norm": 0.0015157641610130668,
-      "learning_rate": 0.00019999992814122335,
-      "loss": 46.0,
-      "step": 5001
-    },
-    {
-      "epoch": 0.38243783091538125,
-      "grad_norm": 0.0007292305817827582,
-      "learning_rate": 0.00019999992811242515,
-      "loss": 46.0,
-      "step": 5002
-    },
-    {
-      "epoch": 0.38251428789877095,
-      "grad_norm": 0.0006928403163328767,
-      "learning_rate": 0.00019999992808362112,
-      "loss": 46.0,
-      "step": 5003
-    },
-    {
-      "epoch": 0.38259074488216066,
-      "grad_norm": 0.0003051522944588214,
-      "learning_rate": 0.00019999992805481135,
-      "loss": 46.0,
-      "step": 5004
-    },
-    {
-      "epoch": 0.3826672018655504,
-      "grad_norm": 0.0514926053583622,
-      "learning_rate": 0.00019999992802599583,
-      "loss": 46.0,
-      "step": 5005
-    },
-    {
-      "epoch": 0.3827436588489401,
-      "grad_norm": 0.006026845425367355,
-      "learning_rate": 0.00019999992799717454,
-      "loss": 46.0,
-      "step": 5006
-    },
-    {
-      "epoch": 0.38282011583232983,
-      "grad_norm": 0.0008158411365002394,
-      "learning_rate": 0.00019999992796834745,
-      "loss": 46.0,
-      "step": 5007
-    },
-    {
-      "epoch": 0.38289657281571954,
-      "grad_norm": 0.00041071680607274175,
-      "learning_rate": 0.00019999992793951464,
-      "loss": 46.0,
-      "step": 5008
-    },
-    {
-      "epoch": 0.3829730297991093,
-      "grad_norm": 0.0009232686134055257,
-      "learning_rate": 0.00019999992791067602,
-      "loss": 46.0,
-      "step": 5009
-    },
-    {
-      "epoch": 0.383049486782499,
-      "grad_norm": 0.0008955828379839659,
-      "learning_rate": 0.00019999992788183164,
-      "loss": 46.0,
-      "step": 5010
-    },
-    {
-      "epoch": 0.3831259437658887,
-      "grad_norm": 0.003606929210945964,
-      "learning_rate": 0.00019999992785298148,
-      "loss": 46.0,
-      "step": 5011
-    },
-    {
-      "epoch": 0.3832024007492784,
-      "grad_norm": 0.0006889833020977676,
-      "learning_rate": 0.00019999992782412557,
-      "loss": 46.0,
-      "step": 5012
-    },
-    {
-      "epoch": 0.3832788577326682,
-      "grad_norm": 0.0033818173687905073,
-      "learning_rate": 0.0001999999277952639,
-      "loss": 46.0,
-      "step": 5013
-    },
-    {
-      "epoch": 0.3833553147160579,
-      "grad_norm": 0.0023649297654628754,
-      "learning_rate": 0.00019999992776639642,
-      "loss": 46.0,
-      "step": 5014
-    },
-    {
-      "epoch": 0.3834317716994476,
-      "grad_norm": 0.0013693240471184254,
-      "learning_rate": 0.00019999992773752322,
-      "loss": 46.0,
-      "step": 5015
-    },
-    {
-      "epoch": 0.3835082286828373,
-      "grad_norm": 0.0009798690443858504,
-      "learning_rate": 0.00019999992770864422,
-      "loss": 46.0,
-      "step": 5016
-    },
-    {
-      "epoch": 0.38358468566622705,
-      "grad_norm": 0.002101196674630046,
-      "learning_rate": 0.00019999992767975945,
-      "loss": 46.0,
-      "step": 5017
-    },
-    {
-      "epoch": 0.38366114264961676,
-      "grad_norm": 0.0006856800755485892,
-      "learning_rate": 0.0001999999276508689,
-      "loss": 46.0,
-      "step": 5018
-    },
-    {
-      "epoch": 0.38373759963300647,
-      "grad_norm": 0.0012978214072063565,
-      "learning_rate": 0.0001999999276219726,
-      "loss": 46.0,
-      "step": 5019
-    },
-    {
-      "epoch": 0.3838140566163962,
-      "grad_norm": 0.0005282756756059825,
-      "learning_rate": 0.00019999992759307054,
-      "loss": 46.0,
-      "step": 5020
-    },
-    {
-      "epoch": 0.38389051359978593,
-      "grad_norm": 0.0015676789917051792,
-      "learning_rate": 0.00019999992756416268,
-      "loss": 46.0,
-      "step": 5021
-    },
-    {
-      "epoch": 0.38396697058317564,
-      "grad_norm": 0.00042119945283047855,
-      "learning_rate": 0.00019999992753524907,
-      "loss": 46.0,
-      "step": 5022
-    },
-    {
-      "epoch": 0.38404342756656534,
-      "grad_norm": 0.0009619808988645673,
-      "learning_rate": 0.0001999999275063297,
-      "loss": 46.0,
-      "step": 5023
-    },
-    {
-      "epoch": 0.3841198845499551,
-      "grad_norm": 0.0013550643343478441,
-      "learning_rate": 0.00019999992747740452,
-      "loss": 46.0,
-      "step": 5024
-    },
-    {
-      "epoch": 0.3841963415333448,
-      "grad_norm": 0.0006222610245458782,
-      "learning_rate": 0.00019999992744847362,
-      "loss": 46.0,
-      "step": 5025
-    },
-    {
-      "epoch": 0.3842727985167345,
-      "grad_norm": 0.0012270138831809163,
-      "learning_rate": 0.0001999999274195369,
-      "loss": 46.0,
-      "step": 5026
-    },
-    {
-      "epoch": 0.3843492555001242,
-      "grad_norm": 0.0007512378506362438,
-      "learning_rate": 0.00019999992739059446,
-      "loss": 46.0,
-      "step": 5027
-    },
-    {
-      "epoch": 0.384425712483514,
-      "grad_norm": 0.005678137298673391,
-      "learning_rate": 0.00019999992736164624,
-      "loss": 46.0,
-      "step": 5028
-    },
-    {
-      "epoch": 0.3845021694669037,
-      "grad_norm": 0.005132416263222694,
-      "learning_rate": 0.00019999992733269224,
-      "loss": 46.0,
-      "step": 5029
-    },
-    {
-      "epoch": 0.3845786264502934,
-      "grad_norm": 0.0011747103417292237,
-      "learning_rate": 0.00019999992730373247,
-      "loss": 46.0,
-      "step": 5030
-    },
-    {
-      "epoch": 0.3846550834336831,
-      "grad_norm": 0.00031244984711520374,
-      "learning_rate": 0.00019999992727476692,
-      "loss": 46.0,
-      "step": 5031
-    },
-    {
-      "epoch": 0.38473154041707286,
-      "grad_norm": 0.0007497024489566684,
-      "learning_rate": 0.00019999992724579563,
-      "loss": 46.0,
-      "step": 5032
-    },
-    {
-      "epoch": 0.38480799740046256,
-      "grad_norm": 0.0008607287309132516,
-      "learning_rate": 0.00019999992721681854,
-      "loss": 46.0,
-      "step": 5033
-    },
-    {
-      "epoch": 0.38488445438385227,
-      "grad_norm": 0.0005297533934935927,
-      "learning_rate": 0.0001999999271878357,
-      "loss": 46.0,
-      "step": 5034
-    },
-    {
-      "epoch": 0.38496091136724203,
-      "grad_norm": 0.0015238580526784062,
-      "learning_rate": 0.0001999999271588471,
-      "loss": 46.0,
-      "step": 5035
-    },
-    {
-      "epoch": 0.38503736835063174,
-      "grad_norm": 0.0007831451366655529,
-      "learning_rate": 0.00019999992712985273,
-      "loss": 46.0,
-      "step": 5036
-    },
-    {
-      "epoch": 0.38511382533402144,
-      "grad_norm": 0.022358976304531097,
-      "learning_rate": 0.00019999992710085254,
-      "loss": 46.0,
-      "step": 5037
-    },
-    {
-      "epoch": 0.38519028231741115,
-      "grad_norm": 0.0008355851750820875,
-      "learning_rate": 0.00019999992707184664,
-      "loss": 46.0,
-      "step": 5038
-    },
-    {
-      "epoch": 0.3852667393008009,
-      "grad_norm": 0.0021296977065503597,
-      "learning_rate": 0.00019999992704283493,
-      "loss": 46.0,
-      "step": 5039
-    },
-    {
-      "epoch": 0.3853431962841906,
-      "grad_norm": 0.0005228519439697266,
-      "learning_rate": 0.00019999992701381746,
-      "loss": 46.0,
-      "step": 5040
-    },
-    {
-      "epoch": 0.3854196532675803,
-      "grad_norm": 0.0010646473383530974,
-      "learning_rate": 0.00019999992698479423,
-      "loss": 46.0,
-      "step": 5041
-    },
-    {
-      "epoch": 0.38549611025097,
-      "grad_norm": 0.002040991559624672,
-      "learning_rate": 0.00019999992695576523,
-      "loss": 46.0,
-      "step": 5042
-    },
-    {
-      "epoch": 0.3855725672343598,
-      "grad_norm": 0.0009962695185095072,
-      "learning_rate": 0.00019999992692673046,
-      "loss": 46.0,
-      "step": 5043
-    },
-    {
-      "epoch": 0.3856490242177495,
-      "grad_norm": 0.0004573659098241478,
-      "learning_rate": 0.00019999992689768994,
-      "loss": 46.0,
-      "step": 5044
-    },
-    {
-      "epoch": 0.3857254812011392,
-      "grad_norm": 0.001267716521397233,
-      "learning_rate": 0.00019999992686864362,
-      "loss": 46.0,
-      "step": 5045
-    },
-    {
-      "epoch": 0.3858019381845289,
-      "grad_norm": 0.0018960109446197748,
-      "learning_rate": 0.00019999992683959153,
-      "loss": 46.0,
-      "step": 5046
-    },
-    {
-      "epoch": 0.38587839516791866,
-      "grad_norm": 0.0011301188496872783,
-      "learning_rate": 0.0001999999268105337,
-      "loss": 46.0,
-      "step": 5047
-    },
-    {
-      "epoch": 0.38595485215130837,
-      "grad_norm": 0.0010853378335013986,
-      "learning_rate": 0.00019999992678147006,
-      "loss": 46.0,
-      "step": 5048
-    },
-    {
-      "epoch": 0.3860313091346981,
-      "grad_norm": 0.001279229880310595,
-      "learning_rate": 0.0001999999267524007,
-      "loss": 46.0,
-      "step": 5049
-    },
-    {
-      "epoch": 0.38610776611808784,
-      "grad_norm": 0.00043137866305187345,
-      "learning_rate": 0.00019999992672332551,
-      "loss": 46.0,
-      "step": 5050
-    },
-    {
-      "epoch": 0.38618422310147754,
-      "grad_norm": 0.0014631374506279826,
-      "learning_rate": 0.00019999992669424458,
-      "loss": 46.0,
-      "step": 5051
-    },
-    {
-      "epoch": 0.38626068008486725,
-      "grad_norm": 0.0014483515406027436,
-      "learning_rate": 0.0001999999266651579,
-      "loss": 46.0,
-      "step": 5052
-    },
-    {
-      "epoch": 0.38633713706825695,
-      "grad_norm": 0.002519422210752964,
-      "learning_rate": 0.00019999992663606543,
-      "loss": 46.0,
-      "step": 5053
-    },
-    {
-      "epoch": 0.3864135940516467,
-      "grad_norm": 0.0011974717490375042,
-      "learning_rate": 0.0001999999266069672,
-      "loss": 46.0,
-      "step": 5054
-    },
-    {
-      "epoch": 0.3864900510350364,
-      "grad_norm": 0.0015806478913873434,
-      "learning_rate": 0.0001999999265778632,
-      "loss": 46.0,
-      "step": 5055
-    },
-    {
-      "epoch": 0.3865665080184261,
-      "grad_norm": 0.0015322653343901038,
-      "learning_rate": 0.0001999999265487534,
-      "loss": 46.0,
-      "step": 5056
-    },
-    {
-      "epoch": 0.38664296500181583,
-      "grad_norm": 0.0014195576077327132,
-      "learning_rate": 0.0001999999265196379,
-      "loss": 46.0,
-      "step": 5057
-    },
-    {
-      "epoch": 0.3867194219852056,
-      "grad_norm": 0.0009716078056953847,
-      "learning_rate": 0.00019999992649051657,
-      "loss": 46.0,
-      "step": 5058
-    },
-    {
-      "epoch": 0.3867958789685953,
-      "grad_norm": 0.0016142894746735692,
-      "learning_rate": 0.00019999992646138948,
-      "loss": 46.0,
-      "step": 5059
-    },
-    {
-      "epoch": 0.386872335951985,
-      "grad_norm": 0.001137088518589735,
-      "learning_rate": 0.00019999992643225664,
-      "loss": 46.0,
-      "step": 5060
-    },
-    {
-      "epoch": 0.3869487929353747,
-      "grad_norm": 0.0006488229846581817,
-      "learning_rate": 0.000199999926403118,
-      "loss": 46.0,
-      "step": 5061
-    },
-    {
-      "epoch": 0.38702524991876447,
-      "grad_norm": 0.0004907710826955736,
-      "learning_rate": 0.00019999992637397362,
-      "loss": 46.0,
-      "step": 5062
-    },
-    {
-      "epoch": 0.3871017069021542,
-      "grad_norm": 0.00034755514934659004,
-      "learning_rate": 0.00019999992634482346,
-      "loss": 46.0,
-      "step": 5063
-    },
-    {
-      "epoch": 0.3871781638855439,
-      "grad_norm": 0.0007590790628455579,
-      "learning_rate": 0.00019999992631566753,
-      "loss": 46.0,
-      "step": 5064
-    },
-    {
-      "epoch": 0.38725462086893364,
-      "grad_norm": 0.009573681280016899,
-      "learning_rate": 0.00019999992628650583,
-      "loss": 46.0,
-      "step": 5065
-    },
-    {
-      "epoch": 0.38733107785232335,
-      "grad_norm": 0.0034562053624540567,
-      "learning_rate": 0.00019999992625733837,
-      "loss": 46.0,
-      "step": 5066
-    },
-    {
-      "epoch": 0.38740753483571305,
-      "grad_norm": 0.0015021811705082655,
-      "learning_rate": 0.00019999992622816512,
-      "loss": 46.0,
-      "step": 5067
-    },
-    {
-      "epoch": 0.38748399181910276,
-      "grad_norm": 0.0008419494261033833,
-      "learning_rate": 0.00019999992619898613,
-      "loss": 46.0,
-      "step": 5068
-    },
-    {
-      "epoch": 0.3875604488024925,
-      "grad_norm": 0.0004314680409152061,
-      "learning_rate": 0.00019999992616980135,
-      "loss": 46.0,
-      "step": 5069
-    },
-    {
-      "epoch": 0.3876369057858822,
-      "grad_norm": 0.000730108586139977,
-      "learning_rate": 0.00019999992614061078,
-      "loss": 46.0,
-      "step": 5070
-    },
-    {
-      "epoch": 0.38771336276927193,
-      "grad_norm": 0.0014029715675860643,
-      "learning_rate": 0.0001999999261114145,
-      "loss": 46.0,
-      "step": 5071
-    },
-    {
-      "epoch": 0.38778981975266164,
-      "grad_norm": 0.0010304381139576435,
-      "learning_rate": 0.00019999992608221237,
-      "loss": 46.0,
-      "step": 5072
-    },
-    {
-      "epoch": 0.3878662767360514,
-      "grad_norm": 0.0039737895131111145,
-      "learning_rate": 0.00019999992605300454,
-      "loss": 46.0,
-      "step": 5073
-    },
-    {
-      "epoch": 0.3879427337194411,
-      "grad_norm": 0.0003692148602567613,
-      "learning_rate": 0.00019999992602379093,
-      "loss": 46.0,
-      "step": 5074
-    },
-    {
-      "epoch": 0.3880191907028308,
-      "grad_norm": 0.0005067788879387081,
-      "learning_rate": 0.00019999992599457152,
-      "loss": 46.0,
-      "step": 5075
-    },
-    {
-      "epoch": 0.3880956476862205,
-      "grad_norm": 0.0016741999424993992,
-      "learning_rate": 0.00019999992596534636,
-      "loss": 46.0,
-      "step": 5076
-    },
-    {
-      "epoch": 0.3881721046696103,
-      "grad_norm": 0.004904313012957573,
-      "learning_rate": 0.00019999992593611543,
-      "loss": 46.0,
-      "step": 5077
-    },
-    {
-      "epoch": 0.388248561653,
-      "grad_norm": 0.0013878638856112957,
-      "learning_rate": 0.00019999992590687872,
-      "loss": 46.0,
-      "step": 5078
-    },
-    {
-      "epoch": 0.3883250186363897,
-      "grad_norm": 0.0012515804264694452,
-      "learning_rate": 0.00019999992587763625,
-      "loss": 46.0,
-      "step": 5079
-    },
-    {
-      "epoch": 0.38840147561977945,
-      "grad_norm": 0.0029100251849740744,
-      "learning_rate": 0.000199999925848388,
-      "loss": 46.0,
-      "step": 5080
-    },
-    {
-      "epoch": 0.38847793260316915,
-      "grad_norm": 0.003508092137053609,
-      "learning_rate": 0.000199999925819134,
-      "loss": 46.0,
-      "step": 5081
-    },
-    {
-      "epoch": 0.38855438958655886,
-      "grad_norm": 0.0012574639404192567,
-      "learning_rate": 0.00019999992578987423,
-      "loss": 46.0,
-      "step": 5082
-    },
-    {
-      "epoch": 0.38863084656994856,
-      "grad_norm": 0.0008746524690650403,
-      "learning_rate": 0.00019999992576060866,
-      "loss": 46.0,
-      "step": 5083
-    },
-    {
-      "epoch": 0.3887073035533383,
-      "grad_norm": 0.0011893465416505933,
-      "learning_rate": 0.00019999992573133737,
-      "loss": 46.0,
-      "step": 5084
-    },
-    {
-      "epoch": 0.38878376053672803,
-      "grad_norm": 0.0013913646107539535,
-      "learning_rate": 0.00019999992570206025,
-      "loss": 46.0,
-      "step": 5085
-    },
-    {
-      "epoch": 0.38886021752011773,
-      "grad_norm": 0.0018540453165769577,
-      "learning_rate": 0.0001999999256727774,
-      "loss": 46.0,
-      "step": 5086
-    },
-    {
-      "epoch": 0.38893667450350744,
-      "grad_norm": 0.001524603576399386,
-      "learning_rate": 0.0001999999256434888,
-      "loss": 46.0,
-      "step": 5087
-    },
-    {
-      "epoch": 0.3890131314868972,
-      "grad_norm": 0.0026149724144488573,
-      "learning_rate": 0.0001999999256141944,
-      "loss": 46.0,
-      "step": 5088
-    },
-    {
-      "epoch": 0.3890895884702869,
-      "grad_norm": 0.0011269188253208995,
-      "learning_rate": 0.00019999992558489423,
-      "loss": 46.0,
-      "step": 5089
-    },
-    {
-      "epoch": 0.3891660454536766,
-      "grad_norm": 0.0004958967329002917,
-      "learning_rate": 0.00019999992555558827,
-      "loss": 46.0,
-      "step": 5090
-    },
-    {
-      "epoch": 0.3892425024370663,
-      "grad_norm": 0.004132285714149475,
-      "learning_rate": 0.00019999992552627657,
-      "loss": 46.0,
-      "step": 5091
-    },
-    {
-      "epoch": 0.3893189594204561,
-      "grad_norm": 0.00184576865285635,
-      "learning_rate": 0.00019999992549695912,
-      "loss": 46.0,
-      "step": 5092
-    },
-    {
-      "epoch": 0.3893954164038458,
-      "grad_norm": 0.0006565260700881481,
-      "learning_rate": 0.00019999992546763587,
-      "loss": 46.0,
-      "step": 5093
-    },
-    {
-      "epoch": 0.3894718733872355,
-      "grad_norm": 0.004384566098451614,
-      "learning_rate": 0.00019999992543830685,
-      "loss": 46.0,
-      "step": 5094
-    },
-    {
-      "epoch": 0.38954833037062525,
-      "grad_norm": 0.0009473259560763836,
-      "learning_rate": 0.00019999992540897208,
-      "loss": 46.0,
-      "step": 5095
-    },
-    {
-      "epoch": 0.38962478735401496,
-      "grad_norm": 0.0009245181572623551,
-      "learning_rate": 0.0001999999253796315,
-      "loss": 46.0,
-      "step": 5096
-    },
-    {
-      "epoch": 0.38970124433740466,
-      "grad_norm": 0.0016667972085997462,
-      "learning_rate": 0.0001999999253502852,
-      "loss": 46.0,
-      "step": 5097
-    },
-    {
-      "epoch": 0.38977770132079437,
-      "grad_norm": 0.0008501725969836116,
-      "learning_rate": 0.0001999999253209331,
-      "loss": 46.0,
-      "step": 5098
-    },
-    {
-      "epoch": 0.38985415830418413,
-      "grad_norm": 0.0021702460944652557,
-      "learning_rate": 0.00019999992529157523,
-      "loss": 46.0,
-      "step": 5099
-    },
-    {
-      "epoch": 0.38993061528757383,
-      "grad_norm": 0.0017222303431481123,
-      "learning_rate": 0.0001999999252622116,
-      "loss": 46.0,
-      "step": 5100
-    },
-    {
-      "epoch": 0.39000707227096354,
-      "grad_norm": 0.001087190699763596,
-      "learning_rate": 0.0001999999252328422,
-      "loss": 46.0,
-      "step": 5101
-    },
-    {
-      "epoch": 0.39008352925435325,
-      "grad_norm": 0.0009349556639790535,
-      "learning_rate": 0.00019999992520346706,
-      "loss": 46.0,
-      "step": 5102
-    },
-    {
-      "epoch": 0.390159986237743,
-      "grad_norm": 0.0029659061692655087,
-      "learning_rate": 0.0001999999251740861,
-      "loss": 46.0,
-      "step": 5103
-    },
-    {
-      "epoch": 0.3902364432211327,
-      "grad_norm": 0.0008405956323258579,
-      "learning_rate": 0.0001999999251446994,
-      "loss": 46.0,
-      "step": 5104
-    },
-    {
-      "epoch": 0.3903129002045224,
-      "grad_norm": 0.0007301776204258204,
-      "learning_rate": 0.00019999992511530695,
-      "loss": 46.0,
-      "step": 5105
-    },
-    {
-      "epoch": 0.3903893571879122,
-      "grad_norm": 0.0008082663407549262,
-      "learning_rate": 0.00019999992508590867,
-      "loss": 46.0,
-      "step": 5106
-    },
-    {
-      "epoch": 0.3904658141713019,
-      "grad_norm": 0.0008237584261223674,
-      "learning_rate": 0.00019999992505650467,
-      "loss": 46.0,
-      "step": 5107
-    },
-    {
-      "epoch": 0.3905422711546916,
-      "grad_norm": 0.001603541779331863,
-      "learning_rate": 0.00019999992502709487,
-      "loss": 46.0,
-      "step": 5108
-    },
-    {
-      "epoch": 0.3906187281380813,
-      "grad_norm": 0.0006512822583317757,
-      "learning_rate": 0.00019999992499767933,
-      "loss": 46.0,
-      "step": 5109
-    },
-    {
-      "epoch": 0.39069518512147106,
-      "grad_norm": 0.0025201989337801933,
-      "learning_rate": 0.000199999924968258,
-      "loss": 46.0,
-      "step": 5110
-    },
-    {
-      "epoch": 0.39077164210486076,
-      "grad_norm": 0.0008589408826082945,
-      "learning_rate": 0.0001999999249388309,
-      "loss": 46.0,
-      "step": 5111
-    },
-    {
-      "epoch": 0.39084809908825047,
-      "grad_norm": 0.0007604301790706813,
-      "learning_rate": 0.00019999992490939806,
-      "loss": 46.0,
-      "step": 5112
-    },
-    {
-      "epoch": 0.3909245560716402,
-      "grad_norm": 0.0029012097511440516,
-      "learning_rate": 0.00019999992487995942,
-      "loss": 46.0,
-      "step": 5113
-    },
-    {
-      "epoch": 0.39100101305502993,
-      "grad_norm": 0.001728267059661448,
-      "learning_rate": 0.000199999924850515,
-      "loss": 46.0,
-      "step": 5114
-    },
-    {
-      "epoch": 0.39107747003841964,
-      "grad_norm": 0.0007799426093697548,
-      "learning_rate": 0.00019999992482106486,
-      "loss": 46.0,
-      "step": 5115
-    },
-    {
-      "epoch": 0.39115392702180934,
-      "grad_norm": 0.0007992628961801529,
-      "learning_rate": 0.0001999999247916089,
-      "loss": 46.0,
-      "step": 5116
-    },
-    {
-      "epoch": 0.39123038400519905,
-      "grad_norm": 0.0012400917476043105,
-      "learning_rate": 0.00019999992476214717,
-      "loss": 46.0,
-      "step": 5117
-    },
-    {
-      "epoch": 0.3913068409885888,
-      "grad_norm": 0.0023228100035339594,
-      "learning_rate": 0.00019999992473267972,
-      "loss": 46.0,
-      "step": 5118
-    },
-    {
-      "epoch": 0.3913832979719785,
-      "grad_norm": 0.0033154755365103483,
-      "learning_rate": 0.00019999992470320644,
-      "loss": 46.0,
-      "step": 5119
-    },
-    {
-      "epoch": 0.3914597549553682,
-      "grad_norm": 0.00548673328012228,
-      "learning_rate": 0.00019999992467372745,
-      "loss": 46.0,
-      "step": 5120
-    },
-    {
-      "epoch": 0.391536211938758,
-      "grad_norm": 0.0013765579788014293,
-      "learning_rate": 0.00019999992464424265,
-      "loss": 46.0,
-      "step": 5121
-    },
-    {
-      "epoch": 0.3916126689221477,
-      "grad_norm": 0.0014996732352301478,
-      "learning_rate": 0.0001999999246147521,
-      "loss": 46.0,
-      "step": 5122
-    },
-    {
-      "epoch": 0.3916891259055374,
-      "grad_norm": 0.0031229890882968903,
-      "learning_rate": 0.00019999992458525576,
-      "loss": 46.0,
-      "step": 5123
-    },
-    {
-      "epoch": 0.3917655828889271,
-      "grad_norm": 0.0011169922072440386,
-      "learning_rate": 0.00019999992455575367,
-      "loss": 46.0,
-      "step": 5124
-    },
-    {
-      "epoch": 0.39184203987231686,
-      "grad_norm": 0.0013352936366572976,
-      "learning_rate": 0.0001999999245262458,
-      "loss": 46.0,
-      "step": 5125
-    },
-    {
-      "epoch": 0.39191849685570657,
-      "grad_norm": 0.0014544575242325664,
-      "learning_rate": 0.00019999992449673217,
-      "loss": 46.0,
-      "step": 5126
-    },
-    {
-      "epoch": 0.39199495383909627,
-      "grad_norm": 0.0006914925761520863,
-      "learning_rate": 0.00019999992446721276,
-      "loss": 46.0,
-      "step": 5127
-    },
-    {
-      "epoch": 0.392071410822486,
-      "grad_norm": 0.0014417931670323014,
-      "learning_rate": 0.0001999999244376876,
-      "loss": 46.0,
-      "step": 5128
-    },
-    {
-      "epoch": 0.39214786780587574,
-      "grad_norm": 0.0005496296798810363,
-      "learning_rate": 0.00019999992440815665,
-      "loss": 46.0,
-      "step": 5129
-    },
-    {
-      "epoch": 0.39222432478926544,
-      "grad_norm": 0.0007964738178998232,
-      "learning_rate": 0.00019999992437861992,
-      "loss": 46.0,
-      "step": 5130
-    },
-    {
-      "epoch": 0.39230078177265515,
-      "grad_norm": 0.0012175141600891948,
-      "learning_rate": 0.00019999992434907747,
-      "loss": 46.0,
-      "step": 5131
-    },
-    {
-      "epoch": 0.39237723875604485,
-      "grad_norm": 0.0024001123383641243,
-      "learning_rate": 0.0001999999243195292,
-      "loss": 46.0,
-      "step": 5132
-    },
-    {
-      "epoch": 0.3924536957394346,
-      "grad_norm": 0.0005558711127378047,
-      "learning_rate": 0.00019999992428997517,
-      "loss": 46.0,
-      "step": 5133
-    },
-    {
-      "epoch": 0.3925301527228243,
-      "grad_norm": 0.004153661895543337,
-      "learning_rate": 0.00019999992426041538,
-      "loss": 46.0,
-      "step": 5134
-    },
-    {
-      "epoch": 0.392606609706214,
-      "grad_norm": 0.0008092347416095436,
-      "learning_rate": 0.0001999999242308498,
-      "loss": 46.0,
-      "step": 5135
-    },
-    {
-      "epoch": 0.3926830666896038,
-      "grad_norm": 0.0009048887877725065,
-      "learning_rate": 0.0001999999242012785,
-      "loss": 46.0,
-      "step": 5136
-    },
-    {
-      "epoch": 0.3927595236729935,
-      "grad_norm": 0.004317007027566433,
-      "learning_rate": 0.0001999999241717014,
-      "loss": 46.0,
-      "step": 5137
-    },
-    {
-      "epoch": 0.3928359806563832,
-      "grad_norm": 0.0023986853193491697,
-      "learning_rate": 0.0001999999241421185,
-      "loss": 46.0,
-      "step": 5138
-    },
-    {
-      "epoch": 0.3929124376397729,
-      "grad_norm": 0.000637244142126292,
-      "learning_rate": 0.00019999992411252987,
-      "loss": 46.0,
-      "step": 5139
-    },
-    {
-      "epoch": 0.39298889462316267,
-      "grad_norm": 0.004860072862356901,
-      "learning_rate": 0.0001999999240829355,
-      "loss": 46.0,
-      "step": 5140
-    },
-    {
-      "epoch": 0.39306535160655237,
-      "grad_norm": 0.0027384189888834953,
-      "learning_rate": 0.0001999999240533353,
-      "loss": 46.0,
-      "step": 5141
-    },
-    {
-      "epoch": 0.3931418085899421,
-      "grad_norm": 0.0015161463525146246,
-      "learning_rate": 0.00019999992402372935,
-      "loss": 46.0,
-      "step": 5142
-    },
-    {
-      "epoch": 0.3932182655733318,
-      "grad_norm": 0.002926499815657735,
-      "learning_rate": 0.00019999992399411765,
-      "loss": 46.0,
-      "step": 5143
-    },
-    {
-      "epoch": 0.39329472255672154,
-      "grad_norm": 0.0041192155331373215,
-      "learning_rate": 0.00019999992396450015,
-      "loss": 46.0,
-      "step": 5144
-    },
-    {
-      "epoch": 0.39337117954011125,
-      "grad_norm": 0.001989888260141015,
-      "learning_rate": 0.0001999999239348769,
-      "loss": 46.0,
-      "step": 5145
-    },
-    {
-      "epoch": 0.39344763652350095,
-      "grad_norm": 0.0006507466896437109,
-      "learning_rate": 0.00019999992390524787,
-      "loss": 46.0,
-      "step": 5146
-    },
-    {
-      "epoch": 0.39352409350689066,
-      "grad_norm": 0.007054603658616543,
-      "learning_rate": 0.00019999992387561308,
-      "loss": 46.0,
-      "step": 5147
-    },
-    {
-      "epoch": 0.3936005504902804,
-      "grad_norm": 0.008480913005769253,
-      "learning_rate": 0.00019999992384597254,
-      "loss": 46.0,
-      "step": 5148
-    },
-    {
-      "epoch": 0.3936770074736701,
-      "grad_norm": 0.0007137679494917393,
-      "learning_rate": 0.0001999999238163262,
-      "loss": 46.0,
-      "step": 5149
-    },
-    {
-      "epoch": 0.39375346445705983,
-      "grad_norm": 0.0012047920608893037,
-      "learning_rate": 0.0001999999237866741,
-      "loss": 46.0,
-      "step": 5150
-    },
-    {
-      "epoch": 0.3938299214404496,
-      "grad_norm": 0.0010212705237790942,
-      "learning_rate": 0.00019999992375701621,
-      "loss": 46.0,
-      "step": 5151
-    },
-    {
-      "epoch": 0.3939063784238393,
-      "grad_norm": 0.000611610827036202,
-      "learning_rate": 0.00019999992372735258,
-      "loss": 46.0,
-      "step": 5152
-    },
-    {
-      "epoch": 0.393982835407229,
-      "grad_norm": 0.0076452926732599735,
-      "learning_rate": 0.00019999992369768317,
-      "loss": 46.0,
-      "step": 5153
-    },
-    {
-      "epoch": 0.3940592923906187,
-      "grad_norm": 0.0012855703243985772,
-      "learning_rate": 0.00019999992366800799,
-      "loss": 46.0,
-      "step": 5154
-    },
-    {
-      "epoch": 0.39413574937400847,
-      "grad_norm": 0.0008539396803826094,
-      "learning_rate": 0.00019999992363832703,
-      "loss": 46.0,
-      "step": 5155
-    },
-    {
-      "epoch": 0.3942122063573982,
-      "grad_norm": 0.0011326392414048314,
-      "learning_rate": 0.00019999992360864033,
-      "loss": 46.0,
-      "step": 5156
-    },
-    {
-      "epoch": 0.3942886633407879,
-      "grad_norm": 0.0008603236638009548,
-      "learning_rate": 0.00019999992357894782,
-      "loss": 46.0,
-      "step": 5157
-    },
-    {
-      "epoch": 0.3943651203241776,
-      "grad_norm": 0.0015567372320219874,
-      "learning_rate": 0.00019999992354924958,
-      "loss": 46.0,
-      "step": 5158
-    },
-    {
-      "epoch": 0.39444157730756735,
-      "grad_norm": 0.0011137907858937979,
-      "learning_rate": 0.00019999992351954555,
-      "loss": 46.0,
-      "step": 5159
-    },
-    {
-      "epoch": 0.39451803429095705,
-      "grad_norm": 0.0011242581531405449,
-      "learning_rate": 0.00019999992348983573,
-      "loss": 46.0,
-      "step": 5160
-    },
-    {
-      "epoch": 0.39459449127434676,
-      "grad_norm": 0.001839528325945139,
-      "learning_rate": 0.0001999999234601202,
-      "loss": 46.0,
-      "step": 5161
-    },
-    {
-      "epoch": 0.39467094825773646,
-      "grad_norm": 0.005966348107904196,
-      "learning_rate": 0.00019999992343039885,
-      "loss": 46.0,
-      "step": 5162
-    },
-    {
-      "epoch": 0.3947474052411262,
-      "grad_norm": 0.00060272152768448,
-      "learning_rate": 0.00019999992340067176,
-      "loss": 46.0,
-      "step": 5163
-    },
-    {
-      "epoch": 0.39482386222451593,
-      "grad_norm": 0.0013761867303401232,
-      "learning_rate": 0.00019999992337093887,
-      "loss": 46.0,
-      "step": 5164
-    },
-    {
-      "epoch": 0.39490031920790564,
-      "grad_norm": 0.0026355090085417032,
-      "learning_rate": 0.0001999999233412002,
-      "loss": 46.0,
-      "step": 5165
-    },
-    {
-      "epoch": 0.3949767761912954,
-      "grad_norm": 0.0004673987568821758,
-      "learning_rate": 0.0001999999233114558,
-      "loss": 46.0,
-      "step": 5166
-    },
-    {
-      "epoch": 0.3950532331746851,
-      "grad_norm": 0.0013201392721384764,
-      "learning_rate": 0.0001999999232817056,
-      "loss": 46.0,
-      "step": 5167
-    },
-    {
-      "epoch": 0.3951296901580748,
-      "grad_norm": 0.0004839313041884452,
-      "learning_rate": 0.00019999992325194968,
-      "loss": 46.0,
-      "step": 5168
-    },
-    {
-      "epoch": 0.3952061471414645,
-      "grad_norm": 0.005619749892503023,
-      "learning_rate": 0.00019999992322218796,
-      "loss": 46.0,
-      "step": 5169
-    },
-    {
-      "epoch": 0.3952826041248543,
-      "grad_norm": 0.0011793739395216107,
-      "learning_rate": 0.00019999992319242048,
-      "loss": 46.0,
-      "step": 5170
-    },
-    {
-      "epoch": 0.395359061108244,
-      "grad_norm": 0.0028332117944955826,
-      "learning_rate": 0.0001999999231626472,
-      "loss": 46.0,
-      "step": 5171
-    },
-    {
-      "epoch": 0.3954355180916337,
-      "grad_norm": 0.0005728068645112216,
-      "learning_rate": 0.00019999992313286818,
-      "loss": 46.0,
-      "step": 5172
-    },
-    {
-      "epoch": 0.3955119750750234,
-      "grad_norm": 0.0030518416315317154,
-      "learning_rate": 0.00019999992310308336,
-      "loss": 46.0,
-      "step": 5173
-    },
-    {
-      "epoch": 0.39558843205841315,
-      "grad_norm": 0.002132006688043475,
-      "learning_rate": 0.0001999999230732928,
-      "loss": 46.0,
-      "step": 5174
-    },
-    {
-      "epoch": 0.39566488904180286,
-      "grad_norm": 0.0009794343495741487,
-      "learning_rate": 0.00019999992304349648,
-      "loss": 46.0,
-      "step": 5175
-    },
-    {
-      "epoch": 0.39574134602519256,
-      "grad_norm": 0.0015904414467513561,
-      "learning_rate": 0.00019999992301369436,
-      "loss": 46.0,
-      "step": 5176
-    },
-    {
-      "epoch": 0.39581780300858227,
-      "grad_norm": 0.0008810659637674689,
-      "learning_rate": 0.00019999992298388647,
-      "loss": 46.0,
-      "step": 5177
-    },
-    {
-      "epoch": 0.39589425999197203,
-      "grad_norm": 0.003126387717202306,
-      "learning_rate": 0.0001999999229540728,
-      "loss": 46.0,
-      "step": 5178
-    },
-    {
-      "epoch": 0.39597071697536174,
-      "grad_norm": 0.003057079389691353,
-      "learning_rate": 0.00019999992292425343,
-      "loss": 46.0,
-      "step": 5179
-    },
-    {
-      "epoch": 0.39604717395875144,
-      "grad_norm": 0.0038163457065820694,
-      "learning_rate": 0.00019999992289442822,
-      "loss": 46.0,
-      "step": 5180
-    },
-    {
-      "epoch": 0.3961236309421412,
-      "grad_norm": 0.005050922278314829,
-      "learning_rate": 0.0001999999228645973,
-      "loss": 46.0,
-      "step": 5181
-    },
-    {
-      "epoch": 0.3962000879255309,
-      "grad_norm": 0.0018765765707939863,
-      "learning_rate": 0.00019999992283476053,
-      "loss": 46.0,
-      "step": 5182
-    },
-    {
-      "epoch": 0.3962765449089206,
-      "grad_norm": 0.0016817814903333783,
-      "learning_rate": 0.00019999992280491803,
-      "loss": 46.0,
-      "step": 5183
-    },
-    {
-      "epoch": 0.3963530018923103,
-      "grad_norm": 0.001378295011818409,
-      "learning_rate": 0.00019999992277506978,
-      "loss": 46.0,
-      "step": 5184
-    },
-    {
-      "epoch": 0.3964294588757001,
-      "grad_norm": 0.003858389565721154,
-      "learning_rate": 0.00019999992274521576,
-      "loss": 46.0,
-      "step": 5185
-    },
-    {
-      "epoch": 0.3965059158590898,
-      "grad_norm": 0.0011563836596906185,
-      "learning_rate": 0.00019999992271535594,
-      "loss": 46.0,
-      "step": 5186
-    },
-    {
-      "epoch": 0.3965823728424795,
-      "grad_norm": 0.000758295354899019,
-      "learning_rate": 0.00019999992268549037,
-      "loss": 46.0,
-      "step": 5187
-    },
-    {
-      "epoch": 0.3966588298258692,
-      "grad_norm": 0.0018489421345293522,
-      "learning_rate": 0.00019999992265561906,
-      "loss": 46.0,
-      "step": 5188
-    },
-    {
-      "epoch": 0.39673528680925896,
-      "grad_norm": 0.002228625351563096,
-      "learning_rate": 0.00019999992262574194,
-      "loss": 46.0,
-      "step": 5189
-    },
-    {
-      "epoch": 0.39681174379264866,
-      "grad_norm": 0.0015381401171907783,
-      "learning_rate": 0.00019999992259585905,
-      "loss": 46.0,
-      "step": 5190
-    },
-    {
-      "epoch": 0.39688820077603837,
-      "grad_norm": 0.002447132719680667,
-      "learning_rate": 0.0001999999225659704,
-      "loss": 46.0,
-      "step": 5191
-    },
-    {
-      "epoch": 0.3969646577594281,
-      "grad_norm": 0.0007145426352508366,
-      "learning_rate": 0.00019999992253607598,
-      "loss": 46.0,
-      "step": 5192
-    },
-    {
-      "epoch": 0.39704111474281784,
-      "grad_norm": 0.001091236132197082,
-      "learning_rate": 0.0001999999225061758,
-      "loss": 46.0,
-      "step": 5193
-    },
-    {
-      "epoch": 0.39711757172620754,
-      "grad_norm": 0.0007018567412160337,
-      "learning_rate": 0.00019999992247626982,
-      "loss": 46.0,
-      "step": 5194
-    },
-    {
-      "epoch": 0.39719402870959725,
-      "grad_norm": 0.010759027674794197,
-      "learning_rate": 0.0001999999224463581,
-      "loss": 46.0,
-      "step": 5195
-    },
-    {
-      "epoch": 0.397270485692987,
-      "grad_norm": 0.0006774477078579366,
-      "learning_rate": 0.00019999992241644061,
-      "loss": 46.0,
-      "step": 5196
-    },
-    {
-      "epoch": 0.3973469426763767,
-      "grad_norm": 0.0025085732340812683,
-      "learning_rate": 0.00019999992238651734,
-      "loss": 46.0,
-      "step": 5197
-    },
-    {
-      "epoch": 0.3974233996597664,
-      "grad_norm": 0.0005426656571216881,
-      "learning_rate": 0.00019999992235658832,
-      "loss": 46.0,
-      "step": 5198
-    },
-    {
-      "epoch": 0.3974998566431561,
-      "grad_norm": 0.002728756284341216,
-      "learning_rate": 0.0001999999223266535,
-      "loss": 46.0,
-      "step": 5199
-    },
-    {
-      "epoch": 0.3975763136265459,
-      "grad_norm": 0.0028018064331263304,
-      "learning_rate": 0.00019999992229671293,
-      "loss": 46.0,
-      "step": 5200
-    },
-    {
-      "epoch": 0.3976527706099356,
-      "grad_norm": 0.0006701347883790731,
-      "learning_rate": 0.00019999992226676659,
-      "loss": 46.0,
-      "step": 5201
-    },
-    {
-      "epoch": 0.3977292275933253,
-      "grad_norm": 0.0027472658548504114,
-      "learning_rate": 0.00019999992223681447,
-      "loss": 46.0,
-      "step": 5202
-    },
-    {
-      "epoch": 0.397805684576715,
-      "grad_norm": 0.0007059333729557693,
-      "learning_rate": 0.00019999992220685658,
-      "loss": 46.0,
-      "step": 5203
-    },
-    {
-      "epoch": 0.39788214156010476,
-      "grad_norm": 0.0009670344879850745,
-      "learning_rate": 0.00019999992217689295,
-      "loss": 46.0,
-      "step": 5204
-    },
-    {
-      "epoch": 0.39795859854349447,
-      "grad_norm": 0.004457896109670401,
-      "learning_rate": 0.0001999999221469235,
-      "loss": 46.0,
-      "step": 5205
-    },
-    {
-      "epoch": 0.3980350555268842,
-      "grad_norm": 0.0008017513900995255,
-      "learning_rate": 0.0001999999221169483,
-      "loss": 46.0,
-      "step": 5206
-    },
-    {
-      "epoch": 0.3981115125102739,
-      "grad_norm": 0.0021098165307193995,
-      "learning_rate": 0.00019999992208696735,
-      "loss": 46.0,
-      "step": 5207
-    },
-    {
-      "epoch": 0.39818796949366364,
-      "grad_norm": 0.002500280039384961,
-      "learning_rate": 0.00019999992205698065,
-      "loss": 46.0,
-      "step": 5208
-    },
-    {
-      "epoch": 0.39826442647705335,
-      "grad_norm": 0.0009139854228124022,
-      "learning_rate": 0.00019999992202698815,
-      "loss": 46.0,
-      "step": 5209
-    },
-    {
-      "epoch": 0.39834088346044305,
-      "grad_norm": 0.000868063245434314,
-      "learning_rate": 0.00019999992199698987,
-      "loss": 46.0,
-      "step": 5210
-    },
-    {
-      "epoch": 0.3984173404438328,
-      "grad_norm": 0.003440618747845292,
-      "learning_rate": 0.00019999992196698582,
-      "loss": 46.0,
-      "step": 5211
-    },
-    {
-      "epoch": 0.3984937974272225,
-      "grad_norm": 0.0007640470284968615,
-      "learning_rate": 0.000199999921936976,
-      "loss": 46.0,
-      "step": 5212
-    },
-    {
-      "epoch": 0.3985702544106122,
-      "grad_norm": 0.00931647326797247,
-      "learning_rate": 0.00019999992190696043,
-      "loss": 46.0,
-      "step": 5213
-    },
-    {
-      "epoch": 0.39864671139400193,
-      "grad_norm": 0.0030859685502946377,
-      "learning_rate": 0.0001999999218769391,
-      "loss": 46.0,
-      "step": 5214
-    },
-    {
-      "epoch": 0.3987231683773917,
-      "grad_norm": 0.0037534020375460386,
-      "learning_rate": 0.00019999992184691198,
-      "loss": 46.0,
-      "step": 5215
-    },
-    {
-      "epoch": 0.3987996253607814,
-      "grad_norm": 0.002515604253858328,
-      "learning_rate": 0.0001999999218168791,
-      "loss": 46.0,
-      "step": 5216
-    },
-    {
-      "epoch": 0.3988760823441711,
-      "grad_norm": 0.0013701661955565214,
-      "learning_rate": 0.00019999992178684043,
-      "loss": 46.0,
-      "step": 5217
-    },
-    {
-      "epoch": 0.3989525393275608,
-      "grad_norm": 0.0018612011335790157,
-      "learning_rate": 0.000199999921756796,
-      "loss": 46.0,
-      "step": 5218
-    },
-    {
-      "epoch": 0.39902899631095057,
-      "grad_norm": 0.004496550187468529,
-      "learning_rate": 0.0001999999217267458,
-      "loss": 46.0,
-      "step": 5219
-    },
-    {
-      "epoch": 0.3991054532943403,
-      "grad_norm": 0.0005126980831846595,
-      "learning_rate": 0.00019999992169668986,
-      "loss": 46.0,
-      "step": 5220
-    },
-    {
-      "epoch": 0.39918191027773,
-      "grad_norm": 0.0009422823204658926,
-      "learning_rate": 0.00019999992166662813,
-      "loss": 46.0,
-      "step": 5221
-    },
-    {
-      "epoch": 0.39925836726111974,
-      "grad_norm": 0.0018354547210037708,
-      "learning_rate": 0.00019999992163656063,
-      "loss": 46.0,
-      "step": 5222
-    },
-    {
-      "epoch": 0.39933482424450945,
-      "grad_norm": 0.0038140076212584972,
-      "learning_rate": 0.00019999992160648735,
-      "loss": 46.0,
-      "step": 5223
-    },
-    {
-      "epoch": 0.39941128122789915,
-      "grad_norm": 0.0018269179854542017,
-      "learning_rate": 0.0001999999215764083,
-      "loss": 46.0,
-      "step": 5224
-    },
-    {
-      "epoch": 0.39948773821128886,
-      "grad_norm": 0.0017675970448181033,
-      "learning_rate": 0.00019999992154632348,
-      "loss": 46.0,
-      "step": 5225
-    },
-    {
-      "epoch": 0.3995641951946786,
-      "grad_norm": 0.002125356113538146,
-      "learning_rate": 0.00019999992151623292,
-      "loss": 46.0,
-      "step": 5226
-    },
-    {
-      "epoch": 0.3996406521780683,
-      "grad_norm": 0.0029736682772636414,
-      "learning_rate": 0.00019999992148613655,
-      "loss": 46.0,
-      "step": 5227
-    },
-    {
-      "epoch": 0.39971710916145803,
-      "grad_norm": 0.0013741985894739628,
-      "learning_rate": 0.00019999992145603446,
-      "loss": 46.0,
-      "step": 5228
-    },
-    {
-      "epoch": 0.39979356614484773,
-      "grad_norm": 0.0005685454234480858,
-      "learning_rate": 0.00019999992142592655,
-      "loss": 46.0,
-      "step": 5229
-    },
-    {
-      "epoch": 0.3998700231282375,
-      "grad_norm": 0.001084899646230042,
-      "learning_rate": 0.00019999992139581289,
-      "loss": 46.0,
-      "step": 5230
-    },
-    {
-      "epoch": 0.3999464801116272,
-      "grad_norm": 0.0014022631803527474,
-      "learning_rate": 0.00019999992136569348,
-      "loss": 46.0,
-      "step": 5231
-    },
-    {
-      "epoch": 0.4000229370950169,
-      "grad_norm": 0.0008176940609700978,
-      "learning_rate": 0.0001999999213355683,
-      "loss": 46.0,
-      "step": 5232
-    },
-    {
-      "epoch": 0.4000993940784066,
-      "grad_norm": 0.0008974050870165229,
-      "learning_rate": 0.00019999992130543732,
-      "loss": 46.0,
-      "step": 5233
-    },
-    {
-      "epoch": 0.4001758510617964,
-      "grad_norm": 0.0012108850060030818,
-      "learning_rate": 0.0001999999212753006,
-      "loss": 46.0,
-      "step": 5234
-    },
-    {
-      "epoch": 0.4002523080451861,
-      "grad_norm": 0.0024835052900016308,
-      "learning_rate": 0.00019999992124515806,
-      "loss": 46.0,
-      "step": 5235
-    },
-    {
-      "epoch": 0.4003287650285758,
-      "grad_norm": 0.0009904626058414578,
-      "learning_rate": 0.00019999992121500979,
-      "loss": 46.0,
-      "step": 5236
-    },
-    {
-      "epoch": 0.40040522201196554,
-      "grad_norm": 0.0016699059633538127,
-      "learning_rate": 0.00019999992118485574,
-      "loss": 46.0,
-      "step": 5237
-    },
-    {
-      "epoch": 0.40048167899535525,
-      "grad_norm": 0.0026765011716634035,
-      "learning_rate": 0.00019999992115469594,
-      "loss": 46.0,
-      "step": 5238
-    },
-    {
-      "epoch": 0.40055813597874496,
-      "grad_norm": 0.0003974362916778773,
-      "learning_rate": 0.00019999992112453035,
-      "loss": 46.0,
-      "step": 5239
-    },
-    {
-      "epoch": 0.40063459296213466,
-      "grad_norm": 0.0038094143383204937,
-      "learning_rate": 0.000199999921094359,
-      "loss": 46.0,
-      "step": 5240
-    },
-    {
-      "epoch": 0.4007110499455244,
-      "grad_norm": 0.0016438699094578624,
-      "learning_rate": 0.00019999992106418187,
-      "loss": 46.0,
-      "step": 5241
-    },
-    {
-      "epoch": 0.40078750692891413,
-      "grad_norm": 0.0009883217280730605,
-      "learning_rate": 0.00019999992103399898,
-      "loss": 46.0,
-      "step": 5242
-    },
-    {
-      "epoch": 0.40086396391230383,
-      "grad_norm": 0.013140978291630745,
-      "learning_rate": 0.00019999992100381032,
-      "loss": 46.0,
-      "step": 5243
-    },
-    {
-      "epoch": 0.40094042089569354,
-      "grad_norm": 0.0010866523953154683,
-      "learning_rate": 0.0001999999209736159,
-      "loss": 46.0,
-      "step": 5244
-    },
-    {
-      "epoch": 0.4010168778790833,
-      "grad_norm": 0.0009235245524905622,
-      "learning_rate": 0.00019999992094341568,
-      "loss": 46.0,
-      "step": 5245
-    },
-    {
-      "epoch": 0.401093334862473,
-      "grad_norm": 0.00375084369443357,
-      "learning_rate": 0.00019999992091320972,
-      "loss": 46.0,
-      "step": 5246
-    },
-    {
-      "epoch": 0.4011697918458627,
-      "grad_norm": 0.0012220956850796938,
-      "learning_rate": 0.000199999920882998,
-      "loss": 46.0,
-      "step": 5247
-    },
-    {
-      "epoch": 0.4012462488292524,
-      "grad_norm": 0.005583017598837614,
-      "learning_rate": 0.00019999992085278047,
-      "loss": 46.0,
-      "step": 5248
-    },
-    {
-      "epoch": 0.4013227058126422,
-      "grad_norm": 0.0007204393041320145,
-      "learning_rate": 0.0001999999208225572,
-      "loss": 46.0,
-      "step": 5249
-    },
-    {
-      "epoch": 0.4013991627960319,
-      "grad_norm": 0.001671968144364655,
-      "learning_rate": 0.00019999992079232818,
-      "loss": 46.0,
-      "step": 5250
-    },
-    {
-      "epoch": 0.4014756197794216,
-      "grad_norm": 0.0012400003615766764,
-      "learning_rate": 0.00019999992076209333,
-      "loss": 46.0,
-      "step": 5251
-    },
-    {
-      "epoch": 0.40155207676281135,
-      "grad_norm": 0.0011241972679272294,
-      "learning_rate": 0.00019999992073185276,
-      "loss": 46.0,
-      "step": 5252
-    },
-    {
-      "epoch": 0.40162853374620106,
-      "grad_norm": 0.00090367830125615,
-      "learning_rate": 0.0001999999207016064,
-      "loss": 46.0,
-      "step": 5253
-    },
-    {
-      "epoch": 0.40170499072959076,
-      "grad_norm": 0.004639594350010157,
-      "learning_rate": 0.00019999992067135428,
-      "loss": 46.0,
-      "step": 5254
-    },
-    {
-      "epoch": 0.40178144771298047,
-      "grad_norm": 0.0009280552621930838,
-      "learning_rate": 0.0001999999206410964,
-      "loss": 46.0,
-      "step": 5255
-    },
-    {
-      "epoch": 0.4018579046963702,
-      "grad_norm": 0.00045384871191345155,
-      "learning_rate": 0.00019999992061083273,
-      "loss": 46.0,
-      "step": 5256
-    },
-    {
-      "epoch": 0.40193436167975993,
-      "grad_norm": 0.01320857647806406,
-      "learning_rate": 0.00019999992058056327,
-      "loss": 46.0,
-      "step": 5257
-    },
-    {
-      "epoch": 0.40201081866314964,
-      "grad_norm": 0.0011672774562612176,
-      "learning_rate": 0.0001999999205502881,
-      "loss": 46.0,
-      "step": 5258
-    },
-    {
-      "epoch": 0.40208727564653934,
-      "grad_norm": 0.002242287155240774,
-      "learning_rate": 0.0001999999205200071,
-      "loss": 46.0,
-      "step": 5259
-    },
-    {
-      "epoch": 0.4021637326299291,
-      "grad_norm": 0.0005927511374466121,
-      "learning_rate": 0.00019999992048972036,
-      "loss": 46.0,
-      "step": 5260
-    },
-    {
-      "epoch": 0.4022401896133188,
-      "grad_norm": 0.0009226332185789943,
-      "learning_rate": 0.00019999992045942783,
-      "loss": 46.0,
-      "step": 5261
-    },
-    {
-      "epoch": 0.4023166465967085,
-      "grad_norm": 0.0016111298464238644,
-      "learning_rate": 0.00019999992042912958,
-      "loss": 46.0,
-      "step": 5262
-    },
-    {
-      "epoch": 0.4023931035800982,
-      "grad_norm": 0.0006652683950960636,
-      "learning_rate": 0.0001999999203988255,
-      "loss": 46.0,
-      "step": 5263
-    },
-    {
-      "epoch": 0.402469560563488,
-      "grad_norm": 0.0025729595217853785,
-      "learning_rate": 0.00019999992036851571,
-      "loss": 46.0,
-      "step": 5264
-    },
-    {
-      "epoch": 0.4025460175468777,
-      "grad_norm": 0.00046787853352725506,
-      "learning_rate": 0.0001999999203382001,
-      "loss": 46.0,
-      "step": 5265
-    },
-    {
-      "epoch": 0.4026224745302674,
-      "grad_norm": 0.0009747404255904257,
-      "learning_rate": 0.00019999992030787875,
-      "loss": 46.0,
-      "step": 5266
-    },
-    {
-      "epoch": 0.40269893151365715,
-      "grad_norm": 0.0012643086956813931,
-      "learning_rate": 0.00019999992027755164,
-      "loss": 46.0,
-      "step": 5267
-    },
-    {
-      "epoch": 0.40277538849704686,
-      "grad_norm": 0.001681259018369019,
-      "learning_rate": 0.00019999992024721875,
-      "loss": 46.0,
-      "step": 5268
-    },
-    {
-      "epoch": 0.40285184548043657,
-      "grad_norm": 0.0027980790473520756,
-      "learning_rate": 0.00019999992021688007,
-      "loss": 46.0,
-      "step": 5269
-    },
-    {
-      "epoch": 0.40292830246382627,
-      "grad_norm": 0.0007206897134892642,
-      "learning_rate": 0.00019999992018653563,
-      "loss": 46.0,
-      "step": 5270
-    },
-    {
-      "epoch": 0.40300475944721603,
-      "grad_norm": 0.0017942889826372266,
-      "learning_rate": 0.00019999992015618543,
-      "loss": 46.0,
-      "step": 5271
-    },
-    {
-      "epoch": 0.40308121643060574,
-      "grad_norm": 0.00030999790760688484,
-      "learning_rate": 0.00019999992012582945,
-      "loss": 46.0,
-      "step": 5272
-    },
-    {
-      "epoch": 0.40315767341399544,
-      "grad_norm": 0.0022347443737089634,
-      "learning_rate": 0.0001999999200954677,
-      "loss": 46.0,
-      "step": 5273
-    },
-    {
-      "epoch": 0.40323413039738515,
-      "grad_norm": 0.0034785778261721134,
-      "learning_rate": 0.0001999999200651002,
-      "loss": 46.0,
-      "step": 5274
-    },
-    {
-      "epoch": 0.4033105873807749,
-      "grad_norm": 0.0009431695216335356,
-      "learning_rate": 0.0001999999200347269,
-      "loss": 46.0,
-      "step": 5275
-    },
-    {
-      "epoch": 0.4033870443641646,
-      "grad_norm": 0.0008784524980001152,
-      "learning_rate": 0.00019999992000434787,
-      "loss": 46.0,
-      "step": 5276
-    },
-    {
-      "epoch": 0.4034635013475543,
-      "grad_norm": 0.0007470921264030039,
-      "learning_rate": 0.00019999991997396302,
-      "loss": 46.0,
-      "step": 5277
-    },
-    {
-      "epoch": 0.403539958330944,
-      "grad_norm": 0.0014381594955921173,
-      "learning_rate": 0.00019999991994357243,
-      "loss": 46.0,
-      "step": 5278
-    },
-    {
-      "epoch": 0.4036164153143338,
-      "grad_norm": 0.002072378294542432,
-      "learning_rate": 0.0001999999199131761,
-      "loss": 46.0,
-      "step": 5279
-    },
-    {
-      "epoch": 0.4036928722977235,
-      "grad_norm": 0.002146427519619465,
-      "learning_rate": 0.00019999991988277395,
-      "loss": 46.0,
-      "step": 5280
-    },
-    {
-      "epoch": 0.4037693292811132,
-      "grad_norm": 0.014893047511577606,
-      "learning_rate": 0.00019999991985236606,
-      "loss": 46.0,
-      "step": 5281
-    },
-    {
-      "epoch": 0.40384578626450296,
-      "grad_norm": 0.001805880106985569,
-      "learning_rate": 0.00019999991982195238,
-      "loss": 46.0,
-      "step": 5282
-    },
-    {
-      "epoch": 0.40392224324789266,
-      "grad_norm": 0.0006186531973071396,
-      "learning_rate": 0.00019999991979153294,
-      "loss": 46.0,
-      "step": 5283
-    },
-    {
-      "epoch": 0.40399870023128237,
-      "grad_norm": 0.0008769570267759264,
-      "learning_rate": 0.0001999999197611077,
-      "loss": 46.0,
-      "step": 5284
-    },
-    {
-      "epoch": 0.4040751572146721,
-      "grad_norm": 0.0013388508232310414,
-      "learning_rate": 0.00019999991973067673,
-      "loss": 46.0,
-      "step": 5285
-    },
-    {
-      "epoch": 0.40415161419806184,
-      "grad_norm": 0.0018591596744954586,
-      "learning_rate": 0.00019999991970024,
-      "loss": 46.0,
-      "step": 5286
-    },
-    {
-      "epoch": 0.40422807118145154,
-      "grad_norm": 0.0010779805015772581,
-      "learning_rate": 0.00019999991966979748,
-      "loss": 46.0,
-      "step": 5287
-    },
-    {
-      "epoch": 0.40430452816484125,
-      "grad_norm": 0.0028559148777276278,
-      "learning_rate": 0.0001999999196393492,
-      "loss": 46.0,
-      "step": 5288
-    },
-    {
-      "epoch": 0.40438098514823095,
-      "grad_norm": 0.010105373337864876,
-      "learning_rate": 0.00019999991960889513,
-      "loss": 46.0,
-      "step": 5289
-    },
-    {
-      "epoch": 0.4044574421316207,
-      "grad_norm": 0.00030190908000804484,
-      "learning_rate": 0.0001999999195784353,
-      "loss": 46.0,
-      "step": 5290
-    },
-    {
-      "epoch": 0.4045338991150104,
-      "grad_norm": 0.0020312201231718063,
-      "learning_rate": 0.00019999991954796972,
-      "loss": 46.0,
-      "step": 5291
-    },
-    {
-      "epoch": 0.4046103560984001,
-      "grad_norm": 0.0010380193125456572,
-      "learning_rate": 0.00019999991951749838,
-      "loss": 46.0,
-      "step": 5292
-    },
-    {
-      "epoch": 0.40468681308178983,
-      "grad_norm": 0.0006743098492734134,
-      "learning_rate": 0.00019999991948702124,
-      "loss": 46.0,
-      "step": 5293
-    },
-    {
-      "epoch": 0.4047632700651796,
-      "grad_norm": 0.0030814683996140957,
-      "learning_rate": 0.00019999991945653833,
-      "loss": 46.0,
-      "step": 5294
-    },
-    {
-      "epoch": 0.4048397270485693,
-      "grad_norm": 0.0006637056358158588,
-      "learning_rate": 0.00019999991942604967,
-      "loss": 46.0,
-      "step": 5295
-    },
-    {
-      "epoch": 0.404916184031959,
-      "grad_norm": 0.0008747492101974785,
-      "learning_rate": 0.0001999999193955552,
-      "loss": 46.0,
-      "step": 5296
-    },
-    {
-      "epoch": 0.40499264101534876,
-      "grad_norm": 0.0016115738544613123,
-      "learning_rate": 0.000199999919365055,
-      "loss": 46.0,
-      "step": 5297
-    },
-    {
-      "epoch": 0.40506909799873847,
-      "grad_norm": 0.0005605354090221226,
-      "learning_rate": 0.00019999991933454902,
-      "loss": 46.0,
-      "step": 5298
-    },
-    {
-      "epoch": 0.4051455549821282,
-      "grad_norm": 0.0011988300830125809,
-      "learning_rate": 0.00019999991930403727,
-      "loss": 46.0,
-      "step": 5299
-    },
-    {
-      "epoch": 0.4052220119655179,
-      "grad_norm": 0.0006165509694255888,
-      "learning_rate": 0.00019999991927351974,
-      "loss": 46.0,
-      "step": 5300
-    },
-    {
-      "epoch": 0.40529846894890764,
-      "grad_norm": 0.0003288340230938047,
-      "learning_rate": 0.00019999991924299647,
-      "loss": 46.0,
-      "step": 5301
-    },
-    {
-      "epoch": 0.40537492593229735,
-      "grad_norm": 0.002198850968852639,
-      "learning_rate": 0.0001999999192124674,
-      "loss": 46.0,
-      "step": 5302
-    },
-    {
-      "epoch": 0.40545138291568705,
-      "grad_norm": 0.0016613348852843046,
-      "learning_rate": 0.00019999991918193257,
-      "loss": 46.0,
-      "step": 5303
-    },
-    {
-      "epoch": 0.40552783989907676,
-      "grad_norm": 0.0013832877157256007,
-      "learning_rate": 0.00019999991915139198,
-      "loss": 46.0,
-      "step": 5304
-    },
-    {
-      "epoch": 0.4056042968824665,
-      "grad_norm": 0.003217297373339534,
-      "learning_rate": 0.00019999991912084562,
-      "loss": 46.0,
-      "step": 5305
-    },
-    {
-      "epoch": 0.4056807538658562,
-      "grad_norm": 0.00459557119756937,
-      "learning_rate": 0.00019999991909029348,
-      "loss": 46.0,
-      "step": 5306
-    },
-    {
-      "epoch": 0.40575721084924593,
-      "grad_norm": 0.0008140060817822814,
-      "learning_rate": 0.00019999991905973556,
-      "loss": 46.0,
-      "step": 5307
-    },
-    {
-      "epoch": 0.40583366783263564,
-      "grad_norm": 0.001100831781513989,
-      "learning_rate": 0.00019999991902917188,
-      "loss": 46.0,
-      "step": 5308
-    },
-    {
-      "epoch": 0.4059101248160254,
-      "grad_norm": 0.0184186939150095,
-      "learning_rate": 0.00019999991899860244,
-      "loss": 46.0,
-      "step": 5309
-    },
-    {
-      "epoch": 0.4059865817994151,
-      "grad_norm": 0.0008051187032833695,
-      "learning_rate": 0.00019999991896802724,
-      "loss": 46.0,
-      "step": 5310
-    },
-    {
-      "epoch": 0.4060630387828048,
-      "grad_norm": 0.003083000658079982,
-      "learning_rate": 0.00019999991893744626,
-      "loss": 46.0,
-      "step": 5311
-    },
-    {
-      "epoch": 0.40613949576619457,
-      "grad_norm": 0.0031733568757772446,
-      "learning_rate": 0.0001999999189068595,
-      "loss": 46.0,
-      "step": 5312
-    },
-    {
-      "epoch": 0.4062159527495843,
-      "grad_norm": 0.0012065396877005696,
-      "learning_rate": 0.00019999991887626698,
-      "loss": 46.0,
-      "step": 5313
-    },
-    {
-      "epoch": 0.406292409732974,
-      "grad_norm": 0.0030355232302099466,
-      "learning_rate": 0.00019999991884566865,
-      "loss": 46.0,
-      "step": 5314
-    },
-    {
-      "epoch": 0.4063688667163637,
-      "grad_norm": 0.000756801397074014,
-      "learning_rate": 0.0001999999188150646,
-      "loss": 46.0,
-      "step": 5315
-    },
-    {
-      "epoch": 0.40644532369975345,
-      "grad_norm": 0.0007659818511456251,
-      "learning_rate": 0.0001999999187844548,
-      "loss": 46.0,
-      "step": 5316
-    },
-    {
-      "epoch": 0.40652178068314315,
-      "grad_norm": 0.0007685855962336063,
-      "learning_rate": 0.0001999999187538392,
-      "loss": 46.0,
-      "step": 5317
-    },
-    {
-      "epoch": 0.40659823766653286,
-      "grad_norm": 0.00038792213308624923,
-      "learning_rate": 0.0001999999187232178,
-      "loss": 46.0,
-      "step": 5318
-    },
-    {
-      "epoch": 0.40667469464992256,
-      "grad_norm": 0.0006638366030529141,
-      "learning_rate": 0.00019999991869259066,
-      "loss": 46.0,
-      "step": 5319
-    },
-    {
-      "epoch": 0.4067511516333123,
-      "grad_norm": 0.0009837057441473007,
-      "learning_rate": 0.00019999991866195778,
-      "loss": 46.0,
-      "step": 5320
-    },
-    {
-      "epoch": 0.40682760861670203,
-      "grad_norm": 0.00202792976051569,
-      "learning_rate": 0.0001999999186313191,
-      "loss": 46.0,
-      "step": 5321
-    },
-    {
-      "epoch": 0.40690406560009174,
-      "grad_norm": 0.003748349379748106,
-      "learning_rate": 0.00019999991860067463,
-      "loss": 46.0,
-      "step": 5322
-    },
-    {
-      "epoch": 0.40698052258348144,
-      "grad_norm": 0.0019859028980135918,
-      "learning_rate": 0.00019999991857002443,
-      "loss": 46.0,
-      "step": 5323
-    },
-    {
-      "epoch": 0.4070569795668712,
-      "grad_norm": 0.0029828983824700117,
-      "learning_rate": 0.00019999991853936845,
-      "loss": 46.0,
-      "step": 5324
-    },
-    {
-      "epoch": 0.4071334365502609,
-      "grad_norm": 0.0022469153627753258,
-      "learning_rate": 0.0001999999185087067,
-      "loss": 46.0,
-      "step": 5325
-    },
-    {
-      "epoch": 0.4072098935336506,
-      "grad_norm": 0.0006606092792935669,
-      "learning_rate": 0.00019999991847803917,
-      "loss": 46.0,
-      "step": 5326
-    },
-    {
-      "epoch": 0.4072863505170404,
-      "grad_norm": 0.004952155519276857,
-      "learning_rate": 0.00019999991844736584,
-      "loss": 46.0,
-      "step": 5327
-    },
-    {
-      "epoch": 0.4073628075004301,
-      "grad_norm": 0.0008421237580478191,
-      "learning_rate": 0.0001999999184166868,
-      "loss": 46.0,
-      "step": 5328
-    },
-    {
-      "epoch": 0.4074392644838198,
-      "grad_norm": 0.0007701292634010315,
-      "learning_rate": 0.00019999991838600195,
-      "loss": 46.0,
-      "step": 5329
-    },
-    {
-      "epoch": 0.4075157214672095,
-      "grad_norm": 0.0025455495342612267,
-      "learning_rate": 0.00019999991835531136,
-      "loss": 46.0,
-      "step": 5330
-    },
-    {
-      "epoch": 0.40759217845059925,
-      "grad_norm": 0.0008730392437428236,
-      "learning_rate": 0.00019999991832461497,
-      "loss": 46.0,
-      "step": 5331
-    },
-    {
-      "epoch": 0.40766863543398896,
-      "grad_norm": 0.0005633785622194409,
-      "learning_rate": 0.00019999991829391286,
-      "loss": 46.0,
-      "step": 5332
-    },
-    {
-      "epoch": 0.40774509241737866,
-      "grad_norm": 0.0009618055773898959,
-      "learning_rate": 0.00019999991826320494,
-      "loss": 46.0,
-      "step": 5333
-    },
-    {
-      "epoch": 0.40782154940076837,
-      "grad_norm": 0.002823295071721077,
-      "learning_rate": 0.00019999991823249126,
-      "loss": 46.0,
-      "step": 5334
-    },
-    {
-      "epoch": 0.40789800638415813,
-      "grad_norm": 0.0013456797460094094,
-      "learning_rate": 0.00019999991820177183,
-      "loss": 46.0,
-      "step": 5335
-    },
-    {
-      "epoch": 0.40797446336754783,
-      "grad_norm": 0.008451217785477638,
-      "learning_rate": 0.00019999991817104657,
-      "loss": 46.0,
-      "step": 5336
-    },
-    {
-      "epoch": 0.40805092035093754,
-      "grad_norm": 0.0044900947250425816,
-      "learning_rate": 0.00019999991814031561,
-      "loss": 46.0,
-      "step": 5337
-    },
-    {
-      "epoch": 0.4081273773343273,
-      "grad_norm": 0.0004067916888743639,
-      "learning_rate": 0.00019999991810957886,
-      "loss": 46.0,
-      "step": 5338
-    },
-    {
-      "epoch": 0.408203834317717,
-      "grad_norm": 0.001792369643226266,
-      "learning_rate": 0.0001999999180788363,
-      "loss": 46.0,
-      "step": 5339
-    },
-    {
-      "epoch": 0.4082802913011067,
-      "grad_norm": 0.0004632076888810843,
-      "learning_rate": 0.000199999918048088,
-      "loss": 46.0,
-      "step": 5340
-    },
-    {
-      "epoch": 0.4083567482844964,
-      "grad_norm": 0.000739187584258616,
-      "learning_rate": 0.00019999991801733394,
-      "loss": 46.0,
-      "step": 5341
-    },
-    {
-      "epoch": 0.4084332052678862,
-      "grad_norm": 0.00245550530962646,
-      "learning_rate": 0.00019999991798657412,
-      "loss": 46.0,
-      "step": 5342
-    },
-    {
-      "epoch": 0.4085096622512759,
-      "grad_norm": 0.0005575842224061489,
-      "learning_rate": 0.0001999999179558085,
-      "loss": 46.0,
-      "step": 5343
-    },
-    {
-      "epoch": 0.4085861192346656,
-      "grad_norm": 0.0010448317043483257,
-      "learning_rate": 0.00019999991792503714,
-      "loss": 46.0,
-      "step": 5344
-    },
-    {
-      "epoch": 0.4086625762180553,
-      "grad_norm": 0.0008700392208993435,
-      "learning_rate": 0.00019999991789425997,
-      "loss": 46.0,
-      "step": 5345
-    },
-    {
-      "epoch": 0.40873903320144506,
-      "grad_norm": 0.0013869213871657848,
-      "learning_rate": 0.00019999991786347706,
-      "loss": 46.0,
-      "step": 5346
-    },
-    {
-      "epoch": 0.40881549018483476,
-      "grad_norm": 0.0014831259613856673,
-      "learning_rate": 0.0001999999178326884,
-      "loss": 46.0,
-      "step": 5347
-    },
-    {
-      "epoch": 0.40889194716822447,
-      "grad_norm": 0.001482793944887817,
-      "learning_rate": 0.00019999991780189394,
-      "loss": 46.0,
-      "step": 5348
-    },
-    {
-      "epoch": 0.4089684041516142,
-      "grad_norm": 0.001930071273818612,
-      "learning_rate": 0.0001999999177710937,
-      "loss": 46.0,
-      "step": 5349
-    },
-    {
-      "epoch": 0.40904486113500393,
-      "grad_norm": 0.001391713391058147,
-      "learning_rate": 0.00019999991774028773,
-      "loss": 46.0,
-      "step": 5350
-    },
-    {
-      "epoch": 0.40912131811839364,
-      "grad_norm": 0.0008259987225756049,
-      "learning_rate": 0.00019999991770947598,
-      "loss": 46.0,
-      "step": 5351
-    },
-    {
-      "epoch": 0.40919777510178335,
-      "grad_norm": 0.0018095808336511254,
-      "learning_rate": 0.00019999991767865843,
-      "loss": 46.0,
-      "step": 5352
-    },
-    {
-      "epoch": 0.4092742320851731,
-      "grad_norm": 0.005817838944494724,
-      "learning_rate": 0.00019999991764783513,
-      "loss": 46.0,
-      "step": 5353
-    },
-    {
-      "epoch": 0.4093506890685628,
-      "grad_norm": 0.0012043097522109747,
-      "learning_rate": 0.00019999991761700608,
-      "loss": 46.0,
-      "step": 5354
-    },
-    {
-      "epoch": 0.4094271460519525,
-      "grad_norm": 0.0006035730475559831,
-      "learning_rate": 0.00019999991758617124,
-      "loss": 46.0,
-      "step": 5355
-    },
-    {
-      "epoch": 0.4095036030353422,
-      "grad_norm": 0.0008884496055543423,
-      "learning_rate": 0.00019999991755533062,
-      "loss": 46.0,
-      "step": 5356
-    },
-    {
-      "epoch": 0.409580060018732,
-      "grad_norm": 0.00195993366651237,
-      "learning_rate": 0.00019999991752448423,
-      "loss": 46.0,
-      "step": 5357
-    },
-    {
-      "epoch": 0.4096565170021217,
-      "grad_norm": 0.0033339839428663254,
-      "learning_rate": 0.0001999999174936321,
-      "loss": 46.0,
-      "step": 5358
-    },
-    {
-      "epoch": 0.4097329739855114,
-      "grad_norm": 0.00281574297696352,
-      "learning_rate": 0.00019999991746277415,
-      "loss": 46.0,
-      "step": 5359
-    },
-    {
-      "epoch": 0.4098094309689011,
-      "grad_norm": 0.0024064439348876476,
-      "learning_rate": 0.0001999999174319105,
-      "loss": 46.0,
-      "step": 5360
-    },
-    {
-      "epoch": 0.40988588795229086,
-      "grad_norm": 0.0013206801377236843,
-      "learning_rate": 0.00019999991740104103,
-      "loss": 46.0,
-      "step": 5361
-    },
-    {
-      "epoch": 0.40996234493568057,
-      "grad_norm": 0.0009637962793931365,
-      "learning_rate": 0.0001999999173701658,
-      "loss": 46.0,
-      "step": 5362
-    },
-    {
-      "epoch": 0.4100388019190703,
-      "grad_norm": 0.0006769727915525436,
-      "learning_rate": 0.0001999999173392848,
-      "loss": 46.0,
-      "step": 5363
-    },
-    {
-      "epoch": 0.41011525890246,
-      "grad_norm": 0.002288165735080838,
-      "learning_rate": 0.00019999991730839804,
-      "loss": 46.0,
-      "step": 5364
-    },
-    {
-      "epoch": 0.41019171588584974,
-      "grad_norm": 0.0014488318702206016,
-      "learning_rate": 0.00019999991727750552,
-      "loss": 46.0,
-      "step": 5365
-    },
-    {
-      "epoch": 0.41026817286923944,
-      "grad_norm": 0.0012545925565063953,
-      "learning_rate": 0.0001999999172466072,
-      "loss": 46.0,
-      "step": 5366
-    },
-    {
-      "epoch": 0.41034462985262915,
-      "grad_norm": 0.00043280309182591736,
-      "learning_rate": 0.00019999991721570312,
-      "loss": 46.0,
-      "step": 5367
-    },
-    {
-      "epoch": 0.4104210868360189,
-      "grad_norm": 0.0018425228772684932,
-      "learning_rate": 0.00019999991718479328,
-      "loss": 46.0,
-      "step": 5368
-    },
-    {
-      "epoch": 0.4104975438194086,
-      "grad_norm": 0.0011807163245975971,
-      "learning_rate": 0.00019999991715387766,
-      "loss": 46.0,
-      "step": 5369
-    },
-    {
-      "epoch": 0.4105740008027983,
-      "grad_norm": 0.002894126810133457,
-      "learning_rate": 0.0001999999171229563,
-      "loss": 46.0,
-      "step": 5370
-    },
-    {
-      "epoch": 0.41065045778618803,
-      "grad_norm": 0.0011282350169494748,
-      "learning_rate": 0.00019999991709202916,
-      "loss": 46.0,
-      "step": 5371
-    },
-    {
-      "epoch": 0.4107269147695778,
-      "grad_norm": 0.003663264913484454,
-      "learning_rate": 0.00019999991706109622,
-      "loss": 46.0,
-      "step": 5372
-    },
-    {
-      "epoch": 0.4108033717529675,
-      "grad_norm": 0.0018011954380199313,
-      "learning_rate": 0.00019999991703015753,
-      "loss": 46.0,
-      "step": 5373
-    },
-    {
-      "epoch": 0.4108798287363572,
-      "grad_norm": 0.0004497392219491303,
-      "learning_rate": 0.00019999991699921307,
-      "loss": 46.0,
-      "step": 5374
-    },
-    {
-      "epoch": 0.4109562857197469,
-      "grad_norm": 0.0007140978123061359,
-      "learning_rate": 0.00019999991696826284,
-      "loss": 46.0,
-      "step": 5375
-    },
-    {
-      "epoch": 0.41103274270313667,
-      "grad_norm": 0.0011017256183549762,
-      "learning_rate": 0.00019999991693730684,
-      "loss": 46.0,
-      "step": 5376
-    },
-    {
-      "epoch": 0.41110919968652637,
-      "grad_norm": 0.0008979947888292372,
-      "learning_rate": 0.0001999999169063451,
-      "loss": 46.0,
-      "step": 5377
-    },
-    {
-      "epoch": 0.4111856566699161,
-      "grad_norm": 0.0008670479874126613,
-      "learning_rate": 0.00019999991687537754,
-      "loss": 46.0,
-      "step": 5378
-    },
-    {
-      "epoch": 0.4112621136533058,
-      "grad_norm": 0.0027278808411210775,
-      "learning_rate": 0.0001999999168444042,
-      "loss": 46.0,
-      "step": 5379
-    },
-    {
-      "epoch": 0.41133857063669554,
-      "grad_norm": 0.0036291528958827257,
-      "learning_rate": 0.00019999991681342517,
-      "loss": 46.0,
-      "step": 5380
-    },
-    {
-      "epoch": 0.41141502762008525,
-      "grad_norm": 0.0005036481888964772,
-      "learning_rate": 0.00019999991678244032,
-      "loss": 46.0,
-      "step": 5381
-    },
-    {
-      "epoch": 0.41149148460347496,
-      "grad_norm": 0.0014522558776661754,
-      "learning_rate": 0.00019999991675144968,
-      "loss": 46.0,
-      "step": 5382
-    },
-    {
-      "epoch": 0.4115679415868647,
-      "grad_norm": 0.0027872086502611637,
-      "learning_rate": 0.00019999991672045334,
-      "loss": 46.0,
-      "step": 5383
-    },
-    {
-      "epoch": 0.4116443985702544,
-      "grad_norm": 0.0008004358969628811,
-      "learning_rate": 0.00019999991668945115,
-      "loss": 46.0,
-      "step": 5384
-    },
-    {
-      "epoch": 0.4117208555536441,
-      "grad_norm": 0.003509378293529153,
-      "learning_rate": 0.00019999991665844324,
-      "loss": 46.0,
-      "step": 5385
-    },
-    {
-      "epoch": 0.41179731253703383,
-      "grad_norm": 0.0010886641684919596,
-      "learning_rate": 0.00019999991662742955,
-      "loss": 46.0,
-      "step": 5386
-    },
-    {
-      "epoch": 0.4118737695204236,
-      "grad_norm": 0.00036593497497960925,
-      "learning_rate": 0.0001999999165964101,
-      "loss": 46.0,
-      "step": 5387
-    },
-    {
-      "epoch": 0.4119502265038133,
-      "grad_norm": 0.0027661616913974285,
-      "learning_rate": 0.00019999991656538486,
-      "loss": 46.0,
-      "step": 5388
-    },
-    {
-      "epoch": 0.412026683487203,
-      "grad_norm": 0.0010463183280080557,
-      "learning_rate": 0.00019999991653435386,
-      "loss": 46.0,
-      "step": 5389
-    },
-    {
-      "epoch": 0.4121031404705927,
-      "grad_norm": 0.0007264424930326641,
-      "learning_rate": 0.00019999991650331708,
-      "loss": 46.0,
-      "step": 5390
-    },
-    {
-      "epoch": 0.41217959745398247,
-      "grad_norm": 0.007100660353899002,
-      "learning_rate": 0.00019999991647227456,
-      "loss": 46.0,
-      "step": 5391
-    },
-    {
-      "epoch": 0.4122560544373722,
-      "grad_norm": 0.0022564150858670473,
-      "learning_rate": 0.00019999991644122623,
-      "loss": 46.0,
-      "step": 5392
-    },
-    {
-      "epoch": 0.4123325114207619,
-      "grad_norm": 0.0013005431974306703,
-      "learning_rate": 0.00019999991641017216,
-      "loss": 46.0,
-      "step": 5393
-    },
-    {
-      "epoch": 0.4124089684041516,
-      "grad_norm": 0.0011177370324730873,
-      "learning_rate": 0.00019999991637911232,
-      "loss": 46.0,
-      "step": 5394
-    },
-    {
-      "epoch": 0.41248542538754135,
-      "grad_norm": 0.0016524213133379817,
-      "learning_rate": 0.0001999999163480467,
-      "loss": 46.0,
-      "step": 5395
-    },
-    {
-      "epoch": 0.41256188237093105,
-      "grad_norm": 0.0004972158349119127,
-      "learning_rate": 0.00019999991631697533,
-      "loss": 46.0,
-      "step": 5396
-    },
-    {
-      "epoch": 0.41263833935432076,
-      "grad_norm": 0.0005656633293256164,
-      "learning_rate": 0.00019999991628589814,
-      "loss": 46.0,
-      "step": 5397
-    },
-    {
-      "epoch": 0.4127147963377105,
-      "grad_norm": 0.0014298890018835664,
-      "learning_rate": 0.00019999991625481523,
-      "loss": 46.0,
-      "step": 5398
-    },
-    {
-      "epoch": 0.4127912533211002,
-      "grad_norm": 0.0006675570621155202,
-      "learning_rate": 0.00019999991622372655,
-      "loss": 46.0,
-      "step": 5399
-    },
-    {
-      "epoch": 0.41286771030448993,
-      "grad_norm": 0.0020149063784629107,
-      "learning_rate": 0.00019999991619263206,
-      "loss": 46.0,
-      "step": 5400
-    },
-    {
-      "epoch": 0.41294416728787964,
-      "grad_norm": 0.0011675046989694238,
-      "learning_rate": 0.00019999991616153183,
-      "loss": 46.0,
-      "step": 5401
-    },
-    {
-      "epoch": 0.4130206242712694,
-      "grad_norm": 0.0007131360471248627,
-      "learning_rate": 0.00019999991613042583,
-      "loss": 46.0,
-      "step": 5402
-    },
-    {
-      "epoch": 0.4130970812546591,
-      "grad_norm": 0.0005054062348790467,
-      "learning_rate": 0.00019999991609931405,
-      "loss": 46.0,
-      "step": 5403
-    },
-    {
-      "epoch": 0.4131735382380488,
-      "grad_norm": 0.0007717363187111914,
-      "learning_rate": 0.0001999999160681965,
-      "loss": 46.0,
-      "step": 5404
-    },
-    {
-      "epoch": 0.4132499952214385,
-      "grad_norm": 0.0008846339769661427,
-      "learning_rate": 0.0001999999160370732,
-      "loss": 46.0,
-      "step": 5405
-    },
-    {
-      "epoch": 0.4133264522048283,
-      "grad_norm": 0.00278662983328104,
-      "learning_rate": 0.00019999991600594413,
-      "loss": 46.0,
-      "step": 5406
-    },
-    {
-      "epoch": 0.413402909188218,
-      "grad_norm": 0.004176516085863113,
-      "learning_rate": 0.00019999991597480926,
-      "loss": 46.0,
-      "step": 5407
-    },
-    {
-      "epoch": 0.4134793661716077,
-      "grad_norm": 0.0008209343650378287,
-      "learning_rate": 0.00019999991594366864,
-      "loss": 46.0,
-      "step": 5408
-    },
-    {
-      "epoch": 0.4135558231549974,
-      "grad_norm": 0.000903637264855206,
-      "learning_rate": 0.00019999991591252228,
-      "loss": 46.0,
-      "step": 5409
-    },
-    {
-      "epoch": 0.41363228013838715,
-      "grad_norm": 0.0033453076612204313,
-      "learning_rate": 0.00019999991588137011,
-      "loss": 46.0,
-      "step": 5410
-    },
-    {
-      "epoch": 0.41370873712177686,
-      "grad_norm": 0.0020626818295568228,
-      "learning_rate": 0.00019999991585021215,
-      "loss": 46.0,
-      "step": 5411
-    },
-    {
-      "epoch": 0.41378519410516656,
-      "grad_norm": 0.002370130503550172,
-      "learning_rate": 0.00019999991581904847,
-      "loss": 46.0,
-      "step": 5412
-    },
-    {
-      "epoch": 0.4138616510885563,
-      "grad_norm": 0.0008303317590616643,
-      "learning_rate": 0.000199999915787879,
-      "loss": 46.0,
-      "step": 5413
-    },
-    {
-      "epoch": 0.41393810807194603,
-      "grad_norm": 0.005083939526230097,
-      "learning_rate": 0.00019999991575670375,
-      "loss": 46.0,
-      "step": 5414
-    },
-    {
-      "epoch": 0.41401456505533574,
-      "grad_norm": 0.001392135163769126,
-      "learning_rate": 0.00019999991572552277,
-      "loss": 46.0,
-      "step": 5415
-    },
-    {
-      "epoch": 0.41409102203872544,
-      "grad_norm": 0.0025353056844323874,
-      "learning_rate": 0.000199999915694336,
-      "loss": 46.0,
-      "step": 5416
-    },
-    {
-      "epoch": 0.4141674790221152,
-      "grad_norm": 0.003599467221647501,
-      "learning_rate": 0.00019999991566314345,
-      "loss": 46.0,
-      "step": 5417
-    },
-    {
-      "epoch": 0.4142439360055049,
-      "grad_norm": 0.0011562509462237358,
-      "learning_rate": 0.00019999991563194512,
-      "loss": 46.0,
-      "step": 5418
-    },
-    {
-      "epoch": 0.4143203929888946,
-      "grad_norm": 0.0017617044504731894,
-      "learning_rate": 0.00019999991560074105,
-      "loss": 46.0,
-      "step": 5419
-    },
-    {
-      "epoch": 0.4143968499722843,
-      "grad_norm": 0.0018615226726979017,
-      "learning_rate": 0.00019999991556953118,
-      "loss": 46.0,
-      "step": 5420
-    },
-    {
-      "epoch": 0.4144733069556741,
-      "grad_norm": 0.0012078463332727551,
-      "learning_rate": 0.00019999991553831556,
-      "loss": 46.0,
-      "step": 5421
-    },
-    {
-      "epoch": 0.4145497639390638,
-      "grad_norm": 0.0008750527631491423,
-      "learning_rate": 0.00019999991550709417,
-      "loss": 46.0,
-      "step": 5422
-    },
-    {
-      "epoch": 0.4146262209224535,
-      "grad_norm": 0.0006168300751596689,
-      "learning_rate": 0.00019999991547586704,
-      "loss": 46.0,
-      "step": 5423
-    },
-    {
-      "epoch": 0.4147026779058432,
-      "grad_norm": 0.0006413274095393717,
-      "learning_rate": 0.0001999999154446341,
-      "loss": 46.0,
-      "step": 5424
-    },
-    {
-      "epoch": 0.41477913488923296,
-      "grad_norm": 0.000596673518884927,
-      "learning_rate": 0.0001999999154133954,
-      "loss": 46.0,
-      "step": 5425
-    },
-    {
-      "epoch": 0.41485559187262266,
-      "grad_norm": 0.0011239356826990843,
-      "learning_rate": 0.00019999991538215093,
-      "loss": 46.0,
-      "step": 5426
-    },
-    {
-      "epoch": 0.41493204885601237,
-      "grad_norm": 0.002828011056408286,
-      "learning_rate": 0.00019999991535090067,
-      "loss": 46.0,
-      "step": 5427
-    },
-    {
-      "epoch": 0.41500850583940213,
-      "grad_norm": 0.000898508820682764,
-      "learning_rate": 0.00019999991531964467,
-      "loss": 46.0,
-      "step": 5428
-    },
-    {
-      "epoch": 0.41508496282279184,
-      "grad_norm": 0.0015465901233255863,
-      "learning_rate": 0.0001999999152883829,
-      "loss": 46.0,
-      "step": 5429
-    },
-    {
-      "epoch": 0.41516141980618154,
-      "grad_norm": 0.0009853861993178725,
-      "learning_rate": 0.00019999991525711534,
-      "loss": 46.0,
-      "step": 5430
-    },
-    {
-      "epoch": 0.41523787678957125,
-      "grad_norm": 0.0017397133633494377,
-      "learning_rate": 0.000199999915225842,
-      "loss": 46.0,
-      "step": 5431
-    },
-    {
-      "epoch": 0.415314333772961,
-      "grad_norm": 0.0005541547434404492,
-      "learning_rate": 0.00019999991519456295,
-      "loss": 46.0,
-      "step": 5432
-    },
-    {
-      "epoch": 0.4153907907563507,
-      "grad_norm": 0.0007481400389224291,
-      "learning_rate": 0.00019999991516327808,
-      "loss": 46.0,
-      "step": 5433
-    },
-    {
-      "epoch": 0.4154672477397404,
-      "grad_norm": 0.0007982912939041853,
-      "learning_rate": 0.00019999991513198744,
-      "loss": 46.0,
-      "step": 5434
-    },
-    {
-      "epoch": 0.4155437047231301,
-      "grad_norm": 0.0035754344426095486,
-      "learning_rate": 0.00019999991510069105,
-      "loss": 46.0,
-      "step": 5435
-    },
-    {
-      "epoch": 0.4156201617065199,
-      "grad_norm": 0.0020971496123820543,
-      "learning_rate": 0.00019999991506938888,
-      "loss": 46.0,
-      "step": 5436
-    },
-    {
-      "epoch": 0.4156966186899096,
-      "grad_norm": 0.0015283911488950253,
-      "learning_rate": 0.00019999991503808094,
-      "loss": 46.0,
-      "step": 5437
-    },
-    {
-      "epoch": 0.4157730756732993,
-      "grad_norm": 0.0027436655946075916,
-      "learning_rate": 0.00019999991500676723,
-      "loss": 46.0,
-      "step": 5438
-    },
-    {
-      "epoch": 0.415849532656689,
-      "grad_norm": 0.0018711426528170705,
-      "learning_rate": 0.00019999991497544778,
-      "loss": 46.0,
-      "step": 5439
-    },
-    {
-      "epoch": 0.41592598964007876,
-      "grad_norm": 0.013819126412272453,
-      "learning_rate": 0.00019999991494412255,
-      "loss": 46.0,
-      "step": 5440
-    },
-    {
-      "epoch": 0.41600244662346847,
-      "grad_norm": 0.0015264011453837156,
-      "learning_rate": 0.00019999991491279152,
-      "loss": 46.0,
-      "step": 5441
-    },
-    {
-      "epoch": 0.4160789036068582,
-      "grad_norm": 0.001619368209503591,
-      "learning_rate": 0.00019999991488145474,
-      "loss": 46.0,
-      "step": 5442
-    },
-    {
-      "epoch": 0.41615536059024794,
-      "grad_norm": 0.0025840725284069777,
-      "learning_rate": 0.0001999999148501122,
-      "loss": 46.0,
-      "step": 5443
-    },
-    {
-      "epoch": 0.41623181757363764,
-      "grad_norm": 0.0010630057659000158,
-      "learning_rate": 0.00019999991481876387,
-      "loss": 46.0,
-      "step": 5444
-    },
-    {
-      "epoch": 0.41630827455702735,
-      "grad_norm": 0.003099302761256695,
-      "learning_rate": 0.00019999991478740977,
-      "loss": 46.0,
-      "step": 5445
-    },
-    {
-      "epoch": 0.41638473154041705,
-      "grad_norm": 0.0006178977782838047,
-      "learning_rate": 0.00019999991475604993,
-      "loss": 46.0,
-      "step": 5446
-    },
-    {
-      "epoch": 0.4164611885238068,
-      "grad_norm": 0.0009201104403473437,
-      "learning_rate": 0.0001999999147246843,
-      "loss": 46.0,
-      "step": 5447
-    },
-    {
-      "epoch": 0.4165376455071965,
-      "grad_norm": 0.0019779487047344446,
-      "learning_rate": 0.0001999999146933129,
-      "loss": 46.0,
-      "step": 5448
-    },
-    {
-      "epoch": 0.4166141024905862,
-      "grad_norm": 0.001476823934353888,
-      "learning_rate": 0.00019999991466193573,
-      "loss": 46.0,
-      "step": 5449
-    },
-    {
-      "epoch": 0.41669055947397593,
-      "grad_norm": 0.00114912586286664,
-      "learning_rate": 0.00019999991463055282,
-      "loss": 46.0,
-      "step": 5450
-    },
-    {
-      "epoch": 0.4167670164573657,
-      "grad_norm": 0.001857487135566771,
-      "learning_rate": 0.00019999991459916408,
-      "loss": 46.0,
-      "step": 5451
-    },
-    {
-      "epoch": 0.4168434734407554,
-      "grad_norm": 0.0025243971031159163,
-      "learning_rate": 0.0001999999145677696,
-      "loss": 46.0,
-      "step": 5452
-    },
-    {
-      "epoch": 0.4169199304241451,
-      "grad_norm": 0.001984777394682169,
-      "learning_rate": 0.00019999991453636937,
-      "loss": 46.0,
-      "step": 5453
-    },
-    {
-      "epoch": 0.41699638740753486,
-      "grad_norm": 0.0016920112539082766,
-      "learning_rate": 0.00019999991450496337,
-      "loss": 46.0,
-      "step": 5454
-    },
-    {
-      "epoch": 0.41707284439092457,
-      "grad_norm": 0.001000522868707776,
-      "learning_rate": 0.00019999991447355156,
-      "loss": 46.0,
-      "step": 5455
-    },
-    {
-      "epoch": 0.4171493013743143,
-      "grad_norm": 0.0017415520269423723,
-      "learning_rate": 0.00019999991444213401,
-      "loss": 46.0,
-      "step": 5456
-    },
-    {
-      "epoch": 0.417225758357704,
-      "grad_norm": 0.0022787717171013355,
-      "learning_rate": 0.0001999999144107107,
-      "loss": 46.0,
-      "step": 5457
-    },
-    {
-      "epoch": 0.41730221534109374,
-      "grad_norm": 0.0020467701833695173,
-      "learning_rate": 0.0001999999143792816,
-      "loss": 46.0,
-      "step": 5458
-    },
-    {
-      "epoch": 0.41737867232448345,
-      "grad_norm": 0.0035578913521021605,
-      "learning_rate": 0.00019999991434784673,
-      "loss": 46.0,
-      "step": 5459
-    },
-    {
-      "epoch": 0.41745512930787315,
-      "grad_norm": 0.0009780119871720672,
-      "learning_rate": 0.0001999999143164061,
-      "loss": 46.0,
-      "step": 5460
-    },
-    {
-      "epoch": 0.41753158629126286,
-      "grad_norm": 0.0006003855960443616,
-      "learning_rate": 0.0001999999142849597,
-      "loss": 46.0,
-      "step": 5461
-    },
-    {
-      "epoch": 0.4176080432746526,
-      "grad_norm": 0.0014704280765727162,
-      "learning_rate": 0.00019999991425350753,
-      "loss": 46.0,
-      "step": 5462
-    },
-    {
-      "epoch": 0.4176845002580423,
-      "grad_norm": 0.0006409023771993816,
-      "learning_rate": 0.00019999991422204957,
-      "loss": 46.0,
-      "step": 5463
-    },
-    {
-      "epoch": 0.41776095724143203,
-      "grad_norm": 0.0013225938891991973,
-      "learning_rate": 0.00019999991419058588,
-      "loss": 46.0,
-      "step": 5464
-    },
-    {
-      "epoch": 0.41783741422482173,
-      "grad_norm": 0.0017730999970808625,
-      "learning_rate": 0.0001999999141591164,
-      "loss": 46.0,
-      "step": 5465
-    },
-    {
-      "epoch": 0.4179138712082115,
-      "grad_norm": 0.0006660941289737821,
-      "learning_rate": 0.00019999991412764114,
-      "loss": 46.0,
-      "step": 5466
-    },
-    {
-      "epoch": 0.4179903281916012,
-      "grad_norm": 0.0005924385040998459,
-      "learning_rate": 0.00019999991409616014,
-      "loss": 46.0,
-      "step": 5467
-    },
-    {
-      "epoch": 0.4180667851749909,
-      "grad_norm": 0.0013160814996808767,
-      "learning_rate": 0.00019999991406467337,
-      "loss": 46.0,
-      "step": 5468
-    },
-    {
-      "epoch": 0.41814324215838067,
-      "grad_norm": 0.0009196484461426735,
-      "learning_rate": 0.0001999999140331808,
-      "loss": 46.0,
-      "step": 5469
-    },
-    {
-      "epoch": 0.4182196991417704,
-      "grad_norm": 0.0006729019223712385,
-      "learning_rate": 0.00019999991400168247,
-      "loss": 46.0,
-      "step": 5470
-    },
-    {
-      "epoch": 0.4182961561251601,
-      "grad_norm": 0.0006046950584277511,
-      "learning_rate": 0.0001999999139701784,
-      "loss": 46.0,
-      "step": 5471
-    },
-    {
-      "epoch": 0.4183726131085498,
-      "grad_norm": 0.0005343265365809202,
-      "learning_rate": 0.0001999999139386685,
-      "loss": 46.0,
-      "step": 5472
-    },
-    {
-      "epoch": 0.41844907009193955,
-      "grad_norm": 0.0011145288590341806,
-      "learning_rate": 0.00019999991390715286,
-      "loss": 46.0,
-      "step": 5473
-    },
-    {
-      "epoch": 0.41852552707532925,
-      "grad_norm": 0.0025561274960637093,
-      "learning_rate": 0.0001999999138756315,
-      "loss": 46.0,
-      "step": 5474
-    },
-    {
-      "epoch": 0.41860198405871896,
-      "grad_norm": 0.0004951556329615414,
-      "learning_rate": 0.0001999999138441043,
-      "loss": 46.0,
-      "step": 5475
-    },
-    {
-      "epoch": 0.41867844104210866,
-      "grad_norm": 0.0014897570945322514,
-      "learning_rate": 0.00019999991381257134,
-      "loss": 46.0,
-      "step": 5476
-    },
-    {
-      "epoch": 0.4187548980254984,
-      "grad_norm": 0.0034681472461670637,
-      "learning_rate": 0.00019999991378103263,
-      "loss": 46.0,
-      "step": 5477
-    },
-    {
-      "epoch": 0.41883135500888813,
-      "grad_norm": 0.00209251930937171,
-      "learning_rate": 0.00019999991374948818,
-      "loss": 46.0,
-      "step": 5478
-    },
-    {
-      "epoch": 0.41890781199227783,
-      "grad_norm": 0.005612328182905912,
-      "learning_rate": 0.00019999991371793792,
-      "loss": 46.0,
-      "step": 5479
-    },
-    {
-      "epoch": 0.41898426897566754,
-      "grad_norm": 0.0005825295811519027,
-      "learning_rate": 0.0001999999136863819,
-      "loss": 46.0,
-      "step": 5480
-    },
-    {
-      "epoch": 0.4190607259590573,
-      "grad_norm": 0.002547320444136858,
-      "learning_rate": 0.00019999991365482012,
-      "loss": 46.0,
-      "step": 5481
-    },
-    {
-      "epoch": 0.419137182942447,
-      "grad_norm": 0.002230370882898569,
-      "learning_rate": 0.00019999991362325254,
-      "loss": 46.0,
-      "step": 5482
-    },
-    {
-      "epoch": 0.4192136399258367,
-      "grad_norm": 0.0015402492135763168,
-      "learning_rate": 0.0001999999135916792,
-      "loss": 46.0,
-      "step": 5483
-    },
-    {
-      "epoch": 0.4192900969092265,
-      "grad_norm": 0.00045340106589719653,
-      "learning_rate": 0.0001999999135601001,
-      "loss": 46.0,
-      "step": 5484
-    },
-    {
-      "epoch": 0.4193665538926162,
-      "grad_norm": 0.0008690215181559324,
-      "learning_rate": 0.00019999991352851522,
-      "loss": 46.0,
-      "step": 5485
-    },
-    {
-      "epoch": 0.4194430108760059,
-      "grad_norm": 0.0009650334832258523,
-      "learning_rate": 0.0001999999134969246,
-      "loss": 46.0,
-      "step": 5486
-    },
-    {
-      "epoch": 0.4195194678593956,
-      "grad_norm": 0.0008257623412646353,
-      "learning_rate": 0.0001999999134653282,
-      "loss": 46.0,
-      "step": 5487
-    },
-    {
-      "epoch": 0.41959592484278535,
-      "grad_norm": 0.0012264044489711523,
-      "learning_rate": 0.00019999991343372603,
-      "loss": 46.0,
-      "step": 5488
-    },
-    {
-      "epoch": 0.41967238182617506,
-      "grad_norm": 0.0012948067160323262,
-      "learning_rate": 0.00019999991340211807,
-      "loss": 46.0,
-      "step": 5489
-    },
-    {
-      "epoch": 0.41974883880956476,
-      "grad_norm": 0.0014125751331448555,
-      "learning_rate": 0.00019999991337050436,
-      "loss": 46.0,
-      "step": 5490
-    },
-    {
-      "epoch": 0.41982529579295447,
-      "grad_norm": 0.002354119438678026,
-      "learning_rate": 0.00019999991333888488,
-      "loss": 46.0,
-      "step": 5491
-    },
-    {
-      "epoch": 0.41990175277634423,
-      "grad_norm": 0.0032949866726994514,
-      "learning_rate": 0.00019999991330725962,
-      "loss": 46.0,
-      "step": 5492
-    },
-    {
-      "epoch": 0.41997820975973393,
-      "grad_norm": 0.0016361873131245375,
-      "learning_rate": 0.0001999999132756286,
-      "loss": 46.0,
-      "step": 5493
-    },
-    {
-      "epoch": 0.42005466674312364,
-      "grad_norm": 0.002817495260387659,
-      "learning_rate": 0.00019999991324399182,
-      "loss": 46.0,
-      "step": 5494
-    },
-    {
-      "epoch": 0.42013112372651334,
-      "grad_norm": 0.003947072196751833,
-      "learning_rate": 0.00019999991321234927,
-      "loss": 46.0,
-      "step": 5495
-    },
-    {
-      "epoch": 0.4202075807099031,
-      "grad_norm": 0.0008553503430448472,
-      "learning_rate": 0.00019999991318070092,
-      "loss": 46.0,
-      "step": 5496
-    },
-    {
-      "epoch": 0.4202840376932928,
-      "grad_norm": 0.0009838691912591457,
-      "learning_rate": 0.00019999991314904682,
-      "loss": 46.0,
-      "step": 5497
-    },
-    {
-      "epoch": 0.4203604946766825,
-      "grad_norm": 0.0006458622519858181,
-      "learning_rate": 0.00019999991311738695,
-      "loss": 46.0,
-      "step": 5498
-    },
-    {
-      "epoch": 0.4204369516600723,
-      "grad_norm": 0.001247179927304387,
-      "learning_rate": 0.0001999999130857213,
-      "loss": 46.0,
-      "step": 5499
-    },
-    {
-      "epoch": 0.420513408643462,
-      "grad_norm": 0.000877166457939893,
-      "learning_rate": 0.00019999991305404987,
-      "loss": 46.0,
-      "step": 5500
-    },
-    {
-      "epoch": 0.4205898656268517,
-      "grad_norm": 0.0005960864364169538,
-      "learning_rate": 0.00019999991302237273,
-      "loss": 46.0,
-      "step": 5501
-    },
-    {
-      "epoch": 0.4206663226102414,
-      "grad_norm": 0.0007253667572513223,
-      "learning_rate": 0.00019999991299068977,
-      "loss": 46.0,
-      "step": 5502
-    },
-    {
-      "epoch": 0.42074277959363116,
-      "grad_norm": 0.0023964273277670145,
-      "learning_rate": 0.00019999991295900104,
-      "loss": 46.0,
-      "step": 5503
-    },
-    {
-      "epoch": 0.42081923657702086,
-      "grad_norm": 0.0016958360793069005,
-      "learning_rate": 0.00019999991292730655,
-      "loss": 46.0,
-      "step": 5504
-    },
-    {
-      "epoch": 0.42089569356041057,
-      "grad_norm": 0.0007424805662594736,
-      "learning_rate": 0.00019999991289560632,
-      "loss": 46.0,
-      "step": 5505
-    },
-    {
-      "epoch": 0.42097215054380027,
-      "grad_norm": 0.0007164208800531924,
-      "learning_rate": 0.0001999999128639003,
-      "loss": 46.0,
-      "step": 5506
-    },
-    {
-      "epoch": 0.42104860752719003,
-      "grad_norm": 0.001105301664210856,
-      "learning_rate": 0.0001999999128321885,
-      "loss": 46.0,
-      "step": 5507
-    },
-    {
-      "epoch": 0.42112506451057974,
-      "grad_norm": 0.000531343393959105,
-      "learning_rate": 0.00019999991280047092,
-      "loss": 46.0,
-      "step": 5508
-    },
-    {
-      "epoch": 0.42120152149396944,
-      "grad_norm": 0.0007272218936122954,
-      "learning_rate": 0.0001999999127687476,
-      "loss": 46.0,
-      "step": 5509
-    },
-    {
-      "epoch": 0.42127797847735915,
-      "grad_norm": 0.005331685766577721,
-      "learning_rate": 0.00019999991273701847,
-      "loss": 46.0,
-      "step": 5510
-    },
-    {
-      "epoch": 0.4213544354607489,
-      "grad_norm": 0.0007073844317346811,
-      "learning_rate": 0.00019999991270528363,
-      "loss": 46.0,
-      "step": 5511
-    },
-    {
-      "epoch": 0.4214308924441386,
-      "grad_norm": 0.0015893859090283513,
-      "learning_rate": 0.00019999991267354296,
-      "loss": 46.0,
-      "step": 5512
-    },
-    {
-      "epoch": 0.4215073494275283,
-      "grad_norm": 0.002210618695244193,
-      "learning_rate": 0.00019999991264179655,
-      "loss": 46.0,
-      "step": 5513
-    },
-    {
-      "epoch": 0.4215838064109181,
-      "grad_norm": 0.001384711591526866,
-      "learning_rate": 0.00019999991261004439,
-      "loss": 46.0,
-      "step": 5514
-    },
-    {
-      "epoch": 0.4216602633943078,
-      "grad_norm": 0.0009261943050660193,
-      "learning_rate": 0.0001999999125782864,
-      "loss": 46.0,
-      "step": 5515
-    },
-    {
-      "epoch": 0.4217367203776975,
-      "grad_norm": 0.001576651236973703,
-      "learning_rate": 0.00019999991254652272,
-      "loss": 46.0,
-      "step": 5516
-    },
-    {
-      "epoch": 0.4218131773610872,
-      "grad_norm": 0.001993928337469697,
-      "learning_rate": 0.0001999999125147532,
-      "loss": 46.0,
-      "step": 5517
-    },
-    {
-      "epoch": 0.42188963434447696,
-      "grad_norm": 0.0009874376701191068,
-      "learning_rate": 0.00019999991248297792,
-      "loss": 46.0,
-      "step": 5518
-    },
-    {
-      "epoch": 0.42196609132786667,
-      "grad_norm": 0.002483912743628025,
-      "learning_rate": 0.00019999991245119692,
-      "loss": 46.0,
-      "step": 5519
-    },
-    {
-      "epoch": 0.42204254831125637,
-      "grad_norm": 0.0018782797269523144,
-      "learning_rate": 0.00019999991241941012,
-      "loss": 46.0,
-      "step": 5520
-    },
-    {
-      "epoch": 0.4221190052946461,
-      "grad_norm": 0.0009184435475617647,
-      "learning_rate": 0.00019999991238761755,
-      "loss": 46.0,
-      "step": 5521
-    },
-    {
-      "epoch": 0.42219546227803584,
-      "grad_norm": 0.0013539415085688233,
-      "learning_rate": 0.0001999999123558192,
-      "loss": 46.0,
-      "step": 5522
-    },
-    {
-      "epoch": 0.42227191926142554,
-      "grad_norm": 0.0007348635117523372,
-      "learning_rate": 0.0001999999123240151,
-      "loss": 46.0,
-      "step": 5523
-    },
-    {
-      "epoch": 0.42234837624481525,
-      "grad_norm": 0.0018051530933007598,
-      "learning_rate": 0.00019999991229220526,
-      "loss": 46.0,
-      "step": 5524
-    },
-    {
-      "epoch": 0.42242483322820495,
-      "grad_norm": 0.0007736569968983531,
-      "learning_rate": 0.0001999999122603896,
-      "loss": 46.0,
-      "step": 5525
-    },
-    {
-      "epoch": 0.4225012902115947,
-      "grad_norm": 0.002512391423806548,
-      "learning_rate": 0.00019999991222856818,
-      "loss": 46.0,
-      "step": 5526
-    },
-    {
-      "epoch": 0.4225777471949844,
-      "grad_norm": 0.0014235122362151742,
-      "learning_rate": 0.000199999912196741,
-      "loss": 46.0,
-      "step": 5527
-    },
-    {
-      "epoch": 0.4226542041783741,
-      "grad_norm": 0.0014216542476788163,
-      "learning_rate": 0.00019999991216490803,
-      "loss": 46.0,
-      "step": 5528
-    },
-    {
-      "epoch": 0.4227306611617639,
-      "grad_norm": 0.0017427793936803937,
-      "learning_rate": 0.0001999999121330693,
-      "loss": 46.0,
-      "step": 5529
-    },
-    {
-      "epoch": 0.4228071181451536,
-      "grad_norm": 0.0021457301918417215,
-      "learning_rate": 0.0001999999121012248,
-      "loss": 46.0,
-      "step": 5530
-    },
-    {
-      "epoch": 0.4228835751285433,
-      "grad_norm": 0.0008998525445349514,
-      "learning_rate": 0.00019999991206937456,
-      "loss": 46.0,
-      "step": 5531
-    },
-    {
-      "epoch": 0.422960032111933,
-      "grad_norm": 0.0026607727631926537,
-      "learning_rate": 0.00019999991203751853,
-      "loss": 46.0,
-      "step": 5532
-    },
-    {
-      "epoch": 0.42303648909532277,
-      "grad_norm": 0.0008612971287220716,
-      "learning_rate": 0.00019999991200565673,
-      "loss": 46.0,
-      "step": 5533
-    },
-    {
-      "epoch": 0.42311294607871247,
-      "grad_norm": 0.00044479413190856576,
-      "learning_rate": 0.00019999991197378915,
-      "loss": 46.0,
-      "step": 5534
-    },
-    {
-      "epoch": 0.4231894030621022,
-      "grad_norm": 0.0010325514012947679,
-      "learning_rate": 0.0001999999119419158,
-      "loss": 46.0,
-      "step": 5535
-    },
-    {
-      "epoch": 0.4232658600454919,
-      "grad_norm": 0.001145778689533472,
-      "learning_rate": 0.0001999999119100367,
-      "loss": 46.0,
-      "step": 5536
-    },
-    {
-      "epoch": 0.42334231702888164,
-      "grad_norm": 0.0009772671619430184,
-      "learning_rate": 0.00019999991187815184,
-      "loss": 46.0,
-      "step": 5537
-    },
-    {
-      "epoch": 0.42341877401227135,
-      "grad_norm": 0.0007602735422551632,
-      "learning_rate": 0.00019999991184626117,
-      "loss": 46.0,
-      "step": 5538
-    },
-    {
-      "epoch": 0.42349523099566105,
-      "grad_norm": 0.00092788360780105,
-      "learning_rate": 0.00019999991181436478,
-      "loss": 46.0,
-      "step": 5539
-    },
-    {
-      "epoch": 0.42357168797905076,
-      "grad_norm": 0.000411903893109411,
-      "learning_rate": 0.00019999991178246257,
-      "loss": 46.0,
-      "step": 5540
-    },
-    {
-      "epoch": 0.4236481449624405,
-      "grad_norm": 0.0018015257082879543,
-      "learning_rate": 0.0001999999117505546,
-      "loss": 46.0,
-      "step": 5541
-    },
-    {
-      "epoch": 0.4237246019458302,
-      "grad_norm": 0.0008646459900774062,
-      "learning_rate": 0.0001999999117186409,
-      "loss": 46.0,
-      "step": 5542
-    },
-    {
-      "epoch": 0.42380105892921993,
-      "grad_norm": 0.0024136521387845278,
-      "learning_rate": 0.0001999999116867214,
-      "loss": 46.0,
-      "step": 5543
-    },
-    {
-      "epoch": 0.4238775159126097,
-      "grad_norm": 0.0017535907682031393,
-      "learning_rate": 0.00019999991165479614,
-      "loss": 46.0,
-      "step": 5544
-    },
-    {
-      "epoch": 0.4239539728959994,
-      "grad_norm": 0.0022689446341246367,
-      "learning_rate": 0.00019999991162286508,
-      "loss": 46.0,
-      "step": 5545
-    },
-    {
-      "epoch": 0.4240304298793891,
-      "grad_norm": 0.002755317371338606,
-      "learning_rate": 0.0001999999115909283,
-      "loss": 46.0,
-      "step": 5546
-    },
-    {
-      "epoch": 0.4241068868627788,
-      "grad_norm": 0.0006008046912029386,
-      "learning_rate": 0.00019999991155898573,
-      "loss": 46.0,
-      "step": 5547
-    },
-    {
-      "epoch": 0.42418334384616857,
-      "grad_norm": 0.0027862340211868286,
-      "learning_rate": 0.00019999991152703736,
-      "loss": 46.0,
-      "step": 5548
-    },
-    {
-      "epoch": 0.4242598008295583,
-      "grad_norm": 0.000724279903806746,
-      "learning_rate": 0.00019999991149508326,
-      "loss": 46.0,
-      "step": 5549
-    },
-    {
-      "epoch": 0.424336257812948,
-      "grad_norm": 0.0008108525071293116,
-      "learning_rate": 0.00019999991146312337,
-      "loss": 46.0,
-      "step": 5550
-    },
-    {
-      "epoch": 0.4244127147963377,
-      "grad_norm": 0.0012870472855865955,
-      "learning_rate": 0.00019999991143115773,
-      "loss": 46.0,
-      "step": 5551
-    },
-    {
-      "epoch": 0.42448917177972745,
-      "grad_norm": 0.0013304775347933173,
-      "learning_rate": 0.0001999999113991863,
-      "loss": 46.0,
-      "step": 5552
-    },
-    {
-      "epoch": 0.42456562876311715,
-      "grad_norm": 0.0013518822379410267,
-      "learning_rate": 0.00019999991136720912,
-      "loss": 46.0,
-      "step": 5553
-    },
-    {
-      "epoch": 0.42464208574650686,
-      "grad_norm": 0.0010198524687439203,
-      "learning_rate": 0.00019999991133522616,
-      "loss": 46.0,
-      "step": 5554
-    },
-    {
-      "epoch": 0.42471854272989656,
-      "grad_norm": 0.005139484535902739,
-      "learning_rate": 0.0001999999113032374,
-      "loss": 46.0,
-      "step": 5555
-    },
-    {
-      "epoch": 0.4247949997132863,
-      "grad_norm": 0.0024663228541612625,
-      "learning_rate": 0.00019999991127124292,
-      "loss": 46.0,
-      "step": 5556
-    },
-    {
-      "epoch": 0.42487145669667603,
-      "grad_norm": 0.0020741941407322884,
-      "learning_rate": 0.00019999991123924264,
-      "loss": 46.0,
-      "step": 5557
-    },
-    {
-      "epoch": 0.42494791368006574,
-      "grad_norm": 0.0011090937769040465,
-      "learning_rate": 0.0001999999112072366,
-      "loss": 46.0,
-      "step": 5558
-    },
-    {
-      "epoch": 0.4250243706634555,
-      "grad_norm": 0.002045979490503669,
-      "learning_rate": 0.00019999991117522484,
-      "loss": 46.0,
-      "step": 5559
-    },
-    {
-      "epoch": 0.4251008276468452,
-      "grad_norm": 0.0011074787471443415,
-      "learning_rate": 0.00019999991114320724,
-      "loss": 46.0,
-      "step": 5560
-    },
-    {
-      "epoch": 0.4251772846302349,
-      "grad_norm": 0.0005929818144068122,
-      "learning_rate": 0.0001999999111111839,
-      "loss": 46.0,
-      "step": 5561
-    },
-    {
-      "epoch": 0.4252537416136246,
-      "grad_norm": 0.00236821873113513,
-      "learning_rate": 0.0001999999110791548,
-      "loss": 46.0,
-      "step": 5562
-    },
-    {
-      "epoch": 0.4253301985970144,
-      "grad_norm": 0.0007213192875497043,
-      "learning_rate": 0.00019999991104711987,
-      "loss": 46.0,
-      "step": 5563
-    },
-    {
-      "epoch": 0.4254066555804041,
-      "grad_norm": 0.0028703149873763323,
-      "learning_rate": 0.00019999991101507923,
-      "loss": 46.0,
-      "step": 5564
-    },
-    {
-      "epoch": 0.4254831125637938,
-      "grad_norm": 0.0013398994924500585,
-      "learning_rate": 0.00019999991098303282,
-      "loss": 46.0,
-      "step": 5565
-    },
-    {
-      "epoch": 0.4255595695471835,
-      "grad_norm": 0.002595379250124097,
-      "learning_rate": 0.0001999999109509806,
-      "loss": 46.0,
-      "step": 5566
-    },
-    {
-      "epoch": 0.42563602653057325,
-      "grad_norm": 0.000621318758931011,
-      "learning_rate": 0.00019999991091892264,
-      "loss": 46.0,
-      "step": 5567
-    },
-    {
-      "epoch": 0.42571248351396296,
-      "grad_norm": 0.0016340581933036447,
-      "learning_rate": 0.0001999999108868589,
-      "loss": 46.0,
-      "step": 5568
-    },
-    {
-      "epoch": 0.42578894049735266,
-      "grad_norm": 0.0007561444654129446,
-      "learning_rate": 0.00019999991085478943,
-      "loss": 46.0,
-      "step": 5569
-    },
-    {
-      "epoch": 0.4258653974807424,
-      "grad_norm": 0.0015946123749017715,
-      "learning_rate": 0.00019999991082271415,
-      "loss": 46.0,
-      "step": 5570
-    },
-    {
-      "epoch": 0.42594185446413213,
-      "grad_norm": 0.0008790738647803664,
-      "learning_rate": 0.00019999991079063312,
-      "loss": 46.0,
-      "step": 5571
-    },
-    {
-      "epoch": 0.42601831144752184,
-      "grad_norm": 0.00113858247641474,
-      "learning_rate": 0.00019999991075854632,
-      "loss": 46.0,
-      "step": 5572
-    },
-    {
-      "epoch": 0.42609476843091154,
-      "grad_norm": 0.0009134138817898929,
-      "learning_rate": 0.00019999991072645372,
-      "loss": 46.0,
-      "step": 5573
-    },
-    {
-      "epoch": 0.4261712254143013,
-      "grad_norm": 0.001085581025108695,
-      "learning_rate": 0.00019999991069435537,
-      "loss": 46.0,
-      "step": 5574
-    },
-    {
-      "epoch": 0.426247682397691,
-      "grad_norm": 0.0015665290411561728,
-      "learning_rate": 0.00019999991066225125,
-      "loss": 46.0,
-      "step": 5575
-    },
-    {
-      "epoch": 0.4263241393810807,
-      "grad_norm": 0.0018858853727579117,
-      "learning_rate": 0.00019999991063014138,
-      "loss": 46.0,
-      "step": 5576
-    },
-    {
-      "epoch": 0.4264005963644704,
-      "grad_norm": 0.009295439347624779,
-      "learning_rate": 0.0001999999105980257,
-      "loss": 46.0,
-      "step": 5577
-    },
-    {
-      "epoch": 0.4264770533478602,
-      "grad_norm": 0.0006506615900434554,
-      "learning_rate": 0.00019999991056590427,
-      "loss": 46.0,
-      "step": 5578
-    },
-    {
-      "epoch": 0.4265535103312499,
-      "grad_norm": 0.0015039339195936918,
-      "learning_rate": 0.0001999999105337771,
-      "loss": 46.0,
-      "step": 5579
-    },
-    {
-      "epoch": 0.4266299673146396,
-      "grad_norm": 0.0003653905587270856,
-      "learning_rate": 0.00019999991050164413,
-      "loss": 46.0,
-      "step": 5580
-    },
-    {
-      "epoch": 0.4267064242980293,
-      "grad_norm": 0.0010432812850922346,
-      "learning_rate": 0.0001999999104695054,
-      "loss": 46.0,
-      "step": 5581
-    },
-    {
-      "epoch": 0.42678288128141906,
-      "grad_norm": 0.0013864922802895308,
-      "learning_rate": 0.00019999991043736088,
-      "loss": 46.0,
-      "step": 5582
-    },
-    {
-      "epoch": 0.42685933826480876,
-      "grad_norm": 0.0008443210972473025,
-      "learning_rate": 0.0001999999104052106,
-      "loss": 46.0,
-      "step": 5583
-    },
-    {
-      "epoch": 0.42693579524819847,
-      "grad_norm": 0.0015386386075988412,
-      "learning_rate": 0.00019999991037305458,
-      "loss": 46.0,
-      "step": 5584
-    },
-    {
-      "epoch": 0.42701225223158823,
-      "grad_norm": 0.00041637918911874294,
-      "learning_rate": 0.00019999991034089275,
-      "loss": 46.0,
-      "step": 5585
-    },
-    {
-      "epoch": 0.42708870921497794,
-      "grad_norm": 0.0012116326251998544,
-      "learning_rate": 0.00019999991030872515,
-      "loss": 46.0,
-      "step": 5586
-    },
-    {
-      "epoch": 0.42716516619836764,
-      "grad_norm": 0.001534649869427085,
-      "learning_rate": 0.00019999991027655183,
-      "loss": 46.0,
-      "step": 5587
-    },
-    {
-      "epoch": 0.42724162318175735,
-      "grad_norm": 0.0006149267428554595,
-      "learning_rate": 0.0001999999102443727,
-      "loss": 46.0,
-      "step": 5588
-    },
-    {
-      "epoch": 0.4273180801651471,
-      "grad_norm": 0.0008732268470339477,
-      "learning_rate": 0.00019999991021218781,
-      "loss": 46.0,
-      "step": 5589
-    },
-    {
-      "epoch": 0.4273945371485368,
-      "grad_norm": 0.004182655829936266,
-      "learning_rate": 0.00019999991017999715,
-      "loss": 46.0,
-      "step": 5590
-    },
-    {
-      "epoch": 0.4274709941319265,
-      "grad_norm": 0.002968008164316416,
-      "learning_rate": 0.00019999991014780073,
-      "loss": 46.0,
-      "step": 5591
-    },
-    {
-      "epoch": 0.4275474511153162,
-      "grad_norm": 0.004070365335792303,
-      "learning_rate": 0.00019999991011559852,
-      "loss": 46.0,
-      "step": 5592
-    },
-    {
-      "epoch": 0.427623908098706,
-      "grad_norm": 0.002037449972704053,
-      "learning_rate": 0.00019999991008339056,
-      "loss": 46.0,
-      "step": 5593
-    },
-    {
-      "epoch": 0.4277003650820957,
-      "grad_norm": 0.0018142735352739692,
-      "learning_rate": 0.00019999991005117683,
-      "loss": 46.0,
-      "step": 5594
-    },
-    {
-      "epoch": 0.4277768220654854,
-      "grad_norm": 0.0020180039573460817,
-      "learning_rate": 0.0001999999100189573,
-      "loss": 46.0,
-      "step": 5595
-    },
-    {
-      "epoch": 0.4278532790488751,
-      "grad_norm": 0.01016896590590477,
-      "learning_rate": 0.00019999990998673204,
-      "loss": 46.0,
-      "step": 5596
-    },
-    {
-      "epoch": 0.42792973603226486,
-      "grad_norm": 0.0009558206656947732,
-      "learning_rate": 0.00019999990995450098,
-      "loss": 46.0,
-      "step": 5597
-    },
-    {
-      "epoch": 0.42800619301565457,
-      "grad_norm": 0.001294442918151617,
-      "learning_rate": 0.00019999990992226416,
-      "loss": 46.0,
-      "step": 5598
-    },
-    {
-      "epoch": 0.4280826499990443,
-      "grad_norm": 0.0005126984906382859,
-      "learning_rate": 0.00019999990989002158,
-      "loss": 46.0,
-      "step": 5599
-    },
-    {
-      "epoch": 0.42815910698243403,
-      "grad_norm": 0.012218517251312733,
-      "learning_rate": 0.00019999990985777324,
-      "loss": 46.0,
-      "step": 5600
-    },
-    {
-      "epoch": 0.42823556396582374,
-      "grad_norm": 0.0019848363008350134,
-      "learning_rate": 0.00019999990982551912,
-      "loss": 46.0,
-      "step": 5601
-    },
-    {
-      "epoch": 0.42831202094921345,
-      "grad_norm": 0.00214760797098279,
-      "learning_rate": 0.00019999990979325922,
-      "loss": 46.0,
-      "step": 5602
-    },
-    {
-      "epoch": 0.42838847793260315,
-      "grad_norm": 0.0008706836379133165,
-      "learning_rate": 0.00019999990976099356,
-      "loss": 46.0,
-      "step": 5603
-    },
-    {
-      "epoch": 0.4284649349159929,
-      "grad_norm": 0.0010077876504510641,
-      "learning_rate": 0.00019999990972872214,
-      "loss": 46.0,
-      "step": 5604
-    },
-    {
-      "epoch": 0.4285413918993826,
-      "grad_norm": 0.0022549619898200035,
-      "learning_rate": 0.00019999990969644493,
-      "loss": 46.0,
-      "step": 5605
-    },
-    {
-      "epoch": 0.4286178488827723,
-      "grad_norm": 0.0020016906782984734,
-      "learning_rate": 0.00019999990966416194,
-      "loss": 46.0,
-      "step": 5606
-    },
-    {
-      "epoch": 0.42869430586616203,
-      "grad_norm": 0.0013492833822965622,
-      "learning_rate": 0.0001999999096318732,
-      "loss": 46.0,
-      "step": 5607
-    },
-    {
-      "epoch": 0.4287707628495518,
-      "grad_norm": 0.002511871512979269,
-      "learning_rate": 0.00019999990959957868,
-      "loss": 46.0,
-      "step": 5608
-    },
-    {
-      "epoch": 0.4288472198329415,
-      "grad_norm": 0.0009711926686577499,
-      "learning_rate": 0.00019999990956727842,
-      "loss": 46.0,
-      "step": 5609
-    },
-    {
-      "epoch": 0.4289236768163312,
-      "grad_norm": 0.0008278907625935972,
-      "learning_rate": 0.00019999990953497237,
-      "loss": 46.0,
-      "step": 5610
-    },
-    {
-      "epoch": 0.4290001337997209,
-      "grad_norm": 0.000699732918292284,
-      "learning_rate": 0.00019999990950266057,
-      "loss": 46.0,
-      "step": 5611
-    },
-    {
-      "epoch": 0.42907659078311067,
-      "grad_norm": 0.0010250251507386565,
-      "learning_rate": 0.00019999990947034294,
-      "loss": 46.0,
-      "step": 5612
-    },
-    {
-      "epoch": 0.4291530477665004,
-      "grad_norm": 0.0015237688785418868,
-      "learning_rate": 0.0001999999094380196,
-      "loss": 46.0,
-      "step": 5613
-    },
-    {
-      "epoch": 0.4292295047498901,
-      "grad_norm": 0.0006184679223224521,
-      "learning_rate": 0.00019999990940569048,
-      "loss": 46.0,
-      "step": 5614
-    },
-    {
-      "epoch": 0.42930596173327984,
-      "grad_norm": 0.000717982416972518,
-      "learning_rate": 0.0001999999093733556,
-      "loss": 46.0,
-      "step": 5615
-    },
-    {
-      "epoch": 0.42938241871666954,
-      "grad_norm": 0.001517292927019298,
-      "learning_rate": 0.00019999990934101492,
-      "loss": 46.0,
-      "step": 5616
-    },
-    {
-      "epoch": 0.42945887570005925,
-      "grad_norm": 0.001716771861538291,
-      "learning_rate": 0.0001999999093086685,
-      "loss": 46.0,
-      "step": 5617
-    },
-    {
-      "epoch": 0.42953533268344896,
-      "grad_norm": 0.0008201595046557486,
-      "learning_rate": 0.0001999999092763163,
-      "loss": 46.0,
-      "step": 5618
-    },
-    {
-      "epoch": 0.4296117896668387,
-      "grad_norm": 0.002338016638532281,
-      "learning_rate": 0.0001999999092439583,
-      "loss": 46.0,
-      "step": 5619
-    },
-    {
-      "epoch": 0.4296882466502284,
-      "grad_norm": 0.0006418954581022263,
-      "learning_rate": 0.00019999990921159454,
-      "loss": 46.0,
-      "step": 5620
-    },
-    {
-      "epoch": 0.42976470363361813,
-      "grad_norm": 0.002266326919198036,
-      "learning_rate": 0.00019999990917922506,
-      "loss": 46.0,
-      "step": 5621
-    },
-    {
-      "epoch": 0.42984116061700783,
-      "grad_norm": 0.0009738274966366589,
-      "learning_rate": 0.00019999990914684976,
-      "loss": 46.0,
-      "step": 5622
-    },
-    {
-      "epoch": 0.4299176176003976,
-      "grad_norm": 0.0008395953918807209,
-      "learning_rate": 0.00019999990911446873,
-      "loss": 46.0,
-      "step": 5623
-    },
-    {
-      "epoch": 0.4299940745837873,
-      "grad_norm": 0.0032531479373574257,
-      "learning_rate": 0.00019999990908208188,
-      "loss": 46.0,
-      "step": 5624
-    },
-    {
-      "epoch": 0.430070531567177,
-      "grad_norm": 0.0042190407402813435,
-      "learning_rate": 0.0001999999090496893,
-      "loss": 46.0,
-      "step": 5625
-    },
-    {
-      "epoch": 0.4301469885505667,
-      "grad_norm": 0.0003855096292681992,
-      "learning_rate": 0.00019999990901729096,
-      "loss": 46.0,
-      "step": 5626
-    },
-    {
-      "epoch": 0.4302234455339565,
-      "grad_norm": 0.0032034313771873713,
-      "learning_rate": 0.00019999990898488684,
-      "loss": 46.0,
-      "step": 5627
-    },
-    {
-      "epoch": 0.4302999025173462,
-      "grad_norm": 0.0020397244952619076,
-      "learning_rate": 0.00019999990895247692,
-      "loss": 46.0,
-      "step": 5628
-    },
-    {
-      "epoch": 0.4303763595007359,
-      "grad_norm": 0.0011068512685596943,
-      "learning_rate": 0.00019999990892006125,
-      "loss": 46.0,
-      "step": 5629
-    },
-    {
-      "epoch": 0.43045281648412564,
-      "grad_norm": 0.002530556870624423,
-      "learning_rate": 0.0001999999088876398,
-      "loss": 46.0,
-      "step": 5630
-    },
-    {
-      "epoch": 0.43052927346751535,
-      "grad_norm": 0.0010821041651070118,
-      "learning_rate": 0.0001999999088552126,
-      "loss": 46.0,
-      "step": 5631
-    },
-    {
-      "epoch": 0.43060573045090506,
-      "grad_norm": 0.011005576699972153,
-      "learning_rate": 0.0001999999088227796,
-      "loss": 46.0,
-      "step": 5632
-    },
-    {
-      "epoch": 0.43068218743429476,
-      "grad_norm": 0.001517979078926146,
-      "learning_rate": 0.00019999990879034088,
-      "loss": 46.0,
-      "step": 5633
-    },
-    {
-      "epoch": 0.4307586444176845,
-      "grad_norm": 0.0010132138850167394,
-      "learning_rate": 0.00019999990875789635,
-      "loss": 46.0,
-      "step": 5634
-    },
-    {
-      "epoch": 0.4308351014010742,
-      "grad_norm": 0.0007832224946469069,
-      "learning_rate": 0.00019999990872544607,
-      "loss": 46.0,
-      "step": 5635
-    },
-    {
-      "epoch": 0.43091155838446393,
-      "grad_norm": 0.0015773362247273326,
-      "learning_rate": 0.00019999990869299002,
-      "loss": 46.0,
-      "step": 5636
-    },
-    {
-      "epoch": 0.43098801536785364,
-      "grad_norm": 0.0015283471439033747,
-      "learning_rate": 0.0001999999086605282,
-      "loss": 46.0,
-      "step": 5637
-    },
-    {
-      "epoch": 0.4310644723512434,
-      "grad_norm": 0.0038945076521486044,
-      "learning_rate": 0.00019999990862806062,
-      "loss": 46.0,
-      "step": 5638
-    },
-    {
-      "epoch": 0.4311409293346331,
-      "grad_norm": 0.0041750576347112656,
-      "learning_rate": 0.00019999990859558724,
-      "loss": 46.0,
-      "step": 5639
-    },
-    {
-      "epoch": 0.4312173863180228,
-      "grad_norm": 0.0019340752623975277,
-      "learning_rate": 0.0001999999085631081,
-      "loss": 46.0,
-      "step": 5640
-    },
-    {
-      "epoch": 0.4312938433014125,
-      "grad_norm": 0.002922413172200322,
-      "learning_rate": 0.0001999999085306232,
-      "loss": 46.0,
-      "step": 5641
-    },
-    {
-      "epoch": 0.4313703002848023,
-      "grad_norm": 0.0004881019704043865,
-      "learning_rate": 0.00019999990849813254,
-      "loss": 46.0,
-      "step": 5642
-    },
-    {
-      "epoch": 0.431446757268192,
-      "grad_norm": 0.0009905296610668302,
-      "learning_rate": 0.0001999999084656361,
-      "loss": 46.0,
-      "step": 5643
-    },
-    {
-      "epoch": 0.4315232142515817,
-      "grad_norm": 0.001366614829748869,
-      "learning_rate": 0.0001999999084331339,
-      "loss": 46.0,
-      "step": 5644
-    },
-    {
-      "epoch": 0.43159967123497145,
-      "grad_norm": 0.0013193455524742603,
-      "learning_rate": 0.0001999999084006259,
-      "loss": 46.0,
-      "step": 5645
-    },
-    {
-      "epoch": 0.43167612821836115,
-      "grad_norm": 0.0019613460171967745,
-      "learning_rate": 0.00019999990836811214,
-      "loss": 46.0,
-      "step": 5646
-    },
-    {
-      "epoch": 0.43175258520175086,
-      "grad_norm": 0.000924398482311517,
-      "learning_rate": 0.00019999990833559266,
-      "loss": 46.0,
-      "step": 5647
-    },
-    {
-      "epoch": 0.43182904218514057,
-      "grad_norm": 0.000894192315172404,
-      "learning_rate": 0.00019999990830306733,
-      "loss": 46.0,
-      "step": 5648
-    },
-    {
-      "epoch": 0.4319054991685303,
-      "grad_norm": 0.0017163888551294804,
-      "learning_rate": 0.0001999999082705363,
-      "loss": 46.0,
-      "step": 5649
-    },
-    {
-      "epoch": 0.43198195615192003,
-      "grad_norm": 0.0011232475517317653,
-      "learning_rate": 0.00019999990823799948,
-      "loss": 46.0,
-      "step": 5650
-    },
-    {
-      "epoch": 0.43205841313530974,
-      "grad_norm": 0.0033304686658084393,
-      "learning_rate": 0.00019999990820545688,
-      "loss": 46.0,
-      "step": 5651
-    },
-    {
-      "epoch": 0.43213487011869944,
-      "grad_norm": 0.00852080900222063,
-      "learning_rate": 0.0001999999081729085,
-      "loss": 46.0,
-      "step": 5652
-    },
-    {
-      "epoch": 0.4322113271020892,
-      "grad_norm": 0.002735903486609459,
-      "learning_rate": 0.0001999999081403544,
-      "loss": 46.0,
-      "step": 5653
-    },
-    {
-      "epoch": 0.4322877840854789,
-      "grad_norm": 0.0007669954793527722,
-      "learning_rate": 0.00019999990810779447,
-      "loss": 46.0,
-      "step": 5654
-    },
-    {
-      "epoch": 0.4323642410688686,
-      "grad_norm": 0.00026303669437766075,
-      "learning_rate": 0.0001999999080752288,
-      "loss": 46.0,
-      "step": 5655
-    },
-    {
-      "epoch": 0.4324406980522583,
-      "grad_norm": 0.0008959940169006586,
-      "learning_rate": 0.00019999990804265734,
-      "loss": 46.0,
-      "step": 5656
-    },
-    {
-      "epoch": 0.4325171550356481,
-      "grad_norm": 0.0012439782731235027,
-      "learning_rate": 0.00019999990801008015,
-      "loss": 46.0,
-      "step": 5657
-    },
-    {
-      "epoch": 0.4325936120190378,
-      "grad_norm": 0.0018033916130661964,
-      "learning_rate": 0.00019999990797749714,
-      "loss": 46.0,
-      "step": 5658
-    },
-    {
-      "epoch": 0.4326700690024275,
-      "grad_norm": 0.0007578761433251202,
-      "learning_rate": 0.0001999999079449084,
-      "loss": 46.0,
-      "step": 5659
-    },
-    {
-      "epoch": 0.43274652598581725,
-      "grad_norm": 0.002927834866568446,
-      "learning_rate": 0.00019999990791231387,
-      "loss": 46.0,
-      "step": 5660
-    },
-    {
-      "epoch": 0.43282298296920696,
-      "grad_norm": 0.0008654286502860487,
-      "learning_rate": 0.0001999999078797136,
-      "loss": 46.0,
-      "step": 5661
-    },
-    {
-      "epoch": 0.43289943995259667,
-      "grad_norm": 0.003591811517253518,
-      "learning_rate": 0.00019999990784710754,
-      "loss": 46.0,
-      "step": 5662
-    },
-    {
-      "epoch": 0.43297589693598637,
-      "grad_norm": 0.00880864355713129,
-      "learning_rate": 0.00019999990781449572,
-      "loss": 46.0,
-      "step": 5663
-    },
-    {
-      "epoch": 0.43305235391937613,
-      "grad_norm": 0.0011075816582888365,
-      "learning_rate": 0.00019999990778187812,
-      "loss": 46.0,
-      "step": 5664
-    },
-    {
-      "epoch": 0.43312881090276584,
-      "grad_norm": 0.0003506146022118628,
-      "learning_rate": 0.00019999990774925475,
-      "loss": 46.0,
-      "step": 5665
-    },
-    {
-      "epoch": 0.43320526788615554,
-      "grad_norm": 0.0024543479084968567,
-      "learning_rate": 0.0001999999077166256,
-      "loss": 46.0,
-      "step": 5666
-    },
-    {
-      "epoch": 0.43328172486954525,
-      "grad_norm": 0.0023912524338811636,
-      "learning_rate": 0.00019999990768399068,
-      "loss": 46.0,
-      "step": 5667
-    },
-    {
-      "epoch": 0.433358181852935,
-      "grad_norm": 0.003390712197870016,
-      "learning_rate": 0.00019999990765135,
-      "loss": 46.0,
-      "step": 5668
-    },
-    {
-      "epoch": 0.4334346388363247,
-      "grad_norm": 0.0027496919501572847,
-      "learning_rate": 0.00019999990761870358,
-      "loss": 46.0,
-      "step": 5669
-    },
-    {
-      "epoch": 0.4335110958197144,
-      "grad_norm": 0.0011278516612946987,
-      "learning_rate": 0.00019999990758605134,
-      "loss": 46.0,
-      "step": 5670
-    },
-    {
-      "epoch": 0.4335875528031041,
-      "grad_norm": 0.0019641281105577946,
-      "learning_rate": 0.00019999990755339338,
-      "loss": 46.0,
-      "step": 5671
-    },
-    {
-      "epoch": 0.4336640097864939,
-      "grad_norm": 0.015833742916584015,
-      "learning_rate": 0.00019999990752072962,
-      "loss": 46.0,
-      "step": 5672
-    },
-    {
-      "epoch": 0.4337404667698836,
-      "grad_norm": 0.0010731312213465571,
-      "learning_rate": 0.0001999999074880601,
-      "loss": 46.0,
-      "step": 5673
-    },
-    {
-      "epoch": 0.4338169237532733,
-      "grad_norm": 0.0010447646491229534,
-      "learning_rate": 0.00019999990745538478,
-      "loss": 46.0,
-      "step": 5674
-    },
-    {
-      "epoch": 0.43389338073666306,
-      "grad_norm": 0.005253052804619074,
-      "learning_rate": 0.00019999990742270373,
-      "loss": 46.0,
-      "step": 5675
-    },
-    {
-      "epoch": 0.43396983772005276,
-      "grad_norm": 0.003013312816619873,
-      "learning_rate": 0.0001999999073900169,
-      "loss": 46.0,
-      "step": 5676
-    },
-    {
-      "epoch": 0.43404629470344247,
-      "grad_norm": 0.003034434514120221,
-      "learning_rate": 0.0001999999073573243,
-      "loss": 46.0,
-      "step": 5677
-    },
-    {
-      "epoch": 0.4341227516868322,
-      "grad_norm": 0.0014418066712096334,
-      "learning_rate": 0.00019999990732462593,
-      "loss": 46.0,
-      "step": 5678
-    },
-    {
-      "epoch": 0.43419920867022194,
-      "grad_norm": 0.003474494442343712,
-      "learning_rate": 0.00019999990729192176,
-      "loss": 46.0,
-      "step": 5679
-    },
-    {
-      "epoch": 0.43427566565361164,
-      "grad_norm": 0.0027134548872709274,
-      "learning_rate": 0.00019999990725921187,
-      "loss": 46.0,
-      "step": 5680
-    },
-    {
-      "epoch": 0.43435212263700135,
-      "grad_norm": 0.0006129696266725659,
-      "learning_rate": 0.0001999999072264962,
-      "loss": 46.0,
-      "step": 5681
-    },
-    {
-      "epoch": 0.43442857962039105,
-      "grad_norm": 0.0012946901842951775,
-      "learning_rate": 0.00019999990719377474,
-      "loss": 46.0,
-      "step": 5682
-    },
-    {
-      "epoch": 0.4345050366037808,
-      "grad_norm": 0.0019250615732744336,
-      "learning_rate": 0.00019999990716104753,
-      "loss": 46.0,
-      "step": 5683
-    },
-    {
-      "epoch": 0.4345814935871705,
-      "grad_norm": 0.004271842073649168,
-      "learning_rate": 0.00019999990712831454,
-      "loss": 46.0,
-      "step": 5684
-    },
-    {
-      "epoch": 0.4346579505705602,
-      "grad_norm": 0.000979638542048633,
-      "learning_rate": 0.00019999990709557576,
-      "loss": 46.0,
-      "step": 5685
-    },
-    {
-      "epoch": 0.43473440755395,
-      "grad_norm": 0.0004195354413241148,
-      "learning_rate": 0.00019999990706283125,
-      "loss": 46.0,
-      "step": 5686
-    },
-    {
-      "epoch": 0.4348108645373397,
-      "grad_norm": 0.0006629640120081604,
-      "learning_rate": 0.00019999990703008095,
-      "loss": 46.0,
-      "step": 5687
-    },
-    {
-      "epoch": 0.4348873215207294,
-      "grad_norm": 0.002123679965734482,
-      "learning_rate": 0.0001999999069973249,
-      "loss": 46.0,
-      "step": 5688
-    },
-    {
-      "epoch": 0.4349637785041191,
-      "grad_norm": 0.0007858469034545124,
-      "learning_rate": 0.00019999990696456304,
-      "loss": 46.0,
-      "step": 5689
-    },
-    {
-      "epoch": 0.43504023548750886,
-      "grad_norm": 0.0010483539663255215,
-      "learning_rate": 0.00019999990693179545,
-      "loss": 46.0,
-      "step": 5690
-    },
-    {
-      "epoch": 0.43511669247089857,
-      "grad_norm": 0.0009564394131302834,
-      "learning_rate": 0.00019999990689902207,
-      "loss": 46.0,
-      "step": 5691
-    },
-    {
-      "epoch": 0.4351931494542883,
-      "grad_norm": 0.0003780309925787151,
-      "learning_rate": 0.00019999990686624293,
-      "loss": 46.0,
-      "step": 5692
-    },
-    {
-      "epoch": 0.435269606437678,
-      "grad_norm": 0.004239376168698072,
-      "learning_rate": 0.000199999906833458,
-      "loss": 46.0,
-      "step": 5693
-    },
-    {
-      "epoch": 0.43534606342106774,
-      "grad_norm": 0.0011086431331932545,
-      "learning_rate": 0.00019999990680066735,
-      "loss": 46.0,
-      "step": 5694
-    },
-    {
-      "epoch": 0.43542252040445745,
-      "grad_norm": 0.0013577269855886698,
-      "learning_rate": 0.00019999990676787088,
-      "loss": 46.0,
-      "step": 5695
-    },
-    {
-      "epoch": 0.43549897738784715,
-      "grad_norm": 0.000948662287555635,
-      "learning_rate": 0.00019999990673506867,
-      "loss": 46.0,
-      "step": 5696
-    },
-    {
-      "epoch": 0.43557543437123686,
-      "grad_norm": 0.0005964995361864567,
-      "learning_rate": 0.00019999990670226066,
-      "loss": 46.0,
-      "step": 5697
-    },
-    {
-      "epoch": 0.4356518913546266,
-      "grad_norm": 0.0008766123792156577,
-      "learning_rate": 0.0001999999066694469,
-      "loss": 46.0,
-      "step": 5698
-    },
-    {
-      "epoch": 0.4357283483380163,
-      "grad_norm": 0.0029147870372980833,
-      "learning_rate": 0.00019999990663662737,
-      "loss": 46.0,
-      "step": 5699
-    },
-    {
-      "epoch": 0.43580480532140603,
-      "grad_norm": 0.0014968147734180093,
-      "learning_rate": 0.0001999999066038021,
-      "loss": 46.0,
-      "step": 5700
-    },
-    {
-      "epoch": 0.4358812623047958,
-      "grad_norm": 0.002592020435258746,
-      "learning_rate": 0.000199999906570971,
-      "loss": 46.0,
-      "step": 5701
-    },
-    {
-      "epoch": 0.4359577192881855,
-      "grad_norm": 0.0006876286934129894,
-      "learning_rate": 0.00019999990653813416,
-      "loss": 46.0,
-      "step": 5702
-    },
-    {
-      "epoch": 0.4360341762715752,
-      "grad_norm": 0.0011866152053698897,
-      "learning_rate": 0.00019999990650529156,
-      "loss": 46.0,
-      "step": 5703
-    },
-    {
-      "epoch": 0.4361106332549649,
-      "grad_norm": 0.0007853118004277349,
-      "learning_rate": 0.0001999999064724432,
-      "loss": 46.0,
-      "step": 5704
-    },
-    {
-      "epoch": 0.43618709023835467,
-      "grad_norm": 0.0013021667255088687,
-      "learning_rate": 0.00019999990643958904,
-      "loss": 46.0,
-      "step": 5705
-    },
-    {
-      "epoch": 0.4362635472217444,
-      "grad_norm": 0.0014558027032762766,
-      "learning_rate": 0.00019999990640672915,
-      "loss": 46.0,
-      "step": 5706
-    },
-    {
-      "epoch": 0.4363400042051341,
-      "grad_norm": 0.0006471550441347063,
-      "learning_rate": 0.00019999990637386344,
-      "loss": 46.0,
-      "step": 5707
-    },
-    {
-      "epoch": 0.4364164611885238,
-      "grad_norm": 0.0007654508226551116,
-      "learning_rate": 0.000199999906340992,
-      "loss": 46.0,
-      "step": 5708
-    },
-    {
-      "epoch": 0.43649291817191355,
-      "grad_norm": 0.0010116776684299111,
-      "learning_rate": 0.00019999990630811476,
-      "loss": 46.0,
-      "step": 5709
-    },
-    {
-      "epoch": 0.43656937515530325,
-      "grad_norm": 0.0016404909547418356,
-      "learning_rate": 0.00019999990627523175,
-      "loss": 46.0,
-      "step": 5710
-    },
-    {
-      "epoch": 0.43664583213869296,
-      "grad_norm": 0.0008823841926641762,
-      "learning_rate": 0.000199999906242343,
-      "loss": 46.0,
-      "step": 5711
-    },
-    {
-      "epoch": 0.43672228912208266,
-      "grad_norm": 0.0008995257667265832,
-      "learning_rate": 0.00019999990620944846,
-      "loss": 46.0,
-      "step": 5712
-    },
-    {
-      "epoch": 0.4367987461054724,
-      "grad_norm": 0.001308804377913475,
-      "learning_rate": 0.00019999990617654815,
-      "loss": 46.0,
-      "step": 5713
-    },
-    {
-      "epoch": 0.43687520308886213,
-      "grad_norm": 0.0011804414680227637,
-      "learning_rate": 0.0001999999061436421,
-      "loss": 46.0,
-      "step": 5714
-    },
-    {
-      "epoch": 0.43695166007225184,
-      "grad_norm": 0.002129241591319442,
-      "learning_rate": 0.00019999990611073028,
-      "loss": 46.0,
-      "step": 5715
-    },
-    {
-      "epoch": 0.4370281170556416,
-      "grad_norm": 0.0037376261316239834,
-      "learning_rate": 0.00019999990607781265,
-      "loss": 46.0,
-      "step": 5716
-    },
-    {
-      "epoch": 0.4371045740390313,
-      "grad_norm": 0.0027529390063136816,
-      "learning_rate": 0.00019999990604488926,
-      "loss": 46.0,
-      "step": 5717
-    },
-    {
-      "epoch": 0.437181031022421,
-      "grad_norm": 0.0007344126934185624,
-      "learning_rate": 0.00019999990601196014,
-      "loss": 46.0,
-      "step": 5718
-    },
-    {
-      "epoch": 0.4372574880058107,
-      "grad_norm": 0.001152716693468392,
-      "learning_rate": 0.0001999999059790252,
-      "loss": 46.0,
-      "step": 5719
-    },
-    {
-      "epoch": 0.4373339449892005,
-      "grad_norm": 0.0012725250562652946,
-      "learning_rate": 0.0001999999059460845,
-      "loss": 46.0,
-      "step": 5720
-    },
-    {
-      "epoch": 0.4374104019725902,
-      "grad_norm": 0.001870090956799686,
-      "learning_rate": 0.00019999990591313806,
-      "loss": 46.0,
-      "step": 5721
-    },
-    {
-      "epoch": 0.4374868589559799,
-      "grad_norm": 0.005571620538830757,
-      "learning_rate": 0.00019999990588018583,
-      "loss": 46.0,
-      "step": 5722
-    },
-    {
-      "epoch": 0.4375633159393696,
-      "grad_norm": 0.0014280268223956227,
-      "learning_rate": 0.00019999990584722784,
-      "loss": 46.0,
-      "step": 5723
-    },
-    {
-      "epoch": 0.43763977292275935,
-      "grad_norm": 0.0005739098414778709,
-      "learning_rate": 0.00019999990581426406,
-      "loss": 46.0,
-      "step": 5724
-    },
-    {
-      "epoch": 0.43771622990614906,
-      "grad_norm": 0.0011787748662754893,
-      "learning_rate": 0.00019999990578129453,
-      "loss": 46.0,
-      "step": 5725
-    },
-    {
-      "epoch": 0.43779268688953876,
-      "grad_norm": 0.001546344137750566,
-      "learning_rate": 0.00019999990574831925,
-      "loss": 46.0,
-      "step": 5726
-    },
-    {
-      "epoch": 0.43786914387292847,
-      "grad_norm": 0.0007155445637181401,
-      "learning_rate": 0.00019999990571533814,
-      "loss": 46.0,
-      "step": 5727
-    },
-    {
-      "epoch": 0.43794560085631823,
-      "grad_norm": 0.0018179194303229451,
-      "learning_rate": 0.0001999999056823513,
-      "loss": 46.0,
-      "step": 5728
-    },
-    {
-      "epoch": 0.43802205783970793,
-      "grad_norm": 0.0032046008855104446,
-      "learning_rate": 0.0001999999056493587,
-      "loss": 46.0,
-      "step": 5729
-    },
-    {
-      "epoch": 0.43809851482309764,
-      "grad_norm": 0.0014087113086134195,
-      "learning_rate": 0.0001999999056163603,
-      "loss": 46.0,
-      "step": 5730
-    },
-    {
-      "epoch": 0.4381749718064874,
-      "grad_norm": 0.001231934642419219,
-      "learning_rate": 0.00019999990558335618,
-      "loss": 46.0,
-      "step": 5731
-    },
-    {
-      "epoch": 0.4382514287898771,
-      "grad_norm": 0.0010691469069570303,
-      "learning_rate": 0.0001999999055503462,
-      "loss": 46.0,
-      "step": 5732
-    },
-    {
-      "epoch": 0.4383278857732668,
-      "grad_norm": 0.0017734867287799716,
-      "learning_rate": 0.00019999990551733055,
-      "loss": 46.0,
-      "step": 5733
-    },
-    {
-      "epoch": 0.4384043427566565,
-      "grad_norm": 0.0005985402385704219,
-      "learning_rate": 0.00019999990548430908,
-      "loss": 46.0,
-      "step": 5734
-    },
-    {
-      "epoch": 0.4384807997400463,
-      "grad_norm": 0.0026119702961295843,
-      "learning_rate": 0.00019999990545128187,
-      "loss": 46.0,
-      "step": 5735
-    },
-    {
-      "epoch": 0.438557256723436,
-      "grad_norm": 0.001037146896123886,
-      "learning_rate": 0.00019999990541824886,
-      "loss": 46.0,
-      "step": 5736
-    },
-    {
-      "epoch": 0.4386337137068257,
-      "grad_norm": 0.0008529619663022459,
-      "learning_rate": 0.0001999999053852101,
-      "loss": 46.0,
-      "step": 5737
-    },
-    {
-      "epoch": 0.4387101706902154,
-      "grad_norm": 0.0009773393394425511,
-      "learning_rate": 0.00019999990535216554,
-      "loss": 46.0,
-      "step": 5738
-    },
-    {
-      "epoch": 0.43878662767360516,
-      "grad_norm": 0.0008442993857897818,
-      "learning_rate": 0.00019999990531911524,
-      "loss": 46.0,
-      "step": 5739
-    },
-    {
-      "epoch": 0.43886308465699486,
-      "grad_norm": 0.0020586398895829916,
-      "learning_rate": 0.00019999990528605916,
-      "loss": 46.0,
-      "step": 5740
-    },
-    {
-      "epoch": 0.43893954164038457,
-      "grad_norm": 0.0028369424398988485,
-      "learning_rate": 0.0001999999052529973,
-      "loss": 46.0,
-      "step": 5741
-    },
-    {
-      "epoch": 0.4390159986237743,
-      "grad_norm": 0.0022347168996930122,
-      "learning_rate": 0.0001999999052199297,
-      "loss": 46.0,
-      "step": 5742
-    },
-    {
-      "epoch": 0.43909245560716403,
-      "grad_norm": 0.0008101914427243173,
-      "learning_rate": 0.00019999990518685632,
-      "loss": 46.0,
-      "step": 5743
-    },
-    {
-      "epoch": 0.43916891259055374,
-      "grad_norm": 0.0004884616355411708,
-      "learning_rate": 0.00019999990515377715,
-      "loss": 46.0,
-      "step": 5744
-    },
-    {
-      "epoch": 0.43924536957394344,
-      "grad_norm": 0.0020497492514550686,
-      "learning_rate": 0.00019999990512069223,
-      "loss": 46.0,
-      "step": 5745
-    },
-    {
-      "epoch": 0.4393218265573332,
-      "grad_norm": 0.0013198519591242075,
-      "learning_rate": 0.00019999990508760154,
-      "loss": 46.0,
-      "step": 5746
-    },
-    {
-      "epoch": 0.4393982835407229,
-      "grad_norm": 0.0008747435640543699,
-      "learning_rate": 0.00019999990505450508,
-      "loss": 46.0,
-      "step": 5747
-    },
-    {
-      "epoch": 0.4394747405241126,
-      "grad_norm": 0.0019397714640945196,
-      "learning_rate": 0.00019999990502140284,
-      "loss": 46.0,
-      "step": 5748
-    },
-    {
-      "epoch": 0.4395511975075023,
-      "grad_norm": 0.0004283189191482961,
-      "learning_rate": 0.00019999990498829483,
-      "loss": 46.0,
-      "step": 5749
-    },
-    {
-      "epoch": 0.4396276544908921,
-      "grad_norm": 0.0016128912102431059,
-      "learning_rate": 0.00019999990495518107,
-      "loss": 46.0,
-      "step": 5750
-    },
-    {
-      "epoch": 0.4397041114742818,
-      "grad_norm": 0.0005147580523043871,
-      "learning_rate": 0.00019999990492206154,
-      "loss": 46.0,
-      "step": 5751
-    },
-    {
-      "epoch": 0.4397805684576715,
-      "grad_norm": 0.0009103418560698628,
-      "learning_rate": 0.00019999990488893624,
-      "loss": 46.0,
-      "step": 5752
-    },
-    {
-      "epoch": 0.4398570254410612,
-      "grad_norm": 0.0009675624896772206,
-      "learning_rate": 0.00019999990485580513,
-      "loss": 46.0,
-      "step": 5753
-    },
-    {
-      "epoch": 0.43993348242445096,
-      "grad_norm": 0.0031458481680601835,
-      "learning_rate": 0.0001999999048226683,
-      "loss": 46.0,
-      "step": 5754
-    },
-    {
-      "epoch": 0.44000993940784067,
-      "grad_norm": 0.000976794632151723,
-      "learning_rate": 0.00019999990478952566,
-      "loss": 46.0,
-      "step": 5755
-    },
-    {
-      "epoch": 0.44008639639123037,
-      "grad_norm": 0.0007982795359566808,
-      "learning_rate": 0.0001999999047563773,
-      "loss": 46.0,
-      "step": 5756
-    },
-    {
-      "epoch": 0.4401628533746201,
-      "grad_norm": 0.005540730431675911,
-      "learning_rate": 0.00019999990472322312,
-      "loss": 46.0,
-      "step": 5757
-    },
-    {
-      "epoch": 0.44023931035800984,
-      "grad_norm": 0.0014101737178862095,
-      "learning_rate": 0.0001999999046900632,
-      "loss": 46.0,
-      "step": 5758
-    },
-    {
-      "epoch": 0.44031576734139954,
-      "grad_norm": 0.001606898382306099,
-      "learning_rate": 0.00019999990465689748,
-      "loss": 46.0,
-      "step": 5759
-    },
-    {
-      "epoch": 0.44039222432478925,
-      "grad_norm": 0.0017210078658536077,
-      "learning_rate": 0.00019999990462372602,
-      "loss": 46.0,
-      "step": 5760
-    },
-    {
-      "epoch": 0.440468681308179,
-      "grad_norm": 0.0007553829345852137,
-      "learning_rate": 0.0001999999045905488,
-      "loss": 46.0,
-      "step": 5761
-    },
-    {
-      "epoch": 0.4405451382915687,
-      "grad_norm": 0.0029103581327944994,
-      "learning_rate": 0.0001999999045573658,
-      "loss": 46.0,
-      "step": 5762
-    },
-    {
-      "epoch": 0.4406215952749584,
-      "grad_norm": 0.0010250663617625833,
-      "learning_rate": 0.00019999990452417702,
-      "loss": 46.0,
-      "step": 5763
-    },
-    {
-      "epoch": 0.4406980522583481,
-      "grad_norm": 0.001137417508289218,
-      "learning_rate": 0.00019999990449098246,
-      "loss": 46.0,
-      "step": 5764
-    },
-    {
-      "epoch": 0.4407745092417379,
-      "grad_norm": 0.0021452028304338455,
-      "learning_rate": 0.00019999990445778215,
-      "loss": 46.0,
-      "step": 5765
-    },
-    {
-      "epoch": 0.4408509662251276,
-      "grad_norm": 0.0005929744220338762,
-      "learning_rate": 0.00019999990442457608,
-      "loss": 46.0,
-      "step": 5766
-    },
-    {
-      "epoch": 0.4409274232085173,
-      "grad_norm": 0.0013729935744777322,
-      "learning_rate": 0.00019999990439136423,
-      "loss": 46.0,
-      "step": 5767
-    },
-    {
-      "epoch": 0.441003880191907,
-      "grad_norm": 0.005637479946017265,
-      "learning_rate": 0.0001999999043581466,
-      "loss": 46.0,
-      "step": 5768
-    },
-    {
-      "epoch": 0.44108033717529677,
-      "grad_norm": 0.002816990949213505,
-      "learning_rate": 0.0001999999043249232,
-      "loss": 46.0,
-      "step": 5769
-    },
-    {
-      "epoch": 0.44115679415868647,
-      "grad_norm": 0.0018337401561439037,
-      "learning_rate": 0.00019999990429169406,
-      "loss": 46.0,
-      "step": 5770
-    },
-    {
-      "epoch": 0.4412332511420762,
-      "grad_norm": 0.001763531006872654,
-      "learning_rate": 0.00019999990425845915,
-      "loss": 46.0,
-      "step": 5771
-    },
-    {
-      "epoch": 0.4413097081254659,
-      "grad_norm": 0.0008495684596709907,
-      "learning_rate": 0.00019999990422521843,
-      "loss": 46.0,
-      "step": 5772
-    },
-    {
-      "epoch": 0.44138616510885564,
-      "grad_norm": 0.0013716942630708218,
-      "learning_rate": 0.00019999990419197197,
-      "loss": 46.0,
-      "step": 5773
-    },
-    {
-      "epoch": 0.44146262209224535,
-      "grad_norm": 0.00035818334436044097,
-      "learning_rate": 0.00019999990415871973,
-      "loss": 46.0,
-      "step": 5774
-    },
-    {
-      "epoch": 0.44153907907563505,
-      "grad_norm": 0.00099306704942137,
-      "learning_rate": 0.0001999999041254617,
-      "loss": 46.0,
-      "step": 5775
-    },
-    {
-      "epoch": 0.4416155360590248,
-      "grad_norm": 0.0016687854658812284,
-      "learning_rate": 0.00019999990409219794,
-      "loss": 46.0,
-      "step": 5776
-    },
-    {
-      "epoch": 0.4416919930424145,
-      "grad_norm": 0.0005627785576507449,
-      "learning_rate": 0.00019999990405892838,
-      "loss": 46.0,
-      "step": 5777
-    },
-    {
-      "epoch": 0.4417684500258042,
-      "grad_norm": 0.002996542025357485,
-      "learning_rate": 0.00019999990402565305,
-      "loss": 46.0,
-      "step": 5778
-    },
-    {
-      "epoch": 0.44184490700919393,
-      "grad_norm": 0.0002453405177220702,
-      "learning_rate": 0.00019999990399237197,
-      "loss": 46.0,
-      "step": 5779
-    },
-    {
-      "epoch": 0.4419213639925837,
-      "grad_norm": 0.0013348313514143229,
-      "learning_rate": 0.00019999990395908512,
-      "loss": 46.0,
-      "step": 5780
-    },
-    {
-      "epoch": 0.4419978209759734,
-      "grad_norm": 0.001339646871201694,
-      "learning_rate": 0.00019999990392579247,
-      "loss": 46.0,
-      "step": 5781
-    },
-    {
-      "epoch": 0.4420742779593631,
-      "grad_norm": 0.004314985126256943,
-      "learning_rate": 0.0001999999038924941,
-      "loss": 46.0,
-      "step": 5782
-    },
-    {
-      "epoch": 0.4421507349427528,
-      "grad_norm": 0.001006556092761457,
-      "learning_rate": 0.00019999990385918993,
-      "loss": 46.0,
-      "step": 5783
-    },
-    {
-      "epoch": 0.44222719192614257,
-      "grad_norm": 0.0005633236723951995,
-      "learning_rate": 0.00019999990382588002,
-      "loss": 46.0,
-      "step": 5784
-    },
-    {
-      "epoch": 0.4423036489095323,
-      "grad_norm": 0.0008139516576193273,
-      "learning_rate": 0.0001999999037925643,
-      "loss": 46.0,
-      "step": 5785
-    },
-    {
-      "epoch": 0.442380105892922,
-      "grad_norm": 0.0018054844113066792,
-      "learning_rate": 0.0001999999037592428,
-      "loss": 46.0,
-      "step": 5786
-    },
-    {
-      "epoch": 0.4424565628763117,
-      "grad_norm": 0.0005708512617275119,
-      "learning_rate": 0.0001999999037259156,
-      "loss": 46.0,
-      "step": 5787
-    },
-    {
-      "epoch": 0.44253301985970145,
-      "grad_norm": 0.0017614421667531133,
-      "learning_rate": 0.0001999999036925826,
-      "loss": 46.0,
-      "step": 5788
-    },
-    {
-      "epoch": 0.44260947684309115,
-      "grad_norm": 0.0024563679471611977,
-      "learning_rate": 0.00019999990365924378,
-      "loss": 46.0,
-      "step": 5789
-    },
-    {
-      "epoch": 0.44268593382648086,
-      "grad_norm": 0.0009294821065850556,
-      "learning_rate": 0.00019999990362589925,
-      "loss": 46.0,
-      "step": 5790
-    },
-    {
-      "epoch": 0.4427623908098706,
-      "grad_norm": 0.0014008278958499432,
-      "learning_rate": 0.00019999990359254892,
-      "loss": 46.0,
-      "step": 5791
-    },
-    {
-      "epoch": 0.4428388477932603,
-      "grad_norm": 0.0008367998525500298,
-      "learning_rate": 0.00019999990355919285,
-      "loss": 46.0,
-      "step": 5792
-    },
-    {
-      "epoch": 0.44291530477665003,
-      "grad_norm": 0.00150088791269809,
-      "learning_rate": 0.000199999903525831,
-      "loss": 46.0,
-      "step": 5793
-    },
-    {
-      "epoch": 0.44299176176003974,
-      "grad_norm": 0.0006685269763693213,
-      "learning_rate": 0.00019999990349246335,
-      "loss": 46.0,
-      "step": 5794
-    },
-    {
-      "epoch": 0.4430682187434295,
-      "grad_norm": 0.0012970648240298033,
-      "learning_rate": 0.00019999990345908998,
-      "loss": 46.0,
-      "step": 5795
-    },
-    {
-      "epoch": 0.4431446757268192,
-      "grad_norm": 0.010080215521156788,
-      "learning_rate": 0.0001999999034257108,
-      "loss": 46.0,
-      "step": 5796
-    },
-    {
-      "epoch": 0.4432211327102089,
-      "grad_norm": 0.004340726882219315,
-      "learning_rate": 0.00019999990339232587,
-      "loss": 46.0,
-      "step": 5797
-    },
-    {
-      "epoch": 0.4432975896935986,
-      "grad_norm": 0.0009696157067082822,
-      "learning_rate": 0.00019999990335893518,
-      "loss": 46.0,
-      "step": 5798
-    },
-    {
-      "epoch": 0.4433740466769884,
-      "grad_norm": 0.002490574959665537,
-      "learning_rate": 0.00019999990332553869,
-      "loss": 46.0,
-      "step": 5799
-    },
-    {
-      "epoch": 0.4434505036603781,
-      "grad_norm": 0.0014003097312524915,
-      "learning_rate": 0.00019999990329213642,
-      "loss": 46.0,
-      "step": 5800
-    },
-    {
-      "epoch": 0.4435269606437678,
-      "grad_norm": 0.001936648041009903,
-      "learning_rate": 0.00019999990325872844,
-      "loss": 46.0,
-      "step": 5801
-    },
-    {
-      "epoch": 0.44360341762715755,
-      "grad_norm": 0.0010092939483001828,
-      "learning_rate": 0.00019999990322531463,
-      "loss": 46.0,
-      "step": 5802
-    },
-    {
-      "epoch": 0.44367987461054725,
-      "grad_norm": 0.0018835609080269933,
-      "learning_rate": 0.0001999999031918951,
-      "loss": 46.0,
-      "step": 5803
-    },
-    {
-      "epoch": 0.44375633159393696,
-      "grad_norm": 0.0011895131319761276,
-      "learning_rate": 0.00019999990315846977,
-      "loss": 46.0,
-      "step": 5804
-    },
-    {
-      "epoch": 0.44383278857732666,
-      "grad_norm": 0.0028711638879030943,
-      "learning_rate": 0.00019999990312503867,
-      "loss": 46.0,
-      "step": 5805
-    },
-    {
-      "epoch": 0.4439092455607164,
-      "grad_norm": 0.001128563773818314,
-      "learning_rate": 0.00019999990309160182,
-      "loss": 46.0,
-      "step": 5806
-    },
-    {
-      "epoch": 0.44398570254410613,
-      "grad_norm": 0.0009731303434818983,
-      "learning_rate": 0.0001999999030581592,
-      "loss": 46.0,
-      "step": 5807
-    },
-    {
-      "epoch": 0.44406215952749584,
-      "grad_norm": 0.003529465990141034,
-      "learning_rate": 0.0001999999030247108,
-      "loss": 46.0,
-      "step": 5808
-    },
-    {
-      "epoch": 0.44413861651088554,
-      "grad_norm": 0.0011232653632760048,
-      "learning_rate": 0.0001999999029912566,
-      "loss": 46.0,
-      "step": 5809
-    },
-    {
-      "epoch": 0.4442150734942753,
-      "grad_norm": 0.0017714614514261484,
-      "learning_rate": 0.00019999990295779666,
-      "loss": 46.0,
-      "step": 5810
-    },
-    {
-      "epoch": 0.444291530477665,
-      "grad_norm": 0.001913544605486095,
-      "learning_rate": 0.00019999990292433097,
-      "loss": 46.0,
-      "step": 5811
-    },
-    {
-      "epoch": 0.4443679874610547,
-      "grad_norm": 0.0010140163358300924,
-      "learning_rate": 0.0001999999028908595,
-      "loss": 46.0,
-      "step": 5812
-    },
-    {
-      "epoch": 0.4444444444444444,
-      "grad_norm": 0.0013631342444568872,
-      "learning_rate": 0.00019999990285738225,
-      "loss": 46.0,
-      "step": 5813
-    },
-    {
-      "epoch": 0.4445209014278342,
-      "grad_norm": 0.0003541119513101876,
-      "learning_rate": 0.00019999990282389924,
-      "loss": 46.0,
-      "step": 5814
-    },
-    {
-      "epoch": 0.4445973584112239,
-      "grad_norm": 0.003940911497920752,
-      "learning_rate": 0.00019999990279041043,
-      "loss": 46.0,
-      "step": 5815
-    },
-    {
-      "epoch": 0.4446738153946136,
-      "grad_norm": 0.005468114744871855,
-      "learning_rate": 0.0001999999027569159,
-      "loss": 46.0,
-      "step": 5816
-    },
-    {
-      "epoch": 0.44475027237800335,
-      "grad_norm": 0.000980605836957693,
-      "learning_rate": 0.00019999990272341557,
-      "loss": 46.0,
-      "step": 5817
-    },
-    {
-      "epoch": 0.44482672936139306,
-      "grad_norm": 0.0004770248488057405,
-      "learning_rate": 0.00019999990268990947,
-      "loss": 46.0,
-      "step": 5818
-    },
-    {
-      "epoch": 0.44490318634478276,
-      "grad_norm": 0.0012241422664374113,
-      "learning_rate": 0.00019999990265639762,
-      "loss": 46.0,
-      "step": 5819
-    },
-    {
-      "epoch": 0.44497964332817247,
-      "grad_norm": 0.0010276573011651635,
-      "learning_rate": 0.00019999990262287997,
-      "loss": 46.0,
-      "step": 5820
-    },
-    {
-      "epoch": 0.44505610031156223,
-      "grad_norm": 0.002679398050531745,
-      "learning_rate": 0.00019999990258935657,
-      "loss": 46.0,
-      "step": 5821
-    },
-    {
-      "epoch": 0.44513255729495194,
-      "grad_norm": 0.0008766966057009995,
-      "learning_rate": 0.0001999999025558274,
-      "loss": 46.0,
-      "step": 5822
-    },
-    {
-      "epoch": 0.44520901427834164,
-      "grad_norm": 0.0024377433583140373,
-      "learning_rate": 0.00019999990252229246,
-      "loss": 46.0,
-      "step": 5823
-    },
-    {
-      "epoch": 0.44528547126173135,
-      "grad_norm": 0.0020526135340332985,
-      "learning_rate": 0.00019999990248875172,
-      "loss": 46.0,
-      "step": 5824
-    },
-    {
-      "epoch": 0.4453619282451211,
-      "grad_norm": 0.0018091736128553748,
-      "learning_rate": 0.00019999990245520529,
-      "loss": 46.0,
-      "step": 5825
-    },
-    {
-      "epoch": 0.4454383852285108,
-      "grad_norm": 0.0027146933134645224,
-      "learning_rate": 0.000199999902421653,
-      "loss": 46.0,
-      "step": 5826
-    },
-    {
-      "epoch": 0.4455148422119005,
-      "grad_norm": 0.0013257615501061082,
-      "learning_rate": 0.00019999990238809501,
-      "loss": 46.0,
-      "step": 5827
-    },
-    {
-      "epoch": 0.4455912991952902,
-      "grad_norm": 0.00398293137550354,
-      "learning_rate": 0.0001999999023545312,
-      "loss": 46.0,
-      "step": 5828
-    },
-    {
-      "epoch": 0.44566775617868,
-      "grad_norm": 0.0012344809947535396,
-      "learning_rate": 0.00019999990232096165,
-      "loss": 46.0,
-      "step": 5829
-    },
-    {
-      "epoch": 0.4457442131620697,
-      "grad_norm": 0.0004443094367161393,
-      "learning_rate": 0.00019999990228738632,
-      "loss": 46.0,
-      "step": 5830
-    },
-    {
-      "epoch": 0.4458206701454594,
-      "grad_norm": 0.006870817393064499,
-      "learning_rate": 0.00019999990225380522,
-      "loss": 46.0,
-      "step": 5831
-    },
-    {
-      "epoch": 0.44589712712884916,
-      "grad_norm": 0.0007931660511530936,
-      "learning_rate": 0.00019999990222021837,
-      "loss": 46.0,
-      "step": 5832
-    },
-    {
-      "epoch": 0.44597358411223886,
-      "grad_norm": 0.0028537327889353037,
-      "learning_rate": 0.00019999990218662572,
-      "loss": 46.0,
-      "step": 5833
-    },
-    {
-      "epoch": 0.44605004109562857,
-      "grad_norm": 0.00420724879950285,
-      "learning_rate": 0.0001999999021530273,
-      "loss": 46.0,
-      "step": 5834
-    },
-    {
-      "epoch": 0.4461264980790183,
-      "grad_norm": 0.0012185847153887153,
-      "learning_rate": 0.00019999990211942316,
-      "loss": 46.0,
-      "step": 5835
-    },
-    {
-      "epoch": 0.44620295506240804,
-      "grad_norm": 0.0011567919282242656,
-      "learning_rate": 0.00019999990208581321,
-      "loss": 46.0,
-      "step": 5836
-    },
-    {
-      "epoch": 0.44627941204579774,
-      "grad_norm": 0.001151342410594225,
-      "learning_rate": 0.00019999990205219747,
-      "loss": 46.0,
-      "step": 5837
-    },
-    {
-      "epoch": 0.44635586902918745,
-      "grad_norm": 0.0008848869474604726,
-      "learning_rate": 0.000199999902018576,
-      "loss": 46.0,
-      "step": 5838
-    },
-    {
-      "epoch": 0.44643232601257715,
-      "grad_norm": 0.001326793571934104,
-      "learning_rate": 0.00019999990198494875,
-      "loss": 46.0,
-      "step": 5839
-    },
-    {
-      "epoch": 0.4465087829959669,
-      "grad_norm": 0.0017756195738911629,
-      "learning_rate": 0.00019999990195131574,
-      "loss": 46.0,
-      "step": 5840
-    },
-    {
-      "epoch": 0.4465852399793566,
-      "grad_norm": 0.00062420783797279,
-      "learning_rate": 0.00019999990191767696,
-      "loss": 46.0,
-      "step": 5841
-    },
-    {
-      "epoch": 0.4466616969627463,
-      "grad_norm": 0.0010073896264657378,
-      "learning_rate": 0.00019999990188403237,
-      "loss": 46.0,
-      "step": 5842
-    },
-    {
-      "epoch": 0.44673815394613603,
-      "grad_norm": 0.002054106444120407,
-      "learning_rate": 0.00019999990185038205,
-      "loss": 46.0,
-      "step": 5843
-    },
-    {
-      "epoch": 0.4468146109295258,
-      "grad_norm": 0.0009332704939879477,
-      "learning_rate": 0.00019999990181672594,
-      "loss": 46.0,
-      "step": 5844
-    },
-    {
-      "epoch": 0.4468910679129155,
-      "grad_norm": 0.0018146971706300974,
-      "learning_rate": 0.0001999999017830641,
-      "loss": 46.0,
-      "step": 5845
-    },
-    {
-      "epoch": 0.4469675248963052,
-      "grad_norm": 0.00193692184984684,
-      "learning_rate": 0.00019999990174939645,
-      "loss": 46.0,
-      "step": 5846
-    },
-    {
-      "epoch": 0.44704398187969496,
-      "grad_norm": 0.0019344044849276543,
-      "learning_rate": 0.00019999990171572303,
-      "loss": 46.0,
-      "step": 5847
-    },
-    {
-      "epoch": 0.44712043886308467,
-      "grad_norm": 0.0016667699674144387,
-      "learning_rate": 0.00019999990168204386,
-      "loss": 46.0,
-      "step": 5848
-    },
-    {
-      "epoch": 0.4471968958464744,
-      "grad_norm": 0.0010337602579966187,
-      "learning_rate": 0.00019999990164835892,
-      "loss": 46.0,
-      "step": 5849
-    },
-    {
-      "epoch": 0.4472733528298641,
-      "grad_norm": 0.001886464306153357,
-      "learning_rate": 0.00019999990161466817,
-      "loss": 46.0,
-      "step": 5850
-    },
-    {
-      "epoch": 0.44734980981325384,
-      "grad_norm": 0.0007727887132205069,
-      "learning_rate": 0.0001999999015809717,
-      "loss": 46.0,
-      "step": 5851
-    },
-    {
-      "epoch": 0.44742626679664355,
-      "grad_norm": 0.0010142027167603374,
-      "learning_rate": 0.00019999990154726945,
-      "loss": 46.0,
-      "step": 5852
-    },
-    {
-      "epoch": 0.44750272378003325,
-      "grad_norm": 0.0023035744670778513,
-      "learning_rate": 0.00019999990151356141,
-      "loss": 46.0,
-      "step": 5853
-    },
-    {
-      "epoch": 0.44757918076342296,
-      "grad_norm": 0.0008648578659631312,
-      "learning_rate": 0.0001999999014798476,
-      "loss": 46.0,
-      "step": 5854
-    },
-    {
-      "epoch": 0.4476556377468127,
-      "grad_norm": 0.0037339525297284126,
-      "learning_rate": 0.00019999990144612805,
-      "loss": 46.0,
-      "step": 5855
-    },
-    {
-      "epoch": 0.4477320947302024,
-      "grad_norm": 0.0024114008992910385,
-      "learning_rate": 0.00019999990141240272,
-      "loss": 46.0,
-      "step": 5856
-    },
-    {
-      "epoch": 0.44780855171359213,
-      "grad_norm": 0.0029424072708934546,
-      "learning_rate": 0.00019999990137867162,
-      "loss": 46.0,
-      "step": 5857
-    },
-    {
-      "epoch": 0.44788500869698183,
-      "grad_norm": 0.0012212153524160385,
-      "learning_rate": 0.00019999990134493477,
-      "loss": 46.0,
-      "step": 5858
-    },
-    {
-      "epoch": 0.4479614656803716,
-      "grad_norm": 0.000532503763679415,
-      "learning_rate": 0.00019999990131119212,
-      "loss": 46.0,
-      "step": 5859
-    },
-    {
-      "epoch": 0.4480379226637613,
-      "grad_norm": 0.0035747734364122152,
-      "learning_rate": 0.0001999999012774437,
-      "loss": 46.0,
-      "step": 5860
-    },
-    {
-      "epoch": 0.448114379647151,
-      "grad_norm": 0.0007307583000510931,
-      "learning_rate": 0.00019999990124368953,
-      "loss": 46.0,
-      "step": 5861
-    },
-    {
-      "epoch": 0.44819083663054077,
-      "grad_norm": 0.007901514880359173,
-      "learning_rate": 0.00019999990120992957,
-      "loss": 46.0,
-      "step": 5862
-    },
-    {
-      "epoch": 0.4482672936139305,
-      "grad_norm": 0.0010505786631256342,
-      "learning_rate": 0.00019999990117616385,
-      "loss": 46.0,
-      "step": 5863
-    },
-    {
-      "epoch": 0.4483437505973202,
-      "grad_norm": 0.001217624288983643,
-      "learning_rate": 0.00019999990114239236,
-      "loss": 46.0,
-      "step": 5864
-    },
-    {
-      "epoch": 0.4484202075807099,
-      "grad_norm": 0.0005843573017045856,
-      "learning_rate": 0.0001999999011086151,
-      "loss": 46.0,
-      "step": 5865
-    },
-    {
-      "epoch": 0.44849666456409965,
-      "grad_norm": 0.0032574895303696394,
-      "learning_rate": 0.0001999999010748321,
-      "loss": 46.0,
-      "step": 5866
-    },
-    {
-      "epoch": 0.44857312154748935,
-      "grad_norm": 0.00047823303611949086,
-      "learning_rate": 0.00019999990104104328,
-      "loss": 46.0,
-      "step": 5867
-    },
-    {
-      "epoch": 0.44864957853087906,
-      "grad_norm": 0.0006631818832829595,
-      "learning_rate": 0.00019999990100724873,
-      "loss": 46.0,
-      "step": 5868
-    },
-    {
-      "epoch": 0.44872603551426876,
-      "grad_norm": 0.0012045379262417555,
-      "learning_rate": 0.00019999990097344838,
-      "loss": 46.0,
-      "step": 5869
-    },
-    {
-      "epoch": 0.4488024924976585,
-      "grad_norm": 0.0004973552422598004,
-      "learning_rate": 0.00019999990093964227,
-      "loss": 46.0,
-      "step": 5870
-    },
-    {
-      "epoch": 0.44887894948104823,
-      "grad_norm": 0.001408767537213862,
-      "learning_rate": 0.00019999990090583037,
-      "loss": 46.0,
-      "step": 5871
-    },
-    {
-      "epoch": 0.44895540646443793,
-      "grad_norm": 0.0006852729711681604,
-      "learning_rate": 0.00019999990087201278,
-      "loss": 46.0,
-      "step": 5872
-    },
-    {
-      "epoch": 0.44903186344782764,
-      "grad_norm": 0.0036148359067738056,
-      "learning_rate": 0.00019999990083818933,
-      "loss": 46.0,
-      "step": 5873
-    },
-    {
-      "epoch": 0.4491083204312174,
-      "grad_norm": 0.0009509313968010247,
-      "learning_rate": 0.00019999990080436016,
-      "loss": 46.0,
-      "step": 5874
-    },
-    {
-      "epoch": 0.4491847774146071,
-      "grad_norm": 0.001048155128955841,
-      "learning_rate": 0.0001999999007705252,
-      "loss": 46.0,
-      "step": 5875
-    },
-    {
-      "epoch": 0.4492612343979968,
-      "grad_norm": 0.00441478518769145,
-      "learning_rate": 0.00019999990073668448,
-      "loss": 46.0,
-      "step": 5876
-    },
-    {
-      "epoch": 0.4493376913813866,
-      "grad_norm": 0.0077624935656785965,
-      "learning_rate": 0.00019999990070283802,
-      "loss": 46.0,
-      "step": 5877
-    },
-    {
-      "epoch": 0.4494141483647763,
-      "grad_norm": 0.0004367143556009978,
-      "learning_rate": 0.00019999990066898573,
-      "loss": 46.0,
-      "step": 5878
-    },
-    {
-      "epoch": 0.449490605348166,
-      "grad_norm": 0.0009758782689459622,
-      "learning_rate": 0.0001999999006351277,
-      "loss": 46.0,
-      "step": 5879
-    },
-    {
-      "epoch": 0.4495670623315557,
-      "grad_norm": 0.0020039204973727465,
-      "learning_rate": 0.00019999990060126391,
-      "loss": 46.0,
-      "step": 5880
-    },
-    {
-      "epoch": 0.44964351931494545,
-      "grad_norm": 0.0015301794046536088,
-      "learning_rate": 0.00019999990056739433,
-      "loss": 46.0,
-      "step": 5881
-    },
-    {
-      "epoch": 0.44971997629833516,
-      "grad_norm": 0.0009870027424767613,
-      "learning_rate": 0.000199999900533519,
-      "loss": 46.0,
-      "step": 5882
-    },
-    {
-      "epoch": 0.44979643328172486,
-      "grad_norm": 0.002041150815784931,
-      "learning_rate": 0.00019999990049963788,
-      "loss": 46.0,
-      "step": 5883
-    },
-    {
-      "epoch": 0.44987289026511457,
-      "grad_norm": 0.0011921870755031705,
-      "learning_rate": 0.00019999990046575098,
-      "loss": 46.0,
-      "step": 5884
-    },
-    {
-      "epoch": 0.4499493472485043,
-      "grad_norm": 0.0007126278942450881,
-      "learning_rate": 0.00019999990043185835,
-      "loss": 46.0,
-      "step": 5885
-    },
-    {
-      "epoch": 0.45002580423189403,
-      "grad_norm": 0.001590652042068541,
-      "learning_rate": 0.00019999990039795993,
-      "loss": 46.0,
-      "step": 5886
-    },
-    {
-      "epoch": 0.45010226121528374,
-      "grad_norm": 0.0006485821213573217,
-      "learning_rate": 0.00019999990036405577,
-      "loss": 46.0,
-      "step": 5887
-    },
-    {
-      "epoch": 0.45017871819867344,
-      "grad_norm": 0.002779987407848239,
-      "learning_rate": 0.0001999999003301458,
-      "loss": 46.0,
-      "step": 5888
-    },
-    {
-      "epoch": 0.4502551751820632,
-      "grad_norm": 0.004117012023925781,
-      "learning_rate": 0.00019999990029623006,
-      "loss": 46.0,
-      "step": 5889
-    },
-    {
-      "epoch": 0.4503316321654529,
-      "grad_norm": 0.0006213849992491305,
-      "learning_rate": 0.00019999990026230857,
-      "loss": 46.0,
-      "step": 5890
-    },
-    {
-      "epoch": 0.4504080891488426,
-      "grad_norm": 0.0013996735215187073,
-      "learning_rate": 0.0001999999002283813,
-      "loss": 46.0,
-      "step": 5891
-    },
-    {
-      "epoch": 0.4504845461322324,
-      "grad_norm": 0.0005633053951896727,
-      "learning_rate": 0.00019999990019444827,
-      "loss": 46.0,
-      "step": 5892
-    },
-    {
-      "epoch": 0.4505610031156221,
-      "grad_norm": 0.0014910736354067922,
-      "learning_rate": 0.00019999990016050947,
-      "loss": 46.0,
-      "step": 5893
-    },
-    {
-      "epoch": 0.4506374600990118,
-      "grad_norm": 0.0011204888578504324,
-      "learning_rate": 0.0001999999001265649,
-      "loss": 46.0,
-      "step": 5894
-    },
-    {
-      "epoch": 0.4507139170824015,
-      "grad_norm": 0.0028882878832519054,
-      "learning_rate": 0.00019999990009261459,
-      "loss": 46.0,
-      "step": 5895
-    },
-    {
-      "epoch": 0.45079037406579126,
-      "grad_norm": 0.003089252393692732,
-      "learning_rate": 0.00019999990005865846,
-      "loss": 46.0,
-      "step": 5896
-    },
-    {
-      "epoch": 0.45086683104918096,
-      "grad_norm": 0.0008939317194744945,
-      "learning_rate": 0.00019999990002469658,
-      "loss": 46.0,
-      "step": 5897
-    },
-    {
-      "epoch": 0.45094328803257067,
-      "grad_norm": 0.0011388816637918353,
-      "learning_rate": 0.00019999989999072894,
-      "loss": 46.0,
-      "step": 5898
-    },
-    {
-      "epoch": 0.45101974501596037,
-      "grad_norm": 0.002678181044757366,
-      "learning_rate": 0.0001999998999567555,
-      "loss": 46.0,
-      "step": 5899
-    },
-    {
-      "epoch": 0.45109620199935013,
-      "grad_norm": 0.0006464957259595394,
-      "learning_rate": 0.0001999998999227763,
-      "loss": 46.0,
-      "step": 5900
-    },
-    {
-      "epoch": 0.45117265898273984,
-      "grad_norm": 0.001799222780391574,
-      "learning_rate": 0.00019999989988879135,
-      "loss": 46.0,
-      "step": 5901
-    },
-    {
-      "epoch": 0.45124911596612954,
-      "grad_norm": 0.007783942390233278,
-      "learning_rate": 0.00019999989985480064,
-      "loss": 46.0,
-      "step": 5902
-    },
-    {
-      "epoch": 0.45132557294951925,
-      "grad_norm": 0.001156967948190868,
-      "learning_rate": 0.00019999989982080412,
-      "loss": 46.0,
-      "step": 5903
-    },
-    {
-      "epoch": 0.451402029932909,
-      "grad_norm": 0.0018953500548377633,
-      "learning_rate": 0.00019999989978680186,
-      "loss": 46.0,
-      "step": 5904
-    },
-    {
-      "epoch": 0.4514784869162987,
-      "grad_norm": 0.0078031993471086025,
-      "learning_rate": 0.00019999989975279383,
-      "loss": 46.0,
-      "step": 5905
-    },
-    {
-      "epoch": 0.4515549438996884,
-      "grad_norm": 0.0007886604289524257,
-      "learning_rate": 0.00019999989971878002,
-      "loss": 46.0,
-      "step": 5906
-    },
-    {
-      "epoch": 0.4516314008830782,
-      "grad_norm": 0.0007624428835697472,
-      "learning_rate": 0.00019999989968476044,
-      "loss": 46.0,
-      "step": 5907
-    },
-    {
-      "epoch": 0.4517078578664679,
-      "grad_norm": 0.002622829517349601,
-      "learning_rate": 0.00019999989965073511,
-      "loss": 46.0,
-      "step": 5908
-    },
-    {
-      "epoch": 0.4517843148498576,
-      "grad_norm": 0.0009754533530212939,
-      "learning_rate": 0.000199999899616704,
-      "loss": 46.0,
-      "step": 5909
-    },
-    {
-      "epoch": 0.4518607718332473,
-      "grad_norm": 0.0009104540222324431,
-      "learning_rate": 0.0001999998995826671,
-      "loss": 46.0,
-      "step": 5910
-    },
-    {
-      "epoch": 0.45193722881663706,
-      "grad_norm": 0.0009701043018139899,
-      "learning_rate": 0.00019999989954862444,
-      "loss": 46.0,
-      "step": 5911
-    },
-    {
-      "epoch": 0.45201368580002677,
-      "grad_norm": 0.0005307946121320128,
-      "learning_rate": 0.00019999989951457602,
-      "loss": 46.0,
-      "step": 5912
-    },
-    {
-      "epoch": 0.45209014278341647,
-      "grad_norm": 0.00223224307410419,
-      "learning_rate": 0.00019999989948052185,
-      "loss": 46.0,
-      "step": 5913
-    },
-    {
-      "epoch": 0.4521665997668062,
-      "grad_norm": 0.0025332423392683268,
-      "learning_rate": 0.00019999989944646188,
-      "loss": 46.0,
-      "step": 5914
-    },
-    {
-      "epoch": 0.45224305675019594,
-      "grad_norm": 0.002023973735049367,
-      "learning_rate": 0.00019999989941239614,
-      "loss": 46.0,
-      "step": 5915
-    },
-    {
-      "epoch": 0.45231951373358564,
-      "grad_norm": 0.0026505168061703444,
-      "learning_rate": 0.00019999989937832463,
-      "loss": 46.0,
-      "step": 5916
-    },
-    {
-      "epoch": 0.45239597071697535,
-      "grad_norm": 0.005159546155482531,
-      "learning_rate": 0.0001999998993442474,
-      "loss": 46.0,
-      "step": 5917
-    },
-    {
-      "epoch": 0.4524724277003651,
-      "grad_norm": 0.001181524246931076,
-      "learning_rate": 0.00019999989931016434,
-      "loss": 46.0,
-      "step": 5918
-    },
-    {
-      "epoch": 0.4525488846837548,
-      "grad_norm": 0.002479558577761054,
-      "learning_rate": 0.00019999989927607553,
-      "loss": 46.0,
-      "step": 5919
-    },
-    {
-      "epoch": 0.4526253416671445,
-      "grad_norm": 0.001620599301531911,
-      "learning_rate": 0.00019999989924198095,
-      "loss": 46.0,
-      "step": 5920
-    },
-    {
-      "epoch": 0.4527017986505342,
-      "grad_norm": 0.0012340190587565303,
-      "learning_rate": 0.0001999998992078806,
-      "loss": 46.0,
-      "step": 5921
-    },
-    {
-      "epoch": 0.452778255633924,
-      "grad_norm": 0.0010031991405412555,
-      "learning_rate": 0.0001999998991737745,
-      "loss": 46.0,
-      "step": 5922
-    },
-    {
-      "epoch": 0.4528547126173137,
-      "grad_norm": 0.0005971941282041371,
-      "learning_rate": 0.0001999998991396626,
-      "loss": 46.0,
-      "step": 5923
-    },
-    {
-      "epoch": 0.4529311696007034,
-      "grad_norm": 0.00413657259196043,
-      "learning_rate": 0.00019999989910554495,
-      "loss": 46.0,
-      "step": 5924
-    },
-    {
-      "epoch": 0.4530076265840931,
-      "grad_norm": 0.0013001210754737258,
-      "learning_rate": 0.00019999989907142153,
-      "loss": 46.0,
-      "step": 5925
-    },
-    {
-      "epoch": 0.45308408356748286,
-      "grad_norm": 0.0014844145625829697,
-      "learning_rate": 0.00019999989903729233,
-      "loss": 46.0,
-      "step": 5926
-    },
-    {
-      "epoch": 0.45316054055087257,
-      "grad_norm": 0.0013648431049659848,
-      "learning_rate": 0.00019999989900315737,
-      "loss": 46.0,
-      "step": 5927
-    },
-    {
-      "epoch": 0.4532369975342623,
-      "grad_norm": 0.0011310586705803871,
-      "learning_rate": 0.00019999989896901663,
-      "loss": 46.0,
-      "step": 5928
-    },
-    {
-      "epoch": 0.453313454517652,
-      "grad_norm": 0.0011114954249933362,
-      "learning_rate": 0.0001999998989348701,
-      "loss": 46.0,
-      "step": 5929
-    },
-    {
-      "epoch": 0.45338991150104174,
-      "grad_norm": 0.0037474429700523615,
-      "learning_rate": 0.00019999989890071785,
-      "loss": 46.0,
-      "step": 5930
-    },
-    {
-      "epoch": 0.45346636848443145,
-      "grad_norm": 0.006916186306625605,
-      "learning_rate": 0.00019999989886655982,
-      "loss": 46.0,
-      "step": 5931
-    },
-    {
-      "epoch": 0.45354282546782115,
-      "grad_norm": 0.0017635890981182456,
-      "learning_rate": 0.00019999989883239598,
-      "loss": 46.0,
-      "step": 5932
-    },
-    {
-      "epoch": 0.4536192824512109,
-      "grad_norm": 0.0023119018878787756,
-      "learning_rate": 0.00019999989879822643,
-      "loss": 46.0,
-      "step": 5933
-    },
-    {
-      "epoch": 0.4536957394346006,
-      "grad_norm": 0.003179059596732259,
-      "learning_rate": 0.00019999989876405105,
-      "loss": 46.0,
-      "step": 5934
-    },
-    {
-      "epoch": 0.4537721964179903,
-      "grad_norm": 0.0011554177617654204,
-      "learning_rate": 0.00019999989872986995,
-      "loss": 46.0,
-      "step": 5935
-    },
-    {
-      "epoch": 0.45384865340138003,
-      "grad_norm": 0.001774788019247353,
-      "learning_rate": 0.00019999989869568308,
-      "loss": 46.0,
-      "step": 5936
-    },
-    {
-      "epoch": 0.4539251103847698,
-      "grad_norm": 0.0020505853462964296,
-      "learning_rate": 0.0001999998986614904,
-      "loss": 46.0,
-      "step": 5937
-    },
-    {
-      "epoch": 0.4540015673681595,
-      "grad_norm": 0.0008048502495512366,
-      "learning_rate": 0.00019999989862729196,
-      "loss": 46.0,
-      "step": 5938
-    },
-    {
-      "epoch": 0.4540780243515492,
-      "grad_norm": 0.0019646461587399244,
-      "learning_rate": 0.00019999989859308776,
-      "loss": 46.0,
-      "step": 5939
-    },
-    {
-      "epoch": 0.4541544813349389,
-      "grad_norm": 0.0013704269658774137,
-      "learning_rate": 0.0001999998985588778,
-      "loss": 46.0,
-      "step": 5940
-    },
-    {
-      "epoch": 0.45423093831832867,
-      "grad_norm": 0.001937161199748516,
-      "learning_rate": 0.00019999989852466206,
-      "loss": 46.0,
-      "step": 5941
-    },
-    {
-      "epoch": 0.4543073953017184,
-      "grad_norm": 0.0017611797666177154,
-      "learning_rate": 0.00019999989849044057,
-      "loss": 46.0,
-      "step": 5942
-    },
-    {
-      "epoch": 0.4543838522851081,
-      "grad_norm": 0.0006365999579429626,
-      "learning_rate": 0.00019999989845621328,
-      "loss": 46.0,
-      "step": 5943
-    },
-    {
-      "epoch": 0.4544603092684978,
-      "grad_norm": 0.001306456862948835,
-      "learning_rate": 0.00019999989842198022,
-      "loss": 46.0,
-      "step": 5944
-    },
-    {
-      "epoch": 0.45453676625188755,
-      "grad_norm": 0.0020896310452371836,
-      "learning_rate": 0.00019999989838774144,
-      "loss": 46.0,
-      "step": 5945
-    },
-    {
-      "epoch": 0.45461322323527725,
-      "grad_norm": 0.0011162115260958672,
-      "learning_rate": 0.00019999989835349684,
-      "loss": 46.0,
-      "step": 5946
-    },
-    {
-      "epoch": 0.45468968021866696,
-      "grad_norm": 0.0044331676326692104,
-      "learning_rate": 0.0001999998983192465,
-      "loss": 46.0,
-      "step": 5947
-    },
-    {
-      "epoch": 0.4547661372020567,
-      "grad_norm": 0.0015705449040979147,
-      "learning_rate": 0.00019999989828499038,
-      "loss": 46.0,
-      "step": 5948
-    },
-    {
-      "epoch": 0.4548425941854464,
-      "grad_norm": 0.0033875214867293835,
-      "learning_rate": 0.00019999989825072846,
-      "loss": 46.0,
-      "step": 5949
-    },
-    {
-      "epoch": 0.45491905116883613,
-      "grad_norm": 0.0010959822684526443,
-      "learning_rate": 0.0001999998982164608,
-      "loss": 46.0,
-      "step": 5950
-    },
-    {
-      "epoch": 0.45499550815222584,
-      "grad_norm": 0.0005235755234025419,
-      "learning_rate": 0.0001999998981821874,
-      "loss": 46.0,
-      "step": 5951
-    },
-    {
-      "epoch": 0.4550719651356156,
-      "grad_norm": 0.0021660521160811186,
-      "learning_rate": 0.00019999989814790817,
-      "loss": 46.0,
-      "step": 5952
-    },
-    {
-      "epoch": 0.4551484221190053,
-      "grad_norm": 0.01906907930970192,
-      "learning_rate": 0.0001999998981136232,
-      "loss": 46.0,
-      "step": 5953
-    },
-    {
-      "epoch": 0.455224879102395,
-      "grad_norm": 0.0008477027877233922,
-      "learning_rate": 0.0001999998980793325,
-      "loss": 46.0,
-      "step": 5954
-    },
-    {
-      "epoch": 0.4553013360857847,
-      "grad_norm": 0.0014201845042407513,
-      "learning_rate": 0.00019999989804503598,
-      "loss": 46.0,
-      "step": 5955
-    },
-    {
-      "epoch": 0.4553777930691745,
-      "grad_norm": 0.0007048692204989493,
-      "learning_rate": 0.0001999998980107337,
-      "loss": 46.0,
-      "step": 5956
-    },
-    {
-      "epoch": 0.4554542500525642,
-      "grad_norm": 0.0034078089520335197,
-      "learning_rate": 0.00019999989797642563,
-      "loss": 46.0,
-      "step": 5957
-    },
-    {
-      "epoch": 0.4555307070359539,
-      "grad_norm": 0.0017970558255910873,
-      "learning_rate": 0.00019999989794211183,
-      "loss": 46.0,
-      "step": 5958
-    },
-    {
-      "epoch": 0.4556071640193436,
-      "grad_norm": 0.00046388880582526326,
-      "learning_rate": 0.00019999989790779225,
-      "loss": 46.0,
-      "step": 5959
-    },
-    {
-      "epoch": 0.45568362100273335,
-      "grad_norm": 0.000561626220587641,
-      "learning_rate": 0.0001999998978734669,
-      "loss": 46.0,
-      "step": 5960
-    },
-    {
-      "epoch": 0.45576007798612306,
-      "grad_norm": 0.0004251794016454369,
-      "learning_rate": 0.00019999989783913574,
-      "loss": 46.0,
-      "step": 5961
-    },
-    {
-      "epoch": 0.45583653496951276,
-      "grad_norm": 0.0011329480912536383,
-      "learning_rate": 0.00019999989780479887,
-      "loss": 46.0,
-      "step": 5962
-    },
-    {
-      "epoch": 0.4559129919529025,
-      "grad_norm": 0.00039562658639624715,
-      "learning_rate": 0.0001999998977704562,
-      "loss": 46.0,
-      "step": 5963
-    },
-    {
-      "epoch": 0.45598944893629223,
-      "grad_norm": 0.004520610440522432,
-      "learning_rate": 0.00019999989773610778,
-      "loss": 46.0,
-      "step": 5964
-    },
-    {
-      "epoch": 0.45606590591968194,
-      "grad_norm": 0.001086476375348866,
-      "learning_rate": 0.00019999989770175356,
-      "loss": 46.0,
-      "step": 5965
-    },
-    {
-      "epoch": 0.45614236290307164,
-      "grad_norm": 0.0010470702545717359,
-      "learning_rate": 0.00019999989766739356,
-      "loss": 46.0,
-      "step": 5966
-    },
-    {
-      "epoch": 0.4562188198864614,
-      "grad_norm": 0.0009131265105679631,
-      "learning_rate": 0.00019999989763302788,
-      "loss": 46.0,
-      "step": 5967
-    },
-    {
-      "epoch": 0.4562952768698511,
-      "grad_norm": 0.002551504410803318,
-      "learning_rate": 0.00019999989759865634,
-      "loss": 46.0,
-      "step": 5968
-    },
-    {
-      "epoch": 0.4563717338532408,
-      "grad_norm": 0.000417816627305001,
-      "learning_rate": 0.00019999989756427905,
-      "loss": 46.0,
-      "step": 5969
-    },
-    {
-      "epoch": 0.4564481908366305,
-      "grad_norm": 0.003151189535856247,
-      "learning_rate": 0.00019999989752989602,
-      "loss": 46.0,
-      "step": 5970
-    },
-    {
-      "epoch": 0.4565246478200203,
-      "grad_norm": 0.0010885881492868066,
-      "learning_rate": 0.0001999998974955072,
-      "loss": 46.0,
-      "step": 5971
-    },
-    {
-      "epoch": 0.45660110480341,
-      "grad_norm": 0.00210529169999063,
-      "learning_rate": 0.00019999989746111258,
-      "loss": 46.0,
-      "step": 5972
-    },
-    {
-      "epoch": 0.4566775617867997,
-      "grad_norm": 0.0015237333718687296,
-      "learning_rate": 0.00019999989742671223,
-      "loss": 46.0,
-      "step": 5973
-    },
-    {
-      "epoch": 0.4567540187701894,
-      "grad_norm": 0.0024765983689576387,
-      "learning_rate": 0.0001999998973923061,
-      "loss": 46.0,
-      "step": 5974
-    },
-    {
-      "epoch": 0.45683047575357916,
-      "grad_norm": 0.0013245424488559365,
-      "learning_rate": 0.00019999989735789423,
-      "loss": 46.0,
-      "step": 5975
-    },
-    {
-      "epoch": 0.45690693273696886,
-      "grad_norm": 0.0010874953586608171,
-      "learning_rate": 0.00019999989732347656,
-      "loss": 46.0,
-      "step": 5976
-    },
-    {
-      "epoch": 0.45698338972035857,
-      "grad_norm": 0.00095972022973001,
-      "learning_rate": 0.0001999998972890531,
-      "loss": 46.0,
-      "step": 5977
-    },
-    {
-      "epoch": 0.45705984670374833,
-      "grad_norm": 0.000726853497326374,
-      "learning_rate": 0.0001999998972546239,
-      "loss": 46.0,
-      "step": 5978
-    },
-    {
-      "epoch": 0.45713630368713803,
-      "grad_norm": 0.0010844114003702998,
-      "learning_rate": 0.00019999989722018893,
-      "loss": 46.0,
-      "step": 5979
-    },
-    {
-      "epoch": 0.45721276067052774,
-      "grad_norm": 0.0014699650928378105,
-      "learning_rate": 0.0001999998971857482,
-      "loss": 46.0,
-      "step": 5980
-    },
-    {
-      "epoch": 0.45728921765391745,
-      "grad_norm": 0.0004494552849791944,
-      "learning_rate": 0.00019999989715130167,
-      "loss": 46.0,
-      "step": 5981
-    },
-    {
-      "epoch": 0.4573656746373072,
-      "grad_norm": 0.001418429659679532,
-      "learning_rate": 0.0001999998971168494,
-      "loss": 46.0,
-      "step": 5982
-    },
-    {
-      "epoch": 0.4574421316206969,
-      "grad_norm": 0.0006162348436191678,
-      "learning_rate": 0.00019999989708239133,
-      "loss": 46.0,
-      "step": 5983
-    },
-    {
-      "epoch": 0.4575185886040866,
-      "grad_norm": 0.001832794165238738,
-      "learning_rate": 0.0001999998970479275,
-      "loss": 46.0,
-      "step": 5984
-    },
-    {
-      "epoch": 0.4575950455874763,
-      "grad_norm": 0.0024114756379276514,
-      "learning_rate": 0.00019999989701345792,
-      "loss": 46.0,
-      "step": 5985
-    },
-    {
-      "epoch": 0.4576715025708661,
-      "grad_norm": 0.0015525853959843516,
-      "learning_rate": 0.00019999989697898257,
-      "loss": 46.0,
-      "step": 5986
-    },
-    {
-      "epoch": 0.4577479595542558,
-      "grad_norm": 0.0010440442711114883,
-      "learning_rate": 0.00019999989694450144,
-      "loss": 46.0,
-      "step": 5987
-    },
-    {
-      "epoch": 0.4578244165376455,
-      "grad_norm": 0.002270434284582734,
-      "learning_rate": 0.00019999989691001454,
-      "loss": 46.0,
-      "step": 5988
-    },
-    {
-      "epoch": 0.4579008735210352,
-      "grad_norm": 0.0013988602440804243,
-      "learning_rate": 0.00019999989687552184,
-      "loss": 46.0,
-      "step": 5989
-    },
-    {
-      "epoch": 0.45797733050442496,
-      "grad_norm": 0.001036861795000732,
-      "learning_rate": 0.0001999998968410234,
-      "loss": 46.0,
-      "step": 5990
-    },
-    {
-      "epoch": 0.45805378748781467,
-      "grad_norm": 0.000586047419346869,
-      "learning_rate": 0.0001999998968065192,
-      "loss": 46.0,
-      "step": 5991
-    },
-    {
-      "epoch": 0.4581302444712044,
-      "grad_norm": 0.002678936580196023,
-      "learning_rate": 0.00019999989677200924,
-      "loss": 46.0,
-      "step": 5992
-    },
-    {
-      "epoch": 0.45820670145459413,
-      "grad_norm": 0.0021975308191031218,
-      "learning_rate": 0.0001999998967374935,
-      "loss": 46.0,
-      "step": 5993
-    },
-    {
-      "epoch": 0.45828315843798384,
-      "grad_norm": 0.00013177323853597045,
-      "learning_rate": 0.00019999989670297196,
-      "loss": 46.0,
-      "step": 5994
-    },
-    {
-      "epoch": 0.45835961542137355,
-      "grad_norm": 0.001314301509410143,
-      "learning_rate": 0.00019999989666844467,
-      "loss": 46.0,
-      "step": 5995
-    },
-    {
-      "epoch": 0.45843607240476325,
-      "grad_norm": 0.0028579598292708397,
-      "learning_rate": 0.00019999989663391162,
-      "loss": 46.0,
-      "step": 5996
-    },
-    {
-      "epoch": 0.458512529388153,
-      "grad_norm": 0.0021432386711239815,
-      "learning_rate": 0.00019999989659937278,
-      "loss": 46.0,
-      "step": 5997
-    },
-    {
-      "epoch": 0.4585889863715427,
-      "grad_norm": 0.0007212893106043339,
-      "learning_rate": 0.0001999998965648282,
-      "loss": 46.0,
-      "step": 5998
-    },
-    {
-      "epoch": 0.4586654433549324,
-      "grad_norm": 0.002305446658283472,
-      "learning_rate": 0.00019999989653027785,
-      "loss": 46.0,
-      "step": 5999
-    },
-    {
-      "epoch": 0.45874190033832213,
-      "grad_norm": 0.003508553374558687,
-      "learning_rate": 0.0001999998964957217,
-      "loss": 46.0,
-      "step": 6000
-    },
-    {
-      "epoch": 0.4588183573217119,
-      "grad_norm": 0.0005807048873975873,
-      "learning_rate": 0.0001999998964611598,
-      "loss": 46.0,
-      "step": 6001
-    },
-    {
-      "epoch": 0.4588948143051016,
-      "grad_norm": 0.0016505331732332706,
-      "learning_rate": 0.0001999998964265921,
-      "loss": 46.0,
-      "step": 6002
-    },
-    {
-      "epoch": 0.4589712712884913,
-      "grad_norm": 0.0007950190338306129,
-      "learning_rate": 0.00019999989639201868,
-      "loss": 46.0,
-      "step": 6003
-    },
-    {
-      "epoch": 0.459047728271881,
-      "grad_norm": 0.000992244342342019,
-      "learning_rate": 0.00019999989635743947,
-      "loss": 46.0,
-      "step": 6004
-    },
-    {
-      "epoch": 0.45912418525527077,
-      "grad_norm": 0.002171586500480771,
-      "learning_rate": 0.0001999998963228545,
-      "loss": 46.0,
-      "step": 6005
-    },
-    {
-      "epoch": 0.4592006422386605,
-      "grad_norm": 0.0018271973822265863,
-      "learning_rate": 0.00019999989628826373,
-      "loss": 46.0,
-      "step": 6006
-    },
-    {
-      "epoch": 0.4592770992220502,
-      "grad_norm": 0.005714506842195988,
-      "learning_rate": 0.00019999989625366722,
-      "loss": 46.0,
-      "step": 6007
-    },
-    {
-      "epoch": 0.45935355620543994,
-      "grad_norm": 0.0015390970511361957,
-      "learning_rate": 0.0001999998962190649,
-      "loss": 46.0,
-      "step": 6008
-    },
-    {
-      "epoch": 0.45943001318882964,
-      "grad_norm": 0.002417241455987096,
-      "learning_rate": 0.00019999989618445688,
-      "loss": 46.0,
-      "step": 6009
-    },
-    {
-      "epoch": 0.45950647017221935,
-      "grad_norm": 0.004362266045063734,
-      "learning_rate": 0.00019999989614984305,
-      "loss": 46.0,
-      "step": 6010
-    },
-    {
-      "epoch": 0.45958292715560906,
-      "grad_norm": 0.0005166534683667123,
-      "learning_rate": 0.00019999989611522344,
-      "loss": 46.0,
-      "step": 6011
-    },
-    {
-      "epoch": 0.4596593841389988,
-      "grad_norm": 0.00039660531911067665,
-      "learning_rate": 0.0001999998960805981,
-      "loss": 46.0,
-      "step": 6012
-    },
-    {
-      "epoch": 0.4597358411223885,
-      "grad_norm": 0.0018095995765179396,
-      "learning_rate": 0.00019999989604596694,
-      "loss": 46.0,
-      "step": 6013
-    },
-    {
-      "epoch": 0.4598122981057782,
-      "grad_norm": 0.0007541960221715271,
-      "learning_rate": 0.00019999989601133,
-      "loss": 46.0,
-      "step": 6014
-    },
-    {
-      "epoch": 0.45988875508916793,
-      "grad_norm": 0.0006454598042182624,
-      "learning_rate": 0.00019999989597668737,
-      "loss": 46.0,
-      "step": 6015
-    },
-    {
-      "epoch": 0.4599652120725577,
-      "grad_norm": 0.006079236976802349,
-      "learning_rate": 0.00019999989594203892,
-      "loss": 46.0,
-      "step": 6016
-    },
-    {
-      "epoch": 0.4600416690559474,
-      "grad_norm": 0.0026544760912656784,
-      "learning_rate": 0.0001999998959073847,
-      "loss": 46.0,
-      "step": 6017
-    },
-    {
-      "epoch": 0.4601181260393371,
-      "grad_norm": 0.0006069413502700627,
-      "learning_rate": 0.0001999998958727247,
-      "loss": 46.0,
-      "step": 6018
-    },
-    {
-      "epoch": 0.4601945830227268,
-      "grad_norm": 0.0013491831487044692,
-      "learning_rate": 0.00019999989583805897,
-      "loss": 46.0,
-      "step": 6019
-    },
-    {
-      "epoch": 0.46027104000611657,
-      "grad_norm": 0.0005287861567921937,
-      "learning_rate": 0.00019999989580338744,
-      "loss": 46.0,
-      "step": 6020
-    },
-    {
-      "epoch": 0.4603474969895063,
-      "grad_norm": 0.0010659437393769622,
-      "learning_rate": 0.00019999989576871015,
-      "loss": 46.0,
-      "step": 6021
-    },
-    {
-      "epoch": 0.460423953972896,
-      "grad_norm": 0.0011204304173588753,
-      "learning_rate": 0.0001999998957340271,
-      "loss": 46.0,
-      "step": 6022
-    },
-    {
-      "epoch": 0.46050041095628574,
-      "grad_norm": 0.004781865514814854,
-      "learning_rate": 0.00019999989569933826,
-      "loss": 46.0,
-      "step": 6023
-    },
-    {
-      "epoch": 0.46057686793967545,
-      "grad_norm": 0.0009274246403947473,
-      "learning_rate": 0.00019999989566464366,
-      "loss": 46.0,
-      "step": 6024
-    },
-    {
-      "epoch": 0.46065332492306515,
-      "grad_norm": 0.002049343893304467,
-      "learning_rate": 0.00019999989562994328,
-      "loss": 46.0,
-      "step": 6025
-    },
-    {
-      "epoch": 0.46072978190645486,
-      "grad_norm": 0.0008803637465462089,
-      "learning_rate": 0.00019999989559523715,
-      "loss": 46.0,
-      "step": 6026
-    },
-    {
-      "epoch": 0.4608062388898446,
-      "grad_norm": 0.002005344955250621,
-      "learning_rate": 0.00019999989556052525,
-      "loss": 46.0,
-      "step": 6027
-    },
-    {
-      "epoch": 0.4608826958732343,
-      "grad_norm": 0.0008307746611535549,
-      "learning_rate": 0.00019999989552580756,
-      "loss": 46.0,
-      "step": 6028
-    },
-    {
-      "epoch": 0.46095915285662403,
-      "grad_norm": 0.006273068021982908,
-      "learning_rate": 0.0001999998954910841,
-      "loss": 46.0,
-      "step": 6029
-    },
-    {
-      "epoch": 0.46103560984001374,
-      "grad_norm": 0.001752551761455834,
-      "learning_rate": 0.0001999998954563549,
-      "loss": 46.0,
-      "step": 6030
-    },
-    {
-      "epoch": 0.4611120668234035,
-      "grad_norm": 0.0035552845802158117,
-      "learning_rate": 0.00019999989542161993,
-      "loss": 46.0,
-      "step": 6031
-    },
-    {
-      "epoch": 0.4611885238067932,
-      "grad_norm": 0.002459985902532935,
-      "learning_rate": 0.00019999989538687916,
-      "loss": 46.0,
-      "step": 6032
-    },
-    {
-      "epoch": 0.4612649807901829,
-      "grad_norm": 0.005091544706374407,
-      "learning_rate": 0.00019999989535213263,
-      "loss": 46.0,
-      "step": 6033
-    },
-    {
-      "epoch": 0.46134143777357267,
-      "grad_norm": 0.0005674569983966649,
-      "learning_rate": 0.0001999998953173803,
-      "loss": 46.0,
-      "step": 6034
-    },
-    {
-      "epoch": 0.4614178947569624,
-      "grad_norm": 0.0004276817780919373,
-      "learning_rate": 0.00019999989528262226,
-      "loss": 46.0,
-      "step": 6035
-    },
-    {
-      "epoch": 0.4614943517403521,
-      "grad_norm": 0.0017261023167520761,
-      "learning_rate": 0.00019999989524785842,
-      "loss": 46.0,
-      "step": 6036
-    },
-    {
-      "epoch": 0.4615708087237418,
-      "grad_norm": 0.002901071682572365,
-      "learning_rate": 0.00019999989521308882,
-      "loss": 46.0,
-      "step": 6037
-    },
-    {
-      "epoch": 0.46164726570713155,
-      "grad_norm": 0.012594775296747684,
-      "learning_rate": 0.00019999989517831347,
-      "loss": 46.0,
-      "step": 6038
-    },
-    {
-      "epoch": 0.46172372269052125,
-      "grad_norm": 0.0008829089347273111,
-      "learning_rate": 0.00019999989514353232,
-      "loss": 46.0,
-      "step": 6039
-    },
-    {
-      "epoch": 0.46180017967391096,
-      "grad_norm": 0.0010119960643351078,
-      "learning_rate": 0.0001999998951087454,
-      "loss": 46.0,
-      "step": 6040
-    },
-    {
-      "epoch": 0.46187663665730067,
-      "grad_norm": 0.0016015958972275257,
-      "learning_rate": 0.00019999989507395275,
-      "loss": 46.0,
-      "step": 6041
-    },
-    {
-      "epoch": 0.4619530936406904,
-      "grad_norm": 0.0009610798442736268,
-      "learning_rate": 0.00019999989503915425,
-      "loss": 46.0,
-      "step": 6042
-    },
-    {
-      "epoch": 0.46202955062408013,
-      "grad_norm": 0.0019293783698230982,
-      "learning_rate": 0.00019999989500435006,
-      "loss": 46.0,
-      "step": 6043
-    },
-    {
-      "epoch": 0.46210600760746984,
-      "grad_norm": 0.001704701455309987,
-      "learning_rate": 0.00019999989496954007,
-      "loss": 46.0,
-      "step": 6044
-    },
-    {
-      "epoch": 0.46218246459085954,
-      "grad_norm": 0.0010965772671625018,
-      "learning_rate": 0.0001999998949347243,
-      "loss": 46.0,
-      "step": 6045
-    },
-    {
-      "epoch": 0.4622589215742493,
-      "grad_norm": 0.0015787002630531788,
-      "learning_rate": 0.00019999989489990276,
-      "loss": 46.0,
-      "step": 6046
-    },
-    {
-      "epoch": 0.462335378557639,
-      "grad_norm": 0.0008652831893414259,
-      "learning_rate": 0.00019999989486507545,
-      "loss": 46.0,
-      "step": 6047
-    },
-    {
-      "epoch": 0.4624118355410287,
-      "grad_norm": 0.003549166489392519,
-      "learning_rate": 0.0001999998948302424,
-      "loss": 46.0,
-      "step": 6048
-    },
-    {
-      "epoch": 0.4624882925244185,
-      "grad_norm": 0.0005808480200357735,
-      "learning_rate": 0.00019999989479540356,
-      "loss": 46.0,
-      "step": 6049
-    },
-    {
-      "epoch": 0.4625647495078082,
-      "grad_norm": 0.0008718917379155755,
-      "learning_rate": 0.000199999894760559,
-      "loss": 46.0,
-      "step": 6050
-    },
-    {
-      "epoch": 0.4626412064911979,
-      "grad_norm": 0.0018762447871267796,
-      "learning_rate": 0.00019999989472570858,
-      "loss": 46.0,
-      "step": 6051
-    },
-    {
-      "epoch": 0.4627176634745876,
-      "grad_norm": 0.0008820659131743014,
-      "learning_rate": 0.00019999989469085246,
-      "loss": 46.0,
-      "step": 6052
-    },
-    {
-      "epoch": 0.46279412045797735,
-      "grad_norm": 0.0009781663538888097,
-      "learning_rate": 0.00019999989465599053,
-      "loss": 46.0,
-      "step": 6053
-    },
-    {
-      "epoch": 0.46287057744136706,
-      "grad_norm": 0.0018254210008308291,
-      "learning_rate": 0.00019999989462112286,
-      "loss": 46.0,
-      "step": 6054
-    },
-    {
-      "epoch": 0.46294703442475676,
-      "grad_norm": 0.0004859469481743872,
-      "learning_rate": 0.0001999998945862494,
-      "loss": 46.0,
-      "step": 6055
-    },
-    {
-      "epoch": 0.46302349140814647,
-      "grad_norm": 0.0009583156206645072,
-      "learning_rate": 0.00019999989455137017,
-      "loss": 46.0,
-      "step": 6056
-    },
-    {
-      "epoch": 0.46309994839153623,
-      "grad_norm": 0.0005690045072697103,
-      "learning_rate": 0.00019999989451648518,
-      "loss": 46.0,
-      "step": 6057
-    },
-    {
-      "epoch": 0.46317640537492594,
-      "grad_norm": 0.0017215729458257556,
-      "learning_rate": 0.00019999989448159442,
-      "loss": 46.0,
-      "step": 6058
-    },
-    {
-      "epoch": 0.46325286235831564,
-      "grad_norm": 0.0015063463943079114,
-      "learning_rate": 0.00019999989444669788,
-      "loss": 46.0,
-      "step": 6059
-    },
-    {
-      "epoch": 0.46332931934170535,
-      "grad_norm": 0.0025609892327338457,
-      "learning_rate": 0.0001999998944117956,
-      "loss": 46.0,
-      "step": 6060
-    },
-    {
-      "epoch": 0.4634057763250951,
-      "grad_norm": 0.0013538898201659322,
-      "learning_rate": 0.00019999989437688754,
-      "loss": 46.0,
-      "step": 6061
-    },
-    {
-      "epoch": 0.4634822333084848,
-      "grad_norm": 0.016068175435066223,
-      "learning_rate": 0.00019999989434197368,
-      "loss": 46.0,
-      "step": 6062
-    },
-    {
-      "epoch": 0.4635586902918745,
-      "grad_norm": 0.003152010263875127,
-      "learning_rate": 0.00019999989430705405,
-      "loss": 46.0,
-      "step": 6063
-    },
-    {
-      "epoch": 0.4636351472752643,
-      "grad_norm": 0.0006886878982186317,
-      "learning_rate": 0.00019999989427212867,
-      "loss": 46.0,
-      "step": 6064
-    },
-    {
-      "epoch": 0.463711604258654,
-      "grad_norm": 0.0049661388620734215,
-      "learning_rate": 0.00019999989423719755,
-      "loss": 46.0,
-      "step": 6065
-    },
-    {
-      "epoch": 0.4637880612420437,
-      "grad_norm": 0.0017762784846127033,
-      "learning_rate": 0.00019999989420226063,
-      "loss": 46.0,
-      "step": 6066
-    },
-    {
-      "epoch": 0.4638645182254334,
-      "grad_norm": 0.002183932811021805,
-      "learning_rate": 0.00019999989416731793,
-      "loss": 46.0,
-      "step": 6067
-    },
-    {
-      "epoch": 0.46394097520882316,
-      "grad_norm": 0.0007665642187930644,
-      "learning_rate": 0.00019999989413236948,
-      "loss": 46.0,
-      "step": 6068
-    },
-    {
-      "epoch": 0.46401743219221286,
-      "grad_norm": 0.0006764632416889071,
-      "learning_rate": 0.00019999989409741527,
-      "loss": 46.0,
-      "step": 6069
-    },
-    {
-      "epoch": 0.46409388917560257,
-      "grad_norm": 0.0012916539562866092,
-      "learning_rate": 0.00019999989406245528,
-      "loss": 46.0,
-      "step": 6070
-    },
-    {
-      "epoch": 0.4641703461589923,
-      "grad_norm": 0.002496344270184636,
-      "learning_rate": 0.0001999998940274895,
-      "loss": 46.0,
-      "step": 6071
-    },
-    {
-      "epoch": 0.46424680314238204,
-      "grad_norm": 0.0009445230825804174,
-      "learning_rate": 0.00019999989399251797,
-      "loss": 46.0,
-      "step": 6072
-    },
-    {
-      "epoch": 0.46432326012577174,
-      "grad_norm": 0.0006523270858451724,
-      "learning_rate": 0.00019999989395754066,
-      "loss": 46.0,
-      "step": 6073
-    },
-    {
-      "epoch": 0.46439971710916145,
-      "grad_norm": 0.0019406828796491027,
-      "learning_rate": 0.00019999989392255758,
-      "loss": 46.0,
-      "step": 6074
-    },
-    {
-      "epoch": 0.46447617409255115,
-      "grad_norm": 0.0013628193410113454,
-      "learning_rate": 0.00019999989388756875,
-      "loss": 46.0,
-      "step": 6075
-    },
-    {
-      "epoch": 0.4645526310759409,
-      "grad_norm": 0.00032049877336248755,
-      "learning_rate": 0.00019999989385257415,
-      "loss": 46.0,
-      "step": 6076
-    },
-    {
-      "epoch": 0.4646290880593306,
-      "grad_norm": 0.0016337080160155892,
-      "learning_rate": 0.00019999989381757374,
-      "loss": 46.0,
-      "step": 6077
-    },
-    {
-      "epoch": 0.4647055450427203,
-      "grad_norm": 0.0011174286482855678,
-      "learning_rate": 0.00019999989378256762,
-      "loss": 46.0,
-      "step": 6078
-    },
-    {
-      "epoch": 0.4647820020261101,
-      "grad_norm": 0.0011314835865050554,
-      "learning_rate": 0.00019999989374755567,
-      "loss": 46.0,
-      "step": 6079
-    },
-    {
-      "epoch": 0.4648584590094998,
-      "grad_norm": 0.001032881555147469,
-      "learning_rate": 0.00019999989371253797,
-      "loss": 46.0,
-      "step": 6080
-    },
-    {
-      "epoch": 0.4649349159928895,
-      "grad_norm": 0.001018154202029109,
-      "learning_rate": 0.00019999989367751453,
-      "loss": 46.0,
-      "step": 6081
-    },
-    {
-      "epoch": 0.4650113729762792,
-      "grad_norm": 0.0014510657638311386,
-      "learning_rate": 0.0001999998936424853,
-      "loss": 46.0,
-      "step": 6082
-    },
-    {
-      "epoch": 0.46508782995966896,
-      "grad_norm": 0.0013799158623442054,
-      "learning_rate": 0.00019999989360745032,
-      "loss": 46.0,
-      "step": 6083
-    },
-    {
-      "epoch": 0.46516428694305867,
-      "grad_norm": 0.0014790084678679705,
-      "learning_rate": 0.00019999989357240953,
-      "loss": 46.0,
-      "step": 6084
-    },
-    {
-      "epoch": 0.4652407439264484,
-      "grad_norm": 0.0023474215995520353,
-      "learning_rate": 0.000199999893537363,
-      "loss": 46.0,
-      "step": 6085
-    },
-    {
-      "epoch": 0.4653172009098381,
-      "grad_norm": 0.0020581530407071114,
-      "learning_rate": 0.00019999989350231068,
-      "loss": 46.0,
-      "step": 6086
-    },
-    {
-      "epoch": 0.46539365789322784,
-      "grad_norm": 0.00048755056923255324,
-      "learning_rate": 0.00019999989346725262,
-      "loss": 46.0,
-      "step": 6087
-    },
-    {
-      "epoch": 0.46547011487661755,
-      "grad_norm": 0.0004982934915460646,
-      "learning_rate": 0.00019999989343218877,
-      "loss": 46.0,
-      "step": 6088
-    },
-    {
-      "epoch": 0.46554657186000725,
-      "grad_norm": 0.00028664973797276616,
-      "learning_rate": 0.00019999989339711916,
-      "loss": 46.0,
-      "step": 6089
-    },
-    {
-      "epoch": 0.46562302884339696,
-      "grad_norm": 0.0008702285122126341,
-      "learning_rate": 0.00019999989336204376,
-      "loss": 46.0,
-      "step": 6090
-    },
-    {
-      "epoch": 0.4656994858267867,
-      "grad_norm": 0.0013627681182697415,
-      "learning_rate": 0.00019999989332696264,
-      "loss": 46.0,
-      "step": 6091
-    },
-    {
-      "epoch": 0.4657759428101764,
-      "grad_norm": 0.0028694255743175745,
-      "learning_rate": 0.00019999989329187569,
-      "loss": 46.0,
-      "step": 6092
-    },
-    {
-      "epoch": 0.46585239979356613,
-      "grad_norm": 0.0009175401064567268,
-      "learning_rate": 0.000199999893256783,
-      "loss": 46.0,
-      "step": 6093
-    },
-    {
-      "epoch": 0.4659288567769559,
-      "grad_norm": 0.0004798599402420223,
-      "learning_rate": 0.00019999989322168452,
-      "loss": 46.0,
-      "step": 6094
-    },
-    {
-      "epoch": 0.4660053137603456,
-      "grad_norm": 0.0006042023305781186,
-      "learning_rate": 0.0001999998931865803,
-      "loss": 46.0,
-      "step": 6095
-    },
-    {
-      "epoch": 0.4660817707437353,
-      "grad_norm": 0.0005490007461048663,
-      "learning_rate": 0.00019999989315147028,
-      "loss": 46.0,
-      "step": 6096
-    },
-    {
-      "epoch": 0.466158227727125,
-      "grad_norm": 0.0011405599070712924,
-      "learning_rate": 0.00019999989311635455,
-      "loss": 46.0,
-      "step": 6097
-    },
-    {
-      "epoch": 0.46623468471051477,
-      "grad_norm": 0.005511615425348282,
-      "learning_rate": 0.000199999893081233,
-      "loss": 46.0,
-      "step": 6098
-    },
-    {
-      "epoch": 0.4663111416939045,
-      "grad_norm": 0.0010426411172375083,
-      "learning_rate": 0.0001999998930461057,
-      "loss": 46.0,
-      "step": 6099
-    },
-    {
-      "epoch": 0.4663875986772942,
-      "grad_norm": 0.0018852632492780685,
-      "learning_rate": 0.00019999989301097262,
-      "loss": 46.0,
-      "step": 6100
-    },
-    {
-      "epoch": 0.4664640556606839,
-      "grad_norm": 0.0032607396133244038,
-      "learning_rate": 0.00019999989297583376,
-      "loss": 46.0,
-      "step": 6101
-    },
-    {
-      "epoch": 0.46654051264407365,
-      "grad_norm": 0.001016563386656344,
-      "learning_rate": 0.00019999989294068916,
-      "loss": 46.0,
-      "step": 6102
-    },
-    {
-      "epoch": 0.46661696962746335,
-      "grad_norm": 0.001638552756048739,
-      "learning_rate": 0.00019999989290553875,
-      "loss": 46.0,
-      "step": 6103
-    },
-    {
-      "epoch": 0.46669342661085306,
-      "grad_norm": 0.004358951468020678,
-      "learning_rate": 0.0001999998928703826,
-      "loss": 46.0,
-      "step": 6104
-    },
-    {
-      "epoch": 0.46676988359424276,
-      "grad_norm": 0.0011524200672283769,
-      "learning_rate": 0.00019999989283522068,
-      "loss": 46.0,
-      "step": 6105
-    },
-    {
-      "epoch": 0.4668463405776325,
-      "grad_norm": 0.0021133748814463615,
-      "learning_rate": 0.00019999989280005298,
-      "loss": 46.0,
-      "step": 6106
-    },
-    {
-      "epoch": 0.46692279756102223,
-      "grad_norm": 0.0026021746452897787,
-      "learning_rate": 0.00019999989276487951,
-      "loss": 46.0,
-      "step": 6107
-    },
-    {
-      "epoch": 0.46699925454441193,
-      "grad_norm": 0.0007394233834929764,
-      "learning_rate": 0.00019999989272970027,
-      "loss": 46.0,
-      "step": 6108
-    },
-    {
-      "epoch": 0.4670757115278017,
-      "grad_norm": 0.0026968824677169323,
-      "learning_rate": 0.00019999989269451528,
-      "loss": 46.0,
-      "step": 6109
-    },
-    {
-      "epoch": 0.4671521685111914,
-      "grad_norm": 0.0011821966618299484,
-      "learning_rate": 0.00019999989265932452,
-      "loss": 46.0,
-      "step": 6110
-    },
-    {
-      "epoch": 0.4672286254945811,
-      "grad_norm": 0.00032955125789158046,
-      "learning_rate": 0.00019999989262412795,
-      "loss": 46.0,
-      "step": 6111
-    },
-    {
-      "epoch": 0.4673050824779708,
-      "grad_norm": 0.001247753738425672,
-      "learning_rate": 0.00019999989258892567,
-      "loss": 46.0,
-      "step": 6112
-    },
-    {
-      "epoch": 0.4673815394613606,
-      "grad_norm": 0.00160357600543648,
-      "learning_rate": 0.00019999989255371756,
-      "loss": 46.0,
-      "step": 6113
-    },
-    {
-      "epoch": 0.4674579964447503,
-      "grad_norm": 0.0028064362704753876,
-      "learning_rate": 0.00019999989251850373,
-      "loss": 46.0,
-      "step": 6114
-    },
-    {
-      "epoch": 0.46753445342814,
-      "grad_norm": 0.0007253466756083071,
-      "learning_rate": 0.0001999998924832841,
-      "loss": 46.0,
-      "step": 6115
-    },
-    {
-      "epoch": 0.4676109104115297,
-      "grad_norm": 0.0016152405878528953,
-      "learning_rate": 0.0001999998924480587,
-      "loss": 46.0,
-      "step": 6116
-    },
-    {
-      "epoch": 0.46768736739491945,
-      "grad_norm": 0.0020839269272983074,
-      "learning_rate": 0.00019999989241282755,
-      "loss": 46.0,
-      "step": 6117
-    },
-    {
-      "epoch": 0.46776382437830916,
-      "grad_norm": 0.001287415623664856,
-      "learning_rate": 0.00019999989237759063,
-      "loss": 46.0,
-      "step": 6118
-    },
-    {
-      "epoch": 0.46784028136169886,
-      "grad_norm": 0.0011231377720832825,
-      "learning_rate": 0.00019999989234234793,
-      "loss": 46.0,
-      "step": 6119
-    },
-    {
-      "epoch": 0.46791673834508857,
-      "grad_norm": 0.001098466687835753,
-      "learning_rate": 0.00019999989230709946,
-      "loss": 46.0,
-      "step": 6120
-    },
-    {
-      "epoch": 0.46799319532847833,
-      "grad_norm": 0.0008879080414772034,
-      "learning_rate": 0.00019999989227184522,
-      "loss": 46.0,
-      "step": 6121
-    },
-    {
-      "epoch": 0.46806965231186803,
-      "grad_norm": 0.010542269796133041,
-      "learning_rate": 0.0001999998922365852,
-      "loss": 46.0,
-      "step": 6122
-    },
-    {
-      "epoch": 0.46814610929525774,
-      "grad_norm": 0.00624800892546773,
-      "learning_rate": 0.00019999989220131944,
-      "loss": 46.0,
-      "step": 6123
-    },
-    {
-      "epoch": 0.4682225662786475,
-      "grad_norm": 0.0024787664879113436,
-      "learning_rate": 0.00019999989216604787,
-      "loss": 46.0,
-      "step": 6124
-    },
-    {
-      "epoch": 0.4682990232620372,
-      "grad_norm": 0.001365824369713664,
-      "learning_rate": 0.00019999989213077056,
-      "loss": 46.0,
-      "step": 6125
-    },
-    {
-      "epoch": 0.4683754802454269,
-      "grad_norm": 0.001040915260091424,
-      "learning_rate": 0.00019999989209548748,
-      "loss": 46.0,
-      "step": 6126
-    },
-    {
-      "epoch": 0.4684519372288166,
-      "grad_norm": 0.0014266117941588163,
-      "learning_rate": 0.00019999989206019865,
-      "loss": 46.0,
-      "step": 6127
-    },
-    {
-      "epoch": 0.4685283942122064,
-      "grad_norm": 0.001384437084197998,
-      "learning_rate": 0.00019999989202490402,
-      "loss": 46.0,
-      "step": 6128
-    },
-    {
-      "epoch": 0.4686048511955961,
-      "grad_norm": 0.0007424904033541679,
-      "learning_rate": 0.00019999989198960362,
-      "loss": 46.0,
-      "step": 6129
-    },
-    {
-      "epoch": 0.4686813081789858,
-      "grad_norm": 0.004829507786780596,
-      "learning_rate": 0.00019999989195429744,
-      "loss": 46.0,
-      "step": 6130
-    },
-    {
-      "epoch": 0.4687577651623755,
-      "grad_norm": 0.0012627897085621953,
-      "learning_rate": 0.00019999989191898552,
-      "loss": 46.0,
-      "step": 6131
-    },
-    {
-      "epoch": 0.46883422214576526,
-      "grad_norm": 0.004621132742613554,
-      "learning_rate": 0.00019999989188366782,
-      "loss": 46.0,
-      "step": 6132
-    },
-    {
-      "epoch": 0.46891067912915496,
-      "grad_norm": 0.0010739917634055018,
-      "learning_rate": 0.00019999989184834436,
-      "loss": 46.0,
-      "step": 6133
-    },
-    {
-      "epoch": 0.46898713611254467,
-      "grad_norm": 0.0024414954241365194,
-      "learning_rate": 0.00019999989181301514,
-      "loss": 46.0,
-      "step": 6134
-    },
-    {
-      "epoch": 0.4690635930959344,
-      "grad_norm": 0.002693266374990344,
-      "learning_rate": 0.0001999998917776801,
-      "loss": 46.0,
-      "step": 6135
-    },
-    {
-      "epoch": 0.46914005007932413,
-      "grad_norm": 0.007640698924660683,
-      "learning_rate": 0.00019999989174233933,
-      "loss": 46.0,
-      "step": 6136
-    },
-    {
-      "epoch": 0.46921650706271384,
-      "grad_norm": 0.0008234454435296357,
-      "learning_rate": 0.00019999989170699277,
-      "loss": 46.0,
-      "step": 6137
-    },
-    {
-      "epoch": 0.46929296404610354,
-      "grad_norm": 0.0005543379229493439,
-      "learning_rate": 0.00019999989167164046,
-      "loss": 46.0,
-      "step": 6138
-    },
-    {
-      "epoch": 0.4693694210294933,
-      "grad_norm": 0.0006760939722880721,
-      "learning_rate": 0.00019999989163628238,
-      "loss": 46.0,
-      "step": 6139
-    },
-    {
-      "epoch": 0.469445878012883,
-      "grad_norm": 0.0007719618151895702,
-      "learning_rate": 0.00019999989160091852,
-      "loss": 46.0,
-      "step": 6140
-    },
-    {
-      "epoch": 0.4695223349962727,
-      "grad_norm": 0.002284866524860263,
-      "learning_rate": 0.0001999998915655489,
-      "loss": 46.0,
-      "step": 6141
-    },
-    {
-      "epoch": 0.4695987919796624,
-      "grad_norm": 0.00119221699424088,
-      "learning_rate": 0.00019999989153017352,
-      "loss": 46.0,
-      "step": 6142
-    },
-    {
-      "epoch": 0.4696752489630522,
-      "grad_norm": 0.0014594650128856301,
-      "learning_rate": 0.00019999989149479234,
-      "loss": 46.0,
-      "step": 6143
-    },
-    {
-      "epoch": 0.4697517059464419,
-      "grad_norm": 0.0007182607660070062,
-      "learning_rate": 0.00019999989145940542,
-      "loss": 46.0,
-      "step": 6144
-    },
-    {
-      "epoch": 0.4698281629298316,
-      "grad_norm": 0.008152276277542114,
-      "learning_rate": 0.0001999998914240127,
-      "loss": 46.0,
-      "step": 6145
-    },
-    {
-      "epoch": 0.4699046199132213,
-      "grad_norm": 0.0008363769738934934,
-      "learning_rate": 0.0001999998913886142,
-      "loss": 46.0,
-      "step": 6146
-    },
-    {
-      "epoch": 0.46998107689661106,
-      "grad_norm": 0.0019262407440692186,
-      "learning_rate": 0.00019999989135320999,
-      "loss": 46.0,
-      "step": 6147
-    },
-    {
-      "epoch": 0.47005753388000077,
-      "grad_norm": 0.0030531391967087984,
-      "learning_rate": 0.00019999989131779997,
-      "loss": 46.0,
-      "step": 6148
-    },
-    {
-      "epoch": 0.47013399086339047,
-      "grad_norm": 0.0013678909745067358,
-      "learning_rate": 0.00019999989128238418,
-      "loss": 46.0,
-      "step": 6149
-    },
-    {
-      "epoch": 0.47021044784678023,
-      "grad_norm": 0.0008489760803058743,
-      "learning_rate": 0.00019999989124696265,
-      "loss": 46.0,
-      "step": 6150
-    },
-    {
-      "epoch": 0.47028690483016994,
-      "grad_norm": 0.0008520630653947592,
-      "learning_rate": 0.00019999989121153534,
-      "loss": 46.0,
-      "step": 6151
-    },
-    {
-      "epoch": 0.47036336181355964,
-      "grad_norm": 0.0008328371914103627,
-      "learning_rate": 0.00019999989117610223,
-      "loss": 46.0,
-      "step": 6152
-    },
-    {
-      "epoch": 0.47043981879694935,
-      "grad_norm": 0.0006129646208137274,
-      "learning_rate": 0.00019999989114066334,
-      "loss": 46.0,
-      "step": 6153
-    },
-    {
-      "epoch": 0.4705162757803391,
-      "grad_norm": 0.003923851530998945,
-      "learning_rate": 0.00019999989110521872,
-      "loss": 46.0,
-      "step": 6154
-    },
-    {
-      "epoch": 0.4705927327637288,
-      "grad_norm": 0.0020324308425188065,
-      "learning_rate": 0.00019999989106976834,
-      "loss": 46.0,
-      "step": 6155
-    },
-    {
-      "epoch": 0.4706691897471185,
-      "grad_norm": 0.004159628413617611,
-      "learning_rate": 0.00019999989103431216,
-      "loss": 46.0,
-      "step": 6156
-    },
-    {
-      "epoch": 0.4707456467305082,
-      "grad_norm": 0.0014560986310243607,
-      "learning_rate": 0.00019999989099885024,
-      "loss": 46.0,
-      "step": 6157
-    },
-    {
-      "epoch": 0.470822103713898,
-      "grad_norm": 0.002111112466081977,
-      "learning_rate": 0.00019999989096338255,
-      "loss": 46.0,
-      "step": 6158
-    },
-    {
-      "epoch": 0.4708985606972877,
-      "grad_norm": 0.013774135150015354,
-      "learning_rate": 0.00019999989092790905,
-      "loss": 46.0,
-      "step": 6159
-    },
-    {
-      "epoch": 0.4709750176806774,
-      "grad_norm": 0.0010287988698109984,
-      "learning_rate": 0.0001999998908924298,
-      "loss": 46.0,
-      "step": 6160
-    },
-    {
-      "epoch": 0.4710514746640671,
-      "grad_norm": 0.002148106461390853,
-      "learning_rate": 0.0001999998908569448,
-      "loss": 46.0,
-      "step": 6161
-    },
-    {
-      "epoch": 0.47112793164745687,
-      "grad_norm": 0.0025398489087820053,
-      "learning_rate": 0.000199999890821454,
-      "loss": 46.0,
-      "step": 6162
-    },
-    {
-      "epoch": 0.47120438863084657,
-      "grad_norm": 0.0011332968715578318,
-      "learning_rate": 0.00019999989078595744,
-      "loss": 46.0,
-      "step": 6163
-    },
-    {
-      "epoch": 0.4712808456142363,
-      "grad_norm": 0.001397020765580237,
-      "learning_rate": 0.00019999989075045513,
-      "loss": 46.0,
-      "step": 6164
-    },
-    {
-      "epoch": 0.47135730259762604,
-      "grad_norm": 0.0066262464970350266,
-      "learning_rate": 0.00019999989071494705,
-      "loss": 46.0,
-      "step": 6165
-    },
-    {
-      "epoch": 0.47143375958101574,
-      "grad_norm": 0.0009376137750223279,
-      "learning_rate": 0.0001999998906794332,
-      "loss": 46.0,
-      "step": 6166
-    },
-    {
-      "epoch": 0.47151021656440545,
-      "grad_norm": 0.004165223799645901,
-      "learning_rate": 0.00019999989064391357,
-      "loss": 46.0,
-      "step": 6167
-    },
-    {
-      "epoch": 0.47158667354779515,
-      "grad_norm": 0.00034537186729721725,
-      "learning_rate": 0.00019999989060838814,
-      "loss": 46.0,
-      "step": 6168
-    },
-    {
-      "epoch": 0.4716631305311849,
-      "grad_norm": 0.0021305617410689592,
-      "learning_rate": 0.00019999989057285696,
-      "loss": 46.0,
-      "step": 6169
-    },
-    {
-      "epoch": 0.4717395875145746,
-      "grad_norm": 0.0006563275237567723,
-      "learning_rate": 0.00019999989053732001,
-      "loss": 46.0,
-      "step": 6170
-    },
-    {
-      "epoch": 0.4718160444979643,
-      "grad_norm": 0.0031731147319078445,
-      "learning_rate": 0.00019999989050177732,
-      "loss": 46.0,
-      "step": 6171
-    },
-    {
-      "epoch": 0.47189250148135403,
-      "grad_norm": 0.003867693245410919,
-      "learning_rate": 0.00019999989046622885,
-      "loss": 46.0,
-      "step": 6172
-    },
-    {
-      "epoch": 0.4719689584647438,
-      "grad_norm": 0.0010971968295052648,
-      "learning_rate": 0.0001999998904306746,
-      "loss": 46.0,
-      "step": 6173
-    },
-    {
-      "epoch": 0.4720454154481335,
-      "grad_norm": 0.0015497637214139104,
-      "learning_rate": 0.00019999989039511457,
-      "loss": 46.0,
-      "step": 6174
-    },
-    {
-      "epoch": 0.4721218724315232,
-      "grad_norm": 0.0017124320147559047,
-      "learning_rate": 0.0001999998903595488,
-      "loss": 46.0,
-      "step": 6175
-    },
-    {
-      "epoch": 0.4721983294149129,
-      "grad_norm": 0.000605948269367218,
-      "learning_rate": 0.00019999989032397724,
-      "loss": 46.0,
-      "step": 6176
-    },
-    {
-      "epoch": 0.47227478639830267,
-      "grad_norm": 0.0018868144834414124,
-      "learning_rate": 0.00019999989028839988,
-      "loss": 46.0,
-      "step": 6177
-    },
-    {
-      "epoch": 0.4723512433816924,
-      "grad_norm": 0.0013617558870464563,
-      "learning_rate": 0.0001999998902528168,
-      "loss": 46.0,
-      "step": 6178
-    },
-    {
-      "epoch": 0.4724277003650821,
-      "grad_norm": 0.0007929742569103837,
-      "learning_rate": 0.00019999989021722794,
-      "loss": 46.0,
-      "step": 6179
-    },
-    {
-      "epoch": 0.47250415734847184,
-      "grad_norm": 0.002169110579416156,
-      "learning_rate": 0.0001999998901816333,
-      "loss": 46.0,
-      "step": 6180
-    },
-    {
-      "epoch": 0.47258061433186155,
-      "grad_norm": 0.0012314191553741693,
-      "learning_rate": 0.00019999989014603291,
-      "loss": 46.0,
-      "step": 6181
-    },
-    {
-      "epoch": 0.47265707131525125,
-      "grad_norm": 0.0012225055834278464,
-      "learning_rate": 0.00019999989011042674,
-      "loss": 46.0,
-      "step": 6182
-    },
-    {
-      "epoch": 0.47273352829864096,
-      "grad_norm": 0.0028150416910648346,
-      "learning_rate": 0.0001999998900748148,
-      "loss": 46.0,
-      "step": 6183
-    },
-    {
-      "epoch": 0.4728099852820307,
-      "grad_norm": 0.0030481473077088594,
-      "learning_rate": 0.00019999989003919707,
-      "loss": 46.0,
-      "step": 6184
-    },
-    {
-      "epoch": 0.4728864422654204,
-      "grad_norm": 0.0008216250571422279,
-      "learning_rate": 0.0001999998900035736,
-      "loss": 46.0,
-      "step": 6185
-    },
-    {
-      "epoch": 0.47296289924881013,
-      "grad_norm": 0.0013899178011342883,
-      "learning_rate": 0.00019999988996794433,
-      "loss": 46.0,
-      "step": 6186
-    },
-    {
-      "epoch": 0.47303935623219984,
-      "grad_norm": 0.0044166324660182,
-      "learning_rate": 0.00019999988993230932,
-      "loss": 46.0,
-      "step": 6187
-    },
-    {
-      "epoch": 0.4731158132155896,
-      "grad_norm": 0.001612075255252421,
-      "learning_rate": 0.00019999988989666853,
-      "loss": 46.0,
-      "step": 6188
-    },
-    {
-      "epoch": 0.4731922701989793,
-      "grad_norm": 0.001719577587209642,
-      "learning_rate": 0.000199999889861022,
-      "loss": 46.0,
-      "step": 6189
-    },
-    {
-      "epoch": 0.473268727182369,
-      "grad_norm": 0.0013245035661384463,
-      "learning_rate": 0.00019999988982536966,
-      "loss": 46.0,
-      "step": 6190
-    },
-    {
-      "epoch": 0.4733451841657587,
-      "grad_norm": 0.0010736206313595176,
-      "learning_rate": 0.00019999988978971155,
-      "loss": 46.0,
-      "step": 6191
-    },
-    {
-      "epoch": 0.4734216411491485,
-      "grad_norm": 0.002807576907798648,
-      "learning_rate": 0.0001999998897540477,
-      "loss": 46.0,
-      "step": 6192
-    },
-    {
-      "epoch": 0.4734980981325382,
-      "grad_norm": 0.002152242697775364,
-      "learning_rate": 0.00019999988971837804,
-      "loss": 46.0,
-      "step": 6193
-    },
-    {
-      "epoch": 0.4735745551159279,
-      "grad_norm": 0.000758540176320821,
-      "learning_rate": 0.0001999998896827026,
-      "loss": 46.0,
-      "step": 6194
-    },
-    {
-      "epoch": 0.47365101209931765,
-      "grad_norm": 0.0031296033412218094,
-      "learning_rate": 0.00019999988964702146,
-      "loss": 46.0,
-      "step": 6195
-    },
-    {
-      "epoch": 0.47372746908270735,
-      "grad_norm": 0.0005724740331061184,
-      "learning_rate": 0.00019999988961133452,
-      "loss": 46.0,
-      "step": 6196
-    },
-    {
-      "epoch": 0.47380392606609706,
-      "grad_norm": 0.0023585292510688305,
-      "learning_rate": 0.0001999998895756418,
-      "loss": 46.0,
-      "step": 6197
-    },
-    {
-      "epoch": 0.47388038304948676,
-      "grad_norm": 0.0009502078755758703,
-      "learning_rate": 0.0001999998895399433,
-      "loss": 46.0,
-      "step": 6198
-    },
-    {
-      "epoch": 0.4739568400328765,
-      "grad_norm": 0.0016020425828173757,
-      "learning_rate": 0.00019999988950423906,
-      "loss": 46.0,
-      "step": 6199
-    },
-    {
-      "epoch": 0.47403329701626623,
-      "grad_norm": 0.000578279432374984,
-      "learning_rate": 0.00019999988946852904,
-      "loss": 46.0,
-      "step": 6200
-    },
-    {
-      "epoch": 0.47410975399965594,
-      "grad_norm": 0.001709281001240015,
-      "learning_rate": 0.00019999988943281326,
-      "loss": 46.0,
-      "step": 6201
-    },
-    {
-      "epoch": 0.47418621098304564,
-      "grad_norm": 0.004529502708464861,
-      "learning_rate": 0.0001999998893970917,
-      "loss": 46.0,
-      "step": 6202
-    },
-    {
-      "epoch": 0.4742626679664354,
-      "grad_norm": 0.0026438753120601177,
-      "learning_rate": 0.00019999988936136436,
-      "loss": 46.0,
-      "step": 6203
-    },
-    {
-      "epoch": 0.4743391249498251,
-      "grad_norm": 0.002564698690548539,
-      "learning_rate": 0.00019999988932563125,
-      "loss": 46.0,
-      "step": 6204
-    },
-    {
-      "epoch": 0.4744155819332148,
-      "grad_norm": 0.0011640641605481505,
-      "learning_rate": 0.0001999998892898924,
-      "loss": 46.0,
-      "step": 6205
-    },
-    {
-      "epoch": 0.4744920389166045,
-      "grad_norm": 0.01073073036968708,
-      "learning_rate": 0.00019999988925414774,
-      "loss": 46.0,
-      "step": 6206
-    },
-    {
-      "epoch": 0.4745684958999943,
-      "grad_norm": 0.0007749986834824085,
-      "learning_rate": 0.00019999988921839731,
-      "loss": 46.0,
-      "step": 6207
-    },
-    {
-      "epoch": 0.474644952883384,
-      "grad_norm": 0.0020734097342938185,
-      "learning_rate": 0.00019999988918264117,
-      "loss": 46.0,
-      "step": 6208
-    },
-    {
-      "epoch": 0.4747214098667737,
-      "grad_norm": 0.002942918334156275,
-      "learning_rate": 0.0001999998891468792,
-      "loss": 46.0,
-      "step": 6209
-    },
-    {
-      "epoch": 0.47479786685016345,
-      "grad_norm": 0.001187040819786489,
-      "learning_rate": 0.0001999998891111115,
-      "loss": 46.0,
-      "step": 6210
-    },
-    {
-      "epoch": 0.47487432383355316,
-      "grad_norm": 0.000389922788599506,
-      "learning_rate": 0.000199999889075338,
-      "loss": 46.0,
-      "step": 6211
-    },
-    {
-      "epoch": 0.47495078081694286,
-      "grad_norm": 0.0026759100146591663,
-      "learning_rate": 0.00019999988903955873,
-      "loss": 46.0,
-      "step": 6212
-    },
-    {
-      "epoch": 0.47502723780033257,
-      "grad_norm": 0.0015849982155486941,
-      "learning_rate": 0.00019999988900377372,
-      "loss": 46.0,
-      "step": 6213
-    },
-    {
-      "epoch": 0.47510369478372233,
-      "grad_norm": 0.0010976759949699044,
-      "learning_rate": 0.0001999998889679829,
-      "loss": 46.0,
-      "step": 6214
-    },
-    {
-      "epoch": 0.47518015176711204,
-      "grad_norm": 0.0004689736815635115,
-      "learning_rate": 0.00019999988893218634,
-      "loss": 46.0,
-      "step": 6215
-    },
-    {
-      "epoch": 0.47525660875050174,
-      "grad_norm": 0.000623491476289928,
-      "learning_rate": 0.000199999888896384,
-      "loss": 46.0,
-      "step": 6216
-    },
-    {
-      "epoch": 0.47533306573389145,
-      "grad_norm": 0.006456580478698015,
-      "learning_rate": 0.0001999998888605759,
-      "loss": 46.0,
-      "step": 6217
-    },
-    {
-      "epoch": 0.4754095227172812,
-      "grad_norm": 0.0010884033981710672,
-      "learning_rate": 0.00019999988882476205,
-      "loss": 46.0,
-      "step": 6218
-    },
-    {
-      "epoch": 0.4754859797006709,
-      "grad_norm": 0.0007241466082632542,
-      "learning_rate": 0.0001999998887889424,
-      "loss": 46.0,
-      "step": 6219
-    },
-    {
-      "epoch": 0.4755624366840606,
-      "grad_norm": 0.0008789958083070815,
-      "learning_rate": 0.000199999888753117,
-      "loss": 46.0,
-      "step": 6220
-    },
-    {
-      "epoch": 0.4756388936674503,
-      "grad_norm": 0.003025518264621496,
-      "learning_rate": 0.00019999988871728582,
-      "loss": 46.0,
-      "step": 6221
-    },
-    {
-      "epoch": 0.4757153506508401,
-      "grad_norm": 0.003622946795076132,
-      "learning_rate": 0.00019999988868144884,
-      "loss": 46.0,
-      "step": 6222
-    },
-    {
-      "epoch": 0.4757918076342298,
-      "grad_norm": 0.0004595489881467074,
-      "learning_rate": 0.00019999988864560612,
-      "loss": 46.0,
-      "step": 6223
-    },
-    {
-      "epoch": 0.4758682646176195,
-      "grad_norm": 0.001185062574222684,
-      "learning_rate": 0.00019999988860975766,
-      "loss": 46.0,
-      "step": 6224
-    },
-    {
-      "epoch": 0.47594472160100926,
-      "grad_norm": 0.0014367899857461452,
-      "learning_rate": 0.0001999998885739034,
-      "loss": 46.0,
-      "step": 6225
-    },
-    {
-      "epoch": 0.47602117858439896,
-      "grad_norm": 0.003645437303930521,
-      "learning_rate": 0.00019999988853804337,
-      "loss": 46.0,
-      "step": 6226
-    },
-    {
-      "epoch": 0.47609763556778867,
-      "grad_norm": 0.0038288708310574293,
-      "learning_rate": 0.00019999988850217756,
-      "loss": 46.0,
-      "step": 6227
-    },
-    {
-      "epoch": 0.4761740925511784,
-      "grad_norm": 0.0010920005152001977,
-      "learning_rate": 0.000199999888466306,
-      "loss": 46.0,
-      "step": 6228
-    },
-    {
-      "epoch": 0.47625054953456814,
-      "grad_norm": 0.0007951435400173068,
-      "learning_rate": 0.00019999988843042864,
-      "loss": 46.0,
-      "step": 6229
-    },
-    {
-      "epoch": 0.47632700651795784,
-      "grad_norm": 0.0006535447319038212,
-      "learning_rate": 0.00019999988839454553,
-      "loss": 46.0,
-      "step": 6230
-    },
-    {
-      "epoch": 0.47640346350134755,
-      "grad_norm": 0.0011695997091010213,
-      "learning_rate": 0.00019999988835865668,
-      "loss": 46.0,
-      "step": 6231
-    },
-    {
-      "epoch": 0.47647992048473725,
-      "grad_norm": 0.0009866493055596948,
-      "learning_rate": 0.00019999988832276202,
-      "loss": 46.0,
-      "step": 6232
-    },
-    {
-      "epoch": 0.476556377468127,
-      "grad_norm": 0.0006724954000674188,
-      "learning_rate": 0.0001999998882868616,
-      "loss": 46.0,
-      "step": 6233
-    },
-    {
-      "epoch": 0.4766328344515167,
-      "grad_norm": 0.0007678009569644928,
-      "learning_rate": 0.00019999988825095542,
-      "loss": 46.0,
-      "step": 6234
-    },
-    {
-      "epoch": 0.4767092914349064,
-      "grad_norm": 0.0009465903276577592,
-      "learning_rate": 0.00019999988821504347,
-      "loss": 46.0,
-      "step": 6235
-    },
-    {
-      "epoch": 0.47678574841829613,
-      "grad_norm": 0.0009502816246822476,
-      "learning_rate": 0.00019999988817912575,
-      "loss": 46.0,
-      "step": 6236
-    },
-    {
-      "epoch": 0.4768622054016859,
-      "grad_norm": 0.0016623563133180141,
-      "learning_rate": 0.00019999988814320226,
-      "loss": 46.0,
-      "step": 6237
-    },
-    {
-      "epoch": 0.4769386623850756,
-      "grad_norm": 0.0018453830853104591,
-      "learning_rate": 0.000199999888107273,
-      "loss": 46.0,
-      "step": 6238
-    },
-    {
-      "epoch": 0.4770151193684653,
-      "grad_norm": 0.0009387017344124615,
-      "learning_rate": 0.00019999988807133795,
-      "loss": 46.0,
-      "step": 6239
-    },
-    {
-      "epoch": 0.47709157635185506,
-      "grad_norm": 0.000980983255431056,
-      "learning_rate": 0.00019999988803539716,
-      "loss": 46.0,
-      "step": 6240
-    },
-    {
-      "epoch": 0.47716803333524477,
-      "grad_norm": 0.0016814226983115077,
-      "learning_rate": 0.0001999998879994506,
-      "loss": 46.0,
-      "step": 6241
-    },
-    {
-      "epoch": 0.4772444903186345,
-      "grad_norm": 0.0006290982710197568,
-      "learning_rate": 0.00019999988796349824,
-      "loss": 46.0,
-      "step": 6242
-    },
-    {
-      "epoch": 0.4773209473020242,
-      "grad_norm": 0.0026615469250828028,
-      "learning_rate": 0.0001999998879275401,
-      "loss": 46.0,
-      "step": 6243
-    },
-    {
-      "epoch": 0.47739740428541394,
-      "grad_norm": 0.0010419313330203295,
-      "learning_rate": 0.00019999988789157626,
-      "loss": 46.0,
-      "step": 6244
-    },
-    {
-      "epoch": 0.47747386126880365,
-      "grad_norm": 0.0010810858802869916,
-      "learning_rate": 0.0001999998878556066,
-      "loss": 46.0,
-      "step": 6245
-    },
-    {
-      "epoch": 0.47755031825219335,
-      "grad_norm": 0.0027360483072698116,
-      "learning_rate": 0.0001999998878196312,
-      "loss": 46.0,
-      "step": 6246
-    },
-    {
-      "epoch": 0.47762677523558306,
-      "grad_norm": 0.011292408220469952,
-      "learning_rate": 0.00019999988778365003,
-      "loss": 46.0,
-      "step": 6247
-    },
-    {
-      "epoch": 0.4777032322189728,
-      "grad_norm": 0.0015824967995285988,
-      "learning_rate": 0.00019999988774766305,
-      "loss": 46.0,
-      "step": 6248
-    },
-    {
-      "epoch": 0.4777796892023625,
-      "grad_norm": 0.0009166456875391304,
-      "learning_rate": 0.0001999998877116703,
-      "loss": 46.0,
-      "step": 6249
-    },
-    {
-      "epoch": 0.47785614618575223,
-      "grad_norm": 0.000988941639661789,
-      "learning_rate": 0.0001999998876756718,
-      "loss": 46.0,
-      "step": 6250
-    },
-    {
-      "epoch": 0.47793260316914193,
-      "grad_norm": 0.0010834858985617757,
-      "learning_rate": 0.00019999988763966755,
-      "loss": 46.0,
-      "step": 6251
-    },
-    {
-      "epoch": 0.4780090601525317,
-      "grad_norm": 0.004990748595446348,
-      "learning_rate": 0.0001999998876036575,
-      "loss": 46.0,
-      "step": 6252
-    },
-    {
-      "epoch": 0.4780855171359214,
-      "grad_norm": 0.0009165873634628952,
-      "learning_rate": 0.00019999988756764172,
-      "loss": 46.0,
-      "step": 6253
-    },
-    {
-      "epoch": 0.4781619741193111,
-      "grad_norm": 0.0022636260837316513,
-      "learning_rate": 0.00019999988753162016,
-      "loss": 46.0,
-      "step": 6254
-    },
-    {
-      "epoch": 0.47823843110270087,
-      "grad_norm": 0.003004265483468771,
-      "learning_rate": 0.0001999998874955928,
-      "loss": 46.0,
-      "step": 6255
-    },
-    {
-      "epoch": 0.4783148880860906,
-      "grad_norm": 0.002727489685639739,
-      "learning_rate": 0.0001999998874595597,
-      "loss": 46.0,
-      "step": 6256
-    },
-    {
-      "epoch": 0.4783913450694803,
-      "grad_norm": 0.004355646204203367,
-      "learning_rate": 0.0001999998874235208,
-      "loss": 46.0,
-      "step": 6257
-    },
-    {
-      "epoch": 0.47846780205287,
-      "grad_norm": 0.0008424715488217771,
-      "learning_rate": 0.00019999988738747616,
-      "loss": 46.0,
-      "step": 6258
-    },
-    {
-      "epoch": 0.47854425903625974,
-      "grad_norm": 0.0011882291873916984,
-      "learning_rate": 0.00019999988735142573,
-      "loss": 46.0,
-      "step": 6259
-    },
-    {
-      "epoch": 0.47862071601964945,
-      "grad_norm": 0.004274935927242041,
-      "learning_rate": 0.00019999988731536953,
-      "loss": 46.0,
-      "step": 6260
-    },
-    {
-      "epoch": 0.47869717300303916,
-      "grad_norm": 0.0005319232004694641,
-      "learning_rate": 0.00019999988727930758,
-      "loss": 46.0,
-      "step": 6261
-    },
-    {
-      "epoch": 0.47877362998642886,
-      "grad_norm": 0.0012364584254100919,
-      "learning_rate": 0.00019999988724323986,
-      "loss": 46.0,
-      "step": 6262
-    },
-    {
-      "epoch": 0.4788500869698186,
-      "grad_norm": 0.0007650853949598968,
-      "learning_rate": 0.00019999988720716637,
-      "loss": 46.0,
-      "step": 6263
-    },
-    {
-      "epoch": 0.47892654395320833,
-      "grad_norm": 0.0005413478938862681,
-      "learning_rate": 0.00019999988717108708,
-      "loss": 46.0,
-      "step": 6264
-    },
-    {
-      "epoch": 0.47900300093659803,
-      "grad_norm": 0.0005395942134782672,
-      "learning_rate": 0.00019999988713500204,
-      "loss": 46.0,
-      "step": 6265
-    },
-    {
-      "epoch": 0.4790794579199878,
-      "grad_norm": 0.006407381501048803,
-      "learning_rate": 0.00019999988709891125,
-      "loss": 46.0,
-      "step": 6266
-    },
-    {
-      "epoch": 0.4791559149033775,
-      "grad_norm": 0.003435243386775255,
-      "learning_rate": 0.00019999988706281466,
-      "loss": 46.0,
-      "step": 6267
-    },
-    {
-      "epoch": 0.4792323718867672,
-      "grad_norm": 0.0004913299344480038,
-      "learning_rate": 0.00019999988702671233,
-      "loss": 46.0,
-      "step": 6268
-    },
-    {
-      "epoch": 0.4793088288701569,
-      "grad_norm": 0.0004139500961173326,
-      "learning_rate": 0.00019999988699060422,
-      "loss": 46.0,
-      "step": 6269
-    },
-    {
-      "epoch": 0.47938528585354667,
-      "grad_norm": 0.0019083141814917326,
-      "learning_rate": 0.00019999988695449032,
-      "loss": 46.0,
-      "step": 6270
-    },
-    {
-      "epoch": 0.4794617428369364,
-      "grad_norm": 0.002150079468265176,
-      "learning_rate": 0.00019999988691837066,
-      "loss": 46.0,
-      "step": 6271
-    },
-    {
-      "epoch": 0.4795381998203261,
-      "grad_norm": 0.0035702658351510763,
-      "learning_rate": 0.00019999988688224524,
-      "loss": 46.0,
-      "step": 6272
-    },
-    {
-      "epoch": 0.4796146568037158,
-      "grad_norm": 0.001005249097943306,
-      "learning_rate": 0.00019999988684611406,
-      "loss": 46.0,
-      "step": 6273
-    },
-    {
-      "epoch": 0.47969111378710555,
-      "grad_norm": 0.007846593856811523,
-      "learning_rate": 0.0001999998868099771,
-      "loss": 46.0,
-      "step": 6274
-    },
-    {
-      "epoch": 0.47976757077049526,
-      "grad_norm": 0.0034158381167799234,
-      "learning_rate": 0.00019999988677383437,
-      "loss": 46.0,
-      "step": 6275
-    },
-    {
-      "epoch": 0.47984402775388496,
-      "grad_norm": 0.0012672198936343193,
-      "learning_rate": 0.00019999988673768585,
-      "loss": 46.0,
-      "step": 6276
-    },
-    {
-      "epoch": 0.47992048473727467,
-      "grad_norm": 0.0008209488005377352,
-      "learning_rate": 0.00019999988670153158,
-      "loss": 46.0,
-      "step": 6277
-    },
-    {
-      "epoch": 0.4799969417206644,
-      "grad_norm": 0.00043715073843486607,
-      "learning_rate": 0.00019999988666537154,
-      "loss": 46.0,
-      "step": 6278
-    },
-    {
-      "epoch": 0.48007339870405413,
-      "grad_norm": 0.0008419574005529284,
-      "learning_rate": 0.00019999988662920576,
-      "loss": 46.0,
-      "step": 6279
-    },
-    {
-      "epoch": 0.48014985568744384,
-      "grad_norm": 0.0011857226490974426,
-      "learning_rate": 0.00019999988659303414,
-      "loss": 46.0,
-      "step": 6280
-    },
-    {
-      "epoch": 0.4802263126708336,
-      "grad_norm": 0.0006404307205229998,
-      "learning_rate": 0.0001999998865568568,
-      "loss": 46.0,
-      "step": 6281
-    },
-    {
-      "epoch": 0.4803027696542233,
-      "grad_norm": 0.003302835626527667,
-      "learning_rate": 0.0001999998865206737,
-      "loss": 46.0,
-      "step": 6282
-    },
-    {
-      "epoch": 0.480379226637613,
-      "grad_norm": 0.01397534180432558,
-      "learning_rate": 0.0001999998864844848,
-      "loss": 46.0,
-      "step": 6283
-    },
-    {
-      "epoch": 0.4804556836210027,
-      "grad_norm": 0.0009030290530063212,
-      "learning_rate": 0.00019999988644829015,
-      "loss": 46.0,
-      "step": 6284
-    },
-    {
-      "epoch": 0.4805321406043925,
-      "grad_norm": 0.0011602970771491528,
-      "learning_rate": 0.00019999988641208972,
-      "loss": 46.0,
-      "step": 6285
-    },
-    {
-      "epoch": 0.4806085975877822,
-      "grad_norm": 0.00275371759198606,
-      "learning_rate": 0.00019999988637588355,
-      "loss": 46.0,
-      "step": 6286
-    },
-    {
-      "epoch": 0.4806850545711719,
-      "grad_norm": 0.001148720970377326,
-      "learning_rate": 0.00019999988633967157,
-      "loss": 46.0,
-      "step": 6287
-    },
-    {
-      "epoch": 0.4807615115545616,
-      "grad_norm": 0.0008375155739486217,
-      "learning_rate": 0.00019999988630345383,
-      "loss": 46.0,
-      "step": 6288
-    },
-    {
-      "epoch": 0.48083796853795135,
-      "grad_norm": 0.0009214160381816328,
-      "learning_rate": 0.0001999998862672303,
-      "loss": 46.0,
-      "step": 6289
-    },
-    {
-      "epoch": 0.48091442552134106,
-      "grad_norm": 0.0025782783050090075,
-      "learning_rate": 0.00019999988623100104,
-      "loss": 46.0,
-      "step": 6290
-    },
-    {
-      "epoch": 0.48099088250473077,
-      "grad_norm": 0.0027072604279965162,
-      "learning_rate": 0.00019999988619476603,
-      "loss": 46.0,
-      "step": 6291
-    },
-    {
-      "epoch": 0.48106733948812047,
-      "grad_norm": 0.005264668725430965,
-      "learning_rate": 0.0001999998861585252,
-      "loss": 46.0,
-      "step": 6292
-    },
-    {
-      "epoch": 0.48114379647151023,
-      "grad_norm": 0.0004953782190568745,
-      "learning_rate": 0.00019999988612227863,
-      "loss": 46.0,
-      "step": 6293
-    },
-    {
-      "epoch": 0.48122025345489994,
-      "grad_norm": 0.000940196739975363,
-      "learning_rate": 0.00019999988608602624,
-      "loss": 46.0,
-      "step": 6294
-    },
-    {
-      "epoch": 0.48129671043828964,
-      "grad_norm": 0.00042951980140060186,
-      "learning_rate": 0.00019999988604976814,
-      "loss": 46.0,
-      "step": 6295
-    },
-    {
-      "epoch": 0.4813731674216794,
-      "grad_norm": 0.0010089338757097721,
-      "learning_rate": 0.00019999988601350426,
-      "loss": 46.0,
-      "step": 6296
-    },
-    {
-      "epoch": 0.4814496244050691,
-      "grad_norm": 0.00034610298462212086,
-      "learning_rate": 0.00019999988597723458,
-      "loss": 46.0,
-      "step": 6297
-    },
-    {
-      "epoch": 0.4815260813884588,
-      "grad_norm": 0.0014529270119965076,
-      "learning_rate": 0.00019999988594095915,
-      "loss": 46.0,
-      "step": 6298
-    },
-    {
-      "epoch": 0.4816025383718485,
-      "grad_norm": 0.041110314428806305,
-      "learning_rate": 0.00019999988590467795,
-      "loss": 46.0,
-      "step": 6299
-    },
-    {
-      "epoch": 0.4816789953552383,
-      "grad_norm": 0.0025845463387668133,
-      "learning_rate": 0.000199999885868391,
-      "loss": 46.0,
-      "step": 6300
-    },
-    {
-      "epoch": 0.481755452338628,
-      "grad_norm": 0.0038914778269827366,
-      "learning_rate": 0.00019999988583209826,
-      "loss": 46.0,
-      "step": 6301
-    },
-    {
-      "epoch": 0.4818319093220177,
-      "grad_norm": 0.0010950975120067596,
-      "learning_rate": 0.00019999988579579974,
-      "loss": 46.0,
-      "step": 6302
-    },
-    {
-      "epoch": 0.4819083663054074,
-      "grad_norm": 0.001826742198318243,
-      "learning_rate": 0.00019999988575949548,
-      "loss": 46.0,
-      "step": 6303
-    },
-    {
-      "epoch": 0.48198482328879716,
-      "grad_norm": 0.0023944347631186247,
-      "learning_rate": 0.0001999998857231854,
-      "loss": 46.0,
-      "step": 6304
-    },
-    {
-      "epoch": 0.48206128027218687,
-      "grad_norm": 0.0033000996336340904,
-      "learning_rate": 0.0001999998856868696,
-      "loss": 46.0,
-      "step": 6305
-    },
-    {
-      "epoch": 0.48213773725557657,
-      "grad_norm": 0.0012456781696528196,
-      "learning_rate": 0.000199999885650548,
-      "loss": 46.0,
-      "step": 6306
-    },
-    {
-      "epoch": 0.4822141942389663,
-      "grad_norm": 0.001440512714907527,
-      "learning_rate": 0.00019999988561422068,
-      "loss": 46.0,
-      "step": 6307
-    },
-    {
-      "epoch": 0.48229065122235604,
-      "grad_norm": 0.0023240000009536743,
-      "learning_rate": 0.00019999988557788755,
-      "loss": 46.0,
-      "step": 6308
-    },
-    {
-      "epoch": 0.48236710820574574,
-      "grad_norm": 0.002259685890749097,
-      "learning_rate": 0.00019999988554154864,
-      "loss": 46.0,
-      "step": 6309
-    },
-    {
-      "epoch": 0.48244356518913545,
-      "grad_norm": 0.0012466987827792764,
-      "learning_rate": 0.000199999885505204,
-      "loss": 46.0,
-      "step": 6310
-    },
-    {
-      "epoch": 0.4825200221725252,
-      "grad_norm": 0.0006265821284614503,
-      "learning_rate": 0.00019999988546885356,
-      "loss": 46.0,
-      "step": 6311
-    },
-    {
-      "epoch": 0.4825964791559149,
-      "grad_norm": 0.0007399387541227043,
-      "learning_rate": 0.00019999988543249736,
-      "loss": 46.0,
-      "step": 6312
-    },
-    {
-      "epoch": 0.4826729361393046,
-      "grad_norm": 0.0017852268647402525,
-      "learning_rate": 0.0001999998853961354,
-      "loss": 46.0,
-      "step": 6313
-    },
-    {
-      "epoch": 0.4827493931226943,
-      "grad_norm": 0.0006445020553655922,
-      "learning_rate": 0.00019999988535976764,
-      "loss": 46.0,
-      "step": 6314
-    },
-    {
-      "epoch": 0.4828258501060841,
-      "grad_norm": 0.0011874273186549544,
-      "learning_rate": 0.00019999988532339412,
-      "loss": 46.0,
-      "step": 6315
-    },
-    {
-      "epoch": 0.4829023070894738,
-      "grad_norm": 0.0010938086779788136,
-      "learning_rate": 0.00019999988528701486,
-      "loss": 46.0,
-      "step": 6316
-    },
-    {
-      "epoch": 0.4829787640728635,
-      "grad_norm": 0.0034574014134705067,
-      "learning_rate": 0.0001999998852506298,
-      "loss": 46.0,
-      "step": 6317
-    },
-    {
-      "epoch": 0.4830552210562532,
-      "grad_norm": 0.003022773191332817,
-      "learning_rate": 0.00019999988521423898,
-      "loss": 46.0,
-      "step": 6318
-    },
-    {
-      "epoch": 0.48313167803964296,
-      "grad_norm": 0.0035729440860450268,
-      "learning_rate": 0.0001999998851778424,
-      "loss": 46.0,
-      "step": 6319
-    },
-    {
-      "epoch": 0.48320813502303267,
-      "grad_norm": 0.0007687106844969094,
-      "learning_rate": 0.00019999988514144006,
-      "loss": 46.0,
-      "step": 6320
-    },
-    {
-      "epoch": 0.4832845920064224,
-      "grad_norm": 0.0017719018505886197,
-      "learning_rate": 0.0001999998851050319,
-      "loss": 46.0,
-      "step": 6321
-    },
-    {
-      "epoch": 0.4833610489898121,
-      "grad_norm": 0.0025150200817734003,
-      "learning_rate": 0.00019999988506861803,
-      "loss": 46.0,
-      "step": 6322
-    },
-    {
-      "epoch": 0.48343750597320184,
-      "grad_norm": 0.0020292329136282206,
-      "learning_rate": 0.00019999988503219835,
-      "loss": 46.0,
-      "step": 6323
-    },
-    {
-      "epoch": 0.48351396295659155,
-      "grad_norm": 0.0011740690097212791,
-      "learning_rate": 0.00019999988499577295,
-      "loss": 46.0,
-      "step": 6324
-    },
-    {
-      "epoch": 0.48359041993998125,
-      "grad_norm": 0.002836063038557768,
-      "learning_rate": 0.00019999988495934172,
-      "loss": 46.0,
-      "step": 6325
-    },
-    {
-      "epoch": 0.483666876923371,
-      "grad_norm": 0.0003468339564278722,
-      "learning_rate": 0.00019999988492290472,
-      "loss": 46.0,
-      "step": 6326
-    },
-    {
-      "epoch": 0.4837433339067607,
-      "grad_norm": 0.0010148851433768868,
-      "learning_rate": 0.000199999884886462,
-      "loss": 46.0,
-      "step": 6327
-    },
-    {
-      "epoch": 0.4838197908901504,
-      "grad_norm": 0.002476799301803112,
-      "learning_rate": 0.00019999988485001349,
-      "loss": 46.0,
-      "step": 6328
-    },
-    {
-      "epoch": 0.48389624787354013,
-      "grad_norm": 0.0008766606915742159,
-      "learning_rate": 0.0001999998848135592,
-      "loss": 46.0,
-      "step": 6329
-    },
-    {
-      "epoch": 0.4839727048569299,
-      "grad_norm": 0.00042470821063034236,
-      "learning_rate": 0.00019999988477709915,
-      "loss": 46.0,
-      "step": 6330
-    },
-    {
-      "epoch": 0.4840491618403196,
-      "grad_norm": 0.0014834204921498895,
-      "learning_rate": 0.00019999988474063334,
-      "loss": 46.0,
-      "step": 6331
-    },
-    {
-      "epoch": 0.4841256188237093,
-      "grad_norm": 0.0006786793819628656,
-      "learning_rate": 0.00019999988470416176,
-      "loss": 46.0,
-      "step": 6332
-    },
-    {
-      "epoch": 0.484202075807099,
-      "grad_norm": 0.001179218990728259,
-      "learning_rate": 0.00019999988466768437,
-      "loss": 46.0,
-      "step": 6333
-    },
-    {
-      "epoch": 0.48427853279048877,
-      "grad_norm": 0.0006261045928113163,
-      "learning_rate": 0.00019999988463120124,
-      "loss": 46.0,
-      "step": 6334
-    },
-    {
-      "epoch": 0.4843549897738785,
-      "grad_norm": 0.0013071447610855103,
-      "learning_rate": 0.00019999988459471236,
-      "loss": 46.0,
-      "step": 6335
-    },
-    {
-      "epoch": 0.4844314467572682,
-      "grad_norm": 0.0010755738476291299,
-      "learning_rate": 0.0001999998845582177,
-      "loss": 46.0,
-      "step": 6336
-    },
-    {
-      "epoch": 0.4845079037406579,
-      "grad_norm": 0.0010401384206488729,
-      "learning_rate": 0.00019999988452171726,
-      "loss": 46.0,
-      "step": 6337
-    },
-    {
-      "epoch": 0.48458436072404765,
-      "grad_norm": 0.0011602689046412706,
-      "learning_rate": 0.00019999988448521103,
-      "loss": 46.0,
-      "step": 6338
-    },
-    {
-      "epoch": 0.48466081770743735,
-      "grad_norm": 0.0009871586225926876,
-      "learning_rate": 0.00019999988444869906,
-      "loss": 46.0,
-      "step": 6339
-    },
-    {
-      "epoch": 0.48473727469082706,
-      "grad_norm": 0.0008016563951969147,
-      "learning_rate": 0.00019999988441218131,
-      "loss": 46.0,
-      "step": 6340
-    },
-    {
-      "epoch": 0.4848137316742168,
-      "grad_norm": 0.001336134155280888,
-      "learning_rate": 0.0001999998843756578,
-      "loss": 46.0,
-      "step": 6341
-    },
-    {
-      "epoch": 0.4848901886576065,
-      "grad_norm": 0.0016448759706690907,
-      "learning_rate": 0.0001999998843391285,
-      "loss": 46.0,
-      "step": 6342
-    },
-    {
-      "epoch": 0.48496664564099623,
-      "grad_norm": 0.0005876056966371834,
-      "learning_rate": 0.00019999988430259347,
-      "loss": 46.0,
-      "step": 6343
-    },
-    {
-      "epoch": 0.48504310262438594,
-      "grad_norm": 0.006785241421312094,
-      "learning_rate": 0.00019999988426605265,
-      "loss": 46.0,
-      "step": 6344
-    },
-    {
-      "epoch": 0.4851195596077757,
-      "grad_norm": 0.0006073467084206641,
-      "learning_rate": 0.00019999988422950604,
-      "loss": 46.0,
-      "step": 6345
-    },
-    {
-      "epoch": 0.4851960165911654,
-      "grad_norm": 0.0005689759855158627,
-      "learning_rate": 0.00019999988419295368,
-      "loss": 46.0,
-      "step": 6346
-    },
-    {
-      "epoch": 0.4852724735745551,
-      "grad_norm": 0.0011022239923477173,
-      "learning_rate": 0.00019999988415639555,
-      "loss": 46.0,
-      "step": 6347
-    },
-    {
-      "epoch": 0.4853489305579448,
-      "grad_norm": 0.0011752459686249495,
-      "learning_rate": 0.00019999988411983165,
-      "loss": 46.0,
-      "step": 6348
-    },
-    {
-      "epoch": 0.4854253875413346,
-      "grad_norm": 0.0012171799317002296,
-      "learning_rate": 0.00019999988408326197,
-      "loss": 46.0,
-      "step": 6349
-    },
-    {
-      "epoch": 0.4855018445247243,
-      "grad_norm": 0.0023077267687767744,
-      "learning_rate": 0.00019999988404668652,
-      "loss": 46.0,
-      "step": 6350
-    },
-    {
-      "epoch": 0.485578301508114,
-      "grad_norm": 0.005109384190291166,
-      "learning_rate": 0.00019999988401010532,
-      "loss": 46.0,
-      "step": 6351
-    },
-    {
-      "epoch": 0.4856547584915037,
-      "grad_norm": 0.004220910370349884,
-      "learning_rate": 0.00019999988397351835,
-      "loss": 46.0,
-      "step": 6352
-    },
-    {
-      "epoch": 0.48573121547489345,
-      "grad_norm": 0.000715788861270994,
-      "learning_rate": 0.0001999998839369256,
-      "loss": 46.0,
-      "step": 6353
-    },
-    {
-      "epoch": 0.48580767245828316,
-      "grad_norm": 0.01046636514365673,
-      "learning_rate": 0.00019999988390032706,
-      "loss": 46.0,
-      "step": 6354
-    },
-    {
-      "epoch": 0.48588412944167286,
-      "grad_norm": 0.003636591136455536,
-      "learning_rate": 0.0001999998838637228,
-      "loss": 46.0,
-      "step": 6355
-    },
-    {
-      "epoch": 0.4859605864250626,
-      "grad_norm": 0.017490627244114876,
-      "learning_rate": 0.00019999988382711275,
-      "loss": 46.0,
-      "step": 6356
-    },
-    {
-      "epoch": 0.48603704340845233,
-      "grad_norm": 0.0006170101114548743,
-      "learning_rate": 0.00019999988379049692,
-      "loss": 46.0,
-      "step": 6357
-    },
-    {
-      "epoch": 0.48611350039184203,
-      "grad_norm": 0.0035188018810003996,
-      "learning_rate": 0.00019999988375387533,
-      "loss": 46.0,
-      "step": 6358
-    },
-    {
-      "epoch": 0.48618995737523174,
-      "grad_norm": 0.003433680161833763,
-      "learning_rate": 0.00019999988371724797,
-      "loss": 46.0,
-      "step": 6359
-    },
-    {
-      "epoch": 0.4862664143586215,
-      "grad_norm": 0.0006563492934219539,
-      "learning_rate": 0.00019999988368061481,
-      "loss": 46.0,
-      "step": 6360
-    },
-    {
-      "epoch": 0.4863428713420112,
-      "grad_norm": 0.009670495055615902,
-      "learning_rate": 0.0001999998836439759,
-      "loss": 46.0,
-      "step": 6361
-    },
-    {
-      "epoch": 0.4864193283254009,
-      "grad_norm": 0.0008141860598698258,
-      "learning_rate": 0.0001999998836073312,
-      "loss": 46.0,
-      "step": 6362
-    },
-    {
-      "epoch": 0.4864957853087906,
-      "grad_norm": 0.0006499075680039823,
-      "learning_rate": 0.0001999998835706808,
-      "loss": 46.0,
-      "step": 6363
-    },
-    {
-      "epoch": 0.4865722422921804,
-      "grad_norm": 0.0013896024320274591,
-      "learning_rate": 0.00019999988353402456,
-      "loss": 46.0,
-      "step": 6364
-    },
-    {
-      "epoch": 0.4866486992755701,
-      "grad_norm": 0.0007991960737854242,
-      "learning_rate": 0.0001999998834973626,
-      "loss": 46.0,
-      "step": 6365
-    },
-    {
-      "epoch": 0.4867251562589598,
-      "grad_norm": 0.0012706021079793572,
-      "learning_rate": 0.00019999988346069484,
-      "loss": 46.0,
-      "step": 6366
-    },
-    {
-      "epoch": 0.4868016132423495,
-      "grad_norm": 0.0010827386286109686,
-      "learning_rate": 0.00019999988342402132,
-      "loss": 46.0,
-      "step": 6367
-    },
-    {
-      "epoch": 0.48687807022573926,
-      "grad_norm": 0.004039994440972805,
-      "learning_rate": 0.00019999988338734203,
-      "loss": 46.0,
-      "step": 6368
-    },
-    {
-      "epoch": 0.48695452720912896,
-      "grad_norm": 0.0011493003694340587,
-      "learning_rate": 0.000199999883350657,
-      "loss": 46.0,
-      "step": 6369
-    },
-    {
-      "epoch": 0.48703098419251867,
-      "grad_norm": 0.0016639565583318472,
-      "learning_rate": 0.00019999988331396615,
-      "loss": 46.0,
-      "step": 6370
-    },
-    {
-      "epoch": 0.48710744117590843,
-      "grad_norm": 0.0009372640633955598,
-      "learning_rate": 0.00019999988327726954,
-      "loss": 46.0,
-      "step": 6371
-    },
-    {
-      "epoch": 0.48718389815929813,
-      "grad_norm": 0.007675940170884132,
-      "learning_rate": 0.00019999988324056719,
-      "loss": 46.0,
-      "step": 6372
-    },
-    {
-      "epoch": 0.48726035514268784,
-      "grad_norm": 0.0007198092062026262,
-      "learning_rate": 0.00019999988320385903,
-      "loss": 46.0,
-      "step": 6373
-    },
-    {
-      "epoch": 0.48733681212607755,
-      "grad_norm": 0.0006535969441756606,
-      "learning_rate": 0.00019999988316714512,
-      "loss": 46.0,
-      "step": 6374
-    },
-    {
-      "epoch": 0.4874132691094673,
-      "grad_norm": 0.0034300717525184155,
-      "learning_rate": 0.00019999988313042544,
-      "loss": 46.0,
-      "step": 6375
-    },
-    {
-      "epoch": 0.487489726092857,
-      "grad_norm": 0.0012840520357713103,
-      "learning_rate": 0.0001999998830937,
-      "loss": 46.0,
-      "step": 6376
-    },
-    {
-      "epoch": 0.4875661830762467,
-      "grad_norm": 0.000656074145808816,
-      "learning_rate": 0.0001999998830569688,
-      "loss": 46.0,
-      "step": 6377
-    },
-    {
-      "epoch": 0.4876426400596364,
-      "grad_norm": 0.0011463371338322759,
-      "learning_rate": 0.0001999998830202318,
-      "loss": 46.0,
-      "step": 6378
-    },
-    {
-      "epoch": 0.4877190970430262,
-      "grad_norm": 0.0008213612018153071,
-      "learning_rate": 0.00019999988298348903,
-      "loss": 46.0,
-      "step": 6379
-    },
-    {
-      "epoch": 0.4877955540264159,
-      "grad_norm": 0.0029335820581763983,
-      "learning_rate": 0.00019999988294674054,
-      "loss": 46.0,
-      "step": 6380
-    },
-    {
-      "epoch": 0.4878720110098056,
-      "grad_norm": 0.0005414159968495369,
-      "learning_rate": 0.00019999988290998625,
-      "loss": 46.0,
-      "step": 6381
-    },
-    {
-      "epoch": 0.48794846799319536,
-      "grad_norm": 0.0004125908308196813,
-      "learning_rate": 0.00019999988287322618,
-      "loss": 46.0,
-      "step": 6382
-    },
-    {
-      "epoch": 0.48802492497658506,
-      "grad_norm": 0.004388558678328991,
-      "learning_rate": 0.00019999988283646034,
-      "loss": 46.0,
-      "step": 6383
-    },
-    {
-      "epoch": 0.48810138195997477,
-      "grad_norm": 0.0010814444394782186,
-      "learning_rate": 0.00019999988279968873,
-      "loss": 46.0,
-      "step": 6384
-    },
-    {
-      "epoch": 0.4881778389433645,
-      "grad_norm": 0.001897557289339602,
-      "learning_rate": 0.00019999988276291135,
-      "loss": 46.0,
-      "step": 6385
-    },
-    {
-      "epoch": 0.48825429592675423,
-      "grad_norm": 0.0008687134250067174,
-      "learning_rate": 0.00019999988272612824,
-      "loss": 46.0,
-      "step": 6386
-    },
-    {
-      "epoch": 0.48833075291014394,
-      "grad_norm": 0.0009987280936911702,
-      "learning_rate": 0.0001999998826893393,
-      "loss": 46.0,
-      "step": 6387
-    },
-    {
-      "epoch": 0.48840720989353364,
-      "grad_norm": 0.0006901868619024754,
-      "learning_rate": 0.00019999988265254463,
-      "loss": 46.0,
-      "step": 6388
-    },
-    {
-      "epoch": 0.48848366687692335,
-      "grad_norm": 0.0005795607576146722,
-      "learning_rate": 0.00019999988261574418,
-      "loss": 46.0,
-      "step": 6389
-    },
-    {
-      "epoch": 0.4885601238603131,
-      "grad_norm": 0.003952940925955772,
-      "learning_rate": 0.000199999882578938,
-      "loss": 46.0,
-      "step": 6390
-    },
-    {
-      "epoch": 0.4886365808437028,
-      "grad_norm": 0.0011986320605501533,
-      "learning_rate": 0.000199999882542126,
-      "loss": 46.0,
-      "step": 6391
-    },
-    {
-      "epoch": 0.4887130378270925,
-      "grad_norm": 0.0006881254375912249,
-      "learning_rate": 0.00019999988250530822,
-      "loss": 46.0,
-      "step": 6392
-    },
-    {
-      "epoch": 0.48878949481048223,
-      "grad_norm": 0.01511130016297102,
-      "learning_rate": 0.0001999998824684847,
-      "loss": 46.0,
-      "step": 6393
-    },
-    {
-      "epoch": 0.488865951793872,
-      "grad_norm": 0.0015410013729706407,
-      "learning_rate": 0.00019999988243165538,
-      "loss": 46.0,
-      "step": 6394
-    },
-    {
-      "epoch": 0.4889424087772617,
-      "grad_norm": 0.0036170308012515306,
-      "learning_rate": 0.00019999988239482032,
-      "loss": 46.0,
-      "step": 6395
-    },
-    {
-      "epoch": 0.4890188657606514,
-      "grad_norm": 0.0007317258277907968,
-      "learning_rate": 0.00019999988235797948,
-      "loss": 46.0,
-      "step": 6396
-    },
-    {
-      "epoch": 0.48909532274404116,
-      "grad_norm": 0.0010901440400630236,
-      "learning_rate": 0.0001999998823211329,
-      "loss": 46.0,
-      "step": 6397
-    },
-    {
-      "epoch": 0.48917177972743087,
-      "grad_norm": 0.0006672409363090992,
-      "learning_rate": 0.0001999998822842805,
-      "loss": 46.0,
-      "step": 6398
-    },
-    {
-      "epoch": 0.48924823671082057,
-      "grad_norm": 0.0011397808557376266,
-      "learning_rate": 0.00019999988224742236,
-      "loss": 46.0,
-      "step": 6399
-    },
-    {
-      "epoch": 0.4893246936942103,
-      "grad_norm": 0.0008003698894754052,
-      "learning_rate": 0.00019999988221055848,
-      "loss": 46.0,
-      "step": 6400
-    },
-    {
-      "epoch": 0.48940115067760004,
-      "grad_norm": 0.002308068098500371,
-      "learning_rate": 0.00019999988217368877,
-      "loss": 46.0,
-      "step": 6401
-    },
-    {
-      "epoch": 0.48947760766098974,
-      "grad_norm": 0.0008845416596159339,
-      "learning_rate": 0.00019999988213681332,
-      "loss": 46.0,
-      "step": 6402
-    },
-    {
-      "epoch": 0.48955406464437945,
-      "grad_norm": 0.0008718667668290436,
-      "learning_rate": 0.00019999988209993213,
-      "loss": 46.0,
-      "step": 6403
-    },
-    {
-      "epoch": 0.48963052162776916,
-      "grad_norm": 0.0007148516015149653,
-      "learning_rate": 0.00019999988206304513,
-      "loss": 46.0,
-      "step": 6404
-    },
-    {
-      "epoch": 0.4897069786111589,
-      "grad_norm": 0.0003723963745869696,
-      "learning_rate": 0.00019999988202615233,
-      "loss": 46.0,
-      "step": 6405
-    },
-    {
-      "epoch": 0.4897834355945486,
-      "grad_norm": 0.0019716841634362936,
-      "learning_rate": 0.00019999988198925381,
-      "loss": 46.0,
-      "step": 6406
-    },
-    {
-      "epoch": 0.4898598925779383,
-      "grad_norm": 0.0017271449323743582,
-      "learning_rate": 0.00019999988195234952,
-      "loss": 46.0,
-      "step": 6407
-    },
-    {
-      "epoch": 0.48993634956132803,
-      "grad_norm": 0.005223521497100592,
-      "learning_rate": 0.00019999988191543946,
-      "loss": 46.0,
-      "step": 6408
-    },
-    {
-      "epoch": 0.4900128065447178,
-      "grad_norm": 0.0006758730160072446,
-      "learning_rate": 0.00019999988187852362,
-      "loss": 46.0,
-      "step": 6409
-    },
-    {
-      "epoch": 0.4900892635281075,
-      "grad_norm": 0.0004181225085631013,
-      "learning_rate": 0.000199999881841602,
-      "loss": 46.0,
-      "step": 6410
-    },
-    {
-      "epoch": 0.4901657205114972,
-      "grad_norm": 0.0006537472945638001,
-      "learning_rate": 0.00019999988180467463,
-      "loss": 46.0,
-      "step": 6411
-    },
-    {
-      "epoch": 0.49024217749488697,
-      "grad_norm": 0.0016621069516986609,
-      "learning_rate": 0.0001999998817677415,
-      "loss": 46.0,
-      "step": 6412
-    },
-    {
-      "epoch": 0.49031863447827667,
-      "grad_norm": 0.0028892322443425655,
-      "learning_rate": 0.00019999988173080257,
-      "loss": 46.0,
-      "step": 6413
-    },
-    {
-      "epoch": 0.4903950914616664,
-      "grad_norm": 0.0016960246721282601,
-      "learning_rate": 0.0001999998816938579,
-      "loss": 46.0,
-      "step": 6414
-    },
-    {
-      "epoch": 0.4904715484450561,
-      "grad_norm": 0.004497898742556572,
-      "learning_rate": 0.00019999988165690744,
-      "loss": 46.0,
-      "step": 6415
-    },
-    {
-      "epoch": 0.49054800542844584,
-      "grad_norm": 0.0006038066931068897,
-      "learning_rate": 0.00019999988161995122,
-      "loss": 46.0,
-      "step": 6416
-    },
-    {
-      "epoch": 0.49062446241183555,
-      "grad_norm": 0.010990196838974953,
-      "learning_rate": 0.00019999988158298925,
-      "loss": 46.0,
-      "step": 6417
-    },
-    {
-      "epoch": 0.49070091939522525,
-      "grad_norm": 0.0009468832286074758,
-      "learning_rate": 0.00019999988154602148,
-      "loss": 46.0,
-      "step": 6418
-    },
-    {
-      "epoch": 0.49077737637861496,
-      "grad_norm": 0.0031462351325899363,
-      "learning_rate": 0.00019999988150904793,
-      "loss": 46.0,
-      "step": 6419
-    },
-    {
-      "epoch": 0.4908538333620047,
-      "grad_norm": 0.0016019238391891122,
-      "learning_rate": 0.00019999988147206862,
-      "loss": 46.0,
-      "step": 6420
-    },
-    {
-      "epoch": 0.4909302903453944,
-      "grad_norm": 0.004620174877345562,
-      "learning_rate": 0.00019999988143508355,
-      "loss": 46.0,
-      "step": 6421
-    },
-    {
-      "epoch": 0.49100674732878413,
-      "grad_norm": 0.002342072082683444,
-      "learning_rate": 0.00019999988139809272,
-      "loss": 46.0,
-      "step": 6422
-    },
-    {
-      "epoch": 0.49108320431217384,
-      "grad_norm": 0.009439565241336823,
-      "learning_rate": 0.0001999998813610961,
-      "loss": 46.0,
-      "step": 6423
-    },
-    {
-      "epoch": 0.4911596612955636,
-      "grad_norm": 0.001219587167724967,
-      "learning_rate": 0.0001999998813240937,
-      "loss": 46.0,
-      "step": 6424
-    },
-    {
-      "epoch": 0.4912361182789533,
-      "grad_norm": 0.0008230343810282648,
-      "learning_rate": 0.00019999988128708557,
-      "loss": 46.0,
-      "step": 6425
-    },
-    {
-      "epoch": 0.491312575262343,
-      "grad_norm": 0.0014162807492539287,
-      "learning_rate": 0.00019999988125007166,
-      "loss": 46.0,
-      "step": 6426
-    },
-    {
-      "epoch": 0.49138903224573277,
-      "grad_norm": 0.0016766769113019109,
-      "learning_rate": 0.00019999988121305196,
-      "loss": 46.0,
-      "step": 6427
-    },
-    {
-      "epoch": 0.4914654892291225,
-      "grad_norm": 0.0006250288570299745,
-      "learning_rate": 0.0001999998811760265,
-      "loss": 46.0,
-      "step": 6428
-    },
-    {
-      "epoch": 0.4915419462125122,
-      "grad_norm": 0.0005541299469769001,
-      "learning_rate": 0.00019999988113899528,
-      "loss": 46.0,
-      "step": 6429
-    },
-    {
-      "epoch": 0.4916184031959019,
-      "grad_norm": 0.0016851290129125118,
-      "learning_rate": 0.0001999998811019583,
-      "loss": 46.0,
-      "step": 6430
-    },
-    {
-      "epoch": 0.49169486017929165,
-      "grad_norm": 0.010401006788015366,
-      "learning_rate": 0.00019999988106491552,
-      "loss": 46.0,
-      "step": 6431
-    },
-    {
-      "epoch": 0.49177131716268135,
-      "grad_norm": 0.001281034667044878,
-      "learning_rate": 0.00019999988102786697,
-      "loss": 46.0,
-      "step": 6432
-    },
-    {
-      "epoch": 0.49184777414607106,
-      "grad_norm": 0.0022689623292535543,
-      "learning_rate": 0.00019999988099081268,
-      "loss": 46.0,
-      "step": 6433
-    },
-    {
-      "epoch": 0.49192423112946077,
-      "grad_norm": 0.018421540036797523,
-      "learning_rate": 0.00019999988095375262,
-      "loss": 46.0,
-      "step": 6434
-    },
-    {
-      "epoch": 0.4920006881128505,
-      "grad_norm": 0.0012801343109458685,
-      "learning_rate": 0.00019999988091668676,
-      "loss": 46.0,
-      "step": 6435
-    },
-    {
-      "epoch": 0.49207714509624023,
-      "grad_norm": 0.0005204727640375495,
-      "learning_rate": 0.00019999988087961515,
-      "loss": 46.0,
-      "step": 6436
-    },
-    {
-      "epoch": 0.49215360207962994,
-      "grad_norm": 0.0020112525671720505,
-      "learning_rate": 0.00019999988084253776,
-      "loss": 46.0,
-      "step": 6437
-    },
-    {
-      "epoch": 0.49223005906301964,
-      "grad_norm": 0.00383200217038393,
-      "learning_rate": 0.0001999998808054546,
-      "loss": 46.0,
-      "step": 6438
-    },
-    {
-      "epoch": 0.4923065160464094,
-      "grad_norm": 0.0014632859965786338,
-      "learning_rate": 0.00019999988076836568,
-      "loss": 46.0,
-      "step": 6439
-    },
-    {
-      "epoch": 0.4923829730297991,
-      "grad_norm": 0.001580241252668202,
-      "learning_rate": 0.000199999880731271,
-      "loss": 46.0,
-      "step": 6440
-    },
-    {
-      "epoch": 0.4924594300131888,
-      "grad_norm": 0.0012626753887161613,
-      "learning_rate": 0.00019999988069417053,
-      "loss": 46.0,
-      "step": 6441
-    },
-    {
-      "epoch": 0.4925358869965786,
-      "grad_norm": 0.0011472193291410804,
-      "learning_rate": 0.0001999998806570643,
-      "loss": 46.0,
-      "step": 6442
-    },
-    {
-      "epoch": 0.4926123439799683,
-      "grad_norm": 0.00302050425671041,
-      "learning_rate": 0.0001999998806199523,
-      "loss": 46.0,
-      "step": 6443
-    },
-    {
-      "epoch": 0.492688800963358,
-      "grad_norm": 0.012528634630143642,
-      "learning_rate": 0.00019999988058283454,
-      "loss": 46.0,
-      "step": 6444
-    },
-    {
-      "epoch": 0.4927652579467477,
-      "grad_norm": 0.002775819506496191,
-      "learning_rate": 0.00019999988054571102,
-      "loss": 46.0,
-      "step": 6445
-    },
-    {
-      "epoch": 0.49284171493013745,
-      "grad_norm": 0.0007796332938596606,
-      "learning_rate": 0.0001999998805085817,
-      "loss": 46.0,
-      "step": 6446
-    },
-    {
-      "epoch": 0.49291817191352716,
-      "grad_norm": 0.003013971960172057,
-      "learning_rate": 0.0001999998804714466,
-      "loss": 46.0,
-      "step": 6447
-    },
-    {
-      "epoch": 0.49299462889691686,
-      "grad_norm": 0.0017489292658865452,
-      "learning_rate": 0.00019999988043430578,
-      "loss": 46.0,
-      "step": 6448
-    },
-    {
-      "epoch": 0.49307108588030657,
-      "grad_norm": 0.00688300421461463,
-      "learning_rate": 0.00019999988039715917,
-      "loss": 46.0,
-      "step": 6449
-    },
-    {
-      "epoch": 0.49314754286369633,
-      "grad_norm": 0.0005748681141994894,
-      "learning_rate": 0.00019999988036000676,
-      "loss": 46.0,
-      "step": 6450
-    },
-    {
-      "epoch": 0.49322399984708604,
-      "grad_norm": 0.0005769344279542565,
-      "learning_rate": 0.00019999988032284863,
-      "loss": 46.0,
-      "step": 6451
-    },
-    {
-      "epoch": 0.49330045683047574,
-      "grad_norm": 0.0007875912124291062,
-      "learning_rate": 0.0001999998802856847,
-      "loss": 46.0,
-      "step": 6452
-    },
-    {
-      "epoch": 0.49337691381386545,
-      "grad_norm": 0.0006917236023582518,
-      "learning_rate": 0.00019999988024851502,
-      "loss": 46.0,
-      "step": 6453
-    },
-    {
-      "epoch": 0.4934533707972552,
-      "grad_norm": 0.006839110050350428,
-      "learning_rate": 0.00019999988021133955,
-      "loss": 46.0,
-      "step": 6454
-    },
-    {
-      "epoch": 0.4935298277806449,
-      "grad_norm": 0.0008490763721056283,
-      "learning_rate": 0.00019999988017415832,
-      "loss": 46.0,
-      "step": 6455
-    },
-    {
-      "epoch": 0.4936062847640346,
-      "grad_norm": 0.0012166728265583515,
-      "learning_rate": 0.0001999998801369713,
-      "loss": 46.0,
-      "step": 6456
-    },
-    {
-      "epoch": 0.4936827417474244,
-      "grad_norm": 0.0005295455921441317,
-      "learning_rate": 0.00019999988009977853,
-      "loss": 46.0,
-      "step": 6457
-    },
-    {
-      "epoch": 0.4937591987308141,
-      "grad_norm": 0.0006590589182451367,
-      "learning_rate": 0.00019999988006258,
-      "loss": 46.0,
-      "step": 6458
-    },
-    {
-      "epoch": 0.4938356557142038,
-      "grad_norm": 0.0014256765134632587,
-      "learning_rate": 0.0001999998800253757,
-      "loss": 46.0,
-      "step": 6459
-    },
-    {
-      "epoch": 0.4939121126975935,
-      "grad_norm": 0.0011379234492778778,
-      "learning_rate": 0.0001999998799881656,
-      "loss": 46.0,
-      "step": 6460
-    },
-    {
-      "epoch": 0.49398856968098326,
-      "grad_norm": 0.0008576003019697964,
-      "learning_rate": 0.00019999987995094975,
-      "loss": 46.0,
-      "step": 6461
-    },
-    {
-      "epoch": 0.49406502666437296,
-      "grad_norm": 0.004929020535200834,
-      "learning_rate": 0.00019999987991372814,
-      "loss": 46.0,
-      "step": 6462
-    },
-    {
-      "epoch": 0.49414148364776267,
-      "grad_norm": 0.0016787319909781218,
-      "learning_rate": 0.00019999987987650073,
-      "loss": 46.0,
-      "step": 6463
-    },
-    {
-      "epoch": 0.4942179406311524,
-      "grad_norm": 0.0008172642556019127,
-      "learning_rate": 0.00019999987983926757,
-      "loss": 46.0,
-      "step": 6464
-    },
-    {
-      "epoch": 0.49429439761454214,
-      "grad_norm": 0.0009970619576051831,
-      "learning_rate": 0.00019999987980202867,
-      "loss": 46.0,
-      "step": 6465
-    },
-    {
-      "epoch": 0.49437085459793184,
-      "grad_norm": 0.000629753281828016,
-      "learning_rate": 0.00019999987976478397,
-      "loss": 46.0,
-      "step": 6466
-    },
-    {
-      "epoch": 0.49444731158132155,
-      "grad_norm": 0.0009540155297145247,
-      "learning_rate": 0.0001999998797275335,
-      "loss": 46.0,
-      "step": 6467
-    },
-    {
-      "epoch": 0.49452376856471125,
-      "grad_norm": 0.0036007454618811607,
-      "learning_rate": 0.00019999987969027724,
-      "loss": 46.0,
-      "step": 6468
-    },
-    {
-      "epoch": 0.494600225548101,
-      "grad_norm": 0.0007068145205266774,
-      "learning_rate": 0.00019999987965301524,
-      "loss": 46.0,
-      "step": 6469
-    },
-    {
-      "epoch": 0.4946766825314907,
-      "grad_norm": 0.003312127897515893,
-      "learning_rate": 0.00019999987961574748,
-      "loss": 46.0,
-      "step": 6470
-    },
-    {
-      "epoch": 0.4947531395148804,
-      "grad_norm": 0.0021120829042047262,
-      "learning_rate": 0.00019999987957847393,
-      "loss": 46.0,
-      "step": 6471
-    },
-    {
-      "epoch": 0.4948295964982702,
-      "grad_norm": 0.002806598087772727,
-      "learning_rate": 0.00019999987954119464,
-      "loss": 46.0,
-      "step": 6472
-    },
-    {
-      "epoch": 0.4949060534816599,
-      "grad_norm": 0.002394828712567687,
-      "learning_rate": 0.00019999987950390955,
-      "loss": 46.0,
-      "step": 6473
-    },
-    {
-      "epoch": 0.4949825104650496,
-      "grad_norm": 0.0004253096121829003,
-      "learning_rate": 0.0001999998794666187,
-      "loss": 46.0,
-      "step": 6474
-    },
-    {
-      "epoch": 0.4950589674484393,
-      "grad_norm": 0.0022463074419647455,
-      "learning_rate": 0.00019999987942932208,
-      "loss": 46.0,
-      "step": 6475
-    },
-    {
-      "epoch": 0.49513542443182906,
-      "grad_norm": 0.0017045411514118314,
-      "learning_rate": 0.0001999998793920197,
-      "loss": 46.0,
-      "step": 6476
-    },
-    {
-      "epoch": 0.49521188141521877,
-      "grad_norm": 0.001246497267857194,
-      "learning_rate": 0.00019999987935471152,
-      "loss": 46.0,
-      "step": 6477
-    },
-    {
-      "epoch": 0.4952883383986085,
-      "grad_norm": 0.0009801244596019387,
-      "learning_rate": 0.0001999998793173976,
-      "loss": 46.0,
-      "step": 6478
-    },
-    {
-      "epoch": 0.4953647953819982,
-      "grad_norm": 0.0005795637262053788,
-      "learning_rate": 0.00019999987928007791,
-      "loss": 46.0,
-      "step": 6479
-    },
-    {
-      "epoch": 0.49544125236538794,
-      "grad_norm": 0.0016051421407610178,
-      "learning_rate": 0.00019999987924275244,
-      "loss": 46.0,
-      "step": 6480
-    },
-    {
-      "epoch": 0.49551770934877765,
-      "grad_norm": 0.0010982013773173094,
-      "learning_rate": 0.0001999998792054212,
-      "loss": 46.0,
-      "step": 6481
-    },
-    {
-      "epoch": 0.49559416633216735,
-      "grad_norm": 0.0018951896345242858,
-      "learning_rate": 0.0001999998791680842,
-      "loss": 46.0,
-      "step": 6482
-    },
-    {
-      "epoch": 0.49567062331555706,
-      "grad_norm": 0.009181102737784386,
-      "learning_rate": 0.0001999998791307414,
-      "loss": 46.0,
-      "step": 6483
-    },
-    {
-      "epoch": 0.4957470802989468,
-      "grad_norm": 0.000797546876128763,
-      "learning_rate": 0.00019999987909339283,
-      "loss": 46.0,
-      "step": 6484
-    },
-    {
-      "epoch": 0.4958235372823365,
-      "grad_norm": 0.000757049594540149,
-      "learning_rate": 0.00019999987905603854,
-      "loss": 46.0,
-      "step": 6485
-    },
-    {
-      "epoch": 0.49589999426572623,
-      "grad_norm": 0.0016351699596270919,
-      "learning_rate": 0.00019999987901867845,
-      "loss": 46.0,
-      "step": 6486
-    },
-    {
-      "epoch": 0.495976451249116,
-      "grad_norm": 0.001320823677815497,
-      "learning_rate": 0.0001999998789813126,
-      "loss": 46.0,
-      "step": 6487
-    },
-    {
-      "epoch": 0.4960529082325057,
-      "grad_norm": 0.0006964000058360398,
-      "learning_rate": 0.00019999987894394095,
-      "loss": 46.0,
-      "step": 6488
-    },
-    {
-      "epoch": 0.4961293652158954,
-      "grad_norm": 0.000624475593212992,
-      "learning_rate": 0.00019999987890656357,
-      "loss": 46.0,
-      "step": 6489
-    },
-    {
-      "epoch": 0.4962058221992851,
-      "grad_norm": 0.0008864251431077719,
-      "learning_rate": 0.00019999987886918041,
-      "loss": 46.0,
-      "step": 6490
-    },
-    {
-      "epoch": 0.49628227918267487,
-      "grad_norm": 0.000672024383675307,
-      "learning_rate": 0.00019999987883179149,
-      "loss": 46.0,
-      "step": 6491
-    },
-    {
-      "epoch": 0.4963587361660646,
-      "grad_norm": 0.00029146266751922667,
-      "learning_rate": 0.00019999987879439676,
-      "loss": 46.0,
-      "step": 6492
-    },
-    {
-      "epoch": 0.4964351931494543,
-      "grad_norm": 0.002330700634047389,
-      "learning_rate": 0.0001999998787569963,
-      "loss": 46.0,
-      "step": 6493
-    },
-    {
-      "epoch": 0.496511650132844,
-      "grad_norm": 0.0022235598880797625,
-      "learning_rate": 0.00019999987871959006,
-      "loss": 46.0,
-      "step": 6494
-    },
-    {
-      "epoch": 0.49658810711623375,
-      "grad_norm": 0.0008372527081519365,
-      "learning_rate": 0.00019999987868217804,
-      "loss": 46.0,
-      "step": 6495
-    },
-    {
-      "epoch": 0.49666456409962345,
-      "grad_norm": 0.0009484909824095666,
-      "learning_rate": 0.0001999998786447603,
-      "loss": 46.0,
-      "step": 6496
-    },
-    {
-      "epoch": 0.49674102108301316,
-      "grad_norm": 0.00572090083733201,
-      "learning_rate": 0.00019999987860733673,
-      "loss": 46.0,
-      "step": 6497
-    },
-    {
-      "epoch": 0.4968174780664029,
-      "grad_norm": 0.0011128135956823826,
-      "learning_rate": 0.0001999998785699074,
-      "loss": 46.0,
-      "step": 6498
-    },
-    {
-      "epoch": 0.4968939350497926,
-      "grad_norm": 0.0007173815974965692,
-      "learning_rate": 0.00019999987853247232,
-      "loss": 46.0,
-      "step": 6499
-    },
-    {
-      "epoch": 0.49697039203318233,
-      "grad_norm": 0.0004026094975415617,
-      "learning_rate": 0.0001999998784950315,
-      "loss": 46.0,
-      "step": 6500
-    },
-    {
-      "epoch": 0.49704684901657203,
-      "grad_norm": 0.0025437013246119022,
-      "learning_rate": 0.00019999987845758483,
-      "loss": 46.0,
-      "step": 6501
-    },
-    {
-      "epoch": 0.4971233059999618,
-      "grad_norm": 0.004859997890889645,
-      "learning_rate": 0.00019999987842013244,
-      "loss": 46.0,
-      "step": 6502
-    },
-    {
-      "epoch": 0.4971997629833515,
-      "grad_norm": 0.0007493896409869194,
-      "learning_rate": 0.0001999998783826743,
-      "loss": 46.0,
-      "step": 6503
-    },
-    {
-      "epoch": 0.4972762199667412,
-      "grad_norm": 0.0010843091877177358,
-      "learning_rate": 0.00019999987834521033,
-      "loss": 46.0,
-      "step": 6504
-    },
-    {
-      "epoch": 0.4973526769501309,
-      "grad_norm": 0.004104179795831442,
-      "learning_rate": 0.00019999987830774063,
-      "loss": 46.0,
-      "step": 6505
-    },
-    {
-      "epoch": 0.4974291339335207,
-      "grad_norm": 0.0033110808581113815,
-      "learning_rate": 0.00019999987827026516,
-      "loss": 46.0,
-      "step": 6506
-    },
-    {
-      "epoch": 0.4975055909169104,
-      "grad_norm": 0.0006934861303307116,
-      "learning_rate": 0.00019999987823278393,
-      "loss": 46.0,
-      "step": 6507
-    },
-    {
-      "epoch": 0.4975820479003001,
-      "grad_norm": 0.006649675313383341,
-      "learning_rate": 0.0001999998781952969,
-      "loss": 46.0,
-      "step": 6508
-    },
-    {
-      "epoch": 0.4976585048836898,
-      "grad_norm": 0.002636313671246171,
-      "learning_rate": 0.00019999987815780412,
-      "loss": 46.0,
-      "step": 6509
-    },
-    {
-      "epoch": 0.49773496186707955,
-      "grad_norm": 0.001595677575096488,
-      "learning_rate": 0.00019999987812030558,
-      "loss": 46.0,
-      "step": 6510
-    },
-    {
-      "epoch": 0.49781141885046926,
-      "grad_norm": 0.001560879754833877,
-      "learning_rate": 0.00019999987808280126,
-      "loss": 46.0,
-      "step": 6511
-    },
-    {
-      "epoch": 0.49788787583385896,
-      "grad_norm": 0.0010170344030484557,
-      "learning_rate": 0.00019999987804529115,
-      "loss": 46.0,
-      "step": 6512
-    },
-    {
-      "epoch": 0.4979643328172487,
-      "grad_norm": 0.0006142422207631171,
-      "learning_rate": 0.00019999987800777528,
-      "loss": 46.0,
-      "step": 6513
-    },
-    {
-      "epoch": 0.49804078980063843,
-      "grad_norm": 0.001040930044837296,
-      "learning_rate": 0.00019999987797025365,
-      "loss": 46.0,
-      "step": 6514
-    },
-    {
-      "epoch": 0.49811724678402813,
-      "grad_norm": 0.0009109100210480392,
-      "learning_rate": 0.00019999987793272627,
-      "loss": 46.0,
-      "step": 6515
-    },
-    {
-      "epoch": 0.49819370376741784,
-      "grad_norm": 0.0010989943984895945,
-      "learning_rate": 0.0001999998778951931,
-      "loss": 46.0,
-      "step": 6516
-    },
-    {
-      "epoch": 0.4982701607508076,
-      "grad_norm": 0.0023832949809730053,
-      "learning_rate": 0.00019999987785765418,
-      "loss": 46.0,
-      "step": 6517
-    },
-    {
-      "epoch": 0.4983466177341973,
-      "grad_norm": 0.0045531452633440495,
-      "learning_rate": 0.00019999987782010946,
-      "loss": 46.0,
-      "step": 6518
-    },
-    {
-      "epoch": 0.498423074717587,
-      "grad_norm": 0.0014620426809415221,
-      "learning_rate": 0.00019999987778255898,
-      "loss": 46.0,
-      "step": 6519
-    },
-    {
-      "epoch": 0.4984995317009767,
-      "grad_norm": 0.001096074585802853,
-      "learning_rate": 0.00019999987774500273,
-      "loss": 46.0,
-      "step": 6520
-    },
-    {
-      "epoch": 0.4985759886843665,
-      "grad_norm": 0.0057863532565534115,
-      "learning_rate": 0.0001999998777074407,
-      "loss": 46.0,
-      "step": 6521
-    },
-    {
-      "epoch": 0.4986524456677562,
-      "grad_norm": 0.0012083373731002212,
-      "learning_rate": 0.00019999987766987292,
-      "loss": 46.0,
-      "step": 6522
-    },
-    {
-      "epoch": 0.4987289026511459,
-      "grad_norm": 0.002958812052384019,
-      "learning_rate": 0.00019999987763229937,
-      "loss": 46.0,
-      "step": 6523
-    },
-    {
-      "epoch": 0.4988053596345356,
-      "grad_norm": 0.008364793844521046,
-      "learning_rate": 0.00019999987759472006,
-      "loss": 46.0,
-      "step": 6524
-    },
-    {
-      "epoch": 0.49888181661792536,
-      "grad_norm": 0.0010773777030408382,
-      "learning_rate": 0.00019999987755713497,
-      "loss": 46.0,
-      "step": 6525
-    },
-    {
-      "epoch": 0.49895827360131506,
-      "grad_norm": 0.003642960684373975,
-      "learning_rate": 0.0001999998775195441,
-      "loss": 46.0,
-      "step": 6526
-    },
-    {
-      "epoch": 0.49903473058470477,
-      "grad_norm": 0.0007070060819387436,
-      "learning_rate": 0.00019999987748194745,
-      "loss": 46.0,
-      "step": 6527
-    },
-    {
-      "epoch": 0.4991111875680945,
-      "grad_norm": 0.0006475977716036141,
-      "learning_rate": 0.00019999987744434504,
-      "loss": 46.0,
-      "step": 6528
-    },
-    {
-      "epoch": 0.49918764455148423,
-      "grad_norm": 0.0013352928217500448,
-      "learning_rate": 0.00019999987740673686,
-      "loss": 46.0,
-      "step": 6529
-    },
-    {
-      "epoch": 0.49926410153487394,
-      "grad_norm": 0.0007934285094961524,
-      "learning_rate": 0.00019999987736912296,
-      "loss": 46.0,
-      "step": 6530
-    },
-    {
-      "epoch": 0.49934055851826364,
-      "grad_norm": 0.0018938110442832112,
-      "learning_rate": 0.00019999987733150323,
-      "loss": 46.0,
-      "step": 6531
-    },
-    {
-      "epoch": 0.4994170155016534,
-      "grad_norm": 0.0006374659715220332,
-      "learning_rate": 0.00019999987729387776,
-      "loss": 46.0,
-      "step": 6532
-    },
-    {
-      "epoch": 0.4994934724850431,
-      "grad_norm": 0.0004182657285127789,
-      "learning_rate": 0.00019999987725624648,
-      "loss": 46.0,
-      "step": 6533
-    },
-    {
-      "epoch": 0.4995699294684328,
-      "grad_norm": 0.001342519884929061,
-      "learning_rate": 0.00019999987721860949,
-      "loss": 46.0,
-      "step": 6534
-    },
-    {
-      "epoch": 0.4996463864518225,
-      "grad_norm": 0.012076162733137608,
-      "learning_rate": 0.00019999987718096672,
-      "loss": 46.0,
-      "step": 6535
-    },
-    {
-      "epoch": 0.4997228434352123,
-      "grad_norm": 0.002149401232600212,
-      "learning_rate": 0.00019999987714331815,
-      "loss": 46.0,
-      "step": 6536
-    },
-    {
-      "epoch": 0.499799300418602,
-      "grad_norm": 0.001540144789032638,
-      "learning_rate": 0.0001999998771056638,
-      "loss": 46.0,
-      "step": 6537
-    },
-    {
-      "epoch": 0.4998757574019917,
-      "grad_norm": 0.0009101862087845802,
-      "learning_rate": 0.00019999987706800372,
-      "loss": 46.0,
-      "step": 6538
-    },
-    {
-      "epoch": 0.4999522143853814,
-      "grad_norm": 0.00200356962159276,
-      "learning_rate": 0.00019999987703033783,
-      "loss": 46.0,
-      "step": 6539
-    },
-    {
-      "epoch": 0.5000286713687712,
-      "grad_norm": 0.0008327013929374516,
-      "learning_rate": 0.00019999987699266623,
-      "loss": 46.0,
-      "step": 6540
-    },
-    {
-      "epoch": 0.5001051283521608,
-      "grad_norm": 0.0006880591390654445,
-      "learning_rate": 0.00019999987695498882,
-      "loss": 46.0,
-      "step": 6541
-    },
-    {
-      "epoch": 0.5001815853355506,
-      "grad_norm": 0.0006306835566647351,
-      "learning_rate": 0.0001999998769173056,
-      "loss": 46.0,
-      "step": 6542
-    },
-    {
-      "epoch": 0.5002580423189403,
-      "grad_norm": 0.0007143211551010609,
-      "learning_rate": 0.0001999998768796167,
-      "loss": 46.0,
-      "step": 6543
-    },
-    {
-      "epoch": 0.50033449930233,
-      "grad_norm": 0.0011272953124716878,
-      "learning_rate": 0.00019999987684192195,
-      "loss": 46.0,
-      "step": 6544
-    },
-    {
-      "epoch": 0.5004109562857197,
-      "grad_norm": 0.0015187429962679744,
-      "learning_rate": 0.0001999998768042215,
-      "loss": 46.0,
-      "step": 6545
-    },
-    {
-      "epoch": 0.5004874132691095,
-      "grad_norm": 0.0007690527127124369,
-      "learning_rate": 0.00019999987676651523,
-      "loss": 46.0,
-      "step": 6546
-    },
-    {
-      "epoch": 0.5005638702524992,
-      "grad_norm": 0.0015275331679731607,
-      "learning_rate": 0.0001999998767288032,
-      "loss": 46.0,
-      "step": 6547
-    },
-    {
-      "epoch": 0.5006403272358889,
-      "grad_norm": 0.003748373594135046,
-      "learning_rate": 0.00019999987669108542,
-      "loss": 46.0,
-      "step": 6548
-    },
-    {
-      "epoch": 0.5007167842192787,
-      "grad_norm": 0.0017606825567781925,
-      "learning_rate": 0.00019999987665336185,
-      "loss": 46.0,
-      "step": 6549
-    },
-    {
-      "epoch": 0.5007932412026683,
-      "grad_norm": 0.00184637529309839,
-      "learning_rate": 0.00019999987661563254,
-      "loss": 46.0,
-      "step": 6550
-    },
-    {
-      "epoch": 0.5008696981860581,
-      "grad_norm": 0.0025023254565894604,
-      "learning_rate": 0.00019999987657789742,
-      "loss": 46.0,
-      "step": 6551
-    },
-    {
-      "epoch": 0.5009461551694477,
-      "grad_norm": 0.0009148131939582527,
-      "learning_rate": 0.00019999987654015656,
-      "loss": 46.0,
-      "step": 6552
-    },
-    {
-      "epoch": 0.5010226121528375,
-      "grad_norm": 0.0003834763483610004,
-      "learning_rate": 0.00019999987650240995,
-      "loss": 46.0,
-      "step": 6553
-    },
-    {
-      "epoch": 0.5010990691362273,
-      "grad_norm": 0.0009837960824370384,
-      "learning_rate": 0.00019999987646465752,
-      "loss": 46.0,
-      "step": 6554
-    },
-    {
-      "epoch": 0.5011755261196169,
-      "grad_norm": 0.0011544080916792154,
-      "learning_rate": 0.00019999987642689934,
-      "loss": 46.0,
-      "step": 6555
-    },
-    {
-      "epoch": 0.5012519831030067,
-      "grad_norm": 0.0033667783718556166,
-      "learning_rate": 0.0001999998763891354,
-      "loss": 46.0,
-      "step": 6556
-    },
-    {
-      "epoch": 0.5013284400863964,
-      "grad_norm": 0.0022614910267293453,
-      "learning_rate": 0.00019999987635136568,
-      "loss": 46.0,
-      "step": 6557
-    },
-    {
-      "epoch": 0.5014048970697861,
-      "grad_norm": 0.0014761054189875722,
-      "learning_rate": 0.00019999987631359018,
-      "loss": 46.0,
-      "step": 6558
-    },
-    {
-      "epoch": 0.5014813540531758,
-      "grad_norm": 0.00044820777839049697,
-      "learning_rate": 0.00019999987627580893,
-      "loss": 46.0,
-      "step": 6559
-    },
-    {
-      "epoch": 0.5015578110365656,
-      "grad_norm": 0.0009060293668881059,
-      "learning_rate": 0.0001999998762380219,
-      "loss": 46.0,
-      "step": 6560
-    },
-    {
-      "epoch": 0.5016342680199553,
-      "grad_norm": 0.002454856876283884,
-      "learning_rate": 0.00019999987620022912,
-      "loss": 46.0,
-      "step": 6561
-    },
-    {
-      "epoch": 0.501710725003345,
-      "grad_norm": 0.0029269622173160315,
-      "learning_rate": 0.00019999987616243058,
-      "loss": 46.0,
-      "step": 6562
-    },
-    {
-      "epoch": 0.5017871819867347,
-      "grad_norm": 0.0012639738852158189,
-      "learning_rate": 0.00019999987612462624,
-      "loss": 46.0,
-      "step": 6563
-    },
-    {
-      "epoch": 0.5018636389701244,
-      "grad_norm": 0.0018493705429136753,
-      "learning_rate": 0.00019999987608681612,
-      "loss": 46.0,
-      "step": 6564
-    },
-    {
-      "epoch": 0.5019400959535142,
-      "grad_norm": 0.0008241417235694826,
-      "learning_rate": 0.00019999987604900026,
-      "loss": 46.0,
-      "step": 6565
-    },
-    {
-      "epoch": 0.5020165529369038,
-      "grad_norm": 0.0014339416520670056,
-      "learning_rate": 0.00019999987601117863,
-      "loss": 46.0,
-      "step": 6566
-    },
-    {
-      "epoch": 0.5020930099202936,
-      "grad_norm": 0.0027079368010163307,
-      "learning_rate": 0.0001999998759733512,
-      "loss": 46.0,
-      "step": 6567
-    },
-    {
-      "epoch": 0.5021694669036834,
-      "grad_norm": 0.02328592538833618,
-      "learning_rate": 0.00019999987593551804,
-      "loss": 46.0,
-      "step": 6568
-    },
-    {
-      "epoch": 0.502245923887073,
-      "grad_norm": 0.0017986586317420006,
-      "learning_rate": 0.0001999998758976791,
-      "loss": 46.0,
-      "step": 6569
-    },
-    {
-      "epoch": 0.5023223808704628,
-      "grad_norm": 0.0018942091846838593,
-      "learning_rate": 0.00019999987585983436,
-      "loss": 46.0,
-      "step": 6570
-    },
-    {
-      "epoch": 0.5023988378538525,
-      "grad_norm": 0.002628538990393281,
-      "learning_rate": 0.00019999987582198386,
-      "loss": 46.0,
-      "step": 6571
-    },
-    {
-      "epoch": 0.5024752948372422,
-      "grad_norm": 0.00043532688869163394,
-      "learning_rate": 0.0001999998757841276,
-      "loss": 46.0,
-      "step": 6572
-    },
-    {
-      "epoch": 0.5025517518206319,
-      "grad_norm": 0.001176826423034072,
-      "learning_rate": 0.0001999998757462656,
-      "loss": 46.0,
-      "step": 6573
-    },
-    {
-      "epoch": 0.5026282088040216,
-      "grad_norm": 0.0012792424531653523,
-      "learning_rate": 0.0001999998757083978,
-      "loss": 46.0,
-      "step": 6574
-    },
-    {
-      "epoch": 0.5027046657874114,
-      "grad_norm": 0.0014877606881782413,
-      "learning_rate": 0.00019999987567052423,
-      "loss": 46.0,
-      "step": 6575
-    },
-    {
-      "epoch": 0.5027811227708011,
-      "grad_norm": 0.0019944391679018736,
-      "learning_rate": 0.0001999998756326449,
-      "loss": 46.0,
-      "step": 6576
-    },
-    {
-      "epoch": 0.5028575797541908,
-      "grad_norm": 0.000718877709005028,
-      "learning_rate": 0.00019999987559475978,
-      "loss": 46.0,
-      "step": 6577
-    },
-    {
-      "epoch": 0.5029340367375805,
-      "grad_norm": 0.0034903271589428186,
-      "learning_rate": 0.0001999998755568689,
-      "loss": 46.0,
-      "step": 6578
-    },
-    {
-      "epoch": 0.5030104937209703,
-      "grad_norm": 0.0020239094737917185,
-      "learning_rate": 0.00019999987551897226,
-      "loss": 46.0,
-      "step": 6579
-    },
-    {
-      "epoch": 0.5030869507043599,
-      "grad_norm": 0.0020184360910207033,
-      "learning_rate": 0.00019999987548106985,
-      "loss": 46.0,
-      "step": 6580
-    },
-    {
-      "epoch": 0.5031634076877497,
-      "grad_norm": 0.000886824622284621,
-      "learning_rate": 0.00019999987544316164,
-      "loss": 46.0,
-      "step": 6581
-    },
-    {
-      "epoch": 0.5032398646711393,
-      "grad_norm": 0.001010234234854579,
-      "learning_rate": 0.00019999987540524772,
-      "loss": 46.0,
-      "step": 6582
-    },
-    {
-      "epoch": 0.5033163216545291,
-      "grad_norm": 0.0035800074692815542,
-      "learning_rate": 0.00019999987536732802,
-      "loss": 46.0,
-      "step": 6583
-    },
-    {
-      "epoch": 0.5033927786379189,
-      "grad_norm": 0.008436190895736217,
-      "learning_rate": 0.0001999998753294025,
-      "loss": 46.0,
-      "step": 6584
-    },
-    {
-      "epoch": 0.5034692356213085,
-      "grad_norm": 0.0009393159998580813,
-      "learning_rate": 0.00019999987529147124,
-      "loss": 46.0,
-      "step": 6585
-    },
-    {
-      "epoch": 0.5035456926046983,
-      "grad_norm": 0.0007460570195689797,
-      "learning_rate": 0.00019999987525353422,
-      "loss": 46.0,
-      "step": 6586
-    },
-    {
-      "epoch": 0.503622149588088,
-      "grad_norm": 0.000494759064167738,
-      "learning_rate": 0.0001999998752155914,
-      "loss": 46.0,
-      "step": 6587
-    },
-    {
-      "epoch": 0.5036986065714777,
-      "grad_norm": 0.0031264338176697493,
-      "learning_rate": 0.00019999987517764286,
-      "loss": 46.0,
-      "step": 6588
-    },
-    {
-      "epoch": 0.5037750635548675,
-      "grad_norm": 0.0006070621311664581,
-      "learning_rate": 0.0001999998751396885,
-      "loss": 46.0,
-      "step": 6589
-    },
-    {
-      "epoch": 0.5038515205382572,
-      "grad_norm": 0.0008838371140882373,
-      "learning_rate": 0.0001999998751017284,
-      "loss": 46.0,
-      "step": 6590
-    },
-    {
-      "epoch": 0.5039279775216469,
-      "grad_norm": 0.0017949964385479689,
-      "learning_rate": 0.00019999987506376252,
-      "loss": 46.0,
-      "step": 6591
-    },
-    {
-      "epoch": 0.5040044345050366,
-      "grad_norm": 0.002289183670654893,
-      "learning_rate": 0.00019999987502579086,
-      "loss": 46.0,
-      "step": 6592
-    },
-    {
-      "epoch": 0.5040808914884263,
-      "grad_norm": 0.0007575471536256373,
-      "learning_rate": 0.00019999987498781345,
-      "loss": 46.0,
-      "step": 6593
-    },
-    {
-      "epoch": 0.504157348471816,
-      "grad_norm": 0.0010387235088273883,
-      "learning_rate": 0.00019999987494983027,
-      "loss": 46.0,
-      "step": 6594
-    },
-    {
-      "epoch": 0.5042338054552058,
-      "grad_norm": 0.0005088611505925655,
-      "learning_rate": 0.00019999987491184132,
-      "loss": 46.0,
-      "step": 6595
-    },
-    {
-      "epoch": 0.5043102624385954,
-      "grad_norm": 0.0009044331382028759,
-      "learning_rate": 0.00019999987487384657,
-      "loss": 46.0,
-      "step": 6596
-    },
-    {
-      "epoch": 0.5043867194219852,
-      "grad_norm": 0.0005453468766063452,
-      "learning_rate": 0.00019999987483584607,
-      "loss": 46.0,
-      "step": 6597
-    },
-    {
-      "epoch": 0.504463176405375,
-      "grad_norm": 0.0006305858842097223,
-      "learning_rate": 0.00019999987479783982,
-      "loss": 46.0,
-      "step": 6598
-    },
-    {
-      "epoch": 0.5045396333887646,
-      "grad_norm": 0.0025635375641286373,
-      "learning_rate": 0.0001999998747598278,
-      "loss": 46.0,
-      "step": 6599
-    },
-    {
-      "epoch": 0.5046160903721544,
-      "grad_norm": 0.001269631553441286,
-      "learning_rate": 0.00019999987472181,
-      "loss": 46.0,
-      "step": 6600
-    },
-    {
-      "epoch": 0.5046925473555441,
-      "grad_norm": 0.005238671321421862,
-      "learning_rate": 0.0001999998746837864,
-      "loss": 46.0,
-      "step": 6601
-    },
-    {
-      "epoch": 0.5047690043389338,
-      "grad_norm": 0.001603403128683567,
-      "learning_rate": 0.00019999987464575707,
-      "loss": 46.0,
-      "step": 6602
-    },
-    {
-      "epoch": 0.5048454613223236,
-      "grad_norm": 0.0021633345168083906,
-      "learning_rate": 0.000199999874607722,
-      "loss": 46.0,
-      "step": 6603
-    },
-    {
-      "epoch": 0.5049219183057132,
-      "grad_norm": 0.001273146946914494,
-      "learning_rate": 0.00019999987456968107,
-      "loss": 46.0,
-      "step": 6604
-    },
-    {
-      "epoch": 0.504998375289103,
-      "grad_norm": 0.0006083719781599939,
-      "learning_rate": 0.00019999987453163444,
-      "loss": 46.0,
-      "step": 6605
-    },
-    {
-      "epoch": 0.5050748322724927,
-      "grad_norm": 0.0014974843943491578,
-      "learning_rate": 0.000199999874493582,
-      "loss": 46.0,
-      "step": 6606
-    },
-    {
-      "epoch": 0.5051512892558824,
-      "grad_norm": 0.0010533076710999012,
-      "learning_rate": 0.00019999987445552383,
-      "loss": 46.0,
-      "step": 6607
-    },
-    {
-      "epoch": 0.5052277462392721,
-      "grad_norm": 0.0009454118553549051,
-      "learning_rate": 0.00019999987441745987,
-      "loss": 46.0,
-      "step": 6608
-    },
-    {
-      "epoch": 0.5053042032226619,
-      "grad_norm": 0.002649895614013076,
-      "learning_rate": 0.00019999987437939015,
-      "loss": 46.0,
-      "step": 6609
-    },
-    {
-      "epoch": 0.5053806602060515,
-      "grad_norm": 0.0005589655484072864,
-      "learning_rate": 0.00019999987434131465,
-      "loss": 46.0,
-      "step": 6610
-    },
-    {
-      "epoch": 0.5054571171894413,
-      "grad_norm": 0.002897490980103612,
-      "learning_rate": 0.00019999987430323338,
-      "loss": 46.0,
-      "step": 6611
-    },
-    {
-      "epoch": 0.505533574172831,
-      "grad_norm": 0.003157546278089285,
-      "learning_rate": 0.00019999987426514636,
-      "loss": 46.0,
-      "step": 6612
-    },
-    {
-      "epoch": 0.5056100311562207,
-      "grad_norm": 0.001781234983354807,
-      "learning_rate": 0.00019999987422705356,
-      "loss": 46.0,
-      "step": 6613
-    },
-    {
-      "epoch": 0.5056864881396105,
-      "grad_norm": 0.002597308252006769,
-      "learning_rate": 0.00019999987418895497,
-      "loss": 46.0,
-      "step": 6614
-    },
-    {
-      "epoch": 0.5057629451230001,
-      "grad_norm": 0.002923625288531184,
-      "learning_rate": 0.00019999987415085063,
-      "loss": 46.0,
-      "step": 6615
-    },
-    {
-      "epoch": 0.5058394021063899,
-      "grad_norm": 0.00042659268365241587,
-      "learning_rate": 0.00019999987411274052,
-      "loss": 46.0,
-      "step": 6616
-    },
-    {
-      "epoch": 0.5059158590897797,
-      "grad_norm": 0.0015453496016561985,
-      "learning_rate": 0.00019999987407462463,
-      "loss": 46.0,
-      "step": 6617
-    },
-    {
-      "epoch": 0.5059923160731693,
-      "grad_norm": 0.007973256520926952,
-      "learning_rate": 0.00019999987403650297,
-      "loss": 46.0,
-      "step": 6618
-    },
-    {
-      "epoch": 0.5060687730565591,
-      "grad_norm": 0.000626304536126554,
-      "learning_rate": 0.00019999987399837554,
-      "loss": 46.0,
-      "step": 6619
-    },
-    {
-      "epoch": 0.5061452300399488,
-      "grad_norm": 0.0013627674197778106,
-      "learning_rate": 0.00019999987396024236,
-      "loss": 46.0,
-      "step": 6620
-    },
-    {
-      "epoch": 0.5062216870233385,
-      "grad_norm": 0.0025127700064331293,
-      "learning_rate": 0.0001999998739221034,
-      "loss": 46.0,
-      "step": 6621
-    },
-    {
-      "epoch": 0.5062981440067282,
-      "grad_norm": 0.0005612922250293195,
-      "learning_rate": 0.00019999987388395865,
-      "loss": 46.0,
-      "step": 6622
-    },
-    {
-      "epoch": 0.5063746009901179,
-      "grad_norm": 0.004672839306294918,
-      "learning_rate": 0.00019999987384580815,
-      "loss": 46.0,
-      "step": 6623
-    },
-    {
-      "epoch": 0.5064510579735076,
-      "grad_norm": 0.0015346667496487498,
-      "learning_rate": 0.0001999998738076519,
-      "loss": 46.0,
-      "step": 6624
-    },
-    {
-      "epoch": 0.5065275149568974,
-      "grad_norm": 0.006579149980098009,
-      "learning_rate": 0.0001999998737694899,
-      "loss": 46.0,
-      "step": 6625
-    },
-    {
-      "epoch": 0.5066039719402871,
-      "grad_norm": 0.000537071842700243,
-      "learning_rate": 0.00019999987373132204,
-      "loss": 46.0,
-      "step": 6626
-    },
-    {
-      "epoch": 0.5066804289236768,
-      "grad_norm": 0.0015940542798489332,
-      "learning_rate": 0.00019999987369314848,
-      "loss": 46.0,
-      "step": 6627
-    },
-    {
-      "epoch": 0.5067568859070666,
-      "grad_norm": 0.0004707959305960685,
-      "learning_rate": 0.00019999987365496914,
-      "loss": 46.0,
-      "step": 6628
-    },
-    {
-      "epoch": 0.5068333428904562,
-      "grad_norm": 0.0019824965856969357,
-      "learning_rate": 0.00019999987361678402,
-      "loss": 46.0,
-      "step": 6629
-    },
-    {
-      "epoch": 0.506909799873846,
-      "grad_norm": 0.0037892009131610394,
-      "learning_rate": 0.00019999987357859314,
-      "loss": 46.0,
-      "step": 6630
-    },
-    {
-      "epoch": 0.5069862568572358,
-      "grad_norm": 0.003043479286134243,
-      "learning_rate": 0.00019999987354039648,
-      "loss": 46.0,
-      "step": 6631
-    },
-    {
-      "epoch": 0.5070627138406254,
-      "grad_norm": 0.0020136937964707613,
-      "learning_rate": 0.00019999987350219405,
-      "loss": 46.0,
-      "step": 6632
-    },
-    {
-      "epoch": 0.5071391708240152,
-      "grad_norm": 0.0012219158234074712,
-      "learning_rate": 0.00019999987346398587,
-      "loss": 46.0,
-      "step": 6633
-    },
-    {
-      "epoch": 0.5072156278074048,
-      "grad_norm": 0.0006614429294131696,
-      "learning_rate": 0.0001999998734257719,
-      "loss": 46.0,
-      "step": 6634
-    },
-    {
-      "epoch": 0.5072920847907946,
-      "grad_norm": 0.0023321930784732103,
-      "learning_rate": 0.00019999987338755216,
-      "loss": 46.0,
-      "step": 6635
-    },
-    {
-      "epoch": 0.5073685417741843,
-      "grad_norm": 0.0014058776432648301,
-      "learning_rate": 0.00019999987334932666,
-      "loss": 46.0,
-      "step": 6636
-    },
-    {
-      "epoch": 0.507444998757574,
-      "grad_norm": 0.0011735991574823856,
-      "learning_rate": 0.0001999998733110954,
-      "loss": 46.0,
-      "step": 6637
-    },
-    {
-      "epoch": 0.5075214557409637,
-      "grad_norm": 0.006963454186916351,
-      "learning_rate": 0.00019999987327285834,
-      "loss": 46.0,
-      "step": 6638
-    },
-    {
-      "epoch": 0.5075979127243535,
-      "grad_norm": 0.005320186261087656,
-      "learning_rate": 0.00019999987323461552,
-      "loss": 46.0,
-      "step": 6639
-    },
-    {
-      "epoch": 0.5076743697077432,
-      "grad_norm": 0.0020673151593655348,
-      "learning_rate": 0.00019999987319636696,
-      "loss": 46.0,
-      "step": 6640
-    },
-    {
-      "epoch": 0.5077508266911329,
-      "grad_norm": 0.00519458157941699,
-      "learning_rate": 0.00019999987315811262,
-      "loss": 46.0,
-      "step": 6641
-    },
-    {
-      "epoch": 0.5078272836745226,
-      "grad_norm": 0.001650924445129931,
-      "learning_rate": 0.00019999987311985248,
-      "loss": 46.0,
-      "step": 6642
-    },
-    {
-      "epoch": 0.5079037406579123,
-      "grad_norm": 0.0034306340385228395,
-      "learning_rate": 0.0001999998730815866,
-      "loss": 46.0,
-      "step": 6643
-    },
-    {
-      "epoch": 0.5079801976413021,
-      "grad_norm": 0.006050763186067343,
-      "learning_rate": 0.00019999987304331494,
-      "loss": 46.0,
-      "step": 6644
-    },
-    {
-      "epoch": 0.5080566546246917,
-      "grad_norm": 0.0003902481694240123,
-      "learning_rate": 0.0001999998730050375,
-      "loss": 46.0,
-      "step": 6645
-    },
-    {
-      "epoch": 0.5081331116080815,
-      "grad_norm": 0.0015269792638719082,
-      "learning_rate": 0.0001999998729667543,
-      "loss": 46.0,
-      "step": 6646
-    },
-    {
-      "epoch": 0.5082095685914713,
-      "grad_norm": 0.0017523665446788073,
-      "learning_rate": 0.00019999987292846535,
-      "loss": 46.0,
-      "step": 6647
-    },
-    {
-      "epoch": 0.5082860255748609,
-      "grad_norm": 0.0006248591234907508,
-      "learning_rate": 0.00019999987289017062,
-      "loss": 46.0,
-      "step": 6648
-    },
-    {
-      "epoch": 0.5083624825582507,
-      "grad_norm": 0.0016106307739391923,
-      "learning_rate": 0.00019999987285187012,
-      "loss": 46.0,
-      "step": 6649
-    },
-    {
-      "epoch": 0.5084389395416404,
-      "grad_norm": 0.003220010083168745,
-      "learning_rate": 0.00019999987281356382,
-      "loss": 46.0,
-      "step": 6650
-    },
-    {
-      "epoch": 0.5085153965250301,
-      "grad_norm": 0.0288386307656765,
-      "learning_rate": 0.00019999987277525178,
-      "loss": 46.0,
-      "step": 6651
-    },
-    {
-      "epoch": 0.5085918535084198,
-      "grad_norm": 0.000946781481616199,
-      "learning_rate": 0.00019999987273693396,
-      "loss": 46.0,
-      "step": 6652
-    },
-    {
-      "epoch": 0.5086683104918095,
-      "grad_norm": 0.001764478045515716,
-      "learning_rate": 0.0001999998726986104,
-      "loss": 46.0,
-      "step": 6653
-    },
-    {
-      "epoch": 0.5087447674751993,
-      "grad_norm": 0.0014122177381068468,
-      "learning_rate": 0.00019999987266028105,
-      "loss": 46.0,
-      "step": 6654
-    },
-    {
-      "epoch": 0.508821224458589,
-      "grad_norm": 0.0006862186128273606,
-      "learning_rate": 0.00019999987262194591,
-      "loss": 46.0,
-      "step": 6655
-    },
-    {
-      "epoch": 0.5088976814419787,
-      "grad_norm": 0.0030939560383558273,
-      "learning_rate": 0.00019999987258360503,
-      "loss": 46.0,
-      "step": 6656
-    },
-    {
-      "epoch": 0.5089741384253684,
-      "grad_norm": 0.005594997201114893,
-      "learning_rate": 0.00019999987254525837,
-      "loss": 46.0,
-      "step": 6657
-    },
-    {
-      "epoch": 0.5090505954087582,
-      "grad_norm": 0.0016314440872520208,
-      "learning_rate": 0.00019999987250690594,
-      "loss": 46.0,
-      "step": 6658
-    },
-    {
-      "epoch": 0.5091270523921478,
-      "grad_norm": 0.0028863984625786543,
-      "learning_rate": 0.00019999987246854773,
-      "loss": 46.0,
-      "step": 6659
-    },
-    {
-      "epoch": 0.5092035093755376,
-      "grad_norm": 0.0007835583528503776,
-      "learning_rate": 0.00019999987243018375,
-      "loss": 46.0,
-      "step": 6660
-    },
-    {
-      "epoch": 0.5092799663589274,
-      "grad_norm": 0.002227367600426078,
-      "learning_rate": 0.00019999987239181403,
-      "loss": 46.0,
-      "step": 6661
-    },
-    {
-      "epoch": 0.509356423342317,
-      "grad_norm": 0.00223711971193552,
-      "learning_rate": 0.0001999998723534385,
-      "loss": 46.0,
-      "step": 6662
-    },
-    {
-      "epoch": 0.5094328803257068,
-      "grad_norm": 0.007872717455029488,
-      "learning_rate": 0.00019999987231505726,
-      "loss": 46.0,
-      "step": 6663
-    },
-    {
-      "epoch": 0.5095093373090964,
-      "grad_norm": 0.0018282405799254775,
-      "learning_rate": 0.00019999987227667019,
-      "loss": 46.0,
-      "step": 6664
-    },
-    {
-      "epoch": 0.5095857942924862,
-      "grad_norm": 0.014191199094057083,
-      "learning_rate": 0.0001999998722382774,
-      "loss": 46.0,
-      "step": 6665
-    },
-    {
-      "epoch": 0.5096622512758759,
-      "grad_norm": 0.001074605737812817,
-      "learning_rate": 0.0001999998721998788,
-      "loss": 46.0,
-      "step": 6666
-    },
-    {
-      "epoch": 0.5097387082592656,
-      "grad_norm": 0.0009614795562811196,
-      "learning_rate": 0.0001999998721614744,
-      "loss": 46.0,
-      "step": 6667
-    },
-    {
-      "epoch": 0.5098151652426554,
-      "grad_norm": 0.0006571012199856341,
-      "learning_rate": 0.0001999998721230643,
-      "loss": 46.0,
-      "step": 6668
-    },
-    {
-      "epoch": 0.5098916222260451,
-      "grad_norm": 0.004921938758343458,
-      "learning_rate": 0.0001999998720846484,
-      "loss": 46.0,
-      "step": 6669
-    },
-    {
-      "epoch": 0.5099680792094348,
-      "grad_norm": 0.0006776495720259845,
-      "learning_rate": 0.00019999987204622675,
-      "loss": 46.0,
-      "step": 6670
-    },
-    {
-      "epoch": 0.5100445361928245,
-      "grad_norm": 0.0012581656919792295,
-      "learning_rate": 0.00019999987200779932,
-      "loss": 46.0,
-      "step": 6671
-    },
-    {
-      "epoch": 0.5101209931762143,
-      "grad_norm": 0.004164766985923052,
-      "learning_rate": 0.00019999987196936612,
-      "loss": 46.0,
-      "step": 6672
-    },
-    {
-      "epoch": 0.5101974501596039,
-      "grad_norm": 0.0020659894216805696,
-      "learning_rate": 0.00019999987193092714,
-      "loss": 46.0,
-      "step": 6673
-    },
-    {
-      "epoch": 0.5102739071429937,
-      "grad_norm": 0.0005768091650679708,
-      "learning_rate": 0.0001999998718924824,
-      "loss": 46.0,
-      "step": 6674
-    },
-    {
-      "epoch": 0.5103503641263833,
-      "grad_norm": 0.003232197370380163,
-      "learning_rate": 0.00019999987185403192,
-      "loss": 46.0,
-      "step": 6675
-    },
-    {
-      "epoch": 0.5104268211097731,
-      "grad_norm": 0.0008823472890071571,
-      "learning_rate": 0.0001999998718155756,
-      "loss": 46.0,
-      "step": 6676
-    },
-    {
-      "epoch": 0.5105032780931629,
-      "grad_norm": 0.0012526483042165637,
-      "learning_rate": 0.00019999987177711357,
-      "loss": 46.0,
-      "step": 6677
-    },
-    {
-      "epoch": 0.5105797350765525,
-      "grad_norm": 0.003952939994633198,
-      "learning_rate": 0.00019999987173864575,
-      "loss": 46.0,
-      "step": 6678
-    },
-    {
-      "epoch": 0.5106561920599423,
-      "grad_norm": 0.0011536605888977647,
-      "learning_rate": 0.00019999987170017216,
-      "loss": 46.0,
-      "step": 6679
-    },
-    {
-      "epoch": 0.510732649043332,
-      "grad_norm": 0.0005209247465245426,
-      "learning_rate": 0.0001999998716616928,
-      "loss": 46.0,
-      "step": 6680
-    },
-    {
-      "epoch": 0.5108091060267217,
-      "grad_norm": 0.000934875919483602,
-      "learning_rate": 0.00019999987162320766,
-      "loss": 46.0,
-      "step": 6681
-    },
-    {
-      "epoch": 0.5108855630101115,
-      "grad_norm": 0.0007178029045462608,
-      "learning_rate": 0.00019999987158471675,
-      "loss": 46.0,
-      "step": 6682
-    },
-    {
-      "epoch": 0.5109620199935011,
-      "grad_norm": 0.004913387820124626,
-      "learning_rate": 0.00019999987154622012,
-      "loss": 46.0,
-      "step": 6683
-    },
-    {
-      "epoch": 0.5110384769768909,
-      "grad_norm": 0.0005487127928063273,
-      "learning_rate": 0.00019999987150771768,
-      "loss": 46.0,
-      "step": 6684
-    },
-    {
-      "epoch": 0.5111149339602806,
-      "grad_norm": 0.00388889922760427,
-      "learning_rate": 0.00019999987146920945,
-      "loss": 46.0,
-      "step": 6685
-    },
-    {
-      "epoch": 0.5111913909436703,
-      "grad_norm": 0.0010851608822122216,
-      "learning_rate": 0.0001999998714306955,
-      "loss": 46.0,
-      "step": 6686
-    },
-    {
-      "epoch": 0.51126784792706,
-      "grad_norm": 0.0006419922574423254,
-      "learning_rate": 0.00019999987139217575,
-      "loss": 46.0,
-      "step": 6687
-    },
-    {
-      "epoch": 0.5113443049104498,
-      "grad_norm": 0.0006653190357610583,
-      "learning_rate": 0.00019999987135365022,
-      "loss": 46.0,
-      "step": 6688
-    },
-    {
-      "epoch": 0.5114207618938394,
-      "grad_norm": 0.0041973283514380455,
-      "learning_rate": 0.00019999987131511895,
-      "loss": 46.0,
-      "step": 6689
-    },
-    {
-      "epoch": 0.5114972188772292,
-      "grad_norm": 0.0015674334717914462,
-      "learning_rate": 0.0001999998712765819,
-      "loss": 46.0,
-      "step": 6690
-    },
-    {
-      "epoch": 0.511573675860619,
-      "grad_norm": 0.0006393588846549392,
-      "learning_rate": 0.00019999987123803906,
-      "loss": 46.0,
-      "step": 6691
-    },
-    {
-      "epoch": 0.5116501328440086,
-      "grad_norm": 0.0014182829763740301,
-      "learning_rate": 0.00019999987119949047,
-      "loss": 46.0,
-      "step": 6692
-    },
-    {
-      "epoch": 0.5117265898273984,
-      "grad_norm": 0.0005151458899490535,
-      "learning_rate": 0.00019999987116093613,
-      "loss": 46.0,
-      "step": 6693
-    },
-    {
-      "epoch": 0.511803046810788,
-      "grad_norm": 0.0007883227663114667,
-      "learning_rate": 0.000199999871122376,
-      "loss": 46.0,
-      "step": 6694
-    },
-    {
-      "epoch": 0.5118795037941778,
-      "grad_norm": 0.0020876929629594088,
-      "learning_rate": 0.0001999998710838101,
-      "loss": 46.0,
-      "step": 6695
-    },
-    {
-      "epoch": 0.5119559607775676,
-      "grad_norm": 0.0011576981050893664,
-      "learning_rate": 0.00019999987104523843,
-      "loss": 46.0,
-      "step": 6696
-    },
-    {
-      "epoch": 0.5120324177609572,
-      "grad_norm": 0.006071015726774931,
-      "learning_rate": 0.00019999987100666097,
-      "loss": 46.0,
-      "step": 6697
-    },
-    {
-      "epoch": 0.512108874744347,
-      "grad_norm": 0.0023508064914494753,
-      "learning_rate": 0.00019999987096807776,
-      "loss": 46.0,
-      "step": 6698
-    },
-    {
-      "epoch": 0.5121853317277367,
-      "grad_norm": 0.011246114037930965,
-      "learning_rate": 0.00019999987092948879,
-      "loss": 46.0,
-      "step": 6699
-    },
-    {
-      "epoch": 0.5122617887111264,
-      "grad_norm": 0.0011908836895599961,
-      "learning_rate": 0.00019999987089089406,
-      "loss": 46.0,
-      "step": 6700
-    },
-    {
-      "epoch": 0.5123382456945161,
-      "grad_norm": 0.0008464244892820716,
-      "learning_rate": 0.00019999987085229354,
-      "loss": 46.0,
-      "step": 6701
-    },
-    {
-      "epoch": 0.5124147026779059,
-      "grad_norm": 0.0011266465298831463,
-      "learning_rate": 0.00019999987081368724,
-      "loss": 46.0,
-      "step": 6702
-    },
-    {
-      "epoch": 0.5124911596612955,
-      "grad_norm": 0.0008307049283757806,
-      "learning_rate": 0.0001999998707750752,
-      "loss": 46.0,
-      "step": 6703
-    },
-    {
-      "epoch": 0.5125676166446853,
-      "grad_norm": 0.0011249450035393238,
-      "learning_rate": 0.00019999987073645738,
-      "loss": 46.0,
-      "step": 6704
-    },
-    {
-      "epoch": 0.512644073628075,
-      "grad_norm": 0.0010429281974211335,
-      "learning_rate": 0.00019999987069783379,
-      "loss": 46.0,
-      "step": 6705
-    },
-    {
-      "epoch": 0.5127205306114647,
-      "grad_norm": 0.0006533378036692739,
-      "learning_rate": 0.00019999987065920442,
-      "loss": 46.0,
-      "step": 6706
-    },
-    {
-      "epoch": 0.5127969875948545,
-      "grad_norm": 0.0013591040624305606,
-      "learning_rate": 0.00019999987062056928,
-      "loss": 46.0,
-      "step": 6707
-    },
-    {
-      "epoch": 0.5128734445782441,
-      "grad_norm": 0.002953270450234413,
-      "learning_rate": 0.00019999987058192837,
-      "loss": 46.0,
-      "step": 6708
-    },
-    {
-      "epoch": 0.5129499015616339,
-      "grad_norm": 0.004673848859965801,
-      "learning_rate": 0.00019999987054328172,
-      "loss": 46.0,
-      "step": 6709
-    },
-    {
-      "epoch": 0.5130263585450237,
-      "grad_norm": 0.001719867461360991,
-      "learning_rate": 0.00019999987050462928,
-      "loss": 46.0,
-      "step": 6710
-    },
-    {
-      "epoch": 0.5131028155284133,
-      "grad_norm": 0.00045337225310504436,
-      "learning_rate": 0.00019999987046597105,
-      "loss": 46.0,
-      "step": 6711
-    },
-    {
-      "epoch": 0.5131792725118031,
-      "grad_norm": 0.0038524160627275705,
-      "learning_rate": 0.00019999987042730708,
-      "loss": 46.0,
-      "step": 6712
-    },
-    {
-      "epoch": 0.5132557294951927,
-      "grad_norm": 0.001208741799928248,
-      "learning_rate": 0.00019999987038863732,
-      "loss": 46.0,
-      "step": 6713
-    },
-    {
-      "epoch": 0.5133321864785825,
-      "grad_norm": 0.005358150694519281,
-      "learning_rate": 0.0001999998703499618,
-      "loss": 46.0,
-      "step": 6714
-    },
-    {
-      "epoch": 0.5134086434619722,
-      "grad_norm": 0.0007636166992597282,
-      "learning_rate": 0.0001999998703112805,
-      "loss": 46.0,
-      "step": 6715
-    },
-    {
-      "epoch": 0.5134851004453619,
-      "grad_norm": 0.0011872861068695784,
-      "learning_rate": 0.00019999987027259346,
-      "loss": 46.0,
-      "step": 6716
-    },
-    {
-      "epoch": 0.5135615574287516,
-      "grad_norm": 0.0014966072048991919,
-      "learning_rate": 0.00019999987023390064,
-      "loss": 46.0,
-      "step": 6717
-    },
-    {
-      "epoch": 0.5136380144121414,
-      "grad_norm": 0.0007914376328699291,
-      "learning_rate": 0.00019999987019520202,
-      "loss": 46.0,
-      "step": 6718
-    },
-    {
-      "epoch": 0.5137144713955311,
-      "grad_norm": 0.003805845510214567,
-      "learning_rate": 0.00019999987015649766,
-      "loss": 46.0,
-      "step": 6719
-    },
-    {
-      "epoch": 0.5137909283789208,
-      "grad_norm": 0.0015213243896141648,
-      "learning_rate": 0.00019999987011778752,
-      "loss": 46.0,
-      "step": 6720
-    },
-    {
-      "epoch": 0.5138673853623106,
-      "grad_norm": 0.0033187270164489746,
-      "learning_rate": 0.00019999987007907164,
-      "loss": 46.0,
-      "step": 6721
-    },
-    {
-      "epoch": 0.5139438423457002,
-      "grad_norm": 0.000449590414064005,
-      "learning_rate": 0.00019999987004034995,
-      "loss": 46.0,
-      "step": 6722
-    },
-    {
-      "epoch": 0.51402029932909,
-      "grad_norm": 0.0016383439069613814,
-      "learning_rate": 0.0001999998700016225,
-      "loss": 46.0,
-      "step": 6723
-    },
-    {
-      "epoch": 0.5140967563124796,
-      "grad_norm": 0.0012532379478216171,
-      "learning_rate": 0.0001999998699628893,
-      "loss": 46.0,
-      "step": 6724
-    },
-    {
-      "epoch": 0.5141732132958694,
-      "grad_norm": 0.0015614822041243315,
-      "learning_rate": 0.00019999986992415031,
-      "loss": 46.0,
-      "step": 6725
-    },
-    {
-      "epoch": 0.5142496702792592,
-      "grad_norm": 0.003305091056972742,
-      "learning_rate": 0.00019999986988540556,
-      "loss": 46.0,
-      "step": 6726
-    },
-    {
-      "epoch": 0.5143261272626488,
-      "grad_norm": 0.0012602624483406544,
-      "learning_rate": 0.000199999869846655,
-      "loss": 46.0,
-      "step": 6727
-    },
-    {
-      "epoch": 0.5144025842460386,
-      "grad_norm": 0.0013412623666226864,
-      "learning_rate": 0.00019999986980789874,
-      "loss": 46.0,
-      "step": 6728
-    },
-    {
-      "epoch": 0.5144790412294283,
-      "grad_norm": 0.0009423950687050819,
-      "learning_rate": 0.00019999986976913667,
-      "loss": 46.0,
-      "step": 6729
-    },
-    {
-      "epoch": 0.514555498212818,
-      "grad_norm": 0.002963624894618988,
-      "learning_rate": 0.00019999986973036883,
-      "loss": 46.0,
-      "step": 6730
-    },
-    {
-      "epoch": 0.5146319551962077,
-      "grad_norm": 0.0029351431876420975,
-      "learning_rate": 0.00019999986969159526,
-      "loss": 46.0,
-      "step": 6731
-    },
-    {
-      "epoch": 0.5147084121795975,
-      "grad_norm": 0.0017759987385943532,
-      "learning_rate": 0.0001999998696528159,
-      "loss": 46.0,
-      "step": 6732
-    },
-    {
-      "epoch": 0.5147848691629872,
-      "grad_norm": 0.0011451707687228918,
-      "learning_rate": 0.00019999986961403074,
-      "loss": 46.0,
-      "step": 6733
-    },
-    {
-      "epoch": 0.5148613261463769,
-      "grad_norm": 0.0007711068028584123,
-      "learning_rate": 0.00019999986957523982,
-      "loss": 46.0,
-      "step": 6734
-    },
-    {
-      "epoch": 0.5149377831297666,
-      "grad_norm": 0.001106961746700108,
-      "learning_rate": 0.00019999986953644317,
-      "loss": 46.0,
-      "step": 6735
-    },
-    {
-      "epoch": 0.5150142401131563,
-      "grad_norm": 0.0007789528463035822,
-      "learning_rate": 0.0001999998694976407,
-      "loss": 46.0,
-      "step": 6736
-    },
-    {
-      "epoch": 0.5150906970965461,
-      "grad_norm": 0.0007508696871809661,
-      "learning_rate": 0.00019999986945883248,
-      "loss": 46.0,
-      "step": 6737
-    },
-    {
-      "epoch": 0.5151671540799357,
-      "grad_norm": 0.0011181732406839728,
-      "learning_rate": 0.0001999998694200185,
-      "loss": 46.0,
-      "step": 6738
-    },
-    {
-      "epoch": 0.5152436110633255,
-      "grad_norm": 0.007998952642083168,
-      "learning_rate": 0.00019999986938119878,
-      "loss": 46.0,
-      "step": 6739
-    },
-    {
-      "epoch": 0.5153200680467153,
-      "grad_norm": 0.0065696947276592255,
-      "learning_rate": 0.00019999986934237323,
-      "loss": 46.0,
-      "step": 6740
-    },
-    {
-      "epoch": 0.5153965250301049,
-      "grad_norm": 0.007332746405154467,
-      "learning_rate": 0.00019999986930354193,
-      "loss": 46.0,
-      "step": 6741
-    },
-    {
-      "epoch": 0.5154729820134947,
-      "grad_norm": 0.0007144419359974563,
-      "learning_rate": 0.00019999986926470486,
-      "loss": 46.0,
-      "step": 6742
-    },
-    {
-      "epoch": 0.5155494389968843,
-      "grad_norm": 0.0007409803802147508,
-      "learning_rate": 0.00019999986922586204,
-      "loss": 46.0,
-      "step": 6743
-    },
-    {
-      "epoch": 0.5156258959802741,
-      "grad_norm": 0.0044411346316337585,
-      "learning_rate": 0.00019999986918701343,
-      "loss": 46.0,
-      "step": 6744
-    },
-    {
-      "epoch": 0.5157023529636638,
-      "grad_norm": 0.0035553663037717342,
-      "learning_rate": 0.00019999986914815904,
-      "loss": 46.0,
-      "step": 6745
-    },
-    {
-      "epoch": 0.5157788099470535,
-      "grad_norm": 0.0007410652469843626,
-      "learning_rate": 0.00019999986910929893,
-      "loss": 46.0,
-      "step": 6746
-    },
-    {
-      "epoch": 0.5158552669304433,
-      "grad_norm": 0.0015842923894524574,
-      "learning_rate": 0.000199999869070433,
-      "loss": 46.0,
-      "step": 6747
-    },
-    {
-      "epoch": 0.515931723913833,
-      "grad_norm": 0.0009804507717490196,
-      "learning_rate": 0.00019999986903156133,
-      "loss": 46.0,
-      "step": 6748
-    },
-    {
-      "epoch": 0.5160081808972227,
-      "grad_norm": 0.0160980224609375,
-      "learning_rate": 0.00019999986899268388,
-      "loss": 46.0,
-      "step": 6749
-    },
-    {
-      "epoch": 0.5160846378806124,
-      "grad_norm": 0.0039010727778077126,
-      "learning_rate": 0.00019999986895380065,
-      "loss": 46.0,
-      "step": 6750
-    },
-    {
-      "epoch": 0.5161610948640022,
-      "grad_norm": 0.0031121016945689917,
-      "learning_rate": 0.00019999986891491167,
-      "loss": 46.0,
-      "step": 6751
-    },
-    {
-      "epoch": 0.5162375518473918,
-      "grad_norm": 0.0006660231156274676,
-      "learning_rate": 0.00019999986887601692,
-      "loss": 46.0,
-      "step": 6752
-    },
-    {
-      "epoch": 0.5163140088307816,
-      "grad_norm": 0.0027548077050596476,
-      "learning_rate": 0.0001999998688371164,
-      "loss": 46.0,
-      "step": 6753
-    },
-    {
-      "epoch": 0.5163904658141713,
-      "grad_norm": 0.0019324087770655751,
-      "learning_rate": 0.0001999998687982101,
-      "loss": 46.0,
-      "step": 6754
-    },
-    {
-      "epoch": 0.516466922797561,
-      "grad_norm": 0.004523304756730795,
-      "learning_rate": 0.000199999868759298,
-      "loss": 46.0,
-      "step": 6755
-    },
-    {
-      "epoch": 0.5165433797809508,
-      "grad_norm": 0.0009916049893945456,
-      "learning_rate": 0.00019999986872038016,
-      "loss": 46.0,
-      "step": 6756
-    },
-    {
-      "epoch": 0.5166198367643404,
-      "grad_norm": 0.0011290484108030796,
-      "learning_rate": 0.0001999998686814566,
-      "loss": 46.0,
-      "step": 6757
-    },
-    {
-      "epoch": 0.5166962937477302,
-      "grad_norm": 0.0005818207282572985,
-      "learning_rate": 0.0001999998686425272,
-      "loss": 46.0,
-      "step": 6758
-    },
-    {
-      "epoch": 0.51677275073112,
-      "grad_norm": 0.00037642737152054906,
-      "learning_rate": 0.00019999986860359207,
-      "loss": 46.0,
-      "step": 6759
-    },
-    {
-      "epoch": 0.5168492077145096,
-      "grad_norm": 0.0014851245796307921,
-      "learning_rate": 0.00019999986856465113,
-      "loss": 46.0,
-      "step": 6760
-    },
-    {
-      "epoch": 0.5169256646978994,
-      "grad_norm": 0.001152483862824738,
-      "learning_rate": 0.00019999986852570447,
-      "loss": 46.0,
-      "step": 6761
-    },
-    {
-      "epoch": 0.5170021216812891,
-      "grad_norm": 0.0012562167830765247,
-      "learning_rate": 0.00019999986848675202,
-      "loss": 46.0,
-      "step": 6762
-    },
-    {
-      "epoch": 0.5170785786646788,
-      "grad_norm": 0.0007644011056981981,
-      "learning_rate": 0.0001999998684477938,
-      "loss": 46.0,
-      "step": 6763
-    },
-    {
-      "epoch": 0.5171550356480685,
-      "grad_norm": 0.0011796654434874654,
-      "learning_rate": 0.00019999986840882978,
-      "loss": 46.0,
-      "step": 6764
-    },
-    {
-      "epoch": 0.5172314926314582,
-      "grad_norm": 0.0017730558756738901,
-      "learning_rate": 0.00019999986836986003,
-      "loss": 46.0,
-      "step": 6765
-    },
-    {
-      "epoch": 0.5173079496148479,
-      "grad_norm": 0.001144835026934743,
-      "learning_rate": 0.00019999986833088448,
-      "loss": 46.0,
-      "step": 6766
-    },
-    {
-      "epoch": 0.5173844065982377,
-      "grad_norm": 0.007220027968287468,
-      "learning_rate": 0.00019999986829190322,
-      "loss": 46.0,
-      "step": 6767
-    },
-    {
-      "epoch": 0.5174608635816274,
-      "grad_norm": 0.0005154610844329,
-      "learning_rate": 0.00019999986825291615,
-      "loss": 46.0,
-      "step": 6768
-    },
-    {
-      "epoch": 0.5175373205650171,
-      "grad_norm": 0.001148174749687314,
-      "learning_rate": 0.0001999998682139233,
-      "loss": 46.0,
-      "step": 6769
-    },
-    {
-      "epoch": 0.5176137775484069,
-      "grad_norm": 0.0029547400772571564,
-      "learning_rate": 0.00019999986817492469,
-      "loss": 46.0,
-      "step": 6770
-    },
-    {
-      "epoch": 0.5176902345317965,
-      "grad_norm": 0.0012961478205397725,
-      "learning_rate": 0.00019999986813592032,
-      "loss": 46.0,
-      "step": 6771
-    },
-    {
-      "epoch": 0.5177666915151863,
-      "grad_norm": 0.0010385812493041158,
-      "learning_rate": 0.00019999986809691016,
-      "loss": 46.0,
-      "step": 6772
-    },
-    {
-      "epoch": 0.5178431484985759,
-      "grad_norm": 0.002846222138032317,
-      "learning_rate": 0.00019999986805789422,
-      "loss": 46.0,
-      "step": 6773
-    },
-    {
-      "epoch": 0.5179196054819657,
-      "grad_norm": 0.0010120623046532273,
-      "learning_rate": 0.00019999986801887257,
-      "loss": 46.0,
-      "step": 6774
-    },
-    {
-      "epoch": 0.5179960624653555,
-      "grad_norm": 0.0028149408753961325,
-      "learning_rate": 0.0001999998679798451,
-      "loss": 46.0,
-      "step": 6775
-    },
-    {
-      "epoch": 0.5180725194487451,
-      "grad_norm": 0.0007786746136844158,
-      "learning_rate": 0.00019999986794081188,
-      "loss": 46.0,
-      "step": 6776
-    },
-    {
-      "epoch": 0.5181489764321349,
-      "grad_norm": 0.0013361842138692737,
-      "learning_rate": 0.0001999998679017729,
-      "loss": 46.0,
-      "step": 6777
-    },
-    {
-      "epoch": 0.5182254334155246,
-      "grad_norm": 0.0024689168203622103,
-      "learning_rate": 0.00019999986786272813,
-      "loss": 46.0,
-      "step": 6778
-    },
-    {
-      "epoch": 0.5183018903989143,
-      "grad_norm": 0.0007742635207250714,
-      "learning_rate": 0.00019999986782367758,
-      "loss": 46.0,
-      "step": 6779
-    },
-    {
-      "epoch": 0.518378347382304,
-      "grad_norm": 0.001180786988697946,
-      "learning_rate": 0.00019999986778462128,
-      "loss": 46.0,
-      "step": 6780
-    },
-    {
-      "epoch": 0.5184548043656938,
-      "grad_norm": 0.0010374423582106829,
-      "learning_rate": 0.0001999998677455592,
-      "loss": 46.0,
-      "step": 6781
-    },
-    {
-      "epoch": 0.5185312613490834,
-      "grad_norm": 0.0008883423288352787,
-      "learning_rate": 0.00019999986770649137,
-      "loss": 46.0,
-      "step": 6782
-    },
-    {
-      "epoch": 0.5186077183324732,
-      "grad_norm": 0.0010320040164515376,
-      "learning_rate": 0.00019999986766741775,
-      "loss": 46.0,
-      "step": 6783
-    },
-    {
-      "epoch": 0.5186841753158629,
-      "grad_norm": 0.0012594616273418069,
-      "learning_rate": 0.0001999998676283384,
-      "loss": 46.0,
-      "step": 6784
-    },
-    {
-      "epoch": 0.5187606322992526,
-      "grad_norm": 0.002100956393405795,
-      "learning_rate": 0.00019999986758925325,
-      "loss": 46.0,
-      "step": 6785
-    },
-    {
-      "epoch": 0.5188370892826424,
-      "grad_norm": 0.0009413774823769927,
-      "learning_rate": 0.00019999986755016232,
-      "loss": 46.0,
-      "step": 6786
-    },
-    {
-      "epoch": 0.518913546266032,
-      "grad_norm": 0.0008941596024669707,
-      "learning_rate": 0.0001999998675110656,
-      "loss": 46.0,
-      "step": 6787
-    },
-    {
-      "epoch": 0.5189900032494218,
-      "grad_norm": 0.0012038570130243897,
-      "learning_rate": 0.00019999986747196315,
-      "loss": 46.0,
-      "step": 6788
-    },
-    {
-      "epoch": 0.5190664602328116,
-      "grad_norm": 0.0013174674240872264,
-      "learning_rate": 0.00019999986743285492,
-      "loss": 46.0,
-      "step": 6789
-    },
-    {
-      "epoch": 0.5191429172162012,
-      "grad_norm": 0.0008081771666184068,
-      "learning_rate": 0.00019999986739374092,
-      "loss": 46.0,
-      "step": 6790
-    },
-    {
-      "epoch": 0.519219374199591,
-      "grad_norm": 0.01545399334281683,
-      "learning_rate": 0.00019999986735462117,
-      "loss": 46.0,
-      "step": 6791
-    },
-    {
-      "epoch": 0.5192958311829807,
-      "grad_norm": 0.0007040402269922197,
-      "learning_rate": 0.00019999986731549565,
-      "loss": 46.0,
-      "step": 6792
-    },
-    {
-      "epoch": 0.5193722881663704,
-      "grad_norm": 0.0011792993173003197,
-      "learning_rate": 0.00019999986727636433,
-      "loss": 46.0,
-      "step": 6793
-    },
-    {
-      "epoch": 0.5194487451497601,
-      "grad_norm": 0.0023760884068906307,
-      "learning_rate": 0.00019999986723722726,
-      "loss": 46.0,
-      "step": 6794
-    },
-    {
-      "epoch": 0.5195252021331498,
-      "grad_norm": 0.000887617701664567,
-      "learning_rate": 0.0001999998671980844,
-      "loss": 46.0,
-      "step": 6795
-    },
-    {
-      "epoch": 0.5196016591165395,
-      "grad_norm": 0.0006348420865833759,
-      "learning_rate": 0.0001999998671589358,
-      "loss": 46.0,
-      "step": 6796
-    },
-    {
-      "epoch": 0.5196781160999293,
-      "grad_norm": 0.0008443936822004616,
-      "learning_rate": 0.0001999998671197814,
-      "loss": 46.0,
-      "step": 6797
-    },
-    {
-      "epoch": 0.519754573083319,
-      "grad_norm": 0.0001689426862867549,
-      "learning_rate": 0.00019999986708062127,
-      "loss": 46.0,
-      "step": 6798
-    },
-    {
-      "epoch": 0.5198310300667087,
-      "grad_norm": 0.0036721364594995975,
-      "learning_rate": 0.00019999986704145534,
-      "loss": 46.0,
-      "step": 6799
-    },
-    {
-      "epoch": 0.5199074870500985,
-      "grad_norm": 0.0016476784367114305,
-      "learning_rate": 0.00019999986700228365,
-      "loss": 46.0,
-      "step": 6800
-    },
-    {
-      "epoch": 0.5199839440334881,
-      "grad_norm": 0.004639696795493364,
-      "learning_rate": 0.0001999998669631062,
-      "loss": 46.0,
-      "step": 6801
-    },
-    {
-      "epoch": 0.5200604010168779,
-      "grad_norm": 0.0019115731120109558,
-      "learning_rate": 0.00019999986692392297,
-      "loss": 46.0,
-      "step": 6802
-    },
-    {
-      "epoch": 0.5201368580002677,
-      "grad_norm": 0.0009177215397357941,
-      "learning_rate": 0.00019999986688473397,
-      "loss": 46.0,
-      "step": 6803
-    },
-    {
-      "epoch": 0.5202133149836573,
-      "grad_norm": 0.000698661373462528,
-      "learning_rate": 0.00019999986684553916,
-      "loss": 46.0,
-      "step": 6804
-    },
-    {
-      "epoch": 0.5202897719670471,
-      "grad_norm": 0.0009078040602616966,
-      "learning_rate": 0.00019999986680633864,
-      "loss": 46.0,
-      "step": 6805
-    },
-    {
-      "epoch": 0.5203662289504367,
-      "grad_norm": 0.0035628413315862417,
-      "learning_rate": 0.00019999986676713235,
-      "loss": 46.0,
-      "step": 6806
-    },
-    {
-      "epoch": 0.5204426859338265,
-      "grad_norm": 0.002117432653903961,
-      "learning_rate": 0.00019999986672792025,
-      "loss": 46.0,
-      "step": 6807
-    },
-    {
-      "epoch": 0.5205191429172162,
-      "grad_norm": 0.000610047543887049,
-      "learning_rate": 0.0001999998666887024,
-      "loss": 46.0,
-      "step": 6808
-    },
-    {
-      "epoch": 0.5205955999006059,
-      "grad_norm": 0.0017189646605402231,
-      "learning_rate": 0.00019999986664947882,
-      "loss": 46.0,
-      "step": 6809
-    },
-    {
-      "epoch": 0.5206720568839956,
-      "grad_norm": 0.0017658865544945002,
-      "learning_rate": 0.0001999998666102494,
-      "loss": 46.0,
-      "step": 6810
-    },
-    {
-      "epoch": 0.5207485138673854,
-      "grad_norm": 0.006992855574935675,
-      "learning_rate": 0.00019999986657101424,
-      "loss": 46.0,
-      "step": 6811
-    },
-    {
-      "epoch": 0.5208249708507751,
-      "grad_norm": 0.0018627009121701121,
-      "learning_rate": 0.0001999998665317733,
-      "loss": 46.0,
-      "step": 6812
-    },
-    {
-      "epoch": 0.5209014278341648,
-      "grad_norm": 0.0045731826685369015,
-      "learning_rate": 0.00019999986649252662,
-      "loss": 46.0,
-      "step": 6813
-    },
-    {
-      "epoch": 0.5209778848175545,
-      "grad_norm": 0.0011791486758738756,
-      "learning_rate": 0.00019999986645327417,
-      "loss": 46.0,
-      "step": 6814
-    },
-    {
-      "epoch": 0.5210543418009442,
-      "grad_norm": 0.004062290769070387,
-      "learning_rate": 0.00019999986641401594,
-      "loss": 46.0,
-      "step": 6815
-    },
-    {
-      "epoch": 0.521130798784334,
-      "grad_norm": 0.0006228104466572404,
-      "learning_rate": 0.00019999986637475194,
-      "loss": 46.0,
-      "step": 6816
-    },
-    {
-      "epoch": 0.5212072557677236,
-      "grad_norm": 0.0008903697016648948,
-      "learning_rate": 0.00019999986633548216,
-      "loss": 46.0,
-      "step": 6817
-    },
-    {
-      "epoch": 0.5212837127511134,
-      "grad_norm": 0.011292941868305206,
-      "learning_rate": 0.0001999998662962066,
-      "loss": 46.0,
-      "step": 6818
-    },
-    {
-      "epoch": 0.5213601697345032,
-      "grad_norm": 0.00036900979466736317,
-      "learning_rate": 0.00019999986625692532,
-      "loss": 46.0,
-      "step": 6819
-    },
-    {
-      "epoch": 0.5214366267178928,
-      "grad_norm": 0.0006371656199917197,
-      "learning_rate": 0.00019999986621763825,
-      "loss": 46.0,
-      "step": 6820
-    },
-    {
-      "epoch": 0.5215130837012826,
-      "grad_norm": 0.0010578773217275739,
-      "learning_rate": 0.00019999986617834535,
-      "loss": 46.0,
-      "step": 6821
-    },
-    {
-      "epoch": 0.5215895406846723,
-      "grad_norm": 0.0007229943294078112,
-      "learning_rate": 0.00019999986613904676,
-      "loss": 46.0,
-      "step": 6822
-    },
-    {
-      "epoch": 0.521665997668062,
-      "grad_norm": 0.0011240994790568948,
-      "learning_rate": 0.00019999986609974237,
-      "loss": 46.0,
-      "step": 6823
-    },
-    {
-      "epoch": 0.5217424546514517,
-      "grad_norm": 0.0005957692046649754,
-      "learning_rate": 0.0001999998660604322,
-      "loss": 46.0,
-      "step": 6824
-    },
-    {
-      "epoch": 0.5218189116348414,
-      "grad_norm": 0.0009057045099325478,
-      "learning_rate": 0.00019999986602111628,
-      "loss": 46.0,
-      "step": 6825
-    },
-    {
-      "epoch": 0.5218953686182312,
-      "grad_norm": 0.0009465592447668314,
-      "learning_rate": 0.0001999998659817946,
-      "loss": 46.0,
-      "step": 6826
-    },
-    {
-      "epoch": 0.5219718256016209,
-      "grad_norm": 0.0051507200114429,
-      "learning_rate": 0.0001999998659424671,
-      "loss": 46.0,
-      "step": 6827
-    },
-    {
-      "epoch": 0.5220482825850106,
-      "grad_norm": 0.00029329105746001005,
-      "learning_rate": 0.00019999986590313388,
-      "loss": 46.0,
-      "step": 6828
-    },
-    {
-      "epoch": 0.5221247395684003,
-      "grad_norm": 0.001336118089966476,
-      "learning_rate": 0.00019999986586379488,
-      "loss": 46.0,
-      "step": 6829
-    },
-    {
-      "epoch": 0.5222011965517901,
-      "grad_norm": 0.0012513954425230622,
-      "learning_rate": 0.0001999998658244501,
-      "loss": 46.0,
-      "step": 6830
-    },
-    {
-      "epoch": 0.5222776535351797,
-      "grad_norm": 0.0006752603221684694,
-      "learning_rate": 0.00019999986578509956,
-      "loss": 46.0,
-      "step": 6831
-    },
-    {
-      "epoch": 0.5223541105185695,
-      "grad_norm": 0.0008080375846475363,
-      "learning_rate": 0.00019999986574574324,
-      "loss": 46.0,
-      "step": 6832
-    },
-    {
-      "epoch": 0.5224305675019593,
-      "grad_norm": 0.0005798717611469328,
-      "learning_rate": 0.00019999986570638117,
-      "loss": 46.0,
-      "step": 6833
-    },
-    {
-      "epoch": 0.5225070244853489,
-      "grad_norm": 0.0009395437664352357,
-      "learning_rate": 0.00019999986566701333,
-      "loss": 46.0,
-      "step": 6834
-    },
-    {
-      "epoch": 0.5225834814687387,
-      "grad_norm": 0.002736675785854459,
-      "learning_rate": 0.00019999986562763968,
-      "loss": 46.0,
-      "step": 6835
-    },
-    {
-      "epoch": 0.5226599384521283,
-      "grad_norm": 0.0017854857724159956,
-      "learning_rate": 0.0001999998655882603,
-      "loss": 46.0,
-      "step": 6836
-    },
-    {
-      "epoch": 0.5227363954355181,
-      "grad_norm": 0.0009235875913873315,
-      "learning_rate": 0.00019999986554887513,
-      "loss": 46.0,
-      "step": 6837
-    },
-    {
-      "epoch": 0.5228128524189078,
-      "grad_norm": 0.0009338136296719313,
-      "learning_rate": 0.00019999986550948423,
-      "loss": 46.0,
-      "step": 6838
-    },
-    {
-      "epoch": 0.5228893094022975,
-      "grad_norm": 0.0005506203160621226,
-      "learning_rate": 0.00019999986547008752,
-      "loss": 46.0,
-      "step": 6839
-    },
-    {
-      "epoch": 0.5229657663856873,
-      "grad_norm": 0.0014457597862929106,
-      "learning_rate": 0.00019999986543068504,
-      "loss": 46.0,
-      "step": 6840
-    },
-    {
-      "epoch": 0.523042223369077,
-      "grad_norm": 0.0004840172769036144,
-      "learning_rate": 0.0001999998653912768,
-      "loss": 46.0,
-      "step": 6841
-    },
-    {
-      "epoch": 0.5231186803524667,
-      "grad_norm": 0.0014432499883696437,
-      "learning_rate": 0.00019999986535186278,
-      "loss": 46.0,
-      "step": 6842
-    },
-    {
-      "epoch": 0.5231951373358564,
-      "grad_norm": 0.0023334503639489412,
-      "learning_rate": 0.000199999865312443,
-      "loss": 46.0,
-      "step": 6843
-    },
-    {
-      "epoch": 0.5232715943192461,
-      "grad_norm": 0.0024276773910969496,
-      "learning_rate": 0.00019999986527301746,
-      "loss": 46.0,
-      "step": 6844
-    },
-    {
-      "epoch": 0.5233480513026358,
-      "grad_norm": 0.0013002149062231183,
-      "learning_rate": 0.00019999986523358616,
-      "loss": 46.0,
-      "step": 6845
-    },
-    {
-      "epoch": 0.5234245082860256,
-      "grad_norm": 0.0018470765789970756,
-      "learning_rate": 0.00019999986519414907,
-      "loss": 46.0,
-      "step": 6846
-    },
-    {
-      "epoch": 0.5235009652694153,
-      "grad_norm": 0.0013910942943766713,
-      "learning_rate": 0.0001999998651547062,
-      "loss": 46.0,
-      "step": 6847
-    },
-    {
-      "epoch": 0.523577422252805,
-      "grad_norm": 0.0009281396633014083,
-      "learning_rate": 0.00019999986511525756,
-      "loss": 46.0,
-      "step": 6848
-    },
-    {
-      "epoch": 0.5236538792361948,
-      "grad_norm": 0.002898489125072956,
-      "learning_rate": 0.00019999986507580317,
-      "loss": 46.0,
-      "step": 6849
-    },
-    {
-      "epoch": 0.5237303362195844,
-      "grad_norm": 0.0014577272813767195,
-      "learning_rate": 0.000199999865036343,
-      "loss": 46.0,
-      "step": 6850
-    },
-    {
-      "epoch": 0.5238067932029742,
-      "grad_norm": 0.0008228104561567307,
-      "learning_rate": 0.00019999986499687707,
-      "loss": 46.0,
-      "step": 6851
-    },
-    {
-      "epoch": 0.523883250186364,
-      "grad_norm": 0.0097061051055789,
-      "learning_rate": 0.0001999998649574054,
-      "loss": 46.0,
-      "step": 6852
-    },
-    {
-      "epoch": 0.5239597071697536,
-      "grad_norm": 0.0008371971198357642,
-      "learning_rate": 0.0001999998649179279,
-      "loss": 46.0,
-      "step": 6853
-    },
-    {
-      "epoch": 0.5240361641531434,
-      "grad_norm": 0.0013694411609321833,
-      "learning_rate": 0.00019999986487844465,
-      "loss": 46.0,
-      "step": 6854
-    },
-    {
-      "epoch": 0.524112621136533,
-      "grad_norm": 0.001899379538372159,
-      "learning_rate": 0.00019999986483895565,
-      "loss": 46.0,
-      "step": 6855
-    },
-    {
-      "epoch": 0.5241890781199228,
-      "grad_norm": 0.006647980771958828,
-      "learning_rate": 0.00019999986479946088,
-      "loss": 46.0,
-      "step": 6856
-    },
-    {
-      "epoch": 0.5242655351033125,
-      "grad_norm": 0.001068764366209507,
-      "learning_rate": 0.0001999998647599603,
-      "loss": 46.0,
-      "step": 6857
-    },
-    {
-      "epoch": 0.5243419920867022,
-      "grad_norm": 0.0006564805516973138,
-      "learning_rate": 0.000199999864720454,
-      "loss": 46.0,
-      "step": 6858
-    },
-    {
-      "epoch": 0.5244184490700919,
-      "grad_norm": 0.0010898180771619081,
-      "learning_rate": 0.0001999998646809419,
-      "loss": 46.0,
-      "step": 6859
-    },
-    {
-      "epoch": 0.5244949060534817,
-      "grad_norm": 0.00036350873415358365,
-      "learning_rate": 0.00019999986464142407,
-      "loss": 46.0,
-      "step": 6860
-    },
-    {
-      "epoch": 0.5245713630368714,
-      "grad_norm": 0.0018155863508582115,
-      "learning_rate": 0.00019999986460190043,
-      "loss": 46.0,
-      "step": 6861
-    },
-    {
-      "epoch": 0.5246478200202611,
-      "grad_norm": 0.0008048060699366033,
-      "learning_rate": 0.00019999986456237104,
-      "loss": 46.0,
-      "step": 6862
-    },
-    {
-      "epoch": 0.5247242770036509,
-      "grad_norm": 0.0026283569168299437,
-      "learning_rate": 0.00019999986452283585,
-      "loss": 46.0,
-      "step": 6863
-    },
-    {
-      "epoch": 0.5248007339870405,
-      "grad_norm": 0.001201946404762566,
-      "learning_rate": 0.00019999986448329492,
-      "loss": 46.0,
-      "step": 6864
-    },
-    {
-      "epoch": 0.5248771909704303,
-      "grad_norm": 0.0016349850920960307,
-      "learning_rate": 0.00019999986444374824,
-      "loss": 46.0,
-      "step": 6865
-    },
-    {
-      "epoch": 0.5249536479538199,
-      "grad_norm": 0.0006353700300678611,
-      "learning_rate": 0.00019999986440419573,
-      "loss": 46.0,
-      "step": 6866
-    },
-    {
-      "epoch": 0.5250301049372097,
-      "grad_norm": 0.00213131052441895,
-      "learning_rate": 0.0001999998643646375,
-      "loss": 46.0,
-      "step": 6867
-    },
-    {
-      "epoch": 0.5251065619205995,
-      "grad_norm": 0.0010265909368172288,
-      "learning_rate": 0.0001999998643250735,
-      "loss": 46.0,
-      "step": 6868
-    },
-    {
-      "epoch": 0.5251830189039891,
-      "grad_norm": 0.0015670554712414742,
-      "learning_rate": 0.0001999998642855037,
-      "loss": 46.0,
-      "step": 6869
-    },
-    {
-      "epoch": 0.5252594758873789,
-      "grad_norm": 0.0018927843775600195,
-      "learning_rate": 0.00019999986424592815,
-      "loss": 46.0,
-      "step": 6870
-    },
-    {
-      "epoch": 0.5253359328707686,
-      "grad_norm": 0.0010991941671818495,
-      "learning_rate": 0.00019999986420634683,
-      "loss": 46.0,
-      "step": 6871
-    },
-    {
-      "epoch": 0.5254123898541583,
-      "grad_norm": 0.001754280412569642,
-      "learning_rate": 0.00019999986416675976,
-      "loss": 46.0,
-      "step": 6872
-    },
-    {
-      "epoch": 0.525488846837548,
-      "grad_norm": 0.0015437264228239655,
-      "learning_rate": 0.0001999998641271669,
-      "loss": 46.0,
-      "step": 6873
-    },
-    {
-      "epoch": 0.5255653038209377,
-      "grad_norm": 0.0006507346406579018,
-      "learning_rate": 0.00019999986408756826,
-      "loss": 46.0,
-      "step": 6874
-    },
-    {
-      "epoch": 0.5256417608043275,
-      "grad_norm": 0.0018157438607886434,
-      "learning_rate": 0.00019999986404796384,
-      "loss": 46.0,
-      "step": 6875
-    },
-    {
-      "epoch": 0.5257182177877172,
-      "grad_norm": 0.001315286848694086,
-      "learning_rate": 0.00019999986400835365,
-      "loss": 46.0,
-      "step": 6876
-    },
-    {
-      "epoch": 0.5257946747711069,
-      "grad_norm": 0.0005139801069162786,
-      "learning_rate": 0.00019999986396873775,
-      "loss": 46.0,
-      "step": 6877
-    },
-    {
-      "epoch": 0.5258711317544966,
-      "grad_norm": 0.002010364318266511,
-      "learning_rate": 0.00019999986392911604,
-      "loss": 46.0,
-      "step": 6878
-    },
-    {
-      "epoch": 0.5259475887378864,
-      "grad_norm": 0.0009020005236379802,
-      "learning_rate": 0.00019999986388948856,
-      "loss": 46.0,
-      "step": 6879
-    },
-    {
-      "epoch": 0.526024045721276,
-      "grad_norm": 0.0010088416747748852,
-      "learning_rate": 0.0001999998638498553,
-      "loss": 46.0,
-      "step": 6880
-    },
-    {
-      "epoch": 0.5261005027046658,
-      "grad_norm": 0.0013785704504698515,
-      "learning_rate": 0.0001999998638102163,
-      "loss": 46.0,
-      "step": 6881
-    },
-    {
-      "epoch": 0.5261769596880556,
-      "grad_norm": 0.0015304688131436706,
-      "learning_rate": 0.00019999986377057153,
-      "loss": 46.0,
-      "step": 6882
-    },
-    {
-      "epoch": 0.5262534166714452,
-      "grad_norm": 0.0003160073247272521,
-      "learning_rate": 0.00019999986373092095,
-      "loss": 46.0,
-      "step": 6883
-    },
-    {
-      "epoch": 0.526329873654835,
-      "grad_norm": 0.0016546716215088964,
-      "learning_rate": 0.00019999986369126463,
-      "loss": 46.0,
-      "step": 6884
-    },
-    {
-      "epoch": 0.5264063306382246,
-      "grad_norm": 0.006996653508394957,
-      "learning_rate": 0.00019999986365160254,
-      "loss": 46.0,
-      "step": 6885
-    },
-    {
-      "epoch": 0.5264827876216144,
-      "grad_norm": 0.0025435148272663355,
-      "learning_rate": 0.00019999986361193467,
-      "loss": 46.0,
-      "step": 6886
-    },
-    {
-      "epoch": 0.5265592446050041,
-      "grad_norm": 0.0007781276945024729,
-      "learning_rate": 0.00019999986357226103,
-      "loss": 46.0,
-      "step": 6887
-    },
-    {
-      "epoch": 0.5266357015883938,
-      "grad_norm": 0.0015560660976916552,
-      "learning_rate": 0.00019999986353258162,
-      "loss": 46.0,
-      "step": 6888
-    },
-    {
-      "epoch": 0.5267121585717836,
-      "grad_norm": 0.004460847470909357,
-      "learning_rate": 0.00019999986349289646,
-      "loss": 46.0,
-      "step": 6889
-    },
-    {
-      "epoch": 0.5267886155551733,
-      "grad_norm": 0.002675467636436224,
-      "learning_rate": 0.00019999986345320552,
-      "loss": 46.0,
-      "step": 6890
-    },
-    {
-      "epoch": 0.526865072538563,
-      "grad_norm": 0.00148398382589221,
-      "learning_rate": 0.00019999986341350881,
-      "loss": 46.0,
-      "step": 6891
-    },
-    {
-      "epoch": 0.5269415295219527,
-      "grad_norm": 0.00960307102650404,
-      "learning_rate": 0.0001999998633738063,
-      "loss": 46.0,
-      "step": 6892
-    },
-    {
-      "epoch": 0.5270179865053425,
-      "grad_norm": 0.0012577949091792107,
-      "learning_rate": 0.00019999986333409808,
-      "loss": 46.0,
-      "step": 6893
-    },
-    {
-      "epoch": 0.5270944434887321,
-      "grad_norm": 0.0005339909694157541,
-      "learning_rate": 0.00019999986329438405,
-      "loss": 46.0,
-      "step": 6894
-    },
-    {
-      "epoch": 0.5271709004721219,
-      "grad_norm": 0.0020182239823043346,
-      "learning_rate": 0.00019999986325466428,
-      "loss": 46.0,
-      "step": 6895
-    },
-    {
-      "epoch": 0.5272473574555115,
-      "grad_norm": 0.0010359484003856778,
-      "learning_rate": 0.00019999986321493868,
-      "loss": 46.0,
-      "step": 6896
-    },
-    {
-      "epoch": 0.5273238144389013,
-      "grad_norm": 0.0010302109876647592,
-      "learning_rate": 0.00019999986317520736,
-      "loss": 46.0,
-      "step": 6897
-    },
-    {
-      "epoch": 0.5274002714222911,
-      "grad_norm": 0.001595202716998756,
-      "learning_rate": 0.0001999998631354703,
-      "loss": 46.0,
-      "step": 6898
-    },
-    {
-      "epoch": 0.5274767284056807,
-      "grad_norm": 0.001114501035772264,
-      "learning_rate": 0.00019999986309572742,
-      "loss": 46.0,
-      "step": 6899
-    },
-    {
-      "epoch": 0.5275531853890705,
-      "grad_norm": 0.0008797789923846722,
-      "learning_rate": 0.00019999986305597878,
-      "loss": 46.0,
-      "step": 6900
-    },
-    {
-      "epoch": 0.5276296423724602,
-      "grad_norm": 0.000978010008111596,
-      "learning_rate": 0.00019999986301622434,
-      "loss": 46.0,
-      "step": 6901
-    },
-    {
-      "epoch": 0.5277060993558499,
-      "grad_norm": 0.00047197192907333374,
-      "learning_rate": 0.0001999998629764642,
-      "loss": 46.0,
-      "step": 6902
-    },
-    {
-      "epoch": 0.5277825563392397,
-      "grad_norm": 0.00043294852366670966,
-      "learning_rate": 0.00019999986293669825,
-      "loss": 46.0,
-      "step": 6903
-    },
-    {
-      "epoch": 0.5278590133226293,
-      "grad_norm": 0.0006556301959790289,
-      "learning_rate": 0.00019999986289692654,
-      "loss": 46.0,
-      "step": 6904
-    },
-    {
-      "epoch": 0.5279354703060191,
-      "grad_norm": 0.002570302924141288,
-      "learning_rate": 0.00019999986285714903,
-      "loss": 46.0,
-      "step": 6905
-    },
-    {
-      "epoch": 0.5280119272894088,
-      "grad_norm": 0.003173999721184373,
-      "learning_rate": 0.0001999998628173658,
-      "loss": 46.0,
-      "step": 6906
-    },
-    {
-      "epoch": 0.5280883842727985,
-      "grad_norm": 0.001547998166643083,
-      "learning_rate": 0.00019999986277757675,
-      "loss": 46.0,
-      "step": 6907
-    },
-    {
-      "epoch": 0.5281648412561882,
-      "grad_norm": 0.0006692355964332819,
-      "learning_rate": 0.00019999986273778198,
-      "loss": 46.0,
-      "step": 6908
-    },
-    {
-      "epoch": 0.528241298239578,
-      "grad_norm": 0.005438718944787979,
-      "learning_rate": 0.0001999998626979814,
-      "loss": 46.0,
-      "step": 6909
-    },
-    {
-      "epoch": 0.5283177552229676,
-      "grad_norm": 0.0006363442516885698,
-      "learning_rate": 0.00019999986265817509,
-      "loss": 46.0,
-      "step": 6910
-    },
-    {
-      "epoch": 0.5283942122063574,
-      "grad_norm": 0.0013391983229666948,
-      "learning_rate": 0.000199999862618363,
-      "loss": 46.0,
-      "step": 6911
-    },
-    {
-      "epoch": 0.5284706691897472,
-      "grad_norm": 0.0017163304146379232,
-      "learning_rate": 0.00019999986257854512,
-      "loss": 46.0,
-      "step": 6912
-    },
-    {
-      "epoch": 0.5285471261731368,
-      "grad_norm": 0.0008894107886590064,
-      "learning_rate": 0.00019999986253872148,
-      "loss": 46.0,
-      "step": 6913
-    },
-    {
-      "epoch": 0.5286235831565266,
-      "grad_norm": 0.0005524918669834733,
-      "learning_rate": 0.0001999998624988921,
-      "loss": 46.0,
-      "step": 6914
-    },
-    {
-      "epoch": 0.5287000401399162,
-      "grad_norm": 0.00254958588629961,
-      "learning_rate": 0.00019999986245905688,
-      "loss": 46.0,
-      "step": 6915
-    },
-    {
-      "epoch": 0.528776497123306,
-      "grad_norm": 0.0010677631944417953,
-      "learning_rate": 0.00019999986241921595,
-      "loss": 46.0,
-      "step": 6916
-    },
-    {
-      "epoch": 0.5288529541066957,
-      "grad_norm": 0.0010262842988595366,
-      "learning_rate": 0.00019999986237936922,
-      "loss": 46.0,
-      "step": 6917
-    },
-    {
-      "epoch": 0.5289294110900854,
-      "grad_norm": 0.0010445505613461137,
-      "learning_rate": 0.00019999986233951674,
-      "loss": 46.0,
-      "step": 6918
-    },
-    {
-      "epoch": 0.5290058680734752,
-      "grad_norm": 0.0013391776010394096,
-      "learning_rate": 0.00019999986229965848,
-      "loss": 46.0,
-      "step": 6919
-    },
-    {
-      "epoch": 0.5290823250568649,
-      "grad_norm": 0.0036920837592333555,
-      "learning_rate": 0.00019999986225979446,
-      "loss": 46.0,
-      "step": 6920
-    },
-    {
-      "epoch": 0.5291587820402546,
-      "grad_norm": 0.0008187103667296469,
-      "learning_rate": 0.00019999986221992466,
-      "loss": 46.0,
-      "step": 6921
-    },
-    {
-      "epoch": 0.5292352390236443,
-      "grad_norm": 0.0009263358078896999,
-      "learning_rate": 0.00019999986218004908,
-      "loss": 46.0,
-      "step": 6922
-    },
-    {
-      "epoch": 0.5293116960070341,
-      "grad_norm": 0.0016071456484496593,
-      "learning_rate": 0.00019999986214016776,
-      "loss": 46.0,
-      "step": 6923
-    },
-    {
-      "epoch": 0.5293881529904237,
-      "grad_norm": 0.0012537785805761814,
-      "learning_rate": 0.00019999986210028067,
-      "loss": 46.0,
-      "step": 6924
-    },
-    {
-      "epoch": 0.5294646099738135,
-      "grad_norm": 0.005476954393088818,
-      "learning_rate": 0.00019999986206038778,
-      "loss": 46.0,
-      "step": 6925
-    },
-    {
-      "epoch": 0.5295410669572032,
-      "grad_norm": 0.0032883668318390846,
-      "learning_rate": 0.00019999986202048916,
-      "loss": 46.0,
-      "step": 6926
-    },
-    {
-      "epoch": 0.5296175239405929,
-      "grad_norm": 0.001330337836407125,
-      "learning_rate": 0.00019999986198058472,
-      "loss": 46.0,
-      "step": 6927
-    },
-    {
-      "epoch": 0.5296939809239827,
-      "grad_norm": 0.0025806999765336514,
-      "learning_rate": 0.00019999986194067456,
-      "loss": 46.0,
-      "step": 6928
-    },
-    {
-      "epoch": 0.5297704379073723,
-      "grad_norm": 0.001325944671407342,
-      "learning_rate": 0.0001999998619007586,
-      "loss": 46.0,
-      "step": 6929
-    },
-    {
-      "epoch": 0.5298468948907621,
-      "grad_norm": 0.0008160117431543767,
-      "learning_rate": 0.00019999986186083687,
-      "loss": 46.0,
-      "step": 6930
-    },
-    {
-      "epoch": 0.5299233518741518,
-      "grad_norm": 0.004486887715756893,
-      "learning_rate": 0.0001999998618209094,
-      "loss": 46.0,
-      "step": 6931
-    },
-    {
-      "epoch": 0.5299998088575415,
-      "grad_norm": 0.0032692733220756054,
-      "learning_rate": 0.00019999986178097614,
-      "loss": 46.0,
-      "step": 6932
-    },
-    {
-      "epoch": 0.5300762658409313,
-      "grad_norm": 0.0015005492605268955,
-      "learning_rate": 0.0001999998617410371,
-      "loss": 46.0,
-      "step": 6933
-    },
-    {
-      "epoch": 0.530152722824321,
-      "grad_norm": 0.001956478226929903,
-      "learning_rate": 0.0001999998617010923,
-      "loss": 46.0,
-      "step": 6934
-    },
-    {
-      "epoch": 0.5302291798077107,
-      "grad_norm": 0.004179133102297783,
-      "learning_rate": 0.00019999986166114176,
-      "loss": 46.0,
-      "step": 6935
-    },
-    {
-      "epoch": 0.5303056367911004,
-      "grad_norm": 0.004741376265883446,
-      "learning_rate": 0.00019999986162118542,
-      "loss": 46.0,
-      "step": 6936
-    },
-    {
-      "epoch": 0.5303820937744901,
-      "grad_norm": 0.0106956847012043,
-      "learning_rate": 0.0001999998615812233,
-      "loss": 46.0,
-      "step": 6937
-    },
-    {
-      "epoch": 0.5304585507578798,
-      "grad_norm": 0.0022359401918947697,
-      "learning_rate": 0.00019999986154125543,
-      "loss": 46.0,
-      "step": 6938
-    },
-    {
-      "epoch": 0.5305350077412696,
-      "grad_norm": 0.001093576429411769,
-      "learning_rate": 0.0001999998615012818,
-      "loss": 46.0,
-      "step": 6939
-    },
-    {
-      "epoch": 0.5306114647246593,
-      "grad_norm": 0.0022686878219246864,
-      "learning_rate": 0.00019999986146130238,
-      "loss": 46.0,
-      "step": 6940
-    },
-    {
-      "epoch": 0.530687921708049,
-      "grad_norm": 0.0009230774012394249,
-      "learning_rate": 0.0001999998614213172,
-      "loss": 46.0,
-      "step": 6941
-    },
-    {
-      "epoch": 0.5307643786914388,
-      "grad_norm": 0.0010031594429165125,
-      "learning_rate": 0.00019999986138132623,
-      "loss": 46.0,
-      "step": 6942
-    },
-    {
-      "epoch": 0.5308408356748284,
-      "grad_norm": 0.0012422491563484073,
-      "learning_rate": 0.0001999998613413295,
-      "loss": 46.0,
-      "step": 6943
-    },
-    {
-      "epoch": 0.5309172926582182,
-      "grad_norm": 0.00034665808198042214,
-      "learning_rate": 0.00019999986130132702,
-      "loss": 46.0,
-      "step": 6944
-    },
-    {
-      "epoch": 0.5309937496416078,
-      "grad_norm": 0.0005560446297749877,
-      "learning_rate": 0.00019999986126131877,
-      "loss": 46.0,
-      "step": 6945
-    },
-    {
-      "epoch": 0.5310702066249976,
-      "grad_norm": 0.009018749929964542,
-      "learning_rate": 0.00019999986122130471,
-      "loss": 46.0,
-      "step": 6946
-    },
-    {
-      "epoch": 0.5311466636083874,
-      "grad_norm": 0.0006785035948269069,
-      "learning_rate": 0.00019999986118128494,
-      "loss": 46.0,
-      "step": 6947
-    },
-    {
-      "epoch": 0.531223120591777,
-      "grad_norm": 0.0043200114741921425,
-      "learning_rate": 0.00019999986114125937,
-      "loss": 46.0,
-      "step": 6948
-    },
-    {
-      "epoch": 0.5312995775751668,
-      "grad_norm": 0.0007045610109344125,
-      "learning_rate": 0.00019999986110122802,
-      "loss": 46.0,
-      "step": 6949
-    },
-    {
-      "epoch": 0.5313760345585565,
-      "grad_norm": 0.00042669090908020735,
-      "learning_rate": 0.00019999986106119093,
-      "loss": 46.0,
-      "step": 6950
-    },
-    {
-      "epoch": 0.5314524915419462,
-      "grad_norm": 0.0005369912250898778,
-      "learning_rate": 0.00019999986102114804,
-      "loss": 46.0,
-      "step": 6951
-    },
-    {
-      "epoch": 0.5315289485253359,
-      "grad_norm": 0.003414823906496167,
-      "learning_rate": 0.0001999998609810994,
-      "loss": 46.0,
-      "step": 6952
-    },
-    {
-      "epoch": 0.5316054055087257,
-      "grad_norm": 0.0012889598729088902,
-      "learning_rate": 0.00019999986094104496,
-      "loss": 46.0,
-      "step": 6953
-    },
-    {
-      "epoch": 0.5316818624921154,
-      "grad_norm": 0.0009127567755058408,
-      "learning_rate": 0.00019999986090098477,
-      "loss": 46.0,
-      "step": 6954
-    },
-    {
-      "epoch": 0.5317583194755051,
-      "grad_norm": 0.0012592733837664127,
-      "learning_rate": 0.0001999998608609188,
-      "loss": 46.0,
-      "step": 6955
-    },
-    {
-      "epoch": 0.5318347764588948,
-      "grad_norm": 0.0006647682166658342,
-      "learning_rate": 0.0001999998608208471,
-      "loss": 46.0,
-      "step": 6956
-    },
-    {
-      "epoch": 0.5319112334422845,
-      "grad_norm": 0.00046009011566638947,
-      "learning_rate": 0.0001999998607807696,
-      "loss": 46.0,
-      "step": 6957
-    },
-    {
-      "epoch": 0.5319876904256743,
-      "grad_norm": 0.0010078222258016467,
-      "learning_rate": 0.00019999986074068632,
-      "loss": 46.0,
-      "step": 6958
-    },
-    {
-      "epoch": 0.5320641474090639,
-      "grad_norm": 0.0005325308884494007,
-      "learning_rate": 0.0001999998607005973,
-      "loss": 46.0,
-      "step": 6959
-    },
-    {
-      "epoch": 0.5321406043924537,
-      "grad_norm": 0.0021776785142719746,
-      "learning_rate": 0.00019999986066050252,
-      "loss": 46.0,
-      "step": 6960
-    },
-    {
-      "epoch": 0.5322170613758435,
-      "grad_norm": 0.0006236338522285223,
-      "learning_rate": 0.00019999986062040192,
-      "loss": 46.0,
-      "step": 6961
-    },
-    {
-      "epoch": 0.5322935183592331,
-      "grad_norm": 0.0080524031072855,
-      "learning_rate": 0.0001999998605802956,
-      "loss": 46.0,
-      "step": 6962
-    },
-    {
-      "epoch": 0.5323699753426229,
-      "grad_norm": 0.0004457864270079881,
-      "learning_rate": 0.00019999986054018346,
-      "loss": 46.0,
-      "step": 6963
-    },
-    {
-      "epoch": 0.5324464323260126,
-      "grad_norm": 0.0008286235970444977,
-      "learning_rate": 0.00019999986050006562,
-      "loss": 46.0,
-      "step": 6964
-    },
-    {
-      "epoch": 0.5325228893094023,
-      "grad_norm": 0.0010822893818840384,
-      "learning_rate": 0.00019999986045994198,
-      "loss": 46.0,
-      "step": 6965
-    },
-    {
-      "epoch": 0.532599346292792,
-      "grad_norm": 0.003305522259324789,
-      "learning_rate": 0.00019999986041981254,
-      "loss": 46.0,
-      "step": 6966
-    },
-    {
-      "epoch": 0.5326758032761817,
-      "grad_norm": 0.0008318547043018043,
-      "learning_rate": 0.00019999986037967733,
-      "loss": 46.0,
-      "step": 6967
-    },
-    {
-      "epoch": 0.5327522602595715,
-      "grad_norm": 0.0012184131192043424,
-      "learning_rate": 0.00019999986033953637,
-      "loss": 46.0,
-      "step": 6968
-    },
-    {
-      "epoch": 0.5328287172429612,
-      "grad_norm": 0.0009557275334373116,
-      "learning_rate": 0.00019999986029938964,
-      "loss": 46.0,
-      "step": 6969
-    },
-    {
-      "epoch": 0.5329051742263509,
-      "grad_norm": 0.0008392874151468277,
-      "learning_rate": 0.00019999986025923716,
-      "loss": 46.0,
-      "step": 6970
-    },
-    {
-      "epoch": 0.5329816312097406,
-      "grad_norm": 0.001108025317080319,
-      "learning_rate": 0.0001999998602190789,
-      "loss": 46.0,
-      "step": 6971
-    },
-    {
-      "epoch": 0.5330580881931304,
-      "grad_norm": 0.0009874349925667048,
-      "learning_rate": 0.00019999986017891485,
-      "loss": 46.0,
-      "step": 6972
-    },
-    {
-      "epoch": 0.53313454517652,
-      "grad_norm": 0.0005841582315042615,
-      "learning_rate": 0.00019999986013874503,
-      "loss": 46.0,
-      "step": 6973
-    },
-    {
-      "epoch": 0.5332110021599098,
-      "grad_norm": 0.0008298710454255342,
-      "learning_rate": 0.00019999986009856948,
-      "loss": 46.0,
-      "step": 6974
-    },
-    {
-      "epoch": 0.5332874591432994,
-      "grad_norm": 0.001268972409889102,
-      "learning_rate": 0.00019999986005838814,
-      "loss": 46.0,
-      "step": 6975
-    },
-    {
-      "epoch": 0.5333639161266892,
-      "grad_norm": 0.0010404494823887944,
-      "learning_rate": 0.00019999986001820102,
-      "loss": 46.0,
-      "step": 6976
-    },
-    {
-      "epoch": 0.533440373110079,
-      "grad_norm": 0.0017261148896068335,
-      "learning_rate": 0.00019999985997800815,
-      "loss": 46.0,
-      "step": 6977
-    },
-    {
-      "epoch": 0.5335168300934686,
-      "grad_norm": 0.000760427734348923,
-      "learning_rate": 0.00019999985993780949,
-      "loss": 46.0,
-      "step": 6978
-    },
-    {
-      "epoch": 0.5335932870768584,
-      "grad_norm": 0.0006444074097089469,
-      "learning_rate": 0.00019999985989760505,
-      "loss": 46.0,
-      "step": 6979
-    },
-    {
-      "epoch": 0.5336697440602481,
-      "grad_norm": 0.0007587940781377256,
-      "learning_rate": 0.00019999985985739486,
-      "loss": 46.0,
-      "step": 6980
-    },
-    {
-      "epoch": 0.5337462010436378,
-      "grad_norm": 0.0015740134986117482,
-      "learning_rate": 0.0001999998598171789,
-      "loss": 46.0,
-      "step": 6981
-    },
-    {
-      "epoch": 0.5338226580270276,
-      "grad_norm": 0.001439502346329391,
-      "learning_rate": 0.00019999985977695717,
-      "loss": 46.0,
-      "step": 6982
-    },
-    {
-      "epoch": 0.5338991150104173,
-      "grad_norm": 0.0013325884938240051,
-      "learning_rate": 0.00019999985973672966,
-      "loss": 46.0,
-      "step": 6983
-    },
-    {
-      "epoch": 0.533975571993807,
-      "grad_norm": 0.0023438294883817434,
-      "learning_rate": 0.00019999985969649639,
-      "loss": 46.0,
-      "step": 6984
-    },
-    {
-      "epoch": 0.5340520289771967,
-      "grad_norm": 0.0006498959846794605,
-      "learning_rate": 0.00019999985965625736,
-      "loss": 46.0,
-      "step": 6985
-    },
-    {
-      "epoch": 0.5341284859605864,
-      "grad_norm": 0.0018395694205537438,
-      "learning_rate": 0.00019999985961601256,
-      "loss": 46.0,
-      "step": 6986
-    },
-    {
-      "epoch": 0.5342049429439761,
-      "grad_norm": 0.000798806082457304,
-      "learning_rate": 0.00019999985957576196,
-      "loss": 46.0,
-      "step": 6987
-    },
-    {
-      "epoch": 0.5342813999273659,
-      "grad_norm": 0.002494304208084941,
-      "learning_rate": 0.00019999985953550562,
-      "loss": 46.0,
-      "step": 6988
-    },
-    {
-      "epoch": 0.5343578569107555,
-      "grad_norm": 0.0006741366232745349,
-      "learning_rate": 0.00019999985949524353,
-      "loss": 46.0,
-      "step": 6989
-    },
-    {
-      "epoch": 0.5344343138941453,
-      "grad_norm": 0.016659339889883995,
-      "learning_rate": 0.00019999985945497563,
-      "loss": 46.0,
-      "step": 6990
-    },
-    {
-      "epoch": 0.5345107708775351,
-      "grad_norm": 0.0033929001074284315,
-      "learning_rate": 0.00019999985941470197,
-      "loss": 46.0,
-      "step": 6991
-    },
-    {
-      "epoch": 0.5345872278609247,
-      "grad_norm": 0.0003985811781603843,
-      "learning_rate": 0.00019999985937442253,
-      "loss": 46.0,
-      "step": 6992
-    },
-    {
-      "epoch": 0.5346636848443145,
-      "grad_norm": 0.002357101533561945,
-      "learning_rate": 0.00019999985933413737,
-      "loss": 46.0,
-      "step": 6993
-    },
-    {
-      "epoch": 0.5347401418277042,
-      "grad_norm": 0.00506207812577486,
-      "learning_rate": 0.00019999985929384639,
-      "loss": 46.0,
-      "step": 6994
-    },
-    {
-      "epoch": 0.5348165988110939,
-      "grad_norm": 0.014972719363868237,
-      "learning_rate": 0.00019999985925354965,
-      "loss": 46.0,
-      "step": 6995
-    },
-    {
-      "epoch": 0.5348930557944837,
-      "grad_norm": 0.0010895323939621449,
-      "learning_rate": 0.00019999985921324712,
-      "loss": 46.0,
-      "step": 6996
-    },
-    {
-      "epoch": 0.5349695127778733,
-      "grad_norm": 0.0007342291064560413,
-      "learning_rate": 0.0001999998591729389,
-      "loss": 46.0,
-      "step": 6997
-    },
-    {
-      "epoch": 0.5350459697612631,
-      "grad_norm": 0.0018762554973363876,
-      "learning_rate": 0.00019999985913262482,
-      "loss": 46.0,
-      "step": 6998
-    },
-    {
-      "epoch": 0.5351224267446528,
-      "grad_norm": 0.0005539553822018206,
-      "learning_rate": 0.00019999985909230502,
-      "loss": 46.0,
-      "step": 6999
-    },
-    {
-      "epoch": 0.5351988837280425,
-      "grad_norm": 0.002060561440885067,
-      "learning_rate": 0.00019999985905197945,
-      "loss": 46.0,
-      "step": 7000
-    },
-    {
-      "epoch": 0.5352753407114322,
-      "grad_norm": 0.001272550318390131,
-      "learning_rate": 0.00019999985901164808,
-      "loss": 46.0,
-      "step": 7001
-    },
-    {
-      "epoch": 0.535351797694822,
-      "grad_norm": 0.0010857011657208204,
-      "learning_rate": 0.00019999985897131098,
-      "loss": 46.0,
-      "step": 7002
-    },
-    {
-      "epoch": 0.5354282546782116,
-      "grad_norm": 0.002568619092926383,
-      "learning_rate": 0.0001999998589309681,
-      "loss": 46.0,
-      "step": 7003
-    },
-    {
-      "epoch": 0.5355047116616014,
-      "grad_norm": 0.0037138089537620544,
-      "learning_rate": 0.00019999985889061943,
-      "loss": 46.0,
-      "step": 7004
-    },
-    {
-      "epoch": 0.5355811686449911,
-      "grad_norm": 0.0009612892754375935,
-      "learning_rate": 0.000199999858850265,
-      "loss": 46.0,
-      "step": 7005
-    },
-    {
-      "epoch": 0.5356576256283808,
-      "grad_norm": 0.002627705456689,
-      "learning_rate": 0.0001999998588099048,
-      "loss": 46.0,
-      "step": 7006
-    },
-    {
-      "epoch": 0.5357340826117706,
-      "grad_norm": 0.00771617516875267,
-      "learning_rate": 0.00019999985876953882,
-      "loss": 46.0,
-      "step": 7007
-    },
-    {
-      "epoch": 0.5358105395951602,
-      "grad_norm": 0.0007642025593668222,
-      "learning_rate": 0.0001999998587291671,
-      "loss": 46.0,
-      "step": 7008
-    },
-    {
-      "epoch": 0.53588699657855,
-      "grad_norm": 0.0040270183235406876,
-      "learning_rate": 0.00019999985868878958,
-      "loss": 46.0,
-      "step": 7009
-    },
-    {
-      "epoch": 0.5359634535619398,
-      "grad_norm": 0.010913661681115627,
-      "learning_rate": 0.0001999998586484063,
-      "loss": 46.0,
-      "step": 7010
-    },
-    {
-      "epoch": 0.5360399105453294,
-      "grad_norm": 0.002033588942140341,
-      "learning_rate": 0.00019999985860801725,
-      "loss": 46.0,
-      "step": 7011
-    },
-    {
-      "epoch": 0.5361163675287192,
-      "grad_norm": 0.010351110249757767,
-      "learning_rate": 0.00019999985856762245,
-      "loss": 46.0,
-      "step": 7012
-    },
-    {
-      "epoch": 0.5361928245121089,
-      "grad_norm": 0.0013268848415464163,
-      "learning_rate": 0.00019999985852722188,
-      "loss": 46.0,
-      "step": 7013
-    },
-    {
-      "epoch": 0.5362692814954986,
-      "grad_norm": 0.0012470822548493743,
-      "learning_rate": 0.0001999998584868155,
-      "loss": 46.0,
-      "step": 7014
-    },
-    {
-      "epoch": 0.5363457384788883,
-      "grad_norm": 0.0014491754118353128,
-      "learning_rate": 0.0001999998584464034,
-      "loss": 46.0,
-      "step": 7015
-    },
-    {
-      "epoch": 0.536422195462278,
-      "grad_norm": 0.0009737068903632462,
-      "learning_rate": 0.0001999998584059855,
-      "loss": 46.0,
-      "step": 7016
-    },
-    {
-      "epoch": 0.5364986524456677,
-      "grad_norm": 0.0014498020755127072,
-      "learning_rate": 0.00019999985836556184,
-      "loss": 46.0,
-      "step": 7017
-    },
-    {
-      "epoch": 0.5365751094290575,
-      "grad_norm": 0.0009031247464008629,
-      "learning_rate": 0.0001999998583251324,
-      "loss": 46.0,
-      "step": 7018
-    },
-    {
-      "epoch": 0.5366515664124472,
-      "grad_norm": 0.001969572389498353,
-      "learning_rate": 0.00019999985828469721,
-      "loss": 46.0,
-      "step": 7019
-    },
-    {
-      "epoch": 0.5367280233958369,
-      "grad_norm": 0.0006924162735231221,
-      "learning_rate": 0.00019999985824425626,
-      "loss": 46.0,
-      "step": 7020
-    },
-    {
-      "epoch": 0.5368044803792267,
-      "grad_norm": 0.000846325303427875,
-      "learning_rate": 0.0001999998582038095,
-      "loss": 46.0,
-      "step": 7021
-    },
-    {
-      "epoch": 0.5368809373626163,
-      "grad_norm": 0.0018873761873692274,
-      "learning_rate": 0.000199999858163357,
-      "loss": 46.0,
-      "step": 7022
-    },
-    {
-      "epoch": 0.5369573943460061,
-      "grad_norm": 0.0010873053688555956,
-      "learning_rate": 0.0001999998581228987,
-      "loss": 46.0,
-      "step": 7023
-    },
-    {
-      "epoch": 0.5370338513293959,
-      "grad_norm": 0.00426768371835351,
-      "learning_rate": 0.00019999985808243466,
-      "loss": 46.0,
-      "step": 7024
-    },
-    {
-      "epoch": 0.5371103083127855,
-      "grad_norm": 0.0008283958886750042,
-      "learning_rate": 0.00019999985804196487,
-      "loss": 46.0,
-      "step": 7025
-    },
-    {
-      "epoch": 0.5371867652961753,
-      "grad_norm": 0.0005977737600915134,
-      "learning_rate": 0.00019999985800148927,
-      "loss": 46.0,
-      "step": 7026
-    },
-    {
-      "epoch": 0.5372632222795649,
-      "grad_norm": 0.00611245259642601,
-      "learning_rate": 0.0001999998579610079,
-      "loss": 46.0,
-      "step": 7027
-    },
-    {
-      "epoch": 0.5373396792629547,
-      "grad_norm": 0.0009862891165539622,
-      "learning_rate": 0.00019999985792052078,
-      "loss": 46.0,
-      "step": 7028
-    },
-    {
-      "epoch": 0.5374161362463444,
-      "grad_norm": 0.004697184078395367,
-      "learning_rate": 0.0001999998578800279,
-      "loss": 46.0,
-      "step": 7029
-    },
-    {
-      "epoch": 0.5374925932297341,
-      "grad_norm": 0.0021030385978519917,
-      "learning_rate": 0.00019999985783952922,
-      "loss": 46.0,
-      "step": 7030
-    },
-    {
-      "epoch": 0.5375690502131238,
-      "grad_norm": 0.001516816671937704,
-      "learning_rate": 0.00019999985779902478,
-      "loss": 46.0,
-      "step": 7031
-    },
-    {
-      "epoch": 0.5376455071965136,
-      "grad_norm": 0.007988082244992256,
-      "learning_rate": 0.00019999985775851457,
-      "loss": 46.0,
-      "step": 7032
-    },
-    {
-      "epoch": 0.5377219641799033,
-      "grad_norm": 0.0012250806903466582,
-      "learning_rate": 0.00019999985771799862,
-      "loss": 46.0,
-      "step": 7033
-    },
-    {
-      "epoch": 0.537798421163293,
-      "grad_norm": 0.0036108752246946096,
-      "learning_rate": 0.00019999985767747686,
-      "loss": 46.0,
-      "step": 7034
-    },
-    {
-      "epoch": 0.5378748781466828,
-      "grad_norm": 0.0016761202132329345,
-      "learning_rate": 0.00019999985763694935,
-      "loss": 46.0,
-      "step": 7035
-    },
-    {
-      "epoch": 0.5379513351300724,
-      "grad_norm": 0.0017803935334086418,
-      "learning_rate": 0.00019999985759641608,
-      "loss": 46.0,
-      "step": 7036
-    },
-    {
-      "epoch": 0.5380277921134622,
-      "grad_norm": 0.001953388098627329,
-      "learning_rate": 0.00019999985755587703,
-      "loss": 46.0,
-      "step": 7037
-    },
-    {
-      "epoch": 0.5381042490968518,
-      "grad_norm": 0.0007448914693668485,
-      "learning_rate": 0.0001999998575153322,
-      "loss": 46.0,
-      "step": 7038
-    },
-    {
-      "epoch": 0.5381807060802416,
-      "grad_norm": 0.0007843797211535275,
-      "learning_rate": 0.00019999985747478163,
-      "loss": 46.0,
-      "step": 7039
-    },
-    {
-      "epoch": 0.5382571630636314,
-      "grad_norm": 0.0031101207714527845,
-      "learning_rate": 0.00019999985743422526,
-      "loss": 46.0,
-      "step": 7040
-    },
-    {
-      "epoch": 0.538333620047021,
-      "grad_norm": 0.00283940602093935,
-      "learning_rate": 0.00019999985739366314,
-      "loss": 46.0,
-      "step": 7041
-    },
-    {
-      "epoch": 0.5384100770304108,
-      "grad_norm": 0.00042539872811175883,
-      "learning_rate": 0.00019999985735309522,
-      "loss": 46.0,
-      "step": 7042
-    },
-    {
-      "epoch": 0.5384865340138005,
-      "grad_norm": 0.00046194077003747225,
-      "learning_rate": 0.00019999985731252156,
-      "loss": 46.0,
-      "step": 7043
-    },
-    {
-      "epoch": 0.5385629909971902,
-      "grad_norm": 0.0007489009294658899,
-      "learning_rate": 0.00019999985727194212,
-      "loss": 46.0,
-      "step": 7044
-    },
-    {
-      "epoch": 0.5386394479805799,
-      "grad_norm": 0.0006209298153407872,
-      "learning_rate": 0.00019999985723135694,
-      "loss": 46.0,
-      "step": 7045
-    },
-    {
-      "epoch": 0.5387159049639696,
-      "grad_norm": 0.0036026679445058107,
-      "learning_rate": 0.00019999985719076595,
-      "loss": 46.0,
-      "step": 7046
-    },
-    {
-      "epoch": 0.5387923619473594,
-      "grad_norm": 0.002750386716797948,
-      "learning_rate": 0.0001999998571501692,
-      "loss": 46.0,
-      "step": 7047
-    },
-    {
-      "epoch": 0.5388688189307491,
-      "grad_norm": 0.0005559534765779972,
-      "learning_rate": 0.0001999998571095667,
-      "loss": 46.0,
-      "step": 7048
-    },
-    {
-      "epoch": 0.5389452759141388,
-      "grad_norm": 0.0007957906927913427,
-      "learning_rate": 0.00019999985706895841,
-      "loss": 46.0,
-      "step": 7049
-    },
-    {
-      "epoch": 0.5390217328975285,
-      "grad_norm": 0.0021421632263809443,
-      "learning_rate": 0.00019999985702834434,
-      "loss": 46.0,
-      "step": 7050
-    },
-    {
-      "epoch": 0.5390981898809183,
-      "grad_norm": 0.0006082300678826869,
-      "learning_rate": 0.0001999998569877245,
-      "loss": 46.0,
-      "step": 7051
-    },
-    {
-      "epoch": 0.5391746468643079,
-      "grad_norm": 0.003624261822551489,
-      "learning_rate": 0.00019999985694709894,
-      "loss": 46.0,
-      "step": 7052
-    },
-    {
-      "epoch": 0.5392511038476977,
-      "grad_norm": 0.002115654991939664,
-      "learning_rate": 0.00019999985690646757,
-      "loss": 46.0,
-      "step": 7053
-    },
-    {
-      "epoch": 0.5393275608310875,
-      "grad_norm": 0.0008541772840544581,
-      "learning_rate": 0.00019999985686583043,
-      "loss": 46.0,
-      "step": 7054
-    },
-    {
-      "epoch": 0.5394040178144771,
-      "grad_norm": 0.00039037197711877525,
-      "learning_rate": 0.00019999985682518754,
-      "loss": 46.0,
-      "step": 7055
-    },
-    {
-      "epoch": 0.5394804747978669,
-      "grad_norm": 0.0018750993767753243,
-      "learning_rate": 0.00019999985678453887,
-      "loss": 46.0,
-      "step": 7056
-    },
-    {
-      "epoch": 0.5395569317812565,
-      "grad_norm": 0.005054390523582697,
-      "learning_rate": 0.0001999998567438844,
-      "loss": 46.0,
-      "step": 7057
-    },
-    {
-      "epoch": 0.5396333887646463,
-      "grad_norm": 0.002898718463256955,
-      "learning_rate": 0.00019999985670322422,
-      "loss": 46.0,
-      "step": 7058
-    },
-    {
-      "epoch": 0.539709845748036,
-      "grad_norm": 0.0029946218710392714,
-      "learning_rate": 0.00019999985666255824,
-      "loss": 46.0,
-      "step": 7059
-    },
-    {
-      "epoch": 0.5397863027314257,
-      "grad_norm": 0.0005702909547835588,
-      "learning_rate": 0.00019999985662188648,
-      "loss": 46.0,
-      "step": 7060
-    },
-    {
-      "epoch": 0.5398627597148155,
-      "grad_norm": 0.001118018408305943,
-      "learning_rate": 0.00019999985658120898,
-      "loss": 46.0,
-      "step": 7061
-    },
-    {
-      "epoch": 0.5399392166982052,
-      "grad_norm": 0.002047475427389145,
-      "learning_rate": 0.0001999998565405257,
-      "loss": 46.0,
-      "step": 7062
-    },
-    {
-      "epoch": 0.5400156736815949,
-      "grad_norm": 0.0029113979544490576,
-      "learning_rate": 0.00019999985649983662,
-      "loss": 46.0,
-      "step": 7063
-    },
-    {
-      "epoch": 0.5400921306649846,
-      "grad_norm": 0.0015522849280387163,
-      "learning_rate": 0.0001999998564591418,
-      "loss": 46.0,
-      "step": 7064
-    },
-    {
-      "epoch": 0.5401685876483744,
-      "grad_norm": 0.0013265385059639812,
-      "learning_rate": 0.0001999998564184412,
-      "loss": 46.0,
-      "step": 7065
-    },
-    {
-      "epoch": 0.540245044631764,
-      "grad_norm": 0.0007060448406264186,
-      "learning_rate": 0.00019999985637773486,
-      "loss": 46.0,
-      "step": 7066
-    },
-    {
-      "epoch": 0.5403215016151538,
-      "grad_norm": 0.0045955488458275795,
-      "learning_rate": 0.00019999985633702272,
-      "loss": 46.0,
-      "step": 7067
-    },
-    {
-      "epoch": 0.5403979585985434,
-      "grad_norm": 0.004033868201076984,
-      "learning_rate": 0.0001999998562963048,
-      "loss": 46.0,
-      "step": 7068
-    },
-    {
-      "epoch": 0.5404744155819332,
-      "grad_norm": 0.0019121209625154734,
-      "learning_rate": 0.00019999985625558116,
-      "loss": 46.0,
-      "step": 7069
-    },
-    {
-      "epoch": 0.540550872565323,
-      "grad_norm": 0.0013592569157481194,
-      "learning_rate": 0.0001999998562148517,
-      "loss": 46.0,
-      "step": 7070
-    },
-    {
-      "epoch": 0.5406273295487126,
-      "grad_norm": 0.0008571575162932277,
-      "learning_rate": 0.0001999998561741165,
-      "loss": 46.0,
-      "step": 7071
-    },
-    {
-      "epoch": 0.5407037865321024,
-      "grad_norm": 0.0030802690889686346,
-      "learning_rate": 0.0001999998561333755,
-      "loss": 46.0,
-      "step": 7072
-    },
-    {
-      "epoch": 0.5407802435154921,
-      "grad_norm": 0.0008141295402310789,
-      "learning_rate": 0.00019999985609262875,
-      "loss": 46.0,
-      "step": 7073
-    },
-    {
-      "epoch": 0.5408567004988818,
-      "grad_norm": 0.005090365651994944,
-      "learning_rate": 0.00019999985605187624,
-      "loss": 46.0,
-      "step": 7074
-    },
-    {
-      "epoch": 0.5409331574822716,
-      "grad_norm": 0.0018800608813762665,
-      "learning_rate": 0.00019999985601111794,
-      "loss": 46.0,
-      "step": 7075
-    },
-    {
-      "epoch": 0.5410096144656612,
-      "grad_norm": 0.0007478751358576119,
-      "learning_rate": 0.0001999998559703539,
-      "loss": 46.0,
-      "step": 7076
-    },
-    {
-      "epoch": 0.541086071449051,
-      "grad_norm": 0.0018071256345137954,
-      "learning_rate": 0.00019999985592958407,
-      "loss": 46.0,
-      "step": 7077
-    },
-    {
-      "epoch": 0.5411625284324407,
-      "grad_norm": 0.003690737532451749,
-      "learning_rate": 0.00019999985588880844,
-      "loss": 46.0,
-      "step": 7078
-    },
-    {
-      "epoch": 0.5412389854158304,
-      "grad_norm": 0.0018988515948876739,
-      "learning_rate": 0.0001999998558480271,
-      "loss": 46.0,
-      "step": 7079
-    },
-    {
-      "epoch": 0.5413154423992201,
-      "grad_norm": 0.0011649978114292026,
-      "learning_rate": 0.00019999985580723996,
-      "loss": 46.0,
-      "step": 7080
-    },
-    {
-      "epoch": 0.5413918993826099,
-      "grad_norm": 0.0005884484271518886,
-      "learning_rate": 0.00019999985576644704,
-      "loss": 46.0,
-      "step": 7081
-    },
-    {
-      "epoch": 0.5414683563659995,
-      "grad_norm": 0.0013894907897338271,
-      "learning_rate": 0.00019999985572564838,
-      "loss": 46.0,
-      "step": 7082
-    },
-    {
-      "epoch": 0.5415448133493893,
-      "grad_norm": 0.00674440385773778,
-      "learning_rate": 0.0001999998556848439,
-      "loss": 46.0,
-      "step": 7083
-    },
-    {
-      "epoch": 0.5416212703327791,
-      "grad_norm": 0.003660250222310424,
-      "learning_rate": 0.00019999985564403373,
-      "loss": 46.0,
-      "step": 7084
-    },
-    {
-      "epoch": 0.5416977273161687,
-      "grad_norm": 0.0015356509247794747,
-      "learning_rate": 0.00019999985560321775,
-      "loss": 46.0,
-      "step": 7085
-    },
-    {
-      "epoch": 0.5417741842995585,
-      "grad_norm": 0.0028624965343624353,
-      "learning_rate": 0.000199999855562396,
-      "loss": 46.0,
-      "step": 7086
-    },
-    {
-      "epoch": 0.5418506412829481,
-      "grad_norm": 0.0012801277916878462,
-      "learning_rate": 0.00019999985552156846,
-      "loss": 46.0,
-      "step": 7087
-    },
-    {
-      "epoch": 0.5419270982663379,
-      "grad_norm": 0.0008986828033812344,
-      "learning_rate": 0.00019999985548073515,
-      "loss": 46.0,
-      "step": 7088
-    },
-    {
-      "epoch": 0.5420035552497277,
-      "grad_norm": 0.0015479095745831728,
-      "learning_rate": 0.0001999998554398961,
-      "loss": 46.0,
-      "step": 7089
-    },
-    {
-      "epoch": 0.5420800122331173,
-      "grad_norm": 0.0010830943938344717,
-      "learning_rate": 0.00019999985539905128,
-      "loss": 46.0,
-      "step": 7090
-    },
-    {
-      "epoch": 0.5421564692165071,
-      "grad_norm": 0.0008617498679086566,
-      "learning_rate": 0.00019999985535820069,
-      "loss": 46.0,
-      "step": 7091
-    },
-    {
-      "epoch": 0.5422329261998968,
-      "grad_norm": 0.00154385797213763,
-      "learning_rate": 0.00019999985531734432,
-      "loss": 46.0,
-      "step": 7092
-    },
-    {
-      "epoch": 0.5423093831832865,
-      "grad_norm": 0.0031907237134873867,
-      "learning_rate": 0.00019999985527648217,
-      "loss": 46.0,
-      "step": 7093
-    },
-    {
-      "epoch": 0.5423858401666762,
-      "grad_norm": 0.0016584784025326371,
-      "learning_rate": 0.00019999985523561426,
-      "loss": 46.0,
-      "step": 7094
-    },
-    {
-      "epoch": 0.542462297150066,
-      "grad_norm": 0.001628034166060388,
-      "learning_rate": 0.0001999998551947406,
-      "loss": 46.0,
-      "step": 7095
-    },
-    {
-      "epoch": 0.5425387541334556,
-      "grad_norm": 0.002685130573809147,
-      "learning_rate": 0.00019999985515386113,
-      "loss": 46.0,
-      "step": 7096
-    },
-    {
-      "epoch": 0.5426152111168454,
-      "grad_norm": 0.003082075621932745,
-      "learning_rate": 0.0001999998551129759,
-      "loss": 46.0,
-      "step": 7097
-    },
-    {
-      "epoch": 0.5426916681002351,
-      "grad_norm": 0.0004978615907020867,
-      "learning_rate": 0.0001999998550720849,
-      "loss": 46.0,
-      "step": 7098
-    },
-    {
-      "epoch": 0.5427681250836248,
-      "grad_norm": 0.00067055260296911,
-      "learning_rate": 0.00019999985503118815,
-      "loss": 46.0,
-      "step": 7099
-    },
-    {
-      "epoch": 0.5428445820670146,
-      "grad_norm": 0.0007644226425327361,
-      "learning_rate": 0.00019999985499028565,
-      "loss": 46.0,
-      "step": 7100
-    },
-    {
-      "epoch": 0.5429210390504042,
-      "grad_norm": 0.004957432858645916,
-      "learning_rate": 0.00019999985494937735,
-      "loss": 46.0,
-      "step": 7101
-    },
-    {
-      "epoch": 0.542997496033794,
-      "grad_norm": 0.0004039366904180497,
-      "learning_rate": 0.00019999985490846327,
-      "loss": 46.0,
-      "step": 7102
-    },
-    {
-      "epoch": 0.5430739530171838,
-      "grad_norm": 0.0006158456089906394,
-      "learning_rate": 0.00019999985486754345,
-      "loss": 46.0,
-      "step": 7103
-    },
-    {
-      "epoch": 0.5431504100005734,
-      "grad_norm": 0.0005012454930692911,
-      "learning_rate": 0.00019999985482661785,
-      "loss": 46.0,
-      "step": 7104
-    },
-    {
-      "epoch": 0.5432268669839632,
-      "grad_norm": 0.0004433618451002985,
-      "learning_rate": 0.00019999985478568648,
-      "loss": 46.0,
-      "step": 7105
-    },
-    {
-      "epoch": 0.5433033239673528,
-      "grad_norm": 0.0007532319286838174,
-      "learning_rate": 0.00019999985474474934,
-      "loss": 46.0,
-      "step": 7106
-    },
-    {
-      "epoch": 0.5433797809507426,
-      "grad_norm": 0.0008327563991770148,
-      "learning_rate": 0.00019999985470380642,
-      "loss": 46.0,
-      "step": 7107
-    },
-    {
-      "epoch": 0.5434562379341323,
-      "grad_norm": 0.0008429056033492088,
-      "learning_rate": 0.00019999985466285773,
-      "loss": 46.0,
-      "step": 7108
-    },
-    {
-      "epoch": 0.543532694917522,
-      "grad_norm": 0.0013189580058678985,
-      "learning_rate": 0.0001999998546219033,
-      "loss": 46.0,
-      "step": 7109
-    },
-    {
-      "epoch": 0.5436091519009117,
-      "grad_norm": 0.0012941764434799552,
-      "learning_rate": 0.0001999998545809431,
-      "loss": 46.0,
-      "step": 7110
-    },
-    {
-      "epoch": 0.5436856088843015,
-      "grad_norm": 0.0017279449384659529,
-      "learning_rate": 0.0001999998545399771,
-      "loss": 46.0,
-      "step": 7111
-    },
-    {
-      "epoch": 0.5437620658676912,
-      "grad_norm": 0.001241541001945734,
-      "learning_rate": 0.00019999985449900535,
-      "loss": 46.0,
-      "step": 7112
-    },
-    {
-      "epoch": 0.5438385228510809,
-      "grad_norm": 0.002491633640602231,
-      "learning_rate": 0.0001999998544580278,
-      "loss": 46.0,
-      "step": 7113
-    },
-    {
-      "epoch": 0.5439149798344707,
-      "grad_norm": 0.000936816621106118,
-      "learning_rate": 0.0001999998544170445,
-      "loss": 46.0,
-      "step": 7114
-    },
-    {
-      "epoch": 0.5439914368178603,
-      "grad_norm": 0.0014724023640155792,
-      "learning_rate": 0.0001999998543760554,
-      "loss": 46.0,
-      "step": 7115
-    },
-    {
-      "epoch": 0.5440678938012501,
-      "grad_norm": 0.0041054049506783485,
-      "learning_rate": 0.0001999998543350606,
-      "loss": 46.0,
-      "step": 7116
-    },
-    {
-      "epoch": 0.5441443507846397,
-      "grad_norm": 0.002560204826295376,
-      "learning_rate": 0.00019999985429406,
-      "loss": 46.0,
-      "step": 7117
-    },
-    {
-      "epoch": 0.5442208077680295,
-      "grad_norm": 0.0015505847986787558,
-      "learning_rate": 0.0001999998542530536,
-      "loss": 46.0,
-      "step": 7118
-    },
-    {
-      "epoch": 0.5442972647514193,
-      "grad_norm": 0.0005005880375392735,
-      "learning_rate": 0.00019999985421204145,
-      "loss": 46.0,
-      "step": 7119
-    },
-    {
-      "epoch": 0.5443737217348089,
-      "grad_norm": 0.0006232415325939655,
-      "learning_rate": 0.00019999985417102357,
-      "loss": 46.0,
-      "step": 7120
-    },
-    {
-      "epoch": 0.5444501787181987,
-      "grad_norm": 0.0017121481942012906,
-      "learning_rate": 0.00019999985412999988,
-      "loss": 46.0,
-      "step": 7121
-    },
-    {
-      "epoch": 0.5445266357015884,
-      "grad_norm": 0.0010269352933391929,
-      "learning_rate": 0.0001999998540889704,
-      "loss": 46.0,
-      "step": 7122
-    },
-    {
-      "epoch": 0.5446030926849781,
-      "grad_norm": 0.0024696041364222765,
-      "learning_rate": 0.00019999985404793518,
-      "loss": 46.0,
-      "step": 7123
-    },
-    {
-      "epoch": 0.5446795496683678,
-      "grad_norm": 0.0011527878232300282,
-      "learning_rate": 0.0001999998540068942,
-      "loss": 46.0,
-      "step": 7124
-    },
-    {
-      "epoch": 0.5447560066517576,
-      "grad_norm": 0.001106124953366816,
-      "learning_rate": 0.00019999985396584744,
-      "loss": 46.0,
-      "step": 7125
-    },
-    {
-      "epoch": 0.5448324636351473,
-      "grad_norm": 0.0011542091378942132,
-      "learning_rate": 0.0001999998539247949,
-      "loss": 46.0,
-      "step": 7126
-    },
-    {
-      "epoch": 0.544908920618537,
-      "grad_norm": 0.0015792502090334892,
-      "learning_rate": 0.0001999998538837366,
-      "loss": 46.0,
-      "step": 7127
-    },
-    {
-      "epoch": 0.5449853776019267,
-      "grad_norm": 0.0016737999394536018,
-      "learning_rate": 0.00019999985384267253,
-      "loss": 46.0,
-      "step": 7128
-    },
-    {
-      "epoch": 0.5450618345853164,
-      "grad_norm": 0.001238030381500721,
-      "learning_rate": 0.0001999998538016027,
-      "loss": 46.0,
-      "step": 7129
-    },
-    {
-      "epoch": 0.5451382915687062,
-      "grad_norm": 0.0007080962532199919,
-      "learning_rate": 0.00019999985376052709,
-      "loss": 46.0,
-      "step": 7130
-    },
-    {
-      "epoch": 0.5452147485520958,
-      "grad_norm": 0.0005658137379214168,
-      "learning_rate": 0.00019999985371944572,
-      "loss": 46.0,
-      "step": 7131
-    },
-    {
-      "epoch": 0.5452912055354856,
-      "grad_norm": 0.0008155421819537878,
-      "learning_rate": 0.00019999985367835857,
-      "loss": 46.0,
-      "step": 7132
-    },
-    {
-      "epoch": 0.5453676625188754,
-      "grad_norm": 0.0021908506751060486,
-      "learning_rate": 0.00019999985363726563,
-      "loss": 46.0,
-      "step": 7133
-    },
-    {
-      "epoch": 0.545444119502265,
-      "grad_norm": 0.0006705531850457191,
-      "learning_rate": 0.00019999985359616694,
-      "loss": 46.0,
-      "step": 7134
-    },
-    {
-      "epoch": 0.5455205764856548,
-      "grad_norm": 0.0010617149528115988,
-      "learning_rate": 0.0001999998535550625,
-      "loss": 46.0,
-      "step": 7135
-    },
-    {
-      "epoch": 0.5455970334690444,
-      "grad_norm": 0.0013305466854944825,
-      "learning_rate": 0.00019999985351395227,
-      "loss": 46.0,
-      "step": 7136
-    },
-    {
-      "epoch": 0.5456734904524342,
-      "grad_norm": 0.0020592212677001953,
-      "learning_rate": 0.00019999985347283626,
-      "loss": 46.0,
-      "step": 7137
-    },
-    {
-      "epoch": 0.545749947435824,
-      "grad_norm": 0.007282910868525505,
-      "learning_rate": 0.0001999998534317145,
-      "loss": 46.0,
-      "step": 7138
-    },
-    {
-      "epoch": 0.5458264044192136,
-      "grad_norm": 0.0008885685238055885,
-      "learning_rate": 0.00019999985339058698,
-      "loss": 46.0,
-      "step": 7139
-    },
-    {
-      "epoch": 0.5459028614026034,
-      "grad_norm": 0.0005298163741827011,
-      "learning_rate": 0.00019999985334945368,
-      "loss": 46.0,
-      "step": 7140
-    },
-    {
-      "epoch": 0.5459793183859931,
-      "grad_norm": 0.0013140361988916993,
-      "learning_rate": 0.0001999998533083146,
-      "loss": 46.0,
-      "step": 7141
-    },
-    {
-      "epoch": 0.5460557753693828,
-      "grad_norm": 0.0007046398241072893,
-      "learning_rate": 0.00019999985326716978,
-      "loss": 46.0,
-      "step": 7142
-    },
-    {
-      "epoch": 0.5461322323527725,
-      "grad_norm": 0.0008345639798790216,
-      "learning_rate": 0.00019999985322601916,
-      "loss": 46.0,
-      "step": 7143
-    },
-    {
-      "epoch": 0.5462086893361623,
-      "grad_norm": 0.005981791764497757,
-      "learning_rate": 0.00019999985318486276,
-      "loss": 46.0,
-      "step": 7144
-    },
-    {
-      "epoch": 0.5462851463195519,
-      "grad_norm": 0.0032875980250537395,
-      "learning_rate": 0.00019999985314370062,
-      "loss": 46.0,
-      "step": 7145
-    },
-    {
-      "epoch": 0.5463616033029417,
-      "grad_norm": 0.0009305764106102288,
-      "learning_rate": 0.0001999998531025327,
-      "loss": 46.0,
-      "step": 7146
-    },
-    {
-      "epoch": 0.5464380602863314,
-      "grad_norm": 0.0010729838395491242,
-      "learning_rate": 0.000199999853061359,
-      "loss": 46.0,
-      "step": 7147
-    },
-    {
-      "epoch": 0.5465145172697211,
-      "grad_norm": 0.0007016872405074537,
-      "learning_rate": 0.00019999985302017955,
-      "loss": 46.0,
-      "step": 7148
-    },
-    {
-      "epoch": 0.5465909742531109,
-      "grad_norm": 0.001024897675961256,
-      "learning_rate": 0.00019999985297899432,
-      "loss": 46.0,
-      "step": 7149
-    },
-    {
-      "epoch": 0.5466674312365005,
-      "grad_norm": 0.0007103008101694286,
-      "learning_rate": 0.00019999985293780333,
-      "loss": 46.0,
-      "step": 7150
-    },
-    {
-      "epoch": 0.5467438882198903,
-      "grad_norm": 0.0005859124939888716,
-      "learning_rate": 0.00019999985289660655,
-      "loss": 46.0,
-      "step": 7151
-    },
-    {
-      "epoch": 0.54682034520328,
-      "grad_norm": 0.0006008106283843517,
-      "learning_rate": 0.00019999985285540402,
-      "loss": 46.0,
-      "step": 7152
-    },
-    {
-      "epoch": 0.5468968021866697,
-      "grad_norm": 0.0008637873688712716,
-      "learning_rate": 0.00019999985281419572,
-      "loss": 46.0,
-      "step": 7153
-    },
-    {
-      "epoch": 0.5469732591700595,
-      "grad_norm": 0.0007222691201604903,
-      "learning_rate": 0.00019999985277298165,
-      "loss": 46.0,
-      "step": 7154
-    },
-    {
-      "epoch": 0.5470497161534492,
-      "grad_norm": 0.002003852976486087,
-      "learning_rate": 0.0001999998527317618,
-      "loss": 46.0,
-      "step": 7155
-    },
-    {
-      "epoch": 0.5471261731368389,
-      "grad_norm": 0.0022850194945931435,
-      "learning_rate": 0.00019999985269053617,
-      "loss": 46.0,
-      "step": 7156
-    },
-    {
-      "epoch": 0.5472026301202286,
-      "grad_norm": 0.000572386896237731,
-      "learning_rate": 0.0001999998526493048,
-      "loss": 46.0,
-      "step": 7157
-    },
-    {
-      "epoch": 0.5472790871036183,
-      "grad_norm": 0.002787049626931548,
-      "learning_rate": 0.00019999985260806764,
-      "loss": 46.0,
-      "step": 7158
-    },
-    {
-      "epoch": 0.547355544087008,
-      "grad_norm": 0.0012862762669101357,
-      "learning_rate": 0.00019999985256682472,
-      "loss": 46.0,
-      "step": 7159
-    },
-    {
-      "epoch": 0.5474320010703978,
-      "grad_norm": 0.0011167486663907766,
-      "learning_rate": 0.00019999985252557603,
-      "loss": 46.0,
-      "step": 7160
-    },
-    {
-      "epoch": 0.5475084580537875,
-      "grad_norm": 0.0021482512820512056,
-      "learning_rate": 0.00019999985248432157,
-      "loss": 46.0,
-      "step": 7161
-    },
-    {
-      "epoch": 0.5475849150371772,
-      "grad_norm": 0.0014115293743088841,
-      "learning_rate": 0.00019999985244306134,
-      "loss": 46.0,
-      "step": 7162
-    },
-    {
-      "epoch": 0.547661372020567,
-      "grad_norm": 0.0011674423003569245,
-      "learning_rate": 0.00019999985240179533,
-      "loss": 46.0,
-      "step": 7163
-    },
-    {
-      "epoch": 0.5477378290039566,
-      "grad_norm": 0.0008137841359712183,
-      "learning_rate": 0.00019999985236052357,
-      "loss": 46.0,
-      "step": 7164
-    },
-    {
-      "epoch": 0.5478142859873464,
-      "grad_norm": 0.0005121002322994173,
-      "learning_rate": 0.00019999985231924602,
-      "loss": 46.0,
-      "step": 7165
-    },
-    {
-      "epoch": 0.5478907429707361,
-      "grad_norm": 0.001126022427342832,
-      "learning_rate": 0.00019999985227796272,
-      "loss": 46.0,
-      "step": 7166
-    },
-    {
-      "epoch": 0.5479671999541258,
-      "grad_norm": 0.00155731406994164,
-      "learning_rate": 0.00019999985223667364,
-      "loss": 46.0,
-      "step": 7167
-    },
-    {
-      "epoch": 0.5480436569375156,
-      "grad_norm": 0.0028543006628751755,
-      "learning_rate": 0.0001999998521953788,
-      "loss": 46.0,
-      "step": 7168
-    },
-    {
-      "epoch": 0.5481201139209052,
-      "grad_norm": 0.0006537465960718691,
-      "learning_rate": 0.00019999985215407817,
-      "loss": 46.0,
-      "step": 7169
-    },
-    {
-      "epoch": 0.548196570904295,
-      "grad_norm": 0.00226815277710557,
-      "learning_rate": 0.00019999985211277178,
-      "loss": 46.0,
-      "step": 7170
-    },
-    {
-      "epoch": 0.5482730278876847,
-      "grad_norm": 0.0025133562739938498,
-      "learning_rate": 0.0001999998520714596,
-      "loss": 46.0,
-      "step": 7171
-    },
-    {
-      "epoch": 0.5483494848710744,
-      "grad_norm": 0.0014155980898067355,
-      "learning_rate": 0.0001999998520301417,
-      "loss": 46.0,
-      "step": 7172
-    },
-    {
-      "epoch": 0.5484259418544641,
-      "grad_norm": 0.004426279105246067,
-      "learning_rate": 0.000199999851988818,
-      "loss": 46.0,
-      "step": 7173
-    },
-    {
-      "epoch": 0.5485023988378539,
-      "grad_norm": 0.0005192711832933128,
-      "learning_rate": 0.00019999985194748857,
-      "loss": 46.0,
-      "step": 7174
-    },
-    {
-      "epoch": 0.5485788558212435,
-      "grad_norm": 0.0006723931874148548,
-      "learning_rate": 0.0001999998519061533,
-      "loss": 46.0,
-      "step": 7175
-    },
-    {
-      "epoch": 0.5486553128046333,
-      "grad_norm": 0.0010698664700612426,
-      "learning_rate": 0.0001999998518648123,
-      "loss": 46.0,
-      "step": 7176
-    },
-    {
-      "epoch": 0.548731769788023,
-      "grad_norm": 0.0011616181582212448,
-      "learning_rate": 0.00019999985182346554,
-      "loss": 46.0,
-      "step": 7177
-    },
-    {
-      "epoch": 0.5488082267714127,
-      "grad_norm": 0.0011422683019191027,
-      "learning_rate": 0.000199999851782113,
-      "loss": 46.0,
-      "step": 7178
-    },
-    {
-      "epoch": 0.5488846837548025,
-      "grad_norm": 0.0023996406234800816,
-      "learning_rate": 0.0001999998517407547,
-      "loss": 46.0,
-      "step": 7179
-    },
-    {
-      "epoch": 0.5489611407381921,
-      "grad_norm": 0.0007847588276490569,
-      "learning_rate": 0.0001999998516993906,
-      "loss": 46.0,
-      "step": 7180
-    },
-    {
-      "epoch": 0.5490375977215819,
-      "grad_norm": 0.0026907974388450384,
-      "learning_rate": 0.00019999985165802074,
-      "loss": 46.0,
-      "step": 7181
-    },
-    {
-      "epoch": 0.5491140547049717,
-      "grad_norm": 0.0020355312153697014,
-      "learning_rate": 0.00019999985161664512,
-      "loss": 46.0,
-      "step": 7182
-    },
-    {
-      "epoch": 0.5491905116883613,
-      "grad_norm": 0.0010263108415529132,
-      "learning_rate": 0.00019999985157526372,
-      "loss": 46.0,
-      "step": 7183
-    },
-    {
-      "epoch": 0.5492669686717511,
-      "grad_norm": 0.0010934211313724518,
-      "learning_rate": 0.00019999985153387658,
-      "loss": 46.0,
-      "step": 7184
-    },
-    {
-      "epoch": 0.5493434256551408,
-      "grad_norm": 0.0007517405902035534,
-      "learning_rate": 0.00019999985149248364,
-      "loss": 46.0,
-      "step": 7185
-    },
-    {
-      "epoch": 0.5494198826385305,
-      "grad_norm": 0.0005203973269090056,
-      "learning_rate": 0.00019999985145108495,
-      "loss": 46.0,
-      "step": 7186
-    },
-    {
-      "epoch": 0.5494963396219202,
-      "grad_norm": 0.002755733672529459,
-      "learning_rate": 0.0001999998514096805,
-      "loss": 46.0,
-      "step": 7187
-    },
-    {
-      "epoch": 0.5495727966053099,
-      "grad_norm": 0.004364221356809139,
-      "learning_rate": 0.00019999985136827023,
-      "loss": 46.0,
-      "step": 7188
-    },
-    {
-      "epoch": 0.5496492535886996,
-      "grad_norm": 0.0014792496804147959,
-      "learning_rate": 0.00019999985132685425,
-      "loss": 46.0,
-      "step": 7189
-    },
-    {
-      "epoch": 0.5497257105720894,
-      "grad_norm": 0.00044737933785654604,
-      "learning_rate": 0.00019999985128543247,
-      "loss": 46.0,
-      "step": 7190
-    },
-    {
-      "epoch": 0.5498021675554791,
-      "grad_norm": 0.000552560028154403,
-      "learning_rate": 0.00019999985124400494,
-      "loss": 46.0,
-      "step": 7191
-    },
-    {
-      "epoch": 0.5498786245388688,
-      "grad_norm": 0.0008303321083076298,
-      "learning_rate": 0.0001999998512025716,
-      "loss": 46.0,
-      "step": 7192
-    },
-    {
-      "epoch": 0.5499550815222586,
-      "grad_norm": 0.0013077287003397942,
-      "learning_rate": 0.0001999998511611325,
-      "loss": 46.0,
-      "step": 7193
-    },
-    {
-      "epoch": 0.5500315385056482,
-      "grad_norm": 0.0029086717404425144,
-      "learning_rate": 0.00019999985111968766,
-      "loss": 46.0,
-      "step": 7194
-    },
-    {
-      "epoch": 0.550107995489038,
-      "grad_norm": 0.001223950763233006,
-      "learning_rate": 0.00019999985107823704,
-      "loss": 46.0,
-      "step": 7195
-    },
-    {
-      "epoch": 0.5501844524724278,
-      "grad_norm": 0.008994979783892632,
-      "learning_rate": 0.00019999985103678065,
-      "loss": 46.0,
-      "step": 7196
-    },
-    {
-      "epoch": 0.5502609094558174,
-      "grad_norm": 0.0031687754672020674,
-      "learning_rate": 0.00019999985099531848,
-      "loss": 46.0,
-      "step": 7197
-    },
-    {
-      "epoch": 0.5503373664392072,
-      "grad_norm": 0.0014216335257515311,
-      "learning_rate": 0.00019999985095385056,
-      "loss": 46.0,
-      "step": 7198
-    },
-    {
-      "epoch": 0.5504138234225968,
-      "grad_norm": 0.0010007154196500778,
-      "learning_rate": 0.00019999985091237688,
-      "loss": 46.0,
-      "step": 7199
-    },
-    {
-      "epoch": 0.5504902804059866,
-      "grad_norm": 0.0035625926684588194,
-      "learning_rate": 0.0001999998508708974,
-      "loss": 46.0,
-      "step": 7200
-    },
-    {
-      "epoch": 0.5505667373893763,
-      "grad_norm": 0.002085289219394326,
-      "learning_rate": 0.00019999985082941215,
-      "loss": 46.0,
-      "step": 7201
-    },
-    {
-      "epoch": 0.550643194372766,
-      "grad_norm": 0.004990634508430958,
-      "learning_rate": 0.00019999985078792115,
-      "loss": 46.0,
-      "step": 7202
-    },
-    {
-      "epoch": 0.5507196513561557,
-      "grad_norm": 0.0041598547250032425,
-      "learning_rate": 0.00019999985074642437,
-      "loss": 46.0,
-      "step": 7203
-    },
-    {
-      "epoch": 0.5507961083395455,
-      "grad_norm": 0.000868777628056705,
-      "learning_rate": 0.0001999998507049218,
-      "loss": 46.0,
-      "step": 7204
-    },
-    {
-      "epoch": 0.5508725653229352,
-      "grad_norm": 0.0008237641886807978,
-      "learning_rate": 0.00019999985066341348,
-      "loss": 46.0,
-      "step": 7205
-    },
-    {
-      "epoch": 0.5509490223063249,
-      "grad_norm": 0.0012723867548629642,
-      "learning_rate": 0.0001999998506218994,
-      "loss": 46.0,
-      "step": 7206
-    },
-    {
-      "epoch": 0.5510254792897146,
-      "grad_norm": 0.0005697850720025599,
-      "learning_rate": 0.00019999985058037956,
-      "loss": 46.0,
-      "step": 7207
-    },
-    {
-      "epoch": 0.5511019362731043,
-      "grad_norm": 0.004647390451282263,
-      "learning_rate": 0.00019999985053885394,
-      "loss": 46.0,
-      "step": 7208
-    },
-    {
-      "epoch": 0.5511783932564941,
-      "grad_norm": 0.0008499856339767575,
-      "learning_rate": 0.00019999985049732252,
-      "loss": 46.0,
-      "step": 7209
-    },
-    {
-      "epoch": 0.5512548502398837,
-      "grad_norm": 0.000891794275958091,
-      "learning_rate": 0.00019999985045578538,
-      "loss": 46.0,
-      "step": 7210
-    },
-    {
-      "epoch": 0.5513313072232735,
-      "grad_norm": 0.0006179591291584074,
-      "learning_rate": 0.00019999985041424244,
-      "loss": 46.0,
-      "step": 7211
-    },
-    {
-      "epoch": 0.5514077642066633,
-      "grad_norm": 0.0005138255655765533,
-      "learning_rate": 0.00019999985037269372,
-      "loss": 46.0,
-      "step": 7212
-    },
-    {
-      "epoch": 0.5514842211900529,
-      "grad_norm": 0.0007363785989582539,
-      "learning_rate": 0.00019999985033113926,
-      "loss": 46.0,
-      "step": 7213
-    },
-    {
-      "epoch": 0.5515606781734427,
-      "grad_norm": 0.0014757682802155614,
-      "learning_rate": 0.00019999985028957903,
-      "loss": 46.0,
-      "step": 7214
-    },
-    {
-      "epoch": 0.5516371351568324,
-      "grad_norm": 0.0030114136170595884,
-      "learning_rate": 0.00019999985024801302,
-      "loss": 46.0,
-      "step": 7215
-    },
-    {
-      "epoch": 0.5517135921402221,
-      "grad_norm": 0.0009294779738411307,
-      "learning_rate": 0.00019999985020644121,
-      "loss": 46.0,
-      "step": 7216
-    },
-    {
-      "epoch": 0.5517900491236118,
-      "grad_norm": 0.0008809330756776035,
-      "learning_rate": 0.00019999985016486366,
-      "loss": 46.0,
-      "step": 7217
-    },
-    {
-      "epoch": 0.5518665061070015,
-      "grad_norm": 0.0016561830416321754,
-      "learning_rate": 0.00019999985012328036,
-      "loss": 46.0,
-      "step": 7218
-    },
-    {
-      "epoch": 0.5519429630903913,
-      "grad_norm": 0.0008260655449703336,
-      "learning_rate": 0.00019999985008169126,
-      "loss": 46.0,
-      "step": 7219
-    },
-    {
-      "epoch": 0.552019420073781,
-      "grad_norm": 0.0005357759655453265,
-      "learning_rate": 0.0001999998500400964,
-      "loss": 46.0,
-      "step": 7220
-    },
-    {
-      "epoch": 0.5520958770571707,
-      "grad_norm": 0.0012031859951093793,
-      "learning_rate": 0.00019999984999849576,
-      "loss": 46.0,
-      "step": 7221
-    },
-    {
-      "epoch": 0.5521723340405604,
-      "grad_norm": 0.00213741441257298,
-      "learning_rate": 0.0001999998499568894,
-      "loss": 46.0,
-      "step": 7222
-    },
-    {
-      "epoch": 0.5522487910239502,
-      "grad_norm": 0.001000562566332519,
-      "learning_rate": 0.00019999984991527723,
-      "loss": 46.0,
-      "step": 7223
-    },
-    {
-      "epoch": 0.5523252480073398,
-      "grad_norm": 0.0037644400727003813,
-      "learning_rate": 0.00019999984987365926,
-      "loss": 46.0,
-      "step": 7224
-    },
-    {
-      "epoch": 0.5524017049907296,
-      "grad_norm": 0.0008638995932415128,
-      "learning_rate": 0.00019999984983203557,
-      "loss": 46.0,
-      "step": 7225
-    },
-    {
-      "epoch": 0.5524781619741194,
-      "grad_norm": 0.0009665734251029789,
-      "learning_rate": 0.0001999998497904061,
-      "loss": 46.0,
-      "step": 7226
-    },
-    {
-      "epoch": 0.552554618957509,
-      "grad_norm": 0.004106669686734676,
-      "learning_rate": 0.00019999984974877085,
-      "loss": 46.0,
-      "step": 7227
-    },
-    {
-      "epoch": 0.5526310759408988,
-      "grad_norm": 0.001969381934031844,
-      "learning_rate": 0.00019999984970712985,
-      "loss": 46.0,
-      "step": 7228
-    },
-    {
-      "epoch": 0.5527075329242884,
-      "grad_norm": 0.0009289950248785317,
-      "learning_rate": 0.00019999984966548304,
-      "loss": 46.0,
-      "step": 7229
-    },
-    {
-      "epoch": 0.5527839899076782,
-      "grad_norm": 0.0009822353022173047,
-      "learning_rate": 0.00019999984962383049,
-      "loss": 46.0,
-      "step": 7230
-    },
-    {
-      "epoch": 0.552860446891068,
-      "grad_norm": 0.0013525179820135236,
-      "learning_rate": 0.00019999984958217219,
-      "loss": 46.0,
-      "step": 7231
-    },
-    {
-      "epoch": 0.5529369038744576,
-      "grad_norm": 0.003662162460386753,
-      "learning_rate": 0.00019999984954050808,
-      "loss": 46.0,
-      "step": 7232
-    },
-    {
-      "epoch": 0.5530133608578474,
-      "grad_norm": 0.0022045017685741186,
-      "learning_rate": 0.0001999998494988382,
-      "loss": 46.0,
-      "step": 7233
-    },
-    {
-      "epoch": 0.5530898178412371,
-      "grad_norm": 0.0016932801809161901,
-      "learning_rate": 0.0001999998494571626,
-      "loss": 46.0,
-      "step": 7234
-    },
-    {
-      "epoch": 0.5531662748246268,
-      "grad_norm": 0.0018906567711383104,
-      "learning_rate": 0.0001999998494154812,
-      "loss": 46.0,
-      "step": 7235
-    },
-    {
-      "epoch": 0.5532427318080165,
-      "grad_norm": 0.0012126766378059983,
-      "learning_rate": 0.000199999849373794,
-      "loss": 46.0,
-      "step": 7236
-    },
-    {
-      "epoch": 0.5533191887914062,
-      "grad_norm": 0.0029551133047789335,
-      "learning_rate": 0.00019999984933210106,
-      "loss": 46.0,
-      "step": 7237
-    },
-    {
-      "epoch": 0.5533956457747959,
-      "grad_norm": 0.0027554978150874376,
-      "learning_rate": 0.00019999984929040238,
-      "loss": 46.0,
-      "step": 7238
-    },
-    {
-      "epoch": 0.5534721027581857,
-      "grad_norm": 0.0003183887165505439,
-      "learning_rate": 0.0001999998492486979,
-      "loss": 46.0,
-      "step": 7239
-    },
-    {
-      "epoch": 0.5535485597415754,
-      "grad_norm": 0.0015790038742125034,
-      "learning_rate": 0.00019999984920698766,
-      "loss": 46.0,
-      "step": 7240
-    },
-    {
-      "epoch": 0.5536250167249651,
-      "grad_norm": 0.0007497859187424183,
-      "learning_rate": 0.00019999984916527162,
-      "loss": 46.0,
-      "step": 7241
-    },
-    {
-      "epoch": 0.5537014737083549,
-      "grad_norm": 0.0012633674778044224,
-      "learning_rate": 0.00019999984912354984,
-      "loss": 46.0,
-      "step": 7242
-    },
-    {
-      "epoch": 0.5537779306917445,
-      "grad_norm": 0.004117726814001799,
-      "learning_rate": 0.0001999998490818223,
-      "loss": 46.0,
-      "step": 7243
-    },
-    {
-      "epoch": 0.5538543876751343,
-      "grad_norm": 0.0010197142837569118,
-      "learning_rate": 0.00019999984904008896,
-      "loss": 46.0,
-      "step": 7244
-    },
-    {
-      "epoch": 0.553930844658524,
-      "grad_norm": 0.0005212892428971827,
-      "learning_rate": 0.00019999984899834986,
-      "loss": 46.0,
-      "step": 7245
-    },
-    {
-      "epoch": 0.5540073016419137,
-      "grad_norm": 0.0004334588011261076,
-      "learning_rate": 0.000199999848956605,
-      "loss": 46.0,
-      "step": 7246
-    },
-    {
-      "epoch": 0.5540837586253035,
-      "grad_norm": 0.001607435755431652,
-      "learning_rate": 0.00019999984891485437,
-      "loss": 46.0,
-      "step": 7247
-    },
-    {
-      "epoch": 0.5541602156086931,
-      "grad_norm": 0.002185296034440398,
-      "learning_rate": 0.00019999984887309795,
-      "loss": 46.0,
-      "step": 7248
-    },
-    {
-      "epoch": 0.5542366725920829,
-      "grad_norm": 0.0016632411861792207,
-      "learning_rate": 0.0001999998488313358,
-      "loss": 46.0,
-      "step": 7249
-    },
-    {
-      "epoch": 0.5543131295754726,
-      "grad_norm": 0.0005494050565175712,
-      "learning_rate": 0.00019999984878956784,
-      "loss": 46.0,
-      "step": 7250
-    },
-    {
-      "epoch": 0.5543895865588623,
-      "grad_norm": 0.0006357296369969845,
-      "learning_rate": 0.00019999984874779413,
-      "loss": 46.0,
-      "step": 7251
-    },
-    {
-      "epoch": 0.554466043542252,
-      "grad_norm": 0.011412625201046467,
-      "learning_rate": 0.00019999984870601464,
-      "loss": 46.0,
-      "step": 7252
-    },
-    {
-      "epoch": 0.5545425005256418,
-      "grad_norm": 0.0039036523085087538,
-      "learning_rate": 0.00019999984866422938,
-      "loss": 46.0,
-      "step": 7253
-    },
-    {
-      "epoch": 0.5546189575090315,
-      "grad_norm": 0.0021096921991556883,
-      "learning_rate": 0.00019999984862243837,
-      "loss": 46.0,
-      "step": 7254
-    },
-    {
-      "epoch": 0.5546954144924212,
-      "grad_norm": 0.0006583980284631252,
-      "learning_rate": 0.00019999984858064157,
-      "loss": 46.0,
-      "step": 7255
-    },
-    {
-      "epoch": 0.554771871475811,
-      "grad_norm": 0.0006962026236578822,
-      "learning_rate": 0.00019999984853883901,
-      "loss": 46.0,
-      "step": 7256
-    },
-    {
-      "epoch": 0.5548483284592006,
-      "grad_norm": 0.0014208783395588398,
-      "learning_rate": 0.00019999984849703071,
-      "loss": 46.0,
-      "step": 7257
-    },
-    {
-      "epoch": 0.5549247854425904,
-      "grad_norm": 0.002004295354709029,
-      "learning_rate": 0.00019999984845521661,
-      "loss": 46.0,
-      "step": 7258
-    },
-    {
-      "epoch": 0.55500124242598,
-      "grad_norm": 0.0006543595809489489,
-      "learning_rate": 0.00019999984841339674,
-      "loss": 46.0,
-      "step": 7259
-    },
-    {
-      "epoch": 0.5550776994093698,
-      "grad_norm": 0.0024939037393778563,
-      "learning_rate": 0.0001999998483715711,
-      "loss": 46.0,
-      "step": 7260
-    },
-    {
-      "epoch": 0.5551541563927596,
-      "grad_norm": 0.006287538446485996,
-      "learning_rate": 0.0001999998483297397,
-      "loss": 46.0,
-      "step": 7261
-    },
-    {
-      "epoch": 0.5552306133761492,
-      "grad_norm": 0.0050195991061627865,
-      "learning_rate": 0.0001999998482879025,
-      "loss": 46.0,
-      "step": 7262
-    },
-    {
-      "epoch": 0.555307070359539,
-      "grad_norm": 0.00040017697028815746,
-      "learning_rate": 0.00019999984824605957,
-      "loss": 46.0,
-      "step": 7263
-    },
-    {
-      "epoch": 0.5553835273429287,
-      "grad_norm": 0.0022776280529797077,
-      "learning_rate": 0.00019999984820421088,
-      "loss": 46.0,
-      "step": 7264
-    },
-    {
-      "epoch": 0.5554599843263184,
-      "grad_norm": 0.0009931365493685007,
-      "learning_rate": 0.00019999984816235637,
-      "loss": 46.0,
-      "step": 7265
-    },
-    {
-      "epoch": 0.5555364413097081,
-      "grad_norm": 0.0003713297483045608,
-      "learning_rate": 0.00019999984812049614,
-      "loss": 46.0,
-      "step": 7266
-    },
-    {
-      "epoch": 0.5556128982930979,
-      "grad_norm": 0.000845150847453624,
-      "learning_rate": 0.0001999998480786301,
-      "loss": 46.0,
-      "step": 7267
-    },
-    {
-      "epoch": 0.5556893552764876,
-      "grad_norm": 0.005847814492881298,
-      "learning_rate": 0.0001999998480367583,
-      "loss": 46.0,
-      "step": 7268
-    },
-    {
-      "epoch": 0.5557658122598773,
-      "grad_norm": 0.0008431893074885011,
-      "learning_rate": 0.00019999984799488074,
-      "loss": 46.0,
-      "step": 7269
-    },
-    {
-      "epoch": 0.555842269243267,
-      "grad_norm": 0.000576764636207372,
-      "learning_rate": 0.00019999984795299742,
-      "loss": 46.0,
-      "step": 7270
-    },
-    {
-      "epoch": 0.5559187262266567,
-      "grad_norm": 0.0009986236691474915,
-      "learning_rate": 0.00019999984791110832,
-      "loss": 46.0,
-      "step": 7271
-    },
-    {
-      "epoch": 0.5559951832100465,
-      "grad_norm": 0.0007525266264565289,
-      "learning_rate": 0.00019999984786921344,
-      "loss": 46.0,
-      "step": 7272
-    },
-    {
-      "epoch": 0.5560716401934361,
-      "grad_norm": 0.0005763444933108985,
-      "learning_rate": 0.0001999998478273128,
-      "loss": 46.0,
-      "step": 7273
-    },
-    {
-      "epoch": 0.5561480971768259,
-      "grad_norm": 0.004906917922198772,
-      "learning_rate": 0.00019999984778540638,
-      "loss": 46.0,
-      "step": 7274
-    },
-    {
-      "epoch": 0.5562245541602157,
-      "grad_norm": 0.0024788579903542995,
-      "learning_rate": 0.0001999998477434942,
-      "loss": 46.0,
-      "step": 7275
-    },
-    {
-      "epoch": 0.5563010111436053,
-      "grad_norm": 0.0006713225739076734,
-      "learning_rate": 0.00019999984770157627,
-      "loss": 46.0,
-      "step": 7276
-    },
-    {
-      "epoch": 0.5563774681269951,
-      "grad_norm": 0.002713465830311179,
-      "learning_rate": 0.00019999984765965256,
-      "loss": 46.0,
-      "step": 7277
-    },
-    {
-      "epoch": 0.5564539251103847,
-      "grad_norm": 0.0006049118819646537,
-      "learning_rate": 0.00019999984761772305,
-      "loss": 46.0,
-      "step": 7278
-    },
-    {
-      "epoch": 0.5565303820937745,
-      "grad_norm": 0.0009185001254081726,
-      "learning_rate": 0.00019999984757578781,
-      "loss": 46.0,
-      "step": 7279
-    },
-    {
-      "epoch": 0.5566068390771642,
-      "grad_norm": 0.0005901929689571261,
-      "learning_rate": 0.0001999998475338468,
-      "loss": 46.0,
-      "step": 7280
-    },
-    {
-      "epoch": 0.5566832960605539,
-      "grad_norm": 0.001059014699421823,
-      "learning_rate": 0.0001999998474919,
-      "loss": 46.0,
-      "step": 7281
-    },
-    {
-      "epoch": 0.5567597530439437,
-      "grad_norm": 0.01876816712319851,
-      "learning_rate": 0.00019999984744994742,
-      "loss": 46.0,
-      "step": 7282
-    },
-    {
-      "epoch": 0.5568362100273334,
-      "grad_norm": 0.0008193185203708708,
-      "learning_rate": 0.00019999984740798907,
-      "loss": 46.0,
-      "step": 7283
-    },
-    {
-      "epoch": 0.5569126670107231,
-      "grad_norm": 0.0029996519442647696,
-      "learning_rate": 0.000199999847366025,
-      "loss": 46.0,
-      "step": 7284
-    },
-    {
-      "epoch": 0.5569891239941128,
-      "grad_norm": 0.0009654250461608171,
-      "learning_rate": 0.0001999998473240551,
-      "loss": 46.0,
-      "step": 7285
-    },
-    {
-      "epoch": 0.5570655809775026,
-      "grad_norm": 0.0018600947223603725,
-      "learning_rate": 0.00019999984728207948,
-      "loss": 46.0,
-      "step": 7286
-    },
-    {
-      "epoch": 0.5571420379608922,
-      "grad_norm": 0.011588250286877155,
-      "learning_rate": 0.00019999984724009806,
-      "loss": 46.0,
-      "step": 7287
-    },
-    {
-      "epoch": 0.557218494944282,
-      "grad_norm": 0.0015380606055259705,
-      "learning_rate": 0.00019999984719811087,
-      "loss": 46.0,
-      "step": 7288
-    },
-    {
-      "epoch": 0.5572949519276716,
-      "grad_norm": 0.00040448809158988297,
-      "learning_rate": 0.00019999984715611793,
-      "loss": 46.0,
-      "step": 7289
-    },
-    {
-      "epoch": 0.5573714089110614,
-      "grad_norm": 0.002322960877791047,
-      "learning_rate": 0.00019999984711411922,
-      "loss": 46.0,
-      "step": 7290
-    },
-    {
-      "epoch": 0.5574478658944512,
-      "grad_norm": 0.0014995394740253687,
-      "learning_rate": 0.00019999984707211473,
-      "loss": 46.0,
-      "step": 7291
-    },
-    {
-      "epoch": 0.5575243228778408,
-      "grad_norm": 0.0012440496357157826,
-      "learning_rate": 0.00019999984703010447,
-      "loss": 46.0,
-      "step": 7292
-    },
-    {
-      "epoch": 0.5576007798612306,
-      "grad_norm": 0.002473707776516676,
-      "learning_rate": 0.00019999984698808844,
-      "loss": 46.0,
-      "step": 7293
-    },
-    {
-      "epoch": 0.5576772368446203,
-      "grad_norm": 0.0009337534429505467,
-      "learning_rate": 0.0001999998469460666,
-      "loss": 46.0,
-      "step": 7294
-    },
-    {
-      "epoch": 0.55775369382801,
-      "grad_norm": 0.005159919150173664,
-      "learning_rate": 0.00019999984690403905,
-      "loss": 46.0,
-      "step": 7295
-    },
-    {
-      "epoch": 0.5578301508113997,
-      "grad_norm": 0.004803015384823084,
-      "learning_rate": 0.00019999984686200573,
-      "loss": 46.0,
-      "step": 7296
-    },
-    {
-      "epoch": 0.5579066077947895,
-      "grad_norm": 0.003702346235513687,
-      "learning_rate": 0.00019999984681996663,
-      "loss": 46.0,
-      "step": 7297
-    },
-    {
-      "epoch": 0.5579830647781792,
-      "grad_norm": 0.000839349115267396,
-      "learning_rate": 0.00019999984677792173,
-      "loss": 46.0,
-      "step": 7298
-    },
-    {
-      "epoch": 0.5580595217615689,
-      "grad_norm": 0.0026783442590385675,
-      "learning_rate": 0.0001999998467358711,
-      "loss": 46.0,
-      "step": 7299
-    },
-    {
-      "epoch": 0.5581359787449586,
-      "grad_norm": 0.0009273099130950868,
-      "learning_rate": 0.0001999998466938147,
-      "loss": 46.0,
-      "step": 7300
-    },
-    {
-      "epoch": 0.5582124357283483,
-      "grad_norm": 0.0013326219050213695,
-      "learning_rate": 0.0001999998466517525,
-      "loss": 46.0,
-      "step": 7301
-    },
-    {
-      "epoch": 0.5582888927117381,
-      "grad_norm": 0.0004435400478541851,
-      "learning_rate": 0.00019999984660968453,
-      "loss": 46.0,
-      "step": 7302
-    },
-    {
-      "epoch": 0.5583653496951277,
-      "grad_norm": 0.0006617952603846788,
-      "learning_rate": 0.00019999984656761082,
-      "loss": 46.0,
-      "step": 7303
-    },
-    {
-      "epoch": 0.5584418066785175,
-      "grad_norm": 0.0008876945939846337,
-      "learning_rate": 0.0001999998465255313,
-      "loss": 46.0,
-      "step": 7304
-    },
-    {
-      "epoch": 0.5585182636619073,
-      "grad_norm": 0.0027649810072034597,
-      "learning_rate": 0.00019999984648344605,
-      "loss": 46.0,
-      "step": 7305
-    },
-    {
-      "epoch": 0.5585947206452969,
-      "grad_norm": 0.0010028013493865728,
-      "learning_rate": 0.00019999984644135504,
-      "loss": 46.0,
-      "step": 7306
-    },
-    {
-      "epoch": 0.5586711776286867,
-      "grad_norm": 0.0015059361467137933,
-      "learning_rate": 0.00019999984639925824,
-      "loss": 46.0,
-      "step": 7307
-    },
-    {
-      "epoch": 0.5587476346120763,
-      "grad_norm": 0.002549107652157545,
-      "learning_rate": 0.00019999984635715566,
-      "loss": 46.0,
-      "step": 7308
-    },
-    {
-      "epoch": 0.5588240915954661,
-      "grad_norm": 0.0016522463411092758,
-      "learning_rate": 0.00019999984631504733,
-      "loss": 46.0,
-      "step": 7309
-    },
-    {
-      "epoch": 0.5589005485788558,
-      "grad_norm": 0.0010461328784003854,
-      "learning_rate": 0.0001999998462729332,
-      "loss": 46.0,
-      "step": 7310
-    },
-    {
-      "epoch": 0.5589770055622455,
-      "grad_norm": 0.00086758635006845,
-      "learning_rate": 0.00019999984623081334,
-      "loss": 46.0,
-      "step": 7311
-    },
-    {
-      "epoch": 0.5590534625456353,
-      "grad_norm": 0.0010320841101929545,
-      "learning_rate": 0.0001999998461886877,
-      "loss": 46.0,
-      "step": 7312
-    },
-    {
-      "epoch": 0.559129919529025,
-      "grad_norm": 0.0038554228376597166,
-      "learning_rate": 0.00019999984614655627,
-      "loss": 46.0,
-      "step": 7313
-    },
-    {
-      "epoch": 0.5592063765124147,
-      "grad_norm": 0.00397162139415741,
-      "learning_rate": 0.00019999984610441908,
-      "loss": 46.0,
-      "step": 7314
-    },
-    {
-      "epoch": 0.5592828334958044,
-      "grad_norm": 0.0019855108112096786,
-      "learning_rate": 0.00019999984606227614,
-      "loss": 46.0,
-      "step": 7315
-    },
-    {
-      "epoch": 0.5593592904791942,
-      "grad_norm": 0.0006833222578279674,
-      "learning_rate": 0.0001999998460201274,
-      "loss": 46.0,
-      "step": 7316
-    },
-    {
-      "epoch": 0.5594357474625838,
-      "grad_norm": 0.0004653666983358562,
-      "learning_rate": 0.00019999984597797292,
-      "loss": 46.0,
-      "step": 7317
-    },
-    {
-      "epoch": 0.5595122044459736,
-      "grad_norm": 0.0012717371573671699,
-      "learning_rate": 0.00019999984593581266,
-      "loss": 46.0,
-      "step": 7318
-    },
-    {
-      "epoch": 0.5595886614293633,
-      "grad_norm": 0.0011538080871105194,
-      "learning_rate": 0.0001999998458936466,
-      "loss": 46.0,
-      "step": 7319
-    },
-    {
-      "epoch": 0.559665118412753,
-      "grad_norm": 0.0020591483917087317,
-      "learning_rate": 0.00019999984585147482,
-      "loss": 46.0,
-      "step": 7320
-    },
-    {
-      "epoch": 0.5597415753961428,
-      "grad_norm": 0.004964011255651712,
-      "learning_rate": 0.00019999984580929721,
-      "loss": 46.0,
-      "step": 7321
-    },
-    {
-      "epoch": 0.5598180323795324,
-      "grad_norm": 0.006484550423920155,
-      "learning_rate": 0.0001999998457671139,
-      "loss": 46.0,
-      "step": 7322
-    },
-    {
-      "epoch": 0.5598944893629222,
-      "grad_norm": 0.00125783437397331,
-      "learning_rate": 0.0001999998457249248,
-      "loss": 46.0,
-      "step": 7323
-    },
-    {
-      "epoch": 0.559970946346312,
-      "grad_norm": 0.0018940666923299432,
-      "learning_rate": 0.0001999998456827299,
-      "loss": 46.0,
-      "step": 7324
-    },
-    {
-      "epoch": 0.5600474033297016,
-      "grad_norm": 0.0011719519970938563,
-      "learning_rate": 0.00019999984564052925,
-      "loss": 46.0,
-      "step": 7325
-    },
-    {
-      "epoch": 0.5601238603130914,
-      "grad_norm": 0.00046957639278844,
-      "learning_rate": 0.00019999984559832283,
-      "loss": 46.0,
-      "step": 7326
-    },
-    {
-      "epoch": 0.5602003172964811,
-      "grad_norm": 0.0015421287389472127,
-      "learning_rate": 0.00019999984555611064,
-      "loss": 46.0,
-      "step": 7327
-    },
-    {
-      "epoch": 0.5602767742798708,
-      "grad_norm": 0.0010119170183315873,
-      "learning_rate": 0.00019999984551389267,
-      "loss": 46.0,
-      "step": 7328
-    },
-    {
-      "epoch": 0.5603532312632605,
-      "grad_norm": 0.0009469077340327203,
-      "learning_rate": 0.00019999984547166893,
-      "loss": 46.0,
-      "step": 7329
-    },
-    {
-      "epoch": 0.5604296882466502,
-      "grad_norm": 0.00043641417869366705,
-      "learning_rate": 0.00019999984542943945,
-      "loss": 46.0,
-      "step": 7330
-    },
-    {
-      "epoch": 0.5605061452300399,
-      "grad_norm": 0.0014040989335626364,
-      "learning_rate": 0.0001999998453872042,
-      "loss": 46.0,
-      "step": 7331
-    },
-    {
-      "epoch": 0.5605826022134297,
-      "grad_norm": 0.00037488597445189953,
-      "learning_rate": 0.00019999984534496313,
-      "loss": 46.0,
-      "step": 7332
-    },
-    {
-      "epoch": 0.5606590591968194,
-      "grad_norm": 0.0006711481255479157,
-      "learning_rate": 0.00019999984530271633,
-      "loss": 46.0,
-      "step": 7333
-    },
-    {
-      "epoch": 0.5607355161802091,
-      "grad_norm": 0.006668330170214176,
-      "learning_rate": 0.00019999984526046375,
-      "loss": 46.0,
-      "step": 7334
-    },
-    {
-      "epoch": 0.5608119731635989,
-      "grad_norm": 0.006237075664103031,
-      "learning_rate": 0.00019999984521820542,
-      "loss": 46.0,
-      "step": 7335
-    },
-    {
-      "epoch": 0.5608884301469885,
-      "grad_norm": 0.0016187290893867612,
-      "learning_rate": 0.0001999998451759413,
-      "loss": 46.0,
-      "step": 7336
-    },
-    {
-      "epoch": 0.5609648871303783,
-      "grad_norm": 0.0005768187111243606,
-      "learning_rate": 0.0001999998451336714,
-      "loss": 46.0,
-      "step": 7337
-    },
-    {
-      "epoch": 0.5610413441137679,
-      "grad_norm": 0.0028132263105362654,
-      "learning_rate": 0.00019999984509139575,
-      "loss": 46.0,
-      "step": 7338
-    },
-    {
-      "epoch": 0.5611178010971577,
-      "grad_norm": 0.0008070542244240642,
-      "learning_rate": 0.00019999984504911434,
-      "loss": 46.0,
-      "step": 7339
-    },
-    {
-      "epoch": 0.5611942580805475,
-      "grad_norm": 0.001138027524575591,
-      "learning_rate": 0.00019999984500682715,
-      "loss": 46.0,
-      "step": 7340
-    },
-    {
-      "epoch": 0.5612707150639371,
-      "grad_norm": 0.0014547263272106647,
-      "learning_rate": 0.00019999984496453418,
-      "loss": 46.0,
-      "step": 7341
-    },
-    {
-      "epoch": 0.5613471720473269,
-      "grad_norm": 0.001572155742906034,
-      "learning_rate": 0.00019999984492223547,
-      "loss": 46.0,
-      "step": 7342
-    },
-    {
-      "epoch": 0.5614236290307166,
-      "grad_norm": 0.003951024264097214,
-      "learning_rate": 0.00019999984487993096,
-      "loss": 46.0,
-      "step": 7343
-    },
-    {
-      "epoch": 0.5615000860141063,
-      "grad_norm": 0.000999454059638083,
-      "learning_rate": 0.0001999998448376207,
-      "loss": 46.0,
-      "step": 7344
-    },
-    {
-      "epoch": 0.561576542997496,
-      "grad_norm": 0.0003573857247829437,
-      "learning_rate": 0.00019999984479530464,
-      "loss": 46.0,
-      "step": 7345
-    },
-    {
-      "epoch": 0.5616529999808858,
-      "grad_norm": 0.00541074899956584,
-      "learning_rate": 0.00019999984475298284,
-      "loss": 46.0,
-      "step": 7346
-    },
-    {
-      "epoch": 0.5617294569642755,
-      "grad_norm": 0.000958868651650846,
-      "learning_rate": 0.00019999984471065526,
-      "loss": 46.0,
-      "step": 7347
-    },
-    {
-      "epoch": 0.5618059139476652,
-      "grad_norm": 0.0005875806673429906,
-      "learning_rate": 0.0001999998446683219,
-      "loss": 46.0,
-      "step": 7348
-    },
-    {
-      "epoch": 0.5618823709310549,
-      "grad_norm": 0.0007857672171667218,
-      "learning_rate": 0.0001999998446259828,
-      "loss": 46.0,
-      "step": 7349
-    },
-    {
-      "epoch": 0.5619588279144446,
-      "grad_norm": 0.0007955322507768869,
-      "learning_rate": 0.0001999998445836379,
-      "loss": 46.0,
-      "step": 7350
-    },
-    {
-      "epoch": 0.5620352848978344,
-      "grad_norm": 0.0018155993893742561,
-      "learning_rate": 0.00019999984454128727,
-      "loss": 46.0,
-      "step": 7351
-    },
-    {
-      "epoch": 0.562111741881224,
-      "grad_norm": 0.0006256432388909161,
-      "learning_rate": 0.00019999984449893082,
-      "loss": 46.0,
-      "step": 7352
-    },
-    {
-      "epoch": 0.5621881988646138,
-      "grad_norm": 0.001237974502146244,
-      "learning_rate": 0.00019999984445656863,
-      "loss": 46.0,
-      "step": 7353
-    },
-    {
-      "epoch": 0.5622646558480036,
-      "grad_norm": 0.002257004613056779,
-      "learning_rate": 0.00019999984441420067,
-      "loss": 46.0,
-      "step": 7354
-    },
-    {
-      "epoch": 0.5623411128313932,
-      "grad_norm": 0.0014228172367438674,
-      "learning_rate": 0.00019999984437182693,
-      "loss": 46.0,
-      "step": 7355
-    },
-    {
-      "epoch": 0.562417569814783,
-      "grad_norm": 0.001369641860947013,
-      "learning_rate": 0.00019999984432944742,
-      "loss": 46.0,
-      "step": 7356
-    },
-    {
-      "epoch": 0.5624940267981727,
-      "grad_norm": 0.001717785489745438,
-      "learning_rate": 0.00019999984428706216,
-      "loss": 46.0,
-      "step": 7357
-    },
-    {
-      "epoch": 0.5625704837815624,
-      "grad_norm": 0.0029476010240614414,
-      "learning_rate": 0.0001999998442446711,
-      "loss": 46.0,
-      "step": 7358
-    },
-    {
-      "epoch": 0.5626469407649521,
-      "grad_norm": 0.0009090182138606906,
-      "learning_rate": 0.00019999984420227432,
-      "loss": 46.0,
-      "step": 7359
-    },
-    {
-      "epoch": 0.5627233977483418,
-      "grad_norm": 0.0008652320830151439,
-      "learning_rate": 0.00019999984415987172,
-      "loss": 46.0,
-      "step": 7360
-    },
-    {
-      "epoch": 0.5627998547317316,
-      "grad_norm": 0.0019179227529093623,
-      "learning_rate": 0.00019999984411746334,
-      "loss": 46.0,
-      "step": 7361
-    },
-    {
-      "epoch": 0.5628763117151213,
-      "grad_norm": 0.0011555092642083764,
-      "learning_rate": 0.00019999984407504924,
-      "loss": 46.0,
-      "step": 7362
-    },
-    {
-      "epoch": 0.562952768698511,
-      "grad_norm": 0.003565003629773855,
-      "learning_rate": 0.00019999984403262934,
-      "loss": 46.0,
-      "step": 7363
-    },
-    {
-      "epoch": 0.5630292256819007,
-      "grad_norm": 0.000671940972097218,
-      "learning_rate": 0.00019999984399020367,
-      "loss": 46.0,
-      "step": 7364
-    },
-    {
-      "epoch": 0.5631056826652905,
-      "grad_norm": 0.0031001626048237085,
-      "learning_rate": 0.00019999984394777226,
-      "loss": 46.0,
-      "step": 7365
-    },
-    {
-      "epoch": 0.5631821396486801,
-      "grad_norm": 0.0006579491309821606,
-      "learning_rate": 0.0001999998439053351,
-      "loss": 46.0,
-      "step": 7366
-    },
-    {
-      "epoch": 0.5632585966320699,
-      "grad_norm": 0.0016551838489249349,
-      "learning_rate": 0.0001999998438628921,
-      "loss": 46.0,
-      "step": 7367
-    },
-    {
-      "epoch": 0.5633350536154595,
-      "grad_norm": 0.0013096745824441314,
-      "learning_rate": 0.00019999984382044336,
-      "loss": 46.0,
-      "step": 7368
-    },
-    {
-      "epoch": 0.5634115105988493,
-      "grad_norm": 0.006040710024535656,
-      "learning_rate": 0.00019999984377798885,
-      "loss": 46.0,
-      "step": 7369
-    },
-    {
-      "epoch": 0.5634879675822391,
-      "grad_norm": 0.0038013113662600517,
-      "learning_rate": 0.0001999998437355286,
-      "loss": 46.0,
-      "step": 7370
-    },
-    {
-      "epoch": 0.5635644245656287,
-      "grad_norm": 0.0007698804838582873,
-      "learning_rate": 0.00019999984369306254,
-      "loss": 46.0,
-      "step": 7371
-    },
-    {
-      "epoch": 0.5636408815490185,
-      "grad_norm": 0.0009979300666600466,
-      "learning_rate": 0.00019999984365059073,
-      "loss": 46.0,
-      "step": 7372
-    },
-    {
-      "epoch": 0.5637173385324082,
-      "grad_norm": 0.0017158391419798136,
-      "learning_rate": 0.00019999984360811316,
-      "loss": 46.0,
-      "step": 7373
-    },
-    {
-      "epoch": 0.5637937955157979,
-      "grad_norm": 0.0009191979188472033,
-      "learning_rate": 0.0001999998435656298,
-      "loss": 46.0,
-      "step": 7374
-    },
-    {
-      "epoch": 0.5638702524991877,
-      "grad_norm": 0.014036411419510841,
-      "learning_rate": 0.00019999984352314068,
-      "loss": 46.0,
-      "step": 7375
-    },
-    {
-      "epoch": 0.5639467094825774,
-      "grad_norm": 0.0007032278226688504,
-      "learning_rate": 0.00019999984348064578,
-      "loss": 46.0,
-      "step": 7376
-    },
-    {
-      "epoch": 0.5640231664659671,
-      "grad_norm": 0.003338852897286415,
-      "learning_rate": 0.0001999998434381451,
-      "loss": 46.0,
-      "step": 7377
-    },
-    {
-      "epoch": 0.5640996234493568,
-      "grad_norm": 0.0019688718020915985,
-      "learning_rate": 0.00019999984339563867,
-      "loss": 46.0,
-      "step": 7378
-    },
-    {
-      "epoch": 0.5641760804327465,
-      "grad_norm": 0.001912647276185453,
-      "learning_rate": 0.0001999998433531265,
-      "loss": 46.0,
-      "step": 7379
-    },
-    {
-      "epoch": 0.5642525374161362,
-      "grad_norm": 0.001072608050890267,
-      "learning_rate": 0.0001999998433106085,
-      "loss": 46.0,
-      "step": 7380
-    },
-    {
-      "epoch": 0.564328994399526,
-      "grad_norm": 0.0010734391398727894,
-      "learning_rate": 0.00019999984326808478,
-      "loss": 46.0,
-      "step": 7381
-    },
-    {
-      "epoch": 0.5644054513829156,
-      "grad_norm": 0.0008450491586700082,
-      "learning_rate": 0.00019999984322555524,
-      "loss": 46.0,
-      "step": 7382
-    },
-    {
-      "epoch": 0.5644819083663054,
-      "grad_norm": 0.0007184364949353039,
-      "learning_rate": 0.00019999984318301998,
-      "loss": 46.0,
-      "step": 7383
-    },
-    {
-      "epoch": 0.5645583653496952,
-      "grad_norm": 0.0035899863578379154,
-      "learning_rate": 0.00019999984314047895,
-      "loss": 46.0,
-      "step": 7384
-    },
-    {
-      "epoch": 0.5646348223330848,
-      "grad_norm": 0.0004965142579749227,
-      "learning_rate": 0.00019999984309793212,
-      "loss": 46.0,
-      "step": 7385
-    },
-    {
-      "epoch": 0.5647112793164746,
-      "grad_norm": 0.0014849104918539524,
-      "learning_rate": 0.00019999984305537954,
-      "loss": 46.0,
-      "step": 7386
-    },
-    {
-      "epoch": 0.5647877362998643,
-      "grad_norm": 0.004753535613417625,
-      "learning_rate": 0.00019999984301282116,
-      "loss": 46.0,
-      "step": 7387
-    },
-    {
-      "epoch": 0.564864193283254,
-      "grad_norm": 0.0007656702073290944,
-      "learning_rate": 0.00019999984297025704,
-      "loss": 46.0,
-      "step": 7388
-    },
-    {
-      "epoch": 0.5649406502666438,
-      "grad_norm": 0.0014321764465421438,
-      "learning_rate": 0.00019999984292768717,
-      "loss": 46.0,
-      "step": 7389
-    },
-    {
-      "epoch": 0.5650171072500334,
-      "grad_norm": 0.0027056585531681776,
-      "learning_rate": 0.0001999998428851115,
-      "loss": 46.0,
-      "step": 7390
-    },
-    {
-      "epoch": 0.5650935642334232,
-      "grad_norm": 0.0007521676598116755,
-      "learning_rate": 0.00019999984284253003,
-      "loss": 46.0,
-      "step": 7391
-    },
-    {
-      "epoch": 0.5651700212168129,
-      "grad_norm": 0.001418567611835897,
-      "learning_rate": 0.00019999984279994287,
-      "loss": 46.0,
-      "step": 7392
-    },
-    {
-      "epoch": 0.5652464782002026,
-      "grad_norm": 0.0028426814824342728,
-      "learning_rate": 0.00019999984275734988,
-      "loss": 46.0,
-      "step": 7393
-    },
-    {
-      "epoch": 0.5653229351835923,
-      "grad_norm": 0.0005968738114461303,
-      "learning_rate": 0.00019999984271475114,
-      "loss": 46.0,
-      "step": 7394
-    },
-    {
-      "epoch": 0.5653993921669821,
-      "grad_norm": 0.0006396625540219247,
-      "learning_rate": 0.00019999984267214663,
-      "loss": 46.0,
-      "step": 7395
-    },
-    {
-      "epoch": 0.5654758491503717,
-      "grad_norm": 0.0004740832664538175,
-      "learning_rate": 0.00019999984262953632,
-      "loss": 46.0,
-      "step": 7396
-    },
-    {
-      "epoch": 0.5655523061337615,
-      "grad_norm": 0.0033132312819361687,
-      "learning_rate": 0.0001999998425869203,
-      "loss": 46.0,
-      "step": 7397
-    },
-    {
-      "epoch": 0.5656287631171513,
-      "grad_norm": 0.0005001610843464732,
-      "learning_rate": 0.00019999984254429846,
-      "loss": 46.0,
-      "step": 7398
-    },
-    {
-      "epoch": 0.5657052201005409,
-      "grad_norm": 0.00477063050493598,
-      "learning_rate": 0.00019999984250167088,
-      "loss": 46.0,
-      "step": 7399
-    },
-    {
-      "epoch": 0.5657816770839307,
-      "grad_norm": 0.001287501654587686,
-      "learning_rate": 0.0001999998424590375,
-      "loss": 46.0,
-      "step": 7400
-    },
-    {
-      "epoch": 0.5658581340673203,
-      "grad_norm": 0.00038623795262537897,
-      "learning_rate": 0.00019999984241639838,
-      "loss": 46.0,
-      "step": 7401
-    },
-    {
-      "epoch": 0.5659345910507101,
-      "grad_norm": 0.0008132167859002948,
-      "learning_rate": 0.0001999998423737535,
-      "loss": 46.0,
-      "step": 7402
-    },
-    {
-      "epoch": 0.5660110480340999,
-      "grad_norm": 0.005862409248948097,
-      "learning_rate": 0.00019999984233110284,
-      "loss": 46.0,
-      "step": 7403
-    },
-    {
-      "epoch": 0.5660875050174895,
-      "grad_norm": 0.0013818725710734725,
-      "learning_rate": 0.0001999998422884464,
-      "loss": 46.0,
-      "step": 7404
-    },
-    {
-      "epoch": 0.5661639620008793,
-      "grad_norm": 0.0013392572291195393,
-      "learning_rate": 0.00019999984224578418,
-      "loss": 46.0,
-      "step": 7405
-    },
-    {
-      "epoch": 0.566240418984269,
-      "grad_norm": 0.002496361965313554,
-      "learning_rate": 0.00019999984220311621,
-      "loss": 46.0,
-      "step": 7406
-    },
-    {
-      "epoch": 0.5663168759676587,
-      "grad_norm": 0.0072517115622758865,
-      "learning_rate": 0.00019999984216044245,
-      "loss": 46.0,
-      "step": 7407
-    },
-    {
-      "epoch": 0.5663933329510484,
-      "grad_norm": 0.0014614968094974756,
-      "learning_rate": 0.00019999984211776294,
-      "loss": 46.0,
-      "step": 7408
-    },
-    {
-      "epoch": 0.5664697899344381,
-      "grad_norm": 0.0031737065874040127,
-      "learning_rate": 0.00019999984207507766,
-      "loss": 46.0,
-      "step": 7409
-    },
-    {
-      "epoch": 0.5665462469178278,
-      "grad_norm": 0.0021332716569304466,
-      "learning_rate": 0.00019999984203238663,
-      "loss": 46.0,
-      "step": 7410
-    },
-    {
-      "epoch": 0.5666227039012176,
-      "grad_norm": 0.0007254000520333648,
-      "learning_rate": 0.0001999998419896898,
-      "loss": 46.0,
-      "step": 7411
-    },
-    {
-      "epoch": 0.5666991608846073,
-      "grad_norm": 0.0005509118782356381,
-      "learning_rate": 0.0001999998419469872,
-      "loss": 46.0,
-      "step": 7412
-    },
-    {
-      "epoch": 0.566775617867997,
-      "grad_norm": 0.000635275209788233,
-      "learning_rate": 0.00019999984190427882,
-      "loss": 46.0,
-      "step": 7413
-    },
-    {
-      "epoch": 0.5668520748513868,
-      "grad_norm": 0.0012590569676831365,
-      "learning_rate": 0.0001999998418615647,
-      "loss": 46.0,
-      "step": 7414
-    },
-    {
-      "epoch": 0.5669285318347764,
-      "grad_norm": 0.0009258652571588755,
-      "learning_rate": 0.00019999984181884482,
-      "loss": 46.0,
-      "step": 7415
-    },
-    {
-      "epoch": 0.5670049888181662,
-      "grad_norm": 0.001466510584577918,
-      "learning_rate": 0.00019999984177611913,
-      "loss": 46.0,
-      "step": 7416
-    },
-    {
-      "epoch": 0.567081445801556,
-      "grad_norm": 0.0009787505259737372,
-      "learning_rate": 0.0001999998417333877,
-      "loss": 46.0,
-      "step": 7417
-    },
-    {
-      "epoch": 0.5671579027849456,
-      "grad_norm": 0.0009220175561495125,
-      "learning_rate": 0.00019999984169065047,
-      "loss": 46.0,
-      "step": 7418
-    },
-    {
-      "epoch": 0.5672343597683354,
-      "grad_norm": 0.0020545015577226877,
-      "learning_rate": 0.0001999998416479075,
-      "loss": 46.0,
-      "step": 7419
-    },
-    {
-      "epoch": 0.567310816751725,
-      "grad_norm": 0.00028155589825473726,
-      "learning_rate": 0.00019999984160515877,
-      "loss": 46.0,
-      "step": 7420
-    },
-    {
-      "epoch": 0.5673872737351148,
-      "grad_norm": 0.005777201149612665,
-      "learning_rate": 0.00019999984156240423,
-      "loss": 46.0,
-      "step": 7421
-    },
-    {
-      "epoch": 0.5674637307185045,
-      "grad_norm": 0.002540904562920332,
-      "learning_rate": 0.00019999984151964395,
-      "loss": 46.0,
-      "step": 7422
-    },
-    {
-      "epoch": 0.5675401877018942,
-      "grad_norm": 0.000998013187199831,
-      "learning_rate": 0.0001999998414768779,
-      "loss": 46.0,
-      "step": 7423
-    },
-    {
-      "epoch": 0.5676166446852839,
-      "grad_norm": 0.001464120578020811,
-      "learning_rate": 0.0001999998414341061,
-      "loss": 46.0,
-      "step": 7424
-    },
-    {
-      "epoch": 0.5676931016686737,
-      "grad_norm": 0.007333226036280394,
-      "learning_rate": 0.00019999984139132848,
-      "loss": 46.0,
-      "step": 7425
-    },
-    {
-      "epoch": 0.5677695586520634,
-      "grad_norm": 0.0004693620721809566,
-      "learning_rate": 0.0001999998413485451,
-      "loss": 46.0,
-      "step": 7426
-    },
-    {
-      "epoch": 0.5678460156354531,
-      "grad_norm": 0.003078486304730177,
-      "learning_rate": 0.00019999984130575598,
-      "loss": 46.0,
-      "step": 7427
-    },
-    {
-      "epoch": 0.5679224726188429,
-      "grad_norm": 0.0005773595767095685,
-      "learning_rate": 0.00019999984126296106,
-      "loss": 46.0,
-      "step": 7428
-    },
-    {
-      "epoch": 0.5679989296022325,
-      "grad_norm": 0.004314554389566183,
-      "learning_rate": 0.0001999998412201604,
-      "loss": 46.0,
-      "step": 7429
-    },
-    {
-      "epoch": 0.5680753865856223,
-      "grad_norm": 0.0009478558786213398,
-      "learning_rate": 0.00019999984117735395,
-      "loss": 46.0,
-      "step": 7430
-    },
-    {
-      "epoch": 0.5681518435690119,
-      "grad_norm": 0.008602998219430447,
-      "learning_rate": 0.00019999984113454176,
-      "loss": 46.0,
-      "step": 7431
-    },
-    {
-      "epoch": 0.5682283005524017,
-      "grad_norm": 0.0003639942442532629,
-      "learning_rate": 0.00019999984109172377,
-      "loss": 46.0,
-      "step": 7432
-    },
-    {
-      "epoch": 0.5683047575357915,
-      "grad_norm": 0.004051700700074434,
-      "learning_rate": 0.0001999998410489,
-      "loss": 46.0,
-      "step": 7433
-    },
-    {
-      "epoch": 0.5683812145191811,
-      "grad_norm": 0.0008075557416304946,
-      "learning_rate": 0.00019999984100607047,
-      "loss": 46.0,
-      "step": 7434
-    },
-    {
-      "epoch": 0.5684576715025709,
-      "grad_norm": 0.0006399169797077775,
-      "learning_rate": 0.0001999998409632352,
-      "loss": 46.0,
-      "step": 7435
-    },
-    {
-      "epoch": 0.5685341284859606,
-      "grad_norm": 0.0016128354473039508,
-      "learning_rate": 0.00019999984092039413,
-      "loss": 46.0,
-      "step": 7436
-    },
-    {
-      "epoch": 0.5686105854693503,
-      "grad_norm": 0.001321173389442265,
-      "learning_rate": 0.0001999998408775473,
-      "loss": 46.0,
-      "step": 7437
-    },
-    {
-      "epoch": 0.56868704245274,
-      "grad_norm": 0.0007738806889392436,
-      "learning_rate": 0.00019999984083469472,
-      "loss": 46.0,
-      "step": 7438
-    },
-    {
-      "epoch": 0.5687634994361297,
-      "grad_norm": 0.0007164178532548249,
-      "learning_rate": 0.00019999984079183632,
-      "loss": 46.0,
-      "step": 7439
-    },
-    {
-      "epoch": 0.5688399564195195,
-      "grad_norm": 0.0008193447138182819,
-      "learning_rate": 0.00019999984074897223,
-      "loss": 46.0,
-      "step": 7440
-    },
-    {
-      "epoch": 0.5689164134029092,
-      "grad_norm": 0.0010059851920232177,
-      "learning_rate": 0.0001999998407061023,
-      "loss": 46.0,
-      "step": 7441
-    },
-    {
-      "epoch": 0.5689928703862989,
-      "grad_norm": 0.004593095742166042,
-      "learning_rate": 0.00019999984066322663,
-      "loss": 46.0,
-      "step": 7442
-    },
-    {
-      "epoch": 0.5690693273696886,
-      "grad_norm": 0.0010517298942431808,
-      "learning_rate": 0.0001999998406203452,
-      "loss": 46.0,
-      "step": 7443
-    },
-    {
-      "epoch": 0.5691457843530784,
-      "grad_norm": 0.0018093701219186187,
-      "learning_rate": 0.00019999984057745795,
-      "loss": 46.0,
-      "step": 7444
-    },
-    {
-      "epoch": 0.569222241336468,
-      "grad_norm": 0.006365989334881306,
-      "learning_rate": 0.00019999984053456499,
-      "loss": 46.0,
-      "step": 7445
-    },
-    {
-      "epoch": 0.5692986983198578,
-      "grad_norm": 0.0016057032626122236,
-      "learning_rate": 0.00019999984049166622,
-      "loss": 46.0,
-      "step": 7446
-    },
-    {
-      "epoch": 0.5693751553032476,
-      "grad_norm": 0.0006403520237654448,
-      "learning_rate": 0.00019999984044876169,
-      "loss": 46.0,
-      "step": 7447
-    },
-    {
-      "epoch": 0.5694516122866372,
-      "grad_norm": 0.0008778295014053583,
-      "learning_rate": 0.0001999998404058514,
-      "loss": 46.0,
-      "step": 7448
-    },
-    {
-      "epoch": 0.569528069270027,
-      "grad_norm": 0.003783304011449218,
-      "learning_rate": 0.00019999984036293535,
-      "loss": 46.0,
-      "step": 7449
-    },
-    {
-      "epoch": 0.5696045262534166,
-      "grad_norm": 0.0012332770274952054,
-      "learning_rate": 0.00019999984032001355,
-      "loss": 46.0,
-      "step": 7450
-    },
-    {
-      "epoch": 0.5696809832368064,
-      "grad_norm": 0.0024706649128347635,
-      "learning_rate": 0.00019999984027708592,
-      "loss": 46.0,
-      "step": 7451
-    },
-    {
-      "epoch": 0.5697574402201961,
-      "grad_norm": 0.001178263803012669,
-      "learning_rate": 0.00019999984023415254,
-      "loss": 46.0,
-      "step": 7452
-    },
-    {
-      "epoch": 0.5698338972035858,
-      "grad_norm": 0.0047165993601083755,
-      "learning_rate": 0.00019999984019121342,
-      "loss": 46.0,
-      "step": 7453
-    },
-    {
-      "epoch": 0.5699103541869756,
-      "grad_norm": 0.0017295429715886712,
-      "learning_rate": 0.00019999984014826852,
-      "loss": 46.0,
-      "step": 7454
-    },
-    {
-      "epoch": 0.5699868111703653,
-      "grad_norm": 0.0021189383696764708,
-      "learning_rate": 0.00019999984010531783,
-      "loss": 46.0,
-      "step": 7455
-    },
-    {
-      "epoch": 0.570063268153755,
-      "grad_norm": 0.0014236309798434377,
-      "learning_rate": 0.00019999984006236136,
-      "loss": 46.0,
-      "step": 7456
-    },
-    {
-      "epoch": 0.5701397251371447,
-      "grad_norm": 0.0012797509552910924,
-      "learning_rate": 0.00019999984001939917,
-      "loss": 46.0,
-      "step": 7457
-    },
-    {
-      "epoch": 0.5702161821205345,
-      "grad_norm": 0.000946762622334063,
-      "learning_rate": 0.00019999983997643118,
-      "loss": 46.0,
-      "step": 7458
-    },
-    {
-      "epoch": 0.5702926391039241,
-      "grad_norm": 0.0022813996765762568,
-      "learning_rate": 0.00019999983993345742,
-      "loss": 46.0,
-      "step": 7459
-    },
-    {
-      "epoch": 0.5703690960873139,
-      "grad_norm": 0.0027368422597646713,
-      "learning_rate": 0.00019999983989047788,
-      "loss": 46.0,
-      "step": 7460
-    },
-    {
-      "epoch": 0.5704455530707035,
-      "grad_norm": 0.003609990468248725,
-      "learning_rate": 0.00019999983984749257,
-      "loss": 46.0,
-      "step": 7461
-    },
-    {
-      "epoch": 0.5705220100540933,
-      "grad_norm": 0.0014453730545938015,
-      "learning_rate": 0.00019999983980450154,
-      "loss": 46.0,
-      "step": 7462
-    },
-    {
-      "epoch": 0.5705984670374831,
-      "grad_norm": 0.0005166811752133071,
-      "learning_rate": 0.00019999983976150469,
-      "loss": 46.0,
-      "step": 7463
-    },
-    {
-      "epoch": 0.5706749240208727,
-      "grad_norm": 0.0007537009660154581,
-      "learning_rate": 0.00019999983971850208,
-      "loss": 46.0,
-      "step": 7464
-    },
-    {
-      "epoch": 0.5707513810042625,
-      "grad_norm": 0.0005977752152830362,
-      "learning_rate": 0.0001999998396754937,
-      "loss": 46.0,
-      "step": 7465
-    },
-    {
-      "epoch": 0.5708278379876522,
-      "grad_norm": 0.0032258043065667152,
-      "learning_rate": 0.00019999983963247956,
-      "loss": 46.0,
-      "step": 7466
-    },
-    {
-      "epoch": 0.5709042949710419,
-      "grad_norm": 0.0008098523248918355,
-      "learning_rate": 0.00019999983958945964,
-      "loss": 46.0,
-      "step": 7467
-    },
-    {
-      "epoch": 0.5709807519544317,
-      "grad_norm": 0.0006344974972307682,
-      "learning_rate": 0.00019999983954643397,
-      "loss": 46.0,
-      "step": 7468
-    },
-    {
-      "epoch": 0.5710572089378213,
-      "grad_norm": 0.0013121215160936117,
-      "learning_rate": 0.00019999983950340252,
-      "loss": 46.0,
-      "step": 7469
-    },
-    {
-      "epoch": 0.5711336659212111,
-      "grad_norm": 0.0006446399493142962,
-      "learning_rate": 0.0001999998394603653,
-      "loss": 46.0,
-      "step": 7470
-    },
-    {
-      "epoch": 0.5712101229046008,
-      "grad_norm": 0.0012016990222036839,
-      "learning_rate": 0.00019999983941732232,
-      "loss": 46.0,
-      "step": 7471
-    },
-    {
-      "epoch": 0.5712865798879905,
-      "grad_norm": 0.002046516863629222,
-      "learning_rate": 0.00019999983937427356,
-      "loss": 46.0,
-      "step": 7472
-    },
-    {
-      "epoch": 0.5713630368713802,
-      "grad_norm": 0.001340606715530157,
-      "learning_rate": 0.00019999983933121902,
-      "loss": 46.0,
-      "step": 7473
-    },
-    {
-      "epoch": 0.57143949385477,
-      "grad_norm": 0.0003149217809550464,
-      "learning_rate": 0.00019999983928815874,
-      "loss": 46.0,
-      "step": 7474
-    },
-    {
-      "epoch": 0.5715159508381596,
-      "grad_norm": 0.0018251878209412098,
-      "learning_rate": 0.00019999983924509266,
-      "loss": 46.0,
-      "step": 7475
-    },
-    {
-      "epoch": 0.5715924078215494,
-      "grad_norm": 0.001373132923617959,
-      "learning_rate": 0.00019999983920202083,
-      "loss": 46.0,
-      "step": 7476
-    },
-    {
-      "epoch": 0.5716688648049392,
-      "grad_norm": 0.0004142786201555282,
-      "learning_rate": 0.00019999983915894323,
-      "loss": 46.0,
-      "step": 7477
-    },
-    {
-      "epoch": 0.5717453217883288,
-      "grad_norm": 0.0012667272239923477,
-      "learning_rate": 0.00019999983911585985,
-      "loss": 46.0,
-      "step": 7478
-    },
-    {
-      "epoch": 0.5718217787717186,
-      "grad_norm": 0.014262200333178043,
-      "learning_rate": 0.00019999983907277068,
-      "loss": 46.0,
-      "step": 7479
-    },
-    {
-      "epoch": 0.5718982357551082,
-      "grad_norm": 0.0013388856314122677,
-      "learning_rate": 0.00019999983902967578,
-      "loss": 46.0,
-      "step": 7480
-    },
-    {
-      "epoch": 0.571974692738498,
-      "grad_norm": 0.0036675715819001198,
-      "learning_rate": 0.0001999998389865751,
-      "loss": 46.0,
-      "step": 7481
-    },
-    {
-      "epoch": 0.5720511497218878,
-      "grad_norm": 0.0008023617556318641,
-      "learning_rate": 0.00019999983894346864,
-      "loss": 46.0,
-      "step": 7482
-    },
-    {
-      "epoch": 0.5721276067052774,
-      "grad_norm": 0.0011382491793483496,
-      "learning_rate": 0.00019999983890035643,
-      "loss": 46.0,
-      "step": 7483
-    },
-    {
-      "epoch": 0.5722040636886672,
-      "grad_norm": 0.0008855733321979642,
-      "learning_rate": 0.00019999983885723844,
-      "loss": 46.0,
-      "step": 7484
-    },
-    {
-      "epoch": 0.5722805206720569,
-      "grad_norm": 0.008134298957884312,
-      "learning_rate": 0.00019999983881411465,
-      "loss": 46.0,
-      "step": 7485
-    },
-    {
-      "epoch": 0.5723569776554466,
-      "grad_norm": 0.007042621728032827,
-      "learning_rate": 0.00019999983877098514,
-      "loss": 46.0,
-      "step": 7486
-    },
-    {
-      "epoch": 0.5724334346388363,
-      "grad_norm": 0.0024729925207793713,
-      "learning_rate": 0.00019999983872784983,
-      "loss": 46.0,
-      "step": 7487
-    },
-    {
-      "epoch": 0.5725098916222261,
-      "grad_norm": 0.0006616662722080946,
-      "learning_rate": 0.00019999983868470878,
-      "loss": 46.0,
-      "step": 7488
-    },
-    {
-      "epoch": 0.5725863486056157,
-      "grad_norm": 0.00438628951087594,
-      "learning_rate": 0.00019999983864156192,
-      "loss": 46.0,
-      "step": 7489
-    },
-    {
-      "epoch": 0.5726628055890055,
-      "grad_norm": 0.0014128299662843347,
-      "learning_rate": 0.00019999983859840932,
-      "loss": 46.0,
-      "step": 7490
-    },
-    {
-      "epoch": 0.5727392625723952,
-      "grad_norm": 0.0017200742149725556,
-      "learning_rate": 0.00019999983855525092,
-      "loss": 46.0,
-      "step": 7491
-    },
-    {
-      "epoch": 0.5728157195557849,
-      "grad_norm": 0.0012010735226795077,
-      "learning_rate": 0.0001999998385120868,
-      "loss": 46.0,
-      "step": 7492
-    },
-    {
-      "epoch": 0.5728921765391747,
-      "grad_norm": 0.00677097961306572,
-      "learning_rate": 0.00019999983846891688,
-      "loss": 46.0,
-      "step": 7493
-    },
-    {
-      "epoch": 0.5729686335225643,
-      "grad_norm": 0.004874946549534798,
-      "learning_rate": 0.00019999983842574118,
-      "loss": 46.0,
-      "step": 7494
-    },
-    {
-      "epoch": 0.5730450905059541,
-      "grad_norm": 0.001123951398767531,
-      "learning_rate": 0.00019999983838255974,
-      "loss": 46.0,
-      "step": 7495
-    },
-    {
-      "epoch": 0.5731215474893439,
-      "grad_norm": 0.0020737103186547756,
-      "learning_rate": 0.0001999998383393725,
-      "loss": 46.0,
-      "step": 7496
-    },
-    {
-      "epoch": 0.5731980044727335,
-      "grad_norm": 0.0005366839468479156,
-      "learning_rate": 0.0001999998382961795,
-      "loss": 46.0,
-      "step": 7497
-    },
-    {
-      "epoch": 0.5732744614561233,
-      "grad_norm": 0.0006064717308618128,
-      "learning_rate": 0.00019999983825298075,
-      "loss": 46.0,
-      "step": 7498
-    },
-    {
-      "epoch": 0.573350918439513,
-      "grad_norm": 0.00137209496460855,
-      "learning_rate": 0.0001999998382097762,
-      "loss": 46.0,
-      "step": 7499
-    },
-    {
-      "epoch": 0.5734273754229027,
-      "grad_norm": 0.0029947252478450537,
-      "learning_rate": 0.0001999998381665659,
-      "loss": 46.0,
-      "step": 7500
-    },
-    {
-      "epoch": 0.5735038324062924,
-      "grad_norm": 0.0044737015850842,
-      "learning_rate": 0.00019999983812334982,
-      "loss": 46.0,
-      "step": 7501
-    },
-    {
-      "epoch": 0.5735802893896821,
-      "grad_norm": 0.0005357933114282787,
-      "learning_rate": 0.000199999838080128,
-      "loss": 46.0,
-      "step": 7502
-    },
-    {
-      "epoch": 0.5736567463730718,
-      "grad_norm": 0.001026822137646377,
-      "learning_rate": 0.00019999983803690037,
-      "loss": 46.0,
-      "step": 7503
-    },
-    {
-      "epoch": 0.5737332033564616,
-      "grad_norm": 0.001595445442944765,
-      "learning_rate": 0.000199999837993667,
-      "loss": 46.0,
-      "step": 7504
-    },
-    {
-      "epoch": 0.5738096603398513,
-      "grad_norm": 0.0010531191946938634,
-      "learning_rate": 0.00019999983795042784,
-      "loss": 46.0,
-      "step": 7505
-    },
-    {
-      "epoch": 0.573886117323241,
-      "grad_norm": 0.0007644860306754708,
-      "learning_rate": 0.00019999983790718292,
-      "loss": 46.0,
-      "step": 7506
-    },
-    {
-      "epoch": 0.5739625743066308,
-      "grad_norm": 0.002083940664306283,
-      "learning_rate": 0.00019999983786393223,
-      "loss": 46.0,
-      "step": 7507
-    },
-    {
-      "epoch": 0.5740390312900204,
-      "grad_norm": 0.0017230160301551223,
-      "learning_rate": 0.00019999983782067579,
-      "loss": 46.0,
-      "step": 7508
-    },
-    {
-      "epoch": 0.5741154882734102,
-      "grad_norm": 0.001040980452671647,
-      "learning_rate": 0.00019999983777741354,
-      "loss": 46.0,
-      "step": 7509
-    },
-    {
-      "epoch": 0.5741919452567998,
-      "grad_norm": 0.0011150571517646313,
-      "learning_rate": 0.00019999983773414556,
-      "loss": 46.0,
-      "step": 7510
-    },
-    {
-      "epoch": 0.5742684022401896,
-      "grad_norm": 0.0009803981520235538,
-      "learning_rate": 0.0001999998376908718,
-      "loss": 46.0,
-      "step": 7511
-    },
-    {
-      "epoch": 0.5743448592235794,
-      "grad_norm": 0.004075199831277132,
-      "learning_rate": 0.00019999983764759226,
-      "loss": 46.0,
-      "step": 7512
-    },
-    {
-      "epoch": 0.574421316206969,
-      "grad_norm": 0.0021375571377575397,
-      "learning_rate": 0.00019999983760430695,
-      "loss": 46.0,
-      "step": 7513
-    },
-    {
-      "epoch": 0.5744977731903588,
-      "grad_norm": 0.0006680492660962045,
-      "learning_rate": 0.00019999983756101587,
-      "loss": 46.0,
-      "step": 7514
-    },
-    {
-      "epoch": 0.5745742301737485,
-      "grad_norm": 0.006769824307411909,
-      "learning_rate": 0.00019999983751771902,
-      "loss": 46.0,
-      "step": 7515
-    },
-    {
-      "epoch": 0.5746506871571382,
-      "grad_norm": 0.001280179712921381,
-      "learning_rate": 0.0001999998374744164,
-      "loss": 46.0,
-      "step": 7516
-    },
-    {
-      "epoch": 0.574727144140528,
-      "grad_norm": 0.0006041454616934061,
-      "learning_rate": 0.000199999837431108,
-      "loss": 46.0,
-      "step": 7517
-    },
-    {
-      "epoch": 0.5748036011239177,
-      "grad_norm": 0.0036631599068641663,
-      "learning_rate": 0.00019999983738779387,
-      "loss": 46.0,
-      "step": 7518
-    },
-    {
-      "epoch": 0.5748800581073074,
-      "grad_norm": 0.002040630904957652,
-      "learning_rate": 0.00019999983734447394,
-      "loss": 46.0,
-      "step": 7519
-    },
-    {
-      "epoch": 0.5749565150906971,
-      "grad_norm": 0.0007929601124487817,
-      "learning_rate": 0.00019999983730114825,
-      "loss": 46.0,
-      "step": 7520
-    },
-    {
-      "epoch": 0.5750329720740868,
-      "grad_norm": 0.0013229057658463717,
-      "learning_rate": 0.00019999983725781678,
-      "loss": 46.0,
-      "step": 7521
-    },
-    {
-      "epoch": 0.5751094290574765,
-      "grad_norm": 0.0031627847347408533,
-      "learning_rate": 0.00019999983721447957,
-      "loss": 46.0,
-      "step": 7522
-    },
-    {
-      "epoch": 0.5751858860408663,
-      "grad_norm": 0.001706066308543086,
-      "learning_rate": 0.00019999983717113655,
-      "loss": 46.0,
-      "step": 7523
-    },
-    {
-      "epoch": 0.5752623430242559,
-      "grad_norm": 0.0009224142995662987,
-      "learning_rate": 0.0001999998371277878,
-      "loss": 46.0,
-      "step": 7524
-    },
-    {
-      "epoch": 0.5753388000076457,
-      "grad_norm": 0.0008795763133093715,
-      "learning_rate": 0.00019999983708443323,
-      "loss": 46.0,
-      "step": 7525
-    },
-    {
-      "epoch": 0.5754152569910355,
-      "grad_norm": 0.0023847175762057304,
-      "learning_rate": 0.00019999983704107295,
-      "loss": 46.0,
-      "step": 7526
-    },
-    {
-      "epoch": 0.5754917139744251,
-      "grad_norm": 0.00121345988009125,
-      "learning_rate": 0.00019999983699770687,
-      "loss": 46.0,
-      "step": 7527
-    },
-    {
-      "epoch": 0.5755681709578149,
-      "grad_norm": 0.007021130528301001,
-      "learning_rate": 0.000199999836954335,
-      "loss": 46.0,
-      "step": 7528
-    },
-    {
-      "epoch": 0.5756446279412046,
-      "grad_norm": 0.001035722205415368,
-      "learning_rate": 0.00019999983691095739,
-      "loss": 46.0,
-      "step": 7529
-    },
-    {
-      "epoch": 0.5757210849245943,
-      "grad_norm": 0.0010075479513034225,
-      "learning_rate": 0.00019999983686757398,
-      "loss": 46.0,
-      "step": 7530
-    },
-    {
-      "epoch": 0.575797541907984,
-      "grad_norm": 0.004983509890735149,
-      "learning_rate": 0.00019999983682418484,
-      "loss": 46.0,
-      "step": 7531
-    },
-    {
-      "epoch": 0.5758739988913737,
-      "grad_norm": 0.0005552451475523412,
-      "learning_rate": 0.00019999983678078992,
-      "loss": 46.0,
-      "step": 7532
-    },
-    {
-      "epoch": 0.5759504558747635,
-      "grad_norm": 0.0005114002269692719,
-      "learning_rate": 0.00019999983673738922,
-      "loss": 46.0,
-      "step": 7533
-    },
-    {
-      "epoch": 0.5760269128581532,
-      "grad_norm": 0.0009269884903915226,
-      "learning_rate": 0.00019999983669398275,
-      "loss": 46.0,
-      "step": 7534
-    },
-    {
-      "epoch": 0.5761033698415429,
-      "grad_norm": 0.0003901891177520156,
-      "learning_rate": 0.0001999998366505705,
-      "loss": 46.0,
-      "step": 7535
-    },
-    {
-      "epoch": 0.5761798268249326,
-      "grad_norm": 0.0006088235531933606,
-      "learning_rate": 0.00019999983660715253,
-      "loss": 46.0,
-      "step": 7536
-    },
-    {
-      "epoch": 0.5762562838083224,
-      "grad_norm": 0.0029156827367842197,
-      "learning_rate": 0.00019999983656372874,
-      "loss": 46.0,
-      "step": 7537
-    },
-    {
-      "epoch": 0.576332740791712,
-      "grad_norm": 0.0011136867105960846,
-      "learning_rate": 0.0001999998365202992,
-      "loss": 46.0,
-      "step": 7538
-    },
-    {
-      "epoch": 0.5764091977751018,
-      "grad_norm": 0.0053533646278083324,
-      "learning_rate": 0.00019999983647686392,
-      "loss": 46.0,
-      "step": 7539
-    },
-    {
-      "epoch": 0.5764856547584915,
-      "grad_norm": 0.0015955432318150997,
-      "learning_rate": 0.00019999983643342282,
-      "loss": 46.0,
-      "step": 7540
-    },
-    {
-      "epoch": 0.5765621117418812,
-      "grad_norm": 0.0009845118038356304,
-      "learning_rate": 0.00019999983638997596,
-      "loss": 46.0,
-      "step": 7541
-    },
-    {
-      "epoch": 0.576638568725271,
-      "grad_norm": 0.0016344516770914197,
-      "learning_rate": 0.00019999983634652336,
-      "loss": 46.0,
-      "step": 7542
-    },
-    {
-      "epoch": 0.5767150257086606,
-      "grad_norm": 0.001754777505993843,
-      "learning_rate": 0.00019999983630306496,
-      "loss": 46.0,
-      "step": 7543
-    },
-    {
-      "epoch": 0.5767914826920504,
-      "grad_norm": 0.0009846682660281658,
-      "learning_rate": 0.00019999983625960079,
-      "loss": 46.0,
-      "step": 7544
-    },
-    {
-      "epoch": 0.5768679396754401,
-      "grad_norm": 0.0007837713928893209,
-      "learning_rate": 0.00019999983621613087,
-      "loss": 46.0,
-      "step": 7545
-    },
-    {
-      "epoch": 0.5769443966588298,
-      "grad_norm": 0.0015967205399647355,
-      "learning_rate": 0.00019999983617265517,
-      "loss": 46.0,
-      "step": 7546
-    },
-    {
-      "epoch": 0.5770208536422196,
-      "grad_norm": 0.004502040334045887,
-      "learning_rate": 0.0001999998361291737,
-      "loss": 46.0,
-      "step": 7547
-    },
-    {
-      "epoch": 0.5770973106256093,
-      "grad_norm": 0.001678119646385312,
-      "learning_rate": 0.00019999983608568646,
-      "loss": 46.0,
-      "step": 7548
-    },
-    {
-      "epoch": 0.577173767608999,
-      "grad_norm": 0.0014648191863670945,
-      "learning_rate": 0.00019999983604219345,
-      "loss": 46.0,
-      "step": 7549
-    },
-    {
-      "epoch": 0.5772502245923887,
-      "grad_norm": 0.003734854981303215,
-      "learning_rate": 0.00019999983599869466,
-      "loss": 46.0,
-      "step": 7550
-    },
-    {
-      "epoch": 0.5773266815757784,
-      "grad_norm": 0.0024289051070809364,
-      "learning_rate": 0.00019999983595519013,
-      "loss": 46.0,
-      "step": 7551
-    },
-    {
-      "epoch": 0.5774031385591681,
-      "grad_norm": 0.005789545364677906,
-      "learning_rate": 0.00019999983591167982,
-      "loss": 46.0,
-      "step": 7552
-    },
-    {
-      "epoch": 0.5774795955425579,
-      "grad_norm": 0.0009317718795500696,
-      "learning_rate": 0.00019999983586816374,
-      "loss": 46.0,
-      "step": 7553
-    },
-    {
-      "epoch": 0.5775560525259475,
-      "grad_norm": 0.0021060071885585785,
-      "learning_rate": 0.00019999983582464191,
-      "loss": 46.0,
-      "step": 7554
-    },
-    {
-      "epoch": 0.5776325095093373,
-      "grad_norm": 0.0016839266754686832,
-      "learning_rate": 0.0001999998357811143,
-      "loss": 46.0,
-      "step": 7555
-    },
-    {
-      "epoch": 0.5777089664927271,
-      "grad_norm": 0.001243123202584684,
-      "learning_rate": 0.00019999983573758089,
-      "loss": 46.0,
-      "step": 7556
-    },
-    {
-      "epoch": 0.5777854234761167,
-      "grad_norm": 0.000614280579611659,
-      "learning_rate": 0.0001999998356940417,
-      "loss": 46.0,
-      "step": 7557
-    },
-    {
-      "epoch": 0.5778618804595065,
-      "grad_norm": 0.002122477162629366,
-      "learning_rate": 0.0001999998356504968,
-      "loss": 46.0,
-      "step": 7558
-    },
-    {
-      "epoch": 0.5779383374428962,
-      "grad_norm": 0.0017634991090744734,
-      "learning_rate": 0.00019999983560694607,
-      "loss": 46.0,
-      "step": 7559
-    },
-    {
-      "epoch": 0.5780147944262859,
-      "grad_norm": 0.0014423084212467074,
-      "learning_rate": 0.0001999998355633896,
-      "loss": 46.0,
-      "step": 7560
-    },
-    {
-      "epoch": 0.5780912514096757,
-      "grad_norm": 0.00041379593312740326,
-      "learning_rate": 0.00019999983551982736,
-      "loss": 46.0,
-      "step": 7561
-    },
-    {
-      "epoch": 0.5781677083930653,
-      "grad_norm": 0.00222440785728395,
-      "learning_rate": 0.00019999983547625935,
-      "loss": 46.0,
-      "step": 7562
-    },
-    {
-      "epoch": 0.5782441653764551,
-      "grad_norm": 0.0009318541851826012,
-      "learning_rate": 0.00019999983543268556,
-      "loss": 46.0,
-      "step": 7563
-    },
-    {
-      "epoch": 0.5783206223598448,
-      "grad_norm": 0.0007590082241222262,
-      "learning_rate": 0.00019999983538910603,
-      "loss": 46.0,
-      "step": 7564
-    },
-    {
-      "epoch": 0.5783970793432345,
-      "grad_norm": 0.004010630305856466,
-      "learning_rate": 0.00019999983534552072,
-      "loss": 46.0,
-      "step": 7565
-    },
-    {
-      "epoch": 0.5784735363266242,
-      "grad_norm": 0.0023934165947139263,
-      "learning_rate": 0.00019999983530192962,
-      "loss": 46.0,
-      "step": 7566
-    },
-    {
-      "epoch": 0.578549993310014,
-      "grad_norm": 0.0020920245442539454,
-      "learning_rate": 0.00019999983525833276,
-      "loss": 46.0,
-      "step": 7567
-    },
-    {
-      "epoch": 0.5786264502934036,
-      "grad_norm": 0.0008549208287149668,
-      "learning_rate": 0.00019999983521473013,
-      "loss": 46.0,
-      "step": 7568
-    },
-    {
-      "epoch": 0.5787029072767934,
-      "grad_norm": 0.002753807231783867,
-      "learning_rate": 0.00019999983517112176,
-      "loss": 46.0,
-      "step": 7569
-    },
-    {
-      "epoch": 0.5787793642601831,
-      "grad_norm": 0.0021882413420826197,
-      "learning_rate": 0.0001999998351275076,
-      "loss": 46.0,
-      "step": 7570
-    },
-    {
-      "epoch": 0.5788558212435728,
-      "grad_norm": 0.001145792193710804,
-      "learning_rate": 0.00019999983508388764,
-      "loss": 46.0,
-      "step": 7571
-    },
-    {
-      "epoch": 0.5789322782269626,
-      "grad_norm": 0.000683027203194797,
-      "learning_rate": 0.00019999983504026195,
-      "loss": 46.0,
-      "step": 7572
-    },
-    {
-      "epoch": 0.5790087352103522,
-      "grad_norm": 0.0017354689771309495,
-      "learning_rate": 0.00019999983499663045,
-      "loss": 46.0,
-      "step": 7573
-    },
-    {
-      "epoch": 0.579085192193742,
-      "grad_norm": 0.0015501282177865505,
-      "learning_rate": 0.00019999983495299324,
-      "loss": 46.0,
-      "step": 7574
-    },
-    {
-      "epoch": 0.5791616491771318,
-      "grad_norm": 0.0053193713538348675,
-      "learning_rate": 0.00019999983490935023,
-      "loss": 46.0,
-      "step": 7575
-    },
-    {
-      "epoch": 0.5792381061605214,
-      "grad_norm": 0.0010329082142561674,
-      "learning_rate": 0.0001999998348657014,
-      "loss": 46.0,
-      "step": 7576
-    },
-    {
-      "epoch": 0.5793145631439112,
-      "grad_norm": 0.008275882340967655,
-      "learning_rate": 0.00019999983482204688,
-      "loss": 46.0,
-      "step": 7577
-    },
-    {
-      "epoch": 0.5793910201273009,
-      "grad_norm": 0.004763545002788305,
-      "learning_rate": 0.00019999983477838655,
-      "loss": 46.0,
-      "step": 7578
-    },
-    {
-      "epoch": 0.5794674771106906,
-      "grad_norm": 0.001086041796952486,
-      "learning_rate": 0.00019999983473472047,
-      "loss": 46.0,
-      "step": 7579
-    },
-    {
-      "epoch": 0.5795439340940803,
-      "grad_norm": 0.0013631421606987715,
-      "learning_rate": 0.0001999998346910486,
-      "loss": 46.0,
-      "step": 7580
-    },
-    {
-      "epoch": 0.57962039107747,
-      "grad_norm": 0.0008691237308084965,
-      "learning_rate": 0.000199999834647371,
-      "loss": 46.0,
-      "step": 7581
-    },
-    {
-      "epoch": 0.5796968480608597,
-      "grad_norm": 0.001286480575799942,
-      "learning_rate": 0.0001999998346036876,
-      "loss": 46.0,
-      "step": 7582
-    },
-    {
-      "epoch": 0.5797733050442495,
-      "grad_norm": 0.0023991360794752836,
-      "learning_rate": 0.0001999998345599984,
-      "loss": 46.0,
-      "step": 7583
-    },
-    {
-      "epoch": 0.5798497620276392,
-      "grad_norm": 0.0008301687194034457,
-      "learning_rate": 0.0001999998345163035,
-      "loss": 46.0,
-      "step": 7584
-    },
-    {
-      "epoch": 0.5799262190110289,
-      "grad_norm": 0.0012167587410658598,
-      "learning_rate": 0.00019999983447260277,
-      "loss": 46.0,
-      "step": 7585
-    },
-    {
-      "epoch": 0.5800026759944187,
-      "grad_norm": 0.00042961325380019844,
-      "learning_rate": 0.00019999983442889628,
-      "loss": 46.0,
-      "step": 7586
-    },
-    {
-      "epoch": 0.5800791329778083,
-      "grad_norm": 0.0027072043158113956,
-      "learning_rate": 0.00019999983438518407,
-      "loss": 46.0,
-      "step": 7587
-    },
-    {
-      "epoch": 0.5801555899611981,
-      "grad_norm": 0.0007777933496981859,
-      "learning_rate": 0.00019999983434146605,
-      "loss": 46.0,
-      "step": 7588
-    },
-    {
-      "epoch": 0.5802320469445879,
-      "grad_norm": 0.0025236657820641994,
-      "learning_rate": 0.00019999983429774227,
-      "loss": 46.0,
-      "step": 7589
-    },
-    {
-      "epoch": 0.5803085039279775,
-      "grad_norm": 0.00039633637061342597,
-      "learning_rate": 0.0001999998342540127,
-      "loss": 46.0,
-      "step": 7590
-    },
-    {
-      "epoch": 0.5803849609113673,
-      "grad_norm": 0.0013137951027601957,
-      "learning_rate": 0.0001999998342102774,
-      "loss": 46.0,
-      "step": 7591
-    },
-    {
-      "epoch": 0.5804614178947569,
-      "grad_norm": 0.00047574235941283405,
-      "learning_rate": 0.0001999998341665363,
-      "loss": 46.0,
-      "step": 7592
-    },
-    {
-      "epoch": 0.5805378748781467,
-      "grad_norm": 0.0009591268608346581,
-      "learning_rate": 0.00019999983412278941,
-      "loss": 46.0,
-      "step": 7593
-    },
-    {
-      "epoch": 0.5806143318615364,
-      "grad_norm": 0.009303290396928787,
-      "learning_rate": 0.0001999998340790368,
-      "loss": 46.0,
-      "step": 7594
-    },
-    {
-      "epoch": 0.5806907888449261,
-      "grad_norm": 0.007136994507163763,
-      "learning_rate": 0.0001999998340352784,
-      "loss": 46.0,
-      "step": 7595
-    },
-    {
-      "epoch": 0.5807672458283158,
-      "grad_norm": 0.0003480181621853262,
-      "learning_rate": 0.00019999983399151422,
-      "loss": 46.0,
-      "step": 7596
-    },
-    {
-      "epoch": 0.5808437028117056,
-      "grad_norm": 0.0011027996661141515,
-      "learning_rate": 0.0001999998339477443,
-      "loss": 46.0,
-      "step": 7597
-    },
-    {
-      "epoch": 0.5809201597950953,
-      "grad_norm": 0.0011567899491637945,
-      "learning_rate": 0.00019999983390396858,
-      "loss": 46.0,
-      "step": 7598
-    },
-    {
-      "epoch": 0.580996616778485,
-      "grad_norm": 0.0067599681206047535,
-      "learning_rate": 0.00019999983386018708,
-      "loss": 46.0,
-      "step": 7599
-    },
-    {
-      "epoch": 0.5810730737618747,
-      "grad_norm": 0.002067169174551964,
-      "learning_rate": 0.00019999983381639984,
-      "loss": 46.0,
-      "step": 7600
-    },
-    {
-      "epoch": 0.5811495307452644,
-      "grad_norm": 0.0005350538995116949,
-      "learning_rate": 0.00019999983377260683,
-      "loss": 46.0,
-      "step": 7601
-    },
-    {
-      "epoch": 0.5812259877286542,
-      "grad_norm": 0.006487096194177866,
-      "learning_rate": 0.00019999983372880805,
-      "loss": 46.0,
-      "step": 7602
-    },
-    {
-      "epoch": 0.5813024447120438,
-      "grad_norm": 0.0007618963718414307,
-      "learning_rate": 0.00019999983368500349,
-      "loss": 46.0,
-      "step": 7603
-    },
-    {
-      "epoch": 0.5813789016954336,
-      "grad_norm": 0.0011088476749137044,
-      "learning_rate": 0.00019999983364119315,
-      "loss": 46.0,
-      "step": 7604
-    },
-    {
-      "epoch": 0.5814553586788234,
-      "grad_norm": 0.0008921415428631008,
-      "learning_rate": 0.00019999983359737707,
-      "loss": 46.0,
-      "step": 7605
-    },
-    {
-      "epoch": 0.581531815662213,
-      "grad_norm": 0.0014126216992735863,
-      "learning_rate": 0.0001999998335535552,
-      "loss": 46.0,
-      "step": 7606
-    },
-    {
-      "epoch": 0.5816082726456028,
-      "grad_norm": 0.001030371873639524,
-      "learning_rate": 0.00019999983350972754,
-      "loss": 46.0,
-      "step": 7607
-    },
-    {
-      "epoch": 0.5816847296289925,
-      "grad_norm": 0.000780138245318085,
-      "learning_rate": 0.00019999983346589417,
-      "loss": 46.0,
-      "step": 7608
-    },
-    {
-      "epoch": 0.5817611866123822,
-      "grad_norm": 0.006118282675743103,
-      "learning_rate": 0.000199999833422055,
-      "loss": 46.0,
-      "step": 7609
-    },
-    {
-      "epoch": 0.581837643595772,
-      "grad_norm": 0.0013880191836506128,
-      "learning_rate": 0.00019999983337821005,
-      "loss": 46.0,
-      "step": 7610
-    },
-    {
-      "epoch": 0.5819141005791616,
-      "grad_norm": 0.0011992092477157712,
-      "learning_rate": 0.00019999983333435933,
-      "loss": 46.0,
-      "step": 7611
-    },
-    {
-      "epoch": 0.5819905575625514,
-      "grad_norm": 0.0006638895138166845,
-      "learning_rate": 0.00019999983329050286,
-      "loss": 46.0,
-      "step": 7612
-    },
-    {
-      "epoch": 0.5820670145459411,
-      "grad_norm": 0.00025403572362847626,
-      "learning_rate": 0.0001999998332466406,
-      "loss": 46.0,
-      "step": 7613
-    },
-    {
-      "epoch": 0.5821434715293308,
-      "grad_norm": 0.0013250509509816766,
-      "learning_rate": 0.00019999983320277259,
-      "loss": 46.0,
-      "step": 7614
-    },
-    {
-      "epoch": 0.5822199285127205,
-      "grad_norm": 0.0005754359299317002,
-      "learning_rate": 0.0001999998331588988,
-      "loss": 46.0,
-      "step": 7615
-    },
-    {
-      "epoch": 0.5822963854961103,
-      "grad_norm": 0.0012737332144752145,
-      "learning_rate": 0.00019999983311501924,
-      "loss": 46.0,
-      "step": 7616
-    },
-    {
-      "epoch": 0.5823728424794999,
-      "grad_norm": 0.0009545432985760272,
-      "learning_rate": 0.0001999998330711339,
-      "loss": 46.0,
-      "step": 7617
-    },
-    {
-      "epoch": 0.5824492994628897,
-      "grad_norm": 0.0007713224622420967,
-      "learning_rate": 0.0001999998330272428,
-      "loss": 46.0,
-      "step": 7618
-    },
-    {
-      "epoch": 0.5825257564462795,
-      "grad_norm": 0.0009839808335527778,
-      "learning_rate": 0.00019999983298334595,
-      "loss": 46.0,
-      "step": 7619
-    },
-    {
-      "epoch": 0.5826022134296691,
-      "grad_norm": 0.001255712122656405,
-      "learning_rate": 0.0001999998329394433,
-      "loss": 46.0,
-      "step": 7620
-    },
-    {
-      "epoch": 0.5826786704130589,
-      "grad_norm": 0.0008356580510735512,
-      "learning_rate": 0.0001999998328955349,
-      "loss": 46.0,
-      "step": 7621
-    },
-    {
-      "epoch": 0.5827551273964485,
-      "grad_norm": 0.00168485043104738,
-      "learning_rate": 0.00019999983285162072,
-      "loss": 46.0,
-      "step": 7622
-    },
-    {
-      "epoch": 0.5828315843798383,
-      "grad_norm": 0.0011813770979642868,
-      "learning_rate": 0.00019999983280770075,
-      "loss": 46.0,
-      "step": 7623
-    },
-    {
-      "epoch": 0.582908041363228,
-      "grad_norm": 0.0008950316114351153,
-      "learning_rate": 0.00019999983276377506,
-      "loss": 46.0,
-      "step": 7624
-    },
-    {
-      "epoch": 0.5829844983466177,
-      "grad_norm": 0.001904475619085133,
-      "learning_rate": 0.00019999983271984357,
-      "loss": 46.0,
-      "step": 7625
-    },
-    {
-      "epoch": 0.5830609553300075,
-      "grad_norm": 0.0005529436166398227,
-      "learning_rate": 0.00019999983267590633,
-      "loss": 46.0,
-      "step": 7626
-    },
-    {
-      "epoch": 0.5831374123133972,
-      "grad_norm": 0.00045964884338900447,
-      "learning_rate": 0.0001999998326319633,
-      "loss": 46.0,
-      "step": 7627
-    },
-    {
-      "epoch": 0.5832138692967869,
-      "grad_norm": 0.00109334965236485,
-      "learning_rate": 0.0001999998325880145,
-      "loss": 46.0,
-      "step": 7628
-    },
-    {
-      "epoch": 0.5832903262801766,
-      "grad_norm": 0.0016947025433182716,
-      "learning_rate": 0.00019999983254405994,
-      "loss": 46.0,
-      "step": 7629
-    },
-    {
-      "epoch": 0.5833667832635664,
-      "grad_norm": 0.0005835481570102274,
-      "learning_rate": 0.0001999998325000996,
-      "loss": 46.0,
-      "step": 7630
-    },
-    {
-      "epoch": 0.583443240246956,
-      "grad_norm": 0.0016982399392873049,
-      "learning_rate": 0.0001999998324561335,
-      "loss": 46.0,
-      "step": 7631
-    },
-    {
-      "epoch": 0.5835196972303458,
-      "grad_norm": 0.00038127953303046525,
-      "learning_rate": 0.00019999983241216165,
-      "loss": 46.0,
-      "step": 7632
-    },
-    {
-      "epoch": 0.5835961542137355,
-      "grad_norm": 0.0013964161043986678,
-      "learning_rate": 0.000199999832368184,
-      "loss": 46.0,
-      "step": 7633
-    },
-    {
-      "epoch": 0.5836726111971252,
-      "grad_norm": 0.0006763490382581949,
-      "learning_rate": 0.00019999983232420058,
-      "loss": 46.0,
-      "step": 7634
-    },
-    {
-      "epoch": 0.583749068180515,
-      "grad_norm": 0.0017157947877421975,
-      "learning_rate": 0.00019999983228021143,
-      "loss": 46.0,
-      "step": 7635
-    },
-    {
-      "epoch": 0.5838255251639046,
-      "grad_norm": 0.0012805898441001773,
-      "learning_rate": 0.00019999983223621649,
-      "loss": 46.0,
-      "step": 7636
-    },
-    {
-      "epoch": 0.5839019821472944,
-      "grad_norm": 0.0017527329036965966,
-      "learning_rate": 0.00019999983219221574,
-      "loss": 46.0,
-      "step": 7637
-    },
-    {
-      "epoch": 0.5839784391306841,
-      "grad_norm": 0.0005397163331508636,
-      "learning_rate": 0.00019999983214820927,
-      "loss": 46.0,
-      "step": 7638
-    },
-    {
-      "epoch": 0.5840548961140738,
-      "grad_norm": 0.0017146962927654386,
-      "learning_rate": 0.000199999832104197,
-      "loss": 46.0,
-      "step": 7639
-    },
-    {
-      "epoch": 0.5841313530974636,
-      "grad_norm": 0.0013553574681282043,
-      "learning_rate": 0.00019999983206017897,
-      "loss": 46.0,
-      "step": 7640
-    },
-    {
-      "epoch": 0.5842078100808532,
-      "grad_norm": 0.001664946204982698,
-      "learning_rate": 0.00019999983201615518,
-      "loss": 46.0,
-      "step": 7641
-    },
-    {
-      "epoch": 0.584284267064243,
-      "grad_norm": 0.01198570430278778,
-      "learning_rate": 0.00019999983197212563,
-      "loss": 46.0,
-      "step": 7642
-    },
-    {
-      "epoch": 0.5843607240476327,
-      "grad_norm": 0.0011399604845792055,
-      "learning_rate": 0.0001999998319280903,
-      "loss": 46.0,
-      "step": 7643
-    },
-    {
-      "epoch": 0.5844371810310224,
-      "grad_norm": 0.004268769174814224,
-      "learning_rate": 0.0001999998318840492,
-      "loss": 46.0,
-      "step": 7644
-    },
-    {
-      "epoch": 0.5845136380144121,
-      "grad_norm": 0.00304241175763309,
-      "learning_rate": 0.0001999998318400023,
-      "loss": 46.0,
-      "step": 7645
-    },
-    {
-      "epoch": 0.5845900949978019,
-      "grad_norm": 0.0013686362653970718,
-      "learning_rate": 0.00019999983179594968,
-      "loss": 46.0,
-      "step": 7646
-    },
-    {
-      "epoch": 0.5846665519811916,
-      "grad_norm": 0.0018477820558473468,
-      "learning_rate": 0.00019999983175189126,
-      "loss": 46.0,
-      "step": 7647
-    },
-    {
-      "epoch": 0.5847430089645813,
-      "grad_norm": 0.008912919089198112,
-      "learning_rate": 0.00019999983170782706,
-      "loss": 46.0,
-      "step": 7648
-    },
-    {
-      "epoch": 0.5848194659479711,
-      "grad_norm": 0.002248775213956833,
-      "learning_rate": 0.00019999983166375712,
-      "loss": 46.0,
-      "step": 7649
-    },
-    {
-      "epoch": 0.5848959229313607,
-      "grad_norm": 0.014130390249192715,
-      "learning_rate": 0.0001999998316196814,
-      "loss": 46.0,
-      "step": 7650
-    },
-    {
-      "epoch": 0.5849723799147505,
-      "grad_norm": 0.0009850258938968182,
-      "learning_rate": 0.0001999998315755999,
-      "loss": 46.0,
-      "step": 7651
-    },
-    {
-      "epoch": 0.5850488368981401,
-      "grad_norm": 0.0012152059935033321,
-      "learning_rate": 0.00019999983153151264,
-      "loss": 46.0,
-      "step": 7652
-    },
-    {
-      "epoch": 0.5851252938815299,
-      "grad_norm": 0.0006443813326768577,
-      "learning_rate": 0.00019999983148741963,
-      "loss": 46.0,
-      "step": 7653
-    },
-    {
-      "epoch": 0.5852017508649197,
-      "grad_norm": 0.006645088084042072,
-      "learning_rate": 0.00019999983144332082,
-      "loss": 46.0,
-      "step": 7654
-    },
-    {
-      "epoch": 0.5852782078483093,
-      "grad_norm": 0.0019394843839108944,
-      "learning_rate": 0.00019999983139921626,
-      "loss": 46.0,
-      "step": 7655
-    },
-    {
-      "epoch": 0.5853546648316991,
-      "grad_norm": 0.0010840287432074547,
-      "learning_rate": 0.00019999983135510593,
-      "loss": 46.0,
-      "step": 7656
-    },
-    {
-      "epoch": 0.5854311218150888,
-      "grad_norm": 0.0036767369601875544,
-      "learning_rate": 0.00019999983131098982,
-      "loss": 46.0,
-      "step": 7657
-    },
-    {
-      "epoch": 0.5855075787984785,
-      "grad_norm": 0.002124917460605502,
-      "learning_rate": 0.00019999983126686792,
-      "loss": 46.0,
-      "step": 7658
-    },
-    {
-      "epoch": 0.5855840357818682,
-      "grad_norm": 0.004471937194466591,
-      "learning_rate": 0.0001999998312227403,
-      "loss": 46.0,
-      "step": 7659
-    },
-    {
-      "epoch": 0.585660492765258,
-      "grad_norm": 0.0017449866281822324,
-      "learning_rate": 0.00019999983117860687,
-      "loss": 46.0,
-      "step": 7660
-    },
-    {
-      "epoch": 0.5857369497486477,
-      "grad_norm": 0.001235291245393455,
-      "learning_rate": 0.0001999998311344677,
-      "loss": 46.0,
-      "step": 7661
-    },
-    {
-      "epoch": 0.5858134067320374,
-      "grad_norm": 0.0008355448953807354,
-      "learning_rate": 0.00019999983109032275,
-      "loss": 46.0,
-      "step": 7662
-    },
-    {
-      "epoch": 0.5858898637154271,
-      "grad_norm": 0.0008515469380654395,
-      "learning_rate": 0.000199999831046172,
-      "loss": 46.0,
-      "step": 7663
-    },
-    {
-      "epoch": 0.5859663206988168,
-      "grad_norm": 0.0006487657665275037,
-      "learning_rate": 0.0001999998310020155,
-      "loss": 46.0,
-      "step": 7664
-    },
-    {
-      "epoch": 0.5860427776822066,
-      "grad_norm": 0.0010451796697452664,
-      "learning_rate": 0.00019999983095785327,
-      "loss": 46.0,
-      "step": 7665
-    },
-    {
-      "epoch": 0.5861192346655962,
-      "grad_norm": 0.0032195597887039185,
-      "learning_rate": 0.00019999983091368524,
-      "loss": 46.0,
-      "step": 7666
-    },
-    {
-      "epoch": 0.586195691648986,
-      "grad_norm": 0.0005423553520813584,
-      "learning_rate": 0.00019999983086951142,
-      "loss": 46.0,
-      "step": 7667
-    },
-    {
-      "epoch": 0.5862721486323758,
-      "grad_norm": 0.002259672386571765,
-      "learning_rate": 0.00019999983082533187,
-      "loss": 46.0,
-      "step": 7668
-    },
-    {
-      "epoch": 0.5863486056157654,
-      "grad_norm": 0.002088018227368593,
-      "learning_rate": 0.0001999998307811465,
-      "loss": 46.0,
-      "step": 7669
-    },
-    {
-      "epoch": 0.5864250625991552,
-      "grad_norm": 0.0031793571542948484,
-      "learning_rate": 0.0001999998307369554,
-      "loss": 46.0,
-      "step": 7670
-    },
-    {
-      "epoch": 0.5865015195825448,
-      "grad_norm": 0.002627237234264612,
-      "learning_rate": 0.00019999983069275855,
-      "loss": 46.0,
-      "step": 7671
-    },
-    {
-      "epoch": 0.5865779765659346,
-      "grad_norm": 0.0046811169013381,
-      "learning_rate": 0.0001999998306485559,
-      "loss": 46.0,
-      "step": 7672
-    },
-    {
-      "epoch": 0.5866544335493243,
-      "grad_norm": 0.0012938586296513677,
-      "learning_rate": 0.00019999983060434745,
-      "loss": 46.0,
-      "step": 7673
-    },
-    {
-      "epoch": 0.586730890532714,
-      "grad_norm": 0.002174362074583769,
-      "learning_rate": 0.00019999983056013328,
-      "loss": 46.0,
-      "step": 7674
-    },
-    {
-      "epoch": 0.5868073475161037,
-      "grad_norm": 0.0023047400172799826,
-      "learning_rate": 0.00019999983051591333,
-      "loss": 46.0,
-      "step": 7675
-    },
-    {
-      "epoch": 0.5868838044994935,
-      "grad_norm": 0.0009289035806432366,
-      "learning_rate": 0.0001999998304716876,
-      "loss": 46.0,
-      "step": 7676
-    },
-    {
-      "epoch": 0.5869602614828832,
-      "grad_norm": 0.0010724244639277458,
-      "learning_rate": 0.0001999998304274561,
-      "loss": 46.0,
-      "step": 7677
-    },
-    {
-      "epoch": 0.5870367184662729,
-      "grad_norm": 0.0007620546966791153,
-      "learning_rate": 0.00019999983038321883,
-      "loss": 46.0,
-      "step": 7678
-    },
-    {
-      "epoch": 0.5871131754496627,
-      "grad_norm": 0.0021984796039760113,
-      "learning_rate": 0.0001999998303389758,
-      "loss": 46.0,
-      "step": 7679
-    },
-    {
-      "epoch": 0.5871896324330523,
-      "grad_norm": 0.001679317676462233,
-      "learning_rate": 0.000199999830294727,
-      "loss": 46.0,
-      "step": 7680
-    },
-    {
-      "epoch": 0.5872660894164421,
-      "grad_norm": 0.0013205293798819184,
-      "learning_rate": 0.00019999983025047242,
-      "loss": 46.0,
-      "step": 7681
-    },
-    {
-      "epoch": 0.5873425463998317,
-      "grad_norm": 0.0008109578629955649,
-      "learning_rate": 0.0001999998302062121,
-      "loss": 46.0,
-      "step": 7682
-    },
-    {
-      "epoch": 0.5874190033832215,
-      "grad_norm": 0.0011380797950550914,
-      "learning_rate": 0.00019999983016194596,
-      "loss": 46.0,
-      "step": 7683
-    },
-    {
-      "epoch": 0.5874954603666113,
-      "grad_norm": 0.0008693085401318967,
-      "learning_rate": 0.00019999983011767408,
-      "loss": 46.0,
-      "step": 7684
-    },
-    {
-      "epoch": 0.5875719173500009,
-      "grad_norm": 0.0016144621185958385,
-      "learning_rate": 0.00019999983007339643,
-      "loss": 46.0,
-      "step": 7685
-    },
-    {
-      "epoch": 0.5876483743333907,
-      "grad_norm": 0.0011661348398774862,
-      "learning_rate": 0.000199999830029113,
-      "loss": 46.0,
-      "step": 7686
-    },
-    {
-      "epoch": 0.5877248313167804,
-      "grad_norm": 0.0020094676874578,
-      "learning_rate": 0.00019999982998482384,
-      "loss": 46.0,
-      "step": 7687
-    },
-    {
-      "epoch": 0.5878012883001701,
-      "grad_norm": 0.003406036412343383,
-      "learning_rate": 0.0001999998299405289,
-      "loss": 46.0,
-      "step": 7688
-    },
-    {
-      "epoch": 0.5878777452835598,
-      "grad_norm": 0.0016322359442710876,
-      "learning_rate": 0.00019999982989622815,
-      "loss": 46.0,
-      "step": 7689
-    },
-    {
-      "epoch": 0.5879542022669496,
-      "grad_norm": 0.003237111261114478,
-      "learning_rate": 0.00019999982985192163,
-      "loss": 46.0,
-      "step": 7690
-    },
-    {
-      "epoch": 0.5880306592503393,
-      "grad_norm": 0.0024859909899532795,
-      "learning_rate": 0.0001999998298076094,
-      "loss": 46.0,
-      "step": 7691
-    },
-    {
-      "epoch": 0.588107116233729,
-      "grad_norm": 0.004241071175783873,
-      "learning_rate": 0.00019999982976329133,
-      "loss": 46.0,
-      "step": 7692
-    },
-    {
-      "epoch": 0.5881835732171187,
-      "grad_norm": 0.0011420173104852438,
-      "learning_rate": 0.00019999982971896754,
-      "loss": 46.0,
-      "step": 7693
-    },
-    {
-      "epoch": 0.5882600302005084,
-      "grad_norm": 0.0020061784889549017,
-      "learning_rate": 0.00019999982967463796,
-      "loss": 46.0,
-      "step": 7694
-    },
-    {
-      "epoch": 0.5883364871838982,
-      "grad_norm": 0.0017800317145884037,
-      "learning_rate": 0.0001999998296303026,
-      "loss": 46.0,
-      "step": 7695
-    },
-    {
-      "epoch": 0.5884129441672878,
-      "grad_norm": 0.002415225375443697,
-      "learning_rate": 0.0001999998295859615,
-      "loss": 46.0,
-      "step": 7696
-    },
-    {
-      "epoch": 0.5884894011506776,
-      "grad_norm": 0.0008627810748293996,
-      "learning_rate": 0.0001999998295416146,
-      "loss": 46.0,
-      "step": 7697
-    },
-    {
-      "epoch": 0.5885658581340674,
-      "grad_norm": 0.0005771710420958698,
-      "learning_rate": 0.00019999982949726194,
-      "loss": 46.0,
-      "step": 7698
-    },
-    {
-      "epoch": 0.588642315117457,
-      "grad_norm": 0.00194309011567384,
-      "learning_rate": 0.00019999982945290352,
-      "loss": 46.0,
-      "step": 7699
-    },
-    {
-      "epoch": 0.5887187721008468,
-      "grad_norm": 0.001984003931283951,
-      "learning_rate": 0.00019999982940853935,
-      "loss": 46.0,
-      "step": 7700
-    },
-    {
-      "epoch": 0.5887952290842364,
-      "grad_norm": 0.0009429525234736502,
-      "learning_rate": 0.00019999982936416938,
-      "loss": 46.0,
-      "step": 7701
-    },
-    {
-      "epoch": 0.5888716860676262,
-      "grad_norm": 0.013471740297973156,
-      "learning_rate": 0.00019999982931979363,
-      "loss": 46.0,
-      "step": 7702
-    },
-    {
-      "epoch": 0.588948143051016,
-      "grad_norm": 0.0010712002404034138,
-      "learning_rate": 0.00019999982927541214,
-      "loss": 46.0,
-      "step": 7703
-    },
-    {
-      "epoch": 0.5890246000344056,
-      "grad_norm": 0.0004987602587789297,
-      "learning_rate": 0.00019999982923102488,
-      "loss": 46.0,
-      "step": 7704
-    },
-    {
-      "epoch": 0.5891010570177954,
-      "grad_norm": 0.0008510476327501237,
-      "learning_rate": 0.00019999982918663184,
-      "loss": 46.0,
-      "step": 7705
-    },
-    {
-      "epoch": 0.5891775140011851,
-      "grad_norm": 0.004711621440947056,
-      "learning_rate": 0.00019999982914223303,
-      "loss": 46.0,
-      "step": 7706
-    },
-    {
-      "epoch": 0.5892539709845748,
-      "grad_norm": 0.0006135948933660984,
-      "learning_rate": 0.00019999982909782844,
-      "loss": 46.0,
-      "step": 7707
-    },
-    {
-      "epoch": 0.5893304279679645,
-      "grad_norm": 0.00043897968134842813,
-      "learning_rate": 0.0001999998290534181,
-      "loss": 46.0,
-      "step": 7708
-    },
-    {
-      "epoch": 0.5894068849513543,
-      "grad_norm": 0.0013876077719032764,
-      "learning_rate": 0.00019999982900900198,
-      "loss": 46.0,
-      "step": 7709
-    },
-    {
-      "epoch": 0.5894833419347439,
-      "grad_norm": 0.0017914950149133801,
-      "learning_rate": 0.0001999998289645801,
-      "loss": 46.0,
-      "step": 7710
-    },
-    {
-      "epoch": 0.5895597989181337,
-      "grad_norm": 0.0011490164324641228,
-      "learning_rate": 0.00019999982892015243,
-      "loss": 46.0,
-      "step": 7711
-    },
-    {
-      "epoch": 0.5896362559015234,
-      "grad_norm": 0.0015233343001455069,
-      "learning_rate": 0.000199999828875719,
-      "loss": 46.0,
-      "step": 7712
-    },
-    {
-      "epoch": 0.5897127128849131,
-      "grad_norm": 0.013940849341452122,
-      "learning_rate": 0.0001999998288312798,
-      "loss": 46.0,
-      "step": 7713
-    },
-    {
-      "epoch": 0.5897891698683029,
-      "grad_norm": 0.0010767108760774136,
-      "learning_rate": 0.00019999982878683486,
-      "loss": 46.0,
-      "step": 7714
-    },
-    {
-      "epoch": 0.5898656268516925,
-      "grad_norm": 0.0030015925876796246,
-      "learning_rate": 0.00019999982874238412,
-      "loss": 46.0,
-      "step": 7715
-    },
-    {
-      "epoch": 0.5899420838350823,
-      "grad_norm": 0.0034688913729041815,
-      "learning_rate": 0.00019999982869792763,
-      "loss": 46.0,
-      "step": 7716
-    },
-    {
-      "epoch": 0.590018540818472,
-      "grad_norm": 0.00441532488912344,
-      "learning_rate": 0.00019999982865346534,
-      "loss": 46.0,
-      "step": 7717
-    },
-    {
-      "epoch": 0.5900949978018617,
-      "grad_norm": 0.002796633169054985,
-      "learning_rate": 0.00019999982860899727,
-      "loss": 46.0,
-      "step": 7718
-    },
-    {
-      "epoch": 0.5901714547852515,
-      "grad_norm": 0.0012663773959502578,
-      "learning_rate": 0.0001999998285645235,
-      "loss": 46.0,
-      "step": 7719
-    },
-    {
-      "epoch": 0.5902479117686412,
-      "grad_norm": 0.0014051677426323295,
-      "learning_rate": 0.0001999998285200439,
-      "loss": 46.0,
-      "step": 7720
-    },
-    {
-      "epoch": 0.5903243687520309,
-      "grad_norm": 0.0014222945319488645,
-      "learning_rate": 0.00019999982847555855,
-      "loss": 46.0,
-      "step": 7721
-    },
-    {
-      "epoch": 0.5904008257354206,
-      "grad_norm": 0.0015413087094202638,
-      "learning_rate": 0.00019999982843106745,
-      "loss": 46.0,
-      "step": 7722
-    },
-    {
-      "epoch": 0.5904772827188103,
-      "grad_norm": 0.0007205578149296343,
-      "learning_rate": 0.00019999982838657054,
-      "loss": 46.0,
-      "step": 7723
-    },
-    {
-      "epoch": 0.5905537397022,
-      "grad_norm": 0.001248725806362927,
-      "learning_rate": 0.00019999982834206786,
-      "loss": 46.0,
-      "step": 7724
-    },
-    {
-      "epoch": 0.5906301966855898,
-      "grad_norm": 0.0006239841459318995,
-      "learning_rate": 0.00019999982829755947,
-      "loss": 46.0,
-      "step": 7725
-    },
-    {
-      "epoch": 0.5907066536689795,
-      "grad_norm": 0.002972880844026804,
-      "learning_rate": 0.00019999982825304527,
-      "loss": 46.0,
-      "step": 7726
-    },
-    {
-      "epoch": 0.5907831106523692,
-      "grad_norm": 0.0017908732406795025,
-      "learning_rate": 0.00019999982820852527,
-      "loss": 46.0,
-      "step": 7727
-    },
-    {
-      "epoch": 0.590859567635759,
-      "grad_norm": 0.00220440118573606,
-      "learning_rate": 0.00019999982816399956,
-      "loss": 46.0,
-      "step": 7728
-    },
-    {
-      "epoch": 0.5909360246191486,
-      "grad_norm": 0.0013985465047881007,
-      "learning_rate": 0.00019999982811946804,
-      "loss": 46.0,
-      "step": 7729
-    },
-    {
-      "epoch": 0.5910124816025384,
-      "grad_norm": 0.0011062863050028682,
-      "learning_rate": 0.00019999982807493075,
-      "loss": 46.0,
-      "step": 7730
-    },
-    {
-      "epoch": 0.5910889385859281,
-      "grad_norm": 0.007403626572340727,
-      "learning_rate": 0.0001999998280303877,
-      "loss": 46.0,
-      "step": 7731
-    },
-    {
-      "epoch": 0.5911653955693178,
-      "grad_norm": 0.0008812104351818562,
-      "learning_rate": 0.00019999982798583893,
-      "loss": 46.0,
-      "step": 7732
-    },
-    {
-      "epoch": 0.5912418525527076,
-      "grad_norm": 0.0009975922293961048,
-      "learning_rate": 0.00019999982794128435,
-      "loss": 46.0,
-      "step": 7733
-    },
-    {
-      "epoch": 0.5913183095360972,
-      "grad_norm": 0.0015556385042145848,
-      "learning_rate": 0.000199999827896724,
-      "loss": 46.0,
-      "step": 7734
-    },
-    {
-      "epoch": 0.591394766519487,
-      "grad_norm": 0.001604054938070476,
-      "learning_rate": 0.00019999982785215786,
-      "loss": 46.0,
-      "step": 7735
-    },
-    {
-      "epoch": 0.5914712235028767,
-      "grad_norm": 0.0004728210042230785,
-      "learning_rate": 0.00019999982780758595,
-      "loss": 46.0,
-      "step": 7736
-    },
-    {
-      "epoch": 0.5915476804862664,
-      "grad_norm": 0.0004678861587308347,
-      "learning_rate": 0.0001999998277630083,
-      "loss": 46.0,
-      "step": 7737
-    },
-    {
-      "epoch": 0.5916241374696561,
-      "grad_norm": 0.004436808172613382,
-      "learning_rate": 0.00019999982771842488,
-      "loss": 46.0,
-      "step": 7738
-    },
-    {
-      "epoch": 0.5917005944530459,
-      "grad_norm": 0.0026647469494491816,
-      "learning_rate": 0.00019999982767383566,
-      "loss": 46.0,
-      "step": 7739
-    },
-    {
-      "epoch": 0.5917770514364356,
-      "grad_norm": 0.0007762060849927366,
-      "learning_rate": 0.0001999998276292407,
-      "loss": 46.0,
-      "step": 7740
-    },
-    {
-      "epoch": 0.5918535084198253,
-      "grad_norm": 0.00167200923897326,
-      "learning_rate": 0.00019999982758463994,
-      "loss": 46.0,
-      "step": 7741
-    },
-    {
-      "epoch": 0.591929965403215,
-      "grad_norm": 0.0010982464300468564,
-      "learning_rate": 0.00019999982754003343,
-      "loss": 46.0,
-      "step": 7742
-    },
-    {
-      "epoch": 0.5920064223866047,
-      "grad_norm": 0.004163967911154032,
-      "learning_rate": 0.00019999982749542116,
-      "loss": 46.0,
-      "step": 7743
-    },
-    {
-      "epoch": 0.5920828793699945,
-      "grad_norm": 0.0007136771455407143,
-      "learning_rate": 0.00019999982745080313,
-      "loss": 46.0,
-      "step": 7744
-    },
-    {
-      "epoch": 0.5921593363533841,
-      "grad_norm": 0.0030117679852992296,
-      "learning_rate": 0.00019999982740617932,
-      "loss": 46.0,
-      "step": 7745
-    },
-    {
-      "epoch": 0.5922357933367739,
-      "grad_norm": 0.0010822394397109747,
-      "learning_rate": 0.0001999998273615497,
-      "loss": 46.0,
-      "step": 7746
-    },
-    {
-      "epoch": 0.5923122503201637,
-      "grad_norm": 0.002164625097066164,
-      "learning_rate": 0.00019999982731691435,
-      "loss": 46.0,
-      "step": 7747
-    },
-    {
-      "epoch": 0.5923887073035533,
-      "grad_norm": 0.0013500876957550645,
-      "learning_rate": 0.00019999982727227322,
-      "loss": 46.0,
-      "step": 7748
-    },
-    {
-      "epoch": 0.5924651642869431,
-      "grad_norm": 0.0006591337732970715,
-      "learning_rate": 0.00019999982722762634,
-      "loss": 46.0,
-      "step": 7749
-    },
-    {
-      "epoch": 0.5925416212703328,
-      "grad_norm": 0.0062608374282717705,
-      "learning_rate": 0.00019999982718297367,
-      "loss": 46.0,
-      "step": 7750
-    },
-    {
-      "epoch": 0.5926180782537225,
-      "grad_norm": 0.0012927285861223936,
-      "learning_rate": 0.00019999982713831522,
-      "loss": 46.0,
-      "step": 7751
-    },
-    {
-      "epoch": 0.5926945352371122,
-      "grad_norm": 0.0009843326406553388,
-      "learning_rate": 0.00019999982709365105,
-      "loss": 46.0,
-      "step": 7752
-    },
-    {
-      "epoch": 0.5927709922205019,
-      "grad_norm": 0.0004142065590713173,
-      "learning_rate": 0.00019999982704898105,
-      "loss": 46.0,
-      "step": 7753
-    },
-    {
-      "epoch": 0.5928474492038917,
-      "grad_norm": 0.0009537625010125339,
-      "learning_rate": 0.0001999998270043053,
-      "loss": 46.0,
-      "step": 7754
-    },
-    {
-      "epoch": 0.5929239061872814,
-      "grad_norm": 0.0022059367038309574,
-      "learning_rate": 0.0001999998269596238,
-      "loss": 46.0,
-      "step": 7755
-    },
-    {
-      "epoch": 0.5930003631706711,
-      "grad_norm": 0.01708313263952732,
-      "learning_rate": 0.00019999982691493653,
-      "loss": 46.0,
-      "step": 7756
-    },
-    {
-      "epoch": 0.5930768201540608,
-      "grad_norm": 0.0021000804845243692,
-      "learning_rate": 0.00019999982687024347,
-      "loss": 46.0,
-      "step": 7757
-    },
-    {
-      "epoch": 0.5931532771374506,
-      "grad_norm": 0.00047445925883948803,
-      "learning_rate": 0.00019999982682554466,
-      "loss": 46.0,
-      "step": 7758
-    },
-    {
-      "epoch": 0.5932297341208402,
-      "grad_norm": 0.0005947981844656169,
-      "learning_rate": 0.00019999982678084007,
-      "loss": 46.0,
-      "step": 7759
-    },
-    {
-      "epoch": 0.59330619110423,
-      "grad_norm": 0.0010240054689347744,
-      "learning_rate": 0.0001999998267361297,
-      "loss": 46.0,
-      "step": 7760
-    },
-    {
-      "epoch": 0.5933826480876198,
-      "grad_norm": 0.0018377226078882813,
-      "learning_rate": 0.00019999982669141356,
-      "loss": 46.0,
-      "step": 7761
-    },
-    {
-      "epoch": 0.5934591050710094,
-      "grad_norm": 0.0004205318109598011,
-      "learning_rate": 0.00019999982664669168,
-      "loss": 46.0,
-      "step": 7762
-    },
-    {
-      "epoch": 0.5935355620543992,
-      "grad_norm": 0.0016403035260736942,
-      "learning_rate": 0.000199999826601964,
-      "loss": 46.0,
-      "step": 7763
-    },
-    {
-      "epoch": 0.5936120190377888,
-      "grad_norm": 0.010571778751909733,
-      "learning_rate": 0.00019999982655723059,
-      "loss": 46.0,
-      "step": 7764
-    },
-    {
-      "epoch": 0.5936884760211786,
-      "grad_norm": 0.0055676233023405075,
-      "learning_rate": 0.00019999982651249136,
-      "loss": 46.0,
-      "step": 7765
-    },
-    {
-      "epoch": 0.5937649330045683,
-      "grad_norm": 0.0028461567126214504,
-      "learning_rate": 0.00019999982646774637,
-      "loss": 46.0,
-      "step": 7766
-    },
-    {
-      "epoch": 0.593841389987958,
-      "grad_norm": 0.001996994251385331,
-      "learning_rate": 0.00019999982642299565,
-      "loss": 46.0,
-      "step": 7767
-    },
-    {
-      "epoch": 0.5939178469713478,
-      "grad_norm": 0.0009702043025754392,
-      "learning_rate": 0.00019999982637823913,
-      "loss": 46.0,
-      "step": 7768
-    },
-    {
-      "epoch": 0.5939943039547375,
-      "grad_norm": 0.004336015321314335,
-      "learning_rate": 0.00019999982633347684,
-      "loss": 46.0,
-      "step": 7769
-    },
-    {
-      "epoch": 0.5940707609381272,
-      "grad_norm": 0.003671329701319337,
-      "learning_rate": 0.00019999982628870878,
-      "loss": 46.0,
-      "step": 7770
-    },
-    {
-      "epoch": 0.5941472179215169,
-      "grad_norm": 0.0006680940859951079,
-      "learning_rate": 0.00019999982624393497,
-      "loss": 46.0,
-      "step": 7771
-    },
-    {
-      "epoch": 0.5942236749049066,
-      "grad_norm": 0.001191614312119782,
-      "learning_rate": 0.00019999982619915536,
-      "loss": 46.0,
-      "step": 7772
-    },
-    {
-      "epoch": 0.5943001318882963,
-      "grad_norm": 0.0011037515942007303,
-      "learning_rate": 0.00019999982615437003,
-      "loss": 46.0,
-      "step": 7773
-    },
-    {
-      "epoch": 0.5943765888716861,
-      "grad_norm": 0.002332849195227027,
-      "learning_rate": 0.00019999982610957888,
-      "loss": 46.0,
-      "step": 7774
-    },
-    {
-      "epoch": 0.5944530458550757,
-      "grad_norm": 0.004200548864901066,
-      "learning_rate": 0.00019999982606478197,
-      "loss": 46.0,
-      "step": 7775
-    },
-    {
-      "epoch": 0.5945295028384655,
-      "grad_norm": 0.0016789293149486184,
-      "learning_rate": 0.0001999998260199793,
-      "loss": 46.0,
-      "step": 7776
-    },
-    {
-      "epoch": 0.5946059598218553,
-      "grad_norm": 0.002002288121730089,
-      "learning_rate": 0.00019999982597517088,
-      "loss": 46.0,
-      "step": 7777
-    },
-    {
-      "epoch": 0.5946824168052449,
-      "grad_norm": 0.0017278622835874557,
-      "learning_rate": 0.00019999982593035665,
-      "loss": 46.0,
-      "step": 7778
-    },
-    {
-      "epoch": 0.5947588737886347,
-      "grad_norm": 0.000411067099776119,
-      "learning_rate": 0.00019999982588553668,
-      "loss": 46.0,
-      "step": 7779
-    },
-    {
-      "epoch": 0.5948353307720244,
-      "grad_norm": 0.0026848837733268738,
-      "learning_rate": 0.00019999982584071094,
-      "loss": 46.0,
-      "step": 7780
-    },
-    {
-      "epoch": 0.5949117877554141,
-      "grad_norm": 0.0009578131139278412,
-      "learning_rate": 0.00019999982579587943,
-      "loss": 46.0,
-      "step": 7781
-    },
-    {
-      "epoch": 0.5949882447388039,
-      "grad_norm": 0.0007319899741560221,
-      "learning_rate": 0.00019999982575104214,
-      "loss": 46.0,
-      "step": 7782
-    },
-    {
-      "epoch": 0.5950647017221935,
-      "grad_norm": 0.0005809739814139903,
-      "learning_rate": 0.00019999982570619907,
-      "loss": 46.0,
-      "step": 7783
-    },
-    {
-      "epoch": 0.5951411587055833,
-      "grad_norm": 0.0007317565032280982,
-      "learning_rate": 0.00019999982566135024,
-      "loss": 46.0,
-      "step": 7784
-    },
-    {
-      "epoch": 0.595217615688973,
-      "grad_norm": 0.0006904666079208255,
-      "learning_rate": 0.00019999982561649566,
-      "loss": 46.0,
-      "step": 7785
-    },
-    {
-      "epoch": 0.5952940726723627,
-      "grad_norm": 0.0070307450369000435,
-      "learning_rate": 0.00019999982557163527,
-      "loss": 46.0,
-      "step": 7786
-    },
-    {
-      "epoch": 0.5953705296557524,
-      "grad_norm": 0.0012231232831254601,
-      "learning_rate": 0.00019999982552676917,
-      "loss": 46.0,
-      "step": 7787
-    },
-    {
-      "epoch": 0.5954469866391422,
-      "grad_norm": 0.0014784516533836722,
-      "learning_rate": 0.00019999982548189724,
-      "loss": 46.0,
-      "step": 7788
-    },
-    {
-      "epoch": 0.5955234436225318,
-      "grad_norm": 0.002267831237986684,
-      "learning_rate": 0.0001999998254370196,
-      "loss": 46.0,
-      "step": 7789
-    },
-    {
-      "epoch": 0.5955999006059216,
-      "grad_norm": 0.0007285760948434472,
-      "learning_rate": 0.00019999982539213614,
-      "loss": 46.0,
-      "step": 7790
-    },
-    {
-      "epoch": 0.5956763575893114,
-      "grad_norm": 0.0023656245321035385,
-      "learning_rate": 0.00019999982534724692,
-      "loss": 46.0,
-      "step": 7791
-    },
-    {
-      "epoch": 0.595752814572701,
-      "grad_norm": 0.0031207932624965906,
-      "learning_rate": 0.00019999982530235192,
-      "loss": 46.0,
-      "step": 7792
-    },
-    {
-      "epoch": 0.5958292715560908,
-      "grad_norm": 0.001985993469133973,
-      "learning_rate": 0.00019999982525745118,
-      "loss": 46.0,
-      "step": 7793
-    },
-    {
-      "epoch": 0.5959057285394804,
-      "grad_norm": 0.005390759091824293,
-      "learning_rate": 0.00019999982521254467,
-      "loss": 46.0,
-      "step": 7794
-    },
-    {
-      "epoch": 0.5959821855228702,
-      "grad_norm": 0.00456582847982645,
-      "learning_rate": 0.00019999982516763238,
-      "loss": 46.0,
-      "step": 7795
-    },
-    {
-      "epoch": 0.59605864250626,
-      "grad_norm": 0.0006566466181538999,
-      "learning_rate": 0.00019999982512271434,
-      "loss": 46.0,
-      "step": 7796
-    },
-    {
-      "epoch": 0.5961350994896496,
-      "grad_norm": 0.0006910919910296798,
-      "learning_rate": 0.00019999982507779048,
-      "loss": 46.0,
-      "step": 7797
-    },
-    {
-      "epoch": 0.5962115564730394,
-      "grad_norm": 0.0011979263508692384,
-      "learning_rate": 0.0001999998250328609,
-      "loss": 46.0,
-      "step": 7798
-    },
-    {
-      "epoch": 0.5962880134564291,
-      "grad_norm": 0.0018503390019759536,
-      "learning_rate": 0.00019999982498792554,
-      "loss": 46.0,
-      "step": 7799
-    },
-    {
-      "epoch": 0.5963644704398188,
-      "grad_norm": 0.0005962568102404475,
-      "learning_rate": 0.0001999998249429844,
-      "loss": 46.0,
-      "step": 7800
-    },
-    {
-      "epoch": 0.5964409274232085,
-      "grad_norm": 0.001471091527491808,
-      "learning_rate": 0.00019999982489803748,
-      "loss": 46.0,
-      "step": 7801
-    },
-    {
-      "epoch": 0.5965173844065982,
-      "grad_norm": 0.0025981185026466846,
-      "learning_rate": 0.0001999998248530848,
-      "loss": 46.0,
-      "step": 7802
-    },
-    {
-      "epoch": 0.5965938413899879,
-      "grad_norm": 0.0013325948966667056,
-      "learning_rate": 0.0001999998248081264,
-      "loss": 46.0,
-      "step": 7803
-    },
-    {
-      "epoch": 0.5966702983733777,
-      "grad_norm": 0.0005899730604141951,
-      "learning_rate": 0.00019999982476316214,
-      "loss": 46.0,
-      "step": 7804
-    },
-    {
-      "epoch": 0.5967467553567674,
-      "grad_norm": 0.0016896016895771027,
-      "learning_rate": 0.00019999982471819217,
-      "loss": 46.0,
-      "step": 7805
-    },
-    {
-      "epoch": 0.5968232123401571,
-      "grad_norm": 0.0009710498852655292,
-      "learning_rate": 0.0001999998246732164,
-      "loss": 46.0,
-      "step": 7806
-    },
-    {
-      "epoch": 0.5968996693235469,
-      "grad_norm": 0.006422850769013166,
-      "learning_rate": 0.0001999998246282349,
-      "loss": 46.0,
-      "step": 7807
-    },
-    {
-      "epoch": 0.5969761263069365,
-      "grad_norm": 0.004061803687363863,
-      "learning_rate": 0.0001999998245832476,
-      "loss": 46.0,
-      "step": 7808
-    },
-    {
-      "epoch": 0.5970525832903263,
-      "grad_norm": 0.0005329784471541643,
-      "learning_rate": 0.0001999998245382545,
-      "loss": 46.0,
-      "step": 7809
-    },
-    {
-      "epoch": 0.597129040273716,
-      "grad_norm": 0.001856053713709116,
-      "learning_rate": 0.00019999982449325573,
-      "loss": 46.0,
-      "step": 7810
-    },
-    {
-      "epoch": 0.5972054972571057,
-      "grad_norm": 0.0009692359599284828,
-      "learning_rate": 0.0001999998244482511,
-      "loss": 46.0,
-      "step": 7811
-    },
-    {
-      "epoch": 0.5972819542404955,
-      "grad_norm": 0.0038092106115072966,
-      "learning_rate": 0.00019999982440324074,
-      "loss": 46.0,
-      "step": 7812
-    },
-    {
-      "epoch": 0.5973584112238851,
-      "grad_norm": 0.00114718871191144,
-      "learning_rate": 0.00019999982435822458,
-      "loss": 46.0,
-      "step": 7813
-    },
-    {
-      "epoch": 0.5974348682072749,
-      "grad_norm": 0.0014407079434022307,
-      "learning_rate": 0.00019999982431320268,
-      "loss": 46.0,
-      "step": 7814
-    },
-    {
-      "epoch": 0.5975113251906646,
-      "grad_norm": 0.000780620553996414,
-      "learning_rate": 0.000199999824268175,
-      "loss": 46.0,
-      "step": 7815
-    },
-    {
-      "epoch": 0.5975877821740543,
-      "grad_norm": 0.0029868101701140404,
-      "learning_rate": 0.00019999982422314155,
-      "loss": 46.0,
-      "step": 7816
-    },
-    {
-      "epoch": 0.597664239157444,
-      "grad_norm": 0.0012581741902977228,
-      "learning_rate": 0.00019999982417810236,
-      "loss": 46.0,
-      "step": 7817
-    },
-    {
-      "epoch": 0.5977406961408338,
-      "grad_norm": 0.002613708609715104,
-      "learning_rate": 0.00019999982413305736,
-      "loss": 46.0,
-      "step": 7818
-    },
-    {
-      "epoch": 0.5978171531242235,
-      "grad_norm": 0.00441998615860939,
-      "learning_rate": 0.0001999998240880066,
-      "loss": 46.0,
-      "step": 7819
-    },
-    {
-      "epoch": 0.5978936101076132,
-      "grad_norm": 0.0007168451556935906,
-      "learning_rate": 0.00019999982404295005,
-      "loss": 46.0,
-      "step": 7820
-    },
-    {
-      "epoch": 0.597970067091003,
-      "grad_norm": 0.0018303974065929651,
-      "learning_rate": 0.00019999982399788777,
-      "loss": 46.0,
-      "step": 7821
-    },
-    {
-      "epoch": 0.5980465240743926,
-      "grad_norm": 0.006136002950370312,
-      "learning_rate": 0.0001999998239528197,
-      "loss": 46.0,
-      "step": 7822
-    },
-    {
-      "epoch": 0.5981229810577824,
-      "grad_norm": 0.0034344345331192017,
-      "learning_rate": 0.00019999982390774587,
-      "loss": 46.0,
-      "step": 7823
-    },
-    {
-      "epoch": 0.598199438041172,
-      "grad_norm": 0.001203719642944634,
-      "learning_rate": 0.00019999982386266626,
-      "loss": 46.0,
-      "step": 7824
-    },
-    {
-      "epoch": 0.5982758950245618,
-      "grad_norm": 0.0009653617744334042,
-      "learning_rate": 0.0001999998238175809,
-      "loss": 46.0,
-      "step": 7825
-    },
-    {
-      "epoch": 0.5983523520079516,
-      "grad_norm": 0.0034613104071468115,
-      "learning_rate": 0.00019999982377248975,
-      "loss": 46.0,
-      "step": 7826
-    },
-    {
-      "epoch": 0.5984288089913412,
-      "grad_norm": 0.0007446022354997694,
-      "learning_rate": 0.00019999982372739282,
-      "loss": 46.0,
-      "step": 7827
-    },
-    {
-      "epoch": 0.598505265974731,
-      "grad_norm": 0.0010518801864236593,
-      "learning_rate": 0.00019999982368229015,
-      "loss": 46.0,
-      "step": 7828
-    },
-    {
-      "epoch": 0.5985817229581207,
-      "grad_norm": 0.0018843668513000011,
-      "learning_rate": 0.0001999998236371817,
-      "loss": 46.0,
-      "step": 7829
-    },
-    {
-      "epoch": 0.5986581799415104,
-      "grad_norm": 0.0016771783120930195,
-      "learning_rate": 0.00019999982359206748,
-      "loss": 46.0,
-      "step": 7830
-    },
-    {
-      "epoch": 0.5987346369249001,
-      "grad_norm": 0.0005323274526745081,
-      "learning_rate": 0.00019999982354694748,
-      "loss": 46.0,
-      "step": 7831
-    },
-    {
-      "epoch": 0.5988110939082898,
-      "grad_norm": 0.0016021371120586991,
-      "learning_rate": 0.00019999982350182174,
-      "loss": 46.0,
-      "step": 7832
-    },
-    {
-      "epoch": 0.5988875508916796,
-      "grad_norm": 0.003104321425780654,
-      "learning_rate": 0.00019999982345669023,
-      "loss": 46.0,
-      "step": 7833
-    },
-    {
-      "epoch": 0.5989640078750693,
-      "grad_norm": 0.002220702823251486,
-      "learning_rate": 0.0001999998234115529,
-      "loss": 46.0,
-      "step": 7834
-    },
-    {
-      "epoch": 0.599040464858459,
-      "grad_norm": 0.0008930038893595338,
-      "learning_rate": 0.00019999982336640985,
-      "loss": 46.0,
-      "step": 7835
-    },
-    {
-      "epoch": 0.5991169218418487,
-      "grad_norm": 0.0012662962544709444,
-      "learning_rate": 0.000199999823321261,
-      "loss": 46.0,
-      "step": 7836
-    },
-    {
-      "epoch": 0.5991933788252385,
-      "grad_norm": 0.008340027183294296,
-      "learning_rate": 0.0001999998232761064,
-      "loss": 46.0,
-      "step": 7837
-    },
-    {
-      "epoch": 0.5992698358086281,
-      "grad_norm": 0.0023606475442647934,
-      "learning_rate": 0.00019999982323094603,
-      "loss": 46.0,
-      "step": 7838
-    },
-    {
-      "epoch": 0.5993462927920179,
-      "grad_norm": 0.003128261771053076,
-      "learning_rate": 0.00019999982318577987,
-      "loss": 46.0,
-      "step": 7839
-    },
-    {
-      "epoch": 0.5994227497754077,
-      "grad_norm": 0.0008913141791708767,
-      "learning_rate": 0.00019999982314060797,
-      "loss": 46.0,
-      "step": 7840
-    },
-    {
-      "epoch": 0.5994992067587973,
-      "grad_norm": 0.0008210894884541631,
-      "learning_rate": 0.0001999998230954303,
-      "loss": 46.0,
-      "step": 7841
-    },
-    {
-      "epoch": 0.5995756637421871,
-      "grad_norm": 0.004353249911218882,
-      "learning_rate": 0.00019999982305024682,
-      "loss": 46.0,
-      "step": 7842
-    },
-    {
-      "epoch": 0.5996521207255767,
-      "grad_norm": 0.0010502181248739362,
-      "learning_rate": 0.00019999982300505763,
-      "loss": 46.0,
-      "step": 7843
-    },
-    {
-      "epoch": 0.5997285777089665,
-      "grad_norm": 0.004726924002170563,
-      "learning_rate": 0.0001999998229598626,
-      "loss": 46.0,
-      "step": 7844
-    },
-    {
-      "epoch": 0.5998050346923562,
-      "grad_norm": 0.0016735497629269958,
-      "learning_rate": 0.00019999982291466184,
-      "loss": 46.0,
-      "step": 7845
-    },
-    {
-      "epoch": 0.5998814916757459,
-      "grad_norm": 0.0004080400103703141,
-      "learning_rate": 0.00019999982286945532,
-      "loss": 46.0,
-      "step": 7846
-    },
-    {
-      "epoch": 0.5999579486591357,
-      "grad_norm": 0.002207372570410371,
-      "learning_rate": 0.000199999822824243,
-      "loss": 46.0,
-      "step": 7847
-    },
-    {
-      "epoch": 0.6000344056425254,
-      "grad_norm": 0.004215470980852842,
-      "learning_rate": 0.00019999982277902495,
-      "loss": 46.0,
-      "step": 7848
-    },
-    {
-      "epoch": 0.6001108626259151,
-      "grad_norm": 0.000501008820720017,
-      "learning_rate": 0.0001999998227338011,
-      "loss": 46.0,
-      "step": 7849
-    },
-    {
-      "epoch": 0.6001873196093048,
-      "grad_norm": 0.0012892954982817173,
-      "learning_rate": 0.0001999998226885715,
-      "loss": 46.0,
-      "step": 7850
-    },
-    {
-      "epoch": 0.6002637765926946,
-      "grad_norm": 0.00043995975283905864,
-      "learning_rate": 0.00019999982264333612,
-      "loss": 46.0,
-      "step": 7851
-    },
-    {
-      "epoch": 0.6003402335760842,
-      "grad_norm": 0.0005717137246392667,
-      "learning_rate": 0.00019999982259809497,
-      "loss": 46.0,
-      "step": 7852
-    },
-    {
-      "epoch": 0.600416690559474,
-      "grad_norm": 0.0047810813412070274,
-      "learning_rate": 0.00019999982255284807,
-      "loss": 46.0,
-      "step": 7853
-    },
-    {
-      "epoch": 0.6004931475428636,
-      "grad_norm": 0.0008196977432817221,
-      "learning_rate": 0.00019999982250759537,
-      "loss": 46.0,
-      "step": 7854
-    },
-    {
-      "epoch": 0.6005696045262534,
-      "grad_norm": 0.0018608559621497989,
-      "learning_rate": 0.00019999982246233692,
-      "loss": 46.0,
-      "step": 7855
-    },
-    {
-      "epoch": 0.6006460615096432,
-      "grad_norm": 0.0017331335693597794,
-      "learning_rate": 0.0001999998224170727,
-      "loss": 46.0,
-      "step": 7856
-    },
-    {
-      "epoch": 0.6007225184930328,
-      "grad_norm": 0.0011089193867519498,
-      "learning_rate": 0.0001999998223718027,
-      "loss": 46.0,
-      "step": 7857
-    },
-    {
-      "epoch": 0.6007989754764226,
-      "grad_norm": 0.0006854445673525333,
-      "learning_rate": 0.0001999998223265269,
-      "loss": 46.0,
-      "step": 7858
-    },
-    {
-      "epoch": 0.6008754324598123,
-      "grad_norm": 0.0005723850918002427,
-      "learning_rate": 0.00019999982228124542,
-      "loss": 46.0,
-      "step": 7859
-    },
-    {
-      "epoch": 0.600951889443202,
-      "grad_norm": 0.0025426107458770275,
-      "learning_rate": 0.00019999982223595808,
-      "loss": 46.0,
-      "step": 7860
-    },
-    {
-      "epoch": 0.6010283464265918,
-      "grad_norm": 0.004472413565963507,
-      "learning_rate": 0.00019999982219066502,
-      "loss": 46.0,
-      "step": 7861
-    },
-    {
-      "epoch": 0.6011048034099815,
-      "grad_norm": 0.0015187521930783987,
-      "learning_rate": 0.00019999982214536618,
-      "loss": 46.0,
-      "step": 7862
-    },
-    {
-      "epoch": 0.6011812603933712,
-      "grad_norm": 0.0010940533829852939,
-      "learning_rate": 0.00019999982210006155,
-      "loss": 46.0,
-      "step": 7863
-    },
-    {
-      "epoch": 0.6012577173767609,
-      "grad_norm": 0.0012296380009502172,
-      "learning_rate": 0.0001999998220547512,
-      "loss": 46.0,
-      "step": 7864
-    },
-    {
-      "epoch": 0.6013341743601506,
-      "grad_norm": 0.004293881822377443,
-      "learning_rate": 0.00019999982200943504,
-      "loss": 46.0,
-      "step": 7865
-    },
-    {
-      "epoch": 0.6014106313435403,
-      "grad_norm": 0.0005095797823742032,
-      "learning_rate": 0.00019999982196411311,
-      "loss": 46.0,
-      "step": 7866
-    },
-    {
-      "epoch": 0.6014870883269301,
-      "grad_norm": 0.0035696239210665226,
-      "learning_rate": 0.00019999982191878544,
-      "loss": 46.0,
-      "step": 7867
-    },
-    {
-      "epoch": 0.6015635453103197,
-      "grad_norm": 0.0011099465191364288,
-      "learning_rate": 0.00019999982187345197,
-      "loss": 46.0,
-      "step": 7868
-    },
-    {
-      "epoch": 0.6016400022937095,
-      "grad_norm": 0.001394714810885489,
-      "learning_rate": 0.00019999982182811272,
-      "loss": 46.0,
-      "step": 7869
-    },
-    {
-      "epoch": 0.6017164592770993,
-      "grad_norm": 0.0007722071604803205,
-      "learning_rate": 0.00019999982178276775,
-      "loss": 46.0,
-      "step": 7870
-    },
-    {
-      "epoch": 0.6017929162604889,
-      "grad_norm": 0.005232236813753843,
-      "learning_rate": 0.00019999982173741696,
-      "loss": 46.0,
-      "step": 7871
-    },
-    {
-      "epoch": 0.6018693732438787,
-      "grad_norm": 0.005517760291695595,
-      "learning_rate": 0.00019999982169206044,
-      "loss": 46.0,
-      "step": 7872
-    },
-    {
-      "epoch": 0.6019458302272683,
-      "grad_norm": 0.0018902813317254186,
-      "learning_rate": 0.00019999982164669813,
-      "loss": 46.0,
-      "step": 7873
-    },
-    {
-      "epoch": 0.6020222872106581,
-      "grad_norm": 0.0006760531687177718,
-      "learning_rate": 0.00019999982160133004,
-      "loss": 46.0,
-      "step": 7874
-    },
-    {
-      "epoch": 0.6020987441940479,
-      "grad_norm": 0.0036640206817537546,
-      "learning_rate": 0.0001999998215559562,
-      "loss": 46.0,
-      "step": 7875
-    },
-    {
-      "epoch": 0.6021752011774375,
-      "grad_norm": 0.0014028297737240791,
-      "learning_rate": 0.0001999998215105766,
-      "loss": 46.0,
-      "step": 7876
-    },
-    {
-      "epoch": 0.6022516581608273,
-      "grad_norm": 0.0012549128150567412,
-      "learning_rate": 0.00019999982146519122,
-      "loss": 46.0,
-      "step": 7877
-    },
-    {
-      "epoch": 0.602328115144217,
-      "grad_norm": 0.0011920961551368237,
-      "learning_rate": 0.00019999982141980004,
-      "loss": 46.0,
-      "step": 7878
-    },
-    {
-      "epoch": 0.6024045721276067,
-      "grad_norm": 0.0015891494695097208,
-      "learning_rate": 0.0001999998213744031,
-      "loss": 46.0,
-      "step": 7879
-    },
-    {
-      "epoch": 0.6024810291109964,
-      "grad_norm": 0.004706768784672022,
-      "learning_rate": 0.00019999982132900044,
-      "loss": 46.0,
-      "step": 7880
-    },
-    {
-      "epoch": 0.6025574860943862,
-      "grad_norm": 0.001879567396827042,
-      "learning_rate": 0.000199999821283592,
-      "loss": 46.0,
-      "step": 7881
-    },
-    {
-      "epoch": 0.6026339430777758,
-      "grad_norm": 0.0005421018577180803,
-      "learning_rate": 0.00019999982123817774,
-      "loss": 46.0,
-      "step": 7882
-    },
-    {
-      "epoch": 0.6027104000611656,
-      "grad_norm": 0.001354339998215437,
-      "learning_rate": 0.00019999982119275775,
-      "loss": 46.0,
-      "step": 7883
-    },
-    {
-      "epoch": 0.6027868570445553,
-      "grad_norm": 0.0010297372937202454,
-      "learning_rate": 0.00019999982114733196,
-      "loss": 46.0,
-      "step": 7884
-    },
-    {
-      "epoch": 0.602863314027945,
-      "grad_norm": 0.0020425557158887386,
-      "learning_rate": 0.00019999982110190044,
-      "loss": 46.0,
-      "step": 7885
-    },
-    {
-      "epoch": 0.6029397710113348,
-      "grad_norm": 0.0009307325235567987,
-      "learning_rate": 0.00019999982105646313,
-      "loss": 46.0,
-      "step": 7886
-    },
-    {
-      "epoch": 0.6030162279947244,
-      "grad_norm": 0.0024466970935463905,
-      "learning_rate": 0.00019999982101102004,
-      "loss": 46.0,
-      "step": 7887
-    },
-    {
-      "epoch": 0.6030926849781142,
-      "grad_norm": 0.002209039404988289,
-      "learning_rate": 0.0001999998209655712,
-      "loss": 46.0,
-      "step": 7888
-    },
-    {
-      "epoch": 0.603169141961504,
-      "grad_norm": 0.009476684033870697,
-      "learning_rate": 0.00019999982092011657,
-      "loss": 46.0,
-      "step": 7889
-    },
-    {
-      "epoch": 0.6032455989448936,
-      "grad_norm": 0.0020331223495304585,
-      "learning_rate": 0.0001999998208746562,
-      "loss": 46.0,
-      "step": 7890
-    },
-    {
-      "epoch": 0.6033220559282834,
-      "grad_norm": 0.0011414854088798165,
-      "learning_rate": 0.00019999982082919004,
-      "loss": 46.0,
-      "step": 7891
-    },
-    {
-      "epoch": 0.6033985129116731,
-      "grad_norm": 0.003357579465955496,
-      "learning_rate": 0.0001999998207837181,
-      "loss": 46.0,
-      "step": 7892
-    },
-    {
-      "epoch": 0.6034749698950628,
-      "grad_norm": 0.0023027893621474504,
-      "learning_rate": 0.00019999982073824038,
-      "loss": 46.0,
-      "step": 7893
-    },
-    {
-      "epoch": 0.6035514268784525,
-      "grad_norm": 0.0022761691361665726,
-      "learning_rate": 0.00019999982069275697,
-      "loss": 46.0,
-      "step": 7894
-    },
-    {
-      "epoch": 0.6036278838618422,
-      "grad_norm": 0.0017769032856449485,
-      "learning_rate": 0.0001999998206472677,
-      "loss": 46.0,
-      "step": 7895
-    },
-    {
-      "epoch": 0.603704340845232,
-      "grad_norm": 0.0011424003168940544,
-      "learning_rate": 0.00019999982060177272,
-      "loss": 46.0,
-      "step": 7896
-    },
-    {
-      "epoch": 0.6037807978286217,
-      "grad_norm": 0.001006170641630888,
-      "learning_rate": 0.00019999982055627193,
-      "loss": 46.0,
-      "step": 7897
-    },
-    {
-      "epoch": 0.6038572548120114,
-      "grad_norm": 0.0008326191455125809,
-      "learning_rate": 0.0001999998205107654,
-      "loss": 46.0,
-      "step": 7898
-    },
-    {
-      "epoch": 0.6039337117954011,
-      "grad_norm": 0.0012525704223662615,
-      "learning_rate": 0.00019999982046525308,
-      "loss": 46.0,
-      "step": 7899
-    },
-    {
-      "epoch": 0.6040101687787909,
-      "grad_norm": 0.0011818106286227703,
-      "learning_rate": 0.000199999820419735,
-      "loss": 46.0,
-      "step": 7900
-    },
-    {
-      "epoch": 0.6040866257621805,
-      "grad_norm": 0.003493251046165824,
-      "learning_rate": 0.00019999982037421116,
-      "loss": 46.0,
-      "step": 7901
-    },
-    {
-      "epoch": 0.6041630827455703,
-      "grad_norm": 0.002517836634069681,
-      "learning_rate": 0.00019999982032868152,
-      "loss": 46.0,
-      "step": 7902
-    },
-    {
-      "epoch": 0.6042395397289599,
-      "grad_norm": 0.0043289246968925,
-      "learning_rate": 0.00019999982028314614,
-      "loss": 46.0,
-      "step": 7903
-    },
-    {
-      "epoch": 0.6043159967123497,
-      "grad_norm": 0.0020103594288229942,
-      "learning_rate": 0.000199999820237605,
-      "loss": 46.0,
-      "step": 7904
-    },
-    {
-      "epoch": 0.6043924536957395,
-      "grad_norm": 0.0015729309525340796,
-      "learning_rate": 0.00019999982019205806,
-      "loss": 46.0,
-      "step": 7905
-    },
-    {
-      "epoch": 0.6044689106791291,
-      "grad_norm": 0.0006856784457340837,
-      "learning_rate": 0.00019999982014650536,
-      "loss": 46.0,
-      "step": 7906
-    },
-    {
-      "epoch": 0.6045453676625189,
-      "grad_norm": 0.0011072318302467465,
-      "learning_rate": 0.0001999998201009469,
-      "loss": 46.0,
-      "step": 7907
-    },
-    {
-      "epoch": 0.6046218246459086,
-      "grad_norm": 0.000863327644765377,
-      "learning_rate": 0.00019999982005538264,
-      "loss": 46.0,
-      "step": 7908
-    },
-    {
-      "epoch": 0.6046982816292983,
-      "grad_norm": 0.0004334252153057605,
-      "learning_rate": 0.00019999982000981262,
-      "loss": 46.0,
-      "step": 7909
-    },
-    {
-      "epoch": 0.604774738612688,
-      "grad_norm": 0.002265630755573511,
-      "learning_rate": 0.00019999981996423686,
-      "loss": 46.0,
-      "step": 7910
-    },
-    {
-      "epoch": 0.6048511955960778,
-      "grad_norm": 0.0028832312673330307,
-      "learning_rate": 0.00019999981991865532,
-      "loss": 46.0,
-      "step": 7911
-    },
-    {
-      "epoch": 0.6049276525794675,
-      "grad_norm": 0.0005825895350426435,
-      "learning_rate": 0.00019999981987306803,
-      "loss": 46.0,
-      "step": 7912
-    },
-    {
-      "epoch": 0.6050041095628572,
-      "grad_norm": 0.002411558758467436,
-      "learning_rate": 0.00019999981982747494,
-      "loss": 46.0,
-      "step": 7913
-    },
-    {
-      "epoch": 0.6050805665462469,
-      "grad_norm": 0.004140479490160942,
-      "learning_rate": 0.00019999981978187608,
-      "loss": 46.0,
-      "step": 7914
-    },
-    {
-      "epoch": 0.6051570235296366,
-      "grad_norm": 0.000905551656614989,
-      "learning_rate": 0.00019999981973627145,
-      "loss": 46.0,
-      "step": 7915
-    },
-    {
-      "epoch": 0.6052334805130264,
-      "grad_norm": 0.0014777554897591472,
-      "learning_rate": 0.00019999981969066104,
-      "loss": 46.0,
-      "step": 7916
-    },
-    {
-      "epoch": 0.605309937496416,
-      "grad_norm": 0.0005173294921405613,
-      "learning_rate": 0.00019999981964504492,
-      "loss": 46.0,
-      "step": 7917
-    },
-    {
-      "epoch": 0.6053863944798058,
-      "grad_norm": 0.000806057418230921,
-      "learning_rate": 0.00019999981959942296,
-      "loss": 46.0,
-      "step": 7918
-    },
-    {
-      "epoch": 0.6054628514631956,
-      "grad_norm": 0.0005943129071965814,
-      "learning_rate": 0.00019999981955379526,
-      "loss": 46.0,
-      "step": 7919
-    },
-    {
-      "epoch": 0.6055393084465852,
-      "grad_norm": 0.0008464242564514279,
-      "learning_rate": 0.0001999998195081618,
-      "loss": 46.0,
-      "step": 7920
-    },
-    {
-      "epoch": 0.605615765429975,
-      "grad_norm": 0.00390784302726388,
-      "learning_rate": 0.00019999981946252255,
-      "loss": 46.0,
-      "step": 7921
-    },
-    {
-      "epoch": 0.6056922224133647,
-      "grad_norm": 0.0017563611036166549,
-      "learning_rate": 0.00019999981941687755,
-      "loss": 46.0,
-      "step": 7922
-    },
-    {
-      "epoch": 0.6057686793967544,
-      "grad_norm": 0.0008239975431933999,
-      "learning_rate": 0.00019999981937122676,
-      "loss": 46.0,
-      "step": 7923
-    },
-    {
-      "epoch": 0.6058451363801441,
-      "grad_norm": 0.0017751911655068398,
-      "learning_rate": 0.00019999981932557022,
-      "loss": 46.0,
-      "step": 7924
-    },
-    {
-      "epoch": 0.6059215933635338,
-      "grad_norm": 0.0009364386787638068,
-      "learning_rate": 0.0001999998192799079,
-      "loss": 46.0,
-      "step": 7925
-    },
-    {
-      "epoch": 0.6059980503469236,
-      "grad_norm": 0.0024566175416111946,
-      "learning_rate": 0.0001999998192342398,
-      "loss": 46.0,
-      "step": 7926
-    },
-    {
-      "epoch": 0.6060745073303133,
-      "grad_norm": 0.00040578775224275887,
-      "learning_rate": 0.00019999981918856596,
-      "loss": 46.0,
-      "step": 7927
-    },
-    {
-      "epoch": 0.606150964313703,
-      "grad_norm": 0.003301683347672224,
-      "learning_rate": 0.00019999981914288633,
-      "loss": 46.0,
-      "step": 7928
-    },
-    {
-      "epoch": 0.6062274212970927,
-      "grad_norm": 0.0013802727917209268,
-      "learning_rate": 0.00019999981909720095,
-      "loss": 46.0,
-      "step": 7929
-    },
-    {
-      "epoch": 0.6063038782804825,
-      "grad_norm": 0.0008085070294328034,
-      "learning_rate": 0.00019999981905150977,
-      "loss": 46.0,
-      "step": 7930
-    },
-    {
-      "epoch": 0.6063803352638721,
-      "grad_norm": 0.0013770516961812973,
-      "learning_rate": 0.00019999981900581284,
-      "loss": 46.0,
-      "step": 7931
-    },
-    {
-      "epoch": 0.6064567922472619,
-      "grad_norm": 0.0006124796345829964,
-      "learning_rate": 0.00019999981896011014,
-      "loss": 46.0,
-      "step": 7932
-    },
-    {
-      "epoch": 0.6065332492306515,
-      "grad_norm": 0.002153716515749693,
-      "learning_rate": 0.00019999981891440167,
-      "loss": 46.0,
-      "step": 7933
-    },
-    {
-      "epoch": 0.6066097062140413,
-      "grad_norm": 0.0024859863333404064,
-      "learning_rate": 0.00019999981886868742,
-      "loss": 46.0,
-      "step": 7934
-    },
-    {
-      "epoch": 0.6066861631974311,
-      "grad_norm": 0.0010924540692940354,
-      "learning_rate": 0.00019999981882296738,
-      "loss": 46.0,
-      "step": 7935
-    },
-    {
-      "epoch": 0.6067626201808207,
-      "grad_norm": 0.0022507929243147373,
-      "learning_rate": 0.0001999998187772416,
-      "loss": 46.0,
-      "step": 7936
-    },
-    {
-      "epoch": 0.6068390771642105,
-      "grad_norm": 0.0013681759592145681,
-      "learning_rate": 0.00019999981873151005,
-      "loss": 46.0,
-      "step": 7937
-    },
-    {
-      "epoch": 0.6069155341476002,
-      "grad_norm": 0.0009938521543517709,
-      "learning_rate": 0.00019999981868577276,
-      "loss": 46.0,
-      "step": 7938
-    },
-    {
-      "epoch": 0.6069919911309899,
-      "grad_norm": 0.0006067179492674768,
-      "learning_rate": 0.00019999981864002965,
-      "loss": 46.0,
-      "step": 7939
-    },
-    {
-      "epoch": 0.6070684481143797,
-      "grad_norm": 0.0008704601787030697,
-      "learning_rate": 0.00019999981859428081,
-      "loss": 46.0,
-      "step": 7940
-    },
-    {
-      "epoch": 0.6071449050977694,
-      "grad_norm": 0.0016925816889852285,
-      "learning_rate": 0.00019999981854852615,
-      "loss": 46.0,
-      "step": 7941
-    },
-    {
-      "epoch": 0.6072213620811591,
-      "grad_norm": 0.000867995957378298,
-      "learning_rate": 0.0001999998185027658,
-      "loss": 46.0,
-      "step": 7942
-    },
-    {
-      "epoch": 0.6072978190645488,
-      "grad_norm": 0.000922888983041048,
-      "learning_rate": 0.0001999998184569996,
-      "loss": 46.0,
-      "step": 7943
-    },
-    {
-      "epoch": 0.6073742760479385,
-      "grad_norm": 0.0008711973787285388,
-      "learning_rate": 0.00019999981841122767,
-      "loss": 46.0,
-      "step": 7944
-    },
-    {
-      "epoch": 0.6074507330313282,
-      "grad_norm": 0.0008109626942314208,
-      "learning_rate": 0.00019999981836544997,
-      "loss": 46.0,
-      "step": 7945
-    },
-    {
-      "epoch": 0.607527190014718,
-      "grad_norm": 0.014748835004866123,
-      "learning_rate": 0.00019999981831966647,
-      "loss": 46.0,
-      "step": 7946
-    },
-    {
-      "epoch": 0.6076036469981076,
-      "grad_norm": 0.0026019506622105837,
-      "learning_rate": 0.00019999981827387723,
-      "loss": 46.0,
-      "step": 7947
-    },
-    {
-      "epoch": 0.6076801039814974,
-      "grad_norm": 0.0014633170794695616,
-      "learning_rate": 0.00019999981822808223,
-      "loss": 46.0,
-      "step": 7948
-    },
-    {
-      "epoch": 0.6077565609648872,
-      "grad_norm": 0.0025430629029870033,
-      "learning_rate": 0.00019999981818228144,
-      "loss": 46.0,
-      "step": 7949
-    },
-    {
-      "epoch": 0.6078330179482768,
-      "grad_norm": 0.0008037923253141344,
-      "learning_rate": 0.00019999981813647488,
-      "loss": 46.0,
-      "step": 7950
-    },
-    {
-      "epoch": 0.6079094749316666,
-      "grad_norm": 0.0009830540511757135,
-      "learning_rate": 0.00019999981809066256,
-      "loss": 46.0,
-      "step": 7951
-    },
-    {
-      "epoch": 0.6079859319150563,
-      "grad_norm": 0.002180629177019,
-      "learning_rate": 0.00019999981804484448,
-      "loss": 46.0,
-      "step": 7952
-    },
-    {
-      "epoch": 0.608062388898446,
-      "grad_norm": 0.0002933908544946462,
-      "learning_rate": 0.00019999981799902062,
-      "loss": 46.0,
-      "step": 7953
-    },
-    {
-      "epoch": 0.6081388458818358,
-      "grad_norm": 0.0020938822999596596,
-      "learning_rate": 0.00019999981795319099,
-      "loss": 46.0,
-      "step": 7954
-    },
-    {
-      "epoch": 0.6082153028652254,
-      "grad_norm": 0.0031283488497138023,
-      "learning_rate": 0.00019999981790735558,
-      "loss": 46.0,
-      "step": 7955
-    },
-    {
-      "epoch": 0.6082917598486152,
-      "grad_norm": 0.0013937586918473244,
-      "learning_rate": 0.00019999981786151443,
-      "loss": 46.0,
-      "step": 7956
-    },
-    {
-      "epoch": 0.6083682168320049,
-      "grad_norm": 0.0025686935987323523,
-      "learning_rate": 0.00019999981781566745,
-      "loss": 46.0,
-      "step": 7957
-    },
-    {
-      "epoch": 0.6084446738153946,
-      "grad_norm": 0.0008471534820273519,
-      "learning_rate": 0.00019999981776981475,
-      "loss": 46.0,
-      "step": 7958
-    },
-    {
-      "epoch": 0.6085211307987843,
-      "grad_norm": 0.0034409454092383385,
-      "learning_rate": 0.00019999981772395628,
-      "loss": 46.0,
-      "step": 7959
-    },
-    {
-      "epoch": 0.6085975877821741,
-      "grad_norm": 0.009748689830303192,
-      "learning_rate": 0.00019999981767809203,
-      "loss": 46.0,
-      "step": 7960
-    },
-    {
-      "epoch": 0.6086740447655637,
-      "grad_norm": 0.0012783779529854655,
-      "learning_rate": 0.000199999817632222,
-      "loss": 46.0,
-      "step": 7961
-    },
-    {
-      "epoch": 0.6087505017489535,
-      "grad_norm": 0.0011554979719221592,
-      "learning_rate": 0.00019999981758634622,
-      "loss": 46.0,
-      "step": 7962
-    },
-    {
-      "epoch": 0.6088269587323433,
-      "grad_norm": 0.0010530215222388506,
-      "learning_rate": 0.00019999981754046465,
-      "loss": 46.0,
-      "step": 7963
-    },
-    {
-      "epoch": 0.6089034157157329,
-      "grad_norm": 0.0032395594753324986,
-      "learning_rate": 0.00019999981749457734,
-      "loss": 46.0,
-      "step": 7964
-    },
-    {
-      "epoch": 0.6089798726991227,
-      "grad_norm": 0.0005449085729196668,
-      "learning_rate": 0.00019999981744868426,
-      "loss": 46.0,
-      "step": 7965
-    },
-    {
-      "epoch": 0.6090563296825123,
-      "grad_norm": 0.007412794977426529,
-      "learning_rate": 0.0001999998174027854,
-      "loss": 46.0,
-      "step": 7966
-    },
-    {
-      "epoch": 0.6091327866659021,
-      "grad_norm": 0.0006506446516141295,
-      "learning_rate": 0.00019999981735688077,
-      "loss": 46.0,
-      "step": 7967
-    },
-    {
-      "epoch": 0.6092092436492919,
-      "grad_norm": 0.0011552934302017093,
-      "learning_rate": 0.00019999981731097036,
-      "loss": 46.0,
-      "step": 7968
-    },
-    {
-      "epoch": 0.6092857006326815,
-      "grad_norm": 0.0006012298399582505,
-      "learning_rate": 0.00019999981726505416,
-      "loss": 46.0,
-      "step": 7969
-    },
-    {
-      "epoch": 0.6093621576160713,
-      "grad_norm": 0.0014015400083735585,
-      "learning_rate": 0.00019999981721913223,
-      "loss": 46.0,
-      "step": 7970
-    },
-    {
-      "epoch": 0.609438614599461,
-      "grad_norm": 0.0005265754880383611,
-      "learning_rate": 0.00019999981717320453,
-      "loss": 46.0,
-      "step": 7971
-    },
-    {
-      "epoch": 0.6095150715828507,
-      "grad_norm": 0.0010723063023760915,
-      "learning_rate": 0.00019999981712727106,
-      "loss": 46.0,
-      "step": 7972
-    },
-    {
-      "epoch": 0.6095915285662404,
-      "grad_norm": 0.0007148168515414,
-      "learning_rate": 0.0001999998170813318,
-      "loss": 46.0,
-      "step": 7973
-    },
-    {
-      "epoch": 0.6096679855496301,
-      "grad_norm": 0.0014183580642566085,
-      "learning_rate": 0.00019999981703538677,
-      "loss": 46.0,
-      "step": 7974
-    },
-    {
-      "epoch": 0.6097444425330198,
-      "grad_norm": 0.0021027768962085247,
-      "learning_rate": 0.00019999981698943598,
-      "loss": 46.0,
-      "step": 7975
-    },
-    {
-      "epoch": 0.6098208995164096,
-      "grad_norm": 0.0005962667055428028,
-      "learning_rate": 0.0001999998169434794,
-      "loss": 46.0,
-      "step": 7976
-    },
-    {
-      "epoch": 0.6098973564997993,
-      "grad_norm": 0.00548368226736784,
-      "learning_rate": 0.0001999998168975171,
-      "loss": 46.0,
-      "step": 7977
-    },
-    {
-      "epoch": 0.609973813483189,
-      "grad_norm": 0.0010928643168881536,
-      "learning_rate": 0.000199999816851549,
-      "loss": 46.0,
-      "step": 7978
-    },
-    {
-      "epoch": 0.6100502704665788,
-      "grad_norm": 0.001230394234880805,
-      "learning_rate": 0.00019999981680557513,
-      "loss": 46.0,
-      "step": 7979
-    },
-    {
-      "epoch": 0.6101267274499684,
-      "grad_norm": 0.002780554350465536,
-      "learning_rate": 0.0001999998167595955,
-      "loss": 46.0,
-      "step": 7980
-    },
-    {
-      "epoch": 0.6102031844333582,
-      "grad_norm": 0.0009225890971720219,
-      "learning_rate": 0.00019999981671361007,
-      "loss": 46.0,
-      "step": 7981
-    },
-    {
-      "epoch": 0.610279641416748,
-      "grad_norm": 0.0007890506531111896,
-      "learning_rate": 0.00019999981666761891,
-      "loss": 46.0,
-      "step": 7982
-    },
-    {
-      "epoch": 0.6103560984001376,
-      "grad_norm": 0.0011693231062963605,
-      "learning_rate": 0.00019999981662162196,
-      "loss": 46.0,
-      "step": 7983
-    },
-    {
-      "epoch": 0.6104325553835274,
-      "grad_norm": 0.0018700744258239865,
-      "learning_rate": 0.00019999981657561926,
-      "loss": 46.0,
-      "step": 7984
-    },
-    {
-      "epoch": 0.610509012366917,
-      "grad_norm": 0.0013759142020717263,
-      "learning_rate": 0.00019999981652961076,
-      "loss": 46.0,
-      "step": 7985
-    },
-    {
-      "epoch": 0.6105854693503068,
-      "grad_norm": 0.0035913782194256783,
-      "learning_rate": 0.00019999981648359652,
-      "loss": 46.0,
-      "step": 7986
-    },
-    {
-      "epoch": 0.6106619263336965,
-      "grad_norm": 0.003008329076692462,
-      "learning_rate": 0.00019999981643757647,
-      "loss": 46.0,
-      "step": 7987
-    },
-    {
-      "epoch": 0.6107383833170862,
-      "grad_norm": 0.0024229709524661303,
-      "learning_rate": 0.0001999998163915507,
-      "loss": 46.0,
-      "step": 7988
-    },
-    {
-      "epoch": 0.610814840300476,
-      "grad_norm": 0.000699841242749244,
-      "learning_rate": 0.00019999981634551914,
-      "loss": 46.0,
-      "step": 7989
-    },
-    {
-      "epoch": 0.6108912972838657,
-      "grad_norm": 0.0006508107762783766,
-      "learning_rate": 0.00019999981629948178,
-      "loss": 46.0,
-      "step": 7990
-    },
-    {
-      "epoch": 0.6109677542672554,
-      "grad_norm": 0.004089084919542074,
-      "learning_rate": 0.00019999981625343872,
-      "loss": 46.0,
-      "step": 7991
-    },
-    {
-      "epoch": 0.6110442112506451,
-      "grad_norm": 0.0006521567120216787,
-      "learning_rate": 0.00019999981620738984,
-      "loss": 46.0,
-      "step": 7992
-    },
-    {
-      "epoch": 0.6111206682340349,
-      "grad_norm": 0.0008699916652403772,
-      "learning_rate": 0.00019999981616133518,
-      "loss": 46.0,
-      "step": 7993
-    },
-    {
-      "epoch": 0.6111971252174245,
-      "grad_norm": 0.0006521597388200462,
-      "learning_rate": 0.00019999981611527477,
-      "loss": 46.0,
-      "step": 7994
-    },
-    {
-      "epoch": 0.6112735822008143,
-      "grad_norm": 0.0009300427045673132,
-      "learning_rate": 0.00019999981606920862,
-      "loss": 46.0,
-      "step": 7995
-    },
-    {
-      "epoch": 0.6113500391842039,
-      "grad_norm": 0.004025891423225403,
-      "learning_rate": 0.00019999981602313667,
-      "loss": 46.0,
-      "step": 7996
-    },
-    {
-      "epoch": 0.6114264961675937,
-      "grad_norm": 0.0010910285636782646,
-      "learning_rate": 0.00019999981597705895,
-      "loss": 46.0,
-      "step": 7997
-    },
-    {
-      "epoch": 0.6115029531509835,
-      "grad_norm": 0.0014916922664269805,
-      "learning_rate": 0.00019999981593097545,
-      "loss": 46.0,
-      "step": 7998
-    },
-    {
-      "epoch": 0.6115794101343731,
-      "grad_norm": 0.0008063392015174031,
-      "learning_rate": 0.00019999981588488618,
-      "loss": 46.0,
-      "step": 7999
-    },
-    {
-      "epoch": 0.6116558671177629,
-      "grad_norm": 0.0016277688555419445,
-      "learning_rate": 0.00019999981583879118,
-      "loss": 46.0,
-      "step": 8000
-    },
-    {
-      "epoch": 0.6117323241011526,
-      "grad_norm": 0.0021985953208059072,
-      "learning_rate": 0.0001999998157926904,
-      "loss": 46.0,
-      "step": 8001
-    },
-    {
-      "epoch": 0.6118087810845423,
-      "grad_norm": 0.0006408132030628622,
-      "learning_rate": 0.0001999998157465838,
-      "loss": 46.0,
-      "step": 8002
-    },
-    {
-      "epoch": 0.611885238067932,
-      "grad_norm": 0.0009227868868038058,
-      "learning_rate": 0.0001999998157004715,
-      "loss": 46.0,
-      "step": 8003
-    },
-    {
-      "epoch": 0.6119616950513217,
-      "grad_norm": 0.0007038955227471888,
-      "learning_rate": 0.00019999981565435338,
-      "loss": 46.0,
-      "step": 8004
-    },
-    {
-      "epoch": 0.6120381520347115,
-      "grad_norm": 0.0008833712199702859,
-      "learning_rate": 0.0001999998156082295,
-      "loss": 46.0,
-      "step": 8005
-    },
-    {
-      "epoch": 0.6121146090181012,
-      "grad_norm": 0.0031500300392508507,
-      "learning_rate": 0.00019999981556209986,
-      "loss": 46.0,
-      "step": 8006
-    },
-    {
-      "epoch": 0.6121910660014909,
-      "grad_norm": 0.0020365710370242596,
-      "learning_rate": 0.00019999981551596446,
-      "loss": 46.0,
-      "step": 8007
-    },
-    {
-      "epoch": 0.6122675229848806,
-      "grad_norm": 0.000716263079084456,
-      "learning_rate": 0.00019999981546982328,
-      "loss": 46.0,
-      "step": 8008
-    },
-    {
-      "epoch": 0.6123439799682704,
-      "grad_norm": 0.0014570709317922592,
-      "learning_rate": 0.00019999981542367633,
-      "loss": 46.0,
-      "step": 8009
-    },
-    {
-      "epoch": 0.61242043695166,
-      "grad_norm": 0.0013562020612880588,
-      "learning_rate": 0.0001999998153775236,
-      "loss": 46.0,
-      "step": 8010
-    },
-    {
-      "epoch": 0.6124968939350498,
-      "grad_norm": 0.0018453149823471904,
-      "learning_rate": 0.00019999981533136508,
-      "loss": 46.0,
-      "step": 8011
-    },
-    {
-      "epoch": 0.6125733509184396,
-      "grad_norm": 0.002312688622623682,
-      "learning_rate": 0.00019999981528520084,
-      "loss": 46.0,
-      "step": 8012
-    },
-    {
-      "epoch": 0.6126498079018292,
-      "grad_norm": 0.0105765201151371,
-      "learning_rate": 0.00019999981523903082,
-      "loss": 46.0,
-      "step": 8013
-    },
-    {
-      "epoch": 0.612726264885219,
-      "grad_norm": 0.005194495897740126,
-      "learning_rate": 0.00019999981519285503,
-      "loss": 46.0,
-      "step": 8014
-    },
-    {
-      "epoch": 0.6128027218686086,
-      "grad_norm": 0.0015117257134988904,
-      "learning_rate": 0.00019999981514667346,
-      "loss": 46.0,
-      "step": 8015
-    },
-    {
-      "epoch": 0.6128791788519984,
-      "grad_norm": 0.0010569387814030051,
-      "learning_rate": 0.0001999998151004861,
-      "loss": 46.0,
-      "step": 8016
-    },
-    {
-      "epoch": 0.6129556358353881,
-      "grad_norm": 0.0006272075697779655,
-      "learning_rate": 0.00019999981505429299,
-      "loss": 46.0,
-      "step": 8017
-    },
-    {
-      "epoch": 0.6130320928187778,
-      "grad_norm": 0.0040042707696557045,
-      "learning_rate": 0.00019999981500809413,
-      "loss": 46.0,
-      "step": 8018
-    },
-    {
-      "epoch": 0.6131085498021676,
-      "grad_norm": 0.005988226737827063,
-      "learning_rate": 0.0001999998149618895,
-      "loss": 46.0,
-      "step": 8019
-    },
-    {
-      "epoch": 0.6131850067855573,
-      "grad_norm": 0.0014884572010487318,
-      "learning_rate": 0.00019999981491567907,
-      "loss": 46.0,
-      "step": 8020
-    },
-    {
-      "epoch": 0.613261463768947,
-      "grad_norm": 0.001359344576485455,
-      "learning_rate": 0.0001999998148694629,
-      "loss": 46.0,
-      "step": 8021
-    },
-    {
-      "epoch": 0.6133379207523367,
-      "grad_norm": 0.003955220337957144,
-      "learning_rate": 0.00019999981482324094,
-      "loss": 46.0,
-      "step": 8022
-    },
-    {
-      "epoch": 0.6134143777357265,
-      "grad_norm": 0.0016917415196076035,
-      "learning_rate": 0.0001999998147770132,
-      "loss": 46.0,
-      "step": 8023
-    },
-    {
-      "epoch": 0.6134908347191161,
-      "grad_norm": 0.0012416922254487872,
-      "learning_rate": 0.00019999981473077974,
-      "loss": 46.0,
-      "step": 8024
-    },
-    {
-      "epoch": 0.6135672917025059,
-      "grad_norm": 0.002291525946930051,
-      "learning_rate": 0.00019999981468454044,
-      "loss": 46.0,
-      "step": 8025
-    },
-    {
-      "epoch": 0.6136437486858956,
-      "grad_norm": 0.001350725069642067,
-      "learning_rate": 0.00019999981463829543,
-      "loss": 46.0,
-      "step": 8026
-    },
-    {
-      "epoch": 0.6137202056692853,
-      "grad_norm": 0.0006322698318399489,
-      "learning_rate": 0.0001999998145920446,
-      "loss": 46.0,
-      "step": 8027
-    },
-    {
-      "epoch": 0.6137966626526751,
-      "grad_norm": 0.0008371872827410698,
-      "learning_rate": 0.00019999981454578807,
-      "loss": 46.0,
-      "step": 8028
-    },
-    {
-      "epoch": 0.6138731196360647,
-      "grad_norm": 0.0026724180206656456,
-      "learning_rate": 0.0001999998144995257,
-      "loss": 46.0,
-      "step": 8029
-    },
-    {
-      "epoch": 0.6139495766194545,
-      "grad_norm": 0.0005557225667871535,
-      "learning_rate": 0.0001999998144532576,
-      "loss": 46.0,
-      "step": 8030
-    },
-    {
-      "epoch": 0.6140260336028442,
-      "grad_norm": 0.0018545619677752256,
-      "learning_rate": 0.0001999998144069837,
-      "loss": 46.0,
-      "step": 8031
-    },
-    {
-      "epoch": 0.6141024905862339,
-      "grad_norm": 0.0029547943267971277,
-      "learning_rate": 0.00019999981436070408,
-      "loss": 46.0,
-      "step": 8032
-    },
-    {
-      "epoch": 0.6141789475696237,
-      "grad_norm": 0.0007139172521419823,
-      "learning_rate": 0.00019999981431441865,
-      "loss": 46.0,
-      "step": 8033
-    },
-    {
-      "epoch": 0.6142554045530133,
-      "grad_norm": 0.0006746540893800557,
-      "learning_rate": 0.00019999981426812747,
-      "loss": 46.0,
-      "step": 8034
-    },
-    {
-      "epoch": 0.6143318615364031,
-      "grad_norm": 0.0074030994437634945,
-      "learning_rate": 0.00019999981422183052,
-      "loss": 46.0,
-      "step": 8035
-    },
-    {
-      "epoch": 0.6144083185197928,
-      "grad_norm": 0.004032945726066828,
-      "learning_rate": 0.0001999998141755278,
-      "loss": 46.0,
-      "step": 8036
-    },
-    {
-      "epoch": 0.6144847755031825,
-      "grad_norm": 0.0027331418823450804,
-      "learning_rate": 0.00019999981412921927,
-      "loss": 46.0,
-      "step": 8037
-    },
-    {
-      "epoch": 0.6145612324865722,
-      "grad_norm": 0.0005232125404290855,
-      "learning_rate": 0.00019999981408290506,
-      "loss": 46.0,
-      "step": 8038
-    },
-    {
-      "epoch": 0.614637689469962,
-      "grad_norm": 0.003908711951225996,
-      "learning_rate": 0.000199999814036585,
-      "loss": 46.0,
-      "step": 8039
-    },
-    {
-      "epoch": 0.6147141464533517,
-      "grad_norm": 0.0006954863201826811,
-      "learning_rate": 0.0001999998139902592,
-      "loss": 46.0,
-      "step": 8040
-    },
-    {
-      "epoch": 0.6147906034367414,
-      "grad_norm": 0.0011021033860743046,
-      "learning_rate": 0.00019999981394392763,
-      "loss": 46.0,
-      "step": 8041
-    },
-    {
-      "epoch": 0.6148670604201312,
-      "grad_norm": 0.0013622321421280503,
-      "learning_rate": 0.0001999998138975903,
-      "loss": 46.0,
-      "step": 8042
-    },
-    {
-      "epoch": 0.6149435174035208,
-      "grad_norm": 0.0008329287520609796,
-      "learning_rate": 0.00019999981385124718,
-      "loss": 46.0,
-      "step": 8043
-    },
-    {
-      "epoch": 0.6150199743869106,
-      "grad_norm": 0.0026834269519895315,
-      "learning_rate": 0.0001999998138048983,
-      "loss": 46.0,
-      "step": 8044
-    },
-    {
-      "epoch": 0.6150964313703002,
-      "grad_norm": 0.0016994967591017485,
-      "learning_rate": 0.00019999981375854364,
-      "loss": 46.0,
-      "step": 8045
-    },
-    {
-      "epoch": 0.61517288835369,
-      "grad_norm": 0.0007733700331300497,
-      "learning_rate": 0.0001999998137121832,
-      "loss": 46.0,
-      "step": 8046
-    },
-    {
-      "epoch": 0.6152493453370798,
-      "grad_norm": 0.0008033501217141747,
-      "learning_rate": 0.00019999981366581704,
-      "loss": 46.0,
-      "step": 8047
-    },
-    {
-      "epoch": 0.6153258023204694,
-      "grad_norm": 0.0006857018452137709,
-      "learning_rate": 0.00019999981361944506,
-      "loss": 46.0,
-      "step": 8048
-    },
-    {
-      "epoch": 0.6154022593038592,
-      "grad_norm": 0.0024543199688196182,
-      "learning_rate": 0.00019999981357306733,
-      "loss": 46.0,
-      "step": 8049
-    },
-    {
-      "epoch": 0.6154787162872489,
-      "grad_norm": 0.0034503464121371508,
-      "learning_rate": 0.00019999981352668384,
-      "loss": 46.0,
-      "step": 8050
-    },
-    {
-      "epoch": 0.6155551732706386,
-      "grad_norm": 0.0011730798287317157,
-      "learning_rate": 0.00019999981348029457,
-      "loss": 46.0,
-      "step": 8051
-    },
-    {
-      "epoch": 0.6156316302540283,
-      "grad_norm": 0.001017491682432592,
-      "learning_rate": 0.00019999981343389952,
-      "loss": 46.0,
-      "step": 8052
-    },
-    {
-      "epoch": 0.6157080872374181,
-      "grad_norm": 0.0011439293157309294,
-      "learning_rate": 0.00019999981338749873,
-      "loss": 46.0,
-      "step": 8053
-    },
-    {
-      "epoch": 0.6157845442208078,
-      "grad_norm": 0.000915037642698735,
-      "learning_rate": 0.00019999981334109217,
-      "loss": 46.0,
-      "step": 8054
-    },
-    {
-      "epoch": 0.6158610012041975,
-      "grad_norm": 0.002680210629478097,
-      "learning_rate": 0.00019999981329467983,
-      "loss": 46.0,
-      "step": 8055
-    },
-    {
-      "epoch": 0.6159374581875872,
-      "grad_norm": 0.002151148859411478,
-      "learning_rate": 0.00019999981324826172,
-      "loss": 46.0,
-      "step": 8056
-    },
-    {
-      "epoch": 0.6160139151709769,
-      "grad_norm": 0.0017771321581676602,
-      "learning_rate": 0.00019999981320183784,
-      "loss": 46.0,
-      "step": 8057
-    },
-    {
-      "epoch": 0.6160903721543667,
-      "grad_norm": 0.0005965745658613741,
-      "learning_rate": 0.00019999981315540815,
-      "loss": 46.0,
-      "step": 8058
-    },
-    {
-      "epoch": 0.6161668291377563,
-      "grad_norm": 0.0004076090408489108,
-      "learning_rate": 0.00019999981310897272,
-      "loss": 46.0,
-      "step": 8059
-    },
-    {
-      "epoch": 0.6162432861211461,
-      "grad_norm": 0.0014622843591496348,
-      "learning_rate": 0.00019999981306253155,
-      "loss": 46.0,
-      "step": 8060
-    },
-    {
-      "epoch": 0.6163197431045359,
-      "grad_norm": 0.0011214729165658355,
-      "learning_rate": 0.0001999998130160846,
-      "loss": 46.0,
-      "step": 8061
-    },
-    {
-      "epoch": 0.6163962000879255,
-      "grad_norm": 0.005945032462477684,
-      "learning_rate": 0.00019999981296963185,
-      "loss": 46.0,
-      "step": 8062
-    },
-    {
-      "epoch": 0.6164726570713153,
-      "grad_norm": 0.001295273075811565,
-      "learning_rate": 0.00019999981292317335,
-      "loss": 46.0,
-      "step": 8063
-    },
-    {
-      "epoch": 0.6165491140547049,
-      "grad_norm": 0.001352186081930995,
-      "learning_rate": 0.00019999981287670908,
-      "loss": 46.0,
-      "step": 8064
-    },
-    {
-      "epoch": 0.6166255710380947,
-      "grad_norm": 0.005008733365684748,
-      "learning_rate": 0.00019999981283023904,
-      "loss": 46.0,
-      "step": 8065
-    },
-    {
-      "epoch": 0.6167020280214844,
-      "grad_norm": 0.0027853958308696747,
-      "learning_rate": 0.00019999981278376322,
-      "loss": 46.0,
-      "step": 8066
-    },
-    {
-      "epoch": 0.6167784850048741,
-      "grad_norm": 0.004503860138356686,
-      "learning_rate": 0.00019999981273728166,
-      "loss": 46.0,
-      "step": 8067
-    },
-    {
-      "epoch": 0.6168549419882638,
-      "grad_norm": 0.0007594541530124843,
-      "learning_rate": 0.00019999981269079432,
-      "loss": 46.0,
-      "step": 8068
-    },
-    {
-      "epoch": 0.6169313989716536,
-      "grad_norm": 0.002667501801624894,
-      "learning_rate": 0.00019999981264430118,
-      "loss": 46.0,
-      "step": 8069
-    },
-    {
-      "epoch": 0.6170078559550433,
-      "grad_norm": 0.0008925874135456979,
-      "learning_rate": 0.0001999998125978023,
-      "loss": 46.0,
-      "step": 8070
-    },
-    {
-      "epoch": 0.617084312938433,
-      "grad_norm": 0.0010283252922818065,
-      "learning_rate": 0.00019999981255129764,
-      "loss": 46.0,
-      "step": 8071
-    },
-    {
-      "epoch": 0.6171607699218228,
-      "grad_norm": 0.0024446446914225817,
-      "learning_rate": 0.00019999981250478724,
-      "loss": 46.0,
-      "step": 8072
-    },
-    {
-      "epoch": 0.6172372269052124,
-      "grad_norm": 0.0012787660816684365,
-      "learning_rate": 0.000199999812458271,
-      "loss": 46.0,
-      "step": 8073
-    },
-    {
-      "epoch": 0.6173136838886022,
-      "grad_norm": 0.0007039426127448678,
-      "learning_rate": 0.0001999998124117491,
-      "loss": 46.0,
-      "step": 8074
-    },
-    {
-      "epoch": 0.6173901408719918,
-      "grad_norm": 0.0008649210212752223,
-      "learning_rate": 0.0001999998123652213,
-      "loss": 46.0,
-      "step": 8075
-    },
-    {
-      "epoch": 0.6174665978553816,
-      "grad_norm": 0.0013115047477185726,
-      "learning_rate": 0.00019999981231868784,
-      "loss": 46.0,
-      "step": 8076
-    },
-    {
-      "epoch": 0.6175430548387714,
-      "grad_norm": 0.001115027698688209,
-      "learning_rate": 0.00019999981227214854,
-      "loss": 46.0,
-      "step": 8077
-    },
-    {
-      "epoch": 0.617619511822161,
-      "grad_norm": 0.0025179898366332054,
-      "learning_rate": 0.0001999998122256035,
-      "loss": 46.0,
-      "step": 8078
-    },
-    {
-      "epoch": 0.6176959688055508,
-      "grad_norm": 0.0010529948631301522,
-      "learning_rate": 0.00019999981217905268,
-      "loss": 46.0,
-      "step": 8079
-    },
-    {
-      "epoch": 0.6177724257889405,
-      "grad_norm": 0.0008856932399794459,
-      "learning_rate": 0.0001999998121324961,
-      "loss": 46.0,
-      "step": 8080
-    },
-    {
-      "epoch": 0.6178488827723302,
-      "grad_norm": 0.0019840889144688845,
-      "learning_rate": 0.00019999981208593376,
-      "loss": 46.0,
-      "step": 8081
-    },
-    {
-      "epoch": 0.61792533975572,
-      "grad_norm": 0.0014206685591489077,
-      "learning_rate": 0.00019999981203936565,
-      "loss": 46.0,
-      "step": 8082
-    },
-    {
-      "epoch": 0.6180017967391097,
-      "grad_norm": 0.0009865479078143835,
-      "learning_rate": 0.00019999981199279177,
-      "loss": 46.0,
-      "step": 8083
-    },
-    {
-      "epoch": 0.6180782537224994,
-      "grad_norm": 0.0011100240517407656,
-      "learning_rate": 0.00019999981194621208,
-      "loss": 46.0,
-      "step": 8084
-    },
-    {
-      "epoch": 0.6181547107058891,
-      "grad_norm": 0.011240546591579914,
-      "learning_rate": 0.00019999981189962668,
-      "loss": 46.0,
-      "step": 8085
-    },
-    {
-      "epoch": 0.6182311676892788,
-      "grad_norm": 0.0008880603709258139,
-      "learning_rate": 0.00019999981185303548,
-      "loss": 46.0,
-      "step": 8086
-    },
-    {
-      "epoch": 0.6183076246726685,
-      "grad_norm": 0.0014324372168630362,
-      "learning_rate": 0.0001999998118064385,
-      "loss": 46.0,
-      "step": 8087
-    },
-    {
-      "epoch": 0.6183840816560583,
-      "grad_norm": 0.001325063407421112,
-      "learning_rate": 0.00019999981175983575,
-      "loss": 46.0,
-      "step": 8088
-    },
-    {
-      "epoch": 0.6184605386394479,
-      "grad_norm": 0.006308611016720533,
-      "learning_rate": 0.00019999981171322725,
-      "loss": 46.0,
-      "step": 8089
-    },
-    {
-      "epoch": 0.6185369956228377,
-      "grad_norm": 0.0012192374560981989,
-      "learning_rate": 0.00019999981166661298,
-      "loss": 46.0,
-      "step": 8090
-    },
-    {
-      "epoch": 0.6186134526062275,
-      "grad_norm": 0.00036524387542158365,
-      "learning_rate": 0.00019999981161999294,
-      "loss": 46.0,
-      "step": 8091
-    },
-    {
-      "epoch": 0.6186899095896171,
-      "grad_norm": 0.0008525089360773563,
-      "learning_rate": 0.00019999981157336712,
-      "loss": 46.0,
-      "step": 8092
-    },
-    {
-      "epoch": 0.6187663665730069,
-      "grad_norm": 0.007759463042020798,
-      "learning_rate": 0.00019999981152673553,
-      "loss": 46.0,
-      "step": 8093
-    },
-    {
-      "epoch": 0.6188428235563966,
-      "grad_norm": 0.0008733192807994783,
-      "learning_rate": 0.0001999998114800982,
-      "loss": 46.0,
-      "step": 8094
-    },
-    {
-      "epoch": 0.6189192805397863,
-      "grad_norm": 0.0014871356543153524,
-      "learning_rate": 0.00019999981143345506,
-      "loss": 46.0,
-      "step": 8095
-    },
-    {
-      "epoch": 0.618995737523176,
-      "grad_norm": 0.00141401297878474,
-      "learning_rate": 0.00019999981138680618,
-      "loss": 46.0,
-      "step": 8096
-    },
-    {
-      "epoch": 0.6190721945065657,
-      "grad_norm": 0.0007656815578229725,
-      "learning_rate": 0.00019999981134015152,
-      "loss": 46.0,
-      "step": 8097
-    },
-    {
-      "epoch": 0.6191486514899555,
-      "grad_norm": 0.0015928716165944934,
-      "learning_rate": 0.00019999981129349107,
-      "loss": 46.0,
-      "step": 8098
-    },
-    {
-      "epoch": 0.6192251084733452,
-      "grad_norm": 0.0012576529989019036,
-      "learning_rate": 0.00019999981124682486,
-      "loss": 46.0,
-      "step": 8099
-    },
-    {
-      "epoch": 0.6193015654567349,
-      "grad_norm": 0.0030896265525370836,
-      "learning_rate": 0.00019999981120015292,
-      "loss": 46.0,
-      "step": 8100
-    },
-    {
-      "epoch": 0.6193780224401246,
-      "grad_norm": 0.010342762805521488,
-      "learning_rate": 0.00019999981115347517,
-      "loss": 46.0,
-      "step": 8101
-    },
-    {
-      "epoch": 0.6194544794235144,
-      "grad_norm": 0.0005783736705780029,
-      "learning_rate": 0.00019999981110679164,
-      "loss": 46.0,
-      "step": 8102
-    },
-    {
-      "epoch": 0.619530936406904,
-      "grad_norm": 0.002448518993332982,
-      "learning_rate": 0.00019999981106010237,
-      "loss": 46.0,
-      "step": 8103
-    },
-    {
-      "epoch": 0.6196073933902938,
-      "grad_norm": 0.0011705290526151657,
-      "learning_rate": 0.00019999981101340733,
-      "loss": 46.0,
-      "step": 8104
-    },
-    {
-      "epoch": 0.6196838503736835,
-      "grad_norm": 0.005643675569444895,
-      "learning_rate": 0.00019999981096670652,
-      "loss": 46.0,
-      "step": 8105
-    },
-    {
-      "epoch": 0.6197603073570732,
-      "grad_norm": 0.001870142761617899,
-      "learning_rate": 0.00019999981091999993,
-      "loss": 46.0,
-      "step": 8106
-    },
-    {
-      "epoch": 0.619836764340463,
-      "grad_norm": 0.00089590362040326,
-      "learning_rate": 0.0001999998108732876,
-      "loss": 46.0,
-      "step": 8107
-    },
-    {
-      "epoch": 0.6199132213238526,
-      "grad_norm": 0.0008799677598290145,
-      "learning_rate": 0.00019999981082656945,
-      "loss": 46.0,
-      "step": 8108
-    },
-    {
-      "epoch": 0.6199896783072424,
-      "grad_norm": 0.002297072671353817,
-      "learning_rate": 0.00019999981077984554,
-      "loss": 46.0,
-      "step": 8109
-    },
-    {
-      "epoch": 0.6200661352906321,
-      "grad_norm": 0.001265001017600298,
-      "learning_rate": 0.0001999998107331159,
-      "loss": 46.0,
-      "step": 8110
-    },
-    {
-      "epoch": 0.6201425922740218,
-      "grad_norm": 0.0022822138853371143,
-      "learning_rate": 0.00019999981068638046,
-      "loss": 46.0,
-      "step": 8111
-    },
-    {
-      "epoch": 0.6202190492574116,
-      "grad_norm": 0.0011523495195433497,
-      "learning_rate": 0.00019999981063963923,
-      "loss": 46.0,
-      "step": 8112
-    },
-    {
-      "epoch": 0.6202955062408013,
-      "grad_norm": 0.0010711951181292534,
-      "learning_rate": 0.00019999981059289228,
-      "loss": 46.0,
-      "step": 8113
-    },
-    {
-      "epoch": 0.620371963224191,
-      "grad_norm": 0.00091460911789909,
-      "learning_rate": 0.00019999981054613953,
-      "loss": 46.0,
-      "step": 8114
-    },
-    {
-      "epoch": 0.6204484202075807,
-      "grad_norm": 0.0050215148366987705,
-      "learning_rate": 0.00019999981049938104,
-      "loss": 46.0,
-      "step": 8115
-    },
-    {
-      "epoch": 0.6205248771909704,
-      "grad_norm": 0.0029333815909922123,
-      "learning_rate": 0.00019999981045261674,
-      "loss": 46.0,
-      "step": 8116
-    },
-    {
-      "epoch": 0.6206013341743601,
-      "grad_norm": 0.0007092771702446043,
-      "learning_rate": 0.0001999998104058467,
-      "loss": 46.0,
-      "step": 8117
-    },
-    {
-      "epoch": 0.6206777911577499,
-      "grad_norm": 0.0018388079479336739,
-      "learning_rate": 0.00019999981035907088,
-      "loss": 46.0,
-      "step": 8118
-    },
-    {
-      "epoch": 0.6207542481411396,
-      "grad_norm": 0.004218053538352251,
-      "learning_rate": 0.00019999981031228927,
-      "loss": 46.0,
-      "step": 8119
-    },
-    {
-      "epoch": 0.6208307051245293,
-      "grad_norm": 0.0007422819035127759,
-      "learning_rate": 0.00019999981026550193,
-      "loss": 46.0,
-      "step": 8120
-    },
-    {
-      "epoch": 0.6209071621079191,
-      "grad_norm": 0.0019210479222238064,
-      "learning_rate": 0.0001999998102187088,
-      "loss": 46.0,
-      "step": 8121
-    },
-    {
-      "epoch": 0.6209836190913087,
-      "grad_norm": 0.005070696584880352,
-      "learning_rate": 0.0001999998101719099,
-      "loss": 46.0,
-      "step": 8122
-    },
-    {
-      "epoch": 0.6210600760746985,
-      "grad_norm": 0.0004544100956991315,
-      "learning_rate": 0.00019999981012510523,
-      "loss": 46.0,
-      "step": 8123
-    },
-    {
-      "epoch": 0.6211365330580882,
-      "grad_norm": 0.0013448148965835571,
-      "learning_rate": 0.0001999998100782948,
-      "loss": 46.0,
-      "step": 8124
-    },
-    {
-      "epoch": 0.6212129900414779,
-      "grad_norm": 0.001338648027740419,
-      "learning_rate": 0.00019999981003147857,
-      "loss": 46.0,
-      "step": 8125
-    },
-    {
-      "epoch": 0.6212894470248677,
-      "grad_norm": 0.003315623616799712,
-      "learning_rate": 0.0001999998099846566,
-      "loss": 46.0,
-      "step": 8126
-    },
-    {
-      "epoch": 0.6213659040082573,
-      "grad_norm": 0.0006857843254692852,
-      "learning_rate": 0.00019999980993782885,
-      "loss": 46.0,
-      "step": 8127
-    },
-    {
-      "epoch": 0.6214423609916471,
-      "grad_norm": 0.0009068864746950567,
-      "learning_rate": 0.00019999980989099533,
-      "loss": 46.0,
-      "step": 8128
-    },
-    {
-      "epoch": 0.6215188179750368,
-      "grad_norm": 0.0015287484275177121,
-      "learning_rate": 0.00019999980984415608,
-      "loss": 46.0,
-      "step": 8129
-    },
-    {
-      "epoch": 0.6215952749584265,
-      "grad_norm": 0.0013803041074424982,
-      "learning_rate": 0.00019999980979731102,
-      "loss": 46.0,
-      "step": 8130
-    },
-    {
-      "epoch": 0.6216717319418162,
-      "grad_norm": 0.011175794526934624,
-      "learning_rate": 0.0001999998097504602,
-      "loss": 46.0,
-      "step": 8131
-    },
-    {
-      "epoch": 0.621748188925206,
-      "grad_norm": 0.0009191324934363365,
-      "learning_rate": 0.00019999980970360358,
-      "loss": 46.0,
-      "step": 8132
-    },
-    {
-      "epoch": 0.6218246459085957,
-      "grad_norm": 0.003293754532933235,
-      "learning_rate": 0.00019999980965674122,
-      "loss": 46.0,
-      "step": 8133
-    },
-    {
-      "epoch": 0.6219011028919854,
-      "grad_norm": 0.001244703191332519,
-      "learning_rate": 0.00019999980960987311,
-      "loss": 46.0,
-      "step": 8134
-    },
-    {
-      "epoch": 0.6219775598753751,
-      "grad_norm": 0.0016437621088698506,
-      "learning_rate": 0.0001999998095629992,
-      "loss": 46.0,
-      "step": 8135
-    },
-    {
-      "epoch": 0.6220540168587648,
-      "grad_norm": 0.001206371234729886,
-      "learning_rate": 0.00019999980951611955,
-      "loss": 46.0,
-      "step": 8136
-    },
-    {
-      "epoch": 0.6221304738421546,
-      "grad_norm": 0.0010340720182284713,
-      "learning_rate": 0.0001999998094692341,
-      "loss": 46.0,
-      "step": 8137
-    },
-    {
-      "epoch": 0.6222069308255442,
-      "grad_norm": 0.0015247574774548411,
-      "learning_rate": 0.0001999998094223429,
-      "loss": 46.0,
-      "step": 8138
-    },
-    {
-      "epoch": 0.622283387808934,
-      "grad_norm": 0.0011224700137972832,
-      "learning_rate": 0.00019999980937544592,
-      "loss": 46.0,
-      "step": 8139
-    },
-    {
-      "epoch": 0.6223598447923238,
-      "grad_norm": 0.0004623421700671315,
-      "learning_rate": 0.00019999980932854317,
-      "loss": 46.0,
-      "step": 8140
-    },
-    {
-      "epoch": 0.6224363017757134,
-      "grad_norm": 0.0005331024876795709,
-      "learning_rate": 0.00019999980928163465,
-      "loss": 46.0,
-      "step": 8141
-    },
-    {
-      "epoch": 0.6225127587591032,
-      "grad_norm": 0.005343261640518904,
-      "learning_rate": 0.00019999980923472035,
-      "loss": 46.0,
-      "step": 8142
-    },
-    {
-      "epoch": 0.6225892157424929,
-      "grad_norm": 0.0006872197845950723,
-      "learning_rate": 0.0001999998091878003,
-      "loss": 46.0,
-      "step": 8143
-    },
-    {
-      "epoch": 0.6226656727258826,
-      "grad_norm": 0.001922086114063859,
-      "learning_rate": 0.0001999998091408745,
-      "loss": 46.0,
-      "step": 8144
-    },
-    {
-      "epoch": 0.6227421297092723,
-      "grad_norm": 0.0020486663561314344,
-      "learning_rate": 0.00019999980909394288,
-      "loss": 46.0,
-      "step": 8145
-    },
-    {
-      "epoch": 0.622818586692662,
-      "grad_norm": 0.001683650421909988,
-      "learning_rate": 0.00019999980904700552,
-      "loss": 46.0,
-      "step": 8146
-    },
-    {
-      "epoch": 0.6228950436760518,
-      "grad_norm": 0.0016485704109072685,
-      "learning_rate": 0.00019999980900006236,
-      "loss": 46.0,
-      "step": 8147
-    },
-    {
-      "epoch": 0.6229715006594415,
-      "grad_norm": 0.0005011818720959127,
-      "learning_rate": 0.00019999980895311347,
-      "loss": 46.0,
-      "step": 8148
-    },
-    {
-      "epoch": 0.6230479576428312,
-      "grad_norm": 0.0012467411579564214,
-      "learning_rate": 0.00019999980890615882,
-      "loss": 46.0,
-      "step": 8149
-    },
-    {
-      "epoch": 0.6231244146262209,
-      "grad_norm": 0.0012904610484838486,
-      "learning_rate": 0.0001999998088591984,
-      "loss": 46.0,
-      "step": 8150
-    },
-    {
-      "epoch": 0.6232008716096107,
-      "grad_norm": 0.006659666076302528,
-      "learning_rate": 0.00019999980881223214,
-      "loss": 46.0,
-      "step": 8151
-    },
-    {
-      "epoch": 0.6232773285930003,
-      "grad_norm": 0.001498241676017642,
-      "learning_rate": 0.00019999980876526016,
-      "loss": 46.0,
-      "step": 8152
-    },
-    {
-      "epoch": 0.6233537855763901,
-      "grad_norm": 0.0018610622501000762,
-      "learning_rate": 0.00019999980871828244,
-      "loss": 46.0,
-      "step": 8153
-    },
-    {
-      "epoch": 0.6234302425597799,
-      "grad_norm": 0.0009072546381503344,
-      "learning_rate": 0.00019999980867129892,
-      "loss": 46.0,
-      "step": 8154
-    },
-    {
-      "epoch": 0.6235066995431695,
-      "grad_norm": 0.002177738817408681,
-      "learning_rate": 0.0001999998086243096,
-      "loss": 46.0,
-      "step": 8155
-    },
-    {
-      "epoch": 0.6235831565265593,
-      "grad_norm": 0.0026074019260704517,
-      "learning_rate": 0.00019999980857731455,
-      "loss": 46.0,
-      "step": 8156
-    },
-    {
-      "epoch": 0.6236596135099489,
-      "grad_norm": 0.0005360660725273192,
-      "learning_rate": 0.00019999980853031374,
-      "loss": 46.0,
-      "step": 8157
-    },
-    {
-      "epoch": 0.6237360704933387,
-      "grad_norm": 0.0013998214853927493,
-      "learning_rate": 0.00019999980848330712,
-      "loss": 46.0,
-      "step": 8158
-    },
-    {
-      "epoch": 0.6238125274767284,
-      "grad_norm": 0.0039712293073534966,
-      "learning_rate": 0.00019999980843629476,
-      "loss": 46.0,
-      "step": 8159
-    },
-    {
-      "epoch": 0.6238889844601181,
-      "grad_norm": 0.0010126747656613588,
-      "learning_rate": 0.00019999980838927663,
-      "loss": 46.0,
-      "step": 8160
-    },
-    {
-      "epoch": 0.6239654414435079,
-      "grad_norm": 0.0031328152399510145,
-      "learning_rate": 0.00019999980834225272,
-      "loss": 46.0,
-      "step": 8161
-    },
-    {
-      "epoch": 0.6240418984268976,
-      "grad_norm": 0.0005414151819422841,
-      "learning_rate": 0.000199999808295223,
-      "loss": 46.0,
-      "step": 8162
-    },
-    {
-      "epoch": 0.6241183554102873,
-      "grad_norm": 0.0018214802257716656,
-      "learning_rate": 0.00019999980824818758,
-      "loss": 46.0,
-      "step": 8163
-    },
-    {
-      "epoch": 0.624194812393677,
-      "grad_norm": 0.0006216823821887374,
-      "learning_rate": 0.00019999980820114635,
-      "loss": 46.0,
-      "step": 8164
-    },
-    {
-      "epoch": 0.6242712693770667,
-      "grad_norm": 0.0010871891863644123,
-      "learning_rate": 0.0001999998081540994,
-      "loss": 46.0,
-      "step": 8165
-    },
-    {
-      "epoch": 0.6243477263604564,
-      "grad_norm": 0.0006974285934120417,
-      "learning_rate": 0.00019999980810704663,
-      "loss": 46.0,
-      "step": 8166
-    },
-    {
-      "epoch": 0.6244241833438462,
-      "grad_norm": 0.001885958481580019,
-      "learning_rate": 0.00019999980805998814,
-      "loss": 46.0,
-      "step": 8167
-    },
-    {
-      "epoch": 0.6245006403272358,
-      "grad_norm": 0.002691855188459158,
-      "learning_rate": 0.00019999980801292382,
-      "loss": 46.0,
-      "step": 8168
-    },
-    {
-      "epoch": 0.6245770973106256,
-      "grad_norm": 0.000914715463295579,
-      "learning_rate": 0.00019999980796585378,
-      "loss": 46.0,
-      "step": 8169
-    },
-    {
-      "epoch": 0.6246535542940154,
-      "grad_norm": 0.010826588608324528,
-      "learning_rate": 0.00019999980791877793,
-      "loss": 46.0,
-      "step": 8170
-    },
-    {
-      "epoch": 0.624730011277405,
-      "grad_norm": 0.0010591283207759261,
-      "learning_rate": 0.00019999980787169635,
-      "loss": 46.0,
-      "step": 8171
-    },
-    {
-      "epoch": 0.6248064682607948,
-      "grad_norm": 0.0016578563954681158,
-      "learning_rate": 0.00019999980782460896,
-      "loss": 46.0,
-      "step": 8172
-    },
-    {
-      "epoch": 0.6248829252441845,
-      "grad_norm": 0.0008777394541539252,
-      "learning_rate": 0.00019999980777751582,
-      "loss": 46.0,
-      "step": 8173
-    },
-    {
-      "epoch": 0.6249593822275742,
-      "grad_norm": 0.012118935585021973,
-      "learning_rate": 0.00019999980773041692,
-      "loss": 46.0,
-      "step": 8174
-    },
-    {
-      "epoch": 0.625035839210964,
-      "grad_norm": 0.005874015856534243,
-      "learning_rate": 0.00019999980768331223,
-      "loss": 46.0,
-      "step": 8175
-    },
-    {
-      "epoch": 0.6251122961943536,
-      "grad_norm": 0.0074394322000443935,
-      "learning_rate": 0.00019999980763620178,
-      "loss": 46.0,
-      "step": 8176
-    },
-    {
-      "epoch": 0.6251887531777434,
-      "grad_norm": 0.002545940689742565,
-      "learning_rate": 0.00019999980758908555,
-      "loss": 46.0,
-      "step": 8177
-    },
-    {
-      "epoch": 0.6252652101611331,
-      "grad_norm": 0.003837330499663949,
-      "learning_rate": 0.0001999998075419636,
-      "loss": 46.0,
-      "step": 8178
-    },
-    {
-      "epoch": 0.6253416671445228,
-      "grad_norm": 0.0022016491275280714,
-      "learning_rate": 0.00019999980749483586,
-      "loss": 46.0,
-      "step": 8179
-    },
-    {
-      "epoch": 0.6254181241279125,
-      "grad_norm": 0.003395914565771818,
-      "learning_rate": 0.0001999998074477023,
-      "loss": 46.0,
-      "step": 8180
-    },
-    {
-      "epoch": 0.6254945811113023,
-      "grad_norm": 0.000628401234280318,
-      "learning_rate": 0.00019999980740056301,
-      "loss": 46.0,
-      "step": 8181
-    },
-    {
-      "epoch": 0.625571038094692,
-      "grad_norm": 0.001131386961787939,
-      "learning_rate": 0.00019999980735341795,
-      "loss": 46.0,
-      "step": 8182
-    },
-    {
-      "epoch": 0.6256474950780817,
-      "grad_norm": 0.0005662293406203389,
-      "learning_rate": 0.0001999998073062671,
-      "loss": 46.0,
-      "step": 8183
-    },
-    {
-      "epoch": 0.6257239520614715,
-      "grad_norm": 0.0011541598942130804,
-      "learning_rate": 0.0001999998072591105,
-      "loss": 46.0,
-      "step": 8184
-    },
-    {
-      "epoch": 0.6258004090448611,
-      "grad_norm": 0.0013944199308753014,
-      "learning_rate": 0.00019999980721194813,
-      "loss": 46.0,
-      "step": 8185
-    },
-    {
-      "epoch": 0.6258768660282509,
-      "grad_norm": 0.0013343654572963715,
-      "learning_rate": 0.00019999980716478,
-      "loss": 46.0,
-      "step": 8186
-    },
-    {
-      "epoch": 0.6259533230116405,
-      "grad_norm": 0.0031992613803595304,
-      "learning_rate": 0.0001999998071176061,
-      "loss": 46.0,
-      "step": 8187
-    },
-    {
-      "epoch": 0.6260297799950303,
-      "grad_norm": 0.0018886931939050555,
-      "learning_rate": 0.0001999998070704264,
-      "loss": 46.0,
-      "step": 8188
-    },
-    {
-      "epoch": 0.62610623697842,
-      "grad_norm": 0.00206000916659832,
-      "learning_rate": 0.00019999980702324095,
-      "loss": 46.0,
-      "step": 8189
-    },
-    {
-      "epoch": 0.6261826939618097,
-      "grad_norm": 0.0018751887837424874,
-      "learning_rate": 0.00019999980697604972,
-      "loss": 46.0,
-      "step": 8190
-    },
-    {
-      "epoch": 0.6262591509451995,
-      "grad_norm": 0.0003227642737329006,
-      "learning_rate": 0.00019999980692885275,
-      "loss": 46.0,
-      "step": 8191
-    },
-    {
-      "epoch": 0.6263356079285892,
-      "grad_norm": 0.007246968802064657,
-      "learning_rate": 0.00019999980688164998,
-      "loss": 46.0,
-      "step": 8192
-    },
-    {
-      "epoch": 0.6264120649119789,
-      "grad_norm": 0.0010604896815493703,
-      "learning_rate": 0.00019999980683444143,
-      "loss": 46.0,
-      "step": 8193
-    },
-    {
-      "epoch": 0.6264885218953686,
-      "grad_norm": 0.00804203748703003,
-      "learning_rate": 0.00019999980678722716,
-      "loss": 46.0,
-      "step": 8194
-    },
-    {
-      "epoch": 0.6265649788787584,
-      "grad_norm": 0.0009484539623372257,
-      "learning_rate": 0.0001999998067400071,
-      "loss": 46.0,
-      "step": 8195
-    },
-    {
-      "epoch": 0.626641435862148,
-      "grad_norm": 0.0018112839898094535,
-      "learning_rate": 0.00019999980669278125,
-      "loss": 46.0,
-      "step": 8196
-    },
-    {
-      "epoch": 0.6267178928455378,
-      "grad_norm": 0.003642122261226177,
-      "learning_rate": 0.00019999980664554964,
-      "loss": 46.0,
-      "step": 8197
-    },
-    {
-      "epoch": 0.6267943498289275,
-      "grad_norm": 0.0020308259408921003,
-      "learning_rate": 0.00019999980659831228,
-      "loss": 46.0,
-      "step": 8198
-    },
-    {
-      "epoch": 0.6268708068123172,
-      "grad_norm": 0.004944133572280407,
-      "learning_rate": 0.00019999980655106914,
-      "loss": 46.0,
-      "step": 8199
-    },
-    {
-      "epoch": 0.626947263795707,
-      "grad_norm": 0.0009574219002388418,
-      "learning_rate": 0.0001999998065038202,
-      "loss": 46.0,
-      "step": 8200
-    },
-    {
-      "epoch": 0.6270237207790966,
-      "grad_norm": 0.001406748779118061,
-      "learning_rate": 0.00019999980645656553,
-      "loss": 46.0,
-      "step": 8201
-    },
-    {
-      "epoch": 0.6271001777624864,
-      "grad_norm": 0.002005796879529953,
-      "learning_rate": 0.00019999980640930508,
-      "loss": 46.0,
-      "step": 8202
-    },
-    {
-      "epoch": 0.6271766347458761,
-      "grad_norm": 0.002738900948315859,
-      "learning_rate": 0.00019999980636203885,
-      "loss": 46.0,
-      "step": 8203
-    },
-    {
-      "epoch": 0.6272530917292658,
-      "grad_norm": 0.0008467985899187624,
-      "learning_rate": 0.00019999980631476687,
-      "loss": 46.0,
-      "step": 8204
-    },
-    {
-      "epoch": 0.6273295487126556,
-      "grad_norm": 0.004121803212910891,
-      "learning_rate": 0.0001999998062674891,
-      "loss": 46.0,
-      "step": 8205
-    },
-    {
-      "epoch": 0.6274060056960452,
-      "grad_norm": 0.002115719486027956,
-      "learning_rate": 0.00019999980622020558,
-      "loss": 46.0,
-      "step": 8206
-    },
-    {
-      "epoch": 0.627482462679435,
-      "grad_norm": 0.0015576917212456465,
-      "learning_rate": 0.00019999980617291626,
-      "loss": 46.0,
-      "step": 8207
-    },
-    {
-      "epoch": 0.6275589196628247,
-      "grad_norm": 0.0008969299378804862,
-      "learning_rate": 0.0001999998061256212,
-      "loss": 46.0,
-      "step": 8208
-    },
-    {
-      "epoch": 0.6276353766462144,
-      "grad_norm": 0.000680507393553853,
-      "learning_rate": 0.00019999980607832035,
-      "loss": 46.0,
-      "step": 8209
-    },
-    {
-      "epoch": 0.6277118336296041,
-      "grad_norm": 0.0009740458917804062,
-      "learning_rate": 0.00019999980603101376,
-      "loss": 46.0,
-      "step": 8210
-    },
-    {
-      "epoch": 0.6277882906129939,
-      "grad_norm": 0.0013327578781172633,
-      "learning_rate": 0.00019999980598370138,
-      "loss": 46.0,
-      "step": 8211
-    },
-    {
-      "epoch": 0.6278647475963836,
-      "grad_norm": 0.01327575370669365,
-      "learning_rate": 0.00019999980593638324,
-      "loss": 46.0,
-      "step": 8212
-    },
-    {
-      "epoch": 0.6279412045797733,
-      "grad_norm": 0.0007239340338855982,
-      "learning_rate": 0.0001999998058890593,
-      "loss": 46.0,
-      "step": 8213
-    },
-    {
-      "epoch": 0.6280176615631631,
-      "grad_norm": 0.0015642091166228056,
-      "learning_rate": 0.0001999998058417296,
-      "loss": 46.0,
-      "step": 8214
-    },
-    {
-      "epoch": 0.6280941185465527,
-      "grad_norm": 0.0007748593343421817,
-      "learning_rate": 0.00019999980579439415,
-      "loss": 46.0,
-      "step": 8215
-    },
-    {
-      "epoch": 0.6281705755299425,
-      "grad_norm": 0.007889986969530582,
-      "learning_rate": 0.00019999980574705295,
-      "loss": 46.0,
-      "step": 8216
-    },
-    {
-      "epoch": 0.6282470325133321,
-      "grad_norm": 0.0004565949202515185,
-      "learning_rate": 0.00019999980569970592,
-      "loss": 46.0,
-      "step": 8217
-    },
-    {
-      "epoch": 0.6283234894967219,
-      "grad_norm": 0.0015378898242488503,
-      "learning_rate": 0.0001999998056523532,
-      "loss": 46.0,
-      "step": 8218
-    },
-    {
-      "epoch": 0.6283999464801117,
-      "grad_norm": 0.000738327915314585,
-      "learning_rate": 0.00019999980560499463,
-      "loss": 46.0,
-      "step": 8219
-    },
-    {
-      "epoch": 0.6284764034635013,
-      "grad_norm": 0.0005164701724424958,
-      "learning_rate": 0.00019999980555763033,
-      "loss": 46.0,
-      "step": 8220
-    },
-    {
-      "epoch": 0.6285528604468911,
-      "grad_norm": 0.0018236987525597215,
-      "learning_rate": 0.00019999980551026027,
-      "loss": 46.0,
-      "step": 8221
-    },
-    {
-      "epoch": 0.6286293174302808,
-      "grad_norm": 0.000812305836006999,
-      "learning_rate": 0.00019999980546288443,
-      "loss": 46.0,
-      "step": 8222
-    },
-    {
-      "epoch": 0.6287057744136705,
-      "grad_norm": 0.0007858831668272614,
-      "learning_rate": 0.00019999980541550284,
-      "loss": 46.0,
-      "step": 8223
-    },
-    {
-      "epoch": 0.6287822313970602,
-      "grad_norm": 0.00037164639797993004,
-      "learning_rate": 0.00019999980536811545,
-      "loss": 46.0,
-      "step": 8224
-    },
-    {
-      "epoch": 0.62885868838045,
-      "grad_norm": 0.0006847066688351333,
-      "learning_rate": 0.0001999998053207223,
-      "loss": 46.0,
-      "step": 8225
-    },
-    {
-      "epoch": 0.6289351453638397,
-      "grad_norm": 0.002156647387892008,
-      "learning_rate": 0.00019999980527332336,
-      "loss": 46.0,
-      "step": 8226
-    },
-    {
-      "epoch": 0.6290116023472294,
-      "grad_norm": 0.0020973009523004293,
-      "learning_rate": 0.00019999980522591865,
-      "loss": 46.0,
-      "step": 8227
-    },
-    {
-      "epoch": 0.6290880593306191,
-      "grad_norm": 0.0009336823131889105,
-      "learning_rate": 0.00019999980517850823,
-      "loss": 46.0,
-      "step": 8228
-    },
-    {
-      "epoch": 0.6291645163140088,
-      "grad_norm": 0.002602536929771304,
-      "learning_rate": 0.00019999980513109197,
-      "loss": 46.0,
-      "step": 8229
-    },
-    {
-      "epoch": 0.6292409732973986,
-      "grad_norm": 0.0010431433329358697,
-      "learning_rate": 0.00019999980508367,
-      "loss": 46.0,
-      "step": 8230
-    },
-    {
-      "epoch": 0.6293174302807882,
-      "grad_norm": 0.002886660862714052,
-      "learning_rate": 0.0001999998050362422,
-      "loss": 46.0,
-      "step": 8231
-    },
-    {
-      "epoch": 0.629393887264178,
-      "grad_norm": 0.0011985368328168988,
-      "learning_rate": 0.00019999980498880868,
-      "loss": 46.0,
-      "step": 8232
-    },
-    {
-      "epoch": 0.6294703442475678,
-      "grad_norm": 0.004041264299303293,
-      "learning_rate": 0.00019999980494136938,
-      "loss": 46.0,
-      "step": 8233
-    },
-    {
-      "epoch": 0.6295468012309574,
-      "grad_norm": 0.0075932457111775875,
-      "learning_rate": 0.00019999980489392432,
-      "loss": 46.0,
-      "step": 8234
-    },
-    {
-      "epoch": 0.6296232582143472,
-      "grad_norm": 0.0013405568897724152,
-      "learning_rate": 0.00019999980484647348,
-      "loss": 46.0,
-      "step": 8235
-    },
-    {
-      "epoch": 0.6296997151977368,
-      "grad_norm": 0.001531782210804522,
-      "learning_rate": 0.00019999980479901687,
-      "loss": 46.0,
-      "step": 8236
-    },
-    {
-      "epoch": 0.6297761721811266,
-      "grad_norm": 0.0022413202095776796,
-      "learning_rate": 0.00019999980475155445,
-      "loss": 46.0,
-      "step": 8237
-    },
-    {
-      "epoch": 0.6298526291645163,
-      "grad_norm": 0.0007129863370209932,
-      "learning_rate": 0.0001999998047040863,
-      "loss": 46.0,
-      "step": 8238
-    },
-    {
-      "epoch": 0.629929086147906,
-      "grad_norm": 0.0027562961913645267,
-      "learning_rate": 0.00019999980465661239,
-      "loss": 46.0,
-      "step": 8239
-    },
-    {
-      "epoch": 0.6300055431312958,
-      "grad_norm": 0.002757111331447959,
-      "learning_rate": 0.0001999998046091327,
-      "loss": 46.0,
-      "step": 8240
-    },
-    {
-      "epoch": 0.6300820001146855,
-      "grad_norm": 0.0022866460494697094,
-      "learning_rate": 0.00019999980456164723,
-      "loss": 46.0,
-      "step": 8241
-    },
-    {
-      "epoch": 0.6301584570980752,
-      "grad_norm": 0.0006270198500715196,
-      "learning_rate": 0.000199999804514156,
-      "loss": 46.0,
-      "step": 8242
-    },
-    {
-      "epoch": 0.6302349140814649,
-      "grad_norm": 0.0008527711615897715,
-      "learning_rate": 0.000199999804466659,
-      "loss": 46.0,
-      "step": 8243
-    },
-    {
-      "epoch": 0.6303113710648547,
-      "grad_norm": 0.0011923557613044977,
-      "learning_rate": 0.00019999980441915623,
-      "loss": 46.0,
-      "step": 8244
-    },
-    {
-      "epoch": 0.6303878280482443,
-      "grad_norm": 0.0007259712438099086,
-      "learning_rate": 0.00019999980437164768,
-      "loss": 46.0,
-      "step": 8245
-    },
-    {
-      "epoch": 0.6304642850316341,
-      "grad_norm": 0.0030345323029905558,
-      "learning_rate": 0.00019999980432413339,
-      "loss": 46.0,
-      "step": 8246
-    },
-    {
-      "epoch": 0.6305407420150237,
-      "grad_norm": 0.0038378837052732706,
-      "learning_rate": 0.00019999980427661332,
-      "loss": 46.0,
-      "step": 8247
-    },
-    {
-      "epoch": 0.6306171989984135,
-      "grad_norm": 0.002348220907151699,
-      "learning_rate": 0.00019999980422908745,
-      "loss": 46.0,
-      "step": 8248
-    },
-    {
-      "epoch": 0.6306936559818033,
-      "grad_norm": 0.0005703881615772843,
-      "learning_rate": 0.00019999980418155584,
-      "loss": 46.0,
-      "step": 8249
-    },
-    {
-      "epoch": 0.6307701129651929,
-      "grad_norm": 0.0022459183819592,
-      "learning_rate": 0.00019999980413401845,
-      "loss": 46.0,
-      "step": 8250
-    },
-    {
-      "epoch": 0.6308465699485827,
-      "grad_norm": 0.001081667491234839,
-      "learning_rate": 0.0001999998040864753,
-      "loss": 46.0,
-      "step": 8251
-    },
-    {
-      "epoch": 0.6309230269319724,
-      "grad_norm": 0.0026280456222593784,
-      "learning_rate": 0.0001999998040389264,
-      "loss": 46.0,
-      "step": 8252
-    },
-    {
-      "epoch": 0.6309994839153621,
-      "grad_norm": 0.0006093858974054456,
-      "learning_rate": 0.00019999980399137165,
-      "loss": 46.0,
-      "step": 8253
-    },
-    {
-      "epoch": 0.6310759408987519,
-      "grad_norm": 0.0010126649867743254,
-      "learning_rate": 0.0001999998039438112,
-      "loss": 46.0,
-      "step": 8254
-    },
-    {
-      "epoch": 0.6311523978821416,
-      "grad_norm": 0.0009760482935234904,
-      "learning_rate": 0.00019999980389624498,
-      "loss": 46.0,
-      "step": 8255
-    },
-    {
-      "epoch": 0.6312288548655313,
-      "grad_norm": 0.00223044422455132,
-      "learning_rate": 0.00019999980384867295,
-      "loss": 46.0,
-      "step": 8256
-    },
-    {
-      "epoch": 0.631305311848921,
-      "grad_norm": 0.0011948707979172468,
-      "learning_rate": 0.0001999998038010952,
-      "loss": 46.0,
-      "step": 8257
-    },
-    {
-      "epoch": 0.6313817688323107,
-      "grad_norm": 0.0007559588411822915,
-      "learning_rate": 0.00019999980375351166,
-      "loss": 46.0,
-      "step": 8258
-    },
-    {
-      "epoch": 0.6314582258157004,
-      "grad_norm": 0.0055906991474330425,
-      "learning_rate": 0.00019999980370592234,
-      "loss": 46.0,
-      "step": 8259
-    },
-    {
-      "epoch": 0.6315346827990902,
-      "grad_norm": 0.0008051712648011744,
-      "learning_rate": 0.00019999980365832727,
-      "loss": 46.0,
-      "step": 8260
-    },
-    {
-      "epoch": 0.6316111397824798,
-      "grad_norm": 0.0005880105891264975,
-      "learning_rate": 0.0001999998036107264,
-      "loss": 46.0,
-      "step": 8261
-    },
-    {
-      "epoch": 0.6316875967658696,
-      "grad_norm": 0.0015040229773148894,
-      "learning_rate": 0.0001999998035631198,
-      "loss": 46.0,
-      "step": 8262
-    },
-    {
-      "epoch": 0.6317640537492594,
-      "grad_norm": 0.0010844470234587789,
-      "learning_rate": 0.00019999980351550743,
-      "loss": 46.0,
-      "step": 8263
-    },
-    {
-      "epoch": 0.631840510732649,
-      "grad_norm": 0.0018527957145124674,
-      "learning_rate": 0.00019999980346788925,
-      "loss": 46.0,
-      "step": 8264
-    },
-    {
-      "epoch": 0.6319169677160388,
-      "grad_norm": 0.0013868644600734115,
-      "learning_rate": 0.00019999980342026531,
-      "loss": 46.0,
-      "step": 8265
-    },
-    {
-      "epoch": 0.6319934246994284,
-      "grad_norm": 0.0009188971016556025,
-      "learning_rate": 0.0001999998033726356,
-      "loss": 46.0,
-      "step": 8266
-    },
-    {
-      "epoch": 0.6320698816828182,
-      "grad_norm": 0.0010394947603344917,
-      "learning_rate": 0.00019999980332500016,
-      "loss": 46.0,
-      "step": 8267
-    },
-    {
-      "epoch": 0.632146338666208,
-      "grad_norm": 0.0007365021738223732,
-      "learning_rate": 0.00019999980327735893,
-      "loss": 46.0,
-      "step": 8268
-    },
-    {
-      "epoch": 0.6322227956495976,
-      "grad_norm": 0.0018793733324855566,
-      "learning_rate": 0.0001999998032297119,
-      "loss": 46.0,
-      "step": 8269
-    },
-    {
-      "epoch": 0.6322992526329874,
-      "grad_norm": 0.0008051790064200759,
-      "learning_rate": 0.00019999980318205913,
-      "loss": 46.0,
-      "step": 8270
-    },
-    {
-      "epoch": 0.6323757096163771,
-      "grad_norm": 0.000983710866421461,
-      "learning_rate": 0.00019999980313440059,
-      "loss": 46.0,
-      "step": 8271
-    },
-    {
-      "epoch": 0.6324521665997668,
-      "grad_norm": 0.0017727516824379563,
-      "learning_rate": 0.00019999980308673627,
-      "loss": 46.0,
-      "step": 8272
-    },
-    {
-      "epoch": 0.6325286235831565,
-      "grad_norm": 0.0009731962345540524,
-      "learning_rate": 0.00019999980303906617,
-      "loss": 46.0,
-      "step": 8273
-    },
-    {
-      "epoch": 0.6326050805665463,
-      "grad_norm": 0.0012013515224680305,
-      "learning_rate": 0.00019999980299139033,
-      "loss": 46.0,
-      "step": 8274
-    },
-    {
-      "epoch": 0.632681537549936,
-      "grad_norm": 0.0020780726335942745,
-      "learning_rate": 0.00019999980294370872,
-      "loss": 46.0,
-      "step": 8275
-    },
-    {
-      "epoch": 0.6327579945333257,
-      "grad_norm": 0.008914541453123093,
-      "learning_rate": 0.0001999998028960213,
-      "loss": 46.0,
-      "step": 8276
-    },
-    {
-      "epoch": 0.6328344515167154,
-      "grad_norm": 0.0010731208603829145,
-      "learning_rate": 0.00019999980284832818,
-      "loss": 46.0,
-      "step": 8277
-    },
-    {
-      "epoch": 0.6329109085001051,
-      "grad_norm": 0.00032802208443172276,
-      "learning_rate": 0.00019999980280062922,
-      "loss": 46.0,
-      "step": 8278
-    },
-    {
-      "epoch": 0.6329873654834949,
-      "grad_norm": 0.004165690392255783,
-      "learning_rate": 0.00019999980275292454,
-      "loss": 46.0,
-      "step": 8279
-    },
-    {
-      "epoch": 0.6330638224668845,
-      "grad_norm": 0.0015048672212287784,
-      "learning_rate": 0.0001999998027052141,
-      "loss": 46.0,
-      "step": 8280
-    },
-    {
-      "epoch": 0.6331402794502743,
-      "grad_norm": 0.0012001678114756942,
-      "learning_rate": 0.00019999980265749783,
-      "loss": 46.0,
-      "step": 8281
-    },
-    {
-      "epoch": 0.633216736433664,
-      "grad_norm": 0.0010596668580546975,
-      "learning_rate": 0.00019999980260977584,
-      "loss": 46.0,
-      "step": 8282
-    },
-    {
-      "epoch": 0.6332931934170537,
-      "grad_norm": 0.0011254140408709645,
-      "learning_rate": 0.000199999802562048,
-      "loss": 46.0,
-      "step": 8283
-    },
-    {
-      "epoch": 0.6333696504004435,
-      "grad_norm": 0.0038946825079619884,
-      "learning_rate": 0.0001999998025143145,
-      "loss": 46.0,
-      "step": 8284
-    },
-    {
-      "epoch": 0.6334461073838332,
-      "grad_norm": 0.000450453459052369,
-      "learning_rate": 0.00019999980246657517,
-      "loss": 46.0,
-      "step": 8285
-    },
-    {
-      "epoch": 0.6335225643672229,
-      "grad_norm": 0.0007265485473908484,
-      "learning_rate": 0.0001999998024188301,
-      "loss": 46.0,
-      "step": 8286
-    },
-    {
-      "epoch": 0.6335990213506126,
-      "grad_norm": 0.001173571334220469,
-      "learning_rate": 0.0001999998023710792,
-      "loss": 46.0,
-      "step": 8287
-    },
-    {
-      "epoch": 0.6336754783340023,
-      "grad_norm": 0.000545788963790983,
-      "learning_rate": 0.0001999998023233226,
-      "loss": 46.0,
-      "step": 8288
-    },
-    {
-      "epoch": 0.633751935317392,
-      "grad_norm": 0.0032594939693808556,
-      "learning_rate": 0.00019999980227556022,
-      "loss": 46.0,
-      "step": 8289
-    },
-    {
-      "epoch": 0.6338283923007818,
-      "grad_norm": 0.003350586863234639,
-      "learning_rate": 0.00019999980222779206,
-      "loss": 46.0,
-      "step": 8290
-    },
-    {
-      "epoch": 0.6339048492841715,
-      "grad_norm": 0.0007236524834297597,
-      "learning_rate": 0.0001999998021800181,
-      "loss": 46.0,
-      "step": 8291
-    },
-    {
-      "epoch": 0.6339813062675612,
-      "grad_norm": 0.0021102663595229387,
-      "learning_rate": 0.00019999980213223842,
-      "loss": 46.0,
-      "step": 8292
-    },
-    {
-      "epoch": 0.634057763250951,
-      "grad_norm": 0.0012011940125375986,
-      "learning_rate": 0.00019999980208445294,
-      "loss": 46.0,
-      "step": 8293
-    },
-    {
-      "epoch": 0.6341342202343406,
-      "grad_norm": 0.0018167115049436688,
-      "learning_rate": 0.00019999980203666171,
-      "loss": 46.0,
-      "step": 8294
-    },
-    {
-      "epoch": 0.6342106772177304,
-      "grad_norm": 0.0015411480562761426,
-      "learning_rate": 0.0001999998019888647,
-      "loss": 46.0,
-      "step": 8295
-    },
-    {
-      "epoch": 0.63428713420112,
-      "grad_norm": 0.0023441424127668142,
-      "learning_rate": 0.00019999980194106192,
-      "loss": 46.0,
-      "step": 8296
-    },
-    {
-      "epoch": 0.6343635911845098,
-      "grad_norm": 0.0012845075689256191,
-      "learning_rate": 0.00019999980189325334,
-      "loss": 46.0,
-      "step": 8297
-    },
-    {
-      "epoch": 0.6344400481678996,
-      "grad_norm": 0.0010803508339449763,
-      "learning_rate": 0.00019999980184543903,
-      "loss": 46.0,
-      "step": 8298
-    },
-    {
-      "epoch": 0.6345165051512892,
-      "grad_norm": 0.002772475127130747,
-      "learning_rate": 0.00019999980179761893,
-      "loss": 46.0,
-      "step": 8299
-    },
-    {
-      "epoch": 0.634592962134679,
-      "grad_norm": 0.0005241383914835751,
-      "learning_rate": 0.0001999998017497931,
-      "loss": 46.0,
-      "step": 8300
-    },
-    {
-      "epoch": 0.6346694191180687,
-      "grad_norm": 0.001759040867909789,
-      "learning_rate": 0.00019999980170196143,
-      "loss": 46.0,
-      "step": 8301
-    },
-    {
-      "epoch": 0.6347458761014584,
-      "grad_norm": 0.0011967982864007354,
-      "learning_rate": 0.00019999980165412407,
-      "loss": 46.0,
-      "step": 8302
-    },
-    {
-      "epoch": 0.6348223330848481,
-      "grad_norm": 0.0032175721134990454,
-      "learning_rate": 0.00019999980160628089,
-      "loss": 46.0,
-      "step": 8303
-    },
-    {
-      "epoch": 0.6348987900682379,
-      "grad_norm": 0.0010704339947551489,
-      "learning_rate": 0.00019999980155843195,
-      "loss": 46.0,
-      "step": 8304
-    },
-    {
-      "epoch": 0.6349752470516276,
-      "grad_norm": 0.01669190265238285,
-      "learning_rate": 0.00019999980151057725,
-      "loss": 46.0,
-      "step": 8305
-    },
-    {
-      "epoch": 0.6350517040350173,
-      "grad_norm": 0.008821866475045681,
-      "learning_rate": 0.00019999980146271677,
-      "loss": 46.0,
-      "step": 8306
-    },
-    {
-      "epoch": 0.635128161018407,
-      "grad_norm": 0.001014237990602851,
-      "learning_rate": 0.00019999980141485052,
-      "loss": 46.0,
-      "step": 8307
-    },
-    {
-      "epoch": 0.6352046180017967,
-      "grad_norm": 0.000294274854240939,
-      "learning_rate": 0.00019999980136697852,
-      "loss": 46.0,
-      "step": 8308
-    },
-    {
-      "epoch": 0.6352810749851865,
-      "grad_norm": 0.004478603135794401,
-      "learning_rate": 0.00019999980131910072,
-      "loss": 46.0,
-      "step": 8309
-    },
-    {
-      "epoch": 0.6353575319685761,
-      "grad_norm": 0.002247787779197097,
-      "learning_rate": 0.00019999980127121718,
-      "loss": 46.0,
-      "step": 8310
-    },
-    {
-      "epoch": 0.6354339889519659,
-      "grad_norm": 0.0011244042543694377,
-      "learning_rate": 0.00019999980122332786,
-      "loss": 46.0,
-      "step": 8311
-    },
-    {
-      "epoch": 0.6355104459353557,
-      "grad_norm": 0.0011448069708421826,
-      "learning_rate": 0.00019999980117543277,
-      "loss": 46.0,
-      "step": 8312
-    },
-    {
-      "epoch": 0.6355869029187453,
-      "grad_norm": 0.0018306816928088665,
-      "learning_rate": 0.00019999980112753193,
-      "loss": 46.0,
-      "step": 8313
-    },
-    {
-      "epoch": 0.6356633599021351,
-      "grad_norm": 0.0006840021233074367,
-      "learning_rate": 0.00019999980107962526,
-      "loss": 46.0,
-      "step": 8314
-    },
-    {
-      "epoch": 0.6357398168855248,
-      "grad_norm": 0.0007227238384075463,
-      "learning_rate": 0.00019999980103171288,
-      "loss": 46.0,
-      "step": 8315
-    },
-    {
-      "epoch": 0.6358162738689145,
-      "grad_norm": 0.0007279314449988306,
-      "learning_rate": 0.00019999980098379472,
-      "loss": 46.0,
-      "step": 8316
-    },
-    {
-      "epoch": 0.6358927308523042,
-      "grad_norm": 0.004097754135727882,
-      "learning_rate": 0.00019999980093587076,
-      "loss": 46.0,
-      "step": 8317
-    },
-    {
-      "epoch": 0.6359691878356939,
-      "grad_norm": 0.001811531838029623,
-      "learning_rate": 0.00019999980088794106,
-      "loss": 46.0,
-      "step": 8318
-    },
-    {
-      "epoch": 0.6360456448190837,
-      "grad_norm": 0.0006689854781143367,
-      "learning_rate": 0.00019999980084000558,
-      "loss": 46.0,
-      "step": 8319
-    },
-    {
-      "epoch": 0.6361221018024734,
-      "grad_norm": 0.005581785459071398,
-      "learning_rate": 0.00019999980079206433,
-      "loss": 46.0,
-      "step": 8320
-    },
-    {
-      "epoch": 0.6361985587858631,
-      "grad_norm": 0.0024075352121144533,
-      "learning_rate": 0.00019999980074411733,
-      "loss": 46.0,
-      "step": 8321
-    },
-    {
-      "epoch": 0.6362750157692528,
-      "grad_norm": 0.005853780545294285,
-      "learning_rate": 0.00019999980069616453,
-      "loss": 46.0,
-      "step": 8322
-    },
-    {
-      "epoch": 0.6363514727526426,
-      "grad_norm": 0.0012639662018045783,
-      "learning_rate": 0.00019999980064820596,
-      "loss": 46.0,
-      "step": 8323
-    },
-    {
-      "epoch": 0.6364279297360322,
-      "grad_norm": 0.0008265635115094483,
-      "learning_rate": 0.00019999980060024164,
-      "loss": 46.0,
-      "step": 8324
-    },
-    {
-      "epoch": 0.636504386719422,
-      "grad_norm": 0.0022197996731847525,
-      "learning_rate": 0.00019999980055227155,
-      "loss": 46.0,
-      "step": 8325
-    },
-    {
-      "epoch": 0.6365808437028118,
-      "grad_norm": 0.0010472809663042426,
-      "learning_rate": 0.0001999998005042957,
-      "loss": 46.0,
-      "step": 8326
-    },
-    {
-      "epoch": 0.6366573006862014,
-      "grad_norm": 0.0014795337338000536,
-      "learning_rate": 0.00019999980045631405,
-      "loss": 46.0,
-      "step": 8327
-    },
-    {
-      "epoch": 0.6367337576695912,
-      "grad_norm": 0.0063463435508310795,
-      "learning_rate": 0.00019999980040832663,
-      "loss": 46.0,
-      "step": 8328
-    },
-    {
-      "epoch": 0.6368102146529808,
-      "grad_norm": 0.0015335968928411603,
-      "learning_rate": 0.00019999980036033348,
-      "loss": 46.0,
-      "step": 8329
-    },
-    {
-      "epoch": 0.6368866716363706,
-      "grad_norm": 0.00292824674397707,
-      "learning_rate": 0.00019999980031233455,
-      "loss": 46.0,
-      "step": 8330
-    },
-    {
-      "epoch": 0.6369631286197603,
-      "grad_norm": 0.002795613370835781,
-      "learning_rate": 0.00019999980026432984,
-      "loss": 46.0,
-      "step": 8331
-    },
-    {
-      "epoch": 0.63703958560315,
-      "grad_norm": 0.002934791147708893,
-      "learning_rate": 0.00019999980021631934,
-      "loss": 46.0,
-      "step": 8332
-    },
-    {
-      "epoch": 0.6371160425865398,
-      "grad_norm": 0.00047825308865867555,
-      "learning_rate": 0.00019999980016830308,
-      "loss": 46.0,
-      "step": 8333
-    },
-    {
-      "epoch": 0.6371924995699295,
-      "grad_norm": 0.0004855692677665502,
-      "learning_rate": 0.00019999980012028106,
-      "loss": 46.0,
-      "step": 8334
-    },
-    {
-      "epoch": 0.6372689565533192,
-      "grad_norm": 0.0018347181612625718,
-      "learning_rate": 0.0001999998000722533,
-      "loss": 46.0,
-      "step": 8335
-    },
-    {
-      "epoch": 0.6373454135367089,
-      "grad_norm": 0.0009170413250103593,
-      "learning_rate": 0.00019999980002421972,
-      "loss": 46.0,
-      "step": 8336
-    },
-    {
-      "epoch": 0.6374218705200986,
-      "grad_norm": 0.0011994520900771022,
-      "learning_rate": 0.0001999997999761804,
-      "loss": 46.0,
-      "step": 8337
-    },
-    {
-      "epoch": 0.6374983275034883,
-      "grad_norm": 0.0035632005892693996,
-      "learning_rate": 0.00019999979992813528,
-      "loss": 46.0,
-      "step": 8338
-    },
-    {
-      "epoch": 0.6375747844868781,
-      "grad_norm": 0.0035316585563123226,
-      "learning_rate": 0.00019999979988008444,
-      "loss": 46.0,
-      "step": 8339
-    },
-    {
-      "epoch": 0.6376512414702677,
-      "grad_norm": 0.001511100446805358,
-      "learning_rate": 0.00019999979983202778,
-      "loss": 46.0,
-      "step": 8340
-    },
-    {
-      "epoch": 0.6377276984536575,
-      "grad_norm": 0.0014686223585158587,
-      "learning_rate": 0.0001999997997839654,
-      "loss": 46.0,
-      "step": 8341
-    },
-    {
-      "epoch": 0.6378041554370473,
-      "grad_norm": 0.0008743414655327797,
-      "learning_rate": 0.0001999997997358972,
-      "loss": 46.0,
-      "step": 8342
-    },
-    {
-      "epoch": 0.6378806124204369,
-      "grad_norm": 0.0006099130841903389,
-      "learning_rate": 0.00019999979968782325,
-      "loss": 46.0,
-      "step": 8343
-    },
-    {
-      "epoch": 0.6379570694038267,
-      "grad_norm": 0.001428297022357583,
-      "learning_rate": 0.00019999979963974355,
-      "loss": 46.0,
-      "step": 8344
-    },
-    {
-      "epoch": 0.6380335263872164,
-      "grad_norm": 0.0022314409725368023,
-      "learning_rate": 0.00019999979959165807,
-      "loss": 46.0,
-      "step": 8345
-    },
-    {
-      "epoch": 0.6381099833706061,
-      "grad_norm": 0.0009802368003875017,
-      "learning_rate": 0.00019999979954356682,
-      "loss": 46.0,
-      "step": 8346
-    },
-    {
-      "epoch": 0.6381864403539959,
-      "grad_norm": 0.0011869882000610232,
-      "learning_rate": 0.0001999997994954698,
-      "loss": 46.0,
-      "step": 8347
-    },
-    {
-      "epoch": 0.6382628973373855,
-      "grad_norm": 0.002710545901209116,
-      "learning_rate": 0.000199999799447367,
-      "loss": 46.0,
-      "step": 8348
-    },
-    {
-      "epoch": 0.6383393543207753,
-      "grad_norm": 0.0013702786527574062,
-      "learning_rate": 0.00019999979939925843,
-      "loss": 46.0,
-      "step": 8349
-    },
-    {
-      "epoch": 0.638415811304165,
-      "grad_norm": 0.005586983636021614,
-      "learning_rate": 0.00019999979935114414,
-      "loss": 46.0,
-      "step": 8350
-    },
-    {
-      "epoch": 0.6384922682875547,
-      "grad_norm": 0.001324632903560996,
-      "learning_rate": 0.00019999979930302402,
-      "loss": 46.0,
-      "step": 8351
-    },
-    {
-      "epoch": 0.6385687252709444,
-      "grad_norm": 0.0011119857663288713,
-      "learning_rate": 0.00019999979925489813,
-      "loss": 46.0,
-      "step": 8352
-    },
-    {
-      "epoch": 0.6386451822543342,
-      "grad_norm": 0.001104339025914669,
-      "learning_rate": 0.00019999979920676652,
-      "loss": 46.0,
-      "step": 8353
-    },
-    {
-      "epoch": 0.6387216392377238,
-      "grad_norm": 0.002143976278603077,
-      "learning_rate": 0.0001999997991586291,
-      "loss": 46.0,
-      "step": 8354
-    },
-    {
-      "epoch": 0.6387980962211136,
-      "grad_norm": 0.006432107649743557,
-      "learning_rate": 0.00019999979911048592,
-      "loss": 46.0,
-      "step": 8355
-    },
-    {
-      "epoch": 0.6388745532045034,
-      "grad_norm": 0.00257153925485909,
-      "learning_rate": 0.00019999979906233696,
-      "loss": 46.0,
-      "step": 8356
-    },
-    {
-      "epoch": 0.638951010187893,
-      "grad_norm": 0.0016004828503355384,
-      "learning_rate": 0.00019999979901418223,
-      "loss": 46.0,
-      "step": 8357
-    },
-    {
-      "epoch": 0.6390274671712828,
-      "grad_norm": 0.0038370173424482346,
-      "learning_rate": 0.00019999979896602175,
-      "loss": 46.0,
-      "step": 8358
-    },
-    {
-      "epoch": 0.6391039241546724,
-      "grad_norm": 0.0010702681029215455,
-      "learning_rate": 0.00019999979891785553,
-      "loss": 46.0,
-      "step": 8359
-    },
-    {
-      "epoch": 0.6391803811380622,
-      "grad_norm": 0.0013894864823669195,
-      "learning_rate": 0.00019999979886968348,
-      "loss": 46.0,
-      "step": 8360
-    },
-    {
-      "epoch": 0.639256838121452,
-      "grad_norm": 0.000811832258477807,
-      "learning_rate": 0.0001999997988215057,
-      "loss": 46.0,
-      "step": 8361
-    },
-    {
-      "epoch": 0.6393332951048416,
-      "grad_norm": 0.003757540602236986,
-      "learning_rate": 0.0001999997987733221,
-      "loss": 46.0,
-      "step": 8362
-    },
-    {
-      "epoch": 0.6394097520882314,
-      "grad_norm": 0.00047705232282169163,
-      "learning_rate": 0.0001999997987251328,
-      "loss": 46.0,
-      "step": 8363
-    },
-    {
-      "epoch": 0.6394862090716211,
-      "grad_norm": 0.0007783086621202528,
-      "learning_rate": 0.0001999997986769377,
-      "loss": 46.0,
-      "step": 8364
-    },
-    {
-      "epoch": 0.6395626660550108,
-      "grad_norm": 0.0009130511316470802,
-      "learning_rate": 0.0001999997986287368,
-      "loss": 46.0,
-      "step": 8365
-    },
-    {
-      "epoch": 0.6396391230384005,
-      "grad_norm": 0.0010191791225224733,
-      "learning_rate": 0.00019999979858053018,
-      "loss": 46.0,
-      "step": 8366
-    },
-    {
-      "epoch": 0.6397155800217902,
-      "grad_norm": 0.0006976840668357909,
-      "learning_rate": 0.00019999979853231776,
-      "loss": 46.0,
-      "step": 8367
-    },
-    {
-      "epoch": 0.63979203700518,
-      "grad_norm": 0.0009047634666785598,
-      "learning_rate": 0.00019999979848409958,
-      "loss": 46.0,
-      "step": 8368
-    },
-    {
-      "epoch": 0.6398684939885697,
-      "grad_norm": 0.002331351861357689,
-      "learning_rate": 0.00019999979843587565,
-      "loss": 46.0,
-      "step": 8369
-    },
-    {
-      "epoch": 0.6399449509719594,
-      "grad_norm": 0.008381582796573639,
-      "learning_rate": 0.00019999979838764592,
-      "loss": 46.0,
-      "step": 8370
-    },
-    {
-      "epoch": 0.6400214079553491,
-      "grad_norm": 0.001435493933968246,
-      "learning_rate": 0.00019999979833941042,
-      "loss": 46.0,
-      "step": 8371
-    },
-    {
-      "epoch": 0.6400978649387389,
-      "grad_norm": 0.0012833019718527794,
-      "learning_rate": 0.00019999979829116916,
-      "loss": 46.0,
-      "step": 8372
-    },
-    {
-      "epoch": 0.6401743219221285,
-      "grad_norm": 0.0005923905991949141,
-      "learning_rate": 0.00019999979824292211,
-      "loss": 46.0,
-      "step": 8373
-    },
-    {
-      "epoch": 0.6402507789055183,
-      "grad_norm": 0.0009728985605761409,
-      "learning_rate": 0.00019999979819466934,
-      "loss": 46.0,
-      "step": 8374
-    },
-    {
-      "epoch": 0.640327235888908,
-      "grad_norm": 0.0011056882794946432,
-      "learning_rate": 0.00019999979814641077,
-      "loss": 46.0,
-      "step": 8375
-    },
-    {
-      "epoch": 0.6404036928722977,
-      "grad_norm": 0.001359409885481,
-      "learning_rate": 0.00019999979809814643,
-      "loss": 46.0,
-      "step": 8376
-    },
-    {
-      "epoch": 0.6404801498556875,
-      "grad_norm": 0.002422122750431299,
-      "learning_rate": 0.0001999997980498763,
-      "loss": 46.0,
-      "step": 8377
-    },
-    {
-      "epoch": 0.6405566068390771,
-      "grad_norm": 0.00031479887547902763,
-      "learning_rate": 0.00019999979800160045,
-      "loss": 46.0,
-      "step": 8378
-    },
-    {
-      "epoch": 0.6406330638224669,
-      "grad_norm": 0.011981228366494179,
-      "learning_rate": 0.0001999997979533188,
-      "loss": 46.0,
-      "step": 8379
-    },
-    {
-      "epoch": 0.6407095208058566,
-      "grad_norm": 0.0006490522646345198,
-      "learning_rate": 0.00019999979790503137,
-      "loss": 46.0,
-      "step": 8380
-    },
-    {
-      "epoch": 0.6407859777892463,
-      "grad_norm": 0.0013353211106732488,
-      "learning_rate": 0.00019999979785673822,
-      "loss": 46.0,
-      "step": 8381
-    },
-    {
-      "epoch": 0.640862434772636,
-      "grad_norm": 0.0013245070585981011,
-      "learning_rate": 0.00019999979780843926,
-      "loss": 46.0,
-      "step": 8382
-    },
-    {
-      "epoch": 0.6409388917560258,
-      "grad_norm": 0.001455620746128261,
-      "learning_rate": 0.00019999979776013453,
-      "loss": 46.0,
-      "step": 8383
-    },
-    {
-      "epoch": 0.6410153487394155,
-      "grad_norm": 0.000765423581469804,
-      "learning_rate": 0.00019999979771182405,
-      "loss": 46.0,
-      "step": 8384
-    },
-    {
-      "epoch": 0.6410918057228052,
-      "grad_norm": 0.0006489277584478259,
-      "learning_rate": 0.00019999979766350778,
-      "loss": 46.0,
-      "step": 8385
-    },
-    {
-      "epoch": 0.641168262706195,
-      "grad_norm": 0.0006296727806329727,
-      "learning_rate": 0.00019999979761518578,
-      "loss": 46.0,
-      "step": 8386
-    },
-    {
-      "epoch": 0.6412447196895846,
-      "grad_norm": 0.012314872816205025,
-      "learning_rate": 0.00019999979756685793,
-      "loss": 46.0,
-      "step": 8387
-    },
-    {
-      "epoch": 0.6413211766729744,
-      "grad_norm": 0.0015245723770931363,
-      "learning_rate": 0.00019999979751852439,
-      "loss": 46.0,
-      "step": 8388
-    },
-    {
-      "epoch": 0.641397633656364,
-      "grad_norm": 0.00026419066125527024,
-      "learning_rate": 0.00019999979747018504,
-      "loss": 46.0,
-      "step": 8389
-    },
-    {
-      "epoch": 0.6414740906397538,
-      "grad_norm": 0.0019564004614949226,
-      "learning_rate": 0.00019999979742183995,
-      "loss": 46.0,
-      "step": 8390
-    },
-    {
-      "epoch": 0.6415505476231436,
-      "grad_norm": 0.0019270264310762286,
-      "learning_rate": 0.00019999979737348906,
-      "loss": 46.0,
-      "step": 8391
-    },
-    {
-      "epoch": 0.6416270046065332,
-      "grad_norm": 0.0012498401338234544,
-      "learning_rate": 0.0001999997973251324,
-      "loss": 46.0,
-      "step": 8392
-    },
-    {
-      "epoch": 0.641703461589923,
-      "grad_norm": 0.0017809005221351981,
-      "learning_rate": 0.00019999979727677,
-      "loss": 46.0,
-      "step": 8393
-    },
-    {
-      "epoch": 0.6417799185733127,
-      "grad_norm": 0.0014072373742237687,
-      "learning_rate": 0.0001999997972284018,
-      "loss": 46.0,
-      "step": 8394
-    },
-    {
-      "epoch": 0.6418563755567024,
-      "grad_norm": 0.001852648681961,
-      "learning_rate": 0.00019999979718002785,
-      "loss": 46.0,
-      "step": 8395
-    },
-    {
-      "epoch": 0.6419328325400921,
-      "grad_norm": 0.0007831387338228524,
-      "learning_rate": 0.00019999979713164812,
-      "loss": 46.0,
-      "step": 8396
-    },
-    {
-      "epoch": 0.6420092895234818,
-      "grad_norm": 0.0017284360947087407,
-      "learning_rate": 0.00019999979708326264,
-      "loss": 46.0,
-      "step": 8397
-    },
-    {
-      "epoch": 0.6420857465068716,
-      "grad_norm": 0.0005251671536825597,
-      "learning_rate": 0.00019999979703487136,
-      "loss": 46.0,
-      "step": 8398
-    },
-    {
-      "epoch": 0.6421622034902613,
-      "grad_norm": 0.0015043087769299746,
-      "learning_rate": 0.00019999979698647434,
-      "loss": 46.0,
-      "step": 8399
-    },
-    {
-      "epoch": 0.642238660473651,
-      "grad_norm": 0.01601600833237171,
-      "learning_rate": 0.00019999979693807155,
-      "loss": 46.0,
-      "step": 8400
-    },
-    {
-      "epoch": 0.6423151174570407,
-      "grad_norm": 0.0008529069600626826,
-      "learning_rate": 0.00019999979688966298,
-      "loss": 46.0,
-      "step": 8401
-    },
-    {
-      "epoch": 0.6423915744404305,
-      "grad_norm": 0.0010558547219261527,
-      "learning_rate": 0.00019999979684124863,
-      "loss": 46.0,
-      "step": 8402
-    },
-    {
-      "epoch": 0.6424680314238201,
-      "grad_norm": 0.001270571257919073,
-      "learning_rate": 0.00019999979679282852,
-      "loss": 46.0,
-      "step": 8403
-    },
-    {
-      "epoch": 0.6425444884072099,
-      "grad_norm": 0.0008635505218990147,
-      "learning_rate": 0.00019999979674440263,
-      "loss": 46.0,
-      "step": 8404
-    },
-    {
-      "epoch": 0.6426209453905997,
-      "grad_norm": 0.0009533761767670512,
-      "learning_rate": 0.000199999796695971,
-      "loss": 46.0,
-      "step": 8405
-    },
-    {
-      "epoch": 0.6426974023739893,
-      "grad_norm": 0.001203278312459588,
-      "learning_rate": 0.00019999979664753355,
-      "loss": 46.0,
-      "step": 8406
-    },
-    {
-      "epoch": 0.6427738593573791,
-      "grad_norm": 0.0006614010781049728,
-      "learning_rate": 0.00019999979659909037,
-      "loss": 46.0,
-      "step": 8407
-    },
-    {
-      "epoch": 0.6428503163407687,
-      "grad_norm": 0.0009135161526501179,
-      "learning_rate": 0.0001999997965506414,
-      "loss": 46.0,
-      "step": 8408
-    },
-    {
-      "epoch": 0.6429267733241585,
-      "grad_norm": 0.0011862055398523808,
-      "learning_rate": 0.00019999979650218668,
-      "loss": 46.0,
-      "step": 8409
-    },
-    {
-      "epoch": 0.6430032303075482,
-      "grad_norm": 0.0013963677920401096,
-      "learning_rate": 0.00019999979645372618,
-      "loss": 46.0,
-      "step": 8410
-    },
-    {
-      "epoch": 0.6430796872909379,
-      "grad_norm": 0.0018271836452186108,
-      "learning_rate": 0.00019999979640525993,
-      "loss": 46.0,
-      "step": 8411
-    },
-    {
-      "epoch": 0.6431561442743277,
-      "grad_norm": 0.0003863652527797967,
-      "learning_rate": 0.0001999997963567879,
-      "loss": 46.0,
-      "step": 8412
-    },
-    {
-      "epoch": 0.6432326012577174,
-      "grad_norm": 0.0006500778836198151,
-      "learning_rate": 0.00019999979630831008,
-      "loss": 46.0,
-      "step": 8413
-    },
-    {
-      "epoch": 0.6433090582411071,
-      "grad_norm": 0.01121943537145853,
-      "learning_rate": 0.0001999997962598265,
-      "loss": 46.0,
-      "step": 8414
-    },
-    {
-      "epoch": 0.6433855152244968,
-      "grad_norm": 0.0007760943844914436,
-      "learning_rate": 0.00019999979621133714,
-      "loss": 46.0,
-      "step": 8415
-    },
-    {
-      "epoch": 0.6434619722078866,
-      "grad_norm": 0.0007321322336792946,
-      "learning_rate": 0.00019999979616284203,
-      "loss": 46.0,
-      "step": 8416
-    },
-    {
-      "epoch": 0.6435384291912762,
-      "grad_norm": 0.0016459973994642496,
-      "learning_rate": 0.00019999979611434117,
-      "loss": 46.0,
-      "step": 8417
-    },
-    {
-      "epoch": 0.643614886174666,
-      "grad_norm": 0.01081672590225935,
-      "learning_rate": 0.0001999997960658345,
-      "loss": 46.0,
-      "step": 8418
-    },
-    {
-      "epoch": 0.6436913431580557,
-      "grad_norm": 0.00067540118470788,
-      "learning_rate": 0.0001999997960173221,
-      "loss": 46.0,
-      "step": 8419
-    },
-    {
-      "epoch": 0.6437678001414454,
-      "grad_norm": 0.0011252046097069979,
-      "learning_rate": 0.00019999979596880388,
-      "loss": 46.0,
-      "step": 8420
-    },
-    {
-      "epoch": 0.6438442571248352,
-      "grad_norm": 0.0007321266457438469,
-      "learning_rate": 0.00019999979592027993,
-      "loss": 46.0,
-      "step": 8421
-    },
-    {
-      "epoch": 0.6439207141082248,
-      "grad_norm": 0.0006748631130903959,
-      "learning_rate": 0.0001999997958717502,
-      "loss": 46.0,
-      "step": 8422
-    },
-    {
-      "epoch": 0.6439971710916146,
-      "grad_norm": 0.0035102139227092266,
-      "learning_rate": 0.00019999979582321472,
-      "loss": 46.0,
-      "step": 8423
-    },
-    {
-      "epoch": 0.6440736280750043,
-      "grad_norm": 0.009357602335512638,
-      "learning_rate": 0.00019999979577467342,
-      "loss": 46.0,
-      "step": 8424
-    },
-    {
-      "epoch": 0.644150085058394,
-      "grad_norm": 0.001554184011183679,
-      "learning_rate": 0.0001999997957261264,
-      "loss": 46.0,
-      "step": 8425
-    },
-    {
-      "epoch": 0.6442265420417838,
-      "grad_norm": 0.0011146211763843894,
-      "learning_rate": 0.00019999979567757357,
-      "loss": 46.0,
-      "step": 8426
-    },
-    {
-      "epoch": 0.6443029990251735,
-      "grad_norm": 0.0004390115791466087,
-      "learning_rate": 0.00019999979562901498,
-      "loss": 46.0,
-      "step": 8427
-    },
-    {
-      "epoch": 0.6443794560085632,
-      "grad_norm": 0.0005093348445370793,
-      "learning_rate": 0.00019999979558045066,
-      "loss": 46.0,
-      "step": 8428
-    },
-    {
-      "epoch": 0.6444559129919529,
-      "grad_norm": 0.0009902993915602565,
-      "learning_rate": 0.00019999979553188054,
-      "loss": 46.0,
-      "step": 8429
-    },
-    {
-      "epoch": 0.6445323699753426,
-      "grad_norm": 0.003410734236240387,
-      "learning_rate": 0.00019999979548330466,
-      "loss": 46.0,
-      "step": 8430
-    },
-    {
-      "epoch": 0.6446088269587323,
-      "grad_norm": 0.001540257129818201,
-      "learning_rate": 0.000199999795434723,
-      "loss": 46.0,
-      "step": 8431
-    },
-    {
-      "epoch": 0.6446852839421221,
-      "grad_norm": 0.0010448395041748881,
-      "learning_rate": 0.00019999979538613558,
-      "loss": 46.0,
-      "step": 8432
-    },
-    {
-      "epoch": 0.6447617409255118,
-      "grad_norm": 0.0009175291052088141,
-      "learning_rate": 0.00019999979533754237,
-      "loss": 46.0,
-      "step": 8433
-    },
-    {
-      "epoch": 0.6448381979089015,
-      "grad_norm": 0.0010905092349275947,
-      "learning_rate": 0.00019999979528894342,
-      "loss": 46.0,
-      "step": 8434
-    },
-    {
-      "epoch": 0.6449146548922913,
-      "grad_norm": 0.0007185473805293441,
-      "learning_rate": 0.0001999997952403387,
-      "loss": 46.0,
-      "step": 8435
-    },
-    {
-      "epoch": 0.6449911118756809,
-      "grad_norm": 0.0008109049522317946,
-      "learning_rate": 0.0001999997951917282,
-      "loss": 46.0,
-      "step": 8436
-    },
-    {
-      "epoch": 0.6450675688590707,
-      "grad_norm": 0.001640122034586966,
-      "learning_rate": 0.00019999979514311188,
-      "loss": 46.0,
-      "step": 8437
-    },
-    {
-      "epoch": 0.6451440258424603,
-      "grad_norm": 0.001342583680525422,
-      "learning_rate": 0.00019999979509448986,
-      "loss": 46.0,
-      "step": 8438
-    },
-    {
-      "epoch": 0.6452204828258501,
-      "grad_norm": 0.0009414669475518167,
-      "learning_rate": 0.00019999979504586204,
-      "loss": 46.0,
-      "step": 8439
-    },
-    {
-      "epoch": 0.6452969398092399,
-      "grad_norm": 0.0004461360222194344,
-      "learning_rate": 0.00019999979499722847,
-      "loss": 46.0,
-      "step": 8440
-    },
-    {
-      "epoch": 0.6453733967926295,
-      "grad_norm": 0.005630668718367815,
-      "learning_rate": 0.00019999979494858913,
-      "loss": 46.0,
-      "step": 8441
-    },
-    {
-      "epoch": 0.6454498537760193,
-      "grad_norm": 0.005121064372360706,
-      "learning_rate": 0.000199999794899944,
-      "loss": 46.0,
-      "step": 8442
-    },
-    {
-      "epoch": 0.645526310759409,
-      "grad_norm": 0.0007692367071285844,
-      "learning_rate": 0.0001999997948512931,
-      "loss": 46.0,
-      "step": 8443
-    },
-    {
-      "epoch": 0.6456027677427987,
-      "grad_norm": 0.000752636871766299,
-      "learning_rate": 0.00019999979480263646,
-      "loss": 46.0,
-      "step": 8444
-    },
-    {
-      "epoch": 0.6456792247261884,
-      "grad_norm": 0.0007875836454331875,
-      "learning_rate": 0.00019999979475397402,
-      "loss": 46.0,
-      "step": 8445
-    },
-    {
-      "epoch": 0.6457556817095782,
-      "grad_norm": 0.0010688817128539085,
-      "learning_rate": 0.00019999979470530584,
-      "loss": 46.0,
-      "step": 8446
-    },
-    {
-      "epoch": 0.6458321386929678,
-      "grad_norm": 0.002124267164617777,
-      "learning_rate": 0.00019999979465663186,
-      "loss": 46.0,
-      "step": 8447
-    },
-    {
-      "epoch": 0.6459085956763576,
-      "grad_norm": 0.0005629121442325413,
-      "learning_rate": 0.00019999979460795213,
-      "loss": 46.0,
-      "step": 8448
-    },
-    {
-      "epoch": 0.6459850526597473,
-      "grad_norm": 0.0009229386341758072,
-      "learning_rate": 0.0001999997945592666,
-      "loss": 46.0,
-      "step": 8449
-    },
-    {
-      "epoch": 0.646061509643137,
-      "grad_norm": 0.000907323497813195,
-      "learning_rate": 0.00019999979451057533,
-      "loss": 46.0,
-      "step": 8450
-    },
-    {
-      "epoch": 0.6461379666265268,
-      "grad_norm": 0.0011599069694057107,
-      "learning_rate": 0.00019999979446187828,
-      "loss": 46.0,
-      "step": 8451
-    },
-    {
-      "epoch": 0.6462144236099164,
-      "grad_norm": 0.0029595019295811653,
-      "learning_rate": 0.00019999979441317548,
-      "loss": 46.0,
-      "step": 8452
-    },
-    {
-      "epoch": 0.6462908805933062,
-      "grad_norm": 0.0013973616296425462,
-      "learning_rate": 0.00019999979436446689,
-      "loss": 46.0,
-      "step": 8453
-    },
-    {
-      "epoch": 0.646367337576696,
-      "grad_norm": 0.0009979789610952139,
-      "learning_rate": 0.00019999979431575254,
-      "loss": 46.0,
-      "step": 8454
-    },
-    {
-      "epoch": 0.6464437945600856,
-      "grad_norm": 0.0008624042966403067,
-      "learning_rate": 0.0001999997942670324,
-      "loss": 46.0,
-      "step": 8455
-    },
-    {
-      "epoch": 0.6465202515434754,
-      "grad_norm": 0.006532927975058556,
-      "learning_rate": 0.00019999979421830654,
-      "loss": 46.0,
-      "step": 8456
-    },
-    {
-      "epoch": 0.6465967085268651,
-      "grad_norm": 0.001143553527072072,
-      "learning_rate": 0.00019999979416957485,
-      "loss": 46.0,
-      "step": 8457
-    },
-    {
-      "epoch": 0.6466731655102548,
-      "grad_norm": 0.00062465516384691,
-      "learning_rate": 0.00019999979412083744,
-      "loss": 46.0,
-      "step": 8458
-    },
-    {
-      "epoch": 0.6467496224936445,
-      "grad_norm": 0.0025251784827560186,
-      "learning_rate": 0.00019999979407209426,
-      "loss": 46.0,
-      "step": 8459
-    },
-    {
-      "epoch": 0.6468260794770342,
-      "grad_norm": 0.0016846662620082498,
-      "learning_rate": 0.00019999979402334528,
-      "loss": 46.0,
-      "step": 8460
-    },
-    {
-      "epoch": 0.646902536460424,
-      "grad_norm": 0.0018301091622561216,
-      "learning_rate": 0.00019999979397459052,
-      "loss": 46.0,
-      "step": 8461
-    },
-    {
-      "epoch": 0.6469789934438137,
-      "grad_norm": 0.0025677981320768595,
-      "learning_rate": 0.00019999979392583002,
-      "loss": 46.0,
-      "step": 8462
-    },
-    {
-      "epoch": 0.6470554504272034,
-      "grad_norm": 0.0011541744461283088,
-      "learning_rate": 0.00019999979387706374,
-      "loss": 46.0,
-      "step": 8463
-    },
-    {
-      "epoch": 0.6471319074105931,
-      "grad_norm": 0.003928376827389002,
-      "learning_rate": 0.0001999997938282917,
-      "loss": 46.0,
-      "step": 8464
-    },
-    {
-      "epoch": 0.6472083643939829,
-      "grad_norm": 0.0005821063532494009,
-      "learning_rate": 0.0001999997937795139,
-      "loss": 46.0,
-      "step": 8465
-    },
-    {
-      "epoch": 0.6472848213773725,
-      "grad_norm": 0.0033516576513648033,
-      "learning_rate": 0.0001999997937307303,
-      "loss": 46.0,
-      "step": 8466
-    },
-    {
-      "epoch": 0.6473612783607623,
-      "grad_norm": 0.00267504109069705,
-      "learning_rate": 0.00019999979368194094,
-      "loss": 46.0,
-      "step": 8467
-    },
-    {
-      "epoch": 0.6474377353441519,
-      "grad_norm": 0.0011814562603831291,
-      "learning_rate": 0.00019999979363314582,
-      "loss": 46.0,
-      "step": 8468
-    },
-    {
-      "epoch": 0.6475141923275417,
-      "grad_norm": 0.001345281838439405,
-      "learning_rate": 0.0001999997935843449,
-      "loss": 46.0,
-      "step": 8469
-    },
-    {
-      "epoch": 0.6475906493109315,
-      "grad_norm": 0.0007618293166160583,
-      "learning_rate": 0.00019999979353553827,
-      "loss": 46.0,
-      "step": 8470
-    },
-    {
-      "epoch": 0.6476671062943211,
-      "grad_norm": 0.010481813922524452,
-      "learning_rate": 0.00019999979348672584,
-      "loss": 46.0,
-      "step": 8471
-    },
-    {
-      "epoch": 0.6477435632777109,
-      "grad_norm": 0.0016783365281298757,
-      "learning_rate": 0.00019999979343790763,
-      "loss": 46.0,
-      "step": 8472
-    },
-    {
-      "epoch": 0.6478200202611006,
-      "grad_norm": 0.0021542124450206757,
-      "learning_rate": 0.00019999979338908365,
-      "loss": 46.0,
-      "step": 8473
-    },
-    {
-      "epoch": 0.6478964772444903,
-      "grad_norm": 0.001814293791539967,
-      "learning_rate": 0.0001999997933402539,
-      "loss": 46.0,
-      "step": 8474
-    },
-    {
-      "epoch": 0.64797293422788,
-      "grad_norm": 0.003838587086647749,
-      "learning_rate": 0.00019999979329141842,
-      "loss": 46.0,
-      "step": 8475
-    },
-    {
-      "epoch": 0.6480493912112698,
-      "grad_norm": 0.0007807528018020093,
-      "learning_rate": 0.00019999979324257711,
-      "loss": 46.0,
-      "step": 8476
-    },
-    {
-      "epoch": 0.6481258481946595,
-      "grad_norm": 0.0013903791550546885,
-      "learning_rate": 0.00019999979319373007,
-      "loss": 46.0,
-      "step": 8477
-    },
-    {
-      "epoch": 0.6482023051780492,
-      "grad_norm": 0.0014009339502081275,
-      "learning_rate": 0.00019999979314487724,
-      "loss": 46.0,
-      "step": 8478
-    },
-    {
-      "epoch": 0.6482787621614389,
-      "grad_norm": 0.0008330792188644409,
-      "learning_rate": 0.00019999979309601865,
-      "loss": 46.0,
-      "step": 8479
-    },
-    {
-      "epoch": 0.6483552191448286,
-      "grad_norm": 0.008371387608349323,
-      "learning_rate": 0.0001999997930471543,
-      "loss": 46.0,
-      "step": 8480
-    },
-    {
-      "epoch": 0.6484316761282184,
-      "grad_norm": 0.0049489461816847324,
-      "learning_rate": 0.00019999979299828417,
-      "loss": 46.0,
-      "step": 8481
-    },
-    {
-      "epoch": 0.648508133111608,
-      "grad_norm": 0.004436735063791275,
-      "learning_rate": 0.00019999979294940828,
-      "loss": 46.0,
-      "step": 8482
-    },
-    {
-      "epoch": 0.6485845900949978,
-      "grad_norm": 0.0019072848372161388,
-      "learning_rate": 0.00019999979290052662,
-      "loss": 46.0,
-      "step": 8483
-    },
-    {
-      "epoch": 0.6486610470783876,
-      "grad_norm": 0.0015642609214410186,
-      "learning_rate": 0.00019999979285163918,
-      "loss": 46.0,
-      "step": 8484
-    },
-    {
-      "epoch": 0.6487375040617772,
-      "grad_norm": 0.002694055438041687,
-      "learning_rate": 0.00019999979280274597,
-      "loss": 46.0,
-      "step": 8485
-    },
-    {
-      "epoch": 0.648813961045167,
-      "grad_norm": 0.0026235098484903574,
-      "learning_rate": 0.000199999792753847,
-      "loss": 46.0,
-      "step": 8486
-    },
-    {
-      "epoch": 0.6488904180285567,
-      "grad_norm": 0.00028394997934810817,
-      "learning_rate": 0.00019999979270494224,
-      "loss": 46.0,
-      "step": 8487
-    },
-    {
-      "epoch": 0.6489668750119464,
-      "grad_norm": 0.0022435025312006474,
-      "learning_rate": 0.00019999979265603176,
-      "loss": 46.0,
-      "step": 8488
-    },
-    {
-      "epoch": 0.6490433319953361,
-      "grad_norm": 0.001888152677565813,
-      "learning_rate": 0.00019999979260711546,
-      "loss": 46.0,
-      "step": 8489
-    },
-    {
-      "epoch": 0.6491197889787258,
-      "grad_norm": 0.004005166701972485,
-      "learning_rate": 0.00019999979255819338,
-      "loss": 46.0,
-      "step": 8490
-    },
-    {
-      "epoch": 0.6491962459621156,
-      "grad_norm": 0.003759379032999277,
-      "learning_rate": 0.0001999997925092656,
-      "loss": 46.0,
-      "step": 8491
-    },
-    {
-      "epoch": 0.6492727029455053,
-      "grad_norm": 0.0005304914084263146,
-      "learning_rate": 0.00019999979246033197,
-      "loss": 46.0,
-      "step": 8492
-    },
-    {
-      "epoch": 0.649349159928895,
-      "grad_norm": 0.0012637015897780657,
-      "learning_rate": 0.00019999979241139263,
-      "loss": 46.0,
-      "step": 8493
-    },
-    {
-      "epoch": 0.6494256169122847,
-      "grad_norm": 0.0011224878253415227,
-      "learning_rate": 0.00019999979236244749,
-      "loss": 46.0,
-      "step": 8494
-    },
-    {
-      "epoch": 0.6495020738956745,
-      "grad_norm": 0.005963853560388088,
-      "learning_rate": 0.0001999997923134966,
-      "loss": 46.0,
-      "step": 8495
-    },
-    {
-      "epoch": 0.6495785308790641,
-      "grad_norm": 0.0006357973325066268,
-      "learning_rate": 0.00019999979226453994,
-      "loss": 46.0,
-      "step": 8496
-    },
-    {
-      "epoch": 0.6496549878624539,
-      "grad_norm": 0.0005483375862240791,
-      "learning_rate": 0.0001999997922155775,
-      "loss": 46.0,
-      "step": 8497
-    },
-    {
-      "epoch": 0.6497314448458436,
-      "grad_norm": 0.001385939191095531,
-      "learning_rate": 0.0001999997921666093,
-      "loss": 46.0,
-      "step": 8498
-    },
-    {
-      "epoch": 0.6498079018292333,
-      "grad_norm": 0.0025952893774956465,
-      "learning_rate": 0.0001999997921176353,
-      "loss": 46.0,
-      "step": 8499
-    },
-    {
-      "epoch": 0.6498843588126231,
-      "grad_norm": 0.0007776801940053701,
-      "learning_rate": 0.00019999979206865556,
-      "loss": 46.0,
-      "step": 8500
-    },
-    {
-      "epoch": 0.6499608157960127,
-      "grad_norm": 0.0005324630183167756,
-      "learning_rate": 0.00019999979201967003,
-      "loss": 46.0,
-      "step": 8501
-    },
-    {
-      "epoch": 0.6500372727794025,
-      "grad_norm": 0.0015339070232585073,
-      "learning_rate": 0.00019999979197067876,
-      "loss": 46.0,
-      "step": 8502
-    },
-    {
-      "epoch": 0.6501137297627922,
-      "grad_norm": 0.0013700415147468448,
-      "learning_rate": 0.0001999997919216817,
-      "loss": 46.0,
-      "step": 8503
-    },
-    {
-      "epoch": 0.6501901867461819,
-      "grad_norm": 0.0020882831886410713,
-      "learning_rate": 0.00019999979187267886,
-      "loss": 46.0,
-      "step": 8504
-    },
-    {
-      "epoch": 0.6502666437295717,
-      "grad_norm": 0.0018810388864949346,
-      "learning_rate": 0.0001999997918236703,
-      "loss": 46.0,
-      "step": 8505
-    },
-    {
-      "epoch": 0.6503431007129614,
-      "grad_norm": 0.0012970364186912775,
-      "learning_rate": 0.0001999997917746559,
-      "loss": 46.0,
-      "step": 8506
-    },
-    {
-      "epoch": 0.6504195576963511,
-      "grad_norm": 0.0009746933355927467,
-      "learning_rate": 0.00019999979172563576,
-      "loss": 46.0,
-      "step": 8507
-    },
-    {
-      "epoch": 0.6504960146797408,
-      "grad_norm": 0.005081235896795988,
-      "learning_rate": 0.00019999979167660987,
-      "loss": 46.0,
-      "step": 8508
-    },
-    {
-      "epoch": 0.6505724716631305,
-      "grad_norm": 0.001871288288384676,
-      "learning_rate": 0.0001999997916275782,
-      "loss": 46.0,
-      "step": 8509
-    },
-    {
-      "epoch": 0.6506489286465202,
-      "grad_norm": 0.002059999154880643,
-      "learning_rate": 0.00019999979157854077,
-      "loss": 46.0,
-      "step": 8510
-    },
-    {
-      "epoch": 0.65072538562991,
-      "grad_norm": 0.002449559746310115,
-      "learning_rate": 0.00019999979152949754,
-      "loss": 46.0,
-      "step": 8511
-    },
-    {
-      "epoch": 0.6508018426132997,
-      "grad_norm": 0.0018699516076594591,
-      "learning_rate": 0.00019999979148044856,
-      "loss": 46.0,
-      "step": 8512
-    },
-    {
-      "epoch": 0.6508782995966894,
-      "grad_norm": 0.0009525343193672597,
-      "learning_rate": 0.00019999979143139383,
-      "loss": 46.0,
-      "step": 8513
-    },
-    {
-      "epoch": 0.6509547565800792,
-      "grad_norm": 0.0006265377742238343,
-      "learning_rate": 0.0001999997913823333,
-      "loss": 46.0,
-      "step": 8514
-    },
-    {
-      "epoch": 0.6510312135634688,
-      "grad_norm": 0.000944600673392415,
-      "learning_rate": 0.000199999791333267,
-      "loss": 46.0,
-      "step": 8515
-    },
-    {
-      "epoch": 0.6511076705468586,
-      "grad_norm": 0.0008313765865750611,
-      "learning_rate": 0.00019999979128419495,
-      "loss": 46.0,
-      "step": 8516
-    },
-    {
-      "epoch": 0.6511841275302483,
-      "grad_norm": 0.0014036077773198485,
-      "learning_rate": 0.0001999997912351171,
-      "loss": 46.0,
-      "step": 8517
-    },
-    {
-      "epoch": 0.651260584513638,
-      "grad_norm": 0.0011903411941602826,
-      "learning_rate": 0.00019999979118603354,
-      "loss": 46.0,
-      "step": 8518
-    },
-    {
-      "epoch": 0.6513370414970278,
-      "grad_norm": 0.0015697837807238102,
-      "learning_rate": 0.00019999979113694417,
-      "loss": 46.0,
-      "step": 8519
-    },
-    {
-      "epoch": 0.6514134984804174,
-      "grad_norm": 0.0015175606822595,
-      "learning_rate": 0.00019999979108784903,
-      "loss": 46.0,
-      "step": 8520
-    },
-    {
-      "epoch": 0.6514899554638072,
-      "grad_norm": 0.0007969412254169583,
-      "learning_rate": 0.00019999979103874812,
-      "loss": 46.0,
-      "step": 8521
-    },
-    {
-      "epoch": 0.6515664124471969,
-      "grad_norm": 0.0008238257141783834,
-      "learning_rate": 0.00019999979098964143,
-      "loss": 46.0,
-      "step": 8522
-    },
-    {
-      "epoch": 0.6516428694305866,
-      "grad_norm": 0.005502438638359308,
-      "learning_rate": 0.000199999790940529,
-      "loss": 46.0,
-      "step": 8523
-    },
-    {
-      "epoch": 0.6517193264139763,
-      "grad_norm": 0.000979154952801764,
-      "learning_rate": 0.0001999997908914108,
-      "loss": 46.0,
-      "step": 8524
-    },
-    {
-      "epoch": 0.6517957833973661,
-      "grad_norm": 0.009840414859354496,
-      "learning_rate": 0.0001999997908422868,
-      "loss": 46.0,
-      "step": 8525
-    },
-    {
-      "epoch": 0.6518722403807558,
-      "grad_norm": 0.0016871714033186436,
-      "learning_rate": 0.00019999979079315705,
-      "loss": 46.0,
-      "step": 8526
-    },
-    {
-      "epoch": 0.6519486973641455,
-      "grad_norm": 0.004615066573023796,
-      "learning_rate": 0.00019999979074402152,
-      "loss": 46.0,
-      "step": 8527
-    },
-    {
-      "epoch": 0.6520251543475352,
-      "grad_norm": 0.0008247296791523695,
-      "learning_rate": 0.00019999979069488025,
-      "loss": 46.0,
-      "step": 8528
-    },
-    {
-      "epoch": 0.6521016113309249,
-      "grad_norm": 0.0007270098431035876,
-      "learning_rate": 0.00019999979064573318,
-      "loss": 46.0,
-      "step": 8529
-    },
-    {
-      "epoch": 0.6521780683143147,
-      "grad_norm": 0.002789735794067383,
-      "learning_rate": 0.00019999979059658033,
-      "loss": 46.0,
-      "step": 8530
-    },
-    {
-      "epoch": 0.6522545252977043,
-      "grad_norm": 0.0013686614111065865,
-      "learning_rate": 0.00019999979054742176,
-      "loss": 46.0,
-      "step": 8531
-    },
-    {
-      "epoch": 0.6523309822810941,
-      "grad_norm": 0.0030278346966952085,
-      "learning_rate": 0.0001999997904982574,
-      "loss": 46.0,
-      "step": 8532
-    },
-    {
-      "epoch": 0.6524074392644839,
-      "grad_norm": 0.0013672002824023366,
-      "learning_rate": 0.00019999979044908725,
-      "loss": 46.0,
-      "step": 8533
-    },
-    {
-      "epoch": 0.6524838962478735,
-      "grad_norm": 0.002181177493184805,
-      "learning_rate": 0.00019999979039991134,
-      "loss": 46.0,
-      "step": 8534
-    },
-    {
-      "epoch": 0.6525603532312633,
-      "grad_norm": 0.0010011212434619665,
-      "learning_rate": 0.00019999979035072965,
-      "loss": 46.0,
-      "step": 8535
-    },
-    {
-      "epoch": 0.652636810214653,
-      "grad_norm": 0.0024739818181842566,
-      "learning_rate": 0.00019999979030154222,
-      "loss": 46.0,
-      "step": 8536
-    },
-    {
-      "epoch": 0.6527132671980427,
-      "grad_norm": 0.0017216530395671725,
-      "learning_rate": 0.00019999979025234899,
-      "loss": 46.0,
-      "step": 8537
-    },
-    {
-      "epoch": 0.6527897241814324,
-      "grad_norm": 0.0016360745066776872,
-      "learning_rate": 0.00019999979020315,
-      "loss": 46.0,
-      "step": 8538
-    },
-    {
-      "epoch": 0.6528661811648221,
-      "grad_norm": 0.0016927840188145638,
-      "learning_rate": 0.00019999979015394525,
-      "loss": 46.0,
-      "step": 8539
-    },
-    {
-      "epoch": 0.6529426381482119,
-      "grad_norm": 0.001046431134454906,
-      "learning_rate": 0.0001999997901047347,
-      "loss": 46.0,
-      "step": 8540
-    },
-    {
-      "epoch": 0.6530190951316016,
-      "grad_norm": 0.0011743007926270366,
-      "learning_rate": 0.00019999979005551842,
-      "loss": 46.0,
-      "step": 8541
-    },
-    {
-      "epoch": 0.6530955521149913,
-      "grad_norm": 0.0008669368689879775,
-      "learning_rate": 0.00019999979000629635,
-      "loss": 46.0,
-      "step": 8542
-    },
-    {
-      "epoch": 0.653172009098381,
-      "grad_norm": 0.00373827968724072,
-      "learning_rate": 0.00019999978995706853,
-      "loss": 46.0,
-      "step": 8543
-    },
-    {
-      "epoch": 0.6532484660817708,
-      "grad_norm": 0.0005279045435599983,
-      "learning_rate": 0.0001999997899078349,
-      "loss": 46.0,
-      "step": 8544
-    },
-    {
-      "epoch": 0.6533249230651604,
-      "grad_norm": 0.0005049153114669025,
-      "learning_rate": 0.00019999978985859554,
-      "loss": 46.0,
-      "step": 8545
-    },
-    {
-      "epoch": 0.6534013800485502,
-      "grad_norm": 0.00194987328723073,
-      "learning_rate": 0.0001999997898093504,
-      "loss": 46.0,
-      "step": 8546
-    },
-    {
-      "epoch": 0.65347783703194,
-      "grad_norm": 0.0013528362615033984,
-      "learning_rate": 0.0001999997897600995,
-      "loss": 46.0,
-      "step": 8547
-    },
-    {
-      "epoch": 0.6535542940153296,
-      "grad_norm": 0.00036748481215909123,
-      "learning_rate": 0.0001999997897108428,
-      "loss": 46.0,
-      "step": 8548
-    },
-    {
-      "epoch": 0.6536307509987194,
-      "grad_norm": 0.003745422698557377,
-      "learning_rate": 0.00019999978966158037,
-      "loss": 46.0,
-      "step": 8549
-    },
-    {
-      "epoch": 0.653707207982109,
-      "grad_norm": 0.00045087194303050637,
-      "learning_rate": 0.00019999978961231216,
-      "loss": 46.0,
-      "step": 8550
-    },
-    {
-      "epoch": 0.6537836649654988,
-      "grad_norm": 0.0016072599682956934,
-      "learning_rate": 0.00019999978956303815,
-      "loss": 46.0,
-      "step": 8551
-    },
-    {
-      "epoch": 0.6538601219488885,
-      "grad_norm": 0.001266929553821683,
-      "learning_rate": 0.00019999978951375843,
-      "loss": 46.0,
-      "step": 8552
-    },
-    {
-      "epoch": 0.6539365789322782,
-      "grad_norm": 0.0007900830241851509,
-      "learning_rate": 0.00019999978946447287,
-      "loss": 46.0,
-      "step": 8553
-    },
-    {
-      "epoch": 0.654013035915668,
-      "grad_norm": 0.0005708296084776521,
-      "learning_rate": 0.00019999978941518157,
-      "loss": 46.0,
-      "step": 8554
-    },
-    {
-      "epoch": 0.6540894928990577,
-      "grad_norm": 0.006334260571748018,
-      "learning_rate": 0.00019999978936588453,
-      "loss": 46.0,
-      "step": 8555
-    },
-    {
-      "epoch": 0.6541659498824474,
-      "grad_norm": 0.0016598799265921116,
-      "learning_rate": 0.00019999978931658168,
-      "loss": 46.0,
-      "step": 8556
-    },
-    {
-      "epoch": 0.6542424068658371,
-      "grad_norm": 0.0004519044596236199,
-      "learning_rate": 0.0001999997892672731,
-      "loss": 46.0,
-      "step": 8557
-    },
-    {
-      "epoch": 0.6543188638492269,
-      "grad_norm": 0.0003759263490792364,
-      "learning_rate": 0.0001999997892179587,
-      "loss": 46.0,
-      "step": 8558
-    },
-    {
-      "epoch": 0.6543953208326165,
-      "grad_norm": 0.002499691443517804,
-      "learning_rate": 0.00019999978916863855,
-      "loss": 46.0,
-      "step": 8559
-    },
-    {
-      "epoch": 0.6544717778160063,
-      "grad_norm": 0.0007688610348850489,
-      "learning_rate": 0.00019999978911931264,
-      "loss": 46.0,
-      "step": 8560
-    },
-    {
-      "epoch": 0.654548234799396,
-      "grad_norm": 0.0012953185942023993,
-      "learning_rate": 0.00019999978906998095,
-      "loss": 46.0,
-      "step": 8561
-    },
-    {
-      "epoch": 0.6546246917827857,
-      "grad_norm": 0.0012482163729146123,
-      "learning_rate": 0.0001999997890206435,
-      "loss": 46.0,
-      "step": 8562
-    },
-    {
-      "epoch": 0.6547011487661755,
-      "grad_norm": 0.0009858411503955722,
-      "learning_rate": 0.0001999997889713003,
-      "loss": 46.0,
-      "step": 8563
-    },
-    {
-      "epoch": 0.6547776057495651,
-      "grad_norm": 0.001155849196948111,
-      "learning_rate": 0.00019999978892195128,
-      "loss": 46.0,
-      "step": 8564
-    },
-    {
-      "epoch": 0.6548540627329549,
-      "grad_norm": 0.00048214488197118044,
-      "learning_rate": 0.00019999978887259653,
-      "loss": 46.0,
-      "step": 8565
-    },
-    {
-      "epoch": 0.6549305197163446,
-      "grad_norm": 0.0011347491526976228,
-      "learning_rate": 0.000199999788823236,
-      "loss": 46.0,
-      "step": 8566
-    },
-    {
-      "epoch": 0.6550069766997343,
-      "grad_norm": 0.004829889163374901,
-      "learning_rate": 0.0001999997887738697,
-      "loss": 46.0,
-      "step": 8567
-    },
-    {
-      "epoch": 0.655083433683124,
-      "grad_norm": 0.0015248326817527413,
-      "learning_rate": 0.00019999978872449763,
-      "loss": 46.0,
-      "step": 8568
-    },
-    {
-      "epoch": 0.6551598906665137,
-      "grad_norm": 0.0005274652503430843,
-      "learning_rate": 0.00019999978867511978,
-      "loss": 46.0,
-      "step": 8569
-    },
-    {
-      "epoch": 0.6552363476499035,
-      "grad_norm": 0.0014829428400844336,
-      "learning_rate": 0.0001999997886257362,
-      "loss": 46.0,
-      "step": 8570
-    },
-    {
-      "epoch": 0.6553128046332932,
-      "grad_norm": 0.0006693553877994418,
-      "learning_rate": 0.0001999997885763468,
-      "loss": 46.0,
-      "step": 8571
-    },
-    {
-      "epoch": 0.6553892616166829,
-      "grad_norm": 0.0012639377964660525,
-      "learning_rate": 0.00019999978852695165,
-      "loss": 46.0,
-      "step": 8572
-    },
-    {
-      "epoch": 0.6554657186000726,
-      "grad_norm": 0.010828603059053421,
-      "learning_rate": 0.00019999978847755074,
-      "loss": 46.0,
-      "step": 8573
-    },
-    {
-      "epoch": 0.6555421755834624,
-      "grad_norm": 0.0010106988484039903,
-      "learning_rate": 0.00019999978842814405,
-      "loss": 46.0,
-      "step": 8574
-    },
-    {
-      "epoch": 0.655618632566852,
-      "grad_norm": 0.0007190277683548629,
-      "learning_rate": 0.0001999997883787316,
-      "loss": 46.0,
-      "step": 8575
-    },
-    {
-      "epoch": 0.6556950895502418,
-      "grad_norm": 0.0035418092738837004,
-      "learning_rate": 0.0001999997883293134,
-      "loss": 46.0,
-      "step": 8576
-    },
-    {
-      "epoch": 0.6557715465336316,
-      "grad_norm": 0.0019862360786646605,
-      "learning_rate": 0.00019999978827988938,
-      "loss": 46.0,
-      "step": 8577
-    },
-    {
-      "epoch": 0.6558480035170212,
-      "grad_norm": 0.0008121223654597998,
-      "learning_rate": 0.00019999978823045963,
-      "loss": 46.0,
-      "step": 8578
-    },
-    {
-      "epoch": 0.655924460500411,
-      "grad_norm": 0.00486897025257349,
-      "learning_rate": 0.0001999997881810241,
-      "loss": 46.0,
-      "step": 8579
-    },
-    {
-      "epoch": 0.6560009174838006,
-      "grad_norm": 0.0011067648883908987,
-      "learning_rate": 0.0001999997881315828,
-      "loss": 46.0,
-      "step": 8580
-    },
-    {
-      "epoch": 0.6560773744671904,
-      "grad_norm": 0.0010947034461423755,
-      "learning_rate": 0.0001999997880821357,
-      "loss": 46.0,
-      "step": 8581
-    },
-    {
-      "epoch": 0.6561538314505801,
-      "grad_norm": 0.005246949382126331,
-      "learning_rate": 0.00019999978803268288,
-      "loss": 46.0,
-      "step": 8582
-    },
-    {
-      "epoch": 0.6562302884339698,
-      "grad_norm": 0.0006006269832141697,
-      "learning_rate": 0.00019999978798322426,
-      "loss": 46.0,
-      "step": 8583
-    },
-    {
-      "epoch": 0.6563067454173596,
-      "grad_norm": 0.0011809815187007189,
-      "learning_rate": 0.0001999997879337599,
-      "loss": 46.0,
-      "step": 8584
-    },
-    {
-      "epoch": 0.6563832024007493,
-      "grad_norm": 0.0010445114457979798,
-      "learning_rate": 0.00019999978788428973,
-      "loss": 46.0,
-      "step": 8585
-    },
-    {
-      "epoch": 0.656459659384139,
-      "grad_norm": 0.0049142492935061455,
-      "learning_rate": 0.00019999978783481382,
-      "loss": 46.0,
-      "step": 8586
-    },
-    {
-      "epoch": 0.6565361163675287,
-      "grad_norm": 0.0014443404506891966,
-      "learning_rate": 0.00019999978778533213,
-      "loss": 46.0,
-      "step": 8587
-    },
-    {
-      "epoch": 0.6566125733509185,
-      "grad_norm": 0.0005549098714254797,
-      "learning_rate": 0.00019999978773584467,
-      "loss": 46.0,
-      "step": 8588
-    },
-    {
-      "epoch": 0.6566890303343081,
-      "grad_norm": 0.002379411133006215,
-      "learning_rate": 0.00019999978768635144,
-      "loss": 46.0,
-      "step": 8589
-    },
-    {
-      "epoch": 0.6567654873176979,
-      "grad_norm": 0.0010705901077017188,
-      "learning_rate": 0.00019999978763685246,
-      "loss": 46.0,
-      "step": 8590
-    },
-    {
-      "epoch": 0.6568419443010876,
-      "grad_norm": 0.0005772141157649457,
-      "learning_rate": 0.00019999978758734768,
-      "loss": 46.0,
-      "step": 8591
-    },
-    {
-      "epoch": 0.6569184012844773,
-      "grad_norm": 0.0047366805374622345,
-      "learning_rate": 0.00019999978753783713,
-      "loss": 46.0,
-      "step": 8592
-    },
-    {
-      "epoch": 0.6569948582678671,
-      "grad_norm": 0.0015679936623200774,
-      "learning_rate": 0.00019999978748832083,
-      "loss": 46.0,
-      "step": 8593
-    },
-    {
-      "epoch": 0.6570713152512567,
-      "grad_norm": 0.0009844069136306643,
-      "learning_rate": 0.00019999978743879876,
-      "loss": 46.0,
-      "step": 8594
-    },
-    {
-      "epoch": 0.6571477722346465,
-      "grad_norm": 0.0007176010985858738,
-      "learning_rate": 0.0001999997873892709,
-      "loss": 46.0,
-      "step": 8595
-    },
-    {
-      "epoch": 0.6572242292180362,
-      "grad_norm": 0.0012983622727915645,
-      "learning_rate": 0.0001999997873397373,
-      "loss": 46.0,
-      "step": 8596
-    },
-    {
-      "epoch": 0.6573006862014259,
-      "grad_norm": 0.0005127433687448502,
-      "learning_rate": 0.00019999978729019793,
-      "loss": 46.0,
-      "step": 8597
-    },
-    {
-      "epoch": 0.6573771431848157,
-      "grad_norm": 0.0007021460914984345,
-      "learning_rate": 0.00019999978724065276,
-      "loss": 46.0,
-      "step": 8598
-    },
-    {
-      "epoch": 0.6574536001682053,
-      "grad_norm": 0.00133537407964468,
-      "learning_rate": 0.00019999978719110185,
-      "loss": 46.0,
-      "step": 8599
-    },
-    {
-      "epoch": 0.6575300571515951,
-      "grad_norm": 0.004666220396757126,
-      "learning_rate": 0.00019999978714154513,
-      "loss": 46.0,
-      "step": 8600
-    },
-    {
-      "epoch": 0.6576065141349848,
-      "grad_norm": 0.0004905962850898504,
-      "learning_rate": 0.0001999997870919827,
-      "loss": 46.0,
-      "step": 8601
-    },
-    {
-      "epoch": 0.6576829711183745,
-      "grad_norm": 0.002547111362218857,
-      "learning_rate": 0.00019999978704241447,
-      "loss": 46.0,
-      "step": 8602
-    },
-    {
-      "epoch": 0.6577594281017642,
-      "grad_norm": 0.001990128308534622,
-      "learning_rate": 0.00019999978699284046,
-      "loss": 46.0,
-      "step": 8603
-    },
-    {
-      "epoch": 0.657835885085154,
-      "grad_norm": 0.0022737362887710333,
-      "learning_rate": 0.00019999978694326068,
-      "loss": 46.0,
-      "step": 8604
-    },
-    {
-      "epoch": 0.6579123420685437,
-      "grad_norm": 0.004663367755711079,
-      "learning_rate": 0.00019999978689367516,
-      "loss": 46.0,
-      "step": 8605
-    },
-    {
-      "epoch": 0.6579887990519334,
-      "grad_norm": 0.0008131011272780597,
-      "learning_rate": 0.00019999978684408386,
-      "loss": 46.0,
-      "step": 8606
-    },
-    {
-      "epoch": 0.6580652560353232,
-      "grad_norm": 0.0016348271165043116,
-      "learning_rate": 0.00019999978679448676,
-      "loss": 46.0,
-      "step": 8607
-    },
-    {
-      "epoch": 0.6581417130187128,
-      "grad_norm": 0.0009887042688205838,
-      "learning_rate": 0.00019999978674488392,
-      "loss": 46.0,
-      "step": 8608
-    },
-    {
-      "epoch": 0.6582181700021026,
-      "grad_norm": 0.008708581328392029,
-      "learning_rate": 0.0001999997866952753,
-      "loss": 46.0,
-      "step": 8609
-    },
-    {
-      "epoch": 0.6582946269854922,
-      "grad_norm": 0.0010642745764926076,
-      "learning_rate": 0.00019999978664566093,
-      "loss": 46.0,
-      "step": 8610
-    },
-    {
-      "epoch": 0.658371083968882,
-      "grad_norm": 0.0014296063454821706,
-      "learning_rate": 0.00019999978659604074,
-      "loss": 46.0,
-      "step": 8611
-    },
-    {
-      "epoch": 0.6584475409522718,
-      "grad_norm": 0.006288490723818541,
-      "learning_rate": 0.00019999978654641485,
-      "loss": 46.0,
-      "step": 8612
-    },
-    {
-      "epoch": 0.6585239979356614,
-      "grad_norm": 0.0026276526041328907,
-      "learning_rate": 0.00019999978649678314,
-      "loss": 46.0,
-      "step": 8613
-    },
-    {
-      "epoch": 0.6586004549190512,
-      "grad_norm": 0.0018835540395230055,
-      "learning_rate": 0.00019999978644714568,
-      "loss": 46.0,
-      "step": 8614
-    },
-    {
-      "epoch": 0.6586769119024409,
-      "grad_norm": 0.0016152536263689399,
-      "learning_rate": 0.00019999978639750245,
-      "loss": 46.0,
-      "step": 8615
-    },
-    {
-      "epoch": 0.6587533688858306,
-      "grad_norm": 0.0010026829550042748,
-      "learning_rate": 0.00019999978634785347,
-      "loss": 46.0,
-      "step": 8616
-    },
-    {
-      "epoch": 0.6588298258692203,
-      "grad_norm": 0.001898156013339758,
-      "learning_rate": 0.0001999997862981987,
-      "loss": 46.0,
-      "step": 8617
-    },
-    {
-      "epoch": 0.6589062828526101,
-      "grad_norm": 0.0030765426345169544,
-      "learning_rate": 0.00019999978624853814,
-      "loss": 46.0,
-      "step": 8618
-    },
-    {
-      "epoch": 0.6589827398359998,
-      "grad_norm": 0.0014737654710188508,
-      "learning_rate": 0.00019999978619887184,
-      "loss": 46.0,
-      "step": 8619
-    },
-    {
-      "epoch": 0.6590591968193895,
-      "grad_norm": 0.0020420304499566555,
-      "learning_rate": 0.00019999978614919974,
-      "loss": 46.0,
-      "step": 8620
-    },
-    {
-      "epoch": 0.6591356538027792,
-      "grad_norm": 0.0014412233140319586,
-      "learning_rate": 0.00019999978609952187,
-      "loss": 46.0,
-      "step": 8621
-    },
-    {
-      "epoch": 0.6592121107861689,
-      "grad_norm": 0.0011346039827913046,
-      "learning_rate": 0.00019999978604983828,
-      "loss": 46.0,
-      "step": 8622
-    },
-    {
-      "epoch": 0.6592885677695587,
-      "grad_norm": 0.004264628514647484,
-      "learning_rate": 0.00019999978600014889,
-      "loss": 46.0,
-      "step": 8623
-    },
-    {
-      "epoch": 0.6593650247529483,
-      "grad_norm": 0.0006827763863839209,
-      "learning_rate": 0.00019999978595045372,
-      "loss": 46.0,
-      "step": 8624
-    },
-    {
-      "epoch": 0.6594414817363381,
-      "grad_norm": 0.0021015778183937073,
-      "learning_rate": 0.00019999978590075278,
-      "loss": 46.0,
-      "step": 8625
-    },
-    {
-      "epoch": 0.6595179387197279,
-      "grad_norm": 0.0024254578165709972,
-      "learning_rate": 0.0001999997858510461,
-      "loss": 46.0,
-      "step": 8626
-    },
-    {
-      "epoch": 0.6595943957031175,
-      "grad_norm": 0.024874890223145485,
-      "learning_rate": 0.00019999978580133364,
-      "loss": 46.0,
-      "step": 8627
-    },
-    {
-      "epoch": 0.6596708526865073,
-      "grad_norm": 0.001570564811117947,
-      "learning_rate": 0.0001999997857516154,
-      "loss": 46.0,
-      "step": 8628
-    },
-    {
-      "epoch": 0.6597473096698969,
-      "grad_norm": 0.0005236054421402514,
-      "learning_rate": 0.0001999997857018914,
-      "loss": 46.0,
-      "step": 8629
-    },
-    {
-      "epoch": 0.6598237666532867,
-      "grad_norm": 0.001178934471681714,
-      "learning_rate": 0.00019999978565216162,
-      "loss": 46.0,
-      "step": 8630
-    },
-    {
-      "epoch": 0.6599002236366764,
-      "grad_norm": 0.000714669527951628,
-      "learning_rate": 0.0001999997856024261,
-      "loss": 46.0,
-      "step": 8631
-    },
-    {
-      "epoch": 0.6599766806200661,
-      "grad_norm": 0.0004759739385917783,
-      "learning_rate": 0.00019999978555268477,
-      "loss": 46.0,
-      "step": 8632
-    },
-    {
-      "epoch": 0.6600531376034559,
-      "grad_norm": 0.0013500095810741186,
-      "learning_rate": 0.00019999978550293767,
-      "loss": 46.0,
-      "step": 8633
-    },
-    {
-      "epoch": 0.6601295945868456,
-      "grad_norm": 0.005612598732113838,
-      "learning_rate": 0.00019999978545318483,
-      "loss": 46.0,
-      "step": 8634
-    },
-    {
-      "epoch": 0.6602060515702353,
-      "grad_norm": 0.0008059635874815285,
-      "learning_rate": 0.0001999997854034262,
-      "loss": 46.0,
-      "step": 8635
-    },
-    {
-      "epoch": 0.660282508553625,
-      "grad_norm": 0.0010862040799111128,
-      "learning_rate": 0.00019999978535366182,
-      "loss": 46.0,
-      "step": 8636
-    },
-    {
-      "epoch": 0.6603589655370148,
-      "grad_norm": 0.001186320441775024,
-      "learning_rate": 0.00019999978530389165,
-      "loss": 46.0,
-      "step": 8637
-    },
-    {
-      "epoch": 0.6604354225204044,
-      "grad_norm": 0.0012938138097524643,
-      "learning_rate": 0.00019999978525411574,
-      "loss": 46.0,
-      "step": 8638
-    },
-    {
-      "epoch": 0.6605118795037942,
-      "grad_norm": 0.0003736805228982121,
-      "learning_rate": 0.00019999978520433403,
-      "loss": 46.0,
-      "step": 8639
-    },
-    {
-      "epoch": 0.6605883364871838,
-      "grad_norm": 0.000617126643192023,
-      "learning_rate": 0.00019999978515454657,
-      "loss": 46.0,
-      "step": 8640
-    },
-    {
-      "epoch": 0.6606647934705736,
-      "grad_norm": 0.00038321109605021775,
-      "learning_rate": 0.0001999997851047533,
-      "loss": 46.0,
-      "step": 8641
-    },
-    {
-      "epoch": 0.6607412504539634,
-      "grad_norm": 0.0010610792087391019,
-      "learning_rate": 0.0001999997850549543,
-      "loss": 46.0,
-      "step": 8642
-    },
-    {
-      "epoch": 0.660817707437353,
-      "grad_norm": 0.0003566528612282127,
-      "learning_rate": 0.00019999978500514953,
-      "loss": 46.0,
-      "step": 8643
-    },
-    {
-      "epoch": 0.6608941644207428,
-      "grad_norm": 0.0018905200995504856,
-      "learning_rate": 0.000199999784955339,
-      "loss": 46.0,
-      "step": 8644
-    },
-    {
-      "epoch": 0.6609706214041325,
-      "grad_norm": 0.0006439249846152961,
-      "learning_rate": 0.00019999978490552268,
-      "loss": 46.0,
-      "step": 8645
-    },
-    {
-      "epoch": 0.6610470783875222,
-      "grad_norm": 0.0014772162539884448,
-      "learning_rate": 0.00019999978485570058,
-      "loss": 46.0,
-      "step": 8646
-    },
-    {
-      "epoch": 0.661123535370912,
-      "grad_norm": 0.000992226880043745,
-      "learning_rate": 0.00019999978480587273,
-      "loss": 46.0,
-      "step": 8647
-    },
-    {
-      "epoch": 0.6611999923543017,
-      "grad_norm": 0.0009126553777605295,
-      "learning_rate": 0.0001999997847560391,
-      "loss": 46.0,
-      "step": 8648
-    },
-    {
-      "epoch": 0.6612764493376914,
-      "grad_norm": 0.009299108758568764,
-      "learning_rate": 0.0001999997847061997,
-      "loss": 46.0,
-      "step": 8649
-    },
-    {
-      "epoch": 0.6613529063210811,
-      "grad_norm": 0.0009371150517836213,
-      "learning_rate": 0.00019999978465635456,
-      "loss": 46.0,
-      "step": 8650
-    },
-    {
-      "epoch": 0.6614293633044708,
-      "grad_norm": 0.0017397486371919513,
-      "learning_rate": 0.00019999978460650362,
-      "loss": 46.0,
-      "step": 8651
-    },
-    {
-      "epoch": 0.6615058202878605,
-      "grad_norm": 0.0018773350166156888,
-      "learning_rate": 0.00019999978455664688,
-      "loss": 46.0,
-      "step": 8652
-    },
-    {
-      "epoch": 0.6615822772712503,
-      "grad_norm": 0.001606195350177586,
-      "learning_rate": 0.00019999978450678443,
-      "loss": 46.0,
-      "step": 8653
-    },
-    {
-      "epoch": 0.66165873425464,
-      "grad_norm": 0.0018110673408955336,
-      "learning_rate": 0.0001999997844569162,
-      "loss": 46.0,
-      "step": 8654
-    },
-    {
-      "epoch": 0.6617351912380297,
-      "grad_norm": 0.0011335978051647544,
-      "learning_rate": 0.0001999997844070422,
-      "loss": 46.0,
-      "step": 8655
-    },
-    {
-      "epoch": 0.6618116482214195,
-      "grad_norm": 0.00031011944520287216,
-      "learning_rate": 0.0001999997843571624,
-      "loss": 46.0,
-      "step": 8656
-    },
-    {
-      "epoch": 0.6618881052048091,
-      "grad_norm": 0.0009702396928332746,
-      "learning_rate": 0.00019999978430727686,
-      "loss": 46.0,
-      "step": 8657
-    },
-    {
-      "epoch": 0.6619645621881989,
-      "grad_norm": 0.004137062001973391,
-      "learning_rate": 0.00019999978425738554,
-      "loss": 46.0,
-      "step": 8658
-    },
-    {
-      "epoch": 0.6620410191715885,
-      "grad_norm": 0.0056906575337052345,
-      "learning_rate": 0.00019999978420748846,
-      "loss": 46.0,
-      "step": 8659
-    },
-    {
-      "epoch": 0.6621174761549783,
-      "grad_norm": 0.002599082188680768,
-      "learning_rate": 0.00019999978415758557,
-      "loss": 46.0,
-      "step": 8660
-    },
-    {
-      "epoch": 0.662193933138368,
-      "grad_norm": 0.0013367700157687068,
-      "learning_rate": 0.00019999978410767697,
-      "loss": 46.0,
-      "step": 8661
-    },
-    {
-      "epoch": 0.6622703901217577,
-      "grad_norm": 0.0017191254300996661,
-      "learning_rate": 0.00019999978405776256,
-      "loss": 46.0,
-      "step": 8662
-    },
-    {
-      "epoch": 0.6623468471051475,
-      "grad_norm": 0.002444988815113902,
-      "learning_rate": 0.00019999978400784242,
-      "loss": 46.0,
-      "step": 8663
-    },
-    {
-      "epoch": 0.6624233040885372,
-      "grad_norm": 0.005234445445239544,
-      "learning_rate": 0.00019999978395791645,
-      "loss": 46.0,
-      "step": 8664
-    },
-    {
-      "epoch": 0.6624997610719269,
-      "grad_norm": 0.0013008839450776577,
-      "learning_rate": 0.00019999978390798474,
-      "loss": 46.0,
-      "step": 8665
-    },
-    {
-      "epoch": 0.6625762180553166,
-      "grad_norm": 0.004277431406080723,
-      "learning_rate": 0.00019999978385804729,
-      "loss": 46.0,
-      "step": 8666
-    },
-    {
-      "epoch": 0.6626526750387064,
-      "grad_norm": 0.0008872722391970456,
-      "learning_rate": 0.00019999978380810405,
-      "loss": 46.0,
-      "step": 8667
-    },
-    {
-      "epoch": 0.662729132022096,
-      "grad_norm": 0.0009823297150433064,
-      "learning_rate": 0.00019999978375815502,
-      "loss": 46.0,
-      "step": 8668
-    },
-    {
-      "epoch": 0.6628055890054858,
-      "grad_norm": 0.0015522469766438007,
-      "learning_rate": 0.00019999978370820024,
-      "loss": 46.0,
-      "step": 8669
-    },
-    {
-      "epoch": 0.6628820459888755,
-      "grad_norm": 0.0016563733806833625,
-      "learning_rate": 0.0001999997836582397,
-      "loss": 46.0,
-      "step": 8670
-    },
-    {
-      "epoch": 0.6629585029722652,
-      "grad_norm": 0.0034450446255505085,
-      "learning_rate": 0.0001999997836082734,
-      "loss": 46.0,
-      "step": 8671
-    },
-    {
-      "epoch": 0.663034959955655,
-      "grad_norm": 0.002170608378946781,
-      "learning_rate": 0.00019999978355830127,
-      "loss": 46.0,
-      "step": 8672
-    },
-    {
-      "epoch": 0.6631114169390446,
-      "grad_norm": 0.002843655413016677,
-      "learning_rate": 0.00019999978350832343,
-      "loss": 46.0,
-      "step": 8673
-    },
-    {
-      "epoch": 0.6631878739224344,
-      "grad_norm": 0.0008554289815947413,
-      "learning_rate": 0.00019999978345833978,
-      "loss": 46.0,
-      "step": 8674
-    },
-    {
-      "epoch": 0.6632643309058242,
-      "grad_norm": 0.001418053638190031,
-      "learning_rate": 0.0001999997834083504,
-      "loss": 46.0,
-      "step": 8675
-    },
-    {
-      "epoch": 0.6633407878892138,
-      "grad_norm": 0.001097179250791669,
-      "learning_rate": 0.0001999997833583552,
-      "loss": 46.0,
-      "step": 8676
-    },
-    {
-      "epoch": 0.6634172448726036,
-      "grad_norm": 0.0012433599913492799,
-      "learning_rate": 0.00019999978330835426,
-      "loss": 46.0,
-      "step": 8677
-    },
-    {
-      "epoch": 0.6634937018559933,
-      "grad_norm": 0.0006016894476488233,
-      "learning_rate": 0.00019999978325834755,
-      "loss": 46.0,
-      "step": 8678
-    },
-    {
-      "epoch": 0.663570158839383,
-      "grad_norm": 0.00232914206571877,
-      "learning_rate": 0.0001999997832083351,
-      "loss": 46.0,
-      "step": 8679
-    },
-    {
-      "epoch": 0.6636466158227727,
-      "grad_norm": 0.0014480187091976404,
-      "learning_rate": 0.00019999978315831686,
-      "loss": 46.0,
-      "step": 8680
-    },
-    {
-      "epoch": 0.6637230728061624,
-      "grad_norm": 0.0008242285111919045,
-      "learning_rate": 0.00019999978310829283,
-      "loss": 46.0,
-      "step": 8681
-    },
-    {
-      "epoch": 0.6637995297895521,
-      "grad_norm": 0.0009306977153755724,
-      "learning_rate": 0.00019999978305826303,
-      "loss": 46.0,
-      "step": 8682
-    },
-    {
-      "epoch": 0.6638759867729419,
-      "grad_norm": 0.0011325350496917963,
-      "learning_rate": 0.0001999997830082275,
-      "loss": 46.0,
-      "step": 8683
-    },
-    {
-      "epoch": 0.6639524437563316,
-      "grad_norm": 0.0006435626419261098,
-      "learning_rate": 0.00019999978295818615,
-      "loss": 46.0,
-      "step": 8684
-    },
-    {
-      "epoch": 0.6640289007397213,
-      "grad_norm": 0.001619804766960442,
-      "learning_rate": 0.00019999978290813908,
-      "loss": 46.0,
-      "step": 8685
-    },
-    {
-      "epoch": 0.6641053577231111,
-      "grad_norm": 0.0015315627679228783,
-      "learning_rate": 0.00019999978285808624,
-      "loss": 46.0,
-      "step": 8686
-    },
-    {
-      "epoch": 0.6641818147065007,
-      "grad_norm": 0.0012504010228440166,
-      "learning_rate": 0.00019999978280802757,
-      "loss": 46.0,
-      "step": 8687
-    },
-    {
-      "epoch": 0.6642582716898905,
-      "grad_norm": 0.0029816979076713324,
-      "learning_rate": 0.00019999978275796315,
-      "loss": 46.0,
-      "step": 8688
-    },
-    {
-      "epoch": 0.6643347286732803,
-      "grad_norm": 0.037114787846803665,
-      "learning_rate": 0.00019999978270789299,
-      "loss": 46.0,
-      "step": 8689
-    },
-    {
-      "epoch": 0.6644111856566699,
-      "grad_norm": 0.002542545786127448,
-      "learning_rate": 0.00019999978265781708,
-      "loss": 46.0,
-      "step": 8690
-    },
-    {
-      "epoch": 0.6644876426400597,
-      "grad_norm": 0.0014440712984651327,
-      "learning_rate": 0.00019999978260773537,
-      "loss": 46.0,
-      "step": 8691
-    },
-    {
-      "epoch": 0.6645640996234493,
-      "grad_norm": 0.0018695704638957977,
-      "learning_rate": 0.00019999978255764788,
-      "loss": 46.0,
-      "step": 8692
-    },
-    {
-      "epoch": 0.6646405566068391,
-      "grad_norm": 0.00045301203499548137,
-      "learning_rate": 0.00019999978250755462,
-      "loss": 46.0,
-      "step": 8693
-    },
-    {
-      "epoch": 0.6647170135902288,
-      "grad_norm": 0.0008245119825005531,
-      "learning_rate": 0.00019999978245745562,
-      "loss": 46.0,
-      "step": 8694
-    },
-    {
-      "epoch": 0.6647934705736185,
-      "grad_norm": 0.0021519893780350685,
-      "learning_rate": 0.00019999978240735082,
-      "loss": 46.0,
-      "step": 8695
-    },
-    {
-      "epoch": 0.6648699275570082,
-      "grad_norm": 0.002213072497397661,
-      "learning_rate": 0.00019999978235724027,
-      "loss": 46.0,
-      "step": 8696
-    },
-    {
-      "epoch": 0.664946384540398,
-      "grad_norm": 0.0012607017997652292,
-      "learning_rate": 0.00019999978230712394,
-      "loss": 46.0,
-      "step": 8697
-    },
-    {
-      "epoch": 0.6650228415237877,
-      "grad_norm": 0.004147354979068041,
-      "learning_rate": 0.00019999978225700184,
-      "loss": 46.0,
-      "step": 8698
-    },
-    {
-      "epoch": 0.6650992985071774,
-      "grad_norm": 0.004091346170753241,
-      "learning_rate": 0.00019999978220687397,
-      "loss": 46.0,
-      "step": 8699
-    },
-    {
-      "epoch": 0.6651757554905671,
-      "grad_norm": 0.0015534093836322427,
-      "learning_rate": 0.00019999978215674033,
-      "loss": 46.0,
-      "step": 8700
-    },
-    {
-      "epoch": 0.6652522124739568,
-      "grad_norm": 0.0006063755135983229,
-      "learning_rate": 0.00019999978210660094,
-      "loss": 46.0,
-      "step": 8701
-    },
-    {
-      "epoch": 0.6653286694573466,
-      "grad_norm": 0.0021936220582574606,
-      "learning_rate": 0.00019999978205645575,
-      "loss": 46.0,
-      "step": 8702
-    },
-    {
-      "epoch": 0.6654051264407362,
-      "grad_norm": 0.0005755945458076894,
-      "learning_rate": 0.00019999978200630484,
-      "loss": 46.0,
-      "step": 8703
-    },
-    {
-      "epoch": 0.665481583424126,
-      "grad_norm": 0.002572490368038416,
-      "learning_rate": 0.0001999997819561481,
-      "loss": 46.0,
-      "step": 8704
-    },
-    {
-      "epoch": 0.6655580404075158,
-      "grad_norm": 0.0002919635735452175,
-      "learning_rate": 0.00019999978190598562,
-      "loss": 46.0,
-      "step": 8705
-    },
-    {
-      "epoch": 0.6656344973909054,
-      "grad_norm": 0.0014955428196117282,
-      "learning_rate": 0.00019999978185581736,
-      "loss": 46.0,
-      "step": 8706
-    },
-    {
-      "epoch": 0.6657109543742952,
-      "grad_norm": 0.0015673916786909103,
-      "learning_rate": 0.00019999978180564333,
-      "loss": 46.0,
-      "step": 8707
-    },
-    {
-      "epoch": 0.6657874113576849,
-      "grad_norm": 0.0006684870459139347,
-      "learning_rate": 0.00019999978175546355,
-      "loss": 46.0,
-      "step": 8708
-    },
-    {
-      "epoch": 0.6658638683410746,
-      "grad_norm": 0.001302677090279758,
-      "learning_rate": 0.00019999978170527798,
-      "loss": 46.0,
-      "step": 8709
-    },
-    {
-      "epoch": 0.6659403253244643,
-      "grad_norm": 0.001877059112302959,
-      "learning_rate": 0.00019999978165508665,
-      "loss": 46.0,
-      "step": 8710
-    },
-    {
-      "epoch": 0.666016782307854,
-      "grad_norm": 0.002077690325677395,
-      "learning_rate": 0.00019999978160488958,
-      "loss": 46.0,
-      "step": 8711
-    },
-    {
-      "epoch": 0.6660932392912438,
-      "grad_norm": 0.0006872828234918416,
-      "learning_rate": 0.0001999997815546867,
-      "loss": 46.0,
-      "step": 8712
-    },
-    {
-      "epoch": 0.6661696962746335,
-      "grad_norm": 0.0012548677623271942,
-      "learning_rate": 0.00019999978150447807,
-      "loss": 46.0,
-      "step": 8713
-    },
-    {
-      "epoch": 0.6662461532580232,
-      "grad_norm": 0.0009545075590722263,
-      "learning_rate": 0.00019999978145426365,
-      "loss": 46.0,
-      "step": 8714
-    },
-    {
-      "epoch": 0.6663226102414129,
-      "grad_norm": 0.001718643237836659,
-      "learning_rate": 0.00019999978140404346,
-      "loss": 46.0,
-      "step": 8715
-    },
-    {
-      "epoch": 0.6663990672248027,
-      "grad_norm": 0.0006353132193908095,
-      "learning_rate": 0.00019999978135381752,
-      "loss": 46.0,
-      "step": 8716
-    },
-    {
-      "epoch": 0.6664755242081923,
-      "grad_norm": 0.001213490148074925,
-      "learning_rate": 0.00019999978130358581,
-      "loss": 46.0,
-      "step": 8717
-    },
-    {
-      "epoch": 0.6665519811915821,
-      "grad_norm": 0.0008136135875247419,
-      "learning_rate": 0.00019999978125334833,
-      "loss": 46.0,
-      "step": 8718
-    },
-    {
-      "epoch": 0.6666284381749719,
-      "grad_norm": 0.001238768920302391,
-      "learning_rate": 0.00019999978120310505,
-      "loss": 46.0,
-      "step": 8719
-    },
-    {
-      "epoch": 0.6667048951583615,
-      "grad_norm": 0.001272666617296636,
-      "learning_rate": 0.00019999978115285604,
-      "loss": 46.0,
-      "step": 8720
-    },
-    {
-      "epoch": 0.6667813521417513,
-      "grad_norm": 0.0008869747398421168,
-      "learning_rate": 0.00019999978110260124,
-      "loss": 46.0,
-      "step": 8721
-    },
-    {
-      "epoch": 0.6668578091251409,
-      "grad_norm": 0.0006244395044632256,
-      "learning_rate": 0.00019999978105234066,
-      "loss": 46.0,
-      "step": 8722
-    },
-    {
-      "epoch": 0.6669342661085307,
-      "grad_norm": 0.0008299718028865755,
-      "learning_rate": 0.00019999978100207434,
-      "loss": 46.0,
-      "step": 8723
-    },
-    {
-      "epoch": 0.6670107230919204,
-      "grad_norm": 0.001618312206119299,
-      "learning_rate": 0.00019999978095180224,
-      "loss": 46.0,
-      "step": 8724
-    },
-    {
-      "epoch": 0.6670871800753101,
-      "grad_norm": 0.0037032710388302803,
-      "learning_rate": 0.00019999978090152437,
-      "loss": 46.0,
-      "step": 8725
-    },
-    {
-      "epoch": 0.6671636370586999,
-      "grad_norm": 0.006620650179684162,
-      "learning_rate": 0.00019999978085124073,
-      "loss": 46.0,
-      "step": 8726
-    },
-    {
-      "epoch": 0.6672400940420896,
-      "grad_norm": 0.0008105770102702081,
-      "learning_rate": 0.0001999997808009513,
-      "loss": 46.0,
-      "step": 8727
-    },
-    {
-      "epoch": 0.6673165510254793,
-      "grad_norm": 0.001030712272040546,
-      "learning_rate": 0.00019999978075065612,
-      "loss": 46.0,
-      "step": 8728
-    },
-    {
-      "epoch": 0.667393008008869,
-      "grad_norm": 0.0026812171563506126,
-      "learning_rate": 0.00019999978070035521,
-      "loss": 46.0,
-      "step": 8729
-    },
-    {
-      "epoch": 0.6674694649922587,
-      "grad_norm": 0.0029469558503478765,
-      "learning_rate": 0.00019999978065004848,
-      "loss": 46.0,
-      "step": 8730
-    },
-    {
-      "epoch": 0.6675459219756484,
-      "grad_norm": 0.0011797200422734022,
-      "learning_rate": 0.000199999780599736,
-      "loss": 46.0,
-      "step": 8731
-    },
-    {
-      "epoch": 0.6676223789590382,
-      "grad_norm": 0.0009308723965659738,
-      "learning_rate": 0.00019999978054941774,
-      "loss": 46.0,
-      "step": 8732
-    },
-    {
-      "epoch": 0.6676988359424278,
-      "grad_norm": 0.0008208405924960971,
-      "learning_rate": 0.0001999997804990937,
-      "loss": 46.0,
-      "step": 8733
-    },
-    {
-      "epoch": 0.6677752929258176,
-      "grad_norm": 0.001317824237048626,
-      "learning_rate": 0.0001999997804487639,
-      "loss": 46.0,
-      "step": 8734
-    },
-    {
-      "epoch": 0.6678517499092074,
-      "grad_norm": 0.0006504436605609953,
-      "learning_rate": 0.00019999978039842833,
-      "loss": 46.0,
-      "step": 8735
-    },
-    {
-      "epoch": 0.667928206892597,
-      "grad_norm": 0.002607166301459074,
-      "learning_rate": 0.000199999780348087,
-      "loss": 46.0,
-      "step": 8736
-    },
-    {
-      "epoch": 0.6680046638759868,
-      "grad_norm": 0.0026411728467792273,
-      "learning_rate": 0.00019999978029773988,
-      "loss": 46.0,
-      "step": 8737
-    },
-    {
-      "epoch": 0.6680811208593765,
-      "grad_norm": 0.0009394661756232381,
-      "learning_rate": 0.000199999780247387,
-      "loss": 46.0,
-      "step": 8738
-    },
-    {
-      "epoch": 0.6681575778427662,
-      "grad_norm": 0.0008242005715146661,
-      "learning_rate": 0.0001999997801970284,
-      "loss": 46.0,
-      "step": 8739
-    },
-    {
-      "epoch": 0.668234034826156,
-      "grad_norm": 0.0014916467480361462,
-      "learning_rate": 0.00019999978014666398,
-      "loss": 46.0,
-      "step": 8740
-    },
-    {
-      "epoch": 0.6683104918095456,
-      "grad_norm": 0.0021101636812090874,
-      "learning_rate": 0.0001999997800962938,
-      "loss": 46.0,
-      "step": 8741
-    },
-    {
-      "epoch": 0.6683869487929354,
-      "grad_norm": 0.0018937013810500503,
-      "learning_rate": 0.00019999978004591783,
-      "loss": 46.0,
-      "step": 8742
-    },
-    {
-      "epoch": 0.6684634057763251,
-      "grad_norm": 0.0007638661772944033,
-      "learning_rate": 0.00019999977999553612,
-      "loss": 46.0,
-      "step": 8743
-    },
-    {
-      "epoch": 0.6685398627597148,
-      "grad_norm": 0.0008760401979088783,
-      "learning_rate": 0.00019999977994514863,
-      "loss": 46.0,
-      "step": 8744
-    },
-    {
-      "epoch": 0.6686163197431045,
-      "grad_norm": 0.0012413676595315337,
-      "learning_rate": 0.00019999977989475535,
-      "loss": 46.0,
-      "step": 8745
-    },
-    {
-      "epoch": 0.6686927767264943,
-      "grad_norm": 0.000798044609837234,
-      "learning_rate": 0.00019999977984435635,
-      "loss": 46.0,
-      "step": 8746
-    },
-    {
-      "epoch": 0.668769233709884,
-      "grad_norm": 0.005128706805408001,
-      "learning_rate": 0.00019999977979395155,
-      "loss": 46.0,
-      "step": 8747
-    },
-    {
-      "epoch": 0.6688456906932737,
-      "grad_norm": 0.004137082956731319,
-      "learning_rate": 0.00019999977974354097,
-      "loss": 46.0,
-      "step": 8748
-    },
-    {
-      "epoch": 0.6689221476766635,
-      "grad_norm": 0.0022486187517642975,
-      "learning_rate": 0.00019999977969312462,
-      "loss": 46.0,
-      "step": 8749
-    },
-    {
-      "epoch": 0.6689986046600531,
-      "grad_norm": 0.001786054577678442,
-      "learning_rate": 0.00019999977964270252,
-      "loss": 46.0,
-      "step": 8750
-    },
-    {
-      "epoch": 0.6690750616434429,
-      "grad_norm": 0.0018166654044762254,
-      "learning_rate": 0.00019999977959227465,
-      "loss": 46.0,
-      "step": 8751
-    },
-    {
-      "epoch": 0.6691515186268325,
-      "grad_norm": 0.0010487374383956194,
-      "learning_rate": 0.00019999977954184098,
-      "loss": 46.0,
-      "step": 8752
-    },
-    {
-      "epoch": 0.6692279756102223,
-      "grad_norm": 0.0004923665546812117,
-      "learning_rate": 0.00019999977949140157,
-      "loss": 46.0,
-      "step": 8753
-    },
-    {
-      "epoch": 0.669304432593612,
-      "grad_norm": 0.0011399348732084036,
-      "learning_rate": 0.0001999997794409564,
-      "loss": 46.0,
-      "step": 8754
-    },
-    {
-      "epoch": 0.6693808895770017,
-      "grad_norm": 0.002236952306702733,
-      "learning_rate": 0.00019999977939050542,
-      "loss": 46.0,
-      "step": 8755
-    },
-    {
-      "epoch": 0.6694573465603915,
-      "grad_norm": 0.001290624146349728,
-      "learning_rate": 0.00019999977934004873,
-      "loss": 46.0,
-      "step": 8756
-    },
-    {
-      "epoch": 0.6695338035437812,
-      "grad_norm": 0.001181522966362536,
-      "learning_rate": 0.00019999977928958622,
-      "loss": 46.0,
-      "step": 8757
-    },
-    {
-      "epoch": 0.6696102605271709,
-      "grad_norm": 0.0014388780109584332,
-      "learning_rate": 0.00019999977923911794,
-      "loss": 46.0,
-      "step": 8758
-    },
-    {
-      "epoch": 0.6696867175105606,
-      "grad_norm": 0.002978184726089239,
-      "learning_rate": 0.0001999997791886439,
-      "loss": 46.0,
-      "step": 8759
-    },
-    {
-      "epoch": 0.6697631744939503,
-      "grad_norm": 0.00031134928576648235,
-      "learning_rate": 0.0001999997791381641,
-      "loss": 46.0,
-      "step": 8760
-    },
-    {
-      "epoch": 0.66983963147734,
-      "grad_norm": 0.0006983547355048358,
-      "learning_rate": 0.00019999977908767856,
-      "loss": 46.0,
-      "step": 8761
-    },
-    {
-      "epoch": 0.6699160884607298,
-      "grad_norm": 0.0008624501642771065,
-      "learning_rate": 0.0001999997790371872,
-      "loss": 46.0,
-      "step": 8762
-    },
-    {
-      "epoch": 0.6699925454441195,
-      "grad_norm": 0.001591513748280704,
-      "learning_rate": 0.0001999997789866901,
-      "loss": 46.0,
-      "step": 8763
-    },
-    {
-      "epoch": 0.6700690024275092,
-      "grad_norm": 0.0013368689687922597,
-      "learning_rate": 0.00019999977893618722,
-      "loss": 46.0,
-      "step": 8764
-    },
-    {
-      "epoch": 0.670145459410899,
-      "grad_norm": 0.0015445018652826548,
-      "learning_rate": 0.00019999977888567858,
-      "loss": 46.0,
-      "step": 8765
-    },
-    {
-      "epoch": 0.6702219163942886,
-      "grad_norm": 0.004002527333796024,
-      "learning_rate": 0.00019999977883516413,
-      "loss": 46.0,
-      "step": 8766
-    },
-    {
-      "epoch": 0.6702983733776784,
-      "grad_norm": 0.005256397649645805,
-      "learning_rate": 0.00019999977878464397,
-      "loss": 46.0,
-      "step": 8767
-    },
-    {
-      "epoch": 0.6703748303610682,
-      "grad_norm": 0.0008160103461705148,
-      "learning_rate": 0.00019999977873411798,
-      "loss": 46.0,
-      "step": 8768
-    },
-    {
-      "epoch": 0.6704512873444578,
-      "grad_norm": 0.002868467476218939,
-      "learning_rate": 0.00019999977868358627,
-      "loss": 46.0,
-      "step": 8769
-    },
-    {
-      "epoch": 0.6705277443278476,
-      "grad_norm": 0.000999022158794105,
-      "learning_rate": 0.0001999997786330488,
-      "loss": 46.0,
-      "step": 8770
-    },
-    {
-      "epoch": 0.6706042013112372,
-      "grad_norm": 0.0007276691612787545,
-      "learning_rate": 0.0001999997785825055,
-      "loss": 46.0,
-      "step": 8771
-    },
-    {
-      "epoch": 0.670680658294627,
-      "grad_norm": 0.005667826626449823,
-      "learning_rate": 0.00019999977853195648,
-      "loss": 46.0,
-      "step": 8772
-    },
-    {
-      "epoch": 0.6707571152780167,
-      "grad_norm": 0.0006296706269495189,
-      "learning_rate": 0.00019999977848140168,
-      "loss": 46.0,
-      "step": 8773
-    },
-    {
-      "epoch": 0.6708335722614064,
-      "grad_norm": 0.0021969785448163748,
-      "learning_rate": 0.0001999997784308411,
-      "loss": 46.0,
-      "step": 8774
-    },
-    {
-      "epoch": 0.6709100292447961,
-      "grad_norm": 0.001086077420040965,
-      "learning_rate": 0.00019999977838027478,
-      "loss": 46.0,
-      "step": 8775
-    },
-    {
-      "epoch": 0.6709864862281859,
-      "grad_norm": 0.0003675601037684828,
-      "learning_rate": 0.00019999977832970266,
-      "loss": 46.0,
-      "step": 8776
-    },
-    {
-      "epoch": 0.6710629432115756,
-      "grad_norm": 0.0008748634718358517,
-      "learning_rate": 0.00019999977827912476,
-      "loss": 46.0,
-      "step": 8777
-    },
-    {
-      "epoch": 0.6711394001949653,
-      "grad_norm": 0.003663973417133093,
-      "learning_rate": 0.00019999977822854112,
-      "loss": 46.0,
-      "step": 8778
-    },
-    {
-      "epoch": 0.6712158571783551,
-      "grad_norm": 0.006362304091453552,
-      "learning_rate": 0.00019999977817795168,
-      "loss": 46.0,
-      "step": 8779
-    },
-    {
-      "epoch": 0.6712923141617447,
-      "grad_norm": 0.01718462072312832,
-      "learning_rate": 0.00019999977812735652,
-      "loss": 46.0,
-      "step": 8780
-    },
-    {
-      "epoch": 0.6713687711451345,
-      "grad_norm": 0.006038782652467489,
-      "learning_rate": 0.00019999977807675555,
-      "loss": 46.0,
-      "step": 8781
-    },
-    {
-      "epoch": 0.6714452281285241,
-      "grad_norm": 0.0007000819896347821,
-      "learning_rate": 0.0001999997780261488,
-      "loss": 46.0,
-      "step": 8782
-    },
-    {
-      "epoch": 0.6715216851119139,
-      "grad_norm": 0.0009150332189165056,
-      "learning_rate": 0.0001999997779755363,
-      "loss": 46.0,
-      "step": 8783
-    },
-    {
-      "epoch": 0.6715981420953037,
-      "grad_norm": 0.0010293731465935707,
-      "learning_rate": 0.00019999977792491805,
-      "loss": 46.0,
-      "step": 8784
-    },
-    {
-      "epoch": 0.6716745990786933,
-      "grad_norm": 0.003607455873861909,
-      "learning_rate": 0.000199999777874294,
-      "loss": 46.0,
-      "step": 8785
-    },
-    {
-      "epoch": 0.6717510560620831,
-      "grad_norm": 0.001263849320821464,
-      "learning_rate": 0.0001999997778236642,
-      "loss": 46.0,
-      "step": 8786
-    },
-    {
-      "epoch": 0.6718275130454728,
-      "grad_norm": 0.001157538965344429,
-      "learning_rate": 0.00019999977777302862,
-      "loss": 46.0,
-      "step": 8787
-    },
-    {
-      "epoch": 0.6719039700288625,
-      "grad_norm": 0.01405640970915556,
-      "learning_rate": 0.00019999977772238727,
-      "loss": 46.0,
-      "step": 8788
-    },
-    {
-      "epoch": 0.6719804270122522,
-      "grad_norm": 0.001509845140390098,
-      "learning_rate": 0.00019999977767174018,
-      "loss": 46.0,
-      "step": 8789
-    },
-    {
-      "epoch": 0.672056883995642,
-      "grad_norm": 0.004036091733723879,
-      "learning_rate": 0.00019999977762108725,
-      "loss": 46.0,
-      "step": 8790
-    },
-    {
-      "epoch": 0.6721333409790317,
-      "grad_norm": 0.0017574878875166178,
-      "learning_rate": 0.00019999977757042864,
-      "loss": 46.0,
-      "step": 8791
-    },
-    {
-      "epoch": 0.6722097979624214,
-      "grad_norm": 0.004809914156794548,
-      "learning_rate": 0.00019999977751976417,
-      "loss": 46.0,
-      "step": 8792
-    },
-    {
-      "epoch": 0.6722862549458111,
-      "grad_norm": 0.0014662700705230236,
-      "learning_rate": 0.000199999777469094,
-      "loss": 46.0,
-      "step": 8793
-    },
-    {
-      "epoch": 0.6723627119292008,
-      "grad_norm": 0.0010704055894166231,
-      "learning_rate": 0.00019999977741841805,
-      "loss": 46.0,
-      "step": 8794
-    },
-    {
-      "epoch": 0.6724391689125906,
-      "grad_norm": 0.0007873719441704452,
-      "learning_rate": 0.0001999997773677363,
-      "loss": 46.0,
-      "step": 8795
-    },
-    {
-      "epoch": 0.6725156258959802,
-      "grad_norm": 0.0006021240260452032,
-      "learning_rate": 0.0001999997773170488,
-      "loss": 46.0,
-      "step": 8796
-    },
-    {
-      "epoch": 0.67259208287937,
-      "grad_norm": 0.0010698512196540833,
-      "learning_rate": 0.00019999977726635552,
-      "loss": 46.0,
-      "step": 8797
-    },
-    {
-      "epoch": 0.6726685398627598,
-      "grad_norm": 0.0013258474646136165,
-      "learning_rate": 0.0001999997772156565,
-      "loss": 46.0,
-      "step": 8798
-    },
-    {
-      "epoch": 0.6727449968461494,
-      "grad_norm": 0.005320283584296703,
-      "learning_rate": 0.0001999997771649517,
-      "loss": 46.0,
-      "step": 8799
-    },
-    {
-      "epoch": 0.6728214538295392,
-      "grad_norm": 0.000826517934910953,
-      "learning_rate": 0.0001999997771142411,
-      "loss": 46.0,
-      "step": 8800
-    },
-    {
-      "epoch": 0.6728979108129288,
-      "grad_norm": 0.0025309508200734854,
-      "learning_rate": 0.00019999977706352474,
-      "loss": 46.0,
-      "step": 8801
-    },
-    {
-      "epoch": 0.6729743677963186,
-      "grad_norm": 0.0007077916525304317,
-      "learning_rate": 0.00019999977701280264,
-      "loss": 46.0,
-      "step": 8802
-    },
-    {
-      "epoch": 0.6730508247797083,
-      "grad_norm": 0.0010015856241807342,
-      "learning_rate": 0.00019999977696207475,
-      "loss": 46.0,
-      "step": 8803
-    },
-    {
-      "epoch": 0.673127281763098,
-      "grad_norm": 0.0018026576144620776,
-      "learning_rate": 0.00019999977691134108,
-      "loss": 46.0,
-      "step": 8804
-    },
-    {
-      "epoch": 0.6732037387464878,
-      "grad_norm": 0.0009123882045969367,
-      "learning_rate": 0.00019999977686060167,
-      "loss": 46.0,
-      "step": 8805
-    },
-    {
-      "epoch": 0.6732801957298775,
-      "grad_norm": 0.0013137912610545754,
-      "learning_rate": 0.00019999977680985648,
-      "loss": 46.0,
-      "step": 8806
-    },
-    {
-      "epoch": 0.6733566527132672,
-      "grad_norm": 0.0033961795270442963,
-      "learning_rate": 0.0001999997767591055,
-      "loss": 46.0,
-      "step": 8807
-    },
-    {
-      "epoch": 0.6734331096966569,
-      "grad_norm": 0.0008632354438304901,
-      "learning_rate": 0.00019999977670834875,
-      "loss": 46.0,
-      "step": 8808
-    },
-    {
-      "epoch": 0.6735095666800467,
-      "grad_norm": 0.0010651986813172698,
-      "learning_rate": 0.00019999977665758625,
-      "loss": 46.0,
-      "step": 8809
-    },
-    {
-      "epoch": 0.6735860236634363,
-      "grad_norm": 0.001592569868080318,
-      "learning_rate": 0.00019999977660681796,
-      "loss": 46.0,
-      "step": 8810
-    },
-    {
-      "epoch": 0.6736624806468261,
-      "grad_norm": 0.007569534238427877,
-      "learning_rate": 0.00019999977655604394,
-      "loss": 46.0,
-      "step": 8811
-    },
-    {
-      "epoch": 0.6737389376302158,
-      "grad_norm": 0.0007504120003432035,
-      "learning_rate": 0.00019999977650526413,
-      "loss": 46.0,
-      "step": 8812
-    },
-    {
-      "epoch": 0.6738153946136055,
-      "grad_norm": 0.005444258451461792,
-      "learning_rate": 0.00019999977645447853,
-      "loss": 46.0,
-      "step": 8813
-    },
-    {
-      "epoch": 0.6738918515969953,
-      "grad_norm": 0.0027945602778345346,
-      "learning_rate": 0.00019999977640368719,
-      "loss": 46.0,
-      "step": 8814
-    },
-    {
-      "epoch": 0.6739683085803849,
-      "grad_norm": 0.0004073910531587899,
-      "learning_rate": 0.00019999977635289006,
-      "loss": 46.0,
-      "step": 8815
-    },
-    {
-      "epoch": 0.6740447655637747,
-      "grad_norm": 0.00032433439628221095,
-      "learning_rate": 0.0001999997763020872,
-      "loss": 46.0,
-      "step": 8816
-    },
-    {
-      "epoch": 0.6741212225471644,
-      "grad_norm": 0.005800970364362001,
-      "learning_rate": 0.0001999997762512785,
-      "loss": 46.0,
-      "step": 8817
-    },
-    {
-      "epoch": 0.6741976795305541,
-      "grad_norm": 0.004759421572089195,
-      "learning_rate": 0.00019999977620046409,
-      "loss": 46.0,
-      "step": 8818
-    },
-    {
-      "epoch": 0.6742741365139439,
-      "grad_norm": 0.0019921069033443928,
-      "learning_rate": 0.0001999997761496439,
-      "loss": 46.0,
-      "step": 8819
-    },
-    {
-      "epoch": 0.6743505934973336,
-      "grad_norm": 0.0009454063256271183,
-      "learning_rate": 0.00019999977609881794,
-      "loss": 46.0,
-      "step": 8820
-    },
-    {
-      "epoch": 0.6744270504807233,
-      "grad_norm": 0.0005759777850471437,
-      "learning_rate": 0.00019999977604798617,
-      "loss": 46.0,
-      "step": 8821
-    },
-    {
-      "epoch": 0.674503507464113,
-      "grad_norm": 0.006274321582168341,
-      "learning_rate": 0.00019999977599714867,
-      "loss": 46.0,
-      "step": 8822
-    },
-    {
-      "epoch": 0.6745799644475027,
-      "grad_norm": 0.0047287954948842525,
-      "learning_rate": 0.00019999977594630539,
-      "loss": 46.0,
-      "step": 8823
-    },
-    {
-      "epoch": 0.6746564214308924,
-      "grad_norm": 0.004649490118026733,
-      "learning_rate": 0.00019999977589545636,
-      "loss": 46.0,
-      "step": 8824
-    },
-    {
-      "epoch": 0.6747328784142822,
-      "grad_norm": 0.002104192739352584,
-      "learning_rate": 0.00019999977584460156,
-      "loss": 46.0,
-      "step": 8825
-    },
-    {
-      "epoch": 0.6748093353976719,
-      "grad_norm": 0.0007555439369753003,
-      "learning_rate": 0.00019999977579374093,
-      "loss": 46.0,
-      "step": 8826
-    },
-    {
-      "epoch": 0.6748857923810616,
-      "grad_norm": 0.0009805556619539857,
-      "learning_rate": 0.0001999997757428746,
-      "loss": 46.0,
-      "step": 8827
-    },
-    {
-      "epoch": 0.6749622493644514,
-      "grad_norm": 0.0012776469811797142,
-      "learning_rate": 0.00019999977569200249,
-      "loss": 46.0,
-      "step": 8828
-    },
-    {
-      "epoch": 0.675038706347841,
-      "grad_norm": 0.0030522739980369806,
-      "learning_rate": 0.00019999977564112456,
-      "loss": 46.0,
-      "step": 8829
-    },
-    {
-      "epoch": 0.6751151633312308,
-      "grad_norm": 0.0009938946459442377,
-      "learning_rate": 0.00019999977559024092,
-      "loss": 46.0,
-      "step": 8830
-    },
-    {
-      "epoch": 0.6751916203146204,
-      "grad_norm": 0.0013739357236772776,
-      "learning_rate": 0.0001999997755393515,
-      "loss": 46.0,
-      "step": 8831
-    },
-    {
-      "epoch": 0.6752680772980102,
-      "grad_norm": 0.0009283007238991559,
-      "learning_rate": 0.0001999997754884563,
-      "loss": 46.0,
-      "step": 8832
-    },
-    {
-      "epoch": 0.6753445342814,
-      "grad_norm": 0.0009329275926575065,
-      "learning_rate": 0.00019999977543755533,
-      "loss": 46.0,
-      "step": 8833
-    },
-    {
-      "epoch": 0.6754209912647896,
-      "grad_norm": 0.000396784336771816,
-      "learning_rate": 0.00019999977538664857,
-      "loss": 46.0,
-      "step": 8834
-    },
-    {
-      "epoch": 0.6754974482481794,
-      "grad_norm": 0.0003582837525755167,
-      "learning_rate": 0.00019999977533573606,
-      "loss": 46.0,
-      "step": 8835
-    },
-    {
-      "epoch": 0.6755739052315691,
-      "grad_norm": 0.0007325966143980622,
-      "learning_rate": 0.00019999977528481776,
-      "loss": 46.0,
-      "step": 8836
-    },
-    {
-      "epoch": 0.6756503622149588,
-      "grad_norm": 0.0003932283434551209,
-      "learning_rate": 0.00019999977523389376,
-      "loss": 46.0,
-      "step": 8837
-    },
-    {
-      "epoch": 0.6757268191983485,
-      "grad_norm": 0.0005122474394738674,
-      "learning_rate": 0.0001999997751829639,
-      "loss": 46.0,
-      "step": 8838
-    },
-    {
-      "epoch": 0.6758032761817383,
-      "grad_norm": 0.0011652312241494656,
-      "learning_rate": 0.00019999977513202833,
-      "loss": 46.0,
-      "step": 8839
-    },
-    {
-      "epoch": 0.675879733165128,
-      "grad_norm": 0.0018340147798880935,
-      "learning_rate": 0.00019999977508108695,
-      "loss": 46.0,
-      "step": 8840
-    },
-    {
-      "epoch": 0.6759561901485177,
-      "grad_norm": 0.0009766505099833012,
-      "learning_rate": 0.00019999977503013986,
-      "loss": 46.0,
-      "step": 8841
-    },
-    {
-      "epoch": 0.6760326471319074,
-      "grad_norm": 0.0013186981668695807,
-      "learning_rate": 0.00019999977497918696,
-      "loss": 46.0,
-      "step": 8842
-    },
-    {
-      "epoch": 0.6761091041152971,
-      "grad_norm": 0.0016813150141388178,
-      "learning_rate": 0.00019999977492822827,
-      "loss": 46.0,
-      "step": 8843
-    },
-    {
-      "epoch": 0.6761855610986869,
-      "grad_norm": 0.0020165392197668552,
-      "learning_rate": 0.00019999977487726386,
-      "loss": 46.0,
-      "step": 8844
-    },
-    {
-      "epoch": 0.6762620180820765,
-      "grad_norm": 0.0019567101262509823,
-      "learning_rate": 0.00019999977482629364,
-      "loss": 46.0,
-      "step": 8845
-    },
-    {
-      "epoch": 0.6763384750654663,
-      "grad_norm": 0.004047292750328779,
-      "learning_rate": 0.00019999977477531768,
-      "loss": 46.0,
-      "step": 8846
-    },
-    {
-      "epoch": 0.676414932048856,
-      "grad_norm": 0.0005070011247880757,
-      "learning_rate": 0.00019999977472433592,
-      "loss": 46.0,
-      "step": 8847
-    },
-    {
-      "epoch": 0.6764913890322457,
-      "grad_norm": 0.0006757982773706317,
-      "learning_rate": 0.0001999997746733484,
-      "loss": 46.0,
-      "step": 8848
-    },
-    {
-      "epoch": 0.6765678460156355,
-      "grad_norm": 0.0025839151348918676,
-      "learning_rate": 0.00019999977462235513,
-      "loss": 46.0,
-      "step": 8849
-    },
-    {
-      "epoch": 0.6766443029990252,
-      "grad_norm": 0.004773588385432959,
-      "learning_rate": 0.00019999977457135608,
-      "loss": 46.0,
-      "step": 8850
-    },
-    {
-      "epoch": 0.6767207599824149,
-      "grad_norm": 0.0009126248769462109,
-      "learning_rate": 0.00019999977452035125,
-      "loss": 46.0,
-      "step": 8851
-    },
-    {
-      "epoch": 0.6767972169658046,
-      "grad_norm": 0.0008942832937464118,
-      "learning_rate": 0.00019999977446934065,
-      "loss": 46.0,
-      "step": 8852
-    },
-    {
-      "epoch": 0.6768736739491943,
-      "grad_norm": 0.0007386683719232678,
-      "learning_rate": 0.0001999997744183243,
-      "loss": 46.0,
-      "step": 8853
-    },
-    {
-      "epoch": 0.676950130932584,
-      "grad_norm": 0.0030498644337058067,
-      "learning_rate": 0.00019999977436730218,
-      "loss": 46.0,
-      "step": 8854
-    },
-    {
-      "epoch": 0.6770265879159738,
-      "grad_norm": 0.0010214113863185048,
-      "learning_rate": 0.0001999997743162743,
-      "loss": 46.0,
-      "step": 8855
-    },
-    {
-      "epoch": 0.6771030448993635,
-      "grad_norm": 0.0022191761527210474,
-      "learning_rate": 0.0001999997742652406,
-      "loss": 46.0,
-      "step": 8856
-    },
-    {
-      "epoch": 0.6771795018827532,
-      "grad_norm": 0.0016494735609740019,
-      "learning_rate": 0.00019999977421420118,
-      "loss": 46.0,
-      "step": 8857
-    },
-    {
-      "epoch": 0.677255958866143,
-      "grad_norm": 0.00178253254853189,
-      "learning_rate": 0.00019999977416315597,
-      "loss": 46.0,
-      "step": 8858
-    },
-    {
-      "epoch": 0.6773324158495326,
-      "grad_norm": 0.0026668112259358168,
-      "learning_rate": 0.000199999774112105,
-      "loss": 46.0,
-      "step": 8859
-    },
-    {
-      "epoch": 0.6774088728329224,
-      "grad_norm": 0.001375620486214757,
-      "learning_rate": 0.00019999977406104824,
-      "loss": 46.0,
-      "step": 8860
-    },
-    {
-      "epoch": 0.677485329816312,
-      "grad_norm": 0.0012475703842937946,
-      "learning_rate": 0.0001999997740099857,
-      "loss": 46.0,
-      "step": 8861
-    },
-    {
-      "epoch": 0.6775617867997018,
-      "grad_norm": 0.0014985468005761504,
-      "learning_rate": 0.00019999977395891746,
-      "loss": 46.0,
-      "step": 8862
-    },
-    {
-      "epoch": 0.6776382437830916,
-      "grad_norm": 0.0013834508135914803,
-      "learning_rate": 0.00019999977390784338,
-      "loss": 46.0,
-      "step": 8863
-    },
-    {
-      "epoch": 0.6777147007664812,
-      "grad_norm": 0.000844665861222893,
-      "learning_rate": 0.00019999977385676355,
-      "loss": 46.0,
-      "step": 8864
-    },
-    {
-      "epoch": 0.677791157749871,
-      "grad_norm": 0.0010332379024475813,
-      "learning_rate": 0.00019999977380567797,
-      "loss": 46.0,
-      "step": 8865
-    },
-    {
-      "epoch": 0.6778676147332607,
-      "grad_norm": 0.00038638152182102203,
-      "learning_rate": 0.0001999997737545866,
-      "loss": 46.0,
-      "step": 8866
-    },
-    {
-      "epoch": 0.6779440717166504,
-      "grad_norm": 0.0012070524971932173,
-      "learning_rate": 0.00019999977370348945,
-      "loss": 46.0,
-      "step": 8867
-    },
-    {
-      "epoch": 0.6780205287000401,
-      "grad_norm": 0.0006206673569977283,
-      "learning_rate": 0.00019999977365238656,
-      "loss": 46.0,
-      "step": 8868
-    },
-    {
-      "epoch": 0.6780969856834299,
-      "grad_norm": 0.006029623094946146,
-      "learning_rate": 0.0001999997736012779,
-      "loss": 46.0,
-      "step": 8869
-    },
-    {
-      "epoch": 0.6781734426668196,
-      "grad_norm": 0.00040032059769146144,
-      "learning_rate": 0.00019999977355016345,
-      "loss": 46.0,
-      "step": 8870
-    },
-    {
-      "epoch": 0.6782498996502093,
-      "grad_norm": 0.0024517986457794905,
-      "learning_rate": 0.00019999977349904327,
-      "loss": 46.0,
-      "step": 8871
-    },
-    {
-      "epoch": 0.678326356633599,
-      "grad_norm": 0.0011035199277102947,
-      "learning_rate": 0.00019999977344791728,
-      "loss": 46.0,
-      "step": 8872
-    },
-    {
-      "epoch": 0.6784028136169887,
-      "grad_norm": 0.0012842753203585744,
-      "learning_rate": 0.00019999977339678552,
-      "loss": 46.0,
-      "step": 8873
-    },
-    {
-      "epoch": 0.6784792706003785,
-      "grad_norm": 0.0007621884578838944,
-      "learning_rate": 0.000199999773345648,
-      "loss": 46.0,
-      "step": 8874
-    },
-    {
-      "epoch": 0.6785557275837681,
-      "grad_norm": 0.0011028656736016273,
-      "learning_rate": 0.0001999997732945047,
-      "loss": 46.0,
-      "step": 8875
-    },
-    {
-      "epoch": 0.6786321845671579,
-      "grad_norm": 0.0025248867459595203,
-      "learning_rate": 0.00019999977324335565,
-      "loss": 46.0,
-      "step": 8876
-    },
-    {
-      "epoch": 0.6787086415505477,
-      "grad_norm": 0.002693133894354105,
-      "learning_rate": 0.00019999977319220085,
-      "loss": 46.0,
-      "step": 8877
-    },
-    {
-      "epoch": 0.6787850985339373,
-      "grad_norm": 0.0029521293472498655,
-      "learning_rate": 0.00019999977314104022,
-      "loss": 46.0,
-      "step": 8878
-    },
-    {
-      "epoch": 0.6788615555173271,
-      "grad_norm": 0.005102098919451237,
-      "learning_rate": 0.00019999977308987385,
-      "loss": 46.0,
-      "step": 8879
-    },
-    {
-      "epoch": 0.6789380125007168,
-      "grad_norm": 0.0018173924181610346,
-      "learning_rate": 0.00019999977303870173,
-      "loss": 46.0,
-      "step": 8880
-    },
-    {
-      "epoch": 0.6790144694841065,
-      "grad_norm": 0.0009427020559087396,
-      "learning_rate": 0.0001999997729875238,
-      "loss": 46.0,
-      "step": 8881
-    },
-    {
-      "epoch": 0.6790909264674962,
-      "grad_norm": 0.0032207202166318893,
-      "learning_rate": 0.00019999977293634017,
-      "loss": 46.0,
-      "step": 8882
-    },
-    {
-      "epoch": 0.6791673834508859,
-      "grad_norm": 0.003478718688711524,
-      "learning_rate": 0.00019999977288515073,
-      "loss": 46.0,
-      "step": 8883
-    },
-    {
-      "epoch": 0.6792438404342757,
-      "grad_norm": 0.0006191230495460331,
-      "learning_rate": 0.0001999997728339555,
-      "loss": 46.0,
-      "step": 8884
-    },
-    {
-      "epoch": 0.6793202974176654,
-      "grad_norm": 0.0007372137042693794,
-      "learning_rate": 0.0001999997727827545,
-      "loss": 46.0,
-      "step": 8885
-    },
-    {
-      "epoch": 0.6793967544010551,
-      "grad_norm": 0.0017002734821289778,
-      "learning_rate": 0.00019999977273154777,
-      "loss": 46.0,
-      "step": 8886
-    },
-    {
-      "epoch": 0.6794732113844448,
-      "grad_norm": 0.0017562441062182188,
-      "learning_rate": 0.00019999977268033524,
-      "loss": 46.0,
-      "step": 8887
-    },
-    {
-      "epoch": 0.6795496683678346,
-      "grad_norm": 0.001531782909296453,
-      "learning_rate": 0.00019999977262911696,
-      "loss": 46.0,
-      "step": 8888
-    },
-    {
-      "epoch": 0.6796261253512242,
-      "grad_norm": 0.0022313494700938463,
-      "learning_rate": 0.0001999997725778929,
-      "loss": 46.0,
-      "step": 8889
-    },
-    {
-      "epoch": 0.679702582334614,
-      "grad_norm": 0.00450655585154891,
-      "learning_rate": 0.00019999977252666308,
-      "loss": 46.0,
-      "step": 8890
-    },
-    {
-      "epoch": 0.6797790393180037,
-      "grad_norm": 0.0012120966566726565,
-      "learning_rate": 0.00019999977247542745,
-      "loss": 46.0,
-      "step": 8891
-    },
-    {
-      "epoch": 0.6798554963013934,
-      "grad_norm": 0.0006009733187966049,
-      "learning_rate": 0.00019999977242418608,
-      "loss": 46.0,
-      "step": 8892
-    },
-    {
-      "epoch": 0.6799319532847832,
-      "grad_norm": 0.008062581531703472,
-      "learning_rate": 0.00019999977237293896,
-      "loss": 46.0,
-      "step": 8893
-    },
-    {
-      "epoch": 0.6800084102681728,
-      "grad_norm": 0.0005266519146971405,
-      "learning_rate": 0.00019999977232168606,
-      "loss": 46.0,
-      "step": 8894
-    },
-    {
-      "epoch": 0.6800848672515626,
-      "grad_norm": 0.0008005801355466247,
-      "learning_rate": 0.0001999997722704274,
-      "loss": 46.0,
-      "step": 8895
-    },
-    {
-      "epoch": 0.6801613242349523,
-      "grad_norm": 0.000989048508927226,
-      "learning_rate": 0.00019999977221916293,
-      "loss": 46.0,
-      "step": 8896
-    },
-    {
-      "epoch": 0.680237781218342,
-      "grad_norm": 0.001267099753022194,
-      "learning_rate": 0.00019999977216789272,
-      "loss": 46.0,
-      "step": 8897
-    },
-    {
-      "epoch": 0.6803142382017318,
-      "grad_norm": 0.002835105871781707,
-      "learning_rate": 0.00019999977211661673,
-      "loss": 46.0,
-      "step": 8898
-    },
-    {
-      "epoch": 0.6803906951851215,
-      "grad_norm": 0.0022383895702660084,
-      "learning_rate": 0.00019999977206533497,
-      "loss": 46.0,
-      "step": 8899
-    },
-    {
-      "epoch": 0.6804671521685112,
-      "grad_norm": 0.0006995209841988981,
-      "learning_rate": 0.00019999977201404744,
-      "loss": 46.0,
-      "step": 8900
-    },
-    {
-      "epoch": 0.6805436091519009,
-      "grad_norm": 0.00225619925186038,
-      "learning_rate": 0.00019999977196275416,
-      "loss": 46.0,
-      "step": 8901
-    },
-    {
-      "epoch": 0.6806200661352906,
-      "grad_norm": 0.0011576348915696144,
-      "learning_rate": 0.0001999997719114551,
-      "loss": 46.0,
-      "step": 8902
-    },
-    {
-      "epoch": 0.6806965231186803,
-      "grad_norm": 0.0013492741854861379,
-      "learning_rate": 0.00019999977186015025,
-      "loss": 46.0,
-      "step": 8903
-    },
-    {
-      "epoch": 0.6807729801020701,
-      "grad_norm": 0.005938329268246889,
-      "learning_rate": 0.00019999977180883965,
-      "loss": 46.0,
-      "step": 8904
-    },
-    {
-      "epoch": 0.6808494370854598,
-      "grad_norm": 0.004824361763894558,
-      "learning_rate": 0.00019999977175752328,
-      "loss": 46.0,
-      "step": 8905
-    },
-    {
-      "epoch": 0.6809258940688495,
-      "grad_norm": 0.0023215790279209614,
-      "learning_rate": 0.00019999977170620113,
-      "loss": 46.0,
-      "step": 8906
-    },
-    {
-      "epoch": 0.6810023510522393,
-      "grad_norm": 0.004163461271673441,
-      "learning_rate": 0.00019999977165487324,
-      "loss": 46.0,
-      "step": 8907
-    },
-    {
-      "epoch": 0.6810788080356289,
-      "grad_norm": 0.0015719932271167636,
-      "learning_rate": 0.00019999977160353955,
-      "loss": 46.0,
-      "step": 8908
-    },
-    {
-      "epoch": 0.6811552650190187,
-      "grad_norm": 0.0004061298677697778,
-      "learning_rate": 0.0001999997715522001,
-      "loss": 46.0,
-      "step": 8909
-    },
-    {
-      "epoch": 0.6812317220024084,
-      "grad_norm": 0.0009165135561488569,
-      "learning_rate": 0.0001999997715008549,
-      "loss": 46.0,
-      "step": 8910
-    },
-    {
-      "epoch": 0.6813081789857981,
-      "grad_norm": 0.0004233189392834902,
-      "learning_rate": 0.0001999997714495039,
-      "loss": 46.0,
-      "step": 8911
-    },
-    {
-      "epoch": 0.6813846359691879,
-      "grad_norm": 0.0029324786737561226,
-      "learning_rate": 0.00019999977139814715,
-      "loss": 46.0,
-      "step": 8912
-    },
-    {
-      "epoch": 0.6814610929525775,
-      "grad_norm": 0.0029001713264733553,
-      "learning_rate": 0.00019999977134678462,
-      "loss": 46.0,
-      "step": 8913
-    },
-    {
-      "epoch": 0.6815375499359673,
-      "grad_norm": 0.0070222048088908195,
-      "learning_rate": 0.00019999977129541634,
-      "loss": 46.0,
-      "step": 8914
-    },
-    {
-      "epoch": 0.681614006919357,
-      "grad_norm": 0.003727530362084508,
-      "learning_rate": 0.00019999977124404226,
-      "loss": 46.0,
-      "step": 8915
-    },
-    {
-      "epoch": 0.6816904639027467,
-      "grad_norm": 0.0010502394288778305,
-      "learning_rate": 0.00019999977119266243,
-      "loss": 46.0,
-      "step": 8916
-    },
-    {
-      "epoch": 0.6817669208861364,
-      "grad_norm": 0.0012994263088330626,
-      "learning_rate": 0.00019999977114127683,
-      "loss": 46.0,
-      "step": 8917
-    },
-    {
-      "epoch": 0.6818433778695262,
-      "grad_norm": 0.0008250034297816455,
-      "learning_rate": 0.00019999977108988546,
-      "loss": 46.0,
-      "step": 8918
-    },
-    {
-      "epoch": 0.6819198348529159,
-      "grad_norm": 0.0052488106302917,
-      "learning_rate": 0.0001999997710384883,
-      "loss": 46.0,
-      "step": 8919
-    },
-    {
-      "epoch": 0.6819962918363056,
-      "grad_norm": 0.0008516021771356463,
-      "learning_rate": 0.00019999977098708537,
-      "loss": 46.0,
-      "step": 8920
-    },
-    {
-      "epoch": 0.6820727488196954,
-      "grad_norm": 0.0012623557122424245,
-      "learning_rate": 0.00019999977093567673,
-      "loss": 46.0,
-      "step": 8921
-    },
-    {
-      "epoch": 0.682149205803085,
-      "grad_norm": 0.0024201348423957825,
-      "learning_rate": 0.00019999977088426226,
-      "loss": 46.0,
-      "step": 8922
-    },
-    {
-      "epoch": 0.6822256627864748,
-      "grad_norm": 0.001519442885182798,
-      "learning_rate": 0.00019999977083284205,
-      "loss": 46.0,
-      "step": 8923
-    },
-    {
-      "epoch": 0.6823021197698644,
-      "grad_norm": 0.0008020659442991018,
-      "learning_rate": 0.00019999977078141606,
-      "loss": 46.0,
-      "step": 8924
-    },
-    {
-      "epoch": 0.6823785767532542,
-      "grad_norm": 0.001165450899861753,
-      "learning_rate": 0.00019999977072998428,
-      "loss": 46.0,
-      "step": 8925
-    },
-    {
-      "epoch": 0.682455033736644,
-      "grad_norm": 0.0011818933999165893,
-      "learning_rate": 0.00019999977067854677,
-      "loss": 46.0,
-      "step": 8926
-    },
-    {
-      "epoch": 0.6825314907200336,
-      "grad_norm": 0.0005220667808316648,
-      "learning_rate": 0.00019999977062710347,
-      "loss": 46.0,
-      "step": 8927
-    },
-    {
-      "epoch": 0.6826079477034234,
-      "grad_norm": 0.007757478393614292,
-      "learning_rate": 0.00019999977057565439,
-      "loss": 46.0,
-      "step": 8928
-    },
-    {
-      "epoch": 0.6826844046868131,
-      "grad_norm": 0.0008062376291491091,
-      "learning_rate": 0.00019999977052419956,
-      "loss": 46.0,
-      "step": 8929
-    },
-    {
-      "epoch": 0.6827608616702028,
-      "grad_norm": 0.001988654024899006,
-      "learning_rate": 0.00019999977047273893,
-      "loss": 46.0,
-      "step": 8930
-    },
-    {
-      "epoch": 0.6828373186535925,
-      "grad_norm": 0.006442064419388771,
-      "learning_rate": 0.00019999977042127256,
-      "loss": 46.0,
-      "step": 8931
-    },
-    {
-      "epoch": 0.6829137756369822,
-      "grad_norm": 0.00390633475035429,
-      "learning_rate": 0.00019999977036980042,
-      "loss": 46.0,
-      "step": 8932
-    },
-    {
-      "epoch": 0.682990232620372,
-      "grad_norm": 0.0003620848001446575,
-      "learning_rate": 0.0001999997703183225,
-      "loss": 46.0,
-      "step": 8933
-    },
-    {
-      "epoch": 0.6830666896037617,
-      "grad_norm": 0.0018519113073125482,
-      "learning_rate": 0.00019999977026683883,
-      "loss": 46.0,
-      "step": 8934
-    },
-    {
-      "epoch": 0.6831431465871514,
-      "grad_norm": 0.00046496186405420303,
-      "learning_rate": 0.00019999977021534937,
-      "loss": 46.0,
-      "step": 8935
-    },
-    {
-      "epoch": 0.6832196035705411,
-      "grad_norm": 0.001361010828986764,
-      "learning_rate": 0.00019999977016385415,
-      "loss": 46.0,
-      "step": 8936
-    },
-    {
-      "epoch": 0.6832960605539309,
-      "grad_norm": 0.0026377395261079073,
-      "learning_rate": 0.00019999977011235317,
-      "loss": 46.0,
-      "step": 8937
-    },
-    {
-      "epoch": 0.6833725175373205,
-      "grad_norm": 0.0013580245431512594,
-      "learning_rate": 0.00019999977006084638,
-      "loss": 46.0,
-      "step": 8938
-    },
-    {
-      "epoch": 0.6834489745207103,
-      "grad_norm": 0.0012145794462412596,
-      "learning_rate": 0.00019999977000933385,
-      "loss": 46.0,
-      "step": 8939
-    },
-    {
-      "epoch": 0.6835254315041001,
-      "grad_norm": 0.0017297255108132958,
-      "learning_rate": 0.00019999976995781554,
-      "loss": 46.0,
-      "step": 8940
-    },
-    {
-      "epoch": 0.6836018884874897,
-      "grad_norm": 0.0010634642094373703,
-      "learning_rate": 0.0001999997699062915,
-      "loss": 46.0,
-      "step": 8941
-    },
-    {
-      "epoch": 0.6836783454708795,
-      "grad_norm": 0.0006453762180171907,
-      "learning_rate": 0.00019999976985476164,
-      "loss": 46.0,
-      "step": 8942
-    },
-    {
-      "epoch": 0.6837548024542691,
-      "grad_norm": 0.002811832819133997,
-      "learning_rate": 0.00019999976980322604,
-      "loss": 46.0,
-      "step": 8943
-    },
-    {
-      "epoch": 0.6838312594376589,
-      "grad_norm": 0.0016413466073572636,
-      "learning_rate": 0.00019999976975168467,
-      "loss": 46.0,
-      "step": 8944
-    },
-    {
-      "epoch": 0.6839077164210486,
-      "grad_norm": 0.00123019446618855,
-      "learning_rate": 0.0001999997697001375,
-      "loss": 46.0,
-      "step": 8945
-    },
-    {
-      "epoch": 0.6839841734044383,
-      "grad_norm": 0.0018856578972190619,
-      "learning_rate": 0.0001999997696485846,
-      "loss": 46.0,
-      "step": 8946
-    },
-    {
-      "epoch": 0.684060630387828,
-      "grad_norm": 0.0008849318255670369,
-      "learning_rate": 0.00019999976959702588,
-      "loss": 46.0,
-      "step": 8947
-    },
-    {
-      "epoch": 0.6841370873712178,
-      "grad_norm": 0.0006442017038352787,
-      "learning_rate": 0.00019999976954546145,
-      "loss": 46.0,
-      "step": 8948
-    },
-    {
-      "epoch": 0.6842135443546075,
-      "grad_norm": 0.0009710351005196571,
-      "learning_rate": 0.00019999976949389123,
-      "loss": 46.0,
-      "step": 8949
-    },
-    {
-      "epoch": 0.6842900013379972,
-      "grad_norm": 0.0006140318582765758,
-      "learning_rate": 0.00019999976944231522,
-      "loss": 46.0,
-      "step": 8950
-    },
-    {
-      "epoch": 0.684366458321387,
-      "grad_norm": 0.0018044919706881046,
-      "learning_rate": 0.00019999976939073346,
-      "loss": 46.0,
-      "step": 8951
-    },
-    {
-      "epoch": 0.6844429153047766,
-      "grad_norm": 0.007445411756634712,
-      "learning_rate": 0.00019999976933914593,
-      "loss": 46.0,
-      "step": 8952
-    },
-    {
-      "epoch": 0.6845193722881664,
-      "grad_norm": 0.0022069085389375687,
-      "learning_rate": 0.00019999976928755263,
-      "loss": 46.0,
-      "step": 8953
-    },
-    {
-      "epoch": 0.684595829271556,
-      "grad_norm": 0.0034716997761279345,
-      "learning_rate": 0.00019999976923595355,
-      "loss": 46.0,
-      "step": 8954
-    },
-    {
-      "epoch": 0.6846722862549458,
-      "grad_norm": 0.00046213821042329073,
-      "learning_rate": 0.0001999997691843487,
-      "loss": 46.0,
-      "step": 8955
-    },
-    {
-      "epoch": 0.6847487432383356,
-      "grad_norm": 0.002667208667844534,
-      "learning_rate": 0.00019999976913273807,
-      "loss": 46.0,
-      "step": 8956
-    },
-    {
-      "epoch": 0.6848252002217252,
-      "grad_norm": 0.0014564176090061665,
-      "learning_rate": 0.00019999976908112173,
-      "loss": 46.0,
-      "step": 8957
-    },
-    {
-      "epoch": 0.684901657205115,
-      "grad_norm": 0.001079651527106762,
-      "learning_rate": 0.00019999976902949955,
-      "loss": 46.0,
-      "step": 8958
-    },
-    {
-      "epoch": 0.6849781141885047,
-      "grad_norm": 0.002777957124635577,
-      "learning_rate": 0.00019999976897787166,
-      "loss": 46.0,
-      "step": 8959
-    },
-    {
-      "epoch": 0.6850545711718944,
-      "grad_norm": 0.00052559346659109,
-      "learning_rate": 0.00019999976892623794,
-      "loss": 46.0,
-      "step": 8960
-    },
-    {
-      "epoch": 0.6851310281552841,
-      "grad_norm": 0.0010202423436567187,
-      "learning_rate": 0.00019999976887459848,
-      "loss": 46.0,
-      "step": 8961
-    },
-    {
-      "epoch": 0.6852074851386738,
-      "grad_norm": 0.0005036952206864953,
-      "learning_rate": 0.00019999976882295327,
-      "loss": 46.0,
-      "step": 8962
-    },
-    {
-      "epoch": 0.6852839421220636,
-      "grad_norm": 0.01150734256953001,
-      "learning_rate": 0.00019999976877130225,
-      "loss": 46.0,
-      "step": 8963
-    },
-    {
-      "epoch": 0.6853603991054533,
-      "grad_norm": 0.0013397192815318704,
-      "learning_rate": 0.0001999997687196455,
-      "loss": 46.0,
-      "step": 8964
-    },
-    {
-      "epoch": 0.685436856088843,
-      "grad_norm": 0.0007708031916990876,
-      "learning_rate": 0.00019999976866798296,
-      "loss": 46.0,
-      "step": 8965
-    },
-    {
-      "epoch": 0.6855133130722327,
-      "grad_norm": 0.015683762729167938,
-      "learning_rate": 0.00019999976861631463,
-      "loss": 46.0,
-      "step": 8966
-    },
-    {
-      "epoch": 0.6855897700556225,
-      "grad_norm": 0.005421179346740246,
-      "learning_rate": 0.00019999976856464055,
-      "loss": 46.0,
-      "step": 8967
-    },
-    {
-      "epoch": 0.6856662270390121,
-      "grad_norm": 0.001229059649631381,
-      "learning_rate": 0.00019999976851296073,
-      "loss": 46.0,
-      "step": 8968
-    },
-    {
-      "epoch": 0.6857426840224019,
-      "grad_norm": 0.0017709207022562623,
-      "learning_rate": 0.0001999997684612751,
-      "loss": 46.0,
-      "step": 8969
-    },
-    {
-      "epoch": 0.6858191410057917,
-      "grad_norm": 0.001023811404593289,
-      "learning_rate": 0.00019999976840958373,
-      "loss": 46.0,
-      "step": 8970
-    },
-    {
-      "epoch": 0.6858955979891813,
-      "grad_norm": 0.0012827676255255938,
-      "learning_rate": 0.00019999976835788656,
-      "loss": 46.0,
-      "step": 8971
-    },
-    {
-      "epoch": 0.6859720549725711,
-      "grad_norm": 0.0009215387399308383,
-      "learning_rate": 0.00019999976830618364,
-      "loss": 46.0,
-      "step": 8972
-    },
-    {
-      "epoch": 0.6860485119559607,
-      "grad_norm": 0.0006364753935486078,
-      "learning_rate": 0.00019999976825447495,
-      "loss": 46.0,
-      "step": 8973
-    },
-    {
-      "epoch": 0.6861249689393505,
-      "grad_norm": 0.004577720537781715,
-      "learning_rate": 0.00019999976820276049,
-      "loss": 46.0,
-      "step": 8974
-    },
-    {
-      "epoch": 0.6862014259227402,
-      "grad_norm": 0.003056132700294256,
-      "learning_rate": 0.00019999976815104025,
-      "loss": 46.0,
-      "step": 8975
-    },
-    {
-      "epoch": 0.6862778829061299,
-      "grad_norm": 0.0026472159661352634,
-      "learning_rate": 0.00019999976809931424,
-      "loss": 46.0,
-      "step": 8976
-    },
-    {
-      "epoch": 0.6863543398895197,
-      "grad_norm": 0.0005493867793120444,
-      "learning_rate": 0.00019999976804758248,
-      "loss": 46.0,
-      "step": 8977
-    },
-    {
-      "epoch": 0.6864307968729094,
-      "grad_norm": 0.0009548880043439567,
-      "learning_rate": 0.00019999976799584495,
-      "loss": 46.0,
-      "step": 8978
-    },
-    {
-      "epoch": 0.6865072538562991,
-      "grad_norm": 0.0005695368745364249,
-      "learning_rate": 0.00019999976794410164,
-      "loss": 46.0,
-      "step": 8979
-    },
-    {
-      "epoch": 0.6865837108396888,
-      "grad_norm": 0.0012504211626946926,
-      "learning_rate": 0.00019999976789235256,
-      "loss": 46.0,
-      "step": 8980
-    },
-    {
-      "epoch": 0.6866601678230786,
-      "grad_norm": 0.0008027313160710037,
-      "learning_rate": 0.0001999997678405977,
-      "loss": 46.0,
-      "step": 8981
-    },
-    {
-      "epoch": 0.6867366248064682,
-      "grad_norm": 0.0011888340814039111,
-      "learning_rate": 0.0001999997677888371,
-      "loss": 46.0,
-      "step": 8982
-    },
-    {
-      "epoch": 0.686813081789858,
-      "grad_norm": 0.0024902618024498224,
-      "learning_rate": 0.00019999976773707072,
-      "loss": 46.0,
-      "step": 8983
-    },
-    {
-      "epoch": 0.6868895387732477,
-      "grad_norm": 0.0011751026613637805,
-      "learning_rate": 0.00019999976768529854,
-      "loss": 46.0,
-      "step": 8984
-    },
-    {
-      "epoch": 0.6869659957566374,
-      "grad_norm": 0.00040528629324398935,
-      "learning_rate": 0.00019999976763352063,
-      "loss": 46.0,
-      "step": 8985
-    },
-    {
-      "epoch": 0.6870424527400272,
-      "grad_norm": 0.000988964457064867,
-      "learning_rate": 0.00019999976758173693,
-      "loss": 46.0,
-      "step": 8986
-    },
-    {
-      "epoch": 0.6871189097234168,
-      "grad_norm": 0.017140034586191177,
-      "learning_rate": 0.00019999976752994747,
-      "loss": 46.0,
-      "step": 8987
-    },
-    {
-      "epoch": 0.6871953667068066,
-      "grad_norm": 0.00042539279093034565,
-      "learning_rate": 0.00019999976747815223,
-      "loss": 46.0,
-      "step": 8988
-    },
-    {
-      "epoch": 0.6872718236901963,
-      "grad_norm": 0.0017999152187258005,
-      "learning_rate": 0.00019999976742635122,
-      "loss": 46.0,
-      "step": 8989
-    },
-    {
-      "epoch": 0.687348280673586,
-      "grad_norm": 0.0004646909364964813,
-      "learning_rate": 0.00019999976737454444,
-      "loss": 46.0,
-      "step": 8990
-    },
-    {
-      "epoch": 0.6874247376569758,
-      "grad_norm": 0.0021967217326164246,
-      "learning_rate": 0.00019999976732273188,
-      "loss": 46.0,
-      "step": 8991
-    },
-    {
-      "epoch": 0.6875011946403654,
-      "grad_norm": 0.0008120503625832498,
-      "learning_rate": 0.0001999997672709136,
-      "loss": 46.0,
-      "step": 8992
-    },
-    {
-      "epoch": 0.6875776516237552,
-      "grad_norm": 0.00040943597559817135,
-      "learning_rate": 0.0001999997672190895,
-      "loss": 46.0,
-      "step": 8993
-    },
-    {
-      "epoch": 0.6876541086071449,
-      "grad_norm": 0.001688003190793097,
-      "learning_rate": 0.00019999976716725964,
-      "loss": 46.0,
-      "step": 8994
-    },
-    {
-      "epoch": 0.6877305655905346,
-      "grad_norm": 0.000609360751695931,
-      "learning_rate": 0.00019999976711542402,
-      "loss": 46.0,
-      "step": 8995
-    },
-    {
-      "epoch": 0.6878070225739243,
-      "grad_norm": 0.0012643372174352407,
-      "learning_rate": 0.00019999976706358265,
-      "loss": 46.0,
-      "step": 8996
-    },
-    {
-      "epoch": 0.6878834795573141,
-      "grad_norm": 0.002649756846949458,
-      "learning_rate": 0.00019999976701173548,
-      "loss": 46.0,
-      "step": 8997
-    },
-    {
-      "epoch": 0.6879599365407038,
-      "grad_norm": 0.0009037270210683346,
-      "learning_rate": 0.00019999976695988256,
-      "loss": 46.0,
-      "step": 8998
-    },
-    {
-      "epoch": 0.6880363935240935,
-      "grad_norm": 0.001698114792816341,
-      "learning_rate": 0.00019999976690802384,
-      "loss": 46.0,
-      "step": 8999
-    },
-    {
-      "epoch": 0.6881128505074833,
-      "grad_norm": 0.0005454815691336989,
-      "learning_rate": 0.0001999997668561594,
-      "loss": 46.0,
-      "step": 9000
-    },
-    {
-      "epoch": 0.6881893074908729,
-      "grad_norm": 0.0011061440454795957,
-      "learning_rate": 0.00019999976680428917,
-      "loss": 46.0,
-      "step": 9001
-    },
-    {
-      "epoch": 0.6882657644742627,
-      "grad_norm": 0.0007159507949836552,
-      "learning_rate": 0.00019999976675241316,
-      "loss": 46.0,
-      "step": 9002
-    },
-    {
-      "epoch": 0.6883422214576523,
-      "grad_norm": 0.004143567755818367,
-      "learning_rate": 0.00019999976670053137,
-      "loss": 46.0,
-      "step": 9003
-    },
-    {
-      "epoch": 0.6884186784410421,
-      "grad_norm": 0.0030515280086547136,
-      "learning_rate": 0.00019999976664864381,
-      "loss": 46.0,
-      "step": 9004
-    },
-    {
-      "epoch": 0.6884951354244319,
-      "grad_norm": 0.0023069658782333136,
-      "learning_rate": 0.00019999976659675054,
-      "loss": 46.0,
-      "step": 9005
-    },
-    {
-      "epoch": 0.6885715924078215,
-      "grad_norm": 0.002587481401860714,
-      "learning_rate": 0.00019999976654485143,
-      "loss": 46.0,
-      "step": 9006
-    },
-    {
-      "epoch": 0.6886480493912113,
-      "grad_norm": 0.0004480658390093595,
-      "learning_rate": 0.00019999976649294658,
-      "loss": 46.0,
-      "step": 9007
-    },
-    {
-      "epoch": 0.688724506374601,
-      "grad_norm": 0.006968831643462181,
-      "learning_rate": 0.00019999976644103596,
-      "loss": 46.0,
-      "step": 9008
-    },
-    {
-      "epoch": 0.6888009633579907,
-      "grad_norm": 0.00040929080569185317,
-      "learning_rate": 0.00019999976638911956,
-      "loss": 46.0,
-      "step": 9009
-    },
-    {
-      "epoch": 0.6888774203413804,
-      "grad_norm": 0.0018370988545939326,
-      "learning_rate": 0.0001999997663371974,
-      "loss": 46.0,
-      "step": 9010
-    },
-    {
-      "epoch": 0.6889538773247702,
-      "grad_norm": 0.0027835245709866285,
-      "learning_rate": 0.00019999976628526947,
-      "loss": 46.0,
-      "step": 9011
-    },
-    {
-      "epoch": 0.6890303343081599,
-      "grad_norm": 0.00038406450767070055,
-      "learning_rate": 0.00019999976623333575,
-      "loss": 46.0,
-      "step": 9012
-    },
-    {
-      "epoch": 0.6891067912915496,
-      "grad_norm": 0.0006710010347887874,
-      "learning_rate": 0.0001999997661813963,
-      "loss": 46.0,
-      "step": 9013
-    },
-    {
-      "epoch": 0.6891832482749393,
-      "grad_norm": 0.003953936044126749,
-      "learning_rate": 0.00019999976612945102,
-      "loss": 46.0,
-      "step": 9014
-    },
-    {
-      "epoch": 0.689259705258329,
-      "grad_norm": 0.001941786496900022,
-      "learning_rate": 0.00019999976607750004,
-      "loss": 46.0,
-      "step": 9015
-    },
-    {
-      "epoch": 0.6893361622417188,
-      "grad_norm": 0.0009593384456820786,
-      "learning_rate": 0.00019999976602554326,
-      "loss": 46.0,
-      "step": 9016
-    },
-    {
-      "epoch": 0.6894126192251084,
-      "grad_norm": 0.0019117763731628656,
-      "learning_rate": 0.0001999997659735807,
-      "loss": 46.0,
-      "step": 9017
-    },
-    {
-      "epoch": 0.6894890762084982,
-      "grad_norm": 0.009926487691700459,
-      "learning_rate": 0.0001999997659216124,
-      "loss": 46.0,
-      "step": 9018
-    },
-    {
-      "epoch": 0.689565533191888,
-      "grad_norm": 0.0008937942329794168,
-      "learning_rate": 0.00019999976586963832,
-      "loss": 46.0,
-      "step": 9019
-    },
-    {
-      "epoch": 0.6896419901752776,
-      "grad_norm": 0.0005348284030333161,
-      "learning_rate": 0.00019999976581765844,
-      "loss": 46.0,
-      "step": 9020
-    },
-    {
-      "epoch": 0.6897184471586674,
-      "grad_norm": 0.0007350753294304013,
-      "learning_rate": 0.00019999976576567284,
-      "loss": 46.0,
-      "step": 9021
-    },
-    {
-      "epoch": 0.6897949041420571,
-      "grad_norm": 0.0008200214942917228,
-      "learning_rate": 0.00019999976571368145,
-      "loss": 46.0,
-      "step": 9022
-    },
-    {
-      "epoch": 0.6898713611254468,
-      "grad_norm": 0.0008654919802211225,
-      "learning_rate": 0.00019999976566168428,
-      "loss": 46.0,
-      "step": 9023
-    },
-    {
-      "epoch": 0.6899478181088365,
-      "grad_norm": 0.0009617213509045541,
-      "learning_rate": 0.00019999976560968133,
-      "loss": 46.0,
-      "step": 9024
-    },
-    {
-      "epoch": 0.6900242750922262,
-      "grad_norm": 0.002543957205489278,
-      "learning_rate": 0.00019999976555767264,
-      "loss": 46.0,
-      "step": 9025
-    },
-    {
-      "epoch": 0.690100732075616,
-      "grad_norm": 0.000933670555241406,
-      "learning_rate": 0.00019999976550565815,
-      "loss": 46.0,
-      "step": 9026
-    },
-    {
-      "epoch": 0.6901771890590057,
-      "grad_norm": 0.0017495944630354643,
-      "learning_rate": 0.0001999997654536379,
-      "loss": 46.0,
-      "step": 9027
-    },
-    {
-      "epoch": 0.6902536460423954,
-      "grad_norm": 0.00033723912201821804,
-      "learning_rate": 0.0001999997654016119,
-      "loss": 46.0,
-      "step": 9028
-    },
-    {
-      "epoch": 0.6903301030257851,
-      "grad_norm": 0.0016255220398306847,
-      "learning_rate": 0.00019999976534958012,
-      "loss": 46.0,
-      "step": 9029
-    },
-    {
-      "epoch": 0.6904065600091749,
-      "grad_norm": 0.0008925366564653814,
-      "learning_rate": 0.00019999976529754256,
-      "loss": 46.0,
-      "step": 9030
-    },
-    {
-      "epoch": 0.6904830169925645,
-      "grad_norm": 0.0007710131467320025,
-      "learning_rate": 0.00019999976524549923,
-      "loss": 46.0,
-      "step": 9031
-    },
-    {
-      "epoch": 0.6905594739759543,
-      "grad_norm": 0.0025350386276841164,
-      "learning_rate": 0.00019999976519345015,
-      "loss": 46.0,
-      "step": 9032
-    },
-    {
-      "epoch": 0.690635930959344,
-      "grad_norm": 0.007017096038907766,
-      "learning_rate": 0.0001999997651413953,
-      "loss": 46.0,
-      "step": 9033
-    },
-    {
-      "epoch": 0.6907123879427337,
-      "grad_norm": 0.0012645142851397395,
-      "learning_rate": 0.00019999976508933465,
-      "loss": 46.0,
-      "step": 9034
-    },
-    {
-      "epoch": 0.6907888449261235,
-      "grad_norm": 0.0017301932675763965,
-      "learning_rate": 0.00019999976503726828,
-      "loss": 46.0,
-      "step": 9035
-    },
-    {
-      "epoch": 0.6908653019095131,
-      "grad_norm": 0.001873756991699338,
-      "learning_rate": 0.00019999976498519611,
-      "loss": 46.0,
-      "step": 9036
-    },
-    {
-      "epoch": 0.6909417588929029,
-      "grad_norm": 0.0015052204253152013,
-      "learning_rate": 0.00019999976493311817,
-      "loss": 46.0,
-      "step": 9037
-    },
-    {
-      "epoch": 0.6910182158762926,
-      "grad_norm": 0.0005426730494946241,
-      "learning_rate": 0.00019999976488103448,
-      "loss": 46.0,
-      "step": 9038
-    },
-    {
-      "epoch": 0.6910946728596823,
-      "grad_norm": 0.0009391424828208983,
-      "learning_rate": 0.000199999764828945,
-      "loss": 46.0,
-      "step": 9039
-    },
-    {
-      "epoch": 0.691171129843072,
-      "grad_norm": 0.0008600588771514595,
-      "learning_rate": 0.00019999976477684975,
-      "loss": 46.0,
-      "step": 9040
-    },
-    {
-      "epoch": 0.6912475868264618,
-      "grad_norm": 0.005597707349807024,
-      "learning_rate": 0.00019999976472474872,
-      "loss": 46.0,
-      "step": 9041
-    },
-    {
-      "epoch": 0.6913240438098515,
-      "grad_norm": 0.0007188781164586544,
-      "learning_rate": 0.00019999976467264193,
-      "loss": 46.0,
-      "step": 9042
-    },
-    {
-      "epoch": 0.6914005007932412,
-      "grad_norm": 0.0011846552370116115,
-      "learning_rate": 0.00019999976462052938,
-      "loss": 46.0,
-      "step": 9043
-    },
-    {
-      "epoch": 0.6914769577766309,
-      "grad_norm": 0.0013120861258357763,
-      "learning_rate": 0.00019999976456841107,
-      "loss": 46.0,
-      "step": 9044
-    },
-    {
-      "epoch": 0.6915534147600206,
-      "grad_norm": 0.0013271295465528965,
-      "learning_rate": 0.00019999976451628697,
-      "loss": 46.0,
-      "step": 9045
-    },
-    {
-      "epoch": 0.6916298717434104,
-      "grad_norm": 0.002557573840022087,
-      "learning_rate": 0.0001999997644641571,
-      "loss": 46.0,
-      "step": 9046
-    },
-    {
-      "epoch": 0.6917063287268,
-      "grad_norm": 0.0007955206092447042,
-      "learning_rate": 0.00019999976441202147,
-      "loss": 46.0,
-      "step": 9047
-    },
-    {
-      "epoch": 0.6917827857101898,
-      "grad_norm": 0.008305965922772884,
-      "learning_rate": 0.00019999976435988007,
-      "loss": 46.0,
-      "step": 9048
-    },
-    {
-      "epoch": 0.6918592426935796,
-      "grad_norm": 0.0017414254834875464,
-      "learning_rate": 0.0001999997643077329,
-      "loss": 46.0,
-      "step": 9049
-    },
-    {
-      "epoch": 0.6919356996769692,
-      "grad_norm": 0.0029711704701185226,
-      "learning_rate": 0.00019999976425557996,
-      "loss": 46.0,
-      "step": 9050
-    },
-    {
-      "epoch": 0.692012156660359,
-      "grad_norm": 0.0006545691285282373,
-      "learning_rate": 0.00019999976420342127,
-      "loss": 46.0,
-      "step": 9051
-    },
-    {
-      "epoch": 0.6920886136437487,
-      "grad_norm": 0.0005934067303314805,
-      "learning_rate": 0.00019999976415125675,
-      "loss": 46.0,
-      "step": 9052
-    },
-    {
-      "epoch": 0.6921650706271384,
-      "grad_norm": 0.0013071164721623063,
-      "learning_rate": 0.00019999976409908654,
-      "loss": 46.0,
-      "step": 9053
-    },
-    {
-      "epoch": 0.6922415276105282,
-      "grad_norm": 0.0006706046406179667,
-      "learning_rate": 0.0001999997640469105,
-      "loss": 46.0,
-      "step": 9054
-    },
-    {
-      "epoch": 0.6923179845939178,
-      "grad_norm": 0.005196481943130493,
-      "learning_rate": 0.00019999976399472872,
-      "loss": 46.0,
-      "step": 9055
-    },
-    {
-      "epoch": 0.6923944415773076,
-      "grad_norm": 0.0010553019819781184,
-      "learning_rate": 0.00019999976394254116,
-      "loss": 46.0,
-      "step": 9056
-    },
-    {
-      "epoch": 0.6924708985606973,
-      "grad_norm": 0.0028756125830113888,
-      "learning_rate": 0.00019999976389034783,
-      "loss": 46.0,
-      "step": 9057
-    },
-    {
-      "epoch": 0.692547355544087,
-      "grad_norm": 0.004338743630796671,
-      "learning_rate": 0.00019999976383814873,
-      "loss": 46.0,
-      "step": 9058
-    },
-    {
-      "epoch": 0.6926238125274767,
-      "grad_norm": 0.0021503502503037453,
-      "learning_rate": 0.00019999976378594388,
-      "loss": 46.0,
-      "step": 9059
-    },
-    {
-      "epoch": 0.6927002695108665,
-      "grad_norm": 0.00505692046135664,
-      "learning_rate": 0.00019999976373373323,
-      "loss": 46.0,
-      "step": 9060
-    },
-    {
-      "epoch": 0.6927767264942561,
-      "grad_norm": 0.0008375363540835679,
-      "learning_rate": 0.00019999976368151684,
-      "loss": 46.0,
-      "step": 9061
-    },
-    {
-      "epoch": 0.6928531834776459,
-      "grad_norm": 0.000604130094870925,
-      "learning_rate": 0.00019999976362929467,
-      "loss": 46.0,
-      "step": 9062
-    },
-    {
-      "epoch": 0.6929296404610356,
-      "grad_norm": 0.00730582932010293,
-      "learning_rate": 0.00019999976357706672,
-      "loss": 46.0,
-      "step": 9063
-    },
-    {
-      "epoch": 0.6930060974444253,
-      "grad_norm": 0.0010035594459623098,
-      "learning_rate": 0.000199999763524833,
-      "loss": 46.0,
-      "step": 9064
-    },
-    {
-      "epoch": 0.6930825544278151,
-      "grad_norm": 0.0008801851072348654,
-      "learning_rate": 0.00019999976347259352,
-      "loss": 46.0,
-      "step": 9065
-    },
-    {
-      "epoch": 0.6931590114112047,
-      "grad_norm": 0.000766176322940737,
-      "learning_rate": 0.00019999976342034828,
-      "loss": 46.0,
-      "step": 9066
-    },
-    {
-      "epoch": 0.6932354683945945,
-      "grad_norm": 0.0018666146788746119,
-      "learning_rate": 0.00019999976336809727,
-      "loss": 46.0,
-      "step": 9067
-    },
-    {
-      "epoch": 0.6933119253779843,
-      "grad_norm": 0.002000202191993594,
-      "learning_rate": 0.00019999976331584046,
-      "loss": 46.0,
-      "step": 9068
-    },
-    {
-      "epoch": 0.6933883823613739,
-      "grad_norm": 0.007076604757457972,
-      "learning_rate": 0.0001999997632635779,
-      "loss": 46.0,
-      "step": 9069
-    },
-    {
-      "epoch": 0.6934648393447637,
-      "grad_norm": 0.0018796473741531372,
-      "learning_rate": 0.00019999976321130957,
-      "loss": 46.0,
-      "step": 9070
-    },
-    {
-      "epoch": 0.6935412963281534,
-      "grad_norm": 0.0034080531913787127,
-      "learning_rate": 0.00019999976315903547,
-      "loss": 46.0,
-      "step": 9071
-    },
-    {
-      "epoch": 0.6936177533115431,
-      "grad_norm": 0.0007652313215658069,
-      "learning_rate": 0.00019999976310675562,
-      "loss": 46.0,
-      "step": 9072
-    },
-    {
-      "epoch": 0.6936942102949328,
-      "grad_norm": 0.026359155774116516,
-      "learning_rate": 0.00019999976305446997,
-      "loss": 46.0,
-      "step": 9073
-    },
-    {
-      "epoch": 0.6937706672783225,
-      "grad_norm": 0.0022392745595425367,
-      "learning_rate": 0.00019999976300217858,
-      "loss": 46.0,
-      "step": 9074
-    },
-    {
-      "epoch": 0.6938471242617122,
-      "grad_norm": 0.0008756750612519681,
-      "learning_rate": 0.0001999997629498814,
-      "loss": 46.0,
-      "step": 9075
-    },
-    {
-      "epoch": 0.693923581245102,
-      "grad_norm": 0.0017607525223866105,
-      "learning_rate": 0.00019999976289757844,
-      "loss": 46.0,
-      "step": 9076
-    },
-    {
-      "epoch": 0.6940000382284917,
-      "grad_norm": 0.0018702505622059107,
-      "learning_rate": 0.00019999976284526972,
-      "loss": 46.0,
-      "step": 9077
-    },
-    {
-      "epoch": 0.6940764952118814,
-      "grad_norm": 0.0010729047935456038,
-      "learning_rate": 0.00019999976279295526,
-      "loss": 46.0,
-      "step": 9078
-    },
-    {
-      "epoch": 0.6941529521952712,
-      "grad_norm": 0.0004929465940222144,
-      "learning_rate": 0.00019999976274063502,
-      "loss": 46.0,
-      "step": 9079
-    },
-    {
-      "epoch": 0.6942294091786608,
-      "grad_norm": 0.020234225317835808,
-      "learning_rate": 0.000199999762688309,
-      "loss": 46.0,
-      "step": 9080
-    },
-    {
-      "epoch": 0.6943058661620506,
-      "grad_norm": 0.000776316097471863,
-      "learning_rate": 0.00019999976263597718,
-      "loss": 46.0,
-      "step": 9081
-    },
-    {
-      "epoch": 0.6943823231454403,
-      "grad_norm": 0.0022005108185112476,
-      "learning_rate": 0.00019999976258363962,
-      "loss": 46.0,
-      "step": 9082
-    },
-    {
-      "epoch": 0.69445878012883,
-      "grad_norm": 0.001105863368138671,
-      "learning_rate": 0.0001999997625312963,
-      "loss": 46.0,
-      "step": 9083
-    },
-    {
-      "epoch": 0.6945352371122198,
-      "grad_norm": 0.005786404013633728,
-      "learning_rate": 0.0001999997624789472,
-      "loss": 46.0,
-      "step": 9084
-    },
-    {
-      "epoch": 0.6946116940956094,
-      "grad_norm": 0.0029700950253754854,
-      "learning_rate": 0.0001999997624265923,
-      "loss": 46.0,
-      "step": 9085
-    },
-    {
-      "epoch": 0.6946881510789992,
-      "grad_norm": 0.0004369357484392822,
-      "learning_rate": 0.0001999997623742317,
-      "loss": 46.0,
-      "step": 9086
-    },
-    {
-      "epoch": 0.6947646080623889,
-      "grad_norm": 0.0018057163106277585,
-      "learning_rate": 0.00019999976232186527,
-      "loss": 46.0,
-      "step": 9087
-    },
-    {
-      "epoch": 0.6948410650457786,
-      "grad_norm": 0.001960674999281764,
-      "learning_rate": 0.00019999976226949307,
-      "loss": 46.0,
-      "step": 9088
-    },
-    {
-      "epoch": 0.6949175220291683,
-      "grad_norm": 0.004416829440742731,
-      "learning_rate": 0.00019999976221711513,
-      "loss": 46.0,
-      "step": 9089
-    },
-    {
-      "epoch": 0.6949939790125581,
-      "grad_norm": 0.0008817437337711453,
-      "learning_rate": 0.0001999997621647314,
-      "loss": 46.0,
-      "step": 9090
-    },
-    {
-      "epoch": 0.6950704359959478,
-      "grad_norm": 0.002371092326939106,
-      "learning_rate": 0.00019999976211234195,
-      "loss": 46.0,
-      "step": 9091
-    },
-    {
-      "epoch": 0.6951468929793375,
-      "grad_norm": 0.0018999106250703335,
-      "learning_rate": 0.0001999997620599467,
-      "loss": 46.0,
-      "step": 9092
-    },
-    {
-      "epoch": 0.6952233499627272,
-      "grad_norm": 0.0011110103223472834,
-      "learning_rate": 0.00019999976200754565,
-      "loss": 46.0,
-      "step": 9093
-    },
-    {
-      "epoch": 0.6952998069461169,
-      "grad_norm": 0.0004885838716290891,
-      "learning_rate": 0.00019999976195513884,
-      "loss": 46.0,
-      "step": 9094
-    },
-    {
-      "epoch": 0.6953762639295067,
-      "grad_norm": 0.003622288117185235,
-      "learning_rate": 0.0001999997619027263,
-      "loss": 46.0,
-      "step": 9095
-    },
-    {
-      "epoch": 0.6954527209128963,
-      "grad_norm": 0.0008509297040291131,
-      "learning_rate": 0.00019999976185030796,
-      "loss": 46.0,
-      "step": 9096
-    },
-    {
-      "epoch": 0.6955291778962861,
-      "grad_norm": 0.0036821668036282063,
-      "learning_rate": 0.00019999976179788386,
-      "loss": 46.0,
-      "step": 9097
-    },
-    {
-      "epoch": 0.6956056348796759,
-      "grad_norm": 0.0013152624014765024,
-      "learning_rate": 0.00019999976174545398,
-      "loss": 46.0,
-      "step": 9098
-    },
-    {
-      "epoch": 0.6956820918630655,
-      "grad_norm": 0.0007734769606031477,
-      "learning_rate": 0.00019999976169301836,
-      "loss": 46.0,
-      "step": 9099
-    },
-    {
-      "epoch": 0.6957585488464553,
-      "grad_norm": 0.0007958488422445953,
-      "learning_rate": 0.0001999997616405769,
-      "loss": 46.0,
-      "step": 9100
-    },
-    {
-      "epoch": 0.695835005829845,
-      "grad_norm": 0.0016632117331027985,
-      "learning_rate": 0.00019999976158812974,
-      "loss": 46.0,
-      "step": 9101
-    },
-    {
-      "epoch": 0.6959114628132347,
-      "grad_norm": 0.0016646585427224636,
-      "learning_rate": 0.0001999997615356768,
-      "loss": 46.0,
-      "step": 9102
-    },
-    {
-      "epoch": 0.6959879197966244,
-      "grad_norm": 0.000806248455774039,
-      "learning_rate": 0.00019999976148321808,
-      "loss": 46.0,
-      "step": 9103
-    },
-    {
-      "epoch": 0.6960643767800141,
-      "grad_norm": 0.0009553435957059264,
-      "learning_rate": 0.0001999997614307536,
-      "loss": 46.0,
-      "step": 9104
-    },
-    {
-      "epoch": 0.6961408337634039,
-      "grad_norm": 0.0034235992934554815,
-      "learning_rate": 0.0001999997613782833,
-      "loss": 46.0,
-      "step": 9105
-    },
-    {
-      "epoch": 0.6962172907467936,
-      "grad_norm": 0.0015158647438511252,
-      "learning_rate": 0.0001999997613258073,
-      "loss": 46.0,
-      "step": 9106
-    },
-    {
-      "epoch": 0.6962937477301833,
-      "grad_norm": 0.002113425638526678,
-      "learning_rate": 0.00019999976127332549,
-      "loss": 46.0,
-      "step": 9107
-    },
-    {
-      "epoch": 0.696370204713573,
-      "grad_norm": 0.003075334010645747,
-      "learning_rate": 0.00019999976122083793,
-      "loss": 46.0,
-      "step": 9108
-    },
-    {
-      "epoch": 0.6964466616969628,
-      "grad_norm": 0.0024250419810414314,
-      "learning_rate": 0.00019999976116834458,
-      "loss": 46.0,
-      "step": 9109
-    },
-    {
-      "epoch": 0.6965231186803524,
-      "grad_norm": 0.001397660467773676,
-      "learning_rate": 0.00019999976111584547,
-      "loss": 46.0,
-      "step": 9110
-    },
-    {
-      "epoch": 0.6965995756637422,
-      "grad_norm": 0.002280839253216982,
-      "learning_rate": 0.00019999976106334062,
-      "loss": 46.0,
-      "step": 9111
-    },
-    {
-      "epoch": 0.696676032647132,
-      "grad_norm": 0.0025200205855071545,
-      "learning_rate": 0.00019999976101082995,
-      "loss": 46.0,
-      "step": 9112
-    },
-    {
-      "epoch": 0.6967524896305216,
-      "grad_norm": 0.003927274141460657,
-      "learning_rate": 0.00019999976095831355,
-      "loss": 46.0,
-      "step": 9113
-    },
-    {
-      "epoch": 0.6968289466139114,
-      "grad_norm": 0.001499826554208994,
-      "learning_rate": 0.00019999976090579136,
-      "loss": 46.0,
-      "step": 9114
-    },
-    {
-      "epoch": 0.696905403597301,
-      "grad_norm": 0.005215655546635389,
-      "learning_rate": 0.00019999976085326342,
-      "loss": 46.0,
-      "step": 9115
-    },
-    {
-      "epoch": 0.6969818605806908,
-      "grad_norm": 0.0010160794481635094,
-      "learning_rate": 0.0001999997608007297,
-      "loss": 46.0,
-      "step": 9116
-    },
-    {
-      "epoch": 0.6970583175640805,
-      "grad_norm": 0.00043587910477072,
-      "learning_rate": 0.0001999997607481902,
-      "loss": 46.0,
-      "step": 9117
-    },
-    {
-      "epoch": 0.6971347745474702,
-      "grad_norm": 0.021117571741342545,
-      "learning_rate": 0.00019999976069564492,
-      "loss": 46.0,
-      "step": 9118
-    },
-    {
-      "epoch": 0.69721123153086,
-      "grad_norm": 0.0010300461435690522,
-      "learning_rate": 0.0001999997606430939,
-      "loss": 46.0,
-      "step": 9119
-    },
-    {
-      "epoch": 0.6972876885142497,
-      "grad_norm": 0.0002722138015087694,
-      "learning_rate": 0.0001999997605905371,
-      "loss": 46.0,
-      "step": 9120
-    },
-    {
-      "epoch": 0.6973641454976394,
-      "grad_norm": 0.0016803985927253962,
-      "learning_rate": 0.00019999976053797452,
-      "loss": 46.0,
-      "step": 9121
-    },
-    {
-      "epoch": 0.6974406024810291,
-      "grad_norm": 0.0006275622290559113,
-      "learning_rate": 0.0001999997604854062,
-      "loss": 46.0,
-      "step": 9122
-    },
-    {
-      "epoch": 0.6975170594644188,
-      "grad_norm": 0.002100080018863082,
-      "learning_rate": 0.0001999997604328321,
-      "loss": 46.0,
-      "step": 9123
-    },
-    {
-      "epoch": 0.6975935164478085,
-      "grad_norm": 0.002103107515722513,
-      "learning_rate": 0.00019999976038025222,
-      "loss": 46.0,
-      "step": 9124
-    },
-    {
-      "epoch": 0.6976699734311983,
-      "grad_norm": 0.0014708503149449825,
-      "learning_rate": 0.00019999976032766657,
-      "loss": 46.0,
-      "step": 9125
-    },
-    {
-      "epoch": 0.697746430414588,
-      "grad_norm": 0.015570555813610554,
-      "learning_rate": 0.00019999976027507515,
-      "loss": 46.0,
-      "step": 9126
-    },
-    {
-      "epoch": 0.6978228873979777,
-      "grad_norm": 0.0009374309447593987,
-      "learning_rate": 0.00019999976022247795,
-      "loss": 46.0,
-      "step": 9127
-    },
-    {
-      "epoch": 0.6978993443813675,
-      "grad_norm": 0.001767691457644105,
-      "learning_rate": 0.000199999760169875,
-      "loss": 46.0,
-      "step": 9128
-    },
-    {
-      "epoch": 0.6979758013647571,
-      "grad_norm": 0.00047994355554692447,
-      "learning_rate": 0.0001999997601172663,
-      "loss": 46.0,
-      "step": 9129
-    },
-    {
-      "epoch": 0.6980522583481469,
-      "grad_norm": 0.0019308087648823857,
-      "learning_rate": 0.00019999976006465178,
-      "loss": 46.0,
-      "step": 9130
-    },
-    {
-      "epoch": 0.6981287153315366,
-      "grad_norm": 0.0012461959850043058,
-      "learning_rate": 0.00019999976001203152,
-      "loss": 46.0,
-      "step": 9131
-    },
-    {
-      "epoch": 0.6982051723149263,
-      "grad_norm": 0.0004380648897495121,
-      "learning_rate": 0.00019999975995940548,
-      "loss": 46.0,
-      "step": 9132
-    },
-    {
-      "epoch": 0.698281629298316,
-      "grad_norm": 0.0007305003819055855,
-      "learning_rate": 0.00019999975990677368,
-      "loss": 46.0,
-      "step": 9133
-    },
-    {
-      "epoch": 0.6983580862817057,
-      "grad_norm": 0.0006421058787964284,
-      "learning_rate": 0.0001999997598541361,
-      "loss": 46.0,
-      "step": 9134
-    },
-    {
-      "epoch": 0.6984345432650955,
-      "grad_norm": 0.0012774879578500986,
-      "learning_rate": 0.00019999975980149274,
-      "loss": 46.0,
-      "step": 9135
-    },
-    {
-      "epoch": 0.6985110002484852,
-      "grad_norm": 0.008576181717216969,
-      "learning_rate": 0.00019999975974884367,
-      "loss": 46.0,
-      "step": 9136
-    },
-    {
-      "epoch": 0.6985874572318749,
-      "grad_norm": 0.0006904984475113451,
-      "learning_rate": 0.00019999975969618876,
-      "loss": 46.0,
-      "step": 9137
-    },
-    {
-      "epoch": 0.6986639142152646,
-      "grad_norm": 0.0010577266803011298,
-      "learning_rate": 0.00019999975964352811,
-      "loss": 46.0,
-      "step": 9138
-    },
-    {
-      "epoch": 0.6987403711986544,
-      "grad_norm": 0.0019243519054725766,
-      "learning_rate": 0.0001999997595908617,
-      "loss": 46.0,
-      "step": 9139
-    },
-    {
-      "epoch": 0.698816828182044,
-      "grad_norm": 0.001724465168081224,
-      "learning_rate": 0.0001999997595381895,
-      "loss": 46.0,
-      "step": 9140
-    },
-    {
-      "epoch": 0.6988932851654338,
-      "grad_norm": 0.0019569003488868475,
-      "learning_rate": 0.00019999975948551156,
-      "loss": 46.0,
-      "step": 9141
-    },
-    {
-      "epoch": 0.6989697421488236,
-      "grad_norm": 0.0033014847431331873,
-      "learning_rate": 0.00019999975943282782,
-      "loss": 46.0,
-      "step": 9142
-    },
-    {
-      "epoch": 0.6990461991322132,
-      "grad_norm": 0.0007960941875353456,
-      "learning_rate": 0.00019999975938013833,
-      "loss": 46.0,
-      "step": 9143
-    },
-    {
-      "epoch": 0.699122656115603,
-      "grad_norm": 0.009571060538291931,
-      "learning_rate": 0.00019999975932744307,
-      "loss": 46.0,
-      "step": 9144
-    },
-    {
-      "epoch": 0.6991991130989926,
-      "grad_norm": 0.0039282264187932014,
-      "learning_rate": 0.000199999759274742,
-      "loss": 46.0,
-      "step": 9145
-    },
-    {
-      "epoch": 0.6992755700823824,
-      "grad_norm": 0.0022329073399305344,
-      "learning_rate": 0.0001999997592220352,
-      "loss": 46.0,
-      "step": 9146
-    },
-    {
-      "epoch": 0.6993520270657722,
-      "grad_norm": 0.001143449218943715,
-      "learning_rate": 0.00019999975916932264,
-      "loss": 46.0,
-      "step": 9147
-    },
-    {
-      "epoch": 0.6994284840491618,
-      "grad_norm": 0.0013885560911148787,
-      "learning_rate": 0.0001999997591166043,
-      "loss": 46.0,
-      "step": 9148
-    },
-    {
-      "epoch": 0.6995049410325516,
-      "grad_norm": 0.004010376520454884,
-      "learning_rate": 0.00019999975906388019,
-      "loss": 46.0,
-      "step": 9149
-    },
-    {
-      "epoch": 0.6995813980159413,
-      "grad_norm": 0.013738351874053478,
-      "learning_rate": 0.00019999975901115028,
-      "loss": 46.0,
-      "step": 9150
-    },
-    {
-      "epoch": 0.699657854999331,
-      "grad_norm": 0.002059882739558816,
-      "learning_rate": 0.00019999975895841466,
-      "loss": 46.0,
-      "step": 9151
-    },
-    {
-      "epoch": 0.6997343119827207,
-      "grad_norm": 0.000560998625587672,
-      "learning_rate": 0.00019999975890567322,
-      "loss": 46.0,
-      "step": 9152
-    },
-    {
-      "epoch": 0.6998107689661105,
-      "grad_norm": 0.0019092325819656253,
-      "learning_rate": 0.00019999975885292602,
-      "loss": 46.0,
-      "step": 9153
-    },
-    {
-      "epoch": 0.6998872259495001,
-      "grad_norm": 0.00193807203322649,
-      "learning_rate": 0.00019999975880017305,
-      "loss": 46.0,
-      "step": 9154
-    },
-    {
-      "epoch": 0.6999636829328899,
-      "grad_norm": 0.0005407581920735538,
-      "learning_rate": 0.0001999997587474143,
-      "loss": 46.0,
-      "step": 9155
-    },
-    {
-      "epoch": 0.7000401399162796,
-      "grad_norm": 0.0022905785590410233,
-      "learning_rate": 0.00019999975869464982,
-      "loss": 46.0,
-      "step": 9156
-    },
-    {
-      "epoch": 0.7001165968996693,
-      "grad_norm": 0.0008564593154005706,
-      "learning_rate": 0.00019999975864187956,
-      "loss": 46.0,
-      "step": 9157
-    },
-    {
-      "epoch": 0.7001930538830591,
-      "grad_norm": 0.00206946418620646,
-      "learning_rate": 0.00019999975858910353,
-      "loss": 46.0,
-      "step": 9158
-    },
-    {
-      "epoch": 0.7002695108664487,
-      "grad_norm": 0.0030055183451622725,
-      "learning_rate": 0.00019999975853632172,
-      "loss": 46.0,
-      "step": 9159
-    },
-    {
-      "epoch": 0.7003459678498385,
-      "grad_norm": 0.00043971664854325354,
-      "learning_rate": 0.00019999975848353414,
-      "loss": 46.0,
-      "step": 9160
-    },
-    {
-      "epoch": 0.7004224248332283,
-      "grad_norm": 0.0006825425662100315,
-      "learning_rate": 0.00019999975843074078,
-      "loss": 46.0,
-      "step": 9161
-    },
-    {
-      "epoch": 0.7004988818166179,
-      "grad_norm": 0.001231030561029911,
-      "learning_rate": 0.00019999975837794168,
-      "loss": 46.0,
-      "step": 9162
-    },
-    {
-      "epoch": 0.7005753388000077,
-      "grad_norm": 0.0006522739422507584,
-      "learning_rate": 0.00019999975832513678,
-      "loss": 46.0,
-      "step": 9163
-    },
-    {
-      "epoch": 0.7006517957833973,
-      "grad_norm": 0.0005894287023693323,
-      "learning_rate": 0.00019999975827232613,
-      "loss": 46.0,
-      "step": 9164
-    },
-    {
-      "epoch": 0.7007282527667871,
-      "grad_norm": 0.0002801214868668467,
-      "learning_rate": 0.00019999975821950971,
-      "loss": 46.0,
-      "step": 9165
-    },
-    {
-      "epoch": 0.7008047097501768,
-      "grad_norm": 0.0009813535725697875,
-      "learning_rate": 0.00019999975816668752,
-      "loss": 46.0,
-      "step": 9166
-    },
-    {
-      "epoch": 0.7008811667335665,
-      "grad_norm": 0.002128161024302244,
-      "learning_rate": 0.00019999975811385952,
-      "loss": 46.0,
-      "step": 9167
-    },
-    {
-      "epoch": 0.7009576237169562,
-      "grad_norm": 0.005774526856839657,
-      "learning_rate": 0.0001999997580610258,
-      "loss": 46.0,
-      "step": 9168
-    },
-    {
-      "epoch": 0.701034080700346,
-      "grad_norm": 0.00047978476504795253,
-      "learning_rate": 0.00019999975800818632,
-      "loss": 46.0,
-      "step": 9169
-    },
-    {
-      "epoch": 0.7011105376837357,
-      "grad_norm": 0.0010162268299609423,
-      "learning_rate": 0.00019999975795534104,
-      "loss": 46.0,
-      "step": 9170
-    },
-    {
-      "epoch": 0.7011869946671254,
-      "grad_norm": 0.0006295115454122424,
-      "learning_rate": 0.00019999975790249,
-      "loss": 46.0,
-      "step": 9171
-    },
-    {
-      "epoch": 0.7012634516505152,
-      "grad_norm": 0.004258931614458561,
-      "learning_rate": 0.0001999997578496332,
-      "loss": 46.0,
-      "step": 9172
-    },
-    {
-      "epoch": 0.7013399086339048,
-      "grad_norm": 0.0014116342645138502,
-      "learning_rate": 0.00019999975779677059,
-      "loss": 46.0,
-      "step": 9173
-    },
-    {
-      "epoch": 0.7014163656172946,
-      "grad_norm": 0.0009487050119787455,
-      "learning_rate": 0.00019999975774390226,
-      "loss": 46.0,
-      "step": 9174
-    },
-    {
-      "epoch": 0.7014928226006842,
-      "grad_norm": 0.0024084998294711113,
-      "learning_rate": 0.00019999975769102813,
-      "loss": 46.0,
-      "step": 9175
-    },
-    {
-      "epoch": 0.701569279584074,
-      "grad_norm": 0.006702831480652094,
-      "learning_rate": 0.00019999975763814823,
-      "loss": 46.0,
-      "step": 9176
-    },
-    {
-      "epoch": 0.7016457365674638,
-      "grad_norm": 0.004745201673358679,
-      "learning_rate": 0.00019999975758526258,
-      "loss": 46.0,
-      "step": 9177
-    },
-    {
-      "epoch": 0.7017221935508534,
-      "grad_norm": 0.0005848890286870301,
-      "learning_rate": 0.00019999975753237114,
-      "loss": 46.0,
-      "step": 9178
-    },
-    {
-      "epoch": 0.7017986505342432,
-      "grad_norm": 0.0007510999566875398,
-      "learning_rate": 0.00019999975747947394,
-      "loss": 46.0,
-      "step": 9179
-    },
-    {
-      "epoch": 0.7018751075176329,
-      "grad_norm": 0.004274856299161911,
-      "learning_rate": 0.00019999975742657097,
-      "loss": 46.0,
-      "step": 9180
-    },
-    {
-      "epoch": 0.7019515645010226,
-      "grad_norm": 0.001858519040979445,
-      "learning_rate": 0.00019999975737366223,
-      "loss": 46.0,
-      "step": 9181
-    },
-    {
-      "epoch": 0.7020280214844123,
-      "grad_norm": 0.001513935625553131,
-      "learning_rate": 0.00019999975732074775,
-      "loss": 46.0,
-      "step": 9182
-    },
-    {
-      "epoch": 0.7021044784678021,
-      "grad_norm": 0.0012599901529029012,
-      "learning_rate": 0.00019999975726782746,
-      "loss": 46.0,
-      "step": 9183
-    },
-    {
-      "epoch": 0.7021809354511918,
-      "grad_norm": 0.0012123690685257316,
-      "learning_rate": 0.0001999997572149014,
-      "loss": 46.0,
-      "step": 9184
-    },
-    {
-      "epoch": 0.7022573924345815,
-      "grad_norm": 0.0011055892100557685,
-      "learning_rate": 0.00019999975716196962,
-      "loss": 46.0,
-      "step": 9185
-    },
-    {
-      "epoch": 0.7023338494179712,
-      "grad_norm": 0.0008655319106765091,
-      "learning_rate": 0.00019999975710903204,
-      "loss": 46.0,
-      "step": 9186
-    },
-    {
-      "epoch": 0.7024103064013609,
-      "grad_norm": 0.001162463566288352,
-      "learning_rate": 0.00019999975705608866,
-      "loss": 46.0,
-      "step": 9187
-    },
-    {
-      "epoch": 0.7024867633847507,
-      "grad_norm": 0.0008038083324208856,
-      "learning_rate": 0.00019999975700313956,
-      "loss": 46.0,
-      "step": 9188
-    },
-    {
-      "epoch": 0.7025632203681403,
-      "grad_norm": 0.0011236275313422084,
-      "learning_rate": 0.00019999975695018468,
-      "loss": 46.0,
-      "step": 9189
-    },
-    {
-      "epoch": 0.7026396773515301,
-      "grad_norm": 0.0024578431621193886,
-      "learning_rate": 0.00019999975689722398,
-      "loss": 46.0,
-      "step": 9190
-    },
-    {
-      "epoch": 0.7027161343349199,
-      "grad_norm": 0.0011110417544841766,
-      "learning_rate": 0.0001999997568442576,
-      "loss": 46.0,
-      "step": 9191
-    },
-    {
-      "epoch": 0.7027925913183095,
-      "grad_norm": 0.0006508385413326323,
-      "learning_rate": 0.00019999975679128537,
-      "loss": 46.0,
-      "step": 9192
-    },
-    {
-      "epoch": 0.7028690483016993,
-      "grad_norm": 0.0013018023455515504,
-      "learning_rate": 0.0001999997567383074,
-      "loss": 46.0,
-      "step": 9193
-    },
-    {
-      "epoch": 0.7029455052850889,
-      "grad_norm": 0.002450330648571253,
-      "learning_rate": 0.00019999975668532366,
-      "loss": 46.0,
-      "step": 9194
-    },
-    {
-      "epoch": 0.7030219622684787,
-      "grad_norm": 0.0014332069549709558,
-      "learning_rate": 0.00019999975663233415,
-      "loss": 46.0,
-      "step": 9195
-    },
-    {
-      "epoch": 0.7030984192518684,
-      "grad_norm": 0.0004257578984834254,
-      "learning_rate": 0.00019999975657933886,
-      "loss": 46.0,
-      "step": 9196
-    },
-    {
-      "epoch": 0.7031748762352581,
-      "grad_norm": 0.002065675798803568,
-      "learning_rate": 0.00019999975652633783,
-      "loss": 46.0,
-      "step": 9197
-    },
-    {
-      "epoch": 0.7032513332186479,
-      "grad_norm": 0.0013056508032605052,
-      "learning_rate": 0.000199999756473331,
-      "loss": 46.0,
-      "step": 9198
-    },
-    {
-      "epoch": 0.7033277902020376,
-      "grad_norm": 0.005665769334882498,
-      "learning_rate": 0.00019999975642031844,
-      "loss": 46.0,
-      "step": 9199
-    },
-    {
-      "epoch": 0.7034042471854273,
-      "grad_norm": 0.0012516052229329944,
-      "learning_rate": 0.00019999975636730006,
-      "loss": 46.0,
-      "step": 9200
-    },
-    {
-      "epoch": 0.703480704168817,
-      "grad_norm": 0.0014841699739918113,
-      "learning_rate": 0.00019999975631427593,
-      "loss": 46.0,
-      "step": 9201
-    },
-    {
-      "epoch": 0.7035571611522068,
-      "grad_norm": 0.005775256548076868,
-      "learning_rate": 0.00019999975626124603,
-      "loss": 46.0,
-      "step": 9202
-    },
-    {
-      "epoch": 0.7036336181355964,
-      "grad_norm": 0.002213146770372987,
-      "learning_rate": 0.00019999975620821036,
-      "loss": 46.0,
-      "step": 9203
-    },
-    {
-      "epoch": 0.7037100751189862,
-      "grad_norm": 0.0024645323865115643,
-      "learning_rate": 0.00019999975615516894,
-      "loss": 46.0,
-      "step": 9204
-    },
-    {
-      "epoch": 0.7037865321023759,
-      "grad_norm": 0.0006344622815959156,
-      "learning_rate": 0.00019999975610212174,
-      "loss": 46.0,
-      "step": 9205
-    },
-    {
-      "epoch": 0.7038629890857656,
-      "grad_norm": 0.0006042519817128778,
-      "learning_rate": 0.00019999975604906878,
-      "loss": 46.0,
-      "step": 9206
-    },
-    {
-      "epoch": 0.7039394460691554,
-      "grad_norm": 0.0026298025622963905,
-      "learning_rate": 0.00019999975599601,
-      "loss": 46.0,
-      "step": 9207
-    },
-    {
-      "epoch": 0.704015903052545,
-      "grad_norm": 0.0004056597244925797,
-      "learning_rate": 0.00019999975594294552,
-      "loss": 46.0,
-      "step": 9208
-    },
-    {
-      "epoch": 0.7040923600359348,
-      "grad_norm": 0.0016110679134726524,
-      "learning_rate": 0.00019999975588987524,
-      "loss": 46.0,
-      "step": 9209
-    },
-    {
-      "epoch": 0.7041688170193245,
-      "grad_norm": 0.0079397764056921,
-      "learning_rate": 0.0001999997558367992,
-      "loss": 46.0,
-      "step": 9210
-    },
-    {
-      "epoch": 0.7042452740027142,
-      "grad_norm": 0.0010373565601184964,
-      "learning_rate": 0.00019999975578371734,
-      "loss": 46.0,
-      "step": 9211
-    },
-    {
-      "epoch": 0.704321730986104,
-      "grad_norm": 0.0010001541813835502,
-      "learning_rate": 0.00019999975573062976,
-      "loss": 46.0,
-      "step": 9212
-    },
-    {
-      "epoch": 0.7043981879694937,
-      "grad_norm": 0.0020129969343543053,
-      "learning_rate": 0.0001999997556775364,
-      "loss": 46.0,
-      "step": 9213
-    },
-    {
-      "epoch": 0.7044746449528834,
-      "grad_norm": 0.001528335502371192,
-      "learning_rate": 0.00019999975562443728,
-      "loss": 46.0,
-      "step": 9214
-    },
-    {
-      "epoch": 0.7045511019362731,
-      "grad_norm": 0.0021050376817584038,
-      "learning_rate": 0.00019999975557133238,
-      "loss": 46.0,
-      "step": 9215
-    },
-    {
-      "epoch": 0.7046275589196628,
-      "grad_norm": 0.0008230862440541387,
-      "learning_rate": 0.0001999997555182217,
-      "loss": 46.0,
-      "step": 9216
-    },
-    {
-      "epoch": 0.7047040159030525,
-      "grad_norm": 0.0007504072273150086,
-      "learning_rate": 0.0001999997554651053,
-      "loss": 46.0,
-      "step": 9217
-    },
-    {
-      "epoch": 0.7047804728864423,
-      "grad_norm": 0.0009644035599194467,
-      "learning_rate": 0.00019999975541198307,
-      "loss": 46.0,
-      "step": 9218
-    },
-    {
-      "epoch": 0.704856929869832,
-      "grad_norm": 0.007080076262354851,
-      "learning_rate": 0.0001999997553588551,
-      "loss": 46.0,
-      "step": 9219
-    },
-    {
-      "epoch": 0.7049333868532217,
-      "grad_norm": 0.0008152625523507595,
-      "learning_rate": 0.00019999975530572134,
-      "loss": 46.0,
-      "step": 9220
-    },
-    {
-      "epoch": 0.7050098438366115,
-      "grad_norm": 0.0013208973687142134,
-      "learning_rate": 0.00019999975525258185,
-      "loss": 46.0,
-      "step": 9221
-    },
-    {
-      "epoch": 0.7050863008200011,
-      "grad_norm": 0.0008160319412127137,
-      "learning_rate": 0.00019999975519943654,
-      "loss": 46.0,
-      "step": 9222
-    },
-    {
-      "epoch": 0.7051627578033909,
-      "grad_norm": 0.030889302492141724,
-      "learning_rate": 0.0001999997551462855,
-      "loss": 46.0,
-      "step": 9223
-    },
-    {
-      "epoch": 0.7052392147867805,
-      "grad_norm": 0.001989106647670269,
-      "learning_rate": 0.0001999997550931287,
-      "loss": 46.0,
-      "step": 9224
-    },
-    {
-      "epoch": 0.7053156717701703,
-      "grad_norm": 0.0018548467196524143,
-      "learning_rate": 0.00019999975503996612,
-      "loss": 46.0,
-      "step": 9225
-    },
-    {
-      "epoch": 0.70539212875356,
-      "grad_norm": 0.0035306604113429785,
-      "learning_rate": 0.00019999975498679774,
-      "loss": 46.0,
-      "step": 9226
-    },
-    {
-      "epoch": 0.7054685857369497,
-      "grad_norm": 0.001718108425848186,
-      "learning_rate": 0.00019999975493362358,
-      "loss": 46.0,
-      "step": 9227
-    },
-    {
-      "epoch": 0.7055450427203395,
-      "grad_norm": 0.004146223422139883,
-      "learning_rate": 0.0001999997548804437,
-      "loss": 46.0,
-      "step": 9228
-    },
-    {
-      "epoch": 0.7056214997037292,
-      "grad_norm": 0.0005129209021106362,
-      "learning_rate": 0.00019999975482725804,
-      "loss": 46.0,
-      "step": 9229
-    },
-    {
-      "epoch": 0.7056979566871189,
-      "grad_norm": 0.0012858293484896421,
-      "learning_rate": 0.0001999997547740666,
-      "loss": 46.0,
-      "step": 9230
-    },
-    {
-      "epoch": 0.7057744136705086,
-      "grad_norm": 0.0014148908667266369,
-      "learning_rate": 0.00019999975472086937,
-      "loss": 46.0,
-      "step": 9231
-    },
-    {
-      "epoch": 0.7058508706538984,
-      "grad_norm": 0.0011945606674998999,
-      "learning_rate": 0.0001999997546676664,
-      "loss": 46.0,
-      "step": 9232
-    },
-    {
-      "epoch": 0.705927327637288,
-      "grad_norm": 0.003520740196108818,
-      "learning_rate": 0.00019999975461445767,
-      "loss": 46.0,
-      "step": 9233
-    },
-    {
-      "epoch": 0.7060037846206778,
-      "grad_norm": 0.0057270401157438755,
-      "learning_rate": 0.00019999975456124313,
-      "loss": 46.0,
-      "step": 9234
-    },
-    {
-      "epoch": 0.7060802416040675,
-      "grad_norm": 0.0006096501601859927,
-      "learning_rate": 0.00019999975450802287,
-      "loss": 46.0,
-      "step": 9235
-    },
-    {
-      "epoch": 0.7061566985874572,
-      "grad_norm": 0.0007158907828852534,
-      "learning_rate": 0.0001999997544547968,
-      "loss": 46.0,
-      "step": 9236
-    },
-    {
-      "epoch": 0.706233155570847,
-      "grad_norm": 0.006918021943420172,
-      "learning_rate": 0.00019999975440156497,
-      "loss": 46.0,
-      "step": 9237
-    },
-    {
-      "epoch": 0.7063096125542366,
-      "grad_norm": 0.0007008587708696723,
-      "learning_rate": 0.00019999975434832737,
-      "loss": 46.0,
-      "step": 9238
-    },
-    {
-      "epoch": 0.7063860695376264,
-      "grad_norm": 0.0029186594765633345,
-      "learning_rate": 0.000199999754295084,
-      "loss": 46.0,
-      "step": 9239
-    },
-    {
-      "epoch": 0.7064625265210162,
-      "grad_norm": 0.0011055928189307451,
-      "learning_rate": 0.0001999997542418349,
-      "loss": 46.0,
-      "step": 9240
-    },
-    {
-      "epoch": 0.7065389835044058,
-      "grad_norm": 0.0013629210880026221,
-      "learning_rate": 0.00019999975418858,
-      "loss": 46.0,
-      "step": 9241
-    },
-    {
-      "epoch": 0.7066154404877956,
-      "grad_norm": 0.008726016618311405,
-      "learning_rate": 0.00019999975413531932,
-      "loss": 46.0,
-      "step": 9242
-    },
-    {
-      "epoch": 0.7066918974711853,
-      "grad_norm": 0.012779823504388332,
-      "learning_rate": 0.00019999975408205284,
-      "loss": 46.0,
-      "step": 9243
-    },
-    {
-      "epoch": 0.706768354454575,
-      "grad_norm": 0.003101082518696785,
-      "learning_rate": 0.00019999975402878065,
-      "loss": 46.0,
-      "step": 9244
-    },
-    {
-      "epoch": 0.7068448114379647,
-      "grad_norm": 0.0008287486271001399,
-      "learning_rate": 0.00019999975397550266,
-      "loss": 46.0,
-      "step": 9245
-    },
-    {
-      "epoch": 0.7069212684213544,
-      "grad_norm": 0.0012902788585051894,
-      "learning_rate": 0.00019999975392221892,
-      "loss": 46.0,
-      "step": 9246
-    },
-    {
-      "epoch": 0.7069977254047441,
-      "grad_norm": 0.0008837492787279189,
-      "learning_rate": 0.0001999997538689294,
-      "loss": 46.0,
-      "step": 9247
-    },
-    {
-      "epoch": 0.7070741823881339,
-      "grad_norm": 0.0036821383982896805,
-      "learning_rate": 0.0001999997538156341,
-      "loss": 46.0,
-      "step": 9248
-    },
-    {
-      "epoch": 0.7071506393715236,
-      "grad_norm": 0.0007667118334211409,
-      "learning_rate": 0.00019999975376233303,
-      "loss": 46.0,
-      "step": 9249
-    },
-    {
-      "epoch": 0.7072270963549133,
-      "grad_norm": 0.0017325804801657796,
-      "learning_rate": 0.00019999975370902623,
-      "loss": 46.0,
-      "step": 9250
-    },
-    {
-      "epoch": 0.7073035533383031,
-      "grad_norm": 0.00407186197116971,
-      "learning_rate": 0.00019999975365571362,
-      "loss": 46.0,
-      "step": 9251
-    },
-    {
-      "epoch": 0.7073800103216927,
-      "grad_norm": 0.0011923268903046846,
-      "learning_rate": 0.00019999975360239524,
-      "loss": 46.0,
-      "step": 9252
-    },
-    {
-      "epoch": 0.7074564673050825,
-      "grad_norm": 0.002721782075241208,
-      "learning_rate": 0.00019999975354907112,
-      "loss": 46.0,
-      "step": 9253
-    },
-    {
-      "epoch": 0.7075329242884723,
-      "grad_norm": 0.0014553384389728308,
-      "learning_rate": 0.00019999975349574122,
-      "loss": 46.0,
-      "step": 9254
-    },
-    {
-      "epoch": 0.7076093812718619,
-      "grad_norm": 0.0015644950326532125,
-      "learning_rate": 0.00019999975344240555,
-      "loss": 46.0,
-      "step": 9255
-    },
-    {
-      "epoch": 0.7076858382552517,
-      "grad_norm": 0.0011328462278470397,
-      "learning_rate": 0.0001999997533890641,
-      "loss": 46.0,
-      "step": 9256
-    },
-    {
-      "epoch": 0.7077622952386413,
-      "grad_norm": 0.0014063374837860465,
-      "learning_rate": 0.00019999975333571688,
-      "loss": 46.0,
-      "step": 9257
-    },
-    {
-      "epoch": 0.7078387522220311,
-      "grad_norm": 0.001097632572054863,
-      "learning_rate": 0.00019999975328236392,
-      "loss": 46.0,
-      "step": 9258
-    },
-    {
-      "epoch": 0.7079152092054208,
-      "grad_norm": 0.0024974090047180653,
-      "learning_rate": 0.00019999975322900515,
-      "loss": 46.0,
-      "step": 9259
-    },
-    {
-      "epoch": 0.7079916661888105,
-      "grad_norm": 0.0012440483551472425,
-      "learning_rate": 0.0001999997531756406,
-      "loss": 46.0,
-      "step": 9260
-    },
-    {
-      "epoch": 0.7080681231722002,
-      "grad_norm": 0.0024394432548433542,
-      "learning_rate": 0.00019999975312227032,
-      "loss": 46.0,
-      "step": 9261
-    },
-    {
-      "epoch": 0.70814458015559,
-      "grad_norm": 0.008031205274164677,
-      "learning_rate": 0.00019999975306889427,
-      "loss": 46.0,
-      "step": 9262
-    },
-    {
-      "epoch": 0.7082210371389797,
-      "grad_norm": 0.004843798466026783,
-      "learning_rate": 0.00019999975301551243,
-      "loss": 46.0,
-      "step": 9263
-    },
-    {
-      "epoch": 0.7082974941223694,
-      "grad_norm": 0.0010757792042568326,
-      "learning_rate": 0.00019999975296212483,
-      "loss": 46.0,
-      "step": 9264
-    },
-    {
-      "epoch": 0.7083739511057591,
-      "grad_norm": 0.0013805648777633905,
-      "learning_rate": 0.00019999975290873148,
-      "loss": 46.0,
-      "step": 9265
-    },
-    {
-      "epoch": 0.7084504080891488,
-      "grad_norm": 0.0012855901150032878,
-      "learning_rate": 0.00019999975285533235,
-      "loss": 46.0,
-      "step": 9266
-    },
-    {
-      "epoch": 0.7085268650725386,
-      "grad_norm": 0.002793646417558193,
-      "learning_rate": 0.00019999975280192742,
-      "loss": 46.0,
-      "step": 9267
-    },
-    {
-      "epoch": 0.7086033220559282,
-      "grad_norm": 0.0009424712625332177,
-      "learning_rate": 0.00019999975274851675,
-      "loss": 46.0,
-      "step": 9268
-    },
-    {
-      "epoch": 0.708679779039318,
-      "grad_norm": 0.00178941257763654,
-      "learning_rate": 0.0001999997526951003,
-      "loss": 46.0,
-      "step": 9269
-    },
-    {
-      "epoch": 0.7087562360227078,
-      "grad_norm": 0.0018387021264061332,
-      "learning_rate": 0.0001999997526416781,
-      "loss": 46.0,
-      "step": 9270
-    },
-    {
-      "epoch": 0.7088326930060974,
-      "grad_norm": 0.0013985419645905495,
-      "learning_rate": 0.0001999997525882501,
-      "loss": 46.0,
-      "step": 9271
-    },
-    {
-      "epoch": 0.7089091499894872,
-      "grad_norm": 0.0015780134126543999,
-      "learning_rate": 0.00019999975253481633,
-      "loss": 46.0,
-      "step": 9272
-    },
-    {
-      "epoch": 0.7089856069728769,
-      "grad_norm": 0.0007537794299423695,
-      "learning_rate": 0.00019999975248137682,
-      "loss": 46.0,
-      "step": 9273
-    },
-    {
-      "epoch": 0.7090620639562666,
-      "grad_norm": 0.0018102899193763733,
-      "learning_rate": 0.00019999975242793153,
-      "loss": 46.0,
-      "step": 9274
-    },
-    {
-      "epoch": 0.7091385209396563,
-      "grad_norm": 0.0011644712649285793,
-      "learning_rate": 0.00019999975237448045,
-      "loss": 46.0,
-      "step": 9275
-    },
-    {
-      "epoch": 0.709214977923046,
-      "grad_norm": 0.0007414617575705051,
-      "learning_rate": 0.00019999975232102361,
-      "loss": 46.0,
-      "step": 9276
-    },
-    {
-      "epoch": 0.7092914349064358,
-      "grad_norm": 0.0005451389588415623,
-      "learning_rate": 0.00019999975226756104,
-      "loss": 46.0,
-      "step": 9277
-    },
-    {
-      "epoch": 0.7093678918898255,
-      "grad_norm": 0.0007670748163945973,
-      "learning_rate": 0.00019999975221409266,
-      "loss": 46.0,
-      "step": 9278
-    },
-    {
-      "epoch": 0.7094443488732152,
-      "grad_norm": 0.002322677057236433,
-      "learning_rate": 0.0001999997521606185,
-      "loss": 46.0,
-      "step": 9279
-    },
-    {
-      "epoch": 0.7095208058566049,
-      "grad_norm": 0.0007842218619771302,
-      "learning_rate": 0.0001999997521071386,
-      "loss": 46.0,
-      "step": 9280
-    },
-    {
-      "epoch": 0.7095972628399947,
-      "grad_norm": 0.0007776504498906434,
-      "learning_rate": 0.0001999997520536529,
-      "loss": 46.0,
-      "step": 9281
-    },
-    {
-      "epoch": 0.7096737198233843,
-      "grad_norm": 0.0037316500674933195,
-      "learning_rate": 0.0001999997520001615,
-      "loss": 46.0,
-      "step": 9282
-    },
-    {
-      "epoch": 0.7097501768067741,
-      "grad_norm": 0.0015603293431922793,
-      "learning_rate": 0.00019999975194666424,
-      "loss": 46.0,
-      "step": 9283
-    },
-    {
-      "epoch": 0.7098266337901639,
-      "grad_norm": 0.003073350293561816,
-      "learning_rate": 0.00019999975189316125,
-      "loss": 46.0,
-      "step": 9284
-    },
-    {
-      "epoch": 0.7099030907735535,
-      "grad_norm": 0.000701656099408865,
-      "learning_rate": 0.0001999997518396525,
-      "loss": 46.0,
-      "step": 9285
-    },
-    {
-      "epoch": 0.7099795477569433,
-      "grad_norm": 0.0029347811359912157,
-      "learning_rate": 0.00019999975178613797,
-      "loss": 46.0,
-      "step": 9286
-    },
-    {
-      "epoch": 0.7100560047403329,
-      "grad_norm": 0.004827440250664949,
-      "learning_rate": 0.00019999975173261766,
-      "loss": 46.0,
-      "step": 9287
-    },
-    {
-      "epoch": 0.7101324617237227,
-      "grad_norm": 0.003475607605651021,
-      "learning_rate": 0.0001999997516790916,
-      "loss": 46.0,
-      "step": 9288
-    },
-    {
-      "epoch": 0.7102089187071124,
-      "grad_norm": 0.0017068135784938931,
-      "learning_rate": 0.00019999975162555977,
-      "loss": 46.0,
-      "step": 9289
-    },
-    {
-      "epoch": 0.7102853756905021,
-      "grad_norm": 0.001960651483386755,
-      "learning_rate": 0.00019999975157202217,
-      "loss": 46.0,
-      "step": 9290
-    },
-    {
-      "epoch": 0.7103618326738919,
-      "grad_norm": 0.0010313292732462287,
-      "learning_rate": 0.0001999997515184788,
-      "loss": 46.0,
-      "step": 9291
-    },
-    {
-      "epoch": 0.7104382896572816,
-      "grad_norm": 0.0017544184811413288,
-      "learning_rate": 0.00019999975146492964,
-      "loss": 46.0,
-      "step": 9292
-    },
-    {
-      "epoch": 0.7105147466406713,
-      "grad_norm": 0.0026029690634459257,
-      "learning_rate": 0.0001999997514113747,
-      "loss": 46.0,
-      "step": 9293
-    },
-    {
-      "epoch": 0.710591203624061,
-      "grad_norm": 0.0009716750937514007,
-      "learning_rate": 0.00019999975135781407,
-      "loss": 46.0,
-      "step": 9294
-    },
-    {
-      "epoch": 0.7106676606074507,
-      "grad_norm": 0.0003008923085872084,
-      "learning_rate": 0.0001999997513042476,
-      "loss": 46.0,
-      "step": 9295
-    },
-    {
-      "epoch": 0.7107441175908404,
-      "grad_norm": 0.0020217609126120806,
-      "learning_rate": 0.00019999975125067535,
-      "loss": 46.0,
-      "step": 9296
-    },
-    {
-      "epoch": 0.7108205745742302,
-      "grad_norm": 0.0019697891548275948,
-      "learning_rate": 0.00019999975119709739,
-      "loss": 46.0,
-      "step": 9297
-    },
-    {
-      "epoch": 0.7108970315576199,
-      "grad_norm": 0.0009867234621196985,
-      "learning_rate": 0.00019999975114351362,
-      "loss": 46.0,
-      "step": 9298
-    },
-    {
-      "epoch": 0.7109734885410096,
-      "grad_norm": 0.001355454558506608,
-      "learning_rate": 0.00019999975108992408,
-      "loss": 46.0,
-      "step": 9299
-    },
-    {
-      "epoch": 0.7110499455243994,
-      "grad_norm": 0.00173101422842592,
-      "learning_rate": 0.00019999975103632877,
-      "loss": 46.0,
-      "step": 9300
-    },
-    {
-      "epoch": 0.711126402507789,
-      "grad_norm": 0.0006403400329872966,
-      "learning_rate": 0.0001999997509827277,
-      "loss": 46.0,
-      "step": 9301
-    },
-    {
-      "epoch": 0.7112028594911788,
-      "grad_norm": 0.0008197331335395575,
-      "learning_rate": 0.00019999975092912088,
-      "loss": 46.0,
-      "step": 9302
-    },
-    {
-      "epoch": 0.7112793164745685,
-      "grad_norm": 0.0034717696253210306,
-      "learning_rate": 0.00019999975087550828,
-      "loss": 46.0,
-      "step": 9303
-    },
-    {
-      "epoch": 0.7113557734579582,
-      "grad_norm": 0.0008460204699076712,
-      "learning_rate": 0.0001999997508218899,
-      "loss": 46.0,
-      "step": 9304
-    },
-    {
-      "epoch": 0.711432230441348,
-      "grad_norm": 0.003026822814717889,
-      "learning_rate": 0.00019999975076826575,
-      "loss": 46.0,
-      "step": 9305
-    },
-    {
-      "epoch": 0.7115086874247376,
-      "grad_norm": 0.003876257222145796,
-      "learning_rate": 0.00019999975071463585,
-      "loss": 46.0,
-      "step": 9306
-    },
-    {
-      "epoch": 0.7115851444081274,
-      "grad_norm": 0.0005720927729271352,
-      "learning_rate": 0.00019999975066100015,
-      "loss": 46.0,
-      "step": 9307
-    },
-    {
-      "epoch": 0.7116616013915171,
-      "grad_norm": 0.0005936491652391851,
-      "learning_rate": 0.0001999997506073587,
-      "loss": 46.0,
-      "step": 9308
-    },
-    {
-      "epoch": 0.7117380583749068,
-      "grad_norm": 0.0007157169166021049,
-      "learning_rate": 0.00019999975055371146,
-      "loss": 46.0,
-      "step": 9309
-    },
-    {
-      "epoch": 0.7118145153582965,
-      "grad_norm": 0.0006979728350415826,
-      "learning_rate": 0.00019999975050005847,
-      "loss": 46.0,
-      "step": 9310
-    },
-    {
-      "epoch": 0.7118909723416863,
-      "grad_norm": 0.005947049241513014,
-      "learning_rate": 0.0001999997504463997,
-      "loss": 46.0,
-      "step": 9311
-    },
-    {
-      "epoch": 0.711967429325076,
-      "grad_norm": 0.0008080280967988074,
-      "learning_rate": 0.00019999975039273517,
-      "loss": 46.0,
-      "step": 9312
-    },
-    {
-      "epoch": 0.7120438863084657,
-      "grad_norm": 0.01812349632382393,
-      "learning_rate": 0.00019999975033906488,
-      "loss": 46.0,
-      "step": 9313
-    },
-    {
-      "epoch": 0.7121203432918555,
-      "grad_norm": 0.0006871453369967639,
-      "learning_rate": 0.0001999997502853888,
-      "loss": 46.0,
-      "step": 9314
-    },
-    {
-      "epoch": 0.7121968002752451,
-      "grad_norm": 0.007056377362459898,
-      "learning_rate": 0.00019999975023170697,
-      "loss": 46.0,
-      "step": 9315
-    },
-    {
-      "epoch": 0.7122732572586349,
-      "grad_norm": 0.0011114017106592655,
-      "learning_rate": 0.00019999975017801936,
-      "loss": 46.0,
-      "step": 9316
-    },
-    {
-      "epoch": 0.7123497142420245,
-      "grad_norm": 0.0007559072109870613,
-      "learning_rate": 0.00019999975012432596,
-      "loss": 46.0,
-      "step": 9317
-    },
-    {
-      "epoch": 0.7124261712254143,
-      "grad_norm": 0.0008278271416202188,
-      "learning_rate": 0.0001999997500706268,
-      "loss": 46.0,
-      "step": 9318
-    },
-    {
-      "epoch": 0.7125026282088041,
-      "grad_norm": 0.001481667859479785,
-      "learning_rate": 0.0001999997500169219,
-      "loss": 46.0,
-      "step": 9319
-    },
-    {
-      "epoch": 0.7125790851921937,
-      "grad_norm": 0.0004982512909919024,
-      "learning_rate": 0.0001999997499632112,
-      "loss": 46.0,
-      "step": 9320
-    },
-    {
-      "epoch": 0.7126555421755835,
-      "grad_norm": 0.0022554430179297924,
-      "learning_rate": 0.00019999974990949474,
-      "loss": 46.0,
-      "step": 9321
-    },
-    {
-      "epoch": 0.7127319991589732,
-      "grad_norm": 0.007220863830298185,
-      "learning_rate": 0.00019999974985577252,
-      "loss": 46.0,
-      "step": 9322
-    },
-    {
-      "epoch": 0.7128084561423629,
-      "grad_norm": 0.0027372404001653194,
-      "learning_rate": 0.00019999974980204453,
-      "loss": 46.0,
-      "step": 9323
-    },
-    {
-      "epoch": 0.7128849131257526,
-      "grad_norm": 0.0011329575208947062,
-      "learning_rate": 0.00019999974974831077,
-      "loss": 46.0,
-      "step": 9324
-    },
-    {
-      "epoch": 0.7129613701091423,
-      "grad_norm": 0.002743436023592949,
-      "learning_rate": 0.00019999974969457123,
-      "loss": 46.0,
-      "step": 9325
-    },
-    {
-      "epoch": 0.713037827092532,
-      "grad_norm": 0.0008240920142270625,
-      "learning_rate": 0.00019999974964082592,
-      "loss": 46.0,
-      "step": 9326
-    },
-    {
-      "epoch": 0.7131142840759218,
-      "grad_norm": 0.0025407446082681417,
-      "learning_rate": 0.00019999974958707483,
-      "loss": 46.0,
-      "step": 9327
-    },
-    {
-      "epoch": 0.7131907410593115,
-      "grad_norm": 0.0006714505143463612,
-      "learning_rate": 0.000199999749533318,
-      "loss": 46.0,
-      "step": 9328
-    },
-    {
-      "epoch": 0.7132671980427012,
-      "grad_norm": 0.0024302825331687927,
-      "learning_rate": 0.0001999997494795554,
-      "loss": 46.0,
-      "step": 9329
-    },
-    {
-      "epoch": 0.713343655026091,
-      "grad_norm": 0.0012964991619810462,
-      "learning_rate": 0.000199999749425787,
-      "loss": 46.0,
-      "step": 9330
-    },
-    {
-      "epoch": 0.7134201120094806,
-      "grad_norm": 0.0013611806789413095,
-      "learning_rate": 0.00019999974937201284,
-      "loss": 46.0,
-      "step": 9331
-    },
-    {
-      "epoch": 0.7134965689928704,
-      "grad_norm": 0.002300181658938527,
-      "learning_rate": 0.00019999974931823294,
-      "loss": 46.0,
-      "step": 9332
-    },
-    {
-      "epoch": 0.7135730259762602,
-      "grad_norm": 0.0010458847973495722,
-      "learning_rate": 0.00019999974926444725,
-      "loss": 46.0,
-      "step": 9333
-    },
-    {
-      "epoch": 0.7136494829596498,
-      "grad_norm": 0.0011069714091718197,
-      "learning_rate": 0.00019999974921065578,
-      "loss": 46.0,
-      "step": 9334
-    },
-    {
-      "epoch": 0.7137259399430396,
-      "grad_norm": 0.0025952591095119715,
-      "learning_rate": 0.00019999974915685856,
-      "loss": 46.0,
-      "step": 9335
-    },
-    {
-      "epoch": 0.7138023969264292,
-      "grad_norm": 0.007478729356080294,
-      "learning_rate": 0.00019999974910305554,
-      "loss": 46.0,
-      "step": 9336
-    },
-    {
-      "epoch": 0.713878853909819,
-      "grad_norm": 0.0013561181258410215,
-      "learning_rate": 0.00019999974904924678,
-      "loss": 46.0,
-      "step": 9337
-    },
-    {
-      "epoch": 0.7139553108932087,
-      "grad_norm": 0.0025987953413277864,
-      "learning_rate": 0.00019999974899543224,
-      "loss": 46.0,
-      "step": 9338
-    },
-    {
-      "epoch": 0.7140317678765984,
-      "grad_norm": 0.001967592863366008,
-      "learning_rate": 0.00019999974894161193,
-      "loss": 46.0,
-      "step": 9339
-    },
-    {
-      "epoch": 0.7141082248599881,
-      "grad_norm": 0.0010752426460385323,
-      "learning_rate": 0.00019999974888778584,
-      "loss": 46.0,
-      "step": 9340
-    },
-    {
-      "epoch": 0.7141846818433779,
-      "grad_norm": 0.0007942437077872455,
-      "learning_rate": 0.000199999748833954,
-      "loss": 46.0,
-      "step": 9341
-    },
-    {
-      "epoch": 0.7142611388267676,
-      "grad_norm": 0.001369924983009696,
-      "learning_rate": 0.00019999974878011638,
-      "loss": 46.0,
-      "step": 9342
-    },
-    {
-      "epoch": 0.7143375958101573,
-      "grad_norm": 0.0008023129776120186,
-      "learning_rate": 0.00019999974872627298,
-      "loss": 46.0,
-      "step": 9343
-    },
-    {
-      "epoch": 0.7144140527935471,
-      "grad_norm": 0.002127670682966709,
-      "learning_rate": 0.00019999974867242383,
-      "loss": 46.0,
-      "step": 9344
-    },
-    {
-      "epoch": 0.7144905097769367,
-      "grad_norm": 0.001691145938821137,
-      "learning_rate": 0.0001999997486185689,
-      "loss": 46.0,
-      "step": 9345
-    },
-    {
-      "epoch": 0.7145669667603265,
-      "grad_norm": 0.0008705343352630734,
-      "learning_rate": 0.00019999974856470823,
-      "loss": 46.0,
-      "step": 9346
-    },
-    {
-      "epoch": 0.7146434237437161,
-      "grad_norm": 0.0011588613269850612,
-      "learning_rate": 0.00019999974851084176,
-      "loss": 46.0,
-      "step": 9347
-    },
-    {
-      "epoch": 0.7147198807271059,
-      "grad_norm": 0.0005489863106049597,
-      "learning_rate": 0.00019999974845696952,
-      "loss": 46.0,
-      "step": 9348
-    },
-    {
-      "epoch": 0.7147963377104957,
-      "grad_norm": 0.0039374688640236855,
-      "learning_rate": 0.00019999974840309153,
-      "loss": 46.0,
-      "step": 9349
-    },
-    {
-      "epoch": 0.7148727946938853,
-      "grad_norm": 0.0008682627812959254,
-      "learning_rate": 0.00019999974834920776,
-      "loss": 46.0,
-      "step": 9350
-    },
-    {
-      "epoch": 0.7149492516772751,
-      "grad_norm": 0.0016881709452718496,
-      "learning_rate": 0.0001999997482953182,
-      "loss": 46.0,
-      "step": 9351
-    },
-    {
-      "epoch": 0.7150257086606648,
-      "grad_norm": 0.001021580072119832,
-      "learning_rate": 0.00019999974824142292,
-      "loss": 46.0,
-      "step": 9352
-    },
-    {
-      "epoch": 0.7151021656440545,
-      "grad_norm": 0.002316372701898217,
-      "learning_rate": 0.0001999997481875218,
-      "loss": 46.0,
-      "step": 9353
-    },
-    {
-      "epoch": 0.7151786226274442,
-      "grad_norm": 0.0009586425148881972,
-      "learning_rate": 0.00019999974813361498,
-      "loss": 46.0,
-      "step": 9354
-    },
-    {
-      "epoch": 0.7152550796108339,
-      "grad_norm": 0.0009738823282532394,
-      "learning_rate": 0.00019999974807970237,
-      "loss": 46.0,
-      "step": 9355
-    },
-    {
-      "epoch": 0.7153315365942237,
-      "grad_norm": 0.0012651185970753431,
-      "learning_rate": 0.00019999974802578397,
-      "loss": 46.0,
-      "step": 9356
-    },
-    {
-      "epoch": 0.7154079935776134,
-      "grad_norm": 0.005341680254787207,
-      "learning_rate": 0.00019999974797185982,
-      "loss": 46.0,
-      "step": 9357
-    },
-    {
-      "epoch": 0.7154844505610031,
-      "grad_norm": 0.004591882228851318,
-      "learning_rate": 0.0001999997479179299,
-      "loss": 46.0,
-      "step": 9358
-    },
-    {
-      "epoch": 0.7155609075443928,
-      "grad_norm": 0.0059111532755196095,
-      "learning_rate": 0.00019999974786399417,
-      "loss": 46.0,
-      "step": 9359
-    },
-    {
-      "epoch": 0.7156373645277826,
-      "grad_norm": 0.0009546828223392367,
-      "learning_rate": 0.00019999974781005273,
-      "loss": 46.0,
-      "step": 9360
-    },
-    {
-      "epoch": 0.7157138215111722,
-      "grad_norm": 0.0016816677525639534,
-      "learning_rate": 0.00019999974775610548,
-      "loss": 46.0,
-      "step": 9361
-    },
-    {
-      "epoch": 0.715790278494562,
-      "grad_norm": 0.0035789271350950003,
-      "learning_rate": 0.00019999974770215247,
-      "loss": 46.0,
-      "step": 9362
-    },
-    {
-      "epoch": 0.7158667354779518,
-      "grad_norm": 0.0019321265863254666,
-      "learning_rate": 0.0001999997476481937,
-      "loss": 46.0,
-      "step": 9363
-    },
-    {
-      "epoch": 0.7159431924613414,
-      "grad_norm": 0.003660425543785095,
-      "learning_rate": 0.00019999974759422914,
-      "loss": 46.0,
-      "step": 9364
-    },
-    {
-      "epoch": 0.7160196494447312,
-      "grad_norm": 0.001513624913059175,
-      "learning_rate": 0.00019999974754025885,
-      "loss": 46.0,
-      "step": 9365
-    },
-    {
-      "epoch": 0.7160961064281208,
-      "grad_norm": 0.002380894962698221,
-      "learning_rate": 0.00019999974748628277,
-      "loss": 46.0,
-      "step": 9366
-    },
-    {
-      "epoch": 0.7161725634115106,
-      "grad_norm": 0.0006959422025829554,
-      "learning_rate": 0.0001999997474323009,
-      "loss": 46.0,
-      "step": 9367
-    },
-    {
-      "epoch": 0.7162490203949003,
-      "grad_norm": 0.000567566545214504,
-      "learning_rate": 0.00019999974737831328,
-      "loss": 46.0,
-      "step": 9368
-    },
-    {
-      "epoch": 0.71632547737829,
-      "grad_norm": 0.0010751682566478848,
-      "learning_rate": 0.0001999997473243199,
-      "loss": 46.0,
-      "step": 9369
-    },
-    {
-      "epoch": 0.7164019343616798,
-      "grad_norm": 0.00132025929633528,
-      "learning_rate": 0.00019999974727032073,
-      "loss": 46.0,
-      "step": 9370
-    },
-    {
-      "epoch": 0.7164783913450695,
-      "grad_norm": 0.0022320637945085764,
-      "learning_rate": 0.0001999997472163158,
-      "loss": 46.0,
-      "step": 9371
-    },
-    {
-      "epoch": 0.7165548483284592,
-      "grad_norm": 0.0006569395773112774,
-      "learning_rate": 0.00019999974716230508,
-      "loss": 46.0,
-      "step": 9372
-    },
-    {
-      "epoch": 0.7166313053118489,
-      "grad_norm": 0.0011366188991814852,
-      "learning_rate": 0.00019999974710828864,
-      "loss": 46.0,
-      "step": 9373
-    },
-    {
-      "epoch": 0.7167077622952387,
-      "grad_norm": 0.0009507978102192283,
-      "learning_rate": 0.0001999997470542664,
-      "loss": 46.0,
-      "step": 9374
-    },
-    {
-      "epoch": 0.7167842192786283,
-      "grad_norm": 0.0014681183965876698,
-      "learning_rate": 0.00019999974700023838,
-      "loss": 46.0,
-      "step": 9375
-    },
-    {
-      "epoch": 0.7168606762620181,
-      "grad_norm": 0.0011173589155077934,
-      "learning_rate": 0.00019999974694620462,
-      "loss": 46.0,
-      "step": 9376
-    },
-    {
-      "epoch": 0.7169371332454078,
-      "grad_norm": 0.0025795476976782084,
-      "learning_rate": 0.00019999974689216505,
-      "loss": 46.0,
-      "step": 9377
-    },
-    {
-      "epoch": 0.7170135902287975,
-      "grad_norm": 0.007265380583703518,
-      "learning_rate": 0.00019999974683811974,
-      "loss": 46.0,
-      "step": 9378
-    },
-    {
-      "epoch": 0.7170900472121873,
-      "grad_norm": 0.0008986399625428021,
-      "learning_rate": 0.0001999997467840687,
-      "loss": 46.0,
-      "step": 9379
-    },
-    {
-      "epoch": 0.7171665041955769,
-      "grad_norm": 0.005378580652177334,
-      "learning_rate": 0.0001999997467300118,
-      "loss": 46.0,
-      "step": 9380
-    },
-    {
-      "epoch": 0.7172429611789667,
-      "grad_norm": 0.000935537158511579,
-      "learning_rate": 0.00019999974667594917,
-      "loss": 46.0,
-      "step": 9381
-    },
-    {
-      "epoch": 0.7173194181623564,
-      "grad_norm": 0.0021799695678055286,
-      "learning_rate": 0.0001999997466218808,
-      "loss": 46.0,
-      "step": 9382
-    },
-    {
-      "epoch": 0.7173958751457461,
-      "grad_norm": 0.0007234615040943027,
-      "learning_rate": 0.00019999974656780662,
-      "loss": 46.0,
-      "step": 9383
-    },
-    {
-      "epoch": 0.7174723321291359,
-      "grad_norm": 0.0031444921623915434,
-      "learning_rate": 0.0001999997465137267,
-      "loss": 46.0,
-      "step": 9384
-    },
-    {
-      "epoch": 0.7175487891125256,
-      "grad_norm": 0.0008040877874009311,
-      "learning_rate": 0.000199999746459641,
-      "loss": 46.0,
-      "step": 9385
-    },
-    {
-      "epoch": 0.7176252460959153,
-      "grad_norm": 0.0009112692205235362,
-      "learning_rate": 0.0001999997464055495,
-      "loss": 46.0,
-      "step": 9386
-    },
-    {
-      "epoch": 0.717701703079305,
-      "grad_norm": 0.0024245227687060833,
-      "learning_rate": 0.0001999997463514523,
-      "loss": 46.0,
-      "step": 9387
-    },
-    {
-      "epoch": 0.7177781600626947,
-      "grad_norm": 0.003240007907152176,
-      "learning_rate": 0.00019999974629734924,
-      "loss": 46.0,
-      "step": 9388
-    },
-    {
-      "epoch": 0.7178546170460844,
-      "grad_norm": 0.0009320462704636157,
-      "learning_rate": 0.00019999974624324048,
-      "loss": 46.0,
-      "step": 9389
-    },
-    {
-      "epoch": 0.7179310740294742,
-      "grad_norm": 0.0011084995931014419,
-      "learning_rate": 0.00019999974618912594,
-      "loss": 46.0,
-      "step": 9390
-    },
-    {
-      "epoch": 0.7180075310128639,
-      "grad_norm": 0.0008703387575224042,
-      "learning_rate": 0.0001999997461350056,
-      "loss": 46.0,
-      "step": 9391
-    },
-    {
-      "epoch": 0.7180839879962536,
-      "grad_norm": 0.003441119333729148,
-      "learning_rate": 0.00019999974608087953,
-      "loss": 46.0,
-      "step": 9392
-    },
-    {
-      "epoch": 0.7181604449796434,
-      "grad_norm": 0.004153742920607328,
-      "learning_rate": 0.00019999974602674767,
-      "loss": 46.0,
-      "step": 9393
-    },
-    {
-      "epoch": 0.718236901963033,
-      "grad_norm": 0.0016669102478772402,
-      "learning_rate": 0.00019999974597261004,
-      "loss": 46.0,
-      "step": 9394
-    },
-    {
-      "epoch": 0.7183133589464228,
-      "grad_norm": 0.001168415299616754,
-      "learning_rate": 0.00019999974591846664,
-      "loss": 46.0,
-      "step": 9395
-    },
-    {
-      "epoch": 0.7183898159298124,
-      "grad_norm": 0.0010830056853592396,
-      "learning_rate": 0.00019999974586431746,
-      "loss": 46.0,
-      "step": 9396
-    },
-    {
-      "epoch": 0.7184662729132022,
-      "grad_norm": 0.0005991084035485983,
-      "learning_rate": 0.00019999974581016254,
-      "loss": 46.0,
-      "step": 9397
-    },
-    {
-      "epoch": 0.718542729896592,
-      "grad_norm": 0.0016091811703518033,
-      "learning_rate": 0.00019999974575600184,
-      "loss": 46.0,
-      "step": 9398
-    },
-    {
-      "epoch": 0.7186191868799816,
-      "grad_norm": 0.0005256792646832764,
-      "learning_rate": 0.00019999974570183534,
-      "loss": 46.0,
-      "step": 9399
-    },
-    {
-      "epoch": 0.7186956438633714,
-      "grad_norm": 0.0030196753796190023,
-      "learning_rate": 0.00019999974564766313,
-      "loss": 46.0,
-      "step": 9400
-    },
-    {
-      "epoch": 0.7187721008467611,
-      "grad_norm": 0.0011832899181172252,
-      "learning_rate": 0.0001999997455934851,
-      "loss": 46.0,
-      "step": 9401
-    },
-    {
-      "epoch": 0.7188485578301508,
-      "grad_norm": 0.0023695374839007854,
-      "learning_rate": 0.00019999974553930132,
-      "loss": 46.0,
-      "step": 9402
-    },
-    {
-      "epoch": 0.7189250148135405,
-      "grad_norm": 0.0012815966038033366,
-      "learning_rate": 0.0001999997454851118,
-      "loss": 46.0,
-      "step": 9403
-    },
-    {
-      "epoch": 0.7190014717969303,
-      "grad_norm": 0.0008994894451461732,
-      "learning_rate": 0.00019999974543091645,
-      "loss": 46.0,
-      "step": 9404
-    },
-    {
-      "epoch": 0.71907792878032,
-      "grad_norm": 0.0035863597877323627,
-      "learning_rate": 0.00019999974537671537,
-      "loss": 46.0,
-      "step": 9405
-    },
-    {
-      "epoch": 0.7191543857637097,
-      "grad_norm": 0.0020268044900149107,
-      "learning_rate": 0.0001999997453225085,
-      "loss": 46.0,
-      "step": 9406
-    },
-    {
-      "epoch": 0.7192308427470994,
-      "grad_norm": 0.0016784875188022852,
-      "learning_rate": 0.00019999974526829588,
-      "loss": 46.0,
-      "step": 9407
-    },
-    {
-      "epoch": 0.7193072997304891,
-      "grad_norm": 0.005460328422486782,
-      "learning_rate": 0.00019999974521407748,
-      "loss": 46.0,
-      "step": 9408
-    },
-    {
-      "epoch": 0.7193837567138789,
-      "grad_norm": 0.0005024038255214691,
-      "learning_rate": 0.0001999997451598533,
-      "loss": 46.0,
-      "step": 9409
-    },
-    {
-      "epoch": 0.7194602136972685,
-      "grad_norm": 0.000776283850427717,
-      "learning_rate": 0.00019999974510562335,
-      "loss": 46.0,
-      "step": 9410
-    },
-    {
-      "epoch": 0.7195366706806583,
-      "grad_norm": 0.0005845642299391329,
-      "learning_rate": 0.00019999974505138766,
-      "loss": 46.0,
-      "step": 9411
-    },
-    {
-      "epoch": 0.7196131276640481,
-      "grad_norm": 0.001419172971509397,
-      "learning_rate": 0.0001999997449971462,
-      "loss": 46.0,
-      "step": 9412
-    },
-    {
-      "epoch": 0.7196895846474377,
-      "grad_norm": 0.002445223508402705,
-      "learning_rate": 0.00019999974494289892,
-      "loss": 46.0,
-      "step": 9413
-    },
-    {
-      "epoch": 0.7197660416308275,
-      "grad_norm": 0.006486148573458195,
-      "learning_rate": 0.00019999974488864593,
-      "loss": 46.0,
-      "step": 9414
-    },
-    {
-      "epoch": 0.7198424986142172,
-      "grad_norm": 0.0013122617965564132,
-      "learning_rate": 0.00019999974483438711,
-      "loss": 46.0,
-      "step": 9415
-    },
-    {
-      "epoch": 0.7199189555976069,
-      "grad_norm": 0.0008856218191795051,
-      "learning_rate": 0.00019999974478012258,
-      "loss": 46.0,
-      "step": 9416
-    },
-    {
-      "epoch": 0.7199954125809966,
-      "grad_norm": 0.0012524225749075413,
-      "learning_rate": 0.00019999974472585224,
-      "loss": 46.0,
-      "step": 9417
-    },
-    {
-      "epoch": 0.7200718695643863,
-      "grad_norm": 0.0009493715479038656,
-      "learning_rate": 0.00019999974467157613,
-      "loss": 46.0,
-      "step": 9418
-    },
-    {
-      "epoch": 0.720148326547776,
-      "grad_norm": 0.0013101128861308098,
-      "learning_rate": 0.0001999997446172943,
-      "loss": 46.0,
-      "step": 9419
-    },
-    {
-      "epoch": 0.7202247835311658,
-      "grad_norm": 0.002712149638682604,
-      "learning_rate": 0.00019999974456300668,
-      "loss": 46.0,
-      "step": 9420
-    },
-    {
-      "epoch": 0.7203012405145555,
-      "grad_norm": 0.0015899811405688524,
-      "learning_rate": 0.00019999974450871327,
-      "loss": 46.0,
-      "step": 9421
-    },
-    {
-      "epoch": 0.7203776974979452,
-      "grad_norm": 0.002752808853983879,
-      "learning_rate": 0.00019999974445441407,
-      "loss": 46.0,
-      "step": 9422
-    },
-    {
-      "epoch": 0.720454154481335,
-      "grad_norm": 0.0015156165463849902,
-      "learning_rate": 0.00019999974440010915,
-      "loss": 46.0,
-      "step": 9423
-    },
-    {
-      "epoch": 0.7205306114647246,
-      "grad_norm": 0.00050859444309026,
-      "learning_rate": 0.00019999974434579842,
-      "loss": 46.0,
-      "step": 9424
-    },
-    {
-      "epoch": 0.7206070684481144,
-      "grad_norm": 0.0006458958378061652,
-      "learning_rate": 0.00019999974429148196,
-      "loss": 46.0,
-      "step": 9425
-    },
-    {
-      "epoch": 0.720683525431504,
-      "grad_norm": 0.0008962931460700929,
-      "learning_rate": 0.00019999974423715971,
-      "loss": 46.0,
-      "step": 9426
-    },
-    {
-      "epoch": 0.7207599824148938,
-      "grad_norm": 0.0012574514839798212,
-      "learning_rate": 0.00019999974418283167,
-      "loss": 46.0,
-      "step": 9427
-    },
-    {
-      "epoch": 0.7208364393982836,
-      "grad_norm": 0.0061298212967813015,
-      "learning_rate": 0.00019999974412849788,
-      "loss": 46.0,
-      "step": 9428
-    },
-    {
-      "epoch": 0.7209128963816732,
-      "grad_norm": 0.0022923534270375967,
-      "learning_rate": 0.00019999974407415835,
-      "loss": 46.0,
-      "step": 9429
-    },
-    {
-      "epoch": 0.720989353365063,
-      "grad_norm": 0.0011839504586532712,
-      "learning_rate": 0.000199999744019813,
-      "loss": 46.0,
-      "step": 9430
-    },
-    {
-      "epoch": 0.7210658103484527,
-      "grad_norm": 0.0017941249534487724,
-      "learning_rate": 0.0001999997439654619,
-      "loss": 46.0,
-      "step": 9431
-    },
-    {
-      "epoch": 0.7211422673318424,
-      "grad_norm": 0.012152974493801594,
-      "learning_rate": 0.00019999974391110505,
-      "loss": 46.0,
-      "step": 9432
-    },
-    {
-      "epoch": 0.7212187243152322,
-      "grad_norm": 0.0014832686865702271,
-      "learning_rate": 0.00019999974385674242,
-      "loss": 46.0,
-      "step": 9433
-    },
-    {
-      "epoch": 0.7212951812986219,
-      "grad_norm": 0.0010577032808214426,
-      "learning_rate": 0.00019999974380237402,
-      "loss": 46.0,
-      "step": 9434
-    },
-    {
-      "epoch": 0.7213716382820116,
-      "grad_norm": 0.006536453031003475,
-      "learning_rate": 0.00019999974374799984,
-      "loss": 46.0,
-      "step": 9435
-    },
-    {
-      "epoch": 0.7214480952654013,
-      "grad_norm": 0.0027409642934799194,
-      "learning_rate": 0.0001999997436936199,
-      "loss": 46.0,
-      "step": 9436
-    },
-    {
-      "epoch": 0.721524552248791,
-      "grad_norm": 0.0027364809066057205,
-      "learning_rate": 0.0001999997436392342,
-      "loss": 46.0,
-      "step": 9437
-    },
-    {
-      "epoch": 0.7216010092321807,
-      "grad_norm": 0.0019776425324380398,
-      "learning_rate": 0.00019999974358484267,
-      "loss": 46.0,
-      "step": 9438
-    },
-    {
-      "epoch": 0.7216774662155705,
-      "grad_norm": 0.0006617572507821023,
-      "learning_rate": 0.00019999974353044543,
-      "loss": 46.0,
-      "step": 9439
-    },
-    {
-      "epoch": 0.7217539231989601,
-      "grad_norm": 0.0010409303940832615,
-      "learning_rate": 0.00019999974347604242,
-      "loss": 46.0,
-      "step": 9440
-    },
-    {
-      "epoch": 0.7218303801823499,
-      "grad_norm": 0.0015258465427905321,
-      "learning_rate": 0.00019999974342163363,
-      "loss": 46.0,
-      "step": 9441
-    },
-    {
-      "epoch": 0.7219068371657397,
-      "grad_norm": 0.0007176099461503327,
-      "learning_rate": 0.00019999974336721907,
-      "loss": 46.0,
-      "step": 9442
-    },
-    {
-      "epoch": 0.7219832941491293,
-      "grad_norm": 0.003987742122262716,
-      "learning_rate": 0.00019999974331279873,
-      "loss": 46.0,
-      "step": 9443
-    },
-    {
-      "epoch": 0.7220597511325191,
-      "grad_norm": 0.001661333255469799,
-      "learning_rate": 0.00019999974325837262,
-      "loss": 46.0,
-      "step": 9444
-    },
-    {
-      "epoch": 0.7221362081159088,
-      "grad_norm": 0.0009441061993129551,
-      "learning_rate": 0.00019999974320394077,
-      "loss": 46.0,
-      "step": 9445
-    },
-    {
-      "epoch": 0.7222126650992985,
-      "grad_norm": 0.0009892223170027137,
-      "learning_rate": 0.00019999974314950314,
-      "loss": 46.0,
-      "step": 9446
-    },
-    {
-      "epoch": 0.7222891220826883,
-      "grad_norm": 0.0009868749184533954,
-      "learning_rate": 0.00019999974309505974,
-      "loss": 46.0,
-      "step": 9447
-    },
-    {
-      "epoch": 0.7223655790660779,
-      "grad_norm": 0.0004595013742800802,
-      "learning_rate": 0.00019999974304061056,
-      "loss": 46.0,
-      "step": 9448
-    },
-    {
-      "epoch": 0.7224420360494677,
-      "grad_norm": 0.0015255939215421677,
-      "learning_rate": 0.0001999997429861556,
-      "loss": 46.0,
-      "step": 9449
-    },
-    {
-      "epoch": 0.7225184930328574,
-      "grad_norm": 0.003566581755876541,
-      "learning_rate": 0.0001999997429316949,
-      "loss": 46.0,
-      "step": 9450
-    },
-    {
-      "epoch": 0.7225949500162471,
-      "grad_norm": 0.0010437725577503443,
-      "learning_rate": 0.0001999997428772284,
-      "loss": 46.0,
-      "step": 9451
-    },
-    {
-      "epoch": 0.7226714069996368,
-      "grad_norm": 0.0011425354750826955,
-      "learning_rate": 0.00019999974282275615,
-      "loss": 46.0,
-      "step": 9452
-    },
-    {
-      "epoch": 0.7227478639830266,
-      "grad_norm": 0.0015790280885994434,
-      "learning_rate": 0.00019999974276827814,
-      "loss": 46.0,
-      "step": 9453
-    },
-    {
-      "epoch": 0.7228243209664162,
-      "grad_norm": 0.002870062133297324,
-      "learning_rate": 0.00019999974271379432,
-      "loss": 46.0,
-      "step": 9454
-    },
-    {
-      "epoch": 0.722900777949806,
-      "grad_norm": 0.0038742884062230587,
-      "learning_rate": 0.00019999974265930476,
-      "loss": 46.0,
-      "step": 9455
-    },
-    {
-      "epoch": 0.7229772349331957,
-      "grad_norm": 0.0011135550448670983,
-      "learning_rate": 0.00019999974260480943,
-      "loss": 46.0,
-      "step": 9456
-    },
-    {
-      "epoch": 0.7230536919165854,
-      "grad_norm": 0.0004728232743218541,
-      "learning_rate": 0.00019999974255030834,
-      "loss": 46.0,
-      "step": 9457
-    },
-    {
-      "epoch": 0.7231301488999752,
-      "grad_norm": 0.00045553603558801115,
-      "learning_rate": 0.00019999974249580146,
-      "loss": 46.0,
-      "step": 9458
-    },
-    {
-      "epoch": 0.7232066058833648,
-      "grad_norm": 0.0006534690619446337,
-      "learning_rate": 0.0001999997424412888,
-      "loss": 46.0,
-      "step": 9459
-    },
-    {
-      "epoch": 0.7232830628667546,
-      "grad_norm": 0.00052379269618541,
-      "learning_rate": 0.0001999997423867704,
-      "loss": 46.0,
-      "step": 9460
-    },
-    {
-      "epoch": 0.7233595198501444,
-      "grad_norm": 0.000713361834641546,
-      "learning_rate": 0.00019999974233224623,
-      "loss": 46.0,
-      "step": 9461
-    },
-    {
-      "epoch": 0.723435976833534,
-      "grad_norm": 0.001291573396883905,
-      "learning_rate": 0.00019999974227771628,
-      "loss": 46.0,
-      "step": 9462
-    },
-    {
-      "epoch": 0.7235124338169238,
-      "grad_norm": 0.0012073114048689604,
-      "learning_rate": 0.00019999974222318056,
-      "loss": 46.0,
-      "step": 9463
-    },
-    {
-      "epoch": 0.7235888908003135,
-      "grad_norm": 0.0005278547760099173,
-      "learning_rate": 0.0001999997421686391,
-      "loss": 46.0,
-      "step": 9464
-    },
-    {
-      "epoch": 0.7236653477837032,
-      "grad_norm": 0.0012327833101153374,
-      "learning_rate": 0.00019999974211409182,
-      "loss": 46.0,
-      "step": 9465
-    },
-    {
-      "epoch": 0.7237418047670929,
-      "grad_norm": 0.0005570958601310849,
-      "learning_rate": 0.0001999997420595388,
-      "loss": 46.0,
-      "step": 9466
-    },
-    {
-      "epoch": 0.7238182617504826,
-      "grad_norm": 0.005379240959882736,
-      "learning_rate": 0.00019999974200498,
-      "loss": 46.0,
-      "step": 9467
-    },
-    {
-      "epoch": 0.7238947187338723,
-      "grad_norm": 0.003918815404176712,
-      "learning_rate": 0.00019999974195041543,
-      "loss": 46.0,
-      "step": 9468
-    },
-    {
-      "epoch": 0.7239711757172621,
-      "grad_norm": 0.0003865146136377007,
-      "learning_rate": 0.0001999997418958451,
-      "loss": 46.0,
-      "step": 9469
-    },
-    {
-      "epoch": 0.7240476327006518,
-      "grad_norm": 0.0012673111632466316,
-      "learning_rate": 0.000199999741841269,
-      "loss": 46.0,
-      "step": 9470
-    },
-    {
-      "epoch": 0.7241240896840415,
-      "grad_norm": 0.0009129495010711253,
-      "learning_rate": 0.0001999997417866871,
-      "loss": 46.0,
-      "step": 9471
-    },
-    {
-      "epoch": 0.7242005466674313,
-      "grad_norm": 0.002420772099867463,
-      "learning_rate": 0.00019999974173209948,
-      "loss": 46.0,
-      "step": 9472
-    },
-    {
-      "epoch": 0.7242770036508209,
-      "grad_norm": 0.0008737086900509894,
-      "learning_rate": 0.00019999974167750605,
-      "loss": 46.0,
-      "step": 9473
-    },
-    {
-      "epoch": 0.7243534606342107,
-      "grad_norm": 0.0014347423566505313,
-      "learning_rate": 0.00019999974162290688,
-      "loss": 46.0,
-      "step": 9474
-    },
-    {
-      "epoch": 0.7244299176176004,
-      "grad_norm": 0.0052198376506567,
-      "learning_rate": 0.00019999974156830193,
-      "loss": 46.0,
-      "step": 9475
-    },
-    {
-      "epoch": 0.7245063746009901,
-      "grad_norm": 0.0011726554948836565,
-      "learning_rate": 0.0001999997415136912,
-      "loss": 46.0,
-      "step": 9476
-    },
-    {
-      "epoch": 0.7245828315843799,
-      "grad_norm": 0.0011018315562978387,
-      "learning_rate": 0.00019999974145907471,
-      "loss": 46.0,
-      "step": 9477
-    },
-    {
-      "epoch": 0.7246592885677695,
-      "grad_norm": 0.0013048032997176051,
-      "learning_rate": 0.00019999974140445245,
-      "loss": 46.0,
-      "step": 9478
-    },
-    {
-      "epoch": 0.7247357455511593,
-      "grad_norm": 0.000518255343195051,
-      "learning_rate": 0.00019999974134982443,
-      "loss": 46.0,
-      "step": 9479
-    },
-    {
-      "epoch": 0.724812202534549,
-      "grad_norm": 0.0013148487778380513,
-      "learning_rate": 0.00019999974129519062,
-      "loss": 46.0,
-      "step": 9480
-    },
-    {
-      "epoch": 0.7248886595179387,
-      "grad_norm": 0.0003296970680821687,
-      "learning_rate": 0.00019999974124055106,
-      "loss": 46.0,
-      "step": 9481
-    },
-    {
-      "epoch": 0.7249651165013284,
-      "grad_norm": 0.0005345446988940239,
-      "learning_rate": 0.00019999974118590572,
-      "loss": 46.0,
-      "step": 9482
-    },
-    {
-      "epoch": 0.7250415734847182,
-      "grad_norm": 0.0007220385014079511,
-      "learning_rate": 0.0001999997411312546,
-      "loss": 46.0,
-      "step": 9483
-    },
-    {
-      "epoch": 0.7251180304681079,
-      "grad_norm": 0.0019289315678179264,
-      "learning_rate": 0.00019999974107659773,
-      "loss": 46.0,
-      "step": 9484
-    },
-    {
-      "epoch": 0.7251944874514976,
-      "grad_norm": 0.004281338769942522,
-      "learning_rate": 0.00019999974102193508,
-      "loss": 46.0,
-      "step": 9485
-    },
-    {
-      "epoch": 0.7252709444348874,
-      "grad_norm": 0.003049361752346158,
-      "learning_rate": 0.00019999974096726665,
-      "loss": 46.0,
-      "step": 9486
-    },
-    {
-      "epoch": 0.725347401418277,
-      "grad_norm": 0.004174086730927229,
-      "learning_rate": 0.00019999974091259248,
-      "loss": 46.0,
-      "step": 9487
-    },
-    {
-      "epoch": 0.7254238584016668,
-      "grad_norm": 0.0007800482562743127,
-      "learning_rate": 0.00019999974085791253,
-      "loss": 46.0,
-      "step": 9488
-    },
-    {
-      "epoch": 0.7255003153850564,
-      "grad_norm": 0.00039280750206671655,
-      "learning_rate": 0.0001999997408032268,
-      "loss": 46.0,
-      "step": 9489
-    },
-    {
-      "epoch": 0.7255767723684462,
-      "grad_norm": 0.0008713495335541666,
-      "learning_rate": 0.0001999997407485353,
-      "loss": 46.0,
-      "step": 9490
-    },
-    {
-      "epoch": 0.725653229351836,
-      "grad_norm": 0.002624642802402377,
-      "learning_rate": 0.00019999974069383804,
-      "loss": 46.0,
-      "step": 9491
-    },
-    {
-      "epoch": 0.7257296863352256,
-      "grad_norm": 0.0021717939525842667,
-      "learning_rate": 0.000199999740639135,
-      "loss": 46.0,
-      "step": 9492
-    },
-    {
-      "epoch": 0.7258061433186154,
-      "grad_norm": 0.0008572438382543623,
-      "learning_rate": 0.00019999974058442622,
-      "loss": 46.0,
-      "step": 9493
-    },
-    {
-      "epoch": 0.7258826003020051,
-      "grad_norm": 0.0007696160464547575,
-      "learning_rate": 0.00019999974052971166,
-      "loss": 46.0,
-      "step": 9494
-    },
-    {
-      "epoch": 0.7259590572853948,
-      "grad_norm": 0.0011571563081815839,
-      "learning_rate": 0.0001999997404749913,
-      "loss": 46.0,
-      "step": 9495
-    },
-    {
-      "epoch": 0.7260355142687845,
-      "grad_norm": 0.0063188462518155575,
-      "learning_rate": 0.0001999997404202652,
-      "loss": 46.0,
-      "step": 9496
-    },
-    {
-      "epoch": 0.7261119712521742,
-      "grad_norm": 0.0014230284141376615,
-      "learning_rate": 0.00019999974036553333,
-      "loss": 46.0,
-      "step": 9497
-    },
-    {
-      "epoch": 0.726188428235564,
-      "grad_norm": 0.0006551842670887709,
-      "learning_rate": 0.00019999974031079568,
-      "loss": 46.0,
-      "step": 9498
-    },
-    {
-      "epoch": 0.7262648852189537,
-      "grad_norm": 0.004869310650974512,
-      "learning_rate": 0.00019999974025605222,
-      "loss": 46.0,
-      "step": 9499
-    },
-    {
-      "epoch": 0.7263413422023434,
-      "grad_norm": 0.0010433326242491603,
-      "learning_rate": 0.00019999974020130305,
-      "loss": 46.0,
-      "step": 9500
-    },
-    {
-      "epoch": 0.7264177991857331,
-      "grad_norm": 0.0059239184483885765,
-      "learning_rate": 0.0001999997401465481,
-      "loss": 46.0,
-      "step": 9501
-    },
-    {
-      "epoch": 0.7264942561691229,
-      "grad_norm": 0.0012868397170677781,
-      "learning_rate": 0.00019999974009178738,
-      "loss": 46.0,
-      "step": 9502
-    },
-    {
-      "epoch": 0.7265707131525125,
-      "grad_norm": 0.0005700111505575478,
-      "learning_rate": 0.00019999974003702086,
-      "loss": 46.0,
-      "step": 9503
-    },
-    {
-      "epoch": 0.7266471701359023,
-      "grad_norm": 0.0010297095868736506,
-      "learning_rate": 0.00019999973998224862,
-      "loss": 46.0,
-      "step": 9504
-    },
-    {
-      "epoch": 0.7267236271192921,
-      "grad_norm": 0.0006494633853435516,
-      "learning_rate": 0.00019999973992747058,
-      "loss": 46.0,
-      "step": 9505
-    },
-    {
-      "epoch": 0.7268000841026817,
-      "grad_norm": 0.0012249941937625408,
-      "learning_rate": 0.00019999973987268677,
-      "loss": 46.0,
-      "step": 9506
-    },
-    {
-      "epoch": 0.7268765410860715,
-      "grad_norm": 0.0019425624050199986,
-      "learning_rate": 0.0001999997398178972,
-      "loss": 46.0,
-      "step": 9507
-    },
-    {
-      "epoch": 0.7269529980694611,
-      "grad_norm": 0.0015153740532696247,
-      "learning_rate": 0.00019999973976310184,
-      "loss": 46.0,
-      "step": 9508
-    },
-    {
-      "epoch": 0.7270294550528509,
-      "grad_norm": 0.0012835317756980658,
-      "learning_rate": 0.00019999973970830074,
-      "loss": 46.0,
-      "step": 9509
-    },
-    {
-      "epoch": 0.7271059120362406,
-      "grad_norm": 0.0008624490583315492,
-      "learning_rate": 0.00019999973965349386,
-      "loss": 46.0,
-      "step": 9510
-    },
-    {
-      "epoch": 0.7271823690196303,
-      "grad_norm": 0.0003058585280086845,
-      "learning_rate": 0.0001999997395986812,
-      "loss": 46.0,
-      "step": 9511
-    },
-    {
-      "epoch": 0.72725882600302,
-      "grad_norm": 0.000640116457361728,
-      "learning_rate": 0.00019999973954386277,
-      "loss": 46.0,
-      "step": 9512
-    },
-    {
-      "epoch": 0.7273352829864098,
-      "grad_norm": 0.004966209176927805,
-      "learning_rate": 0.00019999973948903857,
-      "loss": 46.0,
-      "step": 9513
-    },
-    {
-      "epoch": 0.7274117399697995,
-      "grad_norm": 0.0007462220964953303,
-      "learning_rate": 0.00019999973943420863,
-      "loss": 46.0,
-      "step": 9514
-    },
-    {
-      "epoch": 0.7274881969531892,
-      "grad_norm": 0.0009834475349634886,
-      "learning_rate": 0.0001999997393793729,
-      "loss": 46.0,
-      "step": 9515
-    },
-    {
-      "epoch": 0.727564653936579,
-      "grad_norm": 0.0015895604155957699,
-      "learning_rate": 0.0001999997393245314,
-      "loss": 46.0,
-      "step": 9516
-    },
-    {
-      "epoch": 0.7276411109199686,
-      "grad_norm": 0.003138477448374033,
-      "learning_rate": 0.00019999973926968415,
-      "loss": 46.0,
-      "step": 9517
-    },
-    {
-      "epoch": 0.7277175679033584,
-      "grad_norm": 0.0010540066286921501,
-      "learning_rate": 0.0001999997392148311,
-      "loss": 46.0,
-      "step": 9518
-    },
-    {
-      "epoch": 0.727794024886748,
-      "grad_norm": 0.0012817922979593277,
-      "learning_rate": 0.0001999997391599723,
-      "loss": 46.0,
-      "step": 9519
-    },
-    {
-      "epoch": 0.7278704818701378,
-      "grad_norm": 0.0017212582752108574,
-      "learning_rate": 0.00019999973910510773,
-      "loss": 46.0,
-      "step": 9520
-    },
-    {
-      "epoch": 0.7279469388535276,
-      "grad_norm": 0.0005717522581107914,
-      "learning_rate": 0.00019999973905023737,
-      "loss": 46.0,
-      "step": 9521
-    },
-    {
-      "epoch": 0.7280233958369172,
-      "grad_norm": 0.000797246175352484,
-      "learning_rate": 0.00019999973899536126,
-      "loss": 46.0,
-      "step": 9522
-    },
-    {
-      "epoch": 0.728099852820307,
-      "grad_norm": 0.0013025507796555758,
-      "learning_rate": 0.00019999973894047938,
-      "loss": 46.0,
-      "step": 9523
-    },
-    {
-      "epoch": 0.7281763098036967,
-      "grad_norm": 0.005301415454596281,
-      "learning_rate": 0.0001999997388855917,
-      "loss": 46.0,
-      "step": 9524
-    },
-    {
-      "epoch": 0.7282527667870864,
-      "grad_norm": 0.0012863590382039547,
-      "learning_rate": 0.0001999997388306983,
-      "loss": 46.0,
-      "step": 9525
-    },
-    {
-      "epoch": 0.7283292237704762,
-      "grad_norm": 0.0009101226460188627,
-      "learning_rate": 0.0001999997387757991,
-      "loss": 46.0,
-      "step": 9526
-    },
-    {
-      "epoch": 0.7284056807538658,
-      "grad_norm": 0.0009201113716699183,
-      "learning_rate": 0.00019999973872089416,
-      "loss": 46.0,
-      "step": 9527
-    },
-    {
-      "epoch": 0.7284821377372556,
-      "grad_norm": 0.0028858466539531946,
-      "learning_rate": 0.0001999997386659834,
-      "loss": 46.0,
-      "step": 9528
-    },
-    {
-      "epoch": 0.7285585947206453,
-      "grad_norm": 0.00046491200919263065,
-      "learning_rate": 0.0001999997386110669,
-      "loss": 46.0,
-      "step": 9529
-    },
-    {
-      "epoch": 0.728635051704035,
-      "grad_norm": 0.0006526696961373091,
-      "learning_rate": 0.00019999973855614465,
-      "loss": 46.0,
-      "step": 9530
-    },
-    {
-      "epoch": 0.7287115086874247,
-      "grad_norm": 0.0004188350576441735,
-      "learning_rate": 0.00019999973850121658,
-      "loss": 46.0,
-      "step": 9531
-    },
-    {
-      "epoch": 0.7287879656708145,
-      "grad_norm": 0.008840947411954403,
-      "learning_rate": 0.0001999997384462828,
-      "loss": 46.0,
-      "step": 9532
-    },
-    {
-      "epoch": 0.7288644226542041,
-      "grad_norm": 0.0023184798192232847,
-      "learning_rate": 0.0001999997383913432,
-      "loss": 46.0,
-      "step": 9533
-    },
-    {
-      "epoch": 0.7289408796375939,
-      "grad_norm": 0.00394762959331274,
-      "learning_rate": 0.00019999973833639785,
-      "loss": 46.0,
-      "step": 9534
-    },
-    {
-      "epoch": 0.7290173366209837,
-      "grad_norm": 0.0009867673506960273,
-      "learning_rate": 0.00019999973828144674,
-      "loss": 46.0,
-      "step": 9535
-    },
-    {
-      "epoch": 0.7290937936043733,
-      "grad_norm": 0.00379791553132236,
-      "learning_rate": 0.00019999973822648986,
-      "loss": 46.0,
-      "step": 9536
-    },
-    {
-      "epoch": 0.7291702505877631,
-      "grad_norm": 0.0010544833494350314,
-      "learning_rate": 0.0001999997381715272,
-      "loss": 46.0,
-      "step": 9537
-    },
-    {
-      "epoch": 0.7292467075711527,
-      "grad_norm": 0.0007636480731889606,
-      "learning_rate": 0.00019999973811655875,
-      "loss": 46.0,
-      "step": 9538
-    },
-    {
-      "epoch": 0.7293231645545425,
-      "grad_norm": 0.0008713103015907109,
-      "learning_rate": 0.00019999973806158455,
-      "loss": 46.0,
-      "step": 9539
-    },
-    {
-      "epoch": 0.7293996215379323,
-      "grad_norm": 0.001506561879068613,
-      "learning_rate": 0.00019999973800660458,
-      "loss": 46.0,
-      "step": 9540
-    },
-    {
-      "epoch": 0.7294760785213219,
-      "grad_norm": 0.0005990981007926166,
-      "learning_rate": 0.0001999997379516189,
-      "loss": 46.0,
-      "step": 9541
-    },
-    {
-      "epoch": 0.7295525355047117,
-      "grad_norm": 0.0004713807429652661,
-      "learning_rate": 0.00019999973789662737,
-      "loss": 46.0,
-      "step": 9542
-    },
-    {
-      "epoch": 0.7296289924881014,
-      "grad_norm": 0.0010668407194316387,
-      "learning_rate": 0.00019999973784163007,
-      "loss": 46.0,
-      "step": 9543
-    },
-    {
-      "epoch": 0.7297054494714911,
-      "grad_norm": 0.003675210988149047,
-      "learning_rate": 0.00019999973778662703,
-      "loss": 46.0,
-      "step": 9544
-    },
-    {
-      "epoch": 0.7297819064548808,
-      "grad_norm": 0.0007254960364662111,
-      "learning_rate": 0.00019999973773161825,
-      "loss": 46.0,
-      "step": 9545
-    },
-    {
-      "epoch": 0.7298583634382706,
-      "grad_norm": 0.0009725384297780693,
-      "learning_rate": 0.00019999973767660366,
-      "loss": 46.0,
-      "step": 9546
-    },
-    {
-      "epoch": 0.7299348204216602,
-      "grad_norm": 0.0052064149640500546,
-      "learning_rate": 0.0001999997376215833,
-      "loss": 46.0,
-      "step": 9547
-    },
-    {
-      "epoch": 0.73001127740505,
-      "grad_norm": 0.000842469627968967,
-      "learning_rate": 0.00019999973756655717,
-      "loss": 46.0,
-      "step": 9548
-    },
-    {
-      "epoch": 0.7300877343884397,
-      "grad_norm": 0.0023045376874506474,
-      "learning_rate": 0.0001999997375115253,
-      "loss": 46.0,
-      "step": 9549
-    },
-    {
-      "epoch": 0.7301641913718294,
-      "grad_norm": 0.0016457663150504231,
-      "learning_rate": 0.00019999973745648763,
-      "loss": 46.0,
-      "step": 9550
-    },
-    {
-      "epoch": 0.7302406483552192,
-      "grad_norm": 0.00207923143170774,
-      "learning_rate": 0.0001999997374014442,
-      "loss": 46.0,
-      "step": 9551
-    },
-    {
-      "epoch": 0.7303171053386088,
-      "grad_norm": 0.0013082396471872926,
-      "learning_rate": 0.000199999737346395,
-      "loss": 46.0,
-      "step": 9552
-    },
-    {
-      "epoch": 0.7303935623219986,
-      "grad_norm": 0.0026289978995919228,
-      "learning_rate": 0.00019999973729134003,
-      "loss": 46.0,
-      "step": 9553
-    },
-    {
-      "epoch": 0.7304700193053884,
-      "grad_norm": 0.0012528073275461793,
-      "learning_rate": 0.0001999997372362793,
-      "loss": 46.0,
-      "step": 9554
-    },
-    {
-      "epoch": 0.730546476288778,
-      "grad_norm": 0.003209579735994339,
-      "learning_rate": 0.0001999997371812128,
-      "loss": 46.0,
-      "step": 9555
-    },
-    {
-      "epoch": 0.7306229332721678,
-      "grad_norm": 0.0030588265508413315,
-      "learning_rate": 0.0001999997371261405,
-      "loss": 46.0,
-      "step": 9556
-    },
-    {
-      "epoch": 0.7306993902555574,
-      "grad_norm": 0.0012199330376461148,
-      "learning_rate": 0.00019999973707106246,
-      "loss": 46.0,
-      "step": 9557
-    },
-    {
-      "epoch": 0.7307758472389472,
-      "grad_norm": 0.00031229775049723685,
-      "learning_rate": 0.00019999973701597865,
-      "loss": 46.0,
-      "step": 9558
-    },
-    {
-      "epoch": 0.7308523042223369,
-      "grad_norm": 0.014457184821367264,
-      "learning_rate": 0.00019999973696088906,
-      "loss": 46.0,
-      "step": 9559
-    },
-    {
-      "epoch": 0.7309287612057266,
-      "grad_norm": 0.003756424644961953,
-      "learning_rate": 0.00019999973690579373,
-      "loss": 46.0,
-      "step": 9560
-    },
-    {
-      "epoch": 0.7310052181891163,
-      "grad_norm": 0.000609002890996635,
-      "learning_rate": 0.0001999997368506926,
-      "loss": 46.0,
-      "step": 9561
-    },
-    {
-      "epoch": 0.7310816751725061,
-      "grad_norm": 0.0040803165175020695,
-      "learning_rate": 0.0001999997367955857,
-      "loss": 46.0,
-      "step": 9562
-    },
-    {
-      "epoch": 0.7311581321558958,
-      "grad_norm": 0.0038511387538164854,
-      "learning_rate": 0.00019999973674047304,
-      "loss": 46.0,
-      "step": 9563
-    },
-    {
-      "epoch": 0.7312345891392855,
-      "grad_norm": 0.00056040333583951,
-      "learning_rate": 0.00019999973668535459,
-      "loss": 46.0,
-      "step": 9564
-    },
-    {
-      "epoch": 0.7313110461226753,
-      "grad_norm": 0.0032748880330473185,
-      "learning_rate": 0.00019999973663023039,
-      "loss": 46.0,
-      "step": 9565
-    },
-    {
-      "epoch": 0.7313875031060649,
-      "grad_norm": 0.0007885222439654171,
-      "learning_rate": 0.00019999973657510044,
-      "loss": 46.0,
-      "step": 9566
-    },
-    {
-      "epoch": 0.7314639600894547,
-      "grad_norm": 0.0008641025051474571,
-      "learning_rate": 0.0001999997365199647,
-      "loss": 46.0,
-      "step": 9567
-    },
-    {
-      "epoch": 0.7315404170728443,
-      "grad_norm": 0.01447670254856348,
-      "learning_rate": 0.00019999973646482317,
-      "loss": 46.0,
-      "step": 9568
-    },
-    {
-      "epoch": 0.7316168740562341,
-      "grad_norm": 0.000940754369366914,
-      "learning_rate": 0.0001999997364096759,
-      "loss": 46.0,
-      "step": 9569
-    },
-    {
-      "epoch": 0.7316933310396239,
-      "grad_norm": 0.0031623272225260735,
-      "learning_rate": 0.00019999973635452284,
-      "loss": 46.0,
-      "step": 9570
-    },
-    {
-      "epoch": 0.7317697880230135,
-      "grad_norm": 0.0012078157160431147,
-      "learning_rate": 0.00019999973629936403,
-      "loss": 46.0,
-      "step": 9571
-    },
-    {
-      "epoch": 0.7318462450064033,
-      "grad_norm": 0.0011179299326613545,
-      "learning_rate": 0.00019999973624419944,
-      "loss": 46.0,
-      "step": 9572
-    },
-    {
-      "epoch": 0.731922701989793,
-      "grad_norm": 0.001043855445459485,
-      "learning_rate": 0.00019999973618902908,
-      "loss": 46.0,
-      "step": 9573
-    },
-    {
-      "epoch": 0.7319991589731827,
-      "grad_norm": 0.0009011423098854721,
-      "learning_rate": 0.00019999973613385298,
-      "loss": 46.0,
-      "step": 9574
-    },
-    {
-      "epoch": 0.7320756159565724,
-      "grad_norm": 0.002464559394866228,
-      "learning_rate": 0.00019999973607867107,
-      "loss": 46.0,
-      "step": 9575
-    },
-    {
-      "epoch": 0.7321520729399622,
-      "grad_norm": 0.006772432941943407,
-      "learning_rate": 0.00019999973602348342,
-      "loss": 46.0,
-      "step": 9576
-    },
-    {
-      "epoch": 0.7322285299233519,
-      "grad_norm": 0.003458540653809905,
-      "learning_rate": 0.00019999973596828997,
-      "loss": 46.0,
-      "step": 9577
-    },
-    {
-      "epoch": 0.7323049869067416,
-      "grad_norm": 0.0009902992751449347,
-      "learning_rate": 0.00019999973591309077,
-      "loss": 46.0,
-      "step": 9578
-    },
-    {
-      "epoch": 0.7323814438901313,
-      "grad_norm": 0.0011761031346395612,
-      "learning_rate": 0.0001999997358578858,
-      "loss": 46.0,
-      "step": 9579
-    },
-    {
-      "epoch": 0.732457900873521,
-      "grad_norm": 0.0012663932284340262,
-      "learning_rate": 0.00019999973580267505,
-      "loss": 46.0,
-      "step": 9580
-    },
-    {
-      "epoch": 0.7325343578569108,
-      "grad_norm": 0.0015342329861596227,
-      "learning_rate": 0.0001999997357474585,
-      "loss": 46.0,
-      "step": 9581
-    },
-    {
-      "epoch": 0.7326108148403004,
-      "grad_norm": 0.002045195549726486,
-      "learning_rate": 0.00019999973569223624,
-      "loss": 46.0,
-      "step": 9582
-    },
-    {
-      "epoch": 0.7326872718236902,
-      "grad_norm": 0.001635118154808879,
-      "learning_rate": 0.0001999997356370082,
-      "loss": 46.0,
-      "step": 9583
-    },
-    {
-      "epoch": 0.73276372880708,
-      "grad_norm": 0.0013396850554272532,
-      "learning_rate": 0.00019999973558177439,
-      "loss": 46.0,
-      "step": 9584
-    },
-    {
-      "epoch": 0.7328401857904696,
-      "grad_norm": 0.00120553991291672,
-      "learning_rate": 0.0001999997355265348,
-      "loss": 46.0,
-      "step": 9585
-    },
-    {
-      "epoch": 0.7329166427738594,
-      "grad_norm": 0.0008809276041574776,
-      "learning_rate": 0.00019999973547128944,
-      "loss": 46.0,
-      "step": 9586
-    },
-    {
-      "epoch": 0.732993099757249,
-      "grad_norm": 0.002531450940296054,
-      "learning_rate": 0.00019999973541603828,
-      "loss": 46.0,
-      "step": 9587
-    },
-    {
-      "epoch": 0.7330695567406388,
-      "grad_norm": 0.00129566655959934,
-      "learning_rate": 0.0001999997353607814,
-      "loss": 46.0,
-      "step": 9588
-    },
-    {
-      "epoch": 0.7331460137240285,
-      "grad_norm": 0.0023907015565782785,
-      "learning_rate": 0.00019999973530551875,
-      "loss": 46.0,
-      "step": 9589
-    },
-    {
-      "epoch": 0.7332224707074182,
-      "grad_norm": 0.0010989034781232476,
-      "learning_rate": 0.0001999997352502503,
-      "loss": 46.0,
-      "step": 9590
-    },
-    {
-      "epoch": 0.733298927690808,
-      "grad_norm": 0.0020173180382698774,
-      "learning_rate": 0.0001999997351949761,
-      "loss": 46.0,
-      "step": 9591
-    },
-    {
-      "epoch": 0.7333753846741977,
-      "grad_norm": 0.0014055392239242792,
-      "learning_rate": 0.0001999997351396961,
-      "loss": 46.0,
-      "step": 9592
-    },
-    {
-      "epoch": 0.7334518416575874,
-      "grad_norm": 0.0010951553704217076,
-      "learning_rate": 0.00019999973508441035,
-      "loss": 46.0,
-      "step": 9593
-    },
-    {
-      "epoch": 0.7335282986409771,
-      "grad_norm": 0.003435376100242138,
-      "learning_rate": 0.00019999973502911886,
-      "loss": 46.0,
-      "step": 9594
-    },
-    {
-      "epoch": 0.7336047556243669,
-      "grad_norm": 0.002582656918093562,
-      "learning_rate": 0.0001999997349738216,
-      "loss": 46.0,
-      "step": 9595
-    },
-    {
-      "epoch": 0.7336812126077565,
-      "grad_norm": 0.0032408363185822964,
-      "learning_rate": 0.0001999997349185185,
-      "loss": 46.0,
-      "step": 9596
-    },
-    {
-      "epoch": 0.7337576695911463,
-      "grad_norm": 0.001116379862651229,
-      "learning_rate": 0.0001999997348632097,
-      "loss": 46.0,
-      "step": 9597
-    },
-    {
-      "epoch": 0.733834126574536,
-      "grad_norm": 0.0016293646767735481,
-      "learning_rate": 0.0001999997348078951,
-      "loss": 46.0,
-      "step": 9598
-    },
-    {
-      "epoch": 0.7339105835579257,
-      "grad_norm": 0.0011346143437549472,
-      "learning_rate": 0.00019999973475257475,
-      "loss": 46.0,
-      "step": 9599
-    },
-    {
-      "epoch": 0.7339870405413155,
-      "grad_norm": 0.001031039864756167,
-      "learning_rate": 0.0001999997346972486,
-      "loss": 46.0,
-      "step": 9600
-    },
-    {
-      "epoch": 0.7340634975247051,
-      "grad_norm": 0.0014087734743952751,
-      "learning_rate": 0.0001999997346419167,
-      "loss": 46.0,
-      "step": 9601
-    },
-    {
-      "epoch": 0.7341399545080949,
-      "grad_norm": 0.001699765445664525,
-      "learning_rate": 0.00019999973458657903,
-      "loss": 46.0,
-      "step": 9602
-    },
-    {
-      "epoch": 0.7342164114914846,
-      "grad_norm": 0.0037602009251713753,
-      "learning_rate": 0.0001999997345312356,
-      "loss": 46.0,
-      "step": 9603
-    },
-    {
-      "epoch": 0.7342928684748743,
-      "grad_norm": 0.0031309055630117655,
-      "learning_rate": 0.00019999973447588638,
-      "loss": 46.0,
-      "step": 9604
-    },
-    {
-      "epoch": 0.734369325458264,
-      "grad_norm": 0.001099355984479189,
-      "learning_rate": 0.0001999997344205314,
-      "loss": 46.0,
-      "step": 9605
-    },
-    {
-      "epoch": 0.7344457824416538,
-      "grad_norm": 0.0019889497198164463,
-      "learning_rate": 0.00019999973436517066,
-      "loss": 46.0,
-      "step": 9606
-    },
-    {
-      "epoch": 0.7345222394250435,
-      "grad_norm": 0.001312068896368146,
-      "learning_rate": 0.00019999973430980412,
-      "loss": 46.0,
-      "step": 9607
-    },
-    {
-      "epoch": 0.7345986964084332,
-      "grad_norm": 0.0009766578441485763,
-      "learning_rate": 0.00019999973425443183,
-      "loss": 46.0,
-      "step": 9608
-    },
-    {
-      "epoch": 0.7346751533918229,
-      "grad_norm": 0.0015836225356906652,
-      "learning_rate": 0.0001999997341990538,
-      "loss": 46.0,
-      "step": 9609
-    },
-    {
-      "epoch": 0.7347516103752126,
-      "grad_norm": 0.005435020197182894,
-      "learning_rate": 0.00019999973414366998,
-      "loss": 46.0,
-      "step": 9610
-    },
-    {
-      "epoch": 0.7348280673586024,
-      "grad_norm": 0.0008086469606496394,
-      "learning_rate": 0.00019999973408828036,
-      "loss": 46.0,
-      "step": 9611
-    },
-    {
-      "epoch": 0.734904524341992,
-      "grad_norm": 0.0009257084457203746,
-      "learning_rate": 0.000199999734032885,
-      "loss": 46.0,
-      "step": 9612
-    },
-    {
-      "epoch": 0.7349809813253818,
-      "grad_norm": 0.006190138403326273,
-      "learning_rate": 0.00019999973397748387,
-      "loss": 46.0,
-      "step": 9613
-    },
-    {
-      "epoch": 0.7350574383087716,
-      "grad_norm": 0.0014728892128914595,
-      "learning_rate": 0.00019999973392207697,
-      "loss": 46.0,
-      "step": 9614
-    },
-    {
-      "epoch": 0.7351338952921612,
-      "grad_norm": 0.0016258620889857411,
-      "learning_rate": 0.0001999997338666643,
-      "loss": 46.0,
-      "step": 9615
-    },
-    {
-      "epoch": 0.735210352275551,
-      "grad_norm": 0.0003624729870352894,
-      "learning_rate": 0.00019999973381124587,
-      "loss": 46.0,
-      "step": 9616
-    },
-    {
-      "epoch": 0.7352868092589407,
-      "grad_norm": 0.0011338959448039532,
-      "learning_rate": 0.00019999973375582164,
-      "loss": 46.0,
-      "step": 9617
-    },
-    {
-      "epoch": 0.7353632662423304,
-      "grad_norm": 0.0010770147200673819,
-      "learning_rate": 0.00019999973370039164,
-      "loss": 46.0,
-      "step": 9618
-    },
-    {
-      "epoch": 0.7354397232257202,
-      "grad_norm": 0.00017523637507110834,
-      "learning_rate": 0.0001999997336449559,
-      "loss": 46.0,
-      "step": 9619
-    },
-    {
-      "epoch": 0.7355161802091098,
-      "grad_norm": 0.001183304819278419,
-      "learning_rate": 0.00019999973358951438,
-      "loss": 46.0,
-      "step": 9620
-    },
-    {
-      "epoch": 0.7355926371924996,
-      "grad_norm": 0.0023626259062439203,
-      "learning_rate": 0.00019999973353406709,
-      "loss": 46.0,
-      "step": 9621
-    },
-    {
-      "epoch": 0.7356690941758893,
-      "grad_norm": 0.0006327573210000992,
-      "learning_rate": 0.00019999973347861402,
-      "loss": 46.0,
-      "step": 9622
-    },
-    {
-      "epoch": 0.735745551159279,
-      "grad_norm": 0.0005722071509808302,
-      "learning_rate": 0.0001999997334231552,
-      "loss": 46.0,
-      "step": 9623
-    },
-    {
-      "epoch": 0.7358220081426687,
-      "grad_norm": 0.0014855496119707823,
-      "learning_rate": 0.00019999973336769062,
-      "loss": 46.0,
-      "step": 9624
-    },
-    {
-      "epoch": 0.7358984651260585,
-      "grad_norm": 0.017810819670557976,
-      "learning_rate": 0.00019999973331222024,
-      "loss": 46.0,
-      "step": 9625
-    },
-    {
-      "epoch": 0.7359749221094481,
-      "grad_norm": 0.0010145645355805755,
-      "learning_rate": 0.0001999997332567441,
-      "loss": 46.0,
-      "step": 9626
-    },
-    {
-      "epoch": 0.7360513790928379,
-      "grad_norm": 0.0006681100931018591,
-      "learning_rate": 0.0001999997332012622,
-      "loss": 46.0,
-      "step": 9627
-    },
-    {
-      "epoch": 0.7361278360762276,
-      "grad_norm": 0.0016074924496933818,
-      "learning_rate": 0.00019999973314577453,
-      "loss": 46.0,
-      "step": 9628
-    },
-    {
-      "epoch": 0.7362042930596173,
-      "grad_norm": 0.0013465629890561104,
-      "learning_rate": 0.00019999973309028107,
-      "loss": 46.0,
-      "step": 9629
-    },
-    {
-      "epoch": 0.7362807500430071,
-      "grad_norm": 0.00095395214157179,
-      "learning_rate": 0.00019999973303478185,
-      "loss": 46.0,
-      "step": 9630
-    },
-    {
-      "epoch": 0.7363572070263967,
-      "grad_norm": 0.0012761947000399232,
-      "learning_rate": 0.00019999973297927688,
-      "loss": 46.0,
-      "step": 9631
-    },
-    {
-      "epoch": 0.7364336640097865,
-      "grad_norm": 0.004218399990350008,
-      "learning_rate": 0.00019999973292376613,
-      "loss": 46.0,
-      "step": 9632
-    },
-    {
-      "epoch": 0.7365101209931763,
-      "grad_norm": 0.0012143789790570736,
-      "learning_rate": 0.0001999997328682496,
-      "loss": 46.0,
-      "step": 9633
-    },
-    {
-      "epoch": 0.7365865779765659,
-      "grad_norm": 0.0026241170708090067,
-      "learning_rate": 0.0001999997328127273,
-      "loss": 46.0,
-      "step": 9634
-    },
-    {
-      "epoch": 0.7366630349599557,
-      "grad_norm": 0.0006415036623366177,
-      "learning_rate": 0.00019999973275719923,
-      "loss": 46.0,
-      "step": 9635
-    },
-    {
-      "epoch": 0.7367394919433454,
-      "grad_norm": 0.009661107324063778,
-      "learning_rate": 0.00019999973270166542,
-      "loss": 46.0,
-      "step": 9636
-    },
-    {
-      "epoch": 0.7368159489267351,
-      "grad_norm": 0.0006895619444549084,
-      "learning_rate": 0.0001999997326461258,
-      "loss": 46.0,
-      "step": 9637
-    },
-    {
-      "epoch": 0.7368924059101248,
-      "grad_norm": 0.010084619745612144,
-      "learning_rate": 0.00019999973259058045,
-      "loss": 46.0,
-      "step": 9638
-    },
-    {
-      "epoch": 0.7369688628935145,
-      "grad_norm": 0.0026864074170589447,
-      "learning_rate": 0.0001999997325350293,
-      "loss": 46.0,
-      "step": 9639
-    },
-    {
-      "epoch": 0.7370453198769042,
-      "grad_norm": 0.0007298939162865281,
-      "learning_rate": 0.0001999997324794724,
-      "loss": 46.0,
-      "step": 9640
-    },
-    {
-      "epoch": 0.737121776860294,
-      "grad_norm": 0.0010473444126546383,
-      "learning_rate": 0.0001999997324239097,
-      "loss": 46.0,
-      "step": 9641
-    },
-    {
-      "epoch": 0.7371982338436837,
-      "grad_norm": 0.002117954893037677,
-      "learning_rate": 0.00019999973236834126,
-      "loss": 46.0,
-      "step": 9642
-    },
-    {
-      "epoch": 0.7372746908270734,
-      "grad_norm": 0.001089767785742879,
-      "learning_rate": 0.00019999973231276706,
-      "loss": 46.0,
-      "step": 9643
-    },
-    {
-      "epoch": 0.7373511478104632,
-      "grad_norm": 0.007818181067705154,
-      "learning_rate": 0.00019999973225718706,
-      "loss": 46.0,
-      "step": 9644
-    },
-    {
-      "epoch": 0.7374276047938528,
-      "grad_norm": 0.0018394073704257607,
-      "learning_rate": 0.00019999973220160132,
-      "loss": 46.0,
-      "step": 9645
-    },
-    {
-      "epoch": 0.7375040617772426,
-      "grad_norm": 0.0010977019555866718,
-      "learning_rate": 0.00019999973214600977,
-      "loss": 46.0,
-      "step": 9646
-    },
-    {
-      "epoch": 0.7375805187606324,
-      "grad_norm": 0.0007522209198214114,
-      "learning_rate": 0.00019999973209041248,
-      "loss": 46.0,
-      "step": 9647
-    },
-    {
-      "epoch": 0.737656975744022,
-      "grad_norm": 0.0010153116891160607,
-      "learning_rate": 0.00019999973203480942,
-      "loss": 46.0,
-      "step": 9648
-    },
-    {
-      "epoch": 0.7377334327274118,
-      "grad_norm": 0.0008870738674886525,
-      "learning_rate": 0.00019999973197920058,
-      "loss": 46.0,
-      "step": 9649
-    },
-    {
-      "epoch": 0.7378098897108014,
-      "grad_norm": 0.0005648533115163445,
-      "learning_rate": 0.00019999973192358597,
-      "loss": 46.0,
-      "step": 9650
-    },
-    {
-      "epoch": 0.7378863466941912,
-      "grad_norm": 0.0011166848707944155,
-      "learning_rate": 0.0001999997318679656,
-      "loss": 46.0,
-      "step": 9651
-    },
-    {
-      "epoch": 0.7379628036775809,
-      "grad_norm": 0.005908910185098648,
-      "learning_rate": 0.00019999973181233945,
-      "loss": 46.0,
-      "step": 9652
-    },
-    {
-      "epoch": 0.7380392606609706,
-      "grad_norm": 0.0004542939132079482,
-      "learning_rate": 0.00019999973175670755,
-      "loss": 46.0,
-      "step": 9653
-    },
-    {
-      "epoch": 0.7381157176443603,
-      "grad_norm": 0.0005001468816772103,
-      "learning_rate": 0.00019999973170106987,
-      "loss": 46.0,
-      "step": 9654
-    },
-    {
-      "epoch": 0.7381921746277501,
-      "grad_norm": 0.002464883727952838,
-      "learning_rate": 0.00019999973164542642,
-      "loss": 46.0,
-      "step": 9655
-    },
-    {
-      "epoch": 0.7382686316111398,
-      "grad_norm": 0.0045583778992295265,
-      "learning_rate": 0.0001999997315897772,
-      "loss": 46.0,
-      "step": 9656
-    },
-    {
-      "epoch": 0.7383450885945295,
-      "grad_norm": 0.0022312586661428213,
-      "learning_rate": 0.00019999973153412223,
-      "loss": 46.0,
-      "step": 9657
-    },
-    {
-      "epoch": 0.7384215455779192,
-      "grad_norm": 0.004471499938517809,
-      "learning_rate": 0.00019999973147846145,
-      "loss": 46.0,
-      "step": 9658
-    },
-    {
-      "epoch": 0.7384980025613089,
-      "grad_norm": 0.00023420136130880564,
-      "learning_rate": 0.0001999997314227949,
-      "loss": 46.0,
-      "step": 9659
-    },
-    {
-      "epoch": 0.7385744595446987,
-      "grad_norm": 0.0015983417397364974,
-      "learning_rate": 0.00019999973136712262,
-      "loss": 46.0,
-      "step": 9660
-    },
-    {
-      "epoch": 0.7386509165280883,
-      "grad_norm": 0.0041715409606695175,
-      "learning_rate": 0.00019999973131144456,
-      "loss": 46.0,
-      "step": 9661
-    },
-    {
-      "epoch": 0.7387273735114781,
-      "grad_norm": 0.006407590117305517,
-      "learning_rate": 0.0001999997312557607,
-      "loss": 46.0,
-      "step": 9662
-    },
-    {
-      "epoch": 0.7388038304948679,
-      "grad_norm": 0.001083396840840578,
-      "learning_rate": 0.0001999997312000711,
-      "loss": 46.0,
-      "step": 9663
-    },
-    {
-      "epoch": 0.7388802874782575,
-      "grad_norm": 0.0015013846568763256,
-      "learning_rate": 0.00019999973114437572,
-      "loss": 46.0,
-      "step": 9664
-    },
-    {
-      "epoch": 0.7389567444616473,
-      "grad_norm": 0.0031238263472914696,
-      "learning_rate": 0.0001999997310886746,
-      "loss": 46.0,
-      "step": 9665
-    },
-    {
-      "epoch": 0.739033201445037,
-      "grad_norm": 0.0012393043143674731,
-      "learning_rate": 0.00019999973103296766,
-      "loss": 46.0,
-      "step": 9666
-    },
-    {
-      "epoch": 0.7391096584284267,
-      "grad_norm": 0.0024985058698803186,
-      "learning_rate": 0.00019999973097725498,
-      "loss": 46.0,
-      "step": 9667
-    },
-    {
-      "epoch": 0.7391861154118164,
-      "grad_norm": 0.008887705393135548,
-      "learning_rate": 0.00019999973092153653,
-      "loss": 46.0,
-      "step": 9668
-    },
-    {
-      "epoch": 0.7392625723952061,
-      "grad_norm": 0.0010518525959923863,
-      "learning_rate": 0.0001999997308658123,
-      "loss": 46.0,
-      "step": 9669
-    },
-    {
-      "epoch": 0.7393390293785959,
-      "grad_norm": 0.0009578320314176381,
-      "learning_rate": 0.0001999997308100823,
-      "loss": 46.0,
-      "step": 9670
-    },
-    {
-      "epoch": 0.7394154863619856,
-      "grad_norm": 0.004931585397571325,
-      "learning_rate": 0.00019999973075434654,
-      "loss": 46.0,
-      "step": 9671
-    },
-    {
-      "epoch": 0.7394919433453753,
-      "grad_norm": 0.00102918257471174,
-      "learning_rate": 0.00019999973069860502,
-      "loss": 46.0,
-      "step": 9672
-    },
-    {
-      "epoch": 0.739568400328765,
-      "grad_norm": 0.003197824815288186,
-      "learning_rate": 0.00019999973064285773,
-      "loss": 46.0,
-      "step": 9673
-    },
-    {
-      "epoch": 0.7396448573121548,
-      "grad_norm": 0.0030899906996637583,
-      "learning_rate": 0.00019999973058710467,
-      "loss": 46.0,
-      "step": 9674
-    },
-    {
-      "epoch": 0.7397213142955444,
-      "grad_norm": 0.0005368997808545828,
-      "learning_rate": 0.00019999973053134583,
-      "loss": 46.0,
-      "step": 9675
-    },
-    {
-      "epoch": 0.7397977712789342,
-      "grad_norm": 0.0014108498580753803,
-      "learning_rate": 0.00019999973047558122,
-      "loss": 46.0,
-      "step": 9676
-    },
-    {
-      "epoch": 0.739874228262324,
-      "grad_norm": 0.0004695046809501946,
-      "learning_rate": 0.0001999997304198108,
-      "loss": 46.0,
-      "step": 9677
-    },
-    {
-      "epoch": 0.7399506852457136,
-      "grad_norm": 0.0023242507595568895,
-      "learning_rate": 0.00019999973036403468,
-      "loss": 46.0,
-      "step": 9678
-    },
-    {
-      "epoch": 0.7400271422291034,
-      "grad_norm": 0.0005327652324922383,
-      "learning_rate": 0.00019999973030825278,
-      "loss": 46.0,
-      "step": 9679
-    },
-    {
-      "epoch": 0.740103599212493,
-      "grad_norm": 0.0008565311436541378,
-      "learning_rate": 0.00019999973025246507,
-      "loss": 46.0,
-      "step": 9680
-    },
-    {
-      "epoch": 0.7401800561958828,
-      "grad_norm": 0.002024158136919141,
-      "learning_rate": 0.00019999973019667162,
-      "loss": 46.0,
-      "step": 9681
-    },
-    {
-      "epoch": 0.7402565131792725,
-      "grad_norm": 0.0007599971140734851,
-      "learning_rate": 0.0001999997301408724,
-      "loss": 46.0,
-      "step": 9682
-    },
-    {
-      "epoch": 0.7403329701626622,
-      "grad_norm": 0.001449097995646298,
-      "learning_rate": 0.0001999997300850674,
-      "loss": 46.0,
-      "step": 9683
-    },
-    {
-      "epoch": 0.740409427146052,
-      "grad_norm": 0.0006649132119491696,
-      "learning_rate": 0.00019999973002925666,
-      "loss": 46.0,
-      "step": 9684
-    },
-    {
-      "epoch": 0.7404858841294417,
-      "grad_norm": 0.002705745631828904,
-      "learning_rate": 0.0001999997299734401,
-      "loss": 46.0,
-      "step": 9685
-    },
-    {
-      "epoch": 0.7405623411128314,
-      "grad_norm": 0.0006340504041872919,
-      "learning_rate": 0.0001999997299176178,
-      "loss": 46.0,
-      "step": 9686
-    },
-    {
-      "epoch": 0.7406387980962211,
-      "grad_norm": 0.0028744409792125225,
-      "learning_rate": 0.00019999972986178973,
-      "loss": 46.0,
-      "step": 9687
-    },
-    {
-      "epoch": 0.7407152550796108,
-      "grad_norm": 0.0009442242444492877,
-      "learning_rate": 0.0001999997298059559,
-      "loss": 46.0,
-      "step": 9688
-    },
-    {
-      "epoch": 0.7407917120630005,
-      "grad_norm": 0.0015868137124925852,
-      "learning_rate": 0.00019999972975011626,
-      "loss": 46.0,
-      "step": 9689
-    },
-    {
-      "epoch": 0.7408681690463903,
-      "grad_norm": 0.00102346227504313,
-      "learning_rate": 0.00019999972969427088,
-      "loss": 46.0,
-      "step": 9690
-    },
-    {
-      "epoch": 0.74094462602978,
-      "grad_norm": 0.0021459476556628942,
-      "learning_rate": 0.00019999972963841972,
-      "loss": 46.0,
-      "step": 9691
-    },
-    {
-      "epoch": 0.7410210830131697,
-      "grad_norm": 0.005106296390295029,
-      "learning_rate": 0.00019999972958256281,
-      "loss": 46.0,
-      "step": 9692
-    },
-    {
-      "epoch": 0.7410975399965595,
-      "grad_norm": 0.001213806332089007,
-      "learning_rate": 0.0001999997295267001,
-      "loss": 46.0,
-      "step": 9693
-    },
-    {
-      "epoch": 0.7411739969799491,
-      "grad_norm": 0.0004491073777899146,
-      "learning_rate": 0.0001999997294708317,
-      "loss": 46.0,
-      "step": 9694
-    },
-    {
-      "epoch": 0.7412504539633389,
-      "grad_norm": 0.0002139282296411693,
-      "learning_rate": 0.00019999972941495746,
-      "loss": 46.0,
-      "step": 9695
-    },
-    {
-      "epoch": 0.7413269109467286,
-      "grad_norm": 0.0006848351331427693,
-      "learning_rate": 0.00019999972935907744,
-      "loss": 46.0,
-      "step": 9696
-    },
-    {
-      "epoch": 0.7414033679301183,
-      "grad_norm": 0.012589477933943272,
-      "learning_rate": 0.00019999972930319167,
-      "loss": 46.0,
-      "step": 9697
-    },
-    {
-      "epoch": 0.7414798249135081,
-      "grad_norm": 0.005463951267302036,
-      "learning_rate": 0.00019999972924730015,
-      "loss": 46.0,
-      "step": 9698
-    },
-    {
-      "epoch": 0.7415562818968977,
-      "grad_norm": 0.0004043464723508805,
-      "learning_rate": 0.00019999972919140284,
-      "loss": 46.0,
-      "step": 9699
-    },
-    {
-      "epoch": 0.7416327388802875,
-      "grad_norm": 0.0008798484923318028,
-      "learning_rate": 0.00019999972913549975,
-      "loss": 46.0,
-      "step": 9700
-    },
-    {
-      "epoch": 0.7417091958636772,
-      "grad_norm": 0.0009397157118655741,
-      "learning_rate": 0.00019999972907959094,
-      "loss": 46.0,
-      "step": 9701
-    },
-    {
-      "epoch": 0.7417856528470669,
-      "grad_norm": 0.0006381947896443307,
-      "learning_rate": 0.0001999997290236763,
-      "loss": 46.0,
-      "step": 9702
-    },
-    {
-      "epoch": 0.7418621098304566,
-      "grad_norm": 0.06752842664718628,
-      "learning_rate": 0.00019999972896775592,
-      "loss": 46.0,
-      "step": 9703
-    },
-    {
-      "epoch": 0.7419385668138464,
-      "grad_norm": 0.001153147080913186,
-      "learning_rate": 0.00019999972891182976,
-      "loss": 46.0,
-      "step": 9704
-    },
-    {
-      "epoch": 0.742015023797236,
-      "grad_norm": 0.011964002624154091,
-      "learning_rate": 0.00019999972885589783,
-      "loss": 46.0,
-      "step": 9705
-    },
-    {
-      "epoch": 0.7420914807806258,
-      "grad_norm": 0.0015160597395151854,
-      "learning_rate": 0.00019999972879996015,
-      "loss": 46.0,
-      "step": 9706
-    },
-    {
-      "epoch": 0.7421679377640156,
-      "grad_norm": 0.0010189347667619586,
-      "learning_rate": 0.00019999972874401668,
-      "loss": 46.0,
-      "step": 9707
-    },
-    {
-      "epoch": 0.7422443947474052,
-      "grad_norm": 0.0024226978421211243,
-      "learning_rate": 0.00019999972868806745,
-      "loss": 46.0,
-      "step": 9708
-    },
-    {
-      "epoch": 0.742320851730795,
-      "grad_norm": 0.0012253199238330126,
-      "learning_rate": 0.00019999972863211246,
-      "loss": 46.0,
-      "step": 9709
-    },
-    {
-      "epoch": 0.7423973087141846,
-      "grad_norm": 0.0007613091729581356,
-      "learning_rate": 0.0001999997285761517,
-      "loss": 46.0,
-      "step": 9710
-    },
-    {
-      "epoch": 0.7424737656975744,
-      "grad_norm": 0.0007229610346257687,
-      "learning_rate": 0.00019999972852018517,
-      "loss": 46.0,
-      "step": 9711
-    },
-    {
-      "epoch": 0.7425502226809642,
-      "grad_norm": 0.0011592023074626923,
-      "learning_rate": 0.00019999972846421283,
-      "loss": 46.0,
-      "step": 9712
-    },
-    {
-      "epoch": 0.7426266796643538,
-      "grad_norm": 0.0009097720612771809,
-      "learning_rate": 0.00019999972840823476,
-      "loss": 46.0,
-      "step": 9713
-    },
-    {
-      "epoch": 0.7427031366477436,
-      "grad_norm": 0.0038393689319491386,
-      "learning_rate": 0.0001999997283522509,
-      "loss": 46.0,
-      "step": 9714
-    },
-    {
-      "epoch": 0.7427795936311333,
-      "grad_norm": 0.003027722705155611,
-      "learning_rate": 0.0001999997282962613,
-      "loss": 46.0,
-      "step": 9715
-    },
-    {
-      "epoch": 0.742856050614523,
-      "grad_norm": 0.0016874130815267563,
-      "learning_rate": 0.0001999997282402659,
-      "loss": 46.0,
-      "step": 9716
-    },
-    {
-      "epoch": 0.7429325075979127,
-      "grad_norm": 0.0004744026518892497,
-      "learning_rate": 0.00019999972818426475,
-      "loss": 46.0,
-      "step": 9717
-    },
-    {
-      "epoch": 0.7430089645813025,
-      "grad_norm": 0.000717241142410785,
-      "learning_rate": 0.00019999972812825782,
-      "loss": 46.0,
-      "step": 9718
-    },
-    {
-      "epoch": 0.7430854215646922,
-      "grad_norm": 0.001474284683354199,
-      "learning_rate": 0.00019999972807224512,
-      "loss": 46.0,
-      "step": 9719
-    },
-    {
-      "epoch": 0.7431618785480819,
-      "grad_norm": 0.008101926185190678,
-      "learning_rate": 0.00019999972801622667,
-      "loss": 46.0,
-      "step": 9720
-    },
-    {
-      "epoch": 0.7432383355314716,
-      "grad_norm": 0.0013627730077132583,
-      "learning_rate": 0.00019999972796020242,
-      "loss": 46.0,
-      "step": 9721
-    },
-    {
-      "epoch": 0.7433147925148613,
-      "grad_norm": 0.001118904328905046,
-      "learning_rate": 0.00019999972790417242,
-      "loss": 46.0,
-      "step": 9722
-    },
-    {
-      "epoch": 0.7433912494982511,
-      "grad_norm": 0.003855801885947585,
-      "learning_rate": 0.00019999972784813665,
-      "loss": 46.0,
-      "step": 9723
-    },
-    {
-      "epoch": 0.7434677064816407,
-      "grad_norm": 0.0007810776005499065,
-      "learning_rate": 0.00019999972779209514,
-      "loss": 46.0,
-      "step": 9724
-    },
-    {
-      "epoch": 0.7435441634650305,
-      "grad_norm": 0.004293626639991999,
-      "learning_rate": 0.0001999997277360478,
-      "loss": 46.0,
-      "step": 9725
-    },
-    {
-      "epoch": 0.7436206204484203,
-      "grad_norm": 0.004228371661156416,
-      "learning_rate": 0.0001999997276799947,
-      "loss": 46.0,
-      "step": 9726
-    },
-    {
-      "epoch": 0.7436970774318099,
-      "grad_norm": 0.0007387892110273242,
-      "learning_rate": 0.00019999972762393587,
-      "loss": 46.0,
-      "step": 9727
-    },
-    {
-      "epoch": 0.7437735344151997,
-      "grad_norm": 0.006005508359521627,
-      "learning_rate": 0.00019999972756787123,
-      "loss": 46.0,
-      "step": 9728
-    },
-    {
-      "epoch": 0.7438499913985893,
-      "grad_norm": 0.00045582870370708406,
-      "learning_rate": 0.00019999972751180085,
-      "loss": 46.0,
-      "step": 9729
-    },
-    {
-      "epoch": 0.7439264483819791,
-      "grad_norm": 0.0075628506019711494,
-      "learning_rate": 0.0001999997274557247,
-      "loss": 46.0,
-      "step": 9730
-    },
-    {
-      "epoch": 0.7440029053653688,
-      "grad_norm": 0.0009257580968551338,
-      "learning_rate": 0.0001999997273996428,
-      "loss": 46.0,
-      "step": 9731
-    },
-    {
-      "epoch": 0.7440793623487585,
-      "grad_norm": 0.003533395240083337,
-      "learning_rate": 0.0001999997273435551,
-      "loss": 46.0,
-      "step": 9732
-    },
-    {
-      "epoch": 0.7441558193321482,
-      "grad_norm": 0.0004218671820126474,
-      "learning_rate": 0.0001999997272874616,
-      "loss": 46.0,
-      "step": 9733
-    },
-    {
-      "epoch": 0.744232276315538,
-      "grad_norm": 0.0024166631046682596,
-      "learning_rate": 0.0001999997272313624,
-      "loss": 46.0,
-      "step": 9734
-    },
-    {
-      "epoch": 0.7443087332989277,
-      "grad_norm": 0.0003856563998851925,
-      "learning_rate": 0.00019999972717525737,
-      "loss": 46.0,
-      "step": 9735
-    },
-    {
-      "epoch": 0.7443851902823174,
-      "grad_norm": 0.0014682149048894644,
-      "learning_rate": 0.0001999997271191466,
-      "loss": 46.0,
-      "step": 9736
-    },
-    {
-      "epoch": 0.7444616472657072,
-      "grad_norm": 0.0009955930290743709,
-      "learning_rate": 0.00019999972706303006,
-      "loss": 46.0,
-      "step": 9737
-    },
-    {
-      "epoch": 0.7445381042490968,
-      "grad_norm": 0.005137870088219643,
-      "learning_rate": 0.00019999972700690774,
-      "loss": 46.0,
-      "step": 9738
-    },
-    {
-      "epoch": 0.7446145612324866,
-      "grad_norm": 0.002106881234794855,
-      "learning_rate": 0.00019999972695077965,
-      "loss": 46.0,
-      "step": 9739
-    },
-    {
-      "epoch": 0.7446910182158762,
-      "grad_norm": 0.01727037876844406,
-      "learning_rate": 0.0001999997268946458,
-      "loss": 46.0,
-      "step": 9740
-    },
-    {
-      "epoch": 0.744767475199266,
-      "grad_norm": 0.0017578959232196212,
-      "learning_rate": 0.00019999972683850618,
-      "loss": 46.0,
-      "step": 9741
-    },
-    {
-      "epoch": 0.7448439321826558,
-      "grad_norm": 0.00151543109677732,
-      "learning_rate": 0.0001999997267823608,
-      "loss": 46.0,
-      "step": 9742
-    },
-    {
-      "epoch": 0.7449203891660454,
-      "grad_norm": 0.0031912578269839287,
-      "learning_rate": 0.0001999997267262096,
-      "loss": 46.0,
-      "step": 9743
-    },
-    {
-      "epoch": 0.7449968461494352,
-      "grad_norm": 0.002707613632082939,
-      "learning_rate": 0.00019999972667005268,
-      "loss": 46.0,
-      "step": 9744
-    },
-    {
-      "epoch": 0.7450733031328249,
-      "grad_norm": 0.001352560007944703,
-      "learning_rate": 0.00019999972661388998,
-      "loss": 46.0,
-      "step": 9745
-    },
-    {
-      "epoch": 0.7451497601162146,
-      "grad_norm": 0.0010036175372079015,
-      "learning_rate": 0.0001999997265577215,
-      "loss": 46.0,
-      "step": 9746
-    },
-    {
-      "epoch": 0.7452262170996043,
-      "grad_norm": 0.00435328995808959,
-      "learning_rate": 0.00019999972650154728,
-      "loss": 46.0,
-      "step": 9747
-    },
-    {
-      "epoch": 0.7453026740829941,
-      "grad_norm": 0.0018448621267452836,
-      "learning_rate": 0.00019999972644536726,
-      "loss": 46.0,
-      "step": 9748
-    },
-    {
-      "epoch": 0.7453791310663838,
-      "grad_norm": 0.0012686033733189106,
-      "learning_rate": 0.00019999972638918146,
-      "loss": 46.0,
-      "step": 9749
-    },
-    {
-      "epoch": 0.7454555880497735,
-      "grad_norm": 0.0008322701323777437,
-      "learning_rate": 0.00019999972633298992,
-      "loss": 46.0,
-      "step": 9750
-    },
-    {
-      "epoch": 0.7455320450331632,
-      "grad_norm": 0.0006211692816577852,
-      "learning_rate": 0.00019999972627679263,
-      "loss": 46.0,
-      "step": 9751
-    },
-    {
-      "epoch": 0.7456085020165529,
-      "grad_norm": 0.0003974427527282387,
-      "learning_rate": 0.00019999972622058954,
-      "loss": 46.0,
-      "step": 9752
-    },
-    {
-      "epoch": 0.7456849589999427,
-      "grad_norm": 0.0022102068178355694,
-      "learning_rate": 0.00019999972616438068,
-      "loss": 46.0,
-      "step": 9753
-    },
-    {
-      "epoch": 0.7457614159833323,
-      "grad_norm": 0.0013326670741662383,
-      "learning_rate": 0.00019999972610816607,
-      "loss": 46.0,
-      "step": 9754
-    },
-    {
-      "epoch": 0.7458378729667221,
-      "grad_norm": 0.001092038583010435,
-      "learning_rate": 0.0001999997260519457,
-      "loss": 46.0,
-      "step": 9755
-    },
-    {
-      "epoch": 0.7459143299501119,
-      "grad_norm": 0.0013066742103546858,
-      "learning_rate": 0.0001999997259957195,
-      "loss": 46.0,
-      "step": 9756
-    },
-    {
-      "epoch": 0.7459907869335015,
-      "grad_norm": 0.002295650076121092,
-      "learning_rate": 0.00019999972593948758,
-      "loss": 46.0,
-      "step": 9757
-    },
-    {
-      "epoch": 0.7460672439168913,
-      "grad_norm": 0.00035254412796348333,
-      "learning_rate": 0.00019999972588324988,
-      "loss": 46.0,
-      "step": 9758
-    },
-    {
-      "epoch": 0.7461437009002809,
-      "grad_norm": 0.002182136755436659,
-      "learning_rate": 0.0001999997258270064,
-      "loss": 46.0,
-      "step": 9759
-    },
-    {
-      "epoch": 0.7462201578836707,
-      "grad_norm": 0.0008121103746816516,
-      "learning_rate": 0.00019999972577075715,
-      "loss": 46.0,
-      "step": 9760
-    },
-    {
-      "epoch": 0.7462966148670604,
-      "grad_norm": 0.0012518854346126318,
-      "learning_rate": 0.00019999972571450213,
-      "loss": 46.0,
-      "step": 9761
-    },
-    {
-      "epoch": 0.7463730718504501,
-      "grad_norm": 0.0006228789570741355,
-      "learning_rate": 0.00019999972565824136,
-      "loss": 46.0,
-      "step": 9762
-    },
-    {
-      "epoch": 0.7464495288338399,
-      "grad_norm": 0.001432525459676981,
-      "learning_rate": 0.00019999972560197482,
-      "loss": 46.0,
-      "step": 9763
-    },
-    {
-      "epoch": 0.7465259858172296,
-      "grad_norm": 0.0011126719182357192,
-      "learning_rate": 0.0001999997255457025,
-      "loss": 46.0,
-      "step": 9764
-    },
-    {
-      "epoch": 0.7466024428006193,
-      "grad_norm": 0.002074331045150757,
-      "learning_rate": 0.00019999972548942442,
-      "loss": 46.0,
-      "step": 9765
-    },
-    {
-      "epoch": 0.746678899784009,
-      "grad_norm": 0.002931632800027728,
-      "learning_rate": 0.00019999972543314053,
-      "loss": 46.0,
-      "step": 9766
-    },
-    {
-      "epoch": 0.7467553567673988,
-      "grad_norm": 0.012228799052536488,
-      "learning_rate": 0.0001999997253768509,
-      "loss": 46.0,
-      "step": 9767
-    },
-    {
-      "epoch": 0.7468318137507884,
-      "grad_norm": 0.0009336150833405554,
-      "learning_rate": 0.0001999997253205555,
-      "loss": 46.0,
-      "step": 9768
-    },
-    {
-      "epoch": 0.7469082707341782,
-      "grad_norm": 0.00102690898347646,
-      "learning_rate": 0.00019999972526425435,
-      "loss": 46.0,
-      "step": 9769
-    },
-    {
-      "epoch": 0.7469847277175679,
-      "grad_norm": 0.0003415764949750155,
-      "learning_rate": 0.0001999997252079474,
-      "loss": 46.0,
-      "step": 9770
-    },
-    {
-      "epoch": 0.7470611847009576,
-      "grad_norm": 0.002098718425258994,
-      "learning_rate": 0.0001999997251516347,
-      "loss": 46.0,
-      "step": 9771
-    },
-    {
-      "epoch": 0.7471376416843474,
-      "grad_norm": 0.00022140626970212907,
-      "learning_rate": 0.00019999972509531625,
-      "loss": 46.0,
-      "step": 9772
-    },
-    {
-      "epoch": 0.747214098667737,
-      "grad_norm": 0.0008160950383171439,
-      "learning_rate": 0.00019999972503899197,
-      "loss": 46.0,
-      "step": 9773
-    },
-    {
-      "epoch": 0.7472905556511268,
-      "grad_norm": 0.0030269743874669075,
-      "learning_rate": 0.00019999972498266198,
-      "loss": 46.0,
-      "step": 9774
-    },
-    {
-      "epoch": 0.7473670126345165,
-      "grad_norm": 0.0037928048986941576,
-      "learning_rate": 0.00019999972492632618,
-      "loss": 46.0,
-      "step": 9775
-    },
-    {
-      "epoch": 0.7474434696179062,
-      "grad_norm": 0.001923106610774994,
-      "learning_rate": 0.00019999972486998461,
-      "loss": 46.0,
-      "step": 9776
-    },
-    {
-      "epoch": 0.747519926601296,
-      "grad_norm": 0.0006949757807888091,
-      "learning_rate": 0.00019999972481363733,
-      "loss": 46.0,
-      "step": 9777
-    },
-    {
-      "epoch": 0.7475963835846857,
-      "grad_norm": 0.0025377334095537663,
-      "learning_rate": 0.00019999972475728424,
-      "loss": 46.0,
-      "step": 9778
-    },
-    {
-      "epoch": 0.7476728405680754,
-      "grad_norm": 0.001489555579610169,
-      "learning_rate": 0.00019999972470092535,
-      "loss": 46.0,
-      "step": 9779
-    },
-    {
-      "epoch": 0.7477492975514651,
-      "grad_norm": 0.0019057902973145247,
-      "learning_rate": 0.0001999997246445607,
-      "loss": 46.0,
-      "step": 9780
-    },
-    {
-      "epoch": 0.7478257545348548,
-      "grad_norm": 0.0005225634085945785,
-      "learning_rate": 0.00019999972458819033,
-      "loss": 46.0,
-      "step": 9781
-    },
-    {
-      "epoch": 0.7479022115182445,
-      "grad_norm": 0.0012569152750074863,
-      "learning_rate": 0.00019999972453181415,
-      "loss": 46.0,
-      "step": 9782
-    },
-    {
-      "epoch": 0.7479786685016343,
-      "grad_norm": 0.0033967429772019386,
-      "learning_rate": 0.00019999972447543222,
-      "loss": 46.0,
-      "step": 9783
-    },
-    {
-      "epoch": 0.748055125485024,
-      "grad_norm": 0.0016485553933307528,
-      "learning_rate": 0.00019999972441904452,
-      "loss": 46.0,
-      "step": 9784
-    },
-    {
-      "epoch": 0.7481315824684137,
-      "grad_norm": 0.0008147096959874034,
-      "learning_rate": 0.00019999972436265102,
-      "loss": 46.0,
-      "step": 9785
-    },
-    {
-      "epoch": 0.7482080394518035,
-      "grad_norm": 0.0026291979011148214,
-      "learning_rate": 0.00019999972430625177,
-      "loss": 46.0,
-      "step": 9786
-    },
-    {
-      "epoch": 0.7482844964351931,
-      "grad_norm": 0.0017728663515299559,
-      "learning_rate": 0.00019999972424984678,
-      "loss": 46.0,
-      "step": 9787
-    },
-    {
-      "epoch": 0.7483609534185829,
-      "grad_norm": 0.0016757167177274823,
-      "learning_rate": 0.00019999972419343598,
-      "loss": 46.0,
-      "step": 9788
-    },
-    {
-      "epoch": 0.7484374104019725,
-      "grad_norm": 0.0026348098181188107,
-      "learning_rate": 0.0001999997241370194,
-      "loss": 46.0,
-      "step": 9789
-    },
-    {
-      "epoch": 0.7485138673853623,
-      "grad_norm": 0.025955254212021828,
-      "learning_rate": 0.0001999997240805971,
-      "loss": 46.0,
-      "step": 9790
-    },
-    {
-      "epoch": 0.7485903243687521,
-      "grad_norm": 0.0026988142635673285,
-      "learning_rate": 0.000199999724024169,
-      "loss": 46.0,
-      "step": 9791
-    },
-    {
-      "epoch": 0.7486667813521417,
-      "grad_norm": 0.0017523474525660276,
-      "learning_rate": 0.00019999972396773515,
-      "loss": 46.0,
-      "step": 9792
-    },
-    {
-      "epoch": 0.7487432383355315,
-      "grad_norm": 0.0005799881764687598,
-      "learning_rate": 0.0001999997239112955,
-      "loss": 46.0,
-      "step": 9793
-    },
-    {
-      "epoch": 0.7488196953189212,
-      "grad_norm": 0.0014589010970667005,
-      "learning_rate": 0.0001999997238548501,
-      "loss": 46.0,
-      "step": 9794
-    },
-    {
-      "epoch": 0.7488961523023109,
-      "grad_norm": 0.0022163703106343746,
-      "learning_rate": 0.00019999972379839892,
-      "loss": 46.0,
-      "step": 9795
-    },
-    {
-      "epoch": 0.7489726092857006,
-      "grad_norm": 0.001036088913679123,
-      "learning_rate": 0.00019999972374194197,
-      "loss": 46.0,
-      "step": 9796
-    },
-    {
-      "epoch": 0.7490490662690904,
-      "grad_norm": 0.001170091680251062,
-      "learning_rate": 0.00019999972368547927,
-      "loss": 46.0,
-      "step": 9797
-    },
-    {
-      "epoch": 0.74912552325248,
-      "grad_norm": 0.006794528104364872,
-      "learning_rate": 0.0001999997236290108,
-      "loss": 46.0,
-      "step": 9798
-    },
-    {
-      "epoch": 0.7492019802358698,
-      "grad_norm": 0.002051667310297489,
-      "learning_rate": 0.00019999972357253654,
-      "loss": 46.0,
-      "step": 9799
-    },
-    {
-      "epoch": 0.7492784372192595,
-      "grad_norm": 0.000676435069181025,
-      "learning_rate": 0.00019999972351605655,
-      "loss": 46.0,
-      "step": 9800
-    },
-    {
-      "epoch": 0.7493548942026492,
-      "grad_norm": 0.00033562039607204497,
-      "learning_rate": 0.00019999972345957075,
-      "loss": 46.0,
-      "step": 9801
-    },
-    {
-      "epoch": 0.749431351186039,
-      "grad_norm": 0.0004428086685948074,
-      "learning_rate": 0.0001999997234030792,
-      "loss": 46.0,
-      "step": 9802
-    },
-    {
-      "epoch": 0.7495078081694286,
-      "grad_norm": 0.001040892326273024,
-      "learning_rate": 0.00019999972334658184,
-      "loss": 46.0,
-      "step": 9803
-    },
-    {
-      "epoch": 0.7495842651528184,
-      "grad_norm": 0.0007352923857979476,
-      "learning_rate": 0.00019999972329007876,
-      "loss": 46.0,
-      "step": 9804
-    },
-    {
-      "epoch": 0.7496607221362082,
-      "grad_norm": 0.0024205262307077646,
-      "learning_rate": 0.00019999972323356987,
-      "loss": 46.0,
-      "step": 9805
-    },
-    {
-      "epoch": 0.7497371791195978,
-      "grad_norm": 0.0021224578376859426,
-      "learning_rate": 0.00019999972317705526,
-      "loss": 46.0,
-      "step": 9806
-    },
-    {
-      "epoch": 0.7498136361029876,
-      "grad_norm": 0.0049773999489843845,
-      "learning_rate": 0.00019999972312053485,
-      "loss": 46.0,
-      "step": 9807
-    },
-    {
-      "epoch": 0.7498900930863773,
-      "grad_norm": 0.0021834401413798332,
-      "learning_rate": 0.00019999972306400867,
-      "loss": 46.0,
-      "step": 9808
-    },
-    {
-      "epoch": 0.749966550069767,
-      "grad_norm": 0.01164508331567049,
-      "learning_rate": 0.00019999972300747674,
-      "loss": 46.0,
-      "step": 9809
-    },
-    {
-      "epoch": 0.7500430070531567,
-      "grad_norm": 0.004073922988027334,
-      "learning_rate": 0.00019999972295093902,
-      "loss": 46.0,
-      "step": 9810
-    },
-    {
-      "epoch": 0.7501194640365464,
-      "grad_norm": 0.0008712432463653386,
-      "learning_rate": 0.00019999972289439554,
-      "loss": 46.0,
-      "step": 9811
-    },
-    {
-      "epoch": 0.7501959210199362,
-      "grad_norm": 0.0016555271577090025,
-      "learning_rate": 0.0001999997228378463,
-      "loss": 46.0,
-      "step": 9812
-    },
-    {
-      "epoch": 0.7502723780033259,
-      "grad_norm": 0.0003782497951760888,
-      "learning_rate": 0.00019999972278129127,
-      "loss": 46.0,
-      "step": 9813
-    },
-    {
-      "epoch": 0.7503488349867156,
-      "grad_norm": 0.0021239907946437597,
-      "learning_rate": 0.00019999972272473045,
-      "loss": 46.0,
-      "step": 9814
-    },
-    {
-      "epoch": 0.7504252919701053,
-      "grad_norm": 0.0009066574275493622,
-      "learning_rate": 0.00019999972266816388,
-      "loss": 46.0,
-      "step": 9815
-    },
-    {
-      "epoch": 0.7505017489534951,
-      "grad_norm": 0.0007880771881900728,
-      "learning_rate": 0.0001999997226115916,
-      "loss": 46.0,
-      "step": 9816
-    },
-    {
-      "epoch": 0.7505782059368847,
-      "grad_norm": 0.000987007631920278,
-      "learning_rate": 0.00019999972255501348,
-      "loss": 46.0,
-      "step": 9817
-    },
-    {
-      "epoch": 0.7506546629202745,
-      "grad_norm": 0.0022335790563374758,
-      "learning_rate": 0.00019999972249842962,
-      "loss": 46.0,
-      "step": 9818
-    },
-    {
-      "epoch": 0.7507311199036641,
-      "grad_norm": 0.0009881261503323913,
-      "learning_rate": 0.00019999972244183996,
-      "loss": 46.0,
-      "step": 9819
-    },
-    {
-      "epoch": 0.7508075768870539,
-      "grad_norm": 0.0028058013413101435,
-      "learning_rate": 0.00019999972238524455,
-      "loss": 46.0,
-      "step": 9820
-    },
-    {
-      "epoch": 0.7508840338704437,
-      "grad_norm": 0.0009051478118635714,
-      "learning_rate": 0.0001999997223286434,
-      "loss": 46.0,
-      "step": 9821
-    },
-    {
-      "epoch": 0.7509604908538333,
-      "grad_norm": 0.0011644193436950445,
-      "learning_rate": 0.00019999972227203644,
-      "loss": 46.0,
-      "step": 9822
-    },
-    {
-      "epoch": 0.7510369478372231,
-      "grad_norm": 0.003027724800631404,
-      "learning_rate": 0.00019999972221542372,
-      "loss": 46.0,
-      "step": 9823
-    },
-    {
-      "epoch": 0.7511134048206128,
-      "grad_norm": 0.0034554265439510345,
-      "learning_rate": 0.00019999972215880524,
-      "loss": 46.0,
-      "step": 9824
-    },
-    {
-      "epoch": 0.7511898618040025,
-      "grad_norm": 0.0034576591569930315,
-      "learning_rate": 0.000199999722102181,
-      "loss": 46.0,
-      "step": 9825
-    },
-    {
-      "epoch": 0.7512663187873923,
-      "grad_norm": 0.0009959448361769319,
-      "learning_rate": 0.00019999972204555095,
-      "loss": 46.0,
-      "step": 9826
-    },
-    {
-      "epoch": 0.751342775770782,
-      "grad_norm": 0.0002472540072631091,
-      "learning_rate": 0.00019999972198891515,
-      "loss": 46.0,
-      "step": 9827
-    },
-    {
-      "epoch": 0.7514192327541717,
-      "grad_norm": 0.0032396509777754545,
-      "learning_rate": 0.0001999997219322736,
-      "loss": 46.0,
-      "step": 9828
-    },
-    {
-      "epoch": 0.7514956897375614,
-      "grad_norm": 0.0009097022120840847,
-      "learning_rate": 0.00019999972187562624,
-      "loss": 46.0,
-      "step": 9829
-    },
-    {
-      "epoch": 0.7515721467209511,
-      "grad_norm": 0.0012322955299168825,
-      "learning_rate": 0.00019999972181897318,
-      "loss": 46.0,
-      "step": 9830
-    },
-    {
-      "epoch": 0.7516486037043408,
-      "grad_norm": 0.001780449179932475,
-      "learning_rate": 0.00019999972176231427,
-      "loss": 46.0,
-      "step": 9831
-    },
-    {
-      "epoch": 0.7517250606877306,
-      "grad_norm": 0.0020478214137256145,
-      "learning_rate": 0.00019999972170564964,
-      "loss": 46.0,
-      "step": 9832
-    },
-    {
-      "epoch": 0.7518015176711202,
-      "grad_norm": 0.0005412310129031539,
-      "learning_rate": 0.00019999972164897926,
-      "loss": 46.0,
-      "step": 9833
-    },
-    {
-      "epoch": 0.75187797465451,
-      "grad_norm": 0.0031512987334281206,
-      "learning_rate": 0.00019999972159230308,
-      "loss": 46.0,
-      "step": 9834
-    },
-    {
-      "epoch": 0.7519544316378998,
-      "grad_norm": 0.0008630980155430734,
-      "learning_rate": 0.0001999997215356211,
-      "loss": 46.0,
-      "step": 9835
-    },
-    {
-      "epoch": 0.7520308886212894,
-      "grad_norm": 0.0009712826577015221,
-      "learning_rate": 0.0001999997214789334,
-      "loss": 46.0,
-      "step": 9836
-    },
-    {
-      "epoch": 0.7521073456046792,
-      "grad_norm": 0.0009332008776254952,
-      "learning_rate": 0.0001999997214222399,
-      "loss": 46.0,
-      "step": 9837
-    },
-    {
-      "epoch": 0.7521838025880689,
-      "grad_norm": 0.0010059881024062634,
-      "learning_rate": 0.00019999972136554065,
-      "loss": 46.0,
-      "step": 9838
-    },
-    {
-      "epoch": 0.7522602595714586,
-      "grad_norm": 0.0036950900685042143,
-      "learning_rate": 0.0001999997213088356,
-      "loss": 46.0,
-      "step": 9839
-    },
-    {
-      "epoch": 0.7523367165548484,
-      "grad_norm": 0.0007250310736708343,
-      "learning_rate": 0.00019999972125212483,
-      "loss": 46.0,
-      "step": 9840
-    },
-    {
-      "epoch": 0.752413173538238,
-      "grad_norm": 0.004591570235788822,
-      "learning_rate": 0.00019999972119540824,
-      "loss": 46.0,
-      "step": 9841
-    },
-    {
-      "epoch": 0.7524896305216278,
-      "grad_norm": 0.0012448506895452738,
-      "learning_rate": 0.00019999972113868592,
-      "loss": 46.0,
-      "step": 9842
-    },
-    {
-      "epoch": 0.7525660875050175,
-      "grad_norm": 0.00035542011028155684,
-      "learning_rate": 0.00019999972108195778,
-      "loss": 46.0,
-      "step": 9843
-    },
-    {
-      "epoch": 0.7526425444884072,
-      "grad_norm": 0.0013621420366689563,
-      "learning_rate": 0.00019999972102522392,
-      "loss": 46.0,
-      "step": 9844
-    },
-    {
-      "epoch": 0.7527190014717969,
-      "grad_norm": 0.0008436488569714129,
-      "learning_rate": 0.0001999997209684843,
-      "loss": 46.0,
-      "step": 9845
-    },
-    {
-      "epoch": 0.7527954584551867,
-      "grad_norm": 0.0010542816016823053,
-      "learning_rate": 0.00019999972091173888,
-      "loss": 46.0,
-      "step": 9846
-    },
-    {
-      "epoch": 0.7528719154385763,
-      "grad_norm": 0.006132427137345076,
-      "learning_rate": 0.0001999997208549877,
-      "loss": 46.0,
-      "step": 9847
-    },
-    {
-      "epoch": 0.7529483724219661,
-      "grad_norm": 0.0012286645360291004,
-      "learning_rate": 0.00019999972079823075,
-      "loss": 46.0,
-      "step": 9848
-    },
-    {
-      "epoch": 0.7530248294053559,
-      "grad_norm": 0.000861355394590646,
-      "learning_rate": 0.00019999972074146802,
-      "loss": 46.0,
-      "step": 9849
-    },
-    {
-      "epoch": 0.7531012863887455,
-      "grad_norm": 0.0007934892200864851,
-      "learning_rate": 0.00019999972068469952,
-      "loss": 46.0,
-      "step": 9850
-    },
-    {
-      "epoch": 0.7531777433721353,
-      "grad_norm": 0.003642365802079439,
-      "learning_rate": 0.00019999972062792525,
-      "loss": 46.0,
-      "step": 9851
-    },
-    {
-      "epoch": 0.7532542003555249,
-      "grad_norm": 0.0037795936223119497,
-      "learning_rate": 0.00019999972057114525,
-      "loss": 46.0,
-      "step": 9852
-    },
-    {
-      "epoch": 0.7533306573389147,
-      "grad_norm": 0.0013161622919142246,
-      "learning_rate": 0.00019999972051435943,
-      "loss": 46.0,
-      "step": 9853
-    },
-    {
-      "epoch": 0.7534071143223044,
-      "grad_norm": 0.0008625526097603142,
-      "learning_rate": 0.00019999972045756787,
-      "loss": 46.0,
-      "step": 9854
-    },
-    {
-      "epoch": 0.7534835713056941,
-      "grad_norm": 0.0017944211140275002,
-      "learning_rate": 0.00019999972040077053,
-      "loss": 46.0,
-      "step": 9855
-    },
-    {
-      "epoch": 0.7535600282890839,
-      "grad_norm": 0.0017904648557305336,
-      "learning_rate": 0.0001999997203439674,
-      "loss": 46.0,
-      "step": 9856
-    },
-    {
-      "epoch": 0.7536364852724736,
-      "grad_norm": 0.0009710281156003475,
-      "learning_rate": 0.00019999972028715855,
-      "loss": 46.0,
-      "step": 9857
-    },
-    {
-      "epoch": 0.7537129422558633,
-      "grad_norm": 0.003670301754027605,
-      "learning_rate": 0.0001999997202303439,
-      "loss": 46.0,
-      "step": 9858
-    },
-    {
-      "epoch": 0.753789399239253,
-      "grad_norm": 0.001092874095775187,
-      "learning_rate": 0.00019999972017352348,
-      "loss": 46.0,
-      "step": 9859
-    },
-    {
-      "epoch": 0.7538658562226427,
-      "grad_norm": 0.0006195945898070931,
-      "learning_rate": 0.0001999997201166973,
-      "loss": 46.0,
-      "step": 9860
-    },
-    {
-      "epoch": 0.7539423132060324,
-      "grad_norm": 0.0008521394338458776,
-      "learning_rate": 0.00019999972005986532,
-      "loss": 46.0,
-      "step": 9861
-    },
-    {
-      "epoch": 0.7540187701894222,
-      "grad_norm": 0.0012158494209870696,
-      "learning_rate": 0.0001999997200030276,
-      "loss": 46.0,
-      "step": 9862
-    },
-    {
-      "epoch": 0.7540952271728119,
-      "grad_norm": 0.0013173875631764531,
-      "learning_rate": 0.00019999971994618412,
-      "loss": 46.0,
-      "step": 9863
-    },
-    {
-      "epoch": 0.7541716841562016,
-      "grad_norm": 0.001688745222054422,
-      "learning_rate": 0.00019999971988933485,
-      "loss": 46.0,
-      "step": 9864
-    },
-    {
-      "epoch": 0.7542481411395914,
-      "grad_norm": 0.004590890370309353,
-      "learning_rate": 0.0001999997198324798,
-      "loss": 46.0,
-      "step": 9865
-    },
-    {
-      "epoch": 0.754324598122981,
-      "grad_norm": 0.0009763687266968191,
-      "learning_rate": 0.000199999719775619,
-      "loss": 46.0,
-      "step": 9866
-    },
-    {
-      "epoch": 0.7544010551063708,
-      "grad_norm": 0.006379557307809591,
-      "learning_rate": 0.00019999971971875244,
-      "loss": 46.0,
-      "step": 9867
-    },
-    {
-      "epoch": 0.7544775120897605,
-      "grad_norm": 0.001985793700441718,
-      "learning_rate": 0.0001999997196618801,
-      "loss": 46.0,
-      "step": 9868
-    },
-    {
-      "epoch": 0.7545539690731502,
-      "grad_norm": 0.0010254301596432924,
-      "learning_rate": 0.000199999719605002,
-      "loss": 46.0,
-      "step": 9869
-    },
-    {
-      "epoch": 0.75463042605654,
-      "grad_norm": 0.0014939854154363275,
-      "learning_rate": 0.0001999997195481181,
-      "loss": 46.0,
-      "step": 9870
-    },
-    {
-      "epoch": 0.7547068830399296,
-      "grad_norm": 0.0033801430836319923,
-      "learning_rate": 0.00019999971949122847,
-      "loss": 46.0,
-      "step": 9871
-    },
-    {
-      "epoch": 0.7547833400233194,
-      "grad_norm": 0.0013937511248514056,
-      "learning_rate": 0.00019999971943433304,
-      "loss": 46.0,
-      "step": 9872
-    },
-    {
-      "epoch": 0.7548597970067091,
-      "grad_norm": 0.0010351731907576323,
-      "learning_rate": 0.00019999971937743186,
-      "loss": 46.0,
-      "step": 9873
-    },
-    {
-      "epoch": 0.7549362539900988,
-      "grad_norm": 0.0011348961852490902,
-      "learning_rate": 0.0001999997193205249,
-      "loss": 46.0,
-      "step": 9874
-    },
-    {
-      "epoch": 0.7550127109734885,
-      "grad_norm": 0.0009345394209958613,
-      "learning_rate": 0.00019999971926361218,
-      "loss": 46.0,
-      "step": 9875
-    },
-    {
-      "epoch": 0.7550891679568783,
-      "grad_norm": 0.000685446837451309,
-      "learning_rate": 0.00019999971920669368,
-      "loss": 46.0,
-      "step": 9876
-    },
-    {
-      "epoch": 0.755165624940268,
-      "grad_norm": 0.00556287681683898,
-      "learning_rate": 0.0001999997191497694,
-      "loss": 46.0,
-      "step": 9877
-    },
-    {
-      "epoch": 0.7552420819236577,
-      "grad_norm": 0.00029644460300914943,
-      "learning_rate": 0.00019999971909283936,
-      "loss": 46.0,
-      "step": 9878
-    },
-    {
-      "epoch": 0.7553185389070475,
-      "grad_norm": 0.0007393202395178378,
-      "learning_rate": 0.00019999971903590357,
-      "loss": 46.0,
-      "step": 9879
-    },
-    {
-      "epoch": 0.7553949958904371,
-      "grad_norm": 0.000838323263451457,
-      "learning_rate": 0.000199999718978962,
-      "loss": 46.0,
-      "step": 9880
-    },
-    {
-      "epoch": 0.7554714528738269,
-      "grad_norm": 0.0008017085492610931,
-      "learning_rate": 0.00019999971892201463,
-      "loss": 46.0,
-      "step": 9881
-    },
-    {
-      "epoch": 0.7555479098572165,
-      "grad_norm": 0.0038079828955233097,
-      "learning_rate": 0.00019999971886506152,
-      "loss": 46.0,
-      "step": 9882
-    },
-    {
-      "epoch": 0.7556243668406063,
-      "grad_norm": 0.0025249787140637636,
-      "learning_rate": 0.00019999971880810263,
-      "loss": 46.0,
-      "step": 9883
-    },
-    {
-      "epoch": 0.7557008238239961,
-      "grad_norm": 0.001586354337632656,
-      "learning_rate": 0.000199999718751138,
-      "loss": 46.0,
-      "step": 9884
-    },
-    {
-      "epoch": 0.7557772808073857,
-      "grad_norm": 0.003486678237095475,
-      "learning_rate": 0.0001999997186941676,
-      "loss": 46.0,
-      "step": 9885
-    },
-    {
-      "epoch": 0.7558537377907755,
-      "grad_norm": 0.0005069439648650587,
-      "learning_rate": 0.00019999971863719142,
-      "loss": 46.0,
-      "step": 9886
-    },
-    {
-      "epoch": 0.7559301947741652,
-      "grad_norm": 0.0019650235772132874,
-      "learning_rate": 0.00019999971858020944,
-      "loss": 46.0,
-      "step": 9887
-    },
-    {
-      "epoch": 0.7560066517575549,
-      "grad_norm": 0.0006303850677795708,
-      "learning_rate": 0.0001999997185232217,
-      "loss": 46.0,
-      "step": 9888
-    },
-    {
-      "epoch": 0.7560831087409446,
-      "grad_norm": 0.002826918847858906,
-      "learning_rate": 0.0001999997184662282,
-      "loss": 46.0,
-      "step": 9889
-    },
-    {
-      "epoch": 0.7561595657243343,
-      "grad_norm": 0.0014048358425498009,
-      "learning_rate": 0.00019999971840922894,
-      "loss": 46.0,
-      "step": 9890
-    },
-    {
-      "epoch": 0.756236022707724,
-      "grad_norm": 0.0015641612699255347,
-      "learning_rate": 0.0001999997183522239,
-      "loss": 46.0,
-      "step": 9891
-    },
-    {
-      "epoch": 0.7563124796911138,
-      "grad_norm": 0.001855909125879407,
-      "learning_rate": 0.0001999997182952131,
-      "loss": 46.0,
-      "step": 9892
-    },
-    {
-      "epoch": 0.7563889366745035,
-      "grad_norm": 0.003341649891808629,
-      "learning_rate": 0.00019999971823819653,
-      "loss": 46.0,
-      "step": 9893
-    },
-    {
-      "epoch": 0.7564653936578932,
-      "grad_norm": 0.002157648093998432,
-      "learning_rate": 0.00019999971818117417,
-      "loss": 46.0,
-      "step": 9894
-    },
-    {
-      "epoch": 0.756541850641283,
-      "grad_norm": 0.0012987025547772646,
-      "learning_rate": 0.00019999971812414605,
-      "loss": 46.0,
-      "step": 9895
-    },
-    {
-      "epoch": 0.7566183076246726,
-      "grad_norm": 0.0004675733798649162,
-      "learning_rate": 0.00019999971806711214,
-      "loss": 46.0,
-      "step": 9896
-    },
-    {
-      "epoch": 0.7566947646080624,
-      "grad_norm": 0.0015582921914756298,
-      "learning_rate": 0.0001999997180100725,
-      "loss": 46.0,
-      "step": 9897
-    },
-    {
-      "epoch": 0.7567712215914522,
-      "grad_norm": 0.0011863448889926076,
-      "learning_rate": 0.00019999971795302707,
-      "loss": 46.0,
-      "step": 9898
-    },
-    {
-      "epoch": 0.7568476785748418,
-      "grad_norm": 0.0015999755123630166,
-      "learning_rate": 0.00019999971789597587,
-      "loss": 46.0,
-      "step": 9899
-    },
-    {
-      "epoch": 0.7569241355582316,
-      "grad_norm": 0.0007188166491687298,
-      "learning_rate": 0.00019999971783891894,
-      "loss": 46.0,
-      "step": 9900
-    },
-    {
-      "epoch": 0.7570005925416212,
-      "grad_norm": 0.000480824412079528,
-      "learning_rate": 0.0001999997177818562,
-      "loss": 46.0,
-      "step": 9901
-    },
-    {
-      "epoch": 0.757077049525011,
-      "grad_norm": 0.004491476342082024,
-      "learning_rate": 0.0001999997177247877,
-      "loss": 46.0,
-      "step": 9902
-    },
-    {
-      "epoch": 0.7571535065084007,
-      "grad_norm": 0.0012539024464786053,
-      "learning_rate": 0.00019999971766771342,
-      "loss": 46.0,
-      "step": 9903
-    },
-    {
-      "epoch": 0.7572299634917904,
-      "grad_norm": 0.0012809648178517818,
-      "learning_rate": 0.00019999971761063337,
-      "loss": 46.0,
-      "step": 9904
-    },
-    {
-      "epoch": 0.7573064204751802,
-      "grad_norm": 0.002720392309129238,
-      "learning_rate": 0.00019999971755354755,
-      "loss": 46.0,
-      "step": 9905
-    },
-    {
-      "epoch": 0.7573828774585699,
-      "grad_norm": 0.0015013209776952863,
-      "learning_rate": 0.000199999717496456,
-      "loss": 46.0,
-      "step": 9906
-    },
-    {
-      "epoch": 0.7574593344419596,
-      "grad_norm": 0.0018568446394056082,
-      "learning_rate": 0.00019999971743935865,
-      "loss": 46.0,
-      "step": 9907
-    },
-    {
-      "epoch": 0.7575357914253493,
-      "grad_norm": 0.0007769854855723679,
-      "learning_rate": 0.0001999997173822555,
-      "loss": 46.0,
-      "step": 9908
-    },
-    {
-      "epoch": 0.7576122484087391,
-      "grad_norm": 0.0007741964654996991,
-      "learning_rate": 0.00019999971732514662,
-      "loss": 46.0,
-      "step": 9909
-    },
-    {
-      "epoch": 0.7576887053921287,
-      "grad_norm": 0.0008315082523040473,
-      "learning_rate": 0.00019999971726803196,
-      "loss": 46.0,
-      "step": 9910
-    },
-    {
-      "epoch": 0.7577651623755185,
-      "grad_norm": 0.000988107523880899,
-      "learning_rate": 0.00019999971721091156,
-      "loss": 46.0,
-      "step": 9911
-    },
-    {
-      "epoch": 0.7578416193589081,
-      "grad_norm": 0.004938120488077402,
-      "learning_rate": 0.00019999971715378533,
-      "loss": 46.0,
-      "step": 9912
-    },
-    {
-      "epoch": 0.7579180763422979,
-      "grad_norm": 0.0009589770343154669,
-      "learning_rate": 0.0001999997170966534,
-      "loss": 46.0,
-      "step": 9913
-    },
-    {
-      "epoch": 0.7579945333256877,
-      "grad_norm": 0.0013909179251641035,
-      "learning_rate": 0.00019999971703951562,
-      "loss": 46.0,
-      "step": 9914
-    },
-    {
-      "epoch": 0.7580709903090773,
-      "grad_norm": 0.0011247647926211357,
-      "learning_rate": 0.00019999971698237215,
-      "loss": 46.0,
-      "step": 9915
-    },
-    {
-      "epoch": 0.7581474472924671,
-      "grad_norm": 0.0031595658510923386,
-      "learning_rate": 0.00019999971692522288,
-      "loss": 46.0,
-      "step": 9916
-    },
-    {
-      "epoch": 0.7582239042758568,
-      "grad_norm": 0.0028955473098903894,
-      "learning_rate": 0.00019999971686806783,
-      "loss": 46.0,
-      "step": 9917
-    },
-    {
-      "epoch": 0.7583003612592465,
-      "grad_norm": 0.004822517279535532,
-      "learning_rate": 0.000199999716810907,
-      "loss": 46.0,
-      "step": 9918
-    },
-    {
-      "epoch": 0.7583768182426363,
-      "grad_norm": 0.0007866416126489639,
-      "learning_rate": 0.00019999971675374042,
-      "loss": 46.0,
-      "step": 9919
-    },
-    {
-      "epoch": 0.7584532752260259,
-      "grad_norm": 0.0013370139058679342,
-      "learning_rate": 0.00019999971669656808,
-      "loss": 46.0,
-      "step": 9920
-    },
-    {
-      "epoch": 0.7585297322094157,
-      "grad_norm": 0.005823432933539152,
-      "learning_rate": 0.00019999971663938997,
-      "loss": 46.0,
-      "step": 9921
-    },
-    {
-      "epoch": 0.7586061891928054,
-      "grad_norm": 0.0009634061716496944,
-      "learning_rate": 0.00019999971658220606,
-      "loss": 46.0,
-      "step": 9922
-    },
-    {
-      "epoch": 0.7586826461761951,
-      "grad_norm": 0.0007240860722959042,
-      "learning_rate": 0.0001999997165250164,
-      "loss": 46.0,
-      "step": 9923
-    },
-    {
-      "epoch": 0.7587591031595848,
-      "grad_norm": 0.0006135416333563626,
-      "learning_rate": 0.000199999716467821,
-      "loss": 46.0,
-      "step": 9924
-    },
-    {
-      "epoch": 0.7588355601429746,
-      "grad_norm": 0.00035883820964954793,
-      "learning_rate": 0.00019999971641061976,
-      "loss": 46.0,
-      "step": 9925
-    },
-    {
-      "epoch": 0.7589120171263642,
-      "grad_norm": 0.0023508856538683176,
-      "learning_rate": 0.0001999997163534128,
-      "loss": 46.0,
-      "step": 9926
-    },
-    {
-      "epoch": 0.758988474109754,
-      "grad_norm": 0.003199746599420905,
-      "learning_rate": 0.00019999971629620006,
-      "loss": 46.0,
-      "step": 9927
-    },
-    {
-      "epoch": 0.7590649310931438,
-      "grad_norm": 0.0014498098753392696,
-      "learning_rate": 0.00019999971623898156,
-      "loss": 46.0,
-      "step": 9928
-    },
-    {
-      "epoch": 0.7591413880765334,
-      "grad_norm": 0.0010453228605911136,
-      "learning_rate": 0.00019999971618175728,
-      "loss": 46.0,
-      "step": 9929
-    },
-    {
-      "epoch": 0.7592178450599232,
-      "grad_norm": 0.0013948811683803797,
-      "learning_rate": 0.00019999971612452724,
-      "loss": 46.0,
-      "step": 9930
-    },
-    {
-      "epoch": 0.7592943020433128,
-      "grad_norm": 0.002305847592651844,
-      "learning_rate": 0.00019999971606729142,
-      "loss": 46.0,
-      "step": 9931
-    },
-    {
-      "epoch": 0.7593707590267026,
-      "grad_norm": 0.0024088099598884583,
-      "learning_rate": 0.00019999971601004983,
-      "loss": 46.0,
-      "step": 9932
-    },
-    {
-      "epoch": 0.7594472160100924,
-      "grad_norm": 0.0008411537273786962,
-      "learning_rate": 0.00019999971595280246,
-      "loss": 46.0,
-      "step": 9933
-    },
-    {
-      "epoch": 0.759523672993482,
-      "grad_norm": 0.0027810793835669756,
-      "learning_rate": 0.00019999971589554935,
-      "loss": 46.0,
-      "step": 9934
-    },
-    {
-      "epoch": 0.7596001299768718,
-      "grad_norm": 0.0029860492795705795,
-      "learning_rate": 0.00019999971583829047,
-      "loss": 46.0,
-      "step": 9935
-    },
-    {
-      "epoch": 0.7596765869602615,
-      "grad_norm": 0.0019378914730623364,
-      "learning_rate": 0.00019999971578102578,
-      "loss": 46.0,
-      "step": 9936
-    },
-    {
-      "epoch": 0.7597530439436512,
-      "grad_norm": 0.0011977998074144125,
-      "learning_rate": 0.00019999971572375538,
-      "loss": 46.0,
-      "step": 9937
-    },
-    {
-      "epoch": 0.7598295009270409,
-      "grad_norm": 0.001180144026875496,
-      "learning_rate": 0.00019999971566647914,
-      "loss": 46.0,
-      "step": 9938
-    },
-    {
-      "epoch": 0.7599059579104307,
-      "grad_norm": 0.0007810101960785687,
-      "learning_rate": 0.0001999997156091972,
-      "loss": 46.0,
-      "step": 9939
-    },
-    {
-      "epoch": 0.7599824148938203,
-      "grad_norm": 0.0006825368036516011,
-      "learning_rate": 0.00019999971555190947,
-      "loss": 46.0,
-      "step": 9940
-    },
-    {
-      "epoch": 0.7600588718772101,
-      "grad_norm": 0.0009707232820801437,
-      "learning_rate": 0.00019999971549461594,
-      "loss": 46.0,
-      "step": 9941
-    },
-    {
-      "epoch": 0.7601353288605998,
-      "grad_norm": 0.0010974803008139133,
-      "learning_rate": 0.00019999971543731667,
-      "loss": 46.0,
-      "step": 9942
-    },
-    {
-      "epoch": 0.7602117858439895,
-      "grad_norm": 0.003200179897248745,
-      "learning_rate": 0.0001999997153800116,
-      "loss": 46.0,
-      "step": 9943
-    },
-    {
-      "epoch": 0.7602882428273793,
-      "grad_norm": 0.0021847127936780453,
-      "learning_rate": 0.00019999971532270078,
-      "loss": 46.0,
-      "step": 9944
-    },
-    {
-      "epoch": 0.7603646998107689,
-      "grad_norm": 0.0005863320548087358,
-      "learning_rate": 0.00019999971526538421,
-      "loss": 46.0,
-      "step": 9945
-    },
-    {
-      "epoch": 0.7604411567941587,
-      "grad_norm": 0.008221834897994995,
-      "learning_rate": 0.00019999971520806185,
-      "loss": 46.0,
-      "step": 9946
-    },
-    {
-      "epoch": 0.7605176137775485,
-      "grad_norm": 0.0010193337220698595,
-      "learning_rate": 0.0001999997151507337,
-      "loss": 46.0,
-      "step": 9947
-    },
-    {
-      "epoch": 0.7605940707609381,
-      "grad_norm": 0.0005310768028721213,
-      "learning_rate": 0.00019999971509339983,
-      "loss": 46.0,
-      "step": 9948
-    },
-    {
-      "epoch": 0.7606705277443279,
-      "grad_norm": 0.001330150873400271,
-      "learning_rate": 0.00019999971503606014,
-      "loss": 46.0,
-      "step": 9949
-    },
-    {
-      "epoch": 0.7607469847277176,
-      "grad_norm": 0.001468959730118513,
-      "learning_rate": 0.00019999971497871474,
-      "loss": 46.0,
-      "step": 9950
-    },
-    {
-      "epoch": 0.7608234417111073,
-      "grad_norm": 0.000902983476407826,
-      "learning_rate": 0.00019999971492136353,
-      "loss": 46.0,
-      "step": 9951
-    },
-    {
-      "epoch": 0.760899898694497,
-      "grad_norm": 0.0020855721086263657,
-      "learning_rate": 0.00019999971486400655,
-      "loss": 46.0,
-      "step": 9952
-    },
-    {
-      "epoch": 0.7609763556778867,
-      "grad_norm": 0.0006067457725293934,
-      "learning_rate": 0.0001999997148066438,
-      "loss": 46.0,
-      "step": 9953
-    },
-    {
-      "epoch": 0.7610528126612764,
-      "grad_norm": 0.0014846770791336894,
-      "learning_rate": 0.00019999971474927528,
-      "loss": 46.0,
-      "step": 9954
-    },
-    {
-      "epoch": 0.7611292696446662,
-      "grad_norm": 0.0006153883296065032,
-      "learning_rate": 0.000199999714691901,
-      "loss": 46.0,
-      "step": 9955
-    },
-    {
-      "epoch": 0.7612057266280559,
-      "grad_norm": 0.003313641529530287,
-      "learning_rate": 0.00019999971463452093,
-      "loss": 46.0,
-      "step": 9956
-    },
-    {
-      "epoch": 0.7612821836114456,
-      "grad_norm": 0.007295271381735802,
-      "learning_rate": 0.00019999971457713514,
-      "loss": 46.0,
-      "step": 9957
-    },
-    {
-      "epoch": 0.7613586405948354,
-      "grad_norm": 0.0059340219013392925,
-      "learning_rate": 0.00019999971451974355,
-      "loss": 46.0,
-      "step": 9958
-    },
-    {
-      "epoch": 0.761435097578225,
-      "grad_norm": 0.0018647159449756145,
-      "learning_rate": 0.00019999971446234619,
-      "loss": 46.0,
-      "step": 9959
-    },
-    {
-      "epoch": 0.7615115545616148,
-      "grad_norm": 0.0006109130918048322,
-      "learning_rate": 0.00019999971440494305,
-      "loss": 46.0,
-      "step": 9960
-    },
-    {
-      "epoch": 0.7615880115450044,
-      "grad_norm": 0.0020299204625189304,
-      "learning_rate": 0.00019999971434753416,
-      "loss": 46.0,
-      "step": 9961
-    },
-    {
-      "epoch": 0.7616644685283942,
-      "grad_norm": 0.0012731908354908228,
-      "learning_rate": 0.0001999997142901195,
-      "loss": 46.0,
-      "step": 9962
-    },
-    {
-      "epoch": 0.761740925511784,
-      "grad_norm": 0.011232506483793259,
-      "learning_rate": 0.00019999971423269902,
-      "loss": 46.0,
-      "step": 9963
-    },
-    {
-      "epoch": 0.7618173824951736,
-      "grad_norm": 0.0007498626364395022,
-      "learning_rate": 0.00019999971417527284,
-      "loss": 46.0,
-      "step": 9964
-    },
-    {
-      "epoch": 0.7618938394785634,
-      "grad_norm": 0.006138013210147619,
-      "learning_rate": 0.00019999971411784086,
-      "loss": 46.0,
-      "step": 9965
-    },
-    {
-      "epoch": 0.7619702964619531,
-      "grad_norm": 0.0006603064830414951,
-      "learning_rate": 0.0001999997140604031,
-      "loss": 46.0,
-      "step": 9966
-    },
-    {
-      "epoch": 0.7620467534453428,
-      "grad_norm": 0.0010145963169634342,
-      "learning_rate": 0.00019999971400295961,
-      "loss": 46.0,
-      "step": 9967
-    },
-    {
-      "epoch": 0.7621232104287325,
-      "grad_norm": 0.0020180607680231333,
-      "learning_rate": 0.00019999971394551032,
-      "loss": 46.0,
-      "step": 9968
-    },
-    {
-      "epoch": 0.7621996674121223,
-      "grad_norm": 0.0018007459584623575,
-      "learning_rate": 0.00019999971388805525,
-      "loss": 46.0,
-      "step": 9969
-    },
-    {
-      "epoch": 0.762276124395512,
-      "grad_norm": 0.0006252261810004711,
-      "learning_rate": 0.00019999971383059443,
-      "loss": 46.0,
-      "step": 9970
-    },
-    {
-      "epoch": 0.7623525813789017,
-      "grad_norm": 0.005264383740723133,
-      "learning_rate": 0.00019999971377312786,
-      "loss": 46.0,
-      "step": 9971
-    },
-    {
-      "epoch": 0.7624290383622914,
-      "grad_norm": 0.0010963367531076074,
-      "learning_rate": 0.00019999971371565547,
-      "loss": 46.0,
-      "step": 9972
-    },
-    {
-      "epoch": 0.7625054953456811,
-      "grad_norm": 0.000961070996709168,
-      "learning_rate": 0.00019999971365817736,
-      "loss": 46.0,
-      "step": 9973
-    },
-    {
-      "epoch": 0.7625819523290709,
-      "grad_norm": 0.0021992179099470377,
-      "learning_rate": 0.00019999971360069345,
-      "loss": 46.0,
-      "step": 9974
-    },
-    {
-      "epoch": 0.7626584093124605,
-      "grad_norm": 0.0020298175513744354,
-      "learning_rate": 0.00019999971354320376,
-      "loss": 46.0,
-      "step": 9975
-    },
-    {
-      "epoch": 0.7627348662958503,
-      "grad_norm": 0.00126090447884053,
-      "learning_rate": 0.00019999971348570833,
-      "loss": 46.0,
-      "step": 9976
-    },
-    {
-      "epoch": 0.7628113232792401,
-      "grad_norm": 0.0012229454005137086,
-      "learning_rate": 0.00019999971342820713,
-      "loss": 46.0,
-      "step": 9977
-    },
-    {
-      "epoch": 0.7628877802626297,
-      "grad_norm": 0.0005309944390319288,
-      "learning_rate": 0.00019999971337070015,
-      "loss": 46.0,
-      "step": 9978
-    },
-    {
-      "epoch": 0.7629642372460195,
-      "grad_norm": 0.000510432873852551,
-      "learning_rate": 0.0001999997133131874,
-      "loss": 46.0,
-      "step": 9979
-    },
-    {
-      "epoch": 0.7630406942294092,
-      "grad_norm": 0.002291478216648102,
-      "learning_rate": 0.00019999971325566888,
-      "loss": 46.0,
-      "step": 9980
-    },
-    {
-      "epoch": 0.7631171512127989,
-      "grad_norm": 0.002825989620760083,
-      "learning_rate": 0.00019999971319814458,
-      "loss": 46.0,
-      "step": 9981
-    },
-    {
-      "epoch": 0.7631936081961886,
-      "grad_norm": 0.005555382464081049,
-      "learning_rate": 0.0001999997131406145,
-      "loss": 46.0,
-      "step": 9982
-    },
-    {
-      "epoch": 0.7632700651795783,
-      "grad_norm": 0.0030556845013052225,
-      "learning_rate": 0.00019999971308307872,
-      "loss": 46.0,
-      "step": 9983
-    },
-    {
-      "epoch": 0.763346522162968,
-      "grad_norm": 0.0011028972221538424,
-      "learning_rate": 0.0001999997130255371,
-      "loss": 46.0,
-      "step": 9984
-    },
-    {
-      "epoch": 0.7634229791463578,
-      "grad_norm": 0.0009719515219330788,
-      "learning_rate": 0.00019999971296798976,
-      "loss": 46.0,
-      "step": 9985
-    },
-    {
-      "epoch": 0.7634994361297475,
-      "grad_norm": 0.0018168498063459992,
-      "learning_rate": 0.0001999997129104366,
-      "loss": 46.0,
-      "step": 9986
-    },
-    {
-      "epoch": 0.7635758931131372,
-      "grad_norm": 0.0030886325985193253,
-      "learning_rate": 0.0001999997128528777,
-      "loss": 46.0,
-      "step": 9987
-    },
-    {
-      "epoch": 0.763652350096527,
-      "grad_norm": 0.0023773168213665485,
-      "learning_rate": 0.00019999971279531306,
-      "loss": 46.0,
-      "step": 9988
-    },
-    {
-      "epoch": 0.7637288070799166,
-      "grad_norm": 0.0010059482883661985,
-      "learning_rate": 0.00019999971273774257,
-      "loss": 46.0,
-      "step": 9989
-    },
-    {
-      "epoch": 0.7638052640633064,
-      "grad_norm": 0.00400012219324708,
-      "learning_rate": 0.00019999971268016637,
-      "loss": 46.0,
-      "step": 9990
-    },
-    {
-      "epoch": 0.763881721046696,
-      "grad_norm": 0.0010531320003792644,
-      "learning_rate": 0.0001999997126225844,
-      "loss": 46.0,
-      "step": 9991
-    },
-    {
-      "epoch": 0.7639581780300858,
-      "grad_norm": 0.005792735144495964,
-      "learning_rate": 0.00019999971256499667,
-      "loss": 46.0,
-      "step": 9992
-    },
-    {
-      "epoch": 0.7640346350134756,
-      "grad_norm": 0.0011541342828422785,
-      "learning_rate": 0.00019999971250740312,
-      "loss": 46.0,
-      "step": 9993
-    },
-    {
-      "epoch": 0.7641110919968652,
-      "grad_norm": 0.004007813986390829,
-      "learning_rate": 0.00019999971244980382,
-      "loss": 46.0,
-      "step": 9994
-    },
-    {
-      "epoch": 0.764187548980255,
-      "grad_norm": 0.0011837673373520374,
-      "learning_rate": 0.00019999971239219877,
-      "loss": 46.0,
-      "step": 9995
-    },
-    {
-      "epoch": 0.7642640059636447,
-      "grad_norm": 0.0006242874660529196,
-      "learning_rate": 0.00019999971233458793,
-      "loss": 46.0,
-      "step": 9996
-    },
-    {
-      "epoch": 0.7643404629470344,
-      "grad_norm": 0.0004229389887768775,
-      "learning_rate": 0.00019999971227697134,
-      "loss": 46.0,
-      "step": 9997
-    },
-    {
-      "epoch": 0.7644169199304242,
-      "grad_norm": 0.0009345015860162675,
-      "learning_rate": 0.00019999971221934898,
-      "loss": 46.0,
-      "step": 9998
-    },
-    {
-      "epoch": 0.7644933769138139,
-      "grad_norm": 0.0008252540719695389,
-      "learning_rate": 0.00019999971216172084,
-      "loss": 46.0,
-      "step": 9999
-    },
-    {
-      "epoch": 0.7645698338972036,
-      "grad_norm": 0.0016681652050465345,
-      "learning_rate": 0.00019999971210408695,
-      "loss": 46.0,
-      "step": 10000
-    },
-    {
-      "epoch": 0.7646462908805933,
-      "grad_norm": 0.0009806439047679305,
-      "learning_rate": 0.00019999971204644724,
-      "loss": 46.0,
-      "step": 10001
-    },
-    {
-      "epoch": 0.764722747863983,
-      "grad_norm": 0.0029184853192418814,
-      "learning_rate": 0.0001999997119888018,
-      "loss": 46.0,
-      "step": 10002
-    },
-    {
-      "epoch": 0.7647992048473727,
-      "grad_norm": 0.002406930085271597,
-      "learning_rate": 0.00019999971193115058,
-      "loss": 46.0,
-      "step": 10003
-    },
-    {
-      "epoch": 0.7648756618307625,
-      "grad_norm": 0.0013237128732725978,
-      "learning_rate": 0.00019999971187349363,
-      "loss": 46.0,
-      "step": 10004
-    },
-    {
-      "epoch": 0.7649521188141521,
-      "grad_norm": 0.0009538425947539508,
-      "learning_rate": 0.00019999971181583085,
-      "loss": 46.0,
-      "step": 10005
-    },
-    {
-      "epoch": 0.7650285757975419,
-      "grad_norm": 0.0014584185555577278,
-      "learning_rate": 0.00019999971175816233,
-      "loss": 46.0,
-      "step": 10006
-    },
-    {
-      "epoch": 0.7651050327809317,
-      "grad_norm": 0.001428199466317892,
-      "learning_rate": 0.00019999971170048803,
-      "loss": 46.0,
-      "step": 10007
-    },
-    {
-      "epoch": 0.7651814897643213,
-      "grad_norm": 0.001289650797843933,
-      "learning_rate": 0.00019999971164280796,
-      "loss": 46.0,
-      "step": 10008
-    },
-    {
-      "epoch": 0.7652579467477111,
-      "grad_norm": 0.001247707405127585,
-      "learning_rate": 0.00019999971158512215,
-      "loss": 46.0,
-      "step": 10009
-    },
-    {
-      "epoch": 0.7653344037311008,
-      "grad_norm": 0.003696620464324951,
-      "learning_rate": 0.00019999971152743053,
-      "loss": 46.0,
-      "step": 10010
-    },
-    {
-      "epoch": 0.7654108607144905,
-      "grad_norm": 0.000639944220893085,
-      "learning_rate": 0.00019999971146973317,
-      "loss": 46.0,
-      "step": 10011
-    },
-    {
-      "epoch": 0.7654873176978803,
-      "grad_norm": 0.0026720098685473204,
-      "learning_rate": 0.00019999971141203006,
-      "loss": 46.0,
-      "step": 10012
-    },
-    {
-      "epoch": 0.7655637746812699,
-      "grad_norm": 0.0007816367433406413,
-      "learning_rate": 0.00019999971135432112,
-      "loss": 46.0,
-      "step": 10013
-    },
-    {
-      "epoch": 0.7656402316646597,
-      "grad_norm": 0.001667184173129499,
-      "learning_rate": 0.00019999971129660643,
-      "loss": 46.0,
-      "step": 10014
-    },
-    {
-      "epoch": 0.7657166886480494,
-      "grad_norm": 0.0010481046047061682,
-      "learning_rate": 0.00019999971123888598,
-      "loss": 46.0,
-      "step": 10015
-    },
-    {
-      "epoch": 0.7657931456314391,
-      "grad_norm": 0.0035507171414792538,
-      "learning_rate": 0.0001999997111811598,
-      "loss": 46.0,
-      "step": 10016
-    },
-    {
-      "epoch": 0.7658696026148288,
-      "grad_norm": 0.010692774318158627,
-      "learning_rate": 0.0001999997111234278,
-      "loss": 46.0,
-      "step": 10017
-    },
-    {
-      "epoch": 0.7659460595982186,
-      "grad_norm": 0.0014783920487388968,
-      "learning_rate": 0.00019999971106569002,
-      "loss": 46.0,
-      "step": 10018
-    },
-    {
-      "epoch": 0.7660225165816082,
-      "grad_norm": 0.0014465571148321033,
-      "learning_rate": 0.0001999997110079465,
-      "loss": 46.0,
-      "step": 10019
-    },
-    {
-      "epoch": 0.766098973564998,
-      "grad_norm": 0.0019483668729662895,
-      "learning_rate": 0.0001999997109501972,
-      "loss": 46.0,
-      "step": 10020
-    },
-    {
-      "epoch": 0.7661754305483877,
-      "grad_norm": 0.0007734148530289531,
-      "learning_rate": 0.00019999971089244213,
-      "loss": 46.0,
-      "step": 10021
-    },
-    {
-      "epoch": 0.7662518875317774,
-      "grad_norm": 0.0006514661945402622,
-      "learning_rate": 0.00019999971083468129,
-      "loss": 46.0,
-      "step": 10022
-    },
-    {
-      "epoch": 0.7663283445151672,
-      "grad_norm": 0.0006578920874744654,
-      "learning_rate": 0.0001999997107769147,
-      "loss": 46.0,
-      "step": 10023
-    },
-    {
-      "epoch": 0.7664048014985568,
-      "grad_norm": 0.0010734456591308117,
-      "learning_rate": 0.00019999971071914233,
-      "loss": 46.0,
-      "step": 10024
-    },
-    {
-      "epoch": 0.7664812584819466,
-      "grad_norm": 0.0026550416368991137,
-      "learning_rate": 0.0001999997106613642,
-      "loss": 46.0,
-      "step": 10025
-    },
-    {
-      "epoch": 0.7665577154653364,
-      "grad_norm": 0.001050087041221559,
-      "learning_rate": 0.00019999971060358026,
-      "loss": 46.0,
-      "step": 10026
-    },
-    {
-      "epoch": 0.766634172448726,
-      "grad_norm": 0.0006363586871884763,
-      "learning_rate": 0.0001999997105457906,
-      "loss": 46.0,
-      "step": 10027
-    },
-    {
-      "epoch": 0.7667106294321158,
-      "grad_norm": 0.001123207388445735,
-      "learning_rate": 0.00019999971048799515,
-      "loss": 46.0,
-      "step": 10028
-    },
-    {
-      "epoch": 0.7667870864155055,
-      "grad_norm": 0.0021201076451689005,
-      "learning_rate": 0.00019999971043019392,
-      "loss": 46.0,
-      "step": 10029
-    },
-    {
-      "epoch": 0.7668635433988952,
-      "grad_norm": 0.0008306613890454173,
-      "learning_rate": 0.0001999997103723869,
-      "loss": 46.0,
-      "step": 10030
-    },
-    {
-      "epoch": 0.7669400003822849,
-      "grad_norm": 0.0012050301302224398,
-      "learning_rate": 0.0001999997103145742,
-      "loss": 46.0,
-      "step": 10031
-    },
-    {
-      "epoch": 0.7670164573656746,
-      "grad_norm": 0.0010074771707877517,
-      "learning_rate": 0.00019999971025675564,
-      "loss": 46.0,
-      "step": 10032
-    },
-    {
-      "epoch": 0.7670929143490643,
-      "grad_norm": 0.011882204562425613,
-      "learning_rate": 0.00019999971019893137,
-      "loss": 46.0,
-      "step": 10033
-    },
-    {
-      "epoch": 0.7671693713324541,
-      "grad_norm": 0.005684808362275362,
-      "learning_rate": 0.00019999971014110127,
-      "loss": 46.0,
-      "step": 10034
-    },
-    {
-      "epoch": 0.7672458283158438,
-      "grad_norm": 0.0023781689815223217,
-      "learning_rate": 0.00019999971008326543,
-      "loss": 46.0,
-      "step": 10035
-    },
-    {
-      "epoch": 0.7673222852992335,
-      "grad_norm": 0.01235031895339489,
-      "learning_rate": 0.00019999971002542384,
-      "loss": 46.0,
-      "step": 10036
-    },
-    {
-      "epoch": 0.7673987422826233,
-      "grad_norm": 0.001249405206181109,
-      "learning_rate": 0.00019999970996757648,
-      "loss": 46.0,
-      "step": 10037
-    },
-    {
-      "epoch": 0.7674751992660129,
-      "grad_norm": 0.0008550244965590537,
-      "learning_rate": 0.0001999997099097233,
-      "loss": 46.0,
-      "step": 10038
-    },
-    {
-      "epoch": 0.7675516562494027,
-      "grad_norm": 0.0011739303590729833,
-      "learning_rate": 0.0001999997098518644,
-      "loss": 46.0,
-      "step": 10039
-    },
-    {
-      "epoch": 0.7676281132327925,
-      "grad_norm": 0.0013601342216134071,
-      "learning_rate": 0.00019999970979399972,
-      "loss": 46.0,
-      "step": 10040
-    },
-    {
-      "epoch": 0.7677045702161821,
-      "grad_norm": 0.0012962105683982372,
-      "learning_rate": 0.00019999970973612924,
-      "loss": 46.0,
-      "step": 10041
-    },
-    {
-      "epoch": 0.7677810271995719,
-      "grad_norm": 0.008770779706537724,
-      "learning_rate": 0.00019999970967825303,
-      "loss": 46.0,
-      "step": 10042
-    },
-    {
-      "epoch": 0.7678574841829615,
-      "grad_norm": 0.0013921101344749331,
-      "learning_rate": 0.00019999970962037106,
-      "loss": 46.0,
-      "step": 10043
-    },
-    {
-      "epoch": 0.7679339411663513,
-      "grad_norm": 0.0010800652671605349,
-      "learning_rate": 0.00019999970956248325,
-      "loss": 46.0,
-      "step": 10044
-    },
-    {
-      "epoch": 0.768010398149741,
-      "grad_norm": 0.0005175610422156751,
-      "learning_rate": 0.00019999970950458976,
-      "loss": 46.0,
-      "step": 10045
-    },
-    {
-      "epoch": 0.7680868551331307,
-      "grad_norm": 0.00039670823025517166,
-      "learning_rate": 0.00019999970944669043,
-      "loss": 46.0,
-      "step": 10046
-    },
-    {
-      "epoch": 0.7681633121165204,
-      "grad_norm": 0.002805059775710106,
-      "learning_rate": 0.00019999970938878536,
-      "loss": 46.0,
-      "step": 10047
-    },
-    {
-      "epoch": 0.7682397690999102,
-      "grad_norm": 0.0008304428192786872,
-      "learning_rate": 0.00019999970933087452,
-      "loss": 46.0,
-      "step": 10048
-    },
-    {
-      "epoch": 0.7683162260832999,
-      "grad_norm": 0.0008633029647171497,
-      "learning_rate": 0.00019999970927295793,
-      "loss": 46.0,
-      "step": 10049
-    },
-    {
-      "epoch": 0.7683926830666896,
-      "grad_norm": 0.0013363687321543694,
-      "learning_rate": 0.00019999970921503554,
-      "loss": 46.0,
-      "step": 10050
-    },
-    {
-      "epoch": 0.7684691400500793,
-      "grad_norm": 0.0007525449036620557,
-      "learning_rate": 0.0001999997091571074,
-      "loss": 46.0,
-      "step": 10051
-    },
-    {
-      "epoch": 0.768545597033469,
-      "grad_norm": 0.002808792283758521,
-      "learning_rate": 0.00019999970909917347,
-      "loss": 46.0,
-      "step": 10052
-    },
-    {
-      "epoch": 0.7686220540168588,
-      "grad_norm": 0.0007879525073803961,
-      "learning_rate": 0.0001999997090412338,
-      "loss": 46.0,
-      "step": 10053
-    },
-    {
-      "epoch": 0.7686985110002484,
-      "grad_norm": 0.001690176664851606,
-      "learning_rate": 0.00019999970898328836,
-      "loss": 46.0,
-      "step": 10054
-    },
-    {
-      "epoch": 0.7687749679836382,
-      "grad_norm": 0.004503827542066574,
-      "learning_rate": 0.0001999997089253371,
-      "loss": 46.0,
-      "step": 10055
-    },
-    {
-      "epoch": 0.768851424967028,
-      "grad_norm": 0.00091963866725564,
-      "learning_rate": 0.0001999997088673801,
-      "loss": 46.0,
-      "step": 10056
-    },
-    {
-      "epoch": 0.7689278819504176,
-      "grad_norm": 0.0014203267637640238,
-      "learning_rate": 0.00019999970880941735,
-      "loss": 46.0,
-      "step": 10057
-    },
-    {
-      "epoch": 0.7690043389338074,
-      "grad_norm": 0.003359684254974127,
-      "learning_rate": 0.00019999970875144883,
-      "loss": 46.0,
-      "step": 10058
-    },
-    {
-      "epoch": 0.7690807959171971,
-      "grad_norm": 0.0010892476420849562,
-      "learning_rate": 0.0001999997086934745,
-      "loss": 46.0,
-      "step": 10059
-    },
-    {
-      "epoch": 0.7691572529005868,
-      "grad_norm": 0.000888099311850965,
-      "learning_rate": 0.0001999997086354944,
-      "loss": 46.0,
-      "step": 10060
-    },
-    {
-      "epoch": 0.7692337098839765,
-      "grad_norm": 0.001151447999291122,
-      "learning_rate": 0.0001999997085775086,
-      "loss": 46.0,
-      "step": 10061
-    },
-    {
-      "epoch": 0.7693101668673662,
-      "grad_norm": 0.0020006056874990463,
-      "learning_rate": 0.00019999970851951698,
-      "loss": 46.0,
-      "step": 10062
-    },
-    {
-      "epoch": 0.769386623850756,
-      "grad_norm": 0.0014457236975431442,
-      "learning_rate": 0.0001999997084615196,
-      "loss": 46.0,
-      "step": 10063
-    },
-    {
-      "epoch": 0.7694630808341457,
-      "grad_norm": 0.0010294867679476738,
-      "learning_rate": 0.00019999970840351645,
-      "loss": 46.0,
-      "step": 10064
-    },
-    {
-      "epoch": 0.7695395378175354,
-      "grad_norm": 0.0005748987314291298,
-      "learning_rate": 0.00019999970834550754,
-      "loss": 46.0,
-      "step": 10065
-    },
-    {
-      "epoch": 0.7696159948009251,
-      "grad_norm": 0.001188112422823906,
-      "learning_rate": 0.00019999970828749283,
-      "loss": 46.0,
-      "step": 10066
-    },
-    {
-      "epoch": 0.7696924517843149,
-      "grad_norm": 0.0028988078702241182,
-      "learning_rate": 0.00019999970822947238,
-      "loss": 46.0,
-      "step": 10067
-    },
-    {
-      "epoch": 0.7697689087677045,
-      "grad_norm": 0.0007686301250942051,
-      "learning_rate": 0.00019999970817144615,
-      "loss": 46.0,
-      "step": 10068
-    },
-    {
-      "epoch": 0.7698453657510943,
-      "grad_norm": 0.0038459505885839462,
-      "learning_rate": 0.00019999970811341414,
-      "loss": 46.0,
-      "step": 10069
-    },
-    {
-      "epoch": 0.7699218227344841,
-      "grad_norm": 0.0004141342651564628,
-      "learning_rate": 0.00019999970805537637,
-      "loss": 46.0,
-      "step": 10070
-    },
-    {
-      "epoch": 0.7699982797178737,
-      "grad_norm": 0.0011534850345924497,
-      "learning_rate": 0.00019999970799733285,
-      "loss": 46.0,
-      "step": 10071
-    },
-    {
-      "epoch": 0.7700747367012635,
-      "grad_norm": 0.0019190703751519322,
-      "learning_rate": 0.00019999970793928355,
-      "loss": 46.0,
-      "step": 10072
-    },
-    {
-      "epoch": 0.7701511936846531,
-      "grad_norm": 0.003268068889155984,
-      "learning_rate": 0.00019999970788122845,
-      "loss": 46.0,
-      "step": 10073
-    },
-    {
-      "epoch": 0.7702276506680429,
-      "grad_norm": 0.0019710457418113947,
-      "learning_rate": 0.0001999997078231676,
-      "loss": 46.0,
-      "step": 10074
-    },
-    {
-      "epoch": 0.7703041076514326,
-      "grad_norm": 0.0041036042384803295,
-      "learning_rate": 0.00019999970776510102,
-      "loss": 46.0,
-      "step": 10075
-    },
-    {
-      "epoch": 0.7703805646348223,
-      "grad_norm": 0.0012877883855253458,
-      "learning_rate": 0.00019999970770702863,
-      "loss": 46.0,
-      "step": 10076
-    },
-    {
-      "epoch": 0.7704570216182121,
-      "grad_norm": 0.0023239273577928543,
-      "learning_rate": 0.00019999970764895047,
-      "loss": 46.0,
-      "step": 10077
-    },
-    {
-      "epoch": 0.7705334786016018,
-      "grad_norm": 0.002815867541357875,
-      "learning_rate": 0.00019999970759086654,
-      "loss": 46.0,
-      "step": 10078
-    },
-    {
-      "epoch": 0.7706099355849915,
-      "grad_norm": 0.0008432841277681291,
-      "learning_rate": 0.00019999970753277685,
-      "loss": 46.0,
-      "step": 10079
-    },
-    {
-      "epoch": 0.7706863925683812,
-      "grad_norm": 0.00033620651811361313,
-      "learning_rate": 0.00019999970747468137,
-      "loss": 46.0,
-      "step": 10080
-    },
-    {
-      "epoch": 0.770762849551771,
-      "grad_norm": 0.0012031936785206199,
-      "learning_rate": 0.00019999970741658017,
-      "loss": 46.0,
-      "step": 10081
-    },
-    {
-      "epoch": 0.7708393065351606,
-      "grad_norm": 0.000884325650986284,
-      "learning_rate": 0.00019999970735847317,
-      "loss": 46.0,
-      "step": 10082
-    },
-    {
-      "epoch": 0.7709157635185504,
-      "grad_norm": 0.0004033608711324632,
-      "learning_rate": 0.00019999970730036036,
-      "loss": 46.0,
-      "step": 10083
-    },
-    {
-      "epoch": 0.77099222050194,
-      "grad_norm": 0.0015430657658725977,
-      "learning_rate": 0.00019999970724224184,
-      "loss": 46.0,
-      "step": 10084
-    },
-    {
-      "epoch": 0.7710686774853298,
-      "grad_norm": 0.0012954644626006484,
-      "learning_rate": 0.00019999970718411755,
-      "loss": 46.0,
-      "step": 10085
-    },
-    {
-      "epoch": 0.7711451344687196,
-      "grad_norm": 0.0052140625193715096,
-      "learning_rate": 0.00019999970712598745,
-      "loss": 46.0,
-      "step": 10086
-    },
-    {
-      "epoch": 0.7712215914521092,
-      "grad_norm": 0.0019443245837464929,
-      "learning_rate": 0.0001999997070678516,
-      "loss": 46.0,
-      "step": 10087
-    },
-    {
-      "epoch": 0.771298048435499,
-      "grad_norm": 0.0048497579991817474,
-      "learning_rate": 0.00019999970700970997,
-      "loss": 46.0,
-      "step": 10088
-    },
-    {
-      "epoch": 0.7713745054188887,
-      "grad_norm": 0.0006835806998424232,
-      "learning_rate": 0.0001999997069515626,
-      "loss": 46.0,
-      "step": 10089
-    },
-    {
-      "epoch": 0.7714509624022784,
-      "grad_norm": 0.0033077120315283537,
-      "learning_rate": 0.00019999970689340944,
-      "loss": 46.0,
-      "step": 10090
-    },
-    {
-      "epoch": 0.7715274193856682,
-      "grad_norm": 0.001092192716896534,
-      "learning_rate": 0.0001999997068352505,
-      "loss": 46.0,
-      "step": 10091
-    },
-    {
-      "epoch": 0.7716038763690578,
-      "grad_norm": 0.004216737579554319,
-      "learning_rate": 0.00019999970677708583,
-      "loss": 46.0,
-      "step": 10092
-    },
-    {
-      "epoch": 0.7716803333524476,
-      "grad_norm": 0.001018659444525838,
-      "learning_rate": 0.00019999970671891534,
-      "loss": 46.0,
-      "step": 10093
-    },
-    {
-      "epoch": 0.7717567903358373,
-      "grad_norm": 0.002728919265791774,
-      "learning_rate": 0.00019999970666073911,
-      "loss": 46.0,
-      "step": 10094
-    },
-    {
-      "epoch": 0.771833247319227,
-      "grad_norm": 0.003599038114771247,
-      "learning_rate": 0.00019999970660255714,
-      "loss": 46.0,
-      "step": 10095
-    },
-    {
-      "epoch": 0.7719097043026167,
-      "grad_norm": 0.0007787556387484074,
-      "learning_rate": 0.00019999970654436934,
-      "loss": 46.0,
-      "step": 10096
-    },
-    {
-      "epoch": 0.7719861612860065,
-      "grad_norm": 0.0020864014513790607,
-      "learning_rate": 0.00019999970648617582,
-      "loss": 46.0,
-      "step": 10097
-    },
-    {
-      "epoch": 0.7720626182693962,
-      "grad_norm": 0.0010055698221549392,
-      "learning_rate": 0.0001999997064279765,
-      "loss": 46.0,
-      "step": 10098
-    },
-    {
-      "epoch": 0.7721390752527859,
-      "grad_norm": 0.000786547374445945,
-      "learning_rate": 0.0001999997063697714,
-      "loss": 46.0,
-      "step": 10099
-    },
-    {
-      "epoch": 0.7722155322361757,
-      "grad_norm": 0.0012661585351452231,
-      "learning_rate": 0.00019999970631156058,
-      "loss": 46.0,
-      "step": 10100
-    },
-    {
-      "epoch": 0.7722919892195653,
-      "grad_norm": 0.0011497859377413988,
-      "learning_rate": 0.00019999970625334394,
-      "loss": 46.0,
-      "step": 10101
-    },
-    {
-      "epoch": 0.7723684462029551,
-      "grad_norm": 0.004212136846035719,
-      "learning_rate": 0.00019999970619512153,
-      "loss": 46.0,
-      "step": 10102
-    },
-    {
-      "epoch": 0.7724449031863447,
-      "grad_norm": 0.0006643040687777102,
-      "learning_rate": 0.00019999970613689342,
-      "loss": 46.0,
-      "step": 10103
-    },
-    {
-      "epoch": 0.7725213601697345,
-      "grad_norm": 0.0018662265501916409,
-      "learning_rate": 0.00019999970607865946,
-      "loss": 46.0,
-      "step": 10104
-    },
-    {
-      "epoch": 0.7725978171531243,
-      "grad_norm": 0.0005415039486251771,
-      "learning_rate": 0.00019999970602041977,
-      "loss": 46.0,
-      "step": 10105
-    },
-    {
-      "epoch": 0.7726742741365139,
-      "grad_norm": 0.0031976881437003613,
-      "learning_rate": 0.0001999997059621743,
-      "loss": 46.0,
-      "step": 10106
-    },
-    {
-      "epoch": 0.7727507311199037,
-      "grad_norm": 0.015071452595293522,
-      "learning_rate": 0.00019999970590392306,
-      "loss": 46.0,
-      "step": 10107
-    },
-    {
-      "epoch": 0.7728271881032934,
-      "grad_norm": 0.0006291063036769629,
-      "learning_rate": 0.00019999970584566606,
-      "loss": 46.0,
-      "step": 10108
-    },
-    {
-      "epoch": 0.7729036450866831,
-      "grad_norm": 0.0007333066314458847,
-      "learning_rate": 0.0001999997057874033,
-      "loss": 46.0,
-      "step": 10109
-    },
-    {
-      "epoch": 0.7729801020700728,
-      "grad_norm": 0.003073690924793482,
-      "learning_rate": 0.00019999970572913474,
-      "loss": 46.0,
-      "step": 10110
-    },
-    {
-      "epoch": 0.7730565590534626,
-      "grad_norm": 0.0013207502197474241,
-      "learning_rate": 0.00019999970567086042,
-      "loss": 46.0,
-      "step": 10111
-    },
-    {
-      "epoch": 0.7731330160368522,
-      "grad_norm": 0.0030079903081059456,
-      "learning_rate": 0.00019999970561258032,
-      "loss": 46.0,
-      "step": 10112
-    },
-    {
-      "epoch": 0.773209473020242,
-      "grad_norm": 0.0005014106864109635,
-      "learning_rate": 0.00019999970555429448,
-      "loss": 46.0,
-      "step": 10113
-    },
-    {
-      "epoch": 0.7732859300036317,
-      "grad_norm": 0.00029658337007276714,
-      "learning_rate": 0.00019999970549600287,
-      "loss": 46.0,
-      "step": 10114
-    },
-    {
-      "epoch": 0.7733623869870214,
-      "grad_norm": 0.0013163614785298705,
-      "learning_rate": 0.00019999970543770548,
-      "loss": 46.0,
-      "step": 10115
-    },
-    {
-      "epoch": 0.7734388439704112,
-      "grad_norm": 0.0018276681657880545,
-      "learning_rate": 0.00019999970537940232,
-      "loss": 46.0,
-      "step": 10116
-    },
-    {
-      "epoch": 0.7735153009538008,
-      "grad_norm": 0.004299021791666746,
-      "learning_rate": 0.00019999970532109338,
-      "loss": 46.0,
-      "step": 10117
-    },
-    {
-      "epoch": 0.7735917579371906,
-      "grad_norm": 0.0011009597219526768,
-      "learning_rate": 0.00019999970526277867,
-      "loss": 46.0,
-      "step": 10118
-    },
-    {
-      "epoch": 0.7736682149205804,
-      "grad_norm": 0.000331989664118737,
-      "learning_rate": 0.00019999970520445822,
-      "loss": 46.0,
-      "step": 10119
-    },
-    {
-      "epoch": 0.77374467190397,
-      "grad_norm": 0.00110848608892411,
-      "learning_rate": 0.000199999705146132,
-      "loss": 46.0,
-      "step": 10120
-    },
-    {
-      "epoch": 0.7738211288873598,
-      "grad_norm": 0.0003543139318935573,
-      "learning_rate": 0.00019999970508779996,
-      "loss": 46.0,
-      "step": 10121
-    },
-    {
-      "epoch": 0.7738975858707494,
-      "grad_norm": 0.0002889388124458492,
-      "learning_rate": 0.0001999997050294622,
-      "loss": 46.0,
-      "step": 10122
-    },
-    {
-      "epoch": 0.7739740428541392,
-      "grad_norm": 0.0030940936412662268,
-      "learning_rate": 0.00019999970497111864,
-      "loss": 46.0,
-      "step": 10123
-    },
-    {
-      "epoch": 0.7740504998375289,
-      "grad_norm": 0.0007068327395245433,
-      "learning_rate": 0.00019999970491276932,
-      "loss": 46.0,
-      "step": 10124
-    },
-    {
-      "epoch": 0.7741269568209186,
-      "grad_norm": 0.0019399757729843259,
-      "learning_rate": 0.00019999970485441422,
-      "loss": 46.0,
-      "step": 10125
-    },
-    {
-      "epoch": 0.7742034138043083,
-      "grad_norm": 0.0028171653393656015,
-      "learning_rate": 0.00019999970479605338,
-      "loss": 46.0,
-      "step": 10126
-    },
-    {
-      "epoch": 0.7742798707876981,
-      "grad_norm": 0.001109176897443831,
-      "learning_rate": 0.00019999970473768674,
-      "loss": 46.0,
-      "step": 10127
-    },
-    {
-      "epoch": 0.7743563277710878,
-      "grad_norm": 0.0022520788479596376,
-      "learning_rate": 0.00019999970467931435,
-      "loss": 46.0,
-      "step": 10128
-    },
-    {
-      "epoch": 0.7744327847544775,
-      "grad_norm": 0.004060881678014994,
-      "learning_rate": 0.0001999997046209362,
-      "loss": 46.0,
-      "step": 10129
-    },
-    {
-      "epoch": 0.7745092417378673,
-      "grad_norm": 0.0006325824069790542,
-      "learning_rate": 0.00019999970456255226,
-      "loss": 46.0,
-      "step": 10130
-    },
-    {
-      "epoch": 0.7745856987212569,
-      "grad_norm": 0.009161490947008133,
-      "learning_rate": 0.00019999970450416258,
-      "loss": 46.0,
-      "step": 10131
-    },
-    {
-      "epoch": 0.7746621557046467,
-      "grad_norm": 0.0016569256549701095,
-      "learning_rate": 0.00019999970444576707,
-      "loss": 46.0,
-      "step": 10132
-    },
-    {
-      "epoch": 0.7747386126880363,
-      "grad_norm": 0.0018305061385035515,
-      "learning_rate": 0.00019999970438736584,
-      "loss": 46.0,
-      "step": 10133
-    },
-    {
-      "epoch": 0.7748150696714261,
-      "grad_norm": 0.000851231103297323,
-      "learning_rate": 0.00019999970432895884,
-      "loss": 46.0,
-      "step": 10134
-    },
-    {
-      "epoch": 0.7748915266548159,
-      "grad_norm": 0.0010570806916803122,
-      "learning_rate": 0.00019999970427054604,
-      "loss": 46.0,
-      "step": 10135
-    },
-    {
-      "epoch": 0.7749679836382055,
-      "grad_norm": 0.0009224657551385462,
-      "learning_rate": 0.00019999970421212752,
-      "loss": 46.0,
-      "step": 10136
-    },
-    {
-      "epoch": 0.7750444406215953,
-      "grad_norm": 0.0005287441308610141,
-      "learning_rate": 0.00019999970415370317,
-      "loss": 46.0,
-      "step": 10137
-    },
-    {
-      "epoch": 0.775120897604985,
-      "grad_norm": 0.0024804340209811926,
-      "learning_rate": 0.00019999970409527307,
-      "loss": 46.0,
-      "step": 10138
-    },
-    {
-      "epoch": 0.7751973545883747,
-      "grad_norm": 0.0013349383370950818,
-      "learning_rate": 0.00019999970403683723,
-      "loss": 46.0,
-      "step": 10139
-    },
-    {
-      "epoch": 0.7752738115717644,
-      "grad_norm": 0.0006791549385525286,
-      "learning_rate": 0.0001999997039783956,
-      "loss": 46.0,
-      "step": 10140
-    },
-    {
-      "epoch": 0.7753502685551542,
-      "grad_norm": 0.002364273415878415,
-      "learning_rate": 0.0001999997039199482,
-      "loss": 46.0,
-      "step": 10141
-    },
-    {
-      "epoch": 0.7754267255385439,
-      "grad_norm": 0.003655507694929838,
-      "learning_rate": 0.00019999970386149502,
-      "loss": 46.0,
-      "step": 10142
-    },
-    {
-      "epoch": 0.7755031825219336,
-      "grad_norm": 0.0025535959284752607,
-      "learning_rate": 0.0001999997038030361,
-      "loss": 46.0,
-      "step": 10143
-    },
-    {
-      "epoch": 0.7755796395053233,
-      "grad_norm": 0.0012152957497164607,
-      "learning_rate": 0.00019999970374457137,
-      "loss": 46.0,
-      "step": 10144
-    },
-    {
-      "epoch": 0.775656096488713,
-      "grad_norm": 0.0010598603403195739,
-      "learning_rate": 0.0001999997036861009,
-      "loss": 46.0,
-      "step": 10145
-    },
-    {
-      "epoch": 0.7757325534721028,
-      "grad_norm": 0.0007946204277686775,
-      "learning_rate": 0.00019999970362762466,
-      "loss": 46.0,
-      "step": 10146
-    },
-    {
-      "epoch": 0.7758090104554924,
-      "grad_norm": 0.0011012312024831772,
-      "learning_rate": 0.00019999970356914264,
-      "loss": 46.0,
-      "step": 10147
-    },
-    {
-      "epoch": 0.7758854674388822,
-      "grad_norm": 0.0010971761075779796,
-      "learning_rate": 0.00019999970351065486,
-      "loss": 46.0,
-      "step": 10148
-    },
-    {
-      "epoch": 0.775961924422272,
-      "grad_norm": 0.002020234940573573,
-      "learning_rate": 0.00019999970345216132,
-      "loss": 46.0,
-      "step": 10149
-    },
-    {
-      "epoch": 0.7760383814056616,
-      "grad_norm": 0.003866098588332534,
-      "learning_rate": 0.000199999703393662,
-      "loss": 46.0,
-      "step": 10150
-    },
-    {
-      "epoch": 0.7761148383890514,
-      "grad_norm": 0.0005179843283258379,
-      "learning_rate": 0.0001999997033351569,
-      "loss": 46.0,
-      "step": 10151
-    },
-    {
-      "epoch": 0.776191295372441,
-      "grad_norm": 0.0011416098568588495,
-      "learning_rate": 0.00019999970327664603,
-      "loss": 46.0,
-      "step": 10152
-    },
-    {
-      "epoch": 0.7762677523558308,
-      "grad_norm": 0.0037071979604661465,
-      "learning_rate": 0.00019999970321812942,
-      "loss": 46.0,
-      "step": 10153
-    },
-    {
-      "epoch": 0.7763442093392205,
-      "grad_norm": 0.0022465926595032215,
-      "learning_rate": 0.000199999703159607,
-      "loss": 46.0,
-      "step": 10154
-    },
-    {
-      "epoch": 0.7764206663226102,
-      "grad_norm": 0.000830504868645221,
-      "learning_rate": 0.00019999970310107882,
-      "loss": 46.0,
-      "step": 10155
-    },
-    {
-      "epoch": 0.776497123306,
-      "grad_norm": 0.002061625011265278,
-      "learning_rate": 0.00019999970304254488,
-      "loss": 46.0,
-      "step": 10156
-    },
-    {
-      "epoch": 0.7765735802893897,
-      "grad_norm": 0.0021929454524070024,
-      "learning_rate": 0.00019999970298400518,
-      "loss": 46.0,
-      "step": 10157
-    },
-    {
-      "epoch": 0.7766500372727794,
-      "grad_norm": 0.00038512024912051857,
-      "learning_rate": 0.00019999970292545972,
-      "loss": 46.0,
-      "step": 10158
-    },
-    {
-      "epoch": 0.7767264942561691,
-      "grad_norm": 0.0007712931255809963,
-      "learning_rate": 0.00019999970286690847,
-      "loss": 46.0,
-      "step": 10159
-    },
-    {
-      "epoch": 0.7768029512395589,
-      "grad_norm": 0.0007853909046389163,
-      "learning_rate": 0.00019999970280835144,
-      "loss": 46.0,
-      "step": 10160
-    },
-    {
-      "epoch": 0.7768794082229485,
-      "grad_norm": 0.0010165219428017735,
-      "learning_rate": 0.00019999970274978867,
-      "loss": 46.0,
-      "step": 10161
-    },
-    {
-      "epoch": 0.7769558652063383,
-      "grad_norm": 0.0006136196898296475,
-      "learning_rate": 0.0001999997026912201,
-      "loss": 46.0,
-      "step": 10162
-    },
-    {
-      "epoch": 0.777032322189728,
-      "grad_norm": 0.0019868165254592896,
-      "learning_rate": 0.00019999970263264577,
-      "loss": 46.0,
-      "step": 10163
-    },
-    {
-      "epoch": 0.7771087791731177,
-      "grad_norm": 0.0005031025502830744,
-      "learning_rate": 0.00019999970257406568,
-      "loss": 46.0,
-      "step": 10164
-    },
-    {
-      "epoch": 0.7771852361565075,
-      "grad_norm": 0.00151635252404958,
-      "learning_rate": 0.00019999970251547984,
-      "loss": 46.0,
-      "step": 10165
-    },
-    {
-      "epoch": 0.7772616931398971,
-      "grad_norm": 0.0004623220593202859,
-      "learning_rate": 0.00019999970245688817,
-      "loss": 46.0,
-      "step": 10166
-    },
-    {
-      "epoch": 0.7773381501232869,
-      "grad_norm": 0.0013860411709174514,
-      "learning_rate": 0.00019999970239829078,
-      "loss": 46.0,
-      "step": 10167
-    },
-    {
-      "epoch": 0.7774146071066766,
-      "grad_norm": 0.0015587200177833438,
-      "learning_rate": 0.00019999970233968762,
-      "loss": 46.0,
-      "step": 10168
-    },
-    {
-      "epoch": 0.7774910640900663,
-      "grad_norm": 0.0007903753430582583,
-      "learning_rate": 0.00019999970228107866,
-      "loss": 46.0,
-      "step": 10169
-    },
-    {
-      "epoch": 0.7775675210734561,
-      "grad_norm": 0.003197759622707963,
-      "learning_rate": 0.00019999970222246395,
-      "loss": 46.0,
-      "step": 10170
-    },
-    {
-      "epoch": 0.7776439780568458,
-      "grad_norm": 0.005163267254829407,
-      "learning_rate": 0.00019999970216384347,
-      "loss": 46.0,
-      "step": 10171
-    },
-    {
-      "epoch": 0.7777204350402355,
-      "grad_norm": 0.0005579793942160904,
-      "learning_rate": 0.00019999970210521722,
-      "loss": 46.0,
-      "step": 10172
-    },
-    {
-      "epoch": 0.7777968920236252,
-      "grad_norm": 0.0018871133215725422,
-      "learning_rate": 0.0001999997020465852,
-      "loss": 46.0,
-      "step": 10173
-    },
-    {
-      "epoch": 0.7778733490070149,
-      "grad_norm": 0.001524643157608807,
-      "learning_rate": 0.00019999970198794742,
-      "loss": 46.0,
-      "step": 10174
-    },
-    {
-      "epoch": 0.7779498059904046,
-      "grad_norm": 0.005308306775987148,
-      "learning_rate": 0.00019999970192930387,
-      "loss": 46.0,
-      "step": 10175
-    },
-    {
-      "epoch": 0.7780262629737944,
-      "grad_norm": 0.0004678885452449322,
-      "learning_rate": 0.00019999970187065453,
-      "loss": 46.0,
-      "step": 10176
-    },
-    {
-      "epoch": 0.778102719957184,
-      "grad_norm": 0.001828834880143404,
-      "learning_rate": 0.00019999970181199943,
-      "loss": 46.0,
-      "step": 10177
-    },
-    {
-      "epoch": 0.7781791769405738,
-      "grad_norm": 0.0013521984219551086,
-      "learning_rate": 0.00019999970175333857,
-      "loss": 46.0,
-      "step": 10178
-    },
-    {
-      "epoch": 0.7782556339239636,
-      "grad_norm": 0.001654301187954843,
-      "learning_rate": 0.00019999970169467193,
-      "loss": 46.0,
-      "step": 10179
-    },
-    {
-      "epoch": 0.7783320909073532,
-      "grad_norm": 0.0006634952733293176,
-      "learning_rate": 0.00019999970163599954,
-      "loss": 46.0,
-      "step": 10180
-    },
-    {
-      "epoch": 0.778408547890743,
-      "grad_norm": 0.005076178349554539,
-      "learning_rate": 0.00019999970157732135,
-      "loss": 46.0,
-      "step": 10181
-    },
-    {
-      "epoch": 0.7784850048741326,
-      "grad_norm": 0.0007074660388752818,
-      "learning_rate": 0.00019999970151863742,
-      "loss": 46.0,
-      "step": 10182
-    },
-    {
-      "epoch": 0.7785614618575224,
-      "grad_norm": 0.000865244772285223,
-      "learning_rate": 0.00019999970145994768,
-      "loss": 46.0,
-      "step": 10183
-    },
-    {
-      "epoch": 0.7786379188409122,
-      "grad_norm": 0.0011318132746964693,
-      "learning_rate": 0.0001999997014012522,
-      "loss": 46.0,
-      "step": 10184
-    },
-    {
-      "epoch": 0.7787143758243018,
-      "grad_norm": 0.0006942532490938902,
-      "learning_rate": 0.00019999970134255098,
-      "loss": 46.0,
-      "step": 10185
-    },
-    {
-      "epoch": 0.7787908328076916,
-      "grad_norm": 0.0020117023959755898,
-      "learning_rate": 0.00019999970128384392,
-      "loss": 46.0,
-      "step": 10186
-    },
-    {
-      "epoch": 0.7788672897910813,
-      "grad_norm": 0.000562444212846458,
-      "learning_rate": 0.00019999970122513115,
-      "loss": 46.0,
-      "step": 10187
-    },
-    {
-      "epoch": 0.778943746774471,
-      "grad_norm": 0.0008933614590205252,
-      "learning_rate": 0.00019999970116641258,
-      "loss": 46.0,
-      "step": 10188
-    },
-    {
-      "epoch": 0.7790202037578607,
-      "grad_norm": 0.0021899689454585314,
-      "learning_rate": 0.00019999970110768826,
-      "loss": 46.0,
-      "step": 10189
-    },
-    {
-      "epoch": 0.7790966607412505,
-      "grad_norm": 0.0010486440733075142,
-      "learning_rate": 0.00019999970104895814,
-      "loss": 46.0,
-      "step": 10190
-    },
-    {
-      "epoch": 0.7791731177246402,
-      "grad_norm": 0.0009347231825813651,
-      "learning_rate": 0.00019999970099022227,
-      "loss": 46.0,
-      "step": 10191
-    },
-    {
-      "epoch": 0.7792495747080299,
-      "grad_norm": 0.002102338243275881,
-      "learning_rate": 0.00019999970093148063,
-      "loss": 46.0,
-      "step": 10192
-    },
-    {
-      "epoch": 0.7793260316914196,
-      "grad_norm": 0.0018903817981481552,
-      "learning_rate": 0.00019999970087273324,
-      "loss": 46.0,
-      "step": 10193
-    },
-    {
-      "epoch": 0.7794024886748093,
-      "grad_norm": 0.0010162407997995615,
-      "learning_rate": 0.00019999970081398006,
-      "loss": 46.0,
-      "step": 10194
-    },
-    {
-      "epoch": 0.7794789456581991,
-      "grad_norm": 0.0005231364048086107,
-      "learning_rate": 0.0001999997007552211,
-      "loss": 46.0,
-      "step": 10195
-    },
-    {
-      "epoch": 0.7795554026415887,
-      "grad_norm": 0.0015077412826940417,
-      "learning_rate": 0.0001999997006964564,
-      "loss": 46.0,
-      "step": 10196
-    },
-    {
-      "epoch": 0.7796318596249785,
-      "grad_norm": 0.00283474731259048,
-      "learning_rate": 0.0001999997006376859,
-      "loss": 46.0,
-      "step": 10197
-    },
-    {
-      "epoch": 0.7797083166083683,
-      "grad_norm": 0.002034452510997653,
-      "learning_rate": 0.00019999970057890963,
-      "loss": 46.0,
-      "step": 10198
-    },
-    {
-      "epoch": 0.7797847735917579,
-      "grad_norm": 0.0038028694689273834,
-      "learning_rate": 0.00019999970052012763,
-      "loss": 46.0,
-      "step": 10199
-    },
-    {
-      "epoch": 0.7798612305751477,
-      "grad_norm": 0.0016110887518152595,
-      "learning_rate": 0.00019999970046133983,
-      "loss": 46.0,
-      "step": 10200
-    },
-    {
-      "epoch": 0.7799376875585374,
-      "grad_norm": 0.0029270027298480272,
-      "learning_rate": 0.00019999970040254628,
-      "loss": 46.0,
-      "step": 10201
-    },
-    {
-      "epoch": 0.7800141445419271,
-      "grad_norm": 0.0013306328328326344,
-      "learning_rate": 0.00019999970034374694,
-      "loss": 46.0,
-      "step": 10202
-    },
-    {
-      "epoch": 0.7800906015253168,
-      "grad_norm": 0.0011153396917507052,
-      "learning_rate": 0.00019999970028494182,
-      "loss": 46.0,
-      "step": 10203
-    },
-    {
-      "epoch": 0.7801670585087065,
-      "grad_norm": 0.001664671115577221,
-      "learning_rate": 0.00019999970022613095,
-      "loss": 46.0,
-      "step": 10204
-    },
-    {
-      "epoch": 0.7802435154920963,
-      "grad_norm": 0.001390974037349224,
-      "learning_rate": 0.0001999997001673143,
-      "loss": 46.0,
-      "step": 10205
-    },
-    {
-      "epoch": 0.780319972475486,
-      "grad_norm": 0.0008468308369629085,
-      "learning_rate": 0.0001999997001084919,
-      "loss": 46.0,
-      "step": 10206
-    },
-    {
-      "epoch": 0.7803964294588757,
-      "grad_norm": 0.0034670219756662846,
-      "learning_rate": 0.0001999997000496637,
-      "loss": 46.0,
-      "step": 10207
-    },
-    {
-      "epoch": 0.7804728864422654,
-      "grad_norm": 0.0007849931134842336,
-      "learning_rate": 0.00019999969999082978,
-      "loss": 46.0,
-      "step": 10208
-    },
-    {
-      "epoch": 0.7805493434256552,
-      "grad_norm": 0.005797098390758038,
-      "learning_rate": 0.00019999969993199005,
-      "loss": 46.0,
-      "step": 10209
-    },
-    {
-      "epoch": 0.7806258004090448,
-      "grad_norm": 0.0027351134922355413,
-      "learning_rate": 0.00019999969987314454,
-      "loss": 46.0,
-      "step": 10210
-    },
-    {
-      "epoch": 0.7807022573924346,
-      "grad_norm": 0.004473397042602301,
-      "learning_rate": 0.00019999969981429331,
-      "loss": 46.0,
-      "step": 10211
-    },
-    {
-      "epoch": 0.7807787143758244,
-      "grad_norm": 0.0014184268657118082,
-      "learning_rate": 0.00019999969975543626,
-      "loss": 46.0,
-      "step": 10212
-    },
-    {
-      "epoch": 0.780855171359214,
-      "grad_norm": 0.0017004070105031133,
-      "learning_rate": 0.0001999996996965735,
-      "loss": 46.0,
-      "step": 10213
-    },
-    {
-      "epoch": 0.7809316283426038,
-      "grad_norm": 0.0006384268635883927,
-      "learning_rate": 0.00019999969963770494,
-      "loss": 46.0,
-      "step": 10214
-    },
-    {
-      "epoch": 0.7810080853259934,
-      "grad_norm": 0.0009152454440481961,
-      "learning_rate": 0.00019999969957883057,
-      "loss": 46.0,
-      "step": 10215
-    },
-    {
-      "epoch": 0.7810845423093832,
-      "grad_norm": 0.0047762952744960785,
-      "learning_rate": 0.00019999969951995045,
-      "loss": 46.0,
-      "step": 10216
-    },
-    {
-      "epoch": 0.7811609992927729,
-      "grad_norm": 0.0029480753000825644,
-      "learning_rate": 0.00019999969946106458,
-      "loss": 46.0,
-      "step": 10217
-    },
-    {
-      "epoch": 0.7812374562761626,
-      "grad_norm": 0.0013304011663421988,
-      "learning_rate": 0.00019999969940217297,
-      "loss": 46.0,
-      "step": 10218
-    },
-    {
-      "epoch": 0.7813139132595524,
-      "grad_norm": 0.00200113607570529,
-      "learning_rate": 0.00019999969934327553,
-      "loss": 46.0,
-      "step": 10219
-    },
-    {
-      "epoch": 0.7813903702429421,
-      "grad_norm": 0.0030732930172234774,
-      "learning_rate": 0.00019999969928437237,
-      "loss": 46.0,
-      "step": 10220
-    },
-    {
-      "epoch": 0.7814668272263318,
-      "grad_norm": 0.0006126696243882179,
-      "learning_rate": 0.0001999996992254634,
-      "loss": 46.0,
-      "step": 10221
-    },
-    {
-      "epoch": 0.7815432842097215,
-      "grad_norm": 0.0005815611802972853,
-      "learning_rate": 0.00019999969916654868,
-      "loss": 46.0,
-      "step": 10222
-    },
-    {
-      "epoch": 0.7816197411931112,
-      "grad_norm": 0.0032430393621325493,
-      "learning_rate": 0.0001999996991076282,
-      "loss": 46.0,
-      "step": 10223
-    },
-    {
-      "epoch": 0.7816961981765009,
-      "grad_norm": 0.0007613180205225945,
-      "learning_rate": 0.00019999969904870192,
-      "loss": 46.0,
-      "step": 10224
-    },
-    {
-      "epoch": 0.7817726551598907,
-      "grad_norm": 0.005472178105264902,
-      "learning_rate": 0.0001999996989897699,
-      "loss": 46.0,
-      "step": 10225
-    },
-    {
-      "epoch": 0.7818491121432803,
-      "grad_norm": 0.006193201057612896,
-      "learning_rate": 0.0001999996989308321,
-      "loss": 46.0,
-      "step": 10226
-    },
-    {
-      "epoch": 0.7819255691266701,
-      "grad_norm": 0.004918051417917013,
-      "learning_rate": 0.00019999969887188855,
-      "loss": 46.0,
-      "step": 10227
-    },
-    {
-      "epoch": 0.7820020261100599,
-      "grad_norm": 0.0014459715457633138,
-      "learning_rate": 0.0001999996988129392,
-      "loss": 46.0,
-      "step": 10228
-    },
-    {
-      "epoch": 0.7820784830934495,
-      "grad_norm": 0.0005648498190566897,
-      "learning_rate": 0.00019999969875398408,
-      "loss": 46.0,
-      "step": 10229
-    },
-    {
-      "epoch": 0.7821549400768393,
-      "grad_norm": 0.0006629560375586152,
-      "learning_rate": 0.00019999969869502322,
-      "loss": 46.0,
-      "step": 10230
-    },
-    {
-      "epoch": 0.782231397060229,
-      "grad_norm": 0.0026891680900007486,
-      "learning_rate": 0.00019999969863605655,
-      "loss": 46.0,
-      "step": 10231
-    },
-    {
-      "epoch": 0.7823078540436187,
-      "grad_norm": 0.004870935343205929,
-      "learning_rate": 0.00019999969857708414,
-      "loss": 46.0,
-      "step": 10232
-    },
-    {
-      "epoch": 0.7823843110270084,
-      "grad_norm": 0.0006571028498001397,
-      "learning_rate": 0.00019999969851810595,
-      "loss": 46.0,
-      "step": 10233
-    },
-    {
-      "epoch": 0.7824607680103981,
-      "grad_norm": 0.0005809029680676758,
-      "learning_rate": 0.00019999969845912202,
-      "loss": 46.0,
-      "step": 10234
-    },
-    {
-      "epoch": 0.7825372249937879,
-      "grad_norm": 0.0005806274130009115,
-      "learning_rate": 0.00019999969840013226,
-      "loss": 46.0,
-      "step": 10235
-    },
-    {
-      "epoch": 0.7826136819771776,
-      "grad_norm": 0.0031394949182868004,
-      "learning_rate": 0.00019999969834113678,
-      "loss": 46.0,
-      "step": 10236
-    },
-    {
-      "epoch": 0.7826901389605673,
-      "grad_norm": 0.004411129746586084,
-      "learning_rate": 0.0001999996982821355,
-      "loss": 46.0,
-      "step": 10237
-    },
-    {
-      "epoch": 0.782766595943957,
-      "grad_norm": 0.0006739192176610231,
-      "learning_rate": 0.00019999969822312848,
-      "loss": 46.0,
-      "step": 10238
-    },
-    {
-      "epoch": 0.7828430529273468,
-      "grad_norm": 0.007350389380007982,
-      "learning_rate": 0.00019999969816411568,
-      "loss": 46.0,
-      "step": 10239
-    },
-    {
-      "epoch": 0.7829195099107364,
-      "grad_norm": 0.0009272333700209856,
-      "learning_rate": 0.0001999996981050971,
-      "loss": 46.0,
-      "step": 10240
-    },
-    {
-      "epoch": 0.7829959668941262,
-      "grad_norm": 0.001023614197038114,
-      "learning_rate": 0.00019999969804607276,
-      "loss": 46.0,
-      "step": 10241
-    },
-    {
-      "epoch": 0.783072423877516,
-      "grad_norm": 0.0007502685184590518,
-      "learning_rate": 0.00019999969798704264,
-      "loss": 46.0,
-      "step": 10242
-    },
-    {
-      "epoch": 0.7831488808609056,
-      "grad_norm": 0.0005459418753162026,
-      "learning_rate": 0.00019999969792800678,
-      "loss": 46.0,
-      "step": 10243
-    },
-    {
-      "epoch": 0.7832253378442954,
-      "grad_norm": 0.0034213298931717873,
-      "learning_rate": 0.00019999969786896514,
-      "loss": 46.0,
-      "step": 10244
-    },
-    {
-      "epoch": 0.783301794827685,
-      "grad_norm": 0.0033136524725705385,
-      "learning_rate": 0.00019999969780991773,
-      "loss": 46.0,
-      "step": 10245
-    },
-    {
-      "epoch": 0.7833782518110748,
-      "grad_norm": 0.0016369274817407131,
-      "learning_rate": 0.0001999996977508645,
-      "loss": 46.0,
-      "step": 10246
-    },
-    {
-      "epoch": 0.7834547087944645,
-      "grad_norm": 0.004078489728271961,
-      "learning_rate": 0.00019999969769180558,
-      "loss": 46.0,
-      "step": 10247
-    },
-    {
-      "epoch": 0.7835311657778542,
-      "grad_norm": 0.0018563922494649887,
-      "learning_rate": 0.00019999969763274082,
-      "loss": 46.0,
-      "step": 10248
-    },
-    {
-      "epoch": 0.783607622761244,
-      "grad_norm": 0.001281578792259097,
-      "learning_rate": 0.00019999969757367034,
-      "loss": 46.0,
-      "step": 10249
-    },
-    {
-      "epoch": 0.7836840797446337,
-      "grad_norm": 0.00502871535718441,
-      "learning_rate": 0.0001999996975145941,
-      "loss": 46.0,
-      "step": 10250
-    },
-    {
-      "epoch": 0.7837605367280234,
-      "grad_norm": 0.0013296377146616578,
-      "learning_rate": 0.00019999969745551204,
-      "loss": 46.0,
-      "step": 10251
-    },
-    {
-      "epoch": 0.7838369937114131,
-      "grad_norm": 0.02295924723148346,
-      "learning_rate": 0.00019999969739642424,
-      "loss": 46.0,
-      "step": 10252
-    },
-    {
-      "epoch": 0.7839134506948028,
-      "grad_norm": 0.0017064297571778297,
-      "learning_rate": 0.00019999969733733067,
-      "loss": 46.0,
-      "step": 10253
-    },
-    {
-      "epoch": 0.7839899076781925,
-      "grad_norm": 0.005058642942458391,
-      "learning_rate": 0.00019999969727823132,
-      "loss": 46.0,
-      "step": 10254
-    },
-    {
-      "epoch": 0.7840663646615823,
-      "grad_norm": 0.0006025419570505619,
-      "learning_rate": 0.0001999996972191262,
-      "loss": 46.0,
-      "step": 10255
-    },
-    {
-      "epoch": 0.784142821644972,
-      "grad_norm": 0.0011545703746378422,
-      "learning_rate": 0.0001999996971600153,
-      "loss": 46.0,
-      "step": 10256
-    },
-    {
-      "epoch": 0.7842192786283617,
-      "grad_norm": 0.002726361621171236,
-      "learning_rate": 0.00019999969710089865,
-      "loss": 46.0,
-      "step": 10257
-    },
-    {
-      "epoch": 0.7842957356117515,
-      "grad_norm": 0.004911541007459164,
-      "learning_rate": 0.00019999969704177623,
-      "loss": 46.0,
-      "step": 10258
-    },
-    {
-      "epoch": 0.7843721925951411,
-      "grad_norm": 0.0012584724463522434,
-      "learning_rate": 0.00019999969698264805,
-      "loss": 46.0,
-      "step": 10259
-    },
-    {
-      "epoch": 0.7844486495785309,
-      "grad_norm": 0.0026344831567257643,
-      "learning_rate": 0.0001999996969235141,
-      "loss": 46.0,
-      "step": 10260
-    },
-    {
-      "epoch": 0.7845251065619206,
-      "grad_norm": 0.005456394515931606,
-      "learning_rate": 0.00019999969686437433,
-      "loss": 46.0,
-      "step": 10261
-    },
-    {
-      "epoch": 0.7846015635453103,
-      "grad_norm": 0.0013098577037453651,
-      "learning_rate": 0.00019999969680522885,
-      "loss": 46.0,
-      "step": 10262
-    },
-    {
-      "epoch": 0.7846780205287001,
-      "grad_norm": 0.002115991897881031,
-      "learning_rate": 0.0001999996967460776,
-      "loss": 46.0,
-      "step": 10263
-    },
-    {
-      "epoch": 0.7847544775120897,
-      "grad_norm": 0.008396445773541927,
-      "learning_rate": 0.00019999969668692055,
-      "loss": 46.0,
-      "step": 10264
-    },
-    {
-      "epoch": 0.7848309344954795,
-      "grad_norm": 0.0008707360248081386,
-      "learning_rate": 0.00019999969662775775,
-      "loss": 46.0,
-      "step": 10265
-    },
-    {
-      "epoch": 0.7849073914788692,
-      "grad_norm": 0.003703216090798378,
-      "learning_rate": 0.00019999969656858918,
-      "loss": 46.0,
-      "step": 10266
-    },
-    {
-      "epoch": 0.7849838484622589,
-      "grad_norm": 0.00292229070328176,
-      "learning_rate": 0.0001999996965094148,
-      "loss": 46.0,
-      "step": 10267
-    },
-    {
-      "epoch": 0.7850603054456486,
-      "grad_norm": 0.0012993005802854896,
-      "learning_rate": 0.00019999969645023472,
-      "loss": 46.0,
-      "step": 10268
-    },
-    {
-      "epoch": 0.7851367624290384,
-      "grad_norm": 0.0010618198430165648,
-      "learning_rate": 0.00019999969639104882,
-      "loss": 46.0,
-      "step": 10269
-    },
-    {
-      "epoch": 0.785213219412428,
-      "grad_norm": 0.001083186361938715,
-      "learning_rate": 0.00019999969633185716,
-      "loss": 46.0,
-      "step": 10270
-    },
-    {
-      "epoch": 0.7852896763958178,
-      "grad_norm": 0.0007116859778761864,
-      "learning_rate": 0.00019999969627265975,
-      "loss": 46.0,
-      "step": 10271
-    },
-    {
-      "epoch": 0.7853661333792076,
-      "grad_norm": 0.0013811420649290085,
-      "learning_rate": 0.00019999969621345654,
-      "loss": 46.0,
-      "step": 10272
-    },
-    {
-      "epoch": 0.7854425903625972,
-      "grad_norm": 0.0006520661991089582,
-      "learning_rate": 0.00019999969615424758,
-      "loss": 46.0,
-      "step": 10273
-    },
-    {
-      "epoch": 0.785519047345987,
-      "grad_norm": 0.0034866214264184237,
-      "learning_rate": 0.00019999969609503287,
-      "loss": 46.0,
-      "step": 10274
-    },
-    {
-      "epoch": 0.7855955043293766,
-      "grad_norm": 0.0006679798243567348,
-      "learning_rate": 0.00019999969603581234,
-      "loss": 46.0,
-      "step": 10275
-    },
-    {
-      "epoch": 0.7856719613127664,
-      "grad_norm": 0.0013419750612229109,
-      "learning_rate": 0.0001999996959765861,
-      "loss": 46.0,
-      "step": 10276
-    },
-    {
-      "epoch": 0.7857484182961562,
-      "grad_norm": 0.0017901909304782748,
-      "learning_rate": 0.00019999969591735404,
-      "loss": 46.0,
-      "step": 10277
-    },
-    {
-      "epoch": 0.7858248752795458,
-      "grad_norm": 0.008774053305387497,
-      "learning_rate": 0.0001999996958581162,
-      "loss": 46.0,
-      "step": 10278
-    },
-    {
-      "epoch": 0.7859013322629356,
-      "grad_norm": 0.0010192327899858356,
-      "learning_rate": 0.00019999969579887267,
-      "loss": 46.0,
-      "step": 10279
-    },
-    {
-      "epoch": 0.7859777892463253,
-      "grad_norm": 0.0008977089892141521,
-      "learning_rate": 0.0001999996957396233,
-      "loss": 46.0,
-      "step": 10280
-    },
-    {
-      "epoch": 0.786054246229715,
-      "grad_norm": 0.0007017637835815549,
-      "learning_rate": 0.00019999969568036815,
-      "loss": 46.0,
-      "step": 10281
-    },
-    {
-      "epoch": 0.7861307032131047,
-      "grad_norm": 0.002642162377014756,
-      "learning_rate": 0.0001999996956211073,
-      "loss": 46.0,
-      "step": 10282
-    },
-    {
-      "epoch": 0.7862071601964944,
-      "grad_norm": 0.001508566434495151,
-      "learning_rate": 0.00019999969556184062,
-      "loss": 46.0,
-      "step": 10283
-    },
-    {
-      "epoch": 0.7862836171798842,
-      "grad_norm": 0.0007882867357693613,
-      "learning_rate": 0.0001999996955025682,
-      "loss": 46.0,
-      "step": 10284
-    },
-    {
-      "epoch": 0.7863600741632739,
-      "grad_norm": 0.0009465863695368171,
-      "learning_rate": 0.00019999969544329,
-      "loss": 46.0,
-      "step": 10285
-    },
-    {
-      "epoch": 0.7864365311466636,
-      "grad_norm": 0.005636065267026424,
-      "learning_rate": 0.00019999969538400601,
-      "loss": 46.0,
-      "step": 10286
-    },
-    {
-      "epoch": 0.7865129881300533,
-      "grad_norm": 0.0009969661477953196,
-      "learning_rate": 0.0001999996953247163,
-      "loss": 46.0,
-      "step": 10287
-    },
-    {
-      "epoch": 0.7865894451134431,
-      "grad_norm": 0.0012537481961771846,
-      "learning_rate": 0.0001999996952654208,
-      "loss": 46.0,
-      "step": 10288
-    },
-    {
-      "epoch": 0.7866659020968327,
-      "grad_norm": 0.0007035795715637505,
-      "learning_rate": 0.00019999969520611953,
-      "loss": 46.0,
-      "step": 10289
-    },
-    {
-      "epoch": 0.7867423590802225,
-      "grad_norm": 0.0008786415564827621,
-      "learning_rate": 0.00019999969514681248,
-      "loss": 46.0,
-      "step": 10290
-    },
-    {
-      "epoch": 0.7868188160636123,
-      "grad_norm": 0.0008917577797546983,
-      "learning_rate": 0.00019999969508749965,
-      "loss": 46.0,
-      "step": 10291
-    },
-    {
-      "epoch": 0.7868952730470019,
-      "grad_norm": 0.009642994031310081,
-      "learning_rate": 0.00019999969502818108,
-      "loss": 46.0,
-      "step": 10292
-    },
-    {
-      "epoch": 0.7869717300303917,
-      "grad_norm": 0.0035529329907149076,
-      "learning_rate": 0.00019999969496885673,
-      "loss": 46.0,
-      "step": 10293
-    },
-    {
-      "epoch": 0.7870481870137813,
-      "grad_norm": 0.0007940498762764037,
-      "learning_rate": 0.00019999969490952662,
-      "loss": 46.0,
-      "step": 10294
-    },
-    {
-      "epoch": 0.7871246439971711,
-      "grad_norm": 0.002146205399185419,
-      "learning_rate": 0.00019999969485019073,
-      "loss": 46.0,
-      "step": 10295
-    },
-    {
-      "epoch": 0.7872011009805608,
-      "grad_norm": 0.0012229300336912274,
-      "learning_rate": 0.00019999969479084906,
-      "loss": 46.0,
-      "step": 10296
-    },
-    {
-      "epoch": 0.7872775579639505,
-      "grad_norm": 0.0012784464051946998,
-      "learning_rate": 0.00019999969473150162,
-      "loss": 46.0,
-      "step": 10297
-    },
-    {
-      "epoch": 0.7873540149473403,
-      "grad_norm": 0.00109715829603374,
-      "learning_rate": 0.00019999969467214844,
-      "loss": 46.0,
-      "step": 10298
-    },
-    {
-      "epoch": 0.78743047193073,
-      "grad_norm": 0.004224192816764116,
-      "learning_rate": 0.00019999969461278948,
-      "loss": 46.0,
-      "step": 10299
-    },
-    {
-      "epoch": 0.7875069289141197,
-      "grad_norm": 0.0013706808676943183,
-      "learning_rate": 0.00019999969455342472,
-      "loss": 46.0,
-      "step": 10300
-    },
-    {
-      "epoch": 0.7875833858975094,
-      "grad_norm": 0.0010997304925695062,
-      "learning_rate": 0.00019999969449405422,
-      "loss": 46.0,
-      "step": 10301
-    },
-    {
-      "epoch": 0.7876598428808992,
-      "grad_norm": 0.003662483301013708,
-      "learning_rate": 0.00019999969443467794,
-      "loss": 46.0,
-      "step": 10302
-    },
-    {
-      "epoch": 0.7877362998642888,
-      "grad_norm": 0.0007888961117714643,
-      "learning_rate": 0.0001999996943752959,
-      "loss": 46.0,
-      "step": 10303
-    },
-    {
-      "epoch": 0.7878127568476786,
-      "grad_norm": 0.0008473096531815827,
-      "learning_rate": 0.00019999969431590807,
-      "loss": 46.0,
-      "step": 10304
-    },
-    {
-      "epoch": 0.7878892138310682,
-      "grad_norm": 0.0004846773808822036,
-      "learning_rate": 0.0001999996942565145,
-      "loss": 46.0,
-      "step": 10305
-    },
-    {
-      "epoch": 0.787965670814458,
-      "grad_norm": 0.0019808681681752205,
-      "learning_rate": 0.00019999969419711512,
-      "loss": 46.0,
-      "step": 10306
-    },
-    {
-      "epoch": 0.7880421277978478,
-      "grad_norm": 0.0046212985180318356,
-      "learning_rate": 0.00019999969413771003,
-      "loss": 46.0,
-      "step": 10307
-    },
-    {
-      "epoch": 0.7881185847812374,
-      "grad_norm": 0.0014043668052181602,
-      "learning_rate": 0.00019999969407829911,
-      "loss": 46.0,
-      "step": 10308
-    },
-    {
-      "epoch": 0.7881950417646272,
-      "grad_norm": 0.0004871142446063459,
-      "learning_rate": 0.00019999969401888245,
-      "loss": 46.0,
-      "step": 10309
-    },
-    {
-      "epoch": 0.7882714987480169,
-      "grad_norm": 0.0039797076024115086,
-      "learning_rate": 0.00019999969395946004,
-      "loss": 46.0,
-      "step": 10310
-    },
-    {
-      "epoch": 0.7883479557314066,
-      "grad_norm": 0.0009994247229769826,
-      "learning_rate": 0.0001999996939000318,
-      "loss": 46.0,
-      "step": 10311
-    },
-    {
-      "epoch": 0.7884244127147964,
-      "grad_norm": 0.0012315146159380674,
-      "learning_rate": 0.00019999969384059784,
-      "loss": 46.0,
-      "step": 10312
-    },
-    {
-      "epoch": 0.7885008696981861,
-      "grad_norm": 0.0006372160860337317,
-      "learning_rate": 0.00019999969378115811,
-      "loss": 46.0,
-      "step": 10313
-    },
-    {
-      "epoch": 0.7885773266815758,
-      "grad_norm": 0.0007672370411455631,
-      "learning_rate": 0.0001999996937217126,
-      "loss": 46.0,
-      "step": 10314
-    },
-    {
-      "epoch": 0.7886537836649655,
-      "grad_norm": 0.0013643394922837615,
-      "learning_rate": 0.00019999969366226133,
-      "loss": 46.0,
-      "step": 10315
-    },
-    {
-      "epoch": 0.7887302406483552,
-      "grad_norm": 0.0003806533932220191,
-      "learning_rate": 0.00019999969360280425,
-      "loss": 46.0,
-      "step": 10316
-    },
-    {
-      "epoch": 0.7888066976317449,
-      "grad_norm": 0.002345460932701826,
-      "learning_rate": 0.00019999969354334146,
-      "loss": 46.0,
-      "step": 10317
-    },
-    {
-      "epoch": 0.7888831546151347,
-      "grad_norm": 0.0020158749539405107,
-      "learning_rate": 0.00019999969348387286,
-      "loss": 46.0,
-      "step": 10318
-    },
-    {
-      "epoch": 0.7889596115985243,
-      "grad_norm": 0.0008241494651883841,
-      "learning_rate": 0.00019999969342439852,
-      "loss": 46.0,
-      "step": 10319
-    },
-    {
-      "epoch": 0.7890360685819141,
-      "grad_norm": 0.005219350568950176,
-      "learning_rate": 0.00019999969336491837,
-      "loss": 46.0,
-      "step": 10320
-    },
-    {
-      "epoch": 0.7891125255653039,
-      "grad_norm": 0.002630602801218629,
-      "learning_rate": 0.00019999969330543248,
-      "loss": 46.0,
-      "step": 10321
-    },
-    {
-      "epoch": 0.7891889825486935,
-      "grad_norm": 0.0018792249029502273,
-      "learning_rate": 0.00019999969324594082,
-      "loss": 46.0,
-      "step": 10322
-    },
-    {
-      "epoch": 0.7892654395320833,
-      "grad_norm": 0.007390186656266451,
-      "learning_rate": 0.00019999969318644338,
-      "loss": 46.0,
-      "step": 10323
-    },
-    {
-      "epoch": 0.7893418965154729,
-      "grad_norm": 0.0010319760767742991,
-      "learning_rate": 0.00019999969312694017,
-      "loss": 46.0,
-      "step": 10324
-    },
-    {
-      "epoch": 0.7894183534988627,
-      "grad_norm": 0.002190877916291356,
-      "learning_rate": 0.0001999996930674312,
-      "loss": 46.0,
-      "step": 10325
-    },
-    {
-      "epoch": 0.7894948104822525,
-      "grad_norm": 0.0005150993238203228,
-      "learning_rate": 0.00019999969300791645,
-      "loss": 46.0,
-      "step": 10326
-    },
-    {
-      "epoch": 0.7895712674656421,
-      "grad_norm": 0.0010977080091834068,
-      "learning_rate": 0.00019999969294839595,
-      "loss": 46.0,
-      "step": 10327
-    },
-    {
-      "epoch": 0.7896477244490319,
-      "grad_norm": 0.007419740781188011,
-      "learning_rate": 0.00019999969288886967,
-      "loss": 46.0,
-      "step": 10328
-    },
-    {
-      "epoch": 0.7897241814324216,
-      "grad_norm": 0.0008296669693663716,
-      "learning_rate": 0.00019999969282933762,
-      "loss": 46.0,
-      "step": 10329
-    },
-    {
-      "epoch": 0.7898006384158113,
-      "grad_norm": 0.006710233632475138,
-      "learning_rate": 0.0001999996927697998,
-      "loss": 46.0,
-      "step": 10330
-    },
-    {
-      "epoch": 0.789877095399201,
-      "grad_norm": 0.008683396503329277,
-      "learning_rate": 0.0001999996927102562,
-      "loss": 46.0,
-      "step": 10331
-    },
-    {
-      "epoch": 0.7899535523825908,
-      "grad_norm": 0.0016000174218788743,
-      "learning_rate": 0.00019999969265070683,
-      "loss": 46.0,
-      "step": 10332
-    },
-    {
-      "epoch": 0.7900300093659804,
-      "grad_norm": 0.0019550633151084185,
-      "learning_rate": 0.00019999969259115171,
-      "loss": 46.0,
-      "step": 10333
-    },
-    {
-      "epoch": 0.7901064663493702,
-      "grad_norm": 0.0013147916179150343,
-      "learning_rate": 0.0001999996925315908,
-      "loss": 46.0,
-      "step": 10334
-    },
-    {
-      "epoch": 0.7901829233327599,
-      "grad_norm": 0.0022051376290619373,
-      "learning_rate": 0.00019999969247202413,
-      "loss": 46.0,
-      "step": 10335
-    },
-    {
-      "epoch": 0.7902593803161496,
-      "grad_norm": 0.0014337224420160055,
-      "learning_rate": 0.0001999996924124517,
-      "loss": 46.0,
-      "step": 10336
-    },
-    {
-      "epoch": 0.7903358372995394,
-      "grad_norm": 0.004368383903056383,
-      "learning_rate": 0.00019999969235287349,
-      "loss": 46.0,
-      "step": 10337
-    },
-    {
-      "epoch": 0.790412294282929,
-      "grad_norm": 0.007982072420418262,
-      "learning_rate": 0.00019999969229328953,
-      "loss": 46.0,
-      "step": 10338
-    },
-    {
-      "epoch": 0.7904887512663188,
-      "grad_norm": 0.0023622328881174326,
-      "learning_rate": 0.0001999996922336998,
-      "loss": 46.0,
-      "step": 10339
-    },
-    {
-      "epoch": 0.7905652082497086,
-      "grad_norm": 0.006284235045313835,
-      "learning_rate": 0.00019999969217410427,
-      "loss": 46.0,
-      "step": 10340
-    },
-    {
-      "epoch": 0.7906416652330982,
-      "grad_norm": 0.019389530643820763,
-      "learning_rate": 0.00019999969211450296,
-      "loss": 46.0,
-      "step": 10341
-    },
-    {
-      "epoch": 0.790718122216488,
-      "grad_norm": 0.0011169412173330784,
-      "learning_rate": 0.00019999969205489591,
-      "loss": 46.0,
-      "step": 10342
-    },
-    {
-      "epoch": 0.7907945791998777,
-      "grad_norm": 0.0009455361287109554,
-      "learning_rate": 0.0001999996919952831,
-      "loss": 46.0,
-      "step": 10343
-    },
-    {
-      "epoch": 0.7908710361832674,
-      "grad_norm": 0.0007107259007170796,
-      "learning_rate": 0.0001999996919356645,
-      "loss": 46.0,
-      "step": 10344
-    },
-    {
-      "epoch": 0.7909474931666571,
-      "grad_norm": 0.0011275571305304766,
-      "learning_rate": 0.00019999969187604015,
-      "loss": 46.0,
-      "step": 10345
-    },
-    {
-      "epoch": 0.7910239501500468,
-      "grad_norm": 0.0015255759935826063,
-      "learning_rate": 0.00019999969181641,
-      "loss": 46.0,
-      "step": 10346
-    },
-    {
-      "epoch": 0.7911004071334365,
-      "grad_norm": 0.003709301119670272,
-      "learning_rate": 0.0001999996917567741,
-      "loss": 46.0,
-      "step": 10347
-    },
-    {
-      "epoch": 0.7911768641168263,
-      "grad_norm": 0.0009050164953805506,
-      "learning_rate": 0.00019999969169713243,
-      "loss": 46.0,
-      "step": 10348
-    },
-    {
-      "epoch": 0.791253321100216,
-      "grad_norm": 0.0017321555642411113,
-      "learning_rate": 0.000199999691637485,
-      "loss": 46.0,
-      "step": 10349
-    },
-    {
-      "epoch": 0.7913297780836057,
-      "grad_norm": 0.005666300188750029,
-      "learning_rate": 0.00019999969157783178,
-      "loss": 46.0,
-      "step": 10350
-    },
-    {
-      "epoch": 0.7914062350669955,
-      "grad_norm": 0.0011831017909571528,
-      "learning_rate": 0.00019999969151817282,
-      "loss": 46.0,
-      "step": 10351
-    },
-    {
-      "epoch": 0.7914826920503851,
-      "grad_norm": 0.0010023321956396103,
-      "learning_rate": 0.00019999969145850807,
-      "loss": 46.0,
-      "step": 10352
-    },
-    {
-      "epoch": 0.7915591490337749,
-      "grad_norm": 0.0016831184038892388,
-      "learning_rate": 0.0001999996913988375,
-      "loss": 46.0,
-      "step": 10353
-    },
-    {
-      "epoch": 0.7916356060171645,
-      "grad_norm": 0.0014527245657518506,
-      "learning_rate": 0.00019999969133916126,
-      "loss": 46.0,
-      "step": 10354
-    },
-    {
-      "epoch": 0.7917120630005543,
-      "grad_norm": 0.008887079544365406,
-      "learning_rate": 0.00019999969127947918,
-      "loss": 46.0,
-      "step": 10355
-    },
-    {
-      "epoch": 0.7917885199839441,
-      "grad_norm": 0.0027604512870311737,
-      "learning_rate": 0.00019999969121979136,
-      "loss": 46.0,
-      "step": 10356
-    },
-    {
-      "epoch": 0.7918649769673337,
-      "grad_norm": 0.003108508652076125,
-      "learning_rate": 0.0001999996911600978,
-      "loss": 46.0,
-      "step": 10357
-    },
-    {
-      "epoch": 0.7919414339507235,
-      "grad_norm": 0.0009689987055025995,
-      "learning_rate": 0.0001999996911003984,
-      "loss": 46.0,
-      "step": 10358
-    },
-    {
-      "epoch": 0.7920178909341132,
-      "grad_norm": 0.0038455866742879152,
-      "learning_rate": 0.00019999969104069328,
-      "loss": 46.0,
-      "step": 10359
-    },
-    {
-      "epoch": 0.7920943479175029,
-      "grad_norm": 0.0010078534251078963,
-      "learning_rate": 0.00019999969098098236,
-      "loss": 46.0,
-      "step": 10360
-    },
-    {
-      "epoch": 0.7921708049008926,
-      "grad_norm": 0.0011107171885669231,
-      "learning_rate": 0.0001999996909212657,
-      "loss": 46.0,
-      "step": 10361
-    },
-    {
-      "epoch": 0.7922472618842824,
-      "grad_norm": 0.0013087345287203789,
-      "learning_rate": 0.00019999969086154323,
-      "loss": 46.0,
-      "step": 10362
-    },
-    {
-      "epoch": 0.792323718867672,
-      "grad_norm": 0.0017957051750272512,
-      "learning_rate": 0.00019999969080181505,
-      "loss": 46.0,
-      "step": 10363
-    },
-    {
-      "epoch": 0.7924001758510618,
-      "grad_norm": 0.0013963307719677687,
-      "learning_rate": 0.00019999969074208104,
-      "loss": 46.0,
-      "step": 10364
-    },
-    {
-      "epoch": 0.7924766328344515,
-      "grad_norm": 0.0005416147178038955,
-      "learning_rate": 0.0001999996906823413,
-      "loss": 46.0,
-      "step": 10365
-    },
-    {
-      "epoch": 0.7925530898178412,
-      "grad_norm": 0.00043462254689075053,
-      "learning_rate": 0.00019999969062259578,
-      "loss": 46.0,
-      "step": 10366
-    },
-    {
-      "epoch": 0.792629546801231,
-      "grad_norm": 0.0008399094804190099,
-      "learning_rate": 0.00019999969056284448,
-      "loss": 46.0,
-      "step": 10367
-    },
-    {
-      "epoch": 0.7927060037846206,
-      "grad_norm": 0.005973747931420803,
-      "learning_rate": 0.00019999969050308743,
-      "loss": 46.0,
-      "step": 10368
-    },
-    {
-      "epoch": 0.7927824607680104,
-      "grad_norm": 0.0009809896582737565,
-      "learning_rate": 0.0001999996904433246,
-      "loss": 46.0,
-      "step": 10369
-    },
-    {
-      "epoch": 0.7928589177514002,
-      "grad_norm": 0.002539858454838395,
-      "learning_rate": 0.000199999690383556,
-      "loss": 46.0,
-      "step": 10370
-    },
-    {
-      "epoch": 0.7929353747347898,
-      "grad_norm": 0.0010078662307932973,
-      "learning_rate": 0.00019999969032378164,
-      "loss": 46.0,
-      "step": 10371
-    },
-    {
-      "epoch": 0.7930118317181796,
-      "grad_norm": 0.003399464301764965,
-      "learning_rate": 0.0001999996902640015,
-      "loss": 46.0,
-      "step": 10372
-    },
-    {
-      "epoch": 0.7930882887015693,
-      "grad_norm": 0.0008202534518204629,
-      "learning_rate": 0.00019999969020421558,
-      "loss": 46.0,
-      "step": 10373
-    },
-    {
-      "epoch": 0.793164745684959,
-      "grad_norm": 0.0017233005492016673,
-      "learning_rate": 0.00019999969014442392,
-      "loss": 46.0,
-      "step": 10374
-    },
-    {
-      "epoch": 0.7932412026683487,
-      "grad_norm": 0.010492541827261448,
-      "learning_rate": 0.00019999969008462648,
-      "loss": 46.0,
-      "step": 10375
-    },
-    {
-      "epoch": 0.7933176596517384,
-      "grad_norm": 0.0009122971096076071,
-      "learning_rate": 0.00019999969002482324,
-      "loss": 46.0,
-      "step": 10376
-    },
-    {
-      "epoch": 0.7933941166351282,
-      "grad_norm": 0.0024784451816231012,
-      "learning_rate": 0.0001999996899650143,
-      "loss": 46.0,
-      "step": 10377
-    },
-    {
-      "epoch": 0.7934705736185179,
-      "grad_norm": 0.004751408938318491,
-      "learning_rate": 0.0001999996899051995,
-      "loss": 46.0,
-      "step": 10378
-    },
-    {
-      "epoch": 0.7935470306019076,
-      "grad_norm": 0.0006340565741993487,
-      "learning_rate": 0.000199999689845379,
-      "loss": 46.0,
-      "step": 10379
-    },
-    {
-      "epoch": 0.7936234875852973,
-      "grad_norm": 0.0019558495841920376,
-      "learning_rate": 0.00019999968978555267,
-      "loss": 46.0,
-      "step": 10380
-    },
-    {
-      "epoch": 0.7936999445686871,
-      "grad_norm": 0.006155429407954216,
-      "learning_rate": 0.00019999968972572062,
-      "loss": 46.0,
-      "step": 10381
-    },
-    {
-      "epoch": 0.7937764015520767,
-      "grad_norm": 0.003428464289754629,
-      "learning_rate": 0.0001999996896658828,
-      "loss": 46.0,
-      "step": 10382
-    },
-    {
-      "epoch": 0.7938528585354665,
-      "grad_norm": 0.0026111796032637358,
-      "learning_rate": 0.0001999996896060392,
-      "loss": 46.0,
-      "step": 10383
-    },
-    {
-      "epoch": 0.7939293155188561,
-      "grad_norm": 0.000979086384177208,
-      "learning_rate": 0.00019999968954618983,
-      "loss": 46.0,
-      "step": 10384
-    },
-    {
-      "epoch": 0.7940057725022459,
-      "grad_norm": 0.0015180292539298534,
-      "learning_rate": 0.0001999996894863347,
-      "loss": 46.0,
-      "step": 10385
-    },
-    {
-      "epoch": 0.7940822294856357,
-      "grad_norm": 0.004844078794121742,
-      "learning_rate": 0.00019999968942647378,
-      "loss": 46.0,
-      "step": 10386
-    },
-    {
-      "epoch": 0.7941586864690253,
-      "grad_norm": 0.0005243421765044332,
-      "learning_rate": 0.0001999996893666071,
-      "loss": 46.0,
-      "step": 10387
-    },
-    {
-      "epoch": 0.7942351434524151,
-      "grad_norm": 0.0013097668997943401,
-      "learning_rate": 0.00019999968930673465,
-      "loss": 46.0,
-      "step": 10388
-    },
-    {
-      "epoch": 0.7943116004358048,
-      "grad_norm": 0.0018528528744354844,
-      "learning_rate": 0.00019999968924685644,
-      "loss": 46.0,
-      "step": 10389
-    },
-    {
-      "epoch": 0.7943880574191945,
-      "grad_norm": 0.0007080174400471151,
-      "learning_rate": 0.00019999968918697243,
-      "loss": 46.0,
-      "step": 10390
-    },
-    {
-      "epoch": 0.7944645144025843,
-      "grad_norm": 0.0022853994742035866,
-      "learning_rate": 0.0001999996891270827,
-      "loss": 46.0,
-      "step": 10391
-    },
-    {
-      "epoch": 0.794540971385974,
-      "grad_norm": 0.000591819582041353,
-      "learning_rate": 0.00019999968906718715,
-      "loss": 46.0,
-      "step": 10392
-    },
-    {
-      "epoch": 0.7946174283693637,
-      "grad_norm": 0.00037622853415086865,
-      "learning_rate": 0.00019999968900728584,
-      "loss": 46.0,
-      "step": 10393
-    },
-    {
-      "epoch": 0.7946938853527534,
-      "grad_norm": 0.0004591784963849932,
-      "learning_rate": 0.0001999996889473788,
-      "loss": 46.0,
-      "step": 10394
-    },
-    {
-      "epoch": 0.7947703423361431,
-      "grad_norm": 0.001266356324777007,
-      "learning_rate": 0.00019999968888746597,
-      "loss": 46.0,
-      "step": 10395
-    },
-    {
-      "epoch": 0.7948467993195328,
-      "grad_norm": 0.001246437313966453,
-      "learning_rate": 0.00019999968882754735,
-      "loss": 46.0,
-      "step": 10396
-    },
-    {
-      "epoch": 0.7949232563029226,
-      "grad_norm": 0.0015925847692415118,
-      "learning_rate": 0.00019999968876762298,
-      "loss": 46.0,
-      "step": 10397
-    },
-    {
-      "epoch": 0.7949997132863122,
-      "grad_norm": 0.0017261621542274952,
-      "learning_rate": 0.00019999968870769284,
-      "loss": 46.0,
-      "step": 10398
-    },
-    {
-      "epoch": 0.795076170269702,
-      "grad_norm": 0.001677903113886714,
-      "learning_rate": 0.00019999968864775692,
-      "loss": 46.0,
-      "step": 10399
-    },
-    {
-      "epoch": 0.7951526272530918,
-      "grad_norm": 0.0011187755735591054,
-      "learning_rate": 0.00019999968858781526,
-      "loss": 46.0,
-      "step": 10400
-    },
-    {
-      "epoch": 0.7952290842364814,
-      "grad_norm": 0.0026313646230846643,
-      "learning_rate": 0.0001999996885278678,
-      "loss": 46.0,
-      "step": 10401
-    },
-    {
-      "epoch": 0.7953055412198712,
-      "grad_norm": 0.0035825709346681833,
-      "learning_rate": 0.0001999996884679146,
-      "loss": 46.0,
-      "step": 10402
-    },
-    {
-      "epoch": 0.7953819982032609,
-      "grad_norm": 0.0006700589437969029,
-      "learning_rate": 0.00019999968840795558,
-      "loss": 46.0,
-      "step": 10403
-    },
-    {
-      "epoch": 0.7954584551866506,
-      "grad_norm": 0.0045277769677340984,
-      "learning_rate": 0.00019999968834799082,
-      "loss": 46.0,
-      "step": 10404
-    },
-    {
-      "epoch": 0.7955349121700404,
-      "grad_norm": 0.001390729914419353,
-      "learning_rate": 0.00019999968828802032,
-      "loss": 46.0,
-      "step": 10405
-    },
-    {
-      "epoch": 0.79561136915343,
-      "grad_norm": 0.001328416052274406,
-      "learning_rate": 0.000199999688228044,
-      "loss": 46.0,
-      "step": 10406
-    },
-    {
-      "epoch": 0.7956878261368198,
-      "grad_norm": 0.0007779764127917588,
-      "learning_rate": 0.00019999968816806194,
-      "loss": 46.0,
-      "step": 10407
-    },
-    {
-      "epoch": 0.7957642831202095,
-      "grad_norm": 0.0014895581407472491,
-      "learning_rate": 0.0001999996881080741,
-      "loss": 46.0,
-      "step": 10408
-    },
-    {
-      "epoch": 0.7958407401035992,
-      "grad_norm": 0.0009545899811200798,
-      "learning_rate": 0.0001999996880480805,
-      "loss": 46.0,
-      "step": 10409
-    },
-    {
-      "epoch": 0.7959171970869889,
-      "grad_norm": 0.00044253855594433844,
-      "learning_rate": 0.0001999996879880811,
-      "loss": 46.0,
-      "step": 10410
-    },
-    {
-      "epoch": 0.7959936540703787,
-      "grad_norm": 0.0026480432134121656,
-      "learning_rate": 0.00019999968792807596,
-      "loss": 46.0,
-      "step": 10411
-    },
-    {
-      "epoch": 0.7960701110537683,
-      "grad_norm": 0.0008752365247346461,
-      "learning_rate": 0.00019999968786806505,
-      "loss": 46.0,
-      "step": 10412
-    },
-    {
-      "epoch": 0.7961465680371581,
-      "grad_norm": 0.0005672427942045033,
-      "learning_rate": 0.00019999968780804836,
-      "loss": 46.0,
-      "step": 10413
-    },
-    {
-      "epoch": 0.7962230250205478,
-      "grad_norm": 0.0009923962643370032,
-      "learning_rate": 0.00019999968774802592,
-      "loss": 46.0,
-      "step": 10414
-    },
-    {
-      "epoch": 0.7962994820039375,
-      "grad_norm": 0.0012087089708074927,
-      "learning_rate": 0.00019999968768799771,
-      "loss": 46.0,
-      "step": 10415
-    },
-    {
-      "epoch": 0.7963759389873273,
-      "grad_norm": 0.004050011280924082,
-      "learning_rate": 0.00019999968762796368,
-      "loss": 46.0,
-      "step": 10416
-    },
-    {
-      "epoch": 0.7964523959707169,
-      "grad_norm": 0.0016919528134167194,
-      "learning_rate": 0.00019999968756792392,
-      "loss": 46.0,
-      "step": 10417
-    },
-    {
-      "epoch": 0.7965288529541067,
-      "grad_norm": 0.0013214662903919816,
-      "learning_rate": 0.00019999968750787842,
-      "loss": 46.0,
-      "step": 10418
-    },
-    {
-      "epoch": 0.7966053099374965,
-      "grad_norm": 0.002011113567277789,
-      "learning_rate": 0.00019999968744782712,
-      "loss": 46.0,
-      "step": 10419
-    },
-    {
-      "epoch": 0.7966817669208861,
-      "grad_norm": 0.003301904071122408,
-      "learning_rate": 0.00019999968738777004,
-      "loss": 46.0,
-      "step": 10420
-    },
-    {
-      "epoch": 0.7967582239042759,
-      "grad_norm": 0.0023058359511196613,
-      "learning_rate": 0.0001999996873277072,
-      "loss": 46.0,
-      "step": 10421
-    },
-    {
-      "epoch": 0.7968346808876656,
-      "grad_norm": 0.00020906612917315215,
-      "learning_rate": 0.00019999968726763857,
-      "loss": 46.0,
-      "step": 10422
-    },
-    {
-      "epoch": 0.7969111378710553,
-      "grad_norm": 0.004945096559822559,
-      "learning_rate": 0.00019999968720756418,
-      "loss": 46.0,
-      "step": 10423
-    },
-    {
-      "epoch": 0.796987594854445,
-      "grad_norm": 0.0030127139762043953,
-      "learning_rate": 0.00019999968714748406,
-      "loss": 46.0,
-      "step": 10424
-    },
-    {
-      "epoch": 0.7970640518378347,
-      "grad_norm": 0.00133426277898252,
-      "learning_rate": 0.00019999968708739815,
-      "loss": 46.0,
-      "step": 10425
-    },
-    {
-      "epoch": 0.7971405088212244,
-      "grad_norm": 0.0004186853184364736,
-      "learning_rate": 0.00019999968702730646,
-      "loss": 46.0,
-      "step": 10426
-    },
-    {
-      "epoch": 0.7972169658046142,
-      "grad_norm": 0.0022164955735206604,
-      "learning_rate": 0.000199999686967209,
-      "loss": 46.0,
-      "step": 10427
-    },
-    {
-      "epoch": 0.7972934227880039,
-      "grad_norm": 0.0007602505502291024,
-      "learning_rate": 0.0001999996869071058,
-      "loss": 46.0,
-      "step": 10428
-    },
-    {
-      "epoch": 0.7973698797713936,
-      "grad_norm": 0.0011165005853399634,
-      "learning_rate": 0.00019999968684699678,
-      "loss": 46.0,
-      "step": 10429
-    },
-    {
-      "epoch": 0.7974463367547834,
-      "grad_norm": 0.001903340220451355,
-      "learning_rate": 0.00019999968678688202,
-      "loss": 46.0,
-      "step": 10430
-    },
-    {
-      "epoch": 0.797522793738173,
-      "grad_norm": 0.0011740203481167555,
-      "learning_rate": 0.00019999968672676147,
-      "loss": 46.0,
-      "step": 10431
-    },
-    {
-      "epoch": 0.7975992507215628,
-      "grad_norm": 0.003760506631806493,
-      "learning_rate": 0.00019999968666663517,
-      "loss": 46.0,
-      "step": 10432
-    },
-    {
-      "epoch": 0.7976757077049526,
-      "grad_norm": 0.0007388813537545502,
-      "learning_rate": 0.0001999996866065031,
-      "loss": 46.0,
-      "step": 10433
-    },
-    {
-      "epoch": 0.7977521646883422,
-      "grad_norm": 0.000699860043823719,
-      "learning_rate": 0.00019999968654636524,
-      "loss": 46.0,
-      "step": 10434
-    },
-    {
-      "epoch": 0.797828621671732,
-      "grad_norm": 0.0009509455412626266,
-      "learning_rate": 0.00019999968648622165,
-      "loss": 46.0,
-      "step": 10435
-    },
-    {
-      "epoch": 0.7979050786551216,
-      "grad_norm": 0.0013718181289732456,
-      "learning_rate": 0.00019999968642607225,
-      "loss": 46.0,
-      "step": 10436
-    },
-    {
-      "epoch": 0.7979815356385114,
-      "grad_norm": 0.0018047906924039125,
-      "learning_rate": 0.0001999996863659171,
-      "loss": 46.0,
-      "step": 10437
-    },
-    {
-      "epoch": 0.7980579926219011,
-      "grad_norm": 0.000929017725866288,
-      "learning_rate": 0.0001999996863057562,
-      "loss": 46.0,
-      "step": 10438
-    },
-    {
-      "epoch": 0.7981344496052908,
-      "grad_norm": 0.007691532839089632,
-      "learning_rate": 0.0001999996862455895,
-      "loss": 46.0,
-      "step": 10439
-    },
-    {
-      "epoch": 0.7982109065886805,
-      "grad_norm": 0.0008329108823090792,
-      "learning_rate": 0.00019999968618541705,
-      "loss": 46.0,
-      "step": 10440
-    },
-    {
-      "epoch": 0.7982873635720703,
-      "grad_norm": 0.0005319429328665137,
-      "learning_rate": 0.0001999996861252388,
-      "loss": 46.0,
-      "step": 10441
-    },
-    {
-      "epoch": 0.79836382055546,
-      "grad_norm": 0.0009459342109039426,
-      "learning_rate": 0.0001999996860650548,
-      "loss": 46.0,
-      "step": 10442
-    },
-    {
-      "epoch": 0.7984402775388497,
-      "grad_norm": 0.0009394192602485418,
-      "learning_rate": 0.00019999968600486505,
-      "loss": 46.0,
-      "step": 10443
-    },
-    {
-      "epoch": 0.7985167345222395,
-      "grad_norm": 0.002740062540397048,
-      "learning_rate": 0.00019999968594466952,
-      "loss": 46.0,
-      "step": 10444
-    },
-    {
-      "epoch": 0.7985931915056291,
-      "grad_norm": 0.001409469055943191,
-      "learning_rate": 0.0001999996858844682,
-      "loss": 46.0,
-      "step": 10445
-    },
-    {
-      "epoch": 0.7986696484890189,
-      "grad_norm": 0.004490251187235117,
-      "learning_rate": 0.00019999968582426112,
-      "loss": 46.0,
-      "step": 10446
-    },
-    {
-      "epoch": 0.7987461054724085,
-      "grad_norm": 0.0017022996908053756,
-      "learning_rate": 0.0001999996857640483,
-      "loss": 46.0,
-      "step": 10447
-    },
-    {
-      "epoch": 0.7988225624557983,
-      "grad_norm": 0.001921234535984695,
-      "learning_rate": 0.00019999968570382968,
-      "loss": 46.0,
-      "step": 10448
-    },
-    {
-      "epoch": 0.7988990194391881,
-      "grad_norm": 0.0009661572403274477,
-      "learning_rate": 0.00019999968564360528,
-      "loss": 46.0,
-      "step": 10449
-    },
-    {
-      "epoch": 0.7989754764225777,
-      "grad_norm": 0.0013993048341944814,
-      "learning_rate": 0.00019999968558337514,
-      "loss": 46.0,
-      "step": 10450
-    },
-    {
-      "epoch": 0.7990519334059675,
-      "grad_norm": 0.001107872580178082,
-      "learning_rate": 0.0001999996855231392,
-      "loss": 46.0,
-      "step": 10451
-    },
-    {
-      "epoch": 0.7991283903893572,
-      "grad_norm": 0.0006258583744056523,
-      "learning_rate": 0.0001999996854628975,
-      "loss": 46.0,
-      "step": 10452
-    },
-    {
-      "epoch": 0.7992048473727469,
-      "grad_norm": 0.0011497348314151168,
-      "learning_rate": 0.00019999968540265005,
-      "loss": 46.0,
-      "step": 10453
-    },
-    {
-      "epoch": 0.7992813043561366,
-      "grad_norm": 0.015417809598147869,
-      "learning_rate": 0.00019999968534239684,
-      "loss": 46.0,
-      "step": 10454
-    },
-    {
-      "epoch": 0.7993577613395263,
-      "grad_norm": 0.0012206232640892267,
-      "learning_rate": 0.0001999996852821378,
-      "loss": 46.0,
-      "step": 10455
-    },
-    {
-      "epoch": 0.7994342183229161,
-      "grad_norm": 0.0011034088674932718,
-      "learning_rate": 0.00019999968522187305,
-      "loss": 46.0,
-      "step": 10456
-    },
-    {
-      "epoch": 0.7995106753063058,
-      "grad_norm": 0.0007100504008121789,
-      "learning_rate": 0.00019999968516160252,
-      "loss": 46.0,
-      "step": 10457
-    },
-    {
-      "epoch": 0.7995871322896955,
-      "grad_norm": 0.0025360225699841976,
-      "learning_rate": 0.0001999996851013262,
-      "loss": 46.0,
-      "step": 10458
-    },
-    {
-      "epoch": 0.7996635892730852,
-      "grad_norm": 0.0007951929001137614,
-      "learning_rate": 0.00019999968504104412,
-      "loss": 46.0,
-      "step": 10459
-    },
-    {
-      "epoch": 0.799740046256475,
-      "grad_norm": 0.0011786696268245578,
-      "learning_rate": 0.00019999968498075627,
-      "loss": 46.0,
-      "step": 10460
-    },
-    {
-      "epoch": 0.7998165032398646,
-      "grad_norm": 0.004398555960506201,
-      "learning_rate": 0.00019999968492046265,
-      "loss": 46.0,
-      "step": 10461
-    },
-    {
-      "epoch": 0.7998929602232544,
-      "grad_norm": 0.0007128039142116904,
-      "learning_rate": 0.00019999968486016328,
-      "loss": 46.0,
-      "step": 10462
-    },
-    {
-      "epoch": 0.7999694172066442,
-      "grad_norm": 0.0024774540215730667,
-      "learning_rate": 0.00019999968479985812,
-      "loss": 46.0,
-      "step": 10463
-    },
-    {
-      "epoch": 0.8000458741900338,
-      "grad_norm": 0.007371298037469387,
-      "learning_rate": 0.0001999996847395472,
-      "loss": 46.0,
-      "step": 10464
-    },
-    {
-      "epoch": 0.8001223311734236,
-      "grad_norm": 0.0011722599156200886,
-      "learning_rate": 0.0001999996846792305,
-      "loss": 46.0,
-      "step": 10465
-    },
-    {
-      "epoch": 0.8001987881568132,
-      "grad_norm": 0.00029976724181324244,
-      "learning_rate": 0.000199999684618908,
-      "loss": 46.0,
-      "step": 10466
-    },
-    {
-      "epoch": 0.800275245140203,
-      "grad_norm": 0.0010954602621495724,
-      "learning_rate": 0.0001999996845585798,
-      "loss": 46.0,
-      "step": 10467
-    },
-    {
-      "epoch": 0.8003517021235927,
-      "grad_norm": 0.0011874493211507797,
-      "learning_rate": 0.00019999968449824578,
-      "loss": 46.0,
-      "step": 10468
-    },
-    {
-      "epoch": 0.8004281591069824,
-      "grad_norm": 0.0016141465166583657,
-      "learning_rate": 0.00019999968443790603,
-      "loss": 46.0,
-      "step": 10469
-    },
-    {
-      "epoch": 0.8005046160903722,
-      "grad_norm": 0.0012601014459505677,
-      "learning_rate": 0.00019999968437756048,
-      "loss": 46.0,
-      "step": 10470
-    },
-    {
-      "epoch": 0.8005810730737619,
-      "grad_norm": 0.0020059049129486084,
-      "learning_rate": 0.00019999968431720917,
-      "loss": 46.0,
-      "step": 10471
-    },
-    {
-      "epoch": 0.8006575300571516,
-      "grad_norm": 0.0005422209505923092,
-      "learning_rate": 0.00019999968425685207,
-      "loss": 46.0,
-      "step": 10472
-    },
-    {
-      "epoch": 0.8007339870405413,
-      "grad_norm": 0.0010006512748077512,
-      "learning_rate": 0.00019999968419648925,
-      "loss": 46.0,
-      "step": 10473
-    },
-    {
-      "epoch": 0.8008104440239311,
-      "grad_norm": 0.0040566627867519855,
-      "learning_rate": 0.0001999996841361206,
-      "loss": 46.0,
-      "step": 10474
-    },
-    {
-      "epoch": 0.8008869010073207,
-      "grad_norm": 0.0007985107367858291,
-      "learning_rate": 0.00019999968407574624,
-      "loss": 46.0,
-      "step": 10475
-    },
-    {
-      "epoch": 0.8009633579907105,
-      "grad_norm": 0.0014343177899718285,
-      "learning_rate": 0.00019999968401536607,
-      "loss": 46.0,
-      "step": 10476
-    },
-    {
-      "epoch": 0.8010398149741002,
-      "grad_norm": 0.00050419254694134,
-      "learning_rate": 0.00019999968395498013,
-      "loss": 46.0,
-      "step": 10477
-    },
-    {
-      "epoch": 0.8011162719574899,
-      "grad_norm": 0.0014299041358754039,
-      "learning_rate": 0.00019999968389458844,
-      "loss": 46.0,
-      "step": 10478
-    },
-    {
-      "epoch": 0.8011927289408797,
-      "grad_norm": 0.0025366810150444508,
-      "learning_rate": 0.00019999968383419095,
-      "loss": 46.0,
-      "step": 10479
-    },
-    {
-      "epoch": 0.8012691859242693,
-      "grad_norm": 0.004242269322276115,
-      "learning_rate": 0.00019999968377378772,
-      "loss": 46.0,
-      "step": 10480
-    },
-    {
-      "epoch": 0.8013456429076591,
-      "grad_norm": 0.0012129811802878976,
-      "learning_rate": 0.0001999996837133787,
-      "loss": 46.0,
-      "step": 10481
-    },
-    {
-      "epoch": 0.8014220998910488,
-      "grad_norm": 0.003892101813107729,
-      "learning_rate": 0.00019999968365296396,
-      "loss": 46.0,
-      "step": 10482
-    },
-    {
-      "epoch": 0.8014985568744385,
-      "grad_norm": 0.0007331674569286406,
-      "learning_rate": 0.00019999968359254343,
-      "loss": 46.0,
-      "step": 10483
-    },
-    {
-      "epoch": 0.8015750138578283,
-      "grad_norm": 0.0009749920573085546,
-      "learning_rate": 0.00019999968353211708,
-      "loss": 46.0,
-      "step": 10484
-    },
-    {
-      "epoch": 0.8016514708412179,
-      "grad_norm": 0.009521284140646458,
-      "learning_rate": 0.000199999683471685,
-      "loss": 46.0,
-      "step": 10485
-    },
-    {
-      "epoch": 0.8017279278246077,
-      "grad_norm": 0.0014507188461720943,
-      "learning_rate": 0.00019999968341124716,
-      "loss": 46.0,
-      "step": 10486
-    },
-    {
-      "epoch": 0.8018043848079974,
-      "grad_norm": 0.00235563050955534,
-      "learning_rate": 0.00019999968335080354,
-      "loss": 46.0,
-      "step": 10487
-    },
-    {
-      "epoch": 0.8018808417913871,
-      "grad_norm": 0.010525641962885857,
-      "learning_rate": 0.00019999968329035411,
-      "loss": 46.0,
-      "step": 10488
-    },
-    {
-      "epoch": 0.8019572987747768,
-      "grad_norm": 0.00044382206397131085,
-      "learning_rate": 0.00019999968322989897,
-      "loss": 46.0,
-      "step": 10489
-    },
-    {
-      "epoch": 0.8020337557581666,
-      "grad_norm": 0.0006507197394967079,
-      "learning_rate": 0.00019999968316943803,
-      "loss": 46.0,
-      "step": 10490
-    },
-    {
-      "epoch": 0.8021102127415562,
-      "grad_norm": 0.0005255329888314009,
-      "learning_rate": 0.00019999968310897135,
-      "loss": 46.0,
-      "step": 10491
-    },
-    {
-      "epoch": 0.802186669724946,
-      "grad_norm": 0.0008625674527138472,
-      "learning_rate": 0.00019999968304849889,
-      "loss": 46.0,
-      "step": 10492
-    },
-    {
-      "epoch": 0.8022631267083358,
-      "grad_norm": 0.0030411214102059603,
-      "learning_rate": 0.00019999968298802063,
-      "loss": 46.0,
-      "step": 10493
-    },
-    {
-      "epoch": 0.8023395836917254,
-      "grad_norm": 0.0010720710270106792,
-      "learning_rate": 0.00019999968292753662,
-      "loss": 46.0,
-      "step": 10494
-    },
-    {
-      "epoch": 0.8024160406751152,
-      "grad_norm": 0.003892915789037943,
-      "learning_rate": 0.00019999968286704684,
-      "loss": 46.0,
-      "step": 10495
-    },
-    {
-      "epoch": 0.8024924976585048,
-      "grad_norm": 0.0009724826668389142,
-      "learning_rate": 0.0001999996828065513,
-      "loss": 46.0,
-      "step": 10496
-    },
-    {
-      "epoch": 0.8025689546418946,
-      "grad_norm": 0.0009155191364698112,
-      "learning_rate": 0.00019999968274604998,
-      "loss": 46.0,
-      "step": 10497
-    },
-    {
-      "epoch": 0.8026454116252844,
-      "grad_norm": 0.007622969336807728,
-      "learning_rate": 0.0001999996826855429,
-      "loss": 46.0,
-      "step": 10498
-    },
-    {
-      "epoch": 0.802721868608674,
-      "grad_norm": 0.0004769606457557529,
-      "learning_rate": 0.00019999968262503004,
-      "loss": 46.0,
-      "step": 10499
-    },
-    {
-      "epoch": 0.8027983255920638,
-      "grad_norm": 0.0014206409687176347,
-      "learning_rate": 0.00019999968256451142,
-      "loss": 46.0,
-      "step": 10500
-    },
-    {
-      "epoch": 0.8028747825754535,
-      "grad_norm": 0.0012016526889055967,
-      "learning_rate": 0.00019999968250398702,
-      "loss": 46.0,
-      "step": 10501
-    },
-    {
-      "epoch": 0.8029512395588432,
-      "grad_norm": 0.0016582179814577103,
-      "learning_rate": 0.00019999968244345686,
-      "loss": 46.0,
-      "step": 10502
-    },
-    {
-      "epoch": 0.8030276965422329,
-      "grad_norm": 0.0010918057523667812,
-      "learning_rate": 0.00019999968238292091,
-      "loss": 46.0,
-      "step": 10503
-    },
-    {
-      "epoch": 0.8031041535256227,
-      "grad_norm": 0.0010731570655480027,
-      "learning_rate": 0.00019999968232237923,
-      "loss": 46.0,
-      "step": 10504
-    },
-    {
-      "epoch": 0.8031806105090123,
-      "grad_norm": 0.0007092529558576643,
-      "learning_rate": 0.00019999968226183174,
-      "loss": 46.0,
-      "step": 10505
-    },
-    {
-      "epoch": 0.8032570674924021,
-      "grad_norm": 0.0007144889677874744,
-      "learning_rate": 0.00019999968220127848,
-      "loss": 46.0,
-      "step": 10506
-    },
-    {
-      "epoch": 0.8033335244757918,
-      "grad_norm": 0.0019835426937788725,
-      "learning_rate": 0.0001999996821407195,
-      "loss": 46.0,
-      "step": 10507
-    },
-    {
-      "epoch": 0.8034099814591815,
-      "grad_norm": 0.0013765008188784122,
-      "learning_rate": 0.00019999968208015472,
-      "loss": 46.0,
-      "step": 10508
-    },
-    {
-      "epoch": 0.8034864384425713,
-      "grad_norm": 0.003526052925735712,
-      "learning_rate": 0.00019999968201958417,
-      "loss": 46.0,
-      "step": 10509
-    },
-    {
-      "epoch": 0.8035628954259609,
-      "grad_norm": 0.01059807650744915,
-      "learning_rate": 0.00019999968195900784,
-      "loss": 46.0,
-      "step": 10510
-    },
-    {
-      "epoch": 0.8036393524093507,
-      "grad_norm": 0.00045217445585876703,
-      "learning_rate": 0.00019999968189842574,
-      "loss": 46.0,
-      "step": 10511
-    },
-    {
-      "epoch": 0.8037158093927405,
-      "grad_norm": 0.0024424255825579166,
-      "learning_rate": 0.0001999996818378379,
-      "loss": 46.0,
-      "step": 10512
-    },
-    {
-      "epoch": 0.8037922663761301,
-      "grad_norm": 0.0009442882146686316,
-      "learning_rate": 0.00019999968177724427,
-      "loss": 46.0,
-      "step": 10513
-    },
-    {
-      "epoch": 0.8038687233595199,
-      "grad_norm": 0.0008289010147564113,
-      "learning_rate": 0.00019999968171664488,
-      "loss": 46.0,
-      "step": 10514
-    },
-    {
-      "epoch": 0.8039451803429095,
-      "grad_norm": 0.0006760719697922468,
-      "learning_rate": 0.0001999996816560397,
-      "loss": 46.0,
-      "step": 10515
-    },
-    {
-      "epoch": 0.8040216373262993,
-      "grad_norm": 0.0017378043849021196,
-      "learning_rate": 0.00019999968159542877,
-      "loss": 46.0,
-      "step": 10516
-    },
-    {
-      "epoch": 0.804098094309689,
-      "grad_norm": 0.0046377889811992645,
-      "learning_rate": 0.00019999968153481206,
-      "loss": 46.0,
-      "step": 10517
-    },
-    {
-      "epoch": 0.8041745512930787,
-      "grad_norm": 0.0007898385520093143,
-      "learning_rate": 0.00019999968147418957,
-      "loss": 46.0,
-      "step": 10518
-    },
-    {
-      "epoch": 0.8042510082764684,
-      "grad_norm": 0.00434258533641696,
-      "learning_rate": 0.00019999968141356137,
-      "loss": 46.0,
-      "step": 10519
-    },
-    {
-      "epoch": 0.8043274652598582,
-      "grad_norm": 0.0014697652077302337,
-      "learning_rate": 0.00019999968135292733,
-      "loss": 46.0,
-      "step": 10520
-    },
-    {
-      "epoch": 0.8044039222432479,
-      "grad_norm": 0.001740574254654348,
-      "learning_rate": 0.00019999968129228755,
-      "loss": 46.0,
-      "step": 10521
-    },
-    {
-      "epoch": 0.8044803792266376,
-      "grad_norm": 0.000758209964260459,
-      "learning_rate": 0.000199999681231642,
-      "loss": 46.0,
-      "step": 10522
-    },
-    {
-      "epoch": 0.8045568362100274,
-      "grad_norm": 0.001774105941876769,
-      "learning_rate": 0.00019999968117099067,
-      "loss": 46.0,
-      "step": 10523
-    },
-    {
-      "epoch": 0.804633293193417,
-      "grad_norm": 0.001304331235587597,
-      "learning_rate": 0.00019999968111033357,
-      "loss": 46.0,
-      "step": 10524
-    },
-    {
-      "epoch": 0.8047097501768068,
-      "grad_norm": 0.0006807984900660813,
-      "learning_rate": 0.0001999996810496707,
-      "loss": 46.0,
-      "step": 10525
-    },
-    {
-      "epoch": 0.8047862071601964,
-      "grad_norm": 0.0006133479764685035,
-      "learning_rate": 0.00019999968098900208,
-      "loss": 46.0,
-      "step": 10526
-    },
-    {
-      "epoch": 0.8048626641435862,
-      "grad_norm": 0.0009403585572727025,
-      "learning_rate": 0.0001999996809283277,
-      "loss": 46.0,
-      "step": 10527
-    },
-    {
-      "epoch": 0.804939121126976,
-      "grad_norm": 0.0006824696902185678,
-      "learning_rate": 0.00019999968086764752,
-      "loss": 46.0,
-      "step": 10528
-    },
-    {
-      "epoch": 0.8050155781103656,
-      "grad_norm": 0.001689778408035636,
-      "learning_rate": 0.00019999968080696155,
-      "loss": 46.0,
-      "step": 10529
-    },
-    {
-      "epoch": 0.8050920350937554,
-      "grad_norm": 0.0009304852574132383,
-      "learning_rate": 0.00019999968074626987,
-      "loss": 46.0,
-      "step": 10530
-    },
-    {
-      "epoch": 0.8051684920771451,
-      "grad_norm": 0.0010523987002670765,
-      "learning_rate": 0.0001999996806855724,
-      "loss": 46.0,
-      "step": 10531
-    },
-    {
-      "epoch": 0.8052449490605348,
-      "grad_norm": 0.0015985339414328337,
-      "learning_rate": 0.00019999968062486912,
-      "loss": 46.0,
-      "step": 10532
-    },
-    {
-      "epoch": 0.8053214060439245,
-      "grad_norm": 0.0005110496422275901,
-      "learning_rate": 0.00019999968056416014,
-      "loss": 46.0,
-      "step": 10533
-    },
-    {
-      "epoch": 0.8053978630273143,
-      "grad_norm": 0.0008613889222033322,
-      "learning_rate": 0.00019999968050344534,
-      "loss": 46.0,
-      "step": 10534
-    },
-    {
-      "epoch": 0.805474320010704,
-      "grad_norm": 0.0006032785749994218,
-      "learning_rate": 0.00019999968044272478,
-      "loss": 46.0,
-      "step": 10535
-    },
-    {
-      "epoch": 0.8055507769940937,
-      "grad_norm": 0.0021466233301907778,
-      "learning_rate": 0.00019999968038199846,
-      "loss": 46.0,
-      "step": 10536
-    },
-    {
-      "epoch": 0.8056272339774834,
-      "grad_norm": 0.0006493675173260272,
-      "learning_rate": 0.00019999968032126636,
-      "loss": 46.0,
-      "step": 10537
-    },
-    {
-      "epoch": 0.8057036909608731,
-      "grad_norm": 0.00239465176127851,
-      "learning_rate": 0.00019999968026052848,
-      "loss": 46.0,
-      "step": 10538
-    },
-    {
-      "epoch": 0.8057801479442629,
-      "grad_norm": 0.000779195106588304,
-      "learning_rate": 0.00019999968019978486,
-      "loss": 46.0,
-      "step": 10539
-    },
-    {
-      "epoch": 0.8058566049276525,
-      "grad_norm": 0.005790887400507927,
-      "learning_rate": 0.00019999968013903544,
-      "loss": 46.0,
-      "step": 10540
-    },
-    {
-      "epoch": 0.8059330619110423,
-      "grad_norm": 0.00028756994288414717,
-      "learning_rate": 0.0001999996800782803,
-      "loss": 46.0,
-      "step": 10541
-    },
-    {
-      "epoch": 0.8060095188944321,
-      "grad_norm": 0.0033163148909807205,
-      "learning_rate": 0.00019999968001751937,
-      "loss": 46.0,
-      "step": 10542
-    },
-    {
-      "epoch": 0.8060859758778217,
-      "grad_norm": 0.0010280621936544776,
-      "learning_rate": 0.00019999967995675263,
-      "loss": 46.0,
-      "step": 10543
-    },
-    {
-      "epoch": 0.8061624328612115,
-      "grad_norm": 0.0008250966784544289,
-      "learning_rate": 0.00019999967989598014,
-      "loss": 46.0,
-      "step": 10544
-    },
-    {
-      "epoch": 0.8062388898446012,
-      "grad_norm": 0.0018860476557165384,
-      "learning_rate": 0.0001999996798352019,
-      "loss": 46.0,
-      "step": 10545
-    },
-    {
-      "epoch": 0.8063153468279909,
-      "grad_norm": 0.00048766477266326547,
-      "learning_rate": 0.0001999996797744179,
-      "loss": 46.0,
-      "step": 10546
-    },
-    {
-      "epoch": 0.8063918038113806,
-      "grad_norm": 0.0006458511925302446,
-      "learning_rate": 0.0001999996797136281,
-      "loss": 46.0,
-      "step": 10547
-    },
-    {
-      "epoch": 0.8064682607947703,
-      "grad_norm": 0.0010112955933436751,
-      "learning_rate": 0.00019999967965283254,
-      "loss": 46.0,
-      "step": 10548
-    },
-    {
-      "epoch": 0.8065447177781601,
-      "grad_norm": 0.0039372690953314304,
-      "learning_rate": 0.00019999967959203122,
-      "loss": 46.0,
-      "step": 10549
-    },
-    {
-      "epoch": 0.8066211747615498,
-      "grad_norm": 0.0015946135390549898,
-      "learning_rate": 0.00019999967953122412,
-      "loss": 46.0,
-      "step": 10550
-    },
-    {
-      "epoch": 0.8066976317449395,
-      "grad_norm": 0.0023373279254883528,
-      "learning_rate": 0.00019999967947041124,
-      "loss": 46.0,
-      "step": 10551
-    },
-    {
-      "epoch": 0.8067740887283292,
-      "grad_norm": 0.003955328371375799,
-      "learning_rate": 0.0001999996794095926,
-      "loss": 46.0,
-      "step": 10552
-    },
-    {
-      "epoch": 0.806850545711719,
-      "grad_norm": 0.0017770793056115508,
-      "learning_rate": 0.00019999967934876823,
-      "loss": 46.0,
-      "step": 10553
-    },
-    {
-      "epoch": 0.8069270026951086,
-      "grad_norm": 0.0029069725424051285,
-      "learning_rate": 0.00019999967928793804,
-      "loss": 46.0,
-      "step": 10554
-    },
-    {
-      "epoch": 0.8070034596784984,
-      "grad_norm": 0.0012229240965098143,
-      "learning_rate": 0.00019999967922710207,
-      "loss": 46.0,
-      "step": 10555
-    },
-    {
-      "epoch": 0.807079916661888,
-      "grad_norm": 0.0008900503744371235,
-      "learning_rate": 0.0001999996791662604,
-      "loss": 46.0,
-      "step": 10556
-    },
-    {
-      "epoch": 0.8071563736452778,
-      "grad_norm": 0.000807922100648284,
-      "learning_rate": 0.00019999967910541288,
-      "loss": 46.0,
-      "step": 10557
-    },
-    {
-      "epoch": 0.8072328306286676,
-      "grad_norm": 0.00023124992731027305,
-      "learning_rate": 0.00019999967904455964,
-      "loss": 46.0,
-      "step": 10558
-    },
-    {
-      "epoch": 0.8073092876120572,
-      "grad_norm": 0.0033947890624403954,
-      "learning_rate": 0.0001999996789837006,
-      "loss": 46.0,
-      "step": 10559
-    },
-    {
-      "epoch": 0.807385744595447,
-      "grad_norm": 0.0008910118485800922,
-      "learning_rate": 0.00019999967892283583,
-      "loss": 46.0,
-      "step": 10560
-    },
-    {
-      "epoch": 0.8074622015788367,
-      "grad_norm": 0.011881954967975616,
-      "learning_rate": 0.00019999967886196528,
-      "loss": 46.0,
-      "step": 10561
-    },
-    {
-      "epoch": 0.8075386585622264,
-      "grad_norm": 0.0012975921854376793,
-      "learning_rate": 0.00019999967880108895,
-      "loss": 46.0,
-      "step": 10562
-    },
-    {
-      "epoch": 0.8076151155456162,
-      "grad_norm": 0.011599060148000717,
-      "learning_rate": 0.00019999967874020685,
-      "loss": 46.0,
-      "step": 10563
-    },
-    {
-      "epoch": 0.8076915725290059,
-      "grad_norm": 0.017311716452240944,
-      "learning_rate": 0.00019999967867931898,
-      "loss": 46.0,
-      "step": 10564
-    },
-    {
-      "epoch": 0.8077680295123956,
-      "grad_norm": 0.004300176165997982,
-      "learning_rate": 0.00019999967861842534,
-      "loss": 46.0,
-      "step": 10565
-    },
-    {
-      "epoch": 0.8078444864957853,
-      "grad_norm": 0.001962413312867284,
-      "learning_rate": 0.00019999967855752592,
-      "loss": 46.0,
-      "step": 10566
-    },
-    {
-      "epoch": 0.807920943479175,
-      "grad_norm": 0.0010369386291131377,
-      "learning_rate": 0.00019999967849662072,
-      "loss": 46.0,
-      "step": 10567
-    },
-    {
-      "epoch": 0.8079974004625647,
-      "grad_norm": 0.0007597673102281988,
-      "learning_rate": 0.00019999967843570979,
-      "loss": 46.0,
-      "step": 10568
-    },
-    {
-      "epoch": 0.8080738574459545,
-      "grad_norm": 0.0006968869711272418,
-      "learning_rate": 0.00019999967837479307,
-      "loss": 46.0,
-      "step": 10569
-    },
-    {
-      "epoch": 0.8081503144293442,
-      "grad_norm": 0.0018348144367337227,
-      "learning_rate": 0.0001999996783138706,
-      "loss": 46.0,
-      "step": 10570
-    },
-    {
-      "epoch": 0.8082267714127339,
-      "grad_norm": 0.0006603803485631943,
-      "learning_rate": 0.00019999967825294236,
-      "loss": 46.0,
-      "step": 10571
-    },
-    {
-      "epoch": 0.8083032283961237,
-      "grad_norm": 0.0015955101698637009,
-      "learning_rate": 0.00019999967819200832,
-      "loss": 46.0,
-      "step": 10572
-    },
-    {
-      "epoch": 0.8083796853795133,
-      "grad_norm": 0.00055325822904706,
-      "learning_rate": 0.0001999996781310685,
-      "loss": 46.0,
-      "step": 10573
-    },
-    {
-      "epoch": 0.8084561423629031,
-      "grad_norm": 0.0011604411993175745,
-      "learning_rate": 0.00019999967807012294,
-      "loss": 46.0,
-      "step": 10574
-    },
-    {
-      "epoch": 0.8085325993462928,
-      "grad_norm": 0.0025593554601073265,
-      "learning_rate": 0.00019999967800917161,
-      "loss": 46.0,
-      "step": 10575
-    },
-    {
-      "epoch": 0.8086090563296825,
-      "grad_norm": 0.0004935767501592636,
-      "learning_rate": 0.00019999967794821451,
-      "loss": 46.0,
-      "step": 10576
-    },
-    {
-      "epoch": 0.8086855133130723,
-      "grad_norm": 0.002077204640954733,
-      "learning_rate": 0.00019999967788725164,
-      "loss": 46.0,
-      "step": 10577
-    },
-    {
-      "epoch": 0.8087619702964619,
-      "grad_norm": 0.0006066091009415686,
-      "learning_rate": 0.00019999967782628302,
-      "loss": 46.0,
-      "step": 10578
-    },
-    {
-      "epoch": 0.8088384272798517,
-      "grad_norm": 0.0005193352117203176,
-      "learning_rate": 0.00019999967776530858,
-      "loss": 46.0,
-      "step": 10579
-    },
-    {
-      "epoch": 0.8089148842632414,
-      "grad_norm": 0.0038145852740854025,
-      "learning_rate": 0.0001999996777043284,
-      "loss": 46.0,
-      "step": 10580
-    },
-    {
-      "epoch": 0.8089913412466311,
-      "grad_norm": 0.0009923079051077366,
-      "learning_rate": 0.00019999967764334247,
-      "loss": 46.0,
-      "step": 10581
-    },
-    {
-      "epoch": 0.8090677982300208,
-      "grad_norm": 0.0008851053426042199,
-      "learning_rate": 0.00019999967758235074,
-      "loss": 46.0,
-      "step": 10582
-    },
-    {
-      "epoch": 0.8091442552134106,
-      "grad_norm": 0.002756862435489893,
-      "learning_rate": 0.00019999967752135325,
-      "loss": 46.0,
-      "step": 10583
-    },
-    {
-      "epoch": 0.8092207121968003,
-      "grad_norm": 0.0009387233876623213,
-      "learning_rate": 0.00019999967746035,
-      "loss": 46.0,
-      "step": 10584
-    },
-    {
-      "epoch": 0.80929716918019,
-      "grad_norm": 0.0013590651797130704,
-      "learning_rate": 0.000199999677399341,
-      "loss": 46.0,
-      "step": 10585
-    },
-    {
-      "epoch": 0.8093736261635797,
-      "grad_norm": 0.006492145359516144,
-      "learning_rate": 0.00019999967733832618,
-      "loss": 46.0,
-      "step": 10586
-    },
-    {
-      "epoch": 0.8094500831469694,
-      "grad_norm": 0.0019619599916040897,
-      "learning_rate": 0.00019999967727730563,
-      "loss": 46.0,
-      "step": 10587
-    },
-    {
-      "epoch": 0.8095265401303592,
-      "grad_norm": 0.0008877667132765055,
-      "learning_rate": 0.00019999967721627928,
-      "loss": 46.0,
-      "step": 10588
-    },
-    {
-      "epoch": 0.8096029971137488,
-      "grad_norm": 0.0017375691095367074,
-      "learning_rate": 0.00019999967715524718,
-      "loss": 46.0,
-      "step": 10589
-    },
-    {
-      "epoch": 0.8096794540971386,
-      "grad_norm": 0.005117062479257584,
-      "learning_rate": 0.00019999967709420928,
-      "loss": 46.0,
-      "step": 10590
-    },
-    {
-      "epoch": 0.8097559110805284,
-      "grad_norm": 0.0035905144177377224,
-      "learning_rate": 0.00019999967703316564,
-      "loss": 46.0,
-      "step": 10591
-    },
-    {
-      "epoch": 0.809832368063918,
-      "grad_norm": 0.00300415325909853,
-      "learning_rate": 0.00019999967697211624,
-      "loss": 46.0,
-      "step": 10592
-    },
-    {
-      "epoch": 0.8099088250473078,
-      "grad_norm": 0.000932666240260005,
-      "learning_rate": 0.00019999967691106105,
-      "loss": 46.0,
-      "step": 10593
-    },
-    {
-      "epoch": 0.8099852820306975,
-      "grad_norm": 0.0014116576639935374,
-      "learning_rate": 0.0001999996768500001,
-      "loss": 46.0,
-      "step": 10594
-    },
-    {
-      "epoch": 0.8100617390140872,
-      "grad_norm": 0.0007366169593296945,
-      "learning_rate": 0.00019999967678893337,
-      "loss": 46.0,
-      "step": 10595
-    },
-    {
-      "epoch": 0.8101381959974769,
-      "grad_norm": 0.002435180824249983,
-      "learning_rate": 0.0001999996767278609,
-      "loss": 46.0,
-      "step": 10596
-    },
-    {
-      "epoch": 0.8102146529808666,
-      "grad_norm": 0.0018935957923531532,
-      "learning_rate": 0.0001999996766667826,
-      "loss": 46.0,
-      "step": 10597
-    },
-    {
-      "epoch": 0.8102911099642564,
-      "grad_norm": 0.0006910284864716232,
-      "learning_rate": 0.00019999967660569857,
-      "loss": 46.0,
-      "step": 10598
-    },
-    {
-      "epoch": 0.8103675669476461,
-      "grad_norm": 0.001545571954920888,
-      "learning_rate": 0.0001999996765446088,
-      "loss": 46.0,
-      "step": 10599
-    },
-    {
-      "epoch": 0.8104440239310358,
-      "grad_norm": 0.000960881938226521,
-      "learning_rate": 0.00019999967648351322,
-      "loss": 46.0,
-      "step": 10600
-    },
-    {
-      "epoch": 0.8105204809144255,
-      "grad_norm": 0.0007635874208062887,
-      "learning_rate": 0.0001999996764224119,
-      "loss": 46.0,
-      "step": 10601
-    },
-    {
-      "epoch": 0.8105969378978153,
-      "grad_norm": 0.0011450700694695115,
-      "learning_rate": 0.0001999996763613048,
-      "loss": 46.0,
-      "step": 10602
-    },
-    {
-      "epoch": 0.8106733948812049,
-      "grad_norm": 0.002021638909354806,
-      "learning_rate": 0.0001999996763001919,
-      "loss": 46.0,
-      "step": 10603
-    },
-    {
-      "epoch": 0.8107498518645947,
-      "grad_norm": 0.0009661731892265379,
-      "learning_rate": 0.00019999967623907325,
-      "loss": 46.0,
-      "step": 10604
-    },
-    {
-      "epoch": 0.8108263088479845,
-      "grad_norm": 0.0011234754929319024,
-      "learning_rate": 0.00019999967617794883,
-      "loss": 46.0,
-      "step": 10605
-    },
-    {
-      "epoch": 0.8109027658313741,
-      "grad_norm": 0.0013114104513078928,
-      "learning_rate": 0.00019999967611681867,
-      "loss": 46.0,
-      "step": 10606
-    },
-    {
-      "epoch": 0.8109792228147639,
-      "grad_norm": 0.0027648042887449265,
-      "learning_rate": 0.00019999967605568268,
-      "loss": 46.0,
-      "step": 10607
-    },
-    {
-      "epoch": 0.8110556797981535,
-      "grad_norm": 0.008482534438371658,
-      "learning_rate": 0.00019999967599454096,
-      "loss": 46.0,
-      "step": 10608
-    },
-    {
-      "epoch": 0.8111321367815433,
-      "grad_norm": 0.000731699401512742,
-      "learning_rate": 0.00019999967593339348,
-      "loss": 46.0,
-      "step": 10609
-    },
-    {
-      "epoch": 0.811208593764933,
-      "grad_norm": 0.0006798803224228323,
-      "learning_rate": 0.00019999967587224022,
-      "loss": 46.0,
-      "step": 10610
-    },
-    {
-      "epoch": 0.8112850507483227,
-      "grad_norm": 0.0024888545740395784,
-      "learning_rate": 0.0001999996758110812,
-      "loss": 46.0,
-      "step": 10611
-    },
-    {
-      "epoch": 0.8113615077317125,
-      "grad_norm": 0.0004918433842249215,
-      "learning_rate": 0.00019999967574991639,
-      "loss": 46.0,
-      "step": 10612
-    },
-    {
-      "epoch": 0.8114379647151022,
-      "grad_norm": 0.0011814339086413383,
-      "learning_rate": 0.0001999996756887458,
-      "loss": 46.0,
-      "step": 10613
-    },
-    {
-      "epoch": 0.8115144216984919,
-      "grad_norm": 0.0029570814222097397,
-      "learning_rate": 0.00019999967562756948,
-      "loss": 46.0,
-      "step": 10614
-    },
-    {
-      "epoch": 0.8115908786818816,
-      "grad_norm": 0.0005436950013972819,
-      "learning_rate": 0.00019999967556638736,
-      "loss": 46.0,
-      "step": 10615
-    },
-    {
-      "epoch": 0.8116673356652713,
-      "grad_norm": 0.0006963139167055488,
-      "learning_rate": 0.00019999967550519946,
-      "loss": 46.0,
-      "step": 10616
-    },
-    {
-      "epoch": 0.811743792648661,
-      "grad_norm": 0.00921026524156332,
-      "learning_rate": 0.00019999967544400584,
-      "loss": 46.0,
-      "step": 10617
-    },
-    {
-      "epoch": 0.8118202496320508,
-      "grad_norm": 0.0007407678058370948,
-      "learning_rate": 0.0001999996753828064,
-      "loss": 46.0,
-      "step": 10618
-    },
-    {
-      "epoch": 0.8118967066154404,
-      "grad_norm": 0.0011966772144660354,
-      "learning_rate": 0.00019999967532160123,
-      "loss": 46.0,
-      "step": 10619
-    },
-    {
-      "epoch": 0.8119731635988302,
-      "grad_norm": 0.0012969464296475053,
-      "learning_rate": 0.00019999967526039027,
-      "loss": 46.0,
-      "step": 10620
-    },
-    {
-      "epoch": 0.81204962058222,
-      "grad_norm": 0.0015509049408137798,
-      "learning_rate": 0.00019999967519917356,
-      "loss": 46.0,
-      "step": 10621
-    },
-    {
-      "epoch": 0.8121260775656096,
-      "grad_norm": 0.0048314714804291725,
-      "learning_rate": 0.00019999967513795105,
-      "loss": 46.0,
-      "step": 10622
-    },
-    {
-      "epoch": 0.8122025345489994,
-      "grad_norm": 0.0007092577288858593,
-      "learning_rate": 0.00019999967507672276,
-      "loss": 46.0,
-      "step": 10623
-    },
-    {
-      "epoch": 0.8122789915323891,
-      "grad_norm": 0.0033081660512834787,
-      "learning_rate": 0.00019999967501548876,
-      "loss": 46.0,
-      "step": 10624
-    },
-    {
-      "epoch": 0.8123554485157788,
-      "grad_norm": 0.0013496287865564227,
-      "learning_rate": 0.00019999967495424895,
-      "loss": 46.0,
-      "step": 10625
-    },
-    {
-      "epoch": 0.8124319054991685,
-      "grad_norm": 0.003140041371807456,
-      "learning_rate": 0.00019999967489300335,
-      "loss": 46.0,
-      "step": 10626
-    },
-    {
-      "epoch": 0.8125083624825582,
-      "grad_norm": 0.000635093601886183,
-      "learning_rate": 0.00019999967483175202,
-      "loss": 46.0,
-      "step": 10627
-    },
-    {
-      "epoch": 0.812584819465948,
-      "grad_norm": 0.0047507272101938725,
-      "learning_rate": 0.0001999996747704949,
-      "loss": 46.0,
-      "step": 10628
-    },
-    {
-      "epoch": 0.8126612764493377,
-      "grad_norm": 0.0018666168907657266,
-      "learning_rate": 0.00019999967470923203,
-      "loss": 46.0,
-      "step": 10629
-    },
-    {
-      "epoch": 0.8127377334327274,
-      "grad_norm": 0.0006282232934609056,
-      "learning_rate": 0.00019999967464796336,
-      "loss": 46.0,
-      "step": 10630
-    },
-    {
-      "epoch": 0.8128141904161171,
-      "grad_norm": 0.0007475577294826508,
-      "learning_rate": 0.00019999967458668897,
-      "loss": 46.0,
-      "step": 10631
-    },
-    {
-      "epoch": 0.8128906473995069,
-      "grad_norm": 0.0014038359513506293,
-      "learning_rate": 0.00019999967452540875,
-      "loss": 46.0,
-      "step": 10632
-    },
-    {
-      "epoch": 0.8129671043828965,
-      "grad_norm": 0.0006167926476337016,
-      "learning_rate": 0.0001999996744641228,
-      "loss": 46.0,
-      "step": 10633
-    },
-    {
-      "epoch": 0.8130435613662863,
-      "grad_norm": 0.007281303871423006,
-      "learning_rate": 0.00019999967440283105,
-      "loss": 46.0,
-      "step": 10634
-    },
-    {
-      "epoch": 0.8131200183496761,
-      "grad_norm": 0.004556306637823582,
-      "learning_rate": 0.00019999967434153356,
-      "loss": 46.0,
-      "step": 10635
-    },
-    {
-      "epoch": 0.8131964753330657,
-      "grad_norm": 0.0009176505845971406,
-      "learning_rate": 0.0001999996742802303,
-      "loss": 46.0,
-      "step": 10636
-    },
-    {
-      "epoch": 0.8132729323164555,
-      "grad_norm": 0.0008387612760998309,
-      "learning_rate": 0.00019999967421892127,
-      "loss": 46.0,
-      "step": 10637
-    },
-    {
-      "epoch": 0.8133493892998451,
-      "grad_norm": 0.00030251938733272254,
-      "learning_rate": 0.00019999967415760644,
-      "loss": 46.0,
-      "step": 10638
-    },
-    {
-      "epoch": 0.8134258462832349,
-      "grad_norm": 0.003133848076686263,
-      "learning_rate": 0.00019999967409628586,
-      "loss": 46.0,
-      "step": 10639
-    },
-    {
-      "epoch": 0.8135023032666246,
-      "grad_norm": 0.001175959245301783,
-      "learning_rate": 0.00019999967403495954,
-      "loss": 46.0,
-      "step": 10640
-    },
-    {
-      "epoch": 0.8135787602500143,
-      "grad_norm": 0.0019240969559177756,
-      "learning_rate": 0.0001999996739736274,
-      "loss": 46.0,
-      "step": 10641
-    },
-    {
-      "epoch": 0.8136552172334041,
-      "grad_norm": 0.0012036896077916026,
-      "learning_rate": 0.00019999967391228952,
-      "loss": 46.0,
-      "step": 10642
-    },
-    {
-      "epoch": 0.8137316742167938,
-      "grad_norm": 0.0007384893833659589,
-      "learning_rate": 0.00019999967385094588,
-      "loss": 46.0,
-      "step": 10643
-    },
-    {
-      "epoch": 0.8138081312001835,
-      "grad_norm": 0.0023318736348301172,
-      "learning_rate": 0.00019999967378959646,
-      "loss": 46.0,
-      "step": 10644
-    },
-    {
-      "epoch": 0.8138845881835732,
-      "grad_norm": 0.003980925772339106,
-      "learning_rate": 0.00019999967372824124,
-      "loss": 46.0,
-      "step": 10645
-    },
-    {
-      "epoch": 0.8139610451669629,
-      "grad_norm": 0.0014317728346213698,
-      "learning_rate": 0.00019999967366688028,
-      "loss": 46.0,
-      "step": 10646
-    },
-    {
-      "epoch": 0.8140375021503526,
-      "grad_norm": 0.00195276306476444,
-      "learning_rate": 0.00019999967360551354,
-      "loss": 46.0,
-      "step": 10647
-    },
-    {
-      "epoch": 0.8141139591337424,
-      "grad_norm": 0.002367905108258128,
-      "learning_rate": 0.00019999967354414105,
-      "loss": 46.0,
-      "step": 10648
-    },
-    {
-      "epoch": 0.814190416117132,
-      "grad_norm": 0.00027691235300153494,
-      "learning_rate": 0.00019999967348276277,
-      "loss": 46.0,
-      "step": 10649
-    },
-    {
-      "epoch": 0.8142668731005218,
-      "grad_norm": 0.0026305727660655975,
-      "learning_rate": 0.00019999967342137874,
-      "loss": 46.0,
-      "step": 10650
-    },
-    {
-      "epoch": 0.8143433300839116,
-      "grad_norm": 0.008274825289845467,
-      "learning_rate": 0.00019999967335998894,
-      "loss": 46.0,
-      "step": 10651
-    },
-    {
-      "epoch": 0.8144197870673012,
-      "grad_norm": 0.0010693571530282497,
-      "learning_rate": 0.00019999967329859336,
-      "loss": 46.0,
-      "step": 10652
-    },
-    {
-      "epoch": 0.814496244050691,
-      "grad_norm": 0.0006609584670513868,
-      "learning_rate": 0.000199999673237192,
-      "loss": 46.0,
-      "step": 10653
-    },
-    {
-      "epoch": 0.8145727010340807,
-      "grad_norm": 0.0014914325438439846,
-      "learning_rate": 0.00019999967317578488,
-      "loss": 46.0,
-      "step": 10654
-    },
-    {
-      "epoch": 0.8146491580174704,
-      "grad_norm": 0.0011444612173363566,
-      "learning_rate": 0.000199999673114372,
-      "loss": 46.0,
-      "step": 10655
-    },
-    {
-      "epoch": 0.8147256150008602,
-      "grad_norm": 0.0005577148986048996,
-      "learning_rate": 0.00019999967305295332,
-      "loss": 46.0,
-      "step": 10656
-    },
-    {
-      "epoch": 0.8148020719842498,
-      "grad_norm": 0.0010017806198447943,
-      "learning_rate": 0.0001999996729915289,
-      "loss": 46.0,
-      "step": 10657
-    },
-    {
-      "epoch": 0.8148785289676396,
-      "grad_norm": 0.002900887979194522,
-      "learning_rate": 0.0001999996729300987,
-      "loss": 46.0,
-      "step": 10658
-    },
-    {
-      "epoch": 0.8149549859510293,
-      "grad_norm": 0.005152315832674503,
-      "learning_rate": 0.00019999967286866275,
-      "loss": 46.0,
-      "step": 10659
-    },
-    {
-      "epoch": 0.815031442934419,
-      "grad_norm": 0.0006522878538817167,
-      "learning_rate": 0.000199999672807221,
-      "loss": 46.0,
-      "step": 10660
-    },
-    {
-      "epoch": 0.8151078999178087,
-      "grad_norm": 0.0017765983939170837,
-      "learning_rate": 0.00019999967274577353,
-      "loss": 46.0,
-      "step": 10661
-    },
-    {
-      "epoch": 0.8151843569011985,
-      "grad_norm": 0.0010508804116398096,
-      "learning_rate": 0.00019999967268432024,
-      "loss": 46.0,
-      "step": 10662
-    },
-    {
-      "epoch": 0.8152608138845882,
-      "grad_norm": 0.0006718711229041219,
-      "learning_rate": 0.00019999967262286118,
-      "loss": 46.0,
-      "step": 10663
-    },
-    {
-      "epoch": 0.8153372708679779,
-      "grad_norm": 0.00036310870200395584,
-      "learning_rate": 0.0001999996725613964,
-      "loss": 46.0,
-      "step": 10664
-    },
-    {
-      "epoch": 0.8154137278513677,
-      "grad_norm": 0.0056223091669380665,
-      "learning_rate": 0.0001999996724999258,
-      "loss": 46.0,
-      "step": 10665
-    },
-    {
-      "epoch": 0.8154901848347573,
-      "grad_norm": 0.0005277669406495988,
-      "learning_rate": 0.00019999967243844945,
-      "loss": 46.0,
-      "step": 10666
-    },
-    {
-      "epoch": 0.8155666418181471,
-      "grad_norm": 0.006589485332369804,
-      "learning_rate": 0.00019999967237696733,
-      "loss": 46.0,
-      "step": 10667
-    },
-    {
-      "epoch": 0.8156430988015367,
-      "grad_norm": 0.0006965463981032372,
-      "learning_rate": 0.00019999967231547943,
-      "loss": 46.0,
-      "step": 10668
-    },
-    {
-      "epoch": 0.8157195557849265,
-      "grad_norm": 0.0004891663556918502,
-      "learning_rate": 0.00019999967225398576,
-      "loss": 46.0,
-      "step": 10669
-    },
-    {
-      "epoch": 0.8157960127683163,
-      "grad_norm": 0.0007956529152579606,
-      "learning_rate": 0.00019999967219248635,
-      "loss": 46.0,
-      "step": 10670
-    },
-    {
-      "epoch": 0.8158724697517059,
-      "grad_norm": 0.0005942249554209411,
-      "learning_rate": 0.00019999967213098113,
-      "loss": 46.0,
-      "step": 10671
-    },
-    {
-      "epoch": 0.8159489267350957,
-      "grad_norm": 0.0014059175737202168,
-      "learning_rate": 0.00019999967206947016,
-      "loss": 46.0,
-      "step": 10672
-    },
-    {
-      "epoch": 0.8160253837184854,
-      "grad_norm": 0.006083428859710693,
-      "learning_rate": 0.00019999967200795346,
-      "loss": 46.0,
-      "step": 10673
-    },
-    {
-      "epoch": 0.8161018407018751,
-      "grad_norm": 0.0006601167260669172,
-      "learning_rate": 0.00019999967194643095,
-      "loss": 46.0,
-      "step": 10674
-    },
-    {
-      "epoch": 0.8161782976852648,
-      "grad_norm": 0.0009683783282525837,
-      "learning_rate": 0.00019999967188490266,
-      "loss": 46.0,
-      "step": 10675
-    },
-    {
-      "epoch": 0.8162547546686546,
-      "grad_norm": 0.0008553531370125711,
-      "learning_rate": 0.00019999967182336863,
-      "loss": 46.0,
-      "step": 10676
-    },
-    {
-      "epoch": 0.8163312116520443,
-      "grad_norm": 0.0009982092306017876,
-      "learning_rate": 0.0001999996717618288,
-      "loss": 46.0,
-      "step": 10677
-    },
-    {
-      "epoch": 0.816407668635434,
-      "grad_norm": 0.008247574791312218,
-      "learning_rate": 0.00019999967170028322,
-      "loss": 46.0,
-      "step": 10678
-    },
-    {
-      "epoch": 0.8164841256188237,
-      "grad_norm": 0.00040816611726768315,
-      "learning_rate": 0.00019999967163873185,
-      "loss": 46.0,
-      "step": 10679
-    },
-    {
-      "epoch": 0.8165605826022134,
-      "grad_norm": 0.0005039381212554872,
-      "learning_rate": 0.00019999967157717475,
-      "loss": 46.0,
-      "step": 10680
-    },
-    {
-      "epoch": 0.8166370395856032,
-      "grad_norm": 0.0011511798948049545,
-      "learning_rate": 0.00019999967151561183,
-      "loss": 46.0,
-      "step": 10681
-    },
-    {
-      "epoch": 0.8167134965689928,
-      "grad_norm": 0.0011001338716596365,
-      "learning_rate": 0.00019999967145404316,
-      "loss": 46.0,
-      "step": 10682
-    },
-    {
-      "epoch": 0.8167899535523826,
-      "grad_norm": 0.0010085070971399546,
-      "learning_rate": 0.00019999967139246874,
-      "loss": 46.0,
-      "step": 10683
-    },
-    {
-      "epoch": 0.8168664105357724,
-      "grad_norm": 0.0016138849314302206,
-      "learning_rate": 0.00019999967133088855,
-      "loss": 46.0,
-      "step": 10684
-    },
-    {
-      "epoch": 0.816942867519162,
-      "grad_norm": 0.0036172098480165005,
-      "learning_rate": 0.0001999996712693026,
-      "loss": 46.0,
-      "step": 10685
-    },
-    {
-      "epoch": 0.8170193245025518,
-      "grad_norm": 0.005899055395275354,
-      "learning_rate": 0.00019999967120771082,
-      "loss": 46.0,
-      "step": 10686
-    },
-    {
-      "epoch": 0.8170957814859414,
-      "grad_norm": 0.002268416341394186,
-      "learning_rate": 0.00019999967114611331,
-      "loss": 46.0,
-      "step": 10687
-    },
-    {
-      "epoch": 0.8171722384693312,
-      "grad_norm": 0.0010936105391010642,
-      "learning_rate": 0.00019999967108451006,
-      "loss": 46.0,
-      "step": 10688
-    },
-    {
-      "epoch": 0.8172486954527209,
-      "grad_norm": 0.0028971240390092134,
-      "learning_rate": 0.00019999967102290098,
-      "loss": 46.0,
-      "step": 10689
-    },
-    {
-      "epoch": 0.8173251524361106,
-      "grad_norm": 0.0011245894711464643,
-      "learning_rate": 0.00019999967096128617,
-      "loss": 46.0,
-      "step": 10690
-    },
-    {
-      "epoch": 0.8174016094195004,
-      "grad_norm": 0.0011584118474274874,
-      "learning_rate": 0.0001999996708996656,
-      "loss": 46.0,
-      "step": 10691
-    },
-    {
-      "epoch": 0.8174780664028901,
-      "grad_norm": 0.0005201590829528868,
-      "learning_rate": 0.00019999967083803922,
-      "loss": 46.0,
-      "step": 10692
-    },
-    {
-      "epoch": 0.8175545233862798,
-      "grad_norm": 0.0021525018382817507,
-      "learning_rate": 0.00019999967077640712,
-      "loss": 46.0,
-      "step": 10693
-    },
-    {
-      "epoch": 0.8176309803696695,
-      "grad_norm": 0.0022422950714826584,
-      "learning_rate": 0.00019999967071476923,
-      "loss": 46.0,
-      "step": 10694
-    },
-    {
-      "epoch": 0.8177074373530593,
-      "grad_norm": 0.0012664056848734617,
-      "learning_rate": 0.00019999967065312556,
-      "loss": 46.0,
-      "step": 10695
-    },
-    {
-      "epoch": 0.8177838943364489,
-      "grad_norm": 0.001776279415935278,
-      "learning_rate": 0.00019999967059147611,
-      "loss": 46.0,
-      "step": 10696
-    },
-    {
-      "epoch": 0.8178603513198387,
-      "grad_norm": 0.00044295971747487783,
-      "learning_rate": 0.00019999967052982092,
-      "loss": 46.0,
-      "step": 10697
-    },
-    {
-      "epoch": 0.8179368083032283,
-      "grad_norm": 0.0013613871997222304,
-      "learning_rate": 0.00019999967046815993,
-      "loss": 46.0,
-      "step": 10698
-    },
-    {
-      "epoch": 0.8180132652866181,
-      "grad_norm": 0.007926327176392078,
-      "learning_rate": 0.0001999996704064932,
-      "loss": 46.0,
-      "step": 10699
-    },
-    {
-      "epoch": 0.8180897222700079,
-      "grad_norm": 0.0012601747876033187,
-      "learning_rate": 0.0001999996703448207,
-      "loss": 46.0,
-      "step": 10700
-    },
-    {
-      "epoch": 0.8181661792533975,
-      "grad_norm": 0.0035691813100129366,
-      "learning_rate": 0.0001999996702831424,
-      "loss": 46.0,
-      "step": 10701
-    },
-    {
-      "epoch": 0.8182426362367873,
-      "grad_norm": 0.001062513911165297,
-      "learning_rate": 0.00019999967022145832,
-      "loss": 46.0,
-      "step": 10702
-    },
-    {
-      "epoch": 0.818319093220177,
-      "grad_norm": 0.0005897298105992377,
-      "learning_rate": 0.00019999967015976852,
-      "loss": 46.0,
-      "step": 10703
-    },
-    {
-      "epoch": 0.8183955502035667,
-      "grad_norm": 0.0011999744456261396,
-      "learning_rate": 0.00019999967009807294,
-      "loss": 46.0,
-      "step": 10704
-    },
-    {
-      "epoch": 0.8184720071869565,
-      "grad_norm": 0.0008478070958517492,
-      "learning_rate": 0.0001999996700363716,
-      "loss": 46.0,
-      "step": 10705
-    },
-    {
-      "epoch": 0.8185484641703462,
-      "grad_norm": 0.0010699164122343063,
-      "learning_rate": 0.00019999966997466447,
-      "loss": 46.0,
-      "step": 10706
-    },
-    {
-      "epoch": 0.8186249211537359,
-      "grad_norm": 0.0023799254558980465,
-      "learning_rate": 0.00019999966991295155,
-      "loss": 46.0,
-      "step": 10707
-    },
-    {
-      "epoch": 0.8187013781371256,
-      "grad_norm": 0.0009910025401040912,
-      "learning_rate": 0.00019999966985123288,
-      "loss": 46.0,
-      "step": 10708
-    },
-    {
-      "epoch": 0.8187778351205153,
-      "grad_norm": 0.0004909580457024276,
-      "learning_rate": 0.00019999966978950844,
-      "loss": 46.0,
-      "step": 10709
-    },
-    {
-      "epoch": 0.818854292103905,
-      "grad_norm": 0.000413259316701442,
-      "learning_rate": 0.00019999966972777825,
-      "loss": 46.0,
-      "step": 10710
-    },
-    {
-      "epoch": 0.8189307490872948,
-      "grad_norm": 0.002241626149043441,
-      "learning_rate": 0.00019999966966604226,
-      "loss": 46.0,
-      "step": 10711
-    },
-    {
-      "epoch": 0.8190072060706844,
-      "grad_norm": 0.00330294668674469,
-      "learning_rate": 0.0001999996696043005,
-      "loss": 46.0,
-      "step": 10712
-    },
-    {
-      "epoch": 0.8190836630540742,
-      "grad_norm": 0.002269588178023696,
-      "learning_rate": 0.000199999669542553,
-      "loss": 46.0,
-      "step": 10713
-    },
-    {
-      "epoch": 0.819160120037464,
-      "grad_norm": 0.0020843122620135546,
-      "learning_rate": 0.0001999996694807997,
-      "loss": 46.0,
-      "step": 10714
-    },
-    {
-      "epoch": 0.8192365770208536,
-      "grad_norm": 0.001133975456468761,
-      "learning_rate": 0.00019999966941904065,
-      "loss": 46.0,
-      "step": 10715
-    },
-    {
-      "epoch": 0.8193130340042434,
-      "grad_norm": 0.0030970703810453415,
-      "learning_rate": 0.00019999966935727584,
-      "loss": 46.0,
-      "step": 10716
-    },
-    {
-      "epoch": 0.819389490987633,
-      "grad_norm": 0.0015342723345384002,
-      "learning_rate": 0.00019999966929550524,
-      "loss": 46.0,
-      "step": 10717
-    },
-    {
-      "epoch": 0.8194659479710228,
-      "grad_norm": 0.0004633585922420025,
-      "learning_rate": 0.0001999996692337289,
-      "loss": 46.0,
-      "step": 10718
-    },
-    {
-      "epoch": 0.8195424049544126,
-      "grad_norm": 0.001683680689893663,
-      "learning_rate": 0.00019999966917194674,
-      "loss": 46.0,
-      "step": 10719
-    },
-    {
-      "epoch": 0.8196188619378022,
-      "grad_norm": 0.0003087381483055651,
-      "learning_rate": 0.00019999966911015885,
-      "loss": 46.0,
-      "step": 10720
-    },
-    {
-      "epoch": 0.819695318921192,
-      "grad_norm": 0.001977110281586647,
-      "learning_rate": 0.00019999966904836518,
-      "loss": 46.0,
-      "step": 10721
-    },
-    {
-      "epoch": 0.8197717759045817,
-      "grad_norm": 0.01460284274071455,
-      "learning_rate": 0.00019999966898656574,
-      "loss": 46.0,
-      "step": 10722
-    },
-    {
-      "epoch": 0.8198482328879714,
-      "grad_norm": 0.0020444500260055065,
-      "learning_rate": 0.00019999966892476052,
-      "loss": 46.0,
-      "step": 10723
-    },
-    {
-      "epoch": 0.8199246898713611,
-      "grad_norm": 0.0009977342560887337,
-      "learning_rate": 0.00019999966886294956,
-      "loss": 46.0,
-      "step": 10724
-    },
-    {
-      "epoch": 0.8200011468547509,
-      "grad_norm": 0.0009037595009431243,
-      "learning_rate": 0.0001999996688011328,
-      "loss": 46.0,
-      "step": 10725
-    },
-    {
-      "epoch": 0.8200776038381405,
-      "grad_norm": 0.0007984795374795794,
-      "learning_rate": 0.0001999996687393103,
-      "loss": 46.0,
-      "step": 10726
-    },
-    {
-      "epoch": 0.8201540608215303,
-      "grad_norm": 0.0008109110058285296,
-      "learning_rate": 0.00019999966867748198,
-      "loss": 46.0,
-      "step": 10727
-    },
-    {
-      "epoch": 0.82023051780492,
-      "grad_norm": 0.0009472433011978865,
-      "learning_rate": 0.00019999966861564795,
-      "loss": 46.0,
-      "step": 10728
-    },
-    {
-      "epoch": 0.8203069747883097,
-      "grad_norm": 0.00224841246381402,
-      "learning_rate": 0.0001999996685538081,
-      "loss": 46.0,
-      "step": 10729
-    },
-    {
-      "epoch": 0.8203834317716995,
-      "grad_norm": 0.00199357932433486,
-      "learning_rate": 0.0001999996684919625,
-      "loss": 46.0,
-      "step": 10730
-    },
-    {
-      "epoch": 0.8204598887550891,
-      "grad_norm": 0.013042778708040714,
-      "learning_rate": 0.00019999966843011114,
-      "loss": 46.0,
-      "step": 10731
-    },
-    {
-      "epoch": 0.8205363457384789,
-      "grad_norm": 0.0010406305082142353,
-      "learning_rate": 0.00019999966836825402,
-      "loss": 46.0,
-      "step": 10732
-    },
-    {
-      "epoch": 0.8206128027218687,
-      "grad_norm": 0.000800828798674047,
-      "learning_rate": 0.00019999966830639112,
-      "loss": 46.0,
-      "step": 10733
-    },
-    {
-      "epoch": 0.8206892597052583,
-      "grad_norm": 0.0023250156082212925,
-      "learning_rate": 0.00019999966824452243,
-      "loss": 46.0,
-      "step": 10734
-    },
-    {
-      "epoch": 0.8207657166886481,
-      "grad_norm": 0.0008186423219740391,
-      "learning_rate": 0.00019999966818264799,
-      "loss": 46.0,
-      "step": 10735
-    },
-    {
-      "epoch": 0.8208421736720378,
-      "grad_norm": 0.0007703353185206652,
-      "learning_rate": 0.00019999966812076777,
-      "loss": 46.0,
-      "step": 10736
-    },
-    {
-      "epoch": 0.8209186306554275,
-      "grad_norm": 0.0005793662858195603,
-      "learning_rate": 0.0001999996680588818,
-      "loss": 46.0,
-      "step": 10737
-    },
-    {
-      "epoch": 0.8209950876388172,
-      "grad_norm": 0.0016379449516534805,
-      "learning_rate": 0.00019999966799699005,
-      "loss": 46.0,
-      "step": 10738
-    },
-    {
-      "epoch": 0.8210715446222069,
-      "grad_norm": 0.0008216226706281304,
-      "learning_rate": 0.00019999966793509254,
-      "loss": 46.0,
-      "step": 10739
-    },
-    {
-      "epoch": 0.8211480016055966,
-      "grad_norm": 0.0015827440656721592,
-      "learning_rate": 0.00019999966787318926,
-      "loss": 46.0,
-      "step": 10740
-    },
-    {
-      "epoch": 0.8212244585889864,
-      "grad_norm": 0.0004435455775819719,
-      "learning_rate": 0.0001999996678112802,
-      "loss": 46.0,
-      "step": 10741
-    },
-    {
-      "epoch": 0.8213009155723761,
-      "grad_norm": 0.003094179555773735,
-      "learning_rate": 0.00019999966774936537,
-      "loss": 46.0,
-      "step": 10742
-    },
-    {
-      "epoch": 0.8213773725557658,
-      "grad_norm": 0.005434252321720123,
-      "learning_rate": 0.00019999966768744477,
-      "loss": 46.0,
-      "step": 10743
-    },
-    {
-      "epoch": 0.8214538295391556,
-      "grad_norm": 0.0007276431424543262,
-      "learning_rate": 0.0001999996676255184,
-      "loss": 46.0,
-      "step": 10744
-    },
-    {
-      "epoch": 0.8215302865225452,
-      "grad_norm": 0.015207141637802124,
-      "learning_rate": 0.00019999966756358624,
-      "loss": 46.0,
-      "step": 10745
-    },
-    {
-      "epoch": 0.821606743505935,
-      "grad_norm": 0.0004469765699468553,
-      "learning_rate": 0.00019999966750164835,
-      "loss": 46.0,
-      "step": 10746
-    },
-    {
-      "epoch": 0.8216832004893246,
-      "grad_norm": 0.0018683691741898656,
-      "learning_rate": 0.00019999966743970465,
-      "loss": 46.0,
-      "step": 10747
-    },
-    {
-      "epoch": 0.8217596574727144,
-      "grad_norm": 0.0024904152378439903,
-      "learning_rate": 0.0001999996673777552,
-      "loss": 46.0,
-      "step": 10748
-    },
-    {
-      "epoch": 0.8218361144561042,
-      "grad_norm": 0.0008083500433713198,
-      "learning_rate": 0.0001999996673158,
-      "loss": 46.0,
-      "step": 10749
-    },
-    {
-      "epoch": 0.8219125714394938,
-      "grad_norm": 0.0014286641962826252,
-      "learning_rate": 0.00019999966725383904,
-      "loss": 46.0,
-      "step": 10750
-    },
-    {
-      "epoch": 0.8219890284228836,
-      "grad_norm": 0.002709595952183008,
-      "learning_rate": 0.00019999966719187227,
-      "loss": 46.0,
-      "step": 10751
-    },
-    {
-      "epoch": 0.8220654854062733,
-      "grad_norm": 0.0018716215854510665,
-      "learning_rate": 0.00019999966712989974,
-      "loss": 46.0,
-      "step": 10752
-    },
-    {
-      "epoch": 0.822141942389663,
-      "grad_norm": 0.0027791548054665327,
-      "learning_rate": 0.00019999966706792143,
-      "loss": 46.0,
-      "step": 10753
-    },
-    {
-      "epoch": 0.8222183993730527,
-      "grad_norm": 0.0019235272193327546,
-      "learning_rate": 0.00019999966700593737,
-      "loss": 46.0,
-      "step": 10754
-    },
-    {
-      "epoch": 0.8222948563564425,
-      "grad_norm": 0.0007547183195129037,
-      "learning_rate": 0.00019999966694394757,
-      "loss": 46.0,
-      "step": 10755
-    },
-    {
-      "epoch": 0.8223713133398322,
-      "grad_norm": 0.0007512698648497462,
-      "learning_rate": 0.00019999966688195197,
-      "loss": 46.0,
-      "step": 10756
-    },
-    {
-      "epoch": 0.8224477703232219,
-      "grad_norm": 0.0010405313223600388,
-      "learning_rate": 0.0001999996668199506,
-      "loss": 46.0,
-      "step": 10757
-    },
-    {
-      "epoch": 0.8225242273066116,
-      "grad_norm": 0.0031745682936161757,
-      "learning_rate": 0.00019999966675794345,
-      "loss": 46.0,
-      "step": 10758
-    },
-    {
-      "epoch": 0.8226006842900013,
-      "grad_norm": 0.002374935895204544,
-      "learning_rate": 0.00019999966669593055,
-      "loss": 46.0,
-      "step": 10759
-    },
-    {
-      "epoch": 0.8226771412733911,
-      "grad_norm": 0.0010144548723474145,
-      "learning_rate": 0.00019999966663391186,
-      "loss": 46.0,
-      "step": 10760
-    },
-    {
-      "epoch": 0.8227535982567807,
-      "grad_norm": 0.005825607106089592,
-      "learning_rate": 0.00019999966657188742,
-      "loss": 46.0,
-      "step": 10761
-    },
-    {
-      "epoch": 0.8228300552401705,
-      "grad_norm": 0.0011752373538911343,
-      "learning_rate": 0.0001999996665098572,
-      "loss": 46.0,
-      "step": 10762
-    },
-    {
-      "epoch": 0.8229065122235603,
-      "grad_norm": 0.002835820661857724,
-      "learning_rate": 0.0001999996664478212,
-      "loss": 46.0,
-      "step": 10763
-    },
-    {
-      "epoch": 0.8229829692069499,
-      "grad_norm": 0.0052048806101083755,
-      "learning_rate": 0.00019999966638577945,
-      "loss": 46.0,
-      "step": 10764
-    },
-    {
-      "epoch": 0.8230594261903397,
-      "grad_norm": 0.0040244534611701965,
-      "learning_rate": 0.00019999966632373192,
-      "loss": 46.0,
-      "step": 10765
-    },
-    {
-      "epoch": 0.8231358831737294,
-      "grad_norm": 0.0019150939770042896,
-      "learning_rate": 0.00019999966626167863,
-      "loss": 46.0,
-      "step": 10766
-    },
-    {
-      "epoch": 0.8232123401571191,
-      "grad_norm": 0.005140490364283323,
-      "learning_rate": 0.00019999966619961955,
-      "loss": 46.0,
-      "step": 10767
-    },
-    {
-      "epoch": 0.8232887971405088,
-      "grad_norm": 0.0002480422263033688,
-      "learning_rate": 0.00019999966613755472,
-      "loss": 46.0,
-      "step": 10768
-    },
-    {
-      "epoch": 0.8233652541238985,
-      "grad_norm": 0.001252923859283328,
-      "learning_rate": 0.00019999966607548412,
-      "loss": 46.0,
-      "step": 10769
-    },
-    {
-      "epoch": 0.8234417111072883,
-      "grad_norm": 0.013171762228012085,
-      "learning_rate": 0.00019999966601340775,
-      "loss": 46.0,
-      "step": 10770
-    },
-    {
-      "epoch": 0.823518168090678,
-      "grad_norm": 0.0011907482985407114,
-      "learning_rate": 0.0001999996659513256,
-      "loss": 46.0,
-      "step": 10771
-    },
-    {
-      "epoch": 0.8235946250740677,
-      "grad_norm": 0.004106555134057999,
-      "learning_rate": 0.0001999996658892377,
-      "loss": 46.0,
-      "step": 10772
-    },
-    {
-      "epoch": 0.8236710820574574,
-      "grad_norm": 0.0010761055164039135,
-      "learning_rate": 0.000199999665827144,
-      "loss": 46.0,
-      "step": 10773
-    },
-    {
-      "epoch": 0.8237475390408472,
-      "grad_norm": 0.0006391893839463592,
-      "learning_rate": 0.00019999966576504457,
-      "loss": 46.0,
-      "step": 10774
-    },
-    {
-      "epoch": 0.8238239960242368,
-      "grad_norm": 0.00033305137185379863,
-      "learning_rate": 0.00019999966570293935,
-      "loss": 46.0,
-      "step": 10775
-    },
-    {
-      "epoch": 0.8239004530076266,
-      "grad_norm": 0.0011488631134852767,
-      "learning_rate": 0.00019999966564082837,
-      "loss": 46.0,
-      "step": 10776
-    },
-    {
-      "epoch": 0.8239769099910164,
-      "grad_norm": 0.0009050683584064245,
-      "learning_rate": 0.0001999996655787116,
-      "loss": 46.0,
-      "step": 10777
-    },
-    {
-      "epoch": 0.824053366974406,
-      "grad_norm": 0.00082078215200454,
-      "learning_rate": 0.0001999996655165891,
-      "loss": 46.0,
-      "step": 10778
-    },
-    {
-      "epoch": 0.8241298239577958,
-      "grad_norm": 0.0022648824378848076,
-      "learning_rate": 0.0001999996654544608,
-      "loss": 46.0,
-      "step": 10779
-    },
-    {
-      "epoch": 0.8242062809411854,
-      "grad_norm": 0.012408829294145107,
-      "learning_rate": 0.0001999996653923267,
-      "loss": 46.0,
-      "step": 10780
-    },
-    {
-      "epoch": 0.8242827379245752,
-      "grad_norm": 0.001504181302152574,
-      "learning_rate": 0.00019999966533018685,
-      "loss": 46.0,
-      "step": 10781
-    },
-    {
-      "epoch": 0.8243591949079649,
-      "grad_norm": 0.0017542120767757297,
-      "learning_rate": 0.00019999966526804125,
-      "loss": 46.0,
-      "step": 10782
-    },
-    {
-      "epoch": 0.8244356518913546,
-      "grad_norm": 0.002071010647341609,
-      "learning_rate": 0.00019999966520588988,
-      "loss": 46.0,
-      "step": 10783
-    },
-    {
-      "epoch": 0.8245121088747444,
-      "grad_norm": 0.001121000968851149,
-      "learning_rate": 0.00019999966514373273,
-      "loss": 46.0,
-      "step": 10784
-    },
-    {
-      "epoch": 0.8245885658581341,
-      "grad_norm": 0.00761817954480648,
-      "learning_rate": 0.0001999996650815698,
-      "loss": 46.0,
-      "step": 10785
-    },
-    {
-      "epoch": 0.8246650228415238,
-      "grad_norm": 0.00100034533534199,
-      "learning_rate": 0.00019999966501940114,
-      "loss": 46.0,
-      "step": 10786
-    },
-    {
-      "epoch": 0.8247414798249135,
-      "grad_norm": 0.0016872057458385825,
-      "learning_rate": 0.0001999996649572267,
-      "loss": 46.0,
-      "step": 10787
-    },
-    {
-      "epoch": 0.8248179368083032,
-      "grad_norm": 0.0004919639322906733,
-      "learning_rate": 0.00019999966489504646,
-      "loss": 46.0,
-      "step": 10788
-    },
-    {
-      "epoch": 0.8248943937916929,
-      "grad_norm": 0.003121000248938799,
-      "learning_rate": 0.00019999966483286047,
-      "loss": 46.0,
-      "step": 10789
-    },
-    {
-      "epoch": 0.8249708507750827,
-      "grad_norm": 0.0013548723654821515,
-      "learning_rate": 0.0001999996647706687,
-      "loss": 46.0,
-      "step": 10790
-    },
-    {
-      "epoch": 0.8250473077584723,
-      "grad_norm": 0.001999072963371873,
-      "learning_rate": 0.00019999966470847117,
-      "loss": 46.0,
-      "step": 10791
-    },
-    {
-      "epoch": 0.8251237647418621,
-      "grad_norm": 0.0009184314985759556,
-      "learning_rate": 0.0001999996646462679,
-      "loss": 46.0,
-      "step": 10792
-    },
-    {
-      "epoch": 0.8252002217252519,
-      "grad_norm": 0.0008142598089762032,
-      "learning_rate": 0.0001999996645840588,
-      "loss": 46.0,
-      "step": 10793
-    },
-    {
-      "epoch": 0.8252766787086415,
-      "grad_norm": 0.00563464593142271,
-      "learning_rate": 0.00019999966452184399,
-      "loss": 46.0,
-      "step": 10794
-    },
-    {
-      "epoch": 0.8253531356920313,
-      "grad_norm": 0.0011567280162125826,
-      "learning_rate": 0.00019999966445962336,
-      "loss": 46.0,
-      "step": 10795
-    },
-    {
-      "epoch": 0.825429592675421,
-      "grad_norm": 0.001093081315048039,
-      "learning_rate": 0.00019999966439739698,
-      "loss": 46.0,
-      "step": 10796
-    },
-    {
-      "epoch": 0.8255060496588107,
-      "grad_norm": 0.011824234388768673,
-      "learning_rate": 0.00019999966433516484,
-      "loss": 46.0,
-      "step": 10797
-    },
-    {
-      "epoch": 0.8255825066422005,
-      "grad_norm": 0.00643664225935936,
-      "learning_rate": 0.00019999966427292694,
-      "loss": 46.0,
-      "step": 10798
-    },
-    {
-      "epoch": 0.8256589636255901,
-      "grad_norm": 0.001147162402048707,
-      "learning_rate": 0.00019999966421068325,
-      "loss": 46.0,
-      "step": 10799
-    },
-    {
-      "epoch": 0.8257354206089799,
-      "grad_norm": 0.0072053843177855015,
-      "learning_rate": 0.00019999966414843375,
-      "loss": 46.0,
-      "step": 10800
-    },
-    {
-      "epoch": 0.8258118775923696,
-      "grad_norm": 0.0027590978424996138,
-      "learning_rate": 0.00019999966408617854,
-      "loss": 46.0,
-      "step": 10801
-    },
-    {
-      "epoch": 0.8258883345757593,
-      "grad_norm": 0.0019405200146138668,
-      "learning_rate": 0.00019999966402391755,
-      "loss": 46.0,
-      "step": 10802
-    },
-    {
-      "epoch": 0.825964791559149,
-      "grad_norm": 0.0010096848709508777,
-      "learning_rate": 0.0001999996639616508,
-      "loss": 46.0,
-      "step": 10803
-    },
-    {
-      "epoch": 0.8260412485425388,
-      "grad_norm": 0.0007796257850714028,
-      "learning_rate": 0.00019999966389937828,
-      "loss": 46.0,
-      "step": 10804
-    },
-    {
-      "epoch": 0.8261177055259284,
-      "grad_norm": 0.003733782097697258,
-      "learning_rate": 0.00019999966383709995,
-      "loss": 46.0,
-      "step": 10805
-    },
-    {
-      "epoch": 0.8261941625093182,
-      "grad_norm": 0.001648182631470263,
-      "learning_rate": 0.00019999966377481587,
-      "loss": 46.0,
-      "step": 10806
-    },
-    {
-      "epoch": 0.826270619492708,
-      "grad_norm": 0.0006677035125903785,
-      "learning_rate": 0.00019999966371252604,
-      "loss": 46.0,
-      "step": 10807
-    },
-    {
-      "epoch": 0.8263470764760976,
-      "grad_norm": 0.0012601150665432215,
-      "learning_rate": 0.0001999996636502304,
-      "loss": 46.0,
-      "step": 10808
-    },
-    {
-      "epoch": 0.8264235334594874,
-      "grad_norm": 0.0005845642881467938,
-      "learning_rate": 0.00019999966358792907,
-      "loss": 46.0,
-      "step": 10809
-    },
-    {
-      "epoch": 0.826499990442877,
-      "grad_norm": 0.0007311388035304844,
-      "learning_rate": 0.00019999966352562192,
-      "loss": 46.0,
-      "step": 10810
-    },
-    {
-      "epoch": 0.8265764474262668,
-      "grad_norm": 0.0030390354804694653,
-      "learning_rate": 0.00019999966346330897,
-      "loss": 46.0,
-      "step": 10811
-    },
-    {
-      "epoch": 0.8266529044096566,
-      "grad_norm": 0.0011061836266890168,
-      "learning_rate": 0.00019999966340099028,
-      "loss": 46.0,
-      "step": 10812
-    },
-    {
-      "epoch": 0.8267293613930462,
-      "grad_norm": 0.0008378637721762061,
-      "learning_rate": 0.0001999996633386658,
-      "loss": 46.0,
-      "step": 10813
-    },
-    {
-      "epoch": 0.826805818376436,
-      "grad_norm": 0.001671507372520864,
-      "learning_rate": 0.0001999996632763356,
-      "loss": 46.0,
-      "step": 10814
-    },
-    {
-      "epoch": 0.8268822753598257,
-      "grad_norm": 0.0006811353377997875,
-      "learning_rate": 0.0001999996632139996,
-      "loss": 46.0,
-      "step": 10815
-    },
-    {
-      "epoch": 0.8269587323432154,
-      "grad_norm": 0.0005022775731049478,
-      "learning_rate": 0.00019999966315165785,
-      "loss": 46.0,
-      "step": 10816
-    },
-    {
-      "epoch": 0.8270351893266051,
-      "grad_norm": 0.002233087783679366,
-      "learning_rate": 0.00019999966308931031,
-      "loss": 46.0,
-      "step": 10817
-    },
-    {
-      "epoch": 0.8271116463099948,
-      "grad_norm": 0.0010761056328192353,
-      "learning_rate": 0.00019999966302695698,
-      "loss": 46.0,
-      "step": 10818
-    },
-    {
-      "epoch": 0.8271881032933845,
-      "grad_norm": 0.0008984995656646788,
-      "learning_rate": 0.00019999966296459793,
-      "loss": 46.0,
-      "step": 10819
-    },
-    {
-      "epoch": 0.8272645602767743,
-      "grad_norm": 0.0017700594617053866,
-      "learning_rate": 0.00019999966290223307,
-      "loss": 46.0,
-      "step": 10820
-    },
-    {
-      "epoch": 0.827341017260164,
-      "grad_norm": 0.0027520039584487677,
-      "learning_rate": 0.00019999966283986247,
-      "loss": 46.0,
-      "step": 10821
-    },
-    {
-      "epoch": 0.8274174742435537,
-      "grad_norm": 0.0024779317900538445,
-      "learning_rate": 0.0001999996627774861,
-      "loss": 46.0,
-      "step": 10822
-    },
-    {
-      "epoch": 0.8274939312269435,
-      "grad_norm": 0.0005166740738786757,
-      "learning_rate": 0.00019999966271510393,
-      "loss": 46.0,
-      "step": 10823
-    },
-    {
-      "epoch": 0.8275703882103331,
-      "grad_norm": 0.0012331565376371145,
-      "learning_rate": 0.000199999662652716,
-      "loss": 46.0,
-      "step": 10824
-    },
-    {
-      "epoch": 0.8276468451937229,
-      "grad_norm": 0.0009226638940162957,
-      "learning_rate": 0.0001999996625903223,
-      "loss": 46.0,
-      "step": 10825
-    },
-    {
-      "epoch": 0.8277233021771127,
-      "grad_norm": 0.0009670732542872429,
-      "learning_rate": 0.00019999966252792284,
-      "loss": 46.0,
-      "step": 10826
-    },
-    {
-      "epoch": 0.8277997591605023,
-      "grad_norm": 0.00636687595397234,
-      "learning_rate": 0.00019999966246551763,
-      "loss": 46.0,
-      "step": 10827
-    },
-    {
-      "epoch": 0.8278762161438921,
-      "grad_norm": 0.003914323635399342,
-      "learning_rate": 0.00019999966240310664,
-      "loss": 46.0,
-      "step": 10828
-    },
-    {
-      "epoch": 0.8279526731272817,
-      "grad_norm": 0.0007384219206869602,
-      "learning_rate": 0.00019999966234068986,
-      "loss": 46.0,
-      "step": 10829
-    },
-    {
-      "epoch": 0.8280291301106715,
-      "grad_norm": 0.0010801958851516247,
-      "learning_rate": 0.0001999996622782673,
-      "loss": 46.0,
-      "step": 10830
-    },
-    {
-      "epoch": 0.8281055870940612,
-      "grad_norm": 0.0011917497031390667,
-      "learning_rate": 0.00019999966221583902,
-      "loss": 46.0,
-      "step": 10831
-    },
-    {
-      "epoch": 0.8281820440774509,
-      "grad_norm": 0.006292206700891256,
-      "learning_rate": 0.0001999996621534049,
-      "loss": 46.0,
-      "step": 10832
-    },
-    {
-      "epoch": 0.8282585010608406,
-      "grad_norm": 0.006909584626555443,
-      "learning_rate": 0.00019999966209096508,
-      "loss": 46.0,
-      "step": 10833
-    },
-    {
-      "epoch": 0.8283349580442304,
-      "grad_norm": 0.00443050405010581,
-      "learning_rate": 0.00019999966202851948,
-      "loss": 46.0,
-      "step": 10834
-    },
-    {
-      "epoch": 0.8284114150276201,
-      "grad_norm": 0.002551465993747115,
-      "learning_rate": 0.00019999966196606808,
-      "loss": 46.0,
-      "step": 10835
-    },
-    {
-      "epoch": 0.8284878720110098,
-      "grad_norm": 0.00040333790821023285,
-      "learning_rate": 0.00019999966190361094,
-      "loss": 46.0,
-      "step": 10836
-    },
-    {
-      "epoch": 0.8285643289943996,
-      "grad_norm": 0.0009858980774879456,
-      "learning_rate": 0.000199999661841148,
-      "loss": 46.0,
-      "step": 10837
-    },
-    {
-      "epoch": 0.8286407859777892,
-      "grad_norm": 0.0018890650244429708,
-      "learning_rate": 0.0001999996617786793,
-      "loss": 46.0,
-      "step": 10838
-    },
-    {
-      "epoch": 0.828717242961179,
-      "grad_norm": 0.0012757593067362905,
-      "learning_rate": 0.00019999966171620483,
-      "loss": 46.0,
-      "step": 10839
-    },
-    {
-      "epoch": 0.8287936999445686,
-      "grad_norm": 0.0037090429104864597,
-      "learning_rate": 0.00019999966165372461,
-      "loss": 46.0,
-      "step": 10840
-    },
-    {
-      "epoch": 0.8288701569279584,
-      "grad_norm": 0.00043109917896799743,
-      "learning_rate": 0.0001999996615912386,
-      "loss": 46.0,
-      "step": 10841
-    },
-    {
-      "epoch": 0.8289466139113482,
-      "grad_norm": 0.0008832283201627433,
-      "learning_rate": 0.00019999966152874684,
-      "loss": 46.0,
-      "step": 10842
-    },
-    {
-      "epoch": 0.8290230708947378,
-      "grad_norm": 0.005266947206109762,
-      "learning_rate": 0.00019999966146624928,
-      "loss": 46.0,
-      "step": 10843
-    },
-    {
-      "epoch": 0.8290995278781276,
-      "grad_norm": 0.0007477087783627212,
-      "learning_rate": 0.00019999966140374597,
-      "loss": 46.0,
-      "step": 10844
-    },
-    {
-      "epoch": 0.8291759848615173,
-      "grad_norm": 0.0017315337900072336,
-      "learning_rate": 0.0001999996613412369,
-      "loss": 46.0,
-      "step": 10845
-    },
-    {
-      "epoch": 0.829252441844907,
-      "grad_norm": 0.00042855628998950124,
-      "learning_rate": 0.00019999966127872204,
-      "loss": 46.0,
-      "step": 10846
-    },
-    {
-      "epoch": 0.8293288988282967,
-      "grad_norm": 0.002416914328932762,
-      "learning_rate": 0.00019999966121620144,
-      "loss": 46.0,
-      "step": 10847
-    },
-    {
-      "epoch": 0.8294053558116864,
-      "grad_norm": 0.0008425017585977912,
-      "learning_rate": 0.00019999966115367504,
-      "loss": 46.0,
-      "step": 10848
-    },
-    {
-      "epoch": 0.8294818127950762,
-      "grad_norm": 0.0006613319274038076,
-      "learning_rate": 0.00019999966109114287,
-      "loss": 46.0,
-      "step": 10849
-    },
-    {
-      "epoch": 0.8295582697784659,
-      "grad_norm": 0.0006813930813223124,
-      "learning_rate": 0.00019999966102860495,
-      "loss": 46.0,
-      "step": 10850
-    },
-    {
-      "epoch": 0.8296347267618556,
-      "grad_norm": 0.0009660691721364856,
-      "learning_rate": 0.00019999966096606126,
-      "loss": 46.0,
-      "step": 10851
-    },
-    {
-      "epoch": 0.8297111837452453,
-      "grad_norm": 0.001846193103119731,
-      "learning_rate": 0.0001999996609035118,
-      "loss": 46.0,
-      "step": 10852
-    },
-    {
-      "epoch": 0.8297876407286351,
-      "grad_norm": 0.0004581077373586595,
-      "learning_rate": 0.00019999966084095655,
-      "loss": 46.0,
-      "step": 10853
-    },
-    {
-      "epoch": 0.8298640977120247,
-      "grad_norm": 0.0007270847563631833,
-      "learning_rate": 0.00019999966077839556,
-      "loss": 46.0,
-      "step": 10854
-    },
-    {
-      "epoch": 0.8299405546954145,
-      "grad_norm": 0.003207017667591572,
-      "learning_rate": 0.00019999966071582875,
-      "loss": 46.0,
-      "step": 10855
-    },
-    {
-      "epoch": 0.8300170116788043,
-      "grad_norm": 0.0007058214396238327,
-      "learning_rate": 0.00019999966065325624,
-      "loss": 46.0,
-      "step": 10856
-    },
-    {
-      "epoch": 0.8300934686621939,
-      "grad_norm": 0.0005304698133841157,
-      "learning_rate": 0.00019999966059067788,
-      "loss": 46.0,
-      "step": 10857
-    },
-    {
-      "epoch": 0.8301699256455837,
-      "grad_norm": 0.0007276951801031828,
-      "learning_rate": 0.00019999966052809383,
-      "loss": 46.0,
-      "step": 10858
-    },
-    {
-      "epoch": 0.8302463826289733,
-      "grad_norm": 0.014067895710468292,
-      "learning_rate": 0.00019999966046550398,
-      "loss": 46.0,
-      "step": 10859
-    },
-    {
-      "epoch": 0.8303228396123631,
-      "grad_norm": 0.0038100602105259895,
-      "learning_rate": 0.00019999966040290835,
-      "loss": 46.0,
-      "step": 10860
-    },
-    {
-      "epoch": 0.8303992965957528,
-      "grad_norm": 0.0006950997631065547,
-      "learning_rate": 0.00019999966034030698,
-      "loss": 46.0,
-      "step": 10861
-    },
-    {
-      "epoch": 0.8304757535791425,
-      "grad_norm": 0.0005952907376922667,
-      "learning_rate": 0.0001999996602776998,
-      "loss": 46.0,
-      "step": 10862
-    },
-    {
-      "epoch": 0.8305522105625323,
-      "grad_norm": 0.0015130196698009968,
-      "learning_rate": 0.00019999966021508686,
-      "loss": 46.0,
-      "step": 10863
-    },
-    {
-      "epoch": 0.830628667545922,
-      "grad_norm": 0.001703130081295967,
-      "learning_rate": 0.00019999966015246817,
-      "loss": 46.0,
-      "step": 10864
-    },
-    {
-      "epoch": 0.8307051245293117,
-      "grad_norm": 0.0018298194045200944,
-      "learning_rate": 0.0001999996600898437,
-      "loss": 46.0,
-      "step": 10865
-    },
-    {
-      "epoch": 0.8307815815127014,
-      "grad_norm": 0.000880430219694972,
-      "learning_rate": 0.00019999966002721346,
-      "loss": 46.0,
-      "step": 10866
-    },
-    {
-      "epoch": 0.8308580384960912,
-      "grad_norm": 0.0017140473937615752,
-      "learning_rate": 0.00019999965996457745,
-      "loss": 46.0,
-      "step": 10867
-    },
-    {
-      "epoch": 0.8309344954794808,
-      "grad_norm": 0.002834649058058858,
-      "learning_rate": 0.0001999996599019357,
-      "loss": 46.0,
-      "step": 10868
-    },
-    {
-      "epoch": 0.8310109524628706,
-      "grad_norm": 0.0008349040290340781,
-      "learning_rate": 0.00019999965983928816,
-      "loss": 46.0,
-      "step": 10869
-    },
-    {
-      "epoch": 0.8310874094462603,
-      "grad_norm": 0.0014007369754835963,
-      "learning_rate": 0.00019999965977663482,
-      "loss": 46.0,
-      "step": 10870
-    },
-    {
-      "epoch": 0.83116386642965,
-      "grad_norm": 0.0007154003251343966,
-      "learning_rate": 0.00019999965971397572,
-      "loss": 46.0,
-      "step": 10871
-    },
-    {
-      "epoch": 0.8312403234130398,
-      "grad_norm": 0.0011900861281901598,
-      "learning_rate": 0.00019999965965131086,
-      "loss": 46.0,
-      "step": 10872
-    },
-    {
-      "epoch": 0.8313167803964294,
-      "grad_norm": 0.0006056425627321005,
-      "learning_rate": 0.00019999965958864024,
-      "loss": 46.0,
-      "step": 10873
-    },
-    {
-      "epoch": 0.8313932373798192,
-      "grad_norm": 0.001429014722816646,
-      "learning_rate": 0.00019999965952596384,
-      "loss": 46.0,
-      "step": 10874
-    },
-    {
-      "epoch": 0.8314696943632089,
-      "grad_norm": 0.001696994062513113,
-      "learning_rate": 0.00019999965946328167,
-      "loss": 46.0,
-      "step": 10875
-    },
-    {
-      "epoch": 0.8315461513465986,
-      "grad_norm": 0.0015285846311599016,
-      "learning_rate": 0.00019999965940059375,
-      "loss": 46.0,
-      "step": 10876
-    },
-    {
-      "epoch": 0.8316226083299884,
-      "grad_norm": 0.014070739969611168,
-      "learning_rate": 0.00019999965933790005,
-      "loss": 46.0,
-      "step": 10877
-    },
-    {
-      "epoch": 0.831699065313378,
-      "grad_norm": 0.0012685131514444947,
-      "learning_rate": 0.00019999965927520056,
-      "loss": 46.0,
-      "step": 10878
-    },
-    {
-      "epoch": 0.8317755222967678,
-      "grad_norm": 0.001588876242749393,
-      "learning_rate": 0.00019999965921249535,
-      "loss": 46.0,
-      "step": 10879
-    },
-    {
-      "epoch": 0.8318519792801575,
-      "grad_norm": 0.0022325317841023207,
-      "learning_rate": 0.00019999965914978434,
-      "loss": 46.0,
-      "step": 10880
-    },
-    {
-      "epoch": 0.8319284362635472,
-      "grad_norm": 0.0030568428337574005,
-      "learning_rate": 0.00019999965908706755,
-      "loss": 46.0,
-      "step": 10881
-    },
-    {
-      "epoch": 0.8320048932469369,
-      "grad_norm": 0.0018602528143674135,
-      "learning_rate": 0.00019999965902434502,
-      "loss": 46.0,
-      "step": 10882
-    },
-    {
-      "epoch": 0.8320813502303267,
-      "grad_norm": 0.003781463485211134,
-      "learning_rate": 0.00019999965896161669,
-      "loss": 46.0,
-      "step": 10883
-    },
-    {
-      "epoch": 0.8321578072137163,
-      "grad_norm": 0.001798297162167728,
-      "learning_rate": 0.00019999965889888258,
-      "loss": 46.0,
-      "step": 10884
-    },
-    {
-      "epoch": 0.8322342641971061,
-      "grad_norm": 0.001416568411514163,
-      "learning_rate": 0.00019999965883614273,
-      "loss": 46.0,
-      "step": 10885
-    },
-    {
-      "epoch": 0.8323107211804959,
-      "grad_norm": 0.0013424010248854756,
-      "learning_rate": 0.0001999996587733971,
-      "loss": 46.0,
-      "step": 10886
-    },
-    {
-      "epoch": 0.8323871781638855,
-      "grad_norm": 0.0008791974978521466,
-      "learning_rate": 0.00019999965871064573,
-      "loss": 46.0,
-      "step": 10887
-    },
-    {
-      "epoch": 0.8324636351472753,
-      "grad_norm": 0.0017793815350160003,
-      "learning_rate": 0.00019999965864788856,
-      "loss": 46.0,
-      "step": 10888
-    },
-    {
-      "epoch": 0.8325400921306649,
-      "grad_norm": 0.0031849441584199667,
-      "learning_rate": 0.00019999965858512558,
-      "loss": 46.0,
-      "step": 10889
-    },
-    {
-      "epoch": 0.8326165491140547,
-      "grad_norm": 0.0033617066219449043,
-      "learning_rate": 0.00019999965852235692,
-      "loss": 46.0,
-      "step": 10890
-    },
-    {
-      "epoch": 0.8326930060974445,
-      "grad_norm": 0.0019287305185571313,
-      "learning_rate": 0.00019999965845958243,
-      "loss": 46.0,
-      "step": 10891
-    },
-    {
-      "epoch": 0.8327694630808341,
-      "grad_norm": 0.0017537970561534166,
-      "learning_rate": 0.0001999996583968022,
-      "loss": 46.0,
-      "step": 10892
-    },
-    {
-      "epoch": 0.8328459200642239,
-      "grad_norm": 0.0013972556916996837,
-      "learning_rate": 0.00019999965833401617,
-      "loss": 46.0,
-      "step": 10893
-    },
-    {
-      "epoch": 0.8329223770476136,
-      "grad_norm": 0.002614380093291402,
-      "learning_rate": 0.0001999996582712244,
-      "loss": 46.0,
-      "step": 10894
-    },
-    {
-      "epoch": 0.8329988340310033,
-      "grad_norm": 0.0013157302746549249,
-      "learning_rate": 0.00019999965820842683,
-      "loss": 46.0,
-      "step": 10895
-    },
-    {
-      "epoch": 0.833075291014393,
-      "grad_norm": 0.003121002810075879,
-      "learning_rate": 0.00019999965814562352,
-      "loss": 46.0,
-      "step": 10896
-    },
-    {
-      "epoch": 0.8331517479977828,
-      "grad_norm": 0.0009520279709249735,
-      "learning_rate": 0.00019999965808281442,
-      "loss": 46.0,
-      "step": 10897
-    },
-    {
-      "epoch": 0.8332282049811724,
-      "grad_norm": 0.000802506459876895,
-      "learning_rate": 0.00019999965801999957,
-      "loss": 46.0,
-      "step": 10898
-    },
-    {
-      "epoch": 0.8333046619645622,
-      "grad_norm": 0.0016850921092554927,
-      "learning_rate": 0.00019999965795717891,
-      "loss": 46.0,
-      "step": 10899
-    },
-    {
-      "epoch": 0.8333811189479519,
-      "grad_norm": 0.002314093057066202,
-      "learning_rate": 0.00019999965789435254,
-      "loss": 46.0,
-      "step": 10900
-    },
-    {
-      "epoch": 0.8334575759313416,
-      "grad_norm": 0.0005904283025301993,
-      "learning_rate": 0.00019999965783152037,
-      "loss": 46.0,
-      "step": 10901
-    },
-    {
-      "epoch": 0.8335340329147314,
-      "grad_norm": 0.001012057880870998,
-      "learning_rate": 0.00019999965776868242,
-      "loss": 46.0,
-      "step": 10902
-    },
-    {
-      "epoch": 0.833610489898121,
-      "grad_norm": 0.0033467928878962994,
-      "learning_rate": 0.00019999965770583873,
-      "loss": 46.0,
-      "step": 10903
-    },
-    {
-      "epoch": 0.8336869468815108,
-      "grad_norm": 0.0008123525767587125,
-      "learning_rate": 0.0001999996576429892,
-      "loss": 46.0,
-      "step": 10904
-    },
-    {
-      "epoch": 0.8337634038649006,
-      "grad_norm": 0.002545438939705491,
-      "learning_rate": 0.000199999657580134,
-      "loss": 46.0,
-      "step": 10905
-    },
-    {
-      "epoch": 0.8338398608482902,
-      "grad_norm": 0.003804426407441497,
-      "learning_rate": 0.000199999657517273,
-      "loss": 46.0,
-      "step": 10906
-    },
-    {
-      "epoch": 0.83391631783168,
-      "grad_norm": 0.0008537026587873697,
-      "learning_rate": 0.0001999996574544062,
-      "loss": 46.0,
-      "step": 10907
-    },
-    {
-      "epoch": 0.8339927748150697,
-      "grad_norm": 0.0052169146947562695,
-      "learning_rate": 0.00019999965739153365,
-      "loss": 46.0,
-      "step": 10908
-    },
-    {
-      "epoch": 0.8340692317984594,
-      "grad_norm": 0.000630547758191824,
-      "learning_rate": 0.00019999965732865531,
-      "loss": 46.0,
-      "step": 10909
-    },
-    {
-      "epoch": 0.8341456887818491,
-      "grad_norm": 0.0006478666327893734,
-      "learning_rate": 0.00019999965726577124,
-      "loss": 46.0,
-      "step": 10910
-    },
-    {
-      "epoch": 0.8342221457652388,
-      "grad_norm": 0.0011280137114226818,
-      "learning_rate": 0.00019999965720288136,
-      "loss": 46.0,
-      "step": 10911
-    },
-    {
-      "epoch": 0.8342986027486285,
-      "grad_norm": 0.0007642895798198879,
-      "learning_rate": 0.00019999965713998573,
-      "loss": 46.0,
-      "step": 10912
-    },
-    {
-      "epoch": 0.8343750597320183,
-      "grad_norm": 0.0017754929140210152,
-      "learning_rate": 0.00019999965707708433,
-      "loss": 46.0,
-      "step": 10913
-    },
-    {
-      "epoch": 0.834451516715408,
-      "grad_norm": 0.0012570193503051996,
-      "learning_rate": 0.00019999965701417716,
-      "loss": 46.0,
-      "step": 10914
-    },
-    {
-      "epoch": 0.8345279736987977,
-      "grad_norm": 0.007381140254437923,
-      "learning_rate": 0.00019999965695126422,
-      "loss": 46.0,
-      "step": 10915
-    },
-    {
-      "epoch": 0.8346044306821875,
-      "grad_norm": 0.0016811782261356711,
-      "learning_rate": 0.0001999996568883455,
-      "loss": 46.0,
-      "step": 10916
-    },
-    {
-      "epoch": 0.8346808876655771,
-      "grad_norm": 0.0025787365157157183,
-      "learning_rate": 0.00019999965682542103,
-      "loss": 46.0,
-      "step": 10917
-    },
-    {
-      "epoch": 0.8347573446489669,
-      "grad_norm": 0.002313877223059535,
-      "learning_rate": 0.00019999965676249077,
-      "loss": 46.0,
-      "step": 10918
-    },
-    {
-      "epoch": 0.8348338016323565,
-      "grad_norm": 0.000879984232597053,
-      "learning_rate": 0.00019999965669955473,
-      "loss": 46.0,
-      "step": 10919
-    },
-    {
-      "epoch": 0.8349102586157463,
-      "grad_norm": 0.0037839370779693127,
-      "learning_rate": 0.00019999965663661294,
-      "loss": 46.0,
-      "step": 10920
-    },
-    {
-      "epoch": 0.8349867155991361,
-      "grad_norm": 0.00028221411048434675,
-      "learning_rate": 0.00019999965657366538,
-      "loss": 46.0,
-      "step": 10921
-    },
-    {
-      "epoch": 0.8350631725825257,
-      "grad_norm": 0.002603070577606559,
-      "learning_rate": 0.00019999965651071205,
-      "loss": 46.0,
-      "step": 10922
-    },
-    {
-      "epoch": 0.8351396295659155,
-      "grad_norm": 0.0010584990959614515,
-      "learning_rate": 0.00019999965644775297,
-      "loss": 46.0,
-      "step": 10923
-    },
-    {
-      "epoch": 0.8352160865493052,
-      "grad_norm": 0.0024331440217792988,
-      "learning_rate": 0.00019999965638478812,
-      "loss": 46.0,
-      "step": 10924
-    },
-    {
-      "epoch": 0.8352925435326949,
-      "grad_norm": 0.0020674951374530792,
-      "learning_rate": 0.00019999965632181747,
-      "loss": 46.0,
-      "step": 10925
-    },
-    {
-      "epoch": 0.8353690005160846,
-      "grad_norm": 0.0007721393485553563,
-      "learning_rate": 0.00019999965625884107,
-      "loss": 46.0,
-      "step": 10926
-    },
-    {
-      "epoch": 0.8354454574994744,
-      "grad_norm": 0.005738811567425728,
-      "learning_rate": 0.0001999996561958589,
-      "loss": 46.0,
-      "step": 10927
-    },
-    {
-      "epoch": 0.8355219144828641,
-      "grad_norm": 0.0010936089092865586,
-      "learning_rate": 0.00019999965613287098,
-      "loss": 46.0,
-      "step": 10928
-    },
-    {
-      "epoch": 0.8355983714662538,
-      "grad_norm": 0.0014200042933225632,
-      "learning_rate": 0.00019999965606987724,
-      "loss": 46.0,
-      "step": 10929
-    },
-    {
-      "epoch": 0.8356748284496435,
-      "grad_norm": 0.0028505187947303057,
-      "learning_rate": 0.00019999965600687775,
-      "loss": 46.0,
-      "step": 10930
-    },
-    {
-      "epoch": 0.8357512854330332,
-      "grad_norm": 0.003321971744298935,
-      "learning_rate": 0.0001999996559438725,
-      "loss": 46.0,
-      "step": 10931
-    },
-    {
-      "epoch": 0.835827742416423,
-      "grad_norm": 0.00102518021594733,
-      "learning_rate": 0.0001999996558808615,
-      "loss": 46.0,
-      "step": 10932
-    },
-    {
-      "epoch": 0.8359041993998126,
-      "grad_norm": 0.0007095005712471902,
-      "learning_rate": 0.0001999996558178447,
-      "loss": 46.0,
-      "step": 10933
-    },
-    {
-      "epoch": 0.8359806563832024,
-      "grad_norm": 0.0009429772035218775,
-      "learning_rate": 0.00019999965575482213,
-      "loss": 46.0,
-      "step": 10934
-    },
-    {
-      "epoch": 0.8360571133665922,
-      "grad_norm": 0.0004319078288972378,
-      "learning_rate": 0.0001999996556917938,
-      "loss": 46.0,
-      "step": 10935
-    },
-    {
-      "epoch": 0.8361335703499818,
-      "grad_norm": 0.0011591074289754033,
-      "learning_rate": 0.00019999965562875972,
-      "loss": 46.0,
-      "step": 10936
-    },
-    {
-      "epoch": 0.8362100273333716,
-      "grad_norm": 0.0014750271802768111,
-      "learning_rate": 0.00019999965556571984,
-      "loss": 46.0,
-      "step": 10937
-    },
-    {
-      "epoch": 0.8362864843167613,
-      "grad_norm": 0.0008080319967120886,
-      "learning_rate": 0.0001999996555026742,
-      "loss": 46.0,
-      "step": 10938
-    },
-    {
-      "epoch": 0.836362941300151,
-      "grad_norm": 0.0018655186286196113,
-      "learning_rate": 0.0001999996554396228,
-      "loss": 46.0,
-      "step": 10939
-    },
-    {
-      "epoch": 0.8364393982835407,
-      "grad_norm": 0.0009518834995105863,
-      "learning_rate": 0.00019999965537656562,
-      "loss": 46.0,
-      "step": 10940
-    },
-    {
-      "epoch": 0.8365158552669304,
-      "grad_norm": 0.0027765887789428234,
-      "learning_rate": 0.00019999965531350267,
-      "loss": 46.0,
-      "step": 10941
-    },
-    {
-      "epoch": 0.8365923122503202,
-      "grad_norm": 0.008856285363435745,
-      "learning_rate": 0.00019999965525043396,
-      "loss": 46.0,
-      "step": 10942
-    },
-    {
-      "epoch": 0.8366687692337099,
-      "grad_norm": 0.0009769691387191415,
-      "learning_rate": 0.0001999996551873595,
-      "loss": 46.0,
-      "step": 10943
-    },
-    {
-      "epoch": 0.8367452262170996,
-      "grad_norm": 0.001627656165510416,
-      "learning_rate": 0.00019999965512427923,
-      "loss": 46.0,
-      "step": 10944
-    },
-    {
-      "epoch": 0.8368216832004893,
-      "grad_norm": 0.0008649210794828832,
-      "learning_rate": 0.0001999996550611932,
-      "loss": 46.0,
-      "step": 10945
-    },
-    {
-      "epoch": 0.8368981401838791,
-      "grad_norm": 0.0009970496175810695,
-      "learning_rate": 0.0001999996549981014,
-      "loss": 46.0,
-      "step": 10946
-    },
-    {
-      "epoch": 0.8369745971672687,
-      "grad_norm": 0.0019105859100818634,
-      "learning_rate": 0.00019999965493500382,
-      "loss": 46.0,
-      "step": 10947
-    },
-    {
-      "epoch": 0.8370510541506585,
-      "grad_norm": 0.003146178089082241,
-      "learning_rate": 0.0001999996548719005,
-      "loss": 46.0,
-      "step": 10948
-    },
-    {
-      "epoch": 0.8371275111340482,
-      "grad_norm": 0.0014504408463835716,
-      "learning_rate": 0.0001999996548087914,
-      "loss": 46.0,
-      "step": 10949
-    },
-    {
-      "epoch": 0.8372039681174379,
-      "grad_norm": 0.0009967993246391416,
-      "learning_rate": 0.00019999965474567653,
-      "loss": 46.0,
-      "step": 10950
-    },
-    {
-      "epoch": 0.8372804251008277,
-      "grad_norm": 0.001014632172882557,
-      "learning_rate": 0.0001999996546825559,
-      "loss": 46.0,
-      "step": 10951
-    },
-    {
-      "epoch": 0.8373568820842173,
-      "grad_norm": 0.0010587768629193306,
-      "learning_rate": 0.00019999965461942946,
-      "loss": 46.0,
-      "step": 10952
-    },
-    {
-      "epoch": 0.8374333390676071,
-      "grad_norm": 0.003790617687627673,
-      "learning_rate": 0.0001999996545562973,
-      "loss": 46.0,
-      "step": 10953
-    },
-    {
-      "epoch": 0.8375097960509968,
-      "grad_norm": 0.001789894886314869,
-      "learning_rate": 0.00019999965449315937,
-      "loss": 46.0,
-      "step": 10954
-    },
-    {
-      "epoch": 0.8375862530343865,
-      "grad_norm": 0.0018467131303623319,
-      "learning_rate": 0.00019999965443001565,
-      "loss": 46.0,
-      "step": 10955
-    },
-    {
-      "epoch": 0.8376627100177763,
-      "grad_norm": 0.0013756228145211935,
-      "learning_rate": 0.00019999965436686613,
-      "loss": 46.0,
-      "step": 10956
-    },
-    {
-      "epoch": 0.837739167001166,
-      "grad_norm": 0.00186286517418921,
-      "learning_rate": 0.0001999996543037109,
-      "loss": 46.0,
-      "step": 10957
-    },
-    {
-      "epoch": 0.8378156239845557,
-      "grad_norm": 0.0010459361365064979,
-      "learning_rate": 0.00019999965424054986,
-      "loss": 46.0,
-      "step": 10958
-    },
-    {
-      "epoch": 0.8378920809679454,
-      "grad_norm": 0.005547845270484686,
-      "learning_rate": 0.00019999965417738307,
-      "loss": 46.0,
-      "step": 10959
-    },
-    {
-      "epoch": 0.8379685379513351,
-      "grad_norm": 0.0012270724400877953,
-      "learning_rate": 0.0001999996541142105,
-      "loss": 46.0,
-      "step": 10960
-    },
-    {
-      "epoch": 0.8380449949347248,
-      "grad_norm": 0.0010208089370280504,
-      "learning_rate": 0.00019999965405103219,
-      "loss": 46.0,
-      "step": 10961
-    },
-    {
-      "epoch": 0.8381214519181146,
-      "grad_norm": 0.0015101042808964849,
-      "learning_rate": 0.00019999965398784805,
-      "loss": 46.0,
-      "step": 10962
-    },
-    {
-      "epoch": 0.8381979089015043,
-      "grad_norm": 0.00341527396813035,
-      "learning_rate": 0.00019999965392465818,
-      "loss": 46.0,
-      "step": 10963
-    },
-    {
-      "epoch": 0.838274365884894,
-      "grad_norm": 0.0078042359091341496,
-      "learning_rate": 0.00019999965386146253,
-      "loss": 46.0,
-      "step": 10964
-    },
-    {
-      "epoch": 0.8383508228682838,
-      "grad_norm": 0.000756177119910717,
-      "learning_rate": 0.00019999965379826113,
-      "loss": 46.0,
-      "step": 10965
-    },
-    {
-      "epoch": 0.8384272798516734,
-      "grad_norm": 0.0030392338521778584,
-      "learning_rate": 0.00019999965373505396,
-      "loss": 46.0,
-      "step": 10966
-    },
-    {
-      "epoch": 0.8385037368350632,
-      "grad_norm": 0.0022330654319375753,
-      "learning_rate": 0.000199999653671841,
-      "loss": 46.0,
-      "step": 10967
-    },
-    {
-      "epoch": 0.838580193818453,
-      "grad_norm": 0.0009384745499119163,
-      "learning_rate": 0.00019999965360862227,
-      "loss": 46.0,
-      "step": 10968
-    },
-    {
-      "epoch": 0.8386566508018426,
-      "grad_norm": 0.0008377086487598717,
-      "learning_rate": 0.00019999965354539778,
-      "loss": 46.0,
-      "step": 10969
-    },
-    {
-      "epoch": 0.8387331077852324,
-      "grad_norm": 0.007736325729638338,
-      "learning_rate": 0.00019999965348216754,
-      "loss": 46.0,
-      "step": 10970
-    },
-    {
-      "epoch": 0.838809564768622,
-      "grad_norm": 0.0012282304232940078,
-      "learning_rate": 0.0001999996534189315,
-      "loss": 46.0,
-      "step": 10971
-    },
-    {
-      "epoch": 0.8388860217520118,
-      "grad_norm": 0.001501261256635189,
-      "learning_rate": 0.0001999996533556897,
-      "loss": 46.0,
-      "step": 10972
-    },
-    {
-      "epoch": 0.8389624787354015,
-      "grad_norm": 0.0007189962780103087,
-      "learning_rate": 0.00019999965329244214,
-      "loss": 46.0,
-      "step": 10973
-    },
-    {
-      "epoch": 0.8390389357187912,
-      "grad_norm": 0.0009197764447890222,
-      "learning_rate": 0.00019999965322918878,
-      "loss": 46.0,
-      "step": 10974
-    },
-    {
-      "epoch": 0.8391153927021809,
-      "grad_norm": 0.0011863221880048513,
-      "learning_rate": 0.0001999996531659297,
-      "loss": 46.0,
-      "step": 10975
-    },
-    {
-      "epoch": 0.8391918496855707,
-      "grad_norm": 0.0017392291920259595,
-      "learning_rate": 0.0001999996531026648,
-      "loss": 46.0,
-      "step": 10976
-    },
-    {
-      "epoch": 0.8392683066689604,
-      "grad_norm": 0.0011469278251752257,
-      "learning_rate": 0.00019999965303939417,
-      "loss": 46.0,
-      "step": 10977
-    },
-    {
-      "epoch": 0.8393447636523501,
-      "grad_norm": 0.000676978612318635,
-      "learning_rate": 0.00019999965297611775,
-      "loss": 46.0,
-      "step": 10978
-    },
-    {
-      "epoch": 0.8394212206357398,
-      "grad_norm": 0.0029597305692732334,
-      "learning_rate": 0.00019999965291283558,
-      "loss": 46.0,
-      "step": 10979
-    },
-    {
-      "epoch": 0.8394976776191295,
-      "grad_norm": 0.002962978556752205,
-      "learning_rate": 0.0001999996528495476,
-      "loss": 46.0,
-      "step": 10980
-    },
-    {
-      "epoch": 0.8395741346025193,
-      "grad_norm": 0.007286848500370979,
-      "learning_rate": 0.0001999996527862539,
-      "loss": 46.0,
-      "step": 10981
-    },
-    {
-      "epoch": 0.8396505915859089,
-      "grad_norm": 0.000952498521655798,
-      "learning_rate": 0.0001999996527229544,
-      "loss": 46.0,
-      "step": 10982
-    },
-    {
-      "epoch": 0.8397270485692987,
-      "grad_norm": 0.000845861854031682,
-      "learning_rate": 0.00019999965265964914,
-      "loss": 46.0,
-      "step": 10983
-    },
-    {
-      "epoch": 0.8398035055526885,
-      "grad_norm": 0.0031289902981370687,
-      "learning_rate": 0.00019999965259633807,
-      "loss": 46.0,
-      "step": 10984
-    },
-    {
-      "epoch": 0.8398799625360781,
-      "grad_norm": 0.0004959868383593857,
-      "learning_rate": 0.00019999965253302132,
-      "loss": 46.0,
-      "step": 10985
-    },
-    {
-      "epoch": 0.8399564195194679,
-      "grad_norm": 0.002900330815464258,
-      "learning_rate": 0.0001999996524696987,
-      "loss": 46.0,
-      "step": 10986
-    },
-    {
-      "epoch": 0.8400328765028576,
-      "grad_norm": 0.001075619482435286,
-      "learning_rate": 0.0001999996524063704,
-      "loss": 46.0,
-      "step": 10987
-    },
-    {
-      "epoch": 0.8401093334862473,
-      "grad_norm": 0.0013092114822939038,
-      "learning_rate": 0.00019999965234303625,
-      "loss": 46.0,
-      "step": 10988
-    },
-    {
-      "epoch": 0.840185790469637,
-      "grad_norm": 0.0011865614214912057,
-      "learning_rate": 0.0001999996522796964,
-      "loss": 46.0,
-      "step": 10989
-    },
-    {
-      "epoch": 0.8402622474530267,
-      "grad_norm": 0.00552182924002409,
-      "learning_rate": 0.00019999965221635075,
-      "loss": 46.0,
-      "step": 10990
-    },
-    {
-      "epoch": 0.8403387044364165,
-      "grad_norm": 0.000710930849891156,
-      "learning_rate": 0.00019999965215299935,
-      "loss": 46.0,
-      "step": 10991
-    },
-    {
-      "epoch": 0.8404151614198062,
-      "grad_norm": 0.0005091328639537096,
-      "learning_rate": 0.00019999965208964215,
-      "loss": 46.0,
-      "step": 10992
-    },
-    {
-      "epoch": 0.8404916184031959,
-      "grad_norm": 0.006268938537687063,
-      "learning_rate": 0.00019999965202627918,
-      "loss": 46.0,
-      "step": 10993
-    },
-    {
-      "epoch": 0.8405680753865856,
-      "grad_norm": 0.0009775500511750579,
-      "learning_rate": 0.00019999965196291046,
-      "loss": 46.0,
-      "step": 10994
-    },
-    {
-      "epoch": 0.8406445323699754,
-      "grad_norm": 0.0009568944224156439,
-      "learning_rate": 0.00019999965189953595,
-      "loss": 46.0,
-      "step": 10995
-    },
-    {
-      "epoch": 0.840720989353365,
-      "grad_norm": 0.000858356012031436,
-      "learning_rate": 0.0001999996518361557,
-      "loss": 46.0,
-      "step": 10996
-    },
-    {
-      "epoch": 0.8407974463367548,
-      "grad_norm": 0.0009429023484699428,
-      "learning_rate": 0.00019999965177276967,
-      "loss": 46.0,
-      "step": 10997
-    },
-    {
-      "epoch": 0.8408739033201446,
-      "grad_norm": 0.0012169571127742529,
-      "learning_rate": 0.00019999965170937784,
-      "loss": 46.0,
-      "step": 10998
-    },
-    {
-      "epoch": 0.8409503603035342,
-      "grad_norm": 0.0035304888151586056,
-      "learning_rate": 0.00019999965164598028,
-      "loss": 46.0,
-      "step": 10999
-    },
-    {
-      "epoch": 0.841026817286924,
-      "grad_norm": 0.0006491626263596117,
-      "learning_rate": 0.00019999965158257695,
-      "loss": 46.0,
-      "step": 11000
-    },
-    {
-      "epoch": 0.8411032742703136,
-      "grad_norm": 0.000558013969566673,
-      "learning_rate": 0.00019999965151916782,
-      "loss": 46.0,
-      "step": 11001
-    },
-    {
-      "epoch": 0.8411797312537034,
-      "grad_norm": 0.0011213729158043861,
-      "learning_rate": 0.00019999965145575294,
-      "loss": 46.0,
-      "step": 11002
-    },
-    {
-      "epoch": 0.8412561882370931,
-      "grad_norm": 0.0007211608462966979,
-      "learning_rate": 0.0001999996513923323,
-      "loss": 46.0,
-      "step": 11003
-    },
-    {
-      "epoch": 0.8413326452204828,
-      "grad_norm": 0.0010748731438070536,
-      "learning_rate": 0.00019999965132890587,
-      "loss": 46.0,
-      "step": 11004
-    },
-    {
-      "epoch": 0.8414091022038725,
-      "grad_norm": 0.00047332001850008965,
-      "learning_rate": 0.0001999996512654737,
-      "loss": 46.0,
-      "step": 11005
-    },
-    {
-      "epoch": 0.8414855591872623,
-      "grad_norm": 0.0031150744762271643,
-      "learning_rate": 0.00019999965120203573,
-      "loss": 46.0,
-      "step": 11006
-    },
-    {
-      "epoch": 0.841562016170652,
-      "grad_norm": 0.0020026129204779863,
-      "learning_rate": 0.000199999651138592,
-      "loss": 46.0,
-      "step": 11007
-    },
-    {
-      "epoch": 0.8416384731540417,
-      "grad_norm": 0.003529926994815469,
-      "learning_rate": 0.00019999965107514252,
-      "loss": 46.0,
-      "step": 11008
-    },
-    {
-      "epoch": 0.8417149301374315,
-      "grad_norm": 0.0034201047383248806,
-      "learning_rate": 0.00019999965101168723,
-      "loss": 46.0,
-      "step": 11009
-    },
-    {
-      "epoch": 0.8417913871208211,
-      "grad_norm": 0.0007661358686164021,
-      "learning_rate": 0.0001999996509482262,
-      "loss": 46.0,
-      "step": 11010
-    },
-    {
-      "epoch": 0.8418678441042109,
-      "grad_norm": 0.0006424448220059276,
-      "learning_rate": 0.0001999996508847594,
-      "loss": 46.0,
-      "step": 11011
-    },
-    {
-      "epoch": 0.8419443010876005,
-      "grad_norm": 0.0010926688555628061,
-      "learning_rate": 0.00019999965082128683,
-      "loss": 46.0,
-      "step": 11012
-    },
-    {
-      "epoch": 0.8420207580709903,
-      "grad_norm": 0.000931408314500004,
-      "learning_rate": 0.00019999965075780847,
-      "loss": 46.0,
-      "step": 11013
-    },
-    {
-      "epoch": 0.8420972150543801,
-      "grad_norm": 0.0004905432579107583,
-      "learning_rate": 0.00019999965069432434,
-      "loss": 46.0,
-      "step": 11014
-    },
-    {
-      "epoch": 0.8421736720377697,
-      "grad_norm": 0.0017119819531217217,
-      "learning_rate": 0.00019999965063083447,
-      "loss": 46.0,
-      "step": 11015
-    },
-    {
-      "epoch": 0.8422501290211595,
-      "grad_norm": 0.0004936507903039455,
-      "learning_rate": 0.0001999996505673388,
-      "loss": 46.0,
-      "step": 11016
-    },
-    {
-      "epoch": 0.8423265860045492,
-      "grad_norm": 0.004982360173016787,
-      "learning_rate": 0.00019999965050383737,
-      "loss": 46.0,
-      "step": 11017
-    },
-    {
-      "epoch": 0.8424030429879389,
-      "grad_norm": 0.001137531828135252,
-      "learning_rate": 0.0001999996504403302,
-      "loss": 46.0,
-      "step": 11018
-    },
-    {
-      "epoch": 0.8424794999713286,
-      "grad_norm": 0.004952119197696447,
-      "learning_rate": 0.00019999965037681726,
-      "loss": 46.0,
-      "step": 11019
-    },
-    {
-      "epoch": 0.8425559569547183,
-      "grad_norm": 0.0004241155693307519,
-      "learning_rate": 0.0001999996503132985,
-      "loss": 46.0,
-      "step": 11020
-    },
-    {
-      "epoch": 0.8426324139381081,
-      "grad_norm": 0.0018657136242836714,
-      "learning_rate": 0.000199999650249774,
-      "loss": 46.0,
-      "step": 11021
-    },
-    {
-      "epoch": 0.8427088709214978,
-      "grad_norm": 0.0008627755451016128,
-      "learning_rate": 0.00019999965018624373,
-      "loss": 46.0,
-      "step": 11022
-    },
-    {
-      "epoch": 0.8427853279048875,
-      "grad_norm": 0.00034224175033159554,
-      "learning_rate": 0.0001999996501227077,
-      "loss": 46.0,
-      "step": 11023
-    },
-    {
-      "epoch": 0.8428617848882772,
-      "grad_norm": 0.0003670469159260392,
-      "learning_rate": 0.0001999996500591659,
-      "loss": 46.0,
-      "step": 11024
-    },
-    {
-      "epoch": 0.842938241871667,
-      "grad_norm": 0.0015784529969096184,
-      "learning_rate": 0.0001999996499956183,
-      "loss": 46.0,
-      "step": 11025
-    },
-    {
-      "epoch": 0.8430146988550566,
-      "grad_norm": 0.0021029640920460224,
-      "learning_rate": 0.00019999964993206497,
-      "loss": 46.0,
-      "step": 11026
-    },
-    {
-      "epoch": 0.8430911558384464,
-      "grad_norm": 0.009369504638016224,
-      "learning_rate": 0.00019999964986850584,
-      "loss": 46.0,
-      "step": 11027
-    },
-    {
-      "epoch": 0.8431676128218362,
-      "grad_norm": 0.0016328691272065043,
-      "learning_rate": 0.00019999964980494094,
-      "loss": 46.0,
-      "step": 11028
-    },
-    {
-      "epoch": 0.8432440698052258,
-      "grad_norm": 0.0029586199671030045,
-      "learning_rate": 0.00019999964974137032,
-      "loss": 46.0,
-      "step": 11029
-    },
-    {
-      "epoch": 0.8433205267886156,
-      "grad_norm": 0.001808575470931828,
-      "learning_rate": 0.00019999964967779387,
-      "loss": 46.0,
-      "step": 11030
-    },
-    {
-      "epoch": 0.8433969837720052,
-      "grad_norm": 0.0016076787142083049,
-      "learning_rate": 0.00019999964961421167,
-      "loss": 46.0,
-      "step": 11031
-    },
-    {
-      "epoch": 0.843473440755395,
-      "grad_norm": 0.00102371780667454,
-      "learning_rate": 0.0001999996495506237,
-      "loss": 46.0,
-      "step": 11032
-    },
-    {
-      "epoch": 0.8435498977387847,
-      "grad_norm": 0.0009408775367774069,
-      "learning_rate": 0.00019999964948703,
-      "loss": 46.0,
-      "step": 11033
-    },
-    {
-      "epoch": 0.8436263547221744,
-      "grad_norm": 0.004709445405751467,
-      "learning_rate": 0.00019999964942343044,
-      "loss": 46.0,
-      "step": 11034
-    },
-    {
-      "epoch": 0.8437028117055642,
-      "grad_norm": 0.0003644133103080094,
-      "learning_rate": 0.0001999996493598252,
-      "loss": 46.0,
-      "step": 11035
-    },
-    {
-      "epoch": 0.8437792686889539,
-      "grad_norm": 0.0007538195350207388,
-      "learning_rate": 0.00019999964929621415,
-      "loss": 46.0,
-      "step": 11036
-    },
-    {
-      "epoch": 0.8438557256723436,
-      "grad_norm": 0.004987620282918215,
-      "learning_rate": 0.00019999964923259734,
-      "loss": 46.0,
-      "step": 11037
-    },
-    {
-      "epoch": 0.8439321826557333,
-      "grad_norm": 0.003025482874363661,
-      "learning_rate": 0.00019999964916897475,
-      "loss": 46.0,
-      "step": 11038
-    },
-    {
-      "epoch": 0.8440086396391231,
-      "grad_norm": 0.0007983259274624288,
-      "learning_rate": 0.0001999996491053464,
-      "loss": 46.0,
-      "step": 11039
-    },
-    {
-      "epoch": 0.8440850966225127,
-      "grad_norm": 0.0033262609504163265,
-      "learning_rate": 0.0001999996490417123,
-      "loss": 46.0,
-      "step": 11040
-    },
-    {
-      "epoch": 0.8441615536059025,
-      "grad_norm": 0.0013376990100368857,
-      "learning_rate": 0.0001999996489780724,
-      "loss": 46.0,
-      "step": 11041
-    },
-    {
-      "epoch": 0.8442380105892922,
-      "grad_norm": 0.000613740470726043,
-      "learning_rate": 0.00019999964891442672,
-      "loss": 46.0,
-      "step": 11042
-    },
-    {
-      "epoch": 0.8443144675726819,
-      "grad_norm": 0.0005987072945572436,
-      "learning_rate": 0.00019999964885077532,
-      "loss": 46.0,
-      "step": 11043
-    },
-    {
-      "epoch": 0.8443909245560717,
-      "grad_norm": 0.0016557784983888268,
-      "learning_rate": 0.00019999964878711813,
-      "loss": 46.0,
-      "step": 11044
-    },
-    {
-      "epoch": 0.8444673815394613,
-      "grad_norm": 0.003649662947282195,
-      "learning_rate": 0.00019999964872345513,
-      "loss": 46.0,
-      "step": 11045
-    },
-    {
-      "epoch": 0.8445438385228511,
-      "grad_norm": 0.005795789882540703,
-      "learning_rate": 0.00019999964865978641,
-      "loss": 46.0,
-      "step": 11046
-    },
-    {
-      "epoch": 0.8446202955062408,
-      "grad_norm": 0.0015953467227518559,
-      "learning_rate": 0.0001999996485961119,
-      "loss": 46.0,
-      "step": 11047
-    },
-    {
-      "epoch": 0.8446967524896305,
-      "grad_norm": 0.02038198709487915,
-      "learning_rate": 0.00019999964853243164,
-      "loss": 46.0,
-      "step": 11048
-    },
-    {
-      "epoch": 0.8447732094730203,
-      "grad_norm": 0.00900246761739254,
-      "learning_rate": 0.0001999996484687456,
-      "loss": 46.0,
-      "step": 11049
-    },
-    {
-      "epoch": 0.8448496664564099,
-      "grad_norm": 0.0009198650950565934,
-      "learning_rate": 0.0001999996484050538,
-      "loss": 46.0,
-      "step": 11050
-    },
-    {
-      "epoch": 0.8449261234397997,
-      "grad_norm": 0.004574987571686506,
-      "learning_rate": 0.00019999964834135618,
-      "loss": 46.0,
-      "step": 11051
-    },
-    {
-      "epoch": 0.8450025804231894,
-      "grad_norm": 0.002760454546660185,
-      "learning_rate": 0.00019999964827765283,
-      "loss": 46.0,
-      "step": 11052
-    },
-    {
-      "epoch": 0.8450790374065791,
-      "grad_norm": 0.0028988949488848448,
-      "learning_rate": 0.00019999964821394372,
-      "loss": 46.0,
-      "step": 11053
-    },
-    {
-      "epoch": 0.8451554943899688,
-      "grad_norm": 0.0010574010666459799,
-      "learning_rate": 0.00019999964815022882,
-      "loss": 46.0,
-      "step": 11054
-    },
-    {
-      "epoch": 0.8452319513733586,
-      "grad_norm": 0.0006346292793750763,
-      "learning_rate": 0.00019999964808650815,
-      "loss": 46.0,
-      "step": 11055
-    },
-    {
-      "epoch": 0.8453084083567483,
-      "grad_norm": 0.0007920832140371203,
-      "learning_rate": 0.00019999964802278172,
-      "loss": 46.0,
-      "step": 11056
-    },
-    {
-      "epoch": 0.845384865340138,
-      "grad_norm": 0.005531759932637215,
-      "learning_rate": 0.00019999964795904953,
-      "loss": 46.0,
-      "step": 11057
-    },
-    {
-      "epoch": 0.8454613223235278,
-      "grad_norm": 0.0012296109925955534,
-      "learning_rate": 0.00019999964789531153,
-      "loss": 46.0,
-      "step": 11058
-    },
-    {
-      "epoch": 0.8455377793069174,
-      "grad_norm": 0.0041253287345170975,
-      "learning_rate": 0.00019999964783156782,
-      "loss": 46.0,
-      "step": 11059
-    },
-    {
-      "epoch": 0.8456142362903072,
-      "grad_norm": 0.0007868785760365427,
-      "learning_rate": 0.00019999964776781833,
-      "loss": 46.0,
-      "step": 11060
-    },
-    {
-      "epoch": 0.8456906932736968,
-      "grad_norm": 0.003124518319964409,
-      "learning_rate": 0.00019999964770406304,
-      "loss": 46.0,
-      "step": 11061
-    },
-    {
-      "epoch": 0.8457671502570866,
-      "grad_norm": 0.0015684838872402906,
-      "learning_rate": 0.00019999964764030198,
-      "loss": 46.0,
-      "step": 11062
-    },
-    {
-      "epoch": 0.8458436072404764,
-      "grad_norm": 0.0009703235700726509,
-      "learning_rate": 0.00019999964757653514,
-      "loss": 46.0,
-      "step": 11063
-    },
-    {
-      "epoch": 0.845920064223866,
-      "grad_norm": 0.001825161511078477,
-      "learning_rate": 0.00019999964751276256,
-      "loss": 46.0,
-      "step": 11064
-    },
-    {
-      "epoch": 0.8459965212072558,
-      "grad_norm": 0.0023887422867119312,
-      "learning_rate": 0.0001999996474489842,
-      "loss": 46.0,
-      "step": 11065
-    },
-    {
-      "epoch": 0.8460729781906455,
-      "grad_norm": 0.0004532198654487729,
-      "learning_rate": 0.0001999996473852001,
-      "loss": 46.0,
-      "step": 11066
-    },
-    {
-      "epoch": 0.8461494351740352,
-      "grad_norm": 0.0016221263213083148,
-      "learning_rate": 0.0001999996473214102,
-      "loss": 46.0,
-      "step": 11067
-    },
-    {
-      "epoch": 0.8462258921574249,
-      "grad_norm": 0.000907170819118619,
-      "learning_rate": 0.00019999964725761452,
-      "loss": 46.0,
-      "step": 11068
-    },
-    {
-      "epoch": 0.8463023491408147,
-      "grad_norm": 0.0017541300039738417,
-      "learning_rate": 0.0001999996471938131,
-      "loss": 46.0,
-      "step": 11069
-    },
-    {
-      "epoch": 0.8463788061242044,
-      "grad_norm": 0.0010579756926745176,
-      "learning_rate": 0.0001999996471300059,
-      "loss": 46.0,
-      "step": 11070
-    },
-    {
-      "epoch": 0.8464552631075941,
-      "grad_norm": 0.0019380139419808984,
-      "learning_rate": 0.0001999996470661929,
-      "loss": 46.0,
-      "step": 11071
-    },
-    {
-      "epoch": 0.8465317200909838,
-      "grad_norm": 0.0015164641663432121,
-      "learning_rate": 0.00019999964700237417,
-      "loss": 46.0,
-      "step": 11072
-    },
-    {
-      "epoch": 0.8466081770743735,
-      "grad_norm": 0.0033727106638252735,
-      "learning_rate": 0.00019999964693854965,
-      "loss": 46.0,
-      "step": 11073
-    },
-    {
-      "epoch": 0.8466846340577633,
-      "grad_norm": 0.0005675025749951601,
-      "learning_rate": 0.0001999996468747194,
-      "loss": 46.0,
-      "step": 11074
-    },
-    {
-      "epoch": 0.8467610910411529,
-      "grad_norm": 0.0031972487922757864,
-      "learning_rate": 0.00019999964681088336,
-      "loss": 46.0,
-      "step": 11075
-    },
-    {
-      "epoch": 0.8468375480245427,
-      "grad_norm": 0.001838152064010501,
-      "learning_rate": 0.00019999964674704152,
-      "loss": 46.0,
-      "step": 11076
-    },
-    {
-      "epoch": 0.8469140050079325,
-      "grad_norm": 0.0014730638358741999,
-      "learning_rate": 0.00019999964668319394,
-      "loss": 46.0,
-      "step": 11077
-    },
-    {
-      "epoch": 0.8469904619913221,
-      "grad_norm": 0.0008131054346449673,
-      "learning_rate": 0.00019999964661934056,
-      "loss": 46.0,
-      "step": 11078
-    },
-    {
-      "epoch": 0.8470669189747119,
-      "grad_norm": 0.005546481814235449,
-      "learning_rate": 0.00019999964655548146,
-      "loss": 46.0,
-      "step": 11079
-    },
-    {
-      "epoch": 0.8471433759581015,
-      "grad_norm": 0.00039659501635469496,
-      "learning_rate": 0.00019999964649161655,
-      "loss": 46.0,
-      "step": 11080
-    },
-    {
-      "epoch": 0.8472198329414913,
-      "grad_norm": 0.0009099820163100958,
-      "learning_rate": 0.00019999964642774588,
-      "loss": 46.0,
-      "step": 11081
-    },
-    {
-      "epoch": 0.847296289924881,
-      "grad_norm": 0.0009537608711980283,
-      "learning_rate": 0.00019999964636386946,
-      "loss": 46.0,
-      "step": 11082
-    },
-    {
-      "epoch": 0.8473727469082707,
-      "grad_norm": 0.0013637441443279386,
-      "learning_rate": 0.0001999996462999872,
-      "loss": 46.0,
-      "step": 11083
-    },
-    {
-      "epoch": 0.8474492038916605,
-      "grad_norm": 0.0018717522034421563,
-      "learning_rate": 0.00019999964623609927,
-      "loss": 46.0,
-      "step": 11084
-    },
-    {
-      "epoch": 0.8475256608750502,
-      "grad_norm": 0.0009365485166199505,
-      "learning_rate": 0.00019999964617220553,
-      "loss": 46.0,
-      "step": 11085
-    },
-    {
-      "epoch": 0.8476021178584399,
-      "grad_norm": 0.00041334074921905994,
-      "learning_rate": 0.000199999646108306,
-      "loss": 46.0,
-      "step": 11086
-    },
-    {
-      "epoch": 0.8476785748418296,
-      "grad_norm": 0.001404003007337451,
-      "learning_rate": 0.00019999964604440072,
-      "loss": 46.0,
-      "step": 11087
-    },
-    {
-      "epoch": 0.8477550318252194,
-      "grad_norm": 0.006241766735911369,
-      "learning_rate": 0.0001999996459804897,
-      "loss": 46.0,
-      "step": 11088
-    },
-    {
-      "epoch": 0.847831488808609,
-      "grad_norm": 0.0035682839807122946,
-      "learning_rate": 0.00019999964591657285,
-      "loss": 46.0,
-      "step": 11089
-    },
-    {
-      "epoch": 0.8479079457919988,
-      "grad_norm": 0.0008588465279899538,
-      "learning_rate": 0.00019999964585265024,
-      "loss": 46.0,
-      "step": 11090
-    },
-    {
-      "epoch": 0.8479844027753884,
-      "grad_norm": 0.0008802936063148081,
-      "learning_rate": 0.0001999996457887219,
-      "loss": 46.0,
-      "step": 11091
-    },
-    {
-      "epoch": 0.8480608597587782,
-      "grad_norm": 0.000702622055541724,
-      "learning_rate": 0.00019999964572478776,
-      "loss": 46.0,
-      "step": 11092
-    },
-    {
-      "epoch": 0.848137316742168,
-      "grad_norm": 0.0009303967235609889,
-      "learning_rate": 0.00019999964566084786,
-      "loss": 46.0,
-      "step": 11093
-    },
-    {
-      "epoch": 0.8482137737255576,
-      "grad_norm": 0.0006240560906007886,
-      "learning_rate": 0.00019999964559690218,
-      "loss": 46.0,
-      "step": 11094
-    },
-    {
-      "epoch": 0.8482902307089474,
-      "grad_norm": 0.0005857710493728518,
-      "learning_rate": 0.00019999964553295074,
-      "loss": 46.0,
-      "step": 11095
-    },
-    {
-      "epoch": 0.8483666876923371,
-      "grad_norm": 0.00160969328135252,
-      "learning_rate": 0.00019999964546899354,
-      "loss": 46.0,
-      "step": 11096
-    },
-    {
-      "epoch": 0.8484431446757268,
-      "grad_norm": 0.0006553277489729226,
-      "learning_rate": 0.00019999964540503057,
-      "loss": 46.0,
-      "step": 11097
-    },
-    {
-      "epoch": 0.8485196016591166,
-      "grad_norm": 0.0014790561981499195,
-      "learning_rate": 0.0001999996453410618,
-      "loss": 46.0,
-      "step": 11098
-    },
-    {
-      "epoch": 0.8485960586425063,
-      "grad_norm": 0.003423925954848528,
-      "learning_rate": 0.0001999996452770873,
-      "loss": 46.0,
-      "step": 11099
-    },
-    {
-      "epoch": 0.848672515625896,
-      "grad_norm": 0.0011463325936347246,
-      "learning_rate": 0.000199999645213107,
-      "loss": 46.0,
-      "step": 11100
-    },
-    {
-      "epoch": 0.8487489726092857,
-      "grad_norm": 0.00351628172211349,
-      "learning_rate": 0.00019999964514912094,
-      "loss": 46.0,
-      "step": 11101
-    },
-    {
-      "epoch": 0.8488254295926754,
-      "grad_norm": 0.0008316357852891088,
-      "learning_rate": 0.00019999964508512913,
-      "loss": 46.0,
-      "step": 11102
-    },
-    {
-      "epoch": 0.8489018865760651,
-      "grad_norm": 0.0015575755387544632,
-      "learning_rate": 0.00019999964502113153,
-      "loss": 46.0,
-      "step": 11103
-    },
-    {
-      "epoch": 0.8489783435594549,
-      "grad_norm": 0.005252194125205278,
-      "learning_rate": 0.00019999964495712817,
-      "loss": 46.0,
-      "step": 11104
-    },
-    {
-      "epoch": 0.8490548005428445,
-      "grad_norm": 0.0025980074424296618,
-      "learning_rate": 0.00019999964489311904,
-      "loss": 46.0,
-      "step": 11105
-    },
-    {
-      "epoch": 0.8491312575262343,
-      "grad_norm": 0.0012850436614826322,
-      "learning_rate": 0.00019999964482910414,
-      "loss": 46.0,
-      "step": 11106
-    },
-    {
-      "epoch": 0.8492077145096241,
-      "grad_norm": 0.004240087699145079,
-      "learning_rate": 0.00019999964476508347,
-      "loss": 46.0,
-      "step": 11107
-    },
-    {
-      "epoch": 0.8492841714930137,
-      "grad_norm": 0.00125785858836025,
-      "learning_rate": 0.00019999964470105702,
-      "loss": 46.0,
-      "step": 11108
-    },
-    {
-      "epoch": 0.8493606284764035,
-      "grad_norm": 0.001575937494635582,
-      "learning_rate": 0.0001999996446370248,
-      "loss": 46.0,
-      "step": 11109
-    },
-    {
-      "epoch": 0.8494370854597931,
-      "grad_norm": 0.0029926816932857037,
-      "learning_rate": 0.00019999964457298683,
-      "loss": 46.0,
-      "step": 11110
-    },
-    {
-      "epoch": 0.8495135424431829,
-      "grad_norm": 0.0015967695508152246,
-      "learning_rate": 0.0001999996445089431,
-      "loss": 46.0,
-      "step": 11111
-    },
-    {
-      "epoch": 0.8495899994265727,
-      "grad_norm": 0.002085303422063589,
-      "learning_rate": 0.00019999964444489357,
-      "loss": 46.0,
-      "step": 11112
-    },
-    {
-      "epoch": 0.8496664564099623,
-      "grad_norm": 0.007869662716984749,
-      "learning_rate": 0.00019999964438083826,
-      "loss": 46.0,
-      "step": 11113
-    },
-    {
-      "epoch": 0.8497429133933521,
-      "grad_norm": 0.01517061423510313,
-      "learning_rate": 0.00019999964431677723,
-      "loss": 46.0,
-      "step": 11114
-    },
-    {
-      "epoch": 0.8498193703767418,
-      "grad_norm": 0.0007287377375178039,
-      "learning_rate": 0.0001999996442527104,
-      "loss": 46.0,
-      "step": 11115
-    },
-    {
-      "epoch": 0.8498958273601315,
-      "grad_norm": 0.0007352664833888412,
-      "learning_rate": 0.00019999964418863778,
-      "loss": 46.0,
-      "step": 11116
-    },
-    {
-      "epoch": 0.8499722843435212,
-      "grad_norm": 0.0016861286712810397,
-      "learning_rate": 0.00019999964412455943,
-      "loss": 46.0,
-      "step": 11117
-    },
-    {
-      "epoch": 0.850048741326911,
-      "grad_norm": 0.0006984604406170547,
-      "learning_rate": 0.0001999996440604753,
-      "loss": 46.0,
-      "step": 11118
-    },
-    {
-      "epoch": 0.8501251983103006,
-      "grad_norm": 0.0010811056708917022,
-      "learning_rate": 0.00019999964399638537,
-      "loss": 46.0,
-      "step": 11119
-    },
-    {
-      "epoch": 0.8502016552936904,
-      "grad_norm": 0.000977845280431211,
-      "learning_rate": 0.00019999964393228973,
-      "loss": 46.0,
-      "step": 11120
-    },
-    {
-      "epoch": 0.8502781122770801,
-      "grad_norm": 0.0015058178687468171,
-      "learning_rate": 0.00019999964386818825,
-      "loss": 46.0,
-      "step": 11121
-    },
-    {
-      "epoch": 0.8503545692604698,
-      "grad_norm": 0.0035912117455154657,
-      "learning_rate": 0.00019999964380408106,
-      "loss": 46.0,
-      "step": 11122
-    },
-    {
-      "epoch": 0.8504310262438596,
-      "grad_norm": 0.0021168177481740713,
-      "learning_rate": 0.00019999964373996806,
-      "loss": 46.0,
-      "step": 11123
-    },
-    {
-      "epoch": 0.8505074832272492,
-      "grad_norm": 0.0013354021357372403,
-      "learning_rate": 0.00019999964367584932,
-      "loss": 46.0,
-      "step": 11124
-    },
-    {
-      "epoch": 0.850583940210639,
-      "grad_norm": 0.0007549573783762753,
-      "learning_rate": 0.00019999964361172478,
-      "loss": 46.0,
-      "step": 11125
-    },
-    {
-      "epoch": 0.8506603971940288,
-      "grad_norm": 0.0015961505705490708,
-      "learning_rate": 0.0001999996435475945,
-      "loss": 46.0,
-      "step": 11126
-    },
-    {
-      "epoch": 0.8507368541774184,
-      "grad_norm": 0.0022256355732679367,
-      "learning_rate": 0.00019999964348345843,
-      "loss": 46.0,
-      "step": 11127
-    },
-    {
-      "epoch": 0.8508133111608082,
-      "grad_norm": 0.0009828897891566157,
-      "learning_rate": 0.0001999996434193166,
-      "loss": 46.0,
-      "step": 11128
-    },
-    {
-      "epoch": 0.8508897681441979,
-      "grad_norm": 0.004272941034287214,
-      "learning_rate": 0.00019999964335516902,
-      "loss": 46.0,
-      "step": 11129
-    },
-    {
-      "epoch": 0.8509662251275876,
-      "grad_norm": 0.0005844312254339457,
-      "learning_rate": 0.00019999964329101564,
-      "loss": 46.0,
-      "step": 11130
-    },
-    {
-      "epoch": 0.8510426821109773,
-      "grad_norm": 0.0011244529159739614,
-      "learning_rate": 0.00019999964322685648,
-      "loss": 46.0,
-      "step": 11131
-    },
-    {
-      "epoch": 0.851119139094367,
-      "grad_norm": 0.0015478788409382105,
-      "learning_rate": 0.0001999996431626916,
-      "loss": 46.0,
-      "step": 11132
-    },
-    {
-      "epoch": 0.8511955960777567,
-      "grad_norm": 0.005694926716387272,
-      "learning_rate": 0.00019999964309852094,
-      "loss": 46.0,
-      "step": 11133
-    },
-    {
-      "epoch": 0.8512720530611465,
-      "grad_norm": 0.001463546184822917,
-      "learning_rate": 0.00019999964303434446,
-      "loss": 46.0,
-      "step": 11134
-    },
-    {
-      "epoch": 0.8513485100445362,
-      "grad_norm": 0.003945269156247377,
-      "learning_rate": 0.00019999964297016224,
-      "loss": 46.0,
-      "step": 11135
-    },
-    {
-      "epoch": 0.8514249670279259,
-      "grad_norm": 0.0008292829734273255,
-      "learning_rate": 0.00019999964290597427,
-      "loss": 46.0,
-      "step": 11136
-    },
-    {
-      "epoch": 0.8515014240113157,
-      "grad_norm": 0.0009785188594833016,
-      "learning_rate": 0.00019999964284178053,
-      "loss": 46.0,
-      "step": 11137
-    },
-    {
-      "epoch": 0.8515778809947053,
-      "grad_norm": 0.0005827966961078346,
-      "learning_rate": 0.000199999642777581,
-      "loss": 46.0,
-      "step": 11138
-    },
-    {
-      "epoch": 0.8516543379780951,
-      "grad_norm": 0.0020266349893063307,
-      "learning_rate": 0.00019999964271337568,
-      "loss": 46.0,
-      "step": 11139
-    },
-    {
-      "epoch": 0.8517307949614848,
-      "grad_norm": 0.0021023531444370747,
-      "learning_rate": 0.00019999964264916462,
-      "loss": 46.0,
-      "step": 11140
-    },
-    {
-      "epoch": 0.8518072519448745,
-      "grad_norm": 0.0008576546097174287,
-      "learning_rate": 0.00019999964258494778,
-      "loss": 46.0,
-      "step": 11141
-    },
-    {
-      "epoch": 0.8518837089282643,
-      "grad_norm": 0.0013772320235148072,
-      "learning_rate": 0.00019999964252072518,
-      "loss": 46.0,
-      "step": 11142
-    },
-    {
-      "epoch": 0.8519601659116539,
-      "grad_norm": 0.0018279260257259011,
-      "learning_rate": 0.0001999996424564968,
-      "loss": 46.0,
-      "step": 11143
-    },
-    {
-      "epoch": 0.8520366228950437,
-      "grad_norm": 0.0017454697517678142,
-      "learning_rate": 0.00019999964239226267,
-      "loss": 46.0,
-      "step": 11144
-    },
-    {
-      "epoch": 0.8521130798784334,
-      "grad_norm": 0.001503758248873055,
-      "learning_rate": 0.00019999964232802277,
-      "loss": 46.0,
-      "step": 11145
-    },
-    {
-      "epoch": 0.8521895368618231,
-      "grad_norm": 0.001236366922967136,
-      "learning_rate": 0.0001999996422637771,
-      "loss": 46.0,
-      "step": 11146
-    },
-    {
-      "epoch": 0.8522659938452128,
-      "grad_norm": 0.0011803583474829793,
-      "learning_rate": 0.00019999964219952565,
-      "loss": 46.0,
-      "step": 11147
-    },
-    {
-      "epoch": 0.8523424508286026,
-      "grad_norm": 0.0004411491972859949,
-      "learning_rate": 0.00019999964213526843,
-      "loss": 46.0,
-      "step": 11148
-    },
-    {
-      "epoch": 0.8524189078119923,
-      "grad_norm": 0.0015384680591523647,
-      "learning_rate": 0.00019999964207100543,
-      "loss": 46.0,
-      "step": 11149
-    },
-    {
-      "epoch": 0.852495364795382,
-      "grad_norm": 0.0007610170287080109,
-      "learning_rate": 0.00019999964200673667,
-      "loss": 46.0,
-      "step": 11150
-    },
-    {
-      "epoch": 0.8525718217787717,
-      "grad_norm": 0.0011325698578730226,
-      "learning_rate": 0.00019999964194246215,
-      "loss": 46.0,
-      "step": 11151
-    },
-    {
-      "epoch": 0.8526482787621614,
-      "grad_norm": 0.0018241963116452098,
-      "learning_rate": 0.00019999964187818187,
-      "loss": 46.0,
-      "step": 11152
-    },
-    {
-      "epoch": 0.8527247357455512,
-      "grad_norm": 0.0009631628054194152,
-      "learning_rate": 0.00019999964181389578,
-      "loss": 46.0,
-      "step": 11153
-    },
-    {
-      "epoch": 0.8528011927289408,
-      "grad_norm": 0.005874826572835445,
-      "learning_rate": 0.00019999964174960395,
-      "loss": 46.0,
-      "step": 11154
-    },
-    {
-      "epoch": 0.8528776497123306,
-      "grad_norm": 0.0007128996076062322,
-      "learning_rate": 0.00019999964168530634,
-      "loss": 46.0,
-      "step": 11155
-    },
-    {
-      "epoch": 0.8529541066957204,
-      "grad_norm": 0.00024089418002404273,
-      "learning_rate": 0.00019999964162100296,
-      "loss": 46.0,
-      "step": 11156
-    },
-    {
-      "epoch": 0.85303056367911,
-      "grad_norm": 0.0044003985822200775,
-      "learning_rate": 0.0001999996415566938,
-      "loss": 46.0,
-      "step": 11157
-    },
-    {
-      "epoch": 0.8531070206624998,
-      "grad_norm": 0.000694796210154891,
-      "learning_rate": 0.0001999996414923789,
-      "loss": 46.0,
-      "step": 11158
-    },
-    {
-      "epoch": 0.8531834776458895,
-      "grad_norm": 0.0005199249135330319,
-      "learning_rate": 0.00019999964142805823,
-      "loss": 46.0,
-      "step": 11159
-    },
-    {
-      "epoch": 0.8532599346292792,
-      "grad_norm": 0.0031427121721208096,
-      "learning_rate": 0.00019999964136373178,
-      "loss": 46.0,
-      "step": 11160
-    },
-    {
-      "epoch": 0.8533363916126689,
-      "grad_norm": 0.00044588546734303236,
-      "learning_rate": 0.00019999964129939954,
-      "loss": 46.0,
-      "step": 11161
-    },
-    {
-      "epoch": 0.8534128485960586,
-      "grad_norm": 0.002548330230638385,
-      "learning_rate": 0.00019999964123506157,
-      "loss": 46.0,
-      "step": 11162
-    },
-    {
-      "epoch": 0.8534893055794484,
-      "grad_norm": 0.0016298653790727258,
-      "learning_rate": 0.0001999996411707178,
-      "loss": 46.0,
-      "step": 11163
-    },
-    {
-      "epoch": 0.8535657625628381,
-      "grad_norm": 0.0025588199496269226,
-      "learning_rate": 0.00019999964110636826,
-      "loss": 46.0,
-      "step": 11164
-    },
-    {
-      "epoch": 0.8536422195462278,
-      "grad_norm": 0.0007251531351357698,
-      "learning_rate": 0.00019999964104201298,
-      "loss": 46.0,
-      "step": 11165
-    },
-    {
-      "epoch": 0.8537186765296175,
-      "grad_norm": 0.0007614085334353149,
-      "learning_rate": 0.0001999996409776519,
-      "loss": 46.0,
-      "step": 11166
-    },
-    {
-      "epoch": 0.8537951335130073,
-      "grad_norm": 0.0009560105972923338,
-      "learning_rate": 0.00019999964091328506,
-      "loss": 46.0,
-      "step": 11167
-    },
-    {
-      "epoch": 0.8538715904963969,
-      "grad_norm": 0.00316926883533597,
-      "learning_rate": 0.00019999964084891245,
-      "loss": 46.0,
-      "step": 11168
-    },
-    {
-      "epoch": 0.8539480474797867,
-      "grad_norm": 0.0007053738227114081,
-      "learning_rate": 0.00019999964078453407,
-      "loss": 46.0,
-      "step": 11169
-    },
-    {
-      "epoch": 0.8540245044631765,
-      "grad_norm": 0.001694960636086762,
-      "learning_rate": 0.00019999964072014992,
-      "loss": 46.0,
-      "step": 11170
-    },
-    {
-      "epoch": 0.8541009614465661,
-      "grad_norm": 0.0009114079875871539,
-      "learning_rate": 0.00019999964065576,
-      "loss": 46.0,
-      "step": 11171
-    },
-    {
-      "epoch": 0.8541774184299559,
-      "grad_norm": 0.0010451360139995813,
-      "learning_rate": 0.00019999964059136432,
-      "loss": 46.0,
-      "step": 11172
-    },
-    {
-      "epoch": 0.8542538754133455,
-      "grad_norm": 0.0018855428788810968,
-      "learning_rate": 0.00019999964052696287,
-      "loss": 46.0,
-      "step": 11173
-    },
-    {
-      "epoch": 0.8543303323967353,
-      "grad_norm": 0.0038219306152313948,
-      "learning_rate": 0.00019999964046255565,
-      "loss": 46.0,
-      "step": 11174
-    },
-    {
-      "epoch": 0.854406789380125,
-      "grad_norm": 0.0015638156328350306,
-      "learning_rate": 0.00019999964039814266,
-      "loss": 46.0,
-      "step": 11175
-    },
-    {
-      "epoch": 0.8544832463635147,
-      "grad_norm": 0.0005671442486345768,
-      "learning_rate": 0.0001999996403337239,
-      "loss": 46.0,
-      "step": 11176
-    },
-    {
-      "epoch": 0.8545597033469045,
-      "grad_norm": 0.0006012267549522221,
-      "learning_rate": 0.00019999964026929935,
-      "loss": 46.0,
-      "step": 11177
-    },
-    {
-      "epoch": 0.8546361603302942,
-      "grad_norm": 0.002317205537110567,
-      "learning_rate": 0.00019999964020486907,
-      "loss": 46.0,
-      "step": 11178
-    },
-    {
-      "epoch": 0.8547126173136839,
-      "grad_norm": 0.0008620276930741966,
-      "learning_rate": 0.00019999964014043298,
-      "loss": 46.0,
-      "step": 11179
-    },
-    {
-      "epoch": 0.8547890742970736,
-      "grad_norm": 0.011933354660868645,
-      "learning_rate": 0.00019999964007599115,
-      "loss": 46.0,
-      "step": 11180
-    },
-    {
-      "epoch": 0.8548655312804633,
-      "grad_norm": 0.0005438775406219065,
-      "learning_rate": 0.00019999964001154354,
-      "loss": 46.0,
-      "step": 11181
-    },
-    {
-      "epoch": 0.854941988263853,
-      "grad_norm": 0.00038960689562372863,
-      "learning_rate": 0.00019999963994709016,
-      "loss": 46.0,
-      "step": 11182
-    },
-    {
-      "epoch": 0.8550184452472428,
-      "grad_norm": 0.00303665385581553,
-      "learning_rate": 0.000199999639882631,
-      "loss": 46.0,
-      "step": 11183
-    },
-    {
-      "epoch": 0.8550949022306324,
-      "grad_norm": 0.0013340001460164785,
-      "learning_rate": 0.0001999996398181661,
-      "loss": 46.0,
-      "step": 11184
-    },
-    {
-      "epoch": 0.8551713592140222,
-      "grad_norm": 0.0007283794111572206,
-      "learning_rate": 0.0001999996397536954,
-      "loss": 46.0,
-      "step": 11185
-    },
-    {
-      "epoch": 0.855247816197412,
-      "grad_norm": 0.0010545131517574191,
-      "learning_rate": 0.00019999963968921893,
-      "loss": 46.0,
-      "step": 11186
-    },
-    {
-      "epoch": 0.8553242731808016,
-      "grad_norm": 0.002913212403655052,
-      "learning_rate": 0.00019999963962473671,
-      "loss": 46.0,
-      "step": 11187
-    },
-    {
-      "epoch": 0.8554007301641914,
-      "grad_norm": 0.0012835368979722261,
-      "learning_rate": 0.00019999963956024872,
-      "loss": 46.0,
-      "step": 11188
-    },
-    {
-      "epoch": 0.8554771871475811,
-      "grad_norm": 0.0009773829951882362,
-      "learning_rate": 0.00019999963949575496,
-      "loss": 46.0,
-      "step": 11189
-    },
-    {
-      "epoch": 0.8555536441309708,
-      "grad_norm": 0.002376238815486431,
-      "learning_rate": 0.00019999963943125542,
-      "loss": 46.0,
-      "step": 11190
-    },
-    {
-      "epoch": 0.8556301011143606,
-      "grad_norm": 0.008335355669260025,
-      "learning_rate": 0.00019999963936675013,
-      "loss": 46.0,
-      "step": 11191
-    },
-    {
-      "epoch": 0.8557065580977502,
-      "grad_norm": 0.00114829174708575,
-      "learning_rate": 0.00019999963930223904,
-      "loss": 46.0,
-      "step": 11192
-    },
-    {
-      "epoch": 0.85578301508114,
-      "grad_norm": 0.0018514475086703897,
-      "learning_rate": 0.00019999963923772218,
-      "loss": 46.0,
-      "step": 11193
-    },
-    {
-      "epoch": 0.8558594720645297,
-      "grad_norm": 0.0006004343740642071,
-      "learning_rate": 0.00019999963917319958,
-      "loss": 46.0,
-      "step": 11194
-    },
-    {
-      "epoch": 0.8559359290479194,
-      "grad_norm": 0.0010789603693410754,
-      "learning_rate": 0.0001999996391086712,
-      "loss": 46.0,
-      "step": 11195
-    },
-    {
-      "epoch": 0.8560123860313091,
-      "grad_norm": 0.00040760572301223874,
-      "learning_rate": 0.00019999963904413705,
-      "loss": 46.0,
-      "step": 11196
-    },
-    {
-      "epoch": 0.8560888430146989,
-      "grad_norm": 0.0021404919680207968,
-      "learning_rate": 0.00019999963897959712,
-      "loss": 46.0,
-      "step": 11197
-    },
-    {
-      "epoch": 0.8561652999980885,
-      "grad_norm": 0.0016223929123952985,
-      "learning_rate": 0.00019999963891505145,
-      "loss": 46.0,
-      "step": 11198
-    },
-    {
-      "epoch": 0.8562417569814783,
-      "grad_norm": 0.0014355364255607128,
-      "learning_rate": 0.00019999963885049997,
-      "loss": 46.0,
-      "step": 11199
-    },
-    {
-      "epoch": 0.8563182139648681,
-      "grad_norm": 0.0005289613036438823,
-      "learning_rate": 0.00019999963878594276,
-      "loss": 46.0,
-      "step": 11200
-    },
-    {
-      "epoch": 0.8563946709482577,
-      "grad_norm": 0.00042334923637099564,
-      "learning_rate": 0.00019999963872137974,
-      "loss": 46.0,
-      "step": 11201
-    },
-    {
-      "epoch": 0.8564711279316475,
-      "grad_norm": 0.00808667205274105,
-      "learning_rate": 0.00019999963865681097,
-      "loss": 46.0,
-      "step": 11202
-    },
-    {
-      "epoch": 0.8565475849150371,
-      "grad_norm": 0.0019520564237609506,
-      "learning_rate": 0.00019999963859223643,
-      "loss": 46.0,
-      "step": 11203
-    },
-    {
-      "epoch": 0.8566240418984269,
-      "grad_norm": 0.005279670935124159,
-      "learning_rate": 0.00019999963852765612,
-      "loss": 46.0,
-      "step": 11204
-    },
-    {
-      "epoch": 0.8567004988818167,
-      "grad_norm": 0.0017014906043186784,
-      "learning_rate": 0.00019999963846307006,
-      "loss": 46.0,
-      "step": 11205
-    },
-    {
-      "epoch": 0.8567769558652063,
-      "grad_norm": 0.0012440314749255776,
-      "learning_rate": 0.00019999963839847823,
-      "loss": 46.0,
-      "step": 11206
-    },
-    {
-      "epoch": 0.8568534128485961,
-      "grad_norm": 0.0011593655217438936,
-      "learning_rate": 0.0001999996383338806,
-      "loss": 46.0,
-      "step": 11207
-    },
-    {
-      "epoch": 0.8569298698319858,
-      "grad_norm": 0.0005625446792691946,
-      "learning_rate": 0.00019999963826927721,
-      "loss": 46.0,
-      "step": 11208
-    },
-    {
-      "epoch": 0.8570063268153755,
-      "grad_norm": 0.001205519656650722,
-      "learning_rate": 0.00019999963820466806,
-      "loss": 46.0,
-      "step": 11209
-    },
-    {
-      "epoch": 0.8570827837987652,
-      "grad_norm": 0.002979275770485401,
-      "learning_rate": 0.0001999996381400531,
-      "loss": 46.0,
-      "step": 11210
-    },
-    {
-      "epoch": 0.8571592407821549,
-      "grad_norm": 0.0053017837926745415,
-      "learning_rate": 0.00019999963807543244,
-      "loss": 46.0,
-      "step": 11211
-    },
-    {
-      "epoch": 0.8572356977655446,
-      "grad_norm": 0.001378338783979416,
-      "learning_rate": 0.00019999963801080596,
-      "loss": 46.0,
-      "step": 11212
-    },
-    {
-      "epoch": 0.8573121547489344,
-      "grad_norm": 0.0009639773052185774,
-      "learning_rate": 0.00019999963794617375,
-      "loss": 46.0,
-      "step": 11213
-    },
-    {
-      "epoch": 0.8573886117323241,
-      "grad_norm": 0.000531878205947578,
-      "learning_rate": 0.00019999963788153573,
-      "loss": 46.0,
-      "step": 11214
-    },
-    {
-      "epoch": 0.8574650687157138,
-      "grad_norm": 0.002438716823235154,
-      "learning_rate": 0.00019999963781689196,
-      "loss": 46.0,
-      "step": 11215
-    },
-    {
-      "epoch": 0.8575415256991036,
-      "grad_norm": 0.0016653784550726414,
-      "learning_rate": 0.00019999963775224245,
-      "loss": 46.0,
-      "step": 11216
-    },
-    {
-      "epoch": 0.8576179826824932,
-      "grad_norm": 0.0018046468030661345,
-      "learning_rate": 0.0001999996376875871,
-      "loss": 46.0,
-      "step": 11217
-    },
-    {
-      "epoch": 0.857694439665883,
-      "grad_norm": 0.001049616257660091,
-      "learning_rate": 0.00019999963762292605,
-      "loss": 46.0,
-      "step": 11218
-    },
-    {
-      "epoch": 0.8577708966492728,
-      "grad_norm": 0.0015249074203893542,
-      "learning_rate": 0.0001999996375582592,
-      "loss": 46.0,
-      "step": 11219
-    },
-    {
-      "epoch": 0.8578473536326624,
-      "grad_norm": 0.0028940935153514147,
-      "learning_rate": 0.00019999963749358656,
-      "loss": 46.0,
-      "step": 11220
-    },
-    {
-      "epoch": 0.8579238106160522,
-      "grad_norm": 0.001310089137405157,
-      "learning_rate": 0.00019999963742890818,
-      "loss": 46.0,
-      "step": 11221
-    },
-    {
-      "epoch": 0.8580002675994418,
-      "grad_norm": 0.0037354696542024612,
-      "learning_rate": 0.00019999963736422403,
-      "loss": 46.0,
-      "step": 11222
-    },
-    {
-      "epoch": 0.8580767245828316,
-      "grad_norm": 0.009329929947853088,
-      "learning_rate": 0.00019999963729953408,
-      "loss": 46.0,
-      "step": 11223
-    },
-    {
-      "epoch": 0.8581531815662213,
-      "grad_norm": 0.0006990203983150423,
-      "learning_rate": 0.0001999996372348384,
-      "loss": 46.0,
-      "step": 11224
-    },
-    {
-      "epoch": 0.858229638549611,
-      "grad_norm": 0.0011387696722522378,
-      "learning_rate": 0.00019999963717013693,
-      "loss": 46.0,
-      "step": 11225
-    },
-    {
-      "epoch": 0.8583060955330007,
-      "grad_norm": 0.0014611412771046162,
-      "learning_rate": 0.0001999996371054297,
-      "loss": 46.0,
-      "step": 11226
-    },
-    {
-      "epoch": 0.8583825525163905,
-      "grad_norm": 0.0030784413684159517,
-      "learning_rate": 0.0001999996370407167,
-      "loss": 46.0,
-      "step": 11227
-    },
-    {
-      "epoch": 0.8584590094997802,
-      "grad_norm": 0.0019194413907825947,
-      "learning_rate": 0.00019999963697599793,
-      "loss": 46.0,
-      "step": 11228
-    },
-    {
-      "epoch": 0.8585354664831699,
-      "grad_norm": 0.0009426483884453773,
-      "learning_rate": 0.00019999963691127336,
-      "loss": 46.0,
-      "step": 11229
-    },
-    {
-      "epoch": 0.8586119234665597,
-      "grad_norm": 0.0016573058674111962,
-      "learning_rate": 0.00019999963684654305,
-      "loss": 46.0,
-      "step": 11230
-    },
-    {
-      "epoch": 0.8586883804499493,
-      "grad_norm": 0.000970787659753114,
-      "learning_rate": 0.00019999963678180697,
-      "loss": 46.0,
-      "step": 11231
-    },
-    {
-      "epoch": 0.8587648374333391,
-      "grad_norm": 0.0022142468951642513,
-      "learning_rate": 0.0001999996367170651,
-      "loss": 46.0,
-      "step": 11232
-    },
-    {
-      "epoch": 0.8588412944167287,
-      "grad_norm": 0.0033295524772256613,
-      "learning_rate": 0.0001999996366523175,
-      "loss": 46.0,
-      "step": 11233
-    },
-    {
-      "epoch": 0.8589177514001185,
-      "grad_norm": 0.002869711024686694,
-      "learning_rate": 0.00019999963658756412,
-      "loss": 46.0,
-      "step": 11234
-    },
-    {
-      "epoch": 0.8589942083835083,
-      "grad_norm": 0.0017400338547304273,
-      "learning_rate": 0.00019999963652280494,
-      "loss": 46.0,
-      "step": 11235
-    },
-    {
-      "epoch": 0.8590706653668979,
-      "grad_norm": 0.004591533448547125,
-      "learning_rate": 0.00019999963645804002,
-      "loss": 46.0,
-      "step": 11236
-    },
-    {
-      "epoch": 0.8591471223502877,
-      "grad_norm": 0.0020419990178197622,
-      "learning_rate": 0.00019999963639326932,
-      "loss": 46.0,
-      "step": 11237
-    },
-    {
-      "epoch": 0.8592235793336774,
-      "grad_norm": 0.0014866740675643086,
-      "learning_rate": 0.00019999963632849288,
-      "loss": 46.0,
-      "step": 11238
-    },
-    {
-      "epoch": 0.8593000363170671,
-      "grad_norm": 0.0024214345030486584,
-      "learning_rate": 0.00019999963626371063,
-      "loss": 46.0,
-      "step": 11239
-    },
-    {
-      "epoch": 0.8593764933004568,
-      "grad_norm": 0.0003167380637023598,
-      "learning_rate": 0.0001999996361989226,
-      "loss": 46.0,
-      "step": 11240
-    },
-    {
-      "epoch": 0.8594529502838466,
-      "grad_norm": 0.0008764404919929802,
-      "learning_rate": 0.00019999963613412885,
-      "loss": 46.0,
-      "step": 11241
-    },
-    {
-      "epoch": 0.8595294072672363,
-      "grad_norm": 0.0015325612621381879,
-      "learning_rate": 0.0001999996360693293,
-      "loss": 46.0,
-      "step": 11242
-    },
-    {
-      "epoch": 0.859605864250626,
-      "grad_norm": 0.006553499028086662,
-      "learning_rate": 0.00019999963600452397,
-      "loss": 46.0,
-      "step": 11243
-    },
-    {
-      "epoch": 0.8596823212340157,
-      "grad_norm": 0.0015160355251282454,
-      "learning_rate": 0.00019999963593971288,
-      "loss": 46.0,
-      "step": 11244
-    },
-    {
-      "epoch": 0.8597587782174054,
-      "grad_norm": 0.0014304619980975986,
-      "learning_rate": 0.00019999963587489603,
-      "loss": 46.0,
-      "step": 11245
-    },
-    {
-      "epoch": 0.8598352352007952,
-      "grad_norm": 0.001085482188500464,
-      "learning_rate": 0.00019999963581007345,
-      "loss": 46.0,
-      "step": 11246
-    },
-    {
-      "epoch": 0.8599116921841848,
-      "grad_norm": 0.0038877027109265327,
-      "learning_rate": 0.00019999963574524502,
-      "loss": 46.0,
-      "step": 11247
-    },
-    {
-      "epoch": 0.8599881491675746,
-      "grad_norm": 0.001028078026138246,
-      "learning_rate": 0.00019999963568041086,
-      "loss": 46.0,
-      "step": 11248
-    },
-    {
-      "epoch": 0.8600646061509644,
-      "grad_norm": 0.0021918523125350475,
-      "learning_rate": 0.00019999963561557094,
-      "loss": 46.0,
-      "step": 11249
-    },
-    {
-      "epoch": 0.860141063134354,
-      "grad_norm": 0.0009138314635492861,
-      "learning_rate": 0.00019999963555072524,
-      "loss": 46.0,
-      "step": 11250
-    },
-    {
-      "epoch": 0.8602175201177438,
-      "grad_norm": 0.0034649111330509186,
-      "learning_rate": 0.00019999963548587374,
-      "loss": 46.0,
-      "step": 11251
-    },
-    {
-      "epoch": 0.8602939771011334,
-      "grad_norm": 0.007519397884607315,
-      "learning_rate": 0.00019999963542101652,
-      "loss": 46.0,
-      "step": 11252
-    },
-    {
-      "epoch": 0.8603704340845232,
-      "grad_norm": 0.0009737313375808299,
-      "learning_rate": 0.0001999996353561535,
-      "loss": 46.0,
-      "step": 11253
-    },
-    {
-      "epoch": 0.860446891067913,
-      "grad_norm": 0.006674016825854778,
-      "learning_rate": 0.00019999963529128474,
-      "loss": 46.0,
-      "step": 11254
-    },
-    {
-      "epoch": 0.8605233480513026,
-      "grad_norm": 0.0019541995134204626,
-      "learning_rate": 0.00019999963522641018,
-      "loss": 46.0,
-      "step": 11255
-    },
-    {
-      "epoch": 0.8605998050346924,
-      "grad_norm": 0.0005965818418189883,
-      "learning_rate": 0.00019999963516152987,
-      "loss": 46.0,
-      "step": 11256
-    },
-    {
-      "epoch": 0.8606762620180821,
-      "grad_norm": 0.0006793156499043107,
-      "learning_rate": 0.00019999963509664378,
-      "loss": 46.0,
-      "step": 11257
-    },
-    {
-      "epoch": 0.8607527190014718,
-      "grad_norm": 0.0008767470135353506,
-      "learning_rate": 0.00019999963503175192,
-      "loss": 46.0,
-      "step": 11258
-    },
-    {
-      "epoch": 0.8608291759848615,
-      "grad_norm": 0.0008187260827980936,
-      "learning_rate": 0.0001999996349668543,
-      "loss": 46.0,
-      "step": 11259
-    },
-    {
-      "epoch": 0.8609056329682513,
-      "grad_norm": 0.005235785618424416,
-      "learning_rate": 0.0001999996349019509,
-      "loss": 46.0,
-      "step": 11260
-    },
-    {
-      "epoch": 0.8609820899516409,
-      "grad_norm": 0.0013601203681901097,
-      "learning_rate": 0.00019999963483704173,
-      "loss": 46.0,
-      "step": 11261
-    },
-    {
-      "epoch": 0.8610585469350307,
-      "grad_norm": 0.0025896672159433365,
-      "learning_rate": 0.0001999996347721268,
-      "loss": 46.0,
-      "step": 11262
-    },
-    {
-      "epoch": 0.8611350039184203,
-      "grad_norm": 0.001386967720463872,
-      "learning_rate": 0.00019999963470720608,
-      "loss": 46.0,
-      "step": 11263
-    },
-    {
-      "epoch": 0.8612114609018101,
-      "grad_norm": 0.0002685583021957427,
-      "learning_rate": 0.0001999996346422796,
-      "loss": 46.0,
-      "step": 11264
-    },
-    {
-      "epoch": 0.8612879178851999,
-      "grad_norm": 0.010421804152429104,
-      "learning_rate": 0.0001999996345773474,
-      "loss": 46.0,
-      "step": 11265
-    },
-    {
-      "epoch": 0.8613643748685895,
-      "grad_norm": 0.005579240154474974,
-      "learning_rate": 0.00019999963451240938,
-      "loss": 46.0,
-      "step": 11266
-    },
-    {
-      "epoch": 0.8614408318519793,
-      "grad_norm": 0.0008963888976722956,
-      "learning_rate": 0.00019999963444746559,
-      "loss": 46.0,
-      "step": 11267
-    },
-    {
-      "epoch": 0.861517288835369,
-      "grad_norm": 0.000862924731336534,
-      "learning_rate": 0.00019999963438251602,
-      "loss": 46.0,
-      "step": 11268
-    },
-    {
-      "epoch": 0.8615937458187587,
-      "grad_norm": 0.0012422333238646388,
-      "learning_rate": 0.0001999996343175607,
-      "loss": 46.0,
-      "step": 11269
-    },
-    {
-      "epoch": 0.8616702028021485,
-      "grad_norm": 0.01048423070460558,
-      "learning_rate": 0.00019999963425259965,
-      "loss": 46.0,
-      "step": 11270
-    },
-    {
-      "epoch": 0.8617466597855382,
-      "grad_norm": 0.0011771812569350004,
-      "learning_rate": 0.00019999963418763277,
-      "loss": 46.0,
-      "step": 11271
-    },
-    {
-      "epoch": 0.8618231167689279,
-      "grad_norm": 0.0005978705012239516,
-      "learning_rate": 0.00019999963412266014,
-      "loss": 46.0,
-      "step": 11272
-    },
-    {
-      "epoch": 0.8618995737523176,
-      "grad_norm": 0.0022190730087459087,
-      "learning_rate": 0.00019999963405768173,
-      "loss": 46.0,
-      "step": 11273
-    },
-    {
-      "epoch": 0.8619760307357073,
-      "grad_norm": 0.0024961759336292744,
-      "learning_rate": 0.00019999963399269758,
-      "loss": 46.0,
-      "step": 11274
-    },
-    {
-      "epoch": 0.862052487719097,
-      "grad_norm": 0.0005841923411935568,
-      "learning_rate": 0.00019999963392770763,
-      "loss": 46.0,
-      "step": 11275
-    },
-    {
-      "epoch": 0.8621289447024868,
-      "grad_norm": 0.0007702383445575833,
-      "learning_rate": 0.00019999963386271193,
-      "loss": 46.0,
-      "step": 11276
-    },
-    {
-      "epoch": 0.8622054016858764,
-      "grad_norm": 0.0006330618634819984,
-      "learning_rate": 0.00019999963379771046,
-      "loss": 46.0,
-      "step": 11277
-    },
-    {
-      "epoch": 0.8622818586692662,
-      "grad_norm": 0.004083727952092886,
-      "learning_rate": 0.00019999963373270322,
-      "loss": 46.0,
-      "step": 11278
-    },
-    {
-      "epoch": 0.862358315652656,
-      "grad_norm": 0.0007187266601249576,
-      "learning_rate": 0.0001999996336676902,
-      "loss": 46.0,
-      "step": 11279
-    },
-    {
-      "epoch": 0.8624347726360456,
-      "grad_norm": 0.009648003615438938,
-      "learning_rate": 0.0001999996336026714,
-      "loss": 46.0,
-      "step": 11280
-    },
-    {
-      "epoch": 0.8625112296194354,
-      "grad_norm": 0.0017131298081949353,
-      "learning_rate": 0.00019999963353764687,
-      "loss": 46.0,
-      "step": 11281
-    },
-    {
-      "epoch": 0.862587686602825,
-      "grad_norm": 0.0006826784228906035,
-      "learning_rate": 0.00019999963347261653,
-      "loss": 46.0,
-      "step": 11282
-    },
-    {
-      "epoch": 0.8626641435862148,
-      "grad_norm": 0.002370436443015933,
-      "learning_rate": 0.00019999963340758042,
-      "loss": 46.0,
-      "step": 11283
-    },
-    {
-      "epoch": 0.8627406005696046,
-      "grad_norm": 0.0073270793072879314,
-      "learning_rate": 0.00019999963334253856,
-      "loss": 46.0,
-      "step": 11284
-    },
-    {
-      "epoch": 0.8628170575529942,
-      "grad_norm": 0.0020656436681747437,
-      "learning_rate": 0.00019999963327749093,
-      "loss": 46.0,
-      "step": 11285
-    },
-    {
-      "epoch": 0.862893514536384,
-      "grad_norm": 0.0022376365959644318,
-      "learning_rate": 0.00019999963321243756,
-      "loss": 46.0,
-      "step": 11286
-    },
-    {
-      "epoch": 0.8629699715197737,
-      "grad_norm": 0.0009171944693662226,
-      "learning_rate": 0.00019999963314737838,
-      "loss": 46.0,
-      "step": 11287
-    },
-    {
-      "epoch": 0.8630464285031634,
-      "grad_norm": 0.0007794692646712065,
-      "learning_rate": 0.00019999963308231343,
-      "loss": 46.0,
-      "step": 11288
-    },
-    {
-      "epoch": 0.8631228854865531,
-      "grad_norm": 0.003385237418115139,
-      "learning_rate": 0.0001999996330172427,
-      "loss": 46.0,
-      "step": 11289
-    },
-    {
-      "epoch": 0.8631993424699429,
-      "grad_norm": 0.0006665576947852969,
-      "learning_rate": 0.00019999963295216623,
-      "loss": 46.0,
-      "step": 11290
-    },
-    {
-      "epoch": 0.8632757994533325,
-      "grad_norm": 0.0030527880880981684,
-      "learning_rate": 0.000199999632887084,
-      "loss": 46.0,
-      "step": 11291
-    },
-    {
-      "epoch": 0.8633522564367223,
-      "grad_norm": 0.005277436226606369,
-      "learning_rate": 0.000199999632821996,
-      "loss": 46.0,
-      "step": 11292
-    },
-    {
-      "epoch": 0.863428713420112,
-      "grad_norm": 0.0018622911302372813,
-      "learning_rate": 0.0001999996327569022,
-      "loss": 46.0,
-      "step": 11293
-    },
-    {
-      "epoch": 0.8635051704035017,
-      "grad_norm": 0.0007191606564447284,
-      "learning_rate": 0.00019999963269180264,
-      "loss": 46.0,
-      "step": 11294
-    },
-    {
-      "epoch": 0.8635816273868915,
-      "grad_norm": 0.0006722731050103903,
-      "learning_rate": 0.0001999996326266973,
-      "loss": 46.0,
-      "step": 11295
-    },
-    {
-      "epoch": 0.8636580843702811,
-      "grad_norm": 0.0009683463722467422,
-      "learning_rate": 0.00019999963256158622,
-      "loss": 46.0,
-      "step": 11296
-    },
-    {
-      "epoch": 0.8637345413536709,
-      "grad_norm": 0.0009882425656542182,
-      "learning_rate": 0.00019999963249646936,
-      "loss": 46.0,
-      "step": 11297
-    },
-    {
-      "epoch": 0.8638109983370607,
-      "grad_norm": 0.0004792765248566866,
-      "learning_rate": 0.0001999996324313467,
-      "loss": 46.0,
-      "step": 11298
-    },
-    {
-      "epoch": 0.8638874553204503,
-      "grad_norm": 0.007468396332114935,
-      "learning_rate": 0.00019999963236621833,
-      "loss": 46.0,
-      "step": 11299
-    },
-    {
-      "epoch": 0.8639639123038401,
-      "grad_norm": 0.0005085182492621243,
-      "learning_rate": 0.00019999963230108415,
-      "loss": 46.0,
-      "step": 11300
-    },
-    {
-      "epoch": 0.8640403692872298,
-      "grad_norm": 0.0005188824143260717,
-      "learning_rate": 0.0001999996322359442,
-      "loss": 46.0,
-      "step": 11301
-    },
-    {
-      "epoch": 0.8641168262706195,
-      "grad_norm": 0.0004394619318190962,
-      "learning_rate": 0.00019999963217079848,
-      "loss": 46.0,
-      "step": 11302
-    },
-    {
-      "epoch": 0.8641932832540092,
-      "grad_norm": 0.004168111830949783,
-      "learning_rate": 0.000199999632105647,
-      "loss": 46.0,
-      "step": 11303
-    },
-    {
-      "epoch": 0.8642697402373989,
-      "grad_norm": 0.0009028415661305189,
-      "learning_rate": 0.00019999963204048976,
-      "loss": 46.0,
-      "step": 11304
-    },
-    {
-      "epoch": 0.8643461972207886,
-      "grad_norm": 0.0008435260970145464,
-      "learning_rate": 0.00019999963197532675,
-      "loss": 46.0,
-      "step": 11305
-    },
-    {
-      "epoch": 0.8644226542041784,
-      "grad_norm": 0.002249644137918949,
-      "learning_rate": 0.00019999963191015793,
-      "loss": 46.0,
-      "step": 11306
-    },
-    {
-      "epoch": 0.8644991111875681,
-      "grad_norm": 0.0014918898232281208,
-      "learning_rate": 0.0001999996318449834,
-      "loss": 46.0,
-      "step": 11307
-    },
-    {
-      "epoch": 0.8645755681709578,
-      "grad_norm": 0.0006185891688801348,
-      "learning_rate": 0.00019999963177980305,
-      "loss": 46.0,
-      "step": 11308
-    },
-    {
-      "epoch": 0.8646520251543476,
-      "grad_norm": 0.004601870197802782,
-      "learning_rate": 0.00019999963171461697,
-      "loss": 46.0,
-      "step": 11309
-    },
-    {
-      "epoch": 0.8647284821377372,
-      "grad_norm": 0.0011978121474385262,
-      "learning_rate": 0.0001999996316494251,
-      "loss": 46.0,
-      "step": 11310
-    },
-    {
-      "epoch": 0.864804939121127,
-      "grad_norm": 0.0011850838782265782,
-      "learning_rate": 0.00019999963158422746,
-      "loss": 46.0,
-      "step": 11311
-    },
-    {
-      "epoch": 0.8648813961045166,
-      "grad_norm": 0.000607462483458221,
-      "learning_rate": 0.00019999963151902405,
-      "loss": 46.0,
-      "step": 11312
-    },
-    {
-      "epoch": 0.8649578530879064,
-      "grad_norm": 0.001027140417136252,
-      "learning_rate": 0.00019999963145381488,
-      "loss": 46.0,
-      "step": 11313
-    },
-    {
-      "epoch": 0.8650343100712962,
-      "grad_norm": 0.002121127210557461,
-      "learning_rate": 0.00019999963138859995,
-      "loss": 46.0,
-      "step": 11314
-    },
-    {
-      "epoch": 0.8651107670546858,
-      "grad_norm": 0.003903008298948407,
-      "learning_rate": 0.00019999963132337923,
-      "loss": 46.0,
-      "step": 11315
-    },
-    {
-      "epoch": 0.8651872240380756,
-      "grad_norm": 0.0011306352680549026,
-      "learning_rate": 0.00019999963125815273,
-      "loss": 46.0,
-      "step": 11316
-    },
-    {
-      "epoch": 0.8652636810214653,
-      "grad_norm": 0.006063655484467745,
-      "learning_rate": 0.00019999963119292046,
-      "loss": 46.0,
-      "step": 11317
-    },
-    {
-      "epoch": 0.865340138004855,
-      "grad_norm": 0.0003820773563347757,
-      "learning_rate": 0.00019999963112768244,
-      "loss": 46.0,
-      "step": 11318
-    },
-    {
-      "epoch": 0.8654165949882447,
-      "grad_norm": 0.0009221856016665697,
-      "learning_rate": 0.00019999963106243865,
-      "loss": 46.0,
-      "step": 11319
-    },
-    {
-      "epoch": 0.8654930519716345,
-      "grad_norm": 0.0034840903244912624,
-      "learning_rate": 0.0001999996309971891,
-      "loss": 46.0,
-      "step": 11320
-    },
-    {
-      "epoch": 0.8655695089550242,
-      "grad_norm": 0.0037149747367948294,
-      "learning_rate": 0.00019999963093193378,
-      "loss": 46.0,
-      "step": 11321
-    },
-    {
-      "epoch": 0.8656459659384139,
-      "grad_norm": 0.0015489269280806184,
-      "learning_rate": 0.00019999963086667267,
-      "loss": 46.0,
-      "step": 11322
-    },
-    {
-      "epoch": 0.8657224229218036,
-      "grad_norm": 0.0005270421970635653,
-      "learning_rate": 0.00019999963080140581,
-      "loss": 46.0,
-      "step": 11323
-    },
-    {
-      "epoch": 0.8657988799051933,
-      "grad_norm": 0.002188637852668762,
-      "learning_rate": 0.00019999963073613316,
-      "loss": 46.0,
-      "step": 11324
-    },
-    {
-      "epoch": 0.8658753368885831,
-      "grad_norm": 0.0012622360372915864,
-      "learning_rate": 0.00019999963067085475,
-      "loss": 46.0,
-      "step": 11325
-    },
-    {
-      "epoch": 0.8659517938719727,
-      "grad_norm": 0.00045606374624185264,
-      "learning_rate": 0.00019999963060557058,
-      "loss": 46.0,
-      "step": 11326
-    },
-    {
-      "epoch": 0.8660282508553625,
-      "grad_norm": 0.001388874021358788,
-      "learning_rate": 0.00019999963054028063,
-      "loss": 46.0,
-      "step": 11327
-    },
-    {
-      "epoch": 0.8661047078387523,
-      "grad_norm": 0.0024835302028805017,
-      "learning_rate": 0.0001999996304749849,
-      "loss": 46.0,
-      "step": 11328
-    },
-    {
-      "epoch": 0.8661811648221419,
-      "grad_norm": 0.0007556821801699698,
-      "learning_rate": 0.00019999963040968343,
-      "loss": 46.0,
-      "step": 11329
-    },
-    {
-      "epoch": 0.8662576218055317,
-      "grad_norm": 0.001214748015627265,
-      "learning_rate": 0.00019999963034437616,
-      "loss": 46.0,
-      "step": 11330
-    },
-    {
-      "epoch": 0.8663340787889214,
-      "grad_norm": 0.0018981675384566188,
-      "learning_rate": 0.00019999963027906315,
-      "loss": 46.0,
-      "step": 11331
-    },
-    {
-      "epoch": 0.8664105357723111,
-      "grad_norm": 0.0024927256163209677,
-      "learning_rate": 0.00019999963021374436,
-      "loss": 46.0,
-      "step": 11332
-    },
-    {
-      "epoch": 0.8664869927557008,
-      "grad_norm": 0.0017383882077410817,
-      "learning_rate": 0.0001999996301484198,
-      "loss": 46.0,
-      "step": 11333
-    },
-    {
-      "epoch": 0.8665634497390905,
-      "grad_norm": 0.00047205790178850293,
-      "learning_rate": 0.00019999963008308943,
-      "loss": 46.0,
-      "step": 11334
-    },
-    {
-      "epoch": 0.8666399067224803,
-      "grad_norm": 0.0007868478423915803,
-      "learning_rate": 0.00019999963001775335,
-      "loss": 46.0,
-      "step": 11335
-    },
-    {
-      "epoch": 0.86671636370587,
-      "grad_norm": 0.0018248240230605006,
-      "learning_rate": 0.0001999996299524115,
-      "loss": 46.0,
-      "step": 11336
-    },
-    {
-      "epoch": 0.8667928206892597,
-      "grad_norm": 0.001080757356248796,
-      "learning_rate": 0.00019999962988706383,
-      "loss": 46.0,
-      "step": 11337
-    },
-    {
-      "epoch": 0.8668692776726494,
-      "grad_norm": 0.0012168008834123611,
-      "learning_rate": 0.00019999962982171043,
-      "loss": 46.0,
-      "step": 11338
-    },
-    {
-      "epoch": 0.8669457346560392,
-      "grad_norm": 0.0025758317206054926,
-      "learning_rate": 0.00019999962975635125,
-      "loss": 46.0,
-      "step": 11339
-    },
-    {
-      "epoch": 0.8670221916394288,
-      "grad_norm": 0.003014654852449894,
-      "learning_rate": 0.00019999962969098628,
-      "loss": 46.0,
-      "step": 11340
-    },
-    {
-      "epoch": 0.8670986486228186,
-      "grad_norm": 0.001204540953040123,
-      "learning_rate": 0.00019999962962561555,
-      "loss": 46.0,
-      "step": 11341
-    },
-    {
-      "epoch": 0.8671751056062083,
-      "grad_norm": 0.0004886524984613061,
-      "learning_rate": 0.00019999962956023909,
-      "loss": 46.0,
-      "step": 11342
-    },
-    {
-      "epoch": 0.867251562589598,
-      "grad_norm": 0.0012577750021591783,
-      "learning_rate": 0.00019999962949485682,
-      "loss": 46.0,
-      "step": 11343
-    },
-    {
-      "epoch": 0.8673280195729878,
-      "grad_norm": 0.001744294073432684,
-      "learning_rate": 0.0001999996294294688,
-      "loss": 46.0,
-      "step": 11344
-    },
-    {
-      "epoch": 0.8674044765563774,
-      "grad_norm": 0.0018252218142151833,
-      "learning_rate": 0.00019999962936407498,
-      "loss": 46.0,
-      "step": 11345
-    },
-    {
-      "epoch": 0.8674809335397672,
-      "grad_norm": 0.0011272367555648088,
-      "learning_rate": 0.00019999962929867545,
-      "loss": 46.0,
-      "step": 11346
-    },
-    {
-      "epoch": 0.867557390523157,
-      "grad_norm": 0.0003052992688026279,
-      "learning_rate": 0.00019999962923327008,
-      "loss": 46.0,
-      "step": 11347
-    },
-    {
-      "epoch": 0.8676338475065466,
-      "grad_norm": 0.0009380700648762286,
-      "learning_rate": 0.00019999962916785897,
-      "loss": 46.0,
-      "step": 11348
-    },
-    {
-      "epoch": 0.8677103044899364,
-      "grad_norm": 0.0056645493023097515,
-      "learning_rate": 0.0001999996291024421,
-      "loss": 46.0,
-      "step": 11349
-    },
-    {
-      "epoch": 0.8677867614733261,
-      "grad_norm": 0.003473804797977209,
-      "learning_rate": 0.00019999962903701946,
-      "loss": 46.0,
-      "step": 11350
-    },
-    {
-      "epoch": 0.8678632184567158,
-      "grad_norm": 0.004365224391222,
-      "learning_rate": 0.00019999962897159106,
-      "loss": 46.0,
-      "step": 11351
-    },
-    {
-      "epoch": 0.8679396754401055,
-      "grad_norm": 0.0015035831602290273,
-      "learning_rate": 0.00019999962890615686,
-      "loss": 46.0,
-      "step": 11352
-    },
-    {
-      "epoch": 0.8680161324234952,
-      "grad_norm": 0.0015193622093647718,
-      "learning_rate": 0.00019999962884071693,
-      "loss": 46.0,
-      "step": 11353
-    },
-    {
-      "epoch": 0.8680925894068849,
-      "grad_norm": 0.0034061369951814413,
-      "learning_rate": 0.00019999962877527118,
-      "loss": 46.0,
-      "step": 11354
-    },
-    {
-      "epoch": 0.8681690463902747,
-      "grad_norm": 0.0014110647607594728,
-      "learning_rate": 0.00019999962870981971,
-      "loss": 46.0,
-      "step": 11355
-    },
-    {
-      "epoch": 0.8682455033736644,
-      "grad_norm": 0.0010795751586556435,
-      "learning_rate": 0.00019999962864436244,
-      "loss": 46.0,
-      "step": 11356
-    },
-    {
-      "epoch": 0.8683219603570541,
-      "grad_norm": 0.00052928231889382,
-      "learning_rate": 0.00019999962857889943,
-      "loss": 46.0,
-      "step": 11357
-    },
-    {
-      "epoch": 0.8683984173404439,
-      "grad_norm": 0.00542057491838932,
-      "learning_rate": 0.0001999996285134306,
-      "loss": 46.0,
-      "step": 11358
-    },
-    {
-      "epoch": 0.8684748743238335,
-      "grad_norm": 0.0006601628265343606,
-      "learning_rate": 0.00019999962844795602,
-      "loss": 46.0,
-      "step": 11359
-    },
-    {
-      "epoch": 0.8685513313072233,
-      "grad_norm": 0.012000552378594875,
-      "learning_rate": 0.0001999996283824757,
-      "loss": 46.0,
-      "step": 11360
-    },
-    {
-      "epoch": 0.868627788290613,
-      "grad_norm": 0.0008332718862220645,
-      "learning_rate": 0.0001999996283169896,
-      "loss": 46.0,
-      "step": 11361
-    },
-    {
-      "epoch": 0.8687042452740027,
-      "grad_norm": 0.0010078441118821502,
-      "learning_rate": 0.00019999962825149772,
-      "loss": 46.0,
-      "step": 11362
-    },
-    {
-      "epoch": 0.8687807022573925,
-      "grad_norm": 0.0005545616731978953,
-      "learning_rate": 0.00019999962818600007,
-      "loss": 46.0,
-      "step": 11363
-    },
-    {
-      "epoch": 0.8688571592407821,
-      "grad_norm": 0.001575276255607605,
-      "learning_rate": 0.00019999962812049664,
-      "loss": 46.0,
-      "step": 11364
-    },
-    {
-      "epoch": 0.8689336162241719,
-      "grad_norm": 0.0008677774458192289,
-      "learning_rate": 0.00019999962805498746,
-      "loss": 46.0,
-      "step": 11365
-    },
-    {
-      "epoch": 0.8690100732075616,
-      "grad_norm": 0.0012390395859256387,
-      "learning_rate": 0.0001999996279894725,
-      "loss": 46.0,
-      "step": 11366
-    },
-    {
-      "epoch": 0.8690865301909513,
-      "grad_norm": 0.0005452482146210968,
-      "learning_rate": 0.00019999962792395176,
-      "loss": 46.0,
-      "step": 11367
-    },
-    {
-      "epoch": 0.869162987174341,
-      "grad_norm": 0.0007087248377501965,
-      "learning_rate": 0.0001999996278584253,
-      "loss": 46.0,
-      "step": 11368
-    },
-    {
-      "epoch": 0.8692394441577308,
-      "grad_norm": 0.0018589590908959508,
-      "learning_rate": 0.00019999962779289302,
-      "loss": 46.0,
-      "step": 11369
-    },
-    {
-      "epoch": 0.8693159011411205,
-      "grad_norm": 0.0007085982942953706,
-      "learning_rate": 0.00019999962772735498,
-      "loss": 46.0,
-      "step": 11370
-    },
-    {
-      "epoch": 0.8693923581245102,
-      "grad_norm": 0.008411173708736897,
-      "learning_rate": 0.0001999996276618112,
-      "loss": 46.0,
-      "step": 11371
-    },
-    {
-      "epoch": 0.8694688151079,
-      "grad_norm": 0.0010668684262782335,
-      "learning_rate": 0.00019999962759626163,
-      "loss": 46.0,
-      "step": 11372
-    },
-    {
-      "epoch": 0.8695452720912896,
-      "grad_norm": 0.0014721545157954097,
-      "learning_rate": 0.0001999996275307063,
-      "loss": 46.0,
-      "step": 11373
-    },
-    {
-      "epoch": 0.8696217290746794,
-      "grad_norm": 0.0007569894660264254,
-      "learning_rate": 0.00019999962746514518,
-      "loss": 46.0,
-      "step": 11374
-    },
-    {
-      "epoch": 0.869698186058069,
-      "grad_norm": 0.001771575422026217,
-      "learning_rate": 0.0001999996273995783,
-      "loss": 46.0,
-      "step": 11375
-    },
-    {
-      "epoch": 0.8697746430414588,
-      "grad_norm": 0.0004106743144802749,
-      "learning_rate": 0.00019999962733400565,
-      "loss": 46.0,
-      "step": 11376
-    },
-    {
-      "epoch": 0.8698511000248486,
-      "grad_norm": 0.0017075147479772568,
-      "learning_rate": 0.0001999996272684272,
-      "loss": 46.0,
-      "step": 11377
-    },
-    {
-      "epoch": 0.8699275570082382,
-      "grad_norm": 0.0007278526318259537,
-      "learning_rate": 0.00019999962720284304,
-      "loss": 46.0,
-      "step": 11378
-    },
-    {
-      "epoch": 0.870004013991628,
-      "grad_norm": 0.00116728525608778,
-      "learning_rate": 0.00019999962713725307,
-      "loss": 46.0,
-      "step": 11379
-    },
-    {
-      "epoch": 0.8700804709750177,
-      "grad_norm": 0.0018945814808830619,
-      "learning_rate": 0.00019999962707165737,
-      "loss": 46.0,
-      "step": 11380
-    },
-    {
-      "epoch": 0.8701569279584074,
-      "grad_norm": 0.0013922067591920495,
-      "learning_rate": 0.00019999962700605585,
-      "loss": 46.0,
-      "step": 11381
-    },
-    {
-      "epoch": 0.8702333849417971,
-      "grad_norm": 0.0015502232126891613,
-      "learning_rate": 0.00019999962694044858,
-      "loss": 46.0,
-      "step": 11382
-    },
-    {
-      "epoch": 0.8703098419251868,
-      "grad_norm": 0.0017657580319792032,
-      "learning_rate": 0.00019999962687483556,
-      "loss": 46.0,
-      "step": 11383
-    },
-    {
-      "epoch": 0.8703862989085766,
-      "grad_norm": 0.0009936497081071138,
-      "learning_rate": 0.00019999962680921672,
-      "loss": 46.0,
-      "step": 11384
-    },
-    {
-      "epoch": 0.8704627558919663,
-      "grad_norm": 0.0018627546960487962,
-      "learning_rate": 0.00019999962674359216,
-      "loss": 46.0,
-      "step": 11385
-    },
-    {
-      "epoch": 0.870539212875356,
-      "grad_norm": 0.0011218518484383821,
-      "learning_rate": 0.00019999962667796182,
-      "loss": 46.0,
-      "step": 11386
-    },
-    {
-      "epoch": 0.8706156698587457,
-      "grad_norm": 0.0012299276422709227,
-      "learning_rate": 0.00019999962661232571,
-      "loss": 46.0,
-      "step": 11387
-    },
-    {
-      "epoch": 0.8706921268421355,
-      "grad_norm": 0.0016529296990484,
-      "learning_rate": 0.00019999962654668383,
-      "loss": 46.0,
-      "step": 11388
-    },
-    {
-      "epoch": 0.8707685838255251,
-      "grad_norm": 0.000581545929890126,
-      "learning_rate": 0.00019999962648103615,
-      "loss": 46.0,
-      "step": 11389
-    },
-    {
-      "epoch": 0.8708450408089149,
-      "grad_norm": 0.000293553777737543,
-      "learning_rate": 0.00019999962641538275,
-      "loss": 46.0,
-      "step": 11390
-    },
-    {
-      "epoch": 0.8709214977923047,
-      "grad_norm": 0.002803260926157236,
-      "learning_rate": 0.00019999962634972355,
-      "loss": 46.0,
-      "step": 11391
-    },
-    {
-      "epoch": 0.8709979547756943,
-      "grad_norm": 0.0007345350459218025,
-      "learning_rate": 0.0001999996262840586,
-      "loss": 46.0,
-      "step": 11392
-    },
-    {
-      "epoch": 0.8710744117590841,
-      "grad_norm": 0.001073502586223185,
-      "learning_rate": 0.00019999962621838785,
-      "loss": 46.0,
-      "step": 11393
-    },
-    {
-      "epoch": 0.8711508687424737,
-      "grad_norm": 0.0007213017670437694,
-      "learning_rate": 0.00019999962615271138,
-      "loss": 46.0,
-      "step": 11394
-    },
-    {
-      "epoch": 0.8712273257258635,
-      "grad_norm": 0.0008859593071974814,
-      "learning_rate": 0.0001999996260870291,
-      "loss": 46.0,
-      "step": 11395
-    },
-    {
-      "epoch": 0.8713037827092532,
-      "grad_norm": 0.0007252567447721958,
-      "learning_rate": 0.00019999962602134107,
-      "loss": 46.0,
-      "step": 11396
-    },
-    {
-      "epoch": 0.8713802396926429,
-      "grad_norm": 0.0010628559393808246,
-      "learning_rate": 0.00019999962595564725,
-      "loss": 46.0,
-      "step": 11397
-    },
-    {
-      "epoch": 0.8714566966760326,
-      "grad_norm": 0.0016480223275721073,
-      "learning_rate": 0.00019999962588994767,
-      "loss": 46.0,
-      "step": 11398
-    },
-    {
-      "epoch": 0.8715331536594224,
-      "grad_norm": 0.0021526247728616,
-      "learning_rate": 0.00019999962582424233,
-      "loss": 46.0,
-      "step": 11399
-    },
-    {
-      "epoch": 0.8716096106428121,
-      "grad_norm": 0.0008695118012838066,
-      "learning_rate": 0.00019999962575853122,
-      "loss": 46.0,
-      "step": 11400
-    },
-    {
-      "epoch": 0.8716860676262018,
-      "grad_norm": 0.000764188589528203,
-      "learning_rate": 0.0001999996256928143,
-      "loss": 46.0,
-      "step": 11401
-    },
-    {
-      "epoch": 0.8717625246095916,
-      "grad_norm": 0.001357584842480719,
-      "learning_rate": 0.00019999962562709166,
-      "loss": 46.0,
-      "step": 11402
-    },
-    {
-      "epoch": 0.8718389815929812,
-      "grad_norm": 0.0009459906723350286,
-      "learning_rate": 0.00019999962556136326,
-      "loss": 46.0,
-      "step": 11403
-    },
-    {
-      "epoch": 0.871915438576371,
-      "grad_norm": 0.0006871233927085996,
-      "learning_rate": 0.00019999962549562905,
-      "loss": 46.0,
-      "step": 11404
-    },
-    {
-      "epoch": 0.8719918955597606,
-      "grad_norm": 0.000523898343089968,
-      "learning_rate": 0.00019999962542988908,
-      "loss": 46.0,
-      "step": 11405
-    },
-    {
-      "epoch": 0.8720683525431504,
-      "grad_norm": 0.0005584381869994104,
-      "learning_rate": 0.00019999962536414336,
-      "loss": 46.0,
-      "step": 11406
-    },
-    {
-      "epoch": 0.8721448095265402,
-      "grad_norm": 0.001396314473822713,
-      "learning_rate": 0.00019999962529839186,
-      "loss": 46.0,
-      "step": 11407
-    },
-    {
-      "epoch": 0.8722212665099298,
-      "grad_norm": 0.0014026289572939277,
-      "learning_rate": 0.00019999962523263457,
-      "loss": 46.0,
-      "step": 11408
-    },
-    {
-      "epoch": 0.8722977234933196,
-      "grad_norm": 0.0014972700737416744,
-      "learning_rate": 0.00019999962516687155,
-      "loss": 46.0,
-      "step": 11409
-    },
-    {
-      "epoch": 0.8723741804767093,
-      "grad_norm": 0.0042268093675374985,
-      "learning_rate": 0.00019999962510110274,
-      "loss": 46.0,
-      "step": 11410
-    },
-    {
-      "epoch": 0.872450637460099,
-      "grad_norm": 0.0016035304870456457,
-      "learning_rate": 0.00019999962503532818,
-      "loss": 46.0,
-      "step": 11411
-    },
-    {
-      "epoch": 0.8725270944434887,
-      "grad_norm": 0.000624780310317874,
-      "learning_rate": 0.00019999962496954781,
-      "loss": 46.0,
-      "step": 11412
-    },
-    {
-      "epoch": 0.8726035514268784,
-      "grad_norm": 0.000314099044771865,
-      "learning_rate": 0.00019999962490376168,
-      "loss": 46.0,
-      "step": 11413
-    },
-    {
-      "epoch": 0.8726800084102682,
-      "grad_norm": 0.0017387644620612264,
-      "learning_rate": 0.0001999996248379698,
-      "loss": 46.0,
-      "step": 11414
-    },
-    {
-      "epoch": 0.8727564653936579,
-      "grad_norm": 0.0011003370163962245,
-      "learning_rate": 0.00019999962477217214,
-      "loss": 46.0,
-      "step": 11415
-    },
-    {
-      "epoch": 0.8728329223770476,
-      "grad_norm": 0.002337732817977667,
-      "learning_rate": 0.00019999962470636871,
-      "loss": 46.0,
-      "step": 11416
-    },
-    {
-      "epoch": 0.8729093793604373,
-      "grad_norm": 0.0011581971775740385,
-      "learning_rate": 0.0001999996246405595,
-      "loss": 46.0,
-      "step": 11417
-    },
-    {
-      "epoch": 0.8729858363438271,
-      "grad_norm": 0.0002553130907472223,
-      "learning_rate": 0.00019999962457474454,
-      "loss": 46.0,
-      "step": 11418
-    },
-    {
-      "epoch": 0.8730622933272167,
-      "grad_norm": 0.0037137335166335106,
-      "learning_rate": 0.00019999962450892382,
-      "loss": 46.0,
-      "step": 11419
-    },
-    {
-      "epoch": 0.8731387503106065,
-      "grad_norm": 0.0021945994812995195,
-      "learning_rate": 0.00019999962444309732,
-      "loss": 46.0,
-      "step": 11420
-    },
-    {
-      "epoch": 0.8732152072939963,
-      "grad_norm": 0.000980915268883109,
-      "learning_rate": 0.00019999962437726503,
-      "loss": 46.0,
-      "step": 11421
-    },
-    {
-      "epoch": 0.8732916642773859,
-      "grad_norm": 0.011972527951002121,
-      "learning_rate": 0.000199999624311427,
-      "loss": 46.0,
-      "step": 11422
-    },
-    {
-      "epoch": 0.8733681212607757,
-      "grad_norm": 0.0011863175313919783,
-      "learning_rate": 0.00019999962424558317,
-      "loss": 46.0,
-      "step": 11423
-    },
-    {
-      "epoch": 0.8734445782441653,
-      "grad_norm": 0.001201195758767426,
-      "learning_rate": 0.00019999962417973358,
-      "loss": 46.0,
-      "step": 11424
-    },
-    {
-      "epoch": 0.8735210352275551,
-      "grad_norm": 0.001934562111273408,
-      "learning_rate": 0.00019999962411387822,
-      "loss": 46.0,
-      "step": 11425
-    },
-    {
-      "epoch": 0.8735974922109448,
-      "grad_norm": 0.0010991540038958192,
-      "learning_rate": 0.00019999962404801714,
-      "loss": 46.0,
-      "step": 11426
-    },
-    {
-      "epoch": 0.8736739491943345,
-      "grad_norm": 0.0019634398631751537,
-      "learning_rate": 0.00019999962398215023,
-      "loss": 46.0,
-      "step": 11427
-    },
-    {
-      "epoch": 0.8737504061777243,
-      "grad_norm": 0.00113029929343611,
-      "learning_rate": 0.00019999962391627758,
-      "loss": 46.0,
-      "step": 11428
-    },
-    {
-      "epoch": 0.873826863161114,
-      "grad_norm": 0.0027148055378347635,
-      "learning_rate": 0.00019999962385039915,
-      "loss": 46.0,
-      "step": 11429
-    },
-    {
-      "epoch": 0.8739033201445037,
-      "grad_norm": 0.0009400181588716805,
-      "learning_rate": 0.00019999962378451492,
-      "loss": 46.0,
-      "step": 11430
-    },
-    {
-      "epoch": 0.8739797771278934,
-      "grad_norm": 0.000677421863656491,
-      "learning_rate": 0.00019999962371862497,
-      "loss": 46.0,
-      "step": 11431
-    },
-    {
-      "epoch": 0.8740562341112832,
-      "grad_norm": 0.0012904831673949957,
-      "learning_rate": 0.00019999962365272925,
-      "loss": 46.0,
-      "step": 11432
-    },
-    {
-      "epoch": 0.8741326910946728,
-      "grad_norm": 0.000991379958577454,
-      "learning_rate": 0.00019999962358682773,
-      "loss": 46.0,
-      "step": 11433
-    },
-    {
-      "epoch": 0.8742091480780626,
-      "grad_norm": 0.00108067668043077,
-      "learning_rate": 0.00019999962352092046,
-      "loss": 46.0,
-      "step": 11434
-    },
-    {
-      "epoch": 0.8742856050614523,
-      "grad_norm": 0.00024224857043009251,
-      "learning_rate": 0.0001999996234550074,
-      "loss": 46.0,
-      "step": 11435
-    },
-    {
-      "epoch": 0.874362062044842,
-      "grad_norm": 0.0020552342757582664,
-      "learning_rate": 0.00019999962338908858,
-      "loss": 46.0,
-      "step": 11436
-    },
-    {
-      "epoch": 0.8744385190282318,
-      "grad_norm": 0.002490504877641797,
-      "learning_rate": 0.000199999623323164,
-      "loss": 46.0,
-      "step": 11437
-    },
-    {
-      "epoch": 0.8745149760116214,
-      "grad_norm": 0.003215127857401967,
-      "learning_rate": 0.00019999962325723366,
-      "loss": 46.0,
-      "step": 11438
-    },
-    {
-      "epoch": 0.8745914329950112,
-      "grad_norm": 0.0010782964527606964,
-      "learning_rate": 0.00019999962319129752,
-      "loss": 46.0,
-      "step": 11439
-    },
-    {
-      "epoch": 0.874667889978401,
-      "grad_norm": 0.002760393312200904,
-      "learning_rate": 0.00019999962312535562,
-      "loss": 46.0,
-      "step": 11440
-    },
-    {
-      "epoch": 0.8747443469617906,
-      "grad_norm": 0.008062370121479034,
-      "learning_rate": 0.00019999962305940796,
-      "loss": 46.0,
-      "step": 11441
-    },
-    {
-      "epoch": 0.8748208039451804,
-      "grad_norm": 0.0013652686029672623,
-      "learning_rate": 0.0001999996229934545,
-      "loss": 46.0,
-      "step": 11442
-    },
-    {
-      "epoch": 0.87489726092857,
-      "grad_norm": 0.0006283313850872219,
-      "learning_rate": 0.00019999962292749533,
-      "loss": 46.0,
-      "step": 11443
-    },
-    {
-      "epoch": 0.8749737179119598,
-      "grad_norm": 0.0018913072999566793,
-      "learning_rate": 0.00019999962286153036,
-      "loss": 46.0,
-      "step": 11444
-    },
-    {
-      "epoch": 0.8750501748953495,
-      "grad_norm": 0.0015715928748250008,
-      "learning_rate": 0.0001999996227955596,
-      "loss": 46.0,
-      "step": 11445
-    },
-    {
-      "epoch": 0.8751266318787392,
-      "grad_norm": 0.0009337567607872188,
-      "learning_rate": 0.0001999996227295831,
-      "loss": 46.0,
-      "step": 11446
-    },
-    {
-      "epoch": 0.8752030888621289,
-      "grad_norm": 0.0008720124023966491,
-      "learning_rate": 0.00019999962266360082,
-      "loss": 46.0,
-      "step": 11447
-    },
-    {
-      "epoch": 0.8752795458455187,
-      "grad_norm": 0.0028761515859514475,
-      "learning_rate": 0.00019999962259761278,
-      "loss": 46.0,
-      "step": 11448
-    },
-    {
-      "epoch": 0.8753560028289084,
-      "grad_norm": 0.0021358646918088198,
-      "learning_rate": 0.00019999962253161897,
-      "loss": 46.0,
-      "step": 11449
-    },
-    {
-      "epoch": 0.8754324598122981,
-      "grad_norm": 0.0011539262486621737,
-      "learning_rate": 0.00019999962246561935,
-      "loss": 46.0,
-      "step": 11450
-    },
-    {
-      "epoch": 0.8755089167956879,
-      "grad_norm": 0.0015491455560550094,
-      "learning_rate": 0.00019999962239961402,
-      "loss": 46.0,
-      "step": 11451
-    },
-    {
-      "epoch": 0.8755853737790775,
-      "grad_norm": 0.0028538417536765337,
-      "learning_rate": 0.00019999962233360288,
-      "loss": 46.0,
-      "step": 11452
-    },
-    {
-      "epoch": 0.8756618307624673,
-      "grad_norm": 0.0008607959607616067,
-      "learning_rate": 0.000199999622267586,
-      "loss": 46.0,
-      "step": 11453
-    },
-    {
-      "epoch": 0.8757382877458569,
-      "grad_norm": 0.0011676391586661339,
-      "learning_rate": 0.00019999962220156335,
-      "loss": 46.0,
-      "step": 11454
-    },
-    {
-      "epoch": 0.8758147447292467,
-      "grad_norm": 0.0004308215866331011,
-      "learning_rate": 0.0001999996221355349,
-      "loss": 46.0,
-      "step": 11455
-    },
-    {
-      "epoch": 0.8758912017126365,
-      "grad_norm": 0.0004989307490177453,
-      "learning_rate": 0.0001999996220695007,
-      "loss": 46.0,
-      "step": 11456
-    },
-    {
-      "epoch": 0.8759676586960261,
-      "grad_norm": 0.0005106266471557319,
-      "learning_rate": 0.00019999962200346072,
-      "loss": 46.0,
-      "step": 11457
-    },
-    {
-      "epoch": 0.8760441156794159,
-      "grad_norm": 0.0008801653166301548,
-      "learning_rate": 0.00019999962193741497,
-      "loss": 46.0,
-      "step": 11458
-    },
-    {
-      "epoch": 0.8761205726628056,
-      "grad_norm": 0.0011956244707107544,
-      "learning_rate": 0.00019999962187136348,
-      "loss": 46.0,
-      "step": 11459
-    },
-    {
-      "epoch": 0.8761970296461953,
-      "grad_norm": 0.0030010908376425505,
-      "learning_rate": 0.00019999962180530619,
-      "loss": 46.0,
-      "step": 11460
-    },
-    {
-      "epoch": 0.876273486629585,
-      "grad_norm": 0.0036698300391435623,
-      "learning_rate": 0.00019999962173924312,
-      "loss": 46.0,
-      "step": 11461
-    },
-    {
-      "epoch": 0.8763499436129748,
-      "grad_norm": 0.0029740636236965656,
-      "learning_rate": 0.00019999962167317433,
-      "loss": 46.0,
-      "step": 11462
-    },
-    {
-      "epoch": 0.8764264005963645,
-      "grad_norm": 0.0015087099745869637,
-      "learning_rate": 0.00019999962160709972,
-      "loss": 46.0,
-      "step": 11463
-    },
-    {
-      "epoch": 0.8765028575797542,
-      "grad_norm": 0.0025338975246995687,
-      "learning_rate": 0.00019999962154101936,
-      "loss": 46.0,
-      "step": 11464
-    },
-    {
-      "epoch": 0.8765793145631439,
-      "grad_norm": 0.00045561863225884736,
-      "learning_rate": 0.0001999996214749332,
-      "loss": 46.0,
-      "step": 11465
-    },
-    {
-      "epoch": 0.8766557715465336,
-      "grad_norm": 0.0019635034259408712,
-      "learning_rate": 0.00019999962140884132,
-      "loss": 46.0,
-      "step": 11466
-    },
-    {
-      "epoch": 0.8767322285299234,
-      "grad_norm": 0.000976012903265655,
-      "learning_rate": 0.00019999962134274366,
-      "loss": 46.0,
-      "step": 11467
-    },
-    {
-      "epoch": 0.876808685513313,
-      "grad_norm": 0.0024508091155439615,
-      "learning_rate": 0.00019999962127664023,
-      "loss": 46.0,
-      "step": 11468
-    },
-    {
-      "epoch": 0.8768851424967028,
-      "grad_norm": 0.0013388919178396463,
-      "learning_rate": 0.000199999621210531,
-      "loss": 46.0,
-      "step": 11469
-    },
-    {
-      "epoch": 0.8769615994800926,
-      "grad_norm": 0.0010337402345612645,
-      "learning_rate": 0.00019999962114441603,
-      "loss": 46.0,
-      "step": 11470
-    },
-    {
-      "epoch": 0.8770380564634822,
-      "grad_norm": 0.0010264250449836254,
-      "learning_rate": 0.0001999996210782953,
-      "loss": 46.0,
-      "step": 11471
-    },
-    {
-      "epoch": 0.877114513446872,
-      "grad_norm": 0.0011336618335917592,
-      "learning_rate": 0.00019999962101216877,
-      "loss": 46.0,
-      "step": 11472
-    },
-    {
-      "epoch": 0.8771909704302617,
-      "grad_norm": 0.0020531604532152414,
-      "learning_rate": 0.0001999996209460365,
-      "loss": 46.0,
-      "step": 11473
-    },
-    {
-      "epoch": 0.8772674274136514,
-      "grad_norm": 0.002687373897060752,
-      "learning_rate": 0.00019999962087989843,
-      "loss": 46.0,
-      "step": 11474
-    },
-    {
-      "epoch": 0.8773438843970411,
-      "grad_norm": 0.0014271598774939775,
-      "learning_rate": 0.00019999962081375462,
-      "loss": 46.0,
-      "step": 11475
-    },
-    {
-      "epoch": 0.8774203413804308,
-      "grad_norm": 0.0007046546088531613,
-      "learning_rate": 0.000199999620747605,
-      "loss": 46.0,
-      "step": 11476
-    },
-    {
-      "epoch": 0.8774967983638206,
-      "grad_norm": 0.0013938278425484896,
-      "learning_rate": 0.00019999962068144967,
-      "loss": 46.0,
-      "step": 11477
-    },
-    {
-      "epoch": 0.8775732553472103,
-      "grad_norm": 0.0018538323929533362,
-      "learning_rate": 0.0001999996206152885,
-      "loss": 46.0,
-      "step": 11478
-    },
-    {
-      "epoch": 0.8776497123306,
-      "grad_norm": 0.0005432175239548087,
-      "learning_rate": 0.00019999962054912163,
-      "loss": 46.0,
-      "step": 11479
-    },
-    {
-      "epoch": 0.8777261693139897,
-      "grad_norm": 0.0021084975451231003,
-      "learning_rate": 0.00019999962048294895,
-      "loss": 46.0,
-      "step": 11480
-    },
-    {
-      "epoch": 0.8778026262973795,
-      "grad_norm": 0.004460969474166632,
-      "learning_rate": 0.00019999962041677052,
-      "loss": 46.0,
-      "step": 11481
-    },
-    {
-      "epoch": 0.8778790832807691,
-      "grad_norm": 0.006277133245021105,
-      "learning_rate": 0.0001999996203505863,
-      "loss": 46.0,
-      "step": 11482
-    },
-    {
-      "epoch": 0.8779555402641589,
-      "grad_norm": 0.0010204113787040114,
-      "learning_rate": 0.00019999962028439635,
-      "loss": 46.0,
-      "step": 11483
-    },
-    {
-      "epoch": 0.8780319972475485,
-      "grad_norm": 0.0017665511695668101,
-      "learning_rate": 0.00019999962021820058,
-      "loss": 46.0,
-      "step": 11484
-    },
-    {
-      "epoch": 0.8781084542309383,
-      "grad_norm": 0.0009677269263193011,
-      "learning_rate": 0.00019999962015199906,
-      "loss": 46.0,
-      "step": 11485
-    },
-    {
-      "epoch": 0.8781849112143281,
-      "grad_norm": 0.0011772130383178592,
-      "learning_rate": 0.0001999996200857918,
-      "loss": 46.0,
-      "step": 11486
-    },
-    {
-      "epoch": 0.8782613681977177,
-      "grad_norm": 0.0035982343833893538,
-      "learning_rate": 0.0001999996200195787,
-      "loss": 46.0,
-      "step": 11487
-    },
-    {
-      "epoch": 0.8783378251811075,
-      "grad_norm": 0.0003681460511870682,
-      "learning_rate": 0.00019999961995335988,
-      "loss": 46.0,
-      "step": 11488
-    },
-    {
-      "epoch": 0.8784142821644972,
-      "grad_norm": 0.010014706291258335,
-      "learning_rate": 0.0001999996198871353,
-      "loss": 46.0,
-      "step": 11489
-    },
-    {
-      "epoch": 0.8784907391478869,
-      "grad_norm": 0.0012948894873261452,
-      "learning_rate": 0.0001999996198209049,
-      "loss": 46.0,
-      "step": 11490
-    },
-    {
-      "epoch": 0.8785671961312767,
-      "grad_norm": 0.0012301455717533827,
-      "learning_rate": 0.00019999961975466878,
-      "loss": 46.0,
-      "step": 11491
-    },
-    {
-      "epoch": 0.8786436531146664,
-      "grad_norm": 0.0006541036418639123,
-      "learning_rate": 0.0001999996196884269,
-      "loss": 46.0,
-      "step": 11492
-    },
-    {
-      "epoch": 0.8787201100980561,
-      "grad_norm": 0.0022805549670010805,
-      "learning_rate": 0.00019999961962217922,
-      "loss": 46.0,
-      "step": 11493
-    },
-    {
-      "epoch": 0.8787965670814458,
-      "grad_norm": 0.0005298719624988735,
-      "learning_rate": 0.00019999961955592576,
-      "loss": 46.0,
-      "step": 11494
-    },
-    {
-      "epoch": 0.8788730240648355,
-      "grad_norm": 0.0003304577840026468,
-      "learning_rate": 0.00019999961948966654,
-      "loss": 46.0,
-      "step": 11495
-    },
-    {
-      "epoch": 0.8789494810482252,
-      "grad_norm": 0.0008061114349402487,
-      "learning_rate": 0.00019999961942340156,
-      "loss": 46.0,
-      "step": 11496
-    },
-    {
-      "epoch": 0.879025938031615,
-      "grad_norm": 0.003267186926677823,
-      "learning_rate": 0.00019999961935713084,
-      "loss": 46.0,
-      "step": 11497
-    },
-    {
-      "epoch": 0.8791023950150046,
-      "grad_norm": 0.0009649894782342017,
-      "learning_rate": 0.0001999996192908543,
-      "loss": 46.0,
-      "step": 11498
-    },
-    {
-      "epoch": 0.8791788519983944,
-      "grad_norm": 0.0018604411743581295,
-      "learning_rate": 0.00019999961922457203,
-      "loss": 46.0,
-      "step": 11499
-    },
-    {
-      "epoch": 0.8792553089817842,
-      "grad_norm": 0.001565420301631093,
-      "learning_rate": 0.00019999961915828396,
-      "loss": 46.0,
-      "step": 11500
-    },
-    {
-      "epoch": 0.8793317659651738,
-      "grad_norm": 0.0007320715230889618,
-      "learning_rate": 0.00019999961909199012,
-      "loss": 46.0,
-      "step": 11501
-    },
-    {
-      "epoch": 0.8794082229485636,
-      "grad_norm": 0.001505018793977797,
-      "learning_rate": 0.00019999961902569054,
-      "loss": 46.0,
-      "step": 11502
-    },
-    {
-      "epoch": 0.8794846799319533,
-      "grad_norm": 0.0018255822360515594,
-      "learning_rate": 0.00019999961895938518,
-      "loss": 46.0,
-      "step": 11503
-    },
-    {
-      "epoch": 0.879561136915343,
-      "grad_norm": 0.0010151483584195375,
-      "learning_rate": 0.00019999961889307404,
-      "loss": 46.0,
-      "step": 11504
-    },
-    {
-      "epoch": 0.8796375938987328,
-      "grad_norm": 0.000422679033363238,
-      "learning_rate": 0.00019999961882675714,
-      "loss": 46.0,
-      "step": 11505
-    },
-    {
-      "epoch": 0.8797140508821224,
-      "grad_norm": 0.011802221648395061,
-      "learning_rate": 0.00019999961876043446,
-      "loss": 46.0,
-      "step": 11506
-    },
-    {
-      "epoch": 0.8797905078655122,
-      "grad_norm": 0.0008850020822137594,
-      "learning_rate": 0.000199999618694106,
-      "loss": 46.0,
-      "step": 11507
-    },
-    {
-      "epoch": 0.8798669648489019,
-      "grad_norm": 0.001157082268036902,
-      "learning_rate": 0.00019999961862777178,
-      "loss": 46.0,
-      "step": 11508
-    },
-    {
-      "epoch": 0.8799434218322916,
-      "grad_norm": 0.0022078221663832664,
-      "learning_rate": 0.00019999961856143178,
-      "loss": 46.0,
-      "step": 11509
-    },
-    {
-      "epoch": 0.8800198788156813,
-      "grad_norm": 0.0013158532092347741,
-      "learning_rate": 0.00019999961849508603,
-      "loss": 46.0,
-      "step": 11510
-    },
-    {
-      "epoch": 0.8800963357990711,
-      "grad_norm": 0.00198129890486598,
-      "learning_rate": 0.0001999996184287345,
-      "loss": 46.0,
-      "step": 11511
-    },
-    {
-      "epoch": 0.8801727927824607,
-      "grad_norm": 0.00289785023778677,
-      "learning_rate": 0.00019999961836237722,
-      "loss": 46.0,
-      "step": 11512
-    },
-    {
-      "epoch": 0.8802492497658505,
-      "grad_norm": 0.001195678603835404,
-      "learning_rate": 0.00019999961829601418,
-      "loss": 46.0,
-      "step": 11513
-    },
-    {
-      "epoch": 0.8803257067492402,
-      "grad_norm": 0.0008867033175192773,
-      "learning_rate": 0.00019999961822964534,
-      "loss": 46.0,
-      "step": 11514
-    },
-    {
-      "epoch": 0.8804021637326299,
-      "grad_norm": 0.0015654033049941063,
-      "learning_rate": 0.00019999961816327073,
-      "loss": 46.0,
-      "step": 11515
-    },
-    {
-      "epoch": 0.8804786207160197,
-      "grad_norm": 0.00294181308709085,
-      "learning_rate": 0.00019999961809689037,
-      "loss": 46.0,
-      "step": 11516
-    },
-    {
-      "epoch": 0.8805550776994093,
-      "grad_norm": 0.0011055272771045566,
-      "learning_rate": 0.0001999996180305042,
-      "loss": 46.0,
-      "step": 11517
-    },
-    {
-      "epoch": 0.8806315346827991,
-      "grad_norm": 0.0006200874340720475,
-      "learning_rate": 0.0001999996179641123,
-      "loss": 46.0,
-      "step": 11518
-    },
-    {
-      "epoch": 0.8807079916661888,
-      "grad_norm": 0.0015877856640145183,
-      "learning_rate": 0.00019999961789771465,
-      "loss": 46.0,
-      "step": 11519
-    },
-    {
-      "epoch": 0.8807844486495785,
-      "grad_norm": 0.00277850404381752,
-      "learning_rate": 0.00019999961783131117,
-      "loss": 46.0,
-      "step": 11520
-    },
-    {
-      "epoch": 0.8808609056329683,
-      "grad_norm": 0.0005010172608308494,
-      "learning_rate": 0.00019999961776490197,
-      "loss": 46.0,
-      "step": 11521
-    },
-    {
-      "epoch": 0.880937362616358,
-      "grad_norm": 0.00411998899653554,
-      "learning_rate": 0.00019999961769848697,
-      "loss": 46.0,
-      "step": 11522
-    },
-    {
-      "epoch": 0.8810138195997477,
-      "grad_norm": 0.0008906330913305283,
-      "learning_rate": 0.00019999961763206623,
-      "loss": 46.0,
-      "step": 11523
-    },
-    {
-      "epoch": 0.8810902765831374,
-      "grad_norm": 0.0031794554088264704,
-      "learning_rate": 0.0001999996175656397,
-      "loss": 46.0,
-      "step": 11524
-    },
-    {
-      "epoch": 0.8811667335665271,
-      "grad_norm": 0.000812284357380122,
-      "learning_rate": 0.00019999961749920741,
-      "loss": 46.0,
-      "step": 11525
-    },
-    {
-      "epoch": 0.8812431905499168,
-      "grad_norm": 0.0008765440434217453,
-      "learning_rate": 0.00019999961743276935,
-      "loss": 46.0,
-      "step": 11526
-    },
-    {
-      "epoch": 0.8813196475333066,
-      "grad_norm": 0.009660166688263416,
-      "learning_rate": 0.0001999996173663255,
-      "loss": 46.0,
-      "step": 11527
-    },
-    {
-      "epoch": 0.8813961045166963,
-      "grad_norm": 0.0019364120671525598,
-      "learning_rate": 0.0001999996172998759,
-      "loss": 46.0,
-      "step": 11528
-    },
-    {
-      "epoch": 0.881472561500086,
-      "grad_norm": 0.010289045050740242,
-      "learning_rate": 0.0001999996172334205,
-      "loss": 46.0,
-      "step": 11529
-    },
-    {
-      "epoch": 0.8815490184834758,
-      "grad_norm": 0.0009243633830919862,
-      "learning_rate": 0.00019999961716695938,
-      "loss": 46.0,
-      "step": 11530
-    },
-    {
-      "epoch": 0.8816254754668654,
-      "grad_norm": 0.0007554742624051869,
-      "learning_rate": 0.0001999996171004925,
-      "loss": 46.0,
-      "step": 11531
-    },
-    {
-      "epoch": 0.8817019324502552,
-      "grad_norm": 0.000755980028770864,
-      "learning_rate": 0.0001999996170340198,
-      "loss": 46.0,
-      "step": 11532
-    },
-    {
-      "epoch": 0.881778389433645,
-      "grad_norm": 0.0024595565628260374,
-      "learning_rate": 0.0001999996169675413,
-      "loss": 46.0,
-      "step": 11533
-    },
-    {
-      "epoch": 0.8818548464170346,
-      "grad_norm": 0.0010758370626717806,
-      "learning_rate": 0.00019999961690105714,
-      "loss": 46.0,
-      "step": 11534
-    },
-    {
-      "epoch": 0.8819313034004244,
-      "grad_norm": 0.0009160145418718457,
-      "learning_rate": 0.00019999961683456711,
-      "loss": 46.0,
-      "step": 11535
-    },
-    {
-      "epoch": 0.882007760383814,
-      "grad_norm": 0.0012632475700229406,
-      "learning_rate": 0.00019999961676807137,
-      "loss": 46.0,
-      "step": 11536
-    },
-    {
-      "epoch": 0.8820842173672038,
-      "grad_norm": 0.0011564275482669473,
-      "learning_rate": 0.00019999961670156985,
-      "loss": 46.0,
-      "step": 11537
-    },
-    {
-      "epoch": 0.8821606743505935,
-      "grad_norm": 0.0015774703351780772,
-      "learning_rate": 0.00019999961663506256,
-      "loss": 46.0,
-      "step": 11538
-    },
-    {
-      "epoch": 0.8822371313339832,
-      "grad_norm": 0.001014707493595779,
-      "learning_rate": 0.00019999961656854947,
-      "loss": 46.0,
-      "step": 11539
-    },
-    {
-      "epoch": 0.8823135883173729,
-      "grad_norm": 0.001826371531933546,
-      "learning_rate": 0.00019999961650203063,
-      "loss": 46.0,
-      "step": 11540
-    },
-    {
-      "epoch": 0.8823900453007627,
-      "grad_norm": 0.0008806149126030505,
-      "learning_rate": 0.00019999961643550601,
-      "loss": 46.0,
-      "step": 11541
-    },
-    {
-      "epoch": 0.8824665022841524,
-      "grad_norm": 0.0020287863444536924,
-      "learning_rate": 0.00019999961636897566,
-      "loss": 46.0,
-      "step": 11542
-    },
-    {
-      "epoch": 0.8825429592675421,
-      "grad_norm": 0.008840177208185196,
-      "learning_rate": 0.00019999961630243952,
-      "loss": 46.0,
-      "step": 11543
-    },
-    {
-      "epoch": 0.8826194162509318,
-      "grad_norm": 0.0022187193389981985,
-      "learning_rate": 0.0001999996162358976,
-      "loss": 46.0,
-      "step": 11544
-    },
-    {
-      "epoch": 0.8826958732343215,
-      "grad_norm": 0.000950249785091728,
-      "learning_rate": 0.0001999996161693499,
-      "loss": 46.0,
-      "step": 11545
-    },
-    {
-      "epoch": 0.8827723302177113,
-      "grad_norm": 0.0014825165271759033,
-      "learning_rate": 0.00019999961610279646,
-      "loss": 46.0,
-      "step": 11546
-    },
-    {
-      "epoch": 0.8828487872011009,
-      "grad_norm": 0.0013281184947118163,
-      "learning_rate": 0.00019999961603623723,
-      "loss": 46.0,
-      "step": 11547
-    },
-    {
-      "epoch": 0.8829252441844907,
-      "grad_norm": 0.0014837002381682396,
-      "learning_rate": 0.00019999961596967226,
-      "loss": 46.0,
-      "step": 11548
-    },
-    {
-      "epoch": 0.8830017011678805,
-      "grad_norm": 0.0020646746270358562,
-      "learning_rate": 0.0001999996159031015,
-      "loss": 46.0,
-      "step": 11549
-    },
-    {
-      "epoch": 0.8830781581512701,
-      "grad_norm": 0.004786214325577021,
-      "learning_rate": 0.00019999961583652494,
-      "loss": 46.0,
-      "step": 11550
-    },
-    {
-      "epoch": 0.8831546151346599,
-      "grad_norm": 0.0009910727385431528,
-      "learning_rate": 0.00019999961576994265,
-      "loss": 46.0,
-      "step": 11551
-    },
-    {
-      "epoch": 0.8832310721180496,
-      "grad_norm": 0.004271871875971556,
-      "learning_rate": 0.00019999961570335456,
-      "loss": 46.0,
-      "step": 11552
-    },
-    {
-      "epoch": 0.8833075291014393,
-      "grad_norm": 0.0004526250122580677,
-      "learning_rate": 0.00019999961563676075,
-      "loss": 46.0,
-      "step": 11553
-    },
-    {
-      "epoch": 0.883383986084829,
-      "grad_norm": 0.00037091010017320514,
-      "learning_rate": 0.00019999961557016114,
-      "loss": 46.0,
-      "step": 11554
-    },
-    {
-      "epoch": 0.8834604430682187,
-      "grad_norm": 0.0019659800454974174,
-      "learning_rate": 0.00019999961550355578,
-      "loss": 46.0,
-      "step": 11555
-    },
-    {
-      "epoch": 0.8835369000516085,
-      "grad_norm": 0.003968382719904184,
-      "learning_rate": 0.00019999961543694462,
-      "loss": 46.0,
-      "step": 11556
-    },
-    {
-      "epoch": 0.8836133570349982,
-      "grad_norm": 0.0007019545300863683,
-      "learning_rate": 0.00019999961537032769,
-      "loss": 46.0,
-      "step": 11557
-    },
-    {
-      "epoch": 0.8836898140183879,
-      "grad_norm": 0.00446880841627717,
-      "learning_rate": 0.000199999615303705,
-      "loss": 46.0,
-      "step": 11558
-    },
-    {
-      "epoch": 0.8837662710017776,
-      "grad_norm": 0.0005624861223623157,
-      "learning_rate": 0.00019999961523707656,
-      "loss": 46.0,
-      "step": 11559
-    },
-    {
-      "epoch": 0.8838427279851674,
-      "grad_norm": 0.001459189341403544,
-      "learning_rate": 0.00019999961517044233,
-      "loss": 46.0,
-      "step": 11560
-    },
-    {
-      "epoch": 0.883919184968557,
-      "grad_norm": 0.0010422280756756663,
-      "learning_rate": 0.00019999961510380233,
-      "loss": 46.0,
-      "step": 11561
-    },
-    {
-      "epoch": 0.8839956419519468,
-      "grad_norm": 0.0006864677416160703,
-      "learning_rate": 0.00019999961503715656,
-      "loss": 46.0,
-      "step": 11562
-    },
-    {
-      "epoch": 0.8840720989353366,
-      "grad_norm": 0.0005981221911497414,
-      "learning_rate": 0.00019999961497050504,
-      "loss": 46.0,
-      "step": 11563
-    },
-    {
-      "epoch": 0.8841485559187262,
-      "grad_norm": 0.0005547473556362092,
-      "learning_rate": 0.00019999961490384772,
-      "loss": 46.0,
-      "step": 11564
-    },
-    {
-      "epoch": 0.884225012902116,
-      "grad_norm": 0.001367216114886105,
-      "learning_rate": 0.00019999961483718466,
-      "loss": 46.0,
-      "step": 11565
-    },
-    {
-      "epoch": 0.8843014698855056,
-      "grad_norm": 0.0006109584355726838,
-      "learning_rate": 0.00019999961477051582,
-      "loss": 46.0,
-      "step": 11566
-    },
-    {
-      "epoch": 0.8843779268688954,
-      "grad_norm": 0.0009721220703795552,
-      "learning_rate": 0.0001999996147038412,
-      "loss": 46.0,
-      "step": 11567
-    },
-    {
-      "epoch": 0.8844543838522851,
-      "grad_norm": 0.0004529489378910512,
-      "learning_rate": 0.00019999961463716082,
-      "loss": 46.0,
-      "step": 11568
-    },
-    {
-      "epoch": 0.8845308408356748,
-      "grad_norm": 0.001516729942522943,
-      "learning_rate": 0.00019999961457047466,
-      "loss": 46.0,
-      "step": 11569
-    },
-    {
-      "epoch": 0.8846072978190646,
-      "grad_norm": 0.0006887958734296262,
-      "learning_rate": 0.00019999961450378276,
-      "loss": 46.0,
-      "step": 11570
-    },
-    {
-      "epoch": 0.8846837548024543,
-      "grad_norm": 0.0011303386418148875,
-      "learning_rate": 0.00019999961443708505,
-      "loss": 46.0,
-      "step": 11571
-    },
-    {
-      "epoch": 0.884760211785844,
-      "grad_norm": 0.0038444004021584988,
-      "learning_rate": 0.0001999996143703816,
-      "loss": 46.0,
-      "step": 11572
-    },
-    {
-      "epoch": 0.8848366687692337,
-      "grad_norm": 0.037785839289426804,
-      "learning_rate": 0.00019999961430367238,
-      "loss": 46.0,
-      "step": 11573
-    },
-    {
-      "epoch": 0.8849131257526234,
-      "grad_norm": 0.0011691901599988341,
-      "learning_rate": 0.00019999961423695738,
-      "loss": 46.0,
-      "step": 11574
-    },
-    {
-      "epoch": 0.8849895827360131,
-      "grad_norm": 0.0053506395779550076,
-      "learning_rate": 0.00019999961417023663,
-      "loss": 46.0,
-      "step": 11575
-    },
-    {
-      "epoch": 0.8850660397194029,
-      "grad_norm": 0.0003868041967507452,
-      "learning_rate": 0.0001999996141035101,
-      "loss": 46.0,
-      "step": 11576
-    },
-    {
-      "epoch": 0.8851424967027925,
-      "grad_norm": 0.0008278048480860889,
-      "learning_rate": 0.00019999961403677777,
-      "loss": 46.0,
-      "step": 11577
-    },
-    {
-      "epoch": 0.8852189536861823,
-      "grad_norm": 0.0013785305200144649,
-      "learning_rate": 0.0001999996139700397,
-      "loss": 46.0,
-      "step": 11578
-    },
-    {
-      "epoch": 0.8852954106695721,
-      "grad_norm": 0.0004402343765832484,
-      "learning_rate": 0.00019999961390329584,
-      "loss": 46.0,
-      "step": 11579
-    },
-    {
-      "epoch": 0.8853718676529617,
-      "grad_norm": 0.00043451273813843727,
-      "learning_rate": 0.00019999961383654623,
-      "loss": 46.0,
-      "step": 11580
-    },
-    {
-      "epoch": 0.8854483246363515,
-      "grad_norm": 0.0015124910278245807,
-      "learning_rate": 0.00019999961376979087,
-      "loss": 46.0,
-      "step": 11581
-    },
-    {
-      "epoch": 0.8855247816197412,
-      "grad_norm": 0.0009297903161495924,
-      "learning_rate": 0.0001999996137030297,
-      "loss": 46.0,
-      "step": 11582
-    },
-    {
-      "epoch": 0.8856012386031309,
-      "grad_norm": 0.00126979174092412,
-      "learning_rate": 0.00019999961363626278,
-      "loss": 46.0,
-      "step": 11583
-    },
-    {
-      "epoch": 0.8856776955865207,
-      "grad_norm": 0.0021595198195427656,
-      "learning_rate": 0.00019999961356949008,
-      "loss": 46.0,
-      "step": 11584
-    },
-    {
-      "epoch": 0.8857541525699103,
-      "grad_norm": 0.0016694475198164582,
-      "learning_rate": 0.00019999961350271163,
-      "loss": 46.0,
-      "step": 11585
-    },
-    {
-      "epoch": 0.8858306095533001,
-      "grad_norm": 0.003402827540412545,
-      "learning_rate": 0.0001999996134359274,
-      "loss": 46.0,
-      "step": 11586
-    },
-    {
-      "epoch": 0.8859070665366898,
-      "grad_norm": 0.002378500299528241,
-      "learning_rate": 0.0001999996133691374,
-      "loss": 46.0,
-      "step": 11587
-    },
-    {
-      "epoch": 0.8859835235200795,
-      "grad_norm": 0.0005293606664054096,
-      "learning_rate": 0.00019999961330234163,
-      "loss": 46.0,
-      "step": 11588
-    },
-    {
-      "epoch": 0.8860599805034692,
-      "grad_norm": 0.001085847383365035,
-      "learning_rate": 0.0001999996132355401,
-      "loss": 46.0,
-      "step": 11589
-    },
-    {
-      "epoch": 0.886136437486859,
-      "grad_norm": 0.0016055391170084476,
-      "learning_rate": 0.00019999961316873277,
-      "loss": 46.0,
-      "step": 11590
-    },
-    {
-      "epoch": 0.8862128944702486,
-      "grad_norm": 0.001958891050890088,
-      "learning_rate": 0.0001999996131019197,
-      "loss": 46.0,
-      "step": 11591
-    },
-    {
-      "epoch": 0.8862893514536384,
-      "grad_norm": 0.0007393836276605725,
-      "learning_rate": 0.00019999961303510087,
-      "loss": 46.0,
-      "step": 11592
-    },
-    {
-      "epoch": 0.8863658084370282,
-      "grad_norm": 0.0020159322302788496,
-      "learning_rate": 0.00019999961296827623,
-      "loss": 46.0,
-      "step": 11593
-    },
-    {
-      "epoch": 0.8864422654204178,
-      "grad_norm": 0.001277353148907423,
-      "learning_rate": 0.00019999961290144587,
-      "loss": 46.0,
-      "step": 11594
-    },
-    {
-      "epoch": 0.8865187224038076,
-      "grad_norm": 0.0013615416828542948,
-      "learning_rate": 0.00019999961283460969,
-      "loss": 46.0,
-      "step": 11595
-    },
-    {
-      "epoch": 0.8865951793871972,
-      "grad_norm": 0.0003089065430685878,
-      "learning_rate": 0.00019999961276776778,
-      "loss": 46.0,
-      "step": 11596
-    },
-    {
-      "epoch": 0.886671636370587,
-      "grad_norm": 0.0004677719553001225,
-      "learning_rate": 0.0001999996127009201,
-      "loss": 46.0,
-      "step": 11597
-    },
-    {
-      "epoch": 0.8867480933539768,
-      "grad_norm": 0.0008280706242658198,
-      "learning_rate": 0.00019999961263406663,
-      "loss": 46.0,
-      "step": 11598
-    },
-    {
-      "epoch": 0.8868245503373664,
-      "grad_norm": 0.00181598833296448,
-      "learning_rate": 0.00019999961256720737,
-      "loss": 46.0,
-      "step": 11599
-    },
-    {
-      "epoch": 0.8869010073207562,
-      "grad_norm": 0.002929729875177145,
-      "learning_rate": 0.00019999961250034238,
-      "loss": 46.0,
-      "step": 11600
-    },
-    {
-      "epoch": 0.8869774643041459,
-      "grad_norm": 0.00047074060421437025,
-      "learning_rate": 0.0001999996124334716,
-      "loss": 46.0,
-      "step": 11601
-    },
-    {
-      "epoch": 0.8870539212875356,
-      "grad_norm": 0.0009137978195212781,
-      "learning_rate": 0.00019999961236659506,
-      "loss": 46.0,
-      "step": 11602
-    },
-    {
-      "epoch": 0.8871303782709253,
-      "grad_norm": 0.0007314422400668263,
-      "learning_rate": 0.00019999961229971277,
-      "loss": 46.0,
-      "step": 11603
-    },
-    {
-      "epoch": 0.8872068352543151,
-      "grad_norm": 0.0011030100286006927,
-      "learning_rate": 0.00019999961223282465,
-      "loss": 46.0,
-      "step": 11604
-    },
-    {
-      "epoch": 0.8872832922377047,
-      "grad_norm": 0.00048081023851409554,
-      "learning_rate": 0.00019999961216593081,
-      "loss": 46.0,
-      "step": 11605
-    },
-    {
-      "epoch": 0.8873597492210945,
-      "grad_norm": 0.003363759955391288,
-      "learning_rate": 0.0001999996120990312,
-      "loss": 46.0,
-      "step": 11606
-    },
-    {
-      "epoch": 0.8874362062044842,
-      "grad_norm": 0.0019067656248807907,
-      "learning_rate": 0.00019999961203212582,
-      "loss": 46.0,
-      "step": 11607
-    },
-    {
-      "epoch": 0.8875126631878739,
-      "grad_norm": 0.004441200289875269,
-      "learning_rate": 0.00019999961196521464,
-      "loss": 46.0,
-      "step": 11608
-    },
-    {
-      "epoch": 0.8875891201712637,
-      "grad_norm": 0.00021467953047249466,
-      "learning_rate": 0.00019999961189829773,
-      "loss": 46.0,
-      "step": 11609
-    },
-    {
-      "epoch": 0.8876655771546533,
-      "grad_norm": 0.0012116761645302176,
-      "learning_rate": 0.00019999961183137503,
-      "loss": 46.0,
-      "step": 11610
-    },
-    {
-      "epoch": 0.8877420341380431,
-      "grad_norm": 0.002231928054243326,
-      "learning_rate": 0.00019999961176444658,
-      "loss": 46.0,
-      "step": 11611
-    },
-    {
-      "epoch": 0.8878184911214329,
-      "grad_norm": 0.00046282069524750113,
-      "learning_rate": 0.00019999961169751233,
-      "loss": 46.0,
-      "step": 11612
-    },
-    {
-      "epoch": 0.8878949481048225,
-      "grad_norm": 0.0018814190989360213,
-      "learning_rate": 0.00019999961163057233,
-      "loss": 46.0,
-      "step": 11613
-    },
-    {
-      "epoch": 0.8879714050882123,
-      "grad_norm": 0.008047036826610565,
-      "learning_rate": 0.00019999961156362653,
-      "loss": 46.0,
-      "step": 11614
-    },
-    {
-      "epoch": 0.8880478620716019,
-      "grad_norm": 0.001296870643272996,
-      "learning_rate": 0.00019999961149667499,
-      "loss": 46.0,
-      "step": 11615
-    },
-    {
-      "epoch": 0.8881243190549917,
-      "grad_norm": 0.0011320675257593393,
-      "learning_rate": 0.0001999996114297177,
-      "loss": 46.0,
-      "step": 11616
-    },
-    {
-      "epoch": 0.8882007760383814,
-      "grad_norm": 0.0011050697648897767,
-      "learning_rate": 0.0001999996113627546,
-      "loss": 46.0,
-      "step": 11617
-    },
-    {
-      "epoch": 0.8882772330217711,
-      "grad_norm": 0.001434743870049715,
-      "learning_rate": 0.00019999961129578577,
-      "loss": 46.0,
-      "step": 11618
-    },
-    {
-      "epoch": 0.8883536900051608,
-      "grad_norm": 0.0028370614163577557,
-      "learning_rate": 0.00019999961122881113,
-      "loss": 46.0,
-      "step": 11619
-    },
-    {
-      "epoch": 0.8884301469885506,
-      "grad_norm": 0.0010172122856602073,
-      "learning_rate": 0.00019999961116183075,
-      "loss": 46.0,
-      "step": 11620
-    },
-    {
-      "epoch": 0.8885066039719403,
-      "grad_norm": 0.002100787591189146,
-      "learning_rate": 0.0001999996110948446,
-      "loss": 46.0,
-      "step": 11621
-    },
-    {
-      "epoch": 0.88858306095533,
-      "grad_norm": 0.001675075851380825,
-      "learning_rate": 0.00019999961102785266,
-      "loss": 46.0,
-      "step": 11622
-    },
-    {
-      "epoch": 0.8886595179387198,
-      "grad_norm": 0.0018334781052544713,
-      "learning_rate": 0.00019999961096085495,
-      "loss": 46.0,
-      "step": 11623
-    },
-    {
-      "epoch": 0.8887359749221094,
-      "grad_norm": 0.000838214298710227,
-      "learning_rate": 0.00019999961089385148,
-      "loss": 46.0,
-      "step": 11624
-    },
-    {
-      "epoch": 0.8888124319054992,
-      "grad_norm": 0.0004943794338032603,
-      "learning_rate": 0.00019999961082684225,
-      "loss": 46.0,
-      "step": 11625
-    },
-    {
-      "epoch": 0.8888888888888888,
-      "grad_norm": 0.001850284170359373,
-      "learning_rate": 0.00019999961075982726,
-      "loss": 46.0,
-      "step": 11626
-    },
-    {
-      "epoch": 0.8889653458722786,
-      "grad_norm": 0.004019557498395443,
-      "learning_rate": 0.00019999961069280646,
-      "loss": 46.0,
-      "step": 11627
-    },
-    {
-      "epoch": 0.8890418028556684,
-      "grad_norm": 0.004177835304290056,
-      "learning_rate": 0.0001999996106257799,
-      "loss": 46.0,
-      "step": 11628
-    },
-    {
-      "epoch": 0.889118259839058,
-      "grad_norm": 0.002234953921288252,
-      "learning_rate": 0.0001999996105587476,
-      "loss": 46.0,
-      "step": 11629
-    },
-    {
-      "epoch": 0.8891947168224478,
-      "grad_norm": 0.004928039386868477,
-      "learning_rate": 0.0001999996104917095,
-      "loss": 46.0,
-      "step": 11630
-    },
-    {
-      "epoch": 0.8892711738058375,
-      "grad_norm": 0.002008287003263831,
-      "learning_rate": 0.00019999961042466567,
-      "loss": 46.0,
-      "step": 11631
-    },
-    {
-      "epoch": 0.8893476307892272,
-      "grad_norm": 0.0008996453252620995,
-      "learning_rate": 0.00019999961035761603,
-      "loss": 46.0,
-      "step": 11632
-    },
-    {
-      "epoch": 0.889424087772617,
-      "grad_norm": 0.002622775500640273,
-      "learning_rate": 0.00019999961029056062,
-      "loss": 46.0,
-      "step": 11633
-    },
-    {
-      "epoch": 0.8895005447560067,
-      "grad_norm": 0.0016102243680506945,
-      "learning_rate": 0.00019999961022349946,
-      "loss": 46.0,
-      "step": 11634
-    },
-    {
-      "epoch": 0.8895770017393964,
-      "grad_norm": 0.000692696834448725,
-      "learning_rate": 0.00019999961015643253,
-      "loss": 46.0,
-      "step": 11635
-    },
-    {
-      "epoch": 0.8896534587227861,
-      "grad_norm": 0.0012462381273508072,
-      "learning_rate": 0.00019999961008935983,
-      "loss": 46.0,
-      "step": 11636
-    },
-    {
-      "epoch": 0.8897299157061758,
-      "grad_norm": 0.006560949143022299,
-      "learning_rate": 0.00019999961002228135,
-      "loss": 46.0,
-      "step": 11637
-    },
-    {
-      "epoch": 0.8898063726895655,
-      "grad_norm": 0.0011459467932581902,
-      "learning_rate": 0.00019999960995519713,
-      "loss": 46.0,
-      "step": 11638
-    },
-    {
-      "epoch": 0.8898828296729553,
-      "grad_norm": 0.0008364997338503599,
-      "learning_rate": 0.00019999960988810713,
-      "loss": 46.0,
-      "step": 11639
-    },
-    {
-      "epoch": 0.8899592866563449,
-      "grad_norm": 0.0014103270368650556,
-      "learning_rate": 0.00019999960982101134,
-      "loss": 46.0,
-      "step": 11640
-    },
-    {
-      "epoch": 0.8900357436397347,
-      "grad_norm": 0.0010833328124135733,
-      "learning_rate": 0.0001999996097539098,
-      "loss": 46.0,
-      "step": 11641
-    },
-    {
-      "epoch": 0.8901122006231245,
-      "grad_norm": 0.0007472778088413179,
-      "learning_rate": 0.00019999960968680248,
-      "loss": 46.0,
-      "step": 11642
-    },
-    {
-      "epoch": 0.8901886576065141,
-      "grad_norm": 0.0022222402039915323,
-      "learning_rate": 0.00019999960961968939,
-      "loss": 46.0,
-      "step": 11643
-    },
-    {
-      "epoch": 0.8902651145899039,
-      "grad_norm": 0.0004993730690330267,
-      "learning_rate": 0.00019999960955257052,
-      "loss": 46.0,
-      "step": 11644
-    },
-    {
-      "epoch": 0.8903415715732935,
-      "grad_norm": 0.007205368485301733,
-      "learning_rate": 0.00019999960948544588,
-      "loss": 46.0,
-      "step": 11645
-    },
-    {
-      "epoch": 0.8904180285566833,
-      "grad_norm": 0.0006561383488588035,
-      "learning_rate": 0.0001999996094183155,
-      "loss": 46.0,
-      "step": 11646
-    },
-    {
-      "epoch": 0.890494485540073,
-      "grad_norm": 0.005198752973228693,
-      "learning_rate": 0.00019999960935117934,
-      "loss": 46.0,
-      "step": 11647
-    },
-    {
-      "epoch": 0.8905709425234627,
-      "grad_norm": 0.0007756437989883125,
-      "learning_rate": 0.00019999960928403741,
-      "loss": 46.0,
-      "step": 11648
-    },
-    {
-      "epoch": 0.8906473995068525,
-      "grad_norm": 0.004213810432702303,
-      "learning_rate": 0.00019999960921688968,
-      "loss": 46.0,
-      "step": 11649
-    },
-    {
-      "epoch": 0.8907238564902422,
-      "grad_norm": 0.0007201172411441803,
-      "learning_rate": 0.0001999996091497362,
-      "loss": 46.0,
-      "step": 11650
-    },
-    {
-      "epoch": 0.8908003134736319,
-      "grad_norm": 0.0017625095788389444,
-      "learning_rate": 0.00019999960908257698,
-      "loss": 46.0,
-      "step": 11651
-    },
-    {
-      "epoch": 0.8908767704570216,
-      "grad_norm": 0.001020937692373991,
-      "learning_rate": 0.00019999960901541196,
-      "loss": 46.0,
-      "step": 11652
-    },
-    {
-      "epoch": 0.8909532274404114,
-      "grad_norm": 0.0006207632832229137,
-      "learning_rate": 0.00019999960894824116,
-      "loss": 46.0,
-      "step": 11653
-    },
-    {
-      "epoch": 0.891029684423801,
-      "grad_norm": 0.011001263745129108,
-      "learning_rate": 0.00019999960888106465,
-      "loss": 46.0,
-      "step": 11654
-    },
-    {
-      "epoch": 0.8911061414071908,
-      "grad_norm": 0.0006805288139730692,
-      "learning_rate": 0.00019999960881388228,
-      "loss": 46.0,
-      "step": 11655
-    },
-    {
-      "epoch": 0.8911825983905804,
-      "grad_norm": 0.0016909033292904496,
-      "learning_rate": 0.00019999960874669421,
-      "loss": 46.0,
-      "step": 11656
-    },
-    {
-      "epoch": 0.8912590553739702,
-      "grad_norm": 0.0006358883692882955,
-      "learning_rate": 0.00019999960867950035,
-      "loss": 46.0,
-      "step": 11657
-    },
-    {
-      "epoch": 0.89133551235736,
-      "grad_norm": 0.0017439447110518813,
-      "learning_rate": 0.00019999960861230074,
-      "loss": 46.0,
-      "step": 11658
-    },
-    {
-      "epoch": 0.8914119693407496,
-      "grad_norm": 0.0032483956310898066,
-      "learning_rate": 0.00019999960854509533,
-      "loss": 46.0,
-      "step": 11659
-    },
-    {
-      "epoch": 0.8914884263241394,
-      "grad_norm": 0.0015791610348969698,
-      "learning_rate": 0.00019999960847788417,
-      "loss": 46.0,
-      "step": 11660
-    },
-    {
-      "epoch": 0.8915648833075291,
-      "grad_norm": 0.0009216535254381597,
-      "learning_rate": 0.00019999960841066722,
-      "loss": 46.0,
-      "step": 11661
-    },
-    {
-      "epoch": 0.8916413402909188,
-      "grad_norm": 0.0004906052490696311,
-      "learning_rate": 0.00019999960834344451,
-      "loss": 46.0,
-      "step": 11662
-    },
-    {
-      "epoch": 0.8917177972743086,
-      "grad_norm": 0.0006257439381442964,
-      "learning_rate": 0.000199999608276216,
-      "loss": 46.0,
-      "step": 11663
-    },
-    {
-      "epoch": 0.8917942542576983,
-      "grad_norm": 0.006156679708510637,
-      "learning_rate": 0.00019999960820898182,
-      "loss": 46.0,
-      "step": 11664
-    },
-    {
-      "epoch": 0.891870711241088,
-      "grad_norm": 0.001959191868081689,
-      "learning_rate": 0.00019999960814174177,
-      "loss": 46.0,
-      "step": 11665
-    },
-    {
-      "epoch": 0.8919471682244777,
-      "grad_norm": 0.0006653974414803088,
-      "learning_rate": 0.000199999608074496,
-      "loss": 46.0,
-      "step": 11666
-    },
-    {
-      "epoch": 0.8920236252078674,
-      "grad_norm": 0.0007571681635454297,
-      "learning_rate": 0.00019999960800724443,
-      "loss": 46.0,
-      "step": 11667
-    },
-    {
-      "epoch": 0.8921000821912571,
-      "grad_norm": 0.000815580424387008,
-      "learning_rate": 0.0001999996079399871,
-      "loss": 46.0,
-      "step": 11668
-    },
-    {
-      "epoch": 0.8921765391746469,
-      "grad_norm": 0.0005262845079414546,
-      "learning_rate": 0.00019999960787272402,
-      "loss": 46.0,
-      "step": 11669
-    },
-    {
-      "epoch": 0.8922529961580365,
-      "grad_norm": 0.0008712175185792148,
-      "learning_rate": 0.00019999960780545516,
-      "loss": 46.0,
-      "step": 11670
-    },
-    {
-      "epoch": 0.8923294531414263,
-      "grad_norm": 0.0016177580691874027,
-      "learning_rate": 0.00019999960773818052,
-      "loss": 46.0,
-      "step": 11671
-    },
-    {
-      "epoch": 0.8924059101248161,
-      "grad_norm": 0.005622413009405136,
-      "learning_rate": 0.0001999996076709001,
-      "loss": 46.0,
-      "step": 11672
-    },
-    {
-      "epoch": 0.8924823671082057,
-      "grad_norm": 0.0005647125071845949,
-      "learning_rate": 0.00019999960760361395,
-      "loss": 46.0,
-      "step": 11673
-    },
-    {
-      "epoch": 0.8925588240915955,
-      "grad_norm": 0.0008469205931760371,
-      "learning_rate": 0.000199999607536322,
-      "loss": 46.0,
-      "step": 11674
-    },
-    {
-      "epoch": 0.8926352810749851,
-      "grad_norm": 0.001232604612596333,
-      "learning_rate": 0.0001999996074690243,
-      "loss": 46.0,
-      "step": 11675
-    },
-    {
-      "epoch": 0.8927117380583749,
-      "grad_norm": 0.0005928436294198036,
-      "learning_rate": 0.00019999960740172082,
-      "loss": 46.0,
-      "step": 11676
-    },
-    {
-      "epoch": 0.8927881950417647,
-      "grad_norm": 0.0054781511425971985,
-      "learning_rate": 0.00019999960733441157,
-      "loss": 46.0,
-      "step": 11677
-    },
-    {
-      "epoch": 0.8928646520251543,
-      "grad_norm": 0.0027831760235130787,
-      "learning_rate": 0.00019999960726709655,
-      "loss": 46.0,
-      "step": 11678
-    },
-    {
-      "epoch": 0.8929411090085441,
-      "grad_norm": 0.0008409085567109287,
-      "learning_rate": 0.00019999960719977575,
-      "loss": 46.0,
-      "step": 11679
-    },
-    {
-      "epoch": 0.8930175659919338,
-      "grad_norm": 0.0005970362108200788,
-      "learning_rate": 0.0001999996071324492,
-      "loss": 46.0,
-      "step": 11680
-    },
-    {
-      "epoch": 0.8930940229753235,
-      "grad_norm": 0.0013522456865757704,
-      "learning_rate": 0.0001999996070651169,
-      "loss": 46.0,
-      "step": 11681
-    },
-    {
-      "epoch": 0.8931704799587132,
-      "grad_norm": 0.002104173880070448,
-      "learning_rate": 0.00019999960699777878,
-      "loss": 46.0,
-      "step": 11682
-    },
-    {
-      "epoch": 0.893246936942103,
-      "grad_norm": 0.0014125931775197387,
-      "learning_rate": 0.0001999996069304349,
-      "loss": 46.0,
-      "step": 11683
-    },
-    {
-      "epoch": 0.8933233939254926,
-      "grad_norm": 0.0009136369335465133,
-      "learning_rate": 0.00019999960686308528,
-      "loss": 46.0,
-      "step": 11684
-    },
-    {
-      "epoch": 0.8933998509088824,
-      "grad_norm": 0.0018417006358504295,
-      "learning_rate": 0.0001999996067957299,
-      "loss": 46.0,
-      "step": 11685
-    },
-    {
-      "epoch": 0.8934763078922721,
-      "grad_norm": 0.0033245659433305264,
-      "learning_rate": 0.0001999996067283687,
-      "loss": 46.0,
-      "step": 11686
-    },
-    {
-      "epoch": 0.8935527648756618,
-      "grad_norm": 0.0010629085591062903,
-      "learning_rate": 0.00019999960666100176,
-      "loss": 46.0,
-      "step": 11687
-    },
-    {
-      "epoch": 0.8936292218590516,
-      "grad_norm": 0.0011256822617724538,
-      "learning_rate": 0.00019999960659362905,
-      "loss": 46.0,
-      "step": 11688
-    },
-    {
-      "epoch": 0.8937056788424412,
-      "grad_norm": 0.0003571473353076726,
-      "learning_rate": 0.00019999960652625055,
-      "loss": 46.0,
-      "step": 11689
-    },
-    {
-      "epoch": 0.893782135825831,
-      "grad_norm": 0.002232632366940379,
-      "learning_rate": 0.00019999960645886633,
-      "loss": 46.0,
-      "step": 11690
-    },
-    {
-      "epoch": 0.8938585928092208,
-      "grad_norm": 0.003224589629098773,
-      "learning_rate": 0.00019999960639147628,
-      "loss": 46.0,
-      "step": 11691
-    },
-    {
-      "epoch": 0.8939350497926104,
-      "grad_norm": 0.0012783812126144767,
-      "learning_rate": 0.0001999996063240805,
-      "loss": 46.0,
-      "step": 11692
-    },
-    {
-      "epoch": 0.8940115067760002,
-      "grad_norm": 0.003401727182790637,
-      "learning_rate": 0.00019999960625667894,
-      "loss": 46.0,
-      "step": 11693
-    },
-    {
-      "epoch": 0.8940879637593899,
-      "grad_norm": 0.0010558475041761994,
-      "learning_rate": 0.00019999960618927162,
-      "loss": 46.0,
-      "step": 11694
-    },
-    {
-      "epoch": 0.8941644207427796,
-      "grad_norm": 0.0017676184652373195,
-      "learning_rate": 0.00019999960612185854,
-      "loss": 46.0,
-      "step": 11695
-    },
-    {
-      "epoch": 0.8942408777261693,
-      "grad_norm": 0.005570777226239443,
-      "learning_rate": 0.00019999960605443965,
-      "loss": 46.0,
-      "step": 11696
-    },
-    {
-      "epoch": 0.894317334709559,
-      "grad_norm": 0.005142435431480408,
-      "learning_rate": 0.000199999605987015,
-      "loss": 46.0,
-      "step": 11697
-    },
-    {
-      "epoch": 0.8943937916929487,
-      "grad_norm": 0.0008430549642071128,
-      "learning_rate": 0.0001999996059195846,
-      "loss": 46.0,
-      "step": 11698
-    },
-    {
-      "epoch": 0.8944702486763385,
-      "grad_norm": 0.0010345809860154986,
-      "learning_rate": 0.00019999960585214842,
-      "loss": 46.0,
-      "step": 11699
-    },
-    {
-      "epoch": 0.8945467056597282,
-      "grad_norm": 0.0023996320087462664,
-      "learning_rate": 0.0001999996057847065,
-      "loss": 46.0,
-      "step": 11700
-    },
-    {
-      "epoch": 0.8946231626431179,
-      "grad_norm": 0.008272090926766396,
-      "learning_rate": 0.00019999960571725876,
-      "loss": 46.0,
-      "step": 11701
-    },
-    {
-      "epoch": 0.8946996196265077,
-      "grad_norm": 0.0006001748261041939,
-      "learning_rate": 0.00019999960564980529,
-      "loss": 46.0,
-      "step": 11702
-    },
-    {
-      "epoch": 0.8947760766098973,
-      "grad_norm": 0.0004926119581796229,
-      "learning_rate": 0.00019999960558234604,
-      "loss": 46.0,
-      "step": 11703
-    },
-    {
-      "epoch": 0.8948525335932871,
-      "grad_norm": 0.00027394964126870036,
-      "learning_rate": 0.000199999605514881,
-      "loss": 46.0,
-      "step": 11704
-    },
-    {
-      "epoch": 0.8949289905766769,
-      "grad_norm": 0.001660117064602673,
-      "learning_rate": 0.00019999960544741022,
-      "loss": 46.0,
-      "step": 11705
-    },
-    {
-      "epoch": 0.8950054475600665,
-      "grad_norm": 0.009312672540545464,
-      "learning_rate": 0.00019999960537993365,
-      "loss": 46.0,
-      "step": 11706
-    },
-    {
-      "epoch": 0.8950819045434563,
-      "grad_norm": 0.001353538129478693,
-      "learning_rate": 0.00019999960531245133,
-      "loss": 46.0,
-      "step": 11707
-    },
-    {
-      "epoch": 0.8951583615268459,
-      "grad_norm": 0.00027032053912989795,
-      "learning_rate": 0.00019999960524496322,
-      "loss": 46.0,
-      "step": 11708
-    },
-    {
-      "epoch": 0.8952348185102357,
-      "grad_norm": 0.002126175444573164,
-      "learning_rate": 0.00019999960517746933,
-      "loss": 46.0,
-      "step": 11709
-    },
-    {
-      "epoch": 0.8953112754936254,
-      "grad_norm": 0.0014371026773005724,
-      "learning_rate": 0.0001999996051099697,
-      "loss": 46.0,
-      "step": 11710
-    },
-    {
-      "epoch": 0.8953877324770151,
-      "grad_norm": 0.002477489411830902,
-      "learning_rate": 0.00019999960504246428,
-      "loss": 46.0,
-      "step": 11711
-    },
-    {
-      "epoch": 0.8954641894604048,
-      "grad_norm": 0.0011744264047592878,
-      "learning_rate": 0.0001999996049749531,
-      "loss": 46.0,
-      "step": 11712
-    },
-    {
-      "epoch": 0.8955406464437946,
-      "grad_norm": 0.001536023337393999,
-      "learning_rate": 0.00019999960490743617,
-      "loss": 46.0,
-      "step": 11713
-    },
-    {
-      "epoch": 0.8956171034271843,
-      "grad_norm": 0.001650497317314148,
-      "learning_rate": 0.00019999960483991345,
-      "loss": 46.0,
-      "step": 11714
-    },
-    {
-      "epoch": 0.895693560410574,
-      "grad_norm": 0.001094701001420617,
-      "learning_rate": 0.00019999960477238497,
-      "loss": 46.0,
-      "step": 11715
-    },
-    {
-      "epoch": 0.8957700173939637,
-      "grad_norm": 0.0008322000503540039,
-      "learning_rate": 0.0001999996047048507,
-      "loss": 46.0,
-      "step": 11716
-    },
-    {
-      "epoch": 0.8958464743773534,
-      "grad_norm": 0.007792453747242689,
-      "learning_rate": 0.00019999960463731067,
-      "loss": 46.0,
-      "step": 11717
-    },
-    {
-      "epoch": 0.8959229313607432,
-      "grad_norm": 0.0010417783632874489,
-      "learning_rate": 0.00019999960456976488,
-      "loss": 46.0,
-      "step": 11718
-    },
-    {
-      "epoch": 0.8959993883441328,
-      "grad_norm": 0.008582200855016708,
-      "learning_rate": 0.0001999996045022133,
-      "loss": 46.0,
-      "step": 11719
-    },
-    {
-      "epoch": 0.8960758453275226,
-      "grad_norm": 0.0038891613949090242,
-      "learning_rate": 0.00019999960443465597,
-      "loss": 46.0,
-      "step": 11720
-    },
-    {
-      "epoch": 0.8961523023109124,
-      "grad_norm": 0.0007805288769304752,
-      "learning_rate": 0.00019999960436709285,
-      "loss": 46.0,
-      "step": 11721
-    },
-    {
-      "epoch": 0.896228759294302,
-      "grad_norm": 0.00013388553634285927,
-      "learning_rate": 0.000199999604299524,
-      "loss": 46.0,
-      "step": 11722
-    },
-    {
-      "epoch": 0.8963052162776918,
-      "grad_norm": 0.013632994145154953,
-      "learning_rate": 0.00019999960423194935,
-      "loss": 46.0,
-      "step": 11723
-    },
-    {
-      "epoch": 0.8963816732610815,
-      "grad_norm": 0.0011647128267213702,
-      "learning_rate": 0.00019999960416436895,
-      "loss": 46.0,
-      "step": 11724
-    },
-    {
-      "epoch": 0.8964581302444712,
-      "grad_norm": 0.006610801909118891,
-      "learning_rate": 0.0001999996040967828,
-      "loss": 46.0,
-      "step": 11725
-    },
-    {
-      "epoch": 0.896534587227861,
-      "grad_norm": 0.002584711881354451,
-      "learning_rate": 0.0001999996040291908,
-      "loss": 46.0,
-      "step": 11726
-    },
-    {
-      "epoch": 0.8966110442112506,
-      "grad_norm": 0.0007482058135792613,
-      "learning_rate": 0.0001999996039615931,
-      "loss": 46.0,
-      "step": 11727
-    },
-    {
-      "epoch": 0.8966875011946404,
-      "grad_norm": 0.00039417415973730385,
-      "learning_rate": 0.0001999996038939896,
-      "loss": 46.0,
-      "step": 11728
-    },
-    {
-      "epoch": 0.8967639581780301,
-      "grad_norm": 0.0006242976523935795,
-      "learning_rate": 0.00019999960382638036,
-      "loss": 46.0,
-      "step": 11729
-    },
-    {
-      "epoch": 0.8968404151614198,
-      "grad_norm": 0.0009961030445992947,
-      "learning_rate": 0.0001999996037587653,
-      "loss": 46.0,
-      "step": 11730
-    },
-    {
-      "epoch": 0.8969168721448095,
-      "grad_norm": 0.001067009405232966,
-      "learning_rate": 0.0001999996036911445,
-      "loss": 46.0,
-      "step": 11731
-    },
-    {
-      "epoch": 0.8969933291281993,
-      "grad_norm": 0.008063516579568386,
-      "learning_rate": 0.00019999960362351794,
-      "loss": 46.0,
-      "step": 11732
-    },
-    {
-      "epoch": 0.8970697861115889,
-      "grad_norm": 0.004624022636562586,
-      "learning_rate": 0.00019999960355588563,
-      "loss": 46.0,
-      "step": 11733
-    },
-    {
-      "epoch": 0.8971462430949787,
-      "grad_norm": 0.0015413794899359345,
-      "learning_rate": 0.00019999960348824751,
-      "loss": 46.0,
-      "step": 11734
-    },
-    {
-      "epoch": 0.8972227000783685,
-      "grad_norm": 0.0008905536960810423,
-      "learning_rate": 0.00019999960342060363,
-      "loss": 46.0,
-      "step": 11735
-    },
-    {
-      "epoch": 0.8972991570617581,
-      "grad_norm": 0.0008385450346395373,
-      "learning_rate": 0.000199999603352954,
-      "loss": 46.0,
-      "step": 11736
-    },
-    {
-      "epoch": 0.8973756140451479,
-      "grad_norm": 0.0007168976007960737,
-      "learning_rate": 0.00019999960328529856,
-      "loss": 46.0,
-      "step": 11737
-    },
-    {
-      "epoch": 0.8974520710285375,
-      "grad_norm": 0.002199511043727398,
-      "learning_rate": 0.0001999996032176374,
-      "loss": 46.0,
-      "step": 11738
-    },
-    {
-      "epoch": 0.8975285280119273,
-      "grad_norm": 0.0013263440923765302,
-      "learning_rate": 0.00019999960314997045,
-      "loss": 46.0,
-      "step": 11739
-    },
-    {
-      "epoch": 0.897604984995317,
-      "grad_norm": 0.0005957005196250975,
-      "learning_rate": 0.0001999996030822977,
-      "loss": 46.0,
-      "step": 11740
-    },
-    {
-      "epoch": 0.8976814419787067,
-      "grad_norm": 0.0005969987832941115,
-      "learning_rate": 0.00019999960301461922,
-      "loss": 46.0,
-      "step": 11741
-    },
-    {
-      "epoch": 0.8977578989620965,
-      "grad_norm": 0.0008814689936116338,
-      "learning_rate": 0.00019999960294693497,
-      "loss": 46.0,
-      "step": 11742
-    },
-    {
-      "epoch": 0.8978343559454862,
-      "grad_norm": 0.0024248005356639624,
-      "learning_rate": 0.00019999960287924492,
-      "loss": 46.0,
-      "step": 11743
-    },
-    {
-      "epoch": 0.8979108129288759,
-      "grad_norm": 0.020057395100593567,
-      "learning_rate": 0.00019999960281154913,
-      "loss": 46.0,
-      "step": 11744
-    },
-    {
-      "epoch": 0.8979872699122656,
-      "grad_norm": 0.00041233684169128537,
-      "learning_rate": 0.00019999960274384756,
-      "loss": 46.0,
-      "step": 11745
-    },
-    {
-      "epoch": 0.8980637268956553,
-      "grad_norm": 0.0044325231574475765,
-      "learning_rate": 0.00019999960267614022,
-      "loss": 46.0,
-      "step": 11746
-    },
-    {
-      "epoch": 0.898140183879045,
-      "grad_norm": 0.0007187029696069658,
-      "learning_rate": 0.00019999960260842713,
-      "loss": 46.0,
-      "step": 11747
-    },
-    {
-      "epoch": 0.8982166408624348,
-      "grad_norm": 0.004973837174475193,
-      "learning_rate": 0.0001999996025407082,
-      "loss": 46.0,
-      "step": 11748
-    },
-    {
-      "epoch": 0.8982930978458245,
-      "grad_norm": 0.0007522447849623859,
-      "learning_rate": 0.0001999996024729836,
-      "loss": 46.0,
-      "step": 11749
-    },
-    {
-      "epoch": 0.8983695548292142,
-      "grad_norm": 0.005311708431690931,
-      "learning_rate": 0.00019999960240525314,
-      "loss": 46.0,
-      "step": 11750
-    },
-    {
-      "epoch": 0.898446011812604,
-      "grad_norm": 0.0008693598210811615,
-      "learning_rate": 0.000199999602337517,
-      "loss": 46.0,
-      "step": 11751
-    },
-    {
-      "epoch": 0.8985224687959936,
-      "grad_norm": 0.0028668921440839767,
-      "learning_rate": 0.00019999960226977503,
-      "loss": 46.0,
-      "step": 11752
-    },
-    {
-      "epoch": 0.8985989257793834,
-      "grad_norm": 0.002102965721860528,
-      "learning_rate": 0.00019999960220202728,
-      "loss": 46.0,
-      "step": 11753
-    },
-    {
-      "epoch": 0.8986753827627731,
-      "grad_norm": 0.0024905819445848465,
-      "learning_rate": 0.0001999996021342738,
-      "loss": 46.0,
-      "step": 11754
-    },
-    {
-      "epoch": 0.8987518397461628,
-      "grad_norm": 0.0023743496276438236,
-      "learning_rate": 0.00019999960206651453,
-      "loss": 46.0,
-      "step": 11755
-    },
-    {
-      "epoch": 0.8988282967295526,
-      "grad_norm": 0.0034692531917244196,
-      "learning_rate": 0.0001999996019987495,
-      "loss": 46.0,
-      "step": 11756
-    },
-    {
-      "epoch": 0.8989047537129422,
-      "grad_norm": 0.0008326807292178273,
-      "learning_rate": 0.00019999960193097871,
-      "loss": 46.0,
-      "step": 11757
-    },
-    {
-      "epoch": 0.898981210696332,
-      "grad_norm": 0.0012438230914995074,
-      "learning_rate": 0.00019999960186320212,
-      "loss": 46.0,
-      "step": 11758
-    },
-    {
-      "epoch": 0.8990576676797217,
-      "grad_norm": 0.0006790123879909515,
-      "learning_rate": 0.00019999960179541978,
-      "loss": 46.0,
-      "step": 11759
-    },
-    {
-      "epoch": 0.8991341246631114,
-      "grad_norm": 0.0044629559852182865,
-      "learning_rate": 0.0001999996017276317,
-      "loss": 46.0,
-      "step": 11760
-    },
-    {
-      "epoch": 0.8992105816465011,
-      "grad_norm": 0.0008896850631572306,
-      "learning_rate": 0.00019999960165983778,
-      "loss": 46.0,
-      "step": 11761
-    },
-    {
-      "epoch": 0.8992870386298909,
-      "grad_norm": 0.0016496492316946387,
-      "learning_rate": 0.00019999960159203814,
-      "loss": 46.0,
-      "step": 11762
-    },
-    {
-      "epoch": 0.8993634956132806,
-      "grad_norm": 0.0018810631008818746,
-      "learning_rate": 0.0001999996015242327,
-      "loss": 46.0,
-      "step": 11763
-    },
-    {
-      "epoch": 0.8994399525966703,
-      "grad_norm": 0.002070453716441989,
-      "learning_rate": 0.00019999960145642153,
-      "loss": 46.0,
-      "step": 11764
-    },
-    {
-      "epoch": 0.8995164095800601,
-      "grad_norm": 0.0038174346555024385,
-      "learning_rate": 0.00019999960138860457,
-      "loss": 46.0,
-      "step": 11765
-    },
-    {
-      "epoch": 0.8995928665634497,
-      "grad_norm": 0.0037040775641798973,
-      "learning_rate": 0.00019999960132078184,
-      "loss": 46.0,
-      "step": 11766
-    },
-    {
-      "epoch": 0.8996693235468395,
-      "grad_norm": 0.0005760633503086865,
-      "learning_rate": 0.00019999960125295334,
-      "loss": 46.0,
-      "step": 11767
-    },
-    {
-      "epoch": 0.8997457805302291,
-      "grad_norm": 0.001522030564956367,
-      "learning_rate": 0.00019999960118511907,
-      "loss": 46.0,
-      "step": 11768
-    },
-    {
-      "epoch": 0.8998222375136189,
-      "grad_norm": 0.0015499969013035297,
-      "learning_rate": 0.00019999960111727905,
-      "loss": 46.0,
-      "step": 11769
-    },
-    {
-      "epoch": 0.8998986944970087,
-      "grad_norm": 0.0011950712651014328,
-      "learning_rate": 0.00019999960104943325,
-      "loss": 46.0,
-      "step": 11770
-    },
-    {
-      "epoch": 0.8999751514803983,
-      "grad_norm": 0.0011982048163190484,
-      "learning_rate": 0.00019999960098158166,
-      "loss": 46.0,
-      "step": 11771
-    },
-    {
-      "epoch": 0.9000516084637881,
-      "grad_norm": 0.0005346702528186142,
-      "learning_rate": 0.00019999960091372432,
-      "loss": 46.0,
-      "step": 11772
-    },
-    {
-      "epoch": 0.9001280654471778,
-      "grad_norm": 0.0011791560100391507,
-      "learning_rate": 0.0001999996008458612,
-      "loss": 46.0,
-      "step": 11773
-    },
-    {
-      "epoch": 0.9002045224305675,
-      "grad_norm": 0.002212457824498415,
-      "learning_rate": 0.00019999960077799232,
-      "loss": 46.0,
-      "step": 11774
-    },
-    {
-      "epoch": 0.9002809794139572,
-      "grad_norm": 0.0005708799581043422,
-      "learning_rate": 0.00019999960071011766,
-      "loss": 46.0,
-      "step": 11775
-    },
-    {
-      "epoch": 0.9003574363973469,
-      "grad_norm": 0.0013216623337939382,
-      "learning_rate": 0.00019999960064223725,
-      "loss": 46.0,
-      "step": 11776
-    },
-    {
-      "epoch": 0.9004338933807366,
-      "grad_norm": 0.0006090451497584581,
-      "learning_rate": 0.00019999960057435104,
-      "loss": 46.0,
-      "step": 11777
-    },
-    {
-      "epoch": 0.9005103503641264,
-      "grad_norm": 0.0013295685639604926,
-      "learning_rate": 0.0001999996005064591,
-      "loss": 46.0,
-      "step": 11778
-    },
-    {
-      "epoch": 0.9005868073475161,
-      "grad_norm": 0.0014774148585274816,
-      "learning_rate": 0.00019999960043856136,
-      "loss": 46.0,
-      "step": 11779
-    },
-    {
-      "epoch": 0.9006632643309058,
-      "grad_norm": 0.00426880456507206,
-      "learning_rate": 0.00019999960037065786,
-      "loss": 46.0,
-      "step": 11780
-    },
-    {
-      "epoch": 0.9007397213142956,
-      "grad_norm": 0.0018135968130081892,
-      "learning_rate": 0.00019999960030274858,
-      "loss": 46.0,
-      "step": 11781
-    },
-    {
-      "epoch": 0.9008161782976852,
-      "grad_norm": 0.0010952582815662026,
-      "learning_rate": 0.00019999960023483354,
-      "loss": 46.0,
-      "step": 11782
-    },
-    {
-      "epoch": 0.900892635281075,
-      "grad_norm": 0.0005496160010807216,
-      "learning_rate": 0.00019999960016691274,
-      "loss": 46.0,
-      "step": 11783
-    },
-    {
-      "epoch": 0.9009690922644648,
-      "grad_norm": 0.015186472795903683,
-      "learning_rate": 0.00019999960009898615,
-      "loss": 46.0,
-      "step": 11784
-    },
-    {
-      "epoch": 0.9010455492478544,
-      "grad_norm": 0.001273954170756042,
-      "learning_rate": 0.00019999960003105384,
-      "loss": 46.0,
-      "step": 11785
-    },
-    {
-      "epoch": 0.9011220062312442,
-      "grad_norm": 0.0008249178063124418,
-      "learning_rate": 0.00019999959996311567,
-      "loss": 46.0,
-      "step": 11786
-    },
-    {
-      "epoch": 0.9011984632146338,
-      "grad_norm": 0.00112240226007998,
-      "learning_rate": 0.0001999995998951718,
-      "loss": 46.0,
-      "step": 11787
-    },
-    {
-      "epoch": 0.9012749201980236,
-      "grad_norm": 0.0011776859173551202,
-      "learning_rate": 0.00019999959982722217,
-      "loss": 46.0,
-      "step": 11788
-    },
-    {
-      "epoch": 0.9013513771814133,
-      "grad_norm": 0.0015126215294003487,
-      "learning_rate": 0.00019999959975926674,
-      "loss": 46.0,
-      "step": 11789
-    },
-    {
-      "epoch": 0.901427834164803,
-      "grad_norm": 0.001288728672079742,
-      "learning_rate": 0.00019999959969130556,
-      "loss": 46.0,
-      "step": 11790
-    },
-    {
-      "epoch": 0.9015042911481927,
-      "grad_norm": 0.0008634102996438742,
-      "learning_rate": 0.00019999959962333858,
-      "loss": 46.0,
-      "step": 11791
-    },
-    {
-      "epoch": 0.9015807481315825,
-      "grad_norm": 0.004667697940021753,
-      "learning_rate": 0.00019999959955536585,
-      "loss": 46.0,
-      "step": 11792
-    },
-    {
-      "epoch": 0.9016572051149722,
-      "grad_norm": 0.003966478165239096,
-      "learning_rate": 0.00019999959948738732,
-      "loss": 46.0,
-      "step": 11793
-    },
-    {
-      "epoch": 0.9017336620983619,
-      "grad_norm": 0.003953876905143261,
-      "learning_rate": 0.00019999959941940305,
-      "loss": 46.0,
-      "step": 11794
-    },
-    {
-      "epoch": 0.9018101190817517,
-      "grad_norm": 0.0011055568465963006,
-      "learning_rate": 0.000199999599351413,
-      "loss": 46.0,
-      "step": 11795
-    },
-    {
-      "epoch": 0.9018865760651413,
-      "grad_norm": 0.0010162289254367352,
-      "learning_rate": 0.0001999995992834172,
-      "loss": 46.0,
-      "step": 11796
-    },
-    {
-      "epoch": 0.9019630330485311,
-      "grad_norm": 0.000930530484765768,
-      "learning_rate": 0.00019999959921541562,
-      "loss": 46.0,
-      "step": 11797
-    },
-    {
-      "epoch": 0.9020394900319207,
-      "grad_norm": 0.0034277592785656452,
-      "learning_rate": 0.00019999959914740828,
-      "loss": 46.0,
-      "step": 11798
-    },
-    {
-      "epoch": 0.9021159470153105,
-      "grad_norm": 0.0017827311530709267,
-      "learning_rate": 0.00019999959907939516,
-      "loss": 46.0,
-      "step": 11799
-    },
-    {
-      "epoch": 0.9021924039987003,
-      "grad_norm": 0.0007436890155076981,
-      "learning_rate": 0.00019999959901137628,
-      "loss": 46.0,
-      "step": 11800
-    },
-    {
-      "epoch": 0.9022688609820899,
-      "grad_norm": 0.0025699662510305643,
-      "learning_rate": 0.00019999959894335161,
-      "loss": 46.0,
-      "step": 11801
-    },
-    {
-      "epoch": 0.9023453179654797,
-      "grad_norm": 0.0004999426892027259,
-      "learning_rate": 0.00019999959887532118,
-      "loss": 46.0,
-      "step": 11802
-    },
-    {
-      "epoch": 0.9024217749488694,
-      "grad_norm": 0.001900402596220374,
-      "learning_rate": 0.00019999959880728497,
-      "loss": 46.0,
-      "step": 11803
-    },
-    {
-      "epoch": 0.9024982319322591,
-      "grad_norm": 0.0009961911709979177,
-      "learning_rate": 0.00019999959873924302,
-      "loss": 46.0,
-      "step": 11804
-    },
-    {
-      "epoch": 0.9025746889156488,
-      "grad_norm": 0.002137347124516964,
-      "learning_rate": 0.0001999995986711953,
-      "loss": 46.0,
-      "step": 11805
-    },
-    {
-      "epoch": 0.9026511458990385,
-      "grad_norm": 0.0026213969103991985,
-      "learning_rate": 0.00019999959860314177,
-      "loss": 46.0,
-      "step": 11806
-    },
-    {
-      "epoch": 0.9027276028824283,
-      "grad_norm": 0.001633918029256165,
-      "learning_rate": 0.0001999995985350825,
-      "loss": 46.0,
-      "step": 11807
-    },
-    {
-      "epoch": 0.902804059865818,
-      "grad_norm": 0.0010500723728910089,
-      "learning_rate": 0.00019999959846701745,
-      "loss": 46.0,
-      "step": 11808
-    },
-    {
-      "epoch": 0.9028805168492077,
-      "grad_norm": 0.003920499235391617,
-      "learning_rate": 0.00019999959839894665,
-      "loss": 46.0,
-      "step": 11809
-    },
-    {
-      "epoch": 0.9029569738325974,
-      "grad_norm": 0.002261185785755515,
-      "learning_rate": 0.00019999959833087006,
-      "loss": 46.0,
-      "step": 11810
-    },
-    {
-      "epoch": 0.9030334308159872,
-      "grad_norm": 0.0009962526382878423,
-      "learning_rate": 0.00019999959826278772,
-      "loss": 46.0,
-      "step": 11811
-    },
-    {
-      "epoch": 0.9031098877993768,
-      "grad_norm": 0.004018166568130255,
-      "learning_rate": 0.0001999995981946996,
-      "loss": 46.0,
-      "step": 11812
-    },
-    {
-      "epoch": 0.9031863447827666,
-      "grad_norm": 0.0011788710253313184,
-      "learning_rate": 0.0001999995981266057,
-      "loss": 46.0,
-      "step": 11813
-    },
-    {
-      "epoch": 0.9032628017661564,
-      "grad_norm": 0.0008982756990008056,
-      "learning_rate": 0.00019999959805850603,
-      "loss": 46.0,
-      "step": 11814
-    },
-    {
-      "epoch": 0.903339258749546,
-      "grad_norm": 0.0012094053672626615,
-      "learning_rate": 0.0001999995979904006,
-      "loss": 46.0,
-      "step": 11815
-    },
-    {
-      "epoch": 0.9034157157329358,
-      "grad_norm": 0.0008921209955587983,
-      "learning_rate": 0.00019999959792228942,
-      "loss": 46.0,
-      "step": 11816
-    },
-    {
-      "epoch": 0.9034921727163254,
-      "grad_norm": 0.0032483821269124746,
-      "learning_rate": 0.00019999959785417244,
-      "loss": 46.0,
-      "step": 11817
-    },
-    {
-      "epoch": 0.9035686296997152,
-      "grad_norm": 0.013966198079288006,
-      "learning_rate": 0.00019999959778604969,
-      "loss": 46.0,
-      "step": 11818
-    },
-    {
-      "epoch": 0.903645086683105,
-      "grad_norm": 0.0007771443924866617,
-      "learning_rate": 0.00019999959771792119,
-      "loss": 46.0,
-      "step": 11819
-    },
-    {
-      "epoch": 0.9037215436664946,
-      "grad_norm": 0.0014522760175168514,
-      "learning_rate": 0.0001999995976497869,
-      "loss": 46.0,
-      "step": 11820
-    },
-    {
-      "epoch": 0.9037980006498844,
-      "grad_norm": 0.0007052583387121558,
-      "learning_rate": 0.00019999959758164687,
-      "loss": 46.0,
-      "step": 11821
-    },
-    {
-      "epoch": 0.9038744576332741,
-      "grad_norm": 0.001760429935529828,
-      "learning_rate": 0.00019999959751350105,
-      "loss": 46.0,
-      "step": 11822
-    },
-    {
-      "epoch": 0.9039509146166638,
-      "grad_norm": 0.001570952939800918,
-      "learning_rate": 0.00019999959744534945,
-      "loss": 46.0,
-      "step": 11823
-    },
-    {
-      "epoch": 0.9040273716000535,
-      "grad_norm": 0.0005313638248480856,
-      "learning_rate": 0.0001999995973771921,
-      "loss": 46.0,
-      "step": 11824
-    },
-    {
-      "epoch": 0.9041038285834433,
-      "grad_norm": 0.004057899583131075,
-      "learning_rate": 0.000199999597309029,
-      "loss": 46.0,
-      "step": 11825
-    },
-    {
-      "epoch": 0.9041802855668329,
-      "grad_norm": 0.0005371117731556296,
-      "learning_rate": 0.0001999995972408601,
-      "loss": 46.0,
-      "step": 11826
-    },
-    {
-      "epoch": 0.9042567425502227,
-      "grad_norm": 0.0006020376458764076,
-      "learning_rate": 0.00019999959717268543,
-      "loss": 46.0,
-      "step": 11827
-    },
-    {
-      "epoch": 0.9043331995336124,
-      "grad_norm": 0.004249220713973045,
-      "learning_rate": 0.000199999597104505,
-      "loss": 46.0,
-      "step": 11828
-    },
-    {
-      "epoch": 0.9044096565170021,
-      "grad_norm": 0.0011660712771117687,
-      "learning_rate": 0.0001999995970363188,
-      "loss": 46.0,
-      "step": 11829
-    },
-    {
-      "epoch": 0.9044861135003919,
-      "grad_norm": 0.0050075482577085495,
-      "learning_rate": 0.0001999995969681268,
-      "loss": 46.0,
-      "step": 11830
-    },
-    {
-      "epoch": 0.9045625704837815,
-      "grad_norm": 0.001171441632322967,
-      "learning_rate": 0.00019999959689992908,
-      "loss": 46.0,
-      "step": 11831
-    },
-    {
-      "epoch": 0.9046390274671713,
-      "grad_norm": 0.0019744911696761847,
-      "learning_rate": 0.00019999959683172556,
-      "loss": 46.0,
-      "step": 11832
-    },
-    {
-      "epoch": 0.904715484450561,
-      "grad_norm": 0.0023193301167339087,
-      "learning_rate": 0.00019999959676351628,
-      "loss": 46.0,
-      "step": 11833
-    },
-    {
-      "epoch": 0.9047919414339507,
-      "grad_norm": 0.0021487916819751263,
-      "learning_rate": 0.0001999995966953012,
-      "loss": 46.0,
-      "step": 11834
-    },
-    {
-      "epoch": 0.9048683984173405,
-      "grad_norm": 0.0007565730484202504,
-      "learning_rate": 0.00019999959662708042,
-      "loss": 46.0,
-      "step": 11835
-    },
-    {
-      "epoch": 0.9049448554007302,
-      "grad_norm": 0.0021633936557918787,
-      "learning_rate": 0.00019999959655885382,
-      "loss": 46.0,
-      "step": 11836
-    },
-    {
-      "epoch": 0.9050213123841199,
-      "grad_norm": 0.0017209731740877032,
-      "learning_rate": 0.00019999959649062148,
-      "loss": 46.0,
-      "step": 11837
-    },
-    {
-      "epoch": 0.9050977693675096,
-      "grad_norm": 0.007892737165093422,
-      "learning_rate": 0.00019999959642238334,
-      "loss": 46.0,
-      "step": 11838
-    },
-    {
-      "epoch": 0.9051742263508993,
-      "grad_norm": 0.0010190209141001105,
-      "learning_rate": 0.00019999959635413946,
-      "loss": 46.0,
-      "step": 11839
-    },
-    {
-      "epoch": 0.905250683334289,
-      "grad_norm": 0.0011227353243157268,
-      "learning_rate": 0.00019999959628588977,
-      "loss": 46.0,
-      "step": 11840
-    },
-    {
-      "epoch": 0.9053271403176788,
-      "grad_norm": 0.0023314955178648233,
-      "learning_rate": 0.00019999959621763434,
-      "loss": 46.0,
-      "step": 11841
-    },
-    {
-      "epoch": 0.9054035973010685,
-      "grad_norm": 0.0015706990379840136,
-      "learning_rate": 0.00019999959614937313,
-      "loss": 46.0,
-      "step": 11842
-    },
-    {
-      "epoch": 0.9054800542844582,
-      "grad_norm": 0.0009050035150721669,
-      "learning_rate": 0.00019999959608110615,
-      "loss": 46.0,
-      "step": 11843
-    },
-    {
-      "epoch": 0.905556511267848,
-      "grad_norm": 0.0073762615211308,
-      "learning_rate": 0.00019999959601283343,
-      "loss": 46.0,
-      "step": 11844
-    },
-    {
-      "epoch": 0.9056329682512376,
-      "grad_norm": 0.0005101165734231472,
-      "learning_rate": 0.0001999995959445549,
-      "loss": 46.0,
-      "step": 11845
-    },
-    {
-      "epoch": 0.9057094252346274,
-      "grad_norm": 0.0012179447803646326,
-      "learning_rate": 0.00019999959587627063,
-      "loss": 46.0,
-      "step": 11846
-    },
-    {
-      "epoch": 0.905785882218017,
-      "grad_norm": 0.0008587847696617246,
-      "learning_rate": 0.00019999959580798056,
-      "loss": 46.0,
-      "step": 11847
-    },
-    {
-      "epoch": 0.9058623392014068,
-      "grad_norm": 0.0013332751113921404,
-      "learning_rate": 0.00019999959573968474,
-      "loss": 46.0,
-      "step": 11848
-    },
-    {
-      "epoch": 0.9059387961847966,
-      "grad_norm": 0.0009754467755556107,
-      "learning_rate": 0.00019999959567138314,
-      "loss": 46.0,
-      "step": 11849
-    },
-    {
-      "epoch": 0.9060152531681862,
-      "grad_norm": 0.0006352203781716526,
-      "learning_rate": 0.0001999995956030758,
-      "loss": 46.0,
-      "step": 11850
-    },
-    {
-      "epoch": 0.906091710151576,
-      "grad_norm": 0.0004887196118943393,
-      "learning_rate": 0.00019999959553476267,
-      "loss": 46.0,
-      "step": 11851
-    },
-    {
-      "epoch": 0.9061681671349657,
-      "grad_norm": 0.004350871779024601,
-      "learning_rate": 0.00019999959546644375,
-      "loss": 46.0,
-      "step": 11852
-    },
-    {
-      "epoch": 0.9062446241183554,
-      "grad_norm": 0.0017469220329076052,
-      "learning_rate": 0.0001999995953981191,
-      "loss": 46.0,
-      "step": 11853
-    },
-    {
-      "epoch": 0.9063210811017451,
-      "grad_norm": 0.0005943359574303031,
-      "learning_rate": 0.00019999959532978866,
-      "loss": 46.0,
-      "step": 11854
-    },
-    {
-      "epoch": 0.9063975380851349,
-      "grad_norm": 0.004706950858235359,
-      "learning_rate": 0.00019999959526145246,
-      "loss": 46.0,
-      "step": 11855
-    },
-    {
-      "epoch": 0.9064739950685246,
-      "grad_norm": 0.0005976425018161535,
-      "learning_rate": 0.00019999959519311048,
-      "loss": 46.0,
-      "step": 11856
-    },
-    {
-      "epoch": 0.9065504520519143,
-      "grad_norm": 0.001496283570304513,
-      "learning_rate": 0.00019999959512476272,
-      "loss": 46.0,
-      "step": 11857
-    },
-    {
-      "epoch": 0.906626909035304,
-      "grad_norm": 0.002717680297791958,
-      "learning_rate": 0.00019999959505640922,
-      "loss": 46.0,
-      "step": 11858
-    },
-    {
-      "epoch": 0.9067033660186937,
-      "grad_norm": 0.002634023316204548,
-      "learning_rate": 0.00019999959498804995,
-      "loss": 46.0,
-      "step": 11859
-    },
-    {
-      "epoch": 0.9067798230020835,
-      "grad_norm": 0.0014204670442268252,
-      "learning_rate": 0.00019999959491968488,
-      "loss": 46.0,
-      "step": 11860
-    },
-    {
-      "epoch": 0.9068562799854731,
-      "grad_norm": 0.0014687500661239028,
-      "learning_rate": 0.00019999959485131406,
-      "loss": 46.0,
-      "step": 11861
-    },
-    {
-      "epoch": 0.9069327369688629,
-      "grad_norm": 0.006440106779336929,
-      "learning_rate": 0.00019999959478293747,
-      "loss": 46.0,
-      "step": 11862
-    },
-    {
-      "epoch": 0.9070091939522527,
-      "grad_norm": 0.0011966609163209796,
-      "learning_rate": 0.0001999995947145551,
-      "loss": 46.0,
-      "step": 11863
-    },
-    {
-      "epoch": 0.9070856509356423,
-      "grad_norm": 0.0011822740780189633,
-      "learning_rate": 0.00019999959464616696,
-      "loss": 46.0,
-      "step": 11864
-    },
-    {
-      "epoch": 0.9071621079190321,
-      "grad_norm": 0.0030144276097416878,
-      "learning_rate": 0.00019999959457777305,
-      "loss": 46.0,
-      "step": 11865
-    },
-    {
-      "epoch": 0.9072385649024218,
-      "grad_norm": 0.000799511035438627,
-      "learning_rate": 0.00019999959450937337,
-      "loss": 46.0,
-      "step": 11866
-    },
-    {
-      "epoch": 0.9073150218858115,
-      "grad_norm": 0.0014971902128309011,
-      "learning_rate": 0.00019999959444096793,
-      "loss": 46.0,
-      "step": 11867
-    },
-    {
-      "epoch": 0.9073914788692012,
-      "grad_norm": 0.0023763212375342846,
-      "learning_rate": 0.00019999959437255673,
-      "loss": 46.0,
-      "step": 11868
-    },
-    {
-      "epoch": 0.9074679358525909,
-      "grad_norm": 0.0034972601570189,
-      "learning_rate": 0.00019999959430413978,
-      "loss": 46.0,
-      "step": 11869
-    },
-    {
-      "epoch": 0.9075443928359807,
-      "grad_norm": 0.0026538101956248283,
-      "learning_rate": 0.000199999594235717,
-      "loss": 46.0,
-      "step": 11870
-    },
-    {
-      "epoch": 0.9076208498193704,
-      "grad_norm": 0.0020829658024013042,
-      "learning_rate": 0.0001999995941672885,
-      "loss": 46.0,
-      "step": 11871
-    },
-    {
-      "epoch": 0.9076973068027601,
-      "grad_norm": 0.0010882171336561441,
-      "learning_rate": 0.0001999995940988542,
-      "loss": 46.0,
-      "step": 11872
-    },
-    {
-      "epoch": 0.9077737637861498,
-      "grad_norm": 0.001266672508791089,
-      "learning_rate": 0.00019999959403041415,
-      "loss": 46.0,
-      "step": 11873
-    },
-    {
-      "epoch": 0.9078502207695396,
-      "grad_norm": 0.010262281633913517,
-      "learning_rate": 0.00019999959396196833,
-      "loss": 46.0,
-      "step": 11874
-    },
-    {
-      "epoch": 0.9079266777529292,
-      "grad_norm": 0.0014136049430817366,
-      "learning_rate": 0.00019999959389351672,
-      "loss": 46.0,
-      "step": 11875
-    },
-    {
-      "epoch": 0.908003134736319,
-      "grad_norm": 0.0016992302844300866,
-      "learning_rate": 0.00019999959382505935,
-      "loss": 46.0,
-      "step": 11876
-    },
-    {
-      "epoch": 0.9080795917197086,
-      "grad_norm": 0.0009087096550501883,
-      "learning_rate": 0.0001999995937565962,
-      "loss": 46.0,
-      "step": 11877
-    },
-    {
-      "epoch": 0.9081560487030984,
-      "grad_norm": 0.0008233482367359102,
-      "learning_rate": 0.00019999959368812733,
-      "loss": 46.0,
-      "step": 11878
-    },
-    {
-      "epoch": 0.9082325056864882,
-      "grad_norm": 0.0004427512758411467,
-      "learning_rate": 0.00019999959361965264,
-      "loss": 46.0,
-      "step": 11879
-    },
-    {
-      "epoch": 0.9083089626698778,
-      "grad_norm": 0.0008080598781816661,
-      "learning_rate": 0.00019999959355117218,
-      "loss": 46.0,
-      "step": 11880
-    },
-    {
-      "epoch": 0.9083854196532676,
-      "grad_norm": 0.0011526867747306824,
-      "learning_rate": 0.00019999959348268598,
-      "loss": 46.0,
-      "step": 11881
-    },
-    {
-      "epoch": 0.9084618766366573,
-      "grad_norm": 0.0012742433464154601,
-      "learning_rate": 0.000199999593414194,
-      "loss": 46.0,
-      "step": 11882
-    },
-    {
-      "epoch": 0.908538333620047,
-      "grad_norm": 0.0008148119668476284,
-      "learning_rate": 0.00019999959334569625,
-      "loss": 46.0,
-      "step": 11883
-    },
-    {
-      "epoch": 0.9086147906034368,
-      "grad_norm": 0.0019154173787683249,
-      "learning_rate": 0.00019999959327719272,
-      "loss": 46.0,
-      "step": 11884
-    },
-    {
-      "epoch": 0.9086912475868265,
-      "grad_norm": 0.0005655597778968513,
-      "learning_rate": 0.00019999959320868342,
-      "loss": 46.0,
-      "step": 11885
-    },
-    {
-      "epoch": 0.9087677045702162,
-      "grad_norm": 0.0006353000644594431,
-      "learning_rate": 0.00019999959314016835,
-      "loss": 46.0,
-      "step": 11886
-    },
-    {
-      "epoch": 0.9088441615536059,
-      "grad_norm": 0.0008929286850616336,
-      "learning_rate": 0.00019999959307164756,
-      "loss": 46.0,
-      "step": 11887
-    },
-    {
-      "epoch": 0.9089206185369956,
-      "grad_norm": 0.00280123483389616,
-      "learning_rate": 0.00019999959300312097,
-      "loss": 46.0,
-      "step": 11888
-    },
-    {
-      "epoch": 0.9089970755203853,
-      "grad_norm": 0.0006641016807407141,
-      "learning_rate": 0.00019999959293458857,
-      "loss": 46.0,
-      "step": 11889
-    },
-    {
-      "epoch": 0.9090735325037751,
-      "grad_norm": 0.0028449599631130695,
-      "learning_rate": 0.00019999959286605044,
-      "loss": 46.0,
-      "step": 11890
-    },
-    {
-      "epoch": 0.9091499894871647,
-      "grad_norm": 0.002026725560426712,
-      "learning_rate": 0.00019999959279750655,
-      "loss": 46.0,
-      "step": 11891
-    },
-    {
-      "epoch": 0.9092264464705545,
-      "grad_norm": 0.002606593305245042,
-      "learning_rate": 0.00019999959272895687,
-      "loss": 46.0,
-      "step": 11892
-    },
-    {
-      "epoch": 0.9093029034539443,
-      "grad_norm": 0.0005219989689067006,
-      "learning_rate": 0.00019999959266040143,
-      "loss": 46.0,
-      "step": 11893
-    },
-    {
-      "epoch": 0.9093793604373339,
-      "grad_norm": 0.0028256692457944155,
-      "learning_rate": 0.0001999995925918402,
-      "loss": 46.0,
-      "step": 11894
-    },
-    {
-      "epoch": 0.9094558174207237,
-      "grad_norm": 0.002417660551145673,
-      "learning_rate": 0.0001999995925232732,
-      "loss": 46.0,
-      "step": 11895
-    },
-    {
-      "epoch": 0.9095322744041134,
-      "grad_norm": 0.0008055018843151629,
-      "learning_rate": 0.00019999959245470045,
-      "loss": 46.0,
-      "step": 11896
-    },
-    {
-      "epoch": 0.9096087313875031,
-      "grad_norm": 0.000683775229845196,
-      "learning_rate": 0.00019999959238612195,
-      "loss": 46.0,
-      "step": 11897
-    },
-    {
-      "epoch": 0.9096851883708928,
-      "grad_norm": 0.0032831078860908747,
-      "learning_rate": 0.00019999959231753765,
-      "loss": 46.0,
-      "step": 11898
-    },
-    {
-      "epoch": 0.9097616453542825,
-      "grad_norm": 0.0011444385163486004,
-      "learning_rate": 0.00019999959224894758,
-      "loss": 46.0,
-      "step": 11899
-    },
-    {
-      "epoch": 0.9098381023376723,
-      "grad_norm": 0.003476099343970418,
-      "learning_rate": 0.00019999959218035173,
-      "loss": 46.0,
-      "step": 11900
-    },
-    {
-      "epoch": 0.909914559321062,
-      "grad_norm": 0.0008008545846678317,
-      "learning_rate": 0.00019999959211175014,
-      "loss": 46.0,
-      "step": 11901
-    },
-    {
-      "epoch": 0.9099910163044517,
-      "grad_norm": 0.0021953631658107042,
-      "learning_rate": 0.00019999959204314278,
-      "loss": 46.0,
-      "step": 11902
-    },
-    {
-      "epoch": 0.9100674732878414,
-      "grad_norm": 0.0002964919258374721,
-      "learning_rate": 0.0001999995919745296,
-      "loss": 46.0,
-      "step": 11903
-    },
-    {
-      "epoch": 0.9101439302712312,
-      "grad_norm": 0.001034496701322496,
-      "learning_rate": 0.00019999959190591073,
-      "loss": 46.0,
-      "step": 11904
-    },
-    {
-      "epoch": 0.9102203872546208,
-      "grad_norm": 0.0013356952695176005,
-      "learning_rate": 0.00019999959183728604,
-      "loss": 46.0,
-      "step": 11905
-    },
-    {
-      "epoch": 0.9102968442380106,
-      "grad_norm": 0.02589714340865612,
-      "learning_rate": 0.00019999959176865558,
-      "loss": 46.0,
-      "step": 11906
-    },
-    {
-      "epoch": 0.9103733012214003,
-      "grad_norm": 0.0020736269652843475,
-      "learning_rate": 0.00019999959170001938,
-      "loss": 46.0,
-      "step": 11907
-    },
-    {
-      "epoch": 0.91044975820479,
-      "grad_norm": 0.005621438380330801,
-      "learning_rate": 0.0001999995916313774,
-      "loss": 46.0,
-      "step": 11908
-    },
-    {
-      "epoch": 0.9105262151881798,
-      "grad_norm": 0.0028752838261425495,
-      "learning_rate": 0.00019999959156272965,
-      "loss": 46.0,
-      "step": 11909
-    },
-    {
-      "epoch": 0.9106026721715694,
-      "grad_norm": 0.006113926414400339,
-      "learning_rate": 0.00019999959149407612,
-      "loss": 46.0,
-      "step": 11910
-    },
-    {
-      "epoch": 0.9106791291549592,
-      "grad_norm": 0.0011571331415325403,
-      "learning_rate": 0.00019999959142541683,
-      "loss": 46.0,
-      "step": 11911
-    },
-    {
-      "epoch": 0.910755586138349,
-      "grad_norm": 0.0034599846694618464,
-      "learning_rate": 0.00019999959135675176,
-      "loss": 46.0,
-      "step": 11912
-    },
-    {
-      "epoch": 0.9108320431217386,
-      "grad_norm": 0.0006363982101902366,
-      "learning_rate": 0.0001999995912880809,
-      "loss": 46.0,
-      "step": 11913
-    },
-    {
-      "epoch": 0.9109085001051284,
-      "grad_norm": 0.002225884236395359,
-      "learning_rate": 0.00019999959121940432,
-      "loss": 46.0,
-      "step": 11914
-    },
-    {
-      "epoch": 0.9109849570885181,
-      "grad_norm": 0.01591384783387184,
-      "learning_rate": 0.00019999959115072193,
-      "loss": 46.0,
-      "step": 11915
-    },
-    {
-      "epoch": 0.9110614140719078,
-      "grad_norm": 0.0013474604347720742,
-      "learning_rate": 0.0001999995910820338,
-      "loss": 46.0,
-      "step": 11916
-    },
-    {
-      "epoch": 0.9111378710552975,
-      "grad_norm": 0.0008312687277793884,
-      "learning_rate": 0.00019999959101333988,
-      "loss": 46.0,
-      "step": 11917
-    },
-    {
-      "epoch": 0.9112143280386872,
-      "grad_norm": 0.0008586280164308846,
-      "learning_rate": 0.0001999995909446402,
-      "loss": 46.0,
-      "step": 11918
-    },
-    {
-      "epoch": 0.9112907850220769,
-      "grad_norm": 0.0019733973313122988,
-      "learning_rate": 0.00019999959087593476,
-      "loss": 46.0,
-      "step": 11919
-    },
-    {
-      "epoch": 0.9113672420054667,
-      "grad_norm": 0.005005836486816406,
-      "learning_rate": 0.00019999959080722353,
-      "loss": 46.0,
-      "step": 11920
-    },
-    {
-      "epoch": 0.9114436989888564,
-      "grad_norm": 0.0009364400175400078,
-      "learning_rate": 0.00019999959073850653,
-      "loss": 46.0,
-      "step": 11921
-    },
-    {
-      "epoch": 0.9115201559722461,
-      "grad_norm": 0.002472240710631013,
-      "learning_rate": 0.00019999959066978378,
-      "loss": 46.0,
-      "step": 11922
-    },
-    {
-      "epoch": 0.9115966129556359,
-      "grad_norm": 0.0018947651842609048,
-      "learning_rate": 0.00019999959060105525,
-      "loss": 46.0,
-      "step": 11923
-    },
-    {
-      "epoch": 0.9116730699390255,
-      "grad_norm": 0.0008839494548738003,
-      "learning_rate": 0.00019999959053232095,
-      "loss": 46.0,
-      "step": 11924
-    },
-    {
-      "epoch": 0.9117495269224153,
-      "grad_norm": 0.0013301677536219358,
-      "learning_rate": 0.00019999959046358088,
-      "loss": 46.0,
-      "step": 11925
-    },
-    {
-      "epoch": 0.911825983905805,
-      "grad_norm": 0.0009192051948048174,
-      "learning_rate": 0.00019999959039483504,
-      "loss": 46.0,
-      "step": 11926
-    },
-    {
-      "epoch": 0.9119024408891947,
-      "grad_norm": 0.000963798665907234,
-      "learning_rate": 0.00019999959032608345,
-      "loss": 46.0,
-      "step": 11927
-    },
-    {
-      "epoch": 0.9119788978725845,
-      "grad_norm": 0.000848840456455946,
-      "learning_rate": 0.00019999959025732606,
-      "loss": 46.0,
-      "step": 11928
-    },
-    {
-      "epoch": 0.9120553548559741,
-      "grad_norm": 0.00038978931843303144,
-      "learning_rate": 0.00019999959018856292,
-      "loss": 46.0,
-      "step": 11929
-    },
-    {
-      "epoch": 0.9121318118393639,
-      "grad_norm": 0.00507147703319788,
-      "learning_rate": 0.000199999590119794,
-      "loss": 46.0,
-      "step": 11930
-    },
-    {
-      "epoch": 0.9122082688227536,
-      "grad_norm": 0.0007379365270026028,
-      "learning_rate": 0.00019999959005101932,
-      "loss": 46.0,
-      "step": 11931
-    },
-    {
-      "epoch": 0.9122847258061433,
-      "grad_norm": 0.0013775257393717766,
-      "learning_rate": 0.00019999958998223887,
-      "loss": 46.0,
-      "step": 11932
-    },
-    {
-      "epoch": 0.912361182789533,
-      "grad_norm": 0.0009456261759623885,
-      "learning_rate": 0.00019999958991345263,
-      "loss": 46.0,
-      "step": 11933
-    },
-    {
-      "epoch": 0.9124376397729228,
-      "grad_norm": 0.0014599458081647754,
-      "learning_rate": 0.00019999958984466066,
-      "loss": 46.0,
-      "step": 11934
-    },
-    {
-      "epoch": 0.9125140967563125,
-      "grad_norm": 0.011240809224545956,
-      "learning_rate": 0.00019999958977586288,
-      "loss": 46.0,
-      "step": 11935
-    },
-    {
-      "epoch": 0.9125905537397022,
-      "grad_norm": 0.0004105728876311332,
-      "learning_rate": 0.00019999958970705936,
-      "loss": 46.0,
-      "step": 11936
-    },
-    {
-      "epoch": 0.9126670107230919,
-      "grad_norm": 0.0007375653367489576,
-      "learning_rate": 0.00019999958963825006,
-      "loss": 46.0,
-      "step": 11937
-    },
-    {
-      "epoch": 0.9127434677064816,
-      "grad_norm": 0.0009479140280745924,
-      "learning_rate": 0.000199999589569435,
-      "loss": 46.0,
-      "step": 11938
-    },
-    {
-      "epoch": 0.9128199246898714,
-      "grad_norm": 0.001707016839645803,
-      "learning_rate": 0.00019999958950061414,
-      "loss": 46.0,
-      "step": 11939
-    },
-    {
-      "epoch": 0.912896381673261,
-      "grad_norm": 0.0024091594386845827,
-      "learning_rate": 0.00019999958943178753,
-      "loss": 46.0,
-      "step": 11940
-    },
-    {
-      "epoch": 0.9129728386566508,
-      "grad_norm": 0.009356711991131306,
-      "learning_rate": 0.00019999958936295516,
-      "loss": 46.0,
-      "step": 11941
-    },
-    {
-      "epoch": 0.9130492956400406,
-      "grad_norm": 0.001243910868652165,
-      "learning_rate": 0.00019999958929411702,
-      "loss": 46.0,
-      "step": 11942
-    },
-    {
-      "epoch": 0.9131257526234302,
-      "grad_norm": 0.000702616642229259,
-      "learning_rate": 0.00019999958922527309,
-      "loss": 46.0,
-      "step": 11943
-    },
-    {
-      "epoch": 0.91320220960682,
-      "grad_norm": 0.002289367141202092,
-      "learning_rate": 0.0001999995891564234,
-      "loss": 46.0,
-      "step": 11944
-    },
-    {
-      "epoch": 0.9132786665902097,
-      "grad_norm": 0.0012534311972558498,
-      "learning_rate": 0.00019999958908756795,
-      "loss": 46.0,
-      "step": 11945
-    },
-    {
-      "epoch": 0.9133551235735994,
-      "grad_norm": 0.0011242461623623967,
-      "learning_rate": 0.00019999958901870671,
-      "loss": 46.0,
-      "step": 11946
-    },
-    {
-      "epoch": 0.9134315805569891,
-      "grad_norm": 0.00039433466736227274,
-      "learning_rate": 0.0001999995889498397,
-      "loss": 46.0,
-      "step": 11947
-    },
-    {
-      "epoch": 0.9135080375403788,
-      "grad_norm": 0.003199234139174223,
-      "learning_rate": 0.00019999958888096696,
-      "loss": 46.0,
-      "step": 11948
-    },
-    {
-      "epoch": 0.9135844945237686,
-      "grad_norm": 0.0010197182418778539,
-      "learning_rate": 0.00019999958881208844,
-      "loss": 46.0,
-      "step": 11949
-    },
-    {
-      "epoch": 0.9136609515071583,
-      "grad_norm": 0.0025528890546411276,
-      "learning_rate": 0.0001999995887432041,
-      "loss": 46.0,
-      "step": 11950
-    },
-    {
-      "epoch": 0.913737408490548,
-      "grad_norm": 0.0021053338423371315,
-      "learning_rate": 0.00019999958867431404,
-      "loss": 46.0,
-      "step": 11951
-    },
-    {
-      "epoch": 0.9138138654739377,
-      "grad_norm": 0.0011797461193054914,
-      "learning_rate": 0.0001999995886054182,
-      "loss": 46.0,
-      "step": 11952
-    },
-    {
-      "epoch": 0.9138903224573275,
-      "grad_norm": 0.0015162909403443336,
-      "learning_rate": 0.00019999958853651658,
-      "loss": 46.0,
-      "step": 11953
-    },
-    {
-      "epoch": 0.9139667794407171,
-      "grad_norm": 0.004422291181981564,
-      "learning_rate": 0.00019999958846760922,
-      "loss": 46.0,
-      "step": 11954
-    },
-    {
-      "epoch": 0.9140432364241069,
-      "grad_norm": 0.0011769947595894337,
-      "learning_rate": 0.00019999958839869605,
-      "loss": 46.0,
-      "step": 11955
-    },
-    {
-      "epoch": 0.9141196934074967,
-      "grad_norm": 0.0025571975857019424,
-      "learning_rate": 0.00019999958832977714,
-      "loss": 46.0,
-      "step": 11956
-    },
-    {
-      "epoch": 0.9141961503908863,
-      "grad_norm": 0.00062129448633641,
-      "learning_rate": 0.00019999958826085246,
-      "loss": 46.0,
-      "step": 11957
-    },
-    {
-      "epoch": 0.9142726073742761,
-      "grad_norm": 0.0020329910330474377,
-      "learning_rate": 0.000199999588191922,
-      "loss": 46.0,
-      "step": 11958
-    },
-    {
-      "epoch": 0.9143490643576657,
-      "grad_norm": 0.001169761293567717,
-      "learning_rate": 0.00019999958812298577,
-      "loss": 46.0,
-      "step": 11959
-    },
-    {
-      "epoch": 0.9144255213410555,
-      "grad_norm": 0.002692839130759239,
-      "learning_rate": 0.00019999958805404377,
-      "loss": 46.0,
-      "step": 11960
-    },
-    {
-      "epoch": 0.9145019783244452,
-      "grad_norm": 0.0006688666180707514,
-      "learning_rate": 0.000199999587985096,
-      "loss": 46.0,
-      "step": 11961
-    },
-    {
-      "epoch": 0.9145784353078349,
-      "grad_norm": 0.0013408127706497908,
-      "learning_rate": 0.00019999958791614247,
-      "loss": 46.0,
-      "step": 11962
-    },
-    {
-      "epoch": 0.9146548922912247,
-      "grad_norm": 0.00237915082834661,
-      "learning_rate": 0.00019999958784718314,
-      "loss": 46.0,
-      "step": 11963
-    },
-    {
-      "epoch": 0.9147313492746144,
-      "grad_norm": 0.0014154831878840923,
-      "learning_rate": 0.00019999958777821807,
-      "loss": 46.0,
-      "step": 11964
-    },
-    {
-      "epoch": 0.9148078062580041,
-      "grad_norm": 0.0015954760601744056,
-      "learning_rate": 0.00019999958770924723,
-      "loss": 46.0,
-      "step": 11965
-    },
-    {
-      "epoch": 0.9148842632413938,
-      "grad_norm": 0.005710669793188572,
-      "learning_rate": 0.0001999995876402706,
-      "loss": 46.0,
-      "step": 11966
-    },
-    {
-      "epoch": 0.9149607202247836,
-      "grad_norm": 0.0010667579481378198,
-      "learning_rate": 0.00019999958757128825,
-      "loss": 46.0,
-      "step": 11967
-    },
-    {
-      "epoch": 0.9150371772081732,
-      "grad_norm": 0.0015672474401071668,
-      "learning_rate": 0.0001999995875023001,
-      "loss": 46.0,
-      "step": 11968
-    },
-    {
-      "epoch": 0.915113634191563,
-      "grad_norm": 0.0029464783146977425,
-      "learning_rate": 0.00019999958743330618,
-      "loss": 46.0,
-      "step": 11969
-    },
-    {
-      "epoch": 0.9151900911749526,
-      "grad_norm": 0.004860200919210911,
-      "learning_rate": 0.0001999995873643065,
-      "loss": 46.0,
-      "step": 11970
-    },
-    {
-      "epoch": 0.9152665481583424,
-      "grad_norm": 0.0027938722632825375,
-      "learning_rate": 0.000199999587295301,
-      "loss": 46.0,
-      "step": 11971
-    },
-    {
-      "epoch": 0.9153430051417322,
-      "grad_norm": 0.005090589635074139,
-      "learning_rate": 0.00019999958722628978,
-      "loss": 46.0,
-      "step": 11972
-    },
-    {
-      "epoch": 0.9154194621251218,
-      "grad_norm": 0.001975435996428132,
-      "learning_rate": 0.00019999958715727277,
-      "loss": 46.0,
-      "step": 11973
-    },
-    {
-      "epoch": 0.9154959191085116,
-      "grad_norm": 0.00369916670024395,
-      "learning_rate": 0.00019999958708825002,
-      "loss": 46.0,
-      "step": 11974
-    },
-    {
-      "epoch": 0.9155723760919013,
-      "grad_norm": 0.0041347891092300415,
-      "learning_rate": 0.00019999958701922145,
-      "loss": 46.0,
-      "step": 11975
-    },
-    {
-      "epoch": 0.915648833075291,
-      "grad_norm": 0.001029394450597465,
-      "learning_rate": 0.00019999958695018718,
-      "loss": 46.0,
-      "step": 11976
-    },
-    {
-      "epoch": 0.9157252900586808,
-      "grad_norm": 0.005500374361872673,
-      "learning_rate": 0.00019999958688114708,
-      "loss": 46.0,
-      "step": 11977
-    },
-    {
-      "epoch": 0.9158017470420704,
-      "grad_norm": 0.0014091175980865955,
-      "learning_rate": 0.00019999958681210124,
-      "loss": 46.0,
-      "step": 11978
-    },
-    {
-      "epoch": 0.9158782040254602,
-      "grad_norm": 0.0011467932490631938,
-      "learning_rate": 0.00019999958674304962,
-      "loss": 46.0,
-      "step": 11979
-    },
-    {
-      "epoch": 0.9159546610088499,
-      "grad_norm": 0.0007589886081404984,
-      "learning_rate": 0.00019999958667399223,
-      "loss": 46.0,
-      "step": 11980
-    },
-    {
-      "epoch": 0.9160311179922396,
-      "grad_norm": 0.0007121781236492097,
-      "learning_rate": 0.00019999958660492907,
-      "loss": 46.0,
-      "step": 11981
-    },
-    {
-      "epoch": 0.9161075749756293,
-      "grad_norm": 0.007561048958450556,
-      "learning_rate": 0.00019999958653586016,
-      "loss": 46.0,
-      "step": 11982
-    },
-    {
-      "epoch": 0.9161840319590191,
-      "grad_norm": 0.008352844044566154,
-      "learning_rate": 0.00019999958646678545,
-      "loss": 46.0,
-      "step": 11983
-    },
-    {
-      "epoch": 0.9162604889424087,
-      "grad_norm": 0.001725569716654718,
-      "learning_rate": 0.00019999958639770496,
-      "loss": 46.0,
-      "step": 11984
-    },
-    {
-      "epoch": 0.9163369459257985,
-      "grad_norm": 0.003644380485638976,
-      "learning_rate": 0.00019999958632861876,
-      "loss": 46.0,
-      "step": 11985
-    },
-    {
-      "epoch": 0.9164134029091883,
-      "grad_norm": 0.0012136718723922968,
-      "learning_rate": 0.00019999958625952676,
-      "loss": 46.0,
-      "step": 11986
-    },
-    {
-      "epoch": 0.9164898598925779,
-      "grad_norm": 0.0022887163795530796,
-      "learning_rate": 0.00019999958619042898,
-      "loss": 46.0,
-      "step": 11987
-    },
-    {
-      "epoch": 0.9165663168759677,
-      "grad_norm": 0.011860202997922897,
-      "learning_rate": 0.00019999958612132543,
-      "loss": 46.0,
-      "step": 11988
-    },
-    {
-      "epoch": 0.9166427738593573,
-      "grad_norm": 0.0006480338633991778,
-      "learning_rate": 0.00019999958605221614,
-      "loss": 46.0,
-      "step": 11989
-    },
-    {
-      "epoch": 0.9167192308427471,
-      "grad_norm": 0.0028399424627423286,
-      "learning_rate": 0.00019999958598310104,
-      "loss": 46.0,
-      "step": 11990
-    },
-    {
-      "epoch": 0.9167956878261369,
-      "grad_norm": 0.0005744567606598139,
-      "learning_rate": 0.00019999958591398017,
-      "loss": 46.0,
-      "step": 11991
-    },
-    {
-      "epoch": 0.9168721448095265,
-      "grad_norm": 0.0009081577882170677,
-      "learning_rate": 0.00019999958584485355,
-      "loss": 46.0,
-      "step": 11992
-    },
-    {
-      "epoch": 0.9169486017929163,
-      "grad_norm": 0.0006073405966162682,
-      "learning_rate": 0.0001999995857757212,
-      "loss": 46.0,
-      "step": 11993
-    },
-    {
-      "epoch": 0.917025058776306,
-      "grad_norm": 0.0006281262612901628,
-      "learning_rate": 0.00019999958570658303,
-      "loss": 46.0,
-      "step": 11994
-    },
-    {
-      "epoch": 0.9171015157596957,
-      "grad_norm": 0.0022540143691003323,
-      "learning_rate": 0.0001999995856374391,
-      "loss": 46.0,
-      "step": 11995
-    },
-    {
-      "epoch": 0.9171779727430854,
-      "grad_norm": 0.008779694326221943,
-      "learning_rate": 0.0001999995855682894,
-      "loss": 46.0,
-      "step": 11996
-    },
-    {
-      "epoch": 0.9172544297264752,
-      "grad_norm": 0.0010701692663133144,
-      "learning_rate": 0.00019999958549913392,
-      "loss": 46.0,
-      "step": 11997
-    },
-    {
-      "epoch": 0.9173308867098648,
-      "grad_norm": 0.0010692435316741467,
-      "learning_rate": 0.0001999995854299727,
-      "loss": 46.0,
-      "step": 11998
-    },
-    {
-      "epoch": 0.9174073436932546,
-      "grad_norm": 0.00120479182805866,
-      "learning_rate": 0.0001999995853608057,
-      "loss": 46.0,
-      "step": 11999
-    },
-    {
-      "epoch": 0.9174838006766443,
-      "grad_norm": 0.0023994368966668844,
-      "learning_rate": 0.00019999958529163292,
-      "loss": 46.0,
-      "step": 12000
-    },
-    {
-      "epoch": 0.917560257660034,
-      "grad_norm": 0.003430549055337906,
-      "learning_rate": 0.00019999958522245437,
-      "loss": 46.0,
-      "step": 12001
-    },
-    {
-      "epoch": 0.9176367146434238,
-      "grad_norm": 0.0007764084148220718,
-      "learning_rate": 0.00019999958515327007,
-      "loss": 46.0,
-      "step": 12002
-    },
-    {
-      "epoch": 0.9177131716268134,
-      "grad_norm": 0.0011081916745752096,
-      "learning_rate": 0.00019999958508408,
-      "loss": 46.0,
-      "step": 12003
-    },
-    {
-      "epoch": 0.9177896286102032,
-      "grad_norm": 0.0006723771221004426,
-      "learning_rate": 0.00019999958501488413,
-      "loss": 46.0,
-      "step": 12004
-    },
-    {
-      "epoch": 0.917866085593593,
-      "grad_norm": 0.0004039190534967929,
-      "learning_rate": 0.00019999958494568252,
-      "loss": 46.0,
-      "step": 12005
-    },
-    {
-      "epoch": 0.9179425425769826,
-      "grad_norm": 0.002531598322093487,
-      "learning_rate": 0.00019999958487647513,
-      "loss": 46.0,
-      "step": 12006
-    },
-    {
-      "epoch": 0.9180189995603724,
-      "grad_norm": 0.001886343234218657,
-      "learning_rate": 0.00019999958480726196,
-      "loss": 46.0,
-      "step": 12007
-    },
-    {
-      "epoch": 0.918095456543762,
-      "grad_norm": 0.0007136269705370069,
-      "learning_rate": 0.000199999584738043,
-      "loss": 46.0,
-      "step": 12008
-    },
-    {
-      "epoch": 0.9181719135271518,
-      "grad_norm": 0.0016071625286713243,
-      "learning_rate": 0.00019999958466881832,
-      "loss": 46.0,
-      "step": 12009
-    },
-    {
-      "epoch": 0.9182483705105415,
-      "grad_norm": 0.0032566061709076166,
-      "learning_rate": 0.00019999958459958786,
-      "loss": 46.0,
-      "step": 12010
-    },
-    {
-      "epoch": 0.9183248274939312,
-      "grad_norm": 0.0015204158844426274,
-      "learning_rate": 0.0001999995845303516,
-      "loss": 46.0,
-      "step": 12011
-    },
-    {
-      "epoch": 0.918401284477321,
-      "grad_norm": 0.000433815032010898,
-      "learning_rate": 0.0001999995844611096,
-      "loss": 46.0,
-      "step": 12012
-    },
-    {
-      "epoch": 0.9184777414607107,
-      "grad_norm": 0.0011306294472888112,
-      "learning_rate": 0.00019999958439186183,
-      "loss": 46.0,
-      "step": 12013
-    },
-    {
-      "epoch": 0.9185541984441004,
-      "grad_norm": 0.0009235804900527,
-      "learning_rate": 0.00019999958432260828,
-      "loss": 46.0,
-      "step": 12014
-    },
-    {
-      "epoch": 0.9186306554274901,
-      "grad_norm": 0.0018773576011881232,
-      "learning_rate": 0.00019999958425334896,
-      "loss": 46.0,
-      "step": 12015
-    },
-    {
-      "epoch": 0.9187071124108799,
-      "grad_norm": 0.0010949296411126852,
-      "learning_rate": 0.00019999958418408386,
-      "loss": 46.0,
-      "step": 12016
-    },
-    {
-      "epoch": 0.9187835693942695,
-      "grad_norm": 0.0010235997615382075,
-      "learning_rate": 0.00019999958411481302,
-      "loss": 46.0,
-      "step": 12017
-    },
-    {
-      "epoch": 0.9188600263776593,
-      "grad_norm": 0.002141745062544942,
-      "learning_rate": 0.00019999958404553637,
-      "loss": 46.0,
-      "step": 12018
-    },
-    {
-      "epoch": 0.9189364833610489,
-      "grad_norm": 0.0003547972009982914,
-      "learning_rate": 0.000199999583976254,
-      "loss": 46.0,
-      "step": 12019
-    },
-    {
-      "epoch": 0.9190129403444387,
-      "grad_norm": 0.002120896242558956,
-      "learning_rate": 0.00019999958390696585,
-      "loss": 46.0,
-      "step": 12020
-    },
-    {
-      "epoch": 0.9190893973278285,
-      "grad_norm": 0.005077350419014692,
-      "learning_rate": 0.0001999995838376719,
-      "loss": 46.0,
-      "step": 12021
-    },
-    {
-      "epoch": 0.9191658543112181,
-      "grad_norm": 0.00373087078332901,
-      "learning_rate": 0.0001999995837683722,
-      "loss": 46.0,
-      "step": 12022
-    },
-    {
-      "epoch": 0.9192423112946079,
-      "grad_norm": 0.0008808451821096241,
-      "learning_rate": 0.00019999958369906672,
-      "loss": 46.0,
-      "step": 12023
-    },
-    {
-      "epoch": 0.9193187682779976,
-      "grad_norm": 0.0010077588958665729,
-      "learning_rate": 0.0001999995836297555,
-      "loss": 46.0,
-      "step": 12024
-    },
-    {
-      "epoch": 0.9193952252613873,
-      "grad_norm": 0.0006262955139391124,
-      "learning_rate": 0.00019999958356043846,
-      "loss": 46.0,
-      "step": 12025
-    },
-    {
-      "epoch": 0.919471682244777,
-      "grad_norm": 0.0011571147479116917,
-      "learning_rate": 0.0001999995834911157,
-      "loss": 46.0,
-      "step": 12026
-    },
-    {
-      "epoch": 0.9195481392281668,
-      "grad_norm": 0.001044245669618249,
-      "learning_rate": 0.00019999958342178714,
-      "loss": 46.0,
-      "step": 12027
-    },
-    {
-      "epoch": 0.9196245962115565,
-      "grad_norm": 0.0031220833770930767,
-      "learning_rate": 0.00019999958335245284,
-      "loss": 46.0,
-      "step": 12028
-    },
-    {
-      "epoch": 0.9197010531949462,
-      "grad_norm": 0.0018916547996923327,
-      "learning_rate": 0.00019999958328311275,
-      "loss": 46.0,
-      "step": 12029
-    },
-    {
-      "epoch": 0.9197775101783359,
-      "grad_norm": 0.0017111971974372864,
-      "learning_rate": 0.00019999958321376688,
-      "loss": 46.0,
-      "step": 12030
-    },
-    {
-      "epoch": 0.9198539671617256,
-      "grad_norm": 0.0009090888779610395,
-      "learning_rate": 0.00019999958314441524,
-      "loss": 46.0,
-      "step": 12031
-    },
-    {
-      "epoch": 0.9199304241451154,
-      "grad_norm": 0.0014227371430024505,
-      "learning_rate": 0.00019999958307505785,
-      "loss": 46.0,
-      "step": 12032
-    },
-    {
-      "epoch": 0.920006881128505,
-      "grad_norm": 0.0024583826307207346,
-      "learning_rate": 0.0001999995830056947,
-      "loss": 46.0,
-      "step": 12033
-    },
-    {
-      "epoch": 0.9200833381118948,
-      "grad_norm": 0.004276834893971682,
-      "learning_rate": 0.00019999958293632575,
-      "loss": 46.0,
-      "step": 12034
-    },
-    {
-      "epoch": 0.9201597950952846,
-      "grad_norm": 0.004093404859304428,
-      "learning_rate": 0.00019999958286695107,
-      "loss": 46.0,
-      "step": 12035
-    },
-    {
-      "epoch": 0.9202362520786742,
-      "grad_norm": 0.00042168053914792836,
-      "learning_rate": 0.00019999958279757056,
-      "loss": 46.0,
-      "step": 12036
-    },
-    {
-      "epoch": 0.920312709062064,
-      "grad_norm": 0.0009360152180306613,
-      "learning_rate": 0.00019999958272818433,
-      "loss": 46.0,
-      "step": 12037
-    },
-    {
-      "epoch": 0.9203891660454536,
-      "grad_norm": 0.002095120958983898,
-      "learning_rate": 0.00019999958265879233,
-      "loss": 46.0,
-      "step": 12038
-    },
-    {
-      "epoch": 0.9204656230288434,
-      "grad_norm": 0.001412541838362813,
-      "learning_rate": 0.00019999958258939455,
-      "loss": 46.0,
-      "step": 12039
-    },
-    {
-      "epoch": 0.9205420800122331,
-      "grad_norm": 0.002074155490845442,
-      "learning_rate": 0.00019999958251999098,
-      "loss": 46.0,
-      "step": 12040
-    },
-    {
-      "epoch": 0.9206185369956228,
-      "grad_norm": 0.0013546498958021402,
-      "learning_rate": 0.00019999958245058168,
-      "loss": 46.0,
-      "step": 12041
-    },
-    {
-      "epoch": 0.9206949939790126,
-      "grad_norm": 0.00046360664418898523,
-      "learning_rate": 0.0001999995823811666,
-      "loss": 46.0,
-      "step": 12042
-    },
-    {
-      "epoch": 0.9207714509624023,
-      "grad_norm": 0.0016613882035017014,
-      "learning_rate": 0.00019999958231174572,
-      "loss": 46.0,
-      "step": 12043
-    },
-    {
-      "epoch": 0.920847907945792,
-      "grad_norm": 0.002969025168567896,
-      "learning_rate": 0.00019999958224231908,
-      "loss": 46.0,
-      "step": 12044
-    },
-    {
-      "epoch": 0.9209243649291817,
-      "grad_norm": 0.0013480096822604537,
-      "learning_rate": 0.0001999995821728867,
-      "loss": 46.0,
-      "step": 12045
-    },
-    {
-      "epoch": 0.9210008219125715,
-      "grad_norm": 0.00027013677754439414,
-      "learning_rate": 0.00019999958210344853,
-      "loss": 46.0,
-      "step": 12046
-    },
-    {
-      "epoch": 0.9210772788959611,
-      "grad_norm": 0.010235710069537163,
-      "learning_rate": 0.00019999958203400456,
-      "loss": 46.0,
-      "step": 12047
-    },
-    {
-      "epoch": 0.9211537358793509,
-      "grad_norm": 0.0003867988125421107,
-      "learning_rate": 0.00019999958196455486,
-      "loss": 46.0,
-      "step": 12048
-    },
-    {
-      "epoch": 0.9212301928627405,
-      "grad_norm": 0.003273334354162216,
-      "learning_rate": 0.0001999995818950994,
-      "loss": 46.0,
-      "step": 12049
-    },
-    {
-      "epoch": 0.9213066498461303,
-      "grad_norm": 0.0009624137892387807,
-      "learning_rate": 0.00019999958182563815,
-      "loss": 46.0,
-      "step": 12050
-    },
-    {
-      "epoch": 0.9213831068295201,
-      "grad_norm": 0.0034911471884697676,
-      "learning_rate": 0.00019999958175617112,
-      "loss": 46.0,
-      "step": 12051
-    },
-    {
-      "epoch": 0.9214595638129097,
-      "grad_norm": 0.0014526054728776217,
-      "learning_rate": 0.00019999958168669834,
-      "loss": 46.0,
-      "step": 12052
-    },
-    {
-      "epoch": 0.9215360207962995,
-      "grad_norm": 0.003096493659541011,
-      "learning_rate": 0.0001999995816172198,
-      "loss": 46.0,
-      "step": 12053
-    },
-    {
-      "epoch": 0.9216124777796892,
-      "grad_norm": 0.0016938558546826243,
-      "learning_rate": 0.00019999958154773547,
-      "loss": 46.0,
-      "step": 12054
-    },
-    {
-      "epoch": 0.9216889347630789,
-      "grad_norm": 0.0010012559359893203,
-      "learning_rate": 0.00019999958147824538,
-      "loss": 46.0,
-      "step": 12055
-    },
-    {
-      "epoch": 0.9217653917464687,
-      "grad_norm": 0.0010494185844436288,
-      "learning_rate": 0.0001999995814087495,
-      "loss": 46.0,
-      "step": 12056
-    },
-    {
-      "epoch": 0.9218418487298584,
-      "grad_norm": 0.003322313306853175,
-      "learning_rate": 0.00019999958133924787,
-      "loss": 46.0,
-      "step": 12057
-    },
-    {
-      "epoch": 0.9219183057132481,
-      "grad_norm": 0.001662363763898611,
-      "learning_rate": 0.00019999958126974048,
-      "loss": 46.0,
-      "step": 12058
-    },
-    {
-      "epoch": 0.9219947626966378,
-      "grad_norm": 0.001989088486880064,
-      "learning_rate": 0.00019999958120022732,
-      "loss": 46.0,
-      "step": 12059
-    },
-    {
-      "epoch": 0.9220712196800275,
-      "grad_norm": 0.000909341499209404,
-      "learning_rate": 0.00019999958113070835,
-      "loss": 46.0,
-      "step": 12060
-    },
-    {
-      "epoch": 0.9221476766634172,
-      "grad_norm": 0.0027445184532552958,
-      "learning_rate": 0.00019999958106118367,
-      "loss": 46.0,
-      "step": 12061
-    },
-    {
-      "epoch": 0.922224133646807,
-      "grad_norm": 0.0008998824050650001,
-      "learning_rate": 0.00019999958099165316,
-      "loss": 46.0,
-      "step": 12062
-    },
-    {
-      "epoch": 0.9223005906301966,
-      "grad_norm": 0.002928989240899682,
-      "learning_rate": 0.00019999958092211694,
-      "loss": 46.0,
-      "step": 12063
-    },
-    {
-      "epoch": 0.9223770476135864,
-      "grad_norm": 0.002669268986210227,
-      "learning_rate": 0.00019999958085257488,
-      "loss": 46.0,
-      "step": 12064
-    },
-    {
-      "epoch": 0.9224535045969762,
-      "grad_norm": 0.0023691777605563402,
-      "learning_rate": 0.0001999995807830271,
-      "loss": 46.0,
-      "step": 12065
-    },
-    {
-      "epoch": 0.9225299615803658,
-      "grad_norm": 0.0011108532780781388,
-      "learning_rate": 0.00019999958071347356,
-      "loss": 46.0,
-      "step": 12066
-    },
-    {
-      "epoch": 0.9226064185637556,
-      "grad_norm": 0.0008968866313807666,
-      "learning_rate": 0.00019999958064391424,
-      "loss": 46.0,
-      "step": 12067
-    },
-    {
-      "epoch": 0.9226828755471453,
-      "grad_norm": 0.0010244551813229918,
-      "learning_rate": 0.00019999958057434914,
-      "loss": 46.0,
-      "step": 12068
-    },
-    {
-      "epoch": 0.922759332530535,
-      "grad_norm": 0.001125934300944209,
-      "learning_rate": 0.00019999958050477827,
-      "loss": 46.0,
-      "step": 12069
-    },
-    {
-      "epoch": 0.9228357895139248,
-      "grad_norm": 0.00163006572984159,
-      "learning_rate": 0.00019999958043520163,
-      "loss": 46.0,
-      "step": 12070
-    },
-    {
-      "epoch": 0.9229122464973144,
-      "grad_norm": 0.0016496857861056924,
-      "learning_rate": 0.00019999958036561924,
-      "loss": 46.0,
-      "step": 12071
-    },
-    {
-      "epoch": 0.9229887034807042,
-      "grad_norm": 0.005408870987594128,
-      "learning_rate": 0.00019999958029603106,
-      "loss": 46.0,
-      "step": 12072
-    },
-    {
-      "epoch": 0.9230651604640939,
-      "grad_norm": 0.0012288877042010427,
-      "learning_rate": 0.00019999958022643712,
-      "loss": 46.0,
-      "step": 12073
-    },
-    {
-      "epoch": 0.9231416174474836,
-      "grad_norm": 0.006090317387133837,
-      "learning_rate": 0.0001999995801568374,
-      "loss": 46.0,
-      "step": 12074
-    },
-    {
-      "epoch": 0.9232180744308733,
-      "grad_norm": 0.005107733886688948,
-      "learning_rate": 0.00019999958008723193,
-      "loss": 46.0,
-      "step": 12075
-    },
-    {
-      "epoch": 0.9232945314142631,
-      "grad_norm": 0.003579028183594346,
-      "learning_rate": 0.00019999958001762065,
-      "loss": 46.0,
-      "step": 12076
-    },
-    {
-      "epoch": 0.9233709883976527,
-      "grad_norm": 0.0010870733531191945,
-      "learning_rate": 0.00019999957994800365,
-      "loss": 46.0,
-      "step": 12077
-    },
-    {
-      "epoch": 0.9234474453810425,
-      "grad_norm": 0.0013418698217719793,
-      "learning_rate": 0.00019999957987838085,
-      "loss": 46.0,
-      "step": 12078
-    },
-    {
-      "epoch": 0.9235239023644322,
-      "grad_norm": 0.00044725221232511103,
-      "learning_rate": 0.0001999995798087523,
-      "loss": 46.0,
-      "step": 12079
-    },
-    {
-      "epoch": 0.9236003593478219,
-      "grad_norm": 0.003975850064307451,
-      "learning_rate": 0.00019999957973911795,
-      "loss": 46.0,
-      "step": 12080
-    },
-    {
-      "epoch": 0.9236768163312117,
-      "grad_norm": 0.0017167160985991359,
-      "learning_rate": 0.00019999957966947786,
-      "loss": 46.0,
-      "step": 12081
-    },
-    {
-      "epoch": 0.9237532733146013,
-      "grad_norm": 0.005557652097195387,
-      "learning_rate": 0.000199999579599832,
-      "loss": 46.0,
-      "step": 12082
-    },
-    {
-      "epoch": 0.9238297302979911,
-      "grad_norm": 0.0005236169672571123,
-      "learning_rate": 0.00019999957953018035,
-      "loss": 46.0,
-      "step": 12083
-    },
-    {
-      "epoch": 0.9239061872813809,
-      "grad_norm": 0.0010544555261731148,
-      "learning_rate": 0.00019999957946052293,
-      "loss": 46.0,
-      "step": 12084
-    },
-    {
-      "epoch": 0.9239826442647705,
-      "grad_norm": 0.0007437972817569971,
-      "learning_rate": 0.00019999957939085977,
-      "loss": 46.0,
-      "step": 12085
-    },
-    {
-      "epoch": 0.9240591012481603,
-      "grad_norm": 0.0011064991122111678,
-      "learning_rate": 0.0001999995793211908,
-      "loss": 46.0,
-      "step": 12086
-    },
-    {
-      "epoch": 0.92413555823155,
-      "grad_norm": 0.0014206628547981381,
-      "learning_rate": 0.0001999995792515161,
-      "loss": 46.0,
-      "step": 12087
-    },
-    {
-      "epoch": 0.9242120152149397,
-      "grad_norm": 0.0007029398693703115,
-      "learning_rate": 0.00019999957918183562,
-      "loss": 46.0,
-      "step": 12088
-    },
-    {
-      "epoch": 0.9242884721983294,
-      "grad_norm": 0.0015322756953537464,
-      "learning_rate": 0.00019999957911214937,
-      "loss": 46.0,
-      "step": 12089
-    },
-    {
-      "epoch": 0.9243649291817191,
-      "grad_norm": 0.00215447461232543,
-      "learning_rate": 0.00019999957904245737,
-      "loss": 46.0,
-      "step": 12090
-    },
-    {
-      "epoch": 0.9244413861651088,
-      "grad_norm": 0.0012567087542265654,
-      "learning_rate": 0.00019999957897275957,
-      "loss": 46.0,
-      "step": 12091
-    },
-    {
-      "epoch": 0.9245178431484986,
-      "grad_norm": 0.0009863495361059904,
-      "learning_rate": 0.000199999578903056,
-      "loss": 46.0,
-      "step": 12092
-    },
-    {
-      "epoch": 0.9245943001318883,
-      "grad_norm": 0.0005452686455100775,
-      "learning_rate": 0.00019999957883334664,
-      "loss": 46.0,
-      "step": 12093
-    },
-    {
-      "epoch": 0.924670757115278,
-      "grad_norm": 0.001038632821291685,
-      "learning_rate": 0.00019999957876363155,
-      "loss": 46.0,
-      "step": 12094
-    },
-    {
-      "epoch": 0.9247472140986678,
-      "grad_norm": 0.0009294984047301114,
-      "learning_rate": 0.00019999957869391068,
-      "loss": 46.0,
-      "step": 12095
-    },
-    {
-      "epoch": 0.9248236710820574,
-      "grad_norm": 0.001106839976273477,
-      "learning_rate": 0.00019999957862418404,
-      "loss": 46.0,
-      "step": 12096
-    },
-    {
-      "epoch": 0.9249001280654472,
-      "grad_norm": 0.0019431129330769181,
-      "learning_rate": 0.00019999957855445163,
-      "loss": 46.0,
-      "step": 12097
-    },
-    {
-      "epoch": 0.924976585048837,
-      "grad_norm": 0.002364834537729621,
-      "learning_rate": 0.00019999957848471347,
-      "loss": 46.0,
-      "step": 12098
-    },
-    {
-      "epoch": 0.9250530420322266,
-      "grad_norm": 0.001127138384617865,
-      "learning_rate": 0.00019999957841496953,
-      "loss": 46.0,
-      "step": 12099
-    },
-    {
-      "epoch": 0.9251294990156164,
-      "grad_norm": 0.0011678830487653613,
-      "learning_rate": 0.0001999995783452198,
-      "loss": 46.0,
-      "step": 12100
-    },
-    {
-      "epoch": 0.925205955999006,
-      "grad_norm": 0.0008586464100517333,
-      "learning_rate": 0.00019999957827546432,
-      "loss": 46.0,
-      "step": 12101
-    },
-    {
-      "epoch": 0.9252824129823958,
-      "grad_norm": 0.0038136092480272055,
-      "learning_rate": 0.00019999957820570306,
-      "loss": 46.0,
-      "step": 12102
-    },
-    {
-      "epoch": 0.9253588699657855,
-      "grad_norm": 0.0007142901886254549,
-      "learning_rate": 0.000199999578135936,
-      "loss": 46.0,
-      "step": 12103
-    },
-    {
-      "epoch": 0.9254353269491752,
-      "grad_norm": 0.00046255378401838243,
-      "learning_rate": 0.00019999957806616324,
-      "loss": 46.0,
-      "step": 12104
-    },
-    {
-      "epoch": 0.925511783932565,
-      "grad_norm": 0.0012896499829366803,
-      "learning_rate": 0.00019999957799638466,
-      "loss": 46.0,
-      "step": 12105
-    },
-    {
-      "epoch": 0.9255882409159547,
-      "grad_norm": 0.0020871255546808243,
-      "learning_rate": 0.00019999957792660031,
-      "loss": 46.0,
-      "step": 12106
-    },
-    {
-      "epoch": 0.9256646978993444,
-      "grad_norm": 0.0013764593750238419,
-      "learning_rate": 0.00019999957785681022,
-      "loss": 46.0,
-      "step": 12107
-    },
-    {
-      "epoch": 0.9257411548827341,
-      "grad_norm": 0.0007337106508202851,
-      "learning_rate": 0.00019999957778701435,
-      "loss": 46.0,
-      "step": 12108
-    },
-    {
-      "epoch": 0.9258176118661238,
-      "grad_norm": 0.0021013757213950157,
-      "learning_rate": 0.00019999957771721269,
-      "loss": 46.0,
-      "step": 12109
-    },
-    {
-      "epoch": 0.9258940688495135,
-      "grad_norm": 0.002680626232177019,
-      "learning_rate": 0.0001999995776474053,
-      "loss": 46.0,
-      "step": 12110
-    },
-    {
-      "epoch": 0.9259705258329033,
-      "grad_norm": 0.0012728364672511816,
-      "learning_rate": 0.0001999995775775921,
-      "loss": 46.0,
-      "step": 12111
-    },
-    {
-      "epoch": 0.9260469828162929,
-      "grad_norm": 0.0010688784532248974,
-      "learning_rate": 0.00019999957750777315,
-      "loss": 46.0,
-      "step": 12112
-    },
-    {
-      "epoch": 0.9261234397996827,
-      "grad_norm": 0.0008296940359286964,
-      "learning_rate": 0.00019999957743794844,
-      "loss": 46.0,
-      "step": 12113
-    },
-    {
-      "epoch": 0.9261998967830725,
-      "grad_norm": 0.00026726440410129726,
-      "learning_rate": 0.00019999957736811794,
-      "loss": 46.0,
-      "step": 12114
-    },
-    {
-      "epoch": 0.9262763537664621,
-      "grad_norm": 0.009986361488699913,
-      "learning_rate": 0.00019999957729828168,
-      "loss": 46.0,
-      "step": 12115
-    },
-    {
-      "epoch": 0.9263528107498519,
-      "grad_norm": 0.010265923105180264,
-      "learning_rate": 0.00019999957722843968,
-      "loss": 46.0,
-      "step": 12116
-    },
-    {
-      "epoch": 0.9264292677332416,
-      "grad_norm": 0.0009398945840075612,
-      "learning_rate": 0.00019999957715859185,
-      "loss": 46.0,
-      "step": 12117
-    },
-    {
-      "epoch": 0.9265057247166313,
-      "grad_norm": 0.0010210475884377956,
-      "learning_rate": 0.00019999957708873828,
-      "loss": 46.0,
-      "step": 12118
-    },
-    {
-      "epoch": 0.926582181700021,
-      "grad_norm": 0.002771566156297922,
-      "learning_rate": 0.00019999957701887896,
-      "loss": 46.0,
-      "step": 12119
-    },
-    {
-      "epoch": 0.9266586386834107,
-      "grad_norm": 0.0021849293261766434,
-      "learning_rate": 0.00019999957694901387,
-      "loss": 46.0,
-      "step": 12120
-    },
-    {
-      "epoch": 0.9267350956668005,
-      "grad_norm": 0.00046759648830629885,
-      "learning_rate": 0.00019999957687914297,
-      "loss": 46.0,
-      "step": 12121
-    },
-    {
-      "epoch": 0.9268115526501902,
-      "grad_norm": 0.0012820098781958222,
-      "learning_rate": 0.00019999957680926633,
-      "loss": 46.0,
-      "step": 12122
-    },
-    {
-      "epoch": 0.9268880096335799,
-      "grad_norm": 0.004446228034794331,
-      "learning_rate": 0.0001999995767393839,
-      "loss": 46.0,
-      "step": 12123
-    },
-    {
-      "epoch": 0.9269644666169696,
-      "grad_norm": 0.000262592569924891,
-      "learning_rate": 0.0001999995766694957,
-      "loss": 46.0,
-      "step": 12124
-    },
-    {
-      "epoch": 0.9270409236003594,
-      "grad_norm": 0.0006439832504838705,
-      "learning_rate": 0.00019999957659960177,
-      "loss": 46.0,
-      "step": 12125
-    },
-    {
-      "epoch": 0.927117380583749,
-      "grad_norm": 0.0006937747821211815,
-      "learning_rate": 0.00019999957652970204,
-      "loss": 46.0,
-      "step": 12126
-    },
-    {
-      "epoch": 0.9271938375671388,
-      "grad_norm": 0.0019572488963603973,
-      "learning_rate": 0.00019999957645979656,
-      "loss": 46.0,
-      "step": 12127
-    },
-    {
-      "epoch": 0.9272702945505286,
-      "grad_norm": 0.0122261643409729,
-      "learning_rate": 0.00019999957638988528,
-      "loss": 46.0,
-      "step": 12128
-    },
-    {
-      "epoch": 0.9273467515339182,
-      "grad_norm": 0.0009052679524756968,
-      "learning_rate": 0.00019999957631996825,
-      "loss": 46.0,
-      "step": 12129
-    },
-    {
-      "epoch": 0.927423208517308,
-      "grad_norm": 0.0015283863758668303,
-      "learning_rate": 0.00019999957625004545,
-      "loss": 46.0,
-      "step": 12130
-    },
-    {
-      "epoch": 0.9274996655006976,
-      "grad_norm": 0.0026827522087842226,
-      "learning_rate": 0.0001999995761801169,
-      "loss": 46.0,
-      "step": 12131
-    },
-    {
-      "epoch": 0.9275761224840874,
-      "grad_norm": 0.0002819252840708941,
-      "learning_rate": 0.00019999957611018253,
-      "loss": 46.0,
-      "step": 12132
-    },
-    {
-      "epoch": 0.9276525794674771,
-      "grad_norm": 0.0007557621574960649,
-      "learning_rate": 0.00019999957604024244,
-      "loss": 46.0,
-      "step": 12133
-    },
-    {
-      "epoch": 0.9277290364508668,
-      "grad_norm": 0.0011717005399987102,
-      "learning_rate": 0.00019999957597029657,
-      "loss": 46.0,
-      "step": 12134
-    },
-    {
-      "epoch": 0.9278054934342566,
-      "grad_norm": 0.0007860861369408667,
-      "learning_rate": 0.00019999957590034493,
-      "loss": 46.0,
-      "step": 12135
-    },
-    {
-      "epoch": 0.9278819504176463,
-      "grad_norm": 0.004998686723411083,
-      "learning_rate": 0.00019999957583038752,
-      "loss": 46.0,
-      "step": 12136
-    },
-    {
-      "epoch": 0.927958407401036,
-      "grad_norm": 0.004086183849722147,
-      "learning_rate": 0.00019999957576042433,
-      "loss": 46.0,
-      "step": 12137
-    },
-    {
-      "epoch": 0.9280348643844257,
-      "grad_norm": 0.0010672503849491477,
-      "learning_rate": 0.00019999957569045534,
-      "loss": 46.0,
-      "step": 12138
-    },
-    {
-      "epoch": 0.9281113213678154,
-      "grad_norm": 0.0057138013653457165,
-      "learning_rate": 0.0001999995756204806,
-      "loss": 46.0,
-      "step": 12139
-    },
-    {
-      "epoch": 0.9281877783512051,
-      "grad_norm": 0.0020797853358089924,
-      "learning_rate": 0.00019999957555050016,
-      "loss": 46.0,
-      "step": 12140
-    },
-    {
-      "epoch": 0.9282642353345949,
-      "grad_norm": 0.00601800624281168,
-      "learning_rate": 0.00019999957548051388,
-      "loss": 46.0,
-      "step": 12141
-    },
-    {
-      "epoch": 0.9283406923179846,
-      "grad_norm": 0.002383144572377205,
-      "learning_rate": 0.00019999957541052182,
-      "loss": 46.0,
-      "step": 12142
-    },
-    {
-      "epoch": 0.9284171493013743,
-      "grad_norm": 0.011489329859614372,
-      "learning_rate": 0.00019999957534052402,
-      "loss": 46.0,
-      "step": 12143
-    },
-    {
-      "epoch": 0.9284936062847641,
-      "grad_norm": 0.004554878454655409,
-      "learning_rate": 0.00019999957527052048,
-      "loss": 46.0,
-      "step": 12144
-    },
-    {
-      "epoch": 0.9285700632681537,
-      "grad_norm": 0.0009480147855356336,
-      "learning_rate": 0.0001999995752005111,
-      "loss": 46.0,
-      "step": 12145
-    },
-    {
-      "epoch": 0.9286465202515435,
-      "grad_norm": 0.002283143810927868,
-      "learning_rate": 0.000199999575130496,
-      "loss": 46.0,
-      "step": 12146
-    },
-    {
-      "epoch": 0.9287229772349332,
-      "grad_norm": 0.006962282117456198,
-      "learning_rate": 0.00019999957506047512,
-      "loss": 46.0,
-      "step": 12147
-    },
-    {
-      "epoch": 0.9287994342183229,
-      "grad_norm": 0.0014182819286361337,
-      "learning_rate": 0.00019999957499044845,
-      "loss": 46.0,
-      "step": 12148
-    },
-    {
-      "epoch": 0.9288758912017127,
-      "grad_norm": 0.0007593846530653536,
-      "learning_rate": 0.00019999957492041606,
-      "loss": 46.0,
-      "step": 12149
-    },
-    {
-      "epoch": 0.9289523481851023,
-      "grad_norm": 0.02158031240105629,
-      "learning_rate": 0.00019999957485037788,
-      "loss": 46.0,
-      "step": 12150
-    },
-    {
-      "epoch": 0.9290288051684921,
-      "grad_norm": 0.0009281465318053961,
-      "learning_rate": 0.00019999957478033392,
-      "loss": 46.0,
-      "step": 12151
-    },
-    {
-      "epoch": 0.9291052621518818,
-      "grad_norm": 0.00036368679138831794,
-      "learning_rate": 0.00019999957471028416,
-      "loss": 46.0,
-      "step": 12152
-    },
-    {
-      "epoch": 0.9291817191352715,
-      "grad_norm": 0.006944100372493267,
-      "learning_rate": 0.0001999995746402287,
-      "loss": 46.0,
-      "step": 12153
-    },
-    {
-      "epoch": 0.9292581761186612,
-      "grad_norm": 0.002146515529602766,
-      "learning_rate": 0.0001999995745701674,
-      "loss": 46.0,
-      "step": 12154
-    },
-    {
-      "epoch": 0.929334633102051,
-      "grad_norm": 0.0015281970845535398,
-      "learning_rate": 0.00019999957450010037,
-      "loss": 46.0,
-      "step": 12155
-    },
-    {
-      "epoch": 0.9294110900854406,
-      "grad_norm": 0.001936238957569003,
-      "learning_rate": 0.00019999957443002757,
-      "loss": 46.0,
-      "step": 12156
-    },
-    {
-      "epoch": 0.9294875470688304,
-      "grad_norm": 0.002263550413772464,
-      "learning_rate": 0.00019999957435994897,
-      "loss": 46.0,
-      "step": 12157
-    },
-    {
-      "epoch": 0.9295640040522202,
-      "grad_norm": 0.001114763435907662,
-      "learning_rate": 0.00019999957428986465,
-      "loss": 46.0,
-      "step": 12158
-    },
-    {
-      "epoch": 0.9296404610356098,
-      "grad_norm": 0.0018859588308259845,
-      "learning_rate": 0.00019999957421977453,
-      "loss": 46.0,
-      "step": 12159
-    },
-    {
-      "epoch": 0.9297169180189996,
-      "grad_norm": 0.0029676223639398813,
-      "learning_rate": 0.00019999957414967864,
-      "loss": 46.0,
-      "step": 12160
-    },
-    {
-      "epoch": 0.9297933750023892,
-      "grad_norm": 0.0008736909367144108,
-      "learning_rate": 0.000199999574079577,
-      "loss": 46.0,
-      "step": 12161
-    },
-    {
-      "epoch": 0.929869831985779,
-      "grad_norm": 0.000999182229861617,
-      "learning_rate": 0.00019999957400946956,
-      "loss": 46.0,
-      "step": 12162
-    },
-    {
-      "epoch": 0.9299462889691688,
-      "grad_norm": 0.0012130794348195195,
-      "learning_rate": 0.00019999957393935638,
-      "loss": 46.0,
-      "step": 12163
-    },
-    {
-      "epoch": 0.9300227459525584,
-      "grad_norm": 0.0031265621073544025,
-      "learning_rate": 0.00019999957386923742,
-      "loss": 46.0,
-      "step": 12164
-    },
-    {
-      "epoch": 0.9300992029359482,
-      "grad_norm": 0.00073995441198349,
-      "learning_rate": 0.00019999957379911268,
-      "loss": 46.0,
-      "step": 12165
-    },
-    {
-      "epoch": 0.9301756599193379,
-      "grad_norm": 0.005015944130718708,
-      "learning_rate": 0.0001999995737289822,
-      "loss": 46.0,
-      "step": 12166
-    },
-    {
-      "epoch": 0.9302521169027276,
-      "grad_norm": 0.0020655386615544558,
-      "learning_rate": 0.00019999957365884592,
-      "loss": 46.0,
-      "step": 12167
-    },
-    {
-      "epoch": 0.9303285738861173,
-      "grad_norm": 0.006768765859305859,
-      "learning_rate": 0.00019999957358870387,
-      "loss": 46.0,
-      "step": 12168
-    },
-    {
-      "epoch": 0.930405030869507,
-      "grad_norm": 0.002902182750403881,
-      "learning_rate": 0.00019999957351855607,
-      "loss": 46.0,
-      "step": 12169
-    },
-    {
-      "epoch": 0.9304814878528967,
-      "grad_norm": 0.0009677843190729618,
-      "learning_rate": 0.0001999995734484025,
-      "loss": 46.0,
-      "step": 12170
-    },
-    {
-      "epoch": 0.9305579448362865,
-      "grad_norm": 0.0009656138718128204,
-      "learning_rate": 0.00019999957337824315,
-      "loss": 46.0,
-      "step": 12171
-    },
-    {
-      "epoch": 0.9306344018196762,
-      "grad_norm": 0.0026252625975757837,
-      "learning_rate": 0.00019999957330807803,
-      "loss": 46.0,
-      "step": 12172
-    },
-    {
-      "epoch": 0.9307108588030659,
-      "grad_norm": 0.0032303030602633953,
-      "learning_rate": 0.00019999957323790714,
-      "loss": 46.0,
-      "step": 12173
-    },
-    {
-      "epoch": 0.9307873157864557,
-      "grad_norm": 0.0008673305273987353,
-      "learning_rate": 0.00019999957316773048,
-      "loss": 46.0,
-      "step": 12174
-    },
-    {
-      "epoch": 0.9308637727698453,
-      "grad_norm": 0.0017201764276251197,
-      "learning_rate": 0.00019999957309754806,
-      "loss": 46.0,
-      "step": 12175
-    },
-    {
-      "epoch": 0.9309402297532351,
-      "grad_norm": 0.0009072452085092664,
-      "learning_rate": 0.00019999957302735988,
-      "loss": 46.0,
-      "step": 12176
-    },
-    {
-      "epoch": 0.9310166867366249,
-      "grad_norm": 0.0005744668887928128,
-      "learning_rate": 0.0001999995729571659,
-      "loss": 46.0,
-      "step": 12177
-    },
-    {
-      "epoch": 0.9310931437200145,
-      "grad_norm": 0.0011844061082229018,
-      "learning_rate": 0.00019999957288696619,
-      "loss": 46.0,
-      "step": 12178
-    },
-    {
-      "epoch": 0.9311696007034043,
-      "grad_norm": 0.0006330212927423418,
-      "learning_rate": 0.00019999957281676065,
-      "loss": 46.0,
-      "step": 12179
-    },
-    {
-      "epoch": 0.9312460576867939,
-      "grad_norm": 0.001336416695266962,
-      "learning_rate": 0.0001999995727465494,
-      "loss": 46.0,
-      "step": 12180
-    },
-    {
-      "epoch": 0.9313225146701837,
-      "grad_norm": 0.0029856860637664795,
-      "learning_rate": 0.00019999957267633235,
-      "loss": 46.0,
-      "step": 12181
-    },
-    {
-      "epoch": 0.9313989716535734,
-      "grad_norm": 0.004477132577449083,
-      "learning_rate": 0.00019999957260610955,
-      "loss": 46.0,
-      "step": 12182
-    },
-    {
-      "epoch": 0.9314754286369631,
-      "grad_norm": 0.0030429819598793983,
-      "learning_rate": 0.00019999957253588098,
-      "loss": 46.0,
-      "step": 12183
-    },
-    {
-      "epoch": 0.9315518856203528,
-      "grad_norm": 0.0019116413313895464,
-      "learning_rate": 0.00019999957246564663,
-      "loss": 46.0,
-      "step": 12184
-    },
-    {
-      "epoch": 0.9316283426037426,
-      "grad_norm": 0.0015499474247917533,
-      "learning_rate": 0.0001999995723954065,
-      "loss": 46.0,
-      "step": 12185
-    },
-    {
-      "epoch": 0.9317047995871323,
-      "grad_norm": 0.002475033514201641,
-      "learning_rate": 0.0001999995723251606,
-      "loss": 46.0,
-      "step": 12186
-    },
-    {
-      "epoch": 0.931781256570522,
-      "grad_norm": 0.0020716821309179068,
-      "learning_rate": 0.00019999957225490893,
-      "loss": 46.0,
-      "step": 12187
-    },
-    {
-      "epoch": 0.9318577135539118,
-      "grad_norm": 0.002118935575708747,
-      "learning_rate": 0.00019999957218465154,
-      "loss": 46.0,
-      "step": 12188
-    },
-    {
-      "epoch": 0.9319341705373014,
-      "grad_norm": 0.0021088302601128817,
-      "learning_rate": 0.0001999995721143883,
-      "loss": 46.0,
-      "step": 12189
-    },
-    {
-      "epoch": 0.9320106275206912,
-      "grad_norm": 0.0006917533464729786,
-      "learning_rate": 0.00019999957204411937,
-      "loss": 46.0,
-      "step": 12190
-    },
-    {
-      "epoch": 0.9320870845040808,
-      "grad_norm": 0.0017022602260112762,
-      "learning_rate": 0.0001999995719738446,
-      "loss": 46.0,
-      "step": 12191
-    },
-    {
-      "epoch": 0.9321635414874706,
-      "grad_norm": 0.0008511582273058593,
-      "learning_rate": 0.0001999995719035641,
-      "loss": 46.0,
-      "step": 12192
-    },
-    {
-      "epoch": 0.9322399984708604,
-      "grad_norm": 0.0029274257831275463,
-      "learning_rate": 0.00019999957183327785,
-      "loss": 46.0,
-      "step": 12193
-    },
-    {
-      "epoch": 0.93231645545425,
-      "grad_norm": 0.000857875740621239,
-      "learning_rate": 0.00019999957176298577,
-      "loss": 46.0,
-      "step": 12194
-    },
-    {
-      "epoch": 0.9323929124376398,
-      "grad_norm": 0.0006062813336029649,
-      "learning_rate": 0.00019999957169268798,
-      "loss": 46.0,
-      "step": 12195
-    },
-    {
-      "epoch": 0.9324693694210295,
-      "grad_norm": 0.0007906821556389332,
-      "learning_rate": 0.00019999957162238438,
-      "loss": 46.0,
-      "step": 12196
-    },
-    {
-      "epoch": 0.9325458264044192,
-      "grad_norm": 0.0007957852212712169,
-      "learning_rate": 0.00019999957155207506,
-      "loss": 46.0,
-      "step": 12197
-    },
-    {
-      "epoch": 0.932622283387809,
-      "grad_norm": 0.0011687454534694552,
-      "learning_rate": 0.0001999995714817599,
-      "loss": 46.0,
-      "step": 12198
-    },
-    {
-      "epoch": 0.9326987403711987,
-      "grad_norm": 0.0018368266755715013,
-      "learning_rate": 0.00019999957141143902,
-      "loss": 46.0,
-      "step": 12199
-    },
-    {
-      "epoch": 0.9327751973545884,
-      "grad_norm": 0.0008439357043243945,
-      "learning_rate": 0.00019999957134111238,
-      "loss": 46.0,
-      "step": 12200
-    },
-    {
-      "epoch": 0.9328516543379781,
-      "grad_norm": 0.0007795752608217299,
-      "learning_rate": 0.00019999957127077992,
-      "loss": 46.0,
-      "step": 12201
-    },
-    {
-      "epoch": 0.9329281113213678,
-      "grad_norm": 0.0008637941791675985,
-      "learning_rate": 0.00019999957120044176,
-      "loss": 46.0,
-      "step": 12202
-    },
-    {
-      "epoch": 0.9330045683047575,
-      "grad_norm": 0.0037607033737003803,
-      "learning_rate": 0.00019999957113009774,
-      "loss": 46.0,
-      "step": 12203
-    },
-    {
-      "epoch": 0.9330810252881473,
-      "grad_norm": 0.0033118084538728,
-      "learning_rate": 0.000199999571059748,
-      "loss": 46.0,
-      "step": 12204
-    },
-    {
-      "epoch": 0.9331574822715369,
-      "grad_norm": 0.0006602016510441899,
-      "learning_rate": 0.00019999957098939253,
-      "loss": 46.0,
-      "step": 12205
-    },
-    {
-      "epoch": 0.9332339392549267,
-      "grad_norm": 0.0010778333526104689,
-      "learning_rate": 0.00019999957091903126,
-      "loss": 46.0,
-      "step": 12206
-    },
-    {
-      "epoch": 0.9333103962383165,
-      "grad_norm": 0.0034906051587313414,
-      "learning_rate": 0.0001999995708486642,
-      "loss": 46.0,
-      "step": 12207
-    },
-    {
-      "epoch": 0.9333868532217061,
-      "grad_norm": 0.0019480622140690684,
-      "learning_rate": 0.00019999957077829138,
-      "loss": 46.0,
-      "step": 12208
-    },
-    {
-      "epoch": 0.9334633102050959,
-      "grad_norm": 0.0034833564423024654,
-      "learning_rate": 0.0001999995707079128,
-      "loss": 46.0,
-      "step": 12209
-    },
-    {
-      "epoch": 0.9335397671884855,
-      "grad_norm": 0.0009912169771268964,
-      "learning_rate": 0.00019999957063752843,
-      "loss": 46.0,
-      "step": 12210
-    },
-    {
-      "epoch": 0.9336162241718753,
-      "grad_norm": 0.0009244707180187106,
-      "learning_rate": 0.0001999995705671383,
-      "loss": 46.0,
-      "step": 12211
-    },
-    {
-      "epoch": 0.933692681155265,
-      "grad_norm": 0.0034028750378638506,
-      "learning_rate": 0.00019999957049674242,
-      "loss": 46.0,
-      "step": 12212
-    },
-    {
-      "epoch": 0.9337691381386547,
-      "grad_norm": 0.0011907349107787013,
-      "learning_rate": 0.00019999957042634073,
-      "loss": 46.0,
-      "step": 12213
-    },
-    {
-      "epoch": 0.9338455951220445,
-      "grad_norm": 0.002029647119343281,
-      "learning_rate": 0.00019999957035593332,
-      "loss": 46.0,
-      "step": 12214
-    },
-    {
-      "epoch": 0.9339220521054342,
-      "grad_norm": 0.0009335228241980076,
-      "learning_rate": 0.0001999995702855201,
-      "loss": 46.0,
-      "step": 12215
-    },
-    {
-      "epoch": 0.9339985090888239,
-      "grad_norm": 0.0028611645102500916,
-      "learning_rate": 0.00019999957021510115,
-      "loss": 46.0,
-      "step": 12216
-    },
-    {
-      "epoch": 0.9340749660722136,
-      "grad_norm": 0.0008055506623350084,
-      "learning_rate": 0.00019999957014467642,
-      "loss": 46.0,
-      "step": 12217
-    },
-    {
-      "epoch": 0.9341514230556034,
-      "grad_norm": 0.0021064123138785362,
-      "learning_rate": 0.00019999957007424591,
-      "loss": 46.0,
-      "step": 12218
-    },
-    {
-      "epoch": 0.934227880038993,
-      "grad_norm": 0.0012956928694620728,
-      "learning_rate": 0.00019999957000380964,
-      "loss": 46.0,
-      "step": 12219
-    },
-    {
-      "epoch": 0.9343043370223828,
-      "grad_norm": 0.002445425372570753,
-      "learning_rate": 0.00019999956993336756,
-      "loss": 46.0,
-      "step": 12220
-    },
-    {
-      "epoch": 0.9343807940057725,
-      "grad_norm": 0.005571737419813871,
-      "learning_rate": 0.00019999956986291973,
-      "loss": 46.0,
-      "step": 12221
-    },
-    {
-      "epoch": 0.9344572509891622,
-      "grad_norm": 0.0006551257101818919,
-      "learning_rate": 0.00019999956979246613,
-      "loss": 46.0,
-      "step": 12222
-    },
-    {
-      "epoch": 0.934533707972552,
-      "grad_norm": 0.0006578739848919213,
-      "learning_rate": 0.0001999995697220068,
-      "loss": 46.0,
-      "step": 12223
-    },
-    {
-      "epoch": 0.9346101649559416,
-      "grad_norm": 0.0012671825243160129,
-      "learning_rate": 0.00019999956965154167,
-      "loss": 46.0,
-      "step": 12224
-    },
-    {
-      "epoch": 0.9346866219393314,
-      "grad_norm": 0.0017036617500707507,
-      "learning_rate": 0.00019999956958107075,
-      "loss": 46.0,
-      "step": 12225
-    },
-    {
-      "epoch": 0.9347630789227211,
-      "grad_norm": 0.002068986650556326,
-      "learning_rate": 0.00019999956951059411,
-      "loss": 46.0,
-      "step": 12226
-    },
-    {
-      "epoch": 0.9348395359061108,
-      "grad_norm": 0.00040822228766046464,
-      "learning_rate": 0.00019999956944011168,
-      "loss": 46.0,
-      "step": 12227
-    },
-    {
-      "epoch": 0.9349159928895006,
-      "grad_norm": 0.0007027806714177132,
-      "learning_rate": 0.00019999956936962346,
-      "loss": 46.0,
-      "step": 12228
-    },
-    {
-      "epoch": 0.9349924498728903,
-      "grad_norm": 0.005172847304493189,
-      "learning_rate": 0.00019999956929912948,
-      "loss": 46.0,
-      "step": 12229
-    },
-    {
-      "epoch": 0.93506890685628,
-      "grad_norm": 0.0010630633914843202,
-      "learning_rate": 0.00019999956922862975,
-      "loss": 46.0,
-      "step": 12230
-    },
-    {
-      "epoch": 0.9351453638396697,
-      "grad_norm": 0.0009215231402777135,
-      "learning_rate": 0.00019999956915812424,
-      "loss": 46.0,
-      "step": 12231
-    },
-    {
-      "epoch": 0.9352218208230594,
-      "grad_norm": 0.0010831003310158849,
-      "learning_rate": 0.00019999956908761294,
-      "loss": 46.0,
-      "step": 12232
-    },
-    {
-      "epoch": 0.9352982778064491,
-      "grad_norm": 0.002291974378749728,
-      "learning_rate": 0.0001999995690170959,
-      "loss": 46.0,
-      "step": 12233
-    },
-    {
-      "epoch": 0.9353747347898389,
-      "grad_norm": 0.0016022274503484368,
-      "learning_rate": 0.0001999995689465731,
-      "loss": 46.0,
-      "step": 12234
-    },
-    {
-      "epoch": 0.9354511917732286,
-      "grad_norm": 0.0019400925375521183,
-      "learning_rate": 0.0001999995688760445,
-      "loss": 46.0,
-      "step": 12235
-    },
-    {
-      "epoch": 0.9355276487566183,
-      "grad_norm": 0.0031868666410446167,
-      "learning_rate": 0.00019999956880551012,
-      "loss": 46.0,
-      "step": 12236
-    },
-    {
-      "epoch": 0.9356041057400081,
-      "grad_norm": 0.0010111619485542178,
-      "learning_rate": 0.00019999956873497,
-      "loss": 46.0,
-      "step": 12237
-    },
-    {
-      "epoch": 0.9356805627233977,
-      "grad_norm": 0.0007766479975543916,
-      "learning_rate": 0.00019999956866442408,
-      "loss": 46.0,
-      "step": 12238
-    },
-    {
-      "epoch": 0.9357570197067875,
-      "grad_norm": 0.0019978678319603205,
-      "learning_rate": 0.00019999956859387245,
-      "loss": 46.0,
-      "step": 12239
-    },
-    {
-      "epoch": 0.9358334766901771,
-      "grad_norm": 0.0013941236538812518,
-      "learning_rate": 0.00019999956852331498,
-      "loss": 46.0,
-      "step": 12240
-    },
-    {
-      "epoch": 0.9359099336735669,
-      "grad_norm": 0.0013099727220833302,
-      "learning_rate": 0.0001999995684527518,
-      "loss": 46.0,
-      "step": 12241
-    },
-    {
-      "epoch": 0.9359863906569567,
-      "grad_norm": 0.0008953211945481598,
-      "learning_rate": 0.00019999956838218278,
-      "loss": 46.0,
-      "step": 12242
-    },
-    {
-      "epoch": 0.9360628476403463,
-      "grad_norm": 0.001701546716503799,
-      "learning_rate": 0.00019999956831160805,
-      "loss": 46.0,
-      "step": 12243
-    },
-    {
-      "epoch": 0.9361393046237361,
-      "grad_norm": 0.002749634673818946,
-      "learning_rate": 0.00019999956824102755,
-      "loss": 46.0,
-      "step": 12244
-    },
-    {
-      "epoch": 0.9362157616071258,
-      "grad_norm": 0.0016872003907337785,
-      "learning_rate": 0.00019999956817044124,
-      "loss": 46.0,
-      "step": 12245
-    },
-    {
-      "epoch": 0.9362922185905155,
-      "grad_norm": 0.0015682270750403404,
-      "learning_rate": 0.00019999956809984922,
-      "loss": 46.0,
-      "step": 12246
-    },
-    {
-      "epoch": 0.9363686755739052,
-      "grad_norm": 0.0013069581473246217,
-      "learning_rate": 0.00019999956802925137,
-      "loss": 46.0,
-      "step": 12247
-    },
-    {
-      "epoch": 0.936445132557295,
-      "grad_norm": 0.0005373494932428002,
-      "learning_rate": 0.00019999956795864777,
-      "loss": 46.0,
-      "step": 12248
-    },
-    {
-      "epoch": 0.9365215895406847,
-      "grad_norm": 0.006909727584570646,
-      "learning_rate": 0.00019999956788803843,
-      "loss": 46.0,
-      "step": 12249
-    },
-    {
-      "epoch": 0.9365980465240744,
-      "grad_norm": 0.002061674138531089,
-      "learning_rate": 0.00019999956781742328,
-      "loss": 46.0,
-      "step": 12250
-    },
-    {
-      "epoch": 0.9366745035074641,
-      "grad_norm": 0.0009205021196976304,
-      "learning_rate": 0.00019999956774680236,
-      "loss": 46.0,
-      "step": 12251
-    },
-    {
-      "epoch": 0.9367509604908538,
-      "grad_norm": 0.0012277847854420543,
-      "learning_rate": 0.00019999956767617573,
-      "loss": 46.0,
-      "step": 12252
-    },
-    {
-      "epoch": 0.9368274174742436,
-      "grad_norm": 0.0014238427393138409,
-      "learning_rate": 0.0001999995676055433,
-      "loss": 46.0,
-      "step": 12253
-    },
-    {
-      "epoch": 0.9369038744576332,
-      "grad_norm": 0.004767351783812046,
-      "learning_rate": 0.00019999956753490508,
-      "loss": 46.0,
-      "step": 12254
-    },
-    {
-      "epoch": 0.936980331441023,
-      "grad_norm": 0.005136764608323574,
-      "learning_rate": 0.0001999995674642611,
-      "loss": 46.0,
-      "step": 12255
-    },
-    {
-      "epoch": 0.9370567884244128,
-      "grad_norm": 0.0005738785257562995,
-      "learning_rate": 0.00019999956739361134,
-      "loss": 46.0,
-      "step": 12256
-    },
-    {
-      "epoch": 0.9371332454078024,
-      "grad_norm": 0.001265040598809719,
-      "learning_rate": 0.0001999995673229558,
-      "loss": 46.0,
-      "step": 12257
-    },
-    {
-      "epoch": 0.9372097023911922,
-      "grad_norm": 0.0011224690824747086,
-      "learning_rate": 0.00019999956725229453,
-      "loss": 46.0,
-      "step": 12258
-    },
-    {
-      "epoch": 0.9372861593745819,
-      "grad_norm": 0.003243668470531702,
-      "learning_rate": 0.00019999956718162748,
-      "loss": 46.0,
-      "step": 12259
-    },
-    {
-      "epoch": 0.9373626163579716,
-      "grad_norm": 0.0005513266660273075,
-      "learning_rate": 0.00019999956711095465,
-      "loss": 46.0,
-      "step": 12260
-    },
-    {
-      "epoch": 0.9374390733413613,
-      "grad_norm": 0.007840940728783607,
-      "learning_rate": 0.00019999956704027606,
-      "loss": 46.0,
-      "step": 12261
-    },
-    {
-      "epoch": 0.937515530324751,
-      "grad_norm": 0.002623951528221369,
-      "learning_rate": 0.00019999956696959168,
-      "loss": 46.0,
-      "step": 12262
-    },
-    {
-      "epoch": 0.9375919873081408,
-      "grad_norm": 0.00418903399258852,
-      "learning_rate": 0.00019999956689890157,
-      "loss": 46.0,
-      "step": 12263
-    },
-    {
-      "epoch": 0.9376684442915305,
-      "grad_norm": 0.006563264410942793,
-      "learning_rate": 0.00019999956682820565,
-      "loss": 46.0,
-      "step": 12264
-    },
-    {
-      "epoch": 0.9377449012749202,
-      "grad_norm": 0.0010737586999312043,
-      "learning_rate": 0.00019999956675750396,
-      "loss": 46.0,
-      "step": 12265
-    },
-    {
-      "epoch": 0.9378213582583099,
-      "grad_norm": 0.0031286252196878195,
-      "learning_rate": 0.00019999956668679652,
-      "loss": 46.0,
-      "step": 12266
-    },
-    {
-      "epoch": 0.9378978152416997,
-      "grad_norm": 0.0010336136911064386,
-      "learning_rate": 0.0001999995666160833,
-      "loss": 46.0,
-      "step": 12267
-    },
-    {
-      "epoch": 0.9379742722250893,
-      "grad_norm": 0.0005840155063197017,
-      "learning_rate": 0.00019999956654536433,
-      "loss": 46.0,
-      "step": 12268
-    },
-    {
-      "epoch": 0.9380507292084791,
-      "grad_norm": 0.0002955625532194972,
-      "learning_rate": 0.00019999956647463957,
-      "loss": 46.0,
-      "step": 12269
-    },
-    {
-      "epoch": 0.9381271861918687,
-      "grad_norm": 0.0015363151906058192,
-      "learning_rate": 0.00019999956640390907,
-      "loss": 46.0,
-      "step": 12270
-    },
-    {
-      "epoch": 0.9382036431752585,
-      "grad_norm": 0.001137645565904677,
-      "learning_rate": 0.00019999956633317276,
-      "loss": 46.0,
-      "step": 12271
-    },
-    {
-      "epoch": 0.9382801001586483,
-      "grad_norm": 0.001029876060783863,
-      "learning_rate": 0.0001999995662624307,
-      "loss": 46.0,
-      "step": 12272
-    },
-    {
-      "epoch": 0.9383565571420379,
-      "grad_norm": 0.0008531315252184868,
-      "learning_rate": 0.0001999995661916829,
-      "loss": 46.0,
-      "step": 12273
-    },
-    {
-      "epoch": 0.9384330141254277,
-      "grad_norm": 0.0009198343614116311,
-      "learning_rate": 0.0001999995661209293,
-      "loss": 46.0,
-      "step": 12274
-    },
-    {
-      "epoch": 0.9385094711088174,
-      "grad_norm": 0.001144756213761866,
-      "learning_rate": 0.0001999995660501699,
-      "loss": 46.0,
-      "step": 12275
-    },
-    {
-      "epoch": 0.9385859280922071,
-      "grad_norm": 0.000733811582904309,
-      "learning_rate": 0.00019999956597940478,
-      "loss": 46.0,
-      "step": 12276
-    },
-    {
-      "epoch": 0.9386623850755969,
-      "grad_norm": 0.001582833705469966,
-      "learning_rate": 0.00019999956590863386,
-      "loss": 46.0,
-      "step": 12277
-    },
-    {
-      "epoch": 0.9387388420589866,
-      "grad_norm": 0.0004762540338560939,
-      "learning_rate": 0.0001999995658378572,
-      "loss": 46.0,
-      "step": 12278
-    },
-    {
-      "epoch": 0.9388152990423763,
-      "grad_norm": 0.0022855489514768124,
-      "learning_rate": 0.00019999956576707476,
-      "loss": 46.0,
-      "step": 12279
-    },
-    {
-      "epoch": 0.938891756025766,
-      "grad_norm": 0.0014094322687014937,
-      "learning_rate": 0.00019999956569628652,
-      "loss": 46.0,
-      "step": 12280
-    },
-    {
-      "epoch": 0.9389682130091557,
-      "grad_norm": 0.0006910916999913752,
-      "learning_rate": 0.00019999956562549254,
-      "loss": 46.0,
-      "step": 12281
-    },
-    {
-      "epoch": 0.9390446699925454,
-      "grad_norm": 0.0011549191549420357,
-      "learning_rate": 0.00019999956555469278,
-      "loss": 46.0,
-      "step": 12282
-    },
-    {
-      "epoch": 0.9391211269759352,
-      "grad_norm": 0.0006775555666536093,
-      "learning_rate": 0.00019999956548388728,
-      "loss": 46.0,
-      "step": 12283
-    },
-    {
-      "epoch": 0.9391975839593248,
-      "grad_norm": 0.0043896036222577095,
-      "learning_rate": 0.00019999956541307597,
-      "loss": 46.0,
-      "step": 12284
-    },
-    {
-      "epoch": 0.9392740409427146,
-      "grad_norm": 0.0014866306446492672,
-      "learning_rate": 0.00019999956534225892,
-      "loss": 46.0,
-      "step": 12285
-    },
-    {
-      "epoch": 0.9393504979261044,
-      "grad_norm": 0.0008481458062306046,
-      "learning_rate": 0.00019999956527143607,
-      "loss": 46.0,
-      "step": 12286
-    },
-    {
-      "epoch": 0.939426954909494,
-      "grad_norm": 0.0006656827754341066,
-      "learning_rate": 0.00019999956520060747,
-      "loss": 46.0,
-      "step": 12287
-    },
-    {
-      "epoch": 0.9395034118928838,
-      "grad_norm": 0.0012219274649396539,
-      "learning_rate": 0.0001999995651297731,
-      "loss": 46.0,
-      "step": 12288
-    },
-    {
-      "epoch": 0.9395798688762735,
-      "grad_norm": 0.0010271830251440406,
-      "learning_rate": 0.000199999565058933,
-      "loss": 46.0,
-      "step": 12289
-    },
-    {
-      "epoch": 0.9396563258596632,
-      "grad_norm": 0.00046428918722085655,
-      "learning_rate": 0.00019999956498808704,
-      "loss": 46.0,
-      "step": 12290
-    },
-    {
-      "epoch": 0.939732782843053,
-      "grad_norm": 0.0013342913007363677,
-      "learning_rate": 0.00019999956491723538,
-      "loss": 46.0,
-      "step": 12291
-    },
-    {
-      "epoch": 0.9398092398264426,
-      "grad_norm": 0.0012896527769044042,
-      "learning_rate": 0.00019999956484637794,
-      "loss": 46.0,
-      "step": 12292
-    },
-    {
-      "epoch": 0.9398856968098324,
-      "grad_norm": 0.01178223267197609,
-      "learning_rate": 0.0001999995647755147,
-      "loss": 46.0,
-      "step": 12293
-    },
-    {
-      "epoch": 0.9399621537932221,
-      "grad_norm": 0.0011867908760905266,
-      "learning_rate": 0.00019999956470464572,
-      "loss": 46.0,
-      "step": 12294
-    },
-    {
-      "epoch": 0.9400386107766118,
-      "grad_norm": 0.0029271161183714867,
-      "learning_rate": 0.00019999956463377097,
-      "loss": 46.0,
-      "step": 12295
-    },
-    {
-      "epoch": 0.9401150677600015,
-      "grad_norm": 0.0008796355105005205,
-      "learning_rate": 0.00019999956456289044,
-      "loss": 46.0,
-      "step": 12296
-    },
-    {
-      "epoch": 0.9401915247433913,
-      "grad_norm": 0.0010045936796814203,
-      "learning_rate": 0.00019999956449200413,
-      "loss": 46.0,
-      "step": 12297
-    },
-    {
-      "epoch": 0.9402679817267809,
-      "grad_norm": 0.0005430988967418671,
-      "learning_rate": 0.00019999956442111208,
-      "loss": 46.0,
-      "step": 12298
-    },
-    {
-      "epoch": 0.9403444387101707,
-      "grad_norm": 0.0020924268756061792,
-      "learning_rate": 0.00019999956435021426,
-      "loss": 46.0,
-      "step": 12299
-    },
-    {
-      "epoch": 0.9404208956935605,
-      "grad_norm": 0.0008773243171162903,
-      "learning_rate": 0.00019999956427931064,
-      "loss": 46.0,
-      "step": 12300
-    },
-    {
-      "epoch": 0.9404973526769501,
-      "grad_norm": 0.006207504775375128,
-      "learning_rate": 0.00019999956420840127,
-      "loss": 46.0,
-      "step": 12301
-    },
-    {
-      "epoch": 0.9405738096603399,
-      "grad_norm": 0.017069486901164055,
-      "learning_rate": 0.00019999956413748612,
-      "loss": 46.0,
-      "step": 12302
-    },
-    {
-      "epoch": 0.9406502666437295,
-      "grad_norm": 0.0011084326542913914,
-      "learning_rate": 0.0001999995640665652,
-      "loss": 46.0,
-      "step": 12303
-    },
-    {
-      "epoch": 0.9407267236271193,
-      "grad_norm": 0.00774594210088253,
-      "learning_rate": 0.00019999956399563852,
-      "loss": 46.0,
-      "step": 12304
-    },
-    {
-      "epoch": 0.940803180610509,
-      "grad_norm": 0.0011596804251894355,
-      "learning_rate": 0.00019999956392470608,
-      "loss": 46.0,
-      "step": 12305
-    },
-    {
-      "epoch": 0.9408796375938987,
-      "grad_norm": 0.0007955761393532157,
-      "learning_rate": 0.00019999956385376787,
-      "loss": 46.0,
-      "step": 12306
-    },
-    {
-      "epoch": 0.9409560945772885,
-      "grad_norm": 0.0007006413070484996,
-      "learning_rate": 0.00019999956378282389,
-      "loss": 46.0,
-      "step": 12307
-    },
-    {
-      "epoch": 0.9410325515606782,
-      "grad_norm": 0.0021923750173300505,
-      "learning_rate": 0.0001999995637118741,
-      "loss": 46.0,
-      "step": 12308
-    },
-    {
-      "epoch": 0.9411090085440679,
-      "grad_norm": 0.001045380369760096,
-      "learning_rate": 0.00019999956364091857,
-      "loss": 46.0,
-      "step": 12309
-    },
-    {
-      "epoch": 0.9411854655274576,
-      "grad_norm": 0.0004303822643123567,
-      "learning_rate": 0.00019999956356995727,
-      "loss": 46.0,
-      "step": 12310
-    },
-    {
-      "epoch": 0.9412619225108473,
-      "grad_norm": 0.0014972623903304338,
-      "learning_rate": 0.00019999956349899022,
-      "loss": 46.0,
-      "step": 12311
-    },
-    {
-      "epoch": 0.941338379494237,
-      "grad_norm": 0.00041206591413356364,
-      "learning_rate": 0.00019999956342801737,
-      "loss": 46.0,
-      "step": 12312
-    },
-    {
-      "epoch": 0.9414148364776268,
-      "grad_norm": 0.0027731028385460377,
-      "learning_rate": 0.00019999956335703875,
-      "loss": 46.0,
-      "step": 12313
-    },
-    {
-      "epoch": 0.9414912934610165,
-      "grad_norm": 0.010360583662986755,
-      "learning_rate": 0.00019999956328605438,
-      "loss": 46.0,
-      "step": 12314
-    },
-    {
-      "epoch": 0.9415677504444062,
-      "grad_norm": 0.004714208655059338,
-      "learning_rate": 0.00019999956321506423,
-      "loss": 46.0,
-      "step": 12315
-    },
-    {
-      "epoch": 0.941644207427796,
-      "grad_norm": 0.0022166899871081114,
-      "learning_rate": 0.00019999956314406834,
-      "loss": 46.0,
-      "step": 12316
-    },
-    {
-      "epoch": 0.9417206644111856,
-      "grad_norm": 0.0011904215207323432,
-      "learning_rate": 0.00019999956307306666,
-      "loss": 46.0,
-      "step": 12317
-    },
-    {
-      "epoch": 0.9417971213945754,
-      "grad_norm": 0.0035876703914254904,
-      "learning_rate": 0.0001999995630020592,
-      "loss": 46.0,
-      "step": 12318
-    },
-    {
-      "epoch": 0.9418735783779651,
-      "grad_norm": 0.003940943628549576,
-      "learning_rate": 0.00019999956293104596,
-      "loss": 46.0,
-      "step": 12319
-    },
-    {
-      "epoch": 0.9419500353613548,
-      "grad_norm": 0.0007792055839672685,
-      "learning_rate": 0.000199999562860027,
-      "loss": 46.0,
-      "step": 12320
-    },
-    {
-      "epoch": 0.9420264923447446,
-      "grad_norm": 0.006641231942921877,
-      "learning_rate": 0.00019999956278900222,
-      "loss": 46.0,
-      "step": 12321
-    },
-    {
-      "epoch": 0.9421029493281342,
-      "grad_norm": 0.007249461021274328,
-      "learning_rate": 0.0001999995627179717,
-      "loss": 46.0,
-      "step": 12322
-    },
-    {
-      "epoch": 0.942179406311524,
-      "grad_norm": 0.0008969199261628091,
-      "learning_rate": 0.0001999995626469354,
-      "loss": 46.0,
-      "step": 12323
-    },
-    {
-      "epoch": 0.9422558632949137,
-      "grad_norm": 0.0011842675739899278,
-      "learning_rate": 0.0001999995625758933,
-      "loss": 46.0,
-      "step": 12324
-    },
-    {
-      "epoch": 0.9423323202783034,
-      "grad_norm": 0.0007077105692587793,
-      "learning_rate": 0.00019999956250484546,
-      "loss": 46.0,
-      "step": 12325
-    },
-    {
-      "epoch": 0.9424087772616931,
-      "grad_norm": 0.0017282310873270035,
-      "learning_rate": 0.00019999956243379186,
-      "loss": 46.0,
-      "step": 12326
-    },
-    {
-      "epoch": 0.9424852342450829,
-      "grad_norm": 0.0009200253407470882,
-      "learning_rate": 0.00019999956236273247,
-      "loss": 46.0,
-      "step": 12327
-    },
-    {
-      "epoch": 0.9425616912284726,
-      "grad_norm": 0.0015863269800320268,
-      "learning_rate": 0.00019999956229166735,
-      "loss": 46.0,
-      "step": 12328
-    },
-    {
-      "epoch": 0.9426381482118623,
-      "grad_norm": 0.002071093302220106,
-      "learning_rate": 0.00019999956222059643,
-      "loss": 46.0,
-      "step": 12329
-    },
-    {
-      "epoch": 0.9427146051952521,
-      "grad_norm": 0.0011401161318644881,
-      "learning_rate": 0.00019999956214951974,
-      "loss": 46.0,
-      "step": 12330
-    },
-    {
-      "epoch": 0.9427910621786417,
-      "grad_norm": 0.0006816425593569875,
-      "learning_rate": 0.00019999956207843728,
-      "loss": 46.0,
-      "step": 12331
-    },
-    {
-      "epoch": 0.9428675191620315,
-      "grad_norm": 0.001684514805674553,
-      "learning_rate": 0.00019999956200734907,
-      "loss": 46.0,
-      "step": 12332
-    },
-    {
-      "epoch": 0.9429439761454211,
-      "grad_norm": 0.004684693645685911,
-      "learning_rate": 0.00019999956193625506,
-      "loss": 46.0,
-      "step": 12333
-    },
-    {
-      "epoch": 0.9430204331288109,
-      "grad_norm": 0.0007201653206720948,
-      "learning_rate": 0.0001999995618651553,
-      "loss": 46.0,
-      "step": 12334
-    },
-    {
-      "epoch": 0.9430968901122007,
-      "grad_norm": 0.0011610996443778276,
-      "learning_rate": 0.00019999956179404978,
-      "loss": 46.0,
-      "step": 12335
-    },
-    {
-      "epoch": 0.9431733470955903,
-      "grad_norm": 0.0022103332448750734,
-      "learning_rate": 0.00019999956172293845,
-      "loss": 46.0,
-      "step": 12336
-    },
-    {
-      "epoch": 0.9432498040789801,
-      "grad_norm": 0.001221153186634183,
-      "learning_rate": 0.0001999995616518214,
-      "loss": 46.0,
-      "step": 12337
-    },
-    {
-      "epoch": 0.9433262610623698,
-      "grad_norm": 0.0012803459540009499,
-      "learning_rate": 0.00019999956158069855,
-      "loss": 46.0,
-      "step": 12338
-    },
-    {
-      "epoch": 0.9434027180457595,
-      "grad_norm": 0.00155855983030051,
-      "learning_rate": 0.00019999956150956993,
-      "loss": 46.0,
-      "step": 12339
-    },
-    {
-      "epoch": 0.9434791750291492,
-      "grad_norm": 0.0004491157305892557,
-      "learning_rate": 0.00019999956143843556,
-      "loss": 46.0,
-      "step": 12340
-    },
-    {
-      "epoch": 0.9435556320125389,
-      "grad_norm": 0.003105051815509796,
-      "learning_rate": 0.00019999956136729542,
-      "loss": 46.0,
-      "step": 12341
-    },
-    {
-      "epoch": 0.9436320889959287,
-      "grad_norm": 0.0007455964805558324,
-      "learning_rate": 0.00019999956129614947,
-      "loss": 46.0,
-      "step": 12342
-    },
-    {
-      "epoch": 0.9437085459793184,
-      "grad_norm": 0.0008121364517137408,
-      "learning_rate": 0.00019999956122499778,
-      "loss": 46.0,
-      "step": 12343
-    },
-    {
-      "epoch": 0.9437850029627081,
-      "grad_norm": 0.0010009503457695246,
-      "learning_rate": 0.00019999956115384032,
-      "loss": 46.0,
-      "step": 12344
-    },
-    {
-      "epoch": 0.9438614599460978,
-      "grad_norm": 0.003333230270072818,
-      "learning_rate": 0.0001999995610826771,
-      "loss": 46.0,
-      "step": 12345
-    },
-    {
-      "epoch": 0.9439379169294876,
-      "grad_norm": 0.003051307750865817,
-      "learning_rate": 0.0001999995610115081,
-      "loss": 46.0,
-      "step": 12346
-    },
-    {
-      "epoch": 0.9440143739128772,
-      "grad_norm": 0.0004749305371660739,
-      "learning_rate": 0.00019999956094033335,
-      "loss": 46.0,
-      "step": 12347
-    },
-    {
-      "epoch": 0.944090830896267,
-      "grad_norm": 0.0007354088011197746,
-      "learning_rate": 0.00019999956086915282,
-      "loss": 46.0,
-      "step": 12348
-    },
-    {
-      "epoch": 0.9441672878796568,
-      "grad_norm": 0.0005137072876095772,
-      "learning_rate": 0.00019999956079796652,
-      "loss": 46.0,
-      "step": 12349
-    },
-    {
-      "epoch": 0.9442437448630464,
-      "grad_norm": 0.0006483962060883641,
-      "learning_rate": 0.00019999956072677441,
-      "loss": 46.0,
-      "step": 12350
-    },
-    {
-      "epoch": 0.9443202018464362,
-      "grad_norm": 0.0018804912688210607,
-      "learning_rate": 0.0001999995606555766,
-      "loss": 46.0,
-      "step": 12351
-    },
-    {
-      "epoch": 0.9443966588298258,
-      "grad_norm": 0.0014351321151480079,
-      "learning_rate": 0.00019999956058437297,
-      "loss": 46.0,
-      "step": 12352
-    },
-    {
-      "epoch": 0.9444731158132156,
-      "grad_norm": 0.0006286718999035656,
-      "learning_rate": 0.00019999956051316357,
-      "loss": 46.0,
-      "step": 12353
-    },
-    {
-      "epoch": 0.9445495727966053,
-      "grad_norm": 0.001512345508672297,
-      "learning_rate": 0.0001999995604419484,
-      "loss": 46.0,
-      "step": 12354
-    },
-    {
-      "epoch": 0.944626029779995,
-      "grad_norm": 0.0006150231347419322,
-      "learning_rate": 0.00019999956037072752,
-      "loss": 46.0,
-      "step": 12355
-    },
-    {
-      "epoch": 0.9447024867633848,
-      "grad_norm": 0.0012920809676870704,
-      "learning_rate": 0.0001999995602995008,
-      "loss": 46.0,
-      "step": 12356
-    },
-    {
-      "epoch": 0.9447789437467745,
-      "grad_norm": 0.0013207169249653816,
-      "learning_rate": 0.00019999956022826836,
-      "loss": 46.0,
-      "step": 12357
-    },
-    {
-      "epoch": 0.9448554007301642,
-      "grad_norm": 0.006703382823616266,
-      "learning_rate": 0.00019999956015703013,
-      "loss": 46.0,
-      "step": 12358
-    },
-    {
-      "epoch": 0.9449318577135539,
-      "grad_norm": 0.008000780828297138,
-      "learning_rate": 0.00019999956008578612,
-      "loss": 46.0,
-      "step": 12359
-    },
-    {
-      "epoch": 0.9450083146969437,
-      "grad_norm": 0.0021756684873253107,
-      "learning_rate": 0.00019999956001453634,
-      "loss": 46.0,
-      "step": 12360
-    },
-    {
-      "epoch": 0.9450847716803333,
-      "grad_norm": 0.001456227619200945,
-      "learning_rate": 0.0001999995599432808,
-      "loss": 46.0,
-      "step": 12361
-    },
-    {
-      "epoch": 0.9451612286637231,
-      "grad_norm": 0.0035714353434741497,
-      "learning_rate": 0.0001999995598720195,
-      "loss": 46.0,
-      "step": 12362
-    },
-    {
-      "epoch": 0.9452376856471127,
-      "grad_norm": 0.0011776788160204887,
-      "learning_rate": 0.00019999955980075243,
-      "loss": 46.0,
-      "step": 12363
-    },
-    {
-      "epoch": 0.9453141426305025,
-      "grad_norm": 0.0006154084112495184,
-      "learning_rate": 0.00019999955972947956,
-      "loss": 46.0,
-      "step": 12364
-    },
-    {
-      "epoch": 0.9453905996138923,
-      "grad_norm": 0.0017240220913663507,
-      "learning_rate": 0.00019999955965820093,
-      "loss": 46.0,
-      "step": 12365
-    },
-    {
-      "epoch": 0.9454670565972819,
-      "grad_norm": 0.0017344050575047731,
-      "learning_rate": 0.0001999995595869166,
-      "loss": 46.0,
-      "step": 12366
-    },
-    {
-      "epoch": 0.9455435135806717,
-      "grad_norm": 0.00508534163236618,
-      "learning_rate": 0.00019999955951562642,
-      "loss": 46.0,
-      "step": 12367
-    },
-    {
-      "epoch": 0.9456199705640614,
-      "grad_norm": 0.000537124287802726,
-      "learning_rate": 0.00019999955944433048,
-      "loss": 46.0,
-      "step": 12368
-    },
-    {
-      "epoch": 0.9456964275474511,
-      "grad_norm": 0.0002548867487348616,
-      "learning_rate": 0.0001999995593730288,
-      "loss": 46.0,
-      "step": 12369
-    },
-    {
-      "epoch": 0.9457728845308409,
-      "grad_norm": 0.0026585611049085855,
-      "learning_rate": 0.00019999955930172133,
-      "loss": 46.0,
-      "step": 12370
-    },
-    {
-      "epoch": 0.9458493415142305,
-      "grad_norm": 0.002630931092426181,
-      "learning_rate": 0.0001999995592304081,
-      "loss": 46.0,
-      "step": 12371
-    },
-    {
-      "epoch": 0.9459257984976203,
-      "grad_norm": 0.0030295723117887974,
-      "learning_rate": 0.0001999995591590891,
-      "loss": 46.0,
-      "step": 12372
-    },
-    {
-      "epoch": 0.94600225548101,
-      "grad_norm": 0.0005532665527425706,
-      "learning_rate": 0.0001999995590877643,
-      "loss": 46.0,
-      "step": 12373
-    },
-    {
-      "epoch": 0.9460787124643997,
-      "grad_norm": 0.0017418792704120278,
-      "learning_rate": 0.00019999955901643378,
-      "loss": 46.0,
-      "step": 12374
-    },
-    {
-      "epoch": 0.9461551694477894,
-      "grad_norm": 0.0006412502261810005,
-      "learning_rate": 0.00019999955894509748,
-      "loss": 46.0,
-      "step": 12375
-    },
-    {
-      "epoch": 0.9462316264311792,
-      "grad_norm": 0.0010607157601043582,
-      "learning_rate": 0.0001999995588737554,
-      "loss": 46.0,
-      "step": 12376
-    },
-    {
-      "epoch": 0.9463080834145688,
-      "grad_norm": 0.0014223423786461353,
-      "learning_rate": 0.00019999955880240755,
-      "loss": 46.0,
-      "step": 12377
-    },
-    {
-      "epoch": 0.9463845403979586,
-      "grad_norm": 0.00037162474473007023,
-      "learning_rate": 0.0001999995587310539,
-      "loss": 46.0,
-      "step": 12378
-    },
-    {
-      "epoch": 0.9464609973813484,
-      "grad_norm": 0.0007151460740715265,
-      "learning_rate": 0.00019999955865969454,
-      "loss": 46.0,
-      "step": 12379
-    },
-    {
-      "epoch": 0.946537454364738,
-      "grad_norm": 0.0018177909078076482,
-      "learning_rate": 0.00019999955858832937,
-      "loss": 46.0,
-      "step": 12380
-    },
-    {
-      "epoch": 0.9466139113481278,
-      "grad_norm": 0.0011734383879229426,
-      "learning_rate": 0.00019999955851695845,
-      "loss": 46.0,
-      "step": 12381
-    },
-    {
-      "epoch": 0.9466903683315174,
-      "grad_norm": 0.0010106373811140656,
-      "learning_rate": 0.00019999955844558174,
-      "loss": 46.0,
-      "step": 12382
-    },
-    {
-      "epoch": 0.9467668253149072,
-      "grad_norm": 0.0010133297182619572,
-      "learning_rate": 0.00019999955837419927,
-      "loss": 46.0,
-      "step": 12383
-    },
-    {
-      "epoch": 0.946843282298297,
-      "grad_norm": 0.0007849809480831027,
-      "learning_rate": 0.00019999955830281104,
-      "loss": 46.0,
-      "step": 12384
-    },
-    {
-      "epoch": 0.9469197392816866,
-      "grad_norm": 0.004273064434528351,
-      "learning_rate": 0.00019999955823141706,
-      "loss": 46.0,
-      "step": 12385
-    },
-    {
-      "epoch": 0.9469961962650764,
-      "grad_norm": 0.0011846947018057108,
-      "learning_rate": 0.00019999955816001728,
-      "loss": 46.0,
-      "step": 12386
-    },
-    {
-      "epoch": 0.9470726532484661,
-      "grad_norm": 0.002637983998283744,
-      "learning_rate": 0.00019999955808861172,
-      "loss": 46.0,
-      "step": 12387
-    },
-    {
-      "epoch": 0.9471491102318558,
-      "grad_norm": 0.0005716111045330763,
-      "learning_rate": 0.0001999995580172004,
-      "loss": 46.0,
-      "step": 12388
-    },
-    {
-      "epoch": 0.9472255672152455,
-      "grad_norm": 0.0006461918819695711,
-      "learning_rate": 0.00019999955794578332,
-      "loss": 46.0,
-      "step": 12389
-    },
-    {
-      "epoch": 0.9473020241986353,
-      "grad_norm": 0.002386166714131832,
-      "learning_rate": 0.0001999995578743605,
-      "loss": 46.0,
-      "step": 12390
-    },
-    {
-      "epoch": 0.947378481182025,
-      "grad_norm": 0.000282397901173681,
-      "learning_rate": 0.00019999955780293185,
-      "loss": 46.0,
-      "step": 12391
-    },
-    {
-      "epoch": 0.9474549381654147,
-      "grad_norm": 0.0017616646364331245,
-      "learning_rate": 0.00019999955773149745,
-      "loss": 46.0,
-      "step": 12392
-    },
-    {
-      "epoch": 0.9475313951488044,
-      "grad_norm": 0.0009565998334437609,
-      "learning_rate": 0.00019999955766005729,
-      "loss": 46.0,
-      "step": 12393
-    },
-    {
-      "epoch": 0.9476078521321941,
-      "grad_norm": 0.009229855611920357,
-      "learning_rate": 0.00019999955758861137,
-      "loss": 46.0,
-      "step": 12394
-    },
-    {
-      "epoch": 0.9476843091155839,
-      "grad_norm": 0.0028080889023840427,
-      "learning_rate": 0.00019999955751715966,
-      "loss": 46.0,
-      "step": 12395
-    },
-    {
-      "epoch": 0.9477607660989735,
-      "grad_norm": 0.000975471397396177,
-      "learning_rate": 0.0001999995574457022,
-      "loss": 46.0,
-      "step": 12396
-    },
-    {
-      "epoch": 0.9478372230823633,
-      "grad_norm": 0.0013243749272078276,
-      "learning_rate": 0.00019999955737423893,
-      "loss": 46.0,
-      "step": 12397
-    },
-    {
-      "epoch": 0.947913680065753,
-      "grad_norm": 0.0006749395979568362,
-      "learning_rate": 0.00019999955730276995,
-      "loss": 46.0,
-      "step": 12398
-    },
-    {
-      "epoch": 0.9479901370491427,
-      "grad_norm": 0.0018248808337375522,
-      "learning_rate": 0.00019999955723129517,
-      "loss": 46.0,
-      "step": 12399
-    },
-    {
-      "epoch": 0.9480665940325325,
-      "grad_norm": 0.000542721594683826,
-      "learning_rate": 0.00019999955715981464,
-      "loss": 46.0,
-      "step": 12400
-    },
-    {
-      "epoch": 0.9481430510159221,
-      "grad_norm": 0.0011014159535989165,
-      "learning_rate": 0.00019999955708832832,
-      "loss": 46.0,
-      "step": 12401
-    },
-    {
-      "epoch": 0.9482195079993119,
-      "grad_norm": 0.0021464594174176455,
-      "learning_rate": 0.00019999955701683621,
-      "loss": 46.0,
-      "step": 12402
-    },
-    {
-      "epoch": 0.9482959649827016,
-      "grad_norm": 0.004897886421531439,
-      "learning_rate": 0.0001999995569453384,
-      "loss": 46.0,
-      "step": 12403
-    },
-    {
-      "epoch": 0.9483724219660913,
-      "grad_norm": 0.0016293496591970325,
-      "learning_rate": 0.00019999955687383475,
-      "loss": 46.0,
-      "step": 12404
-    },
-    {
-      "epoch": 0.948448878949481,
-      "grad_norm": 0.0016750118229538202,
-      "learning_rate": 0.00019999955680232535,
-      "loss": 46.0,
-      "step": 12405
-    },
-    {
-      "epoch": 0.9485253359328708,
-      "grad_norm": 0.0031048518139868975,
-      "learning_rate": 0.00019999955673081018,
-      "loss": 46.0,
-      "step": 12406
-    },
-    {
-      "epoch": 0.9486017929162605,
-      "grad_norm": 0.010521024465560913,
-      "learning_rate": 0.00019999955665928924,
-      "loss": 46.0,
-      "step": 12407
-    },
-    {
-      "epoch": 0.9486782498996502,
-      "grad_norm": 0.0007901581702753901,
-      "learning_rate": 0.00019999955658776255,
-      "loss": 46.0,
-      "step": 12408
-    },
-    {
-      "epoch": 0.94875470688304,
-      "grad_norm": 0.0008239212911576033,
-      "learning_rate": 0.0001999995565162301,
-      "loss": 46.0,
-      "step": 12409
-    },
-    {
-      "epoch": 0.9488311638664296,
-      "grad_norm": 0.0004952264134772122,
-      "learning_rate": 0.00019999955644469183,
-      "loss": 46.0,
-      "step": 12410
-    },
-    {
-      "epoch": 0.9489076208498194,
-      "grad_norm": 0.007967387326061726,
-      "learning_rate": 0.00019999955637314782,
-      "loss": 46.0,
-      "step": 12411
-    },
-    {
-      "epoch": 0.948984077833209,
-      "grad_norm": 0.0026131724007427692,
-      "learning_rate": 0.00019999955630159804,
-      "loss": 46.0,
-      "step": 12412
-    },
-    {
-      "epoch": 0.9490605348165988,
-      "grad_norm": 0.0010560484370216727,
-      "learning_rate": 0.00019999955623004252,
-      "loss": 46.0,
-      "step": 12413
-    },
-    {
-      "epoch": 0.9491369917999886,
-      "grad_norm": 0.0011084924917668104,
-      "learning_rate": 0.0001999995561584812,
-      "loss": 46.0,
-      "step": 12414
-    },
-    {
-      "epoch": 0.9492134487833782,
-      "grad_norm": 0.0009459021966904402,
-      "learning_rate": 0.00019999955608691409,
-      "loss": 46.0,
-      "step": 12415
-    },
-    {
-      "epoch": 0.949289905766768,
-      "grad_norm": 0.0019901844207197428,
-      "learning_rate": 0.00019999955601534124,
-      "loss": 46.0,
-      "step": 12416
-    },
-    {
-      "epoch": 0.9493663627501577,
-      "grad_norm": 0.000977960298769176,
-      "learning_rate": 0.00019999955594376262,
-      "loss": 46.0,
-      "step": 12417
-    },
-    {
-      "epoch": 0.9494428197335474,
-      "grad_norm": 0.0002895413781516254,
-      "learning_rate": 0.0001999995558721782,
-      "loss": 46.0,
-      "step": 12418
-    },
-    {
-      "epoch": 0.9495192767169371,
-      "grad_norm": 0.0006636523175984621,
-      "learning_rate": 0.00019999955580058806,
-      "loss": 46.0,
-      "step": 12419
-    },
-    {
-      "epoch": 0.9495957337003269,
-      "grad_norm": 0.001335454755462706,
-      "learning_rate": 0.00019999955572899212,
-      "loss": 46.0,
-      "step": 12420
-    },
-    {
-      "epoch": 0.9496721906837166,
-      "grad_norm": 0.000873195007443428,
-      "learning_rate": 0.0001999995556573904,
-      "loss": 46.0,
-      "step": 12421
-    },
-    {
-      "epoch": 0.9497486476671063,
-      "grad_norm": 0.0046980599872767925,
-      "learning_rate": 0.00019999955558578294,
-      "loss": 46.0,
-      "step": 12422
-    },
-    {
-      "epoch": 0.949825104650496,
-      "grad_norm": 0.00045450410107150674,
-      "learning_rate": 0.00019999955551416968,
-      "loss": 46.0,
-      "step": 12423
-    },
-    {
-      "epoch": 0.9499015616338857,
-      "grad_norm": 0.0032560836989432573,
-      "learning_rate": 0.00019999955544255067,
-      "loss": 46.0,
-      "step": 12424
-    },
-    {
-      "epoch": 0.9499780186172755,
-      "grad_norm": 0.0008326891111209989,
-      "learning_rate": 0.0001999995553709259,
-      "loss": 46.0,
-      "step": 12425
-    },
-    {
-      "epoch": 0.9500544756006651,
-      "grad_norm": 0.0008003672119230032,
-      "learning_rate": 0.00019999955529929534,
-      "loss": 46.0,
-      "step": 12426
-    },
-    {
-      "epoch": 0.9501309325840549,
-      "grad_norm": 0.0005340115167200565,
-      "learning_rate": 0.000199999555227659,
-      "loss": 46.0,
-      "step": 12427
-    },
-    {
-      "epoch": 0.9502073895674447,
-      "grad_norm": 0.0011407658457756042,
-      "learning_rate": 0.00019999955515601694,
-      "loss": 46.0,
-      "step": 12428
-    },
-    {
-      "epoch": 0.9502838465508343,
-      "grad_norm": 0.0012825331650674343,
-      "learning_rate": 0.00019999955508436906,
-      "loss": 46.0,
-      "step": 12429
-    },
-    {
-      "epoch": 0.9503603035342241,
-      "grad_norm": 0.002244018716737628,
-      "learning_rate": 0.00019999955501271544,
-      "loss": 46.0,
-      "step": 12430
-    },
-    {
-      "epoch": 0.9504367605176138,
-      "grad_norm": 0.002747810911387205,
-      "learning_rate": 0.00019999955494105605,
-      "loss": 46.0,
-      "step": 12431
-    },
-    {
-      "epoch": 0.9505132175010035,
-      "grad_norm": 0.000954570947214961,
-      "learning_rate": 0.00019999955486939088,
-      "loss": 46.0,
-      "step": 12432
-    },
-    {
-      "epoch": 0.9505896744843932,
-      "grad_norm": 0.001188487745821476,
-      "learning_rate": 0.00019999955479771994,
-      "loss": 46.0,
-      "step": 12433
-    },
-    {
-      "epoch": 0.9506661314677829,
-      "grad_norm": 0.0026089854072779417,
-      "learning_rate": 0.00019999955472604322,
-      "loss": 46.0,
-      "step": 12434
-    },
-    {
-      "epoch": 0.9507425884511727,
-      "grad_norm": 0.0009334481437690556,
-      "learning_rate": 0.00019999955465436076,
-      "loss": 46.0,
-      "step": 12435
-    },
-    {
-      "epoch": 0.9508190454345624,
-      "grad_norm": 0.0011899399105459452,
-      "learning_rate": 0.0001999995545826725,
-      "loss": 46.0,
-      "step": 12436
-    },
-    {
-      "epoch": 0.9508955024179521,
-      "grad_norm": 0.0013193285558372736,
-      "learning_rate": 0.00019999955451097847,
-      "loss": 46.0,
-      "step": 12437
-    },
-    {
-      "epoch": 0.9509719594013418,
-      "grad_norm": 0.0012993119889870286,
-      "learning_rate": 0.00019999955443927872,
-      "loss": 46.0,
-      "step": 12438
-    },
-    {
-      "epoch": 0.9510484163847316,
-      "grad_norm": 0.0010145334526896477,
-      "learning_rate": 0.00019999955436757314,
-      "loss": 46.0,
-      "step": 12439
-    },
-    {
-      "epoch": 0.9511248733681212,
-      "grad_norm": 0.0008084794972091913,
-      "learning_rate": 0.00019999955429586183,
-      "loss": 46.0,
-      "step": 12440
-    },
-    {
-      "epoch": 0.951201330351511,
-      "grad_norm": 0.0010774113470688462,
-      "learning_rate": 0.0001999995542241447,
-      "loss": 46.0,
-      "step": 12441
-    },
-    {
-      "epoch": 0.9512777873349006,
-      "grad_norm": 0.004175710491836071,
-      "learning_rate": 0.00019999955415242186,
-      "loss": 46.0,
-      "step": 12442
-    },
-    {
-      "epoch": 0.9513542443182904,
-      "grad_norm": 0.00113750493619591,
-      "learning_rate": 0.00019999955408069324,
-      "loss": 46.0,
-      "step": 12443
-    },
-    {
-      "epoch": 0.9514307013016802,
-      "grad_norm": 0.001192736905068159,
-      "learning_rate": 0.00019999955400895885,
-      "loss": 46.0,
-      "step": 12444
-    },
-    {
-      "epoch": 0.9515071582850698,
-      "grad_norm": 0.0011997545370832086,
-      "learning_rate": 0.00019999955393721868,
-      "loss": 46.0,
-      "step": 12445
-    },
-    {
-      "epoch": 0.9515836152684596,
-      "grad_norm": 0.001010085572488606,
-      "learning_rate": 0.00019999955386547274,
-      "loss": 46.0,
-      "step": 12446
-    },
-    {
-      "epoch": 0.9516600722518493,
-      "grad_norm": 0.0008382750093005598,
-      "learning_rate": 0.00019999955379372103,
-      "loss": 46.0,
-      "step": 12447
-    },
-    {
-      "epoch": 0.951736529235239,
-      "grad_norm": 0.0012769688619300723,
-      "learning_rate": 0.00019999955372196354,
-      "loss": 46.0,
-      "step": 12448
-    },
-    {
-      "epoch": 0.9518129862186288,
-      "grad_norm": 0.0012173093855381012,
-      "learning_rate": 0.00019999955365020028,
-      "loss": 46.0,
-      "step": 12449
-    },
-    {
-      "epoch": 0.9518894432020185,
-      "grad_norm": 0.0008360305218957365,
-      "learning_rate": 0.00019999955357843127,
-      "loss": 46.0,
-      "step": 12450
-    },
-    {
-      "epoch": 0.9519659001854082,
-      "grad_norm": 0.000583948043640703,
-      "learning_rate": 0.00019999955350665646,
-      "loss": 46.0,
-      "step": 12451
-    },
-    {
-      "epoch": 0.9520423571687979,
-      "grad_norm": 0.01885116659104824,
-      "learning_rate": 0.00019999955343487594,
-      "loss": 46.0,
-      "step": 12452
-    },
-    {
-      "epoch": 0.9521188141521876,
-      "grad_norm": 0.0010129577713087201,
-      "learning_rate": 0.0001999995533630896,
-      "loss": 46.0,
-      "step": 12453
-    },
-    {
-      "epoch": 0.9521952711355773,
-      "grad_norm": 0.012866014614701271,
-      "learning_rate": 0.00019999955329129748,
-      "loss": 46.0,
-      "step": 12454
-    },
-    {
-      "epoch": 0.9522717281189671,
-      "grad_norm": 0.006403828505426645,
-      "learning_rate": 0.00019999955321949964,
-      "loss": 46.0,
-      "step": 12455
-    },
-    {
-      "epoch": 0.9523481851023567,
-      "grad_norm": 0.0012411960633471608,
-      "learning_rate": 0.00019999955314769602,
-      "loss": 46.0,
-      "step": 12456
-    },
-    {
-      "epoch": 0.9524246420857465,
-      "grad_norm": 0.004769875202327967,
-      "learning_rate": 0.0001999995530758866,
-      "loss": 46.0,
-      "step": 12457
-    },
-    {
-      "epoch": 0.9525010990691363,
-      "grad_norm": 0.0012672748416662216,
-      "learning_rate": 0.00019999955300407143,
-      "loss": 46.0,
-      "step": 12458
-    },
-    {
-      "epoch": 0.9525775560525259,
-      "grad_norm": 0.0007318942225538194,
-      "learning_rate": 0.0001999995529322505,
-      "loss": 46.0,
-      "step": 12459
-    },
-    {
-      "epoch": 0.9526540130359157,
-      "grad_norm": 0.007978350855410099,
-      "learning_rate": 0.00019999955286042375,
-      "loss": 46.0,
-      "step": 12460
-    },
-    {
-      "epoch": 0.9527304700193054,
-      "grad_norm": 0.0011821523075923324,
-      "learning_rate": 0.0001999995527885913,
-      "loss": 46.0,
-      "step": 12461
-    },
-    {
-      "epoch": 0.9528069270026951,
-      "grad_norm": 0.0014150685165077448,
-      "learning_rate": 0.00019999955271675303,
-      "loss": 46.0,
-      "step": 12462
-    },
-    {
-      "epoch": 0.9528833839860849,
-      "grad_norm": 0.004567085299640894,
-      "learning_rate": 0.000199999552644909,
-      "loss": 46.0,
-      "step": 12463
-    },
-    {
-      "epoch": 0.9529598409694745,
-      "grad_norm": 0.00717273959890008,
-      "learning_rate": 0.00019999955257305922,
-      "loss": 46.0,
-      "step": 12464
-    },
-    {
-      "epoch": 0.9530362979528643,
-      "grad_norm": 0.0015309195732697845,
-      "learning_rate": 0.00019999955250120366,
-      "loss": 46.0,
-      "step": 12465
-    },
-    {
-      "epoch": 0.953112754936254,
-      "grad_norm": 0.01209405530244112,
-      "learning_rate": 0.0001999995524293423,
-      "loss": 46.0,
-      "step": 12466
-    },
-    {
-      "epoch": 0.9531892119196437,
-      "grad_norm": 0.0038007544353604317,
-      "learning_rate": 0.00019999955235747524,
-      "loss": 46.0,
-      "step": 12467
-    },
-    {
-      "epoch": 0.9532656689030334,
-      "grad_norm": 0.0006017562700435519,
-      "learning_rate": 0.00019999955228560236,
-      "loss": 46.0,
-      "step": 12468
-    },
-    {
-      "epoch": 0.9533421258864232,
-      "grad_norm": 0.0009332410409115255,
-      "learning_rate": 0.00019999955221372372,
-      "loss": 46.0,
-      "step": 12469
-    },
-    {
-      "epoch": 0.9534185828698128,
-      "grad_norm": 0.0012395849917083979,
-      "learning_rate": 0.00019999955214183932,
-      "loss": 46.0,
-      "step": 12470
-    },
-    {
-      "epoch": 0.9534950398532026,
-      "grad_norm": 0.0014149347553029656,
-      "learning_rate": 0.00019999955206994913,
-      "loss": 46.0,
-      "step": 12471
-    },
-    {
-      "epoch": 0.9535714968365923,
-      "grad_norm": 0.0031176290940493345,
-      "learning_rate": 0.0001999995519980532,
-      "loss": 46.0,
-      "step": 12472
-    },
-    {
-      "epoch": 0.953647953819982,
-      "grad_norm": 0.008678838610649109,
-      "learning_rate": 0.0001999995519261515,
-      "loss": 46.0,
-      "step": 12473
-    },
-    {
-      "epoch": 0.9537244108033718,
-      "grad_norm": 0.000764036609325558,
-      "learning_rate": 0.000199999551854244,
-      "loss": 46.0,
-      "step": 12474
-    },
-    {
-      "epoch": 0.9538008677867614,
-      "grad_norm": 0.0007010448607616127,
-      "learning_rate": 0.00019999955178233073,
-      "loss": 46.0,
-      "step": 12475
-    },
-    {
-      "epoch": 0.9538773247701512,
-      "grad_norm": 0.005646991543471813,
-      "learning_rate": 0.00019999955171041172,
-      "loss": 46.0,
-      "step": 12476
-    },
-    {
-      "epoch": 0.953953781753541,
-      "grad_norm": 0.0020706935320049524,
-      "learning_rate": 0.00019999955163848695,
-      "loss": 46.0,
-      "step": 12477
-    },
-    {
-      "epoch": 0.9540302387369306,
-      "grad_norm": 0.0028199441730976105,
-      "learning_rate": 0.00019999955156655637,
-      "loss": 46.0,
-      "step": 12478
-    },
-    {
-      "epoch": 0.9541066957203204,
-      "grad_norm": 0.002688672160729766,
-      "learning_rate": 0.00019999955149462,
-      "loss": 46.0,
-      "step": 12479
-    },
-    {
-      "epoch": 0.9541831527037101,
-      "grad_norm": 0.0019149164436385036,
-      "learning_rate": 0.0001999995514226779,
-      "loss": 46.0,
-      "step": 12480
-    },
-    {
-      "epoch": 0.9542596096870998,
-      "grad_norm": 0.004896718543022871,
-      "learning_rate": 0.00019999955135073004,
-      "loss": 46.0,
-      "step": 12481
-    },
-    {
-      "epoch": 0.9543360666704895,
-      "grad_norm": 0.0018319346709176898,
-      "learning_rate": 0.00019999955127877642,
-      "loss": 46.0,
-      "step": 12482
-    },
-    {
-      "epoch": 0.9544125236538792,
-      "grad_norm": 0.001564774545840919,
-      "learning_rate": 0.00019999955120681703,
-      "loss": 46.0,
-      "step": 12483
-    },
-    {
-      "epoch": 0.954488980637269,
-      "grad_norm": 0.0007748169591650367,
-      "learning_rate": 0.00019999955113485186,
-      "loss": 46.0,
-      "step": 12484
-    },
-    {
-      "epoch": 0.9545654376206587,
-      "grad_norm": 0.016518425196409225,
-      "learning_rate": 0.0001999995510628809,
-      "loss": 46.0,
-      "step": 12485
-    },
-    {
-      "epoch": 0.9546418946040484,
-      "grad_norm": 0.0025407939683645964,
-      "learning_rate": 0.00019999955099090418,
-      "loss": 46.0,
-      "step": 12486
-    },
-    {
-      "epoch": 0.9547183515874381,
-      "grad_norm": 0.0009022768936119974,
-      "learning_rate": 0.0001999995509189217,
-      "loss": 46.0,
-      "step": 12487
-    },
-    {
-      "epoch": 0.9547948085708279,
-      "grad_norm": 0.0006046746857464314,
-      "learning_rate": 0.00019999955084693344,
-      "loss": 46.0,
-      "step": 12488
-    },
-    {
-      "epoch": 0.9548712655542175,
-      "grad_norm": 0.0035914992913603783,
-      "learning_rate": 0.0001999995507749394,
-      "loss": 46.0,
-      "step": 12489
-    },
-    {
-      "epoch": 0.9549477225376073,
-      "grad_norm": 0.0009309165761806071,
-      "learning_rate": 0.0001999995507029396,
-      "loss": 46.0,
-      "step": 12490
-    },
-    {
-      "epoch": 0.955024179520997,
-      "grad_norm": 0.0028376539703458548,
-      "learning_rate": 0.00019999955063093404,
-      "loss": 46.0,
-      "step": 12491
-    },
-    {
-      "epoch": 0.9551006365043867,
-      "grad_norm": 0.0013215922517701983,
-      "learning_rate": 0.0001999995505589227,
-      "loss": 46.0,
-      "step": 12492
-    },
-    {
-      "epoch": 0.9551770934877765,
-      "grad_norm": 0.0007341204909607768,
-      "learning_rate": 0.0001999995504869056,
-      "loss": 46.0,
-      "step": 12493
-    },
-    {
-      "epoch": 0.9552535504711661,
-      "grad_norm": 0.0008842049282975495,
-      "learning_rate": 0.00019999955041488275,
-      "loss": 46.0,
-      "step": 12494
-    },
-    {
-      "epoch": 0.9553300074545559,
-      "grad_norm": 0.0005446483264677227,
-      "learning_rate": 0.0001999995503428541,
-      "loss": 46.0,
-      "step": 12495
-    },
-    {
-      "epoch": 0.9554064644379456,
-      "grad_norm": 0.004784720484167337,
-      "learning_rate": 0.00019999955027081968,
-      "loss": 46.0,
-      "step": 12496
-    },
-    {
-      "epoch": 0.9554829214213353,
-      "grad_norm": 0.0022217079531401396,
-      "learning_rate": 0.0001999995501987795,
-      "loss": 46.0,
-      "step": 12497
-    },
-    {
-      "epoch": 0.955559378404725,
-      "grad_norm": 0.0006384294247254729,
-      "learning_rate": 0.00019999955012673355,
-      "loss": 46.0,
-      "step": 12498
-    },
-    {
-      "epoch": 0.9556358353881148,
-      "grad_norm": 0.000698225514497608,
-      "learning_rate": 0.0001999995500546818,
-      "loss": 46.0,
-      "step": 12499
-    },
-    {
-      "epoch": 0.9557122923715045,
-      "grad_norm": 0.0022178853396326303,
-      "learning_rate": 0.00019999954998262432,
-      "loss": 46.0,
-      "step": 12500
-    },
-    {
-      "epoch": 0.9557887493548942,
-      "grad_norm": 0.0015768782468512654,
-      "learning_rate": 0.00019999954991056106,
-      "loss": 46.0,
-      "step": 12501
-    },
-    {
-      "epoch": 0.9558652063382839,
-      "grad_norm": 0.001325781224295497,
-      "learning_rate": 0.00019999954983849203,
-      "loss": 46.0,
-      "step": 12502
-    },
-    {
-      "epoch": 0.9559416633216736,
-      "grad_norm": 0.001395052415318787,
-      "learning_rate": 0.00019999954976641723,
-      "loss": 46.0,
-      "step": 12503
-    },
-    {
-      "epoch": 0.9560181203050634,
-      "grad_norm": 0.0008487051236443222,
-      "learning_rate": 0.00019999954969433667,
-      "loss": 46.0,
-      "step": 12504
-    },
-    {
-      "epoch": 0.956094577288453,
-      "grad_norm": 0.00044739051372744143,
-      "learning_rate": 0.00019999954962225035,
-      "loss": 46.0,
-      "step": 12505
-    },
-    {
-      "epoch": 0.9561710342718428,
-      "grad_norm": 0.0011321419151499867,
-      "learning_rate": 0.00019999954955015822,
-      "loss": 46.0,
-      "step": 12506
-    },
-    {
-      "epoch": 0.9562474912552326,
-      "grad_norm": 0.0014656601706519723,
-      "learning_rate": 0.00019999954947806035,
-      "loss": 46.0,
-      "step": 12507
-    },
-    {
-      "epoch": 0.9563239482386222,
-      "grad_norm": 0.001833436544984579,
-      "learning_rate": 0.0001999995494059567,
-      "loss": 46.0,
-      "step": 12508
-    },
-    {
-      "epoch": 0.956400405222012,
-      "grad_norm": 0.0014022770337760448,
-      "learning_rate": 0.0001999995493338473,
-      "loss": 46.0,
-      "step": 12509
-    },
-    {
-      "epoch": 0.9564768622054017,
-      "grad_norm": 0.0027344259433448315,
-      "learning_rate": 0.0001999995492617321,
-      "loss": 46.0,
-      "step": 12510
-    },
-    {
-      "epoch": 0.9565533191887914,
-      "grad_norm": 0.001271241344511509,
-      "learning_rate": 0.00019999954918961118,
-      "loss": 46.0,
-      "step": 12511
-    },
-    {
-      "epoch": 0.9566297761721811,
-      "grad_norm": 0.003318602219223976,
-      "learning_rate": 0.00019999954911748444,
-      "loss": 46.0,
-      "step": 12512
-    },
-    {
-      "epoch": 0.9567062331555708,
-      "grad_norm": 0.0015493493992835283,
-      "learning_rate": 0.00019999954904535193,
-      "loss": 46.0,
-      "step": 12513
-    },
-    {
-      "epoch": 0.9567826901389606,
-      "grad_norm": 0.0019136162009090185,
-      "learning_rate": 0.0001999995489732137,
-      "loss": 46.0,
-      "step": 12514
-    },
-    {
-      "epoch": 0.9568591471223503,
-      "grad_norm": 0.0011575878597795963,
-      "learning_rate": 0.00019999954890106966,
-      "loss": 46.0,
-      "step": 12515
-    },
-    {
-      "epoch": 0.95693560410574,
-      "grad_norm": 0.0031630985904484987,
-      "learning_rate": 0.00019999954882891986,
-      "loss": 46.0,
-      "step": 12516
-    },
-    {
-      "epoch": 0.9570120610891297,
-      "grad_norm": 0.0006764872232452035,
-      "learning_rate": 0.00019999954875676428,
-      "loss": 46.0,
-      "step": 12517
-    },
-    {
-      "epoch": 0.9570885180725195,
-      "grad_norm": 0.0009080265881493688,
-      "learning_rate": 0.00019999954868460295,
-      "loss": 46.0,
-      "step": 12518
-    },
-    {
-      "epoch": 0.9571649750559091,
-      "grad_norm": 0.0010366225615143776,
-      "learning_rate": 0.00019999954861243583,
-      "loss": 46.0,
-      "step": 12519
-    },
-    {
-      "epoch": 0.9572414320392989,
-      "grad_norm": 0.001006502308882773,
-      "learning_rate": 0.00019999954854026296,
-      "loss": 46.0,
-      "step": 12520
-    },
-    {
-      "epoch": 0.9573178890226887,
-      "grad_norm": 0.0032326679211109877,
-      "learning_rate": 0.0001999995484680843,
-      "loss": 46.0,
-      "step": 12521
-    },
-    {
-      "epoch": 0.9573943460060783,
-      "grad_norm": 0.003837198717519641,
-      "learning_rate": 0.0001999995483958999,
-      "loss": 46.0,
-      "step": 12522
-    },
-    {
-      "epoch": 0.9574708029894681,
-      "grad_norm": 0.0007604262791574001,
-      "learning_rate": 0.0001999995483237097,
-      "loss": 46.0,
-      "step": 12523
-    },
-    {
-      "epoch": 0.9575472599728577,
-      "grad_norm": 0.0005066022858954966,
-      "learning_rate": 0.00019999954825151373,
-      "loss": 46.0,
-      "step": 12524
-    },
-    {
-      "epoch": 0.9576237169562475,
-      "grad_norm": 0.002728054765611887,
-      "learning_rate": 0.00019999954817931202,
-      "loss": 46.0,
-      "step": 12525
-    },
-    {
-      "epoch": 0.9577001739396372,
-      "grad_norm": 0.0008250894607044756,
-      "learning_rate": 0.00019999954810710454,
-      "loss": 46.0,
-      "step": 12526
-    },
-    {
-      "epoch": 0.9577766309230269,
-      "grad_norm": 0.0007391558028757572,
-      "learning_rate": 0.00019999954803489125,
-      "loss": 46.0,
-      "step": 12527
-    },
-    {
-      "epoch": 0.9578530879064167,
-      "grad_norm": 0.0015902697341516614,
-      "learning_rate": 0.00019999954796267222,
-      "loss": 46.0,
-      "step": 12528
-    },
-    {
-      "epoch": 0.9579295448898064,
-      "grad_norm": 0.004454911686480045,
-      "learning_rate": 0.00019999954789044741,
-      "loss": 46.0,
-      "step": 12529
-    },
-    {
-      "epoch": 0.9580060018731961,
-      "grad_norm": 0.0015787851298227906,
-      "learning_rate": 0.00019999954781821686,
-      "loss": 46.0,
-      "step": 12530
-    },
-    {
-      "epoch": 0.9580824588565858,
-      "grad_norm": 0.004207047168165445,
-      "learning_rate": 0.0001999995477459805,
-      "loss": 46.0,
-      "step": 12531
-    },
-    {
-      "epoch": 0.9581589158399756,
-      "grad_norm": 0.003194646444171667,
-      "learning_rate": 0.0001999995476737384,
-      "loss": 46.0,
-      "step": 12532
-    },
-    {
-      "epoch": 0.9582353728233652,
-      "grad_norm": 0.0010382853215560317,
-      "learning_rate": 0.00019999954760149051,
-      "loss": 46.0,
-      "step": 12533
-    },
-    {
-      "epoch": 0.958311829806755,
-      "grad_norm": 0.0005894839996472001,
-      "learning_rate": 0.0001999995475292369,
-      "loss": 46.0,
-      "step": 12534
-    },
-    {
-      "epoch": 0.9583882867901447,
-      "grad_norm": 0.0008184568723663688,
-      "learning_rate": 0.00019999954745697745,
-      "loss": 46.0,
-      "step": 12535
-    },
-    {
-      "epoch": 0.9584647437735344,
-      "grad_norm": 0.0004331482632551342,
-      "learning_rate": 0.00019999954738471226,
-      "loss": 46.0,
-      "step": 12536
-    },
-    {
-      "epoch": 0.9585412007569242,
-      "grad_norm": 0.0005351289291866124,
-      "learning_rate": 0.0001999995473124413,
-      "loss": 46.0,
-      "step": 12537
-    },
-    {
-      "epoch": 0.9586176577403138,
-      "grad_norm": 0.0005616553826257586,
-      "learning_rate": 0.00019999954724016455,
-      "loss": 46.0,
-      "step": 12538
-    },
-    {
-      "epoch": 0.9586941147237036,
-      "grad_norm": 0.0013907364336773753,
-      "learning_rate": 0.00019999954716788207,
-      "loss": 46.0,
-      "step": 12539
-    },
-    {
-      "epoch": 0.9587705717070933,
-      "grad_norm": 0.0023821413051337004,
-      "learning_rate": 0.0001999995470955938,
-      "loss": 46.0,
-      "step": 12540
-    },
-    {
-      "epoch": 0.958847028690483,
-      "grad_norm": 0.0005421300302259624,
-      "learning_rate": 0.00019999954702329978,
-      "loss": 46.0,
-      "step": 12541
-    },
-    {
-      "epoch": 0.9589234856738728,
-      "grad_norm": 0.0013058108743280172,
-      "learning_rate": 0.00019999954695099997,
-      "loss": 46.0,
-      "step": 12542
-    },
-    {
-      "epoch": 0.9589999426572624,
-      "grad_norm": 0.0006744128768332303,
-      "learning_rate": 0.00019999954687869437,
-      "loss": 46.0,
-      "step": 12543
-    },
-    {
-      "epoch": 0.9590763996406522,
-      "grad_norm": 0.0012755459174513817,
-      "learning_rate": 0.00019999954680638307,
-      "loss": 46.0,
-      "step": 12544
-    },
-    {
-      "epoch": 0.9591528566240419,
-      "grad_norm": 0.0013100079959258437,
-      "learning_rate": 0.00019999954673406595,
-      "loss": 46.0,
-      "step": 12545
-    },
-    {
-      "epoch": 0.9592293136074316,
-      "grad_norm": 0.001862332341261208,
-      "learning_rate": 0.00019999954666174305,
-      "loss": 46.0,
-      "step": 12546
-    },
-    {
-      "epoch": 0.9593057705908213,
-      "grad_norm": 0.001624559867195785,
-      "learning_rate": 0.0001999995465894144,
-      "loss": 46.0,
-      "step": 12547
-    },
-    {
-      "epoch": 0.9593822275742111,
-      "grad_norm": 0.0007144755800254643,
-      "learning_rate": 0.00019999954651707998,
-      "loss": 46.0,
-      "step": 12548
-    },
-    {
-      "epoch": 0.9594586845576007,
-      "grad_norm": 0.0039685554802417755,
-      "learning_rate": 0.00019999954644473977,
-      "loss": 46.0,
-      "step": 12549
-    },
-    {
-      "epoch": 0.9595351415409905,
-      "grad_norm": 0.006061859428882599,
-      "learning_rate": 0.00019999954637239383,
-      "loss": 46.0,
-      "step": 12550
-    },
-    {
-      "epoch": 0.9596115985243803,
-      "grad_norm": 0.003831411711871624,
-      "learning_rate": 0.0001999995463000421,
-      "loss": 46.0,
-      "step": 12551
-    },
-    {
-      "epoch": 0.9596880555077699,
-      "grad_norm": 0.002832041122019291,
-      "learning_rate": 0.0001999995462276846,
-      "loss": 46.0,
-      "step": 12552
-    },
-    {
-      "epoch": 0.9597645124911597,
-      "grad_norm": 0.0012886578915640712,
-      "learning_rate": 0.00019999954615532132,
-      "loss": 46.0,
-      "step": 12553
-    },
-    {
-      "epoch": 0.9598409694745493,
-      "grad_norm": 0.003579822601750493,
-      "learning_rate": 0.0001999995460829523,
-      "loss": 46.0,
-      "step": 12554
-    },
-    {
-      "epoch": 0.9599174264579391,
-      "grad_norm": 0.0016573643079027534,
-      "learning_rate": 0.00019999954601057746,
-      "loss": 46.0,
-      "step": 12555
-    },
-    {
-      "epoch": 0.9599938834413289,
-      "grad_norm": 0.00741255609318614,
-      "learning_rate": 0.0001999995459381969,
-      "loss": 46.0,
-      "step": 12556
-    },
-    {
-      "epoch": 0.9600703404247185,
-      "grad_norm": 0.005340883042663336,
-      "learning_rate": 0.00019999954586581055,
-      "loss": 46.0,
-      "step": 12557
-    },
-    {
-      "epoch": 0.9601467974081083,
-      "grad_norm": 0.0013376573333516717,
-      "learning_rate": 0.00019999954579341843,
-      "loss": 46.0,
-      "step": 12558
-    },
-    {
-      "epoch": 0.960223254391498,
-      "grad_norm": 0.0012810935731977224,
-      "learning_rate": 0.00019999954572102056,
-      "loss": 46.0,
-      "step": 12559
-    },
-    {
-      "epoch": 0.9602997113748877,
-      "grad_norm": 0.006521451752632856,
-      "learning_rate": 0.0001999995456486169,
-      "loss": 46.0,
-      "step": 12560
-    },
-    {
-      "epoch": 0.9603761683582774,
-      "grad_norm": 0.001418704865500331,
-      "learning_rate": 0.00019999954557620747,
-      "loss": 46.0,
-      "step": 12561
-    },
-    {
-      "epoch": 0.9604526253416672,
-      "grad_norm": 0.006944640539586544,
-      "learning_rate": 0.00019999954550379228,
-      "loss": 46.0,
-      "step": 12562
-    },
-    {
-      "epoch": 0.9605290823250568,
-      "grad_norm": 0.0013539890060201287,
-      "learning_rate": 0.0001999995454313713,
-      "loss": 46.0,
-      "step": 12563
-    },
-    {
-      "epoch": 0.9606055393084466,
-      "grad_norm": 0.0010226204758509994,
-      "learning_rate": 0.00019999954535894458,
-      "loss": 46.0,
-      "step": 12564
-    },
-    {
-      "epoch": 0.9606819962918363,
-      "grad_norm": 0.0019486999372020364,
-      "learning_rate": 0.0001999995452865121,
-      "loss": 46.0,
-      "step": 12565
-    },
-    {
-      "epoch": 0.960758453275226,
-      "grad_norm": 0.0009200394852086902,
-      "learning_rate": 0.0001999995452140738,
-      "loss": 46.0,
-      "step": 12566
-    },
-    {
-      "epoch": 0.9608349102586158,
-      "grad_norm": 0.0016339634312316775,
-      "learning_rate": 0.00019999954514162975,
-      "loss": 46.0,
-      "step": 12567
-    },
-    {
-      "epoch": 0.9609113672420054,
-      "grad_norm": 0.005042479373514652,
-      "learning_rate": 0.00019999954506917994,
-      "loss": 46.0,
-      "step": 12568
-    },
-    {
-      "epoch": 0.9609878242253952,
-      "grad_norm": 0.007448961026966572,
-      "learning_rate": 0.00019999954499672437,
-      "loss": 46.0,
-      "step": 12569
-    },
-    {
-      "epoch": 0.961064281208785,
-      "grad_norm": 0.0008751733112148941,
-      "learning_rate": 0.00019999954492426301,
-      "loss": 46.0,
-      "step": 12570
-    },
-    {
-      "epoch": 0.9611407381921746,
-      "grad_norm": 0.004641556181013584,
-      "learning_rate": 0.00019999954485179592,
-      "loss": 46.0,
-      "step": 12571
-    },
-    {
-      "epoch": 0.9612171951755644,
-      "grad_norm": 0.012658430263400078,
-      "learning_rate": 0.00019999954477932302,
-      "loss": 46.0,
-      "step": 12572
-    },
-    {
-      "epoch": 0.961293652158954,
-      "grad_norm": 0.0006108662346377969,
-      "learning_rate": 0.00019999954470684435,
-      "loss": 46.0,
-      "step": 12573
-    },
-    {
-      "epoch": 0.9613701091423438,
-      "grad_norm": 0.0007247099420055747,
-      "learning_rate": 0.00019999954463435993,
-      "loss": 46.0,
-      "step": 12574
-    },
-    {
-      "epoch": 0.9614465661257335,
-      "grad_norm": 0.0007045581587590277,
-      "learning_rate": 0.00019999954456186974,
-      "loss": 46.0,
-      "step": 12575
-    },
-    {
-      "epoch": 0.9615230231091232,
-      "grad_norm": 0.0021125529892742634,
-      "learning_rate": 0.00019999954448937378,
-      "loss": 46.0,
-      "step": 12576
-    },
-    {
-      "epoch": 0.961599480092513,
-      "grad_norm": 0.0011173088569194078,
-      "learning_rate": 0.00019999954441687204,
-      "loss": 46.0,
-      "step": 12577
-    },
-    {
-      "epoch": 0.9616759370759027,
-      "grad_norm": 0.001079391222447157,
-      "learning_rate": 0.00019999954434436453,
-      "loss": 46.0,
-      "step": 12578
-    },
-    {
-      "epoch": 0.9617523940592924,
-      "grad_norm": 0.004714257083833218,
-      "learning_rate": 0.00019999954427185124,
-      "loss": 46.0,
-      "step": 12579
-    },
-    {
-      "epoch": 0.9618288510426821,
-      "grad_norm": 0.0009884992614388466,
-      "learning_rate": 0.0001999995441993322,
-      "loss": 46.0,
-      "step": 12580
-    },
-    {
-      "epoch": 0.9619053080260719,
-      "grad_norm": 0.0007750187651254237,
-      "learning_rate": 0.00019999954412680738,
-      "loss": 46.0,
-      "step": 12581
-    },
-    {
-      "epoch": 0.9619817650094615,
-      "grad_norm": 0.000613724288996309,
-      "learning_rate": 0.0001999995440542768,
-      "loss": 46.0,
-      "step": 12582
-    },
-    {
-      "epoch": 0.9620582219928513,
-      "grad_norm": 0.0011867601424455643,
-      "learning_rate": 0.00019999954398174045,
-      "loss": 46.0,
-      "step": 12583
-    },
-    {
-      "epoch": 0.9621346789762409,
-      "grad_norm": 0.005172676872462034,
-      "learning_rate": 0.00019999954390919835,
-      "loss": 46.0,
-      "step": 12584
-    },
-    {
-      "epoch": 0.9622111359596307,
-      "grad_norm": 0.003148970426991582,
-      "learning_rate": 0.00019999954383665046,
-      "loss": 46.0,
-      "step": 12585
-    },
-    {
-      "epoch": 0.9622875929430205,
-      "grad_norm": 0.0015872607473284006,
-      "learning_rate": 0.00019999954376409679,
-      "loss": 46.0,
-      "step": 12586
-    },
-    {
-      "epoch": 0.9623640499264101,
-      "grad_norm": 0.0004961260710842907,
-      "learning_rate": 0.00019999954369153734,
-      "loss": 46.0,
-      "step": 12587
-    },
-    {
-      "epoch": 0.9624405069097999,
-      "grad_norm": 0.0038851506542414427,
-      "learning_rate": 0.00019999954361897215,
-      "loss": 46.0,
-      "step": 12588
-    },
-    {
-      "epoch": 0.9625169638931896,
-      "grad_norm": 0.0035033011808991432,
-      "learning_rate": 0.0001999995435464012,
-      "loss": 46.0,
-      "step": 12589
-    },
-    {
-      "epoch": 0.9625934208765793,
-      "grad_norm": 0.0013847750378772616,
-      "learning_rate": 0.00019999954347382442,
-      "loss": 46.0,
-      "step": 12590
-    },
-    {
-      "epoch": 0.962669877859969,
-      "grad_norm": 0.000358601741027087,
-      "learning_rate": 0.00019999954340124194,
-      "loss": 46.0,
-      "step": 12591
-    },
-    {
-      "epoch": 0.9627463348433588,
-      "grad_norm": 0.0013354334514588118,
-      "learning_rate": 0.00019999954332865368,
-      "loss": 46.0,
-      "step": 12592
-    },
-    {
-      "epoch": 0.9628227918267485,
-      "grad_norm": 0.0005576562252826989,
-      "learning_rate": 0.0001999995432560596,
-      "loss": 46.0,
-      "step": 12593
-    },
-    {
-      "epoch": 0.9628992488101382,
-      "grad_norm": 0.014594055712223053,
-      "learning_rate": 0.00019999954318345982,
-      "loss": 46.0,
-      "step": 12594
-    },
-    {
-      "epoch": 0.9629757057935279,
-      "grad_norm": 0.0014263063203543425,
-      "learning_rate": 0.00019999954311085422,
-      "loss": 46.0,
-      "step": 12595
-    },
-    {
-      "epoch": 0.9630521627769176,
-      "grad_norm": 0.0002965877647511661,
-      "learning_rate": 0.00019999954303824287,
-      "loss": 46.0,
-      "step": 12596
-    },
-    {
-      "epoch": 0.9631286197603074,
-      "grad_norm": 0.0016289090272039175,
-      "learning_rate": 0.00019999954296562574,
-      "loss": 46.0,
-      "step": 12597
-    },
-    {
-      "epoch": 0.963205076743697,
-      "grad_norm": 0.0012613135622814298,
-      "learning_rate": 0.00019999954289300284,
-      "loss": 46.0,
-      "step": 12598
-    },
-    {
-      "epoch": 0.9632815337270868,
-      "grad_norm": 0.0008691719849593937,
-      "learning_rate": 0.00019999954282037417,
-      "loss": 46.0,
-      "step": 12599
-    },
-    {
-      "epoch": 0.9633579907104766,
-      "grad_norm": 0.00310651701875031,
-      "learning_rate": 0.00019999954274773973,
-      "loss": 46.0,
-      "step": 12600
-    },
-    {
-      "epoch": 0.9634344476938662,
-      "grad_norm": 0.000629986054264009,
-      "learning_rate": 0.00019999954267509954,
-      "loss": 46.0,
-      "step": 12601
-    },
-    {
-      "epoch": 0.963510904677256,
-      "grad_norm": 0.0020488090813159943,
-      "learning_rate": 0.00019999954260245358,
-      "loss": 46.0,
-      "step": 12602
-    },
-    {
-      "epoch": 0.9635873616606456,
-      "grad_norm": 0.0006753480993211269,
-      "learning_rate": 0.00019999954252980184,
-      "loss": 46.0,
-      "step": 12603
-    },
-    {
-      "epoch": 0.9636638186440354,
-      "grad_norm": 0.0006479251314885914,
-      "learning_rate": 0.00019999954245714433,
-      "loss": 46.0,
-      "step": 12604
-    },
-    {
-      "epoch": 0.9637402756274251,
-      "grad_norm": 0.0009644933161325753,
-      "learning_rate": 0.00019999954238448104,
-      "loss": 46.0,
-      "step": 12605
-    },
-    {
-      "epoch": 0.9638167326108148,
-      "grad_norm": 0.0007472362485714257,
-      "learning_rate": 0.000199999542311812,
-      "loss": 46.0,
-      "step": 12606
-    },
-    {
-      "epoch": 0.9638931895942046,
-      "grad_norm": 0.0012204254744574428,
-      "learning_rate": 0.00019999954223913718,
-      "loss": 46.0,
-      "step": 12607
-    },
-    {
-      "epoch": 0.9639696465775943,
-      "grad_norm": 0.001085710246115923,
-      "learning_rate": 0.00019999954216645658,
-      "loss": 46.0,
-      "step": 12608
-    },
-    {
-      "epoch": 0.964046103560984,
-      "grad_norm": 0.0029610525816679,
-      "learning_rate": 0.00019999954209377023,
-      "loss": 46.0,
-      "step": 12609
-    },
-    {
-      "epoch": 0.9641225605443737,
-      "grad_norm": 0.0007604038692079484,
-      "learning_rate": 0.0001999995420210781,
-      "loss": 46.0,
-      "step": 12610
-    },
-    {
-      "epoch": 0.9641990175277635,
-      "grad_norm": 0.0009824297158047557,
-      "learning_rate": 0.0001999995419483802,
-      "loss": 46.0,
-      "step": 12611
-    },
-    {
-      "epoch": 0.9642754745111531,
-      "grad_norm": 0.0010285725584253669,
-      "learning_rate": 0.00019999954187567657,
-      "loss": 46.0,
-      "step": 12612
-    },
-    {
-      "epoch": 0.9643519314945429,
-      "grad_norm": 0.01939595490694046,
-      "learning_rate": 0.00019999954180296712,
-      "loss": 46.0,
-      "step": 12613
-    },
-    {
-      "epoch": 0.9644283884779326,
-      "grad_norm": 0.008662334643304348,
-      "learning_rate": 0.0001999995417302519,
-      "loss": 46.0,
-      "step": 12614
-    },
-    {
-      "epoch": 0.9645048454613223,
-      "grad_norm": 0.00046927505172789097,
-      "learning_rate": 0.00019999954165753091,
-      "loss": 46.0,
-      "step": 12615
-    },
-    {
-      "epoch": 0.9645813024447121,
-      "grad_norm": 0.004104883410036564,
-      "learning_rate": 0.00019999954158480418,
-      "loss": 46.0,
-      "step": 12616
-    },
-    {
-      "epoch": 0.9646577594281017,
-      "grad_norm": 0.0008808405254967511,
-      "learning_rate": 0.00019999954151207167,
-      "loss": 46.0,
-      "step": 12617
-    },
-    {
-      "epoch": 0.9647342164114915,
-      "grad_norm": 0.0020914131309837103,
-      "learning_rate": 0.00019999954143933338,
-      "loss": 46.0,
-      "step": 12618
-    },
-    {
-      "epoch": 0.9648106733948812,
-      "grad_norm": 0.0008840143564157188,
-      "learning_rate": 0.00019999954136658933,
-      "loss": 46.0,
-      "step": 12619
-    },
-    {
-      "epoch": 0.9648871303782709,
-      "grad_norm": 0.0013798632426187396,
-      "learning_rate": 0.00019999954129383952,
-      "loss": 46.0,
-      "step": 12620
-    },
-    {
-      "epoch": 0.9649635873616607,
-      "grad_norm": 0.0010010011028498411,
-      "learning_rate": 0.00019999954122108392,
-      "loss": 46.0,
-      "step": 12621
-    },
-    {
-      "epoch": 0.9650400443450504,
-      "grad_norm": 0.0016264660516753793,
-      "learning_rate": 0.00019999954114832254,
-      "loss": 46.0,
-      "step": 12622
-    },
-    {
-      "epoch": 0.9651165013284401,
-      "grad_norm": 0.0007057492621243,
-      "learning_rate": 0.00019999954107555542,
-      "loss": 46.0,
-      "step": 12623
-    },
-    {
-      "epoch": 0.9651929583118298,
-      "grad_norm": 0.00040417982381768525,
-      "learning_rate": 0.00019999954100278252,
-      "loss": 46.0,
-      "step": 12624
-    },
-    {
-      "epoch": 0.9652694152952195,
-      "grad_norm": 0.0018019978888332844,
-      "learning_rate": 0.00019999954093000385,
-      "loss": 46.0,
-      "step": 12625
-    },
-    {
-      "epoch": 0.9653458722786092,
-      "grad_norm": 0.0008747209794819355,
-      "learning_rate": 0.00019999954085721944,
-      "loss": 46.0,
-      "step": 12626
-    },
-    {
-      "epoch": 0.965422329261999,
-      "grad_norm": 0.0005741056520491838,
-      "learning_rate": 0.00019999954078442925,
-      "loss": 46.0,
-      "step": 12627
-    },
-    {
-      "epoch": 0.9654987862453887,
-      "grad_norm": 0.002164460252970457,
-      "learning_rate": 0.00019999954071163326,
-      "loss": 46.0,
-      "step": 12628
-    },
-    {
-      "epoch": 0.9655752432287784,
-      "grad_norm": 0.0007059042691253126,
-      "learning_rate": 0.0001999995406388315,
-      "loss": 46.0,
-      "step": 12629
-    },
-    {
-      "epoch": 0.9656517002121682,
-      "grad_norm": 0.0008277862798422575,
-      "learning_rate": 0.00019999954056602398,
-      "loss": 46.0,
-      "step": 12630
-    },
-    {
-      "epoch": 0.9657281571955578,
-      "grad_norm": 0.00032410197309218347,
-      "learning_rate": 0.0001999995404932107,
-      "loss": 46.0,
-      "step": 12631
-    },
-    {
-      "epoch": 0.9658046141789476,
-      "grad_norm": 0.0015394572401419282,
-      "learning_rate": 0.00019999954042039164,
-      "loss": 46.0,
-      "step": 12632
-    },
-    {
-      "epoch": 0.9658810711623372,
-      "grad_norm": 0.0048556337133049965,
-      "learning_rate": 0.00019999954034756681,
-      "loss": 46.0,
-      "step": 12633
-    },
-    {
-      "epoch": 0.965957528145727,
-      "grad_norm": 0.0011588882189244032,
-      "learning_rate": 0.00019999954027473624,
-      "loss": 46.0,
-      "step": 12634
-    },
-    {
-      "epoch": 0.9660339851291168,
-      "grad_norm": 0.0005232239491306245,
-      "learning_rate": 0.0001999995402018999,
-      "loss": 46.0,
-      "step": 12635
-    },
-    {
-      "epoch": 0.9661104421125064,
-      "grad_norm": 0.004094188567250967,
-      "learning_rate": 0.00019999954012905774,
-      "loss": 46.0,
-      "step": 12636
-    },
-    {
-      "epoch": 0.9661868990958962,
-      "grad_norm": 0.004282247740775347,
-      "learning_rate": 0.00019999954005620984,
-      "loss": 46.0,
-      "step": 12637
-    },
-    {
-      "epoch": 0.9662633560792859,
-      "grad_norm": 0.001113204751163721,
-      "learning_rate": 0.00019999953998335617,
-      "loss": 46.0,
-      "step": 12638
-    },
-    {
-      "epoch": 0.9663398130626756,
-      "grad_norm": 0.0012121719773858786,
-      "learning_rate": 0.0001999995399104967,
-      "loss": 46.0,
-      "step": 12639
-    },
-    {
-      "epoch": 0.9664162700460653,
-      "grad_norm": 0.0010293213417753577,
-      "learning_rate": 0.00019999953983763154,
-      "loss": 46.0,
-      "step": 12640
-    },
-    {
-      "epoch": 0.9664927270294551,
-      "grad_norm": 0.0015262255910784006,
-      "learning_rate": 0.00019999953976476052,
-      "loss": 46.0,
-      "step": 12641
-    },
-    {
-      "epoch": 0.9665691840128448,
-      "grad_norm": 0.0011345866369083524,
-      "learning_rate": 0.0001999995396918838,
-      "loss": 46.0,
-      "step": 12642
-    },
-    {
-      "epoch": 0.9666456409962345,
-      "grad_norm": 0.0023789769038558006,
-      "learning_rate": 0.00019999953961900125,
-      "loss": 46.0,
-      "step": 12643
-    },
-    {
-      "epoch": 0.9667220979796242,
-      "grad_norm": 0.00601214962080121,
-      "learning_rate": 0.000199999539546113,
-      "loss": 46.0,
-      "step": 12644
-    },
-    {
-      "epoch": 0.9667985549630139,
-      "grad_norm": 0.0010593831539154053,
-      "learning_rate": 0.00019999953947321894,
-      "loss": 46.0,
-      "step": 12645
-    },
-    {
-      "epoch": 0.9668750119464037,
-      "grad_norm": 0.001333468477241695,
-      "learning_rate": 0.0001999995394003191,
-      "loss": 46.0,
-      "step": 12646
-    },
-    {
-      "epoch": 0.9669514689297933,
-      "grad_norm": 0.0004939683130942285,
-      "learning_rate": 0.0001999995393274135,
-      "loss": 46.0,
-      "step": 12647
-    },
-    {
-      "epoch": 0.9670279259131831,
-      "grad_norm": 0.003756856080144644,
-      "learning_rate": 0.00019999953925450216,
-      "loss": 46.0,
-      "step": 12648
-    },
-    {
-      "epoch": 0.9671043828965729,
-      "grad_norm": 0.000527061230968684,
-      "learning_rate": 0.000199999539181585,
-      "loss": 46.0,
-      "step": 12649
-    },
-    {
-      "epoch": 0.9671808398799625,
-      "grad_norm": 0.0013170158490538597,
-      "learning_rate": 0.0001999995391086621,
-      "loss": 46.0,
-      "step": 12650
-    },
-    {
-      "epoch": 0.9672572968633523,
-      "grad_norm": 0.0006425055908039212,
-      "learning_rate": 0.00019999953903573344,
-      "loss": 46.0,
-      "step": 12651
-    },
-    {
-      "epoch": 0.967333753846742,
-      "grad_norm": 0.0015387942548841238,
-      "learning_rate": 0.00019999953896279897,
-      "loss": 46.0,
-      "step": 12652
-    },
-    {
-      "epoch": 0.9674102108301317,
-      "grad_norm": 0.0023090660106390715,
-      "learning_rate": 0.00019999953888985878,
-      "loss": 46.0,
-      "step": 12653
-    },
-    {
-      "epoch": 0.9674866678135214,
-      "grad_norm": 0.0005368542042560875,
-      "learning_rate": 0.00019999953881691277,
-      "loss": 46.0,
-      "step": 12654
-    },
-    {
-      "epoch": 0.9675631247969111,
-      "grad_norm": 0.0006658294587396085,
-      "learning_rate": 0.00019999953874396106,
-      "loss": 46.0,
-      "step": 12655
-    },
-    {
-      "epoch": 0.9676395817803009,
-      "grad_norm": 0.006444654893130064,
-      "learning_rate": 0.00019999953867100355,
-      "loss": 46.0,
-      "step": 12656
-    },
-    {
-      "epoch": 0.9677160387636906,
-      "grad_norm": 0.000931173621211201,
-      "learning_rate": 0.00019999953859804024,
-      "loss": 46.0,
-      "step": 12657
-    },
-    {
-      "epoch": 0.9677924957470803,
-      "grad_norm": 0.0013718197587877512,
-      "learning_rate": 0.00019999953852507116,
-      "loss": 46.0,
-      "step": 12658
-    },
-    {
-      "epoch": 0.96786895273047,
-      "grad_norm": 0.0008891511242836714,
-      "learning_rate": 0.00019999953845209633,
-      "loss": 46.0,
-      "step": 12659
-    },
-    {
-      "epoch": 0.9679454097138598,
-      "grad_norm": 0.0009287648135796189,
-      "learning_rate": 0.00019999953837911572,
-      "loss": 46.0,
-      "step": 12660
-    },
-    {
-      "epoch": 0.9680218666972494,
-      "grad_norm": 0.0005668742815032601,
-      "learning_rate": 0.00019999953830612937,
-      "loss": 46.0,
-      "step": 12661
-    },
-    {
-      "epoch": 0.9680983236806392,
-      "grad_norm": 0.0014932118356227875,
-      "learning_rate": 0.00019999953823313725,
-      "loss": 46.0,
-      "step": 12662
-    },
-    {
-      "epoch": 0.968174780664029,
-      "grad_norm": 0.001959705725312233,
-      "learning_rate": 0.00019999953816013936,
-      "loss": 46.0,
-      "step": 12663
-    },
-    {
-      "epoch": 0.9682512376474186,
-      "grad_norm": 0.0032820964697748423,
-      "learning_rate": 0.00019999953808713566,
-      "loss": 46.0,
-      "step": 12664
-    },
-    {
-      "epoch": 0.9683276946308084,
-      "grad_norm": 0.0005027010920457542,
-      "learning_rate": 0.00019999953801412622,
-      "loss": 46.0,
-      "step": 12665
-    },
-    {
-      "epoch": 0.968404151614198,
-      "grad_norm": 0.000707439670804888,
-      "learning_rate": 0.000199999537941111,
-      "loss": 46.0,
-      "step": 12666
-    },
-    {
-      "epoch": 0.9684806085975878,
-      "grad_norm": 0.0026558099780231714,
-      "learning_rate": 0.00019999953786809,
-      "loss": 46.0,
-      "step": 12667
-    },
-    {
-      "epoch": 0.9685570655809775,
-      "grad_norm": 0.002453891560435295,
-      "learning_rate": 0.00019999953779506325,
-      "loss": 46.0,
-      "step": 12668
-    },
-    {
-      "epoch": 0.9686335225643672,
-      "grad_norm": 0.001021756324917078,
-      "learning_rate": 0.00019999953772203071,
-      "loss": 46.0,
-      "step": 12669
-    },
-    {
-      "epoch": 0.968709979547757,
-      "grad_norm": 0.0033573920372873545,
-      "learning_rate": 0.00019999953764899243,
-      "loss": 46.0,
-      "step": 12670
-    },
-    {
-      "epoch": 0.9687864365311467,
-      "grad_norm": 0.006432279944419861,
-      "learning_rate": 0.00019999953757594838,
-      "loss": 46.0,
-      "step": 12671
-    },
-    {
-      "epoch": 0.9688628935145364,
-      "grad_norm": 0.0007216979865916073,
-      "learning_rate": 0.00019999953750289855,
-      "loss": 46.0,
-      "step": 12672
-    },
-    {
-      "epoch": 0.9689393504979261,
-      "grad_norm": 0.004251041915267706,
-      "learning_rate": 0.00019999953742984295,
-      "loss": 46.0,
-      "step": 12673
-    },
-    {
-      "epoch": 0.9690158074813158,
-      "grad_norm": 0.0021378234960138798,
-      "learning_rate": 0.0001999995373567816,
-      "loss": 46.0,
-      "step": 12674
-    },
-    {
-      "epoch": 0.9690922644647055,
-      "grad_norm": 0.0002845959970727563,
-      "learning_rate": 0.00019999953728371445,
-      "loss": 46.0,
-      "step": 12675
-    },
-    {
-      "epoch": 0.9691687214480953,
-      "grad_norm": 0.0016557008493691683,
-      "learning_rate": 0.00019999953721064153,
-      "loss": 46.0,
-      "step": 12676
-    },
-    {
-      "epoch": 0.9692451784314849,
-      "grad_norm": 0.0042579397559165955,
-      "learning_rate": 0.00019999953713756286,
-      "loss": 46.0,
-      "step": 12677
-    },
-    {
-      "epoch": 0.9693216354148747,
-      "grad_norm": 0.0010371737880632281,
-      "learning_rate": 0.00019999953706447841,
-      "loss": 46.0,
-      "step": 12678
-    },
-    {
-      "epoch": 0.9693980923982645,
-      "grad_norm": 0.00591895030811429,
-      "learning_rate": 0.0001999995369913882,
-      "loss": 46.0,
-      "step": 12679
-    },
-    {
-      "epoch": 0.9694745493816541,
-      "grad_norm": 0.00045719597255811095,
-      "learning_rate": 0.0001999995369182922,
-      "loss": 46.0,
-      "step": 12680
-    },
-    {
-      "epoch": 0.9695510063650439,
-      "grad_norm": 0.002347303321585059,
-      "learning_rate": 0.00019999953684519045,
-      "loss": 46.0,
-      "step": 12681
-    },
-    {
-      "epoch": 0.9696274633484336,
-      "grad_norm": 0.003925171680748463,
-      "learning_rate": 0.0001999995367720829,
-      "loss": 46.0,
-      "step": 12682
-    },
-    {
-      "epoch": 0.9697039203318233,
-      "grad_norm": 0.0010088527342304587,
-      "learning_rate": 0.00019999953669896963,
-      "loss": 46.0,
-      "step": 12683
-    },
-    {
-      "epoch": 0.969780377315213,
-      "grad_norm": 0.005125761032104492,
-      "learning_rate": 0.00019999953662585055,
-      "loss": 46.0,
-      "step": 12684
-    },
-    {
-      "epoch": 0.9698568342986027,
-      "grad_norm": 0.002419125521555543,
-      "learning_rate": 0.00019999953655272572,
-      "loss": 46.0,
-      "step": 12685
-    },
-    {
-      "epoch": 0.9699332912819925,
-      "grad_norm": 0.0014551643980666995,
-      "learning_rate": 0.00019999953647959512,
-      "loss": 46.0,
-      "step": 12686
-    },
-    {
-      "epoch": 0.9700097482653822,
-      "grad_norm": 0.0025780065916478634,
-      "learning_rate": 0.00019999953640645877,
-      "loss": 46.0,
-      "step": 12687
-    },
-    {
-      "epoch": 0.9700862052487719,
-      "grad_norm": 0.0015552312834188342,
-      "learning_rate": 0.00019999953633331662,
-      "loss": 46.0,
-      "step": 12688
-    },
-    {
-      "epoch": 0.9701626622321616,
-      "grad_norm": 0.0022849254310131073,
-      "learning_rate": 0.0001999995362601687,
-      "loss": 46.0,
-      "step": 12689
-    },
-    {
-      "epoch": 0.9702391192155514,
-      "grad_norm": 0.0013499159831553698,
-      "learning_rate": 0.000199999536187015,
-      "loss": 46.0,
-      "step": 12690
-    },
-    {
-      "epoch": 0.970315576198941,
-      "grad_norm": 0.0007311037043109536,
-      "learning_rate": 0.0001999995361138556,
-      "loss": 46.0,
-      "step": 12691
-    },
-    {
-      "epoch": 0.9703920331823308,
-      "grad_norm": 0.01081959716975689,
-      "learning_rate": 0.00019999953604069034,
-      "loss": 46.0,
-      "step": 12692
-    },
-    {
-      "epoch": 0.9704684901657206,
-      "grad_norm": 0.0008478965610265732,
-      "learning_rate": 0.00019999953596751936,
-      "loss": 46.0,
-      "step": 12693
-    },
-    {
-      "epoch": 0.9705449471491102,
-      "grad_norm": 0.003631509840488434,
-      "learning_rate": 0.0001999995358943426,
-      "loss": 46.0,
-      "step": 12694
-    },
-    {
-      "epoch": 0.9706214041325,
-      "grad_norm": 0.005794324912130833,
-      "learning_rate": 0.00019999953582116006,
-      "loss": 46.0,
-      "step": 12695
-    },
-    {
-      "epoch": 0.9706978611158896,
-      "grad_norm": 0.0017354452284052968,
-      "learning_rate": 0.00019999953574797178,
-      "loss": 46.0,
-      "step": 12696
-    },
-    {
-      "epoch": 0.9707743180992794,
-      "grad_norm": 0.001040946808643639,
-      "learning_rate": 0.0001999995356747777,
-      "loss": 46.0,
-      "step": 12697
-    },
-    {
-      "epoch": 0.9708507750826691,
-      "grad_norm": 0.0013337313430383801,
-      "learning_rate": 0.00019999953560157787,
-      "loss": 46.0,
-      "step": 12698
-    },
-    {
-      "epoch": 0.9709272320660588,
-      "grad_norm": 0.0008756202296353877,
-      "learning_rate": 0.00019999953552837227,
-      "loss": 46.0,
-      "step": 12699
-    },
-    {
-      "epoch": 0.9710036890494486,
-      "grad_norm": 0.0015684895915910602,
-      "learning_rate": 0.0001999995354551609,
-      "loss": 46.0,
-      "step": 12700
-    },
-    {
-      "epoch": 0.9710801460328383,
-      "grad_norm": 0.0011862721294164658,
-      "learning_rate": 0.00019999953538194377,
-      "loss": 46.0,
-      "step": 12701
-    },
-    {
-      "epoch": 0.971156603016228,
-      "grad_norm": 0.007497219834476709,
-      "learning_rate": 0.00019999953530872082,
-      "loss": 46.0,
-      "step": 12702
-    },
-    {
-      "epoch": 0.9712330599996177,
-      "grad_norm": 0.0016313801752403378,
-      "learning_rate": 0.00019999953523549215,
-      "loss": 46.0,
-      "step": 12703
-    },
-    {
-      "epoch": 0.9713095169830074,
-      "grad_norm": 0.003111917292699218,
-      "learning_rate": 0.00019999953516225768,
-      "loss": 46.0,
-      "step": 12704
-    },
-    {
-      "epoch": 0.9713859739663971,
-      "grad_norm": 0.0006332274642772973,
-      "learning_rate": 0.00019999953508901747,
-      "loss": 46.0,
-      "step": 12705
-    },
-    {
-      "epoch": 0.9714624309497869,
-      "grad_norm": 0.0022243268322199583,
-      "learning_rate": 0.00019999953501577148,
-      "loss": 46.0,
-      "step": 12706
-    },
-    {
-      "epoch": 0.9715388879331766,
-      "grad_norm": 0.008653603494167328,
-      "learning_rate": 0.00019999953494251972,
-      "loss": 46.0,
-      "step": 12707
-    },
-    {
-      "epoch": 0.9716153449165663,
-      "grad_norm": 0.0011655132984742522,
-      "learning_rate": 0.00019999953486926218,
-      "loss": 46.0,
-      "step": 12708
-    },
-    {
-      "epoch": 0.9716918018999561,
-      "grad_norm": 0.002224584110081196,
-      "learning_rate": 0.0001999995347959989,
-      "loss": 46.0,
-      "step": 12709
-    },
-    {
-      "epoch": 0.9717682588833457,
-      "grad_norm": 0.0027827557642012835,
-      "learning_rate": 0.00019999953472272982,
-      "loss": 46.0,
-      "step": 12710
-    },
-    {
-      "epoch": 0.9718447158667355,
-      "grad_norm": 0.002156534232199192,
-      "learning_rate": 0.00019999953464945496,
-      "loss": 46.0,
-      "step": 12711
-    },
-    {
-      "epoch": 0.9719211728501252,
-      "grad_norm": 0.0018375476356595755,
-      "learning_rate": 0.00019999953457617436,
-      "loss": 46.0,
-      "step": 12712
-    },
-    {
-      "epoch": 0.9719976298335149,
-      "grad_norm": 0.001420433516614139,
-      "learning_rate": 0.000199999534502888,
-      "loss": 46.0,
-      "step": 12713
-    },
-    {
-      "epoch": 0.9720740868169047,
-      "grad_norm": 0.001599422306753695,
-      "learning_rate": 0.00019999953442959584,
-      "loss": 46.0,
-      "step": 12714
-    },
-    {
-      "epoch": 0.9721505438002943,
-      "grad_norm": 0.0008550670463591814,
-      "learning_rate": 0.00019999953435629792,
-      "loss": 46.0,
-      "step": 12715
-    },
-    {
-      "epoch": 0.9722270007836841,
-      "grad_norm": 0.0008036239305511117,
-      "learning_rate": 0.00019999953428299425,
-      "loss": 46.0,
-      "step": 12716
-    },
-    {
-      "epoch": 0.9723034577670738,
-      "grad_norm": 0.0008184924954548478,
-      "learning_rate": 0.0001999995342096848,
-      "loss": 46.0,
-      "step": 12717
-    },
-    {
-      "epoch": 0.9723799147504635,
-      "grad_norm": 0.0019802553579211235,
-      "learning_rate": 0.00019999953413636957,
-      "loss": 46.0,
-      "step": 12718
-    },
-    {
-      "epoch": 0.9724563717338532,
-      "grad_norm": 0.0005797239718958735,
-      "learning_rate": 0.00019999953406304855,
-      "loss": 46.0,
-      "step": 12719
-    },
-    {
-      "epoch": 0.972532828717243,
-      "grad_norm": 0.0009108522208407521,
-      "learning_rate": 0.00019999953398972182,
-      "loss": 46.0,
-      "step": 12720
-    },
-    {
-      "epoch": 0.9726092857006327,
-      "grad_norm": 0.002286546630784869,
-      "learning_rate": 0.00019999953391638928,
-      "loss": 46.0,
-      "step": 12721
-    },
-    {
-      "epoch": 0.9726857426840224,
-      "grad_norm": 0.001663442817516625,
-      "learning_rate": 0.00019999953384305098,
-      "loss": 46.0,
-      "step": 12722
-    },
-    {
-      "epoch": 0.9727621996674122,
-      "grad_norm": 0.0009871290531009436,
-      "learning_rate": 0.0001999995337697069,
-      "loss": 46.0,
-      "step": 12723
-    },
-    {
-      "epoch": 0.9728386566508018,
-      "grad_norm": 0.0026840870268642902,
-      "learning_rate": 0.00019999953369635707,
-      "loss": 46.0,
-      "step": 12724
-    },
-    {
-      "epoch": 0.9729151136341916,
-      "grad_norm": 0.0010838746093213558,
-      "learning_rate": 0.00019999953362300144,
-      "loss": 46.0,
-      "step": 12725
-    },
-    {
-      "epoch": 0.9729915706175812,
-      "grad_norm": 0.0013851015828549862,
-      "learning_rate": 0.00019999953354964006,
-      "loss": 46.0,
-      "step": 12726
-    },
-    {
-      "epoch": 0.973068027600971,
-      "grad_norm": 0.0010990941664204001,
-      "learning_rate": 0.00019999953347627292,
-      "loss": 46.0,
-      "step": 12727
-    },
-    {
-      "epoch": 0.9731444845843608,
-      "grad_norm": 0.0007844744250178337,
-      "learning_rate": 0.0001999995334029,
-      "loss": 46.0,
-      "step": 12728
-    },
-    {
-      "epoch": 0.9732209415677504,
-      "grad_norm": 0.0009142103954218328,
-      "learning_rate": 0.0001999995333295213,
-      "loss": 46.0,
-      "step": 12729
-    },
-    {
-      "epoch": 0.9732973985511402,
-      "grad_norm": 0.0035712618846446276,
-      "learning_rate": 0.00019999953325613686,
-      "loss": 46.0,
-      "step": 12730
-    },
-    {
-      "epoch": 0.9733738555345299,
-      "grad_norm": 0.002294503152370453,
-      "learning_rate": 0.00019999953318274662,
-      "loss": 46.0,
-      "step": 12731
-    },
-    {
-      "epoch": 0.9734503125179196,
-      "grad_norm": 0.007922537624835968,
-      "learning_rate": 0.00019999953310935063,
-      "loss": 46.0,
-      "step": 12732
-    },
-    {
-      "epoch": 0.9735267695013093,
-      "grad_norm": 0.0007058531045913696,
-      "learning_rate": 0.00019999953303594884,
-      "loss": 46.0,
-      "step": 12733
-    },
-    {
-      "epoch": 0.973603226484699,
-      "grad_norm": 0.005658953450620174,
-      "learning_rate": 0.00019999953296254134,
-      "loss": 46.0,
-      "step": 12734
-    },
-    {
-      "epoch": 0.9736796834680888,
-      "grad_norm": 0.0011385917896404862,
-      "learning_rate": 0.00019999953288912803,
-      "loss": 46.0,
-      "step": 12735
-    },
-    {
-      "epoch": 0.9737561404514785,
-      "grad_norm": 0.0032692786771804094,
-      "learning_rate": 0.00019999953281570895,
-      "loss": 46.0,
-      "step": 12736
-    },
-    {
-      "epoch": 0.9738325974348682,
-      "grad_norm": 0.0006996606243774295,
-      "learning_rate": 0.00019999953274228412,
-      "loss": 46.0,
-      "step": 12737
-    },
-    {
-      "epoch": 0.9739090544182579,
-      "grad_norm": 0.0013061792124062777,
-      "learning_rate": 0.00019999953266885346,
-      "loss": 46.0,
-      "step": 12738
-    },
-    {
-      "epoch": 0.9739855114016477,
-      "grad_norm": 0.000654165749438107,
-      "learning_rate": 0.00019999953259541712,
-      "loss": 46.0,
-      "step": 12739
-    },
-    {
-      "epoch": 0.9740619683850373,
-      "grad_norm": 0.0006127366214059293,
-      "learning_rate": 0.00019999953252197497,
-      "loss": 46.0,
-      "step": 12740
-    },
-    {
-      "epoch": 0.9741384253684271,
-      "grad_norm": 0.0013876823941245675,
-      "learning_rate": 0.00019999953244852702,
-      "loss": 46.0,
-      "step": 12741
-    },
-    {
-      "epoch": 0.9742148823518169,
-      "grad_norm": 0.00219386862590909,
-      "learning_rate": 0.00019999953237507333,
-      "loss": 46.0,
-      "step": 12742
-    },
-    {
-      "epoch": 0.9742913393352065,
-      "grad_norm": 0.0007976947817951441,
-      "learning_rate": 0.0001999995323016139,
-      "loss": 46.0,
-      "step": 12743
-    },
-    {
-      "epoch": 0.9743677963185963,
-      "grad_norm": 0.0011126977624371648,
-      "learning_rate": 0.00019999953222814865,
-      "loss": 46.0,
-      "step": 12744
-    },
-    {
-      "epoch": 0.9744442533019859,
-      "grad_norm": 0.001341003691777587,
-      "learning_rate": 0.00019999953215467766,
-      "loss": 46.0,
-      "step": 12745
-    },
-    {
-      "epoch": 0.9745207102853757,
-      "grad_norm": 0.0020163992885500193,
-      "learning_rate": 0.00019999953208120087,
-      "loss": 46.0,
-      "step": 12746
-    },
-    {
-      "epoch": 0.9745971672687654,
-      "grad_norm": 0.0019039264880120754,
-      "learning_rate": 0.00019999953200771834,
-      "loss": 46.0,
-      "step": 12747
-    },
-    {
-      "epoch": 0.9746736242521551,
-      "grad_norm": 0.0010318623390048742,
-      "learning_rate": 0.00019999953193423006,
-      "loss": 46.0,
-      "step": 12748
-    },
-    {
-      "epoch": 0.9747500812355449,
-      "grad_norm": 0.0017280089668929577,
-      "learning_rate": 0.00019999953186073595,
-      "loss": 46.0,
-      "step": 12749
-    },
-    {
-      "epoch": 0.9748265382189346,
-      "grad_norm": 0.0007378591108135879,
-      "learning_rate": 0.00019999953178723612,
-      "loss": 46.0,
-      "step": 12750
-    },
-    {
-      "epoch": 0.9749029952023243,
-      "grad_norm": 0.0013012961717322469,
-      "learning_rate": 0.0001999995317137305,
-      "loss": 46.0,
-      "step": 12751
-    },
-    {
-      "epoch": 0.974979452185714,
-      "grad_norm": 0.0008283415227197111,
-      "learning_rate": 0.0001999995316402191,
-      "loss": 46.0,
-      "step": 12752
-    },
-    {
-      "epoch": 0.9750559091691038,
-      "grad_norm": 0.0007211164338514209,
-      "learning_rate": 0.00019999953156670195,
-      "loss": 46.0,
-      "step": 12753
-    },
-    {
-      "epoch": 0.9751323661524934,
-      "grad_norm": 0.0016468371031805873,
-      "learning_rate": 0.00019999953149317902,
-      "loss": 46.0,
-      "step": 12754
-    },
-    {
-      "epoch": 0.9752088231358832,
-      "grad_norm": 0.0018547163344919682,
-      "learning_rate": 0.00019999953141965033,
-      "loss": 46.0,
-      "step": 12755
-    },
-    {
-      "epoch": 0.9752852801192728,
-      "grad_norm": 0.001210667658597231,
-      "learning_rate": 0.00019999953134611586,
-      "loss": 46.0,
-      "step": 12756
-    },
-    {
-      "epoch": 0.9753617371026626,
-      "grad_norm": 0.005310072097927332,
-      "learning_rate": 0.00019999953127257562,
-      "loss": 46.0,
-      "step": 12757
-    },
-    {
-      "epoch": 0.9754381940860524,
-      "grad_norm": 0.003956550732254982,
-      "learning_rate": 0.00019999953119902964,
-      "loss": 46.0,
-      "step": 12758
-    },
-    {
-      "epoch": 0.975514651069442,
-      "grad_norm": 0.0024545593187212944,
-      "learning_rate": 0.00019999953112547787,
-      "loss": 46.0,
-      "step": 12759
-    },
-    {
-      "epoch": 0.9755911080528318,
-      "grad_norm": 0.0032583093270659447,
-      "learning_rate": 0.00019999953105192031,
-      "loss": 46.0,
-      "step": 12760
-    },
-    {
-      "epoch": 0.9756675650362215,
-      "grad_norm": 0.0008235130808316171,
-      "learning_rate": 0.000199999530978357,
-      "loss": 46.0,
-      "step": 12761
-    },
-    {
-      "epoch": 0.9757440220196112,
-      "grad_norm": 0.002267898293212056,
-      "learning_rate": 0.00019999953090478793,
-      "loss": 46.0,
-      "step": 12762
-    },
-    {
-      "epoch": 0.975820479003001,
-      "grad_norm": 0.0022681127302348614,
-      "learning_rate": 0.00019999953083121307,
-      "loss": 46.0,
-      "step": 12763
-    },
-    {
-      "epoch": 0.9758969359863907,
-      "grad_norm": 0.0011638295836746693,
-      "learning_rate": 0.00019999953075763247,
-      "loss": 46.0,
-      "step": 12764
-    },
-    {
-      "epoch": 0.9759733929697804,
-      "grad_norm": 0.007835728116333485,
-      "learning_rate": 0.0001999995306840461,
-      "loss": 46.0,
-      "step": 12765
-    },
-    {
-      "epoch": 0.9760498499531701,
-      "grad_norm": 0.0005981057765893638,
-      "learning_rate": 0.00019999953061045392,
-      "loss": 46.0,
-      "step": 12766
-    },
-    {
-      "epoch": 0.9761263069365598,
-      "grad_norm": 0.002416054019704461,
-      "learning_rate": 0.000199999530536856,
-      "loss": 46.0,
-      "step": 12767
-    },
-    {
-      "epoch": 0.9762027639199495,
-      "grad_norm": 0.003757932223379612,
-      "learning_rate": 0.0001999995304632523,
-      "loss": 46.0,
-      "step": 12768
-    },
-    {
-      "epoch": 0.9762792209033393,
-      "grad_norm": 0.003490697592496872,
-      "learning_rate": 0.00019999953038964284,
-      "loss": 46.0,
-      "step": 12769
-    },
-    {
-      "epoch": 0.976355677886729,
-      "grad_norm": 0.001993812620639801,
-      "learning_rate": 0.00019999953031602758,
-      "loss": 46.0,
-      "step": 12770
-    },
-    {
-      "epoch": 0.9764321348701187,
-      "grad_norm": 0.004712095018476248,
-      "learning_rate": 0.00019999953024240662,
-      "loss": 46.0,
-      "step": 12771
-    },
-    {
-      "epoch": 0.9765085918535085,
-      "grad_norm": 0.0052865780889987946,
-      "learning_rate": 0.0001999995301687798,
-      "loss": 46.0,
-      "step": 12772
-    },
-    {
-      "epoch": 0.9765850488368981,
-      "grad_norm": 0.0007005177903920412,
-      "learning_rate": 0.00019999953009514727,
-      "loss": 46.0,
-      "step": 12773
-    },
-    {
-      "epoch": 0.9766615058202879,
-      "grad_norm": 0.001018381561152637,
-      "learning_rate": 0.00019999953002150896,
-      "loss": 46.0,
-      "step": 12774
-    },
-    {
-      "epoch": 0.9767379628036775,
-      "grad_norm": 0.0016934463055804372,
-      "learning_rate": 0.00019999952994786488,
-      "loss": 46.0,
-      "step": 12775
-    },
-    {
-      "epoch": 0.9768144197870673,
-      "grad_norm": 0.013121184892952442,
-      "learning_rate": 0.00019999952987421503,
-      "loss": 46.0,
-      "step": 12776
-    },
-    {
-      "epoch": 0.976890876770457,
-      "grad_norm": 0.0019209851743653417,
-      "learning_rate": 0.0001999995298005594,
-      "loss": 46.0,
-      "step": 12777
-    },
-    {
-      "epoch": 0.9769673337538467,
-      "grad_norm": 0.0007861430640332401,
-      "learning_rate": 0.00019999952972689803,
-      "loss": 46.0,
-      "step": 12778
-    },
-    {
-      "epoch": 0.9770437907372365,
-      "grad_norm": 0.0008459283271804452,
-      "learning_rate": 0.00019999952965323085,
-      "loss": 46.0,
-      "step": 12779
-    },
-    {
-      "epoch": 0.9771202477206262,
-      "grad_norm": 0.0008801782387308776,
-      "learning_rate": 0.00019999952957955793,
-      "loss": 46.0,
-      "step": 12780
-    },
-    {
-      "epoch": 0.9771967047040159,
-      "grad_norm": 0.0008988889749161899,
-      "learning_rate": 0.00019999952950587924,
-      "loss": 46.0,
-      "step": 12781
-    },
-    {
-      "epoch": 0.9772731616874056,
-      "grad_norm": 0.0008732814458198845,
-      "learning_rate": 0.00019999952943219475,
-      "loss": 46.0,
-      "step": 12782
-    },
-    {
-      "epoch": 0.9773496186707954,
-      "grad_norm": 0.001274761976674199,
-      "learning_rate": 0.00019999952935850453,
-      "loss": 46.0,
-      "step": 12783
-    },
-    {
-      "epoch": 0.977426075654185,
-      "grad_norm": 0.0013002234045416117,
-      "learning_rate": 0.00019999952928480852,
-      "loss": 46.0,
-      "step": 12784
-    },
-    {
-      "epoch": 0.9775025326375748,
-      "grad_norm": 0.001604813733138144,
-      "learning_rate": 0.00019999952921110673,
-      "loss": 46.0,
-      "step": 12785
-    },
-    {
-      "epoch": 0.9775789896209645,
-      "grad_norm": 0.0005631727981381118,
-      "learning_rate": 0.0001999995291373992,
-      "loss": 46.0,
-      "step": 12786
-    },
-    {
-      "epoch": 0.9776554466043542,
-      "grad_norm": 0.001688444404862821,
-      "learning_rate": 0.0001999995290636859,
-      "loss": 46.0,
-      "step": 12787
-    },
-    {
-      "epoch": 0.977731903587744,
-      "grad_norm": 0.000580216699745506,
-      "learning_rate": 0.0001999995289899668,
-      "loss": 46.0,
-      "step": 12788
-    },
-    {
-      "epoch": 0.9778083605711336,
-      "grad_norm": 0.001351632410660386,
-      "learning_rate": 0.00019999952891624193,
-      "loss": 46.0,
-      "step": 12789
-    },
-    {
-      "epoch": 0.9778848175545234,
-      "grad_norm": 0.01026106532663107,
-      "learning_rate": 0.0001999995288425113,
-      "loss": 46.0,
-      "step": 12790
-    },
-    {
-      "epoch": 0.9779612745379131,
-      "grad_norm": 0.0030651770066469908,
-      "learning_rate": 0.00019999952876877493,
-      "loss": 46.0,
-      "step": 12791
-    },
-    {
-      "epoch": 0.9780377315213028,
-      "grad_norm": 0.001269552856683731,
-      "learning_rate": 0.00019999952869503276,
-      "loss": 46.0,
-      "step": 12792
-    },
-    {
-      "epoch": 0.9781141885046926,
-      "grad_norm": 0.0006460731965489686,
-      "learning_rate": 0.00019999952862128484,
-      "loss": 46.0,
-      "step": 12793
-    },
-    {
-      "epoch": 0.9781906454880823,
-      "grad_norm": 0.0023850095458328724,
-      "learning_rate": 0.00019999952854753112,
-      "loss": 46.0,
-      "step": 12794
-    },
-    {
-      "epoch": 0.978267102471472,
-      "grad_norm": 0.0010806452482938766,
-      "learning_rate": 0.00019999952847377165,
-      "loss": 46.0,
-      "step": 12795
-    },
-    {
-      "epoch": 0.9783435594548617,
-      "grad_norm": 0.0018902196316048503,
-      "learning_rate": 0.00019999952840000642,
-      "loss": 46.0,
-      "step": 12796
-    },
-    {
-      "epoch": 0.9784200164382514,
-      "grad_norm": 0.0007955400506034493,
-      "learning_rate": 0.0001999995283262354,
-      "loss": 46.0,
-      "step": 12797
-    },
-    {
-      "epoch": 0.9784964734216411,
-      "grad_norm": 0.001599586452357471,
-      "learning_rate": 0.00019999952825245862,
-      "loss": 46.0,
-      "step": 12798
-    },
-    {
-      "epoch": 0.9785729304050309,
-      "grad_norm": 0.00045560463331639767,
-      "learning_rate": 0.00019999952817867608,
-      "loss": 46.0,
-      "step": 12799
-    },
-    {
-      "epoch": 0.9786493873884206,
-      "grad_norm": 0.0010739242425188422,
-      "learning_rate": 0.00019999952810488775,
-      "loss": 46.0,
-      "step": 12800
-    },
-    {
-      "epoch": 0.9787258443718103,
-      "grad_norm": 0.0008624702459201217,
-      "learning_rate": 0.00019999952803109367,
-      "loss": 46.0,
-      "step": 12801
-    },
-    {
-      "epoch": 0.9788023013552001,
-      "grad_norm": 0.0006362621206790209,
-      "learning_rate": 0.0001999995279572938,
-      "loss": 46.0,
-      "step": 12802
-    },
-    {
-      "epoch": 0.9788787583385897,
-      "grad_norm": 0.0010870335390791297,
-      "learning_rate": 0.0001999995278834882,
-      "loss": 46.0,
-      "step": 12803
-    },
-    {
-      "epoch": 0.9789552153219795,
-      "grad_norm": 0.0011031472822651267,
-      "learning_rate": 0.0001999995278096768,
-      "loss": 46.0,
-      "step": 12804
-    },
-    {
-      "epoch": 0.9790316723053691,
-      "grad_norm": 0.0018994042184203863,
-      "learning_rate": 0.00019999952773585964,
-      "loss": 46.0,
-      "step": 12805
-    },
-    {
-      "epoch": 0.9791081292887589,
-      "grad_norm": 0.0002852639590855688,
-      "learning_rate": 0.0001999995276620367,
-      "loss": 46.0,
-      "step": 12806
-    },
-    {
-      "epoch": 0.9791845862721487,
-      "grad_norm": 0.0026290160603821278,
-      "learning_rate": 0.00019999952758820798,
-      "loss": 46.0,
-      "step": 12807
-    },
-    {
-      "epoch": 0.9792610432555383,
-      "grad_norm": 0.001884467783384025,
-      "learning_rate": 0.0001999995275143735,
-      "loss": 46.0,
-      "step": 12808
-    },
-    {
-      "epoch": 0.9793375002389281,
-      "grad_norm": 0.019786866381764412,
-      "learning_rate": 0.00019999952744053327,
-      "loss": 46.0,
-      "step": 12809
-    },
-    {
-      "epoch": 0.9794139572223178,
-      "grad_norm": 0.0008956074598245323,
-      "learning_rate": 0.00019999952736668726,
-      "loss": 46.0,
-      "step": 12810
-    },
-    {
-      "epoch": 0.9794904142057075,
-      "grad_norm": 0.0010076133767142892,
-      "learning_rate": 0.00019999952729283547,
-      "loss": 46.0,
-      "step": 12811
-    },
-    {
-      "epoch": 0.9795668711890972,
-      "grad_norm": 0.0008984691230580211,
-      "learning_rate": 0.00019999952721897791,
-      "loss": 46.0,
-      "step": 12812
-    },
-    {
-      "epoch": 0.979643328172487,
-      "grad_norm": 0.0017784673254936934,
-      "learning_rate": 0.0001999995271451146,
-      "loss": 46.0,
-      "step": 12813
-    },
-    {
-      "epoch": 0.9797197851558767,
-      "grad_norm": 0.001371965860016644,
-      "learning_rate": 0.00019999952707124553,
-      "loss": 46.0,
-      "step": 12814
-    },
-    {
-      "epoch": 0.9797962421392664,
-      "grad_norm": 0.001625708187930286,
-      "learning_rate": 0.00019999952699737065,
-      "loss": 46.0,
-      "step": 12815
-    },
-    {
-      "epoch": 0.9798726991226561,
-      "grad_norm": 0.017182547599077225,
-      "learning_rate": 0.00019999952692349005,
-      "loss": 46.0,
-      "step": 12816
-    },
-    {
-      "epoch": 0.9799491561060458,
-      "grad_norm": 0.00389941711910069,
-      "learning_rate": 0.00019999952684960362,
-      "loss": 46.0,
-      "step": 12817
-    },
-    {
-      "epoch": 0.9800256130894356,
-      "grad_norm": 0.006198948714882135,
-      "learning_rate": 0.00019999952677571145,
-      "loss": 46.0,
-      "step": 12818
-    },
-    {
-      "epoch": 0.9801020700728252,
-      "grad_norm": 0.0012636421015486121,
-      "learning_rate": 0.00019999952670181353,
-      "loss": 46.0,
-      "step": 12819
-    },
-    {
-      "epoch": 0.980178527056215,
-      "grad_norm": 0.0007669960032217205,
-      "learning_rate": 0.0001999995266279098,
-      "loss": 46.0,
-      "step": 12820
-    },
-    {
-      "epoch": 0.9802549840396048,
-      "grad_norm": 0.002090106951072812,
-      "learning_rate": 0.00019999952655400035,
-      "loss": 46.0,
-      "step": 12821
-    },
-    {
-      "epoch": 0.9803314410229944,
-      "grad_norm": 0.0005873393383808434,
-      "learning_rate": 0.00019999952648008508,
-      "loss": 46.0,
-      "step": 12822
-    },
-    {
-      "epoch": 0.9804078980063842,
-      "grad_norm": 0.003474652534350753,
-      "learning_rate": 0.0001999995264061641,
-      "loss": 46.0,
-      "step": 12823
-    },
-    {
-      "epoch": 0.9804843549897739,
-      "grad_norm": 0.0016198530793190002,
-      "learning_rate": 0.0001999995263322373,
-      "loss": 46.0,
-      "step": 12824
-    },
-    {
-      "epoch": 0.9805608119731636,
-      "grad_norm": 0.0009181445930153131,
-      "learning_rate": 0.00019999952625830475,
-      "loss": 46.0,
-      "step": 12825
-    },
-    {
-      "epoch": 0.9806372689565533,
-      "grad_norm": 0.006141019985079765,
-      "learning_rate": 0.00019999952618436644,
-      "loss": 46.0,
-      "step": 12826
-    },
-    {
-      "epoch": 0.980713725939943,
-      "grad_norm": 0.001359390327706933,
-      "learning_rate": 0.00019999952611042234,
-      "loss": 46.0,
-      "step": 12827
-    },
-    {
-      "epoch": 0.9807901829233328,
-      "grad_norm": 0.003163157496601343,
-      "learning_rate": 0.00019999952603647249,
-      "loss": 46.0,
-      "step": 12828
-    },
-    {
-      "epoch": 0.9808666399067225,
-      "grad_norm": 0.0006966502987779677,
-      "learning_rate": 0.00019999952596251683,
-      "loss": 46.0,
-      "step": 12829
-    },
-    {
-      "epoch": 0.9809430968901122,
-      "grad_norm": 0.0005184897454455495,
-      "learning_rate": 0.00019999952588855543,
-      "loss": 46.0,
-      "step": 12830
-    },
-    {
-      "epoch": 0.9810195538735019,
-      "grad_norm": 0.002286858158186078,
-      "learning_rate": 0.00019999952581458826,
-      "loss": 46.0,
-      "step": 12831
-    },
-    {
-      "epoch": 0.9810960108568917,
-      "grad_norm": 0.004737162962555885,
-      "learning_rate": 0.00019999952574061532,
-      "loss": 46.0,
-      "step": 12832
-    },
-    {
-      "epoch": 0.9811724678402813,
-      "grad_norm": 0.000951636815443635,
-      "learning_rate": 0.00019999952566663662,
-      "loss": 46.0,
-      "step": 12833
-    },
-    {
-      "epoch": 0.9812489248236711,
-      "grad_norm": 0.001245854771696031,
-      "learning_rate": 0.00019999952559265213,
-      "loss": 46.0,
-      "step": 12834
-    },
-    {
-      "epoch": 0.9813253818070607,
-      "grad_norm": 0.010222101584076881,
-      "learning_rate": 0.0001999995255186619,
-      "loss": 46.0,
-      "step": 12835
-    },
-    {
-      "epoch": 0.9814018387904505,
-      "grad_norm": 0.0003897756105288863,
-      "learning_rate": 0.00019999952544466588,
-      "loss": 46.0,
-      "step": 12836
-    },
-    {
-      "epoch": 0.9814782957738403,
-      "grad_norm": 0.001301162177696824,
-      "learning_rate": 0.0001999995253706641,
-      "loss": 46.0,
-      "step": 12837
-    },
-    {
-      "epoch": 0.9815547527572299,
-      "grad_norm": 0.0008635082631371915,
-      "learning_rate": 0.00019999952529665654,
-      "loss": 46.0,
-      "step": 12838
-    },
-    {
-      "epoch": 0.9816312097406197,
-      "grad_norm": 0.002238568617030978,
-      "learning_rate": 0.0001999995252226432,
-      "loss": 46.0,
-      "step": 12839
-    },
-    {
-      "epoch": 0.9817076667240094,
-      "grad_norm": 0.0009574776631779969,
-      "learning_rate": 0.0001999995251486241,
-      "loss": 46.0,
-      "step": 12840
-    },
-    {
-      "epoch": 0.9817841237073991,
-      "grad_norm": 0.0015107898507267237,
-      "learning_rate": 0.00019999952507459925,
-      "loss": 46.0,
-      "step": 12841
-    },
-    {
-      "epoch": 0.9818605806907889,
-      "grad_norm": 0.006809369195252657,
-      "learning_rate": 0.00019999952500056862,
-      "loss": 46.0,
-      "step": 12842
-    },
-    {
-      "epoch": 0.9819370376741786,
-      "grad_norm": 0.001495785778388381,
-      "learning_rate": 0.0001999995249265322,
-      "loss": 46.0,
-      "step": 12843
-    },
-    {
-      "epoch": 0.9820134946575683,
-      "grad_norm": 0.0007589769084006548,
-      "learning_rate": 0.00019999952485249005,
-      "loss": 46.0,
-      "step": 12844
-    },
-    {
-      "epoch": 0.982089951640958,
-      "grad_norm": 0.0015061625745147467,
-      "learning_rate": 0.00019999952477844208,
-      "loss": 46.0,
-      "step": 12845
-    },
-    {
-      "epoch": 0.9821664086243477,
-      "grad_norm": 0.001059127040207386,
-      "learning_rate": 0.00019999952470438838,
-      "loss": 46.0,
-      "step": 12846
-    },
-    {
-      "epoch": 0.9822428656077374,
-      "grad_norm": 0.0019423268968239427,
-      "learning_rate": 0.00019999952463032892,
-      "loss": 46.0,
-      "step": 12847
-    },
-    {
-      "epoch": 0.9823193225911272,
-      "grad_norm": 0.002396803814917803,
-      "learning_rate": 0.00019999952455626365,
-      "loss": 46.0,
-      "step": 12848
-    },
-    {
-      "epoch": 0.9823957795745168,
-      "grad_norm": 0.0012942602625116706,
-      "learning_rate": 0.00019999952448219264,
-      "loss": 46.0,
-      "step": 12849
-    },
-    {
-      "epoch": 0.9824722365579066,
-      "grad_norm": 0.001291621127165854,
-      "learning_rate": 0.00019999952440811586,
-      "loss": 46.0,
-      "step": 12850
-    },
-    {
-      "epoch": 0.9825486935412964,
-      "grad_norm": 0.015728944912552834,
-      "learning_rate": 0.00019999952433403327,
-      "loss": 46.0,
-      "step": 12851
-    },
-    {
-      "epoch": 0.982625150524686,
-      "grad_norm": 0.002567258197814226,
-      "learning_rate": 0.00019999952425994497,
-      "loss": 46.0,
-      "step": 12852
-    },
-    {
-      "epoch": 0.9827016075080758,
-      "grad_norm": 0.0008253243868239224,
-      "learning_rate": 0.00019999952418585086,
-      "loss": 46.0,
-      "step": 12853
-    },
-    {
-      "epoch": 0.9827780644914655,
-      "grad_norm": 0.0002872333861887455,
-      "learning_rate": 0.00019999952411175098,
-      "loss": 46.0,
-      "step": 12854
-    },
-    {
-      "epoch": 0.9828545214748552,
-      "grad_norm": 0.0011262326734140515,
-      "learning_rate": 0.00019999952403764536,
-      "loss": 46.0,
-      "step": 12855
-    },
-    {
-      "epoch": 0.982930978458245,
-      "grad_norm": 0.0009918413124978542,
-      "learning_rate": 0.00019999952396353393,
-      "loss": 46.0,
-      "step": 12856
-    },
-    {
-      "epoch": 0.9830074354416346,
-      "grad_norm": 0.0010470869019627571,
-      "learning_rate": 0.0001999995238894168,
-      "loss": 46.0,
-      "step": 12857
-    },
-    {
-      "epoch": 0.9830838924250244,
-      "grad_norm": 0.001586166792549193,
-      "learning_rate": 0.00019999952381529384,
-      "loss": 46.0,
-      "step": 12858
-    },
-    {
-      "epoch": 0.9831603494084141,
-      "grad_norm": 0.004364318680018187,
-      "learning_rate": 0.0001999995237411651,
-      "loss": 46.0,
-      "step": 12859
-    },
-    {
-      "epoch": 0.9832368063918038,
-      "grad_norm": 0.0009643954690545797,
-      "learning_rate": 0.00019999952366703063,
-      "loss": 46.0,
-      "step": 12860
-    },
-    {
-      "epoch": 0.9833132633751935,
-      "grad_norm": 0.0022769563365727663,
-      "learning_rate": 0.0001999995235928904,
-      "loss": 46.0,
-      "step": 12861
-    },
-    {
-      "epoch": 0.9833897203585833,
-      "grad_norm": 0.0022856772411614656,
-      "learning_rate": 0.00019999952351874435,
-      "loss": 46.0,
-      "step": 12862
-    },
-    {
-      "epoch": 0.983466177341973,
-      "grad_norm": 0.0018426288152113557,
-      "learning_rate": 0.00019999952344459257,
-      "loss": 46.0,
-      "step": 12863
-    },
-    {
-      "epoch": 0.9835426343253627,
-      "grad_norm": 0.005390865262597799,
-      "learning_rate": 0.00019999952337043498,
-      "loss": 46.0,
-      "step": 12864
-    },
-    {
-      "epoch": 0.9836190913087524,
-      "grad_norm": 0.0012018068227916956,
-      "learning_rate": 0.00019999952329627168,
-      "loss": 46.0,
-      "step": 12865
-    },
-    {
-      "epoch": 0.9836955482921421,
-      "grad_norm": 0.0018065322656184435,
-      "learning_rate": 0.00019999952322210257,
-      "loss": 46.0,
-      "step": 12866
-    },
-    {
-      "epoch": 0.9837720052755319,
-      "grad_norm": 0.0006826574099250138,
-      "learning_rate": 0.0001999995231479277,
-      "loss": 46.0,
-      "step": 12867
-    },
-    {
-      "epoch": 0.9838484622589215,
-      "grad_norm": 0.009405352175235748,
-      "learning_rate": 0.00019999952307374707,
-      "loss": 46.0,
-      "step": 12868
-    },
-    {
-      "epoch": 0.9839249192423113,
-      "grad_norm": 0.0009445718606002629,
-      "learning_rate": 0.00019999952299956065,
-      "loss": 46.0,
-      "step": 12869
-    },
-    {
-      "epoch": 0.984001376225701,
-      "grad_norm": 0.0024388430174440145,
-      "learning_rate": 0.00019999952292536847,
-      "loss": 46.0,
-      "step": 12870
-    },
-    {
-      "epoch": 0.9840778332090907,
-      "grad_norm": 0.0004667453467845917,
-      "learning_rate": 0.00019999952285117053,
-      "loss": 46.0,
-      "step": 12871
-    },
-    {
-      "epoch": 0.9841542901924805,
-      "grad_norm": 0.0019605106208473444,
-      "learning_rate": 0.0001999995227769668,
-      "loss": 46.0,
-      "step": 12872
-    },
-    {
-      "epoch": 0.9842307471758702,
-      "grad_norm": 0.0015231982106342912,
-      "learning_rate": 0.00019999952270275732,
-      "loss": 46.0,
-      "step": 12873
-    },
-    {
-      "epoch": 0.9843072041592599,
-      "grad_norm": 0.0011590765789151192,
-      "learning_rate": 0.00019999952262854208,
-      "loss": 46.0,
-      "step": 12874
-    },
-    {
-      "epoch": 0.9843836611426496,
-      "grad_norm": 0.0013544122921302915,
-      "learning_rate": 0.00019999952255432104,
-      "loss": 46.0,
-      "step": 12875
-    },
-    {
-      "epoch": 0.9844601181260393,
-      "grad_norm": 0.007144773378968239,
-      "learning_rate": 0.00019999952248009426,
-      "loss": 46.0,
-      "step": 12876
-    },
-    {
-      "epoch": 0.984536575109429,
-      "grad_norm": 0.0005381340743042529,
-      "learning_rate": 0.00019999952240586167,
-      "loss": 46.0,
-      "step": 12877
-    },
-    {
-      "epoch": 0.9846130320928188,
-      "grad_norm": 0.007375964429229498,
-      "learning_rate": 0.00019999952233162337,
-      "loss": 46.0,
-      "step": 12878
-    },
-    {
-      "epoch": 0.9846894890762085,
-      "grad_norm": 0.004662943538278341,
-      "learning_rate": 0.00019999952225737926,
-      "loss": 46.0,
-      "step": 12879
-    },
-    {
-      "epoch": 0.9847659460595982,
-      "grad_norm": 0.0012037480482831597,
-      "learning_rate": 0.00019999952218312936,
-      "loss": 46.0,
-      "step": 12880
-    },
-    {
-      "epoch": 0.984842403042988,
-      "grad_norm": 0.0010006591910496354,
-      "learning_rate": 0.00019999952210887373,
-      "loss": 46.0,
-      "step": 12881
-    },
-    {
-      "epoch": 0.9849188600263776,
-      "grad_norm": 0.0024163899943232536,
-      "learning_rate": 0.00019999952203461234,
-      "loss": 46.0,
-      "step": 12882
-    },
-    {
-      "epoch": 0.9849953170097674,
-      "grad_norm": 0.0008803425589576364,
-      "learning_rate": 0.00019999952196034514,
-      "loss": 46.0,
-      "step": 12883
-    },
-    {
-      "epoch": 0.9850717739931572,
-      "grad_norm": 0.0007207217277027667,
-      "learning_rate": 0.0001999995218860722,
-      "loss": 46.0,
-      "step": 12884
-    },
-    {
-      "epoch": 0.9851482309765468,
-      "grad_norm": 0.0009455467225052416,
-      "learning_rate": 0.00019999952181179348,
-      "loss": 46.0,
-      "step": 12885
-    },
-    {
-      "epoch": 0.9852246879599366,
-      "grad_norm": 0.0005324353114701807,
-      "learning_rate": 0.00019999952173750898,
-      "loss": 46.0,
-      "step": 12886
-    },
-    {
-      "epoch": 0.9853011449433262,
-      "grad_norm": 0.004103631712496281,
-      "learning_rate": 0.00019999952166321872,
-      "loss": 46.0,
-      "step": 12887
-    },
-    {
-      "epoch": 0.985377601926716,
-      "grad_norm": 0.0023021239321678877,
-      "learning_rate": 0.0001999995215889227,
-      "loss": 46.0,
-      "step": 12888
-    },
-    {
-      "epoch": 0.9854540589101057,
-      "grad_norm": 0.004153313580900431,
-      "learning_rate": 0.00019999952151462092,
-      "loss": 46.0,
-      "step": 12889
-    },
-    {
-      "epoch": 0.9855305158934954,
-      "grad_norm": 0.0007513034506700933,
-      "learning_rate": 0.00019999952144031334,
-      "loss": 46.0,
-      "step": 12890
-    },
-    {
-      "epoch": 0.9856069728768851,
-      "grad_norm": 0.0020820791833102703,
-      "learning_rate": 0.000199999521366,
-      "loss": 46.0,
-      "step": 12891
-    },
-    {
-      "epoch": 0.9856834298602749,
-      "grad_norm": 0.0007823725463822484,
-      "learning_rate": 0.0001999995212916809,
-      "loss": 46.0,
-      "step": 12892
-    },
-    {
-      "epoch": 0.9857598868436646,
-      "grad_norm": 0.001226668944582343,
-      "learning_rate": 0.00019999952121735603,
-      "loss": 46.0,
-      "step": 12893
-    },
-    {
-      "epoch": 0.9858363438270543,
-      "grad_norm": 0.00089498411398381,
-      "learning_rate": 0.00019999952114302537,
-      "loss": 46.0,
-      "step": 12894
-    },
-    {
-      "epoch": 0.9859128008104441,
-      "grad_norm": 0.001373312552459538,
-      "learning_rate": 0.00019999952106868895,
-      "loss": 46.0,
-      "step": 12895
-    },
-    {
-      "epoch": 0.9859892577938337,
-      "grad_norm": 0.0038569029420614243,
-      "learning_rate": 0.00019999952099434678,
-      "loss": 46.0,
-      "step": 12896
-    },
-    {
-      "epoch": 0.9860657147772235,
-      "grad_norm": 0.005535216070711613,
-      "learning_rate": 0.00019999952091999884,
-      "loss": 46.0,
-      "step": 12897
-    },
-    {
-      "epoch": 0.9861421717606131,
-      "grad_norm": 0.0007432788843289018,
-      "learning_rate": 0.00019999952084564512,
-      "loss": 46.0,
-      "step": 12898
-    },
-    {
-      "epoch": 0.9862186287440029,
-      "grad_norm": 0.0024454211816191673,
-      "learning_rate": 0.0001999995207712856,
-      "loss": 46.0,
-      "step": 12899
-    },
-    {
-      "epoch": 0.9862950857273927,
-      "grad_norm": 0.0028503509238362312,
-      "learning_rate": 0.00019999952069692036,
-      "loss": 46.0,
-      "step": 12900
-    },
-    {
-      "epoch": 0.9863715427107823,
-      "grad_norm": 0.0021564550697803497,
-      "learning_rate": 0.00019999952062254935,
-      "loss": 46.0,
-      "step": 12901
-    },
-    {
-      "epoch": 0.9864479996941721,
-      "grad_norm": 0.002211756305769086,
-      "learning_rate": 0.0001999995205481725,
-      "loss": 46.0,
-      "step": 12902
-    },
-    {
-      "epoch": 0.9865244566775618,
-      "grad_norm": 0.0018920553848147392,
-      "learning_rate": 0.00019999952047378996,
-      "loss": 46.0,
-      "step": 12903
-    },
-    {
-      "epoch": 0.9866009136609515,
-      "grad_norm": 0.0025991916190832853,
-      "learning_rate": 0.00019999952039940162,
-      "loss": 46.0,
-      "step": 12904
-    },
-    {
-      "epoch": 0.9866773706443412,
-      "grad_norm": 0.000843998568598181,
-      "learning_rate": 0.00019999952032500752,
-      "loss": 46.0,
-      "step": 12905
-    },
-    {
-      "epoch": 0.9867538276277309,
-      "grad_norm": 0.0016047506360337138,
-      "learning_rate": 0.00019999952025060762,
-      "loss": 46.0,
-      "step": 12906
-    },
-    {
-      "epoch": 0.9868302846111207,
-      "grad_norm": 0.0007460570195689797,
-      "learning_rate": 0.000199999520176202,
-      "loss": 46.0,
-      "step": 12907
-    },
-    {
-      "epoch": 0.9869067415945104,
-      "grad_norm": 0.004684058949351311,
-      "learning_rate": 0.00019999952010179057,
-      "loss": 46.0,
-      "step": 12908
-    },
-    {
-      "epoch": 0.9869831985779001,
-      "grad_norm": 0.006778838112950325,
-      "learning_rate": 0.0001999995200273734,
-      "loss": 46.0,
-      "step": 12909
-    },
-    {
-      "epoch": 0.9870596555612898,
-      "grad_norm": 0.0008412369061261415,
-      "learning_rate": 0.00019999951995295043,
-      "loss": 46.0,
-      "step": 12910
-    },
-    {
-      "epoch": 0.9871361125446796,
-      "grad_norm": 0.004750713240355253,
-      "learning_rate": 0.0001999995198785217,
-      "loss": 46.0,
-      "step": 12911
-    },
-    {
-      "epoch": 0.9872125695280692,
-      "grad_norm": 0.0012419220292940736,
-      "learning_rate": 0.0001999995198040872,
-      "loss": 46.0,
-      "step": 12912
-    },
-    {
-      "epoch": 0.987289026511459,
-      "grad_norm": 0.005220053251832724,
-      "learning_rate": 0.00019999951972964695,
-      "loss": 46.0,
-      "step": 12913
-    },
-    {
-      "epoch": 0.9873654834948488,
-      "grad_norm": 0.0013945535756647587,
-      "learning_rate": 0.00019999951965520092,
-      "loss": 46.0,
-      "step": 12914
-    },
-    {
-      "epoch": 0.9874419404782384,
-      "grad_norm": 0.0021088188514113426,
-      "learning_rate": 0.0001999995195807491,
-      "loss": 46.0,
-      "step": 12915
-    },
-    {
-      "epoch": 0.9875183974616282,
-      "grad_norm": 0.00351086538285017,
-      "learning_rate": 0.00019999951950629155,
-      "loss": 46.0,
-      "step": 12916
-    },
-    {
-      "epoch": 0.9875948544450178,
-      "grad_norm": 0.0007880371995270252,
-      "learning_rate": 0.00019999951943182822,
-      "loss": 46.0,
-      "step": 12917
-    },
-    {
-      "epoch": 0.9876713114284076,
-      "grad_norm": 0.0011672667460516095,
-      "learning_rate": 0.0001999995193573591,
-      "loss": 46.0,
-      "step": 12918
-    },
-    {
-      "epoch": 0.9877477684117973,
-      "grad_norm": 0.01388244517147541,
-      "learning_rate": 0.0001999995192828842,
-      "loss": 46.0,
-      "step": 12919
-    },
-    {
-      "epoch": 0.987824225395187,
-      "grad_norm": 0.002916842233389616,
-      "learning_rate": 0.00019999951920840356,
-      "loss": 46.0,
-      "step": 12920
-    },
-    {
-      "epoch": 0.9879006823785768,
-      "grad_norm": 0.002946749795228243,
-      "learning_rate": 0.00019999951913391714,
-      "loss": 46.0,
-      "step": 12921
-    },
-    {
-      "epoch": 0.9879771393619665,
-      "grad_norm": 0.0003848739725071937,
-      "learning_rate": 0.00019999951905942497,
-      "loss": 46.0,
-      "step": 12922
-    },
-    {
-      "epoch": 0.9880535963453562,
-      "grad_norm": 0.000876496487762779,
-      "learning_rate": 0.000199999518984927,
-      "loss": 46.0,
-      "step": 12923
-    },
-    {
-      "epoch": 0.9881300533287459,
-      "grad_norm": 0.0003296476788818836,
-      "learning_rate": 0.00019999951891042328,
-      "loss": 46.0,
-      "step": 12924
-    },
-    {
-      "epoch": 0.9882065103121357,
-      "grad_norm": 0.002046069363132119,
-      "learning_rate": 0.0001999995188359138,
-      "loss": 46.0,
-      "step": 12925
-    },
-    {
-      "epoch": 0.9882829672955253,
-      "grad_norm": 0.0017047772416844964,
-      "learning_rate": 0.0001999995187613985,
-      "loss": 46.0,
-      "step": 12926
-    },
-    {
-      "epoch": 0.9883594242789151,
-      "grad_norm": 0.00694029126316309,
-      "learning_rate": 0.0001999995186868775,
-      "loss": 46.0,
-      "step": 12927
-    },
-    {
-      "epoch": 0.9884358812623047,
-      "grad_norm": 0.002277107909321785,
-      "learning_rate": 0.00019999951861235068,
-      "loss": 46.0,
-      "step": 12928
-    },
-    {
-      "epoch": 0.9885123382456945,
-      "grad_norm": 0.0008195373229682446,
-      "learning_rate": 0.00019999951853781812,
-      "loss": 46.0,
-      "step": 12929
-    },
-    {
-      "epoch": 0.9885887952290843,
-      "grad_norm": 0.001289637410081923,
-      "learning_rate": 0.00019999951846327976,
-      "loss": 46.0,
-      "step": 12930
-    },
-    {
-      "epoch": 0.9886652522124739,
-      "grad_norm": 0.0011808108538389206,
-      "learning_rate": 0.00019999951838873563,
-      "loss": 46.0,
-      "step": 12931
-    },
-    {
-      "epoch": 0.9887417091958637,
-      "grad_norm": 0.0015951062086969614,
-      "learning_rate": 0.00019999951831418575,
-      "loss": 46.0,
-      "step": 12932
-    },
-    {
-      "epoch": 0.9888181661792534,
-      "grad_norm": 0.0011097746901214123,
-      "learning_rate": 0.0001999995182396301,
-      "loss": 46.0,
-      "step": 12933
-    },
-    {
-      "epoch": 0.9888946231626431,
-      "grad_norm": 0.00202350364997983,
-      "learning_rate": 0.00019999951816506868,
-      "loss": 46.0,
-      "step": 12934
-    },
-    {
-      "epoch": 0.9889710801460329,
-      "grad_norm": 0.0007234001532196999,
-      "learning_rate": 0.00019999951809050148,
-      "loss": 46.0,
-      "step": 12935
-    },
-    {
-      "epoch": 0.9890475371294225,
-      "grad_norm": 0.004777347669005394,
-      "learning_rate": 0.00019999951801592854,
-      "loss": 46.0,
-      "step": 12936
-    },
-    {
-      "epoch": 0.9891239941128123,
-      "grad_norm": 0.0010399366728961468,
-      "learning_rate": 0.00019999951794134982,
-      "loss": 46.0,
-      "step": 12937
-    },
-    {
-      "epoch": 0.989200451096202,
-      "grad_norm": 0.016312703490257263,
-      "learning_rate": 0.0001999995178667653,
-      "loss": 46.0,
-      "step": 12938
-    },
-    {
-      "epoch": 0.9892769080795917,
-      "grad_norm": 0.0014975982485339046,
-      "learning_rate": 0.00019999951779217504,
-      "loss": 46.0,
-      "step": 12939
-    },
-    {
-      "epoch": 0.9893533650629814,
-      "grad_norm": 0.0010700864950194955,
-      "learning_rate": 0.00019999951771757898,
-      "loss": 46.0,
-      "step": 12940
-    },
-    {
-      "epoch": 0.9894298220463712,
-      "grad_norm": 0.0015709844883531332,
-      "learning_rate": 0.0001999995176429772,
-      "loss": 46.0,
-      "step": 12941
-    },
-    {
-      "epoch": 0.9895062790297608,
-      "grad_norm": 0.0007254788069985807,
-      "learning_rate": 0.00019999951756836964,
-      "loss": 46.0,
-      "step": 12942
-    },
-    {
-      "epoch": 0.9895827360131506,
-      "grad_norm": 0.0008481065160594881,
-      "learning_rate": 0.00019999951749375628,
-      "loss": 46.0,
-      "step": 12943
-    },
-    {
-      "epoch": 0.9896591929965404,
-      "grad_norm": 0.005952953360974789,
-      "learning_rate": 0.00019999951741913715,
-      "loss": 46.0,
-      "step": 12944
-    },
-    {
-      "epoch": 0.98973564997993,
-      "grad_norm": 0.0022617268841713667,
-      "learning_rate": 0.00019999951734451225,
-      "loss": 46.0,
-      "step": 12945
-    },
-    {
-      "epoch": 0.9898121069633198,
-      "grad_norm": 0.000859533145558089,
-      "learning_rate": 0.00019999951726988163,
-      "loss": 46.0,
-      "step": 12946
-    },
-    {
-      "epoch": 0.9898885639467094,
-      "grad_norm": 0.0007730942452326417,
-      "learning_rate": 0.0001999995171952452,
-      "loss": 46.0,
-      "step": 12947
-    },
-    {
-      "epoch": 0.9899650209300992,
-      "grad_norm": 0.0013047890970483422,
-      "learning_rate": 0.000199999517120603,
-      "loss": 46.0,
-      "step": 12948
-    },
-    {
-      "epoch": 0.990041477913489,
-      "grad_norm": 0.002887323033064604,
-      "learning_rate": 0.00019999951704595504,
-      "loss": 46.0,
-      "step": 12949
-    },
-    {
-      "epoch": 0.9901179348968786,
-      "grad_norm": 0.012278687208890915,
-      "learning_rate": 0.0001999995169713013,
-      "loss": 46.0,
-      "step": 12950
-    },
-    {
-      "epoch": 0.9901943918802684,
-      "grad_norm": 0.00662717130035162,
-      "learning_rate": 0.0001999995168966418,
-      "loss": 46.0,
-      "step": 12951
-    },
-    {
-      "epoch": 0.9902708488636581,
-      "grad_norm": 0.00028733929502777755,
-      "learning_rate": 0.00019999951682197654,
-      "loss": 46.0,
-      "step": 12952
-    },
-    {
-      "epoch": 0.9903473058470478,
-      "grad_norm": 0.0008203035686165094,
-      "learning_rate": 0.00019999951674730548,
-      "loss": 46.0,
-      "step": 12953
-    },
-    {
-      "epoch": 0.9904237628304375,
-      "grad_norm": 0.0008644326007924974,
-      "learning_rate": 0.0001999995166726287,
-      "loss": 46.0,
-      "step": 12954
-    },
-    {
-      "epoch": 0.9905002198138273,
-      "grad_norm": 0.0007698223926126957,
-      "learning_rate": 0.0001999995165979461,
-      "loss": 46.0,
-      "step": 12955
-    },
-    {
-      "epoch": 0.990576676797217,
-      "grad_norm": 0.0032603333238512278,
-      "learning_rate": 0.00019999951652325778,
-      "loss": 46.0,
-      "step": 12956
-    },
-    {
-      "epoch": 0.9906531337806067,
-      "grad_norm": 0.0008211845415644348,
-      "learning_rate": 0.00019999951644856365,
-      "loss": 46.0,
-      "step": 12957
-    },
-    {
-      "epoch": 0.9907295907639964,
-      "grad_norm": 0.001471293973736465,
-      "learning_rate": 0.00019999951637386375,
-      "loss": 46.0,
-      "step": 12958
-    },
-    {
-      "epoch": 0.9908060477473861,
-      "grad_norm": 0.0007434911094605923,
-      "learning_rate": 0.0001999995162991581,
-      "loss": 46.0,
-      "step": 12959
-    },
-    {
-      "epoch": 0.9908825047307759,
-      "grad_norm": 0.0005880248500034213,
-      "learning_rate": 0.00019999951622444665,
-      "loss": 46.0,
-      "step": 12960
-    },
-    {
-      "epoch": 0.9909589617141655,
-      "grad_norm": 0.0018656188622117043,
-      "learning_rate": 0.00019999951614972948,
-      "loss": 46.0,
-      "step": 12961
-    },
-    {
-      "epoch": 0.9910354186975553,
-      "grad_norm": 0.0009604257065802813,
-      "learning_rate": 0.0001999995160750065,
-      "loss": 46.0,
-      "step": 12962
-    },
-    {
-      "epoch": 0.991111875680945,
-      "grad_norm": 0.0029789903201162815,
-      "learning_rate": 0.00019999951600027776,
-      "loss": 46.0,
-      "step": 12963
-    },
-    {
-      "epoch": 0.9911883326643347,
-      "grad_norm": 0.0010843200143426657,
-      "learning_rate": 0.00019999951592554328,
-      "loss": 46.0,
-      "step": 12964
-    },
-    {
-      "epoch": 0.9912647896477245,
-      "grad_norm": 0.0012757823569700122,
-      "learning_rate": 0.000199999515850803,
-      "loss": 46.0,
-      "step": 12965
-    },
-    {
-      "epoch": 0.9913412466311141,
-      "grad_norm": 0.0004343360778875649,
-      "learning_rate": 0.00019999951577605695,
-      "loss": 46.0,
-      "step": 12966
-    },
-    {
-      "epoch": 0.9914177036145039,
-      "grad_norm": 0.0005968990735709667,
-      "learning_rate": 0.00019999951570130514,
-      "loss": 46.0,
-      "step": 12967
-    },
-    {
-      "epoch": 0.9914941605978936,
-      "grad_norm": 0.0009484199108555913,
-      "learning_rate": 0.00019999951562654756,
-      "loss": 46.0,
-      "step": 12968
-    },
-    {
-      "epoch": 0.9915706175812833,
-      "grad_norm": 0.0032046486157923937,
-      "learning_rate": 0.0001999995155517842,
-      "loss": 46.0,
-      "step": 12969
-    },
-    {
-      "epoch": 0.991647074564673,
-      "grad_norm": 0.0008889741147868335,
-      "learning_rate": 0.00019999951547701507,
-      "loss": 46.0,
-      "step": 12970
-    },
-    {
-      "epoch": 0.9917235315480628,
-      "grad_norm": 0.002994607901200652,
-      "learning_rate": 0.0001999995154022402,
-      "loss": 46.0,
-      "step": 12971
-    },
-    {
-      "epoch": 0.9917999885314525,
-      "grad_norm": 0.001974208513274789,
-      "learning_rate": 0.00019999951532745955,
-      "loss": 46.0,
-      "step": 12972
-    },
-    {
-      "epoch": 0.9918764455148422,
-      "grad_norm": 0.0020898994989693165,
-      "learning_rate": 0.0001999995152526731,
-      "loss": 46.0,
-      "step": 12973
-    },
-    {
-      "epoch": 0.991952902498232,
-      "grad_norm": 0.00042279413901269436,
-      "learning_rate": 0.00019999951517788093,
-      "loss": 46.0,
-      "step": 12974
-    },
-    {
-      "epoch": 0.9920293594816216,
-      "grad_norm": 0.0012046464253216982,
-      "learning_rate": 0.00019999951510308296,
-      "loss": 46.0,
-      "step": 12975
-    },
-    {
-      "epoch": 0.9921058164650114,
-      "grad_norm": 0.0009185183444060385,
-      "learning_rate": 0.00019999951502827921,
-      "loss": 46.0,
-      "step": 12976
-    },
-    {
-      "epoch": 0.992182273448401,
-      "grad_norm": 0.001820918289013207,
-      "learning_rate": 0.0001999995149534697,
-      "loss": 46.0,
-      "step": 12977
-    },
-    {
-      "epoch": 0.9922587304317908,
-      "grad_norm": 0.0009391932399012148,
-      "learning_rate": 0.00019999951487865444,
-      "loss": 46.0,
-      "step": 12978
-    },
-    {
-      "epoch": 0.9923351874151806,
-      "grad_norm": 0.0007493648445233703,
-      "learning_rate": 0.0001999995148038334,
-      "loss": 46.0,
-      "step": 12979
-    },
-    {
-      "epoch": 0.9924116443985702,
-      "grad_norm": 0.008705347776412964,
-      "learning_rate": 0.00019999951472900656,
-      "loss": 46.0,
-      "step": 12980
-    },
-    {
-      "epoch": 0.99248810138196,
-      "grad_norm": 0.007132470607757568,
-      "learning_rate": 0.00019999951465417398,
-      "loss": 46.0,
-      "step": 12981
-    },
-    {
-      "epoch": 0.9925645583653497,
-      "grad_norm": 0.0007482675136998296,
-      "learning_rate": 0.00019999951457933562,
-      "loss": 46.0,
-      "step": 12982
-    },
-    {
-      "epoch": 0.9926410153487394,
-      "grad_norm": 0.005840052384883165,
-      "learning_rate": 0.0001999995145044915,
-      "loss": 46.0,
-      "step": 12983
-    },
-    {
-      "epoch": 0.9927174723321291,
-      "grad_norm": 0.0007353696273639798,
-      "learning_rate": 0.0001999995144296416,
-      "loss": 46.0,
-      "step": 12984
-    },
-    {
-      "epoch": 0.9927939293155189,
-      "grad_norm": 0.0005675255670212209,
-      "learning_rate": 0.00019999951435478594,
-      "loss": 46.0,
-      "step": 12985
-    },
-    {
-      "epoch": 0.9928703862989086,
-      "grad_norm": 0.0013363094767555594,
-      "learning_rate": 0.00019999951427992452,
-      "loss": 46.0,
-      "step": 12986
-    },
-    {
-      "epoch": 0.9929468432822983,
-      "grad_norm": 0.0011422872776165605,
-      "learning_rate": 0.00019999951420505733,
-      "loss": 46.0,
-      "step": 12987
-    },
-    {
-      "epoch": 0.993023300265688,
-      "grad_norm": 0.000899986713193357,
-      "learning_rate": 0.00019999951413018433,
-      "loss": 46.0,
-      "step": 12988
-    },
-    {
-      "epoch": 0.9930997572490777,
-      "grad_norm": 0.001702246954664588,
-      "learning_rate": 0.00019999951405530561,
-      "loss": 46.0,
-      "step": 12989
-    },
-    {
-      "epoch": 0.9931762142324675,
-      "grad_norm": 0.0013333183014765382,
-      "learning_rate": 0.0001999995139804211,
-      "loss": 46.0,
-      "step": 12990
-    },
-    {
-      "epoch": 0.9932526712158571,
-      "grad_norm": 0.0005960882408544421,
-      "learning_rate": 0.00019999951390553084,
-      "loss": 46.0,
-      "step": 12991
-    },
-    {
-      "epoch": 0.9933291281992469,
-      "grad_norm": 0.0015765996649861336,
-      "learning_rate": 0.00019999951383063477,
-      "loss": 46.0,
-      "step": 12992
-    },
-    {
-      "epoch": 0.9934055851826367,
-      "grad_norm": 0.001805830281227827,
-      "learning_rate": 0.00019999951375573296,
-      "loss": 46.0,
-      "step": 12993
-    },
-    {
-      "epoch": 0.9934820421660263,
-      "grad_norm": 0.0014324636431410909,
-      "learning_rate": 0.00019999951368082535,
-      "loss": 46.0,
-      "step": 12994
-    },
-    {
-      "epoch": 0.9935584991494161,
-      "grad_norm": 0.0017241350142285228,
-      "learning_rate": 0.00019999951360591203,
-      "loss": 46.0,
-      "step": 12995
-    },
-    {
-      "epoch": 0.9936349561328058,
-      "grad_norm": 0.004104727879166603,
-      "learning_rate": 0.0001999995135309929,
-      "loss": 46.0,
-      "step": 12996
-    },
-    {
-      "epoch": 0.9937114131161955,
-      "grad_norm": 0.000588090973906219,
-      "learning_rate": 0.00019999951345606797,
-      "loss": 46.0,
-      "step": 12997
-    },
-    {
-      "epoch": 0.9937878700995852,
-      "grad_norm": 0.0009377379319630563,
-      "learning_rate": 0.00019999951338113735,
-      "loss": 46.0,
-      "step": 12998
-    },
-    {
-      "epoch": 0.9938643270829749,
-      "grad_norm": 0.00223111012019217,
-      "learning_rate": 0.00019999951330620087,
-      "loss": 46.0,
-      "step": 12999
-    },
-    {
-      "epoch": 0.9939407840663647,
-      "grad_norm": 0.0014962464338168502,
-      "learning_rate": 0.0001999995132312587,
-      "loss": 46.0,
-      "step": 13000
-    },
-    {
-      "epoch": 0.9940172410497544,
-      "grad_norm": 0.0025067706592381,
-      "learning_rate": 0.0001999995131563107,
-      "loss": 46.0,
-      "step": 13001
-    },
-    {
-      "epoch": 0.9940936980331441,
-      "grad_norm": 0.0007811136892996728,
-      "learning_rate": 0.00019999951308135696,
-      "loss": 46.0,
-      "step": 13002
-    },
-    {
-      "epoch": 0.9941701550165338,
-      "grad_norm": 0.0003123429778497666,
-      "learning_rate": 0.00019999951300639745,
-      "loss": 46.0,
-      "step": 13003
-    },
-    {
-      "epoch": 0.9942466119999236,
-      "grad_norm": 0.0007085145334713161,
-      "learning_rate": 0.00019999951293143219,
-      "loss": 46.0,
-      "step": 13004
-    },
-    {
-      "epoch": 0.9943230689833132,
-      "grad_norm": 0.0009960256284102798,
-      "learning_rate": 0.00019999951285646112,
-      "loss": 46.0,
-      "step": 13005
-    },
-    {
-      "epoch": 0.994399525966703,
-      "grad_norm": 0.001250269589945674,
-      "learning_rate": 0.0001999995127814843,
-      "loss": 46.0,
-      "step": 13006
-    },
-    {
-      "epoch": 0.9944759829500927,
-      "grad_norm": 0.0006877474370412529,
-      "learning_rate": 0.00019999951270650173,
-      "loss": 46.0,
-      "step": 13007
-    },
-    {
-      "epoch": 0.9945524399334824,
-      "grad_norm": 0.00038921754457987845,
-      "learning_rate": 0.00019999951263151338,
-      "loss": 46.0,
-      "step": 13008
-    },
-    {
-      "epoch": 0.9946288969168722,
-      "grad_norm": 0.006857279222458601,
-      "learning_rate": 0.00019999951255651922,
-      "loss": 46.0,
-      "step": 13009
-    },
-    {
-      "epoch": 0.9947053539002618,
-      "grad_norm": 0.0006055301637388766,
-      "learning_rate": 0.00019999951248151935,
-      "loss": 46.0,
-      "step": 13010
-    },
-    {
-      "epoch": 0.9947818108836516,
-      "grad_norm": 0.004898465238511562,
-      "learning_rate": 0.00019999951240651367,
-      "loss": 46.0,
-      "step": 13011
-    },
-    {
-      "epoch": 0.9948582678670413,
-      "grad_norm": 0.0009586650994606316,
-      "learning_rate": 0.00019999951233150225,
-      "loss": 46.0,
-      "step": 13012
-    },
-    {
-      "epoch": 0.994934724850431,
-      "grad_norm": 0.005516965873539448,
-      "learning_rate": 0.00019999951225648503,
-      "loss": 46.0,
-      "step": 13013
-    },
-    {
-      "epoch": 0.9950111818338208,
-      "grad_norm": 0.0011231043608859181,
-      "learning_rate": 0.00019999951218146206,
-      "loss": 46.0,
-      "step": 13014
-    },
-    {
-      "epoch": 0.9950876388172105,
-      "grad_norm": 0.000755190965719521,
-      "learning_rate": 0.00019999951210643332,
-      "loss": 46.0,
-      "step": 13015
-    },
-    {
-      "epoch": 0.9951640958006002,
-      "grad_norm": 0.001508427201770246,
-      "learning_rate": 0.0001999995120313988,
-      "loss": 46.0,
-      "step": 13016
-    },
-    {
-      "epoch": 0.9952405527839899,
-      "grad_norm": 0.0011397219495847821,
-      "learning_rate": 0.0001999995119563585,
-      "loss": 46.0,
-      "step": 13017
-    },
-    {
-      "epoch": 0.9953170097673796,
-      "grad_norm": 0.0012535445857793093,
-      "learning_rate": 0.00019999951188131245,
-      "loss": 46.0,
-      "step": 13018
-    },
-    {
-      "epoch": 0.9953934667507693,
-      "grad_norm": 0.00167899951338768,
-      "learning_rate": 0.00019999951180626064,
-      "loss": 46.0,
-      "step": 13019
-    },
-    {
-      "epoch": 0.9954699237341591,
-      "grad_norm": 0.0012216983595862985,
-      "learning_rate": 0.00019999951173120306,
-      "loss": 46.0,
-      "step": 13020
-    },
-    {
-      "epoch": 0.9955463807175488,
-      "grad_norm": 0.0009084090124815702,
-      "learning_rate": 0.00019999951165613968,
-      "loss": 46.0,
-      "step": 13021
-    },
-    {
-      "epoch": 0.9956228377009385,
-      "grad_norm": 0.0032085846178233624,
-      "learning_rate": 0.00019999951158107055,
-      "loss": 46.0,
-      "step": 13022
-    },
-    {
-      "epoch": 0.9956992946843283,
-      "grad_norm": 0.0009710604790598154,
-      "learning_rate": 0.00019999951150599565,
-      "loss": 46.0,
-      "step": 13023
-    },
-    {
-      "epoch": 0.9957757516677179,
-      "grad_norm": 0.0004906139802187681,
-      "learning_rate": 0.000199999511430915,
-      "loss": 46.0,
-      "step": 13024
-    },
-    {
-      "epoch": 0.9958522086511077,
-      "grad_norm": 0.0015517802676185966,
-      "learning_rate": 0.00019999951135582852,
-      "loss": 46.0,
-      "step": 13025
-    },
-    {
-      "epoch": 0.9959286656344974,
-      "grad_norm": 0.0008460269309580326,
-      "learning_rate": 0.00019999951128073633,
-      "loss": 46.0,
-      "step": 13026
-    },
-    {
-      "epoch": 0.9960051226178871,
-      "grad_norm": 0.003025216283276677,
-      "learning_rate": 0.00019999951120563836,
-      "loss": 46.0,
-      "step": 13027
-    },
-    {
-      "epoch": 0.9960815796012769,
-      "grad_norm": 0.0029430841095745564,
-      "learning_rate": 0.00019999951113053462,
-      "loss": 46.0,
-      "step": 13028
-    },
-    {
-      "epoch": 0.9961580365846665,
-      "grad_norm": 0.0014440554659813643,
-      "learning_rate": 0.0001999995110554251,
-      "loss": 46.0,
-      "step": 13029
-    },
-    {
-      "epoch": 0.9962344935680563,
-      "grad_norm": 0.0015573466662317514,
-      "learning_rate": 0.0001999995109803098,
-      "loss": 46.0,
-      "step": 13030
-    },
-    {
-      "epoch": 0.996310950551446,
-      "grad_norm": 0.0021596390288323164,
-      "learning_rate": 0.00019999951090518875,
-      "loss": 46.0,
-      "step": 13031
-    },
-    {
-      "epoch": 0.9963874075348357,
-      "grad_norm": 0.003103324444964528,
-      "learning_rate": 0.00019999951083006195,
-      "loss": 46.0,
-      "step": 13032
-    },
-    {
-      "epoch": 0.9964638645182254,
-      "grad_norm": 0.0006132619455456734,
-      "learning_rate": 0.00019999951075492934,
-      "loss": 46.0,
-      "step": 13033
-    },
-    {
-      "epoch": 0.9965403215016152,
-      "grad_norm": 0.0036434493958950043,
-      "learning_rate": 0.00019999951067979096,
-      "loss": 46.0,
-      "step": 13034
-    },
-    {
-      "epoch": 0.9966167784850049,
-      "grad_norm": 0.005200007930397987,
-      "learning_rate": 0.00019999951060464683,
-      "loss": 46.0,
-      "step": 13035
-    },
-    {
-      "epoch": 0.9966932354683946,
-      "grad_norm": 0.0008730081608518958,
-      "learning_rate": 0.00019999951052949695,
-      "loss": 46.0,
-      "step": 13036
-    },
-    {
-      "epoch": 0.9967696924517843,
-      "grad_norm": 0.000802867638412863,
-      "learning_rate": 0.00019999951045434128,
-      "loss": 46.0,
-      "step": 13037
-    },
-    {
-      "epoch": 0.996846149435174,
-      "grad_norm": 0.0015725247794762254,
-      "learning_rate": 0.0001999995103791798,
-      "loss": 46.0,
-      "step": 13038
-    },
-    {
-      "epoch": 0.9969226064185638,
-      "grad_norm": 0.0009133379790000618,
-      "learning_rate": 0.0001999995103040126,
-      "loss": 46.0,
-      "step": 13039
-    },
-    {
-      "epoch": 0.9969990634019534,
-      "grad_norm": 0.0016108219278976321,
-      "learning_rate": 0.00019999951022883961,
-      "loss": 46.0,
-      "step": 13040
-    },
-    {
-      "epoch": 0.9970755203853432,
-      "grad_norm": 0.0011954944347962737,
-      "learning_rate": 0.00019999951015366087,
-      "loss": 46.0,
-      "step": 13041
-    },
-    {
-      "epoch": 0.997151977368733,
-      "grad_norm": 0.0008768549305386841,
-      "learning_rate": 0.00019999951007847636,
-      "loss": 46.0,
-      "step": 13042
-    },
-    {
-      "epoch": 0.9972284343521226,
-      "grad_norm": 0.0012309534940868616,
-      "learning_rate": 0.00019999951000328607,
-      "loss": 46.0,
-      "step": 13043
-    },
-    {
-      "epoch": 0.9973048913355124,
-      "grad_norm": 0.0010584152769297361,
-      "learning_rate": 0.00019999950992809004,
-      "loss": 46.0,
-      "step": 13044
-    },
-    {
-      "epoch": 0.9973813483189021,
-      "grad_norm": 0.0006732388865202665,
-      "learning_rate": 0.00019999950985288817,
-      "loss": 46.0,
-      "step": 13045
-    },
-    {
-      "epoch": 0.9974578053022918,
-      "grad_norm": 0.0024629042018204927,
-      "learning_rate": 0.0001999995097776806,
-      "loss": 46.0,
-      "step": 13046
-    },
-    {
-      "epoch": 0.9975342622856815,
-      "grad_norm": 0.0008411156595684588,
-      "learning_rate": 0.0001999995097024672,
-      "loss": 46.0,
-      "step": 13047
-    },
-    {
-      "epoch": 0.9976107192690712,
-      "grad_norm": 0.006107902619987726,
-      "learning_rate": 0.00019999950962724808,
-      "loss": 46.0,
-      "step": 13048
-    },
-    {
-      "epoch": 0.997687176252461,
-      "grad_norm": 0.0011961539275944233,
-      "learning_rate": 0.00019999950955202318,
-      "loss": 46.0,
-      "step": 13049
-    },
-    {
-      "epoch": 0.9977636332358507,
-      "grad_norm": 0.0010772492969408631,
-      "learning_rate": 0.0001999995094767925,
-      "loss": 46.0,
-      "step": 13050
-    },
-    {
-      "epoch": 0.9978400902192404,
-      "grad_norm": 0.0015731386374682188,
-      "learning_rate": 0.00019999950940155606,
-      "loss": 46.0,
-      "step": 13051
-    },
-    {
-      "epoch": 0.9979165472026301,
-      "grad_norm": 0.0006390456692315638,
-      "learning_rate": 0.00019999950932631384,
-      "loss": 46.0,
-      "step": 13052
-    },
-    {
-      "epoch": 0.9979930041860199,
-      "grad_norm": 0.0008038360392674804,
-      "learning_rate": 0.00019999950925106584,
-      "loss": 46.0,
-      "step": 13053
-    },
-    {
-      "epoch": 0.9980694611694095,
-      "grad_norm": 0.0027667242102324963,
-      "learning_rate": 0.0001999995091758121,
-      "loss": 46.0,
-      "step": 13054
-    },
-    {
-      "epoch": 0.9981459181527993,
-      "grad_norm": 0.0018887505866587162,
-      "learning_rate": 0.00019999950910055262,
-      "loss": 46.0,
-      "step": 13055
-    },
-    {
-      "epoch": 0.998222375136189,
-      "grad_norm": 0.001135754631832242,
-      "learning_rate": 0.0001999995090252873,
-      "loss": 46.0,
-      "step": 13056
-    },
-    {
-      "epoch": 0.9982988321195787,
-      "grad_norm": 0.0006053533870726824,
-      "learning_rate": 0.00019999950895001624,
-      "loss": 46.0,
-      "step": 13057
-    },
-    {
-      "epoch": 0.9983752891029685,
-      "grad_norm": 0.0021726926788687706,
-      "learning_rate": 0.0001999995088747394,
-      "loss": 46.0,
-      "step": 13058
-    },
-    {
-      "epoch": 0.9984517460863581,
-      "grad_norm": 0.0007626822334714234,
-      "learning_rate": 0.00019999950879945682,
-      "loss": 46.0,
-      "step": 13059
-    },
-    {
-      "epoch": 0.9985282030697479,
-      "grad_norm": 0.014977667480707169,
-      "learning_rate": 0.00019999950872416844,
-      "loss": 46.0,
-      "step": 13060
-    },
-    {
-      "epoch": 0.9986046600531376,
-      "grad_norm": 0.0008592166122980416,
-      "learning_rate": 0.00019999950864887432,
-      "loss": 46.0,
-      "step": 13061
-    },
-    {
-      "epoch": 0.9986811170365273,
-      "grad_norm": 0.004858432337641716,
-      "learning_rate": 0.00019999950857357442,
-      "loss": 46.0,
-      "step": 13062
-    },
-    {
-      "epoch": 0.998757574019917,
-      "grad_norm": 0.0010461441706866026,
-      "learning_rate": 0.00019999950849826871,
-      "loss": 46.0,
-      "step": 13063
-    },
-    {
-      "epoch": 0.9988340310033068,
-      "grad_norm": 0.0016068416880443692,
-      "learning_rate": 0.00019999950842295727,
-      "loss": 46.0,
-      "step": 13064
-    },
-    {
-      "epoch": 0.9989104879866965,
-      "grad_norm": 0.0017768868710845709,
-      "learning_rate": 0.00019999950834764007,
-      "loss": 46.0,
-      "step": 13065
-    },
-    {
-      "epoch": 0.9989869449700862,
-      "grad_norm": 0.001869536703452468,
-      "learning_rate": 0.00019999950827231708,
-      "loss": 46.0,
-      "step": 13066
-    },
-    {
-      "epoch": 0.9990634019534759,
-      "grad_norm": 0.0028397131245583296,
-      "learning_rate": 0.0001999995081969883,
-      "loss": 46.0,
-      "step": 13067
-    },
-    {
-      "epoch": 0.9991398589368656,
-      "grad_norm": 0.006175707560032606,
-      "learning_rate": 0.00019999950812165377,
-      "loss": 46.0,
-      "step": 13068
-    },
-    {
-      "epoch": 0.9992163159202554,
-      "grad_norm": 0.0005917723174206913,
-      "learning_rate": 0.0001999995080463135,
-      "loss": 46.0,
-      "step": 13069
-    },
-    {
-      "epoch": 0.999292772903645,
-      "grad_norm": 0.001969987526535988,
-      "learning_rate": 0.00019999950797096742,
-      "loss": 46.0,
-      "step": 13070
-    },
-    {
-      "epoch": 0.9993692298870348,
-      "grad_norm": 0.0016098202904686332,
-      "learning_rate": 0.00019999950789561561,
-      "loss": 46.0,
-      "step": 13071
-    },
-    {
-      "epoch": 0.9994456868704246,
-      "grad_norm": 0.002032308839261532,
-      "learning_rate": 0.000199999507820258,
-      "loss": 46.0,
-      "step": 13072
-    },
-    {
-      "epoch": 0.9995221438538142,
-      "grad_norm": 0.003767122281715274,
-      "learning_rate": 0.00019999950774489463,
-      "loss": 46.0,
-      "step": 13073
-    },
-    {
-      "epoch": 0.999598600837204,
-      "grad_norm": 0.0013361971359699965,
-      "learning_rate": 0.0001999995076695255,
-      "loss": 46.0,
-      "step": 13074
-    },
-    {
-      "epoch": 0.9996750578205937,
-      "grad_norm": 0.0003991764970123768,
-      "learning_rate": 0.00019999950759415057,
-      "loss": 46.0,
-      "step": 13075
-    },
-    {
-      "epoch": 0.9997515148039834,
-      "grad_norm": 0.0013790132943540812,
-      "learning_rate": 0.0001999995075187699,
-      "loss": 46.0,
-      "step": 13076
-    },
-    {
-      "epoch": 0.9998279717873731,
-      "grad_norm": 0.004516522400081158,
-      "learning_rate": 0.00019999950744338345,
-      "loss": 46.0,
-      "step": 13077
-    },
-    {
-      "epoch": 0.9999044287707628,
-      "grad_norm": 0.0006289720768108964,
-      "learning_rate": 0.00019999950736799123,
-      "loss": 46.0,
-      "step": 13078
-    },
-    {
-      "epoch": 0.9999808857541526,
-      "grad_norm": 0.000514974060934037,
-      "learning_rate": 0.00019999950729259323,
-      "loss": 46.0,
-      "step": 13079
-    },
-    {
-      "epoch": 0.9999808857541526,
-      "eval_loss": 11.5,
-      "eval_runtime": 31.2973,
-      "eval_samples_per_second": 175.99,
-      "eval_steps_per_second": 87.995,
-      "step": 13079
-    },
-    {
-      "epoch": 1.0000573427375423,
-      "grad_norm": 0.0016246383311226964,
-      "learning_rate": 0.0001999995072171895,
-      "loss": 46.0,
-      "step": 13080
-    },
-    {
-      "epoch": 1.000133799720932,
-      "grad_norm": 0.0010487493127584457,
-      "learning_rate": 0.00019999950714177995,
-      "loss": 46.0,
-      "step": 13081
-    },
-    {
-      "epoch": 1.0002102567043216,
-      "grad_norm": 0.015009724535048008,
-      "learning_rate": 0.00019999950706636467,
-      "loss": 46.0,
-      "step": 13082
-    },
-    {
-      "epoch": 1.0002867136877114,
-      "grad_norm": 0.0030437775421887636,
-      "learning_rate": 0.00019999950699094358,
-      "loss": 46.0,
-      "step": 13083
-    },
-    {
-      "epoch": 1.0003631706711011,
-      "grad_norm": 0.003403489477932453,
-      "learning_rate": 0.00019999950691551677,
-      "loss": 46.0,
-      "step": 13084
-    },
-    {
-      "epoch": 1.000439627654491,
-      "grad_norm": 0.0054220883175730705,
-      "learning_rate": 0.00019999950684008416,
-      "loss": 46.0,
-      "step": 13085
-    },
-    {
-      "epoch": 1.0005160846378807,
-      "grad_norm": 0.0017335679149255157,
-      "learning_rate": 0.00019999950676464578,
-      "loss": 46.0,
-      "step": 13086
-    },
-    {
-      "epoch": 1.0005925416212704,
-      "grad_norm": 0.0010313049424439669,
-      "learning_rate": 0.00019999950668920163,
-      "loss": 46.0,
-      "step": 13087
-    },
-    {
-      "epoch": 1.00066899860466,
-      "grad_norm": 0.006911544594913721,
-      "learning_rate": 0.00019999950661375173,
-      "loss": 46.0,
-      "step": 13088
-    },
-    {
-      "epoch": 1.0007454555880497,
-      "grad_norm": 0.002955998294055462,
-      "learning_rate": 0.00019999950653829606,
-      "loss": 46.0,
-      "step": 13089
-    },
-    {
-      "epoch": 1.0008219125714395,
-      "grad_norm": 0.001319556962698698,
-      "learning_rate": 0.0001999995064628346,
-      "loss": 46.0,
-      "step": 13090
-    },
-    {
-      "epoch": 1.0008983695548292,
-      "grad_norm": 0.0020879770163446665,
-      "learning_rate": 0.00019999950638736736,
-      "loss": 46.0,
-      "step": 13091
-    },
-    {
-      "epoch": 1.000974826538219,
-      "grad_norm": 0.00044731638627126813,
-      "learning_rate": 0.00019999950631189437,
-      "loss": 46.0,
-      "step": 13092
-    },
-    {
-      "epoch": 1.0010512835216085,
-      "grad_norm": 0.0010726484470069408,
-      "learning_rate": 0.00019999950623641563,
-      "loss": 46.0,
-      "step": 13093
-    },
-    {
-      "epoch": 1.0011277405049983,
-      "grad_norm": 0.00800649356096983,
-      "learning_rate": 0.0001999995061609311,
-      "loss": 46.0,
-      "step": 13094
-    },
-    {
-      "epoch": 1.001204197488388,
-      "grad_norm": 0.000554498634301126,
-      "learning_rate": 0.0001999995060854408,
-      "loss": 46.0,
-      "step": 13095
-    },
-    {
-      "epoch": 1.0012806544717778,
-      "grad_norm": 0.0008250930113717914,
-      "learning_rate": 0.00019999950600994471,
-      "loss": 46.0,
-      "step": 13096
-    },
-    {
-      "epoch": 1.0013571114551676,
-      "grad_norm": 0.0049845450557768345,
-      "learning_rate": 0.00019999950593444288,
-      "loss": 46.0,
-      "step": 13097
-    },
-    {
-      "epoch": 1.0014335684385574,
-      "grad_norm": 0.0013318548444658518,
-      "learning_rate": 0.00019999950585893527,
-      "loss": 46.0,
-      "step": 13098
-    },
-    {
-      "epoch": 1.001510025421947,
-      "grad_norm": 0.0008032101904973388,
-      "learning_rate": 0.0001999995057834219,
-      "loss": 46.0,
-      "step": 13099
-    },
-    {
-      "epoch": 1.0015864824053367,
-      "grad_norm": 0.0021659035701304674,
-      "learning_rate": 0.00019999950570790277,
-      "loss": 46.0,
-      "step": 13100
-    },
-    {
-      "epoch": 1.0016629393887264,
-      "grad_norm": 0.0013529403368011117,
-      "learning_rate": 0.00019999950563237784,
-      "loss": 46.0,
-      "step": 13101
-    },
-    {
-      "epoch": 1.0017393963721162,
-      "grad_norm": 0.0020392779260873795,
-      "learning_rate": 0.00019999950555684717,
-      "loss": 46.0,
-      "step": 13102
-    },
-    {
-      "epoch": 1.001815853355506,
-      "grad_norm": 0.0012425929307937622,
-      "learning_rate": 0.00019999950548131072,
-      "loss": 46.0,
-      "step": 13103
-    },
-    {
-      "epoch": 1.0018923103388955,
-      "grad_norm": 0.015269949100911617,
-      "learning_rate": 0.00019999950540576847,
-      "loss": 46.0,
-      "step": 13104
-    },
-    {
-      "epoch": 1.0019687673222852,
-      "grad_norm": 0.0005601951852440834,
-      "learning_rate": 0.0001999995053302205,
-      "loss": 46.0,
-      "step": 13105
-    },
-    {
-      "epoch": 1.002045224305675,
-      "grad_norm": 0.0028679859824478626,
-      "learning_rate": 0.00019999950525466674,
-      "loss": 46.0,
-      "step": 13106
-    },
-    {
-      "epoch": 1.0021216812890648,
-      "grad_norm": 0.0007895689341239631,
-      "learning_rate": 0.0001999995051791072,
-      "loss": 46.0,
-      "step": 13107
-    },
-    {
-      "epoch": 1.0021981382724545,
-      "grad_norm": 0.0011641173623502254,
-      "learning_rate": 0.00019999950510354189,
-      "loss": 46.0,
-      "step": 13108
-    },
-    {
-      "epoch": 1.0022745952558443,
-      "grad_norm": 0.0010575753403827548,
-      "learning_rate": 0.00019999950502797083,
-      "loss": 46.0,
-      "step": 13109
-    },
-    {
-      "epoch": 1.0023510522392338,
-      "grad_norm": 0.0012260411167517304,
-      "learning_rate": 0.000199999504952394,
-      "loss": 46.0,
-      "step": 13110
-    },
-    {
-      "epoch": 1.0024275092226236,
-      "grad_norm": 0.0023909483570605516,
-      "learning_rate": 0.00019999950487681139,
-      "loss": 46.0,
-      "step": 13111
-    },
-    {
-      "epoch": 1.0025039662060133,
-      "grad_norm": 0.00043188981362618506,
-      "learning_rate": 0.000199999504801223,
-      "loss": 46.0,
-      "step": 13112
-    },
-    {
-      "epoch": 1.002580423189403,
-      "grad_norm": 0.0014382440131157637,
-      "learning_rate": 0.00019999950472562885,
-      "loss": 46.0,
-      "step": 13113
-    },
-    {
-      "epoch": 1.0026568801727929,
-      "grad_norm": 0.0017269690288230777,
-      "learning_rate": 0.00019999950465002893,
-      "loss": 46.0,
-      "step": 13114
-    },
-    {
-      "epoch": 1.0027333371561824,
-      "grad_norm": 0.002471608342602849,
-      "learning_rate": 0.00019999950457442325,
-      "loss": 46.0,
-      "step": 13115
-    },
-    {
-      "epoch": 1.0028097941395722,
-      "grad_norm": 0.0006464608595706522,
-      "learning_rate": 0.0001999995044988118,
-      "loss": 46.0,
-      "step": 13116
-    },
-    {
-      "epoch": 1.002886251122962,
-      "grad_norm": 0.0016540633514523506,
-      "learning_rate": 0.00019999950442319456,
-      "loss": 46.0,
-      "step": 13117
-    },
-    {
-      "epoch": 1.0029627081063517,
-      "grad_norm": 0.0014798224437981844,
-      "learning_rate": 0.00019999950434757157,
-      "loss": 46.0,
-      "step": 13118
-    },
-    {
-      "epoch": 1.0030391650897414,
-      "grad_norm": 0.002955448580905795,
-      "learning_rate": 0.0001999995042719428,
-      "loss": 46.0,
-      "step": 13119
-    },
-    {
-      "epoch": 1.0031156220731312,
-      "grad_norm": 0.0012454278767108917,
-      "learning_rate": 0.00019999950419630826,
-      "loss": 46.0,
-      "step": 13120
-    },
-    {
-      "epoch": 1.0031920790565207,
-      "grad_norm": 0.0008011875906959176,
-      "learning_rate": 0.00019999950412066797,
-      "loss": 46.0,
-      "step": 13121
-    },
-    {
-      "epoch": 1.0032685360399105,
-      "grad_norm": 0.0013702984433621168,
-      "learning_rate": 0.00019999950404502191,
-      "loss": 46.0,
-      "step": 13122
-    },
-    {
-      "epoch": 1.0033449930233003,
-      "grad_norm": 0.0023872603196650743,
-      "learning_rate": 0.00019999950396937005,
-      "loss": 46.0,
-      "step": 13123
-    },
-    {
-      "epoch": 1.00342145000669,
-      "grad_norm": 0.002514572348445654,
-      "learning_rate": 0.00019999950389371245,
-      "loss": 46.0,
-      "step": 13124
-    },
-    {
-      "epoch": 1.0034979069900798,
-      "grad_norm": 0.002194985980167985,
-      "learning_rate": 0.00019999950381804907,
-      "loss": 46.0,
-      "step": 13125
-    },
-    {
-      "epoch": 1.0035743639734693,
-      "grad_norm": 0.0006772606866434216,
-      "learning_rate": 0.0001999995037423799,
-      "loss": 46.0,
-      "step": 13126
-    },
-    {
-      "epoch": 1.003650820956859,
-      "grad_norm": 0.0011174071114510298,
-      "learning_rate": 0.00019999950366670496,
-      "loss": 46.0,
-      "step": 13127
-    },
-    {
-      "epoch": 1.0037272779402489,
-      "grad_norm": 0.00036141840973868966,
-      "learning_rate": 0.00019999950359102432,
-      "loss": 46.0,
-      "step": 13128
-    },
-    {
-      "epoch": 1.0038037349236386,
-      "grad_norm": 0.001346914446912706,
-      "learning_rate": 0.00019999950351533782,
-      "loss": 46.0,
-      "step": 13129
-    },
-    {
-      "epoch": 1.0038801919070284,
-      "grad_norm": 0.0006093299016356468,
-      "learning_rate": 0.00019999950343964562,
-      "loss": 46.0,
-      "step": 13130
-    },
-    {
-      "epoch": 1.0039566488904181,
-      "grad_norm": 0.0038393668364733458,
-      "learning_rate": 0.0001999995033639476,
-      "loss": 46.0,
-      "step": 13131
-    },
-    {
-      "epoch": 1.0040331058738077,
-      "grad_norm": 0.0022960868664085865,
-      "learning_rate": 0.00019999950328824384,
-      "loss": 46.0,
-      "step": 13132
-    },
-    {
-      "epoch": 1.0041095628571974,
-      "grad_norm": 0.0014905341668054461,
-      "learning_rate": 0.0001999995032125343,
-      "loss": 46.0,
-      "step": 13133
-    },
-    {
-      "epoch": 1.0041860198405872,
-      "grad_norm": 0.0013899640180170536,
-      "learning_rate": 0.000199999503136819,
-      "loss": 46.0,
-      "step": 13134
-    },
-    {
-      "epoch": 1.004262476823977,
-      "grad_norm": 0.0014219989534467459,
-      "learning_rate": 0.00019999950306109793,
-      "loss": 46.0,
-      "step": 13135
-    },
-    {
-      "epoch": 1.0043389338073667,
-      "grad_norm": 0.012358224019408226,
-      "learning_rate": 0.0001999995029853711,
-      "loss": 46.0,
-      "step": 13136
-    },
-    {
-      "epoch": 1.0044153907907563,
-      "grad_norm": 0.0007344927871599793,
-      "learning_rate": 0.00019999950290963846,
-      "loss": 46.0,
-      "step": 13137
-    },
-    {
-      "epoch": 1.004491847774146,
-      "grad_norm": 0.002723429352045059,
-      "learning_rate": 0.00019999950283390008,
-      "loss": 46.0,
-      "step": 13138
-    },
-    {
-      "epoch": 1.0045683047575358,
-      "grad_norm": 0.003247004933655262,
-      "learning_rate": 0.0001999995027581559,
-      "loss": 46.0,
-      "step": 13139
-    },
-    {
-      "epoch": 1.0046447617409255,
-      "grad_norm": 0.0010048558469861746,
-      "learning_rate": 0.000199999502682406,
-      "loss": 46.0,
-      "step": 13140
-    },
-    {
-      "epoch": 1.0047212187243153,
-      "grad_norm": 0.0008429336594417691,
-      "learning_rate": 0.0001999995026066503,
-      "loss": 46.0,
-      "step": 13141
-    },
-    {
-      "epoch": 1.004797675707705,
-      "grad_norm": 0.0013114274479448795,
-      "learning_rate": 0.00019999950253088886,
-      "loss": 46.0,
-      "step": 13142
-    },
-    {
-      "epoch": 1.0048741326910946,
-      "grad_norm": 0.005130447447299957,
-      "learning_rate": 0.0001999995024551216,
-      "loss": 46.0,
-      "step": 13143
-    },
-    {
-      "epoch": 1.0049505896744844,
-      "grad_norm": 0.0170944444835186,
-      "learning_rate": 0.00019999950237934862,
-      "loss": 46.0,
-      "step": 13144
-    },
-    {
-      "epoch": 1.0050270466578741,
-      "grad_norm": 0.0013413139386102557,
-      "learning_rate": 0.00019999950230356985,
-      "loss": 46.0,
-      "step": 13145
-    },
-    {
-      "epoch": 1.0051035036412639,
-      "grad_norm": 0.0005037693190388381,
-      "learning_rate": 0.00019999950222778531,
-      "loss": 46.0,
-      "step": 13146
-    },
-    {
-      "epoch": 1.0051799606246536,
-      "grad_norm": 0.0011129215126857162,
-      "learning_rate": 0.000199999502151995,
-      "loss": 46.0,
-      "step": 13147
-    },
-    {
-      "epoch": 1.0052564176080432,
-      "grad_norm": 0.001147244474850595,
-      "learning_rate": 0.00019999950207619894,
-      "loss": 46.0,
-      "step": 13148
-    },
-    {
-      "epoch": 1.005332874591433,
-      "grad_norm": 0.0036876939702779055,
-      "learning_rate": 0.00019999950200039708,
-      "loss": 46.0,
-      "step": 13149
-    },
-    {
-      "epoch": 1.0054093315748227,
-      "grad_norm": 0.0028461748734116554,
-      "learning_rate": 0.00019999950192458945,
-      "loss": 46.0,
-      "step": 13150
-    },
-    {
-      "epoch": 1.0054857885582125,
-      "grad_norm": 0.0005899151437915862,
-      "learning_rate": 0.00019999950184877604,
-      "loss": 46.0,
-      "step": 13151
-    },
-    {
-      "epoch": 1.0055622455416022,
-      "grad_norm": 0.00482617923989892,
-      "learning_rate": 0.00019999950177295692,
-      "loss": 46.0,
-      "step": 13152
-    },
-    {
-      "epoch": 1.0056387025249918,
-      "grad_norm": 0.003291077446192503,
-      "learning_rate": 0.00019999950169713197,
-      "loss": 46.0,
-      "step": 13153
-    },
-    {
-      "epoch": 1.0057151595083815,
-      "grad_norm": 0.0009646614198572934,
-      "learning_rate": 0.0001999995016213013,
-      "loss": 46.0,
-      "step": 13154
-    },
-    {
-      "epoch": 1.0057916164917713,
-      "grad_norm": 0.0016465745866298676,
-      "learning_rate": 0.00019999950154546482,
-      "loss": 46.0,
-      "step": 13155
-    },
-    {
-      "epoch": 1.005868073475161,
-      "grad_norm": 0.0007563810795545578,
-      "learning_rate": 0.00019999950146962258,
-      "loss": 46.0,
-      "step": 13156
-    },
-    {
-      "epoch": 1.0059445304585508,
-      "grad_norm": 0.0047136046923696995,
-      "learning_rate": 0.00019999950139377458,
-      "loss": 46.0,
-      "step": 13157
-    },
-    {
-      "epoch": 1.0060209874419406,
-      "grad_norm": 0.006196219008415937,
-      "learning_rate": 0.00019999950131792082,
-      "loss": 46.0,
-      "step": 13158
-    },
-    {
-      "epoch": 1.0060974444253301,
-      "grad_norm": 0.008243533782660961,
-      "learning_rate": 0.00019999950124206128,
-      "loss": 46.0,
-      "step": 13159
-    },
-    {
-      "epoch": 1.0061739014087199,
-      "grad_norm": 0.0009325897553935647,
-      "learning_rate": 0.00019999950116619597,
-      "loss": 46.0,
-      "step": 13160
-    },
-    {
-      "epoch": 1.0062503583921096,
-      "grad_norm": 0.0012052697129547596,
-      "learning_rate": 0.00019999950109032488,
-      "loss": 46.0,
-      "step": 13161
-    },
-    {
-      "epoch": 1.0063268153754994,
-      "grad_norm": 0.0006250923033803701,
-      "learning_rate": 0.00019999950101444802,
-      "loss": 46.0,
-      "step": 13162
-    },
-    {
-      "epoch": 1.0064032723588892,
-      "grad_norm": 0.0007843942730687559,
-      "learning_rate": 0.00019999950093856542,
-      "loss": 46.0,
-      "step": 13163
-    },
-    {
-      "epoch": 1.0064797293422787,
-      "grad_norm": 0.0008219070732593536,
-      "learning_rate": 0.00019999950086267704,
-      "loss": 46.0,
-      "step": 13164
-    },
-    {
-      "epoch": 1.0065561863256685,
-      "grad_norm": 0.0008874632185325027,
-      "learning_rate": 0.00019999950078678286,
-      "loss": 46.0,
-      "step": 13165
-    },
-    {
-      "epoch": 1.0066326433090582,
-      "grad_norm": 0.0010848186211660504,
-      "learning_rate": 0.00019999950071088293,
-      "loss": 46.0,
-      "step": 13166
-    },
-    {
-      "epoch": 1.006709100292448,
-      "grad_norm": 0.0026227065827697515,
-      "learning_rate": 0.00019999950063497726,
-      "loss": 46.0,
-      "step": 13167
-    },
-    {
-      "epoch": 1.0067855572758377,
-      "grad_norm": 0.0010908190160989761,
-      "learning_rate": 0.0001999995005590658,
-      "loss": 46.0,
-      "step": 13168
-    },
-    {
-      "epoch": 1.0068620142592275,
-      "grad_norm": 0.005107661709189415,
-      "learning_rate": 0.00019999950048314852,
-      "loss": 46.0,
-      "step": 13169
-    },
-    {
-      "epoch": 1.006938471242617,
-      "grad_norm": 0.0012098741717636585,
-      "learning_rate": 0.00019999950040722552,
-      "loss": 46.0,
-      "step": 13170
-    },
-    {
-      "epoch": 1.0070149282260068,
-      "grad_norm": 0.0022341248113662004,
-      "learning_rate": 0.00019999950033129676,
-      "loss": 46.0,
-      "step": 13171
-    },
-    {
-      "epoch": 1.0070913852093966,
-      "grad_norm": 0.002104635816067457,
-      "learning_rate": 0.0001999995002553622,
-      "loss": 46.0,
-      "step": 13172
-    },
-    {
-      "epoch": 1.0071678421927863,
-      "grad_norm": 0.0008151038200594485,
-      "learning_rate": 0.00019999950017942188,
-      "loss": 46.0,
-      "step": 13173
-    },
-    {
-      "epoch": 1.007244299176176,
-      "grad_norm": 0.0011404327815398574,
-      "learning_rate": 0.00019999950010347582,
-      "loss": 46.0,
-      "step": 13174
-    },
-    {
-      "epoch": 1.0073207561595656,
-      "grad_norm": 0.001159876468591392,
-      "learning_rate": 0.00019999950002752396,
-      "loss": 46.0,
-      "step": 13175
-    },
-    {
-      "epoch": 1.0073972131429554,
-      "grad_norm": 0.0019495253218337893,
-      "learning_rate": 0.00019999949995156633,
-      "loss": 46.0,
-      "step": 13176
-    },
-    {
-      "epoch": 1.0074736701263451,
-      "grad_norm": 0.0019082440994679928,
-      "learning_rate": 0.00019999949987560295,
-      "loss": 46.0,
-      "step": 13177
-    },
-    {
-      "epoch": 1.007550127109735,
-      "grad_norm": 0.0011245633941143751,
-      "learning_rate": 0.00019999949979963378,
-      "loss": 46.0,
-      "step": 13178
-    },
-    {
-      "epoch": 1.0076265840931247,
-      "grad_norm": 0.0007572720642201602,
-      "learning_rate": 0.00019999949972365885,
-      "loss": 46.0,
-      "step": 13179
-    },
-    {
-      "epoch": 1.0077030410765144,
-      "grad_norm": 0.001434324774891138,
-      "learning_rate": 0.00019999949964767818,
-      "loss": 46.0,
-      "step": 13180
-    },
-    {
-      "epoch": 1.007779498059904,
-      "grad_norm": 0.0024929086212068796,
-      "learning_rate": 0.00019999949957169168,
-      "loss": 46.0,
-      "step": 13181
-    },
-    {
-      "epoch": 1.0078559550432937,
-      "grad_norm": 0.0015253580641001463,
-      "learning_rate": 0.00019999949949569946,
-      "loss": 46.0,
-      "step": 13182
-    },
-    {
-      "epoch": 1.0079324120266835,
-      "grad_norm": 0.01165300328284502,
-      "learning_rate": 0.00019999949941970144,
-      "loss": 46.0,
-      "step": 13183
-    },
-    {
-      "epoch": 1.0080088690100732,
-      "grad_norm": 0.0026303925551474094,
-      "learning_rate": 0.00019999949934369765,
-      "loss": 46.0,
-      "step": 13184
-    },
-    {
-      "epoch": 1.008085325993463,
-      "grad_norm": 0.001410634838975966,
-      "learning_rate": 0.00019999949926768814,
-      "loss": 46.0,
-      "step": 13185
-    },
-    {
-      "epoch": 1.0081617829768525,
-      "grad_norm": 0.0008460862445645034,
-      "learning_rate": 0.0001999994991916728,
-      "loss": 46.0,
-      "step": 13186
-    },
-    {
-      "epoch": 1.0082382399602423,
-      "grad_norm": 0.0015290474984794855,
-      "learning_rate": 0.0001999994991156517,
-      "loss": 46.0,
-      "step": 13187
-    },
-    {
-      "epoch": 1.008314696943632,
-      "grad_norm": 0.009030179120600224,
-      "learning_rate": 0.00019999949903962488,
-      "loss": 46.0,
-      "step": 13188
-    },
-    {
-      "epoch": 1.0083911539270218,
-      "grad_norm": 0.0005173471290618181,
-      "learning_rate": 0.00019999949896359223,
-      "loss": 46.0,
-      "step": 13189
-    },
-    {
-      "epoch": 1.0084676109104116,
-      "grad_norm": 0.015396231785416603,
-      "learning_rate": 0.00019999949888755385,
-      "loss": 46.0,
-      "step": 13190
-    },
-    {
-      "epoch": 1.0085440678938014,
-      "grad_norm": 0.001245530555024743,
-      "learning_rate": 0.00019999949881150967,
-      "loss": 46.0,
-      "step": 13191
-    },
-    {
-      "epoch": 1.008620524877191,
-      "grad_norm": 0.0016823558835312724,
-      "learning_rate": 0.00019999949873545974,
-      "loss": 46.0,
-      "step": 13192
-    },
-    {
-      "epoch": 1.0086969818605807,
-      "grad_norm": 0.0013860964681953192,
-      "learning_rate": 0.00019999949865940405,
-      "loss": 46.0,
-      "step": 13193
-    },
-    {
-      "epoch": 1.0087734388439704,
-      "grad_norm": 0.0006061822059564292,
-      "learning_rate": 0.00019999949858334257,
-      "loss": 46.0,
-      "step": 13194
-    },
-    {
-      "epoch": 1.0088498958273602,
-      "grad_norm": 0.0014511196641251445,
-      "learning_rate": 0.00019999949850727533,
-      "loss": 46.0,
-      "step": 13195
-    },
-    {
-      "epoch": 1.00892635281075,
-      "grad_norm": 0.012013912200927734,
-      "learning_rate": 0.0001999994984312023,
-      "loss": 46.0,
-      "step": 13196
-    },
-    {
-      "epoch": 1.0090028097941395,
-      "grad_norm": 0.0006593851139768958,
-      "learning_rate": 0.00019999949835512352,
-      "loss": 46.0,
-      "step": 13197
-    },
-    {
-      "epoch": 1.0090792667775292,
-      "grad_norm": 0.001966786803677678,
-      "learning_rate": 0.00019999949827903898,
-      "loss": 46.0,
-      "step": 13198
-    },
-    {
-      "epoch": 1.009155723760919,
-      "grad_norm": 0.0030168199446052313,
-      "learning_rate": 0.00019999949820294867,
-      "loss": 46.0,
-      "step": 13199
-    },
-    {
-      "epoch": 1.0092321807443088,
-      "grad_norm": 0.0005610954249277711,
-      "learning_rate": 0.00019999949812685256,
-      "loss": 46.0,
-      "step": 13200
-    },
-    {
-      "epoch": 1.0093086377276985,
-      "grad_norm": 0.004524660762399435,
-      "learning_rate": 0.00019999949805075073,
-      "loss": 46.0,
-      "step": 13201
-    },
-    {
-      "epoch": 1.0093850947110883,
-      "grad_norm": 0.0009732217877171934,
-      "learning_rate": 0.00019999949797464307,
-      "loss": 46.0,
-      "step": 13202
-    },
-    {
-      "epoch": 1.0094615516944778,
-      "grad_norm": 0.0009303790284320712,
-      "learning_rate": 0.0001999994978985297,
-      "loss": 46.0,
-      "step": 13203
-    },
-    {
-      "epoch": 1.0095380086778676,
-      "grad_norm": 0.0014858862850815058,
-      "learning_rate": 0.00019999949782241054,
-      "loss": 46.0,
-      "step": 13204
-    },
-    {
-      "epoch": 1.0096144656612573,
-      "grad_norm": 0.013205749914050102,
-      "learning_rate": 0.0001999994977462856,
-      "loss": 46.0,
-      "step": 13205
-    },
-    {
-      "epoch": 1.009690922644647,
-      "grad_norm": 0.0023184148594737053,
-      "learning_rate": 0.0001999994976701549,
-      "loss": 46.0,
-      "step": 13206
-    },
-    {
-      "epoch": 1.0097673796280369,
-      "grad_norm": 0.0007458882755599916,
-      "learning_rate": 0.00019999949759401842,
-      "loss": 46.0,
-      "step": 13207
-    },
-    {
-      "epoch": 1.0098438366114264,
-      "grad_norm": 0.0008488466846756637,
-      "learning_rate": 0.0001999994975178762,
-      "loss": 46.0,
-      "step": 13208
-    },
-    {
-      "epoch": 1.0099202935948162,
-      "grad_norm": 0.0012341069523245096,
-      "learning_rate": 0.00019999949744172816,
-      "loss": 46.0,
-      "step": 13209
-    },
-    {
-      "epoch": 1.009996750578206,
-      "grad_norm": 0.0029805428348481655,
-      "learning_rate": 0.00019999949736557437,
-      "loss": 46.0,
-      "step": 13210
-    },
-    {
-      "epoch": 1.0100732075615957,
-      "grad_norm": 0.000832177815027535,
-      "learning_rate": 0.00019999949728941483,
-      "loss": 46.0,
-      "step": 13211
-    },
-    {
-      "epoch": 1.0101496645449854,
-      "grad_norm": 0.0015722366515547037,
-      "learning_rate": 0.00019999949721324952,
-      "loss": 46.0,
-      "step": 13212
-    },
-    {
-      "epoch": 1.0102261215283752,
-      "grad_norm": 0.00213468074798584,
-      "learning_rate": 0.0001999994971370784,
-      "loss": 46.0,
-      "step": 13213
-    },
-    {
-      "epoch": 1.0103025785117647,
-      "grad_norm": 0.000951695314142853,
-      "learning_rate": 0.00019999949706090155,
-      "loss": 46.0,
-      "step": 13214
-    },
-    {
-      "epoch": 1.0103790354951545,
-      "grad_norm": 0.0011788548436015844,
-      "learning_rate": 0.00019999949698471894,
-      "loss": 46.0,
-      "step": 13215
-    },
-    {
-      "epoch": 1.0104554924785443,
-      "grad_norm": 0.0008633707766421139,
-      "learning_rate": 0.00019999949690853054,
-      "loss": 46.0,
-      "step": 13216
-    },
-    {
-      "epoch": 1.010531949461934,
-      "grad_norm": 0.002745748497545719,
-      "learning_rate": 0.00019999949683233633,
-      "loss": 46.0,
-      "step": 13217
-    },
-    {
-      "epoch": 1.0106084064453238,
-      "grad_norm": 0.0008100612321868539,
-      "learning_rate": 0.0001999994967561364,
-      "loss": 46.0,
-      "step": 13218
-    },
-    {
-      "epoch": 1.0106848634287133,
-      "grad_norm": 0.0012199390912428498,
-      "learning_rate": 0.00019999949667993068,
-      "loss": 46.0,
-      "step": 13219
-    },
-    {
-      "epoch": 1.010761320412103,
-      "grad_norm": 0.0004941097577102482,
-      "learning_rate": 0.00019999949660371921,
-      "loss": 46.0,
-      "step": 13220
-    },
-    {
-      "epoch": 1.0108377773954929,
-      "grad_norm": 0.0009790302719920874,
-      "learning_rate": 0.000199999496527502,
-      "loss": 46.0,
-      "step": 13221
-    },
-    {
-      "epoch": 1.0109142343788826,
-      "grad_norm": 0.0009613094734959304,
-      "learning_rate": 0.00019999949645127898,
-      "loss": 46.0,
-      "step": 13222
-    },
-    {
-      "epoch": 1.0109906913622724,
-      "grad_norm": 0.0006092821713536978,
-      "learning_rate": 0.0001999994963750502,
-      "loss": 46.0,
-      "step": 13223
-    },
-    {
-      "epoch": 1.011067148345662,
-      "grad_norm": 0.0016754708485677838,
-      "learning_rate": 0.00019999949629881562,
-      "loss": 46.0,
-      "step": 13224
-    },
-    {
-      "epoch": 1.0111436053290517,
-      "grad_norm": 0.0004980962257832289,
-      "learning_rate": 0.0001999994962225753,
-      "loss": 46.0,
-      "step": 13225
-    },
-    {
-      "epoch": 1.0112200623124414,
-      "grad_norm": 0.0007395015563815832,
-      "learning_rate": 0.00019999949614632923,
-      "loss": 46.0,
-      "step": 13226
-    },
-    {
-      "epoch": 1.0112965192958312,
-      "grad_norm": 0.0007635345100425184,
-      "learning_rate": 0.00019999949607007734,
-      "loss": 46.0,
-      "step": 13227
-    },
-    {
-      "epoch": 1.011372976279221,
-      "grad_norm": 0.000687832129187882,
-      "learning_rate": 0.0001999994959938197,
-      "loss": 46.0,
-      "step": 13228
-    },
-    {
-      "epoch": 1.0114494332626107,
-      "grad_norm": 0.0007053890149109066,
-      "learning_rate": 0.0001999994959175563,
-      "loss": 46.0,
-      "step": 13229
-    },
-    {
-      "epoch": 1.0115258902460003,
-      "grad_norm": 0.0016934218583628535,
-      "learning_rate": 0.00019999949584128713,
-      "loss": 46.0,
-      "step": 13230
-    },
-    {
-      "epoch": 1.01160234722939,
-      "grad_norm": 0.00045626488281413913,
-      "learning_rate": 0.0001999994957650122,
-      "loss": 46.0,
-      "step": 13231
-    },
-    {
-      "epoch": 1.0116788042127798,
-      "grad_norm": 0.0004521886585280299,
-      "learning_rate": 0.00019999949568873148,
-      "loss": 46.0,
-      "step": 13232
-    },
-    {
-      "epoch": 1.0117552611961695,
-      "grad_norm": 0.0012220500502735376,
-      "learning_rate": 0.00019999949561244499,
-      "loss": 46.0,
-      "step": 13233
-    },
-    {
-      "epoch": 1.0118317181795593,
-      "grad_norm": 0.0003151035634800792,
-      "learning_rate": 0.00019999949553615277,
-      "loss": 46.0,
-      "step": 13234
-    },
-    {
-      "epoch": 1.0119081751629488,
-      "grad_norm": 0.003835513722151518,
-      "learning_rate": 0.00019999949545985475,
-      "loss": 46.0,
-      "step": 13235
-    },
-    {
-      "epoch": 1.0119846321463386,
-      "grad_norm": 0.005143537186086178,
-      "learning_rate": 0.00019999949538355096,
-      "loss": 46.0,
-      "step": 13236
-    },
-    {
-      "epoch": 1.0120610891297284,
-      "grad_norm": 0.0028145890682935715,
-      "learning_rate": 0.0001999994953072414,
-      "loss": 46.0,
-      "step": 13237
-    },
-    {
-      "epoch": 1.0121375461131181,
-      "grad_norm": 0.00553949223831296,
-      "learning_rate": 0.00019999949523092606,
-      "loss": 46.0,
-      "step": 13238
-    },
-    {
-      "epoch": 1.0122140030965079,
-      "grad_norm": 0.0038574826903641224,
-      "learning_rate": 0.00019999949515460497,
-      "loss": 46.0,
-      "step": 13239
-    },
-    {
-      "epoch": 1.0122904600798976,
-      "grad_norm": 0.0019628724548965693,
-      "learning_rate": 0.00019999949507827812,
-      "loss": 46.0,
-      "step": 13240
-    },
-    {
-      "epoch": 1.0123669170632872,
-      "grad_norm": 0.0009287242428399622,
-      "learning_rate": 0.0001999994950019455,
-      "loss": 46.0,
-      "step": 13241
-    },
-    {
-      "epoch": 1.012443374046677,
-      "grad_norm": 0.0003255531773902476,
-      "learning_rate": 0.00019999949492560708,
-      "loss": 46.0,
-      "step": 13242
-    },
-    {
-      "epoch": 1.0125198310300667,
-      "grad_norm": 0.000495244050398469,
-      "learning_rate": 0.0001999994948492629,
-      "loss": 46.0,
-      "step": 13243
-    },
-    {
-      "epoch": 1.0125962880134565,
-      "grad_norm": 0.003493133233860135,
-      "learning_rate": 0.00019999949477291296,
-      "loss": 46.0,
-      "step": 13244
-    },
-    {
-      "epoch": 1.0126727449968462,
-      "grad_norm": 0.0012319928500801325,
-      "learning_rate": 0.00019999949469655723,
-      "loss": 46.0,
-      "step": 13245
-    },
-    {
-      "epoch": 1.0127492019802358,
-      "grad_norm": 0.008995313197374344,
-      "learning_rate": 0.00019999949462019576,
-      "loss": 46.0,
-      "step": 13246
-    },
-    {
-      "epoch": 1.0128256589636255,
-      "grad_norm": 0.0009984509088099003,
-      "learning_rate": 0.00019999949454382852,
-      "loss": 46.0,
-      "step": 13247
-    },
-    {
-      "epoch": 1.0129021159470153,
-      "grad_norm": 0.001639196532778442,
-      "learning_rate": 0.0001999994944674555,
-      "loss": 46.0,
-      "step": 13248
-    },
-    {
-      "epoch": 1.012978572930405,
-      "grad_norm": 0.0011900431709364057,
-      "learning_rate": 0.0001999994943910767,
-      "loss": 46.0,
-      "step": 13249
-    },
-    {
-      "epoch": 1.0130550299137948,
-      "grad_norm": 0.00172789569478482,
-      "learning_rate": 0.00019999949431469215,
-      "loss": 46.0,
-      "step": 13250
-    },
-    {
-      "epoch": 1.0131314868971846,
-      "grad_norm": 0.0007873403374105692,
-      "learning_rate": 0.0001999994942383018,
-      "loss": 46.0,
-      "step": 13251
-    },
-    {
-      "epoch": 1.0132079438805741,
-      "grad_norm": 0.0007505659596063197,
-      "learning_rate": 0.0001999994941619057,
-      "loss": 46.0,
-      "step": 13252
-    },
-    {
-      "epoch": 1.0132844008639639,
-      "grad_norm": 0.0012867376208305359,
-      "learning_rate": 0.00019999949408550384,
-      "loss": 46.0,
-      "step": 13253
-    },
-    {
-      "epoch": 1.0133608578473536,
-      "grad_norm": 0.00035619662958197296,
-      "learning_rate": 0.0001999994940090962,
-      "loss": 46.0,
-      "step": 13254
-    },
-    {
-      "epoch": 1.0134373148307434,
-      "grad_norm": 0.0005880212993361056,
-      "learning_rate": 0.0001999994939326828,
-      "loss": 46.0,
-      "step": 13255
-    },
-    {
-      "epoch": 1.0135137718141332,
-      "grad_norm": 0.0019242139533162117,
-      "learning_rate": 0.00019999949385626363,
-      "loss": 46.0,
-      "step": 13256
-    },
-    {
-      "epoch": 1.0135902287975227,
-      "grad_norm": 0.0005807740380987525,
-      "learning_rate": 0.00019999949377983865,
-      "loss": 46.0,
-      "step": 13257
-    },
-    {
-      "epoch": 1.0136666857809125,
-      "grad_norm": 0.002704866463318467,
-      "learning_rate": 0.00019999949370340798,
-      "loss": 46.0,
-      "step": 13258
-    },
-    {
-      "epoch": 1.0137431427643022,
-      "grad_norm": 0.0008134253439493477,
-      "learning_rate": 0.00019999949362697149,
-      "loss": 46.0,
-      "step": 13259
-    },
-    {
-      "epoch": 1.013819599747692,
-      "grad_norm": 0.0030122932512313128,
-      "learning_rate": 0.00019999949355052922,
-      "loss": 46.0,
-      "step": 13260
-    },
-    {
-      "epoch": 1.0138960567310817,
-      "grad_norm": 0.0007891881396062672,
-      "learning_rate": 0.0001999994934740812,
-      "loss": 46.0,
-      "step": 13261
-    },
-    {
-      "epoch": 1.0139725137144715,
-      "grad_norm": 0.0003403721784707159,
-      "learning_rate": 0.00019999949339762744,
-      "loss": 46.0,
-      "step": 13262
-    },
-    {
-      "epoch": 1.014048970697861,
-      "grad_norm": 0.0008272282429970801,
-      "learning_rate": 0.00019999949332116785,
-      "loss": 46.0,
-      "step": 13263
-    },
-    {
-      "epoch": 1.0141254276812508,
-      "grad_norm": 0.0017194539541378617,
-      "learning_rate": 0.0001999994932447025,
-      "loss": 46.0,
-      "step": 13264
-    },
-    {
-      "epoch": 1.0142018846646406,
-      "grad_norm": 0.0006940565654076636,
-      "learning_rate": 0.0001999994931682314,
-      "loss": 46.0,
-      "step": 13265
-    },
-    {
-      "epoch": 1.0142783416480303,
-      "grad_norm": 0.0007200895925052464,
-      "learning_rate": 0.00019999949309175454,
-      "loss": 46.0,
-      "step": 13266
-    },
-    {
-      "epoch": 1.01435479863142,
-      "grad_norm": 0.003316490910947323,
-      "learning_rate": 0.0001999994930152719,
-      "loss": 46.0,
-      "step": 13267
-    },
-    {
-      "epoch": 1.0144312556148096,
-      "grad_norm": 0.017602525651454926,
-      "learning_rate": 0.00019999949293878348,
-      "loss": 46.0,
-      "step": 13268
-    },
-    {
-      "epoch": 1.0145077125981994,
-      "grad_norm": 0.001880124444141984,
-      "learning_rate": 0.0001999994928622893,
-      "loss": 46.0,
-      "step": 13269
-    },
-    {
-      "epoch": 1.0145841695815891,
-      "grad_norm": 0.0006637490005232394,
-      "learning_rate": 0.00019999949278578938,
-      "loss": 46.0,
-      "step": 13270
-    },
-    {
-      "epoch": 1.014660626564979,
-      "grad_norm": 0.0006572098936885595,
-      "learning_rate": 0.00019999949270928363,
-      "loss": 46.0,
-      "step": 13271
-    },
-    {
-      "epoch": 1.0147370835483687,
-      "grad_norm": 0.0007339658332057297,
-      "learning_rate": 0.0001999994926327722,
-      "loss": 46.0,
-      "step": 13272
-    },
-    {
-      "epoch": 1.0148135405317584,
-      "grad_norm": 0.002393106697127223,
-      "learning_rate": 0.0001999994925562549,
-      "loss": 46.0,
-      "step": 13273
-    },
-    {
-      "epoch": 1.014889997515148,
-      "grad_norm": 0.00019928716938011348,
-      "learning_rate": 0.00019999949247973187,
-      "loss": 46.0,
-      "step": 13274
-    },
-    {
-      "epoch": 1.0149664544985377,
-      "grad_norm": 0.0009872004156932235,
-      "learning_rate": 0.0001999994924032031,
-      "loss": 46.0,
-      "step": 13275
-    },
-    {
-      "epoch": 1.0150429114819275,
-      "grad_norm": 0.003996880725026131,
-      "learning_rate": 0.00019999949232666852,
-      "loss": 46.0,
-      "step": 13276
-    },
-    {
-      "epoch": 1.0151193684653173,
-      "grad_norm": 0.0033545023761689663,
-      "learning_rate": 0.00019999949225012819,
-      "loss": 46.0,
-      "step": 13277
-    },
-    {
-      "epoch": 1.015195825448707,
-      "grad_norm": 0.0006512168911285698,
-      "learning_rate": 0.00019999949217358208,
-      "loss": 46.0,
-      "step": 13278
-    },
-    {
-      "epoch": 1.0152722824320966,
-      "grad_norm": 0.00299880956299603,
-      "learning_rate": 0.00019999949209703022,
-      "loss": 46.0,
-      "step": 13279
-    },
-    {
-      "epoch": 1.0153487394154863,
-      "grad_norm": 0.008543360978364944,
-      "learning_rate": 0.00019999949202047256,
-      "loss": 46.0,
-      "step": 13280
-    },
-    {
-      "epoch": 1.015425196398876,
-      "grad_norm": 0.003519149264320731,
-      "learning_rate": 0.00019999949194390916,
-      "loss": 46.0,
-      "step": 13281
-    },
-    {
-      "epoch": 1.0155016533822658,
-      "grad_norm": 0.0004817867302335799,
-      "learning_rate": 0.00019999949186733998,
-      "loss": 46.0,
-      "step": 13282
-    },
-    {
-      "epoch": 1.0155781103656556,
-      "grad_norm": 0.004248388111591339,
-      "learning_rate": 0.00019999949179076503,
-      "loss": 46.0,
-      "step": 13283
-    },
-    {
-      "epoch": 1.0156545673490451,
-      "grad_norm": 0.0012315039057284594,
-      "learning_rate": 0.0001999994917141843,
-      "loss": 46.0,
-      "step": 13284
-    },
-    {
-      "epoch": 1.015731024332435,
-      "grad_norm": 0.0005654659471474588,
-      "learning_rate": 0.00019999949163759781,
-      "loss": 46.0,
-      "step": 13285
-    },
-    {
-      "epoch": 1.0158074813158247,
-      "grad_norm": 0.0018208416877314448,
-      "learning_rate": 0.00019999949156100554,
-      "loss": 46.0,
-      "step": 13286
-    },
-    {
-      "epoch": 1.0158839382992144,
-      "grad_norm": 0.0006994802388362586,
-      "learning_rate": 0.00019999949148440753,
-      "loss": 46.0,
-      "step": 13287
-    },
-    {
-      "epoch": 1.0159603952826042,
-      "grad_norm": 0.004237137734889984,
-      "learning_rate": 0.00019999949140780374,
-      "loss": 46.0,
-      "step": 13288
-    },
-    {
-      "epoch": 1.016036852265994,
-      "grad_norm": 0.0010248580947518349,
-      "learning_rate": 0.00019999949133119418,
-      "loss": 46.0,
-      "step": 13289
-    },
-    {
-      "epoch": 1.0161133092493835,
-      "grad_norm": 0.01344903465360403,
-      "learning_rate": 0.00019999949125457884,
-      "loss": 46.0,
-      "step": 13290
-    },
-    {
-      "epoch": 1.0161897662327732,
-      "grad_norm": 0.0041988142766058445,
-      "learning_rate": 0.00019999949117795773,
-      "loss": 46.0,
-      "step": 13291
-    },
-    {
-      "epoch": 1.016266223216163,
-      "grad_norm": 0.0003930846869479865,
-      "learning_rate": 0.00019999949110133085,
-      "loss": 46.0,
-      "step": 13292
-    },
-    {
-      "epoch": 1.0163426801995528,
-      "grad_norm": 0.0017059551319107413,
-      "learning_rate": 0.0001999994910246982,
-      "loss": 46.0,
-      "step": 13293
-    },
-    {
-      "epoch": 1.0164191371829425,
-      "grad_norm": 0.0024940853472799063,
-      "learning_rate": 0.0001999994909480598,
-      "loss": 46.0,
-      "step": 13294
-    },
-    {
-      "epoch": 1.016495594166332,
-      "grad_norm": 0.01116472389549017,
-      "learning_rate": 0.0001999994908714156,
-      "loss": 46.0,
-      "step": 13295
-    },
-    {
-      "epoch": 1.0165720511497218,
-      "grad_norm": 0.0014680898748338223,
-      "learning_rate": 0.00019999949079476566,
-      "loss": 46.0,
-      "step": 13296
-    },
-    {
-      "epoch": 1.0166485081331116,
-      "grad_norm": 0.000381072866730392,
-      "learning_rate": 0.00019999949071810994,
-      "loss": 46.0,
-      "step": 13297
-    },
-    {
-      "epoch": 1.0167249651165013,
-      "grad_norm": 0.0006147310486994684,
-      "learning_rate": 0.00019999949064144844,
-      "loss": 46.0,
-      "step": 13298
-    },
-    {
-      "epoch": 1.016801422099891,
-      "grad_norm": 0.0005943061551079154,
-      "learning_rate": 0.00019999949056478117,
-      "loss": 46.0,
-      "step": 13299
-    },
-    {
-      "epoch": 1.0168778790832809,
-      "grad_norm": 0.0005558032426051795,
-      "learning_rate": 0.00019999949048810813,
-      "loss": 46.0,
-      "step": 13300
-    },
-    {
-      "epoch": 1.0169543360666704,
-      "grad_norm": 0.003409595927223563,
-      "learning_rate": 0.00019999949041142934,
-      "loss": 46.0,
-      "step": 13301
-    },
-    {
-      "epoch": 1.0170307930500602,
-      "grad_norm": 0.0012485369807109237,
-      "learning_rate": 0.00019999949033474478,
-      "loss": 46.0,
-      "step": 13302
-    },
-    {
-      "epoch": 1.01710725003345,
-      "grad_norm": 0.0011931437766179442,
-      "learning_rate": 0.00019999949025805442,
-      "loss": 46.0,
-      "step": 13303
-    },
-    {
-      "epoch": 1.0171837070168397,
-      "grad_norm": 0.0007169584278017282,
-      "learning_rate": 0.00019999949018135833,
-      "loss": 46.0,
-      "step": 13304
-    },
-    {
-      "epoch": 1.0172601640002294,
-      "grad_norm": 0.0010900953784585,
-      "learning_rate": 0.00019999949010465645,
-      "loss": 46.0,
-      "step": 13305
-    },
-    {
-      "epoch": 1.017336620983619,
-      "grad_norm": 0.000607198104262352,
-      "learning_rate": 0.0001999994900279488,
-      "loss": 46.0,
-      "step": 13306
-    },
-    {
-      "epoch": 1.0174130779670087,
-      "grad_norm": 0.001631378778256476,
-      "learning_rate": 0.0001999994899512354,
-      "loss": 46.0,
-      "step": 13307
-    },
-    {
-      "epoch": 1.0174895349503985,
-      "grad_norm": 0.0014307759702205658,
-      "learning_rate": 0.0001999994898745162,
-      "loss": 46.0,
-      "step": 13308
-    },
-    {
-      "epoch": 1.0175659919337883,
-      "grad_norm": 0.0016589186852797866,
-      "learning_rate": 0.00019999948979779124,
-      "loss": 46.0,
-      "step": 13309
-    },
-    {
-      "epoch": 1.017642448917178,
-      "grad_norm": 0.0007353859255090356,
-      "learning_rate": 0.00019999948972106052,
-      "loss": 46.0,
-      "step": 13310
-    },
-    {
-      "epoch": 1.0177189059005678,
-      "grad_norm": 0.0008505401201546192,
-      "learning_rate": 0.00019999948964432402,
-      "loss": 46.0,
-      "step": 13311
-    },
-    {
-      "epoch": 1.0177953628839573,
-      "grad_norm": 0.0011174505343660712,
-      "learning_rate": 0.00019999948956758175,
-      "loss": 46.0,
-      "step": 13312
-    },
-    {
-      "epoch": 1.017871819867347,
-      "grad_norm": 0.0049111987464129925,
-      "learning_rate": 0.00019999948949083374,
-      "loss": 46.0,
-      "step": 13313
-    },
-    {
-      "epoch": 1.0179482768507369,
-      "grad_norm": 0.0017285100184381008,
-      "learning_rate": 0.00019999948941407992,
-      "loss": 46.0,
-      "step": 13314
-    },
-    {
-      "epoch": 1.0180247338341266,
-      "grad_norm": 0.004865770693868399,
-      "learning_rate": 0.00019999948933732036,
-      "loss": 46.0,
-      "step": 13315
-    },
-    {
-      "epoch": 1.0181011908175164,
-      "grad_norm": 0.0007946233381517231,
-      "learning_rate": 0.000199999489260555,
-      "loss": 46.0,
-      "step": 13316
-    },
-    {
-      "epoch": 1.018177647800906,
-      "grad_norm": 0.0014810017310082912,
-      "learning_rate": 0.0001999994891837839,
-      "loss": 46.0,
-      "step": 13317
-    },
-    {
-      "epoch": 1.0182541047842957,
-      "grad_norm": 0.000969009764958173,
-      "learning_rate": 0.000199999489107007,
-      "loss": 46.0,
-      "step": 13318
-    },
-    {
-      "epoch": 1.0183305617676854,
-      "grad_norm": 0.0005195743287913501,
-      "learning_rate": 0.00019999948903022435,
-      "loss": 46.0,
-      "step": 13319
-    },
-    {
-      "epoch": 1.0184070187510752,
-      "grad_norm": 0.0011100205592811108,
-      "learning_rate": 0.00019999948895343595,
-      "loss": 46.0,
-      "step": 13320
-    },
-    {
-      "epoch": 1.018483475734465,
-      "grad_norm": 0.000961338693741709,
-      "learning_rate": 0.00019999948887664175,
-      "loss": 46.0,
-      "step": 13321
-    },
-    {
-      "epoch": 1.0185599327178547,
-      "grad_norm": 0.0011881001992151141,
-      "learning_rate": 0.0001999994887998418,
-      "loss": 46.0,
-      "step": 13322
-    },
-    {
-      "epoch": 1.0186363897012443,
-      "grad_norm": 0.00022212143812794238,
-      "learning_rate": 0.00019999948872303607,
-      "loss": 46.0,
-      "step": 13323
-    },
-    {
-      "epoch": 1.018712846684634,
-      "grad_norm": 0.000817084452137351,
-      "learning_rate": 0.00019999948864622458,
-      "loss": 46.0,
-      "step": 13324
-    },
-    {
-      "epoch": 1.0187893036680238,
-      "grad_norm": 0.0029416445177048445,
-      "learning_rate": 0.0001999994885694073,
-      "loss": 46.0,
-      "step": 13325
-    },
-    {
-      "epoch": 1.0188657606514135,
-      "grad_norm": 0.0013460393529385328,
-      "learning_rate": 0.00019999948849258427,
-      "loss": 46.0,
-      "step": 13326
-    },
-    {
-      "epoch": 1.0189422176348033,
-      "grad_norm": 0.001525513012893498,
-      "learning_rate": 0.00019999948841575545,
-      "loss": 46.0,
-      "step": 13327
-    },
-    {
-      "epoch": 1.0190186746181928,
-      "grad_norm": 0.0045488192699849606,
-      "learning_rate": 0.00019999948833892092,
-      "loss": 46.0,
-      "step": 13328
-    },
-    {
-      "epoch": 1.0190951316015826,
-      "grad_norm": 0.0013691463973373175,
-      "learning_rate": 0.00019999948826208055,
-      "loss": 46.0,
-      "step": 13329
-    },
-    {
-      "epoch": 1.0191715885849724,
-      "grad_norm": 0.0034634231124073267,
-      "learning_rate": 0.00019999948818523442,
-      "loss": 46.0,
-      "step": 13330
-    },
-    {
-      "epoch": 1.0192480455683621,
-      "grad_norm": 0.00534837506711483,
-      "learning_rate": 0.00019999948810838256,
-      "loss": 46.0,
-      "step": 13331
-    },
-    {
-      "epoch": 1.0193245025517519,
-      "grad_norm": 0.0017972958739846945,
-      "learning_rate": 0.00019999948803152488,
-      "loss": 46.0,
-      "step": 13332
-    },
-    {
-      "epoch": 1.0194009595351416,
-      "grad_norm": 0.0009605927625671029,
-      "learning_rate": 0.00019999948795466148,
-      "loss": 46.0,
-      "step": 13333
-    },
-    {
-      "epoch": 1.0194774165185312,
-      "grad_norm": 0.0020247125066816807,
-      "learning_rate": 0.00019999948787779228,
-      "loss": 46.0,
-      "step": 13334
-    },
-    {
-      "epoch": 1.019553873501921,
-      "grad_norm": 0.0022912139538675547,
-      "learning_rate": 0.00019999948780091733,
-      "loss": 46.0,
-      "step": 13335
-    },
-    {
-      "epoch": 1.0196303304853107,
-      "grad_norm": 0.003716285340487957,
-      "learning_rate": 0.00019999948772403658,
-      "loss": 46.0,
-      "step": 13336
-    },
-    {
-      "epoch": 1.0197067874687005,
-      "grad_norm": 0.0010237148962914944,
-      "learning_rate": 0.00019999948764715008,
-      "loss": 46.0,
-      "step": 13337
-    },
-    {
-      "epoch": 1.0197832444520902,
-      "grad_norm": 0.0006742149125784636,
-      "learning_rate": 0.00019999948757025782,
-      "loss": 46.0,
-      "step": 13338
-    },
-    {
-      "epoch": 1.0198597014354798,
-      "grad_norm": 0.0009522874606773257,
-      "learning_rate": 0.00019999948749335977,
-      "loss": 46.0,
-      "step": 13339
-    },
-    {
-      "epoch": 1.0199361584188695,
-      "grad_norm": 0.0011698524467647076,
-      "learning_rate": 0.00019999948741645599,
-      "loss": 46.0,
-      "step": 13340
-    },
-    {
-      "epoch": 1.0200126154022593,
-      "grad_norm": 0.0035062176175415516,
-      "learning_rate": 0.0001999994873395464,
-      "loss": 46.0,
-      "step": 13341
-    },
-    {
-      "epoch": 1.020089072385649,
-      "grad_norm": 0.002914591459557414,
-      "learning_rate": 0.00019999948726263104,
-      "loss": 46.0,
-      "step": 13342
-    },
-    {
-      "epoch": 1.0201655293690388,
-      "grad_norm": 0.019517388194799423,
-      "learning_rate": 0.00019999948718570995,
-      "loss": 46.0,
-      "step": 13343
-    },
-    {
-      "epoch": 1.0202419863524286,
-      "grad_norm": 0.0014743616338819265,
-      "learning_rate": 0.00019999948710878304,
-      "loss": 46.0,
-      "step": 13344
-    },
-    {
-      "epoch": 1.0203184433358181,
-      "grad_norm": 0.002358692465350032,
-      "learning_rate": 0.00019999948703185042,
-      "loss": 46.0,
-      "step": 13345
-    },
-    {
-      "epoch": 1.0203949003192079,
-      "grad_norm": 0.0009612751891836524,
-      "learning_rate": 0.00019999948695491196,
-      "loss": 46.0,
-      "step": 13346
-    },
-    {
-      "epoch": 1.0204713573025976,
-      "grad_norm": 0.0011470431927591562,
-      "learning_rate": 0.00019999948687796776,
-      "loss": 46.0,
-      "step": 13347
-    },
-    {
-      "epoch": 1.0205478142859874,
-      "grad_norm": 0.0013494361191987991,
-      "learning_rate": 0.0001999994868010178,
-      "loss": 46.0,
-      "step": 13348
-    },
-    {
-      "epoch": 1.0206242712693772,
-      "grad_norm": 0.000935558695346117,
-      "learning_rate": 0.0001999994867240621,
-      "loss": 46.0,
-      "step": 13349
-    },
-    {
-      "epoch": 1.0207007282527667,
-      "grad_norm": 0.0006529902457259595,
-      "learning_rate": 0.00019999948664710057,
-      "loss": 46.0,
-      "step": 13350
-    },
-    {
-      "epoch": 1.0207771852361565,
-      "grad_norm": 0.0035267509520053864,
-      "learning_rate": 0.0001999994865701333,
-      "loss": 46.0,
-      "step": 13351
-    },
-    {
-      "epoch": 1.0208536422195462,
-      "grad_norm": 0.0013048126129433513,
-      "learning_rate": 0.00019999948649316026,
-      "loss": 46.0,
-      "step": 13352
-    },
-    {
-      "epoch": 1.020930099202936,
-      "grad_norm": 0.013325194828212261,
-      "learning_rate": 0.00019999948641618147,
-      "loss": 46.0,
-      "step": 13353
-    },
-    {
-      "epoch": 1.0210065561863257,
-      "grad_norm": 0.000999304698780179,
-      "learning_rate": 0.00019999948633919688,
-      "loss": 46.0,
-      "step": 13354
-    },
-    {
-      "epoch": 1.0210830131697153,
-      "grad_norm": 0.0035504517145454884,
-      "learning_rate": 0.00019999948626220652,
-      "loss": 46.0,
-      "step": 13355
-    },
-    {
-      "epoch": 1.021159470153105,
-      "grad_norm": 0.00038704698090441525,
-      "learning_rate": 0.0001999994861852104,
-      "loss": 46.0,
-      "step": 13356
-    },
-    {
-      "epoch": 1.0212359271364948,
-      "grad_norm": 0.0003472507814876735,
-      "learning_rate": 0.00019999948610820853,
-      "loss": 46.0,
-      "step": 13357
-    },
-    {
-      "epoch": 1.0213123841198846,
-      "grad_norm": 0.0015108231455087662,
-      "learning_rate": 0.00019999948603120088,
-      "loss": 46.0,
-      "step": 13358
-    },
-    {
-      "epoch": 1.0213888411032743,
-      "grad_norm": 0.0012180168414488435,
-      "learning_rate": 0.00019999948595418742,
-      "loss": 46.0,
-      "step": 13359
-    },
-    {
-      "epoch": 1.021465298086664,
-      "grad_norm": 0.004298059735447168,
-      "learning_rate": 0.00019999948587716825,
-      "loss": 46.0,
-      "step": 13360
-    },
-    {
-      "epoch": 1.0215417550700536,
-      "grad_norm": 0.0006079935701563954,
-      "learning_rate": 0.00019999948580014327,
-      "loss": 46.0,
-      "step": 13361
-    },
-    {
-      "epoch": 1.0216182120534434,
-      "grad_norm": 0.009199530817568302,
-      "learning_rate": 0.00019999948572311252,
-      "loss": 46.0,
-      "step": 13362
-    },
-    {
-      "epoch": 1.0216946690368331,
-      "grad_norm": 0.0019859843887388706,
-      "learning_rate": 0.00019999948564607603,
-      "loss": 46.0,
-      "step": 13363
-    },
-    {
-      "epoch": 1.021771126020223,
-      "grad_norm": 0.0011677610455080867,
-      "learning_rate": 0.00019999948556903376,
-      "loss": 46.0,
-      "step": 13364
-    },
-    {
-      "epoch": 1.0218475830036127,
-      "grad_norm": 0.0007705016760155559,
-      "learning_rate": 0.00019999948549198572,
-      "loss": 46.0,
-      "step": 13365
-    },
-    {
-      "epoch": 1.0219240399870022,
-      "grad_norm": 0.002353726653382182,
-      "learning_rate": 0.0001999994854149319,
-      "loss": 46.0,
-      "step": 13366
-    },
-    {
-      "epoch": 1.022000496970392,
-      "grad_norm": 0.000638998462818563,
-      "learning_rate": 0.00019999948533787234,
-      "loss": 46.0,
-      "step": 13367
-    },
-    {
-      "epoch": 1.0220769539537817,
-      "grad_norm": 0.0008855049381963909,
-      "learning_rate": 0.00019999948526080695,
-      "loss": 46.0,
-      "step": 13368
-    },
-    {
-      "epoch": 1.0221534109371715,
-      "grad_norm": 0.0015744090778753161,
-      "learning_rate": 0.00019999948518373585,
-      "loss": 46.0,
-      "step": 13369
-    },
-    {
-      "epoch": 1.0222298679205613,
-      "grad_norm": 0.0008924994617700577,
-      "learning_rate": 0.00019999948510665897,
-      "loss": 46.0,
-      "step": 13370
-    },
-    {
-      "epoch": 1.022306324903951,
-      "grad_norm": 0.0006267475546337664,
-      "learning_rate": 0.00019999948502957628,
-      "loss": 46.0,
-      "step": 13371
-    },
-    {
-      "epoch": 1.0223827818873406,
-      "grad_norm": 0.0006603386136703193,
-      "learning_rate": 0.00019999948495248786,
-      "loss": 46.0,
-      "step": 13372
-    },
-    {
-      "epoch": 1.0224592388707303,
-      "grad_norm": 0.0005316564347594976,
-      "learning_rate": 0.00019999948487539365,
-      "loss": 46.0,
-      "step": 13373
-    },
-    {
-      "epoch": 1.02253569585412,
-      "grad_norm": 0.0010533621534705162,
-      "learning_rate": 0.00019999948479829368,
-      "loss": 46.0,
-      "step": 13374
-    },
-    {
-      "epoch": 1.0226121528375098,
-      "grad_norm": 0.005168898031115532,
-      "learning_rate": 0.00019999948472118796,
-      "loss": 46.0,
-      "step": 13375
-    },
-    {
-      "epoch": 1.0226886098208996,
-      "grad_norm": 0.0019177552312612534,
-      "learning_rate": 0.00019999948464407646,
-      "loss": 46.0,
-      "step": 13376
-    },
-    {
-      "epoch": 1.0227650668042891,
-      "grad_norm": 0.0007928262930363417,
-      "learning_rate": 0.00019999948456695917,
-      "loss": 46.0,
-      "step": 13377
-    },
-    {
-      "epoch": 1.022841523787679,
-      "grad_norm": 0.001959329703822732,
-      "learning_rate": 0.00019999948448983613,
-      "loss": 46.0,
-      "step": 13378
-    },
-    {
-      "epoch": 1.0229179807710687,
-      "grad_norm": 0.0010179789969697595,
-      "learning_rate": 0.00019999948441270731,
-      "loss": 46.0,
-      "step": 13379
-    },
-    {
-      "epoch": 1.0229944377544584,
-      "grad_norm": 0.0007430230616591871,
-      "learning_rate": 0.00019999948433557273,
-      "loss": 46.0,
-      "step": 13380
-    },
-    {
-      "epoch": 1.0230708947378482,
-      "grad_norm": 0.00038856040919199586,
-      "learning_rate": 0.00019999948425843237,
-      "loss": 46.0,
-      "step": 13381
-    },
-    {
-      "epoch": 1.023147351721238,
-      "grad_norm": 0.002925224369391799,
-      "learning_rate": 0.00019999948418128626,
-      "loss": 46.0,
-      "step": 13382
-    },
-    {
-      "epoch": 1.0232238087046275,
-      "grad_norm": 0.008724424056708813,
-      "learning_rate": 0.00019999948410413435,
-      "loss": 46.0,
-      "step": 13383
-    },
-    {
-      "epoch": 1.0233002656880172,
-      "grad_norm": 0.0011514967773109674,
-      "learning_rate": 0.0001999994840269767,
-      "loss": 46.0,
-      "step": 13384
-    },
-    {
-      "epoch": 1.023376722671407,
-      "grad_norm": 0.0021016153041273355,
-      "learning_rate": 0.00019999948394981327,
-      "loss": 46.0,
-      "step": 13385
-    },
-    {
-      "epoch": 1.0234531796547968,
-      "grad_norm": 0.0003936931607313454,
-      "learning_rate": 0.00019999948387264407,
-      "loss": 46.0,
-      "step": 13386
-    },
-    {
-      "epoch": 1.0235296366381865,
-      "grad_norm": 0.0008404591935686767,
-      "learning_rate": 0.00019999948379546912,
-      "loss": 46.0,
-      "step": 13387
-    },
-    {
-      "epoch": 1.023606093621576,
-      "grad_norm": 0.005145175848156214,
-      "learning_rate": 0.00019999948371828837,
-      "loss": 46.0,
-      "step": 13388
-    },
-    {
-      "epoch": 1.0236825506049658,
-      "grad_norm": 0.0010951557196676731,
-      "learning_rate": 0.00019999948364110182,
-      "loss": 46.0,
-      "step": 13389
-    },
-    {
-      "epoch": 1.0237590075883556,
-      "grad_norm": 0.0028131543658673763,
-      "learning_rate": 0.00019999948356390958,
-      "loss": 46.0,
-      "step": 13390
-    },
-    {
-      "epoch": 1.0238354645717453,
-      "grad_norm": 0.0021544003393501043,
-      "learning_rate": 0.00019999948348671152,
-      "loss": 46.0,
-      "step": 13391
-    },
-    {
-      "epoch": 1.023911921555135,
-      "grad_norm": 0.0024513881653547287,
-      "learning_rate": 0.0001999994834095077,
-      "loss": 46.0,
-      "step": 13392
-    },
-    {
-      "epoch": 1.0239883785385249,
-      "grad_norm": 0.0012055312981829047,
-      "learning_rate": 0.0001999994833322981,
-      "loss": 46.0,
-      "step": 13393
-    },
-    {
-      "epoch": 1.0240648355219144,
-      "grad_norm": 0.0018854618538171053,
-      "learning_rate": 0.00019999948325508275,
-      "loss": 46.0,
-      "step": 13394
-    },
-    {
-      "epoch": 1.0241412925053042,
-      "grad_norm": 0.0010407872032374144,
-      "learning_rate": 0.00019999948317786165,
-      "loss": 46.0,
-      "step": 13395
-    },
-    {
-      "epoch": 1.024217749488694,
-      "grad_norm": 0.005921361036598682,
-      "learning_rate": 0.00019999948310063474,
-      "loss": 46.0,
-      "step": 13396
-    },
-    {
-      "epoch": 1.0242942064720837,
-      "grad_norm": 0.0005181310698390007,
-      "learning_rate": 0.00019999948302340209,
-      "loss": 46.0,
-      "step": 13397
-    },
-    {
-      "epoch": 1.0243706634554735,
-      "grad_norm": 0.000706980936229229,
-      "learning_rate": 0.00019999948294616363,
-      "loss": 46.0,
-      "step": 13398
-    },
-    {
-      "epoch": 1.024447120438863,
-      "grad_norm": 0.0009316926589235663,
-      "learning_rate": 0.00019999948286891943,
-      "loss": 46.0,
-      "step": 13399
-    },
-    {
-      "epoch": 1.0245235774222528,
-      "grad_norm": 0.001712973229587078,
-      "learning_rate": 0.00019999948279166946,
-      "loss": 46.0,
-      "step": 13400
-    },
-    {
-      "epoch": 1.0246000344056425,
-      "grad_norm": 0.0007629567408002913,
-      "learning_rate": 0.00019999948271441374,
-      "loss": 46.0,
-      "step": 13401
-    },
-    {
-      "epoch": 1.0246764913890323,
-      "grad_norm": 0.000793912447988987,
-      "learning_rate": 0.00019999948263715222,
-      "loss": 46.0,
-      "step": 13402
-    },
-    {
-      "epoch": 1.024752948372422,
-      "grad_norm": 0.0046837893314659595,
-      "learning_rate": 0.00019999948255988492,
-      "loss": 46.0,
-      "step": 13403
-    },
-    {
-      "epoch": 1.0248294053558118,
-      "grad_norm": 0.0014919465174898505,
-      "learning_rate": 0.00019999948248261185,
-      "loss": 46.0,
-      "step": 13404
-    },
-    {
-      "epoch": 1.0249058623392013,
-      "grad_norm": 0.00041987377335317433,
-      "learning_rate": 0.00019999948240533307,
-      "loss": 46.0,
-      "step": 13405
-    },
-    {
-      "epoch": 1.024982319322591,
-      "grad_norm": 0.0010832062689587474,
-      "learning_rate": 0.00019999948232804845,
-      "loss": 46.0,
-      "step": 13406
-    },
-    {
-      "epoch": 1.0250587763059809,
-      "grad_norm": 0.00044721554149873555,
-      "learning_rate": 0.0001999994822507581,
-      "loss": 46.0,
-      "step": 13407
-    },
-    {
-      "epoch": 1.0251352332893706,
-      "grad_norm": 0.0005603706231340766,
-      "learning_rate": 0.00019999948217346196,
-      "loss": 46.0,
-      "step": 13408
-    },
-    {
-      "epoch": 1.0252116902727604,
-      "grad_norm": 0.002323447959497571,
-      "learning_rate": 0.00019999948209616005,
-      "loss": 46.0,
-      "step": 13409
-    },
-    {
-      "epoch": 1.02528814725615,
-      "grad_norm": 0.000664880673866719,
-      "learning_rate": 0.0001999994820188524,
-      "loss": 46.0,
-      "step": 13410
-    },
-    {
-      "epoch": 1.0253646042395397,
-      "grad_norm": 0.002324435394257307,
-      "learning_rate": 0.00019999948194153897,
-      "loss": 46.0,
-      "step": 13411
-    },
-    {
-      "epoch": 1.0254410612229294,
-      "grad_norm": 0.0006417757831513882,
-      "learning_rate": 0.00019999948186421977,
-      "loss": 46.0,
-      "step": 13412
-    },
-    {
-      "epoch": 1.0255175182063192,
-      "grad_norm": 0.0009849769994616508,
-      "learning_rate": 0.00019999948178689477,
-      "loss": 46.0,
-      "step": 13413
-    },
-    {
-      "epoch": 1.025593975189709,
-      "grad_norm": 0.008826961740851402,
-      "learning_rate": 0.00019999948170956405,
-      "loss": 46.0,
-      "step": 13414
-    },
-    {
-      "epoch": 1.0256704321730985,
-      "grad_norm": 0.00046254077460616827,
-      "learning_rate": 0.00019999948163222753,
-      "loss": 46.0,
-      "step": 13415
-    },
-    {
-      "epoch": 1.0257468891564883,
-      "grad_norm": 0.0015090582892298698,
-      "learning_rate": 0.00019999948155488526,
-      "loss": 46.0,
-      "step": 13416
-    },
-    {
-      "epoch": 1.025823346139878,
-      "grad_norm": 0.0014890835154801607,
-      "learning_rate": 0.00019999948147753717,
-      "loss": 46.0,
-      "step": 13417
-    },
-    {
-      "epoch": 1.0258998031232678,
-      "grad_norm": 0.0030610349494963884,
-      "learning_rate": 0.00019999948140018335,
-      "loss": 46.0,
-      "step": 13418
-    },
-    {
-      "epoch": 1.0259762601066575,
-      "grad_norm": 0.0007048352854326367,
-      "learning_rate": 0.00019999948132282377,
-      "loss": 46.0,
-      "step": 13419
-    },
-    {
-      "epoch": 1.0260527170900473,
-      "grad_norm": 0.0008247590158134699,
-      "learning_rate": 0.0001999994812454584,
-      "loss": 46.0,
-      "step": 13420
-    },
-    {
-      "epoch": 1.0261291740734368,
-      "grad_norm": 0.001291550463065505,
-      "learning_rate": 0.0001999994811680873,
-      "loss": 46.0,
-      "step": 13421
-    },
-    {
-      "epoch": 1.0262056310568266,
-      "grad_norm": 0.008750289678573608,
-      "learning_rate": 0.0001999994810907104,
-      "loss": 46.0,
-      "step": 13422
-    },
-    {
-      "epoch": 1.0262820880402164,
-      "grad_norm": 0.00046774253132753074,
-      "learning_rate": 0.00019999948101332771,
-      "loss": 46.0,
-      "step": 13423
-    },
-    {
-      "epoch": 1.0263585450236061,
-      "grad_norm": 0.0005651692044921219,
-      "learning_rate": 0.0001999994809359393,
-      "loss": 46.0,
-      "step": 13424
-    },
-    {
-      "epoch": 1.026435002006996,
-      "grad_norm": 0.0030300442595034838,
-      "learning_rate": 0.0001999994808585451,
-      "loss": 46.0,
-      "step": 13425
-    },
-    {
-      "epoch": 1.0265114589903854,
-      "grad_norm": 0.0016867343802005053,
-      "learning_rate": 0.0001999994807811451,
-      "loss": 46.0,
-      "step": 13426
-    },
-    {
-      "epoch": 1.0265879159737752,
-      "grad_norm": 0.0009941098978742957,
-      "learning_rate": 0.00019999948070373934,
-      "loss": 46.0,
-      "step": 13427
-    },
-    {
-      "epoch": 1.026664372957165,
-      "grad_norm": 0.006308149080723524,
-      "learning_rate": 0.0001999994806263278,
-      "loss": 46.0,
-      "step": 13428
-    },
-    {
-      "epoch": 1.0267408299405547,
-      "grad_norm": 0.008601488545536995,
-      "learning_rate": 0.00019999948054891053,
-      "loss": 46.0,
-      "step": 13429
-    },
-    {
-      "epoch": 1.0268172869239445,
-      "grad_norm": 0.000377606600522995,
-      "learning_rate": 0.0001999994804714875,
-      "loss": 46.0,
-      "step": 13430
-    },
-    {
-      "epoch": 1.0268937439073342,
-      "grad_norm": 0.0009216060279868543,
-      "learning_rate": 0.00019999948039405865,
-      "loss": 46.0,
-      "step": 13431
-    },
-    {
-      "epoch": 1.0269702008907238,
-      "grad_norm": 0.004727725405246019,
-      "learning_rate": 0.00019999948031662406,
-      "loss": 46.0,
-      "step": 13432
-    },
-    {
-      "epoch": 1.0270466578741135,
-      "grad_norm": 0.002066485583782196,
-      "learning_rate": 0.0001999994802391837,
-      "loss": 46.0,
-      "step": 13433
-    },
-    {
-      "epoch": 1.0271231148575033,
-      "grad_norm": 0.0011169894132763147,
-      "learning_rate": 0.00019999948016173757,
-      "loss": 46.0,
-      "step": 13434
-    },
-    {
-      "epoch": 1.027199571840893,
-      "grad_norm": 0.002267979783937335,
-      "learning_rate": 0.00019999948008428566,
-      "loss": 46.0,
-      "step": 13435
-    },
-    {
-      "epoch": 1.0272760288242828,
-      "grad_norm": 0.0012321372050791979,
-      "learning_rate": 0.00019999948000682798,
-      "loss": 46.0,
-      "step": 13436
-    },
-    {
-      "epoch": 1.0273524858076724,
-      "grad_norm": 0.0006722132093273103,
-      "learning_rate": 0.00019999947992936455,
-      "loss": 46.0,
-      "step": 13437
-    },
-    {
-      "epoch": 1.0274289427910621,
-      "grad_norm": 0.0029457705095410347,
-      "learning_rate": 0.00019999947985189533,
-      "loss": 46.0,
-      "step": 13438
-    },
-    {
-      "epoch": 1.0275053997744519,
-      "grad_norm": 0.000724061974324286,
-      "learning_rate": 0.00019999947977442035,
-      "loss": 46.0,
-      "step": 13439
-    },
-    {
-      "epoch": 1.0275818567578416,
-      "grad_norm": 0.0013748782221227884,
-      "learning_rate": 0.0001999994796969396,
-      "loss": 46.0,
-      "step": 13440
-    },
-    {
-      "epoch": 1.0276583137412314,
-      "grad_norm": 0.00034714952926151454,
-      "learning_rate": 0.0001999994796194531,
-      "loss": 46.0,
-      "step": 13441
-    },
-    {
-      "epoch": 1.0277347707246212,
-      "grad_norm": 0.0020628017373383045,
-      "learning_rate": 0.0001999994795419608,
-      "loss": 46.0,
-      "step": 13442
-    },
-    {
-      "epoch": 1.0278112277080107,
-      "grad_norm": 0.00254381587728858,
-      "learning_rate": 0.0001999994794644627,
-      "loss": 46.0,
-      "step": 13443
-    },
-    {
-      "epoch": 1.0278876846914005,
-      "grad_norm": 0.0014569807099178433,
-      "learning_rate": 0.0001999994793869589,
-      "loss": 46.0,
-      "step": 13444
-    },
-    {
-      "epoch": 1.0279641416747902,
-      "grad_norm": 0.0006399656413123012,
-      "learning_rate": 0.0001999994793094493,
-      "loss": 46.0,
-      "step": 13445
-    },
-    {
-      "epoch": 1.02804059865818,
-      "grad_norm": 0.005551642272621393,
-      "learning_rate": 0.00019999947923193394,
-      "loss": 46.0,
-      "step": 13446
-    },
-    {
-      "epoch": 1.0281170556415697,
-      "grad_norm": 0.0009210671414621174,
-      "learning_rate": 0.0001999994791544128,
-      "loss": 46.0,
-      "step": 13447
-    },
-    {
-      "epoch": 1.0281935126249593,
-      "grad_norm": 0.0012172829592600465,
-      "learning_rate": 0.0001999994790768859,
-      "loss": 46.0,
-      "step": 13448
-    },
-    {
-      "epoch": 1.028269969608349,
-      "grad_norm": 0.0009926899801939726,
-      "learning_rate": 0.0001999994789993532,
-      "loss": 46.0,
-      "step": 13449
-    },
-    {
-      "epoch": 1.0283464265917388,
-      "grad_norm": 0.008660213090479374,
-      "learning_rate": 0.0001999994789218148,
-      "loss": 46.0,
-      "step": 13450
-    },
-    {
-      "epoch": 1.0284228835751286,
-      "grad_norm": 0.0016274573281407356,
-      "learning_rate": 0.00019999947884427054,
-      "loss": 46.0,
-      "step": 13451
-    },
-    {
-      "epoch": 1.0284993405585183,
-      "grad_norm": 0.0007088819984346628,
-      "learning_rate": 0.00019999947876672057,
-      "loss": 46.0,
-      "step": 13452
-    },
-    {
-      "epoch": 1.028575797541908,
-      "grad_norm": 0.0013249666662886739,
-      "learning_rate": 0.00019999947868916483,
-      "loss": 46.0,
-      "step": 13453
-    },
-    {
-      "epoch": 1.0286522545252976,
-      "grad_norm": 0.0014633532846346498,
-      "learning_rate": 0.0001999994786116033,
-      "loss": 46.0,
-      "step": 13454
-    },
-    {
-      "epoch": 1.0287287115086874,
-      "grad_norm": 0.0028745331801474094,
-      "learning_rate": 0.000199999478534036,
-      "loss": 46.0,
-      "step": 13455
-    },
-    {
-      "epoch": 1.0288051684920771,
-      "grad_norm": 0.0043944101780653,
-      "learning_rate": 0.00019999947845646295,
-      "loss": 46.0,
-      "step": 13456
-    },
-    {
-      "epoch": 1.028881625475467,
-      "grad_norm": 0.0012361662229523063,
-      "learning_rate": 0.0001999994783788841,
-      "loss": 46.0,
-      "step": 13457
-    },
-    {
-      "epoch": 1.0289580824588567,
-      "grad_norm": 0.016332441940903664,
-      "learning_rate": 0.00019999947830129952,
-      "loss": 46.0,
-      "step": 13458
-    },
-    {
-      "epoch": 1.0290345394422462,
-      "grad_norm": 0.003959211986511946,
-      "learning_rate": 0.00019999947822370913,
-      "loss": 46.0,
-      "step": 13459
-    },
-    {
-      "epoch": 1.029110996425636,
-      "grad_norm": 0.003940400201827288,
-      "learning_rate": 0.00019999947814611303,
-      "loss": 46.0,
-      "step": 13460
-    },
-    {
-      "epoch": 1.0291874534090257,
-      "grad_norm": 0.0007510966970585287,
-      "learning_rate": 0.0001999994780685111,
-      "loss": 46.0,
-      "step": 13461
-    },
-    {
-      "epoch": 1.0292639103924155,
-      "grad_norm": 0.001795685151591897,
-      "learning_rate": 0.00019999947799090344,
-      "loss": 46.0,
-      "step": 13462
-    },
-    {
-      "epoch": 1.0293403673758053,
-      "grad_norm": 0.0020094038918614388,
-      "learning_rate": 0.00019999947791328996,
-      "loss": 46.0,
-      "step": 13463
-    },
-    {
-      "epoch": 1.029416824359195,
-      "grad_norm": 0.0009090380626730621,
-      "learning_rate": 0.00019999947783567076,
-      "loss": 46.0,
-      "step": 13464
-    },
-    {
-      "epoch": 1.0294932813425846,
-      "grad_norm": 0.0010834747226908803,
-      "learning_rate": 0.00019999947775804577,
-      "loss": 46.0,
-      "step": 13465
-    },
-    {
-      "epoch": 1.0295697383259743,
-      "grad_norm": 0.00221562129445374,
-      "learning_rate": 0.00019999947768041502,
-      "loss": 46.0,
-      "step": 13466
-    },
-    {
-      "epoch": 1.029646195309364,
-      "grad_norm": 0.0038680091965943575,
-      "learning_rate": 0.00019999947760277847,
-      "loss": 46.0,
-      "step": 13467
-    },
-    {
-      "epoch": 1.0297226522927538,
-      "grad_norm": 0.0021474480163306,
-      "learning_rate": 0.00019999947752513618,
-      "loss": 46.0,
-      "step": 13468
-    },
-    {
-      "epoch": 1.0297991092761436,
-      "grad_norm": 0.001359705813229084,
-      "learning_rate": 0.00019999947744748811,
-      "loss": 46.0,
-      "step": 13469
-    },
-    {
-      "epoch": 1.0298755662595331,
-      "grad_norm": 0.002832723781466484,
-      "learning_rate": 0.0001999994773698343,
-      "loss": 46.0,
-      "step": 13470
-    },
-    {
-      "epoch": 1.029952023242923,
-      "grad_norm": 0.001258998061530292,
-      "learning_rate": 0.00019999947729217472,
-      "loss": 46.0,
-      "step": 13471
-    },
-    {
-      "epoch": 1.0300284802263127,
-      "grad_norm": 0.0026318260934203863,
-      "learning_rate": 0.00019999947721450933,
-      "loss": 46.0,
-      "step": 13472
-    },
-    {
-      "epoch": 1.0301049372097024,
-      "grad_norm": 0.00022138512576930225,
-      "learning_rate": 0.00019999947713683817,
-      "loss": 46.0,
-      "step": 13473
-    },
-    {
-      "epoch": 1.0301813941930922,
-      "grad_norm": 0.0032568189781159163,
-      "learning_rate": 0.0001999994770591613,
-      "loss": 46.0,
-      "step": 13474
-    },
-    {
-      "epoch": 1.030257851176482,
-      "grad_norm": 0.0010018192697316408,
-      "learning_rate": 0.0001999994769814786,
-      "loss": 46.0,
-      "step": 13475
-    },
-    {
-      "epoch": 1.0303343081598715,
-      "grad_norm": 0.00046532382839359343,
-      "learning_rate": 0.00019999947690379016,
-      "loss": 46.0,
-      "step": 13476
-    },
-    {
-      "epoch": 1.0304107651432612,
-      "grad_norm": 0.000701883458532393,
-      "learning_rate": 0.00019999947682609594,
-      "loss": 46.0,
-      "step": 13477
-    },
-    {
-      "epoch": 1.030487222126651,
-      "grad_norm": 0.006138428580015898,
-      "learning_rate": 0.00019999947674839594,
-      "loss": 46.0,
-      "step": 13478
-    },
-    {
-      "epoch": 1.0305636791100408,
-      "grad_norm": 0.002016335027292371,
-      "learning_rate": 0.0001999994766706902,
-      "loss": 46.0,
-      "step": 13479
-    },
-    {
-      "epoch": 1.0306401360934305,
-      "grad_norm": 0.0007726931944489479,
-      "learning_rate": 0.00019999947659297864,
-      "loss": 46.0,
-      "step": 13480
-    },
-    {
-      "epoch": 1.03071659307682,
-      "grad_norm": 0.002231534570455551,
-      "learning_rate": 0.00019999947651526135,
-      "loss": 46.0,
-      "step": 13481
-    },
-    {
-      "epoch": 1.0307930500602098,
-      "grad_norm": 0.0018946636701002717,
-      "learning_rate": 0.0001999994764375383,
-      "loss": 46.0,
-      "step": 13482
-    },
-    {
-      "epoch": 1.0308695070435996,
-      "grad_norm": 0.00237390142865479,
-      "learning_rate": 0.00019999947635980947,
-      "loss": 46.0,
-      "step": 13483
-    },
-    {
-      "epoch": 1.0309459640269893,
-      "grad_norm": 0.0006835960666649044,
-      "learning_rate": 0.00019999947628207486,
-      "loss": 46.0,
-      "step": 13484
-    },
-    {
-      "epoch": 1.031022421010379,
-      "grad_norm": 0.0016942026559263468,
-      "learning_rate": 0.00019999947620433448,
-      "loss": 46.0,
-      "step": 13485
-    },
-    {
-      "epoch": 1.0310988779937686,
-      "grad_norm": 0.002672442002221942,
-      "learning_rate": 0.00019999947612658834,
-      "loss": 46.0,
-      "step": 13486
-    },
-    {
-      "epoch": 1.0311753349771584,
-      "grad_norm": 0.0019218804081901908,
-      "learning_rate": 0.0001999994760488364,
-      "loss": 46.0,
-      "step": 13487
-    },
-    {
-      "epoch": 1.0312517919605482,
-      "grad_norm": 0.0005724010989069939,
-      "learning_rate": 0.00019999947597107873,
-      "loss": 46.0,
-      "step": 13488
-    },
-    {
-      "epoch": 1.031328248943938,
-      "grad_norm": 0.01681784726679325,
-      "learning_rate": 0.00019999947589331528,
-      "loss": 46.0,
-      "step": 13489
-    },
-    {
-      "epoch": 1.0314047059273277,
-      "grad_norm": 0.0030642254278063774,
-      "learning_rate": 0.00019999947581554608,
-      "loss": 46.0,
-      "step": 13490
-    },
-    {
-      "epoch": 1.0314811629107175,
-      "grad_norm": 0.0012512184912338853,
-      "learning_rate": 0.00019999947573777106,
-      "loss": 46.0,
-      "step": 13491
-    },
-    {
-      "epoch": 1.031557619894107,
-      "grad_norm": 0.001450938987545669,
-      "learning_rate": 0.0001999994756599903,
-      "loss": 46.0,
-      "step": 13492
-    },
-    {
-      "epoch": 1.0316340768774968,
-      "grad_norm": 0.001800187979824841,
-      "learning_rate": 0.0001999994755822038,
-      "loss": 46.0,
-      "step": 13493
-    },
-    {
-      "epoch": 1.0317105338608865,
-      "grad_norm": 0.0008096079109236598,
-      "learning_rate": 0.0001999994755044115,
-      "loss": 46.0,
-      "step": 13494
-    },
-    {
-      "epoch": 1.0317869908442763,
-      "grad_norm": 0.006515221670269966,
-      "learning_rate": 0.0001999994754266134,
-      "loss": 46.0,
-      "step": 13495
-    },
-    {
-      "epoch": 1.031863447827666,
-      "grad_norm": 0.0010154324118047953,
-      "learning_rate": 0.0001999994753488096,
-      "loss": 46.0,
-      "step": 13496
-    },
-    {
-      "epoch": 1.0319399048110556,
-      "grad_norm": 0.010628794319927692,
-      "learning_rate": 0.00019999947527099999,
-      "loss": 46.0,
-      "step": 13497
-    },
-    {
-      "epoch": 1.0320163617944453,
-      "grad_norm": 0.005647844634950161,
-      "learning_rate": 0.00019999947519318457,
-      "loss": 46.0,
-      "step": 13498
-    },
-    {
-      "epoch": 1.032092818777835,
-      "grad_norm": 0.001832620589993894,
-      "learning_rate": 0.00019999947511536347,
-      "loss": 46.0,
-      "step": 13499
-    },
-    {
-      "epoch": 1.0321692757612249,
-      "grad_norm": 0.0016396817518398166,
-      "learning_rate": 0.00019999947503753654,
-      "loss": 46.0,
-      "step": 13500
-    },
-    {
-      "epoch": 1.0322457327446146,
-      "grad_norm": 0.00513737415894866,
-      "learning_rate": 0.00019999947495970383,
-      "loss": 46.0,
-      "step": 13501
-    },
-    {
-      "epoch": 1.0323221897280044,
-      "grad_norm": 0.0011204457841813564,
-      "learning_rate": 0.00019999947488186538,
-      "loss": 46.0,
-      "step": 13502
-    },
-    {
-      "epoch": 1.032398646711394,
-      "grad_norm": 0.0009296801290474832,
-      "learning_rate": 0.00019999947480402118,
-      "loss": 46.0,
-      "step": 13503
-    },
-    {
-      "epoch": 1.0324751036947837,
-      "grad_norm": 0.0006482386379502714,
-      "learning_rate": 0.00019999947472617118,
-      "loss": 46.0,
-      "step": 13504
-    },
-    {
-      "epoch": 1.0325515606781734,
-      "grad_norm": 0.004908979404717684,
-      "learning_rate": 0.0001999994746483154,
-      "loss": 46.0,
-      "step": 13505
-    },
-    {
-      "epoch": 1.0326280176615632,
-      "grad_norm": 0.0009123889612965286,
-      "learning_rate": 0.00019999947457045386,
-      "loss": 46.0,
-      "step": 13506
-    },
-    {
-      "epoch": 1.032704474644953,
-      "grad_norm": 0.0045522041618824005,
-      "learning_rate": 0.00019999947449258657,
-      "loss": 46.0,
-      "step": 13507
-    },
-    {
-      "epoch": 1.0327809316283425,
-      "grad_norm": 0.0005408361903391778,
-      "learning_rate": 0.0001999994744147135,
-      "loss": 46.0,
-      "step": 13508
-    },
-    {
-      "epoch": 1.0328573886117323,
-      "grad_norm": 0.000514801824465394,
-      "learning_rate": 0.00019999947433683467,
-      "loss": 46.0,
-      "step": 13509
-    },
-    {
-      "epoch": 1.032933845595122,
-      "grad_norm": 0.0005614846013486385,
-      "learning_rate": 0.00019999947425895006,
-      "loss": 46.0,
-      "step": 13510
-    },
-    {
-      "epoch": 1.0330103025785118,
-      "grad_norm": 0.0024432719219475985,
-      "learning_rate": 0.00019999947418105967,
-      "loss": 46.0,
-      "step": 13511
-    },
-    {
-      "epoch": 1.0330867595619015,
-      "grad_norm": 0.0012248132843524218,
-      "learning_rate": 0.00019999947410316354,
-      "loss": 46.0,
-      "step": 13512
-    },
-    {
-      "epoch": 1.0331632165452913,
-      "grad_norm": 0.0020208240021020174,
-      "learning_rate": 0.00019999947402526164,
-      "loss": 46.0,
-      "step": 13513
-    },
-    {
-      "epoch": 1.0332396735286808,
-      "grad_norm": 0.0003774506039917469,
-      "learning_rate": 0.00019999947394735393,
-      "loss": 46.0,
-      "step": 13514
-    },
-    {
-      "epoch": 1.0333161305120706,
-      "grad_norm": 0.0009529279777780175,
-      "learning_rate": 0.00019999947386944048,
-      "loss": 46.0,
-      "step": 13515
-    },
-    {
-      "epoch": 1.0333925874954604,
-      "grad_norm": 0.001439249492250383,
-      "learning_rate": 0.00019999947379152123,
-      "loss": 46.0,
-      "step": 13516
-    },
-    {
-      "epoch": 1.0334690444788501,
-      "grad_norm": 0.0006790959741920233,
-      "learning_rate": 0.00019999947371359623,
-      "loss": 46.0,
-      "step": 13517
-    },
-    {
-      "epoch": 1.03354550146224,
-      "grad_norm": 0.0010327092604711652,
-      "learning_rate": 0.00019999947363566549,
-      "loss": 46.0,
-      "step": 13518
-    },
-    {
-      "epoch": 1.0336219584456294,
-      "grad_norm": 0.0011972899083048105,
-      "learning_rate": 0.00019999947355772894,
-      "loss": 46.0,
-      "step": 13519
-    },
-    {
-      "epoch": 1.0336984154290192,
-      "grad_norm": 0.004191542509943247,
-      "learning_rate": 0.00019999947347978662,
-      "loss": 46.0,
-      "step": 13520
-    },
-    {
-      "epoch": 1.033774872412409,
-      "grad_norm": 0.0007471860735677183,
-      "learning_rate": 0.00019999947340183858,
-      "loss": 46.0,
-      "step": 13521
-    },
-    {
-      "epoch": 1.0338513293957987,
-      "grad_norm": 0.0002676327130757272,
-      "learning_rate": 0.00019999947332388472,
-      "loss": 46.0,
-      "step": 13522
-    },
-    {
-      "epoch": 1.0339277863791885,
-      "grad_norm": 0.0010614984203130007,
-      "learning_rate": 0.0001999994732459251,
-      "loss": 46.0,
-      "step": 13523
-    },
-    {
-      "epoch": 1.0340042433625782,
-      "grad_norm": 0.0027158628217875957,
-      "learning_rate": 0.00019999947316795972,
-      "loss": 46.0,
-      "step": 13524
-    },
-    {
-      "epoch": 1.0340807003459678,
-      "grad_norm": 0.0010642081033438444,
-      "learning_rate": 0.00019999947308998856,
-      "loss": 46.0,
-      "step": 13525
-    },
-    {
-      "epoch": 1.0341571573293575,
-      "grad_norm": 0.0011307941749691963,
-      "learning_rate": 0.00019999947301201166,
-      "loss": 46.0,
-      "step": 13526
-    },
-    {
-      "epoch": 1.0342336143127473,
-      "grad_norm": 0.0034480879548937082,
-      "learning_rate": 0.00019999947293402896,
-      "loss": 46.0,
-      "step": 13527
-    },
-    {
-      "epoch": 1.034310071296137,
-      "grad_norm": 0.005681936163455248,
-      "learning_rate": 0.0001999994728560405,
-      "loss": 46.0,
-      "step": 13528
-    },
-    {
-      "epoch": 1.0343865282795268,
-      "grad_norm": 0.001056214328855276,
-      "learning_rate": 0.00019999947277804628,
-      "loss": 46.0,
-      "step": 13529
-    },
-    {
-      "epoch": 1.0344629852629164,
-      "grad_norm": 0.006874502636492252,
-      "learning_rate": 0.00019999947270004625,
-      "loss": 46.0,
-      "step": 13530
-    },
-    {
-      "epoch": 1.0345394422463061,
-      "grad_norm": 0.0016898735193535686,
-      "learning_rate": 0.0001999994726220405,
-      "loss": 46.0,
-      "step": 13531
-    },
-    {
-      "epoch": 1.0346158992296959,
-      "grad_norm": 0.0019312641816213727,
-      "learning_rate": 0.00019999947254402897,
-      "loss": 46.0,
-      "step": 13532
-    },
-    {
-      "epoch": 1.0346923562130856,
-      "grad_norm": 0.000637902005109936,
-      "learning_rate": 0.00019999947246601165,
-      "loss": 46.0,
-      "step": 13533
-    },
-    {
-      "epoch": 1.0347688131964754,
-      "grad_norm": 0.00584821030497551,
-      "learning_rate": 0.00019999947238798858,
-      "loss": 46.0,
-      "step": 13534
-    },
-    {
-      "epoch": 1.034845270179865,
-      "grad_norm": 0.0010185374412685633,
-      "learning_rate": 0.00019999947230995975,
-      "loss": 46.0,
-      "step": 13535
-    },
-    {
-      "epoch": 1.0349217271632547,
-      "grad_norm": 0.0011475110659375787,
-      "learning_rate": 0.00019999947223192513,
-      "loss": 46.0,
-      "step": 13536
-    },
-    {
-      "epoch": 1.0349981841466445,
-      "grad_norm": 0.0015137637965381145,
-      "learning_rate": 0.00019999947215388475,
-      "loss": 46.0,
-      "step": 13537
-    },
-    {
-      "epoch": 1.0350746411300342,
-      "grad_norm": 0.002390458947047591,
-      "learning_rate": 0.0001999994720758386,
-      "loss": 46.0,
-      "step": 13538
-    },
-    {
-      "epoch": 1.035151098113424,
-      "grad_norm": 0.0007472530705854297,
-      "learning_rate": 0.0001999994719977867,
-      "loss": 46.0,
-      "step": 13539
-    },
-    {
-      "epoch": 1.0352275550968137,
-      "grad_norm": 0.0023285613860934973,
-      "learning_rate": 0.00019999947191972898,
-      "loss": 46.0,
-      "step": 13540
-    },
-    {
-      "epoch": 1.0353040120802033,
-      "grad_norm": 0.0013641807017847896,
-      "learning_rate": 0.00019999947184166553,
-      "loss": 46.0,
-      "step": 13541
-    },
-    {
-      "epoch": 1.035380469063593,
-      "grad_norm": 0.0007180684478953481,
-      "learning_rate": 0.00019999947176359628,
-      "loss": 46.0,
-      "step": 13542
-    },
-    {
-      "epoch": 1.0354569260469828,
-      "grad_norm": 0.0014429963193833828,
-      "learning_rate": 0.00019999947168552126,
-      "loss": 46.0,
-      "step": 13543
-    },
-    {
-      "epoch": 1.0355333830303726,
-      "grad_norm": 0.0010568283032625914,
-      "learning_rate": 0.0001999994716074405,
-      "loss": 46.0,
-      "step": 13544
-    },
-    {
-      "epoch": 1.0356098400137623,
-      "grad_norm": 0.0009685680270195007,
-      "learning_rate": 0.00019999947152935397,
-      "loss": 46.0,
-      "step": 13545
-    },
-    {
-      "epoch": 1.0356862969971519,
-      "grad_norm": 0.0007094335742294788,
-      "learning_rate": 0.00019999947145126165,
-      "loss": 46.0,
-      "step": 13546
-    },
-    {
-      "epoch": 1.0357627539805416,
-      "grad_norm": 0.0008215317502617836,
-      "learning_rate": 0.00019999947137316356,
-      "loss": 46.0,
-      "step": 13547
-    },
-    {
-      "epoch": 1.0358392109639314,
-      "grad_norm": 0.0006712996400892735,
-      "learning_rate": 0.00019999947129505975,
-      "loss": 46.0,
-      "step": 13548
-    },
-    {
-      "epoch": 1.0359156679473212,
-      "grad_norm": 0.0013593286275863647,
-      "learning_rate": 0.00019999947121695014,
-      "loss": 46.0,
-      "step": 13549
-    },
-    {
-      "epoch": 1.035992124930711,
-      "grad_norm": 0.001702674780972302,
-      "learning_rate": 0.00019999947113883473,
-      "loss": 46.0,
-      "step": 13550
-    },
-    {
-      "epoch": 1.0360685819141007,
-      "grad_norm": 0.008379071019589901,
-      "learning_rate": 0.00019999947106071357,
-      "loss": 46.0,
-      "step": 13551
-    },
-    {
-      "epoch": 1.0361450388974902,
-      "grad_norm": 0.006596278864890337,
-      "learning_rate": 0.00019999947098258664,
-      "loss": 46.0,
-      "step": 13552
-    },
-    {
-      "epoch": 1.03622149588088,
-      "grad_norm": 0.0026221596635878086,
-      "learning_rate": 0.00019999947090445396,
-      "loss": 46.0,
-      "step": 13553
-    },
-    {
-      "epoch": 1.0362979528642697,
-      "grad_norm": 0.0005150894867256284,
-      "learning_rate": 0.00019999947082631548,
-      "loss": 46.0,
-      "step": 13554
-    },
-    {
-      "epoch": 1.0363744098476595,
-      "grad_norm": 0.0010322367306798697,
-      "learning_rate": 0.00019999947074817126,
-      "loss": 46.0,
-      "step": 13555
-    },
-    {
-      "epoch": 1.0364508668310493,
-      "grad_norm": 0.001431052340194583,
-      "learning_rate": 0.00019999947067002126,
-      "loss": 46.0,
-      "step": 13556
-    },
-    {
-      "epoch": 1.0365273238144388,
-      "grad_norm": 0.002187419915571809,
-      "learning_rate": 0.0001999994705918655,
-      "loss": 46.0,
-      "step": 13557
-    },
-    {
-      "epoch": 1.0366037807978286,
-      "grad_norm": 0.00047102628741413355,
-      "learning_rate": 0.00019999947051370395,
-      "loss": 46.0,
-      "step": 13558
-    },
-    {
-      "epoch": 1.0366802377812183,
-      "grad_norm": 0.002510851714760065,
-      "learning_rate": 0.00019999947043553663,
-      "loss": 46.0,
-      "step": 13559
-    },
-    {
-      "epoch": 1.036756694764608,
-      "grad_norm": 0.0009445094037801027,
-      "learning_rate": 0.00019999947035736354,
-      "loss": 46.0,
-      "step": 13560
-    },
-    {
-      "epoch": 1.0368331517479978,
-      "grad_norm": 0.0013167369179427624,
-      "learning_rate": 0.00019999947027918467,
-      "loss": 46.0,
-      "step": 13561
-    },
-    {
-      "epoch": 1.0369096087313876,
-      "grad_norm": 0.0020650813821703196,
-      "learning_rate": 0.00019999947020100006,
-      "loss": 46.0,
-      "step": 13562
-    },
-    {
-      "epoch": 1.0369860657147771,
-      "grad_norm": 0.0011621693847700953,
-      "learning_rate": 0.00019999947012280968,
-      "loss": 46.0,
-      "step": 13563
-    },
-    {
-      "epoch": 1.037062522698167,
-      "grad_norm": 0.0004166075377725065,
-      "learning_rate": 0.00019999947004461352,
-      "loss": 46.0,
-      "step": 13564
-    },
-    {
-      "epoch": 1.0371389796815567,
-      "grad_norm": 0.0010638912208378315,
-      "learning_rate": 0.0001999994699664116,
-      "loss": 46.0,
-      "step": 13565
-    },
-    {
-      "epoch": 1.0372154366649464,
-      "grad_norm": 0.0005540067795664072,
-      "learning_rate": 0.00019999946988820392,
-      "loss": 46.0,
-      "step": 13566
-    },
-    {
-      "epoch": 1.0372918936483362,
-      "grad_norm": 0.0022389935329556465,
-      "learning_rate": 0.00019999946980999044,
-      "loss": 46.0,
-      "step": 13567
-    },
-    {
-      "epoch": 1.0373683506317257,
-      "grad_norm": 0.00045577631681226194,
-      "learning_rate": 0.0001999994697317712,
-      "loss": 46.0,
-      "step": 13568
-    },
-    {
-      "epoch": 1.0374448076151155,
-      "grad_norm": 0.0014733450952917337,
-      "learning_rate": 0.0001999994696535462,
-      "loss": 46.0,
-      "step": 13569
-    },
-    {
-      "epoch": 1.0375212645985052,
-      "grad_norm": 0.0006990301189944148,
-      "learning_rate": 0.00019999946957531542,
-      "loss": 46.0,
-      "step": 13570
-    },
-    {
-      "epoch": 1.037597721581895,
-      "grad_norm": 0.0008845215197652578,
-      "learning_rate": 0.00019999946949707885,
-      "loss": 46.0,
-      "step": 13571
-    },
-    {
-      "epoch": 1.0376741785652848,
-      "grad_norm": 0.006686788517981768,
-      "learning_rate": 0.00019999946941883656,
-      "loss": 46.0,
-      "step": 13572
-    },
-    {
-      "epoch": 1.0377506355486745,
-      "grad_norm": 0.00149017374496907,
-      "learning_rate": 0.00019999946934058847,
-      "loss": 46.0,
-      "step": 13573
-    },
-    {
-      "epoch": 1.037827092532064,
-      "grad_norm": 0.0012148062232881784,
-      "learning_rate": 0.0001999994692623346,
-      "loss": 46.0,
-      "step": 13574
-    },
-    {
-      "epoch": 1.0379035495154538,
-      "grad_norm": 0.0005045654252171516,
-      "learning_rate": 0.000199999469184075,
-      "loss": 46.0,
-      "step": 13575
-    },
-    {
-      "epoch": 1.0379800064988436,
-      "grad_norm": 0.001179555431008339,
-      "learning_rate": 0.0001999994691058096,
-      "loss": 46.0,
-      "step": 13576
-    },
-    {
-      "epoch": 1.0380564634822333,
-      "grad_norm": 0.00018153031123802066,
-      "learning_rate": 0.00019999946902753845,
-      "loss": 46.0,
-      "step": 13577
-    },
-    {
-      "epoch": 1.038132920465623,
-      "grad_norm": 0.0008283884380944073,
-      "learning_rate": 0.00019999946894926152,
-      "loss": 46.0,
-      "step": 13578
-    },
-    {
-      "epoch": 1.0382093774490126,
-      "grad_norm": 0.0009526748908683658,
-      "learning_rate": 0.0001999994688709788,
-      "loss": 46.0,
-      "step": 13579
-    },
-    {
-      "epoch": 1.0382858344324024,
-      "grad_norm": 0.0004294952377676964,
-      "learning_rate": 0.00019999946879269034,
-      "loss": 46.0,
-      "step": 13580
-    },
-    {
-      "epoch": 1.0383622914157922,
-      "grad_norm": 0.0006923009641468525,
-      "learning_rate": 0.0001999994687143961,
-      "loss": 46.0,
-      "step": 13581
-    },
-    {
-      "epoch": 1.038438748399182,
-      "grad_norm": 0.0028068027459084988,
-      "learning_rate": 0.0001999994686360961,
-      "loss": 46.0,
-      "step": 13582
-    },
-    {
-      "epoch": 1.0385152053825717,
-      "grad_norm": 0.0006975121796131134,
-      "learning_rate": 0.0001999994685577903,
-      "loss": 46.0,
-      "step": 13583
-    },
-    {
-      "epoch": 1.0385916623659615,
-      "grad_norm": 0.001403622212819755,
-      "learning_rate": 0.00019999946847947876,
-      "loss": 46.0,
-      "step": 13584
-    },
-    {
-      "epoch": 1.038668119349351,
-      "grad_norm": 0.0008122682920657098,
-      "learning_rate": 0.00019999946840116147,
-      "loss": 46.0,
-      "step": 13585
-    },
-    {
-      "epoch": 1.0387445763327408,
-      "grad_norm": 0.000543102971278131,
-      "learning_rate": 0.00019999946832283838,
-      "loss": 46.0,
-      "step": 13586
-    },
-    {
-      "epoch": 1.0388210333161305,
-      "grad_norm": 0.011759413406252861,
-      "learning_rate": 0.0001999994682445095,
-      "loss": 46.0,
-      "step": 13587
-    },
-    {
-      "epoch": 1.0388974902995203,
-      "grad_norm": 0.0027144672349095345,
-      "learning_rate": 0.0001999994681661749,
-      "loss": 46.0,
-      "step": 13588
-    },
-    {
-      "epoch": 1.03897394728291,
-      "grad_norm": 0.006971284281462431,
-      "learning_rate": 0.0001999994680878345,
-      "loss": 46.0,
-      "step": 13589
-    },
-    {
-      "epoch": 1.0390504042662996,
-      "grad_norm": 0.004064250737428665,
-      "learning_rate": 0.00019999946800948834,
-      "loss": 46.0,
-      "step": 13590
-    },
-    {
-      "epoch": 1.0391268612496893,
-      "grad_norm": 0.001461939886212349,
-      "learning_rate": 0.00019999946793113638,
-      "loss": 46.0,
-      "step": 13591
-    },
-    {
-      "epoch": 1.039203318233079,
-      "grad_norm": 0.0006457848357968032,
-      "learning_rate": 0.0001999994678527787,
-      "loss": 46.0,
-      "step": 13592
-    },
-    {
-      "epoch": 1.0392797752164689,
-      "grad_norm": 0.0010811688844114542,
-      "learning_rate": 0.0001999994677744152,
-      "loss": 46.0,
-      "step": 13593
-    },
-    {
-      "epoch": 1.0393562321998586,
-      "grad_norm": 0.006942572072148323,
-      "learning_rate": 0.00019999946769604597,
-      "loss": 46.0,
-      "step": 13594
-    },
-    {
-      "epoch": 1.0394326891832484,
-      "grad_norm": 0.000904568936675787,
-      "learning_rate": 0.00019999946761767098,
-      "loss": 46.0,
-      "step": 13595
-    },
-    {
-      "epoch": 1.039509146166638,
-      "grad_norm": 0.0005813430761918426,
-      "learning_rate": 0.00019999946753929018,
-      "loss": 46.0,
-      "step": 13596
-    },
-    {
-      "epoch": 1.0395856031500277,
-      "grad_norm": 0.001077859429642558,
-      "learning_rate": 0.00019999946746090364,
-      "loss": 46.0,
-      "step": 13597
-    },
-    {
-      "epoch": 1.0396620601334174,
-      "grad_norm": 0.0039033321663737297,
-      "learning_rate": 0.00019999946738251132,
-      "loss": 46.0,
-      "step": 13598
-    },
-    {
-      "epoch": 1.0397385171168072,
-      "grad_norm": 0.001438381033949554,
-      "learning_rate": 0.00019999946730411323,
-      "loss": 46.0,
-      "step": 13599
-    },
-    {
-      "epoch": 1.039814974100197,
-      "grad_norm": 0.011141834780573845,
-      "learning_rate": 0.00019999946722570937,
-      "loss": 46.0,
-      "step": 13600
-    },
-    {
-      "epoch": 1.0398914310835865,
-      "grad_norm": 0.021015118807554245,
-      "learning_rate": 0.00019999946714729973,
-      "loss": 46.0,
-      "step": 13601
-    },
-    {
-      "epoch": 1.0399678880669763,
-      "grad_norm": 0.00244560232385993,
-      "learning_rate": 0.00019999946706888432,
-      "loss": 46.0,
-      "step": 13602
-    },
-    {
-      "epoch": 1.040044345050366,
-      "grad_norm": 0.0006874041282571852,
-      "learning_rate": 0.0001999994669904632,
-      "loss": 46.0,
-      "step": 13603
-    },
-    {
-      "epoch": 1.0401208020337558,
-      "grad_norm": 0.0007708551711402833,
-      "learning_rate": 0.00019999946691203624,
-      "loss": 46.0,
-      "step": 13604
-    },
-    {
-      "epoch": 1.0401972590171455,
-      "grad_norm": 0.0008431808673776686,
-      "learning_rate": 0.00019999946683360353,
-      "loss": 46.0,
-      "step": 13605
-    },
-    {
-      "epoch": 1.0402737160005353,
-      "grad_norm": 0.0008816320332698524,
-      "learning_rate": 0.00019999946675516506,
-      "loss": 46.0,
-      "step": 13606
-    },
-    {
-      "epoch": 1.0403501729839248,
-      "grad_norm": 0.0035544056445360184,
-      "learning_rate": 0.00019999946667672083,
-      "loss": 46.0,
-      "step": 13607
-    },
-    {
-      "epoch": 1.0404266299673146,
-      "grad_norm": 0.0009899329161271453,
-      "learning_rate": 0.0001999994665982708,
-      "loss": 46.0,
-      "step": 13608
-    },
-    {
-      "epoch": 1.0405030869507044,
-      "grad_norm": 0.0028448542580008507,
-      "learning_rate": 0.00019999946651981501,
-      "loss": 46.0,
-      "step": 13609
-    },
-    {
-      "epoch": 1.0405795439340941,
-      "grad_norm": 0.0007092030718922615,
-      "learning_rate": 0.00019999946644135344,
-      "loss": 46.0,
-      "step": 13610
-    },
-    {
-      "epoch": 1.040656000917484,
-      "grad_norm": 0.002470811130478978,
-      "learning_rate": 0.00019999946636288613,
-      "loss": 46.0,
-      "step": 13611
-    },
-    {
-      "epoch": 1.0407324579008734,
-      "grad_norm": 0.003848015097901225,
-      "learning_rate": 0.00019999946628441304,
-      "loss": 46.0,
-      "step": 13612
-    },
-    {
-      "epoch": 1.0408089148842632,
-      "grad_norm": 0.0022235563956201077,
-      "learning_rate": 0.0001999994662059342,
-      "loss": 46.0,
-      "step": 13613
-    },
-    {
-      "epoch": 1.040885371867653,
-      "grad_norm": 0.0016022929921746254,
-      "learning_rate": 0.00019999946612744957,
-      "loss": 46.0,
-      "step": 13614
-    },
-    {
-      "epoch": 1.0409618288510427,
-      "grad_norm": 0.0015884211752563715,
-      "learning_rate": 0.00019999946604895916,
-      "loss": 46.0,
-      "step": 13615
-    },
-    {
-      "epoch": 1.0410382858344325,
-      "grad_norm": 0.0031271979678422213,
-      "learning_rate": 0.000199999465970463,
-      "loss": 46.0,
-      "step": 13616
-    },
-    {
-      "epoch": 1.0411147428178222,
-      "grad_norm": 0.004668507259339094,
-      "learning_rate": 0.00019999946589196104,
-      "loss": 46.0,
-      "step": 13617
-    },
-    {
-      "epoch": 1.0411911998012118,
-      "grad_norm": 0.0006736156647093594,
-      "learning_rate": 0.00019999946581345334,
-      "loss": 46.0,
-      "step": 13618
-    },
-    {
-      "epoch": 1.0412676567846015,
-      "grad_norm": 0.0006623711669817567,
-      "learning_rate": 0.00019999946573493987,
-      "loss": 46.0,
-      "step": 13619
-    },
-    {
-      "epoch": 1.0413441137679913,
-      "grad_norm": 0.0010033831931650639,
-      "learning_rate": 0.00019999946565642062,
-      "loss": 46.0,
-      "step": 13620
-    },
-    {
-      "epoch": 1.041420570751381,
-      "grad_norm": 0.00031629044678993523,
-      "learning_rate": 0.00019999946557789562,
-      "loss": 46.0,
-      "step": 13621
-    },
-    {
-      "epoch": 1.0414970277347708,
-      "grad_norm": 0.0015300210798159242,
-      "learning_rate": 0.00019999946549936482,
-      "loss": 46.0,
-      "step": 13622
-    },
-    {
-      "epoch": 1.0415734847181604,
-      "grad_norm": 0.002132994355633855,
-      "learning_rate": 0.00019999946542082828,
-      "loss": 46.0,
-      "step": 13623
-    },
-    {
-      "epoch": 1.0416499417015501,
-      "grad_norm": 0.002365228720009327,
-      "learning_rate": 0.00019999946534228594,
-      "loss": 46.0,
-      "step": 13624
-    },
-    {
-      "epoch": 1.0417263986849399,
-      "grad_norm": 0.004612469580024481,
-      "learning_rate": 0.00019999946526373785,
-      "loss": 46.0,
-      "step": 13625
-    },
-    {
-      "epoch": 1.0418028556683296,
-      "grad_norm": 0.00740779796615243,
-      "learning_rate": 0.000199999465185184,
-      "loss": 46.0,
-      "step": 13626
-    },
-    {
-      "epoch": 1.0418793126517194,
-      "grad_norm": 0.0011563485022634268,
-      "learning_rate": 0.00019999946510662435,
-      "loss": 46.0,
-      "step": 13627
-    },
-    {
-      "epoch": 1.041955769635109,
-      "grad_norm": 0.0008899495005607605,
-      "learning_rate": 0.00019999946502805894,
-      "loss": 46.0,
-      "step": 13628
-    },
-    {
-      "epoch": 1.0420322266184987,
-      "grad_norm": 0.003935758024454117,
-      "learning_rate": 0.00019999946494948779,
-      "loss": 46.0,
-      "step": 13629
-    },
-    {
-      "epoch": 1.0421086836018885,
-      "grad_norm": 0.0012259908253327012,
-      "learning_rate": 0.00019999946487091083,
-      "loss": 46.0,
-      "step": 13630
-    },
-    {
-      "epoch": 1.0421851405852782,
-      "grad_norm": 0.0007822164334356785,
-      "learning_rate": 0.00019999946479232813,
-      "loss": 46.0,
-      "step": 13631
-    },
-    {
-      "epoch": 1.042261597568668,
-      "grad_norm": 0.0005279415054246783,
-      "learning_rate": 0.00019999946471373965,
-      "loss": 46.0,
-      "step": 13632
-    },
-    {
-      "epoch": 1.0423380545520577,
-      "grad_norm": 0.0008386989939026535,
-      "learning_rate": 0.0001999994646351454,
-      "loss": 46.0,
-      "step": 13633
-    },
-    {
-      "epoch": 1.0424145115354473,
-      "grad_norm": 0.0028680155519396067,
-      "learning_rate": 0.00019999946455654538,
-      "loss": 46.0,
-      "step": 13634
-    },
-    {
-      "epoch": 1.042490968518837,
-      "grad_norm": 0.0035199858248233795,
-      "learning_rate": 0.00019999946447793958,
-      "loss": 46.0,
-      "step": 13635
-    },
-    {
-      "epoch": 1.0425674255022268,
-      "grad_norm": 0.000766197161283344,
-      "learning_rate": 0.00019999946439932804,
-      "loss": 46.0,
-      "step": 13636
-    },
-    {
-      "epoch": 1.0426438824856166,
-      "grad_norm": 0.00039460984407924116,
-      "learning_rate": 0.00019999946432071073,
-      "loss": 46.0,
-      "step": 13637
-    },
-    {
-      "epoch": 1.0427203394690063,
-      "grad_norm": 0.0006706234416924417,
-      "learning_rate": 0.0001999994642420876,
-      "loss": 46.0,
-      "step": 13638
-    },
-    {
-      "epoch": 1.0427967964523959,
-      "grad_norm": 0.00034934974974021316,
-      "learning_rate": 0.00019999946416345875,
-      "loss": 46.0,
-      "step": 13639
-    },
-    {
-      "epoch": 1.0428732534357856,
-      "grad_norm": 0.0050964984111487865,
-      "learning_rate": 0.00019999946408482414,
-      "loss": 46.0,
-      "step": 13640
-    },
-    {
-      "epoch": 1.0429497104191754,
-      "grad_norm": 0.001377904205583036,
-      "learning_rate": 0.0001999994640061837,
-      "loss": 46.0,
-      "step": 13641
-    },
-    {
-      "epoch": 1.0430261674025652,
-      "grad_norm": 0.002273187041282654,
-      "learning_rate": 0.00019999946392753755,
-      "loss": 46.0,
-      "step": 13642
-    },
-    {
-      "epoch": 1.043102624385955,
-      "grad_norm": 0.006097042933106422,
-      "learning_rate": 0.0001999994638488856,
-      "loss": 46.0,
-      "step": 13643
-    },
-    {
-      "epoch": 1.0431790813693447,
-      "grad_norm": 0.002980379853397608,
-      "learning_rate": 0.00019999946377022786,
-      "loss": 46.0,
-      "step": 13644
-    },
-    {
-      "epoch": 1.0432555383527342,
-      "grad_norm": 0.002494513988494873,
-      "learning_rate": 0.0001999994636915644,
-      "loss": 46.0,
-      "step": 13645
-    },
-    {
-      "epoch": 1.043331995336124,
-      "grad_norm": 0.0011300815967842937,
-      "learning_rate": 0.00019999946361289514,
-      "loss": 46.0,
-      "step": 13646
-    },
-    {
-      "epoch": 1.0434084523195137,
-      "grad_norm": 0.0005182846216484904,
-      "learning_rate": 0.00019999946353422012,
-      "loss": 46.0,
-      "step": 13647
-    },
-    {
-      "epoch": 1.0434849093029035,
-      "grad_norm": 0.00644265441223979,
-      "learning_rate": 0.00019999946345553932,
-      "loss": 46.0,
-      "step": 13648
-    },
-    {
-      "epoch": 1.0435613662862933,
-      "grad_norm": 0.005353832151740789,
-      "learning_rate": 0.00019999946337685278,
-      "loss": 46.0,
-      "step": 13649
-    },
-    {
-      "epoch": 1.0436378232696828,
-      "grad_norm": 0.0018770033493638039,
-      "learning_rate": 0.00019999946329816046,
-      "loss": 46.0,
-      "step": 13650
-    },
-    {
-      "epoch": 1.0437142802530726,
-      "grad_norm": 0.0013232798082754016,
-      "learning_rate": 0.00019999946321946235,
-      "loss": 46.0,
-      "step": 13651
-    },
-    {
-      "epoch": 1.0437907372364623,
-      "grad_norm": 0.0009263827814720571,
-      "learning_rate": 0.00019999946314075849,
-      "loss": 46.0,
-      "step": 13652
-    },
-    {
-      "epoch": 1.043867194219852,
-      "grad_norm": 0.0035563067067414522,
-      "learning_rate": 0.00019999946306204885,
-      "loss": 46.0,
-      "step": 13653
-    },
-    {
-      "epoch": 1.0439436512032418,
-      "grad_norm": 0.005331353284418583,
-      "learning_rate": 0.00019999946298333344,
-      "loss": 46.0,
-      "step": 13654
-    },
-    {
-      "epoch": 1.0440201081866316,
-      "grad_norm": 0.001637037261389196,
-      "learning_rate": 0.00019999946290461223,
-      "loss": 46.0,
-      "step": 13655
-    },
-    {
-      "epoch": 1.0440965651700211,
-      "grad_norm": 0.0008109994232654572,
-      "learning_rate": 0.0001999994628258853,
-      "loss": 46.0,
-      "step": 13656
-    },
-    {
-      "epoch": 1.044173022153411,
-      "grad_norm": 0.004891208838671446,
-      "learning_rate": 0.0001999994627471526,
-      "loss": 46.0,
-      "step": 13657
-    },
-    {
-      "epoch": 1.0442494791368007,
-      "grad_norm": 0.004336840473115444,
-      "learning_rate": 0.00019999946266841413,
-      "loss": 46.0,
-      "step": 13658
-    },
-    {
-      "epoch": 1.0443259361201904,
-      "grad_norm": 0.0029382656794041395,
-      "learning_rate": 0.00019999946258966985,
-      "loss": 46.0,
-      "step": 13659
-    },
-    {
-      "epoch": 1.0444023931035802,
-      "grad_norm": 0.0004281848669052124,
-      "learning_rate": 0.00019999946251091983,
-      "loss": 46.0,
-      "step": 13660
-    },
-    {
-      "epoch": 1.0444788500869697,
-      "grad_norm": 0.00045410022721625865,
-      "learning_rate": 0.00019999946243216403,
-      "loss": 46.0,
-      "step": 13661
-    },
-    {
-      "epoch": 1.0445553070703595,
-      "grad_norm": 0.0012513186084106565,
-      "learning_rate": 0.00019999946235340247,
-      "loss": 46.0,
-      "step": 13662
-    },
-    {
-      "epoch": 1.0446317640537492,
-      "grad_norm": 0.0007714648963883519,
-      "learning_rate": 0.00019999946227463515,
-      "loss": 46.0,
-      "step": 13663
-    },
-    {
-      "epoch": 1.044708221037139,
-      "grad_norm": 0.0010255188681185246,
-      "learning_rate": 0.00019999946219586203,
-      "loss": 46.0,
-      "step": 13664
-    },
-    {
-      "epoch": 1.0447846780205288,
-      "grad_norm": 0.0007217352977022529,
-      "learning_rate": 0.00019999946211708317,
-      "loss": 46.0,
-      "step": 13665
-    },
-    {
-      "epoch": 1.0448611350039183,
-      "grad_norm": 0.003992976620793343,
-      "learning_rate": 0.00019999946203829854,
-      "loss": 46.0,
-      "step": 13666
-    },
-    {
-      "epoch": 1.044937591987308,
-      "grad_norm": 0.012705071829259396,
-      "learning_rate": 0.00019999946195950813,
-      "loss": 46.0,
-      "step": 13667
-    },
-    {
-      "epoch": 1.0450140489706978,
-      "grad_norm": 0.009934223257005215,
-      "learning_rate": 0.00019999946188071195,
-      "loss": 46.0,
-      "step": 13668
-    },
-    {
-      "epoch": 1.0450905059540876,
-      "grad_norm": 0.003383256494998932,
-      "learning_rate": 0.00019999946180191,
-      "loss": 46.0,
-      "step": 13669
-    },
-    {
-      "epoch": 1.0451669629374774,
-      "grad_norm": 0.0009678475908003747,
-      "learning_rate": 0.0001999994617231023,
-      "loss": 46.0,
-      "step": 13670
-    },
-    {
-      "epoch": 1.0452434199208671,
-      "grad_norm": 0.0013291980139911175,
-      "learning_rate": 0.0001999994616442888,
-      "loss": 46.0,
-      "step": 13671
-    },
-    {
-      "epoch": 1.0453198769042567,
-      "grad_norm": 0.0007834390853531659,
-      "learning_rate": 0.00019999946156546954,
-      "loss": 46.0,
-      "step": 13672
-    },
-    {
-      "epoch": 1.0453963338876464,
-      "grad_norm": 0.004882368259131908,
-      "learning_rate": 0.0001999994614866445,
-      "loss": 46.0,
-      "step": 13673
-    },
-    {
-      "epoch": 1.0454727908710362,
-      "grad_norm": 0.005486754700541496,
-      "learning_rate": 0.00019999946140781372,
-      "loss": 46.0,
-      "step": 13674
-    },
-    {
-      "epoch": 1.045549247854426,
-      "grad_norm": 0.0036457020323723555,
-      "learning_rate": 0.00019999946132897716,
-      "loss": 46.0,
-      "step": 13675
-    },
-    {
-      "epoch": 1.0456257048378157,
-      "grad_norm": 0.0005810261936858296,
-      "learning_rate": 0.0001999994612501348,
-      "loss": 46.0,
-      "step": 13676
-    },
-    {
-      "epoch": 1.0457021618212052,
-      "grad_norm": 0.0019671572372317314,
-      "learning_rate": 0.00019999946117128673,
-      "loss": 46.0,
-      "step": 13677
-    },
-    {
-      "epoch": 1.045778618804595,
-      "grad_norm": 0.0008908930467441678,
-      "learning_rate": 0.00019999946109243284,
-      "loss": 46.0,
-      "step": 13678
-    },
-    {
-      "epoch": 1.0458550757879848,
-      "grad_norm": 0.001128511969000101,
-      "learning_rate": 0.00019999946101357318,
-      "loss": 46.0,
-      "step": 13679
-    },
-    {
-      "epoch": 1.0459315327713745,
-      "grad_norm": 0.0010615490609779954,
-      "learning_rate": 0.0001999994609347078,
-      "loss": 46.0,
-      "step": 13680
-    },
-    {
-      "epoch": 1.0460079897547643,
-      "grad_norm": 0.001050188671797514,
-      "learning_rate": 0.00019999946085583661,
-      "loss": 46.0,
-      "step": 13681
-    },
-    {
-      "epoch": 1.046084446738154,
-      "grad_norm": 0.0023927355650812387,
-      "learning_rate": 0.00019999946077695966,
-      "loss": 46.0,
-      "step": 13682
-    },
-    {
-      "epoch": 1.0461609037215436,
-      "grad_norm": 0.0008692205883562565,
-      "learning_rate": 0.00019999946069807693,
-      "loss": 46.0,
-      "step": 13683
-    },
-    {
-      "epoch": 1.0462373607049333,
-      "grad_norm": 0.008289349265396595,
-      "learning_rate": 0.00019999946061918843,
-      "loss": 46.0,
-      "step": 13684
-    },
-    {
-      "epoch": 1.046313817688323,
-      "grad_norm": 0.001233914983458817,
-      "learning_rate": 0.00019999946054029418,
-      "loss": 46.0,
-      "step": 13685
-    },
-    {
-      "epoch": 1.0463902746717129,
-      "grad_norm": 0.002012679586187005,
-      "learning_rate": 0.00019999946046139416,
-      "loss": 46.0,
-      "step": 13686
-    },
-    {
-      "epoch": 1.0464667316551026,
-      "grad_norm": 0.00028954027220606804,
-      "learning_rate": 0.00019999946038248836,
-      "loss": 46.0,
-      "step": 13687
-    },
-    {
-      "epoch": 1.0465431886384922,
-      "grad_norm": 0.007841721177101135,
-      "learning_rate": 0.0001999994603035768,
-      "loss": 46.0,
-      "step": 13688
-    },
-    {
-      "epoch": 1.046619645621882,
-      "grad_norm": 0.0012884187744930387,
-      "learning_rate": 0.00019999946022465946,
-      "loss": 46.0,
-      "step": 13689
-    },
-    {
-      "epoch": 1.0466961026052717,
-      "grad_norm": 0.004484914243221283,
-      "learning_rate": 0.00019999946014573637,
-      "loss": 46.0,
-      "step": 13690
-    },
-    {
-      "epoch": 1.0467725595886614,
-      "grad_norm": 0.0009876226540654898,
-      "learning_rate": 0.00019999946006680748,
-      "loss": 46.0,
-      "step": 13691
-    },
-    {
-      "epoch": 1.0468490165720512,
-      "grad_norm": 0.0017544401343911886,
-      "learning_rate": 0.00019999945998787282,
-      "loss": 46.0,
-      "step": 13692
-    },
-    {
-      "epoch": 1.046925473555441,
-      "grad_norm": 0.001603399170562625,
-      "learning_rate": 0.0001999994599089324,
-      "loss": 46.0,
-      "step": 13693
-    },
-    {
-      "epoch": 1.0470019305388305,
-      "grad_norm": 0.0014639829751104116,
-      "learning_rate": 0.00019999945982998623,
-      "loss": 46.0,
-      "step": 13694
-    },
-    {
-      "epoch": 1.0470783875222203,
-      "grad_norm": 0.0008471490582451224,
-      "learning_rate": 0.00019999945975103427,
-      "loss": 46.0,
-      "step": 13695
-    },
-    {
-      "epoch": 1.04715484450561,
-      "grad_norm": 0.003734071273356676,
-      "learning_rate": 0.00019999945967207655,
-      "loss": 46.0,
-      "step": 13696
-    },
-    {
-      "epoch": 1.0472313014889998,
-      "grad_norm": 0.013389692641794682,
-      "learning_rate": 0.00019999945959311305,
-      "loss": 46.0,
-      "step": 13697
-    },
-    {
-      "epoch": 1.0473077584723895,
-      "grad_norm": 0.00100217224098742,
-      "learning_rate": 0.0001999994595141438,
-      "loss": 46.0,
-      "step": 13698
-    },
-    {
-      "epoch": 1.047384215455779,
-      "grad_norm": 0.0021251337602734566,
-      "learning_rate": 0.00019999945943516875,
-      "loss": 46.0,
-      "step": 13699
-    },
-    {
-      "epoch": 1.0474606724391688,
-      "grad_norm": 0.001842330559156835,
-      "learning_rate": 0.00019999945935618798,
-      "loss": 46.0,
-      "step": 13700
-    },
-    {
-      "epoch": 1.0475371294225586,
-      "grad_norm": 0.000933229923248291,
-      "learning_rate": 0.0001999994592772014,
-      "loss": 46.0,
-      "step": 13701
-    },
-    {
-      "epoch": 1.0476135864059484,
-      "grad_norm": 0.0006351879565045238,
-      "learning_rate": 0.00019999945919820907,
-      "loss": 46.0,
-      "step": 13702
-    },
-    {
-      "epoch": 1.0476900433893381,
-      "grad_norm": 0.001479899394325912,
-      "learning_rate": 0.00019999945911921096,
-      "loss": 46.0,
-      "step": 13703
-    },
-    {
-      "epoch": 1.047766500372728,
-      "grad_norm": 0.0015070316148921847,
-      "learning_rate": 0.00019999945904020707,
-      "loss": 46.0,
-      "step": 13704
-    },
-    {
-      "epoch": 1.0478429573561174,
-      "grad_norm": 0.001733145210891962,
-      "learning_rate": 0.00019999945896119744,
-      "loss": 46.0,
-      "step": 13705
-    },
-    {
-      "epoch": 1.0479194143395072,
-      "grad_norm": 0.0010924865491688251,
-      "learning_rate": 0.000199999458882182,
-      "loss": 46.0,
-      "step": 13706
-    },
-    {
-      "epoch": 1.047995871322897,
-      "grad_norm": 0.0009933550609275699,
-      "learning_rate": 0.00019999945880316082,
-      "loss": 46.0,
-      "step": 13707
-    },
-    {
-      "epoch": 1.0480723283062867,
-      "grad_norm": 0.003499661572277546,
-      "learning_rate": 0.00019999945872413387,
-      "loss": 46.0,
-      "step": 13708
-    },
-    {
-      "epoch": 1.0481487852896765,
-      "grad_norm": 0.001802286715246737,
-      "learning_rate": 0.00019999945864510114,
-      "loss": 46.0,
-      "step": 13709
-    },
-    {
-      "epoch": 1.048225242273066,
-      "grad_norm": 0.004978041164577007,
-      "learning_rate": 0.00019999945856606264,
-      "loss": 46.0,
-      "step": 13710
-    },
-    {
-      "epoch": 1.0483016992564558,
-      "grad_norm": 0.0018099646549671888,
-      "learning_rate": 0.00019999945848701836,
-      "loss": 46.0,
-      "step": 13711
-    },
-    {
-      "epoch": 1.0483781562398455,
-      "grad_norm": 0.001607256941497326,
-      "learning_rate": 0.00019999945840796834,
-      "loss": 46.0,
-      "step": 13712
-    },
-    {
-      "epoch": 1.0484546132232353,
-      "grad_norm": 0.0012299392838031054,
-      "learning_rate": 0.00019999945832891255,
-      "loss": 46.0,
-      "step": 13713
-    },
-    {
-      "epoch": 1.048531070206625,
-      "grad_norm": 0.0019190805032849312,
-      "learning_rate": 0.00019999945824985096,
-      "loss": 46.0,
-      "step": 13714
-    },
-    {
-      "epoch": 1.0486075271900148,
-      "grad_norm": 0.0013620478566735983,
-      "learning_rate": 0.00019999945817078361,
-      "loss": 46.0,
-      "step": 13715
-    },
-    {
-      "epoch": 1.0486839841734044,
-      "grad_norm": 0.0010639394167810678,
-      "learning_rate": 0.00019999945809171053,
-      "loss": 46.0,
-      "step": 13716
-    },
-    {
-      "epoch": 1.0487604411567941,
-      "grad_norm": 0.002282089786604047,
-      "learning_rate": 0.00019999945801263164,
-      "loss": 46.0,
-      "step": 13717
-    },
-    {
-      "epoch": 1.0488368981401839,
-      "grad_norm": 0.0007381599280051887,
-      "learning_rate": 0.000199999457933547,
-      "loss": 46.0,
-      "step": 13718
-    },
-    {
-      "epoch": 1.0489133551235736,
-      "grad_norm": 0.0020886831916868687,
-      "learning_rate": 0.00019999945785445657,
-      "loss": 46.0,
-      "step": 13719
-    },
-    {
-      "epoch": 1.0489898121069634,
-      "grad_norm": 0.0005917309317737818,
-      "learning_rate": 0.00019999945777536036,
-      "loss": 46.0,
-      "step": 13720
-    },
-    {
-      "epoch": 1.049066269090353,
-      "grad_norm": 0.002708891173824668,
-      "learning_rate": 0.0001999994576962584,
-      "loss": 46.0,
-      "step": 13721
-    },
-    {
-      "epoch": 1.0491427260737427,
-      "grad_norm": 0.001117902691476047,
-      "learning_rate": 0.0001999994576171507,
-      "loss": 46.0,
-      "step": 13722
-    },
-    {
-      "epoch": 1.0492191830571325,
-      "grad_norm": 0.002134914742782712,
-      "learning_rate": 0.00019999945753803718,
-      "loss": 46.0,
-      "step": 13723
-    },
-    {
-      "epoch": 1.0492956400405222,
-      "grad_norm": 0.0005679893074557185,
-      "learning_rate": 0.00019999945745891794,
-      "loss": 46.0,
-      "step": 13724
-    },
-    {
-      "epoch": 1.049372097023912,
-      "grad_norm": 0.0011671686079353094,
-      "learning_rate": 0.0001999994573797929,
-      "loss": 46.0,
-      "step": 13725
-    },
-    {
-      "epoch": 1.0494485540073017,
-      "grad_norm": 0.0013793562538921833,
-      "learning_rate": 0.0001999994573006621,
-      "loss": 46.0,
-      "step": 13726
-    },
-    {
-      "epoch": 1.0495250109906913,
-      "grad_norm": 0.009731952100992203,
-      "learning_rate": 0.0001999994572215255,
-      "loss": 46.0,
-      "step": 13727
-    },
-    {
-      "epoch": 1.049601467974081,
-      "grad_norm": 0.009387202560901642,
-      "learning_rate": 0.0001999994571423832,
-      "loss": 46.0,
-      "step": 13728
-    },
-    {
-      "epoch": 1.0496779249574708,
-      "grad_norm": 0.0028804149478673935,
-      "learning_rate": 0.00019999945706323505,
-      "loss": 46.0,
-      "step": 13729
-    },
-    {
-      "epoch": 1.0497543819408606,
-      "grad_norm": 0.0013176982756704092,
-      "learning_rate": 0.00019999945698408116,
-      "loss": 46.0,
-      "step": 13730
-    },
-    {
-      "epoch": 1.0498308389242503,
-      "grad_norm": 0.0037029809318482876,
-      "learning_rate": 0.00019999945690492152,
-      "loss": 46.0,
-      "step": 13731
-    },
-    {
-      "epoch": 1.0499072959076399,
-      "grad_norm": 0.0010358724975958467,
-      "learning_rate": 0.0001999994568257561,
-      "loss": 46.0,
-      "step": 13732
-    },
-    {
-      "epoch": 1.0499837528910296,
-      "grad_norm": 0.004200923256576061,
-      "learning_rate": 0.0001999994567465849,
-      "loss": 46.0,
-      "step": 13733
-    },
-    {
-      "epoch": 1.0500602098744194,
-      "grad_norm": 0.0006061145686544478,
-      "learning_rate": 0.00019999945666740793,
-      "loss": 46.0,
-      "step": 13734
-    },
-    {
-      "epoch": 1.0501366668578092,
-      "grad_norm": 0.0010573656763881445,
-      "learning_rate": 0.00019999945658822523,
-      "loss": 46.0,
-      "step": 13735
-    },
-    {
-      "epoch": 1.050213123841199,
-      "grad_norm": 0.0007051253342069685,
-      "learning_rate": 0.0001999994565090367,
-      "loss": 46.0,
-      "step": 13736
-    },
-    {
-      "epoch": 1.0502895808245887,
-      "grad_norm": 0.0010889818659052253,
-      "learning_rate": 0.00019999945642984246,
-      "loss": 46.0,
-      "step": 13737
-    },
-    {
-      "epoch": 1.0503660378079782,
-      "grad_norm": 0.0014793929876759648,
-      "learning_rate": 0.0001999994563506424,
-      "loss": 46.0,
-      "step": 13738
-    },
-    {
-      "epoch": 1.050442494791368,
-      "grad_norm": 0.000877274782396853,
-      "learning_rate": 0.0001999994562714366,
-      "loss": 46.0,
-      "step": 13739
-    },
-    {
-      "epoch": 1.0505189517747577,
-      "grad_norm": 0.005213289987295866,
-      "learning_rate": 0.00019999945619222502,
-      "loss": 46.0,
-      "step": 13740
-    },
-    {
-      "epoch": 1.0505954087581475,
-      "grad_norm": 0.0015559031162410975,
-      "learning_rate": 0.00019999945611300768,
-      "loss": 46.0,
-      "step": 13741
-    },
-    {
-      "epoch": 1.0506718657415373,
-      "grad_norm": 0.0010985876433551311,
-      "learning_rate": 0.00019999945603378457,
-      "loss": 46.0,
-      "step": 13742
-    },
-    {
-      "epoch": 1.0507483227249268,
-      "grad_norm": 0.001478587044402957,
-      "learning_rate": 0.00019999945595455568,
-      "loss": 46.0,
-      "step": 13743
-    },
-    {
-      "epoch": 1.0508247797083166,
-      "grad_norm": 0.0006309555028565228,
-      "learning_rate": 0.000199999455875321,
-      "loss": 46.0,
-      "step": 13744
-    },
-    {
-      "epoch": 1.0509012366917063,
-      "grad_norm": 0.004757370334118605,
-      "learning_rate": 0.0001999994557960806,
-      "loss": 46.0,
-      "step": 13745
-    },
-    {
-      "epoch": 1.050977693675096,
-      "grad_norm": 0.004981026519089937,
-      "learning_rate": 0.0001999994557168344,
-      "loss": 46.0,
-      "step": 13746
-    },
-    {
-      "epoch": 1.0510541506584858,
-      "grad_norm": 0.0008750940323807299,
-      "learning_rate": 0.00019999945563758243,
-      "loss": 46.0,
-      "step": 13747
-    },
-    {
-      "epoch": 1.0511306076418756,
-      "grad_norm": 0.0010408051311969757,
-      "learning_rate": 0.0001999994555583247,
-      "loss": 46.0,
-      "step": 13748
-    },
-    {
-      "epoch": 1.0512070646252651,
-      "grad_norm": 0.0021473548840731382,
-      "learning_rate": 0.0001999994554790612,
-      "loss": 46.0,
-      "step": 13749
-    },
-    {
-      "epoch": 1.051283521608655,
-      "grad_norm": 0.0005332798464223742,
-      "learning_rate": 0.00019999945539979193,
-      "loss": 46.0,
-      "step": 13750
-    },
-    {
-      "epoch": 1.0513599785920447,
-      "grad_norm": 0.0025775327812880278,
-      "learning_rate": 0.00019999945532051688,
-      "loss": 46.0,
-      "step": 13751
-    },
-    {
-      "epoch": 1.0514364355754344,
-      "grad_norm": 0.014799004420638084,
-      "learning_rate": 0.0001999994552412361,
-      "loss": 46.0,
-      "step": 13752
-    },
-    {
-      "epoch": 1.0515128925588242,
-      "grad_norm": 0.002512461505830288,
-      "learning_rate": 0.00019999945516194952,
-      "loss": 46.0,
-      "step": 13753
-    },
-    {
-      "epoch": 1.0515893495422137,
-      "grad_norm": 0.011496616527438164,
-      "learning_rate": 0.00019999945508265715,
-      "loss": 46.0,
-      "step": 13754
-    },
-    {
-      "epoch": 1.0516658065256035,
-      "grad_norm": 0.0006701357197016478,
-      "learning_rate": 0.00019999945500335904,
-      "loss": 46.0,
-      "step": 13755
-    },
-    {
-      "epoch": 1.0517422635089932,
-      "grad_norm": 0.00208885595202446,
-      "learning_rate": 0.00019999945492405515,
-      "loss": 46.0,
-      "step": 13756
-    },
-    {
-      "epoch": 1.051818720492383,
-      "grad_norm": 0.0008675071294419467,
-      "learning_rate": 0.00019999945484474547,
-      "loss": 46.0,
-      "step": 13757
-    },
-    {
-      "epoch": 1.0518951774757728,
-      "grad_norm": 0.0010657686507329345,
-      "learning_rate": 0.00019999945476543006,
-      "loss": 46.0,
-      "step": 13758
-    },
-    {
-      "epoch": 1.0519716344591623,
-      "grad_norm": 0.0007977762725204229,
-      "learning_rate": 0.00019999945468610885,
-      "loss": 46.0,
-      "step": 13759
-    },
-    {
-      "epoch": 1.052048091442552,
-      "grad_norm": 0.0009384271106682718,
-      "learning_rate": 0.0001999994546067819,
-      "loss": 46.0,
-      "step": 13760
-    },
-    {
-      "epoch": 1.0521245484259418,
-      "grad_norm": 0.0009604932856746018,
-      "learning_rate": 0.00019999945452744915,
-      "loss": 46.0,
-      "step": 13761
-    },
-    {
-      "epoch": 1.0522010054093316,
-      "grad_norm": 0.0028188966680318117,
-      "learning_rate": 0.00019999945444811065,
-      "loss": 46.0,
-      "step": 13762
-    },
-    {
-      "epoch": 1.0522774623927214,
-      "grad_norm": 0.004643087275326252,
-      "learning_rate": 0.0001999994543687664,
-      "loss": 46.0,
-      "step": 13763
-    },
-    {
-      "epoch": 1.0523539193761111,
-      "grad_norm": 0.0006218428025022149,
-      "learning_rate": 0.00019999945428941635,
-      "loss": 46.0,
-      "step": 13764
-    },
-    {
-      "epoch": 1.0524303763595007,
-      "grad_norm": 0.007419173140078783,
-      "learning_rate": 0.00019999945421006053,
-      "loss": 46.0,
-      "step": 13765
-    },
-    {
-      "epoch": 1.0525068333428904,
-      "grad_norm": 0.00118886714335531,
-      "learning_rate": 0.00019999945413069897,
-      "loss": 46.0,
-      "step": 13766
-    },
-    {
-      "epoch": 1.0525832903262802,
-      "grad_norm": 0.00042946290341205895,
-      "learning_rate": 0.0001999994540513316,
-      "loss": 46.0,
-      "step": 13767
-    },
-    {
-      "epoch": 1.05265974730967,
-      "grad_norm": 0.0006862473674118519,
-      "learning_rate": 0.00019999945397195846,
-      "loss": 46.0,
-      "step": 13768
-    },
-    {
-      "epoch": 1.0527362042930597,
-      "grad_norm": 0.00039609955274499953,
-      "learning_rate": 0.00019999945389257958,
-      "loss": 46.0,
-      "step": 13769
-    },
-    {
-      "epoch": 1.0528126612764492,
-      "grad_norm": 0.0013059170451015234,
-      "learning_rate": 0.00019999945381319492,
-      "loss": 46.0,
-      "step": 13770
-    },
-    {
-      "epoch": 1.052889118259839,
-      "grad_norm": 0.00047861848725005984,
-      "learning_rate": 0.00019999945373380448,
-      "loss": 46.0,
-      "step": 13771
-    },
-    {
-      "epoch": 1.0529655752432288,
-      "grad_norm": 0.0002761677314992994,
-      "learning_rate": 0.00019999945365440828,
-      "loss": 46.0,
-      "step": 13772
-    },
-    {
-      "epoch": 1.0530420322266185,
-      "grad_norm": 0.0013266659807413816,
-      "learning_rate": 0.00019999945357500632,
-      "loss": 46.0,
-      "step": 13773
-    },
-    {
-      "epoch": 1.0531184892100083,
-      "grad_norm": 0.0019480689661577344,
-      "learning_rate": 0.0001999994534955986,
-      "loss": 46.0,
-      "step": 13774
-    },
-    {
-      "epoch": 1.053194946193398,
-      "grad_norm": 0.001726043876260519,
-      "learning_rate": 0.0001999994534161851,
-      "loss": 46.0,
-      "step": 13775
-    },
-    {
-      "epoch": 1.0532714031767876,
-      "grad_norm": 0.0012734434567391872,
-      "learning_rate": 0.0001999994533367658,
-      "loss": 46.0,
-      "step": 13776
-    },
-    {
-      "epoch": 1.0533478601601773,
-      "grad_norm": 0.0017384429229423404,
-      "learning_rate": 0.00019999945325734075,
-      "loss": 46.0,
-      "step": 13777
-    },
-    {
-      "epoch": 1.053424317143567,
-      "grad_norm": 0.0011614080285653472,
-      "learning_rate": 0.00019999945317790996,
-      "loss": 46.0,
-      "step": 13778
-    },
-    {
-      "epoch": 1.0535007741269569,
-      "grad_norm": 0.0009231917792931199,
-      "learning_rate": 0.00019999945309847337,
-      "loss": 46.0,
-      "step": 13779
-    },
-    {
-      "epoch": 1.0535772311103466,
-      "grad_norm": 0.0010538025526329875,
-      "learning_rate": 0.000199999453019031,
-      "loss": 46.0,
-      "step": 13780
-    },
-    {
-      "epoch": 1.0536536880937362,
-      "grad_norm": 0.00501682935282588,
-      "learning_rate": 0.00019999945293958286,
-      "loss": 46.0,
-      "step": 13781
-    },
-    {
-      "epoch": 1.053730145077126,
-      "grad_norm": 0.002531415782868862,
-      "learning_rate": 0.000199999452860129,
-      "loss": 46.0,
-      "step": 13782
-    },
-    {
-      "epoch": 1.0538066020605157,
-      "grad_norm": 0.0007454706355929375,
-      "learning_rate": 0.00019999945278066932,
-      "loss": 46.0,
-      "step": 13783
-    },
-    {
-      "epoch": 1.0538830590439054,
-      "grad_norm": 0.004449669271707535,
-      "learning_rate": 0.0001999994527012039,
-      "loss": 46.0,
-      "step": 13784
-    },
-    {
-      "epoch": 1.0539595160272952,
-      "grad_norm": 0.0010712840594351292,
-      "learning_rate": 0.00019999945262173268,
-      "loss": 46.0,
-      "step": 13785
-    },
-    {
-      "epoch": 1.054035973010685,
-      "grad_norm": 0.002205733209848404,
-      "learning_rate": 0.00019999945254225572,
-      "loss": 46.0,
-      "step": 13786
-    },
-    {
-      "epoch": 1.0541124299940745,
-      "grad_norm": 0.001104493043385446,
-      "learning_rate": 0.00019999945246277297,
-      "loss": 46.0,
-      "step": 13787
-    },
-    {
-      "epoch": 1.0541888869774643,
-      "grad_norm": 0.0007203186396509409,
-      "learning_rate": 0.00019999945238328447,
-      "loss": 46.0,
-      "step": 13788
-    },
-    {
-      "epoch": 1.054265343960854,
-      "grad_norm": 0.002434931695461273,
-      "learning_rate": 0.0001999994523037902,
-      "loss": 46.0,
-      "step": 13789
-    },
-    {
-      "epoch": 1.0543418009442438,
-      "grad_norm": 0.0015634875744581223,
-      "learning_rate": 0.00019999945222429015,
-      "loss": 46.0,
-      "step": 13790
-    },
-    {
-      "epoch": 1.0544182579276336,
-      "grad_norm": 0.0012279485817998648,
-      "learning_rate": 0.00019999945214478433,
-      "loss": 46.0,
-      "step": 13791
-    },
-    {
-      "epoch": 1.054494714911023,
-      "grad_norm": 0.0005795487086288631,
-      "learning_rate": 0.00019999945206527274,
-      "loss": 46.0,
-      "step": 13792
-    },
-    {
-      "epoch": 1.0545711718944129,
-      "grad_norm": 0.0017037650104612112,
-      "learning_rate": 0.00019999945198575538,
-      "loss": 46.0,
-      "step": 13793
-    },
-    {
-      "epoch": 1.0546476288778026,
-      "grad_norm": 0.004591943230479956,
-      "learning_rate": 0.00019999945190623226,
-      "loss": 46.0,
-      "step": 13794
-    },
-    {
-      "epoch": 1.0547240858611924,
-      "grad_norm": 0.0007427557138726115,
-      "learning_rate": 0.00019999945182670335,
-      "loss": 46.0,
-      "step": 13795
-    },
-    {
-      "epoch": 1.0548005428445821,
-      "grad_norm": 0.0005532317445613444,
-      "learning_rate": 0.0001999994517471687,
-      "loss": 46.0,
-      "step": 13796
-    },
-    {
-      "epoch": 1.0548769998279717,
-      "grad_norm": 0.00041654807864688337,
-      "learning_rate": 0.00019999945166762826,
-      "loss": 46.0,
-      "step": 13797
-    },
-    {
-      "epoch": 1.0549534568113614,
-      "grad_norm": 0.0021832596976310015,
-      "learning_rate": 0.00019999945158808205,
-      "loss": 46.0,
-      "step": 13798
-    },
-    {
-      "epoch": 1.0550299137947512,
-      "grad_norm": 0.0008539077243767679,
-      "learning_rate": 0.00019999945150853007,
-      "loss": 46.0,
-      "step": 13799
-    },
-    {
-      "epoch": 1.055106370778141,
-      "grad_norm": 0.0006561922491528094,
-      "learning_rate": 0.00019999945142897235,
-      "loss": 46.0,
-      "step": 13800
-    },
-    {
-      "epoch": 1.0551828277615307,
-      "grad_norm": 0.00170541659463197,
-      "learning_rate": 0.00019999945134940882,
-      "loss": 46.0,
-      "step": 13801
-    },
-    {
-      "epoch": 1.0552592847449205,
-      "grad_norm": 0.009879326447844505,
-      "learning_rate": 0.00019999945126983955,
-      "loss": 46.0,
-      "step": 13802
-    },
-    {
-      "epoch": 1.05533574172831,
-      "grad_norm": 0.00253881374374032,
-      "learning_rate": 0.00019999945119026448,
-      "loss": 46.0,
-      "step": 13803
-    },
-    {
-      "epoch": 1.0554121987116998,
-      "grad_norm": 0.004980718716979027,
-      "learning_rate": 0.0001999994511106837,
-      "loss": 46.0,
-      "step": 13804
-    },
-    {
-      "epoch": 1.0554886556950895,
-      "grad_norm": 0.0040793651714921,
-      "learning_rate": 0.0001999994510310971,
-      "loss": 46.0,
-      "step": 13805
-    },
-    {
-      "epoch": 1.0555651126784793,
-      "grad_norm": 0.0013489745324477553,
-      "learning_rate": 0.00019999945095150473,
-      "loss": 46.0,
-      "step": 13806
-    },
-    {
-      "epoch": 1.055641569661869,
-      "grad_norm": 0.0007245230954140425,
-      "learning_rate": 0.0001999994508719066,
-      "loss": 46.0,
-      "step": 13807
-    },
-    {
-      "epoch": 1.0557180266452586,
-      "grad_norm": 0.0034304941073060036,
-      "learning_rate": 0.00019999945079230268,
-      "loss": 46.0,
-      "step": 13808
-    },
-    {
-      "epoch": 1.0557944836286484,
-      "grad_norm": 0.0015015583485364914,
-      "learning_rate": 0.00019999945071269305,
-      "loss": 46.0,
-      "step": 13809
-    },
-    {
-      "epoch": 1.0558709406120381,
-      "grad_norm": 0.022670798003673553,
-      "learning_rate": 0.0001999994506330776,
-      "loss": 46.0,
-      "step": 13810
-    },
-    {
-      "epoch": 1.0559473975954279,
-      "grad_norm": 0.0004668761102948338,
-      "learning_rate": 0.00019999945055345638,
-      "loss": 46.0,
-      "step": 13811
-    },
-    {
-      "epoch": 1.0560238545788176,
-      "grad_norm": 0.0020051011815667152,
-      "learning_rate": 0.0001999994504738294,
-      "loss": 46.0,
-      "step": 13812
-    },
-    {
-      "epoch": 1.0561003115622074,
-      "grad_norm": 0.0006668036221526563,
-      "learning_rate": 0.00019999945039419665,
-      "loss": 46.0,
-      "step": 13813
-    },
-    {
-      "epoch": 1.056176768545597,
-      "grad_norm": 0.0025760773569345474,
-      "learning_rate": 0.00019999945031455812,
-      "loss": 46.0,
-      "step": 13814
-    },
-    {
-      "epoch": 1.0562532255289867,
-      "grad_norm": 0.002122888108715415,
-      "learning_rate": 0.00019999945023491385,
-      "loss": 46.0,
-      "step": 13815
-    },
-    {
-      "epoch": 1.0563296825123765,
-      "grad_norm": 0.0006113943527452648,
-      "learning_rate": 0.00019999945015526383,
-      "loss": 46.0,
-      "step": 13816
-    },
-    {
-      "epoch": 1.0564061394957662,
-      "grad_norm": 0.004621585365384817,
-      "learning_rate": 0.000199999450075608,
-      "loss": 46.0,
-      "step": 13817
-    },
-    {
-      "epoch": 1.056482596479156,
-      "grad_norm": 0.0005622573080472648,
-      "learning_rate": 0.0001999994499959464,
-      "loss": 46.0,
-      "step": 13818
-    },
-    {
-      "epoch": 1.0565590534625455,
-      "grad_norm": 0.0010672372300177813,
-      "learning_rate": 0.00019999944991627903,
-      "loss": 46.0,
-      "step": 13819
-    },
-    {
-      "epoch": 1.0566355104459353,
-      "grad_norm": 0.0035669091157615185,
-      "learning_rate": 0.0001999994498366059,
-      "loss": 46.0,
-      "step": 13820
-    },
-    {
-      "epoch": 1.056711967429325,
-      "grad_norm": 0.0020136700477451086,
-      "learning_rate": 0.00019999944975692698,
-      "loss": 46.0,
-      "step": 13821
-    },
-    {
-      "epoch": 1.0567884244127148,
-      "grad_norm": 0.002263610251247883,
-      "learning_rate": 0.00019999944967724232,
-      "loss": 46.0,
-      "step": 13822
-    },
-    {
-      "epoch": 1.0568648813961046,
-      "grad_norm": 0.0007125269039534032,
-      "learning_rate": 0.0001999994495975519,
-      "loss": 46.0,
-      "step": 13823
-    },
-    {
-      "epoch": 1.0569413383794943,
-      "grad_norm": 0.0004161091346759349,
-      "learning_rate": 0.00019999944951785569,
-      "loss": 46.0,
-      "step": 13824
-    },
-    {
-      "epoch": 1.0570177953628839,
-      "grad_norm": 0.0005981248104944825,
-      "learning_rate": 0.0001999994494381537,
-      "loss": 46.0,
-      "step": 13825
-    },
-    {
-      "epoch": 1.0570942523462736,
-      "grad_norm": 0.0007105072145350277,
-      "learning_rate": 0.00019999944935844595,
-      "loss": 46.0,
-      "step": 13826
-    },
-    {
-      "epoch": 1.0571707093296634,
-      "grad_norm": 0.006244134623557329,
-      "learning_rate": 0.00019999944927873243,
-      "loss": 46.0,
-      "step": 13827
-    },
-    {
-      "epoch": 1.0572471663130532,
-      "grad_norm": 0.000786393997259438,
-      "learning_rate": 0.00019999944919901316,
-      "loss": 46.0,
-      "step": 13828
-    },
-    {
-      "epoch": 1.057323623296443,
-      "grad_norm": 0.0021577521692961454,
-      "learning_rate": 0.00019999944911928809,
-      "loss": 46.0,
-      "step": 13829
-    },
-    {
-      "epoch": 1.0574000802798325,
-      "grad_norm": 0.0015617961762472987,
-      "learning_rate": 0.00019999944903955727,
-      "loss": 46.0,
-      "step": 13830
-    },
-    {
-      "epoch": 1.0574765372632222,
-      "grad_norm": 0.00087475631153211,
-      "learning_rate": 0.00019999944895982068,
-      "loss": 46.0,
-      "step": 13831
-    },
-    {
-      "epoch": 1.057552994246612,
-      "grad_norm": 0.0037890796083956957,
-      "learning_rate": 0.0001999994488800783,
-      "loss": 46.0,
-      "step": 13832
-    },
-    {
-      "epoch": 1.0576294512300017,
-      "grad_norm": 0.0032942306715995073,
-      "learning_rate": 0.0001999994488003302,
-      "loss": 46.0,
-      "step": 13833
-    },
-    {
-      "epoch": 1.0577059082133915,
-      "grad_norm": 0.0011341753415763378,
-      "learning_rate": 0.0001999994487205763,
-      "loss": 46.0,
-      "step": 13834
-    },
-    {
-      "epoch": 1.0577823651967813,
-      "grad_norm": 0.00255045248195529,
-      "learning_rate": 0.0001999994486408166,
-      "loss": 46.0,
-      "step": 13835
-    },
-    {
-      "epoch": 1.0578588221801708,
-      "grad_norm": 0.0007576187490485609,
-      "learning_rate": 0.00019999944856105114,
-      "loss": 46.0,
-      "step": 13836
-    },
-    {
-      "epoch": 1.0579352791635606,
-      "grad_norm": 0.0030599667225033045,
-      "learning_rate": 0.00019999944848127994,
-      "loss": 46.0,
-      "step": 13837
-    },
-    {
-      "epoch": 1.0580117361469503,
-      "grad_norm": 0.0007769312942400575,
-      "learning_rate": 0.000199999448401503,
-      "loss": 46.0,
-      "step": 13838
-    },
-    {
-      "epoch": 1.05808819313034,
-      "grad_norm": 0.003815098898485303,
-      "learning_rate": 0.0001999994483217202,
-      "loss": 46.0,
-      "step": 13839
-    },
-    {
-      "epoch": 1.0581646501137298,
-      "grad_norm": 0.0030538265127688646,
-      "learning_rate": 0.0001999994482419317,
-      "loss": 46.0,
-      "step": 13840
-    },
-    {
-      "epoch": 1.0582411070971194,
-      "grad_norm": 0.0007212893688119948,
-      "learning_rate": 0.0001999994481621374,
-      "loss": 46.0,
-      "step": 13841
-    },
-    {
-      "epoch": 1.0583175640805091,
-      "grad_norm": 0.0012864685850217938,
-      "learning_rate": 0.00019999944808233737,
-      "loss": 46.0,
-      "step": 13842
-    },
-    {
-      "epoch": 1.058394021063899,
-      "grad_norm": 0.0011859576916322112,
-      "learning_rate": 0.00019999944800253152,
-      "loss": 46.0,
-      "step": 13843
-    },
-    {
-      "epoch": 1.0584704780472887,
-      "grad_norm": 0.001851612119935453,
-      "learning_rate": 0.00019999944792271993,
-      "loss": 46.0,
-      "step": 13844
-    },
-    {
-      "epoch": 1.0585469350306784,
-      "grad_norm": 0.0010871737031266093,
-      "learning_rate": 0.00019999944784290257,
-      "loss": 46.0,
-      "step": 13845
-    },
-    {
-      "epoch": 1.0586233920140682,
-      "grad_norm": 0.0018016340909525752,
-      "learning_rate": 0.00019999944776307943,
-      "loss": 46.0,
-      "step": 13846
-    },
-    {
-      "epoch": 1.0586998489974577,
-      "grad_norm": 0.0023099295794963837,
-      "learning_rate": 0.00019999944768325052,
-      "loss": 46.0,
-      "step": 13847
-    },
-    {
-      "epoch": 1.0587763059808475,
-      "grad_norm": 0.0021356171928346157,
-      "learning_rate": 0.00019999944760341583,
-      "loss": 46.0,
-      "step": 13848
-    },
-    {
-      "epoch": 1.0588527629642372,
-      "grad_norm": 0.0010564925614744425,
-      "learning_rate": 0.0001999994475235754,
-      "loss": 46.0,
-      "step": 13849
-    },
-    {
-      "epoch": 1.058929219947627,
-      "grad_norm": 0.004075905773788691,
-      "learning_rate": 0.00019999944744372917,
-      "loss": 46.0,
-      "step": 13850
-    },
-    {
-      "epoch": 1.0590056769310168,
-      "grad_norm": 0.0004121228994335979,
-      "learning_rate": 0.0001999994473638772,
-      "loss": 46.0,
-      "step": 13851
-    },
-    {
-      "epoch": 1.0590821339144063,
-      "grad_norm": 0.0009071128442883492,
-      "learning_rate": 0.00019999944728401944,
-      "loss": 46.0,
-      "step": 13852
-    },
-    {
-      "epoch": 1.059158590897796,
-      "grad_norm": 0.000884173670783639,
-      "learning_rate": 0.00019999944720415594,
-      "loss": 46.0,
-      "step": 13853
-    },
-    {
-      "epoch": 1.0592350478811858,
-      "grad_norm": 0.0018070179503411055,
-      "learning_rate": 0.00019999944712428665,
-      "loss": 46.0,
-      "step": 13854
-    },
-    {
-      "epoch": 1.0593115048645756,
-      "grad_norm": 0.001158177386969328,
-      "learning_rate": 0.00019999944704441157,
-      "loss": 46.0,
-      "step": 13855
-    },
-    {
-      "epoch": 1.0593879618479654,
-      "grad_norm": 0.002257951535284519,
-      "learning_rate": 0.00019999944696453076,
-      "loss": 46.0,
-      "step": 13856
-    },
-    {
-      "epoch": 1.0594644188313551,
-      "grad_norm": 0.0010425011860206723,
-      "learning_rate": 0.00019999944688464414,
-      "loss": 46.0,
-      "step": 13857
-    },
-    {
-      "epoch": 1.0595408758147447,
-      "grad_norm": 0.0020225043408572674,
-      "learning_rate": 0.00019999944680475177,
-      "loss": 46.0,
-      "step": 13858
-    },
-    {
-      "epoch": 1.0596173327981344,
-      "grad_norm": 0.000977058312855661,
-      "learning_rate": 0.00019999944672485364,
-      "loss": 46.0,
-      "step": 13859
-    },
-    {
-      "epoch": 1.0596937897815242,
-      "grad_norm": 0.007345973514020443,
-      "learning_rate": 0.00019999944664494975,
-      "loss": 46.0,
-      "step": 13860
-    },
-    {
-      "epoch": 1.059770246764914,
-      "grad_norm": 0.0009060087613761425,
-      "learning_rate": 0.00019999944656504007,
-      "loss": 46.0,
-      "step": 13861
-    },
-    {
-      "epoch": 1.0598467037483037,
-      "grad_norm": 0.0014056694926694036,
-      "learning_rate": 0.0001999994464851246,
-      "loss": 46.0,
-      "step": 13862
-    },
-    {
-      "epoch": 1.0599231607316932,
-      "grad_norm": 0.0012392988428473473,
-      "learning_rate": 0.00019999944640520338,
-      "loss": 46.0,
-      "step": 13863
-    },
-    {
-      "epoch": 1.059999617715083,
-      "grad_norm": 0.009871658869087696,
-      "learning_rate": 0.0001999994463252764,
-      "loss": 46.0,
-      "step": 13864
-    },
-    {
-      "epoch": 1.0600760746984728,
-      "grad_norm": 0.0010715080425143242,
-      "learning_rate": 0.00019999944624534365,
-      "loss": 46.0,
-      "step": 13865
-    },
-    {
-      "epoch": 1.0601525316818625,
-      "grad_norm": 0.0026220898143947124,
-      "learning_rate": 0.00019999944616540513,
-      "loss": 46.0,
-      "step": 13866
-    },
-    {
-      "epoch": 1.0602289886652523,
-      "grad_norm": 0.000647886423394084,
-      "learning_rate": 0.00019999944608546083,
-      "loss": 46.0,
-      "step": 13867
-    },
-    {
-      "epoch": 1.060305445648642,
-      "grad_norm": 0.0003339028626214713,
-      "learning_rate": 0.00019999944600551078,
-      "loss": 46.0,
-      "step": 13868
-    },
-    {
-      "epoch": 1.0603819026320316,
-      "grad_norm": 0.0009584177169017494,
-      "learning_rate": 0.00019999944592555494,
-      "loss": 46.0,
-      "step": 13869
-    },
-    {
-      "epoch": 1.0604583596154213,
-      "grad_norm": 0.0011102084536105394,
-      "learning_rate": 0.00019999944584559335,
-      "loss": 46.0,
-      "step": 13870
-    },
-    {
-      "epoch": 1.060534816598811,
-      "grad_norm": 0.002412444679066539,
-      "learning_rate": 0.00019999944576562596,
-      "loss": 46.0,
-      "step": 13871
-    },
-    {
-      "epoch": 1.0606112735822009,
-      "grad_norm": 0.0015999983297660947,
-      "learning_rate": 0.00019999944568565282,
-      "loss": 46.0,
-      "step": 13872
-    },
-    {
-      "epoch": 1.0606877305655906,
-      "grad_norm": 0.0015892237424850464,
-      "learning_rate": 0.0001999994456056739,
-      "loss": 46.0,
-      "step": 13873
-    },
-    {
-      "epoch": 1.0607641875489802,
-      "grad_norm": 0.0005685322685167193,
-      "learning_rate": 0.00019999944552568922,
-      "loss": 46.0,
-      "step": 13874
-    },
-    {
-      "epoch": 1.06084064453237,
-      "grad_norm": 0.0012501970632001758,
-      "learning_rate": 0.00019999944544569877,
-      "loss": 46.0,
-      "step": 13875
-    },
-    {
-      "epoch": 1.0609171015157597,
-      "grad_norm": 0.0011882673716172576,
-      "learning_rate": 0.00019999944536570256,
-      "loss": 46.0,
-      "step": 13876
-    },
-    {
-      "epoch": 1.0609935584991494,
-      "grad_norm": 0.0011230031959712505,
-      "learning_rate": 0.00019999944528570059,
-      "loss": 46.0,
-      "step": 13877
-    },
-    {
-      "epoch": 1.0610700154825392,
-      "grad_norm": 0.0005707996897399426,
-      "learning_rate": 0.0001999994452056928,
-      "loss": 46.0,
-      "step": 13878
-    },
-    {
-      "epoch": 1.061146472465929,
-      "grad_norm": 0.0005501255509443581,
-      "learning_rate": 0.00019999944512567928,
-      "loss": 46.0,
-      "step": 13879
-    },
-    {
-      "epoch": 1.0612229294493185,
-      "grad_norm": 0.0024445634335279465,
-      "learning_rate": 0.00019999944504566,
-      "loss": 46.0,
-      "step": 13880
-    },
-    {
-      "epoch": 1.0612993864327083,
-      "grad_norm": 0.0014774432638660073,
-      "learning_rate": 0.00019999944496563492,
-      "loss": 46.0,
-      "step": 13881
-    },
-    {
-      "epoch": 1.061375843416098,
-      "grad_norm": 0.0016902334755286574,
-      "learning_rate": 0.0001999994448856041,
-      "loss": 46.0,
-      "step": 13882
-    },
-    {
-      "epoch": 1.0614523003994878,
-      "grad_norm": 0.003523385850712657,
-      "learning_rate": 0.00019999944480556748,
-      "loss": 46.0,
-      "step": 13883
-    },
-    {
-      "epoch": 1.0615287573828776,
-      "grad_norm": 0.0012848739279434085,
-      "learning_rate": 0.00019999944472552512,
-      "loss": 46.0,
-      "step": 13884
-    },
-    {
-      "epoch": 1.061605214366267,
-      "grad_norm": 0.0045204805210232735,
-      "learning_rate": 0.00019999944464547698,
-      "loss": 46.0,
-      "step": 13885
-    },
-    {
-      "epoch": 1.0616816713496569,
-      "grad_norm": 0.0009408780606463552,
-      "learning_rate": 0.00019999944456542304,
-      "loss": 46.0,
-      "step": 13886
-    },
-    {
-      "epoch": 1.0617581283330466,
-      "grad_norm": 0.001686860341578722,
-      "learning_rate": 0.00019999944448536336,
-      "loss": 46.0,
-      "step": 13887
-    },
-    {
-      "epoch": 1.0618345853164364,
-      "grad_norm": 0.0042866868898272514,
-      "learning_rate": 0.00019999944440529793,
-      "loss": 46.0,
-      "step": 13888
-    },
-    {
-      "epoch": 1.0619110422998261,
-      "grad_norm": 0.000781725684646517,
-      "learning_rate": 0.0001999994443252267,
-      "loss": 46.0,
-      "step": 13889
-    },
-    {
-      "epoch": 1.0619874992832157,
-      "grad_norm": 0.0012019751593470573,
-      "learning_rate": 0.0001999994442451497,
-      "loss": 46.0,
-      "step": 13890
-    },
-    {
-      "epoch": 1.0620639562666054,
-      "grad_norm": 0.0017092638881877065,
-      "learning_rate": 0.00019999944416506694,
-      "loss": 46.0,
-      "step": 13891
-    },
-    {
-      "epoch": 1.0621404132499952,
-      "grad_norm": 0.0011406762059777975,
-      "learning_rate": 0.00019999944408497842,
-      "loss": 46.0,
-      "step": 13892
-    },
-    {
-      "epoch": 1.062216870233385,
-      "grad_norm": 0.005513755138963461,
-      "learning_rate": 0.0001999994440048841,
-      "loss": 46.0,
-      "step": 13893
-    },
-    {
-      "epoch": 1.0622933272167747,
-      "grad_norm": 0.001303394790738821,
-      "learning_rate": 0.00019999944392478405,
-      "loss": 46.0,
-      "step": 13894
-    },
-    {
-      "epoch": 1.0623697842001645,
-      "grad_norm": 0.0011032327311113477,
-      "learning_rate": 0.0001999994438446782,
-      "loss": 46.0,
-      "step": 13895
-    },
-    {
-      "epoch": 1.062446241183554,
-      "grad_norm": 0.0010128116700798273,
-      "learning_rate": 0.00019999944376456662,
-      "loss": 46.0,
-      "step": 13896
-    },
-    {
-      "epoch": 1.0625226981669438,
-      "grad_norm": 0.001919205067679286,
-      "learning_rate": 0.00019999944368444923,
-      "loss": 46.0,
-      "step": 13897
-    },
-    {
-      "epoch": 1.0625991551503335,
-      "grad_norm": 0.008968381211161613,
-      "learning_rate": 0.00019999944360432606,
-      "loss": 46.0,
-      "step": 13898
-    },
-    {
-      "epoch": 1.0626756121337233,
-      "grad_norm": 0.0024009079206734896,
-      "learning_rate": 0.00019999944352419718,
-      "loss": 46.0,
-      "step": 13899
-    },
-    {
-      "epoch": 1.062752069117113,
-      "grad_norm": 0.00323579553514719,
-      "learning_rate": 0.0001999994434440625,
-      "loss": 46.0,
-      "step": 13900
-    },
-    {
-      "epoch": 1.0628285261005026,
-      "grad_norm": 0.0005370948347263038,
-      "learning_rate": 0.000199999443363922,
-      "loss": 46.0,
-      "step": 13901
-    },
-    {
-      "epoch": 1.0629049830838924,
-      "grad_norm": 0.0016549831489101052,
-      "learning_rate": 0.0001999994432837758,
-      "loss": 46.0,
-      "step": 13902
-    },
-    {
-      "epoch": 1.0629814400672821,
-      "grad_norm": 0.0028080902993679047,
-      "learning_rate": 0.0001999994432036238,
-      "loss": 46.0,
-      "step": 13903
-    },
-    {
-      "epoch": 1.0630578970506719,
-      "grad_norm": 0.002084887120872736,
-      "learning_rate": 0.00019999944312346606,
-      "loss": 46.0,
-      "step": 13904
-    },
-    {
-      "epoch": 1.0631343540340616,
-      "grad_norm": 0.0009972843108698726,
-      "learning_rate": 0.0001999994430433025,
-      "loss": 46.0,
-      "step": 13905
-    },
-    {
-      "epoch": 1.0632108110174514,
-      "grad_norm": 0.0006952547118999064,
-      "learning_rate": 0.0001999994429631332,
-      "loss": 46.0,
-      "step": 13906
-    },
-    {
-      "epoch": 1.063287268000841,
-      "grad_norm": 0.0023219604045152664,
-      "learning_rate": 0.00019999944288295814,
-      "loss": 46.0,
-      "step": 13907
-    },
-    {
-      "epoch": 1.0633637249842307,
-      "grad_norm": 0.0016072430880740285,
-      "learning_rate": 0.0001999994428027773,
-      "loss": 46.0,
-      "step": 13908
-    },
-    {
-      "epoch": 1.0634401819676205,
-      "grad_norm": 0.0006178435869514942,
-      "learning_rate": 0.00019999944272259068,
-      "loss": 46.0,
-      "step": 13909
-    },
-    {
-      "epoch": 1.0635166389510102,
-      "grad_norm": 0.0014596887631341815,
-      "learning_rate": 0.00019999944264239831,
-      "loss": 46.0,
-      "step": 13910
-    },
-    {
-      "epoch": 1.0635930959344,
-      "grad_norm": 0.0007467910181730986,
-      "learning_rate": 0.00019999944256220018,
-      "loss": 46.0,
-      "step": 13911
-    },
-    {
-      "epoch": 1.0636695529177895,
-      "grad_norm": 0.0003401925787329674,
-      "learning_rate": 0.00019999944248199624,
-      "loss": 46.0,
-      "step": 13912
-    },
-    {
-      "epoch": 1.0637460099011793,
-      "grad_norm": 0.0006558423629030585,
-      "learning_rate": 0.00019999944240178656,
-      "loss": 46.0,
-      "step": 13913
-    },
-    {
-      "epoch": 1.063822466884569,
-      "grad_norm": 0.005127561744302511,
-      "learning_rate": 0.0001999994423215711,
-      "loss": 46.0,
-      "step": 13914
-    },
-    {
-      "epoch": 1.0638989238679588,
-      "grad_norm": 0.0004356475837994367,
-      "learning_rate": 0.00019999944224134987,
-      "loss": 46.0,
-      "step": 13915
-    },
-    {
-      "epoch": 1.0639753808513486,
-      "grad_norm": 0.0006102005718275905,
-      "learning_rate": 0.00019999944216112287,
-      "loss": 46.0,
-      "step": 13916
-    },
-    {
-      "epoch": 1.0640518378347383,
-      "grad_norm": 0.010889898054301739,
-      "learning_rate": 0.00019999944208089012,
-      "loss": 46.0,
-      "step": 13917
-    },
-    {
-      "epoch": 1.0641282948181279,
-      "grad_norm": 0.005070434417575598,
-      "learning_rate": 0.0001999994420006516,
-      "loss": 46.0,
-      "step": 13918
-    },
-    {
-      "epoch": 1.0642047518015176,
-      "grad_norm": 0.010388080030679703,
-      "learning_rate": 0.0001999994419204073,
-      "loss": 46.0,
-      "step": 13919
-    },
-    {
-      "epoch": 1.0642812087849074,
-      "grad_norm": 0.0015205597737804055,
-      "learning_rate": 0.00019999944184015723,
-      "loss": 46.0,
-      "step": 13920
-    },
-    {
-      "epoch": 1.0643576657682972,
-      "grad_norm": 0.0013914147857576609,
-      "learning_rate": 0.00019999944175990139,
-      "loss": 46.0,
-      "step": 13921
-    },
-    {
-      "epoch": 1.064434122751687,
-      "grad_norm": 0.0008867255528457463,
-      "learning_rate": 0.00019999944167963974,
-      "loss": 46.0,
-      "step": 13922
-    },
-    {
-      "epoch": 1.0645105797350765,
-      "grad_norm": 0.0012207137187942863,
-      "learning_rate": 0.00019999944159937235,
-      "loss": 46.0,
-      "step": 13923
-    },
-    {
-      "epoch": 1.0645870367184662,
-      "grad_norm": 0.0053460123017430305,
-      "learning_rate": 0.00019999944151909922,
-      "loss": 46.0,
-      "step": 13924
-    },
-    {
-      "epoch": 1.064663493701856,
-      "grad_norm": 0.00437916861847043,
-      "learning_rate": 0.0001999994414388203,
-      "loss": 46.0,
-      "step": 13925
-    },
-    {
-      "epoch": 1.0647399506852457,
-      "grad_norm": 0.0012223742669448256,
-      "learning_rate": 0.00019999944135853562,
-      "loss": 46.0,
-      "step": 13926
-    },
-    {
-      "epoch": 1.0648164076686355,
-      "grad_norm": 0.0006850723293609917,
-      "learning_rate": 0.00019999944127824514,
-      "loss": 46.0,
-      "step": 13927
-    },
-    {
-      "epoch": 1.064892864652025,
-      "grad_norm": 0.002215558895841241,
-      "learning_rate": 0.00019999944119794894,
-      "loss": 46.0,
-      "step": 13928
-    },
-    {
-      "epoch": 1.0649693216354148,
-      "grad_norm": 0.002368104411289096,
-      "learning_rate": 0.00019999944111764693,
-      "loss": 46.0,
-      "step": 13929
-    },
-    {
-      "epoch": 1.0650457786188046,
-      "grad_norm": 0.0043951282277703285,
-      "learning_rate": 0.00019999944103733913,
-      "loss": 46.0,
-      "step": 13930
-    },
-    {
-      "epoch": 1.0651222356021943,
-      "grad_norm": 0.002747457241639495,
-      "learning_rate": 0.0001999994409570256,
-      "loss": 46.0,
-      "step": 13931
-    },
-    {
-      "epoch": 1.065198692585584,
-      "grad_norm": 0.0009195831953547895,
-      "learning_rate": 0.0001999994408767063,
-      "loss": 46.0,
-      "step": 13932
-    },
-    {
-      "epoch": 1.0652751495689738,
-      "grad_norm": 0.0004528456483967602,
-      "learning_rate": 0.00019999944079638121,
-      "loss": 46.0,
-      "step": 13933
-    },
-    {
-      "epoch": 1.0653516065523634,
-      "grad_norm": 0.0024599151220172644,
-      "learning_rate": 0.0001999994407160504,
-      "loss": 46.0,
-      "step": 13934
-    },
-    {
-      "epoch": 1.0654280635357531,
-      "grad_norm": 0.0012116293655708432,
-      "learning_rate": 0.00019999944063571378,
-      "loss": 46.0,
-      "step": 13935
-    },
-    {
-      "epoch": 1.065504520519143,
-      "grad_norm": 0.001292297849431634,
-      "learning_rate": 0.0001999994405553714,
-      "loss": 46.0,
-      "step": 13936
-    },
-    {
-      "epoch": 1.0655809775025327,
-      "grad_norm": 0.0027012350037693977,
-      "learning_rate": 0.00019999944047502323,
-      "loss": 46.0,
-      "step": 13937
-    },
-    {
-      "epoch": 1.0656574344859224,
-      "grad_norm": 0.002673080191016197,
-      "learning_rate": 0.00019999944039466932,
-      "loss": 46.0,
-      "step": 13938
-    },
-    {
-      "epoch": 1.065733891469312,
-      "grad_norm": 0.0005960490670986474,
-      "learning_rate": 0.0001999994403143096,
-      "loss": 46.0,
-      "step": 13939
-    },
-    {
-      "epoch": 1.0658103484527017,
-      "grad_norm": 0.000893922580871731,
-      "learning_rate": 0.00019999944023394415,
-      "loss": 46.0,
-      "step": 13940
-    },
-    {
-      "epoch": 1.0658868054360915,
-      "grad_norm": 0.0030416965018957853,
-      "learning_rate": 0.00019999944015357293,
-      "loss": 46.0,
-      "step": 13941
-    },
-    {
-      "epoch": 1.0659632624194813,
-      "grad_norm": 0.0006262491806410253,
-      "learning_rate": 0.00019999944007319592,
-      "loss": 46.0,
-      "step": 13942
-    },
-    {
-      "epoch": 1.066039719402871,
-      "grad_norm": 0.0006563802598975599,
-      "learning_rate": 0.00019999943999281317,
-      "loss": 46.0,
-      "step": 13943
-    },
-    {
-      "epoch": 1.0661161763862608,
-      "grad_norm": 0.0006599270273000002,
-      "learning_rate": 0.00019999943991242462,
-      "loss": 46.0,
-      "step": 13944
-    },
-    {
-      "epoch": 1.0661926333696503,
-      "grad_norm": 0.002420566976070404,
-      "learning_rate": 0.0001999994398320303,
-      "loss": 46.0,
-      "step": 13945
-    },
-    {
-      "epoch": 1.06626909035304,
-      "grad_norm": 0.0007779768202453852,
-      "learning_rate": 0.00019999943975163023,
-      "loss": 46.0,
-      "step": 13946
-    },
-    {
-      "epoch": 1.0663455473364298,
-      "grad_norm": 0.001933785737492144,
-      "learning_rate": 0.0001999994396712244,
-      "loss": 46.0,
-      "step": 13947
-    },
-    {
-      "epoch": 1.0664220043198196,
-      "grad_norm": 0.001142044086009264,
-      "learning_rate": 0.00019999943959081277,
-      "loss": 46.0,
-      "step": 13948
-    },
-    {
-      "epoch": 1.0664984613032094,
-      "grad_norm": 0.0010103305103257298,
-      "learning_rate": 0.00019999943951039536,
-      "loss": 46.0,
-      "step": 13949
-    },
-    {
-      "epoch": 1.066574918286599,
-      "grad_norm": 0.0038067286368459463,
-      "learning_rate": 0.00019999943942997225,
-      "loss": 46.0,
-      "step": 13950
-    },
-    {
-      "epoch": 1.0666513752699887,
-      "grad_norm": 0.0010027054231613874,
-      "learning_rate": 0.00019999943934954329,
-      "loss": 46.0,
-      "step": 13951
-    },
-    {
-      "epoch": 1.0667278322533784,
-      "grad_norm": 0.0021999781019985676,
-      "learning_rate": 0.0001999994392691086,
-      "loss": 46.0,
-      "step": 13952
-    },
-    {
-      "epoch": 1.0668042892367682,
-      "grad_norm": 0.0010803771438077092,
-      "learning_rate": 0.00019999943918866815,
-      "loss": 46.0,
-      "step": 13953
-    },
-    {
-      "epoch": 1.066880746220158,
-      "grad_norm": 0.003673393977805972,
-      "learning_rate": 0.00019999943910822192,
-      "loss": 46.0,
-      "step": 13954
-    },
-    {
-      "epoch": 1.0669572032035477,
-      "grad_norm": 0.0018245381070300937,
-      "learning_rate": 0.0001999994390277699,
-      "loss": 46.0,
-      "step": 13955
-    },
-    {
-      "epoch": 1.0670336601869372,
-      "grad_norm": 0.000570030533708632,
-      "learning_rate": 0.0001999994389473121,
-      "loss": 46.0,
-      "step": 13956
-    },
-    {
-      "epoch": 1.067110117170327,
-      "grad_norm": 0.0035906797274947166,
-      "learning_rate": 0.0001999994388668486,
-      "loss": 46.0,
-      "step": 13957
-    },
-    {
-      "epoch": 1.0671865741537168,
-      "grad_norm": 0.0031293847132474184,
-      "learning_rate": 0.00019999943878637927,
-      "loss": 46.0,
-      "step": 13958
-    },
-    {
-      "epoch": 1.0672630311371065,
-      "grad_norm": 0.012548170052468777,
-      "learning_rate": 0.00019999943870590417,
-      "loss": 46.0,
-      "step": 13959
-    },
-    {
-      "epoch": 1.0673394881204963,
-      "grad_norm": 0.0019971649162471294,
-      "learning_rate": 0.00019999943862542333,
-      "loss": 46.0,
-      "step": 13960
-    },
-    {
-      "epoch": 1.0674159451038858,
-      "grad_norm": 0.001103279646486044,
-      "learning_rate": 0.00019999943854493671,
-      "loss": 46.0,
-      "step": 13961
-    },
-    {
-      "epoch": 1.0674924020872756,
-      "grad_norm": 0.001971827819943428,
-      "learning_rate": 0.00019999943846444433,
-      "loss": 46.0,
-      "step": 13962
-    },
-    {
-      "epoch": 1.0675688590706653,
-      "grad_norm": 0.0006237890338525176,
-      "learning_rate": 0.00019999943838394616,
-      "loss": 46.0,
-      "step": 13963
-    },
-    {
-      "epoch": 1.067645316054055,
-      "grad_norm": 0.0047789085656404495,
-      "learning_rate": 0.00019999943830344225,
-      "loss": 46.0,
-      "step": 13964
-    },
-    {
-      "epoch": 1.0677217730374449,
-      "grad_norm": 0.0009100134484469891,
-      "learning_rate": 0.00019999943822293257,
-      "loss": 46.0,
-      "step": 13965
-    },
-    {
-      "epoch": 1.0677982300208346,
-      "grad_norm": 0.0016656355001032352,
-      "learning_rate": 0.0001999994381424171,
-      "loss": 46.0,
-      "step": 13966
-    },
-    {
-      "epoch": 1.0678746870042242,
-      "grad_norm": 0.002566606504842639,
-      "learning_rate": 0.00019999943806189586,
-      "loss": 46.0,
-      "step": 13967
-    },
-    {
-      "epoch": 1.067951143987614,
-      "grad_norm": 0.0009039713768288493,
-      "learning_rate": 0.00019999943798136883,
-      "loss": 46.0,
-      "step": 13968
-    },
-    {
-      "epoch": 1.0680276009710037,
-      "grad_norm": 0.00263177789747715,
-      "learning_rate": 0.00019999943790083606,
-      "loss": 46.0,
-      "step": 13969
-    },
-    {
-      "epoch": 1.0681040579543934,
-      "grad_norm": 0.00045178047730587423,
-      "learning_rate": 0.00019999943782029753,
-      "loss": 46.0,
-      "step": 13970
-    },
-    {
-      "epoch": 1.0681805149377832,
-      "grad_norm": 0.000992119312286377,
-      "learning_rate": 0.0001999994377397532,
-      "loss": 46.0,
-      "step": 13971
-    },
-    {
-      "epoch": 1.0682569719211727,
-      "grad_norm": 0.0010817080037668347,
-      "learning_rate": 0.00019999943765920312,
-      "loss": 46.0,
-      "step": 13972
-    },
-    {
-      "epoch": 1.0683334289045625,
-      "grad_norm": 0.002219643909484148,
-      "learning_rate": 0.00019999943757864727,
-      "loss": 46.0,
-      "step": 13973
-    },
-    {
-      "epoch": 1.0684098858879523,
-      "grad_norm": 0.006232650950551033,
-      "learning_rate": 0.00019999943749808566,
-      "loss": 46.0,
-      "step": 13974
-    },
-    {
-      "epoch": 1.068486342871342,
-      "grad_norm": 0.0008987795445136726,
-      "learning_rate": 0.00019999943741751824,
-      "loss": 46.0,
-      "step": 13975
-    },
-    {
-      "epoch": 1.0685627998547318,
-      "grad_norm": 0.007680040318518877,
-      "learning_rate": 0.0001999994373369451,
-      "loss": 46.0,
-      "step": 13976
-    },
-    {
-      "epoch": 1.0686392568381216,
-      "grad_norm": 0.0015218533808365464,
-      "learning_rate": 0.00019999943725636617,
-      "loss": 46.0,
-      "step": 13977
-    },
-    {
-      "epoch": 1.068715713821511,
-      "grad_norm": 0.001155595644377172,
-      "learning_rate": 0.00019999943717578146,
-      "loss": 46.0,
-      "step": 13978
-    },
-    {
-      "epoch": 1.0687921708049009,
-      "grad_norm": 0.0014901005197316408,
-      "learning_rate": 0.000199999437095191,
-      "loss": 46.0,
-      "step": 13979
-    },
-    {
-      "epoch": 1.0688686277882906,
-      "grad_norm": 0.008625698275864124,
-      "learning_rate": 0.00019999943701459475,
-      "loss": 46.0,
-      "step": 13980
-    },
-    {
-      "epoch": 1.0689450847716804,
-      "grad_norm": 0.005076452158391476,
-      "learning_rate": 0.00019999943693399278,
-      "loss": 46.0,
-      "step": 13981
-    },
-    {
-      "epoch": 1.0690215417550701,
-      "grad_norm": 0.0022425646893680096,
-      "learning_rate": 0.00019999943685338498,
-      "loss": 46.0,
-      "step": 13982
-    },
-    {
-      "epoch": 1.0690979987384597,
-      "grad_norm": 0.007618380710482597,
-      "learning_rate": 0.00019999943677277143,
-      "loss": 46.0,
-      "step": 13983
-    },
-    {
-      "epoch": 1.0691744557218494,
-      "grad_norm": 0.0003344593569636345,
-      "learning_rate": 0.0001999994366921521,
-      "loss": 46.0,
-      "step": 13984
-    },
-    {
-      "epoch": 1.0692509127052392,
-      "grad_norm": 0.0008194109541364014,
-      "learning_rate": 0.00019999943661152704,
-      "loss": 46.0,
-      "step": 13985
-    },
-    {
-      "epoch": 1.069327369688629,
-      "grad_norm": 0.0005785655812360346,
-      "learning_rate": 0.00019999943653089617,
-      "loss": 46.0,
-      "step": 13986
-    },
-    {
-      "epoch": 1.0694038266720187,
-      "grad_norm": 0.0063834101893007755,
-      "learning_rate": 0.00019999943645025953,
-      "loss": 46.0,
-      "step": 13987
-    },
-    {
-      "epoch": 1.0694802836554085,
-      "grad_norm": 0.0012502353638410568,
-      "learning_rate": 0.00019999943636961717,
-      "loss": 46.0,
-      "step": 13988
-    },
-    {
-      "epoch": 1.069556740638798,
-      "grad_norm": 0.0014219083823263645,
-      "learning_rate": 0.000199999436288969,
-      "loss": 46.0,
-      "step": 13989
-    },
-    {
-      "epoch": 1.0696331976221878,
-      "grad_norm": 0.0020772735588252544,
-      "learning_rate": 0.00019999943620831504,
-      "loss": 46.0,
-      "step": 13990
-    },
-    {
-      "epoch": 1.0697096546055775,
-      "grad_norm": 0.0006585913361050189,
-      "learning_rate": 0.00019999943612765536,
-      "loss": 46.0,
-      "step": 13991
-    },
-    {
-      "epoch": 1.0697861115889673,
-      "grad_norm": 0.0013312958180904388,
-      "learning_rate": 0.0001999994360469899,
-      "loss": 46.0,
-      "step": 13992
-    },
-    {
-      "epoch": 1.069862568572357,
-      "grad_norm": 0.003061628434807062,
-      "learning_rate": 0.00019999943596631865,
-      "loss": 46.0,
-      "step": 13993
-    },
-    {
-      "epoch": 1.0699390255557466,
-      "grad_norm": 0.0015037890989333391,
-      "learning_rate": 0.00019999943588564162,
-      "loss": 46.0,
-      "step": 13994
-    },
-    {
-      "epoch": 1.0700154825391364,
-      "grad_norm": 0.001705870614387095,
-      "learning_rate": 0.00019999943580495885,
-      "loss": 46.0,
-      "step": 13995
-    },
-    {
-      "epoch": 1.0700919395225261,
-      "grad_norm": 0.0010505091631785035,
-      "learning_rate": 0.00019999943572427032,
-      "loss": 46.0,
-      "step": 13996
-    },
-    {
-      "epoch": 1.0701683965059159,
-      "grad_norm": 0.0013376179849728942,
-      "learning_rate": 0.00019999943564357598,
-      "loss": 46.0,
-      "step": 13997
-    },
-    {
-      "epoch": 1.0702448534893056,
-      "grad_norm": 0.000491575978230685,
-      "learning_rate": 0.0001999994355628759,
-      "loss": 46.0,
-      "step": 13998
-    },
-    {
-      "epoch": 1.0703213104726954,
-      "grad_norm": 0.0008970610215328634,
-      "learning_rate": 0.00019999943548217004,
-      "loss": 46.0,
-      "step": 13999
-    },
-    {
-      "epoch": 1.070397767456085,
-      "grad_norm": 0.0007986085838638246,
-      "learning_rate": 0.0001999994354014584,
-      "loss": 46.0,
-      "step": 14000
-    },
-    {
-      "epoch": 1.0704742244394747,
-      "grad_norm": 0.005095066037029028,
-      "learning_rate": 0.000199999435320741,
-      "loss": 46.0,
-      "step": 14001
-    },
-    {
-      "epoch": 1.0705506814228645,
-      "grad_norm": 0.0006853225640952587,
-      "learning_rate": 0.00019999943524001785,
-      "loss": 46.0,
-      "step": 14002
-    },
-    {
-      "epoch": 1.0706271384062542,
-      "grad_norm": 0.0014812907902523875,
-      "learning_rate": 0.00019999943515928891,
-      "loss": 46.0,
-      "step": 14003
-    },
-    {
-      "epoch": 1.070703595389644,
-      "grad_norm": 0.0007540222723037004,
-      "learning_rate": 0.0001999994350785542,
-      "loss": 46.0,
-      "step": 14004
-    },
-    {
-      "epoch": 1.0707800523730335,
-      "grad_norm": 0.0016401633620262146,
-      "learning_rate": 0.00019999943499781372,
-      "loss": 46.0,
-      "step": 14005
-    },
-    {
-      "epoch": 1.0708565093564233,
-      "grad_norm": 0.0012590049300342798,
-      "learning_rate": 0.00019999943491706747,
-      "loss": 46.0,
-      "step": 14006
-    },
-    {
-      "epoch": 1.070932966339813,
-      "grad_norm": 0.010890955105423927,
-      "learning_rate": 0.00019999943483631547,
-      "loss": 46.0,
-      "step": 14007
-    },
-    {
-      "epoch": 1.0710094233232028,
-      "grad_norm": 0.0007714315433986485,
-      "learning_rate": 0.0001999994347555577,
-      "loss": 46.0,
-      "step": 14008
-    },
-    {
-      "epoch": 1.0710858803065926,
-      "grad_norm": 0.0007708133780397475,
-      "learning_rate": 0.00019999943467479415,
-      "loss": 46.0,
-      "step": 14009
-    },
-    {
-      "epoch": 1.0711623372899823,
-      "grad_norm": 0.0004983276594430208,
-      "learning_rate": 0.00019999943459402482,
-      "loss": 46.0,
-      "step": 14010
-    },
-    {
-      "epoch": 1.0712387942733719,
-      "grad_norm": 0.0007447259267792106,
-      "learning_rate": 0.00019999943451324973,
-      "loss": 46.0,
-      "step": 14011
-    },
-    {
-      "epoch": 1.0713152512567616,
-      "grad_norm": 0.004360907711088657,
-      "learning_rate": 0.00019999943443246886,
-      "loss": 46.0,
-      "step": 14012
-    },
-    {
-      "epoch": 1.0713917082401514,
-      "grad_norm": 0.0011546981986612082,
-      "learning_rate": 0.00019999943435168222,
-      "loss": 46.0,
-      "step": 14013
-    },
-    {
-      "epoch": 1.0714681652235412,
-      "grad_norm": 0.004667202010750771,
-      "learning_rate": 0.00019999943427088983,
-      "loss": 46.0,
-      "step": 14014
-    },
-    {
-      "epoch": 1.071544622206931,
-      "grad_norm": 0.002899933373555541,
-      "learning_rate": 0.00019999943419009167,
-      "loss": 46.0,
-      "step": 14015
-    },
-    {
-      "epoch": 1.0716210791903205,
-      "grad_norm": 0.002609699033200741,
-      "learning_rate": 0.0001999994341092877,
-      "loss": 46.0,
-      "step": 14016
-    },
-    {
-      "epoch": 1.0716975361737102,
-      "grad_norm": 0.0009058783762156963,
-      "learning_rate": 0.00019999943402847803,
-      "loss": 46.0,
-      "step": 14017
-    },
-    {
-      "epoch": 1.0717739931571,
-      "grad_norm": 0.0005103914299979806,
-      "learning_rate": 0.00019999943394766257,
-      "loss": 46.0,
-      "step": 14018
-    },
-    {
-      "epoch": 1.0718504501404897,
-      "grad_norm": 0.0012112745316699147,
-      "learning_rate": 0.0001999994338668413,
-      "loss": 46.0,
-      "step": 14019
-    },
-    {
-      "epoch": 1.0719269071238795,
-      "grad_norm": 0.0012157124001532793,
-      "learning_rate": 0.0001999994337860143,
-      "loss": 46.0,
-      "step": 14020
-    },
-    {
-      "epoch": 1.0720033641072693,
-      "grad_norm": 0.005270917434245348,
-      "learning_rate": 0.00019999943370518152,
-      "loss": 46.0,
-      "step": 14021
-    },
-    {
-      "epoch": 1.0720798210906588,
-      "grad_norm": 0.0014197340933606029,
-      "learning_rate": 0.00019999943362434294,
-      "loss": 46.0,
-      "step": 14022
-    },
-    {
-      "epoch": 1.0721562780740486,
-      "grad_norm": 0.0008277800516225398,
-      "learning_rate": 0.00019999943354349862,
-      "loss": 46.0,
-      "step": 14023
-    },
-    {
-      "epoch": 1.0722327350574383,
-      "grad_norm": 0.0012856264365836978,
-      "learning_rate": 0.00019999943346264853,
-      "loss": 46.0,
-      "step": 14024
-    },
-    {
-      "epoch": 1.072309192040828,
-      "grad_norm": 0.0010629127500578761,
-      "learning_rate": 0.00019999943338179269,
-      "loss": 46.0,
-      "step": 14025
-    },
-    {
-      "epoch": 1.0723856490242178,
-      "grad_norm": 0.0011625560000538826,
-      "learning_rate": 0.00019999943330093104,
-      "loss": 46.0,
-      "step": 14026
-    },
-    {
-      "epoch": 1.0724621060076074,
-      "grad_norm": 0.0010466283420100808,
-      "learning_rate": 0.00019999943322006366,
-      "loss": 46.0,
-      "step": 14027
-    },
-    {
-      "epoch": 1.0725385629909971,
-      "grad_norm": 0.0007803745684213936,
-      "learning_rate": 0.00019999943313919047,
-      "loss": 46.0,
-      "step": 14028
-    },
-    {
-      "epoch": 1.072615019974387,
-      "grad_norm": 0.0005933826323598623,
-      "learning_rate": 0.00019999943305831154,
-      "loss": 46.0,
-      "step": 14029
-    },
-    {
-      "epoch": 1.0726914769577767,
-      "grad_norm": 0.0007032674620859325,
-      "learning_rate": 0.0001999994329774268,
-      "loss": 46.0,
-      "step": 14030
-    },
-    {
-      "epoch": 1.0727679339411664,
-      "grad_norm": 0.0032815239392220974,
-      "learning_rate": 0.00019999943289653632,
-      "loss": 46.0,
-      "step": 14031
-    },
-    {
-      "epoch": 1.072844390924556,
-      "grad_norm": 0.0029132612980902195,
-      "learning_rate": 0.0001999994328156401,
-      "loss": 46.0,
-      "step": 14032
-    },
-    {
-      "epoch": 1.0729208479079457,
-      "grad_norm": 0.001137091894634068,
-      "learning_rate": 0.00019999943273473806,
-      "loss": 46.0,
-      "step": 14033
-    },
-    {
-      "epoch": 1.0729973048913355,
-      "grad_norm": 0.0009258515201508999,
-      "learning_rate": 0.00019999943265383026,
-      "loss": 46.0,
-      "step": 14034
-    },
-    {
-      "epoch": 1.0730737618747253,
-      "grad_norm": 0.0021736854687333107,
-      "learning_rate": 0.00019999943257291674,
-      "loss": 46.0,
-      "step": 14035
-    },
-    {
-      "epoch": 1.073150218858115,
-      "grad_norm": 0.00047871595597825944,
-      "learning_rate": 0.0001999994324919974,
-      "loss": 46.0,
-      "step": 14036
-    },
-    {
-      "epoch": 1.0732266758415048,
-      "grad_norm": 0.0007176236249506474,
-      "learning_rate": 0.00019999943241107227,
-      "loss": 46.0,
-      "step": 14037
-    },
-    {
-      "epoch": 1.0733031328248943,
-      "grad_norm": 0.0057198782451450825,
-      "learning_rate": 0.00019999943233014143,
-      "loss": 46.0,
-      "step": 14038
-    },
-    {
-      "epoch": 1.073379589808284,
-      "grad_norm": 0.00047919657663442194,
-      "learning_rate": 0.0001999994322492048,
-      "loss": 46.0,
-      "step": 14039
-    },
-    {
-      "epoch": 1.0734560467916738,
-      "grad_norm": 0.0011984321754425764,
-      "learning_rate": 0.0001999994321682624,
-      "loss": 46.0,
-      "step": 14040
-    },
-    {
-      "epoch": 1.0735325037750636,
-      "grad_norm": 0.0012020234717056155,
-      "learning_rate": 0.00019999943208731422,
-      "loss": 46.0,
-      "step": 14041
-    },
-    {
-      "epoch": 1.0736089607584534,
-      "grad_norm": 0.0012850005878135562,
-      "learning_rate": 0.00019999943200636028,
-      "loss": 46.0,
-      "step": 14042
-    },
-    {
-      "epoch": 1.073685417741843,
-      "grad_norm": 0.001039457623846829,
-      "learning_rate": 0.00019999943192540055,
-      "loss": 46.0,
-      "step": 14043
-    },
-    {
-      "epoch": 1.0737618747252327,
-      "grad_norm": 0.0015597695019096136,
-      "learning_rate": 0.00019999943184443507,
-      "loss": 46.0,
-      "step": 14044
-    },
-    {
-      "epoch": 1.0738383317086224,
-      "grad_norm": 0.0021370318718254566,
-      "learning_rate": 0.00019999943176346384,
-      "loss": 46.0,
-      "step": 14045
-    },
-    {
-      "epoch": 1.0739147886920122,
-      "grad_norm": 0.0011766912648454309,
-      "learning_rate": 0.00019999943168248682,
-      "loss": 46.0,
-      "step": 14046
-    },
-    {
-      "epoch": 1.073991245675402,
-      "grad_norm": 0.004398695193231106,
-      "learning_rate": 0.00019999943160150401,
-      "loss": 46.0,
-      "step": 14047
-    },
-    {
-      "epoch": 1.0740677026587917,
-      "grad_norm": 0.0015525203198194504,
-      "learning_rate": 0.00019999943152051547,
-      "loss": 46.0,
-      "step": 14048
-    },
-    {
-      "epoch": 1.0741441596421812,
-      "grad_norm": 0.0021211395505815744,
-      "learning_rate": 0.00019999943143952115,
-      "loss": 46.0,
-      "step": 14049
-    },
-    {
-      "epoch": 1.074220616625571,
-      "grad_norm": 0.0017788648838177323,
-      "learning_rate": 0.00019999943135852103,
-      "loss": 46.0,
-      "step": 14050
-    },
-    {
-      "epoch": 1.0742970736089608,
-      "grad_norm": 0.009273013100028038,
-      "learning_rate": 0.00019999943127751516,
-      "loss": 46.0,
-      "step": 14051
-    },
-    {
-      "epoch": 1.0743735305923505,
-      "grad_norm": 0.0010953949531540275,
-      "learning_rate": 0.00019999943119650352,
-      "loss": 46.0,
-      "step": 14052
-    },
-    {
-      "epoch": 1.0744499875757403,
-      "grad_norm": 0.00035844757803715765,
-      "learning_rate": 0.0001999994311154861,
-      "loss": 46.0,
-      "step": 14053
-    },
-    {
-      "epoch": 1.0745264445591298,
-      "grad_norm": 0.0010386542417109013,
-      "learning_rate": 0.00019999943103446294,
-      "loss": 46.0,
-      "step": 14054
-    },
-    {
-      "epoch": 1.0746029015425196,
-      "grad_norm": 0.0019077634206041694,
-      "learning_rate": 0.000199999430953434,
-      "loss": 46.0,
-      "step": 14055
-    },
-    {
-      "epoch": 1.0746793585259093,
-      "grad_norm": 0.014803217723965645,
-      "learning_rate": 0.0001999994308723993,
-      "loss": 46.0,
-      "step": 14056
-    },
-    {
-      "epoch": 1.074755815509299,
-      "grad_norm": 0.0065595852211117744,
-      "learning_rate": 0.00019999943079135882,
-      "loss": 46.0,
-      "step": 14057
-    },
-    {
-      "epoch": 1.0748322724926889,
-      "grad_norm": 0.00048247791710309684,
-      "learning_rate": 0.00019999943071031254,
-      "loss": 46.0,
-      "step": 14058
-    },
-    {
-      "epoch": 1.0749087294760784,
-      "grad_norm": 0.0014820725191384554,
-      "learning_rate": 0.00019999943062926052,
-      "loss": 46.0,
-      "step": 14059
-    },
-    {
-      "epoch": 1.0749851864594682,
-      "grad_norm": 0.0007685491000302136,
-      "learning_rate": 0.00019999943054820274,
-      "loss": 46.0,
-      "step": 14060
-    },
-    {
-      "epoch": 1.075061643442858,
-      "grad_norm": 0.0010234734509140253,
-      "learning_rate": 0.00019999943046713917,
-      "loss": 46.0,
-      "step": 14061
-    },
-    {
-      "epoch": 1.0751381004262477,
-      "grad_norm": 0.010823758319020271,
-      "learning_rate": 0.00019999943038606985,
-      "loss": 46.0,
-      "step": 14062
-    },
-    {
-      "epoch": 1.0752145574096375,
-      "grad_norm": 0.011111236177384853,
-      "learning_rate": 0.00019999943030499476,
-      "loss": 46.0,
-      "step": 14063
-    },
-    {
-      "epoch": 1.0752910143930272,
-      "grad_norm": 0.004077092744410038,
-      "learning_rate": 0.0001999994302239139,
-      "loss": 46.0,
-      "step": 14064
-    },
-    {
-      "epoch": 1.0753674713764168,
-      "grad_norm": 0.0010589115554466844,
-      "learning_rate": 0.00019999943014282722,
-      "loss": 46.0,
-      "step": 14065
-    },
-    {
-      "epoch": 1.0754439283598065,
-      "grad_norm": 0.0036376789212226868,
-      "learning_rate": 0.00019999943006173483,
-      "loss": 46.0,
-      "step": 14066
-    },
-    {
-      "epoch": 1.0755203853431963,
-      "grad_norm": 0.004602636676281691,
-      "learning_rate": 0.00019999942998063665,
-      "loss": 46.0,
-      "step": 14067
-    },
-    {
-      "epoch": 1.075596842326586,
-      "grad_norm": 0.0030424189753830433,
-      "learning_rate": 0.0001999994298995327,
-      "loss": 46.0,
-      "step": 14068
-    },
-    {
-      "epoch": 1.0756732993099758,
-      "grad_norm": 0.001877476810477674,
-      "learning_rate": 0.00019999942981842298,
-      "loss": 46.0,
-      "step": 14069
-    },
-    {
-      "epoch": 1.0757497562933653,
-      "grad_norm": 0.0006097674486227334,
-      "learning_rate": 0.0001999994297373075,
-      "loss": 46.0,
-      "step": 14070
-    },
-    {
-      "epoch": 1.075826213276755,
-      "grad_norm": 0.0006602031062357128,
-      "learning_rate": 0.00019999942965618622,
-      "loss": 46.0,
-      "step": 14071
-    },
-    {
-      "epoch": 1.0759026702601449,
-      "grad_norm": 0.0026901427190750837,
-      "learning_rate": 0.00019999942957505922,
-      "loss": 46.0,
-      "step": 14072
-    },
-    {
-      "epoch": 1.0759791272435346,
-      "grad_norm": 0.0005531682982109487,
-      "learning_rate": 0.00019999942949392642,
-      "loss": 46.0,
-      "step": 14073
-    },
-    {
-      "epoch": 1.0760555842269244,
-      "grad_norm": 0.0009872823720797896,
-      "learning_rate": 0.00019999942941278787,
-      "loss": 46.0,
-      "step": 14074
-    },
-    {
-      "epoch": 1.0761320412103141,
-      "grad_norm": 0.0010065939277410507,
-      "learning_rate": 0.00019999942933164353,
-      "loss": 46.0,
-      "step": 14075
-    },
-    {
-      "epoch": 1.0762084981937037,
-      "grad_norm": 0.004798435606062412,
-      "learning_rate": 0.0001999994292504934,
-      "loss": 46.0,
-      "step": 14076
-    },
-    {
-      "epoch": 1.0762849551770934,
-      "grad_norm": 0.0008685302454978228,
-      "learning_rate": 0.00019999942916933754,
-      "loss": 46.0,
-      "step": 14077
-    },
-    {
-      "epoch": 1.0763614121604832,
-      "grad_norm": 0.0015082533936947584,
-      "learning_rate": 0.0001999994290881759,
-      "loss": 46.0,
-      "step": 14078
-    },
-    {
-      "epoch": 1.076437869143873,
-      "grad_norm": 0.006136310752481222,
-      "learning_rate": 0.0001999994290070085,
-      "loss": 46.0,
-      "step": 14079
-    },
-    {
-      "epoch": 1.0765143261272627,
-      "grad_norm": 0.0010586829157546163,
-      "learning_rate": 0.0001999994289258353,
-      "loss": 46.0,
-      "step": 14080
-    },
-    {
-      "epoch": 1.0765907831106523,
-      "grad_norm": 0.002820434281602502,
-      "learning_rate": 0.00019999942884465637,
-      "loss": 46.0,
-      "step": 14081
-    },
-    {
-      "epoch": 1.076667240094042,
-      "grad_norm": 0.0017263127956539392,
-      "learning_rate": 0.00019999942876347163,
-      "loss": 46.0,
-      "step": 14082
-    },
-    {
-      "epoch": 1.0767436970774318,
-      "grad_norm": 0.0010990806622430682,
-      "learning_rate": 0.00019999942868228115,
-      "loss": 46.0,
-      "step": 14083
-    },
-    {
-      "epoch": 1.0768201540608215,
-      "grad_norm": 0.0011168146738782525,
-      "learning_rate": 0.0001999994286010849,
-      "loss": 46.0,
-      "step": 14084
-    },
-    {
-      "epoch": 1.0768966110442113,
-      "grad_norm": 0.005691499914973974,
-      "learning_rate": 0.00019999942851988287,
-      "loss": 46.0,
-      "step": 14085
-    },
-    {
-      "epoch": 1.076973068027601,
-      "grad_norm": 0.0013034566072747111,
-      "learning_rate": 0.00019999942843867507,
-      "loss": 46.0,
-      "step": 14086
-    },
-    {
-      "epoch": 1.0770495250109906,
-      "grad_norm": 0.001498423283919692,
-      "learning_rate": 0.0001999994283574615,
-      "loss": 46.0,
-      "step": 14087
-    },
-    {
-      "epoch": 1.0771259819943804,
-      "grad_norm": 0.002232861705124378,
-      "learning_rate": 0.00019999942827624216,
-      "loss": 46.0,
-      "step": 14088
-    },
-    {
-      "epoch": 1.0772024389777701,
-      "grad_norm": 0.006442929618060589,
-      "learning_rate": 0.00019999942819501704,
-      "loss": 46.0,
-      "step": 14089
-    },
-    {
-      "epoch": 1.0772788959611599,
-      "grad_norm": 0.0007409933023154736,
-      "learning_rate": 0.00019999942811378617,
-      "loss": 46.0,
-      "step": 14090
-    },
-    {
-      "epoch": 1.0773553529445496,
-      "grad_norm": 0.0011904374696314335,
-      "learning_rate": 0.00019999942803254953,
-      "loss": 46.0,
-      "step": 14091
-    },
-    {
-      "epoch": 1.0774318099279392,
-      "grad_norm": 0.012527956627309322,
-      "learning_rate": 0.00019999942795130712,
-      "loss": 46.0,
-      "step": 14092
-    },
-    {
-      "epoch": 1.077508266911329,
-      "grad_norm": 0.0034234528429806232,
-      "learning_rate": 0.00019999942787005893,
-      "loss": 46.0,
-      "step": 14093
-    },
-    {
-      "epoch": 1.0775847238947187,
-      "grad_norm": 0.002114152768626809,
-      "learning_rate": 0.000199999427788805,
-      "loss": 46.0,
-      "step": 14094
-    },
-    {
-      "epoch": 1.0776611808781085,
-      "grad_norm": 0.004000728949904442,
-      "learning_rate": 0.00019999942770754526,
-      "loss": 46.0,
-      "step": 14095
-    },
-    {
-      "epoch": 1.0777376378614982,
-      "grad_norm": 0.004133324604481459,
-      "learning_rate": 0.00019999942762627976,
-      "loss": 46.0,
-      "step": 14096
-    },
-    {
-      "epoch": 1.077814094844888,
-      "grad_norm": 0.0016991019947454333,
-      "learning_rate": 0.0001999994275450085,
-      "loss": 46.0,
-      "step": 14097
-    },
-    {
-      "epoch": 1.0778905518282775,
-      "grad_norm": 0.0006427473272196949,
-      "learning_rate": 0.00019999942746373148,
-      "loss": 46.0,
-      "step": 14098
-    },
-    {
-      "epoch": 1.0779670088116673,
-      "grad_norm": 0.0022385362535715103,
-      "learning_rate": 0.00019999942738244868,
-      "loss": 46.0,
-      "step": 14099
-    },
-    {
-      "epoch": 1.078043465795057,
-      "grad_norm": 0.00048014981439337134,
-      "learning_rate": 0.0001999994273011601,
-      "loss": 46.0,
-      "step": 14100
-    },
-    {
-      "epoch": 1.0781199227784468,
-      "grad_norm": 0.0008626414928585291,
-      "learning_rate": 0.0001999994272198658,
-      "loss": 46.0,
-      "step": 14101
-    },
-    {
-      "epoch": 1.0781963797618366,
-      "grad_norm": 0.0013959245989099145,
-      "learning_rate": 0.00019999942713856567,
-      "loss": 46.0,
-      "step": 14102
-    },
-    {
-      "epoch": 1.0782728367452261,
-      "grad_norm": 0.0007344413315877318,
-      "learning_rate": 0.00019999942705725977,
-      "loss": 46.0,
-      "step": 14103
-    },
-    {
-      "epoch": 1.0783492937286159,
-      "grad_norm": 0.0024672059807926416,
-      "learning_rate": 0.00019999942697594816,
-      "loss": 46.0,
-      "step": 14104
-    },
-    {
-      "epoch": 1.0784257507120056,
-      "grad_norm": 0.0009161752532236278,
-      "learning_rate": 0.00019999942689463072,
-      "loss": 46.0,
-      "step": 14105
-    },
-    {
-      "epoch": 1.0785022076953954,
-      "grad_norm": 0.0011544388253241777,
-      "learning_rate": 0.00019999942681330754,
-      "loss": 46.0,
-      "step": 14106
-    },
-    {
-      "epoch": 1.0785786646787852,
-      "grad_norm": 0.0035547870211303234,
-      "learning_rate": 0.0001999994267319786,
-      "loss": 46.0,
-      "step": 14107
-    },
-    {
-      "epoch": 1.078655121662175,
-      "grad_norm": 0.003957994747906923,
-      "learning_rate": 0.00019999942665064387,
-      "loss": 46.0,
-      "step": 14108
-    },
-    {
-      "epoch": 1.0787315786455645,
-      "grad_norm": 0.0004292784142307937,
-      "learning_rate": 0.00019999942656930337,
-      "loss": 46.0,
-      "step": 14109
-    },
-    {
-      "epoch": 1.0788080356289542,
-      "grad_norm": 0.0008417435456067324,
-      "learning_rate": 0.0001999994264879571,
-      "loss": 46.0,
-      "step": 14110
-    },
-    {
-      "epoch": 1.078884492612344,
-      "grad_norm": 0.001191833638586104,
-      "learning_rate": 0.00019999942640660506,
-      "loss": 46.0,
-      "step": 14111
-    },
-    {
-      "epoch": 1.0789609495957337,
-      "grad_norm": 0.0007495269528590143,
-      "learning_rate": 0.0001999994263252473,
-      "loss": 46.0,
-      "step": 14112
-    },
-    {
-      "epoch": 1.0790374065791235,
-      "grad_norm": 0.0006646562251262367,
-      "learning_rate": 0.00019999942624388372,
-      "loss": 46.0,
-      "step": 14113
-    },
-    {
-      "epoch": 1.079113863562513,
-      "grad_norm": 0.0009241909720003605,
-      "learning_rate": 0.00019999942616251434,
-      "loss": 46.0,
-      "step": 14114
-    },
-    {
-      "epoch": 1.0791903205459028,
-      "grad_norm": 0.0022906421218067408,
-      "learning_rate": 0.00019999942608113925,
-      "loss": 46.0,
-      "step": 14115
-    },
-    {
-      "epoch": 1.0792667775292926,
-      "grad_norm": 0.0015027154004201293,
-      "learning_rate": 0.00019999942599975836,
-      "loss": 46.0,
-      "step": 14116
-    },
-    {
-      "epoch": 1.0793432345126823,
-      "grad_norm": 0.0005942407879047096,
-      "learning_rate": 0.00019999942591837172,
-      "loss": 46.0,
-      "step": 14117
-    },
-    {
-      "epoch": 1.079419691496072,
-      "grad_norm": 0.0018964800983667374,
-      "learning_rate": 0.0001999994258369793,
-      "loss": 46.0,
-      "step": 14118
-    },
-    {
-      "epoch": 1.0794961484794618,
-      "grad_norm": 0.0037390545476228,
-      "learning_rate": 0.0001999994257555811,
-      "loss": 46.0,
-      "step": 14119
-    },
-    {
-      "epoch": 1.0795726054628514,
-      "grad_norm": 0.0034300799015909433,
-      "learning_rate": 0.00019999942567417716,
-      "loss": 46.0,
-      "step": 14120
-    },
-    {
-      "epoch": 1.0796490624462411,
-      "grad_norm": 0.0009461540030315518,
-      "learning_rate": 0.00019999942559276743,
-      "loss": 46.0,
-      "step": 14121
-    },
-    {
-      "epoch": 1.079725519429631,
-      "grad_norm": 0.0010551678715273738,
-      "learning_rate": 0.00019999942551135192,
-      "loss": 46.0,
-      "step": 14122
-    },
-    {
-      "epoch": 1.0798019764130207,
-      "grad_norm": 0.0015043900348246098,
-      "learning_rate": 0.00019999942542993067,
-      "loss": 46.0,
-      "step": 14123
-    },
-    {
-      "epoch": 1.0798784333964104,
-      "grad_norm": 0.011879284866154194,
-      "learning_rate": 0.00019999942534850362,
-      "loss": 46.0,
-      "step": 14124
-    },
-    {
-      "epoch": 1.0799548903798,
-      "grad_norm": 0.0015790078323334455,
-      "learning_rate": 0.00019999942526707082,
-      "loss": 46.0,
-      "step": 14125
-    },
-    {
-      "epoch": 1.0800313473631897,
-      "grad_norm": 0.0008118515834212303,
-      "learning_rate": 0.00019999942518563225,
-      "loss": 46.0,
-      "step": 14126
-    },
-    {
-      "epoch": 1.0801078043465795,
-      "grad_norm": 0.0010812345426529646,
-      "learning_rate": 0.0001999994251041879,
-      "loss": 46.0,
-      "step": 14127
-    },
-    {
-      "epoch": 1.0801842613299693,
-      "grad_norm": 0.0014956349041312933,
-      "learning_rate": 0.0001999994250227378,
-      "loss": 46.0,
-      "step": 14128
-    },
-    {
-      "epoch": 1.080260718313359,
-      "grad_norm": 0.0059073143638670444,
-      "learning_rate": 0.0001999994249412819,
-      "loss": 46.0,
-      "step": 14129
-    },
-    {
-      "epoch": 1.0803371752967488,
-      "grad_norm": 0.006054010707885027,
-      "learning_rate": 0.00019999942485982025,
-      "loss": 46.0,
-      "step": 14130
-    },
-    {
-      "epoch": 1.0804136322801383,
-      "grad_norm": 0.001565291895531118,
-      "learning_rate": 0.00019999942477835284,
-      "loss": 46.0,
-      "step": 14131
-    },
-    {
-      "epoch": 1.080490089263528,
-      "grad_norm": 0.001726433401927352,
-      "learning_rate": 0.00019999942469687965,
-      "loss": 46.0,
-      "step": 14132
-    },
-    {
-      "epoch": 1.0805665462469178,
-      "grad_norm": 0.00841713696718216,
-      "learning_rate": 0.0001999994246154007,
-      "loss": 46.0,
-      "step": 14133
-    },
-    {
-      "epoch": 1.0806430032303076,
-      "grad_norm": 0.0015397778479382396,
-      "learning_rate": 0.00019999942453391596,
-      "loss": 46.0,
-      "step": 14134
-    },
-    {
-      "epoch": 1.0807194602136974,
-      "grad_norm": 0.0026642829179763794,
-      "learning_rate": 0.00019999942445242546,
-      "loss": 46.0,
-      "step": 14135
-    },
-    {
-      "epoch": 1.080795917197087,
-      "grad_norm": 0.0017836529295891523,
-      "learning_rate": 0.0001999994243709292,
-      "loss": 46.0,
-      "step": 14136
-    },
-    {
-      "epoch": 1.0808723741804767,
-      "grad_norm": 0.001951273763552308,
-      "learning_rate": 0.00019999942428942715,
-      "loss": 46.0,
-      "step": 14137
-    },
-    {
-      "epoch": 1.0809488311638664,
-      "grad_norm": 0.0010551137384027243,
-      "learning_rate": 0.00019999942420791933,
-      "loss": 46.0,
-      "step": 14138
-    },
-    {
-      "epoch": 1.0810252881472562,
-      "grad_norm": 0.00663244491443038,
-      "learning_rate": 0.00019999942412640578,
-      "loss": 46.0,
-      "step": 14139
-    },
-    {
-      "epoch": 1.081101745130646,
-      "grad_norm": 0.002032683929428458,
-      "learning_rate": 0.0001999994240448864,
-      "loss": 46.0,
-      "step": 14140
-    },
-    {
-      "epoch": 1.0811782021140357,
-      "grad_norm": 0.0005868744920007885,
-      "learning_rate": 0.00019999942396336132,
-      "loss": 46.0,
-      "step": 14141
-    },
-    {
-      "epoch": 1.0812546590974252,
-      "grad_norm": 0.0022092147264629602,
-      "learning_rate": 0.00019999942388183043,
-      "loss": 46.0,
-      "step": 14142
-    },
-    {
-      "epoch": 1.081331116080815,
-      "grad_norm": 0.003351550316438079,
-      "learning_rate": 0.00019999942380029376,
-      "loss": 46.0,
-      "step": 14143
-    },
-    {
-      "epoch": 1.0814075730642048,
-      "grad_norm": 0.0007190115284174681,
-      "learning_rate": 0.00019999942371875132,
-      "loss": 46.0,
-      "step": 14144
-    },
-    {
-      "epoch": 1.0814840300475945,
-      "grad_norm": 0.0009946164209395647,
-      "learning_rate": 0.00019999942363720314,
-      "loss": 46.0,
-      "step": 14145
-    },
-    {
-      "epoch": 1.0815604870309843,
-      "grad_norm": 0.0005807209527119994,
-      "learning_rate": 0.00019999942355564918,
-      "loss": 46.0,
-      "step": 14146
-    },
-    {
-      "epoch": 1.0816369440143738,
-      "grad_norm": 0.004871238023042679,
-      "learning_rate": 0.00019999942347408945,
-      "loss": 46.0,
-      "step": 14147
-    },
-    {
-      "epoch": 1.0817134009977636,
-      "grad_norm": 0.000403507990995422,
-      "learning_rate": 0.00019999942339252394,
-      "loss": 46.0,
-      "step": 14148
-    },
-    {
-      "epoch": 1.0817898579811533,
-      "grad_norm": 0.001539861666969955,
-      "learning_rate": 0.0001999994233109527,
-      "loss": 46.0,
-      "step": 14149
-    },
-    {
-      "epoch": 1.081866314964543,
-      "grad_norm": 0.0012488304637372494,
-      "learning_rate": 0.00019999942322937564,
-      "loss": 46.0,
-      "step": 14150
-    },
-    {
-      "epoch": 1.0819427719479329,
-      "grad_norm": 0.0035214044619351625,
-      "learning_rate": 0.00019999942314779284,
-      "loss": 46.0,
-      "step": 14151
-    },
-    {
-      "epoch": 1.0820192289313226,
-      "grad_norm": 0.004506431985646486,
-      "learning_rate": 0.00019999942306620424,
-      "loss": 46.0,
-      "step": 14152
-    },
-    {
-      "epoch": 1.0820956859147122,
-      "grad_norm": 0.014677898958325386,
-      "learning_rate": 0.00019999942298460992,
-      "loss": 46.0,
-      "step": 14153
-    },
-    {
-      "epoch": 1.082172142898102,
-      "grad_norm": 0.0026089835446327925,
-      "learning_rate": 0.00019999942290300978,
-      "loss": 46.0,
-      "step": 14154
-    },
-    {
-      "epoch": 1.0822485998814917,
-      "grad_norm": 0.0006066914065741003,
-      "learning_rate": 0.0001999994228214039,
-      "loss": 46.0,
-      "step": 14155
-    },
-    {
-      "epoch": 1.0823250568648815,
-      "grad_norm": 0.0045068091712892056,
-      "learning_rate": 0.00019999942273979225,
-      "loss": 46.0,
-      "step": 14156
-    },
-    {
-      "epoch": 1.0824015138482712,
-      "grad_norm": 0.0010878955945372581,
-      "learning_rate": 0.0001999994226581748,
-      "loss": 46.0,
-      "step": 14157
-    },
-    {
-      "epoch": 1.0824779708316608,
-      "grad_norm": 0.001035849447362125,
-      "learning_rate": 0.00019999942257655163,
-      "loss": 46.0,
-      "step": 14158
-    },
-    {
-      "epoch": 1.0825544278150505,
-      "grad_norm": 0.0008703760686330497,
-      "learning_rate": 0.00019999942249492264,
-      "loss": 46.0,
-      "step": 14159
-    },
-    {
-      "epoch": 1.0826308847984403,
-      "grad_norm": 0.004856419749557972,
-      "learning_rate": 0.0001999994224132879,
-      "loss": 46.0,
-      "step": 14160
-    },
-    {
-      "epoch": 1.08270734178183,
-      "grad_norm": 0.0011521837441250682,
-      "learning_rate": 0.0001999994223316474,
-      "loss": 46.0,
-      "step": 14161
-    },
-    {
-      "epoch": 1.0827837987652198,
-      "grad_norm": 0.00044389814138412476,
-      "learning_rate": 0.00019999942225000115,
-      "loss": 46.0,
-      "step": 14162
-    },
-    {
-      "epoch": 1.0828602557486093,
-      "grad_norm": 0.0030648878309875727,
-      "learning_rate": 0.0001999994221683491,
-      "loss": 46.0,
-      "step": 14163
-    },
-    {
-      "epoch": 1.082936712731999,
-      "grad_norm": 0.0009003924205899239,
-      "learning_rate": 0.0001999994220866913,
-      "loss": 46.0,
-      "step": 14164
-    },
-    {
-      "epoch": 1.0830131697153889,
-      "grad_norm": 0.006464676465839148,
-      "learning_rate": 0.0001999994220050277,
-      "loss": 46.0,
-      "step": 14165
-    },
-    {
-      "epoch": 1.0830896266987786,
-      "grad_norm": 0.003375225467607379,
-      "learning_rate": 0.00019999942192335833,
-      "loss": 46.0,
-      "step": 14166
-    },
-    {
-      "epoch": 1.0831660836821684,
-      "grad_norm": 0.0016525730025023222,
-      "learning_rate": 0.00019999942184168324,
-      "loss": 46.0,
-      "step": 14167
-    },
-    {
-      "epoch": 1.0832425406655581,
-      "grad_norm": 0.0007938342751003802,
-      "learning_rate": 0.00019999942176000235,
-      "loss": 46.0,
-      "step": 14168
-    },
-    {
-      "epoch": 1.0833189976489477,
-      "grad_norm": 0.0006924756453372538,
-      "learning_rate": 0.00019999942167831569,
-      "loss": 46.0,
-      "step": 14169
-    },
-    {
-      "epoch": 1.0833954546323374,
-      "grad_norm": 0.002503820229321718,
-      "learning_rate": 0.00019999942159662325,
-      "loss": 46.0,
-      "step": 14170
-    },
-    {
-      "epoch": 1.0834719116157272,
-      "grad_norm": 0.0006507441867142916,
-      "learning_rate": 0.00019999942151492506,
-      "loss": 46.0,
-      "step": 14171
-    },
-    {
-      "epoch": 1.083548368599117,
-      "grad_norm": 0.013293273746967316,
-      "learning_rate": 0.00019999942143322108,
-      "loss": 46.0,
-      "step": 14172
-    },
-    {
-      "epoch": 1.0836248255825067,
-      "grad_norm": 0.009504067711532116,
-      "learning_rate": 0.00019999942135151135,
-      "loss": 46.0,
-      "step": 14173
-    },
-    {
-      "epoch": 1.0837012825658963,
-      "grad_norm": 0.019417013972997665,
-      "learning_rate": 0.00019999942126979584,
-      "loss": 46.0,
-      "step": 14174
-    },
-    {
-      "epoch": 1.083777739549286,
-      "grad_norm": 0.0016870127292349935,
-      "learning_rate": 0.00019999942118807456,
-      "loss": 46.0,
-      "step": 14175
-    },
-    {
-      "epoch": 1.0838541965326758,
-      "grad_norm": 0.0017147115431725979,
-      "learning_rate": 0.00019999942110634754,
-      "loss": 46.0,
-      "step": 14176
-    },
-    {
-      "epoch": 1.0839306535160655,
-      "grad_norm": 0.003814919851720333,
-      "learning_rate": 0.00019999942102461472,
-      "loss": 46.0,
-      "step": 14177
-    },
-    {
-      "epoch": 1.0840071104994553,
-      "grad_norm": 0.002881775377318263,
-      "learning_rate": 0.00019999942094287612,
-      "loss": 46.0,
-      "step": 14178
-    },
-    {
-      "epoch": 1.084083567482845,
-      "grad_norm": 0.002160491421818733,
-      "learning_rate": 0.00019999942086113177,
-      "loss": 46.0,
-      "step": 14179
-    },
-    {
-      "epoch": 1.0841600244662346,
-      "grad_norm": 0.015525463968515396,
-      "learning_rate": 0.00019999942077938166,
-      "loss": 46.0,
-      "step": 14180
-    },
-    {
-      "epoch": 1.0842364814496244,
-      "grad_norm": 0.0011640343582257628,
-      "learning_rate": 0.00019999942069762576,
-      "loss": 46.0,
-      "step": 14181
-    },
-    {
-      "epoch": 1.0843129384330141,
-      "grad_norm": 0.001267034444026649,
-      "learning_rate": 0.0001999994206158641,
-      "loss": 46.0,
-      "step": 14182
-    },
-    {
-      "epoch": 1.084389395416404,
-      "grad_norm": 0.0023554738145321608,
-      "learning_rate": 0.00019999942053409666,
-      "loss": 46.0,
-      "step": 14183
-    },
-    {
-      "epoch": 1.0844658523997937,
-      "grad_norm": 0.00042725124512799084,
-      "learning_rate": 0.00019999942045232345,
-      "loss": 46.0,
-      "step": 14184
-    },
-    {
-      "epoch": 1.0845423093831832,
-      "grad_norm": 0.0007296939729712903,
-      "learning_rate": 0.00019999942037054452,
-      "loss": 46.0,
-      "step": 14185
-    },
-    {
-      "epoch": 1.084618766366573,
-      "grad_norm": 0.0006084716296754777,
-      "learning_rate": 0.00019999942028875976,
-      "loss": 46.0,
-      "step": 14186
-    },
-    {
-      "epoch": 1.0846952233499627,
-      "grad_norm": 0.005822858773171902,
-      "learning_rate": 0.00019999942020696926,
-      "loss": 46.0,
-      "step": 14187
-    },
-    {
-      "epoch": 1.0847716803333525,
-      "grad_norm": 0.0015475107356905937,
-      "learning_rate": 0.00019999942012517295,
-      "loss": 46.0,
-      "step": 14188
-    },
-    {
-      "epoch": 1.0848481373167422,
-      "grad_norm": 0.0009447706979699433,
-      "learning_rate": 0.00019999942004337093,
-      "loss": 46.0,
-      "step": 14189
-    },
-    {
-      "epoch": 1.0849245943001318,
-      "grad_norm": 0.003099769353866577,
-      "learning_rate": 0.0001999994199615631,
-      "loss": 46.0,
-      "step": 14190
-    },
-    {
-      "epoch": 1.0850010512835215,
-      "grad_norm": 0.0008022661786526442,
-      "learning_rate": 0.0001999994198797495,
-      "loss": 46.0,
-      "step": 14191
-    },
-    {
-      "epoch": 1.0850775082669113,
-      "grad_norm": 0.0009172812569886446,
-      "learning_rate": 0.00019999941979793016,
-      "loss": 46.0,
-      "step": 14192
-    },
-    {
-      "epoch": 1.085153965250301,
-      "grad_norm": 0.0014615554828196764,
-      "learning_rate": 0.00019999941971610502,
-      "loss": 46.0,
-      "step": 14193
-    },
-    {
-      "epoch": 1.0852304222336908,
-      "grad_norm": 0.0009250359726138413,
-      "learning_rate": 0.00019999941963427413,
-      "loss": 46.0,
-      "step": 14194
-    },
-    {
-      "epoch": 1.0853068792170806,
-      "grad_norm": 0.0008587997872382402,
-      "learning_rate": 0.00019999941955243746,
-      "loss": 46.0,
-      "step": 14195
-    },
-    {
-      "epoch": 1.0853833362004701,
-      "grad_norm": 0.0009336332441307604,
-      "learning_rate": 0.00019999941947059505,
-      "loss": 46.0,
-      "step": 14196
-    },
-    {
-      "epoch": 1.0854597931838599,
-      "grad_norm": 0.0009571525151841342,
-      "learning_rate": 0.00019999941938874684,
-      "loss": 46.0,
-      "step": 14197
-    },
-    {
-      "epoch": 1.0855362501672496,
-      "grad_norm": 0.0025165551342070103,
-      "learning_rate": 0.00019999941930689286,
-      "loss": 46.0,
-      "step": 14198
-    },
-    {
-      "epoch": 1.0856127071506394,
-      "grad_norm": 0.0033422091510146856,
-      "learning_rate": 0.00019999941922503313,
-      "loss": 46.0,
-      "step": 14199
-    },
-    {
-      "epoch": 1.0856891641340292,
-      "grad_norm": 0.0013745806645601988,
-      "learning_rate": 0.0001999994191431676,
-      "loss": 46.0,
-      "step": 14200
-    },
-    {
-      "epoch": 1.0857656211174187,
-      "grad_norm": 0.00043467935756780207,
-      "learning_rate": 0.00019999941906129635,
-      "loss": 46.0,
-      "step": 14201
-    },
-    {
-      "epoch": 1.0858420781008085,
-      "grad_norm": 0.0010825492208823562,
-      "learning_rate": 0.0001999994189794193,
-      "loss": 46.0,
-      "step": 14202
-    },
-    {
-      "epoch": 1.0859185350841982,
-      "grad_norm": 0.0031072243582457304,
-      "learning_rate": 0.00019999941889753647,
-      "loss": 46.0,
-      "step": 14203
-    },
-    {
-      "epoch": 1.085994992067588,
-      "grad_norm": 0.0006837751134298742,
-      "learning_rate": 0.00019999941881564787,
-      "loss": 46.0,
-      "step": 14204
-    },
-    {
-      "epoch": 1.0860714490509777,
-      "grad_norm": 0.0008400525548495352,
-      "learning_rate": 0.00019999941873375353,
-      "loss": 46.0,
-      "step": 14205
-    },
-    {
-      "epoch": 1.0861479060343675,
-      "grad_norm": 0.003376114647835493,
-      "learning_rate": 0.00019999941865185339,
-      "loss": 46.0,
-      "step": 14206
-    },
-    {
-      "epoch": 1.086224363017757,
-      "grad_norm": 0.0009143169154413044,
-      "learning_rate": 0.0001999994185699475,
-      "loss": 46.0,
-      "step": 14207
-    },
-    {
-      "epoch": 1.0863008200011468,
-      "grad_norm": 0.001540070166811347,
-      "learning_rate": 0.00019999941848803583,
-      "loss": 46.0,
-      "step": 14208
-    },
-    {
-      "epoch": 1.0863772769845366,
-      "grad_norm": 0.000881193729583174,
-      "learning_rate": 0.00019999941840611837,
-      "loss": 46.0,
-      "step": 14209
-    },
-    {
-      "epoch": 1.0864537339679263,
-      "grad_norm": 0.0014265382196754217,
-      "learning_rate": 0.0001999994183241952,
-      "loss": 46.0,
-      "step": 14210
-    },
-    {
-      "epoch": 1.086530190951316,
-      "grad_norm": 0.0006530576501972973,
-      "learning_rate": 0.0001999994182422662,
-      "loss": 46.0,
-      "step": 14211
-    },
-    {
-      "epoch": 1.0866066479347056,
-      "grad_norm": 0.0011609486537054181,
-      "learning_rate": 0.00019999941816033147,
-      "loss": 46.0,
-      "step": 14212
-    },
-    {
-      "epoch": 1.0866831049180954,
-      "grad_norm": 0.00048583035822957754,
-      "learning_rate": 0.00019999941807839097,
-      "loss": 46.0,
-      "step": 14213
-    },
-    {
-      "epoch": 1.0867595619014851,
-      "grad_norm": 0.0005649944650940597,
-      "learning_rate": 0.00019999941799644466,
-      "loss": 46.0,
-      "step": 14214
-    },
-    {
-      "epoch": 1.086836018884875,
-      "grad_norm": 0.0011280899634584785,
-      "learning_rate": 0.0001999994179144926,
-      "loss": 46.0,
-      "step": 14215
-    },
-    {
-      "epoch": 1.0869124758682647,
-      "grad_norm": 0.005256341770291328,
-      "learning_rate": 0.0001999994178325348,
-      "loss": 46.0,
-      "step": 14216
-    },
-    {
-      "epoch": 1.0869889328516544,
-      "grad_norm": 0.0026654910761862993,
-      "learning_rate": 0.00019999941775057122,
-      "loss": 46.0,
-      "step": 14217
-    },
-    {
-      "epoch": 1.087065389835044,
-      "grad_norm": 0.0005470442119985819,
-      "learning_rate": 0.00019999941766860185,
-      "loss": 46.0,
-      "step": 14218
-    },
-    {
-      "epoch": 1.0871418468184337,
-      "grad_norm": 0.0008824032265692949,
-      "learning_rate": 0.0001999994175866267,
-      "loss": 46.0,
-      "step": 14219
-    },
-    {
-      "epoch": 1.0872183038018235,
-      "grad_norm": 0.001202802057377994,
-      "learning_rate": 0.0001999994175046458,
-      "loss": 46.0,
-      "step": 14220
-    },
-    {
-      "epoch": 1.0872947607852133,
-      "grad_norm": 0.0006799806142225862,
-      "learning_rate": 0.00019999941742265915,
-      "loss": 46.0,
-      "step": 14221
-    },
-    {
-      "epoch": 1.087371217768603,
-      "grad_norm": 0.006414244882762432,
-      "learning_rate": 0.00019999941734066668,
-      "loss": 46.0,
-      "step": 14222
-    },
-    {
-      "epoch": 1.0874476747519926,
-      "grad_norm": 0.0012257150374352932,
-      "learning_rate": 0.0001999994172586685,
-      "loss": 46.0,
-      "step": 14223
-    },
-    {
-      "epoch": 1.0875241317353823,
-      "grad_norm": 0.0013132316526025534,
-      "learning_rate": 0.00019999941717666452,
-      "loss": 46.0,
-      "step": 14224
-    },
-    {
-      "epoch": 1.087600588718772,
-      "grad_norm": 0.0031813608948141336,
-      "learning_rate": 0.00019999941709465476,
-      "loss": 46.0,
-      "step": 14225
-    },
-    {
-      "epoch": 1.0876770457021618,
-      "grad_norm": 0.0008511891937814653,
-      "learning_rate": 0.00019999941701263926,
-      "loss": 46.0,
-      "step": 14226
-    },
-    {
-      "epoch": 1.0877535026855516,
-      "grad_norm": 0.001913353567942977,
-      "learning_rate": 0.00019999941693061795,
-      "loss": 46.0,
-      "step": 14227
-    },
-    {
-      "epoch": 1.0878299596689414,
-      "grad_norm": 0.0020528610330075026,
-      "learning_rate": 0.0001999994168485909,
-      "loss": 46.0,
-      "step": 14228
-    },
-    {
-      "epoch": 1.087906416652331,
-      "grad_norm": 0.011435126885771751,
-      "learning_rate": 0.00019999941676655808,
-      "loss": 46.0,
-      "step": 14229
-    },
-    {
-      "epoch": 1.0879828736357207,
-      "grad_norm": 0.0023141042329370975,
-      "learning_rate": 0.00019999941668451948,
-      "loss": 46.0,
-      "step": 14230
-    },
-    {
-      "epoch": 1.0880593306191104,
-      "grad_norm": 0.002171831438317895,
-      "learning_rate": 0.00019999941660247514,
-      "loss": 46.0,
-      "step": 14231
-    },
-    {
-      "epoch": 1.0881357876025002,
-      "grad_norm": 0.005008396692574024,
-      "learning_rate": 0.000199999416520425,
-      "loss": 46.0,
-      "step": 14232
-    },
-    {
-      "epoch": 1.08821224458589,
-      "grad_norm": 0.0005496269441209733,
-      "learning_rate": 0.00019999941643836908,
-      "loss": 46.0,
-      "step": 14233
-    },
-    {
-      "epoch": 1.0882887015692795,
-      "grad_norm": 0.0019480016781017184,
-      "learning_rate": 0.00019999941635630744,
-      "loss": 46.0,
-      "step": 14234
-    },
-    {
-      "epoch": 1.0883651585526692,
-      "grad_norm": 0.0005285037914291024,
-      "learning_rate": 0.00019999941627423998,
-      "loss": 46.0,
-      "step": 14235
-    },
-    {
-      "epoch": 1.088441615536059,
-      "grad_norm": 0.0030830376781523228,
-      "learning_rate": 0.00019999941619216677,
-      "loss": 46.0,
-      "step": 14236
-    },
-    {
-      "epoch": 1.0885180725194488,
-      "grad_norm": 0.0007691544597037137,
-      "learning_rate": 0.0001999994161100878,
-      "loss": 46.0,
-      "step": 14237
-    },
-    {
-      "epoch": 1.0885945295028385,
-      "grad_norm": 0.0013154344633221626,
-      "learning_rate": 0.00019999941602800303,
-      "loss": 46.0,
-      "step": 14238
-    },
-    {
-      "epoch": 1.0886709864862283,
-      "grad_norm": 0.0021835200022906065,
-      "learning_rate": 0.00019999941594591253,
-      "loss": 46.0,
-      "step": 14239
-    },
-    {
-      "epoch": 1.0887474434696178,
-      "grad_norm": 0.0011576777324080467,
-      "learning_rate": 0.00019999941586381622,
-      "loss": 46.0,
-      "step": 14240
-    },
-    {
-      "epoch": 1.0888239004530076,
-      "grad_norm": 0.0003792611532844603,
-      "learning_rate": 0.00019999941578171417,
-      "loss": 46.0,
-      "step": 14241
-    },
-    {
-      "epoch": 1.0889003574363973,
-      "grad_norm": 0.0009405299206264317,
-      "learning_rate": 0.00019999941569960635,
-      "loss": 46.0,
-      "step": 14242
-    },
-    {
-      "epoch": 1.088976814419787,
-      "grad_norm": 0.0006741826073266566,
-      "learning_rate": 0.00019999941561749275,
-      "loss": 46.0,
-      "step": 14243
-    },
-    {
-      "epoch": 1.0890532714031769,
-      "grad_norm": 0.004020855762064457,
-      "learning_rate": 0.0001999994155353734,
-      "loss": 46.0,
-      "step": 14244
-    },
-    {
-      "epoch": 1.0891297283865664,
-      "grad_norm": 0.000288337905658409,
-      "learning_rate": 0.00019999941545324824,
-      "loss": 46.0,
-      "step": 14245
-    },
-    {
-      "epoch": 1.0892061853699562,
-      "grad_norm": 0.0007682301802560687,
-      "learning_rate": 0.00019999941537111735,
-      "loss": 46.0,
-      "step": 14246
-    },
-    {
-      "epoch": 1.089282642353346,
-      "grad_norm": 0.0005076015368103981,
-      "learning_rate": 0.00019999941528898066,
-      "loss": 46.0,
-      "step": 14247
-    },
-    {
-      "epoch": 1.0893590993367357,
-      "grad_norm": 0.0020966168958693743,
-      "learning_rate": 0.00019999941520683822,
-      "loss": 46.0,
-      "step": 14248
-    },
-    {
-      "epoch": 1.0894355563201255,
-      "grad_norm": 0.0007726805051788688,
-      "learning_rate": 0.00019999941512469002,
-      "loss": 46.0,
-      "step": 14249
-    },
-    {
-      "epoch": 1.0895120133035152,
-      "grad_norm": 0.0025917242746800184,
-      "learning_rate": 0.00019999941504253603,
-      "loss": 46.0,
-      "step": 14250
-    },
-    {
-      "epoch": 1.0895884702869048,
-      "grad_norm": 0.00030111370142549276,
-      "learning_rate": 0.00019999941496037628,
-      "loss": 46.0,
-      "step": 14251
-    },
-    {
-      "epoch": 1.0896649272702945,
-      "grad_norm": 0.0038051693700253963,
-      "learning_rate": 0.00019999941487821077,
-      "loss": 46.0,
-      "step": 14252
-    },
-    {
-      "epoch": 1.0897413842536843,
-      "grad_norm": 0.0014256901340559125,
-      "learning_rate": 0.00019999941479603947,
-      "loss": 46.0,
-      "step": 14253
-    },
-    {
-      "epoch": 1.089817841237074,
-      "grad_norm": 0.0007533254683949053,
-      "learning_rate": 0.00019999941471386242,
-      "loss": 46.0,
-      "step": 14254
-    },
-    {
-      "epoch": 1.0898942982204638,
-      "grad_norm": 0.004037715960294008,
-      "learning_rate": 0.00019999941463167957,
-      "loss": 46.0,
-      "step": 14255
-    },
-    {
-      "epoch": 1.0899707552038533,
-      "grad_norm": 0.008206482045352459,
-      "learning_rate": 0.00019999941454949097,
-      "loss": 46.0,
-      "step": 14256
-    },
-    {
-      "epoch": 1.090047212187243,
-      "grad_norm": 0.0006208682316355407,
-      "learning_rate": 0.00019999941446729663,
-      "loss": 46.0,
-      "step": 14257
-    },
-    {
-      "epoch": 1.0901236691706329,
-      "grad_norm": 0.000864917878061533,
-      "learning_rate": 0.00019999941438509646,
-      "loss": 46.0,
-      "step": 14258
-    },
-    {
-      "epoch": 1.0902001261540226,
-      "grad_norm": 0.0031939952168613672,
-      "learning_rate": 0.00019999941430289057,
-      "loss": 46.0,
-      "step": 14259
-    },
-    {
-      "epoch": 1.0902765831374124,
-      "grad_norm": 0.0006800976698286831,
-      "learning_rate": 0.00019999941422067888,
-      "loss": 46.0,
-      "step": 14260
-    },
-    {
-      "epoch": 1.0903530401208021,
-      "grad_norm": 0.0012649616692215204,
-      "learning_rate": 0.00019999941413846145,
-      "loss": 46.0,
-      "step": 14261
-    },
-    {
-      "epoch": 1.0904294971041917,
-      "grad_norm": 0.0011895967181771994,
-      "learning_rate": 0.00019999941405623824,
-      "loss": 46.0,
-      "step": 14262
-    },
-    {
-      "epoch": 1.0905059540875814,
-      "grad_norm": 0.0017825077520683408,
-      "learning_rate": 0.00019999941397400925,
-      "loss": 46.0,
-      "step": 14263
-    },
-    {
-      "epoch": 1.0905824110709712,
-      "grad_norm": 0.001082477392628789,
-      "learning_rate": 0.0001999994138917745,
-      "loss": 46.0,
-      "step": 14264
-    },
-    {
-      "epoch": 1.090658868054361,
-      "grad_norm": 0.0014291038969531655,
-      "learning_rate": 0.00019999941380953397,
-      "loss": 46.0,
-      "step": 14265
-    },
-    {
-      "epoch": 1.0907353250377507,
-      "grad_norm": 0.002422285731881857,
-      "learning_rate": 0.00019999941372728767,
-      "loss": 46.0,
-      "step": 14266
-    },
-    {
-      "epoch": 1.0908117820211403,
-      "grad_norm": 0.0010261788265779614,
-      "learning_rate": 0.00019999941364503562,
-      "loss": 46.0,
-      "step": 14267
-    },
-    {
-      "epoch": 1.09088823900453,
-      "grad_norm": 0.0014432704774662852,
-      "learning_rate": 0.0001999994135627778,
-      "loss": 46.0,
-      "step": 14268
-    },
-    {
-      "epoch": 1.0909646959879198,
-      "grad_norm": 0.009557048790156841,
-      "learning_rate": 0.0001999994134805142,
-      "loss": 46.0,
-      "step": 14269
-    },
-    {
-      "epoch": 1.0910411529713095,
-      "grad_norm": 0.001137599116191268,
-      "learning_rate": 0.0001999994133982448,
-      "loss": 46.0,
-      "step": 14270
-    },
-    {
-      "epoch": 1.0911176099546993,
-      "grad_norm": 0.002625234192237258,
-      "learning_rate": 0.0001999994133159697,
-      "loss": 46.0,
-      "step": 14271
-    },
-    {
-      "epoch": 1.091194066938089,
-      "grad_norm": 0.0010089966235682368,
-      "learning_rate": 0.00019999941323368877,
-      "loss": 46.0,
-      "step": 14272
-    },
-    {
-      "epoch": 1.0912705239214786,
-      "grad_norm": 0.0017524787690490484,
-      "learning_rate": 0.00019999941315140208,
-      "loss": 46.0,
-      "step": 14273
-    },
-    {
-      "epoch": 1.0913469809048684,
-      "grad_norm": 0.0013489775592461228,
-      "learning_rate": 0.00019999941306910962,
-      "loss": 46.0,
-      "step": 14274
-    },
-    {
-      "epoch": 1.0914234378882581,
-      "grad_norm": 0.012160534970462322,
-      "learning_rate": 0.0001999994129868114,
-      "loss": 46.0,
-      "step": 14275
-    },
-    {
-      "epoch": 1.091499894871648,
-      "grad_norm": 0.0024522680323570967,
-      "learning_rate": 0.00019999941290450743,
-      "loss": 46.0,
-      "step": 14276
-    },
-    {
-      "epoch": 1.0915763518550377,
-      "grad_norm": 0.0019066098611801863,
-      "learning_rate": 0.00019999941282219767,
-      "loss": 46.0,
-      "step": 14277
-    },
-    {
-      "epoch": 1.0916528088384272,
-      "grad_norm": 0.003900516778230667,
-      "learning_rate": 0.00019999941273988214,
-      "loss": 46.0,
-      "step": 14278
-    },
-    {
-      "epoch": 1.091729265821817,
-      "grad_norm": 0.004702542442828417,
-      "learning_rate": 0.00019999941265756084,
-      "loss": 46.0,
-      "step": 14279
-    },
-    {
-      "epoch": 1.0918057228052067,
-      "grad_norm": 0.003026853082701564,
-      "learning_rate": 0.0001999994125752338,
-      "loss": 46.0,
-      "step": 14280
-    },
-    {
-      "epoch": 1.0918821797885965,
-      "grad_norm": 0.0005396785563789308,
-      "learning_rate": 0.00019999941249290094,
-      "loss": 46.0,
-      "step": 14281
-    },
-    {
-      "epoch": 1.0919586367719862,
-      "grad_norm": 0.0006609810516238213,
-      "learning_rate": 0.00019999941241056235,
-      "loss": 46.0,
-      "step": 14282
-    },
-    {
-      "epoch": 1.092035093755376,
-      "grad_norm": 0.0009154931176453829,
-      "learning_rate": 0.00019999941232821798,
-      "loss": 46.0,
-      "step": 14283
-    },
-    {
-      "epoch": 1.0921115507387655,
-      "grad_norm": 0.003077547065913677,
-      "learning_rate": 0.00019999941224586783,
-      "loss": 46.0,
-      "step": 14284
-    },
-    {
-      "epoch": 1.0921880077221553,
-      "grad_norm": 0.0012425247114151716,
-      "learning_rate": 0.0001999994121635119,
-      "loss": 46.0,
-      "step": 14285
-    },
-    {
-      "epoch": 1.092264464705545,
-      "grad_norm": 0.0006585618830285966,
-      "learning_rate": 0.00019999941208115023,
-      "loss": 46.0,
-      "step": 14286
-    },
-    {
-      "epoch": 1.0923409216889348,
-      "grad_norm": 0.0006826651515439153,
-      "learning_rate": 0.0001999994119987828,
-      "loss": 46.0,
-      "step": 14287
-    },
-    {
-      "epoch": 1.0924173786723246,
-      "grad_norm": 0.0012108294758945704,
-      "learning_rate": 0.00019999941191640956,
-      "loss": 46.0,
-      "step": 14288
-    },
-    {
-      "epoch": 1.0924938356557141,
-      "grad_norm": 0.011197608895599842,
-      "learning_rate": 0.00019999941183403058,
-      "loss": 46.0,
-      "step": 14289
-    },
-    {
-      "epoch": 1.0925702926391039,
-      "grad_norm": 0.007502470165491104,
-      "learning_rate": 0.0001999994117516458,
-      "loss": 46.0,
-      "step": 14290
-    },
-    {
-      "epoch": 1.0926467496224936,
-      "grad_norm": 0.017458975315093994,
-      "learning_rate": 0.0001999994116692553,
-      "loss": 46.0,
-      "step": 14291
-    },
-    {
-      "epoch": 1.0927232066058834,
-      "grad_norm": 0.0008480674587190151,
-      "learning_rate": 0.000199999411586859,
-      "loss": 46.0,
-      "step": 14292
-    },
-    {
-      "epoch": 1.0927996635892732,
-      "grad_norm": 0.002832941245287657,
-      "learning_rate": 0.00019999941150445691,
-      "loss": 46.0,
-      "step": 14293
-    },
-    {
-      "epoch": 1.092876120572663,
-      "grad_norm": 0.005975499749183655,
-      "learning_rate": 0.0001999994114220491,
-      "loss": 46.0,
-      "step": 14294
-    },
-    {
-      "epoch": 1.0929525775560525,
-      "grad_norm": 0.003008169587701559,
-      "learning_rate": 0.00019999941133963547,
-      "loss": 46.0,
-      "step": 14295
-    },
-    {
-      "epoch": 1.0930290345394422,
-      "grad_norm": 0.001098928158171475,
-      "learning_rate": 0.0001999994112572161,
-      "loss": 46.0,
-      "step": 14296
-    },
-    {
-      "epoch": 1.093105491522832,
-      "grad_norm": 0.0005411769379861653,
-      "learning_rate": 0.00019999941117479096,
-      "loss": 46.0,
-      "step": 14297
-    },
-    {
-      "epoch": 1.0931819485062217,
-      "grad_norm": 0.0016416804865002632,
-      "learning_rate": 0.00019999941109236004,
-      "loss": 46.0,
-      "step": 14298
-    },
-    {
-      "epoch": 1.0932584054896115,
-      "grad_norm": 0.0010592266917228699,
-      "learning_rate": 0.00019999941100992336,
-      "loss": 46.0,
-      "step": 14299
-    },
-    {
-      "epoch": 1.093334862473001,
-      "grad_norm": 0.002182136056944728,
-      "learning_rate": 0.0001999994109274809,
-      "loss": 46.0,
-      "step": 14300
-    },
-    {
-      "epoch": 1.0934113194563908,
-      "grad_norm": 0.0008108122856356204,
-      "learning_rate": 0.00019999941084503268,
-      "loss": 46.0,
-      "step": 14301
-    },
-    {
-      "epoch": 1.0934877764397806,
-      "grad_norm": 0.003072872292250395,
-      "learning_rate": 0.00019999941076257868,
-      "loss": 46.0,
-      "step": 14302
-    },
-    {
-      "epoch": 1.0935642334231703,
-      "grad_norm": 0.00027640291955322027,
-      "learning_rate": 0.00019999941068011895,
-      "loss": 46.0,
-      "step": 14303
-    },
-    {
-      "epoch": 1.09364069040656,
-      "grad_norm": 0.005797708407044411,
-      "learning_rate": 0.0001999994105976534,
-      "loss": 46.0,
-      "step": 14304
-    },
-    {
-      "epoch": 1.0937171473899496,
-      "grad_norm": 0.002766105579212308,
-      "learning_rate": 0.0001999994105151821,
-      "loss": 46.0,
-      "step": 14305
-    },
-    {
-      "epoch": 1.0937936043733394,
-      "grad_norm": 0.0011098323157057166,
-      "learning_rate": 0.00019999941043270501,
-      "loss": 46.0,
-      "step": 14306
-    },
-    {
-      "epoch": 1.0938700613567292,
-      "grad_norm": 0.0007261222344823182,
-      "learning_rate": 0.0001999994103502222,
-      "loss": 46.0,
-      "step": 14307
-    },
-    {
-      "epoch": 1.093946518340119,
-      "grad_norm": 0.0014111533528193831,
-      "learning_rate": 0.00019999941026773357,
-      "loss": 46.0,
-      "step": 14308
-    },
-    {
-      "epoch": 1.0940229753235087,
-      "grad_norm": 0.0020265085622668266,
-      "learning_rate": 0.0001999994101852392,
-      "loss": 46.0,
-      "step": 14309
-    },
-    {
-      "epoch": 1.0940994323068984,
-      "grad_norm": 0.0014865364646539092,
-      "learning_rate": 0.00019999941010273906,
-      "loss": 46.0,
-      "step": 14310
-    },
-    {
-      "epoch": 1.094175889290288,
-      "grad_norm": 0.0005349652492441237,
-      "learning_rate": 0.00019999941002023315,
-      "loss": 46.0,
-      "step": 14311
-    },
-    {
-      "epoch": 1.0942523462736777,
-      "grad_norm": 0.0008251930703409016,
-      "learning_rate": 0.00019999940993772146,
-      "loss": 46.0,
-      "step": 14312
-    },
-    {
-      "epoch": 1.0943288032570675,
-      "grad_norm": 0.0005950809572823346,
-      "learning_rate": 0.000199999409855204,
-      "loss": 46.0,
-      "step": 14313
-    },
-    {
-      "epoch": 1.0944052602404573,
-      "grad_norm": 0.004502730909734964,
-      "learning_rate": 0.00019999940977268076,
-      "loss": 46.0,
-      "step": 14314
-    },
-    {
-      "epoch": 1.094481717223847,
-      "grad_norm": 0.001285791746340692,
-      "learning_rate": 0.00019999940969015175,
-      "loss": 46.0,
-      "step": 14315
-    },
-    {
-      "epoch": 1.0945581742072366,
-      "grad_norm": 0.0010087513364851475,
-      "learning_rate": 0.000199999409607617,
-      "loss": 46.0,
-      "step": 14316
-    },
-    {
-      "epoch": 1.0946346311906263,
-      "grad_norm": 0.0014537750976160169,
-      "learning_rate": 0.00019999940952507647,
-      "loss": 46.0,
-      "step": 14317
-    },
-    {
-      "epoch": 1.094711088174016,
-      "grad_norm": 0.0007825226639397442,
-      "learning_rate": 0.00019999940944253017,
-      "loss": 46.0,
-      "step": 14318
-    },
-    {
-      "epoch": 1.0947875451574058,
-      "grad_norm": 0.0012522010365501046,
-      "learning_rate": 0.0001999994093599781,
-      "loss": 46.0,
-      "step": 14319
-    },
-    {
-      "epoch": 1.0948640021407956,
-      "grad_norm": 0.0015893217641860247,
-      "learning_rate": 0.00019999940927742027,
-      "loss": 46.0,
-      "step": 14320
-    },
-    {
-      "epoch": 1.0949404591241851,
-      "grad_norm": 0.0011654389090836048,
-      "learning_rate": 0.00019999940919485665,
-      "loss": 46.0,
-      "step": 14321
-    },
-    {
-      "epoch": 1.095016916107575,
-      "grad_norm": 0.0013971192529425025,
-      "learning_rate": 0.00019999940911228728,
-      "loss": 46.0,
-      "step": 14322
-    },
-    {
-      "epoch": 1.0950933730909647,
-      "grad_norm": 0.0007000422920100391,
-      "learning_rate": 0.0001999994090297121,
-      "loss": 46.0,
-      "step": 14323
-    },
-    {
-      "epoch": 1.0951698300743544,
-      "grad_norm": 0.004778878763318062,
-      "learning_rate": 0.00019999940894713117,
-      "loss": 46.0,
-      "step": 14324
-    },
-    {
-      "epoch": 1.0952462870577442,
-      "grad_norm": 0.0007299938006326556,
-      "learning_rate": 0.0001999994088645445,
-      "loss": 46.0,
-      "step": 14325
-    },
-    {
-      "epoch": 1.095322744041134,
-      "grad_norm": 0.0012658416526392102,
-      "learning_rate": 0.00019999940878195205,
-      "loss": 46.0,
-      "step": 14326
-    },
-    {
-      "epoch": 1.0953992010245235,
-      "grad_norm": 0.001013493980281055,
-      "learning_rate": 0.00019999940869935384,
-      "loss": 46.0,
-      "step": 14327
-    },
-    {
-      "epoch": 1.0954756580079132,
-      "grad_norm": 0.0020321214105933905,
-      "learning_rate": 0.0001999994086167498,
-      "loss": 46.0,
-      "step": 14328
-    },
-    {
-      "epoch": 1.095552114991303,
-      "grad_norm": 0.0014854243490844965,
-      "learning_rate": 0.00019999940853414005,
-      "loss": 46.0,
-      "step": 14329
-    },
-    {
-      "epoch": 1.0956285719746928,
-      "grad_norm": 0.0029417916666716337,
-      "learning_rate": 0.00019999940845152452,
-      "loss": 46.0,
-      "step": 14330
-    },
-    {
-      "epoch": 1.0957050289580825,
-      "grad_norm": 0.0057570659555494785,
-      "learning_rate": 0.00019999940836890322,
-      "loss": 46.0,
-      "step": 14331
-    },
-    {
-      "epoch": 1.095781485941472,
-      "grad_norm": 0.0041068303398787975,
-      "learning_rate": 0.00019999940828627615,
-      "loss": 46.0,
-      "step": 14332
-    },
-    {
-      "epoch": 1.0958579429248618,
-      "grad_norm": 0.0025582164525985718,
-      "learning_rate": 0.0001999994082036433,
-      "loss": 46.0,
-      "step": 14333
-    },
-    {
-      "epoch": 1.0959343999082516,
-      "grad_norm": 0.0004594225320033729,
-      "learning_rate": 0.00019999940812100468,
-      "loss": 46.0,
-      "step": 14334
-    },
-    {
-      "epoch": 1.0960108568916413,
-      "grad_norm": 0.007679132744669914,
-      "learning_rate": 0.0001999994080383603,
-      "loss": 46.0,
-      "step": 14335
-    },
-    {
-      "epoch": 1.096087313875031,
-      "grad_norm": 0.0033311406150460243,
-      "learning_rate": 0.00019999940795571014,
-      "loss": 46.0,
-      "step": 14336
-    },
-    {
-      "epoch": 1.0961637708584209,
-      "grad_norm": 0.002244324656203389,
-      "learning_rate": 0.00019999940787305423,
-      "loss": 46.0,
-      "step": 14337
-    },
-    {
-      "epoch": 1.0962402278418104,
-      "grad_norm": 0.002362343482673168,
-      "learning_rate": 0.0001999994077903925,
-      "loss": 46.0,
-      "step": 14338
-    },
-    {
-      "epoch": 1.0963166848252002,
-      "grad_norm": 0.0023635150864720345,
-      "learning_rate": 0.00019999940770772508,
-      "loss": 46.0,
-      "step": 14339
-    },
-    {
-      "epoch": 1.09639314180859,
-      "grad_norm": 0.0016179463127627969,
-      "learning_rate": 0.00019999940762505184,
-      "loss": 46.0,
-      "step": 14340
-    },
-    {
-      "epoch": 1.0964695987919797,
-      "grad_norm": 0.0005304119549691677,
-      "learning_rate": 0.00019999940754237284,
-      "loss": 46.0,
-      "step": 14341
-    },
-    {
-      "epoch": 1.0965460557753695,
-      "grad_norm": 0.0058393292129039764,
-      "learning_rate": 0.00019999940745968808,
-      "loss": 46.0,
-      "step": 14342
-    },
-    {
-      "epoch": 1.096622512758759,
-      "grad_norm": 0.004538960754871368,
-      "learning_rate": 0.00019999940737699753,
-      "loss": 46.0,
-      "step": 14343
-    },
-    {
-      "epoch": 1.0966989697421488,
-      "grad_norm": 0.0038289304357022047,
-      "learning_rate": 0.00019999940729430123,
-      "loss": 46.0,
-      "step": 14344
-    },
-    {
-      "epoch": 1.0967754267255385,
-      "grad_norm": 0.001197592355310917,
-      "learning_rate": 0.00019999940721159915,
-      "loss": 46.0,
-      "step": 14345
-    },
-    {
-      "epoch": 1.0968518837089283,
-      "grad_norm": 0.0008619913132861257,
-      "learning_rate": 0.0001999994071288913,
-      "loss": 46.0,
-      "step": 14346
-    },
-    {
-      "epoch": 1.096928340692318,
-      "grad_norm": 0.0010824321070685983,
-      "learning_rate": 0.0001999994070461777,
-      "loss": 46.0,
-      "step": 14347
-    },
-    {
-      "epoch": 1.0970047976757078,
-      "grad_norm": 0.0008617830462753773,
-      "learning_rate": 0.00019999940696345831,
-      "loss": 46.0,
-      "step": 14348
-    },
-    {
-      "epoch": 1.0970812546590973,
-      "grad_norm": 0.0014405191177502275,
-      "learning_rate": 0.00019999940688073315,
-      "loss": 46.0,
-      "step": 14349
-    },
-    {
-      "epoch": 1.097157711642487,
-      "grad_norm": 0.0021947685163468122,
-      "learning_rate": 0.00019999940679800223,
-      "loss": 46.0,
-      "step": 14350
-    },
-    {
-      "epoch": 1.0972341686258769,
-      "grad_norm": 0.0004924390814267099,
-      "learning_rate": 0.00019999940671526555,
-      "loss": 46.0,
-      "step": 14351
-    },
-    {
-      "epoch": 1.0973106256092666,
-      "grad_norm": 0.0009457893902435899,
-      "learning_rate": 0.00019999940663252308,
-      "loss": 46.0,
-      "step": 14352
-    },
-    {
-      "epoch": 1.0973870825926564,
-      "grad_norm": 0.001278024516068399,
-      "learning_rate": 0.00019999940654977482,
-      "loss": 46.0,
-      "step": 14353
-    },
-    {
-      "epoch": 1.097463539576046,
-      "grad_norm": 0.04226132109761238,
-      "learning_rate": 0.00019999940646702084,
-      "loss": 46.0,
-      "step": 14354
-    },
-    {
-      "epoch": 1.0975399965594357,
-      "grad_norm": 0.0015559179009869695,
-      "learning_rate": 0.00019999940638426103,
-      "loss": 46.0,
-      "step": 14355
-    },
-    {
-      "epoch": 1.0976164535428254,
-      "grad_norm": 0.005004845093935728,
-      "learning_rate": 0.0001999994063014955,
-      "loss": 46.0,
-      "step": 14356
-    },
-    {
-      "epoch": 1.0976929105262152,
-      "grad_norm": 0.0012471572263166308,
-      "learning_rate": 0.0001999994062187242,
-      "loss": 46.0,
-      "step": 14357
-    },
-    {
-      "epoch": 1.097769367509605,
-      "grad_norm": 0.0009397243848070502,
-      "learning_rate": 0.00019999940613594713,
-      "loss": 46.0,
-      "step": 14358
-    },
-    {
-      "epoch": 1.0978458244929947,
-      "grad_norm": 0.00046177723561413586,
-      "learning_rate": 0.00019999940605316429,
-      "loss": 46.0,
-      "step": 14359
-    },
-    {
-      "epoch": 1.0979222814763843,
-      "grad_norm": 0.0020263090264052153,
-      "learning_rate": 0.00019999940597037566,
-      "loss": 46.0,
-      "step": 14360
-    },
-    {
-      "epoch": 1.097998738459774,
-      "grad_norm": 0.0009293400798924267,
-      "learning_rate": 0.00019999940588758127,
-      "loss": 46.0,
-      "step": 14361
-    },
-    {
-      "epoch": 1.0980751954431638,
-      "grad_norm": 0.0010265014134347439,
-      "learning_rate": 0.0001999994058047811,
-      "loss": 46.0,
-      "step": 14362
-    },
-    {
-      "epoch": 1.0981516524265535,
-      "grad_norm": 0.003921338357031345,
-      "learning_rate": 0.00019999940572197516,
-      "loss": 46.0,
-      "step": 14363
-    },
-    {
-      "epoch": 1.0982281094099433,
-      "grad_norm": 0.0016031806590035558,
-      "learning_rate": 0.0001999994056391635,
-      "loss": 46.0,
-      "step": 14364
-    },
-    {
-      "epoch": 1.0983045663933328,
-      "grad_norm": 0.002569162053987384,
-      "learning_rate": 0.00019999940555634601,
-      "loss": 46.0,
-      "step": 14365
-    },
-    {
-      "epoch": 1.0983810233767226,
-      "grad_norm": 0.000720013864338398,
-      "learning_rate": 0.0001999994054735228,
-      "loss": 46.0,
-      "step": 14366
-    },
-    {
-      "epoch": 1.0984574803601124,
-      "grad_norm": 0.002361343940719962,
-      "learning_rate": 0.00019999940539069377,
-      "loss": 46.0,
-      "step": 14367
-    },
-    {
-      "epoch": 1.0985339373435021,
-      "grad_norm": 0.000996338319964707,
-      "learning_rate": 0.00019999940530785902,
-      "loss": 46.0,
-      "step": 14368
-    },
-    {
-      "epoch": 1.098610394326892,
-      "grad_norm": 0.0010644549038261175,
-      "learning_rate": 0.00019999940522501846,
-      "loss": 46.0,
-      "step": 14369
-    },
-    {
-      "epoch": 1.0986868513102817,
-      "grad_norm": 0.0018155085854232311,
-      "learning_rate": 0.00019999940514217216,
-      "loss": 46.0,
-      "step": 14370
-    },
-    {
-      "epoch": 1.0987633082936712,
-      "grad_norm": 0.0008115923847071826,
-      "learning_rate": 0.0001999994050593201,
-      "loss": 46.0,
-      "step": 14371
-    },
-    {
-      "epoch": 1.098839765277061,
-      "grad_norm": 0.001588737010024488,
-      "learning_rate": 0.00019999940497646224,
-      "loss": 46.0,
-      "step": 14372
-    },
-    {
-      "epoch": 1.0989162222604507,
-      "grad_norm": 0.0013079143827781081,
-      "learning_rate": 0.0001999994048935986,
-      "loss": 46.0,
-      "step": 14373
-    },
-    {
-      "epoch": 1.0989926792438405,
-      "grad_norm": 0.0016037931200116873,
-      "learning_rate": 0.0001999994048107292,
-      "loss": 46.0,
-      "step": 14374
-    },
-    {
-      "epoch": 1.0990691362272302,
-      "grad_norm": 0.0015312456525862217,
-      "learning_rate": 0.00019999940472785406,
-      "loss": 46.0,
-      "step": 14375
-    },
-    {
-      "epoch": 1.0991455932106198,
-      "grad_norm": 0.004522494971752167,
-      "learning_rate": 0.00019999940464497312,
-      "loss": 46.0,
-      "step": 14376
-    },
-    {
-      "epoch": 1.0992220501940095,
-      "grad_norm": 0.0021111182868480682,
-      "learning_rate": 0.0001999994045620864,
-      "loss": 46.0,
-      "step": 14377
-    },
-    {
-      "epoch": 1.0992985071773993,
-      "grad_norm": 0.0008027053554542363,
-      "learning_rate": 0.00019999940447919397,
-      "loss": 46.0,
-      "step": 14378
-    },
-    {
-      "epoch": 1.099374964160789,
-      "grad_norm": 0.0007732062367722392,
-      "learning_rate": 0.00019999940439629571,
-      "loss": 46.0,
-      "step": 14379
-    },
-    {
-      "epoch": 1.0994514211441788,
-      "grad_norm": 0.001535898307338357,
-      "learning_rate": 0.0001999994043133917,
-      "loss": 46.0,
-      "step": 14380
-    },
-    {
-      "epoch": 1.0995278781275686,
-      "grad_norm": 0.0068520293571054935,
-      "learning_rate": 0.00019999940423048193,
-      "loss": 46.0,
-      "step": 14381
-    },
-    {
-      "epoch": 1.0996043351109581,
-      "grad_norm": 0.002485432429239154,
-      "learning_rate": 0.0001999994041475664,
-      "loss": 46.0,
-      "step": 14382
-    },
-    {
-      "epoch": 1.0996807920943479,
-      "grad_norm": 0.001896600821055472,
-      "learning_rate": 0.00019999940406464507,
-      "loss": 46.0,
-      "step": 14383
-    },
-    {
-      "epoch": 1.0997572490777376,
-      "grad_norm": 0.001922485651448369,
-      "learning_rate": 0.000199999403981718,
-      "loss": 46.0,
-      "step": 14384
-    },
-    {
-      "epoch": 1.0998337060611274,
-      "grad_norm": 0.002252592472359538,
-      "learning_rate": 0.00019999940389878513,
-      "loss": 46.0,
-      "step": 14385
-    },
-    {
-      "epoch": 1.0999101630445172,
-      "grad_norm": 0.0007943834061734378,
-      "learning_rate": 0.0001999994038158465,
-      "loss": 46.0,
-      "step": 14386
-    },
-    {
-      "epoch": 1.0999866200279067,
-      "grad_norm": 0.002882985631003976,
-      "learning_rate": 0.00019999940373290214,
-      "loss": 46.0,
-      "step": 14387
-    },
-    {
-      "epoch": 1.1000630770112965,
-      "grad_norm": 0.0019011124968528748,
-      "learning_rate": 0.00019999940364995194,
-      "loss": 46.0,
-      "step": 14388
-    },
-    {
-      "epoch": 1.1001395339946862,
-      "grad_norm": 0.0033900458365678787,
-      "learning_rate": 0.00019999940356699603,
-      "loss": 46.0,
-      "step": 14389
-    },
-    {
-      "epoch": 1.100215990978076,
-      "grad_norm": 0.0015945249469950795,
-      "learning_rate": 0.00019999940348403432,
-      "loss": 46.0,
-      "step": 14390
-    },
-    {
-      "epoch": 1.1002924479614657,
-      "grad_norm": 0.0021856757812201977,
-      "learning_rate": 0.00019999940340106686,
-      "loss": 46.0,
-      "step": 14391
-    },
-    {
-      "epoch": 1.1003689049448555,
-      "grad_norm": 0.00046254173503257334,
-      "learning_rate": 0.00019999940331809362,
-      "loss": 46.0,
-      "step": 14392
-    },
-    {
-      "epoch": 1.100445361928245,
-      "grad_norm": 0.0010494188172742724,
-      "learning_rate": 0.0001999994032351146,
-      "loss": 46.0,
-      "step": 14393
-    },
-    {
-      "epoch": 1.1005218189116348,
-      "grad_norm": 0.0008125773165374994,
-      "learning_rate": 0.0001999994031521298,
-      "loss": 46.0,
-      "step": 14394
-    },
-    {
-      "epoch": 1.1005982758950246,
-      "grad_norm": 0.001067085424438119,
-      "learning_rate": 0.00019999940306913926,
-      "loss": 46.0,
-      "step": 14395
-    },
-    {
-      "epoch": 1.1006747328784143,
-      "grad_norm": 0.0013594641350209713,
-      "learning_rate": 0.00019999940298614296,
-      "loss": 46.0,
-      "step": 14396
-    },
-    {
-      "epoch": 1.100751189861804,
-      "grad_norm": 0.0024234233424067497,
-      "learning_rate": 0.00019999940290314088,
-      "loss": 46.0,
-      "step": 14397
-    },
-    {
-      "epoch": 1.1008276468451936,
-      "grad_norm": 0.0005430607707239687,
-      "learning_rate": 0.00019999940282013304,
-      "loss": 46.0,
-      "step": 14398
-    },
-    {
-      "epoch": 1.1009041038285834,
-      "grad_norm": 0.0011346307583153248,
-      "learning_rate": 0.0001999994027371194,
-      "loss": 46.0,
-      "step": 14399
-    },
-    {
-      "epoch": 1.1009805608119732,
-      "grad_norm": 0.019765105098485947,
-      "learning_rate": 0.0001999994026541,
-      "loss": 46.0,
-      "step": 14400
-    },
-    {
-      "epoch": 1.101057017795363,
-      "grad_norm": 0.0009854196105152369,
-      "learning_rate": 0.00019999940257107483,
-      "loss": 46.0,
-      "step": 14401
-    },
-    {
-      "epoch": 1.1011334747787527,
-      "grad_norm": 0.0010334094986319542,
-      "learning_rate": 0.0001999994024880439,
-      "loss": 46.0,
-      "step": 14402
-    },
-    {
-      "epoch": 1.1012099317621424,
-      "grad_norm": 0.0016409477684646845,
-      "learning_rate": 0.00019999940240500718,
-      "loss": 46.0,
-      "step": 14403
-    },
-    {
-      "epoch": 1.101286388745532,
-      "grad_norm": 0.004089921712875366,
-      "learning_rate": 0.00019999940232196472,
-      "loss": 46.0,
-      "step": 14404
-    },
-    {
-      "epoch": 1.1013628457289217,
-      "grad_norm": 0.0032342243939638138,
-      "learning_rate": 0.00019999940223891649,
-      "loss": 46.0,
-      "step": 14405
-    },
-    {
-      "epoch": 1.1014393027123115,
-      "grad_norm": 0.0013204023707658052,
-      "learning_rate": 0.00019999940215586248,
-      "loss": 46.0,
-      "step": 14406
-    },
-    {
-      "epoch": 1.1015157596957013,
-      "grad_norm": 0.0014779911143705249,
-      "learning_rate": 0.0001999994020728027,
-      "loss": 46.0,
-      "step": 14407
-    },
-    {
-      "epoch": 1.101592216679091,
-      "grad_norm": 0.0029300181195139885,
-      "learning_rate": 0.00019999940198973715,
-      "loss": 46.0,
-      "step": 14408
-    },
-    {
-      "epoch": 1.1016686736624806,
-      "grad_norm": 0.008498414419591427,
-      "learning_rate": 0.00019999940190666582,
-      "loss": 46.0,
-      "step": 14409
-    },
-    {
-      "epoch": 1.1017451306458703,
-      "grad_norm": 0.0032741765026003122,
-      "learning_rate": 0.00019999940182358872,
-      "loss": 46.0,
-      "step": 14410
-    },
-    {
-      "epoch": 1.10182158762926,
-      "grad_norm": 0.0019324973691254854,
-      "learning_rate": 0.00019999940174050587,
-      "loss": 46.0,
-      "step": 14411
-    },
-    {
-      "epoch": 1.1018980446126498,
-      "grad_norm": 0.0022848667576909065,
-      "learning_rate": 0.00019999940165741725,
-      "loss": 46.0,
-      "step": 14412
-    },
-    {
-      "epoch": 1.1019745015960396,
-      "grad_norm": 0.0010458447504788637,
-      "learning_rate": 0.00019999940157432283,
-      "loss": 46.0,
-      "step": 14413
-    },
-    {
-      "epoch": 1.1020509585794294,
-      "grad_norm": 0.0016525558894500136,
-      "learning_rate": 0.00019999940149122267,
-      "loss": 46.0,
-      "step": 14414
-    },
-    {
-      "epoch": 1.102127415562819,
-      "grad_norm": 0.0020195390097796917,
-      "learning_rate": 0.00019999940140811675,
-      "loss": 46.0,
-      "step": 14415
-    },
-    {
-      "epoch": 1.1022038725462087,
-      "grad_norm": 0.003891491098329425,
-      "learning_rate": 0.00019999940132500504,
-      "loss": 46.0,
-      "step": 14416
-    },
-    {
-      "epoch": 1.1022803295295984,
-      "grad_norm": 0.0005561204743571579,
-      "learning_rate": 0.00019999940124188753,
-      "loss": 46.0,
-      "step": 14417
-    },
-    {
-      "epoch": 1.1023567865129882,
-      "grad_norm": 0.0009384403238072991,
-      "learning_rate": 0.0001999994011587643,
-      "loss": 46.0,
-      "step": 14418
-    },
-    {
-      "epoch": 1.102433243496378,
-      "grad_norm": 0.0015144681092351675,
-      "learning_rate": 0.0001999994010756353,
-      "loss": 46.0,
-      "step": 14419
-    },
-    {
-      "epoch": 1.1025097004797675,
-      "grad_norm": 0.0014589742058888078,
-      "learning_rate": 0.00019999940099250048,
-      "loss": 46.0,
-      "step": 14420
-    },
-    {
-      "epoch": 1.1025861574631572,
-      "grad_norm": 0.0035015870817005634,
-      "learning_rate": 0.00019999940090935996,
-      "loss": 46.0,
-      "step": 14421
-    },
-    {
-      "epoch": 1.102662614446547,
-      "grad_norm": 0.0008231739629991353,
-      "learning_rate": 0.00019999940082621366,
-      "loss": 46.0,
-      "step": 14422
-    },
-    {
-      "epoch": 1.1027390714299368,
-      "grad_norm": 0.0009577129385434091,
-      "learning_rate": 0.00019999940074306153,
-      "loss": 46.0,
-      "step": 14423
-    },
-    {
-      "epoch": 1.1028155284133265,
-      "grad_norm": 0.0007714586681686342,
-      "learning_rate": 0.00019999940065990369,
-      "loss": 46.0,
-      "step": 14424
-    },
-    {
-      "epoch": 1.1028919853967163,
-      "grad_norm": 0.0008904701680876315,
-      "learning_rate": 0.00019999940057674004,
-      "loss": 46.0,
-      "step": 14425
-    },
-    {
-      "epoch": 1.1029684423801058,
-      "grad_norm": 0.003224317217245698,
-      "learning_rate": 0.00019999940049357065,
-      "loss": 46.0,
-      "step": 14426
-    },
-    {
-      "epoch": 1.1030448993634956,
-      "grad_norm": 0.0016535185277462006,
-      "learning_rate": 0.00019999940041039545,
-      "loss": 46.0,
-      "step": 14427
-    },
-    {
-      "epoch": 1.1031213563468854,
-      "grad_norm": 0.0012804112629964948,
-      "learning_rate": 0.00019999940032721454,
-      "loss": 46.0,
-      "step": 14428
-    },
-    {
-      "epoch": 1.1031978133302751,
-      "grad_norm": 0.0022527556866407394,
-      "learning_rate": 0.00019999940024402783,
-      "loss": 46.0,
-      "step": 14429
-    },
-    {
-      "epoch": 1.1032742703136649,
-      "grad_norm": 0.002016973914578557,
-      "learning_rate": 0.00019999940016083537,
-      "loss": 46.0,
-      "step": 14430
-    },
-    {
-      "epoch": 1.1033507272970544,
-      "grad_norm": 0.000759428832679987,
-      "learning_rate": 0.0001999994000776371,
-      "loss": 46.0,
-      "step": 14431
-    },
-    {
-      "epoch": 1.1034271842804442,
-      "grad_norm": 0.0037978622131049633,
-      "learning_rate": 0.00019999939999443308,
-      "loss": 46.0,
-      "step": 14432
-    },
-    {
-      "epoch": 1.103503641263834,
-      "grad_norm": 0.0011195025872439146,
-      "learning_rate": 0.0001999993999112233,
-      "loss": 46.0,
-      "step": 14433
-    },
-    {
-      "epoch": 1.1035800982472237,
-      "grad_norm": 0.0005680260364897549,
-      "learning_rate": 0.00019999939982800775,
-      "loss": 46.0,
-      "step": 14434
-    },
-    {
-      "epoch": 1.1036565552306135,
-      "grad_norm": 0.001830149907618761,
-      "learning_rate": 0.00019999939974478642,
-      "loss": 46.0,
-      "step": 14435
-    },
-    {
-      "epoch": 1.103733012214003,
-      "grad_norm": 0.013707278296351433,
-      "learning_rate": 0.00019999939966155932,
-      "loss": 46.0,
-      "step": 14436
-    },
-    {
-      "epoch": 1.1038094691973928,
-      "grad_norm": 0.0010511912405490875,
-      "learning_rate": 0.00019999939957832648,
-      "loss": 46.0,
-      "step": 14437
-    },
-    {
-      "epoch": 1.1038859261807825,
-      "grad_norm": 0.005906635895371437,
-      "learning_rate": 0.0001999993994950878,
-      "loss": 46.0,
-      "step": 14438
-    },
-    {
-      "epoch": 1.1039623831641723,
-      "grad_norm": 0.003412651363760233,
-      "learning_rate": 0.00019999939941184344,
-      "loss": 46.0,
-      "step": 14439
-    },
-    {
-      "epoch": 1.104038840147562,
-      "grad_norm": 0.0021113960538059473,
-      "learning_rate": 0.00019999939932859327,
-      "loss": 46.0,
-      "step": 14440
-    },
-    {
-      "epoch": 1.1041152971309518,
-      "grad_norm": 0.0010071038268506527,
-      "learning_rate": 0.00019999939924533733,
-      "loss": 46.0,
-      "step": 14441
-    },
-    {
-      "epoch": 1.1041917541143413,
-      "grad_norm": 0.0021579128224402666,
-      "learning_rate": 0.00019999939916207562,
-      "loss": 46.0,
-      "step": 14442
-    },
-    {
-      "epoch": 1.104268211097731,
-      "grad_norm": 0.007896361872553825,
-      "learning_rate": 0.00019999939907880813,
-      "loss": 46.0,
-      "step": 14443
-    },
-    {
-      "epoch": 1.1043446680811209,
-      "grad_norm": 0.0011740727350115776,
-      "learning_rate": 0.00019999939899553487,
-      "loss": 46.0,
-      "step": 14444
-    },
-    {
-      "epoch": 1.1044211250645106,
-      "grad_norm": 0.0008192333043552935,
-      "learning_rate": 0.00019999939891225584,
-      "loss": 46.0,
-      "step": 14445
-    },
-    {
-      "epoch": 1.1044975820479004,
-      "grad_norm": 0.00061931146774441,
-      "learning_rate": 0.00019999939882897106,
-      "loss": 46.0,
-      "step": 14446
-    },
-    {
-      "epoch": 1.10457403903129,
-      "grad_norm": 0.002528889337554574,
-      "learning_rate": 0.0001999993987456805,
-      "loss": 46.0,
-      "step": 14447
-    },
-    {
-      "epoch": 1.1046504960146797,
-      "grad_norm": 0.0010466751409694552,
-      "learning_rate": 0.00019999939866238419,
-      "loss": 46.0,
-      "step": 14448
-    },
-    {
-      "epoch": 1.1047269529980694,
-      "grad_norm": 0.0010379026643931866,
-      "learning_rate": 0.0001999993985790821,
-      "loss": 46.0,
-      "step": 14449
-    },
-    {
-      "epoch": 1.1048034099814592,
-      "grad_norm": 0.001050668186508119,
-      "learning_rate": 0.00019999939849577421,
-      "loss": 46.0,
-      "step": 14450
-    },
-    {
-      "epoch": 1.104879866964849,
-      "grad_norm": 0.0033809831365942955,
-      "learning_rate": 0.00019999939841246057,
-      "loss": 46.0,
-      "step": 14451
-    },
-    {
-      "epoch": 1.1049563239482385,
-      "grad_norm": 0.000771119084674865,
-      "learning_rate": 0.00019999939832914118,
-      "loss": 46.0,
-      "step": 14452
-    },
-    {
-      "epoch": 1.1050327809316283,
-      "grad_norm": 0.003728782758116722,
-      "learning_rate": 0.00019999939824581598,
-      "loss": 46.0,
-      "step": 14453
-    },
-    {
-      "epoch": 1.105109237915018,
-      "grad_norm": 0.002532541286200285,
-      "learning_rate": 0.00019999939816248505,
-      "loss": 46.0,
-      "step": 14454
-    },
-    {
-      "epoch": 1.1051856948984078,
-      "grad_norm": 0.000899175473023206,
-      "learning_rate": 0.00019999939807914833,
-      "loss": 46.0,
-      "step": 14455
-    },
-    {
-      "epoch": 1.1052621518817975,
-      "grad_norm": 0.0028807383496314287,
-      "learning_rate": 0.00019999939799580588,
-      "loss": 46.0,
-      "step": 14456
-    },
-    {
-      "epoch": 1.1053386088651873,
-      "grad_norm": 0.00307116424664855,
-      "learning_rate": 0.00019999939791245762,
-      "loss": 46.0,
-      "step": 14457
-    },
-    {
-      "epoch": 1.1054150658485768,
-      "grad_norm": 0.0006544000934809446,
-      "learning_rate": 0.00019999939782910358,
-      "loss": 46.0,
-      "step": 14458
-    },
-    {
-      "epoch": 1.1054915228319666,
-      "grad_norm": 0.0010353238321840763,
-      "learning_rate": 0.00019999939774574378,
-      "loss": 46.0,
-      "step": 14459
-    },
-    {
-      "epoch": 1.1055679798153564,
-      "grad_norm": 0.0013588287401944399,
-      "learning_rate": 0.00019999939766237823,
-      "loss": 46.0,
-      "step": 14460
-    },
-    {
-      "epoch": 1.1056444367987461,
-      "grad_norm": 0.015461375936865807,
-      "learning_rate": 0.00019999939757900693,
-      "loss": 46.0,
-      "step": 14461
-    },
-    {
-      "epoch": 1.105720893782136,
-      "grad_norm": 0.020595377311110497,
-      "learning_rate": 0.0001999993974956298,
-      "loss": 46.0,
-      "step": 14462
-    },
-    {
-      "epoch": 1.1057973507655254,
-      "grad_norm": 0.0017215183470398188,
-      "learning_rate": 0.00019999939741224693,
-      "loss": 46.0,
-      "step": 14463
-    },
-    {
-      "epoch": 1.1058738077489152,
-      "grad_norm": 0.0008422309765592217,
-      "learning_rate": 0.0001999993973288583,
-      "loss": 46.0,
-      "step": 14464
-    },
-    {
-      "epoch": 1.105950264732305,
-      "grad_norm": 0.005359334871172905,
-      "learning_rate": 0.0001999993972454639,
-      "loss": 46.0,
-      "step": 14465
-    },
-    {
-      "epoch": 1.1060267217156947,
-      "grad_norm": 0.0011890229070559144,
-      "learning_rate": 0.00019999939716206373,
-      "loss": 46.0,
-      "step": 14466
-    },
-    {
-      "epoch": 1.1061031786990845,
-      "grad_norm": 0.003403395414352417,
-      "learning_rate": 0.0001999993970786578,
-      "loss": 46.0,
-      "step": 14467
-    },
-    {
-      "epoch": 1.1061796356824742,
-      "grad_norm": 0.004768850281834602,
-      "learning_rate": 0.00019999939699524605,
-      "loss": 46.0,
-      "step": 14468
-    },
-    {
-      "epoch": 1.1062560926658638,
-      "grad_norm": 0.0024453282821923494,
-      "learning_rate": 0.00019999939691182857,
-      "loss": 46.0,
-      "step": 14469
-    },
-    {
-      "epoch": 1.1063325496492535,
-      "grad_norm": 0.030039310455322266,
-      "learning_rate": 0.0001999993968284053,
-      "loss": 46.0,
-      "step": 14470
-    },
-    {
-      "epoch": 1.1064090066326433,
-      "grad_norm": 0.0004251937789376825,
-      "learning_rate": 0.0001999993967449763,
-      "loss": 46.0,
-      "step": 14471
-    },
-    {
-      "epoch": 1.106485463616033,
-      "grad_norm": 0.0009537142468616366,
-      "learning_rate": 0.00019999939666154147,
-      "loss": 46.0,
-      "step": 14472
-    },
-    {
-      "epoch": 1.1065619205994228,
-      "grad_norm": 0.0007189179304987192,
-      "learning_rate": 0.00019999939657810095,
-      "loss": 46.0,
-      "step": 14473
-    },
-    {
-      "epoch": 1.1066383775828124,
-      "grad_norm": 0.0009143691859208047,
-      "learning_rate": 0.0001999993964946546,
-      "loss": 46.0,
-      "step": 14474
-    },
-    {
-      "epoch": 1.1067148345662021,
-      "grad_norm": 0.0005709800170734525,
-      "learning_rate": 0.0001999993964112025,
-      "loss": 46.0,
-      "step": 14475
-    },
-    {
-      "epoch": 1.1067912915495919,
-      "grad_norm": 0.001992768608033657,
-      "learning_rate": 0.00019999939632774462,
-      "loss": 46.0,
-      "step": 14476
-    },
-    {
-      "epoch": 1.1068677485329816,
-      "grad_norm": 0.0008526010788045824,
-      "learning_rate": 0.00019999939624428098,
-      "loss": 46.0,
-      "step": 14477
-    },
-    {
-      "epoch": 1.1069442055163714,
-      "grad_norm": 0.000938041543122381,
-      "learning_rate": 0.0001999993961608116,
-      "loss": 46.0,
-      "step": 14478
-    },
-    {
-      "epoch": 1.1070206624997612,
-      "grad_norm": 0.0035874040331691504,
-      "learning_rate": 0.0001999993960773364,
-      "loss": 46.0,
-      "step": 14479
-    },
-    {
-      "epoch": 1.1070971194831507,
-      "grad_norm": 0.0009905361803248525,
-      "learning_rate": 0.00019999939599385543,
-      "loss": 46.0,
-      "step": 14480
-    },
-    {
-      "epoch": 1.1071735764665405,
-      "grad_norm": 0.0015274417819455266,
-      "learning_rate": 0.00019999939591036872,
-      "loss": 46.0,
-      "step": 14481
-    },
-    {
-      "epoch": 1.1072500334499302,
-      "grad_norm": 0.0014322145143523812,
-      "learning_rate": 0.00019999939582687623,
-      "loss": 46.0,
-      "step": 14482
-    },
-    {
-      "epoch": 1.10732649043332,
-      "grad_norm": 0.0005529415211640298,
-      "learning_rate": 0.00019999939574337798,
-      "loss": 46.0,
-      "step": 14483
-    },
-    {
-      "epoch": 1.1074029474167097,
-      "grad_norm": 0.000556205166503787,
-      "learning_rate": 0.00019999939565987394,
-      "loss": 46.0,
-      "step": 14484
-    },
-    {
-      "epoch": 1.1074794044000993,
-      "grad_norm": 0.00035779859172180295,
-      "learning_rate": 0.00019999939557636417,
-      "loss": 46.0,
-      "step": 14485
-    },
-    {
-      "epoch": 1.107555861383489,
-      "grad_norm": 0.0037278644740581512,
-      "learning_rate": 0.00019999939549284862,
-      "loss": 46.0,
-      "step": 14486
-    },
-    {
-      "epoch": 1.1076323183668788,
-      "grad_norm": 0.0003472397511359304,
-      "learning_rate": 0.00019999939540932726,
-      "loss": 46.0,
-      "step": 14487
-    },
-    {
-      "epoch": 1.1077087753502686,
-      "grad_norm": 0.0032691857777535915,
-      "learning_rate": 0.00019999939532580014,
-      "loss": 46.0,
-      "step": 14488
-    },
-    {
-      "epoch": 1.1077852323336583,
-      "grad_norm": 0.0006711469613946974,
-      "learning_rate": 0.00019999939524226727,
-      "loss": 46.0,
-      "step": 14489
-    },
-    {
-      "epoch": 1.107861689317048,
-      "grad_norm": 0.0008515675435774028,
-      "learning_rate": 0.00019999939515872862,
-      "loss": 46.0,
-      "step": 14490
-    },
-    {
-      "epoch": 1.1079381463004376,
-      "grad_norm": 0.0011870659654960036,
-      "learning_rate": 0.0001999993950751842,
-      "loss": 46.0,
-      "step": 14491
-    },
-    {
-      "epoch": 1.1080146032838274,
-      "grad_norm": 0.0009953827830031514,
-      "learning_rate": 0.000199999394991634,
-      "loss": 46.0,
-      "step": 14492
-    },
-    {
-      "epoch": 1.1080910602672172,
-      "grad_norm": 0.0007408856763504446,
-      "learning_rate": 0.00019999939490807807,
-      "loss": 46.0,
-      "step": 14493
-    },
-    {
-      "epoch": 1.108167517250607,
-      "grad_norm": 0.0003907881327904761,
-      "learning_rate": 0.00019999939482451636,
-      "loss": 46.0,
-      "step": 14494
-    },
-    {
-      "epoch": 1.1082439742339967,
-      "grad_norm": 0.005370740778744221,
-      "learning_rate": 0.00019999939474094885,
-      "loss": 46.0,
-      "step": 14495
-    },
-    {
-      "epoch": 1.1083204312173862,
-      "grad_norm": 0.003085968317463994,
-      "learning_rate": 0.00019999939465737562,
-      "loss": 46.0,
-      "step": 14496
-    },
-    {
-      "epoch": 1.108396888200776,
-      "grad_norm": 0.0015167782548815012,
-      "learning_rate": 0.0001999993945737966,
-      "loss": 46.0,
-      "step": 14497
-    },
-    {
-      "epoch": 1.1084733451841657,
-      "grad_norm": 0.0006963883060961962,
-      "learning_rate": 0.00019999939449021179,
-      "loss": 46.0,
-      "step": 14498
-    },
-    {
-      "epoch": 1.1085498021675555,
-      "grad_norm": 0.0033322281669825315,
-      "learning_rate": 0.00019999939440662123,
-      "loss": 46.0,
-      "step": 14499
-    },
-    {
-      "epoch": 1.1086262591509453,
-      "grad_norm": 0.0028548554982990026,
-      "learning_rate": 0.0001999993943230249,
-      "loss": 46.0,
-      "step": 14500
-    },
-    {
-      "epoch": 1.108702716134335,
-      "grad_norm": 0.00474637420848012,
-      "learning_rate": 0.00019999939423942279,
-      "loss": 46.0,
-      "step": 14501
-    },
-    {
-      "epoch": 1.1087791731177246,
-      "grad_norm": 0.0037463712505996227,
-      "learning_rate": 0.00019999939415581491,
-      "loss": 46.0,
-      "step": 14502
-    },
-    {
-      "epoch": 1.1088556301011143,
-      "grad_norm": 0.00046203105011954904,
-      "learning_rate": 0.00019999939407220127,
-      "loss": 46.0,
-      "step": 14503
-    },
-    {
-      "epoch": 1.108932087084504,
-      "grad_norm": 0.0007735805120319128,
-      "learning_rate": 0.00019999939398858185,
-      "loss": 46.0,
-      "step": 14504
-    },
-    {
-      "epoch": 1.1090085440678938,
-      "grad_norm": 0.0018727260176092386,
-      "learning_rate": 0.00019999939390495666,
-      "loss": 46.0,
-      "step": 14505
-    },
-    {
-      "epoch": 1.1090850010512836,
-      "grad_norm": 0.0008407104178331792,
-      "learning_rate": 0.0001999993938213257,
-      "loss": 46.0,
-      "step": 14506
-    },
-    {
-      "epoch": 1.1091614580346731,
-      "grad_norm": 0.0007268865592777729,
-      "learning_rate": 0.00019999939373768899,
-      "loss": 46.0,
-      "step": 14507
-    },
-    {
-      "epoch": 1.109237915018063,
-      "grad_norm": 0.0015527295181527734,
-      "learning_rate": 0.00019999939365404647,
-      "loss": 46.0,
-      "step": 14508
-    },
-    {
-      "epoch": 1.1093143720014527,
-      "grad_norm": 0.00415242463350296,
-      "learning_rate": 0.00019999939357039824,
-      "loss": 46.0,
-      "step": 14509
-    },
-    {
-      "epoch": 1.1093908289848424,
-      "grad_norm": 0.0024282834492623806,
-      "learning_rate": 0.00019999939348674419,
-      "loss": 46.0,
-      "step": 14510
-    },
-    {
-      "epoch": 1.1094672859682322,
-      "grad_norm": 0.0038267658092081547,
-      "learning_rate": 0.00019999939340308438,
-      "loss": 46.0,
-      "step": 14511
-    },
-    {
-      "epoch": 1.109543742951622,
-      "grad_norm": 0.0013181344838812947,
-      "learning_rate": 0.0001999993933194188,
-      "loss": 46.0,
-      "step": 14512
-    },
-    {
-      "epoch": 1.1096201999350115,
-      "grad_norm": 0.0017209653742611408,
-      "learning_rate": 0.00019999939323574748,
-      "loss": 46.0,
-      "step": 14513
-    },
-    {
-      "epoch": 1.1096966569184012,
-      "grad_norm": 0.0015143873170018196,
-      "learning_rate": 0.00019999939315207038,
-      "loss": 46.0,
-      "step": 14514
-    },
-    {
-      "epoch": 1.109773113901791,
-      "grad_norm": 0.0033869887702167034,
-      "learning_rate": 0.00019999939306838748,
-      "loss": 46.0,
-      "step": 14515
-    },
-    {
-      "epoch": 1.1098495708851808,
-      "grad_norm": 0.07970468699932098,
-      "learning_rate": 0.00019999939298469884,
-      "loss": 46.0,
-      "step": 14516
-    },
-    {
-      "epoch": 1.1099260278685705,
-      "grad_norm": 0.0011768367839977145,
-      "learning_rate": 0.00019999939290100445,
-      "loss": 46.0,
-      "step": 14517
-    },
-    {
-      "epoch": 1.11000248485196,
-      "grad_norm": 0.0009448364726267755,
-      "learning_rate": 0.00019999939281730423,
-      "loss": 46.0,
-      "step": 14518
-    },
-    {
-      "epoch": 1.1100789418353498,
-      "grad_norm": 0.0035344057250767946,
-      "learning_rate": 0.0001999993927335983,
-      "loss": 46.0,
-      "step": 14519
-    },
-    {
-      "epoch": 1.1101553988187396,
-      "grad_norm": 0.0007960295770317316,
-      "learning_rate": 0.00019999939264988656,
-      "loss": 46.0,
-      "step": 14520
-    },
-    {
-      "epoch": 1.1102318558021294,
-      "grad_norm": 0.0033600088208913803,
-      "learning_rate": 0.00019999939256616907,
-      "loss": 46.0,
-      "step": 14521
-    },
-    {
-      "epoch": 1.1103083127855191,
-      "grad_norm": 0.0007676472305320203,
-      "learning_rate": 0.0001999993924824458,
-      "loss": 46.0,
-      "step": 14522
-    },
-    {
-      "epoch": 1.1103847697689089,
-      "grad_norm": 0.0005052352789789438,
-      "learning_rate": 0.00019999939239871676,
-      "loss": 46.0,
-      "step": 14523
-    },
-    {
-      "epoch": 1.1104612267522984,
-      "grad_norm": 0.0006214513559825718,
-      "learning_rate": 0.00019999939231498198,
-      "loss": 46.0,
-      "step": 14524
-    },
-    {
-      "epoch": 1.1105376837356882,
-      "grad_norm": 0.006707735825330019,
-      "learning_rate": 0.0001999993922312414,
-      "loss": 46.0,
-      "step": 14525
-    },
-    {
-      "epoch": 1.110614140719078,
-      "grad_norm": 0.002007142873480916,
-      "learning_rate": 0.00019999939214749505,
-      "loss": 46.0,
-      "step": 14526
-    },
-    {
-      "epoch": 1.1106905977024677,
-      "grad_norm": 0.002970101311802864,
-      "learning_rate": 0.00019999939206374293,
-      "loss": 46.0,
-      "step": 14527
-    },
-    {
-      "epoch": 1.1107670546858575,
-      "grad_norm": 0.004510781727731228,
-      "learning_rate": 0.00019999939197998506,
-      "loss": 46.0,
-      "step": 14528
-    },
-    {
-      "epoch": 1.110843511669247,
-      "grad_norm": 0.002100117038935423,
-      "learning_rate": 0.00019999939189622141,
-      "loss": 46.0,
-      "step": 14529
-    },
-    {
-      "epoch": 1.1109199686526368,
-      "grad_norm": 0.0030245014932006598,
-      "learning_rate": 0.000199999391812452,
-      "loss": 46.0,
-      "step": 14530
-    },
-    {
-      "epoch": 1.1109964256360265,
-      "grad_norm": 0.0015051182126626372,
-      "learning_rate": 0.00019999939172867678,
-      "loss": 46.0,
-      "step": 14531
-    },
-    {
-      "epoch": 1.1110728826194163,
-      "grad_norm": 0.0005448606098070741,
-      "learning_rate": 0.00019999939164489584,
-      "loss": 46.0,
-      "step": 14532
-    },
-    {
-      "epoch": 1.111149339602806,
-      "grad_norm": 0.0005574658280238509,
-      "learning_rate": 0.0001999993915611091,
-      "loss": 46.0,
-      "step": 14533
-    },
-    {
-      "epoch": 1.1112257965861958,
-      "grad_norm": 0.000571928801946342,
-      "learning_rate": 0.00019999939147731662,
-      "loss": 46.0,
-      "step": 14534
-    },
-    {
-      "epoch": 1.1113022535695853,
-      "grad_norm": 0.002085441257804632,
-      "learning_rate": 0.00019999939139351836,
-      "loss": 46.0,
-      "step": 14535
-    },
-    {
-      "epoch": 1.111378710552975,
-      "grad_norm": 0.0004395263676997274,
-      "learning_rate": 0.00019999939130971433,
-      "loss": 46.0,
-      "step": 14536
-    },
-    {
-      "epoch": 1.1114551675363649,
-      "grad_norm": 0.0010859448229894042,
-      "learning_rate": 0.00019999939122590453,
-      "loss": 46.0,
-      "step": 14537
-    },
-    {
-      "epoch": 1.1115316245197546,
-      "grad_norm": 0.0020721242763102055,
-      "learning_rate": 0.00019999939114208895,
-      "loss": 46.0,
-      "step": 14538
-    },
-    {
-      "epoch": 1.1116080815031444,
-      "grad_norm": 0.002610336057841778,
-      "learning_rate": 0.0001999993910582676,
-      "loss": 46.0,
-      "step": 14539
-    },
-    {
-      "epoch": 1.111684538486534,
-      "grad_norm": 0.0015839739935472608,
-      "learning_rate": 0.00019999939097444048,
-      "loss": 46.0,
-      "step": 14540
-    },
-    {
-      "epoch": 1.1117609954699237,
-      "grad_norm": 0.00048155890544876456,
-      "learning_rate": 0.00019999939089060758,
-      "loss": 46.0,
-      "step": 14541
-    },
-    {
-      "epoch": 1.1118374524533134,
-      "grad_norm": 0.001979676540941,
-      "learning_rate": 0.00019999939080676894,
-      "loss": 46.0,
-      "step": 14542
-    },
-    {
-      "epoch": 1.1119139094367032,
-      "grad_norm": 0.0007771134260110557,
-      "learning_rate": 0.0001999993907229245,
-      "loss": 46.0,
-      "step": 14543
-    },
-    {
-      "epoch": 1.111990366420093,
-      "grad_norm": 0.00046945270150899887,
-      "learning_rate": 0.00019999939063907433,
-      "loss": 46.0,
-      "step": 14544
-    },
-    {
-      "epoch": 1.1120668234034827,
-      "grad_norm": 0.0010695139644667506,
-      "learning_rate": 0.00019999939055521834,
-      "loss": 46.0,
-      "step": 14545
-    },
-    {
-      "epoch": 1.1121432803868723,
-      "grad_norm": 0.002555902348831296,
-      "learning_rate": 0.00019999939047135663,
-      "loss": 46.0,
-      "step": 14546
-    },
-    {
-      "epoch": 1.112219737370262,
-      "grad_norm": 0.0034649772569537163,
-      "learning_rate": 0.00019999939038748915,
-      "loss": 46.0,
-      "step": 14547
-    },
-    {
-      "epoch": 1.1122961943536518,
-      "grad_norm": 0.0009688063291832805,
-      "learning_rate": 0.00019999939030361586,
-      "loss": 46.0,
-      "step": 14548
-    },
-    {
-      "epoch": 1.1123726513370416,
-      "grad_norm": 0.0011843398679047823,
-      "learning_rate": 0.0001999993902197368,
-      "loss": 46.0,
-      "step": 14549
-    },
-    {
-      "epoch": 1.1124491083204313,
-      "grad_norm": 0.0027196116279810667,
-      "learning_rate": 0.00019999939013585203,
-      "loss": 46.0,
-      "step": 14550
-    },
-    {
-      "epoch": 1.1125255653038209,
-      "grad_norm": 0.004095969721674919,
-      "learning_rate": 0.00019999939005196145,
-      "loss": 46.0,
-      "step": 14551
-    },
-    {
-      "epoch": 1.1126020222872106,
-      "grad_norm": 0.0020090772304683924,
-      "learning_rate": 0.00019999938996806508,
-      "loss": 46.0,
-      "step": 14552
-    },
-    {
-      "epoch": 1.1126784792706004,
-      "grad_norm": 0.002264356706291437,
-      "learning_rate": 0.00019999938988416295,
-      "loss": 46.0,
-      "step": 14553
-    },
-    {
-      "epoch": 1.1127549362539901,
-      "grad_norm": 0.0006671007722616196,
-      "learning_rate": 0.0001999993898002551,
-      "loss": 46.0,
-      "step": 14554
-    },
-    {
-      "epoch": 1.11283139323738,
-      "grad_norm": 0.0007893171277828515,
-      "learning_rate": 0.0001999993897163414,
-      "loss": 46.0,
-      "step": 14555
-    },
-    {
-      "epoch": 1.1129078502207697,
-      "grad_norm": 0.0007784123881720006,
-      "learning_rate": 0.00019999938963242202,
-      "loss": 46.0,
-      "step": 14556
-    },
-    {
-      "epoch": 1.1129843072041592,
-      "grad_norm": 0.0009553050622344017,
-      "learning_rate": 0.0001999993895484968,
-      "loss": 46.0,
-      "step": 14557
-    },
-    {
-      "epoch": 1.113060764187549,
-      "grad_norm": 0.001965997740626335,
-      "learning_rate": 0.00019999938946456584,
-      "loss": 46.0,
-      "step": 14558
-    },
-    {
-      "epoch": 1.1131372211709387,
-      "grad_norm": 0.0005890214815735817,
-      "learning_rate": 0.00019999938938062913,
-      "loss": 46.0,
-      "step": 14559
-    },
-    {
-      "epoch": 1.1132136781543285,
-      "grad_norm": 0.007633336819708347,
-      "learning_rate": 0.00019999938929668662,
-      "loss": 46.0,
-      "step": 14560
-    },
-    {
-      "epoch": 1.1132901351377182,
-      "grad_norm": 0.0012276744237169623,
-      "learning_rate": 0.00019999938921273834,
-      "loss": 46.0,
-      "step": 14561
-    },
-    {
-      "epoch": 1.1133665921211078,
-      "grad_norm": 0.0003334197390358895,
-      "learning_rate": 0.0001999993891287843,
-      "loss": 46.0,
-      "step": 14562
-    },
-    {
-      "epoch": 1.1134430491044975,
-      "grad_norm": 0.0008329752599820495,
-      "learning_rate": 0.00019999938904482448,
-      "loss": 46.0,
-      "step": 14563
-    },
-    {
-      "epoch": 1.1135195060878873,
-      "grad_norm": 0.002505991840735078,
-      "learning_rate": 0.0001999993889608589,
-      "loss": 46.0,
-      "step": 14564
-    },
-    {
-      "epoch": 1.113595963071277,
-      "grad_norm": 0.004677955061197281,
-      "learning_rate": 0.00019999938887688755,
-      "loss": 46.0,
-      "step": 14565
-    },
-    {
-      "epoch": 1.1136724200546668,
-      "grad_norm": 0.0009772832272574306,
-      "learning_rate": 0.00019999938879291046,
-      "loss": 46.0,
-      "step": 14566
-    },
-    {
-      "epoch": 1.1137488770380564,
-      "grad_norm": 0.00270912260748446,
-      "learning_rate": 0.00019999938870892753,
-      "loss": 46.0,
-      "step": 14567
-    },
-    {
-      "epoch": 1.1138253340214461,
-      "grad_norm": 0.002612000796943903,
-      "learning_rate": 0.0001999993886249389,
-      "loss": 46.0,
-      "step": 14568
-    },
-    {
-      "epoch": 1.1139017910048359,
-      "grad_norm": 0.0009519907762296498,
-      "learning_rate": 0.00019999938854094447,
-      "loss": 46.0,
-      "step": 14569
-    },
-    {
-      "epoch": 1.1139782479882256,
-      "grad_norm": 0.0007542958483099937,
-      "learning_rate": 0.00019999938845694426,
-      "loss": 46.0,
-      "step": 14570
-    },
-    {
-      "epoch": 1.1140547049716154,
-      "grad_norm": 0.0011578942649066448,
-      "learning_rate": 0.0001999993883729383,
-      "loss": 46.0,
-      "step": 14571
-    },
-    {
-      "epoch": 1.1141311619550052,
-      "grad_norm": 0.002856897423043847,
-      "learning_rate": 0.00019999938828892656,
-      "loss": 46.0,
-      "step": 14572
-    },
-    {
-      "epoch": 1.1142076189383947,
-      "grad_norm": 0.000751401181332767,
-      "learning_rate": 0.00019999938820490908,
-      "loss": 46.0,
-      "step": 14573
-    },
-    {
-      "epoch": 1.1142840759217845,
-      "grad_norm": 0.000736706075258553,
-      "learning_rate": 0.0001999993881208858,
-      "loss": 46.0,
-      "step": 14574
-    },
-    {
-      "epoch": 1.1143605329051742,
-      "grad_norm": 0.006852846127003431,
-      "learning_rate": 0.00019999938803685674,
-      "loss": 46.0,
-      "step": 14575
-    },
-    {
-      "epoch": 1.114436989888564,
-      "grad_norm": 0.0009242200758308172,
-      "learning_rate": 0.0001999993879528219,
-      "loss": 46.0,
-      "step": 14576
-    },
-    {
-      "epoch": 1.1145134468719538,
-      "grad_norm": 0.0026618086267262697,
-      "learning_rate": 0.00019999938786878136,
-      "loss": 46.0,
-      "step": 14577
-    },
-    {
-      "epoch": 1.1145899038553433,
-      "grad_norm": 0.001850903034210205,
-      "learning_rate": 0.00019999938778473498,
-      "loss": 46.0,
-      "step": 14578
-    },
-    {
-      "epoch": 1.114666360838733,
-      "grad_norm": 0.0036684207152575254,
-      "learning_rate": 0.00019999938770068289,
-      "loss": 46.0,
-      "step": 14579
-    },
-    {
-      "epoch": 1.1147428178221228,
-      "grad_norm": 0.0013093024026602507,
-      "learning_rate": 0.000199999387616625,
-      "loss": 46.0,
-      "step": 14580
-    },
-    {
-      "epoch": 1.1148192748055126,
-      "grad_norm": 0.001117786392569542,
-      "learning_rate": 0.00019999938753256135,
-      "loss": 46.0,
-      "step": 14581
-    },
-    {
-      "epoch": 1.1148957317889023,
-      "grad_norm": 0.0009027719497680664,
-      "learning_rate": 0.0001999993874484919,
-      "loss": 46.0,
-      "step": 14582
-    },
-    {
-      "epoch": 1.1149721887722919,
-      "grad_norm": 0.0017418510979041457,
-      "learning_rate": 0.0001999993873644167,
-      "loss": 46.0,
-      "step": 14583
-    },
-    {
-      "epoch": 1.1150486457556816,
-      "grad_norm": 0.003381724003702402,
-      "learning_rate": 0.00019999938728033572,
-      "loss": 46.0,
-      "step": 14584
-    },
-    {
-      "epoch": 1.1151251027390714,
-      "grad_norm": 0.004773173946887255,
-      "learning_rate": 0.000199999387196249,
-      "loss": 46.0,
-      "step": 14585
-    },
-    {
-      "epoch": 1.1152015597224612,
-      "grad_norm": 0.0015445406315848231,
-      "learning_rate": 0.00019999938711215648,
-      "loss": 46.0,
-      "step": 14586
-    },
-    {
-      "epoch": 1.115278016705851,
-      "grad_norm": 0.0010909107513725758,
-      "learning_rate": 0.0001999993870280582,
-      "loss": 46.0,
-      "step": 14587
-    },
-    {
-      "epoch": 1.1153544736892407,
-      "grad_norm": 0.0048500304110348225,
-      "learning_rate": 0.00019999938694395417,
-      "loss": 46.0,
-      "step": 14588
-    },
-    {
-      "epoch": 1.1154309306726302,
-      "grad_norm": 0.0009982322808355093,
-      "learning_rate": 0.00019999938685984434,
-      "loss": 46.0,
-      "step": 14589
-    },
-    {
-      "epoch": 1.11550738765602,
-      "grad_norm": 0.0021196540910750628,
-      "learning_rate": 0.00019999938677572876,
-      "loss": 46.0,
-      "step": 14590
-    },
-    {
-      "epoch": 1.1155838446394097,
-      "grad_norm": 0.002493401989340782,
-      "learning_rate": 0.0001999993866916074,
-      "loss": 46.0,
-      "step": 14591
-    },
-    {
-      "epoch": 1.1156603016227995,
-      "grad_norm": 0.0035793951246887445,
-      "learning_rate": 0.0001999993866074803,
-      "loss": 46.0,
-      "step": 14592
-    },
-    {
-      "epoch": 1.1157367586061893,
-      "grad_norm": 0.001330891507677734,
-      "learning_rate": 0.00019999938652334737,
-      "loss": 46.0,
-      "step": 14593
-    },
-    {
-      "epoch": 1.1158132155895788,
-      "grad_norm": 0.004177140071988106,
-      "learning_rate": 0.00019999938643920872,
-      "loss": 46.0,
-      "step": 14594
-    },
-    {
-      "epoch": 1.1158896725729686,
-      "grad_norm": 0.0014069881290197372,
-      "learning_rate": 0.0001999993863550643,
-      "loss": 46.0,
-      "step": 14595
-    },
-    {
-      "epoch": 1.1159661295563583,
-      "grad_norm": 0.0009850873611867428,
-      "learning_rate": 0.0001999993862709141,
-      "loss": 46.0,
-      "step": 14596
-    },
-    {
-      "epoch": 1.116042586539748,
-      "grad_norm": 0.0017023030668497086,
-      "learning_rate": 0.00019999938618675813,
-      "loss": 46.0,
-      "step": 14597
-    },
-    {
-      "epoch": 1.1161190435231378,
-      "grad_norm": 0.0029525081627070904,
-      "learning_rate": 0.0001999993861025964,
-      "loss": 46.0,
-      "step": 14598
-    },
-    {
-      "epoch": 1.1161955005065276,
-      "grad_norm": 0.0005272942944429815,
-      "learning_rate": 0.00019999938601842886,
-      "loss": 46.0,
-      "step": 14599
-    },
-    {
-      "epoch": 1.1162719574899171,
-      "grad_norm": 0.0010052170837298036,
-      "learning_rate": 0.00019999938593425558,
-      "loss": 46.0,
-      "step": 14600
-    },
-    {
-      "epoch": 1.116348414473307,
-      "grad_norm": 0.001506473054178059,
-      "learning_rate": 0.00019999938585007655,
-      "loss": 46.0,
-      "step": 14601
-    },
-    {
-      "epoch": 1.1164248714566967,
-      "grad_norm": 0.001979702850803733,
-      "learning_rate": 0.00019999938576589172,
-      "loss": 46.0,
-      "step": 14602
-    },
-    {
-      "epoch": 1.1165013284400864,
-      "grad_norm": 0.0018389724427834153,
-      "learning_rate": 0.00019999938568170114,
-      "loss": 46.0,
-      "step": 14603
-    },
-    {
-      "epoch": 1.1165777854234762,
-      "grad_norm": 0.0006304451962932944,
-      "learning_rate": 0.00019999938559750477,
-      "loss": 46.0,
-      "step": 14604
-    },
-    {
-      "epoch": 1.1166542424068657,
-      "grad_norm": 0.0008731254492886364,
-      "learning_rate": 0.00019999938551330267,
-      "loss": 46.0,
-      "step": 14605
-    },
-    {
-      "epoch": 1.1167306993902555,
-      "grad_norm": 0.0010184376733377576,
-      "learning_rate": 0.00019999938542909475,
-      "loss": 46.0,
-      "step": 14606
-    },
-    {
-      "epoch": 1.1168071563736452,
-      "grad_norm": 0.0070178876630961895,
-      "learning_rate": 0.00019999938534488108,
-      "loss": 46.0,
-      "step": 14607
-    },
-    {
-      "epoch": 1.116883613357035,
-      "grad_norm": 0.0007276375545188785,
-      "learning_rate": 0.00019999938526066164,
-      "loss": 46.0,
-      "step": 14608
-    },
-    {
-      "epoch": 1.1169600703404248,
-      "grad_norm": 0.0014580200659111142,
-      "learning_rate": 0.00019999938517643645,
-      "loss": 46.0,
-      "step": 14609
-    },
-    {
-      "epoch": 1.1170365273238145,
-      "grad_norm": 0.0038676836993545294,
-      "learning_rate": 0.00019999938509220549,
-      "loss": 46.0,
-      "step": 14610
-    },
-    {
-      "epoch": 1.117112984307204,
-      "grad_norm": 0.019074708223342896,
-      "learning_rate": 0.00019999938500796872,
-      "loss": 46.0,
-      "step": 14611
-    },
-    {
-      "epoch": 1.1171894412905938,
-      "grad_norm": 0.001472771167755127,
-      "learning_rate": 0.00019999938492372624,
-      "loss": 46.0,
-      "step": 14612
-    },
-    {
-      "epoch": 1.1172658982739836,
-      "grad_norm": 0.0005961598362773657,
-      "learning_rate": 0.00019999938483947793,
-      "loss": 46.0,
-      "step": 14613
-    },
-    {
-      "epoch": 1.1173423552573734,
-      "grad_norm": 0.003704892238602042,
-      "learning_rate": 0.00019999938475522388,
-      "loss": 46.0,
-      "step": 14614
-    },
-    {
-      "epoch": 1.1174188122407631,
-      "grad_norm": 0.001822069170884788,
-      "learning_rate": 0.00019999938467096408,
-      "loss": 46.0,
-      "step": 14615
-    },
-    {
-      "epoch": 1.1174952692241527,
-      "grad_norm": 0.0007231037016026676,
-      "learning_rate": 0.0001999993845866985,
-      "loss": 46.0,
-      "step": 14616
-    },
-    {
-      "epoch": 1.1175717262075424,
-      "grad_norm": 0.0031787478365004063,
-      "learning_rate": 0.00019999938450242712,
-      "loss": 46.0,
-      "step": 14617
-    },
-    {
-      "epoch": 1.1176481831909322,
-      "grad_norm": 0.0012274907203391194,
-      "learning_rate": 0.00019999938441814998,
-      "loss": 46.0,
-      "step": 14618
-    },
-    {
-      "epoch": 1.117724640174322,
-      "grad_norm": 0.0009310300811193883,
-      "learning_rate": 0.00019999938433386708,
-      "loss": 46.0,
-      "step": 14619
-    },
-    {
-      "epoch": 1.1178010971577117,
-      "grad_norm": 0.020797275006771088,
-      "learning_rate": 0.00019999938424957844,
-      "loss": 46.0,
-      "step": 14620
-    },
-    {
-      "epoch": 1.1178775541411015,
-      "grad_norm": 0.005265621468424797,
-      "learning_rate": 0.000199999384165284,
-      "loss": 46.0,
-      "step": 14621
-    },
-    {
-      "epoch": 1.117954011124491,
-      "grad_norm": 0.0006121473270468414,
-      "learning_rate": 0.00019999938408098378,
-      "loss": 46.0,
-      "step": 14622
-    },
-    {
-      "epoch": 1.1180304681078808,
-      "grad_norm": 0.0018558199517428875,
-      "learning_rate": 0.0001999993839966778,
-      "loss": 46.0,
-      "step": 14623
-    },
-    {
-      "epoch": 1.1181069250912705,
-      "grad_norm": 0.0022252004127949476,
-      "learning_rate": 0.00019999938391236606,
-      "loss": 46.0,
-      "step": 14624
-    },
-    {
-      "epoch": 1.1181833820746603,
-      "grad_norm": 0.0009050408843904734,
-      "learning_rate": 0.00019999938382804855,
-      "loss": 46.0,
-      "step": 14625
-    },
-    {
-      "epoch": 1.11825983905805,
-      "grad_norm": 0.0010619487147778273,
-      "learning_rate": 0.00019999938374372527,
-      "loss": 46.0,
-      "step": 14626
-    },
-    {
-      "epoch": 1.1183362960414396,
-      "grad_norm": 0.0008505067089572549,
-      "learning_rate": 0.0001999993836593962,
-      "loss": 46.0,
-      "step": 14627
-    },
-    {
-      "epoch": 1.1184127530248293,
-      "grad_norm": 0.007907591760158539,
-      "learning_rate": 0.00019999938357506138,
-      "loss": 46.0,
-      "step": 14628
-    },
-    {
-      "epoch": 1.118489210008219,
-      "grad_norm": 0.0020460919477045536,
-      "learning_rate": 0.00019999938349072078,
-      "loss": 46.0,
-      "step": 14629
-    },
-    {
-      "epoch": 1.1185656669916089,
-      "grad_norm": 0.0015377800446003675,
-      "learning_rate": 0.00019999938340637443,
-      "loss": 46.0,
-      "step": 14630
-    },
-    {
-      "epoch": 1.1186421239749986,
-      "grad_norm": 0.0008151123765856028,
-      "learning_rate": 0.0001999993833220223,
-      "loss": 46.0,
-      "step": 14631
-    },
-    {
-      "epoch": 1.1187185809583884,
-      "grad_norm": 0.001854747999459505,
-      "learning_rate": 0.00019999938323766442,
-      "loss": 46.0,
-      "step": 14632
-    },
-    {
-      "epoch": 1.118795037941778,
-      "grad_norm": 0.0012854618253186345,
-      "learning_rate": 0.00019999938315330075,
-      "loss": 46.0,
-      "step": 14633
-    },
-    {
-      "epoch": 1.1188714949251677,
-      "grad_norm": 0.000622910913079977,
-      "learning_rate": 0.00019999938306893128,
-      "loss": 46.0,
-      "step": 14634
-    },
-    {
-      "epoch": 1.1189479519085574,
-      "grad_norm": 0.0028844457119703293,
-      "learning_rate": 0.00019999938298455606,
-      "loss": 46.0,
-      "step": 14635
-    },
-    {
-      "epoch": 1.1190244088919472,
-      "grad_norm": 0.004551828373223543,
-      "learning_rate": 0.0001999993829001751,
-      "loss": 46.0,
-      "step": 14636
-    },
-    {
-      "epoch": 1.119100865875337,
-      "grad_norm": 0.0008320729248225689,
-      "learning_rate": 0.00019999938281578837,
-      "loss": 46.0,
-      "step": 14637
-    },
-    {
-      "epoch": 1.1191773228587265,
-      "grad_norm": 0.00041060420335270464,
-      "learning_rate": 0.00019999938273139583,
-      "loss": 46.0,
-      "step": 14638
-    },
-    {
-      "epoch": 1.1192537798421163,
-      "grad_norm": 0.002745057689025998,
-      "learning_rate": 0.00019999938264699755,
-      "loss": 46.0,
-      "step": 14639
-    },
-    {
-      "epoch": 1.119330236825506,
-      "grad_norm": 0.0022486604284495115,
-      "learning_rate": 0.0001999993825625935,
-      "loss": 46.0,
-      "step": 14640
-    },
-    {
-      "epoch": 1.1194066938088958,
-      "grad_norm": 0.0021962022874504328,
-      "learning_rate": 0.00019999938247818367,
-      "loss": 46.0,
-      "step": 14641
-    },
-    {
-      "epoch": 1.1194831507922856,
-      "grad_norm": 0.0005597159615717828,
-      "learning_rate": 0.00019999938239376807,
-      "loss": 46.0,
-      "step": 14642
-    },
-    {
-      "epoch": 1.1195596077756753,
-      "grad_norm": 0.0004508441488724202,
-      "learning_rate": 0.00019999938230934672,
-      "loss": 46.0,
-      "step": 14643
-    },
-    {
-      "epoch": 1.1196360647590649,
-      "grad_norm": 0.0014569774502888322,
-      "learning_rate": 0.0001999993822249196,
-      "loss": 46.0,
-      "step": 14644
-    },
-    {
-      "epoch": 1.1197125217424546,
-      "grad_norm": 0.008597142994403839,
-      "learning_rate": 0.00019999938214048667,
-      "loss": 46.0,
-      "step": 14645
-    },
-    {
-      "epoch": 1.1197889787258444,
-      "grad_norm": 0.004099604208022356,
-      "learning_rate": 0.000199999382056048,
-      "loss": 46.0,
-      "step": 14646
-    },
-    {
-      "epoch": 1.1198654357092341,
-      "grad_norm": 0.0025996100157499313,
-      "learning_rate": 0.00019999938197160354,
-      "loss": 46.0,
-      "step": 14647
-    },
-    {
-      "epoch": 1.119941892692624,
-      "grad_norm": 0.0016507940599694848,
-      "learning_rate": 0.00019999938188715335,
-      "loss": 46.0,
-      "step": 14648
-    },
-    {
-      "epoch": 1.1200183496760134,
-      "grad_norm": 0.0009958160808309913,
-      "learning_rate": 0.00019999938180269736,
-      "loss": 46.0,
-      "step": 14649
-    },
-    {
-      "epoch": 1.1200948066594032,
-      "grad_norm": 0.0006161693017929792,
-      "learning_rate": 0.0001999993817182356,
-      "loss": 46.0,
-      "step": 14650
-    },
-    {
-      "epoch": 1.120171263642793,
-      "grad_norm": 0.0014730727998539805,
-      "learning_rate": 0.0001999993816337681,
-      "loss": 46.0,
-      "step": 14651
-    },
-    {
-      "epoch": 1.1202477206261827,
-      "grad_norm": 0.0004374423297122121,
-      "learning_rate": 0.0001999993815492948,
-      "loss": 46.0,
-      "step": 14652
-    },
-    {
-      "epoch": 1.1203241776095725,
-      "grad_norm": 0.001653553918004036,
-      "learning_rate": 0.00019999938146481573,
-      "loss": 46.0,
-      "step": 14653
-    },
-    {
-      "epoch": 1.1204006345929622,
-      "grad_norm": 0.0005684678326360881,
-      "learning_rate": 0.00019999938138033093,
-      "loss": 46.0,
-      "step": 14654
-    },
-    {
-      "epoch": 1.1204770915763518,
-      "grad_norm": 0.004800945520401001,
-      "learning_rate": 0.00019999938129584033,
-      "loss": 46.0,
-      "step": 14655
-    },
-    {
-      "epoch": 1.1205535485597415,
-      "grad_norm": 0.0007770388037897646,
-      "learning_rate": 0.00019999938121134395,
-      "loss": 46.0,
-      "step": 14656
-    },
-    {
-      "epoch": 1.1206300055431313,
-      "grad_norm": 0.0005487881717272103,
-      "learning_rate": 0.00019999938112684183,
-      "loss": 46.0,
-      "step": 14657
-    },
-    {
-      "epoch": 1.120706462526521,
-      "grad_norm": 0.0012454709503799677,
-      "learning_rate": 0.0001999993810423339,
-      "loss": 46.0,
-      "step": 14658
-    },
-    {
-      "epoch": 1.1207829195099108,
-      "grad_norm": 0.0006934825796633959,
-      "learning_rate": 0.00019999938095782022,
-      "loss": 46.0,
-      "step": 14659
-    },
-    {
-      "epoch": 1.1208593764933004,
-      "grad_norm": 0.0016738050617277622,
-      "learning_rate": 0.0001999993808733008,
-      "loss": 46.0,
-      "step": 14660
-    },
-    {
-      "epoch": 1.1209358334766901,
-      "grad_norm": 0.0009980673203244805,
-      "learning_rate": 0.0001999993807887756,
-      "loss": 46.0,
-      "step": 14661
-    },
-    {
-      "epoch": 1.1210122904600799,
-      "grad_norm": 0.0011886612046509981,
-      "learning_rate": 0.0001999993807042446,
-      "loss": 46.0,
-      "step": 14662
-    },
-    {
-      "epoch": 1.1210887474434696,
-      "grad_norm": 0.0004915367462672293,
-      "learning_rate": 0.00019999938061970784,
-      "loss": 46.0,
-      "step": 14663
-    },
-    {
-      "epoch": 1.1211652044268594,
-      "grad_norm": 0.0027914049569517374,
-      "learning_rate": 0.00019999938053516533,
-      "loss": 46.0,
-      "step": 14664
-    },
-    {
-      "epoch": 1.1212416614102492,
-      "grad_norm": 0.0018157025333493948,
-      "learning_rate": 0.00019999938045061705,
-      "loss": 46.0,
-      "step": 14665
-    },
-    {
-      "epoch": 1.1213181183936387,
-      "grad_norm": 0.001068687648512423,
-      "learning_rate": 0.00019999938036606297,
-      "loss": 46.0,
-      "step": 14666
-    },
-    {
-      "epoch": 1.1213945753770285,
-      "grad_norm": 0.0029094095807522535,
-      "learning_rate": 0.00019999938028150314,
-      "loss": 46.0,
-      "step": 14667
-    },
-    {
-      "epoch": 1.1214710323604182,
-      "grad_norm": 0.003869979875162244,
-      "learning_rate": 0.00019999938019693754,
-      "loss": 46.0,
-      "step": 14668
-    },
-    {
-      "epoch": 1.121547489343808,
-      "grad_norm": 0.0025615976192057133,
-      "learning_rate": 0.00019999938011236616,
-      "loss": 46.0,
-      "step": 14669
-    },
-    {
-      "epoch": 1.1216239463271978,
-      "grad_norm": 0.0006732771289534867,
-      "learning_rate": 0.00019999938002778904,
-      "loss": 46.0,
-      "step": 14670
-    },
-    {
-      "epoch": 1.1217004033105873,
-      "grad_norm": 0.0014956679660826921,
-      "learning_rate": 0.00019999937994320612,
-      "loss": 46.0,
-      "step": 14671
-    },
-    {
-      "epoch": 1.121776860293977,
-      "grad_norm": 0.005253876093775034,
-      "learning_rate": 0.00019999937985861746,
-      "loss": 46.0,
-      "step": 14672
-    },
-    {
-      "epoch": 1.1218533172773668,
-      "grad_norm": 0.003075046231970191,
-      "learning_rate": 0.000199999379774023,
-      "loss": 46.0,
-      "step": 14673
-    },
-    {
-      "epoch": 1.1219297742607566,
-      "grad_norm": 0.0010864182841032743,
-      "learning_rate": 0.0001999993796894228,
-      "loss": 46.0,
-      "step": 14674
-    },
-    {
-      "epoch": 1.1220062312441463,
-      "grad_norm": 0.00045925957965664566,
-      "learning_rate": 0.0001999993796048168,
-      "loss": 46.0,
-      "step": 14675
-    },
-    {
-      "epoch": 1.122082688227536,
-      "grad_norm": 0.004406673833727837,
-      "learning_rate": 0.00019999937952020505,
-      "loss": 46.0,
-      "step": 14676
-    },
-    {
-      "epoch": 1.1221591452109256,
-      "grad_norm": 0.0021077890414744616,
-      "learning_rate": 0.00019999937943558752,
-      "loss": 46.0,
-      "step": 14677
-    },
-    {
-      "epoch": 1.1222356021943154,
-      "grad_norm": 0.0007649871986359358,
-      "learning_rate": 0.00019999937935096424,
-      "loss": 46.0,
-      "step": 14678
-    },
-    {
-      "epoch": 1.1223120591777052,
-      "grad_norm": 0.004702277481555939,
-      "learning_rate": 0.00019999937926633516,
-      "loss": 46.0,
-      "step": 14679
-    },
-    {
-      "epoch": 1.122388516161095,
-      "grad_norm": 0.0013283280422911048,
-      "learning_rate": 0.00019999937918170033,
-      "loss": 46.0,
-      "step": 14680
-    },
-    {
-      "epoch": 1.1224649731444847,
-      "grad_norm": 0.0005173981771804392,
-      "learning_rate": 0.00019999937909705973,
-      "loss": 46.0,
-      "step": 14681
-    },
-    {
-      "epoch": 1.1225414301278742,
-      "grad_norm": 0.005242025945335627,
-      "learning_rate": 0.00019999937901241338,
-      "loss": 46.0,
-      "step": 14682
-    },
-    {
-      "epoch": 1.122617887111264,
-      "grad_norm": 0.0023013725876808167,
-      "learning_rate": 0.0001999993789277612,
-      "loss": 46.0,
-      "step": 14683
-    },
-    {
-      "epoch": 1.1226943440946537,
-      "grad_norm": 0.00021133584959898144,
-      "learning_rate": 0.00019999937884310328,
-      "loss": 46.0,
-      "step": 14684
-    },
-    {
-      "epoch": 1.1227708010780435,
-      "grad_norm": 0.03928922489285469,
-      "learning_rate": 0.00019999937875843964,
-      "loss": 46.0,
-      "step": 14685
-    },
-    {
-      "epoch": 1.1228472580614333,
-      "grad_norm": 0.0007286184700205922,
-      "learning_rate": 0.00019999937867377018,
-      "loss": 46.0,
-      "step": 14686
-    },
-    {
-      "epoch": 1.122923715044823,
-      "grad_norm": 0.0019904032815247774,
-      "learning_rate": 0.00019999937858909496,
-      "loss": 46.0,
-      "step": 14687
-    },
-    {
-      "epoch": 1.1230001720282126,
-      "grad_norm": 0.0006558075547218323,
-      "learning_rate": 0.00019999937850441398,
-      "loss": 46.0,
-      "step": 14688
-    },
-    {
-      "epoch": 1.1230766290116023,
-      "grad_norm": 0.011617771349847317,
-      "learning_rate": 0.0001999993784197272,
-      "loss": 46.0,
-      "step": 14689
-    },
-    {
-      "epoch": 1.123153085994992,
-      "grad_norm": 0.0013864102074876428,
-      "learning_rate": 0.00019999937833503468,
-      "loss": 46.0,
-      "step": 14690
-    },
-    {
-      "epoch": 1.1232295429783818,
-      "grad_norm": 0.0004177753289695829,
-      "learning_rate": 0.0001999993782503364,
-      "loss": 46.0,
-      "step": 14691
-    },
-    {
-      "epoch": 1.1233059999617716,
-      "grad_norm": 0.0035847309045493603,
-      "learning_rate": 0.00019999937816563232,
-      "loss": 46.0,
-      "step": 14692
-    },
-    {
-      "epoch": 1.1233824569451611,
-      "grad_norm": 0.0009797184029594064,
-      "learning_rate": 0.0001999993780809225,
-      "loss": 46.0,
-      "step": 14693
-    },
-    {
-      "epoch": 1.123458913928551,
-      "grad_norm": 0.0007893379661254585,
-      "learning_rate": 0.00019999937799620687,
-      "loss": 46.0,
-      "step": 14694
-    },
-    {
-      "epoch": 1.1235353709119407,
-      "grad_norm": 0.0011319953482598066,
-      "learning_rate": 0.00019999937791148552,
-      "loss": 46.0,
-      "step": 14695
-    },
-    {
-      "epoch": 1.1236118278953304,
-      "grad_norm": 0.009061713702976704,
-      "learning_rate": 0.00019999937782675837,
-      "loss": 46.0,
-      "step": 14696
-    },
-    {
-      "epoch": 1.1236882848787202,
-      "grad_norm": 0.001112721860408783,
-      "learning_rate": 0.00019999937774202545,
-      "loss": 46.0,
-      "step": 14697
-    },
-    {
-      "epoch": 1.1237647418621097,
-      "grad_norm": 0.0010878619505092502,
-      "learning_rate": 0.00019999937765728678,
-      "loss": 46.0,
-      "step": 14698
-    },
-    {
-      "epoch": 1.1238411988454995,
-      "grad_norm": 0.0012675655307248235,
-      "learning_rate": 0.00019999937757254232,
-      "loss": 46.0,
-      "step": 14699
-    },
-    {
-      "epoch": 1.1239176558288893,
-      "grad_norm": 0.002784379757940769,
-      "learning_rate": 0.0001999993774877921,
-      "loss": 46.0,
-      "step": 14700
-    },
-    {
-      "epoch": 1.123994112812279,
-      "grad_norm": 0.0014011251041665673,
-      "learning_rate": 0.00019999937740303612,
-      "loss": 46.0,
-      "step": 14701
-    },
-    {
-      "epoch": 1.1240705697956688,
-      "grad_norm": 0.001019007759168744,
-      "learning_rate": 0.00019999937731827436,
-      "loss": 46.0,
-      "step": 14702
-    },
-    {
-      "epoch": 1.1241470267790585,
-      "grad_norm": 0.005811325740069151,
-      "learning_rate": 0.00019999937723350682,
-      "loss": 46.0,
-      "step": 14703
-    },
-    {
-      "epoch": 1.124223483762448,
-      "grad_norm": 0.010807032696902752,
-      "learning_rate": 0.00019999937714873354,
-      "loss": 46.0,
-      "step": 14704
-    },
-    {
-      "epoch": 1.1242999407458378,
-      "grad_norm": 0.012497744522988796,
-      "learning_rate": 0.00019999937706395446,
-      "loss": 46.0,
-      "step": 14705
-    },
-    {
-      "epoch": 1.1243763977292276,
-      "grad_norm": 0.00172276864759624,
-      "learning_rate": 0.0001999993769791696,
-      "loss": 46.0,
-      "step": 14706
-    },
-    {
-      "epoch": 1.1244528547126174,
-      "grad_norm": 0.0007320116274058819,
-      "learning_rate": 0.000199999376894379,
-      "loss": 46.0,
-      "step": 14707
-    },
-    {
-      "epoch": 1.1245293116960071,
-      "grad_norm": 0.00480728829279542,
-      "learning_rate": 0.00019999937680958263,
-      "loss": 46.0,
-      "step": 14708
-    },
-    {
-      "epoch": 1.1246057686793967,
-      "grad_norm": 0.0018890599021688104,
-      "learning_rate": 0.0001999993767247805,
-      "loss": 46.0,
-      "step": 14709
-    },
-    {
-      "epoch": 1.1246822256627864,
-      "grad_norm": 0.0009019332937896252,
-      "learning_rate": 0.00019999937663997257,
-      "loss": 46.0,
-      "step": 14710
-    },
-    {
-      "epoch": 1.1247586826461762,
-      "grad_norm": 0.0010031972778961062,
-      "learning_rate": 0.00019999937655515887,
-      "loss": 46.0,
-      "step": 14711
-    },
-    {
-      "epoch": 1.124835139629566,
-      "grad_norm": 0.001480329898186028,
-      "learning_rate": 0.00019999937647033943,
-      "loss": 46.0,
-      "step": 14712
-    },
-    {
-      "epoch": 1.1249115966129557,
-      "grad_norm": 0.000891718256752938,
-      "learning_rate": 0.00019999937638551422,
-      "loss": 46.0,
-      "step": 14713
-    },
-    {
-      "epoch": 1.1249880535963452,
-      "grad_norm": 0.006402339320629835,
-      "learning_rate": 0.0001999993763006832,
-      "loss": 46.0,
-      "step": 14714
-    },
-    {
-      "epoch": 1.125064510579735,
-      "grad_norm": 0.0023910673335194588,
-      "learning_rate": 0.00019999937621584645,
-      "loss": 46.0,
-      "step": 14715
-    },
-    {
-      "epoch": 1.1251409675631248,
-      "grad_norm": 0.0026635779067873955,
-      "learning_rate": 0.00019999937613100391,
-      "loss": 46.0,
-      "step": 14716
-    },
-    {
-      "epoch": 1.1252174245465145,
-      "grad_norm": 0.0008241531904786825,
-      "learning_rate": 0.0001999993760461556,
-      "loss": 46.0,
-      "step": 14717
-    },
-    {
-      "epoch": 1.1252938815299043,
-      "grad_norm": 0.006887366995215416,
-      "learning_rate": 0.00019999937596130155,
-      "loss": 46.0,
-      "step": 14718
-    },
-    {
-      "epoch": 1.125370338513294,
-      "grad_norm": 0.007253400515764952,
-      "learning_rate": 0.0001999993758764417,
-      "loss": 46.0,
-      "step": 14719
-    },
-    {
-      "epoch": 1.1254467954966836,
-      "grad_norm": 0.004404946696013212,
-      "learning_rate": 0.0001999993757915761,
-      "loss": 46.0,
-      "step": 14720
-    },
-    {
-      "epoch": 1.1255232524800733,
-      "grad_norm": 0.0009496635757386684,
-      "learning_rate": 0.00019999937570670473,
-      "loss": 46.0,
-      "step": 14721
-    },
-    {
-      "epoch": 1.125599709463463,
-      "grad_norm": 0.002452364657074213,
-      "learning_rate": 0.00019999937562182758,
-      "loss": 46.0,
-      "step": 14722
-    },
-    {
-      "epoch": 1.1256761664468529,
-      "grad_norm": 0.0047640809789299965,
-      "learning_rate": 0.00019999937553694466,
-      "loss": 46.0,
-      "step": 14723
-    },
-    {
-      "epoch": 1.1257526234302426,
-      "grad_norm": 0.0006687306449748576,
-      "learning_rate": 0.000199999375452056,
-      "loss": 46.0,
-      "step": 14724
-    },
-    {
-      "epoch": 1.1258290804136322,
-      "grad_norm": 0.0014523823047056794,
-      "learning_rate": 0.0001999993753671615,
-      "loss": 46.0,
-      "step": 14725
-    },
-    {
-      "epoch": 1.125905537397022,
-      "grad_norm": 0.0005080315750092268,
-      "learning_rate": 0.00019999937528226126,
-      "loss": 46.0,
-      "step": 14726
-    },
-    {
-      "epoch": 1.1259819943804117,
-      "grad_norm": 0.0014898290392011404,
-      "learning_rate": 0.00019999937519735527,
-      "loss": 46.0,
-      "step": 14727
-    },
-    {
-      "epoch": 1.1260584513638014,
-      "grad_norm": 0.0032106288708746433,
-      "learning_rate": 0.00019999937511244352,
-      "loss": 46.0,
-      "step": 14728
-    },
-    {
-      "epoch": 1.1261349083471912,
-      "grad_norm": 0.0011365648824721575,
-      "learning_rate": 0.00019999937502752598,
-      "loss": 46.0,
-      "step": 14729
-    },
-    {
-      "epoch": 1.126211365330581,
-      "grad_norm": 0.0010857610031962395,
-      "learning_rate": 0.0001999993749426027,
-      "loss": 46.0,
-      "step": 14730
-    },
-    {
-      "epoch": 1.1262878223139705,
-      "grad_norm": 0.0010345472255721688,
-      "learning_rate": 0.00019999937485767362,
-      "loss": 46.0,
-      "step": 14731
-    },
-    {
-      "epoch": 1.1263642792973603,
-      "grad_norm": 0.0025925934314727783,
-      "learning_rate": 0.00019999937477273877,
-      "loss": 46.0,
-      "step": 14732
-    },
-    {
-      "epoch": 1.12644073628075,
-      "grad_norm": 0.003173497272655368,
-      "learning_rate": 0.00019999937468779817,
-      "loss": 46.0,
-      "step": 14733
-    },
-    {
-      "epoch": 1.1265171932641398,
-      "grad_norm": 0.004784189164638519,
-      "learning_rate": 0.00019999937460285177,
-      "loss": 46.0,
-      "step": 14734
-    },
-    {
-      "epoch": 1.1265936502475296,
-      "grad_norm": 0.0006555349100381136,
-      "learning_rate": 0.00019999937451789962,
-      "loss": 46.0,
-      "step": 14735
-    },
-    {
-      "epoch": 1.126670107230919,
-      "grad_norm": 0.0005574798560701311,
-      "learning_rate": 0.00019999937443294173,
-      "loss": 46.0,
-      "step": 14736
-    },
-    {
-      "epoch": 1.1267465642143089,
-      "grad_norm": 0.0004293904057703912,
-      "learning_rate": 0.000199999374347978,
-      "loss": 46.0,
-      "step": 14737
-    },
-    {
-      "epoch": 1.1268230211976986,
-      "grad_norm": 0.002910550683736801,
-      "learning_rate": 0.00019999937426300857,
-      "loss": 46.0,
-      "step": 14738
-    },
-    {
-      "epoch": 1.1268994781810884,
-      "grad_norm": 0.005198165308684111,
-      "learning_rate": 0.0001999993741780333,
-      "loss": 46.0,
-      "step": 14739
-    },
-    {
-      "epoch": 1.1269759351644781,
-      "grad_norm": 0.0028310639318078756,
-      "learning_rate": 0.00019999937409305235,
-      "loss": 46.0,
-      "step": 14740
-    },
-    {
-      "epoch": 1.127052392147868,
-      "grad_norm": 0.0007997811189852655,
-      "learning_rate": 0.00019999937400806556,
-      "loss": 46.0,
-      "step": 14741
-    },
-    {
-      "epoch": 1.1271288491312574,
-      "grad_norm": 0.001436564140021801,
-      "learning_rate": 0.00019999937392307305,
-      "loss": 46.0,
-      "step": 14742
-    },
-    {
-      "epoch": 1.1272053061146472,
-      "grad_norm": 0.01144907996058464,
-      "learning_rate": 0.00019999937383807472,
-      "loss": 46.0,
-      "step": 14743
-    },
-    {
-      "epoch": 1.127281763098037,
-      "grad_norm": 0.0005288422107696533,
-      "learning_rate": 0.00019999937375307064,
-      "loss": 46.0,
-      "step": 14744
-    },
-    {
-      "epoch": 1.1273582200814267,
-      "grad_norm": 0.004587346687912941,
-      "learning_rate": 0.00019999937366806082,
-      "loss": 46.0,
-      "step": 14745
-    },
-    {
-      "epoch": 1.1274346770648165,
-      "grad_norm": 0.0006087794899940491,
-      "learning_rate": 0.0001999993735830452,
-      "loss": 46.0,
-      "step": 14746
-    },
-    {
-      "epoch": 1.127511134048206,
-      "grad_norm": 0.0010498225456103683,
-      "learning_rate": 0.0001999993734980238,
-      "loss": 46.0,
-      "step": 14747
-    },
-    {
-      "epoch": 1.1275875910315958,
-      "grad_norm": 0.0013251262716948986,
-      "learning_rate": 0.00019999937341299667,
-      "loss": 46.0,
-      "step": 14748
-    },
-    {
-      "epoch": 1.1276640480149855,
-      "grad_norm": 0.001476874342188239,
-      "learning_rate": 0.00019999937332796375,
-      "loss": 46.0,
-      "step": 14749
-    },
-    {
-      "epoch": 1.1277405049983753,
-      "grad_norm": 0.002560973633080721,
-      "learning_rate": 0.00019999937324292503,
-      "loss": 46.0,
-      "step": 14750
-    },
-    {
-      "epoch": 1.127816961981765,
-      "grad_norm": 0.009889591485261917,
-      "learning_rate": 0.00019999937315788057,
-      "loss": 46.0,
-      "step": 14751
-    },
-    {
-      "epoch": 1.1278934189651548,
-      "grad_norm": 0.0006871015066280961,
-      "learning_rate": 0.00019999937307283035,
-      "loss": 46.0,
-      "step": 14752
-    },
-    {
-      "epoch": 1.1279698759485444,
-      "grad_norm": 0.007508058566600084,
-      "learning_rate": 0.00019999937298777437,
-      "loss": 46.0,
-      "step": 14753
-    },
-    {
-      "epoch": 1.1280463329319341,
-      "grad_norm": 0.000655336189083755,
-      "learning_rate": 0.00019999937290271258,
-      "loss": 46.0,
-      "step": 14754
-    },
-    {
-      "epoch": 1.1281227899153239,
-      "grad_norm": 0.0004830614780075848,
-      "learning_rate": 0.00019999937281764502,
-      "loss": 46.0,
-      "step": 14755
-    },
-    {
-      "epoch": 1.1281992468987136,
-      "grad_norm": 0.0006952632684260607,
-      "learning_rate": 0.00019999937273257174,
-      "loss": 46.0,
-      "step": 14756
-    },
-    {
-      "epoch": 1.1282757038821034,
-      "grad_norm": 0.00028462946647778153,
-      "learning_rate": 0.00019999937264749264,
-      "loss": 46.0,
-      "step": 14757
-    },
-    {
-      "epoch": 1.128352160865493,
-      "grad_norm": 0.0007418460445478559,
-      "learning_rate": 0.0001999993725624078,
-      "loss": 46.0,
-      "step": 14758
-    },
-    {
-      "epoch": 1.1284286178488827,
-      "grad_norm": 0.00320740370079875,
-      "learning_rate": 0.00019999937247731716,
-      "loss": 46.0,
-      "step": 14759
-    },
-    {
-      "epoch": 1.1285050748322725,
-      "grad_norm": 0.000649608438834548,
-      "learning_rate": 0.00019999937239222079,
-      "loss": 46.0,
-      "step": 14760
-    },
-    {
-      "epoch": 1.1285815318156622,
-      "grad_norm": 0.0008807418635115027,
-      "learning_rate": 0.00019999937230711864,
-      "loss": 46.0,
-      "step": 14761
-    },
-    {
-      "epoch": 1.128657988799052,
-      "grad_norm": 0.0010668383911252022,
-      "learning_rate": 0.00019999937222201072,
-      "loss": 46.0,
-      "step": 14762
-    },
-    {
-      "epoch": 1.1287344457824418,
-      "grad_norm": 0.0008774464367888868,
-      "learning_rate": 0.00019999937213689703,
-      "loss": 46.0,
-      "step": 14763
-    },
-    {
-      "epoch": 1.1288109027658313,
-      "grad_norm": 0.0014249954838305712,
-      "learning_rate": 0.00019999937205177756,
-      "loss": 46.0,
-      "step": 14764
-    },
-    {
-      "epoch": 1.128887359749221,
-      "grad_norm": 0.0014622171875089407,
-      "learning_rate": 0.00019999937196665232,
-      "loss": 46.0,
-      "step": 14765
-    },
-    {
-      "epoch": 1.1289638167326108,
-      "grad_norm": 0.0015475729014724493,
-      "learning_rate": 0.0001999993718815213,
-      "loss": 46.0,
-      "step": 14766
-    },
-    {
-      "epoch": 1.1290402737160006,
-      "grad_norm": 0.0008325300877913833,
-      "learning_rate": 0.00019999937179638455,
-      "loss": 46.0,
-      "step": 14767
-    },
-    {
-      "epoch": 1.1291167306993903,
-      "grad_norm": 0.0060522365383803844,
-      "learning_rate": 0.00019999937171124202,
-      "loss": 46.0,
-      "step": 14768
-    },
-    {
-      "epoch": 1.1291931876827799,
-      "grad_norm": 0.007760223001241684,
-      "learning_rate": 0.0001999993716260937,
-      "loss": 46.0,
-      "step": 14769
-    },
-    {
-      "epoch": 1.1292696446661696,
-      "grad_norm": 0.0018330486491322517,
-      "learning_rate": 0.0001999993715409396,
-      "loss": 46.0,
-      "step": 14770
-    },
-    {
-      "epoch": 1.1293461016495594,
-      "grad_norm": 0.0007562093669548631,
-      "learning_rate": 0.00019999937145577978,
-      "loss": 46.0,
-      "step": 14771
-    },
-    {
-      "epoch": 1.1294225586329492,
-      "grad_norm": 0.005484177730977535,
-      "learning_rate": 0.00019999937137061416,
-      "loss": 46.0,
-      "step": 14772
-    },
-    {
-      "epoch": 1.129499015616339,
-      "grad_norm": 0.008028887212276459,
-      "learning_rate": 0.00019999937128544276,
-      "loss": 46.0,
-      "step": 14773
-    },
-    {
-      "epoch": 1.1295754725997287,
-      "grad_norm": 0.005428973585367203,
-      "learning_rate": 0.0001999993712002656,
-      "loss": 46.0,
-      "step": 14774
-    },
-    {
-      "epoch": 1.1296519295831182,
-      "grad_norm": 0.0040501998737454414,
-      "learning_rate": 0.00019999937111508267,
-      "loss": 46.0,
-      "step": 14775
-    },
-    {
-      "epoch": 1.129728386566508,
-      "grad_norm": 0.009396358393132687,
-      "learning_rate": 0.00019999937102989398,
-      "loss": 46.0,
-      "step": 14776
-    },
-    {
-      "epoch": 1.1298048435498977,
-      "grad_norm": 0.0012576605658978224,
-      "learning_rate": 0.0001999993709446995,
-      "loss": 46.0,
-      "step": 14777
-    },
-    {
-      "epoch": 1.1298813005332875,
-      "grad_norm": 0.0014676004648208618,
-      "learning_rate": 0.0001999993708594993,
-      "loss": 46.0,
-      "step": 14778
-    },
-    {
-      "epoch": 1.1299577575166773,
-      "grad_norm": 0.0005422924878075719,
-      "learning_rate": 0.00019999937077429326,
-      "loss": 46.0,
-      "step": 14779
-    },
-    {
-      "epoch": 1.1300342145000668,
-      "grad_norm": 0.004032416269183159,
-      "learning_rate": 0.0001999993706890815,
-      "loss": 46.0,
-      "step": 14780
-    },
-    {
-      "epoch": 1.1301106714834566,
-      "grad_norm": 0.0003497939324006438,
-      "learning_rate": 0.00019999937060386394,
-      "loss": 46.0,
-      "step": 14781
-    },
-    {
-      "epoch": 1.1301871284668463,
-      "grad_norm": 0.002158332848921418,
-      "learning_rate": 0.00019999937051864064,
-      "loss": 46.0,
-      "step": 14782
-    },
-    {
-      "epoch": 1.130263585450236,
-      "grad_norm": 0.0035231029614806175,
-      "learning_rate": 0.00019999937043341156,
-      "loss": 46.0,
-      "step": 14783
-    },
-    {
-      "epoch": 1.1303400424336258,
-      "grad_norm": 0.0015833841171115637,
-      "learning_rate": 0.0001999993703481767,
-      "loss": 46.0,
-      "step": 14784
-    },
-    {
-      "epoch": 1.1304164994170156,
-      "grad_norm": 0.000613283016718924,
-      "learning_rate": 0.00019999937026293608,
-      "loss": 46.0,
-      "step": 14785
-    },
-    {
-      "epoch": 1.1304929564004051,
-      "grad_norm": 0.0013062895741313696,
-      "learning_rate": 0.0001999993701776897,
-      "loss": 46.0,
-      "step": 14786
-    },
-    {
-      "epoch": 1.130569413383795,
-      "grad_norm": 0.0019704641308635473,
-      "learning_rate": 0.00019999937009243754,
-      "loss": 46.0,
-      "step": 14787
-    },
-    {
-      "epoch": 1.1306458703671847,
-      "grad_norm": 0.00115984829608351,
-      "learning_rate": 0.00019999937000717962,
-      "loss": 46.0,
-      "step": 14788
-    },
-    {
-      "epoch": 1.1307223273505744,
-      "grad_norm": 0.01768254116177559,
-      "learning_rate": 0.0001999993699219159,
-      "loss": 46.0,
-      "step": 14789
-    },
-    {
-      "epoch": 1.1307987843339642,
-      "grad_norm": 0.0011324993101879954,
-      "learning_rate": 0.0001999993698366464,
-      "loss": 46.0,
-      "step": 14790
-    },
-    {
-      "epoch": 1.1308752413173537,
-      "grad_norm": 0.0007432815618813038,
-      "learning_rate": 0.00019999936975137122,
-      "loss": 46.0,
-      "step": 14791
-    },
-    {
-      "epoch": 1.1309516983007435,
-      "grad_norm": 0.00291697820648551,
-      "learning_rate": 0.00019999936966609019,
-      "loss": 46.0,
-      "step": 14792
-    },
-    {
-      "epoch": 1.1310281552841333,
-      "grad_norm": 0.0016537210904061794,
-      "learning_rate": 0.0001999993695808034,
-      "loss": 46.0,
-      "step": 14793
-    },
-    {
-      "epoch": 1.131104612267523,
-      "grad_norm": 0.0007090248982422054,
-      "learning_rate": 0.00019999936949551087,
-      "loss": 46.0,
-      "step": 14794
-    },
-    {
-      "epoch": 1.1311810692509128,
-      "grad_norm": 0.0029994596261531115,
-      "learning_rate": 0.00019999936941021256,
-      "loss": 46.0,
-      "step": 14795
-    },
-    {
-      "epoch": 1.1312575262343025,
-      "grad_norm": 0.0015521880704909563,
-      "learning_rate": 0.00019999936932490849,
-      "loss": 46.0,
-      "step": 14796
-    },
-    {
-      "epoch": 1.131333983217692,
-      "grad_norm": 0.0008390407892875373,
-      "learning_rate": 0.0001999993692395986,
-      "loss": 46.0,
-      "step": 14797
-    },
-    {
-      "epoch": 1.1314104402010818,
-      "grad_norm": 0.0036280702333897352,
-      "learning_rate": 0.00019999936915428298,
-      "loss": 46.0,
-      "step": 14798
-    },
-    {
-      "epoch": 1.1314868971844716,
-      "grad_norm": 0.0016299005364999175,
-      "learning_rate": 0.00019999936906896158,
-      "loss": 46.0,
-      "step": 14799
-    },
-    {
-      "epoch": 1.1315633541678614,
-      "grad_norm": 0.0050012883730232716,
-      "learning_rate": 0.00019999936898363444,
-      "loss": 46.0,
-      "step": 14800
-    },
-    {
-      "epoch": 1.1316398111512511,
-      "grad_norm": 0.0006390110356733203,
-      "learning_rate": 0.00019999936889830152,
-      "loss": 46.0,
-      "step": 14801
-    },
-    {
-      "epoch": 1.1317162681346407,
-      "grad_norm": 0.002643123734742403,
-      "learning_rate": 0.0001999993688129628,
-      "loss": 46.0,
-      "step": 14802
-    },
-    {
-      "epoch": 1.1317927251180304,
-      "grad_norm": 0.002462599892169237,
-      "learning_rate": 0.00019999936872761834,
-      "loss": 46.0,
-      "step": 14803
-    },
-    {
-      "epoch": 1.1318691821014202,
-      "grad_norm": 0.0023767855018377304,
-      "learning_rate": 0.0001999993686422681,
-      "loss": 46.0,
-      "step": 14804
-    },
-    {
-      "epoch": 1.13194563908481,
-      "grad_norm": 0.009208094328641891,
-      "learning_rate": 0.00019999936855691206,
-      "loss": 46.0,
-      "step": 14805
-    },
-    {
-      "epoch": 1.1320220960681997,
-      "grad_norm": 0.0023866461124271154,
-      "learning_rate": 0.0001999993684715503,
-      "loss": 46.0,
-      "step": 14806
-    },
-    {
-      "epoch": 1.1320985530515895,
-      "grad_norm": 0.0004645321168936789,
-      "learning_rate": 0.00019999936838618277,
-      "loss": 46.0,
-      "step": 14807
-    },
-    {
-      "epoch": 1.132175010034979,
-      "grad_norm": 0.0008993617375381291,
-      "learning_rate": 0.00019999936830080944,
-      "loss": 46.0,
-      "step": 14808
-    },
-    {
-      "epoch": 1.1322514670183688,
-      "grad_norm": 0.001164270332083106,
-      "learning_rate": 0.00019999936821543034,
-      "loss": 46.0,
-      "step": 14809
-    },
-    {
-      "epoch": 1.1323279240017585,
-      "grad_norm": 0.008869229815900326,
-      "learning_rate": 0.00019999936813004548,
-      "loss": 46.0,
-      "step": 14810
-    },
-    {
-      "epoch": 1.1324043809851483,
-      "grad_norm": 0.0012709078146144748,
-      "learning_rate": 0.00019999936804465483,
-      "loss": 46.0,
-      "step": 14811
-    },
-    {
-      "epoch": 1.132480837968538,
-      "grad_norm": 0.001199951395392418,
-      "learning_rate": 0.00019999936795925846,
-      "loss": 46.0,
-      "step": 14812
-    },
-    {
-      "epoch": 1.1325572949519276,
-      "grad_norm": 0.0010795732960104942,
-      "learning_rate": 0.0001999993678738563,
-      "loss": 46.0,
-      "step": 14813
-    },
-    {
-      "epoch": 1.1326337519353173,
-      "grad_norm": 0.001172151300124824,
-      "learning_rate": 0.00019999936778844837,
-      "loss": 46.0,
-      "step": 14814
-    },
-    {
-      "epoch": 1.132710208918707,
-      "grad_norm": 0.0009043826721608639,
-      "learning_rate": 0.00019999936770303465,
-      "loss": 46.0,
-      "step": 14815
-    },
-    {
-      "epoch": 1.1327866659020969,
-      "grad_norm": 0.001527887536212802,
-      "learning_rate": 0.0001999993676176152,
-      "loss": 46.0,
-      "step": 14816
-    },
-    {
-      "epoch": 1.1328631228854866,
-      "grad_norm": 0.002399386838078499,
-      "learning_rate": 0.00019999936753218992,
-      "loss": 46.0,
-      "step": 14817
-    },
-    {
-      "epoch": 1.1329395798688764,
-      "grad_norm": 0.0062965815886855125,
-      "learning_rate": 0.00019999936744675894,
-      "loss": 46.0,
-      "step": 14818
-    },
-    {
-      "epoch": 1.133016036852266,
-      "grad_norm": 0.005739616230130196,
-      "learning_rate": 0.00019999936736132216,
-      "loss": 46.0,
-      "step": 14819
-    },
-    {
-      "epoch": 1.1330924938356557,
-      "grad_norm": 0.001907101133838296,
-      "learning_rate": 0.00019999936727587957,
-      "loss": 46.0,
-      "step": 14820
-    },
-    {
-      "epoch": 1.1331689508190455,
-      "grad_norm": 0.004465166013687849,
-      "learning_rate": 0.00019999936719043127,
-      "loss": 46.0,
-      "step": 14821
-    },
-    {
-      "epoch": 1.1332454078024352,
-      "grad_norm": 0.0007419039611704648,
-      "learning_rate": 0.0001999993671049772,
-      "loss": 46.0,
-      "step": 14822
-    },
-    {
-      "epoch": 1.1333218647858248,
-      "grad_norm": 0.00021928591013420373,
-      "learning_rate": 0.0001999993670195173,
-      "loss": 46.0,
-      "step": 14823
-    },
-    {
-      "epoch": 1.1333983217692145,
-      "grad_norm": 0.0007787608774378896,
-      "learning_rate": 0.00019999936693405169,
-      "loss": 46.0,
-      "step": 14824
-    },
-    {
-      "epoch": 1.1334747787526043,
-      "grad_norm": 0.0008408714784309268,
-      "learning_rate": 0.0001999993668485803,
-      "loss": 46.0,
-      "step": 14825
-    },
-    {
-      "epoch": 1.133551235735994,
-      "grad_norm": 0.0009409827180206776,
-      "learning_rate": 0.00019999936676310314,
-      "loss": 46.0,
-      "step": 14826
-    },
-    {
-      "epoch": 1.1336276927193838,
-      "grad_norm": 0.0008453232585452497,
-      "learning_rate": 0.00019999936667762017,
-      "loss": 46.0,
-      "step": 14827
-    },
-    {
-      "epoch": 1.1337041497027736,
-      "grad_norm": 0.0014677041908726096,
-      "learning_rate": 0.0001999993665921315,
-      "loss": 46.0,
-      "step": 14828
-    },
-    {
-      "epoch": 1.1337806066861633,
-      "grad_norm": 0.015801239758729935,
-      "learning_rate": 0.000199999366506637,
-      "loss": 46.0,
-      "step": 14829
-    },
-    {
-      "epoch": 1.1338570636695529,
-      "grad_norm": 0.0012309295125305653,
-      "learning_rate": 0.00019999936642113678,
-      "loss": 46.0,
-      "step": 14830
-    },
-    {
-      "epoch": 1.1339335206529426,
-      "grad_norm": 0.000673692615237087,
-      "learning_rate": 0.00019999936633563074,
-      "loss": 46.0,
-      "step": 14831
-    },
-    {
-      "epoch": 1.1340099776363324,
-      "grad_norm": 0.011002810671925545,
-      "learning_rate": 0.00019999936625011896,
-      "loss": 46.0,
-      "step": 14832
-    },
-    {
-      "epoch": 1.1340864346197221,
-      "grad_norm": 0.0012958996230736375,
-      "learning_rate": 0.0001999993661646014,
-      "loss": 46.0,
-      "step": 14833
-    },
-    {
-      "epoch": 1.1341628916031117,
-      "grad_norm": 0.004489955957978964,
-      "learning_rate": 0.00019999936607907807,
-      "loss": 46.0,
-      "step": 14834
-    },
-    {
-      "epoch": 1.1342393485865014,
-      "grad_norm": 0.001114579732529819,
-      "learning_rate": 0.000199999365993549,
-      "loss": 46.0,
-      "step": 14835
-    },
-    {
-      "epoch": 1.1343158055698912,
-      "grad_norm": 0.00208867690525949,
-      "learning_rate": 0.00019999936590801414,
-      "loss": 46.0,
-      "step": 14836
-    },
-    {
-      "epoch": 1.134392262553281,
-      "grad_norm": 0.00457307044416666,
-      "learning_rate": 0.00019999936582247352,
-      "loss": 46.0,
-      "step": 14837
-    },
-    {
-      "epoch": 1.1344687195366707,
-      "grad_norm": 0.0015747388824820518,
-      "learning_rate": 0.0001999993657369271,
-      "loss": 46.0,
-      "step": 14838
-    },
-    {
-      "epoch": 1.1345451765200605,
-      "grad_norm": 0.001406220137141645,
-      "learning_rate": 0.00019999936565137495,
-      "loss": 46.0,
-      "step": 14839
-    },
-    {
-      "epoch": 1.1346216335034502,
-      "grad_norm": 0.0006022056913934648,
-      "learning_rate": 0.000199999365565817,
-      "loss": 46.0,
-      "step": 14840
-    },
-    {
-      "epoch": 1.1346980904868398,
-      "grad_norm": 0.0013228778261691332,
-      "learning_rate": 0.0001999993654802533,
-      "loss": 46.0,
-      "step": 14841
-    },
-    {
-      "epoch": 1.1347745474702295,
-      "grad_norm": 0.00700360769405961,
-      "learning_rate": 0.0001999993653946838,
-      "loss": 46.0,
-      "step": 14842
-    },
-    {
-      "epoch": 1.1348510044536193,
-      "grad_norm": 0.010039308108389378,
-      "learning_rate": 0.00019999936530910856,
-      "loss": 46.0,
-      "step": 14843
-    },
-    {
-      "epoch": 1.134927461437009,
-      "grad_norm": 0.000553904683329165,
-      "learning_rate": 0.00019999936522352752,
-      "loss": 46.0,
-      "step": 14844
-    },
-    {
-      "epoch": 1.1350039184203986,
-      "grad_norm": 0.0009194056037813425,
-      "learning_rate": 0.00019999936513794076,
-      "loss": 46.0,
-      "step": 14845
-    },
-    {
-      "epoch": 1.1350803754037884,
-      "grad_norm": 0.0023718979209661484,
-      "learning_rate": 0.0001999993650523482,
-      "loss": 46.0,
-      "step": 14846
-    },
-    {
-      "epoch": 1.1351568323871781,
-      "grad_norm": 0.0012134637217968702,
-      "learning_rate": 0.00019999936496674988,
-      "loss": 46.0,
-      "step": 14847
-    },
-    {
-      "epoch": 1.1352332893705679,
-      "grad_norm": 0.0012317672371864319,
-      "learning_rate": 0.0001999993648811458,
-      "loss": 46.0,
-      "step": 14848
-    },
-    {
-      "epoch": 1.1353097463539576,
-      "grad_norm": 0.0016546202823519707,
-      "learning_rate": 0.00019999936479553592,
-      "loss": 46.0,
-      "step": 14849
-    },
-    {
-      "epoch": 1.1353862033373474,
-      "grad_norm": 0.003088191617280245,
-      "learning_rate": 0.00019999936470992027,
-      "loss": 46.0,
-      "step": 14850
-    },
-    {
-      "epoch": 1.135462660320737,
-      "grad_norm": 0.001624019700102508,
-      "learning_rate": 0.0001999993646242989,
-      "loss": 46.0,
-      "step": 14851
-    },
-    {
-      "epoch": 1.1355391173041267,
-      "grad_norm": 0.0010510454885661602,
-      "learning_rate": 0.0001999993645386717,
-      "loss": 46.0,
-      "step": 14852
-    },
-    {
-      "epoch": 1.1356155742875165,
-      "grad_norm": 0.0020558179821819067,
-      "learning_rate": 0.0001999993644530388,
-      "loss": 46.0,
-      "step": 14853
-    },
-    {
-      "epoch": 1.1356920312709062,
-      "grad_norm": 0.0007484521484002471,
-      "learning_rate": 0.00019999936436740004,
-      "loss": 46.0,
-      "step": 14854
-    },
-    {
-      "epoch": 1.135768488254296,
-      "grad_norm": 0.0004398217424750328,
-      "learning_rate": 0.00019999936428175555,
-      "loss": 46.0,
-      "step": 14855
-    },
-    {
-      "epoch": 1.1358449452376855,
-      "grad_norm": 0.002432475332170725,
-      "learning_rate": 0.00019999936419610531,
-      "loss": 46.0,
-      "step": 14856
-    },
-    {
-      "epoch": 1.1359214022210753,
-      "grad_norm": 0.03173724561929703,
-      "learning_rate": 0.0001999993641104493,
-      "loss": 46.0,
-      "step": 14857
-    },
-    {
-      "epoch": 1.135997859204465,
-      "grad_norm": 0.000692440546117723,
-      "learning_rate": 0.00019999936402478752,
-      "loss": 46.0,
-      "step": 14858
-    },
-    {
-      "epoch": 1.1360743161878548,
-      "grad_norm": 0.003717514919117093,
-      "learning_rate": 0.00019999936393911996,
-      "loss": 46.0,
-      "step": 14859
-    },
-    {
-      "epoch": 1.1361507731712446,
-      "grad_norm": 0.002513560699298978,
-      "learning_rate": 0.0001999993638534466,
-      "loss": 46.0,
-      "step": 14860
-    },
-    {
-      "epoch": 1.1362272301546343,
-      "grad_norm": 0.0017767185345292091,
-      "learning_rate": 0.0001999993637677675,
-      "loss": 46.0,
-      "step": 14861
-    },
-    {
-      "epoch": 1.1363036871380239,
-      "grad_norm": 0.003988855052739382,
-      "learning_rate": 0.00019999936368208268,
-      "loss": 46.0,
-      "step": 14862
-    },
-    {
-      "epoch": 1.1363801441214136,
-      "grad_norm": 0.0011938252719119191,
-      "learning_rate": 0.00019999936359639203,
-      "loss": 46.0,
-      "step": 14863
-    },
-    {
-      "epoch": 1.1364566011048034,
-      "grad_norm": 0.004000297747552395,
-      "learning_rate": 0.0001999993635106956,
-      "loss": 46.0,
-      "step": 14864
-    },
-    {
-      "epoch": 1.1365330580881932,
-      "grad_norm": 0.000504208030179143,
-      "learning_rate": 0.00019999936342499344,
-      "loss": 46.0,
-      "step": 14865
-    },
-    {
-      "epoch": 1.136609515071583,
-      "grad_norm": 0.0006104611675255001,
-      "learning_rate": 0.00019999936333928552,
-      "loss": 46.0,
-      "step": 14866
-    },
-    {
-      "epoch": 1.1366859720549725,
-      "grad_norm": 0.002088473876938224,
-      "learning_rate": 0.00019999936325357177,
-      "loss": 46.0,
-      "step": 14867
-    },
-    {
-      "epoch": 1.1367624290383622,
-      "grad_norm": 0.002962731057778001,
-      "learning_rate": 0.0001999993631678523,
-      "loss": 46.0,
-      "step": 14868
-    },
-    {
-      "epoch": 1.136838886021752,
-      "grad_norm": 0.001260658958926797,
-      "learning_rate": 0.00019999936308212702,
-      "loss": 46.0,
-      "step": 14869
-    },
-    {
-      "epoch": 1.1369153430051417,
-      "grad_norm": 0.00048768112901598215,
-      "learning_rate": 0.000199999362996396,
-      "loss": 46.0,
-      "step": 14870
-    },
-    {
-      "epoch": 1.1369917999885315,
-      "grad_norm": 0.0006740997778251767,
-      "learning_rate": 0.00019999936291065923,
-      "loss": 46.0,
-      "step": 14871
-    },
-    {
-      "epoch": 1.1370682569719213,
-      "grad_norm": 0.0013945174869149923,
-      "learning_rate": 0.00019999936282491667,
-      "loss": 46.0,
-      "step": 14872
-    },
-    {
-      "epoch": 1.1371447139553108,
-      "grad_norm": 0.0009060294833034277,
-      "learning_rate": 0.00019999936273916834,
-      "loss": 46.0,
-      "step": 14873
-    },
-    {
-      "epoch": 1.1372211709387006,
-      "grad_norm": 0.0005468170274980366,
-      "learning_rate": 0.00019999936265341424,
-      "loss": 46.0,
-      "step": 14874
-    },
-    {
-      "epoch": 1.1372976279220903,
-      "grad_norm": 0.0011170461075380445,
-      "learning_rate": 0.00019999936256765436,
-      "loss": 46.0,
-      "step": 14875
-    },
-    {
-      "epoch": 1.13737408490548,
-      "grad_norm": 0.0028317549731582403,
-      "learning_rate": 0.00019999936248188874,
-      "loss": 46.0,
-      "step": 14876
-    },
-    {
-      "epoch": 1.1374505418888698,
-      "grad_norm": 0.0015321020036935806,
-      "learning_rate": 0.00019999936239611734,
-      "loss": 46.0,
-      "step": 14877
-    },
-    {
-      "epoch": 1.1375269988722594,
-      "grad_norm": 0.0006664778920821846,
-      "learning_rate": 0.00019999936231034014,
-      "loss": 46.0,
-      "step": 14878
-    },
-    {
-      "epoch": 1.1376034558556491,
-      "grad_norm": 0.002765545854344964,
-      "learning_rate": 0.0001999993622245572,
-      "loss": 46.0,
-      "step": 14879
-    },
-    {
-      "epoch": 1.137679912839039,
-      "grad_norm": 0.007046765647828579,
-      "learning_rate": 0.00019999936213876848,
-      "loss": 46.0,
-      "step": 14880
-    },
-    {
-      "epoch": 1.1377563698224287,
-      "grad_norm": 0.0010540410876274109,
-      "learning_rate": 0.000199999362052974,
-      "loss": 46.0,
-      "step": 14881
-    },
-    {
-      "epoch": 1.1378328268058184,
-      "grad_norm": 0.0010335259139537811,
-      "learning_rate": 0.00019999936196717376,
-      "loss": 46.0,
-      "step": 14882
-    },
-    {
-      "epoch": 1.1379092837892082,
-      "grad_norm": 0.0007436656742356718,
-      "learning_rate": 0.00019999936188136772,
-      "loss": 46.0,
-      "step": 14883
-    },
-    {
-      "epoch": 1.1379857407725977,
-      "grad_norm": 0.0006247290875762701,
-      "learning_rate": 0.0001999993617955559,
-      "loss": 46.0,
-      "step": 14884
-    },
-    {
-      "epoch": 1.1380621977559875,
-      "grad_norm": 0.012216565199196339,
-      "learning_rate": 0.00019999936170973835,
-      "loss": 46.0,
-      "step": 14885
-    },
-    {
-      "epoch": 1.1381386547393773,
-      "grad_norm": 0.0003892823006026447,
-      "learning_rate": 0.00019999936162391505,
-      "loss": 46.0,
-      "step": 14886
-    },
-    {
-      "epoch": 1.138215111722767,
-      "grad_norm": 0.0015606341185048223,
-      "learning_rate": 0.00019999936153808592,
-      "loss": 46.0,
-      "step": 14887
-    },
-    {
-      "epoch": 1.1382915687061568,
-      "grad_norm": 0.004554451908916235,
-      "learning_rate": 0.00019999936145225107,
-      "loss": 46.0,
-      "step": 14888
-    },
-    {
-      "epoch": 1.1383680256895463,
-      "grad_norm": 0.00164883048273623,
-      "learning_rate": 0.00019999936136641042,
-      "loss": 46.0,
-      "step": 14889
-    },
-    {
-      "epoch": 1.138444482672936,
-      "grad_norm": 0.0018709847936406732,
-      "learning_rate": 0.000199999361280564,
-      "loss": 46.0,
-      "step": 14890
-    },
-    {
-      "epoch": 1.1385209396563258,
-      "grad_norm": 0.002143275924026966,
-      "learning_rate": 0.00019999936119471183,
-      "loss": 46.0,
-      "step": 14891
-    },
-    {
-      "epoch": 1.1385973966397156,
-      "grad_norm": 0.0024356436915695667,
-      "learning_rate": 0.00019999936110885389,
-      "loss": 46.0,
-      "step": 14892
-    },
-    {
-      "epoch": 1.1386738536231054,
-      "grad_norm": 0.0006338706007227302,
-      "learning_rate": 0.00019999936102299017,
-      "loss": 46.0,
-      "step": 14893
-    },
-    {
-      "epoch": 1.1387503106064951,
-      "grad_norm": 0.0005174280377104878,
-      "learning_rate": 0.00019999936093712068,
-      "loss": 46.0,
-      "step": 14894
-    },
-    {
-      "epoch": 1.1388267675898847,
-      "grad_norm": 0.02072383277118206,
-      "learning_rate": 0.00019999936085124542,
-      "loss": 46.0,
-      "step": 14895
-    },
-    {
-      "epoch": 1.1389032245732744,
-      "grad_norm": 0.0012328787706792355,
-      "learning_rate": 0.00019999936076536438,
-      "loss": 46.0,
-      "step": 14896
-    },
-    {
-      "epoch": 1.1389796815566642,
-      "grad_norm": 0.0037673902697861195,
-      "learning_rate": 0.00019999936067947757,
-      "loss": 46.0,
-      "step": 14897
-    },
-    {
-      "epoch": 1.139056138540054,
-      "grad_norm": 0.0032261626329272985,
-      "learning_rate": 0.00019999936059358504,
-      "loss": 46.0,
-      "step": 14898
-    },
-    {
-      "epoch": 1.1391325955234437,
-      "grad_norm": 0.0017686389619484544,
-      "learning_rate": 0.00019999936050768669,
-      "loss": 46.0,
-      "step": 14899
-    },
-    {
-      "epoch": 1.1392090525068332,
-      "grad_norm": 0.00190397456753999,
-      "learning_rate": 0.00019999936042178258,
-      "loss": 46.0,
-      "step": 14900
-    },
-    {
-      "epoch": 1.139285509490223,
-      "grad_norm": 0.010475380346179008,
-      "learning_rate": 0.0001999993603358727,
-      "loss": 46.0,
-      "step": 14901
-    },
-    {
-      "epoch": 1.1393619664736128,
-      "grad_norm": 0.002703100675716996,
-      "learning_rate": 0.00019999936024995706,
-      "loss": 46.0,
-      "step": 14902
-    },
-    {
-      "epoch": 1.1394384234570025,
-      "grad_norm": 0.0007554986514151096,
-      "learning_rate": 0.00019999936016403566,
-      "loss": 46.0,
-      "step": 14903
-    },
-    {
-      "epoch": 1.1395148804403923,
-      "grad_norm": 0.0015145642682909966,
-      "learning_rate": 0.00019999936007810846,
-      "loss": 46.0,
-      "step": 14904
-    },
-    {
-      "epoch": 1.139591337423782,
-      "grad_norm": 0.0011756264138966799,
-      "learning_rate": 0.00019999935999217552,
-      "loss": 46.0,
-      "step": 14905
-    },
-    {
-      "epoch": 1.1396677944071716,
-      "grad_norm": 0.003527546999976039,
-      "learning_rate": 0.0001999993599062368,
-      "loss": 46.0,
-      "step": 14906
-    },
-    {
-      "epoch": 1.1397442513905613,
-      "grad_norm": 0.0009252013405784965,
-      "learning_rate": 0.00019999935982029232,
-      "loss": 46.0,
-      "step": 14907
-    },
-    {
-      "epoch": 1.139820708373951,
-      "grad_norm": 0.002993206027895212,
-      "learning_rate": 0.00019999935973434205,
-      "loss": 46.0,
-      "step": 14908
-    },
-    {
-      "epoch": 1.1398971653573409,
-      "grad_norm": 0.0014797229086980224,
-      "learning_rate": 0.00019999935964838602,
-      "loss": 46.0,
-      "step": 14909
-    },
-    {
-      "epoch": 1.1399736223407306,
-      "grad_norm": 0.0007859288016334176,
-      "learning_rate": 0.0001999993595624242,
-      "loss": 46.0,
-      "step": 14910
-    },
-    {
-      "epoch": 1.1400500793241202,
-      "grad_norm": 0.003251680638641119,
-      "learning_rate": 0.00019999935947645665,
-      "loss": 46.0,
-      "step": 14911
-    },
-    {
-      "epoch": 1.14012653630751,
-      "grad_norm": 0.0010924285743385553,
-      "learning_rate": 0.00019999935939048332,
-      "loss": 46.0,
-      "step": 14912
-    },
-    {
-      "epoch": 1.1402029932908997,
-      "grad_norm": 0.0015300287632271647,
-      "learning_rate": 0.0001999993593045042,
-      "loss": 46.0,
-      "step": 14913
-    },
-    {
-      "epoch": 1.1402794502742895,
-      "grad_norm": 0.001611140789464116,
-      "learning_rate": 0.00019999935921851934,
-      "loss": 46.0,
-      "step": 14914
-    },
-    {
-      "epoch": 1.1403559072576792,
-      "grad_norm": 0.0006278762593865395,
-      "learning_rate": 0.0001999993591325287,
-      "loss": 46.0,
-      "step": 14915
-    },
-    {
-      "epoch": 1.140432364241069,
-      "grad_norm": 0.0013426505029201508,
-      "learning_rate": 0.00019999935904653227,
-      "loss": 46.0,
-      "step": 14916
-    },
-    {
-      "epoch": 1.1405088212244585,
-      "grad_norm": 0.0015826657181605697,
-      "learning_rate": 0.00019999935896053008,
-      "loss": 46.0,
-      "step": 14917
-    },
-    {
-      "epoch": 1.1405852782078483,
-      "grad_norm": 0.002817973028868437,
-      "learning_rate": 0.00019999935887452213,
-      "loss": 46.0,
-      "step": 14918
-    },
-    {
-      "epoch": 1.140661735191238,
-      "grad_norm": 0.0003659868089016527,
-      "learning_rate": 0.00019999935878850842,
-      "loss": 46.0,
-      "step": 14919
-    },
-    {
-      "epoch": 1.1407381921746278,
-      "grad_norm": 0.0009584235958755016,
-      "learning_rate": 0.0001999993587024889,
-      "loss": 46.0,
-      "step": 14920
-    },
-    {
-      "epoch": 1.1408146491580176,
-      "grad_norm": 0.003864546772092581,
-      "learning_rate": 0.00019999935861646364,
-      "loss": 46.0,
-      "step": 14921
-    },
-    {
-      "epoch": 1.140891106141407,
-      "grad_norm": 0.0009550279937684536,
-      "learning_rate": 0.00019999935853043263,
-      "loss": 46.0,
-      "step": 14922
-    },
-    {
-      "epoch": 1.1409675631247969,
-      "grad_norm": 0.0002660670434124768,
-      "learning_rate": 0.00019999935844439582,
-      "loss": 46.0,
-      "step": 14923
-    },
-    {
-      "epoch": 1.1410440201081866,
-      "grad_norm": 0.0017768017714843154,
-      "learning_rate": 0.00019999935835835327,
-      "loss": 46.0,
-      "step": 14924
-    },
-    {
-      "epoch": 1.1411204770915764,
-      "grad_norm": 0.0035320480819791555,
-      "learning_rate": 0.00019999935827230488,
-      "loss": 46.0,
-      "step": 14925
-    },
-    {
-      "epoch": 1.1411969340749661,
-      "grad_norm": 0.0099811851978302,
-      "learning_rate": 0.0001999993581862508,
-      "loss": 46.0,
-      "step": 14926
-    },
-    {
-      "epoch": 1.141273391058356,
-      "grad_norm": 0.0008795639732852578,
-      "learning_rate": 0.00019999935810019093,
-      "loss": 46.0,
-      "step": 14927
-    },
-    {
-      "epoch": 1.1413498480417454,
-      "grad_norm": 0.0010169106535613537,
-      "learning_rate": 0.00019999935801412528,
-      "loss": 46.0,
-      "step": 14928
-    },
-    {
-      "epoch": 1.1414263050251352,
-      "grad_norm": 0.0011053842026740313,
-      "learning_rate": 0.00019999935792805386,
-      "loss": 46.0,
-      "step": 14929
-    },
-    {
-      "epoch": 1.141502762008525,
-      "grad_norm": 0.001032682484947145,
-      "learning_rate": 0.00019999935784197667,
-      "loss": 46.0,
-      "step": 14930
-    },
-    {
-      "epoch": 1.1415792189919147,
-      "grad_norm": 0.0029977222438901663,
-      "learning_rate": 0.00019999935775589372,
-      "loss": 46.0,
-      "step": 14931
-    },
-    {
-      "epoch": 1.1416556759753045,
-      "grad_norm": 0.00665133586153388,
-      "learning_rate": 0.00019999935766980498,
-      "loss": 46.0,
-      "step": 14932
-    },
-    {
-      "epoch": 1.141732132958694,
-      "grad_norm": 0.0018333197804167867,
-      "learning_rate": 0.0001999993575837105,
-      "loss": 46.0,
-      "step": 14933
-    },
-    {
-      "epoch": 1.1418085899420838,
-      "grad_norm": 0.0008564876043237746,
-      "learning_rate": 0.00019999935749761023,
-      "loss": 46.0,
-      "step": 14934
-    },
-    {
-      "epoch": 1.1418850469254735,
-      "grad_norm": 0.0006230600411072373,
-      "learning_rate": 0.00019999935741150417,
-      "loss": 46.0,
-      "step": 14935
-    },
-    {
-      "epoch": 1.1419615039088633,
-      "grad_norm": 0.0005235027056187391,
-      "learning_rate": 0.00019999935732539239,
-      "loss": 46.0,
-      "step": 14936
-    },
-    {
-      "epoch": 1.142037960892253,
-      "grad_norm": 0.005327706225216389,
-      "learning_rate": 0.0001999993572392748,
-      "loss": 46.0,
-      "step": 14937
-    },
-    {
-      "epoch": 1.1421144178756428,
-      "grad_norm": 0.000640437996480614,
-      "learning_rate": 0.00019999935715315145,
-      "loss": 46.0,
-      "step": 14938
-    },
-    {
-      "epoch": 1.1421908748590324,
-      "grad_norm": 0.0012529651867225766,
-      "learning_rate": 0.00019999935706702235,
-      "loss": 46.0,
-      "step": 14939
-    },
-    {
-      "epoch": 1.1422673318424221,
-      "grad_norm": 0.0015473649837076664,
-      "learning_rate": 0.00019999935698088747,
-      "loss": 46.0,
-      "step": 14940
-    },
-    {
-      "epoch": 1.142343788825812,
-      "grad_norm": 0.0034063924103975296,
-      "learning_rate": 0.00019999935689474682,
-      "loss": 46.0,
-      "step": 14941
-    },
-    {
-      "epoch": 1.1424202458092017,
-      "grad_norm": 0.003460923209786415,
-      "learning_rate": 0.0001999993568086004,
-      "loss": 46.0,
-      "step": 14942
-    },
-    {
-      "epoch": 1.1424967027925914,
-      "grad_norm": 0.006585083436220884,
-      "learning_rate": 0.0001999993567224482,
-      "loss": 46.0,
-      "step": 14943
-    },
-    {
-      "epoch": 1.142573159775981,
-      "grad_norm": 0.001133933779783547,
-      "learning_rate": 0.00019999935663629026,
-      "loss": 46.0,
-      "step": 14944
-    },
-    {
-      "epoch": 1.1426496167593707,
-      "grad_norm": 0.0010075286263599992,
-      "learning_rate": 0.00019999935655012655,
-      "loss": 46.0,
-      "step": 14945
-    },
-    {
-      "epoch": 1.1427260737427605,
-      "grad_norm": 0.001493255840614438,
-      "learning_rate": 0.00019999935646395703,
-      "loss": 46.0,
-      "step": 14946
-    },
-    {
-      "epoch": 1.1428025307261502,
-      "grad_norm": 0.0006314993952400982,
-      "learning_rate": 0.00019999935637778177,
-      "loss": 46.0,
-      "step": 14947
-    },
-    {
-      "epoch": 1.14287898770954,
-      "grad_norm": 0.0011521636042743921,
-      "learning_rate": 0.0001999993562916007,
-      "loss": 46.0,
-      "step": 14948
-    },
-    {
-      "epoch": 1.1429554446929298,
-      "grad_norm": 0.005184476729482412,
-      "learning_rate": 0.00019999935620541393,
-      "loss": 46.0,
-      "step": 14949
-    },
-    {
-      "epoch": 1.1430319016763193,
-      "grad_norm": 0.0009690604638308287,
-      "learning_rate": 0.00019999935611922137,
-      "loss": 46.0,
-      "step": 14950
-    },
-    {
-      "epoch": 1.143108358659709,
-      "grad_norm": 0.0032997315283864737,
-      "learning_rate": 0.000199999356033023,
-      "loss": 46.0,
-      "step": 14951
-    },
-    {
-      "epoch": 1.1431848156430988,
-      "grad_norm": 0.0010366393253207207,
-      "learning_rate": 0.0001999993559468189,
-      "loss": 46.0,
-      "step": 14952
-    },
-    {
-      "epoch": 1.1432612726264886,
-      "grad_norm": 0.0010884796502068639,
-      "learning_rate": 0.000199999355860609,
-      "loss": 46.0,
-      "step": 14953
-    },
-    {
-      "epoch": 1.1433377296098783,
-      "grad_norm": 0.0012642787769436836,
-      "learning_rate": 0.00019999935577439334,
-      "loss": 46.0,
-      "step": 14954
-    },
-    {
-      "epoch": 1.1434141865932679,
-      "grad_norm": 0.0012666076654568315,
-      "learning_rate": 0.00019999935568817194,
-      "loss": 46.0,
-      "step": 14955
-    },
-    {
-      "epoch": 1.1434906435766576,
-      "grad_norm": 0.0025166980922222137,
-      "learning_rate": 0.00019999935560194475,
-      "loss": 46.0,
-      "step": 14956
-    },
-    {
-      "epoch": 1.1435671005600474,
-      "grad_norm": 0.0009621352655813098,
-      "learning_rate": 0.00019999935551571175,
-      "loss": 46.0,
-      "step": 14957
-    },
-    {
-      "epoch": 1.1436435575434372,
-      "grad_norm": 0.001618271810002625,
-      "learning_rate": 0.00019999935542947307,
-      "loss": 46.0,
-      "step": 14958
-    },
-    {
-      "epoch": 1.143720014526827,
-      "grad_norm": 0.0007134682964533567,
-      "learning_rate": 0.00019999935534322852,
-      "loss": 46.0,
-      "step": 14959
-    },
-    {
-      "epoch": 1.1437964715102167,
-      "grad_norm": 0.0006969352834858,
-      "learning_rate": 0.0001999993552569783,
-      "loss": 46.0,
-      "step": 14960
-    },
-    {
-      "epoch": 1.1438729284936062,
-      "grad_norm": 0.0005706926458515227,
-      "learning_rate": 0.00019999935517072223,
-      "loss": 46.0,
-      "step": 14961
-    },
-    {
-      "epoch": 1.143949385476996,
-      "grad_norm": 0.0019901113118976355,
-      "learning_rate": 0.00019999935508446042,
-      "loss": 46.0,
-      "step": 14962
-    },
-    {
-      "epoch": 1.1440258424603857,
-      "grad_norm": 0.001561901532113552,
-      "learning_rate": 0.00019999935499819284,
-      "loss": 46.0,
-      "step": 14963
-    },
-    {
-      "epoch": 1.1441022994437755,
-      "grad_norm": 0.004461059346795082,
-      "learning_rate": 0.0001999993549119195,
-      "loss": 46.0,
-      "step": 14964
-    },
-    {
-      "epoch": 1.144178756427165,
-      "grad_norm": 0.000552699260879308,
-      "learning_rate": 0.00019999935482564038,
-      "loss": 46.0,
-      "step": 14965
-    },
-    {
-      "epoch": 1.1442552134105548,
-      "grad_norm": 0.0015074230032041669,
-      "learning_rate": 0.0001999993547393555,
-      "loss": 46.0,
-      "step": 14966
-    },
-    {
-      "epoch": 1.1443316703939446,
-      "grad_norm": 0.008560001850128174,
-      "learning_rate": 0.00019999935465306483,
-      "loss": 46.0,
-      "step": 14967
-    },
-    {
-      "epoch": 1.1444081273773343,
-      "grad_norm": 0.0015120639000087976,
-      "learning_rate": 0.0001999993545667684,
-      "loss": 46.0,
-      "step": 14968
-    },
-    {
-      "epoch": 1.144484584360724,
-      "grad_norm": 0.0004665811429731548,
-      "learning_rate": 0.00019999935448046621,
-      "loss": 46.0,
-      "step": 14969
-    },
-    {
-      "epoch": 1.1445610413441138,
-      "grad_norm": 0.003878061193972826,
-      "learning_rate": 0.00019999935439415825,
-      "loss": 46.0,
-      "step": 14970
-    },
-    {
-      "epoch": 1.1446374983275036,
-      "grad_norm": 0.0005604481557384133,
-      "learning_rate": 0.00019999935430784453,
-      "loss": 46.0,
-      "step": 14971
-    },
-    {
-      "epoch": 1.1447139553108931,
-      "grad_norm": 0.0025031897239387035,
-      "learning_rate": 0.00019999935422152502,
-      "loss": 46.0,
-      "step": 14972
-    },
-    {
-      "epoch": 1.144790412294283,
-      "grad_norm": 0.0003243081155233085,
-      "learning_rate": 0.00019999935413519976,
-      "loss": 46.0,
-      "step": 14973
-    },
-    {
-      "epoch": 1.1448668692776727,
-      "grad_norm": 0.0030771391466259956,
-      "learning_rate": 0.00019999935404886872,
-      "loss": 46.0,
-      "step": 14974
-    },
-    {
-      "epoch": 1.1449433262610624,
-      "grad_norm": 0.0013102870434522629,
-      "learning_rate": 0.0001999993539625319,
-      "loss": 46.0,
-      "step": 14975
-    },
-    {
-      "epoch": 1.145019783244452,
-      "grad_norm": 0.0011032220209017396,
-      "learning_rate": 0.00019999935387618933,
-      "loss": 46.0,
-      "step": 14976
-    },
-    {
-      "epoch": 1.1450962402278417,
-      "grad_norm": 0.01567205600440502,
-      "learning_rate": 0.00019999935378984098,
-      "loss": 46.0,
-      "step": 14977
-    },
-    {
-      "epoch": 1.1451726972112315,
-      "grad_norm": 0.000965586572419852,
-      "learning_rate": 0.00019999935370348685,
-      "loss": 46.0,
-      "step": 14978
-    },
-    {
-      "epoch": 1.1452491541946213,
-      "grad_norm": 0.002214935142546892,
-      "learning_rate": 0.00019999935361712695,
-      "loss": 46.0,
-      "step": 14979
-    },
-    {
-      "epoch": 1.145325611178011,
-      "grad_norm": 0.0034965798258781433,
-      "learning_rate": 0.0001999993535307613,
-      "loss": 46.0,
-      "step": 14980
-    },
-    {
-      "epoch": 1.1454020681614008,
-      "grad_norm": 0.0009097771253436804,
-      "learning_rate": 0.00019999935344438988,
-      "loss": 46.0,
-      "step": 14981
-    },
-    {
-      "epoch": 1.1454785251447903,
-      "grad_norm": 0.0008807270787656307,
-      "learning_rate": 0.00019999935335801268,
-      "loss": 46.0,
-      "step": 14982
-    },
-    {
-      "epoch": 1.14555498212818,
-      "grad_norm": 0.0016269584884867072,
-      "learning_rate": 0.00019999935327162972,
-      "loss": 46.0,
-      "step": 14983
-    },
-    {
-      "epoch": 1.1456314391115698,
-      "grad_norm": 0.002810464473441243,
-      "learning_rate": 0.00019999935318524098,
-      "loss": 46.0,
-      "step": 14984
-    },
-    {
-      "epoch": 1.1457078960949596,
-      "grad_norm": 0.0016543669626116753,
-      "learning_rate": 0.0001999993530988465,
-      "loss": 46.0,
-      "step": 14985
-    },
-    {
-      "epoch": 1.1457843530783494,
-      "grad_norm": 0.0019010150572285056,
-      "learning_rate": 0.0001999993530124462,
-      "loss": 46.0,
-      "step": 14986
-    },
-    {
-      "epoch": 1.145860810061739,
-      "grad_norm": 0.006113330367952585,
-      "learning_rate": 0.00019999935292604014,
-      "loss": 46.0,
-      "step": 14987
-    },
-    {
-      "epoch": 1.1459372670451287,
-      "grad_norm": 0.0009544951608404517,
-      "learning_rate": 0.00019999935283962833,
-      "loss": 46.0,
-      "step": 14988
-    },
-    {
-      "epoch": 1.1460137240285184,
-      "grad_norm": 0.01300711091607809,
-      "learning_rate": 0.00019999935275321075,
-      "loss": 46.0,
-      "step": 14989
-    },
-    {
-      "epoch": 1.1460901810119082,
-      "grad_norm": 0.001527132117189467,
-      "learning_rate": 0.0001999993526667874,
-      "loss": 46.0,
-      "step": 14990
-    },
-    {
-      "epoch": 1.146166637995298,
-      "grad_norm": 0.0007927576662041247,
-      "learning_rate": 0.0001999993525803583,
-      "loss": 46.0,
-      "step": 14991
-    },
-    {
-      "epoch": 1.1462430949786877,
-      "grad_norm": 0.0008664650958962739,
-      "learning_rate": 0.0001999993524939234,
-      "loss": 46.0,
-      "step": 14992
-    },
-    {
-      "epoch": 1.1463195519620772,
-      "grad_norm": 0.003534761257469654,
-      "learning_rate": 0.00019999935240748272,
-      "loss": 46.0,
-      "step": 14993
-    },
-    {
-      "epoch": 1.146396008945467,
-      "grad_norm": 0.0017882920801639557,
-      "learning_rate": 0.0001999993523210363,
-      "loss": 46.0,
-      "step": 14994
-    },
-    {
-      "epoch": 1.1464724659288568,
-      "grad_norm": 0.0007550594164058566,
-      "learning_rate": 0.0001999993522345841,
-      "loss": 46.0,
-      "step": 14995
-    },
-    {
-      "epoch": 1.1465489229122465,
-      "grad_norm": 0.0011703919153660536,
-      "learning_rate": 0.00019999935214812614,
-      "loss": 46.0,
-      "step": 14996
-    },
-    {
-      "epoch": 1.1466253798956363,
-      "grad_norm": 0.0008181068697012961,
-      "learning_rate": 0.0001999993520616624,
-      "loss": 46.0,
-      "step": 14997
-    },
-    {
-      "epoch": 1.1467018368790258,
-      "grad_norm": 0.0012540104798972607,
-      "learning_rate": 0.00019999935197519288,
-      "loss": 46.0,
-      "step": 14998
-    },
-    {
-      "epoch": 1.1467782938624156,
-      "grad_norm": 0.0034341567661613226,
-      "learning_rate": 0.00019999935188871762,
-      "loss": 46.0,
-      "step": 14999
-    },
-    {
-      "epoch": 1.1468547508458053,
-      "grad_norm": 0.011878249235451221,
-      "learning_rate": 0.00019999935180223656,
-      "loss": 46.0,
-      "step": 15000
-    },
-    {
-      "epoch": 1.146931207829195,
-      "grad_norm": 0.0005729121621698141,
-      "learning_rate": 0.00019999935171574975,
-      "loss": 46.0,
-      "step": 15001
-    },
-    {
-      "epoch": 1.1470076648125849,
-      "grad_norm": 0.0005691108526661992,
-      "learning_rate": 0.00019999935162925717,
-      "loss": 46.0,
-      "step": 15002
-    },
-    {
-      "epoch": 1.1470841217959746,
-      "grad_norm": 0.004101775120943785,
-      "learning_rate": 0.00019999935154275882,
-      "loss": 46.0,
-      "step": 15003
-    },
-    {
-      "epoch": 1.1471605787793642,
-      "grad_norm": 0.0024868238251656294,
-      "learning_rate": 0.0001999993514562547,
-      "loss": 46.0,
-      "step": 15004
-    },
-    {
-      "epoch": 1.147237035762754,
-      "grad_norm": 0.002159201307222247,
-      "learning_rate": 0.0001999993513697448,
-      "loss": 46.0,
-      "step": 15005
-    },
-    {
-      "epoch": 1.1473134927461437,
-      "grad_norm": 0.0010787959909066558,
-      "learning_rate": 0.00019999935128322914,
-      "loss": 46.0,
-      "step": 15006
-    },
-    {
-      "epoch": 1.1473899497295335,
-      "grad_norm": 0.0008919485262595117,
-      "learning_rate": 0.0001999993511967077,
-      "loss": 46.0,
-      "step": 15007
-    },
-    {
-      "epoch": 1.1474664067129232,
-      "grad_norm": 0.004554085433483124,
-      "learning_rate": 0.0001999993511101805,
-      "loss": 46.0,
-      "step": 15008
-    },
-    {
-      "epoch": 1.1475428636963128,
-      "grad_norm": 0.0007534997421316803,
-      "learning_rate": 0.00019999935102364753,
-      "loss": 46.0,
-      "step": 15009
-    },
-    {
-      "epoch": 1.1476193206797025,
-      "grad_norm": 0.0012508330401033163,
-      "learning_rate": 0.0001999993509371088,
-      "loss": 46.0,
-      "step": 15010
-    },
-    {
-      "epoch": 1.1476957776630923,
-      "grad_norm": 0.0013874380383640528,
-      "learning_rate": 0.00019999935085056428,
-      "loss": 46.0,
-      "step": 15011
-    },
-    {
-      "epoch": 1.147772234646482,
-      "grad_norm": 0.0008722414495423436,
-      "learning_rate": 0.000199999350764014,
-      "loss": 46.0,
-      "step": 15012
-    },
-    {
-      "epoch": 1.1478486916298718,
-      "grad_norm": 0.0006791193154640496,
-      "learning_rate": 0.00019999935067745793,
-      "loss": 46.0,
-      "step": 15013
-    },
-    {
-      "epoch": 1.1479251486132616,
-      "grad_norm": 0.0013268411858007312,
-      "learning_rate": 0.00019999935059089615,
-      "loss": 46.0,
-      "step": 15014
-    },
-    {
-      "epoch": 1.148001605596651,
-      "grad_norm": 0.00099524250254035,
-      "learning_rate": 0.00019999935050432854,
-      "loss": 46.0,
-      "step": 15015
-    },
-    {
-      "epoch": 1.1480780625800409,
-      "grad_norm": 0.0029780941549688578,
-      "learning_rate": 0.00019999935041775516,
-      "loss": 46.0,
-      "step": 15016
-    },
-    {
-      "epoch": 1.1481545195634306,
-      "grad_norm": 0.002819076180458069,
-      "learning_rate": 0.00019999935033117604,
-      "loss": 46.0,
-      "step": 15017
-    },
-    {
-      "epoch": 1.1482309765468204,
-      "grad_norm": 0.0006981000187806785,
-      "learning_rate": 0.00019999935024459116,
-      "loss": 46.0,
-      "step": 15018
-    },
-    {
-      "epoch": 1.1483074335302101,
-      "grad_norm": 0.0009209704003296793,
-      "learning_rate": 0.00019999935015800052,
-      "loss": 46.0,
-      "step": 15019
-    },
-    {
-      "epoch": 1.1483838905135997,
-      "grad_norm": 0.00589301623404026,
-      "learning_rate": 0.00019999935007140404,
-      "loss": 46.0,
-      "step": 15020
-    },
-    {
-      "epoch": 1.1484603474969894,
-      "grad_norm": 0.0038544354028999805,
-      "learning_rate": 0.00019999934998480185,
-      "loss": 46.0,
-      "step": 15021
-    },
-    {
-      "epoch": 1.1485368044803792,
-      "grad_norm": 0.004403791856020689,
-      "learning_rate": 0.00019999934989819388,
-      "loss": 46.0,
-      "step": 15022
-    },
-    {
-      "epoch": 1.148613261463769,
-      "grad_norm": 0.0008765459060668945,
-      "learning_rate": 0.00019999934981158014,
-      "loss": 46.0,
-      "step": 15023
-    },
-    {
-      "epoch": 1.1486897184471587,
-      "grad_norm": 0.003624152159318328,
-      "learning_rate": 0.00019999934972496063,
-      "loss": 46.0,
-      "step": 15024
-    },
-    {
-      "epoch": 1.1487661754305485,
-      "grad_norm": 0.0012451920192688704,
-      "learning_rate": 0.00019999934963833534,
-      "loss": 46.0,
-      "step": 15025
-    },
-    {
-      "epoch": 1.148842632413938,
-      "grad_norm": 0.004042399115860462,
-      "learning_rate": 0.0001999993495517043,
-      "loss": 46.0,
-      "step": 15026
-    },
-    {
-      "epoch": 1.1489190893973278,
-      "grad_norm": 0.0003782711864914745,
-      "learning_rate": 0.00019999934946506747,
-      "loss": 46.0,
-      "step": 15027
-    },
-    {
-      "epoch": 1.1489955463807175,
-      "grad_norm": 0.000507171789649874,
-      "learning_rate": 0.0001999993493784249,
-      "loss": 46.0,
-      "step": 15028
-    },
-    {
-      "epoch": 1.1490720033641073,
-      "grad_norm": 0.002404289785772562,
-      "learning_rate": 0.00019999934929177654,
-      "loss": 46.0,
-      "step": 15029
-    },
-    {
-      "epoch": 1.149148460347497,
-      "grad_norm": 0.0005509738111868501,
-      "learning_rate": 0.00019999934920512239,
-      "loss": 46.0,
-      "step": 15030
-    },
-    {
-      "epoch": 1.1492249173308866,
-      "grad_norm": 0.01184279378503561,
-      "learning_rate": 0.00019999934911846249,
-      "loss": 46.0,
-      "step": 15031
-    },
-    {
-      "epoch": 1.1493013743142764,
-      "grad_norm": 0.0009135470609180629,
-      "learning_rate": 0.0001999993490317968,
-      "loss": 46.0,
-      "step": 15032
-    },
-    {
-      "epoch": 1.1493778312976661,
-      "grad_norm": 0.0019471608102321625,
-      "learning_rate": 0.00019999934894512537,
-      "loss": 46.0,
-      "step": 15033
-    },
-    {
-      "epoch": 1.149454288281056,
-      "grad_norm": 0.0006821328424848616,
-      "learning_rate": 0.00019999934885844817,
-      "loss": 46.0,
-      "step": 15034
-    },
-    {
-      "epoch": 1.1495307452644457,
-      "grad_norm": 0.0006921023596078157,
-      "learning_rate": 0.0001999993487717652,
-      "loss": 46.0,
-      "step": 15035
-    },
-    {
-      "epoch": 1.1496072022478354,
-      "grad_norm": 0.0015824009897187352,
-      "learning_rate": 0.00019999934868507647,
-      "loss": 46.0,
-      "step": 15036
-    },
-    {
-      "epoch": 1.149683659231225,
-      "grad_norm": 0.002302803099155426,
-      "learning_rate": 0.00019999934859838195,
-      "loss": 46.0,
-      "step": 15037
-    },
-    {
-      "epoch": 1.1497601162146147,
-      "grad_norm": 0.0007952466839924455,
-      "learning_rate": 0.00019999934851168164,
-      "loss": 46.0,
-      "step": 15038
-    },
-    {
-      "epoch": 1.1498365731980045,
-      "grad_norm": 0.0022488797549158335,
-      "learning_rate": 0.0001999993484249756,
-      "loss": 46.0,
-      "step": 15039
-    },
-    {
-      "epoch": 1.1499130301813942,
-      "grad_norm": 0.0004454887821339071,
-      "learning_rate": 0.0001999993483382638,
-      "loss": 46.0,
-      "step": 15040
-    },
-    {
-      "epoch": 1.149989487164784,
-      "grad_norm": 0.001770614180713892,
-      "learning_rate": 0.0001999993482515462,
-      "loss": 46.0,
-      "step": 15041
-    },
-    {
-      "epoch": 1.1500659441481735,
-      "grad_norm": 0.0008196835406124592,
-      "learning_rate": 0.00019999934816482284,
-      "loss": 46.0,
-      "step": 15042
-    },
-    {
-      "epoch": 1.1501424011315633,
-      "grad_norm": 0.002438412979245186,
-      "learning_rate": 0.0001999993480780937,
-      "loss": 46.0,
-      "step": 15043
-    },
-    {
-      "epoch": 1.150218858114953,
-      "grad_norm": 0.0010210862383246422,
-      "learning_rate": 0.00019999934799135879,
-      "loss": 46.0,
-      "step": 15044
-    },
-    {
-      "epoch": 1.1502953150983428,
-      "grad_norm": 0.006213759537786245,
-      "learning_rate": 0.00019999934790461814,
-      "loss": 46.0,
-      "step": 15045
-    },
-    {
-      "epoch": 1.1503717720817326,
-      "grad_norm": 0.0007969638681970537,
-      "learning_rate": 0.00019999934781787172,
-      "loss": 46.0,
-      "step": 15046
-    },
-    {
-      "epoch": 1.1504482290651223,
-      "grad_norm": 0.0006172886351123452,
-      "learning_rate": 0.00019999934773111947,
-      "loss": 46.0,
-      "step": 15047
-    },
-    {
-      "epoch": 1.1505246860485119,
-      "grad_norm": 0.0005070014740340412,
-      "learning_rate": 0.0001999993476443615,
-      "loss": 46.0,
-      "step": 15048
-    },
-    {
-      "epoch": 1.1506011430319016,
-      "grad_norm": 0.0013464834773913026,
-      "learning_rate": 0.00019999934755759777,
-      "loss": 46.0,
-      "step": 15049
-    },
-    {
-      "epoch": 1.1506776000152914,
-      "grad_norm": 0.001961823320016265,
-      "learning_rate": 0.00019999934747082825,
-      "loss": 46.0,
-      "step": 15050
-    },
-    {
-      "epoch": 1.1507540569986812,
-      "grad_norm": 0.003623869502916932,
-      "learning_rate": 0.00019999934738405294,
-      "loss": 46.0,
-      "step": 15051
-    },
-    {
-      "epoch": 1.150830513982071,
-      "grad_norm": 0.0025356500409543514,
-      "learning_rate": 0.0001999993472972719,
-      "loss": 46.0,
-      "step": 15052
-    },
-    {
-      "epoch": 1.1509069709654605,
-      "grad_norm": 0.0015827063471078873,
-      "learning_rate": 0.00019999934721048507,
-      "loss": 46.0,
-      "step": 15053
-    },
-    {
-      "epoch": 1.1509834279488502,
-      "grad_norm": 0.0007500315550714731,
-      "learning_rate": 0.0001999993471236925,
-      "loss": 46.0,
-      "step": 15054
-    },
-    {
-      "epoch": 1.15105988493224,
-      "grad_norm": 0.0035137522500008345,
-      "learning_rate": 0.00019999934703689412,
-      "loss": 46.0,
-      "step": 15055
-    },
-    {
-      "epoch": 1.1511363419156297,
-      "grad_norm": 0.000715860107447952,
-      "learning_rate": 0.00019999934695009,
-      "loss": 46.0,
-      "step": 15056
-    },
-    {
-      "epoch": 1.1512127988990195,
-      "grad_norm": 0.001137072453275323,
-      "learning_rate": 0.0001999993468632801,
-      "loss": 46.0,
-      "step": 15057
-    },
-    {
-      "epoch": 1.1512892558824093,
-      "grad_norm": 0.010411955416202545,
-      "learning_rate": 0.0001999993467764644,
-      "loss": 46.0,
-      "step": 15058
-    },
-    {
-      "epoch": 1.1513657128657988,
-      "grad_norm": 0.004144825506955385,
-      "learning_rate": 0.00019999934668964294,
-      "loss": 46.0,
-      "step": 15059
-    },
-    {
-      "epoch": 1.1514421698491886,
-      "grad_norm": 0.0005411454476416111,
-      "learning_rate": 0.00019999934660281575,
-      "loss": 46.0,
-      "step": 15060
-    },
-    {
-      "epoch": 1.1515186268325783,
-      "grad_norm": 0.0057448456063866615,
-      "learning_rate": 0.00019999934651598276,
-      "loss": 46.0,
-      "step": 15061
-    },
-    {
-      "epoch": 1.151595083815968,
-      "grad_norm": 0.009231047704815865,
-      "learning_rate": 0.00019999934642914402,
-      "loss": 46.0,
-      "step": 15062
-    },
-    {
-      "epoch": 1.1516715407993579,
-      "grad_norm": 0.00044466342660598457,
-      "learning_rate": 0.0001999993463422995,
-      "loss": 46.0,
-      "step": 15063
-    },
-    {
-      "epoch": 1.1517479977827474,
-      "grad_norm": 0.0007833400159142911,
-      "learning_rate": 0.0001999993462554492,
-      "loss": 46.0,
-      "step": 15064
-    },
-    {
-      "epoch": 1.1518244547661372,
-      "grad_norm": 0.0011725116055458784,
-      "learning_rate": 0.00019999934616859316,
-      "loss": 46.0,
-      "step": 15065
-    },
-    {
-      "epoch": 1.151900911749527,
-      "grad_norm": 0.0009265861008316278,
-      "learning_rate": 0.00019999934608173133,
-      "loss": 46.0,
-      "step": 15066
-    },
-    {
-      "epoch": 1.1519773687329167,
-      "grad_norm": 0.0018999343737959862,
-      "learning_rate": 0.00019999934599486375,
-      "loss": 46.0,
-      "step": 15067
-    },
-    {
-      "epoch": 1.1520538257163064,
-      "grad_norm": 0.003532078582793474,
-      "learning_rate": 0.00019999934590799037,
-      "loss": 46.0,
-      "step": 15068
-    },
-    {
-      "epoch": 1.1521302826996962,
-      "grad_norm": 0.0007617921219207346,
-      "learning_rate": 0.00019999934582111122,
-      "loss": 46.0,
-      "step": 15069
-    },
-    {
-      "epoch": 1.1522067396830857,
-      "grad_norm": 0.00045757039333693683,
-      "learning_rate": 0.00019999934573422632,
-      "loss": 46.0,
-      "step": 15070
-    },
-    {
-      "epoch": 1.1522831966664755,
-      "grad_norm": 0.002881947671994567,
-      "learning_rate": 0.00019999934564733564,
-      "loss": 46.0,
-      "step": 15071
-    },
-    {
-      "epoch": 1.1523596536498653,
-      "grad_norm": 0.0039932820945978165,
-      "learning_rate": 0.0001999993455604392,
-      "loss": 46.0,
-      "step": 15072
-    },
-    {
-      "epoch": 1.152436110633255,
-      "grad_norm": 0.001974108163267374,
-      "learning_rate": 0.00019999934547353698,
-      "loss": 46.0,
-      "step": 15073
-    },
-    {
-      "epoch": 1.1525125676166448,
-      "grad_norm": 0.0056515466421842575,
-      "learning_rate": 0.00019999934538662899,
-      "loss": 46.0,
-      "step": 15074
-    },
-    {
-      "epoch": 1.1525890246000343,
-      "grad_norm": 0.0034237743820995092,
-      "learning_rate": 0.00019999934529971525,
-      "loss": 46.0,
-      "step": 15075
-    },
-    {
-      "epoch": 1.152665481583424,
-      "grad_norm": 0.0010519057977944613,
-      "learning_rate": 0.00019999934521279573,
-      "loss": 46.0,
-      "step": 15076
-    },
-    {
-      "epoch": 1.1527419385668138,
-      "grad_norm": 0.000748187187127769,
-      "learning_rate": 0.00019999934512587045,
-      "loss": 46.0,
-      "step": 15077
-    },
-    {
-      "epoch": 1.1528183955502036,
-      "grad_norm": 0.002633391646668315,
-      "learning_rate": 0.0001999993450389394,
-      "loss": 46.0,
-      "step": 15078
-    },
-    {
-      "epoch": 1.1528948525335934,
-      "grad_norm": 0.0005842336686328053,
-      "learning_rate": 0.00019999934495200253,
-      "loss": 46.0,
-      "step": 15079
-    },
-    {
-      "epoch": 1.1529713095169831,
-      "grad_norm": 0.004706832114607096,
-      "learning_rate": 0.00019999934486505992,
-      "loss": 46.0,
-      "step": 15080
-    },
-    {
-      "epoch": 1.1530477665003727,
-      "grad_norm": 0.001201618812046945,
-      "learning_rate": 0.00019999934477811157,
-      "loss": 46.0,
-      "step": 15081
-    },
-    {
-      "epoch": 1.1531242234837624,
-      "grad_norm": 0.0020892757456749678,
-      "learning_rate": 0.00019999934469115745,
-      "loss": 46.0,
-      "step": 15082
-    },
-    {
-      "epoch": 1.1532006804671522,
-      "grad_norm": 0.0012676959158852696,
-      "learning_rate": 0.00019999934460419755,
-      "loss": 46.0,
-      "step": 15083
-    },
-    {
-      "epoch": 1.153277137450542,
-      "grad_norm": 0.0008154280949383974,
-      "learning_rate": 0.00019999934451723185,
-      "loss": 46.0,
-      "step": 15084
-    },
-    {
-      "epoch": 1.1533535944339317,
-      "grad_norm": 0.0011737341992557049,
-      "learning_rate": 0.0001999993444302604,
-      "loss": 46.0,
-      "step": 15085
-    },
-    {
-      "epoch": 1.1534300514173212,
-      "grad_norm": 0.0024835539516061544,
-      "learning_rate": 0.00019999934434328318,
-      "loss": 46.0,
-      "step": 15086
-    },
-    {
-      "epoch": 1.153506508400711,
-      "grad_norm": 0.0006917276768945158,
-      "learning_rate": 0.0001999993442563002,
-      "loss": 46.0,
-      "step": 15087
-    },
-    {
-      "epoch": 1.1535829653841008,
-      "grad_norm": 0.003550690831616521,
-      "learning_rate": 0.00019999934416931145,
-      "loss": 46.0,
-      "step": 15088
-    },
-    {
-      "epoch": 1.1536594223674905,
-      "grad_norm": 0.0011154643725603819,
-      "learning_rate": 0.00019999934408231694,
-      "loss": 46.0,
-      "step": 15089
-    },
-    {
-      "epoch": 1.1537358793508803,
-      "grad_norm": 0.00127691391389817,
-      "learning_rate": 0.00019999934399531663,
-      "loss": 46.0,
-      "step": 15090
-    },
-    {
-      "epoch": 1.15381233633427,
-      "grad_norm": 0.0006531690014526248,
-      "learning_rate": 0.00019999934390831057,
-      "loss": 46.0,
-      "step": 15091
-    },
-    {
-      "epoch": 1.1538887933176596,
-      "grad_norm": 0.010128902271389961,
-      "learning_rate": 0.00019999934382129874,
-      "loss": 46.0,
-      "step": 15092
-    },
-    {
-      "epoch": 1.1539652503010494,
-      "grad_norm": 0.006677805911749601,
-      "learning_rate": 0.00019999934373428113,
-      "loss": 46.0,
-      "step": 15093
-    },
-    {
-      "epoch": 1.1540417072844391,
-      "grad_norm": 0.0012262585805729032,
-      "learning_rate": 0.00019999934364725775,
-      "loss": 46.0,
-      "step": 15094
-    },
-    {
-      "epoch": 1.1541181642678289,
-      "grad_norm": 0.0006395959062501788,
-      "learning_rate": 0.00019999934356022863,
-      "loss": 46.0,
-      "step": 15095
-    },
-    {
-      "epoch": 1.1541946212512184,
-      "grad_norm": 0.0011814252939075232,
-      "learning_rate": 0.0001999993434731937,
-      "loss": 46.0,
-      "step": 15096
-    },
-    {
-      "epoch": 1.1542710782346082,
-      "grad_norm": 0.0007847616798244417,
-      "learning_rate": 0.00019999934338615303,
-      "loss": 46.0,
-      "step": 15097
-    },
-    {
-      "epoch": 1.154347535217998,
-      "grad_norm": 0.001342223258689046,
-      "learning_rate": 0.00019999934329910656,
-      "loss": 46.0,
-      "step": 15098
-    },
-    {
-      "epoch": 1.1544239922013877,
-      "grad_norm": 0.0013589911395683885,
-      "learning_rate": 0.0001999993432120544,
-      "loss": 46.0,
-      "step": 15099
-    },
-    {
-      "epoch": 1.1545004491847775,
-      "grad_norm": 0.0010037136962637305,
-      "learning_rate": 0.00019999934312499637,
-      "loss": 46.0,
-      "step": 15100
-    },
-    {
-      "epoch": 1.1545769061681672,
-      "grad_norm": 0.0028625307604670525,
-      "learning_rate": 0.0001999993430379326,
-      "loss": 46.0,
-      "step": 15101
-    },
-    {
-      "epoch": 1.154653363151557,
-      "grad_norm": 0.005549272056668997,
-      "learning_rate": 0.0001999993429508631,
-      "loss": 46.0,
-      "step": 15102
-    },
-    {
-      "epoch": 1.1547298201349465,
-      "grad_norm": 0.0011536511592566967,
-      "learning_rate": 0.0001999993428637878,
-      "loss": 46.0,
-      "step": 15103
-    },
-    {
-      "epoch": 1.1548062771183363,
-      "grad_norm": 0.0021739182993769646,
-      "learning_rate": 0.00019999934277670673,
-      "loss": 46.0,
-      "step": 15104
-    },
-    {
-      "epoch": 1.154882734101726,
-      "grad_norm": 0.0008432441391050816,
-      "learning_rate": 0.0001999993426896199,
-      "loss": 46.0,
-      "step": 15105
-    },
-    {
-      "epoch": 1.1549591910851158,
-      "grad_norm": 0.000681453850120306,
-      "learning_rate": 0.0001999993426025273,
-      "loss": 46.0,
-      "step": 15106
-    },
-    {
-      "epoch": 1.1550356480685053,
-      "grad_norm": 0.0010257838293910027,
-      "learning_rate": 0.0001999993425154289,
-      "loss": 46.0,
-      "step": 15107
-    },
-    {
-      "epoch": 1.155112105051895,
-      "grad_norm": 0.007447650656104088,
-      "learning_rate": 0.00019999934242832476,
-      "loss": 46.0,
-      "step": 15108
-    },
-    {
-      "epoch": 1.1551885620352849,
-      "grad_norm": 0.0004147738800384104,
-      "learning_rate": 0.00019999934234121486,
-      "loss": 46.0,
-      "step": 15109
-    },
-    {
-      "epoch": 1.1552650190186746,
-      "grad_norm": 0.0017721919575706124,
-      "learning_rate": 0.0001999993422540992,
-      "loss": 46.0,
-      "step": 15110
-    },
-    {
-      "epoch": 1.1553414760020644,
-      "grad_norm": 0.005435622297227383,
-      "learning_rate": 0.00019999934216697774,
-      "loss": 46.0,
-      "step": 15111
-    },
-    {
-      "epoch": 1.1554179329854541,
-      "grad_norm": 0.002470600651577115,
-      "learning_rate": 0.0001999993420798505,
-      "loss": 46.0,
-      "step": 15112
-    },
-    {
-      "epoch": 1.1554943899688437,
-      "grad_norm": 0.001980425789952278,
-      "learning_rate": 0.00019999934199271753,
-      "loss": 46.0,
-      "step": 15113
-    },
-    {
-      "epoch": 1.1555708469522334,
-      "grad_norm": 0.0009720762027427554,
-      "learning_rate": 0.00019999934190557874,
-      "loss": 46.0,
-      "step": 15114
-    },
-    {
-      "epoch": 1.1556473039356232,
-      "grad_norm": 0.000672674854286015,
-      "learning_rate": 0.00019999934181843423,
-      "loss": 46.0,
-      "step": 15115
-    },
-    {
-      "epoch": 1.155723760919013,
-      "grad_norm": 0.0014648701762780547,
-      "learning_rate": 0.00019999934173128394,
-      "loss": 46.0,
-      "step": 15116
-    },
-    {
-      "epoch": 1.1558002179024027,
-      "grad_norm": 0.005455412901937962,
-      "learning_rate": 0.00019999934164412786,
-      "loss": 46.0,
-      "step": 15117
-    },
-    {
-      "epoch": 1.1558766748857923,
-      "grad_norm": 0.0016148581635206938,
-      "learning_rate": 0.00019999934155696603,
-      "loss": 46.0,
-      "step": 15118
-    },
-    {
-      "epoch": 1.155953131869182,
-      "grad_norm": 0.006246815901249647,
-      "learning_rate": 0.00019999934146979842,
-      "loss": 46.0,
-      "step": 15119
-    },
-    {
-      "epoch": 1.1560295888525718,
-      "grad_norm": 0.0014239230658859015,
-      "learning_rate": 0.00019999934138262504,
-      "loss": 46.0,
-      "step": 15120
-    },
-    {
-      "epoch": 1.1561060458359615,
-      "grad_norm": 0.0011944967554882169,
-      "learning_rate": 0.00019999934129544592,
-      "loss": 46.0,
-      "step": 15121
-    },
-    {
-      "epoch": 1.1561825028193513,
-      "grad_norm": 0.0007652255007997155,
-      "learning_rate": 0.000199999341208261,
-      "loss": 46.0,
-      "step": 15122
-    },
-    {
-      "epoch": 1.156258959802741,
-      "grad_norm": 0.0005660508759319782,
-      "learning_rate": 0.0001999993411210703,
-      "loss": 46.0,
-      "step": 15123
-    },
-    {
-      "epoch": 1.1563354167861306,
-      "grad_norm": 0.0013085348764434457,
-      "learning_rate": 0.00019999934103387385,
-      "loss": 46.0,
-      "step": 15124
-    },
-    {
-      "epoch": 1.1564118737695204,
-      "grad_norm": 0.0021195111330598593,
-      "learning_rate": 0.0001999993409466716,
-      "loss": 46.0,
-      "step": 15125
-    },
-    {
-      "epoch": 1.1564883307529101,
-      "grad_norm": 0.0313098318874836,
-      "learning_rate": 0.00019999934085946364,
-      "loss": 46.0,
-      "step": 15126
-    },
-    {
-      "epoch": 1.1565647877363,
-      "grad_norm": 0.009222094900906086,
-      "learning_rate": 0.00019999934077224988,
-      "loss": 46.0,
-      "step": 15127
-    },
-    {
-      "epoch": 1.1566412447196897,
-      "grad_norm": 0.0013312717201188207,
-      "learning_rate": 0.0001999993406850303,
-      "loss": 46.0,
-      "step": 15128
-    },
-    {
-      "epoch": 1.1567177017030792,
-      "grad_norm": 0.0012715579941868782,
-      "learning_rate": 0.00019999934059780503,
-      "loss": 46.0,
-      "step": 15129
-    },
-    {
-      "epoch": 1.156794158686469,
-      "grad_norm": 0.0012400245759636164,
-      "learning_rate": 0.00019999934051057397,
-      "loss": 46.0,
-      "step": 15130
-    },
-    {
-      "epoch": 1.1568706156698587,
-      "grad_norm": 0.0012981371255591512,
-      "learning_rate": 0.0001999993404233371,
-      "loss": 46.0,
-      "step": 15131
-    },
-    {
-      "epoch": 1.1569470726532485,
-      "grad_norm": 0.004702863283455372,
-      "learning_rate": 0.0001999993403360945,
-      "loss": 46.0,
-      "step": 15132
-    },
-    {
-      "epoch": 1.1570235296366382,
-      "grad_norm": 0.0005895852227695286,
-      "learning_rate": 0.00019999934024884613,
-      "loss": 46.0,
-      "step": 15133
-    },
-    {
-      "epoch": 1.157099986620028,
-      "grad_norm": 0.004708996042609215,
-      "learning_rate": 0.00019999934016159198,
-      "loss": 46.0,
-      "step": 15134
-    },
-    {
-      "epoch": 1.1571764436034175,
-      "grad_norm": 0.001433222321793437,
-      "learning_rate": 0.00019999934007433208,
-      "loss": 46.0,
-      "step": 15135
-    },
-    {
-      "epoch": 1.1572529005868073,
-      "grad_norm": 0.000706484483089298,
-      "learning_rate": 0.00019999933998706638,
-      "loss": 46.0,
-      "step": 15136
-    },
-    {
-      "epoch": 1.157329357570197,
-      "grad_norm": 0.00033455152879469097,
-      "learning_rate": 0.00019999933989979494,
-      "loss": 46.0,
-      "step": 15137
-    },
-    {
-      "epoch": 1.1574058145535868,
-      "grad_norm": 0.0008692506817169487,
-      "learning_rate": 0.0001999993398125177,
-      "loss": 46.0,
-      "step": 15138
-    },
-    {
-      "epoch": 1.1574822715369766,
-      "grad_norm": 0.002206777920946479,
-      "learning_rate": 0.0001999993397252347,
-      "loss": 46.0,
-      "step": 15139
-    },
-    {
-      "epoch": 1.1575587285203661,
-      "grad_norm": 0.0025811726227402687,
-      "learning_rate": 0.00019999933963794594,
-      "loss": 46.0,
-      "step": 15140
-    },
-    {
-      "epoch": 1.1576351855037559,
-      "grad_norm": 0.0015727608697488904,
-      "learning_rate": 0.0001999993395506514,
-      "loss": 46.0,
-      "step": 15141
-    },
-    {
-      "epoch": 1.1577116424871456,
-      "grad_norm": 0.0016923267394304276,
-      "learning_rate": 0.00019999933946335109,
-      "loss": 46.0,
-      "step": 15142
-    },
-    {
-      "epoch": 1.1577880994705354,
-      "grad_norm": 0.0017254312988370657,
-      "learning_rate": 0.00019999933937604503,
-      "loss": 46.0,
-      "step": 15143
-    },
-    {
-      "epoch": 1.1578645564539252,
-      "grad_norm": 0.0008460857206955552,
-      "learning_rate": 0.00019999933928873317,
-      "loss": 46.0,
-      "step": 15144
-    },
-    {
-      "epoch": 1.157941013437315,
-      "grad_norm": 0.0011523258872330189,
-      "learning_rate": 0.0001999993392014156,
-      "loss": 46.0,
-      "step": 15145
-    },
-    {
-      "epoch": 1.1580174704207045,
-      "grad_norm": 0.0005783882807008922,
-      "learning_rate": 0.0001999993391140922,
-      "loss": 46.0,
-      "step": 15146
-    },
-    {
-      "epoch": 1.1580939274040942,
-      "grad_norm": 0.0016667454037815332,
-      "learning_rate": 0.00019999933902676304,
-      "loss": 46.0,
-      "step": 15147
-    },
-    {
-      "epoch": 1.158170384387484,
-      "grad_norm": 0.001644982141442597,
-      "learning_rate": 0.00019999933893942814,
-      "loss": 46.0,
-      "step": 15148
-    },
-    {
-      "epoch": 1.1582468413708737,
-      "grad_norm": 0.0028538370970636606,
-      "learning_rate": 0.00019999933885208741,
-      "loss": 46.0,
-      "step": 15149
-    },
-    {
-      "epoch": 1.1583232983542635,
-      "grad_norm": 0.0012513879919424653,
-      "learning_rate": 0.00019999933876474097,
-      "loss": 46.0,
-      "step": 15150
-    },
-    {
-      "epoch": 1.158399755337653,
-      "grad_norm": 0.001760341809131205,
-      "learning_rate": 0.00019999933867738875,
-      "loss": 46.0,
-      "step": 15151
-    },
-    {
-      "epoch": 1.1584762123210428,
-      "grad_norm": 0.0009005492902360857,
-      "learning_rate": 0.00019999933859003076,
-      "loss": 46.0,
-      "step": 15152
-    },
-    {
-      "epoch": 1.1585526693044326,
-      "grad_norm": 0.01334104873239994,
-      "learning_rate": 0.00019999933850266697,
-      "loss": 46.0,
-      "step": 15153
-    },
-    {
-      "epoch": 1.1586291262878223,
-      "grad_norm": 0.002504281932488084,
-      "learning_rate": 0.00019999933841529743,
-      "loss": 46.0,
-      "step": 15154
-    },
-    {
-      "epoch": 1.158705583271212,
-      "grad_norm": 0.0009778373641893268,
-      "learning_rate": 0.00019999933832792212,
-      "loss": 46.0,
-      "step": 15155
-    },
-    {
-      "epoch": 1.1587820402546019,
-      "grad_norm": 0.0006808859179727733,
-      "learning_rate": 0.00019999933824054104,
-      "loss": 46.0,
-      "step": 15156
-    },
-    {
-      "epoch": 1.1588584972379914,
-      "grad_norm": 0.0021752764005213976,
-      "learning_rate": 0.0001999993381531542,
-      "loss": 46.0,
-      "step": 15157
-    },
-    {
-      "epoch": 1.1589349542213812,
-      "grad_norm": 0.0010882660280913115,
-      "learning_rate": 0.0001999993380657616,
-      "loss": 46.0,
-      "step": 15158
-    },
-    {
-      "epoch": 1.159011411204771,
-      "grad_norm": 0.000635226140730083,
-      "learning_rate": 0.0001999993379783632,
-      "loss": 46.0,
-      "step": 15159
-    },
-    {
-      "epoch": 1.1590878681881607,
-      "grad_norm": 0.0009248583228327334,
-      "learning_rate": 0.00019999933789095905,
-      "loss": 46.0,
-      "step": 15160
-    },
-    {
-      "epoch": 1.1591643251715504,
-      "grad_norm": 0.0019124189857393503,
-      "learning_rate": 0.00019999933780354912,
-      "loss": 46.0,
-      "step": 15161
-    },
-    {
-      "epoch": 1.15924078215494,
-      "grad_norm": 0.0016719320556148887,
-      "learning_rate": 0.00019999933771613345,
-      "loss": 46.0,
-      "step": 15162
-    },
-    {
-      "epoch": 1.1593172391383297,
-      "grad_norm": 0.0021364176645874977,
-      "learning_rate": 0.00019999933762871198,
-      "loss": 46.0,
-      "step": 15163
-    },
-    {
-      "epoch": 1.1593936961217195,
-      "grad_norm": 0.0009226591791957617,
-      "learning_rate": 0.00019999933754128474,
-      "loss": 46.0,
-      "step": 15164
-    },
-    {
-      "epoch": 1.1594701531051093,
-      "grad_norm": 0.0008954951772466302,
-      "learning_rate": 0.00019999933745385175,
-      "loss": 46.0,
-      "step": 15165
-    },
-    {
-      "epoch": 1.159546610088499,
-      "grad_norm": 0.02052871510386467,
-      "learning_rate": 0.00019999933736641298,
-      "loss": 46.0,
-      "step": 15166
-    },
-    {
-      "epoch": 1.1596230670718888,
-      "grad_norm": 0.00204788101837039,
-      "learning_rate": 0.00019999933727896845,
-      "loss": 46.0,
-      "step": 15167
-    },
-    {
-      "epoch": 1.1596995240552783,
-      "grad_norm": 0.0005477396189235151,
-      "learning_rate": 0.00019999933719151814,
-      "loss": 46.0,
-      "step": 15168
-    },
-    {
-      "epoch": 1.159775981038668,
-      "grad_norm": 0.008746067062020302,
-      "learning_rate": 0.00019999933710406208,
-      "loss": 46.0,
-      "step": 15169
-    },
-    {
-      "epoch": 1.1598524380220578,
-      "grad_norm": 0.012607166543602943,
-      "learning_rate": 0.00019999933701660022,
-      "loss": 46.0,
-      "step": 15170
-    },
-    {
-      "epoch": 1.1599288950054476,
-      "grad_norm": 0.001968152355402708,
-      "learning_rate": 0.0001999993369291326,
-      "loss": 46.0,
-      "step": 15171
-    },
-    {
-      "epoch": 1.1600053519888374,
-      "grad_norm": 0.0019370965892449021,
-      "learning_rate": 0.00019999933684165921,
-      "loss": 46.0,
-      "step": 15172
-    },
-    {
-      "epoch": 1.160081808972227,
-      "grad_norm": 0.0014788837870582938,
-      "learning_rate": 0.00019999933675418006,
-      "loss": 46.0,
-      "step": 15173
-    },
-    {
-      "epoch": 1.1601582659556167,
-      "grad_norm": 0.002217283006757498,
-      "learning_rate": 0.00019999933666669514,
-      "loss": 46.0,
-      "step": 15174
-    },
-    {
-      "epoch": 1.1602347229390064,
-      "grad_norm": 0.0005758915795013309,
-      "learning_rate": 0.00019999933657920444,
-      "loss": 46.0,
-      "step": 15175
-    },
-    {
-      "epoch": 1.1603111799223962,
-      "grad_norm": 0.0012430837377905846,
-      "learning_rate": 0.00019999933649170797,
-      "loss": 46.0,
-      "step": 15176
-    },
-    {
-      "epoch": 1.160387636905786,
-      "grad_norm": 0.001995112979784608,
-      "learning_rate": 0.00019999933640420573,
-      "loss": 46.0,
-      "step": 15177
-    },
-    {
-      "epoch": 1.1604640938891757,
-      "grad_norm": 0.001302272779867053,
-      "learning_rate": 0.00019999933631669774,
-      "loss": 46.0,
-      "step": 15178
-    },
-    {
-      "epoch": 1.1605405508725652,
-      "grad_norm": 0.0006122771301306784,
-      "learning_rate": 0.00019999933622918397,
-      "loss": 46.0,
-      "step": 15179
-    },
-    {
-      "epoch": 1.160617007855955,
-      "grad_norm": 0.0008690704125910997,
-      "learning_rate": 0.0001999993361416644,
-      "loss": 46.0,
-      "step": 15180
-    },
-    {
-      "epoch": 1.1606934648393448,
-      "grad_norm": 0.003428082913160324,
-      "learning_rate": 0.00019999933605413913,
-      "loss": 46.0,
-      "step": 15181
-    },
-    {
-      "epoch": 1.1607699218227345,
-      "grad_norm": 0.0006868264172226191,
-      "learning_rate": 0.00019999933596660804,
-      "loss": 46.0,
-      "step": 15182
-    },
-    {
-      "epoch": 1.1608463788061243,
-      "grad_norm": 0.0018730575684458017,
-      "learning_rate": 0.00019999933587907119,
-      "loss": 46.0,
-      "step": 15183
-    },
-    {
-      "epoch": 1.1609228357895138,
-      "grad_norm": 0.0013282742584124207,
-      "learning_rate": 0.00019999933579152858,
-      "loss": 46.0,
-      "step": 15184
-    },
-    {
-      "epoch": 1.1609992927729036,
-      "grad_norm": 0.0035859534982591867,
-      "learning_rate": 0.00019999933570398018,
-      "loss": 46.0,
-      "step": 15185
-    },
-    {
-      "epoch": 1.1610757497562934,
-      "grad_norm": 0.000959887751378119,
-      "learning_rate": 0.00019999933561642603,
-      "loss": 46.0,
-      "step": 15186
-    },
-    {
-      "epoch": 1.1611522067396831,
-      "grad_norm": 0.0009075164562091231,
-      "learning_rate": 0.0001999993355288661,
-      "loss": 46.0,
-      "step": 15187
-    },
-    {
-      "epoch": 1.1612286637230729,
-      "grad_norm": 0.0012018652632832527,
-      "learning_rate": 0.00019999933544130038,
-      "loss": 46.0,
-      "step": 15188
-    },
-    {
-      "epoch": 1.1613051207064626,
-      "grad_norm": 0.016878055408596992,
-      "learning_rate": 0.0001999993353537289,
-      "loss": 46.0,
-      "step": 15189
-    },
-    {
-      "epoch": 1.1613815776898522,
-      "grad_norm": 0.0034380739089101553,
-      "learning_rate": 0.0001999993352661517,
-      "loss": 46.0,
-      "step": 15190
-    },
-    {
-      "epoch": 1.161458034673242,
-      "grad_norm": 0.0026983842253684998,
-      "learning_rate": 0.00019999933517856868,
-      "loss": 46.0,
-      "step": 15191
-    },
-    {
-      "epoch": 1.1615344916566317,
-      "grad_norm": 0.0031875669956207275,
-      "learning_rate": 0.00019999933509097991,
-      "loss": 46.0,
-      "step": 15192
-    },
-    {
-      "epoch": 1.1616109486400215,
-      "grad_norm": 0.0016166269779205322,
-      "learning_rate": 0.00019999933500338538,
-      "loss": 46.0,
-      "step": 15193
-    },
-    {
-      "epoch": 1.1616874056234112,
-      "grad_norm": 0.00036780114169232547,
-      "learning_rate": 0.00019999933491578507,
-      "loss": 46.0,
-      "step": 15194
-    },
-    {
-      "epoch": 1.1617638626068008,
-      "grad_norm": 0.001250248751603067,
-      "learning_rate": 0.00019999933482817896,
-      "loss": 46.0,
-      "step": 15195
-    },
-    {
-      "epoch": 1.1618403195901905,
-      "grad_norm": 0.0019274428486824036,
-      "learning_rate": 0.00019999933474056713,
-      "loss": 46.0,
-      "step": 15196
-    },
-    {
-      "epoch": 1.1619167765735803,
-      "grad_norm": 0.0012433610390871763,
-      "learning_rate": 0.0001999993346529495,
-      "loss": 46.0,
-      "step": 15197
-    },
-    {
-      "epoch": 1.16199323355697,
-      "grad_norm": 0.0017420609947293997,
-      "learning_rate": 0.00019999933456532612,
-      "loss": 46.0,
-      "step": 15198
-    },
-    {
-      "epoch": 1.1620696905403598,
-      "grad_norm": 0.0038535750936716795,
-      "learning_rate": 0.00019999933447769697,
-      "loss": 46.0,
-      "step": 15199
-    },
-    {
-      "epoch": 1.1621461475237496,
-      "grad_norm": 0.0011651928070932627,
-      "learning_rate": 0.00019999933439006202,
-      "loss": 46.0,
-      "step": 15200
-    },
-    {
-      "epoch": 1.162222604507139,
-      "grad_norm": 0.003509982954710722,
-      "learning_rate": 0.00019999933430242132,
-      "loss": 46.0,
-      "step": 15201
-    },
-    {
-      "epoch": 1.1622990614905289,
-      "grad_norm": 0.010439963079988956,
-      "learning_rate": 0.00019999933421477485,
-      "loss": 46.0,
-      "step": 15202
-    },
-    {
-      "epoch": 1.1623755184739186,
-      "grad_norm": 0.002135409275069833,
-      "learning_rate": 0.0001999993341271226,
-      "loss": 46.0,
-      "step": 15203
-    },
-    {
-      "epoch": 1.1624519754573084,
-      "grad_norm": 0.001027404796332121,
-      "learning_rate": 0.00019999933403946462,
-      "loss": 46.0,
-      "step": 15204
-    },
-    {
-      "epoch": 1.1625284324406981,
-      "grad_norm": 0.0007074300665408373,
-      "learning_rate": 0.00019999933395180083,
-      "loss": 46.0,
-      "step": 15205
-    },
-    {
-      "epoch": 1.1626048894240877,
-      "grad_norm": 0.001136760925874114,
-      "learning_rate": 0.00019999933386413127,
-      "loss": 46.0,
-      "step": 15206
-    },
-    {
-      "epoch": 1.1626813464074774,
-      "grad_norm": 0.0005381845403462648,
-      "learning_rate": 0.00019999933377645599,
-      "loss": 46.0,
-      "step": 15207
-    },
-    {
-      "epoch": 1.1627578033908672,
-      "grad_norm": 0.0016024968354031444,
-      "learning_rate": 0.00019999933368877487,
-      "loss": 46.0,
-      "step": 15208
-    },
-    {
-      "epoch": 1.162834260374257,
-      "grad_norm": 0.0007701287395320833,
-      "learning_rate": 0.00019999933360108805,
-      "loss": 46.0,
-      "step": 15209
-    },
-    {
-      "epoch": 1.1629107173576467,
-      "grad_norm": 0.0010702564613893628,
-      "learning_rate": 0.00019999933351339542,
-      "loss": 46.0,
-      "step": 15210
-    },
-    {
-      "epoch": 1.1629871743410365,
-      "grad_norm": 0.002136446302756667,
-      "learning_rate": 0.000199999333425697,
-      "loss": 46.0,
-      "step": 15211
-    },
-    {
-      "epoch": 1.163063631324426,
-      "grad_norm": 0.0020883698016405106,
-      "learning_rate": 0.00019999933333799286,
-      "loss": 46.0,
-      "step": 15212
-    },
-    {
-      "epoch": 1.1631400883078158,
-      "grad_norm": 0.002453526249155402,
-      "learning_rate": 0.00019999933325028294,
-      "loss": 46.0,
-      "step": 15213
-    },
-    {
-      "epoch": 1.1632165452912056,
-      "grad_norm": 0.0009737057844176888,
-      "learning_rate": 0.00019999933316256722,
-      "loss": 46.0,
-      "step": 15214
-    },
-    {
-      "epoch": 1.1632930022745953,
-      "grad_norm": 0.0015718861250206828,
-      "learning_rate": 0.00019999933307484575,
-      "loss": 46.0,
-      "step": 15215
-    },
-    {
-      "epoch": 1.163369459257985,
-      "grad_norm": 0.0008530642371624708,
-      "learning_rate": 0.0001999993329871185,
-      "loss": 46.0,
-      "step": 15216
-    },
-    {
-      "epoch": 1.1634459162413746,
-      "grad_norm": 0.0011651882668957114,
-      "learning_rate": 0.00019999933289938551,
-      "loss": 46.0,
-      "step": 15217
-    },
-    {
-      "epoch": 1.1635223732247644,
-      "grad_norm": 0.003901093266904354,
-      "learning_rate": 0.00019999933281164673,
-      "loss": 46.0,
-      "step": 15218
-    },
-    {
-      "epoch": 1.1635988302081541,
-      "grad_norm": 0.0011217995779588819,
-      "learning_rate": 0.00019999933272390216,
-      "loss": 46.0,
-      "step": 15219
-    },
-    {
-      "epoch": 1.163675287191544,
-      "grad_norm": 0.0007204018183983862,
-      "learning_rate": 0.00019999933263615188,
-      "loss": 46.0,
-      "step": 15220
-    },
-    {
-      "epoch": 1.1637517441749337,
-      "grad_norm": 0.0008575300453230739,
-      "learning_rate": 0.00019999933254839577,
-      "loss": 46.0,
-      "step": 15221
-    },
-    {
-      "epoch": 1.1638282011583234,
-      "grad_norm": 0.0011869131121784449,
-      "learning_rate": 0.0001999993324606339,
-      "loss": 46.0,
-      "step": 15222
-    },
-    {
-      "epoch": 1.163904658141713,
-      "grad_norm": 0.0005873568588867784,
-      "learning_rate": 0.0001999993323728663,
-      "loss": 46.0,
-      "step": 15223
-    },
-    {
-      "epoch": 1.1639811151251027,
-      "grad_norm": 0.0010079137282446027,
-      "learning_rate": 0.00019999933228509288,
-      "loss": 46.0,
-      "step": 15224
-    },
-    {
-      "epoch": 1.1640575721084925,
-      "grad_norm": 0.0026000505313277245,
-      "learning_rate": 0.00019999933219731373,
-      "loss": 46.0,
-      "step": 15225
-    },
-    {
-      "epoch": 1.1641340290918822,
-      "grad_norm": 0.0026131710037589073,
-      "learning_rate": 0.00019999933210952878,
-      "loss": 46.0,
-      "step": 15226
-    },
-    {
-      "epoch": 1.1642104860752718,
-      "grad_norm": 0.0009959496092051268,
-      "learning_rate": 0.00019999933202173809,
-      "loss": 46.0,
-      "step": 15227
-    },
-    {
-      "epoch": 1.1642869430586615,
-      "grad_norm": 0.0009695867775008082,
-      "learning_rate": 0.00019999933193394162,
-      "loss": 46.0,
-      "step": 15228
-    },
-    {
-      "epoch": 1.1643634000420513,
-      "grad_norm": 0.0026348461396992207,
-      "learning_rate": 0.00019999933184613937,
-      "loss": 46.0,
-      "step": 15229
-    },
-    {
-      "epoch": 1.164439857025441,
-      "grad_norm": 0.0005475819925777614,
-      "learning_rate": 0.00019999933175833136,
-      "loss": 46.0,
-      "step": 15230
-    },
-    {
-      "epoch": 1.1645163140088308,
-      "grad_norm": 0.0009518524166196585,
-      "learning_rate": 0.0001999993316705176,
-      "loss": 46.0,
-      "step": 15231
-    },
-    {
-      "epoch": 1.1645927709922206,
-      "grad_norm": 0.0016149907605722547,
-      "learning_rate": 0.000199999331582698,
-      "loss": 46.0,
-      "step": 15232
-    },
-    {
-      "epoch": 1.1646692279756103,
-      "grad_norm": 0.0010324977338314056,
-      "learning_rate": 0.0001999993314948727,
-      "loss": 46.0,
-      "step": 15233
-    },
-    {
-      "epoch": 1.1647456849589999,
-      "grad_norm": 0.0015724231489002705,
-      "learning_rate": 0.00019999933140704161,
-      "loss": 46.0,
-      "step": 15234
-    },
-    {
-      "epoch": 1.1648221419423896,
-      "grad_norm": 0.0008942470885813236,
-      "learning_rate": 0.00019999933131920476,
-      "loss": 46.0,
-      "step": 15235
-    },
-    {
-      "epoch": 1.1648985989257794,
-      "grad_norm": 0.0013806776842102408,
-      "learning_rate": 0.00019999933123136213,
-      "loss": 46.0,
-      "step": 15236
-    },
-    {
-      "epoch": 1.1649750559091692,
-      "grad_norm": 0.001410484779626131,
-      "learning_rate": 0.00019999933114351373,
-      "loss": 46.0,
-      "step": 15237
-    },
-    {
-      "epoch": 1.1650515128925587,
-      "grad_norm": 0.0005561124416999519,
-      "learning_rate": 0.00019999933105565955,
-      "loss": 46.0,
-      "step": 15238
-    },
-    {
-      "epoch": 1.1651279698759485,
-      "grad_norm": 0.0013579585356637836,
-      "learning_rate": 0.0001999993309677996,
-      "loss": 46.0,
-      "step": 15239
-    },
-    {
-      "epoch": 1.1652044268593382,
-      "grad_norm": 0.004409249406307936,
-      "learning_rate": 0.00019999933087993393,
-      "loss": 46.0,
-      "step": 15240
-    },
-    {
-      "epoch": 1.165280883842728,
-      "grad_norm": 0.0008311390993185341,
-      "learning_rate": 0.00019999933079206246,
-      "loss": 46.0,
-      "step": 15241
-    },
-    {
-      "epoch": 1.1653573408261177,
-      "grad_norm": 0.0010167277650907636,
-      "learning_rate": 0.0001999993307041852,
-      "loss": 46.0,
-      "step": 15242
-    },
-    {
-      "epoch": 1.1654337978095075,
-      "grad_norm": 0.001312794047407806,
-      "learning_rate": 0.00019999933061630218,
-      "loss": 46.0,
-      "step": 15243
-    },
-    {
-      "epoch": 1.165510254792897,
-      "grad_norm": 0.002374960808083415,
-      "learning_rate": 0.0001999993305284134,
-      "loss": 46.0,
-      "step": 15244
-    },
-    {
-      "epoch": 1.1655867117762868,
-      "grad_norm": 0.001582706463523209,
-      "learning_rate": 0.00019999933044051885,
-      "loss": 46.0,
-      "step": 15245
-    },
-    {
-      "epoch": 1.1656631687596766,
-      "grad_norm": 0.0022503496147692204,
-      "learning_rate": 0.00019999933035261852,
-      "loss": 46.0,
-      "step": 15246
-    },
-    {
-      "epoch": 1.1657396257430663,
-      "grad_norm": 0.001016007736325264,
-      "learning_rate": 0.00019999933026471244,
-      "loss": 46.0,
-      "step": 15247
-    },
-    {
-      "epoch": 1.165816082726456,
-      "grad_norm": 0.0012851026840507984,
-      "learning_rate": 0.00019999933017680058,
-      "loss": 46.0,
-      "step": 15248
-    },
-    {
-      "epoch": 1.1658925397098456,
-      "grad_norm": 0.0005490654148161411,
-      "learning_rate": 0.00019999933008888295,
-      "loss": 46.0,
-      "step": 15249
-    },
-    {
-      "epoch": 1.1659689966932354,
-      "grad_norm": 0.003530229674652219,
-      "learning_rate": 0.00019999933000095955,
-      "loss": 46.0,
-      "step": 15250
-    },
-    {
-      "epoch": 1.1660454536766252,
-      "grad_norm": 0.0014833117602393031,
-      "learning_rate": 0.00019999932991303037,
-      "loss": 46.0,
-      "step": 15251
-    },
-    {
-      "epoch": 1.166121910660015,
-      "grad_norm": 0.0028618767391890287,
-      "learning_rate": 0.00019999932982509542,
-      "loss": 46.0,
-      "step": 15252
-    },
-    {
-      "epoch": 1.1661983676434047,
-      "grad_norm": 0.001250334084033966,
-      "learning_rate": 0.00019999932973715473,
-      "loss": 46.0,
-      "step": 15253
-    },
-    {
-      "epoch": 1.1662748246267944,
-      "grad_norm": 0.0014182317536324263,
-      "learning_rate": 0.00019999932964920823,
-      "loss": 46.0,
-      "step": 15254
-    },
-    {
-      "epoch": 1.166351281610184,
-      "grad_norm": 0.0007662430871278048,
-      "learning_rate": 0.000199999329561256,
-      "loss": 46.0,
-      "step": 15255
-    },
-    {
-      "epoch": 1.1664277385935737,
-      "grad_norm": 0.001694596721790731,
-      "learning_rate": 0.00019999932947329798,
-      "loss": 46.0,
-      "step": 15256
-    },
-    {
-      "epoch": 1.1665041955769635,
-      "grad_norm": 0.0006246630218811333,
-      "learning_rate": 0.00019999932938533421,
-      "loss": 46.0,
-      "step": 15257
-    },
-    {
-      "epoch": 1.1665806525603533,
-      "grad_norm": 0.00046615154133178294,
-      "learning_rate": 0.00019999932929736463,
-      "loss": 46.0,
-      "step": 15258
-    },
-    {
-      "epoch": 1.166657109543743,
-      "grad_norm": 0.0010403483174741268,
-      "learning_rate": 0.00019999932920938932,
-      "loss": 46.0,
-      "step": 15259
-    },
-    {
-      "epoch": 1.1667335665271326,
-      "grad_norm": 0.0010131682502105832,
-      "learning_rate": 0.0001999993291214082,
-      "loss": 46.0,
-      "step": 15260
-    },
-    {
-      "epoch": 1.1668100235105223,
-      "grad_norm": 0.0009167783427983522,
-      "learning_rate": 0.00019999932903342135,
-      "loss": 46.0,
-      "step": 15261
-    },
-    {
-      "epoch": 1.166886480493912,
-      "grad_norm": 0.0019855392165482044,
-      "learning_rate": 0.00019999932894542872,
-      "loss": 46.0,
-      "step": 15262
-    },
-    {
-      "epoch": 1.1669629374773018,
-      "grad_norm": 0.0005558494594879448,
-      "learning_rate": 0.00019999932885743032,
-      "loss": 46.0,
-      "step": 15263
-    },
-    {
-      "epoch": 1.1670393944606916,
-      "grad_norm": 0.000635394302662462,
-      "learning_rate": 0.00019999932876942615,
-      "loss": 46.0,
-      "step": 15264
-    },
-    {
-      "epoch": 1.1671158514440814,
-      "grad_norm": 0.03312766179442406,
-      "learning_rate": 0.00019999932868141622,
-      "loss": 46.0,
-      "step": 15265
-    },
-    {
-      "epoch": 1.167192308427471,
-      "grad_norm": 0.0019652668852359056,
-      "learning_rate": 0.0001999993285934005,
-      "loss": 46.0,
-      "step": 15266
-    },
-    {
-      "epoch": 1.1672687654108607,
-      "grad_norm": 0.0008170937653630972,
-      "learning_rate": 0.000199999328505379,
-      "loss": 46.0,
-      "step": 15267
-    },
-    {
-      "epoch": 1.1673452223942504,
-      "grad_norm": 0.0009267200948670506,
-      "learning_rate": 0.00019999932841735177,
-      "loss": 46.0,
-      "step": 15268
-    },
-    {
-      "epoch": 1.1674216793776402,
-      "grad_norm": 0.001527606975287199,
-      "learning_rate": 0.00019999932832931875,
-      "loss": 46.0,
-      "step": 15269
-    },
-    {
-      "epoch": 1.16749813636103,
-      "grad_norm": 0.003069679019972682,
-      "learning_rate": 0.00019999932824127996,
-      "loss": 46.0,
-      "step": 15270
-    },
-    {
-      "epoch": 1.1675745933444195,
-      "grad_norm": 0.001260369666852057,
-      "learning_rate": 0.0001999993281532354,
-      "loss": 46.0,
-      "step": 15271
-    },
-    {
-      "epoch": 1.1676510503278092,
-      "grad_norm": 0.005360051989555359,
-      "learning_rate": 0.00019999932806518506,
-      "loss": 46.0,
-      "step": 15272
-    },
-    {
-      "epoch": 1.167727507311199,
-      "grad_norm": 0.0005134766106493771,
-      "learning_rate": 0.00019999932797712896,
-      "loss": 46.0,
-      "step": 15273
-    },
-    {
-      "epoch": 1.1678039642945888,
-      "grad_norm": 0.0006678166682831943,
-      "learning_rate": 0.0001999993278890671,
-      "loss": 46.0,
-      "step": 15274
-    },
-    {
-      "epoch": 1.1678804212779785,
-      "grad_norm": 0.0007406029617413878,
-      "learning_rate": 0.00019999932780099945,
-      "loss": 46.0,
-      "step": 15275
-    },
-    {
-      "epoch": 1.1679568782613683,
-      "grad_norm": 0.0006844205199740827,
-      "learning_rate": 0.00019999932771292607,
-      "loss": 46.0,
-      "step": 15276
-    },
-    {
-      "epoch": 1.1680333352447578,
-      "grad_norm": 0.0009097892907448113,
-      "learning_rate": 0.0001999993276248469,
-      "loss": 46.0,
-      "step": 15277
-    },
-    {
-      "epoch": 1.1681097922281476,
-      "grad_norm": 0.0005681724287569523,
-      "learning_rate": 0.00019999932753676195,
-      "loss": 46.0,
-      "step": 15278
-    },
-    {
-      "epoch": 1.1681862492115374,
-      "grad_norm": 0.010308535769581795,
-      "learning_rate": 0.00019999932744867125,
-      "loss": 46.0,
-      "step": 15279
-    },
-    {
-      "epoch": 1.1682627061949271,
-      "grad_norm": 0.00038089058944024146,
-      "learning_rate": 0.00019999932736057473,
-      "loss": 46.0,
-      "step": 15280
-    },
-    {
-      "epoch": 1.1683391631783169,
-      "grad_norm": 0.0005786475376226008,
-      "learning_rate": 0.0001999993272724725,
-      "loss": 46.0,
-      "step": 15281
-    },
-    {
-      "epoch": 1.1684156201617064,
-      "grad_norm": 0.009335752576589584,
-      "learning_rate": 0.0001999993271843645,
-      "loss": 46.0,
-      "step": 15282
-    },
-    {
-      "epoch": 1.1684920771450962,
-      "grad_norm": 0.0018412502249702811,
-      "learning_rate": 0.0001999993270962507,
-      "loss": 46.0,
-      "step": 15283
-    },
-    {
-      "epoch": 1.168568534128486,
-      "grad_norm": 0.000612473173532635,
-      "learning_rate": 0.00019999932700813113,
-      "loss": 46.0,
-      "step": 15284
-    },
-    {
-      "epoch": 1.1686449911118757,
-      "grad_norm": 0.0006646934198215604,
-      "learning_rate": 0.0001999993269200058,
-      "loss": 46.0,
-      "step": 15285
-    },
-    {
-      "epoch": 1.1687214480952655,
-      "grad_norm": 0.0009676276822574437,
-      "learning_rate": 0.0001999993268318747,
-      "loss": 46.0,
-      "step": 15286
-    },
-    {
-      "epoch": 1.1687979050786552,
-      "grad_norm": 0.0018264454556629062,
-      "learning_rate": 0.00019999932674373783,
-      "loss": 46.0,
-      "step": 15287
-    },
-    {
-      "epoch": 1.1688743620620448,
-      "grad_norm": 0.0006531497347168624,
-      "learning_rate": 0.0001999993266555952,
-      "loss": 46.0,
-      "step": 15288
-    },
-    {
-      "epoch": 1.1689508190454345,
-      "grad_norm": 0.004945653956383467,
-      "learning_rate": 0.00019999932656744677,
-      "loss": 46.0,
-      "step": 15289
-    },
-    {
-      "epoch": 1.1690272760288243,
-      "grad_norm": 0.010995850898325443,
-      "learning_rate": 0.0001999993264792926,
-      "loss": 46.0,
-      "step": 15290
-    },
-    {
-      "epoch": 1.169103733012214,
-      "grad_norm": 0.0026049462612718344,
-      "learning_rate": 0.00019999932639113265,
-      "loss": 46.0,
-      "step": 15291
-    },
-    {
-      "epoch": 1.1691801899956038,
-      "grad_norm": 0.0010155064519494772,
-      "learning_rate": 0.00019999932630296693,
-      "loss": 46.0,
-      "step": 15292
-    },
-    {
-      "epoch": 1.1692566469789933,
-      "grad_norm": 0.0028305945452302694,
-      "learning_rate": 0.00019999932621479546,
-      "loss": 46.0,
-      "step": 15293
-    },
-    {
-      "epoch": 1.169333103962383,
-      "grad_norm": 0.0050291153602302074,
-      "learning_rate": 0.0001999993261266182,
-      "loss": 46.0,
-      "step": 15294
-    },
-    {
-      "epoch": 1.1694095609457729,
-      "grad_norm": 0.0024297942873090506,
-      "learning_rate": 0.00019999932603843518,
-      "loss": 46.0,
-      "step": 15295
-    },
-    {
-      "epoch": 1.1694860179291626,
-      "grad_norm": 0.0013381576864048839,
-      "learning_rate": 0.0001999993259502464,
-      "loss": 46.0,
-      "step": 15296
-    },
-    {
-      "epoch": 1.1695624749125524,
-      "grad_norm": 0.0008254920248873532,
-      "learning_rate": 0.00019999932586205183,
-      "loss": 46.0,
-      "step": 15297
-    },
-    {
-      "epoch": 1.1696389318959421,
-      "grad_norm": 0.0008895752253010869,
-      "learning_rate": 0.0001999993257738515,
-      "loss": 46.0,
-      "step": 15298
-    },
-    {
-      "epoch": 1.1697153888793317,
-      "grad_norm": 0.0012544800993055105,
-      "learning_rate": 0.00019999932568564538,
-      "loss": 46.0,
-      "step": 15299
-    },
-    {
-      "epoch": 1.1697918458627214,
-      "grad_norm": 0.00042710232082754374,
-      "learning_rate": 0.0001999993255974335,
-      "loss": 46.0,
-      "step": 15300
-    },
-    {
-      "epoch": 1.1698683028461112,
-      "grad_norm": 0.0010928551200777292,
-      "learning_rate": 0.00019999932550921588,
-      "loss": 46.0,
-      "step": 15301
-    },
-    {
-      "epoch": 1.169944759829501,
-      "grad_norm": 0.0018931017257273197,
-      "learning_rate": 0.00019999932542099247,
-      "loss": 46.0,
-      "step": 15302
-    },
-    {
-      "epoch": 1.1700212168128907,
-      "grad_norm": 0.0011676972499117255,
-      "learning_rate": 0.0001999993253327633,
-      "loss": 46.0,
-      "step": 15303
-    },
-    {
-      "epoch": 1.1700976737962803,
-      "grad_norm": 0.0007247917237691581,
-      "learning_rate": 0.00019999932524452833,
-      "loss": 46.0,
-      "step": 15304
-    },
-    {
-      "epoch": 1.17017413077967,
-      "grad_norm": 0.001112442696467042,
-      "learning_rate": 0.00019999932515628763,
-      "loss": 46.0,
-      "step": 15305
-    },
-    {
-      "epoch": 1.1702505877630598,
-      "grad_norm": 0.0006946864305064082,
-      "learning_rate": 0.00019999932506804114,
-      "loss": 46.0,
-      "step": 15306
-    },
-    {
-      "epoch": 1.1703270447464496,
-      "grad_norm": 0.00289551867172122,
-      "learning_rate": 0.00019999932497978887,
-      "loss": 46.0,
-      "step": 15307
-    },
-    {
-      "epoch": 1.1704035017298393,
-      "grad_norm": 0.007863664999604225,
-      "learning_rate": 0.00019999932489153083,
-      "loss": 46.0,
-      "step": 15308
-    },
-    {
-      "epoch": 1.170479958713229,
-      "grad_norm": 0.004753883462399244,
-      "learning_rate": 0.00019999932480326704,
-      "loss": 46.0,
-      "step": 15309
-    },
-    {
-      "epoch": 1.1705564156966186,
-      "grad_norm": 0.004676952958106995,
-      "learning_rate": 0.0001999993247149975,
-      "loss": 46.0,
-      "step": 15310
-    },
-    {
-      "epoch": 1.1706328726800084,
-      "grad_norm": 0.0017181761795654893,
-      "learning_rate": 0.00019999932462672214,
-      "loss": 46.0,
-      "step": 15311
-    },
-    {
-      "epoch": 1.1707093296633981,
-      "grad_norm": 0.0011043314589187503,
-      "learning_rate": 0.00019999932453844106,
-      "loss": 46.0,
-      "step": 15312
-    },
-    {
-      "epoch": 1.170785786646788,
-      "grad_norm": 0.0007979258080013096,
-      "learning_rate": 0.00019999932445015416,
-      "loss": 46.0,
-      "step": 15313
-    },
-    {
-      "epoch": 1.1708622436301777,
-      "grad_norm": 0.0013872137060388923,
-      "learning_rate": 0.00019999932436186153,
-      "loss": 46.0,
-      "step": 15314
-    },
-    {
-      "epoch": 1.1709387006135672,
-      "grad_norm": 0.0009669964201748371,
-      "learning_rate": 0.00019999932427356313,
-      "loss": 46.0,
-      "step": 15315
-    },
-    {
-      "epoch": 1.171015157596957,
-      "grad_norm": 0.0008406496490351856,
-      "learning_rate": 0.00019999932418525895,
-      "loss": 46.0,
-      "step": 15316
-    },
-    {
-      "epoch": 1.1710916145803467,
-      "grad_norm": 0.000770802958868444,
-      "learning_rate": 0.00019999932409694898,
-      "loss": 46.0,
-      "step": 15317
-    },
-    {
-      "epoch": 1.1711680715637365,
-      "grad_norm": 0.0017223728355020285,
-      "learning_rate": 0.00019999932400863326,
-      "loss": 46.0,
-      "step": 15318
-    },
-    {
-      "epoch": 1.1712445285471262,
-      "grad_norm": 0.004088039044290781,
-      "learning_rate": 0.00019999932392031176,
-      "loss": 46.0,
-      "step": 15319
-    },
-    {
-      "epoch": 1.171320985530516,
-      "grad_norm": 0.0021303328685462475,
-      "learning_rate": 0.00019999932383198452,
-      "loss": 46.0,
-      "step": 15320
-    },
-    {
-      "epoch": 1.1713974425139055,
-      "grad_norm": 0.0007540914812125266,
-      "learning_rate": 0.00019999932374365148,
-      "loss": 46.0,
-      "step": 15321
-    },
-    {
-      "epoch": 1.1714738994972953,
-      "grad_norm": 0.004459446761757135,
-      "learning_rate": 0.0001999993236553127,
-      "loss": 46.0,
-      "step": 15322
-    },
-    {
-      "epoch": 1.171550356480685,
-      "grad_norm": 0.0017783168004825711,
-      "learning_rate": 0.0001999993235669681,
-      "loss": 46.0,
-      "step": 15323
-    },
-    {
-      "epoch": 1.1716268134640748,
-      "grad_norm": 0.0010124341351911426,
-      "learning_rate": 0.00019999932347861777,
-      "loss": 46.0,
-      "step": 15324
-    },
-    {
-      "epoch": 1.1717032704474646,
-      "grad_norm": 0.0007500700303353369,
-      "learning_rate": 0.0001999993233902617,
-      "loss": 46.0,
-      "step": 15325
-    },
-    {
-      "epoch": 1.1717797274308541,
-      "grad_norm": 0.0005886171711608768,
-      "learning_rate": 0.00019999932330189979,
-      "loss": 46.0,
-      "step": 15326
-    },
-    {
-      "epoch": 1.1718561844142439,
-      "grad_norm": 0.004187110345810652,
-      "learning_rate": 0.00019999932321353216,
-      "loss": 46.0,
-      "step": 15327
-    },
-    {
-      "epoch": 1.1719326413976336,
-      "grad_norm": 0.00308572081848979,
-      "learning_rate": 0.00019999932312515873,
-      "loss": 46.0,
-      "step": 15328
-    },
-    {
-      "epoch": 1.1720090983810234,
-      "grad_norm": 0.001711120828986168,
-      "learning_rate": 0.00019999932303677956,
-      "loss": 46.0,
-      "step": 15329
-    },
-    {
-      "epoch": 1.1720855553644132,
-      "grad_norm": 0.0035854196175932884,
-      "learning_rate": 0.0001999993229483946,
-      "loss": 46.0,
-      "step": 15330
-    },
-    {
-      "epoch": 1.172162012347803,
-      "grad_norm": 0.007313880138099194,
-      "learning_rate": 0.0001999993228600039,
-      "loss": 46.0,
-      "step": 15331
-    },
-    {
-      "epoch": 1.1722384693311925,
-      "grad_norm": 0.0008720503537915647,
-      "learning_rate": 0.0001999993227716074,
-      "loss": 46.0,
-      "step": 15332
-    },
-    {
-      "epoch": 1.1723149263145822,
-      "grad_norm": 0.001349899685010314,
-      "learning_rate": 0.00019999932268320513,
-      "loss": 46.0,
-      "step": 15333
-    },
-    {
-      "epoch": 1.172391383297972,
-      "grad_norm": 0.011800352483987808,
-      "learning_rate": 0.0001999993225947971,
-      "loss": 46.0,
-      "step": 15334
-    },
-    {
-      "epoch": 1.1724678402813618,
-      "grad_norm": 0.0007968770805746317,
-      "learning_rate": 0.0001999993225063833,
-      "loss": 46.0,
-      "step": 15335
-    },
-    {
-      "epoch": 1.1725442972647515,
-      "grad_norm": 0.001050457707606256,
-      "learning_rate": 0.00019999932241796374,
-      "loss": 46.0,
-      "step": 15336
-    },
-    {
-      "epoch": 1.172620754248141,
-      "grad_norm": 0.018532712012529373,
-      "learning_rate": 0.0001999993223295384,
-      "loss": 46.0,
-      "step": 15337
-    },
-    {
-      "epoch": 1.1726972112315308,
-      "grad_norm": 0.0005391244194470346,
-      "learning_rate": 0.00019999932224110727,
-      "loss": 46.0,
-      "step": 15338
-    },
-    {
-      "epoch": 1.1727736682149206,
-      "grad_norm": 0.0005859649972990155,
-      "learning_rate": 0.00019999932215267042,
-      "loss": 46.0,
-      "step": 15339
-    },
-    {
-      "epoch": 1.1728501251983103,
-      "grad_norm": 0.0023180313874036074,
-      "learning_rate": 0.00019999932206422777,
-      "loss": 46.0,
-      "step": 15340
-    },
-    {
-      "epoch": 1.1729265821817,
-      "grad_norm": 0.0014467989094555378,
-      "learning_rate": 0.00019999932197577934,
-      "loss": 46.0,
-      "step": 15341
-    },
-    {
-      "epoch": 1.1730030391650899,
-      "grad_norm": 0.009292705915868282,
-      "learning_rate": 0.00019999932188732516,
-      "loss": 46.0,
-      "step": 15342
-    },
-    {
-      "epoch": 1.1730794961484794,
-      "grad_norm": 0.001998856896534562,
-      "learning_rate": 0.0001999993217988652,
-      "loss": 46.0,
-      "step": 15343
-    },
-    {
-      "epoch": 1.1731559531318692,
-      "grad_norm": 0.0036292888689786196,
-      "learning_rate": 0.00019999932171039947,
-      "loss": 46.0,
-      "step": 15344
-    },
-    {
-      "epoch": 1.173232410115259,
-      "grad_norm": 0.005679688882082701,
-      "learning_rate": 0.00019999932162192798,
-      "loss": 46.0,
-      "step": 15345
-    },
-    {
-      "epoch": 1.1733088670986487,
-      "grad_norm": 0.0007781201275065541,
-      "learning_rate": 0.0001999993215334507,
-      "loss": 46.0,
-      "step": 15346
-    },
-    {
-      "epoch": 1.1733853240820384,
-      "grad_norm": 0.0010257113026455045,
-      "learning_rate": 0.00019999932144496767,
-      "loss": 46.0,
-      "step": 15347
-    },
-    {
-      "epoch": 1.173461781065428,
-      "grad_norm": 0.010551296174526215,
-      "learning_rate": 0.00019999932135647888,
-      "loss": 46.0,
-      "step": 15348
-    },
-    {
-      "epoch": 1.1735382380488177,
-      "grad_norm": 0.002150070620700717,
-      "learning_rate": 0.0001999993212679843,
-      "loss": 46.0,
-      "step": 15349
-    },
-    {
-      "epoch": 1.1736146950322075,
-      "grad_norm": 0.0012724722037091851,
-      "learning_rate": 0.00019999932117948394,
-      "loss": 46.0,
-      "step": 15350
-    },
-    {
-      "epoch": 1.1736911520155973,
-      "grad_norm": 0.002970330184325576,
-      "learning_rate": 0.00019999932109097786,
-      "loss": 46.0,
-      "step": 15351
-    },
-    {
-      "epoch": 1.173767608998987,
-      "grad_norm": 0.0003113442216999829,
-      "learning_rate": 0.00019999932100246598,
-      "loss": 46.0,
-      "step": 15352
-    },
-    {
-      "epoch": 1.1738440659823768,
-      "grad_norm": 0.001013904926367104,
-      "learning_rate": 0.0001999993209139483,
-      "loss": 46.0,
-      "step": 15353
-    },
-    {
-      "epoch": 1.1739205229657663,
-      "grad_norm": 0.0009218161576427519,
-      "learning_rate": 0.0001999993208254249,
-      "loss": 46.0,
-      "step": 15354
-    },
-    {
-      "epoch": 1.173996979949156,
-      "grad_norm": 0.0011633316753432155,
-      "learning_rate": 0.00019999932073689572,
-      "loss": 46.0,
-      "step": 15355
-    },
-    {
-      "epoch": 1.1740734369325458,
-      "grad_norm": 0.0016000056639313698,
-      "learning_rate": 0.00019999932064836075,
-      "loss": 46.0,
-      "step": 15356
-    },
-    {
-      "epoch": 1.1741498939159356,
-      "grad_norm": 0.0007826977525837719,
-      "learning_rate": 0.00019999932055982003,
-      "loss": 46.0,
-      "step": 15357
-    },
-    {
-      "epoch": 1.1742263508993251,
-      "grad_norm": 0.02654682844877243,
-      "learning_rate": 0.0001999993204712735,
-      "loss": 46.0,
-      "step": 15358
-    },
-    {
-      "epoch": 1.174302807882715,
-      "grad_norm": 0.0009169411496259272,
-      "learning_rate": 0.00019999932038272124,
-      "loss": 46.0,
-      "step": 15359
-    },
-    {
-      "epoch": 1.1743792648661047,
-      "grad_norm": 0.0005046732840128243,
-      "learning_rate": 0.00019999932029416323,
-      "loss": 46.0,
-      "step": 15360
-    },
-    {
-      "epoch": 1.1744557218494944,
-      "grad_norm": 0.015056449919939041,
-      "learning_rate": 0.00019999932020559942,
-      "loss": 46.0,
-      "step": 15361
-    },
-    {
-      "epoch": 1.1745321788328842,
-      "grad_norm": 0.0007006344385445118,
-      "learning_rate": 0.00019999932011702986,
-      "loss": 46.0,
-      "step": 15362
-    },
-    {
-      "epoch": 1.174608635816274,
-      "grad_norm": 0.0004977277712896466,
-      "learning_rate": 0.0001999993200284545,
-      "loss": 46.0,
-      "step": 15363
-    },
-    {
-      "epoch": 1.1746850927996637,
-      "grad_norm": 0.0022992698941379786,
-      "learning_rate": 0.0001999993199398734,
-      "loss": 46.0,
-      "step": 15364
-    },
-    {
-      "epoch": 1.1747615497830532,
-      "grad_norm": 0.0015222210204228759,
-      "learning_rate": 0.0001999993198512865,
-      "loss": 46.0,
-      "step": 15365
-    },
-    {
-      "epoch": 1.174838006766443,
-      "grad_norm": 0.0020082297269254923,
-      "learning_rate": 0.00019999931976269383,
-      "loss": 46.0,
-      "step": 15366
-    },
-    {
-      "epoch": 1.1749144637498328,
-      "grad_norm": 0.0039048115722835064,
-      "learning_rate": 0.00019999931967409543,
-      "loss": 46.0,
-      "step": 15367
-    },
-    {
-      "epoch": 1.1749909207332225,
-      "grad_norm": 0.0003600383934099227,
-      "learning_rate": 0.00019999931958549123,
-      "loss": 46.0,
-      "step": 15368
-    },
-    {
-      "epoch": 1.175067377716612,
-      "grad_norm": 0.00027772699831984937,
-      "learning_rate": 0.00019999931949688126,
-      "loss": 46.0,
-      "step": 15369
-    },
-    {
-      "epoch": 1.1751438347000018,
-      "grad_norm": 0.0011163987219333649,
-      "learning_rate": 0.00019999931940826554,
-      "loss": 46.0,
-      "step": 15370
-    },
-    {
-      "epoch": 1.1752202916833916,
-      "grad_norm": 0.009584103710949421,
-      "learning_rate": 0.00019999931931964404,
-      "loss": 46.0,
-      "step": 15371
-    },
-    {
-      "epoch": 1.1752967486667814,
-      "grad_norm": 0.0008097888203337789,
-      "learning_rate": 0.00019999931923101678,
-      "loss": 46.0,
-      "step": 15372
-    },
-    {
-      "epoch": 1.1753732056501711,
-      "grad_norm": 0.001861053635366261,
-      "learning_rate": 0.0001999993191423837,
-      "loss": 46.0,
-      "step": 15373
-    },
-    {
-      "epoch": 1.1754496626335609,
-      "grad_norm": 0.0006373692885972559,
-      "learning_rate": 0.00019999931905374492,
-      "loss": 46.0,
-      "step": 15374
-    },
-    {
-      "epoch": 1.1755261196169506,
-      "grad_norm": 0.0019834719132632017,
-      "learning_rate": 0.00019999931896510034,
-      "loss": 46.0,
-      "step": 15375
-    },
-    {
-      "epoch": 1.1756025766003402,
-      "grad_norm": 0.0006228814017958939,
-      "learning_rate": 0.00019999931887645,
-      "loss": 46.0,
-      "step": 15376
-    },
-    {
-      "epoch": 1.17567903358373,
-      "grad_norm": 0.0011599621502682567,
-      "learning_rate": 0.0001999993187877939,
-      "loss": 46.0,
-      "step": 15377
-    },
-    {
-      "epoch": 1.1757554905671197,
-      "grad_norm": 0.0009037195122800767,
-      "learning_rate": 0.000199999318699132,
-      "loss": 46.0,
-      "step": 15378
-    },
-    {
-      "epoch": 1.1758319475505095,
-      "grad_norm": 0.001184728229418397,
-      "learning_rate": 0.00019999931861046434,
-      "loss": 46.0,
-      "step": 15379
-    },
-    {
-      "epoch": 1.175908404533899,
-      "grad_norm": 0.0014615757390856743,
-      "learning_rate": 0.00019999931852179091,
-      "loss": 46.0,
-      "step": 15380
-    },
-    {
-      "epoch": 1.1759848615172888,
-      "grad_norm": 0.000806040596216917,
-      "learning_rate": 0.00019999931843311171,
-      "loss": 46.0,
-      "step": 15381
-    },
-    {
-      "epoch": 1.1760613185006785,
-      "grad_norm": 0.0049487450160086155,
-      "learning_rate": 0.00019999931834442677,
-      "loss": 46.0,
-      "step": 15382
-    },
-    {
-      "epoch": 1.1761377754840683,
-      "grad_norm": 0.0005893089110031724,
-      "learning_rate": 0.00019999931825573602,
-      "loss": 46.0,
-      "step": 15383
-    },
-    {
-      "epoch": 1.176214232467458,
-      "grad_norm": 0.0011899803066626191,
-      "learning_rate": 0.00019999931816703953,
-      "loss": 46.0,
-      "step": 15384
-    },
-    {
-      "epoch": 1.1762906894508478,
-      "grad_norm": 0.001896370085887611,
-      "learning_rate": 0.00019999931807833726,
-      "loss": 46.0,
-      "step": 15385
-    },
-    {
-      "epoch": 1.1763671464342373,
-      "grad_norm": 0.002269502729177475,
-      "learning_rate": 0.0001999993179896292,
-      "loss": 46.0,
-      "step": 15386
-    },
-    {
-      "epoch": 1.176443603417627,
-      "grad_norm": 0.0029634120874106884,
-      "learning_rate": 0.0001999993179009154,
-      "loss": 46.0,
-      "step": 15387
-    },
-    {
-      "epoch": 1.1765200604010169,
-      "grad_norm": 0.0008897338411770761,
-      "learning_rate": 0.00019999931781219582,
-      "loss": 46.0,
-      "step": 15388
-    },
-    {
-      "epoch": 1.1765965173844066,
-      "grad_norm": 0.001576107693836093,
-      "learning_rate": 0.00019999931772347047,
-      "loss": 46.0,
-      "step": 15389
-    },
-    {
-      "epoch": 1.1766729743677964,
-      "grad_norm": 0.00033984603942371905,
-      "learning_rate": 0.00019999931763473936,
-      "loss": 46.0,
-      "step": 15390
-    },
-    {
-      "epoch": 1.176749431351186,
-      "grad_norm": 0.00216719601303339,
-      "learning_rate": 0.00019999931754600248,
-      "loss": 46.0,
-      "step": 15391
-    },
-    {
-      "epoch": 1.1768258883345757,
-      "grad_norm": 0.00038681377191096544,
-      "learning_rate": 0.00019999931745725983,
-      "loss": 46.0,
-      "step": 15392
-    },
-    {
-      "epoch": 1.1769023453179654,
-      "grad_norm": 0.0017777879256755114,
-      "learning_rate": 0.00019999931736851137,
-      "loss": 46.0,
-      "step": 15393
-    },
-    {
-      "epoch": 1.1769788023013552,
-      "grad_norm": 0.0004222371499054134,
-      "learning_rate": 0.00019999931727975718,
-      "loss": 46.0,
-      "step": 15394
-    },
-    {
-      "epoch": 1.177055259284745,
-      "grad_norm": 0.0016078681219369173,
-      "learning_rate": 0.00019999931719099723,
-      "loss": 46.0,
-      "step": 15395
-    },
-    {
-      "epoch": 1.1771317162681347,
-      "grad_norm": 0.0007601361139677465,
-      "learning_rate": 0.0001999993171022315,
-      "loss": 46.0,
-      "step": 15396
-    },
-    {
-      "epoch": 1.1772081732515243,
-      "grad_norm": 0.0007044863305054605,
-      "learning_rate": 0.00019999931701345996,
-      "loss": 46.0,
-      "step": 15397
-    },
-    {
-      "epoch": 1.177284630234914,
-      "grad_norm": 0.002174082677811384,
-      "learning_rate": 0.00019999931692468273,
-      "loss": 46.0,
-      "step": 15398
-    },
-    {
-      "epoch": 1.1773610872183038,
-      "grad_norm": 0.0003637280606199056,
-      "learning_rate": 0.0001999993168358997,
-      "loss": 46.0,
-      "step": 15399
-    },
-    {
-      "epoch": 1.1774375442016936,
-      "grad_norm": 0.0012934559490531683,
-      "learning_rate": 0.00019999931674711085,
-      "loss": 46.0,
-      "step": 15400
-    },
-    {
-      "epoch": 1.1775140011850833,
-      "grad_norm": 0.0026614877860993147,
-      "learning_rate": 0.0001999993166583163,
-      "loss": 46.0,
-      "step": 15401
-    },
-    {
-      "epoch": 1.1775904581684729,
-      "grad_norm": 0.001914922846481204,
-      "learning_rate": 0.00019999931656951593,
-      "loss": 46.0,
-      "step": 15402
-    },
-    {
-      "epoch": 1.1776669151518626,
-      "grad_norm": 0.001040788134559989,
-      "learning_rate": 0.00019999931648070982,
-      "loss": 46.0,
-      "step": 15403
-    },
-    {
-      "epoch": 1.1777433721352524,
-      "grad_norm": 0.0005309241823852062,
-      "learning_rate": 0.00019999931639189792,
-      "loss": 46.0,
-      "step": 15404
-    },
-    {
-      "epoch": 1.1778198291186421,
-      "grad_norm": 0.001082441653124988,
-      "learning_rate": 0.00019999931630308027,
-      "loss": 46.0,
-      "step": 15405
-    },
-    {
-      "epoch": 1.177896286102032,
-      "grad_norm": 0.005845897365361452,
-      "learning_rate": 0.00019999931621425684,
-      "loss": 46.0,
-      "step": 15406
-    },
-    {
-      "epoch": 1.1779727430854217,
-      "grad_norm": 0.0017694375710561872,
-      "learning_rate": 0.00019999931612542761,
-      "loss": 46.0,
-      "step": 15407
-    },
-    {
-      "epoch": 1.1780492000688112,
-      "grad_norm": 0.000950800662394613,
-      "learning_rate": 0.00019999931603659264,
-      "loss": 46.0,
-      "step": 15408
-    },
-    {
-      "epoch": 1.178125657052201,
-      "grad_norm": 0.003270723158493638,
-      "learning_rate": 0.00019999931594775192,
-      "loss": 46.0,
-      "step": 15409
-    },
-    {
-      "epoch": 1.1782021140355907,
-      "grad_norm": 0.004383347928524017,
-      "learning_rate": 0.0001999993158589054,
-      "loss": 46.0,
-      "step": 15410
-    },
-    {
-      "epoch": 1.1782785710189805,
-      "grad_norm": 0.0009847903857007623,
-      "learning_rate": 0.00019999931577005314,
-      "loss": 46.0,
-      "step": 15411
-    },
-    {
-      "epoch": 1.1783550280023702,
-      "grad_norm": 0.0020054453052580357,
-      "learning_rate": 0.0001999993156811951,
-      "loss": 46.0,
-      "step": 15412
-    },
-    {
-      "epoch": 1.1784314849857598,
-      "grad_norm": 0.000637597928289324,
-      "learning_rate": 0.0001999993155923313,
-      "loss": 46.0,
-      "step": 15413
-    },
-    {
-      "epoch": 1.1785079419691495,
-      "grad_norm": 0.0009336243383586407,
-      "learning_rate": 0.0001999993155034617,
-      "loss": 46.0,
-      "step": 15414
-    },
-    {
-      "epoch": 1.1785843989525393,
-      "grad_norm": 0.0024396292865276337,
-      "learning_rate": 0.00019999931541458634,
-      "loss": 46.0,
-      "step": 15415
-    },
-    {
-      "epoch": 1.178660855935929,
-      "grad_norm": 0.00648860028013587,
-      "learning_rate": 0.0001999993153257052,
-      "loss": 46.0,
-      "step": 15416
-    },
-    {
-      "epoch": 1.1787373129193188,
-      "grad_norm": 0.0016205405117943883,
-      "learning_rate": 0.0001999993152368183,
-      "loss": 46.0,
-      "step": 15417
-    },
-    {
-      "epoch": 1.1788137699027086,
-      "grad_norm": 0.0008968322072178125,
-      "learning_rate": 0.00019999931514792568,
-      "loss": 46.0,
-      "step": 15418
-    },
-    {
-      "epoch": 1.1788902268860981,
-      "grad_norm": 0.0019769337959587574,
-      "learning_rate": 0.00019999931505902723,
-      "loss": 46.0,
-      "step": 15419
-    },
-    {
-      "epoch": 1.1789666838694879,
-      "grad_norm": 0.0015798007370904088,
-      "learning_rate": 0.00019999931497012303,
-      "loss": 46.0,
-      "step": 15420
-    },
-    {
-      "epoch": 1.1790431408528776,
-      "grad_norm": 0.001267299405299127,
-      "learning_rate": 0.00019999931488121306,
-      "loss": 46.0,
-      "step": 15421
-    },
-    {
-      "epoch": 1.1791195978362674,
-      "grad_norm": 0.0009020359138958156,
-      "learning_rate": 0.0001999993147922973,
-      "loss": 46.0,
-      "step": 15422
-    },
-    {
-      "epoch": 1.1791960548196572,
-      "grad_norm": 0.0023591441567987204,
-      "learning_rate": 0.0001999993147033758,
-      "loss": 46.0,
-      "step": 15423
-    },
-    {
-      "epoch": 1.1792725118030467,
-      "grad_norm": 0.0015569181414321065,
-      "learning_rate": 0.00019999931461444853,
-      "loss": 46.0,
-      "step": 15424
-    },
-    {
-      "epoch": 1.1793489687864365,
-      "grad_norm": 0.0011056180810555816,
-      "learning_rate": 0.0001999993145255155,
-      "loss": 46.0,
-      "step": 15425
-    },
-    {
-      "epoch": 1.1794254257698262,
-      "grad_norm": 0.0012619401095435023,
-      "learning_rate": 0.00019999931443657668,
-      "loss": 46.0,
-      "step": 15426
-    },
-    {
-      "epoch": 1.179501882753216,
-      "grad_norm": 0.003567162435501814,
-      "learning_rate": 0.0001999993143476321,
-      "loss": 46.0,
-      "step": 15427
-    },
-    {
-      "epoch": 1.1795783397366058,
-      "grad_norm": 0.012163173407316208,
-      "learning_rate": 0.00019999931425868173,
-      "loss": 46.0,
-      "step": 15428
-    },
-    {
-      "epoch": 1.1796547967199955,
-      "grad_norm": 0.0016579240327700973,
-      "learning_rate": 0.0001999993141697256,
-      "loss": 46.0,
-      "step": 15429
-    },
-    {
-      "epoch": 1.179731253703385,
-      "grad_norm": 0.000664646562654525,
-      "learning_rate": 0.00019999931408076372,
-      "loss": 46.0,
-      "step": 15430
-    },
-    {
-      "epoch": 1.1798077106867748,
-      "grad_norm": 0.0017933192430064082,
-      "learning_rate": 0.00019999931399179604,
-      "loss": 46.0,
-      "step": 15431
-    },
-    {
-      "epoch": 1.1798841676701646,
-      "grad_norm": 0.0009983469499275088,
-      "learning_rate": 0.0001999993139028226,
-      "loss": 46.0,
-      "step": 15432
-    },
-    {
-      "epoch": 1.1799606246535543,
-      "grad_norm": 0.0006373559008352458,
-      "learning_rate": 0.0001999993138138434,
-      "loss": 46.0,
-      "step": 15433
-    },
-    {
-      "epoch": 1.180037081636944,
-      "grad_norm": 0.005174398887902498,
-      "learning_rate": 0.00019999931372485842,
-      "loss": 46.0,
-      "step": 15434
-    },
-    {
-      "epoch": 1.1801135386203336,
-      "grad_norm": 0.0029068132862448692,
-      "learning_rate": 0.00019999931363586768,
-      "loss": 46.0,
-      "step": 15435
-    },
-    {
-      "epoch": 1.1801899956037234,
-      "grad_norm": 0.0005682732444256544,
-      "learning_rate": 0.00019999931354687119,
-      "loss": 46.0,
-      "step": 15436
-    },
-    {
-      "epoch": 1.1802664525871132,
-      "grad_norm": 0.0015646591782569885,
-      "learning_rate": 0.0001999993134578689,
-      "loss": 46.0,
-      "step": 15437
-    },
-    {
-      "epoch": 1.180342909570503,
-      "grad_norm": 0.00571898277848959,
-      "learning_rate": 0.00019999931336886083,
-      "loss": 46.0,
-      "step": 15438
-    },
-    {
-      "epoch": 1.1804193665538927,
-      "grad_norm": 0.0016845166683197021,
-      "learning_rate": 0.00019999931327984702,
-      "loss": 46.0,
-      "step": 15439
-    },
-    {
-      "epoch": 1.1804958235372824,
-      "grad_norm": 0.0006869092467240989,
-      "learning_rate": 0.00019999931319082743,
-      "loss": 46.0,
-      "step": 15440
-    },
-    {
-      "epoch": 1.180572280520672,
-      "grad_norm": 0.001669710618443787,
-      "learning_rate": 0.0001999993131018021,
-      "loss": 46.0,
-      "step": 15441
-    },
-    {
-      "epoch": 1.1806487375040617,
-      "grad_norm": 0.0006997182499617338,
-      "learning_rate": 0.00019999931301277094,
-      "loss": 46.0,
-      "step": 15442
-    },
-    {
-      "epoch": 1.1807251944874515,
-      "grad_norm": 0.0011786954710260034,
-      "learning_rate": 0.00019999931292373404,
-      "loss": 46.0,
-      "step": 15443
-    },
-    {
-      "epoch": 1.1808016514708413,
-      "grad_norm": 0.0031874889973551035,
-      "learning_rate": 0.0001999993128346914,
-      "loss": 46.0,
-      "step": 15444
-    },
-    {
-      "epoch": 1.180878108454231,
-      "grad_norm": 0.0019556907936930656,
-      "learning_rate": 0.00019999931274564293,
-      "loss": 46.0,
-      "step": 15445
-    },
-    {
-      "epoch": 1.1809545654376206,
-      "grad_norm": 0.0014699024613946676,
-      "learning_rate": 0.00019999931265658876,
-      "loss": 46.0,
-      "step": 15446
-    },
-    {
-      "epoch": 1.1810310224210103,
-      "grad_norm": 0.004740988835692406,
-      "learning_rate": 0.00019999931256752874,
-      "loss": 46.0,
-      "step": 15447
-    },
-    {
-      "epoch": 1.1811074794044,
-      "grad_norm": 0.002042998792603612,
-      "learning_rate": 0.00019999931247846302,
-      "loss": 46.0,
-      "step": 15448
-    },
-    {
-      "epoch": 1.1811839363877898,
-      "grad_norm": 0.0009597280877642334,
-      "learning_rate": 0.0001999993123893915,
-      "loss": 46.0,
-      "step": 15449
-    },
-    {
-      "epoch": 1.1812603933711796,
-      "grad_norm": 0.0006218540947884321,
-      "learning_rate": 0.00019999931230031424,
-      "loss": 46.0,
-      "step": 15450
-    },
-    {
-      "epoch": 1.1813368503545694,
-      "grad_norm": 0.0005163172609172761,
-      "learning_rate": 0.00019999931221123117,
-      "loss": 46.0,
-      "step": 15451
-    },
-    {
-      "epoch": 1.181413307337959,
-      "grad_norm": 0.02556164562702179,
-      "learning_rate": 0.00019999931212214236,
-      "loss": 46.0,
-      "step": 15452
-    },
-    {
-      "epoch": 1.1814897643213487,
-      "grad_norm": 0.0012519657611846924,
-      "learning_rate": 0.00019999931203304775,
-      "loss": 46.0,
-      "step": 15453
-    },
-    {
-      "epoch": 1.1815662213047384,
-      "grad_norm": 0.0013388078659772873,
-      "learning_rate": 0.0001999993119439474,
-      "loss": 46.0,
-      "step": 15454
-    },
-    {
-      "epoch": 1.1816426782881282,
-      "grad_norm": 0.0005547038745135069,
-      "learning_rate": 0.00019999931185484126,
-      "loss": 46.0,
-      "step": 15455
-    },
-    {
-      "epoch": 1.181719135271518,
-      "grad_norm": 0.001275600166991353,
-      "learning_rate": 0.00019999931176572936,
-      "loss": 46.0,
-      "step": 15456
-    },
-    {
-      "epoch": 1.1817955922549075,
-      "grad_norm": 0.0003953548439312726,
-      "learning_rate": 0.0001999993116766117,
-      "loss": 46.0,
-      "step": 15457
-    },
-    {
-      "epoch": 1.1818720492382973,
-      "grad_norm": 0.0006889212527312338,
-      "learning_rate": 0.00019999931158748825,
-      "loss": 46.0,
-      "step": 15458
-    },
-    {
-      "epoch": 1.181948506221687,
-      "grad_norm": 0.0020947728771716356,
-      "learning_rate": 0.00019999931149835906,
-      "loss": 46.0,
-      "step": 15459
-    },
-    {
-      "epoch": 1.1820249632050768,
-      "grad_norm": 0.00659290561452508,
-      "learning_rate": 0.00019999931140922406,
-      "loss": 46.0,
-      "step": 15460
-    },
-    {
-      "epoch": 1.1821014201884665,
-      "grad_norm": 0.0021299654617905617,
-      "learning_rate": 0.0001999993113200833,
-      "loss": 46.0,
-      "step": 15461
-    },
-    {
-      "epoch": 1.1821778771718563,
-      "grad_norm": 0.0005596533883363008,
-      "learning_rate": 0.0001999993112309368,
-      "loss": 46.0,
-      "step": 15462
-    },
-    {
-      "epoch": 1.1822543341552458,
-      "grad_norm": 0.00132686214055866,
-      "learning_rate": 0.00019999931114178453,
-      "loss": 46.0,
-      "step": 15463
-    },
-    {
-      "epoch": 1.1823307911386356,
-      "grad_norm": 0.0015958151780068874,
-      "learning_rate": 0.00019999931105262647,
-      "loss": 46.0,
-      "step": 15464
-    },
-    {
-      "epoch": 1.1824072481220254,
-      "grad_norm": 0.0008263010531663895,
-      "learning_rate": 0.00019999931096346263,
-      "loss": 46.0,
-      "step": 15465
-    },
-    {
-      "epoch": 1.1824837051054151,
-      "grad_norm": 0.0008682474144734442,
-      "learning_rate": 0.00019999931087429304,
-      "loss": 46.0,
-      "step": 15466
-    },
-    {
-      "epoch": 1.1825601620888049,
-      "grad_norm": 0.001114193699322641,
-      "learning_rate": 0.0001999993107851177,
-      "loss": 46.0,
-      "step": 15467
-    },
-    {
-      "epoch": 1.1826366190721944,
-      "grad_norm": 0.0018916726112365723,
-      "learning_rate": 0.00019999931069593653,
-      "loss": 46.0,
-      "step": 15468
-    },
-    {
-      "epoch": 1.1827130760555842,
-      "grad_norm": 0.010582629591226578,
-      "learning_rate": 0.00019999931060674965,
-      "loss": 46.0,
-      "step": 15469
-    },
-    {
-      "epoch": 1.182789533038974,
-      "grad_norm": 0.0010482128709554672,
-      "learning_rate": 0.00019999931051755697,
-      "loss": 46.0,
-      "step": 15470
-    },
-    {
-      "epoch": 1.1828659900223637,
-      "grad_norm": 0.0016996449558064342,
-      "learning_rate": 0.00019999931042835852,
-      "loss": 46.0,
-      "step": 15471
-    },
-    {
-      "epoch": 1.1829424470057535,
-      "grad_norm": 0.0012100361054763198,
-      "learning_rate": 0.00019999931033915433,
-      "loss": 46.0,
-      "step": 15472
-    },
-    {
-      "epoch": 1.1830189039891432,
-      "grad_norm": 0.0006674345931969583,
-      "learning_rate": 0.00019999931024994435,
-      "loss": 46.0,
-      "step": 15473
-    },
-    {
-      "epoch": 1.1830953609725328,
-      "grad_norm": 0.014526882208883762,
-      "learning_rate": 0.00019999931016072858,
-      "loss": 46.0,
-      "step": 15474
-    },
-    {
-      "epoch": 1.1831718179559225,
-      "grad_norm": 0.0011232286924496293,
-      "learning_rate": 0.00019999931007150707,
-      "loss": 46.0,
-      "step": 15475
-    },
-    {
-      "epoch": 1.1832482749393123,
-      "grad_norm": 0.000940681085921824,
-      "learning_rate": 0.0001999993099822798,
-      "loss": 46.0,
-      "step": 15476
-    },
-    {
-      "epoch": 1.183324731922702,
-      "grad_norm": 0.0012857902329415083,
-      "learning_rate": 0.00019999930989304674,
-      "loss": 46.0,
-      "step": 15477
-    },
-    {
-      "epoch": 1.1834011889060918,
-      "grad_norm": 0.008986872620880604,
-      "learning_rate": 0.0001999993098038079,
-      "loss": 46.0,
-      "step": 15478
-    },
-    {
-      "epoch": 1.1834776458894813,
-      "grad_norm": 0.0015535138081759214,
-      "learning_rate": 0.00019999930971456332,
-      "loss": 46.0,
-      "step": 15479
-    },
-    {
-      "epoch": 1.183554102872871,
-      "grad_norm": 0.0013885722728446126,
-      "learning_rate": 0.00019999930962531296,
-      "loss": 46.0,
-      "step": 15480
-    },
-    {
-      "epoch": 1.1836305598562609,
-      "grad_norm": 0.008818816393613815,
-      "learning_rate": 0.0001999993095360568,
-      "loss": 46.0,
-      "step": 15481
-    },
-    {
-      "epoch": 1.1837070168396506,
-      "grad_norm": 0.0011947766179218888,
-      "learning_rate": 0.0001999993094467949,
-      "loss": 46.0,
-      "step": 15482
-    },
-    {
-      "epoch": 1.1837834738230404,
-      "grad_norm": 0.0004282281151972711,
-      "learning_rate": 0.00019999930935752725,
-      "loss": 46.0,
-      "step": 15483
-    },
-    {
-      "epoch": 1.1838599308064301,
-      "grad_norm": 0.00035026593832299113,
-      "learning_rate": 0.0001999993092682538,
-      "loss": 46.0,
-      "step": 15484
-    },
-    {
-      "epoch": 1.1839363877898197,
-      "grad_norm": 0.0010041380301117897,
-      "learning_rate": 0.00019999930917897457,
-      "loss": 46.0,
-      "step": 15485
-    },
-    {
-      "epoch": 1.1840128447732094,
-      "grad_norm": 0.0021436784882098436,
-      "learning_rate": 0.00019999930908968957,
-      "loss": 46.0,
-      "step": 15486
-    },
-    {
-      "epoch": 1.1840893017565992,
-      "grad_norm": 0.0012605394003912807,
-      "learning_rate": 0.00019999930900039883,
-      "loss": 46.0,
-      "step": 15487
-    },
-    {
-      "epoch": 1.184165758739989,
-      "grad_norm": 0.0013191248290240765,
-      "learning_rate": 0.0001999993089111023,
-      "loss": 46.0,
-      "step": 15488
-    },
-    {
-      "epoch": 1.1842422157233785,
-      "grad_norm": 0.0004925571847707033,
-      "learning_rate": 0.00019999930882180002,
-      "loss": 46.0,
-      "step": 15489
-    },
-    {
-      "epoch": 1.1843186727067683,
-      "grad_norm": 0.0007125950651243329,
-      "learning_rate": 0.00019999930873249199,
-      "loss": 46.0,
-      "step": 15490
-    },
-    {
-      "epoch": 1.184395129690158,
-      "grad_norm": 0.0007753656827844679,
-      "learning_rate": 0.00019999930864317812,
-      "loss": 46.0,
-      "step": 15491
-    },
-    {
-      "epoch": 1.1844715866735478,
-      "grad_norm": 0.0009097495931200683,
-      "learning_rate": 0.00019999930855385856,
-      "loss": 46.0,
-      "step": 15492
-    },
-    {
-      "epoch": 1.1845480436569376,
-      "grad_norm": 0.00044157583033666015,
-      "learning_rate": 0.00019999930846453315,
-      "loss": 46.0,
-      "step": 15493
-    },
-    {
-      "epoch": 1.1846245006403273,
-      "grad_norm": 0.0010635844664648175,
-      "learning_rate": 0.00019999930837520205,
-      "loss": 46.0,
-      "step": 15494
-    },
-    {
-      "epoch": 1.184700957623717,
-      "grad_norm": 0.0007624759455211461,
-      "learning_rate": 0.00019999930828586512,
-      "loss": 46.0,
-      "step": 15495
-    },
-    {
-      "epoch": 1.1847774146071066,
-      "grad_norm": 0.0028700088150799274,
-      "learning_rate": 0.00019999930819652244,
-      "loss": 46.0,
-      "step": 15496
-    },
-    {
-      "epoch": 1.1848538715904964,
-      "grad_norm": 0.0014099212130531669,
-      "learning_rate": 0.000199999308107174,
-      "loss": 46.0,
-      "step": 15497
-    },
-    {
-      "epoch": 1.1849303285738861,
-      "grad_norm": 0.005258242599666119,
-      "learning_rate": 0.0001999993080178198,
-      "loss": 46.0,
-      "step": 15498
-    },
-    {
-      "epoch": 1.185006785557276,
-      "grad_norm": 0.0009431655053049326,
-      "learning_rate": 0.0001999993079284598,
-      "loss": 46.0,
-      "step": 15499
-    },
-    {
-      "epoch": 1.1850832425406654,
-      "grad_norm": 0.0009974036365747452,
-      "learning_rate": 0.00019999930783909405,
-      "loss": 46.0,
-      "step": 15500
-    },
-    {
-      "epoch": 1.1851596995240552,
-      "grad_norm": 0.0033718496561050415,
-      "learning_rate": 0.00019999930774972254,
-      "loss": 46.0,
-      "step": 15501
-    },
-    {
-      "epoch": 1.185236156507445,
-      "grad_norm": 0.0018586317310109735,
-      "learning_rate": 0.00019999930766034525,
-      "loss": 46.0,
-      "step": 15502
-    },
-    {
-      "epoch": 1.1853126134908347,
-      "grad_norm": 0.0027830402832478285,
-      "learning_rate": 0.00019999930757096216,
-      "loss": 46.0,
-      "step": 15503
-    },
-    {
-      "epoch": 1.1853890704742245,
-      "grad_norm": 0.0248353723436594,
-      "learning_rate": 0.00019999930748157335,
-      "loss": 46.0,
-      "step": 15504
-    },
-    {
-      "epoch": 1.1854655274576142,
-      "grad_norm": 0.0005697284359484911,
-      "learning_rate": 0.00019999930739217874,
-      "loss": 46.0,
-      "step": 15505
-    },
-    {
-      "epoch": 1.185541984441004,
-      "grad_norm": 0.00223549478687346,
-      "learning_rate": 0.00019999930730277838,
-      "loss": 46.0,
-      "step": 15506
-    },
-    {
-      "epoch": 1.1856184414243935,
-      "grad_norm": 0.002657330362126231,
-      "learning_rate": 0.00019999930721337222,
-      "loss": 46.0,
-      "step": 15507
-    },
-    {
-      "epoch": 1.1856948984077833,
-      "grad_norm": 0.0014047704171389341,
-      "learning_rate": 0.00019999930712396032,
-      "loss": 46.0,
-      "step": 15508
-    },
-    {
-      "epoch": 1.185771355391173,
-      "grad_norm": 0.0018662635702639818,
-      "learning_rate": 0.00019999930703454264,
-      "loss": 46.0,
-      "step": 15509
-    },
-    {
-      "epoch": 1.1858478123745628,
-      "grad_norm": 0.002144875703379512,
-      "learning_rate": 0.0001999993069451192,
-      "loss": 46.0,
-      "step": 15510
-    },
-    {
-      "epoch": 1.1859242693579524,
-      "grad_norm": 0.0012508862419053912,
-      "learning_rate": 0.00019999930685568997,
-      "loss": 46.0,
-      "step": 15511
-    },
-    {
-      "epoch": 1.1860007263413421,
-      "grad_norm": 0.0010779559379443526,
-      "learning_rate": 0.00019999930676625497,
-      "loss": 46.0,
-      "step": 15512
-    },
-    {
-      "epoch": 1.1860771833247319,
-      "grad_norm": 0.0007783265900798142,
-      "learning_rate": 0.00019999930667681423,
-      "loss": 46.0,
-      "step": 15513
-    },
-    {
-      "epoch": 1.1861536403081216,
-      "grad_norm": 0.0008492347551509738,
-      "learning_rate": 0.0001999993065873677,
-      "loss": 46.0,
-      "step": 15514
-    },
-    {
-      "epoch": 1.1862300972915114,
-      "grad_norm": 0.0004294106038287282,
-      "learning_rate": 0.0001999993064979154,
-      "loss": 46.0,
-      "step": 15515
-    },
-    {
-      "epoch": 1.1863065542749012,
-      "grad_norm": 0.0006817835965193808,
-      "learning_rate": 0.00019999930640845733,
-      "loss": 46.0,
-      "step": 15516
-    },
-    {
-      "epoch": 1.1863830112582907,
-      "grad_norm": 0.011327757500112057,
-      "learning_rate": 0.0001999993063189935,
-      "loss": 46.0,
-      "step": 15517
-    },
-    {
-      "epoch": 1.1864594682416805,
-      "grad_norm": 0.0007910567219369113,
-      "learning_rate": 0.0001999993062295239,
-      "loss": 46.0,
-      "step": 15518
-    },
-    {
-      "epoch": 1.1865359252250702,
-      "grad_norm": 0.0013478505425155163,
-      "learning_rate": 0.00019999930614004853,
-      "loss": 46.0,
-      "step": 15519
-    },
-    {
-      "epoch": 1.18661238220846,
-      "grad_norm": 0.001307833124883473,
-      "learning_rate": 0.00019999930605056737,
-      "loss": 46.0,
-      "step": 15520
-    },
-    {
-      "epoch": 1.1866888391918498,
-      "grad_norm": 0.0005885678110644221,
-      "learning_rate": 0.00019999930596108047,
-      "loss": 46.0,
-      "step": 15521
-    },
-    {
-      "epoch": 1.1867652961752393,
-      "grad_norm": 0.0012011270737275481,
-      "learning_rate": 0.0001999993058715878,
-      "loss": 46.0,
-      "step": 15522
-    },
-    {
-      "epoch": 1.186841753158629,
-      "grad_norm": 0.0006303370464593172,
-      "learning_rate": 0.00019999930578208934,
-      "loss": 46.0,
-      "step": 15523
-    },
-    {
-      "epoch": 1.1869182101420188,
-      "grad_norm": 0.0007026963867247105,
-      "learning_rate": 0.00019999930569258512,
-      "loss": 46.0,
-      "step": 15524
-    },
-    {
-      "epoch": 1.1869946671254086,
-      "grad_norm": 0.0029053364414721727,
-      "learning_rate": 0.00019999930560307515,
-      "loss": 46.0,
-      "step": 15525
-    },
-    {
-      "epoch": 1.1870711241087983,
-      "grad_norm": 0.00027197255985811353,
-      "learning_rate": 0.00019999930551355938,
-      "loss": 46.0,
-      "step": 15526
-    },
-    {
-      "epoch": 1.187147581092188,
-      "grad_norm": 0.0013747927732765675,
-      "learning_rate": 0.00019999930542403783,
-      "loss": 46.0,
-      "step": 15527
-    },
-    {
-      "epoch": 1.1872240380755776,
-      "grad_norm": 0.019425805658102036,
-      "learning_rate": 0.00019999930533451057,
-      "loss": 46.0,
-      "step": 15528
-    },
-    {
-      "epoch": 1.1873004950589674,
-      "grad_norm": 0.002812436316162348,
-      "learning_rate": 0.00019999930524497748,
-      "loss": 46.0,
-      "step": 15529
-    },
-    {
-      "epoch": 1.1873769520423572,
-      "grad_norm": 0.0008677076548337936,
-      "learning_rate": 0.00019999930515543865,
-      "loss": 46.0,
-      "step": 15530
-    },
-    {
-      "epoch": 1.187453409025747,
-      "grad_norm": 0.0010049112606793642,
-      "learning_rate": 0.00019999930506589406,
-      "loss": 46.0,
-      "step": 15531
-    },
-    {
-      "epoch": 1.1875298660091367,
-      "grad_norm": 0.0019073245348408818,
-      "learning_rate": 0.00019999930497634368,
-      "loss": 46.0,
-      "step": 15532
-    },
-    {
-      "epoch": 1.1876063229925262,
-      "grad_norm": 0.0046845655888319016,
-      "learning_rate": 0.0001999993048867875,
-      "loss": 46.0,
-      "step": 15533
-    },
-    {
-      "epoch": 1.187682779975916,
-      "grad_norm": 0.0016535703325644135,
-      "learning_rate": 0.0001999993047972256,
-      "loss": 46.0,
-      "step": 15534
-    },
-    {
-      "epoch": 1.1877592369593057,
-      "grad_norm": 0.0011481159599497914,
-      "learning_rate": 0.00019999930470765794,
-      "loss": 46.0,
-      "step": 15535
-    },
-    {
-      "epoch": 1.1878356939426955,
-      "grad_norm": 0.002079533413052559,
-      "learning_rate": 0.00019999930461808447,
-      "loss": 46.0,
-      "step": 15536
-    },
-    {
-      "epoch": 1.1879121509260853,
-      "grad_norm": 0.003487946232780814,
-      "learning_rate": 0.00019999930452850527,
-      "loss": 46.0,
-      "step": 15537
-    },
-    {
-      "epoch": 1.187988607909475,
-      "grad_norm": 0.000580307561904192,
-      "learning_rate": 0.00019999930443892025,
-      "loss": 46.0,
-      "step": 15538
-    },
-    {
-      "epoch": 1.1880650648928646,
-      "grad_norm": 0.0003372525388840586,
-      "learning_rate": 0.0001999993043493295,
-      "loss": 46.0,
-      "step": 15539
-    },
-    {
-      "epoch": 1.1881415218762543,
-      "grad_norm": 0.00035768383531831205,
-      "learning_rate": 0.00019999930425973296,
-      "loss": 46.0,
-      "step": 15540
-    },
-    {
-      "epoch": 1.188217978859644,
-      "grad_norm": 0.005122461821883917,
-      "learning_rate": 0.00019999930417013067,
-      "loss": 46.0,
-      "step": 15541
-    },
-    {
-      "epoch": 1.1882944358430338,
-      "grad_norm": 0.002018338069319725,
-      "learning_rate": 0.00019999930408052258,
-      "loss": 46.0,
-      "step": 15542
-    },
-    {
-      "epoch": 1.1883708928264236,
-      "grad_norm": 0.0025058388710021973,
-      "learning_rate": 0.00019999930399090875,
-      "loss": 46.0,
-      "step": 15543
-    },
-    {
-      "epoch": 1.1884473498098131,
-      "grad_norm": 0.0035456903278827667,
-      "learning_rate": 0.00019999930390128916,
-      "loss": 46.0,
-      "step": 15544
-    },
-    {
-      "epoch": 1.188523806793203,
-      "grad_norm": 0.0039965189062058926,
-      "learning_rate": 0.00019999930381166378,
-      "loss": 46.0,
-      "step": 15545
-    },
-    {
-      "epoch": 1.1886002637765927,
-      "grad_norm": 0.0009805349400267005,
-      "learning_rate": 0.00019999930372203263,
-      "loss": 46.0,
-      "step": 15546
-    },
-    {
-      "epoch": 1.1886767207599824,
-      "grad_norm": 0.0008931240881793201,
-      "learning_rate": 0.00019999930363239572,
-      "loss": 46.0,
-      "step": 15547
-    },
-    {
-      "epoch": 1.1887531777433722,
-      "grad_norm": 0.0036219563335180283,
-      "learning_rate": 0.00019999930354275302,
-      "loss": 46.0,
-      "step": 15548
-    },
-    {
-      "epoch": 1.188829634726762,
-      "grad_norm": 0.0014206778723746538,
-      "learning_rate": 0.00019999930345310457,
-      "loss": 46.0,
-      "step": 15549
-    },
-    {
-      "epoch": 1.1889060917101515,
-      "grad_norm": 0.0008011855534277856,
-      "learning_rate": 0.00019999930336345035,
-      "loss": 46.0,
-      "step": 15550
-    },
-    {
-      "epoch": 1.1889825486935413,
-      "grad_norm": 0.0005782452644780278,
-      "learning_rate": 0.00019999930327379035,
-      "loss": 46.0,
-      "step": 15551
-    },
-    {
-      "epoch": 1.189059005676931,
-      "grad_norm": 0.0016597473295405507,
-      "learning_rate": 0.00019999930318412458,
-      "loss": 46.0,
-      "step": 15552
-    },
-    {
-      "epoch": 1.1891354626603208,
-      "grad_norm": 0.0007053731824271381,
-      "learning_rate": 0.00019999930309445304,
-      "loss": 46.0,
-      "step": 15553
-    },
-    {
-      "epoch": 1.1892119196437105,
-      "grad_norm": 0.0008106050663627684,
-      "learning_rate": 0.00019999930300477575,
-      "loss": 46.0,
-      "step": 15554
-    },
-    {
-      "epoch": 1.1892883766271,
-      "grad_norm": 0.004580857697874308,
-      "learning_rate": 0.0001999993029150927,
-      "loss": 46.0,
-      "step": 15555
-    },
-    {
-      "epoch": 1.1893648336104898,
-      "grad_norm": 0.0007276267861016095,
-      "learning_rate": 0.00019999930282540383,
-      "loss": 46.0,
-      "step": 15556
-    },
-    {
-      "epoch": 1.1894412905938796,
-      "grad_norm": 0.0011253143893554807,
-      "learning_rate": 0.00019999930273570922,
-      "loss": 46.0,
-      "step": 15557
-    },
-    {
-      "epoch": 1.1895177475772694,
-      "grad_norm": 0.003680725581943989,
-      "learning_rate": 0.00019999930264600883,
-      "loss": 46.0,
-      "step": 15558
-    },
-    {
-      "epoch": 1.1895942045606591,
-      "grad_norm": 0.0020579949487000704,
-      "learning_rate": 0.00019999930255630268,
-      "loss": 46.0,
-      "step": 15559
-    },
-    {
-      "epoch": 1.1896706615440489,
-      "grad_norm": 0.0021691475994884968,
-      "learning_rate": 0.00019999930246659078,
-      "loss": 46.0,
-      "step": 15560
-    },
-    {
-      "epoch": 1.1897471185274384,
-      "grad_norm": 0.0016062804497778416,
-      "learning_rate": 0.0001999993023768731,
-      "loss": 46.0,
-      "step": 15561
-    },
-    {
-      "epoch": 1.1898235755108282,
-      "grad_norm": 0.00032115174690261483,
-      "learning_rate": 0.00019999930228714962,
-      "loss": 46.0,
-      "step": 15562
-    },
-    {
-      "epoch": 1.189900032494218,
-      "grad_norm": 0.001098536653444171,
-      "learning_rate": 0.0001999993021974204,
-      "loss": 46.0,
-      "step": 15563
-    },
-    {
-      "epoch": 1.1899764894776077,
-      "grad_norm": 0.0012571340193971992,
-      "learning_rate": 0.0001999993021076854,
-      "loss": 46.0,
-      "step": 15564
-    },
-    {
-      "epoch": 1.1900529464609975,
-      "grad_norm": 0.0016245145816355944,
-      "learning_rate": 0.00019999930201794466,
-      "loss": 46.0,
-      "step": 15565
-    },
-    {
-      "epoch": 1.190129403444387,
-      "grad_norm": 0.0012835495872423053,
-      "learning_rate": 0.0001999993019281981,
-      "loss": 46.0,
-      "step": 15566
-    },
-    {
-      "epoch": 1.1902058604277768,
-      "grad_norm": 0.0005703620845451951,
-      "learning_rate": 0.0001999993018384458,
-      "loss": 46.0,
-      "step": 15567
-    },
-    {
-      "epoch": 1.1902823174111665,
-      "grad_norm": 0.0012046168558299541,
-      "learning_rate": 0.00019999930174868772,
-      "loss": 46.0,
-      "step": 15568
-    },
-    {
-      "epoch": 1.1903587743945563,
-      "grad_norm": 0.004526298958808184,
-      "learning_rate": 0.00019999930165892388,
-      "loss": 46.0,
-      "step": 15569
-    },
-    {
-      "epoch": 1.190435231377946,
-      "grad_norm": 0.002534755039960146,
-      "learning_rate": 0.00019999930156915427,
-      "loss": 46.0,
-      "step": 15570
-    },
-    {
-      "epoch": 1.1905116883613358,
-      "grad_norm": 0.0031370320357382298,
-      "learning_rate": 0.0001999993014793789,
-      "loss": 46.0,
-      "step": 15571
-    },
-    {
-      "epoch": 1.1905881453447253,
-      "grad_norm": 0.0008780405623838305,
-      "learning_rate": 0.00019999930138959773,
-      "loss": 46.0,
-      "step": 15572
-    },
-    {
-      "epoch": 1.190664602328115,
-      "grad_norm": 0.0004659658297896385,
-      "learning_rate": 0.0001999993012998108,
-      "loss": 46.0,
-      "step": 15573
-    },
-    {
-      "epoch": 1.1907410593115049,
-      "grad_norm": 0.0009031351655721664,
-      "learning_rate": 0.00019999930121001813,
-      "loss": 46.0,
-      "step": 15574
-    },
-    {
-      "epoch": 1.1908175162948946,
-      "grad_norm": 0.0008859423687681556,
-      "learning_rate": 0.00019999930112021965,
-      "loss": 46.0,
-      "step": 15575
-    },
-    {
-      "epoch": 1.1908939732782844,
-      "grad_norm": 0.0036954539828002453,
-      "learning_rate": 0.00019999930103041543,
-      "loss": 46.0,
-      "step": 15576
-    },
-    {
-      "epoch": 1.190970430261674,
-      "grad_norm": 0.002833851845934987,
-      "learning_rate": 0.00019999930094060544,
-      "loss": 46.0,
-      "step": 15577
-    },
-    {
-      "epoch": 1.1910468872450637,
-      "grad_norm": 0.0007078189519234002,
-      "learning_rate": 0.00019999930085078967,
-      "loss": 46.0,
-      "step": 15578
-    },
-    {
-      "epoch": 1.1911233442284535,
-      "grad_norm": 0.002464585704728961,
-      "learning_rate": 0.00019999930076096813,
-      "loss": 46.0,
-      "step": 15579
-    },
-    {
-      "epoch": 1.1911998012118432,
-      "grad_norm": 0.00043234441545791924,
-      "learning_rate": 0.00019999930067114084,
-      "loss": 46.0,
-      "step": 15580
-    },
-    {
-      "epoch": 1.191276258195233,
-      "grad_norm": 0.001243620296008885,
-      "learning_rate": 0.00019999930058130775,
-      "loss": 46.0,
-      "step": 15581
-    },
-    {
-      "epoch": 1.1913527151786227,
-      "grad_norm": 0.0021027387119829655,
-      "learning_rate": 0.00019999930049146891,
-      "loss": 46.0,
-      "step": 15582
-    },
-    {
-      "epoch": 1.1914291721620123,
-      "grad_norm": 0.0007718808483332396,
-      "learning_rate": 0.00019999930040162428,
-      "loss": 46.0,
-      "step": 15583
-    },
-    {
-      "epoch": 1.191505629145402,
-      "grad_norm": 0.0051543209701776505,
-      "learning_rate": 0.0001999993003117739,
-      "loss": 46.0,
-      "step": 15584
-    },
-    {
-      "epoch": 1.1915820861287918,
-      "grad_norm": 0.003042373340576887,
-      "learning_rate": 0.00019999930022191774,
-      "loss": 46.0,
-      "step": 15585
-    },
-    {
-      "epoch": 1.1916585431121816,
-      "grad_norm": 0.0007157353684306145,
-      "learning_rate": 0.00019999930013205584,
-      "loss": 46.0,
-      "step": 15586
-    },
-    {
-      "epoch": 1.1917350000955713,
-      "grad_norm": 0.0048282877542078495,
-      "learning_rate": 0.0001999993000421881,
-      "loss": 46.0,
-      "step": 15587
-    },
-    {
-      "epoch": 1.1918114570789609,
-      "grad_norm": 0.0007894468144513667,
-      "learning_rate": 0.0001999992999523147,
-      "loss": 46.0,
-      "step": 15588
-    },
-    {
-      "epoch": 1.1918879140623506,
-      "grad_norm": 0.0008379890932701528,
-      "learning_rate": 0.0001999992998624354,
-      "loss": 46.0,
-      "step": 15589
-    },
-    {
-      "epoch": 1.1919643710457404,
-      "grad_norm": 0.001715177670121193,
-      "learning_rate": 0.00019999929977255042,
-      "loss": 46.0,
-      "step": 15590
-    },
-    {
-      "epoch": 1.1920408280291301,
-      "grad_norm": 0.0018882370786741376,
-      "learning_rate": 0.00019999929968265968,
-      "loss": 46.0,
-      "step": 15591
-    },
-    {
-      "epoch": 1.19211728501252,
-      "grad_norm": 0.0028970823623239994,
-      "learning_rate": 0.00019999929959276313,
-      "loss": 46.0,
-      "step": 15592
-    },
-    {
-      "epoch": 1.1921937419959097,
-      "grad_norm": 0.0004898144397884607,
-      "learning_rate": 0.00019999929950286082,
-      "loss": 46.0,
-      "step": 15593
-    },
-    {
-      "epoch": 1.1922701989792992,
-      "grad_norm": 0.001743406057357788,
-      "learning_rate": 0.00019999929941295276,
-      "loss": 46.0,
-      "step": 15594
-    },
-    {
-      "epoch": 1.192346655962689,
-      "grad_norm": 0.0008911142940632999,
-      "learning_rate": 0.0001999992993230389,
-      "loss": 46.0,
-      "step": 15595
-    },
-    {
-      "epoch": 1.1924231129460787,
-      "grad_norm": 0.0005246399086900055,
-      "learning_rate": 0.00019999929923311926,
-      "loss": 46.0,
-      "step": 15596
-    },
-    {
-      "epoch": 1.1924995699294685,
-      "grad_norm": 0.0019381000893190503,
-      "learning_rate": 0.00019999929914319385,
-      "loss": 46.0,
-      "step": 15597
-    },
-    {
-      "epoch": 1.1925760269128582,
-      "grad_norm": 0.0005452848272398114,
-      "learning_rate": 0.00019999929905326272,
-      "loss": 46.0,
-      "step": 15598
-    },
-    {
-      "epoch": 1.1926524838962478,
-      "grad_norm": 0.004592359997332096,
-      "learning_rate": 0.0001999992989633258,
-      "loss": 46.0,
-      "step": 15599
-    },
-    {
-      "epoch": 1.1927289408796375,
-      "grad_norm": 0.0026152986101806164,
-      "learning_rate": 0.00019999929887338312,
-      "loss": 46.0,
-      "step": 15600
-    },
-    {
-      "epoch": 1.1928053978630273,
-      "grad_norm": 0.0006257917848415673,
-      "learning_rate": 0.00019999929878343465,
-      "loss": 46.0,
-      "step": 15601
-    },
-    {
-      "epoch": 1.192881854846417,
-      "grad_norm": 0.0011624919716268778,
-      "learning_rate": 0.0001999992986934804,
-      "loss": 46.0,
-      "step": 15602
-    },
-    {
-      "epoch": 1.1929583118298068,
-      "grad_norm": 0.0005555449170060456,
-      "learning_rate": 0.0001999992986035204,
-      "loss": 46.0,
-      "step": 15603
-    },
-    {
-      "epoch": 1.1930347688131966,
-      "grad_norm": 0.0034399430733174086,
-      "learning_rate": 0.00019999929851355464,
-      "loss": 46.0,
-      "step": 15604
-    },
-    {
-      "epoch": 1.1931112257965861,
-      "grad_norm": 0.0009292967733927071,
-      "learning_rate": 0.0001999992984235831,
-      "loss": 46.0,
-      "step": 15605
-    },
-    {
-      "epoch": 1.1931876827799759,
-      "grad_norm": 0.0012462940067052841,
-      "learning_rate": 0.00019999929833360578,
-      "loss": 46.0,
-      "step": 15606
-    },
-    {
-      "epoch": 1.1932641397633656,
-      "grad_norm": 0.0008562547154724598,
-      "learning_rate": 0.0001999992982436227,
-      "loss": 46.0,
-      "step": 15607
-    },
-    {
-      "epoch": 1.1933405967467554,
-      "grad_norm": 0.0008484327117912471,
-      "learning_rate": 0.00019999929815363383,
-      "loss": 46.0,
-      "step": 15608
-    },
-    {
-      "epoch": 1.1934170537301452,
-      "grad_norm": 0.0032459385693073273,
-      "learning_rate": 0.00019999929806363922,
-      "loss": 46.0,
-      "step": 15609
-    },
-    {
-      "epoch": 1.1934935107135347,
-      "grad_norm": 0.000585326284635812,
-      "learning_rate": 0.00019999929797363884,
-      "loss": 46.0,
-      "step": 15610
-    },
-    {
-      "epoch": 1.1935699676969245,
-      "grad_norm": 0.0009977323934435844,
-      "learning_rate": 0.00019999929788363266,
-      "loss": 46.0,
-      "step": 15611
-    },
-    {
-      "epoch": 1.1936464246803142,
-      "grad_norm": 0.0008985247113741934,
-      "learning_rate": 0.00019999929779362076,
-      "loss": 46.0,
-      "step": 15612
-    },
-    {
-      "epoch": 1.193722881663704,
-      "grad_norm": 0.0008775783353485167,
-      "learning_rate": 0.00019999929770360303,
-      "loss": 46.0,
-      "step": 15613
-    },
-    {
-      "epoch": 1.1937993386470938,
-      "grad_norm": 0.0013121296651661396,
-      "learning_rate": 0.00019999929761357955,
-      "loss": 46.0,
-      "step": 15614
-    },
-    {
-      "epoch": 1.1938757956304835,
-      "grad_norm": 0.009874706156551838,
-      "learning_rate": 0.00019999929752355033,
-      "loss": 46.0,
-      "step": 15615
-    },
-    {
-      "epoch": 1.193952252613873,
-      "grad_norm": 0.002934205112978816,
-      "learning_rate": 0.00019999929743351534,
-      "loss": 46.0,
-      "step": 15616
-    },
-    {
-      "epoch": 1.1940287095972628,
-      "grad_norm": 0.0009753283229656518,
-      "learning_rate": 0.00019999929734347454,
-      "loss": 46.0,
-      "step": 15617
-    },
-    {
-      "epoch": 1.1941051665806526,
-      "grad_norm": 0.0009777714731171727,
-      "learning_rate": 0.000199999297253428,
-      "loss": 46.0,
-      "step": 15618
-    },
-    {
-      "epoch": 1.1941816235640423,
-      "grad_norm": 0.0043449425138533115,
-      "learning_rate": 0.0001999992971633757,
-      "loss": 46.0,
-      "step": 15619
-    },
-    {
-      "epoch": 1.1942580805474319,
-      "grad_norm": 0.09789062291383743,
-      "learning_rate": 0.00019999929707331763,
-      "loss": 46.0,
-      "step": 15620
-    },
-    {
-      "epoch": 1.1943345375308216,
-      "grad_norm": 0.011709102429449558,
-      "learning_rate": 0.00019999929698325377,
-      "loss": 46.0,
-      "step": 15621
-    },
-    {
-      "epoch": 1.1944109945142114,
-      "grad_norm": 0.0004991114838048816,
-      "learning_rate": 0.00019999929689318413,
-      "loss": 46.0,
-      "step": 15622
-    },
-    {
-      "epoch": 1.1944874514976012,
-      "grad_norm": 0.001007484970614314,
-      "learning_rate": 0.00019999929680310875,
-      "loss": 46.0,
-      "step": 15623
-    },
-    {
-      "epoch": 1.194563908480991,
-      "grad_norm": 0.0025881242472678423,
-      "learning_rate": 0.00019999929671302757,
-      "loss": 46.0,
-      "step": 15624
-    },
-    {
-      "epoch": 1.1946403654643807,
-      "grad_norm": 0.0012657179031521082,
-      "learning_rate": 0.00019999929662294067,
-      "loss": 46.0,
-      "step": 15625
-    },
-    {
-      "epoch": 1.1947168224477704,
-      "grad_norm": 0.0005705134244635701,
-      "learning_rate": 0.00019999929653284797,
-      "loss": 46.0,
-      "step": 15626
-    },
-    {
-      "epoch": 1.19479327943116,
-      "grad_norm": 0.0005474984645843506,
-      "learning_rate": 0.0001999992964427495,
-      "loss": 46.0,
-      "step": 15627
-    },
-    {
-      "epoch": 1.1948697364145497,
-      "grad_norm": 0.0007833965355530381,
-      "learning_rate": 0.00019999929635264524,
-      "loss": 46.0,
-      "step": 15628
-    },
-    {
-      "epoch": 1.1949461933979395,
-      "grad_norm": 0.0006912931567057967,
-      "learning_rate": 0.00019999929626253525,
-      "loss": 46.0,
-      "step": 15629
-    },
-    {
-      "epoch": 1.1950226503813293,
-      "grad_norm": 0.0016190021997317672,
-      "learning_rate": 0.00019999929617241946,
-      "loss": 46.0,
-      "step": 15630
-    },
-    {
-      "epoch": 1.1950991073647188,
-      "grad_norm": 0.0013823227491229773,
-      "learning_rate": 0.00019999929608229791,
-      "loss": 46.0,
-      "step": 15631
-    },
-    {
-      "epoch": 1.1951755643481086,
-      "grad_norm": 0.002389416564255953,
-      "learning_rate": 0.0001999992959921706,
-      "loss": 46.0,
-      "step": 15632
-    },
-    {
-      "epoch": 1.1952520213314983,
-      "grad_norm": 0.0012995239812880754,
-      "learning_rate": 0.00019999929590203749,
-      "loss": 46.0,
-      "step": 15633
-    },
-    {
-      "epoch": 1.195328478314888,
-      "grad_norm": 0.003739702282473445,
-      "learning_rate": 0.00019999929581189868,
-      "loss": 46.0,
-      "step": 15634
-    },
-    {
-      "epoch": 1.1954049352982778,
-      "grad_norm": 0.005378869362175465,
-      "learning_rate": 0.00019999929572175402,
-      "loss": 46.0,
-      "step": 15635
-    },
-    {
-      "epoch": 1.1954813922816676,
-      "grad_norm": 0.003659531706944108,
-      "learning_rate": 0.00019999929563160364,
-      "loss": 46.0,
-      "step": 15636
-    },
-    {
-      "epoch": 1.1955578492650574,
-      "grad_norm": 0.0006310660974122584,
-      "learning_rate": 0.00019999929554144748,
-      "loss": 46.0,
-      "step": 15637
-    },
-    {
-      "epoch": 1.195634306248447,
-      "grad_norm": 0.0013169602025300264,
-      "learning_rate": 0.00019999929545128553,
-      "loss": 46.0,
-      "step": 15638
-    },
-    {
-      "epoch": 1.1957107632318367,
-      "grad_norm": 0.001478321268223226,
-      "learning_rate": 0.00019999929536111783,
-      "loss": 46.0,
-      "step": 15639
-    },
-    {
-      "epoch": 1.1957872202152264,
-      "grad_norm": 0.0013395135756582022,
-      "learning_rate": 0.00019999929527094435,
-      "loss": 46.0,
-      "step": 15640
-    },
-    {
-      "epoch": 1.1958636771986162,
-      "grad_norm": 0.0013497895561158657,
-      "learning_rate": 0.00019999929518076513,
-      "loss": 46.0,
-      "step": 15641
-    },
-    {
-      "epoch": 1.1959401341820057,
-      "grad_norm": 0.0008274232386611402,
-      "learning_rate": 0.0001999992950905801,
-      "loss": 46.0,
-      "step": 15642
-    },
-    {
-      "epoch": 1.1960165911653955,
-      "grad_norm": 0.0005337708862498403,
-      "learning_rate": 0.00019999929500038932,
-      "loss": 46.0,
-      "step": 15643
-    },
-    {
-      "epoch": 1.1960930481487853,
-      "grad_norm": 0.0018749083392322063,
-      "learning_rate": 0.00019999929491019278,
-      "loss": 46.0,
-      "step": 15644
-    },
-    {
-      "epoch": 1.196169505132175,
-      "grad_norm": 0.0007693894440308213,
-      "learning_rate": 0.00019999929481999046,
-      "loss": 46.0,
-      "step": 15645
-    },
-    {
-      "epoch": 1.1962459621155648,
-      "grad_norm": 0.029752876609563828,
-      "learning_rate": 0.00019999929472978235,
-      "loss": 46.0,
-      "step": 15646
-    },
-    {
-      "epoch": 1.1963224190989545,
-      "grad_norm": 0.0005141844158060849,
-      "learning_rate": 0.00019999929463956852,
-      "loss": 46.0,
-      "step": 15647
-    },
-    {
-      "epoch": 1.196398876082344,
-      "grad_norm": 0.0018940542358905077,
-      "learning_rate": 0.00019999929454934888,
-      "loss": 46.0,
-      "step": 15648
-    },
-    {
-      "epoch": 1.1964753330657338,
-      "grad_norm": 0.005433386657387018,
-      "learning_rate": 0.0001999992944591235,
-      "loss": 46.0,
-      "step": 15649
-    },
-    {
-      "epoch": 1.1965517900491236,
-      "grad_norm": 0.0018717003986239433,
-      "learning_rate": 0.00019999929436889232,
-      "loss": 46.0,
-      "step": 15650
-    },
-    {
-      "epoch": 1.1966282470325134,
-      "grad_norm": 0.002483527874574065,
-      "learning_rate": 0.0001999992942786554,
-      "loss": 46.0,
-      "step": 15651
-    },
-    {
-      "epoch": 1.1967047040159031,
-      "grad_norm": 0.003623990109190345,
-      "learning_rate": 0.00019999929418841267,
-      "loss": 46.0,
-      "step": 15652
-    },
-    {
-      "epoch": 1.1967811609992927,
-      "grad_norm": 0.03166210651397705,
-      "learning_rate": 0.00019999929409816422,
-      "loss": 46.0,
-      "step": 15653
-    },
-    {
-      "epoch": 1.1968576179826824,
-      "grad_norm": 0.0007963934913277626,
-      "learning_rate": 0.00019999929400790997,
-      "loss": 46.0,
-      "step": 15654
-    },
-    {
-      "epoch": 1.1969340749660722,
-      "grad_norm": 0.0017357331234961748,
-      "learning_rate": 0.00019999929391764995,
-      "loss": 46.0,
-      "step": 15655
-    },
-    {
-      "epoch": 1.197010531949462,
-      "grad_norm": 0.0056830416433513165,
-      "learning_rate": 0.00019999929382738416,
-      "loss": 46.0,
-      "step": 15656
-    },
-    {
-      "epoch": 1.1970869889328517,
-      "grad_norm": 0.0014931554906070232,
-      "learning_rate": 0.00019999929373711262,
-      "loss": 46.0,
-      "step": 15657
-    },
-    {
-      "epoch": 1.1971634459162415,
-      "grad_norm": 0.009246180765330791,
-      "learning_rate": 0.0001999992936468353,
-      "loss": 46.0,
-      "step": 15658
-    },
-    {
-      "epoch": 1.197239902899631,
-      "grad_norm": 0.003291910281404853,
-      "learning_rate": 0.00019999929355655222,
-      "loss": 46.0,
-      "step": 15659
-    },
-    {
-      "epoch": 1.1973163598830208,
-      "grad_norm": 0.003330525942146778,
-      "learning_rate": 0.00019999929346626333,
-      "loss": 46.0,
-      "step": 15660
-    },
-    {
-      "epoch": 1.1973928168664105,
-      "grad_norm": 0.0008768620318733156,
-      "learning_rate": 0.00019999929337596872,
-      "loss": 46.0,
-      "step": 15661
-    },
-    {
-      "epoch": 1.1974692738498003,
-      "grad_norm": 0.0006625146488659084,
-      "learning_rate": 0.00019999929328566834,
-      "loss": 46.0,
-      "step": 15662
-    },
-    {
-      "epoch": 1.19754573083319,
-      "grad_norm": 0.0009174922597594559,
-      "learning_rate": 0.00019999929319536213,
-      "loss": 46.0,
-      "step": 15663
-    },
-    {
-      "epoch": 1.1976221878165796,
-      "grad_norm": 0.0017498661763966084,
-      "learning_rate": 0.0001999992931050502,
-      "loss": 46.0,
-      "step": 15664
-    },
-    {
-      "epoch": 1.1976986447999693,
-      "grad_norm": 0.012102453038096428,
-      "learning_rate": 0.0001999992930147325,
-      "loss": 46.0,
-      "step": 15665
-    },
-    {
-      "epoch": 1.197775101783359,
-      "grad_norm": 0.0007139602093957365,
-      "learning_rate": 0.00019999929292440903,
-      "loss": 46.0,
-      "step": 15666
-    },
-    {
-      "epoch": 1.1978515587667489,
-      "grad_norm": 0.0013389961095526814,
-      "learning_rate": 0.0001999992928340798,
-      "loss": 46.0,
-      "step": 15667
-    },
-    {
-      "epoch": 1.1979280157501386,
-      "grad_norm": 0.002476147376000881,
-      "learning_rate": 0.00019999929274374477,
-      "loss": 46.0,
-      "step": 15668
-    },
-    {
-      "epoch": 1.1980044727335284,
-      "grad_norm": 0.004321780521422625,
-      "learning_rate": 0.00019999929265340397,
-      "loss": 46.0,
-      "step": 15669
-    },
-    {
-      "epoch": 1.198080929716918,
-      "grad_norm": 0.001915768370963633,
-      "learning_rate": 0.00019999929256305743,
-      "loss": 46.0,
-      "step": 15670
-    },
-    {
-      "epoch": 1.1981573867003077,
-      "grad_norm": 0.000593613600358367,
-      "learning_rate": 0.0001999992924727051,
-      "loss": 46.0,
-      "step": 15671
-    },
-    {
-      "epoch": 1.1982338436836975,
-      "grad_norm": 0.0013915409799665213,
-      "learning_rate": 0.000199999292382347,
-      "loss": 46.0,
-      "step": 15672
-    },
-    {
-      "epoch": 1.1983103006670872,
-      "grad_norm": 0.0040787504985928535,
-      "learning_rate": 0.00019999929229198315,
-      "loss": 46.0,
-      "step": 15673
-    },
-    {
-      "epoch": 1.198386757650477,
-      "grad_norm": 0.0006661030347459018,
-      "learning_rate": 0.0001999992922016135,
-      "loss": 46.0,
-      "step": 15674
-    },
-    {
-      "epoch": 1.1984632146338665,
-      "grad_norm": 0.0006008045747876167,
-      "learning_rate": 0.0001999992921112381,
-      "loss": 46.0,
-      "step": 15675
-    },
-    {
-      "epoch": 1.1985396716172563,
-      "grad_norm": 0.010750112123787403,
-      "learning_rate": 0.00019999929202085695,
-      "loss": 46.0,
-      "step": 15676
-    },
-    {
-      "epoch": 1.198616128600646,
-      "grad_norm": 0.004843415226787329,
-      "learning_rate": 0.00019999929193047,
-      "loss": 46.0,
-      "step": 15677
-    },
-    {
-      "epoch": 1.1986925855840358,
-      "grad_norm": 0.0017501342808827758,
-      "learning_rate": 0.00019999929184007727,
-      "loss": 46.0,
-      "step": 15678
-    },
-    {
-      "epoch": 1.1987690425674256,
-      "grad_norm": 0.0016653920756652951,
-      "learning_rate": 0.00019999929174967883,
-      "loss": 46.0,
-      "step": 15679
-    },
-    {
-      "epoch": 1.1988454995508153,
-      "grad_norm": 0.001998928841203451,
-      "learning_rate": 0.00019999929165927455,
-      "loss": 46.0,
-      "step": 15680
-    },
-    {
-      "epoch": 1.1989219565342049,
-      "grad_norm": 0.0014376409817487001,
-      "learning_rate": 0.00019999929156886453,
-      "loss": 46.0,
-      "step": 15681
-    },
-    {
-      "epoch": 1.1989984135175946,
-      "grad_norm": 0.0010311650112271309,
-      "learning_rate": 0.00019999929147844877,
-      "loss": 46.0,
-      "step": 15682
-    },
-    {
-      "epoch": 1.1990748705009844,
-      "grad_norm": 0.0006080266321077943,
-      "learning_rate": 0.0001999992913880272,
-      "loss": 46.0,
-      "step": 15683
-    },
-    {
-      "epoch": 1.1991513274843741,
-      "grad_norm": 0.0011371842119842768,
-      "learning_rate": 0.00019999929129759986,
-      "loss": 46.0,
-      "step": 15684
-    },
-    {
-      "epoch": 1.199227784467764,
-      "grad_norm": 0.0012255256297066808,
-      "learning_rate": 0.00019999929120716677,
-      "loss": 46.0,
-      "step": 15685
-    },
-    {
-      "epoch": 1.1993042414511534,
-      "grad_norm": 0.003824906889349222,
-      "learning_rate": 0.00019999929111672794,
-      "loss": 46.0,
-      "step": 15686
-    },
-    {
-      "epoch": 1.1993806984345432,
-      "grad_norm": 0.0009307143627665937,
-      "learning_rate": 0.00019999929102628328,
-      "loss": 46.0,
-      "step": 15687
-    },
-    {
-      "epoch": 1.199457155417933,
-      "grad_norm": 0.004113251343369484,
-      "learning_rate": 0.00019999929093583287,
-      "loss": 46.0,
-      "step": 15688
-    },
-    {
-      "epoch": 1.1995336124013227,
-      "grad_norm": 0.0006817805697210133,
-      "learning_rate": 0.0001999992908453767,
-      "loss": 46.0,
-      "step": 15689
-    },
-    {
-      "epoch": 1.1996100693847125,
-      "grad_norm": 0.007668792735785246,
-      "learning_rate": 0.00019999929075491477,
-      "loss": 46.0,
-      "step": 15690
-    },
-    {
-      "epoch": 1.1996865263681022,
-      "grad_norm": 0.0011725733056664467,
-      "learning_rate": 0.00019999929066444704,
-      "loss": 46.0,
-      "step": 15691
-    },
-    {
-      "epoch": 1.1997629833514918,
-      "grad_norm": 0.0013711429201066494,
-      "learning_rate": 0.00019999929057397354,
-      "loss": 46.0,
-      "step": 15692
-    },
-    {
-      "epoch": 1.1998394403348815,
-      "grad_norm": 0.014447052031755447,
-      "learning_rate": 0.00019999929048349432,
-      "loss": 46.0,
-      "step": 15693
-    },
-    {
-      "epoch": 1.1999158973182713,
-      "grad_norm": 0.0028197262436151505,
-      "learning_rate": 0.00019999929039300927,
-      "loss": 46.0,
-      "step": 15694
-    },
-    {
-      "epoch": 1.199992354301661,
-      "grad_norm": 0.0005684084026142955,
-      "learning_rate": 0.0001999992903025185,
-      "loss": 46.0,
-      "step": 15695
-    },
-    {
-      "epoch": 1.2000688112850508,
-      "grad_norm": 0.0021674875169992447,
-      "learning_rate": 0.00019999929021202192,
-      "loss": 46.0,
-      "step": 15696
-    },
-    {
-      "epoch": 1.2001452682684404,
-      "grad_norm": 0.0014034503838047385,
-      "learning_rate": 0.0001999992901215196,
-      "loss": 46.0,
-      "step": 15697
-    },
-    {
-      "epoch": 1.2002217252518301,
-      "grad_norm": 0.0005880245007574558,
-      "learning_rate": 0.00019999929003101152,
-      "loss": 46.0,
-      "step": 15698
-    },
-    {
-      "epoch": 1.20029818223522,
-      "grad_norm": 0.0023854663595557213,
-      "learning_rate": 0.00019999928994049766,
-      "loss": 46.0,
-      "step": 15699
-    },
-    {
-      "epoch": 1.2003746392186097,
-      "grad_norm": 0.0008366689435206354,
-      "learning_rate": 0.00019999928984997802,
-      "loss": 46.0,
-      "step": 15700
-    },
-    {
-      "epoch": 1.2004510962019994,
-      "grad_norm": 0.0010805936763063073,
-      "learning_rate": 0.00019999928975945262,
-      "loss": 46.0,
-      "step": 15701
-    },
-    {
-      "epoch": 1.2005275531853892,
-      "grad_norm": 0.0017285803332924843,
-      "learning_rate": 0.00019999928966892144,
-      "loss": 46.0,
-      "step": 15702
-    },
-    {
-      "epoch": 1.2006040101687787,
-      "grad_norm": 0.0006802756688557565,
-      "learning_rate": 0.00019999928957838449,
-      "loss": 46.0,
-      "step": 15703
-    },
-    {
-      "epoch": 1.2006804671521685,
-      "grad_norm": 0.0007277424447238445,
-      "learning_rate": 0.00019999928948784176,
-      "loss": 46.0,
-      "step": 15704
-    },
-    {
-      "epoch": 1.2007569241355582,
-      "grad_norm": 0.0008758124895393848,
-      "learning_rate": 0.0001999992893972933,
-      "loss": 46.0,
-      "step": 15705
-    },
-    {
-      "epoch": 1.200833381118948,
-      "grad_norm": 0.0008830575970932841,
-      "learning_rate": 0.00019999928930673904,
-      "loss": 46.0,
-      "step": 15706
-    },
-    {
-      "epoch": 1.2009098381023378,
-      "grad_norm": 0.0011764157097786665,
-      "learning_rate": 0.00019999928921617902,
-      "loss": 46.0,
-      "step": 15707
-    },
-    {
-      "epoch": 1.2009862950857273,
-      "grad_norm": 0.001333172433078289,
-      "learning_rate": 0.00019999928912561323,
-      "loss": 46.0,
-      "step": 15708
-    },
-    {
-      "epoch": 1.201062752069117,
-      "grad_norm": 0.0013649067841470242,
-      "learning_rate": 0.00019999928903504166,
-      "loss": 46.0,
-      "step": 15709
-    },
-    {
-      "epoch": 1.2011392090525068,
-      "grad_norm": 0.002075194614008069,
-      "learning_rate": 0.00019999928894446432,
-      "loss": 46.0,
-      "step": 15710
-    },
-    {
-      "epoch": 1.2012156660358966,
-      "grad_norm": 0.0008259046007879078,
-      "learning_rate": 0.0001999992888538812,
-      "loss": 46.0,
-      "step": 15711
-    },
-    {
-      "epoch": 1.2012921230192863,
-      "grad_norm": 0.002426484599709511,
-      "learning_rate": 0.00019999928876329235,
-      "loss": 46.0,
-      "step": 15712
-    },
-    {
-      "epoch": 1.201368580002676,
-      "grad_norm": 0.0007950731087476015,
-      "learning_rate": 0.0001999992886726977,
-      "loss": 46.0,
-      "step": 15713
-    },
-    {
-      "epoch": 1.2014450369860656,
-      "grad_norm": 0.0007124730036593974,
-      "learning_rate": 0.0001999992885820973,
-      "loss": 46.0,
-      "step": 15714
-    },
-    {
-      "epoch": 1.2015214939694554,
-      "grad_norm": 0.0008882875554263592,
-      "learning_rate": 0.00019999928849149113,
-      "loss": 46.0,
-      "step": 15715
-    },
-    {
-      "epoch": 1.2015979509528452,
-      "grad_norm": 0.009240842424333096,
-      "learning_rate": 0.00019999928840087918,
-      "loss": 46.0,
-      "step": 15716
-    },
-    {
-      "epoch": 1.201674407936235,
-      "grad_norm": 0.0008655826677568257,
-      "learning_rate": 0.00019999928831026146,
-      "loss": 46.0,
-      "step": 15717
-    },
-    {
-      "epoch": 1.2017508649196247,
-      "grad_norm": 0.0038563332054764032,
-      "learning_rate": 0.00019999928821963798,
-      "loss": 46.0,
-      "step": 15718
-    },
-    {
-      "epoch": 1.2018273219030142,
-      "grad_norm": 0.0011683328775689006,
-      "learning_rate": 0.0001999992881290087,
-      "loss": 46.0,
-      "step": 15719
-    },
-    {
-      "epoch": 1.201903778886404,
-      "grad_norm": 0.00226473156362772,
-      "learning_rate": 0.0001999992880383737,
-      "loss": 46.0,
-      "step": 15720
-    },
-    {
-      "epoch": 1.2019802358697937,
-      "grad_norm": 0.020189478993415833,
-      "learning_rate": 0.0001999992879477329,
-      "loss": 46.0,
-      "step": 15721
-    },
-    {
-      "epoch": 1.2020566928531835,
-      "grad_norm": 0.004795554094016552,
-      "learning_rate": 0.00019999928785708636,
-      "loss": 46.0,
-      "step": 15722
-    },
-    {
-      "epoch": 1.2021331498365733,
-      "grad_norm": 0.003608362516388297,
-      "learning_rate": 0.00019999928776643402,
-      "loss": 46.0,
-      "step": 15723
-    },
-    {
-      "epoch": 1.202209606819963,
-      "grad_norm": 0.001722640939988196,
-      "learning_rate": 0.0001999992876757759,
-      "loss": 46.0,
-      "step": 15724
-    },
-    {
-      "epoch": 1.2022860638033526,
-      "grad_norm": 0.019515084102749825,
-      "learning_rate": 0.00019999928758511205,
-      "loss": 46.0,
-      "step": 15725
-    },
-    {
-      "epoch": 1.2023625207867423,
-      "grad_norm": 0.003746159840375185,
-      "learning_rate": 0.0001999992874944424,
-      "loss": 46.0,
-      "step": 15726
-    },
-    {
-      "epoch": 1.202438977770132,
-      "grad_norm": 0.0007993843173608184,
-      "learning_rate": 0.00019999928740376696,
-      "loss": 46.0,
-      "step": 15727
-    },
-    {
-      "epoch": 1.2025154347535219,
-      "grad_norm": 0.0006316457875072956,
-      "learning_rate": 0.0001999992873130858,
-      "loss": 46.0,
-      "step": 15728
-    },
-    {
-      "epoch": 1.2025918917369116,
-      "grad_norm": 0.0013367807259783149,
-      "learning_rate": 0.00019999928722239885,
-      "loss": 46.0,
-      "step": 15729
-    },
-    {
-      "epoch": 1.2026683487203012,
-      "grad_norm": 0.001325565972365439,
-      "learning_rate": 0.00019999928713170615,
-      "loss": 46.0,
-      "step": 15730
-    },
-    {
-      "epoch": 1.202744805703691,
-      "grad_norm": 0.0017381255747750401,
-      "learning_rate": 0.00019999928704100763,
-      "loss": 46.0,
-      "step": 15731
-    },
-    {
-      "epoch": 1.2028212626870807,
-      "grad_norm": 0.0007777658174745739,
-      "learning_rate": 0.00019999928695030338,
-      "loss": 46.0,
-      "step": 15732
-    },
-    {
-      "epoch": 1.2028977196704704,
-      "grad_norm": 0.0052057718858122826,
-      "learning_rate": 0.00019999928685959336,
-      "loss": 46.0,
-      "step": 15733
-    },
-    {
-      "epoch": 1.2029741766538602,
-      "grad_norm": 0.0016010685358196497,
-      "learning_rate": 0.00019999928676887755,
-      "loss": 46.0,
-      "step": 15734
-    },
-    {
-      "epoch": 1.20305063363725,
-      "grad_norm": 0.008380594663321972,
-      "learning_rate": 0.000199999286678156,
-      "loss": 46.0,
-      "step": 15735
-    },
-    {
-      "epoch": 1.2031270906206395,
-      "grad_norm": 0.0013300662394613028,
-      "learning_rate": 0.00019999928658742867,
-      "loss": 46.0,
-      "step": 15736
-    },
-    {
-      "epoch": 1.2032035476040293,
-      "grad_norm": 0.0024072073865681887,
-      "learning_rate": 0.00019999928649669556,
-      "loss": 46.0,
-      "step": 15737
-    },
-    {
-      "epoch": 1.203280004587419,
-      "grad_norm": 0.0009984906064346433,
-      "learning_rate": 0.0001999992864059567,
-      "loss": 46.0,
-      "step": 15738
-    },
-    {
-      "epoch": 1.2033564615708088,
-      "grad_norm": 0.0010214403737336397,
-      "learning_rate": 0.00019999928631521204,
-      "loss": 46.0,
-      "step": 15739
-    },
-    {
-      "epoch": 1.2034329185541985,
-      "grad_norm": 0.0011432041646912694,
-      "learning_rate": 0.00019999928622446163,
-      "loss": 46.0,
-      "step": 15740
-    },
-    {
-      "epoch": 1.203509375537588,
-      "grad_norm": 0.0008477675728499889,
-      "learning_rate": 0.00019999928613370545,
-      "loss": 46.0,
-      "step": 15741
-    },
-    {
-      "epoch": 1.2035858325209778,
-      "grad_norm": 0.0018691600998863578,
-      "learning_rate": 0.0001999992860429435,
-      "loss": 46.0,
-      "step": 15742
-    },
-    {
-      "epoch": 1.2036622895043676,
-      "grad_norm": 0.0011603889288380742,
-      "learning_rate": 0.00019999928595217575,
-      "loss": 46.0,
-      "step": 15743
-    },
-    {
-      "epoch": 1.2037387464877574,
-      "grad_norm": 0.0018194845179095864,
-      "learning_rate": 0.00019999928586140228,
-      "loss": 46.0,
-      "step": 15744
-    },
-    {
-      "epoch": 1.2038152034711471,
-      "grad_norm": 0.005773098673671484,
-      "learning_rate": 0.000199999285770623,
-      "loss": 46.0,
-      "step": 15745
-    },
-    {
-      "epoch": 1.2038916604545369,
-      "grad_norm": 0.0022576279006898403,
-      "learning_rate": 0.000199999285679838,
-      "loss": 46.0,
-      "step": 15746
-    },
-    {
-      "epoch": 1.2039681174379264,
-      "grad_norm": 0.0007069289567880332,
-      "learning_rate": 0.00019999928558904717,
-      "loss": 46.0,
-      "step": 15747
-    },
-    {
-      "epoch": 1.2040445744213162,
-      "grad_norm": 0.001302338787354529,
-      "learning_rate": 0.0001999992854982506,
-      "loss": 46.0,
-      "step": 15748
-    },
-    {
-      "epoch": 1.204121031404706,
-      "grad_norm": 0.0061085522174835205,
-      "learning_rate": 0.00019999928540744827,
-      "loss": 46.0,
-      "step": 15749
-    },
-    {
-      "epoch": 1.2041974883880957,
-      "grad_norm": 0.0017046869033947587,
-      "learning_rate": 0.00019999928531664018,
-      "loss": 46.0,
-      "step": 15750
-    },
-    {
-      "epoch": 1.2042739453714852,
-      "grad_norm": 0.003390026278793812,
-      "learning_rate": 0.0001999992852258263,
-      "loss": 46.0,
-      "step": 15751
-    },
-    {
-      "epoch": 1.204350402354875,
-      "grad_norm": 0.0013632539194077253,
-      "learning_rate": 0.00019999928513500664,
-      "loss": 46.0,
-      "step": 15752
-    },
-    {
-      "epoch": 1.2044268593382648,
-      "grad_norm": 0.003184331115335226,
-      "learning_rate": 0.00019999928504418123,
-      "loss": 46.0,
-      "step": 15753
-    },
-    {
-      "epoch": 1.2045033163216545,
-      "grad_norm": 0.0018485623877495527,
-      "learning_rate": 0.00019999928495335003,
-      "loss": 46.0,
-      "step": 15754
-    },
-    {
-      "epoch": 1.2045797733050443,
-      "grad_norm": 0.0009776364313438535,
-      "learning_rate": 0.0001999992848625131,
-      "loss": 46.0,
-      "step": 15755
-    },
-    {
-      "epoch": 1.204656230288434,
-      "grad_norm": 0.001214486313983798,
-      "learning_rate": 0.00019999928477167038,
-      "loss": 46.0,
-      "step": 15756
-    },
-    {
-      "epoch": 1.2047326872718238,
-      "grad_norm": 0.004157719202339649,
-      "learning_rate": 0.00019999928468082188,
-      "loss": 46.0,
-      "step": 15757
-    },
-    {
-      "epoch": 1.2048091442552133,
-      "grad_norm": 0.0006044196197763085,
-      "learning_rate": 0.0001999992845899676,
-      "loss": 46.0,
-      "step": 15758
-    },
-    {
-      "epoch": 1.204885601238603,
-      "grad_norm": 0.0008829439175315201,
-      "learning_rate": 0.00019999928449910756,
-      "loss": 46.0,
-      "step": 15759
-    },
-    {
-      "epoch": 1.2049620582219929,
-      "grad_norm": 0.0009564216015860438,
-      "learning_rate": 0.00019999928440824177,
-      "loss": 46.0,
-      "step": 15760
-    },
-    {
-      "epoch": 1.2050385152053826,
-      "grad_norm": 0.002625989029183984,
-      "learning_rate": 0.0001999992843173702,
-      "loss": 46.0,
-      "step": 15761
-    },
-    {
-      "epoch": 1.2051149721887722,
-      "grad_norm": 0.0009223964880220592,
-      "learning_rate": 0.00019999928422649287,
-      "loss": 46.0,
-      "step": 15762
-    },
-    {
-      "epoch": 1.205191429172162,
-      "grad_norm": 0.0006487914943136275,
-      "learning_rate": 0.00019999928413560976,
-      "loss": 46.0,
-      "step": 15763
-    },
-    {
-      "epoch": 1.2052678861555517,
-      "grad_norm": 0.0027385635767132044,
-      "learning_rate": 0.00019999928404472087,
-      "loss": 46.0,
-      "step": 15764
-    },
-    {
-      "epoch": 1.2053443431389415,
-      "grad_norm": 0.00044901890214532614,
-      "learning_rate": 0.00019999928395382624,
-      "loss": 46.0,
-      "step": 15765
-    },
-    {
-      "epoch": 1.2054208001223312,
-      "grad_norm": 0.0007760389125905931,
-      "learning_rate": 0.0001999992838629258,
-      "loss": 46.0,
-      "step": 15766
-    },
-    {
-      "epoch": 1.205497257105721,
-      "grad_norm": 0.0009632749133743346,
-      "learning_rate": 0.00019999928377201963,
-      "loss": 46.0,
-      "step": 15767
-    },
-    {
-      "epoch": 1.2055737140891107,
-      "grad_norm": 0.0009677984635345638,
-      "learning_rate": 0.00019999928368110768,
-      "loss": 46.0,
-      "step": 15768
-    },
-    {
-      "epoch": 1.2056501710725003,
-      "grad_norm": 0.0023266051430255175,
-      "learning_rate": 0.00019999928359018993,
-      "loss": 46.0,
-      "step": 15769
-    },
-    {
-      "epoch": 1.20572662805589,
-      "grad_norm": 0.0021874792873859406,
-      "learning_rate": 0.00019999928349926646,
-      "loss": 46.0,
-      "step": 15770
-    },
-    {
-      "epoch": 1.2058030850392798,
-      "grad_norm": 0.0007334948168136179,
-      "learning_rate": 0.00019999928340833718,
-      "loss": 46.0,
-      "step": 15771
-    },
-    {
-      "epoch": 1.2058795420226696,
-      "grad_norm": 0.01449010893702507,
-      "learning_rate": 0.00019999928331740214,
-      "loss": 46.0,
-      "step": 15772
-    },
-    {
-      "epoch": 1.205955999006059,
-      "grad_norm": 0.00066584930755198,
-      "learning_rate": 0.00019999928322646135,
-      "loss": 46.0,
-      "step": 15773
-    },
-    {
-      "epoch": 1.2060324559894489,
-      "grad_norm": 0.002816611435264349,
-      "learning_rate": 0.00019999928313551478,
-      "loss": 46.0,
-      "step": 15774
-    },
-    {
-      "epoch": 1.2061089129728386,
-      "grad_norm": 0.0020622105803340673,
-      "learning_rate": 0.00019999928304456242,
-      "loss": 46.0,
-      "step": 15775
-    },
-    {
-      "epoch": 1.2061853699562284,
-      "grad_norm": 0.001504861400462687,
-      "learning_rate": 0.00019999928295360433,
-      "loss": 46.0,
-      "step": 15776
-    },
-    {
-      "epoch": 1.2062618269396181,
-      "grad_norm": 0.001224591745994985,
-      "learning_rate": 0.00019999928286264042,
-      "loss": 46.0,
-      "step": 15777
-    },
-    {
-      "epoch": 1.206338283923008,
-      "grad_norm": 0.0012241953518241644,
-      "learning_rate": 0.0001999992827716708,
-      "loss": 46.0,
-      "step": 15778
-    },
-    {
-      "epoch": 1.2064147409063974,
-      "grad_norm": 0.005175963044166565,
-      "learning_rate": 0.00019999928268069533,
-      "loss": 46.0,
-      "step": 15779
-    },
-    {
-      "epoch": 1.2064911978897872,
-      "grad_norm": 0.000688257219735533,
-      "learning_rate": 0.00019999928258971416,
-      "loss": 46.0,
-      "step": 15780
-    },
-    {
-      "epoch": 1.206567654873177,
-      "grad_norm": 0.0015347795560956001,
-      "learning_rate": 0.0001999992824987272,
-      "loss": 46.0,
-      "step": 15781
-    },
-    {
-      "epoch": 1.2066441118565667,
-      "grad_norm": 0.0006922337342984974,
-      "learning_rate": 0.00019999928240773448,
-      "loss": 46.0,
-      "step": 15782
-    },
-    {
-      "epoch": 1.2067205688399565,
-      "grad_norm": 0.006682609673589468,
-      "learning_rate": 0.00019999928231673598,
-      "loss": 46.0,
-      "step": 15783
-    },
-    {
-      "epoch": 1.206797025823346,
-      "grad_norm": 0.0013085160171613097,
-      "learning_rate": 0.0001999992822257317,
-      "loss": 46.0,
-      "step": 15784
-    },
-    {
-      "epoch": 1.2068734828067358,
-      "grad_norm": 0.0023341807536780834,
-      "learning_rate": 0.00019999928213472167,
-      "loss": 46.0,
-      "step": 15785
-    },
-    {
-      "epoch": 1.2069499397901255,
-      "grad_norm": 0.0010500644566491246,
-      "learning_rate": 0.00019999928204370585,
-      "loss": 46.0,
-      "step": 15786
-    },
-    {
-      "epoch": 1.2070263967735153,
-      "grad_norm": 0.006038394756615162,
-      "learning_rate": 0.00019999928195268428,
-      "loss": 46.0,
-      "step": 15787
-    },
-    {
-      "epoch": 1.207102853756905,
-      "grad_norm": 0.0005848530563525856,
-      "learning_rate": 0.00019999928186165695,
-      "loss": 46.0,
-      "step": 15788
-    },
-    {
-      "epoch": 1.2071793107402948,
-      "grad_norm": 0.001309465034864843,
-      "learning_rate": 0.00019999928177062384,
-      "loss": 46.0,
-      "step": 15789
-    },
-    {
-      "epoch": 1.2072557677236844,
-      "grad_norm": 0.028371086344122887,
-      "learning_rate": 0.00019999928167958495,
-      "loss": 46.0,
-      "step": 15790
-    },
-    {
-      "epoch": 1.2073322247070741,
-      "grad_norm": 0.0012720482191070914,
-      "learning_rate": 0.00019999928158854032,
-      "loss": 46.0,
-      "step": 15791
-    },
-    {
-      "epoch": 1.207408681690464,
-      "grad_norm": 0.0008237158763222396,
-      "learning_rate": 0.0001999992814974899,
-      "loss": 46.0,
-      "step": 15792
-    },
-    {
-      "epoch": 1.2074851386738537,
-      "grad_norm": 0.0007866528467275202,
-      "learning_rate": 0.00019999928140643366,
-      "loss": 46.0,
-      "step": 15793
-    },
-    {
-      "epoch": 1.2075615956572434,
-      "grad_norm": 0.004056129604578018,
-      "learning_rate": 0.0001999992813153717,
-      "loss": 46.0,
-      "step": 15794
-    },
-    {
-      "epoch": 1.207638052640633,
-      "grad_norm": 0.0009275604388676584,
-      "learning_rate": 0.00019999928122430398,
-      "loss": 46.0,
-      "step": 15795
-    },
-    {
-      "epoch": 1.2077145096240227,
-      "grad_norm": 0.0013814476551488042,
-      "learning_rate": 0.00019999928113323049,
-      "loss": 46.0,
-      "step": 15796
-    },
-    {
-      "epoch": 1.2077909666074125,
-      "grad_norm": 0.001514368923380971,
-      "learning_rate": 0.00019999928104215121,
-      "loss": 46.0,
-      "step": 15797
-    },
-    {
-      "epoch": 1.2078674235908022,
-      "grad_norm": 0.0010194851784035563,
-      "learning_rate": 0.0001999992809510662,
-      "loss": 46.0,
-      "step": 15798
-    },
-    {
-      "epoch": 1.207943880574192,
-      "grad_norm": 0.000640751444734633,
-      "learning_rate": 0.00019999928085997538,
-      "loss": 46.0,
-      "step": 15799
-    },
-    {
-      "epoch": 1.2080203375575818,
-      "grad_norm": 0.0011753039434552193,
-      "learning_rate": 0.0001999992807688788,
-      "loss": 46.0,
-      "step": 15800
-    },
-    {
-      "epoch": 1.2080967945409713,
-      "grad_norm": 0.003666070057079196,
-      "learning_rate": 0.00019999928067777645,
-      "loss": 46.0,
-      "step": 15801
-    },
-    {
-      "epoch": 1.208173251524361,
-      "grad_norm": 0.0012108872178941965,
-      "learning_rate": 0.00019999928058666831,
-      "loss": 46.0,
-      "step": 15802
-    },
-    {
-      "epoch": 1.2082497085077508,
-      "grad_norm": 0.004139328841120005,
-      "learning_rate": 0.00019999928049555443,
-      "loss": 46.0,
-      "step": 15803
-    },
-    {
-      "epoch": 1.2083261654911406,
-      "grad_norm": 0.00129911326803267,
-      "learning_rate": 0.00019999928040443477,
-      "loss": 46.0,
-      "step": 15804
-    },
-    {
-      "epoch": 1.2084026224745303,
-      "grad_norm": 0.0005257275770418346,
-      "learning_rate": 0.00019999928031330937,
-      "loss": 46.0,
-      "step": 15805
-    },
-    {
-      "epoch": 1.2084790794579199,
-      "grad_norm": 0.0003099743917118758,
-      "learning_rate": 0.00019999928022217816,
-      "loss": 46.0,
-      "step": 15806
-    },
-    {
-      "epoch": 1.2085555364413096,
-      "grad_norm": 0.0007760750595480204,
-      "learning_rate": 0.00019999928013104121,
-      "loss": 46.0,
-      "step": 15807
-    },
-    {
-      "epoch": 1.2086319934246994,
-      "grad_norm": 0.003667765064164996,
-      "learning_rate": 0.00019999928003989846,
-      "loss": 46.0,
-      "step": 15808
-    },
-    {
-      "epoch": 1.2087084504080892,
-      "grad_norm": 0.0008092631469480693,
-      "learning_rate": 0.00019999927994874997,
-      "loss": 46.0,
-      "step": 15809
-    },
-    {
-      "epoch": 1.208784907391479,
-      "grad_norm": 0.002482498763129115,
-      "learning_rate": 0.0001999992798575957,
-      "loss": 46.0,
-      "step": 15810
-    },
-    {
-      "epoch": 1.2088613643748687,
-      "grad_norm": 0.00486728036776185,
-      "learning_rate": 0.00019999927976643565,
-      "loss": 46.0,
-      "step": 15811
-    },
-    {
-      "epoch": 1.2089378213582582,
-      "grad_norm": 0.003524020779877901,
-      "learning_rate": 0.00019999927967526983,
-      "loss": 46.0,
-      "step": 15812
-    },
-    {
-      "epoch": 1.209014278341648,
-      "grad_norm": 0.005361601244658232,
-      "learning_rate": 0.00019999927958409827,
-      "loss": 46.0,
-      "step": 15813
-    },
-    {
-      "epoch": 1.2090907353250377,
-      "grad_norm": 0.001418828614987433,
-      "learning_rate": 0.0001999992794929209,
-      "loss": 46.0,
-      "step": 15814
-    },
-    {
-      "epoch": 1.2091671923084275,
-      "grad_norm": 0.006174227222800255,
-      "learning_rate": 0.0001999992794017378,
-      "loss": 46.0,
-      "step": 15815
-    },
-    {
-      "epoch": 1.2092436492918173,
-      "grad_norm": 0.002934941789135337,
-      "learning_rate": 0.0001999992793105489,
-      "loss": 46.0,
-      "step": 15816
-    },
-    {
-      "epoch": 1.2093201062752068,
-      "grad_norm": 0.0010737967677414417,
-      "learning_rate": 0.00019999927921935423,
-      "loss": 46.0,
-      "step": 15817
-    },
-    {
-      "epoch": 1.2093965632585966,
-      "grad_norm": 0.022198457270860672,
-      "learning_rate": 0.00019999927912815383,
-      "loss": 46.0,
-      "step": 15818
-    },
-    {
-      "epoch": 1.2094730202419863,
-      "grad_norm": 0.024242263287305832,
-      "learning_rate": 0.00019999927903694762,
-      "loss": 46.0,
-      "step": 15819
-    },
-    {
-      "epoch": 1.209549477225376,
-      "grad_norm": 0.0035455338656902313,
-      "learning_rate": 0.00019999927894573567,
-      "loss": 46.0,
-      "step": 15820
-    },
-    {
-      "epoch": 1.2096259342087659,
-      "grad_norm": 0.0033642714843153954,
-      "learning_rate": 0.00019999927885451792,
-      "loss": 46.0,
-      "step": 15821
-    },
-    {
-      "epoch": 1.2097023911921556,
-      "grad_norm": 0.0016957588959485292,
-      "learning_rate": 0.0001999992787632944,
-      "loss": 46.0,
-      "step": 15822
-    },
-    {
-      "epoch": 1.2097788481755452,
-      "grad_norm": 0.0007913652225397527,
-      "learning_rate": 0.00019999927867206513,
-      "loss": 46.0,
-      "step": 15823
-    },
-    {
-      "epoch": 1.209855305158935,
-      "grad_norm": 0.0010948361596092582,
-      "learning_rate": 0.00019999927858083008,
-      "loss": 46.0,
-      "step": 15824
-    },
-    {
-      "epoch": 1.2099317621423247,
-      "grad_norm": 0.002287833718582988,
-      "learning_rate": 0.0001999992784895893,
-      "loss": 46.0,
-      "step": 15825
-    },
-    {
-      "epoch": 1.2100082191257144,
-      "grad_norm": 0.0016990774311125278,
-      "learning_rate": 0.0001999992783983427,
-      "loss": 46.0,
-      "step": 15826
-    },
-    {
-      "epoch": 1.2100846761091042,
-      "grad_norm": 0.0011581816943362355,
-      "learning_rate": 0.00019999927830709037,
-      "loss": 46.0,
-      "step": 15827
-    },
-    {
-      "epoch": 1.2101611330924937,
-      "grad_norm": 0.0010288050398230553,
-      "learning_rate": 0.00019999927821583223,
-      "loss": 46.0,
-      "step": 15828
-    },
-    {
-      "epoch": 1.2102375900758835,
-      "grad_norm": 0.0007644432480446994,
-      "learning_rate": 0.00019999927812456835,
-      "loss": 46.0,
-      "step": 15829
-    },
-    {
-      "epoch": 1.2103140470592733,
-      "grad_norm": 0.0010509173152968287,
-      "learning_rate": 0.0001999992780332987,
-      "loss": 46.0,
-      "step": 15830
-    },
-    {
-      "epoch": 1.210390504042663,
-      "grad_norm": 0.0006509299273602664,
-      "learning_rate": 0.00019999927794202323,
-      "loss": 46.0,
-      "step": 15831
-    },
-    {
-      "epoch": 1.2104669610260528,
-      "grad_norm": 0.0009767917217686772,
-      "learning_rate": 0.00019999927785074203,
-      "loss": 46.0,
-      "step": 15832
-    },
-    {
-      "epoch": 1.2105434180094425,
-      "grad_norm": 0.003716397564858198,
-      "learning_rate": 0.00019999927775945508,
-      "loss": 46.0,
-      "step": 15833
-    },
-    {
-      "epoch": 1.210619874992832,
-      "grad_norm": 0.0033332237508147955,
-      "learning_rate": 0.00019999927766816233,
-      "loss": 46.0,
-      "step": 15834
-    },
-    {
-      "epoch": 1.2106963319762218,
-      "grad_norm": 0.0008651018142700195,
-      "learning_rate": 0.00019999927757686383,
-      "loss": 46.0,
-      "step": 15835
-    },
-    {
-      "epoch": 1.2107727889596116,
-      "grad_norm": 0.002342688851058483,
-      "learning_rate": 0.00019999927748555956,
-      "loss": 46.0,
-      "step": 15836
-    },
-    {
-      "epoch": 1.2108492459430014,
-      "grad_norm": 0.001168787362985313,
-      "learning_rate": 0.00019999927739424952,
-      "loss": 46.0,
-      "step": 15837
-    },
-    {
-      "epoch": 1.2109257029263911,
-      "grad_norm": 0.003096845233812928,
-      "learning_rate": 0.00019999927730293367,
-      "loss": 46.0,
-      "step": 15838
-    },
-    {
-      "epoch": 1.2110021599097807,
-      "grad_norm": 0.004865511320531368,
-      "learning_rate": 0.0001999992772116121,
-      "loss": 46.0,
-      "step": 15839
-    },
-    {
-      "epoch": 1.2110786168931704,
-      "grad_norm": 0.0017508199671283364,
-      "learning_rate": 0.00019999927712028478,
-      "loss": 46.0,
-      "step": 15840
-    },
-    {
-      "epoch": 1.2111550738765602,
-      "grad_norm": 0.001202625804580748,
-      "learning_rate": 0.00019999927702895164,
-      "loss": 46.0,
-      "step": 15841
-    },
-    {
-      "epoch": 1.21123153085995,
-      "grad_norm": 0.0012115223798900843,
-      "learning_rate": 0.00019999927693761273,
-      "loss": 46.0,
-      "step": 15842
-    },
-    {
-      "epoch": 1.2113079878433397,
-      "grad_norm": 0.0060264067724347115,
-      "learning_rate": 0.00019999927684626807,
-      "loss": 46.0,
-      "step": 15843
-    },
-    {
-      "epoch": 1.2113844448267295,
-      "grad_norm": 0.0016420769970864058,
-      "learning_rate": 0.00019999927675491764,
-      "loss": 46.0,
-      "step": 15844
-    },
-    {
-      "epoch": 1.211460901810119,
-      "grad_norm": 0.0007236411329358816,
-      "learning_rate": 0.00019999927666356144,
-      "loss": 46.0,
-      "step": 15845
-    },
-    {
-      "epoch": 1.2115373587935088,
-      "grad_norm": 0.004076045472174883,
-      "learning_rate": 0.00019999927657219946,
-      "loss": 46.0,
-      "step": 15846
-    },
-    {
-      "epoch": 1.2116138157768985,
-      "grad_norm": 0.0010735122486948967,
-      "learning_rate": 0.0001999992764808317,
-      "loss": 46.0,
-      "step": 15847
-    },
-    {
-      "epoch": 1.2116902727602883,
-      "grad_norm": 0.0016806970816105604,
-      "learning_rate": 0.00019999927638945822,
-      "loss": 46.0,
-      "step": 15848
-    },
-    {
-      "epoch": 1.211766729743678,
-      "grad_norm": 0.0011726424563676119,
-      "learning_rate": 0.00019999927629807895,
-      "loss": 46.0,
-      "step": 15849
-    },
-    {
-      "epoch": 1.2118431867270676,
-      "grad_norm": 0.0009754498605616391,
-      "learning_rate": 0.00019999927620669388,
-      "loss": 46.0,
-      "step": 15850
-    },
-    {
-      "epoch": 1.2119196437104574,
-      "grad_norm": 0.0008477416122332215,
-      "learning_rate": 0.00019999927611530306,
-      "loss": 46.0,
-      "step": 15851
-    },
-    {
-      "epoch": 1.2119961006938471,
-      "grad_norm": 0.00562383234500885,
-      "learning_rate": 0.0001999992760239065,
-      "loss": 46.0,
-      "step": 15852
-    },
-    {
-      "epoch": 1.2120725576772369,
-      "grad_norm": 0.0027025260496884584,
-      "learning_rate": 0.00019999927593250413,
-      "loss": 46.0,
-      "step": 15853
-    },
-    {
-      "epoch": 1.2121490146606266,
-      "grad_norm": 0.0019046923844143748,
-      "learning_rate": 0.00019999927584109603,
-      "loss": 46.0,
-      "step": 15854
-    },
-    {
-      "epoch": 1.2122254716440164,
-      "grad_norm": 0.024615593254566193,
-      "learning_rate": 0.00019999927574968212,
-      "loss": 46.0,
-      "step": 15855
-    },
-    {
-      "epoch": 1.212301928627406,
-      "grad_norm": 0.0008551036007702351,
-      "learning_rate": 0.00019999927565826243,
-      "loss": 46.0,
-      "step": 15856
-    },
-    {
-      "epoch": 1.2123783856107957,
-      "grad_norm": 0.0010178949451074004,
-      "learning_rate": 0.000199999275566837,
-      "loss": 46.0,
-      "step": 15857
-    },
-    {
-      "epoch": 1.2124548425941855,
-      "grad_norm": 0.0010859669419005513,
-      "learning_rate": 0.0001999992754754058,
-      "loss": 46.0,
-      "step": 15858
-    },
-    {
-      "epoch": 1.2125312995775752,
-      "grad_norm": 0.0024663677904754877,
-      "learning_rate": 0.00019999927538396882,
-      "loss": 46.0,
-      "step": 15859
-    },
-    {
-      "epoch": 1.212607756560965,
-      "grad_norm": 0.0017164589371532202,
-      "learning_rate": 0.00019999927529252607,
-      "loss": 46.0,
-      "step": 15860
-    },
-    {
-      "epoch": 1.2126842135443545,
-      "grad_norm": 0.0015444419113919139,
-      "learning_rate": 0.00019999927520107758,
-      "loss": 46.0,
-      "step": 15861
-    },
-    {
-      "epoch": 1.2127606705277443,
-      "grad_norm": 0.000503760005813092,
-      "learning_rate": 0.00019999927510962328,
-      "loss": 46.0,
-      "step": 15862
-    },
-    {
-      "epoch": 1.212837127511134,
-      "grad_norm": 0.0008513595093972981,
-      "learning_rate": 0.0001999992750181632,
-      "loss": 46.0,
-      "step": 15863
-    },
-    {
-      "epoch": 1.2129135844945238,
-      "grad_norm": 0.0015667357947677374,
-      "learning_rate": 0.00019999927492669742,
-      "loss": 46.0,
-      "step": 15864
-    },
-    {
-      "epoch": 1.2129900414779136,
-      "grad_norm": 0.0013385817874222994,
-      "learning_rate": 0.00019999927483522583,
-      "loss": 46.0,
-      "step": 15865
-    },
-    {
-      "epoch": 1.2130664984613033,
-      "grad_norm": 0.0030473442748188972,
-      "learning_rate": 0.00019999927474374847,
-      "loss": 46.0,
-      "step": 15866
-    },
-    {
-      "epoch": 1.2131429554446929,
-      "grad_norm": 0.012189347296953201,
-      "learning_rate": 0.00019999927465226533,
-      "loss": 46.0,
-      "step": 15867
-    },
-    {
-      "epoch": 1.2132194124280826,
-      "grad_norm": 0.005485172383487225,
-      "learning_rate": 0.00019999927456077642,
-      "loss": 46.0,
-      "step": 15868
-    },
-    {
-      "epoch": 1.2132958694114724,
-      "grad_norm": 0.0005193467368371785,
-      "learning_rate": 0.00019999927446928177,
-      "loss": 46.0,
-      "step": 15869
-    },
-    {
-      "epoch": 1.2133723263948621,
-      "grad_norm": 0.00326589890755713,
-      "learning_rate": 0.00019999927437778134,
-      "loss": 46.0,
-      "step": 15870
-    },
-    {
-      "epoch": 1.213448783378252,
-      "grad_norm": 0.0008399309008382261,
-      "learning_rate": 0.0001999992742862751,
-      "loss": 46.0,
-      "step": 15871
-    },
-    {
-      "epoch": 1.2135252403616414,
-      "grad_norm": 0.0007985203992575407,
-      "learning_rate": 0.00019999927419476316,
-      "loss": 46.0,
-      "step": 15872
-    },
-    {
-      "epoch": 1.2136016973450312,
-      "grad_norm": 0.0007578706135973334,
-      "learning_rate": 0.00019999927410324538,
-      "loss": 46.0,
-      "step": 15873
-    },
-    {
-      "epoch": 1.213678154328421,
-      "grad_norm": 0.00224702013656497,
-      "learning_rate": 0.0001999992740117219,
-      "loss": 46.0,
-      "step": 15874
-    },
-    {
-      "epoch": 1.2137546113118107,
-      "grad_norm": 0.0006704268744215369,
-      "learning_rate": 0.00019999927392019262,
-      "loss": 46.0,
-      "step": 15875
-    },
-    {
-      "epoch": 1.2138310682952005,
-      "grad_norm": 0.0015335469506680965,
-      "learning_rate": 0.00019999927382865755,
-      "loss": 46.0,
-      "step": 15876
-    },
-    {
-      "epoch": 1.2139075252785902,
-      "grad_norm": 0.0004962708917446434,
-      "learning_rate": 0.00019999927373711674,
-      "loss": 46.0,
-      "step": 15877
-    },
-    {
-      "epoch": 1.2139839822619798,
-      "grad_norm": 0.0012502699391916394,
-      "learning_rate": 0.00019999927364557015,
-      "loss": 46.0,
-      "step": 15878
-    },
-    {
-      "epoch": 1.2140604392453695,
-      "grad_norm": 0.0037831610534340143,
-      "learning_rate": 0.00019999927355401778,
-      "loss": 46.0,
-      "step": 15879
-    },
-    {
-      "epoch": 1.2141368962287593,
-      "grad_norm": 0.0030314449686557055,
-      "learning_rate": 0.00019999927346245965,
-      "loss": 46.0,
-      "step": 15880
-    },
-    {
-      "epoch": 1.214213353212149,
-      "grad_norm": 0.001109353732317686,
-      "learning_rate": 0.00019999927337089574,
-      "loss": 46.0,
-      "step": 15881
-    },
-    {
-      "epoch": 1.2142898101955386,
-      "grad_norm": 0.0020769829861819744,
-      "learning_rate": 0.00019999927327932608,
-      "loss": 46.0,
-      "step": 15882
-    },
-    {
-      "epoch": 1.2143662671789284,
-      "grad_norm": 0.0017611763905733824,
-      "learning_rate": 0.00019999927318775065,
-      "loss": 46.0,
-      "step": 15883
-    },
-    {
-      "epoch": 1.2144427241623181,
-      "grad_norm": 0.0014140050625428557,
-      "learning_rate": 0.00019999927309616942,
-      "loss": 46.0,
-      "step": 15884
-    },
-    {
-      "epoch": 1.214519181145708,
-      "grad_norm": 0.0008226445643231273,
-      "learning_rate": 0.00019999927300458245,
-      "loss": 46.0,
-      "step": 15885
-    },
-    {
-      "epoch": 1.2145956381290977,
-      "grad_norm": 0.0018748317379504442,
-      "learning_rate": 0.0001999992729129897,
-      "loss": 46.0,
-      "step": 15886
-    },
-    {
-      "epoch": 1.2146720951124874,
-      "grad_norm": 0.0012664722744375467,
-      "learning_rate": 0.00019999927282139118,
-      "loss": 46.0,
-      "step": 15887
-    },
-    {
-      "epoch": 1.2147485520958772,
-      "grad_norm": 0.0008940821280702949,
-      "learning_rate": 0.00019999927272978688,
-      "loss": 46.0,
-      "step": 15888
-    },
-    {
-      "epoch": 1.2148250090792667,
-      "grad_norm": 0.004960604943335056,
-      "learning_rate": 0.00019999927263817684,
-      "loss": 46.0,
-      "step": 15889
-    },
-    {
-      "epoch": 1.2149014660626565,
-      "grad_norm": 0.0009728364530019462,
-      "learning_rate": 0.00019999927254656102,
-      "loss": 46.0,
-      "step": 15890
-    },
-    {
-      "epoch": 1.2149779230460462,
-      "grad_norm": 0.004020273685455322,
-      "learning_rate": 0.0001999992724549394,
-      "loss": 46.0,
-      "step": 15891
-    },
-    {
-      "epoch": 1.215054380029436,
-      "grad_norm": 0.0008358367485925555,
-      "learning_rate": 0.00019999927236331207,
-      "loss": 46.0,
-      "step": 15892
-    },
-    {
-      "epoch": 1.2151308370128255,
-      "grad_norm": 0.0017120903357863426,
-      "learning_rate": 0.00019999927227167894,
-      "loss": 46.0,
-      "step": 15893
-    },
-    {
-      "epoch": 1.2152072939962153,
-      "grad_norm": 0.0012376890517771244,
-      "learning_rate": 0.00019999927218004003,
-      "loss": 46.0,
-      "step": 15894
-    },
-    {
-      "epoch": 1.215283750979605,
-      "grad_norm": 0.0014771816786378622,
-      "learning_rate": 0.00019999927208839535,
-      "loss": 46.0,
-      "step": 15895
-    },
-    {
-      "epoch": 1.2153602079629948,
-      "grad_norm": 0.0011830950388684869,
-      "learning_rate": 0.00019999927199674492,
-      "loss": 46.0,
-      "step": 15896
-    },
-    {
-      "epoch": 1.2154366649463846,
-      "grad_norm": 0.002167725469917059,
-      "learning_rate": 0.00019999927190508872,
-      "loss": 46.0,
-      "step": 15897
-    },
-    {
-      "epoch": 1.2155131219297743,
-      "grad_norm": 0.00478803226724267,
-      "learning_rate": 0.00019999927181342671,
-      "loss": 46.0,
-      "step": 15898
-    },
-    {
-      "epoch": 1.215589578913164,
-      "grad_norm": 0.0007764878682792187,
-      "learning_rate": 0.00019999927172175897,
-      "loss": 46.0,
-      "step": 15899
-    },
-    {
-      "epoch": 1.2156660358965536,
-      "grad_norm": 0.0007188431336544454,
-      "learning_rate": 0.00019999927163008544,
-      "loss": 46.0,
-      "step": 15900
-    },
-    {
-      "epoch": 1.2157424928799434,
-      "grad_norm": 0.0004835437284782529,
-      "learning_rate": 0.00019999927153840617,
-      "loss": 46.0,
-      "step": 15901
-    },
-    {
-      "epoch": 1.2158189498633332,
-      "grad_norm": 0.0016761115984991193,
-      "learning_rate": 0.0001999992714467211,
-      "loss": 46.0,
-      "step": 15902
-    },
-    {
-      "epoch": 1.215895406846723,
-      "grad_norm": 0.004625473637133837,
-      "learning_rate": 0.00019999927135503026,
-      "loss": 46.0,
-      "step": 15903
-    },
-    {
-      "epoch": 1.2159718638301125,
-      "grad_norm": 0.0007580881356261671,
-      "learning_rate": 0.00019999927126333368,
-      "loss": 46.0,
-      "step": 15904
-    },
-    {
-      "epoch": 1.2160483208135022,
-      "grad_norm": 0.0003252858587075025,
-      "learning_rate": 0.0001999992711716313,
-      "loss": 46.0,
-      "step": 15905
-    },
-    {
-      "epoch": 1.216124777796892,
-      "grad_norm": 0.0025529228150844574,
-      "learning_rate": 0.00019999927107992318,
-      "loss": 46.0,
-      "step": 15906
-    },
-    {
-      "epoch": 1.2162012347802817,
-      "grad_norm": 0.008447384461760521,
-      "learning_rate": 0.00019999927098820927,
-      "loss": 46.0,
-      "step": 15907
-    },
-    {
-      "epoch": 1.2162776917636715,
-      "grad_norm": 0.0009964253986254334,
-      "learning_rate": 0.0001999992708964896,
-      "loss": 46.0,
-      "step": 15908
-    },
-    {
-      "epoch": 1.2163541487470613,
-      "grad_norm": 0.001493568066507578,
-      "learning_rate": 0.00019999927080476416,
-      "loss": 46.0,
-      "step": 15909
-    },
-    {
-      "epoch": 1.2164306057304508,
-      "grad_norm": 0.0016631204634904861,
-      "learning_rate": 0.00019999927071303293,
-      "loss": 46.0,
-      "step": 15910
-    },
-    {
-      "epoch": 1.2165070627138406,
-      "grad_norm": 0.0005618016002699733,
-      "learning_rate": 0.00019999927062129596,
-      "loss": 46.0,
-      "step": 15911
-    },
-    {
-      "epoch": 1.2165835196972303,
-      "grad_norm": 0.0015689197462052107,
-      "learning_rate": 0.0001999992705295532,
-      "loss": 46.0,
-      "step": 15912
-    },
-    {
-      "epoch": 1.21665997668062,
-      "grad_norm": 0.0033365788403898478,
-      "learning_rate": 0.00019999927043780469,
-      "loss": 46.0,
-      "step": 15913
-    },
-    {
-      "epoch": 1.2167364336640099,
-      "grad_norm": 0.0007817473378963768,
-      "learning_rate": 0.0001999992703460504,
-      "loss": 46.0,
-      "step": 15914
-    },
-    {
-      "epoch": 1.2168128906473994,
-      "grad_norm": 0.005644321441650391,
-      "learning_rate": 0.00019999927025429032,
-      "loss": 46.0,
-      "step": 15915
-    },
-    {
-      "epoch": 1.2168893476307892,
-      "grad_norm": 0.001306253601796925,
-      "learning_rate": 0.00019999927016252448,
-      "loss": 46.0,
-      "step": 15916
-    },
-    {
-      "epoch": 1.216965804614179,
-      "grad_norm": 0.0012445839820429683,
-      "learning_rate": 0.0001999992700707529,
-      "loss": 46.0,
-      "step": 15917
-    },
-    {
-      "epoch": 1.2170422615975687,
-      "grad_norm": 0.0006187583203427494,
-      "learning_rate": 0.00019999926997897553,
-      "loss": 46.0,
-      "step": 15918
-    },
-    {
-      "epoch": 1.2171187185809584,
-      "grad_norm": 0.0010500020580366254,
-      "learning_rate": 0.0001999992698871924,
-      "loss": 46.0,
-      "step": 15919
-    },
-    {
-      "epoch": 1.2171951755643482,
-      "grad_norm": 0.0005063980934210122,
-      "learning_rate": 0.0001999992697954035,
-      "loss": 46.0,
-      "step": 15920
-    },
-    {
-      "epoch": 1.2172716325477377,
-      "grad_norm": 0.00218991213478148,
-      "learning_rate": 0.0001999992697036088,
-      "loss": 46.0,
-      "step": 15921
-    },
-    {
-      "epoch": 1.2173480895311275,
-      "grad_norm": 0.0005154482205398381,
-      "learning_rate": 0.00019999926961180835,
-      "loss": 46.0,
-      "step": 15922
-    },
-    {
-      "epoch": 1.2174245465145173,
-      "grad_norm": 0.0011830269359052181,
-      "learning_rate": 0.00019999926952000215,
-      "loss": 46.0,
-      "step": 15923
-    },
-    {
-      "epoch": 1.217501003497907,
-      "grad_norm": 0.0032510936725884676,
-      "learning_rate": 0.00019999926942819017,
-      "loss": 46.0,
-      "step": 15924
-    },
-    {
-      "epoch": 1.2175774604812968,
-      "grad_norm": 0.002212245250120759,
-      "learning_rate": 0.0001999992693363724,
-      "loss": 46.0,
-      "step": 15925
-    },
-    {
-      "epoch": 1.2176539174646863,
-      "grad_norm": 0.0006792893982492387,
-      "learning_rate": 0.00019999926924454888,
-      "loss": 46.0,
-      "step": 15926
-    },
-    {
-      "epoch": 1.217730374448076,
-      "grad_norm": 0.00031252449844032526,
-      "learning_rate": 0.00019999926915271958,
-      "loss": 46.0,
-      "step": 15927
-    },
-    {
-      "epoch": 1.2178068314314658,
-      "grad_norm": 0.0009779686806723475,
-      "learning_rate": 0.00019999926906088451,
-      "loss": 46.0,
-      "step": 15928
-    },
-    {
-      "epoch": 1.2178832884148556,
-      "grad_norm": 0.005132970865815878,
-      "learning_rate": 0.0001999992689690437,
-      "loss": 46.0,
-      "step": 15929
-    },
-    {
-      "epoch": 1.2179597453982454,
-      "grad_norm": 0.000386443076422438,
-      "learning_rate": 0.00019999926887719709,
-      "loss": 46.0,
-      "step": 15930
-    },
-    {
-      "epoch": 1.2180362023816351,
-      "grad_norm": 0.007529250346124172,
-      "learning_rate": 0.00019999926878534472,
-      "loss": 46.0,
-      "step": 15931
-    },
-    {
-      "epoch": 1.2181126593650247,
-      "grad_norm": 0.0010429989779368043,
-      "learning_rate": 0.00019999926869348656,
-      "loss": 46.0,
-      "step": 15932
-    },
-    {
-      "epoch": 1.2181891163484144,
-      "grad_norm": 0.0007154787890613079,
-      "learning_rate": 0.00019999926860162268,
-      "loss": 46.0,
-      "step": 15933
-    },
-    {
-      "epoch": 1.2182655733318042,
-      "grad_norm": 0.0008657712023705244,
-      "learning_rate": 0.000199999268509753,
-      "loss": 46.0,
-      "step": 15934
-    },
-    {
-      "epoch": 1.218342030315194,
-      "grad_norm": 0.0021054905373603106,
-      "learning_rate": 0.00019999926841787752,
-      "loss": 46.0,
-      "step": 15935
-    },
-    {
-      "epoch": 1.2184184872985837,
-      "grad_norm": 0.0012357636587694287,
-      "learning_rate": 0.00019999926832599632,
-      "loss": 46.0,
-      "step": 15936
-    },
-    {
-      "epoch": 1.2184949442819732,
-      "grad_norm": 0.007126253563910723,
-      "learning_rate": 0.00019999926823410932,
-      "loss": 46.0,
-      "step": 15937
-    },
-    {
-      "epoch": 1.218571401265363,
-      "grad_norm": 0.0004661525599658489,
-      "learning_rate": 0.00019999926814221657,
-      "loss": 46.0,
-      "step": 15938
-    },
-    {
-      "epoch": 1.2186478582487528,
-      "grad_norm": 0.0057141780853271484,
-      "learning_rate": 0.00019999926805031805,
-      "loss": 46.0,
-      "step": 15939
-    },
-    {
-      "epoch": 1.2187243152321425,
-      "grad_norm": 0.0018218666082248092,
-      "learning_rate": 0.00019999926795841378,
-      "loss": 46.0,
-      "step": 15940
-    },
-    {
-      "epoch": 1.2188007722155323,
-      "grad_norm": 0.0008860841626301408,
-      "learning_rate": 0.00019999926786650368,
-      "loss": 46.0,
-      "step": 15941
-    },
-    {
-      "epoch": 1.218877229198922,
-      "grad_norm": 0.0005197604186832905,
-      "learning_rate": 0.00019999926777458787,
-      "loss": 46.0,
-      "step": 15942
-    },
-    {
-      "epoch": 1.2189536861823116,
-      "grad_norm": 0.0021949796937406063,
-      "learning_rate": 0.00019999926768266626,
-      "loss": 46.0,
-      "step": 15943
-    },
-    {
-      "epoch": 1.2190301431657014,
-      "grad_norm": 0.005042894743382931,
-      "learning_rate": 0.0001999992675907389,
-      "loss": 46.0,
-      "step": 15944
-    },
-    {
-      "epoch": 1.2191066001490911,
-      "grad_norm": 0.0009545614011585712,
-      "learning_rate": 0.00019999926749880573,
-      "loss": 46.0,
-      "step": 15945
-    },
-    {
-      "epoch": 1.2191830571324809,
-      "grad_norm": 0.0034013711847364902,
-      "learning_rate": 0.00019999926740686683,
-      "loss": 46.0,
-      "step": 15946
-    },
-    {
-      "epoch": 1.2192595141158706,
-      "grad_norm": 0.0008844583644531667,
-      "learning_rate": 0.00019999926731492214,
-      "loss": 46.0,
-      "step": 15947
-    },
-    {
-      "epoch": 1.2193359710992602,
-      "grad_norm": 0.010152323171496391,
-      "learning_rate": 0.0001999992672229717,
-      "loss": 46.0,
-      "step": 15948
-    },
-    {
-      "epoch": 1.21941242808265,
-      "grad_norm": 0.0011236346326768398,
-      "learning_rate": 0.00019999926713101546,
-      "loss": 46.0,
-      "step": 15949
-    },
-    {
-      "epoch": 1.2194888850660397,
-      "grad_norm": 0.001233525574207306,
-      "learning_rate": 0.0001999992670390535,
-      "loss": 46.0,
-      "step": 15950
-    },
-    {
-      "epoch": 1.2195653420494295,
-      "grad_norm": 0.0005170265794731677,
-      "learning_rate": 0.00019999926694708574,
-      "loss": 46.0,
-      "step": 15951
-    },
-    {
-      "epoch": 1.2196417990328192,
-      "grad_norm": 0.0017057224176824093,
-      "learning_rate": 0.0001999992668551122,
-      "loss": 46.0,
-      "step": 15952
-    },
-    {
-      "epoch": 1.219718256016209,
-      "grad_norm": 0.001143014756962657,
-      "learning_rate": 0.0001999992667631329,
-      "loss": 46.0,
-      "step": 15953
-    },
-    {
-      "epoch": 1.2197947129995985,
-      "grad_norm": 0.0004931085277348757,
-      "learning_rate": 0.00019999926667114783,
-      "loss": 46.0,
-      "step": 15954
-    },
-    {
-      "epoch": 1.2198711699829883,
-      "grad_norm": 0.003175918245688081,
-      "learning_rate": 0.00019999926657915696,
-      "loss": 46.0,
-      "step": 15955
-    },
-    {
-      "epoch": 1.219947626966378,
-      "grad_norm": 0.0005682227783836424,
-      "learning_rate": 0.00019999926648716038,
-      "loss": 46.0,
-      "step": 15956
-    },
-    {
-      "epoch": 1.2200240839497678,
-      "grad_norm": 0.0011381481308490038,
-      "learning_rate": 0.00019999926639515802,
-      "loss": 46.0,
-      "step": 15957
-    },
-    {
-      "epoch": 1.2201005409331576,
-      "grad_norm": 0.0007836158038116992,
-      "learning_rate": 0.00019999926630314985,
-      "loss": 46.0,
-      "step": 15958
-    },
-    {
-      "epoch": 1.220176997916547,
-      "grad_norm": 0.006223226897418499,
-      "learning_rate": 0.00019999926621113595,
-      "loss": 46.0,
-      "step": 15959
-    },
-    {
-      "epoch": 1.2202534548999369,
-      "grad_norm": 0.0034408390056341887,
-      "learning_rate": 0.00019999926611911627,
-      "loss": 46.0,
-      "step": 15960
-    },
-    {
-      "epoch": 1.2203299118833266,
-      "grad_norm": 0.0010396104771643877,
-      "learning_rate": 0.0001999992660270908,
-      "loss": 46.0,
-      "step": 15961
-    },
-    {
-      "epoch": 1.2204063688667164,
-      "grad_norm": 0.0015676913317292929,
-      "learning_rate": 0.00019999926593505958,
-      "loss": 46.0,
-      "step": 15962
-    },
-    {
-      "epoch": 1.2204828258501061,
-      "grad_norm": 0.0023836365435272455,
-      "learning_rate": 0.0001999992658430226,
-      "loss": 46.0,
-      "step": 15963
-    },
-    {
-      "epoch": 1.220559282833496,
-      "grad_norm": 0.0006649569259025156,
-      "learning_rate": 0.00019999926575097984,
-      "loss": 46.0,
-      "step": 15964
-    },
-    {
-      "epoch": 1.2206357398168854,
-      "grad_norm": 0.0008707884699106216,
-      "learning_rate": 0.0001999992656589313,
-      "loss": 46.0,
-      "step": 15965
-    },
-    {
-      "epoch": 1.2207121968002752,
-      "grad_norm": 0.0004146408464293927,
-      "learning_rate": 0.000199999265566877,
-      "loss": 46.0,
-      "step": 15966
-    },
-    {
-      "epoch": 1.220788653783665,
-      "grad_norm": 0.0031799226999282837,
-      "learning_rate": 0.00019999926547481693,
-      "loss": 46.0,
-      "step": 15967
-    },
-    {
-      "epoch": 1.2208651107670547,
-      "grad_norm": 0.0059667713940143585,
-      "learning_rate": 0.0001999992653827511,
-      "loss": 46.0,
-      "step": 15968
-    },
-    {
-      "epoch": 1.2209415677504445,
-      "grad_norm": 0.0009155656443908811,
-      "learning_rate": 0.00019999926529067947,
-      "loss": 46.0,
-      "step": 15969
-    },
-    {
-      "epoch": 1.221018024733834,
-      "grad_norm": 0.004582240246236324,
-      "learning_rate": 0.0001999992651986021,
-      "loss": 46.0,
-      "step": 15970
-    },
-    {
-      "epoch": 1.2210944817172238,
-      "grad_norm": 0.004969425965100527,
-      "learning_rate": 0.00019999926510651895,
-      "loss": 46.0,
-      "step": 15971
-    },
-    {
-      "epoch": 1.2211709387006136,
-      "grad_norm": 0.0029141867998987436,
-      "learning_rate": 0.00019999926501443005,
-      "loss": 46.0,
-      "step": 15972
-    },
-    {
-      "epoch": 1.2212473956840033,
-      "grad_norm": 0.0014684441266581416,
-      "learning_rate": 0.00019999926492233536,
-      "loss": 46.0,
-      "step": 15973
-    },
-    {
-      "epoch": 1.221323852667393,
-      "grad_norm": 0.0008456144132651389,
-      "learning_rate": 0.0001999992648302349,
-      "loss": 46.0,
-      "step": 15974
-    },
-    {
-      "epoch": 1.2214003096507828,
-      "grad_norm": 0.02805967815220356,
-      "learning_rate": 0.00019999926473812866,
-      "loss": 46.0,
-      "step": 15975
-    },
-    {
-      "epoch": 1.2214767666341724,
-      "grad_norm": 0.001692122663371265,
-      "learning_rate": 0.00019999926464601668,
-      "loss": 46.0,
-      "step": 15976
-    },
-    {
-      "epoch": 1.2215532236175621,
-      "grad_norm": 0.004000423941761255,
-      "learning_rate": 0.00019999926455389894,
-      "loss": 46.0,
-      "step": 15977
-    },
-    {
-      "epoch": 1.221629680600952,
-      "grad_norm": 0.0032135627698153257,
-      "learning_rate": 0.00019999926446177536,
-      "loss": 46.0,
-      "step": 15978
-    },
-    {
-      "epoch": 1.2217061375843417,
-      "grad_norm": 0.0023271837271749973,
-      "learning_rate": 0.0001999992643696461,
-      "loss": 46.0,
-      "step": 15979
-    },
-    {
-      "epoch": 1.2217825945677314,
-      "grad_norm": 0.0017594281816855073,
-      "learning_rate": 0.000199999264277511,
-      "loss": 46.0,
-      "step": 15980
-    },
-    {
-      "epoch": 1.221859051551121,
-      "grad_norm": 0.002090575871989131,
-      "learning_rate": 0.00019999926418537016,
-      "loss": 46.0,
-      "step": 15981
-    },
-    {
-      "epoch": 1.2219355085345107,
-      "grad_norm": 0.0010836290894076228,
-      "learning_rate": 0.00019999926409322355,
-      "loss": 46.0,
-      "step": 15982
-    },
-    {
-      "epoch": 1.2220119655179005,
-      "grad_norm": 0.0003954892745241523,
-      "learning_rate": 0.0001999992640010712,
-      "loss": 46.0,
-      "step": 15983
-    },
-    {
-      "epoch": 1.2220884225012902,
-      "grad_norm": 0.0006336772930808365,
-      "learning_rate": 0.00019999926390891303,
-      "loss": 46.0,
-      "step": 15984
-    },
-    {
-      "epoch": 1.22216487948468,
-      "grad_norm": 0.0014205609913915396,
-      "learning_rate": 0.00019999926381674912,
-      "loss": 46.0,
-      "step": 15985
-    },
-    {
-      "epoch": 1.2222413364680698,
-      "grad_norm": 0.001966850133612752,
-      "learning_rate": 0.0001999992637245794,
-      "loss": 46.0,
-      "step": 15986
-    },
-    {
-      "epoch": 1.2223177934514593,
-      "grad_norm": 0.0007035090238787234,
-      "learning_rate": 0.00019999926363240396,
-      "loss": 46.0,
-      "step": 15987
-    },
-    {
-      "epoch": 1.222394250434849,
-      "grad_norm": 0.0005249667447060347,
-      "learning_rate": 0.0001999992635402227,
-      "loss": 46.0,
-      "step": 15988
-    },
-    {
-      "epoch": 1.2224707074182388,
-      "grad_norm": 0.0007026847451925278,
-      "learning_rate": 0.00019999926344803573,
-      "loss": 46.0,
-      "step": 15989
-    },
-    {
-      "epoch": 1.2225471644016286,
-      "grad_norm": 0.003269130364060402,
-      "learning_rate": 0.00019999926335584296,
-      "loss": 46.0,
-      "step": 15990
-    },
-    {
-      "epoch": 1.2226236213850183,
-      "grad_norm": 0.008514025248587132,
-      "learning_rate": 0.00019999926326364444,
-      "loss": 46.0,
-      "step": 15991
-    },
-    {
-      "epoch": 1.2227000783684079,
-      "grad_norm": 0.0012790159089490771,
-      "learning_rate": 0.00019999926317144012,
-      "loss": 46.0,
-      "step": 15992
-    },
-    {
-      "epoch": 1.2227765353517976,
-      "grad_norm": 0.0014809739077463746,
-      "learning_rate": 0.00019999926307923005,
-      "loss": 46.0,
-      "step": 15993
-    },
-    {
-      "epoch": 1.2228529923351874,
-      "grad_norm": 0.0012516764691099524,
-      "learning_rate": 0.0001999992629870142,
-      "loss": 46.0,
-      "step": 15994
-    },
-    {
-      "epoch": 1.2229294493185772,
-      "grad_norm": 0.003328055376186967,
-      "learning_rate": 0.0001999992628947926,
-      "loss": 46.0,
-      "step": 15995
-    },
-    {
-      "epoch": 1.223005906301967,
-      "grad_norm": 0.0009070641244761646,
-      "learning_rate": 0.0001999992628025652,
-      "loss": 46.0,
-      "step": 15996
-    },
-    {
-      "epoch": 1.2230823632853567,
-      "grad_norm": 0.0011334576411172748,
-      "learning_rate": 0.00019999926271033208,
-      "loss": 46.0,
-      "step": 15997
-    },
-    {
-      "epoch": 1.2231588202687462,
-      "grad_norm": 0.0008738017058931291,
-      "learning_rate": 0.00019999926261809315,
-      "loss": 46.0,
-      "step": 15998
-    },
-    {
-      "epoch": 1.223235277252136,
-      "grad_norm": 0.0007524918182753026,
-      "learning_rate": 0.00019999926252584844,
-      "loss": 46.0,
-      "step": 15999
-    },
-    {
-      "epoch": 1.2233117342355257,
-      "grad_norm": 0.0015665973769500852,
-      "learning_rate": 0.00019999926243359796,
-      "loss": 46.0,
-      "step": 16000
-    },
-    {
-      "epoch": 1.2233881912189155,
-      "grad_norm": 0.0012447729241102934,
-      "learning_rate": 0.00019999926234134176,
-      "loss": 46.0,
-      "step": 16001
-    },
-    {
-      "epoch": 1.2234646482023053,
-      "grad_norm": 0.0008318267646245658,
-      "learning_rate": 0.00019999926224907973,
-      "loss": 46.0,
-      "step": 16002
-    },
-    {
-      "epoch": 1.2235411051856948,
-      "grad_norm": 0.00046407562331296504,
-      "learning_rate": 0.00019999926215681199,
-      "loss": 46.0,
-      "step": 16003
-    },
-    {
-      "epoch": 1.2236175621690846,
-      "grad_norm": 0.00404116977006197,
-      "learning_rate": 0.00019999926206453844,
-      "loss": 46.0,
-      "step": 16004
-    },
-    {
-      "epoch": 1.2236940191524743,
-      "grad_norm": 0.0013177929213270545,
-      "learning_rate": 0.00019999926197225915,
-      "loss": 46.0,
-      "step": 16005
-    },
-    {
-      "epoch": 1.223770476135864,
-      "grad_norm": 0.0012153750285506248,
-      "learning_rate": 0.00019999926187997405,
-      "loss": 46.0,
-      "step": 16006
-    },
-    {
-      "epoch": 1.2238469331192539,
-      "grad_norm": 0.004689784254878759,
-      "learning_rate": 0.0001999992617876832,
-      "loss": 46.0,
-      "step": 16007
-    },
-    {
-      "epoch": 1.2239233901026436,
-      "grad_norm": 0.0010637850500643253,
-      "learning_rate": 0.0001999992616953866,
-      "loss": 46.0,
-      "step": 16008
-    },
-    {
-      "epoch": 1.2239998470860332,
-      "grad_norm": 0.0007532556192018092,
-      "learning_rate": 0.00019999926160308421,
-      "loss": 46.0,
-      "step": 16009
-    },
-    {
-      "epoch": 1.224076304069423,
-      "grad_norm": 0.001208620727993548,
-      "learning_rate": 0.00019999926151077605,
-      "loss": 46.0,
-      "step": 16010
-    },
-    {
-      "epoch": 1.2241527610528127,
-      "grad_norm": 0.0009078102302737534,
-      "learning_rate": 0.00019999926141846212,
-      "loss": 46.0,
-      "step": 16011
-    },
-    {
-      "epoch": 1.2242292180362024,
-      "grad_norm": 0.0014089127071201801,
-      "learning_rate": 0.00019999926132614244,
-      "loss": 46.0,
-      "step": 16012
-    },
-    {
-      "epoch": 1.2243056750195922,
-      "grad_norm": 0.0007508535054512322,
-      "learning_rate": 0.000199999261233817,
-      "loss": 46.0,
-      "step": 16013
-    },
-    {
-      "epoch": 1.2243821320029817,
-      "grad_norm": 0.0008192521636374295,
-      "learning_rate": 0.00019999926114148576,
-      "loss": 46.0,
-      "step": 16014
-    },
-    {
-      "epoch": 1.2244585889863715,
-      "grad_norm": 0.0005333675653673708,
-      "learning_rate": 0.00019999926104914874,
-      "loss": 46.0,
-      "step": 16015
-    },
-    {
-      "epoch": 1.2245350459697613,
-      "grad_norm": 0.0017077181255444884,
-      "learning_rate": 0.000199999260956806,
-      "loss": 46.0,
-      "step": 16016
-    },
-    {
-      "epoch": 1.224611502953151,
-      "grad_norm": 0.0009263677056878805,
-      "learning_rate": 0.00019999926086445742,
-      "loss": 46.0,
-      "step": 16017
-    },
-    {
-      "epoch": 1.2246879599365408,
-      "grad_norm": 0.0008773449226282537,
-      "learning_rate": 0.00019999926077210312,
-      "loss": 46.0,
-      "step": 16018
-    },
-    {
-      "epoch": 1.2247644169199305,
-      "grad_norm": 0.0018951385281980038,
-      "learning_rate": 0.00019999926067974303,
-      "loss": 46.0,
-      "step": 16019
-    },
-    {
-      "epoch": 1.22484087390332,
-      "grad_norm": 0.0007082127267494798,
-      "learning_rate": 0.0001999992605873772,
-      "loss": 46.0,
-      "step": 16020
-    },
-    {
-      "epoch": 1.2249173308867098,
-      "grad_norm": 0.0006573075661435723,
-      "learning_rate": 0.00019999926049500558,
-      "loss": 46.0,
-      "step": 16021
-    },
-    {
-      "epoch": 1.2249937878700996,
-      "grad_norm": 0.001450154697522521,
-      "learning_rate": 0.00019999926040262817,
-      "loss": 46.0,
-      "step": 16022
-    },
-    {
-      "epoch": 1.2250702448534894,
-      "grad_norm": 0.0005226099165156484,
-      "learning_rate": 0.00019999926031024503,
-      "loss": 46.0,
-      "step": 16023
-    },
-    {
-      "epoch": 1.225146701836879,
-      "grad_norm": 0.0005636019050143659,
-      "learning_rate": 0.0001999992602178561,
-      "loss": 46.0,
-      "step": 16024
-    },
-    {
-      "epoch": 1.2252231588202687,
-      "grad_norm": 0.0010586412390694022,
-      "learning_rate": 0.0001999992601254614,
-      "loss": 46.0,
-      "step": 16025
-    },
-    {
-      "epoch": 1.2252996158036584,
-      "grad_norm": 0.000290626980131492,
-      "learning_rate": 0.00019999926003306094,
-      "loss": 46.0,
-      "step": 16026
-    },
-    {
-      "epoch": 1.2253760727870482,
-      "grad_norm": 0.0012102006003260612,
-      "learning_rate": 0.0001999992599406547,
-      "loss": 46.0,
-      "step": 16027
-    },
-    {
-      "epoch": 1.225452529770438,
-      "grad_norm": 0.0013221726985648274,
-      "learning_rate": 0.00019999925984824272,
-      "loss": 46.0,
-      "step": 16028
-    },
-    {
-      "epoch": 1.2255289867538277,
-      "grad_norm": 0.0005828901776112616,
-      "learning_rate": 0.00019999925975582494,
-      "loss": 46.0,
-      "step": 16029
-    },
-    {
-      "epoch": 1.2256054437372175,
-      "grad_norm": 0.0010008827084675431,
-      "learning_rate": 0.00019999925966340137,
-      "loss": 46.0,
-      "step": 16030
-    },
-    {
-      "epoch": 1.225681900720607,
-      "grad_norm": 0.0033580453600734472,
-      "learning_rate": 0.00019999925957097208,
-      "loss": 46.0,
-      "step": 16031
-    },
-    {
-      "epoch": 1.2257583577039968,
-      "grad_norm": 0.002998590236529708,
-      "learning_rate": 0.00019999925947853699,
-      "loss": 46.0,
-      "step": 16032
-    },
-    {
-      "epoch": 1.2258348146873865,
-      "grad_norm": 0.022819610312581062,
-      "learning_rate": 0.00019999925938609615,
-      "loss": 46.0,
-      "step": 16033
-    },
-    {
-      "epoch": 1.2259112716707763,
-      "grad_norm": 0.002703476930037141,
-      "learning_rate": 0.00019999925929364953,
-      "loss": 46.0,
-      "step": 16034
-    },
-    {
-      "epoch": 1.2259877286541658,
-      "grad_norm": 0.006659628823399544,
-      "learning_rate": 0.00019999925920119712,
-      "loss": 46.0,
-      "step": 16035
-    },
-    {
-      "epoch": 1.2260641856375556,
-      "grad_norm": 0.0012546078069135547,
-      "learning_rate": 0.000199999259108739,
-      "loss": 46.0,
-      "step": 16036
-    },
-    {
-      "epoch": 1.2261406426209454,
-      "grad_norm": 0.0009955652058124542,
-      "learning_rate": 0.00019999925901627506,
-      "loss": 46.0,
-      "step": 16037
-    },
-    {
-      "epoch": 1.2262170996043351,
-      "grad_norm": 0.0034153307788074017,
-      "learning_rate": 0.00019999925892380532,
-      "loss": 46.0,
-      "step": 16038
-    },
-    {
-      "epoch": 1.2262935565877249,
-      "grad_norm": 0.002630155300721526,
-      "learning_rate": 0.00019999925883132987,
-      "loss": 46.0,
-      "step": 16039
-    },
-    {
-      "epoch": 1.2263700135711146,
-      "grad_norm": 0.0012720349477604032,
-      "learning_rate": 0.00019999925873884865,
-      "loss": 46.0,
-      "step": 16040
-    },
-    {
-      "epoch": 1.2264464705545042,
-      "grad_norm": 0.000962450576480478,
-      "learning_rate": 0.00019999925864636162,
-      "loss": 46.0,
-      "step": 16041
-    },
-    {
-      "epoch": 1.226522927537894,
-      "grad_norm": 0.0006845833268016577,
-      "learning_rate": 0.00019999925855386885,
-      "loss": 46.0,
-      "step": 16042
-    },
-    {
-      "epoch": 1.2265993845212837,
-      "grad_norm": 0.0006671029841527343,
-      "learning_rate": 0.0001999992584613703,
-      "loss": 46.0,
-      "step": 16043
-    },
-    {
-      "epoch": 1.2266758415046735,
-      "grad_norm": 0.001622217707335949,
-      "learning_rate": 0.00019999925836886598,
-      "loss": 46.0,
-      "step": 16044
-    },
-    {
-      "epoch": 1.2267522984880632,
-      "grad_norm": 0.0038421335630118847,
-      "learning_rate": 0.00019999925827635592,
-      "loss": 46.0,
-      "step": 16045
-    },
-    {
-      "epoch": 1.2268287554714528,
-      "grad_norm": 0.000969208253081888,
-      "learning_rate": 0.00019999925818384008,
-      "loss": 46.0,
-      "step": 16046
-    },
-    {
-      "epoch": 1.2269052124548425,
-      "grad_norm": 0.0009379900875501335,
-      "learning_rate": 0.00019999925809131847,
-      "loss": 46.0,
-      "step": 16047
-    },
-    {
-      "epoch": 1.2269816694382323,
-      "grad_norm": 0.002720875898376107,
-      "learning_rate": 0.00019999925799879105,
-      "loss": 46.0,
-      "step": 16048
-    },
-    {
-      "epoch": 1.227058126421622,
-      "grad_norm": 0.0004442320205271244,
-      "learning_rate": 0.0001999992579062579,
-      "loss": 46.0,
-      "step": 16049
-    },
-    {
-      "epoch": 1.2271345834050118,
-      "grad_norm": 0.0006628509145230055,
-      "learning_rate": 0.00019999925781371896,
-      "loss": 46.0,
-      "step": 16050
-    },
-    {
-      "epoch": 1.2272110403884016,
-      "grad_norm": 0.0015220519853755832,
-      "learning_rate": 0.00019999925772117426,
-      "loss": 46.0,
-      "step": 16051
-    },
-    {
-      "epoch": 1.227287497371791,
-      "grad_norm": 0.0014778684126213193,
-      "learning_rate": 0.00019999925762862378,
-      "loss": 46.0,
-      "step": 16052
-    },
-    {
-      "epoch": 1.2273639543551809,
-      "grad_norm": 0.000373422255506739,
-      "learning_rate": 0.00019999925753606755,
-      "loss": 46.0,
-      "step": 16053
-    },
-    {
-      "epoch": 1.2274404113385706,
-      "grad_norm": 0.0007266036118380725,
-      "learning_rate": 0.00019999925744350553,
-      "loss": 46.0,
-      "step": 16054
-    },
-    {
-      "epoch": 1.2275168683219604,
-      "grad_norm": 0.0012402933789417148,
-      "learning_rate": 0.00019999925735093775,
-      "loss": 46.0,
-      "step": 16055
-    },
-    {
-      "epoch": 1.2275933253053501,
-      "grad_norm": 0.0009391485946252942,
-      "learning_rate": 0.0001999992572583642,
-      "loss": 46.0,
-      "step": 16056
-    },
-    {
-      "epoch": 1.2276697822887397,
-      "grad_norm": 0.000821779016405344,
-      "learning_rate": 0.0001999992571657849,
-      "loss": 46.0,
-      "step": 16057
-    },
-    {
-      "epoch": 1.2277462392721294,
-      "grad_norm": 0.021309809759259224,
-      "learning_rate": 0.0001999992570731998,
-      "loss": 46.0,
-      "step": 16058
-    },
-    {
-      "epoch": 1.2278226962555192,
-      "grad_norm": 0.001841805293224752,
-      "learning_rate": 0.00019999925698060896,
-      "loss": 46.0,
-      "step": 16059
-    },
-    {
-      "epoch": 1.227899153238909,
-      "grad_norm": 0.0003897084097843617,
-      "learning_rate": 0.00019999925688801232,
-      "loss": 46.0,
-      "step": 16060
-    },
-    {
-      "epoch": 1.2279756102222987,
-      "grad_norm": 0.0020316436421126127,
-      "learning_rate": 0.0001999992567954099,
-      "loss": 46.0,
-      "step": 16061
-    },
-    {
-      "epoch": 1.2280520672056885,
-      "grad_norm": 0.0007745440234430134,
-      "learning_rate": 0.00019999925670280178,
-      "loss": 46.0,
-      "step": 16062
-    },
-    {
-      "epoch": 1.228128524189078,
-      "grad_norm": 0.0015158121241256595,
-      "learning_rate": 0.00019999925661018785,
-      "loss": 46.0,
-      "step": 16063
-    },
-    {
-      "epoch": 1.2282049811724678,
-      "grad_norm": 0.0005278295138850808,
-      "learning_rate": 0.0001999992565175681,
-      "loss": 46.0,
-      "step": 16064
-    },
-    {
-      "epoch": 1.2282814381558576,
-      "grad_norm": 0.0008000683737918735,
-      "learning_rate": 0.00019999925642494266,
-      "loss": 46.0,
-      "step": 16065
-    },
-    {
-      "epoch": 1.2283578951392473,
-      "grad_norm": 0.0019807324279099703,
-      "learning_rate": 0.00019999925633231144,
-      "loss": 46.0,
-      "step": 16066
-    },
-    {
-      "epoch": 1.228434352122637,
-      "grad_norm": 0.0012882484588772058,
-      "learning_rate": 0.0001999992562396744,
-      "loss": 46.0,
-      "step": 16067
-    },
-    {
-      "epoch": 1.2285108091060266,
-      "grad_norm": 0.0006577378953807056,
-      "learning_rate": 0.00019999925614703164,
-      "loss": 46.0,
-      "step": 16068
-    },
-    {
-      "epoch": 1.2285872660894164,
-      "grad_norm": 0.0014593867817893624,
-      "learning_rate": 0.00019999925605438307,
-      "loss": 46.0,
-      "step": 16069
-    },
-    {
-      "epoch": 1.2286637230728061,
-      "grad_norm": 0.002570970682427287,
-      "learning_rate": 0.00019999925596172875,
-      "loss": 46.0,
-      "step": 16070
-    },
-    {
-      "epoch": 1.228740180056196,
-      "grad_norm": 0.0006125358049757779,
-      "learning_rate": 0.00019999925586906866,
-      "loss": 46.0,
-      "step": 16071
-    },
-    {
-      "epoch": 1.2288166370395857,
-      "grad_norm": 0.0014641006710007787,
-      "learning_rate": 0.00019999925577640282,
-      "loss": 46.0,
-      "step": 16072
-    },
-    {
-      "epoch": 1.2288930940229754,
-      "grad_norm": 0.0004070083377882838,
-      "learning_rate": 0.00019999925568373118,
-      "loss": 46.0,
-      "step": 16073
-    },
-    {
-      "epoch": 1.228969551006365,
-      "grad_norm": 0.0017690513050183654,
-      "learning_rate": 0.0001999992555910538,
-      "loss": 46.0,
-      "step": 16074
-    },
-    {
-      "epoch": 1.2290460079897547,
-      "grad_norm": 0.0005106092430651188,
-      "learning_rate": 0.0001999992554983706,
-      "loss": 46.0,
-      "step": 16075
-    },
-    {
-      "epoch": 1.2291224649731445,
-      "grad_norm": 0.0007438038592226803,
-      "learning_rate": 0.00019999925540568168,
-      "loss": 46.0,
-      "step": 16076
-    },
-    {
-      "epoch": 1.2291989219565342,
-      "grad_norm": 0.0016169827431440353,
-      "learning_rate": 0.00019999925531298697,
-      "loss": 46.0,
-      "step": 16077
-    },
-    {
-      "epoch": 1.229275378939924,
-      "grad_norm": 0.003142925212159753,
-      "learning_rate": 0.0001999992552202865,
-      "loss": 46.0,
-      "step": 16078
-    },
-    {
-      "epoch": 1.2293518359233135,
-      "grad_norm": 0.005689569283276796,
-      "learning_rate": 0.00019999925512758027,
-      "loss": 46.0,
-      "step": 16079
-    },
-    {
-      "epoch": 1.2294282929067033,
-      "grad_norm": 0.0010265561286360025,
-      "learning_rate": 0.00019999925503486824,
-      "loss": 46.0,
-      "step": 16080
-    },
-    {
-      "epoch": 1.229504749890093,
-      "grad_norm": 0.0004349384398665279,
-      "learning_rate": 0.00019999925494215047,
-      "loss": 46.0,
-      "step": 16081
-    },
-    {
-      "epoch": 1.2295812068734828,
-      "grad_norm": 0.0007582813268527389,
-      "learning_rate": 0.0001999992548494269,
-      "loss": 46.0,
-      "step": 16082
-    },
-    {
-      "epoch": 1.2296576638568726,
-      "grad_norm": 0.010492806322872639,
-      "learning_rate": 0.00019999925475669758,
-      "loss": 46.0,
-      "step": 16083
-    },
-    {
-      "epoch": 1.2297341208402623,
-      "grad_norm": 0.0022323697339743376,
-      "learning_rate": 0.0001999992546639625,
-      "loss": 46.0,
-      "step": 16084
-    },
-    {
-      "epoch": 1.2298105778236519,
-      "grad_norm": 0.004740293137729168,
-      "learning_rate": 0.00019999925457122163,
-      "loss": 46.0,
-      "step": 16085
-    },
-    {
-      "epoch": 1.2298870348070416,
-      "grad_norm": 0.004808230325579643,
-      "learning_rate": 0.00019999925447847502,
-      "loss": 46.0,
-      "step": 16086
-    },
-    {
-      "epoch": 1.2299634917904314,
-      "grad_norm": 0.0005458660889416933,
-      "learning_rate": 0.00019999925438572263,
-      "loss": 46.0,
-      "step": 16087
-    },
-    {
-      "epoch": 1.2300399487738212,
-      "grad_norm": 0.0008397402707487345,
-      "learning_rate": 0.00019999925429296442,
-      "loss": 46.0,
-      "step": 16088
-    },
-    {
-      "epoch": 1.230116405757211,
-      "grad_norm": 0.000881789019331336,
-      "learning_rate": 0.0001999992542002005,
-      "loss": 46.0,
-      "step": 16089
-    },
-    {
-      "epoch": 1.2301928627406005,
-      "grad_norm": 0.0006640120409429073,
-      "learning_rate": 0.00019999925410743078,
-      "loss": 46.0,
-      "step": 16090
-    },
-    {
-      "epoch": 1.2302693197239902,
-      "grad_norm": 0.0007411792175844312,
-      "learning_rate": 0.0001999992540146553,
-      "loss": 46.0,
-      "step": 16091
-    },
-    {
-      "epoch": 1.23034577670738,
-      "grad_norm": 0.0011047892039641738,
-      "learning_rate": 0.00019999925392187405,
-      "loss": 46.0,
-      "step": 16092
-    },
-    {
-      "epoch": 1.2304222336907698,
-      "grad_norm": 0.00347514683380723,
-      "learning_rate": 0.00019999925382908703,
-      "loss": 46.0,
-      "step": 16093
-    },
-    {
-      "epoch": 1.2304986906741595,
-      "grad_norm": 0.0007427798700518906,
-      "learning_rate": 0.00019999925373629426,
-      "loss": 46.0,
-      "step": 16094
-    },
-    {
-      "epoch": 1.2305751476575493,
-      "grad_norm": 0.0010663170833140612,
-      "learning_rate": 0.0001999992536434957,
-      "loss": 46.0,
-      "step": 16095
-    },
-    {
-      "epoch": 1.2306516046409388,
-      "grad_norm": 0.0009400284616276622,
-      "learning_rate": 0.00019999925355069137,
-      "loss": 46.0,
-      "step": 16096
-    },
-    {
-      "epoch": 1.2307280616243286,
-      "grad_norm": 0.0030948964413255453,
-      "learning_rate": 0.0001999992534578813,
-      "loss": 46.0,
-      "step": 16097
-    },
-    {
-      "epoch": 1.2308045186077183,
-      "grad_norm": 0.0014139021513983607,
-      "learning_rate": 0.0001999992533650654,
-      "loss": 46.0,
-      "step": 16098
-    },
-    {
-      "epoch": 1.230880975591108,
-      "grad_norm": 0.0008110702619887888,
-      "learning_rate": 0.0001999992532722438,
-      "loss": 46.0,
-      "step": 16099
-    },
-    {
-      "epoch": 1.2309574325744979,
-      "grad_norm": 0.0005633847904391587,
-      "learning_rate": 0.00019999925317941642,
-      "loss": 46.0,
-      "step": 16100
-    },
-    {
-      "epoch": 1.2310338895578874,
-      "grad_norm": 0.0006365786539390683,
-      "learning_rate": 0.00019999925308658323,
-      "loss": 46.0,
-      "step": 16101
-    },
-    {
-      "epoch": 1.2311103465412772,
-      "grad_norm": 0.0013632565969601274,
-      "learning_rate": 0.0001999992529937443,
-      "loss": 46.0,
-      "step": 16102
-    },
-    {
-      "epoch": 1.231186803524667,
-      "grad_norm": 0.0016504349187016487,
-      "learning_rate": 0.00019999925290089957,
-      "loss": 46.0,
-      "step": 16103
-    },
-    {
-      "epoch": 1.2312632605080567,
-      "grad_norm": 0.0015550582902505994,
-      "learning_rate": 0.0001999992528080491,
-      "loss": 46.0,
-      "step": 16104
-    },
-    {
-      "epoch": 1.2313397174914464,
-      "grad_norm": 0.0005089414771646261,
-      "learning_rate": 0.00019999925271519287,
-      "loss": 46.0,
-      "step": 16105
-    },
-    {
-      "epoch": 1.2314161744748362,
-      "grad_norm": 0.0006099442252889276,
-      "learning_rate": 0.00019999925262233084,
-      "loss": 46.0,
-      "step": 16106
-    },
-    {
-      "epoch": 1.2314926314582257,
-      "grad_norm": 0.001864170772023499,
-      "learning_rate": 0.00019999925252946302,
-      "loss": 46.0,
-      "step": 16107
-    },
-    {
-      "epoch": 1.2315690884416155,
-      "grad_norm": 0.0010190855246037245,
-      "learning_rate": 0.0001999992524365895,
-      "loss": 46.0,
-      "step": 16108
-    },
-    {
-      "epoch": 1.2316455454250053,
-      "grad_norm": 0.0019764506723731756,
-      "learning_rate": 0.00019999925234371016,
-      "loss": 46.0,
-      "step": 16109
-    },
-    {
-      "epoch": 1.231722002408395,
-      "grad_norm": 0.004341428633779287,
-      "learning_rate": 0.00019999925225082507,
-      "loss": 46.0,
-      "step": 16110
-    },
-    {
-      "epoch": 1.2317984593917848,
-      "grad_norm": 0.003148481948301196,
-      "learning_rate": 0.0001999992521579342,
-      "loss": 46.0,
-      "step": 16111
-    },
-    {
-      "epoch": 1.2318749163751743,
-      "grad_norm": 0.003891546744853258,
-      "learning_rate": 0.00019999925206503756,
-      "loss": 46.0,
-      "step": 16112
-    },
-    {
-      "epoch": 1.231951373358564,
-      "grad_norm": 0.01181508507579565,
-      "learning_rate": 0.00019999925197213515,
-      "loss": 46.0,
-      "step": 16113
-    },
-    {
-      "epoch": 1.2320278303419538,
-      "grad_norm": 0.002067372901365161,
-      "learning_rate": 0.000199999251879227,
-      "loss": 46.0,
-      "step": 16114
-    },
-    {
-      "epoch": 1.2321042873253436,
-      "grad_norm": 0.0033666668459773064,
-      "learning_rate": 0.00019999925178631304,
-      "loss": 46.0,
-      "step": 16115
-    },
-    {
-      "epoch": 1.2321807443087334,
-      "grad_norm": 0.005725543480366468,
-      "learning_rate": 0.00019999925169339333,
-      "loss": 46.0,
-      "step": 16116
-    },
-    {
-      "epoch": 1.2322572012921231,
-      "grad_norm": 0.0016691512428224087,
-      "learning_rate": 0.00019999925160046783,
-      "loss": 46.0,
-      "step": 16117
-    },
-    {
-      "epoch": 1.2323336582755127,
-      "grad_norm": 0.001649119658395648,
-      "learning_rate": 0.0001999992515075366,
-      "loss": 46.0,
-      "step": 16118
-    },
-    {
-      "epoch": 1.2324101152589024,
-      "grad_norm": 0.006222618278115988,
-      "learning_rate": 0.00019999925141459958,
-      "loss": 46.0,
-      "step": 16119
-    },
-    {
-      "epoch": 1.2324865722422922,
-      "grad_norm": 0.0028519434854388237,
-      "learning_rate": 0.00019999925132165678,
-      "loss": 46.0,
-      "step": 16120
-    },
-    {
-      "epoch": 1.232563029225682,
-      "grad_norm": 0.0006758978706784546,
-      "learning_rate": 0.00019999925122870824,
-      "loss": 46.0,
-      "step": 16121
-    },
-    {
-      "epoch": 1.2326394862090717,
-      "grad_norm": 0.00047451394493691623,
-      "learning_rate": 0.0001999992511357539,
-      "loss": 46.0,
-      "step": 16122
-    },
-    {
-      "epoch": 1.2327159431924612,
-      "grad_norm": 0.0030632740817964077,
-      "learning_rate": 0.0001999992510427938,
-      "loss": 46.0,
-      "step": 16123
-    },
-    {
-      "epoch": 1.232792400175851,
-      "grad_norm": 0.0018083520699292421,
-      "learning_rate": 0.00019999925094982794,
-      "loss": 46.0,
-      "step": 16124
-    },
-    {
-      "epoch": 1.2328688571592408,
-      "grad_norm": 0.003031206550076604,
-      "learning_rate": 0.0001999992508568563,
-      "loss": 46.0,
-      "step": 16125
-    },
-    {
-      "epoch": 1.2329453141426305,
-      "grad_norm": 0.0008447115542367101,
-      "learning_rate": 0.0001999992507638789,
-      "loss": 46.0,
-      "step": 16126
-    },
-    {
-      "epoch": 1.2330217711260203,
-      "grad_norm": 0.0005253884010016918,
-      "learning_rate": 0.0001999992506708957,
-      "loss": 46.0,
-      "step": 16127
-    },
-    {
-      "epoch": 1.23309822810941,
-      "grad_norm": 0.00044152833288535476,
-      "learning_rate": 0.00019999925057790678,
-      "loss": 46.0,
-      "step": 16128
-    },
-    {
-      "epoch": 1.2331746850927996,
-      "grad_norm": 0.0015555663267150521,
-      "learning_rate": 0.00019999925048491205,
-      "loss": 46.0,
-      "step": 16129
-    },
-    {
-      "epoch": 1.2332511420761894,
-      "grad_norm": 0.0009968159720301628,
-      "learning_rate": 0.00019999925039191157,
-      "loss": 46.0,
-      "step": 16130
-    },
-    {
-      "epoch": 1.2333275990595791,
-      "grad_norm": 0.0025702675338834524,
-      "learning_rate": 0.00019999925029890532,
-      "loss": 46.0,
-      "step": 16131
-    },
-    {
-      "epoch": 1.2334040560429689,
-      "grad_norm": 0.001372497878037393,
-      "learning_rate": 0.0001999992502058933,
-      "loss": 46.0,
-      "step": 16132
-    },
-    {
-      "epoch": 1.2334805130263586,
-      "grad_norm": 0.002881584921851754,
-      "learning_rate": 0.0001999992501128755,
-      "loss": 46.0,
-      "step": 16133
-    },
-    {
-      "epoch": 1.2335569700097482,
-      "grad_norm": 0.0025717185344547033,
-      "learning_rate": 0.00019999925001985193,
-      "loss": 46.0,
-      "step": 16134
-    },
-    {
-      "epoch": 1.233633426993138,
-      "grad_norm": 0.00585021311417222,
-      "learning_rate": 0.0001999992499268226,
-      "loss": 46.0,
-      "step": 16135
-    },
-    {
-      "epoch": 1.2337098839765277,
-      "grad_norm": 0.002392655238509178,
-      "learning_rate": 0.0001999992498337875,
-      "loss": 46.0,
-      "step": 16136
-    },
-    {
-      "epoch": 1.2337863409599175,
-      "grad_norm": 0.001367436139844358,
-      "learning_rate": 0.00019999924974074663,
-      "loss": 46.0,
-      "step": 16137
-    },
-    {
-      "epoch": 1.2338627979433072,
-      "grad_norm": 0.010383804328739643,
-      "learning_rate": 0.0001999992496477,
-      "loss": 46.0,
-      "step": 16138
-    },
-    {
-      "epoch": 1.233939254926697,
-      "grad_norm": 0.00501530896872282,
-      "learning_rate": 0.00019999924955464758,
-      "loss": 46.0,
-      "step": 16139
-    },
-    {
-      "epoch": 1.2340157119100865,
-      "grad_norm": 0.0011613249080255628,
-      "learning_rate": 0.0001999992494615894,
-      "loss": 46.0,
-      "step": 16140
-    },
-    {
-      "epoch": 1.2340921688934763,
-      "grad_norm": 0.00086108228424564,
-      "learning_rate": 0.00019999924936852547,
-      "loss": 46.0,
-      "step": 16141
-    },
-    {
-      "epoch": 1.234168625876866,
-      "grad_norm": 0.00201953062787652,
-      "learning_rate": 0.00019999924927545574,
-      "loss": 46.0,
-      "step": 16142
-    },
-    {
-      "epoch": 1.2342450828602558,
-      "grad_norm": 0.0024786172434687614,
-      "learning_rate": 0.00019999924918238026,
-      "loss": 46.0,
-      "step": 16143
-    },
-    {
-      "epoch": 1.2343215398436456,
-      "grad_norm": 0.0011446401476860046,
-      "learning_rate": 0.000199999249089299,
-      "loss": 46.0,
-      "step": 16144
-    },
-    {
-      "epoch": 1.234397996827035,
-      "grad_norm": 0.0010201006662100554,
-      "learning_rate": 0.000199999248996212,
-      "loss": 46.0,
-      "step": 16145
-    },
-    {
-      "epoch": 1.2344744538104249,
-      "grad_norm": 0.003638614434748888,
-      "learning_rate": 0.0001999992489031192,
-      "loss": 46.0,
-      "step": 16146
-    },
-    {
-      "epoch": 1.2345509107938146,
-      "grad_norm": 0.00087597721721977,
-      "learning_rate": 0.00019999924881002062,
-      "loss": 46.0,
-      "step": 16147
-    },
-    {
-      "epoch": 1.2346273677772044,
-      "grad_norm": 0.0005804141983389854,
-      "learning_rate": 0.00019999924871691628,
-      "loss": 46.0,
-      "step": 16148
-    },
-    {
-      "epoch": 1.2347038247605941,
-      "grad_norm": 0.004878742154687643,
-      "learning_rate": 0.0001999992486238062,
-      "loss": 46.0,
-      "step": 16149
-    },
-    {
-      "epoch": 1.234780281743984,
-      "grad_norm": 0.009944730438292027,
-      "learning_rate": 0.00019999924853069032,
-      "loss": 46.0,
-      "step": 16150
-    },
-    {
-      "epoch": 1.2348567387273734,
-      "grad_norm": 0.0006810511113144457,
-      "learning_rate": 0.0001999992484375687,
-      "loss": 46.0,
-      "step": 16151
-    },
-    {
-      "epoch": 1.2349331957107632,
-      "grad_norm": 0.0020122802816331387,
-      "learning_rate": 0.00019999924834444125,
-      "loss": 46.0,
-      "step": 16152
-    },
-    {
-      "epoch": 1.235009652694153,
-      "grad_norm": 0.0026216183323413134,
-      "learning_rate": 0.0001999992482513081,
-      "loss": 46.0,
-      "step": 16153
-    },
-    {
-      "epoch": 1.2350861096775427,
-      "grad_norm": 0.0010264270240440965,
-      "learning_rate": 0.00019999924815816914,
-      "loss": 46.0,
-      "step": 16154
-    },
-    {
-      "epoch": 1.2351625666609323,
-      "grad_norm": 0.0011440090602263808,
-      "learning_rate": 0.0001999992480650244,
-      "loss": 46.0,
-      "step": 16155
-    },
-    {
-      "epoch": 1.235239023644322,
-      "grad_norm": 0.0011800434440374374,
-      "learning_rate": 0.00019999924797187393,
-      "loss": 46.0,
-      "step": 16156
-    },
-    {
-      "epoch": 1.2353154806277118,
-      "grad_norm": 0.0014037586515769362,
-      "learning_rate": 0.00019999924787871765,
-      "loss": 46.0,
-      "step": 16157
-    },
-    {
-      "epoch": 1.2353919376111016,
-      "grad_norm": 0.0006325557478703558,
-      "learning_rate": 0.00019999924778555563,
-      "loss": 46.0,
-      "step": 16158
-    },
-    {
-      "epoch": 1.2354683945944913,
-      "grad_norm": 0.0014421932864934206,
-      "learning_rate": 0.00019999924769238783,
-      "loss": 46.0,
-      "step": 16159
-    },
-    {
-      "epoch": 1.235544851577881,
-      "grad_norm": 0.0009284376865252852,
-      "learning_rate": 0.0001999992475992143,
-      "loss": 46.0,
-      "step": 16160
-    },
-    {
-      "epoch": 1.2356213085612708,
-      "grad_norm": 0.001163533772341907,
-      "learning_rate": 0.00019999924750603495,
-      "loss": 46.0,
-      "step": 16161
-    },
-    {
-      "epoch": 1.2356977655446604,
-      "grad_norm": 0.0024985643103718758,
-      "learning_rate": 0.00019999924741284983,
-      "loss": 46.0,
-      "step": 16162
-    },
-    {
-      "epoch": 1.2357742225280501,
-      "grad_norm": 0.0006378112593665719,
-      "learning_rate": 0.00019999924731965897,
-      "loss": 46.0,
-      "step": 16163
-    },
-    {
-      "epoch": 1.23585067951144,
-      "grad_norm": 0.0007794823031872511,
-      "learning_rate": 0.00019999924722646233,
-      "loss": 46.0,
-      "step": 16164
-    },
-    {
-      "epoch": 1.2359271364948297,
-      "grad_norm": 0.0008109778864309192,
-      "learning_rate": 0.0001999992471332599,
-      "loss": 46.0,
-      "step": 16165
-    },
-    {
-      "epoch": 1.2360035934782192,
-      "grad_norm": 0.0007147433934733272,
-      "learning_rate": 0.0001999992470400517,
-      "loss": 46.0,
-      "step": 16166
-    },
-    {
-      "epoch": 1.236080050461609,
-      "grad_norm": 0.00229713530279696,
-      "learning_rate": 0.00019999924694683778,
-      "loss": 46.0,
-      "step": 16167
-    },
-    {
-      "epoch": 1.2361565074449987,
-      "grad_norm": 0.0012973527191206813,
-      "learning_rate": 0.00019999924685361803,
-      "loss": 46.0,
-      "step": 16168
-    },
-    {
-      "epoch": 1.2362329644283885,
-      "grad_norm": 0.0017626202898100019,
-      "learning_rate": 0.00019999924676039258,
-      "loss": 46.0,
-      "step": 16169
-    },
-    {
-      "epoch": 1.2363094214117782,
-      "grad_norm": 0.0006364518194459379,
-      "learning_rate": 0.0001999992466671613,
-      "loss": 46.0,
-      "step": 16170
-    },
-    {
-      "epoch": 1.236385878395168,
-      "grad_norm": 0.002682435791939497,
-      "learning_rate": 0.00019999924657392425,
-      "loss": 46.0,
-      "step": 16171
-    },
-    {
-      "epoch": 1.2364623353785575,
-      "grad_norm": 0.0010396478464826941,
-      "learning_rate": 0.00019999924648068148,
-      "loss": 46.0,
-      "step": 16172
-    },
-    {
-      "epoch": 1.2365387923619473,
-      "grad_norm": 0.0009788823081180453,
-      "learning_rate": 0.0001999992463874329,
-      "loss": 46.0,
-      "step": 16173
-    },
-    {
-      "epoch": 1.236615249345337,
-      "grad_norm": 0.0010156398639082909,
-      "learning_rate": 0.00019999924629417857,
-      "loss": 46.0,
-      "step": 16174
-    },
-    {
-      "epoch": 1.2366917063287268,
-      "grad_norm": 0.000712582899723202,
-      "learning_rate": 0.00019999924620091845,
-      "loss": 46.0,
-      "step": 16175
-    },
-    {
-      "epoch": 1.2367681633121166,
-      "grad_norm": 0.0018316407222300768,
-      "learning_rate": 0.00019999924610765256,
-      "loss": 46.0,
-      "step": 16176
-    },
-    {
-      "epoch": 1.2368446202955061,
-      "grad_norm": 0.0013650762848556042,
-      "learning_rate": 0.00019999924601438093,
-      "loss": 46.0,
-      "step": 16177
-    },
-    {
-      "epoch": 1.2369210772788959,
-      "grad_norm": 0.0006641428335569799,
-      "learning_rate": 0.00019999924592110352,
-      "loss": 46.0,
-      "step": 16178
-    },
-    {
-      "epoch": 1.2369975342622856,
-      "grad_norm": 0.0006430278299376369,
-      "learning_rate": 0.00019999924582782033,
-      "loss": 46.0,
-      "step": 16179
-    },
-    {
-      "epoch": 1.2370739912456754,
-      "grad_norm": 0.001062024850398302,
-      "learning_rate": 0.00019999924573453138,
-      "loss": 46.0,
-      "step": 16180
-    },
-    {
-      "epoch": 1.2371504482290652,
-      "grad_norm": 0.0004638636310119182,
-      "learning_rate": 0.00019999924564123665,
-      "loss": 46.0,
-      "step": 16181
-    },
-    {
-      "epoch": 1.237226905212455,
-      "grad_norm": 0.0006842865259386599,
-      "learning_rate": 0.00019999924554793617,
-      "loss": 46.0,
-      "step": 16182
-    },
-    {
-      "epoch": 1.2373033621958445,
-      "grad_norm": 0.000827309675514698,
-      "learning_rate": 0.0001999992454546299,
-      "loss": 46.0,
-      "step": 16183
-    },
-    {
-      "epoch": 1.2373798191792342,
-      "grad_norm": 0.000634370488114655,
-      "learning_rate": 0.00019999924536131785,
-      "loss": 46.0,
-      "step": 16184
-    },
-    {
-      "epoch": 1.237456276162624,
-      "grad_norm": 0.00038265628973022103,
-      "learning_rate": 0.00019999924526800005,
-      "loss": 46.0,
-      "step": 16185
-    },
-    {
-      "epoch": 1.2375327331460138,
-      "grad_norm": 0.0005271471454761922,
-      "learning_rate": 0.00019999924517467648,
-      "loss": 46.0,
-      "step": 16186
-    },
-    {
-      "epoch": 1.2376091901294035,
-      "grad_norm": 0.013602126389741898,
-      "learning_rate": 0.00019999924508134714,
-      "loss": 46.0,
-      "step": 16187
-    },
-    {
-      "epoch": 1.237685647112793,
-      "grad_norm": 0.0012449331115931273,
-      "learning_rate": 0.00019999924498801205,
-      "loss": 46.0,
-      "step": 16188
-    },
-    {
-      "epoch": 1.2377621040961828,
-      "grad_norm": 0.0010873565915971994,
-      "learning_rate": 0.00019999924489467116,
-      "loss": 46.0,
-      "step": 16189
-    },
-    {
-      "epoch": 1.2378385610795726,
-      "grad_norm": 0.0019398077856749296,
-      "learning_rate": 0.0001999992448013245,
-      "loss": 46.0,
-      "step": 16190
-    },
-    {
-      "epoch": 1.2379150180629623,
-      "grad_norm": 0.0007611886248923838,
-      "learning_rate": 0.0001999992447079721,
-      "loss": 46.0,
-      "step": 16191
-    },
-    {
-      "epoch": 1.237991475046352,
-      "grad_norm": 0.006044069305062294,
-      "learning_rate": 0.0001999992446146139,
-      "loss": 46.0,
-      "step": 16192
-    },
-    {
-      "epoch": 1.2380679320297419,
-      "grad_norm": 0.002332279458642006,
-      "learning_rate": 0.00019999924452124995,
-      "loss": 46.0,
-      "step": 16193
-    },
-    {
-      "epoch": 1.2381443890131314,
-      "grad_norm": 0.0009618547046557069,
-      "learning_rate": 0.0001999992444278802,
-      "loss": 46.0,
-      "step": 16194
-    },
-    {
-      "epoch": 1.2382208459965212,
-      "grad_norm": 0.0028223982080817223,
-      "learning_rate": 0.00019999924433450472,
-      "loss": 46.0,
-      "step": 16195
-    },
-    {
-      "epoch": 1.238297302979911,
-      "grad_norm": 0.0026741530746221542,
-      "learning_rate": 0.00019999924424112347,
-      "loss": 46.0,
-      "step": 16196
-    },
-    {
-      "epoch": 1.2383737599633007,
-      "grad_norm": 0.0014414218021556735,
-      "learning_rate": 0.00019999924414773642,
-      "loss": 46.0,
-      "step": 16197
-    },
-    {
-      "epoch": 1.2384502169466904,
-      "grad_norm": 0.0011779188644140959,
-      "learning_rate": 0.00019999924405434362,
-      "loss": 46.0,
-      "step": 16198
-    },
-    {
-      "epoch": 1.23852667393008,
-      "grad_norm": 0.0012218063930049539,
-      "learning_rate": 0.00019999924396094505,
-      "loss": 46.0,
-      "step": 16199
-    },
-    {
-      "epoch": 1.2386031309134697,
-      "grad_norm": 0.0026084252167493105,
-      "learning_rate": 0.00019999924386754068,
-      "loss": 46.0,
-      "step": 16200
-    },
-    {
-      "epoch": 1.2386795878968595,
-      "grad_norm": 0.0030180804897099733,
-      "learning_rate": 0.0001999992437741306,
-      "loss": 46.0,
-      "step": 16201
-    },
-    {
-      "epoch": 1.2387560448802493,
-      "grad_norm": 0.003117549931630492,
-      "learning_rate": 0.00019999924368071473,
-      "loss": 46.0,
-      "step": 16202
-    },
-    {
-      "epoch": 1.238832501863639,
-      "grad_norm": 0.0025547968689352274,
-      "learning_rate": 0.00019999924358729304,
-      "loss": 46.0,
-      "step": 16203
-    },
-    {
-      "epoch": 1.2389089588470288,
-      "grad_norm": 0.0014640373410657048,
-      "learning_rate": 0.00019999924349386566,
-      "loss": 46.0,
-      "step": 16204
-    },
-    {
-      "epoch": 1.2389854158304183,
-      "grad_norm": 0.00039131238008849323,
-      "learning_rate": 0.00019999924340043243,
-      "loss": 46.0,
-      "step": 16205
-    },
-    {
-      "epoch": 1.239061872813808,
-      "grad_norm": 0.0013695830712094903,
-      "learning_rate": 0.0001999992433069935,
-      "loss": 46.0,
-      "step": 16206
-    },
-    {
-      "epoch": 1.2391383297971978,
-      "grad_norm": 0.0011735250009223819,
-      "learning_rate": 0.00019999924321354877,
-      "loss": 46.0,
-      "step": 16207
-    },
-    {
-      "epoch": 1.2392147867805876,
-      "grad_norm": 0.001279689371585846,
-      "learning_rate": 0.00019999924312009827,
-      "loss": 46.0,
-      "step": 16208
-    },
-    {
-      "epoch": 1.2392912437639774,
-      "grad_norm": 0.0012681097723543644,
-      "learning_rate": 0.000199999243026642,
-      "loss": 46.0,
-      "step": 16209
-    },
-    {
-      "epoch": 1.239367700747367,
-      "grad_norm": 0.0022884411737322807,
-      "learning_rate": 0.00019999924293317997,
-      "loss": 46.0,
-      "step": 16210
-    },
-    {
-      "epoch": 1.2394441577307567,
-      "grad_norm": 0.005618903785943985,
-      "learning_rate": 0.00019999924283971217,
-      "loss": 46.0,
-      "step": 16211
-    },
-    {
-      "epoch": 1.2395206147141464,
-      "grad_norm": 0.001149306888692081,
-      "learning_rate": 0.00019999924274623858,
-      "loss": 46.0,
-      "step": 16212
-    },
-    {
-      "epoch": 1.2395970716975362,
-      "grad_norm": 0.0005931926425546408,
-      "learning_rate": 0.00019999924265275923,
-      "loss": 46.0,
-      "step": 16213
-    },
-    {
-      "epoch": 1.239673528680926,
-      "grad_norm": 0.0034078762400895357,
-      "learning_rate": 0.00019999924255927412,
-      "loss": 46.0,
-      "step": 16214
-    },
-    {
-      "epoch": 1.2397499856643157,
-      "grad_norm": 0.0017527914606034756,
-      "learning_rate": 0.00019999924246578323,
-      "loss": 46.0,
-      "step": 16215
-    },
-    {
-      "epoch": 1.2398264426477053,
-      "grad_norm": 0.006149209104478359,
-      "learning_rate": 0.00019999924237228657,
-      "loss": 46.0,
-      "step": 16216
-    },
-    {
-      "epoch": 1.239902899631095,
-      "grad_norm": 0.0046256110072135925,
-      "learning_rate": 0.00019999924227878416,
-      "loss": 46.0,
-      "step": 16217
-    },
-    {
-      "epoch": 1.2399793566144848,
-      "grad_norm": 0.0004790711391251534,
-      "learning_rate": 0.00019999924218527598,
-      "loss": 46.0,
-      "step": 16218
-    },
-    {
-      "epoch": 1.2400558135978745,
-      "grad_norm": 0.0002844065602403134,
-      "learning_rate": 0.00019999924209176202,
-      "loss": 46.0,
-      "step": 16219
-    },
-    {
-      "epoch": 1.2401322705812643,
-      "grad_norm": 0.00136388442479074,
-      "learning_rate": 0.00019999924199824227,
-      "loss": 46.0,
-      "step": 16220
-    },
-    {
-      "epoch": 1.2402087275646538,
-      "grad_norm": 0.001815884024836123,
-      "learning_rate": 0.00019999924190471676,
-      "loss": 46.0,
-      "step": 16221
-    },
-    {
-      "epoch": 1.2402851845480436,
-      "grad_norm": 0.0018731511663645506,
-      "learning_rate": 0.0001999992418111855,
-      "loss": 46.0,
-      "step": 16222
-    },
-    {
-      "epoch": 1.2403616415314334,
-      "grad_norm": 0.0005997954867780209,
-      "learning_rate": 0.00019999924171764847,
-      "loss": 46.0,
-      "step": 16223
-    },
-    {
-      "epoch": 1.2404380985148231,
-      "grad_norm": 0.002003435045480728,
-      "learning_rate": 0.00019999924162410567,
-      "loss": 46.0,
-      "step": 16224
-    },
-    {
-      "epoch": 1.2405145554982129,
-      "grad_norm": 0.006114141549915075,
-      "learning_rate": 0.00019999924153055708,
-      "loss": 46.0,
-      "step": 16225
-    },
-    {
-      "epoch": 1.2405910124816026,
-      "grad_norm": 0.0009145902586169541,
-      "learning_rate": 0.00019999924143700274,
-      "loss": 46.0,
-      "step": 16226
-    },
-    {
-      "epoch": 1.2406674694649922,
-      "grad_norm": 0.0016500097699463367,
-      "learning_rate": 0.00019999924134344262,
-      "loss": 46.0,
-      "step": 16227
-    },
-    {
-      "epoch": 1.240743926448382,
-      "grad_norm": 0.0005913766217418015,
-      "learning_rate": 0.00019999924124987673,
-      "loss": 46.0,
-      "step": 16228
-    },
-    {
-      "epoch": 1.2408203834317717,
-      "grad_norm": 0.0038779121823608875,
-      "learning_rate": 0.00019999924115630507,
-      "loss": 46.0,
-      "step": 16229
-    },
-    {
-      "epoch": 1.2408968404151615,
-      "grad_norm": 0.0003662245871964842,
-      "learning_rate": 0.00019999924106272766,
-      "loss": 46.0,
-      "step": 16230
-    },
-    {
-      "epoch": 1.2409732973985512,
-      "grad_norm": 0.0008042265544645488,
-      "learning_rate": 0.00019999924096914445,
-      "loss": 46.0,
-      "step": 16231
-    },
-    {
-      "epoch": 1.2410497543819408,
-      "grad_norm": 0.0015300572849810123,
-      "learning_rate": 0.00019999924087555547,
-      "loss": 46.0,
-      "step": 16232
-    },
-    {
-      "epoch": 1.2411262113653305,
-      "grad_norm": 0.0005596036207862198,
-      "learning_rate": 0.00019999924078196074,
-      "loss": 46.0,
-      "step": 16233
-    },
-    {
-      "epoch": 1.2412026683487203,
-      "grad_norm": 0.0005785702960565686,
-      "learning_rate": 0.00019999924068836027,
-      "loss": 46.0,
-      "step": 16234
-    },
-    {
-      "epoch": 1.24127912533211,
-      "grad_norm": 0.016834190115332603,
-      "learning_rate": 0.000199999240594754,
-      "loss": 46.0,
-      "step": 16235
-    },
-    {
-      "epoch": 1.2413555823154998,
-      "grad_norm": 0.002775345230475068,
-      "learning_rate": 0.00019999924050114194,
-      "loss": 46.0,
-      "step": 16236
-    },
-    {
-      "epoch": 1.2414320392988896,
-      "grad_norm": 0.0033433123026043177,
-      "learning_rate": 0.00019999924040752412,
-      "loss": 46.0,
-      "step": 16237
-    },
-    {
-      "epoch": 1.241508496282279,
-      "grad_norm": 0.0027108630165457726,
-      "learning_rate": 0.00019999924031390055,
-      "loss": 46.0,
-      "step": 16238
-    },
-    {
-      "epoch": 1.2415849532656689,
-      "grad_norm": 0.0018398347310721874,
-      "learning_rate": 0.00019999924022027119,
-      "loss": 46.0,
-      "step": 16239
-    },
-    {
-      "epoch": 1.2416614102490586,
-      "grad_norm": 0.002263805130496621,
-      "learning_rate": 0.0001999992401266361,
-      "loss": 46.0,
-      "step": 16240
-    },
-    {
-      "epoch": 1.2417378672324484,
-      "grad_norm": 0.0023306889925152063,
-      "learning_rate": 0.00019999924003299518,
-      "loss": 46.0,
-      "step": 16241
-    },
-    {
-      "epoch": 1.2418143242158382,
-      "grad_norm": 0.0023174022790044546,
-      "learning_rate": 0.00019999923993934852,
-      "loss": 46.0,
-      "step": 16242
-    },
-    {
-      "epoch": 1.2418907811992277,
-      "grad_norm": 0.0031036718282848597,
-      "learning_rate": 0.0001999992398456961,
-      "loss": 46.0,
-      "step": 16243
-    },
-    {
-      "epoch": 1.2419672381826175,
-      "grad_norm": 0.00155420892406255,
-      "learning_rate": 0.0001999992397520379,
-      "loss": 46.0,
-      "step": 16244
-    },
-    {
-      "epoch": 1.2420436951660072,
-      "grad_norm": 0.0039010613691061735,
-      "learning_rate": 0.00019999923965837395,
-      "loss": 46.0,
-      "step": 16245
-    },
-    {
-      "epoch": 1.242120152149397,
-      "grad_norm": 0.0008912440971471369,
-      "learning_rate": 0.00019999923956470422,
-      "loss": 46.0,
-      "step": 16246
-    },
-    {
-      "epoch": 1.2421966091327867,
-      "grad_norm": 0.000617080891970545,
-      "learning_rate": 0.0001999992394710287,
-      "loss": 46.0,
-      "step": 16247
-    },
-    {
-      "epoch": 1.2422730661161765,
-      "grad_norm": 0.0014821469085291028,
-      "learning_rate": 0.00019999923937734742,
-      "loss": 46.0,
-      "step": 16248
-    },
-    {
-      "epoch": 1.242349523099566,
-      "grad_norm": 0.0006066728383302689,
-      "learning_rate": 0.00019999923928366037,
-      "loss": 46.0,
-      "step": 16249
-    },
-    {
-      "epoch": 1.2424259800829558,
-      "grad_norm": 0.002139391377568245,
-      "learning_rate": 0.00019999923918996758,
-      "loss": 46.0,
-      "step": 16250
-    },
-    {
-      "epoch": 1.2425024370663456,
-      "grad_norm": 0.0019193944754078984,
-      "learning_rate": 0.00019999923909626898,
-      "loss": 46.0,
-      "step": 16251
-    },
-    {
-      "epoch": 1.2425788940497353,
-      "grad_norm": 0.008647779934108257,
-      "learning_rate": 0.00019999923900256464,
-      "loss": 46.0,
-      "step": 16252
-    },
-    {
-      "epoch": 1.242655351033125,
-      "grad_norm": 0.000417533447034657,
-      "learning_rate": 0.00019999923890885453,
-      "loss": 46.0,
-      "step": 16253
-    },
-    {
-      "epoch": 1.2427318080165146,
-      "grad_norm": 0.004375928081572056,
-      "learning_rate": 0.0001999992388151386,
-      "loss": 46.0,
-      "step": 16254
-    },
-    {
-      "epoch": 1.2428082649999044,
-      "grad_norm": 0.004379493650048971,
-      "learning_rate": 0.00019999923872141695,
-      "loss": 46.0,
-      "step": 16255
-    },
-    {
-      "epoch": 1.2428847219832941,
-      "grad_norm": 0.0010251826606690884,
-      "learning_rate": 0.00019999923862768952,
-      "loss": 46.0,
-      "step": 16256
-    },
-    {
-      "epoch": 1.242961178966684,
-      "grad_norm": 0.0006800846313126385,
-      "learning_rate": 0.00019999923853395633,
-      "loss": 46.0,
-      "step": 16257
-    },
-    {
-      "epoch": 1.2430376359500737,
-      "grad_norm": 0.002811568323522806,
-      "learning_rate": 0.00019999923844021735,
-      "loss": 46.0,
-      "step": 16258
-    },
-    {
-      "epoch": 1.2431140929334634,
-      "grad_norm": 0.0007100341026671231,
-      "learning_rate": 0.00019999923834647263,
-      "loss": 46.0,
-      "step": 16259
-    },
-    {
-      "epoch": 1.243190549916853,
-      "grad_norm": 0.0012785578146576881,
-      "learning_rate": 0.00019999923825272212,
-      "loss": 46.0,
-      "step": 16260
-    },
-    {
-      "epoch": 1.2432670069002427,
-      "grad_norm": 0.0009894692339003086,
-      "learning_rate": 0.00019999923815896585,
-      "loss": 46.0,
-      "step": 16261
-    },
-    {
-      "epoch": 1.2433434638836325,
-      "grad_norm": 0.00040542345959693193,
-      "learning_rate": 0.0001999992380652038,
-      "loss": 46.0,
-      "step": 16262
-    },
-    {
-      "epoch": 1.2434199208670222,
-      "grad_norm": 0.0017142135184258223,
-      "learning_rate": 0.00019999923797143598,
-      "loss": 46.0,
-      "step": 16263
-    },
-    {
-      "epoch": 1.243496377850412,
-      "grad_norm": 0.000728804909158498,
-      "learning_rate": 0.00019999923787766239,
-      "loss": 46.0,
-      "step": 16264
-    },
-    {
-      "epoch": 1.2435728348338015,
-      "grad_norm": 0.0009789259638637304,
-      "learning_rate": 0.00019999923778388302,
-      "loss": 46.0,
-      "step": 16265
-    },
-    {
-      "epoch": 1.2436492918171913,
-      "grad_norm": 0.001567603787407279,
-      "learning_rate": 0.00019999923769009793,
-      "loss": 46.0,
-      "step": 16266
-    },
-    {
-      "epoch": 1.243725748800581,
-      "grad_norm": 0.0006959529710002244,
-      "learning_rate": 0.00019999923759630704,
-      "loss": 46.0,
-      "step": 16267
-    },
-    {
-      "epoch": 1.2438022057839708,
-      "grad_norm": 0.0014233523979783058,
-      "learning_rate": 0.00019999923750251038,
-      "loss": 46.0,
-      "step": 16268
-    },
-    {
-      "epoch": 1.2438786627673606,
-      "grad_norm": 0.0011403318494558334,
-      "learning_rate": 0.00019999923740870795,
-      "loss": 46.0,
-      "step": 16269
-    },
-    {
-      "epoch": 1.2439551197507503,
-      "grad_norm": 0.0023819133639335632,
-      "learning_rate": 0.00019999923731489974,
-      "loss": 46.0,
-      "step": 16270
-    },
-    {
-      "epoch": 1.2440315767341399,
-      "grad_norm": 0.0026477393694221973,
-      "learning_rate": 0.00019999923722108576,
-      "loss": 46.0,
-      "step": 16271
-    },
-    {
-      "epoch": 1.2441080337175296,
-      "grad_norm": 0.0011295088334009051,
-      "learning_rate": 0.000199999237127266,
-      "loss": 46.0,
-      "step": 16272
-    },
-    {
-      "epoch": 1.2441844907009194,
-      "grad_norm": 0.0010174043709412217,
-      "learning_rate": 0.0001999992370334405,
-      "loss": 46.0,
-      "step": 16273
-    },
-    {
-      "epoch": 1.2442609476843092,
-      "grad_norm": 0.000810265657491982,
-      "learning_rate": 0.00019999923693960923,
-      "loss": 46.0,
-      "step": 16274
-    },
-    {
-      "epoch": 1.244337404667699,
-      "grad_norm": 0.0014384017558768392,
-      "learning_rate": 0.00019999923684577218,
-      "loss": 46.0,
-      "step": 16275
-    },
-    {
-      "epoch": 1.2444138616510885,
-      "grad_norm": 0.0015048125060275197,
-      "learning_rate": 0.00019999923675192936,
-      "loss": 46.0,
-      "step": 16276
-    },
-    {
-      "epoch": 1.2444903186344782,
-      "grad_norm": 0.001139685045927763,
-      "learning_rate": 0.00019999923665808077,
-      "loss": 46.0,
-      "step": 16277
-    },
-    {
-      "epoch": 1.244566775617868,
-      "grad_norm": 0.001738716964609921,
-      "learning_rate": 0.0001999992365642264,
-      "loss": 46.0,
-      "step": 16278
-    },
-    {
-      "epoch": 1.2446432326012578,
-      "grad_norm": 0.0017025260021910071,
-      "learning_rate": 0.00019999923647036629,
-      "loss": 46.0,
-      "step": 16279
-    },
-    {
-      "epoch": 1.2447196895846475,
-      "grad_norm": 0.012858227826654911,
-      "learning_rate": 0.0001999992363765004,
-      "loss": 46.0,
-      "step": 16280
-    },
-    {
-      "epoch": 1.2447961465680373,
-      "grad_norm": 0.0013719985727220774,
-      "learning_rate": 0.00019999923628262874,
-      "loss": 46.0,
-      "step": 16281
-    },
-    {
-      "epoch": 1.2448726035514268,
-      "grad_norm": 0.0010038783075287938,
-      "learning_rate": 0.0001999992361887513,
-      "loss": 46.0,
-      "step": 16282
-    },
-    {
-      "epoch": 1.2449490605348166,
-      "grad_norm": 0.0017927760491147637,
-      "learning_rate": 0.0001999992360948681,
-      "loss": 46.0,
-      "step": 16283
-    },
-    {
-      "epoch": 1.2450255175182063,
-      "grad_norm": 0.00580955296754837,
-      "learning_rate": 0.00019999923600097912,
-      "loss": 46.0,
-      "step": 16284
-    },
-    {
-      "epoch": 1.245101974501596,
-      "grad_norm": 0.004721891600638628,
-      "learning_rate": 0.00019999923590708436,
-      "loss": 46.0,
-      "step": 16285
-    },
-    {
-      "epoch": 1.2451784314849856,
-      "grad_norm": 0.0008275595610029995,
-      "learning_rate": 0.00019999923581318386,
-      "loss": 46.0,
-      "step": 16286
-    },
-    {
-      "epoch": 1.2452548884683754,
-      "grad_norm": 0.0012638730695471168,
-      "learning_rate": 0.00019999923571927756,
-      "loss": 46.0,
-      "step": 16287
-    },
-    {
-      "epoch": 1.2453313454517652,
-      "grad_norm": 0.001253433059900999,
-      "learning_rate": 0.00019999923562536554,
-      "loss": 46.0,
-      "step": 16288
-    },
-    {
-      "epoch": 1.245407802435155,
-      "grad_norm": 0.002018025843426585,
-      "learning_rate": 0.00019999923553144772,
-      "loss": 46.0,
-      "step": 16289
-    },
-    {
-      "epoch": 1.2454842594185447,
-      "grad_norm": 0.003162020118907094,
-      "learning_rate": 0.00019999923543752413,
-      "loss": 46.0,
-      "step": 16290
-    },
-    {
-      "epoch": 1.2455607164019344,
-      "grad_norm": 0.0006781438132748008,
-      "learning_rate": 0.00019999923534359476,
-      "loss": 46.0,
-      "step": 16291
-    },
-    {
-      "epoch": 1.2456371733853242,
-      "grad_norm": 0.0013762114103883505,
-      "learning_rate": 0.00019999923524965962,
-      "loss": 46.0,
-      "step": 16292
-    },
-    {
-      "epoch": 1.2457136303687137,
-      "grad_norm": 0.0008886960567906499,
-      "learning_rate": 0.00019999923515571873,
-      "loss": 46.0,
-      "step": 16293
-    },
-    {
-      "epoch": 1.2457900873521035,
-      "grad_norm": 0.009898100048303604,
-      "learning_rate": 0.00019999923506177207,
-      "loss": 46.0,
-      "step": 16294
-    },
-    {
-      "epoch": 1.2458665443354933,
-      "grad_norm": 0.0008863803232088685,
-      "learning_rate": 0.00019999923496781964,
-      "loss": 46.0,
-      "step": 16295
-    },
-    {
-      "epoch": 1.245943001318883,
-      "grad_norm": 0.0010941906366497278,
-      "learning_rate": 0.00019999923487386143,
-      "loss": 46.0,
-      "step": 16296
-    },
-    {
-      "epoch": 1.2460194583022726,
-      "grad_norm": 0.0005877632647752762,
-      "learning_rate": 0.00019999923477989745,
-      "loss": 46.0,
-      "step": 16297
-    },
-    {
-      "epoch": 1.2460959152856623,
-      "grad_norm": 0.0010235217632725835,
-      "learning_rate": 0.00019999923468592772,
-      "loss": 46.0,
-      "step": 16298
-    },
-    {
-      "epoch": 1.246172372269052,
-      "grad_norm": 0.010449610650539398,
-      "learning_rate": 0.00019999923459195217,
-      "loss": 46.0,
-      "step": 16299
-    },
-    {
-      "epoch": 1.2462488292524418,
-      "grad_norm": 0.002556039486080408,
-      "learning_rate": 0.00019999923449797092,
-      "loss": 46.0,
-      "step": 16300
-    },
-    {
-      "epoch": 1.2463252862358316,
-      "grad_norm": 0.0046071987599134445,
-      "learning_rate": 0.00019999923440398385,
-      "loss": 46.0,
-      "step": 16301
-    },
-    {
-      "epoch": 1.2464017432192214,
-      "grad_norm": 0.003930953796952963,
-      "learning_rate": 0.00019999923430999103,
-      "loss": 46.0,
-      "step": 16302
-    },
-    {
-      "epoch": 1.2464782002026111,
-      "grad_norm": 0.00046681874664500356,
-      "learning_rate": 0.00019999923421599243,
-      "loss": 46.0,
-      "step": 16303
-    },
-    {
-      "epoch": 1.2465546571860007,
-      "grad_norm": 0.0007205181173048913,
-      "learning_rate": 0.00019999923412198807,
-      "loss": 46.0,
-      "step": 16304
-    },
-    {
-      "epoch": 1.2466311141693904,
-      "grad_norm": 0.004459925461560488,
-      "learning_rate": 0.00019999923402797793,
-      "loss": 46.0,
-      "step": 16305
-    },
-    {
-      "epoch": 1.2467075711527802,
-      "grad_norm": 0.0005660425522364676,
-      "learning_rate": 0.00019999923393396204,
-      "loss": 46.0,
-      "step": 16306
-    },
-    {
-      "epoch": 1.24678402813617,
-      "grad_norm": 0.0022302453871816397,
-      "learning_rate": 0.00019999923383994038,
-      "loss": 46.0,
-      "step": 16307
-    },
-    {
-      "epoch": 1.2468604851195595,
-      "grad_norm": 0.0012974118581041694,
-      "learning_rate": 0.00019999923374591295,
-      "loss": 46.0,
-      "step": 16308
-    },
-    {
-      "epoch": 1.2469369421029493,
-      "grad_norm": 0.0009810287738218904,
-      "learning_rate": 0.0001999992336518797,
-      "loss": 46.0,
-      "step": 16309
-    },
-    {
-      "epoch": 1.247013399086339,
-      "grad_norm": 0.0011373079614713788,
-      "learning_rate": 0.00019999923355784076,
-      "loss": 46.0,
-      "step": 16310
-    },
-    {
-      "epoch": 1.2470898560697288,
-      "grad_norm": 0.0012776160147041082,
-      "learning_rate": 0.00019999923346379598,
-      "loss": 46.0,
-      "step": 16311
-    },
-    {
-      "epoch": 1.2471663130531185,
-      "grad_norm": 0.004559324588626623,
-      "learning_rate": 0.00019999923336974548,
-      "loss": 46.0,
-      "step": 16312
-    },
-    {
-      "epoch": 1.2472427700365083,
-      "grad_norm": 0.0019243181450292468,
-      "learning_rate": 0.0001999992332756892,
-      "loss": 46.0,
-      "step": 16313
-    },
-    {
-      "epoch": 1.2473192270198978,
-      "grad_norm": 0.0018928120844066143,
-      "learning_rate": 0.00019999923318162713,
-      "loss": 46.0,
-      "step": 16314
-    },
-    {
-      "epoch": 1.2473956840032876,
-      "grad_norm": 0.001409701886586845,
-      "learning_rate": 0.0001999992330875593,
-      "loss": 46.0,
-      "step": 16315
-    },
-    {
-      "epoch": 1.2474721409866774,
-      "grad_norm": 0.0011028624139726162,
-      "learning_rate": 0.00019999923299348572,
-      "loss": 46.0,
-      "step": 16316
-    },
-    {
-      "epoch": 1.2475485979700671,
-      "grad_norm": 0.006825182121247053,
-      "learning_rate": 0.00019999923289940635,
-      "loss": 46.0,
-      "step": 16317
-    },
-    {
-      "epoch": 1.2476250549534569,
-      "grad_norm": 0.0010181937832385302,
-      "learning_rate": 0.0001999992328053212,
-      "loss": 46.0,
-      "step": 16318
-    },
-    {
-      "epoch": 1.2477015119368464,
-      "grad_norm": 0.0013277959078550339,
-      "learning_rate": 0.00019999923271123033,
-      "loss": 46.0,
-      "step": 16319
-    },
-    {
-      "epoch": 1.2477779689202362,
-      "grad_norm": 0.0024626338854432106,
-      "learning_rate": 0.00019999923261713364,
-      "loss": 46.0,
-      "step": 16320
-    },
-    {
-      "epoch": 1.247854425903626,
-      "grad_norm": 0.0015709082363173366,
-      "learning_rate": 0.0001999992325230312,
-      "loss": 46.0,
-      "step": 16321
-    },
-    {
-      "epoch": 1.2479308828870157,
-      "grad_norm": 0.0012152184499427676,
-      "learning_rate": 0.000199999232428923,
-      "loss": 46.0,
-      "step": 16322
-    },
-    {
-      "epoch": 1.2480073398704055,
-      "grad_norm": 0.0012776012299582362,
-      "learning_rate": 0.00019999923233480902,
-      "loss": 46.0,
-      "step": 16323
-    },
-    {
-      "epoch": 1.2480837968537952,
-      "grad_norm": 0.0006866256590001285,
-      "learning_rate": 0.00019999923224068927,
-      "loss": 46.0,
-      "step": 16324
-    },
-    {
-      "epoch": 1.2481602538371848,
-      "grad_norm": 0.0012240622891113162,
-      "learning_rate": 0.00019999923214656377,
-      "loss": 46.0,
-      "step": 16325
-    },
-    {
-      "epoch": 1.2482367108205745,
-      "grad_norm": 0.001377153443172574,
-      "learning_rate": 0.00019999923205243247,
-      "loss": 46.0,
-      "step": 16326
-    },
-    {
-      "epoch": 1.2483131678039643,
-      "grad_norm": 0.020239299163222313,
-      "learning_rate": 0.00019999923195829542,
-      "loss": 46.0,
-      "step": 16327
-    },
-    {
-      "epoch": 1.248389624787354,
-      "grad_norm": 0.0008103779400698841,
-      "learning_rate": 0.0001999992318641526,
-      "loss": 46.0,
-      "step": 16328
-    },
-    {
-      "epoch": 1.2484660817707438,
-      "grad_norm": 0.0009015205432660878,
-      "learning_rate": 0.00019999923177000398,
-      "loss": 46.0,
-      "step": 16329
-    },
-    {
-      "epoch": 1.2485425387541333,
-      "grad_norm": 0.0009062756435014307,
-      "learning_rate": 0.00019999923167584961,
-      "loss": 46.0,
-      "step": 16330
-    },
-    {
-      "epoch": 1.248618995737523,
-      "grad_norm": 0.0006869305507279932,
-      "learning_rate": 0.00019999923158168947,
-      "loss": 46.0,
-      "step": 16331
-    },
-    {
-      "epoch": 1.2486954527209129,
-      "grad_norm": 0.0020838575437664986,
-      "learning_rate": 0.0001999992314875236,
-      "loss": 46.0,
-      "step": 16332
-    },
-    {
-      "epoch": 1.2487719097043026,
-      "grad_norm": 0.0006608259282074869,
-      "learning_rate": 0.0001999992313933519,
-      "loss": 46.0,
-      "step": 16333
-    },
-    {
-      "epoch": 1.2488483666876924,
-      "grad_norm": 0.0007762082386761904,
-      "learning_rate": 0.00019999923129917447,
-      "loss": 46.0,
-      "step": 16334
-    },
-    {
-      "epoch": 1.2489248236710822,
-      "grad_norm": 0.0011478044325485826,
-      "learning_rate": 0.00019999923120499124,
-      "loss": 46.0,
-      "step": 16335
-    },
-    {
-      "epoch": 1.2490012806544717,
-      "grad_norm": 0.005378944333642721,
-      "learning_rate": 0.00019999923111080228,
-      "loss": 46.0,
-      "step": 16336
-    },
-    {
-      "epoch": 1.2490777376378615,
-      "grad_norm": 0.006752186920493841,
-      "learning_rate": 0.0001999992310166075,
-      "loss": 46.0,
-      "step": 16337
-    },
-    {
-      "epoch": 1.2491541946212512,
-      "grad_norm": 0.004907096736133099,
-      "learning_rate": 0.00019999923092240698,
-      "loss": 46.0,
-      "step": 16338
-    },
-    {
-      "epoch": 1.249230651604641,
-      "grad_norm": 0.0009439613786526024,
-      "learning_rate": 0.00019999923082820068,
-      "loss": 46.0,
-      "step": 16339
-    },
-    {
-      "epoch": 1.2493071085880307,
-      "grad_norm": 0.0007309425855055451,
-      "learning_rate": 0.00019999923073398863,
-      "loss": 46.0,
-      "step": 16340
-    },
-    {
-      "epoch": 1.2493835655714203,
-      "grad_norm": 0.0007836582371965051,
-      "learning_rate": 0.00019999923063977078,
-      "loss": 46.0,
-      "step": 16341
-    },
-    {
-      "epoch": 1.24946002255481,
-      "grad_norm": 0.001868736813776195,
-      "learning_rate": 0.00019999923054554722,
-      "loss": 46.0,
-      "step": 16342
-    },
-    {
-      "epoch": 1.2495364795381998,
-      "grad_norm": 0.0005757371545769274,
-      "learning_rate": 0.00019999923045131785,
-      "loss": 46.0,
-      "step": 16343
-    },
-    {
-      "epoch": 1.2496129365215896,
-      "grad_norm": 0.0006243189563974738,
-      "learning_rate": 0.00019999923035708271,
-      "loss": 46.0,
-      "step": 16344
-    },
-    {
-      "epoch": 1.2496893935049793,
-      "grad_norm": 0.02133634313941002,
-      "learning_rate": 0.0001999992302628418,
-      "loss": 46.0,
-      "step": 16345
-    },
-    {
-      "epoch": 1.249765850488369,
-      "grad_norm": 0.0007615424110554159,
-      "learning_rate": 0.00019999923016859511,
-      "loss": 46.0,
-      "step": 16346
-    },
-    {
-      "epoch": 1.2498423074717586,
-      "grad_norm": 0.0019910961855202913,
-      "learning_rate": 0.00019999923007434268,
-      "loss": 46.0,
-      "step": 16347
-    },
-    {
-      "epoch": 1.2499187644551484,
-      "grad_norm": 0.0017621485749259591,
-      "learning_rate": 0.00019999922998008445,
-      "loss": 46.0,
-      "step": 16348
-    },
-    {
-      "epoch": 1.2499952214385381,
-      "grad_norm": 0.0059636677615344524,
-      "learning_rate": 0.00019999922988582047,
-      "loss": 46.0,
-      "step": 16349
-    },
-    {
-      "epoch": 1.250071678421928,
-      "grad_norm": 0.002571824472397566,
-      "learning_rate": 0.00019999922979155072,
-      "loss": 46.0,
-      "step": 16350
-    },
-    {
-      "epoch": 1.2501481354053177,
-      "grad_norm": 0.0012950774980708957,
-      "learning_rate": 0.0001999992296972752,
-      "loss": 46.0,
-      "step": 16351
-    },
-    {
-      "epoch": 1.2502245923887072,
-      "grad_norm": 0.001501788618043065,
-      "learning_rate": 0.00019999922960299392,
-      "loss": 46.0,
-      "step": 16352
-    },
-    {
-      "epoch": 1.250301049372097,
-      "grad_norm": 0.0005776687758043408,
-      "learning_rate": 0.00019999922950870685,
-      "loss": 46.0,
-      "step": 16353
-    },
-    {
-      "epoch": 1.2503775063554867,
-      "grad_norm": 0.007493641227483749,
-      "learning_rate": 0.000199999229414414,
-      "loss": 46.0,
-      "step": 16354
-    },
-    {
-      "epoch": 1.2504539633388765,
-      "grad_norm": 0.001138480263762176,
-      "learning_rate": 0.00019999922932011538,
-      "loss": 46.0,
-      "step": 16355
-    },
-    {
-      "epoch": 1.2505304203222662,
-      "grad_norm": 0.0009020980796776712,
-      "learning_rate": 0.00019999922922581104,
-      "loss": 46.0,
-      "step": 16356
-    },
-    {
-      "epoch": 1.250606877305656,
-      "grad_norm": 0.002299938118085265,
-      "learning_rate": 0.00019999922913150088,
-      "loss": 46.0,
-      "step": 16357
-    },
-    {
-      "epoch": 1.2506833342890455,
-      "grad_norm": 0.0025025885552167892,
-      "learning_rate": 0.000199999229037185,
-      "loss": 46.0,
-      "step": 16358
-    },
-    {
-      "epoch": 1.2507597912724353,
-      "grad_norm": 0.0017581349238753319,
-      "learning_rate": 0.0001999992289428633,
-      "loss": 46.0,
-      "step": 16359
-    },
-    {
-      "epoch": 1.250836248255825,
-      "grad_norm": 0.0018445730675011873,
-      "learning_rate": 0.00019999922884853585,
-      "loss": 46.0,
-      "step": 16360
-    },
-    {
-      "epoch": 1.2509127052392148,
-      "grad_norm": 0.0010989966103807092,
-      "learning_rate": 0.00019999922875420264,
-      "loss": 46.0,
-      "step": 16361
-    },
-    {
-      "epoch": 1.2509891622226046,
-      "grad_norm": 0.0010360617889091372,
-      "learning_rate": 0.00019999922865986366,
-      "loss": 46.0,
-      "step": 16362
-    },
-    {
-      "epoch": 1.2510656192059941,
-      "grad_norm": 0.004308904986828566,
-      "learning_rate": 0.00019999922856551888,
-      "loss": 46.0,
-      "step": 16363
-    },
-    {
-      "epoch": 1.251142076189384,
-      "grad_norm": 0.0013923459919169545,
-      "learning_rate": 0.00019999922847116836,
-      "loss": 46.0,
-      "step": 16364
-    },
-    {
-      "epoch": 1.2512185331727737,
-      "grad_norm": 0.0007284162566065788,
-      "learning_rate": 0.00019999922837681208,
-      "loss": 46.0,
-      "step": 16365
-    },
-    {
-      "epoch": 1.2512949901561634,
-      "grad_norm": 0.0030336237978190184,
-      "learning_rate": 0.00019999922828245,
-      "loss": 46.0,
-      "step": 16366
-    },
-    {
-      "epoch": 1.2513714471395532,
-      "grad_norm": 0.0015833905199542642,
-      "learning_rate": 0.00019999922818808217,
-      "loss": 46.0,
-      "step": 16367
-    },
-    {
-      "epoch": 1.251447904122943,
-      "grad_norm": 0.00913267768919468,
-      "learning_rate": 0.00019999922809370857,
-      "loss": 46.0,
-      "step": 16368
-    },
-    {
-      "epoch": 1.2515243611063325,
-      "grad_norm": 0.0007435894221998751,
-      "learning_rate": 0.0001999992279993292,
-      "loss": 46.0,
-      "step": 16369
-    },
-    {
-      "epoch": 1.2516008180897222,
-      "grad_norm": 0.013490833342075348,
-      "learning_rate": 0.00019999922790494404,
-      "loss": 46.0,
-      "step": 16370
-    },
-    {
-      "epoch": 1.251677275073112,
-      "grad_norm": 0.000607972324360162,
-      "learning_rate": 0.00019999922781055313,
-      "loss": 46.0,
-      "step": 16371
-    },
-    {
-      "epoch": 1.2517537320565018,
-      "grad_norm": 0.002596711041405797,
-      "learning_rate": 0.00019999922771615647,
-      "loss": 46.0,
-      "step": 16372
-    },
-    {
-      "epoch": 1.2518301890398913,
-      "grad_norm": 0.0031988900154829025,
-      "learning_rate": 0.000199999227621754,
-      "loss": 46.0,
-      "step": 16373
-    },
-    {
-      "epoch": 1.251906646023281,
-      "grad_norm": 0.002775750122964382,
-      "learning_rate": 0.00019999922752734578,
-      "loss": 46.0,
-      "step": 16374
-    },
-    {
-      "epoch": 1.2519831030066708,
-      "grad_norm": 0.0031500759068876505,
-      "learning_rate": 0.0001999992274329318,
-      "loss": 46.0,
-      "step": 16375
-    },
-    {
-      "epoch": 1.2520595599900606,
-      "grad_norm": 0.005923932418227196,
-      "learning_rate": 0.00019999922733851205,
-      "loss": 46.0,
-      "step": 16376
-    },
-    {
-      "epoch": 1.2521360169734503,
-      "grad_norm": 0.0006573188002221286,
-      "learning_rate": 0.00019999922724408652,
-      "loss": 46.0,
-      "step": 16377
-    },
-    {
-      "epoch": 1.25221247395684,
-      "grad_norm": 0.0006369486800394952,
-      "learning_rate": 0.00019999922714965523,
-      "loss": 46.0,
-      "step": 16378
-    },
-    {
-      "epoch": 1.2522889309402299,
-      "grad_norm": 0.04541425406932831,
-      "learning_rate": 0.00019999922705521815,
-      "loss": 46.0,
-      "step": 16379
-    },
-    {
-      "epoch": 1.2523653879236194,
-      "grad_norm": 0.0007402290357276797,
-      "learning_rate": 0.0001999992269607753,
-      "loss": 46.0,
-      "step": 16380
-    },
-    {
-      "epoch": 1.2524418449070092,
-      "grad_norm": 0.0026812339201569557,
-      "learning_rate": 0.00019999922686632672,
-      "loss": 46.0,
-      "step": 16381
-    },
-    {
-      "epoch": 1.252518301890399,
-      "grad_norm": 0.004508398473262787,
-      "learning_rate": 0.00019999922677187232,
-      "loss": 46.0,
-      "step": 16382
-    },
-    {
-      "epoch": 1.2525947588737887,
-      "grad_norm": 0.0007145089912228286,
-      "learning_rate": 0.00019999922667741219,
-      "loss": 46.0,
-      "step": 16383
-    },
-    {
-      "epoch": 1.2526712158571782,
-      "grad_norm": 0.001860719989053905,
-      "learning_rate": 0.00019999922658294627,
-      "loss": 46.0,
-      "step": 16384
-    },
-    {
-      "epoch": 1.252747672840568,
-      "grad_norm": 0.001473874319344759,
-      "learning_rate": 0.0001999992264884746,
-      "loss": 46.0,
-      "step": 16385
-    },
-    {
-      "epoch": 1.2528241298239577,
-      "grad_norm": 0.0007973352330736816,
-      "learning_rate": 0.00019999922639399716,
-      "loss": 46.0,
-      "step": 16386
-    },
-    {
-      "epoch": 1.2529005868073475,
-      "grad_norm": 0.001900269417092204,
-      "learning_rate": 0.0001999992262995139,
-      "loss": 46.0,
-      "step": 16387
-    },
-    {
-      "epoch": 1.2529770437907373,
-      "grad_norm": 0.0016751679359003901,
-      "learning_rate": 0.00019999922620502495,
-      "loss": 46.0,
-      "step": 16388
-    },
-    {
-      "epoch": 1.253053500774127,
-      "grad_norm": 0.0024422432761639357,
-      "learning_rate": 0.00019999922611053017,
-      "loss": 46.0,
-      "step": 16389
-    },
-    {
-      "epoch": 1.2531299577575168,
-      "grad_norm": 0.0011167627526447177,
-      "learning_rate": 0.00019999922601602962,
-      "loss": 46.0,
-      "step": 16390
-    },
-    {
-      "epoch": 1.2532064147409063,
-      "grad_norm": 0.0005226141074672341,
-      "learning_rate": 0.00019999922592152332,
-      "loss": 46.0,
-      "step": 16391
-    },
-    {
-      "epoch": 1.253282871724296,
-      "grad_norm": 0.001044654636643827,
-      "learning_rate": 0.00019999922582701127,
-      "loss": 46.0,
-      "step": 16392
-    },
-    {
-      "epoch": 1.2533593287076858,
-      "grad_norm": 0.0019333363743498921,
-      "learning_rate": 0.0001999992257324934,
-      "loss": 46.0,
-      "step": 16393
-    },
-    {
-      "epoch": 1.2534357856910756,
-      "grad_norm": 0.0014803933445364237,
-      "learning_rate": 0.0001999992256379698,
-      "loss": 46.0,
-      "step": 16394
-    },
-    {
-      "epoch": 1.2535122426744651,
-      "grad_norm": 0.00046341324923560023,
-      "learning_rate": 0.00019999922554344044,
-      "loss": 46.0,
-      "step": 16395
-    },
-    {
-      "epoch": 1.253588699657855,
-      "grad_norm": 0.003914657514542341,
-      "learning_rate": 0.00019999922544890528,
-      "loss": 46.0,
-      "step": 16396
-    },
-    {
-      "epoch": 1.2536651566412447,
-      "grad_norm": 0.0016827434301376343,
-      "learning_rate": 0.00019999922535436437,
-      "loss": 46.0,
-      "step": 16397
-    },
-    {
-      "epoch": 1.2537416136246344,
-      "grad_norm": 0.0011291344417259097,
-      "learning_rate": 0.00019999922525981768,
-      "loss": 46.0,
-      "step": 16398
-    },
-    {
-      "epoch": 1.2538180706080242,
-      "grad_norm": 0.01156060490757227,
-      "learning_rate": 0.00019999922516526522,
-      "loss": 46.0,
-      "step": 16399
-    },
-    {
-      "epoch": 1.253894527591414,
-      "grad_norm": 0.0011152446968480945,
-      "learning_rate": 0.00019999922507070702,
-      "loss": 46.0,
-      "step": 16400
-    },
-    {
-      "epoch": 1.2539709845748037,
-      "grad_norm": 0.002592396456748247,
-      "learning_rate": 0.00019999922497614301,
-      "loss": 46.0,
-      "step": 16401
-    },
-    {
-      "epoch": 1.2540474415581933,
-      "grad_norm": 0.004796263761818409,
-      "learning_rate": 0.00019999922488157324,
-      "loss": 46.0,
-      "step": 16402
-    },
-    {
-      "epoch": 1.254123898541583,
-      "grad_norm": 0.0004613221681211144,
-      "learning_rate": 0.00019999922478699774,
-      "loss": 46.0,
-      "step": 16403
-    },
-    {
-      "epoch": 1.2542003555249728,
-      "grad_norm": 0.004102225415408611,
-      "learning_rate": 0.00019999922469241641,
-      "loss": 46.0,
-      "step": 16404
-    },
-    {
-      "epoch": 1.2542768125083625,
-      "grad_norm": 0.00032626657048240304,
-      "learning_rate": 0.00019999922459782934,
-      "loss": 46.0,
-      "step": 16405
-    },
-    {
-      "epoch": 1.254353269491752,
-      "grad_norm": 0.005635222885757685,
-      "learning_rate": 0.0001999992245032365,
-      "loss": 46.0,
-      "step": 16406
-    },
-    {
-      "epoch": 1.2544297264751418,
-      "grad_norm": 0.001193670672364533,
-      "learning_rate": 0.0001999992244086379,
-      "loss": 46.0,
-      "step": 16407
-    },
-    {
-      "epoch": 1.2545061834585316,
-      "grad_norm": 0.0007480288622900844,
-      "learning_rate": 0.00019999922431403351,
-      "loss": 46.0,
-      "step": 16408
-    },
-    {
-      "epoch": 1.2545826404419214,
-      "grad_norm": 0.0004027009126730263,
-      "learning_rate": 0.00019999922421942338,
-      "loss": 46.0,
-      "step": 16409
-    },
-    {
-      "epoch": 1.2546590974253111,
-      "grad_norm": 0.0007107963901944458,
-      "learning_rate": 0.00019999922412480744,
-      "loss": 46.0,
-      "step": 16410
-    },
-    {
-      "epoch": 1.2547355544087009,
-      "grad_norm": 0.0015318809309974313,
-      "learning_rate": 0.00019999922403018573,
-      "loss": 46.0,
-      "step": 16411
-    },
-    {
-      "epoch": 1.2548120113920906,
-      "grad_norm": 0.0008197858696803451,
-      "learning_rate": 0.0001999992239355583,
-      "loss": 46.0,
-      "step": 16412
-    },
-    {
-      "epoch": 1.2548884683754802,
-      "grad_norm": 0.0007833583513274789,
-      "learning_rate": 0.00019999922384092506,
-      "loss": 46.0,
-      "step": 16413
-    },
-    {
-      "epoch": 1.25496492535887,
-      "grad_norm": 0.0025514462031424046,
-      "learning_rate": 0.0001999992237462861,
-      "loss": 46.0,
-      "step": 16414
-    },
-    {
-      "epoch": 1.2550413823422597,
-      "grad_norm": 0.0016124843386933208,
-      "learning_rate": 0.0001999992236516413,
-      "loss": 46.0,
-      "step": 16415
-    },
-    {
-      "epoch": 1.2551178393256495,
-      "grad_norm": 0.00036243052454665303,
-      "learning_rate": 0.00019999922355699076,
-      "loss": 46.0,
-      "step": 16416
-    },
-    {
-      "epoch": 1.255194296309039,
-      "grad_norm": 0.0009067085920833051,
-      "learning_rate": 0.00019999922346233446,
-      "loss": 46.0,
-      "step": 16417
-    },
-    {
-      "epoch": 1.2552707532924288,
-      "grad_norm": 0.000285412126686424,
-      "learning_rate": 0.00019999922336767242,
-      "loss": 46.0,
-      "step": 16418
-    },
-    {
-      "epoch": 1.2553472102758185,
-      "grad_norm": 0.0016339175635948777,
-      "learning_rate": 0.00019999922327300454,
-      "loss": 46.0,
-      "step": 16419
-    },
-    {
-      "epoch": 1.2554236672592083,
-      "grad_norm": 0.001466345856897533,
-      "learning_rate": 0.00019999922317833093,
-      "loss": 46.0,
-      "step": 16420
-    },
-    {
-      "epoch": 1.255500124242598,
-      "grad_norm": 0.005923457909375429,
-      "learning_rate": 0.00019999922308365156,
-      "loss": 46.0,
-      "step": 16421
-    },
-    {
-      "epoch": 1.2555765812259878,
-      "grad_norm": 0.0009513382101431489,
-      "learning_rate": 0.0001999992229889664,
-      "loss": 46.0,
-      "step": 16422
-    },
-    {
-      "epoch": 1.2556530382093776,
-      "grad_norm": 0.0007420122274197638,
-      "learning_rate": 0.00019999922289427549,
-      "loss": 46.0,
-      "step": 16423
-    },
-    {
-      "epoch": 1.255729495192767,
-      "grad_norm": 0.00023287672956939787,
-      "learning_rate": 0.0001999992227995788,
-      "loss": 46.0,
-      "step": 16424
-    },
-    {
-      "epoch": 1.2558059521761569,
-      "grad_norm": 0.01207447238266468,
-      "learning_rate": 0.00019999922270487632,
-      "loss": 46.0,
-      "step": 16425
-    },
-    {
-      "epoch": 1.2558824091595466,
-      "grad_norm": 0.0009858588455244899,
-      "learning_rate": 0.0001999992226101681,
-      "loss": 46.0,
-      "step": 16426
-    },
-    {
-      "epoch": 1.2559588661429364,
-      "grad_norm": 0.0011797369224950671,
-      "learning_rate": 0.0001999992225154541,
-      "loss": 46.0,
-      "step": 16427
-    },
-    {
-      "epoch": 1.256035323126326,
-      "grad_norm": 0.0026717614382505417,
-      "learning_rate": 0.00019999922242073433,
-      "loss": 46.0,
-      "step": 16428
-    },
-    {
-      "epoch": 1.2561117801097157,
-      "grad_norm": 0.0006869658827781677,
-      "learning_rate": 0.0001999992223260088,
-      "loss": 46.0,
-      "step": 16429
-    },
-    {
-      "epoch": 1.2561882370931055,
-      "grad_norm": 0.0010481844656169415,
-      "learning_rate": 0.00019999922223127748,
-      "loss": 46.0,
-      "step": 16430
-    },
-    {
-      "epoch": 1.2562646940764952,
-      "grad_norm": 0.000877156388014555,
-      "learning_rate": 0.0001999992221365404,
-      "loss": 46.0,
-      "step": 16431
-    },
-    {
-      "epoch": 1.256341151059885,
-      "grad_norm": 0.004499401897192001,
-      "learning_rate": 0.00019999922204179757,
-      "loss": 46.0,
-      "step": 16432
-    },
-    {
-      "epoch": 1.2564176080432747,
-      "grad_norm": 0.0030122101306915283,
-      "learning_rate": 0.00019999922194704898,
-      "loss": 46.0,
-      "step": 16433
-    },
-    {
-      "epoch": 1.2564940650266645,
-      "grad_norm": 0.00827631913125515,
-      "learning_rate": 0.00019999922185229456,
-      "loss": 46.0,
-      "step": 16434
-    },
-    {
-      "epoch": 1.256570522010054,
-      "grad_norm": 0.0010939062340185046,
-      "learning_rate": 0.00019999922175753442,
-      "loss": 46.0,
-      "step": 16435
-    },
-    {
-      "epoch": 1.2566469789934438,
-      "grad_norm": 0.002193512162193656,
-      "learning_rate": 0.0001999992216627685,
-      "loss": 46.0,
-      "step": 16436
-    },
-    {
-      "epoch": 1.2567234359768336,
-      "grad_norm": 0.0020986529998481274,
-      "learning_rate": 0.0001999992215679968,
-      "loss": 46.0,
-      "step": 16437
-    },
-    {
-      "epoch": 1.2567998929602233,
-      "grad_norm": 0.0006867117481306195,
-      "learning_rate": 0.00019999922147321932,
-      "loss": 46.0,
-      "step": 16438
-    },
-    {
-      "epoch": 1.2568763499436129,
-      "grad_norm": 0.015324806794524193,
-      "learning_rate": 0.00019999922137843608,
-      "loss": 46.0,
-      "step": 16439
-    },
-    {
-      "epoch": 1.2569528069270026,
-      "grad_norm": 0.0013696304522454739,
-      "learning_rate": 0.0001999992212836471,
-      "loss": 46.0,
-      "step": 16440
-    },
-    {
-      "epoch": 1.2570292639103924,
-      "grad_norm": 0.003059870097786188,
-      "learning_rate": 0.00019999922118885233,
-      "loss": 46.0,
-      "step": 16441
-    },
-    {
-      "epoch": 1.2571057208937821,
-      "grad_norm": 0.0016380450688302517,
-      "learning_rate": 0.00019999922109405178,
-      "loss": 46.0,
-      "step": 16442
-    },
-    {
-      "epoch": 1.257182177877172,
-      "grad_norm": 0.0006603061337955296,
-      "learning_rate": 0.00019999922099924548,
-      "loss": 46.0,
-      "step": 16443
-    },
-    {
-      "epoch": 1.2572586348605617,
-      "grad_norm": 0.003353565465658903,
-      "learning_rate": 0.0001999992209044334,
-      "loss": 46.0,
-      "step": 16444
-    },
-    {
-      "epoch": 1.2573350918439514,
-      "grad_norm": 0.0006715510971844196,
-      "learning_rate": 0.00019999922080961554,
-      "loss": 46.0,
-      "step": 16445
-    },
-    {
-      "epoch": 1.257411548827341,
-      "grad_norm": 0.0006737725343555212,
-      "learning_rate": 0.00019999922071479192,
-      "loss": 46.0,
-      "step": 16446
-    },
-    {
-      "epoch": 1.2574880058107307,
-      "grad_norm": 0.001551430788822472,
-      "learning_rate": 0.00019999922061996256,
-      "loss": 46.0,
-      "step": 16447
-    },
-    {
-      "epoch": 1.2575644627941205,
-      "grad_norm": 0.00261330115608871,
-      "learning_rate": 0.0001999992205251274,
-      "loss": 46.0,
-      "step": 16448
-    },
-    {
-      "epoch": 1.2576409197775102,
-      "grad_norm": 0.0008056391961872578,
-      "learning_rate": 0.00019999922043028646,
-      "loss": 46.0,
-      "step": 16449
-    },
-    {
-      "epoch": 1.2577173767608998,
-      "grad_norm": 0.013351313769817352,
-      "learning_rate": 0.00019999922033543977,
-      "loss": 46.0,
-      "step": 16450
-    },
-    {
-      "epoch": 1.2577938337442895,
-      "grad_norm": 0.001033368636853993,
-      "learning_rate": 0.00019999922024058732,
-      "loss": 46.0,
-      "step": 16451
-    },
-    {
-      "epoch": 1.2578702907276793,
-      "grad_norm": 0.002594680991023779,
-      "learning_rate": 0.00019999922014572906,
-      "loss": 46.0,
-      "step": 16452
-    },
-    {
-      "epoch": 1.257946747711069,
-      "grad_norm": 0.001639623544178903,
-      "learning_rate": 0.00019999922005086506,
-      "loss": 46.0,
-      "step": 16453
-    },
-    {
-      "epoch": 1.2580232046944588,
-      "grad_norm": 0.0030052606016397476,
-      "learning_rate": 0.00019999921995599528,
-      "loss": 46.0,
-      "step": 16454
-    },
-    {
-      "epoch": 1.2580996616778486,
-      "grad_norm": 0.0014609363861382008,
-      "learning_rate": 0.00019999921986111976,
-      "loss": 46.0,
-      "step": 16455
-    },
-    {
-      "epoch": 1.2581761186612384,
-      "grad_norm": 0.0023911858443170786,
-      "learning_rate": 0.00019999921976623846,
-      "loss": 46.0,
-      "step": 16456
-    },
-    {
-      "epoch": 1.258252575644628,
-      "grad_norm": 0.0013884411891922355,
-      "learning_rate": 0.00019999921967135136,
-      "loss": 46.0,
-      "step": 16457
-    },
-    {
-      "epoch": 1.2583290326280177,
-      "grad_norm": 0.0007670379709452391,
-      "learning_rate": 0.00019999921957645852,
-      "loss": 46.0,
-      "step": 16458
-    },
-    {
-      "epoch": 1.2584054896114074,
-      "grad_norm": 0.001012188266031444,
-      "learning_rate": 0.00019999921948155987,
-      "loss": 46.0,
-      "step": 16459
-    },
-    {
-      "epoch": 1.2584819465947972,
-      "grad_norm": 0.0008938998798839748,
-      "learning_rate": 0.0001999992193866555,
-      "loss": 46.0,
-      "step": 16460
-    },
-    {
-      "epoch": 1.2585584035781867,
-      "grad_norm": 0.004603705368936062,
-      "learning_rate": 0.00019999921929174535,
-      "loss": 46.0,
-      "step": 16461
-    },
-    {
-      "epoch": 1.2586348605615765,
-      "grad_norm": 0.0006540778558701277,
-      "learning_rate": 0.0001999992191968294,
-      "loss": 46.0,
-      "step": 16462
-    },
-    {
-      "epoch": 1.2587113175449662,
-      "grad_norm": 0.0017041172832250595,
-      "learning_rate": 0.00019999921910190773,
-      "loss": 46.0,
-      "step": 16463
-    },
-    {
-      "epoch": 1.258787774528356,
-      "grad_norm": 0.001365963602438569,
-      "learning_rate": 0.00019999921900698024,
-      "loss": 46.0,
-      "step": 16464
-    },
-    {
-      "epoch": 1.2588642315117458,
-      "grad_norm": 0.0024636252783238888,
-      "learning_rate": 0.00019999921891204704,
-      "loss": 46.0,
-      "step": 16465
-    },
-    {
-      "epoch": 1.2589406884951355,
-      "grad_norm": 0.0012807559687644243,
-      "learning_rate": 0.00019999921881710798,
-      "loss": 46.0,
-      "step": 16466
-    },
-    {
-      "epoch": 1.2590171454785253,
-      "grad_norm": 0.0008748608524911106,
-      "learning_rate": 0.00019999921872216323,
-      "loss": 46.0,
-      "step": 16467
-    },
-    {
-      "epoch": 1.2590936024619148,
-      "grad_norm": 0.003088687313720584,
-      "learning_rate": 0.00019999921862721268,
-      "loss": 46.0,
-      "step": 16468
-    },
-    {
-      "epoch": 1.2591700594453046,
-      "grad_norm": 0.0011356067843735218,
-      "learning_rate": 0.00019999921853225639,
-      "loss": 46.0,
-      "step": 16469
-    },
-    {
-      "epoch": 1.2592465164286943,
-      "grad_norm": 0.0016523645026609302,
-      "learning_rate": 0.0001999992184372943,
-      "loss": 46.0,
-      "step": 16470
-    },
-    {
-      "epoch": 1.259322973412084,
-      "grad_norm": 0.0014879752416163683,
-      "learning_rate": 0.00019999921834232644,
-      "loss": 46.0,
-      "step": 16471
-    },
-    {
-      "epoch": 1.2593994303954736,
-      "grad_norm": 0.0013844635104760528,
-      "learning_rate": 0.00019999921824735283,
-      "loss": 46.0,
-      "step": 16472
-    },
-    {
-      "epoch": 1.2594758873788634,
-      "grad_norm": 0.00357668474316597,
-      "learning_rate": 0.0001999992181523734,
-      "loss": 46.0,
-      "step": 16473
-    },
-    {
-      "epoch": 1.2595523443622532,
-      "grad_norm": 0.001873683650046587,
-      "learning_rate": 0.00019999921805738825,
-      "loss": 46.0,
-      "step": 16474
-    },
-    {
-      "epoch": 1.259628801345643,
-      "grad_norm": 0.022805245593190193,
-      "learning_rate": 0.00019999921796239734,
-      "loss": 46.0,
-      "step": 16475
-    },
-    {
-      "epoch": 1.2597052583290327,
-      "grad_norm": 0.0007727384218014777,
-      "learning_rate": 0.00019999921786740063,
-      "loss": 46.0,
-      "step": 16476
-    },
-    {
-      "epoch": 1.2597817153124224,
-      "grad_norm": 0.0037208418361842632,
-      "learning_rate": 0.00019999921777239817,
-      "loss": 46.0,
-      "step": 16477
-    },
-    {
-      "epoch": 1.2598581722958122,
-      "grad_norm": 0.0011424771510064602,
-      "learning_rate": 0.00019999921767738994,
-      "loss": 46.0,
-      "step": 16478
-    },
-    {
-      "epoch": 1.2599346292792017,
-      "grad_norm": 0.0008388548740185797,
-      "learning_rate": 0.0001999992175823759,
-      "loss": 46.0,
-      "step": 16479
-    },
-    {
-      "epoch": 1.2600110862625915,
-      "grad_norm": 0.0014203950995579362,
-      "learning_rate": 0.00019999921748735613,
-      "loss": 46.0,
-      "step": 16480
-    },
-    {
-      "epoch": 1.2600875432459813,
-      "grad_norm": 0.000518681132234633,
-      "learning_rate": 0.00019999921739233058,
-      "loss": 46.0,
-      "step": 16481
-    },
-    {
-      "epoch": 1.260164000229371,
-      "grad_norm": 0.007198650855571032,
-      "learning_rate": 0.00019999921729729926,
-      "loss": 46.0,
-      "step": 16482
-    },
-    {
-      "epoch": 1.2602404572127606,
-      "grad_norm": 0.0006486611091531813,
-      "learning_rate": 0.0001999992172022622,
-      "loss": 46.0,
-      "step": 16483
-    },
-    {
-      "epoch": 1.2603169141961503,
-      "grad_norm": 0.011363430880010128,
-      "learning_rate": 0.00019999921710721935,
-      "loss": 46.0,
-      "step": 16484
-    },
-    {
-      "epoch": 1.26039337117954,
-      "grad_norm": 0.00170981977134943,
-      "learning_rate": 0.00019999921701217073,
-      "loss": 46.0,
-      "step": 16485
-    },
-    {
-      "epoch": 1.2604698281629299,
-      "grad_norm": 0.004637775011360645,
-      "learning_rate": 0.00019999921691711632,
-      "loss": 46.0,
-      "step": 16486
-    },
-    {
-      "epoch": 1.2605462851463196,
-      "grad_norm": 0.003388921497389674,
-      "learning_rate": 0.00019999921682205615,
-      "loss": 46.0,
-      "step": 16487
-    },
-    {
-      "epoch": 1.2606227421297094,
-      "grad_norm": 0.0015615829033777118,
-      "learning_rate": 0.00019999921672699022,
-      "loss": 46.0,
-      "step": 16488
-    },
-    {
-      "epoch": 1.260699199113099,
-      "grad_norm": 0.0005641995812766254,
-      "learning_rate": 0.0001999992166319185,
-      "loss": 46.0,
-      "step": 16489
-    },
-    {
-      "epoch": 1.2607756560964887,
-      "grad_norm": 0.0007029682747088373,
-      "learning_rate": 0.00019999921653684105,
-      "loss": 46.0,
-      "step": 16490
-    },
-    {
-      "epoch": 1.2608521130798784,
-      "grad_norm": 0.0007309545180760324,
-      "learning_rate": 0.00019999921644175782,
-      "loss": 46.0,
-      "step": 16491
-    },
-    {
-      "epoch": 1.2609285700632682,
-      "grad_norm": 0.000503484916407615,
-      "learning_rate": 0.00019999921634666882,
-      "loss": 46.0,
-      "step": 16492
-    },
-    {
-      "epoch": 1.261005027046658,
-      "grad_norm": 0.0013351273955777287,
-      "learning_rate": 0.00019999921625157404,
-      "loss": 46.0,
-      "step": 16493
-    },
-    {
-      "epoch": 1.2610814840300475,
-      "grad_norm": 0.0012642602669075131,
-      "learning_rate": 0.0001999992161564735,
-      "loss": 46.0,
-      "step": 16494
-    },
-    {
-      "epoch": 1.2611579410134373,
-      "grad_norm": 0.002974753500893712,
-      "learning_rate": 0.00019999921606136717,
-      "loss": 46.0,
-      "step": 16495
-    },
-    {
-      "epoch": 1.261234397996827,
-      "grad_norm": 0.0011573872761800885,
-      "learning_rate": 0.00019999921596625507,
-      "loss": 46.0,
-      "step": 16496
-    },
-    {
-      "epoch": 1.2613108549802168,
-      "grad_norm": 0.027800822630524635,
-      "learning_rate": 0.0001999992158711372,
-      "loss": 46.0,
-      "step": 16497
-    },
-    {
-      "epoch": 1.2613873119636065,
-      "grad_norm": 0.001148067880421877,
-      "learning_rate": 0.00019999921577601356,
-      "loss": 46.0,
-      "step": 16498
-    },
-    {
-      "epoch": 1.2614637689469963,
-      "grad_norm": 0.0018769362941384315,
-      "learning_rate": 0.00019999921568088417,
-      "loss": 46.0,
-      "step": 16499
-    },
-    {
-      "epoch": 1.2615402259303858,
-      "grad_norm": 0.0010231048800051212,
-      "learning_rate": 0.000199999215585749,
-      "loss": 46.0,
-      "step": 16500
-    },
-    {
-      "epoch": 1.2616166829137756,
-      "grad_norm": 0.002108842832967639,
-      "learning_rate": 0.00019999921549060807,
-      "loss": 46.0,
-      "step": 16501
-    },
-    {
-      "epoch": 1.2616931398971654,
-      "grad_norm": 0.001610041013918817,
-      "learning_rate": 0.0001999992153954614,
-      "loss": 46.0,
-      "step": 16502
-    },
-    {
-      "epoch": 1.2617695968805551,
-      "grad_norm": 0.0017317309975624084,
-      "learning_rate": 0.0001999992153003089,
-      "loss": 46.0,
-      "step": 16503
-    },
-    {
-      "epoch": 1.2618460538639447,
-      "grad_norm": 0.007911157794296741,
-      "learning_rate": 0.00019999921520515065,
-      "loss": 46.0,
-      "step": 16504
-    },
-    {
-      "epoch": 1.2619225108473344,
-      "grad_norm": 0.0010015369625762105,
-      "learning_rate": 0.00019999921510998662,
-      "loss": 46.0,
-      "step": 16505
-    },
-    {
-      "epoch": 1.2619989678307242,
-      "grad_norm": 0.0024152202531695366,
-      "learning_rate": 0.00019999921501481687,
-      "loss": 46.0,
-      "step": 16506
-    },
-    {
-      "epoch": 1.262075424814114,
-      "grad_norm": 0.009238668717443943,
-      "learning_rate": 0.00019999921491964132,
-      "loss": 46.0,
-      "step": 16507
-    },
-    {
-      "epoch": 1.2621518817975037,
-      "grad_norm": 0.002034083940088749,
-      "learning_rate": 0.00019999921482446,
-      "loss": 46.0,
-      "step": 16508
-    },
-    {
-      "epoch": 1.2622283387808935,
-      "grad_norm": 0.0010638341773301363,
-      "learning_rate": 0.0001999992147292729,
-      "loss": 46.0,
-      "step": 16509
-    },
-    {
-      "epoch": 1.2623047957642832,
-      "grad_norm": 0.0013406927464529872,
-      "learning_rate": 0.00019999921463408003,
-      "loss": 46.0,
-      "step": 16510
-    },
-    {
-      "epoch": 1.2623812527476728,
-      "grad_norm": 0.001091621583327651,
-      "learning_rate": 0.00019999921453888142,
-      "loss": 46.0,
-      "step": 16511
-    },
-    {
-      "epoch": 1.2624577097310625,
-      "grad_norm": 0.0014226384228095412,
-      "learning_rate": 0.000199999214443677,
-      "loss": 46.0,
-      "step": 16512
-    },
-    {
-      "epoch": 1.2625341667144523,
-      "grad_norm": 0.007693960331380367,
-      "learning_rate": 0.00019999921434846684,
-      "loss": 46.0,
-      "step": 16513
-    },
-    {
-      "epoch": 1.262610623697842,
-      "grad_norm": 0.0028738912660628557,
-      "learning_rate": 0.0001999992142532509,
-      "loss": 46.0,
-      "step": 16514
-    },
-    {
-      "epoch": 1.2626870806812316,
-      "grad_norm": 0.0008677901932969689,
-      "learning_rate": 0.0001999992141580292,
-      "loss": 46.0,
-      "step": 16515
-    },
-    {
-      "epoch": 1.2627635376646213,
-      "grad_norm": 0.0029121434781700373,
-      "learning_rate": 0.0001999992140628017,
-      "loss": 46.0,
-      "step": 16516
-    },
-    {
-      "epoch": 1.262839994648011,
-      "grad_norm": 0.005586137063801289,
-      "learning_rate": 0.00019999921396756846,
-      "loss": 46.0,
-      "step": 16517
-    },
-    {
-      "epoch": 1.2629164516314009,
-      "grad_norm": 0.0021744370460510254,
-      "learning_rate": 0.00019999921387232945,
-      "loss": 46.0,
-      "step": 16518
-    },
-    {
-      "epoch": 1.2629929086147906,
-      "grad_norm": 0.0011308243265375495,
-      "learning_rate": 0.00019999921377708468,
-      "loss": 46.0,
-      "step": 16519
-    },
-    {
-      "epoch": 1.2630693655981804,
-      "grad_norm": 0.002094999188557267,
-      "learning_rate": 0.00019999921368183413,
-      "loss": 46.0,
-      "step": 16520
-    },
-    {
-      "epoch": 1.2631458225815702,
-      "grad_norm": 0.0010615659411996603,
-      "learning_rate": 0.0001999992135865778,
-      "loss": 46.0,
-      "step": 16521
-    },
-    {
-      "epoch": 1.2632222795649597,
-      "grad_norm": 0.0029781595803797245,
-      "learning_rate": 0.00019999921349131568,
-      "loss": 46.0,
-      "step": 16522
-    },
-    {
-      "epoch": 1.2632987365483495,
-      "grad_norm": 0.0010432334383949637,
-      "learning_rate": 0.00019999921339604787,
-      "loss": 46.0,
-      "step": 16523
-    },
-    {
-      "epoch": 1.2633751935317392,
-      "grad_norm": 0.0023265474010258913,
-      "learning_rate": 0.0001999992133007742,
-      "loss": 46.0,
-      "step": 16524
-    },
-    {
-      "epoch": 1.263451650515129,
-      "grad_norm": 0.001020620809867978,
-      "learning_rate": 0.0001999992132054948,
-      "loss": 46.0,
-      "step": 16525
-    },
-    {
-      "epoch": 1.2635281074985185,
-      "grad_norm": 0.0011690454557538033,
-      "learning_rate": 0.00019999921311020965,
-      "loss": 46.0,
-      "step": 16526
-    },
-    {
-      "epoch": 1.2636045644819083,
-      "grad_norm": 0.0007669496117159724,
-      "learning_rate": 0.0001999992130149187,
-      "loss": 46.0,
-      "step": 16527
-    },
-    {
-      "epoch": 1.263681021465298,
-      "grad_norm": 0.0033987811766564846,
-      "learning_rate": 0.00019999921291962198,
-      "loss": 46.0,
-      "step": 16528
-    },
-    {
-      "epoch": 1.2637574784486878,
-      "grad_norm": 0.0017866194248199463,
-      "learning_rate": 0.0001999992128243195,
-      "loss": 46.0,
-      "step": 16529
-    },
-    {
-      "epoch": 1.2638339354320776,
-      "grad_norm": 0.002379220211878419,
-      "learning_rate": 0.00019999921272901127,
-      "loss": 46.0,
-      "step": 16530
-    },
-    {
-      "epoch": 1.2639103924154673,
-      "grad_norm": 0.0007606594008393586,
-      "learning_rate": 0.00019999921263369726,
-      "loss": 46.0,
-      "step": 16531
-    },
-    {
-      "epoch": 1.263986849398857,
-      "grad_norm": 0.0007884561782702804,
-      "learning_rate": 0.0001999992125383775,
-      "loss": 46.0,
-      "step": 16532
-    },
-    {
-      "epoch": 1.2640633063822466,
-      "grad_norm": 0.001040615257807076,
-      "learning_rate": 0.0001999992124430519,
-      "loss": 46.0,
-      "step": 16533
-    },
-    {
-      "epoch": 1.2641397633656364,
-      "grad_norm": 0.0005108760087750852,
-      "learning_rate": 0.00019999921234772056,
-      "loss": 46.0,
-      "step": 16534
-    },
-    {
-      "epoch": 1.2642162203490261,
-      "grad_norm": 0.0006389761110767722,
-      "learning_rate": 0.0001999992122523835,
-      "loss": 46.0,
-      "step": 16535
-    },
-    {
-      "epoch": 1.264292677332416,
-      "grad_norm": 0.0016258525429293513,
-      "learning_rate": 0.0001999992121570406,
-      "loss": 46.0,
-      "step": 16536
-    },
-    {
-      "epoch": 1.2643691343158054,
-      "grad_norm": 0.0044453563168644905,
-      "learning_rate": 0.00019999921206169199,
-      "loss": 46.0,
-      "step": 16537
-    },
-    {
-      "epoch": 1.2644455912991952,
-      "grad_norm": 0.0009650862193666399,
-      "learning_rate": 0.0001999992119663376,
-      "loss": 46.0,
-      "step": 16538
-    },
-    {
-      "epoch": 1.264522048282585,
-      "grad_norm": 0.0007522368105128407,
-      "learning_rate": 0.0001999992118709774,
-      "loss": 46.0,
-      "step": 16539
-    },
-    {
-      "epoch": 1.2645985052659747,
-      "grad_norm": 0.0012720072409138083,
-      "learning_rate": 0.00019999921177561147,
-      "loss": 46.0,
-      "step": 16540
-    },
-    {
-      "epoch": 1.2646749622493645,
-      "grad_norm": 0.001897201407700777,
-      "learning_rate": 0.00019999921168023976,
-      "loss": 46.0,
-      "step": 16541
-    },
-    {
-      "epoch": 1.2647514192327542,
-      "grad_norm": 0.0023809215053915977,
-      "learning_rate": 0.00019999921158486228,
-      "loss": 46.0,
-      "step": 16542
-    },
-    {
-      "epoch": 1.264827876216144,
-      "grad_norm": 0.0017961254343390465,
-      "learning_rate": 0.00019999921148947905,
-      "loss": 46.0,
-      "step": 16543
-    },
-    {
-      "epoch": 1.2649043331995335,
-      "grad_norm": 0.0012066778726875782,
-      "learning_rate": 0.00019999921139409003,
-      "loss": 46.0,
-      "step": 16544
-    },
-    {
-      "epoch": 1.2649807901829233,
-      "grad_norm": 0.003078483510762453,
-      "learning_rate": 0.00019999921129869522,
-      "loss": 46.0,
-      "step": 16545
-    },
-    {
-      "epoch": 1.265057247166313,
-      "grad_norm": 0.010181276127696037,
-      "learning_rate": 0.00019999921120329465,
-      "loss": 46.0,
-      "step": 16546
-    },
-    {
-      "epoch": 1.2651337041497028,
-      "grad_norm": 0.007794594392180443,
-      "learning_rate": 0.00019999921110788833,
-      "loss": 46.0,
-      "step": 16547
-    },
-    {
-      "epoch": 1.2652101611330924,
-      "grad_norm": 0.0167773999273777,
-      "learning_rate": 0.00019999921101247626,
-      "loss": 46.0,
-      "step": 16548
-    },
-    {
-      "epoch": 1.2652866181164821,
-      "grad_norm": 0.0004972476162947714,
-      "learning_rate": 0.00019999921091705836,
-      "loss": 46.0,
-      "step": 16549
-    },
-    {
-      "epoch": 1.265363075099872,
-      "grad_norm": 0.001543243881314993,
-      "learning_rate": 0.00019999921082163472,
-      "loss": 46.0,
-      "step": 16550
-    },
-    {
-      "epoch": 1.2654395320832617,
-      "grad_norm": 0.0011297330493107438,
-      "learning_rate": 0.00019999921072620533,
-      "loss": 46.0,
-      "step": 16551
-    },
-    {
-      "epoch": 1.2655159890666514,
-      "grad_norm": 0.001778118428774178,
-      "learning_rate": 0.00019999921063077017,
-      "loss": 46.0,
-      "step": 16552
-    },
-    {
-      "epoch": 1.2655924460500412,
-      "grad_norm": 0.003555284347385168,
-      "learning_rate": 0.0001999992105353292,
-      "loss": 46.0,
-      "step": 16553
-    },
-    {
-      "epoch": 1.265668903033431,
-      "grad_norm": 0.0029371276032179594,
-      "learning_rate": 0.0001999992104398825,
-      "loss": 46.0,
-      "step": 16554
-    },
-    {
-      "epoch": 1.2657453600168205,
-      "grad_norm": 0.003292506095021963,
-      "learning_rate": 0.00019999921034443002,
-      "loss": 46.0,
-      "step": 16555
-    },
-    {
-      "epoch": 1.2658218170002102,
-      "grad_norm": 0.0027380939573049545,
-      "learning_rate": 0.00019999921024897174,
-      "loss": 46.0,
-      "step": 16556
-    },
-    {
-      "epoch": 1.2658982739836,
-      "grad_norm": 0.0006956110009923577,
-      "learning_rate": 0.00019999921015350774,
-      "loss": 46.0,
-      "step": 16557
-    },
-    {
-      "epoch": 1.2659747309669898,
-      "grad_norm": 0.002119495067745447,
-      "learning_rate": 0.00019999921005803793,
-      "loss": 46.0,
-      "step": 16558
-    },
-    {
-      "epoch": 1.2660511879503793,
-      "grad_norm": 0.003707997966557741,
-      "learning_rate": 0.00019999920996256239,
-      "loss": 46.0,
-      "step": 16559
-    },
-    {
-      "epoch": 1.266127644933769,
-      "grad_norm": 0.0006579714827239513,
-      "learning_rate": 0.00019999920986708104,
-      "loss": 46.0,
-      "step": 16560
-    },
-    {
-      "epoch": 1.2662041019171588,
-      "grad_norm": 0.0011696142610162497,
-      "learning_rate": 0.00019999920977159397,
-      "loss": 46.0,
-      "step": 16561
-    },
-    {
-      "epoch": 1.2662805589005486,
-      "grad_norm": 0.0007162902620621026,
-      "learning_rate": 0.00019999920967610107,
-      "loss": 46.0,
-      "step": 16562
-    },
-    {
-      "epoch": 1.2663570158839383,
-      "grad_norm": 0.0015866734320297837,
-      "learning_rate": 0.00019999920958060243,
-      "loss": 46.0,
-      "step": 16563
-    },
-    {
-      "epoch": 1.266433472867328,
-      "grad_norm": 0.0010536729823797941,
-      "learning_rate": 0.00019999920948509802,
-      "loss": 46.0,
-      "step": 16564
-    },
-    {
-      "epoch": 1.2665099298507179,
-      "grad_norm": 0.003634421853348613,
-      "learning_rate": 0.00019999920938958786,
-      "loss": 46.0,
-      "step": 16565
-    },
-    {
-      "epoch": 1.2665863868341074,
-      "grad_norm": 0.00461319275200367,
-      "learning_rate": 0.0001999992092940719,
-      "loss": 46.0,
-      "step": 16566
-    },
-    {
-      "epoch": 1.2666628438174972,
-      "grad_norm": 0.0012473630486056209,
-      "learning_rate": 0.0001999992091985502,
-      "loss": 46.0,
-      "step": 16567
-    },
-    {
-      "epoch": 1.266739300800887,
-      "grad_norm": 0.007341197691857815,
-      "learning_rate": 0.00019999920910302273,
-      "loss": 46.0,
-      "step": 16568
-    },
-    {
-      "epoch": 1.2668157577842767,
-      "grad_norm": 0.0007693300140090287,
-      "learning_rate": 0.00019999920900748945,
-      "loss": 46.0,
-      "step": 16569
-    },
-    {
-      "epoch": 1.2668922147676662,
-      "grad_norm": 0.002891362411901355,
-      "learning_rate": 0.00019999920891195042,
-      "loss": 46.0,
-      "step": 16570
-    },
-    {
-      "epoch": 1.266968671751056,
-      "grad_norm": 0.0004631927004083991,
-      "learning_rate": 0.00019999920881640565,
-      "loss": 46.0,
-      "step": 16571
-    },
-    {
-      "epoch": 1.2670451287344457,
-      "grad_norm": 0.0014737516175955534,
-      "learning_rate": 0.0001999992087208551,
-      "loss": 46.0,
-      "step": 16572
-    },
-    {
-      "epoch": 1.2671215857178355,
-      "grad_norm": 0.007577573414891958,
-      "learning_rate": 0.00019999920862529873,
-      "loss": 46.0,
-      "step": 16573
-    },
-    {
-      "epoch": 1.2671980427012253,
-      "grad_norm": 0.00051100057316944,
-      "learning_rate": 0.00019999920852973663,
-      "loss": 46.0,
-      "step": 16574
-    },
-    {
-      "epoch": 1.267274499684615,
-      "grad_norm": 0.0013006648514419794,
-      "learning_rate": 0.00019999920843416876,
-      "loss": 46.0,
-      "step": 16575
-    },
-    {
-      "epoch": 1.2673509566680048,
-      "grad_norm": 0.0004796080756932497,
-      "learning_rate": 0.00019999920833859512,
-      "loss": 46.0,
-      "step": 16576
-    },
-    {
-      "epoch": 1.2674274136513943,
-      "grad_norm": 0.007776820100843906,
-      "learning_rate": 0.00019999920824301573,
-      "loss": 46.0,
-      "step": 16577
-    },
-    {
-      "epoch": 1.267503870634784,
-      "grad_norm": 0.00033089928911067545,
-      "learning_rate": 0.00019999920814743055,
-      "loss": 46.0,
-      "step": 16578
-    },
-    {
-      "epoch": 1.2675803276181739,
-      "grad_norm": 0.0030644612852483988,
-      "learning_rate": 0.00019999920805183958,
-      "loss": 46.0,
-      "step": 16579
-    },
-    {
-      "epoch": 1.2676567846015636,
-      "grad_norm": 0.0009644554811529815,
-      "learning_rate": 0.00019999920795624288,
-      "loss": 46.0,
-      "step": 16580
-    },
-    {
-      "epoch": 1.2677332415849532,
-      "grad_norm": 0.00048184592742472887,
-      "learning_rate": 0.0001999992078606404,
-      "loss": 46.0,
-      "step": 16581
-    },
-    {
-      "epoch": 1.267809698568343,
-      "grad_norm": 0.0012677001068368554,
-      "learning_rate": 0.00019999920776503211,
-      "loss": 46.0,
-      "step": 16582
-    },
-    {
-      "epoch": 1.2678861555517327,
-      "grad_norm": 0.0009831043425947428,
-      "learning_rate": 0.00019999920766941811,
-      "loss": 46.0,
-      "step": 16583
-    },
-    {
-      "epoch": 1.2679626125351224,
-      "grad_norm": 0.0007564874831587076,
-      "learning_rate": 0.0001999992075737983,
-      "loss": 46.0,
-      "step": 16584
-    },
-    {
-      "epoch": 1.2680390695185122,
-      "grad_norm": 0.00402864022180438,
-      "learning_rate": 0.00019999920747817274,
-      "loss": 46.0,
-      "step": 16585
-    },
-    {
-      "epoch": 1.268115526501902,
-      "grad_norm": 0.0007438174216076732,
-      "learning_rate": 0.00019999920738254142,
-      "loss": 46.0,
-      "step": 16586
-    },
-    {
-      "epoch": 1.2681919834852917,
-      "grad_norm": 0.0009092689142562449,
-      "learning_rate": 0.00019999920728690432,
-      "loss": 46.0,
-      "step": 16587
-    },
-    {
-      "epoch": 1.2682684404686813,
-      "grad_norm": 0.002842287067323923,
-      "learning_rate": 0.00019999920719126143,
-      "loss": 46.0,
-      "step": 16588
-    },
-    {
-      "epoch": 1.268344897452071,
-      "grad_norm": 0.0014212834648787975,
-      "learning_rate": 0.0001999992070956128,
-      "loss": 46.0,
-      "step": 16589
-    },
-    {
-      "epoch": 1.2684213544354608,
-      "grad_norm": 0.0015836736420169473,
-      "learning_rate": 0.00019999920699995837,
-      "loss": 46.0,
-      "step": 16590
-    },
-    {
-      "epoch": 1.2684978114188505,
-      "grad_norm": 0.007913281209766865,
-      "learning_rate": 0.00019999920690429819,
-      "loss": 46.0,
-      "step": 16591
-    },
-    {
-      "epoch": 1.26857426840224,
-      "grad_norm": 0.0007570114685222507,
-      "learning_rate": 0.00019999920680863222,
-      "loss": 46.0,
-      "step": 16592
-    },
-    {
-      "epoch": 1.2686507253856298,
-      "grad_norm": 0.0006269912701100111,
-      "learning_rate": 0.00019999920671296052,
-      "loss": 46.0,
-      "step": 16593
-    },
-    {
-      "epoch": 1.2687271823690196,
-      "grad_norm": 0.001360012567602098,
-      "learning_rate": 0.000199999206617283,
-      "loss": 46.0,
-      "step": 16594
-    },
-    {
-      "epoch": 1.2688036393524094,
-      "grad_norm": 0.0006117227021604776,
-      "learning_rate": 0.00019999920652159978,
-      "loss": 46.0,
-      "step": 16595
-    },
-    {
-      "epoch": 1.2688800963357991,
-      "grad_norm": 0.0008045897702686489,
-      "learning_rate": 0.00019999920642591073,
-      "loss": 46.0,
-      "step": 16596
-    },
-    {
-      "epoch": 1.2689565533191889,
-      "grad_norm": 0.0013260227860882878,
-      "learning_rate": 0.00019999920633021593,
-      "loss": 46.0,
-      "step": 16597
-    },
-    {
-      "epoch": 1.2690330103025786,
-      "grad_norm": 0.002823147689923644,
-      "learning_rate": 0.00019999920623451538,
-      "loss": 46.0,
-      "step": 16598
-    },
-    {
-      "epoch": 1.2691094672859682,
-      "grad_norm": 0.0012219519121572375,
-      "learning_rate": 0.00019999920613880906,
-      "loss": 46.0,
-      "step": 16599
-    },
-    {
-      "epoch": 1.269185924269358,
-      "grad_norm": 0.002700086683034897,
-      "learning_rate": 0.00019999920604309694,
-      "loss": 46.0,
-      "step": 16600
-    },
-    {
-      "epoch": 1.2692623812527477,
-      "grad_norm": 0.0026994612999260426,
-      "learning_rate": 0.00019999920594737907,
-      "loss": 46.0,
-      "step": 16601
-    },
-    {
-      "epoch": 1.2693388382361375,
-      "grad_norm": 0.0013872861163690686,
-      "learning_rate": 0.0001999992058516554,
-      "loss": 46.0,
-      "step": 16602
-    },
-    {
-      "epoch": 1.269415295219527,
-      "grad_norm": 0.0008902488625608385,
-      "learning_rate": 0.000199999205755926,
-      "loss": 46.0,
-      "step": 16603
-    },
-    {
-      "epoch": 1.2694917522029168,
-      "grad_norm": 0.0019012822303920984,
-      "learning_rate": 0.0001999992056601908,
-      "loss": 46.0,
-      "step": 16604
-    },
-    {
-      "epoch": 1.2695682091863065,
-      "grad_norm": 0.0006907540373504162,
-      "learning_rate": 0.00019999920556444987,
-      "loss": 46.0,
-      "step": 16605
-    },
-    {
-      "epoch": 1.2696446661696963,
-      "grad_norm": 0.0022475533187389374,
-      "learning_rate": 0.00019999920546870313,
-      "loss": 46.0,
-      "step": 16606
-    },
-    {
-      "epoch": 1.269721123153086,
-      "grad_norm": 0.0009680610382929444,
-      "learning_rate": 0.00019999920537295065,
-      "loss": 46.0,
-      "step": 16607
-    },
-    {
-      "epoch": 1.2697975801364758,
-      "grad_norm": 0.0036767732817679644,
-      "learning_rate": 0.00019999920527719237,
-      "loss": 46.0,
-      "step": 16608
-    },
-    {
-      "epoch": 1.2698740371198656,
-      "grad_norm": 0.003197531448677182,
-      "learning_rate": 0.00019999920518142835,
-      "loss": 46.0,
-      "step": 16609
-    },
-    {
-      "epoch": 1.2699504941032551,
-      "grad_norm": 0.001533971750177443,
-      "learning_rate": 0.00019999920508565855,
-      "loss": 46.0,
-      "step": 16610
-    },
-    {
-      "epoch": 1.2700269510866449,
-      "grad_norm": 0.0074510653503239155,
-      "learning_rate": 0.00019999920498988297,
-      "loss": 46.0,
-      "step": 16611
-    },
-    {
-      "epoch": 1.2701034080700346,
-      "grad_norm": 0.001222128514200449,
-      "learning_rate": 0.00019999920489410165,
-      "loss": 46.0,
-      "step": 16612
-    },
-    {
-      "epoch": 1.2701798650534244,
-      "grad_norm": 0.005164609290659428,
-      "learning_rate": 0.00019999920479831453,
-      "loss": 46.0,
-      "step": 16613
-    },
-    {
-      "epoch": 1.270256322036814,
-      "grad_norm": 0.0008039168897084892,
-      "learning_rate": 0.00019999920470252164,
-      "loss": 46.0,
-      "step": 16614
-    },
-    {
-      "epoch": 1.2703327790202037,
-      "grad_norm": 0.007434156257659197,
-      "learning_rate": 0.000199999204606723,
-      "loss": 46.0,
-      "step": 16615
-    },
-    {
-      "epoch": 1.2704092360035935,
-      "grad_norm": 0.0014897106448188424,
-      "learning_rate": 0.00019999920451091858,
-      "loss": 46.0,
-      "step": 16616
-    },
-    {
-      "epoch": 1.2704856929869832,
-      "grad_norm": 0.0005934524815529585,
-      "learning_rate": 0.0001999992044151084,
-      "loss": 46.0,
-      "step": 16617
-    },
-    {
-      "epoch": 1.270562149970373,
-      "grad_norm": 0.0008495784713886678,
-      "learning_rate": 0.00019999920431929244,
-      "loss": 46.0,
-      "step": 16618
-    },
-    {
-      "epoch": 1.2706386069537627,
-      "grad_norm": 0.041476089507341385,
-      "learning_rate": 0.0001999992042234707,
-      "loss": 46.0,
-      "step": 16619
-    },
-    {
-      "epoch": 1.2707150639371523,
-      "grad_norm": 0.0004496494948398322,
-      "learning_rate": 0.00019999920412764325,
-      "loss": 46.0,
-      "step": 16620
-    },
-    {
-      "epoch": 1.270791520920542,
-      "grad_norm": 0.006153932306915522,
-      "learning_rate": 0.00019999920403180997,
-      "loss": 46.0,
-      "step": 16621
-    },
-    {
-      "epoch": 1.2708679779039318,
-      "grad_norm": 0.002771378494799137,
-      "learning_rate": 0.00019999920393597091,
-      "loss": 46.0,
-      "step": 16622
-    },
-    {
-      "epoch": 1.2709444348873216,
-      "grad_norm": 0.003782296320423484,
-      "learning_rate": 0.00019999920384012611,
-      "loss": 46.0,
-      "step": 16623
-    },
-    {
-      "epoch": 1.2710208918707113,
-      "grad_norm": 0.001720154657959938,
-      "learning_rate": 0.00019999920374427554,
-      "loss": 46.0,
-      "step": 16624
-    },
-    {
-      "epoch": 1.2710973488541009,
-      "grad_norm": 0.0026643683668226004,
-      "learning_rate": 0.0001999992036484192,
-      "loss": 46.0,
-      "step": 16625
-    },
-    {
-      "epoch": 1.2711738058374906,
-      "grad_norm": 0.0022460664622485638,
-      "learning_rate": 0.0001999992035525571,
-      "loss": 46.0,
-      "step": 16626
-    },
-    {
-      "epoch": 1.2712502628208804,
-      "grad_norm": 0.015322674065828323,
-      "learning_rate": 0.00019999920345668923,
-      "loss": 46.0,
-      "step": 16627
-    },
-    {
-      "epoch": 1.2713267198042701,
-      "grad_norm": 0.007272724993526936,
-      "learning_rate": 0.00019999920336081557,
-      "loss": 46.0,
-      "step": 16628
-    },
-    {
-      "epoch": 1.27140317678766,
-      "grad_norm": 0.0018612624844536185,
-      "learning_rate": 0.00019999920326493613,
-      "loss": 46.0,
-      "step": 16629
-    },
-    {
-      "epoch": 1.2714796337710497,
-      "grad_norm": 0.0007866800879128277,
-      "learning_rate": 0.00019999920316905097,
-      "loss": 46.0,
-      "step": 16630
-    },
-    {
-      "epoch": 1.2715560907544392,
-      "grad_norm": 0.005086428951472044,
-      "learning_rate": 0.00019999920307316,
-      "loss": 46.0,
-      "step": 16631
-    },
-    {
-      "epoch": 1.271632547737829,
-      "grad_norm": 0.0009069970110431314,
-      "learning_rate": 0.00019999920297726327,
-      "loss": 46.0,
-      "step": 16632
-    },
-    {
-      "epoch": 1.2717090047212187,
-      "grad_norm": 0.001986677059903741,
-      "learning_rate": 0.0001999992028813608,
-      "loss": 46.0,
-      "step": 16633
-    },
-    {
-      "epoch": 1.2717854617046085,
-      "grad_norm": 0.003573087742552161,
-      "learning_rate": 0.0001999992027854525,
-      "loss": 46.0,
-      "step": 16634
-    },
-    {
-      "epoch": 1.271861918687998,
-      "grad_norm": 0.013164808973670006,
-      "learning_rate": 0.00019999920268953846,
-      "loss": 46.0,
-      "step": 16635
-    },
-    {
-      "epoch": 1.2719383756713878,
-      "grad_norm": 0.001385536976158619,
-      "learning_rate": 0.0001999992025936187,
-      "loss": 46.0,
-      "step": 16636
-    },
-    {
-      "epoch": 1.2720148326547775,
-      "grad_norm": 0.0006866052863188088,
-      "learning_rate": 0.00019999920249769311,
-      "loss": 46.0,
-      "step": 16637
-    },
-    {
-      "epoch": 1.2720912896381673,
-      "grad_norm": 0.00137321837246418,
-      "learning_rate": 0.00019999920240176177,
-      "loss": 46.0,
-      "step": 16638
-    },
-    {
-      "epoch": 1.272167746621557,
-      "grad_norm": 0.0012887525372207165,
-      "learning_rate": 0.00019999920230582462,
-      "loss": 46.0,
-      "step": 16639
-    },
-    {
-      "epoch": 1.2722442036049468,
-      "grad_norm": 0.0023031088057905436,
-      "learning_rate": 0.00019999920220988175,
-      "loss": 46.0,
-      "step": 16640
-    },
-    {
-      "epoch": 1.2723206605883366,
-      "grad_norm": 0.002139696851372719,
-      "learning_rate": 0.00019999920211393311,
-      "loss": 46.0,
-      "step": 16641
-    },
-    {
-      "epoch": 1.2723971175717261,
-      "grad_norm": 0.005255787633359432,
-      "learning_rate": 0.00019999920201797867,
-      "loss": 46.0,
-      "step": 16642
-    },
-    {
-      "epoch": 1.272473574555116,
-      "grad_norm": 0.00397008890286088,
-      "learning_rate": 0.0001999992019220185,
-      "loss": 46.0,
-      "step": 16643
-    },
-    {
-      "epoch": 1.2725500315385057,
-      "grad_norm": 0.002200365299358964,
-      "learning_rate": 0.00019999920182605253,
-      "loss": 46.0,
-      "step": 16644
-    },
-    {
-      "epoch": 1.2726264885218954,
-      "grad_norm": 0.0005313935107551515,
-      "learning_rate": 0.0001999992017300808,
-      "loss": 46.0,
-      "step": 16645
-    },
-    {
-      "epoch": 1.272702945505285,
-      "grad_norm": 0.006090338341891766,
-      "learning_rate": 0.0001999992016341033,
-      "loss": 46.0,
-      "step": 16646
-    },
-    {
-      "epoch": 1.2727794024886747,
-      "grad_norm": 0.0013891661074012518,
-      "learning_rate": 0.00019999920153812,
-      "loss": 46.0,
-      "step": 16647
-    },
-    {
-      "epoch": 1.2728558594720645,
-      "grad_norm": 0.0023395169991999865,
-      "learning_rate": 0.00019999920144213098,
-      "loss": 46.0,
-      "step": 16648
-    },
-    {
-      "epoch": 1.2729323164554542,
-      "grad_norm": 0.002753504319116473,
-      "learning_rate": 0.00019999920134613616,
-      "loss": 46.0,
-      "step": 16649
-    },
-    {
-      "epoch": 1.273008773438844,
-      "grad_norm": 0.014427597634494305,
-      "learning_rate": 0.0001999992012501356,
-      "loss": 46.0,
-      "step": 16650
-    },
-    {
-      "epoch": 1.2730852304222338,
-      "grad_norm": 0.0012178807519376278,
-      "learning_rate": 0.00019999920115412926,
-      "loss": 46.0,
-      "step": 16651
-    },
-    {
-      "epoch": 1.2731616874056235,
-      "grad_norm": 0.001439434476196766,
-      "learning_rate": 0.00019999920105811714,
-      "loss": 46.0,
-      "step": 16652
-    },
-    {
-      "epoch": 1.273238144389013,
-      "grad_norm": 0.001111507066525519,
-      "learning_rate": 0.00019999920096209925,
-      "loss": 46.0,
-      "step": 16653
-    },
-    {
-      "epoch": 1.2733146013724028,
-      "grad_norm": 0.0007047646213322878,
-      "learning_rate": 0.0001999992008660756,
-      "loss": 46.0,
-      "step": 16654
-    },
-    {
-      "epoch": 1.2733910583557926,
-      "grad_norm": 0.007054263260215521,
-      "learning_rate": 0.0001999992007700462,
-      "loss": 46.0,
-      "step": 16655
-    },
-    {
-      "epoch": 1.2734675153391823,
-      "grad_norm": 0.0009644195670261979,
-      "learning_rate": 0.00019999920067401099,
-      "loss": 46.0,
-      "step": 16656
-    },
-    {
-      "epoch": 1.2735439723225719,
-      "grad_norm": 0.0018699506763368845,
-      "learning_rate": 0.00019999920057797003,
-      "loss": 46.0,
-      "step": 16657
-    },
-    {
-      "epoch": 1.2736204293059616,
-      "grad_norm": 0.001187073066830635,
-      "learning_rate": 0.0001999992004819233,
-      "loss": 46.0,
-      "step": 16658
-    },
-    {
-      "epoch": 1.2736968862893514,
-      "grad_norm": 0.0007307162159122527,
-      "learning_rate": 0.0001999992003858708,
-      "loss": 46.0,
-      "step": 16659
-    },
-    {
-      "epoch": 1.2737733432727412,
-      "grad_norm": 0.0028837842401117086,
-      "learning_rate": 0.0001999992002898125,
-      "loss": 46.0,
-      "step": 16660
-    },
-    {
-      "epoch": 1.273849800256131,
-      "grad_norm": 0.002315081190317869,
-      "learning_rate": 0.00019999920019374848,
-      "loss": 46.0,
-      "step": 16661
-    },
-    {
-      "epoch": 1.2739262572395207,
-      "grad_norm": 0.0014030288439244032,
-      "learning_rate": 0.00019999920009767868,
-      "loss": 46.0,
-      "step": 16662
-    },
-    {
-      "epoch": 1.2740027142229104,
-      "grad_norm": 0.0015715938061475754,
-      "learning_rate": 0.0001999992000016031,
-      "loss": 46.0,
-      "step": 16663
-    },
-    {
-      "epoch": 1.2740791712063,
-      "grad_norm": 0.0003001201548613608,
-      "learning_rate": 0.00019999919990552174,
-      "loss": 46.0,
-      "step": 16664
-    },
-    {
-      "epoch": 1.2741556281896897,
-      "grad_norm": 0.0005279534379951656,
-      "learning_rate": 0.00019999919980943462,
-      "loss": 46.0,
-      "step": 16665
-    },
-    {
-      "epoch": 1.2742320851730795,
-      "grad_norm": 0.0014168646885082126,
-      "learning_rate": 0.00019999919971334172,
-      "loss": 46.0,
-      "step": 16666
-    },
-    {
-      "epoch": 1.2743085421564693,
-      "grad_norm": 0.005152455996721983,
-      "learning_rate": 0.00019999919961724308,
-      "loss": 46.0,
-      "step": 16667
-    },
-    {
-      "epoch": 1.2743849991398588,
-      "grad_norm": 0.0014782946091145277,
-      "learning_rate": 0.00019999919952113865,
-      "loss": 46.0,
-      "step": 16668
-    },
-    {
-      "epoch": 1.2744614561232486,
-      "grad_norm": 0.001206699525937438,
-      "learning_rate": 0.00019999919942502846,
-      "loss": 46.0,
-      "step": 16669
-    },
-    {
-      "epoch": 1.2745379131066383,
-      "grad_norm": 0.0012304575648158789,
-      "learning_rate": 0.0001999991993289125,
-      "loss": 46.0,
-      "step": 16670
-    },
-    {
-      "epoch": 1.274614370090028,
-      "grad_norm": 0.0014089810429140925,
-      "learning_rate": 0.00019999919923279077,
-      "loss": 46.0,
-      "step": 16671
-    },
-    {
-      "epoch": 1.2746908270734179,
-      "grad_norm": 0.005282597150653601,
-      "learning_rate": 0.00019999919913666326,
-      "loss": 46.0,
-      "step": 16672
-    },
-    {
-      "epoch": 1.2747672840568076,
-      "grad_norm": 0.001720866421237588,
-      "learning_rate": 0.00019999919904052998,
-      "loss": 46.0,
-      "step": 16673
-    },
-    {
-      "epoch": 1.2748437410401974,
-      "grad_norm": 0.0008292746497318149,
-      "learning_rate": 0.00019999919894439093,
-      "loss": 46.0,
-      "step": 16674
-    },
-    {
-      "epoch": 1.274920198023587,
-      "grad_norm": 0.0014621547888964415,
-      "learning_rate": 0.00019999919884824613,
-      "loss": 46.0,
-      "step": 16675
-    },
-    {
-      "epoch": 1.2749966550069767,
-      "grad_norm": 0.00047751498641446233,
-      "learning_rate": 0.00019999919875209553,
-      "loss": 46.0,
-      "step": 16676
-    },
-    {
-      "epoch": 1.2750731119903664,
-      "grad_norm": 0.0038491645827889442,
-      "learning_rate": 0.00019999919865593919,
-      "loss": 46.0,
-      "step": 16677
-    },
-    {
-      "epoch": 1.2751495689737562,
-      "grad_norm": 0.0006406709435395896,
-      "learning_rate": 0.00019999919855977707,
-      "loss": 46.0,
-      "step": 16678
-    },
-    {
-      "epoch": 1.2752260259571457,
-      "grad_norm": 0.0006662397645413876,
-      "learning_rate": 0.0001999991984636092,
-      "loss": 46.0,
-      "step": 16679
-    },
-    {
-      "epoch": 1.2753024829405355,
-      "grad_norm": 0.0007661851705051959,
-      "learning_rate": 0.00019999919836743554,
-      "loss": 46.0,
-      "step": 16680
-    },
-    {
-      "epoch": 1.2753789399239253,
-      "grad_norm": 0.0015701422234997153,
-      "learning_rate": 0.00019999919827125607,
-      "loss": 46.0,
-      "step": 16681
-    },
-    {
-      "epoch": 1.275455396907315,
-      "grad_norm": 0.0017563244327902794,
-      "learning_rate": 0.00019999919817507088,
-      "loss": 46.0,
-      "step": 16682
-    },
-    {
-      "epoch": 1.2755318538907048,
-      "grad_norm": 0.000649426830932498,
-      "learning_rate": 0.00019999919807887993,
-      "loss": 46.0,
-      "step": 16683
-    },
-    {
-      "epoch": 1.2756083108740945,
-      "grad_norm": 0.0009337930241599679,
-      "learning_rate": 0.0001999991979826832,
-      "loss": 46.0,
-      "step": 16684
-    },
-    {
-      "epoch": 1.2756847678574843,
-      "grad_norm": 0.0007409833488054574,
-      "learning_rate": 0.00019999919788648066,
-      "loss": 46.0,
-      "step": 16685
-    },
-    {
-      "epoch": 1.2757612248408738,
-      "grad_norm": 0.0012036182451993227,
-      "learning_rate": 0.0001999991977902724,
-      "loss": 46.0,
-      "step": 16686
-    },
-    {
-      "epoch": 1.2758376818242636,
-      "grad_norm": 0.002025241730734706,
-      "learning_rate": 0.00019999919769405838,
-      "loss": 46.0,
-      "step": 16687
-    },
-    {
-      "epoch": 1.2759141388076534,
-      "grad_norm": 0.0032368816900998354,
-      "learning_rate": 0.00019999919759783853,
-      "loss": 46.0,
-      "step": 16688
-    },
-    {
-      "epoch": 1.2759905957910431,
-      "grad_norm": 0.0009148273384198546,
-      "learning_rate": 0.00019999919750161299,
-      "loss": 46.0,
-      "step": 16689
-    },
-    {
-      "epoch": 1.2760670527744327,
-      "grad_norm": 0.0033446482848376036,
-      "learning_rate": 0.0001999991974053816,
-      "loss": 46.0,
-      "step": 16690
-    },
-    {
-      "epoch": 1.2761435097578224,
-      "grad_norm": 0.002166281221434474,
-      "learning_rate": 0.0001999991973091445,
-      "loss": 46.0,
-      "step": 16691
-    },
-    {
-      "epoch": 1.2762199667412122,
-      "grad_norm": 0.002013568300753832,
-      "learning_rate": 0.0001999991972129016,
-      "loss": 46.0,
-      "step": 16692
-    },
-    {
-      "epoch": 1.276296423724602,
-      "grad_norm": 0.001305202255025506,
-      "learning_rate": 0.00019999919711665294,
-      "loss": 46.0,
-      "step": 16693
-    },
-    {
-      "epoch": 1.2763728807079917,
-      "grad_norm": 0.0012782372068613768,
-      "learning_rate": 0.0001999991970203985,
-      "loss": 46.0,
-      "step": 16694
-    },
-    {
-      "epoch": 1.2764493376913815,
-      "grad_norm": 0.0006937553989700973,
-      "learning_rate": 0.0001999991969241383,
-      "loss": 46.0,
-      "step": 16695
-    },
-    {
-      "epoch": 1.2765257946747712,
-      "grad_norm": 0.001616218825802207,
-      "learning_rate": 0.00019999919682787233,
-      "loss": 46.0,
-      "step": 16696
-    },
-    {
-      "epoch": 1.2766022516581608,
-      "grad_norm": 0.002052373019978404,
-      "learning_rate": 0.00019999919673160062,
-      "loss": 46.0,
-      "step": 16697
-    },
-    {
-      "epoch": 1.2766787086415505,
-      "grad_norm": 0.0010528018465265632,
-      "learning_rate": 0.0001999991966353231,
-      "loss": 46.0,
-      "step": 16698
-    },
-    {
-      "epoch": 1.2767551656249403,
-      "grad_norm": 0.0025244427379220724,
-      "learning_rate": 0.0001999991965390398,
-      "loss": 46.0,
-      "step": 16699
-    },
-    {
-      "epoch": 1.27683162260833,
-      "grad_norm": 0.0011942755663767457,
-      "learning_rate": 0.00019999919644275076,
-      "loss": 46.0,
-      "step": 16700
-    },
-    {
-      "epoch": 1.2769080795917196,
-      "grad_norm": 0.00030391293694265187,
-      "learning_rate": 0.00019999919634645593,
-      "loss": 46.0,
-      "step": 16701
-    },
-    {
-      "epoch": 1.2769845365751094,
-      "grad_norm": 0.00041188261820934713,
-      "learning_rate": 0.00019999919625015536,
-      "loss": 46.0,
-      "step": 16702
-    },
-    {
-      "epoch": 1.2770609935584991,
-      "grad_norm": 0.0017732786945998669,
-      "learning_rate": 0.000199999196153849,
-      "loss": 46.0,
-      "step": 16703
-    },
-    {
-      "epoch": 1.2771374505418889,
-      "grad_norm": 0.0050637805834412575,
-      "learning_rate": 0.00019999919605753687,
-      "loss": 46.0,
-      "step": 16704
-    },
-    {
-      "epoch": 1.2772139075252786,
-      "grad_norm": 0.006125276442617178,
-      "learning_rate": 0.00019999919596121898,
-      "loss": 46.0,
-      "step": 16705
-    },
-    {
-      "epoch": 1.2772903645086684,
-      "grad_norm": 0.0006493722321465611,
-      "learning_rate": 0.00019999919586489534,
-      "loss": 46.0,
-      "step": 16706
-    },
-    {
-      "epoch": 1.2773668214920582,
-      "grad_norm": 0.0010154986521229148,
-      "learning_rate": 0.0001999991957685659,
-      "loss": 46.0,
-      "step": 16707
-    },
-    {
-      "epoch": 1.2774432784754477,
-      "grad_norm": 0.004606302827596664,
-      "learning_rate": 0.0001999991956722307,
-      "loss": 46.0,
-      "step": 16708
-    },
-    {
-      "epoch": 1.2775197354588375,
-      "grad_norm": 0.00040772423380985856,
-      "learning_rate": 0.0001999991955758897,
-      "loss": 46.0,
-      "step": 16709
-    },
-    {
-      "epoch": 1.2775961924422272,
-      "grad_norm": 0.0010258068796247244,
-      "learning_rate": 0.00019999919547954295,
-      "loss": 46.0,
-      "step": 16710
-    },
-    {
-      "epoch": 1.277672649425617,
-      "grad_norm": 0.0014730957336723804,
-      "learning_rate": 0.00019999919538319047,
-      "loss": 46.0,
-      "step": 16711
-    },
-    {
-      "epoch": 1.2777491064090065,
-      "grad_norm": 0.0018360507674515247,
-      "learning_rate": 0.00019999919528683216,
-      "loss": 46.0,
-      "step": 16712
-    },
-    {
-      "epoch": 1.2778255633923963,
-      "grad_norm": 0.0031213832553476095,
-      "learning_rate": 0.00019999919519046814,
-      "loss": 46.0,
-      "step": 16713
-    },
-    {
-      "epoch": 1.277902020375786,
-      "grad_norm": 0.0005239760503172874,
-      "learning_rate": 0.00019999919509409831,
-      "loss": 46.0,
-      "step": 16714
-    },
-    {
-      "epoch": 1.2779784773591758,
-      "grad_norm": 0.002066908171400428,
-      "learning_rate": 0.00019999919499772272,
-      "loss": 46.0,
-      "step": 16715
-    },
-    {
-      "epoch": 1.2780549343425656,
-      "grad_norm": 0.007595262955874205,
-      "learning_rate": 0.00019999919490134137,
-      "loss": 46.0,
-      "step": 16716
-    },
-    {
-      "epoch": 1.2781313913259553,
-      "grad_norm": 0.0014205316547304392,
-      "learning_rate": 0.00019999919480495423,
-      "loss": 46.0,
-      "step": 16717
-    },
-    {
-      "epoch": 1.278207848309345,
-      "grad_norm": 0.00043894999544136226,
-      "learning_rate": 0.00019999919470856133,
-      "loss": 46.0,
-      "step": 16718
-    },
-    {
-      "epoch": 1.2782843052927346,
-      "grad_norm": 0.007440910208970308,
-      "learning_rate": 0.00019999919461216267,
-      "loss": 46.0,
-      "step": 16719
-    },
-    {
-      "epoch": 1.2783607622761244,
-      "grad_norm": 0.0008081428823061287,
-      "learning_rate": 0.00019999919451575823,
-      "loss": 46.0,
-      "step": 16720
-    },
-    {
-      "epoch": 1.2784372192595141,
-      "grad_norm": 0.00144693732727319,
-      "learning_rate": 0.00019999919441934805,
-      "loss": 46.0,
-      "step": 16721
-    },
-    {
-      "epoch": 1.278513676242904,
-      "grad_norm": 0.001247951528057456,
-      "learning_rate": 0.00019999919432293206,
-      "loss": 46.0,
-      "step": 16722
-    },
-    {
-      "epoch": 1.2785901332262934,
-      "grad_norm": 0.0061896494589746,
-      "learning_rate": 0.00019999919422651033,
-      "loss": 46.0,
-      "step": 16723
-    },
-    {
-      "epoch": 1.2786665902096832,
-      "grad_norm": 0.0013113584136590362,
-      "learning_rate": 0.00019999919413008283,
-      "loss": 46.0,
-      "step": 16724
-    },
-    {
-      "epoch": 1.278743047193073,
-      "grad_norm": 0.000813615566585213,
-      "learning_rate": 0.0001999991940336495,
-      "loss": 46.0,
-      "step": 16725
-    },
-    {
-      "epoch": 1.2788195041764627,
-      "grad_norm": 0.0031911423429846764,
-      "learning_rate": 0.00019999919393721047,
-      "loss": 46.0,
-      "step": 16726
-    },
-    {
-      "epoch": 1.2788959611598525,
-      "grad_norm": 0.0012606719974428415,
-      "learning_rate": 0.00019999919384076567,
-      "loss": 46.0,
-      "step": 16727
-    },
-    {
-      "epoch": 1.2789724181432423,
-      "grad_norm": 0.0007722353911958635,
-      "learning_rate": 0.00019999919374431507,
-      "loss": 46.0,
-      "step": 16728
-    },
-    {
-      "epoch": 1.279048875126632,
-      "grad_norm": 0.0006220923969522119,
-      "learning_rate": 0.00019999919364785867,
-      "loss": 46.0,
-      "step": 16729
-    },
-    {
-      "epoch": 1.2791253321100216,
-      "grad_norm": 0.0006802931311540306,
-      "learning_rate": 0.00019999919355139656,
-      "loss": 46.0,
-      "step": 16730
-    },
-    {
-      "epoch": 1.2792017890934113,
-      "grad_norm": 0.001854546251706779,
-      "learning_rate": 0.00019999919345492867,
-      "loss": 46.0,
-      "step": 16731
-    },
-    {
-      "epoch": 1.279278246076801,
-      "grad_norm": 0.00047614064533263445,
-      "learning_rate": 0.00019999919335845497,
-      "loss": 46.0,
-      "step": 16732
-    },
-    {
-      "epoch": 1.2793547030601908,
-      "grad_norm": 0.0013608257286250591,
-      "learning_rate": 0.00019999919326197556,
-      "loss": 46.0,
-      "step": 16733
-    },
-    {
-      "epoch": 1.2794311600435804,
-      "grad_norm": 0.0005788191920146346,
-      "learning_rate": 0.00019999919316549035,
-      "loss": 46.0,
-      "step": 16734
-    },
-    {
-      "epoch": 1.2795076170269701,
-      "grad_norm": 0.0009505003690719604,
-      "learning_rate": 0.00019999919306899937,
-      "loss": 46.0,
-      "step": 16735
-    },
-    {
-      "epoch": 1.27958407401036,
-      "grad_norm": 0.0007994129555299878,
-      "learning_rate": 0.00019999919297250264,
-      "loss": 46.0,
-      "step": 16736
-    },
-    {
-      "epoch": 1.2796605309937497,
-      "grad_norm": 0.0009163281647488475,
-      "learning_rate": 0.0001999991928760001,
-      "loss": 46.0,
-      "step": 16737
-    },
-    {
-      "epoch": 1.2797369879771394,
-      "grad_norm": 0.0012442166917026043,
-      "learning_rate": 0.00019999919277949183,
-      "loss": 46.0,
-      "step": 16738
-    },
-    {
-      "epoch": 1.2798134449605292,
-      "grad_norm": 0.0014432481257244945,
-      "learning_rate": 0.00019999919268297778,
-      "loss": 46.0,
-      "step": 16739
-    },
-    {
-      "epoch": 1.279889901943919,
-      "grad_norm": 0.0018776024226099253,
-      "learning_rate": 0.00019999919258645795,
-      "loss": 46.0,
-      "step": 16740
-    },
-    {
-      "epoch": 1.2799663589273085,
-      "grad_norm": 0.004317415878176689,
-      "learning_rate": 0.00019999919248993235,
-      "loss": 46.0,
-      "step": 16741
-    },
-    {
-      "epoch": 1.2800428159106982,
-      "grad_norm": 0.003129079006612301,
-      "learning_rate": 0.00019999919239340098,
-      "loss": 46.0,
-      "step": 16742
-    },
-    {
-      "epoch": 1.280119272894088,
-      "grad_norm": 0.003770609153434634,
-      "learning_rate": 0.00019999919229686386,
-      "loss": 46.0,
-      "step": 16743
-    },
-    {
-      "epoch": 1.2801957298774778,
-      "grad_norm": 0.007101863622665405,
-      "learning_rate": 0.00019999919220032095,
-      "loss": 46.0,
-      "step": 16744
-    },
-    {
-      "epoch": 1.2802721868608673,
-      "grad_norm": 0.0007034234004095197,
-      "learning_rate": 0.00019999919210377228,
-      "loss": 46.0,
-      "step": 16745
-    },
-    {
-      "epoch": 1.280348643844257,
-      "grad_norm": 0.004554283805191517,
-      "learning_rate": 0.00019999919200721787,
-      "loss": 46.0,
-      "step": 16746
-    },
-    {
-      "epoch": 1.2804251008276468,
-      "grad_norm": 0.0004629794566426426,
-      "learning_rate": 0.00019999919191065763,
-      "loss": 46.0,
-      "step": 16747
-    },
-    {
-      "epoch": 1.2805015578110366,
-      "grad_norm": 0.002565753413364291,
-      "learning_rate": 0.00019999919181409168,
-      "loss": 46.0,
-      "step": 16748
-    },
-    {
-      "epoch": 1.2805780147944263,
-      "grad_norm": 0.0008360464707948267,
-      "learning_rate": 0.00019999919171751992,
-      "loss": 46.0,
-      "step": 16749
-    },
-    {
-      "epoch": 1.280654471777816,
-      "grad_norm": 0.00041501910891383886,
-      "learning_rate": 0.00019999919162094241,
-      "loss": 46.0,
-      "step": 16750
-    },
-    {
-      "epoch": 1.2807309287612056,
-      "grad_norm": 0.006867795251309872,
-      "learning_rate": 0.0001999991915243591,
-      "loss": 46.0,
-      "step": 16751
-    },
-    {
-      "epoch": 1.2808073857445954,
-      "grad_norm": 0.0028635147027671337,
-      "learning_rate": 0.00019999919142777006,
-      "loss": 46.0,
-      "step": 16752
-    },
-    {
-      "epoch": 1.2808838427279852,
-      "grad_norm": 0.0019897243473678827,
-      "learning_rate": 0.0001999991913311752,
-      "loss": 46.0,
-      "step": 16753
-    },
-    {
-      "epoch": 1.280960299711375,
-      "grad_norm": 0.001525622676126659,
-      "learning_rate": 0.0001999991912345746,
-      "loss": 46.0,
-      "step": 16754
-    },
-    {
-      "epoch": 1.2810367566947647,
-      "grad_norm": 0.005283045582473278,
-      "learning_rate": 0.00019999919113796827,
-      "loss": 46.0,
-      "step": 16755
-    },
-    {
-      "epoch": 1.2811132136781542,
-      "grad_norm": 0.001264419755898416,
-      "learning_rate": 0.00019999919104135615,
-      "loss": 46.0,
-      "step": 16756
-    },
-    {
-      "epoch": 1.281189670661544,
-      "grad_norm": 0.0012440038844943047,
-      "learning_rate": 0.00019999919094473826,
-      "loss": 46.0,
-      "step": 16757
-    },
-    {
-      "epoch": 1.2812661276449337,
-      "grad_norm": 0.0007672092178836465,
-      "learning_rate": 0.00019999919084811457,
-      "loss": 46.0,
-      "step": 16758
-    },
-    {
-      "epoch": 1.2813425846283235,
-      "grad_norm": 0.00561955152079463,
-      "learning_rate": 0.00019999919075148513,
-      "loss": 46.0,
-      "step": 16759
-    },
-    {
-      "epoch": 1.2814190416117133,
-      "grad_norm": 0.0029164638835936785,
-      "learning_rate": 0.00019999919065484992,
-      "loss": 46.0,
-      "step": 16760
-    },
-    {
-      "epoch": 1.281495498595103,
-      "grad_norm": 0.0009613786824047565,
-      "learning_rate": 0.00019999919055820894,
-      "loss": 46.0,
-      "step": 16761
-    },
-    {
-      "epoch": 1.2815719555784926,
-      "grad_norm": 0.0004206848971080035,
-      "learning_rate": 0.00019999919046156218,
-      "loss": 46.0,
-      "step": 16762
-    },
-    {
-      "epoch": 1.2816484125618823,
-      "grad_norm": 0.0011929139727726579,
-      "learning_rate": 0.00019999919036490965,
-      "loss": 46.0,
-      "step": 16763
-    },
-    {
-      "epoch": 1.281724869545272,
-      "grad_norm": 0.00257196556776762,
-      "learning_rate": 0.00019999919026825137,
-      "loss": 46.0,
-      "step": 16764
-    },
-    {
-      "epoch": 1.2818013265286619,
-      "grad_norm": 0.0005610408261418343,
-      "learning_rate": 0.0001999991901715873,
-      "loss": 46.0,
-      "step": 16765
-    },
-    {
-      "epoch": 1.2818777835120514,
-      "grad_norm": 0.0007481915527023375,
-      "learning_rate": 0.00019999919007491747,
-      "loss": 46.0,
-      "step": 16766
-    },
-    {
-      "epoch": 1.2819542404954412,
-      "grad_norm": 0.0008595797698944807,
-      "learning_rate": 0.00019999918997824187,
-      "loss": 46.0,
-      "step": 16767
-    },
-    {
-      "epoch": 1.282030697478831,
-      "grad_norm": 0.0005751067074015737,
-      "learning_rate": 0.00019999918988156053,
-      "loss": 46.0,
-      "step": 16768
-    },
-    {
-      "epoch": 1.2821071544622207,
-      "grad_norm": 0.0008936603553593159,
-      "learning_rate": 0.00019999918978487338,
-      "loss": 46.0,
-      "step": 16769
-    },
-    {
-      "epoch": 1.2821836114456104,
-      "grad_norm": 0.006716905161738396,
-      "learning_rate": 0.00019999918968818047,
-      "loss": 46.0,
-      "step": 16770
-    },
-    {
-      "epoch": 1.2822600684290002,
-      "grad_norm": 0.0010301817674189806,
-      "learning_rate": 0.0001999991895914818,
-      "loss": 46.0,
-      "step": 16771
-    },
-    {
-      "epoch": 1.28233652541239,
-      "grad_norm": 0.0013083780650049448,
-      "learning_rate": 0.00019999918949477734,
-      "loss": 46.0,
-      "step": 16772
-    },
-    {
-      "epoch": 1.2824129823957795,
-      "grad_norm": 0.0007963533280417323,
-      "learning_rate": 0.00019999918939806716,
-      "loss": 46.0,
-      "step": 16773
-    },
-    {
-      "epoch": 1.2824894393791693,
-      "grad_norm": 0.0004952663439325988,
-      "learning_rate": 0.00019999918930135115,
-      "loss": 46.0,
-      "step": 16774
-    },
-    {
-      "epoch": 1.282565896362559,
-      "grad_norm": 0.0019545843824744225,
-      "learning_rate": 0.0001999991892046294,
-      "loss": 46.0,
-      "step": 16775
-    },
-    {
-      "epoch": 1.2826423533459488,
-      "grad_norm": 0.0017712283879518509,
-      "learning_rate": 0.00019999918910790188,
-      "loss": 46.0,
-      "step": 16776
-    },
-    {
-      "epoch": 1.2827188103293383,
-      "grad_norm": 0.0007526252884417772,
-      "learning_rate": 0.00019999918901116858,
-      "loss": 46.0,
-      "step": 16777
-    },
-    {
-      "epoch": 1.282795267312728,
-      "grad_norm": 0.0017684220802038908,
-      "learning_rate": 0.00019999918891442953,
-      "loss": 46.0,
-      "step": 16778
-    },
-    {
-      "epoch": 1.2828717242961178,
-      "grad_norm": 0.0010477288160473108,
-      "learning_rate": 0.00019999918881768468,
-      "loss": 46.0,
-      "step": 16779
-    },
-    {
-      "epoch": 1.2829481812795076,
-      "grad_norm": 0.008022102527320385,
-      "learning_rate": 0.00019999918872093408,
-      "loss": 46.0,
-      "step": 16780
-    },
-    {
-      "epoch": 1.2830246382628974,
-      "grad_norm": 0.000900499930139631,
-      "learning_rate": 0.0001999991886241777,
-      "loss": 46.0,
-      "step": 16781
-    },
-    {
-      "epoch": 1.2831010952462871,
-      "grad_norm": 0.002912320662289858,
-      "learning_rate": 0.0001999991885274156,
-      "loss": 46.0,
-      "step": 16782
-    },
-    {
-      "epoch": 1.2831775522296769,
-      "grad_norm": 0.000426529353717342,
-      "learning_rate": 0.00019999918843064768,
-      "loss": 46.0,
-      "step": 16783
-    },
-    {
-      "epoch": 1.2832540092130664,
-      "grad_norm": 0.0009704563417471945,
-      "learning_rate": 0.00019999918833387402,
-      "loss": 46.0,
-      "step": 16784
-    },
-    {
-      "epoch": 1.2833304661964562,
-      "grad_norm": 0.001059980597347021,
-      "learning_rate": 0.00019999918823709453,
-      "loss": 46.0,
-      "step": 16785
-    },
-    {
-      "epoch": 1.283406923179846,
-      "grad_norm": 0.0007888384279794991,
-      "learning_rate": 0.00019999918814030934,
-      "loss": 46.0,
-      "step": 16786
-    },
-    {
-      "epoch": 1.2834833801632357,
-      "grad_norm": 0.00038376240991055965,
-      "learning_rate": 0.00019999918804351836,
-      "loss": 46.0,
-      "step": 16787
-    },
-    {
-      "epoch": 1.2835598371466252,
-      "grad_norm": 0.0007514795288443565,
-      "learning_rate": 0.0001999991879467216,
-      "loss": 46.0,
-      "step": 16788
-    },
-    {
-      "epoch": 1.283636294130015,
-      "grad_norm": 0.002230469137430191,
-      "learning_rate": 0.00019999918784991907,
-      "loss": 46.0,
-      "step": 16789
-    },
-    {
-      "epoch": 1.2837127511134048,
-      "grad_norm": 0.0016391673125326633,
-      "learning_rate": 0.00019999918775311077,
-      "loss": 46.0,
-      "step": 16790
-    },
-    {
-      "epoch": 1.2837892080967945,
-      "grad_norm": 0.002981659723445773,
-      "learning_rate": 0.00019999918765629672,
-      "loss": 46.0,
-      "step": 16791
-    },
-    {
-      "epoch": 1.2838656650801843,
-      "grad_norm": 0.0009505358757451177,
-      "learning_rate": 0.0001999991875594769,
-      "loss": 46.0,
-      "step": 16792
-    },
-    {
-      "epoch": 1.283942122063574,
-      "grad_norm": 0.001292443834245205,
-      "learning_rate": 0.00019999918746265127,
-      "loss": 46.0,
-      "step": 16793
-    },
-    {
-      "epoch": 1.2840185790469638,
-      "grad_norm": 0.0011380905052646995,
-      "learning_rate": 0.0001999991873658199,
-      "loss": 46.0,
-      "step": 16794
-    },
-    {
-      "epoch": 1.2840950360303534,
-      "grad_norm": 0.0011754799634218216,
-      "learning_rate": 0.00019999918726898276,
-      "loss": 46.0,
-      "step": 16795
-    },
-    {
-      "epoch": 1.2841714930137431,
-      "grad_norm": 0.006151310168206692,
-      "learning_rate": 0.00019999918717213987,
-      "loss": 46.0,
-      "step": 16796
-    },
-    {
-      "epoch": 1.2842479499971329,
-      "grad_norm": 0.0008278985624201596,
-      "learning_rate": 0.00019999918707529118,
-      "loss": 46.0,
-      "step": 16797
-    },
-    {
-      "epoch": 1.2843244069805226,
-      "grad_norm": 0.007651502266526222,
-      "learning_rate": 0.00019999918697843674,
-      "loss": 46.0,
-      "step": 16798
-    },
-    {
-      "epoch": 1.2844008639639122,
-      "grad_norm": 0.000510715355630964,
-      "learning_rate": 0.0001999991868815765,
-      "loss": 46.0,
-      "step": 16799
-    },
-    {
-      "epoch": 1.284477320947302,
-      "grad_norm": 0.0029057872015982866,
-      "learning_rate": 0.00019999918678471052,
-      "loss": 46.0,
-      "step": 16800
-    },
-    {
-      "epoch": 1.2845537779306917,
-      "grad_norm": 0.005568479187786579,
-      "learning_rate": 0.00019999918668783874,
-      "loss": 46.0,
-      "step": 16801
-    },
-    {
-      "epoch": 1.2846302349140815,
-      "grad_norm": 0.013746390119194984,
-      "learning_rate": 0.0001999991865909612,
-      "loss": 46.0,
-      "step": 16802
-    },
-    {
-      "epoch": 1.2847066918974712,
-      "grad_norm": 0.0011699412716552615,
-      "learning_rate": 0.00019999918649407793,
-      "loss": 46.0,
-      "step": 16803
-    },
-    {
-      "epoch": 1.284783148880861,
-      "grad_norm": 0.0008903414709493518,
-      "learning_rate": 0.00019999918639718888,
-      "loss": 46.0,
-      "step": 16804
-    },
-    {
-      "epoch": 1.2848596058642507,
-      "grad_norm": 0.002062085783109069,
-      "learning_rate": 0.00019999918630029403,
-      "loss": 46.0,
-      "step": 16805
-    },
-    {
-      "epoch": 1.2849360628476403,
-      "grad_norm": 0.0009825292509049177,
-      "learning_rate": 0.0001999991862033934,
-      "loss": 46.0,
-      "step": 16806
-    },
-    {
-      "epoch": 1.28501251983103,
-      "grad_norm": 0.017173558473587036,
-      "learning_rate": 0.00019999918610648707,
-      "loss": 46.0,
-      "step": 16807
-    },
-    {
-      "epoch": 1.2850889768144198,
-      "grad_norm": 0.005197874736040831,
-      "learning_rate": 0.0001999991860095749,
-      "loss": 46.0,
-      "step": 16808
-    },
-    {
-      "epoch": 1.2851654337978096,
-      "grad_norm": 0.0014744378859177232,
-      "learning_rate": 0.000199999185912657,
-      "loss": 46.0,
-      "step": 16809
-    },
-    {
-      "epoch": 1.285241890781199,
-      "grad_norm": 0.0020845697727054358,
-      "learning_rate": 0.00019999918581573335,
-      "loss": 46.0,
-      "step": 16810
-    },
-    {
-      "epoch": 1.2853183477645889,
-      "grad_norm": 0.002136238617822528,
-      "learning_rate": 0.00019999918571880388,
-      "loss": 46.0,
-      "step": 16811
-    },
-    {
-      "epoch": 1.2853948047479786,
-      "grad_norm": 0.0004366362700238824,
-      "learning_rate": 0.00019999918562186865,
-      "loss": 46.0,
-      "step": 16812
-    },
-    {
-      "epoch": 1.2854712617313684,
-      "grad_norm": 0.0007503421511501074,
-      "learning_rate": 0.00019999918552492764,
-      "loss": 46.0,
-      "step": 16813
-    },
-    {
-      "epoch": 1.2855477187147581,
-      "grad_norm": 0.0012404288863763213,
-      "learning_rate": 0.00019999918542798088,
-      "loss": 46.0,
-      "step": 16814
-    },
-    {
-      "epoch": 1.285624175698148,
-      "grad_norm": 0.0015721333911642432,
-      "learning_rate": 0.00019999918533102838,
-      "loss": 46.0,
-      "step": 16815
-    },
-    {
-      "epoch": 1.2857006326815377,
-      "grad_norm": 0.0011280224425718188,
-      "learning_rate": 0.00019999918523407007,
-      "loss": 46.0,
-      "step": 16816
-    },
-    {
-      "epoch": 1.2857770896649272,
-      "grad_norm": 0.0013504691887646914,
-      "learning_rate": 0.000199999185137106,
-      "loss": 46.0,
-      "step": 16817
-    },
-    {
-      "epoch": 1.285853546648317,
-      "grad_norm": 0.001519326469860971,
-      "learning_rate": 0.00019999918504013615,
-      "loss": 46.0,
-      "step": 16818
-    },
-    {
-      "epoch": 1.2859300036317067,
-      "grad_norm": 0.011493152938783169,
-      "learning_rate": 0.00019999918494316058,
-      "loss": 46.0,
-      "step": 16819
-    },
-    {
-      "epoch": 1.2860064606150965,
-      "grad_norm": 0.0015346946893259883,
-      "learning_rate": 0.00019999918484617916,
-      "loss": 46.0,
-      "step": 16820
-    },
-    {
-      "epoch": 1.286082917598486,
-      "grad_norm": 0.005602580960839987,
-      "learning_rate": 0.00019999918474919204,
-      "loss": 46.0,
-      "step": 16821
-    },
-    {
-      "epoch": 1.2861593745818758,
-      "grad_norm": 0.0030354626942425966,
-      "learning_rate": 0.00019999918465219912,
-      "loss": 46.0,
-      "step": 16822
-    },
-    {
-      "epoch": 1.2862358315652656,
-      "grad_norm": 0.0030874423682689667,
-      "learning_rate": 0.00019999918455520043,
-      "loss": 46.0,
-      "step": 16823
-    },
-    {
-      "epoch": 1.2863122885486553,
-      "grad_norm": 0.00119735614862293,
-      "learning_rate": 0.000199999184458196,
-      "loss": 46.0,
-      "step": 16824
-    },
-    {
-      "epoch": 1.286388745532045,
-      "grad_norm": 0.0014501403784379363,
-      "learning_rate": 0.00019999918436118574,
-      "loss": 46.0,
-      "step": 16825
-    },
-    {
-      "epoch": 1.2864652025154348,
-      "grad_norm": 0.0013016184093430638,
-      "learning_rate": 0.00019999918426416978,
-      "loss": 46.0,
-      "step": 16826
-    },
-    {
-      "epoch": 1.2865416594988246,
-      "grad_norm": 0.00046273902989923954,
-      "learning_rate": 0.000199999184167148,
-      "loss": 46.0,
-      "step": 16827
-    },
-    {
-      "epoch": 1.2866181164822141,
-      "grad_norm": 0.0003741539840120822,
-      "learning_rate": 0.0001999991840701205,
-      "loss": 46.0,
-      "step": 16828
-    },
-    {
-      "epoch": 1.286694573465604,
-      "grad_norm": 0.0036199484020471573,
-      "learning_rate": 0.00019999918397308717,
-      "loss": 46.0,
-      "step": 16829
-    },
-    {
-      "epoch": 1.2867710304489937,
-      "grad_norm": 0.0005321458447724581,
-      "learning_rate": 0.0001999991838760481,
-      "loss": 46.0,
-      "step": 16830
-    },
-    {
-      "epoch": 1.2868474874323834,
-      "grad_norm": 0.001334922038950026,
-      "learning_rate": 0.00019999918377900327,
-      "loss": 46.0,
-      "step": 16831
-    },
-    {
-      "epoch": 1.286923944415773,
-      "grad_norm": 0.0008652047836221755,
-      "learning_rate": 0.00019999918368195264,
-      "loss": 46.0,
-      "step": 16832
-    },
-    {
-      "epoch": 1.2870004013991627,
-      "grad_norm": 0.0028909649699926376,
-      "learning_rate": 0.00019999918358489628,
-      "loss": 46.0,
-      "step": 16833
-    },
-    {
-      "epoch": 1.2870768583825525,
-      "grad_norm": 0.0012722325045615435,
-      "learning_rate": 0.00019999918348783413,
-      "loss": 46.0,
-      "step": 16834
-    },
-    {
-      "epoch": 1.2871533153659422,
-      "grad_norm": 0.0003154508594889194,
-      "learning_rate": 0.00019999918339076622,
-      "loss": 46.0,
-      "step": 16835
-    },
-    {
-      "epoch": 1.287229772349332,
-      "grad_norm": 0.0012075634440407157,
-      "learning_rate": 0.00019999918329369253,
-      "loss": 46.0,
-      "step": 16836
-    },
-    {
-      "epoch": 1.2873062293327218,
-      "grad_norm": 0.0005677743465639651,
-      "learning_rate": 0.00019999918319661307,
-      "loss": 46.0,
-      "step": 16837
-    },
-    {
-      "epoch": 1.2873826863161115,
-      "grad_norm": 0.0008385126129724085,
-      "learning_rate": 0.00019999918309952786,
-      "loss": 46.0,
-      "step": 16838
-    },
-    {
-      "epoch": 1.287459143299501,
-      "grad_norm": 0.0005314136506058276,
-      "learning_rate": 0.00019999918300243685,
-      "loss": 46.0,
-      "step": 16839
-    },
-    {
-      "epoch": 1.2875356002828908,
-      "grad_norm": 0.0015994678251445293,
-      "learning_rate": 0.00019999918290534006,
-      "loss": 46.0,
-      "step": 16840
-    },
-    {
-      "epoch": 1.2876120572662806,
-      "grad_norm": 0.0006531765102408826,
-      "learning_rate": 0.00019999918280823756,
-      "loss": 46.0,
-      "step": 16841
-    },
-    {
-      "epoch": 1.2876885142496703,
-      "grad_norm": 0.003536122152581811,
-      "learning_rate": 0.00019999918271112926,
-      "loss": 46.0,
-      "step": 16842
-    },
-    {
-      "epoch": 1.2877649712330599,
-      "grad_norm": 0.001129776705056429,
-      "learning_rate": 0.00019999918261401516,
-      "loss": 46.0,
-      "step": 16843
-    },
-    {
-      "epoch": 1.2878414282164496,
-      "grad_norm": 0.0003413664235267788,
-      "learning_rate": 0.00019999918251689534,
-      "loss": 46.0,
-      "step": 16844
-    },
-    {
-      "epoch": 1.2879178851998394,
-      "grad_norm": 0.0012691678712144494,
-      "learning_rate": 0.0001999991824197697,
-      "loss": 46.0,
-      "step": 16845
-    },
-    {
-      "epoch": 1.2879943421832292,
-      "grad_norm": 0.0020315167494118214,
-      "learning_rate": 0.00019999918232263832,
-      "loss": 46.0,
-      "step": 16846
-    },
-    {
-      "epoch": 1.288070799166619,
-      "grad_norm": 0.005115443374961615,
-      "learning_rate": 0.00019999918222550118,
-      "loss": 46.0,
-      "step": 16847
-    },
-    {
-      "epoch": 1.2881472561500087,
-      "grad_norm": 0.0011568405898287892,
-      "learning_rate": 0.00019999918212835826,
-      "loss": 46.0,
-      "step": 16848
-    },
-    {
-      "epoch": 1.2882237131333985,
-      "grad_norm": 0.02212458848953247,
-      "learning_rate": 0.00019999918203120957,
-      "loss": 46.0,
-      "step": 16849
-    },
-    {
-      "epoch": 1.288300170116788,
-      "grad_norm": 0.0038435286842286587,
-      "learning_rate": 0.0001999991819340551,
-      "loss": 46.0,
-      "step": 16850
-    },
-    {
-      "epoch": 1.2883766271001778,
-      "grad_norm": 0.0004945640102960169,
-      "learning_rate": 0.00019999918183689487,
-      "loss": 46.0,
-      "step": 16851
-    },
-    {
-      "epoch": 1.2884530840835675,
-      "grad_norm": 0.001998421037569642,
-      "learning_rate": 0.0001999991817397289,
-      "loss": 46.0,
-      "step": 16852
-    },
-    {
-      "epoch": 1.2885295410669573,
-      "grad_norm": 0.0010479297488927841,
-      "learning_rate": 0.00019999918164255714,
-      "loss": 46.0,
-      "step": 16853
-    },
-    {
-      "epoch": 1.2886059980503468,
-      "grad_norm": 0.0008820774964988232,
-      "learning_rate": 0.00019999918154537958,
-      "loss": 46.0,
-      "step": 16854
-    },
-    {
-      "epoch": 1.2886824550337366,
-      "grad_norm": 0.0012558652088046074,
-      "learning_rate": 0.0001999991814481963,
-      "loss": 46.0,
-      "step": 16855
-    },
-    {
-      "epoch": 1.2887589120171263,
-      "grad_norm": 0.0008983712759800255,
-      "learning_rate": 0.0001999991813510072,
-      "loss": 46.0,
-      "step": 16856
-    },
-    {
-      "epoch": 1.288835369000516,
-      "grad_norm": 0.0009755120845511556,
-      "learning_rate": 0.00019999918125381235,
-      "loss": 46.0,
-      "step": 16857
-    },
-    {
-      "epoch": 1.2889118259839059,
-      "grad_norm": 0.002427182625979185,
-      "learning_rate": 0.00019999918115661176,
-      "loss": 46.0,
-      "step": 16858
-    },
-    {
-      "epoch": 1.2889882829672956,
-      "grad_norm": 0.001303202356211841,
-      "learning_rate": 0.0001999991810594054,
-      "loss": 46.0,
-      "step": 16859
-    },
-    {
-      "epoch": 1.2890647399506854,
-      "grad_norm": 0.0004270741774234921,
-      "learning_rate": 0.00019999918096219322,
-      "loss": 46.0,
-      "step": 16860
-    },
-    {
-      "epoch": 1.289141196934075,
-      "grad_norm": 0.0030658901669085026,
-      "learning_rate": 0.0001999991808649753,
-      "loss": 46.0,
-      "step": 16861
-    },
-    {
-      "epoch": 1.2892176539174647,
-      "grad_norm": 0.004079016856849194,
-      "learning_rate": 0.00019999918076775162,
-      "loss": 46.0,
-      "step": 16862
-    },
-    {
-      "epoch": 1.2892941109008544,
-      "grad_norm": 0.0013592157047241926,
-      "learning_rate": 0.00019999918067052216,
-      "loss": 46.0,
-      "step": 16863
-    },
-    {
-      "epoch": 1.2893705678842442,
-      "grad_norm": 0.004121098667383194,
-      "learning_rate": 0.0001999991805732869,
-      "loss": 46.0,
-      "step": 16864
-    },
-    {
-      "epoch": 1.2894470248676337,
-      "grad_norm": 0.005282412748783827,
-      "learning_rate": 0.0001999991804760459,
-      "loss": 46.0,
-      "step": 16865
-    },
-    {
-      "epoch": 1.2895234818510235,
-      "grad_norm": 0.0012981975451111794,
-      "learning_rate": 0.00019999918037879916,
-      "loss": 46.0,
-      "step": 16866
-    },
-    {
-      "epoch": 1.2895999388344133,
-      "grad_norm": 0.0004448969557415694,
-      "learning_rate": 0.0001999991802815466,
-      "loss": 46.0,
-      "step": 16867
-    },
-    {
-      "epoch": 1.289676395817803,
-      "grad_norm": 0.001530495355837047,
-      "learning_rate": 0.0001999991801842883,
-      "loss": 46.0,
-      "step": 16868
-    },
-    {
-      "epoch": 1.2897528528011928,
-      "grad_norm": 0.0016936227912083268,
-      "learning_rate": 0.00019999918008702423,
-      "loss": 46.0,
-      "step": 16869
-    },
-    {
-      "epoch": 1.2898293097845825,
-      "grad_norm": 0.000989087508060038,
-      "learning_rate": 0.00019999917998975438,
-      "loss": 46.0,
-      "step": 16870
-    },
-    {
-      "epoch": 1.2899057667679723,
-      "grad_norm": 0.0010438649915158749,
-      "learning_rate": 0.00019999917989247876,
-      "loss": 46.0,
-      "step": 16871
-    },
-    {
-      "epoch": 1.2899822237513618,
-      "grad_norm": 0.0006205153185874224,
-      "learning_rate": 0.0001999991797951974,
-      "loss": 46.0,
-      "step": 16872
-    },
-    {
-      "epoch": 1.2900586807347516,
-      "grad_norm": 0.0044082277454435825,
-      "learning_rate": 0.00019999917969791022,
-      "loss": 46.0,
-      "step": 16873
-    },
-    {
-      "epoch": 1.2901351377181414,
-      "grad_norm": 0.0012309845769777894,
-      "learning_rate": 0.0001999991796006173,
-      "loss": 46.0,
-      "step": 16874
-    },
-    {
-      "epoch": 1.2902115947015311,
-      "grad_norm": 0.0016119048232212663,
-      "learning_rate": 0.00019999917950331862,
-      "loss": 46.0,
-      "step": 16875
-    },
-    {
-      "epoch": 1.2902880516849207,
-      "grad_norm": 0.0013921381905674934,
-      "learning_rate": 0.00019999917940601413,
-      "loss": 46.0,
-      "step": 16876
-    },
-    {
-      "epoch": 1.2903645086683104,
-      "grad_norm": 0.0010991496965289116,
-      "learning_rate": 0.0001999991793087039,
-      "loss": 46.0,
-      "step": 16877
-    },
-    {
-      "epoch": 1.2904409656517002,
-      "grad_norm": 0.0008678184822201729,
-      "learning_rate": 0.0001999991792113879,
-      "loss": 46.0,
-      "step": 16878
-    },
-    {
-      "epoch": 1.29051742263509,
-      "grad_norm": 0.0007273616502061486,
-      "learning_rate": 0.0001999991791140661,
-      "loss": 46.0,
-      "step": 16879
-    },
-    {
-      "epoch": 1.2905938796184797,
-      "grad_norm": 0.002260018140077591,
-      "learning_rate": 0.0001999991790167386,
-      "loss": 46.0,
-      "step": 16880
-    },
-    {
-      "epoch": 1.2906703366018695,
-      "grad_norm": 0.0005118157132528722,
-      "learning_rate": 0.0001999991789194053,
-      "loss": 46.0,
-      "step": 16881
-    },
-    {
-      "epoch": 1.290746793585259,
-      "grad_norm": 0.0017869130242615938,
-      "learning_rate": 0.0001999991788220662,
-      "loss": 46.0,
-      "step": 16882
-    },
-    {
-      "epoch": 1.2908232505686488,
-      "grad_norm": 0.0008558796253055334,
-      "learning_rate": 0.00019999917872472135,
-      "loss": 46.0,
-      "step": 16883
-    },
-    {
-      "epoch": 1.2908997075520385,
-      "grad_norm": 0.0007911978755146265,
-      "learning_rate": 0.00019999917862737073,
-      "loss": 46.0,
-      "step": 16884
-    },
-    {
-      "epoch": 1.2909761645354283,
-      "grad_norm": 0.002293990459293127,
-      "learning_rate": 0.00019999917853001433,
-      "loss": 46.0,
-      "step": 16885
-    },
-    {
-      "epoch": 1.291052621518818,
-      "grad_norm": 0.003016027621924877,
-      "learning_rate": 0.0001999991784326522,
-      "loss": 46.0,
-      "step": 16886
-    },
-    {
-      "epoch": 1.2911290785022076,
-      "grad_norm": 0.0010313489474356174,
-      "learning_rate": 0.00019999917833528428,
-      "loss": 46.0,
-      "step": 16887
-    },
-    {
-      "epoch": 1.2912055354855974,
-      "grad_norm": 0.001564008998684585,
-      "learning_rate": 0.00019999917823791056,
-      "loss": 46.0,
-      "step": 16888
-    },
-    {
-      "epoch": 1.2912819924689871,
-      "grad_norm": 0.001408661832101643,
-      "learning_rate": 0.00019999917814053107,
-      "loss": 46.0,
-      "step": 16889
-    },
-    {
-      "epoch": 1.2913584494523769,
-      "grad_norm": 0.0010653232457116246,
-      "learning_rate": 0.00019999917804314587,
-      "loss": 46.0,
-      "step": 16890
-    },
-    {
-      "epoch": 1.2914349064357666,
-      "grad_norm": 0.002050354378297925,
-      "learning_rate": 0.00019999917794575486,
-      "loss": 46.0,
-      "step": 16891
-    },
-    {
-      "epoch": 1.2915113634191564,
-      "grad_norm": 0.0010547222336754203,
-      "learning_rate": 0.00019999917784835808,
-      "loss": 46.0,
-      "step": 16892
-    },
-    {
-      "epoch": 1.291587820402546,
-      "grad_norm": 0.002576560713350773,
-      "learning_rate": 0.00019999917775095555,
-      "loss": 46.0,
-      "step": 16893
-    },
-    {
-      "epoch": 1.2916642773859357,
-      "grad_norm": 0.0010254953522235155,
-      "learning_rate": 0.00019999917765354722,
-      "loss": 46.0,
-      "step": 16894
-    },
-    {
-      "epoch": 1.2917407343693255,
-      "grad_norm": 0.0006489516817964613,
-      "learning_rate": 0.00019999917755613315,
-      "loss": 46.0,
-      "step": 16895
-    },
-    {
-      "epoch": 1.2918171913527152,
-      "grad_norm": 0.001210812944918871,
-      "learning_rate": 0.0001999991774587133,
-      "loss": 46.0,
-      "step": 16896
-    },
-    {
-      "epoch": 1.2918936483361048,
-      "grad_norm": 0.002377795986831188,
-      "learning_rate": 0.00019999917736128768,
-      "loss": 46.0,
-      "step": 16897
-    },
-    {
-      "epoch": 1.2919701053194945,
-      "grad_norm": 0.00047870652633719146,
-      "learning_rate": 0.00019999917726385628,
-      "loss": 46.0,
-      "step": 16898
-    },
-    {
-      "epoch": 1.2920465623028843,
-      "grad_norm": 0.0013301590224727988,
-      "learning_rate": 0.00019999917716641914,
-      "loss": 46.0,
-      "step": 16899
-    },
-    {
-      "epoch": 1.292123019286274,
-      "grad_norm": 0.0005451434990391135,
-      "learning_rate": 0.0001999991770689762,
-      "loss": 46.0,
-      "step": 16900
-    },
-    {
-      "epoch": 1.2921994762696638,
-      "grad_norm": 0.0005562631413340569,
-      "learning_rate": 0.00019999917697152751,
-      "loss": 46.0,
-      "step": 16901
-    },
-    {
-      "epoch": 1.2922759332530536,
-      "grad_norm": 0.0021114330738782883,
-      "learning_rate": 0.00019999917687407303,
-      "loss": 46.0,
-      "step": 16902
-    },
-    {
-      "epoch": 1.2923523902364433,
-      "grad_norm": 0.002595886355265975,
-      "learning_rate": 0.0001999991767766128,
-      "loss": 46.0,
-      "step": 16903
-    },
-    {
-      "epoch": 1.2924288472198329,
-      "grad_norm": 0.0013018480967730284,
-      "learning_rate": 0.00019999917667914678,
-      "loss": 46.0,
-      "step": 16904
-    },
-    {
-      "epoch": 1.2925053042032226,
-      "grad_norm": 0.003693181788548827,
-      "learning_rate": 0.00019999917658167503,
-      "loss": 46.0,
-      "step": 16905
-    },
-    {
-      "epoch": 1.2925817611866124,
-      "grad_norm": 0.0013611132744699717,
-      "learning_rate": 0.00019999917648419748,
-      "loss": 46.0,
-      "step": 16906
-    },
-    {
-      "epoch": 1.2926582181700021,
-      "grad_norm": 0.0011091484921053052,
-      "learning_rate": 0.00019999917638671417,
-      "loss": 46.0,
-      "step": 16907
-    },
-    {
-      "epoch": 1.2927346751533917,
-      "grad_norm": 0.0007690757629461586,
-      "learning_rate": 0.0001999991762892251,
-      "loss": 46.0,
-      "step": 16908
-    },
-    {
-      "epoch": 1.2928111321367814,
-      "grad_norm": 0.006611744873225689,
-      "learning_rate": 0.00019999917619173023,
-      "loss": 46.0,
-      "step": 16909
-    },
-    {
-      "epoch": 1.2928875891201712,
-      "grad_norm": 0.0007839055033400655,
-      "learning_rate": 0.0001999991760942296,
-      "loss": 46.0,
-      "step": 16910
-    },
-    {
-      "epoch": 1.292964046103561,
-      "grad_norm": 0.000695764843840152,
-      "learning_rate": 0.0001999991759967232,
-      "loss": 46.0,
-      "step": 16911
-    },
-    {
-      "epoch": 1.2930405030869507,
-      "grad_norm": 0.002346856752410531,
-      "learning_rate": 0.00019999917589921104,
-      "loss": 46.0,
-      "step": 16912
-    },
-    {
-      "epoch": 1.2931169600703405,
-      "grad_norm": 0.0006746028084307909,
-      "learning_rate": 0.00019999917580169313,
-      "loss": 46.0,
-      "step": 16913
-    },
-    {
-      "epoch": 1.2931934170537303,
-      "grad_norm": 0.0008049709722399712,
-      "learning_rate": 0.0001999991757041694,
-      "loss": 46.0,
-      "step": 16914
-    },
-    {
-      "epoch": 1.2932698740371198,
-      "grad_norm": 0.0037269913591444492,
-      "learning_rate": 0.00019999917560663995,
-      "loss": 46.0,
-      "step": 16915
-    },
-    {
-      "epoch": 1.2933463310205096,
-      "grad_norm": 0.0007973831379786134,
-      "learning_rate": 0.0001999991755091047,
-      "loss": 46.0,
-      "step": 16916
-    },
-    {
-      "epoch": 1.2934227880038993,
-      "grad_norm": 0.0016177480574697256,
-      "learning_rate": 0.00019999917541156369,
-      "loss": 46.0,
-      "step": 16917
-    },
-    {
-      "epoch": 1.293499244987289,
-      "grad_norm": 0.0010194308124482632,
-      "learning_rate": 0.0001999991753140169,
-      "loss": 46.0,
-      "step": 16918
-    },
-    {
-      "epoch": 1.2935757019706786,
-      "grad_norm": 0.00223959656432271,
-      "learning_rate": 0.00019999917521646435,
-      "loss": 46.0,
-      "step": 16919
-    },
-    {
-      "epoch": 1.2936521589540684,
-      "grad_norm": 0.0007535201730206609,
-      "learning_rate": 0.00019999917511890605,
-      "loss": 46.0,
-      "step": 16920
-    },
-    {
-      "epoch": 1.2937286159374581,
-      "grad_norm": 0.0009935133857652545,
-      "learning_rate": 0.00019999917502134195,
-      "loss": 46.0,
-      "step": 16921
-    },
-    {
-      "epoch": 1.293805072920848,
-      "grad_norm": 0.0007746369228698313,
-      "learning_rate": 0.0001999991749237721,
-      "loss": 46.0,
-      "step": 16922
-    },
-    {
-      "epoch": 1.2938815299042377,
-      "grad_norm": 0.001072959857992828,
-      "learning_rate": 0.00019999917482619648,
-      "loss": 46.0,
-      "step": 16923
-    },
-    {
-      "epoch": 1.2939579868876274,
-      "grad_norm": 0.0007886750972829759,
-      "learning_rate": 0.0001999991747286151,
-      "loss": 46.0,
-      "step": 16924
-    },
-    {
-      "epoch": 1.2940344438710172,
-      "grad_norm": 0.0013376976130530238,
-      "learning_rate": 0.00019999917463102792,
-      "loss": 46.0,
-      "step": 16925
-    },
-    {
-      "epoch": 1.2941109008544067,
-      "grad_norm": 0.0019853590056300163,
-      "learning_rate": 0.000199999174533435,
-      "loss": 46.0,
-      "step": 16926
-    },
-    {
-      "epoch": 1.2941873578377965,
-      "grad_norm": 0.0013454477302730083,
-      "learning_rate": 0.0001999991744358363,
-      "loss": 46.0,
-      "step": 16927
-    },
-    {
-      "epoch": 1.2942638148211862,
-      "grad_norm": 0.0008105112356133759,
-      "learning_rate": 0.0001999991743382318,
-      "loss": 46.0,
-      "step": 16928
-    },
-    {
-      "epoch": 1.294340271804576,
-      "grad_norm": 0.0012198216281831264,
-      "learning_rate": 0.00019999917424062157,
-      "loss": 46.0,
-      "step": 16929
-    },
-    {
-      "epoch": 1.2944167287879655,
-      "grad_norm": 0.0009669942664913833,
-      "learning_rate": 0.00019999917414300557,
-      "loss": 46.0,
-      "step": 16930
-    },
-    {
-      "epoch": 1.2944931857713553,
-      "grad_norm": 0.0006907760398462415,
-      "learning_rate": 0.00019999917404538378,
-      "loss": 46.0,
-      "step": 16931
-    },
-    {
-      "epoch": 1.294569642754745,
-      "grad_norm": 0.0011938433162868023,
-      "learning_rate": 0.00019999917394775626,
-      "loss": 46.0,
-      "step": 16932
-    },
-    {
-      "epoch": 1.2946460997381348,
-      "grad_norm": 0.002145028207451105,
-      "learning_rate": 0.00019999917385012293,
-      "loss": 46.0,
-      "step": 16933
-    },
-    {
-      "epoch": 1.2947225567215246,
-      "grad_norm": 0.0037043741904199123,
-      "learning_rate": 0.00019999917375248383,
-      "loss": 46.0,
-      "step": 16934
-    },
-    {
-      "epoch": 1.2947990137049143,
-      "grad_norm": 0.0011335520539432764,
-      "learning_rate": 0.00019999917365483896,
-      "loss": 46.0,
-      "step": 16935
-    },
-    {
-      "epoch": 1.294875470688304,
-      "grad_norm": 0.0009578640456311405,
-      "learning_rate": 0.00019999917355718834,
-      "loss": 46.0,
-      "step": 16936
-    },
-    {
-      "epoch": 1.2949519276716936,
-      "grad_norm": 0.0009613609872758389,
-      "learning_rate": 0.00019999917345953194,
-      "loss": 46.0,
-      "step": 16937
-    },
-    {
-      "epoch": 1.2950283846550834,
-      "grad_norm": 0.0034558584447950125,
-      "learning_rate": 0.00019999917336186978,
-      "loss": 46.0,
-      "step": 16938
-    },
-    {
-      "epoch": 1.2951048416384732,
-      "grad_norm": 0.0004144326376263052,
-      "learning_rate": 0.00019999917326420184,
-      "loss": 46.0,
-      "step": 16939
-    },
-    {
-      "epoch": 1.295181298621863,
-      "grad_norm": 0.005028862506151199,
-      "learning_rate": 0.00019999917316652815,
-      "loss": 46.0,
-      "step": 16940
-    },
-    {
-      "epoch": 1.2952577556052525,
-      "grad_norm": 0.0014026965945959091,
-      "learning_rate": 0.0001999991730688487,
-      "loss": 46.0,
-      "step": 16941
-    },
-    {
-      "epoch": 1.2953342125886422,
-      "grad_norm": 0.0018483380554243922,
-      "learning_rate": 0.00019999917297116343,
-      "loss": 46.0,
-      "step": 16942
-    },
-    {
-      "epoch": 1.295410669572032,
-      "grad_norm": 0.001309492508880794,
-      "learning_rate": 0.0001999991728734724,
-      "loss": 46.0,
-      "step": 16943
-    },
-    {
-      "epoch": 1.2954871265554218,
-      "grad_norm": 0.0041581010445952415,
-      "learning_rate": 0.00019999917277577561,
-      "loss": 46.0,
-      "step": 16944
-    },
-    {
-      "epoch": 1.2955635835388115,
-      "grad_norm": 0.0006954861455596983,
-      "learning_rate": 0.00019999917267807306,
-      "loss": 46.0,
-      "step": 16945
-    },
-    {
-      "epoch": 1.2956400405222013,
-      "grad_norm": 0.010062218643724918,
-      "learning_rate": 0.00019999917258036476,
-      "loss": 46.0,
-      "step": 16946
-    },
-    {
-      "epoch": 1.295716497505591,
-      "grad_norm": 0.0006993044517003,
-      "learning_rate": 0.00019999917248265066,
-      "loss": 46.0,
-      "step": 16947
-    },
-    {
-      "epoch": 1.2957929544889806,
-      "grad_norm": 0.01372715923935175,
-      "learning_rate": 0.0001999991723849308,
-      "loss": 46.0,
-      "step": 16948
-    },
-    {
-      "epoch": 1.2958694114723703,
-      "grad_norm": 0.0014995898818597198,
-      "learning_rate": 0.00019999917228720517,
-      "loss": 46.0,
-      "step": 16949
-    },
-    {
-      "epoch": 1.29594586845576,
-      "grad_norm": 0.0010075741447508335,
-      "learning_rate": 0.00019999917218947377,
-      "loss": 46.0,
-      "step": 16950
-    },
-    {
-      "epoch": 1.2960223254391499,
-      "grad_norm": 0.0011956244707107544,
-      "learning_rate": 0.0001999991720917366,
-      "loss": 46.0,
-      "step": 16951
-    },
-    {
-      "epoch": 1.2960987824225394,
-      "grad_norm": 0.001037903013639152,
-      "learning_rate": 0.00019999917199399367,
-      "loss": 46.0,
-      "step": 16952
-    },
-    {
-      "epoch": 1.2961752394059292,
-      "grad_norm": 0.0011192620731890202,
-      "learning_rate": 0.00019999917189624495,
-      "loss": 46.0,
-      "step": 16953
-    },
-    {
-      "epoch": 1.296251696389319,
-      "grad_norm": 0.004891249351203442,
-      "learning_rate": 0.0001999991717984905,
-      "loss": 46.0,
-      "step": 16954
-    },
-    {
-      "epoch": 1.2963281533727087,
-      "grad_norm": 0.0010662992717698216,
-      "learning_rate": 0.0001999991717007302,
-      "loss": 46.0,
-      "step": 16955
-    },
-    {
-      "epoch": 1.2964046103560984,
-      "grad_norm": 0.003124087816104293,
-      "learning_rate": 0.00019999917160296423,
-      "loss": 46.0,
-      "step": 16956
-    },
-    {
-      "epoch": 1.2964810673394882,
-      "grad_norm": 0.0015029701171442866,
-      "learning_rate": 0.00019999917150519242,
-      "loss": 46.0,
-      "step": 16957
-    },
-    {
-      "epoch": 1.296557524322878,
-      "grad_norm": 0.005907348357141018,
-      "learning_rate": 0.0001999991714074149,
-      "loss": 46.0,
-      "step": 16958
-    },
-    {
-      "epoch": 1.2966339813062675,
-      "grad_norm": 0.0004651906783692539,
-      "learning_rate": 0.00019999917130963154,
-      "loss": 46.0,
-      "step": 16959
-    },
-    {
-      "epoch": 1.2967104382896573,
-      "grad_norm": 0.0007895607850514352,
-      "learning_rate": 0.00019999917121184247,
-      "loss": 46.0,
-      "step": 16960
-    },
-    {
-      "epoch": 1.296786895273047,
-      "grad_norm": 0.0018282203236594796,
-      "learning_rate": 0.0001999991711140476,
-      "loss": 46.0,
-      "step": 16961
-    },
-    {
-      "epoch": 1.2968633522564368,
-      "grad_norm": 0.002710596891120076,
-      "learning_rate": 0.00019999917101624697,
-      "loss": 46.0,
-      "step": 16962
-    },
-    {
-      "epoch": 1.2969398092398263,
-      "grad_norm": 0.0019121470395475626,
-      "learning_rate": 0.00019999917091844055,
-      "loss": 46.0,
-      "step": 16963
-    },
-    {
-      "epoch": 1.297016266223216,
-      "grad_norm": 0.0026673662941902876,
-      "learning_rate": 0.0001999991708206284,
-      "loss": 46.0,
-      "step": 16964
-    },
-    {
-      "epoch": 1.2970927232066058,
-      "grad_norm": 0.0010244291042909026,
-      "learning_rate": 0.00019999917072281045,
-      "loss": 46.0,
-      "step": 16965
-    },
-    {
-      "epoch": 1.2971691801899956,
-      "grad_norm": 0.002440034644678235,
-      "learning_rate": 0.00019999917062498674,
-      "loss": 46.0,
-      "step": 16966
-    },
-    {
-      "epoch": 1.2972456371733854,
-      "grad_norm": 0.000626893131993711,
-      "learning_rate": 0.00019999917052715728,
-      "loss": 46.0,
-      "step": 16967
-    },
-    {
-      "epoch": 1.2973220941567751,
-      "grad_norm": 0.0013734777458012104,
-      "learning_rate": 0.00019999917042932202,
-      "loss": 46.0,
-      "step": 16968
-    },
-    {
-      "epoch": 1.297398551140165,
-      "grad_norm": 0.0008380646468140185,
-      "learning_rate": 0.000199999170331481,
-      "loss": 46.0,
-      "step": 16969
-    },
-    {
-      "epoch": 1.2974750081235544,
-      "grad_norm": 0.0020817413460463285,
-      "learning_rate": 0.0001999991702336342,
-      "loss": 46.0,
-      "step": 16970
-    },
-    {
-      "epoch": 1.2975514651069442,
-      "grad_norm": 0.006914359051734209,
-      "learning_rate": 0.00019999917013578165,
-      "loss": 46.0,
-      "step": 16971
-    },
-    {
-      "epoch": 1.297627922090334,
-      "grad_norm": 0.0016369130462408066,
-      "learning_rate": 0.00019999917003792332,
-      "loss": 46.0,
-      "step": 16972
-    },
-    {
-      "epoch": 1.2977043790737237,
-      "grad_norm": 0.0008318163454532623,
-      "learning_rate": 0.00019999916994005922,
-      "loss": 46.0,
-      "step": 16973
-    },
-    {
-      "epoch": 1.2977808360571133,
-      "grad_norm": 0.002666981192305684,
-      "learning_rate": 0.00019999916984218938,
-      "loss": 46.0,
-      "step": 16974
-    },
-    {
-      "epoch": 1.297857293040503,
-      "grad_norm": 0.000722285476513207,
-      "learning_rate": 0.00019999916974431373,
-      "loss": 46.0,
-      "step": 16975
-    },
-    {
-      "epoch": 1.2979337500238928,
-      "grad_norm": 0.0008893163176253438,
-      "learning_rate": 0.00019999916964643234,
-      "loss": 46.0,
-      "step": 16976
-    },
-    {
-      "epoch": 1.2980102070072825,
-      "grad_norm": 0.007473403587937355,
-      "learning_rate": 0.00019999916954854517,
-      "loss": 46.0,
-      "step": 16977
-    },
-    {
-      "epoch": 1.2980866639906723,
-      "grad_norm": 0.0036508669145405293,
-      "learning_rate": 0.00019999916945065223,
-      "loss": 46.0,
-      "step": 16978
-    },
-    {
-      "epoch": 1.298163120974062,
-      "grad_norm": 0.0019909648690372705,
-      "learning_rate": 0.00019999916935275352,
-      "loss": 46.0,
-      "step": 16979
-    },
-    {
-      "epoch": 1.2982395779574518,
-      "grad_norm": 0.005040519405156374,
-      "learning_rate": 0.00019999916925484904,
-      "loss": 46.0,
-      "step": 16980
-    },
-    {
-      "epoch": 1.2983160349408414,
-      "grad_norm": 0.0005798061029054224,
-      "learning_rate": 0.00019999916915693878,
-      "loss": 46.0,
-      "step": 16981
-    },
-    {
-      "epoch": 1.2983924919242311,
-      "grad_norm": 0.004145279061049223,
-      "learning_rate": 0.00019999916905902274,
-      "loss": 46.0,
-      "step": 16982
-    },
-    {
-      "epoch": 1.2984689489076209,
-      "grad_norm": 0.0011156536638736725,
-      "learning_rate": 0.00019999916896110096,
-      "loss": 46.0,
-      "step": 16983
-    },
-    {
-      "epoch": 1.2985454058910106,
-      "grad_norm": 0.0006065593333914876,
-      "learning_rate": 0.0001999991688631734,
-      "loss": 46.0,
-      "step": 16984
-    },
-    {
-      "epoch": 1.2986218628744002,
-      "grad_norm": 0.0014073837082833052,
-      "learning_rate": 0.00019999916876524009,
-      "loss": 46.0,
-      "step": 16985
-    },
-    {
-      "epoch": 1.29869831985779,
-      "grad_norm": 0.0006809005280956626,
-      "learning_rate": 0.00019999916866730099,
-      "loss": 46.0,
-      "step": 16986
-    },
-    {
-      "epoch": 1.2987747768411797,
-      "grad_norm": 0.001131216180510819,
-      "learning_rate": 0.0001999991685693561,
-      "loss": 46.0,
-      "step": 16987
-    },
-    {
-      "epoch": 1.2988512338245695,
-      "grad_norm": 0.0003314427740406245,
-      "learning_rate": 0.00019999916847140547,
-      "loss": 46.0,
-      "step": 16988
-    },
-    {
-      "epoch": 1.2989276908079592,
-      "grad_norm": 0.002919338643550873,
-      "learning_rate": 0.00019999916837344905,
-      "loss": 46.0,
-      "step": 16989
-    },
-    {
-      "epoch": 1.299004147791349,
-      "grad_norm": 0.004110758192837238,
-      "learning_rate": 0.00019999916827548688,
-      "loss": 46.0,
-      "step": 16990
-    },
-    {
-      "epoch": 1.2990806047747387,
-      "grad_norm": 0.00038438753108493984,
-      "learning_rate": 0.00019999916817751894,
-      "loss": 46.0,
-      "step": 16991
-    },
-    {
-      "epoch": 1.2991570617581283,
-      "grad_norm": 0.001421051798388362,
-      "learning_rate": 0.00019999916807954523,
-      "loss": 46.0,
-      "step": 16992
-    },
-    {
-      "epoch": 1.299233518741518,
-      "grad_norm": 0.0003969838726334274,
-      "learning_rate": 0.00019999916798156574,
-      "loss": 46.0,
-      "step": 16993
-    },
-    {
-      "epoch": 1.2993099757249078,
-      "grad_norm": 0.003521139267832041,
-      "learning_rate": 0.0001999991678835805,
-      "loss": 46.0,
-      "step": 16994
-    },
-    {
-      "epoch": 1.2993864327082976,
-      "grad_norm": 0.0036777553614228964,
-      "learning_rate": 0.00019999916778558948,
-      "loss": 46.0,
-      "step": 16995
-    },
-    {
-      "epoch": 1.299462889691687,
-      "grad_norm": 0.0008138920529745519,
-      "learning_rate": 0.00019999916768759267,
-      "loss": 46.0,
-      "step": 16996
-    },
-    {
-      "epoch": 1.2995393466750769,
-      "grad_norm": 0.003016428090631962,
-      "learning_rate": 0.00019999916758959012,
-      "loss": 46.0,
-      "step": 16997
-    },
-    {
-      "epoch": 1.2996158036584666,
-      "grad_norm": 0.0014221933670341969,
-      "learning_rate": 0.0001999991674915818,
-      "loss": 46.0,
-      "step": 16998
-    },
-    {
-      "epoch": 1.2996922606418564,
-      "grad_norm": 0.002771981991827488,
-      "learning_rate": 0.00019999916739356767,
-      "loss": 46.0,
-      "step": 16999
-    },
-    {
-      "epoch": 1.2997687176252462,
-      "grad_norm": 0.0006767380400560796,
-      "learning_rate": 0.00019999916729554782,
-      "loss": 46.0,
-      "step": 17000
-    },
-    {
-      "epoch": 1.299845174608636,
-      "grad_norm": 0.0018132395343855023,
-      "learning_rate": 0.00019999916719752218,
-      "loss": 46.0,
-      "step": 17001
-    },
-    {
-      "epoch": 1.2999216315920257,
-      "grad_norm": 0.002206960925832391,
-      "learning_rate": 0.0001999991670994908,
-      "loss": 46.0,
-      "step": 17002
-    },
-    {
-      "epoch": 1.2999980885754152,
-      "grad_norm": 0.0010303910821676254,
-      "learning_rate": 0.0001999991670014536,
-      "loss": 46.0,
-      "step": 17003
-    },
-    {
-      "epoch": 1.300074545558805,
-      "grad_norm": 0.00044133688788861036,
-      "learning_rate": 0.00019999916690341066,
-      "loss": 46.0,
-      "step": 17004
-    },
-    {
-      "epoch": 1.3001510025421947,
-      "grad_norm": 0.0031654739286750555,
-      "learning_rate": 0.00019999916680536192,
-      "loss": 46.0,
-      "step": 17005
-    },
-    {
-      "epoch": 1.3002274595255845,
-      "grad_norm": 0.0017270793905481696,
-      "learning_rate": 0.00019999916670730743,
-      "loss": 46.0,
-      "step": 17006
-    },
-    {
-      "epoch": 1.300303916508974,
-      "grad_norm": 0.0026358310133218765,
-      "learning_rate": 0.0001999991666092472,
-      "loss": 46.0,
-      "step": 17007
-    },
-    {
-      "epoch": 1.3003803734923638,
-      "grad_norm": 0.0010879419278353453,
-      "learning_rate": 0.00019999916651118117,
-      "loss": 46.0,
-      "step": 17008
-    },
-    {
-      "epoch": 1.3004568304757536,
-      "grad_norm": 0.0011486713774502277,
-      "learning_rate": 0.0001999991664131094,
-      "loss": 46.0,
-      "step": 17009
-    },
-    {
-      "epoch": 1.3005332874591433,
-      "grad_norm": 0.000679433869663626,
-      "learning_rate": 0.00019999916631503178,
-      "loss": 46.0,
-      "step": 17010
-    },
-    {
-      "epoch": 1.300609744442533,
-      "grad_norm": 0.003507207613438368,
-      "learning_rate": 0.00019999916621694848,
-      "loss": 46.0,
-      "step": 17011
-    },
-    {
-      "epoch": 1.3006862014259228,
-      "grad_norm": 0.0022981667425483465,
-      "learning_rate": 0.00019999916611885938,
-      "loss": 46.0,
-      "step": 17012
-    },
-    {
-      "epoch": 1.3007626584093124,
-      "grad_norm": 0.0037976689636707306,
-      "learning_rate": 0.00019999916602076449,
-      "loss": 46.0,
-      "step": 17013
-    },
-    {
-      "epoch": 1.3008391153927021,
-      "grad_norm": 0.0007833215058781207,
-      "learning_rate": 0.00019999916592266387,
-      "loss": 46.0,
-      "step": 17014
-    },
-    {
-      "epoch": 1.300915572376092,
-      "grad_norm": 0.008463768288493156,
-      "learning_rate": 0.00019999916582455745,
-      "loss": 46.0,
-      "step": 17015
-    },
-    {
-      "epoch": 1.3009920293594817,
-      "grad_norm": 0.000680348020978272,
-      "learning_rate": 0.00019999916572644528,
-      "loss": 46.0,
-      "step": 17016
-    },
-    {
-      "epoch": 1.3010684863428714,
-      "grad_norm": 0.0005476119695231318,
-      "learning_rate": 0.00019999916562832732,
-      "loss": 46.0,
-      "step": 17017
-    },
-    {
-      "epoch": 1.301144943326261,
-      "grad_norm": 0.0007602603873237967,
-      "learning_rate": 0.00019999916553020358,
-      "loss": 46.0,
-      "step": 17018
-    },
-    {
-      "epoch": 1.3012214003096507,
-      "grad_norm": 0.00031254449277184904,
-      "learning_rate": 0.00019999916543207412,
-      "loss": 46.0,
-      "step": 17019
-    },
-    {
-      "epoch": 1.3012978572930405,
-      "grad_norm": 0.0008787038386799395,
-      "learning_rate": 0.00019999916533393886,
-      "loss": 46.0,
-      "step": 17020
-    },
-    {
-      "epoch": 1.3013743142764302,
-      "grad_norm": 0.015780337154865265,
-      "learning_rate": 0.00019999916523579783,
-      "loss": 46.0,
-      "step": 17021
-    },
-    {
-      "epoch": 1.30145077125982,
-      "grad_norm": 0.0013650988694280386,
-      "learning_rate": 0.00019999916513765102,
-      "loss": 46.0,
-      "step": 17022
-    },
-    {
-      "epoch": 1.3015272282432098,
-      "grad_norm": 0.004020048771053553,
-      "learning_rate": 0.00019999916503949847,
-      "loss": 46.0,
-      "step": 17023
-    },
-    {
-      "epoch": 1.3016036852265993,
-      "grad_norm": 0.0007510558352805674,
-      "learning_rate": 0.00019999916494134015,
-      "loss": 46.0,
-      "step": 17024
-    },
-    {
-      "epoch": 1.301680142209989,
-      "grad_norm": 0.0009251527953892946,
-      "learning_rate": 0.00019999916484317602,
-      "loss": 46.0,
-      "step": 17025
-    },
-    {
-      "epoch": 1.3017565991933788,
-      "grad_norm": 0.0006468918872997165,
-      "learning_rate": 0.00019999916474500615,
-      "loss": 46.0,
-      "step": 17026
-    },
-    {
-      "epoch": 1.3018330561767686,
-      "grad_norm": 0.002583061344921589,
-      "learning_rate": 0.0001999991646468305,
-      "loss": 46.0,
-      "step": 17027
-    },
-    {
-      "epoch": 1.3019095131601581,
-      "grad_norm": 0.0011829755967482924,
-      "learning_rate": 0.00019999916454864911,
-      "loss": 46.0,
-      "step": 17028
-    },
-    {
-      "epoch": 1.3019859701435479,
-      "grad_norm": 0.0015390904154628515,
-      "learning_rate": 0.00019999916445046192,
-      "loss": 46.0,
-      "step": 17029
-    },
-    {
-      "epoch": 1.3020624271269376,
-      "grad_norm": 0.001295155263505876,
-      "learning_rate": 0.00019999916435226896,
-      "loss": 46.0,
-      "step": 17030
-    },
-    {
-      "epoch": 1.3021388841103274,
-      "grad_norm": 0.0013396060094237328,
-      "learning_rate": 0.00019999916425407025,
-      "loss": 46.0,
-      "step": 17031
-    },
-    {
-      "epoch": 1.3022153410937172,
-      "grad_norm": 0.004085657186806202,
-      "learning_rate": 0.00019999916415586576,
-      "loss": 46.0,
-      "step": 17032
-    },
-    {
-      "epoch": 1.302291798077107,
-      "grad_norm": 0.0021983031183481216,
-      "learning_rate": 0.00019999916405765548,
-      "loss": 46.0,
-      "step": 17033
-    },
-    {
-      "epoch": 1.3023682550604967,
-      "grad_norm": 0.0012879796558991075,
-      "learning_rate": 0.00019999916395943944,
-      "loss": 46.0,
-      "step": 17034
-    },
-    {
-      "epoch": 1.3024447120438862,
-      "grad_norm": 0.0009685040568001568,
-      "learning_rate": 0.00019999916386121767,
-      "loss": 46.0,
-      "step": 17035
-    },
-    {
-      "epoch": 1.302521169027276,
-      "grad_norm": 0.0006651482544839382,
-      "learning_rate": 0.00019999916376299011,
-      "loss": 46.0,
-      "step": 17036
-    },
-    {
-      "epoch": 1.3025976260106658,
-      "grad_norm": 0.0007602511323057115,
-      "learning_rate": 0.00019999916366475676,
-      "loss": 46.0,
-      "step": 17037
-    },
-    {
-      "epoch": 1.3026740829940555,
-      "grad_norm": 0.003833473427221179,
-      "learning_rate": 0.00019999916356651764,
-      "loss": 46.0,
-      "step": 17038
-    },
-    {
-      "epoch": 1.302750539977445,
-      "grad_norm": 0.0036222226917743683,
-      "learning_rate": 0.0001999991634682728,
-      "loss": 46.0,
-      "step": 17039
-    },
-    {
-      "epoch": 1.3028269969608348,
-      "grad_norm": 0.0024131571408361197,
-      "learning_rate": 0.00019999916337002215,
-      "loss": 46.0,
-      "step": 17040
-    },
-    {
-      "epoch": 1.3029034539442246,
-      "grad_norm": 0.0005607795319519937,
-      "learning_rate": 0.0001999991632717657,
-      "loss": 46.0,
-      "step": 17041
-    },
-    {
-      "epoch": 1.3029799109276143,
-      "grad_norm": 0.0035813169088214636,
-      "learning_rate": 0.0001999991631735035,
-      "loss": 46.0,
-      "step": 17042
-    },
-    {
-      "epoch": 1.303056367911004,
-      "grad_norm": 0.0017612132942304015,
-      "learning_rate": 0.00019999916307523557,
-      "loss": 46.0,
-      "step": 17043
-    },
-    {
-      "epoch": 1.3031328248943939,
-      "grad_norm": 0.0031446327921003103,
-      "learning_rate": 0.00019999916297696184,
-      "loss": 46.0,
-      "step": 17044
-    },
-    {
-      "epoch": 1.3032092818777836,
-      "grad_norm": 0.0038262035232037306,
-      "learning_rate": 0.00019999916287868238,
-      "loss": 46.0,
-      "step": 17045
-    },
-    {
-      "epoch": 1.3032857388611732,
-      "grad_norm": 0.0005556940450333059,
-      "learning_rate": 0.0001999991627803971,
-      "loss": 46.0,
-      "step": 17046
-    },
-    {
-      "epoch": 1.303362195844563,
-      "grad_norm": 0.0035446039400994778,
-      "learning_rate": 0.00019999916268210606,
-      "loss": 46.0,
-      "step": 17047
-    },
-    {
-      "epoch": 1.3034386528279527,
-      "grad_norm": 0.0019526415271684527,
-      "learning_rate": 0.00019999916258380926,
-      "loss": 46.0,
-      "step": 17048
-    },
-    {
-      "epoch": 1.3035151098113424,
-      "grad_norm": 0.002811564365401864,
-      "learning_rate": 0.0001999991624855067,
-      "loss": 46.0,
-      "step": 17049
-    },
-    {
-      "epoch": 1.303591566794732,
-      "grad_norm": 0.0012579062022268772,
-      "learning_rate": 0.00019999916238719835,
-      "loss": 46.0,
-      "step": 17050
-    },
-    {
-      "epoch": 1.3036680237781217,
-      "grad_norm": 0.0019145956030115485,
-      "learning_rate": 0.00019999916228888426,
-      "loss": 46.0,
-      "step": 17051
-    },
-    {
-      "epoch": 1.3037444807615115,
-      "grad_norm": 0.0012322311522439122,
-      "learning_rate": 0.00019999916219056436,
-      "loss": 46.0,
-      "step": 17052
-    },
-    {
-      "epoch": 1.3038209377449013,
-      "grad_norm": 0.00039088213816285133,
-      "learning_rate": 0.00019999916209223871,
-      "loss": 46.0,
-      "step": 17053
-    },
-    {
-      "epoch": 1.303897394728291,
-      "grad_norm": 0.0014892910839989781,
-      "learning_rate": 0.00019999916199390732,
-      "loss": 46.0,
-      "step": 17054
-    },
-    {
-      "epoch": 1.3039738517116808,
-      "grad_norm": 0.005496213678270578,
-      "learning_rate": 0.0001999991618955701,
-      "loss": 46.0,
-      "step": 17055
-    },
-    {
-      "epoch": 1.3040503086950705,
-      "grad_norm": 0.000989635824225843,
-      "learning_rate": 0.00019999916179722714,
-      "loss": 46.0,
-      "step": 17056
-    },
-    {
-      "epoch": 1.30412676567846,
-      "grad_norm": 0.0012343868147581816,
-      "learning_rate": 0.00019999916169887843,
-      "loss": 46.0,
-      "step": 17057
-    },
-    {
-      "epoch": 1.3042032226618498,
-      "grad_norm": 0.01438385434448719,
-      "learning_rate": 0.00019999916160052394,
-      "loss": 46.0,
-      "step": 17058
-    },
-    {
-      "epoch": 1.3042796796452396,
-      "grad_norm": 0.0009648837149143219,
-      "learning_rate": 0.00019999916150216366,
-      "loss": 46.0,
-      "step": 17059
-    },
-    {
-      "epoch": 1.3043561366286294,
-      "grad_norm": 0.0018759606173262,
-      "learning_rate": 0.00019999916140379763,
-      "loss": 46.0,
-      "step": 17060
-    },
-    {
-      "epoch": 1.304432593612019,
-      "grad_norm": 0.0007930463762022555,
-      "learning_rate": 0.00019999916130542582,
-      "loss": 46.0,
-      "step": 17061
-    },
-    {
-      "epoch": 1.3045090505954087,
-      "grad_norm": 0.0004368225345388055,
-      "learning_rate": 0.00019999916120704825,
-      "loss": 46.0,
-      "step": 17062
-    },
-    {
-      "epoch": 1.3045855075787984,
-      "grad_norm": 0.0051443264819681644,
-      "learning_rate": 0.00019999916110866492,
-      "loss": 46.0,
-      "step": 17063
-    },
-    {
-      "epoch": 1.3046619645621882,
-      "grad_norm": 0.0034380012657493353,
-      "learning_rate": 0.0001999991610102758,
-      "loss": 46.0,
-      "step": 17064
-    },
-    {
-      "epoch": 1.304738421545578,
-      "grad_norm": 0.005695089232176542,
-      "learning_rate": 0.00019999916091188093,
-      "loss": 46.0,
-      "step": 17065
-    },
-    {
-      "epoch": 1.3048148785289677,
-      "grad_norm": 0.0024394141510128975,
-      "learning_rate": 0.00019999916081348026,
-      "loss": 46.0,
-      "step": 17066
-    },
-    {
-      "epoch": 1.3048913355123575,
-      "grad_norm": 0.0006356213125400245,
-      "learning_rate": 0.00019999916071507387,
-      "loss": 46.0,
-      "step": 17067
-    },
-    {
-      "epoch": 1.304967792495747,
-      "grad_norm": 0.0007458970067091286,
-      "learning_rate": 0.00019999916061666167,
-      "loss": 46.0,
-      "step": 17068
-    },
-    {
-      "epoch": 1.3050442494791368,
-      "grad_norm": 0.0006066332571208477,
-      "learning_rate": 0.0001999991605182437,
-      "loss": 46.0,
-      "step": 17069
-    },
-    {
-      "epoch": 1.3051207064625265,
-      "grad_norm": 0.0012583734933286905,
-      "learning_rate": 0.00019999916041982,
-      "loss": 46.0,
-      "step": 17070
-    },
-    {
-      "epoch": 1.3051971634459163,
-      "grad_norm": 0.0007267092005349696,
-      "learning_rate": 0.0001999991603213905,
-      "loss": 46.0,
-      "step": 17071
-    },
-    {
-      "epoch": 1.3052736204293058,
-      "grad_norm": 0.00033966515911743045,
-      "learning_rate": 0.0001999991602229552,
-      "loss": 46.0,
-      "step": 17072
-    },
-    {
-      "epoch": 1.3053500774126956,
-      "grad_norm": 0.004110801964998245,
-      "learning_rate": 0.00019999916012451417,
-      "loss": 46.0,
-      "step": 17073
-    },
-    {
-      "epoch": 1.3054265343960854,
-      "grad_norm": 0.0015570273390039802,
-      "learning_rate": 0.00019999916002606737,
-      "loss": 46.0,
-      "step": 17074
-    },
-    {
-      "epoch": 1.3055029913794751,
-      "grad_norm": 0.0010826793732121587,
-      "learning_rate": 0.0001999991599276148,
-      "loss": 46.0,
-      "step": 17075
-    },
-    {
-      "epoch": 1.3055794483628649,
-      "grad_norm": 0.0029062326066195965,
-      "learning_rate": 0.00019999915982915644,
-      "loss": 46.0,
-      "step": 17076
-    },
-    {
-      "epoch": 1.3056559053462546,
-      "grad_norm": 0.0006346172885969281,
-      "learning_rate": 0.00019999915973069234,
-      "loss": 46.0,
-      "step": 17077
-    },
-    {
-      "epoch": 1.3057323623296444,
-      "grad_norm": 0.0013415678404271603,
-      "learning_rate": 0.00019999915963222247,
-      "loss": 46.0,
-      "step": 17078
-    },
-    {
-      "epoch": 1.305808819313034,
-      "grad_norm": 0.0010799997253343463,
-      "learning_rate": 0.0001999991595337468,
-      "loss": 46.0,
-      "step": 17079
-    },
-    {
-      "epoch": 1.3058852762964237,
-      "grad_norm": 0.0012914076214656234,
-      "learning_rate": 0.00019999915943526539,
-      "loss": 46.0,
-      "step": 17080
-    },
-    {
-      "epoch": 1.3059617332798135,
-      "grad_norm": 0.002412175526842475,
-      "learning_rate": 0.0001999991593367782,
-      "loss": 46.0,
-      "step": 17081
-    },
-    {
-      "epoch": 1.3060381902632032,
-      "grad_norm": 0.0010982458479702473,
-      "learning_rate": 0.00019999915923828523,
-      "loss": 46.0,
-      "step": 17082
-    },
-    {
-      "epoch": 1.3061146472465928,
-      "grad_norm": 0.002270329277962446,
-      "learning_rate": 0.00019999915913978652,
-      "loss": 46.0,
-      "step": 17083
-    },
-    {
-      "epoch": 1.3061911042299825,
-      "grad_norm": 0.0010062523651868105,
-      "learning_rate": 0.000199999159041282,
-      "loss": 46.0,
-      "step": 17084
-    },
-    {
-      "epoch": 1.3062675612133723,
-      "grad_norm": 0.004939588252454996,
-      "learning_rate": 0.00019999915894277175,
-      "loss": 46.0,
-      "step": 17085
-    },
-    {
-      "epoch": 1.306344018196762,
-      "grad_norm": 0.0010617101797834039,
-      "learning_rate": 0.0001999991588442557,
-      "loss": 46.0,
-      "step": 17086
-    },
-    {
-      "epoch": 1.3064204751801518,
-      "grad_norm": 0.0007467318791896105,
-      "learning_rate": 0.0001999991587457339,
-      "loss": 46.0,
-      "step": 17087
-    },
-    {
-      "epoch": 1.3064969321635416,
-      "grad_norm": 0.0021483786404132843,
-      "learning_rate": 0.00019999915864720631,
-      "loss": 46.0,
-      "step": 17088
-    },
-    {
-      "epoch": 1.3065733891469313,
-      "grad_norm": 0.0027480274438858032,
-      "learning_rate": 0.00019999915854867296,
-      "loss": 46.0,
-      "step": 17089
-    },
-    {
-      "epoch": 1.3066498461303209,
-      "grad_norm": 0.0008927136077545583,
-      "learning_rate": 0.00019999915845013387,
-      "loss": 46.0,
-      "step": 17090
-    },
-    {
-      "epoch": 1.3067263031137106,
-      "grad_norm": 0.0013238799292594194,
-      "learning_rate": 0.00019999915835158894,
-      "loss": 46.0,
-      "step": 17091
-    },
-    {
-      "epoch": 1.3068027600971004,
-      "grad_norm": 0.0004264836315996945,
-      "learning_rate": 0.00019999915825303833,
-      "loss": 46.0,
-      "step": 17092
-    },
-    {
-      "epoch": 1.3068792170804902,
-      "grad_norm": 0.011239280924201012,
-      "learning_rate": 0.00019999915815448188,
-      "loss": 46.0,
-      "step": 17093
-    },
-    {
-      "epoch": 1.3069556740638797,
-      "grad_norm": 0.002597997896373272,
-      "learning_rate": 0.0001999991580559197,
-      "loss": 46.0,
-      "step": 17094
-    },
-    {
-      "epoch": 1.3070321310472695,
-      "grad_norm": 0.0017309971153736115,
-      "learning_rate": 0.00019999915795735173,
-      "loss": 46.0,
-      "step": 17095
-    },
-    {
-      "epoch": 1.3071085880306592,
-      "grad_norm": 0.007089436985552311,
-      "learning_rate": 0.00019999915785877802,
-      "loss": 46.0,
-      "step": 17096
-    },
-    {
-      "epoch": 1.307185045014049,
-      "grad_norm": 0.001025989418849349,
-      "learning_rate": 0.0001999991577601985,
-      "loss": 46.0,
-      "step": 17097
-    },
-    {
-      "epoch": 1.3072615019974387,
-      "grad_norm": 0.013778255321085453,
-      "learning_rate": 0.00019999915766161322,
-      "loss": 46.0,
-      "step": 17098
-    },
-    {
-      "epoch": 1.3073379589808285,
-      "grad_norm": 0.001568926265463233,
-      "learning_rate": 0.0001999991575630222,
-      "loss": 46.0,
-      "step": 17099
-    },
-    {
-      "epoch": 1.3074144159642183,
-      "grad_norm": 0.0010469084372743964,
-      "learning_rate": 0.00019999915746442536,
-      "loss": 46.0,
-      "step": 17100
-    },
-    {
-      "epoch": 1.3074908729476078,
-      "grad_norm": 0.0010955313919112086,
-      "learning_rate": 0.00019999915736582279,
-      "loss": 46.0,
-      "step": 17101
-    },
-    {
-      "epoch": 1.3075673299309976,
-      "grad_norm": 0.008914983831346035,
-      "learning_rate": 0.00019999915726721446,
-      "loss": 46.0,
-      "step": 17102
-    },
-    {
-      "epoch": 1.3076437869143873,
-      "grad_norm": 0.002404893049970269,
-      "learning_rate": 0.0001999991571686003,
-      "loss": 46.0,
-      "step": 17103
-    },
-    {
-      "epoch": 1.307720243897777,
-      "grad_norm": 0.010579340159893036,
-      "learning_rate": 0.00019999915706998044,
-      "loss": 46.0,
-      "step": 17104
-    },
-    {
-      "epoch": 1.3077967008811666,
-      "grad_norm": 0.005604743491858244,
-      "learning_rate": 0.00019999915697135477,
-      "loss": 46.0,
-      "step": 17105
-    },
-    {
-      "epoch": 1.3078731578645564,
-      "grad_norm": 0.006471514236181974,
-      "learning_rate": 0.00019999915687272336,
-      "loss": 46.0,
-      "step": 17106
-    },
-    {
-      "epoch": 1.3079496148479461,
-      "grad_norm": 0.0014159809798002243,
-      "learning_rate": 0.00019999915677408617,
-      "loss": 46.0,
-      "step": 17107
-    },
-    {
-      "epoch": 1.308026071831336,
-      "grad_norm": 0.0013574615586549044,
-      "learning_rate": 0.0001999991566754432,
-      "loss": 46.0,
-      "step": 17108
-    },
-    {
-      "epoch": 1.3081025288147257,
-      "grad_norm": 0.005113136954605579,
-      "learning_rate": 0.00019999915657679447,
-      "loss": 46.0,
-      "step": 17109
-    },
-    {
-      "epoch": 1.3081789857981154,
-      "grad_norm": 0.001255523762665689,
-      "learning_rate": 0.00019999915647813996,
-      "loss": 46.0,
-      "step": 17110
-    },
-    {
-      "epoch": 1.3082554427815052,
-      "grad_norm": 0.0008455773931927979,
-      "learning_rate": 0.00019999915637947967,
-      "loss": 46.0,
-      "step": 17111
-    },
-    {
-      "epoch": 1.3083318997648947,
-      "grad_norm": 0.00045527919428423047,
-      "learning_rate": 0.00019999915628081361,
-      "loss": 46.0,
-      "step": 17112
-    },
-    {
-      "epoch": 1.3084083567482845,
-      "grad_norm": 0.005838301964104176,
-      "learning_rate": 0.0001999991561821418,
-      "loss": 46.0,
-      "step": 17113
-    },
-    {
-      "epoch": 1.3084848137316742,
-      "grad_norm": 0.00553513178601861,
-      "learning_rate": 0.00019999915608346424,
-      "loss": 46.0,
-      "step": 17114
-    },
-    {
-      "epoch": 1.308561270715064,
-      "grad_norm": 0.004545619245618582,
-      "learning_rate": 0.00019999915598478089,
-      "loss": 46.0,
-      "step": 17115
-    },
-    {
-      "epoch": 1.3086377276984535,
-      "grad_norm": 0.0017334987642243505,
-      "learning_rate": 0.00019999915588609176,
-      "loss": 46.0,
-      "step": 17116
-    },
-    {
-      "epoch": 1.3087141846818433,
-      "grad_norm": 0.0009968295926228166,
-      "learning_rate": 0.0001999991557873969,
-      "loss": 46.0,
-      "step": 17117
-    },
-    {
-      "epoch": 1.308790641665233,
-      "grad_norm": 0.0004681750142481178,
-      "learning_rate": 0.00019999915568869622,
-      "loss": 46.0,
-      "step": 17118
-    },
-    {
-      "epoch": 1.3088670986486228,
-      "grad_norm": 0.0014000753872096539,
-      "learning_rate": 0.00019999915558998978,
-      "loss": 46.0,
-      "step": 17119
-    },
-    {
-      "epoch": 1.3089435556320126,
-      "grad_norm": 0.0009876416297629476,
-      "learning_rate": 0.0001999991554912776,
-      "loss": 46.0,
-      "step": 17120
-    },
-    {
-      "epoch": 1.3090200126154024,
-      "grad_norm": 0.0016021830961108208,
-      "learning_rate": 0.00019999915539255963,
-      "loss": 46.0,
-      "step": 17121
-    },
-    {
-      "epoch": 1.3090964695987921,
-      "grad_norm": 0.006823969539254904,
-      "learning_rate": 0.0001999991552938359,
-      "loss": 46.0,
-      "step": 17122
-    },
-    {
-      "epoch": 1.3091729265821817,
-      "grad_norm": 0.001259143347851932,
-      "learning_rate": 0.00019999915519510638,
-      "loss": 46.0,
-      "step": 17123
-    },
-    {
-      "epoch": 1.3092493835655714,
-      "grad_norm": 0.0028713110368698835,
-      "learning_rate": 0.00019999915509637112,
-      "loss": 46.0,
-      "step": 17124
-    },
-    {
-      "epoch": 1.3093258405489612,
-      "grad_norm": 0.0008640232845209539,
-      "learning_rate": 0.00019999915499763007,
-      "loss": 46.0,
-      "step": 17125
-    },
-    {
-      "epoch": 1.309402297532351,
-      "grad_norm": 0.00041096031782217324,
-      "learning_rate": 0.00019999915489888324,
-      "loss": 46.0,
-      "step": 17126
-    },
-    {
-      "epoch": 1.3094787545157405,
-      "grad_norm": 0.0018318830989301205,
-      "learning_rate": 0.00019999915480013066,
-      "loss": 46.0,
-      "step": 17127
-    },
-    {
-      "epoch": 1.3095552114991302,
-      "grad_norm": 0.0029584444127976894,
-      "learning_rate": 0.0001999991547013723,
-      "loss": 46.0,
-      "step": 17128
-    },
-    {
-      "epoch": 1.30963166848252,
-      "grad_norm": 0.0010804666671901941,
-      "learning_rate": 0.0001999991546026082,
-      "loss": 46.0,
-      "step": 17129
-    },
-    {
-      "epoch": 1.3097081254659098,
-      "grad_norm": 0.0009533656993880868,
-      "learning_rate": 0.0001999991545038383,
-      "loss": 46.0,
-      "step": 17130
-    },
-    {
-      "epoch": 1.3097845824492995,
-      "grad_norm": 0.0014663183828815818,
-      "learning_rate": 0.00019999915440506265,
-      "loss": 46.0,
-      "step": 17131
-    },
-    {
-      "epoch": 1.3098610394326893,
-      "grad_norm": 0.0003289435408078134,
-      "learning_rate": 0.00019999915430628123,
-      "loss": 46.0,
-      "step": 17132
-    },
-    {
-      "epoch": 1.309937496416079,
-      "grad_norm": 0.0017163041047751904,
-      "learning_rate": 0.00019999915420749402,
-      "loss": 46.0,
-      "step": 17133
-    },
-    {
-      "epoch": 1.3100139533994686,
-      "grad_norm": 0.0010907113319262862,
-      "learning_rate": 0.00019999915410870105,
-      "loss": 46.0,
-      "step": 17134
-    },
-    {
-      "epoch": 1.3100904103828583,
-      "grad_norm": 0.004672758746892214,
-      "learning_rate": 0.0001999991540099023,
-      "loss": 46.0,
-      "step": 17135
-    },
-    {
-      "epoch": 1.310166867366248,
-      "grad_norm": 0.0004823078343179077,
-      "learning_rate": 0.00019999915391109778,
-      "loss": 46.0,
-      "step": 17136
-    },
-    {
-      "epoch": 1.3102433243496379,
-      "grad_norm": 0.016972461715340614,
-      "learning_rate": 0.0001999991538122875,
-      "loss": 46.0,
-      "step": 17137
-    },
-    {
-      "epoch": 1.3103197813330274,
-      "grad_norm": 0.0007133137551136315,
-      "learning_rate": 0.00019999915371347147,
-      "loss": 46.0,
-      "step": 17138
-    },
-    {
-      "epoch": 1.3103962383164172,
-      "grad_norm": 0.003391512203961611,
-      "learning_rate": 0.00019999915361464964,
-      "loss": 46.0,
-      "step": 17139
-    },
-    {
-      "epoch": 1.310472695299807,
-      "grad_norm": 0.0032656979747116566,
-      "learning_rate": 0.00019999915351582206,
-      "loss": 46.0,
-      "step": 17140
-    },
-    {
-      "epoch": 1.3105491522831967,
-      "grad_norm": 0.002001130487769842,
-      "learning_rate": 0.00019999915341698872,
-      "loss": 46.0,
-      "step": 17141
-    },
-    {
-      "epoch": 1.3106256092665864,
-      "grad_norm": 0.0005703952629119158,
-      "learning_rate": 0.0001999991533181496,
-      "loss": 46.0,
-      "step": 17142
-    },
-    {
-      "epoch": 1.3107020662499762,
-      "grad_norm": 0.003303299192339182,
-      "learning_rate": 0.00019999915321930467,
-      "loss": 46.0,
-      "step": 17143
-    },
-    {
-      "epoch": 1.3107785232333657,
-      "grad_norm": 0.0022562930826097727,
-      "learning_rate": 0.000199999153120454,
-      "loss": 46.0,
-      "step": 17144
-    },
-    {
-      "epoch": 1.3108549802167555,
-      "grad_norm": 0.0009510604431852698,
-      "learning_rate": 0.00019999915302159758,
-      "loss": 46.0,
-      "step": 17145
-    },
-    {
-      "epoch": 1.3109314372001453,
-      "grad_norm": 0.0012925302144140005,
-      "learning_rate": 0.0001999991529227354,
-      "loss": 46.0,
-      "step": 17146
-    },
-    {
-      "epoch": 1.311007894183535,
-      "grad_norm": 0.0010041811037808657,
-      "learning_rate": 0.0001999991528238674,
-      "loss": 46.0,
-      "step": 17147
-    },
-    {
-      "epoch": 1.3110843511669248,
-      "grad_norm": 0.0016535257454961538,
-      "learning_rate": 0.00019999915272499367,
-      "loss": 46.0,
-      "step": 17148
-    },
-    {
-      "epoch": 1.3111608081503143,
-      "grad_norm": 0.0018469089409336448,
-      "learning_rate": 0.00019999915262611416,
-      "loss": 46.0,
-      "step": 17149
-    },
-    {
-      "epoch": 1.311237265133704,
-      "grad_norm": 0.001431250129826367,
-      "learning_rate": 0.0001999991525272289,
-      "loss": 46.0,
-      "step": 17150
-    },
-    {
-      "epoch": 1.3113137221170938,
-      "grad_norm": 0.0006772562628611922,
-      "learning_rate": 0.00019999915242833782,
-      "loss": 46.0,
-      "step": 17151
-    },
-    {
-      "epoch": 1.3113901791004836,
-      "grad_norm": 0.015381153672933578,
-      "learning_rate": 0.00019999915232944102,
-      "loss": 46.0,
-      "step": 17152
-    },
-    {
-      "epoch": 1.3114666360838734,
-      "grad_norm": 0.004695856012403965,
-      "learning_rate": 0.00019999915223053842,
-      "loss": 46.0,
-      "step": 17153
-    },
-    {
-      "epoch": 1.3115430930672631,
-      "grad_norm": 0.001373407314531505,
-      "learning_rate": 0.00019999915213163004,
-      "loss": 46.0,
-      "step": 17154
-    },
-    {
-      "epoch": 1.3116195500506527,
-      "grad_norm": 0.0005978519329801202,
-      "learning_rate": 0.00019999915203271592,
-      "loss": 46.0,
-      "step": 17155
-    },
-    {
-      "epoch": 1.3116960070340424,
-      "grad_norm": 0.0014906262513250113,
-      "learning_rate": 0.00019999915193379602,
-      "loss": 46.0,
-      "step": 17156
-    },
-    {
-      "epoch": 1.3117724640174322,
-      "grad_norm": 0.001896207220852375,
-      "learning_rate": 0.00019999915183487038,
-      "loss": 46.0,
-      "step": 17157
-    },
-    {
-      "epoch": 1.311848921000822,
-      "grad_norm": 0.0012717708013951778,
-      "learning_rate": 0.00019999915173593894,
-      "loss": 46.0,
-      "step": 17158
-    },
-    {
-      "epoch": 1.3119253779842117,
-      "grad_norm": 0.0013867573579773307,
-      "learning_rate": 0.00019999915163700172,
-      "loss": 46.0,
-      "step": 17159
-    },
-    {
-      "epoch": 1.3120018349676013,
-      "grad_norm": 0.0012509622611105442,
-      "learning_rate": 0.00019999915153805876,
-      "loss": 46.0,
-      "step": 17160
-    },
-    {
-      "epoch": 1.312078291950991,
-      "grad_norm": 0.005528236739337444,
-      "learning_rate": 0.00019999915143911,
-      "loss": 46.0,
-      "step": 17161
-    },
-    {
-      "epoch": 1.3121547489343808,
-      "grad_norm": 0.0029104817658662796,
-      "learning_rate": 0.0001999991513401555,
-      "loss": 46.0,
-      "step": 17162
-    },
-    {
-      "epoch": 1.3122312059177705,
-      "grad_norm": 0.010189109481871128,
-      "learning_rate": 0.0001999991512411952,
-      "loss": 46.0,
-      "step": 17163
-    },
-    {
-      "epoch": 1.3123076629011603,
-      "grad_norm": 0.000931020185817033,
-      "learning_rate": 0.00019999915114222918,
-      "loss": 46.0,
-      "step": 17164
-    },
-    {
-      "epoch": 1.31238411988455,
-      "grad_norm": 0.0012041906593367457,
-      "learning_rate": 0.00019999915104325735,
-      "loss": 46.0,
-      "step": 17165
-    },
-    {
-      "epoch": 1.3124605768679396,
-      "grad_norm": 0.002006001304835081,
-      "learning_rate": 0.00019999915094427975,
-      "loss": 46.0,
-      "step": 17166
-    },
-    {
-      "epoch": 1.3125370338513294,
-      "grad_norm": 0.0005529527552425861,
-      "learning_rate": 0.0001999991508452964,
-      "loss": 46.0,
-      "step": 17167
-    },
-    {
-      "epoch": 1.3126134908347191,
-      "grad_norm": 0.0008109658374451101,
-      "learning_rate": 0.00019999915074630725,
-      "loss": 46.0,
-      "step": 17168
-    },
-    {
-      "epoch": 1.3126899478181089,
-      "grad_norm": 0.004001515917479992,
-      "learning_rate": 0.00019999915064731235,
-      "loss": 46.0,
-      "step": 17169
-    },
-    {
-      "epoch": 1.3127664048014984,
-      "grad_norm": 0.0010594894411042333,
-      "learning_rate": 0.00019999915054831169,
-      "loss": 46.0,
-      "step": 17170
-    },
-    {
-      "epoch": 1.3128428617848882,
-      "grad_norm": 0.003379775444045663,
-      "learning_rate": 0.00019999915044930527,
-      "loss": 46.0,
-      "step": 17171
-    },
-    {
-      "epoch": 1.312919318768278,
-      "grad_norm": 0.0011823073728010058,
-      "learning_rate": 0.00019999915035029305,
-      "loss": 46.0,
-      "step": 17172
-    },
-    {
-      "epoch": 1.3129957757516677,
-      "grad_norm": 0.0024823795538395643,
-      "learning_rate": 0.00019999915025127507,
-      "loss": 46.0,
-      "step": 17173
-    },
-    {
-      "epoch": 1.3130722327350575,
-      "grad_norm": 0.0009069254738278687,
-      "learning_rate": 0.00019999915015225136,
-      "loss": 46.0,
-      "step": 17174
-    },
-    {
-      "epoch": 1.3131486897184472,
-      "grad_norm": 0.002019510604441166,
-      "learning_rate": 0.0001999991500532218,
-      "loss": 46.0,
-      "step": 17175
-    },
-    {
-      "epoch": 1.313225146701837,
-      "grad_norm": 0.0005911023472435772,
-      "learning_rate": 0.00019999914995418654,
-      "loss": 46.0,
-      "step": 17176
-    },
-    {
-      "epoch": 1.3133016036852265,
-      "grad_norm": 0.0009922138415277004,
-      "learning_rate": 0.00019999914985514548,
-      "loss": 46.0,
-      "step": 17177
-    },
-    {
-      "epoch": 1.3133780606686163,
-      "grad_norm": 0.0013644836144521832,
-      "learning_rate": 0.00019999914975609863,
-      "loss": 46.0,
-      "step": 17178
-    },
-    {
-      "epoch": 1.313454517652006,
-      "grad_norm": 0.002014856319874525,
-      "learning_rate": 0.00019999914965704605,
-      "loss": 46.0,
-      "step": 17179
-    },
-    {
-      "epoch": 1.3135309746353958,
-      "grad_norm": 0.0006061993772163987,
-      "learning_rate": 0.00019999914955798768,
-      "loss": 46.0,
-      "step": 17180
-    },
-    {
-      "epoch": 1.3136074316187853,
-      "grad_norm": 0.00035414070589467883,
-      "learning_rate": 0.00019999914945892356,
-      "loss": 46.0,
-      "step": 17181
-    },
-    {
-      "epoch": 1.313683888602175,
-      "grad_norm": 0.00031240942189469934,
-      "learning_rate": 0.00019999914935985366,
-      "loss": 46.0,
-      "step": 17182
-    },
-    {
-      "epoch": 1.3137603455855649,
-      "grad_norm": 0.0005707665113732219,
-      "learning_rate": 0.00019999914926077802,
-      "loss": 46.0,
-      "step": 17183
-    },
-    {
-      "epoch": 1.3138368025689546,
-      "grad_norm": 0.002884496469050646,
-      "learning_rate": 0.00019999914916169655,
-      "loss": 46.0,
-      "step": 17184
-    },
-    {
-      "epoch": 1.3139132595523444,
-      "grad_norm": 0.00043921807082369924,
-      "learning_rate": 0.00019999914906260934,
-      "loss": 46.0,
-      "step": 17185
-    },
-    {
-      "epoch": 1.3139897165357342,
-      "grad_norm": 0.0010038636391982436,
-      "learning_rate": 0.00019999914896351637,
-      "loss": 46.0,
-      "step": 17186
-    },
-    {
-      "epoch": 1.314066173519124,
-      "grad_norm": 0.0007171165780164301,
-      "learning_rate": 0.0001999991488644176,
-      "loss": 46.0,
-      "step": 17187
-    },
-    {
-      "epoch": 1.3141426305025135,
-      "grad_norm": 0.0010055151069536805,
-      "learning_rate": 0.0001999991487653131,
-      "loss": 46.0,
-      "step": 17188
-    },
-    {
-      "epoch": 1.3142190874859032,
-      "grad_norm": 0.0028502445202320814,
-      "learning_rate": 0.00019999914866620282,
-      "loss": 46.0,
-      "step": 17189
-    },
-    {
-      "epoch": 1.314295544469293,
-      "grad_norm": 0.0005379535141400993,
-      "learning_rate": 0.00019999914856708677,
-      "loss": 46.0,
-      "step": 17190
-    },
-    {
-      "epoch": 1.3143720014526827,
-      "grad_norm": 0.006283284164965153,
-      "learning_rate": 0.0001999991484679649,
-      "loss": 46.0,
-      "step": 17191
-    },
-    {
-      "epoch": 1.3144484584360723,
-      "grad_norm": 0.0006603418732993305,
-      "learning_rate": 0.00019999914836883734,
-      "loss": 46.0,
-      "step": 17192
-    },
-    {
-      "epoch": 1.314524915419462,
-      "grad_norm": 0.0006712102331221104,
-      "learning_rate": 0.000199999148269704,
-      "loss": 46.0,
-      "step": 17193
-    },
-    {
-      "epoch": 1.3146013724028518,
-      "grad_norm": 0.022888246923685074,
-      "learning_rate": 0.00019999914817056484,
-      "loss": 46.0,
-      "step": 17194
-    },
-    {
-      "epoch": 1.3146778293862416,
-      "grad_norm": 0.0027552032843232155,
-      "learning_rate": 0.00019999914807141992,
-      "loss": 46.0,
-      "step": 17195
-    },
-    {
-      "epoch": 1.3147542863696313,
-      "grad_norm": 0.0013505418319255114,
-      "learning_rate": 0.00019999914797226925,
-      "loss": 46.0,
-      "step": 17196
-    },
-    {
-      "epoch": 1.314830743353021,
-      "grad_norm": 0.003091516438871622,
-      "learning_rate": 0.00019999914787311284,
-      "loss": 46.0,
-      "step": 17197
-    },
-    {
-      "epoch": 1.3149072003364108,
-      "grad_norm": 0.0006567321834154427,
-      "learning_rate": 0.0001999991477739506,
-      "loss": 46.0,
-      "step": 17198
-    },
-    {
-      "epoch": 1.3149836573198004,
-      "grad_norm": 0.0005596888950094581,
-      "learning_rate": 0.00019999914767478263,
-      "loss": 46.0,
-      "step": 17199
-    },
-    {
-      "epoch": 1.3150601143031901,
-      "grad_norm": 0.0004952656454406679,
-      "learning_rate": 0.00019999914757560887,
-      "loss": 46.0,
-      "step": 17200
-    },
-    {
-      "epoch": 1.31513657128658,
-      "grad_norm": 0.0009215171448886395,
-      "learning_rate": 0.00019999914747642934,
-      "loss": 46.0,
-      "step": 17201
-    },
-    {
-      "epoch": 1.3152130282699697,
-      "grad_norm": 0.001959695480763912,
-      "learning_rate": 0.00019999914737724406,
-      "loss": 46.0,
-      "step": 17202
-    },
-    {
-      "epoch": 1.3152894852533592,
-      "grad_norm": 0.0008185189217329025,
-      "learning_rate": 0.000199999147278053,
-      "loss": 46.0,
-      "step": 17203
-    },
-    {
-      "epoch": 1.315365942236749,
-      "grad_norm": 0.0008231941028498113,
-      "learning_rate": 0.00019999914717885617,
-      "loss": 46.0,
-      "step": 17204
-    },
-    {
-      "epoch": 1.3154423992201387,
-      "grad_norm": 0.0007910511340014637,
-      "learning_rate": 0.00019999914707965357,
-      "loss": 46.0,
-      "step": 17205
-    },
-    {
-      "epoch": 1.3155188562035285,
-      "grad_norm": 0.0017435974441468716,
-      "learning_rate": 0.0001999991469804452,
-      "loss": 46.0,
-      "step": 17206
-    },
-    {
-      "epoch": 1.3155953131869182,
-      "grad_norm": 0.0004577609652187675,
-      "learning_rate": 0.00019999914688123108,
-      "loss": 46.0,
-      "step": 17207
-    },
-    {
-      "epoch": 1.315671770170308,
-      "grad_norm": 0.0011333676520735025,
-      "learning_rate": 0.00019999914678201118,
-      "loss": 46.0,
-      "step": 17208
-    },
-    {
-      "epoch": 1.3157482271536978,
-      "grad_norm": 0.0003037470451090485,
-      "learning_rate": 0.00019999914668278549,
-      "loss": 46.0,
-      "step": 17209
-    },
-    {
-      "epoch": 1.3158246841370873,
-      "grad_norm": 0.0004673453513532877,
-      "learning_rate": 0.00019999914658355405,
-      "loss": 46.0,
-      "step": 17210
-    },
-    {
-      "epoch": 1.315901141120477,
-      "grad_norm": 0.0026928263250738382,
-      "learning_rate": 0.00019999914648431683,
-      "loss": 46.0,
-      "step": 17211
-    },
-    {
-      "epoch": 1.3159775981038668,
-      "grad_norm": 0.001289918553084135,
-      "learning_rate": 0.00019999914638507384,
-      "loss": 46.0,
-      "step": 17212
-    },
-    {
-      "epoch": 1.3160540550872566,
-      "grad_norm": 0.0018694362370297313,
-      "learning_rate": 0.0001999991462858251,
-      "loss": 46.0,
-      "step": 17213
-    },
-    {
-      "epoch": 1.3161305120706461,
-      "grad_norm": 0.0022590463049709797,
-      "learning_rate": 0.00019999914618657057,
-      "loss": 46.0,
-      "step": 17214
-    },
-    {
-      "epoch": 1.316206969054036,
-      "grad_norm": 0.001186096458695829,
-      "learning_rate": 0.0001999991460873103,
-      "loss": 46.0,
-      "step": 17215
-    },
-    {
-      "epoch": 1.3162834260374257,
-      "grad_norm": 0.000799805682618171,
-      "learning_rate": 0.00019999914598804424,
-      "loss": 46.0,
-      "step": 17216
-    },
-    {
-      "epoch": 1.3163598830208154,
-      "grad_norm": 0.0035685899201780558,
-      "learning_rate": 0.0001999991458887724,
-      "loss": 46.0,
-      "step": 17217
-    },
-    {
-      "epoch": 1.3164363400042052,
-      "grad_norm": 0.0009092389955185354,
-      "learning_rate": 0.00019999914578949478,
-      "loss": 46.0,
-      "step": 17218
-    },
-    {
-      "epoch": 1.316512796987595,
-      "grad_norm": 0.0006872125086374581,
-      "learning_rate": 0.0001999991456902114,
-      "loss": 46.0,
-      "step": 17219
-    },
-    {
-      "epoch": 1.3165892539709847,
-      "grad_norm": 0.0009588262182660401,
-      "learning_rate": 0.0001999991455909223,
-      "loss": 46.0,
-      "step": 17220
-    },
-    {
-      "epoch": 1.3166657109543742,
-      "grad_norm": 0.0010056920582428575,
-      "learning_rate": 0.00019999914549162737,
-      "loss": 46.0,
-      "step": 17221
-    },
-    {
-      "epoch": 1.316742167937764,
-      "grad_norm": 0.0012327306903898716,
-      "learning_rate": 0.00019999914539232673,
-      "loss": 46.0,
-      "step": 17222
-    },
-    {
-      "epoch": 1.3168186249211538,
-      "grad_norm": 0.017056334763765335,
-      "learning_rate": 0.00019999914529302023,
-      "loss": 46.0,
-      "step": 17223
-    },
-    {
-      "epoch": 1.3168950819045435,
-      "grad_norm": 0.00018474803073331714,
-      "learning_rate": 0.00019999914519370804,
-      "loss": 46.0,
-      "step": 17224
-    },
-    {
-      "epoch": 1.316971538887933,
-      "grad_norm": 0.0005765540990978479,
-      "learning_rate": 0.00019999914509439006,
-      "loss": 46.0,
-      "step": 17225
-    },
-    {
-      "epoch": 1.3170479958713228,
-      "grad_norm": 0.0021213290747255087,
-      "learning_rate": 0.00019999914499506632,
-      "loss": 46.0,
-      "step": 17226
-    },
-    {
-      "epoch": 1.3171244528547126,
-      "grad_norm": 0.001359756919555366,
-      "learning_rate": 0.00019999914489573676,
-      "loss": 46.0,
-      "step": 17227
-    },
-    {
-      "epoch": 1.3172009098381023,
-      "grad_norm": 0.0009897955460473895,
-      "learning_rate": 0.0001999991447964015,
-      "loss": 46.0,
-      "step": 17228
-    },
-    {
-      "epoch": 1.317277366821492,
-      "grad_norm": 0.001963048242032528,
-      "learning_rate": 0.00019999914469706043,
-      "loss": 46.0,
-      "step": 17229
-    },
-    {
-      "epoch": 1.3173538238048819,
-      "grad_norm": 0.0011945561273023486,
-      "learning_rate": 0.00019999914459771357,
-      "loss": 46.0,
-      "step": 17230
-    },
-    {
-      "epoch": 1.3174302807882716,
-      "grad_norm": 0.0017985154408961535,
-      "learning_rate": 0.00019999914449836097,
-      "loss": 46.0,
-      "step": 17231
-    },
-    {
-      "epoch": 1.3175067377716612,
-      "grad_norm": 0.0009320020908489823,
-      "learning_rate": 0.00019999914439900262,
-      "loss": 46.0,
-      "step": 17232
-    },
-    {
-      "epoch": 1.317583194755051,
-      "grad_norm": 0.00032047543209046125,
-      "learning_rate": 0.00019999914429963848,
-      "loss": 46.0,
-      "step": 17233
-    },
-    {
-      "epoch": 1.3176596517384407,
-      "grad_norm": 0.0014175066025927663,
-      "learning_rate": 0.00019999914420026858,
-      "loss": 46.0,
-      "step": 17234
-    },
-    {
-      "epoch": 1.3177361087218304,
-      "grad_norm": 0.007659487891942263,
-      "learning_rate": 0.00019999914410089286,
-      "loss": 46.0,
-      "step": 17235
-    },
-    {
-      "epoch": 1.31781256570522,
-      "grad_norm": 0.0006222569500096142,
-      "learning_rate": 0.00019999914400151142,
-      "loss": 46.0,
-      "step": 17236
-    },
-    {
-      "epoch": 1.3178890226886097,
-      "grad_norm": 0.005294390022754669,
-      "learning_rate": 0.00019999914390212423,
-      "loss": 46.0,
-      "step": 17237
-    },
-    {
-      "epoch": 1.3179654796719995,
-      "grad_norm": 0.0004505709221120924,
-      "learning_rate": 0.00019999914380273125,
-      "loss": 46.0,
-      "step": 17238
-    },
-    {
-      "epoch": 1.3180419366553893,
-      "grad_norm": 0.004525754135102034,
-      "learning_rate": 0.00019999914370333248,
-      "loss": 46.0,
-      "step": 17239
-    },
-    {
-      "epoch": 1.318118393638779,
-      "grad_norm": 0.0014177690027281642,
-      "learning_rate": 0.00019999914360392795,
-      "loss": 46.0,
-      "step": 17240
-    },
-    {
-      "epoch": 1.3181948506221688,
-      "grad_norm": 0.0018735553603619337,
-      "learning_rate": 0.00019999914350451764,
-      "loss": 46.0,
-      "step": 17241
-    },
-    {
-      "epoch": 1.3182713076055586,
-      "grad_norm": 0.0017055986681953073,
-      "learning_rate": 0.00019999914340510156,
-      "loss": 46.0,
-      "step": 17242
-    },
-    {
-      "epoch": 1.318347764588948,
-      "grad_norm": 0.0007548227440565825,
-      "learning_rate": 0.00019999914330567974,
-      "loss": 46.0,
-      "step": 17243
-    },
-    {
-      "epoch": 1.3184242215723379,
-      "grad_norm": 0.0006721281679347157,
-      "learning_rate": 0.00019999914320625216,
-      "loss": 46.0,
-      "step": 17244
-    },
-    {
-      "epoch": 1.3185006785557276,
-      "grad_norm": 0.004983718506991863,
-      "learning_rate": 0.00019999914310681876,
-      "loss": 46.0,
-      "step": 17245
-    },
-    {
-      "epoch": 1.3185771355391174,
-      "grad_norm": 0.0030971700325608253,
-      "learning_rate": 0.00019999914300737964,
-      "loss": 46.0,
-      "step": 17246
-    },
-    {
-      "epoch": 1.318653592522507,
-      "grad_norm": 0.0027738006319850683,
-      "learning_rate": 0.00019999914290793472,
-      "loss": 46.0,
-      "step": 17247
-    },
-    {
-      "epoch": 1.3187300495058967,
-      "grad_norm": 0.001320280833169818,
-      "learning_rate": 0.00019999914280848403,
-      "loss": 46.0,
-      "step": 17248
-    },
-    {
-      "epoch": 1.3188065064892864,
-      "grad_norm": 0.003970456309616566,
-      "learning_rate": 0.00019999914270902759,
-      "loss": 46.0,
-      "step": 17249
-    },
-    {
-      "epoch": 1.3188829634726762,
-      "grad_norm": 0.0009836690733209252,
-      "learning_rate": 0.00019999914260956537,
-      "loss": 46.0,
-      "step": 17250
-    },
-    {
-      "epoch": 1.318959420456066,
-      "grad_norm": 0.0006394994561560452,
-      "learning_rate": 0.00019999914251009736,
-      "loss": 46.0,
-      "step": 17251
-    },
-    {
-      "epoch": 1.3190358774394557,
-      "grad_norm": 0.0011349063133820891,
-      "learning_rate": 0.0001999991424106236,
-      "loss": 46.0,
-      "step": 17252
-    },
-    {
-      "epoch": 1.3191123344228455,
-      "grad_norm": 0.001036398345604539,
-      "learning_rate": 0.00019999914231114406,
-      "loss": 46.0,
-      "step": 17253
-    },
-    {
-      "epoch": 1.319188791406235,
-      "grad_norm": 0.0017536289524286985,
-      "learning_rate": 0.00019999914221165876,
-      "loss": 46.0,
-      "step": 17254
-    },
-    {
-      "epoch": 1.3192652483896248,
-      "grad_norm": 0.0008155809482559562,
-      "learning_rate": 0.0001999991421121677,
-      "loss": 46.0,
-      "step": 17255
-    },
-    {
-      "epoch": 1.3193417053730145,
-      "grad_norm": 0.0010370647069066763,
-      "learning_rate": 0.00019999914201267088,
-      "loss": 46.0,
-      "step": 17256
-    },
-    {
-      "epoch": 1.3194181623564043,
-      "grad_norm": 0.0043812645599246025,
-      "learning_rate": 0.00019999914191316828,
-      "loss": 46.0,
-      "step": 17257
-    },
-    {
-      "epoch": 1.3194946193397938,
-      "grad_norm": 0.0005749440169893205,
-      "learning_rate": 0.00019999914181365988,
-      "loss": 46.0,
-      "step": 17258
-    },
-    {
-      "epoch": 1.3195710763231836,
-      "grad_norm": 0.004461383447051048,
-      "learning_rate": 0.00019999914171414576,
-      "loss": 46.0,
-      "step": 17259
-    },
-    {
-      "epoch": 1.3196475333065734,
-      "grad_norm": 0.002153238747268915,
-      "learning_rate": 0.00019999914161462584,
-      "loss": 46.0,
-      "step": 17260
-    },
-    {
-      "epoch": 1.3197239902899631,
-      "grad_norm": 0.00180660595651716,
-      "learning_rate": 0.00019999914151510014,
-      "loss": 46.0,
-      "step": 17261
-    },
-    {
-      "epoch": 1.3198004472733529,
-      "grad_norm": 0.0013833047123625875,
-      "learning_rate": 0.00019999914141556868,
-      "loss": 46.0,
-      "step": 17262
-    },
-    {
-      "epoch": 1.3198769042567426,
-      "grad_norm": 0.0014423176180571318,
-      "learning_rate": 0.00019999914131603146,
-      "loss": 46.0,
-      "step": 17263
-    },
-    {
-      "epoch": 1.3199533612401324,
-      "grad_norm": 0.0005700822221115232,
-      "learning_rate": 0.00019999914121648848,
-      "loss": 46.0,
-      "step": 17264
-    },
-    {
-      "epoch": 1.320029818223522,
-      "grad_norm": 0.0008115441305562854,
-      "learning_rate": 0.00019999914111693971,
-      "loss": 46.0,
-      "step": 17265
-    },
-    {
-      "epoch": 1.3201062752069117,
-      "grad_norm": 0.004543479066342115,
-      "learning_rate": 0.00019999914101738518,
-      "loss": 46.0,
-      "step": 17266
-    },
-    {
-      "epoch": 1.3201827321903015,
-      "grad_norm": 0.00228821556083858,
-      "learning_rate": 0.00019999914091782487,
-      "loss": 46.0,
-      "step": 17267
-    },
-    {
-      "epoch": 1.3202591891736912,
-      "grad_norm": 0.0040205069817602634,
-      "learning_rate": 0.00019999914081825882,
-      "loss": 46.0,
-      "step": 17268
-    },
-    {
-      "epoch": 1.3203356461570808,
-      "grad_norm": 0.0066104792058467865,
-      "learning_rate": 0.00019999914071868697,
-      "loss": 46.0,
-      "step": 17269
-    },
-    {
-      "epoch": 1.3204121031404705,
-      "grad_norm": 0.0067365881986916065,
-      "learning_rate": 0.00019999914061910937,
-      "loss": 46.0,
-      "step": 17270
-    },
-    {
-      "epoch": 1.3204885601238603,
-      "grad_norm": 0.0016946530668064952,
-      "learning_rate": 0.00019999914051952597,
-      "loss": 46.0,
-      "step": 17271
-    },
-    {
-      "epoch": 1.32056501710725,
-      "grad_norm": 0.002005101880058646,
-      "learning_rate": 0.00019999914041993685,
-      "loss": 46.0,
-      "step": 17272
-    },
-    {
-      "epoch": 1.3206414740906398,
-      "grad_norm": 0.003516450524330139,
-      "learning_rate": 0.00019999914032034193,
-      "loss": 46.0,
-      "step": 17273
-    },
-    {
-      "epoch": 1.3207179310740296,
-      "grad_norm": 0.0017321519553661346,
-      "learning_rate": 0.00019999914022074123,
-      "loss": 46.0,
-      "step": 17274
-    },
-    {
-      "epoch": 1.3207943880574193,
-      "grad_norm": 0.004706586245447397,
-      "learning_rate": 0.0001999991401211348,
-      "loss": 46.0,
-      "step": 17275
-    },
-    {
-      "epoch": 1.3208708450408089,
-      "grad_norm": 0.0007195802172645926,
-      "learning_rate": 0.00019999914002152256,
-      "loss": 46.0,
-      "step": 17276
-    },
-    {
-      "epoch": 1.3209473020241986,
-      "grad_norm": 0.00174514704849571,
-      "learning_rate": 0.00019999913992190454,
-      "loss": 46.0,
-      "step": 17277
-    },
-    {
-      "epoch": 1.3210237590075884,
-      "grad_norm": 0.005880001001060009,
-      "learning_rate": 0.00019999913982228078,
-      "loss": 46.0,
-      "step": 17278
-    },
-    {
-      "epoch": 1.3211002159909782,
-      "grad_norm": 0.0021151211112737656,
-      "learning_rate": 0.00019999913972265128,
-      "loss": 46.0,
-      "step": 17279
-    },
-    {
-      "epoch": 1.3211766729743677,
-      "grad_norm": 0.0013669688487425447,
-      "learning_rate": 0.00019999913962301594,
-      "loss": 46.0,
-      "step": 17280
-    },
-    {
-      "epoch": 1.3212531299577575,
-      "grad_norm": 0.0009188097901642323,
-      "learning_rate": 0.0001999991395233749,
-      "loss": 46.0,
-      "step": 17281
-    },
-    {
-      "epoch": 1.3213295869411472,
-      "grad_norm": 0.0017044630367308855,
-      "learning_rate": 0.00019999913942372804,
-      "loss": 46.0,
-      "step": 17282
-    },
-    {
-      "epoch": 1.321406043924537,
-      "grad_norm": 0.0013585459673777223,
-      "learning_rate": 0.00019999913932407544,
-      "loss": 46.0,
-      "step": 17283
-    },
-    {
-      "epoch": 1.3214825009079267,
-      "grad_norm": 0.0010362127795815468,
-      "learning_rate": 0.00019999913922441706,
-      "loss": 46.0,
-      "step": 17284
-    },
-    {
-      "epoch": 1.3215589578913165,
-      "grad_norm": 0.0021890075877308846,
-      "learning_rate": 0.0001999991391247529,
-      "loss": 46.0,
-      "step": 17285
-    },
-    {
-      "epoch": 1.321635414874706,
-      "grad_norm": 0.0016954620368778706,
-      "learning_rate": 0.00019999913902508297,
-      "loss": 46.0,
-      "step": 17286
-    },
-    {
-      "epoch": 1.3217118718580958,
-      "grad_norm": 0.0016661520348861814,
-      "learning_rate": 0.0001999991389254073,
-      "loss": 46.0,
-      "step": 17287
-    },
-    {
-      "epoch": 1.3217883288414856,
-      "grad_norm": 0.0007691810606047511,
-      "learning_rate": 0.00019999913882572584,
-      "loss": 46.0,
-      "step": 17288
-    },
-    {
-      "epoch": 1.3218647858248753,
-      "grad_norm": 0.0009226563270203769,
-      "learning_rate": 0.00019999913872603863,
-      "loss": 46.0,
-      "step": 17289
-    },
-    {
-      "epoch": 1.321941242808265,
-      "grad_norm": 0.0021697941701859236,
-      "learning_rate": 0.0001999991386263456,
-      "loss": 46.0,
-      "step": 17290
-    },
-    {
-      "epoch": 1.3220176997916546,
-      "grad_norm": 0.0017355961026623845,
-      "learning_rate": 0.00019999913852664685,
-      "loss": 46.0,
-      "step": 17291
-    },
-    {
-      "epoch": 1.3220941567750444,
-      "grad_norm": 0.0017698290757834911,
-      "learning_rate": 0.00019999913842694232,
-      "loss": 46.0,
-      "step": 17292
-    },
-    {
-      "epoch": 1.3221706137584341,
-      "grad_norm": 0.004600329790264368,
-      "learning_rate": 0.00019999913832723201,
-      "loss": 46.0,
-      "step": 17293
-    },
-    {
-      "epoch": 1.322247070741824,
-      "grad_norm": 0.0003432325320318341,
-      "learning_rate": 0.00019999913822751593,
-      "loss": 46.0,
-      "step": 17294
-    },
-    {
-      "epoch": 1.3223235277252137,
-      "grad_norm": 0.006027223076671362,
-      "learning_rate": 0.00019999913812779408,
-      "loss": 46.0,
-      "step": 17295
-    },
-    {
-      "epoch": 1.3223999847086034,
-      "grad_norm": 0.0016782063758000731,
-      "learning_rate": 0.00019999913802806646,
-      "loss": 46.0,
-      "step": 17296
-    },
-    {
-      "epoch": 1.322476441691993,
-      "grad_norm": 0.0010369824012741446,
-      "learning_rate": 0.00019999913792833308,
-      "loss": 46.0,
-      "step": 17297
-    },
-    {
-      "epoch": 1.3225528986753827,
-      "grad_norm": 0.00047666847240179777,
-      "learning_rate": 0.0001999991378285939,
-      "loss": 46.0,
-      "step": 17298
-    },
-    {
-      "epoch": 1.3226293556587725,
-      "grad_norm": 0.0015767350560054183,
-      "learning_rate": 0.00019999913772884902,
-      "loss": 46.0,
-      "step": 17299
-    },
-    {
-      "epoch": 1.3227058126421622,
-      "grad_norm": 0.0014668862568214536,
-      "learning_rate": 0.00019999913762909832,
-      "loss": 46.0,
-      "step": 17300
-    },
-    {
-      "epoch": 1.3227822696255518,
-      "grad_norm": 0.0008369010174646974,
-      "learning_rate": 0.00019999913752934189,
-      "loss": 46.0,
-      "step": 17301
-    },
-    {
-      "epoch": 1.3228587266089415,
-      "grad_norm": 0.000895605597179383,
-      "learning_rate": 0.00019999913742957962,
-      "loss": 46.0,
-      "step": 17302
-    },
-    {
-      "epoch": 1.3229351835923313,
-      "grad_norm": 0.0012225806713104248,
-      "learning_rate": 0.00019999913732981163,
-      "loss": 46.0,
-      "step": 17303
-    },
-    {
-      "epoch": 1.323011640575721,
-      "grad_norm": 0.0007701747235842049,
-      "learning_rate": 0.00019999913723003787,
-      "loss": 46.0,
-      "step": 17304
-    },
-    {
-      "epoch": 1.3230880975591108,
-      "grad_norm": 0.0016745930770412087,
-      "learning_rate": 0.00019999913713025834,
-      "loss": 46.0,
-      "step": 17305
-    },
-    {
-      "epoch": 1.3231645545425006,
-      "grad_norm": 0.002976943738758564,
-      "learning_rate": 0.000199999137030473,
-      "loss": 46.0,
-      "step": 17306
-    },
-    {
-      "epoch": 1.3232410115258904,
-      "grad_norm": 0.004786735400557518,
-      "learning_rate": 0.00019999913693068196,
-      "loss": 46.0,
-      "step": 17307
-    },
-    {
-      "epoch": 1.32331746850928,
-      "grad_norm": 0.01070769689977169,
-      "learning_rate": 0.0001999991368308851,
-      "loss": 46.0,
-      "step": 17308
-    },
-    {
-      "epoch": 1.3233939254926697,
-      "grad_norm": 0.000870837306138128,
-      "learning_rate": 0.00019999913673108248,
-      "loss": 46.0,
-      "step": 17309
-    },
-    {
-      "epoch": 1.3234703824760594,
-      "grad_norm": 0.003156396560370922,
-      "learning_rate": 0.0001999991366312741,
-      "loss": 46.0,
-      "step": 17310
-    },
-    {
-      "epoch": 1.3235468394594492,
-      "grad_norm": 0.0007843500934541225,
-      "learning_rate": 0.00019999913653145993,
-      "loss": 46.0,
-      "step": 17311
-    },
-    {
-      "epoch": 1.3236232964428387,
-      "grad_norm": 0.001117914798669517,
-      "learning_rate": 0.00019999913643164004,
-      "loss": 46.0,
-      "step": 17312
-    },
-    {
-      "epoch": 1.3236997534262285,
-      "grad_norm": 0.016299236565828323,
-      "learning_rate": 0.00019999913633181432,
-      "loss": 46.0,
-      "step": 17313
-    },
-    {
-      "epoch": 1.3237762104096182,
-      "grad_norm": 0.0009847644250839949,
-      "learning_rate": 0.00019999913623198286,
-      "loss": 46.0,
-      "step": 17314
-    },
-    {
-      "epoch": 1.323852667393008,
-      "grad_norm": 0.001436815713532269,
-      "learning_rate": 0.00019999913613214562,
-      "loss": 46.0,
-      "step": 17315
-    },
-    {
-      "epoch": 1.3239291243763978,
-      "grad_norm": 0.00039300520438700914,
-      "learning_rate": 0.00019999913603230263,
-      "loss": 46.0,
-      "step": 17316
-    },
-    {
-      "epoch": 1.3240055813597875,
-      "grad_norm": 0.005245849024504423,
-      "learning_rate": 0.00019999913593245384,
-      "loss": 46.0,
-      "step": 17317
-    },
-    {
-      "epoch": 1.3240820383431773,
-      "grad_norm": 0.0007793670520186424,
-      "learning_rate": 0.0001999991358325993,
-      "loss": 46.0,
-      "step": 17318
-    },
-    {
-      "epoch": 1.3241584953265668,
-      "grad_norm": 0.0005983553710393608,
-      "learning_rate": 0.000199999135732739,
-      "loss": 46.0,
-      "step": 17319
-    },
-    {
-      "epoch": 1.3242349523099566,
-      "grad_norm": 0.001569002284668386,
-      "learning_rate": 0.00019999913563287293,
-      "loss": 46.0,
-      "step": 17320
-    },
-    {
-      "epoch": 1.3243114092933463,
-      "grad_norm": 0.001614886336028576,
-      "learning_rate": 0.0001999991355330011,
-      "loss": 46.0,
-      "step": 17321
-    },
-    {
-      "epoch": 1.324387866276736,
-      "grad_norm": 0.0013751612277701497,
-      "learning_rate": 0.00019999913543312348,
-      "loss": 46.0,
-      "step": 17322
-    },
-    {
-      "epoch": 1.3244643232601256,
-      "grad_norm": 0.0022565105464309454,
-      "learning_rate": 0.00019999913533324008,
-      "loss": 46.0,
-      "step": 17323
-    },
-    {
-      "epoch": 1.3245407802435154,
-      "grad_norm": 0.002738203154876828,
-      "learning_rate": 0.00019999913523335093,
-      "loss": 46.0,
-      "step": 17324
-    },
-    {
-      "epoch": 1.3246172372269052,
-      "grad_norm": 0.000623198866378516,
-      "learning_rate": 0.000199999135133456,
-      "loss": 46.0,
-      "step": 17325
-    },
-    {
-      "epoch": 1.324693694210295,
-      "grad_norm": 0.0009948090882971883,
-      "learning_rate": 0.0001999991350335553,
-      "loss": 46.0,
-      "step": 17326
-    },
-    {
-      "epoch": 1.3247701511936847,
-      "grad_norm": 0.00345505285076797,
-      "learning_rate": 0.00019999913493364883,
-      "loss": 46.0,
-      "step": 17327
-    },
-    {
-      "epoch": 1.3248466081770744,
-      "grad_norm": 0.0011185671901330352,
-      "learning_rate": 0.0001999991348337366,
-      "loss": 46.0,
-      "step": 17328
-    },
-    {
-      "epoch": 1.3249230651604642,
-      "grad_norm": 0.002106211381033063,
-      "learning_rate": 0.0001999991347338186,
-      "loss": 46.0,
-      "step": 17329
-    },
-    {
-      "epoch": 1.3249995221438537,
-      "grad_norm": 0.0007250930066220462,
-      "learning_rate": 0.00019999913463389482,
-      "loss": 46.0,
-      "step": 17330
-    },
-    {
-      "epoch": 1.3250759791272435,
-      "grad_norm": 0.0012351027689874172,
-      "learning_rate": 0.00019999913453396526,
-      "loss": 46.0,
-      "step": 17331
-    },
-    {
-      "epoch": 1.3251524361106333,
-      "grad_norm": 0.0006779627874493599,
-      "learning_rate": 0.00019999913443402996,
-      "loss": 46.0,
-      "step": 17332
-    },
-    {
-      "epoch": 1.325228893094023,
-      "grad_norm": 0.0008195066475309432,
-      "learning_rate": 0.00019999913433408888,
-      "loss": 46.0,
-      "step": 17333
-    },
-    {
-      "epoch": 1.3253053500774126,
-      "grad_norm": 0.0010583183029666543,
-      "learning_rate": 0.00019999913423414202,
-      "loss": 46.0,
-      "step": 17334
-    },
-    {
-      "epoch": 1.3253818070608023,
-      "grad_norm": 0.0011396575719118118,
-      "learning_rate": 0.0001999991341341894,
-      "loss": 46.0,
-      "step": 17335
-    },
-    {
-      "epoch": 1.325458264044192,
-      "grad_norm": 0.0007618501549586654,
-      "learning_rate": 0.000199999134034231,
-      "loss": 46.0,
-      "step": 17336
-    },
-    {
-      "epoch": 1.3255347210275819,
-      "grad_norm": 0.0013859316240996122,
-      "learning_rate": 0.00019999913393426686,
-      "loss": 46.0,
-      "step": 17337
-    },
-    {
-      "epoch": 1.3256111780109716,
-      "grad_norm": 0.002333087846636772,
-      "learning_rate": 0.00019999913383429694,
-      "loss": 46.0,
-      "step": 17338
-    },
-    {
-      "epoch": 1.3256876349943614,
-      "grad_norm": 0.0005865683197043836,
-      "learning_rate": 0.00019999913373432122,
-      "loss": 46.0,
-      "step": 17339
-    },
-    {
-      "epoch": 1.3257640919777511,
-      "grad_norm": 0.0016845760401338339,
-      "learning_rate": 0.00019999913363433978,
-      "loss": 46.0,
-      "step": 17340
-    },
-    {
-      "epoch": 1.3258405489611407,
-      "grad_norm": 0.0005724859656766057,
-      "learning_rate": 0.00019999913353435254,
-      "loss": 46.0,
-      "step": 17341
-    },
-    {
-      "epoch": 1.3259170059445304,
-      "grad_norm": 0.0020432285964488983,
-      "learning_rate": 0.00019999913343435953,
-      "loss": 46.0,
-      "step": 17342
-    },
-    {
-      "epoch": 1.3259934629279202,
-      "grad_norm": 0.0016204017447307706,
-      "learning_rate": 0.00019999913333436074,
-      "loss": 46.0,
-      "step": 17343
-    },
-    {
-      "epoch": 1.32606991991131,
-      "grad_norm": 0.0010332120582461357,
-      "learning_rate": 0.0001999991332343562,
-      "loss": 46.0,
-      "step": 17344
-    },
-    {
-      "epoch": 1.3261463768946995,
-      "grad_norm": 0.00043746328447014093,
-      "learning_rate": 0.0001999991331343459,
-      "loss": 46.0,
-      "step": 17345
-    },
-    {
-      "epoch": 1.3262228338780893,
-      "grad_norm": 0.0010063236113637686,
-      "learning_rate": 0.00019999913303432983,
-      "loss": 46.0,
-      "step": 17346
-    },
-    {
-      "epoch": 1.326299290861479,
-      "grad_norm": 0.004033395554870367,
-      "learning_rate": 0.00019999913293430798,
-      "loss": 46.0,
-      "step": 17347
-    },
-    {
-      "epoch": 1.3263757478448688,
-      "grad_norm": 0.0006785108125768602,
-      "learning_rate": 0.00019999913283428033,
-      "loss": 46.0,
-      "step": 17348
-    },
-    {
-      "epoch": 1.3264522048282585,
-      "grad_norm": 0.0010622142581269145,
-      "learning_rate": 0.00019999913273424693,
-      "loss": 46.0,
-      "step": 17349
-    },
-    {
-      "epoch": 1.3265286618116483,
-      "grad_norm": 0.0031055437866598368,
-      "learning_rate": 0.00019999913263420778,
-      "loss": 46.0,
-      "step": 17350
-    },
-    {
-      "epoch": 1.326605118795038,
-      "grad_norm": 0.00039451938937418163,
-      "learning_rate": 0.00019999913253416284,
-      "loss": 46.0,
-      "step": 17351
-    },
-    {
-      "epoch": 1.3266815757784276,
-      "grad_norm": 0.0008647080976516008,
-      "learning_rate": 0.00019999913243411215,
-      "loss": 46.0,
-      "step": 17352
-    },
-    {
-      "epoch": 1.3267580327618174,
-      "grad_norm": 0.001688301214016974,
-      "learning_rate": 0.00019999913233405568,
-      "loss": 46.0,
-      "step": 17353
-    },
-    {
-      "epoch": 1.3268344897452071,
-      "grad_norm": 0.0012700185179710388,
-      "learning_rate": 0.00019999913223399344,
-      "loss": 46.0,
-      "step": 17354
-    },
-    {
-      "epoch": 1.3269109467285969,
-      "grad_norm": 0.0019978515338152647,
-      "learning_rate": 0.00019999913213392546,
-      "loss": 46.0,
-      "step": 17355
-    },
-    {
-      "epoch": 1.3269874037119864,
-      "grad_norm": 0.0013060306664556265,
-      "learning_rate": 0.00019999913203385165,
-      "loss": 46.0,
-      "step": 17356
-    },
-    {
-      "epoch": 1.3270638606953762,
-      "grad_norm": 0.0015286528505384922,
-      "learning_rate": 0.00019999913193377212,
-      "loss": 46.0,
-      "step": 17357
-    },
-    {
-      "epoch": 1.327140317678766,
-      "grad_norm": 0.005105131771415472,
-      "learning_rate": 0.00019999913183368678,
-      "loss": 46.0,
-      "step": 17358
-    },
-    {
-      "epoch": 1.3272167746621557,
-      "grad_norm": 0.000762317271437496,
-      "learning_rate": 0.0001999991317335957,
-      "loss": 46.0,
-      "step": 17359
-    },
-    {
-      "epoch": 1.3272932316455455,
-      "grad_norm": 0.0011469654273241758,
-      "learning_rate": 0.00019999913163349885,
-      "loss": 46.0,
-      "step": 17360
-    },
-    {
-      "epoch": 1.3273696886289352,
-      "grad_norm": 0.002654470968991518,
-      "learning_rate": 0.00019999913153339623,
-      "loss": 46.0,
-      "step": 17361
-    },
-    {
-      "epoch": 1.327446145612325,
-      "grad_norm": 0.0003295818460173905,
-      "learning_rate": 0.00019999913143328783,
-      "loss": 46.0,
-      "step": 17362
-    },
-    {
-      "epoch": 1.3275226025957145,
-      "grad_norm": 0.0011115336092188954,
-      "learning_rate": 0.00019999913133317366,
-      "loss": 46.0,
-      "step": 17363
-    },
-    {
-      "epoch": 1.3275990595791043,
-      "grad_norm": 0.008734781295061111,
-      "learning_rate": 0.00019999913123305374,
-      "loss": 46.0,
-      "step": 17364
-    },
-    {
-      "epoch": 1.327675516562494,
-      "grad_norm": 0.003452907083556056,
-      "learning_rate": 0.00019999913113292802,
-      "loss": 46.0,
-      "step": 17365
-    },
-    {
-      "epoch": 1.3277519735458838,
-      "grad_norm": 0.0008871277095749974,
-      "learning_rate": 0.00019999913103279656,
-      "loss": 46.0,
-      "step": 17366
-    },
-    {
-      "epoch": 1.3278284305292734,
-      "grad_norm": 0.0012537275906652212,
-      "learning_rate": 0.00019999913093265932,
-      "loss": 46.0,
-      "step": 17367
-    },
-    {
-      "epoch": 1.3279048875126631,
-      "grad_norm": 0.001192585565149784,
-      "learning_rate": 0.0001999991308325163,
-      "loss": 46.0,
-      "step": 17368
-    },
-    {
-      "epoch": 1.3279813444960529,
-      "grad_norm": 0.0018154066056013107,
-      "learning_rate": 0.00019999913073236753,
-      "loss": 46.0,
-      "step": 17369
-    },
-    {
-      "epoch": 1.3280578014794426,
-      "grad_norm": 0.0013291678624227643,
-      "learning_rate": 0.000199999130632213,
-      "loss": 46.0,
-      "step": 17370
-    },
-    {
-      "epoch": 1.3281342584628324,
-      "grad_norm": 0.0005590153159573674,
-      "learning_rate": 0.00019999913053205266,
-      "loss": 46.0,
-      "step": 17371
-    },
-    {
-      "epoch": 1.3282107154462222,
-      "grad_norm": 0.0012035639956593513,
-      "learning_rate": 0.00019999913043188659,
-      "loss": 46.0,
-      "step": 17372
-    },
-    {
-      "epoch": 1.328287172429612,
-      "grad_norm": 0.0034630370792001486,
-      "learning_rate": 0.0001999991303317147,
-      "loss": 46.0,
-      "step": 17373
-    },
-    {
-      "epoch": 1.3283636294130015,
-      "grad_norm": 0.0009711846360005438,
-      "learning_rate": 0.00019999913023153708,
-      "loss": 46.0,
-      "step": 17374
-    },
-    {
-      "epoch": 1.3284400863963912,
-      "grad_norm": 0.004778137430548668,
-      "learning_rate": 0.00019999913013135368,
-      "loss": 46.0,
-      "step": 17375
-    },
-    {
-      "epoch": 1.328516543379781,
-      "grad_norm": 0.0011376874754205346,
-      "learning_rate": 0.00019999913003116454,
-      "loss": 46.0,
-      "step": 17376
-    },
-    {
-      "epoch": 1.3285930003631707,
-      "grad_norm": 0.00118243636097759,
-      "learning_rate": 0.0001999991299309696,
-      "loss": 46.0,
-      "step": 17377
-    },
-    {
-      "epoch": 1.3286694573465603,
-      "grad_norm": 0.004341739695519209,
-      "learning_rate": 0.0001999991298307689,
-      "loss": 46.0,
-      "step": 17378
-    },
-    {
-      "epoch": 1.32874591432995,
-      "grad_norm": 0.0004418192838784307,
-      "learning_rate": 0.00019999912973056244,
-      "loss": 46.0,
-      "step": 17379
-    },
-    {
-      "epoch": 1.3288223713133398,
-      "grad_norm": 0.00048639284796081483,
-      "learning_rate": 0.00019999912963035018,
-      "loss": 46.0,
-      "step": 17380
-    },
-    {
-      "epoch": 1.3288988282967296,
-      "grad_norm": 0.0012268339050933719,
-      "learning_rate": 0.00019999912953013217,
-      "loss": 46.0,
-      "step": 17381
-    },
-    {
-      "epoch": 1.3289752852801193,
-      "grad_norm": 0.005768474191427231,
-      "learning_rate": 0.00019999912942990838,
-      "loss": 46.0,
-      "step": 17382
-    },
-    {
-      "epoch": 1.329051742263509,
-      "grad_norm": 0.00395636260509491,
-      "learning_rate": 0.00019999912932967882,
-      "loss": 46.0,
-      "step": 17383
-    },
-    {
-      "epoch": 1.3291281992468988,
-      "grad_norm": 0.0019664298743009567,
-      "learning_rate": 0.0001999991292294435,
-      "loss": 46.0,
-      "step": 17384
-    },
-    {
-      "epoch": 1.3292046562302884,
-      "grad_norm": 0.0016989157302305102,
-      "learning_rate": 0.0001999991291292024,
-      "loss": 46.0,
-      "step": 17385
-    },
-    {
-      "epoch": 1.3292811132136781,
-      "grad_norm": 0.00261163292452693,
-      "learning_rate": 0.00019999912902895556,
-      "loss": 46.0,
-      "step": 17386
-    },
-    {
-      "epoch": 1.329357570197068,
-      "grad_norm": 0.002815736923366785,
-      "learning_rate": 0.0001999991289287029,
-      "loss": 46.0,
-      "step": 17387
-    },
-    {
-      "epoch": 1.3294340271804577,
-      "grad_norm": 0.0008977506658993661,
-      "learning_rate": 0.00019999912882844454,
-      "loss": 46.0,
-      "step": 17388
-    },
-    {
-      "epoch": 1.3295104841638472,
-      "grad_norm": 0.0004119258082937449,
-      "learning_rate": 0.00019999912872818034,
-      "loss": 46.0,
-      "step": 17389
-    },
-    {
-      "epoch": 1.329586941147237,
-      "grad_norm": 0.0009729574667289853,
-      "learning_rate": 0.00019999912862791043,
-      "loss": 46.0,
-      "step": 17390
-    },
-    {
-      "epoch": 1.3296633981306267,
-      "grad_norm": 0.0008385512628592551,
-      "learning_rate": 0.0001999991285276347,
-      "loss": 46.0,
-      "step": 17391
-    },
-    {
-      "epoch": 1.3297398551140165,
-      "grad_norm": 0.0010143026011064649,
-      "learning_rate": 0.00019999912842735325,
-      "loss": 46.0,
-      "step": 17392
-    },
-    {
-      "epoch": 1.3298163120974063,
-      "grad_norm": 0.0022218613885343075,
-      "learning_rate": 0.000199999128327066,
-      "loss": 46.0,
-      "step": 17393
-    },
-    {
-      "epoch": 1.329892769080796,
-      "grad_norm": 0.0006187410326674581,
-      "learning_rate": 0.00019999912822677297,
-      "loss": 46.0,
-      "step": 17394
-    },
-    {
-      "epoch": 1.3299692260641858,
-      "grad_norm": 0.001159863080829382,
-      "learning_rate": 0.0001999991281264742,
-      "loss": 46.0,
-      "step": 17395
-    },
-    {
-      "epoch": 1.3300456830475753,
-      "grad_norm": 0.003690300974994898,
-      "learning_rate": 0.00019999912802616963,
-      "loss": 46.0,
-      "step": 17396
-    },
-    {
-      "epoch": 1.330122140030965,
-      "grad_norm": 0.0008360066567547619,
-      "learning_rate": 0.0001999991279258593,
-      "loss": 46.0,
-      "step": 17397
-    },
-    {
-      "epoch": 1.3301985970143548,
-      "grad_norm": 0.004908973351120949,
-      "learning_rate": 0.00019999912782554322,
-      "loss": 46.0,
-      "step": 17398
-    },
-    {
-      "epoch": 1.3302750539977446,
-      "grad_norm": 0.0005259255995042622,
-      "learning_rate": 0.00019999912772522137,
-      "loss": 46.0,
-      "step": 17399
-    },
-    {
-      "epoch": 1.3303515109811341,
-      "grad_norm": 0.00280459295026958,
-      "learning_rate": 0.00019999912762489372,
-      "loss": 46.0,
-      "step": 17400
-    },
-    {
-      "epoch": 1.330427967964524,
-      "grad_norm": 0.00237468839623034,
-      "learning_rate": 0.00019999912752456035,
-      "loss": 46.0,
-      "step": 17401
-    },
-    {
-      "epoch": 1.3305044249479137,
-      "grad_norm": 0.0003370768390595913,
-      "learning_rate": 0.00019999912742422115,
-      "loss": 46.0,
-      "step": 17402
-    },
-    {
-      "epoch": 1.3305808819313034,
-      "grad_norm": 0.0010442129569128156,
-      "learning_rate": 0.0001999991273238762,
-      "loss": 46.0,
-      "step": 17403
-    },
-    {
-      "epoch": 1.3306573389146932,
-      "grad_norm": 0.0016302593285217881,
-      "learning_rate": 0.0001999991272235255,
-      "loss": 46.0,
-      "step": 17404
-    },
-    {
-      "epoch": 1.330733795898083,
-      "grad_norm": 0.0019970976281911135,
-      "learning_rate": 0.00019999912712316903,
-      "loss": 46.0,
-      "step": 17405
-    },
-    {
-      "epoch": 1.3308102528814727,
-      "grad_norm": 0.0015073231188580394,
-      "learning_rate": 0.0001999991270228068,
-      "loss": 46.0,
-      "step": 17406
-    },
-    {
-      "epoch": 1.3308867098648622,
-      "grad_norm": 0.003061566036194563,
-      "learning_rate": 0.00019999912692243875,
-      "loss": 46.0,
-      "step": 17407
-    },
-    {
-      "epoch": 1.330963166848252,
-      "grad_norm": 0.003795386990532279,
-      "learning_rate": 0.000199999126822065,
-      "loss": 46.0,
-      "step": 17408
-    },
-    {
-      "epoch": 1.3310396238316418,
-      "grad_norm": 0.0045354003086686134,
-      "learning_rate": 0.0001999991267216854,
-      "loss": 46.0,
-      "step": 17409
-    },
-    {
-      "epoch": 1.3311160808150315,
-      "grad_norm": 0.000849285046570003,
-      "learning_rate": 0.00019999912662130008,
-      "loss": 46.0,
-      "step": 17410
-    },
-    {
-      "epoch": 1.331192537798421,
-      "grad_norm": 0.0013559279032051563,
-      "learning_rate": 0.000199999126520909,
-      "loss": 46.0,
-      "step": 17411
-    },
-    {
-      "epoch": 1.3312689947818108,
-      "grad_norm": 0.0004952421295456588,
-      "learning_rate": 0.00019999912642051215,
-      "loss": 46.0,
-      "step": 17412
-    },
-    {
-      "epoch": 1.3313454517652006,
-      "grad_norm": 0.0004372815601527691,
-      "learning_rate": 0.00019999912632010948,
-      "loss": 46.0,
-      "step": 17413
-    },
-    {
-      "epoch": 1.3314219087485903,
-      "grad_norm": 0.003960534930229187,
-      "learning_rate": 0.0001999991262197011,
-      "loss": 46.0,
-      "step": 17414
-    },
-    {
-      "epoch": 1.33149836573198,
-      "grad_norm": 0.0007412740960717201,
-      "learning_rate": 0.00019999912611928694,
-      "loss": 46.0,
-      "step": 17415
-    },
-    {
-      "epoch": 1.3315748227153699,
-      "grad_norm": 0.00043966012890450656,
-      "learning_rate": 0.000199999126018867,
-      "loss": 46.0,
-      "step": 17416
-    },
-    {
-      "epoch": 1.3316512796987594,
-      "grad_norm": 0.015806972980499268,
-      "learning_rate": 0.00019999912591844128,
-      "loss": 46.0,
-      "step": 17417
-    },
-    {
-      "epoch": 1.3317277366821492,
-      "grad_norm": 0.0011110759805887938,
-      "learning_rate": 0.00019999912581800979,
-      "loss": 46.0,
-      "step": 17418
-    },
-    {
-      "epoch": 1.331804193665539,
-      "grad_norm": 0.0011276552686467767,
-      "learning_rate": 0.00019999912571757252,
-      "loss": 46.0,
-      "step": 17419
-    },
-    {
-      "epoch": 1.3318806506489287,
-      "grad_norm": 0.0028547486290335655,
-      "learning_rate": 0.0001999991256171295,
-      "loss": 46.0,
-      "step": 17420
-    },
-    {
-      "epoch": 1.3319571076323184,
-      "grad_norm": 0.005724901333451271,
-      "learning_rate": 0.00019999912551668073,
-      "loss": 46.0,
-      "step": 17421
-    },
-    {
-      "epoch": 1.332033564615708,
-      "grad_norm": 0.001001560129225254,
-      "learning_rate": 0.00019999912541622617,
-      "loss": 46.0,
-      "step": 17422
-    },
-    {
-      "epoch": 1.3321100215990977,
-      "grad_norm": 0.011771823279559612,
-      "learning_rate": 0.00019999912531576587,
-      "loss": 46.0,
-      "step": 17423
-    },
-    {
-      "epoch": 1.3321864785824875,
-      "grad_norm": 0.0005923372809775174,
-      "learning_rate": 0.00019999912521529977,
-      "loss": 46.0,
-      "step": 17424
-    },
-    {
-      "epoch": 1.3322629355658773,
-      "grad_norm": 0.0007965003605931997,
-      "learning_rate": 0.0001999991251148279,
-      "loss": 46.0,
-      "step": 17425
-    },
-    {
-      "epoch": 1.332339392549267,
-      "grad_norm": 0.0034073968417942524,
-      "learning_rate": 0.00019999912501435027,
-      "loss": 46.0,
-      "step": 17426
-    },
-    {
-      "epoch": 1.3324158495326568,
-      "grad_norm": 0.0006852917140349746,
-      "learning_rate": 0.00019999912491386684,
-      "loss": 46.0,
-      "step": 17427
-    },
-    {
-      "epoch": 1.3324923065160463,
-      "grad_norm": 0.0002649962261784822,
-      "learning_rate": 0.00019999912481337767,
-      "loss": 46.0,
-      "step": 17428
-    },
-    {
-      "epoch": 1.332568763499436,
-      "grad_norm": 0.00125643250066787,
-      "learning_rate": 0.00019999912471288273,
-      "loss": 46.0,
-      "step": 17429
-    },
-    {
-      "epoch": 1.3326452204828259,
-      "grad_norm": 0.008375952951610088,
-      "learning_rate": 0.000199999124612382,
-      "loss": 46.0,
-      "step": 17430
-    },
-    {
-      "epoch": 1.3327216774662156,
-      "grad_norm": 0.002601398155093193,
-      "learning_rate": 0.00019999912451187552,
-      "loss": 46.0,
-      "step": 17431
-    },
-    {
-      "epoch": 1.3327981344496052,
-      "grad_norm": 0.002572769531980157,
-      "learning_rate": 0.00019999912441136329,
-      "loss": 46.0,
-      "step": 17432
-    },
-    {
-      "epoch": 1.332874591432995,
-      "grad_norm": 0.007171506993472576,
-      "learning_rate": 0.00019999912431084528,
-      "loss": 46.0,
-      "step": 17433
-    },
-    {
-      "epoch": 1.3329510484163847,
-      "grad_norm": 0.0010910426499322057,
-      "learning_rate": 0.00019999912421032147,
-      "loss": 46.0,
-      "step": 17434
-    },
-    {
-      "epoch": 1.3330275053997744,
-      "grad_norm": 0.001661907765083015,
-      "learning_rate": 0.0001999991241097919,
-      "loss": 46.0,
-      "step": 17435
-    },
-    {
-      "epoch": 1.3331039623831642,
-      "grad_norm": 0.004399863071739674,
-      "learning_rate": 0.00019999912400925658,
-      "loss": 46.0,
-      "step": 17436
-    },
-    {
-      "epoch": 1.333180419366554,
-      "grad_norm": 0.0007236532401293516,
-      "learning_rate": 0.00019999912390871548,
-      "loss": 46.0,
-      "step": 17437
-    },
-    {
-      "epoch": 1.3332568763499437,
-      "grad_norm": 0.0009657396003603935,
-      "learning_rate": 0.0001999991238081686,
-      "loss": 46.0,
-      "step": 17438
-    },
-    {
-      "epoch": 1.3333333333333333,
-      "grad_norm": 0.0004946592380292714,
-      "learning_rate": 0.00019999912370761598,
-      "loss": 46.0,
-      "step": 17439
-    },
-    {
-      "epoch": 1.333409790316723,
-      "grad_norm": 0.0009828481124714017,
-      "learning_rate": 0.00019999912360705758,
-      "loss": 46.0,
-      "step": 17440
-    },
-    {
-      "epoch": 1.3334862473001128,
-      "grad_norm": 0.0016624336130917072,
-      "learning_rate": 0.00019999912350649338,
-      "loss": 46.0,
-      "step": 17441
-    },
-    {
-      "epoch": 1.3335627042835025,
-      "grad_norm": 0.0024372676853090525,
-      "learning_rate": 0.00019999912340592344,
-      "loss": 46.0,
-      "step": 17442
-    },
-    {
-      "epoch": 1.333639161266892,
-      "grad_norm": 0.004044575151056051,
-      "learning_rate": 0.00019999912330534772,
-      "loss": 46.0,
-      "step": 17443
-    },
-    {
-      "epoch": 1.3337156182502818,
-      "grad_norm": 0.002666729735210538,
-      "learning_rate": 0.00019999912320476623,
-      "loss": 46.0,
-      "step": 17444
-    },
-    {
-      "epoch": 1.3337920752336716,
-      "grad_norm": 0.0012696458725258708,
-      "learning_rate": 0.000199999123104179,
-      "loss": 46.0,
-      "step": 17445
-    },
-    {
-      "epoch": 1.3338685322170614,
-      "grad_norm": 0.0024716819170862436,
-      "learning_rate": 0.00019999912300358593,
-      "loss": 46.0,
-      "step": 17446
-    },
-    {
-      "epoch": 1.3339449892004511,
-      "grad_norm": 0.017424043267965317,
-      "learning_rate": 0.00019999912290298718,
-      "loss": 46.0,
-      "step": 17447
-    },
-    {
-      "epoch": 1.3340214461838409,
-      "grad_norm": 0.0008413971518166363,
-      "learning_rate": 0.0001999991228023826,
-      "loss": 46.0,
-      "step": 17448
-    },
-    {
-      "epoch": 1.3340979031672306,
-      "grad_norm": 0.0008915279759094119,
-      "learning_rate": 0.00019999912270177227,
-      "loss": 46.0,
-      "step": 17449
-    },
-    {
-      "epoch": 1.3341743601506202,
-      "grad_norm": 0.0019343190360814333,
-      "learning_rate": 0.00019999912260115616,
-      "loss": 46.0,
-      "step": 17450
-    },
-    {
-      "epoch": 1.33425081713401,
-      "grad_norm": 0.0018069229554384947,
-      "learning_rate": 0.0001999991225005343,
-      "loss": 46.0,
-      "step": 17451
-    },
-    {
-      "epoch": 1.3343272741173997,
-      "grad_norm": 0.005727724637836218,
-      "learning_rate": 0.00019999912239990666,
-      "loss": 46.0,
-      "step": 17452
-    },
-    {
-      "epoch": 1.3344037311007895,
-      "grad_norm": 0.0010423107305541635,
-      "learning_rate": 0.00019999912229927324,
-      "loss": 46.0,
-      "step": 17453
-    },
-    {
-      "epoch": 1.334480188084179,
-      "grad_norm": 0.0004155821225140244,
-      "learning_rate": 0.00019999912219863407,
-      "loss": 46.0,
-      "step": 17454
-    },
-    {
-      "epoch": 1.3345566450675688,
-      "grad_norm": 0.0008572808001190424,
-      "learning_rate": 0.00019999912209798913,
-      "loss": 46.0,
-      "step": 17455
-    },
-    {
-      "epoch": 1.3346331020509585,
-      "grad_norm": 0.0019973027519881725,
-      "learning_rate": 0.0001999991219973384,
-      "loss": 46.0,
-      "step": 17456
-    },
-    {
-      "epoch": 1.3347095590343483,
-      "grad_norm": 0.0005126166506670415,
-      "learning_rate": 0.0001999991218966819,
-      "loss": 46.0,
-      "step": 17457
-    },
-    {
-      "epoch": 1.334786016017738,
-      "grad_norm": 0.0015972558176144958,
-      "learning_rate": 0.00019999912179601966,
-      "loss": 46.0,
-      "step": 17458
-    },
-    {
-      "epoch": 1.3348624730011278,
-      "grad_norm": 0.003226458327844739,
-      "learning_rate": 0.00019999912169535162,
-      "loss": 46.0,
-      "step": 17459
-    },
-    {
-      "epoch": 1.3349389299845176,
-      "grad_norm": 0.0010951668955385685,
-      "learning_rate": 0.00019999912159467784,
-      "loss": 46.0,
-      "step": 17460
-    },
-    {
-      "epoch": 1.3350153869679071,
-      "grad_norm": 0.0008140777354128659,
-      "learning_rate": 0.00019999912149399826,
-      "loss": 46.0,
-      "step": 17461
-    },
-    {
-      "epoch": 1.3350918439512969,
-      "grad_norm": 0.0005094542866572738,
-      "learning_rate": 0.00019999912139331293,
-      "loss": 46.0,
-      "step": 17462
-    },
-    {
-      "epoch": 1.3351683009346866,
-      "grad_norm": 0.0013758456334471703,
-      "learning_rate": 0.00019999912129262183,
-      "loss": 46.0,
-      "step": 17463
-    },
-    {
-      "epoch": 1.3352447579180764,
-      "grad_norm": 0.002713556634262204,
-      "learning_rate": 0.00019999912119192495,
-      "loss": 46.0,
-      "step": 17464
-    },
-    {
-      "epoch": 1.335321214901466,
-      "grad_norm": 0.0011309538967907429,
-      "learning_rate": 0.0001999991210912223,
-      "loss": 46.0,
-      "step": 17465
-    },
-    {
-      "epoch": 1.3353976718848557,
-      "grad_norm": 0.0012767646694555879,
-      "learning_rate": 0.0001999991209905139,
-      "loss": 46.0,
-      "step": 17466
-    },
-    {
-      "epoch": 1.3354741288682455,
-      "grad_norm": 0.001357071683742106,
-      "learning_rate": 0.00019999912088979974,
-      "loss": 46.0,
-      "step": 17467
-    },
-    {
-      "epoch": 1.3355505858516352,
-      "grad_norm": 0.009967680089175701,
-      "learning_rate": 0.00019999912078907977,
-      "loss": 46.0,
-      "step": 17468
-    },
-    {
-      "epoch": 1.335627042835025,
-      "grad_norm": 0.0008076328667812049,
-      "learning_rate": 0.00019999912068835405,
-      "loss": 46.0,
-      "step": 17469
-    },
-    {
-      "epoch": 1.3357034998184147,
-      "grad_norm": 0.001356467604637146,
-      "learning_rate": 0.00019999912058762256,
-      "loss": 46.0,
-      "step": 17470
-    },
-    {
-      "epoch": 1.3357799568018045,
-      "grad_norm": 0.0007080992218106985,
-      "learning_rate": 0.0001999991204868853,
-      "loss": 46.0,
-      "step": 17471
-    },
-    {
-      "epoch": 1.335856413785194,
-      "grad_norm": 0.0005376964691095054,
-      "learning_rate": 0.00019999912038614227,
-      "loss": 46.0,
-      "step": 17472
-    },
-    {
-      "epoch": 1.3359328707685838,
-      "grad_norm": 0.0011065345024690032,
-      "learning_rate": 0.00019999912028539348,
-      "loss": 46.0,
-      "step": 17473
-    },
-    {
-      "epoch": 1.3360093277519736,
-      "grad_norm": 0.0005677349399775267,
-      "learning_rate": 0.0001999991201846389,
-      "loss": 46.0,
-      "step": 17474
-    },
-    {
-      "epoch": 1.3360857847353633,
-      "grad_norm": 0.0005541481077671051,
-      "learning_rate": 0.00019999912008387857,
-      "loss": 46.0,
-      "step": 17475
-    },
-    {
-      "epoch": 1.3361622417187529,
-      "grad_norm": 0.0010768037755042315,
-      "learning_rate": 0.00019999911998311247,
-      "loss": 46.0,
-      "step": 17476
-    },
-    {
-      "epoch": 1.3362386987021426,
-      "grad_norm": 0.0036428302992135286,
-      "learning_rate": 0.0001999991198823406,
-      "loss": 46.0,
-      "step": 17477
-    },
-    {
-      "epoch": 1.3363151556855324,
-      "grad_norm": 0.0006980306934565306,
-      "learning_rate": 0.00019999911978156294,
-      "loss": 46.0,
-      "step": 17478
-    },
-    {
-      "epoch": 1.3363916126689221,
-      "grad_norm": 0.004891026299446821,
-      "learning_rate": 0.00019999911968077952,
-      "loss": 46.0,
-      "step": 17479
-    },
-    {
-      "epoch": 1.336468069652312,
-      "grad_norm": 0.0007515577599406242,
-      "learning_rate": 0.00019999911957999035,
-      "loss": 46.0,
-      "step": 17480
-    },
-    {
-      "epoch": 1.3365445266357017,
-      "grad_norm": 0.0016693056095391512,
-      "learning_rate": 0.00019999911947919538,
-      "loss": 46.0,
-      "step": 17481
-    },
-    {
-      "epoch": 1.3366209836190914,
-      "grad_norm": 0.0007030356791801751,
-      "learning_rate": 0.0001999991193783947,
-      "loss": 46.0,
-      "step": 17482
-    },
-    {
-      "epoch": 1.336697440602481,
-      "grad_norm": 0.0011476923245936632,
-      "learning_rate": 0.00019999911927758815,
-      "loss": 46.0,
-      "step": 17483
-    },
-    {
-      "epoch": 1.3367738975858707,
-      "grad_norm": 0.0023628147318959236,
-      "learning_rate": 0.00019999911917677592,
-      "loss": 46.0,
-      "step": 17484
-    },
-    {
-      "epoch": 1.3368503545692605,
-      "grad_norm": 0.0007388531230390072,
-      "learning_rate": 0.0001999991190759579,
-      "loss": 46.0,
-      "step": 17485
-    },
-    {
-      "epoch": 1.3369268115526503,
-      "grad_norm": 0.001888773636892438,
-      "learning_rate": 0.00019999911897513407,
-      "loss": 46.0,
-      "step": 17486
-    },
-    {
-      "epoch": 1.3370032685360398,
-      "grad_norm": 0.0008290036930702627,
-      "learning_rate": 0.00019999911887430452,
-      "loss": 46.0,
-      "step": 17487
-    },
-    {
-      "epoch": 1.3370797255194296,
-      "grad_norm": 0.0006322208209894598,
-      "learning_rate": 0.00019999911877346916,
-      "loss": 46.0,
-      "step": 17488
-    },
-    {
-      "epoch": 1.3371561825028193,
-      "grad_norm": 0.0021140817552804947,
-      "learning_rate": 0.0001999991186726281,
-      "loss": 46.0,
-      "step": 17489
-    },
-    {
-      "epoch": 1.337232639486209,
-      "grad_norm": 0.004914324264973402,
-      "learning_rate": 0.00019999911857178119,
-      "loss": 46.0,
-      "step": 17490
-    },
-    {
-      "epoch": 1.3373090964695988,
-      "grad_norm": 0.0008868437143974006,
-      "learning_rate": 0.00019999911847092854,
-      "loss": 46.0,
-      "step": 17491
-    },
-    {
-      "epoch": 1.3373855534529886,
-      "grad_norm": 0.003679433139041066,
-      "learning_rate": 0.00019999911837007011,
-      "loss": 46.0,
-      "step": 17492
-    },
-    {
-      "epoch": 1.3374620104363784,
-      "grad_norm": 0.0010031504789367318,
-      "learning_rate": 0.00019999911826920595,
-      "loss": 46.0,
-      "step": 17493
-    },
-    {
-      "epoch": 1.337538467419768,
-      "grad_norm": 0.0012689121067523956,
-      "learning_rate": 0.00019999911816833598,
-      "loss": 46.0,
-      "step": 17494
-    },
-    {
-      "epoch": 1.3376149244031577,
-      "grad_norm": 0.0010104383109137416,
-      "learning_rate": 0.00019999911806746026,
-      "loss": 46.0,
-      "step": 17495
-    },
-    {
-      "epoch": 1.3376913813865474,
-      "grad_norm": 0.0006006374023854733,
-      "learning_rate": 0.00019999911796657877,
-      "loss": 46.0,
-      "step": 17496
-    },
-    {
-      "epoch": 1.3377678383699372,
-      "grad_norm": 0.0014534505316987634,
-      "learning_rate": 0.0001999991178656915,
-      "loss": 46.0,
-      "step": 17497
-    },
-    {
-      "epoch": 1.3378442953533267,
-      "grad_norm": 0.0008841780945658684,
-      "learning_rate": 0.00019999911776479848,
-      "loss": 46.0,
-      "step": 17498
-    },
-    {
-      "epoch": 1.3379207523367165,
-      "grad_norm": 0.0021158214658498764,
-      "learning_rate": 0.00019999911766389967,
-      "loss": 46.0,
-      "step": 17499
-    },
-    {
-      "epoch": 1.3379972093201062,
-      "grad_norm": 0.004220927599817514,
-      "learning_rate": 0.00019999911756299509,
-      "loss": 46.0,
-      "step": 17500
-    },
-    {
-      "epoch": 1.338073666303496,
-      "grad_norm": 0.0014245593920350075,
-      "learning_rate": 0.00019999911746208476,
-      "loss": 46.0,
-      "step": 17501
-    },
-    {
-      "epoch": 1.3381501232868858,
-      "grad_norm": 0.0011016873177140951,
-      "learning_rate": 0.00019999911736116866,
-      "loss": 46.0,
-      "step": 17502
-    },
-    {
-      "epoch": 1.3382265802702755,
-      "grad_norm": 0.00121684733312577,
-      "learning_rate": 0.00019999911726024675,
-      "loss": 46.0,
-      "step": 17503
-    },
-    {
-      "epoch": 1.3383030372536653,
-      "grad_norm": 0.002816010732203722,
-      "learning_rate": 0.0001999991171593191,
-      "loss": 46.0,
-      "step": 17504
-    },
-    {
-      "epoch": 1.3383794942370548,
-      "grad_norm": 0.0015124849742278457,
-      "learning_rate": 0.0001999991170583857,
-      "loss": 46.0,
-      "step": 17505
-    },
-    {
-      "epoch": 1.3384559512204446,
-      "grad_norm": 0.006849550176411867,
-      "learning_rate": 0.00019999911695744652,
-      "loss": 46.0,
-      "step": 17506
-    },
-    {
-      "epoch": 1.3385324082038343,
-      "grad_norm": 0.0012817404931411147,
-      "learning_rate": 0.00019999911685650155,
-      "loss": 46.0,
-      "step": 17507
-    },
-    {
-      "epoch": 1.338608865187224,
-      "grad_norm": 0.000825156515929848,
-      "learning_rate": 0.00019999911675555083,
-      "loss": 46.0,
-      "step": 17508
-    },
-    {
-      "epoch": 1.3386853221706136,
-      "grad_norm": 0.0016330397920683026,
-      "learning_rate": 0.00019999911665459432,
-      "loss": 46.0,
-      "step": 17509
-    },
-    {
-      "epoch": 1.3387617791540034,
-      "grad_norm": 0.0031566356774419546,
-      "learning_rate": 0.00019999911655363206,
-      "loss": 46.0,
-      "step": 17510
-    },
-    {
-      "epoch": 1.3388382361373932,
-      "grad_norm": 0.002160874893888831,
-      "learning_rate": 0.00019999911645266402,
-      "loss": 46.0,
-      "step": 17511
-    },
-    {
-      "epoch": 1.338914693120783,
-      "grad_norm": 0.003055114531889558,
-      "learning_rate": 0.0001999991163516902,
-      "loss": 46.0,
-      "step": 17512
-    },
-    {
-      "epoch": 1.3389911501041727,
-      "grad_norm": 0.0018241537036374211,
-      "learning_rate": 0.00019999911625071066,
-      "loss": 46.0,
-      "step": 17513
-    },
-    {
-      "epoch": 1.3390676070875625,
-      "grad_norm": 0.0007302055601030588,
-      "learning_rate": 0.0001999991161497253,
-      "loss": 46.0,
-      "step": 17514
-    },
-    {
-      "epoch": 1.3391440640709522,
-      "grad_norm": 0.0003395903331693262,
-      "learning_rate": 0.00019999911604873417,
-      "loss": 46.0,
-      "step": 17515
-    },
-    {
-      "epoch": 1.3392205210543418,
-      "grad_norm": 0.0009740596287883818,
-      "learning_rate": 0.0001999991159477373,
-      "loss": 46.0,
-      "step": 17516
-    },
-    {
-      "epoch": 1.3392969780377315,
-      "grad_norm": 0.0013622610131278634,
-      "learning_rate": 0.00019999911584673465,
-      "loss": 46.0,
-      "step": 17517
-    },
-    {
-      "epoch": 1.3393734350211213,
-      "grad_norm": 0.0014347624965012074,
-      "learning_rate": 0.00019999911574572626,
-      "loss": 46.0,
-      "step": 17518
-    },
-    {
-      "epoch": 1.339449892004511,
-      "grad_norm": 0.0010314049432054162,
-      "learning_rate": 0.00019999911564471203,
-      "loss": 46.0,
-      "step": 17519
-    },
-    {
-      "epoch": 1.3395263489879006,
-      "grad_norm": 0.00046051066601648927,
-      "learning_rate": 0.0001999991155436921,
-      "loss": 46.0,
-      "step": 17520
-    },
-    {
-      "epoch": 1.3396028059712903,
-      "grad_norm": 0.00027199723990634084,
-      "learning_rate": 0.00019999911544266635,
-      "loss": 46.0,
-      "step": 17521
-    },
-    {
-      "epoch": 1.33967926295468,
-      "grad_norm": 0.0012747214641422033,
-      "learning_rate": 0.00019999911534163486,
-      "loss": 46.0,
-      "step": 17522
-    },
-    {
-      "epoch": 1.3397557199380699,
-      "grad_norm": 0.0023008554708212614,
-      "learning_rate": 0.0001999991152405976,
-      "loss": 46.0,
-      "step": 17523
-    },
-    {
-      "epoch": 1.3398321769214596,
-      "grad_norm": 0.0017643632600083947,
-      "learning_rate": 0.00019999911513955457,
-      "loss": 46.0,
-      "step": 17524
-    },
-    {
-      "epoch": 1.3399086339048494,
-      "grad_norm": 0.0009348672465421259,
-      "learning_rate": 0.00019999911503850573,
-      "loss": 46.0,
-      "step": 17525
-    },
-    {
-      "epoch": 1.3399850908882391,
-      "grad_norm": 0.000584596476983279,
-      "learning_rate": 0.00019999911493745115,
-      "loss": 46.0,
-      "step": 17526
-    },
-    {
-      "epoch": 1.3400615478716287,
-      "grad_norm": 0.0036607247311621904,
-      "learning_rate": 0.00019999911483639082,
-      "loss": 46.0,
-      "step": 17527
-    },
-    {
-      "epoch": 1.3401380048550184,
-      "grad_norm": 0.0008729560649953783,
-      "learning_rate": 0.00019999911473532472,
-      "loss": 46.0,
-      "step": 17528
-    },
-    {
-      "epoch": 1.3402144618384082,
-      "grad_norm": 0.0009502782486379147,
-      "learning_rate": 0.00019999911463425282,
-      "loss": 46.0,
-      "step": 17529
-    },
-    {
-      "epoch": 1.340290918821798,
-      "grad_norm": 0.0015565053327009082,
-      "learning_rate": 0.00019999911453317517,
-      "loss": 46.0,
-      "step": 17530
-    },
-    {
-      "epoch": 1.3403673758051875,
-      "grad_norm": 0.0023390448186546564,
-      "learning_rate": 0.00019999911443209175,
-      "loss": 46.0,
-      "step": 17531
-    },
-    {
-      "epoch": 1.3404438327885773,
-      "grad_norm": 0.0009186035022139549,
-      "learning_rate": 0.00019999911433100256,
-      "loss": 46.0,
-      "step": 17532
-    },
-    {
-      "epoch": 1.340520289771967,
-      "grad_norm": 0.0011317494791001081,
-      "learning_rate": 0.0001999991142299076,
-      "loss": 46.0,
-      "step": 17533
-    },
-    {
-      "epoch": 1.3405967467553568,
-      "grad_norm": 0.005681768525391817,
-      "learning_rate": 0.00019999911412880685,
-      "loss": 46.0,
-      "step": 17534
-    },
-    {
-      "epoch": 1.3406732037387465,
-      "grad_norm": 0.0020753771532326937,
-      "learning_rate": 0.00019999911402770036,
-      "loss": 46.0,
-      "step": 17535
-    },
-    {
-      "epoch": 1.3407496607221363,
-      "grad_norm": 0.001430074917152524,
-      "learning_rate": 0.00019999911392658807,
-      "loss": 46.0,
-      "step": 17536
-    },
-    {
-      "epoch": 1.340826117705526,
-      "grad_norm": 0.002051582559943199,
-      "learning_rate": 0.00019999911382547004,
-      "loss": 46.0,
-      "step": 17537
-    },
-    {
-      "epoch": 1.3409025746889156,
-      "grad_norm": 0.0010138758225366473,
-      "learning_rate": 0.00019999911372434626,
-      "loss": 46.0,
-      "step": 17538
-    },
-    {
-      "epoch": 1.3409790316723054,
-      "grad_norm": 0.0005562843871302903,
-      "learning_rate": 0.00019999911362321665,
-      "loss": 46.0,
-      "step": 17539
-    },
-    {
-      "epoch": 1.3410554886556951,
-      "grad_norm": 0.01306749228388071,
-      "learning_rate": 0.00019999911352208132,
-      "loss": 46.0,
-      "step": 17540
-    },
-    {
-      "epoch": 1.3411319456390849,
-      "grad_norm": 0.005159082822501659,
-      "learning_rate": 0.0001999991134209402,
-      "loss": 46.0,
-      "step": 17541
-    },
-    {
-      "epoch": 1.3412084026224744,
-      "grad_norm": 0.0009739922825247049,
-      "learning_rate": 0.0001999991133197933,
-      "loss": 46.0,
-      "step": 17542
-    },
-    {
-      "epoch": 1.3412848596058642,
-      "grad_norm": 0.0014173141680657864,
-      "learning_rate": 0.00019999911321864065,
-      "loss": 46.0,
-      "step": 17543
-    },
-    {
-      "epoch": 1.341361316589254,
-      "grad_norm": 0.012223972007632256,
-      "learning_rate": 0.00019999911311748223,
-      "loss": 46.0,
-      "step": 17544
-    },
-    {
-      "epoch": 1.3414377735726437,
-      "grad_norm": 0.00590596254914999,
-      "learning_rate": 0.00019999911301631803,
-      "loss": 46.0,
-      "step": 17545
-    },
-    {
-      "epoch": 1.3415142305560335,
-      "grad_norm": 0.008157900534570217,
-      "learning_rate": 0.0001999991129151481,
-      "loss": 46.0,
-      "step": 17546
-    },
-    {
-      "epoch": 1.3415906875394232,
-      "grad_norm": 0.0003982759080827236,
-      "learning_rate": 0.00019999911281397235,
-      "loss": 46.0,
-      "step": 17547
-    },
-    {
-      "epoch": 1.3416671445228128,
-      "grad_norm": 0.0034173447638750076,
-      "learning_rate": 0.00019999911271279084,
-      "loss": 46.0,
-      "step": 17548
-    },
-    {
-      "epoch": 1.3417436015062025,
-      "grad_norm": 0.0007977579371072352,
-      "learning_rate": 0.00019999911261160355,
-      "loss": 46.0,
-      "step": 17549
-    },
-    {
-      "epoch": 1.3418200584895923,
-      "grad_norm": 0.0013559350045397878,
-      "learning_rate": 0.00019999911251041051,
-      "loss": 46.0,
-      "step": 17550
-    },
-    {
-      "epoch": 1.341896515472982,
-      "grad_norm": 0.0009482645546086133,
-      "learning_rate": 0.0001999991124092117,
-      "loss": 46.0,
-      "step": 17551
-    },
-    {
-      "epoch": 1.3419729724563718,
-      "grad_norm": 0.0034379465505480766,
-      "learning_rate": 0.00019999911230800713,
-      "loss": 46.0,
-      "step": 17552
-    },
-    {
-      "epoch": 1.3420494294397614,
-      "grad_norm": 0.002199098700657487,
-      "learning_rate": 0.0001999991122067968,
-      "loss": 46.0,
-      "step": 17553
-    },
-    {
-      "epoch": 1.3421258864231511,
-      "grad_norm": 0.0014237448340281844,
-      "learning_rate": 0.00019999911210558064,
-      "loss": 46.0,
-      "step": 17554
-    },
-    {
-      "epoch": 1.3422023434065409,
-      "grad_norm": 0.0008657604339532554,
-      "learning_rate": 0.00019999911200435877,
-      "loss": 46.0,
-      "step": 17555
-    },
-    {
-      "epoch": 1.3422788003899306,
-      "grad_norm": 0.0005057230591773987,
-      "learning_rate": 0.00019999911190313112,
-      "loss": 46.0,
-      "step": 17556
-    },
-    {
-      "epoch": 1.3423552573733204,
-      "grad_norm": 0.005004973150789738,
-      "learning_rate": 0.00019999911180189768,
-      "loss": 46.0,
-      "step": 17557
-    },
-    {
-      "epoch": 1.3424317143567102,
-      "grad_norm": 0.001214786316268146,
-      "learning_rate": 0.00019999911170065848,
-      "loss": 46.0,
-      "step": 17558
-    },
-    {
-      "epoch": 1.3425081713400997,
-      "grad_norm": 0.0015020164428278804,
-      "learning_rate": 0.00019999911159941351,
-      "loss": 46.0,
-      "step": 17559
-    },
-    {
-      "epoch": 1.3425846283234895,
-      "grad_norm": 0.0029497521463781595,
-      "learning_rate": 0.0001999991114981628,
-      "loss": 46.0,
-      "step": 17560
-    },
-    {
-      "epoch": 1.3426610853068792,
-      "grad_norm": 0.01663339138031006,
-      "learning_rate": 0.0001999991113969063,
-      "loss": 46.0,
-      "step": 17561
-    },
-    {
-      "epoch": 1.342737542290269,
-      "grad_norm": 0.0014462413964793086,
-      "learning_rate": 0.000199999111295644,
-      "loss": 46.0,
-      "step": 17562
-    },
-    {
-      "epoch": 1.3428139992736585,
-      "grad_norm": 0.003001823555678129,
-      "learning_rate": 0.00019999911119437597,
-      "loss": 46.0,
-      "step": 17563
-    },
-    {
-      "epoch": 1.3428904562570483,
-      "grad_norm": 0.0007143891416490078,
-      "learning_rate": 0.00019999911109310213,
-      "loss": 46.0,
-      "step": 17564
-    },
-    {
-      "epoch": 1.342966913240438,
-      "grad_norm": 0.0032529369927942753,
-      "learning_rate": 0.00019999911099182255,
-      "loss": 46.0,
-      "step": 17565
-    },
-    {
-      "epoch": 1.3430433702238278,
-      "grad_norm": 0.0010566123528406024,
-      "learning_rate": 0.0001999991108905372,
-      "loss": 46.0,
-      "step": 17566
-    },
-    {
-      "epoch": 1.3431198272072176,
-      "grad_norm": 0.0012466214830055833,
-      "learning_rate": 0.0001999991107892461,
-      "loss": 46.0,
-      "step": 17567
-    },
-    {
-      "epoch": 1.3431962841906073,
-      "grad_norm": 0.000706339196767658,
-      "learning_rate": 0.0001999991106879492,
-      "loss": 46.0,
-      "step": 17568
-    },
-    {
-      "epoch": 1.343272741173997,
-      "grad_norm": 0.0004942166269756854,
-      "learning_rate": 0.00019999911058664652,
-      "loss": 46.0,
-      "step": 17569
-    },
-    {
-      "epoch": 1.3433491981573866,
-      "grad_norm": 0.0004397607408463955,
-      "learning_rate": 0.0001999991104853381,
-      "loss": 46.0,
-      "step": 17570
-    },
-    {
-      "epoch": 1.3434256551407764,
-      "grad_norm": 0.0012350472388789058,
-      "learning_rate": 0.0001999991103840239,
-      "loss": 46.0,
-      "step": 17571
-    },
-    {
-      "epoch": 1.3435021121241661,
-      "grad_norm": 0.0006688169087283313,
-      "learning_rate": 0.00019999911028270394,
-      "loss": 46.0,
-      "step": 17572
-    },
-    {
-      "epoch": 1.343578569107556,
-      "grad_norm": 0.0017817559419199824,
-      "learning_rate": 0.0001999991101813782,
-      "loss": 46.0,
-      "step": 17573
-    },
-    {
-      "epoch": 1.3436550260909454,
-      "grad_norm": 0.001308659091591835,
-      "learning_rate": 0.0001999991100800467,
-      "loss": 46.0,
-      "step": 17574
-    },
-    {
-      "epoch": 1.3437314830743352,
-      "grad_norm": 0.0012899389257654548,
-      "learning_rate": 0.0001999991099787094,
-      "loss": 46.0,
-      "step": 17575
-    },
-    {
-      "epoch": 1.343807940057725,
-      "grad_norm": 0.0010276848915964365,
-      "learning_rate": 0.00019999910987736637,
-      "loss": 46.0,
-      "step": 17576
-    },
-    {
-      "epoch": 1.3438843970411147,
-      "grad_norm": 0.00026489104493521154,
-      "learning_rate": 0.00019999910977601753,
-      "loss": 46.0,
-      "step": 17577
-    },
-    {
-      "epoch": 1.3439608540245045,
-      "grad_norm": 0.0007294208044186234,
-      "learning_rate": 0.00019999910967466295,
-      "loss": 46.0,
-      "step": 17578
-    },
-    {
-      "epoch": 1.3440373110078943,
-      "grad_norm": 0.005262431222945452,
-      "learning_rate": 0.0001999991095733026,
-      "loss": 46.0,
-      "step": 17579
-    },
-    {
-      "epoch": 1.344113767991284,
-      "grad_norm": 0.0023152728099375963,
-      "learning_rate": 0.0001999991094719365,
-      "loss": 46.0,
-      "step": 17580
-    },
-    {
-      "epoch": 1.3441902249746736,
-      "grad_norm": 0.0003386722528375685,
-      "learning_rate": 0.0001999991093705646,
-      "loss": 46.0,
-      "step": 17581
-    },
-    {
-      "epoch": 1.3442666819580633,
-      "grad_norm": 0.002151747699826956,
-      "learning_rate": 0.00019999910926918693,
-      "loss": 46.0,
-      "step": 17582
-    },
-    {
-      "epoch": 1.344343138941453,
-      "grad_norm": 0.0023805673699826,
-      "learning_rate": 0.0001999991091678035,
-      "loss": 46.0,
-      "step": 17583
-    },
-    {
-      "epoch": 1.3444195959248428,
-      "grad_norm": 0.0023206949699670076,
-      "learning_rate": 0.00019999910906641431,
-      "loss": 46.0,
-      "step": 17584
-    },
-    {
-      "epoch": 1.3444960529082324,
-      "grad_norm": 0.0012594126164913177,
-      "learning_rate": 0.00019999910896501932,
-      "loss": 46.0,
-      "step": 17585
-    },
-    {
-      "epoch": 1.3445725098916221,
-      "grad_norm": 0.0030643094796687365,
-      "learning_rate": 0.00019999910886361858,
-      "loss": 46.0,
-      "step": 17586
-    },
-    {
-      "epoch": 1.344648966875012,
-      "grad_norm": 0.0012008552439510822,
-      "learning_rate": 0.0001999991087622121,
-      "loss": 46.0,
-      "step": 17587
-    },
-    {
-      "epoch": 1.3447254238584017,
-      "grad_norm": 0.0010384246706962585,
-      "learning_rate": 0.00019999910866079978,
-      "loss": 46.0,
-      "step": 17588
-    },
-    {
-      "epoch": 1.3448018808417914,
-      "grad_norm": 0.0003308470477350056,
-      "learning_rate": 0.00019999910855938175,
-      "loss": 46.0,
-      "step": 17589
-    },
-    {
-      "epoch": 1.3448783378251812,
-      "grad_norm": 0.0005131775978952646,
-      "learning_rate": 0.00019999910845795794,
-      "loss": 46.0,
-      "step": 17590
-    },
-    {
-      "epoch": 1.344954794808571,
-      "grad_norm": 0.0008817252237349749,
-      "learning_rate": 0.00019999910835652833,
-      "loss": 46.0,
-      "step": 17591
-    },
-    {
-      "epoch": 1.3450312517919605,
-      "grad_norm": 0.0006807656609453261,
-      "learning_rate": 0.00019999910825509298,
-      "loss": 46.0,
-      "step": 17592
-    },
-    {
-      "epoch": 1.3451077087753502,
-      "grad_norm": 0.00342154735699296,
-      "learning_rate": 0.00019999910815365185,
-      "loss": 46.0,
-      "step": 17593
-    },
-    {
-      "epoch": 1.34518416575874,
-      "grad_norm": 0.0003578035393729806,
-      "learning_rate": 0.00019999910805220495,
-      "loss": 46.0,
-      "step": 17594
-    },
-    {
-      "epoch": 1.3452606227421298,
-      "grad_norm": 0.00356718129478395,
-      "learning_rate": 0.0001999991079507523,
-      "loss": 46.0,
-      "step": 17595
-    },
-    {
-      "epoch": 1.3453370797255193,
-      "grad_norm": 0.004658654797822237,
-      "learning_rate": 0.00019999910784929386,
-      "loss": 46.0,
-      "step": 17596
-    },
-    {
-      "epoch": 1.345413536708909,
-      "grad_norm": 0.0007298219134099782,
-      "learning_rate": 0.00019999910774782967,
-      "loss": 46.0,
-      "step": 17597
-    },
-    {
-      "epoch": 1.3454899936922988,
-      "grad_norm": 0.0007636790396645665,
-      "learning_rate": 0.0001999991076463597,
-      "loss": 46.0,
-      "step": 17598
-    },
-    {
-      "epoch": 1.3455664506756886,
-      "grad_norm": 0.0003657451015897095,
-      "learning_rate": 0.00019999910754488394,
-      "loss": 46.0,
-      "step": 17599
-    },
-    {
-      "epoch": 1.3456429076590783,
-      "grad_norm": 0.001634453539736569,
-      "learning_rate": 0.00019999910744340245,
-      "loss": 46.0,
-      "step": 17600
-    },
-    {
-      "epoch": 1.345719364642468,
-      "grad_norm": 0.0015206843381747603,
-      "learning_rate": 0.00019999910734191514,
-      "loss": 46.0,
-      "step": 17601
-    },
-    {
-      "epoch": 1.3457958216258579,
-      "grad_norm": 0.0006551849655807018,
-      "learning_rate": 0.0001999991072404221,
-      "loss": 46.0,
-      "step": 17602
-    },
-    {
-      "epoch": 1.3458722786092474,
-      "grad_norm": 0.0009461393929086626,
-      "learning_rate": 0.0001999991071389233,
-      "loss": 46.0,
-      "step": 17603
-    },
-    {
-      "epoch": 1.3459487355926372,
-      "grad_norm": 0.007163412403315306,
-      "learning_rate": 0.0001999991070374187,
-      "loss": 46.0,
-      "step": 17604
-    },
-    {
-      "epoch": 1.346025192576027,
-      "grad_norm": 0.002033101161941886,
-      "learning_rate": 0.00019999910693590834,
-      "loss": 46.0,
-      "step": 17605
-    },
-    {
-      "epoch": 1.3461016495594167,
-      "grad_norm": 0.001575839938595891,
-      "learning_rate": 0.0001999991068343922,
-      "loss": 46.0,
-      "step": 17606
-    },
-    {
-      "epoch": 1.3461781065428062,
-      "grad_norm": 0.00047944457037374377,
-      "learning_rate": 0.0001999991067328703,
-      "loss": 46.0,
-      "step": 17607
-    },
-    {
-      "epoch": 1.346254563526196,
-      "grad_norm": 0.002485336037352681,
-      "learning_rate": 0.00019999910663134264,
-      "loss": 46.0,
-      "step": 17608
-    },
-    {
-      "epoch": 1.3463310205095858,
-      "grad_norm": 0.002915919991210103,
-      "learning_rate": 0.0001999991065298092,
-      "loss": 46.0,
-      "step": 17609
-    },
-    {
-      "epoch": 1.3464074774929755,
-      "grad_norm": 0.0003705063136294484,
-      "learning_rate": 0.00019999910642826997,
-      "loss": 46.0,
-      "step": 17610
-    },
-    {
-      "epoch": 1.3464839344763653,
-      "grad_norm": 0.0006521191098727286,
-      "learning_rate": 0.00019999910632672503,
-      "loss": 46.0,
-      "step": 17611
-    },
-    {
-      "epoch": 1.346560391459755,
-      "grad_norm": 0.0012793421046808362,
-      "learning_rate": 0.00019999910622517427,
-      "loss": 46.0,
-      "step": 17612
-    },
-    {
-      "epoch": 1.3466368484431448,
-      "grad_norm": 0.000768558937124908,
-      "learning_rate": 0.00019999910612361775,
-      "loss": 46.0,
-      "step": 17613
-    },
-    {
-      "epoch": 1.3467133054265343,
-      "grad_norm": 0.0012492822716012597,
-      "learning_rate": 0.00019999910602205547,
-      "loss": 46.0,
-      "step": 17614
-    },
-    {
-      "epoch": 1.346789762409924,
-      "grad_norm": 0.0017069990281015635,
-      "learning_rate": 0.0001999991059204874,
-      "loss": 46.0,
-      "step": 17615
-    },
-    {
-      "epoch": 1.3468662193933139,
-      "grad_norm": 0.001446478650905192,
-      "learning_rate": 0.0001999991058189136,
-      "loss": 46.0,
-      "step": 17616
-    },
-    {
-      "epoch": 1.3469426763767036,
-      "grad_norm": 0.0005476575461216271,
-      "learning_rate": 0.000199999105717334,
-      "loss": 46.0,
-      "step": 17617
-    },
-    {
-      "epoch": 1.3470191333600932,
-      "grad_norm": 0.027798477560281754,
-      "learning_rate": 0.00019999910561574864,
-      "loss": 46.0,
-      "step": 17618
-    },
-    {
-      "epoch": 1.347095590343483,
-      "grad_norm": 0.0017784852534532547,
-      "learning_rate": 0.0001999991055141575,
-      "loss": 46.0,
-      "step": 17619
-    },
-    {
-      "epoch": 1.3471720473268727,
-      "grad_norm": 0.005920881405472755,
-      "learning_rate": 0.00019999910541256062,
-      "loss": 46.0,
-      "step": 17620
-    },
-    {
-      "epoch": 1.3472485043102624,
-      "grad_norm": 0.005008005537092686,
-      "learning_rate": 0.00019999910531095795,
-      "loss": 46.0,
-      "step": 17621
-    },
-    {
-      "epoch": 1.3473249612936522,
-      "grad_norm": 0.003869800129905343,
-      "learning_rate": 0.0001999991052093495,
-      "loss": 46.0,
-      "step": 17622
-    },
-    {
-      "epoch": 1.347401418277042,
-      "grad_norm": 0.001331445761024952,
-      "learning_rate": 0.0001999991051077353,
-      "loss": 46.0,
-      "step": 17623
-    },
-    {
-      "epoch": 1.3474778752604317,
-      "grad_norm": 0.0034771051723510027,
-      "learning_rate": 0.00019999910500611531,
-      "loss": 46.0,
-      "step": 17624
-    },
-    {
-      "epoch": 1.3475543322438213,
-      "grad_norm": 0.0010703509906306863,
-      "learning_rate": 0.00019999910490448958,
-      "loss": 46.0,
-      "step": 17625
-    },
-    {
-      "epoch": 1.347630789227211,
-      "grad_norm": 0.0052464609034359455,
-      "learning_rate": 0.00019999910480285804,
-      "loss": 46.0,
-      "step": 17626
-    },
-    {
-      "epoch": 1.3477072462106008,
-      "grad_norm": 0.0014775327872484922,
-      "learning_rate": 0.00019999910470122075,
-      "loss": 46.0,
-      "step": 17627
-    },
-    {
-      "epoch": 1.3477837031939905,
-      "grad_norm": 0.004175929818302393,
-      "learning_rate": 0.00019999910459957772,
-      "loss": 46.0,
-      "step": 17628
-    },
-    {
-      "epoch": 1.34786016017738,
-      "grad_norm": 0.0022350838407874107,
-      "learning_rate": 0.00019999910449792889,
-      "loss": 46.0,
-      "step": 17629
-    },
-    {
-      "epoch": 1.3479366171607698,
-      "grad_norm": 0.000799885077867657,
-      "learning_rate": 0.0001999991043962743,
-      "loss": 46.0,
-      "step": 17630
-    },
-    {
-      "epoch": 1.3480130741441596,
-      "grad_norm": 0.0013833839911967516,
-      "learning_rate": 0.00019999910429461393,
-      "loss": 46.0,
-      "step": 17631
-    },
-    {
-      "epoch": 1.3480895311275494,
-      "grad_norm": 0.0018418688559904695,
-      "learning_rate": 0.00019999910419294777,
-      "loss": 46.0,
-      "step": 17632
-    },
-    {
-      "epoch": 1.3481659881109391,
-      "grad_norm": 0.0002597272105049342,
-      "learning_rate": 0.0001999991040912759,
-      "loss": 46.0,
-      "step": 17633
-    },
-    {
-      "epoch": 1.348242445094329,
-      "grad_norm": 0.0009844328742474318,
-      "learning_rate": 0.00019999910398959823,
-      "loss": 46.0,
-      "step": 17634
-    },
-    {
-      "epoch": 1.3483189020777187,
-      "grad_norm": 0.014189861714839935,
-      "learning_rate": 0.00019999910388791478,
-      "loss": 46.0,
-      "step": 17635
-    },
-    {
-      "epoch": 1.3483953590611082,
-      "grad_norm": 0.0007297651027329266,
-      "learning_rate": 0.00019999910378622557,
-      "loss": 46.0,
-      "step": 17636
-    },
-    {
-      "epoch": 1.348471816044498,
-      "grad_norm": 0.004691058304160833,
-      "learning_rate": 0.0001999991036845306,
-      "loss": 46.0,
-      "step": 17637
-    },
-    {
-      "epoch": 1.3485482730278877,
-      "grad_norm": 0.0008430343004874885,
-      "learning_rate": 0.00019999910358282983,
-      "loss": 46.0,
-      "step": 17638
-    },
-    {
-      "epoch": 1.3486247300112775,
-      "grad_norm": 0.0005567933549173176,
-      "learning_rate": 0.00019999910348112332,
-      "loss": 46.0,
-      "step": 17639
-    },
-    {
-      "epoch": 1.348701186994667,
-      "grad_norm": 0.0010431535774841905,
-      "learning_rate": 0.00019999910337941104,
-      "loss": 46.0,
-      "step": 17640
-    },
-    {
-      "epoch": 1.3487776439780568,
-      "grad_norm": 0.0009747219155542552,
-      "learning_rate": 0.00019999910327769298,
-      "loss": 46.0,
-      "step": 17641
-    },
-    {
-      "epoch": 1.3488541009614465,
-      "grad_norm": 0.004696002695709467,
-      "learning_rate": 0.00019999910317596915,
-      "loss": 46.0,
-      "step": 17642
-    },
-    {
-      "epoch": 1.3489305579448363,
-      "grad_norm": 0.0005317625473253429,
-      "learning_rate": 0.00019999910307423954,
-      "loss": 46.0,
-      "step": 17643
-    },
-    {
-      "epoch": 1.349007014928226,
-      "grad_norm": 0.0007092943997122347,
-      "learning_rate": 0.0001999991029725042,
-      "loss": 46.0,
-      "step": 17644
-    },
-    {
-      "epoch": 1.3490834719116158,
-      "grad_norm": 0.002394125098362565,
-      "learning_rate": 0.00019999910287076306,
-      "loss": 46.0,
-      "step": 17645
-    },
-    {
-      "epoch": 1.3491599288950056,
-      "grad_norm": 0.006621365435421467,
-      "learning_rate": 0.00019999910276901614,
-      "loss": 46.0,
-      "step": 17646
-    },
-    {
-      "epoch": 1.3492363858783951,
-      "grad_norm": 0.0010238155955448747,
-      "learning_rate": 0.00019999910266726346,
-      "loss": 46.0,
-      "step": 17647
-    },
-    {
-      "epoch": 1.3493128428617849,
-      "grad_norm": 0.0027714355383068323,
-      "learning_rate": 0.00019999910256550502,
-      "loss": 46.0,
-      "step": 17648
-    },
-    {
-      "epoch": 1.3493892998451746,
-      "grad_norm": 0.006141017656773329,
-      "learning_rate": 0.00019999910246374083,
-      "loss": 46.0,
-      "step": 17649
-    },
-    {
-      "epoch": 1.3494657568285644,
-      "grad_norm": 0.0004310406220611185,
-      "learning_rate": 0.00019999910236197084,
-      "loss": 46.0,
-      "step": 17650
-    },
-    {
-      "epoch": 1.349542213811954,
-      "grad_norm": 0.0005662787007167935,
-      "learning_rate": 0.00019999910226019507,
-      "loss": 46.0,
-      "step": 17651
-    },
-    {
-      "epoch": 1.3496186707953437,
-      "grad_norm": 0.003674947889521718,
-      "learning_rate": 0.00019999910215841356,
-      "loss": 46.0,
-      "step": 17652
-    },
-    {
-      "epoch": 1.3496951277787335,
-      "grad_norm": 0.0012099051382392645,
-      "learning_rate": 0.00019999910205662627,
-      "loss": 46.0,
-      "step": 17653
-    },
-    {
-      "epoch": 1.3497715847621232,
-      "grad_norm": 0.009085611440241337,
-      "learning_rate": 0.00019999910195483321,
-      "loss": 46.0,
-      "step": 17654
-    },
-    {
-      "epoch": 1.349848041745513,
-      "grad_norm": 0.0013135239714756608,
-      "learning_rate": 0.00019999910185303438,
-      "loss": 46.0,
-      "step": 17655
-    },
-    {
-      "epoch": 1.3499244987289027,
-      "grad_norm": 0.0012159928446635604,
-      "learning_rate": 0.0001999991017512298,
-      "loss": 46.0,
-      "step": 17656
-    },
-    {
-      "epoch": 1.3500009557122925,
-      "grad_norm": 0.0005517210811376572,
-      "learning_rate": 0.00019999910164941943,
-      "loss": 46.0,
-      "step": 17657
-    },
-    {
-      "epoch": 1.350077412695682,
-      "grad_norm": 0.002159824827685952,
-      "learning_rate": 0.00019999910154760327,
-      "loss": 46.0,
-      "step": 17658
-    },
-    {
-      "epoch": 1.3501538696790718,
-      "grad_norm": 0.001300903270021081,
-      "learning_rate": 0.00019999910144578138,
-      "loss": 46.0,
-      "step": 17659
-    },
-    {
-      "epoch": 1.3502303266624616,
-      "grad_norm": 0.005021155346184969,
-      "learning_rate": 0.0001999991013439537,
-      "loss": 46.0,
-      "step": 17660
-    },
-    {
-      "epoch": 1.3503067836458513,
-      "grad_norm": 0.002784095937386155,
-      "learning_rate": 0.00019999910124212026,
-      "loss": 46.0,
-      "step": 17661
-    },
-    {
-      "epoch": 1.3503832406292409,
-      "grad_norm": 0.0018474971875548363,
-      "learning_rate": 0.00019999910114028104,
-      "loss": 46.0,
-      "step": 17662
-    },
-    {
-      "epoch": 1.3504596976126306,
-      "grad_norm": 0.009056279435753822,
-      "learning_rate": 0.00019999910103843608,
-      "loss": 46.0,
-      "step": 17663
-    },
-    {
-      "epoch": 1.3505361545960204,
-      "grad_norm": 0.0019982350058853626,
-      "learning_rate": 0.0001999991009365853,
-      "loss": 46.0,
-      "step": 17664
-    },
-    {
-      "epoch": 1.3506126115794101,
-      "grad_norm": 0.0012308531440794468,
-      "learning_rate": 0.00019999910083472877,
-      "loss": 46.0,
-      "step": 17665
-    },
-    {
-      "epoch": 1.3506890685628,
-      "grad_norm": 0.012501278892159462,
-      "learning_rate": 0.00019999910073286649,
-      "loss": 46.0,
-      "step": 17666
-    },
-    {
-      "epoch": 1.3507655255461897,
-      "grad_norm": 0.0017271460965275764,
-      "learning_rate": 0.00019999910063099843,
-      "loss": 46.0,
-      "step": 17667
-    },
-    {
-      "epoch": 1.3508419825295794,
-      "grad_norm": 0.0011376762995496392,
-      "learning_rate": 0.0001999991005291246,
-      "loss": 46.0,
-      "step": 17668
-    },
-    {
-      "epoch": 1.350918439512969,
-      "grad_norm": 0.0017403539968654513,
-      "learning_rate": 0.000199999100427245,
-      "loss": 46.0,
-      "step": 17669
-    },
-    {
-      "epoch": 1.3509948964963587,
-      "grad_norm": 0.0010393804404884577,
-      "learning_rate": 0.0001999991003253596,
-      "loss": 46.0,
-      "step": 17670
-    },
-    {
-      "epoch": 1.3510713534797485,
-      "grad_norm": 0.0010857520392164588,
-      "learning_rate": 0.0001999991002234685,
-      "loss": 46.0,
-      "step": 17671
-    },
-    {
-      "epoch": 1.3511478104631383,
-      "grad_norm": 0.0012325705029070377,
-      "learning_rate": 0.00019999910012157156,
-      "loss": 46.0,
-      "step": 17672
-    },
-    {
-      "epoch": 1.3512242674465278,
-      "grad_norm": 0.0016186099965125322,
-      "learning_rate": 0.00019999910001966892,
-      "loss": 46.0,
-      "step": 17673
-    },
-    {
-      "epoch": 1.3513007244299176,
-      "grad_norm": 0.004259428009390831,
-      "learning_rate": 0.00019999909991776045,
-      "loss": 46.0,
-      "step": 17674
-    },
-    {
-      "epoch": 1.3513771814133073,
-      "grad_norm": 0.0007236300152726471,
-      "learning_rate": 0.0001999990998158462,
-      "loss": 46.0,
-      "step": 17675
-    },
-    {
-      "epoch": 1.351453638396697,
-      "grad_norm": 0.001005738042294979,
-      "learning_rate": 0.00019999909971392624,
-      "loss": 46.0,
-      "step": 17676
-    },
-    {
-      "epoch": 1.3515300953800868,
-      "grad_norm": 0.0008229472441598773,
-      "learning_rate": 0.00019999909961200047,
-      "loss": 46.0,
-      "step": 17677
-    },
-    {
-      "epoch": 1.3516065523634766,
-      "grad_norm": 0.0011688934173434973,
-      "learning_rate": 0.00019999909951006893,
-      "loss": 46.0,
-      "step": 17678
-    },
-    {
-      "epoch": 1.3516830093468661,
-      "grad_norm": 0.00225238804705441,
-      "learning_rate": 0.00019999909940813165,
-      "loss": 46.0,
-      "step": 17679
-    },
-    {
-      "epoch": 1.351759466330256,
-      "grad_norm": 0.0045624286867678165,
-      "learning_rate": 0.00019999909930618862,
-      "loss": 46.0,
-      "step": 17680
-    },
-    {
-      "epoch": 1.3518359233136457,
-      "grad_norm": 0.0020937968511134386,
-      "learning_rate": 0.00019999909920423976,
-      "loss": 46.0,
-      "step": 17681
-    },
-    {
-      "epoch": 1.3519123802970354,
-      "grad_norm": 0.0017957761883735657,
-      "learning_rate": 0.00019999909910228516,
-      "loss": 46.0,
-      "step": 17682
-    },
-    {
-      "epoch": 1.3519888372804252,
-      "grad_norm": 0.008097846060991287,
-      "learning_rate": 0.00019999909900032478,
-      "loss": 46.0,
-      "step": 17683
-    },
-    {
-      "epoch": 1.3520652942638147,
-      "grad_norm": 0.00241546961478889,
-      "learning_rate": 0.00019999909889835865,
-      "loss": 46.0,
-      "step": 17684
-    },
-    {
-      "epoch": 1.3521417512472045,
-      "grad_norm": 0.0011286619119346142,
-      "learning_rate": 0.00019999909879638673,
-      "loss": 46.0,
-      "step": 17685
-    },
-    {
-      "epoch": 1.3522182082305942,
-      "grad_norm": 0.0008477772353217006,
-      "learning_rate": 0.00019999909869440906,
-      "loss": 46.0,
-      "step": 17686
-    },
-    {
-      "epoch": 1.352294665213984,
-      "grad_norm": 0.0014911523321643472,
-      "learning_rate": 0.0001999990985924256,
-      "loss": 46.0,
-      "step": 17687
-    },
-    {
-      "epoch": 1.3523711221973738,
-      "grad_norm": 0.004975670017302036,
-      "learning_rate": 0.0001999990984904364,
-      "loss": 46.0,
-      "step": 17688
-    },
-    {
-      "epoch": 1.3524475791807635,
-      "grad_norm": 0.007092161104083061,
-      "learning_rate": 0.0001999990983884414,
-      "loss": 46.0,
-      "step": 17689
-    },
-    {
-      "epoch": 1.352524036164153,
-      "grad_norm": 0.0006369981565512717,
-      "learning_rate": 0.00019999909828644064,
-      "loss": 46.0,
-      "step": 17690
-    },
-    {
-      "epoch": 1.3526004931475428,
-      "grad_norm": 0.00046385114546865225,
-      "learning_rate": 0.0001999990981844341,
-      "loss": 46.0,
-      "step": 17691
-    },
-    {
-      "epoch": 1.3526769501309326,
-      "grad_norm": 0.004023105837404728,
-      "learning_rate": 0.00019999909808242182,
-      "loss": 46.0,
-      "step": 17692
-    },
-    {
-      "epoch": 1.3527534071143223,
-      "grad_norm": 0.0018335661152377725,
-      "learning_rate": 0.00019999909798040376,
-      "loss": 46.0,
-      "step": 17693
-    },
-    {
-      "epoch": 1.3528298640977119,
-      "grad_norm": 0.0005817157798446715,
-      "learning_rate": 0.00019999909787837993,
-      "loss": 46.0,
-      "step": 17694
-    },
-    {
-      "epoch": 1.3529063210811016,
-      "grad_norm": 0.000753791187889874,
-      "learning_rate": 0.0001999990977763503,
-      "loss": 46.0,
-      "step": 17695
-    },
-    {
-      "epoch": 1.3529827780644914,
-      "grad_norm": 0.0009575884905643761,
-      "learning_rate": 0.00019999909767431492,
-      "loss": 46.0,
-      "step": 17696
-    },
-    {
-      "epoch": 1.3530592350478812,
-      "grad_norm": 0.0030338154174387455,
-      "learning_rate": 0.00019999909757227377,
-      "loss": 46.0,
-      "step": 17697
-    },
-    {
-      "epoch": 1.353135692031271,
-      "grad_norm": 0.0007067607948556542,
-      "learning_rate": 0.00019999909747022687,
-      "loss": 46.0,
-      "step": 17698
-    },
-    {
-      "epoch": 1.3532121490146607,
-      "grad_norm": 0.0006542197079397738,
-      "learning_rate": 0.0001999990973681742,
-      "loss": 46.0,
-      "step": 17699
-    },
-    {
-      "epoch": 1.3532886059980505,
-      "grad_norm": 0.0011140276910737157,
-      "learning_rate": 0.00019999909726611576,
-      "loss": 46.0,
-      "step": 17700
-    },
-    {
-      "epoch": 1.35336506298144,
-      "grad_norm": 0.0007084396784193814,
-      "learning_rate": 0.0001999990971640515,
-      "loss": 46.0,
-      "step": 17701
-    },
-    {
-      "epoch": 1.3534415199648298,
-      "grad_norm": 0.001138114370405674,
-      "learning_rate": 0.00019999909706198152,
-      "loss": 46.0,
-      "step": 17702
-    },
-    {
-      "epoch": 1.3535179769482195,
-      "grad_norm": 0.001708924537524581,
-      "learning_rate": 0.00019999909695990576,
-      "loss": 46.0,
-      "step": 17703
-    },
-    {
-      "epoch": 1.3535944339316093,
-      "grad_norm": 0.0005078091635368764,
-      "learning_rate": 0.00019999909685782422,
-      "loss": 46.0,
-      "step": 17704
-    },
-    {
-      "epoch": 1.3536708909149988,
-      "grad_norm": 0.0020662799943238497,
-      "learning_rate": 0.00019999909675573693,
-      "loss": 46.0,
-      "step": 17705
-    },
-    {
-      "epoch": 1.3537473478983886,
-      "grad_norm": 0.00034990065614692867,
-      "learning_rate": 0.00019999909665364388,
-      "loss": 46.0,
-      "step": 17706
-    },
-    {
-      "epoch": 1.3538238048817783,
-      "grad_norm": 0.0024372527841478586,
-      "learning_rate": 0.00019999909655154502,
-      "loss": 46.0,
-      "step": 17707
-    },
-    {
-      "epoch": 1.353900261865168,
-      "grad_norm": 0.0006608974654227495,
-      "learning_rate": 0.00019999909644944044,
-      "loss": 46.0,
-      "step": 17708
-    },
-    {
-      "epoch": 1.3539767188485579,
-      "grad_norm": 0.0021569731179624796,
-      "learning_rate": 0.00019999909634733004,
-      "loss": 46.0,
-      "step": 17709
-    },
-    {
-      "epoch": 1.3540531758319476,
-      "grad_norm": 0.0010395536664873362,
-      "learning_rate": 0.0001999990962452139,
-      "loss": 46.0,
-      "step": 17710
-    },
-    {
-      "epoch": 1.3541296328153374,
-      "grad_norm": 0.0011058994568884373,
-      "learning_rate": 0.000199999096143092,
-      "loss": 46.0,
-      "step": 17711
-    },
-    {
-      "epoch": 1.354206089798727,
-      "grad_norm": 0.0020159909036010504,
-      "learning_rate": 0.0001999990960409643,
-      "loss": 46.0,
-      "step": 17712
-    },
-    {
-      "epoch": 1.3542825467821167,
-      "grad_norm": 0.005407491698861122,
-      "learning_rate": 0.00019999909593883085,
-      "loss": 46.0,
-      "step": 17713
-    },
-    {
-      "epoch": 1.3543590037655064,
-      "grad_norm": 0.00033945226459763944,
-      "learning_rate": 0.0001999990958366916,
-      "loss": 46.0,
-      "step": 17714
-    },
-    {
-      "epoch": 1.3544354607488962,
-      "grad_norm": 0.0035249083302915096,
-      "learning_rate": 0.0001999990957345466,
-      "loss": 46.0,
-      "step": 17715
-    },
-    {
-      "epoch": 1.3545119177322857,
-      "grad_norm": 0.0014321065973490477,
-      "learning_rate": 0.00019999909563239588,
-      "loss": 46.0,
-      "step": 17716
-    },
-    {
-      "epoch": 1.3545883747156755,
-      "grad_norm": 0.0012607151875272393,
-      "learning_rate": 0.0001999990955302393,
-      "loss": 46.0,
-      "step": 17717
-    },
-    {
-      "epoch": 1.3546648316990653,
-      "grad_norm": 0.0012724926928058267,
-      "learning_rate": 0.00019999909542807706,
-      "loss": 46.0,
-      "step": 17718
-    },
-    {
-      "epoch": 1.354741288682455,
-      "grad_norm": 0.0028961985372006893,
-      "learning_rate": 0.00019999909532590894,
-      "loss": 46.0,
-      "step": 17719
-    },
-    {
-      "epoch": 1.3548177456658448,
-      "grad_norm": 0.007116749417036772,
-      "learning_rate": 0.00019999909522373511,
-      "loss": 46.0,
-      "step": 17720
-    },
-    {
-      "epoch": 1.3548942026492345,
-      "grad_norm": 0.0005645353230647743,
-      "learning_rate": 0.0001999990951215555,
-      "loss": 46.0,
-      "step": 17721
-    },
-    {
-      "epoch": 1.3549706596326243,
-      "grad_norm": 0.0023879176005721092,
-      "learning_rate": 0.00019999909501937013,
-      "loss": 46.0,
-      "step": 17722
-    },
-    {
-      "epoch": 1.3550471166160138,
-      "grad_norm": 0.0007873786962591112,
-      "learning_rate": 0.00019999909491717896,
-      "loss": 46.0,
-      "step": 17723
-    },
-    {
-      "epoch": 1.3551235735994036,
-      "grad_norm": 0.0006270324229262769,
-      "learning_rate": 0.00019999909481498206,
-      "loss": 46.0,
-      "step": 17724
-    },
-    {
-      "epoch": 1.3552000305827934,
-      "grad_norm": 0.001199326361529529,
-      "learning_rate": 0.0001999990947127794,
-      "loss": 46.0,
-      "step": 17725
-    },
-    {
-      "epoch": 1.3552764875661831,
-      "grad_norm": 0.0004179398238193244,
-      "learning_rate": 0.00019999909461057092,
-      "loss": 46.0,
-      "step": 17726
-    },
-    {
-      "epoch": 1.3553529445495727,
-      "grad_norm": 0.0007768411887809634,
-      "learning_rate": 0.0001999990945083567,
-      "loss": 46.0,
-      "step": 17727
-    },
-    {
-      "epoch": 1.3554294015329624,
-      "grad_norm": 0.0009438990382477641,
-      "learning_rate": 0.00019999909440613668,
-      "loss": 46.0,
-      "step": 17728
-    },
-    {
-      "epoch": 1.3555058585163522,
-      "grad_norm": 0.0007272938382811844,
-      "learning_rate": 0.00019999909430391092,
-      "loss": 46.0,
-      "step": 17729
-    },
-    {
-      "epoch": 1.355582315499742,
-      "grad_norm": 0.0012425988679751754,
-      "learning_rate": 0.00019999909420167938,
-      "loss": 46.0,
-      "step": 17730
-    },
-    {
-      "epoch": 1.3556587724831317,
-      "grad_norm": 0.001140510430559516,
-      "learning_rate": 0.0001999990940994421,
-      "loss": 46.0,
-      "step": 17731
-    },
-    {
-      "epoch": 1.3557352294665215,
-      "grad_norm": 0.0007697140681557357,
-      "learning_rate": 0.00019999909399719902,
-      "loss": 46.0,
-      "step": 17732
-    },
-    {
-      "epoch": 1.3558116864499112,
-      "grad_norm": 0.0032814880833029747,
-      "learning_rate": 0.00019999909389495019,
-      "loss": 46.0,
-      "step": 17733
-    },
-    {
-      "epoch": 1.3558881434333008,
-      "grad_norm": 0.0018105126218870282,
-      "learning_rate": 0.00019999909379269558,
-      "loss": 46.0,
-      "step": 17734
-    },
-    {
-      "epoch": 1.3559646004166905,
-      "grad_norm": 0.0013273388613015413,
-      "learning_rate": 0.00019999909369043518,
-      "loss": 46.0,
-      "step": 17735
-    },
-    {
-      "epoch": 1.3560410574000803,
-      "grad_norm": 0.006241875700652599,
-      "learning_rate": 0.00019999909358816903,
-      "loss": 46.0,
-      "step": 17736
-    },
-    {
-      "epoch": 1.35611751438347,
-      "grad_norm": 0.00621512345969677,
-      "learning_rate": 0.0001999990934858971,
-      "loss": 46.0,
-      "step": 17737
-    },
-    {
-      "epoch": 1.3561939713668596,
-      "grad_norm": 0.0018622925272211432,
-      "learning_rate": 0.0001999990933836194,
-      "loss": 46.0,
-      "step": 17738
-    },
-    {
-      "epoch": 1.3562704283502494,
-      "grad_norm": 0.0009286011918447912,
-      "learning_rate": 0.00019999909328133594,
-      "loss": 46.0,
-      "step": 17739
-    },
-    {
-      "epoch": 1.3563468853336391,
-      "grad_norm": 0.0007228510221466422,
-      "learning_rate": 0.00019999909317904675,
-      "loss": 46.0,
-      "step": 17740
-    },
-    {
-      "epoch": 1.3564233423170289,
-      "grad_norm": 0.0010326404590159655,
-      "learning_rate": 0.00019999909307675173,
-      "loss": 46.0,
-      "step": 17741
-    },
-    {
-      "epoch": 1.3564997993004186,
-      "grad_norm": 0.0007617344381287694,
-      "learning_rate": 0.000199999092974451,
-      "loss": 46.0,
-      "step": 17742
-    },
-    {
-      "epoch": 1.3565762562838084,
-      "grad_norm": 0.0007650285842828453,
-      "learning_rate": 0.00019999909287214443,
-      "loss": 46.0,
-      "step": 17743
-    },
-    {
-      "epoch": 1.3566527132671982,
-      "grad_norm": 0.0009211569558829069,
-      "learning_rate": 0.00019999909276983212,
-      "loss": 46.0,
-      "step": 17744
-    },
-    {
-      "epoch": 1.3567291702505877,
-      "grad_norm": 0.005173215176910162,
-      "learning_rate": 0.00019999909266751406,
-      "loss": 46.0,
-      "step": 17745
-    },
-    {
-      "epoch": 1.3568056272339775,
-      "grad_norm": 0.00498534832149744,
-      "learning_rate": 0.0001999990925651902,
-      "loss": 46.0,
-      "step": 17746
-    },
-    {
-      "epoch": 1.3568820842173672,
-      "grad_norm": 0.0003701482492033392,
-      "learning_rate": 0.0001999990924628606,
-      "loss": 46.0,
-      "step": 17747
-    },
-    {
-      "epoch": 1.356958541200757,
-      "grad_norm": 0.000933669856749475,
-      "learning_rate": 0.0001999990923605252,
-      "loss": 46.0,
-      "step": 17748
-    },
-    {
-      "epoch": 1.3570349981841465,
-      "grad_norm": 0.0011430637678131461,
-      "learning_rate": 0.00019999909225818405,
-      "loss": 46.0,
-      "step": 17749
-    },
-    {
-      "epoch": 1.3571114551675363,
-      "grad_norm": 0.002336802426725626,
-      "learning_rate": 0.00019999909215583715,
-      "loss": 46.0,
-      "step": 17750
-    },
-    {
-      "epoch": 1.357187912150926,
-      "grad_norm": 0.0013836455764248967,
-      "learning_rate": 0.00019999909205348443,
-      "loss": 46.0,
-      "step": 17751
-    },
-    {
-      "epoch": 1.3572643691343158,
-      "grad_norm": 0.006308353040367365,
-      "learning_rate": 0.000199999091951126,
-      "loss": 46.0,
-      "step": 17752
-    },
-    {
-      "epoch": 1.3573408261177056,
-      "grad_norm": 0.003316121641546488,
-      "learning_rate": 0.00019999909184876174,
-      "loss": 46.0,
-      "step": 17753
-    },
-    {
-      "epoch": 1.3574172831010953,
-      "grad_norm": 0.0011328560067340732,
-      "learning_rate": 0.00019999909174639175,
-      "loss": 46.0,
-      "step": 17754
-    },
-    {
-      "epoch": 1.357493740084485,
-      "grad_norm": 0.00046925709466449916,
-      "learning_rate": 0.000199999091644016,
-      "loss": 46.0,
-      "step": 17755
-    },
-    {
-      "epoch": 1.3575701970678746,
-      "grad_norm": 0.003665717551484704,
-      "learning_rate": 0.00019999909154163446,
-      "loss": 46.0,
-      "step": 17756
-    },
-    {
-      "epoch": 1.3576466540512644,
-      "grad_norm": 0.0008024421986192465,
-      "learning_rate": 0.00019999909143924715,
-      "loss": 46.0,
-      "step": 17757
-    },
-    {
-      "epoch": 1.3577231110346542,
-      "grad_norm": 0.000460057461168617,
-      "learning_rate": 0.00019999909133685406,
-      "loss": 46.0,
-      "step": 17758
-    },
-    {
-      "epoch": 1.357799568018044,
-      "grad_norm": 0.0011262510670349002,
-      "learning_rate": 0.0001999990912344552,
-      "loss": 46.0,
-      "step": 17759
-    },
-    {
-      "epoch": 1.3578760250014335,
-      "grad_norm": 0.00027881463756784797,
-      "learning_rate": 0.0001999990911320506,
-      "loss": 46.0,
-      "step": 17760
-    },
-    {
-      "epoch": 1.3579524819848232,
-      "grad_norm": 0.0037160865031182766,
-      "learning_rate": 0.0001999990910296402,
-      "loss": 46.0,
-      "step": 17761
-    },
-    {
-      "epoch": 1.358028938968213,
-      "grad_norm": 0.002092040376737714,
-      "learning_rate": 0.00019999909092722408,
-      "loss": 46.0,
-      "step": 17762
-    },
-    {
-      "epoch": 1.3581053959516027,
-      "grad_norm": 0.00047612396883778274,
-      "learning_rate": 0.00019999909082480213,
-      "loss": 46.0,
-      "step": 17763
-    },
-    {
-      "epoch": 1.3581818529349925,
-      "grad_norm": 0.010135921649634838,
-      "learning_rate": 0.00019999909072237443,
-      "loss": 46.0,
-      "step": 17764
-    },
-    {
-      "epoch": 1.3582583099183823,
-      "grad_norm": 0.0006043854518793523,
-      "learning_rate": 0.000199999090619941,
-      "loss": 46.0,
-      "step": 17765
-    },
-    {
-      "epoch": 1.358334766901772,
-      "grad_norm": 0.0006798132671974599,
-      "learning_rate": 0.00019999909051750175,
-      "loss": 46.0,
-      "step": 17766
-    },
-    {
-      "epoch": 1.3584112238851616,
-      "grad_norm": 0.0009309955057688057,
-      "learning_rate": 0.00019999909041505673,
-      "loss": 46.0,
-      "step": 17767
-    },
-    {
-      "epoch": 1.3584876808685513,
-      "grad_norm": 0.001363671850413084,
-      "learning_rate": 0.00019999909031260597,
-      "loss": 46.0,
-      "step": 17768
-    },
-    {
-      "epoch": 1.358564137851941,
-      "grad_norm": 0.0024964939802885056,
-      "learning_rate": 0.00019999909021014943,
-      "loss": 46.0,
-      "step": 17769
-    },
-    {
-      "epoch": 1.3586405948353308,
-      "grad_norm": 0.002011209260672331,
-      "learning_rate": 0.00019999909010768712,
-      "loss": 46.0,
-      "step": 17770
-    },
-    {
-      "epoch": 1.3587170518187204,
-      "grad_norm": 0.012724400497972965,
-      "learning_rate": 0.00019999909000521904,
-      "loss": 46.0,
-      "step": 17771
-    },
-    {
-      "epoch": 1.3587935088021101,
-      "grad_norm": 0.002366705797612667,
-      "learning_rate": 0.00019999908990274518,
-      "loss": 46.0,
-      "step": 17772
-    },
-    {
-      "epoch": 1.3588699657855,
-      "grad_norm": 0.001034746179357171,
-      "learning_rate": 0.00019999908980026555,
-      "loss": 46.0,
-      "step": 17773
-    },
-    {
-      "epoch": 1.3589464227688897,
-      "grad_norm": 0.0011732979910448194,
-      "learning_rate": 0.00019999908969778018,
-      "loss": 46.0,
-      "step": 17774
-    },
-    {
-      "epoch": 1.3590228797522794,
-      "grad_norm": 0.0012751020258292556,
-      "learning_rate": 0.000199999089595289,
-      "loss": 46.0,
-      "step": 17775
-    },
-    {
-      "epoch": 1.3590993367356692,
-      "grad_norm": 0.0007026111707091331,
-      "learning_rate": 0.0001999990894927921,
-      "loss": 46.0,
-      "step": 17776
-    },
-    {
-      "epoch": 1.359175793719059,
-      "grad_norm": 0.0005206980276852846,
-      "learning_rate": 0.0001999990893902894,
-      "loss": 46.0,
-      "step": 17777
-    },
-    {
-      "epoch": 1.3592522507024485,
-      "grad_norm": 0.0006734158378094435,
-      "learning_rate": 0.00019999908928778094,
-      "loss": 46.0,
-      "step": 17778
-    },
-    {
-      "epoch": 1.3593287076858382,
-      "grad_norm": 0.0009815518278628588,
-      "learning_rate": 0.0001999990891852667,
-      "loss": 46.0,
-      "step": 17779
-    },
-    {
-      "epoch": 1.359405164669228,
-      "grad_norm": 0.0008330151904374361,
-      "learning_rate": 0.0001999990890827467,
-      "loss": 46.0,
-      "step": 17780
-    },
-    {
-      "epoch": 1.3594816216526178,
-      "grad_norm": 0.0009843712905421853,
-      "learning_rate": 0.00019999908898022092,
-      "loss": 46.0,
-      "step": 17781
-    },
-    {
-      "epoch": 1.3595580786360073,
-      "grad_norm": 0.0014770576963201165,
-      "learning_rate": 0.00019999908887768935,
-      "loss": 46.0,
-      "step": 17782
-    },
-    {
-      "epoch": 1.359634535619397,
-      "grad_norm": 0.000717276765499264,
-      "learning_rate": 0.00019999908877515205,
-      "loss": 46.0,
-      "step": 17783
-    },
-    {
-      "epoch": 1.3597109926027868,
-      "grad_norm": 0.001629303558729589,
-      "learning_rate": 0.000199999088672609,
-      "loss": 46.0,
-      "step": 17784
-    },
-    {
-      "epoch": 1.3597874495861766,
-      "grad_norm": 0.0011074998183175921,
-      "learning_rate": 0.00019999908857006013,
-      "loss": 46.0,
-      "step": 17785
-    },
-    {
-      "epoch": 1.3598639065695663,
-      "grad_norm": 0.0013326950138434768,
-      "learning_rate": 0.0001999990884675055,
-      "loss": 46.0,
-      "step": 17786
-    },
-    {
-      "epoch": 1.359940363552956,
-      "grad_norm": 0.0006041023880243301,
-      "learning_rate": 0.00019999908836494513,
-      "loss": 46.0,
-      "step": 17787
-    },
-    {
-      "epoch": 1.3600168205363459,
-      "grad_norm": 0.00262673432007432,
-      "learning_rate": 0.00019999908826237895,
-      "loss": 46.0,
-      "step": 17788
-    },
-    {
-      "epoch": 1.3600932775197354,
-      "grad_norm": 0.0012383033754304051,
-      "learning_rate": 0.000199999088159807,
-      "loss": 46.0,
-      "step": 17789
-    },
-    {
-      "epoch": 1.3601697345031252,
-      "grad_norm": 0.0019016830483451486,
-      "learning_rate": 0.00019999908805722934,
-      "loss": 46.0,
-      "step": 17790
-    },
-    {
-      "epoch": 1.360246191486515,
-      "grad_norm": 0.0008480129181407392,
-      "learning_rate": 0.00019999908795464587,
-      "loss": 46.0,
-      "step": 17791
-    },
-    {
-      "epoch": 1.3603226484699047,
-      "grad_norm": 0.0023631234653294086,
-      "learning_rate": 0.00019999908785205662,
-      "loss": 46.0,
-      "step": 17792
-    },
-    {
-      "epoch": 1.3603991054532942,
-      "grad_norm": 0.0026122103445231915,
-      "learning_rate": 0.0001999990877494616,
-      "loss": 46.0,
-      "step": 17793
-    },
-    {
-      "epoch": 1.360475562436684,
-      "grad_norm": 0.0030562952160835266,
-      "learning_rate": 0.00019999908764686085,
-      "loss": 46.0,
-      "step": 17794
-    },
-    {
-      "epoch": 1.3605520194200738,
-      "grad_norm": 0.0009380441042594612,
-      "learning_rate": 0.00019999908754425428,
-      "loss": 46.0,
-      "step": 17795
-    },
-    {
-      "epoch": 1.3606284764034635,
-      "grad_norm": 0.00028944300720468163,
-      "learning_rate": 0.00019999908744164197,
-      "loss": 46.0,
-      "step": 17796
-    },
-    {
-      "epoch": 1.3607049333868533,
-      "grad_norm": 0.00037508911918848753,
-      "learning_rate": 0.0001999990873390239,
-      "loss": 46.0,
-      "step": 17797
-    },
-    {
-      "epoch": 1.360781390370243,
-      "grad_norm": 0.002429352840408683,
-      "learning_rate": 0.00019999908723640004,
-      "loss": 46.0,
-      "step": 17798
-    },
-    {
-      "epoch": 1.3608578473536328,
-      "grad_norm": 0.001250554691068828,
-      "learning_rate": 0.00019999908713377043,
-      "loss": 46.0,
-      "step": 17799
-    },
-    {
-      "epoch": 1.3609343043370223,
-      "grad_norm": 0.0020091719925403595,
-      "learning_rate": 0.000199999087031135,
-      "loss": 46.0,
-      "step": 17800
-    },
-    {
-      "epoch": 1.361010761320412,
-      "grad_norm": 0.0009023642633110285,
-      "learning_rate": 0.00019999908692849388,
-      "loss": 46.0,
-      "step": 17801
-    },
-    {
-      "epoch": 1.3610872183038019,
-      "grad_norm": 0.0016051430720835924,
-      "learning_rate": 0.00019999908682584693,
-      "loss": 46.0,
-      "step": 17802
-    },
-    {
-      "epoch": 1.3611636752871916,
-      "grad_norm": 0.0009749067248776555,
-      "learning_rate": 0.00019999908672319424,
-      "loss": 46.0,
-      "step": 17803
-    },
-    {
-      "epoch": 1.3612401322705812,
-      "grad_norm": 0.0029462212696671486,
-      "learning_rate": 0.00019999908662053577,
-      "loss": 46.0,
-      "step": 17804
-    },
-    {
-      "epoch": 1.361316589253971,
-      "grad_norm": 0.0022877671290189028,
-      "learning_rate": 0.00019999908651787153,
-      "loss": 46.0,
-      "step": 17805
-    },
-    {
-      "epoch": 1.3613930462373607,
-      "grad_norm": 0.012493717484176159,
-      "learning_rate": 0.0001999990864152015,
-      "loss": 46.0,
-      "step": 17806
-    },
-    {
-      "epoch": 1.3614695032207504,
-      "grad_norm": 0.0013477298198267817,
-      "learning_rate": 0.00019999908631252572,
-      "loss": 46.0,
-      "step": 17807
-    },
-    {
-      "epoch": 1.3615459602041402,
-      "grad_norm": 0.0006354089127853513,
-      "learning_rate": 0.0001999990862098442,
-      "loss": 46.0,
-      "step": 17808
-    },
-    {
-      "epoch": 1.36162241718753,
-      "grad_norm": 0.0030554740224033594,
-      "learning_rate": 0.00019999908610715688,
-      "loss": 46.0,
-      "step": 17809
-    },
-    {
-      "epoch": 1.3616988741709195,
-      "grad_norm": 0.001531387330032885,
-      "learning_rate": 0.00019999908600446377,
-      "loss": 46.0,
-      "step": 17810
-    },
-    {
-      "epoch": 1.3617753311543093,
-      "grad_norm": 0.001600883319042623,
-      "learning_rate": 0.00019999908590176492,
-      "loss": 46.0,
-      "step": 17811
-    },
-    {
-      "epoch": 1.361851788137699,
-      "grad_norm": 0.0020242915488779545,
-      "learning_rate": 0.00019999908579906031,
-      "loss": 46.0,
-      "step": 17812
-    },
-    {
-      "epoch": 1.3619282451210888,
-      "grad_norm": 0.0034273462370038033,
-      "learning_rate": 0.0001999990856963499,
-      "loss": 46.0,
-      "step": 17813
-    },
-    {
-      "epoch": 1.3620047021044785,
-      "grad_norm": 0.0010653403587639332,
-      "learning_rate": 0.00019999908559363374,
-      "loss": 46.0,
-      "step": 17814
-    },
-    {
-      "epoch": 1.362081159087868,
-      "grad_norm": 0.0006787891616113484,
-      "learning_rate": 0.00019999908549091182,
-      "loss": 46.0,
-      "step": 17815
-    },
-    {
-      "epoch": 1.3621576160712578,
-      "grad_norm": 0.0017184378812089562,
-      "learning_rate": 0.0001999990853881841,
-      "loss": 46.0,
-      "step": 17816
-    },
-    {
-      "epoch": 1.3622340730546476,
-      "grad_norm": 0.002074718941003084,
-      "learning_rate": 0.00019999908528545062,
-      "loss": 46.0,
-      "step": 17817
-    },
-    {
-      "epoch": 1.3623105300380374,
-      "grad_norm": 0.004496516659855843,
-      "learning_rate": 0.00019999908518271138,
-      "loss": 46.0,
-      "step": 17818
-    },
-    {
-      "epoch": 1.3623869870214271,
-      "grad_norm": 0.0006063260952942073,
-      "learning_rate": 0.0001999990850799664,
-      "loss": 46.0,
-      "step": 17819
-    },
-    {
-      "epoch": 1.362463444004817,
-      "grad_norm": 0.0015051227528601885,
-      "learning_rate": 0.0001999990849772156,
-      "loss": 46.0,
-      "step": 17820
-    },
-    {
-      "epoch": 1.3625399009882064,
-      "grad_norm": 0.00028397294227033854,
-      "learning_rate": 0.00019999908487445907,
-      "loss": 46.0,
-      "step": 17821
-    },
-    {
-      "epoch": 1.3626163579715962,
-      "grad_norm": 0.000749338127207011,
-      "learning_rate": 0.00019999908477169674,
-      "loss": 46.0,
-      "step": 17822
-    },
-    {
-      "epoch": 1.362692814954986,
-      "grad_norm": 0.0006876124534755945,
-      "learning_rate": 0.00019999908466892863,
-      "loss": 46.0,
-      "step": 17823
-    },
-    {
-      "epoch": 1.3627692719383757,
-      "grad_norm": 0.0008136221440508962,
-      "learning_rate": 0.0001999990845661548,
-      "loss": 46.0,
-      "step": 17824
-    },
-    {
-      "epoch": 1.3628457289217653,
-      "grad_norm": 0.0007484632078558207,
-      "learning_rate": 0.00019999908446337514,
-      "loss": 46.0,
-      "step": 17825
-    },
-    {
-      "epoch": 1.362922185905155,
-      "grad_norm": 0.0013365275226533413,
-      "learning_rate": 0.00019999908436058977,
-      "loss": 46.0,
-      "step": 17826
-    },
-    {
-      "epoch": 1.3629986428885448,
-      "grad_norm": 0.006817058194428682,
-      "learning_rate": 0.0001999990842577986,
-      "loss": 46.0,
-      "step": 17827
-    },
-    {
-      "epoch": 1.3630750998719345,
-      "grad_norm": 0.000941616075579077,
-      "learning_rate": 0.00019999908415500165,
-      "loss": 46.0,
-      "step": 17828
-    },
-    {
-      "epoch": 1.3631515568553243,
-      "grad_norm": 0.001861086580902338,
-      "learning_rate": 0.00019999908405219892,
-      "loss": 46.0,
-      "step": 17829
-    },
-    {
-      "epoch": 1.363228013838714,
-      "grad_norm": 0.002122849691659212,
-      "learning_rate": 0.00019999908394939046,
-      "loss": 46.0,
-      "step": 17830
-    },
-    {
-      "epoch": 1.3633044708221038,
-      "grad_norm": 0.002680432517081499,
-      "learning_rate": 0.00019999908384657621,
-      "loss": 46.0,
-      "step": 17831
-    },
-    {
-      "epoch": 1.3633809278054934,
-      "grad_norm": 0.0007977135828696191,
-      "learning_rate": 0.00019999908374375623,
-      "loss": 46.0,
-      "step": 17832
-    },
-    {
-      "epoch": 1.3634573847888831,
-      "grad_norm": 0.0005986588657833636,
-      "learning_rate": 0.00019999908364093044,
-      "loss": 46.0,
-      "step": 17833
-    },
-    {
-      "epoch": 1.3635338417722729,
-      "grad_norm": 0.0006033839308656752,
-      "learning_rate": 0.00019999908353809888,
-      "loss": 46.0,
-      "step": 17834
-    },
-    {
-      "epoch": 1.3636102987556626,
-      "grad_norm": 0.0010023170616477728,
-      "learning_rate": 0.00019999908343526157,
-      "loss": 46.0,
-      "step": 17835
-    },
-    {
-      "epoch": 1.3636867557390522,
-      "grad_norm": 0.000915006035938859,
-      "learning_rate": 0.0001999990833324185,
-      "loss": 46.0,
-      "step": 17836
-    },
-    {
-      "epoch": 1.363763212722442,
-      "grad_norm": 0.0014478982193395495,
-      "learning_rate": 0.0001999990832295696,
-      "loss": 46.0,
-      "step": 17837
-    },
-    {
-      "epoch": 1.3638396697058317,
-      "grad_norm": 0.0011957855895161629,
-      "learning_rate": 0.00019999908312671498,
-      "loss": 46.0,
-      "step": 17838
-    },
-    {
-      "epoch": 1.3639161266892215,
-      "grad_norm": 0.0023915900383144617,
-      "learning_rate": 0.0001999990830238546,
-      "loss": 46.0,
-      "step": 17839
-    },
-    {
-      "epoch": 1.3639925836726112,
-      "grad_norm": 0.0036710863932967186,
-      "learning_rate": 0.0001999990829209884,
-      "loss": 46.0,
-      "step": 17840
-    },
-    {
-      "epoch": 1.364069040656001,
-      "grad_norm": 0.002569563454017043,
-      "learning_rate": 0.00019999908281811648,
-      "loss": 46.0,
-      "step": 17841
-    },
-    {
-      "epoch": 1.3641454976393907,
-      "grad_norm": 0.0015425558667629957,
-      "learning_rate": 0.00019999908271523876,
-      "loss": 46.0,
-      "step": 17842
-    },
-    {
-      "epoch": 1.3642219546227803,
-      "grad_norm": 0.0015317416982725263,
-      "learning_rate": 0.0001999990826123553,
-      "loss": 46.0,
-      "step": 17843
-    },
-    {
-      "epoch": 1.36429841160617,
-      "grad_norm": 0.0010371280368417501,
-      "learning_rate": 0.00019999908250946605,
-      "loss": 46.0,
-      "step": 17844
-    },
-    {
-      "epoch": 1.3643748685895598,
-      "grad_norm": 0.0010485831880941987,
-      "learning_rate": 0.00019999908240657104,
-      "loss": 46.0,
-      "step": 17845
-    },
-    {
-      "epoch": 1.3644513255729496,
-      "grad_norm": 0.006667489185929298,
-      "learning_rate": 0.00019999908230367025,
-      "loss": 46.0,
-      "step": 17846
-    },
-    {
-      "epoch": 1.364527782556339,
-      "grad_norm": 0.003453640267252922,
-      "learning_rate": 0.00019999908220076369,
-      "loss": 46.0,
-      "step": 17847
-    },
-    {
-      "epoch": 1.3646042395397289,
-      "grad_norm": 0.0016167510766535997,
-      "learning_rate": 0.00019999908209785135,
-      "loss": 46.0,
-      "step": 17848
-    },
-    {
-      "epoch": 1.3646806965231186,
-      "grad_norm": 0.0010021318448707461,
-      "learning_rate": 0.0001999990819949333,
-      "loss": 46.0,
-      "step": 17849
-    },
-    {
-      "epoch": 1.3647571535065084,
-      "grad_norm": 0.0006228694110177457,
-      "learning_rate": 0.0001999990818920094,
-      "loss": 46.0,
-      "step": 17850
-    },
-    {
-      "epoch": 1.3648336104898982,
-      "grad_norm": 0.0014177424600347877,
-      "learning_rate": 0.0001999990817890798,
-      "loss": 46.0,
-      "step": 17851
-    },
-    {
-      "epoch": 1.364910067473288,
-      "grad_norm": 0.00393153028562665,
-      "learning_rate": 0.0001999990816861444,
-      "loss": 46.0,
-      "step": 17852
-    },
-    {
-      "epoch": 1.3649865244566777,
-      "grad_norm": 0.0010082251392304897,
-      "learning_rate": 0.0001999990815832032,
-      "loss": 46.0,
-      "step": 17853
-    },
-    {
-      "epoch": 1.3650629814400672,
-      "grad_norm": 0.0018287351122125983,
-      "learning_rate": 0.00019999908148025627,
-      "loss": 46.0,
-      "step": 17854
-    },
-    {
-      "epoch": 1.365139438423457,
-      "grad_norm": 0.004699264653027058,
-      "learning_rate": 0.00019999908137730357,
-      "loss": 46.0,
-      "step": 17855
-    },
-    {
-      "epoch": 1.3652158954068467,
-      "grad_norm": 0.014791145920753479,
-      "learning_rate": 0.00019999908127434508,
-      "loss": 46.0,
-      "step": 17856
-    },
-    {
-      "epoch": 1.3652923523902365,
-      "grad_norm": 0.0007130418671295047,
-      "learning_rate": 0.0001999990811713808,
-      "loss": 46.0,
-      "step": 17857
-    },
-    {
-      "epoch": 1.365368809373626,
-      "grad_norm": 0.0011027216678485274,
-      "learning_rate": 0.00019999908106841082,
-      "loss": 46.0,
-      "step": 17858
-    },
-    {
-      "epoch": 1.3654452663570158,
-      "grad_norm": 0.0023836635518819094,
-      "learning_rate": 0.00019999908096543503,
-      "loss": 46.0,
-      "step": 17859
-    },
-    {
-      "epoch": 1.3655217233404056,
-      "grad_norm": 0.0022195016499608755,
-      "learning_rate": 0.00019999908086245347,
-      "loss": 46.0,
-      "step": 17860
-    },
-    {
-      "epoch": 1.3655981803237953,
-      "grad_norm": 0.0012998756719753146,
-      "learning_rate": 0.00019999908075946614,
-      "loss": 46.0,
-      "step": 17861
-    },
-    {
-      "epoch": 1.365674637307185,
-      "grad_norm": 0.0008149457862600684,
-      "learning_rate": 0.00019999908065647306,
-      "loss": 46.0,
-      "step": 17862
-    },
-    {
-      "epoch": 1.3657510942905748,
-      "grad_norm": 0.0020915286149829626,
-      "learning_rate": 0.00019999908055347418,
-      "loss": 46.0,
-      "step": 17863
-    },
-    {
-      "epoch": 1.3658275512739646,
-      "grad_norm": 0.0012571917613968253,
-      "learning_rate": 0.00019999908045046952,
-      "loss": 46.0,
-      "step": 17864
-    },
-    {
-      "epoch": 1.3659040082573541,
-      "grad_norm": 0.01209266483783722,
-      "learning_rate": 0.00019999908034745912,
-      "loss": 46.0,
-      "step": 17865
-    },
-    {
-      "epoch": 1.365980465240744,
-      "grad_norm": 0.0008113992516882718,
-      "learning_rate": 0.00019999908024444295,
-      "loss": 46.0,
-      "step": 17866
-    },
-    {
-      "epoch": 1.3660569222241337,
-      "grad_norm": 0.0009873210219666362,
-      "learning_rate": 0.00019999908014142103,
-      "loss": 46.0,
-      "step": 17867
-    },
-    {
-      "epoch": 1.3661333792075234,
-      "grad_norm": 0.0007087065605446696,
-      "learning_rate": 0.0001999990800383933,
-      "loss": 46.0,
-      "step": 17868
-    },
-    {
-      "epoch": 1.366209836190913,
-      "grad_norm": 0.0008767751860432327,
-      "learning_rate": 0.00019999907993535984,
-      "loss": 46.0,
-      "step": 17869
-    },
-    {
-      "epoch": 1.3662862931743027,
-      "grad_norm": 0.0004419413744471967,
-      "learning_rate": 0.00019999907983232057,
-      "loss": 46.0,
-      "step": 17870
-    },
-    {
-      "epoch": 1.3663627501576925,
-      "grad_norm": 0.005212228279560804,
-      "learning_rate": 0.00019999907972927556,
-      "loss": 46.0,
-      "step": 17871
-    },
-    {
-      "epoch": 1.3664392071410822,
-      "grad_norm": 0.006762003526091576,
-      "learning_rate": 0.00019999907962622477,
-      "loss": 46.0,
-      "step": 17872
-    },
-    {
-      "epoch": 1.366515664124472,
-      "grad_norm": 0.0020630399230867624,
-      "learning_rate": 0.00019999907952316824,
-      "loss": 46.0,
-      "step": 17873
-    },
-    {
-      "epoch": 1.3665921211078618,
-      "grad_norm": 0.0007178988307714462,
-      "learning_rate": 0.0001999990794201059,
-      "loss": 46.0,
-      "step": 17874
-    },
-    {
-      "epoch": 1.3666685780912515,
-      "grad_norm": 0.00037824458559043705,
-      "learning_rate": 0.0001999990793170378,
-      "loss": 46.0,
-      "step": 17875
-    },
-    {
-      "epoch": 1.366745035074641,
-      "grad_norm": 0.0027194602880626917,
-      "learning_rate": 0.00019999907921396391,
-      "loss": 46.0,
-      "step": 17876
-    },
-    {
-      "epoch": 1.3668214920580308,
-      "grad_norm": 0.0009563557105138898,
-      "learning_rate": 0.0001999990791108843,
-      "loss": 46.0,
-      "step": 17877
-    },
-    {
-      "epoch": 1.3668979490414206,
-      "grad_norm": 0.001081207301467657,
-      "learning_rate": 0.0001999990790077989,
-      "loss": 46.0,
-      "step": 17878
-    },
-    {
-      "epoch": 1.3669744060248104,
-      "grad_norm": 0.0022381143644452095,
-      "learning_rate": 0.0001999990789047077,
-      "loss": 46.0,
-      "step": 17879
-    },
-    {
-      "epoch": 1.3670508630082,
-      "grad_norm": 0.0019142135279253125,
-      "learning_rate": 0.00019999907880161074,
-      "loss": 46.0,
-      "step": 17880
-    },
-    {
-      "epoch": 1.3671273199915897,
-      "grad_norm": 0.0004608039162121713,
-      "learning_rate": 0.00019999907869850804,
-      "loss": 46.0,
-      "step": 17881
-    },
-    {
-      "epoch": 1.3672037769749794,
-      "grad_norm": 0.0015253640012815595,
-      "learning_rate": 0.00019999907859539958,
-      "loss": 46.0,
-      "step": 17882
-    },
-    {
-      "epoch": 1.3672802339583692,
-      "grad_norm": 0.0005132431979291141,
-      "learning_rate": 0.0001999990784922853,
-      "loss": 46.0,
-      "step": 17883
-    },
-    {
-      "epoch": 1.367356690941759,
-      "grad_norm": 0.0008245253120549023,
-      "learning_rate": 0.0001999990783891653,
-      "loss": 46.0,
-      "step": 17884
-    },
-    {
-      "epoch": 1.3674331479251487,
-      "grad_norm": 0.0005842418177053332,
-      "learning_rate": 0.0001999990782860395,
-      "loss": 46.0,
-      "step": 17885
-    },
-    {
-      "epoch": 1.3675096049085385,
-      "grad_norm": 0.0039068120531737804,
-      "learning_rate": 0.00019999907818290792,
-      "loss": 46.0,
-      "step": 17886
-    },
-    {
-      "epoch": 1.367586061891928,
-      "grad_norm": 0.002817972796037793,
-      "learning_rate": 0.00019999907807977062,
-      "loss": 46.0,
-      "step": 17887
-    },
-    {
-      "epoch": 1.3676625188753178,
-      "grad_norm": 0.0013007987290620804,
-      "learning_rate": 0.0001999990779766275,
-      "loss": 46.0,
-      "step": 17888
-    },
-    {
-      "epoch": 1.3677389758587075,
-      "grad_norm": 0.0021931403316557407,
-      "learning_rate": 0.00019999907787347866,
-      "loss": 46.0,
-      "step": 17889
-    },
-    {
-      "epoch": 1.3678154328420973,
-      "grad_norm": 0.001547219348140061,
-      "learning_rate": 0.000199999077770324,
-      "loss": 46.0,
-      "step": 17890
-    },
-    {
-      "epoch": 1.3678918898254868,
-      "grad_norm": 0.0011261365143582225,
-      "learning_rate": 0.0001999990776671636,
-      "loss": 46.0,
-      "step": 17891
-    },
-    {
-      "epoch": 1.3679683468088766,
-      "grad_norm": 0.0013355721021071076,
-      "learning_rate": 0.0001999990775639974,
-      "loss": 46.0,
-      "step": 17892
-    },
-    {
-      "epoch": 1.3680448037922663,
-      "grad_norm": 0.0009573460556566715,
-      "learning_rate": 0.00019999907746082548,
-      "loss": 46.0,
-      "step": 17893
-    },
-    {
-      "epoch": 1.368121260775656,
-      "grad_norm": 0.005690068006515503,
-      "learning_rate": 0.00019999907735764776,
-      "loss": 46.0,
-      "step": 17894
-    },
-    {
-      "epoch": 1.3681977177590459,
-      "grad_norm": 0.0011770413257181644,
-      "learning_rate": 0.00019999907725446424,
-      "loss": 46.0,
-      "step": 17895
-    },
-    {
-      "epoch": 1.3682741747424356,
-      "grad_norm": 0.0020774193108081818,
-      "learning_rate": 0.00019999907715127503,
-      "loss": 46.0,
-      "step": 17896
-    },
-    {
-      "epoch": 1.3683506317258254,
-      "grad_norm": 0.0014860432129353285,
-      "learning_rate": 0.00019999907704808,
-      "loss": 46.0,
-      "step": 17897
-    },
-    {
-      "epoch": 1.368427088709215,
-      "grad_norm": 0.0005194481927901506,
-      "learning_rate": 0.00019999907694487917,
-      "loss": 46.0,
-      "step": 17898
-    },
-    {
-      "epoch": 1.3685035456926047,
-      "grad_norm": 0.007208625786006451,
-      "learning_rate": 0.00019999907684167264,
-      "loss": 46.0,
-      "step": 17899
-    },
-    {
-      "epoch": 1.3685800026759944,
-      "grad_norm": 0.0011275600409135222,
-      "learning_rate": 0.0001999990767384603,
-      "loss": 46.0,
-      "step": 17900
-    },
-    {
-      "epoch": 1.3686564596593842,
-      "grad_norm": 0.0010615979554131627,
-      "learning_rate": 0.0001999990766352422,
-      "loss": 46.0,
-      "step": 17901
-    },
-    {
-      "epoch": 1.3687329166427737,
-      "grad_norm": 0.0016583160031586885,
-      "learning_rate": 0.00019999907653201832,
-      "loss": 46.0,
-      "step": 17902
-    },
-    {
-      "epoch": 1.3688093736261635,
-      "grad_norm": 0.0018497579731047153,
-      "learning_rate": 0.00019999907642878867,
-      "loss": 46.0,
-      "step": 17903
-    },
-    {
-      "epoch": 1.3688858306095533,
-      "grad_norm": 0.0030268337577581406,
-      "learning_rate": 0.00019999907632555327,
-      "loss": 46.0,
-      "step": 17904
-    },
-    {
-      "epoch": 1.368962287592943,
-      "grad_norm": 0.0031173464376479387,
-      "learning_rate": 0.00019999907622231212,
-      "loss": 46.0,
-      "step": 17905
-    },
-    {
-      "epoch": 1.3690387445763328,
-      "grad_norm": 0.0028509721159934998,
-      "learning_rate": 0.00019999907611906515,
-      "loss": 46.0,
-      "step": 17906
-    },
-    {
-      "epoch": 1.3691152015597225,
-      "grad_norm": 0.0009712688042782247,
-      "learning_rate": 0.00019999907601581243,
-      "loss": 46.0,
-      "step": 17907
-    },
-    {
-      "epoch": 1.3691916585431123,
-      "grad_norm": 0.0008491530315950513,
-      "learning_rate": 0.00019999907591255393,
-      "loss": 46.0,
-      "step": 17908
-    },
-    {
-      "epoch": 1.3692681155265018,
-      "grad_norm": 0.0006790999323129654,
-      "learning_rate": 0.0001999990758092897,
-      "loss": 46.0,
-      "step": 17909
-    },
-    {
-      "epoch": 1.3693445725098916,
-      "grad_norm": 0.0007914750021882355,
-      "learning_rate": 0.00019999907570601965,
-      "loss": 46.0,
-      "step": 17910
-    },
-    {
-      "epoch": 1.3694210294932814,
-      "grad_norm": 0.0009068988147191703,
-      "learning_rate": 0.00019999907560274384,
-      "loss": 46.0,
-      "step": 17911
-    },
-    {
-      "epoch": 1.3694974864766711,
-      "grad_norm": 0.002898185048252344,
-      "learning_rate": 0.00019999907549946228,
-      "loss": 46.0,
-      "step": 17912
-    },
-    {
-      "epoch": 1.3695739434600607,
-      "grad_norm": 0.00536075234413147,
-      "learning_rate": 0.00019999907539617495,
-      "loss": 46.0,
-      "step": 17913
-    },
-    {
-      "epoch": 1.3696504004434504,
-      "grad_norm": 0.0004926254041492939,
-      "learning_rate": 0.00019999907529288184,
-      "loss": 46.0,
-      "step": 17914
-    },
-    {
-      "epoch": 1.3697268574268402,
-      "grad_norm": 0.0009255745098926127,
-      "learning_rate": 0.000199999075189583,
-      "loss": 46.0,
-      "step": 17915
-    },
-    {
-      "epoch": 1.36980331441023,
-      "grad_norm": 0.001889142906293273,
-      "learning_rate": 0.00019999907508627836,
-      "loss": 46.0,
-      "step": 17916
-    },
-    {
-      "epoch": 1.3698797713936197,
-      "grad_norm": 0.00608845567330718,
-      "learning_rate": 0.00019999907498296794,
-      "loss": 46.0,
-      "step": 17917
-    },
-    {
-      "epoch": 1.3699562283770095,
-      "grad_norm": 0.0018852971261367202,
-      "learning_rate": 0.00019999907487965174,
-      "loss": 46.0,
-      "step": 17918
-    },
-    {
-      "epoch": 1.3700326853603992,
-      "grad_norm": 0.0007477473118342459,
-      "learning_rate": 0.0001999990747763298,
-      "loss": 46.0,
-      "step": 17919
-    },
-    {
-      "epoch": 1.3701091423437888,
-      "grad_norm": 0.0013447540113702416,
-      "learning_rate": 0.00019999907467300207,
-      "loss": 46.0,
-      "step": 17920
-    },
-    {
-      "epoch": 1.3701855993271785,
-      "grad_norm": 0.0011717418674379587,
-      "learning_rate": 0.00019999907456966858,
-      "loss": 46.0,
-      "step": 17921
-    },
-    {
-      "epoch": 1.3702620563105683,
-      "grad_norm": 0.0016690271440893412,
-      "learning_rate": 0.00019999907446632934,
-      "loss": 46.0,
-      "step": 17922
-    },
-    {
-      "epoch": 1.370338513293958,
-      "grad_norm": 0.0013515135506168008,
-      "learning_rate": 0.0001999990743629843,
-      "loss": 46.0,
-      "step": 17923
-    },
-    {
-      "epoch": 1.3704149702773476,
-      "grad_norm": 0.0006318299565464258,
-      "learning_rate": 0.00019999907425963348,
-      "loss": 46.0,
-      "step": 17924
-    },
-    {
-      "epoch": 1.3704914272607374,
-      "grad_norm": 0.0007922223303467035,
-      "learning_rate": 0.00019999907415627692,
-      "loss": 46.0,
-      "step": 17925
-    },
-    {
-      "epoch": 1.3705678842441271,
-      "grad_norm": 0.0012414443772286177,
-      "learning_rate": 0.0001999990740529146,
-      "loss": 46.0,
-      "step": 17926
-    },
-    {
-      "epoch": 1.3706443412275169,
-      "grad_norm": 0.002843524096533656,
-      "learning_rate": 0.00019999907394954649,
-      "loss": 46.0,
-      "step": 17927
-    },
-    {
-      "epoch": 1.3707207982109066,
-      "grad_norm": 0.0016238440293818712,
-      "learning_rate": 0.0001999990738461726,
-      "loss": 46.0,
-      "step": 17928
-    },
-    {
-      "epoch": 1.3707972551942964,
-      "grad_norm": 0.007920495234429836,
-      "learning_rate": 0.00019999907374279295,
-      "loss": 46.0,
-      "step": 17929
-    },
-    {
-      "epoch": 1.3708737121776862,
-      "grad_norm": 0.0012356085935607553,
-      "learning_rate": 0.00019999907363940755,
-      "loss": 46.0,
-      "step": 17930
-    },
-    {
-      "epoch": 1.3709501691610757,
-      "grad_norm": 0.003255319781601429,
-      "learning_rate": 0.00019999907353601638,
-      "loss": 46.0,
-      "step": 17931
-    },
-    {
-      "epoch": 1.3710266261444655,
-      "grad_norm": 0.0024844917934387922,
-      "learning_rate": 0.0001999990734326194,
-      "loss": 46.0,
-      "step": 17932
-    },
-    {
-      "epoch": 1.3711030831278552,
-      "grad_norm": 0.0020016005728393793,
-      "learning_rate": 0.0001999990733292167,
-      "loss": 46.0,
-      "step": 17933
-    },
-    {
-      "epoch": 1.371179540111245,
-      "grad_norm": 0.0007648197934031487,
-      "learning_rate": 0.0001999990732258082,
-      "loss": 46.0,
-      "step": 17934
-    },
-    {
-      "epoch": 1.3712559970946345,
-      "grad_norm": 0.000825755181722343,
-      "learning_rate": 0.00019999907312239393,
-      "loss": 46.0,
-      "step": 17935
-    },
-    {
-      "epoch": 1.3713324540780243,
-      "grad_norm": 0.004587918519973755,
-      "learning_rate": 0.0001999990730189739,
-      "loss": 46.0,
-      "step": 17936
-    },
-    {
-      "epoch": 1.371408911061414,
-      "grad_norm": 0.0035124497953802347,
-      "learning_rate": 0.00019999907291554808,
-      "loss": 46.0,
-      "step": 17937
-    },
-    {
-      "epoch": 1.3714853680448038,
-      "grad_norm": 0.0048817857168614864,
-      "learning_rate": 0.00019999907281211652,
-      "loss": 46.0,
-      "step": 17938
-    },
-    {
-      "epoch": 1.3715618250281936,
-      "grad_norm": 0.0010551700834184885,
-      "learning_rate": 0.00019999907270867919,
-      "loss": 46.0,
-      "step": 17939
-    },
-    {
-      "epoch": 1.3716382820115833,
-      "grad_norm": 0.0010012377751991153,
-      "learning_rate": 0.00019999907260523608,
-      "loss": 46.0,
-      "step": 17940
-    },
-    {
-      "epoch": 1.3717147389949729,
-      "grad_norm": 0.0027756786439567804,
-      "learning_rate": 0.0001999990725017872,
-      "loss": 46.0,
-      "step": 17941
-    },
-    {
-      "epoch": 1.3717911959783626,
-      "grad_norm": 0.00627170642837882,
-      "learning_rate": 0.00019999907239833255,
-      "loss": 46.0,
-      "step": 17942
-    },
-    {
-      "epoch": 1.3718676529617524,
-      "grad_norm": 0.0006332259508781135,
-      "learning_rate": 0.00019999907229487215,
-      "loss": 46.0,
-      "step": 17943
-    },
-    {
-      "epoch": 1.3719441099451422,
-      "grad_norm": 0.0006545702344737947,
-      "learning_rate": 0.00019999907219140598,
-      "loss": 46.0,
-      "step": 17944
-    },
-    {
-      "epoch": 1.372020566928532,
-      "grad_norm": 0.00031641146051697433,
-      "learning_rate": 0.000199999072087934,
-      "loss": 46.0,
-      "step": 17945
-    },
-    {
-      "epoch": 1.3720970239119215,
-      "grad_norm": 0.0009163955692201853,
-      "learning_rate": 0.00019999907198445626,
-      "loss": 46.0,
-      "step": 17946
-    },
-    {
-      "epoch": 1.3721734808953112,
-      "grad_norm": 0.0012735750060528517,
-      "learning_rate": 0.00019999907188097277,
-      "loss": 46.0,
-      "step": 17947
-    },
-    {
-      "epoch": 1.372249937878701,
-      "grad_norm": 0.0017701095202937722,
-      "learning_rate": 0.00019999907177748353,
-      "loss": 46.0,
-      "step": 17948
-    },
-    {
-      "epoch": 1.3723263948620907,
-      "grad_norm": 0.0012463940074667335,
-      "learning_rate": 0.0001999990716739885,
-      "loss": 46.0,
-      "step": 17949
-    },
-    {
-      "epoch": 1.3724028518454805,
-      "grad_norm": 0.0007806146750226617,
-      "learning_rate": 0.00019999907157048768,
-      "loss": 46.0,
-      "step": 17950
-    },
-    {
-      "epoch": 1.3724793088288703,
-      "grad_norm": 0.0014912097249180079,
-      "learning_rate": 0.0001999990714669811,
-      "loss": 46.0,
-      "step": 17951
-    },
-    {
-      "epoch": 1.3725557658122598,
-      "grad_norm": 0.000257108302321285,
-      "learning_rate": 0.00019999907136346878,
-      "loss": 46.0,
-      "step": 17952
-    },
-    {
-      "epoch": 1.3726322227956496,
-      "grad_norm": 0.003247505519539118,
-      "learning_rate": 0.00019999907125995065,
-      "loss": 46.0,
-      "step": 17953
-    },
-    {
-      "epoch": 1.3727086797790393,
-      "grad_norm": 0.0015863614389672875,
-      "learning_rate": 0.00019999907115642677,
-      "loss": 46.0,
-      "step": 17954
-    },
-    {
-      "epoch": 1.372785136762429,
-      "grad_norm": 0.000813423132058233,
-      "learning_rate": 0.00019999907105289712,
-      "loss": 46.0,
-      "step": 17955
-    },
-    {
-      "epoch": 1.3728615937458186,
-      "grad_norm": 0.0008148307679221034,
-      "learning_rate": 0.00019999907094936172,
-      "loss": 46.0,
-      "step": 17956
-    },
-    {
-      "epoch": 1.3729380507292084,
-      "grad_norm": 0.0012644513044506311,
-      "learning_rate": 0.0001999990708458205,
-      "loss": 46.0,
-      "step": 17957
-    },
-    {
-      "epoch": 1.3730145077125981,
-      "grad_norm": 0.0009246161789633334,
-      "learning_rate": 0.00019999907074227355,
-      "loss": 46.0,
-      "step": 17958
-    },
-    {
-      "epoch": 1.373090964695988,
-      "grad_norm": 0.00284808618016541,
-      "learning_rate": 0.00019999907063872083,
-      "loss": 46.0,
-      "step": 17959
-    },
-    {
-      "epoch": 1.3731674216793777,
-      "grad_norm": 0.008393158204853535,
-      "learning_rate": 0.00019999907053516234,
-      "loss": 46.0,
-      "step": 17960
-    },
-    {
-      "epoch": 1.3732438786627674,
-      "grad_norm": 0.0004486207035370171,
-      "learning_rate": 0.00019999907043159807,
-      "loss": 46.0,
-      "step": 17961
-    },
-    {
-      "epoch": 1.3733203356461572,
-      "grad_norm": 0.0007089965511113405,
-      "learning_rate": 0.00019999907032802804,
-      "loss": 46.0,
-      "step": 17962
-    },
-    {
-      "epoch": 1.3733967926295467,
-      "grad_norm": 0.0009626184473745525,
-      "learning_rate": 0.00019999907022445222,
-      "loss": 46.0,
-      "step": 17963
-    },
-    {
-      "epoch": 1.3734732496129365,
-      "grad_norm": 0.0025398533325642347,
-      "learning_rate": 0.00019999907012087067,
-      "loss": 46.0,
-      "step": 17964
-    },
-    {
-      "epoch": 1.3735497065963262,
-      "grad_norm": 0.010438891127705574,
-      "learning_rate": 0.0001999990700172833,
-      "loss": 46.0,
-      "step": 17965
-    },
-    {
-      "epoch": 1.373626163579716,
-      "grad_norm": 0.0007563194376416504,
-      "learning_rate": 0.0001999990699136902,
-      "loss": 46.0,
-      "step": 17966
-    },
-    {
-      "epoch": 1.3737026205631055,
-      "grad_norm": 0.0009807052556425333,
-      "learning_rate": 0.00019999906981009132,
-      "loss": 46.0,
-      "step": 17967
-    },
-    {
-      "epoch": 1.3737790775464953,
-      "grad_norm": 0.0006703404360450804,
-      "learning_rate": 0.00019999906970648667,
-      "loss": 46.0,
-      "step": 17968
-    },
-    {
-      "epoch": 1.373855534529885,
-      "grad_norm": 0.0011624944163486362,
-      "learning_rate": 0.00019999906960287625,
-      "loss": 46.0,
-      "step": 17969
-    },
-    {
-      "epoch": 1.3739319915132748,
-      "grad_norm": 0.0015916782431304455,
-      "learning_rate": 0.00019999906949926005,
-      "loss": 46.0,
-      "step": 17970
-    },
-    {
-      "epoch": 1.3740084484966646,
-      "grad_norm": 0.0005979355191811919,
-      "learning_rate": 0.00019999906939563807,
-      "loss": 46.0,
-      "step": 17971
-    },
-    {
-      "epoch": 1.3740849054800544,
-      "grad_norm": 0.000791784783359617,
-      "learning_rate": 0.00019999906929201036,
-      "loss": 46.0,
-      "step": 17972
-    },
-    {
-      "epoch": 1.3741613624634441,
-      "grad_norm": 0.0009857387049123645,
-      "learning_rate": 0.00019999906918837686,
-      "loss": 46.0,
-      "step": 17973
-    },
-    {
-      "epoch": 1.3742378194468337,
-      "grad_norm": 0.0004861295747105032,
-      "learning_rate": 0.0001999990690847376,
-      "loss": 46.0,
-      "step": 17974
-    },
-    {
-      "epoch": 1.3743142764302234,
-      "grad_norm": 0.0010758782736957073,
-      "learning_rate": 0.0001999990689810926,
-      "loss": 46.0,
-      "step": 17975
-    },
-    {
-      "epoch": 1.3743907334136132,
-      "grad_norm": 0.0011926807928830385,
-      "learning_rate": 0.00019999906887744175,
-      "loss": 46.0,
-      "step": 17976
-    },
-    {
-      "epoch": 1.374467190397003,
-      "grad_norm": 0.0020258103031665087,
-      "learning_rate": 0.00019999906877378516,
-      "loss": 46.0,
-      "step": 17977
-    },
-    {
-      "epoch": 1.3745436473803925,
-      "grad_norm": 0.002082027029246092,
-      "learning_rate": 0.00019999906867012283,
-      "loss": 46.0,
-      "step": 17978
-    },
-    {
-      "epoch": 1.3746201043637822,
-      "grad_norm": 0.0010071509750559926,
-      "learning_rate": 0.00019999906856645473,
-      "loss": 46.0,
-      "step": 17979
-    },
-    {
-      "epoch": 1.374696561347172,
-      "grad_norm": 0.0007822625921107829,
-      "learning_rate": 0.00019999906846278082,
-      "loss": 46.0,
-      "step": 17980
-    },
-    {
-      "epoch": 1.3747730183305618,
-      "grad_norm": 0.0021289994474500418,
-      "learning_rate": 0.00019999906835910117,
-      "loss": 46.0,
-      "step": 17981
-    },
-    {
-      "epoch": 1.3748494753139515,
-      "grad_norm": 0.0009119003661908209,
-      "learning_rate": 0.00019999906825541577,
-      "loss": 46.0,
-      "step": 17982
-    },
-    {
-      "epoch": 1.3749259322973413,
-      "grad_norm": 0.0012132251868024468,
-      "learning_rate": 0.00019999906815172455,
-      "loss": 46.0,
-      "step": 17983
-    },
-    {
-      "epoch": 1.375002389280731,
-      "grad_norm": 0.0029130529146641493,
-      "learning_rate": 0.0001999990680480276,
-      "loss": 46.0,
-      "step": 17984
-    },
-    {
-      "epoch": 1.3750788462641206,
-      "grad_norm": 0.0032822180073708296,
-      "learning_rate": 0.00019999906794432488,
-      "loss": 46.0,
-      "step": 17985
-    },
-    {
-      "epoch": 1.3751553032475103,
-      "grad_norm": 0.011254685930907726,
-      "learning_rate": 0.00019999906784061636,
-      "loss": 46.0,
-      "step": 17986
-    },
-    {
-      "epoch": 1.3752317602309,
-      "grad_norm": 0.000657734926789999,
-      "learning_rate": 0.0001999990677369021,
-      "loss": 46.0,
-      "step": 17987
-    },
-    {
-      "epoch": 1.3753082172142899,
-      "grad_norm": 0.001019347575493157,
-      "learning_rate": 0.00019999906763318206,
-      "loss": 46.0,
-      "step": 17988
-    },
-    {
-      "epoch": 1.3753846741976794,
-      "grad_norm": 0.0015836822567507625,
-      "learning_rate": 0.00019999906752945625,
-      "loss": 46.0,
-      "step": 17989
-    },
-    {
-      "epoch": 1.3754611311810692,
-      "grad_norm": 0.0008894450729712844,
-      "learning_rate": 0.00019999906742572464,
-      "loss": 46.0,
-      "step": 17990
-    },
-    {
-      "epoch": 1.375537588164459,
-      "grad_norm": 0.0017351461574435234,
-      "learning_rate": 0.00019999906732198733,
-      "loss": 46.0,
-      "step": 17991
-    },
-    {
-      "epoch": 1.3756140451478487,
-      "grad_norm": 0.0009439553832635283,
-      "learning_rate": 0.0001999990672182442,
-      "loss": 46.0,
-      "step": 17992
-    },
-    {
-      "epoch": 1.3756905021312384,
-      "grad_norm": 0.0010688863694667816,
-      "learning_rate": 0.0001999990671144953,
-      "loss": 46.0,
-      "step": 17993
-    },
-    {
-      "epoch": 1.3757669591146282,
-      "grad_norm": 0.004524154122918844,
-      "learning_rate": 0.00019999906701074065,
-      "loss": 46.0,
-      "step": 17994
-    },
-    {
-      "epoch": 1.375843416098018,
-      "grad_norm": 0.0012322546681389213,
-      "learning_rate": 0.00019999906690698022,
-      "loss": 46.0,
-      "step": 17995
-    },
-    {
-      "epoch": 1.3759198730814075,
-      "grad_norm": 0.0018652474973350763,
-      "learning_rate": 0.00019999906680321402,
-      "loss": 46.0,
-      "step": 17996
-    },
-    {
-      "epoch": 1.3759963300647973,
-      "grad_norm": 0.0015578032471239567,
-      "learning_rate": 0.00019999906669944205,
-      "loss": 46.0,
-      "step": 17997
-    },
-    {
-      "epoch": 1.376072787048187,
-      "grad_norm": 0.0012477594427764416,
-      "learning_rate": 0.00019999906659566433,
-      "loss": 46.0,
-      "step": 17998
-    },
-    {
-      "epoch": 1.3761492440315768,
-      "grad_norm": 0.005300607997924089,
-      "learning_rate": 0.00019999906649188084,
-      "loss": 46.0,
-      "step": 17999
-    },
-    {
-      "epoch": 1.3762257010149663,
-      "grad_norm": 0.0008351447177119553,
-      "learning_rate": 0.00019999906638809155,
-      "loss": 46.0,
-      "step": 18000
-    },
-    {
-      "epoch": 1.376302157998356,
-      "grad_norm": 0.0007692829240113497,
-      "learning_rate": 0.0001999990662842965,
-      "loss": 46.0,
-      "step": 18001
-    },
-    {
-      "epoch": 1.3763786149817459,
-      "grad_norm": 0.0037672400940209627,
-      "learning_rate": 0.0001999990661804957,
-      "loss": 46.0,
-      "step": 18002
-    },
-    {
-      "epoch": 1.3764550719651356,
-      "grad_norm": 0.0014659576117992401,
-      "learning_rate": 0.00019999906607668912,
-      "loss": 46.0,
-      "step": 18003
-    },
-    {
-      "epoch": 1.3765315289485254,
-      "grad_norm": 0.0017991299973800778,
-      "learning_rate": 0.00019999906597287676,
-      "loss": 46.0,
-      "step": 18004
-    },
-    {
-      "epoch": 1.3766079859319151,
-      "grad_norm": 0.0021526277996599674,
-      "learning_rate": 0.00019999906586905866,
-      "loss": 46.0,
-      "step": 18005
-    },
-    {
-      "epoch": 1.376684442915305,
-      "grad_norm": 0.0014948190655559301,
-      "learning_rate": 0.00019999906576523478,
-      "loss": 46.0,
-      "step": 18006
-    },
-    {
-      "epoch": 1.3767608998986944,
-      "grad_norm": 0.0003963281342294067,
-      "learning_rate": 0.00019999906566140513,
-      "loss": 46.0,
-      "step": 18007
-    },
-    {
-      "epoch": 1.3768373568820842,
-      "grad_norm": 0.008857641369104385,
-      "learning_rate": 0.00019999906555756967,
-      "loss": 46.0,
-      "step": 18008
-    },
-    {
-      "epoch": 1.376913813865474,
-      "grad_norm": 0.0010356189450249076,
-      "learning_rate": 0.00019999906545372848,
-      "loss": 46.0,
-      "step": 18009
-    },
-    {
-      "epoch": 1.3769902708488637,
-      "grad_norm": 0.0034679740201681852,
-      "learning_rate": 0.0001999990653498815,
-      "loss": 46.0,
-      "step": 18010
-    },
-    {
-      "epoch": 1.3770667278322533,
-      "grad_norm": 0.0016193451592698693,
-      "learning_rate": 0.0001999990652460288,
-      "loss": 46.0,
-      "step": 18011
-    },
-    {
-      "epoch": 1.377143184815643,
-      "grad_norm": 0.0035377254243940115,
-      "learning_rate": 0.00019999906514217027,
-      "loss": 46.0,
-      "step": 18012
-    },
-    {
-      "epoch": 1.3772196417990328,
-      "grad_norm": 0.0008031788165681064,
-      "learning_rate": 0.00019999906503830598,
-      "loss": 46.0,
-      "step": 18013
-    },
-    {
-      "epoch": 1.3772960987824225,
-      "grad_norm": 0.0006095413118600845,
-      "learning_rate": 0.00019999906493443597,
-      "loss": 46.0,
-      "step": 18014
-    },
-    {
-      "epoch": 1.3773725557658123,
-      "grad_norm": 0.006988126784563065,
-      "learning_rate": 0.00019999906483056016,
-      "loss": 46.0,
-      "step": 18015
-    },
-    {
-      "epoch": 1.377449012749202,
-      "grad_norm": 0.0007553836330771446,
-      "learning_rate": 0.00019999906472667855,
-      "loss": 46.0,
-      "step": 18016
-    },
-    {
-      "epoch": 1.3775254697325918,
-      "grad_norm": 0.0019366928609088063,
-      "learning_rate": 0.00019999906462279122,
-      "loss": 46.0,
-      "step": 18017
-    },
-    {
-      "epoch": 1.3776019267159814,
-      "grad_norm": 0.0012283315882086754,
-      "learning_rate": 0.00019999906451889808,
-      "loss": 46.0,
-      "step": 18018
-    },
-    {
-      "epoch": 1.3776783836993711,
-      "grad_norm": 0.0007402935298159719,
-      "learning_rate": 0.0001999990644149992,
-      "loss": 46.0,
-      "step": 18019
-    },
-    {
-      "epoch": 1.3777548406827609,
-      "grad_norm": 0.0009826201712712646,
-      "learning_rate": 0.00019999906431109453,
-      "loss": 46.0,
-      "step": 18020
-    },
-    {
-      "epoch": 1.3778312976661506,
-      "grad_norm": 0.000709379673935473,
-      "learning_rate": 0.0001999990642071841,
-      "loss": 46.0,
-      "step": 18021
-    },
-    {
-      "epoch": 1.3779077546495402,
-      "grad_norm": 0.0015795029466971755,
-      "learning_rate": 0.0001999990641032679,
-      "loss": 46.0,
-      "step": 18022
-    },
-    {
-      "epoch": 1.37798421163293,
-      "grad_norm": 0.0005581254372373223,
-      "learning_rate": 0.00019999906399934594,
-      "loss": 46.0,
-      "step": 18023
-    },
-    {
-      "epoch": 1.3780606686163197,
-      "grad_norm": 0.0004443573416210711,
-      "learning_rate": 0.0001999990638954182,
-      "loss": 46.0,
-      "step": 18024
-    },
-    {
-      "epoch": 1.3781371255997095,
-      "grad_norm": 0.0005796657060272992,
-      "learning_rate": 0.00019999906379148468,
-      "loss": 46.0,
-      "step": 18025
-    },
-    {
-      "epoch": 1.3782135825830992,
-      "grad_norm": 0.002661475446075201,
-      "learning_rate": 0.0001999990636875454,
-      "loss": 46.0,
-      "step": 18026
-    },
-    {
-      "epoch": 1.378290039566489,
-      "grad_norm": 0.0007935474277473986,
-      "learning_rate": 0.00019999906358360037,
-      "loss": 46.0,
-      "step": 18027
-    },
-    {
-      "epoch": 1.3783664965498788,
-      "grad_norm": 0.00486373994499445,
-      "learning_rate": 0.00019999906347964954,
-      "loss": 46.0,
-      "step": 18028
-    },
-    {
-      "epoch": 1.3784429535332683,
-      "grad_norm": 0.003100744681432843,
-      "learning_rate": 0.00019999906337569295,
-      "loss": 46.0,
-      "step": 18029
-    },
-    {
-      "epoch": 1.378519410516658,
-      "grad_norm": 0.0008301998022943735,
-      "learning_rate": 0.00019999906327173062,
-      "loss": 46.0,
-      "step": 18030
-    },
-    {
-      "epoch": 1.3785958675000478,
-      "grad_norm": 0.0011180470464751124,
-      "learning_rate": 0.0001999990631677625,
-      "loss": 46.0,
-      "step": 18031
-    },
-    {
-      "epoch": 1.3786723244834376,
-      "grad_norm": 0.00048506358871236444,
-      "learning_rate": 0.0001999990630637886,
-      "loss": 46.0,
-      "step": 18032
-    },
-    {
-      "epoch": 1.3787487814668271,
-      "grad_norm": 0.0005775447352789342,
-      "learning_rate": 0.00019999906295980894,
-      "loss": 46.0,
-      "step": 18033
-    },
-    {
-      "epoch": 1.3788252384502169,
-      "grad_norm": 0.0025686314329504967,
-      "learning_rate": 0.00019999906285582349,
-      "loss": 46.0,
-      "step": 18034
-    },
-    {
-      "epoch": 1.3789016954336066,
-      "grad_norm": 0.0007559625082649291,
-      "learning_rate": 0.00019999906275183232,
-      "loss": 46.0,
-      "step": 18035
-    },
-    {
-      "epoch": 1.3789781524169964,
-      "grad_norm": 0.0007519559003412724,
-      "learning_rate": 0.00019999906264783532,
-      "loss": 46.0,
-      "step": 18036
-    },
-    {
-      "epoch": 1.3790546094003862,
-      "grad_norm": 0.0005074836080893874,
-      "learning_rate": 0.00019999906254383257,
-      "loss": 46.0,
-      "step": 18037
-    },
-    {
-      "epoch": 1.379131066383776,
-      "grad_norm": 0.0018584885401651263,
-      "learning_rate": 0.00019999906243982406,
-      "loss": 46.0,
-      "step": 18038
-    },
-    {
-      "epoch": 1.3792075233671657,
-      "grad_norm": 0.0008913018391467631,
-      "learning_rate": 0.0001999990623358098,
-      "loss": 46.0,
-      "step": 18039
-    },
-    {
-      "epoch": 1.3792839803505552,
-      "grad_norm": 0.0002679518365766853,
-      "learning_rate": 0.00019999906223178976,
-      "loss": 46.0,
-      "step": 18040
-    },
-    {
-      "epoch": 1.379360437333945,
-      "grad_norm": 0.012079217471182346,
-      "learning_rate": 0.00019999906212776392,
-      "loss": 46.0,
-      "step": 18041
-    },
-    {
-      "epoch": 1.3794368943173347,
-      "grad_norm": 0.0012403877917677164,
-      "learning_rate": 0.00019999906202373234,
-      "loss": 46.0,
-      "step": 18042
-    },
-    {
-      "epoch": 1.3795133513007245,
-      "grad_norm": 0.001639240887016058,
-      "learning_rate": 0.00019999906191969498,
-      "loss": 46.0,
-      "step": 18043
-    },
-    {
-      "epoch": 1.379589808284114,
-      "grad_norm": 0.0009506415808573365,
-      "learning_rate": 0.00019999906181565185,
-      "loss": 46.0,
-      "step": 18044
-    },
-    {
-      "epoch": 1.3796662652675038,
-      "grad_norm": 0.005230034701526165,
-      "learning_rate": 0.00019999906171160297,
-      "loss": 46.0,
-      "step": 18045
-    },
-    {
-      "epoch": 1.3797427222508936,
-      "grad_norm": 0.0006512513500638306,
-      "learning_rate": 0.00019999906160754832,
-      "loss": 46.0,
-      "step": 18046
-    },
-    {
-      "epoch": 1.3798191792342833,
-      "grad_norm": 0.00037874715053476393,
-      "learning_rate": 0.00019999906150348787,
-      "loss": 46.0,
-      "step": 18047
-    },
-    {
-      "epoch": 1.379895636217673,
-      "grad_norm": 0.0018193217692896724,
-      "learning_rate": 0.00019999906139942167,
-      "loss": 46.0,
-      "step": 18048
-    },
-    {
-      "epoch": 1.3799720932010628,
-      "grad_norm": 0.0005614160909317434,
-      "learning_rate": 0.00019999906129534968,
-      "loss": 46.0,
-      "step": 18049
-    },
-    {
-      "epoch": 1.3800485501844526,
-      "grad_norm": 0.00171806407161057,
-      "learning_rate": 0.00019999906119127193,
-      "loss": 46.0,
-      "step": 18050
-    },
-    {
-      "epoch": 1.3801250071678421,
-      "grad_norm": 0.001787136192433536,
-      "learning_rate": 0.00019999906108718842,
-      "loss": 46.0,
-      "step": 18051
-    },
-    {
-      "epoch": 1.380201464151232,
-      "grad_norm": 0.001263280282728374,
-      "learning_rate": 0.00019999906098309915,
-      "loss": 46.0,
-      "step": 18052
-    },
-    {
-      "epoch": 1.3802779211346217,
-      "grad_norm": 0.0009184384834952652,
-      "learning_rate": 0.00019999906087900412,
-      "loss": 46.0,
-      "step": 18053
-    },
-    {
-      "epoch": 1.3803543781180114,
-      "grad_norm": 0.00200935872271657,
-      "learning_rate": 0.00019999906077490328,
-      "loss": 46.0,
-      "step": 18054
-    },
-    {
-      "epoch": 1.380430835101401,
-      "grad_norm": 0.0013755237450823188,
-      "learning_rate": 0.0001999990606707967,
-      "loss": 46.0,
-      "step": 18055
-    },
-    {
-      "epoch": 1.3805072920847907,
-      "grad_norm": 0.00067552225664258,
-      "learning_rate": 0.00019999906056668434,
-      "loss": 46.0,
-      "step": 18056
-    },
-    {
-      "epoch": 1.3805837490681805,
-      "grad_norm": 0.001759067177772522,
-      "learning_rate": 0.0001999990604625662,
-      "loss": 46.0,
-      "step": 18057
-    },
-    {
-      "epoch": 1.3806602060515702,
-      "grad_norm": 0.0011331499554216862,
-      "learning_rate": 0.0001999990603584423,
-      "loss": 46.0,
-      "step": 18058
-    },
-    {
-      "epoch": 1.38073666303496,
-      "grad_norm": 0.0015485257608816028,
-      "learning_rate": 0.00019999906025431263,
-      "loss": 46.0,
-      "step": 18059
-    },
-    {
-      "epoch": 1.3808131200183498,
-      "grad_norm": 0.0003694407641887665,
-      "learning_rate": 0.0001999990601501772,
-      "loss": 46.0,
-      "step": 18060
-    },
-    {
-      "epoch": 1.3808895770017395,
-      "grad_norm": 0.0037997965700924397,
-      "learning_rate": 0.00019999906004603598,
-      "loss": 46.0,
-      "step": 18061
-    },
-    {
-      "epoch": 1.380966033985129,
-      "grad_norm": 0.0008745102677494287,
-      "learning_rate": 0.000199999059941889,
-      "loss": 46.0,
-      "step": 18062
-    },
-    {
-      "epoch": 1.3810424909685188,
-      "grad_norm": 0.0009955859277397394,
-      "learning_rate": 0.00019999905983773627,
-      "loss": 46.0,
-      "step": 18063
-    },
-    {
-      "epoch": 1.3811189479519086,
-      "grad_norm": 0.0020191327203065157,
-      "learning_rate": 0.00019999905973357775,
-      "loss": 46.0,
-      "step": 18064
-    },
-    {
-      "epoch": 1.3811954049352984,
-      "grad_norm": 0.0013388830702751875,
-      "learning_rate": 0.0001999990596294135,
-      "loss": 46.0,
-      "step": 18065
-    },
-    {
-      "epoch": 1.381271861918688,
-      "grad_norm": 0.03064756467938423,
-      "learning_rate": 0.0001999990595252434,
-      "loss": 46.0,
-      "step": 18066
-    },
-    {
-      "epoch": 1.3813483189020777,
-      "grad_norm": 0.002888164948672056,
-      "learning_rate": 0.00019999905942106762,
-      "loss": 46.0,
-      "step": 18067
-    },
-    {
-      "epoch": 1.3814247758854674,
-      "grad_norm": 0.001091856975108385,
-      "learning_rate": 0.000199999059316886,
-      "loss": 46.0,
-      "step": 18068
-    },
-    {
-      "epoch": 1.3815012328688572,
-      "grad_norm": 0.002184961922466755,
-      "learning_rate": 0.00019999905921269865,
-      "loss": 46.0,
-      "step": 18069
-    },
-    {
-      "epoch": 1.381577689852247,
-      "grad_norm": 0.0020398579072207212,
-      "learning_rate": 0.00019999905910850552,
-      "loss": 46.0,
-      "step": 18070
-    },
-    {
-      "epoch": 1.3816541468356367,
-      "grad_norm": 0.0006634164601564407,
-      "learning_rate": 0.0001999990590043066,
-      "loss": 46.0,
-      "step": 18071
-    },
-    {
-      "epoch": 1.3817306038190262,
-      "grad_norm": 0.0009516172576695681,
-      "learning_rate": 0.00019999905890010197,
-      "loss": 46.0,
-      "step": 18072
-    },
-    {
-      "epoch": 1.381807060802416,
-      "grad_norm": 0.0010657728416845202,
-      "learning_rate": 0.0001999990587958915,
-      "loss": 46.0,
-      "step": 18073
-    },
-    {
-      "epoch": 1.3818835177858058,
-      "grad_norm": 0.0009695056360214949,
-      "learning_rate": 0.0001999990586916753,
-      "loss": 46.0,
-      "step": 18074
-    },
-    {
-      "epoch": 1.3819599747691955,
-      "grad_norm": 0.0032378961332142353,
-      "learning_rate": 0.00019999905858745332,
-      "loss": 46.0,
-      "step": 18075
-    },
-    {
-      "epoch": 1.3820364317525853,
-      "grad_norm": 0.011132714338600636,
-      "learning_rate": 0.00019999905848322558,
-      "loss": 46.0,
-      "step": 18076
-    },
-    {
-      "epoch": 1.3821128887359748,
-      "grad_norm": 0.0009438168490305543,
-      "learning_rate": 0.00019999905837899204,
-      "loss": 46.0,
-      "step": 18077
-    },
-    {
-      "epoch": 1.3821893457193646,
-      "grad_norm": 0.0013492620782926679,
-      "learning_rate": 0.00019999905827475275,
-      "loss": 46.0,
-      "step": 18078
-    },
-    {
-      "epoch": 1.3822658027027543,
-      "grad_norm": 0.0009250510483980179,
-      "learning_rate": 0.0001999990581705077,
-      "loss": 46.0,
-      "step": 18079
-    },
-    {
-      "epoch": 1.382342259686144,
-      "grad_norm": 0.01226492878049612,
-      "learning_rate": 0.0001999990580662569,
-      "loss": 46.0,
-      "step": 18080
-    },
-    {
-      "epoch": 1.3824187166695339,
-      "grad_norm": 0.002441051183268428,
-      "learning_rate": 0.0001999990579620003,
-      "loss": 46.0,
-      "step": 18081
-    },
-    {
-      "epoch": 1.3824951736529236,
-      "grad_norm": 0.005781848449259996,
-      "learning_rate": 0.00019999905785773794,
-      "loss": 46.0,
-      "step": 18082
-    },
-    {
-      "epoch": 1.3825716306363132,
-      "grad_norm": 0.0007825401844456792,
-      "learning_rate": 0.00019999905775346978,
-      "loss": 46.0,
-      "step": 18083
-    },
-    {
-      "epoch": 1.382648087619703,
-      "grad_norm": 0.0014283062191680074,
-      "learning_rate": 0.0001999990576491959,
-      "loss": 46.0,
-      "step": 18084
-    },
-    {
-      "epoch": 1.3827245446030927,
-      "grad_norm": 0.0014039864763617516,
-      "learning_rate": 0.0001999990575449162,
-      "loss": 46.0,
-      "step": 18085
-    },
-    {
-      "epoch": 1.3828010015864824,
-      "grad_norm": 0.01543811522424221,
-      "learning_rate": 0.00019999905744063078,
-      "loss": 46.0,
-      "step": 18086
-    },
-    {
-      "epoch": 1.3828774585698722,
-      "grad_norm": 0.0005554198869504035,
-      "learning_rate": 0.00019999905733633955,
-      "loss": 46.0,
-      "step": 18087
-    },
-    {
-      "epoch": 1.3829539155532617,
-      "grad_norm": 0.0008181114681065083,
-      "learning_rate": 0.00019999905723204259,
-      "loss": 46.0,
-      "step": 18088
-    },
-    {
-      "epoch": 1.3830303725366515,
-      "grad_norm": 0.0018580456962808967,
-      "learning_rate": 0.00019999905712773984,
-      "loss": 46.0,
-      "step": 18089
-    },
-    {
-      "epoch": 1.3831068295200413,
-      "grad_norm": 0.005458520259708166,
-      "learning_rate": 0.00019999905702343133,
-      "loss": 46.0,
-      "step": 18090
-    },
-    {
-      "epoch": 1.383183286503431,
-      "grad_norm": 0.0006446646875701845,
-      "learning_rate": 0.000199999056919117,
-      "loss": 46.0,
-      "step": 18091
-    },
-    {
-      "epoch": 1.3832597434868208,
-      "grad_norm": 0.01109933853149414,
-      "learning_rate": 0.00019999905681479697,
-      "loss": 46.0,
-      "step": 18092
-    },
-    {
-      "epoch": 1.3833362004702106,
-      "grad_norm": 0.0034960834309458733,
-      "learning_rate": 0.00019999905671047117,
-      "loss": 46.0,
-      "step": 18093
-    },
-    {
-      "epoch": 1.3834126574536,
-      "grad_norm": 0.001120292698033154,
-      "learning_rate": 0.00019999905660613956,
-      "loss": 46.0,
-      "step": 18094
-    },
-    {
-      "epoch": 1.3834891144369899,
-      "grad_norm": 0.0008797007612884045,
-      "learning_rate": 0.0001999990565018022,
-      "loss": 46.0,
-      "step": 18095
-    },
-    {
-      "epoch": 1.3835655714203796,
-      "grad_norm": 0.001252058893442154,
-      "learning_rate": 0.00019999905639745907,
-      "loss": 46.0,
-      "step": 18096
-    },
-    {
-      "epoch": 1.3836420284037694,
-      "grad_norm": 0.0007532025338150561,
-      "learning_rate": 0.00019999905629311014,
-      "loss": 46.0,
-      "step": 18097
-    },
-    {
-      "epoch": 1.383718485387159,
-      "grad_norm": 0.0006492463871836662,
-      "learning_rate": 0.00019999905618875547,
-      "loss": 46.0,
-      "step": 18098
-    },
-    {
-      "epoch": 1.3837949423705487,
-      "grad_norm": 0.0011558914557099342,
-      "learning_rate": 0.00019999905608439504,
-      "loss": 46.0,
-      "step": 18099
-    },
-    {
-      "epoch": 1.3838713993539384,
-      "grad_norm": 0.0006250928272493184,
-      "learning_rate": 0.00019999905598002885,
-      "loss": 46.0,
-      "step": 18100
-    },
-    {
-      "epoch": 1.3839478563373282,
-      "grad_norm": 0.0006683696992695332,
-      "learning_rate": 0.00019999905587565685,
-      "loss": 46.0,
-      "step": 18101
-    },
-    {
-      "epoch": 1.384024313320718,
-      "grad_norm": 0.0007870956906117499,
-      "learning_rate": 0.00019999905577127908,
-      "loss": 46.0,
-      "step": 18102
-    },
-    {
-      "epoch": 1.3841007703041077,
-      "grad_norm": 0.00279410881921649,
-      "learning_rate": 0.00019999905566689557,
-      "loss": 46.0,
-      "step": 18103
-    },
-    {
-      "epoch": 1.3841772272874975,
-      "grad_norm": 0.0008921432308852673,
-      "learning_rate": 0.00019999905556250628,
-      "loss": 46.0,
-      "step": 18104
-    },
-    {
-      "epoch": 1.384253684270887,
-      "grad_norm": 0.000745880592148751,
-      "learning_rate": 0.00019999905545811121,
-      "loss": 46.0,
-      "step": 18105
-    },
-    {
-      "epoch": 1.3843301412542768,
-      "grad_norm": 0.0004097307682968676,
-      "learning_rate": 0.00019999905535371038,
-      "loss": 46.0,
-      "step": 18106
-    },
-    {
-      "epoch": 1.3844065982376665,
-      "grad_norm": 0.00046161108184605837,
-      "learning_rate": 0.0001999990552493038,
-      "loss": 46.0,
-      "step": 18107
-    },
-    {
-      "epoch": 1.3844830552210563,
-      "grad_norm": 0.0007116220076568425,
-      "learning_rate": 0.00019999905514489141,
-      "loss": 46.0,
-      "step": 18108
-    },
-    {
-      "epoch": 1.3845595122044458,
-      "grad_norm": 0.0009194620070047677,
-      "learning_rate": 0.00019999905504047329,
-      "loss": 46.0,
-      "step": 18109
-    },
-    {
-      "epoch": 1.3846359691878356,
-      "grad_norm": 0.0036174501292407513,
-      "learning_rate": 0.00019999905493604938,
-      "loss": 46.0,
-      "step": 18110
-    },
-    {
-      "epoch": 1.3847124261712254,
-      "grad_norm": 0.0023814463056623936,
-      "learning_rate": 0.00019999905483161968,
-      "loss": 46.0,
-      "step": 18111
-    },
-    {
-      "epoch": 1.3847888831546151,
-      "grad_norm": 0.002753094071522355,
-      "learning_rate": 0.00019999905472718426,
-      "loss": 46.0,
-      "step": 18112
-    },
-    {
-      "epoch": 1.3848653401380049,
-      "grad_norm": 0.0006366602028720081,
-      "learning_rate": 0.00019999905462274304,
-      "loss": 46.0,
-      "step": 18113
-    },
-    {
-      "epoch": 1.3849417971213946,
-      "grad_norm": 0.00701924879103899,
-      "learning_rate": 0.00019999905451829607,
-      "loss": 46.0,
-      "step": 18114
-    },
-    {
-      "epoch": 1.3850182541047844,
-      "grad_norm": 0.0017806018004193902,
-      "learning_rate": 0.0001999990544138433,
-      "loss": 46.0,
-      "step": 18115
-    },
-    {
-      "epoch": 1.385094711088174,
-      "grad_norm": 0.0009679956128820777,
-      "learning_rate": 0.00019999905430938478,
-      "loss": 46.0,
-      "step": 18116
-    },
-    {
-      "epoch": 1.3851711680715637,
-      "grad_norm": 0.0004925800021737814,
-      "learning_rate": 0.0001999990542049205,
-      "loss": 46.0,
-      "step": 18117
-    },
-    {
-      "epoch": 1.3852476250549535,
-      "grad_norm": 0.0009191755088977516,
-      "learning_rate": 0.00019999905410045043,
-      "loss": 46.0,
-      "step": 18118
-    },
-    {
-      "epoch": 1.3853240820383432,
-      "grad_norm": 0.0038608727045357227,
-      "learning_rate": 0.0001999990539959746,
-      "loss": 46.0,
-      "step": 18119
-    },
-    {
-      "epoch": 1.3854005390217328,
-      "grad_norm": 0.0008928577299229801,
-      "learning_rate": 0.00019999905389149301,
-      "loss": 46.0,
-      "step": 18120
-    },
-    {
-      "epoch": 1.3854769960051225,
-      "grad_norm": 0.0006483289762400091,
-      "learning_rate": 0.0001999990537870056,
-      "loss": 46.0,
-      "step": 18121
-    },
-    {
-      "epoch": 1.3855534529885123,
-      "grad_norm": 0.0017912021139636636,
-      "learning_rate": 0.00019999905368251248,
-      "loss": 46.0,
-      "step": 18122
-    },
-    {
-      "epoch": 1.385629909971902,
-      "grad_norm": 0.005643123760819435,
-      "learning_rate": 0.00019999905357801357,
-      "loss": 46.0,
-      "step": 18123
-    },
-    {
-      "epoch": 1.3857063669552918,
-      "grad_norm": 0.0035028192214667797,
-      "learning_rate": 0.00019999905347350893,
-      "loss": 46.0,
-      "step": 18124
-    },
-    {
-      "epoch": 1.3857828239386816,
-      "grad_norm": 0.006375194061547518,
-      "learning_rate": 0.00019999905336899845,
-      "loss": 46.0,
-      "step": 18125
-    },
-    {
-      "epoch": 1.3858592809220713,
-      "grad_norm": 0.0007990702288225293,
-      "learning_rate": 0.00019999905326448223,
-      "loss": 46.0,
-      "step": 18126
-    },
-    {
-      "epoch": 1.3859357379054609,
-      "grad_norm": 0.001139629865065217,
-      "learning_rate": 0.00019999905315996026,
-      "loss": 46.0,
-      "step": 18127
-    },
-    {
-      "epoch": 1.3860121948888506,
-      "grad_norm": 0.002823038026690483,
-      "learning_rate": 0.0001999990530554325,
-      "loss": 46.0,
-      "step": 18128
-    },
-    {
-      "epoch": 1.3860886518722404,
-      "grad_norm": 0.0005093346117064357,
-      "learning_rate": 0.00019999905295089895,
-      "loss": 46.0,
-      "step": 18129
-    },
-    {
-      "epoch": 1.3861651088556302,
-      "grad_norm": 0.006516624707728624,
-      "learning_rate": 0.00019999905284635966,
-      "loss": 46.0,
-      "step": 18130
-    },
-    {
-      "epoch": 1.3862415658390197,
-      "grad_norm": 0.0016051542479544878,
-      "learning_rate": 0.0001999990527418146,
-      "loss": 46.0,
-      "step": 18131
-    },
-    {
-      "epoch": 1.3863180228224095,
-      "grad_norm": 0.0006347924936562777,
-      "learning_rate": 0.00019999905263726376,
-      "loss": 46.0,
-      "step": 18132
-    },
-    {
-      "epoch": 1.3863944798057992,
-      "grad_norm": 0.002941821003332734,
-      "learning_rate": 0.00019999905253270715,
-      "loss": 46.0,
-      "step": 18133
-    },
-    {
-      "epoch": 1.386470936789189,
-      "grad_norm": 0.0010585044510662556,
-      "learning_rate": 0.0001999990524281448,
-      "loss": 46.0,
-      "step": 18134
-    },
-    {
-      "epoch": 1.3865473937725787,
-      "grad_norm": 0.0011076467344537377,
-      "learning_rate": 0.00019999905232357664,
-      "loss": 46.0,
-      "step": 18135
-    },
-    {
-      "epoch": 1.3866238507559685,
-      "grad_norm": 0.0016473408322781324,
-      "learning_rate": 0.00019999905221900274,
-      "loss": 46.0,
-      "step": 18136
-    },
-    {
-      "epoch": 1.3867003077393583,
-      "grad_norm": 0.0028028725646436214,
-      "learning_rate": 0.00019999905211442307,
-      "loss": 46.0,
-      "step": 18137
-    },
-    {
-      "epoch": 1.3867767647227478,
-      "grad_norm": 0.0036406780127435923,
-      "learning_rate": 0.00019999905200983762,
-      "loss": 46.0,
-      "step": 18138
-    },
-    {
-      "epoch": 1.3868532217061376,
-      "grad_norm": 0.0006841887370683253,
-      "learning_rate": 0.0001999990519052464,
-      "loss": 46.0,
-      "step": 18139
-    },
-    {
-      "epoch": 1.3869296786895273,
-      "grad_norm": 0.0030098850838840008,
-      "learning_rate": 0.00019999905180064943,
-      "loss": 46.0,
-      "step": 18140
-    },
-    {
-      "epoch": 1.387006135672917,
-      "grad_norm": 0.0015919606667011976,
-      "learning_rate": 0.00019999905169604663,
-      "loss": 46.0,
-      "step": 18141
-    },
-    {
-      "epoch": 1.3870825926563066,
-      "grad_norm": 0.0008367933914996684,
-      "learning_rate": 0.00019999905159143812,
-      "loss": 46.0,
-      "step": 18142
-    },
-    {
-      "epoch": 1.3871590496396964,
-      "grad_norm": 0.0021825837902724743,
-      "learning_rate": 0.00019999905148682383,
-      "loss": 46.0,
-      "step": 18143
-    },
-    {
-      "epoch": 1.3872355066230861,
-      "grad_norm": 0.002459099283441901,
-      "learning_rate": 0.00019999905138220377,
-      "loss": 46.0,
-      "step": 18144
-    },
-    {
-      "epoch": 1.387311963606476,
-      "grad_norm": 0.0010476986644789577,
-      "learning_rate": 0.0001999990512775779,
-      "loss": 46.0,
-      "step": 18145
-    },
-    {
-      "epoch": 1.3873884205898657,
-      "grad_norm": 0.0003944080963265151,
-      "learning_rate": 0.00019999905117294633,
-      "loss": 46.0,
-      "step": 18146
-    },
-    {
-      "epoch": 1.3874648775732554,
-      "grad_norm": 0.000341695238603279,
-      "learning_rate": 0.00019999905106830895,
-      "loss": 46.0,
-      "step": 18147
-    },
-    {
-      "epoch": 1.3875413345566452,
-      "grad_norm": 0.002485455246642232,
-      "learning_rate": 0.0001999990509636658,
-      "loss": 46.0,
-      "step": 18148
-    },
-    {
-      "epoch": 1.3876177915400347,
-      "grad_norm": 0.0016624669078737497,
-      "learning_rate": 0.0001999990508590169,
-      "loss": 46.0,
-      "step": 18149
-    },
-    {
-      "epoch": 1.3876942485234245,
-      "grad_norm": 0.0015724037075415254,
-      "learning_rate": 0.00019999905075436221,
-      "loss": 46.0,
-      "step": 18150
-    },
-    {
-      "epoch": 1.3877707055068143,
-      "grad_norm": 0.0010537775233387947,
-      "learning_rate": 0.00019999905064970174,
-      "loss": 46.0,
-      "step": 18151
-    },
-    {
-      "epoch": 1.387847162490204,
-      "grad_norm": 0.003523198189213872,
-      "learning_rate": 0.00019999905054503554,
-      "loss": 46.0,
-      "step": 18152
-    },
-    {
-      "epoch": 1.3879236194735936,
-      "grad_norm": 0.002381999744102359,
-      "learning_rate": 0.00019999905044036352,
-      "loss": 46.0,
-      "step": 18153
-    },
-    {
-      "epoch": 1.3880000764569833,
-      "grad_norm": 0.001294403919018805,
-      "learning_rate": 0.00019999905033568578,
-      "loss": 46.0,
-      "step": 18154
-    },
-    {
-      "epoch": 1.388076533440373,
-      "grad_norm": 0.0008013628539629281,
-      "learning_rate": 0.00019999905023100224,
-      "loss": 46.0,
-      "step": 18155
-    },
-    {
-      "epoch": 1.3881529904237628,
-      "grad_norm": 0.0018153274431824684,
-      "learning_rate": 0.00019999905012631295,
-      "loss": 46.0,
-      "step": 18156
-    },
-    {
-      "epoch": 1.3882294474071526,
-      "grad_norm": 0.0008623525500297546,
-      "learning_rate": 0.00019999905002161786,
-      "loss": 46.0,
-      "step": 18157
-    },
-    {
-      "epoch": 1.3883059043905424,
-      "grad_norm": 0.0004039740015286952,
-      "learning_rate": 0.00019999904991691706,
-      "loss": 46.0,
-      "step": 18158
-    },
-    {
-      "epoch": 1.3883823613739321,
-      "grad_norm": 0.000994433299638331,
-      "learning_rate": 0.00019999904981221045,
-      "loss": 46.0,
-      "step": 18159
-    },
-    {
-      "epoch": 1.3884588183573217,
-      "grad_norm": 0.007976262830197811,
-      "learning_rate": 0.00019999904970749807,
-      "loss": 46.0,
-      "step": 18160
-    },
-    {
-      "epoch": 1.3885352753407114,
-      "grad_norm": 0.0012299882946535945,
-      "learning_rate": 0.0001999990496027799,
-      "loss": 46.0,
-      "step": 18161
-    },
-    {
-      "epoch": 1.3886117323241012,
-      "grad_norm": 0.0017116060480475426,
-      "learning_rate": 0.000199999049498056,
-      "loss": 46.0,
-      "step": 18162
-    },
-    {
-      "epoch": 1.388688189307491,
-      "grad_norm": 0.0010554678738117218,
-      "learning_rate": 0.0001999990493933263,
-      "loss": 46.0,
-      "step": 18163
-    },
-    {
-      "epoch": 1.3887646462908805,
-      "grad_norm": 0.000657076423522085,
-      "learning_rate": 0.00019999904928859086,
-      "loss": 46.0,
-      "step": 18164
-    },
-    {
-      "epoch": 1.3888411032742702,
-      "grad_norm": 0.0016456126468256116,
-      "learning_rate": 0.00019999904918384961,
-      "loss": 46.0,
-      "step": 18165
-    },
-    {
-      "epoch": 1.38891756025766,
-      "grad_norm": 0.000919406593311578,
-      "learning_rate": 0.00019999904907910265,
-      "loss": 46.0,
-      "step": 18166
-    },
-    {
-      "epoch": 1.3889940172410498,
-      "grad_norm": 0.0007722167647443712,
-      "learning_rate": 0.00019999904897434988,
-      "loss": 46.0,
-      "step": 18167
-    },
-    {
-      "epoch": 1.3890704742244395,
-      "grad_norm": 0.003925792407244444,
-      "learning_rate": 0.00019999904886959136,
-      "loss": 46.0,
-      "step": 18168
-    },
-    {
-      "epoch": 1.3891469312078293,
-      "grad_norm": 0.001052317675203085,
-      "learning_rate": 0.00019999904876482705,
-      "loss": 46.0,
-      "step": 18169
-    },
-    {
-      "epoch": 1.389223388191219,
-      "grad_norm": 0.0009575384319759905,
-      "learning_rate": 0.00019999904866005696,
-      "loss": 46.0,
-      "step": 18170
-    },
-    {
-      "epoch": 1.3892998451746086,
-      "grad_norm": 0.0005681662005372345,
-      "learning_rate": 0.00019999904855528113,
-      "loss": 46.0,
-      "step": 18171
-    },
-    {
-      "epoch": 1.3893763021579983,
-      "grad_norm": 0.0010825723875313997,
-      "learning_rate": 0.00019999904845049952,
-      "loss": 46.0,
-      "step": 18172
-    },
-    {
-      "epoch": 1.389452759141388,
-      "grad_norm": 0.0005776237812824547,
-      "learning_rate": 0.00019999904834571214,
-      "loss": 46.0,
-      "step": 18173
-    },
-    {
-      "epoch": 1.3895292161247779,
-      "grad_norm": 0.0010564846452325583,
-      "learning_rate": 0.00019999904824091898,
-      "loss": 46.0,
-      "step": 18174
-    },
-    {
-      "epoch": 1.3896056731081674,
-      "grad_norm": 0.002812911756336689,
-      "learning_rate": 0.00019999904813612008,
-      "loss": 46.0,
-      "step": 18175
-    },
-    {
-      "epoch": 1.3896821300915572,
-      "grad_norm": 0.0008109891205094755,
-      "learning_rate": 0.00019999904803131538,
-      "loss": 46.0,
-      "step": 18176
-    },
-    {
-      "epoch": 1.389758587074947,
-      "grad_norm": 0.002363978885114193,
-      "learning_rate": 0.00019999904792650494,
-      "loss": 46.0,
-      "step": 18177
-    },
-    {
-      "epoch": 1.3898350440583367,
-      "grad_norm": 0.0009998250752687454,
-      "learning_rate": 0.00019999904782168871,
-      "loss": 46.0,
-      "step": 18178
-    },
-    {
-      "epoch": 1.3899115010417264,
-      "grad_norm": 0.001122571644373238,
-      "learning_rate": 0.00019999904771686672,
-      "loss": 46.0,
-      "step": 18179
-    },
-    {
-      "epoch": 1.3899879580251162,
-      "grad_norm": 0.0007013622089289129,
-      "learning_rate": 0.00019999904761203893,
-      "loss": 46.0,
-      "step": 18180
-    },
-    {
-      "epoch": 1.390064415008506,
-      "grad_norm": 0.0003700793313328177,
-      "learning_rate": 0.0001999990475072054,
-      "loss": 46.0,
-      "step": 18181
-    },
-    {
-      "epoch": 1.3901408719918955,
-      "grad_norm": 0.0012531543616205454,
-      "learning_rate": 0.00019999904740236612,
-      "loss": 46.0,
-      "step": 18182
-    },
-    {
-      "epoch": 1.3902173289752853,
-      "grad_norm": 0.002273143734782934,
-      "learning_rate": 0.00019999904729752104,
-      "loss": 46.0,
-      "step": 18183
-    },
-    {
-      "epoch": 1.390293785958675,
-      "grad_norm": 0.001236573327332735,
-      "learning_rate": 0.0001999990471926702,
-      "loss": 46.0,
-      "step": 18184
-    },
-    {
-      "epoch": 1.3903702429420648,
-      "grad_norm": 0.0004166388534940779,
-      "learning_rate": 0.0001999990470878136,
-      "loss": 46.0,
-      "step": 18185
-    },
-    {
-      "epoch": 1.3904466999254543,
-      "grad_norm": 0.0013104334939271212,
-      "learning_rate": 0.00019999904698295121,
-      "loss": 46.0,
-      "step": 18186
-    },
-    {
-      "epoch": 1.390523156908844,
-      "grad_norm": 0.0015591164119541645,
-      "learning_rate": 0.00019999904687808306,
-      "loss": 46.0,
-      "step": 18187
-    },
-    {
-      "epoch": 1.3905996138922339,
-      "grad_norm": 0.006224802695214748,
-      "learning_rate": 0.00019999904677320913,
-      "loss": 46.0,
-      "step": 18188
-    },
-    {
-      "epoch": 1.3906760708756236,
-      "grad_norm": 0.001653607003390789,
-      "learning_rate": 0.00019999904666832946,
-      "loss": 46.0,
-      "step": 18189
-    },
-    {
-      "epoch": 1.3907525278590134,
-      "grad_norm": 0.0017317173769697547,
-      "learning_rate": 0.00019999904656344399,
-      "loss": 46.0,
-      "step": 18190
-    },
-    {
-      "epoch": 1.3908289848424031,
-      "grad_norm": 0.0013346589403226972,
-      "learning_rate": 0.00019999904645855276,
-      "loss": 46.0,
-      "step": 18191
-    },
-    {
-      "epoch": 1.390905441825793,
-      "grad_norm": 0.0019391510868445039,
-      "learning_rate": 0.00019999904635365577,
-      "loss": 46.0,
-      "step": 18192
-    },
-    {
-      "epoch": 1.3909818988091824,
-      "grad_norm": 0.005857347045093775,
-      "learning_rate": 0.000199999046248753,
-      "loss": 46.0,
-      "step": 18193
-    },
-    {
-      "epoch": 1.3910583557925722,
-      "grad_norm": 0.0011941107222810388,
-      "learning_rate": 0.00019999904614384446,
-      "loss": 46.0,
-      "step": 18194
-    },
-    {
-      "epoch": 1.391134812775962,
-      "grad_norm": 0.009722065180540085,
-      "learning_rate": 0.00019999904603893015,
-      "loss": 46.0,
-      "step": 18195
-    },
-    {
-      "epoch": 1.3912112697593517,
-      "grad_norm": 0.0006942698382772505,
-      "learning_rate": 0.00019999904593401006,
-      "loss": 46.0,
-      "step": 18196
-    },
-    {
-      "epoch": 1.3912877267427413,
-      "grad_norm": 0.0006114181596785784,
-      "learning_rate": 0.00019999904582908423,
-      "loss": 46.0,
-      "step": 18197
-    },
-    {
-      "epoch": 1.391364183726131,
-      "grad_norm": 0.0012399927945807576,
-      "learning_rate": 0.00019999904572415262,
-      "loss": 46.0,
-      "step": 18198
-    },
-    {
-      "epoch": 1.3914406407095208,
-      "grad_norm": 0.0009988698875531554,
-      "learning_rate": 0.00019999904561921524,
-      "loss": 46.0,
-      "step": 18199
-    },
-    {
-      "epoch": 1.3915170976929105,
-      "grad_norm": 0.000732092303223908,
-      "learning_rate": 0.0001999990455142721,
-      "loss": 46.0,
-      "step": 18200
-    },
-    {
-      "epoch": 1.3915935546763003,
-      "grad_norm": 0.0012976337457075715,
-      "learning_rate": 0.00019999904540932316,
-      "loss": 46.0,
-      "step": 18201
-    },
-    {
-      "epoch": 1.39167001165969,
-      "grad_norm": 0.0012231545988470316,
-      "learning_rate": 0.00019999904530436849,
-      "loss": 46.0,
-      "step": 18202
-    },
-    {
-      "epoch": 1.3917464686430798,
-      "grad_norm": 0.003263626014813781,
-      "learning_rate": 0.00019999904519940799,
-      "loss": 46.0,
-      "step": 18203
-    },
-    {
-      "epoch": 1.3918229256264694,
-      "grad_norm": 0.0005668877274729311,
-      "learning_rate": 0.0001999990450944418,
-      "loss": 46.0,
-      "step": 18204
-    },
-    {
-      "epoch": 1.3918993826098591,
-      "grad_norm": 0.0015439471462741494,
-      "learning_rate": 0.0001999990449894698,
-      "loss": 46.0,
-      "step": 18205
-    },
-    {
-      "epoch": 1.3919758395932489,
-      "grad_norm": 0.0019080775091424584,
-      "learning_rate": 0.000199999044884492,
-      "loss": 46.0,
-      "step": 18206
-    },
-    {
-      "epoch": 1.3920522965766386,
-      "grad_norm": 0.0003980291076004505,
-      "learning_rate": 0.00019999904477950846,
-      "loss": 46.0,
-      "step": 18207
-    },
-    {
-      "epoch": 1.3921287535600282,
-      "grad_norm": 0.0006080828607082367,
-      "learning_rate": 0.00019999904467451915,
-      "loss": 46.0,
-      "step": 18208
-    },
-    {
-      "epoch": 1.392205210543418,
-      "grad_norm": 0.002454567700624466,
-      "learning_rate": 0.00019999904456952406,
-      "loss": 46.0,
-      "step": 18209
-    },
-    {
-      "epoch": 1.3922816675268077,
-      "grad_norm": 0.0016060691559687257,
-      "learning_rate": 0.00019999904446452326,
-      "loss": 46.0,
-      "step": 18210
-    },
-    {
-      "epoch": 1.3923581245101975,
-      "grad_norm": 0.0007170222816057503,
-      "learning_rate": 0.00019999904435951665,
-      "loss": 46.0,
-      "step": 18211
-    },
-    {
-      "epoch": 1.3924345814935872,
-      "grad_norm": 0.0011072804918512702,
-      "learning_rate": 0.00019999904425450424,
-      "loss": 46.0,
-      "step": 18212
-    },
-    {
-      "epoch": 1.392511038476977,
-      "grad_norm": 0.004278427455574274,
-      "learning_rate": 0.00019999904414948606,
-      "loss": 46.0,
-      "step": 18213
-    },
-    {
-      "epoch": 1.3925874954603665,
-      "grad_norm": 0.0011637897696346045,
-      "learning_rate": 0.00019999904404446216,
-      "loss": 46.0,
-      "step": 18214
-    },
-    {
-      "epoch": 1.3926639524437563,
-      "grad_norm": 0.0004555878695100546,
-      "learning_rate": 0.00019999904393943246,
-      "loss": 46.0,
-      "step": 18215
-    },
-    {
-      "epoch": 1.392740409427146,
-      "grad_norm": 0.0012224672827869654,
-      "learning_rate": 0.00019999904383439702,
-      "loss": 46.0,
-      "step": 18216
-    },
-    {
-      "epoch": 1.3928168664105358,
-      "grad_norm": 0.0016161407111212611,
-      "learning_rate": 0.0001999990437293558,
-      "loss": 46.0,
-      "step": 18217
-    },
-    {
-      "epoch": 1.3928933233939256,
-      "grad_norm": 0.0005803713575005531,
-      "learning_rate": 0.00019999904362430878,
-      "loss": 46.0,
-      "step": 18218
-    },
-    {
-      "epoch": 1.3929697803773151,
-      "grad_norm": 0.0010446751257404685,
-      "learning_rate": 0.000199999043519256,
-      "loss": 46.0,
-      "step": 18219
-    },
-    {
-      "epoch": 1.3930462373607049,
-      "grad_norm": 0.0009772925404831767,
-      "learning_rate": 0.00019999904341419747,
-      "loss": 46.0,
-      "step": 18220
-    },
-    {
-      "epoch": 1.3931226943440946,
-      "grad_norm": 0.0005918702227063477,
-      "learning_rate": 0.00019999904330913316,
-      "loss": 46.0,
-      "step": 18221
-    },
-    {
-      "epoch": 1.3931991513274844,
-      "grad_norm": 0.001663452247157693,
-      "learning_rate": 0.0001999990432040631,
-      "loss": 46.0,
-      "step": 18222
-    },
-    {
-      "epoch": 1.3932756083108742,
-      "grad_norm": 0.004550378769636154,
-      "learning_rate": 0.00019999904309898724,
-      "loss": 46.0,
-      "step": 18223
-    },
-    {
-      "epoch": 1.393352065294264,
-      "grad_norm": 0.00024089851649478078,
-      "learning_rate": 0.0001999990429939056,
-      "loss": 46.0,
-      "step": 18224
-    },
-    {
-      "epoch": 1.3934285222776535,
-      "grad_norm": 0.0008896503131836653,
-      "learning_rate": 0.00019999904288881822,
-      "loss": 46.0,
-      "step": 18225
-    },
-    {
-      "epoch": 1.3935049792610432,
-      "grad_norm": 0.0010326530318707228,
-      "learning_rate": 0.00019999904278372504,
-      "loss": 46.0,
-      "step": 18226
-    },
-    {
-      "epoch": 1.393581436244433,
-      "grad_norm": 0.0008250580867752433,
-      "learning_rate": 0.00019999904267862614,
-      "loss": 46.0,
-      "step": 18227
-    },
-    {
-      "epoch": 1.3936578932278227,
-      "grad_norm": 0.0008015081402845681,
-      "learning_rate": 0.00019999904257352142,
-      "loss": 46.0,
-      "step": 18228
-    },
-    {
-      "epoch": 1.3937343502112123,
-      "grad_norm": 0.0025446105282753706,
-      "learning_rate": 0.00019999904246841097,
-      "loss": 46.0,
-      "step": 18229
-    },
-    {
-      "epoch": 1.393810807194602,
-      "grad_norm": 0.00039046400343067944,
-      "learning_rate": 0.00019999904236329475,
-      "loss": 46.0,
-      "step": 18230
-    },
-    {
-      "epoch": 1.3938872641779918,
-      "grad_norm": 0.0021942162420600653,
-      "learning_rate": 0.00019999904225817273,
-      "loss": 46.0,
-      "step": 18231
-    },
-    {
-      "epoch": 1.3939637211613816,
-      "grad_norm": 0.010255785658955574,
-      "learning_rate": 0.00019999904215304496,
-      "loss": 46.0,
-      "step": 18232
-    },
-    {
-      "epoch": 1.3940401781447713,
-      "grad_norm": 0.0015067668864503503,
-      "learning_rate": 0.00019999904204791142,
-      "loss": 46.0,
-      "step": 18233
-    },
-    {
-      "epoch": 1.394116635128161,
-      "grad_norm": 0.0006419286364689469,
-      "learning_rate": 0.0001999990419427721,
-      "loss": 46.0,
-      "step": 18234
-    },
-    {
-      "epoch": 1.3941930921115508,
-      "grad_norm": 0.001984857954084873,
-      "learning_rate": 0.000199999041837627,
-      "loss": 46.0,
-      "step": 18235
-    },
-    {
-      "epoch": 1.3942695490949404,
-      "grad_norm": 0.0015746374847367406,
-      "learning_rate": 0.00019999904173247617,
-      "loss": 46.0,
-      "step": 18236
-    },
-    {
-      "epoch": 1.3943460060783301,
-      "grad_norm": 0.0007287380867637694,
-      "learning_rate": 0.00019999904162731956,
-      "loss": 46.0,
-      "step": 18237
-    },
-    {
-      "epoch": 1.39442246306172,
-      "grad_norm": 0.0018226653337478638,
-      "learning_rate": 0.00019999904152215715,
-      "loss": 46.0,
-      "step": 18238
-    },
-    {
-      "epoch": 1.3944989200451097,
-      "grad_norm": 0.0035836955066770315,
-      "learning_rate": 0.000199999041416989,
-      "loss": 46.0,
-      "step": 18239
-    },
-    {
-      "epoch": 1.3945753770284992,
-      "grad_norm": 0.0006668817368336022,
-      "learning_rate": 0.00019999904131181507,
-      "loss": 46.0,
-      "step": 18240
-    },
-    {
-      "epoch": 1.394651834011889,
-      "grad_norm": 0.001640597707591951,
-      "learning_rate": 0.00019999904120663537,
-      "loss": 46.0,
-      "step": 18241
-    },
-    {
-      "epoch": 1.3947282909952787,
-      "grad_norm": 0.0020899372175335884,
-      "learning_rate": 0.0001999990411014499,
-      "loss": 46.0,
-      "step": 18242
-    },
-    {
-      "epoch": 1.3948047479786685,
-      "grad_norm": 0.0014405716210603714,
-      "learning_rate": 0.00019999904099625865,
-      "loss": 46.0,
-      "step": 18243
-    },
-    {
-      "epoch": 1.3948812049620583,
-      "grad_norm": 0.0013223255518823862,
-      "learning_rate": 0.00019999904089106166,
-      "loss": 46.0,
-      "step": 18244
-    },
-    {
-      "epoch": 1.394957661945448,
-      "grad_norm": 0.008614566177129745,
-      "learning_rate": 0.0001999990407858589,
-      "loss": 46.0,
-      "step": 18245
-    },
-    {
-      "epoch": 1.3950341189288378,
-      "grad_norm": 0.0011943380814045668,
-      "learning_rate": 0.00019999904068065033,
-      "loss": 46.0,
-      "step": 18246
-    },
-    {
-      "epoch": 1.3951105759122273,
-      "grad_norm": 0.010545573197305202,
-      "learning_rate": 0.00019999904057543604,
-      "loss": 46.0,
-      "step": 18247
-    },
-    {
-      "epoch": 1.395187032895617,
-      "grad_norm": 0.0006156945019029081,
-      "learning_rate": 0.00019999904047021593,
-      "loss": 46.0,
-      "step": 18248
-    },
-    {
-      "epoch": 1.3952634898790068,
-      "grad_norm": 0.0012297678040340543,
-      "learning_rate": 0.0001999990403649901,
-      "loss": 46.0,
-      "step": 18249
-    },
-    {
-      "epoch": 1.3953399468623966,
-      "grad_norm": 0.002259167144075036,
-      "learning_rate": 0.00019999904025975847,
-      "loss": 46.0,
-      "step": 18250
-    },
-    {
-      "epoch": 1.3954164038457861,
-      "grad_norm": 0.0022718163672834635,
-      "learning_rate": 0.0001999990401545211,
-      "loss": 46.0,
-      "step": 18251
-    },
-    {
-      "epoch": 1.395492860829176,
-      "grad_norm": 0.0020323318894952536,
-      "learning_rate": 0.0001999990400492779,
-      "loss": 46.0,
-      "step": 18252
-    },
-    {
-      "epoch": 1.3955693178125657,
-      "grad_norm": 0.0017350508132949471,
-      "learning_rate": 0.00019999903994402898,
-      "loss": 46.0,
-      "step": 18253
-    },
-    {
-      "epoch": 1.3956457747959554,
-      "grad_norm": 0.0012287198333069682,
-      "learning_rate": 0.00019999903983877428,
-      "loss": 46.0,
-      "step": 18254
-    },
-    {
-      "epoch": 1.3957222317793452,
-      "grad_norm": 0.0009139716858044267,
-      "learning_rate": 0.0001999990397335138,
-      "loss": 46.0,
-      "step": 18255
-    },
-    {
-      "epoch": 1.395798688762735,
-      "grad_norm": 0.001420615124516189,
-      "learning_rate": 0.00019999903962824756,
-      "loss": 46.0,
-      "step": 18256
-    },
-    {
-      "epoch": 1.3958751457461247,
-      "grad_norm": 0.0005012263427488506,
-      "learning_rate": 0.00019999903952297557,
-      "loss": 46.0,
-      "step": 18257
-    },
-    {
-      "epoch": 1.3959516027295142,
-      "grad_norm": 0.0016221682308241725,
-      "learning_rate": 0.00019999903941769778,
-      "loss": 46.0,
-      "step": 18258
-    },
-    {
-      "epoch": 1.396028059712904,
-      "grad_norm": 0.0007155595812946558,
-      "learning_rate": 0.00019999903931241427,
-      "loss": 46.0,
-      "step": 18259
-    },
-    {
-      "epoch": 1.3961045166962938,
-      "grad_norm": 0.007044223137199879,
-      "learning_rate": 0.00019999903920712493,
-      "loss": 46.0,
-      "step": 18260
-    },
-    {
-      "epoch": 1.3961809736796835,
-      "grad_norm": 0.0011055180802941322,
-      "learning_rate": 0.00019999903910182984,
-      "loss": 46.0,
-      "step": 18261
-    },
-    {
-      "epoch": 1.396257430663073,
-      "grad_norm": 0.0015097169671207666,
-      "learning_rate": 0.00019999903899652898,
-      "loss": 46.0,
-      "step": 18262
-    },
-    {
-      "epoch": 1.3963338876464628,
-      "grad_norm": 0.0043243844993412495,
-      "learning_rate": 0.00019999903889122235,
-      "loss": 46.0,
-      "step": 18263
-    },
-    {
-      "epoch": 1.3964103446298526,
-      "grad_norm": 0.0012168345274403691,
-      "learning_rate": 0.00019999903878590994,
-      "loss": 46.0,
-      "step": 18264
-    },
-    {
-      "epoch": 1.3964868016132423,
-      "grad_norm": 0.0007755389669910073,
-      "learning_rate": 0.0001999990386805918,
-      "loss": 46.0,
-      "step": 18265
-    },
-    {
-      "epoch": 1.396563258596632,
-      "grad_norm": 0.0021192762069404125,
-      "learning_rate": 0.00019999903857526787,
-      "loss": 46.0,
-      "step": 18266
-    },
-    {
-      "epoch": 1.3966397155800219,
-      "grad_norm": 0.002402530750259757,
-      "learning_rate": 0.00019999903846993817,
-      "loss": 46.0,
-      "step": 18267
-    },
-    {
-      "epoch": 1.3967161725634116,
-      "grad_norm": 0.0004343707987572998,
-      "learning_rate": 0.0001999990383646027,
-      "loss": 46.0,
-      "step": 18268
-    },
-    {
-      "epoch": 1.3967926295468012,
-      "grad_norm": 0.0013917690375819802,
-      "learning_rate": 0.00019999903825926145,
-      "loss": 46.0,
-      "step": 18269
-    },
-    {
-      "epoch": 1.396869086530191,
-      "grad_norm": 0.0006006306502968073,
-      "learning_rate": 0.00019999903815391443,
-      "loss": 46.0,
-      "step": 18270
-    },
-    {
-      "epoch": 1.3969455435135807,
-      "grad_norm": 0.0033171798568218946,
-      "learning_rate": 0.00019999903804856166,
-      "loss": 46.0,
-      "step": 18271
-    },
-    {
-      "epoch": 1.3970220004969705,
-      "grad_norm": 0.0005021097022108734,
-      "learning_rate": 0.00019999903794320313,
-      "loss": 46.0,
-      "step": 18272
-    },
-    {
-      "epoch": 1.39709845748036,
-      "grad_norm": 0.0017547993920743465,
-      "learning_rate": 0.00019999903783783879,
-      "loss": 46.0,
-      "step": 18273
-    },
-    {
-      "epoch": 1.3971749144637498,
-      "grad_norm": 0.0010602496331557631,
-      "learning_rate": 0.0001999990377324687,
-      "loss": 46.0,
-      "step": 18274
-    },
-    {
-      "epoch": 1.3972513714471395,
-      "grad_norm": 0.000917400699108839,
-      "learning_rate": 0.00019999903762709287,
-      "loss": 46.0,
-      "step": 18275
-    },
-    {
-      "epoch": 1.3973278284305293,
-      "grad_norm": 0.0012330766767263412,
-      "learning_rate": 0.0001999990375217112,
-      "loss": 46.0,
-      "step": 18276
-    },
-    {
-      "epoch": 1.397404285413919,
-      "grad_norm": 0.0012281337985768914,
-      "learning_rate": 0.00019999903741632383,
-      "loss": 46.0,
-      "step": 18277
-    },
-    {
-      "epoch": 1.3974807423973088,
-      "grad_norm": 0.0008549666381441057,
-      "learning_rate": 0.00019999903731093065,
-      "loss": 46.0,
-      "step": 18278
-    },
-    {
-      "epoch": 1.3975571993806986,
-      "grad_norm": 0.004524443298578262,
-      "learning_rate": 0.00019999903720553173,
-      "loss": 46.0,
-      "step": 18279
-    },
-    {
-      "epoch": 1.397633656364088,
-      "grad_norm": 0.0015528372023254633,
-      "learning_rate": 0.00019999903710012703,
-      "loss": 46.0,
-      "step": 18280
-    },
-    {
-      "epoch": 1.3977101133474779,
-      "grad_norm": 0.0013199535897001624,
-      "learning_rate": 0.00019999903699471653,
-      "loss": 46.0,
-      "step": 18281
-    },
-    {
-      "epoch": 1.3977865703308676,
-      "grad_norm": 0.006697415374219418,
-      "learning_rate": 0.0001999990368893003,
-      "loss": 46.0,
-      "step": 18282
-    },
-    {
-      "epoch": 1.3978630273142574,
-      "grad_norm": 0.006428098306059837,
-      "learning_rate": 0.0001999990367838783,
-      "loss": 46.0,
-      "step": 18283
-    },
-    {
-      "epoch": 1.397939484297647,
-      "grad_norm": 0.0013741186121478677,
-      "learning_rate": 0.0001999990366784505,
-      "loss": 46.0,
-      "step": 18284
-    },
-    {
-      "epoch": 1.3980159412810367,
-      "grad_norm": 0.014019043184816837,
-      "learning_rate": 0.00019999903657301696,
-      "loss": 46.0,
-      "step": 18285
-    },
-    {
-      "epoch": 1.3980923982644264,
-      "grad_norm": 0.0014207991771399975,
-      "learning_rate": 0.00019999903646757762,
-      "loss": 46.0,
-      "step": 18286
-    },
-    {
-      "epoch": 1.3981688552478162,
-      "grad_norm": 0.009901546873152256,
-      "learning_rate": 0.00019999903636213256,
-      "loss": 46.0,
-      "step": 18287
-    },
-    {
-      "epoch": 1.398245312231206,
-      "grad_norm": 0.0015407156897708774,
-      "learning_rate": 0.00019999903625668168,
-      "loss": 46.0,
-      "step": 18288
-    },
-    {
-      "epoch": 1.3983217692145957,
-      "grad_norm": 0.0014143314911052585,
-      "learning_rate": 0.00019999903615122507,
-      "loss": 46.0,
-      "step": 18289
-    },
-    {
-      "epoch": 1.3983982261979855,
-      "grad_norm": 0.0013123456155881286,
-      "learning_rate": 0.00019999903604576264,
-      "loss": 46.0,
-      "step": 18290
-    },
-    {
-      "epoch": 1.398474683181375,
-      "grad_norm": 0.0016088540432974696,
-      "learning_rate": 0.0001999990359402945,
-      "loss": 46.0,
-      "step": 18291
-    },
-    {
-      "epoch": 1.3985511401647648,
-      "grad_norm": 0.00041369968676008284,
-      "learning_rate": 0.00019999903583482056,
-      "loss": 46.0,
-      "step": 18292
-    },
-    {
-      "epoch": 1.3986275971481545,
-      "grad_norm": 0.004795182961970568,
-      "learning_rate": 0.00019999903572934084,
-      "loss": 46.0,
-      "step": 18293
-    },
-    {
-      "epoch": 1.3987040541315443,
-      "grad_norm": 0.0006917525315657258,
-      "learning_rate": 0.00019999903562385537,
-      "loss": 46.0,
-      "step": 18294
-    },
-    {
-      "epoch": 1.3987805111149338,
-      "grad_norm": 0.003914724104106426,
-      "learning_rate": 0.00019999903551836412,
-      "loss": 46.0,
-      "step": 18295
-    },
-    {
-      "epoch": 1.3988569680983236,
-      "grad_norm": 0.0014889624435454607,
-      "learning_rate": 0.0001999990354128671,
-      "loss": 46.0,
-      "step": 18296
-    },
-    {
-      "epoch": 1.3989334250817134,
-      "grad_norm": 0.0008348251576535404,
-      "learning_rate": 0.0001999990353073643,
-      "loss": 46.0,
-      "step": 18297
-    },
-    {
-      "epoch": 1.3990098820651031,
-      "grad_norm": 0.009671281091868877,
-      "learning_rate": 0.0001999990352018558,
-      "loss": 46.0,
-      "step": 18298
-    },
-    {
-      "epoch": 1.3990863390484929,
-      "grad_norm": 0.008715366944670677,
-      "learning_rate": 0.00019999903509634144,
-      "loss": 46.0,
-      "step": 18299
-    },
-    {
-      "epoch": 1.3991627960318826,
-      "grad_norm": 0.0030145812779664993,
-      "learning_rate": 0.00019999903499082135,
-      "loss": 46.0,
-      "step": 18300
-    },
-    {
-      "epoch": 1.3992392530152724,
-      "grad_norm": 0.0010817403672263026,
-      "learning_rate": 0.0001999990348852955,
-      "loss": 46.0,
-      "step": 18301
-    },
-    {
-      "epoch": 1.399315709998662,
-      "grad_norm": 0.0003958876186516136,
-      "learning_rate": 0.00019999903477976386,
-      "loss": 46.0,
-      "step": 18302
-    },
-    {
-      "epoch": 1.3993921669820517,
-      "grad_norm": 0.0022777714766561985,
-      "learning_rate": 0.00019999903467422648,
-      "loss": 46.0,
-      "step": 18303
-    },
-    {
-      "epoch": 1.3994686239654415,
-      "grad_norm": 0.0016170493327081203,
-      "learning_rate": 0.0001999990345686833,
-      "loss": 46.0,
-      "step": 18304
-    },
-    {
-      "epoch": 1.3995450809488312,
-      "grad_norm": 0.0036381310783326626,
-      "learning_rate": 0.00019999903446313435,
-      "loss": 46.0,
-      "step": 18305
-    },
-    {
-      "epoch": 1.3996215379322208,
-      "grad_norm": 0.002684554550796747,
-      "learning_rate": 0.00019999903435757963,
-      "loss": 46.0,
-      "step": 18306
-    },
-    {
-      "epoch": 1.3996979949156105,
-      "grad_norm": 0.0026757996529340744,
-      "learning_rate": 0.00019999903425201918,
-      "loss": 46.0,
-      "step": 18307
-    },
-    {
-      "epoch": 1.3997744518990003,
-      "grad_norm": 0.0011949578765779734,
-      "learning_rate": 0.0001999990341464529,
-      "loss": 46.0,
-      "step": 18308
-    },
-    {
-      "epoch": 1.39985090888239,
-      "grad_norm": 0.0006952470284886658,
-      "learning_rate": 0.00019999903404088092,
-      "loss": 46.0,
-      "step": 18309
-    },
-    {
-      "epoch": 1.3999273658657798,
-      "grad_norm": 0.002998712472617626,
-      "learning_rate": 0.0001999990339353031,
-      "loss": 46.0,
-      "step": 18310
-    },
-    {
-      "epoch": 1.4000038228491696,
-      "grad_norm": 0.0013851657276973128,
-      "learning_rate": 0.00019999903382971956,
-      "loss": 46.0,
-      "step": 18311
-    },
-    {
-      "epoch": 1.4000802798325593,
-      "grad_norm": 0.0017672590911388397,
-      "learning_rate": 0.00019999903372413025,
-      "loss": 46.0,
-      "step": 18312
-    },
-    {
-      "epoch": 1.4001567368159489,
-      "grad_norm": 0.001287979306653142,
-      "learning_rate": 0.00019999903361853514,
-      "loss": 46.0,
-      "step": 18313
-    },
-    {
-      "epoch": 1.4002331937993386,
-      "grad_norm": 0.0008470408502034843,
-      "learning_rate": 0.00019999903351293428,
-      "loss": 46.0,
-      "step": 18314
-    },
-    {
-      "epoch": 1.4003096507827284,
-      "grad_norm": 0.0009091580868698657,
-      "learning_rate": 0.00019999903340732765,
-      "loss": 46.0,
-      "step": 18315
-    },
-    {
-      "epoch": 1.4003861077661182,
-      "grad_norm": 0.0037146888207644224,
-      "learning_rate": 0.00019999903330171525,
-      "loss": 46.0,
-      "step": 18316
-    },
-    {
-      "epoch": 1.4004625647495077,
-      "grad_norm": 0.0018429117044433951,
-      "learning_rate": 0.00019999903319609707,
-      "loss": 46.0,
-      "step": 18317
-    },
-    {
-      "epoch": 1.4005390217328975,
-      "grad_norm": 0.0010084389941766858,
-      "learning_rate": 0.00019999903309047312,
-      "loss": 46.0,
-      "step": 18318
-    },
-    {
-      "epoch": 1.4006154787162872,
-      "grad_norm": 0.002414534566923976,
-      "learning_rate": 0.00019999903298484342,
-      "loss": 46.0,
-      "step": 18319
-    },
-    {
-      "epoch": 1.400691935699677,
-      "grad_norm": 0.0010621643159538507,
-      "learning_rate": 0.00019999903287920795,
-      "loss": 46.0,
-      "step": 18320
-    },
-    {
-      "epoch": 1.4007683926830667,
-      "grad_norm": 0.0004849625111091882,
-      "learning_rate": 0.00019999903277356668,
-      "loss": 46.0,
-      "step": 18321
-    },
-    {
-      "epoch": 1.4008448496664565,
-      "grad_norm": 0.0011082657147198915,
-      "learning_rate": 0.00019999903266791966,
-      "loss": 46.0,
-      "step": 18322
-    },
-    {
-      "epoch": 1.4009213066498463,
-      "grad_norm": 0.0006692179595120251,
-      "learning_rate": 0.0001999990325622669,
-      "loss": 46.0,
-      "step": 18323
-    },
-    {
-      "epoch": 1.4009977636332358,
-      "grad_norm": 0.0028096118476241827,
-      "learning_rate": 0.0001999990324566083,
-      "loss": 46.0,
-      "step": 18324
-    },
-    {
-      "epoch": 1.4010742206166256,
-      "grad_norm": 0.009814753197133541,
-      "learning_rate": 0.000199999032350944,
-      "loss": 46.0,
-      "step": 18325
-    },
-    {
-      "epoch": 1.4011506776000153,
-      "grad_norm": 0.00365639035589993,
-      "learning_rate": 0.0001999990322452739,
-      "loss": 46.0,
-      "step": 18326
-    },
-    {
-      "epoch": 1.401227134583405,
-      "grad_norm": 0.003469970775768161,
-      "learning_rate": 0.00019999903213959802,
-      "loss": 46.0,
-      "step": 18327
-    },
-    {
-      "epoch": 1.4013035915667946,
-      "grad_norm": 0.0012035678373649716,
-      "learning_rate": 0.0001999990320339164,
-      "loss": 46.0,
-      "step": 18328
-    },
-    {
-      "epoch": 1.4013800485501844,
-      "grad_norm": 0.0005106648895889521,
-      "learning_rate": 0.000199999031928229,
-      "loss": 46.0,
-      "step": 18329
-    },
-    {
-      "epoch": 1.4014565055335741,
-      "grad_norm": 0.002731070388108492,
-      "learning_rate": 0.0001999990318225358,
-      "loss": 46.0,
-      "step": 18330
-    },
-    {
-      "epoch": 1.401532962516964,
-      "grad_norm": 0.0008909603348001838,
-      "learning_rate": 0.00019999903171683686,
-      "loss": 46.0,
-      "step": 18331
-    },
-    {
-      "epoch": 1.4016094195003537,
-      "grad_norm": 0.0011393693275749683,
-      "learning_rate": 0.00019999903161113216,
-      "loss": 46.0,
-      "step": 18332
-    },
-    {
-      "epoch": 1.4016858764837434,
-      "grad_norm": 0.0022736121900379658,
-      "learning_rate": 0.00019999903150542166,
-      "loss": 46.0,
-      "step": 18333
-    },
-    {
-      "epoch": 1.4017623334671332,
-      "grad_norm": 0.0006398377008736134,
-      "learning_rate": 0.00019999903139970542,
-      "loss": 46.0,
-      "step": 18334
-    },
-    {
-      "epoch": 1.4018387904505227,
-      "grad_norm": 0.000655961106531322,
-      "learning_rate": 0.0001999990312939834,
-      "loss": 46.0,
-      "step": 18335
-    },
-    {
-      "epoch": 1.4019152474339125,
-      "grad_norm": 0.001291886786930263,
-      "learning_rate": 0.0001999990311882556,
-      "loss": 46.0,
-      "step": 18336
-    },
-    {
-      "epoch": 1.4019917044173023,
-      "grad_norm": 0.002854206133633852,
-      "learning_rate": 0.00019999903108252205,
-      "loss": 46.0,
-      "step": 18337
-    },
-    {
-      "epoch": 1.402068161400692,
-      "grad_norm": 0.001123277936130762,
-      "learning_rate": 0.0001999990309767827,
-      "loss": 46.0,
-      "step": 18338
-    },
-    {
-      "epoch": 1.4021446183840816,
-      "grad_norm": 0.0006095334538258612,
-      "learning_rate": 0.00019999903087103763,
-      "loss": 46.0,
-      "step": 18339
-    },
-    {
-      "epoch": 1.4022210753674713,
-      "grad_norm": 0.0033795267809182405,
-      "learning_rate": 0.00019999903076528677,
-      "loss": 46.0,
-      "step": 18340
-    },
-    {
-      "epoch": 1.402297532350861,
-      "grad_norm": 0.00105930189602077,
-      "learning_rate": 0.0001999990306595301,
-      "loss": 46.0,
-      "step": 18341
-    },
-    {
-      "epoch": 1.4023739893342508,
-      "grad_norm": 0.01124873198568821,
-      "learning_rate": 0.0001999990305537677,
-      "loss": 46.0,
-      "step": 18342
-    },
-    {
-      "epoch": 1.4024504463176406,
-      "grad_norm": 0.0012958410661667585,
-      "learning_rate": 0.00019999903044799953,
-      "loss": 46.0,
-      "step": 18343
-    },
-    {
-      "epoch": 1.4025269033010304,
-      "grad_norm": 0.002868752693757415,
-      "learning_rate": 0.00019999903034222558,
-      "loss": 46.0,
-      "step": 18344
-    },
-    {
-      "epoch": 1.40260336028442,
-      "grad_norm": 0.0014347676187753677,
-      "learning_rate": 0.00019999903023644588,
-      "loss": 46.0,
-      "step": 18345
-    },
-    {
-      "epoch": 1.4026798172678097,
-      "grad_norm": 0.00029974826611578465,
-      "learning_rate": 0.00019999903013066036,
-      "loss": 46.0,
-      "step": 18346
-    },
-    {
-      "epoch": 1.4027562742511994,
-      "grad_norm": 0.0026474210899323225,
-      "learning_rate": 0.00019999903002486914,
-      "loss": 46.0,
-      "step": 18347
-    },
-    {
-      "epoch": 1.4028327312345892,
-      "grad_norm": 0.0010696279350668192,
-      "learning_rate": 0.00019999902991907212,
-      "loss": 46.0,
-      "step": 18348
-    },
-    {
-      "epoch": 1.402909188217979,
-      "grad_norm": 0.0006355736986733973,
-      "learning_rate": 0.00019999902981326933,
-      "loss": 46.0,
-      "step": 18349
-    },
-    {
-      "epoch": 1.4029856452013685,
-      "grad_norm": 0.0008076782687567174,
-      "learning_rate": 0.00019999902970746077,
-      "loss": 46.0,
-      "step": 18350
-    },
-    {
-      "epoch": 1.4030621021847582,
-      "grad_norm": 0.0005658628651872277,
-      "learning_rate": 0.00019999902960164643,
-      "loss": 46.0,
-      "step": 18351
-    },
-    {
-      "epoch": 1.403138559168148,
-      "grad_norm": 0.0008830587030388415,
-      "learning_rate": 0.00019999902949582632,
-      "loss": 46.0,
-      "step": 18352
-    },
-    {
-      "epoch": 1.4032150161515378,
-      "grad_norm": 0.0007348970975726843,
-      "learning_rate": 0.00019999902939000044,
-      "loss": 46.0,
-      "step": 18353
-    },
-    {
-      "epoch": 1.4032914731349275,
-      "grad_norm": 0.0007917945040389895,
-      "learning_rate": 0.0001999990292841688,
-      "loss": 46.0,
-      "step": 18354
-    },
-    {
-      "epoch": 1.4033679301183173,
-      "grad_norm": 0.0025544478558003902,
-      "learning_rate": 0.0001999990291783314,
-      "loss": 46.0,
-      "step": 18355
-    },
-    {
-      "epoch": 1.4034443871017068,
-      "grad_norm": 0.0004933216259814799,
-      "learning_rate": 0.00019999902907248823,
-      "loss": 46.0,
-      "step": 18356
-    },
-    {
-      "epoch": 1.4035208440850966,
-      "grad_norm": 0.0007026851526461542,
-      "learning_rate": 0.00019999902896663928,
-      "loss": 46.0,
-      "step": 18357
-    },
-    {
-      "epoch": 1.4035973010684863,
-      "grad_norm": 0.0010201854165643454,
-      "learning_rate": 0.00019999902886078458,
-      "loss": 46.0,
-      "step": 18358
-    },
-    {
-      "epoch": 1.403673758051876,
-      "grad_norm": 0.0006736062350682914,
-      "learning_rate": 0.00019999902875492405,
-      "loss": 46.0,
-      "step": 18359
-    },
-    {
-      "epoch": 1.4037502150352656,
-      "grad_norm": 0.0006356733501888812,
-      "learning_rate": 0.00019999902864905778,
-      "loss": 46.0,
-      "step": 18360
-    },
-    {
-      "epoch": 1.4038266720186554,
-      "grad_norm": 0.0006490937666967511,
-      "learning_rate": 0.00019999902854318577,
-      "loss": 46.0,
-      "step": 18361
-    },
-    {
-      "epoch": 1.4039031290020452,
-      "grad_norm": 0.0063923317939043045,
-      "learning_rate": 0.00019999902843730798,
-      "loss": 46.0,
-      "step": 18362
-    },
-    {
-      "epoch": 1.403979585985435,
-      "grad_norm": 0.003219634760171175,
-      "learning_rate": 0.0001999990283314244,
-      "loss": 46.0,
-      "step": 18363
-    },
-    {
-      "epoch": 1.4040560429688247,
-      "grad_norm": 0.0013041739584878087,
-      "learning_rate": 0.0001999990282255351,
-      "loss": 46.0,
-      "step": 18364
-    },
-    {
-      "epoch": 1.4041324999522145,
-      "grad_norm": 0.0015903398161754012,
-      "learning_rate": 0.00019999902811964,
-      "loss": 46.0,
-      "step": 18365
-    },
-    {
-      "epoch": 1.4042089569356042,
-      "grad_norm": 0.004123296122997999,
-      "learning_rate": 0.0001999990280137391,
-      "loss": 46.0,
-      "step": 18366
-    },
-    {
-      "epoch": 1.4042854139189938,
-      "grad_norm": 0.0008914932841435075,
-      "learning_rate": 0.00019999902790783248,
-      "loss": 46.0,
-      "step": 18367
-    },
-    {
-      "epoch": 1.4043618709023835,
-      "grad_norm": 0.0006523121846839786,
-      "learning_rate": 0.00019999902780192005,
-      "loss": 46.0,
-      "step": 18368
-    },
-    {
-      "epoch": 1.4044383278857733,
-      "grad_norm": 0.007997400127351284,
-      "learning_rate": 0.00019999902769600187,
-      "loss": 46.0,
-      "step": 18369
-    },
-    {
-      "epoch": 1.404514784869163,
-      "grad_norm": 0.0005990521749481559,
-      "learning_rate": 0.00019999902759007792,
-      "loss": 46.0,
-      "step": 18370
-    },
-    {
-      "epoch": 1.4045912418525526,
-      "grad_norm": 0.0008489358006045222,
-      "learning_rate": 0.0001999990274841482,
-      "loss": 46.0,
-      "step": 18371
-    },
-    {
-      "epoch": 1.4046676988359423,
-      "grad_norm": 0.0013013811549171805,
-      "learning_rate": 0.00019999902737821273,
-      "loss": 46.0,
-      "step": 18372
-    },
-    {
-      "epoch": 1.404744155819332,
-      "grad_norm": 0.003888198174536228,
-      "learning_rate": 0.00019999902727227146,
-      "loss": 46.0,
-      "step": 18373
-    },
-    {
-      "epoch": 1.4048206128027219,
-      "grad_norm": 0.0003828374610748142,
-      "learning_rate": 0.00019999902716632444,
-      "loss": 46.0,
-      "step": 18374
-    },
-    {
-      "epoch": 1.4048970697861116,
-      "grad_norm": 0.0018283678218722343,
-      "learning_rate": 0.00019999902706037162,
-      "loss": 46.0,
-      "step": 18375
-    },
-    {
-      "epoch": 1.4049735267695014,
-      "grad_norm": 0.00234527001157403,
-      "learning_rate": 0.00019999902695441306,
-      "loss": 46.0,
-      "step": 18376
-    },
-    {
-      "epoch": 1.4050499837528911,
-      "grad_norm": 0.001431879703886807,
-      "learning_rate": 0.00019999902684844872,
-      "loss": 46.0,
-      "step": 18377
-    },
-    {
-      "epoch": 1.4051264407362807,
-      "grad_norm": 0.000716458831448108,
-      "learning_rate": 0.00019999902674247864,
-      "loss": 46.0,
-      "step": 18378
-    },
-    {
-      "epoch": 1.4052028977196704,
-      "grad_norm": 0.003407655283808708,
-      "learning_rate": 0.00019999902663650273,
-      "loss": 46.0,
-      "step": 18379
-    },
-    {
-      "epoch": 1.4052793547030602,
-      "grad_norm": 0.0023236097767949104,
-      "learning_rate": 0.00019999902653052113,
-      "loss": 46.0,
-      "step": 18380
-    },
-    {
-      "epoch": 1.40535581168645,
-      "grad_norm": 0.0033848697785288095,
-      "learning_rate": 0.0001999990264245337,
-      "loss": 46.0,
-      "step": 18381
-    },
-    {
-      "epoch": 1.4054322686698395,
-      "grad_norm": 0.0005153603269718587,
-      "learning_rate": 0.0001999990263185405,
-      "loss": 46.0,
-      "step": 18382
-    },
-    {
-      "epoch": 1.4055087256532293,
-      "grad_norm": 0.00100629439111799,
-      "learning_rate": 0.00019999902621254157,
-      "loss": 46.0,
-      "step": 18383
-    },
-    {
-      "epoch": 1.405585182636619,
-      "grad_norm": 0.005232179071754217,
-      "learning_rate": 0.00019999902610653685,
-      "loss": 46.0,
-      "step": 18384
-    },
-    {
-      "epoch": 1.4056616396200088,
-      "grad_norm": 0.00042024534195661545,
-      "learning_rate": 0.00019999902600052635,
-      "loss": 46.0,
-      "step": 18385
-    },
-    {
-      "epoch": 1.4057380966033985,
-      "grad_norm": 0.0026354603469371796,
-      "learning_rate": 0.00019999902589451008,
-      "loss": 46.0,
-      "step": 18386
-    },
-    {
-      "epoch": 1.4058145535867883,
-      "grad_norm": 0.005128634627908468,
-      "learning_rate": 0.00019999902578848807,
-      "loss": 46.0,
-      "step": 18387
-    },
-    {
-      "epoch": 1.405891010570178,
-      "grad_norm": 0.0005527346511371434,
-      "learning_rate": 0.00019999902568246025,
-      "loss": 46.0,
-      "step": 18388
-    },
-    {
-      "epoch": 1.4059674675535676,
-      "grad_norm": 0.0005377604975365102,
-      "learning_rate": 0.00019999902557642666,
-      "loss": 46.0,
-      "step": 18389
-    },
-    {
-      "epoch": 1.4060439245369574,
-      "grad_norm": 0.0015182136557996273,
-      "learning_rate": 0.00019999902547038735,
-      "loss": 46.0,
-      "step": 18390
-    },
-    {
-      "epoch": 1.4061203815203471,
-      "grad_norm": 0.0008125728345476091,
-      "learning_rate": 0.00019999902536434224,
-      "loss": 46.0,
-      "step": 18391
-    },
-    {
-      "epoch": 1.406196838503737,
-      "grad_norm": 0.0009747163858264685,
-      "learning_rate": 0.00019999902525829138,
-      "loss": 46.0,
-      "step": 18392
-    },
-    {
-      "epoch": 1.4062732954871264,
-      "grad_norm": 0.0013950306456536055,
-      "learning_rate": 0.0001999990251522347,
-      "loss": 46.0,
-      "step": 18393
-    },
-    {
-      "epoch": 1.4063497524705162,
-      "grad_norm": 0.0017438828945159912,
-      "learning_rate": 0.0001999990250461723,
-      "loss": 46.0,
-      "step": 18394
-    },
-    {
-      "epoch": 1.406426209453906,
-      "grad_norm": 0.0008534836815670133,
-      "learning_rate": 0.00019999902494010412,
-      "loss": 46.0,
-      "step": 18395
-    },
-    {
-      "epoch": 1.4065026664372957,
-      "grad_norm": 0.0009154222207143903,
-      "learning_rate": 0.00019999902483403017,
-      "loss": 46.0,
-      "step": 18396
-    },
-    {
-      "epoch": 1.4065791234206855,
-      "grad_norm": 0.0009027472697198391,
-      "learning_rate": 0.00019999902472795045,
-      "loss": 46.0,
-      "step": 18397
-    },
-    {
-      "epoch": 1.4066555804040752,
-      "grad_norm": 0.003171935910359025,
-      "learning_rate": 0.00019999902462186495,
-      "loss": 46.0,
-      "step": 18398
-    },
-    {
-      "epoch": 1.406732037387465,
-      "grad_norm": 0.004676215350627899,
-      "learning_rate": 0.00019999902451577368,
-      "loss": 46.0,
-      "step": 18399
-    },
-    {
-      "epoch": 1.4068084943708545,
-      "grad_norm": 0.0008767013205215335,
-      "learning_rate": 0.00019999902440967664,
-      "loss": 46.0,
-      "step": 18400
-    },
-    {
-      "epoch": 1.4068849513542443,
-      "grad_norm": 0.0007084507960826159,
-      "learning_rate": 0.00019999902430357385,
-      "loss": 46.0,
-      "step": 18401
-    },
-    {
-      "epoch": 1.406961408337634,
-      "grad_norm": 0.001167476293630898,
-      "learning_rate": 0.0001999990241974653,
-      "loss": 46.0,
-      "step": 18402
-    },
-    {
-      "epoch": 1.4070378653210238,
-      "grad_norm": 0.00039947484037838876,
-      "learning_rate": 0.00019999902409135092,
-      "loss": 46.0,
-      "step": 18403
-    },
-    {
-      "epoch": 1.4071143223044134,
-      "grad_norm": 0.0010710599599406123,
-      "learning_rate": 0.00019999902398523081,
-      "loss": 46.0,
-      "step": 18404
-    },
-    {
-      "epoch": 1.4071907792878031,
-      "grad_norm": 0.0005583700840361416,
-      "learning_rate": 0.00019999902387910493,
-      "loss": 46.0,
-      "step": 18405
-    },
-    {
-      "epoch": 1.4072672362711929,
-      "grad_norm": 0.0008522698190063238,
-      "learning_rate": 0.00019999902377297328,
-      "loss": 46.0,
-      "step": 18406
-    },
-    {
-      "epoch": 1.4073436932545826,
-      "grad_norm": 0.0007379850139841437,
-      "learning_rate": 0.00019999902366683587,
-      "loss": 46.0,
-      "step": 18407
-    },
-    {
-      "epoch": 1.4074201502379724,
-      "grad_norm": 0.00615641800686717,
-      "learning_rate": 0.0001999990235606927,
-      "loss": 46.0,
-      "step": 18408
-    },
-    {
-      "epoch": 1.4074966072213622,
-      "grad_norm": 0.0008752090507186949,
-      "learning_rate": 0.00019999902345454372,
-      "loss": 46.0,
-      "step": 18409
-    },
-    {
-      "epoch": 1.407573064204752,
-      "grad_norm": 0.0003531481488607824,
-      "learning_rate": 0.000199999023348389,
-      "loss": 46.0,
-      "step": 18410
-    },
-    {
-      "epoch": 1.4076495211881415,
-      "grad_norm": 0.0014633395476266742,
-      "learning_rate": 0.0001999990232422285,
-      "loss": 46.0,
-      "step": 18411
-    },
-    {
-      "epoch": 1.4077259781715312,
-      "grad_norm": 0.0010693443473428488,
-      "learning_rate": 0.00019999902313606226,
-      "loss": 46.0,
-      "step": 18412
-    },
-    {
-      "epoch": 1.407802435154921,
-      "grad_norm": 0.0009467890486121178,
-      "learning_rate": 0.00019999902302989022,
-      "loss": 46.0,
-      "step": 18413
-    },
-    {
-      "epoch": 1.4078788921383107,
-      "grad_norm": 0.0047689140774309635,
-      "learning_rate": 0.0001999990229237124,
-      "loss": 46.0,
-      "step": 18414
-    },
-    {
-      "epoch": 1.4079553491217003,
-      "grad_norm": 0.004703803453594446,
-      "learning_rate": 0.00019999902281752884,
-      "loss": 46.0,
-      "step": 18415
-    },
-    {
-      "epoch": 1.40803180610509,
-      "grad_norm": 0.0007608246523886919,
-      "learning_rate": 0.00019999902271133948,
-      "loss": 46.0,
-      "step": 18416
-    },
-    {
-      "epoch": 1.4081082630884798,
-      "grad_norm": 0.003276802832260728,
-      "learning_rate": 0.0001999990226051444,
-      "loss": 46.0,
-      "step": 18417
-    },
-    {
-      "epoch": 1.4081847200718696,
-      "grad_norm": 0.0008925948641262949,
-      "learning_rate": 0.0001999990224989435,
-      "loss": 46.0,
-      "step": 18418
-    },
-    {
-      "epoch": 1.4082611770552593,
-      "grad_norm": 0.00042892477358691394,
-      "learning_rate": 0.00019999902239273686,
-      "loss": 46.0,
-      "step": 18419
-    },
-    {
-      "epoch": 1.408337634038649,
-      "grad_norm": 0.001327955978922546,
-      "learning_rate": 0.00019999902228652443,
-      "loss": 46.0,
-      "step": 18420
-    },
-    {
-      "epoch": 1.4084140910220388,
-      "grad_norm": 0.0019752285443246365,
-      "learning_rate": 0.00019999902218030625,
-      "loss": 46.0,
-      "step": 18421
-    },
-    {
-      "epoch": 1.4084905480054284,
-      "grad_norm": 0.00046541838673874736,
-      "learning_rate": 0.00019999902207408227,
-      "loss": 46.0,
-      "step": 18422
-    },
-    {
-      "epoch": 1.4085670049888181,
-      "grad_norm": 0.0009978982852771878,
-      "learning_rate": 0.00019999902196785255,
-      "loss": 46.0,
-      "step": 18423
-    },
-    {
-      "epoch": 1.408643461972208,
-      "grad_norm": 0.0006274883635342121,
-      "learning_rate": 0.00019999902186161706,
-      "loss": 46.0,
-      "step": 18424
-    },
-    {
-      "epoch": 1.4087199189555977,
-      "grad_norm": 0.001348936348222196,
-      "learning_rate": 0.00019999902175537576,
-      "loss": 46.0,
-      "step": 18425
-    },
-    {
-      "epoch": 1.4087963759389872,
-      "grad_norm": 0.0012112503172829747,
-      "learning_rate": 0.00019999902164912877,
-      "loss": 46.0,
-      "step": 18426
-    },
-    {
-      "epoch": 1.408872832922377,
-      "grad_norm": 0.001995028229430318,
-      "learning_rate": 0.00019999902154287593,
-      "loss": 46.0,
-      "step": 18427
-    },
-    {
-      "epoch": 1.4089492899057667,
-      "grad_norm": 0.0014687899965792894,
-      "learning_rate": 0.00019999902143661737,
-      "loss": 46.0,
-      "step": 18428
-    },
-    {
-      "epoch": 1.4090257468891565,
-      "grad_norm": 0.0006656462792307138,
-      "learning_rate": 0.00019999902133035303,
-      "loss": 46.0,
-      "step": 18429
-    },
-    {
-      "epoch": 1.4091022038725463,
-      "grad_norm": 0.0026355283334851265,
-      "learning_rate": 0.0001999990212240829,
-      "loss": 46.0,
-      "step": 18430
-    },
-    {
-      "epoch": 1.409178660855936,
-      "grad_norm": 0.0028784542810171843,
-      "learning_rate": 0.00019999902111780701,
-      "loss": 46.0,
-      "step": 18431
-    },
-    {
-      "epoch": 1.4092551178393258,
-      "grad_norm": 0.0012396732345223427,
-      "learning_rate": 0.00019999902101152536,
-      "loss": 46.0,
-      "step": 18432
-    },
-    {
-      "epoch": 1.4093315748227153,
-      "grad_norm": 0.0003747464215848595,
-      "learning_rate": 0.00019999902090523796,
-      "loss": 46.0,
-      "step": 18433
-    },
-    {
-      "epoch": 1.409408031806105,
-      "grad_norm": 0.001326856086961925,
-      "learning_rate": 0.00019999902079894475,
-      "loss": 46.0,
-      "step": 18434
-    },
-    {
-      "epoch": 1.4094844887894948,
-      "grad_norm": 0.0017358335899189115,
-      "learning_rate": 0.00019999902069264578,
-      "loss": 46.0,
-      "step": 18435
-    },
-    {
-      "epoch": 1.4095609457728846,
-      "grad_norm": 0.0005002904217690229,
-      "learning_rate": 0.00019999902058634106,
-      "loss": 46.0,
-      "step": 18436
-    },
-    {
-      "epoch": 1.4096374027562741,
-      "grad_norm": 0.000988552812486887,
-      "learning_rate": 0.00019999902048003056,
-      "loss": 46.0,
-      "step": 18437
-    },
-    {
-      "epoch": 1.409713859739664,
-      "grad_norm": 0.000578157021664083,
-      "learning_rate": 0.0001999990203737143,
-      "loss": 46.0,
-      "step": 18438
-    },
-    {
-      "epoch": 1.4097903167230537,
-      "grad_norm": 0.0007497709011659026,
-      "learning_rate": 0.00019999902026739225,
-      "loss": 46.0,
-      "step": 18439
-    },
-    {
-      "epoch": 1.4098667737064434,
-      "grad_norm": 0.0009373456705361605,
-      "learning_rate": 0.00019999902016106444,
-      "loss": 46.0,
-      "step": 18440
-    },
-    {
-      "epoch": 1.4099432306898332,
-      "grad_norm": 0.0007314086542464793,
-      "learning_rate": 0.00019999902005473085,
-      "loss": 46.0,
-      "step": 18441
-    },
-    {
-      "epoch": 1.410019687673223,
-      "grad_norm": 0.0011600431753322482,
-      "learning_rate": 0.0001999990199483915,
-      "loss": 46.0,
-      "step": 18442
-    },
-    {
-      "epoch": 1.4100961446566127,
-      "grad_norm": 0.00025432309485040605,
-      "learning_rate": 0.0001999990198420464,
-      "loss": 46.0,
-      "step": 18443
-    },
-    {
-      "epoch": 1.4101726016400022,
-      "grad_norm": 0.023697514086961746,
-      "learning_rate": 0.00019999901973569552,
-      "loss": 46.0,
-      "step": 18444
-    },
-    {
-      "epoch": 1.410249058623392,
-      "grad_norm": 0.0023732506670057774,
-      "learning_rate": 0.00019999901962933887,
-      "loss": 46.0,
-      "step": 18445
-    },
-    {
-      "epoch": 1.4103255156067818,
-      "grad_norm": 0.0023167117033153772,
-      "learning_rate": 0.00019999901952297644,
-      "loss": 46.0,
-      "step": 18446
-    },
-    {
-      "epoch": 1.4104019725901715,
-      "grad_norm": 0.00045706008677370846,
-      "learning_rate": 0.00019999901941660823,
-      "loss": 46.0,
-      "step": 18447
-    },
-    {
-      "epoch": 1.410478429573561,
-      "grad_norm": 0.0009085984784178436,
-      "learning_rate": 0.0001999990193102343,
-      "loss": 46.0,
-      "step": 18448
-    },
-    {
-      "epoch": 1.4105548865569508,
-      "grad_norm": 0.010532423853874207,
-      "learning_rate": 0.00019999901920385456,
-      "loss": 46.0,
-      "step": 18449
-    },
-    {
-      "epoch": 1.4106313435403406,
-      "grad_norm": 0.001632206724025309,
-      "learning_rate": 0.00019999901909746904,
-      "loss": 46.0,
-      "step": 18450
-    },
-    {
-      "epoch": 1.4107078005237303,
-      "grad_norm": 0.0016392648685723543,
-      "learning_rate": 0.00019999901899107777,
-      "loss": 46.0,
-      "step": 18451
-    },
-    {
-      "epoch": 1.41078425750712,
-      "grad_norm": 0.002026929520070553,
-      "learning_rate": 0.00019999901888468073,
-      "loss": 46.0,
-      "step": 18452
-    },
-    {
-      "epoch": 1.4108607144905099,
-      "grad_norm": 0.0008087048772722483,
-      "learning_rate": 0.00019999901877827792,
-      "loss": 46.0,
-      "step": 18453
-    },
-    {
-      "epoch": 1.4109371714738996,
-      "grad_norm": 0.002954391296952963,
-      "learning_rate": 0.00019999901867186936,
-      "loss": 46.0,
-      "step": 18454
-    },
-    {
-      "epoch": 1.4110136284572892,
-      "grad_norm": 0.004516988061368465,
-      "learning_rate": 0.000199999018565455,
-      "loss": 46.0,
-      "step": 18455
-    },
-    {
-      "epoch": 1.411090085440679,
-      "grad_norm": 0.003901523072272539,
-      "learning_rate": 0.00019999901845903489,
-      "loss": 46.0,
-      "step": 18456
-    },
-    {
-      "epoch": 1.4111665424240687,
-      "grad_norm": 0.0011401647934690118,
-      "learning_rate": 0.000199999018352609,
-      "loss": 46.0,
-      "step": 18457
-    },
-    {
-      "epoch": 1.4112429994074585,
-      "grad_norm": 0.0014308291720226407,
-      "learning_rate": 0.00019999901824617732,
-      "loss": 46.0,
-      "step": 18458
-    },
-    {
-      "epoch": 1.411319456390848,
-      "grad_norm": 0.004948511719703674,
-      "learning_rate": 0.0001999990181397399,
-      "loss": 46.0,
-      "step": 18459
-    },
-    {
-      "epoch": 1.4113959133742378,
-      "grad_norm": 0.000722626515198499,
-      "learning_rate": 0.0001999990180332967,
-      "loss": 46.0,
-      "step": 18460
-    },
-    {
-      "epoch": 1.4114723703576275,
-      "grad_norm": 0.0021766405552625656,
-      "learning_rate": 0.00019999901792684774,
-      "loss": 46.0,
-      "step": 18461
-    },
-    {
-      "epoch": 1.4115488273410173,
-      "grad_norm": 0.0012948965886607766,
-      "learning_rate": 0.000199999017820393,
-      "loss": 46.0,
-      "step": 18462
-    },
-    {
-      "epoch": 1.411625284324407,
-      "grad_norm": 0.00854575913399458,
-      "learning_rate": 0.0001999990177139325,
-      "loss": 46.0,
-      "step": 18463
-    },
-    {
-      "epoch": 1.4117017413077968,
-      "grad_norm": 0.0036632472183555365,
-      "learning_rate": 0.00019999901760746623,
-      "loss": 46.0,
-      "step": 18464
-    },
-    {
-      "epoch": 1.4117781982911866,
-      "grad_norm": 0.0010747485794126987,
-      "learning_rate": 0.0001999990175009942,
-      "loss": 46.0,
-      "step": 18465
-    },
-    {
-      "epoch": 1.411854655274576,
-      "grad_norm": 0.004225791897624731,
-      "learning_rate": 0.00019999901739451638,
-      "loss": 46.0,
-      "step": 18466
-    },
-    {
-      "epoch": 1.4119311122579659,
-      "grad_norm": 0.001447275746613741,
-      "learning_rate": 0.0001999990172880328,
-      "loss": 46.0,
-      "step": 18467
-    },
-    {
-      "epoch": 1.4120075692413556,
-      "grad_norm": 0.001079175970517099,
-      "learning_rate": 0.00019999901718154343,
-      "loss": 46.0,
-      "step": 18468
-    },
-    {
-      "epoch": 1.4120840262247454,
-      "grad_norm": 0.0010275662643834949,
-      "learning_rate": 0.00019999901707504832,
-      "loss": 46.0,
-      "step": 18469
-    },
-    {
-      "epoch": 1.412160483208135,
-      "grad_norm": 0.0005107459146529436,
-      "learning_rate": 0.0001999990169685474,
-      "loss": 46.0,
-      "step": 18470
-    },
-    {
-      "epoch": 1.4122369401915247,
-      "grad_norm": 0.0029556278605014086,
-      "learning_rate": 0.00019999901686204076,
-      "loss": 46.0,
-      "step": 18471
-    },
-    {
-      "epoch": 1.4123133971749144,
-      "grad_norm": 0.0016287070466205478,
-      "learning_rate": 0.00019999901675552836,
-      "loss": 46.0,
-      "step": 18472
-    },
-    {
-      "epoch": 1.4123898541583042,
-      "grad_norm": 0.0010849777609109879,
-      "learning_rate": 0.00019999901664901013,
-      "loss": 46.0,
-      "step": 18473
-    },
-    {
-      "epoch": 1.412466311141694,
-      "grad_norm": 0.0010581731330603361,
-      "learning_rate": 0.00019999901654248618,
-      "loss": 46.0,
-      "step": 18474
-    },
-    {
-      "epoch": 1.4125427681250837,
-      "grad_norm": 0.005207499954849482,
-      "learning_rate": 0.00019999901643595643,
-      "loss": 46.0,
-      "step": 18475
-    },
-    {
-      "epoch": 1.4126192251084733,
-      "grad_norm": 0.01932896487414837,
-      "learning_rate": 0.00019999901632942094,
-      "loss": 46.0,
-      "step": 18476
-    },
-    {
-      "epoch": 1.412695682091863,
-      "grad_norm": 0.0028620220255106688,
-      "learning_rate": 0.00019999901622287967,
-      "loss": 46.0,
-      "step": 18477
-    },
-    {
-      "epoch": 1.4127721390752528,
-      "grad_norm": 0.0013349989894777536,
-      "learning_rate": 0.0001999990161163326,
-      "loss": 46.0,
-      "step": 18478
-    },
-    {
-      "epoch": 1.4128485960586425,
-      "grad_norm": 0.0021573402918875217,
-      "learning_rate": 0.00019999901600977979,
-      "loss": 46.0,
-      "step": 18479
-    },
-    {
-      "epoch": 1.4129250530420323,
-      "grad_norm": 0.0010723735904321074,
-      "learning_rate": 0.0001999990159032212,
-      "loss": 46.0,
-      "step": 18480
-    },
-    {
-      "epoch": 1.4130015100254218,
-      "grad_norm": 0.00990714505314827,
-      "learning_rate": 0.00019999901579665686,
-      "loss": 46.0,
-      "step": 18481
-    },
-    {
-      "epoch": 1.4130779670088116,
-      "grad_norm": 0.0011306923115625978,
-      "learning_rate": 0.00019999901569008676,
-      "loss": 46.0,
-      "step": 18482
-    },
-    {
-      "epoch": 1.4131544239922014,
-      "grad_norm": 0.0011009775334969163,
-      "learning_rate": 0.00019999901558351085,
-      "loss": 46.0,
-      "step": 18483
-    },
-    {
-      "epoch": 1.4132308809755911,
-      "grad_norm": 0.0013946243561804295,
-      "learning_rate": 0.00019999901547692917,
-      "loss": 46.0,
-      "step": 18484
-    },
-    {
-      "epoch": 1.413307337958981,
-      "grad_norm": 0.0011130532948300242,
-      "learning_rate": 0.00019999901537034177,
-      "loss": 46.0,
-      "step": 18485
-    },
-    {
-      "epoch": 1.4133837949423707,
-      "grad_norm": 0.0018272692104801536,
-      "learning_rate": 0.00019999901526374854,
-      "loss": 46.0,
-      "step": 18486
-    },
-    {
-      "epoch": 1.4134602519257602,
-      "grad_norm": 0.001515660434961319,
-      "learning_rate": 0.0001999990151571496,
-      "loss": 46.0,
-      "step": 18487
-    },
-    {
-      "epoch": 1.41353670890915,
-      "grad_norm": 0.0005353334126994014,
-      "learning_rate": 0.00019999901505054484,
-      "loss": 46.0,
-      "step": 18488
-    },
-    {
-      "epoch": 1.4136131658925397,
-      "grad_norm": 0.0019273837096989155,
-      "learning_rate": 0.00019999901494393432,
-      "loss": 46.0,
-      "step": 18489
-    },
-    {
-      "epoch": 1.4136896228759295,
-      "grad_norm": 0.003083708230406046,
-      "learning_rate": 0.00019999901483731805,
-      "loss": 46.0,
-      "step": 18490
-    },
-    {
-      "epoch": 1.413766079859319,
-      "grad_norm": 0.001123978290706873,
-      "learning_rate": 0.00019999901473069599,
-      "loss": 46.0,
-      "step": 18491
-    },
-    {
-      "epoch": 1.4138425368427088,
-      "grad_norm": 0.0005934460205025971,
-      "learning_rate": 0.00019999901462406817,
-      "loss": 46.0,
-      "step": 18492
-    },
-    {
-      "epoch": 1.4139189938260985,
-      "grad_norm": 0.0016057519242167473,
-      "learning_rate": 0.00019999901451743458,
-      "loss": 46.0,
-      "step": 18493
-    },
-    {
-      "epoch": 1.4139954508094883,
-      "grad_norm": 0.0028722987044602633,
-      "learning_rate": 0.00019999901441079525,
-      "loss": 46.0,
-      "step": 18494
-    },
-    {
-      "epoch": 1.414071907792878,
-      "grad_norm": 0.0031471208203583956,
-      "learning_rate": 0.00019999901430415014,
-      "loss": 46.0,
-      "step": 18495
-    },
-    {
-      "epoch": 1.4141483647762678,
-      "grad_norm": 0.0006638827035203576,
-      "learning_rate": 0.00019999901419749924,
-      "loss": 46.0,
-      "step": 18496
-    },
-    {
-      "epoch": 1.4142248217596576,
-      "grad_norm": 0.0016214716015383601,
-      "learning_rate": 0.00019999901409084255,
-      "loss": 46.0,
-      "step": 18497
-    },
-    {
-      "epoch": 1.4143012787430471,
-      "grad_norm": 0.001752912299707532,
-      "learning_rate": 0.00019999901398418013,
-      "loss": 46.0,
-      "step": 18498
-    },
-    {
-      "epoch": 1.4143777357264369,
-      "grad_norm": 0.0023880701046437025,
-      "learning_rate": 0.00019999901387751195,
-      "loss": 46.0,
-      "step": 18499
-    },
-    {
-      "epoch": 1.4144541927098266,
-      "grad_norm": 0.0004587325674947351,
-      "learning_rate": 0.00019999901377083795,
-      "loss": 46.0,
-      "step": 18500
-    },
-    {
-      "epoch": 1.4145306496932164,
-      "grad_norm": 0.003012463916093111,
-      "learning_rate": 0.0001999990136641582,
-      "loss": 46.0,
-      "step": 18501
-    },
-    {
-      "epoch": 1.414607106676606,
-      "grad_norm": 0.0010055694729089737,
-      "learning_rate": 0.0001999990135574727,
-      "loss": 46.0,
-      "step": 18502
-    },
-    {
-      "epoch": 1.4146835636599957,
-      "grad_norm": 0.0007959576905705035,
-      "learning_rate": 0.00019999901345078142,
-      "loss": 46.0,
-      "step": 18503
-    },
-    {
-      "epoch": 1.4147600206433855,
-      "grad_norm": 0.0008114899392239749,
-      "learning_rate": 0.00019999901334408438,
-      "loss": 46.0,
-      "step": 18504
-    },
-    {
-      "epoch": 1.4148364776267752,
-      "grad_norm": 0.003362583927810192,
-      "learning_rate": 0.00019999901323738156,
-      "loss": 46.0,
-      "step": 18505
-    },
-    {
-      "epoch": 1.414912934610165,
-      "grad_norm": 0.0010398655431345105,
-      "learning_rate": 0.00019999901313067295,
-      "loss": 46.0,
-      "step": 18506
-    },
-    {
-      "epoch": 1.4149893915935547,
-      "grad_norm": 0.0007882789359427989,
-      "learning_rate": 0.00019999901302395861,
-      "loss": 46.0,
-      "step": 18507
-    },
-    {
-      "epoch": 1.4150658485769445,
-      "grad_norm": 0.001123824156820774,
-      "learning_rate": 0.00019999901291723848,
-      "loss": 46.0,
-      "step": 18508
-    },
-    {
-      "epoch": 1.415142305560334,
-      "grad_norm": 0.0004243047733325511,
-      "learning_rate": 0.0001999990128105126,
-      "loss": 46.0,
-      "step": 18509
-    },
-    {
-      "epoch": 1.4152187625437238,
-      "grad_norm": 0.004101932048797607,
-      "learning_rate": 0.00019999901270378092,
-      "loss": 46.0,
-      "step": 18510
-    },
-    {
-      "epoch": 1.4152952195271136,
-      "grad_norm": 0.01726161688566208,
-      "learning_rate": 0.00019999901259704347,
-      "loss": 46.0,
-      "step": 18511
-    },
-    {
-      "epoch": 1.4153716765105033,
-      "grad_norm": 0.000661838857922703,
-      "learning_rate": 0.00019999901249030026,
-      "loss": 46.0,
-      "step": 18512
-    },
-    {
-      "epoch": 1.4154481334938929,
-      "grad_norm": 0.0009470888180658221,
-      "learning_rate": 0.0001999990123835513,
-      "loss": 46.0,
-      "step": 18513
-    },
-    {
-      "epoch": 1.4155245904772826,
-      "grad_norm": 0.0024587218649685383,
-      "learning_rate": 0.00019999901227679657,
-      "loss": 46.0,
-      "step": 18514
-    },
-    {
-      "epoch": 1.4156010474606724,
-      "grad_norm": 0.001808356842957437,
-      "learning_rate": 0.00019999901217003605,
-      "loss": 46.0,
-      "step": 18515
-    },
-    {
-      "epoch": 1.4156775044440622,
-      "grad_norm": 0.0018504871986806393,
-      "learning_rate": 0.00019999901206326976,
-      "loss": 46.0,
-      "step": 18516
-    },
-    {
-      "epoch": 1.415753961427452,
-      "grad_norm": 0.002093084156513214,
-      "learning_rate": 0.00019999901195649772,
-      "loss": 46.0,
-      "step": 18517
-    },
-    {
-      "epoch": 1.4158304184108417,
-      "grad_norm": 0.0016352457460016012,
-      "learning_rate": 0.00019999901184971988,
-      "loss": 46.0,
-      "step": 18518
-    },
-    {
-      "epoch": 1.4159068753942314,
-      "grad_norm": 0.0013365078484639525,
-      "learning_rate": 0.00019999901174293632,
-      "loss": 46.0,
-      "step": 18519
-    },
-    {
-      "epoch": 1.415983332377621,
-      "grad_norm": 0.0006704757106490433,
-      "learning_rate": 0.00019999901163614693,
-      "loss": 46.0,
-      "step": 18520
-    },
-    {
-      "epoch": 1.4160597893610107,
-      "grad_norm": 0.0017654619878157973,
-      "learning_rate": 0.00019999901152935182,
-      "loss": 46.0,
-      "step": 18521
-    },
-    {
-      "epoch": 1.4161362463444005,
-      "grad_norm": 0.0015793463680893183,
-      "learning_rate": 0.00019999901142255091,
-      "loss": 46.0,
-      "step": 18522
-    },
-    {
-      "epoch": 1.4162127033277903,
-      "grad_norm": 0.0007882977952249348,
-      "learning_rate": 0.00019999901131574426,
-      "loss": 46.0,
-      "step": 18523
-    },
-    {
-      "epoch": 1.4162891603111798,
-      "grad_norm": 0.0008499923278577626,
-      "learning_rate": 0.0001999990112089318,
-      "loss": 46.0,
-      "step": 18524
-    },
-    {
-      "epoch": 1.4163656172945696,
-      "grad_norm": 0.004672562703490257,
-      "learning_rate": 0.00019999901110211358,
-      "loss": 46.0,
-      "step": 18525
-    },
-    {
-      "epoch": 1.4164420742779593,
-      "grad_norm": 0.0005966693279333413,
-      "learning_rate": 0.0001999990109952896,
-      "loss": 46.0,
-      "step": 18526
-    },
-    {
-      "epoch": 1.416518531261349,
-      "grad_norm": 0.0005770426942035556,
-      "learning_rate": 0.00019999901088845986,
-      "loss": 46.0,
-      "step": 18527
-    },
-    {
-      "epoch": 1.4165949882447388,
-      "grad_norm": 0.0002567377232480794,
-      "learning_rate": 0.00019999901078162434,
-      "loss": 46.0,
-      "step": 18528
-    },
-    {
-      "epoch": 1.4166714452281286,
-      "grad_norm": 0.0005484905559569597,
-      "learning_rate": 0.00019999901067478307,
-      "loss": 46.0,
-      "step": 18529
-    },
-    {
-      "epoch": 1.4167479022115184,
-      "grad_norm": 0.00044046645052731037,
-      "learning_rate": 0.000199999010567936,
-      "loss": 46.0,
-      "step": 18530
-    },
-    {
-      "epoch": 1.416824359194908,
-      "grad_norm": 0.0013616152573376894,
-      "learning_rate": 0.0001999990104610832,
-      "loss": 46.0,
-      "step": 18531
-    },
-    {
-      "epoch": 1.4169008161782977,
-      "grad_norm": 0.0009252128074876964,
-      "learning_rate": 0.00019999901035422458,
-      "loss": 46.0,
-      "step": 18532
-    },
-    {
-      "epoch": 1.4169772731616874,
-      "grad_norm": 0.0017123956931754947,
-      "learning_rate": 0.00019999901024736025,
-      "loss": 46.0,
-      "step": 18533
-    },
-    {
-      "epoch": 1.4170537301450772,
-      "grad_norm": 0.0014507912565022707,
-      "learning_rate": 0.0001999990101404901,
-      "loss": 46.0,
-      "step": 18534
-    },
-    {
-      "epoch": 1.4171301871284667,
-      "grad_norm": 0.00126954959705472,
-      "learning_rate": 0.0001999990100336142,
-      "loss": 46.0,
-      "step": 18535
-    },
-    {
-      "epoch": 1.4172066441118565,
-      "grad_norm": 0.003204815788194537,
-      "learning_rate": 0.00019999900992673255,
-      "loss": 46.0,
-      "step": 18536
-    },
-    {
-      "epoch": 1.4172831010952462,
-      "grad_norm": 0.002077037002891302,
-      "learning_rate": 0.0001999990098198451,
-      "loss": 46.0,
-      "step": 18537
-    },
-    {
-      "epoch": 1.417359558078636,
-      "grad_norm": 0.000843019166495651,
-      "learning_rate": 0.0001999990097129519,
-      "loss": 46.0,
-      "step": 18538
-    },
-    {
-      "epoch": 1.4174360150620258,
-      "grad_norm": 0.0029420412611216307,
-      "learning_rate": 0.0001999990096060529,
-      "loss": 46.0,
-      "step": 18539
-    },
-    {
-      "epoch": 1.4175124720454155,
-      "grad_norm": 0.0011912168702110648,
-      "learning_rate": 0.00019999900949914815,
-      "loss": 46.0,
-      "step": 18540
-    },
-    {
-      "epoch": 1.4175889290288053,
-      "grad_norm": 0.0011484661372378469,
-      "learning_rate": 0.00019999900939223766,
-      "loss": 46.0,
-      "step": 18541
-    },
-    {
-      "epoch": 1.4176653860121948,
-      "grad_norm": 0.002181977266445756,
-      "learning_rate": 0.00019999900928532137,
-      "loss": 46.0,
-      "step": 18542
-    },
-    {
-      "epoch": 1.4177418429955846,
-      "grad_norm": 0.002635022159665823,
-      "learning_rate": 0.0001999990091783993,
-      "loss": 46.0,
-      "step": 18543
-    },
-    {
-      "epoch": 1.4178182999789744,
-      "grad_norm": 0.004045007284730673,
-      "learning_rate": 0.0001999990090714715,
-      "loss": 46.0,
-      "step": 18544
-    },
-    {
-      "epoch": 1.4178947569623641,
-      "grad_norm": 0.0023879518266767263,
-      "learning_rate": 0.00019999900896453787,
-      "loss": 46.0,
-      "step": 18545
-    },
-    {
-      "epoch": 1.4179712139457537,
-      "grad_norm": 0.011243748478591442,
-      "learning_rate": 0.0001999990088575985,
-      "loss": 46.0,
-      "step": 18546
-    },
-    {
-      "epoch": 1.4180476709291434,
-      "grad_norm": 0.0013129001017659903,
-      "learning_rate": 0.00019999900875065338,
-      "loss": 46.0,
-      "step": 18547
-    },
-    {
-      "epoch": 1.4181241279125332,
-      "grad_norm": 0.0005883431294932961,
-      "learning_rate": 0.00019999900864370247,
-      "loss": 46.0,
-      "step": 18548
-    },
-    {
-      "epoch": 1.418200584895923,
-      "grad_norm": 0.0014756311429664493,
-      "learning_rate": 0.0001999990085367458,
-      "loss": 46.0,
-      "step": 18549
-    },
-    {
-      "epoch": 1.4182770418793127,
-      "grad_norm": 0.0028583332896232605,
-      "learning_rate": 0.00019999900842978337,
-      "loss": 46.0,
-      "step": 18550
-    },
-    {
-      "epoch": 1.4183534988627025,
-      "grad_norm": 0.0009094447013922036,
-      "learning_rate": 0.00019999900832281514,
-      "loss": 46.0,
-      "step": 18551
-    },
-    {
-      "epoch": 1.4184299558460922,
-      "grad_norm": 0.0030003993306308985,
-      "learning_rate": 0.00019999900821584117,
-      "loss": 46.0,
-      "step": 18552
-    },
-    {
-      "epoch": 1.4185064128294818,
-      "grad_norm": 0.001980611588805914,
-      "learning_rate": 0.00019999900810886142,
-      "loss": 46.0,
-      "step": 18553
-    },
-    {
-      "epoch": 1.4185828698128715,
-      "grad_norm": 0.0010967858834192157,
-      "learning_rate": 0.00019999900800187593,
-      "loss": 46.0,
-      "step": 18554
-    },
-    {
-      "epoch": 1.4186593267962613,
-      "grad_norm": 0.014430834911763668,
-      "learning_rate": 0.0001999990078948846,
-      "loss": 46.0,
-      "step": 18555
-    },
-    {
-      "epoch": 1.418735783779651,
-      "grad_norm": 0.0009640708449296653,
-      "learning_rate": 0.00019999900778788757,
-      "loss": 46.0,
-      "step": 18556
-    },
-    {
-      "epoch": 1.4188122407630406,
-      "grad_norm": 0.0006808165344409645,
-      "learning_rate": 0.00019999900768088473,
-      "loss": 46.0,
-      "step": 18557
-    },
-    {
-      "epoch": 1.4188886977464303,
-      "grad_norm": 0.002339055296033621,
-      "learning_rate": 0.00019999900757387612,
-      "loss": 46.0,
-      "step": 18558
-    },
-    {
-      "epoch": 1.41896515472982,
-      "grad_norm": 0.008080870844423771,
-      "learning_rate": 0.00019999900746686176,
-      "loss": 46.0,
-      "step": 18559
-    },
-    {
-      "epoch": 1.4190416117132099,
-      "grad_norm": 0.0031779950950294733,
-      "learning_rate": 0.00019999900735984163,
-      "loss": 46.0,
-      "step": 18560
-    },
-    {
-      "epoch": 1.4191180686965996,
-      "grad_norm": 0.0008830921724438667,
-      "learning_rate": 0.00019999900725281572,
-      "loss": 46.0,
-      "step": 18561
-    },
-    {
-      "epoch": 1.4191945256799894,
-      "grad_norm": 0.00106416130438447,
-      "learning_rate": 0.00019999900714578407,
-      "loss": 46.0,
-      "step": 18562
-    },
-    {
-      "epoch": 1.4192709826633791,
-      "grad_norm": 0.0028730027843266726,
-      "learning_rate": 0.00019999900703874661,
-      "loss": 46.0,
-      "step": 18563
-    },
-    {
-      "epoch": 1.4193474396467687,
-      "grad_norm": 0.0025810750667005777,
-      "learning_rate": 0.0001999990069317034,
-      "loss": 46.0,
-      "step": 18564
-    },
-    {
-      "epoch": 1.4194238966301584,
-      "grad_norm": 0.004935820586979389,
-      "learning_rate": 0.0001999990068246544,
-      "loss": 46.0,
-      "step": 18565
-    },
-    {
-      "epoch": 1.4195003536135482,
-      "grad_norm": 0.001362173818051815,
-      "learning_rate": 0.00019999900671759967,
-      "loss": 46.0,
-      "step": 18566
-    },
-    {
-      "epoch": 1.419576810596938,
-      "grad_norm": 0.0067724683322012424,
-      "learning_rate": 0.00019999900661053915,
-      "loss": 46.0,
-      "step": 18567
-    },
-    {
-      "epoch": 1.4196532675803275,
-      "grad_norm": 0.003695362014696002,
-      "learning_rate": 0.00019999900650347286,
-      "loss": 46.0,
-      "step": 18568
-    },
-    {
-      "epoch": 1.4197297245637173,
-      "grad_norm": 0.0011306408559903502,
-      "learning_rate": 0.0001999990063964008,
-      "loss": 46.0,
-      "step": 18569
-    },
-    {
-      "epoch": 1.419806181547107,
-      "grad_norm": 0.0008153924718499184,
-      "learning_rate": 0.00019999900628932295,
-      "loss": 46.0,
-      "step": 18570
-    },
-    {
-      "epoch": 1.4198826385304968,
-      "grad_norm": 0.0007796114659868181,
-      "learning_rate": 0.00019999900618223934,
-      "loss": 46.0,
-      "step": 18571
-    },
-    {
-      "epoch": 1.4199590955138865,
-      "grad_norm": 0.0006157926400192082,
-      "learning_rate": 0.00019999900607514998,
-      "loss": 46.0,
-      "step": 18572
-    },
-    {
-      "epoch": 1.4200355524972763,
-      "grad_norm": 0.0008048530435189605,
-      "learning_rate": 0.00019999900596805485,
-      "loss": 46.0,
-      "step": 18573
-    },
-    {
-      "epoch": 1.420112009480666,
-      "grad_norm": 0.0011103039141744375,
-      "learning_rate": 0.00019999900586095394,
-      "loss": 46.0,
-      "step": 18574
-    },
-    {
-      "epoch": 1.4201884664640556,
-      "grad_norm": 0.0008789126295596361,
-      "learning_rate": 0.00019999900575384726,
-      "loss": 46.0,
-      "step": 18575
-    },
-    {
-      "epoch": 1.4202649234474454,
-      "grad_norm": 0.002710483269765973,
-      "learning_rate": 0.0001999990056467348,
-      "loss": 46.0,
-      "step": 18576
-    },
-    {
-      "epoch": 1.4203413804308351,
-      "grad_norm": 0.0008579171262681484,
-      "learning_rate": 0.00019999900553961658,
-      "loss": 46.0,
-      "step": 18577
-    },
-    {
-      "epoch": 1.420417837414225,
-      "grad_norm": 0.0021600141189992428,
-      "learning_rate": 0.0001999990054324926,
-      "loss": 46.0,
-      "step": 18578
-    },
-    {
-      "epoch": 1.4204942943976144,
-      "grad_norm": 0.001998534658923745,
-      "learning_rate": 0.00019999900532536287,
-      "loss": 46.0,
-      "step": 18579
-    },
-    {
-      "epoch": 1.4205707513810042,
-      "grad_norm": 0.007283096667379141,
-      "learning_rate": 0.00019999900521822735,
-      "loss": 46.0,
-      "step": 18580
-    },
-    {
-      "epoch": 1.420647208364394,
-      "grad_norm": 0.0022521934006363153,
-      "learning_rate": 0.00019999900511108605,
-      "loss": 46.0,
-      "step": 18581
-    },
-    {
-      "epoch": 1.4207236653477837,
-      "grad_norm": 0.0009552362025715411,
-      "learning_rate": 0.000199999005003939,
-      "loss": 46.0,
-      "step": 18582
-    },
-    {
-      "epoch": 1.4208001223311735,
-      "grad_norm": 0.004177308641374111,
-      "learning_rate": 0.00019999900489678615,
-      "loss": 46.0,
-      "step": 18583
-    },
-    {
-      "epoch": 1.4208765793145632,
-      "grad_norm": 0.0016907198587432504,
-      "learning_rate": 0.00019999900478962754,
-      "loss": 46.0,
-      "step": 18584
-    },
-    {
-      "epoch": 1.420953036297953,
-      "grad_norm": 0.004697458818554878,
-      "learning_rate": 0.00019999900468246318,
-      "loss": 46.0,
-      "step": 18585
-    },
-    {
-      "epoch": 1.4210294932813425,
-      "grad_norm": 0.0005175459664314985,
-      "learning_rate": 0.00019999900457529305,
-      "loss": 46.0,
-      "step": 18586
-    },
-    {
-      "epoch": 1.4211059502647323,
-      "grad_norm": 0.00032005272805690765,
-      "learning_rate": 0.00019999900446811711,
-      "loss": 46.0,
-      "step": 18587
-    },
-    {
-      "epoch": 1.421182407248122,
-      "grad_norm": 0.0009023714810609818,
-      "learning_rate": 0.00019999900436093543,
-      "loss": 46.0,
-      "step": 18588
-    },
-    {
-      "epoch": 1.4212588642315118,
-      "grad_norm": 0.00041148753371089697,
-      "learning_rate": 0.000199999004253748,
-      "loss": 46.0,
-      "step": 18589
-    },
-    {
-      "epoch": 1.4213353212149014,
-      "grad_norm": 0.0015865247696638107,
-      "learning_rate": 0.00019999900414655478,
-      "loss": 46.0,
-      "step": 18590
-    },
-    {
-      "epoch": 1.4214117781982911,
-      "grad_norm": 0.0012406307505443692,
-      "learning_rate": 0.00019999900403935578,
-      "loss": 46.0,
-      "step": 18591
-    },
-    {
-      "epoch": 1.4214882351816809,
-      "grad_norm": 0.001070024911314249,
-      "learning_rate": 0.00019999900393215104,
-      "loss": 46.0,
-      "step": 18592
-    },
-    {
-      "epoch": 1.4215646921650706,
-      "grad_norm": 0.0011345661478117108,
-      "learning_rate": 0.0001999990038249405,
-      "loss": 46.0,
-      "step": 18593
-    },
-    {
-      "epoch": 1.4216411491484604,
-      "grad_norm": 0.0009538857848383486,
-      "learning_rate": 0.00019999900371772423,
-      "loss": 46.0,
-      "step": 18594
-    },
-    {
-      "epoch": 1.4217176061318502,
-      "grad_norm": 0.0006144797662273049,
-      "learning_rate": 0.00019999900361050216,
-      "loss": 46.0,
-      "step": 18595
-    },
-    {
-      "epoch": 1.42179406311524,
-      "grad_norm": 0.0021742782555520535,
-      "learning_rate": 0.00019999900350327432,
-      "loss": 46.0,
-      "step": 18596
-    },
-    {
-      "epoch": 1.4218705200986295,
-      "grad_norm": 0.0012468538479879498,
-      "learning_rate": 0.0001999990033960407,
-      "loss": 46.0,
-      "step": 18597
-    },
-    {
-      "epoch": 1.4219469770820192,
-      "grad_norm": 0.0013210023753345013,
-      "learning_rate": 0.00019999900328880135,
-      "loss": 46.0,
-      "step": 18598
-    },
-    {
-      "epoch": 1.422023434065409,
-      "grad_norm": 0.001301935059018433,
-      "learning_rate": 0.0001999990031815562,
-      "loss": 46.0,
-      "step": 18599
-    },
-    {
-      "epoch": 1.4220998910487987,
-      "grad_norm": 0.0007719448185525835,
-      "learning_rate": 0.0001999990030743053,
-      "loss": 46.0,
-      "step": 18600
-    },
-    {
-      "epoch": 1.4221763480321883,
-      "grad_norm": 0.0016647737938910723,
-      "learning_rate": 0.0001999990029670486,
-      "loss": 46.0,
-      "step": 18601
-    },
-    {
-      "epoch": 1.422252805015578,
-      "grad_norm": 0.0007330256048589945,
-      "learning_rate": 0.00019999900285978613,
-      "loss": 46.0,
-      "step": 18602
-    },
-    {
-      "epoch": 1.4223292619989678,
-      "grad_norm": 0.0008366921101696789,
-      "learning_rate": 0.00019999900275251796,
-      "loss": 46.0,
-      "step": 18603
-    },
-    {
-      "epoch": 1.4224057189823576,
-      "grad_norm": 0.0009837065590545535,
-      "learning_rate": 0.00019999900264524394,
-      "loss": 46.0,
-      "step": 18604
-    },
-    {
-      "epoch": 1.4224821759657473,
-      "grad_norm": 0.0010996019700542092,
-      "learning_rate": 0.0001999990025379642,
-      "loss": 46.0,
-      "step": 18605
-    },
-    {
-      "epoch": 1.422558632949137,
-      "grad_norm": 0.0003169424890074879,
-      "learning_rate": 0.00019999900243067867,
-      "loss": 46.0,
-      "step": 18606
-    },
-    {
-      "epoch": 1.4226350899325266,
-      "grad_norm": 0.0011030015302821994,
-      "learning_rate": 0.00019999900232338735,
-      "loss": 46.0,
-      "step": 18607
-    },
-    {
-      "epoch": 1.4227115469159164,
-      "grad_norm": 0.0015582737978547812,
-      "learning_rate": 0.0001999990022160903,
-      "loss": 46.0,
-      "step": 18608
-    },
-    {
-      "epoch": 1.4227880038993062,
-      "grad_norm": 0.0009905400220304728,
-      "learning_rate": 0.00019999900210878745,
-      "loss": 46.0,
-      "step": 18609
-    },
-    {
-      "epoch": 1.422864460882696,
-      "grad_norm": 0.0008396326447837055,
-      "learning_rate": 0.00019999900200147886,
-      "loss": 46.0,
-      "step": 18610
-    },
-    {
-      "epoch": 1.4229409178660857,
-      "grad_norm": 0.0005049839965067804,
-      "learning_rate": 0.00019999900189416448,
-      "loss": 46.0,
-      "step": 18611
-    },
-    {
-      "epoch": 1.4230173748494752,
-      "grad_norm": 0.003065577009692788,
-      "learning_rate": 0.00019999900178684432,
-      "loss": 46.0,
-      "step": 18612
-    },
-    {
-      "epoch": 1.423093831832865,
-      "grad_norm": 0.0027007723692804575,
-      "learning_rate": 0.00019999900167951844,
-      "loss": 46.0,
-      "step": 18613
-    },
-    {
-      "epoch": 1.4231702888162547,
-      "grad_norm": 0.003133367747068405,
-      "learning_rate": 0.00019999900157218674,
-      "loss": 46.0,
-      "step": 18614
-    },
-    {
-      "epoch": 1.4232467457996445,
-      "grad_norm": 0.0007613941561430693,
-      "learning_rate": 0.00019999900146484929,
-      "loss": 46.0,
-      "step": 18615
-    },
-    {
-      "epoch": 1.4233232027830343,
-      "grad_norm": 0.001023577991873026,
-      "learning_rate": 0.00019999900135750606,
-      "loss": 46.0,
-      "step": 18616
-    },
-    {
-      "epoch": 1.423399659766424,
-      "grad_norm": 0.0007779033039696515,
-      "learning_rate": 0.00019999900125015706,
-      "loss": 46.0,
-      "step": 18617
-    },
-    {
-      "epoch": 1.4234761167498136,
-      "grad_norm": 0.0006752707413397729,
-      "learning_rate": 0.0001999990011428023,
-      "loss": 46.0,
-      "step": 18618
-    },
-    {
-      "epoch": 1.4235525737332033,
-      "grad_norm": 0.0015332060866057873,
-      "learning_rate": 0.00019999900103544177,
-      "loss": 46.0,
-      "step": 18619
-    },
-    {
-      "epoch": 1.423629030716593,
-      "grad_norm": 0.018377667292952538,
-      "learning_rate": 0.00019999900092807548,
-      "loss": 46.0,
-      "step": 18620
-    },
-    {
-      "epoch": 1.4237054876999828,
-      "grad_norm": 0.0014321247581392527,
-      "learning_rate": 0.00019999900082070342,
-      "loss": 46.0,
-      "step": 18621
-    },
-    {
-      "epoch": 1.4237819446833724,
-      "grad_norm": 0.006086652632802725,
-      "learning_rate": 0.00019999900071332558,
-      "loss": 46.0,
-      "step": 18622
-    },
-    {
-      "epoch": 1.4238584016667621,
-      "grad_norm": 0.0026824716478586197,
-      "learning_rate": 0.00019999900060594197,
-      "loss": 46.0,
-      "step": 18623
-    },
-    {
-      "epoch": 1.423934858650152,
-      "grad_norm": 0.0016962835798040032,
-      "learning_rate": 0.00019999900049855258,
-      "loss": 46.0,
-      "step": 18624
-    },
-    {
-      "epoch": 1.4240113156335417,
-      "grad_norm": 0.002232568571344018,
-      "learning_rate": 0.00019999900039115745,
-      "loss": 46.0,
-      "step": 18625
-    },
-    {
-      "epoch": 1.4240877726169314,
-      "grad_norm": 0.0014266332145780325,
-      "learning_rate": 0.00019999900028375652,
-      "loss": 46.0,
-      "step": 18626
-    },
-    {
-      "epoch": 1.4241642296003212,
-      "grad_norm": 0.004583638161420822,
-      "learning_rate": 0.00019999900017634981,
-      "loss": 46.0,
-      "step": 18627
-    },
-    {
-      "epoch": 1.424240686583711,
-      "grad_norm": 0.0007547771674580872,
-      "learning_rate": 0.0001999990000689374,
-      "loss": 46.0,
-      "step": 18628
-    },
-    {
-      "epoch": 1.4243171435671005,
-      "grad_norm": 0.0008986523025669158,
-      "learning_rate": 0.00019999899996151917,
-      "loss": 46.0,
-      "step": 18629
-    },
-    {
-      "epoch": 1.4243936005504902,
-      "grad_norm": 0.0010006674565374851,
-      "learning_rate": 0.00019999899985409517,
-      "loss": 46.0,
-      "step": 18630
-    },
-    {
-      "epoch": 1.42447005753388,
-      "grad_norm": 0.0004649788897950202,
-      "learning_rate": 0.0001999989997466654,
-      "loss": 46.0,
-      "step": 18631
-    },
-    {
-      "epoch": 1.4245465145172698,
-      "grad_norm": 0.0016545712715014815,
-      "learning_rate": 0.00019999899963922988,
-      "loss": 46.0,
-      "step": 18632
-    },
-    {
-      "epoch": 1.4246229715006593,
-      "grad_norm": 0.0027371577452868223,
-      "learning_rate": 0.0001999989995317886,
-      "loss": 46.0,
-      "step": 18633
-    },
-    {
-      "epoch": 1.424699428484049,
-      "grad_norm": 0.0026675439439713955,
-      "learning_rate": 0.0001999989994243415,
-      "loss": 46.0,
-      "step": 18634
-    },
-    {
-      "epoch": 1.4247758854674388,
-      "grad_norm": 0.0018106952775269747,
-      "learning_rate": 0.00019999899931688866,
-      "loss": 46.0,
-      "step": 18635
-    },
-    {
-      "epoch": 1.4248523424508286,
-      "grad_norm": 0.0012644518865272403,
-      "learning_rate": 0.00019999899920943005,
-      "loss": 46.0,
-      "step": 18636
-    },
-    {
-      "epoch": 1.4249287994342184,
-      "grad_norm": 0.0004560202651191503,
-      "learning_rate": 0.00019999899910196566,
-      "loss": 46.0,
-      "step": 18637
-    },
-    {
-      "epoch": 1.4250052564176081,
-      "grad_norm": 0.0034558644983917475,
-      "learning_rate": 0.00019999899899449553,
-      "loss": 46.0,
-      "step": 18638
-    },
-    {
-      "epoch": 1.4250817134009979,
-      "grad_norm": 0.0016475735465064645,
-      "learning_rate": 0.0001999989988870196,
-      "loss": 46.0,
-      "step": 18639
-    },
-    {
-      "epoch": 1.4251581703843874,
-      "grad_norm": 0.0036439159885048866,
-      "learning_rate": 0.00019999899877953792,
-      "loss": 46.0,
-      "step": 18640
-    },
-    {
-      "epoch": 1.4252346273677772,
-      "grad_norm": 0.0033887161407619715,
-      "learning_rate": 0.00019999899867205044,
-      "loss": 46.0,
-      "step": 18641
-    },
-    {
-      "epoch": 1.425311084351167,
-      "grad_norm": 0.0007457028259523213,
-      "learning_rate": 0.00019999899856455722,
-      "loss": 46.0,
-      "step": 18642
-    },
-    {
-      "epoch": 1.4253875413345567,
-      "grad_norm": 0.006611408665776253,
-      "learning_rate": 0.00019999899845705822,
-      "loss": 46.0,
-      "step": 18643
-    },
-    {
-      "epoch": 1.4254639983179462,
-      "grad_norm": 0.0007065095123834908,
-      "learning_rate": 0.00019999899834955348,
-      "loss": 46.0,
-      "step": 18644
-    },
-    {
-      "epoch": 1.425540455301336,
-      "grad_norm": 0.0006840868736617267,
-      "learning_rate": 0.00019999899824204293,
-      "loss": 46.0,
-      "step": 18645
-    },
-    {
-      "epoch": 1.4256169122847258,
-      "grad_norm": 0.0018743122927844524,
-      "learning_rate": 0.00019999899813452661,
-      "loss": 46.0,
-      "step": 18646
-    },
-    {
-      "epoch": 1.4256933692681155,
-      "grad_norm": 0.0008903193520382047,
-      "learning_rate": 0.00019999899802700455,
-      "loss": 46.0,
-      "step": 18647
-    },
-    {
-      "epoch": 1.4257698262515053,
-      "grad_norm": 0.0016152093885466456,
-      "learning_rate": 0.0001999989979194767,
-      "loss": 46.0,
-      "step": 18648
-    },
-    {
-      "epoch": 1.425846283234895,
-      "grad_norm": 0.004433429334312677,
-      "learning_rate": 0.0001999989978119431,
-      "loss": 46.0,
-      "step": 18649
-    },
-    {
-      "epoch": 1.4259227402182848,
-      "grad_norm": 0.0011305814841762185,
-      "learning_rate": 0.00019999899770440372,
-      "loss": 46.0,
-      "step": 18650
-    },
-    {
-      "epoch": 1.4259991972016743,
-      "grad_norm": 0.007987190037965775,
-      "learning_rate": 0.0001999989975968586,
-      "loss": 46.0,
-      "step": 18651
-    },
-    {
-      "epoch": 1.426075654185064,
-      "grad_norm": 0.00039009845932014287,
-      "learning_rate": 0.00019999899748930763,
-      "loss": 46.0,
-      "step": 18652
-    },
-    {
-      "epoch": 1.4261521111684539,
-      "grad_norm": 0.007430423516780138,
-      "learning_rate": 0.00019999899738175095,
-      "loss": 46.0,
-      "step": 18653
-    },
-    {
-      "epoch": 1.4262285681518436,
-      "grad_norm": 0.003537822747603059,
-      "learning_rate": 0.00019999899727418853,
-      "loss": 46.0,
-      "step": 18654
-    },
-    {
-      "epoch": 1.4263050251352332,
-      "grad_norm": 0.0020287923980504274,
-      "learning_rate": 0.00019999899716662028,
-      "loss": 46.0,
-      "step": 18655
-    },
-    {
-      "epoch": 1.426381482118623,
-      "grad_norm": 0.013133603148162365,
-      "learning_rate": 0.00019999899705904628,
-      "loss": 46.0,
-      "step": 18656
-    },
-    {
-      "epoch": 1.4264579391020127,
-      "grad_norm": 0.001169879105873406,
-      "learning_rate": 0.0001999989969514665,
-      "loss": 46.0,
-      "step": 18657
-    },
-    {
-      "epoch": 1.4265343960854024,
-      "grad_norm": 0.0032761956099420786,
-      "learning_rate": 0.00019999899684388096,
-      "loss": 46.0,
-      "step": 18658
-    },
-    {
-      "epoch": 1.4266108530687922,
-      "grad_norm": 0.0006516337743960321,
-      "learning_rate": 0.00019999899673628967,
-      "loss": 46.0,
-      "step": 18659
-    },
-    {
-      "epoch": 1.426687310052182,
-      "grad_norm": 0.0021252119913697243,
-      "learning_rate": 0.00019999899662869258,
-      "loss": 46.0,
-      "step": 18660
-    },
-    {
-      "epoch": 1.4267637670355717,
-      "grad_norm": 0.0009915714617818594,
-      "learning_rate": 0.00019999899652108972,
-      "loss": 46.0,
-      "step": 18661
-    },
-    {
-      "epoch": 1.4268402240189613,
-      "grad_norm": 0.0036959603894501925,
-      "learning_rate": 0.00019999899641348114,
-      "loss": 46.0,
-      "step": 18662
-    },
-    {
-      "epoch": 1.426916681002351,
-      "grad_norm": 0.0005464826244860888,
-      "learning_rate": 0.00019999899630586672,
-      "loss": 46.0,
-      "step": 18663
-    },
-    {
-      "epoch": 1.4269931379857408,
-      "grad_norm": 0.0007365245837718248,
-      "learning_rate": 0.0001999989961982466,
-      "loss": 46.0,
-      "step": 18664
-    },
-    {
-      "epoch": 1.4270695949691306,
-      "grad_norm": 0.0010078215273097157,
-      "learning_rate": 0.00019999899609062066,
-      "loss": 46.0,
-      "step": 18665
-    },
-    {
-      "epoch": 1.42714605195252,
-      "grad_norm": 0.005040076561272144,
-      "learning_rate": 0.00019999899598298896,
-      "loss": 46.0,
-      "step": 18666
-    },
-    {
-      "epoch": 1.4272225089359099,
-      "grad_norm": 0.004677411634474993,
-      "learning_rate": 0.0001999989958753515,
-      "loss": 46.0,
-      "step": 18667
-    },
-    {
-      "epoch": 1.4272989659192996,
-      "grad_norm": 0.0009450827492401004,
-      "learning_rate": 0.00019999899576770828,
-      "loss": 46.0,
-      "step": 18668
-    },
-    {
-      "epoch": 1.4273754229026894,
-      "grad_norm": 0.0006946150679141283,
-      "learning_rate": 0.0001999989956600593,
-      "loss": 46.0,
-      "step": 18669
-    },
-    {
-      "epoch": 1.4274518798860791,
-      "grad_norm": 0.0008818952483125031,
-      "learning_rate": 0.00019999899555240452,
-      "loss": 46.0,
-      "step": 18670
-    },
-    {
-      "epoch": 1.427528336869469,
-      "grad_norm": 0.0015344680286943913,
-      "learning_rate": 0.00019999899544474397,
-      "loss": 46.0,
-      "step": 18671
-    },
-    {
-      "epoch": 1.4276047938528587,
-      "grad_norm": 0.0009433648083359003,
-      "learning_rate": 0.00019999899533707766,
-      "loss": 46.0,
-      "step": 18672
-    },
-    {
-      "epoch": 1.4276812508362482,
-      "grad_norm": 0.004973521456122398,
-      "learning_rate": 0.0001999989952294056,
-      "loss": 46.0,
-      "step": 18673
-    },
-    {
-      "epoch": 1.427757707819638,
-      "grad_norm": 0.001084906398318708,
-      "learning_rate": 0.00019999899512172773,
-      "loss": 46.0,
-      "step": 18674
-    },
-    {
-      "epoch": 1.4278341648030277,
-      "grad_norm": 0.001618018141016364,
-      "learning_rate": 0.00019999899501404415,
-      "loss": 46.0,
-      "step": 18675
-    },
-    {
-      "epoch": 1.4279106217864175,
-      "grad_norm": 0.0013044291408732533,
-      "learning_rate": 0.00019999899490635473,
-      "loss": 46.0,
-      "step": 18676
-    },
-    {
-      "epoch": 1.427987078769807,
-      "grad_norm": 0.0006601489149034023,
-      "learning_rate": 0.00019999899479865958,
-      "loss": 46.0,
-      "step": 18677
-    },
-    {
-      "epoch": 1.4280635357531968,
-      "grad_norm": 0.0010181114776059985,
-      "learning_rate": 0.00019999899469095867,
-      "loss": 46.0,
-      "step": 18678
-    },
-    {
-      "epoch": 1.4281399927365865,
-      "grad_norm": 0.0048792739398777485,
-      "learning_rate": 0.00019999899458325197,
-      "loss": 46.0,
-      "step": 18679
-    },
-    {
-      "epoch": 1.4282164497199763,
-      "grad_norm": 0.0006001926958560944,
-      "learning_rate": 0.00019999899447553952,
-      "loss": 46.0,
-      "step": 18680
-    },
-    {
-      "epoch": 1.428292906703366,
-      "grad_norm": 0.003065509255975485,
-      "learning_rate": 0.00019999899436782127,
-      "loss": 46.0,
-      "step": 18681
-    },
-    {
-      "epoch": 1.4283693636867558,
-      "grad_norm": 0.012520805932581425,
-      "learning_rate": 0.00019999899426009727,
-      "loss": 46.0,
-      "step": 18682
-    },
-    {
-      "epoch": 1.4284458206701456,
-      "grad_norm": 0.0010263193398714066,
-      "learning_rate": 0.0001999989941523675,
-      "loss": 46.0,
-      "step": 18683
-    },
-    {
-      "epoch": 1.4285222776535351,
-      "grad_norm": 0.0033806944265961647,
-      "learning_rate": 0.00019999899404463196,
-      "loss": 46.0,
-      "step": 18684
-    },
-    {
-      "epoch": 1.4285987346369249,
-      "grad_norm": 0.000461573654320091,
-      "learning_rate": 0.00019999899393689064,
-      "loss": 46.0,
-      "step": 18685
-    },
-    {
-      "epoch": 1.4286751916203146,
-      "grad_norm": 0.0005505571025423706,
-      "learning_rate": 0.00019999899382914355,
-      "loss": 46.0,
-      "step": 18686
-    },
-    {
-      "epoch": 1.4287516486037044,
-      "grad_norm": 0.0016493485309183598,
-      "learning_rate": 0.00019999899372139072,
-      "loss": 46.0,
-      "step": 18687
-    },
-    {
-      "epoch": 1.428828105587094,
-      "grad_norm": 0.0012330812169238925,
-      "learning_rate": 0.0001999989936136321,
-      "loss": 46.0,
-      "step": 18688
-    },
-    {
-      "epoch": 1.4289045625704837,
-      "grad_norm": 0.0018370734760537744,
-      "learning_rate": 0.0001999989935058677,
-      "loss": 46.0,
-      "step": 18689
-    },
-    {
-      "epoch": 1.4289810195538735,
-      "grad_norm": 0.000623841944616288,
-      "learning_rate": 0.00019999899339809757,
-      "loss": 46.0,
-      "step": 18690
-    },
-    {
-      "epoch": 1.4290574765372632,
-      "grad_norm": 0.001196195837110281,
-      "learning_rate": 0.00019999899329032164,
-      "loss": 46.0,
-      "step": 18691
-    },
-    {
-      "epoch": 1.429133933520653,
-      "grad_norm": 0.0033842779230326414,
-      "learning_rate": 0.00019999899318253993,
-      "loss": 46.0,
-      "step": 18692
-    },
-    {
-      "epoch": 1.4292103905040427,
-      "grad_norm": 0.0008980220300145447,
-      "learning_rate": 0.00019999899307475248,
-      "loss": 46.0,
-      "step": 18693
-    },
-    {
-      "epoch": 1.4292868474874325,
-      "grad_norm": 0.0004935725592076778,
-      "learning_rate": 0.00019999899296695923,
-      "loss": 46.0,
-      "step": 18694
-    },
-    {
-      "epoch": 1.429363304470822,
-      "grad_norm": 0.008362212218344212,
-      "learning_rate": 0.0001999989928591602,
-      "loss": 46.0,
-      "step": 18695
-    },
-    {
-      "epoch": 1.4294397614542118,
-      "grad_norm": 0.004225720185786486,
-      "learning_rate": 0.00019999899275135547,
-      "loss": 46.0,
-      "step": 18696
-    },
-    {
-      "epoch": 1.4295162184376016,
-      "grad_norm": 0.0020503634586930275,
-      "learning_rate": 0.00019999899264354492,
-      "loss": 46.0,
-      "step": 18697
-    },
-    {
-      "epoch": 1.4295926754209913,
-      "grad_norm": 0.0024455529637634754,
-      "learning_rate": 0.00019999899253572858,
-      "loss": 46.0,
-      "step": 18698
-    },
-    {
-      "epoch": 1.4296691324043809,
-      "grad_norm": 0.001271538552828133,
-      "learning_rate": 0.00019999899242790652,
-      "loss": 46.0,
-      "step": 18699
-    },
-    {
-      "epoch": 1.4297455893877706,
-      "grad_norm": 0.0020179194398224354,
-      "learning_rate": 0.00019999899232007865,
-      "loss": 46.0,
-      "step": 18700
-    },
-    {
-      "epoch": 1.4298220463711604,
-      "grad_norm": 0.0009006079635582864,
-      "learning_rate": 0.00019999899221224504,
-      "loss": 46.0,
-      "step": 18701
-    },
-    {
-      "epoch": 1.4298985033545502,
-      "grad_norm": 0.0010742960730567575,
-      "learning_rate": 0.00019999899210440566,
-      "loss": 46.0,
-      "step": 18702
-    },
-    {
-      "epoch": 1.42997496033794,
-      "grad_norm": 0.002057841047644615,
-      "learning_rate": 0.0001999989919965605,
-      "loss": 46.0,
-      "step": 18703
-    },
-    {
-      "epoch": 1.4300514173213297,
-      "grad_norm": 0.001576918177306652,
-      "learning_rate": 0.00019999899188870957,
-      "loss": 46.0,
-      "step": 18704
-    },
-    {
-      "epoch": 1.4301278743047194,
-      "grad_norm": 0.0008519926341250539,
-      "learning_rate": 0.00019999899178085287,
-      "loss": 46.0,
-      "step": 18705
-    },
-    {
-      "epoch": 1.430204331288109,
-      "grad_norm": 0.0007095919572748244,
-      "learning_rate": 0.0001999989916729904,
-      "loss": 46.0,
-      "step": 18706
-    },
-    {
-      "epoch": 1.4302807882714987,
-      "grad_norm": 0.0015185867669060826,
-      "learning_rate": 0.00019999899156512217,
-      "loss": 46.0,
-      "step": 18707
-    },
-    {
-      "epoch": 1.4303572452548885,
-      "grad_norm": 0.0009592008427716792,
-      "learning_rate": 0.00019999899145724817,
-      "loss": 46.0,
-      "step": 18708
-    },
-    {
-      "epoch": 1.4304337022382783,
-      "grad_norm": 0.0009776568040251732,
-      "learning_rate": 0.00019999899134936838,
-      "loss": 46.0,
-      "step": 18709
-    },
-    {
-      "epoch": 1.4305101592216678,
-      "grad_norm": 0.001985912909731269,
-      "learning_rate": 0.00019999899124148283,
-      "loss": 46.0,
-      "step": 18710
-    },
-    {
-      "epoch": 1.4305866162050576,
-      "grad_norm": 0.0006811233470216393,
-      "learning_rate": 0.00019999899113359152,
-      "loss": 46.0,
-      "step": 18711
-    },
-    {
-      "epoch": 1.4306630731884473,
-      "grad_norm": 0.0062431227415800095,
-      "learning_rate": 0.00019999899102569443,
-      "loss": 46.0,
-      "step": 18712
-    },
-    {
-      "epoch": 1.430739530171837,
-      "grad_norm": 0.0007096232729963958,
-      "learning_rate": 0.00019999899091779157,
-      "loss": 46.0,
-      "step": 18713
-    },
-    {
-      "epoch": 1.4308159871552268,
-      "grad_norm": 0.0015456915134564042,
-      "learning_rate": 0.00019999899080988296,
-      "loss": 46.0,
-      "step": 18714
-    },
-    {
-      "epoch": 1.4308924441386166,
-      "grad_norm": 0.018905820325016975,
-      "learning_rate": 0.00019999899070196855,
-      "loss": 46.0,
-      "step": 18715
-    },
-    {
-      "epoch": 1.4309689011220064,
-      "grad_norm": 0.003829828230664134,
-      "learning_rate": 0.0001999989905940484,
-      "loss": 46.0,
-      "step": 18716
-    },
-    {
-      "epoch": 1.431045358105396,
-      "grad_norm": 0.0004912882577627897,
-      "learning_rate": 0.00019999899048612246,
-      "loss": 46.0,
-      "step": 18717
-    },
-    {
-      "epoch": 1.4311218150887857,
-      "grad_norm": 0.001517701894044876,
-      "learning_rate": 0.00019999899037819076,
-      "loss": 46.0,
-      "step": 18718
-    },
-    {
-      "epoch": 1.4311982720721754,
-      "grad_norm": 0.0017941041151061654,
-      "learning_rate": 0.0001999989902702533,
-      "loss": 46.0,
-      "step": 18719
-    },
-    {
-      "epoch": 1.4312747290555652,
-      "grad_norm": 0.000769377569667995,
-      "learning_rate": 0.00019999899016231006,
-      "loss": 46.0,
-      "step": 18720
-    },
-    {
-      "epoch": 1.4313511860389547,
-      "grad_norm": 0.0012285992270335555,
-      "learning_rate": 0.00019999899005436104,
-      "loss": 46.0,
-      "step": 18721
-    },
-    {
-      "epoch": 1.4314276430223445,
-      "grad_norm": 0.007775334641337395,
-      "learning_rate": 0.00019999898994640627,
-      "loss": 46.0,
-      "step": 18722
-    },
-    {
-      "epoch": 1.4315041000057342,
-      "grad_norm": 0.0014218116411939263,
-      "learning_rate": 0.00019999898983844575,
-      "loss": 46.0,
-      "step": 18723
-    },
-    {
-      "epoch": 1.431580556989124,
-      "grad_norm": 0.0006094217533245683,
-      "learning_rate": 0.00019999898973047938,
-      "loss": 46.0,
-      "step": 18724
-    },
-    {
-      "epoch": 1.4316570139725138,
-      "grad_norm": 0.00151549291331321,
-      "learning_rate": 0.00019999898962250732,
-      "loss": 46.0,
-      "step": 18725
-    },
-    {
-      "epoch": 1.4317334709559035,
-      "grad_norm": 0.0015793833881616592,
-      "learning_rate": 0.00019999898951452946,
-      "loss": 46.0,
-      "step": 18726
-    },
-    {
-      "epoch": 1.4318099279392933,
-      "grad_norm": 0.0007776551647111773,
-      "learning_rate": 0.00019999898940654585,
-      "loss": 46.0,
-      "step": 18727
-    },
-    {
-      "epoch": 1.4318863849226828,
-      "grad_norm": 0.0027849250473082066,
-      "learning_rate": 0.00019999898929855644,
-      "loss": 46.0,
-      "step": 18728
-    },
-    {
-      "epoch": 1.4319628419060726,
-      "grad_norm": 0.0006977255106903613,
-      "learning_rate": 0.00019999898919056128,
-      "loss": 46.0,
-      "step": 18729
-    },
-    {
-      "epoch": 1.4320392988894624,
-      "grad_norm": 0.00579830864444375,
-      "learning_rate": 0.00019999898908256035,
-      "loss": 46.0,
-      "step": 18730
-    },
-    {
-      "epoch": 1.4321157558728521,
-      "grad_norm": 0.0015832193894311786,
-      "learning_rate": 0.00019999898897455365,
-      "loss": 46.0,
-      "step": 18731
-    },
-    {
-      "epoch": 1.4321922128562417,
-      "grad_norm": 0.001268958323635161,
-      "learning_rate": 0.00019999898886654117,
-      "loss": 46.0,
-      "step": 18732
-    },
-    {
-      "epoch": 1.4322686698396314,
-      "grad_norm": 0.0023746297229081392,
-      "learning_rate": 0.00019999898875852292,
-      "loss": 46.0,
-      "step": 18733
-    },
-    {
-      "epoch": 1.4323451268230212,
-      "grad_norm": 0.003861360251903534,
-      "learning_rate": 0.00019999898865049893,
-      "loss": 46.0,
-      "step": 18734
-    },
-    {
-      "epoch": 1.432421583806411,
-      "grad_norm": 0.0011787962866947055,
-      "learning_rate": 0.00019999898854246913,
-      "loss": 46.0,
-      "step": 18735
-    },
-    {
-      "epoch": 1.4324980407898007,
-      "grad_norm": 0.012241599150002003,
-      "learning_rate": 0.0001999989884344336,
-      "loss": 46.0,
-      "step": 18736
-    },
-    {
-      "epoch": 1.4325744977731905,
-      "grad_norm": 0.0024612946435809135,
-      "learning_rate": 0.00019999898832639228,
-      "loss": 46.0,
-      "step": 18737
-    },
-    {
-      "epoch": 1.43265095475658,
-      "grad_norm": 0.0006849869969300926,
-      "learning_rate": 0.00019999898821834519,
-      "loss": 46.0,
-      "step": 18738
-    },
-    {
-      "epoch": 1.4327274117399698,
-      "grad_norm": 0.001151581876911223,
-      "learning_rate": 0.00019999898811029235,
-      "loss": 46.0,
-      "step": 18739
-    },
-    {
-      "epoch": 1.4328038687233595,
-      "grad_norm": 0.0014457560610026121,
-      "learning_rate": 0.00019999898800223372,
-      "loss": 46.0,
-      "step": 18740
-    },
-    {
-      "epoch": 1.4328803257067493,
-      "grad_norm": 0.0006206260877661407,
-      "learning_rate": 0.0001999989878941693,
-      "loss": 46.0,
-      "step": 18741
-    },
-    {
-      "epoch": 1.432956782690139,
-      "grad_norm": 0.0013611504109576344,
-      "learning_rate": 0.00019999898778609912,
-      "loss": 46.0,
-      "step": 18742
-    },
-    {
-      "epoch": 1.4330332396735286,
-      "grad_norm": 0.0021817791275680065,
-      "learning_rate": 0.0001999989876780232,
-      "loss": 46.0,
-      "step": 18743
-    },
-    {
-      "epoch": 1.4331096966569183,
-      "grad_norm": 0.002085766987875104,
-      "learning_rate": 0.00019999898756994152,
-      "loss": 46.0,
-      "step": 18744
-    },
-    {
-      "epoch": 1.433186153640308,
-      "grad_norm": 0.0009521907195448875,
-      "learning_rate": 0.00019999898746185402,
-      "loss": 46.0,
-      "step": 18745
-    },
-    {
-      "epoch": 1.4332626106236979,
-      "grad_norm": 0.0008955611265264452,
-      "learning_rate": 0.0001999989873537608,
-      "loss": 46.0,
-      "step": 18746
-    },
-    {
-      "epoch": 1.4333390676070876,
-      "grad_norm": 0.0026764387730509043,
-      "learning_rate": 0.00019999898724566175,
-      "loss": 46.0,
-      "step": 18747
-    },
-    {
-      "epoch": 1.4334155245904774,
-      "grad_norm": 0.0013781870948150754,
-      "learning_rate": 0.00019999898713755698,
-      "loss": 46.0,
-      "step": 18748
-    },
-    {
-      "epoch": 1.433491981573867,
-      "grad_norm": 0.0023273660335689783,
-      "learning_rate": 0.00019999898702944643,
-      "loss": 46.0,
-      "step": 18749
-    },
-    {
-      "epoch": 1.4335684385572567,
-      "grad_norm": 0.005769746843725443,
-      "learning_rate": 0.00019999898692133012,
-      "loss": 46.0,
-      "step": 18750
-    },
-    {
-      "epoch": 1.4336448955406464,
-      "grad_norm": 0.007147619500756264,
-      "learning_rate": 0.00019999898681320803,
-      "loss": 46.0,
-      "step": 18751
-    },
-    {
-      "epoch": 1.4337213525240362,
-      "grad_norm": 0.000937281409278512,
-      "learning_rate": 0.00019999898670508017,
-      "loss": 46.0,
-      "step": 18752
-    },
-    {
-      "epoch": 1.4337978095074257,
-      "grad_norm": 0.0015842615393921733,
-      "learning_rate": 0.00019999898659694653,
-      "loss": 46.0,
-      "step": 18753
-    },
-    {
-      "epoch": 1.4338742664908155,
-      "grad_norm": 0.004984136670827866,
-      "learning_rate": 0.00019999898648880712,
-      "loss": 46.0,
-      "step": 18754
-    },
-    {
-      "epoch": 1.4339507234742053,
-      "grad_norm": 0.0006235511391423643,
-      "learning_rate": 0.00019999898638066197,
-      "loss": 46.0,
-      "step": 18755
-    },
-    {
-      "epoch": 1.434027180457595,
-      "grad_norm": 0.0013173272600397468,
-      "learning_rate": 0.00019999898627251104,
-      "loss": 46.0,
-      "step": 18756
-    },
-    {
-      "epoch": 1.4341036374409848,
-      "grad_norm": 0.004447820130735636,
-      "learning_rate": 0.0001999989861643543,
-      "loss": 46.0,
-      "step": 18757
-    },
-    {
-      "epoch": 1.4341800944243746,
-      "grad_norm": 0.005511179566383362,
-      "learning_rate": 0.00019999898605619186,
-      "loss": 46.0,
-      "step": 18758
-    },
-    {
-      "epoch": 1.4342565514077643,
-      "grad_norm": 0.0011743898503482342,
-      "learning_rate": 0.0001999989859480236,
-      "loss": 46.0,
-      "step": 18759
-    },
-    {
-      "epoch": 1.4343330083911539,
-      "grad_norm": 0.006367901805788279,
-      "learning_rate": 0.00019999898583984957,
-      "loss": 46.0,
-      "step": 18760
-    },
-    {
-      "epoch": 1.4344094653745436,
-      "grad_norm": 0.0007368992082774639,
-      "learning_rate": 0.0001999989857316698,
-      "loss": 46.0,
-      "step": 18761
-    },
-    {
-      "epoch": 1.4344859223579334,
-      "grad_norm": 0.001369076082482934,
-      "learning_rate": 0.00019999898562348426,
-      "loss": 46.0,
-      "step": 18762
-    },
-    {
-      "epoch": 1.4345623793413231,
-      "grad_norm": 0.0005207036156207323,
-      "learning_rate": 0.0001999989855152929,
-      "loss": 46.0,
-      "step": 18763
-    },
-    {
-      "epoch": 1.4346388363247127,
-      "grad_norm": 0.001141222775913775,
-      "learning_rate": 0.00019999898540709582,
-      "loss": 46.0,
-      "step": 18764
-    },
-    {
-      "epoch": 1.4347152933081024,
-      "grad_norm": 0.002879117848351598,
-      "learning_rate": 0.00019999898529889296,
-      "loss": 46.0,
-      "step": 18765
-    },
-    {
-      "epoch": 1.4347917502914922,
-      "grad_norm": 0.0013714451342821121,
-      "learning_rate": 0.00019999898519068433,
-      "loss": 46.0,
-      "step": 18766
-    },
-    {
-      "epoch": 1.434868207274882,
-      "grad_norm": 0.0007758744759485126,
-      "learning_rate": 0.00019999898508246995,
-      "loss": 46.0,
-      "step": 18767
-    },
-    {
-      "epoch": 1.4349446642582717,
-      "grad_norm": 0.0003821845748461783,
-      "learning_rate": 0.00019999898497424976,
-      "loss": 46.0,
-      "step": 18768
-    },
-    {
-      "epoch": 1.4350211212416615,
-      "grad_norm": 0.005240944214165211,
-      "learning_rate": 0.0001999989848660238,
-      "loss": 46.0,
-      "step": 18769
-    },
-    {
-      "epoch": 1.4350975782250512,
-      "grad_norm": 0.003266120096668601,
-      "learning_rate": 0.0001999989847577921,
-      "loss": 46.0,
-      "step": 18770
-    },
-    {
-      "epoch": 1.4351740352084408,
-      "grad_norm": 0.0032120102550834417,
-      "learning_rate": 0.00019999898464955463,
-      "loss": 46.0,
-      "step": 18771
-    },
-    {
-      "epoch": 1.4352504921918305,
-      "grad_norm": 0.000991167384199798,
-      "learning_rate": 0.00019999898454131138,
-      "loss": 46.0,
-      "step": 18772
-    },
-    {
-      "epoch": 1.4353269491752203,
-      "grad_norm": 0.004238905850797892,
-      "learning_rate": 0.00019999898443306236,
-      "loss": 46.0,
-      "step": 18773
-    },
-    {
-      "epoch": 1.43540340615861,
-      "grad_norm": 0.0014413670869544148,
-      "learning_rate": 0.00019999898432480757,
-      "loss": 46.0,
-      "step": 18774
-    },
-    {
-      "epoch": 1.4354798631419996,
-      "grad_norm": 0.0007632928900420666,
-      "learning_rate": 0.00019999898421654703,
-      "loss": 46.0,
-      "step": 18775
-    },
-    {
-      "epoch": 1.4355563201253894,
-      "grad_norm": 0.0008038313826546073,
-      "learning_rate": 0.00019999898410828068,
-      "loss": 46.0,
-      "step": 18776
-    },
-    {
-      "epoch": 1.4356327771087791,
-      "grad_norm": 0.001329036895185709,
-      "learning_rate": 0.0001999989840000086,
-      "loss": 46.0,
-      "step": 18777
-    },
-    {
-      "epoch": 1.4357092340921689,
-      "grad_norm": 0.00276779243722558,
-      "learning_rate": 0.00019999898389173076,
-      "loss": 46.0,
-      "step": 18778
-    },
-    {
-      "epoch": 1.4357856910755586,
-      "grad_norm": 0.002485651755705476,
-      "learning_rate": 0.0001999989837834471,
-      "loss": 46.0,
-      "step": 18779
-    },
-    {
-      "epoch": 1.4358621480589484,
-      "grad_norm": 0.0012535244459286332,
-      "learning_rate": 0.0001999989836751577,
-      "loss": 46.0,
-      "step": 18780
-    },
-    {
-      "epoch": 1.4359386050423382,
-      "grad_norm": 0.0006342001724988222,
-      "learning_rate": 0.00019999898356686254,
-      "loss": 46.0,
-      "step": 18781
-    },
-    {
-      "epoch": 1.4360150620257277,
-      "grad_norm": 0.0014903387054800987,
-      "learning_rate": 0.0001999989834585616,
-      "loss": 46.0,
-      "step": 18782
-    },
-    {
-      "epoch": 1.4360915190091175,
-      "grad_norm": 0.0011879093945026398,
-      "learning_rate": 0.00019999898335025485,
-      "loss": 46.0,
-      "step": 18783
-    },
-    {
-      "epoch": 1.4361679759925072,
-      "grad_norm": 0.0007911497377790511,
-      "learning_rate": 0.0001999989832419424,
-      "loss": 46.0,
-      "step": 18784
-    },
-    {
-      "epoch": 1.436244432975897,
-      "grad_norm": 0.0009299229714088142,
-      "learning_rate": 0.00019999898313362416,
-      "loss": 46.0,
-      "step": 18785
-    },
-    {
-      "epoch": 1.4363208899592865,
-      "grad_norm": 0.0009314122726209462,
-      "learning_rate": 0.00019999898302530014,
-      "loss": 46.0,
-      "step": 18786
-    },
-    {
-      "epoch": 1.4363973469426763,
-      "grad_norm": 0.004738044925034046,
-      "learning_rate": 0.00019999898291697034,
-      "loss": 46.0,
-      "step": 18787
-    },
-    {
-      "epoch": 1.436473803926066,
-      "grad_norm": 0.0011719597969204187,
-      "learning_rate": 0.00019999898280863477,
-      "loss": 46.0,
-      "step": 18788
-    },
-    {
-      "epoch": 1.4365502609094558,
-      "grad_norm": 0.0007641953998245299,
-      "learning_rate": 0.00019999898270029346,
-      "loss": 46.0,
-      "step": 18789
-    },
-    {
-      "epoch": 1.4366267178928456,
-      "grad_norm": 0.0008409270667470992,
-      "learning_rate": 0.00019999898259194637,
-      "loss": 46.0,
-      "step": 18790
-    },
-    {
-      "epoch": 1.4367031748762353,
-      "grad_norm": 0.0011571451323106885,
-      "learning_rate": 0.0001999989824835935,
-      "loss": 46.0,
-      "step": 18791
-    },
-    {
-      "epoch": 1.436779631859625,
-      "grad_norm": 0.0006027471390552819,
-      "learning_rate": 0.00019999898237523485,
-      "loss": 46.0,
-      "step": 18792
-    },
-    {
-      "epoch": 1.4368560888430146,
-      "grad_norm": 0.0018605849472805858,
-      "learning_rate": 0.00019999898226687047,
-      "loss": 46.0,
-      "step": 18793
-    },
-    {
-      "epoch": 1.4369325458264044,
-      "grad_norm": 0.001404240378178656,
-      "learning_rate": 0.00019999898215850026,
-      "loss": 46.0,
-      "step": 18794
-    },
-    {
-      "epoch": 1.4370090028097942,
-      "grad_norm": 0.008444486185908318,
-      "learning_rate": 0.00019999898205012433,
-      "loss": 46.0,
-      "step": 18795
-    },
-    {
-      "epoch": 1.437085459793184,
-      "grad_norm": 0.0052825771272182465,
-      "learning_rate": 0.00019999898194174263,
-      "loss": 46.0,
-      "step": 18796
-    },
-    {
-      "epoch": 1.4371619167765735,
-      "grad_norm": 0.0006845666794106364,
-      "learning_rate": 0.00019999898183335516,
-      "loss": 46.0,
-      "step": 18797
-    },
-    {
-      "epoch": 1.4372383737599632,
-      "grad_norm": 0.004222303628921509,
-      "learning_rate": 0.0001999989817249619,
-      "loss": 46.0,
-      "step": 18798
-    },
-    {
-      "epoch": 1.437314830743353,
-      "grad_norm": 0.0011709652608260512,
-      "learning_rate": 0.0001999989816165629,
-      "loss": 46.0,
-      "step": 18799
-    },
-    {
-      "epoch": 1.4373912877267427,
-      "grad_norm": 0.000967221858445555,
-      "learning_rate": 0.00019999898150815807,
-      "loss": 46.0,
-      "step": 18800
-    },
-    {
-      "epoch": 1.4374677447101325,
-      "grad_norm": 0.0009400346898473799,
-      "learning_rate": 0.0001999989813997475,
-      "loss": 46.0,
-      "step": 18801
-    },
-    {
-      "epoch": 1.4375442016935223,
-      "grad_norm": 0.0038634752854704857,
-      "learning_rate": 0.00019999898129133116,
-      "loss": 46.0,
-      "step": 18802
-    },
-    {
-      "epoch": 1.437620658676912,
-      "grad_norm": 0.00157473748549819,
-      "learning_rate": 0.0001999989811829091,
-      "loss": 46.0,
-      "step": 18803
-    },
-    {
-      "epoch": 1.4376971156603016,
-      "grad_norm": 0.0012032538652420044,
-      "learning_rate": 0.0001999989810744812,
-      "loss": 46.0,
-      "step": 18804
-    },
-    {
-      "epoch": 1.4377735726436913,
-      "grad_norm": 0.004058492835611105,
-      "learning_rate": 0.0001999989809660476,
-      "loss": 46.0,
-      "step": 18805
-    },
-    {
-      "epoch": 1.437850029627081,
-      "grad_norm": 0.0007875863229855895,
-      "learning_rate": 0.00019999898085760817,
-      "loss": 46.0,
-      "step": 18806
-    },
-    {
-      "epoch": 1.4379264866104708,
-      "grad_norm": 0.0020004352554678917,
-      "learning_rate": 0.00019999898074916299,
-      "loss": 46.0,
-      "step": 18807
-    },
-    {
-      "epoch": 1.4380029435938604,
-      "grad_norm": 0.0006731065222993493,
-      "learning_rate": 0.00019999898064071206,
-      "loss": 46.0,
-      "step": 18808
-    },
-    {
-      "epoch": 1.4380794005772501,
-      "grad_norm": 0.0009487764327786863,
-      "learning_rate": 0.00019999898053225533,
-      "loss": 46.0,
-      "step": 18809
-    },
-    {
-      "epoch": 1.43815585756064,
-      "grad_norm": 0.008388428017497063,
-      "learning_rate": 0.00019999898042379286,
-      "loss": 46.0,
-      "step": 18810
-    },
-    {
-      "epoch": 1.4382323145440297,
-      "grad_norm": 0.001875359215773642,
-      "learning_rate": 0.0001999989803153246,
-      "loss": 46.0,
-      "step": 18811
-    },
-    {
-      "epoch": 1.4383087715274194,
-      "grad_norm": 0.0007180998218245804,
-      "learning_rate": 0.00019999898020685056,
-      "loss": 46.0,
-      "step": 18812
-    },
-    {
-      "epoch": 1.4383852285108092,
-      "grad_norm": 0.0015995518770068884,
-      "learning_rate": 0.00019999898009837077,
-      "loss": 46.0,
-      "step": 18813
-    },
-    {
-      "epoch": 1.438461685494199,
-      "grad_norm": 0.002107905223965645,
-      "learning_rate": 0.0001999989799898852,
-      "loss": 46.0,
-      "step": 18814
-    },
-    {
-      "epoch": 1.4385381424775885,
-      "grad_norm": 0.001844717189669609,
-      "learning_rate": 0.00019999897988139389,
-      "loss": 46.0,
-      "step": 18815
-    },
-    {
-      "epoch": 1.4386145994609782,
-      "grad_norm": 0.0005767298280261457,
-      "learning_rate": 0.00019999897977289677,
-      "loss": 46.0,
-      "step": 18816
-    },
-    {
-      "epoch": 1.438691056444368,
-      "grad_norm": 0.0004050273564644158,
-      "learning_rate": 0.0001999989796643939,
-      "loss": 46.0,
-      "step": 18817
-    },
-    {
-      "epoch": 1.4387675134277578,
-      "grad_norm": 0.0009034311515279114,
-      "learning_rate": 0.00019999897955588528,
-      "loss": 46.0,
-      "step": 18818
-    },
-    {
-      "epoch": 1.4388439704111473,
-      "grad_norm": 0.0034355544485151768,
-      "learning_rate": 0.00019999897944737084,
-      "loss": 46.0,
-      "step": 18819
-    },
-    {
-      "epoch": 1.438920427394537,
-      "grad_norm": 0.0009344511199742556,
-      "learning_rate": 0.0001999989793388507,
-      "loss": 46.0,
-      "step": 18820
-    },
-    {
-      "epoch": 1.4389968843779268,
-      "grad_norm": 0.0022184192202985287,
-      "learning_rate": 0.00019999897923032473,
-      "loss": 46.0,
-      "step": 18821
-    },
-    {
-      "epoch": 1.4390733413613166,
-      "grad_norm": 0.00031321513233706355,
-      "learning_rate": 0.000199998979121793,
-      "loss": 46.0,
-      "step": 18822
-    },
-    {
-      "epoch": 1.4391497983447064,
-      "grad_norm": 0.004710352048277855,
-      "learning_rate": 0.00019999897901325553,
-      "loss": 46.0,
-      "step": 18823
-    },
-    {
-      "epoch": 1.4392262553280961,
-      "grad_norm": 0.006851649843156338,
-      "learning_rate": 0.00019999897890471226,
-      "loss": 46.0,
-      "step": 18824
-    },
-    {
-      "epoch": 1.4393027123114859,
-      "grad_norm": 0.001001182128675282,
-      "learning_rate": 0.00019999897879616324,
-      "loss": 46.0,
-      "step": 18825
-    },
-    {
-      "epoch": 1.4393791692948754,
-      "grad_norm": 0.001211144495755434,
-      "learning_rate": 0.00019999897868760847,
-      "loss": 46.0,
-      "step": 18826
-    },
-    {
-      "epoch": 1.4394556262782652,
-      "grad_norm": 0.0019999388605356216,
-      "learning_rate": 0.00019999897857904788,
-      "loss": 46.0,
-      "step": 18827
-    },
-    {
-      "epoch": 1.439532083261655,
-      "grad_norm": 0.005038689821958542,
-      "learning_rate": 0.00019999897847048154,
-      "loss": 46.0,
-      "step": 18828
-    },
-    {
-      "epoch": 1.4396085402450447,
-      "grad_norm": 0.0009320495883002877,
-      "learning_rate": 0.00019999897836190945,
-      "loss": 46.0,
-      "step": 18829
-    },
-    {
-      "epoch": 1.4396849972284342,
-      "grad_norm": 0.0018318586517125368,
-      "learning_rate": 0.0001999989782533316,
-      "loss": 46.0,
-      "step": 18830
-    },
-    {
-      "epoch": 1.439761454211824,
-      "grad_norm": 0.0012748222798109055,
-      "learning_rate": 0.00019999897814474793,
-      "loss": 46.0,
-      "step": 18831
-    },
-    {
-      "epoch": 1.4398379111952138,
-      "grad_norm": 0.0051794955506920815,
-      "learning_rate": 0.0001999989780361585,
-      "loss": 46.0,
-      "step": 18832
-    },
-    {
-      "epoch": 1.4399143681786035,
-      "grad_norm": 0.0009334392962045968,
-      "learning_rate": 0.00019999897792756331,
-      "loss": 46.0,
-      "step": 18833
-    },
-    {
-      "epoch": 1.4399908251619933,
-      "grad_norm": 0.0013264057924970984,
-      "learning_rate": 0.00019999897781896236,
-      "loss": 46.0,
-      "step": 18834
-    },
-    {
-      "epoch": 1.440067282145383,
-      "grad_norm": 0.0003454741963651031,
-      "learning_rate": 0.00019999897771035566,
-      "loss": 46.0,
-      "step": 18835
-    },
-    {
-      "epoch": 1.4401437391287728,
-      "grad_norm": 0.0011132069630548358,
-      "learning_rate": 0.0001999989776017432,
-      "loss": 46.0,
-      "step": 18836
-    },
-    {
-      "epoch": 1.4402201961121623,
-      "grad_norm": 0.0006953852134756744,
-      "learning_rate": 0.00019999897749312494,
-      "loss": 46.0,
-      "step": 18837
-    },
-    {
-      "epoch": 1.440296653095552,
-      "grad_norm": 0.0011303108185529709,
-      "learning_rate": 0.0001999989773845009,
-      "loss": 46.0,
-      "step": 18838
-    },
-    {
-      "epoch": 1.4403731100789419,
-      "grad_norm": 0.0009725252748467028,
-      "learning_rate": 0.0001999989772758711,
-      "loss": 46.0,
-      "step": 18839
-    },
-    {
-      "epoch": 1.4404495670623316,
-      "grad_norm": 0.001734154182486236,
-      "learning_rate": 0.00019999897716723553,
-      "loss": 46.0,
-      "step": 18840
-    },
-    {
-      "epoch": 1.4405260240457212,
-      "grad_norm": 0.0009559121681377292,
-      "learning_rate": 0.0001999989770585942,
-      "loss": 46.0,
-      "step": 18841
-    },
-    {
-      "epoch": 1.440602481029111,
-      "grad_norm": 0.00429040752351284,
-      "learning_rate": 0.00019999897694994708,
-      "loss": 46.0,
-      "step": 18842
-    },
-    {
-      "epoch": 1.4406789380125007,
-      "grad_norm": 0.0003997149469796568,
-      "learning_rate": 0.0001999989768412942,
-      "loss": 46.0,
-      "step": 18843
-    },
-    {
-      "epoch": 1.4407553949958904,
-      "grad_norm": 0.007469318341463804,
-      "learning_rate": 0.00019999897673263556,
-      "loss": 46.0,
-      "step": 18844
-    },
-    {
-      "epoch": 1.4408318519792802,
-      "grad_norm": 0.003976054023951292,
-      "learning_rate": 0.00019999897662397115,
-      "loss": 46.0,
-      "step": 18845
-    },
-    {
-      "epoch": 1.44090830896267,
-      "grad_norm": 0.0007124468102119863,
-      "learning_rate": 0.00019999897651530097,
-      "loss": 46.0,
-      "step": 18846
-    },
-    {
-      "epoch": 1.4409847659460597,
-      "grad_norm": 0.0011592571390792727,
-      "learning_rate": 0.00019999897640662502,
-      "loss": 46.0,
-      "step": 18847
-    },
-    {
-      "epoch": 1.4410612229294493,
-      "grad_norm": 0.0010226726299151778,
-      "learning_rate": 0.0001999989762979433,
-      "loss": 46.0,
-      "step": 18848
-    },
-    {
-      "epoch": 1.441137679912839,
-      "grad_norm": 0.0007074595778249204,
-      "learning_rate": 0.0001999989761892558,
-      "loss": 46.0,
-      "step": 18849
-    },
-    {
-      "epoch": 1.4412141368962288,
-      "grad_norm": 0.001997965620830655,
-      "learning_rate": 0.00019999897608056255,
-      "loss": 46.0,
-      "step": 18850
-    },
-    {
-      "epoch": 1.4412905938796186,
-      "grad_norm": 0.0005131909274496138,
-      "learning_rate": 0.0001999989759718635,
-      "loss": 46.0,
-      "step": 18851
-    },
-    {
-      "epoch": 1.441367050863008,
-      "grad_norm": 0.0017190531361848116,
-      "learning_rate": 0.00019999897586315873,
-      "loss": 46.0,
-      "step": 18852
-    },
-    {
-      "epoch": 1.4414435078463979,
-      "grad_norm": 0.001054232008755207,
-      "learning_rate": 0.00019999897575444814,
-      "loss": 46.0,
-      "step": 18853
-    },
-    {
-      "epoch": 1.4415199648297876,
-      "grad_norm": 0.001867811195552349,
-      "learning_rate": 0.0001999989756457318,
-      "loss": 46.0,
-      "step": 18854
-    },
-    {
-      "epoch": 1.4415964218131774,
-      "grad_norm": 0.0007233214564621449,
-      "learning_rate": 0.0001999989755370097,
-      "loss": 46.0,
-      "step": 18855
-    },
-    {
-      "epoch": 1.4416728787965671,
-      "grad_norm": 0.004756170324981213,
-      "learning_rate": 0.00019999897542828183,
-      "loss": 46.0,
-      "step": 18856
-    },
-    {
-      "epoch": 1.441749335779957,
-      "grad_norm": 0.0009767322335392237,
-      "learning_rate": 0.0001999989753195482,
-      "loss": 46.0,
-      "step": 18857
-    },
-    {
-      "epoch": 1.4418257927633467,
-      "grad_norm": 0.0009141616756096482,
-      "learning_rate": 0.00019999897521080876,
-      "loss": 46.0,
-      "step": 18858
-    },
-    {
-      "epoch": 1.4419022497467362,
-      "grad_norm": 0.0006841244176030159,
-      "learning_rate": 0.00019999897510206358,
-      "loss": 46.0,
-      "step": 18859
-    },
-    {
-      "epoch": 1.441978706730126,
-      "grad_norm": 0.000625805405434221,
-      "learning_rate": 0.00019999897499331263,
-      "loss": 46.0,
-      "step": 18860
-    },
-    {
-      "epoch": 1.4420551637135157,
-      "grad_norm": 0.005052138119935989,
-      "learning_rate": 0.00019999897488455587,
-      "loss": 46.0,
-      "step": 18861
-    },
-    {
-      "epoch": 1.4421316206969055,
-      "grad_norm": 0.0003998340980615467,
-      "learning_rate": 0.0001999989747757934,
-      "loss": 46.0,
-      "step": 18862
-    },
-    {
-      "epoch": 1.442208077680295,
-      "grad_norm": 0.0015206149546429515,
-      "learning_rate": 0.00019999897466702513,
-      "loss": 46.0,
-      "step": 18863
-    },
-    {
-      "epoch": 1.4422845346636848,
-      "grad_norm": 0.000785012380219996,
-      "learning_rate": 0.0001999989745582511,
-      "loss": 46.0,
-      "step": 18864
-    },
-    {
-      "epoch": 1.4423609916470745,
-      "grad_norm": 0.0013361851451918483,
-      "learning_rate": 0.0001999989744494713,
-      "loss": 46.0,
-      "step": 18865
-    },
-    {
-      "epoch": 1.4424374486304643,
-      "grad_norm": 0.00223688711412251,
-      "learning_rate": 0.00019999897434068572,
-      "loss": 46.0,
-      "step": 18866
-    },
-    {
-      "epoch": 1.442513905613854,
-      "grad_norm": 0.0008344555390067399,
-      "learning_rate": 0.0001999989742318944,
-      "loss": 46.0,
-      "step": 18867
-    },
-    {
-      "epoch": 1.4425903625972438,
-      "grad_norm": 0.0009108957601711154,
-      "learning_rate": 0.0001999989741230973,
-      "loss": 46.0,
-      "step": 18868
-    },
-    {
-      "epoch": 1.4426668195806334,
-      "grad_norm": 0.001850196160376072,
-      "learning_rate": 0.0001999989740142944,
-      "loss": 46.0,
-      "step": 18869
-    },
-    {
-      "epoch": 1.4427432765640231,
-      "grad_norm": 0.0010056825121864676,
-      "learning_rate": 0.00019999897390548578,
-      "loss": 46.0,
-      "step": 18870
-    },
-    {
-      "epoch": 1.4428197335474129,
-      "grad_norm": 0.0005432313773781061,
-      "learning_rate": 0.00019999897379667134,
-      "loss": 46.0,
-      "step": 18871
-    },
-    {
-      "epoch": 1.4428961905308026,
-      "grad_norm": 0.0024772018659859896,
-      "learning_rate": 0.00019999897368785114,
-      "loss": 46.0,
-      "step": 18872
-    },
-    {
-      "epoch": 1.4429726475141924,
-      "grad_norm": 0.003192031057551503,
-      "learning_rate": 0.00019999897357902518,
-      "loss": 46.0,
-      "step": 18873
-    },
-    {
-      "epoch": 1.443049104497582,
-      "grad_norm": 0.0018787558656185865,
-      "learning_rate": 0.00019999897347019348,
-      "loss": 46.0,
-      "step": 18874
-    },
-    {
-      "epoch": 1.4431255614809717,
-      "grad_norm": 0.003805401735007763,
-      "learning_rate": 0.00019999897336135596,
-      "loss": 46.0,
-      "step": 18875
-    },
-    {
-      "epoch": 1.4432020184643615,
-      "grad_norm": 0.0006748136365786195,
-      "learning_rate": 0.0001999989732525127,
-      "loss": 46.0,
-      "step": 18876
-    },
-    {
-      "epoch": 1.4432784754477512,
-      "grad_norm": 0.0007874740986153483,
-      "learning_rate": 0.00019999897314366367,
-      "loss": 46.0,
-      "step": 18877
-    },
-    {
-      "epoch": 1.443354932431141,
-      "grad_norm": 0.005053788423538208,
-      "learning_rate": 0.00019999897303480885,
-      "loss": 46.0,
-      "step": 18878
-    },
-    {
-      "epoch": 1.4434313894145308,
-      "grad_norm": 0.0004960333462804556,
-      "learning_rate": 0.00019999897292594828,
-      "loss": 46.0,
-      "step": 18879
-    },
-    {
-      "epoch": 1.4435078463979203,
-      "grad_norm": 0.005428021308034658,
-      "learning_rate": 0.00019999897281708194,
-      "loss": 46.0,
-      "step": 18880
-    },
-    {
-      "epoch": 1.44358430338131,
-      "grad_norm": 0.001701239263638854,
-      "learning_rate": 0.00019999897270820986,
-      "loss": 46.0,
-      "step": 18881
-    },
-    {
-      "epoch": 1.4436607603646998,
-      "grad_norm": 0.0010987732093781233,
-      "learning_rate": 0.00019999897259933197,
-      "loss": 46.0,
-      "step": 18882
-    },
-    {
-      "epoch": 1.4437372173480896,
-      "grad_norm": 0.0034763775765895844,
-      "learning_rate": 0.0001999989724904483,
-      "loss": 46.0,
-      "step": 18883
-    },
-    {
-      "epoch": 1.4438136743314791,
-      "grad_norm": 0.0007811313262209296,
-      "learning_rate": 0.0001999989723815589,
-      "loss": 46.0,
-      "step": 18884
-    },
-    {
-      "epoch": 1.4438901313148689,
-      "grad_norm": 0.0023446024861186743,
-      "learning_rate": 0.0001999989722726637,
-      "loss": 46.0,
-      "step": 18885
-    },
-    {
-      "epoch": 1.4439665882982586,
-      "grad_norm": 0.0031363684684038162,
-      "learning_rate": 0.00019999897216376275,
-      "loss": 46.0,
-      "step": 18886
-    },
-    {
-      "epoch": 1.4440430452816484,
-      "grad_norm": 0.002317397389560938,
-      "learning_rate": 0.000199998972054856,
-      "loss": 46.0,
-      "step": 18887
-    },
-    {
-      "epoch": 1.4441195022650382,
-      "grad_norm": 0.000758698966819793,
-      "learning_rate": 0.0001999989719459435,
-      "loss": 46.0,
-      "step": 18888
-    },
-    {
-      "epoch": 1.444195959248428,
-      "grad_norm": 0.0005982097354717553,
-      "learning_rate": 0.00019999897183702525,
-      "loss": 46.0,
-      "step": 18889
-    },
-    {
-      "epoch": 1.4442724162318177,
-      "grad_norm": 0.0007652511703781784,
-      "learning_rate": 0.0001999989717281012,
-      "loss": 46.0,
-      "step": 18890
-    },
-    {
-      "epoch": 1.4443488732152072,
-      "grad_norm": 0.0014397187624126673,
-      "learning_rate": 0.0001999989716191714,
-      "loss": 46.0,
-      "step": 18891
-    },
-    {
-      "epoch": 1.444425330198597,
-      "grad_norm": 0.0007819320890121162,
-      "learning_rate": 0.00019999897151023582,
-      "loss": 46.0,
-      "step": 18892
-    },
-    {
-      "epoch": 1.4445017871819867,
-      "grad_norm": 0.0026882081292569637,
-      "learning_rate": 0.00019999897140129448,
-      "loss": 46.0,
-      "step": 18893
-    },
-    {
-      "epoch": 1.4445782441653765,
-      "grad_norm": 0.0012461873702704906,
-      "learning_rate": 0.00019999897129234737,
-      "loss": 46.0,
-      "step": 18894
-    },
-    {
-      "epoch": 1.444654701148766,
-      "grad_norm": 0.0004840005422011018,
-      "learning_rate": 0.00019999897118339448,
-      "loss": 46.0,
-      "step": 18895
-    },
-    {
-      "epoch": 1.4447311581321558,
-      "grad_norm": 0.000709084328263998,
-      "learning_rate": 0.00019999897107443585,
-      "loss": 46.0,
-      "step": 18896
-    },
-    {
-      "epoch": 1.4448076151155456,
-      "grad_norm": 0.002557893982157111,
-      "learning_rate": 0.0001999989709654714,
-      "loss": 46.0,
-      "step": 18897
-    },
-    {
-      "epoch": 1.4448840720989353,
-      "grad_norm": 0.0005654958658851683,
-      "learning_rate": 0.00019999897085650124,
-      "loss": 46.0,
-      "step": 18898
-    },
-    {
-      "epoch": 1.444960529082325,
-      "grad_norm": 0.001185168162919581,
-      "learning_rate": 0.00019999897074752526,
-      "loss": 46.0,
-      "step": 18899
-    },
-    {
-      "epoch": 1.4450369860657148,
-      "grad_norm": 0.0006020786240696907,
-      "learning_rate": 0.00019999897063854353,
-      "loss": 46.0,
-      "step": 18900
-    },
-    {
-      "epoch": 1.4451134430491046,
-      "grad_norm": 0.029584430158138275,
-      "learning_rate": 0.00019999897052955603,
-      "loss": 46.0,
-      "step": 18901
-    },
-    {
-      "epoch": 1.4451899000324941,
-      "grad_norm": 0.0012202764628455043,
-      "learning_rate": 0.00019999897042056276,
-      "loss": 46.0,
-      "step": 18902
-    },
-    {
-      "epoch": 1.445266357015884,
-      "grad_norm": 0.0004320077132433653,
-      "learning_rate": 0.00019999897031156371,
-      "loss": 46.0,
-      "step": 18903
-    },
-    {
-      "epoch": 1.4453428139992737,
-      "grad_norm": 0.0008896985091269016,
-      "learning_rate": 0.00019999897020255892,
-      "loss": 46.0,
-      "step": 18904
-    },
-    {
-      "epoch": 1.4454192709826634,
-      "grad_norm": 0.0007843612693250179,
-      "learning_rate": 0.00019999897009354833,
-      "loss": 46.0,
-      "step": 18905
-    },
-    {
-      "epoch": 1.445495727966053,
-      "grad_norm": 0.0011643936159089208,
-      "learning_rate": 0.000199998969984532,
-      "loss": 46.0,
-      "step": 18906
-    },
-    {
-      "epoch": 1.4455721849494427,
-      "grad_norm": 0.0007157287909649312,
-      "learning_rate": 0.00019999896987550985,
-      "loss": 46.0,
-      "step": 18907
-    },
-    {
-      "epoch": 1.4456486419328325,
-      "grad_norm": 0.0010913871228694916,
-      "learning_rate": 0.000199998969766482,
-      "loss": 46.0,
-      "step": 18908
-    },
-    {
-      "epoch": 1.4457250989162223,
-      "grad_norm": 0.0015709202270954847,
-      "learning_rate": 0.00019999896965744833,
-      "loss": 46.0,
-      "step": 18909
-    },
-    {
-      "epoch": 1.445801555899612,
-      "grad_norm": 0.003016118425875902,
-      "learning_rate": 0.00019999896954840888,
-      "loss": 46.0,
-      "step": 18910
-    },
-    {
-      "epoch": 1.4458780128830018,
-      "grad_norm": 0.0004957798519171774,
-      "learning_rate": 0.00019999896943936372,
-      "loss": 46.0,
-      "step": 18911
-    },
-    {
-      "epoch": 1.4459544698663915,
-      "grad_norm": 0.0101634431630373,
-      "learning_rate": 0.00019999896933031275,
-      "loss": 46.0,
-      "step": 18912
-    },
-    {
-      "epoch": 1.446030926849781,
-      "grad_norm": 0.001034548506140709,
-      "learning_rate": 0.000199998969221256,
-      "loss": 46.0,
-      "step": 18913
-    },
-    {
-      "epoch": 1.4461073838331708,
-      "grad_norm": 0.0015693142777308822,
-      "learning_rate": 0.00019999896911219352,
-      "loss": 46.0,
-      "step": 18914
-    },
-    {
-      "epoch": 1.4461838408165606,
-      "grad_norm": 0.0011236666468903422,
-      "learning_rate": 0.00019999896900312525,
-      "loss": 46.0,
-      "step": 18915
-    },
-    {
-      "epoch": 1.4462602977999504,
-      "grad_norm": 0.0019882076885551214,
-      "learning_rate": 0.0001999989688940512,
-      "loss": 46.0,
-      "step": 18916
-    },
-    {
-      "epoch": 1.44633675478334,
-      "grad_norm": 0.0009653542074374855,
-      "learning_rate": 0.00019999896878497139,
-      "loss": 46.0,
-      "step": 18917
-    },
-    {
-      "epoch": 1.4464132117667297,
-      "grad_norm": 0.0003363323921803385,
-      "learning_rate": 0.00019999896867588582,
-      "loss": 46.0,
-      "step": 18918
-    },
-    {
-      "epoch": 1.4464896687501194,
-      "grad_norm": 0.008416110649704933,
-      "learning_rate": 0.00019999896856679445,
-      "loss": 46.0,
-      "step": 18919
-    },
-    {
-      "epoch": 1.4465661257335092,
-      "grad_norm": 0.0011363705853000283,
-      "learning_rate": 0.00019999896845769734,
-      "loss": 46.0,
-      "step": 18920
-    },
-    {
-      "epoch": 1.446642582716899,
-      "grad_norm": 0.0014391785953193903,
-      "learning_rate": 0.00019999896834859446,
-      "loss": 46.0,
-      "step": 18921
-    },
-    {
-      "epoch": 1.4467190397002887,
-      "grad_norm": 0.002617859048768878,
-      "learning_rate": 0.00019999896823948577,
-      "loss": 46.0,
-      "step": 18922
-    },
-    {
-      "epoch": 1.4467954966836785,
-      "grad_norm": 0.0021110600791871548,
-      "learning_rate": 0.00019999896813037137,
-      "loss": 46.0,
-      "step": 18923
-    },
-    {
-      "epoch": 1.446871953667068,
-      "grad_norm": 0.0017390345456078649,
-      "learning_rate": 0.0001999989680212512,
-      "loss": 46.0,
-      "step": 18924
-    },
-    {
-      "epoch": 1.4469484106504578,
-      "grad_norm": 0.0017242267495021224,
-      "learning_rate": 0.0001999989679121252,
-      "loss": 46.0,
-      "step": 18925
-    },
-    {
-      "epoch": 1.4470248676338475,
-      "grad_norm": 0.0009134160936810076,
-      "learning_rate": 0.00019999896780299349,
-      "loss": 46.0,
-      "step": 18926
-    },
-    {
-      "epoch": 1.4471013246172373,
-      "grad_norm": 0.0040166620165109634,
-      "learning_rate": 0.000199998967693856,
-      "loss": 46.0,
-      "step": 18927
-    },
-    {
-      "epoch": 1.4471777816006268,
-      "grad_norm": 0.0008931373013183475,
-      "learning_rate": 0.00019999896758471272,
-      "loss": 46.0,
-      "step": 18928
-    },
-    {
-      "epoch": 1.4472542385840166,
-      "grad_norm": 0.0004858648171648383,
-      "learning_rate": 0.00019999896747556367,
-      "loss": 46.0,
-      "step": 18929
-    },
-    {
-      "epoch": 1.4473306955674063,
-      "grad_norm": 0.004265401978045702,
-      "learning_rate": 0.00019999896736640885,
-      "loss": 46.0,
-      "step": 18930
-    },
-    {
-      "epoch": 1.447407152550796,
-      "grad_norm": 0.0032247009221464396,
-      "learning_rate": 0.00019999896725724826,
-      "loss": 46.0,
-      "step": 18931
-    },
-    {
-      "epoch": 1.4474836095341859,
-      "grad_norm": 0.0016347304917871952,
-      "learning_rate": 0.0001999989671480819,
-      "loss": 46.0,
-      "step": 18932
-    },
-    {
-      "epoch": 1.4475600665175756,
-      "grad_norm": 0.001962438225746155,
-      "learning_rate": 0.00019999896703890978,
-      "loss": 46.0,
-      "step": 18933
-    },
-    {
-      "epoch": 1.4476365235009654,
-      "grad_norm": 0.0010024956427514553,
-      "learning_rate": 0.0001999989669297319,
-      "loss": 46.0,
-      "step": 18934
-    },
-    {
-      "epoch": 1.447712980484355,
-      "grad_norm": 0.001242920639924705,
-      "learning_rate": 0.00019999896682054824,
-      "loss": 46.0,
-      "step": 18935
-    },
-    {
-      "epoch": 1.4477894374677447,
-      "grad_norm": 0.004915470723062754,
-      "learning_rate": 0.00019999896671135884,
-      "loss": 46.0,
-      "step": 18936
-    },
-    {
-      "epoch": 1.4478658944511344,
-      "grad_norm": 0.0003239080833736807,
-      "learning_rate": 0.00019999896660216363,
-      "loss": 46.0,
-      "step": 18937
-    },
-    {
-      "epoch": 1.4479423514345242,
-      "grad_norm": 0.000989253749139607,
-      "learning_rate": 0.00019999896649296263,
-      "loss": 46.0,
-      "step": 18938
-    },
-    {
-      "epoch": 1.4480188084179137,
-      "grad_norm": 0.0006176682072691619,
-      "learning_rate": 0.0001999989663837559,
-      "loss": 46.0,
-      "step": 18939
-    },
-    {
-      "epoch": 1.4480952654013035,
-      "grad_norm": 0.0016268908511847258,
-      "learning_rate": 0.0001999989662745434,
-      "loss": 46.0,
-      "step": 18940
-    },
-    {
-      "epoch": 1.4481717223846933,
-      "grad_norm": 0.0005626184865832329,
-      "learning_rate": 0.00019999896616532513,
-      "loss": 46.0,
-      "step": 18941
-    },
-    {
-      "epoch": 1.448248179368083,
-      "grad_norm": 0.004932803101837635,
-      "learning_rate": 0.00019999896605610106,
-      "loss": 46.0,
-      "step": 18942
-    },
-    {
-      "epoch": 1.4483246363514728,
-      "grad_norm": 0.0011716191656887531,
-      "learning_rate": 0.00019999896594687127,
-      "loss": 46.0,
-      "step": 18943
-    },
-    {
-      "epoch": 1.4484010933348626,
-      "grad_norm": 0.0014760260237380862,
-      "learning_rate": 0.00019999896583763568,
-      "loss": 46.0,
-      "step": 18944
-    },
-    {
-      "epoch": 1.4484775503182523,
-      "grad_norm": 0.0014312447747215629,
-      "learning_rate": 0.0001999989657283943,
-      "loss": 46.0,
-      "step": 18945
-    },
-    {
-      "epoch": 1.4485540073016419,
-      "grad_norm": 0.0019218989182263613,
-      "learning_rate": 0.00019999896561914723,
-      "loss": 46.0,
-      "step": 18946
-    },
-    {
-      "epoch": 1.4486304642850316,
-      "grad_norm": 0.0011618298012763262,
-      "learning_rate": 0.0001999989655098943,
-      "loss": 46.0,
-      "step": 18947
-    },
-    {
-      "epoch": 1.4487069212684214,
-      "grad_norm": 0.0011045156279578805,
-      "learning_rate": 0.00019999896540063566,
-      "loss": 46.0,
-      "step": 18948
-    },
-    {
-      "epoch": 1.4487833782518111,
-      "grad_norm": 0.0010301691945642233,
-      "learning_rate": 0.00019999896529137123,
-      "loss": 46.0,
-      "step": 18949
-    },
-    {
-      "epoch": 1.4488598352352007,
-      "grad_norm": 0.0005085246521048248,
-      "learning_rate": 0.00019999896518210102,
-      "loss": 46.0,
-      "step": 18950
-    },
-    {
-      "epoch": 1.4489362922185904,
-      "grad_norm": 0.0012716705678030849,
-      "learning_rate": 0.00019999896507282504,
-      "loss": 46.0,
-      "step": 18951
-    },
-    {
-      "epoch": 1.4490127492019802,
-      "grad_norm": 0.0007319659925997257,
-      "learning_rate": 0.00019999896496354332,
-      "loss": 46.0,
-      "step": 18952
-    },
-    {
-      "epoch": 1.44908920618537,
-      "grad_norm": 0.0005724075599573553,
-      "learning_rate": 0.00019999896485425582,
-      "loss": 46.0,
-      "step": 18953
-    },
-    {
-      "epoch": 1.4491656631687597,
-      "grad_norm": 0.0008692061528563499,
-      "learning_rate": 0.00019999896474496252,
-      "loss": 46.0,
-      "step": 18954
-    },
-    {
-      "epoch": 1.4492421201521495,
-      "grad_norm": 0.0025127530097961426,
-      "learning_rate": 0.00019999896463566348,
-      "loss": 46.0,
-      "step": 18955
-    },
-    {
-      "epoch": 1.4493185771355392,
-      "grad_norm": 0.0008072750060819089,
-      "learning_rate": 0.00019999896452635866,
-      "loss": 46.0,
-      "step": 18956
-    },
-    {
-      "epoch": 1.4493950341189288,
-      "grad_norm": 0.001020344439893961,
-      "learning_rate": 0.00019999896441704807,
-      "loss": 46.0,
-      "step": 18957
-    },
-    {
-      "epoch": 1.4494714911023185,
-      "grad_norm": 0.003414277918636799,
-      "learning_rate": 0.0001999989643077317,
-      "loss": 46.0,
-      "step": 18958
-    },
-    {
-      "epoch": 1.4495479480857083,
-      "grad_norm": 0.0019378105644136667,
-      "learning_rate": 0.0001999989641984096,
-      "loss": 46.0,
-      "step": 18959
-    },
-    {
-      "epoch": 1.449624405069098,
-      "grad_norm": 0.000865251524373889,
-      "learning_rate": 0.0001999989640890817,
-      "loss": 46.0,
-      "step": 18960
-    },
-    {
-      "epoch": 1.4497008620524876,
-      "grad_norm": 0.0004280477878637612,
-      "learning_rate": 0.00019999896397974805,
-      "loss": 46.0,
-      "step": 18961
-    },
-    {
-      "epoch": 1.4497773190358774,
-      "grad_norm": 0.0020386280957609415,
-      "learning_rate": 0.0001999989638704086,
-      "loss": 46.0,
-      "step": 18962
-    },
-    {
-      "epoch": 1.4498537760192671,
-      "grad_norm": 0.000983902020379901,
-      "learning_rate": 0.0001999989637610634,
-      "loss": 46.0,
-      "step": 18963
-    },
-    {
-      "epoch": 1.4499302330026569,
-      "grad_norm": 0.0009496128186583519,
-      "learning_rate": 0.00019999896365171241,
-      "loss": 46.0,
-      "step": 18964
-    },
-    {
-      "epoch": 1.4500066899860466,
-      "grad_norm": 0.0017147546168416739,
-      "learning_rate": 0.0001999989635423557,
-      "loss": 46.0,
-      "step": 18965
-    },
-    {
-      "epoch": 1.4500831469694364,
-      "grad_norm": 0.0011938734678551555,
-      "learning_rate": 0.0001999989634329932,
-      "loss": 46.0,
-      "step": 18966
-    },
-    {
-      "epoch": 1.4501596039528262,
-      "grad_norm": 0.002902195556089282,
-      "learning_rate": 0.00019999896332362492,
-      "loss": 46.0,
-      "step": 18967
-    },
-    {
-      "epoch": 1.4502360609362157,
-      "grad_norm": 0.001075799111276865,
-      "learning_rate": 0.00019999896321425085,
-      "loss": 46.0,
-      "step": 18968
-    },
-    {
-      "epoch": 1.4503125179196055,
-      "grad_norm": 0.0005118598346598446,
-      "learning_rate": 0.00019999896310487103,
-      "loss": 46.0,
-      "step": 18969
-    },
-    {
-      "epoch": 1.4503889749029952,
-      "grad_norm": 0.0009796771919354796,
-      "learning_rate": 0.00019999896299548544,
-      "loss": 46.0,
-      "step": 18970
-    },
-    {
-      "epoch": 1.450465431886385,
-      "grad_norm": 0.0016069040866568685,
-      "learning_rate": 0.0001999989628860941,
-      "loss": 46.0,
-      "step": 18971
-    },
-    {
-      "epoch": 1.4505418888697745,
-      "grad_norm": 0.0007837492157705128,
-      "learning_rate": 0.00019999896277669697,
-      "loss": 46.0,
-      "step": 18972
-    },
-    {
-      "epoch": 1.4506183458531643,
-      "grad_norm": 0.0008090370683930814,
-      "learning_rate": 0.00019999896266729406,
-      "loss": 46.0,
-      "step": 18973
-    },
-    {
-      "epoch": 1.450694802836554,
-      "grad_norm": 0.0007583677652291954,
-      "learning_rate": 0.0001999989625578854,
-      "loss": 46.0,
-      "step": 18974
-    },
-    {
-      "epoch": 1.4507712598199438,
-      "grad_norm": 0.009012933820486069,
-      "learning_rate": 0.00019999896244847097,
-      "loss": 46.0,
-      "step": 18975
-    },
-    {
-      "epoch": 1.4508477168033336,
-      "grad_norm": 0.0018240658100694418,
-      "learning_rate": 0.00019999896233905076,
-      "loss": 46.0,
-      "step": 18976
-    },
-    {
-      "epoch": 1.4509241737867233,
-      "grad_norm": 0.0019005857175216079,
-      "learning_rate": 0.0001999989622296248,
-      "loss": 46.0,
-      "step": 18977
-    },
-    {
-      "epoch": 1.451000630770113,
-      "grad_norm": 0.00193652359303087,
-      "learning_rate": 0.00019999896212019304,
-      "loss": 46.0,
-      "step": 18978
-    },
-    {
-      "epoch": 1.4510770877535026,
-      "grad_norm": 0.001566311577335,
-      "learning_rate": 0.00019999896201075554,
-      "loss": 46.0,
-      "step": 18979
-    },
-    {
-      "epoch": 1.4511535447368924,
-      "grad_norm": 0.0009208136470988393,
-      "learning_rate": 0.00019999896190131224,
-      "loss": 46.0,
-      "step": 18980
-    },
-    {
-      "epoch": 1.4512300017202822,
-      "grad_norm": 0.0008543008589185774,
-      "learning_rate": 0.0001999989617918632,
-      "loss": 46.0,
-      "step": 18981
-    },
-    {
-      "epoch": 1.451306458703672,
-      "grad_norm": 0.0014147782931104302,
-      "learning_rate": 0.00019999896168240838,
-      "loss": 46.0,
-      "step": 18982
-    },
-    {
-      "epoch": 1.4513829156870615,
-      "grad_norm": 0.0014980046544224024,
-      "learning_rate": 0.0001999989615729478,
-      "loss": 46.0,
-      "step": 18983
-    },
-    {
-      "epoch": 1.4514593726704512,
-      "grad_norm": 0.0008350092102773488,
-      "learning_rate": 0.00019999896146348143,
-      "loss": 46.0,
-      "step": 18984
-    },
-    {
-      "epoch": 1.451535829653841,
-      "grad_norm": 0.0006741269025951624,
-      "learning_rate": 0.0001999989613540093,
-      "loss": 46.0,
-      "step": 18985
-    },
-    {
-      "epoch": 1.4516122866372307,
-      "grad_norm": 0.0005927701713517308,
-      "learning_rate": 0.00019999896124453138,
-      "loss": 46.0,
-      "step": 18986
-    },
-    {
-      "epoch": 1.4516887436206205,
-      "grad_norm": 0.0008082900894805789,
-      "learning_rate": 0.00019999896113504772,
-      "loss": 46.0,
-      "step": 18987
-    },
-    {
-      "epoch": 1.4517652006040103,
-      "grad_norm": 0.005529089365154505,
-      "learning_rate": 0.0001999989610255583,
-      "loss": 46.0,
-      "step": 18988
-    },
-    {
-      "epoch": 1.4518416575874,
-      "grad_norm": 0.0012249990832060575,
-      "learning_rate": 0.00019999896091606306,
-      "loss": 46.0,
-      "step": 18989
-    },
-    {
-      "epoch": 1.4519181145707896,
-      "grad_norm": 0.000879644590895623,
-      "learning_rate": 0.0001999989608065621,
-      "loss": 46.0,
-      "step": 18990
-    },
-    {
-      "epoch": 1.4519945715541793,
-      "grad_norm": 0.0020269686356186867,
-      "learning_rate": 0.00019999896069705536,
-      "loss": 46.0,
-      "step": 18991
-    },
-    {
-      "epoch": 1.452071028537569,
-      "grad_norm": 0.0027563904877752066,
-      "learning_rate": 0.00019999896058754284,
-      "loss": 46.0,
-      "step": 18992
-    },
-    {
-      "epoch": 1.4521474855209588,
-      "grad_norm": 0.0037514863070100546,
-      "learning_rate": 0.00019999896047802457,
-      "loss": 46.0,
-      "step": 18993
-    },
-    {
-      "epoch": 1.4522239425043484,
-      "grad_norm": 0.0015333560295403004,
-      "learning_rate": 0.0001999989603685005,
-      "loss": 46.0,
-      "step": 18994
-    },
-    {
-      "epoch": 1.4523003994877381,
-      "grad_norm": 0.0031711405608803034,
-      "learning_rate": 0.00019999896025897068,
-      "loss": 46.0,
-      "step": 18995
-    },
-    {
-      "epoch": 1.452376856471128,
-      "grad_norm": 0.0009732060716487467,
-      "learning_rate": 0.00019999896014943512,
-      "loss": 46.0,
-      "step": 18996
-    },
-    {
-      "epoch": 1.4524533134545177,
-      "grad_norm": 0.002221033675596118,
-      "learning_rate": 0.00019999896003989372,
-      "loss": 46.0,
-      "step": 18997
-    },
-    {
-      "epoch": 1.4525297704379074,
-      "grad_norm": 0.0010713287629187107,
-      "learning_rate": 0.00019999895993034662,
-      "loss": 46.0,
-      "step": 18998
-    },
-    {
-      "epoch": 1.4526062274212972,
-      "grad_norm": 0.003200495382770896,
-      "learning_rate": 0.0001999989598207937,
-      "loss": 46.0,
-      "step": 18999
-    },
-    {
-      "epoch": 1.4526826844046867,
-      "grad_norm": 0.000680546450894326,
-      "learning_rate": 0.00019999895971123502,
-      "loss": 46.0,
-      "step": 19000
-    },
-    {
-      "epoch": 1.4527591413880765,
-      "grad_norm": 0.0010694727534428239,
-      "learning_rate": 0.00019999895960167062,
-      "loss": 46.0,
-      "step": 19001
-    },
-    {
-      "epoch": 1.4528355983714663,
-      "grad_norm": 0.0013278444530442357,
-      "learning_rate": 0.0001999989594921004,
-      "loss": 46.0,
-      "step": 19002
-    },
-    {
-      "epoch": 1.452912055354856,
-      "grad_norm": 0.0005983138689771295,
-      "learning_rate": 0.0001999989593825244,
-      "loss": 46.0,
-      "step": 19003
-    },
-    {
-      "epoch": 1.4529885123382458,
-      "grad_norm": 0.0009986660443246365,
-      "learning_rate": 0.00019999895927294266,
-      "loss": 46.0,
-      "step": 19004
-    },
-    {
-      "epoch": 1.4530649693216353,
-      "grad_norm": 0.0006804827717132866,
-      "learning_rate": 0.00019999895916335514,
-      "loss": 46.0,
-      "step": 19005
-    },
-    {
-      "epoch": 1.453141426305025,
-      "grad_norm": 0.007900598458945751,
-      "learning_rate": 0.00019999895905376187,
-      "loss": 46.0,
-      "step": 19006
-    },
-    {
-      "epoch": 1.4532178832884148,
-      "grad_norm": 0.0008294080616906285,
-      "learning_rate": 0.0001999989589441628,
-      "loss": 46.0,
-      "step": 19007
-    },
-    {
-      "epoch": 1.4532943402718046,
-      "grad_norm": 0.017306895926594734,
-      "learning_rate": 0.00019999895883455798,
-      "loss": 46.0,
-      "step": 19008
-    },
-    {
-      "epoch": 1.4533707972551944,
-      "grad_norm": 0.004924267530441284,
-      "learning_rate": 0.0001999989587249474,
-      "loss": 46.0,
-      "step": 19009
-    },
-    {
-      "epoch": 1.4534472542385841,
-      "grad_norm": 0.0024102837778627872,
-      "learning_rate": 0.00019999895861533103,
-      "loss": 46.0,
-      "step": 19010
-    },
-    {
-      "epoch": 1.4535237112219737,
-      "grad_norm": 0.0006523790070787072,
-      "learning_rate": 0.0001999989585057089,
-      "loss": 46.0,
-      "step": 19011
-    },
-    {
-      "epoch": 1.4536001682053634,
-      "grad_norm": 0.01680338755249977,
-      "learning_rate": 0.00019999895839608098,
-      "loss": 46.0,
-      "step": 19012
-    },
-    {
-      "epoch": 1.4536766251887532,
-      "grad_norm": 0.0029436489567160606,
-      "learning_rate": 0.0001999989582864473,
-      "loss": 46.0,
-      "step": 19013
-    },
-    {
-      "epoch": 1.453753082172143,
-      "grad_norm": 0.0008789649582467973,
-      "learning_rate": 0.00019999895817680787,
-      "loss": 46.0,
-      "step": 19014
-    },
-    {
-      "epoch": 1.4538295391555325,
-      "grad_norm": 0.00046684793778695166,
-      "learning_rate": 0.00019999895806716266,
-      "loss": 46.0,
-      "step": 19015
-    },
-    {
-      "epoch": 1.4539059961389222,
-      "grad_norm": 0.0024449429474771023,
-      "learning_rate": 0.00019999895795751166,
-      "loss": 46.0,
-      "step": 19016
-    },
-    {
-      "epoch": 1.453982453122312,
-      "grad_norm": 0.0016013651620596647,
-      "learning_rate": 0.00019999895784785494,
-      "loss": 46.0,
-      "step": 19017
-    },
-    {
-      "epoch": 1.4540589101057018,
-      "grad_norm": 0.0036540080327540636,
-      "learning_rate": 0.0001999989577381924,
-      "loss": 46.0,
-      "step": 19018
-    },
-    {
-      "epoch": 1.4541353670890915,
-      "grad_norm": 0.003510485403239727,
-      "learning_rate": 0.00019999895762852412,
-      "loss": 46.0,
-      "step": 19019
-    },
-    {
-      "epoch": 1.4542118240724813,
-      "grad_norm": 0.013017495162785053,
-      "learning_rate": 0.00019999895751885008,
-      "loss": 46.0,
-      "step": 19020
-    },
-    {
-      "epoch": 1.454288281055871,
-      "grad_norm": 0.0017165071330964565,
-      "learning_rate": 0.00019999895740917023,
-      "loss": 46.0,
-      "step": 19021
-    },
-    {
-      "epoch": 1.4543647380392606,
-      "grad_norm": 0.0026019043289124966,
-      "learning_rate": 0.00019999895729948464,
-      "loss": 46.0,
-      "step": 19022
-    },
-    {
-      "epoch": 1.4544411950226503,
-      "grad_norm": 0.0004465322708711028,
-      "learning_rate": 0.00019999895718979328,
-      "loss": 46.0,
-      "step": 19023
-    },
-    {
-      "epoch": 1.45451765200604,
-      "grad_norm": 0.014807252213358879,
-      "learning_rate": 0.00019999895708009614,
-      "loss": 46.0,
-      "step": 19024
-    },
-    {
-      "epoch": 1.4545941089894299,
-      "grad_norm": 0.001151958596892655,
-      "learning_rate": 0.00019999895697039323,
-      "loss": 46.0,
-      "step": 19025
-    },
-    {
-      "epoch": 1.4546705659728194,
-      "grad_norm": 0.00103965203743428,
-      "learning_rate": 0.00019999895686068455,
-      "loss": 46.0,
-      "step": 19026
-    },
-    {
-      "epoch": 1.4547470229562092,
-      "grad_norm": 0.003957048524171114,
-      "learning_rate": 0.00019999895675097012,
-      "loss": 46.0,
-      "step": 19027
-    },
-    {
-      "epoch": 1.454823479939599,
-      "grad_norm": 0.0008143301820382476,
-      "learning_rate": 0.0001999989566412499,
-      "loss": 46.0,
-      "step": 19028
-    },
-    {
-      "epoch": 1.4548999369229887,
-      "grad_norm": 0.002168800914660096,
-      "learning_rate": 0.00019999895653152392,
-      "loss": 46.0,
-      "step": 19029
-    },
-    {
-      "epoch": 1.4549763939063785,
-      "grad_norm": 0.0007615119102410972,
-      "learning_rate": 0.00019999895642179217,
-      "loss": 46.0,
-      "step": 19030
-    },
-    {
-      "epoch": 1.4550528508897682,
-      "grad_norm": 0.0005023975973017514,
-      "learning_rate": 0.00019999895631205464,
-      "loss": 46.0,
-      "step": 19031
-    },
-    {
-      "epoch": 1.455129307873158,
-      "grad_norm": 0.004793549422174692,
-      "learning_rate": 0.00019999895620231135,
-      "loss": 46.0,
-      "step": 19032
-    },
-    {
-      "epoch": 1.4552057648565475,
-      "grad_norm": 0.0013992143794894218,
-      "learning_rate": 0.0001999989560925623,
-      "loss": 46.0,
-      "step": 19033
-    },
-    {
-      "epoch": 1.4552822218399373,
-      "grad_norm": 0.004013485740870237,
-      "learning_rate": 0.00019999895598280746,
-      "loss": 46.0,
-      "step": 19034
-    },
-    {
-      "epoch": 1.455358678823327,
-      "grad_norm": 0.0008225878700613976,
-      "learning_rate": 0.00019999895587304687,
-      "loss": 46.0,
-      "step": 19035
-    },
-    {
-      "epoch": 1.4554351358067168,
-      "grad_norm": 0.0013154208427295089,
-      "learning_rate": 0.00019999895576328048,
-      "loss": 46.0,
-      "step": 19036
-    },
-    {
-      "epoch": 1.4555115927901063,
-      "grad_norm": 0.0010862386552616954,
-      "learning_rate": 0.00019999895565350837,
-      "loss": 46.0,
-      "step": 19037
-    },
-    {
-      "epoch": 1.455588049773496,
-      "grad_norm": 0.0018405833980068564,
-      "learning_rate": 0.00019999895554373044,
-      "loss": 46.0,
-      "step": 19038
-    },
-    {
-      "epoch": 1.4556645067568859,
-      "grad_norm": 0.005517409183084965,
-      "learning_rate": 0.00019999895543394678,
-      "loss": 46.0,
-      "step": 19039
-    },
-    {
-      "epoch": 1.4557409637402756,
-      "grad_norm": 0.0025462207850068808,
-      "learning_rate": 0.00019999895532415732,
-      "loss": 46.0,
-      "step": 19040
-    },
-    {
-      "epoch": 1.4558174207236654,
-      "grad_norm": 0.0007382428157143295,
-      "learning_rate": 0.00019999895521436212,
-      "loss": 46.0,
-      "step": 19041
-    },
-    {
-      "epoch": 1.4558938777070551,
-      "grad_norm": 0.0011158391134813428,
-      "learning_rate": 0.00019999895510456112,
-      "loss": 46.0,
-      "step": 19042
-    },
-    {
-      "epoch": 1.455970334690445,
-      "grad_norm": 0.0013417394366115332,
-      "learning_rate": 0.0001999989549947544,
-      "loss": 46.0,
-      "step": 19043
-    },
-    {
-      "epoch": 1.4560467916738344,
-      "grad_norm": 0.0006569102988578379,
-      "learning_rate": 0.00019999895488494187,
-      "loss": 46.0,
-      "step": 19044
-    },
-    {
-      "epoch": 1.4561232486572242,
-      "grad_norm": 0.0022195298224687576,
-      "learning_rate": 0.00019999895477512355,
-      "loss": 46.0,
-      "step": 19045
-    },
-    {
-      "epoch": 1.456199705640614,
-      "grad_norm": 0.010601436719298363,
-      "learning_rate": 0.0001999989546652995,
-      "loss": 46.0,
-      "step": 19046
-    },
-    {
-      "epoch": 1.4562761626240037,
-      "grad_norm": 0.0007732558296993375,
-      "learning_rate": 0.00019999895455546967,
-      "loss": 46.0,
-      "step": 19047
-    },
-    {
-      "epoch": 1.4563526196073933,
-      "grad_norm": 0.000681392615661025,
-      "learning_rate": 0.00019999895444563408,
-      "loss": 46.0,
-      "step": 19048
-    },
-    {
-      "epoch": 1.456429076590783,
-      "grad_norm": 0.003260162193328142,
-      "learning_rate": 0.0001999989543357927,
-      "loss": 46.0,
-      "step": 19049
-    },
-    {
-      "epoch": 1.4565055335741728,
-      "grad_norm": 0.0014880695380270481,
-      "learning_rate": 0.00019999895422594555,
-      "loss": 46.0,
-      "step": 19050
-    },
-    {
-      "epoch": 1.4565819905575625,
-      "grad_norm": 0.0012432746589183807,
-      "learning_rate": 0.00019999895411609267,
-      "loss": 46.0,
-      "step": 19051
-    },
-    {
-      "epoch": 1.4566584475409523,
-      "grad_norm": 0.03235870227217674,
-      "learning_rate": 0.00019999895400623396,
-      "loss": 46.0,
-      "step": 19052
-    },
-    {
-      "epoch": 1.456734904524342,
-      "grad_norm": 0.001498910249210894,
-      "learning_rate": 0.00019999895389636953,
-      "loss": 46.0,
-      "step": 19053
-    },
-    {
-      "epoch": 1.4568113615077318,
-      "grad_norm": 0.0007728941855020821,
-      "learning_rate": 0.0001999989537864993,
-      "loss": 46.0,
-      "step": 19054
-    },
-    {
-      "epoch": 1.4568878184911214,
-      "grad_norm": 0.0005912454798817635,
-      "learning_rate": 0.0001999989536766233,
-      "loss": 46.0,
-      "step": 19055
-    },
-    {
-      "epoch": 1.4569642754745111,
-      "grad_norm": 0.0008412210736423731,
-      "learning_rate": 0.00019999895356674158,
-      "loss": 46.0,
-      "step": 19056
-    },
-    {
-      "epoch": 1.4570407324579009,
-      "grad_norm": 0.013740879483520985,
-      "learning_rate": 0.00019999895345685406,
-      "loss": 46.0,
-      "step": 19057
-    },
-    {
-      "epoch": 1.4571171894412907,
-      "grad_norm": 0.001597710303030908,
-      "learning_rate": 0.00019999895334696076,
-      "loss": 46.0,
-      "step": 19058
-    },
-    {
-      "epoch": 1.4571936464246802,
-      "grad_norm": 0.001536812400445342,
-      "learning_rate": 0.0001999989532370617,
-      "loss": 46.0,
-      "step": 19059
-    },
-    {
-      "epoch": 1.45727010340807,
-      "grad_norm": 0.0016400080639868975,
-      "learning_rate": 0.00019999895312715685,
-      "loss": 46.0,
-      "step": 19060
-    },
-    {
-      "epoch": 1.4573465603914597,
-      "grad_norm": 0.0024575062561780214,
-      "learning_rate": 0.00019999895301724623,
-      "loss": 46.0,
-      "step": 19061
-    },
-    {
-      "epoch": 1.4574230173748495,
-      "grad_norm": 0.013304968364536762,
-      "learning_rate": 0.00019999895290732987,
-      "loss": 46.0,
-      "step": 19062
-    },
-    {
-      "epoch": 1.4574994743582392,
-      "grad_norm": 0.0007338508148677647,
-      "learning_rate": 0.00019999895279740774,
-      "loss": 46.0,
-      "step": 19063
-    },
-    {
-      "epoch": 1.457575931341629,
-      "grad_norm": 0.0008978633559308946,
-      "learning_rate": 0.0001999989526874798,
-      "loss": 46.0,
-      "step": 19064
-    },
-    {
-      "epoch": 1.4576523883250188,
-      "grad_norm": 0.0007933261804282665,
-      "learning_rate": 0.00019999895257754612,
-      "loss": 46.0,
-      "step": 19065
-    },
-    {
-      "epoch": 1.4577288453084083,
-      "grad_norm": 0.0020673710387200117,
-      "learning_rate": 0.0001999989524676067,
-      "loss": 46.0,
-      "step": 19066
-    },
-    {
-      "epoch": 1.457805302291798,
-      "grad_norm": 0.0012937564169988036,
-      "learning_rate": 0.00019999895235766146,
-      "loss": 46.0,
-      "step": 19067
-    },
-    {
-      "epoch": 1.4578817592751878,
-      "grad_norm": 0.0028017491567879915,
-      "learning_rate": 0.00019999895224771049,
-      "loss": 46.0,
-      "step": 19068
-    },
-    {
-      "epoch": 1.4579582162585776,
-      "grad_norm": 0.006538200657814741,
-      "learning_rate": 0.00019999895213775374,
-      "loss": 46.0,
-      "step": 19069
-    },
-    {
-      "epoch": 1.4580346732419671,
-      "grad_norm": 0.0007572290487587452,
-      "learning_rate": 0.0001999989520277912,
-      "loss": 46.0,
-      "step": 19070
-    },
-    {
-      "epoch": 1.4581111302253569,
-      "grad_norm": 0.0012289370642974973,
-      "learning_rate": 0.0001999989519178229,
-      "loss": 46.0,
-      "step": 19071
-    },
-    {
-      "epoch": 1.4581875872087466,
-      "grad_norm": 0.0025619827210903168,
-      "learning_rate": 0.00019999895180784885,
-      "loss": 46.0,
-      "step": 19072
-    },
-    {
-      "epoch": 1.4582640441921364,
-      "grad_norm": 0.0011546517489477992,
-      "learning_rate": 0.000199998951697869,
-      "loss": 46.0,
-      "step": 19073
-    },
-    {
-      "epoch": 1.4583405011755262,
-      "grad_norm": 0.001893850276246667,
-      "learning_rate": 0.00019999895158788337,
-      "loss": 46.0,
-      "step": 19074
-    },
-    {
-      "epoch": 1.458416958158916,
-      "grad_norm": 0.00036115458351559937,
-      "learning_rate": 0.00019999895147789203,
-      "loss": 46.0,
-      "step": 19075
-    },
-    {
-      "epoch": 1.4584934151423057,
-      "grad_norm": 0.0012535308487713337,
-      "learning_rate": 0.00019999895136789487,
-      "loss": 46.0,
-      "step": 19076
-    },
-    {
-      "epoch": 1.4585698721256952,
-      "grad_norm": 0.009620990604162216,
-      "learning_rate": 0.00019999895125789196,
-      "loss": 46.0,
-      "step": 19077
-    },
-    {
-      "epoch": 1.458646329109085,
-      "grad_norm": 0.00458752503618598,
-      "learning_rate": 0.00019999895114788328,
-      "loss": 46.0,
-      "step": 19078
-    },
-    {
-      "epoch": 1.4587227860924747,
-      "grad_norm": 0.002012411365285516,
-      "learning_rate": 0.0001999989510378688,
-      "loss": 46.0,
-      "step": 19079
-    },
-    {
-      "epoch": 1.4587992430758645,
-      "grad_norm": 0.0032266732305288315,
-      "learning_rate": 0.0001999989509278486,
-      "loss": 46.0,
-      "step": 19080
-    },
-    {
-      "epoch": 1.458875700059254,
-      "grad_norm": 0.0014690555399283767,
-      "learning_rate": 0.0001999989508178226,
-      "loss": 46.0,
-      "step": 19081
-    },
-    {
-      "epoch": 1.4589521570426438,
-      "grad_norm": 0.0009337826631963253,
-      "learning_rate": 0.00019999895070779084,
-      "loss": 46.0,
-      "step": 19082
-    },
-    {
-      "epoch": 1.4590286140260336,
-      "grad_norm": 0.001110545010305941,
-      "learning_rate": 0.0001999989505977533,
-      "loss": 46.0,
-      "step": 19083
-    },
-    {
-      "epoch": 1.4591050710094233,
-      "grad_norm": 0.0039435699582099915,
-      "learning_rate": 0.00019999895048771003,
-      "loss": 46.0,
-      "step": 19084
-    },
-    {
-      "epoch": 1.459181527992813,
-      "grad_norm": 0.003430826822295785,
-      "learning_rate": 0.00019999895037766096,
-      "loss": 46.0,
-      "step": 19085
-    },
-    {
-      "epoch": 1.4592579849762028,
-      "grad_norm": 0.0004946888657286763,
-      "learning_rate": 0.00019999895026760612,
-      "loss": 46.0,
-      "step": 19086
-    },
-    {
-      "epoch": 1.4593344419595926,
-      "grad_norm": 0.0012747155269607902,
-      "learning_rate": 0.0001999989501575455,
-      "loss": 46.0,
-      "step": 19087
-    },
-    {
-      "epoch": 1.4594108989429821,
-      "grad_norm": 0.0010246805613860488,
-      "learning_rate": 0.00019999895004747911,
-      "loss": 46.0,
-      "step": 19088
-    },
-    {
-      "epoch": 1.459487355926372,
-      "grad_norm": 0.0024050110951066017,
-      "learning_rate": 0.000199998949937407,
-      "loss": 46.0,
-      "step": 19089
-    },
-    {
-      "epoch": 1.4595638129097617,
-      "grad_norm": 0.0014022196410223842,
-      "learning_rate": 0.00019999894982732907,
-      "loss": 46.0,
-      "step": 19090
-    },
-    {
-      "epoch": 1.4596402698931514,
-      "grad_norm": 0.004510878119617701,
-      "learning_rate": 0.00019999894971724536,
-      "loss": 46.0,
-      "step": 19091
-    },
-    {
-      "epoch": 1.459716726876541,
-      "grad_norm": 0.0013228988973423839,
-      "learning_rate": 0.00019999894960715593,
-      "loss": 46.0,
-      "step": 19092
-    },
-    {
-      "epoch": 1.4597931838599307,
-      "grad_norm": 0.003901219228282571,
-      "learning_rate": 0.0001999989494970607,
-      "loss": 46.0,
-      "step": 19093
-    },
-    {
-      "epoch": 1.4598696408433205,
-      "grad_norm": 0.0011969677871093154,
-      "learning_rate": 0.0001999989493869597,
-      "loss": 46.0,
-      "step": 19094
-    },
-    {
-      "epoch": 1.4599460978267103,
-      "grad_norm": 0.0005538155674003065,
-      "learning_rate": 0.00019999894927685293,
-      "loss": 46.0,
-      "step": 19095
-    },
-    {
-      "epoch": 1.4600225548101,
-      "grad_norm": 0.0014049196615815163,
-      "learning_rate": 0.00019999894916674044,
-      "loss": 46.0,
-      "step": 19096
-    },
-    {
-      "epoch": 1.4600990117934898,
-      "grad_norm": 0.00385579327121377,
-      "learning_rate": 0.00019999894905662214,
-      "loss": 46.0,
-      "step": 19097
-    },
-    {
-      "epoch": 1.4601754687768795,
-      "grad_norm": 0.0010067732073366642,
-      "learning_rate": 0.00019999894894649805,
-      "loss": 46.0,
-      "step": 19098
-    },
-    {
-      "epoch": 1.460251925760269,
-      "grad_norm": 0.0005056728259660304,
-      "learning_rate": 0.0001999989488363682,
-      "loss": 46.0,
-      "step": 19099
-    },
-    {
-      "epoch": 1.4603283827436588,
-      "grad_norm": 0.0012051729718223214,
-      "learning_rate": 0.0001999989487262326,
-      "loss": 46.0,
-      "step": 19100
-    },
-    {
-      "epoch": 1.4604048397270486,
-      "grad_norm": 0.0005369476857595146,
-      "learning_rate": 0.00019999894861609123,
-      "loss": 46.0,
-      "step": 19101
-    },
-    {
-      "epoch": 1.4604812967104384,
-      "grad_norm": 0.0015683204401284456,
-      "learning_rate": 0.00019999894850594407,
-      "loss": 46.0,
-      "step": 19102
-    },
-    {
-      "epoch": 1.460557753693828,
-      "grad_norm": 0.004127866122871637,
-      "learning_rate": 0.00019999894839579113,
-      "loss": 46.0,
-      "step": 19103
-    },
-    {
-      "epoch": 1.4606342106772177,
-      "grad_norm": 0.0008356585749424994,
-      "learning_rate": 0.00019999894828563245,
-      "loss": 46.0,
-      "step": 19104
-    },
-    {
-      "epoch": 1.4607106676606074,
-      "grad_norm": 0.0010605361312627792,
-      "learning_rate": 0.000199998948175468,
-      "loss": 46.0,
-      "step": 19105
-    },
-    {
-      "epoch": 1.4607871246439972,
-      "grad_norm": 0.0037868469953536987,
-      "learning_rate": 0.00019999894806529777,
-      "loss": 46.0,
-      "step": 19106
-    },
-    {
-      "epoch": 1.460863581627387,
-      "grad_norm": 0.0011186314513906837,
-      "learning_rate": 0.0001999989479551218,
-      "loss": 46.0,
-      "step": 19107
-    },
-    {
-      "epoch": 1.4609400386107767,
-      "grad_norm": 0.0052525680512189865,
-      "learning_rate": 0.00019999894784494005,
-      "loss": 46.0,
-      "step": 19108
-    },
-    {
-      "epoch": 1.4610164955941665,
-      "grad_norm": 0.003108471166342497,
-      "learning_rate": 0.0001999989477347525,
-      "loss": 46.0,
-      "step": 19109
-    },
-    {
-      "epoch": 1.461092952577556,
-      "grad_norm": 0.0016261577839031816,
-      "learning_rate": 0.00019999894762455918,
-      "loss": 46.0,
-      "step": 19110
-    },
-    {
-      "epoch": 1.4611694095609458,
-      "grad_norm": 0.0011216031853109598,
-      "learning_rate": 0.00019999894751436014,
-      "loss": 46.0,
-      "step": 19111
-    },
-    {
-      "epoch": 1.4612458665443355,
-      "grad_norm": 0.004440872464329004,
-      "learning_rate": 0.00019999894740415527,
-      "loss": 46.0,
-      "step": 19112
-    },
-    {
-      "epoch": 1.4613223235277253,
-      "grad_norm": 0.002481652656570077,
-      "learning_rate": 0.00019999894729394465,
-      "loss": 46.0,
-      "step": 19113
-    },
-    {
-      "epoch": 1.4613987805111148,
-      "grad_norm": 0.0017673727124929428,
-      "learning_rate": 0.0001999989471837283,
-      "loss": 46.0,
-      "step": 19114
-    },
-    {
-      "epoch": 1.4614752374945046,
-      "grad_norm": 0.0007568387663923204,
-      "learning_rate": 0.00019999894707350616,
-      "loss": 46.0,
-      "step": 19115
-    },
-    {
-      "epoch": 1.4615516944778943,
-      "grad_norm": 0.00285971793346107,
-      "learning_rate": 0.00019999894696327822,
-      "loss": 46.0,
-      "step": 19116
-    },
-    {
-      "epoch": 1.461628151461284,
-      "grad_norm": 0.002128881635144353,
-      "learning_rate": 0.00019999894685304452,
-      "loss": 46.0,
-      "step": 19117
-    },
-    {
-      "epoch": 1.4617046084446739,
-      "grad_norm": 0.0010401071049273014,
-      "learning_rate": 0.00019999894674280506,
-      "loss": 46.0,
-      "step": 19118
-    },
-    {
-      "epoch": 1.4617810654280636,
-      "grad_norm": 0.0010843260679394007,
-      "learning_rate": 0.00019999894663255983,
-      "loss": 46.0,
-      "step": 19119
-    },
-    {
-      "epoch": 1.4618575224114534,
-      "grad_norm": 0.00039553752867504954,
-      "learning_rate": 0.00019999894652230883,
-      "loss": 46.0,
-      "step": 19120
-    },
-    {
-      "epoch": 1.461933979394843,
-      "grad_norm": 0.0015113131375983357,
-      "learning_rate": 0.00019999894641205206,
-      "loss": 46.0,
-      "step": 19121
-    },
-    {
-      "epoch": 1.4620104363782327,
-      "grad_norm": 0.0009209364070557058,
-      "learning_rate": 0.00019999894630178954,
-      "loss": 46.0,
-      "step": 19122
-    },
-    {
-      "epoch": 1.4620868933616225,
-      "grad_norm": 0.0020361284259706736,
-      "learning_rate": 0.00019999894619152122,
-      "loss": 46.0,
-      "step": 19123
-    },
-    {
-      "epoch": 1.4621633503450122,
-      "grad_norm": 0.003316224319860339,
-      "learning_rate": 0.00019999894608124718,
-      "loss": 46.0,
-      "step": 19124
-    },
-    {
-      "epoch": 1.4622398073284018,
-      "grad_norm": 0.004525564145296812,
-      "learning_rate": 0.0001999989459709673,
-      "loss": 46.0,
-      "step": 19125
-    },
-    {
-      "epoch": 1.4623162643117915,
-      "grad_norm": 0.0013780089793726802,
-      "learning_rate": 0.0001999989458606817,
-      "loss": 46.0,
-      "step": 19126
-    },
-    {
-      "epoch": 1.4623927212951813,
-      "grad_norm": 0.0015095982234925032,
-      "learning_rate": 0.0001999989457503903,
-      "loss": 46.0,
-      "step": 19127
-    },
-    {
-      "epoch": 1.462469178278571,
-      "grad_norm": 0.0011300451587885618,
-      "learning_rate": 0.00019999894564009317,
-      "loss": 46.0,
-      "step": 19128
-    },
-    {
-      "epoch": 1.4625456352619608,
-      "grad_norm": 0.0009763225098140538,
-      "learning_rate": 0.00019999894552979024,
-      "loss": 46.0,
-      "step": 19129
-    },
-    {
-      "epoch": 1.4626220922453506,
-      "grad_norm": 0.0012847681064158678,
-      "learning_rate": 0.00019999894541948153,
-      "loss": 46.0,
-      "step": 19130
-    },
-    {
-      "epoch": 1.46269854922874,
-      "grad_norm": 0.0009895693510770798,
-      "learning_rate": 0.0001999989453091671,
-      "loss": 46.0,
-      "step": 19131
-    },
-    {
-      "epoch": 1.4627750062121299,
-      "grad_norm": 0.0012944055488333106,
-      "learning_rate": 0.00019999894519884688,
-      "loss": 46.0,
-      "step": 19132
-    },
-    {
-      "epoch": 1.4628514631955196,
-      "grad_norm": 0.0019849692471325397,
-      "learning_rate": 0.00019999894508852085,
-      "loss": 46.0,
-      "step": 19133
-    },
-    {
-      "epoch": 1.4629279201789094,
-      "grad_norm": 0.0009527428192086518,
-      "learning_rate": 0.0001999989449781891,
-      "loss": 46.0,
-      "step": 19134
-    },
-    {
-      "epoch": 1.4630043771622991,
-      "grad_norm": 0.0008851123275235295,
-      "learning_rate": 0.00019999894486785155,
-      "loss": 46.0,
-      "step": 19135
-    },
-    {
-      "epoch": 1.4630808341456887,
-      "grad_norm": 0.0009851803770288825,
-      "learning_rate": 0.00019999894475750823,
-      "loss": 46.0,
-      "step": 19136
-    },
-    {
-      "epoch": 1.4631572911290784,
-      "grad_norm": 0.0020150537602603436,
-      "learning_rate": 0.00019999894464715917,
-      "loss": 46.0,
-      "step": 19137
-    },
-    {
-      "epoch": 1.4632337481124682,
-      "grad_norm": 0.0011949003674089909,
-      "learning_rate": 0.00019999894453680433,
-      "loss": 46.0,
-      "step": 19138
-    },
-    {
-      "epoch": 1.463310205095858,
-      "grad_norm": 0.000945518200751394,
-      "learning_rate": 0.0001999989444264437,
-      "loss": 46.0,
-      "step": 19139
-    },
-    {
-      "epoch": 1.4633866620792477,
-      "grad_norm": 0.003644590498879552,
-      "learning_rate": 0.00019999894431607732,
-      "loss": 46.0,
-      "step": 19140
-    },
-    {
-      "epoch": 1.4634631190626375,
-      "grad_norm": 0.0043143476359546185,
-      "learning_rate": 0.00019999894420570516,
-      "loss": 46.0,
-      "step": 19141
-    },
-    {
-      "epoch": 1.463539576046027,
-      "grad_norm": 0.0015756627544760704,
-      "learning_rate": 0.00019999894409532726,
-      "loss": 46.0,
-      "step": 19142
-    },
-    {
-      "epoch": 1.4636160330294168,
-      "grad_norm": 0.0007920751231722534,
-      "learning_rate": 0.00019999894398494355,
-      "loss": 46.0,
-      "step": 19143
-    },
-    {
-      "epoch": 1.4636924900128065,
-      "grad_norm": 0.0004181976546533406,
-      "learning_rate": 0.0001999989438745541,
-      "loss": 46.0,
-      "step": 19144
-    },
-    {
-      "epoch": 1.4637689469961963,
-      "grad_norm": 0.0029525617137551308,
-      "learning_rate": 0.00019999894376415887,
-      "loss": 46.0,
-      "step": 19145
-    },
-    {
-      "epoch": 1.463845403979586,
-      "grad_norm": 0.003163623157888651,
-      "learning_rate": 0.00019999894365375784,
-      "loss": 46.0,
-      "step": 19146
-    },
-    {
-      "epoch": 1.4639218609629756,
-      "grad_norm": 0.0004567409341689199,
-      "learning_rate": 0.0001999989435433511,
-      "loss": 46.0,
-      "step": 19147
-    },
-    {
-      "epoch": 1.4639983179463654,
-      "grad_norm": 0.0003693280159495771,
-      "learning_rate": 0.00019999894343293855,
-      "loss": 46.0,
-      "step": 19148
-    },
-    {
-      "epoch": 1.4640747749297551,
-      "grad_norm": 0.00269862893037498,
-      "learning_rate": 0.00019999894332252023,
-      "loss": 46.0,
-      "step": 19149
-    },
-    {
-      "epoch": 1.464151231913145,
-      "grad_norm": 0.0007455289014615119,
-      "learning_rate": 0.00019999894321209616,
-      "loss": 46.0,
-      "step": 19150
-    },
-    {
-      "epoch": 1.4642276888965347,
-      "grad_norm": 0.003705678042024374,
-      "learning_rate": 0.00019999894310166632,
-      "loss": 46.0,
-      "step": 19151
-    },
-    {
-      "epoch": 1.4643041458799244,
-      "grad_norm": 0.0006435929099097848,
-      "learning_rate": 0.0001999989429912307,
-      "loss": 46.0,
-      "step": 19152
-    },
-    {
-      "epoch": 1.464380602863314,
-      "grad_norm": 0.0011181682348251343,
-      "learning_rate": 0.00019999894288078932,
-      "loss": 46.0,
-      "step": 19153
-    },
-    {
-      "epoch": 1.4644570598467037,
-      "grad_norm": 0.0007335380068980157,
-      "learning_rate": 0.00019999894277034216,
-      "loss": 46.0,
-      "step": 19154
-    },
-    {
-      "epoch": 1.4645335168300935,
-      "grad_norm": 0.000914985139388591,
-      "learning_rate": 0.0001999989426598892,
-      "loss": 46.0,
-      "step": 19155
-    },
-    {
-      "epoch": 1.4646099738134832,
-      "grad_norm": 0.0005888041923753917,
-      "learning_rate": 0.00019999894254943052,
-      "loss": 46.0,
-      "step": 19156
-    },
-    {
-      "epoch": 1.4646864307968728,
-      "grad_norm": 0.0014561886200681329,
-      "learning_rate": 0.00019999894243896607,
-      "loss": 46.0,
-      "step": 19157
-    },
-    {
-      "epoch": 1.4647628877802625,
-      "grad_norm": 0.017310000956058502,
-      "learning_rate": 0.00019999894232849584,
-      "loss": 46.0,
-      "step": 19158
-    },
-    {
-      "epoch": 1.4648393447636523,
-      "grad_norm": 0.00172267179004848,
-      "learning_rate": 0.0001999989422180198,
-      "loss": 46.0,
-      "step": 19159
-    },
-    {
-      "epoch": 1.464915801747042,
-      "grad_norm": 0.001299018389545381,
-      "learning_rate": 0.00019999894210753804,
-      "loss": 46.0,
-      "step": 19160
-    },
-    {
-      "epoch": 1.4649922587304318,
-      "grad_norm": 0.003932367078959942,
-      "learning_rate": 0.0001999989419970505,
-      "loss": 46.0,
-      "step": 19161
-    },
-    {
-      "epoch": 1.4650687157138216,
-      "grad_norm": 0.0007361496100202203,
-      "learning_rate": 0.0001999989418865572,
-      "loss": 46.0,
-      "step": 19162
-    },
-    {
-      "epoch": 1.4651451726972113,
-      "grad_norm": 0.003676505759358406,
-      "learning_rate": 0.0001999989417760581,
-      "loss": 46.0,
-      "step": 19163
-    },
-    {
-      "epoch": 1.4652216296806009,
-      "grad_norm": 0.0025701585691422224,
-      "learning_rate": 0.00019999894166555326,
-      "loss": 46.0,
-      "step": 19164
-    },
-    {
-      "epoch": 1.4652980866639906,
-      "grad_norm": 0.0011493987403810024,
-      "learning_rate": 0.00019999894155504265,
-      "loss": 46.0,
-      "step": 19165
-    },
-    {
-      "epoch": 1.4653745436473804,
-      "grad_norm": 0.001910841092467308,
-      "learning_rate": 0.00019999894144452624,
-      "loss": 46.0,
-      "step": 19166
-    },
-    {
-      "epoch": 1.4654510006307702,
-      "grad_norm": 0.0006643445231020451,
-      "learning_rate": 0.0001999989413340041,
-      "loss": 46.0,
-      "step": 19167
-    },
-    {
-      "epoch": 1.4655274576141597,
-      "grad_norm": 0.0007738910499028862,
-      "learning_rate": 0.00019999894122347617,
-      "loss": 46.0,
-      "step": 19168
-    },
-    {
-      "epoch": 1.4656039145975495,
-      "grad_norm": 0.0019675877410918474,
-      "learning_rate": 0.00019999894111294246,
-      "loss": 46.0,
-      "step": 19169
-    },
-    {
-      "epoch": 1.4656803715809392,
-      "grad_norm": 0.0012034904211759567,
-      "learning_rate": 0.000199998941002403,
-      "loss": 46.0,
-      "step": 19170
-    },
-    {
-      "epoch": 1.465756828564329,
-      "grad_norm": 0.00350357498973608,
-      "learning_rate": 0.00019999894089185776,
-      "loss": 46.0,
-      "step": 19171
-    },
-    {
-      "epoch": 1.4658332855477187,
-      "grad_norm": 0.0006386736058630049,
-      "learning_rate": 0.00019999894078130676,
-      "loss": 46.0,
-      "step": 19172
-    },
-    {
-      "epoch": 1.4659097425311085,
-      "grad_norm": 0.0006145481602288783,
-      "learning_rate": 0.00019999894067075,
-      "loss": 46.0,
-      "step": 19173
-    },
-    {
-      "epoch": 1.4659861995144983,
-      "grad_norm": 0.006934434175491333,
-      "learning_rate": 0.00019999894056018744,
-      "loss": 46.0,
-      "step": 19174
-    },
-    {
-      "epoch": 1.4660626564978878,
-      "grad_norm": 0.001964472234249115,
-      "learning_rate": 0.00019999894044961915,
-      "loss": 46.0,
-      "step": 19175
-    },
-    {
-      "epoch": 1.4661391134812776,
-      "grad_norm": 0.0022962279617786407,
-      "learning_rate": 0.00019999894033904505,
-      "loss": 46.0,
-      "step": 19176
-    },
-    {
-      "epoch": 1.4662155704646673,
-      "grad_norm": 0.0008476219372823834,
-      "learning_rate": 0.00019999894022846519,
-      "loss": 46.0,
-      "step": 19177
-    },
-    {
-      "epoch": 1.466292027448057,
-      "grad_norm": 0.005569208413362503,
-      "learning_rate": 0.00019999894011787957,
-      "loss": 46.0,
-      "step": 19178
-    },
-    {
-      "epoch": 1.4663684844314466,
-      "grad_norm": 0.004880499094724655,
-      "learning_rate": 0.00019999894000728819,
-      "loss": 46.0,
-      "step": 19179
-    },
-    {
-      "epoch": 1.4664449414148364,
-      "grad_norm": 0.00301688932813704,
-      "learning_rate": 0.000199998939896691,
-      "loss": 46.0,
-      "step": 19180
-    },
-    {
-      "epoch": 1.4665213983982262,
-      "grad_norm": 0.0043104891665279865,
-      "learning_rate": 0.0001999989397860881,
-      "loss": 46.0,
-      "step": 19181
-    },
-    {
-      "epoch": 1.466597855381616,
-      "grad_norm": 0.0006606430979445577,
-      "learning_rate": 0.0001999989396754794,
-      "loss": 46.0,
-      "step": 19182
-    },
-    {
-      "epoch": 1.4666743123650057,
-      "grad_norm": 0.0013644114369526505,
-      "learning_rate": 0.00019999893956486493,
-      "loss": 46.0,
-      "step": 19183
-    },
-    {
-      "epoch": 1.4667507693483954,
-      "grad_norm": 0.0006841550348326564,
-      "learning_rate": 0.00019999893945424468,
-      "loss": 46.0,
-      "step": 19184
-    },
-    {
-      "epoch": 1.4668272263317852,
-      "grad_norm": 0.0017201006412506104,
-      "learning_rate": 0.00019999893934361865,
-      "loss": 46.0,
-      "step": 19185
-    },
-    {
-      "epoch": 1.4669036833151747,
-      "grad_norm": 0.0004923130036331713,
-      "learning_rate": 0.00019999893923298688,
-      "loss": 46.0,
-      "step": 19186
-    },
-    {
-      "epoch": 1.4669801402985645,
-      "grad_norm": 0.0010550328297540545,
-      "learning_rate": 0.00019999893912234934,
-      "loss": 46.0,
-      "step": 19187
-    },
-    {
-      "epoch": 1.4670565972819543,
-      "grad_norm": 0.0011936826631426811,
-      "learning_rate": 0.00019999893901170604,
-      "loss": 46.0,
-      "step": 19188
-    },
-    {
-      "epoch": 1.467133054265344,
-      "grad_norm": 0.0018220498459413648,
-      "learning_rate": 0.00019999893890105695,
-      "loss": 46.0,
-      "step": 19189
-    },
-    {
-      "epoch": 1.4672095112487336,
-      "grad_norm": 0.0015751337632536888,
-      "learning_rate": 0.00019999893879040208,
-      "loss": 46.0,
-      "step": 19190
-    },
-    {
-      "epoch": 1.4672859682321233,
-      "grad_norm": 0.00822540931403637,
-      "learning_rate": 0.00019999893867974147,
-      "loss": 46.0,
-      "step": 19191
-    },
-    {
-      "epoch": 1.467362425215513,
-      "grad_norm": 0.0007038480835035443,
-      "learning_rate": 0.00019999893856907509,
-      "loss": 46.0,
-      "step": 19192
-    },
-    {
-      "epoch": 1.4674388821989028,
-      "grad_norm": 0.0032394074369221926,
-      "learning_rate": 0.00019999893845840293,
-      "loss": 46.0,
-      "step": 19193
-    },
-    {
-      "epoch": 1.4675153391822926,
-      "grad_norm": 0.00037621636874973774,
-      "learning_rate": 0.000199998938347725,
-      "loss": 46.0,
-      "step": 19194
-    },
-    {
-      "epoch": 1.4675917961656824,
-      "grad_norm": 0.0005880384123884141,
-      "learning_rate": 0.0001999989382370413,
-      "loss": 46.0,
-      "step": 19195
-    },
-    {
-      "epoch": 1.4676682531490721,
-      "grad_norm": 0.000750627601519227,
-      "learning_rate": 0.0001999989381263518,
-      "loss": 46.0,
-      "step": 19196
-    },
-    {
-      "epoch": 1.4677447101324617,
-      "grad_norm": 0.0022673553321510553,
-      "learning_rate": 0.00019999893801565655,
-      "loss": 46.0,
-      "step": 19197
-    },
-    {
-      "epoch": 1.4678211671158514,
-      "grad_norm": 0.002134941751137376,
-      "learning_rate": 0.00019999893790495556,
-      "loss": 46.0,
-      "step": 19198
-    },
-    {
-      "epoch": 1.4678976240992412,
-      "grad_norm": 0.0019374073017388582,
-      "learning_rate": 0.00019999893779424878,
-      "loss": 46.0,
-      "step": 19199
-    },
-    {
-      "epoch": 1.467974081082631,
-      "grad_norm": 0.002374977571889758,
-      "learning_rate": 0.00019999893768353624,
-      "loss": 46.0,
-      "step": 19200
-    },
-    {
-      "epoch": 1.4680505380660205,
-      "grad_norm": 0.002467510523274541,
-      "learning_rate": 0.00019999893757281792,
-      "loss": 46.0,
-      "step": 19201
-    },
-    {
-      "epoch": 1.4681269950494102,
-      "grad_norm": 0.0027188737876713276,
-      "learning_rate": 0.00019999893746209383,
-      "loss": 46.0,
-      "step": 19202
-    },
-    {
-      "epoch": 1.4682034520328,
-      "grad_norm": 0.0017604512395337224,
-      "learning_rate": 0.00019999893735136396,
-      "loss": 46.0,
-      "step": 19203
-    },
-    {
-      "epoch": 1.4682799090161898,
-      "grad_norm": 0.001243614824488759,
-      "learning_rate": 0.00019999893724062835,
-      "loss": 46.0,
-      "step": 19204
-    },
-    {
-      "epoch": 1.4683563659995795,
-      "grad_norm": 0.0015066171763464808,
-      "learning_rate": 0.00019999893712988693,
-      "loss": 46.0,
-      "step": 19205
-    },
-    {
-      "epoch": 1.4684328229829693,
-      "grad_norm": 0.0005808811401948333,
-      "learning_rate": 0.00019999893701913977,
-      "loss": 46.0,
-      "step": 19206
-    },
-    {
-      "epoch": 1.468509279966359,
-      "grad_norm": 0.006335408892482519,
-      "learning_rate": 0.00019999893690838684,
-      "loss": 46.0,
-      "step": 19207
-    },
-    {
-      "epoch": 1.4685857369497486,
-      "grad_norm": 0.0008780763600952923,
-      "learning_rate": 0.00019999893679762814,
-      "loss": 46.0,
-      "step": 19208
-    },
-    {
-      "epoch": 1.4686621939331383,
-      "grad_norm": 0.0014805751852691174,
-      "learning_rate": 0.00019999893668686366,
-      "loss": 46.0,
-      "step": 19209
-    },
-    {
-      "epoch": 1.468738650916528,
-      "grad_norm": 0.0006930132512934506,
-      "learning_rate": 0.00019999893657609343,
-      "loss": 46.0,
-      "step": 19210
-    },
-    {
-      "epoch": 1.4688151078999179,
-      "grad_norm": 0.0015620873309671879,
-      "learning_rate": 0.00019999893646531743,
-      "loss": 46.0,
-      "step": 19211
-    },
-    {
-      "epoch": 1.4688915648833074,
-      "grad_norm": 0.0011235555866733193,
-      "learning_rate": 0.00019999893635453563,
-      "loss": 46.0,
-      "step": 19212
-    },
-    {
-      "epoch": 1.4689680218666972,
-      "grad_norm": 0.0007692488725297153,
-      "learning_rate": 0.0001999989362437481,
-      "loss": 46.0,
-      "step": 19213
-    },
-    {
-      "epoch": 1.469044478850087,
-      "grad_norm": 0.006577357184141874,
-      "learning_rate": 0.00019999893613295474,
-      "loss": 46.0,
-      "step": 19214
-    },
-    {
-      "epoch": 1.4691209358334767,
-      "grad_norm": 0.002416061470285058,
-      "learning_rate": 0.00019999893602215568,
-      "loss": 46.0,
-      "step": 19215
-    },
-    {
-      "epoch": 1.4691973928168665,
-      "grad_norm": 0.0009787181625142694,
-      "learning_rate": 0.0001999989359113508,
-      "loss": 46.0,
-      "step": 19216
-    },
-    {
-      "epoch": 1.4692738498002562,
-      "grad_norm": 0.0017808849224820733,
-      "learning_rate": 0.0001999989358005402,
-      "loss": 46.0,
-      "step": 19217
-    },
-    {
-      "epoch": 1.469350306783646,
-      "grad_norm": 0.00116780458483845,
-      "learning_rate": 0.0001999989356897238,
-      "loss": 46.0,
-      "step": 19218
-    },
-    {
-      "epoch": 1.4694267637670355,
-      "grad_norm": 0.0007117206114344299,
-      "learning_rate": 0.00019999893557890163,
-      "loss": 46.0,
-      "step": 19219
-    },
-    {
-      "epoch": 1.4695032207504253,
-      "grad_norm": 0.0008770351414568722,
-      "learning_rate": 0.0001999989354680737,
-      "loss": 46.0,
-      "step": 19220
-    },
-    {
-      "epoch": 1.469579677733815,
-      "grad_norm": 0.001342743868008256,
-      "learning_rate": 0.00019999893535723996,
-      "loss": 46.0,
-      "step": 19221
-    },
-    {
-      "epoch": 1.4696561347172048,
-      "grad_norm": 0.0008404088439419866,
-      "learning_rate": 0.0001999989352464005,
-      "loss": 46.0,
-      "step": 19222
-    },
-    {
-      "epoch": 1.4697325917005943,
-      "grad_norm": 0.000537903280928731,
-      "learning_rate": 0.00019999893513555526,
-      "loss": 46.0,
-      "step": 19223
-    },
-    {
-      "epoch": 1.469809048683984,
-      "grad_norm": 0.0006145606748759747,
-      "learning_rate": 0.00019999893502470423,
-      "loss": 46.0,
-      "step": 19224
-    },
-    {
-      "epoch": 1.4698855056673739,
-      "grad_norm": 0.0010471732821315527,
-      "learning_rate": 0.00019999893491384743,
-      "loss": 46.0,
-      "step": 19225
-    },
-    {
-      "epoch": 1.4699619626507636,
-      "grad_norm": 0.0010731476359069347,
-      "learning_rate": 0.00019999893480298492,
-      "loss": 46.0,
-      "step": 19226
-    },
-    {
-      "epoch": 1.4700384196341534,
-      "grad_norm": 0.003976867068558931,
-      "learning_rate": 0.0001999989346921166,
-      "loss": 46.0,
-      "step": 19227
-    },
-    {
-      "epoch": 1.4701148766175431,
-      "grad_norm": 0.0024218300823122263,
-      "learning_rate": 0.00019999893458124248,
-      "loss": 46.0,
-      "step": 19228
-    },
-    {
-      "epoch": 1.470191333600933,
-      "grad_norm": 0.0007920422940514982,
-      "learning_rate": 0.00019999893447036264,
-      "loss": 46.0,
-      "step": 19229
-    },
-    {
-      "epoch": 1.4702677905843224,
-      "grad_norm": 0.002143591409549117,
-      "learning_rate": 0.000199998934359477,
-      "loss": 46.0,
-      "step": 19230
-    },
-    {
-      "epoch": 1.4703442475677122,
-      "grad_norm": 0.000491861894261092,
-      "learning_rate": 0.0001999989342485856,
-      "loss": 46.0,
-      "step": 19231
-    },
-    {
-      "epoch": 1.470420704551102,
-      "grad_norm": 0.0012849452905356884,
-      "learning_rate": 0.00019999893413768843,
-      "loss": 46.0,
-      "step": 19232
-    },
-    {
-      "epoch": 1.4704971615344917,
-      "grad_norm": 0.0015341893304139376,
-      "learning_rate": 0.0001999989340267855,
-      "loss": 46.0,
-      "step": 19233
-    },
-    {
-      "epoch": 1.4705736185178813,
-      "grad_norm": 0.00040873538819141686,
-      "learning_rate": 0.0001999989339158768,
-      "loss": 46.0,
-      "step": 19234
-    },
-    {
-      "epoch": 1.470650075501271,
-      "grad_norm": 0.000625201384536922,
-      "learning_rate": 0.00019999893380496232,
-      "loss": 46.0,
-      "step": 19235
-    },
-    {
-      "epoch": 1.4707265324846608,
-      "grad_norm": 0.0022729034535586834,
-      "learning_rate": 0.0001999989336940421,
-      "loss": 46.0,
-      "step": 19236
-    },
-    {
-      "epoch": 1.4708029894680505,
-      "grad_norm": 0.0008619332802481949,
-      "learning_rate": 0.00019999893358311607,
-      "loss": 46.0,
-      "step": 19237
-    },
-    {
-      "epoch": 1.4708794464514403,
-      "grad_norm": 0.0010886061936616898,
-      "learning_rate": 0.00019999893347218427,
-      "loss": 46.0,
-      "step": 19238
-    },
-    {
-      "epoch": 1.47095590343483,
-      "grad_norm": 0.0009552035480737686,
-      "learning_rate": 0.00019999893336124672,
-      "loss": 46.0,
-      "step": 19239
-    },
-    {
-      "epoch": 1.4710323604182198,
-      "grad_norm": 0.0009725200361572206,
-      "learning_rate": 0.00019999893325030338,
-      "loss": 46.0,
-      "step": 19240
-    },
-    {
-      "epoch": 1.4711088174016094,
-      "grad_norm": 0.001992037519812584,
-      "learning_rate": 0.00019999893313935428,
-      "loss": 46.0,
-      "step": 19241
-    },
-    {
-      "epoch": 1.4711852743849991,
-      "grad_norm": 0.00086099689360708,
-      "learning_rate": 0.00019999893302839942,
-      "loss": 46.0,
-      "step": 19242
-    },
-    {
-      "epoch": 1.471261731368389,
-      "grad_norm": 0.00737357884645462,
-      "learning_rate": 0.0001999989329174388,
-      "loss": 46.0,
-      "step": 19243
-    },
-    {
-      "epoch": 1.4713381883517787,
-      "grad_norm": 0.0017700770404189825,
-      "learning_rate": 0.0001999989328064724,
-      "loss": 46.0,
-      "step": 19244
-    },
-    {
-      "epoch": 1.4714146453351682,
-      "grad_norm": 0.0010289129568263888,
-      "learning_rate": 0.0001999989326955002,
-      "loss": 46.0,
-      "step": 19245
-    },
-    {
-      "epoch": 1.471491102318558,
-      "grad_norm": 0.0013079507043585181,
-      "learning_rate": 0.00019999893258452228,
-      "loss": 46.0,
-      "step": 19246
-    },
-    {
-      "epoch": 1.4715675593019477,
-      "grad_norm": 0.0015224862145259976,
-      "learning_rate": 0.00019999893247353855,
-      "loss": 46.0,
-      "step": 19247
-    },
-    {
-      "epoch": 1.4716440162853375,
-      "grad_norm": 0.0011894813505932689,
-      "learning_rate": 0.00019999893236254907,
-      "loss": 46.0,
-      "step": 19248
-    },
-    {
-      "epoch": 1.4717204732687272,
-      "grad_norm": 0.0008942015119828284,
-      "learning_rate": 0.00019999893225155384,
-      "loss": 46.0,
-      "step": 19249
-    },
-    {
-      "epoch": 1.471796930252117,
-      "grad_norm": 0.0007221175474114716,
-      "learning_rate": 0.00019999893214055282,
-      "loss": 46.0,
-      "step": 19250
-    },
-    {
-      "epoch": 1.4718733872355068,
-      "grad_norm": 0.0005632580723613501,
-      "learning_rate": 0.00019999893202954602,
-      "loss": 46.0,
-      "step": 19251
-    },
-    {
-      "epoch": 1.4719498442188963,
-      "grad_norm": 0.002287228126078844,
-      "learning_rate": 0.0001999989319185335,
-      "loss": 46.0,
-      "step": 19252
-    },
-    {
-      "epoch": 1.472026301202286,
-      "grad_norm": 0.004281922243535519,
-      "learning_rate": 0.00019999893180751513,
-      "loss": 46.0,
-      "step": 19253
-    },
-    {
-      "epoch": 1.4721027581856758,
-      "grad_norm": 0.0004845394869334996,
-      "learning_rate": 0.00019999893169649107,
-      "loss": 46.0,
-      "step": 19254
-    },
-    {
-      "epoch": 1.4721792151690656,
-      "grad_norm": 0.0028777297120541334,
-      "learning_rate": 0.0001999989315854612,
-      "loss": 46.0,
-      "step": 19255
-    },
-    {
-      "epoch": 1.4722556721524551,
-      "grad_norm": 0.000936887867283076,
-      "learning_rate": 0.00019999893147442554,
-      "loss": 46.0,
-      "step": 19256
-    },
-    {
-      "epoch": 1.4723321291358449,
-      "grad_norm": 0.0031726856250315905,
-      "learning_rate": 0.00019999893136338415,
-      "loss": 46.0,
-      "step": 19257
-    },
-    {
-      "epoch": 1.4724085861192346,
-      "grad_norm": 0.0007614083588123322,
-      "learning_rate": 0.00019999893125233697,
-      "loss": 46.0,
-      "step": 19258
-    },
-    {
-      "epoch": 1.4724850431026244,
-      "grad_norm": 0.004533240105956793,
-      "learning_rate": 0.00019999893114128404,
-      "loss": 46.0,
-      "step": 19259
-    },
-    {
-      "epoch": 1.4725615000860142,
-      "grad_norm": 0.0026064226403832436,
-      "learning_rate": 0.0001999989310302253,
-      "loss": 46.0,
-      "step": 19260
-    },
-    {
-      "epoch": 1.472637957069404,
-      "grad_norm": 0.0017717110458761454,
-      "learning_rate": 0.00019999893091916085,
-      "loss": 46.0,
-      "step": 19261
-    },
-    {
-      "epoch": 1.4727144140527937,
-      "grad_norm": 0.005900966934859753,
-      "learning_rate": 0.0001999989308080906,
-      "loss": 46.0,
-      "step": 19262
-    },
-    {
-      "epoch": 1.4727908710361832,
-      "grad_norm": 0.0006830496131442487,
-      "learning_rate": 0.00019999893069701455,
-      "loss": 46.0,
-      "step": 19263
-    },
-    {
-      "epoch": 1.472867328019573,
-      "grad_norm": 0.0004614562203641981,
-      "learning_rate": 0.00019999893058593278,
-      "loss": 46.0,
-      "step": 19264
-    },
-    {
-      "epoch": 1.4729437850029627,
-      "grad_norm": 0.0009629227570258081,
-      "learning_rate": 0.0001999989304748452,
-      "loss": 46.0,
-      "step": 19265
-    },
-    {
-      "epoch": 1.4730202419863525,
-      "grad_norm": 0.0014378144405782223,
-      "learning_rate": 0.0001999989303637519,
-      "loss": 46.0,
-      "step": 19266
-    },
-    {
-      "epoch": 1.473096698969742,
-      "grad_norm": 0.0012453735107555985,
-      "learning_rate": 0.0001999989302526528,
-      "loss": 46.0,
-      "step": 19267
-    },
-    {
-      "epoch": 1.4731731559531318,
-      "grad_norm": 0.0014825640246272087,
-      "learning_rate": 0.0001999989301415479,
-      "loss": 46.0,
-      "step": 19268
-    },
-    {
-      "epoch": 1.4732496129365216,
-      "grad_norm": 0.0011866736458614469,
-      "learning_rate": 0.00019999893003043727,
-      "loss": 46.0,
-      "step": 19269
-    },
-    {
-      "epoch": 1.4733260699199113,
-      "grad_norm": 0.0011308465618640184,
-      "learning_rate": 0.0001999989299193209,
-      "loss": 46.0,
-      "step": 19270
-    },
-    {
-      "epoch": 1.473402526903301,
-      "grad_norm": 0.0007920544594526291,
-      "learning_rate": 0.0001999989298081987,
-      "loss": 46.0,
-      "step": 19271
-    },
-    {
-      "epoch": 1.4734789838866909,
-      "grad_norm": 0.004185117781162262,
-      "learning_rate": 0.00019999892969707077,
-      "loss": 46.0,
-      "step": 19272
-    },
-    {
-      "epoch": 1.4735554408700804,
-      "grad_norm": 0.0012875179527327418,
-      "learning_rate": 0.00019999892958593704,
-      "loss": 46.0,
-      "step": 19273
-    },
-    {
-      "epoch": 1.4736318978534702,
-      "grad_norm": 0.001267644576728344,
-      "learning_rate": 0.00019999892947479756,
-      "loss": 46.0,
-      "step": 19274
-    },
-    {
-      "epoch": 1.47370835483686,
-      "grad_norm": 0.0005683074123226106,
-      "learning_rate": 0.0001999989293636523,
-      "loss": 46.0,
-      "step": 19275
-    },
-    {
-      "epoch": 1.4737848118202497,
-      "grad_norm": 0.002309659728780389,
-      "learning_rate": 0.0001999989292525013,
-      "loss": 46.0,
-      "step": 19276
-    },
-    {
-      "epoch": 1.4738612688036394,
-      "grad_norm": 0.0006356228259392083,
-      "learning_rate": 0.0001999989291413445,
-      "loss": 46.0,
-      "step": 19277
-    },
-    {
-      "epoch": 1.473937725787029,
-      "grad_norm": 0.0013314933748915792,
-      "learning_rate": 0.00019999892903018195,
-      "loss": 46.0,
-      "step": 19278
-    },
-    {
-      "epoch": 1.4740141827704187,
-      "grad_norm": 0.0023864246904850006,
-      "learning_rate": 0.00019999892891901363,
-      "loss": 46.0,
-      "step": 19279
-    },
-    {
-      "epoch": 1.4740906397538085,
-      "grad_norm": 0.0019366954220458865,
-      "learning_rate": 0.0001999989288078395,
-      "loss": 46.0,
-      "step": 19280
-    },
-    {
-      "epoch": 1.4741670967371983,
-      "grad_norm": 0.0012591093545779586,
-      "learning_rate": 0.00019999892869665965,
-      "loss": 46.0,
-      "step": 19281
-    },
-    {
-      "epoch": 1.474243553720588,
-      "grad_norm": 0.0012208328116685152,
-      "learning_rate": 0.000199998928585474,
-      "loss": 46.0,
-      "step": 19282
-    },
-    {
-      "epoch": 1.4743200107039778,
-      "grad_norm": 0.0017227797070518136,
-      "learning_rate": 0.0001999989284742826,
-      "loss": 46.0,
-      "step": 19283
-    },
-    {
-      "epoch": 1.4743964676873673,
-      "grad_norm": 0.002297386759892106,
-      "learning_rate": 0.0001999989283630854,
-      "loss": 46.0,
-      "step": 19284
-    },
-    {
-      "epoch": 1.474472924670757,
-      "grad_norm": 0.0006985300569795072,
-      "learning_rate": 0.00019999892825188246,
-      "loss": 46.0,
-      "step": 19285
-    },
-    {
-      "epoch": 1.4745493816541468,
-      "grad_norm": 0.006447529420256615,
-      "learning_rate": 0.00019999892814067372,
-      "loss": 46.0,
-      "step": 19286
-    },
-    {
-      "epoch": 1.4746258386375366,
-      "grad_norm": 0.0004198014212306589,
-      "learning_rate": 0.00019999892802945927,
-      "loss": 46.0,
-      "step": 19287
-    },
-    {
-      "epoch": 1.4747022956209261,
-      "grad_norm": 0.0024661975912749767,
-      "learning_rate": 0.000199998927918239,
-      "loss": 46.0,
-      "step": 19288
-    },
-    {
-      "epoch": 1.474778752604316,
-      "grad_norm": 0.0011110918130725622,
-      "learning_rate": 0.000199998927807013,
-      "loss": 46.0,
-      "step": 19289
-    },
-    {
-      "epoch": 1.4748552095877057,
-      "grad_norm": 0.002359352307394147,
-      "learning_rate": 0.0001999989276957812,
-      "loss": 46.0,
-      "step": 19290
-    },
-    {
-      "epoch": 1.4749316665710954,
-      "grad_norm": 0.00275266170501709,
-      "learning_rate": 0.0001999989275845436,
-      "loss": 46.0,
-      "step": 19291
-    },
-    {
-      "epoch": 1.4750081235544852,
-      "grad_norm": 0.007068646140396595,
-      "learning_rate": 0.0001999989274733003,
-      "loss": 46.0,
-      "step": 19292
-    },
-    {
-      "epoch": 1.475084580537875,
-      "grad_norm": 0.0005153819220140576,
-      "learning_rate": 0.0001999989273620512,
-      "loss": 46.0,
-      "step": 19293
-    },
-    {
-      "epoch": 1.4751610375212647,
-      "grad_norm": 0.0012687172275036573,
-      "learning_rate": 0.00019999892725079633,
-      "loss": 46.0,
-      "step": 19294
-    },
-    {
-      "epoch": 1.4752374945046542,
-      "grad_norm": 0.00040213772444985807,
-      "learning_rate": 0.00019999892713953567,
-      "loss": 46.0,
-      "step": 19295
-    },
-    {
-      "epoch": 1.475313951488044,
-      "grad_norm": 0.00597192021086812,
-      "learning_rate": 0.00019999892702826926,
-      "loss": 46.0,
-      "step": 19296
-    },
-    {
-      "epoch": 1.4753904084714338,
-      "grad_norm": 0.0008220398449338973,
-      "learning_rate": 0.00019999892691699707,
-      "loss": 46.0,
-      "step": 19297
-    },
-    {
-      "epoch": 1.4754668654548235,
-      "grad_norm": 0.002113528084009886,
-      "learning_rate": 0.00019999892680571914,
-      "loss": 46.0,
-      "step": 19298
-    },
-    {
-      "epoch": 1.475543322438213,
-      "grad_norm": 0.002578044543042779,
-      "learning_rate": 0.0001999989266944354,
-      "loss": 46.0,
-      "step": 19299
-    },
-    {
-      "epoch": 1.4756197794216028,
-      "grad_norm": 0.0014027304714545608,
-      "learning_rate": 0.00019999892658314594,
-      "loss": 46.0,
-      "step": 19300
-    },
-    {
-      "epoch": 1.4756962364049926,
-      "grad_norm": 0.0008738245232962072,
-      "learning_rate": 0.00019999892647185069,
-      "loss": 46.0,
-      "step": 19301
-    },
-    {
-      "epoch": 1.4757726933883824,
-      "grad_norm": 0.001580814947374165,
-      "learning_rate": 0.00019999892636054963,
-      "loss": 46.0,
-      "step": 19302
-    },
-    {
-      "epoch": 1.4758491503717721,
-      "grad_norm": 0.0008109455229714513,
-      "learning_rate": 0.00019999892624924286,
-      "loss": 46.0,
-      "step": 19303
-    },
-    {
-      "epoch": 1.4759256073551619,
-      "grad_norm": 0.008592908270657063,
-      "learning_rate": 0.0001999989261379303,
-      "loss": 46.0,
-      "step": 19304
-    },
-    {
-      "epoch": 1.4760020643385516,
-      "grad_norm": 0.001697972766123712,
-      "learning_rate": 0.00019999892602661198,
-      "loss": 46.0,
-      "step": 19305
-    },
-    {
-      "epoch": 1.4760785213219412,
-      "grad_norm": 0.0009819112019613385,
-      "learning_rate": 0.00019999892591528786,
-      "loss": 46.0,
-      "step": 19306
-    },
-    {
-      "epoch": 1.476154978305331,
-      "grad_norm": 0.006307743024080992,
-      "learning_rate": 0.00019999892580395797,
-      "loss": 46.0,
-      "step": 19307
-    },
-    {
-      "epoch": 1.4762314352887207,
-      "grad_norm": 0.0012942017056047916,
-      "learning_rate": 0.00019999892569262233,
-      "loss": 46.0,
-      "step": 19308
-    },
-    {
-      "epoch": 1.4763078922721105,
-      "grad_norm": 0.0005027378792874515,
-      "learning_rate": 0.0001999989255812809,
-      "loss": 46.0,
-      "step": 19309
-    },
-    {
-      "epoch": 1.4763843492555,
-      "grad_norm": 0.0006243587704375386,
-      "learning_rate": 0.00019999892546993374,
-      "loss": 46.0,
-      "step": 19310
-    },
-    {
-      "epoch": 1.4764608062388898,
-      "grad_norm": 0.0005012330366298556,
-      "learning_rate": 0.00019999892535858078,
-      "loss": 46.0,
-      "step": 19311
-    },
-    {
-      "epoch": 1.4765372632222795,
-      "grad_norm": 0.004242443013936281,
-      "learning_rate": 0.00019999892524722208,
-      "loss": 46.0,
-      "step": 19312
-    },
-    {
-      "epoch": 1.4766137202056693,
-      "grad_norm": 0.002433671150356531,
-      "learning_rate": 0.00019999892513585757,
-      "loss": 46.0,
-      "step": 19313
-    },
-    {
-      "epoch": 1.476690177189059,
-      "grad_norm": 0.0004711611836683005,
-      "learning_rate": 0.00019999892502448732,
-      "loss": 46.0,
-      "step": 19314
-    },
-    {
-      "epoch": 1.4767666341724488,
-      "grad_norm": 0.000834117759950459,
-      "learning_rate": 0.0001999989249131113,
-      "loss": 46.0,
-      "step": 19315
-    },
-    {
-      "epoch": 1.4768430911558386,
-      "grad_norm": 0.00785927101969719,
-      "learning_rate": 0.0001999989248017295,
-      "loss": 46.0,
-      "step": 19316
-    },
-    {
-      "epoch": 1.476919548139228,
-      "grad_norm": 0.004790124483406544,
-      "learning_rate": 0.00019999892469034193,
-      "loss": 46.0,
-      "step": 19317
-    },
-    {
-      "epoch": 1.4769960051226179,
-      "grad_norm": 0.0037100291810929775,
-      "learning_rate": 0.0001999989245789486,
-      "loss": 46.0,
-      "step": 19318
-    },
-    {
-      "epoch": 1.4770724621060076,
-      "grad_norm": 0.007419370114803314,
-      "learning_rate": 0.00019999892446754947,
-      "loss": 46.0,
-      "step": 19319
-    },
-    {
-      "epoch": 1.4771489190893974,
-      "grad_norm": 0.004334686789661646,
-      "learning_rate": 0.0001999989243561446,
-      "loss": 46.0,
-      "step": 19320
-    },
-    {
-      "epoch": 1.477225376072787,
-      "grad_norm": 0.0007274855161085725,
-      "learning_rate": 0.00019999892424473395,
-      "loss": 46.0,
-      "step": 19321
-    },
-    {
-      "epoch": 1.4773018330561767,
-      "grad_norm": 0.0012469470966607332,
-      "learning_rate": 0.00019999892413331756,
-      "loss": 46.0,
-      "step": 19322
-    },
-    {
-      "epoch": 1.4773782900395664,
-      "grad_norm": 0.00805369671434164,
-      "learning_rate": 0.00019999892402189535,
-      "loss": 46.0,
-      "step": 19323
-    },
-    {
-      "epoch": 1.4774547470229562,
-      "grad_norm": 0.0037581881042569876,
-      "learning_rate": 0.00019999892391046742,
-      "loss": 46.0,
-      "step": 19324
-    },
-    {
-      "epoch": 1.477531204006346,
-      "grad_norm": 0.0024633165448904037,
-      "learning_rate": 0.0001999989237990337,
-      "loss": 46.0,
-      "step": 19325
-    },
-    {
-      "epoch": 1.4776076609897357,
-      "grad_norm": 0.002172968117520213,
-      "learning_rate": 0.0001999989236875942,
-      "loss": 46.0,
-      "step": 19326
-    },
-    {
-      "epoch": 1.4776841179731255,
-      "grad_norm": 0.0007751628290861845,
-      "learning_rate": 0.00019999892357614894,
-      "loss": 46.0,
-      "step": 19327
-    },
-    {
-      "epoch": 1.477760574956515,
-      "grad_norm": 0.0016616367502138019,
-      "learning_rate": 0.00019999892346469792,
-      "loss": 46.0,
-      "step": 19328
-    },
-    {
-      "epoch": 1.4778370319399048,
-      "grad_norm": 0.0017188034253194928,
-      "learning_rate": 0.0001999989233532411,
-      "loss": 46.0,
-      "step": 19329
-    },
-    {
-      "epoch": 1.4779134889232945,
-      "grad_norm": 0.0013235813239589334,
-      "learning_rate": 0.00019999892324177852,
-      "loss": 46.0,
-      "step": 19330
-    },
-    {
-      "epoch": 1.4779899459066843,
-      "grad_norm": 0.0013228385942056775,
-      "learning_rate": 0.00019999892313031018,
-      "loss": 46.0,
-      "step": 19331
-    },
-    {
-      "epoch": 1.4780664028900738,
-      "grad_norm": 0.0006751374457962811,
-      "learning_rate": 0.00019999892301883606,
-      "loss": 46.0,
-      "step": 19332
-    },
-    {
-      "epoch": 1.4781428598734636,
-      "grad_norm": 0.006896035745739937,
-      "learning_rate": 0.0001999989229073562,
-      "loss": 46.0,
-      "step": 19333
-    },
-    {
-      "epoch": 1.4782193168568534,
-      "grad_norm": 0.002090553054586053,
-      "learning_rate": 0.00019999892279587057,
-      "loss": 46.0,
-      "step": 19334
-    },
-    {
-      "epoch": 1.4782957738402431,
-      "grad_norm": 0.003226600820198655,
-      "learning_rate": 0.00019999892268437913,
-      "loss": 46.0,
-      "step": 19335
-    },
-    {
-      "epoch": 1.478372230823633,
-      "grad_norm": 0.001297247363254428,
-      "learning_rate": 0.00019999892257288195,
-      "loss": 46.0,
-      "step": 19336
-    },
-    {
-      "epoch": 1.4784486878070227,
-      "grad_norm": 0.0007756868144497275,
-      "learning_rate": 0.00019999892246137902,
-      "loss": 46.0,
-      "step": 19337
-    },
-    {
-      "epoch": 1.4785251447904124,
-      "grad_norm": 0.0009608445107005537,
-      "learning_rate": 0.0001999989223498703,
-      "loss": 46.0,
-      "step": 19338
-    },
-    {
-      "epoch": 1.478601601773802,
-      "grad_norm": 0.0019543610978871584,
-      "learning_rate": 0.00019999892223835576,
-      "loss": 46.0,
-      "step": 19339
-    },
-    {
-      "epoch": 1.4786780587571917,
-      "grad_norm": 0.0020117731764912605,
-      "learning_rate": 0.0001999989221268355,
-      "loss": 46.0,
-      "step": 19340
-    },
-    {
-      "epoch": 1.4787545157405815,
-      "grad_norm": 0.0012466547777876258,
-      "learning_rate": 0.00019999892201530948,
-      "loss": 46.0,
-      "step": 19341
-    },
-    {
-      "epoch": 1.4788309727239712,
-      "grad_norm": 0.000765244709327817,
-      "learning_rate": 0.00019999892190377766,
-      "loss": 46.0,
-      "step": 19342
-    },
-    {
-      "epoch": 1.4789074297073608,
-      "grad_norm": 0.0014841114170849323,
-      "learning_rate": 0.00019999892179224012,
-      "loss": 46.0,
-      "step": 19343
-    },
-    {
-      "epoch": 1.4789838866907505,
-      "grad_norm": 0.0016742039006203413,
-      "learning_rate": 0.00019999892168069678,
-      "loss": 46.0,
-      "step": 19344
-    },
-    {
-      "epoch": 1.4790603436741403,
-      "grad_norm": 0.006475646514445543,
-      "learning_rate": 0.00019999892156914766,
-      "loss": 46.0,
-      "step": 19345
-    },
-    {
-      "epoch": 1.47913680065753,
-      "grad_norm": 0.001939265406690538,
-      "learning_rate": 0.00019999892145759277,
-      "loss": 46.0,
-      "step": 19346
-    },
-    {
-      "epoch": 1.4792132576409198,
-      "grad_norm": 0.0008946038433350623,
-      "learning_rate": 0.00019999892134603213,
-      "loss": 46.0,
-      "step": 19347
-    },
-    {
-      "epoch": 1.4792897146243096,
-      "grad_norm": 0.0015546358190476894,
-      "learning_rate": 0.0001999989212344657,
-      "loss": 46.0,
-      "step": 19348
-    },
-    {
-      "epoch": 1.4793661716076993,
-      "grad_norm": 0.002928095869719982,
-      "learning_rate": 0.00019999892112289351,
-      "loss": 46.0,
-      "step": 19349
-    },
-    {
-      "epoch": 1.4794426285910889,
-      "grad_norm": 0.0011154068633913994,
-      "learning_rate": 0.00019999892101131556,
-      "loss": 46.0,
-      "step": 19350
-    },
-    {
-      "epoch": 1.4795190855744786,
-      "grad_norm": 0.0015208248514682055,
-      "learning_rate": 0.00019999892089973186,
-      "loss": 46.0,
-      "step": 19351
-    },
-    {
-      "epoch": 1.4795955425578684,
-      "grad_norm": 0.0008519775001332164,
-      "learning_rate": 0.00019999892078814235,
-      "loss": 46.0,
-      "step": 19352
-    },
-    {
-      "epoch": 1.4796719995412582,
-      "grad_norm": 0.0022986182011663914,
-      "learning_rate": 0.00019999892067654708,
-      "loss": 46.0,
-      "step": 19353
-    },
-    {
-      "epoch": 1.4797484565246477,
-      "grad_norm": 0.003029866376891732,
-      "learning_rate": 0.00019999892056494605,
-      "loss": 46.0,
-      "step": 19354
-    },
-    {
-      "epoch": 1.4798249135080375,
-      "grad_norm": 0.0005599893629550934,
-      "learning_rate": 0.00019999892045333923,
-      "loss": 46.0,
-      "step": 19355
-    },
-    {
-      "epoch": 1.4799013704914272,
-      "grad_norm": 0.006208906881511211,
-      "learning_rate": 0.00019999892034172664,
-      "loss": 46.0,
-      "step": 19356
-    },
-    {
-      "epoch": 1.479977827474817,
-      "grad_norm": 0.0008485438884235919,
-      "learning_rate": 0.00019999892023010832,
-      "loss": 46.0,
-      "step": 19357
-    },
-    {
-      "epoch": 1.4800542844582067,
-      "grad_norm": 0.0005953279323875904,
-      "learning_rate": 0.0001999989201184842,
-      "loss": 46.0,
-      "step": 19358
-    },
-    {
-      "epoch": 1.4801307414415965,
-      "grad_norm": 0.001970952143892646,
-      "learning_rate": 0.00019999892000685431,
-      "loss": 46.0,
-      "step": 19359
-    },
-    {
-      "epoch": 1.4802071984249863,
-      "grad_norm": 0.0007726777112111449,
-      "learning_rate": 0.00019999891989521865,
-      "loss": 46.0,
-      "step": 19360
-    },
-    {
-      "epoch": 1.4802836554083758,
-      "grad_norm": 0.0034486232325434685,
-      "learning_rate": 0.00019999891978357724,
-      "loss": 46.0,
-      "step": 19361
-    },
-    {
-      "epoch": 1.4803601123917656,
-      "grad_norm": 0.0002802393864840269,
-      "learning_rate": 0.0001999989196719301,
-      "loss": 46.0,
-      "step": 19362
-    },
-    {
-      "epoch": 1.4804365693751553,
-      "grad_norm": 0.0007574927294626832,
-      "learning_rate": 0.0001999989195602771,
-      "loss": 46.0,
-      "step": 19363
-    },
-    {
-      "epoch": 1.480513026358545,
-      "grad_norm": 0.0012183692306280136,
-      "learning_rate": 0.00019999891944861838,
-      "loss": 46.0,
-      "step": 19364
-    },
-    {
-      "epoch": 1.4805894833419346,
-      "grad_norm": 0.002184170065447688,
-      "learning_rate": 0.00019999891933695387,
-      "loss": 46.0,
-      "step": 19365
-    },
-    {
-      "epoch": 1.4806659403253244,
-      "grad_norm": 0.0009818640537559986,
-      "learning_rate": 0.00019999891922528362,
-      "loss": 46.0,
-      "step": 19366
-    },
-    {
-      "epoch": 1.4807423973087142,
-      "grad_norm": 0.0053679752163589,
-      "learning_rate": 0.00019999891911360757,
-      "loss": 46.0,
-      "step": 19367
-    },
-    {
-      "epoch": 1.480818854292104,
-      "grad_norm": 0.0030079404823482037,
-      "learning_rate": 0.00019999891900192575,
-      "loss": 46.0,
-      "step": 19368
-    },
-    {
-      "epoch": 1.4808953112754937,
-      "grad_norm": 0.0009387653553858399,
-      "learning_rate": 0.00019999891889023818,
-      "loss": 46.0,
-      "step": 19369
-    },
-    {
-      "epoch": 1.4809717682588834,
-      "grad_norm": 0.0024747024290263653,
-      "learning_rate": 0.00019999891877854484,
-      "loss": 46.0,
-      "step": 19370
-    },
-    {
-      "epoch": 1.4810482252422732,
-      "grad_norm": 0.0010155820054933429,
-      "learning_rate": 0.00019999891866684573,
-      "loss": 46.0,
-      "step": 19371
-    },
-    {
-      "epoch": 1.4811246822256627,
-      "grad_norm": 0.002400304889306426,
-      "learning_rate": 0.00019999891855514084,
-      "loss": 46.0,
-      "step": 19372
-    },
-    {
-      "epoch": 1.4812011392090525,
-      "grad_norm": 0.012223457917571068,
-      "learning_rate": 0.0001999989184434302,
-      "loss": 46.0,
-      "step": 19373
-    },
-    {
-      "epoch": 1.4812775961924423,
-      "grad_norm": 0.001629875274375081,
-      "learning_rate": 0.0001999989183317138,
-      "loss": 46.0,
-      "step": 19374
-    },
-    {
-      "epoch": 1.481354053175832,
-      "grad_norm": 0.0008832475286908448,
-      "learning_rate": 0.00019999891821999158,
-      "loss": 46.0,
-      "step": 19375
-    },
-    {
-      "epoch": 1.4814305101592216,
-      "grad_norm": 0.009094862267374992,
-      "learning_rate": 0.0001999989181082636,
-      "loss": 46.0,
-      "step": 19376
-    },
-    {
-      "epoch": 1.4815069671426113,
-      "grad_norm": 0.0006835999083705246,
-      "learning_rate": 0.00019999891799652987,
-      "loss": 46.0,
-      "step": 19377
-    },
-    {
-      "epoch": 1.481583424126001,
-      "grad_norm": 0.0019940081983804703,
-      "learning_rate": 0.0001999989178847904,
-      "loss": 46.0,
-      "step": 19378
-    },
-    {
-      "epoch": 1.4816598811093908,
-      "grad_norm": 0.0066498215310275555,
-      "learning_rate": 0.00019999891777304512,
-      "loss": 46.0,
-      "step": 19379
-    },
-    {
-      "epoch": 1.4817363380927806,
-      "grad_norm": 0.0008446233114227653,
-      "learning_rate": 0.00019999891766129407,
-      "loss": 46.0,
-      "step": 19380
-    },
-    {
-      "epoch": 1.4818127950761704,
-      "grad_norm": 0.0005438279476948082,
-      "learning_rate": 0.00019999891754953725,
-      "loss": 46.0,
-      "step": 19381
-    },
-    {
-      "epoch": 1.4818892520595601,
-      "grad_norm": 0.0013539448846131563,
-      "learning_rate": 0.00019999891743777468,
-      "loss": 46.0,
-      "step": 19382
-    },
-    {
-      "epoch": 1.4819657090429497,
-      "grad_norm": 0.0013285381719470024,
-      "learning_rate": 0.00019999891732600634,
-      "loss": 46.0,
-      "step": 19383
-    },
-    {
-      "epoch": 1.4820421660263394,
-      "grad_norm": 0.0017920794198289514,
-      "learning_rate": 0.0001999989172142322,
-      "loss": 46.0,
-      "step": 19384
-    },
-    {
-      "epoch": 1.4821186230097292,
-      "grad_norm": 0.0007969422149471939,
-      "learning_rate": 0.00019999891710245234,
-      "loss": 46.0,
-      "step": 19385
-    },
-    {
-      "epoch": 1.482195079993119,
-      "grad_norm": 0.0008265781216323376,
-      "learning_rate": 0.00019999891699066667,
-      "loss": 46.0,
-      "step": 19386
-    },
-    {
-      "epoch": 1.4822715369765085,
-      "grad_norm": 0.0007214270299300551,
-      "learning_rate": 0.00019999891687887526,
-      "loss": 46.0,
-      "step": 19387
-    },
-    {
-      "epoch": 1.4823479939598982,
-      "grad_norm": 0.0065918490290641785,
-      "learning_rate": 0.00019999891676707808,
-      "loss": 46.0,
-      "step": 19388
-    },
-    {
-      "epoch": 1.482424450943288,
-      "grad_norm": 0.0022306328173726797,
-      "learning_rate": 0.0001999989166552751,
-      "loss": 46.0,
-      "step": 19389
-    },
-    {
-      "epoch": 1.4825009079266778,
-      "grad_norm": 0.0005873983609490097,
-      "learning_rate": 0.00019999891654346637,
-      "loss": 46.0,
-      "step": 19390
-    },
-    {
-      "epoch": 1.4825773649100675,
-      "grad_norm": 0.0021284841932356358,
-      "learning_rate": 0.00019999891643165184,
-      "loss": 46.0,
-      "step": 19391
-    },
-    {
-      "epoch": 1.4826538218934573,
-      "grad_norm": 0.004474198445677757,
-      "learning_rate": 0.0001999989163198316,
-      "loss": 46.0,
-      "step": 19392
-    },
-    {
-      "epoch": 1.482730278876847,
-      "grad_norm": 0.0007538386271335185,
-      "learning_rate": 0.00019999891620800555,
-      "loss": 46.0,
-      "step": 19393
-    },
-    {
-      "epoch": 1.4828067358602366,
-      "grad_norm": 0.0006229786667972803,
-      "learning_rate": 0.00019999891609617375,
-      "loss": 46.0,
-      "step": 19394
-    },
-    {
-      "epoch": 1.4828831928436264,
-      "grad_norm": 0.0014410275034606457,
-      "learning_rate": 0.00019999891598433616,
-      "loss": 46.0,
-      "step": 19395
-    },
-    {
-      "epoch": 1.4829596498270161,
-      "grad_norm": 0.002674880437552929,
-      "learning_rate": 0.00019999891587249281,
-      "loss": 46.0,
-      "step": 19396
-    },
-    {
-      "epoch": 1.4830361068104059,
-      "grad_norm": 0.0015902069862931967,
-      "learning_rate": 0.0001999989157606437,
-      "loss": 46.0,
-      "step": 19397
-    },
-    {
-      "epoch": 1.4831125637937954,
-      "grad_norm": 0.0031280554831027985,
-      "learning_rate": 0.00019999891564878878,
-      "loss": 46.0,
-      "step": 19398
-    },
-    {
-      "epoch": 1.4831890207771852,
-      "grad_norm": 0.0032418814953416586,
-      "learning_rate": 0.00019999891553692812,
-      "loss": 46.0,
-      "step": 19399
-    },
-    {
-      "epoch": 1.483265477760575,
-      "grad_norm": 0.0013350851368159056,
-      "learning_rate": 0.00019999891542506171,
-      "loss": 46.0,
-      "step": 19400
-    },
-    {
-      "epoch": 1.4833419347439647,
-      "grad_norm": 0.0021014041267335415,
-      "learning_rate": 0.0001999989153131895,
-      "loss": 46.0,
-      "step": 19401
-    },
-    {
-      "epoch": 1.4834183917273545,
-      "grad_norm": 0.0011488395975902677,
-      "learning_rate": 0.00019999891520131155,
-      "loss": 46.0,
-      "step": 19402
-    },
-    {
-      "epoch": 1.4834948487107442,
-      "grad_norm": 0.0008808790007606149,
-      "learning_rate": 0.00019999891508942782,
-      "loss": 46.0,
-      "step": 19403
-    },
-    {
-      "epoch": 1.4835713056941338,
-      "grad_norm": 0.0009377410751767457,
-      "learning_rate": 0.00019999891497753832,
-      "loss": 46.0,
-      "step": 19404
-    },
-    {
-      "epoch": 1.4836477626775235,
-      "grad_norm": 0.0009179608314298093,
-      "learning_rate": 0.00019999891486564305,
-      "loss": 46.0,
-      "step": 19405
-    },
-    {
-      "epoch": 1.4837242196609133,
-      "grad_norm": 0.002284203888848424,
-      "learning_rate": 0.000199998914753742,
-      "loss": 46.0,
-      "step": 19406
-    },
-    {
-      "epoch": 1.483800676644303,
-      "grad_norm": 0.0010577746434137225,
-      "learning_rate": 0.00019999891464183518,
-      "loss": 46.0,
-      "step": 19407
-    },
-    {
-      "epoch": 1.4838771336276928,
-      "grad_norm": 0.0014719637110829353,
-      "learning_rate": 0.00019999891452992258,
-      "loss": 46.0,
-      "step": 19408
-    },
-    {
-      "epoch": 1.4839535906110823,
-      "grad_norm": 0.0009698682115413249,
-      "learning_rate": 0.00019999891441800424,
-      "loss": 46.0,
-      "step": 19409
-    },
-    {
-      "epoch": 1.484030047594472,
-      "grad_norm": 0.0007479399209842086,
-      "learning_rate": 0.00019999891430608012,
-      "loss": 46.0,
-      "step": 19410
-    },
-    {
-      "epoch": 1.4841065045778619,
-      "grad_norm": 0.0005773433367721736,
-      "learning_rate": 0.00019999891419415024,
-      "loss": 46.0,
-      "step": 19411
-    },
-    {
-      "epoch": 1.4841829615612516,
-      "grad_norm": 0.0007472485885955393,
-      "learning_rate": 0.00019999891408221458,
-      "loss": 46.0,
-      "step": 19412
-    },
-    {
-      "epoch": 1.4842594185446414,
-      "grad_norm": 0.004900270141661167,
-      "learning_rate": 0.00019999891397027314,
-      "loss": 46.0,
-      "step": 19413
-    },
-    {
-      "epoch": 1.4843358755280311,
-      "grad_norm": 0.004642863292247057,
-      "learning_rate": 0.00019999891385832593,
-      "loss": 46.0,
-      "step": 19414
-    },
-    {
-      "epoch": 1.4844123325114207,
-      "grad_norm": 0.005004368722438812,
-      "learning_rate": 0.00019999891374637298,
-      "loss": 46.0,
-      "step": 19415
-    },
-    {
-      "epoch": 1.4844887894948104,
-      "grad_norm": 0.003215061966329813,
-      "learning_rate": 0.00019999891363441425,
-      "loss": 46.0,
-      "step": 19416
-    },
-    {
-      "epoch": 1.4845652464782002,
-      "grad_norm": 0.011967454105615616,
-      "learning_rate": 0.00019999891352244972,
-      "loss": 46.0,
-      "step": 19417
-    },
-    {
-      "epoch": 1.48464170346159,
-      "grad_norm": 0.0008361469954252243,
-      "learning_rate": 0.00019999891341047947,
-      "loss": 46.0,
-      "step": 19418
-    },
-    {
-      "epoch": 1.4847181604449795,
-      "grad_norm": 0.001079672365449369,
-      "learning_rate": 0.0001999989132985034,
-      "loss": 46.0,
-      "step": 19419
-    },
-    {
-      "epoch": 1.4847946174283693,
-      "grad_norm": 0.005496449302881956,
-      "learning_rate": 0.0001999989131865216,
-      "loss": 46.0,
-      "step": 19420
-    },
-    {
-      "epoch": 1.484871074411759,
-      "grad_norm": 0.005404963623732328,
-      "learning_rate": 0.000199998913074534,
-      "loss": 46.0,
-      "step": 19421
-    },
-    {
-      "epoch": 1.4849475313951488,
-      "grad_norm": 0.0014856626512482762,
-      "learning_rate": 0.00019999891296254067,
-      "loss": 46.0,
-      "step": 19422
-    },
-    {
-      "epoch": 1.4850239883785386,
-      "grad_norm": 0.004496196284890175,
-      "learning_rate": 0.00019999891285054155,
-      "loss": 46.0,
-      "step": 19423
-    },
-    {
-      "epoch": 1.4851004453619283,
-      "grad_norm": 0.001068907673470676,
-      "learning_rate": 0.00019999891273853667,
-      "loss": 46.0,
-      "step": 19424
-    },
-    {
-      "epoch": 1.485176902345318,
-      "grad_norm": 0.006409113295376301,
-      "learning_rate": 0.00019999891262652598,
-      "loss": 46.0,
-      "step": 19425
-    },
-    {
-      "epoch": 1.4852533593287076,
-      "grad_norm": 0.0012095402926206589,
-      "learning_rate": 0.00019999891251450954,
-      "loss": 46.0,
-      "step": 19426
-    },
-    {
-      "epoch": 1.4853298163120974,
-      "grad_norm": 0.001035945606417954,
-      "learning_rate": 0.00019999891240248734,
-      "loss": 46.0,
-      "step": 19427
-    },
-    {
-      "epoch": 1.4854062732954871,
-      "grad_norm": 0.0006595029844902456,
-      "learning_rate": 0.00019999891229045938,
-      "loss": 46.0,
-      "step": 19428
-    },
-    {
-      "epoch": 1.485482730278877,
-      "grad_norm": 0.0012759588425979018,
-      "learning_rate": 0.00019999891217842563,
-      "loss": 46.0,
-      "step": 19429
-    },
-    {
-      "epoch": 1.4855591872622664,
-      "grad_norm": 0.001290992135182023,
-      "learning_rate": 0.00019999891206638615,
-      "loss": 46.0,
-      "step": 19430
-    },
-    {
-      "epoch": 1.4856356442456562,
-      "grad_norm": 0.0007580990786664188,
-      "learning_rate": 0.00019999891195434085,
-      "loss": 46.0,
-      "step": 19431
-    },
-    {
-      "epoch": 1.485712101229046,
-      "grad_norm": 0.0008399473736062646,
-      "learning_rate": 0.0001999989118422898,
-      "loss": 46.0,
-      "step": 19432
-    },
-    {
-      "epoch": 1.4857885582124357,
-      "grad_norm": 0.00062708550831303,
-      "learning_rate": 0.00019999891173023298,
-      "loss": 46.0,
-      "step": 19433
-    },
-    {
-      "epoch": 1.4858650151958255,
-      "grad_norm": 0.0054834159091115,
-      "learning_rate": 0.00019999891161817042,
-      "loss": 46.0,
-      "step": 19434
-    },
-    {
-      "epoch": 1.4859414721792152,
-      "grad_norm": 0.002074608113616705,
-      "learning_rate": 0.00019999891150610205,
-      "loss": 46.0,
-      "step": 19435
-    },
-    {
-      "epoch": 1.486017929162605,
-      "grad_norm": 0.0014773256843909621,
-      "learning_rate": 0.00019999891139402793,
-      "loss": 46.0,
-      "step": 19436
-    },
-    {
-      "epoch": 1.4860943861459945,
-      "grad_norm": 0.0006069326191209257,
-      "learning_rate": 0.00019999891128194802,
-      "loss": 46.0,
-      "step": 19437
-    },
-    {
-      "epoch": 1.4861708431293843,
-      "grad_norm": 0.001013072906062007,
-      "learning_rate": 0.00019999891116986236,
-      "loss": 46.0,
-      "step": 19438
-    },
-    {
-      "epoch": 1.486247300112774,
-      "grad_norm": 0.0008229255327023566,
-      "learning_rate": 0.00019999891105777092,
-      "loss": 46.0,
-      "step": 19439
-    },
-    {
-      "epoch": 1.4863237570961638,
-      "grad_norm": 0.0009191535646095872,
-      "learning_rate": 0.00019999891094567374,
-      "loss": 46.0,
-      "step": 19440
-    },
-    {
-      "epoch": 1.4864002140795534,
-      "grad_norm": 0.0007151351310312748,
-      "learning_rate": 0.00019999891083357074,
-      "loss": 46.0,
-      "step": 19441
-    },
-    {
-      "epoch": 1.4864766710629431,
-      "grad_norm": 0.0014817918417975307,
-      "learning_rate": 0.000199998910721462,
-      "loss": 46.0,
-      "step": 19442
-    },
-    {
-      "epoch": 1.4865531280463329,
-      "grad_norm": 0.0009195443708449602,
-      "learning_rate": 0.0001999989106093475,
-      "loss": 46.0,
-      "step": 19443
-    },
-    {
-      "epoch": 1.4866295850297226,
-      "grad_norm": 0.0005606548511423171,
-      "learning_rate": 0.0001999989104972272,
-      "loss": 46.0,
-      "step": 19444
-    },
-    {
-      "epoch": 1.4867060420131124,
-      "grad_norm": 0.0019231239566579461,
-      "learning_rate": 0.00019999891038510116,
-      "loss": 46.0,
-      "step": 19445
-    },
-    {
-      "epoch": 1.4867824989965022,
-      "grad_norm": 0.003348397556692362,
-      "learning_rate": 0.00019999891027296937,
-      "loss": 46.0,
-      "step": 19446
-    },
-    {
-      "epoch": 1.486858955979892,
-      "grad_norm": 0.0009724876726977527,
-      "learning_rate": 0.00019999891016083177,
-      "loss": 46.0,
-      "step": 19447
-    },
-    {
-      "epoch": 1.4869354129632815,
-      "grad_norm": 0.0013627978041768074,
-      "learning_rate": 0.0001999989100486884,
-      "loss": 46.0,
-      "step": 19448
-    },
-    {
-      "epoch": 1.4870118699466712,
-      "grad_norm": 0.002209307625889778,
-      "learning_rate": 0.00019999890993653926,
-      "loss": 46.0,
-      "step": 19449
-    },
-    {
-      "epoch": 1.487088326930061,
-      "grad_norm": 0.0014282738557085395,
-      "learning_rate": 0.00019999890982438438,
-      "loss": 46.0,
-      "step": 19450
-    },
-    {
-      "epoch": 1.4871647839134507,
-      "grad_norm": 0.010506403632462025,
-      "learning_rate": 0.00019999890971222372,
-      "loss": 46.0,
-      "step": 19451
-    },
-    {
-      "epoch": 1.4872412408968403,
-      "grad_norm": 0.0008174251415766776,
-      "learning_rate": 0.00019999890960005728,
-      "loss": 46.0,
-      "step": 19452
-    },
-    {
-      "epoch": 1.48731769788023,
-      "grad_norm": 0.0007639850373379886,
-      "learning_rate": 0.0001999989094878851,
-      "loss": 46.0,
-      "step": 19453
-    },
-    {
-      "epoch": 1.4873941548636198,
-      "grad_norm": 0.0011500392574816942,
-      "learning_rate": 0.0001999989093757071,
-      "loss": 46.0,
-      "step": 19454
-    },
-    {
-      "epoch": 1.4874706118470096,
-      "grad_norm": 0.0007135348860174417,
-      "learning_rate": 0.00019999890926352337,
-      "loss": 46.0,
-      "step": 19455
-    },
-    {
-      "epoch": 1.4875470688303993,
-      "grad_norm": 0.0017271365504711866,
-      "learning_rate": 0.00019999890915133387,
-      "loss": 46.0,
-      "step": 19456
-    },
-    {
-      "epoch": 1.487623525813789,
-      "grad_norm": 0.0016721192514523864,
-      "learning_rate": 0.00019999890903913857,
-      "loss": 46.0,
-      "step": 19457
-    },
-    {
-      "epoch": 1.4876999827971789,
-      "grad_norm": 0.0010404117638245225,
-      "learning_rate": 0.00019999890892693752,
-      "loss": 46.0,
-      "step": 19458
-    },
-    {
-      "epoch": 1.4877764397805684,
-      "grad_norm": 0.003896439913660288,
-      "learning_rate": 0.0001999989088147307,
-      "loss": 46.0,
-      "step": 19459
-    },
-    {
-      "epoch": 1.4878528967639582,
-      "grad_norm": 0.006754432339221239,
-      "learning_rate": 0.00019999890870251813,
-      "loss": 46.0,
-      "step": 19460
-    },
-    {
-      "epoch": 1.487929353747348,
-      "grad_norm": 0.0007660723058506846,
-      "learning_rate": 0.00019999890859029974,
-      "loss": 46.0,
-      "step": 19461
-    },
-    {
-      "epoch": 1.4880058107307377,
-      "grad_norm": 0.0006175947492010891,
-      "learning_rate": 0.0001999989084780756,
-      "loss": 46.0,
-      "step": 19462
-    },
-    {
-      "epoch": 1.4880822677141272,
-      "grad_norm": 0.001723306137137115,
-      "learning_rate": 0.0001999989083658457,
-      "loss": 46.0,
-      "step": 19463
-    },
-    {
-      "epoch": 1.488158724697517,
-      "grad_norm": 0.0030678361654281616,
-      "learning_rate": 0.00019999890825361005,
-      "loss": 46.0,
-      "step": 19464
-    },
-    {
-      "epoch": 1.4882351816809067,
-      "grad_norm": 0.0008425671840086579,
-      "learning_rate": 0.00019999890814136862,
-      "loss": 46.0,
-      "step": 19465
-    },
-    {
-      "epoch": 1.4883116386642965,
-      "grad_norm": 0.00228483765386045,
-      "learning_rate": 0.0001999989080291214,
-      "loss": 46.0,
-      "step": 19466
-    },
-    {
-      "epoch": 1.4883880956476863,
-      "grad_norm": 0.0017840571235865355,
-      "learning_rate": 0.00019999890791686843,
-      "loss": 46.0,
-      "step": 19467
-    },
-    {
-      "epoch": 1.488464552631076,
-      "grad_norm": 0.0010684585431590676,
-      "learning_rate": 0.00019999890780460967,
-      "loss": 46.0,
-      "step": 19468
-    },
-    {
-      "epoch": 1.4885410096144658,
-      "grad_norm": 0.005592750385403633,
-      "learning_rate": 0.00019999890769234515,
-      "loss": 46.0,
-      "step": 19469
-    },
-    {
-      "epoch": 1.4886174665978553,
-      "grad_norm": 0.0050817388109862804,
-      "learning_rate": 0.00019999890758007487,
-      "loss": 46.0,
-      "step": 19470
-    },
-    {
-      "epoch": 1.488693923581245,
-      "grad_norm": 0.001060408540070057,
-      "learning_rate": 0.0001999989074677988,
-      "loss": 46.0,
-      "step": 19471
-    },
-    {
-      "epoch": 1.4887703805646348,
-      "grad_norm": 0.004463254474103451,
-      "learning_rate": 0.000199998907355517,
-      "loss": 46.0,
-      "step": 19472
-    },
-    {
-      "epoch": 1.4888468375480246,
-      "grad_norm": 0.0006602173089049757,
-      "learning_rate": 0.0001999989072432294,
-      "loss": 46.0,
-      "step": 19473
-    },
-    {
-      "epoch": 1.4889232945314141,
-      "grad_norm": 0.000831605342682451,
-      "learning_rate": 0.00019999890713093605,
-      "loss": 46.0,
-      "step": 19474
-    },
-    {
-      "epoch": 1.488999751514804,
-      "grad_norm": 0.0028210084419697523,
-      "learning_rate": 0.0001999989070186369,
-      "loss": 46.0,
-      "step": 19475
-    },
-    {
-      "epoch": 1.4890762084981937,
-      "grad_norm": 0.002045969944447279,
-      "learning_rate": 0.00019999890690633202,
-      "loss": 46.0,
-      "step": 19476
-    },
-    {
-      "epoch": 1.4891526654815834,
-      "grad_norm": 0.000600756611675024,
-      "learning_rate": 0.00019999890679402133,
-      "loss": 46.0,
-      "step": 19477
-    },
-    {
-      "epoch": 1.4892291224649732,
-      "grad_norm": 0.0020963200367987156,
-      "learning_rate": 0.0001999989066817049,
-      "loss": 46.0,
-      "step": 19478
-    },
-    {
-      "epoch": 1.489305579448363,
-      "grad_norm": 0.0006956481374800205,
-      "learning_rate": 0.0001999989065693827,
-      "loss": 46.0,
-      "step": 19479
-    },
-    {
-      "epoch": 1.4893820364317527,
-      "grad_norm": 0.002297037048265338,
-      "learning_rate": 0.0001999989064570547,
-      "loss": 46.0,
-      "step": 19480
-    },
-    {
-      "epoch": 1.4894584934151422,
-      "grad_norm": 0.001409986987709999,
-      "learning_rate": 0.00019999890634472096,
-      "loss": 46.0,
-      "step": 19481
-    },
-    {
-      "epoch": 1.489534950398532,
-      "grad_norm": 0.003171502845361829,
-      "learning_rate": 0.00019999890623238143,
-      "loss": 46.0,
-      "step": 19482
-    },
-    {
-      "epoch": 1.4896114073819218,
-      "grad_norm": 0.00182953046169132,
-      "learning_rate": 0.00019999890612003616,
-      "loss": 46.0,
-      "step": 19483
-    },
-    {
-      "epoch": 1.4896878643653115,
-      "grad_norm": 0.002752818865701556,
-      "learning_rate": 0.00019999890600768509,
-      "loss": 46.0,
-      "step": 19484
-    },
-    {
-      "epoch": 1.489764321348701,
-      "grad_norm": 0.0014071421464905143,
-      "learning_rate": 0.0001999989058953283,
-      "loss": 46.0,
-      "step": 19485
-    },
-    {
-      "epoch": 1.4898407783320908,
-      "grad_norm": 0.004534035921096802,
-      "learning_rate": 0.0001999989057829657,
-      "loss": 46.0,
-      "step": 19486
-    },
-    {
-      "epoch": 1.4899172353154806,
-      "grad_norm": 0.0033098235726356506,
-      "learning_rate": 0.0001999989056705973,
-      "loss": 46.0,
-      "step": 19487
-    },
-    {
-      "epoch": 1.4899936922988704,
-      "grad_norm": 0.0017915000207722187,
-      "learning_rate": 0.0001999989055582232,
-      "loss": 46.0,
-      "step": 19488
-    },
-    {
-      "epoch": 1.4900701492822601,
-      "grad_norm": 0.008325501345098019,
-      "learning_rate": 0.00019999890544584328,
-      "loss": 46.0,
-      "step": 19489
-    },
-    {
-      "epoch": 1.4901466062656499,
-      "grad_norm": 0.0020200293511152267,
-      "learning_rate": 0.00019999890533345762,
-      "loss": 46.0,
-      "step": 19490
-    },
-    {
-      "epoch": 1.4902230632490396,
-      "grad_norm": 0.0010951198637485504,
-      "learning_rate": 0.0001999989052210662,
-      "loss": 46.0,
-      "step": 19491
-    },
-    {
-      "epoch": 1.4902995202324292,
-      "grad_norm": 0.0022767281625419855,
-      "learning_rate": 0.00019999890510866895,
-      "loss": 46.0,
-      "step": 19492
-    },
-    {
-      "epoch": 1.490375977215819,
-      "grad_norm": 0.0007755453698337078,
-      "learning_rate": 0.00019999890499626597,
-      "loss": 46.0,
-      "step": 19493
-    },
-    {
-      "epoch": 1.4904524341992087,
-      "grad_norm": 0.0006250563892535865,
-      "learning_rate": 0.00019999890488385722,
-      "loss": 46.0,
-      "step": 19494
-    },
-    {
-      "epoch": 1.4905288911825985,
-      "grad_norm": 0.0030151105020195246,
-      "learning_rate": 0.00019999890477144272,
-      "loss": 46.0,
-      "step": 19495
-    },
-    {
-      "epoch": 1.490605348165988,
-      "grad_norm": 0.0007749577052891254,
-      "learning_rate": 0.00019999890465902242,
-      "loss": 46.0,
-      "step": 19496
-    },
-    {
-      "epoch": 1.4906818051493778,
-      "grad_norm": 0.0008271206170320511,
-      "learning_rate": 0.00019999890454659638,
-      "loss": 46.0,
-      "step": 19497
-    },
-    {
-      "epoch": 1.4907582621327675,
-      "grad_norm": 0.004292960744351149,
-      "learning_rate": 0.00019999890443416453,
-      "loss": 46.0,
-      "step": 19498
-    },
-    {
-      "epoch": 1.4908347191161573,
-      "grad_norm": 0.0010203635320067406,
-      "learning_rate": 0.00019999890432172694,
-      "loss": 46.0,
-      "step": 19499
-    },
-    {
-      "epoch": 1.490911176099547,
-      "grad_norm": 0.0018442235887050629,
-      "learning_rate": 0.00019999890420928357,
-      "loss": 46.0,
-      "step": 19500
-    },
-    {
-      "epoch": 1.4909876330829368,
-      "grad_norm": 0.0005117595428600907,
-      "learning_rate": 0.00019999890409683443,
-      "loss": 46.0,
-      "step": 19501
-    },
-    {
-      "epoch": 1.4910640900663266,
-      "grad_norm": 0.0009238338097929955,
-      "learning_rate": 0.00019999890398437952,
-      "loss": 46.0,
-      "step": 19502
-    },
-    {
-      "epoch": 1.491140547049716,
-      "grad_norm": 0.0006527479854412377,
-      "learning_rate": 0.00019999890387191886,
-      "loss": 46.0,
-      "step": 19503
-    },
-    {
-      "epoch": 1.4912170040331059,
-      "grad_norm": 0.001469060662202537,
-      "learning_rate": 0.00019999890375945243,
-      "loss": 46.0,
-      "step": 19504
-    },
-    {
-      "epoch": 1.4912934610164956,
-      "grad_norm": 0.000661347818095237,
-      "learning_rate": 0.00019999890364698022,
-      "loss": 46.0,
-      "step": 19505
-    },
-    {
-      "epoch": 1.4913699179998854,
-      "grad_norm": 0.0010039459448307753,
-      "learning_rate": 0.00019999890353450224,
-      "loss": 46.0,
-      "step": 19506
-    },
-    {
-      "epoch": 1.491446374983275,
-      "grad_norm": 0.0006407740875147283,
-      "learning_rate": 0.0001999989034220185,
-      "loss": 46.0,
-      "step": 19507
-    },
-    {
-      "epoch": 1.4915228319666647,
-      "grad_norm": 0.00034887384390458465,
-      "learning_rate": 0.00019999890330952896,
-      "loss": 46.0,
-      "step": 19508
-    },
-    {
-      "epoch": 1.4915992889500544,
-      "grad_norm": 0.0010198383824899793,
-      "learning_rate": 0.00019999890319703366,
-      "loss": 46.0,
-      "step": 19509
-    },
-    {
-      "epoch": 1.4916757459334442,
-      "grad_norm": 0.002623027889057994,
-      "learning_rate": 0.0001999989030845326,
-      "loss": 46.0,
-      "step": 19510
-    },
-    {
-      "epoch": 1.491752202916834,
-      "grad_norm": 0.0005774474120698869,
-      "learning_rate": 0.00019999890297202577,
-      "loss": 46.0,
-      "step": 19511
-    },
-    {
-      "epoch": 1.4918286599002237,
-      "grad_norm": 0.002018352970480919,
-      "learning_rate": 0.00019999890285951318,
-      "loss": 46.0,
-      "step": 19512
-    },
-    {
-      "epoch": 1.4919051168836135,
-      "grad_norm": 0.0003837211988866329,
-      "learning_rate": 0.0001999989027469948,
-      "loss": 46.0,
-      "step": 19513
-    },
-    {
-      "epoch": 1.491981573867003,
-      "grad_norm": 0.0021206915844231844,
-      "learning_rate": 0.00019999890263447067,
-      "loss": 46.0,
-      "step": 19514
-    },
-    {
-      "epoch": 1.4920580308503928,
-      "grad_norm": 0.0008061110856942832,
-      "learning_rate": 0.00019999890252194076,
-      "loss": 46.0,
-      "step": 19515
-    },
-    {
-      "epoch": 1.4921344878337826,
-      "grad_norm": 0.0017214424442499876,
-      "learning_rate": 0.00019999890240940507,
-      "loss": 46.0,
-      "step": 19516
-    },
-    {
-      "epoch": 1.4922109448171723,
-      "grad_norm": 0.0011911719338968396,
-      "learning_rate": 0.00019999890229686361,
-      "loss": 46.0,
-      "step": 19517
-    },
-    {
-      "epoch": 1.4922874018005619,
-      "grad_norm": 0.00245038908906281,
-      "learning_rate": 0.0001999989021843164,
-      "loss": 46.0,
-      "step": 19518
-    },
-    {
-      "epoch": 1.4923638587839516,
-      "grad_norm": 0.002539834240451455,
-      "learning_rate": 0.00019999890207176343,
-      "loss": 46.0,
-      "step": 19519
-    },
-    {
-      "epoch": 1.4924403157673414,
-      "grad_norm": 0.0006074015982449055,
-      "learning_rate": 0.00019999890195920468,
-      "loss": 46.0,
-      "step": 19520
-    },
-    {
-      "epoch": 1.4925167727507311,
-      "grad_norm": 0.0009795281803235412,
-      "learning_rate": 0.00019999890184664015,
-      "loss": 46.0,
-      "step": 19521
-    },
-    {
-      "epoch": 1.492593229734121,
-      "grad_norm": 0.0007095045875757933,
-      "learning_rate": 0.00019999890173406985,
-      "loss": 46.0,
-      "step": 19522
-    },
-    {
-      "epoch": 1.4926696867175107,
-      "grad_norm": 0.002733758883550763,
-      "learning_rate": 0.0001999989016214938,
-      "loss": 46.0,
-      "step": 19523
-    },
-    {
-      "epoch": 1.4927461437009004,
-      "grad_norm": 0.0013536748010665178,
-      "learning_rate": 0.00019999890150891196,
-      "loss": 46.0,
-      "step": 19524
-    },
-    {
-      "epoch": 1.49282260068429,
-      "grad_norm": 0.001242319936864078,
-      "learning_rate": 0.00019999890139632437,
-      "loss": 46.0,
-      "step": 19525
-    },
-    {
-      "epoch": 1.4928990576676797,
-      "grad_norm": 0.0012543570483103395,
-      "learning_rate": 0.000199998901283731,
-      "loss": 46.0,
-      "step": 19526
-    },
-    {
-      "epoch": 1.4929755146510695,
-      "grad_norm": 0.002490056911483407,
-      "learning_rate": 0.00019999890117113184,
-      "loss": 46.0,
-      "step": 19527
-    },
-    {
-      "epoch": 1.4930519716344592,
-      "grad_norm": 0.0014810101129114628,
-      "learning_rate": 0.00019999890105852695,
-      "loss": 46.0,
-      "step": 19528
-    },
-    {
-      "epoch": 1.4931284286178488,
-      "grad_norm": 0.0009366166195832193,
-      "learning_rate": 0.00019999890094591627,
-      "loss": 46.0,
-      "step": 19529
-    },
-    {
-      "epoch": 1.4932048856012385,
-      "grad_norm": 0.011300221085548401,
-      "learning_rate": 0.0001999989008332998,
-      "loss": 46.0,
-      "step": 19530
-    },
-    {
-      "epoch": 1.4932813425846283,
-      "grad_norm": 0.0009357829112559557,
-      "learning_rate": 0.0001999989007206776,
-      "loss": 46.0,
-      "step": 19531
-    },
-    {
-      "epoch": 1.493357799568018,
-      "grad_norm": 0.0012869485653936863,
-      "learning_rate": 0.00019999890060804962,
-      "loss": 46.0,
-      "step": 19532
-    },
-    {
-      "epoch": 1.4934342565514078,
-      "grad_norm": 0.00031263078562915325,
-      "learning_rate": 0.00019999890049541587,
-      "loss": 46.0,
-      "step": 19533
-    },
-    {
-      "epoch": 1.4935107135347976,
-      "grad_norm": 0.0006317394436337054,
-      "learning_rate": 0.00019999890038277634,
-      "loss": 46.0,
-      "step": 19534
-    },
-    {
-      "epoch": 1.4935871705181871,
-      "grad_norm": 0.0019006740767508745,
-      "learning_rate": 0.00019999890027013105,
-      "loss": 46.0,
-      "step": 19535
-    },
-    {
-      "epoch": 1.4936636275015769,
-      "grad_norm": 0.0007488125120289624,
-      "learning_rate": 0.00019999890015747997,
-      "loss": 46.0,
-      "step": 19536
-    },
-    {
-      "epoch": 1.4937400844849666,
-      "grad_norm": 0.003511018818244338,
-      "learning_rate": 0.00019999890004482315,
-      "loss": 46.0,
-      "step": 19537
-    },
-    {
-      "epoch": 1.4938165414683564,
-      "grad_norm": 0.0006438705022446811,
-      "learning_rate": 0.00019999889993216054,
-      "loss": 46.0,
-      "step": 19538
-    },
-    {
-      "epoch": 1.4938929984517462,
-      "grad_norm": 0.0010862246854230762,
-      "learning_rate": 0.00019999889981949217,
-      "loss": 46.0,
-      "step": 19539
-    },
-    {
-      "epoch": 1.4939694554351357,
-      "grad_norm": 0.0017551570199429989,
-      "learning_rate": 0.00019999889970681803,
-      "loss": 46.0,
-      "step": 19540
-    },
-    {
-      "epoch": 1.4940459124185255,
-      "grad_norm": 0.0004184254794381559,
-      "learning_rate": 0.00019999889959413812,
-      "loss": 46.0,
-      "step": 19541
-    },
-    {
-      "epoch": 1.4941223694019152,
-      "grad_norm": 0.0005411518504843116,
-      "learning_rate": 0.00019999889948145243,
-      "loss": 46.0,
-      "step": 19542
-    },
-    {
-      "epoch": 1.494198826385305,
-      "grad_norm": 0.0023433968890458345,
-      "learning_rate": 0.00019999889936876098,
-      "loss": 46.0,
-      "step": 19543
-    },
-    {
-      "epoch": 1.4942752833686948,
-      "grad_norm": 0.0049154614098370075,
-      "learning_rate": 0.00019999889925606377,
-      "loss": 46.0,
-      "step": 19544
-    },
-    {
-      "epoch": 1.4943517403520845,
-      "grad_norm": 0.00227510672993958,
-      "learning_rate": 0.0001999988991433608,
-      "loss": 46.0,
-      "step": 19545
-    },
-    {
-      "epoch": 1.494428197335474,
-      "grad_norm": 0.0005902040284126997,
-      "learning_rate": 0.000199998899030652,
-      "loss": 46.0,
-      "step": 19546
-    },
-    {
-      "epoch": 1.4945046543188638,
-      "grad_norm": 0.0036403939593583345,
-      "learning_rate": 0.00019999889891793751,
-      "loss": 46.0,
-      "step": 19547
-    },
-    {
-      "epoch": 1.4945811113022536,
-      "grad_norm": 0.0021516934502869844,
-      "learning_rate": 0.0001999988988052172,
-      "loss": 46.0,
-      "step": 19548
-    },
-    {
-      "epoch": 1.4946575682856433,
-      "grad_norm": 0.004841632209718227,
-      "learning_rate": 0.00019999889869249114,
-      "loss": 46.0,
-      "step": 19549
-    },
-    {
-      "epoch": 1.4947340252690329,
-      "grad_norm": 0.0008717633900232613,
-      "learning_rate": 0.0001999988985797593,
-      "loss": 46.0,
-      "step": 19550
-    },
-    {
-      "epoch": 1.4948104822524226,
-      "grad_norm": 0.0004530810983851552,
-      "learning_rate": 0.00019999889846702168,
-      "loss": 46.0,
-      "step": 19551
-    },
-    {
-      "epoch": 1.4948869392358124,
-      "grad_norm": 0.0013703095028176904,
-      "learning_rate": 0.00019999889835427831,
-      "loss": 46.0,
-      "step": 19552
-    },
-    {
-      "epoch": 1.4949633962192022,
-      "grad_norm": 0.0008979664999060333,
-      "learning_rate": 0.00019999889824152915,
-      "loss": 46.0,
-      "step": 19553
-    },
-    {
-      "epoch": 1.495039853202592,
-      "grad_norm": 0.007226672489196062,
-      "learning_rate": 0.00019999889812877426,
-      "loss": 46.0,
-      "step": 19554
-    },
-    {
-      "epoch": 1.4951163101859817,
-      "grad_norm": 0.00529554532840848,
-      "learning_rate": 0.00019999889801601358,
-      "loss": 46.0,
-      "step": 19555
-    },
-    {
-      "epoch": 1.4951927671693714,
-      "grad_norm": 0.0008369798306375742,
-      "learning_rate": 0.00019999889790324712,
-      "loss": 46.0,
-      "step": 19556
-    },
-    {
-      "epoch": 1.495269224152761,
-      "grad_norm": 0.0005549624329432845,
-      "learning_rate": 0.0001999988977904749,
-      "loss": 46.0,
-      "step": 19557
-    },
-    {
-      "epoch": 1.4953456811361507,
-      "grad_norm": 0.0020292503759264946,
-      "learning_rate": 0.0001999988976776969,
-      "loss": 46.0,
-      "step": 19558
-    },
-    {
-      "epoch": 1.4954221381195405,
-      "grad_norm": 0.002541651949286461,
-      "learning_rate": 0.00019999889756491313,
-      "loss": 46.0,
-      "step": 19559
-    },
-    {
-      "epoch": 1.4954985951029303,
-      "grad_norm": 0.002015515463426709,
-      "learning_rate": 0.00019999889745212363,
-      "loss": 46.0,
-      "step": 19560
-    },
-    {
-      "epoch": 1.4955750520863198,
-      "grad_norm": 0.005844128783792257,
-      "learning_rate": 0.0001999988973393283,
-      "loss": 46.0,
-      "step": 19561
-    },
-    {
-      "epoch": 1.4956515090697096,
-      "grad_norm": 0.0005108119221404195,
-      "learning_rate": 0.00019999889722652724,
-      "loss": 46.0,
-      "step": 19562
-    },
-    {
-      "epoch": 1.4957279660530993,
-      "grad_norm": 0.0018171415431424975,
-      "learning_rate": 0.00019999889711372042,
-      "loss": 46.0,
-      "step": 19563
-    },
-    {
-      "epoch": 1.495804423036489,
-      "grad_norm": 0.0030105323530733585,
-      "learning_rate": 0.0001999988970009078,
-      "loss": 46.0,
-      "step": 19564
-    },
-    {
-      "epoch": 1.4958808800198788,
-      "grad_norm": 0.0022451234981417656,
-      "learning_rate": 0.00019999889688808944,
-      "loss": 46.0,
-      "step": 19565
-    },
-    {
-      "epoch": 1.4959573370032686,
-      "grad_norm": 0.0031978993210941553,
-      "learning_rate": 0.0001999988967752653,
-      "loss": 46.0,
-      "step": 19566
-    },
-    {
-      "epoch": 1.4960337939866584,
-      "grad_norm": 0.005774857476353645,
-      "learning_rate": 0.00019999889666243536,
-      "loss": 46.0,
-      "step": 19567
-    },
-    {
-      "epoch": 1.496110250970048,
-      "grad_norm": 0.0005562927690334618,
-      "learning_rate": 0.00019999889654959968,
-      "loss": 46.0,
-      "step": 19568
-    },
-    {
-      "epoch": 1.4961867079534377,
-      "grad_norm": 0.021002499386668205,
-      "learning_rate": 0.00019999889643675824,
-      "loss": 46.0,
-      "step": 19569
-    },
-    {
-      "epoch": 1.4962631649368274,
-      "grad_norm": 0.00046855604159645736,
-      "learning_rate": 0.000199998896323911,
-      "loss": 46.0,
-      "step": 19570
-    },
-    {
-      "epoch": 1.4963396219202172,
-      "grad_norm": 0.00031454942654818296,
-      "learning_rate": 0.00019999889621105803,
-      "loss": 46.0,
-      "step": 19571
-    },
-    {
-      "epoch": 1.4964160789036067,
-      "grad_norm": 0.00027260222123004496,
-      "learning_rate": 0.00019999889609819923,
-      "loss": 46.0,
-      "step": 19572
-    },
-    {
-      "epoch": 1.4964925358869965,
-      "grad_norm": 0.0005289419204927981,
-      "learning_rate": 0.00019999889598533473,
-      "loss": 46.0,
-      "step": 19573
-    },
-    {
-      "epoch": 1.4965689928703862,
-      "grad_norm": 0.0016695429803803563,
-      "learning_rate": 0.00019999889587246443,
-      "loss": 46.0,
-      "step": 19574
-    },
-    {
-      "epoch": 1.496645449853776,
-      "grad_norm": 0.0008459216332994401,
-      "learning_rate": 0.00019999889575958834,
-      "loss": 46.0,
-      "step": 19575
-    },
-    {
-      "epoch": 1.4967219068371658,
-      "grad_norm": 0.0031055586878210306,
-      "learning_rate": 0.00019999889564670652,
-      "loss": 46.0,
-      "step": 19576
-    },
-    {
-      "epoch": 1.4967983638205555,
-      "grad_norm": 0.0011080635013058782,
-      "learning_rate": 0.0001999988955338189,
-      "loss": 46.0,
-      "step": 19577
-    },
-    {
-      "epoch": 1.4968748208039453,
-      "grad_norm": 0.001750386436469853,
-      "learning_rate": 0.00019999889542092554,
-      "loss": 46.0,
-      "step": 19578
-    },
-    {
-      "epoch": 1.4969512777873348,
-      "grad_norm": 0.0021380395628511906,
-      "learning_rate": 0.00019999889530802637,
-      "loss": 46.0,
-      "step": 19579
-    },
-    {
-      "epoch": 1.4970277347707246,
-      "grad_norm": 0.001577091170474887,
-      "learning_rate": 0.00019999889519512146,
-      "loss": 46.0,
-      "step": 19580
-    },
-    {
-      "epoch": 1.4971041917541144,
-      "grad_norm": 0.005649860482662916,
-      "learning_rate": 0.00019999889508221077,
-      "loss": 46.0,
-      "step": 19581
-    },
-    {
-      "epoch": 1.4971806487375041,
-      "grad_norm": 0.0021517756395041943,
-      "learning_rate": 0.00019999889496929432,
-      "loss": 46.0,
-      "step": 19582
-    },
-    {
-      "epoch": 1.4972571057208937,
-      "grad_norm": 0.0026646198239177465,
-      "learning_rate": 0.00019999889485637209,
-      "loss": 46.0,
-      "step": 19583
-    },
-    {
-      "epoch": 1.4973335627042834,
-      "grad_norm": 0.0013939602067694068,
-      "learning_rate": 0.00019999889474344408,
-      "loss": 46.0,
-      "step": 19584
-    },
-    {
-      "epoch": 1.4974100196876732,
-      "grad_norm": 0.0006275244522839785,
-      "learning_rate": 0.00019999889463051033,
-      "loss": 46.0,
-      "step": 19585
-    },
-    {
-      "epoch": 1.497486476671063,
-      "grad_norm": 0.0008980878628790379,
-      "learning_rate": 0.00019999889451757078,
-      "loss": 46.0,
-      "step": 19586
-    },
-    {
-      "epoch": 1.4975629336544527,
-      "grad_norm": 0.0012236405164003372,
-      "learning_rate": 0.00019999889440462548,
-      "loss": 46.0,
-      "step": 19587
-    },
-    {
-      "epoch": 1.4976393906378425,
-      "grad_norm": 0.001069837948307395,
-      "learning_rate": 0.0001999988942916744,
-      "loss": 46.0,
-      "step": 19588
-    },
-    {
-      "epoch": 1.4977158476212322,
-      "grad_norm": 0.0023093358613550663,
-      "learning_rate": 0.00019999889417871757,
-      "loss": 46.0,
-      "step": 19589
-    },
-    {
-      "epoch": 1.4977923046046218,
-      "grad_norm": 0.0009686610428616405,
-      "learning_rate": 0.00019999889406575495,
-      "loss": 46.0,
-      "step": 19590
-    },
-    {
-      "epoch": 1.4978687615880115,
-      "grad_norm": 0.002314019249752164,
-      "learning_rate": 0.00019999889395278658,
-      "loss": 46.0,
-      "step": 19591
-    },
-    {
-      "epoch": 1.4979452185714013,
-      "grad_norm": 0.0045213657431304455,
-      "learning_rate": 0.00019999889383981242,
-      "loss": 46.0,
-      "step": 19592
-    },
-    {
-      "epoch": 1.498021675554791,
-      "grad_norm": 0.0038004473317414522,
-      "learning_rate": 0.0001999988937268325,
-      "loss": 46.0,
-      "step": 19593
-    },
-    {
-      "epoch": 1.4980981325381806,
-      "grad_norm": 0.002153785200789571,
-      "learning_rate": 0.00019999889361384682,
-      "loss": 46.0,
-      "step": 19594
-    },
-    {
-      "epoch": 1.4981745895215703,
-      "grad_norm": 0.001252977759577334,
-      "learning_rate": 0.00019999889350085537,
-      "loss": 46.0,
-      "step": 19595
-    },
-    {
-      "epoch": 1.49825104650496,
-      "grad_norm": 0.006883320864289999,
-      "learning_rate": 0.00019999889338785814,
-      "loss": 46.0,
-      "step": 19596
-    },
-    {
-      "epoch": 1.4983275034883499,
-      "grad_norm": 0.0012721295934170485,
-      "learning_rate": 0.00019999889327485516,
-      "loss": 46.0,
-      "step": 19597
-    },
-    {
-      "epoch": 1.4984039604717396,
-      "grad_norm": 0.0016243761638179421,
-      "learning_rate": 0.00019999889316184638,
-      "loss": 46.0,
-      "step": 19598
-    },
-    {
-      "epoch": 1.4984804174551294,
-      "grad_norm": 0.0016149788862094283,
-      "learning_rate": 0.00019999889304883183,
-      "loss": 46.0,
-      "step": 19599
-    },
-    {
-      "epoch": 1.4985568744385191,
-      "grad_norm": 0.03335101529955864,
-      "learning_rate": 0.00019999889293581153,
-      "loss": 46.0,
-      "step": 19600
-    },
-    {
-      "epoch": 1.4986333314219087,
-      "grad_norm": 0.0009267142158932984,
-      "learning_rate": 0.00019999889282278546,
-      "loss": 46.0,
-      "step": 19601
-    },
-    {
-      "epoch": 1.4987097884052984,
-      "grad_norm": 0.0025871070101857185,
-      "learning_rate": 0.00019999889270975362,
-      "loss": 46.0,
-      "step": 19602
-    },
-    {
-      "epoch": 1.4987862453886882,
-      "grad_norm": 0.0006616840837523341,
-      "learning_rate": 0.000199998892596716,
-      "loss": 46.0,
-      "step": 19603
-    },
-    {
-      "epoch": 1.498862702372078,
-      "grad_norm": 0.0013267133617773652,
-      "learning_rate": 0.0001999988924836726,
-      "loss": 46.0,
-      "step": 19604
-    },
-    {
-      "epoch": 1.4989391593554675,
-      "grad_norm": 0.006896014790982008,
-      "learning_rate": 0.00019999889237062347,
-      "loss": 46.0,
-      "step": 19605
-    },
-    {
-      "epoch": 1.4990156163388573,
-      "grad_norm": 0.002806306118145585,
-      "learning_rate": 0.00019999889225756854,
-      "loss": 46.0,
-      "step": 19606
-    },
-    {
-      "epoch": 1.499092073322247,
-      "grad_norm": 0.006940712220966816,
-      "learning_rate": 0.00019999889214450788,
-      "loss": 46.0,
-      "step": 19607
-    },
-    {
-      "epoch": 1.4991685303056368,
-      "grad_norm": 0.0008994821691885591,
-      "learning_rate": 0.00019999889203144137,
-      "loss": 46.0,
-      "step": 19608
-    },
-    {
-      "epoch": 1.4992449872890266,
-      "grad_norm": 0.0034189713187515736,
-      "learning_rate": 0.00019999889191836916,
-      "loss": 46.0,
-      "step": 19609
-    },
-    {
-      "epoch": 1.4993214442724163,
-      "grad_norm": 0.0022839833982288837,
-      "learning_rate": 0.00019999889180529116,
-      "loss": 46.0,
-      "step": 19610
-    },
-    {
-      "epoch": 1.499397901255806,
-      "grad_norm": 0.003786930348724127,
-      "learning_rate": 0.0001999988916922074,
-      "loss": 46.0,
-      "step": 19611
-    },
-    {
-      "epoch": 1.4994743582391956,
-      "grad_norm": 0.0026628703344613314,
-      "learning_rate": 0.00019999889157911786,
-      "loss": 46.0,
-      "step": 19612
-    },
-    {
-      "epoch": 1.4995508152225854,
-      "grad_norm": 0.0037180662620812654,
-      "learning_rate": 0.00019999889146602256,
-      "loss": 46.0,
-      "step": 19613
-    },
-    {
-      "epoch": 1.4996272722059751,
-      "grad_norm": 0.003466070396825671,
-      "learning_rate": 0.00019999889135292146,
-      "loss": 46.0,
-      "step": 19614
-    },
-    {
-      "epoch": 1.499703729189365,
-      "grad_norm": 0.0005706932279281318,
-      "learning_rate": 0.00019999889123981462,
-      "loss": 46.0,
-      "step": 19615
-    },
-    {
-      "epoch": 1.4997801861727544,
-      "grad_norm": 0.0008191209635697305,
-      "learning_rate": 0.00019999889112670203,
-      "loss": 46.0,
-      "step": 19616
-    },
-    {
-      "epoch": 1.4998566431561442,
-      "grad_norm": 0.002530982717871666,
-      "learning_rate": 0.0001999988910135836,
-      "loss": 46.0,
-      "step": 19617
-    },
-    {
-      "epoch": 1.499933100139534,
-      "grad_norm": 0.0015611827839165926,
-      "learning_rate": 0.00019999889090045948,
-      "loss": 46.0,
-      "step": 19618
-    },
-    {
-      "epoch": 1.5000095571229237,
-      "grad_norm": 0.0015027644112706184,
-      "learning_rate": 0.00019999889078732954,
-      "loss": 46.0,
-      "step": 19619
-    },
-    {
-      "epoch": 1.5000860141063135,
-      "grad_norm": 0.0010443015489727259,
-      "learning_rate": 0.00019999889067419385,
-      "loss": 46.0,
-      "step": 19620
-    },
-    {
-      "epoch": 1.5001624710897032,
-      "grad_norm": 0.0006005424074828625,
-      "learning_rate": 0.0001999988905610524,
-      "loss": 46.0,
-      "step": 19621
-    },
-    {
-      "epoch": 1.500238928073093,
-      "grad_norm": 0.0046441699378192425,
-      "learning_rate": 0.00019999889044790517,
-      "loss": 46.0,
-      "step": 19622
-    },
-    {
-      "epoch": 1.5003153850564825,
-      "grad_norm": 0.001452158554457128,
-      "learning_rate": 0.00019999889033475216,
-      "loss": 46.0,
-      "step": 19623
-    },
-    {
-      "epoch": 1.5003918420398723,
-      "grad_norm": 0.009183228015899658,
-      "learning_rate": 0.0001999988902215934,
-      "loss": 46.0,
-      "step": 19624
-    },
-    {
-      "epoch": 1.500468299023262,
-      "grad_norm": 0.001078345812857151,
-      "learning_rate": 0.00019999889010842884,
-      "loss": 46.0,
-      "step": 19625
-    },
-    {
-      "epoch": 1.5005447560066516,
-      "grad_norm": 0.0058320192620158195,
-      "learning_rate": 0.00019999888999525857,
-      "loss": 46.0,
-      "step": 19626
-    },
-    {
-      "epoch": 1.5006212129900414,
-      "grad_norm": 0.0004143244295846671,
-      "learning_rate": 0.00019999888988208247,
-      "loss": 46.0,
-      "step": 19627
-    },
-    {
-      "epoch": 1.5006976699734311,
-      "grad_norm": 0.0020292585249990225,
-      "learning_rate": 0.0001999988897689006,
-      "loss": 46.0,
-      "step": 19628
-    },
-    {
-      "epoch": 1.5007741269568209,
-      "grad_norm": 0.0009465735056437552,
-      "learning_rate": 0.000199998889655713,
-      "loss": 46.0,
-      "step": 19629
-    },
-    {
-      "epoch": 1.5008505839402106,
-      "grad_norm": 0.0013611565809696913,
-      "learning_rate": 0.0001999988895425196,
-      "loss": 46.0,
-      "step": 19630
-    },
-    {
-      "epoch": 1.5009270409236004,
-      "grad_norm": 0.0036437571980059147,
-      "learning_rate": 0.00019999888942932046,
-      "loss": 46.0,
-      "step": 19631
-    },
-    {
-      "epoch": 1.5010034979069902,
-      "grad_norm": 0.0006971657858230174,
-      "learning_rate": 0.00019999888931611552,
-      "loss": 46.0,
-      "step": 19632
-    },
-    {
-      "epoch": 1.50107995489038,
-      "grad_norm": 0.002104934537783265,
-      "learning_rate": 0.0001999988892029048,
-      "loss": 46.0,
-      "step": 19633
-    },
-    {
-      "epoch": 1.5011564118737695,
-      "grad_norm": 0.0009927240898832679,
-      "learning_rate": 0.00019999888908968835,
-      "loss": 46.0,
-      "step": 19634
-    },
-    {
-      "epoch": 1.5012328688571592,
-      "grad_norm": 0.0010329802753403783,
-      "learning_rate": 0.00019999888897646612,
-      "loss": 46.0,
-      "step": 19635
-    },
-    {
-      "epoch": 1.501309325840549,
-      "grad_norm": 0.0012246761471033096,
-      "learning_rate": 0.00019999888886323812,
-      "loss": 46.0,
-      "step": 19636
-    },
-    {
-      "epoch": 1.5013857828239385,
-      "grad_norm": 0.0004657434765249491,
-      "learning_rate": 0.00019999888875000434,
-      "loss": 46.0,
-      "step": 19637
-    },
-    {
-      "epoch": 1.5014622398073283,
-      "grad_norm": 0.002692838665097952,
-      "learning_rate": 0.00019999888863676482,
-      "loss": 46.0,
-      "step": 19638
-    },
-    {
-      "epoch": 1.501538696790718,
-      "grad_norm": 0.0014246393693611026,
-      "learning_rate": 0.00019999888852351952,
-      "loss": 46.0,
-      "step": 19639
-    },
-    {
-      "epoch": 1.5016151537741078,
-      "grad_norm": 0.0006924595218151808,
-      "learning_rate": 0.00019999888841026843,
-      "loss": 46.0,
-      "step": 19640
-    },
-    {
-      "epoch": 1.5016916107574976,
-      "grad_norm": 0.0008101038401946425,
-      "learning_rate": 0.00019999888829701155,
-      "loss": 46.0,
-      "step": 19641
-    },
-    {
-      "epoch": 1.5017680677408873,
-      "grad_norm": 0.0018298059003427625,
-      "learning_rate": 0.00019999888818374894,
-      "loss": 46.0,
-      "step": 19642
-    },
-    {
-      "epoch": 1.501844524724277,
-      "grad_norm": 0.0005969632184132934,
-      "learning_rate": 0.00019999888807048058,
-      "loss": 46.0,
-      "step": 19643
-    },
-    {
-      "epoch": 1.5019209817076669,
-      "grad_norm": 0.0038788749370723963,
-      "learning_rate": 0.0001999988879572064,
-      "loss": 46.0,
-      "step": 19644
-    },
-    {
-      "epoch": 1.5019974386910564,
-      "grad_norm": 0.002375392708927393,
-      "learning_rate": 0.00019999888784392648,
-      "loss": 46.0,
-      "step": 19645
-    },
-    {
-      "epoch": 1.5020738956744462,
-      "grad_norm": 0.0006361521664075553,
-      "learning_rate": 0.00019999888773064077,
-      "loss": 46.0,
-      "step": 19646
-    },
-    {
-      "epoch": 1.502150352657836,
-      "grad_norm": 0.0008676521829329431,
-      "learning_rate": 0.0001999988876173493,
-      "loss": 46.0,
-      "step": 19647
-    },
-    {
-      "epoch": 1.5022268096412255,
-      "grad_norm": 0.0005255150608718395,
-      "learning_rate": 0.00019999888750405208,
-      "loss": 46.0,
-      "step": 19648
-    },
-    {
-      "epoch": 1.5023032666246152,
-      "grad_norm": 0.00031688372837379575,
-      "learning_rate": 0.00019999888739074905,
-      "loss": 46.0,
-      "step": 19649
-    },
-    {
-      "epoch": 1.502379723608005,
-      "grad_norm": 0.00045653030974790454,
-      "learning_rate": 0.0001999988872774403,
-      "loss": 46.0,
-      "step": 19650
-    },
-    {
-      "epoch": 1.5024561805913947,
-      "grad_norm": 0.005554562900215387,
-      "learning_rate": 0.00019999888716412575,
-      "loss": 46.0,
-      "step": 19651
-    },
-    {
-      "epoch": 1.5025326375747845,
-      "grad_norm": 0.0020736439619213343,
-      "learning_rate": 0.00019999888705080543,
-      "loss": 46.0,
-      "step": 19652
-    },
-    {
-      "epoch": 1.5026090945581743,
-      "grad_norm": 0.004107438027858734,
-      "learning_rate": 0.00019999888693747936,
-      "loss": 46.0,
-      "step": 19653
-    },
-    {
-      "epoch": 1.502685551541564,
-      "grad_norm": 0.0016275091329589486,
-      "learning_rate": 0.0001999988868241475,
-      "loss": 46.0,
-      "step": 19654
-    },
-    {
-      "epoch": 1.5027620085249538,
-      "grad_norm": 0.001553536974824965,
-      "learning_rate": 0.00019999888671080987,
-      "loss": 46.0,
-      "step": 19655
-    },
-    {
-      "epoch": 1.5028384655083433,
-      "grad_norm": 0.0013049402041360736,
-      "learning_rate": 0.0001999988865974665,
-      "loss": 46.0,
-      "step": 19656
-    },
-    {
-      "epoch": 1.502914922491733,
-      "grad_norm": 0.0013029269175603986,
-      "learning_rate": 0.00019999888648411732,
-      "loss": 46.0,
-      "step": 19657
-    },
-    {
-      "epoch": 1.5029913794751228,
-      "grad_norm": 0.0009008460910990834,
-      "learning_rate": 0.00019999888637076238,
-      "loss": 46.0,
-      "step": 19658
-    },
-    {
-      "epoch": 1.5030678364585124,
-      "grad_norm": 0.0007676975801587105,
-      "learning_rate": 0.00019999888625740167,
-      "loss": 46.0,
-      "step": 19659
-    },
-    {
-      "epoch": 1.5031442934419021,
-      "grad_norm": 0.003085235133767128,
-      "learning_rate": 0.00019999888614403524,
-      "loss": 46.0,
-      "step": 19660
-    },
-    {
-      "epoch": 1.503220750425292,
-      "grad_norm": 0.0006421257858164608,
-      "learning_rate": 0.000199998886030663,
-      "loss": 46.0,
-      "step": 19661
-    },
-    {
-      "epoch": 1.5032972074086817,
-      "grad_norm": 0.002629264257848263,
-      "learning_rate": 0.00019999888591728499,
-      "loss": 46.0,
-      "step": 19662
-    },
-    {
-      "epoch": 1.5033736643920714,
-      "grad_norm": 0.000902424450032413,
-      "learning_rate": 0.0001999988858039012,
-      "loss": 46.0,
-      "step": 19663
-    },
-    {
-      "epoch": 1.5034501213754612,
-      "grad_norm": 0.002749947365373373,
-      "learning_rate": 0.00019999888569051166,
-      "loss": 46.0,
-      "step": 19664
-    },
-    {
-      "epoch": 1.503526578358851,
-      "grad_norm": 0.0008771510329097509,
-      "learning_rate": 0.00019999888557711634,
-      "loss": 46.0,
-      "step": 19665
-    },
-    {
-      "epoch": 1.5036030353422407,
-      "grad_norm": 0.0018273703753948212,
-      "learning_rate": 0.00019999888546371527,
-      "loss": 46.0,
-      "step": 19666
-    },
-    {
-      "epoch": 1.5036794923256303,
-      "grad_norm": 0.0004970564623363316,
-      "learning_rate": 0.0001999988853503084,
-      "loss": 46.0,
-      "step": 19667
-    },
-    {
-      "epoch": 1.50375594930902,
-      "grad_norm": 0.0012448756024241447,
-      "learning_rate": 0.00019999888523689578,
-      "loss": 46.0,
-      "step": 19668
-    },
-    {
-      "epoch": 1.5038324062924098,
-      "grad_norm": 0.0006642649532295763,
-      "learning_rate": 0.0001999988851234774,
-      "loss": 46.0,
-      "step": 19669
-    },
-    {
-      "epoch": 1.5039088632757993,
-      "grad_norm": 0.000801940041128546,
-      "learning_rate": 0.00019999888501005323,
-      "loss": 46.0,
-      "step": 19670
-    },
-    {
-      "epoch": 1.503985320259189,
-      "grad_norm": 0.0022543715313076973,
-      "learning_rate": 0.0001999988848966233,
-      "loss": 46.0,
-      "step": 19671
-    },
-    {
-      "epoch": 1.5040617772425788,
-      "grad_norm": 0.0014665884664282203,
-      "learning_rate": 0.0001999988847831876,
-      "loss": 46.0,
-      "step": 19672
-    },
-    {
-      "epoch": 1.5041382342259686,
-      "grad_norm": 0.0003214908065274358,
-      "learning_rate": 0.00019999888466974613,
-      "loss": 46.0,
-      "step": 19673
-    },
-    {
-      "epoch": 1.5042146912093584,
-      "grad_norm": 0.001157070160843432,
-      "learning_rate": 0.00019999888455629887,
-      "loss": 46.0,
-      "step": 19674
-    },
-    {
-      "epoch": 1.5042911481927481,
-      "grad_norm": 0.014479960314929485,
-      "learning_rate": 0.0001999988844428459,
-      "loss": 46.0,
-      "step": 19675
-    },
-    {
-      "epoch": 1.5043676051761379,
-      "grad_norm": 0.000796623935457319,
-      "learning_rate": 0.0001999988843293871,
-      "loss": 46.0,
-      "step": 19676
-    },
-    {
-      "epoch": 1.5044440621595276,
-      "grad_norm": 0.0005131034413352609,
-      "learning_rate": 0.00019999888421592255,
-      "loss": 46.0,
-      "step": 19677
-    },
-    {
-      "epoch": 1.5045205191429172,
-      "grad_norm": 0.0030730010475963354,
-      "learning_rate": 0.00019999888410245222,
-      "loss": 46.0,
-      "step": 19678
-    },
-    {
-      "epoch": 1.504596976126307,
-      "grad_norm": 0.0005181065644137561,
-      "learning_rate": 0.00019999888398897615,
-      "loss": 46.0,
-      "step": 19679
-    },
-    {
-      "epoch": 1.5046734331096967,
-      "grad_norm": 0.0009171762503683567,
-      "learning_rate": 0.00019999888387549428,
-      "loss": 46.0,
-      "step": 19680
-    },
-    {
-      "epoch": 1.5047498900930862,
-      "grad_norm": 0.0012362569104880095,
-      "learning_rate": 0.00019999888376200667,
-      "loss": 46.0,
-      "step": 19681
-    },
-    {
-      "epoch": 1.504826347076476,
-      "grad_norm": 0.001237946911714971,
-      "learning_rate": 0.00019999888364851328,
-      "loss": 46.0,
-      "step": 19682
-    },
-    {
-      "epoch": 1.5049028040598658,
-      "grad_norm": 0.0007379649905487895,
-      "learning_rate": 0.0001999988835350141,
-      "loss": 46.0,
-      "step": 19683
-    },
-    {
-      "epoch": 1.5049792610432555,
-      "grad_norm": 0.001304334495216608,
-      "learning_rate": 0.00019999888342150916,
-      "loss": 46.0,
-      "step": 19684
-    },
-    {
-      "epoch": 1.5050557180266453,
-      "grad_norm": 0.0013156364439055324,
-      "learning_rate": 0.00019999888330799847,
-      "loss": 46.0,
-      "step": 19685
-    },
-    {
-      "epoch": 1.505132175010035,
-      "grad_norm": 0.0022053250577300787,
-      "learning_rate": 0.000199998883194482,
-      "loss": 46.0,
-      "step": 19686
-    },
-    {
-      "epoch": 1.5052086319934248,
-      "grad_norm": 0.0009780323598533869,
-      "learning_rate": 0.00019999888308095973,
-      "loss": 46.0,
-      "step": 19687
-    },
-    {
-      "epoch": 1.5052850889768146,
-      "grad_norm": 0.001505318097770214,
-      "learning_rate": 0.00019999888296743173,
-      "loss": 46.0,
-      "step": 19688
-    },
-    {
-      "epoch": 1.505361545960204,
-      "grad_norm": 0.0036668111570179462,
-      "learning_rate": 0.00019999888285389793,
-      "loss": 46.0,
-      "step": 19689
-    },
-    {
-      "epoch": 1.5054380029435939,
-      "grad_norm": 0.0027202495839446783,
-      "learning_rate": 0.0001999988827403584,
-      "loss": 46.0,
-      "step": 19690
-    },
-    {
-      "epoch": 1.5055144599269836,
-      "grad_norm": 0.008233198896050453,
-      "learning_rate": 0.0001999988826268131,
-      "loss": 46.0,
-      "step": 19691
-    },
-    {
-      "epoch": 1.5055909169103732,
-      "grad_norm": 0.0010036530438810587,
-      "learning_rate": 0.000199998882513262,
-      "loss": 46.0,
-      "step": 19692
-    },
-    {
-      "epoch": 1.505667373893763,
-      "grad_norm": 0.0010734088718891144,
-      "learning_rate": 0.00019999888239970512,
-      "loss": 46.0,
-      "step": 19693
-    },
-    {
-      "epoch": 1.5057438308771527,
-      "grad_norm": 0.001148715615272522,
-      "learning_rate": 0.0001999988822861425,
-      "loss": 46.0,
-      "step": 19694
-    },
-    {
-      "epoch": 1.5058202878605425,
-      "grad_norm": 0.0015884805470705032,
-      "learning_rate": 0.0001999988821725741,
-      "loss": 46.0,
-      "step": 19695
-    },
-    {
-      "epoch": 1.5058967448439322,
-      "grad_norm": 0.0012611215934157372,
-      "learning_rate": 0.00019999888205899996,
-      "loss": 46.0,
-      "step": 19696
-    },
-    {
-      "epoch": 1.505973201827322,
-      "grad_norm": 0.008940557949244976,
-      "learning_rate": 0.00019999888194542002,
-      "loss": 46.0,
-      "step": 19697
-    },
-    {
-      "epoch": 1.5060496588107117,
-      "grad_norm": 0.0007845713407732546,
-      "learning_rate": 0.0001999988818318343,
-      "loss": 46.0,
-      "step": 19698
-    },
-    {
-      "epoch": 1.5061261157941015,
-      "grad_norm": 0.0018031923100352287,
-      "learning_rate": 0.0001999988817182428,
-      "loss": 46.0,
-      "step": 19699
-    },
-    {
-      "epoch": 1.506202572777491,
-      "grad_norm": 0.001237313961610198,
-      "learning_rate": 0.0001999988816046456,
-      "loss": 46.0,
-      "step": 19700
-    },
-    {
-      "epoch": 1.5062790297608808,
-      "grad_norm": 0.01438084151595831,
-      "learning_rate": 0.00019999888149104255,
-      "loss": 46.0,
-      "step": 19701
-    },
-    {
-      "epoch": 1.5063554867442706,
-      "grad_norm": 0.0011346966493874788,
-      "learning_rate": 0.0001999988813774338,
-      "loss": 46.0,
-      "step": 19702
-    },
-    {
-      "epoch": 1.50643194372766,
-      "grad_norm": 0.0008850598824210465,
-      "learning_rate": 0.00019999888126381925,
-      "loss": 46.0,
-      "step": 19703
-    },
-    {
-      "epoch": 1.5065084007110499,
-      "grad_norm": 0.0011766334064304829,
-      "learning_rate": 0.00019999888115019893,
-      "loss": 46.0,
-      "step": 19704
-    },
-    {
-      "epoch": 1.5065848576944396,
-      "grad_norm": 0.0009010198409669101,
-      "learning_rate": 0.00019999888103657283,
-      "loss": 46.0,
-      "step": 19705
-    },
-    {
-      "epoch": 1.5066613146778294,
-      "grad_norm": 0.0010107802227139473,
-      "learning_rate": 0.00019999888092294094,
-      "loss": 46.0,
-      "step": 19706
-    },
-    {
-      "epoch": 1.5067377716612191,
-      "grad_norm": 0.0011119669070467353,
-      "learning_rate": 0.00019999888080930335,
-      "loss": 46.0,
-      "step": 19707
-    },
-    {
-      "epoch": 1.506814228644609,
-      "grad_norm": 0.0011931966291740537,
-      "learning_rate": 0.00019999888069565993,
-      "loss": 46.0,
-      "step": 19708
-    },
-    {
-      "epoch": 1.5068906856279987,
-      "grad_norm": 0.0009842299623414874,
-      "learning_rate": 0.00019999888058201077,
-      "loss": 46.0,
-      "step": 19709
-    },
-    {
-      "epoch": 1.5069671426113884,
-      "grad_norm": 0.0014397201593965292,
-      "learning_rate": 0.0001999988804683558,
-      "loss": 46.0,
-      "step": 19710
-    },
-    {
-      "epoch": 1.507043599594778,
-      "grad_norm": 0.0009246676345355809,
-      "learning_rate": 0.00019999888035469513,
-      "loss": 46.0,
-      "step": 19711
-    },
-    {
-      "epoch": 1.5071200565781677,
-      "grad_norm": 0.010779195465147495,
-      "learning_rate": 0.00019999888024102865,
-      "loss": 46.0,
-      "step": 19712
-    },
-    {
-      "epoch": 1.5071965135615575,
-      "grad_norm": 0.000838337407913059,
-      "learning_rate": 0.0001999988801273564,
-      "loss": 46.0,
-      "step": 19713
-    },
-    {
-      "epoch": 1.507272970544947,
-      "grad_norm": 0.0021979818120598793,
-      "learning_rate": 0.00019999888001367836,
-      "loss": 46.0,
-      "step": 19714
-    },
-    {
-      "epoch": 1.5073494275283368,
-      "grad_norm": 0.0008863132097758353,
-      "learning_rate": 0.0001999988798999946,
-      "loss": 46.0,
-      "step": 19715
-    },
-    {
-      "epoch": 1.5074258845117265,
-      "grad_norm": 0.0014464984415099025,
-      "learning_rate": 0.00019999887978630504,
-      "loss": 46.0,
-      "step": 19716
-    },
-    {
-      "epoch": 1.5075023414951163,
-      "grad_norm": 0.001759712235070765,
-      "learning_rate": 0.00019999887967260972,
-      "loss": 46.0,
-      "step": 19717
-    },
-    {
-      "epoch": 1.507578798478506,
-      "grad_norm": 0.005756087601184845,
-      "learning_rate": 0.00019999887955890865,
-      "loss": 46.0,
-      "step": 19718
-    },
-    {
-      "epoch": 1.5076552554618958,
-      "grad_norm": 0.0007727245683781803,
-      "learning_rate": 0.00019999887944520178,
-      "loss": 46.0,
-      "step": 19719
-    },
-    {
-      "epoch": 1.5077317124452856,
-      "grad_norm": 0.0018911795923486352,
-      "learning_rate": 0.00019999887933148914,
-      "loss": 46.0,
-      "step": 19720
-    },
-    {
-      "epoch": 1.5078081694286753,
-      "grad_norm": 0.0010581546230241656,
-      "learning_rate": 0.00019999887921777072,
-      "loss": 46.0,
-      "step": 19721
-    },
-    {
-      "epoch": 1.5078846264120649,
-      "grad_norm": 0.004947170149534941,
-      "learning_rate": 0.00019999887910404656,
-      "loss": 46.0,
-      "step": 19722
-    },
-    {
-      "epoch": 1.5079610833954546,
-      "grad_norm": 0.005442941095679998,
-      "learning_rate": 0.00019999887899031663,
-      "loss": 46.0,
-      "step": 19723
-    },
-    {
-      "epoch": 1.5080375403788442,
-      "grad_norm": 0.0009650898864492774,
-      "learning_rate": 0.00019999887887658092,
-      "loss": 46.0,
-      "step": 19724
-    },
-    {
-      "epoch": 1.508113997362234,
-      "grad_norm": 0.0005430366145446897,
-      "learning_rate": 0.00019999887876283944,
-      "loss": 46.0,
-      "step": 19725
-    },
-    {
-      "epoch": 1.5081904543456237,
-      "grad_norm": 0.0025717385578900576,
-      "learning_rate": 0.00019999887864909218,
-      "loss": 46.0,
-      "step": 19726
-    },
-    {
-      "epoch": 1.5082669113290135,
-      "grad_norm": 0.005509734619408846,
-      "learning_rate": 0.00019999887853533915,
-      "loss": 46.0,
-      "step": 19727
-    },
-    {
-      "epoch": 1.5083433683124032,
-      "grad_norm": 0.0022998247295618057,
-      "learning_rate": 0.00019999887842158035,
-      "loss": 46.0,
-      "step": 19728
-    },
-    {
-      "epoch": 1.508419825295793,
-      "grad_norm": 0.0012572853593155742,
-      "learning_rate": 0.0001999988783078158,
-      "loss": 46.0,
-      "step": 19729
-    },
-    {
-      "epoch": 1.5084962822791828,
-      "grad_norm": 0.0007268390618264675,
-      "learning_rate": 0.00019999887819404548,
-      "loss": 46.0,
-      "step": 19730
-    },
-    {
-      "epoch": 1.5085727392625725,
-      "grad_norm": 0.0019113688031211495,
-      "learning_rate": 0.0001999988780802694,
-      "loss": 46.0,
-      "step": 19731
-    },
-    {
-      "epoch": 1.5086491962459623,
-      "grad_norm": 0.0024600664619356394,
-      "learning_rate": 0.00019999887796648752,
-      "loss": 46.0,
-      "step": 19732
-    },
-    {
-      "epoch": 1.5087256532293518,
-      "grad_norm": 0.0015150255057960749,
-      "learning_rate": 0.0001999988778526999,
-      "loss": 46.0,
-      "step": 19733
-    },
-    {
-      "epoch": 1.5088021102127416,
-      "grad_norm": 0.0005535635282285511,
-      "learning_rate": 0.0001999988777389065,
-      "loss": 46.0,
-      "step": 19734
-    },
-    {
-      "epoch": 1.5088785671961311,
-      "grad_norm": 0.0010429403046146035,
-      "learning_rate": 0.0001999988776251073,
-      "loss": 46.0,
-      "step": 19735
-    },
-    {
-      "epoch": 1.5089550241795209,
-      "grad_norm": 0.001272533554583788,
-      "learning_rate": 0.00019999887751130237,
-      "loss": 46.0,
-      "step": 19736
-    },
-    {
-      "epoch": 1.5090314811629106,
-      "grad_norm": 0.0007790165254846215,
-      "learning_rate": 0.0001999988773974917,
-      "loss": 46.0,
-      "step": 19737
-    },
-    {
-      "epoch": 1.5091079381463004,
-      "grad_norm": 0.0024776593782007694,
-      "learning_rate": 0.00019999887728367518,
-      "loss": 46.0,
-      "step": 19738
-    },
-    {
-      "epoch": 1.5091843951296902,
-      "grad_norm": 0.0009201911743730307,
-      "learning_rate": 0.00019999887716985293,
-      "loss": 46.0,
-      "step": 19739
-    },
-    {
-      "epoch": 1.50926085211308,
-      "grad_norm": 0.0014332858845591545,
-      "learning_rate": 0.00019999887705602492,
-      "loss": 46.0,
-      "step": 19740
-    },
-    {
-      "epoch": 1.5093373090964697,
-      "grad_norm": 0.0010543070966377854,
-      "learning_rate": 0.00019999887694219112,
-      "loss": 46.0,
-      "step": 19741
-    },
-    {
-      "epoch": 1.5094137660798594,
-      "grad_norm": 0.002088436158373952,
-      "learning_rate": 0.00019999887682835158,
-      "loss": 46.0,
-      "step": 19742
-    },
-    {
-      "epoch": 1.5094902230632492,
-      "grad_norm": 0.0014168419875204563,
-      "learning_rate": 0.00019999887671450623,
-      "loss": 46.0,
-      "step": 19743
-    },
-    {
-      "epoch": 1.5095666800466387,
-      "grad_norm": 0.005782569758594036,
-      "learning_rate": 0.00019999887660065513,
-      "loss": 46.0,
-      "step": 19744
-    },
-    {
-      "epoch": 1.5096431370300285,
-      "grad_norm": 0.0011829511495307088,
-      "learning_rate": 0.00019999887648679827,
-      "loss": 46.0,
-      "step": 19745
-    },
-    {
-      "epoch": 1.509719594013418,
-      "grad_norm": 0.0007292631780728698,
-      "learning_rate": 0.00019999887637293565,
-      "loss": 46.0,
-      "step": 19746
-    },
-    {
-      "epoch": 1.5097960509968078,
-      "grad_norm": 0.0018247476546093822,
-      "learning_rate": 0.00019999887625906724,
-      "loss": 46.0,
-      "step": 19747
-    },
-    {
-      "epoch": 1.5098725079801976,
-      "grad_norm": 0.0010375638958066702,
-      "learning_rate": 0.00019999887614519305,
-      "loss": 46.0,
-      "step": 19748
-    },
-    {
-      "epoch": 1.5099489649635873,
-      "grad_norm": 0.0006080397288314998,
-      "learning_rate": 0.00019999887603131312,
-      "loss": 46.0,
-      "step": 19749
-    },
-    {
-      "epoch": 1.510025421946977,
-      "grad_norm": 0.0009214449673891068,
-      "learning_rate": 0.00019999887591742738,
-      "loss": 46.0,
-      "step": 19750
-    },
-    {
-      "epoch": 1.5101018789303668,
-      "grad_norm": 0.00048188489745371044,
-      "learning_rate": 0.0001999988758035359,
-      "loss": 46.0,
-      "step": 19751
-    },
-    {
-      "epoch": 1.5101783359137566,
-      "grad_norm": 0.0010743956081569195,
-      "learning_rate": 0.00019999887568963867,
-      "loss": 46.0,
-      "step": 19752
-    },
-    {
-      "epoch": 1.5102547928971464,
-      "grad_norm": 0.0010220925323665142,
-      "learning_rate": 0.00019999887557573565,
-      "loss": 46.0,
-      "step": 19753
-    },
-    {
-      "epoch": 1.510331249880536,
-      "grad_norm": 0.0004688806366175413,
-      "learning_rate": 0.00019999887546182684,
-      "loss": 46.0,
-      "step": 19754
-    },
-    {
-      "epoch": 1.5104077068639257,
-      "grad_norm": 0.0008299436885863543,
-      "learning_rate": 0.0001999988753479123,
-      "loss": 46.0,
-      "step": 19755
-    },
-    {
-      "epoch": 1.5104841638473154,
-      "grad_norm": 0.006093153730034828,
-      "learning_rate": 0.00019999887523399195,
-      "loss": 46.0,
-      "step": 19756
-    },
-    {
-      "epoch": 1.510560620830705,
-      "grad_norm": 0.0011010019807145,
-      "learning_rate": 0.00019999887512006588,
-      "loss": 46.0,
-      "step": 19757
-    },
-    {
-      "epoch": 1.5106370778140947,
-      "grad_norm": 0.0006050923257134855,
-      "learning_rate": 0.000199998875006134,
-      "loss": 46.0,
-      "step": 19758
-    },
-    {
-      "epoch": 1.5107135347974845,
-      "grad_norm": 0.0036128507927060127,
-      "learning_rate": 0.00019999887489219635,
-      "loss": 46.0,
-      "step": 19759
-    },
-    {
-      "epoch": 1.5107899917808743,
-      "grad_norm": 0.0005929625476710498,
-      "learning_rate": 0.00019999887477825293,
-      "loss": 46.0,
-      "step": 19760
-    },
-    {
-      "epoch": 1.510866448764264,
-      "grad_norm": 0.002497768495231867,
-      "learning_rate": 0.00019999887466430377,
-      "loss": 46.0,
-      "step": 19761
-    },
-    {
-      "epoch": 1.5109429057476538,
-      "grad_norm": 0.0007081806543283165,
-      "learning_rate": 0.0001999988745503488,
-      "loss": 46.0,
-      "step": 19762
-    },
-    {
-      "epoch": 1.5110193627310435,
-      "grad_norm": 0.0013643552083522081,
-      "learning_rate": 0.0001999988744363881,
-      "loss": 46.0,
-      "step": 19763
-    },
-    {
-      "epoch": 1.5110958197144333,
-      "grad_norm": 0.000939934398047626,
-      "learning_rate": 0.00019999887432242162,
-      "loss": 46.0,
-      "step": 19764
-    },
-    {
-      "epoch": 1.5111722766978228,
-      "grad_norm": 0.0010947646806016564,
-      "learning_rate": 0.00019999887420844937,
-      "loss": 46.0,
-      "step": 19765
-    },
-    {
-      "epoch": 1.5112487336812126,
-      "grad_norm": 0.0008887078147381544,
-      "learning_rate": 0.00019999887409447134,
-      "loss": 46.0,
-      "step": 19766
-    },
-    {
-      "epoch": 1.5113251906646024,
-      "grad_norm": 0.00033823965350165963,
-      "learning_rate": 0.00019999887398048754,
-      "loss": 46.0,
-      "step": 19767
-    },
-    {
-      "epoch": 1.511401647647992,
-      "grad_norm": 0.00070320995291695,
-      "learning_rate": 0.000199998873866498,
-      "loss": 46.0,
-      "step": 19768
-    },
-    {
-      "epoch": 1.5114781046313817,
-      "grad_norm": 0.00603750953450799,
-      "learning_rate": 0.00019999887375250265,
-      "loss": 46.0,
-      "step": 19769
-    },
-    {
-      "epoch": 1.5115545616147714,
-      "grad_norm": 0.0007358907023444772,
-      "learning_rate": 0.00019999887363850153,
-      "loss": 46.0,
-      "step": 19770
-    },
-    {
-      "epoch": 1.5116310185981612,
-      "grad_norm": 0.0017141265561804175,
-      "learning_rate": 0.00019999887352449469,
-      "loss": 46.0,
-      "step": 19771
-    },
-    {
-      "epoch": 1.511707475581551,
-      "grad_norm": 0.0012082947650924325,
-      "learning_rate": 0.00019999887341048205,
-      "loss": 46.0,
-      "step": 19772
-    },
-    {
-      "epoch": 1.5117839325649407,
-      "grad_norm": 0.002776392502710223,
-      "learning_rate": 0.00019999887329646363,
-      "loss": 46.0,
-      "step": 19773
-    },
-    {
-      "epoch": 1.5118603895483305,
-      "grad_norm": 0.0014532109489664435,
-      "learning_rate": 0.00019999887318243947,
-      "loss": 46.0,
-      "step": 19774
-    },
-    {
-      "epoch": 1.5119368465317202,
-      "grad_norm": 0.008840211667120457,
-      "learning_rate": 0.0001999988730684095,
-      "loss": 46.0,
-      "step": 19775
-    },
-    {
-      "epoch": 1.5120133035151098,
-      "grad_norm": 0.002735511865466833,
-      "learning_rate": 0.00019999887295437378,
-      "loss": 46.0,
-      "step": 19776
-    },
-    {
-      "epoch": 1.5120897604984995,
-      "grad_norm": 0.0020389084238559008,
-      "learning_rate": 0.0001999988728403323,
-      "loss": 46.0,
-      "step": 19777
-    },
-    {
-      "epoch": 1.5121662174818893,
-      "grad_norm": 0.004298189654946327,
-      "learning_rate": 0.00019999887272628507,
-      "loss": 46.0,
-      "step": 19778
-    },
-    {
-      "epoch": 1.5122426744652788,
-      "grad_norm": 0.0005396513734012842,
-      "learning_rate": 0.00019999887261223202,
-      "loss": 46.0,
-      "step": 19779
-    },
-    {
-      "epoch": 1.5123191314486686,
-      "grad_norm": 0.0008666576468385756,
-      "learning_rate": 0.00019999887249817322,
-      "loss": 46.0,
-      "step": 19780
-    },
-    {
-      "epoch": 1.5123955884320583,
-      "grad_norm": 0.005232043098658323,
-      "learning_rate": 0.00019999887238410867,
-      "loss": 46.0,
-      "step": 19781
-    },
-    {
-      "epoch": 1.512472045415448,
-      "grad_norm": 0.001679981709457934,
-      "learning_rate": 0.00019999887227003832,
-      "loss": 46.0,
-      "step": 19782
-    },
-    {
-      "epoch": 1.5125485023988379,
-      "grad_norm": 0.0016673285281285644,
-      "learning_rate": 0.0001999988721559622,
-      "loss": 46.0,
-      "step": 19783
-    },
-    {
-      "epoch": 1.5126249593822276,
-      "grad_norm": 0.005448025651276112,
-      "learning_rate": 0.00019999887204188033,
-      "loss": 46.0,
-      "step": 19784
-    },
-    {
-      "epoch": 1.5127014163656174,
-      "grad_norm": 0.0016198459779843688,
-      "learning_rate": 0.0001999988719277927,
-      "loss": 46.0,
-      "step": 19785
-    },
-    {
-      "epoch": 1.5127778733490072,
-      "grad_norm": 0.0003494654083624482,
-      "learning_rate": 0.0001999988718136993,
-      "loss": 46.0,
-      "step": 19786
-    },
-    {
-      "epoch": 1.5128543303323967,
-      "grad_norm": 0.0015478498535230756,
-      "learning_rate": 0.0001999988716996001,
-      "loss": 46.0,
-      "step": 19787
-    },
-    {
-      "epoch": 1.5129307873157865,
-      "grad_norm": 0.0007455118466168642,
-      "learning_rate": 0.00019999887158549516,
-      "loss": 46.0,
-      "step": 19788
-    },
-    {
-      "epoch": 1.5130072442991762,
-      "grad_norm": 0.0004042270011268556,
-      "learning_rate": 0.00019999887147138443,
-      "loss": 46.0,
-      "step": 19789
-    },
-    {
-      "epoch": 1.5130837012825658,
-      "grad_norm": 0.005060931202024221,
-      "learning_rate": 0.00019999887135726795,
-      "loss": 46.0,
-      "step": 19790
-    },
-    {
-      "epoch": 1.5131601582659555,
-      "grad_norm": 0.03275544196367264,
-      "learning_rate": 0.0001999988712431457,
-      "loss": 46.0,
-      "step": 19791
-    },
-    {
-      "epoch": 1.5132366152493453,
-      "grad_norm": 0.0015940743032842875,
-      "learning_rate": 0.00019999887112901767,
-      "loss": 46.0,
-      "step": 19792
-    },
-    {
-      "epoch": 1.513313072232735,
-      "grad_norm": 0.001091941841877997,
-      "learning_rate": 0.00019999887101488387,
-      "loss": 46.0,
-      "step": 19793
-    },
-    {
-      "epoch": 1.5133895292161248,
-      "grad_norm": 0.003730242606252432,
-      "learning_rate": 0.0001999988709007443,
-      "loss": 46.0,
-      "step": 19794
-    },
-    {
-      "epoch": 1.5134659861995146,
-      "grad_norm": 0.0009056238923221827,
-      "learning_rate": 0.00019999887078659895,
-      "loss": 46.0,
-      "step": 19795
-    },
-    {
-      "epoch": 1.5135424431829043,
-      "grad_norm": 0.000796723528765142,
-      "learning_rate": 0.00019999887067244788,
-      "loss": 46.0,
-      "step": 19796
-    },
-    {
-      "epoch": 1.513618900166294,
-      "grad_norm": 0.0008455587667413056,
-      "learning_rate": 0.000199998870558291,
-      "loss": 46.0,
-      "step": 19797
-    },
-    {
-      "epoch": 1.5136953571496836,
-      "grad_norm": 0.0009108296362683177,
-      "learning_rate": 0.00019999887044412832,
-      "loss": 46.0,
-      "step": 19798
-    },
-    {
-      "epoch": 1.5137718141330734,
-      "grad_norm": 0.0012390697374939919,
-      "learning_rate": 0.0001999988703299599,
-      "loss": 46.0,
-      "step": 19799
-    },
-    {
-      "epoch": 1.5138482711164631,
-      "grad_norm": 0.0021715802140533924,
-      "learning_rate": 0.00019999887021578575,
-      "loss": 46.0,
-      "step": 19800
-    },
-    {
-      "epoch": 1.5139247280998527,
-      "grad_norm": 0.0022135537583380938,
-      "learning_rate": 0.0001999988701016058,
-      "loss": 46.0,
-      "step": 19801
-    },
-    {
-      "epoch": 1.5140011850832424,
-      "grad_norm": 0.0004373548144940287,
-      "learning_rate": 0.00019999886998742005,
-      "loss": 46.0,
-      "step": 19802
-    },
-    {
-      "epoch": 1.5140776420666322,
-      "grad_norm": 0.0006472120294347405,
-      "learning_rate": 0.00019999886987322858,
-      "loss": 46.0,
-      "step": 19803
-    },
-    {
-      "epoch": 1.514154099050022,
-      "grad_norm": 0.0003856404800899327,
-      "learning_rate": 0.00019999886975903132,
-      "loss": 46.0,
-      "step": 19804
-    },
-    {
-      "epoch": 1.5142305560334117,
-      "grad_norm": 0.0009939444717019796,
-      "learning_rate": 0.00019999886964482827,
-      "loss": 46.0,
-      "step": 19805
-    },
-    {
-      "epoch": 1.5143070130168015,
-      "grad_norm": 0.0025241789408028126,
-      "learning_rate": 0.00019999886953061947,
-      "loss": 46.0,
-      "step": 19806
-    },
-    {
-      "epoch": 1.5143834700001912,
-      "grad_norm": 0.002797146327793598,
-      "learning_rate": 0.00019999886941640492,
-      "loss": 46.0,
-      "step": 19807
-    },
-    {
-      "epoch": 1.514459926983581,
-      "grad_norm": 0.002307654358446598,
-      "learning_rate": 0.00019999886930218458,
-      "loss": 46.0,
-      "step": 19808
-    },
-    {
-      "epoch": 1.5145363839669705,
-      "grad_norm": 0.016860200092196465,
-      "learning_rate": 0.00019999886918795848,
-      "loss": 46.0,
-      "step": 19809
-    },
-    {
-      "epoch": 1.5146128409503603,
-      "grad_norm": 0.004295938648283482,
-      "learning_rate": 0.0001999988690737266,
-      "loss": 46.0,
-      "step": 19810
-    },
-    {
-      "epoch": 1.51468929793375,
-      "grad_norm": 0.0009333654306828976,
-      "learning_rate": 0.00019999886895948895,
-      "loss": 46.0,
-      "step": 19811
-    },
-    {
-      "epoch": 1.5147657549171396,
-      "grad_norm": 0.0008975553791970015,
-      "learning_rate": 0.0001999988688452455,
-      "loss": 46.0,
-      "step": 19812
-    },
-    {
-      "epoch": 1.5148422119005294,
-      "grad_norm": 0.001588331419043243,
-      "learning_rate": 0.00019999886873099632,
-      "loss": 46.0,
-      "step": 19813
-    },
-    {
-      "epoch": 1.5149186688839191,
-      "grad_norm": 0.0015657334588468075,
-      "learning_rate": 0.00019999886861674136,
-      "loss": 46.0,
-      "step": 19814
-    },
-    {
-      "epoch": 1.514995125867309,
-      "grad_norm": 0.0013467337703332305,
-      "learning_rate": 0.00019999886850248066,
-      "loss": 46.0,
-      "step": 19815
-    },
-    {
-      "epoch": 1.5150715828506987,
-      "grad_norm": 0.0026081660762429237,
-      "learning_rate": 0.00019999886838821418,
-      "loss": 46.0,
-      "step": 19816
-    },
-    {
-      "epoch": 1.5151480398340884,
-      "grad_norm": 0.0019258181564509869,
-      "learning_rate": 0.0001999988682739419,
-      "loss": 46.0,
-      "step": 19817
-    },
-    {
-      "epoch": 1.5152244968174782,
-      "grad_norm": 0.005441577173769474,
-      "learning_rate": 0.0001999988681596639,
-      "loss": 46.0,
-      "step": 19818
-    },
-    {
-      "epoch": 1.515300953800868,
-      "grad_norm": 0.0010827280348166823,
-      "learning_rate": 0.00019999886804538007,
-      "loss": 46.0,
-      "step": 19819
-    },
-    {
-      "epoch": 1.5153774107842575,
-      "grad_norm": 0.0018788437591865659,
-      "learning_rate": 0.0001999988679310905,
-      "loss": 46.0,
-      "step": 19820
-    },
-    {
-      "epoch": 1.5154538677676472,
-      "grad_norm": 0.0027129354421049356,
-      "learning_rate": 0.00019999886781679518,
-      "loss": 46.0,
-      "step": 19821
-    },
-    {
-      "epoch": 1.515530324751037,
-      "grad_norm": 0.0003765798464883119,
-      "learning_rate": 0.00019999886770249406,
-      "loss": 46.0,
-      "step": 19822
-    },
-    {
-      "epoch": 1.5156067817344265,
-      "grad_norm": 0.004867858719080687,
-      "learning_rate": 0.00019999886758818717,
-      "loss": 46.0,
-      "step": 19823
-    },
-    {
-      "epoch": 1.5156832387178163,
-      "grad_norm": 0.00627858517691493,
-      "learning_rate": 0.00019999886747387453,
-      "loss": 46.0,
-      "step": 19824
-    },
-    {
-      "epoch": 1.515759695701206,
-      "grad_norm": 0.0004949085414409637,
-      "learning_rate": 0.00019999886735955612,
-      "loss": 46.0,
-      "step": 19825
-    },
-    {
-      "epoch": 1.5158361526845958,
-      "grad_norm": 0.00047921508667059243,
-      "learning_rate": 0.00019999886724523193,
-      "loss": 46.0,
-      "step": 19826
-    },
-    {
-      "epoch": 1.5159126096679856,
-      "grad_norm": 0.003148201620206237,
-      "learning_rate": 0.00019999886713090197,
-      "loss": 46.0,
-      "step": 19827
-    },
-    {
-      "epoch": 1.5159890666513753,
-      "grad_norm": 0.0010536847403272986,
-      "learning_rate": 0.00019999886701656624,
-      "loss": 46.0,
-      "step": 19828
-    },
-    {
-      "epoch": 1.516065523634765,
-      "grad_norm": 0.0011822348460555077,
-      "learning_rate": 0.00019999886690222476,
-      "loss": 46.0,
-      "step": 19829
-    },
-    {
-      "epoch": 1.5161419806181549,
-      "grad_norm": 0.005962466821074486,
-      "learning_rate": 0.00019999886678787748,
-      "loss": 46.0,
-      "step": 19830
-    },
-    {
-      "epoch": 1.5162184376015444,
-      "grad_norm": 0.0014695709105581045,
-      "learning_rate": 0.00019999886667352446,
-      "loss": 46.0,
-      "step": 19831
-    },
-    {
-      "epoch": 1.5162948945849342,
-      "grad_norm": 0.0014733609277755022,
-      "learning_rate": 0.00019999886655916563,
-      "loss": 46.0,
-      "step": 19832
-    },
-    {
-      "epoch": 1.516371351568324,
-      "grad_norm": 0.0011888614390045404,
-      "learning_rate": 0.00019999886644480106,
-      "loss": 46.0,
-      "step": 19833
-    },
-    {
-      "epoch": 1.5164478085517135,
-      "grad_norm": 0.0009545217035338283,
-      "learning_rate": 0.0001999988663304307,
-      "loss": 46.0,
-      "step": 19834
-    },
-    {
-      "epoch": 1.5165242655351032,
-      "grad_norm": 0.007244802080094814,
-      "learning_rate": 0.00019999886621605462,
-      "loss": 46.0,
-      "step": 19835
-    },
-    {
-      "epoch": 1.516600722518493,
-      "grad_norm": 0.0058004772290587425,
-      "learning_rate": 0.00019999886610167273,
-      "loss": 46.0,
-      "step": 19836
-    },
-    {
-      "epoch": 1.5166771795018827,
-      "grad_norm": 0.0005330651765689254,
-      "learning_rate": 0.0001999988659872851,
-      "loss": 46.0,
-      "step": 19837
-    },
-    {
-      "epoch": 1.5167536364852725,
-      "grad_norm": 0.0052987937815487385,
-      "learning_rate": 0.00019999886587289165,
-      "loss": 46.0,
-      "step": 19838
-    },
-    {
-      "epoch": 1.5168300934686623,
-      "grad_norm": 0.001190151902846992,
-      "learning_rate": 0.00019999886575849246,
-      "loss": 46.0,
-      "step": 19839
-    },
-    {
-      "epoch": 1.516906550452052,
-      "grad_norm": 0.0006867880583740771,
-      "learning_rate": 0.00019999886564408753,
-      "loss": 46.0,
-      "step": 19840
-    },
-    {
-      "epoch": 1.5169830074354418,
-      "grad_norm": 0.0002723844372667372,
-      "learning_rate": 0.00019999886552967677,
-      "loss": 46.0,
-      "step": 19841
-    },
-    {
-      "epoch": 1.5170594644188313,
-      "grad_norm": 0.005987521726638079,
-      "learning_rate": 0.0001999988654152603,
-      "loss": 46.0,
-      "step": 19842
-    },
-    {
-      "epoch": 1.517135921402221,
-      "grad_norm": 0.002486526034772396,
-      "learning_rate": 0.000199998865300838,
-      "loss": 46.0,
-      "step": 19843
-    },
-    {
-      "epoch": 1.5172123783856108,
-      "grad_norm": 0.0009461008012294769,
-      "learning_rate": 0.00019999886518641,
-      "loss": 46.0,
-      "step": 19844
-    },
-    {
-      "epoch": 1.5172888353690004,
-      "grad_norm": 0.003981179557740688,
-      "learning_rate": 0.0001999988650719762,
-      "loss": 46.0,
-      "step": 19845
-    },
-    {
-      "epoch": 1.5173652923523901,
-      "grad_norm": 0.0012268285499885678,
-      "learning_rate": 0.00019999886495753662,
-      "loss": 46.0,
-      "step": 19846
-    },
-    {
-      "epoch": 1.51744174933578,
-      "grad_norm": 0.0007840031757950783,
-      "learning_rate": 0.00019999886484309127,
-      "loss": 46.0,
-      "step": 19847
-    },
-    {
-      "epoch": 1.5175182063191697,
-      "grad_norm": 0.0009198356419801712,
-      "learning_rate": 0.00019999886472864015,
-      "loss": 46.0,
-      "step": 19848
-    },
-    {
-      "epoch": 1.5175946633025594,
-      "grad_norm": 0.0009282948449254036,
-      "learning_rate": 0.00019999886461418326,
-      "loss": 46.0,
-      "step": 19849
-    },
-    {
-      "epoch": 1.5176711202859492,
-      "grad_norm": 0.0022731777280569077,
-      "learning_rate": 0.0001999988644997206,
-      "loss": 46.0,
-      "step": 19850
-    },
-    {
-      "epoch": 1.517747577269339,
-      "grad_norm": 0.0013992219464853406,
-      "learning_rate": 0.00019999886438525218,
-      "loss": 46.0,
-      "step": 19851
-    },
-    {
-      "epoch": 1.5178240342527287,
-      "grad_norm": 0.0009773056954145432,
-      "learning_rate": 0.000199998864270778,
-      "loss": 46.0,
-      "step": 19852
-    },
-    {
-      "epoch": 1.5179004912361183,
-      "grad_norm": 0.0011099019320681691,
-      "learning_rate": 0.00019999886415629804,
-      "loss": 46.0,
-      "step": 19853
-    },
-    {
-      "epoch": 1.517976948219508,
-      "grad_norm": 0.0029436720069497824,
-      "learning_rate": 0.0001999988640418123,
-      "loss": 46.0,
-      "step": 19854
-    },
-    {
-      "epoch": 1.5180534052028976,
-      "grad_norm": 0.0017158595146611333,
-      "learning_rate": 0.0001999988639273208,
-      "loss": 46.0,
-      "step": 19855
-    },
-    {
-      "epoch": 1.5181298621862873,
-      "grad_norm": 0.007042300421744585,
-      "learning_rate": 0.00019999886381282352,
-      "loss": 46.0,
-      "step": 19856
-    },
-    {
-      "epoch": 1.518206319169677,
-      "grad_norm": 0.0006916631828062236,
-      "learning_rate": 0.0001999988636983205,
-      "loss": 46.0,
-      "step": 19857
-    },
-    {
-      "epoch": 1.5182827761530668,
-      "grad_norm": 0.0009145598160102963,
-      "learning_rate": 0.0001999988635838117,
-      "loss": 46.0,
-      "step": 19858
-    },
-    {
-      "epoch": 1.5183592331364566,
-      "grad_norm": 0.0010668050963431597,
-      "learning_rate": 0.00019999886346929713,
-      "loss": 46.0,
-      "step": 19859
-    },
-    {
-      "epoch": 1.5184356901198464,
-      "grad_norm": 0.002123479265719652,
-      "learning_rate": 0.00019999886335477678,
-      "loss": 46.0,
-      "step": 19860
-    },
-    {
-      "epoch": 1.5185121471032361,
-      "grad_norm": 0.0005922801210545003,
-      "learning_rate": 0.00019999886324025066,
-      "loss": 46.0,
-      "step": 19861
-    },
-    {
-      "epoch": 1.5185886040866259,
-      "grad_norm": 0.0012449015630409122,
-      "learning_rate": 0.00019999886312571877,
-      "loss": 46.0,
-      "step": 19862
-    },
-    {
-      "epoch": 1.5186650610700156,
-      "grad_norm": 0.0007242189603857696,
-      "learning_rate": 0.0001999988630111811,
-      "loss": 46.0,
-      "step": 19863
-    },
-    {
-      "epoch": 1.5187415180534052,
-      "grad_norm": 0.00037359242560341954,
-      "learning_rate": 0.0001999988628966377,
-      "loss": 46.0,
-      "step": 19864
-    },
-    {
-      "epoch": 1.518817975036795,
-      "grad_norm": 0.0015802932903170586,
-      "learning_rate": 0.00019999886278208848,
-      "loss": 46.0,
-      "step": 19865
-    },
-    {
-      "epoch": 1.5188944320201845,
-      "grad_norm": 0.0010674564400687814,
-      "learning_rate": 0.00019999886266753355,
-      "loss": 46.0,
-      "step": 19866
-    },
-    {
-      "epoch": 1.5189708890035742,
-      "grad_norm": 0.0022782045416533947,
-      "learning_rate": 0.0001999988625529728,
-      "loss": 46.0,
-      "step": 19867
-    },
-    {
-      "epoch": 1.519047345986964,
-      "grad_norm": 0.001007290557026863,
-      "learning_rate": 0.00019999886243840629,
-      "loss": 46.0,
-      "step": 19868
-    },
-    {
-      "epoch": 1.5191238029703538,
-      "grad_norm": 0.0009211021242663264,
-      "learning_rate": 0.00019999886232383403,
-      "loss": 46.0,
-      "step": 19869
-    },
-    {
-      "epoch": 1.5192002599537435,
-      "grad_norm": 0.0013268961338326335,
-      "learning_rate": 0.00019999886220925598,
-      "loss": 46.0,
-      "step": 19870
-    },
-    {
-      "epoch": 1.5192767169371333,
-      "grad_norm": 0.0005880377138964832,
-      "learning_rate": 0.00019999886209467218,
-      "loss": 46.0,
-      "step": 19871
-    },
-    {
-      "epoch": 1.519353173920523,
-      "grad_norm": 0.005156193859875202,
-      "learning_rate": 0.00019999886198008259,
-      "loss": 46.0,
-      "step": 19872
-    },
-    {
-      "epoch": 1.5194296309039128,
-      "grad_norm": 0.0006965100765228271,
-      "learning_rate": 0.00019999886186548724,
-      "loss": 46.0,
-      "step": 19873
-    },
-    {
-      "epoch": 1.5195060878873026,
-      "grad_norm": 0.000582943030167371,
-      "learning_rate": 0.00019999886175088612,
-      "loss": 46.0,
-      "step": 19874
-    },
-    {
-      "epoch": 1.519582544870692,
-      "grad_norm": 0.000550413562450558,
-      "learning_rate": 0.00019999886163627926,
-      "loss": 46.0,
-      "step": 19875
-    },
-    {
-      "epoch": 1.5196590018540819,
-      "grad_norm": 0.003940922673791647,
-      "learning_rate": 0.0001999988615216666,
-      "loss": 46.0,
-      "step": 19876
-    },
-    {
-      "epoch": 1.5197354588374714,
-      "grad_norm": 0.0006562622729688883,
-      "learning_rate": 0.00019999886140704815,
-      "loss": 46.0,
-      "step": 19877
-    },
-    {
-      "epoch": 1.5198119158208612,
-      "grad_norm": 0.0006262643728405237,
-      "learning_rate": 0.00019999886129242394,
-      "loss": 46.0,
-      "step": 19878
-    },
-    {
-      "epoch": 1.519888372804251,
-      "grad_norm": 0.0006632976001128554,
-      "learning_rate": 0.000199998861177794,
-      "loss": 46.0,
-      "step": 19879
-    },
-    {
-      "epoch": 1.5199648297876407,
-      "grad_norm": 0.0005016404902562499,
-      "learning_rate": 0.00019999886106315825,
-      "loss": 46.0,
-      "step": 19880
-    },
-    {
-      "epoch": 1.5200412867710305,
-      "grad_norm": 0.0022389309015125036,
-      "learning_rate": 0.00019999886094851678,
-      "loss": 46.0,
-      "step": 19881
-    },
-    {
-      "epoch": 1.5201177437544202,
-      "grad_norm": 0.0009701202507130802,
-      "learning_rate": 0.00019999886083386947,
-      "loss": 46.0,
-      "step": 19882
-    },
-    {
-      "epoch": 1.52019420073781,
-      "grad_norm": 0.0020565709564834833,
-      "learning_rate": 0.00019999886071921642,
-      "loss": 46.0,
-      "step": 19883
-    },
-    {
-      "epoch": 1.5202706577211997,
-      "grad_norm": 0.0010162091348320246,
-      "learning_rate": 0.00019999886060455765,
-      "loss": 46.0,
-      "step": 19884
-    },
-    {
-      "epoch": 1.5203471147045893,
-      "grad_norm": 0.0009761560359038413,
-      "learning_rate": 0.00019999886048989305,
-      "loss": 46.0,
-      "step": 19885
-    },
-    {
-      "epoch": 1.520423571687979,
-      "grad_norm": 0.000699535827152431,
-      "learning_rate": 0.0001999988603752227,
-      "loss": 46.0,
-      "step": 19886
-    },
-    {
-      "epoch": 1.5205000286713688,
-      "grad_norm": 0.0007347283535636961,
-      "learning_rate": 0.00019999886026054656,
-      "loss": 46.0,
-      "step": 19887
-    },
-    {
-      "epoch": 1.5205764856547583,
-      "grad_norm": 0.004425846505910158,
-      "learning_rate": 0.0001999988601458647,
-      "loss": 46.0,
-      "step": 19888
-    },
-    {
-      "epoch": 1.520652942638148,
-      "grad_norm": 0.004591597244143486,
-      "learning_rate": 0.00019999886003117703,
-      "loss": 46.0,
-      "step": 19889
-    },
-    {
-      "epoch": 1.5207293996215379,
-      "grad_norm": 0.002363857114687562,
-      "learning_rate": 0.00019999885991648362,
-      "loss": 46.0,
-      "step": 19890
-    },
-    {
-      "epoch": 1.5208058566049276,
-      "grad_norm": 0.002615045290440321,
-      "learning_rate": 0.0001999988598017844,
-      "loss": 46.0,
-      "step": 19891
-    },
-    {
-      "epoch": 1.5208823135883174,
-      "grad_norm": 0.0018921426963061094,
-      "learning_rate": 0.00019999885968707942,
-      "loss": 46.0,
-      "step": 19892
-    },
-    {
-      "epoch": 1.5209587705717071,
-      "grad_norm": 0.0009470746736042202,
-      "learning_rate": 0.0001999988595723687,
-      "loss": 46.0,
-      "step": 19893
-    },
-    {
-      "epoch": 1.521035227555097,
-      "grad_norm": 0.0013329192297533154,
-      "learning_rate": 0.0001999988594576522,
-      "loss": 46.0,
-      "step": 19894
-    },
-    {
-      "epoch": 1.5211116845384867,
-      "grad_norm": 0.0018824591534212232,
-      "learning_rate": 0.0001999988593429299,
-      "loss": 46.0,
-      "step": 19895
-    },
-    {
-      "epoch": 1.5211881415218762,
-      "grad_norm": 0.0019770858343690634,
-      "learning_rate": 0.00019999885922820189,
-      "loss": 46.0,
-      "step": 19896
-    },
-    {
-      "epoch": 1.521264598505266,
-      "grad_norm": 0.0010360676096752286,
-      "learning_rate": 0.00019999885911346806,
-      "loss": 46.0,
-      "step": 19897
-    },
-    {
-      "epoch": 1.5213410554886557,
-      "grad_norm": 0.0022221815306693316,
-      "learning_rate": 0.0001999988589987285,
-      "loss": 46.0,
-      "step": 19898
-    },
-    {
-      "epoch": 1.5214175124720453,
-      "grad_norm": 0.002833024365827441,
-      "learning_rate": 0.00019999885888398312,
-      "loss": 46.0,
-      "step": 19899
-    },
-    {
-      "epoch": 1.521493969455435,
-      "grad_norm": 0.0014749213587492704,
-      "learning_rate": 0.000199998858769232,
-      "loss": 46.0,
-      "step": 19900
-    },
-    {
-      "epoch": 1.5215704264388248,
-      "grad_norm": 0.0015110863605514169,
-      "learning_rate": 0.0001999988586544751,
-      "loss": 46.0,
-      "step": 19901
-    },
-    {
-      "epoch": 1.5216468834222145,
-      "grad_norm": 0.0012851685751229525,
-      "learning_rate": 0.00019999885853971245,
-      "loss": 46.0,
-      "step": 19902
-    },
-    {
-      "epoch": 1.5217233404056043,
-      "grad_norm": 0.0008480880642309785,
-      "learning_rate": 0.000199998858424944,
-      "loss": 46.0,
-      "step": 19903
-    },
-    {
-      "epoch": 1.521799797388994,
-      "grad_norm": 0.0013265982270240784,
-      "learning_rate": 0.00019999885831016982,
-      "loss": 46.0,
-      "step": 19904
-    },
-    {
-      "epoch": 1.5218762543723838,
-      "grad_norm": 0.0019406657665967941,
-      "learning_rate": 0.00019999885819538987,
-      "loss": 46.0,
-      "step": 19905
-    },
-    {
-      "epoch": 1.5219527113557736,
-      "grad_norm": 0.003281353274360299,
-      "learning_rate": 0.00019999885808060414,
-      "loss": 46.0,
-      "step": 19906
-    },
-    {
-      "epoch": 1.5220291683391631,
-      "grad_norm": 0.0010717188706621528,
-      "learning_rate": 0.0001999988579658126,
-      "loss": 46.0,
-      "step": 19907
-    },
-    {
-      "epoch": 1.522105625322553,
-      "grad_norm": 0.0017535180086269975,
-      "learning_rate": 0.00019999885785101533,
-      "loss": 46.0,
-      "step": 19908
-    },
-    {
-      "epoch": 1.5221820823059427,
-      "grad_norm": 0.0034215175546705723,
-      "learning_rate": 0.00019999885773621228,
-      "loss": 46.0,
-      "step": 19909
-    },
-    {
-      "epoch": 1.5222585392893322,
-      "grad_norm": 0.0012956680729985237,
-      "learning_rate": 0.00019999885762140348,
-      "loss": 46.0,
-      "step": 19910
-    },
-    {
-      "epoch": 1.522334996272722,
-      "grad_norm": 0.0026397292967885733,
-      "learning_rate": 0.00019999885750658888,
-      "loss": 46.0,
-      "step": 19911
-    },
-    {
-      "epoch": 1.5224114532561117,
-      "grad_norm": 0.0011051550973206758,
-      "learning_rate": 0.0001999988573917685,
-      "loss": 46.0,
-      "step": 19912
-    },
-    {
-      "epoch": 1.5224879102395015,
-      "grad_norm": 0.00885475892573595,
-      "learning_rate": 0.00019999885727694242,
-      "loss": 46.0,
-      "step": 19913
-    },
-    {
-      "epoch": 1.5225643672228912,
-      "grad_norm": 0.0034929709509015083,
-      "learning_rate": 0.0001999988571621105,
-      "loss": 46.0,
-      "step": 19914
-    },
-    {
-      "epoch": 1.522640824206281,
-      "grad_norm": 0.0005535755190066993,
-      "learning_rate": 0.00019999885704727284,
-      "loss": 46.0,
-      "step": 19915
-    },
-    {
-      "epoch": 1.5227172811896708,
-      "grad_norm": 0.0007332603563554585,
-      "learning_rate": 0.0001999988569324294,
-      "loss": 46.0,
-      "step": 19916
-    },
-    {
-      "epoch": 1.5227937381730605,
-      "grad_norm": 0.0023711537942290306,
-      "learning_rate": 0.00019999885681758022,
-      "loss": 46.0,
-      "step": 19917
-    },
-    {
-      "epoch": 1.52287019515645,
-      "grad_norm": 0.0012392589123919606,
-      "learning_rate": 0.00019999885670272526,
-      "loss": 46.0,
-      "step": 19918
-    },
-    {
-      "epoch": 1.5229466521398398,
-      "grad_norm": 0.0011114421067759395,
-      "learning_rate": 0.0001999988565878645,
-      "loss": 46.0,
-      "step": 19919
-    },
-    {
-      "epoch": 1.5230231091232296,
-      "grad_norm": 0.0013310209615156054,
-      "learning_rate": 0.000199998856472998,
-      "loss": 46.0,
-      "step": 19920
-    },
-    {
-      "epoch": 1.5230995661066191,
-      "grad_norm": 0.005071427207440138,
-      "learning_rate": 0.00019999885635812572,
-      "loss": 46.0,
-      "step": 19921
-    },
-    {
-      "epoch": 1.5231760230900089,
-      "grad_norm": 0.0017212532693520188,
-      "learning_rate": 0.00019999885624324767,
-      "loss": 46.0,
-      "step": 19922
-    },
-    {
-      "epoch": 1.5232524800733986,
-      "grad_norm": 0.0013851674739271402,
-      "learning_rate": 0.00019999885612836385,
-      "loss": 46.0,
-      "step": 19923
-    },
-    {
-      "epoch": 1.5233289370567884,
-      "grad_norm": 0.001247505540959537,
-      "learning_rate": 0.00019999885601347425,
-      "loss": 46.0,
-      "step": 19924
-    },
-    {
-      "epoch": 1.5234053940401782,
-      "grad_norm": 0.0015065122861415148,
-      "learning_rate": 0.0001999988558985789,
-      "loss": 46.0,
-      "step": 19925
-    },
-    {
-      "epoch": 1.523481851023568,
-      "grad_norm": 0.000630170397926122,
-      "learning_rate": 0.0001999988557836778,
-      "loss": 46.0,
-      "step": 19926
-    },
-    {
-      "epoch": 1.5235583080069577,
-      "grad_norm": 0.0005325032398104668,
-      "learning_rate": 0.00019999885566877087,
-      "loss": 46.0,
-      "step": 19927
-    },
-    {
-      "epoch": 1.5236347649903474,
-      "grad_norm": 0.0009795564692467451,
-      "learning_rate": 0.0001999988555538582,
-      "loss": 46.0,
-      "step": 19928
-    },
-    {
-      "epoch": 1.523711221973737,
-      "grad_norm": 0.0048400405794382095,
-      "learning_rate": 0.00019999885543893977,
-      "loss": 46.0,
-      "step": 19929
-    },
-    {
-      "epoch": 1.5237876789571267,
-      "grad_norm": 0.0008271417464129627,
-      "learning_rate": 0.00019999885532401556,
-      "loss": 46.0,
-      "step": 19930
-    },
-    {
-      "epoch": 1.5238641359405165,
-      "grad_norm": 0.002531171077862382,
-      "learning_rate": 0.0001999988552090856,
-      "loss": 46.0,
-      "step": 19931
-    },
-    {
-      "epoch": 1.523940592923906,
-      "grad_norm": 0.002875386504456401,
-      "learning_rate": 0.00019999885509414987,
-      "loss": 46.0,
-      "step": 19932
-    },
-    {
-      "epoch": 1.5240170499072958,
-      "grad_norm": 0.001793494215235114,
-      "learning_rate": 0.00019999885497920837,
-      "loss": 46.0,
-      "step": 19933
-    },
-    {
-      "epoch": 1.5240935068906856,
-      "grad_norm": 0.0012160597834736109,
-      "learning_rate": 0.00019999885486426106,
-      "loss": 46.0,
-      "step": 19934
-    },
-    {
-      "epoch": 1.5241699638740753,
-      "grad_norm": 0.0020554724615067244,
-      "learning_rate": 0.000199998854749308,
-      "loss": 46.0,
-      "step": 19935
-    },
-    {
-      "epoch": 1.524246420857465,
-      "grad_norm": 0.0023764052893966436,
-      "learning_rate": 0.0001999988546343492,
-      "loss": 46.0,
-      "step": 19936
-    },
-    {
-      "epoch": 1.5243228778408549,
-      "grad_norm": 0.0009446701151318848,
-      "learning_rate": 0.0001999988545193846,
-      "loss": 46.0,
-      "step": 19937
-    },
-    {
-      "epoch": 1.5243993348242446,
-      "grad_norm": 0.000792208593338728,
-      "learning_rate": 0.00019999885440441425,
-      "loss": 46.0,
-      "step": 19938
-    },
-    {
-      "epoch": 1.5244757918076344,
-      "grad_norm": 0.0006929371156729758,
-      "learning_rate": 0.00019999885428943813,
-      "loss": 46.0,
-      "step": 19939
-    },
-    {
-      "epoch": 1.524552248791024,
-      "grad_norm": 0.0008702266495674849,
-      "learning_rate": 0.00019999885417445624,
-      "loss": 46.0,
-      "step": 19940
-    },
-    {
-      "epoch": 1.5246287057744137,
-      "grad_norm": 0.001397086656652391,
-      "learning_rate": 0.00019999885405946858,
-      "loss": 46.0,
-      "step": 19941
-    },
-    {
-      "epoch": 1.5247051627578034,
-      "grad_norm": 0.013896516524255276,
-      "learning_rate": 0.00019999885394447514,
-      "loss": 46.0,
-      "step": 19942
-    },
-    {
-      "epoch": 1.524781619741193,
-      "grad_norm": 0.0003093729610554874,
-      "learning_rate": 0.00019999885382947593,
-      "loss": 46.0,
-      "step": 19943
-    },
-    {
-      "epoch": 1.5248580767245827,
-      "grad_norm": 0.0013066452229395509,
-      "learning_rate": 0.00019999885371447095,
-      "loss": 46.0,
-      "step": 19944
-    },
-    {
-      "epoch": 1.5249345337079725,
-      "grad_norm": 0.0005022439872846007,
-      "learning_rate": 0.0001999988535994602,
-      "loss": 46.0,
-      "step": 19945
-    },
-    {
-      "epoch": 1.5250109906913623,
-      "grad_norm": 0.0031297921668738127,
-      "learning_rate": 0.0001999988534844437,
-      "loss": 46.0,
-      "step": 19946
-    },
-    {
-      "epoch": 1.525087447674752,
-      "grad_norm": 0.0038235660176724195,
-      "learning_rate": 0.00019999885336942138,
-      "loss": 46.0,
-      "step": 19947
-    },
-    {
-      "epoch": 1.5251639046581418,
-      "grad_norm": 0.0017294284189119935,
-      "learning_rate": 0.00019999885325439336,
-      "loss": 46.0,
-      "step": 19948
-    },
-    {
-      "epoch": 1.5252403616415315,
-      "grad_norm": 0.0014719052705913782,
-      "learning_rate": 0.0001999988531393595,
-      "loss": 46.0,
-      "step": 19949
-    },
-    {
-      "epoch": 1.5253168186249213,
-      "grad_norm": 0.0021882113069295883,
-      "learning_rate": 0.00019999885302431994,
-      "loss": 46.0,
-      "step": 19950
-    },
-    {
-      "epoch": 1.5253932756083108,
-      "grad_norm": 0.001612881664186716,
-      "learning_rate": 0.00019999885290927457,
-      "loss": 46.0,
-      "step": 19951
-    },
-    {
-      "epoch": 1.5254697325917006,
-      "grad_norm": 0.0007277579279616475,
-      "learning_rate": 0.00019999885279422345,
-      "loss": 46.0,
-      "step": 19952
-    },
-    {
-      "epoch": 1.5255461895750904,
-      "grad_norm": 0.0017669732915237546,
-      "learning_rate": 0.00019999885267916654,
-      "loss": 46.0,
-      "step": 19953
-    },
-    {
-      "epoch": 1.52562264655848,
-      "grad_norm": 0.0012967296643182635,
-      "learning_rate": 0.0001999988525641039,
-      "loss": 46.0,
-      "step": 19954
-    },
-    {
-      "epoch": 1.5256991035418697,
-      "grad_norm": 0.010033457539975643,
-      "learning_rate": 0.00019999885244903544,
-      "loss": 46.0,
-      "step": 19955
-    },
-    {
-      "epoch": 1.5257755605252594,
-      "grad_norm": 0.0009488180512562394,
-      "learning_rate": 0.00019999885233396123,
-      "loss": 46.0,
-      "step": 19956
-    },
-    {
-      "epoch": 1.5258520175086492,
-      "grad_norm": 0.0005836423952132463,
-      "learning_rate": 0.00019999885221888124,
-      "loss": 46.0,
-      "step": 19957
-    },
-    {
-      "epoch": 1.525928474492039,
-      "grad_norm": 0.0040775868110358715,
-      "learning_rate": 0.0001999988521037955,
-      "loss": 46.0,
-      "step": 19958
-    },
-    {
-      "epoch": 1.5260049314754287,
-      "grad_norm": 0.012530580163002014,
-      "learning_rate": 0.00019999885198870398,
-      "loss": 46.0,
-      "step": 19959
-    },
-    {
-      "epoch": 1.5260813884588185,
-      "grad_norm": 0.0007836756412871182,
-      "learning_rate": 0.0001999988518736067,
-      "loss": 46.0,
-      "step": 19960
-    },
-    {
-      "epoch": 1.5261578454422082,
-      "grad_norm": 0.0009908212814480066,
-      "learning_rate": 0.00019999885175850363,
-      "loss": 46.0,
-      "step": 19961
-    },
-    {
-      "epoch": 1.5262343024255978,
-      "grad_norm": 0.002022718545049429,
-      "learning_rate": 0.00019999885164339484,
-      "loss": 46.0,
-      "step": 19962
-    },
-    {
-      "epoch": 1.5263107594089875,
-      "grad_norm": 0.0010559409856796265,
-      "learning_rate": 0.00019999885152828024,
-      "loss": 46.0,
-      "step": 19963
-    },
-    {
-      "epoch": 1.5263872163923773,
-      "grad_norm": 0.0021514487452805042,
-      "learning_rate": 0.00019999885141315987,
-      "loss": 46.0,
-      "step": 19964
-    },
-    {
-      "epoch": 1.5264636733757668,
-      "grad_norm": 0.0011832613963633776,
-      "learning_rate": 0.00019999885129803373,
-      "loss": 46.0,
-      "step": 19965
-    },
-    {
-      "epoch": 1.5265401303591566,
-      "grad_norm": 0.0013538129860535264,
-      "learning_rate": 0.00019999885118290184,
-      "loss": 46.0,
-      "step": 19966
-    },
-    {
-      "epoch": 1.5266165873425463,
-      "grad_norm": 0.002653195522725582,
-      "learning_rate": 0.00019999885106776417,
-      "loss": 46.0,
-      "step": 19967
-    },
-    {
-      "epoch": 1.526693044325936,
-      "grad_norm": 0.000602905114647001,
-      "learning_rate": 0.0001999988509526207,
-      "loss": 46.0,
-      "step": 19968
-    },
-    {
-      "epoch": 1.5267695013093259,
-      "grad_norm": 0.0027083309832960367,
-      "learning_rate": 0.0001999988508374715,
-      "loss": 46.0,
-      "step": 19969
-    },
-    {
-      "epoch": 1.5268459582927156,
-      "grad_norm": 0.0029829933773726225,
-      "learning_rate": 0.00019999885072231654,
-      "loss": 46.0,
-      "step": 19970
-    },
-    {
-      "epoch": 1.5269224152761054,
-      "grad_norm": 0.0008750649867579341,
-      "learning_rate": 0.00019999885060715576,
-      "loss": 46.0,
-      "step": 19971
-    },
-    {
-      "epoch": 1.5269988722594952,
-      "grad_norm": 0.0007617862429469824,
-      "learning_rate": 0.00019999885049198929,
-      "loss": 46.0,
-      "step": 19972
-    },
-    {
-      "epoch": 1.5270753292428847,
-      "grad_norm": 0.0010432065464556217,
-      "learning_rate": 0.00019999885037681698,
-      "loss": 46.0,
-      "step": 19973
-    },
-    {
-      "epoch": 1.5271517862262745,
-      "grad_norm": 0.0012251805746927857,
-      "learning_rate": 0.0001999988502616389,
-      "loss": 46.0,
-      "step": 19974
-    },
-    {
-      "epoch": 1.5272282432096642,
-      "grad_norm": 0.0009974499698728323,
-      "learning_rate": 0.00019999885014645508,
-      "loss": 46.0,
-      "step": 19975
-    },
-    {
-      "epoch": 1.5273047001930538,
-      "grad_norm": 0.00042729786946438253,
-      "learning_rate": 0.00019999885003126551,
-      "loss": 46.0,
-      "step": 19976
-    },
-    {
-      "epoch": 1.5273811571764435,
-      "grad_norm": 0.0006383644649758935,
-      "learning_rate": 0.00019999884991607014,
-      "loss": 46.0,
-      "step": 19977
-    },
-    {
-      "epoch": 1.5274576141598333,
-      "grad_norm": 0.0004573434707708657,
-      "learning_rate": 0.000199998849800869,
-      "loss": 46.0,
-      "step": 19978
-    },
-    {
-      "epoch": 1.527534071143223,
-      "grad_norm": 0.0008911238401196897,
-      "learning_rate": 0.00019999884968566209,
-      "loss": 46.0,
-      "step": 19979
-    },
-    {
-      "epoch": 1.5276105281266128,
-      "grad_norm": 0.001186067471280694,
-      "learning_rate": 0.0001999988495704494,
-      "loss": 46.0,
-      "step": 19980
-    },
-    {
-      "epoch": 1.5276869851100026,
-      "grad_norm": 0.001374711631797254,
-      "learning_rate": 0.000199998849455231,
-      "loss": 46.0,
-      "step": 19981
-    },
-    {
-      "epoch": 1.5277634420933923,
-      "grad_norm": 0.0020042192190885544,
-      "learning_rate": 0.00019999884934000678,
-      "loss": 46.0,
-      "step": 19982
-    },
-    {
-      "epoch": 1.527839899076782,
-      "grad_norm": 0.0006101718172430992,
-      "learning_rate": 0.0001999988492247768,
-      "loss": 46.0,
-      "step": 19983
-    },
-    {
-      "epoch": 1.5279163560601716,
-      "grad_norm": 0.0026644417084753513,
-      "learning_rate": 0.00019999884910954104,
-      "loss": 46.0,
-      "step": 19984
-    },
-    {
-      "epoch": 1.5279928130435614,
-      "grad_norm": 0.0017771368147805333,
-      "learning_rate": 0.00019999884899429954,
-      "loss": 46.0,
-      "step": 19985
-    },
-    {
-      "epoch": 1.528069270026951,
-      "grad_norm": 0.0013287786860018969,
-      "learning_rate": 0.00019999884887905223,
-      "loss": 46.0,
-      "step": 19986
-    },
-    {
-      "epoch": 1.5281457270103407,
-      "grad_norm": 0.002531816717237234,
-      "learning_rate": 0.00019999884876379916,
-      "loss": 46.0,
-      "step": 19987
-    },
-    {
-      "epoch": 1.5282221839937304,
-      "grad_norm": 0.00046768636093474925,
-      "learning_rate": 0.00019999884864854034,
-      "loss": 46.0,
-      "step": 19988
-    },
-    {
-      "epoch": 1.5282986409771202,
-      "grad_norm": 0.0010499063646420836,
-      "learning_rate": 0.00019999884853327574,
-      "loss": 46.0,
-      "step": 19989
-    },
-    {
-      "epoch": 1.52837509796051,
-      "grad_norm": 0.0012143907370045781,
-      "learning_rate": 0.00019999884841800537,
-      "loss": 46.0,
-      "step": 19990
-    },
-    {
-      "epoch": 1.5284515549438997,
-      "grad_norm": 0.001626562443561852,
-      "learning_rate": 0.00019999884830272923,
-      "loss": 46.0,
-      "step": 19991
-    },
-    {
-      "epoch": 1.5285280119272895,
-      "grad_norm": 0.004600904881954193,
-      "learning_rate": 0.00019999884818744734,
-      "loss": 46.0,
-      "step": 19992
-    },
-    {
-      "epoch": 1.5286044689106792,
-      "grad_norm": 0.012294813990592957,
-      "learning_rate": 0.00019999884807215965,
-      "loss": 46.0,
-      "step": 19993
-    },
-    {
-      "epoch": 1.528680925894069,
-      "grad_norm": 0.0006995838484726846,
-      "learning_rate": 0.00019999884795686621,
-      "loss": 46.0,
-      "step": 19994
-    },
-    {
-      "epoch": 1.5287573828774585,
-      "grad_norm": 0.007329861633479595,
-      "learning_rate": 0.000199998847841567,
-      "loss": 46.0,
-      "step": 19995
-    },
-    {
-      "epoch": 1.5288338398608483,
-      "grad_norm": 0.000449772720457986,
-      "learning_rate": 0.00019999884772626202,
-      "loss": 46.0,
-      "step": 19996
-    },
-    {
-      "epoch": 1.5289102968442378,
-      "grad_norm": 0.0013943086378276348,
-      "learning_rate": 0.00019999884761095127,
-      "loss": 46.0,
-      "step": 19997
-    },
-    {
-      "epoch": 1.5289867538276276,
-      "grad_norm": 0.0011242623440921307,
-      "learning_rate": 0.00019999884749563474,
-      "loss": 46.0,
-      "step": 19998
-    },
-    {
-      "epoch": 1.5290632108110174,
-      "grad_norm": 0.0015805922448635101,
-      "learning_rate": 0.00019999884738031244,
-      "loss": 46.0,
-      "step": 19999
-    },
-    {
-      "epoch": 1.5291396677944071,
-      "grad_norm": 0.0013938507763668895,
-      "learning_rate": 0.0001999988472649844,
-      "loss": 46.0,
-      "step": 20000
-    },
-    {
-      "epoch": 1.529216124777797,
-      "grad_norm": 0.0010079655330628157,
-      "learning_rate": 0.00019999884714965056,
-      "loss": 46.0,
-      "step": 20001
-    },
-    {
-      "epoch": 1.5292925817611867,
-      "grad_norm": 0.0005260030156932771,
-      "learning_rate": 0.00019999884703431094,
-      "loss": 46.0,
-      "step": 20002
-    },
-    {
-      "epoch": 1.5293690387445764,
-      "grad_norm": 0.002525837393477559,
-      "learning_rate": 0.00019999884691896557,
-      "loss": 46.0,
-      "step": 20003
-    },
-    {
-      "epoch": 1.5294454957279662,
-      "grad_norm": 0.0007301702862605453,
-      "learning_rate": 0.00019999884680361446,
-      "loss": 46.0,
-      "step": 20004
-    },
-    {
-      "epoch": 1.529521952711356,
-      "grad_norm": 0.003996156621724367,
-      "learning_rate": 0.00019999884668825754,
-      "loss": 46.0,
-      "step": 20005
-    },
-    {
-      "epoch": 1.5295984096947455,
-      "grad_norm": 0.0011266169603914022,
-      "learning_rate": 0.00019999884657289488,
-      "loss": 46.0,
-      "step": 20006
-    },
-    {
-      "epoch": 1.5296748666781352,
-      "grad_norm": 0.0011759857879951596,
-      "learning_rate": 0.00019999884645752642,
-      "loss": 46.0,
-      "step": 20007
-    },
-    {
-      "epoch": 1.5297513236615248,
-      "grad_norm": 0.0010739343706518412,
-      "learning_rate": 0.0001999988463421522,
-      "loss": 46.0,
-      "step": 20008
-    },
-    {
-      "epoch": 1.5298277806449145,
-      "grad_norm": 0.0019064394291490316,
-      "learning_rate": 0.0001999988462267722,
-      "loss": 46.0,
-      "step": 20009
-    },
-    {
-      "epoch": 1.5299042376283043,
-      "grad_norm": 0.0009081195457838476,
-      "learning_rate": 0.00019999884611138647,
-      "loss": 46.0,
-      "step": 20010
-    },
-    {
-      "epoch": 1.529980694611694,
-      "grad_norm": 0.0017312541604042053,
-      "learning_rate": 0.00019999884599599494,
-      "loss": 46.0,
-      "step": 20011
-    },
-    {
-      "epoch": 1.5300571515950838,
-      "grad_norm": 0.000858985586091876,
-      "learning_rate": 0.00019999884588059764,
-      "loss": 46.0,
-      "step": 20012
-    },
-    {
-      "epoch": 1.5301336085784736,
-      "grad_norm": 0.0005749059491790831,
-      "learning_rate": 0.0001999988457651946,
-      "loss": 46.0,
-      "step": 20013
-    },
-    {
-      "epoch": 1.5302100655618633,
-      "grad_norm": 0.0017046816647052765,
-      "learning_rate": 0.00019999884564978574,
-      "loss": 46.0,
-      "step": 20014
-    },
-    {
-      "epoch": 1.530286522545253,
-      "grad_norm": 0.0017081631813198328,
-      "learning_rate": 0.00019999884553437115,
-      "loss": 46.0,
-      "step": 20015
-    },
-    {
-      "epoch": 1.5303629795286429,
-      "grad_norm": 0.000795937841758132,
-      "learning_rate": 0.00019999884541895078,
-      "loss": 46.0,
-      "step": 20016
-    },
-    {
-      "epoch": 1.5304394365120324,
-      "grad_norm": 0.0007827605004422367,
-      "learning_rate": 0.00019999884530352464,
-      "loss": 46.0,
-      "step": 20017
-    },
-    {
-      "epoch": 1.5305158934954222,
-      "grad_norm": 0.0008895203936845064,
-      "learning_rate": 0.00019999884518809272,
-      "loss": 46.0,
-      "step": 20018
-    },
-    {
-      "epoch": 1.5305923504788117,
-      "grad_norm": 0.0006428833003155887,
-      "learning_rate": 0.00019999884507265506,
-      "loss": 46.0,
-      "step": 20019
-    },
-    {
-      "epoch": 1.5306688074622015,
-      "grad_norm": 0.0007569598965346813,
-      "learning_rate": 0.0001999988449572116,
-      "loss": 46.0,
-      "step": 20020
-    },
-    {
-      "epoch": 1.5307452644455912,
-      "grad_norm": 0.0015019163256511092,
-      "learning_rate": 0.00019999884484176236,
-      "loss": 46.0,
-      "step": 20021
-    },
-    {
-      "epoch": 1.530821721428981,
-      "grad_norm": 0.0016402594046667218,
-      "learning_rate": 0.00019999884472630738,
-      "loss": 46.0,
-      "step": 20022
-    },
-    {
-      "epoch": 1.5308981784123707,
-      "grad_norm": 0.0005509136826731265,
-      "learning_rate": 0.00019999884461084663,
-      "loss": 46.0,
-      "step": 20023
-    },
-    {
-      "epoch": 1.5309746353957605,
-      "grad_norm": 0.005408469121903181,
-      "learning_rate": 0.0001999988444953801,
-      "loss": 46.0,
-      "step": 20024
-    },
-    {
-      "epoch": 1.5310510923791503,
-      "grad_norm": 0.0016071624122560024,
-      "learning_rate": 0.00019999884437990782,
-      "loss": 46.0,
-      "step": 20025
-    },
-    {
-      "epoch": 1.53112754936254,
-      "grad_norm": 0.002094559371471405,
-      "learning_rate": 0.00019999884426442975,
-      "loss": 46.0,
-      "step": 20026
-    },
-    {
-      "epoch": 1.5312040063459296,
-      "grad_norm": 0.008855417370796204,
-      "learning_rate": 0.00019999884414894593,
-      "loss": 46.0,
-      "step": 20027
-    },
-    {
-      "epoch": 1.5312804633293193,
-      "grad_norm": 0.0005544427549466491,
-      "learning_rate": 0.00019999884403345633,
-      "loss": 46.0,
-      "step": 20028
-    },
-    {
-      "epoch": 1.531356920312709,
-      "grad_norm": 0.0009228939888998866,
-      "learning_rate": 0.00019999884391796094,
-      "loss": 46.0,
-      "step": 20029
-    },
-    {
-      "epoch": 1.5314333772960986,
-      "grad_norm": 0.0007227399037219584,
-      "learning_rate": 0.00019999884380245982,
-      "loss": 46.0,
-      "step": 20030
-    },
-    {
-      "epoch": 1.5315098342794884,
-      "grad_norm": 0.001131095690652728,
-      "learning_rate": 0.00019999884368695288,
-      "loss": 46.0,
-      "step": 20031
-    },
-    {
-      "epoch": 1.5315862912628782,
-      "grad_norm": 0.0007045971578918397,
-      "learning_rate": 0.00019999884357144022,
-      "loss": 46.0,
-      "step": 20032
-    },
-    {
-      "epoch": 1.531662748246268,
-      "grad_norm": 0.0007540889200754464,
-      "learning_rate": 0.00019999884345592176,
-      "loss": 46.0,
-      "step": 20033
-    },
-    {
-      "epoch": 1.5317392052296577,
-      "grad_norm": 0.005252319388091564,
-      "learning_rate": 0.00019999884334039752,
-      "loss": 46.0,
-      "step": 20034
-    },
-    {
-      "epoch": 1.5318156622130474,
-      "grad_norm": 0.0013515064492821693,
-      "learning_rate": 0.00019999884322486754,
-      "loss": 46.0,
-      "step": 20035
-    },
-    {
-      "epoch": 1.5318921191964372,
-      "grad_norm": 0.0013417586451396346,
-      "learning_rate": 0.00019999884310933179,
-      "loss": 46.0,
-      "step": 20036
-    },
-    {
-      "epoch": 1.531968576179827,
-      "grad_norm": 0.0006675160257145762,
-      "learning_rate": 0.00019999884299379026,
-      "loss": 46.0,
-      "step": 20037
-    },
-    {
-      "epoch": 1.5320450331632165,
-      "grad_norm": 0.0012373344507068396,
-      "learning_rate": 0.00019999884287824298,
-      "loss": 46.0,
-      "step": 20038
-    },
-    {
-      "epoch": 1.5321214901466063,
-      "grad_norm": 0.002380277495831251,
-      "learning_rate": 0.0001999988427626899,
-      "loss": 46.0,
-      "step": 20039
-    },
-    {
-      "epoch": 1.532197947129996,
-      "grad_norm": 0.005655379965901375,
-      "learning_rate": 0.00019999884264713106,
-      "loss": 46.0,
-      "step": 20040
-    },
-    {
-      "epoch": 1.5322744041133856,
-      "grad_norm": 0.0016430789837613702,
-      "learning_rate": 0.00019999884253156646,
-      "loss": 46.0,
-      "step": 20041
-    },
-    {
-      "epoch": 1.5323508610967753,
-      "grad_norm": 0.0007284291787073016,
-      "learning_rate": 0.00019999884241599607,
-      "loss": 46.0,
-      "step": 20042
-    },
-    {
-      "epoch": 1.532427318080165,
-      "grad_norm": 0.0016675929073244333,
-      "learning_rate": 0.00019999884230041993,
-      "loss": 46.0,
-      "step": 20043
-    },
-    {
-      "epoch": 1.5325037750635548,
-      "grad_norm": 0.0006671563605777919,
-      "learning_rate": 0.00019999884218483801,
-      "loss": 46.0,
-      "step": 20044
-    },
-    {
-      "epoch": 1.5325802320469446,
-      "grad_norm": 0.0015132996486499906,
-      "learning_rate": 0.00019999884206925033,
-      "loss": 46.0,
-      "step": 20045
-    },
-    {
-      "epoch": 1.5326566890303344,
-      "grad_norm": 0.0006749456515535712,
-      "learning_rate": 0.0001999988419536569,
-      "loss": 46.0,
-      "step": 20046
-    },
-    {
-      "epoch": 1.5327331460137241,
-      "grad_norm": 0.00534209655597806,
-      "learning_rate": 0.00019999884183805766,
-      "loss": 46.0,
-      "step": 20047
-    },
-    {
-      "epoch": 1.5328096029971139,
-      "grad_norm": 0.0009820647537708282,
-      "learning_rate": 0.00019999884172245268,
-      "loss": 46.0,
-      "step": 20048
-    },
-    {
-      "epoch": 1.5328860599805034,
-      "grad_norm": 0.0027369658928364515,
-      "learning_rate": 0.0001999988416068419,
-      "loss": 46.0,
-      "step": 20049
-    },
-    {
-      "epoch": 1.5329625169638932,
-      "grad_norm": 0.0008373742457479239,
-      "learning_rate": 0.00019999884149122537,
-      "loss": 46.0,
-      "step": 20050
-    },
-    {
-      "epoch": 1.533038973947283,
-      "grad_norm": 0.004651294555515051,
-      "learning_rate": 0.00019999884137560306,
-      "loss": 46.0,
-      "step": 20051
-    },
-    {
-      "epoch": 1.5331154309306725,
-      "grad_norm": 0.004516802728176117,
-      "learning_rate": 0.00019999884125997502,
-      "loss": 46.0,
-      "step": 20052
-    },
-    {
-      "epoch": 1.5331918879140622,
-      "grad_norm": 0.0005958874826319516,
-      "learning_rate": 0.00019999884114434114,
-      "loss": 46.0,
-      "step": 20053
-    },
-    {
-      "epoch": 1.533268344897452,
-      "grad_norm": 0.0006082784384489059,
-      "learning_rate": 0.00019999884102870157,
-      "loss": 46.0,
-      "step": 20054
-    },
-    {
-      "epoch": 1.5333448018808418,
-      "grad_norm": 0.0018846383318305016,
-      "learning_rate": 0.00019999884091305618,
-      "loss": 46.0,
-      "step": 20055
-    },
-    {
-      "epoch": 1.5334212588642315,
-      "grad_norm": 0.0006319937529042363,
-      "learning_rate": 0.00019999884079740504,
-      "loss": 46.0,
-      "step": 20056
-    },
-    {
-      "epoch": 1.5334977158476213,
-      "grad_norm": 0.0004855721490457654,
-      "learning_rate": 0.00019999884068174812,
-      "loss": 46.0,
-      "step": 20057
-    },
-    {
-      "epoch": 1.533574172831011,
-      "grad_norm": 0.0020239066798239946,
-      "learning_rate": 0.00019999884056608544,
-      "loss": 46.0,
-      "step": 20058
-    },
-    {
-      "epoch": 1.5336506298144008,
-      "grad_norm": 0.003815114265307784,
-      "learning_rate": 0.000199998840450417,
-      "loss": 46.0,
-      "step": 20059
-    },
-    {
-      "epoch": 1.5337270867977904,
-      "grad_norm": 0.004444007761776447,
-      "learning_rate": 0.00019999884033474277,
-      "loss": 46.0,
-      "step": 20060
-    },
-    {
-      "epoch": 1.5338035437811801,
-      "grad_norm": 0.0005429207230918109,
-      "learning_rate": 0.00019999884021906276,
-      "loss": 46.0,
-      "step": 20061
-    },
-    {
-      "epoch": 1.5338800007645699,
-      "grad_norm": 0.000800707726739347,
-      "learning_rate": 0.000199998840103377,
-      "loss": 46.0,
-      "step": 20062
-    },
-    {
-      "epoch": 1.5339564577479594,
-      "grad_norm": 0.0013812314718961716,
-      "learning_rate": 0.00019999883998768548,
-      "loss": 46.0,
-      "step": 20063
-    },
-    {
-      "epoch": 1.5340329147313492,
-      "grad_norm": 0.0008911380427889526,
-      "learning_rate": 0.00019999883987198818,
-      "loss": 46.0,
-      "step": 20064
-    },
-    {
-      "epoch": 1.534109371714739,
-      "grad_norm": 0.0007954948232509196,
-      "learning_rate": 0.00019999883975628508,
-      "loss": 46.0,
-      "step": 20065
-    },
-    {
-      "epoch": 1.5341858286981287,
-      "grad_norm": 0.0008387573761865497,
-      "learning_rate": 0.00019999883964057626,
-      "loss": 46.0,
-      "step": 20066
-    },
-    {
-      "epoch": 1.5342622856815185,
-      "grad_norm": 0.0012238919734954834,
-      "learning_rate": 0.00019999883952486163,
-      "loss": 46.0,
-      "step": 20067
-    },
-    {
-      "epoch": 1.5343387426649082,
-      "grad_norm": 0.0007302372832782567,
-      "learning_rate": 0.00019999883940914127,
-      "loss": 46.0,
-      "step": 20068
-    },
-    {
-      "epoch": 1.534415199648298,
-      "grad_norm": 0.0011135886888951063,
-      "learning_rate": 0.00019999883929341513,
-      "loss": 46.0,
-      "step": 20069
-    },
-    {
-      "epoch": 1.5344916566316877,
-      "grad_norm": 0.0008723319042474031,
-      "learning_rate": 0.0001999988391776832,
-      "loss": 46.0,
-      "step": 20070
-    },
-    {
-      "epoch": 1.5345681136150773,
-      "grad_norm": 0.0005100786220282316,
-      "learning_rate": 0.00019999883906194553,
-      "loss": 46.0,
-      "step": 20071
-    },
-    {
-      "epoch": 1.534644570598467,
-      "grad_norm": 0.0008733382564969361,
-      "learning_rate": 0.00019999883894620206,
-      "loss": 46.0,
-      "step": 20072
-    },
-    {
-      "epoch": 1.5347210275818568,
-      "grad_norm": 0.0013452529674395919,
-      "learning_rate": 0.00019999883883045283,
-      "loss": 46.0,
-      "step": 20073
-    },
-    {
-      "epoch": 1.5347974845652463,
-      "grad_norm": 0.003339141607284546,
-      "learning_rate": 0.00019999883871469782,
-      "loss": 46.0,
-      "step": 20074
-    },
-    {
-      "epoch": 1.534873941548636,
-      "grad_norm": 0.0014340804191306233,
-      "learning_rate": 0.00019999883859893707,
-      "loss": 46.0,
-      "step": 20075
-    },
-    {
-      "epoch": 1.5349503985320259,
-      "grad_norm": 0.011916508898139,
-      "learning_rate": 0.00019999883848317054,
-      "loss": 46.0,
-      "step": 20076
-    },
-    {
-      "epoch": 1.5350268555154156,
-      "grad_norm": 0.0015547092771157622,
-      "learning_rate": 0.00019999883836739824,
-      "loss": 46.0,
-      "step": 20077
-    },
-    {
-      "epoch": 1.5351033124988054,
-      "grad_norm": 0.00432837288826704,
-      "learning_rate": 0.00019999883825162017,
-      "loss": 46.0,
-      "step": 20078
-    },
-    {
-      "epoch": 1.5351797694821951,
-      "grad_norm": 0.0004542127426248044,
-      "learning_rate": 0.00019999883813583632,
-      "loss": 46.0,
-      "step": 20079
-    },
-    {
-      "epoch": 1.535256226465585,
-      "grad_norm": 0.002094621304422617,
-      "learning_rate": 0.0001999988380200467,
-      "loss": 46.0,
-      "step": 20080
-    },
-    {
-      "epoch": 1.5353326834489747,
-      "grad_norm": 0.0033219624310731888,
-      "learning_rate": 0.0001999988379042513,
-      "loss": 46.0,
-      "step": 20081
-    },
-    {
-      "epoch": 1.5354091404323642,
-      "grad_norm": 0.0010463689686730504,
-      "learning_rate": 0.00019999883778845016,
-      "loss": 46.0,
-      "step": 20082
-    },
-    {
-      "epoch": 1.535485597415754,
-      "grad_norm": 0.0016901319613680243,
-      "learning_rate": 0.00019999883767264325,
-      "loss": 46.0,
-      "step": 20083
-    },
-    {
-      "epoch": 1.5355620543991437,
-      "grad_norm": 0.001070959959179163,
-      "learning_rate": 0.00019999883755683056,
-      "loss": 46.0,
-      "step": 20084
-    },
-    {
-      "epoch": 1.5356385113825333,
-      "grad_norm": 0.002557846950367093,
-      "learning_rate": 0.00019999883744101208,
-      "loss": 46.0,
-      "step": 20085
-    },
-    {
-      "epoch": 1.535714968365923,
-      "grad_norm": 0.0010747192427515984,
-      "learning_rate": 0.00019999883732518787,
-      "loss": 46.0,
-      "step": 20086
-    },
-    {
-      "epoch": 1.5357914253493128,
-      "grad_norm": 0.0009346804581582546,
-      "learning_rate": 0.0001999988372093579,
-      "loss": 46.0,
-      "step": 20087
-    },
-    {
-      "epoch": 1.5358678823327025,
-      "grad_norm": 0.0012615770101547241,
-      "learning_rate": 0.0001999988370935221,
-      "loss": 46.0,
-      "step": 20088
-    },
-    {
-      "epoch": 1.5359443393160923,
-      "grad_norm": 0.0010398214217275381,
-      "learning_rate": 0.00019999883697768058,
-      "loss": 46.0,
-      "step": 20089
-    },
-    {
-      "epoch": 1.536020796299482,
-      "grad_norm": 0.00044385544606484473,
-      "learning_rate": 0.00019999883686183328,
-      "loss": 46.0,
-      "step": 20090
-    },
-    {
-      "epoch": 1.5360972532828718,
-      "grad_norm": 0.002626515459269285,
-      "learning_rate": 0.00019999883674598018,
-      "loss": 46.0,
-      "step": 20091
-    },
-    {
-      "epoch": 1.5361737102662616,
-      "grad_norm": 0.0006823960575275123,
-      "learning_rate": 0.00019999883663012136,
-      "loss": 46.0,
-      "step": 20092
-    },
-    {
-      "epoch": 1.5362501672496511,
-      "grad_norm": 0.0021544608753174543,
-      "learning_rate": 0.00019999883651425674,
-      "loss": 46.0,
-      "step": 20093
-    },
-    {
-      "epoch": 1.536326624233041,
-      "grad_norm": 0.0011576790129765868,
-      "learning_rate": 0.00019999883639838635,
-      "loss": 46.0,
-      "step": 20094
-    },
-    {
-      "epoch": 1.5364030812164307,
-      "grad_norm": 0.0009752301266416907,
-      "learning_rate": 0.0001999988362825102,
-      "loss": 46.0,
-      "step": 20095
-    },
-    {
-      "epoch": 1.5364795381998202,
-      "grad_norm": 0.01034811232239008,
-      "learning_rate": 0.0001999988361666283,
-      "loss": 46.0,
-      "step": 20096
-    },
-    {
-      "epoch": 1.53655599518321,
-      "grad_norm": 0.0010354340774938464,
-      "learning_rate": 0.00019999883605074058,
-      "loss": 46.0,
-      "step": 20097
-    },
-    {
-      "epoch": 1.5366324521665997,
-      "grad_norm": 0.0014967684401199222,
-      "learning_rate": 0.00019999883593484712,
-      "loss": 46.0,
-      "step": 20098
-    },
-    {
-      "epoch": 1.5367089091499895,
-      "grad_norm": 0.005033082794398069,
-      "learning_rate": 0.00019999883581894789,
-      "loss": 46.0,
-      "step": 20099
-    },
-    {
-      "epoch": 1.5367853661333792,
-      "grad_norm": 0.0038119973614811897,
-      "learning_rate": 0.0001999988357030429,
-      "loss": 46.0,
-      "step": 20100
-    },
-    {
-      "epoch": 1.536861823116769,
-      "grad_norm": 0.0019197850488126278,
-      "learning_rate": 0.00019999883558713213,
-      "loss": 46.0,
-      "step": 20101
-    },
-    {
-      "epoch": 1.5369382801001588,
-      "grad_norm": 0.0009453543461859226,
-      "learning_rate": 0.00019999883547121557,
-      "loss": 46.0,
-      "step": 20102
-    },
-    {
-      "epoch": 1.5370147370835485,
-      "grad_norm": 0.0007630455656908453,
-      "learning_rate": 0.0001999988353552933,
-      "loss": 46.0,
-      "step": 20103
-    },
-    {
-      "epoch": 1.537091194066938,
-      "grad_norm": 0.002025666879490018,
-      "learning_rate": 0.0001999988352393652,
-      "loss": 46.0,
-      "step": 20104
-    },
-    {
-      "epoch": 1.5371676510503278,
-      "grad_norm": 0.004234236665070057,
-      "learning_rate": 0.00019999883512343135,
-      "loss": 46.0,
-      "step": 20105
-    },
-    {
-      "epoch": 1.5372441080337176,
-      "grad_norm": 0.005363223142921925,
-      "learning_rate": 0.00019999883500749173,
-      "loss": 46.0,
-      "step": 20106
-    },
-    {
-      "epoch": 1.5373205650171071,
-      "grad_norm": 0.004469661973416805,
-      "learning_rate": 0.00019999883489154634,
-      "loss": 46.0,
-      "step": 20107
-    },
-    {
-      "epoch": 1.5373970220004969,
-      "grad_norm": 0.000897183024790138,
-      "learning_rate": 0.0001999988347755952,
-      "loss": 46.0,
-      "step": 20108
-    },
-    {
-      "epoch": 1.5374734789838866,
-      "grad_norm": 0.0028845383785665035,
-      "learning_rate": 0.00019999883465963828,
-      "loss": 46.0,
-      "step": 20109
-    },
-    {
-      "epoch": 1.5375499359672764,
-      "grad_norm": 0.00042767953709699214,
-      "learning_rate": 0.00019999883454367557,
-      "loss": 46.0,
-      "step": 20110
-    },
-    {
-      "epoch": 1.5376263929506662,
-      "grad_norm": 0.0009216518956236541,
-      "learning_rate": 0.0001999988344277071,
-      "loss": 46.0,
-      "step": 20111
-    },
-    {
-      "epoch": 1.537702849934056,
-      "grad_norm": 0.0008002610993571579,
-      "learning_rate": 0.0001999988343117329,
-      "loss": 46.0,
-      "step": 20112
-    },
-    {
-      "epoch": 1.5377793069174457,
-      "grad_norm": 0.001917055924423039,
-      "learning_rate": 0.00019999883419575287,
-      "loss": 46.0,
-      "step": 20113
-    },
-    {
-      "epoch": 1.5378557639008354,
-      "grad_norm": 0.001889210776425898,
-      "learning_rate": 0.00019999883407976712,
-      "loss": 46.0,
-      "step": 20114
-    },
-    {
-      "epoch": 1.537932220884225,
-      "grad_norm": 0.001500417129136622,
-      "learning_rate": 0.0001999988339637756,
-      "loss": 46.0,
-      "step": 20115
-    },
-    {
-      "epoch": 1.5380086778676147,
-      "grad_norm": 0.0012968312948942184,
-      "learning_rate": 0.00019999883384777827,
-      "loss": 46.0,
-      "step": 20116
-    },
-    {
-      "epoch": 1.5380851348510043,
-      "grad_norm": 0.003965589217841625,
-      "learning_rate": 0.00019999883373177517,
-      "loss": 46.0,
-      "step": 20117
-    },
-    {
-      "epoch": 1.538161591834394,
-      "grad_norm": 0.0021684097591787577,
-      "learning_rate": 0.00019999883361576632,
-      "loss": 46.0,
-      "step": 20118
-    },
-    {
-      "epoch": 1.5382380488177838,
-      "grad_norm": 0.001090253354050219,
-      "learning_rate": 0.0001999988334997517,
-      "loss": 46.0,
-      "step": 20119
-    },
-    {
-      "epoch": 1.5383145058011736,
-      "grad_norm": 0.0009661591029725969,
-      "learning_rate": 0.00019999883338373133,
-      "loss": 46.0,
-      "step": 20120
-    },
-    {
-      "epoch": 1.5383909627845633,
-      "grad_norm": 0.010315361432731152,
-      "learning_rate": 0.0001999988332677052,
-      "loss": 46.0,
-      "step": 20121
-    },
-    {
-      "epoch": 1.538467419767953,
-      "grad_norm": 0.0015649535926058888,
-      "learning_rate": 0.00019999883315167325,
-      "loss": 46.0,
-      "step": 20122
-    },
-    {
-      "epoch": 1.5385438767513429,
-      "grad_norm": 0.0015382713172584772,
-      "learning_rate": 0.00019999883303563557,
-      "loss": 46.0,
-      "step": 20123
-    },
-    {
-      "epoch": 1.5386203337347326,
-      "grad_norm": 0.0022082326468080282,
-      "learning_rate": 0.00019999883291959208,
-      "loss": 46.0,
-      "step": 20124
-    },
-    {
-      "epoch": 1.5386967907181224,
-      "grad_norm": 0.000669982167892158,
-      "learning_rate": 0.00019999883280354285,
-      "loss": 46.0,
-      "step": 20125
-    },
-    {
-      "epoch": 1.538773247701512,
-      "grad_norm": 0.0016096193576231599,
-      "learning_rate": 0.00019999883268748784,
-      "loss": 46.0,
-      "step": 20126
-    },
-    {
-      "epoch": 1.5388497046849017,
-      "grad_norm": 0.0007320628501474857,
-      "learning_rate": 0.0001999988325714271,
-      "loss": 46.0,
-      "step": 20127
-    },
-    {
-      "epoch": 1.5389261616682912,
-      "grad_norm": 0.001097666914574802,
-      "learning_rate": 0.00019999883245536054,
-      "loss": 46.0,
-      "step": 20128
-    },
-    {
-      "epoch": 1.539002618651681,
-      "grad_norm": 0.0013889233814552426,
-      "learning_rate": 0.0001999988323392882,
-      "loss": 46.0,
-      "step": 20129
-    },
-    {
-      "epoch": 1.5390790756350707,
-      "grad_norm": 0.0020983247086405754,
-      "learning_rate": 0.00019999883222321014,
-      "loss": 46.0,
-      "step": 20130
-    },
-    {
-      "epoch": 1.5391555326184605,
-      "grad_norm": 0.0011359930504113436,
-      "learning_rate": 0.0001999988321071263,
-      "loss": 46.0,
-      "step": 20131
-    },
-    {
-      "epoch": 1.5392319896018503,
-      "grad_norm": 0.010866709053516388,
-      "learning_rate": 0.00019999883199103667,
-      "loss": 46.0,
-      "step": 20132
-    },
-    {
-      "epoch": 1.53930844658524,
-      "grad_norm": 0.0008480479009449482,
-      "learning_rate": 0.00019999883187494128,
-      "loss": 46.0,
-      "step": 20133
-    },
-    {
-      "epoch": 1.5393849035686298,
-      "grad_norm": 0.0010555190965533257,
-      "learning_rate": 0.0001999988317588401,
-      "loss": 46.0,
-      "step": 20134
-    },
-    {
-      "epoch": 1.5394613605520195,
-      "grad_norm": 0.0010322027374058962,
-      "learning_rate": 0.0001999988316427332,
-      "loss": 46.0,
-      "step": 20135
-    },
-    {
-      "epoch": 1.5395378175354093,
-      "grad_norm": 0.0011218597646802664,
-      "learning_rate": 0.0001999988315266205,
-      "loss": 46.0,
-      "step": 20136
-    },
-    {
-      "epoch": 1.5396142745187988,
-      "grad_norm": 0.001945384545251727,
-      "learning_rate": 0.00019999883141050205,
-      "loss": 46.0,
-      "step": 20137
-    },
-    {
-      "epoch": 1.5396907315021886,
-      "grad_norm": 0.0014750653645023704,
-      "learning_rate": 0.0001999988312943778,
-      "loss": 46.0,
-      "step": 20138
-    },
-    {
-      "epoch": 1.5397671884855781,
-      "grad_norm": 0.0015266623813658953,
-      "learning_rate": 0.0001999988311782478,
-      "loss": 46.0,
-      "step": 20139
-    },
-    {
-      "epoch": 1.539843645468968,
-      "grad_norm": 0.0012810645857825875,
-      "learning_rate": 0.000199998831062112,
-      "loss": 46.0,
-      "step": 20140
-    },
-    {
-      "epoch": 1.5399201024523577,
-      "grad_norm": 0.003351570339873433,
-      "learning_rate": 0.00019999883094597048,
-      "loss": 46.0,
-      "step": 20141
-    },
-    {
-      "epoch": 1.5399965594357474,
-      "grad_norm": 0.0015526647912338376,
-      "learning_rate": 0.00019999883082982316,
-      "loss": 46.0,
-      "step": 20142
-    },
-    {
-      "epoch": 1.5400730164191372,
-      "grad_norm": 0.0012627204414457083,
-      "learning_rate": 0.00019999883071367006,
-      "loss": 46.0,
-      "step": 20143
-    },
-    {
-      "epoch": 1.540149473402527,
-      "grad_norm": 0.002908539492636919,
-      "learning_rate": 0.0001999988305975112,
-      "loss": 46.0,
-      "step": 20144
-    },
-    {
-      "epoch": 1.5402259303859167,
-      "grad_norm": 0.0026809454429894686,
-      "learning_rate": 0.0001999988304813466,
-      "loss": 46.0,
-      "step": 20145
-    },
-    {
-      "epoch": 1.5403023873693065,
-      "grad_norm": 0.006418232340365648,
-      "learning_rate": 0.00019999883036517623,
-      "loss": 46.0,
-      "step": 20146
-    },
-    {
-      "epoch": 1.5403788443526962,
-      "grad_norm": 0.001138602034188807,
-      "learning_rate": 0.00019999883024900003,
-      "loss": 46.0,
-      "step": 20147
-    },
-    {
-      "epoch": 1.5404553013360858,
-      "grad_norm": 0.0009076328715309501,
-      "learning_rate": 0.00019999883013281812,
-      "loss": 46.0,
-      "step": 20148
-    },
-    {
-      "epoch": 1.5405317583194755,
-      "grad_norm": 0.0023125596344470978,
-      "learning_rate": 0.0001999988300166304,
-      "loss": 46.0,
-      "step": 20149
-    },
-    {
-      "epoch": 1.540608215302865,
-      "grad_norm": 0.007807811722159386,
-      "learning_rate": 0.00019999882990043695,
-      "loss": 46.0,
-      "step": 20150
-    },
-    {
-      "epoch": 1.5406846722862548,
-      "grad_norm": 0.0021241719368845224,
-      "learning_rate": 0.00019999882978423772,
-      "loss": 46.0,
-      "step": 20151
-    },
-    {
-      "epoch": 1.5407611292696446,
-      "grad_norm": 0.0007885193917900324,
-      "learning_rate": 0.00019999882966803269,
-      "loss": 46.0,
-      "step": 20152
-    },
-    {
-      "epoch": 1.5408375862530344,
-      "grad_norm": 0.0008994427626021206,
-      "learning_rate": 0.0001999988295518219,
-      "loss": 46.0,
-      "step": 20153
-    },
-    {
-      "epoch": 1.5409140432364241,
-      "grad_norm": 0.004090954549610615,
-      "learning_rate": 0.00019999882943560538,
-      "loss": 46.0,
-      "step": 20154
-    },
-    {
-      "epoch": 1.5409905002198139,
-      "grad_norm": 0.001574847148731351,
-      "learning_rate": 0.00019999882931938308,
-      "loss": 46.0,
-      "step": 20155
-    },
-    {
-      "epoch": 1.5410669572032036,
-      "grad_norm": 0.000622278661467135,
-      "learning_rate": 0.00019999882920315498,
-      "loss": 46.0,
-      "step": 20156
-    },
-    {
-      "epoch": 1.5411434141865934,
-      "grad_norm": 0.0005950985359959304,
-      "learning_rate": 0.0001999988290869211,
-      "loss": 46.0,
-      "step": 20157
-    },
-    {
-      "epoch": 1.541219871169983,
-      "grad_norm": 0.0012609973782673478,
-      "learning_rate": 0.0001999988289706815,
-      "loss": 46.0,
-      "step": 20158
-    },
-    {
-      "epoch": 1.5412963281533727,
-      "grad_norm": 0.0012368971947580576,
-      "learning_rate": 0.0001999988288544361,
-      "loss": 46.0,
-      "step": 20159
-    },
-    {
-      "epoch": 1.5413727851367625,
-      "grad_norm": 0.002296286867931485,
-      "learning_rate": 0.00019999882873818493,
-      "loss": 46.0,
-      "step": 20160
-    },
-    {
-      "epoch": 1.541449242120152,
-      "grad_norm": 0.0006178400362841785,
-      "learning_rate": 0.000199998828621928,
-      "loss": 46.0,
-      "step": 20161
-    },
-    {
-      "epoch": 1.5415256991035418,
-      "grad_norm": 0.000837925064843148,
-      "learning_rate": 0.0001999988285056653,
-      "loss": 46.0,
-      "step": 20162
-    },
-    {
-      "epoch": 1.5416021560869315,
-      "grad_norm": 0.0009972784901037812,
-      "learning_rate": 0.00019999882838939682,
-      "loss": 46.0,
-      "step": 20163
-    },
-    {
-      "epoch": 1.5416786130703213,
-      "grad_norm": 0.0009486893541179597,
-      "learning_rate": 0.0001999988282731226,
-      "loss": 46.0,
-      "step": 20164
-    },
-    {
-      "epoch": 1.541755070053711,
-      "grad_norm": 0.005483708810061216,
-      "learning_rate": 0.0001999988281568426,
-      "loss": 46.0,
-      "step": 20165
-    },
-    {
-      "epoch": 1.5418315270371008,
-      "grad_norm": 0.0035777557641267776,
-      "learning_rate": 0.0001999988280405568,
-      "loss": 46.0,
-      "step": 20166
-    },
-    {
-      "epoch": 1.5419079840204906,
-      "grad_norm": 0.0019115410977974534,
-      "learning_rate": 0.00019999882792426526,
-      "loss": 46.0,
-      "step": 20167
-    },
-    {
-      "epoch": 1.5419844410038803,
-      "grad_norm": 0.003945767879486084,
-      "learning_rate": 0.00019999882780796796,
-      "loss": 46.0,
-      "step": 20168
-    },
-    {
-      "epoch": 1.5420608979872699,
-      "grad_norm": 0.0011247425572946668,
-      "learning_rate": 0.00019999882769166486,
-      "loss": 46.0,
-      "step": 20169
-    },
-    {
-      "epoch": 1.5421373549706596,
-      "grad_norm": 0.0007622959674336016,
-      "learning_rate": 0.000199998827575356,
-      "loss": 46.0,
-      "step": 20170
-    },
-    {
-      "epoch": 1.5422138119540494,
-      "grad_norm": 0.0007714423118159175,
-      "learning_rate": 0.00019999882745904137,
-      "loss": 46.0,
-      "step": 20171
-    },
-    {
-      "epoch": 1.542290268937439,
-      "grad_norm": 0.0012912163510918617,
-      "learning_rate": 0.00019999882734272098,
-      "loss": 46.0,
-      "step": 20172
-    },
-    {
-      "epoch": 1.5423667259208287,
-      "grad_norm": 0.002913554199039936,
-      "learning_rate": 0.0001999988272263948,
-      "loss": 46.0,
-      "step": 20173
-    },
-    {
-      "epoch": 1.5424431829042184,
-      "grad_norm": 0.0004806739743798971,
-      "learning_rate": 0.00019999882711006287,
-      "loss": 46.0,
-      "step": 20174
-    },
-    {
-      "epoch": 1.5425196398876082,
-      "grad_norm": 0.008131139911711216,
-      "learning_rate": 0.0001999988269937252,
-      "loss": 46.0,
-      "step": 20175
-    },
-    {
-      "epoch": 1.542596096870998,
-      "grad_norm": 0.0006606580573134124,
-      "learning_rate": 0.0001999988268773817,
-      "loss": 46.0,
-      "step": 20176
-    },
-    {
-      "epoch": 1.5426725538543877,
-      "grad_norm": 0.0037447737995535135,
-      "learning_rate": 0.00019999882676103244,
-      "loss": 46.0,
-      "step": 20177
-    },
-    {
-      "epoch": 1.5427490108377775,
-      "grad_norm": 0.0024279500357806683,
-      "learning_rate": 0.00019999882664467744,
-      "loss": 46.0,
-      "step": 20178
-    },
-    {
-      "epoch": 1.5428254678211673,
-      "grad_norm": 0.0014981512213125825,
-      "learning_rate": 0.0001999988265283167,
-      "loss": 46.0,
-      "step": 20179
-    },
-    {
-      "epoch": 1.5429019248045568,
-      "grad_norm": 0.00046638367348350585,
-      "learning_rate": 0.0001999988264119501,
-      "loss": 46.0,
-      "step": 20180
-    },
-    {
-      "epoch": 1.5429783817879466,
-      "grad_norm": 0.0008069496252574027,
-      "learning_rate": 0.00019999882629557778,
-      "loss": 46.0,
-      "step": 20181
-    },
-    {
-      "epoch": 1.5430548387713363,
-      "grad_norm": 0.0025949811097234488,
-      "learning_rate": 0.0001999988261791997,
-      "loss": 46.0,
-      "step": 20182
-    },
-    {
-      "epoch": 1.5431312957547259,
-      "grad_norm": 0.002164746169000864,
-      "learning_rate": 0.00019999882606281584,
-      "loss": 46.0,
-      "step": 20183
-    },
-    {
-      "epoch": 1.5432077527381156,
-      "grad_norm": 0.010595188476145267,
-      "learning_rate": 0.00019999882594642622,
-      "loss": 46.0,
-      "step": 20184
-    },
-    {
-      "epoch": 1.5432842097215054,
-      "grad_norm": 0.003314277622848749,
-      "learning_rate": 0.00019999882583003083,
-      "loss": 46.0,
-      "step": 20185
-    },
-    {
-      "epoch": 1.5433606667048951,
-      "grad_norm": 0.00048742940998636186,
-      "learning_rate": 0.00019999882571362964,
-      "loss": 46.0,
-      "step": 20186
-    },
-    {
-      "epoch": 1.543437123688285,
-      "grad_norm": 0.0005593310343101621,
-      "learning_rate": 0.00019999882559722273,
-      "loss": 46.0,
-      "step": 20187
-    },
-    {
-      "epoch": 1.5435135806716747,
-      "grad_norm": 0.0014772203285247087,
-      "learning_rate": 0.00019999882548081001,
-      "loss": 46.0,
-      "step": 20188
-    },
-    {
-      "epoch": 1.5435900376550644,
-      "grad_norm": 0.004009888041764498,
-      "learning_rate": 0.00019999882536439153,
-      "loss": 46.0,
-      "step": 20189
-    },
-    {
-      "epoch": 1.5436664946384542,
-      "grad_norm": 0.0011133728548884392,
-      "learning_rate": 0.0001999988252479673,
-      "loss": 46.0,
-      "step": 20190
-    },
-    {
-      "epoch": 1.5437429516218437,
-      "grad_norm": 0.0008045951835811138,
-      "learning_rate": 0.00019999882513153727,
-      "loss": 46.0,
-      "step": 20191
-    },
-    {
-      "epoch": 1.5438194086052335,
-      "grad_norm": 0.0016953253652900457,
-      "learning_rate": 0.0001999988250151015,
-      "loss": 46.0,
-      "step": 20192
-    },
-    {
-      "epoch": 1.5438958655886232,
-      "grad_norm": 0.004482306074351072,
-      "learning_rate": 0.00019999882489865996,
-      "loss": 46.0,
-      "step": 20193
-    },
-    {
-      "epoch": 1.5439723225720128,
-      "grad_norm": 0.010717046447098255,
-      "learning_rate": 0.0001999988247822126,
-      "loss": 46.0,
-      "step": 20194
-    },
-    {
-      "epoch": 1.5440487795554025,
-      "grad_norm": 0.0038280882872641087,
-      "learning_rate": 0.00019999882466575954,
-      "loss": 46.0,
-      "step": 20195
-    },
-    {
-      "epoch": 1.5441252365387923,
-      "grad_norm": 0.0015682651428505778,
-      "learning_rate": 0.00019999882454930067,
-      "loss": 46.0,
-      "step": 20196
-    },
-    {
-      "epoch": 1.544201693522182,
-      "grad_norm": 0.0009037533891387284,
-      "learning_rate": 0.00019999882443283605,
-      "loss": 46.0,
-      "step": 20197
-    },
-    {
-      "epoch": 1.5442781505055718,
-      "grad_norm": 0.0016367345815524459,
-      "learning_rate": 0.00019999882431636566,
-      "loss": 46.0,
-      "step": 20198
-    },
-    {
-      "epoch": 1.5443546074889616,
-      "grad_norm": 0.0007118707871995866,
-      "learning_rate": 0.00019999882419988947,
-      "loss": 46.0,
-      "step": 20199
-    },
-    {
-      "epoch": 1.5444310644723513,
-      "grad_norm": 0.006440271623432636,
-      "learning_rate": 0.00019999882408340753,
-      "loss": 46.0,
-      "step": 20200
-    },
-    {
-      "epoch": 1.544507521455741,
-      "grad_norm": 0.0005165501497685909,
-      "learning_rate": 0.00019999882396691982,
-      "loss": 46.0,
-      "step": 20201
-    },
-    {
-      "epoch": 1.5445839784391306,
-      "grad_norm": 0.0011839722283184528,
-      "learning_rate": 0.00019999882385042633,
-      "loss": 46.0,
-      "step": 20202
-    },
-    {
-      "epoch": 1.5446604354225204,
-      "grad_norm": 0.0017505462747067213,
-      "learning_rate": 0.0001999988237339271,
-      "loss": 46.0,
-      "step": 20203
-    },
-    {
-      "epoch": 1.5447368924059102,
-      "grad_norm": 0.001137514947913587,
-      "learning_rate": 0.00019999882361742207,
-      "loss": 46.0,
-      "step": 20204
-    },
-    {
-      "epoch": 1.5448133493892997,
-      "grad_norm": 0.0036035377997905016,
-      "learning_rate": 0.0001999988235009113,
-      "loss": 46.0,
-      "step": 20205
-    },
-    {
-      "epoch": 1.5448898063726895,
-      "grad_norm": 0.002373239491134882,
-      "learning_rate": 0.00019999882338439474,
-      "loss": 46.0,
-      "step": 20206
-    },
-    {
-      "epoch": 1.5449662633560792,
-      "grad_norm": 0.011920620687305927,
-      "learning_rate": 0.0001999988232678724,
-      "loss": 46.0,
-      "step": 20207
-    },
-    {
-      "epoch": 1.545042720339469,
-      "grad_norm": 0.0007025506929494441,
-      "learning_rate": 0.00019999882315134432,
-      "loss": 46.0,
-      "step": 20208
-    },
-    {
-      "epoch": 1.5451191773228588,
-      "grad_norm": 0.0006856209365651011,
-      "learning_rate": 0.00019999882303481048,
-      "loss": 46.0,
-      "step": 20209
-    },
-    {
-      "epoch": 1.5451956343062485,
-      "grad_norm": 0.0031736339442431927,
-      "learning_rate": 0.00019999882291827083,
-      "loss": 46.0,
-      "step": 20210
-    },
-    {
-      "epoch": 1.5452720912896383,
-      "grad_norm": 0.0026848490815609694,
-      "learning_rate": 0.0001999988228017254,
-      "loss": 46.0,
-      "step": 20211
-    },
-    {
-      "epoch": 1.545348548273028,
-      "grad_norm": 0.0007010195986367762,
-      "learning_rate": 0.00019999882268517425,
-      "loss": 46.0,
-      "step": 20212
-    },
-    {
-      "epoch": 1.5454250052564176,
-      "grad_norm": 0.0015660380013287067,
-      "learning_rate": 0.0001999988225686173,
-      "loss": 46.0,
-      "step": 20213
-    },
-    {
-      "epoch": 1.5455014622398073,
-      "grad_norm": 0.0009423309820704162,
-      "learning_rate": 0.0001999988224520546,
-      "loss": 46.0,
-      "step": 20214
-    },
-    {
-      "epoch": 1.545577919223197,
-      "grad_norm": 0.002163567114621401,
-      "learning_rate": 0.00019999882233548614,
-      "loss": 46.0,
-      "step": 20215
-    },
-    {
-      "epoch": 1.5456543762065866,
-      "grad_norm": 0.0007394609274342656,
-      "learning_rate": 0.00019999882221891188,
-      "loss": 46.0,
-      "step": 20216
-    },
-    {
-      "epoch": 1.5457308331899764,
-      "grad_norm": 0.0004886023234575987,
-      "learning_rate": 0.00019999882210233185,
-      "loss": 46.0,
-      "step": 20217
-    },
-    {
-      "epoch": 1.5458072901733662,
-      "grad_norm": 0.002775841159746051,
-      "learning_rate": 0.00019999882198574608,
-      "loss": 46.0,
-      "step": 20218
-    },
-    {
-      "epoch": 1.545883747156756,
-      "grad_norm": 0.0008558969129808247,
-      "learning_rate": 0.0001999988218691545,
-      "loss": 46.0,
-      "step": 20219
-    },
-    {
-      "epoch": 1.5459602041401457,
-      "grad_norm": 0.002447514096274972,
-      "learning_rate": 0.0001999988217525572,
-      "loss": 46.0,
-      "step": 20220
-    },
-    {
-      "epoch": 1.5460366611235354,
-      "grad_norm": 0.001322864554822445,
-      "learning_rate": 0.0001999988216359541,
-      "loss": 46.0,
-      "step": 20221
-    },
-    {
-      "epoch": 1.5461131181069252,
-      "grad_norm": 0.0018312036991119385,
-      "learning_rate": 0.00019999882151934523,
-      "loss": 46.0,
-      "step": 20222
-    },
-    {
-      "epoch": 1.546189575090315,
-      "grad_norm": 0.0009744654526002705,
-      "learning_rate": 0.0001999988214027306,
-      "loss": 46.0,
-      "step": 20223
-    },
-    {
-      "epoch": 1.5462660320737045,
-      "grad_norm": 0.001992936013266444,
-      "learning_rate": 0.00019999882128611017,
-      "loss": 46.0,
-      "step": 20224
-    },
-    {
-      "epoch": 1.5463424890570943,
-      "grad_norm": 0.0021744153928011656,
-      "learning_rate": 0.000199998821169484,
-      "loss": 46.0,
-      "step": 20225
-    },
-    {
-      "epoch": 1.546418946040484,
-      "grad_norm": 0.000638212775811553,
-      "learning_rate": 0.00019999882105285207,
-      "loss": 46.0,
-      "step": 20226
-    },
-    {
-      "epoch": 1.5464954030238736,
-      "grad_norm": 0.0009381708805449307,
-      "learning_rate": 0.00019999882093621438,
-      "loss": 46.0,
-      "step": 20227
-    },
-    {
-      "epoch": 1.5465718600072633,
-      "grad_norm": 0.0008068294264376163,
-      "learning_rate": 0.00019999882081957087,
-      "loss": 46.0,
-      "step": 20228
-    },
-    {
-      "epoch": 1.546648316990653,
-      "grad_norm": 0.0024747317656874657,
-      "learning_rate": 0.00019999882070292164,
-      "loss": 46.0,
-      "step": 20229
-    },
-    {
-      "epoch": 1.5467247739740428,
-      "grad_norm": 0.002904994646087289,
-      "learning_rate": 0.0001999988205862666,
-      "loss": 46.0,
-      "step": 20230
-    },
-    {
-      "epoch": 1.5468012309574326,
-      "grad_norm": 0.0014590652426704764,
-      "learning_rate": 0.00019999882046960584,
-      "loss": 46.0,
-      "step": 20231
-    },
-    {
-      "epoch": 1.5468776879408224,
-      "grad_norm": 0.0013190620811656117,
-      "learning_rate": 0.00019999882035293926,
-      "loss": 46.0,
-      "step": 20232
-    },
-    {
-      "epoch": 1.5469541449242121,
-      "grad_norm": 0.0006903313333168626,
-      "learning_rate": 0.00019999882023626696,
-      "loss": 46.0,
-      "step": 20233
-    },
-    {
-      "epoch": 1.5470306019076019,
-      "grad_norm": 0.0008527478785254061,
-      "learning_rate": 0.00019999882011958886,
-      "loss": 46.0,
-      "step": 20234
-    },
-    {
-      "epoch": 1.5471070588909914,
-      "grad_norm": 0.0011800432112067938,
-      "learning_rate": 0.00019999882000290497,
-      "loss": 46.0,
-      "step": 20235
-    },
-    {
-      "epoch": 1.5471835158743812,
-      "grad_norm": 0.0005883224657736719,
-      "learning_rate": 0.00019999881988621535,
-      "loss": 46.0,
-      "step": 20236
-    },
-    {
-      "epoch": 1.547259972857771,
-      "grad_norm": 0.002735986141487956,
-      "learning_rate": 0.00019999881976951996,
-      "loss": 46.0,
-      "step": 20237
-    },
-    {
-      "epoch": 1.5473364298411605,
-      "grad_norm": 0.0007256532553583384,
-      "learning_rate": 0.00019999881965281877,
-      "loss": 46.0,
-      "step": 20238
-    },
-    {
-      "epoch": 1.5474128868245502,
-      "grad_norm": 0.0011123957810923457,
-      "learning_rate": 0.0001999988195361118,
-      "loss": 46.0,
-      "step": 20239
-    },
-    {
-      "epoch": 1.54748934380794,
-      "grad_norm": 0.0010934810852631927,
-      "learning_rate": 0.00019999881941939912,
-      "loss": 46.0,
-      "step": 20240
-    },
-    {
-      "epoch": 1.5475658007913298,
-      "grad_norm": 0.001546977087855339,
-      "learning_rate": 0.00019999881930268064,
-      "loss": 46.0,
-      "step": 20241
-    },
-    {
-      "epoch": 1.5476422577747195,
-      "grad_norm": 0.0026557340752333403,
-      "learning_rate": 0.00019999881918595635,
-      "loss": 46.0,
-      "step": 20242
-    },
-    {
-      "epoch": 1.5477187147581093,
-      "grad_norm": 0.0004453608999028802,
-      "learning_rate": 0.00019999881906922635,
-      "loss": 46.0,
-      "step": 20243
-    },
-    {
-      "epoch": 1.547795171741499,
-      "grad_norm": 0.0009327111765742302,
-      "learning_rate": 0.00019999881895249054,
-      "loss": 46.0,
-      "step": 20244
-    },
-    {
-      "epoch": 1.5478716287248888,
-      "grad_norm": 0.003226051339879632,
-      "learning_rate": 0.00019999881883574897,
-      "loss": 46.0,
-      "step": 20245
-    },
-    {
-      "epoch": 1.5479480857082784,
-      "grad_norm": 0.0032770477700978518,
-      "learning_rate": 0.00019999881871900167,
-      "loss": 46.0,
-      "step": 20246
-    },
-    {
-      "epoch": 1.5480245426916681,
-      "grad_norm": 0.0011336986208334565,
-      "learning_rate": 0.00019999881860224857,
-      "loss": 46.0,
-      "step": 20247
-    },
-    {
-      "epoch": 1.5481009996750579,
-      "grad_norm": 0.0022451074328273535,
-      "learning_rate": 0.0001999988184854897,
-      "loss": 46.0,
-      "step": 20248
-    },
-    {
-      "epoch": 1.5481774566584474,
-      "grad_norm": 0.001434828620404005,
-      "learning_rate": 0.00019999881836872506,
-      "loss": 46.0,
-      "step": 20249
-    },
-    {
-      "epoch": 1.5482539136418372,
-      "grad_norm": 0.0011933945352211595,
-      "learning_rate": 0.00019999881825195464,
-      "loss": 46.0,
-      "step": 20250
-    },
-    {
-      "epoch": 1.548330370625227,
-      "grad_norm": 0.0009847356704995036,
-      "learning_rate": 0.00019999881813517848,
-      "loss": 46.0,
-      "step": 20251
-    },
-    {
-      "epoch": 1.5484068276086167,
-      "grad_norm": 0.0013555812183767557,
-      "learning_rate": 0.00019999881801839651,
-      "loss": 46.0,
-      "step": 20252
-    },
-    {
-      "epoch": 1.5484832845920065,
-      "grad_norm": 0.0015788115561008453,
-      "learning_rate": 0.0001999988179016088,
-      "loss": 46.0,
-      "step": 20253
-    },
-    {
-      "epoch": 1.5485597415753962,
-      "grad_norm": 0.0005741255008615553,
-      "learning_rate": 0.00019999881778481532,
-      "loss": 46.0,
-      "step": 20254
-    },
-    {
-      "epoch": 1.548636198558786,
-      "grad_norm": 0.0008115789969451725,
-      "learning_rate": 0.00019999881766801606,
-      "loss": 46.0,
-      "step": 20255
-    },
-    {
-      "epoch": 1.5487126555421757,
-      "grad_norm": 0.002358938567340374,
-      "learning_rate": 0.00019999881755121103,
-      "loss": 46.0,
-      "step": 20256
-    },
-    {
-      "epoch": 1.5487891125255653,
-      "grad_norm": 0.00048008712474256754,
-      "learning_rate": 0.00019999881743440026,
-      "loss": 46.0,
-      "step": 20257
-    },
-    {
-      "epoch": 1.548865569508955,
-      "grad_norm": 0.010214579291641712,
-      "learning_rate": 0.0001999988173175837,
-      "loss": 46.0,
-      "step": 20258
-    },
-    {
-      "epoch": 1.5489420264923446,
-      "grad_norm": 0.0048224530182778835,
-      "learning_rate": 0.00019999881720076136,
-      "loss": 46.0,
-      "step": 20259
-    },
-    {
-      "epoch": 1.5490184834757343,
-      "grad_norm": 0.002420478267595172,
-      "learning_rate": 0.00019999881708393326,
-      "loss": 46.0,
-      "step": 20260
-    },
-    {
-      "epoch": 1.549094940459124,
-      "grad_norm": 0.002419776748865843,
-      "learning_rate": 0.0001999988169670994,
-      "loss": 46.0,
-      "step": 20261
-    },
-    {
-      "epoch": 1.5491713974425139,
-      "grad_norm": 0.0010610036551952362,
-      "learning_rate": 0.00019999881685025975,
-      "loss": 46.0,
-      "step": 20262
-    },
-    {
-      "epoch": 1.5492478544259036,
-      "grad_norm": 0.0023971563205122948,
-      "learning_rate": 0.00019999881673341433,
-      "loss": 46.0,
-      "step": 20263
-    },
-    {
-      "epoch": 1.5493243114092934,
-      "grad_norm": 0.0010830481769517064,
-      "learning_rate": 0.00019999881661656317,
-      "loss": 46.0,
-      "step": 20264
-    },
-    {
-      "epoch": 1.5494007683926831,
-      "grad_norm": 0.0005202654283493757,
-      "learning_rate": 0.0001999988164997062,
-      "loss": 46.0,
-      "step": 20265
-    },
-    {
-      "epoch": 1.549477225376073,
-      "grad_norm": 0.0006369129405356944,
-      "learning_rate": 0.0001999988163828435,
-      "loss": 46.0,
-      "step": 20266
-    },
-    {
-      "epoch": 1.5495536823594627,
-      "grad_norm": 0.0012664797250181437,
-      "learning_rate": 0.000199998816265975,
-      "loss": 46.0,
-      "step": 20267
-    },
-    {
-      "epoch": 1.5496301393428522,
-      "grad_norm": 0.0009975587017834187,
-      "learning_rate": 0.00019999881614910075,
-      "loss": 46.0,
-      "step": 20268
-    },
-    {
-      "epoch": 1.549706596326242,
-      "grad_norm": 0.001327601377852261,
-      "learning_rate": 0.00019999881603222072,
-      "loss": 46.0,
-      "step": 20269
-    },
-    {
-      "epoch": 1.5497830533096315,
-      "grad_norm": 0.0007339943549595773,
-      "learning_rate": 0.00019999881591533495,
-      "loss": 46.0,
-      "step": 20270
-    },
-    {
-      "epoch": 1.5498595102930213,
-      "grad_norm": 0.0029646975453943014,
-      "learning_rate": 0.00019999881579844337,
-      "loss": 46.0,
-      "step": 20271
-    },
-    {
-      "epoch": 1.549935967276411,
-      "grad_norm": 0.0009326311992481351,
-      "learning_rate": 0.00019999881568154605,
-      "loss": 46.0,
-      "step": 20272
-    },
-    {
-      "epoch": 1.5500124242598008,
-      "grad_norm": 0.0031023994088172913,
-      "learning_rate": 0.00019999881556464292,
-      "loss": 46.0,
-      "step": 20273
-    },
-    {
-      "epoch": 1.5500888812431906,
-      "grad_norm": 0.0019929411355406046,
-      "learning_rate": 0.00019999881544773405,
-      "loss": 46.0,
-      "step": 20274
-    },
-    {
-      "epoch": 1.5501653382265803,
-      "grad_norm": 0.0015137135051190853,
-      "learning_rate": 0.0001999988153308194,
-      "loss": 46.0,
-      "step": 20275
-    },
-    {
-      "epoch": 1.55024179520997,
-      "grad_norm": 0.0008138444391079247,
-      "learning_rate": 0.000199998815213899,
-      "loss": 46.0,
-      "step": 20276
-    },
-    {
-      "epoch": 1.5503182521933598,
-      "grad_norm": 0.004656719043850899,
-      "learning_rate": 0.0001999988150969728,
-      "loss": 46.0,
-      "step": 20277
-    },
-    {
-      "epoch": 1.5503947091767496,
-      "grad_norm": 0.0005335908499546349,
-      "learning_rate": 0.00019999881498004087,
-      "loss": 46.0,
-      "step": 20278
-    },
-    {
-      "epoch": 1.5504711661601391,
-      "grad_norm": 0.0022118152119219303,
-      "learning_rate": 0.00019999881486310313,
-      "loss": 46.0,
-      "step": 20279
-    },
-    {
-      "epoch": 1.550547623143529,
-      "grad_norm": 0.0011276851873844862,
-      "learning_rate": 0.00019999881474615965,
-      "loss": 46.0,
-      "step": 20280
-    },
-    {
-      "epoch": 1.5506240801269184,
-      "grad_norm": 0.0010222881101071835,
-      "learning_rate": 0.0001999988146292104,
-      "loss": 46.0,
-      "step": 20281
-    },
-    {
-      "epoch": 1.5507005371103082,
-      "grad_norm": 0.001826761057600379,
-      "learning_rate": 0.00019999881451225536,
-      "loss": 46.0,
-      "step": 20282
-    },
-    {
-      "epoch": 1.550776994093698,
-      "grad_norm": 0.0014074405189603567,
-      "learning_rate": 0.0001999988143952946,
-      "loss": 46.0,
-      "step": 20283
-    },
-    {
-      "epoch": 1.5508534510770877,
-      "grad_norm": 0.0007083318778313696,
-      "learning_rate": 0.000199998814278328,
-      "loss": 46.0,
-      "step": 20284
-    },
-    {
-      "epoch": 1.5509299080604775,
-      "grad_norm": 0.0011526060989126563,
-      "learning_rate": 0.00019999881416135566,
-      "loss": 46.0,
-      "step": 20285
-    },
-    {
-      "epoch": 1.5510063650438672,
-      "grad_norm": 0.0005087851313874125,
-      "learning_rate": 0.00019999881404437757,
-      "loss": 46.0,
-      "step": 20286
-    },
-    {
-      "epoch": 1.551082822027257,
-      "grad_norm": 0.001099860412068665,
-      "learning_rate": 0.0001999988139273937,
-      "loss": 46.0,
-      "step": 20287
-    },
-    {
-      "epoch": 1.5511592790106468,
-      "grad_norm": 0.0009887852938845754,
-      "learning_rate": 0.00019999881381040403,
-      "loss": 46.0,
-      "step": 20288
-    },
-    {
-      "epoch": 1.5512357359940363,
-      "grad_norm": 0.0009057808783836663,
-      "learning_rate": 0.00019999881369340864,
-      "loss": 46.0,
-      "step": 20289
-    },
-    {
-      "epoch": 1.551312192977426,
-      "grad_norm": 0.004380343947559595,
-      "learning_rate": 0.00019999881357640745,
-      "loss": 46.0,
-      "step": 20290
-    },
-    {
-      "epoch": 1.5513886499608158,
-      "grad_norm": 0.0009076056885533035,
-      "learning_rate": 0.0001999988134594005,
-      "loss": 46.0,
-      "step": 20291
-    },
-    {
-      "epoch": 1.5514651069442054,
-      "grad_norm": 0.000971232249867171,
-      "learning_rate": 0.0001999988133423878,
-      "loss": 46.0,
-      "step": 20292
-    },
-    {
-      "epoch": 1.5515415639275951,
-      "grad_norm": 0.011418294161558151,
-      "learning_rate": 0.0001999988132253693,
-      "loss": 46.0,
-      "step": 20293
-    },
-    {
-      "epoch": 1.5516180209109849,
-      "grad_norm": 0.00544810201972723,
-      "learning_rate": 0.00019999881310834504,
-      "loss": 46.0,
-      "step": 20294
-    },
-    {
-      "epoch": 1.5516944778943746,
-      "grad_norm": 0.0012689060531556606,
-      "learning_rate": 0.000199998812991315,
-      "loss": 46.0,
-      "step": 20295
-    },
-    {
-      "epoch": 1.5517709348777644,
-      "grad_norm": 0.0004988238215446472,
-      "learning_rate": 0.0001999988128742792,
-      "loss": 46.0,
-      "step": 20296
-    },
-    {
-      "epoch": 1.5518473918611542,
-      "grad_norm": 0.0017380461795255542,
-      "learning_rate": 0.00019999881275723763,
-      "loss": 46.0,
-      "step": 20297
-    },
-    {
-      "epoch": 1.551923848844544,
-      "grad_norm": 0.001257389667443931,
-      "learning_rate": 0.00019999881264019028,
-      "loss": 46.0,
-      "step": 20298
-    },
-    {
-      "epoch": 1.5520003058279337,
-      "grad_norm": 0.001535171759314835,
-      "learning_rate": 0.00019999881252313716,
-      "loss": 46.0,
-      "step": 20299
-    },
-    {
-      "epoch": 1.5520767628113232,
-      "grad_norm": 0.011539758183062077,
-      "learning_rate": 0.00019999881240607831,
-      "loss": 46.0,
-      "step": 20300
-    },
-    {
-      "epoch": 1.552153219794713,
-      "grad_norm": 0.007358429953455925,
-      "learning_rate": 0.00019999881228901365,
-      "loss": 46.0,
-      "step": 20301
-    },
-    {
-      "epoch": 1.5522296767781028,
-      "grad_norm": 0.0010508454870432615,
-      "learning_rate": 0.00019999881217194326,
-      "loss": 46.0,
-      "step": 20302
-    },
-    {
-      "epoch": 1.5523061337614923,
-      "grad_norm": 0.004711613059043884,
-      "learning_rate": 0.00019999881205486707,
-      "loss": 46.0,
-      "step": 20303
-    },
-    {
-      "epoch": 1.552382590744882,
-      "grad_norm": 0.0010608049342408776,
-      "learning_rate": 0.0001999988119377851,
-      "loss": 46.0,
-      "step": 20304
-    },
-    {
-      "epoch": 1.5524590477282718,
-      "grad_norm": 0.00034053970011882484,
-      "learning_rate": 0.00019999881182069737,
-      "loss": 46.0,
-      "step": 20305
-    },
-    {
-      "epoch": 1.5525355047116616,
-      "grad_norm": 0.007171119097620249,
-      "learning_rate": 0.00019999881170360389,
-      "loss": 46.0,
-      "step": 20306
-    },
-    {
-      "epoch": 1.5526119616950513,
-      "grad_norm": 0.0009845270542427897,
-      "learning_rate": 0.00019999881158650463,
-      "loss": 46.0,
-      "step": 20307
-    },
-    {
-      "epoch": 1.552688418678441,
-      "grad_norm": 0.0009325945866294205,
-      "learning_rate": 0.00019999881146939957,
-      "loss": 46.0,
-      "step": 20308
-    },
-    {
-      "epoch": 1.5527648756618309,
-      "grad_norm": 0.0014653062680736184,
-      "learning_rate": 0.00019999881135228877,
-      "loss": 46.0,
-      "step": 20309
-    },
-    {
-      "epoch": 1.5528413326452206,
-      "grad_norm": 0.0006368522881530225,
-      "learning_rate": 0.00019999881123517222,
-      "loss": 46.0,
-      "step": 20310
-    },
-    {
-      "epoch": 1.5529177896286102,
-      "grad_norm": 0.004681960679590702,
-      "learning_rate": 0.00019999881111804987,
-      "loss": 46.0,
-      "step": 20311
-    },
-    {
-      "epoch": 1.552994246612,
-      "grad_norm": 0.0007058492046780884,
-      "learning_rate": 0.00019999881100092178,
-      "loss": 46.0,
-      "step": 20312
-    },
-    {
-      "epoch": 1.5530707035953897,
-      "grad_norm": 0.0009261932573281229,
-      "learning_rate": 0.00019999881088378788,
-      "loss": 46.0,
-      "step": 20313
-    },
-    {
-      "epoch": 1.5531471605787792,
-      "grad_norm": 0.006575034931302071,
-      "learning_rate": 0.00019999881076664824,
-      "loss": 46.0,
-      "step": 20314
-    },
-    {
-      "epoch": 1.553223617562169,
-      "grad_norm": 0.00036820978857576847,
-      "learning_rate": 0.0001999988106495028,
-      "loss": 46.0,
-      "step": 20315
-    },
-    {
-      "epoch": 1.5533000745455587,
-      "grad_norm": 0.0007867935346439481,
-      "learning_rate": 0.00019999881053235164,
-      "loss": 46.0,
-      "step": 20316
-    },
-    {
-      "epoch": 1.5533765315289485,
-      "grad_norm": 0.0007679632399231195,
-      "learning_rate": 0.00019999881041519467,
-      "loss": 46.0,
-      "step": 20317
-    },
-    {
-      "epoch": 1.5534529885123383,
-      "grad_norm": 0.0014141318388283253,
-      "learning_rate": 0.00019999881029803197,
-      "loss": 46.0,
-      "step": 20318
-    },
-    {
-      "epoch": 1.553529445495728,
-      "grad_norm": 0.001984365750104189,
-      "learning_rate": 0.00019999881018086346,
-      "loss": 46.0,
-      "step": 20319
-    },
-    {
-      "epoch": 1.5536059024791178,
-      "grad_norm": 0.0010668091708794236,
-      "learning_rate": 0.0001999988100636892,
-      "loss": 46.0,
-      "step": 20320
-    },
-    {
-      "epoch": 1.5536823594625075,
-      "grad_norm": 0.0019426877843216062,
-      "learning_rate": 0.00019999880994650915,
-      "loss": 46.0,
-      "step": 20321
-    },
-    {
-      "epoch": 1.553758816445897,
-      "grad_norm": 0.0005831917515024543,
-      "learning_rate": 0.00019999880982932334,
-      "loss": 46.0,
-      "step": 20322
-    },
-    {
-      "epoch": 1.5538352734292868,
-      "grad_norm": 0.0013772249221801758,
-      "learning_rate": 0.0001999988097121318,
-      "loss": 46.0,
-      "step": 20323
-    },
-    {
-      "epoch": 1.5539117304126766,
-      "grad_norm": 0.0024889749474823475,
-      "learning_rate": 0.00019999880959493445,
-      "loss": 46.0,
-      "step": 20324
-    },
-    {
-      "epoch": 1.5539881873960661,
-      "grad_norm": 0.0010758873540908098,
-      "learning_rate": 0.00019999880947773132,
-      "loss": 46.0,
-      "step": 20325
-    },
-    {
-      "epoch": 1.554064644379456,
-      "grad_norm": 0.005038782954216003,
-      "learning_rate": 0.00019999880936052246,
-      "loss": 46.0,
-      "step": 20326
-    },
-    {
-      "epoch": 1.5541411013628457,
-      "grad_norm": 0.0004562149115372449,
-      "learning_rate": 0.0001999988092433078,
-      "loss": 46.0,
-      "step": 20327
-    },
-    {
-      "epoch": 1.5542175583462354,
-      "grad_norm": 0.0022854830604046583,
-      "learning_rate": 0.00019999880912608737,
-      "loss": 46.0,
-      "step": 20328
-    },
-    {
-      "epoch": 1.5542940153296252,
-      "grad_norm": 0.0008763526566326618,
-      "learning_rate": 0.00019999880900886118,
-      "loss": 46.0,
-      "step": 20329
-    },
-    {
-      "epoch": 1.554370472313015,
-      "grad_norm": 0.001301970100030303,
-      "learning_rate": 0.00019999880889162922,
-      "loss": 46.0,
-      "step": 20330
-    },
-    {
-      "epoch": 1.5544469292964047,
-      "grad_norm": 0.0016204144340008497,
-      "learning_rate": 0.0001999988087743915,
-      "loss": 46.0,
-      "step": 20331
-    },
-    {
-      "epoch": 1.5545233862797945,
-      "grad_norm": 0.0008843140094541013,
-      "learning_rate": 0.000199998808657148,
-      "loss": 46.0,
-      "step": 20332
-    },
-    {
-      "epoch": 1.554599843263184,
-      "grad_norm": 0.0006387456669472158,
-      "learning_rate": 0.00019999880853989875,
-      "loss": 46.0,
-      "step": 20333
-    },
-    {
-      "epoch": 1.5546763002465738,
-      "grad_norm": 0.0008754501468501985,
-      "learning_rate": 0.00019999880842264372,
-      "loss": 46.0,
-      "step": 20334
-    },
-    {
-      "epoch": 1.5547527572299635,
-      "grad_norm": 0.0004958927165716887,
-      "learning_rate": 0.0001999988083053829,
-      "loss": 46.0,
-      "step": 20335
-    },
-    {
-      "epoch": 1.554829214213353,
-      "grad_norm": 0.0007865107618272305,
-      "learning_rate": 0.00019999880818811634,
-      "loss": 46.0,
-      "step": 20336
-    },
-    {
-      "epoch": 1.5549056711967428,
-      "grad_norm": 0.0013830723473802209,
-      "learning_rate": 0.00019999880807084397,
-      "loss": 46.0,
-      "step": 20337
-    },
-    {
-      "epoch": 1.5549821281801326,
-      "grad_norm": 0.0020435727201402187,
-      "learning_rate": 0.00019999880795356587,
-      "loss": 46.0,
-      "step": 20338
-    },
-    {
-      "epoch": 1.5550585851635224,
-      "grad_norm": 0.0009164935327135026,
-      "learning_rate": 0.000199998807836282,
-      "loss": 46.0,
-      "step": 20339
-    },
-    {
-      "epoch": 1.5551350421469121,
-      "grad_norm": 0.001196700381115079,
-      "learning_rate": 0.00019999880771899234,
-      "loss": 46.0,
-      "step": 20340
-    },
-    {
-      "epoch": 1.5552114991303019,
-      "grad_norm": 0.0022680407855659723,
-      "learning_rate": 0.00019999880760169692,
-      "loss": 46.0,
-      "step": 20341
-    },
-    {
-      "epoch": 1.5552879561136916,
-      "grad_norm": 0.0005042555276304483,
-      "learning_rate": 0.00019999880748439573,
-      "loss": 46.0,
-      "step": 20342
-    },
-    {
-      "epoch": 1.5553644130970814,
-      "grad_norm": 0.0016276228707283735,
-      "learning_rate": 0.00019999880736708877,
-      "loss": 46.0,
-      "step": 20343
-    },
-    {
-      "epoch": 1.555440870080471,
-      "grad_norm": 0.000963117927312851,
-      "learning_rate": 0.00019999880724977604,
-      "loss": 46.0,
-      "step": 20344
-    },
-    {
-      "epoch": 1.5555173270638607,
-      "grad_norm": 0.0010502267396077514,
-      "learning_rate": 0.00019999880713245753,
-      "loss": 46.0,
-      "step": 20345
-    },
-    {
-      "epoch": 1.5555937840472505,
-      "grad_norm": 0.0013835520949214697,
-      "learning_rate": 0.00019999880701513325,
-      "loss": 46.0,
-      "step": 20346
-    },
-    {
-      "epoch": 1.55567024103064,
-      "grad_norm": 0.0008860574453137815,
-      "learning_rate": 0.00019999880689780324,
-      "loss": 46.0,
-      "step": 20347
-    },
-    {
-      "epoch": 1.5557466980140298,
-      "grad_norm": 0.0024157913867384195,
-      "learning_rate": 0.00019999880678046742,
-      "loss": 46.0,
-      "step": 20348
-    },
-    {
-      "epoch": 1.5558231549974195,
-      "grad_norm": 0.003710251534357667,
-      "learning_rate": 0.00019999880666312584,
-      "loss": 46.0,
-      "step": 20349
-    },
-    {
-      "epoch": 1.5558996119808093,
-      "grad_norm": 0.0009747496224008501,
-      "learning_rate": 0.0001999988065457785,
-      "loss": 46.0,
-      "step": 20350
-    },
-    {
-      "epoch": 1.555976068964199,
-      "grad_norm": 0.0016732370713725686,
-      "learning_rate": 0.00019999880642842537,
-      "loss": 46.0,
-      "step": 20351
-    },
-    {
-      "epoch": 1.5560525259475888,
-      "grad_norm": 0.00039919669507071376,
-      "learning_rate": 0.00019999880631106648,
-      "loss": 46.0,
-      "step": 20352
-    },
-    {
-      "epoch": 1.5561289829309786,
-      "grad_norm": 0.0008389498107135296,
-      "learning_rate": 0.00019999880619370184,
-      "loss": 46.0,
-      "step": 20353
-    },
-    {
-      "epoch": 1.5562054399143683,
-      "grad_norm": 0.0010824992787092924,
-      "learning_rate": 0.00019999880607633142,
-      "loss": 46.0,
-      "step": 20354
-    },
-    {
-      "epoch": 1.5562818968977579,
-      "grad_norm": 0.0010296195978298783,
-      "learning_rate": 0.0001999988059589552,
-      "loss": 46.0,
-      "step": 20355
-    },
-    {
-      "epoch": 1.5563583538811476,
-      "grad_norm": 0.004622041247785091,
-      "learning_rate": 0.00019999880584157324,
-      "loss": 46.0,
-      "step": 20356
-    },
-    {
-      "epoch": 1.5564348108645374,
-      "grad_norm": 0.0009110590908676386,
-      "learning_rate": 0.0001999988057241855,
-      "loss": 46.0,
-      "step": 20357
-    },
-    {
-      "epoch": 1.556511267847927,
-      "grad_norm": 0.0015861720312386751,
-      "learning_rate": 0.00019999880560679203,
-      "loss": 46.0,
-      "step": 20358
-    },
-    {
-      "epoch": 1.5565877248313167,
-      "grad_norm": 0.0007488686824217439,
-      "learning_rate": 0.00019999880548939275,
-      "loss": 46.0,
-      "step": 20359
-    },
-    {
-      "epoch": 1.5566641818147064,
-      "grad_norm": 0.001435458310879767,
-      "learning_rate": 0.00019999880537198772,
-      "loss": 46.0,
-      "step": 20360
-    },
-    {
-      "epoch": 1.5567406387980962,
-      "grad_norm": 0.0018419409170746803,
-      "learning_rate": 0.0001999988052545769,
-      "loss": 46.0,
-      "step": 20361
-    },
-    {
-      "epoch": 1.556817095781486,
-      "grad_norm": 0.0011355690658092499,
-      "learning_rate": 0.00019999880513716032,
-      "loss": 46.0,
-      "step": 20362
-    },
-    {
-      "epoch": 1.5568935527648757,
-      "grad_norm": 0.0019013368291780353,
-      "learning_rate": 0.00019999880501973797,
-      "loss": 46.0,
-      "step": 20363
-    },
-    {
-      "epoch": 1.5569700097482655,
-      "grad_norm": 0.012250037863850594,
-      "learning_rate": 0.00019999880490230987,
-      "loss": 46.0,
-      "step": 20364
-    },
-    {
-      "epoch": 1.5570464667316553,
-      "grad_norm": 0.0035572205670177937,
-      "learning_rate": 0.00019999880478487598,
-      "loss": 46.0,
-      "step": 20365
-    },
-    {
-      "epoch": 1.5571229237150448,
-      "grad_norm": 0.0017253459664061666,
-      "learning_rate": 0.0001999988046674363,
-      "loss": 46.0,
-      "step": 20366
-    },
-    {
-      "epoch": 1.5571993806984346,
-      "grad_norm": 0.0010187855223193765,
-      "learning_rate": 0.0001999988045499909,
-      "loss": 46.0,
-      "step": 20367
-    },
-    {
-      "epoch": 1.5572758376818243,
-      "grad_norm": 0.0011968839680776,
-      "learning_rate": 0.00019999880443253968,
-      "loss": 46.0,
-      "step": 20368
-    },
-    {
-      "epoch": 1.5573522946652139,
-      "grad_norm": 0.0007393784471787512,
-      "learning_rate": 0.00019999880431508272,
-      "loss": 46.0,
-      "step": 20369
-    },
-    {
-      "epoch": 1.5574287516486036,
-      "grad_norm": 0.0054636672139167786,
-      "learning_rate": 0.00019999880419762001,
-      "loss": 46.0,
-      "step": 20370
-    },
-    {
-      "epoch": 1.5575052086319934,
-      "grad_norm": 0.0013797142310068011,
-      "learning_rate": 0.00019999880408015148,
-      "loss": 46.0,
-      "step": 20371
-    },
-    {
-      "epoch": 1.5575816656153831,
-      "grad_norm": 0.0006277625216171145,
-      "learning_rate": 0.00019999880396267723,
-      "loss": 46.0,
-      "step": 20372
-    },
-    {
-      "epoch": 1.557658122598773,
-      "grad_norm": 0.0021176051814109087,
-      "learning_rate": 0.00019999880384519717,
-      "loss": 46.0,
-      "step": 20373
-    },
-    {
-      "epoch": 1.5577345795821627,
-      "grad_norm": 0.0010102363303303719,
-      "learning_rate": 0.00019999880372771137,
-      "loss": 46.0,
-      "step": 20374
-    },
-    {
-      "epoch": 1.5578110365655524,
-      "grad_norm": 0.0019918824546039104,
-      "learning_rate": 0.0001999988036102198,
-      "loss": 46.0,
-      "step": 20375
-    },
-    {
-      "epoch": 1.5578874935489422,
-      "grad_norm": 0.0008416366181336343,
-      "learning_rate": 0.00019999880349272245,
-      "loss": 46.0,
-      "step": 20376
-    },
-    {
-      "epoch": 1.5579639505323317,
-      "grad_norm": 0.0039924620650708675,
-      "learning_rate": 0.00019999880337521933,
-      "loss": 46.0,
-      "step": 20377
-    },
-    {
-      "epoch": 1.5580404075157215,
-      "grad_norm": 0.001947305747307837,
-      "learning_rate": 0.00019999880325771043,
-      "loss": 46.0,
-      "step": 20378
-    },
-    {
-      "epoch": 1.5581168644991112,
-      "grad_norm": 0.0015651760622859001,
-      "learning_rate": 0.00019999880314019577,
-      "loss": 46.0,
-      "step": 20379
-    },
-    {
-      "epoch": 1.5581933214825008,
-      "grad_norm": 0.0019158568466082215,
-      "learning_rate": 0.00019999880302267532,
-      "loss": 46.0,
-      "step": 20380
-    },
-    {
-      "epoch": 1.5582697784658905,
-      "grad_norm": 0.005281593184918165,
-      "learning_rate": 0.00019999880290514916,
-      "loss": 46.0,
-      "step": 20381
-    },
-    {
-      "epoch": 1.5583462354492803,
-      "grad_norm": 0.0023288815282285213,
-      "learning_rate": 0.00019999880278761718,
-      "loss": 46.0,
-      "step": 20382
-    },
-    {
-      "epoch": 1.55842269243267,
-      "grad_norm": 0.0035321833565831184,
-      "learning_rate": 0.00019999880267007944,
-      "loss": 46.0,
-      "step": 20383
-    },
-    {
-      "epoch": 1.5584991494160598,
-      "grad_norm": 0.0005555053357966244,
-      "learning_rate": 0.00019999880255253593,
-      "loss": 46.0,
-      "step": 20384
-    },
-    {
-      "epoch": 1.5585756063994496,
-      "grad_norm": 0.0017697765724733472,
-      "learning_rate": 0.00019999880243498668,
-      "loss": 46.0,
-      "step": 20385
-    },
-    {
-      "epoch": 1.5586520633828393,
-      "grad_norm": 0.0005751054850406945,
-      "learning_rate": 0.00019999880231743163,
-      "loss": 46.0,
-      "step": 20386
-    },
-    {
-      "epoch": 1.558728520366229,
-      "grad_norm": 0.0010335806291550398,
-      "learning_rate": 0.0001999988021998708,
-      "loss": 46.0,
-      "step": 20387
-    },
-    {
-      "epoch": 1.5588049773496186,
-      "grad_norm": 0.0007465775706805289,
-      "learning_rate": 0.00019999880208230423,
-      "loss": 46.0,
-      "step": 20388
-    },
-    {
-      "epoch": 1.5588814343330084,
-      "grad_norm": 0.0005362386582419276,
-      "learning_rate": 0.00019999880196473185,
-      "loss": 46.0,
-      "step": 20389
-    },
-    {
-      "epoch": 1.558957891316398,
-      "grad_norm": 0.000868162838742137,
-      "learning_rate": 0.00019999880184715373,
-      "loss": 46.0,
-      "step": 20390
-    },
-    {
-      "epoch": 1.5590343482997877,
-      "grad_norm": 0.0016189463203772902,
-      "learning_rate": 0.00019999880172956986,
-      "loss": 46.0,
-      "step": 20391
-    },
-    {
-      "epoch": 1.5591108052831775,
-      "grad_norm": 0.0016034849686548114,
-      "learning_rate": 0.00019999880161198022,
-      "loss": 46.0,
-      "step": 20392
-    },
-    {
-      "epoch": 1.5591872622665672,
-      "grad_norm": 0.005704442970454693,
-      "learning_rate": 0.00019999880149438476,
-      "loss": 46.0,
-      "step": 20393
-    },
-    {
-      "epoch": 1.559263719249957,
-      "grad_norm": 0.0015032096998766065,
-      "learning_rate": 0.00019999880137678357,
-      "loss": 46.0,
-      "step": 20394
-    },
-    {
-      "epoch": 1.5593401762333468,
-      "grad_norm": 0.0014174515381455421,
-      "learning_rate": 0.0001999988012591766,
-      "loss": 46.0,
-      "step": 20395
-    },
-    {
-      "epoch": 1.5594166332167365,
-      "grad_norm": 0.0010963774984702468,
-      "learning_rate": 0.00019999880114156387,
-      "loss": 46.0,
-      "step": 20396
-    },
-    {
-      "epoch": 1.5594930902001263,
-      "grad_norm": 0.0008113204967230558,
-      "learning_rate": 0.00019999880102394534,
-      "loss": 46.0,
-      "step": 20397
-    },
-    {
-      "epoch": 1.559569547183516,
-      "grad_norm": 0.006441914476454258,
-      "learning_rate": 0.00019999880090632109,
-      "loss": 46.0,
-      "step": 20398
-    },
-    {
-      "epoch": 1.5596460041669056,
-      "grad_norm": 0.0019139398355036974,
-      "learning_rate": 0.00019999880078869103,
-      "loss": 46.0,
-      "step": 20399
-    },
-    {
-      "epoch": 1.5597224611502953,
-      "grad_norm": 0.0010233854409307241,
-      "learning_rate": 0.00019999880067105523,
-      "loss": 46.0,
-      "step": 20400
-    },
-    {
-      "epoch": 1.5597989181336849,
-      "grad_norm": 0.0006365884328261018,
-      "learning_rate": 0.00019999880055341366,
-      "loss": 46.0,
-      "step": 20401
-    },
-    {
-      "epoch": 1.5598753751170746,
-      "grad_norm": 0.0016257824609056115,
-      "learning_rate": 0.00019999880043576628,
-      "loss": 46.0,
-      "step": 20402
-    },
-    {
-      "epoch": 1.5599518321004644,
-      "grad_norm": 0.0005208643851801753,
-      "learning_rate": 0.00019999880031811316,
-      "loss": 46.0,
-      "step": 20403
-    },
-    {
-      "epoch": 1.5600282890838542,
-      "grad_norm": 0.0019847780931741,
-      "learning_rate": 0.00019999880020045424,
-      "loss": 46.0,
-      "step": 20404
-    },
-    {
-      "epoch": 1.560104746067244,
-      "grad_norm": 0.0008785325335338712,
-      "learning_rate": 0.0001999988000827896,
-      "loss": 46.0,
-      "step": 20405
-    },
-    {
-      "epoch": 1.5601812030506337,
-      "grad_norm": 0.013772883452475071,
-      "learning_rate": 0.00019999879996511916,
-      "loss": 46.0,
-      "step": 20406
-    },
-    {
-      "epoch": 1.5602576600340234,
-      "grad_norm": 0.001103578950278461,
-      "learning_rate": 0.00019999879984744298,
-      "loss": 46.0,
-      "step": 20407
-    },
-    {
-      "epoch": 1.5603341170174132,
-      "grad_norm": 0.009055916219949722,
-      "learning_rate": 0.000199998799729761,
-      "loss": 46.0,
-      "step": 20408
-    },
-    {
-      "epoch": 1.560410574000803,
-      "grad_norm": 0.0012917376589030027,
-      "learning_rate": 0.00019999879961207326,
-      "loss": 46.0,
-      "step": 20409
-    },
-    {
-      "epoch": 1.5604870309841925,
-      "grad_norm": 0.0007482227520085871,
-      "learning_rate": 0.00019999879949437978,
-      "loss": 46.0,
-      "step": 20410
-    },
-    {
-      "epoch": 1.5605634879675823,
-      "grad_norm": 0.0004723152087535709,
-      "learning_rate": 0.00019999879937668047,
-      "loss": 46.0,
-      "step": 20411
-    },
-    {
-      "epoch": 1.5606399449509718,
-      "grad_norm": 0.004322065506130457,
-      "learning_rate": 0.00019999879925897542,
-      "loss": 46.0,
-      "step": 20412
-    },
-    {
-      "epoch": 1.5607164019343616,
-      "grad_norm": 0.0008148546330630779,
-      "learning_rate": 0.00019999879914126462,
-      "loss": 46.0,
-      "step": 20413
-    },
-    {
-      "epoch": 1.5607928589177513,
-      "grad_norm": 0.0003856477269437164,
-      "learning_rate": 0.00019999879902354804,
-      "loss": 46.0,
-      "step": 20414
-    },
-    {
-      "epoch": 1.560869315901141,
-      "grad_norm": 0.0010075336322188377,
-      "learning_rate": 0.00019999879890582567,
-      "loss": 46.0,
-      "step": 20415
-    },
-    {
-      "epoch": 1.5609457728845308,
-      "grad_norm": 0.005647373851388693,
-      "learning_rate": 0.00019999879878809752,
-      "loss": 46.0,
-      "step": 20416
-    },
-    {
-      "epoch": 1.5610222298679206,
-      "grad_norm": 0.0021779201924800873,
-      "learning_rate": 0.00019999879867036363,
-      "loss": 46.0,
-      "step": 20417
-    },
-    {
-      "epoch": 1.5610986868513104,
-      "grad_norm": 0.001866922597400844,
-      "learning_rate": 0.000199998798552624,
-      "loss": 46.0,
-      "step": 20418
-    },
-    {
-      "epoch": 1.5611751438347001,
-      "grad_norm": 0.001715793740004301,
-      "learning_rate": 0.00019999879843487855,
-      "loss": 46.0,
-      "step": 20419
-    },
-    {
-      "epoch": 1.5612516008180897,
-      "grad_norm": 0.0006460725562646985,
-      "learning_rate": 0.00019999879831712733,
-      "loss": 46.0,
-      "step": 20420
-    },
-    {
-      "epoch": 1.5613280578014794,
-      "grad_norm": 0.0024972131941467524,
-      "learning_rate": 0.00019999879819937037,
-      "loss": 46.0,
-      "step": 20421
-    },
-    {
-      "epoch": 1.5614045147848692,
-      "grad_norm": 0.0012341371038928628,
-      "learning_rate": 0.00019999879808160764,
-      "loss": 46.0,
-      "step": 20422
-    },
-    {
-      "epoch": 1.5614809717682587,
-      "grad_norm": 0.0026142382994294167,
-      "learning_rate": 0.0001999987979638391,
-      "loss": 46.0,
-      "step": 20423
-    },
-    {
-      "epoch": 1.5615574287516485,
-      "grad_norm": 0.0022262735292315483,
-      "learning_rate": 0.00019999879784606483,
-      "loss": 46.0,
-      "step": 20424
-    },
-    {
-      "epoch": 1.5616338857350383,
-      "grad_norm": 0.0008153364178724587,
-      "learning_rate": 0.00019999879772828477,
-      "loss": 46.0,
-      "step": 20425
-    },
-    {
-      "epoch": 1.561710342718428,
-      "grad_norm": 0.0008112417417578399,
-      "learning_rate": 0.00019999879761049898,
-      "loss": 46.0,
-      "step": 20426
-    },
-    {
-      "epoch": 1.5617867997018178,
-      "grad_norm": 0.0006853495142422616,
-      "learning_rate": 0.00019999879749270737,
-      "loss": 46.0,
-      "step": 20427
-    },
-    {
-      "epoch": 1.5618632566852075,
-      "grad_norm": 0.0013695686357095838,
-      "learning_rate": 0.00019999879737491,
-      "loss": 46.0,
-      "step": 20428
-    },
-    {
-      "epoch": 1.5619397136685973,
-      "grad_norm": 0.001403912785463035,
-      "learning_rate": 0.0001999987972571069,
-      "loss": 46.0,
-      "step": 20429
-    },
-    {
-      "epoch": 1.562016170651987,
-      "grad_norm": 0.0006473068497143686,
-      "learning_rate": 0.000199998797139298,
-      "loss": 46.0,
-      "step": 20430
-    },
-    {
-      "epoch": 1.5620926276353766,
-      "grad_norm": 0.001170830219052732,
-      "learning_rate": 0.00019999879702148335,
-      "loss": 46.0,
-      "step": 20431
-    },
-    {
-      "epoch": 1.5621690846187664,
-      "grad_norm": 0.001028629019856453,
-      "learning_rate": 0.00019999879690366288,
-      "loss": 46.0,
-      "step": 20432
-    },
-    {
-      "epoch": 1.5622455416021561,
-      "grad_norm": 0.0033223647624254227,
-      "learning_rate": 0.0001999987967858367,
-      "loss": 46.0,
-      "step": 20433
-    },
-    {
-      "epoch": 1.5623219985855457,
-      "grad_norm": 0.0022415989078581333,
-      "learning_rate": 0.0001999987966680047,
-      "loss": 46.0,
-      "step": 20434
-    },
-    {
-      "epoch": 1.5623984555689354,
-      "grad_norm": 0.0015029300702735782,
-      "learning_rate": 0.00019999879655016698,
-      "loss": 46.0,
-      "step": 20435
-    },
-    {
-      "epoch": 1.5624749125523252,
-      "grad_norm": 0.004798507783561945,
-      "learning_rate": 0.00019999879643232347,
-      "loss": 46.0,
-      "step": 20436
-    },
-    {
-      "epoch": 1.562551369535715,
-      "grad_norm": 0.0008729594992473722,
-      "learning_rate": 0.0001999987963144742,
-      "loss": 46.0,
-      "step": 20437
-    },
-    {
-      "epoch": 1.5626278265191047,
-      "grad_norm": 0.002530557569116354,
-      "learning_rate": 0.0001999987961966191,
-      "loss": 46.0,
-      "step": 20438
-    },
-    {
-      "epoch": 1.5627042835024945,
-      "grad_norm": 0.0009950305102393031,
-      "learning_rate": 0.00019999879607875828,
-      "loss": 46.0,
-      "step": 20439
-    },
-    {
-      "epoch": 1.5627807404858842,
-      "grad_norm": 0.0016097306506708264,
-      "learning_rate": 0.0001999987959608917,
-      "loss": 46.0,
-      "step": 20440
-    },
-    {
-      "epoch": 1.562857197469274,
-      "grad_norm": 0.0011646109633147717,
-      "learning_rate": 0.00019999879584301934,
-      "loss": 46.0,
-      "step": 20441
-    },
-    {
-      "epoch": 1.5629336544526635,
-      "grad_norm": 0.00990972202271223,
-      "learning_rate": 0.00019999879572514122,
-      "loss": 46.0,
-      "step": 20442
-    },
-    {
-      "epoch": 1.5630101114360533,
-      "grad_norm": 0.0029427348636090755,
-      "learning_rate": 0.00019999879560725733,
-      "loss": 46.0,
-      "step": 20443
-    },
-    {
-      "epoch": 1.563086568419443,
-      "grad_norm": 0.001082059694454074,
-      "learning_rate": 0.00019999879548936766,
-      "loss": 46.0,
-      "step": 20444
-    },
-    {
-      "epoch": 1.5631630254028326,
-      "grad_norm": 0.0010490723652765155,
-      "learning_rate": 0.00019999879537147222,
-      "loss": 46.0,
-      "step": 20445
-    },
-    {
-      "epoch": 1.5632394823862223,
-      "grad_norm": 0.002688538981601596,
-      "learning_rate": 0.00019999879525357098,
-      "loss": 46.0,
-      "step": 20446
-    },
-    {
-      "epoch": 1.563315939369612,
-      "grad_norm": 0.001219554920680821,
-      "learning_rate": 0.00019999879513566402,
-      "loss": 46.0,
-      "step": 20447
-    },
-    {
-      "epoch": 1.5633923963530019,
-      "grad_norm": 0.0008803309174254537,
-      "learning_rate": 0.00019999879501775126,
-      "loss": 46.0,
-      "step": 20448
-    },
-    {
-      "epoch": 1.5634688533363916,
-      "grad_norm": 0.0011195058468729258,
-      "learning_rate": 0.00019999879489983276,
-      "loss": 46.0,
-      "step": 20449
-    },
-    {
-      "epoch": 1.5635453103197814,
-      "grad_norm": 0.0022292984649538994,
-      "learning_rate": 0.00019999879478190848,
-      "loss": 46.0,
-      "step": 20450
-    },
-    {
-      "epoch": 1.5636217673031712,
-      "grad_norm": 0.002224868396297097,
-      "learning_rate": 0.00019999879466397845,
-      "loss": 46.0,
-      "step": 20451
-    },
-    {
-      "epoch": 1.563698224286561,
-      "grad_norm": 0.0018027651822194457,
-      "learning_rate": 0.0001999987945460426,
-      "loss": 46.0,
-      "step": 20452
-    },
-    {
-      "epoch": 1.5637746812699505,
-      "grad_norm": 0.0011839448707178235,
-      "learning_rate": 0.000199998794428101,
-      "loss": 46.0,
-      "step": 20453
-    },
-    {
-      "epoch": 1.5638511382533402,
-      "grad_norm": 0.0015217720065265894,
-      "learning_rate": 0.00019999879431015365,
-      "loss": 46.0,
-      "step": 20454
-    },
-    {
-      "epoch": 1.56392759523673,
-      "grad_norm": 0.0007957973284646869,
-      "learning_rate": 0.00019999879419220053,
-      "loss": 46.0,
-      "step": 20455
-    },
-    {
-      "epoch": 1.5640040522201195,
-      "grad_norm": 0.0010524482931941748,
-      "learning_rate": 0.00019999879407424161,
-      "loss": 46.0,
-      "step": 20456
-    },
-    {
-      "epoch": 1.5640805092035093,
-      "grad_norm": 0.0006354114739224315,
-      "learning_rate": 0.00019999879395627695,
-      "loss": 46.0,
-      "step": 20457
-    },
-    {
-      "epoch": 1.564156966186899,
-      "grad_norm": 0.0016668433090671897,
-      "learning_rate": 0.0001999987938383065,
-      "loss": 46.0,
-      "step": 20458
-    },
-    {
-      "epoch": 1.5642334231702888,
-      "grad_norm": 0.0011136078974232078,
-      "learning_rate": 0.0001999987937203303,
-      "loss": 46.0,
-      "step": 20459
-    },
-    {
-      "epoch": 1.5643098801536786,
-      "grad_norm": 0.0038942249957472086,
-      "learning_rate": 0.0001999987936023483,
-      "loss": 46.0,
-      "step": 20460
-    },
-    {
-      "epoch": 1.5643863371370683,
-      "grad_norm": 0.004990273155272007,
-      "learning_rate": 0.00019999879348436058,
-      "loss": 46.0,
-      "step": 20461
-    },
-    {
-      "epoch": 1.564462794120458,
-      "grad_norm": 0.0032325657084584236,
-      "learning_rate": 0.00019999879336636705,
-      "loss": 46.0,
-      "step": 20462
-    },
-    {
-      "epoch": 1.5645392511038478,
-      "grad_norm": 0.0012750310124829412,
-      "learning_rate": 0.00019999879324836777,
-      "loss": 46.0,
-      "step": 20463
-    },
-    {
-      "epoch": 1.5646157080872374,
-      "grad_norm": 0.0010060176718980074,
-      "learning_rate": 0.00019999879313036272,
-      "loss": 46.0,
-      "step": 20464
-    },
-    {
-      "epoch": 1.5646921650706271,
-      "grad_norm": 0.0006512170075438917,
-      "learning_rate": 0.0001999987930123519,
-      "loss": 46.0,
-      "step": 20465
-    },
-    {
-      "epoch": 1.564768622054017,
-      "grad_norm": 0.0012688649585470557,
-      "learning_rate": 0.0001999987928943353,
-      "loss": 46.0,
-      "step": 20466
-    },
-    {
-      "epoch": 1.5648450790374064,
-      "grad_norm": 0.00587441073730588,
-      "learning_rate": 0.00019999879277631292,
-      "loss": 46.0,
-      "step": 20467
-    },
-    {
-      "epoch": 1.5649215360207962,
-      "grad_norm": 0.0014448395231738687,
-      "learning_rate": 0.00019999879265828477,
-      "loss": 46.0,
-      "step": 20468
-    },
-    {
-      "epoch": 1.564997993004186,
-      "grad_norm": 0.0010669083567336202,
-      "learning_rate": 0.00019999879254025088,
-      "loss": 46.0,
-      "step": 20469
-    },
-    {
-      "epoch": 1.5650744499875757,
-      "grad_norm": 0.0026821040082722902,
-      "learning_rate": 0.00019999879242221121,
-      "loss": 46.0,
-      "step": 20470
-    },
-    {
-      "epoch": 1.5651509069709655,
-      "grad_norm": 0.000854135665576905,
-      "learning_rate": 0.00019999879230416578,
-      "loss": 46.0,
-      "step": 20471
-    },
-    {
-      "epoch": 1.5652273639543552,
-      "grad_norm": 0.0017521621193736792,
-      "learning_rate": 0.00019999879218611456,
-      "loss": 46.0,
-      "step": 20472
-    },
-    {
-      "epoch": 1.565303820937745,
-      "grad_norm": 0.0003465542977210134,
-      "learning_rate": 0.00019999879206805758,
-      "loss": 46.0,
-      "step": 20473
-    },
-    {
-      "epoch": 1.5653802779211348,
-      "grad_norm": 0.00042865925934165716,
-      "learning_rate": 0.00019999879194999482,
-      "loss": 46.0,
-      "step": 20474
-    },
-    {
-      "epoch": 1.5654567349045243,
-      "grad_norm": 0.0010470196139067411,
-      "learning_rate": 0.0001999987918319263,
-      "loss": 46.0,
-      "step": 20475
-    },
-    {
-      "epoch": 1.565533191887914,
-      "grad_norm": 0.0008275810978375375,
-      "learning_rate": 0.00019999879171385203,
-      "loss": 46.0,
-      "step": 20476
-    },
-    {
-      "epoch": 1.5656096488713038,
-      "grad_norm": 0.0010817410657182336,
-      "learning_rate": 0.00019999879159577198,
-      "loss": 46.0,
-      "step": 20477
-    },
-    {
-      "epoch": 1.5656861058546934,
-      "grad_norm": 0.0007110472652129829,
-      "learning_rate": 0.00019999879147768613,
-      "loss": 46.0,
-      "step": 20478
-    },
-    {
-      "epoch": 1.5657625628380831,
-      "grad_norm": 0.0036815183702856302,
-      "learning_rate": 0.00019999879135959456,
-      "loss": 46.0,
-      "step": 20479
-    },
-    {
-      "epoch": 1.5658390198214729,
-      "grad_norm": 0.001409601652994752,
-      "learning_rate": 0.00019999879124149719,
-      "loss": 46.0,
-      "step": 20480
-    },
-    {
-      "epoch": 1.5659154768048626,
-      "grad_norm": 0.0006302836118265986,
-      "learning_rate": 0.00019999879112339404,
-      "loss": 46.0,
-      "step": 20481
-    },
-    {
-      "epoch": 1.5659919337882524,
-      "grad_norm": 0.0012520537711679935,
-      "learning_rate": 0.00019999879100528512,
-      "loss": 46.0,
-      "step": 20482
-    },
-    {
-      "epoch": 1.5660683907716422,
-      "grad_norm": 0.0032005272805690765,
-      "learning_rate": 0.00019999879088717046,
-      "loss": 46.0,
-      "step": 20483
-    },
-    {
-      "epoch": 1.566144847755032,
-      "grad_norm": 0.0020078090019524097,
-      "learning_rate": 0.00019999879076905002,
-      "loss": 46.0,
-      "step": 20484
-    },
-    {
-      "epoch": 1.5662213047384217,
-      "grad_norm": 0.0019918247126042843,
-      "learning_rate": 0.0001999987906509238,
-      "loss": 46.0,
-      "step": 20485
-    },
-    {
-      "epoch": 1.5662977617218112,
-      "grad_norm": 0.002960031619295478,
-      "learning_rate": 0.00019999879053279182,
-      "loss": 46.0,
-      "step": 20486
-    },
-    {
-      "epoch": 1.566374218705201,
-      "grad_norm": 0.0006002835580147803,
-      "learning_rate": 0.0001999987904146541,
-      "loss": 46.0,
-      "step": 20487
-    },
-    {
-      "epoch": 1.5664506756885908,
-      "grad_norm": 0.0034434457775205374,
-      "learning_rate": 0.00019999879029651053,
-      "loss": 46.0,
-      "step": 20488
-    },
-    {
-      "epoch": 1.5665271326719803,
-      "grad_norm": 0.002080704551190138,
-      "learning_rate": 0.00019999879017836128,
-      "loss": 46.0,
-      "step": 20489
-    },
-    {
-      "epoch": 1.56660358965537,
-      "grad_norm": 0.001218809629790485,
-      "learning_rate": 0.0001999987900602062,
-      "loss": 46.0,
-      "step": 20490
-    },
-    {
-      "epoch": 1.5666800466387598,
-      "grad_norm": 0.00241008005104959,
-      "learning_rate": 0.00019999878994204538,
-      "loss": 46.0,
-      "step": 20491
-    },
-    {
-      "epoch": 1.5667565036221496,
-      "grad_norm": 0.0023835906758904457,
-      "learning_rate": 0.00019999878982387878,
-      "loss": 46.0,
-      "step": 20492
-    },
-    {
-      "epoch": 1.5668329606055393,
-      "grad_norm": 0.0010999118676409125,
-      "learning_rate": 0.0001999987897057064,
-      "loss": 46.0,
-      "step": 20493
-    },
-    {
-      "epoch": 1.566909417588929,
-      "grad_norm": 0.0008885576971806586,
-      "learning_rate": 0.00019999878958752826,
-      "loss": 46.0,
-      "step": 20494
-    },
-    {
-      "epoch": 1.5669858745723189,
-      "grad_norm": 0.006827152334153652,
-      "learning_rate": 0.00019999878946934437,
-      "loss": 46.0,
-      "step": 20495
-    },
-    {
-      "epoch": 1.5670623315557086,
-      "grad_norm": 0.0013334945542737842,
-      "learning_rate": 0.00019999878935115468,
-      "loss": 46.0,
-      "step": 20496
-    },
-    {
-      "epoch": 1.5671387885390982,
-      "grad_norm": 0.0016205388819798827,
-      "learning_rate": 0.00019999878923295924,
-      "loss": 46.0,
-      "step": 20497
-    },
-    {
-      "epoch": 1.567215245522488,
-      "grad_norm": 0.0006641803774982691,
-      "learning_rate": 0.00019999878911475803,
-      "loss": 46.0,
-      "step": 20498
-    },
-    {
-      "epoch": 1.5672917025058777,
-      "grad_norm": 0.008667604066431522,
-      "learning_rate": 0.00019999878899655104,
-      "loss": 46.0,
-      "step": 20499
-    },
-    {
-      "epoch": 1.5673681594892672,
-      "grad_norm": 0.0015205841045826674,
-      "learning_rate": 0.00019999878887833828,
-      "loss": 46.0,
-      "step": 20500
-    },
-    {
-      "epoch": 1.567444616472657,
-      "grad_norm": 0.0013605996500700712,
-      "learning_rate": 0.00019999878876011975,
-      "loss": 46.0,
-      "step": 20501
-    },
-    {
-      "epoch": 1.5675210734560467,
-      "grad_norm": 0.005175224505364895,
-      "learning_rate": 0.00019999878864189547,
-      "loss": 46.0,
-      "step": 20502
-    },
-    {
-      "epoch": 1.5675975304394365,
-      "grad_norm": 0.001720156753435731,
-      "learning_rate": 0.0001999987885236654,
-      "loss": 46.0,
-      "step": 20503
-    },
-    {
-      "epoch": 1.5676739874228263,
-      "grad_norm": 0.006137765012681484,
-      "learning_rate": 0.00019999878840542957,
-      "loss": 46.0,
-      "step": 20504
-    },
-    {
-      "epoch": 1.567750444406216,
-      "grad_norm": 0.004961061757057905,
-      "learning_rate": 0.00019999878828718797,
-      "loss": 46.0,
-      "step": 20505
-    },
-    {
-      "epoch": 1.5678269013896058,
-      "grad_norm": 0.001239399309270084,
-      "learning_rate": 0.0001999987881689406,
-      "loss": 46.0,
-      "step": 20506
-    },
-    {
-      "epoch": 1.5679033583729955,
-      "grad_norm": 0.0023358126636594534,
-      "learning_rate": 0.00019999878805068746,
-      "loss": 46.0,
-      "step": 20507
-    },
-    {
-      "epoch": 1.567979815356385,
-      "grad_norm": 0.0011175277177244425,
-      "learning_rate": 0.00019999878793242854,
-      "loss": 46.0,
-      "step": 20508
-    },
-    {
-      "epoch": 1.5680562723397748,
-      "grad_norm": 0.0007174051133915782,
-      "learning_rate": 0.00019999878781416387,
-      "loss": 46.0,
-      "step": 20509
-    },
-    {
-      "epoch": 1.5681327293231646,
-      "grad_norm": 0.00573556125164032,
-      "learning_rate": 0.00019999878769589343,
-      "loss": 46.0,
-      "step": 20510
-    },
-    {
-      "epoch": 1.5682091863065541,
-      "grad_norm": 0.0022795796394348145,
-      "learning_rate": 0.0001999987875776172,
-      "loss": 46.0,
-      "step": 20511
-    },
-    {
-      "epoch": 1.568285643289944,
-      "grad_norm": 0.0012685037218034267,
-      "learning_rate": 0.0001999987874593352,
-      "loss": 46.0,
-      "step": 20512
-    },
-    {
-      "epoch": 1.5683621002733337,
-      "grad_norm": 0.001142531749792397,
-      "learning_rate": 0.00019999878734104745,
-      "loss": 46.0,
-      "step": 20513
-    },
-    {
-      "epoch": 1.5684385572567234,
-      "grad_norm": 0.001449443749152124,
-      "learning_rate": 0.00019999878722275395,
-      "loss": 46.0,
-      "step": 20514
-    },
-    {
-      "epoch": 1.5685150142401132,
-      "grad_norm": 0.0059852758422493935,
-      "learning_rate": 0.00019999878710445464,
-      "loss": 46.0,
-      "step": 20515
-    },
-    {
-      "epoch": 1.568591471223503,
-      "grad_norm": 0.0008569424971938133,
-      "learning_rate": 0.00019999878698614957,
-      "loss": 46.0,
-      "step": 20516
-    },
-    {
-      "epoch": 1.5686679282068927,
-      "grad_norm": 0.0003648698329925537,
-      "learning_rate": 0.00019999878686783874,
-      "loss": 46.0,
-      "step": 20517
-    },
-    {
-      "epoch": 1.5687443851902825,
-      "grad_norm": 0.0007804068154655397,
-      "learning_rate": 0.00019999878674952214,
-      "loss": 46.0,
-      "step": 20518
-    },
-    {
-      "epoch": 1.568820842173672,
-      "grad_norm": 0.0008888784795999527,
-      "learning_rate": 0.00019999878663119977,
-      "loss": 46.0,
-      "step": 20519
-    },
-    {
-      "epoch": 1.5688972991570618,
-      "grad_norm": 0.001737418700940907,
-      "learning_rate": 0.00019999878651287163,
-      "loss": 46.0,
-      "step": 20520
-    },
-    {
-      "epoch": 1.5689737561404513,
-      "grad_norm": 0.0015959254233166575,
-      "learning_rate": 0.00019999878639453774,
-      "loss": 46.0,
-      "step": 20521
-    },
-    {
-      "epoch": 1.569050213123841,
-      "grad_norm": 0.0033062328584492207,
-      "learning_rate": 0.00019999878627619805,
-      "loss": 46.0,
-      "step": 20522
-    },
-    {
-      "epoch": 1.5691266701072308,
-      "grad_norm": 0.0009737692889757454,
-      "learning_rate": 0.00019999878615785258,
-      "loss": 46.0,
-      "step": 20523
-    },
-    {
-      "epoch": 1.5692031270906206,
-      "grad_norm": 0.0012240028008818626,
-      "learning_rate": 0.00019999878603950137,
-      "loss": 46.0,
-      "step": 20524
-    },
-    {
-      "epoch": 1.5692795840740104,
-      "grad_norm": 0.0020970734767615795,
-      "learning_rate": 0.00019999878592114439,
-      "loss": 46.0,
-      "step": 20525
-    },
-    {
-      "epoch": 1.5693560410574001,
-      "grad_norm": 0.002257255371659994,
-      "learning_rate": 0.00019999878580278163,
-      "loss": 46.0,
-      "step": 20526
-    },
-    {
-      "epoch": 1.5694324980407899,
-      "grad_norm": 0.006322106812149286,
-      "learning_rate": 0.0001999987856844131,
-      "loss": 46.0,
-      "step": 20527
-    },
-    {
-      "epoch": 1.5695089550241796,
-      "grad_norm": 0.0007832669652998447,
-      "learning_rate": 0.0001999987855660388,
-      "loss": 46.0,
-      "step": 20528
-    },
-    {
-      "epoch": 1.5695854120075694,
-      "grad_norm": 0.0039178491570055485,
-      "learning_rate": 0.00019999878544765874,
-      "loss": 46.0,
-      "step": 20529
-    },
-    {
-      "epoch": 1.569661868990959,
-      "grad_norm": 0.0010992736788466573,
-      "learning_rate": 0.0001999987853292729,
-      "loss": 46.0,
-      "step": 20530
-    },
-    {
-      "epoch": 1.5697383259743487,
-      "grad_norm": 0.0040085711516439915,
-      "learning_rate": 0.0001999987852108813,
-      "loss": 46.0,
-      "step": 20531
-    },
-    {
-      "epoch": 1.5698147829577382,
-      "grad_norm": 0.0010158132063224912,
-      "learning_rate": 0.00019999878509248392,
-      "loss": 46.0,
-      "step": 20532
-    },
-    {
-      "epoch": 1.569891239941128,
-      "grad_norm": 0.0010326748015359044,
-      "learning_rate": 0.0001999987849740808,
-      "loss": 46.0,
-      "step": 20533
-    },
-    {
-      "epoch": 1.5699676969245178,
-      "grad_norm": 0.0009951175889000297,
-      "learning_rate": 0.00019999878485567186,
-      "loss": 46.0,
-      "step": 20534
-    },
-    {
-      "epoch": 1.5700441539079075,
-      "grad_norm": 0.0010285226162523031,
-      "learning_rate": 0.0001999987847372572,
-      "loss": 46.0,
-      "step": 20535
-    },
-    {
-      "epoch": 1.5701206108912973,
-      "grad_norm": 0.0018726319540292025,
-      "learning_rate": 0.00019999878461883673,
-      "loss": 46.0,
-      "step": 20536
-    },
-    {
-      "epoch": 1.570197067874687,
-      "grad_norm": 0.0010387040674686432,
-      "learning_rate": 0.00019999878450041052,
-      "loss": 46.0,
-      "step": 20537
-    },
-    {
-      "epoch": 1.5702735248580768,
-      "grad_norm": 0.003290874185040593,
-      "learning_rate": 0.0001999987843819785,
-      "loss": 46.0,
-      "step": 20538
-    },
-    {
-      "epoch": 1.5703499818414666,
-      "grad_norm": 0.0007973746978677809,
-      "learning_rate": 0.00019999878426354078,
-      "loss": 46.0,
-      "step": 20539
-    },
-    {
-      "epoch": 1.5704264388248563,
-      "grad_norm": 0.0008850062149576843,
-      "learning_rate": 0.00019999878414509725,
-      "loss": 46.0,
-      "step": 20540
-    },
-    {
-      "epoch": 1.5705028958082459,
-      "grad_norm": 0.0010700584389269352,
-      "learning_rate": 0.00019999878402664794,
-      "loss": 46.0,
-      "step": 20541
-    },
-    {
-      "epoch": 1.5705793527916356,
-      "grad_norm": 0.00105974730104208,
-      "learning_rate": 0.00019999878390819287,
-      "loss": 46.0,
-      "step": 20542
-    },
-    {
-      "epoch": 1.5706558097750252,
-      "grad_norm": 0.002016504528000951,
-      "learning_rate": 0.00019999878378973204,
-      "loss": 46.0,
-      "step": 20543
-    },
-    {
-      "epoch": 1.570732266758415,
-      "grad_norm": 0.0009395172819495201,
-      "learning_rate": 0.00019999878367126545,
-      "loss": 46.0,
-      "step": 20544
-    },
-    {
-      "epoch": 1.5708087237418047,
-      "grad_norm": 0.0008172324160113931,
-      "learning_rate": 0.00019999878355279305,
-      "loss": 46.0,
-      "step": 20545
-    },
-    {
-      "epoch": 1.5708851807251945,
-      "grad_norm": 0.0006112594273872674,
-      "learning_rate": 0.0001999987834343149,
-      "loss": 46.0,
-      "step": 20546
-    },
-    {
-      "epoch": 1.5709616377085842,
-      "grad_norm": 0.0021788757294416428,
-      "learning_rate": 0.000199998783315831,
-      "loss": 46.0,
-      "step": 20547
-    },
-    {
-      "epoch": 1.571038094691974,
-      "grad_norm": 0.00084165227599442,
-      "learning_rate": 0.00019999878319734132,
-      "loss": 46.0,
-      "step": 20548
-    },
-    {
-      "epoch": 1.5711145516753637,
-      "grad_norm": 0.0029515374917536974,
-      "learning_rate": 0.00019999878307884586,
-      "loss": 46.0,
-      "step": 20549
-    },
-    {
-      "epoch": 1.5711910086587535,
-      "grad_norm": 0.0006206241669133306,
-      "learning_rate": 0.00019999878296034465,
-      "loss": 46.0,
-      "step": 20550
-    },
-    {
-      "epoch": 1.571267465642143,
-      "grad_norm": 0.0008314374717883766,
-      "learning_rate": 0.00019999878284183764,
-      "loss": 46.0,
-      "step": 20551
-    },
-    {
-      "epoch": 1.5713439226255328,
-      "grad_norm": 0.0006867795600555837,
-      "learning_rate": 0.00019999878272332485,
-      "loss": 46.0,
-      "step": 20552
-    },
-    {
-      "epoch": 1.5714203796089226,
-      "grad_norm": 0.0009752354235388339,
-      "learning_rate": 0.00019999878260480632,
-      "loss": 46.0,
-      "step": 20553
-    },
-    {
-      "epoch": 1.571496836592312,
-      "grad_norm": 0.0022557354532182217,
-      "learning_rate": 0.00019999878248628202,
-      "loss": 46.0,
-      "step": 20554
-    },
-    {
-      "epoch": 1.5715732935757019,
-      "grad_norm": 0.003024192526936531,
-      "learning_rate": 0.00019999878236775197,
-      "loss": 46.0,
-      "step": 20555
-    },
-    {
-      "epoch": 1.5716497505590916,
-      "grad_norm": 0.04782325029373169,
-      "learning_rate": 0.00019999878224921615,
-      "loss": 46.0,
-      "step": 20556
-    },
-    {
-      "epoch": 1.5717262075424814,
-      "grad_norm": 0.002820422640070319,
-      "learning_rate": 0.00019999878213067452,
-      "loss": 46.0,
-      "step": 20557
-    },
-    {
-      "epoch": 1.5718026645258711,
-      "grad_norm": 0.001744549022987485,
-      "learning_rate": 0.00019999878201212715,
-      "loss": 46.0,
-      "step": 20558
-    },
-    {
-      "epoch": 1.571879121509261,
-      "grad_norm": 0.002101277466863394,
-      "learning_rate": 0.000199998781893574,
-      "loss": 46.0,
-      "step": 20559
-    },
-    {
-      "epoch": 1.5719555784926507,
-      "grad_norm": 0.0019647928420454264,
-      "learning_rate": 0.00019999878177501506,
-      "loss": 46.0,
-      "step": 20560
-    },
-    {
-      "epoch": 1.5720320354760404,
-      "grad_norm": 0.0008673628326505423,
-      "learning_rate": 0.0001999987816564504,
-      "loss": 46.0,
-      "step": 20561
-    },
-    {
-      "epoch": 1.57210849245943,
-      "grad_norm": 0.002075694967061281,
-      "learning_rate": 0.00019999878153787994,
-      "loss": 46.0,
-      "step": 20562
-    },
-    {
-      "epoch": 1.5721849494428197,
-      "grad_norm": 0.002163885161280632,
-      "learning_rate": 0.0001999987814193037,
-      "loss": 46.0,
-      "step": 20563
-    },
-    {
-      "epoch": 1.5722614064262095,
-      "grad_norm": 0.0017930259928107262,
-      "learning_rate": 0.00019999878130072172,
-      "loss": 46.0,
-      "step": 20564
-    },
-    {
-      "epoch": 1.572337863409599,
-      "grad_norm": 0.004732145927846432,
-      "learning_rate": 0.00019999878118213396,
-      "loss": 46.0,
-      "step": 20565
-    },
-    {
-      "epoch": 1.5724143203929888,
-      "grad_norm": 0.0011254495475441217,
-      "learning_rate": 0.00019999878106354043,
-      "loss": 46.0,
-      "step": 20566
-    },
-    {
-      "epoch": 1.5724907773763785,
-      "grad_norm": 0.004453013651072979,
-      "learning_rate": 0.0001999987809449411,
-      "loss": 46.0,
-      "step": 20567
-    },
-    {
-      "epoch": 1.5725672343597683,
-      "grad_norm": 0.003936637192964554,
-      "learning_rate": 0.00019999878082633605,
-      "loss": 46.0,
-      "step": 20568
-    },
-    {
-      "epoch": 1.572643691343158,
-      "grad_norm": 0.00412857998162508,
-      "learning_rate": 0.00019999878070772523,
-      "loss": 46.0,
-      "step": 20569
-    },
-    {
-      "epoch": 1.5727201483265478,
-      "grad_norm": 0.0025975273456424475,
-      "learning_rate": 0.0001999987805891086,
-      "loss": 46.0,
-      "step": 20570
-    },
-    {
-      "epoch": 1.5727966053099376,
-      "grad_norm": 0.004237269517034292,
-      "learning_rate": 0.0001999987804704862,
-      "loss": 46.0,
-      "step": 20571
-    },
-    {
-      "epoch": 1.5728730622933274,
-      "grad_norm": 0.0011645270278677344,
-      "learning_rate": 0.00019999878035185806,
-      "loss": 46.0,
-      "step": 20572
-    },
-    {
-      "epoch": 1.572949519276717,
-      "grad_norm": 0.0016765110194683075,
-      "learning_rate": 0.00019999878023322417,
-      "loss": 46.0,
-      "step": 20573
-    },
-    {
-      "epoch": 1.5730259762601067,
-      "grad_norm": 0.0006472570239566267,
-      "learning_rate": 0.00019999878011458445,
-      "loss": 46.0,
-      "step": 20574
-    },
-    {
-      "epoch": 1.5731024332434964,
-      "grad_norm": 0.000870082585606724,
-      "learning_rate": 0.000199998779995939,
-      "loss": 46.0,
-      "step": 20575
-    },
-    {
-      "epoch": 1.573178890226886,
-      "grad_norm": 0.0007107230485416949,
-      "learning_rate": 0.00019999877987728775,
-      "loss": 46.0,
-      "step": 20576
-    },
-    {
-      "epoch": 1.5732553472102757,
-      "grad_norm": 0.0020948899909853935,
-      "learning_rate": 0.00019999877975863077,
-      "loss": 46.0,
-      "step": 20577
-    },
-    {
-      "epoch": 1.5733318041936655,
-      "grad_norm": 0.0005416327039711177,
-      "learning_rate": 0.00019999877963996801,
-      "loss": 46.0,
-      "step": 20578
-    },
-    {
-      "epoch": 1.5734082611770552,
-      "grad_norm": 0.001961984671652317,
-      "learning_rate": 0.00019999877952129946,
-      "loss": 46.0,
-      "step": 20579
-    },
-    {
-      "epoch": 1.573484718160445,
-      "grad_norm": 0.0012567578814923763,
-      "learning_rate": 0.00019999877940262518,
-      "loss": 46.0,
-      "step": 20580
-    },
-    {
-      "epoch": 1.5735611751438348,
-      "grad_norm": 0.009453502483665943,
-      "learning_rate": 0.0001999987792839451,
-      "loss": 46.0,
-      "step": 20581
-    },
-    {
-      "epoch": 1.5736376321272245,
-      "grad_norm": 0.00041271818918175995,
-      "learning_rate": 0.00019999877916525925,
-      "loss": 46.0,
-      "step": 20582
-    },
-    {
-      "epoch": 1.5737140891106143,
-      "grad_norm": 0.0016757945995777845,
-      "learning_rate": 0.00019999877904656766,
-      "loss": 46.0,
-      "step": 20583
-    },
-    {
-      "epoch": 1.5737905460940038,
-      "grad_norm": 0.00207105977460742,
-      "learning_rate": 0.00019999877892787023,
-      "loss": 46.0,
-      "step": 20584
-    },
-    {
-      "epoch": 1.5738670030773936,
-      "grad_norm": 0.0008442188263870776,
-      "learning_rate": 0.00019999877880916712,
-      "loss": 46.0,
-      "step": 20585
-    },
-    {
-      "epoch": 1.5739434600607833,
-      "grad_norm": 0.0008008594159036875,
-      "learning_rate": 0.0001999987786904582,
-      "loss": 46.0,
-      "step": 20586
-    },
-    {
-      "epoch": 1.5740199170441729,
-      "grad_norm": 0.0011094157816842198,
-      "learning_rate": 0.00019999877857174349,
-      "loss": 46.0,
-      "step": 20587
-    },
-    {
-      "epoch": 1.5740963740275626,
-      "grad_norm": 0.0004447736428119242,
-      "learning_rate": 0.00019999877845302305,
-      "loss": 46.0,
-      "step": 20588
-    },
-    {
-      "epoch": 1.5741728310109524,
-      "grad_norm": 0.0006118929595686495,
-      "learning_rate": 0.0001999987783342968,
-      "loss": 46.0,
-      "step": 20589
-    },
-    {
-      "epoch": 1.5742492879943422,
-      "grad_norm": 0.0006569484248757362,
-      "learning_rate": 0.0001999987782155648,
-      "loss": 46.0,
-      "step": 20590
-    },
-    {
-      "epoch": 1.574325744977732,
-      "grad_norm": 0.004677473101764917,
-      "learning_rate": 0.00019999877809682705,
-      "loss": 46.0,
-      "step": 20591
-    },
-    {
-      "epoch": 1.5744022019611217,
-      "grad_norm": 0.0007658423273824155,
-      "learning_rate": 0.00019999877797808352,
-      "loss": 46.0,
-      "step": 20592
-    },
-    {
-      "epoch": 1.5744786589445114,
-      "grad_norm": 0.002214341424405575,
-      "learning_rate": 0.0001999987778593342,
-      "loss": 46.0,
-      "step": 20593
-    },
-    {
-      "epoch": 1.5745551159279012,
-      "grad_norm": 0.0021820280235260725,
-      "learning_rate": 0.00019999877774057914,
-      "loss": 46.0,
-      "step": 20594
-    },
-    {
-      "epoch": 1.5746315729112907,
-      "grad_norm": 0.0028867812361568213,
-      "learning_rate": 0.00019999877762181826,
-      "loss": 46.0,
-      "step": 20595
-    },
-    {
-      "epoch": 1.5747080298946805,
-      "grad_norm": 0.0019501956412568688,
-      "learning_rate": 0.00019999877750305167,
-      "loss": 46.0,
-      "step": 20596
-    },
-    {
-      "epoch": 1.5747844868780703,
-      "grad_norm": 0.010670505464076996,
-      "learning_rate": 0.00019999877738427927,
-      "loss": 46.0,
-      "step": 20597
-    },
-    {
-      "epoch": 1.5748609438614598,
-      "grad_norm": 0.0008644146146252751,
-      "learning_rate": 0.00019999877726550113,
-      "loss": 46.0,
-      "step": 20598
-    },
-    {
-      "epoch": 1.5749374008448496,
-      "grad_norm": 0.0017332402057945728,
-      "learning_rate": 0.0001999987771467172,
-      "loss": 46.0,
-      "step": 20599
-    },
-    {
-      "epoch": 1.5750138578282393,
-      "grad_norm": 0.005928058177232742,
-      "learning_rate": 0.00019999877702792752,
-      "loss": 46.0,
-      "step": 20600
-    },
-    {
-      "epoch": 1.575090314811629,
-      "grad_norm": 0.0026727814693003893,
-      "learning_rate": 0.00019999877690913206,
-      "loss": 46.0,
-      "step": 20601
-    },
-    {
-      "epoch": 1.5751667717950188,
-      "grad_norm": 0.0022742270957678556,
-      "learning_rate": 0.00019999877679033082,
-      "loss": 46.0,
-      "step": 20602
-    },
-    {
-      "epoch": 1.5752432287784086,
-      "grad_norm": 0.0019309110939502716,
-      "learning_rate": 0.0001999987766715238,
-      "loss": 46.0,
-      "step": 20603
-    },
-    {
-      "epoch": 1.5753196857617984,
-      "grad_norm": 0.0012425679014995694,
-      "learning_rate": 0.00019999877655271106,
-      "loss": 46.0,
-      "step": 20604
-    },
-    {
-      "epoch": 1.5753961427451881,
-      "grad_norm": 0.0011150309583172202,
-      "learning_rate": 0.0001999987764338925,
-      "loss": 46.0,
-      "step": 20605
-    },
-    {
-      "epoch": 1.5754725997285777,
-      "grad_norm": 0.007273803930729628,
-      "learning_rate": 0.0001999987763150682,
-      "loss": 46.0,
-      "step": 20606
-    },
-    {
-      "epoch": 1.5755490567119674,
-      "grad_norm": 0.002322570187970996,
-      "learning_rate": 0.00019999877619623812,
-      "loss": 46.0,
-      "step": 20607
-    },
-    {
-      "epoch": 1.5756255136953572,
-      "grad_norm": 0.001206431072205305,
-      "learning_rate": 0.00019999877607740227,
-      "loss": 46.0,
-      "step": 20608
-    },
-    {
-      "epoch": 1.5757019706787467,
-      "grad_norm": 0.0004714473325293511,
-      "learning_rate": 0.00019999877595856065,
-      "loss": 46.0,
-      "step": 20609
-    },
-    {
-      "epoch": 1.5757784276621365,
-      "grad_norm": 0.0010275548556819558,
-      "learning_rate": 0.00019999877583971328,
-      "loss": 46.0,
-      "step": 20610
-    },
-    {
-      "epoch": 1.5758548846455263,
-      "grad_norm": 0.0010984393302351236,
-      "learning_rate": 0.0001999987757208601,
-      "loss": 46.0,
-      "step": 20611
-    },
-    {
-      "epoch": 1.575931341628916,
-      "grad_norm": 0.0033590730745345354,
-      "learning_rate": 0.0001999987756020012,
-      "loss": 46.0,
-      "step": 20612
-    },
-    {
-      "epoch": 1.5760077986123058,
-      "grad_norm": 0.0015435887034982443,
-      "learning_rate": 0.0001999987754831365,
-      "loss": 46.0,
-      "step": 20613
-    },
-    {
-      "epoch": 1.5760842555956955,
-      "grad_norm": 0.0014658686704933643,
-      "learning_rate": 0.00019999877536426604,
-      "loss": 46.0,
-      "step": 20614
-    },
-    {
-      "epoch": 1.5761607125790853,
-      "grad_norm": 0.0015555268619209528,
-      "learning_rate": 0.00019999877524538978,
-      "loss": 46.0,
-      "step": 20615
-    },
-    {
-      "epoch": 1.576237169562475,
-      "grad_norm": 0.006283890455961227,
-      "learning_rate": 0.0001999987751265078,
-      "loss": 46.0,
-      "step": 20616
-    },
-    {
-      "epoch": 1.5763136265458646,
-      "grad_norm": 0.0006554757128469646,
-      "learning_rate": 0.00019999877500762004,
-      "loss": 46.0,
-      "step": 20617
-    },
-    {
-      "epoch": 1.5763900835292544,
-      "grad_norm": 0.0010277057299390435,
-      "learning_rate": 0.00019999877488872646,
-      "loss": 46.0,
-      "step": 20618
-    },
-    {
-      "epoch": 1.5764665405126441,
-      "grad_norm": 0.01042028609663248,
-      "learning_rate": 0.00019999877476982718,
-      "loss": 46.0,
-      "step": 20619
-    },
-    {
-      "epoch": 1.5765429974960337,
-      "grad_norm": 0.0014580290298908949,
-      "learning_rate": 0.00019999877465092208,
-      "loss": 46.0,
-      "step": 20620
-    },
-    {
-      "epoch": 1.5766194544794234,
-      "grad_norm": 0.000923613493796438,
-      "learning_rate": 0.00019999877453201123,
-      "loss": 46.0,
-      "step": 20621
-    },
-    {
-      "epoch": 1.5766959114628132,
-      "grad_norm": 0.0004414485301822424,
-      "learning_rate": 0.00019999877441309464,
-      "loss": 46.0,
-      "step": 20622
-    },
-    {
-      "epoch": 1.576772368446203,
-      "grad_norm": 0.0008916439837776124,
-      "learning_rate": 0.00019999877429417224,
-      "loss": 46.0,
-      "step": 20623
-    },
-    {
-      "epoch": 1.5768488254295927,
-      "grad_norm": 0.0014385569375008345,
-      "learning_rate": 0.0001999987741752441,
-      "loss": 46.0,
-      "step": 20624
-    },
-    {
-      "epoch": 1.5769252824129825,
-      "grad_norm": 0.0008587023476138711,
-      "learning_rate": 0.00019999877405631016,
-      "loss": 46.0,
-      "step": 20625
-    },
-    {
-      "epoch": 1.5770017393963722,
-      "grad_norm": 0.0012934551341459155,
-      "learning_rate": 0.00019999877393737044,
-      "loss": 46.0,
-      "step": 20626
-    },
-    {
-      "epoch": 1.577078196379762,
-      "grad_norm": 0.004118271172046661,
-      "learning_rate": 0.00019999877381842498,
-      "loss": 46.0,
-      "step": 20627
-    },
-    {
-      "epoch": 1.5771546533631515,
-      "grad_norm": 0.003280634991824627,
-      "learning_rate": 0.00019999877369947377,
-      "loss": 46.0,
-      "step": 20628
-    },
-    {
-      "epoch": 1.5772311103465413,
-      "grad_norm": 0.0030874786898493767,
-      "learning_rate": 0.00019999877358051676,
-      "loss": 46.0,
-      "step": 20629
-    },
-    {
-      "epoch": 1.577307567329931,
-      "grad_norm": 0.0005862012039870024,
-      "learning_rate": 0.00019999877346155398,
-      "loss": 46.0,
-      "step": 20630
-    },
-    {
-      "epoch": 1.5773840243133206,
-      "grad_norm": 0.0010861713672056794,
-      "learning_rate": 0.00019999877334258542,
-      "loss": 46.0,
-      "step": 20631
-    },
-    {
-      "epoch": 1.5774604812967103,
-      "grad_norm": 0.0014244478661566973,
-      "learning_rate": 0.00019999877322361115,
-      "loss": 46.0,
-      "step": 20632
-    },
-    {
-      "epoch": 1.5775369382801,
-      "grad_norm": 0.0007127447752282023,
-      "learning_rate": 0.00019999877310463105,
-      "loss": 46.0,
-      "step": 20633
-    },
-    {
-      "epoch": 1.5776133952634899,
-      "grad_norm": 0.0009441408910788596,
-      "learning_rate": 0.00019999877298564517,
-      "loss": 46.0,
-      "step": 20634
-    },
-    {
-      "epoch": 1.5776898522468796,
-      "grad_norm": 0.0007822144543752074,
-      "learning_rate": 0.00019999877286665355,
-      "loss": 46.0,
-      "step": 20635
-    },
-    {
-      "epoch": 1.5777663092302694,
-      "grad_norm": 0.0006972999544814229,
-      "learning_rate": 0.00019999877274765618,
-      "loss": 46.0,
-      "step": 20636
-    },
-    {
-      "epoch": 1.5778427662136592,
-      "grad_norm": 0.0010089692659676075,
-      "learning_rate": 0.000199998772628653,
-      "loss": 46.0,
-      "step": 20637
-    },
-    {
-      "epoch": 1.577919223197049,
-      "grad_norm": 0.000784090138040483,
-      "learning_rate": 0.0001999987725096441,
-      "loss": 46.0,
-      "step": 20638
-    },
-    {
-      "epoch": 1.5779956801804385,
-      "grad_norm": 0.0022438494488596916,
-      "learning_rate": 0.0001999987723906294,
-      "loss": 46.0,
-      "step": 20639
-    },
-    {
-      "epoch": 1.5780721371638282,
-      "grad_norm": 0.0030220877379179,
-      "learning_rate": 0.00019999877227160892,
-      "loss": 46.0,
-      "step": 20640
-    },
-    {
-      "epoch": 1.578148594147218,
-      "grad_norm": 0.011319195851683617,
-      "learning_rate": 0.00019999877215258268,
-      "loss": 46.0,
-      "step": 20641
-    },
-    {
-      "epoch": 1.5782250511306075,
-      "grad_norm": 0.010835738852620125,
-      "learning_rate": 0.00019999877203355067,
-      "loss": 46.0,
-      "step": 20642
-    },
-    {
-      "epoch": 1.5783015081139973,
-      "grad_norm": 0.0028837614227086306,
-      "learning_rate": 0.00019999877191451292,
-      "loss": 46.0,
-      "step": 20643
-    },
-    {
-      "epoch": 1.578377965097387,
-      "grad_norm": 0.00233117095194757,
-      "learning_rate": 0.00019999877179546936,
-      "loss": 46.0,
-      "step": 20644
-    },
-    {
-      "epoch": 1.5784544220807768,
-      "grad_norm": 0.000977079151198268,
-      "learning_rate": 0.00019999877167642006,
-      "loss": 46.0,
-      "step": 20645
-    },
-    {
-      "epoch": 1.5785308790641666,
-      "grad_norm": 0.0034887176007032394,
-      "learning_rate": 0.00019999877155736496,
-      "loss": 46.0,
-      "step": 20646
-    },
-    {
-      "epoch": 1.5786073360475563,
-      "grad_norm": 0.0014209384098649025,
-      "learning_rate": 0.0001999987714383041,
-      "loss": 46.0,
-      "step": 20647
-    },
-    {
-      "epoch": 1.578683793030946,
-      "grad_norm": 0.00106794573366642,
-      "learning_rate": 0.0001999987713192375,
-      "loss": 46.0,
-      "step": 20648
-    },
-    {
-      "epoch": 1.5787602500143358,
-      "grad_norm": 0.0006753666675649583,
-      "learning_rate": 0.0001999987712001651,
-      "loss": 46.0,
-      "step": 20649
-    },
-    {
-      "epoch": 1.5788367069977254,
-      "grad_norm": 0.003812754061073065,
-      "learning_rate": 0.00019999877108108693,
-      "loss": 46.0,
-      "step": 20650
-    },
-    {
-      "epoch": 1.5789131639811151,
-      "grad_norm": 0.0013617555378004909,
-      "learning_rate": 0.00019999877096200298,
-      "loss": 46.0,
-      "step": 20651
-    },
-    {
-      "epoch": 1.5789896209645047,
-      "grad_norm": 0.005027902778238058,
-      "learning_rate": 0.00019999877084291332,
-      "loss": 46.0,
-      "step": 20652
-    },
-    {
-      "epoch": 1.5790660779478944,
-      "grad_norm": 0.0022772192023694515,
-      "learning_rate": 0.00019999877072381783,
-      "loss": 46.0,
-      "step": 20653
-    },
-    {
-      "epoch": 1.5791425349312842,
-      "grad_norm": 0.0010735277319326997,
-      "learning_rate": 0.0001999987706047166,
-      "loss": 46.0,
-      "step": 20654
-    },
-    {
-      "epoch": 1.579218991914674,
-      "grad_norm": 0.002673623850569129,
-      "learning_rate": 0.00019999877048560956,
-      "loss": 46.0,
-      "step": 20655
-    },
-    {
-      "epoch": 1.5792954488980637,
-      "grad_norm": 0.0008719094912521541,
-      "learning_rate": 0.0001999987703664968,
-      "loss": 46.0,
-      "step": 20656
-    },
-    {
-      "epoch": 1.5793719058814535,
-      "grad_norm": 0.0009952655527740717,
-      "learning_rate": 0.00019999877024737828,
-      "loss": 46.0,
-      "step": 20657
-    },
-    {
-      "epoch": 1.5794483628648432,
-      "grad_norm": 0.002484864555299282,
-      "learning_rate": 0.00019999877012825395,
-      "loss": 46.0,
-      "step": 20658
-    },
-    {
-      "epoch": 1.579524819848233,
-      "grad_norm": 0.00041910645086318254,
-      "learning_rate": 0.00019999877000912385,
-      "loss": 46.0,
-      "step": 20659
-    },
-    {
-      "epoch": 1.5796012768316228,
-      "grad_norm": 0.0013854647986590862,
-      "learning_rate": 0.000199998769889988,
-      "loss": 46.0,
-      "step": 20660
-    },
-    {
-      "epoch": 1.5796777338150123,
-      "grad_norm": 0.0016841775504872203,
-      "learning_rate": 0.0001999987697708464,
-      "loss": 46.0,
-      "step": 20661
-    },
-    {
-      "epoch": 1.579754190798402,
-      "grad_norm": 0.003285430371761322,
-      "learning_rate": 0.00019999876965169899,
-      "loss": 46.0,
-      "step": 20662
-    },
-    {
-      "epoch": 1.5798306477817916,
-      "grad_norm": 0.0012333591002970934,
-      "learning_rate": 0.00019999876953254582,
-      "loss": 46.0,
-      "step": 20663
-    },
-    {
-      "epoch": 1.5799071047651814,
-      "grad_norm": 0.0013427617959678173,
-      "learning_rate": 0.00019999876941338688,
-      "loss": 46.0,
-      "step": 20664
-    },
-    {
-      "epoch": 1.5799835617485711,
-      "grad_norm": 0.000605247390922159,
-      "learning_rate": 0.0001999987692942222,
-      "loss": 46.0,
-      "step": 20665
-    },
-    {
-      "epoch": 1.580060018731961,
-      "grad_norm": 0.0016831998946145177,
-      "learning_rate": 0.00019999876917505173,
-      "loss": 46.0,
-      "step": 20666
-    },
-    {
-      "epoch": 1.5801364757153507,
-      "grad_norm": 0.002568417228758335,
-      "learning_rate": 0.00019999876905587547,
-      "loss": 46.0,
-      "step": 20667
-    },
-    {
-      "epoch": 1.5802129326987404,
-      "grad_norm": 0.000863865832798183,
-      "learning_rate": 0.00019999876893669346,
-      "loss": 46.0,
-      "step": 20668
-    },
-    {
-      "epoch": 1.5802893896821302,
-      "grad_norm": 0.001574463676661253,
-      "learning_rate": 0.0001999987688175057,
-      "loss": 46.0,
-      "step": 20669
-    },
-    {
-      "epoch": 1.58036584666552,
-      "grad_norm": 0.0005493666394613683,
-      "learning_rate": 0.00019999876869831212,
-      "loss": 46.0,
-      "step": 20670
-    },
-    {
-      "epoch": 1.5804423036489097,
-      "grad_norm": 0.0020681836176663637,
-      "learning_rate": 0.00019999876857911282,
-      "loss": 46.0,
-      "step": 20671
-    },
-    {
-      "epoch": 1.5805187606322992,
-      "grad_norm": 0.00034496409352868795,
-      "learning_rate": 0.00019999876845990772,
-      "loss": 46.0,
-      "step": 20672
-    },
-    {
-      "epoch": 1.580595217615689,
-      "grad_norm": 0.0012792296474799514,
-      "learning_rate": 0.0001999987683406969,
-      "loss": 46.0,
-      "step": 20673
-    },
-    {
-      "epoch": 1.5806716745990785,
-      "grad_norm": 0.0008240740280598402,
-      "learning_rate": 0.00019999876822148025,
-      "loss": 46.0,
-      "step": 20674
-    },
-    {
-      "epoch": 1.5807481315824683,
-      "grad_norm": 0.0009947834769263864,
-      "learning_rate": 0.00019999876810225783,
-      "loss": 46.0,
-      "step": 20675
-    },
-    {
-      "epoch": 1.580824588565858,
-      "grad_norm": 0.000829110445920378,
-      "learning_rate": 0.0001999987679830297,
-      "loss": 46.0,
-      "step": 20676
-    },
-    {
-      "epoch": 1.5809010455492478,
-      "grad_norm": 0.0026388815604150295,
-      "learning_rate": 0.00019999876786379575,
-      "loss": 46.0,
-      "step": 20677
-    },
-    {
-      "epoch": 1.5809775025326376,
-      "grad_norm": 0.001018973533064127,
-      "learning_rate": 0.00019999876774455606,
-      "loss": 46.0,
-      "step": 20678
-    },
-    {
-      "epoch": 1.5810539595160273,
-      "grad_norm": 0.000981897464953363,
-      "learning_rate": 0.00019999876762531054,
-      "loss": 46.0,
-      "step": 20679
-    },
-    {
-      "epoch": 1.581130416499417,
-      "grad_norm": 0.001099984860047698,
-      "learning_rate": 0.00019999876750605934,
-      "loss": 46.0,
-      "step": 20680
-    },
-    {
-      "epoch": 1.5812068734828069,
-      "grad_norm": 0.001517529017291963,
-      "learning_rate": 0.00019999876738680233,
-      "loss": 46.0,
-      "step": 20681
-    },
-    {
-      "epoch": 1.5812833304661964,
-      "grad_norm": 0.0002112446236424148,
-      "learning_rate": 0.00019999876726753955,
-      "loss": 46.0,
-      "step": 20682
-    },
-    {
-      "epoch": 1.5813597874495862,
-      "grad_norm": 0.0005338817136362195,
-      "learning_rate": 0.000199998767148271,
-      "loss": 46.0,
-      "step": 20683
-    },
-    {
-      "epoch": 1.581436244432976,
-      "grad_norm": 0.00727510591968894,
-      "learning_rate": 0.00019999876702899667,
-      "loss": 46.0,
-      "step": 20684
-    },
-    {
-      "epoch": 1.5815127014163655,
-      "grad_norm": 0.0014092057244852185,
-      "learning_rate": 0.00019999876690971656,
-      "loss": 46.0,
-      "step": 20685
-    },
-    {
-      "epoch": 1.5815891583997552,
-      "grad_norm": 0.0008078165119513869,
-      "learning_rate": 0.0001999987667904307,
-      "loss": 46.0,
-      "step": 20686
-    },
-    {
-      "epoch": 1.581665615383145,
-      "grad_norm": 0.00041847664397209883,
-      "learning_rate": 0.00019999876667113907,
-      "loss": 46.0,
-      "step": 20687
-    },
-    {
-      "epoch": 1.5817420723665347,
-      "grad_norm": 0.0012829189654439688,
-      "learning_rate": 0.00019999876655184167,
-      "loss": 46.0,
-      "step": 20688
-    },
-    {
-      "epoch": 1.5818185293499245,
-      "grad_norm": 0.0007831389084458351,
-      "learning_rate": 0.0001999987664325385,
-      "loss": 46.0,
-      "step": 20689
-    },
-    {
-      "epoch": 1.5818949863333143,
-      "grad_norm": 0.0027099738363176584,
-      "learning_rate": 0.0001999987663132296,
-      "loss": 46.0,
-      "step": 20690
-    },
-    {
-      "epoch": 1.581971443316704,
-      "grad_norm": 0.002286385279148817,
-      "learning_rate": 0.00019999876619391488,
-      "loss": 46.0,
-      "step": 20691
-    },
-    {
-      "epoch": 1.5820479003000938,
-      "grad_norm": 0.0015743793919682503,
-      "learning_rate": 0.0001999987660745944,
-      "loss": 46.0,
-      "step": 20692
-    },
-    {
-      "epoch": 1.5821243572834833,
-      "grad_norm": 0.0008356235921382904,
-      "learning_rate": 0.00019999876595526813,
-      "loss": 46.0,
-      "step": 20693
-    },
-    {
-      "epoch": 1.582200814266873,
-      "grad_norm": 0.0007721131551079452,
-      "learning_rate": 0.00019999876583593612,
-      "loss": 46.0,
-      "step": 20694
-    },
-    {
-      "epoch": 1.5822772712502629,
-      "grad_norm": 0.0007326969061978161,
-      "learning_rate": 0.00019999876571659834,
-      "loss": 46.0,
-      "step": 20695
-    },
-    {
-      "epoch": 1.5823537282336524,
-      "grad_norm": 0.0013834059936925769,
-      "learning_rate": 0.0001999987655972548,
-      "loss": 46.0,
-      "step": 20696
-    },
-    {
-      "epoch": 1.5824301852170422,
-      "grad_norm": 0.0014844231773167849,
-      "learning_rate": 0.0001999987654779055,
-      "loss": 46.0,
-      "step": 20697
-    },
-    {
-      "epoch": 1.582506642200432,
-      "grad_norm": 0.0014713979326188564,
-      "learning_rate": 0.00019999876535855039,
-      "loss": 46.0,
-      "step": 20698
-    },
-    {
-      "epoch": 1.5825830991838217,
-      "grad_norm": 0.003678596578538418,
-      "learning_rate": 0.0001999987652391895,
-      "loss": 46.0,
-      "step": 20699
-    },
-    {
-      "epoch": 1.5826595561672114,
-      "grad_norm": 0.0004370861279312521,
-      "learning_rate": 0.0001999987651198229,
-      "loss": 46.0,
-      "step": 20700
-    },
-    {
-      "epoch": 1.5827360131506012,
-      "grad_norm": 0.0005486364243552089,
-      "learning_rate": 0.0001999987650004505,
-      "loss": 46.0,
-      "step": 20701
-    },
-    {
-      "epoch": 1.582812470133991,
-      "grad_norm": 0.00366935133934021,
-      "learning_rate": 0.00019999876488107233,
-      "loss": 46.0,
-      "step": 20702
-    },
-    {
-      "epoch": 1.5828889271173807,
-      "grad_norm": 0.0012095060665160418,
-      "learning_rate": 0.0001999987647616884,
-      "loss": 46.0,
-      "step": 20703
-    },
-    {
-      "epoch": 1.5829653841007703,
-      "grad_norm": 0.001610441948287189,
-      "learning_rate": 0.00019999876464229868,
-      "loss": 46.0,
-      "step": 20704
-    },
-    {
-      "epoch": 1.58304184108416,
-      "grad_norm": 0.0005798356723971665,
-      "learning_rate": 0.00019999876452290321,
-      "loss": 46.0,
-      "step": 20705
-    },
-    {
-      "epoch": 1.5831182980675498,
-      "grad_norm": 0.005967284552752972,
-      "learning_rate": 0.00019999876440350195,
-      "loss": 46.0,
-      "step": 20706
-    },
-    {
-      "epoch": 1.5831947550509393,
-      "grad_norm": 0.004606550559401512,
-      "learning_rate": 0.00019999876428409495,
-      "loss": 46.0,
-      "step": 20707
-    },
-    {
-      "epoch": 1.583271212034329,
-      "grad_norm": 0.00360392895527184,
-      "learning_rate": 0.00019999876416468214,
-      "loss": 46.0,
-      "step": 20708
-    },
-    {
-      "epoch": 1.5833476690177188,
-      "grad_norm": 0.0006789357867091894,
-      "learning_rate": 0.00019999876404526359,
-      "loss": 46.0,
-      "step": 20709
-    },
-    {
-      "epoch": 1.5834241260011086,
-      "grad_norm": 0.0015642885118722916,
-      "learning_rate": 0.00019999876392583926,
-      "loss": 46.0,
-      "step": 20710
-    },
-    {
-      "epoch": 1.5835005829844984,
-      "grad_norm": 0.00783013366162777,
-      "learning_rate": 0.00019999876380640919,
-      "loss": 46.0,
-      "step": 20711
-    },
-    {
-      "epoch": 1.5835770399678881,
-      "grad_norm": 0.001398614956997335,
-      "learning_rate": 0.0001999987636869733,
-      "loss": 46.0,
-      "step": 20712
-    },
-    {
-      "epoch": 1.5836534969512779,
-      "grad_norm": 0.0005407409043982625,
-      "learning_rate": 0.0001999987635675317,
-      "loss": 46.0,
-      "step": 20713
-    },
-    {
-      "epoch": 1.5837299539346676,
-      "grad_norm": 0.005178102757781744,
-      "learning_rate": 0.00019999876344808427,
-      "loss": 46.0,
-      "step": 20714
-    },
-    {
-      "epoch": 1.5838064109180572,
-      "grad_norm": 0.0008475991198793054,
-      "learning_rate": 0.0001999987633286311,
-      "loss": 46.0,
-      "step": 20715
-    },
-    {
-      "epoch": 1.583882867901447,
-      "grad_norm": 0.0022023834753781557,
-      "learning_rate": 0.00019999876320917216,
-      "loss": 46.0,
-      "step": 20716
-    },
-    {
-      "epoch": 1.5839593248848367,
-      "grad_norm": 0.0016136984340846539,
-      "learning_rate": 0.00019999876308970748,
-      "loss": 46.0,
-      "step": 20717
-    },
-    {
-      "epoch": 1.5840357818682262,
-      "grad_norm": 0.00853719748556614,
-      "learning_rate": 0.000199998762970237,
-      "loss": 46.0,
-      "step": 20718
-    },
-    {
-      "epoch": 1.584112238851616,
-      "grad_norm": 0.0005609886138699949,
-      "learning_rate": 0.00019999876285076073,
-      "loss": 46.0,
-      "step": 20719
-    },
-    {
-      "epoch": 1.5841886958350058,
-      "grad_norm": 0.0009174311999231577,
-      "learning_rate": 0.00019999876273127872,
-      "loss": 46.0,
-      "step": 20720
-    },
-    {
-      "epoch": 1.5842651528183955,
-      "grad_norm": 0.001000142190605402,
-      "learning_rate": 0.00019999876261179091,
-      "loss": 46.0,
-      "step": 20721
-    },
-    {
-      "epoch": 1.5843416098017853,
-      "grad_norm": 0.0019016934093087912,
-      "learning_rate": 0.0001999987624922974,
-      "loss": 46.0,
-      "step": 20722
-    },
-    {
-      "epoch": 1.584418066785175,
-      "grad_norm": 0.000705777492839843,
-      "learning_rate": 0.00019999876237279803,
-      "loss": 46.0,
-      "step": 20723
-    },
-    {
-      "epoch": 1.5844945237685648,
-      "grad_norm": 0.0008949976763688028,
-      "learning_rate": 0.00019999876225329296,
-      "loss": 46.0,
-      "step": 20724
-    },
-    {
-      "epoch": 1.5845709807519546,
-      "grad_norm": 0.0016376987332478166,
-      "learning_rate": 0.0001999987621337821,
-      "loss": 46.0,
-      "step": 20725
-    },
-    {
-      "epoch": 1.584647437735344,
-      "grad_norm": 0.0020857248455286026,
-      "learning_rate": 0.00019999876201426547,
-      "loss": 46.0,
-      "step": 20726
-    },
-    {
-      "epoch": 1.5847238947187339,
-      "grad_norm": 0.0008070914191193879,
-      "learning_rate": 0.00019999876189474305,
-      "loss": 46.0,
-      "step": 20727
-    },
-    {
-      "epoch": 1.5848003517021236,
-      "grad_norm": 0.00403270497918129,
-      "learning_rate": 0.00019999876177521485,
-      "loss": 46.0,
-      "step": 20728
-    },
-    {
-      "epoch": 1.5848768086855132,
-      "grad_norm": 0.004228347912430763,
-      "learning_rate": 0.0001999987616556809,
-      "loss": 46.0,
-      "step": 20729
-    },
-    {
-      "epoch": 1.584953265668903,
-      "grad_norm": 0.0013364336919039488,
-      "learning_rate": 0.0001999987615361412,
-      "loss": 46.0,
-      "step": 20730
-    },
-    {
-      "epoch": 1.5850297226522927,
-      "grad_norm": 0.0017118168761953712,
-      "learning_rate": 0.0001999987614165957,
-      "loss": 46.0,
-      "step": 20731
-    },
-    {
-      "epoch": 1.5851061796356825,
-      "grad_norm": 0.004465586040169001,
-      "learning_rate": 0.00019999876129704448,
-      "loss": 46.0,
-      "step": 20732
-    },
-    {
-      "epoch": 1.5851826366190722,
-      "grad_norm": 0.0030279208440333605,
-      "learning_rate": 0.00019999876117748747,
-      "loss": 46.0,
-      "step": 20733
-    },
-    {
-      "epoch": 1.585259093602462,
-      "grad_norm": 0.0004620306135620922,
-      "learning_rate": 0.00019999876105792467,
-      "loss": 46.0,
-      "step": 20734
-    },
-    {
-      "epoch": 1.5853355505858517,
-      "grad_norm": 0.0067252651788294315,
-      "learning_rate": 0.0001999987609383561,
-      "loss": 46.0,
-      "step": 20735
-    },
-    {
-      "epoch": 1.5854120075692415,
-      "grad_norm": 0.001134400488808751,
-      "learning_rate": 0.0001999987608187818,
-      "loss": 46.0,
-      "step": 20736
-    },
-    {
-      "epoch": 1.585488464552631,
-      "grad_norm": 0.00219959719106555,
-      "learning_rate": 0.0001999987606992017,
-      "loss": 46.0,
-      "step": 20737
-    },
-    {
-      "epoch": 1.5855649215360208,
-      "grad_norm": 0.0007221494452096522,
-      "learning_rate": 0.00019999876057961581,
-      "loss": 46.0,
-      "step": 20738
-    },
-    {
-      "epoch": 1.5856413785194106,
-      "grad_norm": 0.0009620852069929242,
-      "learning_rate": 0.0001999987604600242,
-      "loss": 46.0,
-      "step": 20739
-    },
-    {
-      "epoch": 1.5857178355028,
-      "grad_norm": 0.0038043714594095945,
-      "learning_rate": 0.0001999987603404268,
-      "loss": 46.0,
-      "step": 20740
-    },
-    {
-      "epoch": 1.5857942924861899,
-      "grad_norm": 0.00460520526394248,
-      "learning_rate": 0.00019999876022082363,
-      "loss": 46.0,
-      "step": 20741
-    },
-    {
-      "epoch": 1.5858707494695796,
-      "grad_norm": 0.0009562601917423308,
-      "learning_rate": 0.00019999876010121467,
-      "loss": 46.0,
-      "step": 20742
-    },
-    {
-      "epoch": 1.5859472064529694,
-      "grad_norm": 0.0016995855839923024,
-      "learning_rate": 0.00019999875998159995,
-      "loss": 46.0,
-      "step": 20743
-    },
-    {
-      "epoch": 1.5860236634363591,
-      "grad_norm": 0.0008635168196633458,
-      "learning_rate": 0.00019999875986197947,
-      "loss": 46.0,
-      "step": 20744
-    },
-    {
-      "epoch": 1.586100120419749,
-      "grad_norm": 0.0006289152661338449,
-      "learning_rate": 0.0001999987597423532,
-      "loss": 46.0,
-      "step": 20745
-    },
-    {
-      "epoch": 1.5861765774031387,
-      "grad_norm": 0.04407087340950966,
-      "learning_rate": 0.00019999875962272118,
-      "loss": 46.0,
-      "step": 20746
-    },
-    {
-      "epoch": 1.5862530343865284,
-      "grad_norm": 0.0005366257973946631,
-      "learning_rate": 0.0001999987595030834,
-      "loss": 46.0,
-      "step": 20747
-    },
-    {
-      "epoch": 1.586329491369918,
-      "grad_norm": 0.0011152320075780153,
-      "learning_rate": 0.00019999875938343984,
-      "loss": 46.0,
-      "step": 20748
-    },
-    {
-      "epoch": 1.5864059483533077,
-      "grad_norm": 0.0016431775875389576,
-      "learning_rate": 0.0001999987592637905,
-      "loss": 46.0,
-      "step": 20749
-    },
-    {
-      "epoch": 1.5864824053366975,
-      "grad_norm": 0.008029848337173462,
-      "learning_rate": 0.00019999875914413542,
-      "loss": 46.0,
-      "step": 20750
-    },
-    {
-      "epoch": 1.586558862320087,
-      "grad_norm": 0.0025705730076879263,
-      "learning_rate": 0.00019999875902447454,
-      "loss": 46.0,
-      "step": 20751
-    },
-    {
-      "epoch": 1.5866353193034768,
-      "grad_norm": 0.004085468128323555,
-      "learning_rate": 0.0001999987589048079,
-      "loss": 46.0,
-      "step": 20752
-    },
-    {
-      "epoch": 1.5867117762868665,
-      "grad_norm": 0.0039822980761528015,
-      "learning_rate": 0.00019999875878513548,
-      "loss": 46.0,
-      "step": 20753
-    },
-    {
-      "epoch": 1.5867882332702563,
-      "grad_norm": 0.0006404825253412127,
-      "learning_rate": 0.00019999875866545728,
-      "loss": 46.0,
-      "step": 20754
-    },
-    {
-      "epoch": 1.586864690253646,
-      "grad_norm": 0.0008114344673231244,
-      "learning_rate": 0.00019999875854577337,
-      "loss": 46.0,
-      "step": 20755
-    },
-    {
-      "epoch": 1.5869411472370358,
-      "grad_norm": 0.0010592964245006442,
-      "learning_rate": 0.00019999875842608366,
-      "loss": 46.0,
-      "step": 20756
-    },
-    {
-      "epoch": 1.5870176042204256,
-      "grad_norm": 0.002433561021462083,
-      "learning_rate": 0.00019999875830638817,
-      "loss": 46.0,
-      "step": 20757
-    },
-    {
-      "epoch": 1.5870940612038154,
-      "grad_norm": 0.0029834541492164135,
-      "learning_rate": 0.0001999987581866869,
-      "loss": 46.0,
-      "step": 20758
-    },
-    {
-      "epoch": 1.587170518187205,
-      "grad_norm": 0.002690690103918314,
-      "learning_rate": 0.0001999987580669799,
-      "loss": 46.0,
-      "step": 20759
-    },
-    {
-      "epoch": 1.5872469751705947,
-      "grad_norm": 0.0016464430373162031,
-      "learning_rate": 0.00019999875794726707,
-      "loss": 46.0,
-      "step": 20760
-    },
-    {
-      "epoch": 1.5873234321539844,
-      "grad_norm": 0.0013718388509005308,
-      "learning_rate": 0.00019999875782754852,
-      "loss": 46.0,
-      "step": 20761
-    },
-    {
-      "epoch": 1.587399889137374,
-      "grad_norm": 0.0010634150821715593,
-      "learning_rate": 0.00019999875770782422,
-      "loss": 46.0,
-      "step": 20762
-    },
-    {
-      "epoch": 1.5874763461207637,
-      "grad_norm": 0.0021020257845520973,
-      "learning_rate": 0.0001999987575880941,
-      "loss": 46.0,
-      "step": 20763
-    },
-    {
-      "epoch": 1.5875528031041535,
-      "grad_norm": 0.0013116709887981415,
-      "learning_rate": 0.00019999875746835822,
-      "loss": 46.0,
-      "step": 20764
-    },
-    {
-      "epoch": 1.5876292600875432,
-      "grad_norm": 0.0010398468002676964,
-      "learning_rate": 0.00019999875734861658,
-      "loss": 46.0,
-      "step": 20765
-    },
-    {
-      "epoch": 1.587705717070933,
-      "grad_norm": 0.002630737377330661,
-      "learning_rate": 0.00019999875722886916,
-      "loss": 46.0,
-      "step": 20766
-    },
-    {
-      "epoch": 1.5877821740543228,
-      "grad_norm": 0.0012553774286061525,
-      "learning_rate": 0.000199998757109116,
-      "loss": 46.0,
-      "step": 20767
-    },
-    {
-      "epoch": 1.5878586310377125,
-      "grad_norm": 0.0010421781335026026,
-      "learning_rate": 0.00019999875698935705,
-      "loss": 46.0,
-      "step": 20768
-    },
-    {
-      "epoch": 1.5879350880211023,
-      "grad_norm": 0.0014124239096418023,
-      "learning_rate": 0.00019999875686959234,
-      "loss": 46.0,
-      "step": 20769
-    },
-    {
-      "epoch": 1.5880115450044918,
-      "grad_norm": 0.002958934521302581,
-      "learning_rate": 0.00019999875674982183,
-      "loss": 46.0,
-      "step": 20770
-    },
-    {
-      "epoch": 1.5880880019878816,
-      "grad_norm": 0.0010132124880328774,
-      "learning_rate": 0.0001999987566300456,
-      "loss": 46.0,
-      "step": 20771
-    },
-    {
-      "epoch": 1.5881644589712713,
-      "grad_norm": 0.006387178786098957,
-      "learning_rate": 0.00019999875651026356,
-      "loss": 46.0,
-      "step": 20772
-    },
-    {
-      "epoch": 1.5882409159546609,
-      "grad_norm": 0.0022145016118884087,
-      "learning_rate": 0.00019999875639047576,
-      "loss": 46.0,
-      "step": 20773
-    },
-    {
-      "epoch": 1.5883173729380506,
-      "grad_norm": 0.0007645189180038869,
-      "learning_rate": 0.0001999987562706822,
-      "loss": 46.0,
-      "step": 20774
-    },
-    {
-      "epoch": 1.5883938299214404,
-      "grad_norm": 0.0008706299122422934,
-      "learning_rate": 0.00019999875615088285,
-      "loss": 46.0,
-      "step": 20775
-    },
-    {
-      "epoch": 1.5884702869048302,
-      "grad_norm": 0.0010577547363936901,
-      "learning_rate": 0.00019999875603107775,
-      "loss": 46.0,
-      "step": 20776
-    },
-    {
-      "epoch": 1.58854674388822,
-      "grad_norm": 0.0009517275029793382,
-      "learning_rate": 0.00019999875591126688,
-      "loss": 46.0,
-      "step": 20777
-    },
-    {
-      "epoch": 1.5886232008716097,
-      "grad_norm": 0.01015094667673111,
-      "learning_rate": 0.00019999875579145024,
-      "loss": 46.0,
-      "step": 20778
-    },
-    {
-      "epoch": 1.5886996578549994,
-      "grad_norm": 0.0011491603218019009,
-      "learning_rate": 0.00019999875567162784,
-      "loss": 46.0,
-      "step": 20779
-    },
-    {
-      "epoch": 1.5887761148383892,
-      "grad_norm": 0.0012401501880958676,
-      "learning_rate": 0.00019999875555179965,
-      "loss": 46.0,
-      "step": 20780
-    },
-    {
-      "epoch": 1.5888525718217787,
-      "grad_norm": 0.0016608699224889278,
-      "learning_rate": 0.00019999875543196569,
-      "loss": 46.0,
-      "step": 20781
-    },
-    {
-      "epoch": 1.5889290288051685,
-      "grad_norm": 0.022490549832582474,
-      "learning_rate": 0.000199998755312126,
-      "loss": 46.0,
-      "step": 20782
-    },
-    {
-      "epoch": 1.589005485788558,
-      "grad_norm": 0.002403588267043233,
-      "learning_rate": 0.0001999987551922805,
-      "loss": 46.0,
-      "step": 20783
-    },
-    {
-      "epoch": 1.5890819427719478,
-      "grad_norm": 0.0007447859970852733,
-      "learning_rate": 0.00019999875507242926,
-      "loss": 46.0,
-      "step": 20784
-    },
-    {
-      "epoch": 1.5891583997553376,
-      "grad_norm": 0.0008410013979300857,
-      "learning_rate": 0.0001999987549525722,
-      "loss": 46.0,
-      "step": 20785
-    },
-    {
-      "epoch": 1.5892348567387273,
-      "grad_norm": 0.005045631900429726,
-      "learning_rate": 0.0001999987548327094,
-      "loss": 46.0,
-      "step": 20786
-    },
-    {
-      "epoch": 1.589311313722117,
-      "grad_norm": 0.00039256393210962415,
-      "learning_rate": 0.00019999875471284084,
-      "loss": 46.0,
-      "step": 20787
-    },
-    {
-      "epoch": 1.5893877707055069,
-      "grad_norm": 0.0008464088896289468,
-      "learning_rate": 0.0001999987545929665,
-      "loss": 46.0,
-      "step": 20788
-    },
-    {
-      "epoch": 1.5894642276888966,
-      "grad_norm": 0.0005065621226094663,
-      "learning_rate": 0.00019999875447308641,
-      "loss": 46.0,
-      "step": 20789
-    },
-    {
-      "epoch": 1.5895406846722864,
-      "grad_norm": 0.0009137816377915442,
-      "learning_rate": 0.00019999875435320052,
-      "loss": 46.0,
-      "step": 20790
-    },
-    {
-      "epoch": 1.5896171416556761,
-      "grad_norm": 0.0018472615629434586,
-      "learning_rate": 0.00019999875423330887,
-      "loss": 46.0,
-      "step": 20791
-    },
-    {
-      "epoch": 1.5896935986390657,
-      "grad_norm": 0.0076946658082306385,
-      "learning_rate": 0.00019999875411341148,
-      "loss": 46.0,
-      "step": 20792
-    },
-    {
-      "epoch": 1.5897700556224554,
-      "grad_norm": 0.001870285370387137,
-      "learning_rate": 0.00019999875399350829,
-      "loss": 46.0,
-      "step": 20793
-    },
-    {
-      "epoch": 1.589846512605845,
-      "grad_norm": 0.0007157549844123423,
-      "learning_rate": 0.00019999875387359932,
-      "loss": 46.0,
-      "step": 20794
-    },
-    {
-      "epoch": 1.5899229695892347,
-      "grad_norm": 0.0024230354465544224,
-      "learning_rate": 0.0001999987537536846,
-      "loss": 46.0,
-      "step": 20795
-    },
-    {
-      "epoch": 1.5899994265726245,
-      "grad_norm": 0.0018404213478788733,
-      "learning_rate": 0.0001999987536337641,
-      "loss": 46.0,
-      "step": 20796
-    },
-    {
-      "epoch": 1.5900758835560143,
-      "grad_norm": 0.0014037626096978784,
-      "learning_rate": 0.00019999875351383787,
-      "loss": 46.0,
-      "step": 20797
-    },
-    {
-      "epoch": 1.590152340539404,
-      "grad_norm": 0.0028544871602207422,
-      "learning_rate": 0.00019999875339390583,
-      "loss": 46.0,
-      "step": 20798
-    },
-    {
-      "epoch": 1.5902287975227938,
-      "grad_norm": 0.003408226650208235,
-      "learning_rate": 0.00019999875327396805,
-      "loss": 46.0,
-      "step": 20799
-    },
-    {
-      "epoch": 1.5903052545061835,
-      "grad_norm": 0.0007468569674529135,
-      "learning_rate": 0.00019999875315402445,
-      "loss": 46.0,
-      "step": 20800
-    },
-    {
-      "epoch": 1.5903817114895733,
-      "grad_norm": 0.002749278210103512,
-      "learning_rate": 0.00019999875303407512,
-      "loss": 46.0,
-      "step": 20801
-    },
-    {
-      "epoch": 1.590458168472963,
-      "grad_norm": 0.0016064891824498773,
-      "learning_rate": 0.00019999875291412003,
-      "loss": 46.0,
-      "step": 20802
-    },
-    {
-      "epoch": 1.5905346254563526,
-      "grad_norm": 0.0005981055437587202,
-      "learning_rate": 0.00019999875279415913,
-      "loss": 46.0,
-      "step": 20803
-    },
-    {
-      "epoch": 1.5906110824397424,
-      "grad_norm": 0.005029712803661823,
-      "learning_rate": 0.00019999875267419248,
-      "loss": 46.0,
-      "step": 20804
-    },
-    {
-      "epoch": 1.590687539423132,
-      "grad_norm": 0.00731795746833086,
-      "learning_rate": 0.00019999875255422006,
-      "loss": 46.0,
-      "step": 20805
-    },
-    {
-      "epoch": 1.5907639964065217,
-      "grad_norm": 0.0022763372398912907,
-      "learning_rate": 0.0001999987524342419,
-      "loss": 46.0,
-      "step": 20806
-    },
-    {
-      "epoch": 1.5908404533899114,
-      "grad_norm": 0.000999784329906106,
-      "learning_rate": 0.00019999875231425793,
-      "loss": 46.0,
-      "step": 20807
-    },
-    {
-      "epoch": 1.5909169103733012,
-      "grad_norm": 0.0031457243021577597,
-      "learning_rate": 0.0001999987521942682,
-      "loss": 46.0,
-      "step": 20808
-    },
-    {
-      "epoch": 1.590993367356691,
-      "grad_norm": 0.002259228378534317,
-      "learning_rate": 0.0001999987520742727,
-      "loss": 46.0,
-      "step": 20809
-    },
-    {
-      "epoch": 1.5910698243400807,
-      "grad_norm": 0.0005722450441680849,
-      "learning_rate": 0.00019999875195427148,
-      "loss": 46.0,
-      "step": 20810
-    },
-    {
-      "epoch": 1.5911462813234705,
-      "grad_norm": 0.0035994506906718016,
-      "learning_rate": 0.00019999875183426442,
-      "loss": 46.0,
-      "step": 20811
-    },
-    {
-      "epoch": 1.5912227383068602,
-      "grad_norm": 0.0035778535529971123,
-      "learning_rate": 0.00019999875171425161,
-      "loss": 46.0,
-      "step": 20812
-    },
-    {
-      "epoch": 1.5912991952902498,
-      "grad_norm": 0.0017123211873695254,
-      "learning_rate": 0.00019999875159423304,
-      "loss": 46.0,
-      "step": 20813
-    },
-    {
-      "epoch": 1.5913756522736395,
-      "grad_norm": 0.004178970120847225,
-      "learning_rate": 0.0001999987514742087,
-      "loss": 46.0,
-      "step": 20814
-    },
-    {
-      "epoch": 1.5914521092570293,
-      "grad_norm": 0.0010184430284425616,
-      "learning_rate": 0.0001999987513541786,
-      "loss": 46.0,
-      "step": 20815
-    },
-    {
-      "epoch": 1.5915285662404188,
-      "grad_norm": 0.0008125404710881412,
-      "learning_rate": 0.00019999875123414271,
-      "loss": 46.0,
-      "step": 20816
-    },
-    {
-      "epoch": 1.5916050232238086,
-      "grad_norm": 0.0008357223705388606,
-      "learning_rate": 0.00019999875111410107,
-      "loss": 46.0,
-      "step": 20817
-    },
-    {
-      "epoch": 1.5916814802071984,
-      "grad_norm": 0.003372203791514039,
-      "learning_rate": 0.00019999875099405365,
-      "loss": 46.0,
-      "step": 20818
-    },
-    {
-      "epoch": 1.5917579371905881,
-      "grad_norm": 0.007136847358196974,
-      "learning_rate": 0.00019999875087400046,
-      "loss": 46.0,
-      "step": 20819
-    },
-    {
-      "epoch": 1.5918343941739779,
-      "grad_norm": 0.0007031827699393034,
-      "learning_rate": 0.0001999987507539415,
-      "loss": 46.0,
-      "step": 20820
-    },
-    {
-      "epoch": 1.5919108511573676,
-      "grad_norm": 0.0008745537488721311,
-      "learning_rate": 0.00019999875063387678,
-      "loss": 46.0,
-      "step": 20821
-    },
-    {
-      "epoch": 1.5919873081407574,
-      "grad_norm": 0.0006434021051973104,
-      "learning_rate": 0.0001999987505138063,
-      "loss": 46.0,
-      "step": 20822
-    },
-    {
-      "epoch": 1.5920637651241472,
-      "grad_norm": 0.0007673674845136702,
-      "learning_rate": 0.00019999875039373004,
-      "loss": 46.0,
-      "step": 20823
-    },
-    {
-      "epoch": 1.5921402221075367,
-      "grad_norm": 0.0003213771269656718,
-      "learning_rate": 0.000199998750273648,
-      "loss": 46.0,
-      "step": 20824
-    },
-    {
-      "epoch": 1.5922166790909265,
-      "grad_norm": 0.0010954872705042362,
-      "learning_rate": 0.0001999987501535602,
-      "loss": 46.0,
-      "step": 20825
-    },
-    {
-      "epoch": 1.5922931360743162,
-      "grad_norm": 0.0029426219407469034,
-      "learning_rate": 0.00019999875003346663,
-      "loss": 46.0,
-      "step": 20826
-    },
-    {
-      "epoch": 1.5923695930577058,
-      "grad_norm": 0.0006779970717616379,
-      "learning_rate": 0.00019999874991336728,
-      "loss": 46.0,
-      "step": 20827
-    },
-    {
-      "epoch": 1.5924460500410955,
-      "grad_norm": 0.001310917316004634,
-      "learning_rate": 0.00019999874979326215,
-      "loss": 46.0,
-      "step": 20828
-    },
-    {
-      "epoch": 1.5925225070244853,
-      "grad_norm": 0.0008367177797481418,
-      "learning_rate": 0.00019999874967315128,
-      "loss": 46.0,
-      "step": 20829
-    },
-    {
-      "epoch": 1.592598964007875,
-      "grad_norm": 0.0020137790124863386,
-      "learning_rate": 0.0001999987495530346,
-      "loss": 46.0,
-      "step": 20830
-    },
-    {
-      "epoch": 1.5926754209912648,
-      "grad_norm": 0.002945945831015706,
-      "learning_rate": 0.00019999874943291222,
-      "loss": 46.0,
-      "step": 20831
-    },
-    {
-      "epoch": 1.5927518779746546,
-      "grad_norm": 0.0027137494180351496,
-      "learning_rate": 0.00019999874931278403,
-      "loss": 46.0,
-      "step": 20832
-    },
-    {
-      "epoch": 1.5928283349580443,
-      "grad_norm": 0.0009312950423918664,
-      "learning_rate": 0.00019999874919265003,
-      "loss": 46.0,
-      "step": 20833
-    },
-    {
-      "epoch": 1.592904791941434,
-      "grad_norm": 0.0016654873033985496,
-      "learning_rate": 0.00019999874907251032,
-      "loss": 46.0,
-      "step": 20834
-    },
-    {
-      "epoch": 1.5929812489248236,
-      "grad_norm": 0.0023353395517915487,
-      "learning_rate": 0.00019999874895236484,
-      "loss": 46.0,
-      "step": 20835
-    },
-    {
-      "epoch": 1.5930577059082134,
-      "grad_norm": 0.0016996051417663693,
-      "learning_rate": 0.00019999874883221355,
-      "loss": 46.0,
-      "step": 20836
-    },
-    {
-      "epoch": 1.5931341628916031,
-      "grad_norm": 0.0015450542559847236,
-      "learning_rate": 0.00019999874871205652,
-      "loss": 46.0,
-      "step": 20837
-    },
-    {
-      "epoch": 1.5932106198749927,
-      "grad_norm": 0.005874384194612503,
-      "learning_rate": 0.00019999874859189372,
-      "loss": 46.0,
-      "step": 20838
-    },
-    {
-      "epoch": 1.5932870768583824,
-      "grad_norm": 0.001090421574190259,
-      "learning_rate": 0.00019999874847172514,
-      "loss": 46.0,
-      "step": 20839
-    },
-    {
-      "epoch": 1.5933635338417722,
-      "grad_norm": 0.0009212449076585472,
-      "learning_rate": 0.00019999874835155082,
-      "loss": 46.0,
-      "step": 20840
-    },
-    {
-      "epoch": 1.593439990825162,
-      "grad_norm": 0.0010577606735751033,
-      "learning_rate": 0.0001999987482313707,
-      "loss": 46.0,
-      "step": 20841
-    },
-    {
-      "epoch": 1.5935164478085517,
-      "grad_norm": 0.0018657715991139412,
-      "learning_rate": 0.00019999874811118482,
-      "loss": 46.0,
-      "step": 20842
-    },
-    {
-      "epoch": 1.5935929047919415,
-      "grad_norm": 0.0012882158625870943,
-      "learning_rate": 0.00019999874799099315,
-      "loss": 46.0,
-      "step": 20843
-    },
-    {
-      "epoch": 1.5936693617753313,
-      "grad_norm": 0.002236689440906048,
-      "learning_rate": 0.00019999874787079573,
-      "loss": 46.0,
-      "step": 20844
-    },
-    {
-      "epoch": 1.593745818758721,
-      "grad_norm": 0.0011569999624043703,
-      "learning_rate": 0.00019999874775059254,
-      "loss": 46.0,
-      "step": 20845
-    },
-    {
-      "epoch": 1.5938222757421106,
-      "grad_norm": 0.0011731341946870089,
-      "learning_rate": 0.0001999987476303836,
-      "loss": 46.0,
-      "step": 20846
-    },
-    {
-      "epoch": 1.5938987327255003,
-      "grad_norm": 0.0016871416009962559,
-      "learning_rate": 0.00019999874751016887,
-      "loss": 46.0,
-      "step": 20847
-    },
-    {
-      "epoch": 1.59397518970889,
-      "grad_norm": 0.0012930063530802727,
-      "learning_rate": 0.00019999874738994838,
-      "loss": 46.0,
-      "step": 20848
-    },
-    {
-      "epoch": 1.5940516466922796,
-      "grad_norm": 0.0011039967648684978,
-      "learning_rate": 0.0001999987472697221,
-      "loss": 46.0,
-      "step": 20849
-    },
-    {
-      "epoch": 1.5941281036756694,
-      "grad_norm": 0.0024709177669137716,
-      "learning_rate": 0.00019999874714949004,
-      "loss": 46.0,
-      "step": 20850
-    },
-    {
-      "epoch": 1.5942045606590591,
-      "grad_norm": 0.00043582136277109385,
-      "learning_rate": 0.00019999874702925224,
-      "loss": 46.0,
-      "step": 20851
-    },
-    {
-      "epoch": 1.594281017642449,
-      "grad_norm": 0.0008094763616099954,
-      "learning_rate": 0.0001999987469090087,
-      "loss": 46.0,
-      "step": 20852
-    },
-    {
-      "epoch": 1.5943574746258387,
-      "grad_norm": 0.0012028083438053727,
-      "learning_rate": 0.0001999987467887593,
-      "loss": 46.0,
-      "step": 20853
-    },
-    {
-      "epoch": 1.5944339316092284,
-      "grad_norm": 0.0019079288467764854,
-      "learning_rate": 0.0001999987466685042,
-      "loss": 46.0,
-      "step": 20854
-    },
-    {
-      "epoch": 1.5945103885926182,
-      "grad_norm": 0.0005420685047283769,
-      "learning_rate": 0.00019999874654824331,
-      "loss": 46.0,
-      "step": 20855
-    },
-    {
-      "epoch": 1.594586845576008,
-      "grad_norm": 0.0006356465164572,
-      "learning_rate": 0.00019999874642797667,
-      "loss": 46.0,
-      "step": 20856
-    },
-    {
-      "epoch": 1.5946633025593975,
-      "grad_norm": 0.002818293171003461,
-      "learning_rate": 0.00019999874630770425,
-      "loss": 46.0,
-      "step": 20857
-    },
-    {
-      "epoch": 1.5947397595427872,
-      "grad_norm": 0.010402724146842957,
-      "learning_rate": 0.00019999874618742606,
-      "loss": 46.0,
-      "step": 20858
-    },
-    {
-      "epoch": 1.594816216526177,
-      "grad_norm": 0.0012574056163430214,
-      "learning_rate": 0.00019999874606714207,
-      "loss": 46.0,
-      "step": 20859
-    },
-    {
-      "epoch": 1.5948926735095665,
-      "grad_norm": 0.0017806269461289048,
-      "learning_rate": 0.00019999874594685236,
-      "loss": 46.0,
-      "step": 20860
-    },
-    {
-      "epoch": 1.5949691304929563,
-      "grad_norm": 0.0008180948789231479,
-      "learning_rate": 0.00019999874582655688,
-      "loss": 46.0,
-      "step": 20861
-    },
-    {
-      "epoch": 1.595045587476346,
-      "grad_norm": 0.0005747161339968443,
-      "learning_rate": 0.0001999987457062556,
-      "loss": 46.0,
-      "step": 20862
-    },
-    {
-      "epoch": 1.5951220444597358,
-      "grad_norm": 0.0013234395300969481,
-      "learning_rate": 0.00019999874558594854,
-      "loss": 46.0,
-      "step": 20863
-    },
-    {
-      "epoch": 1.5951985014431256,
-      "grad_norm": 0.0012940064771100879,
-      "learning_rate": 0.00019999874546563576,
-      "loss": 46.0,
-      "step": 20864
-    },
-    {
-      "epoch": 1.5952749584265153,
-      "grad_norm": 0.00130782974883914,
-      "learning_rate": 0.00019999874534531715,
-      "loss": 46.0,
-      "step": 20865
-    },
-    {
-      "epoch": 1.595351415409905,
-      "grad_norm": 0.0016017970629036427,
-      "learning_rate": 0.00019999874522499283,
-      "loss": 46.0,
-      "step": 20866
-    },
-    {
-      "epoch": 1.5954278723932949,
-      "grad_norm": 0.0006534263375215232,
-      "learning_rate": 0.00019999874510466268,
-      "loss": 46.0,
-      "step": 20867
-    },
-    {
-      "epoch": 1.5955043293766844,
-      "grad_norm": 0.0019020965555682778,
-      "learning_rate": 0.0001999987449843268,
-      "loss": 46.0,
-      "step": 20868
-    },
-    {
-      "epoch": 1.5955807863600742,
-      "grad_norm": 0.0007338155410252512,
-      "learning_rate": 0.00019999874486398517,
-      "loss": 46.0,
-      "step": 20869
-    },
-    {
-      "epoch": 1.595657243343464,
-      "grad_norm": 0.0007633164641447365,
-      "learning_rate": 0.00019999874474363772,
-      "loss": 46.0,
-      "step": 20870
-    },
-    {
-      "epoch": 1.5957337003268535,
-      "grad_norm": 0.0007111151935532689,
-      "learning_rate": 0.00019999874462328453,
-      "loss": 46.0,
-      "step": 20871
-    },
-    {
-      "epoch": 1.5958101573102432,
-      "grad_norm": 0.0005665522185154259,
-      "learning_rate": 0.00019999874450292557,
-      "loss": 46.0,
-      "step": 20872
-    },
-    {
-      "epoch": 1.595886614293633,
-      "grad_norm": 0.00092048489023,
-      "learning_rate": 0.00019999874438256083,
-      "loss": 46.0,
-      "step": 20873
-    },
-    {
-      "epoch": 1.5959630712770227,
-      "grad_norm": 0.0006376648088917136,
-      "learning_rate": 0.00019999874426219032,
-      "loss": 46.0,
-      "step": 20874
-    },
-    {
-      "epoch": 1.5960395282604125,
-      "grad_norm": 0.0003015449910890311,
-      "learning_rate": 0.00019999874414181406,
-      "loss": 46.0,
-      "step": 20875
-    },
-    {
-      "epoch": 1.5961159852438023,
-      "grad_norm": 0.002682760590687394,
-      "learning_rate": 0.000199998744021432,
-      "loss": 46.0,
-      "step": 20876
-    },
-    {
-      "epoch": 1.596192442227192,
-      "grad_norm": 0.0023874102625995874,
-      "learning_rate": 0.0001999987439010442,
-      "loss": 46.0,
-      "step": 20877
-    },
-    {
-      "epoch": 1.5962688992105818,
-      "grad_norm": 0.0016277419636026025,
-      "learning_rate": 0.00019999874378065063,
-      "loss": 46.0,
-      "step": 20878
-    },
-    {
-      "epoch": 1.5963453561939713,
-      "grad_norm": 0.0009841523133218288,
-      "learning_rate": 0.00019999874366025128,
-      "loss": 46.0,
-      "step": 20879
-    },
-    {
-      "epoch": 1.596421813177361,
-      "grad_norm": 0.000902264378964901,
-      "learning_rate": 0.00019999874353984615,
-      "loss": 46.0,
-      "step": 20880
-    },
-    {
-      "epoch": 1.5964982701607509,
-      "grad_norm": 0.006537141744047403,
-      "learning_rate": 0.00019999874341943526,
-      "loss": 46.0,
-      "step": 20881
-    },
-    {
-      "epoch": 1.5965747271441404,
-      "grad_norm": 0.0009693100000731647,
-      "learning_rate": 0.00019999874329901859,
-      "loss": 46.0,
-      "step": 20882
-    },
-    {
-      "epoch": 1.5966511841275302,
-      "grad_norm": 0.0008807822014205158,
-      "learning_rate": 0.00019999874317859617,
-      "loss": 46.0,
-      "step": 20883
-    },
-    {
-      "epoch": 1.59672764111092,
-      "grad_norm": 0.002462441800162196,
-      "learning_rate": 0.00019999874305816798,
-      "loss": 46.0,
-      "step": 20884
-    },
-    {
-      "epoch": 1.5968040980943097,
-      "grad_norm": 0.0009589497931301594,
-      "learning_rate": 0.00019999874293773402,
-      "loss": 46.0,
-      "step": 20885
-    },
-    {
-      "epoch": 1.5968805550776994,
-      "grad_norm": 0.0006952695548534393,
-      "learning_rate": 0.0001999987428172943,
-      "loss": 46.0,
-      "step": 20886
-    },
-    {
-      "epoch": 1.5969570120610892,
-      "grad_norm": 0.002531966660171747,
-      "learning_rate": 0.00019999874269684877,
-      "loss": 46.0,
-      "step": 20887
-    },
-    {
-      "epoch": 1.597033469044479,
-      "grad_norm": 0.0014390654396265745,
-      "learning_rate": 0.0001999987425763975,
-      "loss": 46.0,
-      "step": 20888
-    },
-    {
-      "epoch": 1.5971099260278687,
-      "grad_norm": 0.000946647603996098,
-      "learning_rate": 0.00019999874245594046,
-      "loss": 46.0,
-      "step": 20889
-    },
-    {
-      "epoch": 1.5971863830112583,
-      "grad_norm": 0.0008359703933820128,
-      "learning_rate": 0.00019999874233547765,
-      "loss": 46.0,
-      "step": 20890
-    },
-    {
-      "epoch": 1.597262839994648,
-      "grad_norm": 0.001937451888807118,
-      "learning_rate": 0.00019999874221500905,
-      "loss": 46.0,
-      "step": 20891
-    },
-    {
-      "epoch": 1.5973392969780378,
-      "grad_norm": 0.0006890198565088212,
-      "learning_rate": 0.0001999987420945347,
-      "loss": 46.0,
-      "step": 20892
-    },
-    {
-      "epoch": 1.5974157539614273,
-      "grad_norm": 0.000873405544552952,
-      "learning_rate": 0.0001999987419740546,
-      "loss": 46.0,
-      "step": 20893
-    },
-    {
-      "epoch": 1.597492210944817,
-      "grad_norm": 0.0019404280465096235,
-      "learning_rate": 0.0001999987418535687,
-      "loss": 46.0,
-      "step": 20894
-    },
-    {
-      "epoch": 1.5975686679282068,
-      "grad_norm": 0.0011197124840691686,
-      "learning_rate": 0.00019999874173307704,
-      "loss": 46.0,
-      "step": 20895
-    },
-    {
-      "epoch": 1.5976451249115966,
-      "grad_norm": 0.002604343928396702,
-      "learning_rate": 0.00019999874161257962,
-      "loss": 46.0,
-      "step": 20896
-    },
-    {
-      "epoch": 1.5977215818949864,
-      "grad_norm": 0.0009899816941469908,
-      "learning_rate": 0.0001999987414920764,
-      "loss": 46.0,
-      "step": 20897
-    },
-    {
-      "epoch": 1.5977980388783761,
-      "grad_norm": 0.004315212368965149,
-      "learning_rate": 0.00019999874137156744,
-      "loss": 46.0,
-      "step": 20898
-    },
-    {
-      "epoch": 1.5978744958617659,
-      "grad_norm": 0.004477735608816147,
-      "learning_rate": 0.0001999987412510527,
-      "loss": 46.0,
-      "step": 20899
-    },
-    {
-      "epoch": 1.5979509528451556,
-      "grad_norm": 0.0009104431373998523,
-      "learning_rate": 0.0001999987411305322,
-      "loss": 46.0,
-      "step": 20900
-    },
-    {
-      "epoch": 1.5980274098285452,
-      "grad_norm": 0.006579105276614428,
-      "learning_rate": 0.00019999874101000594,
-      "loss": 46.0,
-      "step": 20901
-    },
-    {
-      "epoch": 1.598103866811935,
-      "grad_norm": 0.00044072544551454484,
-      "learning_rate": 0.00019999874088947385,
-      "loss": 46.0,
-      "step": 20902
-    },
-    {
-      "epoch": 1.5981803237953247,
-      "grad_norm": 0.0009522698237560689,
-      "learning_rate": 0.00019999874076893605,
-      "loss": 46.0,
-      "step": 20903
-    },
-    {
-      "epoch": 1.5982567807787142,
-      "grad_norm": 0.0013952743029221892,
-      "learning_rate": 0.00019999874064839248,
-      "loss": 46.0,
-      "step": 20904
-    },
-    {
-      "epoch": 1.598333237762104,
-      "grad_norm": 0.006217521149665117,
-      "learning_rate": 0.00019999874052784313,
-      "loss": 46.0,
-      "step": 20905
-    },
-    {
-      "epoch": 1.5984096947454938,
-      "grad_norm": 0.0009118167217820883,
-      "learning_rate": 0.000199998740407288,
-      "loss": 46.0,
-      "step": 20906
-    },
-    {
-      "epoch": 1.5984861517288835,
-      "grad_norm": 0.0006965245120227337,
-      "learning_rate": 0.0001999987402867271,
-      "loss": 46.0,
-      "step": 20907
-    },
-    {
-      "epoch": 1.5985626087122733,
-      "grad_norm": 0.004919230006635189,
-      "learning_rate": 0.00019999874016616044,
-      "loss": 46.0,
-      "step": 20908
-    },
-    {
-      "epoch": 1.598639065695663,
-      "grad_norm": 0.0011556344106793404,
-      "learning_rate": 0.000199998740045588,
-      "loss": 46.0,
-      "step": 20909
-    },
-    {
-      "epoch": 1.5987155226790528,
-      "grad_norm": 0.0003047606151085347,
-      "learning_rate": 0.0001999987399250098,
-      "loss": 46.0,
-      "step": 20910
-    },
-    {
-      "epoch": 1.5987919796624426,
-      "grad_norm": 0.0007800646126270294,
-      "learning_rate": 0.00019999873980442584,
-      "loss": 46.0,
-      "step": 20911
-    },
-    {
-      "epoch": 1.5988684366458321,
-      "grad_norm": 0.0009797647362574935,
-      "learning_rate": 0.00019999873968383608,
-      "loss": 46.0,
-      "step": 20912
-    },
-    {
-      "epoch": 1.5989448936292219,
-      "grad_norm": 0.002188715385273099,
-      "learning_rate": 0.0001999987395632406,
-      "loss": 46.0,
-      "step": 20913
-    },
-    {
-      "epoch": 1.5990213506126114,
-      "grad_norm": 0.0010862976778298616,
-      "learning_rate": 0.0001999987394426393,
-      "loss": 46.0,
-      "step": 20914
-    },
-    {
-      "epoch": 1.5990978075960012,
-      "grad_norm": 0.002701367950066924,
-      "learning_rate": 0.00019999873932203226,
-      "loss": 46.0,
-      "step": 20915
-    },
-    {
-      "epoch": 1.599174264579391,
-      "grad_norm": 0.0008128510671667755,
-      "learning_rate": 0.00019999873920141945,
-      "loss": 46.0,
-      "step": 20916
-    },
-    {
-      "epoch": 1.5992507215627807,
-      "grad_norm": 0.0008816164336167276,
-      "learning_rate": 0.00019999873908080088,
-      "loss": 46.0,
-      "step": 20917
-    },
-    {
-      "epoch": 1.5993271785461705,
-      "grad_norm": 0.0026887725107371807,
-      "learning_rate": 0.0001999987389601765,
-      "loss": 46.0,
-      "step": 20918
-    },
-    {
-      "epoch": 1.5994036355295602,
-      "grad_norm": 0.003197918413206935,
-      "learning_rate": 0.00019999873883954638,
-      "loss": 46.0,
-      "step": 20919
-    },
-    {
-      "epoch": 1.59948009251295,
-      "grad_norm": 0.0008864808478392661,
-      "learning_rate": 0.00019999873871891048,
-      "loss": 46.0,
-      "step": 20920
-    },
-    {
-      "epoch": 1.5995565494963397,
-      "grad_norm": 0.0007778560393489897,
-      "learning_rate": 0.00019999873859826881,
-      "loss": 46.0,
-      "step": 20921
-    },
-    {
-      "epoch": 1.5996330064797295,
-      "grad_norm": 0.0011391224106773734,
-      "learning_rate": 0.00019999873847762137,
-      "loss": 46.0,
-      "step": 20922
-    },
-    {
-      "epoch": 1.599709463463119,
-      "grad_norm": 0.0007007248350419104,
-      "learning_rate": 0.00019999873835696818,
-      "loss": 46.0,
-      "step": 20923
-    },
-    {
-      "epoch": 1.5997859204465088,
-      "grad_norm": 0.0010847756639122963,
-      "learning_rate": 0.00019999873823630922,
-      "loss": 46.0,
-      "step": 20924
-    },
-    {
-      "epoch": 1.5998623774298983,
-      "grad_norm": 0.0015344604616984725,
-      "learning_rate": 0.00019999873811564446,
-      "loss": 46.0,
-      "step": 20925
-    },
-    {
-      "epoch": 1.599938834413288,
-      "grad_norm": 0.0028038558084517717,
-      "learning_rate": 0.00019999873799497397,
-      "loss": 46.0,
-      "step": 20926
-    },
-    {
-      "epoch": 1.6000152913966779,
-      "grad_norm": 0.001368249300867319,
-      "learning_rate": 0.0001999987378742977,
-      "loss": 46.0,
-      "step": 20927
-    },
-    {
-      "epoch": 1.6000917483800676,
-      "grad_norm": 0.0017033469630405307,
-      "learning_rate": 0.0001999987377536156,
-      "loss": 46.0,
-      "step": 20928
-    },
-    {
-      "epoch": 1.6001682053634574,
-      "grad_norm": 0.0012224729871377349,
-      "learning_rate": 0.0001999987376329278,
-      "loss": 46.0,
-      "step": 20929
-    },
-    {
-      "epoch": 1.6002446623468471,
-      "grad_norm": 0.0011936567025259137,
-      "learning_rate": 0.0001999987375122342,
-      "loss": 46.0,
-      "step": 20930
-    },
-    {
-      "epoch": 1.600321119330237,
-      "grad_norm": 0.0052709211595356464,
-      "learning_rate": 0.00019999873739153486,
-      "loss": 46.0,
-      "step": 20931
-    },
-    {
-      "epoch": 1.6003975763136267,
-      "grad_norm": 0.0028194868937134743,
-      "learning_rate": 0.00019999873727082973,
-      "loss": 46.0,
-      "step": 20932
-    },
-    {
-      "epoch": 1.6004740332970164,
-      "grad_norm": 0.0010508736595511436,
-      "learning_rate": 0.00019999873715011886,
-      "loss": 46.0,
-      "step": 20933
-    },
-    {
-      "epoch": 1.600550490280406,
-      "grad_norm": 0.0015400223201140761,
-      "learning_rate": 0.00019999873702940217,
-      "loss": 46.0,
-      "step": 20934
-    },
-    {
-      "epoch": 1.6006269472637957,
-      "grad_norm": 0.0016773439710959792,
-      "learning_rate": 0.00019999873690867975,
-      "loss": 46.0,
-      "step": 20935
-    },
-    {
-      "epoch": 1.6007034042471853,
-      "grad_norm": 0.0011647220235317945,
-      "learning_rate": 0.00019999873678795154,
-      "loss": 46.0,
-      "step": 20936
-    },
-    {
-      "epoch": 1.600779861230575,
-      "grad_norm": 0.0006939746672287583,
-      "learning_rate": 0.00019999873666721755,
-      "loss": 46.0,
-      "step": 20937
-    },
-    {
-      "epoch": 1.6008563182139648,
-      "grad_norm": 0.0012253641616553068,
-      "learning_rate": 0.0001999987365464778,
-      "loss": 46.0,
-      "step": 20938
-    },
-    {
-      "epoch": 1.6009327751973546,
-      "grad_norm": 0.013783898204565048,
-      "learning_rate": 0.0001999987364257323,
-      "loss": 46.0,
-      "step": 20939
-    },
-    {
-      "epoch": 1.6010092321807443,
-      "grad_norm": 0.0006799410912208259,
-      "learning_rate": 0.00019999873630498102,
-      "loss": 46.0,
-      "step": 20940
-    },
-    {
-      "epoch": 1.601085689164134,
-      "grad_norm": 0.0005718697793781757,
-      "learning_rate": 0.00019999873618422397,
-      "loss": 46.0,
-      "step": 20941
-    },
-    {
-      "epoch": 1.6011621461475238,
-      "grad_norm": 0.0028528645634651184,
-      "learning_rate": 0.00019999873606346116,
-      "loss": 46.0,
-      "step": 20942
-    },
-    {
-      "epoch": 1.6012386031309136,
-      "grad_norm": 0.0030135787092149258,
-      "learning_rate": 0.00019999873594269256,
-      "loss": 46.0,
-      "step": 20943
-    },
-    {
-      "epoch": 1.6013150601143031,
-      "grad_norm": 0.0033525920007377863,
-      "learning_rate": 0.00019999873582191819,
-      "loss": 46.0,
-      "step": 20944
-    },
-    {
-      "epoch": 1.601391517097693,
-      "grad_norm": 0.0006587335374206305,
-      "learning_rate": 0.00019999873570113806,
-      "loss": 46.0,
-      "step": 20945
-    },
-    {
-      "epoch": 1.6014679740810827,
-      "grad_norm": 0.001272045192308724,
-      "learning_rate": 0.00019999873558035217,
-      "loss": 46.0,
-      "step": 20946
-    },
-    {
-      "epoch": 1.6015444310644722,
-      "grad_norm": 0.0009788313182070851,
-      "learning_rate": 0.0001999987354595605,
-      "loss": 46.0,
-      "step": 20947
-    },
-    {
-      "epoch": 1.601620888047862,
-      "grad_norm": 0.0021034665405750275,
-      "learning_rate": 0.00019999873533876308,
-      "loss": 46.0,
-      "step": 20948
-    },
-    {
-      "epoch": 1.6016973450312517,
-      "grad_norm": 0.0012123460182920098,
-      "learning_rate": 0.00019999873521795984,
-      "loss": 46.0,
-      "step": 20949
-    },
-    {
-      "epoch": 1.6017738020146415,
-      "grad_norm": 0.004423484206199646,
-      "learning_rate": 0.00019999873509715088,
-      "loss": 46.0,
-      "step": 20950
-    },
-    {
-      "epoch": 1.6018502589980312,
-      "grad_norm": 0.0004724509490188211,
-      "learning_rate": 0.00019999873497633614,
-      "loss": 46.0,
-      "step": 20951
-    },
-    {
-      "epoch": 1.601926715981421,
-      "grad_norm": 0.0017719129100441933,
-      "learning_rate": 0.00019999873485551564,
-      "loss": 46.0,
-      "step": 20952
-    },
-    {
-      "epoch": 1.6020031729648108,
-      "grad_norm": 0.0010541756637394428,
-      "learning_rate": 0.00019999873473468935,
-      "loss": 46.0,
-      "step": 20953
-    },
-    {
-      "epoch": 1.6020796299482005,
-      "grad_norm": 0.0007369638769887388,
-      "learning_rate": 0.0001999987346138573,
-      "loss": 46.0,
-      "step": 20954
-    },
-    {
-      "epoch": 1.60215608693159,
-      "grad_norm": 0.0012938340660184622,
-      "learning_rate": 0.00019999873449301947,
-      "loss": 46.0,
-      "step": 20955
-    },
-    {
-      "epoch": 1.6022325439149798,
-      "grad_norm": 0.0014175311662256718,
-      "learning_rate": 0.00019999873437217587,
-      "loss": 46.0,
-      "step": 20956
-    },
-    {
-      "epoch": 1.6023090008983696,
-      "grad_norm": 0.014053162187337875,
-      "learning_rate": 0.0001999987342513265,
-      "loss": 46.0,
-      "step": 20957
-    },
-    {
-      "epoch": 1.6023854578817591,
-      "grad_norm": 0.0017105075530707836,
-      "learning_rate": 0.0001999987341304714,
-      "loss": 46.0,
-      "step": 20958
-    },
-    {
-      "epoch": 1.6024619148651489,
-      "grad_norm": 0.0030711700674146414,
-      "learning_rate": 0.00019999873400961048,
-      "loss": 46.0,
-      "step": 20959
-    },
-    {
-      "epoch": 1.6025383718485386,
-      "grad_norm": 0.0005176718113943934,
-      "learning_rate": 0.0001999987338887438,
-      "loss": 46.0,
-      "step": 20960
-    },
-    {
-      "epoch": 1.6026148288319284,
-      "grad_norm": 0.0023311355616897345,
-      "learning_rate": 0.00019999873376787134,
-      "loss": 46.0,
-      "step": 20961
-    },
-    {
-      "epoch": 1.6026912858153182,
-      "grad_norm": 0.0007144382107071579,
-      "learning_rate": 0.00019999873364699315,
-      "loss": 46.0,
-      "step": 20962
-    },
-    {
-      "epoch": 1.602767742798708,
-      "grad_norm": 0.00022900894691701978,
-      "learning_rate": 0.0001999987335261092,
-      "loss": 46.0,
-      "step": 20963
-    },
-    {
-      "epoch": 1.6028441997820977,
-      "grad_norm": 0.0007813775446265936,
-      "learning_rate": 0.00019999873340521943,
-      "loss": 46.0,
-      "step": 20964
-    },
-    {
-      "epoch": 1.6029206567654875,
-      "grad_norm": 0.0035930254962295294,
-      "learning_rate": 0.00019999873328432392,
-      "loss": 46.0,
-      "step": 20965
-    },
-    {
-      "epoch": 1.602997113748877,
-      "grad_norm": 0.0005185171612538397,
-      "learning_rate": 0.00019999873316342264,
-      "loss": 46.0,
-      "step": 20966
-    },
-    {
-      "epoch": 1.6030735707322668,
-      "grad_norm": 0.0007214799406938255,
-      "learning_rate": 0.00019999873304251558,
-      "loss": 46.0,
-      "step": 20967
-    },
-    {
-      "epoch": 1.6031500277156565,
-      "grad_norm": 0.0005261466722004116,
-      "learning_rate": 0.00019999873292160273,
-      "loss": 46.0,
-      "step": 20968
-    },
-    {
-      "epoch": 1.603226484699046,
-      "grad_norm": 0.0008981827413663268,
-      "learning_rate": 0.00019999873280068415,
-      "loss": 46.0,
-      "step": 20969
-    },
-    {
-      "epoch": 1.6033029416824358,
-      "grad_norm": 0.002676150994375348,
-      "learning_rate": 0.0001999987326797598,
-      "loss": 46.0,
-      "step": 20970
-    },
-    {
-      "epoch": 1.6033793986658256,
-      "grad_norm": 0.0009556770091876388,
-      "learning_rate": 0.00019999873255882966,
-      "loss": 46.0,
-      "step": 20971
-    },
-    {
-      "epoch": 1.6034558556492153,
-      "grad_norm": 0.002996955532580614,
-      "learning_rate": 0.00019999873243789376,
-      "loss": 46.0,
-      "step": 20972
-    },
-    {
-      "epoch": 1.603532312632605,
-      "grad_norm": 0.0009080239688046277,
-      "learning_rate": 0.0001999987323169521,
-      "loss": 46.0,
-      "step": 20973
-    },
-    {
-      "epoch": 1.6036087696159949,
-      "grad_norm": 0.002595087280496955,
-      "learning_rate": 0.00019999873219600465,
-      "loss": 46.0,
-      "step": 20974
-    },
-    {
-      "epoch": 1.6036852265993846,
-      "grad_norm": 0.003918865229934454,
-      "learning_rate": 0.0001999987320750514,
-      "loss": 46.0,
-      "step": 20975
-    },
-    {
-      "epoch": 1.6037616835827744,
-      "grad_norm": 0.0011954326182603836,
-      "learning_rate": 0.00019999873195409245,
-      "loss": 46.0,
-      "step": 20976
-    },
-    {
-      "epoch": 1.603838140566164,
-      "grad_norm": 0.0012244610115885735,
-      "learning_rate": 0.0001999987318331277,
-      "loss": 46.0,
-      "step": 20977
-    },
-    {
-      "epoch": 1.6039145975495537,
-      "grad_norm": 0.00048343127127736807,
-      "learning_rate": 0.00019999873171215718,
-      "loss": 46.0,
-      "step": 20978
-    },
-    {
-      "epoch": 1.6039910545329434,
-      "grad_norm": 0.007794278673827648,
-      "learning_rate": 0.00019999873159118087,
-      "loss": 46.0,
-      "step": 20979
-    },
-    {
-      "epoch": 1.604067511516333,
-      "grad_norm": 0.0030588796362280846,
-      "learning_rate": 0.00019999873147019884,
-      "loss": 46.0,
-      "step": 20980
-    },
-    {
-      "epoch": 1.6041439684997227,
-      "grad_norm": 0.0007419758476316929,
-      "learning_rate": 0.00019999873134921101,
-      "loss": 46.0,
-      "step": 20981
-    },
-    {
-      "epoch": 1.6042204254831125,
-      "grad_norm": 0.0017668596701696515,
-      "learning_rate": 0.0001999987312282174,
-      "loss": 46.0,
-      "step": 20982
-    },
-    {
-      "epoch": 1.6042968824665023,
-      "grad_norm": 0.0019268132746219635,
-      "learning_rate": 0.00019999873110721804,
-      "loss": 46.0,
-      "step": 20983
-    },
-    {
-      "epoch": 1.604373339449892,
-      "grad_norm": 0.000637273711618036,
-      "learning_rate": 0.00019999873098621292,
-      "loss": 46.0,
-      "step": 20984
-    },
-    {
-      "epoch": 1.6044497964332818,
-      "grad_norm": 0.0017091365298256278,
-      "learning_rate": 0.000199998730865202,
-      "loss": 46.0,
-      "step": 20985
-    },
-    {
-      "epoch": 1.6045262534166715,
-      "grad_norm": 0.0008401759550906718,
-      "learning_rate": 0.00019999873074418533,
-      "loss": 46.0,
-      "step": 20986
-    },
-    {
-      "epoch": 1.6046027104000613,
-      "grad_norm": 0.0016504711238667369,
-      "learning_rate": 0.0001999987306231629,
-      "loss": 46.0,
-      "step": 20987
-    },
-    {
-      "epoch": 1.6046791673834508,
-      "grad_norm": 0.0012186329113319516,
-      "learning_rate": 0.0001999987305021347,
-      "loss": 46.0,
-      "step": 20988
-    },
-    {
-      "epoch": 1.6047556243668406,
-      "grad_norm": 0.0006863982416689396,
-      "learning_rate": 0.0001999987303811007,
-      "loss": 46.0,
-      "step": 20989
-    },
-    {
-      "epoch": 1.6048320813502304,
-      "grad_norm": 0.0009601805941201746,
-      "learning_rate": 0.00019999873026006095,
-      "loss": 46.0,
-      "step": 20990
-    },
-    {
-      "epoch": 1.60490853833362,
-      "grad_norm": 0.0014521460980176926,
-      "learning_rate": 0.00019999873013901542,
-      "loss": 46.0,
-      "step": 20991
-    },
-    {
-      "epoch": 1.6049849953170097,
-      "grad_norm": 0.002574555343016982,
-      "learning_rate": 0.0001999987300179641,
-      "loss": 46.0,
-      "step": 20992
-    },
-    {
-      "epoch": 1.6050614523003994,
-      "grad_norm": 0.000844193622469902,
-      "learning_rate": 0.00019999872989690708,
-      "loss": 46.0,
-      "step": 20993
-    },
-    {
-      "epoch": 1.6051379092837892,
-      "grad_norm": 0.004621168132871389,
-      "learning_rate": 0.00019999872977584425,
-      "loss": 46.0,
-      "step": 20994
-    },
-    {
-      "epoch": 1.605214366267179,
-      "grad_norm": 0.0035156519152224064,
-      "learning_rate": 0.00019999872965477565,
-      "loss": 46.0,
-      "step": 20995
-    },
-    {
-      "epoch": 1.6052908232505687,
-      "grad_norm": 0.0033848511520773172,
-      "learning_rate": 0.00019999872953370128,
-      "loss": 46.0,
-      "step": 20996
-    },
-    {
-      "epoch": 1.6053672802339585,
-      "grad_norm": 0.0004745798360090703,
-      "learning_rate": 0.00019999872941262116,
-      "loss": 46.0,
-      "step": 20997
-    },
-    {
-      "epoch": 1.6054437372173482,
-      "grad_norm": 0.0028230210300534964,
-      "learning_rate": 0.00019999872929153523,
-      "loss": 46.0,
-      "step": 20998
-    },
-    {
-      "epoch": 1.6055201942007378,
-      "grad_norm": 0.0004764936165884137,
-      "learning_rate": 0.00019999872917044357,
-      "loss": 46.0,
-      "step": 20999
-    },
-    {
-      "epoch": 1.6055966511841275,
-      "grad_norm": 0.00035234479582868516,
-      "learning_rate": 0.0001999987290493461,
-      "loss": 46.0,
-      "step": 21000
-    },
-    {
-      "epoch": 1.6056731081675173,
-      "grad_norm": 0.00030260623316280544,
-      "learning_rate": 0.00019999872892824288,
-      "loss": 46.0,
-      "step": 21001
-    },
-    {
-      "epoch": 1.6057495651509068,
-      "grad_norm": 0.0012927382485941052,
-      "learning_rate": 0.00019999872880713392,
-      "loss": 46.0,
-      "step": 21002
-    },
-    {
-      "epoch": 1.6058260221342966,
-      "grad_norm": 0.002184497192502022,
-      "learning_rate": 0.00019999872868601916,
-      "loss": 46.0,
-      "step": 21003
-    },
-    {
-      "epoch": 1.6059024791176864,
-      "grad_norm": 0.0009127986850216985,
-      "learning_rate": 0.00019999872856489863,
-      "loss": 46.0,
-      "step": 21004
-    },
-    {
-      "epoch": 1.6059789361010761,
-      "grad_norm": 0.0005975909298285842,
-      "learning_rate": 0.00019999872844377238,
-      "loss": 46.0,
-      "step": 21005
-    },
-    {
-      "epoch": 1.6060553930844659,
-      "grad_norm": 0.0009857243858277798,
-      "learning_rate": 0.0001999987283226403,
-      "loss": 46.0,
-      "step": 21006
-    },
-    {
-      "epoch": 1.6061318500678556,
-      "grad_norm": 0.0007776133716106415,
-      "learning_rate": 0.00019999872820150247,
-      "loss": 46.0,
-      "step": 21007
-    },
-    {
-      "epoch": 1.6062083070512454,
-      "grad_norm": 0.0022794189862906933,
-      "learning_rate": 0.00019999872808035884,
-      "loss": 46.0,
-      "step": 21008
-    },
-    {
-      "epoch": 1.6062847640346352,
-      "grad_norm": 0.001021401141770184,
-      "learning_rate": 0.0001999987279592095,
-      "loss": 46.0,
-      "step": 21009
-    },
-    {
-      "epoch": 1.6063612210180247,
-      "grad_norm": 0.00104917970020324,
-      "learning_rate": 0.00019999872783805434,
-      "loss": 46.0,
-      "step": 21010
-    },
-    {
-      "epoch": 1.6064376780014145,
-      "grad_norm": 0.0013767699711024761,
-      "learning_rate": 0.00019999872771689345,
-      "loss": 46.0,
-      "step": 21011
-    },
-    {
-      "epoch": 1.6065141349848042,
-      "grad_norm": 0.000705127080436796,
-      "learning_rate": 0.00019999872759572678,
-      "loss": 46.0,
-      "step": 21012
-    },
-    {
-      "epoch": 1.6065905919681938,
-      "grad_norm": 0.0005498520331457257,
-      "learning_rate": 0.00019999872747455432,
-      "loss": 46.0,
-      "step": 21013
-    },
-    {
-      "epoch": 1.6066670489515835,
-      "grad_norm": 0.00036141107557341456,
-      "learning_rate": 0.0001999987273533761,
-      "loss": 46.0,
-      "step": 21014
-    },
-    {
-      "epoch": 1.6067435059349733,
-      "grad_norm": 0.0013553696917369962,
-      "learning_rate": 0.00019999872723219211,
-      "loss": 46.0,
-      "step": 21015
-    },
-    {
-      "epoch": 1.606819962918363,
-      "grad_norm": 0.002571397228166461,
-      "learning_rate": 0.00019999872711100235,
-      "loss": 46.0,
-      "step": 21016
-    },
-    {
-      "epoch": 1.6068964199017528,
-      "grad_norm": 0.0021386868320405483,
-      "learning_rate": 0.00019999872698980685,
-      "loss": 46.0,
-      "step": 21017
-    },
-    {
-      "epoch": 1.6069728768851426,
-      "grad_norm": 0.0035288440994918346,
-      "learning_rate": 0.0001999987268686055,
-      "loss": 46.0,
-      "step": 21018
-    },
-    {
-      "epoch": 1.6070493338685323,
-      "grad_norm": 0.004163360223174095,
-      "learning_rate": 0.00019999872674739849,
-      "loss": 46.0,
-      "step": 21019
-    },
-    {
-      "epoch": 1.607125790851922,
-      "grad_norm": 0.002563976449891925,
-      "learning_rate": 0.00019999872662618563,
-      "loss": 46.0,
-      "step": 21020
-    },
-    {
-      "epoch": 1.6072022478353116,
-      "grad_norm": 0.00160535320173949,
-      "learning_rate": 0.00019999872650496703,
-      "loss": 46.0,
-      "step": 21021
-    },
-    {
-      "epoch": 1.6072787048187014,
-      "grad_norm": 0.0015865828609094024,
-      "learning_rate": 0.00019999872638374266,
-      "loss": 46.0,
-      "step": 21022
-    },
-    {
-      "epoch": 1.6073551618020911,
-      "grad_norm": 0.0014409509021788836,
-      "learning_rate": 0.00019999872626251254,
-      "loss": 46.0,
-      "step": 21023
-    },
-    {
-      "epoch": 1.6074316187854807,
-      "grad_norm": 0.0006730647874064744,
-      "learning_rate": 0.00019999872614127662,
-      "loss": 46.0,
-      "step": 21024
-    },
-    {
-      "epoch": 1.6075080757688704,
-      "grad_norm": 0.0020042809192091227,
-      "learning_rate": 0.00019999872602003492,
-      "loss": 46.0,
-      "step": 21025
-    },
-    {
-      "epoch": 1.6075845327522602,
-      "grad_norm": 0.0018369007157161832,
-      "learning_rate": 0.00019999872589878748,
-      "loss": 46.0,
-      "step": 21026
-    },
-    {
-      "epoch": 1.60766098973565,
-      "grad_norm": 0.0008264737552963197,
-      "learning_rate": 0.00019999872577753427,
-      "loss": 46.0,
-      "step": 21027
-    },
-    {
-      "epoch": 1.6077374467190397,
-      "grad_norm": 0.010046363808214664,
-      "learning_rate": 0.00019999872565627528,
-      "loss": 46.0,
-      "step": 21028
-    },
-    {
-      "epoch": 1.6078139037024295,
-      "grad_norm": 0.0007724351598881185,
-      "learning_rate": 0.00019999872553501052,
-      "loss": 46.0,
-      "step": 21029
-    },
-    {
-      "epoch": 1.6078903606858193,
-      "grad_norm": 0.0007708275225013494,
-      "learning_rate": 0.00019999872541374001,
-      "loss": 46.0,
-      "step": 21030
-    },
-    {
-      "epoch": 1.607966817669209,
-      "grad_norm": 0.003397758584469557,
-      "learning_rate": 0.0001999987252924637,
-      "loss": 46.0,
-      "step": 21031
-    },
-    {
-      "epoch": 1.6080432746525986,
-      "grad_norm": 0.0015222474467009306,
-      "learning_rate": 0.00019999872517118163,
-      "loss": 46.0,
-      "step": 21032
-    },
-    {
-      "epoch": 1.6081197316359883,
-      "grad_norm": 0.0029681578744202852,
-      "learning_rate": 0.0001999987250498938,
-      "loss": 46.0,
-      "step": 21033
-    },
-    {
-      "epoch": 1.608196188619378,
-      "grad_norm": 0.0019044431392103434,
-      "learning_rate": 0.0001999987249286002,
-      "loss": 46.0,
-      "step": 21034
-    },
-    {
-      "epoch": 1.6082726456027676,
-      "grad_norm": 0.002594572026282549,
-      "learning_rate": 0.00019999872480730085,
-      "loss": 46.0,
-      "step": 21035
-    },
-    {
-      "epoch": 1.6083491025861574,
-      "grad_norm": 0.003152968129143119,
-      "learning_rate": 0.0001999987246859957,
-      "loss": 46.0,
-      "step": 21036
-    },
-    {
-      "epoch": 1.6084255595695471,
-      "grad_norm": 0.001157531514763832,
-      "learning_rate": 0.00019999872456468479,
-      "loss": 46.0,
-      "step": 21037
-    },
-    {
-      "epoch": 1.608502016552937,
-      "grad_norm": 0.0016753802774474025,
-      "learning_rate": 0.0001999987244433681,
-      "loss": 46.0,
-      "step": 21038
-    },
-    {
-      "epoch": 1.6085784735363267,
-      "grad_norm": 0.0009453432867303491,
-      "learning_rate": 0.00019999872432204563,
-      "loss": 46.0,
-      "step": 21039
-    },
-    {
-      "epoch": 1.6086549305197164,
-      "grad_norm": 0.002466905862092972,
-      "learning_rate": 0.0001999987242007174,
-      "loss": 46.0,
-      "step": 21040
-    },
-    {
-      "epoch": 1.6087313875031062,
-      "grad_norm": 0.000778735731728375,
-      "learning_rate": 0.00019999872407938343,
-      "loss": 46.0,
-      "step": 21041
-    },
-    {
-      "epoch": 1.608807844486496,
-      "grad_norm": 0.0022954270243644714,
-      "learning_rate": 0.00019999872395804367,
-      "loss": 46.0,
-      "step": 21042
-    },
-    {
-      "epoch": 1.6088843014698855,
-      "grad_norm": 0.0006725925486534834,
-      "learning_rate": 0.00019999872383669813,
-      "loss": 46.0,
-      "step": 21043
-    },
-    {
-      "epoch": 1.6089607584532752,
-      "grad_norm": 0.004365310538560152,
-      "learning_rate": 0.00019999872371534685,
-      "loss": 46.0,
-      "step": 21044
-    },
-    {
-      "epoch": 1.6090372154366648,
-      "grad_norm": 0.0019086155807599425,
-      "learning_rate": 0.0001999987235939898,
-      "loss": 46.0,
-      "step": 21045
-    },
-    {
-      "epoch": 1.6091136724200545,
-      "grad_norm": 0.0006107357912696898,
-      "learning_rate": 0.00019999872347262695,
-      "loss": 46.0,
-      "step": 21046
-    },
-    {
-      "epoch": 1.6091901294034443,
-      "grad_norm": 0.00041029637213796377,
-      "learning_rate": 0.00019999872335125835,
-      "loss": 46.0,
-      "step": 21047
-    },
-    {
-      "epoch": 1.609266586386834,
-      "grad_norm": 0.0011750655248761177,
-      "learning_rate": 0.00019999872322988395,
-      "loss": 46.0,
-      "step": 21048
-    },
-    {
-      "epoch": 1.6093430433702238,
-      "grad_norm": 0.002391480142250657,
-      "learning_rate": 0.00019999872310850383,
-      "loss": 46.0,
-      "step": 21049
-    },
-    {
-      "epoch": 1.6094195003536136,
-      "grad_norm": 0.00286471308209002,
-      "learning_rate": 0.0001999987229871179,
-      "loss": 46.0,
-      "step": 21050
-    },
-    {
-      "epoch": 1.6094959573370033,
-      "grad_norm": 0.008997954428195953,
-      "learning_rate": 0.0001999987228657262,
-      "loss": 46.0,
-      "step": 21051
-    },
-    {
-      "epoch": 1.609572414320393,
-      "grad_norm": 0.0017753532156348228,
-      "learning_rate": 0.00019999872274432877,
-      "loss": 46.0,
-      "step": 21052
-    },
-    {
-      "epoch": 1.6096488713037829,
-      "grad_norm": 0.003001385135576129,
-      "learning_rate": 0.00019999872262292553,
-      "loss": 46.0,
-      "step": 21053
-    },
-    {
-      "epoch": 1.6097253282871724,
-      "grad_norm": 0.0018249507993459702,
-      "learning_rate": 0.00019999872250151655,
-      "loss": 46.0,
-      "step": 21054
-    },
-    {
-      "epoch": 1.6098017852705622,
-      "grad_norm": 0.0012306715361773968,
-      "learning_rate": 0.0001999987223801018,
-      "loss": 46.0,
-      "step": 21055
-    },
-    {
-      "epoch": 1.6098782422539517,
-      "grad_norm": 0.0009357483359053731,
-      "learning_rate": 0.00019999872225868125,
-      "loss": 46.0,
-      "step": 21056
-    },
-    {
-      "epoch": 1.6099546992373415,
-      "grad_norm": 0.002308095572516322,
-      "learning_rate": 0.00019999872213725497,
-      "loss": 46.0,
-      "step": 21057
-    },
-    {
-      "epoch": 1.6100311562207312,
-      "grad_norm": 0.0007877701427787542,
-      "learning_rate": 0.0001999987220158229,
-      "loss": 46.0,
-      "step": 21058
-    },
-    {
-      "epoch": 1.610107613204121,
-      "grad_norm": 0.0009794340003281832,
-      "learning_rate": 0.00019999872189438504,
-      "loss": 46.0,
-      "step": 21059
-    },
-    {
-      "epoch": 1.6101840701875108,
-      "grad_norm": 0.0013161645038053393,
-      "learning_rate": 0.00019999872177294147,
-      "loss": 46.0,
-      "step": 21060
-    },
-    {
-      "epoch": 1.6102605271709005,
-      "grad_norm": 0.0018214213196188211,
-      "learning_rate": 0.00019999872165149207,
-      "loss": 46.0,
-      "step": 21061
-    },
-    {
-      "epoch": 1.6103369841542903,
-      "grad_norm": 0.0011455663479864597,
-      "learning_rate": 0.00019999872153003692,
-      "loss": 46.0,
-      "step": 21062
-    },
-    {
-      "epoch": 1.61041344113768,
-      "grad_norm": 0.0019520290661603212,
-      "learning_rate": 0.00019999872140857598,
-      "loss": 46.0,
-      "step": 21063
-    },
-    {
-      "epoch": 1.6104898981210698,
-      "grad_norm": 0.0005044558201916516,
-      "learning_rate": 0.00019999872128710934,
-      "loss": 46.0,
-      "step": 21064
-    },
-    {
-      "epoch": 1.6105663551044593,
-      "grad_norm": 0.0005181654705666006,
-      "learning_rate": 0.00019999872116563684,
-      "loss": 46.0,
-      "step": 21065
-    },
-    {
-      "epoch": 1.610642812087849,
-      "grad_norm": 0.0013047849060967565,
-      "learning_rate": 0.00019999872104415863,
-      "loss": 46.0,
-      "step": 21066
-    },
-    {
-      "epoch": 1.6107192690712386,
-      "grad_norm": 0.0010704946471378207,
-      "learning_rate": 0.00019999872092267464,
-      "loss": 46.0,
-      "step": 21067
-    },
-    {
-      "epoch": 1.6107957260546284,
-      "grad_norm": 0.0009067317587323487,
-      "learning_rate": 0.00019999872080118486,
-      "loss": 46.0,
-      "step": 21068
-    },
-    {
-      "epoch": 1.6108721830380182,
-      "grad_norm": 0.001266265520825982,
-      "learning_rate": 0.00019999872067968935,
-      "loss": 46.0,
-      "step": 21069
-    },
-    {
-      "epoch": 1.610948640021408,
-      "grad_norm": 0.00047098510549403727,
-      "learning_rate": 0.00019999872055818805,
-      "loss": 46.0,
-      "step": 21070
-    },
-    {
-      "epoch": 1.6110250970047977,
-      "grad_norm": 0.000976894167251885,
-      "learning_rate": 0.00019999872043668097,
-      "loss": 46.0,
-      "step": 21071
-    },
-    {
-      "epoch": 1.6111015539881874,
-      "grad_norm": 0.0010359307052567601,
-      "learning_rate": 0.00019999872031516814,
-      "loss": 46.0,
-      "step": 21072
-    },
-    {
-      "epoch": 1.6111780109715772,
-      "grad_norm": 0.0003913237014785409,
-      "learning_rate": 0.00019999872019364951,
-      "loss": 46.0,
-      "step": 21073
-    },
-    {
-      "epoch": 1.611254467954967,
-      "grad_norm": 0.0019767077174037695,
-      "learning_rate": 0.00019999872007212514,
-      "loss": 46.0,
-      "step": 21074
-    },
-    {
-      "epoch": 1.6113309249383567,
-      "grad_norm": 0.0036628753878176212,
-      "learning_rate": 0.000199998719950595,
-      "loss": 46.0,
-      "step": 21075
-    },
-    {
-      "epoch": 1.6114073819217463,
-      "grad_norm": 0.0015950137749314308,
-      "learning_rate": 0.00019999871982905908,
-      "loss": 46.0,
-      "step": 21076
-    },
-    {
-      "epoch": 1.611483838905136,
-      "grad_norm": 0.0016126794507727027,
-      "learning_rate": 0.0001999987197075174,
-      "loss": 46.0,
-      "step": 21077
-    },
-    {
-      "epoch": 1.6115602958885256,
-      "grad_norm": 0.0012470823712646961,
-      "learning_rate": 0.00019999871958596992,
-      "loss": 46.0,
-      "step": 21078
-    },
-    {
-      "epoch": 1.6116367528719153,
-      "grad_norm": 0.0007921934593468904,
-      "learning_rate": 0.0001999987194644167,
-      "loss": 46.0,
-      "step": 21079
-    },
-    {
-      "epoch": 1.611713209855305,
-      "grad_norm": 0.0032635461539030075,
-      "learning_rate": 0.00019999871934285772,
-      "loss": 46.0,
-      "step": 21080
-    },
-    {
-      "epoch": 1.6117896668386948,
-      "grad_norm": 0.0022898688912391663,
-      "learning_rate": 0.00019999871922129293,
-      "loss": 46.0,
-      "step": 21081
-    },
-    {
-      "epoch": 1.6118661238220846,
-      "grad_norm": 0.000564819376450032,
-      "learning_rate": 0.0001999987190997224,
-      "loss": 46.0,
-      "step": 21082
-    },
-    {
-      "epoch": 1.6119425808054744,
-      "grad_norm": 0.004197360016405582,
-      "learning_rate": 0.0001999987189781461,
-      "loss": 46.0,
-      "step": 21083
-    },
-    {
-      "epoch": 1.6120190377888641,
-      "grad_norm": 0.0005882505211047828,
-      "learning_rate": 0.00019999871885656404,
-      "loss": 46.0,
-      "step": 21084
-    },
-    {
-      "epoch": 1.612095494772254,
-      "grad_norm": 0.0019654470961540937,
-      "learning_rate": 0.0001999987187349762,
-      "loss": 46.0,
-      "step": 21085
-    },
-    {
-      "epoch": 1.6121719517556434,
-      "grad_norm": 0.002143718535080552,
-      "learning_rate": 0.00019999871861338256,
-      "loss": 46.0,
-      "step": 21086
-    },
-    {
-      "epoch": 1.6122484087390332,
-      "grad_norm": 0.0017097517848014832,
-      "learning_rate": 0.0001999987184917832,
-      "loss": 46.0,
-      "step": 21087
-    },
-    {
-      "epoch": 1.612324865722423,
-      "grad_norm": 0.007222683634608984,
-      "learning_rate": 0.00019999871837017805,
-      "loss": 46.0,
-      "step": 21088
-    },
-    {
-      "epoch": 1.6124013227058125,
-      "grad_norm": 0.0007527847192250192,
-      "learning_rate": 0.00019999871824856713,
-      "loss": 46.0,
-      "step": 21089
-    },
-    {
-      "epoch": 1.6124777796892023,
-      "grad_norm": 0.0006051480304449797,
-      "learning_rate": 0.00019999871812695043,
-      "loss": 46.0,
-      "step": 21090
-    },
-    {
-      "epoch": 1.612554236672592,
-      "grad_norm": 0.0019459350733086467,
-      "learning_rate": 0.000199998718005328,
-      "loss": 46.0,
-      "step": 21091
-    },
-    {
-      "epoch": 1.6126306936559818,
-      "grad_norm": 0.006585201248526573,
-      "learning_rate": 0.00019999871788369976,
-      "loss": 46.0,
-      "step": 21092
-    },
-    {
-      "epoch": 1.6127071506393715,
-      "grad_norm": 0.0038099808152765036,
-      "learning_rate": 0.00019999871776206574,
-      "loss": 46.0,
-      "step": 21093
-    },
-    {
-      "epoch": 1.6127836076227613,
-      "grad_norm": 0.003795116674154997,
-      "learning_rate": 0.00019999871764042596,
-      "loss": 46.0,
-      "step": 21094
-    },
-    {
-      "epoch": 1.612860064606151,
-      "grad_norm": 0.0031860575545579195,
-      "learning_rate": 0.00019999871751878045,
-      "loss": 46.0,
-      "step": 21095
-    },
-    {
-      "epoch": 1.6129365215895408,
-      "grad_norm": 0.0007133377366699278,
-      "learning_rate": 0.00019999871739712915,
-      "loss": 46.0,
-      "step": 21096
-    },
-    {
-      "epoch": 1.6130129785729304,
-      "grad_norm": 0.0025748417247086763,
-      "learning_rate": 0.00019999871727547207,
-      "loss": 46.0,
-      "step": 21097
-    },
-    {
-      "epoch": 1.6130894355563201,
-      "grad_norm": 0.0005095878150314093,
-      "learning_rate": 0.00019999871715380922,
-      "loss": 46.0,
-      "step": 21098
-    },
-    {
-      "epoch": 1.6131658925397099,
-      "grad_norm": 0.00039984722388908267,
-      "learning_rate": 0.00019999871703214062,
-      "loss": 46.0,
-      "step": 21099
-    },
-    {
-      "epoch": 1.6132423495230994,
-      "grad_norm": 0.0013893496943637729,
-      "learning_rate": 0.00019999871691046625,
-      "loss": 46.0,
-      "step": 21100
-    },
-    {
-      "epoch": 1.6133188065064892,
-      "grad_norm": 0.0025907151866704226,
-      "learning_rate": 0.00019999871678878607,
-      "loss": 46.0,
-      "step": 21101
-    },
-    {
-      "epoch": 1.613395263489879,
-      "grad_norm": 0.0005438943626359105,
-      "learning_rate": 0.00019999871666710015,
-      "loss": 46.0,
-      "step": 21102
-    },
-    {
-      "epoch": 1.6134717204732687,
-      "grad_norm": 0.0031791573856025934,
-      "learning_rate": 0.00019999871654540846,
-      "loss": 46.0,
-      "step": 21103
-    },
-    {
-      "epoch": 1.6135481774566585,
-      "grad_norm": 0.0009783757850527763,
-      "learning_rate": 0.000199998716423711,
-      "loss": 46.0,
-      "step": 21104
-    },
-    {
-      "epoch": 1.6136246344400482,
-      "grad_norm": 0.0034628184512257576,
-      "learning_rate": 0.00019999871630200776,
-      "loss": 46.0,
-      "step": 21105
-    },
-    {
-      "epoch": 1.613701091423438,
-      "grad_norm": 0.0014042137190699577,
-      "learning_rate": 0.00019999871618029877,
-      "loss": 46.0,
-      "step": 21106
-    },
-    {
-      "epoch": 1.6137775484068277,
-      "grad_norm": 0.0008561168215237558,
-      "learning_rate": 0.00019999871605858401,
-      "loss": 46.0,
-      "step": 21107
-    },
-    {
-      "epoch": 1.6138540053902173,
-      "grad_norm": 0.0017930029425770044,
-      "learning_rate": 0.00019999871593686348,
-      "loss": 46.0,
-      "step": 21108
-    },
-    {
-      "epoch": 1.613930462373607,
-      "grad_norm": 0.011705402284860611,
-      "learning_rate": 0.00019999871581513715,
-      "loss": 46.0,
-      "step": 21109
-    },
-    {
-      "epoch": 1.6140069193569968,
-      "grad_norm": 0.0010347990319132805,
-      "learning_rate": 0.00019999871569340507,
-      "loss": 46.0,
-      "step": 21110
-    },
-    {
-      "epoch": 1.6140833763403863,
-      "grad_norm": 0.001082845265045762,
-      "learning_rate": 0.00019999871557166725,
-      "loss": 46.0,
-      "step": 21111
-    },
-    {
-      "epoch": 1.614159833323776,
-      "grad_norm": 0.0012829214101657271,
-      "learning_rate": 0.00019999871544992362,
-      "loss": 46.0,
-      "step": 21112
-    },
-    {
-      "epoch": 1.6142362903071659,
-      "grad_norm": 0.0012234133901074529,
-      "learning_rate": 0.00019999871532817425,
-      "loss": 46.0,
-      "step": 21113
-    },
-    {
-      "epoch": 1.6143127472905556,
-      "grad_norm": 0.000985841965302825,
-      "learning_rate": 0.00019999871520641905,
-      "loss": 46.0,
-      "step": 21114
-    },
-    {
-      "epoch": 1.6143892042739454,
-      "grad_norm": 0.0015574110439047217,
-      "learning_rate": 0.00019999871508465816,
-      "loss": 46.0,
-      "step": 21115
-    },
-    {
-      "epoch": 1.6144656612573351,
-      "grad_norm": 0.0017301106126978993,
-      "learning_rate": 0.00019999871496289147,
-      "loss": 46.0,
-      "step": 21116
-    },
-    {
-      "epoch": 1.614542118240725,
-      "grad_norm": 0.0008492065244354308,
-      "learning_rate": 0.000199998714841119,
-      "loss": 46.0,
-      "step": 21117
-    },
-    {
-      "epoch": 1.6146185752241147,
-      "grad_norm": 0.0010417180601507425,
-      "learning_rate": 0.00019999871471934076,
-      "loss": 46.0,
-      "step": 21118
-    },
-    {
-      "epoch": 1.6146950322075042,
-      "grad_norm": 0.0028450845275074244,
-      "learning_rate": 0.00019999871459755678,
-      "loss": 46.0,
-      "step": 21119
-    },
-    {
-      "epoch": 1.614771489190894,
-      "grad_norm": 0.0013455734588205814,
-      "learning_rate": 0.000199998714475767,
-      "loss": 46.0,
-      "step": 21120
-    },
-    {
-      "epoch": 1.6148479461742837,
-      "grad_norm": 0.0004979423829354346,
-      "learning_rate": 0.00019999871435397143,
-      "loss": 46.0,
-      "step": 21121
-    },
-    {
-      "epoch": 1.6149244031576733,
-      "grad_norm": 0.0006722980760969222,
-      "learning_rate": 0.00019999871423217016,
-      "loss": 46.0,
-      "step": 21122
-    },
-    {
-      "epoch": 1.615000860141063,
-      "grad_norm": 0.0014136051759123802,
-      "learning_rate": 0.00019999871411036305,
-      "loss": 46.0,
-      "step": 21123
-    },
-    {
-      "epoch": 1.6150773171244528,
-      "grad_norm": 0.0016931983409449458,
-      "learning_rate": 0.0001999987139885502,
-      "loss": 46.0,
-      "step": 21124
-    },
-    {
-      "epoch": 1.6151537741078426,
-      "grad_norm": 0.0006579659529961646,
-      "learning_rate": 0.0001999987138667316,
-      "loss": 46.0,
-      "step": 21125
-    },
-    {
-      "epoch": 1.6152302310912323,
-      "grad_norm": 0.0007514195749536157,
-      "learning_rate": 0.00019999871374490723,
-      "loss": 46.0,
-      "step": 21126
-    },
-    {
-      "epoch": 1.615306688074622,
-      "grad_norm": 0.0030831689946353436,
-      "learning_rate": 0.00019999871362307706,
-      "loss": 46.0,
-      "step": 21127
-    },
-    {
-      "epoch": 1.6153831450580118,
-      "grad_norm": 0.0005621747695840895,
-      "learning_rate": 0.00019999871350124114,
-      "loss": 46.0,
-      "step": 21128
-    },
-    {
-      "epoch": 1.6154596020414016,
-      "grad_norm": 0.0003873935202136636,
-      "learning_rate": 0.00019999871337939945,
-      "loss": 46.0,
-      "step": 21129
-    },
-    {
-      "epoch": 1.6155360590247911,
-      "grad_norm": 0.0018026749603450298,
-      "learning_rate": 0.00019999871325755198,
-      "loss": 46.0,
-      "step": 21130
-    },
-    {
-      "epoch": 1.615612516008181,
-      "grad_norm": 0.0023601707071065903,
-      "learning_rate": 0.00019999871313569875,
-      "loss": 46.0,
-      "step": 21131
-    },
-    {
-      "epoch": 1.6156889729915707,
-      "grad_norm": 0.0008154114475473762,
-      "learning_rate": 0.00019999871301383973,
-      "loss": 46.0,
-      "step": 21132
-    },
-    {
-      "epoch": 1.6157654299749602,
-      "grad_norm": 0.0023265972267836332,
-      "learning_rate": 0.00019999871289197495,
-      "loss": 46.0,
-      "step": 21133
-    },
-    {
-      "epoch": 1.61584188695835,
-      "grad_norm": 0.0010747710475698113,
-      "learning_rate": 0.0001999987127701044,
-      "loss": 46.0,
-      "step": 21134
-    },
-    {
-      "epoch": 1.6159183439417397,
-      "grad_norm": 0.018500735983252525,
-      "learning_rate": 0.00019999871264822811,
-      "loss": 46.0,
-      "step": 21135
-    },
-    {
-      "epoch": 1.6159948009251295,
-      "grad_norm": 0.0019123135134577751,
-      "learning_rate": 0.000199998712526346,
-      "loss": 46.0,
-      "step": 21136
-    },
-    {
-      "epoch": 1.6160712579085192,
-      "grad_norm": 0.0006049704388715327,
-      "learning_rate": 0.00019999871240445818,
-      "loss": 46.0,
-      "step": 21137
-    },
-    {
-      "epoch": 1.616147714891909,
-      "grad_norm": 0.002371120033785701,
-      "learning_rate": 0.00019999871228256456,
-      "loss": 46.0,
-      "step": 21138
-    },
-    {
-      "epoch": 1.6162241718752988,
-      "grad_norm": 0.001425352762453258,
-      "learning_rate": 0.00019999871216066516,
-      "loss": 46.0,
-      "step": 21139
-    },
-    {
-      "epoch": 1.6163006288586885,
-      "grad_norm": 0.0008812497835606337,
-      "learning_rate": 0.00019999871203876,
-      "loss": 46.0,
-      "step": 21140
-    },
-    {
-      "epoch": 1.616377085842078,
-      "grad_norm": 0.0015051300870254636,
-      "learning_rate": 0.00019999871191684907,
-      "loss": 46.0,
-      "step": 21141
-    },
-    {
-      "epoch": 1.6164535428254678,
-      "grad_norm": 0.009064078330993652,
-      "learning_rate": 0.00019999871179493238,
-      "loss": 46.0,
-      "step": 21142
-    },
-    {
-      "epoch": 1.6165299998088576,
-      "grad_norm": 0.0006886118790134788,
-      "learning_rate": 0.00019999871167300992,
-      "loss": 46.0,
-      "step": 21143
-    },
-    {
-      "epoch": 1.6166064567922471,
-      "grad_norm": 0.0003229297581128776,
-      "learning_rate": 0.0001999987115510817,
-      "loss": 46.0,
-      "step": 21144
-    },
-    {
-      "epoch": 1.6166829137756369,
-      "grad_norm": 0.003070235252380371,
-      "learning_rate": 0.00019999871142914767,
-      "loss": 46.0,
-      "step": 21145
-    },
-    {
-      "epoch": 1.6167593707590266,
-      "grad_norm": 0.000722051365301013,
-      "learning_rate": 0.00019999871130720788,
-      "loss": 46.0,
-      "step": 21146
-    },
-    {
-      "epoch": 1.6168358277424164,
-      "grad_norm": 0.0008982019498944283,
-      "learning_rate": 0.00019999871118526238,
-      "loss": 46.0,
-      "step": 21147
-    },
-    {
-      "epoch": 1.6169122847258062,
-      "grad_norm": 0.0038732513785362244,
-      "learning_rate": 0.00019999871106331102,
-      "loss": 46.0,
-      "step": 21148
-    },
-    {
-      "epoch": 1.616988741709196,
-      "grad_norm": 0.001466869842261076,
-      "learning_rate": 0.00019999871094135397,
-      "loss": 46.0,
-      "step": 21149
-    },
-    {
-      "epoch": 1.6170651986925857,
-      "grad_norm": 0.0037480571772903204,
-      "learning_rate": 0.00019999871081939112,
-      "loss": 46.0,
-      "step": 21150
-    },
-    {
-      "epoch": 1.6171416556759755,
-      "grad_norm": 0.002543717622756958,
-      "learning_rate": 0.0001999987106974225,
-      "loss": 46.0,
-      "step": 21151
-    },
-    {
-      "epoch": 1.617218112659365,
-      "grad_norm": 0.0022300435230135918,
-      "learning_rate": 0.0001999987105754481,
-      "loss": 46.0,
-      "step": 21152
-    },
-    {
-      "epoch": 1.6172945696427548,
-      "grad_norm": 0.0007123122923076153,
-      "learning_rate": 0.00019999871045346793,
-      "loss": 46.0,
-      "step": 21153
-    },
-    {
-      "epoch": 1.6173710266261445,
-      "grad_norm": 0.0015811588382348418,
-      "learning_rate": 0.000199998710331482,
-      "loss": 46.0,
-      "step": 21154
-    },
-    {
-      "epoch": 1.617447483609534,
-      "grad_norm": 0.001919200294651091,
-      "learning_rate": 0.0001999987102094903,
-      "loss": 46.0,
-      "step": 21155
-    },
-    {
-      "epoch": 1.6175239405929238,
-      "grad_norm": 0.0018038144335150719,
-      "learning_rate": 0.00019999871008749285,
-      "loss": 46.0,
-      "step": 21156
-    },
-    {
-      "epoch": 1.6176003975763136,
-      "grad_norm": 0.0008698327583260834,
-      "learning_rate": 0.0001999987099654896,
-      "loss": 46.0,
-      "step": 21157
-    },
-    {
-      "epoch": 1.6176768545597033,
-      "grad_norm": 0.0010338870342820883,
-      "learning_rate": 0.0001999987098434806,
-      "loss": 46.0,
-      "step": 21158
-    },
-    {
-      "epoch": 1.617753311543093,
-      "grad_norm": 0.00043855590047314763,
-      "learning_rate": 0.00019999870972146582,
-      "loss": 46.0,
-      "step": 21159
-    },
-    {
-      "epoch": 1.6178297685264829,
-      "grad_norm": 0.0024216121528297663,
-      "learning_rate": 0.00019999870959944526,
-      "loss": 46.0,
-      "step": 21160
-    },
-    {
-      "epoch": 1.6179062255098726,
-      "grad_norm": 0.0011566090397536755,
-      "learning_rate": 0.00019999870947741893,
-      "loss": 46.0,
-      "step": 21161
-    },
-    {
-      "epoch": 1.6179826824932624,
-      "grad_norm": 0.0005988717311993241,
-      "learning_rate": 0.00019999870935538688,
-      "loss": 46.0,
-      "step": 21162
-    },
-    {
-      "epoch": 1.618059139476652,
-      "grad_norm": 0.001366671989671886,
-      "learning_rate": 0.00019999870923334903,
-      "loss": 46.0,
-      "step": 21163
-    },
-    {
-      "epoch": 1.6181355964600417,
-      "grad_norm": 0.0007778751896694303,
-      "learning_rate": 0.0001999987091113054,
-      "loss": 46.0,
-      "step": 21164
-    },
-    {
-      "epoch": 1.6182120534434314,
-      "grad_norm": 0.004723059944808483,
-      "learning_rate": 0.000199998708989256,
-      "loss": 46.0,
-      "step": 21165
-    },
-    {
-      "epoch": 1.618288510426821,
-      "grad_norm": 0.001515114912763238,
-      "learning_rate": 0.00019999870886720084,
-      "loss": 46.0,
-      "step": 21166
-    },
-    {
-      "epoch": 1.6183649674102107,
-      "grad_norm": 0.0005236520082689822,
-      "learning_rate": 0.0001999987087451399,
-      "loss": 46.0,
-      "step": 21167
-    },
-    {
-      "epoch": 1.6184414243936005,
-      "grad_norm": 0.0003598094917833805,
-      "learning_rate": 0.0001999987086230732,
-      "loss": 46.0,
-      "step": 21168
-    },
-    {
-      "epoch": 1.6185178813769903,
-      "grad_norm": 0.0014772940194234252,
-      "learning_rate": 0.00019999870850100074,
-      "loss": 46.0,
-      "step": 21169
-    },
-    {
-      "epoch": 1.61859433836038,
-      "grad_norm": 0.0005886157159693539,
-      "learning_rate": 0.0001999987083789225,
-      "loss": 46.0,
-      "step": 21170
-    },
-    {
-      "epoch": 1.6186707953437698,
-      "grad_norm": 0.0009562487248331308,
-      "learning_rate": 0.0001999987082568385,
-      "loss": 46.0,
-      "step": 21171
-    },
-    {
-      "epoch": 1.6187472523271595,
-      "grad_norm": 0.0011610715882852674,
-      "learning_rate": 0.0001999987081347487,
-      "loss": 46.0,
-      "step": 21172
-    },
-    {
-      "epoch": 1.6188237093105493,
-      "grad_norm": 0.0016038455069065094,
-      "learning_rate": 0.00019999870801265318,
-      "loss": 46.0,
-      "step": 21173
-    },
-    {
-      "epoch": 1.6189001662939388,
-      "grad_norm": 0.001127162715420127,
-      "learning_rate": 0.00019999870789055185,
-      "loss": 46.0,
-      "step": 21174
-    },
-    {
-      "epoch": 1.6189766232773286,
-      "grad_norm": 0.0011866713175550103,
-      "learning_rate": 0.00019999870776844477,
-      "loss": 46.0,
-      "step": 21175
-    },
-    {
-      "epoch": 1.6190530802607181,
-      "grad_norm": 0.0033893210347741842,
-      "learning_rate": 0.0001999987076463319,
-      "loss": 46.0,
-      "step": 21176
-    },
-    {
-      "epoch": 1.619129537244108,
-      "grad_norm": 0.0013858703896403313,
-      "learning_rate": 0.00019999870752421327,
-      "loss": 46.0,
-      "step": 21177
-    },
-    {
-      "epoch": 1.6192059942274977,
-      "grad_norm": 0.0009029399370774627,
-      "learning_rate": 0.00019999870740208887,
-      "loss": 46.0,
-      "step": 21178
-    },
-    {
-      "epoch": 1.6192824512108874,
-      "grad_norm": 0.0006544990465044975,
-      "learning_rate": 0.00019999870727995873,
-      "loss": 46.0,
-      "step": 21179
-    },
-    {
-      "epoch": 1.6193589081942772,
-      "grad_norm": 0.0007578975055366755,
-      "learning_rate": 0.00019999870715782278,
-      "loss": 46.0,
-      "step": 21180
-    },
-    {
-      "epoch": 1.619435365177667,
-      "grad_norm": 0.0005427899304777384,
-      "learning_rate": 0.00019999870703568107,
-      "loss": 46.0,
-      "step": 21181
-    },
-    {
-      "epoch": 1.6195118221610567,
-      "grad_norm": 0.0015042303130030632,
-      "learning_rate": 0.00019999870691353363,
-      "loss": 46.0,
-      "step": 21182
-    },
-    {
-      "epoch": 1.6195882791444465,
-      "grad_norm": 0.0019409341039136052,
-      "learning_rate": 0.00019999870679138036,
-      "loss": 46.0,
-      "step": 21183
-    },
-    {
-      "epoch": 1.6196647361278362,
-      "grad_norm": 0.001799962599761784,
-      "learning_rate": 0.00019999870666922135,
-      "loss": 46.0,
-      "step": 21184
-    },
-    {
-      "epoch": 1.6197411931112258,
-      "grad_norm": 0.000986735220067203,
-      "learning_rate": 0.00019999870654705657,
-      "loss": 46.0,
-      "step": 21185
-    },
-    {
-      "epoch": 1.6198176500946155,
-      "grad_norm": 0.007555858232080936,
-      "learning_rate": 0.00019999870642488601,
-      "loss": 46.0,
-      "step": 21186
-    },
-    {
-      "epoch": 1.619894107078005,
-      "grad_norm": 0.004765776917338371,
-      "learning_rate": 0.00019999870630270968,
-      "loss": 46.0,
-      "step": 21187
-    },
-    {
-      "epoch": 1.6199705640613948,
-      "grad_norm": 0.0004920654464513063,
-      "learning_rate": 0.0001999987061805276,
-      "loss": 46.0,
-      "step": 21188
-    },
-    {
-      "epoch": 1.6200470210447846,
-      "grad_norm": 0.0005713792634196579,
-      "learning_rate": 0.00019999870605833976,
-      "loss": 46.0,
-      "step": 21189
-    },
-    {
-      "epoch": 1.6201234780281744,
-      "grad_norm": 0.0009968169033527374,
-      "learning_rate": 0.0001999987059361461,
-      "loss": 46.0,
-      "step": 21190
-    },
-    {
-      "epoch": 1.6201999350115641,
-      "grad_norm": 0.0007498997147195041,
-      "learning_rate": 0.00019999870581394674,
-      "loss": 46.0,
-      "step": 21191
-    },
-    {
-      "epoch": 1.6202763919949539,
-      "grad_norm": 0.000861411914229393,
-      "learning_rate": 0.00019999870569174157,
-      "loss": 46.0,
-      "step": 21192
-    },
-    {
-      "epoch": 1.6203528489783436,
-      "grad_norm": 0.0004635646764654666,
-      "learning_rate": 0.00019999870556953062,
-      "loss": 46.0,
-      "step": 21193
-    },
-    {
-      "epoch": 1.6204293059617334,
-      "grad_norm": 0.0007969801663421094,
-      "learning_rate": 0.00019999870544731393,
-      "loss": 46.0,
-      "step": 21194
-    },
-    {
-      "epoch": 1.6205057629451232,
-      "grad_norm": 0.001320715295150876,
-      "learning_rate": 0.00019999870532509144,
-      "loss": 46.0,
-      "step": 21195
-    },
-    {
-      "epoch": 1.6205822199285127,
-      "grad_norm": 0.001197740202769637,
-      "learning_rate": 0.0001999987052028632,
-      "loss": 46.0,
-      "step": 21196
-    },
-    {
-      "epoch": 1.6206586769119025,
-      "grad_norm": 0.001280443393625319,
-      "learning_rate": 0.0001999987050806292,
-      "loss": 46.0,
-      "step": 21197
-    },
-    {
-      "epoch": 1.620735133895292,
-      "grad_norm": 0.0006781668635085225,
-      "learning_rate": 0.0001999987049583894,
-      "loss": 46.0,
-      "step": 21198
-    },
-    {
-      "epoch": 1.6208115908786818,
-      "grad_norm": 0.006907659117132425,
-      "learning_rate": 0.00019999870483614386,
-      "loss": 46.0,
-      "step": 21199
-    },
-    {
-      "epoch": 1.6208880478620715,
-      "grad_norm": 0.001500763464719057,
-      "learning_rate": 0.00019999870471389252,
-      "loss": 46.0,
-      "step": 21200
-    },
-    {
-      "epoch": 1.6209645048454613,
-      "grad_norm": 0.003449493320658803,
-      "learning_rate": 0.00019999870459163542,
-      "loss": 46.0,
-      "step": 21201
-    },
-    {
-      "epoch": 1.621040961828851,
-      "grad_norm": 0.0005858216900378466,
-      "learning_rate": 0.0001999987044693726,
-      "loss": 46.0,
-      "step": 21202
-    },
-    {
-      "epoch": 1.6211174188122408,
-      "grad_norm": 0.003766548354178667,
-      "learning_rate": 0.00019999870434710398,
-      "loss": 46.0,
-      "step": 21203
-    },
-    {
-      "epoch": 1.6211938757956306,
-      "grad_norm": 0.0012942487373948097,
-      "learning_rate": 0.00019999870422482955,
-      "loss": 46.0,
-      "step": 21204
-    },
-    {
-      "epoch": 1.6212703327790203,
-      "grad_norm": 0.0017242156900465488,
-      "learning_rate": 0.00019999870410254938,
-      "loss": 46.0,
-      "step": 21205
-    },
-    {
-      "epoch": 1.62134678976241,
-      "grad_norm": 0.0007253267103806138,
-      "learning_rate": 0.00019999870398026347,
-      "loss": 46.0,
-      "step": 21206
-    },
-    {
-      "epoch": 1.6214232467457996,
-      "grad_norm": 0.0013813120312988758,
-      "learning_rate": 0.00019999870385797175,
-      "loss": 46.0,
-      "step": 21207
-    },
-    {
-      "epoch": 1.6214997037291894,
-      "grad_norm": 0.0028068660758435726,
-      "learning_rate": 0.00019999870373567426,
-      "loss": 46.0,
-      "step": 21208
-    },
-    {
-      "epoch": 1.621576160712579,
-      "grad_norm": 0.001091246958822012,
-      "learning_rate": 0.00019999870361337102,
-      "loss": 46.0,
-      "step": 21209
-    },
-    {
-      "epoch": 1.6216526176959687,
-      "grad_norm": 0.0030145449563860893,
-      "learning_rate": 0.000199998703491062,
-      "loss": 46.0,
-      "step": 21210
-    },
-    {
-      "epoch": 1.6217290746793585,
-      "grad_norm": 0.004196884576231241,
-      "learning_rate": 0.00019999870336874723,
-      "loss": 46.0,
-      "step": 21211
-    },
-    {
-      "epoch": 1.6218055316627482,
-      "grad_norm": 0.0019495775923132896,
-      "learning_rate": 0.0001999987032464267,
-      "loss": 46.0,
-      "step": 21212
-    },
-    {
-      "epoch": 1.621881988646138,
-      "grad_norm": 0.0009065186022780836,
-      "learning_rate": 0.00019999870312410037,
-      "loss": 46.0,
-      "step": 21213
-    },
-    {
-      "epoch": 1.6219584456295277,
-      "grad_norm": 0.006810011342167854,
-      "learning_rate": 0.00019999870300176827,
-      "loss": 46.0,
-      "step": 21214
-    },
-    {
-      "epoch": 1.6220349026129175,
-      "grad_norm": 0.0029011250007897615,
-      "learning_rate": 0.0001999987028794304,
-      "loss": 46.0,
-      "step": 21215
-    },
-    {
-      "epoch": 1.6221113595963073,
-      "grad_norm": 0.0010498181218281388,
-      "learning_rate": 0.00019999870275708677,
-      "loss": 46.0,
-      "step": 21216
-    },
-    {
-      "epoch": 1.6221878165796968,
-      "grad_norm": 0.0014272972475737333,
-      "learning_rate": 0.00019999870263473737,
-      "loss": 46.0,
-      "step": 21217
-    },
-    {
-      "epoch": 1.6222642735630866,
-      "grad_norm": 0.0006101102917455137,
-      "learning_rate": 0.0001999987025123822,
-      "loss": 46.0,
-      "step": 21218
-    },
-    {
-      "epoch": 1.6223407305464763,
-      "grad_norm": 0.007509036920964718,
-      "learning_rate": 0.00019999870239002126,
-      "loss": 46.0,
-      "step": 21219
-    },
-    {
-      "epoch": 1.6224171875298659,
-      "grad_norm": 0.001004537334665656,
-      "learning_rate": 0.00019999870226765457,
-      "loss": 46.0,
-      "step": 21220
-    },
-    {
-      "epoch": 1.6224936445132556,
-      "grad_norm": 0.005441855173557997,
-      "learning_rate": 0.00019999870214528208,
-      "loss": 46.0,
-      "step": 21221
-    },
-    {
-      "epoch": 1.6225701014966454,
-      "grad_norm": 0.0012853537918999791,
-      "learning_rate": 0.00019999870202290382,
-      "loss": 46.0,
-      "step": 21222
-    },
-    {
-      "epoch": 1.6226465584800351,
-      "grad_norm": 0.004511438775807619,
-      "learning_rate": 0.00019999870190051978,
-      "loss": 46.0,
-      "step": 21223
-    },
-    {
-      "epoch": 1.622723015463425,
-      "grad_norm": 0.0020690823439508677,
-      "learning_rate": 0.00019999870177813,
-      "loss": 46.0,
-      "step": 21224
-    },
-    {
-      "epoch": 1.6227994724468147,
-      "grad_norm": 0.0010296042310073972,
-      "learning_rate": 0.00019999870165573447,
-      "loss": 46.0,
-      "step": 21225
-    },
-    {
-      "epoch": 1.6228759294302044,
-      "grad_norm": 0.001650592079386115,
-      "learning_rate": 0.00019999870153333314,
-      "loss": 46.0,
-      "step": 21226
-    },
-    {
-      "epoch": 1.6229523864135942,
-      "grad_norm": 0.001578155905008316,
-      "learning_rate": 0.00019999870141092604,
-      "loss": 46.0,
-      "step": 21227
-    },
-    {
-      "epoch": 1.6230288433969837,
-      "grad_norm": 0.0035745149943977594,
-      "learning_rate": 0.00019999870128851316,
-      "loss": 46.0,
-      "step": 21228
-    },
-    {
-      "epoch": 1.6231053003803735,
-      "grad_norm": 0.0019117856863886118,
-      "learning_rate": 0.00019999870116609456,
-      "loss": 46.0,
-      "step": 21229
-    },
-    {
-      "epoch": 1.6231817573637632,
-      "grad_norm": 0.011858200654387474,
-      "learning_rate": 0.00019999870104367014,
-      "loss": 46.0,
-      "step": 21230
-    },
-    {
-      "epoch": 1.6232582143471528,
-      "grad_norm": 0.0009259468642994761,
-      "learning_rate": 0.00019999870092123997,
-      "loss": 46.0,
-      "step": 21231
-    },
-    {
-      "epoch": 1.6233346713305425,
-      "grad_norm": 0.002775243017822504,
-      "learning_rate": 0.00019999870079880403,
-      "loss": 46.0,
-      "step": 21232
-    },
-    {
-      "epoch": 1.6234111283139323,
-      "grad_norm": 0.004227730445563793,
-      "learning_rate": 0.00019999870067636231,
-      "loss": 46.0,
-      "step": 21233
-    },
-    {
-      "epoch": 1.623487585297322,
-      "grad_norm": 0.002377475146204233,
-      "learning_rate": 0.00019999870055391482,
-      "loss": 46.0,
-      "step": 21234
-    },
-    {
-      "epoch": 1.6235640422807118,
-      "grad_norm": 0.0006435289396904409,
-      "learning_rate": 0.0001999987004314616,
-      "loss": 46.0,
-      "step": 21235
-    },
-    {
-      "epoch": 1.6236404992641016,
-      "grad_norm": 0.008548357523977757,
-      "learning_rate": 0.00019999870030900255,
-      "loss": 46.0,
-      "step": 21236
-    },
-    {
-      "epoch": 1.6237169562474913,
-      "grad_norm": 0.0010516246547922492,
-      "learning_rate": 0.00019999870018653777,
-      "loss": 46.0,
-      "step": 21237
-    },
-    {
-      "epoch": 1.623793413230881,
-      "grad_norm": 0.0008543775766156614,
-      "learning_rate": 0.00019999870006406721,
-      "loss": 46.0,
-      "step": 21238
-    },
-    {
-      "epoch": 1.6238698702142706,
-      "grad_norm": 0.01090431772172451,
-      "learning_rate": 0.0001999986999415909,
-      "loss": 46.0,
-      "step": 21239
-    },
-    {
-      "epoch": 1.6239463271976604,
-      "grad_norm": 0.0021908299531787634,
-      "learning_rate": 0.00019999869981910878,
-      "loss": 46.0,
-      "step": 21240
-    },
-    {
-      "epoch": 1.6240227841810502,
-      "grad_norm": 0.0008095631492324173,
-      "learning_rate": 0.00019999869969662096,
-      "loss": 46.0,
-      "step": 21241
-    },
-    {
-      "epoch": 1.6240992411644397,
-      "grad_norm": 0.0007362104952335358,
-      "learning_rate": 0.00019999869957412728,
-      "loss": 46.0,
-      "step": 21242
-    },
-    {
-      "epoch": 1.6241756981478295,
-      "grad_norm": 0.010303488932549953,
-      "learning_rate": 0.00019999869945162792,
-      "loss": 46.0,
-      "step": 21243
-    },
-    {
-      "epoch": 1.6242521551312192,
-      "grad_norm": 0.0012417277321219444,
-      "learning_rate": 0.00019999869932912272,
-      "loss": 46.0,
-      "step": 21244
-    },
-    {
-      "epoch": 1.624328612114609,
-      "grad_norm": 0.0008282453054562211,
-      "learning_rate": 0.00019999869920661178,
-      "loss": 46.0,
-      "step": 21245
-    },
-    {
-      "epoch": 1.6244050690979988,
-      "grad_norm": 0.004407276399433613,
-      "learning_rate": 0.00019999869908409506,
-      "loss": 46.0,
-      "step": 21246
-    },
-    {
-      "epoch": 1.6244815260813885,
-      "grad_norm": 0.0006785824662074447,
-      "learning_rate": 0.00019999869896157257,
-      "loss": 46.0,
-      "step": 21247
-    },
-    {
-      "epoch": 1.6245579830647783,
-      "grad_norm": 0.0015478740679100156,
-      "learning_rate": 0.0001999986988390443,
-      "loss": 46.0,
-      "step": 21248
-    },
-    {
-      "epoch": 1.624634440048168,
-      "grad_norm": 0.0010299028363078833,
-      "learning_rate": 0.0001999986987165103,
-      "loss": 46.0,
-      "step": 21249
-    },
-    {
-      "epoch": 1.6247108970315576,
-      "grad_norm": 0.000814037979580462,
-      "learning_rate": 0.00019999869859397055,
-      "loss": 46.0,
-      "step": 21250
-    },
-    {
-      "epoch": 1.6247873540149473,
-      "grad_norm": 0.0003908071666955948,
-      "learning_rate": 0.00019999869847142496,
-      "loss": 46.0,
-      "step": 21251
-    },
-    {
-      "epoch": 1.624863810998337,
-      "grad_norm": 0.0013096685288473964,
-      "learning_rate": 0.00019999869834887363,
-      "loss": 46.0,
-      "step": 21252
-    },
-    {
-      "epoch": 1.6249402679817266,
-      "grad_norm": 0.0027929276693612337,
-      "learning_rate": 0.00019999869822631653,
-      "loss": 46.0,
-      "step": 21253
-    },
-    {
-      "epoch": 1.6250167249651164,
-      "grad_norm": 0.0011807382106781006,
-      "learning_rate": 0.00019999869810375366,
-      "loss": 46.0,
-      "step": 21254
-    },
-    {
-      "epoch": 1.6250931819485062,
-      "grad_norm": 0.000671778223477304,
-      "learning_rate": 0.00019999869798118503,
-      "loss": 46.0,
-      "step": 21255
-    },
-    {
-      "epoch": 1.625169638931896,
-      "grad_norm": 0.0007618090603500605,
-      "learning_rate": 0.00019999869785861064,
-      "loss": 46.0,
-      "step": 21256
-    },
-    {
-      "epoch": 1.6252460959152857,
-      "grad_norm": 0.0003669520956464112,
-      "learning_rate": 0.00019999869773603047,
-      "loss": 46.0,
-      "step": 21257
-    },
-    {
-      "epoch": 1.6253225528986754,
-      "grad_norm": 0.0021568224765360355,
-      "learning_rate": 0.00019999869761344453,
-      "loss": 46.0,
-      "step": 21258
-    },
-    {
-      "epoch": 1.6253990098820652,
-      "grad_norm": 0.0021585659123957157,
-      "learning_rate": 0.0001999986974908528,
-      "loss": 46.0,
-      "step": 21259
-    },
-    {
-      "epoch": 1.625475466865455,
-      "grad_norm": 0.0013506005052477121,
-      "learning_rate": 0.00019999869736825532,
-      "loss": 46.0,
-      "step": 21260
-    },
-    {
-      "epoch": 1.6255519238488445,
-      "grad_norm": 0.0021476324182003736,
-      "learning_rate": 0.00019999869724565206,
-      "loss": 46.0,
-      "step": 21261
-    },
-    {
-      "epoch": 1.6256283808322343,
-      "grad_norm": 0.0017908531008288264,
-      "learning_rate": 0.00019999869712304303,
-      "loss": 46.0,
-      "step": 21262
-    },
-    {
-      "epoch": 1.625704837815624,
-      "grad_norm": 0.0031097340397536755,
-      "learning_rate": 0.00019999869700042824,
-      "loss": 46.0,
-      "step": 21263
-    },
-    {
-      "epoch": 1.6257812947990136,
-      "grad_norm": 0.0018535548588261008,
-      "learning_rate": 0.0001999986968778077,
-      "loss": 46.0,
-      "step": 21264
-    },
-    {
-      "epoch": 1.6258577517824033,
-      "grad_norm": 0.002143446123227477,
-      "learning_rate": 0.00019999869675518133,
-      "loss": 46.0,
-      "step": 21265
-    },
-    {
-      "epoch": 1.625934208765793,
-      "grad_norm": 0.006431363523006439,
-      "learning_rate": 0.00019999869663254923,
-      "loss": 46.0,
-      "step": 21266
-    },
-    {
-      "epoch": 1.6260106657491828,
-      "grad_norm": 0.0025239151436835527,
-      "learning_rate": 0.00019999869650991138,
-      "loss": 46.0,
-      "step": 21267
-    },
-    {
-      "epoch": 1.6260871227325726,
-      "grad_norm": 0.002566578332334757,
-      "learning_rate": 0.00019999869638726773,
-      "loss": 46.0,
-      "step": 21268
-    },
-    {
-      "epoch": 1.6261635797159624,
-      "grad_norm": 0.0012660878710448742,
-      "learning_rate": 0.0001999986962646183,
-      "loss": 46.0,
-      "step": 21269
-    },
-    {
-      "epoch": 1.6262400366993521,
-      "grad_norm": 0.0008707016822881997,
-      "learning_rate": 0.00019999869614196314,
-      "loss": 46.0,
-      "step": 21270
-    },
-    {
-      "epoch": 1.626316493682742,
-      "grad_norm": 0.009448220022022724,
-      "learning_rate": 0.0001999986960193022,
-      "loss": 46.0,
-      "step": 21271
-    },
-    {
-      "epoch": 1.6263929506661314,
-      "grad_norm": 0.0014334573643282056,
-      "learning_rate": 0.0001999986958966355,
-      "loss": 46.0,
-      "step": 21272
-    },
-    {
-      "epoch": 1.6264694076495212,
-      "grad_norm": 0.002474881475791335,
-      "learning_rate": 0.000199998695773963,
-      "loss": 46.0,
-      "step": 21273
-    },
-    {
-      "epoch": 1.626545864632911,
-      "grad_norm": 0.0009059604490175843,
-      "learning_rate": 0.00019999869565128476,
-      "loss": 46.0,
-      "step": 21274
-    },
-    {
-      "epoch": 1.6266223216163005,
-      "grad_norm": 0.033154770731925964,
-      "learning_rate": 0.0001999986955286007,
-      "loss": 46.0,
-      "step": 21275
-    },
-    {
-      "epoch": 1.6266987785996903,
-      "grad_norm": 0.004225148353725672,
-      "learning_rate": 0.00019999869540591095,
-      "loss": 46.0,
-      "step": 21276
-    },
-    {
-      "epoch": 1.62677523558308,
-      "grad_norm": 0.00220974232070148,
-      "learning_rate": 0.00019999869528321537,
-      "loss": 46.0,
-      "step": 21277
-    },
-    {
-      "epoch": 1.6268516925664698,
-      "grad_norm": 0.0009935535490512848,
-      "learning_rate": 0.00019999869516051404,
-      "loss": 46.0,
-      "step": 21278
-    },
-    {
-      "epoch": 1.6269281495498595,
-      "grad_norm": 0.001757741323672235,
-      "learning_rate": 0.00019999869503780694,
-      "loss": 46.0,
-      "step": 21279
-    },
-    {
-      "epoch": 1.6270046065332493,
-      "grad_norm": 0.0009551630355417728,
-      "learning_rate": 0.00019999869491509403,
-      "loss": 46.0,
-      "step": 21280
-    },
-    {
-      "epoch": 1.627081063516639,
-      "grad_norm": 0.0005284722428768873,
-      "learning_rate": 0.0001999986947923754,
-      "loss": 46.0,
-      "step": 21281
-    },
-    {
-      "epoch": 1.6271575205000288,
-      "grad_norm": 0.0009793429635465145,
-      "learning_rate": 0.000199998694669651,
-      "loss": 46.0,
-      "step": 21282
-    },
-    {
-      "epoch": 1.6272339774834184,
-      "grad_norm": 0.004317453131079674,
-      "learning_rate": 0.00019999869454692082,
-      "loss": 46.0,
-      "step": 21283
-    },
-    {
-      "epoch": 1.6273104344668081,
-      "grad_norm": 0.0006739678210578859,
-      "learning_rate": 0.00019999869442418485,
-      "loss": 46.0,
-      "step": 21284
-    },
-    {
-      "epoch": 1.6273868914501979,
-      "grad_norm": 0.0038984795100986958,
-      "learning_rate": 0.00019999869430144317,
-      "loss": 46.0,
-      "step": 21285
-    },
-    {
-      "epoch": 1.6274633484335874,
-      "grad_norm": 0.0009606571402400732,
-      "learning_rate": 0.00019999869417869568,
-      "loss": 46.0,
-      "step": 21286
-    },
-    {
-      "epoch": 1.6275398054169772,
-      "grad_norm": 0.0009748979937285185,
-      "learning_rate": 0.00019999869405594242,
-      "loss": 46.0,
-      "step": 21287
-    },
-    {
-      "epoch": 1.627616262400367,
-      "grad_norm": 0.001311673317104578,
-      "learning_rate": 0.0001999986939331834,
-      "loss": 46.0,
-      "step": 21288
-    },
-    {
-      "epoch": 1.6276927193837567,
-      "grad_norm": 0.005168382544070482,
-      "learning_rate": 0.00019999869381041857,
-      "loss": 46.0,
-      "step": 21289
-    },
-    {
-      "epoch": 1.6277691763671465,
-      "grad_norm": 0.0009775989456102252,
-      "learning_rate": 0.00019999869368764802,
-      "loss": 46.0,
-      "step": 21290
-    },
-    {
-      "epoch": 1.6278456333505362,
-      "grad_norm": 0.0008270779508166015,
-      "learning_rate": 0.0001999986935648717,
-      "loss": 46.0,
-      "step": 21291
-    },
-    {
-      "epoch": 1.627922090333926,
-      "grad_norm": 0.0009698302601464093,
-      "learning_rate": 0.0001999986934420896,
-      "loss": 46.0,
-      "step": 21292
-    },
-    {
-      "epoch": 1.6279985473173157,
-      "grad_norm": 0.0003710102755576372,
-      "learning_rate": 0.0001999986933193017,
-      "loss": 46.0,
-      "step": 21293
-    },
-    {
-      "epoch": 1.6280750043007053,
-      "grad_norm": 0.0009183627553284168,
-      "learning_rate": 0.00019999869319650807,
-      "loss": 46.0,
-      "step": 21294
-    },
-    {
-      "epoch": 1.628151461284095,
-      "grad_norm": 0.001310950261540711,
-      "learning_rate": 0.00019999869307370865,
-      "loss": 46.0,
-      "step": 21295
-    },
-    {
-      "epoch": 1.6282279182674848,
-      "grad_norm": 0.0015147174708545208,
-      "learning_rate": 0.00019999869295090348,
-      "loss": 46.0,
-      "step": 21296
-    },
-    {
-      "epoch": 1.6283043752508743,
-      "grad_norm": 0.0018670111894607544,
-      "learning_rate": 0.00019999869282809254,
-      "loss": 46.0,
-      "step": 21297
-    },
-    {
-      "epoch": 1.628380832234264,
-      "grad_norm": 0.007048591505736113,
-      "learning_rate": 0.0001999986927052758,
-      "loss": 46.0,
-      "step": 21298
-    },
-    {
-      "epoch": 1.6284572892176539,
-      "grad_norm": 0.006009828764945269,
-      "learning_rate": 0.0001999986925824533,
-      "loss": 46.0,
-      "step": 21299
-    },
-    {
-      "epoch": 1.6285337462010436,
-      "grad_norm": 0.0004770963860210031,
-      "learning_rate": 0.00019999869245962505,
-      "loss": 46.0,
-      "step": 21300
-    },
-    {
-      "epoch": 1.6286102031844334,
-      "grad_norm": 0.0013777134008705616,
-      "learning_rate": 0.00019999869233679104,
-      "loss": 46.0,
-      "step": 21301
-    },
-    {
-      "epoch": 1.6286866601678232,
-      "grad_norm": 0.0007388416561298072,
-      "learning_rate": 0.00019999869221395123,
-      "loss": 46.0,
-      "step": 21302
-    },
-    {
-      "epoch": 1.628763117151213,
-      "grad_norm": 0.005754141602665186,
-      "learning_rate": 0.00019999869209110568,
-      "loss": 46.0,
-      "step": 21303
-    },
-    {
-      "epoch": 1.6288395741346027,
-      "grad_norm": 0.006719463504850864,
-      "learning_rate": 0.00019999869196825435,
-      "loss": 46.0,
-      "step": 21304
-    },
-    {
-      "epoch": 1.6289160311179922,
-      "grad_norm": 0.0008256753790192306,
-      "learning_rate": 0.00019999869184539725,
-      "loss": 46.0,
-      "step": 21305
-    },
-    {
-      "epoch": 1.628992488101382,
-      "grad_norm": 0.0019987996201962233,
-      "learning_rate": 0.00019999869172253435,
-      "loss": 46.0,
-      "step": 21306
-    },
-    {
-      "epoch": 1.6290689450847717,
-      "grad_norm": 0.0007657418609596789,
-      "learning_rate": 0.0001999986915996657,
-      "loss": 46.0,
-      "step": 21307
-    },
-    {
-      "epoch": 1.6291454020681613,
-      "grad_norm": 0.0013484855880960822,
-      "learning_rate": 0.00019999869147679128,
-      "loss": 46.0,
-      "step": 21308
-    },
-    {
-      "epoch": 1.629221859051551,
-      "grad_norm": 0.0009328161831945181,
-      "learning_rate": 0.00019999869135391108,
-      "loss": 46.0,
-      "step": 21309
-    },
-    {
-      "epoch": 1.6292983160349408,
-      "grad_norm": 0.002216699765995145,
-      "learning_rate": 0.00019999869123102514,
-      "loss": 46.0,
-      "step": 21310
-    },
-    {
-      "epoch": 1.6293747730183306,
-      "grad_norm": 0.0004925195826217532,
-      "learning_rate": 0.00019999869110813343,
-      "loss": 46.0,
-      "step": 21311
-    },
-    {
-      "epoch": 1.6294512300017203,
-      "grad_norm": 0.0007043244550004601,
-      "learning_rate": 0.00019999869098523594,
-      "loss": 46.0,
-      "step": 21312
-    },
-    {
-      "epoch": 1.62952768698511,
-      "grad_norm": 0.000749769969843328,
-      "learning_rate": 0.00019999869086233268,
-      "loss": 46.0,
-      "step": 21313
-    },
-    {
-      "epoch": 1.6296041439684998,
-      "grad_norm": 0.0010399612365290523,
-      "learning_rate": 0.00019999869073942364,
-      "loss": 46.0,
-      "step": 21314
-    },
-    {
-      "epoch": 1.6296806009518896,
-      "grad_norm": 0.00022684960276819766,
-      "learning_rate": 0.00019999869061650884,
-      "loss": 46.0,
-      "step": 21315
-    },
-    {
-      "epoch": 1.6297570579352791,
-      "grad_norm": 0.0020600163843482733,
-      "learning_rate": 0.00019999869049358826,
-      "loss": 46.0,
-      "step": 21316
-    },
-    {
-      "epoch": 1.629833514918669,
-      "grad_norm": 0.0009904297767207026,
-      "learning_rate": 0.00019999869037066193,
-      "loss": 46.0,
-      "step": 21317
-    },
-    {
-      "epoch": 1.6299099719020584,
-      "grad_norm": 0.0018723539542406797,
-      "learning_rate": 0.0001999986902477298,
-      "loss": 46.0,
-      "step": 21318
-    },
-    {
-      "epoch": 1.6299864288854482,
-      "grad_norm": 0.0008943119901232421,
-      "learning_rate": 0.00019999869012479195,
-      "loss": 46.0,
-      "step": 21319
-    },
-    {
-      "epoch": 1.630062885868838,
-      "grad_norm": 0.006255750544369221,
-      "learning_rate": 0.00019999869000184833,
-      "loss": 46.0,
-      "step": 21320
-    },
-    {
-      "epoch": 1.6301393428522277,
-      "grad_norm": 0.0009223469533026218,
-      "learning_rate": 0.00019999868987889889,
-      "loss": 46.0,
-      "step": 21321
-    },
-    {
-      "epoch": 1.6302157998356175,
-      "grad_norm": 0.000810697500128299,
-      "learning_rate": 0.0001999986897559437,
-      "loss": 46.0,
-      "step": 21322
-    },
-    {
-      "epoch": 1.6302922568190072,
-      "grad_norm": 0.0009278429206460714,
-      "learning_rate": 0.00019999868963298275,
-      "loss": 46.0,
-      "step": 21323
-    },
-    {
-      "epoch": 1.630368713802397,
-      "grad_norm": 0.0007479425403289497,
-      "learning_rate": 0.000199998689510016,
-      "loss": 46.0,
-      "step": 21324
-    },
-    {
-      "epoch": 1.6304451707857868,
-      "grad_norm": 0.0029310144018381834,
-      "learning_rate": 0.00019999868938704352,
-      "loss": 46.0,
-      "step": 21325
-    },
-    {
-      "epoch": 1.6305216277691765,
-      "grad_norm": 0.00308302347548306,
-      "learning_rate": 0.00019999868926406526,
-      "loss": 46.0,
-      "step": 21326
-    },
-    {
-      "epoch": 1.630598084752566,
-      "grad_norm": 0.0009149739053100348,
-      "learning_rate": 0.00019999868914108123,
-      "loss": 46.0,
-      "step": 21327
-    },
-    {
-      "epoch": 1.6306745417359558,
-      "grad_norm": 0.0006685683038085699,
-      "learning_rate": 0.00019999868901809142,
-      "loss": 46.0,
-      "step": 21328
-    },
-    {
-      "epoch": 1.6307509987193454,
-      "grad_norm": 0.0008936495869420469,
-      "learning_rate": 0.00019999868889509587,
-      "loss": 46.0,
-      "step": 21329
-    },
-    {
-      "epoch": 1.6308274557027351,
-      "grad_norm": 0.0007385073695331812,
-      "learning_rate": 0.0001999986887720945,
-      "loss": 46.0,
-      "step": 21330
-    },
-    {
-      "epoch": 1.630903912686125,
-      "grad_norm": 0.004384136293083429,
-      "learning_rate": 0.0001999986886490874,
-      "loss": 46.0,
-      "step": 21331
-    },
-    {
-      "epoch": 1.6309803696695147,
-      "grad_norm": 0.0014456650242209435,
-      "learning_rate": 0.00019999868852607454,
-      "loss": 46.0,
-      "step": 21332
-    },
-    {
-      "epoch": 1.6310568266529044,
-      "grad_norm": 0.0010554405162110925,
-      "learning_rate": 0.0001999986884030559,
-      "loss": 46.0,
-      "step": 21333
-    },
-    {
-      "epoch": 1.6311332836362942,
-      "grad_norm": 0.0010284145828336477,
-      "learning_rate": 0.00019999868828003144,
-      "loss": 46.0,
-      "step": 21334
-    },
-    {
-      "epoch": 1.631209740619684,
-      "grad_norm": 0.0005547644104808569,
-      "learning_rate": 0.00019999868815700128,
-      "loss": 46.0,
-      "step": 21335
-    },
-    {
-      "epoch": 1.6312861976030737,
-      "grad_norm": 0.0006440222496166825,
-      "learning_rate": 0.0001999986880339653,
-      "loss": 46.0,
-      "step": 21336
-    },
-    {
-      "epoch": 1.6313626545864635,
-      "grad_norm": 0.001036867150105536,
-      "learning_rate": 0.0001999986879109236,
-      "loss": 46.0,
-      "step": 21337
-    },
-    {
-      "epoch": 1.631439111569853,
-      "grad_norm": 0.000969358894508332,
-      "learning_rate": 0.0001999986877878761,
-      "loss": 46.0,
-      "step": 21338
-    },
-    {
-      "epoch": 1.6315155685532428,
-      "grad_norm": 0.00198317295871675,
-      "learning_rate": 0.00019999868766482285,
-      "loss": 46.0,
-      "step": 21339
-    },
-    {
-      "epoch": 1.6315920255366323,
-      "grad_norm": 0.0007343141478486359,
-      "learning_rate": 0.00019999868754176382,
-      "loss": 46.0,
-      "step": 21340
-    },
-    {
-      "epoch": 1.631668482520022,
-      "grad_norm": 0.0020955230575054884,
-      "learning_rate": 0.00019999868741869898,
-      "loss": 46.0,
-      "step": 21341
-    },
-    {
-      "epoch": 1.6317449395034118,
-      "grad_norm": 0.012542137876152992,
-      "learning_rate": 0.00019999868729562843,
-      "loss": 46.0,
-      "step": 21342
-    },
-    {
-      "epoch": 1.6318213964868016,
-      "grad_norm": 0.0013200690736994147,
-      "learning_rate": 0.00019999868717255207,
-      "loss": 46.0,
-      "step": 21343
-    },
-    {
-      "epoch": 1.6318978534701913,
-      "grad_norm": 0.0005991157959215343,
-      "learning_rate": 0.00019999868704946997,
-      "loss": 46.0,
-      "step": 21344
-    },
-    {
-      "epoch": 1.631974310453581,
-      "grad_norm": 0.0012595861917361617,
-      "learning_rate": 0.00019999868692638207,
-      "loss": 46.0,
-      "step": 21345
-    },
-    {
-      "epoch": 1.6320507674369709,
-      "grad_norm": 0.001718417857773602,
-      "learning_rate": 0.00019999868680328843,
-      "loss": 46.0,
-      "step": 21346
-    },
-    {
-      "epoch": 1.6321272244203606,
-      "grad_norm": 0.0015305429697036743,
-      "learning_rate": 0.000199998686680189,
-      "loss": 46.0,
-      "step": 21347
-    },
-    {
-      "epoch": 1.6322036814037502,
-      "grad_norm": 0.0014196389820426702,
-      "learning_rate": 0.0001999986865570838,
-      "loss": 46.0,
-      "step": 21348
-    },
-    {
-      "epoch": 1.63228013838714,
-      "grad_norm": 0.00282644503749907,
-      "learning_rate": 0.00019999868643397285,
-      "loss": 46.0,
-      "step": 21349
-    },
-    {
-      "epoch": 1.6323565953705297,
-      "grad_norm": 0.0004768500220961869,
-      "learning_rate": 0.00019999868631085613,
-      "loss": 46.0,
-      "step": 21350
-    },
-    {
-      "epoch": 1.6324330523539192,
-      "grad_norm": 0.001017568982206285,
-      "learning_rate": 0.00019999868618773365,
-      "loss": 46.0,
-      "step": 21351
-    },
-    {
-      "epoch": 1.632509509337309,
-      "grad_norm": 0.0008996358956210315,
-      "learning_rate": 0.0001999986860646054,
-      "loss": 46.0,
-      "step": 21352
-    },
-    {
-      "epoch": 1.6325859663206987,
-      "grad_norm": 0.001769957598298788,
-      "learning_rate": 0.00019999868594147135,
-      "loss": 46.0,
-      "step": 21353
-    },
-    {
-      "epoch": 1.6326624233040885,
-      "grad_norm": 0.0006543158669956028,
-      "learning_rate": 0.00019999868581833155,
-      "loss": 46.0,
-      "step": 21354
-    },
-    {
-      "epoch": 1.6327388802874783,
-      "grad_norm": 0.0026587098836898804,
-      "learning_rate": 0.00019999868569518597,
-      "loss": 46.0,
-      "step": 21355
-    },
-    {
-      "epoch": 1.632815337270868,
-      "grad_norm": 0.003121469169855118,
-      "learning_rate": 0.0001999986855720346,
-      "loss": 46.0,
-      "step": 21356
-    },
-    {
-      "epoch": 1.6328917942542578,
-      "grad_norm": 0.0007932318840175867,
-      "learning_rate": 0.0001999986854488775,
-      "loss": 46.0,
-      "step": 21357
-    },
-    {
-      "epoch": 1.6329682512376476,
-      "grad_norm": 0.0013700078707188368,
-      "learning_rate": 0.0001999986853257146,
-      "loss": 46.0,
-      "step": 21358
-    },
-    {
-      "epoch": 1.633044708221037,
-      "grad_norm": 0.0024322369135916233,
-      "learning_rate": 0.00019999868520254597,
-      "loss": 46.0,
-      "step": 21359
-    },
-    {
-      "epoch": 1.6331211652044269,
-      "grad_norm": 0.0010843066265806556,
-      "learning_rate": 0.00019999868507937155,
-      "loss": 46.0,
-      "step": 21360
-    },
-    {
-      "epoch": 1.6331976221878166,
-      "grad_norm": 0.0011280658654868603,
-      "learning_rate": 0.00019999868495619135,
-      "loss": 46.0,
-      "step": 21361
-    },
-    {
-      "epoch": 1.6332740791712062,
-      "grad_norm": 0.0014752084389328957,
-      "learning_rate": 0.00019999868483300541,
-      "loss": 46.0,
-      "step": 21362
-    },
-    {
-      "epoch": 1.633350536154596,
-      "grad_norm": 0.002738103736191988,
-      "learning_rate": 0.00019999868470981367,
-      "loss": 46.0,
-      "step": 21363
-    },
-    {
-      "epoch": 1.6334269931379857,
-      "grad_norm": 0.004372116178274155,
-      "learning_rate": 0.00019999868458661616,
-      "loss": 46.0,
-      "step": 21364
-    },
-    {
-      "epoch": 1.6335034501213754,
-      "grad_norm": 0.0017591556534171104,
-      "learning_rate": 0.0001999986844634129,
-      "loss": 46.0,
-      "step": 21365
-    },
-    {
-      "epoch": 1.6335799071047652,
-      "grad_norm": 0.0006350170588120818,
-      "learning_rate": 0.00019999868434020387,
-      "loss": 46.0,
-      "step": 21366
-    },
-    {
-      "epoch": 1.633656364088155,
-      "grad_norm": 0.0009469349170103669,
-      "learning_rate": 0.00019999868421698906,
-      "loss": 46.0,
-      "step": 21367
-    },
-    {
-      "epoch": 1.6337328210715447,
-      "grad_norm": 0.0034909015521407127,
-      "learning_rate": 0.00019999868409376848,
-      "loss": 46.0,
-      "step": 21368
-    },
-    {
-      "epoch": 1.6338092780549345,
-      "grad_norm": 0.0030245762318372726,
-      "learning_rate": 0.00019999868397054213,
-      "loss": 46.0,
-      "step": 21369
-    },
-    {
-      "epoch": 1.633885735038324,
-      "grad_norm": 0.0003537854354362935,
-      "learning_rate": 0.00019999868384731003,
-      "loss": 46.0,
-      "step": 21370
-    },
-    {
-      "epoch": 1.6339621920217138,
-      "grad_norm": 0.004920335486531258,
-      "learning_rate": 0.00019999868372407213,
-      "loss": 46.0,
-      "step": 21371
-    },
-    {
-      "epoch": 1.6340386490051035,
-      "grad_norm": 0.0012120850151404738,
-      "learning_rate": 0.00019999868360082848,
-      "loss": 46.0,
-      "step": 21372
-    },
-    {
-      "epoch": 1.634115105988493,
-      "grad_norm": 0.005476017948240042,
-      "learning_rate": 0.00019999868347757906,
-      "loss": 46.0,
-      "step": 21373
-    },
-    {
-      "epoch": 1.6341915629718828,
-      "grad_norm": 0.0003926352073904127,
-      "learning_rate": 0.00019999868335432387,
-      "loss": 46.0,
-      "step": 21374
-    },
-    {
-      "epoch": 1.6342680199552726,
-      "grad_norm": 0.001729327137582004,
-      "learning_rate": 0.0001999986832310629,
-      "loss": 46.0,
-      "step": 21375
-    },
-    {
-      "epoch": 1.6343444769386624,
-      "grad_norm": 0.0011682361364364624,
-      "learning_rate": 0.00019999868310779614,
-      "loss": 46.0,
-      "step": 21376
-    },
-    {
-      "epoch": 1.6344209339220521,
-      "grad_norm": 0.0027087063062936068,
-      "learning_rate": 0.00019999868298452368,
-      "loss": 46.0,
-      "step": 21377
-    },
-    {
-      "epoch": 1.6344973909054419,
-      "grad_norm": 0.002049011643975973,
-      "learning_rate": 0.0001999986828612454,
-      "loss": 46.0,
-      "step": 21378
-    },
-    {
-      "epoch": 1.6345738478888316,
-      "grad_norm": 0.0008791850996203721,
-      "learning_rate": 0.00019999868273796136,
-      "loss": 46.0,
-      "step": 21379
-    },
-    {
-      "epoch": 1.6346503048722214,
-      "grad_norm": 0.00478854775428772,
-      "learning_rate": 0.00019999868261467155,
-      "loss": 46.0,
-      "step": 21380
-    },
-    {
-      "epoch": 1.634726761855611,
-      "grad_norm": 0.0020329647231847048,
-      "learning_rate": 0.00019999868249137597,
-      "loss": 46.0,
-      "step": 21381
-    },
-    {
-      "epoch": 1.6348032188390007,
-      "grad_norm": 0.005377105437219143,
-      "learning_rate": 0.00019999868236807462,
-      "loss": 46.0,
-      "step": 21382
-    },
-    {
-      "epoch": 1.6348796758223905,
-      "grad_norm": 0.0012790110195055604,
-      "learning_rate": 0.0001999986822447675,
-      "loss": 46.0,
-      "step": 21383
-    },
-    {
-      "epoch": 1.63495613280578,
-      "grad_norm": 0.0016248412430286407,
-      "learning_rate": 0.0001999986821214546,
-      "loss": 46.0,
-      "step": 21384
-    },
-    {
-      "epoch": 1.6350325897891698,
-      "grad_norm": 0.001474945922382176,
-      "learning_rate": 0.00019999868199813595,
-      "loss": 46.0,
-      "step": 21385
-    },
-    {
-      "epoch": 1.6351090467725595,
-      "grad_norm": 0.0008074503275565803,
-      "learning_rate": 0.00019999868187481156,
-      "loss": 46.0,
-      "step": 21386
-    },
-    {
-      "epoch": 1.6351855037559493,
-      "grad_norm": 0.003150628414005041,
-      "learning_rate": 0.00019999868175148134,
-      "loss": 46.0,
-      "step": 21387
-    },
-    {
-      "epoch": 1.635261960739339,
-      "grad_norm": 0.0004442027129698545,
-      "learning_rate": 0.00019999868162814537,
-      "loss": 46.0,
-      "step": 21388
-    },
-    {
-      "epoch": 1.6353384177227288,
-      "grad_norm": 0.0025579649955034256,
-      "learning_rate": 0.00019999868150480363,
-      "loss": 46.0,
-      "step": 21389
-    },
-    {
-      "epoch": 1.6354148747061186,
-      "grad_norm": 0.0005267733940854669,
-      "learning_rate": 0.00019999868138145615,
-      "loss": 46.0,
-      "step": 21390
-    },
-    {
-      "epoch": 1.6354913316895083,
-      "grad_norm": 0.0007746985065750778,
-      "learning_rate": 0.00019999868125810286,
-      "loss": 46.0,
-      "step": 21391
-    },
-    {
-      "epoch": 1.6355677886728979,
-      "grad_norm": 0.0010149333393201232,
-      "learning_rate": 0.0001999986811347438,
-      "loss": 46.0,
-      "step": 21392
-    },
-    {
-      "epoch": 1.6356442456562876,
-      "grad_norm": 0.0008023123955354095,
-      "learning_rate": 0.00019999868101137902,
-      "loss": 46.0,
-      "step": 21393
-    },
-    {
-      "epoch": 1.6357207026396774,
-      "grad_norm": 0.0007437619497068226,
-      "learning_rate": 0.00019999868088800842,
-      "loss": 46.0,
-      "step": 21394
-    },
-    {
-      "epoch": 1.635797159623067,
-      "grad_norm": 0.0012289745500311255,
-      "learning_rate": 0.0001999986807646321,
-      "loss": 46.0,
-      "step": 21395
-    },
-    {
-      "epoch": 1.6358736166064567,
-      "grad_norm": 0.001823923666961491,
-      "learning_rate": 0.00019999868064124996,
-      "loss": 46.0,
-      "step": 21396
-    },
-    {
-      "epoch": 1.6359500735898465,
-      "grad_norm": 0.0006801133276894689,
-      "learning_rate": 0.00019999868051786206,
-      "loss": 46.0,
-      "step": 21397
-    },
-    {
-      "epoch": 1.6360265305732362,
-      "grad_norm": 0.002444860991090536,
-      "learning_rate": 0.00019999868039446842,
-      "loss": 46.0,
-      "step": 21398
-    },
-    {
-      "epoch": 1.636102987556626,
-      "grad_norm": 0.0004933615564368665,
-      "learning_rate": 0.00019999868027106897,
-      "loss": 46.0,
-      "step": 21399
-    },
-    {
-      "epoch": 1.6361794445400157,
-      "grad_norm": 0.000761988281738013,
-      "learning_rate": 0.00019999868014766378,
-      "loss": 46.0,
-      "step": 21400
-    },
-    {
-      "epoch": 1.6362559015234055,
-      "grad_norm": 0.0011164663592353463,
-      "learning_rate": 0.00019999868002425282,
-      "loss": 46.0,
-      "step": 21401
-    },
-    {
-      "epoch": 1.6363323585067953,
-      "grad_norm": 0.003488842397928238,
-      "learning_rate": 0.00019999867990083608,
-      "loss": 46.0,
-      "step": 21402
-    },
-    {
-      "epoch": 1.6364088154901848,
-      "grad_norm": 0.0008522930438630283,
-      "learning_rate": 0.0001999986797774136,
-      "loss": 46.0,
-      "step": 21403
-    },
-    {
-      "epoch": 1.6364852724735746,
-      "grad_norm": 0.0003183636872563511,
-      "learning_rate": 0.0001999986796539853,
-      "loss": 46.0,
-      "step": 21404
-    },
-    {
-      "epoch": 1.6365617294569643,
-      "grad_norm": 0.0007623909041285515,
-      "learning_rate": 0.00019999867953055125,
-      "loss": 46.0,
-      "step": 21405
-    },
-    {
-      "epoch": 1.6366381864403539,
-      "grad_norm": 0.0006663136300630867,
-      "learning_rate": 0.00019999867940711147,
-      "loss": 46.0,
-      "step": 21406
-    },
-    {
-      "epoch": 1.6367146434237436,
-      "grad_norm": 0.0012197361793369055,
-      "learning_rate": 0.00019999867928366586,
-      "loss": 46.0,
-      "step": 21407
-    },
-    {
-      "epoch": 1.6367911004071334,
-      "grad_norm": 0.001278865383937955,
-      "learning_rate": 0.0001999986791602145,
-      "loss": 46.0,
-      "step": 21408
-    },
-    {
-      "epoch": 1.6368675573905231,
-      "grad_norm": 0.0020506444852799177,
-      "learning_rate": 0.00019999867903675738,
-      "loss": 46.0,
-      "step": 21409
-    },
-    {
-      "epoch": 1.636944014373913,
-      "grad_norm": 0.0021305012051016092,
-      "learning_rate": 0.0001999986789132945,
-      "loss": 46.0,
-      "step": 21410
-    },
-    {
-      "epoch": 1.6370204713573027,
-      "grad_norm": 0.002022538334131241,
-      "learning_rate": 0.00019999867878982587,
-      "loss": 46.0,
-      "step": 21411
-    },
-    {
-      "epoch": 1.6370969283406924,
-      "grad_norm": 0.0009814834920689464,
-      "learning_rate": 0.00019999867866635142,
-      "loss": 46.0,
-      "step": 21412
-    },
-    {
-      "epoch": 1.6371733853240822,
-      "grad_norm": 0.0008672121912240982,
-      "learning_rate": 0.00019999867854287123,
-      "loss": 46.0,
-      "step": 21413
-    },
-    {
-      "epoch": 1.6372498423074717,
-      "grad_norm": 0.00041513825999572873,
-      "learning_rate": 0.00019999867841938524,
-      "loss": 46.0,
-      "step": 21414
-    },
-    {
-      "epoch": 1.6373262992908615,
-      "grad_norm": 0.0009369790786877275,
-      "learning_rate": 0.0001999986782958935,
-      "loss": 46.0,
-      "step": 21415
-    },
-    {
-      "epoch": 1.6374027562742512,
-      "grad_norm": 0.001369163510389626,
-      "learning_rate": 0.000199998678172396,
-      "loss": 46.0,
-      "step": 21416
-    },
-    {
-      "epoch": 1.6374792132576408,
-      "grad_norm": 0.0008466416038572788,
-      "learning_rate": 0.00019999867804889273,
-      "loss": 46.0,
-      "step": 21417
-    },
-    {
-      "epoch": 1.6375556702410305,
-      "grad_norm": 0.0012737085344269872,
-      "learning_rate": 0.0001999986779253837,
-      "loss": 46.0,
-      "step": 21418
-    },
-    {
-      "epoch": 1.6376321272244203,
-      "grad_norm": 0.0007173936464823782,
-      "learning_rate": 0.00019999867780186886,
-      "loss": 46.0,
-      "step": 21419
-    },
-    {
-      "epoch": 1.63770858420781,
-      "grad_norm": 0.001023541553877294,
-      "learning_rate": 0.00019999867767834828,
-      "loss": 46.0,
-      "step": 21420
-    },
-    {
-      "epoch": 1.6377850411911998,
-      "grad_norm": 0.0007021892233751714,
-      "learning_rate": 0.00019999867755482193,
-      "loss": 46.0,
-      "step": 21421
-    },
-    {
-      "epoch": 1.6378614981745896,
-      "grad_norm": 0.0006749724852852523,
-      "learning_rate": 0.0001999986774312898,
-      "loss": 46.0,
-      "step": 21422
-    },
-    {
-      "epoch": 1.6379379551579794,
-      "grad_norm": 0.0006043238681741059,
-      "learning_rate": 0.00019999867730775194,
-      "loss": 46.0,
-      "step": 21423
-    },
-    {
-      "epoch": 1.6380144121413691,
-      "grad_norm": 0.0008882411057129502,
-      "learning_rate": 0.00019999867718420826,
-      "loss": 46.0,
-      "step": 21424
-    },
-    {
-      "epoch": 1.6380908691247587,
-      "grad_norm": 0.0065286909230053425,
-      "learning_rate": 0.00019999867706065882,
-      "loss": 46.0,
-      "step": 21425
-    },
-    {
-      "epoch": 1.6381673261081484,
-      "grad_norm": 0.007178504019975662,
-      "learning_rate": 0.0001999986769371036,
-      "loss": 46.0,
-      "step": 21426
-    },
-    {
-      "epoch": 1.6382437830915382,
-      "grad_norm": 0.0009327356819994748,
-      "learning_rate": 0.00019999867681354263,
-      "loss": 46.0,
-      "step": 21427
-    },
-    {
-      "epoch": 1.6383202400749277,
-      "grad_norm": 0.000739293813239783,
-      "learning_rate": 0.00019999867668997592,
-      "loss": 46.0,
-      "step": 21428
-    },
-    {
-      "epoch": 1.6383966970583175,
-      "grad_norm": 0.0010062269866466522,
-      "learning_rate": 0.00019999867656640338,
-      "loss": 46.0,
-      "step": 21429
-    },
-    {
-      "epoch": 1.6384731540417072,
-      "grad_norm": 0.0012885708129033446,
-      "learning_rate": 0.00019999867644282513,
-      "loss": 46.0,
-      "step": 21430
-    },
-    {
-      "epoch": 1.638549611025097,
-      "grad_norm": 0.0013641210971400142,
-      "learning_rate": 0.00019999867631924107,
-      "loss": 46.0,
-      "step": 21431
-    },
-    {
-      "epoch": 1.6386260680084868,
-      "grad_norm": 0.0011096522212028503,
-      "learning_rate": 0.00019999867619565124,
-      "loss": 46.0,
-      "step": 21432
-    },
-    {
-      "epoch": 1.6387025249918765,
-      "grad_norm": 0.0016244127182289958,
-      "learning_rate": 0.00019999867607205566,
-      "loss": 46.0,
-      "step": 21433
-    },
-    {
-      "epoch": 1.6387789819752663,
-      "grad_norm": 0.0024805685970932245,
-      "learning_rate": 0.0001999986759484543,
-      "loss": 46.0,
-      "step": 21434
-    },
-    {
-      "epoch": 1.638855438958656,
-      "grad_norm": 0.0026701297610998154,
-      "learning_rate": 0.00019999867582484718,
-      "loss": 46.0,
-      "step": 21435
-    },
-    {
-      "epoch": 1.6389318959420456,
-      "grad_norm": 0.0009438165579922497,
-      "learning_rate": 0.00019999867570123428,
-      "loss": 46.0,
-      "step": 21436
-    },
-    {
-      "epoch": 1.6390083529254353,
-      "grad_norm": 0.027668753638863564,
-      "learning_rate": 0.0001999986755776156,
-      "loss": 46.0,
-      "step": 21437
-    },
-    {
-      "epoch": 1.639084809908825,
-      "grad_norm": 0.0008696553413756192,
-      "learning_rate": 0.0001999986754539912,
-      "loss": 46.0,
-      "step": 21438
-    },
-    {
-      "epoch": 1.6391612668922146,
-      "grad_norm": 0.007601591292768717,
-      "learning_rate": 0.000199998675330361,
-      "loss": 46.0,
-      "step": 21439
-    },
-    {
-      "epoch": 1.6392377238756044,
-      "grad_norm": 0.004165851976722479,
-      "learning_rate": 0.000199998675206725,
-      "loss": 46.0,
-      "step": 21440
-    },
-    {
-      "epoch": 1.6393141808589942,
-      "grad_norm": 0.0006563612259924412,
-      "learning_rate": 0.00019999867508308327,
-      "loss": 46.0,
-      "step": 21441
-    },
-    {
-      "epoch": 1.639390637842384,
-      "grad_norm": 0.005062736105173826,
-      "learning_rate": 0.00019999867495943576,
-      "loss": 46.0,
-      "step": 21442
-    },
-    {
-      "epoch": 1.6394670948257737,
-      "grad_norm": 0.0008686263463459909,
-      "learning_rate": 0.0001999986748357825,
-      "loss": 46.0,
-      "step": 21443
-    },
-    {
-      "epoch": 1.6395435518091634,
-      "grad_norm": 0.0016146207926794887,
-      "learning_rate": 0.00019999867471212344,
-      "loss": 46.0,
-      "step": 21444
-    },
-    {
-      "epoch": 1.6396200087925532,
-      "grad_norm": 0.0006938331061974168,
-      "learning_rate": 0.0001999986745884586,
-      "loss": 46.0,
-      "step": 21445
-    },
-    {
-      "epoch": 1.639696465775943,
-      "grad_norm": 0.0037709628231823444,
-      "learning_rate": 0.000199998674464788,
-      "loss": 46.0,
-      "step": 21446
-    },
-    {
-      "epoch": 1.6397729227593325,
-      "grad_norm": 0.003228316782042384,
-      "learning_rate": 0.00019999867434111166,
-      "loss": 46.0,
-      "step": 21447
-    },
-    {
-      "epoch": 1.6398493797427223,
-      "grad_norm": 0.0020065733697265387,
-      "learning_rate": 0.00019999867421742956,
-      "loss": 46.0,
-      "step": 21448
-    },
-    {
-      "epoch": 1.6399258367261118,
-      "grad_norm": 0.000771641731262207,
-      "learning_rate": 0.00019999867409374163,
-      "loss": 46.0,
-      "step": 21449
-    },
-    {
-      "epoch": 1.6400022937095016,
-      "grad_norm": 0.0006261492962948978,
-      "learning_rate": 0.000199998673970048,
-      "loss": 46.0,
-      "step": 21450
-    },
-    {
-      "epoch": 1.6400787506928913,
-      "grad_norm": 0.0011512351920828223,
-      "learning_rate": 0.00019999867384634854,
-      "loss": 46.0,
-      "step": 21451
-    },
-    {
-      "epoch": 1.640155207676281,
-      "grad_norm": 0.0023112036287784576,
-      "learning_rate": 0.00019999867372264335,
-      "loss": 46.0,
-      "step": 21452
-    },
-    {
-      "epoch": 1.6402316646596709,
-      "grad_norm": 0.0009989881655201316,
-      "learning_rate": 0.00019999867359893236,
-      "loss": 46.0,
-      "step": 21453
-    },
-    {
-      "epoch": 1.6403081216430606,
-      "grad_norm": 0.002040344290435314,
-      "learning_rate": 0.00019999867347521562,
-      "loss": 46.0,
-      "step": 21454
-    },
-    {
-      "epoch": 1.6403845786264504,
-      "grad_norm": 0.0007470930577255785,
-      "learning_rate": 0.0001999986733514931,
-      "loss": 46.0,
-      "step": 21455
-    },
-    {
-      "epoch": 1.6404610356098401,
-      "grad_norm": 0.0016769059002399445,
-      "learning_rate": 0.00019999867322776483,
-      "loss": 46.0,
-      "step": 21456
-    },
-    {
-      "epoch": 1.64053749259323,
-      "grad_norm": 0.003037337912246585,
-      "learning_rate": 0.00019999867310403077,
-      "loss": 46.0,
-      "step": 21457
-    },
-    {
-      "epoch": 1.6406139495766194,
-      "grad_norm": 0.004207718186080456,
-      "learning_rate": 0.00019999867298029097,
-      "loss": 46.0,
-      "step": 21458
-    },
-    {
-      "epoch": 1.6406904065600092,
-      "grad_norm": 0.000677183736115694,
-      "learning_rate": 0.00019999867285654536,
-      "loss": 46.0,
-      "step": 21459
-    },
-    {
-      "epoch": 1.6407668635433987,
-      "grad_norm": 0.0018042910378426313,
-      "learning_rate": 0.000199998672732794,
-      "loss": 46.0,
-      "step": 21460
-    },
-    {
-      "epoch": 1.6408433205267885,
-      "grad_norm": 0.0009706224082037807,
-      "learning_rate": 0.00019999867260903689,
-      "loss": 46.0,
-      "step": 21461
-    },
-    {
-      "epoch": 1.6409197775101783,
-      "grad_norm": 0.0009095891728065908,
-      "learning_rate": 0.000199998672485274,
-      "loss": 46.0,
-      "step": 21462
-    },
-    {
-      "epoch": 1.640996234493568,
-      "grad_norm": 0.003760226536542177,
-      "learning_rate": 0.00019999867236150532,
-      "loss": 46.0,
-      "step": 21463
-    },
-    {
-      "epoch": 1.6410726914769578,
-      "grad_norm": 0.001143462723121047,
-      "learning_rate": 0.00019999867223773087,
-      "loss": 46.0,
-      "step": 21464
-    },
-    {
-      "epoch": 1.6411491484603475,
-      "grad_norm": 0.0015298080397769809,
-      "learning_rate": 0.00019999867211395068,
-      "loss": 46.0,
-      "step": 21465
-    },
-    {
-      "epoch": 1.6412256054437373,
-      "grad_norm": 0.001853389199823141,
-      "learning_rate": 0.00019999867199016472,
-      "loss": 46.0,
-      "step": 21466
-    },
-    {
-      "epoch": 1.641302062427127,
-      "grad_norm": 0.0014462117105722427,
-      "learning_rate": 0.00019999867186637293,
-      "loss": 46.0,
-      "step": 21467
-    },
-    {
-      "epoch": 1.6413785194105168,
-      "grad_norm": 0.0007944151293486357,
-      "learning_rate": 0.00019999867174257542,
-      "loss": 46.0,
-      "step": 21468
-    },
-    {
-      "epoch": 1.6414549763939064,
-      "grad_norm": 0.003324038116261363,
-      "learning_rate": 0.00019999867161877216,
-      "loss": 46.0,
-      "step": 21469
-    },
-    {
-      "epoch": 1.6415314333772961,
-      "grad_norm": 0.006871282588690519,
-      "learning_rate": 0.0001999986714949631,
-      "loss": 46.0,
-      "step": 21470
-    },
-    {
-      "epoch": 1.6416078903606857,
-      "grad_norm": 0.001411988283507526,
-      "learning_rate": 0.00019999867137114827,
-      "loss": 46.0,
-      "step": 21471
-    },
-    {
-      "epoch": 1.6416843473440754,
-      "grad_norm": 0.003897459479048848,
-      "learning_rate": 0.00019999867124732766,
-      "loss": 46.0,
-      "step": 21472
-    },
-    {
-      "epoch": 1.6417608043274652,
-      "grad_norm": 0.0010658982209861279,
-      "learning_rate": 0.00019999867112350131,
-      "loss": 46.0,
-      "step": 21473
-    },
-    {
-      "epoch": 1.641837261310855,
-      "grad_norm": 0.0009889076463878155,
-      "learning_rate": 0.0001999986709996692,
-      "loss": 46.0,
-      "step": 21474
-    },
-    {
-      "epoch": 1.6419137182942447,
-      "grad_norm": 0.0019584798719733953,
-      "learning_rate": 0.0001999986708758313,
-      "loss": 46.0,
-      "step": 21475
-    },
-    {
-      "epoch": 1.6419901752776345,
-      "grad_norm": 0.0006531606777571142,
-      "learning_rate": 0.00019999867075198762,
-      "loss": 46.0,
-      "step": 21476
-    },
-    {
-      "epoch": 1.6420666322610242,
-      "grad_norm": 0.002930920571088791,
-      "learning_rate": 0.00019999867062813818,
-      "loss": 46.0,
-      "step": 21477
-    },
-    {
-      "epoch": 1.642143089244414,
-      "grad_norm": 0.0014221802121028304,
-      "learning_rate": 0.000199998670504283,
-      "loss": 46.0,
-      "step": 21478
-    },
-    {
-      "epoch": 1.6422195462278035,
-      "grad_norm": 0.00303996866568923,
-      "learning_rate": 0.000199998670380422,
-      "loss": 46.0,
-      "step": 21479
-    },
-    {
-      "epoch": 1.6422960032111933,
-      "grad_norm": 0.009048588573932648,
-      "learning_rate": 0.00019999867025655523,
-      "loss": 46.0,
-      "step": 21480
-    },
-    {
-      "epoch": 1.642372460194583,
-      "grad_norm": 0.001439518528059125,
-      "learning_rate": 0.00019999867013268272,
-      "loss": 46.0,
-      "step": 21481
-    },
-    {
-      "epoch": 1.6424489171779726,
-      "grad_norm": 0.0004770209197886288,
-      "learning_rate": 0.00019999867000880444,
-      "loss": 46.0,
-      "step": 21482
-    },
-    {
-      "epoch": 1.6425253741613624,
-      "grad_norm": 0.0009524385677650571,
-      "learning_rate": 0.00019999866988492038,
-      "loss": 46.0,
-      "step": 21483
-    },
-    {
-      "epoch": 1.6426018311447521,
-      "grad_norm": 0.0008762957295402884,
-      "learning_rate": 0.00019999866976103058,
-      "loss": 46.0,
-      "step": 21484
-    },
-    {
-      "epoch": 1.6426782881281419,
-      "grad_norm": 0.001765785156749189,
-      "learning_rate": 0.00019999866963713497,
-      "loss": 46.0,
-      "step": 21485
-    },
-    {
-      "epoch": 1.6427547451115316,
-      "grad_norm": 0.0008420029189437628,
-      "learning_rate": 0.0001999986695132336,
-      "loss": 46.0,
-      "step": 21486
-    },
-    {
-      "epoch": 1.6428312020949214,
-      "grad_norm": 0.00047600921243429184,
-      "learning_rate": 0.00019999866938932647,
-      "loss": 46.0,
-      "step": 21487
-    },
-    {
-      "epoch": 1.6429076590783112,
-      "grad_norm": 0.0012290027225390077,
-      "learning_rate": 0.00019999866926541357,
-      "loss": 46.0,
-      "step": 21488
-    },
-    {
-      "epoch": 1.642984116061701,
-      "grad_norm": 0.000704701931681484,
-      "learning_rate": 0.0001999986691414949,
-      "loss": 46.0,
-      "step": 21489
-    },
-    {
-      "epoch": 1.6430605730450905,
-      "grad_norm": 0.0005110579659231007,
-      "learning_rate": 0.00019999866901757046,
-      "loss": 46.0,
-      "step": 21490
-    },
-    {
-      "epoch": 1.6431370300284802,
-      "grad_norm": 0.0004316684789955616,
-      "learning_rate": 0.00019999866889364024,
-      "loss": 46.0,
-      "step": 21491
-    },
-    {
-      "epoch": 1.64321348701187,
-      "grad_norm": 0.002363208681344986,
-      "learning_rate": 0.00019999866876970428,
-      "loss": 46.0,
-      "step": 21492
-    },
-    {
-      "epoch": 1.6432899439952595,
-      "grad_norm": 0.0027257229667156935,
-      "learning_rate": 0.00019999866864576254,
-      "loss": 46.0,
-      "step": 21493
-    },
-    {
-      "epoch": 1.6433664009786493,
-      "grad_norm": 0.004278544336557388,
-      "learning_rate": 0.000199998668521815,
-      "loss": 46.0,
-      "step": 21494
-    },
-    {
-      "epoch": 1.643442857962039,
-      "grad_norm": 0.0007887862739153206,
-      "learning_rate": 0.00019999866839786172,
-      "loss": 46.0,
-      "step": 21495
-    },
-    {
-      "epoch": 1.6435193149454288,
-      "grad_norm": 0.0006672461167909205,
-      "learning_rate": 0.00019999866827390266,
-      "loss": 46.0,
-      "step": 21496
-    },
-    {
-      "epoch": 1.6435957719288186,
-      "grad_norm": 0.005941648036241531,
-      "learning_rate": 0.0001999986681499378,
-      "loss": 46.0,
-      "step": 21497
-    },
-    {
-      "epoch": 1.6436722289122083,
-      "grad_norm": 0.0008345708483830094,
-      "learning_rate": 0.00019999866802596726,
-      "loss": 46.0,
-      "step": 21498
-    },
-    {
-      "epoch": 1.643748685895598,
-      "grad_norm": 0.0032663140445947647,
-      "learning_rate": 0.00019999866790199085,
-      "loss": 46.0,
-      "step": 21499
-    },
-    {
-      "epoch": 1.6438251428789878,
-      "grad_norm": 0.001309213344939053,
-      "learning_rate": 0.00019999866777800873,
-      "loss": 46.0,
-      "step": 21500
-    },
-    {
-      "epoch": 1.6439015998623774,
-      "grad_norm": 0.00046075653517618775,
-      "learning_rate": 0.00019999866765402083,
-      "loss": 46.0,
-      "step": 21501
-    },
-    {
-      "epoch": 1.6439780568457671,
-      "grad_norm": 0.011375710368156433,
-      "learning_rate": 0.00019999866753002716,
-      "loss": 46.0,
-      "step": 21502
-    },
-    {
-      "epoch": 1.644054513829157,
-      "grad_norm": 0.002224414609372616,
-      "learning_rate": 0.00019999866740602772,
-      "loss": 46.0,
-      "step": 21503
-    },
-    {
-      "epoch": 1.6441309708125464,
-      "grad_norm": 0.001404323847964406,
-      "learning_rate": 0.0001999986672820225,
-      "loss": 46.0,
-      "step": 21504
-    },
-    {
-      "epoch": 1.6442074277959362,
-      "grad_norm": 0.00245850533246994,
-      "learning_rate": 0.0001999986671580115,
-      "loss": 46.0,
-      "step": 21505
-    },
-    {
-      "epoch": 1.644283884779326,
-      "grad_norm": 0.011049183085560799,
-      "learning_rate": 0.00019999866703399478,
-      "loss": 46.0,
-      "step": 21506
-    },
-    {
-      "epoch": 1.6443603417627157,
-      "grad_norm": 0.002516641980037093,
-      "learning_rate": 0.00019999866690997227,
-      "loss": 46.0,
-      "step": 21507
-    },
-    {
-      "epoch": 1.6444367987461055,
-      "grad_norm": 0.0012511474778875709,
-      "learning_rate": 0.00019999866678594396,
-      "loss": 46.0,
-      "step": 21508
-    },
-    {
-      "epoch": 1.6445132557294952,
-      "grad_norm": 0.0019315023673698306,
-      "learning_rate": 0.0001999986666619099,
-      "loss": 46.0,
-      "step": 21509
-    },
-    {
-      "epoch": 1.644589712712885,
-      "grad_norm": 0.0017806924879550934,
-      "learning_rate": 0.00019999866653787007,
-      "loss": 46.0,
-      "step": 21510
-    },
-    {
-      "epoch": 1.6446661696962748,
-      "grad_norm": 0.003197327023372054,
-      "learning_rate": 0.00019999866641382447,
-      "loss": 46.0,
-      "step": 21511
-    },
-    {
-      "epoch": 1.6447426266796643,
-      "grad_norm": 0.001833573216572404,
-      "learning_rate": 0.00019999866628977312,
-      "loss": 46.0,
-      "step": 21512
-    },
-    {
-      "epoch": 1.644819083663054,
-      "grad_norm": 0.003872358938679099,
-      "learning_rate": 0.00019999866616571597,
-      "loss": 46.0,
-      "step": 21513
-    },
-    {
-      "epoch": 1.6448955406464438,
-      "grad_norm": 0.0017271749675273895,
-      "learning_rate": 0.00019999866604165307,
-      "loss": 46.0,
-      "step": 21514
-    },
-    {
-      "epoch": 1.6449719976298334,
-      "grad_norm": 0.001248883199878037,
-      "learning_rate": 0.0001999986659175844,
-      "loss": 46.0,
-      "step": 21515
-    },
-    {
-      "epoch": 1.6450484546132231,
-      "grad_norm": 0.0010162307880818844,
-      "learning_rate": 0.00019999866579350996,
-      "loss": 46.0,
-      "step": 21516
-    },
-    {
-      "epoch": 1.645124911596613,
-      "grad_norm": 0.0021877228282392025,
-      "learning_rate": 0.00019999866566942974,
-      "loss": 46.0,
-      "step": 21517
-    },
-    {
-      "epoch": 1.6452013685800027,
-      "grad_norm": 0.0012758949305862188,
-      "learning_rate": 0.00019999866554534375,
-      "loss": 46.0,
-      "step": 21518
-    },
-    {
-      "epoch": 1.6452778255633924,
-      "grad_norm": 0.00048259375034831464,
-      "learning_rate": 0.000199998665421252,
-      "loss": 46.0,
-      "step": 21519
-    },
-    {
-      "epoch": 1.6453542825467822,
-      "grad_norm": 0.005213888827711344,
-      "learning_rate": 0.00019999866529715448,
-      "loss": 46.0,
-      "step": 21520
-    },
-    {
-      "epoch": 1.645430739530172,
-      "grad_norm": 0.0007712020887993276,
-      "learning_rate": 0.00019999866517305117,
-      "loss": 46.0,
-      "step": 21521
-    },
-    {
-      "epoch": 1.6455071965135617,
-      "grad_norm": 0.0008918406092561781,
-      "learning_rate": 0.00019999866504894214,
-      "loss": 46.0,
-      "step": 21522
-    },
-    {
-      "epoch": 1.6455836534969512,
-      "grad_norm": 0.0010949833085760474,
-      "learning_rate": 0.00019999866492482728,
-      "loss": 46.0,
-      "step": 21523
-    },
-    {
-      "epoch": 1.645660110480341,
-      "grad_norm": 0.004753062501549721,
-      "learning_rate": 0.0001999986648007067,
-      "loss": 46.0,
-      "step": 21524
-    },
-    {
-      "epoch": 1.6457365674637308,
-      "grad_norm": 0.0017014964250847697,
-      "learning_rate": 0.0001999986646765803,
-      "loss": 46.0,
-      "step": 21525
-    },
-    {
-      "epoch": 1.6458130244471203,
-      "grad_norm": 0.0008219338487833738,
-      "learning_rate": 0.00019999866455244818,
-      "loss": 46.0,
-      "step": 21526
-    },
-    {
-      "epoch": 1.64588948143051,
-      "grad_norm": 0.0010213457280769944,
-      "learning_rate": 0.00019999866442831028,
-      "loss": 46.0,
-      "step": 21527
-    },
-    {
-      "epoch": 1.6459659384138998,
-      "grad_norm": 0.0012580167967826128,
-      "learning_rate": 0.0001999986643041666,
-      "loss": 46.0,
-      "step": 21528
-    },
-    {
-      "epoch": 1.6460423953972896,
-      "grad_norm": 0.0007823933265171945,
-      "learning_rate": 0.00019999866418001717,
-      "loss": 46.0,
-      "step": 21529
-    },
-    {
-      "epoch": 1.6461188523806793,
-      "grad_norm": 0.000719325733371079,
-      "learning_rate": 0.00019999866405586193,
-      "loss": 46.0,
-      "step": 21530
-    },
-    {
-      "epoch": 1.646195309364069,
-      "grad_norm": 0.0010906400857493281,
-      "learning_rate": 0.00019999866393170094,
-      "loss": 46.0,
-      "step": 21531
-    },
-    {
-      "epoch": 1.6462717663474589,
-      "grad_norm": 0.0010455127339810133,
-      "learning_rate": 0.00019999866380753417,
-      "loss": 46.0,
-      "step": 21532
-    },
-    {
-      "epoch": 1.6463482233308486,
-      "grad_norm": 0.0010679762344807386,
-      "learning_rate": 0.0001999986636833617,
-      "loss": 46.0,
-      "step": 21533
-    },
-    {
-      "epoch": 1.6464246803142382,
-      "grad_norm": 0.003093340899795294,
-      "learning_rate": 0.00019999866355918338,
-      "loss": 46.0,
-      "step": 21534
-    },
-    {
-      "epoch": 1.646501137297628,
-      "grad_norm": 0.0002861547109205276,
-      "learning_rate": 0.0001999986634349993,
-      "loss": 46.0,
-      "step": 21535
-    },
-    {
-      "epoch": 1.6465775942810177,
-      "grad_norm": 0.0014112535864114761,
-      "learning_rate": 0.00019999866331080947,
-      "loss": 46.0,
-      "step": 21536
-    },
-    {
-      "epoch": 1.6466540512644072,
-      "grad_norm": 0.0005738164763897657,
-      "learning_rate": 0.0001999986631866139,
-      "loss": 46.0,
-      "step": 21537
-    },
-    {
-      "epoch": 1.646730508247797,
-      "grad_norm": 0.0006993873394094408,
-      "learning_rate": 0.00019999866306241252,
-      "loss": 46.0,
-      "step": 21538
-    },
-    {
-      "epoch": 1.6468069652311867,
-      "grad_norm": 0.0016875143628567457,
-      "learning_rate": 0.00019999866293820537,
-      "loss": 46.0,
-      "step": 21539
-    },
-    {
-      "epoch": 1.6468834222145765,
-      "grad_norm": 0.000774713815189898,
-      "learning_rate": 0.00019999866281399244,
-      "loss": 46.0,
-      "step": 21540
-    },
-    {
-      "epoch": 1.6469598791979663,
-      "grad_norm": 0.0006773581262677908,
-      "learning_rate": 0.00019999866268977378,
-      "loss": 46.0,
-      "step": 21541
-    },
-    {
-      "epoch": 1.647036336181356,
-      "grad_norm": 0.0005350045394152403,
-      "learning_rate": 0.00019999866256554933,
-      "loss": 46.0,
-      "step": 21542
-    },
-    {
-      "epoch": 1.6471127931647458,
-      "grad_norm": 0.0012855910463258624,
-      "learning_rate": 0.00019999866244131912,
-      "loss": 46.0,
-      "step": 21543
-    },
-    {
-      "epoch": 1.6471892501481356,
-      "grad_norm": 0.0009674620814621449,
-      "learning_rate": 0.00019999866231708313,
-      "loss": 46.0,
-      "step": 21544
-    },
-    {
-      "epoch": 1.647265707131525,
-      "grad_norm": 0.0010357644641771913,
-      "learning_rate": 0.00019999866219284136,
-      "loss": 46.0,
-      "step": 21545
-    },
-    {
-      "epoch": 1.6473421641149149,
-      "grad_norm": 0.009009824134409428,
-      "learning_rate": 0.00019999866206859386,
-      "loss": 46.0,
-      "step": 21546
-    },
-    {
-      "epoch": 1.6474186210983046,
-      "grad_norm": 0.0007428620592691004,
-      "learning_rate": 0.00019999866194434055,
-      "loss": 46.0,
-      "step": 21547
-    },
-    {
-      "epoch": 1.6474950780816942,
-      "grad_norm": 0.0008998431731015444,
-      "learning_rate": 0.0001999986618200815,
-      "loss": 46.0,
-      "step": 21548
-    },
-    {
-      "epoch": 1.647571535065084,
-      "grad_norm": 0.0007695354288443923,
-      "learning_rate": 0.00019999866169581663,
-      "loss": 46.0,
-      "step": 21549
-    },
-    {
-      "epoch": 1.6476479920484737,
-      "grad_norm": 0.0033494688104838133,
-      "learning_rate": 0.00019999866157154603,
-      "loss": 46.0,
-      "step": 21550
-    },
-    {
-      "epoch": 1.6477244490318634,
-      "grad_norm": 0.0008527569589205086,
-      "learning_rate": 0.00019999866144726966,
-      "loss": 46.0,
-      "step": 21551
-    },
-    {
-      "epoch": 1.6478009060152532,
-      "grad_norm": 0.010698262602090836,
-      "learning_rate": 0.00019999866132298753,
-      "loss": 46.0,
-      "step": 21552
-    },
-    {
-      "epoch": 1.647877362998643,
-      "grad_norm": 0.00197147810831666,
-      "learning_rate": 0.0001999986611986996,
-      "loss": 46.0,
-      "step": 21553
-    },
-    {
-      "epoch": 1.6479538199820327,
-      "grad_norm": 0.0014603988965973258,
-      "learning_rate": 0.00019999866107440592,
-      "loss": 46.0,
-      "step": 21554
-    },
-    {
-      "epoch": 1.6480302769654225,
-      "grad_norm": 0.001671516802161932,
-      "learning_rate": 0.0001999986609501065,
-      "loss": 46.0,
-      "step": 21555
-    },
-    {
-      "epoch": 1.648106733948812,
-      "grad_norm": 0.0004867609532084316,
-      "learning_rate": 0.00019999866082580128,
-      "loss": 46.0,
-      "step": 21556
-    },
-    {
-      "epoch": 1.6481831909322018,
-      "grad_norm": 0.001656443695537746,
-      "learning_rate": 0.00019999866070149027,
-      "loss": 46.0,
-      "step": 21557
-    },
-    {
-      "epoch": 1.6482596479155915,
-      "grad_norm": 0.005509183742105961,
-      "learning_rate": 0.00019999866057717353,
-      "loss": 46.0,
-      "step": 21558
-    },
-    {
-      "epoch": 1.648336104898981,
-      "grad_norm": 0.0006206129328347743,
-      "learning_rate": 0.00019999866045285097,
-      "loss": 46.0,
-      "step": 21559
-    },
-    {
-      "epoch": 1.6484125618823708,
-      "grad_norm": 0.0006834531086497009,
-      "learning_rate": 0.00019999866032852272,
-      "loss": 46.0,
-      "step": 21560
-    },
-    {
-      "epoch": 1.6484890188657606,
-      "grad_norm": 0.0007297497941181064,
-      "learning_rate": 0.00019999866020418863,
-      "loss": 46.0,
-      "step": 21561
-    },
-    {
-      "epoch": 1.6485654758491504,
-      "grad_norm": 0.0005961584392935038,
-      "learning_rate": 0.00019999866007984878,
-      "loss": 46.0,
-      "step": 21562
-    },
-    {
-      "epoch": 1.6486419328325401,
-      "grad_norm": 0.0014104496221989393,
-      "learning_rate": 0.00019999865995550317,
-      "loss": 46.0,
-      "step": 21563
-    },
-    {
-      "epoch": 1.6487183898159299,
-      "grad_norm": 0.0009982379851862788,
-      "learning_rate": 0.0001999986598311518,
-      "loss": 46.0,
-      "step": 21564
-    },
-    {
-      "epoch": 1.6487948467993196,
-      "grad_norm": 0.008005477488040924,
-      "learning_rate": 0.00019999865970679465,
-      "loss": 46.0,
-      "step": 21565
-    },
-    {
-      "epoch": 1.6488713037827094,
-      "grad_norm": 0.0005726784584112465,
-      "learning_rate": 0.00019999865958243175,
-      "loss": 46.0,
-      "step": 21566
-    },
-    {
-      "epoch": 1.648947760766099,
-      "grad_norm": 0.002822288079187274,
-      "learning_rate": 0.00019999865945806309,
-      "loss": 46.0,
-      "step": 21567
-    },
-    {
-      "epoch": 1.6490242177494887,
-      "grad_norm": 0.0009812784846872091,
-      "learning_rate": 0.00019999865933368862,
-      "loss": 46.0,
-      "step": 21568
-    },
-    {
-      "epoch": 1.6491006747328785,
-      "grad_norm": 0.0009202967048622668,
-      "learning_rate": 0.00019999865920930837,
-      "loss": 46.0,
-      "step": 21569
-    },
-    {
-      "epoch": 1.649177131716268,
-      "grad_norm": 0.0016039526090025902,
-      "learning_rate": 0.0001999986590849224,
-      "loss": 46.0,
-      "step": 21570
-    },
-    {
-      "epoch": 1.6492535886996578,
-      "grad_norm": 0.002659932244569063,
-      "learning_rate": 0.00019999865896053065,
-      "loss": 46.0,
-      "step": 21571
-    },
-    {
-      "epoch": 1.6493300456830475,
-      "grad_norm": 0.0006635301397182047,
-      "learning_rate": 0.00019999865883613312,
-      "loss": 46.0,
-      "step": 21572
-    },
-    {
-      "epoch": 1.6494065026664373,
-      "grad_norm": 0.0020059943199157715,
-      "learning_rate": 0.00019999865871172983,
-      "loss": 46.0,
-      "step": 21573
-    },
-    {
-      "epoch": 1.649482959649827,
-      "grad_norm": 0.0014936156803742051,
-      "learning_rate": 0.00019999865858732075,
-      "loss": 46.0,
-      "step": 21574
-    },
-    {
-      "epoch": 1.6495594166332168,
-      "grad_norm": 0.015905240550637245,
-      "learning_rate": 0.0001999986584629059,
-      "loss": 46.0,
-      "step": 21575
-    },
-    {
-      "epoch": 1.6496358736166066,
-      "grad_norm": 0.0026930151507258415,
-      "learning_rate": 0.0001999986583384853,
-      "loss": 46.0,
-      "step": 21576
-    },
-    {
-      "epoch": 1.6497123305999963,
-      "grad_norm": 0.0018401573179289699,
-      "learning_rate": 0.00019999865821405892,
-      "loss": 46.0,
-      "step": 21577
-    },
-    {
-      "epoch": 1.6497887875833859,
-      "grad_norm": 0.00048216519644483924,
-      "learning_rate": 0.0001999986580896268,
-      "loss": 46.0,
-      "step": 21578
-    },
-    {
-      "epoch": 1.6498652445667756,
-      "grad_norm": 0.0005768570699729025,
-      "learning_rate": 0.00019999865796518888,
-      "loss": 46.0,
-      "step": 21579
-    },
-    {
-      "epoch": 1.6499417015501652,
-      "grad_norm": 0.0007254622178152204,
-      "learning_rate": 0.00019999865784074515,
-      "loss": 46.0,
-      "step": 21580
-    },
-    {
-      "epoch": 1.650018158533555,
-      "grad_norm": 0.0016321848379448056,
-      "learning_rate": 0.00019999865771629574,
-      "loss": 46.0,
-      "step": 21581
-    },
-    {
-      "epoch": 1.6500946155169447,
-      "grad_norm": 0.0006128469249233603,
-      "learning_rate": 0.0001999986575918405,
-      "loss": 46.0,
-      "step": 21582
-    },
-    {
-      "epoch": 1.6501710725003345,
-      "grad_norm": 0.0010214586509391665,
-      "learning_rate": 0.0001999986574673795,
-      "loss": 46.0,
-      "step": 21583
-    },
-    {
-      "epoch": 1.6502475294837242,
-      "grad_norm": 0.000726212514564395,
-      "learning_rate": 0.00019999865734291275,
-      "loss": 46.0,
-      "step": 21584
-    },
-    {
-      "epoch": 1.650323986467114,
-      "grad_norm": 0.0023472574539482594,
-      "learning_rate": 0.0001999986572184402,
-      "loss": 46.0,
-      "step": 21585
-    },
-    {
-      "epoch": 1.6504004434505037,
-      "grad_norm": 0.008027667179703712,
-      "learning_rate": 0.0001999986570939619,
-      "loss": 46.0,
-      "step": 21586
-    },
-    {
-      "epoch": 1.6504769004338935,
-      "grad_norm": 0.002492302330210805,
-      "learning_rate": 0.00019999865696947785,
-      "loss": 46.0,
-      "step": 21587
-    },
-    {
-      "epoch": 1.6505533574172833,
-      "grad_norm": 0.0010119705693796277,
-      "learning_rate": 0.000199998656844988,
-      "loss": 46.0,
-      "step": 21588
-    },
-    {
-      "epoch": 1.6506298144006728,
-      "grad_norm": 0.0008986667962744832,
-      "learning_rate": 0.0001999986567204924,
-      "loss": 46.0,
-      "step": 21589
-    },
-    {
-      "epoch": 1.6507062713840626,
-      "grad_norm": 0.0012087725335732102,
-      "learning_rate": 0.00019999865659599102,
-      "loss": 46.0,
-      "step": 21590
-    },
-    {
-      "epoch": 1.650782728367452,
-      "grad_norm": 0.001363583141937852,
-      "learning_rate": 0.00019999865647148387,
-      "loss": 46.0,
-      "step": 21591
-    },
-    {
-      "epoch": 1.6508591853508419,
-      "grad_norm": 0.001039830851368606,
-      "learning_rate": 0.00019999865634697095,
-      "loss": 46.0,
-      "step": 21592
-    },
-    {
-      "epoch": 1.6509356423342316,
-      "grad_norm": 0.00040533390711061656,
-      "learning_rate": 0.00019999865622245225,
-      "loss": 46.0,
-      "step": 21593
-    },
-    {
-      "epoch": 1.6510120993176214,
-      "grad_norm": 0.0009514043922536075,
-      "learning_rate": 0.0001999986560979278,
-      "loss": 46.0,
-      "step": 21594
-    },
-    {
-      "epoch": 1.6510885563010111,
-      "grad_norm": 0.003054453991353512,
-      "learning_rate": 0.00019999865597339757,
-      "loss": 46.0,
-      "step": 21595
-    },
-    {
-      "epoch": 1.651165013284401,
-      "grad_norm": 0.0006716137286275625,
-      "learning_rate": 0.00019999865584886158,
-      "loss": 46.0,
-      "step": 21596
-    },
-    {
-      "epoch": 1.6512414702677907,
-      "grad_norm": 0.0003735281643457711,
-      "learning_rate": 0.00019999865572431982,
-      "loss": 46.0,
-      "step": 21597
-    },
-    {
-      "epoch": 1.6513179272511804,
-      "grad_norm": 0.002730264561250806,
-      "learning_rate": 0.00019999865559977228,
-      "loss": 46.0,
-      "step": 21598
-    },
-    {
-      "epoch": 1.6513943842345702,
-      "grad_norm": 0.0010241935960948467,
-      "learning_rate": 0.00019999865547521898,
-      "loss": 46.0,
-      "step": 21599
-    },
-    {
-      "epoch": 1.6514708412179597,
-      "grad_norm": 0.0010458551114425063,
-      "learning_rate": 0.00019999865535065992,
-      "loss": 46.0,
-      "step": 21600
-    },
-    {
-      "epoch": 1.6515472982013495,
-      "grad_norm": 0.0010004040086641908,
-      "learning_rate": 0.00019999865522609507,
-      "loss": 46.0,
-      "step": 21601
-    },
-    {
-      "epoch": 1.651623755184739,
-      "grad_norm": 0.0009689674479886889,
-      "learning_rate": 0.00019999865510152444,
-      "loss": 46.0,
-      "step": 21602
-    },
-    {
-      "epoch": 1.6517002121681288,
-      "grad_norm": 0.0027522125747054815,
-      "learning_rate": 0.00019999865497694806,
-      "loss": 46.0,
-      "step": 21603
-    },
-    {
-      "epoch": 1.6517766691515186,
-      "grad_norm": 0.003077751724049449,
-      "learning_rate": 0.00019999865485236594,
-      "loss": 46.0,
-      "step": 21604
-    },
-    {
-      "epoch": 1.6518531261349083,
-      "grad_norm": 0.0006341529078781605,
-      "learning_rate": 0.00019999865472777802,
-      "loss": 46.0,
-      "step": 21605
-    },
-    {
-      "epoch": 1.651929583118298,
-      "grad_norm": 0.001420557964593172,
-      "learning_rate": 0.00019999865460318433,
-      "loss": 46.0,
-      "step": 21606
-    },
-    {
-      "epoch": 1.6520060401016878,
-      "grad_norm": 0.0007528112037107348,
-      "learning_rate": 0.00019999865447858488,
-      "loss": 46.0,
-      "step": 21607
-    },
-    {
-      "epoch": 1.6520824970850776,
-      "grad_norm": 0.0011315782321617007,
-      "learning_rate": 0.00019999865435397962,
-      "loss": 46.0,
-      "step": 21608
-    },
-    {
-      "epoch": 1.6521589540684674,
-      "grad_norm": 0.004409399814903736,
-      "learning_rate": 0.00019999865422936863,
-      "loss": 46.0,
-      "step": 21609
-    },
-    {
-      "epoch": 1.652235411051857,
-      "grad_norm": 0.0015307499561458826,
-      "learning_rate": 0.00019999865410475187,
-      "loss": 46.0,
-      "step": 21610
-    },
-    {
-      "epoch": 1.6523118680352467,
-      "grad_norm": 0.0009448578348383307,
-      "learning_rate": 0.00019999865398012933,
-      "loss": 46.0,
-      "step": 21611
-    },
-    {
-      "epoch": 1.6523883250186364,
-      "grad_norm": 0.006631122436374426,
-      "learning_rate": 0.00019999865385550103,
-      "loss": 46.0,
-      "step": 21612
-    },
-    {
-      "epoch": 1.652464782002026,
-      "grad_norm": 0.0006825471064075828,
-      "learning_rate": 0.00019999865373086694,
-      "loss": 46.0,
-      "step": 21613
-    },
-    {
-      "epoch": 1.6525412389854157,
-      "grad_norm": 0.0008091607014648616,
-      "learning_rate": 0.00019999865360622712,
-      "loss": 46.0,
-      "step": 21614
-    },
-    {
-      "epoch": 1.6526176959688055,
-      "grad_norm": 0.0038639530539512634,
-      "learning_rate": 0.00019999865348158152,
-      "loss": 46.0,
-      "step": 21615
-    },
-    {
-      "epoch": 1.6526941529521952,
-      "grad_norm": 0.0011462783440947533,
-      "learning_rate": 0.00019999865335693011,
-      "loss": 46.0,
-      "step": 21616
-    },
-    {
-      "epoch": 1.652770609935585,
-      "grad_norm": 0.0011677076108753681,
-      "learning_rate": 0.000199998653232273,
-      "loss": 46.0,
-      "step": 21617
-    },
-    {
-      "epoch": 1.6528470669189748,
-      "grad_norm": 0.002352225361391902,
-      "learning_rate": 0.00019999865310761005,
-      "loss": 46.0,
-      "step": 21618
-    },
-    {
-      "epoch": 1.6529235239023645,
-      "grad_norm": 0.001397239277139306,
-      "learning_rate": 0.00019999865298294135,
-      "loss": 46.0,
-      "step": 21619
-    },
-    {
-      "epoch": 1.6529999808857543,
-      "grad_norm": 0.0011672712862491608,
-      "learning_rate": 0.0001999986528582669,
-      "loss": 46.0,
-      "step": 21620
-    },
-    {
-      "epoch": 1.6530764378691438,
-      "grad_norm": 0.0015168029349297285,
-      "learning_rate": 0.00019999865273358667,
-      "loss": 46.0,
-      "step": 21621
-    },
-    {
-      "epoch": 1.6531528948525336,
-      "grad_norm": 0.0020129806362092495,
-      "learning_rate": 0.00019999865260890068,
-      "loss": 46.0,
-      "step": 21622
-    },
-    {
-      "epoch": 1.6532293518359233,
-      "grad_norm": 0.0007061193464323878,
-      "learning_rate": 0.0001999986524842089,
-      "loss": 46.0,
-      "step": 21623
-    },
-    {
-      "epoch": 1.6533058088193129,
-      "grad_norm": 0.0007871821289882064,
-      "learning_rate": 0.00019999865235951136,
-      "loss": 46.0,
-      "step": 21624
-    },
-    {
-      "epoch": 1.6533822658027026,
-      "grad_norm": 0.0015681508230045438,
-      "learning_rate": 0.00019999865223480808,
-      "loss": 46.0,
-      "step": 21625
-    },
-    {
-      "epoch": 1.6534587227860924,
-      "grad_norm": 0.0007553251925855875,
-      "learning_rate": 0.00019999865211009897,
-      "loss": 46.0,
-      "step": 21626
-    },
-    {
-      "epoch": 1.6535351797694822,
-      "grad_norm": 0.0009992470731958747,
-      "learning_rate": 0.00019999865198538412,
-      "loss": 46.0,
-      "step": 21627
-    },
-    {
-      "epoch": 1.653611636752872,
-      "grad_norm": 0.0017561583081260324,
-      "learning_rate": 0.00019999865186066354,
-      "loss": 46.0,
-      "step": 21628
-    },
-    {
-      "epoch": 1.6536880937362617,
-      "grad_norm": 0.0018788323504850268,
-      "learning_rate": 0.00019999865173593711,
-      "loss": 46.0,
-      "step": 21629
-    },
-    {
-      "epoch": 1.6537645507196514,
-      "grad_norm": 0.0023352557327598333,
-      "learning_rate": 0.000199998651611205,
-      "loss": 46.0,
-      "step": 21630
-    },
-    {
-      "epoch": 1.6538410077030412,
-      "grad_norm": 0.0010183880804106593,
-      "learning_rate": 0.00019999865148646705,
-      "loss": 46.0,
-      "step": 21631
-    },
-    {
-      "epoch": 1.6539174646864307,
-      "grad_norm": 0.00117160112131387,
-      "learning_rate": 0.00019999865136172335,
-      "loss": 46.0,
-      "step": 21632
-    },
-    {
-      "epoch": 1.6539939216698205,
-      "grad_norm": 0.005221403203904629,
-      "learning_rate": 0.0001999986512369739,
-      "loss": 46.0,
-      "step": 21633
-    },
-    {
-      "epoch": 1.6540703786532103,
-      "grad_norm": 0.0033547745551913977,
-      "learning_rate": 0.00019999865111221867,
-      "loss": 46.0,
-      "step": 21634
-    },
-    {
-      "epoch": 1.6541468356365998,
-      "grad_norm": 0.002293757162988186,
-      "learning_rate": 0.00019999865098745766,
-      "loss": 46.0,
-      "step": 21635
-    },
-    {
-      "epoch": 1.6542232926199896,
-      "grad_norm": 0.0030303706880658865,
-      "learning_rate": 0.0001999986508626909,
-      "loss": 46.0,
-      "step": 21636
-    },
-    {
-      "epoch": 1.6542997496033793,
-      "grad_norm": 0.0033912723883986473,
-      "learning_rate": 0.00019999865073791834,
-      "loss": 46.0,
-      "step": 21637
-    },
-    {
-      "epoch": 1.654376206586769,
-      "grad_norm": 0.002571378368884325,
-      "learning_rate": 0.00019999865061314006,
-      "loss": 46.0,
-      "step": 21638
-    },
-    {
-      "epoch": 1.6544526635701589,
-      "grad_norm": 0.0011502320412546396,
-      "learning_rate": 0.00019999865048835597,
-      "loss": 46.0,
-      "step": 21639
-    },
-    {
-      "epoch": 1.6545291205535486,
-      "grad_norm": 0.00027156888972967863,
-      "learning_rate": 0.00019999865036356612,
-      "loss": 46.0,
-      "step": 21640
-    },
-    {
-      "epoch": 1.6546055775369384,
-      "grad_norm": 0.0009653330780565739,
-      "learning_rate": 0.00019999865023877052,
-      "loss": 46.0,
-      "step": 21641
-    },
-    {
-      "epoch": 1.6546820345203281,
-      "grad_norm": 0.0011527067981660366,
-      "learning_rate": 0.00019999865011396912,
-      "loss": 46.0,
-      "step": 21642
-    },
-    {
-      "epoch": 1.6547584915037177,
-      "grad_norm": 0.0010315327672287822,
-      "learning_rate": 0.00019999864998916195,
-      "loss": 46.0,
-      "step": 21643
-    },
-    {
-      "epoch": 1.6548349484871074,
-      "grad_norm": 0.0012466766638681293,
-      "learning_rate": 0.00019999864986434905,
-      "loss": 46.0,
-      "step": 21644
-    },
-    {
-      "epoch": 1.6549114054704972,
-      "grad_norm": 0.01666574738919735,
-      "learning_rate": 0.00019999864973953036,
-      "loss": 46.0,
-      "step": 21645
-    },
-    {
-      "epoch": 1.6549878624538867,
-      "grad_norm": 0.001451404532417655,
-      "learning_rate": 0.0001999986496147059,
-      "loss": 46.0,
-      "step": 21646
-    },
-    {
-      "epoch": 1.6550643194372765,
-      "grad_norm": 0.001533798291347921,
-      "learning_rate": 0.00019999864948987565,
-      "loss": 46.0,
-      "step": 21647
-    },
-    {
-      "epoch": 1.6551407764206663,
-      "grad_norm": 0.0017557170940563083,
-      "learning_rate": 0.00019999864936503964,
-      "loss": 46.0,
-      "step": 21648
-    },
-    {
-      "epoch": 1.655217233404056,
-      "grad_norm": 0.0028071885462850332,
-      "learning_rate": 0.00019999864924019788,
-      "loss": 46.0,
-      "step": 21649
-    },
-    {
-      "epoch": 1.6552936903874458,
-      "grad_norm": 0.001328642014414072,
-      "learning_rate": 0.00019999864911535034,
-      "loss": 46.0,
-      "step": 21650
-    },
-    {
-      "epoch": 1.6553701473708355,
-      "grad_norm": 0.0012538135051727295,
-      "learning_rate": 0.00019999864899049704,
-      "loss": 46.0,
-      "step": 21651
-    },
-    {
-      "epoch": 1.6554466043542253,
-      "grad_norm": 0.0010261310962960124,
-      "learning_rate": 0.00019999864886563796,
-      "loss": 46.0,
-      "step": 21652
-    },
-    {
-      "epoch": 1.655523061337615,
-      "grad_norm": 0.002302152570337057,
-      "learning_rate": 0.0001999986487407731,
-      "loss": 46.0,
-      "step": 21653
-    },
-    {
-      "epoch": 1.6555995183210046,
-      "grad_norm": 0.001607475453056395,
-      "learning_rate": 0.0001999986486159025,
-      "loss": 46.0,
-      "step": 21654
-    },
-    {
-      "epoch": 1.6556759753043944,
-      "grad_norm": 0.0013634710339829326,
-      "learning_rate": 0.00019999864849102607,
-      "loss": 46.0,
-      "step": 21655
-    },
-    {
-      "epoch": 1.6557524322877841,
-      "grad_norm": 0.0014689142117276788,
-      "learning_rate": 0.00019999864836614393,
-      "loss": 46.0,
-      "step": 21656
-    },
-    {
-      "epoch": 1.6558288892711737,
-      "grad_norm": 0.006453211419284344,
-      "learning_rate": 0.000199998648241256,
-      "loss": 46.0,
-      "step": 21657
-    },
-    {
-      "epoch": 1.6559053462545634,
-      "grad_norm": 0.0011455732164904475,
-      "learning_rate": 0.00019999864811636231,
-      "loss": 46.0,
-      "step": 21658
-    },
-    {
-      "epoch": 1.6559818032379532,
-      "grad_norm": 0.003112090053036809,
-      "learning_rate": 0.00019999864799146285,
-      "loss": 46.0,
-      "step": 21659
-    },
-    {
-      "epoch": 1.656058260221343,
-      "grad_norm": 0.004679406061768532,
-      "learning_rate": 0.0001999986478665576,
-      "loss": 46.0,
-      "step": 21660
-    },
-    {
-      "epoch": 1.6561347172047327,
-      "grad_norm": 0.0004650199261959642,
-      "learning_rate": 0.0001999986477416466,
-      "loss": 46.0,
-      "step": 21661
-    },
-    {
-      "epoch": 1.6562111741881225,
-      "grad_norm": 0.0006998194148764014,
-      "learning_rate": 0.00019999864761672983,
-      "loss": 46.0,
-      "step": 21662
-    },
-    {
-      "epoch": 1.6562876311715122,
-      "grad_norm": 0.0003703190595842898,
-      "learning_rate": 0.00019999864749180727,
-      "loss": 46.0,
-      "step": 21663
-    },
-    {
-      "epoch": 1.656364088154902,
-      "grad_norm": 0.011942723765969276,
-      "learning_rate": 0.000199998647366879,
-      "loss": 46.0,
-      "step": 21664
-    },
-    {
-      "epoch": 1.6564405451382915,
-      "grad_norm": 0.0004818958695977926,
-      "learning_rate": 0.0001999986472419449,
-      "loss": 46.0,
-      "step": 21665
-    },
-    {
-      "epoch": 1.6565170021216813,
-      "grad_norm": 0.0007157226791605353,
-      "learning_rate": 0.00019999864711700503,
-      "loss": 46.0,
-      "step": 21666
-    },
-    {
-      "epoch": 1.656593459105071,
-      "grad_norm": 0.002656287979334593,
-      "learning_rate": 0.00019999864699205943,
-      "loss": 46.0,
-      "step": 21667
-    },
-    {
-      "epoch": 1.6566699160884606,
-      "grad_norm": 0.0006410392234101892,
-      "learning_rate": 0.00019999864686710803,
-      "loss": 46.0,
-      "step": 21668
-    },
-    {
-      "epoch": 1.6567463730718504,
-      "grad_norm": 0.011426913551986217,
-      "learning_rate": 0.0001999986467421509,
-      "loss": 46.0,
-      "step": 21669
-    },
-    {
-      "epoch": 1.6568228300552401,
-      "grad_norm": 0.0008393619209527969,
-      "learning_rate": 0.00019999864661718794,
-      "loss": 46.0,
-      "step": 21670
-    },
-    {
-      "epoch": 1.6568992870386299,
-      "grad_norm": 0.0008457890362478793,
-      "learning_rate": 0.00019999864649221925,
-      "loss": 46.0,
-      "step": 21671
-    },
-    {
-      "epoch": 1.6569757440220196,
-      "grad_norm": 0.0005599437281489372,
-      "learning_rate": 0.00019999864636724478,
-      "loss": 46.0,
-      "step": 21672
-    },
-    {
-      "epoch": 1.6570522010054094,
-      "grad_norm": 0.0021990975365042686,
-      "learning_rate": 0.00019999864624226454,
-      "loss": 46.0,
-      "step": 21673
-    },
-    {
-      "epoch": 1.6571286579887992,
-      "grad_norm": 0.0010131970047950745,
-      "learning_rate": 0.00019999864611727853,
-      "loss": 46.0,
-      "step": 21674
-    },
-    {
-      "epoch": 1.657205114972189,
-      "grad_norm": 0.0009348964085802436,
-      "learning_rate": 0.00019999864599228677,
-      "loss": 46.0,
-      "step": 21675
-    },
-    {
-      "epoch": 1.6572815719555785,
-      "grad_norm": 0.0022120086941868067,
-      "learning_rate": 0.0001999986458672892,
-      "loss": 46.0,
-      "step": 21676
-    },
-    {
-      "epoch": 1.6573580289389682,
-      "grad_norm": 0.0009733859915286303,
-      "learning_rate": 0.0001999986457422859,
-      "loss": 46.0,
-      "step": 21677
-    },
-    {
-      "epoch": 1.657434485922358,
-      "grad_norm": 0.0010268380865454674,
-      "learning_rate": 0.00019999864561727682,
-      "loss": 46.0,
-      "step": 21678
-    },
-    {
-      "epoch": 1.6575109429057475,
-      "grad_norm": 0.0009617885807529092,
-      "learning_rate": 0.00019999864549226194,
-      "loss": 46.0,
-      "step": 21679
-    },
-    {
-      "epoch": 1.6575873998891373,
-      "grad_norm": 0.0027607481461018324,
-      "learning_rate": 0.00019999864536724134,
-      "loss": 46.0,
-      "step": 21680
-    },
-    {
-      "epoch": 1.657663856872527,
-      "grad_norm": 0.005382539238780737,
-      "learning_rate": 0.00019999864524221494,
-      "loss": 46.0,
-      "step": 21681
-    },
-    {
-      "epoch": 1.6577403138559168,
-      "grad_norm": 0.0008170593064278364,
-      "learning_rate": 0.00019999864511718277,
-      "loss": 46.0,
-      "step": 21682
-    },
-    {
-      "epoch": 1.6578167708393066,
-      "grad_norm": 0.0009224091772921383,
-      "learning_rate": 0.00019999864499214485,
-      "loss": 46.0,
-      "step": 21683
-    },
-    {
-      "epoch": 1.6578932278226963,
-      "grad_norm": 0.0004626004374586046,
-      "learning_rate": 0.00019999864486710116,
-      "loss": 46.0,
-      "step": 21684
-    },
-    {
-      "epoch": 1.657969684806086,
-      "grad_norm": 0.0006666372064501047,
-      "learning_rate": 0.0001999986447420517,
-      "loss": 46.0,
-      "step": 21685
-    },
-    {
-      "epoch": 1.6580461417894758,
-      "grad_norm": 0.0020343840587884188,
-      "learning_rate": 0.00019999864461699645,
-      "loss": 46.0,
-      "step": 21686
-    },
-    {
-      "epoch": 1.6581225987728654,
-      "grad_norm": 0.0007447120733559132,
-      "learning_rate": 0.00019999864449193544,
-      "loss": 46.0,
-      "step": 21687
-    },
-    {
-      "epoch": 1.6581990557562551,
-      "grad_norm": 0.00390347046777606,
-      "learning_rate": 0.00019999864436686865,
-      "loss": 46.0,
-      "step": 21688
-    },
-    {
-      "epoch": 1.658275512739645,
-      "grad_norm": 0.0008550811326131225,
-      "learning_rate": 0.00019999864424179612,
-      "loss": 46.0,
-      "step": 21689
-    },
-    {
-      "epoch": 1.6583519697230344,
-      "grad_norm": 0.003112517297267914,
-      "learning_rate": 0.0001999986441167178,
-      "loss": 46.0,
-      "step": 21690
-    },
-    {
-      "epoch": 1.6584284267064242,
-      "grad_norm": 0.001175080076791346,
-      "learning_rate": 0.0001999986439916337,
-      "loss": 46.0,
-      "step": 21691
-    },
-    {
-      "epoch": 1.658504883689814,
-      "grad_norm": 0.0012133476557210088,
-      "learning_rate": 0.00019999864386654385,
-      "loss": 46.0,
-      "step": 21692
-    },
-    {
-      "epoch": 1.6585813406732037,
-      "grad_norm": 0.0014472611946985126,
-      "learning_rate": 0.00019999864374144823,
-      "loss": 46.0,
-      "step": 21693
-    },
-    {
-      "epoch": 1.6586577976565935,
-      "grad_norm": 0.00025356828700751066,
-      "learning_rate": 0.00019999864361634683,
-      "loss": 46.0,
-      "step": 21694
-    },
-    {
-      "epoch": 1.6587342546399833,
-      "grad_norm": 0.003534167306497693,
-      "learning_rate": 0.00019999864349123966,
-      "loss": 46.0,
-      "step": 21695
-    },
-    {
-      "epoch": 1.658810711623373,
-      "grad_norm": 0.0037848169449716806,
-      "learning_rate": 0.00019999864336612674,
-      "loss": 46.0,
-      "step": 21696
-    },
-    {
-      "epoch": 1.6588871686067628,
-      "grad_norm": 0.007858806289732456,
-      "learning_rate": 0.00019999864324100804,
-      "loss": 46.0,
-      "step": 21697
-    },
-    {
-      "epoch": 1.6589636255901523,
-      "grad_norm": 0.0003288577136117965,
-      "learning_rate": 0.00019999864311588358,
-      "loss": 46.0,
-      "step": 21698
-    },
-    {
-      "epoch": 1.659040082573542,
-      "grad_norm": 0.0005986799951642752,
-      "learning_rate": 0.00019999864299075334,
-      "loss": 46.0,
-      "step": 21699
-    },
-    {
-      "epoch": 1.6591165395569318,
-      "grad_norm": 0.000887929811142385,
-      "learning_rate": 0.00019999864286561733,
-      "loss": 46.0,
-      "step": 21700
-    },
-    {
-      "epoch": 1.6591929965403214,
-      "grad_norm": 0.001999678323045373,
-      "learning_rate": 0.00019999864274047554,
-      "loss": 46.0,
-      "step": 21701
-    },
-    {
-      "epoch": 1.6592694535237111,
-      "grad_norm": 0.0022027036175131798,
-      "learning_rate": 0.00019999864261532798,
-      "loss": 46.0,
-      "step": 21702
-    },
-    {
-      "epoch": 1.659345910507101,
-      "grad_norm": 0.0006224642274901271,
-      "learning_rate": 0.00019999864249017467,
-      "loss": 46.0,
-      "step": 21703
-    },
-    {
-      "epoch": 1.6594223674904907,
-      "grad_norm": 0.0006737412768416107,
-      "learning_rate": 0.00019999864236501557,
-      "loss": 46.0,
-      "step": 21704
-    },
-    {
-      "epoch": 1.6594988244738804,
-      "grad_norm": 0.0006587451207451522,
-      "learning_rate": 0.00019999864223985072,
-      "loss": 46.0,
-      "step": 21705
-    },
-    {
-      "epoch": 1.6595752814572702,
-      "grad_norm": 0.0009604567312635481,
-      "learning_rate": 0.0001999986421146801,
-      "loss": 46.0,
-      "step": 21706
-    },
-    {
-      "epoch": 1.65965173844066,
-      "grad_norm": 0.0011488718446344137,
-      "learning_rate": 0.00019999864198950372,
-      "loss": 46.0,
-      "step": 21707
-    },
-    {
-      "epoch": 1.6597281954240497,
-      "grad_norm": 0.0037051609251648188,
-      "learning_rate": 0.00019999864186432155,
-      "loss": 46.0,
-      "step": 21708
-    },
-    {
-      "epoch": 1.6598046524074392,
-      "grad_norm": 0.0055769700556993484,
-      "learning_rate": 0.0001999986417391336,
-      "loss": 46.0,
-      "step": 21709
-    },
-    {
-      "epoch": 1.659881109390829,
-      "grad_norm": 0.0006354156066663563,
-      "learning_rate": 0.0001999986416139399,
-      "loss": 46.0,
-      "step": 21710
-    },
-    {
-      "epoch": 1.6599575663742185,
-      "grad_norm": 0.0010919993510469794,
-      "learning_rate": 0.00019999864148874044,
-      "loss": 46.0,
-      "step": 21711
-    },
-    {
-      "epoch": 1.6600340233576083,
-      "grad_norm": 0.0010993661126121879,
-      "learning_rate": 0.00019999864136353518,
-      "loss": 46.0,
-      "step": 21712
-    },
-    {
-      "epoch": 1.660110480340998,
-      "grad_norm": 0.005171710625290871,
-      "learning_rate": 0.00019999864123832416,
-      "loss": 46.0,
-      "step": 21713
-    },
-    {
-      "epoch": 1.6601869373243878,
-      "grad_norm": 0.0040898979641497135,
-      "learning_rate": 0.0001999986411131074,
-      "loss": 46.0,
-      "step": 21714
-    },
-    {
-      "epoch": 1.6602633943077776,
-      "grad_norm": 0.0017966422019526362,
-      "learning_rate": 0.00019999864098788485,
-      "loss": 46.0,
-      "step": 21715
-    },
-    {
-      "epoch": 1.6603398512911673,
-      "grad_norm": 0.0013107439735904336,
-      "learning_rate": 0.00019999864086265654,
-      "loss": 46.0,
-      "step": 21716
-    },
-    {
-      "epoch": 1.660416308274557,
-      "grad_norm": 0.0007092754240147769,
-      "learning_rate": 0.00019999864073742246,
-      "loss": 46.0,
-      "step": 21717
-    },
-    {
-      "epoch": 1.6604927652579469,
-      "grad_norm": 0.0018959550652652979,
-      "learning_rate": 0.00019999864061218258,
-      "loss": 46.0,
-      "step": 21718
-    },
-    {
-      "epoch": 1.6605692222413366,
-      "grad_norm": 0.004546552896499634,
-      "learning_rate": 0.00019999864048693696,
-      "loss": 46.0,
-      "step": 21719
-    },
-    {
-      "epoch": 1.6606456792247262,
-      "grad_norm": 0.005949110724031925,
-      "learning_rate": 0.00019999864036168553,
-      "loss": 46.0,
-      "step": 21720
-    },
-    {
-      "epoch": 1.660722136208116,
-      "grad_norm": 0.0012279408983886242,
-      "learning_rate": 0.0001999986402364284,
-      "loss": 46.0,
-      "step": 21721
-    },
-    {
-      "epoch": 1.6607985931915055,
-      "grad_norm": 0.0006957650766707957,
-      "learning_rate": 0.00019999864011116544,
-      "loss": 46.0,
-      "step": 21722
-    },
-    {
-      "epoch": 1.6608750501748952,
-      "grad_norm": 0.004193790722638369,
-      "learning_rate": 0.00019999863998589675,
-      "loss": 46.0,
-      "step": 21723
-    },
-    {
-      "epoch": 1.660951507158285,
-      "grad_norm": 0.0019929578993469477,
-      "learning_rate": 0.00019999863986062226,
-      "loss": 46.0,
-      "step": 21724
-    },
-    {
-      "epoch": 1.6610279641416748,
-      "grad_norm": 0.0020179208368062973,
-      "learning_rate": 0.00019999863973534202,
-      "loss": 46.0,
-      "step": 21725
-    },
-    {
-      "epoch": 1.6611044211250645,
-      "grad_norm": 0.0011595117393881083,
-      "learning_rate": 0.00019999863961005603,
-      "loss": 46.0,
-      "step": 21726
-    },
-    {
-      "epoch": 1.6611808781084543,
-      "grad_norm": 0.004813872277736664,
-      "learning_rate": 0.00019999863948476422,
-      "loss": 46.0,
-      "step": 21727
-    },
-    {
-      "epoch": 1.661257335091844,
-      "grad_norm": 0.0007569942972622812,
-      "learning_rate": 0.0001999986393594667,
-      "loss": 46.0,
-      "step": 21728
-    },
-    {
-      "epoch": 1.6613337920752338,
-      "grad_norm": 0.0005347812548279762,
-      "learning_rate": 0.00019999863923416336,
-      "loss": 46.0,
-      "step": 21729
-    },
-    {
-      "epoch": 1.6614102490586236,
-      "grad_norm": 0.0021815341897308826,
-      "learning_rate": 0.00019999863910885428,
-      "loss": 46.0,
-      "step": 21730
-    },
-    {
-      "epoch": 1.661486706042013,
-      "grad_norm": 0.001619148999452591,
-      "learning_rate": 0.0001999986389835394,
-      "loss": 46.0,
-      "step": 21731
-    },
-    {
-      "epoch": 1.6615631630254029,
-      "grad_norm": 0.0012437907280400395,
-      "learning_rate": 0.00019999863885821877,
-      "loss": 46.0,
-      "step": 21732
-    },
-    {
-      "epoch": 1.6616396200087924,
-      "grad_norm": 0.0004048449045512825,
-      "learning_rate": 0.00019999863873289237,
-      "loss": 46.0,
-      "step": 21733
-    },
-    {
-      "epoch": 1.6617160769921822,
-      "grad_norm": 0.0020641072187572718,
-      "learning_rate": 0.0001999986386075602,
-      "loss": 46.0,
-      "step": 21734
-    },
-    {
-      "epoch": 1.661792533975572,
-      "grad_norm": 0.0007592297624796629,
-      "learning_rate": 0.00019999863848222226,
-      "loss": 46.0,
-      "step": 21735
-    },
-    {
-      "epoch": 1.6618689909589617,
-      "grad_norm": 0.00298447348177433,
-      "learning_rate": 0.00019999863835687857,
-      "loss": 46.0,
-      "step": 21736
-    },
-    {
-      "epoch": 1.6619454479423514,
-      "grad_norm": 0.0010736165568232536,
-      "learning_rate": 0.0001999986382315291,
-      "loss": 46.0,
-      "step": 21737
-    },
-    {
-      "epoch": 1.6620219049257412,
-      "grad_norm": 0.0012079750886186957,
-      "learning_rate": 0.00019999863810617384,
-      "loss": 46.0,
-      "step": 21738
-    },
-    {
-      "epoch": 1.662098361909131,
-      "grad_norm": 0.0012040602741762996,
-      "learning_rate": 0.00019999863798081282,
-      "loss": 46.0,
-      "step": 21739
-    },
-    {
-      "epoch": 1.6621748188925207,
-      "grad_norm": 0.004665082786232233,
-      "learning_rate": 0.00019999863785544604,
-      "loss": 46.0,
-      "step": 21740
-    },
-    {
-      "epoch": 1.6622512758759103,
-      "grad_norm": 0.0017301623011007905,
-      "learning_rate": 0.00019999863773007348,
-      "loss": 46.0,
-      "step": 21741
-    },
-    {
-      "epoch": 1.6623277328593,
-      "grad_norm": 0.0015062422025948763,
-      "learning_rate": 0.00019999863760469518,
-      "loss": 46.0,
-      "step": 21742
-    },
-    {
-      "epoch": 1.6624041898426898,
-      "grad_norm": 0.0022936842869967222,
-      "learning_rate": 0.00019999863747931107,
-      "loss": 46.0,
-      "step": 21743
-    },
-    {
-      "epoch": 1.6624806468260793,
-      "grad_norm": 0.0012678923085331917,
-      "learning_rate": 0.0001999986373539212,
-      "loss": 46.0,
-      "step": 21744
-    },
-    {
-      "epoch": 1.662557103809469,
-      "grad_norm": 0.0014153685187920928,
-      "learning_rate": 0.00019999863722852557,
-      "loss": 46.0,
-      "step": 21745
-    },
-    {
-      "epoch": 1.6626335607928588,
-      "grad_norm": 0.0016152793541550636,
-      "learning_rate": 0.00019999863710312417,
-      "loss": 46.0,
-      "step": 21746
-    },
-    {
-      "epoch": 1.6627100177762486,
-      "grad_norm": 0.001285429927520454,
-      "learning_rate": 0.00019999863697771702,
-      "loss": 46.0,
-      "step": 21747
-    },
-    {
-      "epoch": 1.6627864747596384,
-      "grad_norm": 0.0008521474082954228,
-      "learning_rate": 0.00019999863685230405,
-      "loss": 46.0,
-      "step": 21748
-    },
-    {
-      "epoch": 1.6628629317430281,
-      "grad_norm": 0.0005075836088508368,
-      "learning_rate": 0.00019999863672688533,
-      "loss": 46.0,
-      "step": 21749
-    },
-    {
-      "epoch": 1.6629393887264179,
-      "grad_norm": 0.0015887904446572065,
-      "learning_rate": 0.00019999863660146087,
-      "loss": 46.0,
-      "step": 21750
-    },
-    {
-      "epoch": 1.6630158457098076,
-      "grad_norm": 0.0013447246747091413,
-      "learning_rate": 0.00019999863647603063,
-      "loss": 46.0,
-      "step": 21751
-    },
-    {
-      "epoch": 1.6630923026931972,
-      "grad_norm": 0.0022178690414875746,
-      "learning_rate": 0.00019999863635059462,
-      "loss": 46.0,
-      "step": 21752
-    },
-    {
-      "epoch": 1.663168759676587,
-      "grad_norm": 0.001922041061334312,
-      "learning_rate": 0.00019999863622515283,
-      "loss": 46.0,
-      "step": 21753
-    },
-    {
-      "epoch": 1.6632452166599767,
-      "grad_norm": 0.0012107868678867817,
-      "learning_rate": 0.00019999863609970525,
-      "loss": 46.0,
-      "step": 21754
-    },
-    {
-      "epoch": 1.6633216736433662,
-      "grad_norm": 0.0007862246129661798,
-      "learning_rate": 0.00019999863597425192,
-      "loss": 46.0,
-      "step": 21755
-    },
-    {
-      "epoch": 1.663398130626756,
-      "grad_norm": 0.0012303428957238793,
-      "learning_rate": 0.00019999863584879284,
-      "loss": 46.0,
-      "step": 21756
-    },
-    {
-      "epoch": 1.6634745876101458,
-      "grad_norm": 0.0025524809025228024,
-      "learning_rate": 0.000199998635723328,
-      "loss": 46.0,
-      "step": 21757
-    },
-    {
-      "epoch": 1.6635510445935355,
-      "grad_norm": 0.0014661196619272232,
-      "learning_rate": 0.00019999863559785734,
-      "loss": 46.0,
-      "step": 21758
-    },
-    {
-      "epoch": 1.6636275015769253,
-      "grad_norm": 0.0012701055966317654,
-      "learning_rate": 0.00019999863547238094,
-      "loss": 46.0,
-      "step": 21759
-    },
-    {
-      "epoch": 1.663703958560315,
-      "grad_norm": 0.002395395655184984,
-      "learning_rate": 0.00019999863534689877,
-      "loss": 46.0,
-      "step": 21760
-    },
-    {
-      "epoch": 1.6637804155437048,
-      "grad_norm": 0.0024629172403365374,
-      "learning_rate": 0.00019999863522141082,
-      "loss": 46.0,
-      "step": 21761
-    },
-    {
-      "epoch": 1.6638568725270946,
-      "grad_norm": 0.0006907899514771998,
-      "learning_rate": 0.0001999986350959171,
-      "loss": 46.0,
-      "step": 21762
-    },
-    {
-      "epoch": 1.6639333295104841,
-      "grad_norm": 0.000729088787920773,
-      "learning_rate": 0.00019999863497041761,
-      "loss": 46.0,
-      "step": 21763
-    },
-    {
-      "epoch": 1.6640097864938739,
-      "grad_norm": 0.0012969098752364516,
-      "learning_rate": 0.00019999863484491238,
-      "loss": 46.0,
-      "step": 21764
-    },
-    {
-      "epoch": 1.6640862434772636,
-      "grad_norm": 0.0008993445662781596,
-      "learning_rate": 0.00019999863471940134,
-      "loss": 46.0,
-      "step": 21765
-    },
-    {
-      "epoch": 1.6641627004606532,
-      "grad_norm": 0.0010818386217579246,
-      "learning_rate": 0.00019999863459388458,
-      "loss": 46.0,
-      "step": 21766
-    },
-    {
-      "epoch": 1.664239157444043,
-      "grad_norm": 0.0005669799284078181,
-      "learning_rate": 0.00019999863446836202,
-      "loss": 46.0,
-      "step": 21767
-    },
-    {
-      "epoch": 1.6643156144274327,
-      "grad_norm": 0.001311693456955254,
-      "learning_rate": 0.0001999986343428337,
-      "loss": 46.0,
-      "step": 21768
-    },
-    {
-      "epoch": 1.6643920714108225,
-      "grad_norm": 0.0006725798011757433,
-      "learning_rate": 0.0001999986342172996,
-      "loss": 46.0,
-      "step": 21769
-    },
-    {
-      "epoch": 1.6644685283942122,
-      "grad_norm": 0.0023361369967460632,
-      "learning_rate": 0.0001999986340917597,
-      "loss": 46.0,
-      "step": 21770
-    },
-    {
-      "epoch": 1.664544985377602,
-      "grad_norm": 0.001291585504077375,
-      "learning_rate": 0.0001999986339662141,
-      "loss": 46.0,
-      "step": 21771
-    },
-    {
-      "epoch": 1.6646214423609917,
-      "grad_norm": 0.001266077859327197,
-      "learning_rate": 0.0001999986338406627,
-      "loss": 46.0,
-      "step": 21772
-    },
-    {
-      "epoch": 1.6646978993443815,
-      "grad_norm": 0.0005462805274873972,
-      "learning_rate": 0.0001999986337151055,
-      "loss": 46.0,
-      "step": 21773
-    },
-    {
-      "epoch": 1.664774356327771,
-      "grad_norm": 0.0016762202139943838,
-      "learning_rate": 0.00019999863358954257,
-      "loss": 46.0,
-      "step": 21774
-    },
-    {
-      "epoch": 1.6648508133111608,
-      "grad_norm": 0.0026368971448391676,
-      "learning_rate": 0.00019999863346397385,
-      "loss": 46.0,
-      "step": 21775
-    },
-    {
-      "epoch": 1.6649272702945506,
-      "grad_norm": 0.000918002100661397,
-      "learning_rate": 0.00019999863333839936,
-      "loss": 46.0,
-      "step": 21776
-    },
-    {
-      "epoch": 1.66500372727794,
-      "grad_norm": 0.0015913642710074782,
-      "learning_rate": 0.0001999986332128191,
-      "loss": 46.0,
-      "step": 21777
-    },
-    {
-      "epoch": 1.6650801842613299,
-      "grad_norm": 0.001786625711247325,
-      "learning_rate": 0.0001999986330872331,
-      "loss": 46.0,
-      "step": 21778
-    },
-    {
-      "epoch": 1.6651566412447196,
-      "grad_norm": 0.0013834276469424367,
-      "learning_rate": 0.00019999863296164133,
-      "loss": 46.0,
-      "step": 21779
-    },
-    {
-      "epoch": 1.6652330982281094,
-      "grad_norm": 0.003367658006027341,
-      "learning_rate": 0.00019999863283604375,
-      "loss": 46.0,
-      "step": 21780
-    },
-    {
-      "epoch": 1.6653095552114991,
-      "grad_norm": 0.0006114642601460218,
-      "learning_rate": 0.00019999863271044041,
-      "loss": 46.0,
-      "step": 21781
-    },
-    {
-      "epoch": 1.665386012194889,
-      "grad_norm": 0.0014529459876939654,
-      "learning_rate": 0.0001999986325848313,
-      "loss": 46.0,
-      "step": 21782
-    },
-    {
-      "epoch": 1.6654624691782787,
-      "grad_norm": 0.028930600732564926,
-      "learning_rate": 0.00019999863245921646,
-      "loss": 46.0,
-      "step": 21783
-    },
-    {
-      "epoch": 1.6655389261616684,
-      "grad_norm": 0.0012406535679474473,
-      "learning_rate": 0.0001999986323335958,
-      "loss": 46.0,
-      "step": 21784
-    },
-    {
-      "epoch": 1.665615383145058,
-      "grad_norm": 0.0006059122388251126,
-      "learning_rate": 0.00019999863220796938,
-      "loss": 46.0,
-      "step": 21785
-    },
-    {
-      "epoch": 1.6656918401284477,
-      "grad_norm": 0.000924452324397862,
-      "learning_rate": 0.00019999863208233721,
-      "loss": 46.0,
-      "step": 21786
-    },
-    {
-      "epoch": 1.6657682971118375,
-      "grad_norm": 0.000638169061858207,
-      "learning_rate": 0.00019999863195669927,
-      "loss": 46.0,
-      "step": 21787
-    },
-    {
-      "epoch": 1.665844754095227,
-      "grad_norm": 0.0006390910129994154,
-      "learning_rate": 0.00019999863183105555,
-      "loss": 46.0,
-      "step": 21788
-    },
-    {
-      "epoch": 1.6659212110786168,
-      "grad_norm": 0.0007224271539598703,
-      "learning_rate": 0.00019999863170540606,
-      "loss": 46.0,
-      "step": 21789
-    },
-    {
-      "epoch": 1.6659976680620066,
-      "grad_norm": 0.016078229993581772,
-      "learning_rate": 0.00019999863157975082,
-      "loss": 46.0,
-      "step": 21790
-    },
-    {
-      "epoch": 1.6660741250453963,
-      "grad_norm": 0.001869305968284607,
-      "learning_rate": 0.00019999863145408981,
-      "loss": 46.0,
-      "step": 21791
-    },
-    {
-      "epoch": 1.666150582028786,
-      "grad_norm": 0.001883415738120675,
-      "learning_rate": 0.000199998631328423,
-      "loss": 46.0,
-      "step": 21792
-    },
-    {
-      "epoch": 1.6662270390121758,
-      "grad_norm": 0.0008455296629108489,
-      "learning_rate": 0.00019999863120275047,
-      "loss": 46.0,
-      "step": 21793
-    },
-    {
-      "epoch": 1.6663034959955656,
-      "grad_norm": 0.001925777760334313,
-      "learning_rate": 0.00019999863107707212,
-      "loss": 46.0,
-      "step": 21794
-    },
-    {
-      "epoch": 1.6663799529789554,
-      "grad_norm": 0.0011103162541985512,
-      "learning_rate": 0.000199998630951388,
-      "loss": 46.0,
-      "step": 21795
-    },
-    {
-      "epoch": 1.666456409962345,
-      "grad_norm": 0.001968169817700982,
-      "learning_rate": 0.00019999863082569816,
-      "loss": 46.0,
-      "step": 21796
-    },
-    {
-      "epoch": 1.6665328669457347,
-      "grad_norm": 0.003113889368250966,
-      "learning_rate": 0.0001999986307000025,
-      "loss": 46.0,
-      "step": 21797
-    },
-    {
-      "epoch": 1.6666093239291244,
-      "grad_norm": 0.001178809441626072,
-      "learning_rate": 0.0001999986305743011,
-      "loss": 46.0,
-      "step": 21798
-    },
-    {
-      "epoch": 1.666685780912514,
-      "grad_norm": 0.0037485749926418066,
-      "learning_rate": 0.00019999863044859392,
-      "loss": 46.0,
-      "step": 21799
-    },
-    {
-      "epoch": 1.6667622378959037,
-      "grad_norm": 0.00220527988858521,
-      "learning_rate": 0.00019999863032288097,
-      "loss": 46.0,
-      "step": 21800
-    },
-    {
-      "epoch": 1.6668386948792935,
-      "grad_norm": 0.0013013853458687663,
-      "learning_rate": 0.00019999863019716228,
-      "loss": 46.0,
-      "step": 21801
-    },
-    {
-      "epoch": 1.6669151518626832,
-      "grad_norm": 0.0015391850611194968,
-      "learning_rate": 0.00019999863007143776,
-      "loss": 46.0,
-      "step": 21802
-    },
-    {
-      "epoch": 1.666991608846073,
-      "grad_norm": 0.008504524827003479,
-      "learning_rate": 0.0001999986299457075,
-      "loss": 46.0,
-      "step": 21803
-    },
-    {
-      "epoch": 1.6670680658294628,
-      "grad_norm": 0.0033453081268817186,
-      "learning_rate": 0.00019999862981997152,
-      "loss": 46.0,
-      "step": 21804
-    },
-    {
-      "epoch": 1.6671445228128525,
-      "grad_norm": 0.0007444752845913172,
-      "learning_rate": 0.00019999862969422968,
-      "loss": 46.0,
-      "step": 21805
-    },
-    {
-      "epoch": 1.6672209797962423,
-      "grad_norm": 0.004083487205207348,
-      "learning_rate": 0.00019999862956848215,
-      "loss": 46.0,
-      "step": 21806
-    },
-    {
-      "epoch": 1.6672974367796318,
-      "grad_norm": 0.00040972669376060367,
-      "learning_rate": 0.0001999986294427288,
-      "loss": 46.0,
-      "step": 21807
-    },
-    {
-      "epoch": 1.6673738937630216,
-      "grad_norm": 0.0005900819669477642,
-      "learning_rate": 0.0001999986293169697,
-      "loss": 46.0,
-      "step": 21808
-    },
-    {
-      "epoch": 1.6674503507464113,
-      "grad_norm": 0.0023416285403072834,
-      "learning_rate": 0.0001999986291912048,
-      "loss": 46.0,
-      "step": 21809
-    },
-    {
-      "epoch": 1.6675268077298009,
-      "grad_norm": 0.0019489447586238384,
-      "learning_rate": 0.0001999986290654342,
-      "loss": 46.0,
-      "step": 21810
-    },
-    {
-      "epoch": 1.6676032647131906,
-      "grad_norm": 0.002405709819868207,
-      "learning_rate": 0.00019999862893965776,
-      "loss": 46.0,
-      "step": 21811
-    },
-    {
-      "epoch": 1.6676797216965804,
-      "grad_norm": 0.0006889355136081576,
-      "learning_rate": 0.0001999986288138756,
-      "loss": 46.0,
-      "step": 21812
-    },
-    {
-      "epoch": 1.6677561786799702,
-      "grad_norm": 0.0017988828476518393,
-      "learning_rate": 0.00019999862868808762,
-      "loss": 46.0,
-      "step": 21813
-    },
-    {
-      "epoch": 1.66783263566336,
-      "grad_norm": 0.0013023015344515443,
-      "learning_rate": 0.00019999862856229393,
-      "loss": 46.0,
-      "step": 21814
-    },
-    {
-      "epoch": 1.6679090926467497,
-      "grad_norm": 0.001345370663329959,
-      "learning_rate": 0.00019999862843649444,
-      "loss": 46.0,
-      "step": 21815
-    },
-    {
-      "epoch": 1.6679855496301395,
-      "grad_norm": 0.0011189555516466498,
-      "learning_rate": 0.00019999862831068918,
-      "loss": 46.0,
-      "step": 21816
-    },
-    {
-      "epoch": 1.6680620066135292,
-      "grad_norm": 0.001372084952890873,
-      "learning_rate": 0.00019999862818487814,
-      "loss": 46.0,
-      "step": 21817
-    },
-    {
-      "epoch": 1.6681384635969188,
-      "grad_norm": 0.003140259301289916,
-      "learning_rate": 0.00019999862805906136,
-      "loss": 46.0,
-      "step": 21818
-    },
-    {
-      "epoch": 1.6682149205803085,
-      "grad_norm": 0.00125603296328336,
-      "learning_rate": 0.00019999862793323878,
-      "loss": 46.0,
-      "step": 21819
-    },
-    {
-      "epoch": 1.6682913775636983,
-      "grad_norm": 0.006496195215731859,
-      "learning_rate": 0.00019999862780741045,
-      "loss": 46.0,
-      "step": 21820
-    },
-    {
-      "epoch": 1.6683678345470878,
-      "grad_norm": 0.004713891074061394,
-      "learning_rate": 0.00019999862768157634,
-      "loss": 46.0,
-      "step": 21821
-    },
-    {
-      "epoch": 1.6684442915304776,
-      "grad_norm": 0.00048765799147076905,
-      "learning_rate": 0.0001999986275557365,
-      "loss": 46.0,
-      "step": 21822
-    },
-    {
-      "epoch": 1.6685207485138673,
-      "grad_norm": 0.0018604713259264827,
-      "learning_rate": 0.00019999862742989084,
-      "loss": 46.0,
-      "step": 21823
-    },
-    {
-      "epoch": 1.668597205497257,
-      "grad_norm": 0.00674502644687891,
-      "learning_rate": 0.00019999862730403942,
-      "loss": 46.0,
-      "step": 21824
-    },
-    {
-      "epoch": 1.6686736624806469,
-      "grad_norm": 0.0006744790589436889,
-      "learning_rate": 0.00019999862717818222,
-      "loss": 46.0,
-      "step": 21825
-    },
-    {
-      "epoch": 1.6687501194640366,
-      "grad_norm": 0.002573287347331643,
-      "learning_rate": 0.00019999862705231928,
-      "loss": 46.0,
-      "step": 21826
-    },
-    {
-      "epoch": 1.6688265764474264,
-      "grad_norm": 0.006537720561027527,
-      "learning_rate": 0.00019999862692645056,
-      "loss": 46.0,
-      "step": 21827
-    },
-    {
-      "epoch": 1.6689030334308161,
-      "grad_norm": 0.002559009939432144,
-      "learning_rate": 0.00019999862680057607,
-      "loss": 46.0,
-      "step": 21828
-    },
-    {
-      "epoch": 1.6689794904142057,
-      "grad_norm": 0.003728767391294241,
-      "learning_rate": 0.0001999986266746958,
-      "loss": 46.0,
-      "step": 21829
-    },
-    {
-      "epoch": 1.6690559473975954,
-      "grad_norm": 0.006121995393186808,
-      "learning_rate": 0.00019999862654880977,
-      "loss": 46.0,
-      "step": 21830
-    },
-    {
-      "epoch": 1.6691324043809852,
-      "grad_norm": 0.0006884766626171768,
-      "learning_rate": 0.000199998626422918,
-      "loss": 46.0,
-      "step": 21831
-    },
-    {
-      "epoch": 1.6692088613643747,
-      "grad_norm": 0.0010282910661771894,
-      "learning_rate": 0.0001999986262970204,
-      "loss": 46.0,
-      "step": 21832
-    },
-    {
-      "epoch": 1.6692853183477645,
-      "grad_norm": 0.002935074269771576,
-      "learning_rate": 0.00019999862617111708,
-      "loss": 46.0,
-      "step": 21833
-    },
-    {
-      "epoch": 1.6693617753311543,
-      "grad_norm": 0.0022740764543414116,
-      "learning_rate": 0.00019999862604520798,
-      "loss": 46.0,
-      "step": 21834
-    },
-    {
-      "epoch": 1.669438232314544,
-      "grad_norm": 0.0006617013132199645,
-      "learning_rate": 0.0001999986259192931,
-      "loss": 46.0,
-      "step": 21835
-    },
-    {
-      "epoch": 1.6695146892979338,
-      "grad_norm": 0.0005657104193232954,
-      "learning_rate": 0.00019999862579337245,
-      "loss": 46.0,
-      "step": 21836
-    },
-    {
-      "epoch": 1.6695911462813235,
-      "grad_norm": 0.0014485052088275552,
-      "learning_rate": 0.00019999862566744605,
-      "loss": 46.0,
-      "step": 21837
-    },
-    {
-      "epoch": 1.6696676032647133,
-      "grad_norm": 0.0004112808674108237,
-      "learning_rate": 0.00019999862554151388,
-      "loss": 46.0,
-      "step": 21838
-    },
-    {
-      "epoch": 1.669744060248103,
-      "grad_norm": 0.0008636778802610934,
-      "learning_rate": 0.00019999862541557591,
-      "loss": 46.0,
-      "step": 21839
-    },
-    {
-      "epoch": 1.6698205172314926,
-      "grad_norm": 0.0013015868607908487,
-      "learning_rate": 0.0001999986252896322,
-      "loss": 46.0,
-      "step": 21840
-    },
-    {
-      "epoch": 1.6698969742148824,
-      "grad_norm": 0.0005731317214667797,
-      "learning_rate": 0.00019999862516368268,
-      "loss": 46.0,
-      "step": 21841
-    },
-    {
-      "epoch": 1.669973431198272,
-      "grad_norm": 0.000936092168558389,
-      "learning_rate": 0.00019999862503772745,
-      "loss": 46.0,
-      "step": 21842
-    },
-    {
-      "epoch": 1.6700498881816617,
-      "grad_norm": 0.001647431286983192,
-      "learning_rate": 0.0001999986249117664,
-      "loss": 46.0,
-      "step": 21843
-    },
-    {
-      "epoch": 1.6701263451650514,
-      "grad_norm": 0.0013484007213264704,
-      "learning_rate": 0.0001999986247857996,
-      "loss": 46.0,
-      "step": 21844
-    },
-    {
-      "epoch": 1.6702028021484412,
-      "grad_norm": 0.003102868562564254,
-      "learning_rate": 0.00019999862465982702,
-      "loss": 46.0,
-      "step": 21845
-    },
-    {
-      "epoch": 1.670279259131831,
-      "grad_norm": 0.0012187802931293845,
-      "learning_rate": 0.0001999986245338487,
-      "loss": 46.0,
-      "step": 21846
-    },
-    {
-      "epoch": 1.6703557161152207,
-      "grad_norm": 0.0012246815022081137,
-      "learning_rate": 0.00019999862440786458,
-      "loss": 46.0,
-      "step": 21847
-    },
-    {
-      "epoch": 1.6704321730986105,
-      "grad_norm": 0.0004368004738353193,
-      "learning_rate": 0.0001999986242818747,
-      "loss": 46.0,
-      "step": 21848
-    },
-    {
-      "epoch": 1.6705086300820002,
-      "grad_norm": 0.0021131793037056923,
-      "learning_rate": 0.00019999862415587906,
-      "loss": 46.0,
-      "step": 21849
-    },
-    {
-      "epoch": 1.67058508706539,
-      "grad_norm": 0.0055949329398572445,
-      "learning_rate": 0.00019999862402987764,
-      "loss": 46.0,
-      "step": 21850
-    },
-    {
-      "epoch": 1.6706615440487795,
-      "grad_norm": 0.002044996013864875,
-      "learning_rate": 0.00019999862390387047,
-      "loss": 46.0,
-      "step": 21851
-    },
-    {
-      "epoch": 1.6707380010321693,
-      "grad_norm": 0.0003055713605135679,
-      "learning_rate": 0.0001999986237778575,
-      "loss": 46.0,
-      "step": 21852
-    },
-    {
-      "epoch": 1.6708144580155588,
-      "grad_norm": 0.0019037730526179075,
-      "learning_rate": 0.00019999862365183878,
-      "loss": 46.0,
-      "step": 21853
-    },
-    {
-      "epoch": 1.6708909149989486,
-      "grad_norm": 0.0004695192910730839,
-      "learning_rate": 0.0001999986235258143,
-      "loss": 46.0,
-      "step": 21854
-    },
-    {
-      "epoch": 1.6709673719823384,
-      "grad_norm": 0.0014583783922716975,
-      "learning_rate": 0.000199998623399784,
-      "loss": 46.0,
-      "step": 21855
-    },
-    {
-      "epoch": 1.6710438289657281,
-      "grad_norm": 0.0012723759282380342,
-      "learning_rate": 0.000199998623273748,
-      "loss": 46.0,
-      "step": 21856
-    },
-    {
-      "epoch": 1.6711202859491179,
-      "grad_norm": 0.00046787571045570076,
-      "learning_rate": 0.00019999862314770618,
-      "loss": 46.0,
-      "step": 21857
-    },
-    {
-      "epoch": 1.6711967429325076,
-      "grad_norm": 0.0005144478054717183,
-      "learning_rate": 0.00019999862302165863,
-      "loss": 46.0,
-      "step": 21858
-    },
-    {
-      "epoch": 1.6712731999158974,
-      "grad_norm": 0.0008232161053456366,
-      "learning_rate": 0.00019999862289560527,
-      "loss": 46.0,
-      "step": 21859
-    },
-    {
-      "epoch": 1.6713496568992872,
-      "grad_norm": 0.0023117554374039173,
-      "learning_rate": 0.00019999862276954614,
-      "loss": 46.0,
-      "step": 21860
-    },
-    {
-      "epoch": 1.671426113882677,
-      "grad_norm": 0.0007232401403598487,
-      "learning_rate": 0.0001999986226434813,
-      "loss": 46.0,
-      "step": 21861
-    },
-    {
-      "epoch": 1.6715025708660665,
-      "grad_norm": 0.0013719139387831092,
-      "learning_rate": 0.00019999862251741064,
-      "loss": 46.0,
-      "step": 21862
-    },
-    {
-      "epoch": 1.6715790278494562,
-      "grad_norm": 0.0010404265485703945,
-      "learning_rate": 0.00019999862239133422,
-      "loss": 46.0,
-      "step": 21863
-    },
-    {
-      "epoch": 1.6716554848328458,
-      "grad_norm": 0.0009005528991110623,
-      "learning_rate": 0.00019999862226525203,
-      "loss": 46.0,
-      "step": 21864
-    },
-    {
-      "epoch": 1.6717319418162355,
-      "grad_norm": 0.0008152939262799919,
-      "learning_rate": 0.00019999862213916408,
-      "loss": 46.0,
-      "step": 21865
-    },
-    {
-      "epoch": 1.6718083987996253,
-      "grad_norm": 0.0020366471726447344,
-      "learning_rate": 0.00019999862201307034,
-      "loss": 46.0,
-      "step": 21866
-    },
-    {
-      "epoch": 1.671884855783015,
-      "grad_norm": 0.0038110660389065742,
-      "learning_rate": 0.00019999862188697085,
-      "loss": 46.0,
-      "step": 21867
-    },
-    {
-      "epoch": 1.6719613127664048,
-      "grad_norm": 0.0011139587732031941,
-      "learning_rate": 0.0001999986217608656,
-      "loss": 46.0,
-      "step": 21868
-    },
-    {
-      "epoch": 1.6720377697497946,
-      "grad_norm": 0.001885499688796699,
-      "learning_rate": 0.00019999862163475455,
-      "loss": 46.0,
-      "step": 21869
-    },
-    {
-      "epoch": 1.6721142267331843,
-      "grad_norm": 0.0012409884948283434,
-      "learning_rate": 0.00019999862150863775,
-      "loss": 46.0,
-      "step": 21870
-    },
-    {
-      "epoch": 1.672190683716574,
-      "grad_norm": 0.0007964686374180019,
-      "learning_rate": 0.00019999862138251516,
-      "loss": 46.0,
-      "step": 21871
-    },
-    {
-      "epoch": 1.6722671406999636,
-      "grad_norm": 0.000891521864105016,
-      "learning_rate": 0.0001999986212563868,
-      "loss": 46.0,
-      "step": 21872
-    },
-    {
-      "epoch": 1.6723435976833534,
-      "grad_norm": 0.0012375818332657218,
-      "learning_rate": 0.0001999986211302527,
-      "loss": 46.0,
-      "step": 21873
-    },
-    {
-      "epoch": 1.6724200546667431,
-      "grad_norm": 0.000987158971838653,
-      "learning_rate": 0.00019999862100411283,
-      "loss": 46.0,
-      "step": 21874
-    },
-    {
-      "epoch": 1.6724965116501327,
-      "grad_norm": 0.0033816248178482056,
-      "learning_rate": 0.00019999862087796718,
-      "loss": 46.0,
-      "step": 21875
-    },
-    {
-      "epoch": 1.6725729686335224,
-      "grad_norm": 0.001214248826727271,
-      "learning_rate": 0.00019999862075181576,
-      "loss": 46.0,
-      "step": 21876
-    },
-    {
-      "epoch": 1.6726494256169122,
-      "grad_norm": 0.001549255452118814,
-      "learning_rate": 0.0001999986206256586,
-      "loss": 46.0,
-      "step": 21877
-    },
-    {
-      "epoch": 1.672725882600302,
-      "grad_norm": 0.0010053038131445646,
-      "learning_rate": 0.00019999862049949562,
-      "loss": 46.0,
-      "step": 21878
-    },
-    {
-      "epoch": 1.6728023395836917,
-      "grad_norm": 0.0024029631167650223,
-      "learning_rate": 0.00019999862037332688,
-      "loss": 46.0,
-      "step": 21879
-    },
-    {
-      "epoch": 1.6728787965670815,
-      "grad_norm": 0.0015450253849849105,
-      "learning_rate": 0.0001999986202471524,
-      "loss": 46.0,
-      "step": 21880
-    },
-    {
-      "epoch": 1.6729552535504713,
-      "grad_norm": 0.001079551293514669,
-      "learning_rate": 0.00019999862012097213,
-      "loss": 46.0,
-      "step": 21881
-    },
-    {
-      "epoch": 1.673031710533861,
-      "grad_norm": 0.002845742041245103,
-      "learning_rate": 0.0001999986199947861,
-      "loss": 46.0,
-      "step": 21882
-    },
-    {
-      "epoch": 1.6731081675172506,
-      "grad_norm": 0.0027554119005799294,
-      "learning_rate": 0.00019999861986859428,
-      "loss": 46.0,
-      "step": 21883
-    },
-    {
-      "epoch": 1.6731846245006403,
-      "grad_norm": 0.0008502207929268479,
-      "learning_rate": 0.0001999986197423967,
-      "loss": 46.0,
-      "step": 21884
-    },
-    {
-      "epoch": 1.67326108148403,
-      "grad_norm": 0.0010531614534556866,
-      "learning_rate": 0.00019999861961619337,
-      "loss": 46.0,
-      "step": 21885
-    },
-    {
-      "epoch": 1.6733375384674196,
-      "grad_norm": 0.0039682453498244286,
-      "learning_rate": 0.00019999861948998425,
-      "loss": 46.0,
-      "step": 21886
-    },
-    {
-      "epoch": 1.6734139954508094,
-      "grad_norm": 0.0023424874525517225,
-      "learning_rate": 0.00019999861936376937,
-      "loss": 46.0,
-      "step": 21887
-    },
-    {
-      "epoch": 1.6734904524341991,
-      "grad_norm": 0.0013596605276688933,
-      "learning_rate": 0.00019999861923754872,
-      "loss": 46.0,
-      "step": 21888
-    },
-    {
-      "epoch": 1.673566909417589,
-      "grad_norm": 0.019456440582871437,
-      "learning_rate": 0.0001999986191113223,
-      "loss": 46.0,
-      "step": 21889
-    },
-    {
-      "epoch": 1.6736433664009787,
-      "grad_norm": 0.0010776534909382463,
-      "learning_rate": 0.0001999986189850901,
-      "loss": 46.0,
-      "step": 21890
-    },
-    {
-      "epoch": 1.6737198233843684,
-      "grad_norm": 0.00044219361734576523,
-      "learning_rate": 0.00019999861885885216,
-      "loss": 46.0,
-      "step": 21891
-    },
-    {
-      "epoch": 1.6737962803677582,
-      "grad_norm": 0.0012572972336784005,
-      "learning_rate": 0.00019999861873260842,
-      "loss": 46.0,
-      "step": 21892
-    },
-    {
-      "epoch": 1.673872737351148,
-      "grad_norm": 0.001081867958419025,
-      "learning_rate": 0.0001999986186063589,
-      "loss": 46.0,
-      "step": 21893
-    },
-    {
-      "epoch": 1.6739491943345375,
-      "grad_norm": 0.0007973661413416266,
-      "learning_rate": 0.00019999861848010367,
-      "loss": 46.0,
-      "step": 21894
-    },
-    {
-      "epoch": 1.6740256513179272,
-      "grad_norm": 0.001230223453603685,
-      "learning_rate": 0.0001999986183538426,
-      "loss": 46.0,
-      "step": 21895
-    },
-    {
-      "epoch": 1.674102108301317,
-      "grad_norm": 0.0044013215228915215,
-      "learning_rate": 0.0001999986182275758,
-      "loss": 46.0,
-      "step": 21896
-    },
-    {
-      "epoch": 1.6741785652847065,
-      "grad_norm": 0.0005416969652287662,
-      "learning_rate": 0.00019999861810130322,
-      "loss": 46.0,
-      "step": 21897
-    },
-    {
-      "epoch": 1.6742550222680963,
-      "grad_norm": 0.0005605234182439744,
-      "learning_rate": 0.0001999986179750249,
-      "loss": 46.0,
-      "step": 21898
-    },
-    {
-      "epoch": 1.674331479251486,
-      "grad_norm": 0.006108017172664404,
-      "learning_rate": 0.00019999861784874076,
-      "loss": 46.0,
-      "step": 21899
-    },
-    {
-      "epoch": 1.6744079362348758,
-      "grad_norm": 0.0013146072160452604,
-      "learning_rate": 0.0001999986177224509,
-      "loss": 46.0,
-      "step": 21900
-    },
-    {
-      "epoch": 1.6744843932182656,
-      "grad_norm": 0.0007573894108645618,
-      "learning_rate": 0.00019999861759615524,
-      "loss": 46.0,
-      "step": 21901
-    },
-    {
-      "epoch": 1.6745608502016553,
-      "grad_norm": 0.0016217163065448403,
-      "learning_rate": 0.0001999986174698538,
-      "loss": 46.0,
-      "step": 21902
-    },
-    {
-      "epoch": 1.674637307185045,
-      "grad_norm": 0.0018594225402921438,
-      "learning_rate": 0.0001999986173435466,
-      "loss": 46.0,
-      "step": 21903
-    },
-    {
-      "epoch": 1.6747137641684349,
-      "grad_norm": 0.0007377619622275233,
-      "learning_rate": 0.00019999861721723363,
-      "loss": 46.0,
-      "step": 21904
-    },
-    {
-      "epoch": 1.6747902211518244,
-      "grad_norm": 0.000803124625235796,
-      "learning_rate": 0.00019999861709091489,
-      "loss": 46.0,
-      "step": 21905
-    },
-    {
-      "epoch": 1.6748666781352142,
-      "grad_norm": 0.0007471116841770709,
-      "learning_rate": 0.00019999861696459042,
-      "loss": 46.0,
-      "step": 21906
-    },
-    {
-      "epoch": 1.674943135118604,
-      "grad_norm": 0.001843982026912272,
-      "learning_rate": 0.00019999861683826014,
-      "loss": 46.0,
-      "step": 21907
-    },
-    {
-      "epoch": 1.6750195921019935,
-      "grad_norm": 0.0005107785691507161,
-      "learning_rate": 0.0001999986167119241,
-      "loss": 46.0,
-      "step": 21908
-    },
-    {
-      "epoch": 1.6750960490853832,
-      "grad_norm": 0.00047672796063125134,
-      "learning_rate": 0.00019999861658558227,
-      "loss": 46.0,
-      "step": 21909
-    },
-    {
-      "epoch": 1.675172506068773,
-      "grad_norm": 0.0011195311089977622,
-      "learning_rate": 0.0001999986164592347,
-      "loss": 46.0,
-      "step": 21910
-    },
-    {
-      "epoch": 1.6752489630521628,
-      "grad_norm": 0.0035118686500936747,
-      "learning_rate": 0.00019999861633288136,
-      "loss": 46.0,
-      "step": 21911
-    },
-    {
-      "epoch": 1.6753254200355525,
-      "grad_norm": 0.0008547678589820862,
-      "learning_rate": 0.00019999861620652226,
-      "loss": 46.0,
-      "step": 21912
-    },
-    {
-      "epoch": 1.6754018770189423,
-      "grad_norm": 0.001727780792862177,
-      "learning_rate": 0.00019999861608015735,
-      "loss": 46.0,
-      "step": 21913
-    },
-    {
-      "epoch": 1.675478334002332,
-      "grad_norm": 0.0006705062696710229,
-      "learning_rate": 0.0001999986159537867,
-      "loss": 46.0,
-      "step": 21914
-    },
-    {
-      "epoch": 1.6755547909857218,
-      "grad_norm": 0.0005745262023992836,
-      "learning_rate": 0.00019999861582741026,
-      "loss": 46.0,
-      "step": 21915
-    },
-    {
-      "epoch": 1.6756312479691113,
-      "grad_norm": 0.0036399506498128176,
-      "learning_rate": 0.00019999861570102806,
-      "loss": 46.0,
-      "step": 21916
-    },
-    {
-      "epoch": 1.675707704952501,
-      "grad_norm": 0.0003371548082213849,
-      "learning_rate": 0.0001999986155746401,
-      "loss": 46.0,
-      "step": 21917
-    },
-    {
-      "epoch": 1.6757841619358909,
-      "grad_norm": 0.001034217537380755,
-      "learning_rate": 0.00019999861544824638,
-      "loss": 46.0,
-      "step": 21918
-    },
-    {
-      "epoch": 1.6758606189192804,
-      "grad_norm": 0.0004881095082964748,
-      "learning_rate": 0.0001999986153218469,
-      "loss": 46.0,
-      "step": 21919
-    },
-    {
-      "epoch": 1.6759370759026702,
-      "grad_norm": 0.000510350102558732,
-      "learning_rate": 0.0001999986151954416,
-      "loss": 46.0,
-      "step": 21920
-    },
-    {
-      "epoch": 1.67601353288606,
-      "grad_norm": 0.0012808021856471896,
-      "learning_rate": 0.00019999861506903057,
-      "loss": 46.0,
-      "step": 21921
-    },
-    {
-      "epoch": 1.6760899898694497,
-      "grad_norm": 0.017940206453204155,
-      "learning_rate": 0.00019999861494261376,
-      "loss": 46.0,
-      "step": 21922
-    },
-    {
-      "epoch": 1.6761664468528394,
-      "grad_norm": 0.007467540446668863,
-      "learning_rate": 0.00019999861481619118,
-      "loss": 46.0,
-      "step": 21923
-    },
-    {
-      "epoch": 1.6762429038362292,
-      "grad_norm": 0.000876472913660109,
-      "learning_rate": 0.00019999861468976283,
-      "loss": 46.0,
-      "step": 21924
-    },
-    {
-      "epoch": 1.676319360819619,
-      "grad_norm": 0.000634226540569216,
-      "learning_rate": 0.0001999986145633287,
-      "loss": 46.0,
-      "step": 21925
-    },
-    {
-      "epoch": 1.6763958178030087,
-      "grad_norm": 0.0019975327886641026,
-      "learning_rate": 0.00019999861443688882,
-      "loss": 46.0,
-      "step": 21926
-    },
-    {
-      "epoch": 1.6764722747863983,
-      "grad_norm": 0.0007047424442134798,
-      "learning_rate": 0.00019999861431044315,
-      "loss": 46.0,
-      "step": 21927
-    },
-    {
-      "epoch": 1.676548731769788,
-      "grad_norm": 0.0020711985416710377,
-      "learning_rate": 0.00019999861418399173,
-      "loss": 46.0,
-      "step": 21928
-    },
-    {
-      "epoch": 1.6766251887531778,
-      "grad_norm": 0.00022046097728889436,
-      "learning_rate": 0.00019999861405753453,
-      "loss": 46.0,
-      "step": 21929
-    },
-    {
-      "epoch": 1.6767016457365673,
-      "grad_norm": 0.0004052590811625123,
-      "learning_rate": 0.00019999861393107157,
-      "loss": 46.0,
-      "step": 21930
-    },
-    {
-      "epoch": 1.676778102719957,
-      "grad_norm": 0.0007094019092619419,
-      "learning_rate": 0.00019999861380460283,
-      "loss": 46.0,
-      "step": 21931
-    },
-    {
-      "epoch": 1.6768545597033468,
-      "grad_norm": 0.0010485637467354536,
-      "learning_rate": 0.0001999986136781283,
-      "loss": 46.0,
-      "step": 21932
-    },
-    {
-      "epoch": 1.6769310166867366,
-      "grad_norm": 0.0050408439710736275,
-      "learning_rate": 0.00019999861355164805,
-      "loss": 46.0,
-      "step": 21933
-    },
-    {
-      "epoch": 1.6770074736701264,
-      "grad_norm": 0.0013136015040799975,
-      "learning_rate": 0.000199998613425162,
-      "loss": 46.0,
-      "step": 21934
-    },
-    {
-      "epoch": 1.6770839306535161,
-      "grad_norm": 0.0008111601346172392,
-      "learning_rate": 0.0001999986132986702,
-      "loss": 46.0,
-      "step": 21935
-    },
-    {
-      "epoch": 1.677160387636906,
-      "grad_norm": 0.002402213402092457,
-      "learning_rate": 0.0001999986131721726,
-      "loss": 46.0,
-      "step": 21936
-    },
-    {
-      "epoch": 1.6772368446202957,
-      "grad_norm": 0.0009191042045131326,
-      "learning_rate": 0.00019999861304566925,
-      "loss": 46.0,
-      "step": 21937
-    },
-    {
-      "epoch": 1.6773133016036852,
-      "grad_norm": 0.0007864030194468796,
-      "learning_rate": 0.00019999861291916015,
-      "loss": 46.0,
-      "step": 21938
-    },
-    {
-      "epoch": 1.677389758587075,
-      "grad_norm": 0.001963469898328185,
-      "learning_rate": 0.00019999861279264525,
-      "loss": 46.0,
-      "step": 21939
-    },
-    {
-      "epoch": 1.6774662155704647,
-      "grad_norm": 0.001459066872484982,
-      "learning_rate": 0.0001999986126661246,
-      "loss": 46.0,
-      "step": 21940
-    },
-    {
-      "epoch": 1.6775426725538543,
-      "grad_norm": 0.003184190485626459,
-      "learning_rate": 0.00019999861253959815,
-      "loss": 46.0,
-      "step": 21941
-    },
-    {
-      "epoch": 1.677619129537244,
-      "grad_norm": 0.0009452649974264205,
-      "learning_rate": 0.00019999861241306596,
-      "loss": 46.0,
-      "step": 21942
-    },
-    {
-      "epoch": 1.6776955865206338,
-      "grad_norm": 0.0012913054088130593,
-      "learning_rate": 0.000199998612286528,
-      "loss": 46.0,
-      "step": 21943
-    },
-    {
-      "epoch": 1.6777720435040235,
-      "grad_norm": 0.008002624846994877,
-      "learning_rate": 0.00019999861215998425,
-      "loss": 46.0,
-      "step": 21944
-    },
-    {
-      "epoch": 1.6778485004874133,
-      "grad_norm": 0.0013766029151156545,
-      "learning_rate": 0.00019999861203343476,
-      "loss": 46.0,
-      "step": 21945
-    },
-    {
-      "epoch": 1.677924957470803,
-      "grad_norm": 0.0022248669993132353,
-      "learning_rate": 0.00019999861190687947,
-      "loss": 46.0,
-      "step": 21946
-    },
-    {
-      "epoch": 1.6780014144541928,
-      "grad_norm": 0.000477284484077245,
-      "learning_rate": 0.0001999986117803184,
-      "loss": 46.0,
-      "step": 21947
-    },
-    {
-      "epoch": 1.6780778714375826,
-      "grad_norm": 0.00070261798100546,
-      "learning_rate": 0.0001999986116537516,
-      "loss": 46.0,
-      "step": 21948
-    },
-    {
-      "epoch": 1.6781543284209721,
-      "grad_norm": 0.0015849824994802475,
-      "learning_rate": 0.00019999861152717903,
-      "loss": 46.0,
-      "step": 21949
-    },
-    {
-      "epoch": 1.6782307854043619,
-      "grad_norm": 0.0003068811201956123,
-      "learning_rate": 0.00019999861140060067,
-      "loss": 46.0,
-      "step": 21950
-    },
-    {
-      "epoch": 1.6783072423877516,
-      "grad_norm": 0.001097612432204187,
-      "learning_rate": 0.00019999861127401654,
-      "loss": 46.0,
-      "step": 21951
-    },
-    {
-      "epoch": 1.6783836993711412,
-      "grad_norm": 0.00123211566824466,
-      "learning_rate": 0.00019999861114742667,
-      "loss": 46.0,
-      "step": 21952
-    },
-    {
-      "epoch": 1.678460156354531,
-      "grad_norm": 0.0005299783661030233,
-      "learning_rate": 0.000199998611020831,
-      "loss": 46.0,
-      "step": 21953
-    },
-    {
-      "epoch": 1.6785366133379207,
-      "grad_norm": 0.0003358338144607842,
-      "learning_rate": 0.00019999861089422955,
-      "loss": 46.0,
-      "step": 21954
-    },
-    {
-      "epoch": 1.6786130703213105,
-      "grad_norm": 0.0005786007968708873,
-      "learning_rate": 0.00019999861076762236,
-      "loss": 46.0,
-      "step": 21955
-    },
-    {
-      "epoch": 1.6786895273047002,
-      "grad_norm": 0.002668971661478281,
-      "learning_rate": 0.0001999986106410094,
-      "loss": 46.0,
-      "step": 21956
-    },
-    {
-      "epoch": 1.67876598428809,
-      "grad_norm": 0.0004163929261267185,
-      "learning_rate": 0.00019999861051439065,
-      "loss": 46.0,
-      "step": 21957
-    },
-    {
-      "epoch": 1.6788424412714797,
-      "grad_norm": 0.00147600460331887,
-      "learning_rate": 0.00019999861038776616,
-      "loss": 46.0,
-      "step": 21958
-    },
-    {
-      "epoch": 1.6789188982548695,
-      "grad_norm": 0.009283826686441898,
-      "learning_rate": 0.00019999861026113587,
-      "loss": 46.0,
-      "step": 21959
-    },
-    {
-      "epoch": 1.678995355238259,
-      "grad_norm": 0.0038927216082811356,
-      "learning_rate": 0.0001999986101344998,
-      "loss": 46.0,
-      "step": 21960
-    },
-    {
-      "epoch": 1.6790718122216488,
-      "grad_norm": 0.0035042238887399435,
-      "learning_rate": 0.000199998610007858,
-      "loss": 46.0,
-      "step": 21961
-    },
-    {
-      "epoch": 1.6791482692050386,
-      "grad_norm": 0.0008100098930299282,
-      "learning_rate": 0.00019999860988121043,
-      "loss": 46.0,
-      "step": 21962
-    },
-    {
-      "epoch": 1.679224726188428,
-      "grad_norm": 0.0034163829404860735,
-      "learning_rate": 0.00019999860975455707,
-      "loss": 46.0,
-      "step": 21963
-    },
-    {
-      "epoch": 1.6793011831718179,
-      "grad_norm": 0.0012321972753852606,
-      "learning_rate": 0.00019999860962789795,
-      "loss": 46.0,
-      "step": 21964
-    },
-    {
-      "epoch": 1.6793776401552076,
-      "grad_norm": 0.0032783031929284334,
-      "learning_rate": 0.00019999860950123305,
-      "loss": 46.0,
-      "step": 21965
-    },
-    {
-      "epoch": 1.6794540971385974,
-      "grad_norm": 0.005596384406089783,
-      "learning_rate": 0.00019999860937456237,
-      "loss": 46.0,
-      "step": 21966
-    },
-    {
-      "epoch": 1.6795305541219872,
-      "grad_norm": 0.0012995415600016713,
-      "learning_rate": 0.00019999860924788595,
-      "loss": 46.0,
-      "step": 21967
-    },
-    {
-      "epoch": 1.679607011105377,
-      "grad_norm": 0.001801166101358831,
-      "learning_rate": 0.00019999860912120373,
-      "loss": 46.0,
-      "step": 21968
-    },
-    {
-      "epoch": 1.6796834680887667,
-      "grad_norm": 0.007909854874014854,
-      "learning_rate": 0.00019999860899451576,
-      "loss": 46.0,
-      "step": 21969
-    },
-    {
-      "epoch": 1.6797599250721564,
-      "grad_norm": 0.0007382303010672331,
-      "learning_rate": 0.00019999860886782202,
-      "loss": 46.0,
-      "step": 21970
-    },
-    {
-      "epoch": 1.679836382055546,
-      "grad_norm": 0.003825595136731863,
-      "learning_rate": 0.00019999860874112254,
-      "loss": 46.0,
-      "step": 21971
-    },
-    {
-      "epoch": 1.6799128390389357,
-      "grad_norm": 0.0005994772654958069,
-      "learning_rate": 0.00019999860861441725,
-      "loss": 46.0,
-      "step": 21972
-    },
-    {
-      "epoch": 1.6799892960223253,
-      "grad_norm": 0.0005374700995162129,
-      "learning_rate": 0.00019999860848770616,
-      "loss": 46.0,
-      "step": 21973
-    },
-    {
-      "epoch": 1.680065753005715,
-      "grad_norm": 0.0009538496960885823,
-      "learning_rate": 0.00019999860836098936,
-      "loss": 46.0,
-      "step": 21974
-    },
-    {
-      "epoch": 1.6801422099891048,
-      "grad_norm": 0.002697878982871771,
-      "learning_rate": 0.00019999860823426678,
-      "loss": 46.0,
-      "step": 21975
-    },
-    {
-      "epoch": 1.6802186669724946,
-      "grad_norm": 0.002903758082538843,
-      "learning_rate": 0.0001999986081075384,
-      "loss": 46.0,
-      "step": 21976
-    },
-    {
-      "epoch": 1.6802951239558843,
-      "grad_norm": 0.00045755933388136327,
-      "learning_rate": 0.0001999986079808043,
-      "loss": 46.0,
-      "step": 21977
-    },
-    {
-      "epoch": 1.680371580939274,
-      "grad_norm": 0.0014379234053194523,
-      "learning_rate": 0.00019999860785406442,
-      "loss": 46.0,
-      "step": 21978
-    },
-    {
-      "epoch": 1.6804480379226638,
-      "grad_norm": 0.0012884469470009208,
-      "learning_rate": 0.00019999860772731872,
-      "loss": 46.0,
-      "step": 21979
-    },
-    {
-      "epoch": 1.6805244949060536,
-      "grad_norm": 0.0007561820675618947,
-      "learning_rate": 0.0001999986076005673,
-      "loss": 46.0,
-      "step": 21980
-    },
-    {
-      "epoch": 1.6806009518894434,
-      "grad_norm": 0.012626275420188904,
-      "learning_rate": 0.00019999860747381008,
-      "loss": 46.0,
-      "step": 21981
-    },
-    {
-      "epoch": 1.680677408872833,
-      "grad_norm": 0.0007733111851848662,
-      "learning_rate": 0.00019999860734704712,
-      "loss": 46.0,
-      "step": 21982
-    },
-    {
-      "epoch": 1.6807538658562227,
-      "grad_norm": 0.001948415650986135,
-      "learning_rate": 0.00019999860722027838,
-      "loss": 46.0,
-      "step": 21983
-    },
-    {
-      "epoch": 1.6808303228396122,
-      "grad_norm": 0.002721319906413555,
-      "learning_rate": 0.00019999860709350386,
-      "loss": 46.0,
-      "step": 21984
-    },
-    {
-      "epoch": 1.680906779823002,
-      "grad_norm": 0.0008817558991722763,
-      "learning_rate": 0.00019999860696672358,
-      "loss": 46.0,
-      "step": 21985
-    },
-    {
-      "epoch": 1.6809832368063917,
-      "grad_norm": 0.0022498629987239838,
-      "learning_rate": 0.00019999860683993754,
-      "loss": 46.0,
-      "step": 21986
-    },
-    {
-      "epoch": 1.6810596937897815,
-      "grad_norm": 0.0007718690321780741,
-      "learning_rate": 0.0001999986067131457,
-      "loss": 46.0,
-      "step": 21987
-    },
-    {
-      "epoch": 1.6811361507731712,
-      "grad_norm": 0.0011238427832722664,
-      "learning_rate": 0.00019999860658634813,
-      "loss": 46.0,
-      "step": 21988
-    },
-    {
-      "epoch": 1.681212607756561,
-      "grad_norm": 0.0008793903980404139,
-      "learning_rate": 0.00019999860645954475,
-      "loss": 46.0,
-      "step": 21989
-    },
-    {
-      "epoch": 1.6812890647399508,
-      "grad_norm": 0.0007311938679777086,
-      "learning_rate": 0.00019999860633273563,
-      "loss": 46.0,
-      "step": 21990
-    },
-    {
-      "epoch": 1.6813655217233405,
-      "grad_norm": 0.0015717670321464539,
-      "learning_rate": 0.00019999860620592075,
-      "loss": 46.0,
-      "step": 21991
-    },
-    {
-      "epoch": 1.6814419787067303,
-      "grad_norm": 0.0011497556697577238,
-      "learning_rate": 0.00019999860607910005,
-      "loss": 46.0,
-      "step": 21992
-    },
-    {
-      "epoch": 1.6815184356901198,
-      "grad_norm": 0.001142602995969355,
-      "learning_rate": 0.00019999860595227363,
-      "loss": 46.0,
-      "step": 21993
-    },
-    {
-      "epoch": 1.6815948926735096,
-      "grad_norm": 0.0018448163755238056,
-      "learning_rate": 0.00019999860582544144,
-      "loss": 46.0,
-      "step": 21994
-    },
-    {
-      "epoch": 1.6816713496568991,
-      "grad_norm": 0.0006297669024206698,
-      "learning_rate": 0.00019999860569860345,
-      "loss": 46.0,
-      "step": 21995
-    },
-    {
-      "epoch": 1.681747806640289,
-      "grad_norm": 0.0019050653791055083,
-      "learning_rate": 0.0001999986055717597,
-      "loss": 46.0,
-      "step": 21996
-    },
-    {
-      "epoch": 1.6818242636236787,
-      "grad_norm": 0.0074720801785588264,
-      "learning_rate": 0.0001999986054449102,
-      "loss": 46.0,
-      "step": 21997
-    },
-    {
-      "epoch": 1.6819007206070684,
-      "grad_norm": 0.0007136424537748098,
-      "learning_rate": 0.0001999986053180549,
-      "loss": 46.0,
-      "step": 21998
-    },
-    {
-      "epoch": 1.6819771775904582,
-      "grad_norm": 0.0007414708379656076,
-      "learning_rate": 0.00019999860519119388,
-      "loss": 46.0,
-      "step": 21999
-    },
-    {
-      "epoch": 1.682053634573848,
-      "grad_norm": 0.0008026451105251908,
-      "learning_rate": 0.00019999860506432704,
-      "loss": 46.0,
-      "step": 22000
-    },
-    {
-      "epoch": 1.6821300915572377,
-      "grad_norm": 0.003655873006209731,
-      "learning_rate": 0.00019999860493745444,
-      "loss": 46.0,
-      "step": 22001
-    },
-    {
-      "epoch": 1.6822065485406275,
-      "grad_norm": 0.0025371690280735493,
-      "learning_rate": 0.00019999860481057608,
-      "loss": 46.0,
-      "step": 22002
-    },
-    {
-      "epoch": 1.682283005524017,
-      "grad_norm": 0.0003799954429268837,
-      "learning_rate": 0.00019999860468369196,
-      "loss": 46.0,
-      "step": 22003
-    },
-    {
-      "epoch": 1.6823594625074068,
-      "grad_norm": 0.0007769875228404999,
-      "learning_rate": 0.00019999860455680203,
-      "loss": 46.0,
-      "step": 22004
-    },
-    {
-      "epoch": 1.6824359194907965,
-      "grad_norm": 0.0006201844080351293,
-      "learning_rate": 0.00019999860442990639,
-      "loss": 46.0,
-      "step": 22005
-    },
-    {
-      "epoch": 1.682512376474186,
-      "grad_norm": 0.0010021042544394732,
-      "learning_rate": 0.00019999860430300494,
-      "loss": 46.0,
-      "step": 22006
-    },
-    {
-      "epoch": 1.6825888334575758,
-      "grad_norm": 0.0004711336805485189,
-      "learning_rate": 0.00019999860417609772,
-      "loss": 46.0,
-      "step": 22007
-    },
-    {
-      "epoch": 1.6826652904409656,
-      "grad_norm": 0.0008106256718747318,
-      "learning_rate": 0.00019999860404918475,
-      "loss": 46.0,
-      "step": 22008
-    },
-    {
-      "epoch": 1.6827417474243553,
-      "grad_norm": 0.001056780805811286,
-      "learning_rate": 0.00019999860392226602,
-      "loss": 46.0,
-      "step": 22009
-    },
-    {
-      "epoch": 1.682818204407745,
-      "grad_norm": 0.0006996591109782457,
-      "learning_rate": 0.0001999986037953415,
-      "loss": 46.0,
-      "step": 22010
-    },
-    {
-      "epoch": 1.6828946613911349,
-      "grad_norm": 0.0012410633498802781,
-      "learning_rate": 0.00019999860366841122,
-      "loss": 46.0,
-      "step": 22011
-    },
-    {
-      "epoch": 1.6829711183745246,
-      "grad_norm": 0.0012923083268105984,
-      "learning_rate": 0.00019999860354147516,
-      "loss": 46.0,
-      "step": 22012
-    },
-    {
-      "epoch": 1.6830475753579144,
-      "grad_norm": 0.005224398337304592,
-      "learning_rate": 0.00019999860341453332,
-      "loss": 46.0,
-      "step": 22013
-    },
-    {
-      "epoch": 1.683124032341304,
-      "grad_norm": 0.002212190069258213,
-      "learning_rate": 0.00019999860328758575,
-      "loss": 46.0,
-      "step": 22014
-    },
-    {
-      "epoch": 1.6832004893246937,
-      "grad_norm": 0.0008398907375521958,
-      "learning_rate": 0.00019999860316063237,
-      "loss": 46.0,
-      "step": 22015
-    },
-    {
-      "epoch": 1.6832769463080834,
-      "grad_norm": 0.0017157564871013165,
-      "learning_rate": 0.00019999860303367324,
-      "loss": 46.0,
-      "step": 22016
-    },
-    {
-      "epoch": 1.683353403291473,
-      "grad_norm": 0.003142619039863348,
-      "learning_rate": 0.00019999860290670834,
-      "loss": 46.0,
-      "step": 22017
-    },
-    {
-      "epoch": 1.6834298602748627,
-      "grad_norm": 0.002326208632439375,
-      "learning_rate": 0.00019999860277973767,
-      "loss": 46.0,
-      "step": 22018
-    },
-    {
-      "epoch": 1.6835063172582525,
-      "grad_norm": 0.0015836707316339016,
-      "learning_rate": 0.00019999860265276122,
-      "loss": 46.0,
-      "step": 22019
-    },
-    {
-      "epoch": 1.6835827742416423,
-      "grad_norm": 0.003529695561155677,
-      "learning_rate": 0.000199998602525779,
-      "loss": 46.0,
-      "step": 22020
-    },
-    {
-      "epoch": 1.683659231225032,
-      "grad_norm": 0.00117284688167274,
-      "learning_rate": 0.000199998602398791,
-      "loss": 46.0,
-      "step": 22021
-    },
-    {
-      "epoch": 1.6837356882084218,
-      "grad_norm": 0.0024197855964303017,
-      "learning_rate": 0.00019999860227179727,
-      "loss": 46.0,
-      "step": 22022
-    },
-    {
-      "epoch": 1.6838121451918115,
-      "grad_norm": 0.0017137534450739622,
-      "learning_rate": 0.00019999860214479776,
-      "loss": 46.0,
-      "step": 22023
-    },
-    {
-      "epoch": 1.6838886021752013,
-      "grad_norm": 0.004155252128839493,
-      "learning_rate": 0.00019999860201779247,
-      "loss": 46.0,
-      "step": 22024
-    },
-    {
-      "epoch": 1.6839650591585908,
-      "grad_norm": 0.0011570446658879519,
-      "learning_rate": 0.00019999860189078144,
-      "loss": 46.0,
-      "step": 22025
-    },
-    {
-      "epoch": 1.6840415161419806,
-      "grad_norm": 0.0005555233801715076,
-      "learning_rate": 0.00019999860176376458,
-      "loss": 46.0,
-      "step": 22026
-    },
-    {
-      "epoch": 1.6841179731253704,
-      "grad_norm": 0.00028855595155619085,
-      "learning_rate": 0.000199998601636742,
-      "loss": 46.0,
-      "step": 22027
-    },
-    {
-      "epoch": 1.68419443010876,
-      "grad_norm": 0.0018521032761782408,
-      "learning_rate": 0.00019999860150971363,
-      "loss": 46.0,
-      "step": 22028
-    },
-    {
-      "epoch": 1.6842708870921497,
-      "grad_norm": 0.00781367439776659,
-      "learning_rate": 0.00019999860138267953,
-      "loss": 46.0,
-      "step": 22029
-    },
-    {
-      "epoch": 1.6843473440755394,
-      "grad_norm": 0.001226624008268118,
-      "learning_rate": 0.0001999986012556396,
-      "loss": 46.0,
-      "step": 22030
-    },
-    {
-      "epoch": 1.6844238010589292,
-      "grad_norm": 0.0008058040984906256,
-      "learning_rate": 0.00019999860112859393,
-      "loss": 46.0,
-      "step": 22031
-    },
-    {
-      "epoch": 1.684500258042319,
-      "grad_norm": 0.0002205067139584571,
-      "learning_rate": 0.00019999860100154248,
-      "loss": 46.0,
-      "step": 22032
-    },
-    {
-      "epoch": 1.6845767150257087,
-      "grad_norm": 0.0011257849400863051,
-      "learning_rate": 0.00019999860087448526,
-      "loss": 46.0,
-      "step": 22033
-    },
-    {
-      "epoch": 1.6846531720090985,
-      "grad_norm": 0.023186737671494484,
-      "learning_rate": 0.0001999986007474223,
-      "loss": 46.0,
-      "step": 22034
-    },
-    {
-      "epoch": 1.6847296289924882,
-      "grad_norm": 0.0032532988116145134,
-      "learning_rate": 0.00019999860062035353,
-      "loss": 46.0,
-      "step": 22035
-    },
-    {
-      "epoch": 1.6848060859758778,
-      "grad_norm": 0.0022807603236287832,
-      "learning_rate": 0.00019999860049327902,
-      "loss": 46.0,
-      "step": 22036
-    },
-    {
-      "epoch": 1.6848825429592675,
-      "grad_norm": 0.001121692592278123,
-      "learning_rate": 0.00019999860036619874,
-      "loss": 46.0,
-      "step": 22037
-    },
-    {
-      "epoch": 1.6849589999426573,
-      "grad_norm": 0.0007382062030956149,
-      "learning_rate": 0.00019999860023911268,
-      "loss": 46.0,
-      "step": 22038
-    },
-    {
-      "epoch": 1.6850354569260468,
-      "grad_norm": 0.0029226462356746197,
-      "learning_rate": 0.00019999860011202085,
-      "loss": 46.0,
-      "step": 22039
-    },
-    {
-      "epoch": 1.6851119139094366,
-      "grad_norm": 0.004737196955829859,
-      "learning_rate": 0.00019999859998492324,
-      "loss": 46.0,
-      "step": 22040
-    },
-    {
-      "epoch": 1.6851883708928264,
-      "grad_norm": 0.000391368375858292,
-      "learning_rate": 0.0001999985998578199,
-      "loss": 46.0,
-      "step": 22041
-    },
-    {
-      "epoch": 1.6852648278762161,
-      "grad_norm": 0.0011650958331301808,
-      "learning_rate": 0.00019999859973071074,
-      "loss": 46.0,
-      "step": 22042
-    },
-    {
-      "epoch": 1.6853412848596059,
-      "grad_norm": 0.00641118036583066,
-      "learning_rate": 0.00019999859960359584,
-      "loss": 46.0,
-      "step": 22043
-    },
-    {
-      "epoch": 1.6854177418429956,
-      "grad_norm": 0.00103643792681396,
-      "learning_rate": 0.00019999859947647517,
-      "loss": 46.0,
-      "step": 22044
-    },
-    {
-      "epoch": 1.6854941988263854,
-      "grad_norm": 0.00042371300514787436,
-      "learning_rate": 0.00019999859934934872,
-      "loss": 46.0,
-      "step": 22045
-    },
-    {
-      "epoch": 1.6855706558097752,
-      "grad_norm": 0.006192030850797892,
-      "learning_rate": 0.0001999985992222165,
-      "loss": 46.0,
-      "step": 22046
-    },
-    {
-      "epoch": 1.6856471127931647,
-      "grad_norm": 0.004003859590739012,
-      "learning_rate": 0.0001999985990950785,
-      "loss": 46.0,
-      "step": 22047
-    },
-    {
-      "epoch": 1.6857235697765545,
-      "grad_norm": 0.0009548881207592785,
-      "learning_rate": 0.00019999859896793474,
-      "loss": 46.0,
-      "step": 22048
-    },
-    {
-      "epoch": 1.6858000267599442,
-      "grad_norm": 0.0014837206108495593,
-      "learning_rate": 0.00019999859884078523,
-      "loss": 46.0,
-      "step": 22049
-    },
-    {
-      "epoch": 1.6858764837433338,
-      "grad_norm": 0.00445706257596612,
-      "learning_rate": 0.00019999859871362995,
-      "loss": 46.0,
-      "step": 22050
-    },
-    {
-      "epoch": 1.6859529407267235,
-      "grad_norm": 0.0015673998277634382,
-      "learning_rate": 0.0001999985985864689,
-      "loss": 46.0,
-      "step": 22051
-    },
-    {
-      "epoch": 1.6860293977101133,
-      "grad_norm": 0.0016404592897742987,
-      "learning_rate": 0.00019999859845930206,
-      "loss": 46.0,
-      "step": 22052
-    },
-    {
-      "epoch": 1.686105854693503,
-      "grad_norm": 0.00034314801450818777,
-      "learning_rate": 0.00019999859833212948,
-      "loss": 46.0,
-      "step": 22053
-    },
-    {
-      "epoch": 1.6861823116768928,
-      "grad_norm": 0.0005453019984997809,
-      "learning_rate": 0.0001999985982049511,
-      "loss": 46.0,
-      "step": 22054
-    },
-    {
-      "epoch": 1.6862587686602826,
-      "grad_norm": 0.001099586603231728,
-      "learning_rate": 0.00019999859807776695,
-      "loss": 46.0,
-      "step": 22055
-    },
-    {
-      "epoch": 1.6863352256436723,
-      "grad_norm": 0.0011352287838235497,
-      "learning_rate": 0.00019999859795057705,
-      "loss": 46.0,
-      "step": 22056
-    },
-    {
-      "epoch": 1.686411682627062,
-      "grad_norm": 0.0012024266179651022,
-      "learning_rate": 0.00019999859782338138,
-      "loss": 46.0,
-      "step": 22057
-    },
-    {
-      "epoch": 1.6864881396104516,
-      "grad_norm": 0.0006635950412601233,
-      "learning_rate": 0.00019999859769617994,
-      "loss": 46.0,
-      "step": 22058
-    },
-    {
-      "epoch": 1.6865645965938414,
-      "grad_norm": 0.0006319055682979524,
-      "learning_rate": 0.00019999859756897272,
-      "loss": 46.0,
-      "step": 22059
-    },
-    {
-      "epoch": 1.6866410535772312,
-      "grad_norm": 0.0017972607165575027,
-      "learning_rate": 0.00019999859744175973,
-      "loss": 46.0,
-      "step": 22060
-    },
-    {
-      "epoch": 1.6867175105606207,
-      "grad_norm": 0.005334995221346617,
-      "learning_rate": 0.000199998597314541,
-      "loss": 46.0,
-      "step": 22061
-    },
-    {
-      "epoch": 1.6867939675440105,
-      "grad_norm": 0.0014196924166753888,
-      "learning_rate": 0.00019999859718731645,
-      "loss": 46.0,
-      "step": 22062
-    },
-    {
-      "epoch": 1.6868704245274002,
-      "grad_norm": 0.0009492482058703899,
-      "learning_rate": 0.00019999859706008616,
-      "loss": 46.0,
-      "step": 22063
-    },
-    {
-      "epoch": 1.68694688151079,
-      "grad_norm": 0.0010091520380228758,
-      "learning_rate": 0.0001999985969328501,
-      "loss": 46.0,
-      "step": 22064
-    },
-    {
-      "epoch": 1.6870233384941797,
-      "grad_norm": 0.0013195306528359652,
-      "learning_rate": 0.00019999859680560827,
-      "loss": 46.0,
-      "step": 22065
-    },
-    {
-      "epoch": 1.6870997954775695,
-      "grad_norm": 0.0011403732933104038,
-      "learning_rate": 0.00019999859667836067,
-      "loss": 46.0,
-      "step": 22066
-    },
-    {
-      "epoch": 1.6871762524609593,
-      "grad_norm": 0.006182524375617504,
-      "learning_rate": 0.00019999859655110732,
-      "loss": 46.0,
-      "step": 22067
-    },
-    {
-      "epoch": 1.687252709444349,
-      "grad_norm": 0.0009352777851745486,
-      "learning_rate": 0.00019999859642384817,
-      "loss": 46.0,
-      "step": 22068
-    },
-    {
-      "epoch": 1.6873291664277386,
-      "grad_norm": 0.0008126057800836861,
-      "learning_rate": 0.00019999859629658327,
-      "loss": 46.0,
-      "step": 22069
-    },
-    {
-      "epoch": 1.6874056234111283,
-      "grad_norm": 0.002511648228392005,
-      "learning_rate": 0.00019999859616931257,
-      "loss": 46.0,
-      "step": 22070
-    },
-    {
-      "epoch": 1.687482080394518,
-      "grad_norm": 0.002141204196959734,
-      "learning_rate": 0.00019999859604203613,
-      "loss": 46.0,
-      "step": 22071
-    },
-    {
-      "epoch": 1.6875585373779076,
-      "grad_norm": 0.0004095903132110834,
-      "learning_rate": 0.00019999859591475388,
-      "loss": 46.0,
-      "step": 22072
-    },
-    {
-      "epoch": 1.6876349943612974,
-      "grad_norm": 0.0013014693977311254,
-      "learning_rate": 0.00019999859578746592,
-      "loss": 46.0,
-      "step": 22073
-    },
-    {
-      "epoch": 1.6877114513446871,
-      "grad_norm": 0.001256956602446735,
-      "learning_rate": 0.00019999859566017215,
-      "loss": 46.0,
-      "step": 22074
-    },
-    {
-      "epoch": 1.687787908328077,
-      "grad_norm": 0.0008801307994872332,
-      "learning_rate": 0.00019999859553287264,
-      "loss": 46.0,
-      "step": 22075
-    },
-    {
-      "epoch": 1.6878643653114667,
-      "grad_norm": 0.006607298273593187,
-      "learning_rate": 0.00019999859540556733,
-      "loss": 46.0,
-      "step": 22076
-    },
-    {
-      "epoch": 1.6879408222948564,
-      "grad_norm": 0.0016905938973650336,
-      "learning_rate": 0.00019999859527825627,
-      "loss": 46.0,
-      "step": 22077
-    },
-    {
-      "epoch": 1.6880172792782462,
-      "grad_norm": 0.0010199955431744456,
-      "learning_rate": 0.00019999859515093944,
-      "loss": 46.0,
-      "step": 22078
-    },
-    {
-      "epoch": 1.688093736261636,
-      "grad_norm": 0.0016149061266332865,
-      "learning_rate": 0.00019999859502361684,
-      "loss": 46.0,
-      "step": 22079
-    },
-    {
-      "epoch": 1.6881701932450255,
-      "grad_norm": 0.0013768686912953854,
-      "learning_rate": 0.00019999859489628846,
-      "loss": 46.0,
-      "step": 22080
-    },
-    {
-      "epoch": 1.6882466502284152,
-      "grad_norm": 0.002957918681204319,
-      "learning_rate": 0.00019999859476895433,
-      "loss": 46.0,
-      "step": 22081
-    },
-    {
-      "epoch": 1.688323107211805,
-      "grad_norm": 0.004262669011950493,
-      "learning_rate": 0.0001999985946416144,
-      "loss": 46.0,
-      "step": 22082
-    },
-    {
-      "epoch": 1.6883995641951945,
-      "grad_norm": 0.0011622327147051692,
-      "learning_rate": 0.00019999859451426874,
-      "loss": 46.0,
-      "step": 22083
-    },
-    {
-      "epoch": 1.6884760211785843,
-      "grad_norm": 0.0012830978957936168,
-      "learning_rate": 0.0001999985943869173,
-      "loss": 46.0,
-      "step": 22084
-    },
-    {
-      "epoch": 1.688552478161974,
-      "grad_norm": 0.0007717488333582878,
-      "learning_rate": 0.00019999859425956008,
-      "loss": 46.0,
-      "step": 22085
-    },
-    {
-      "epoch": 1.6886289351453638,
-      "grad_norm": 0.0010372751858085394,
-      "learning_rate": 0.00019999859413219709,
-      "loss": 46.0,
-      "step": 22086
-    },
-    {
-      "epoch": 1.6887053921287536,
-      "grad_norm": 0.0010127536952495575,
-      "learning_rate": 0.00019999859400482832,
-      "loss": 46.0,
-      "step": 22087
-    },
-    {
-      "epoch": 1.6887818491121434,
-      "grad_norm": 0.014839183539152145,
-      "learning_rate": 0.00019999859387745378,
-      "loss": 46.0,
-      "step": 22088
-    },
-    {
-      "epoch": 1.6888583060955331,
-      "grad_norm": 0.002051344607025385,
-      "learning_rate": 0.0001999985937500735,
-      "loss": 46.0,
-      "step": 22089
-    },
-    {
-      "epoch": 1.6889347630789229,
-      "grad_norm": 0.0018079476431012154,
-      "learning_rate": 0.00019999859362268742,
-      "loss": 46.0,
-      "step": 22090
-    },
-    {
-      "epoch": 1.6890112200623124,
-      "grad_norm": 0.0011286528315395117,
-      "learning_rate": 0.00019999859349529558,
-      "loss": 46.0,
-      "step": 22091
-    },
-    {
-      "epoch": 1.6890876770457022,
-      "grad_norm": 0.0006381383864209056,
-      "learning_rate": 0.00019999859336789798,
-      "loss": 46.0,
-      "step": 22092
-    },
-    {
-      "epoch": 1.689164134029092,
-      "grad_norm": 0.0007902693469077349,
-      "learning_rate": 0.0001999985932404946,
-      "loss": 46.0,
-      "step": 22093
-    },
-    {
-      "epoch": 1.6892405910124815,
-      "grad_norm": 0.0005956924869678915,
-      "learning_rate": 0.00019999859311308548,
-      "loss": 46.0,
-      "step": 22094
-    },
-    {
-      "epoch": 1.6893170479958712,
-      "grad_norm": 0.00042516615940257907,
-      "learning_rate": 0.00019999859298567055,
-      "loss": 46.0,
-      "step": 22095
-    },
-    {
-      "epoch": 1.689393504979261,
-      "grad_norm": 0.004786976613104343,
-      "learning_rate": 0.00019999859285824986,
-      "loss": 46.0,
-      "step": 22096
-    },
-    {
-      "epoch": 1.6894699619626508,
-      "grad_norm": 0.0013289989437907934,
-      "learning_rate": 0.0001999985927308234,
-      "loss": 46.0,
-      "step": 22097
-    },
-    {
-      "epoch": 1.6895464189460405,
-      "grad_norm": 0.0011059901444241405,
-      "learning_rate": 0.0001999985926033912,
-      "loss": 46.0,
-      "step": 22098
-    },
-    {
-      "epoch": 1.6896228759294303,
-      "grad_norm": 0.0006552828708663583,
-      "learning_rate": 0.00019999859247595323,
-      "loss": 46.0,
-      "step": 22099
-    },
-    {
-      "epoch": 1.68969933291282,
-      "grad_norm": 0.0008212397224269807,
-      "learning_rate": 0.00019999859234850944,
-      "loss": 46.0,
-      "step": 22100
-    },
-    {
-      "epoch": 1.6897757898962098,
-      "grad_norm": 0.003853896399959922,
-      "learning_rate": 0.00019999859222105993,
-      "loss": 46.0,
-      "step": 22101
-    },
-    {
-      "epoch": 1.6898522468795993,
-      "grad_norm": 0.0007604114944115281,
-      "learning_rate": 0.00019999859209360462,
-      "loss": 46.0,
-      "step": 22102
-    },
-    {
-      "epoch": 1.689928703862989,
-      "grad_norm": 0.0011506242444738746,
-      "learning_rate": 0.00019999859196614356,
-      "loss": 46.0,
-      "step": 22103
-    },
-    {
-      "epoch": 1.6900051608463786,
-      "grad_norm": 0.000643672130536288,
-      "learning_rate": 0.00019999859183867673,
-      "loss": 46.0,
-      "step": 22104
-    },
-    {
-      "epoch": 1.6900816178297684,
-      "grad_norm": 0.001015696907415986,
-      "learning_rate": 0.00019999859171120413,
-      "loss": 46.0,
-      "step": 22105
-    },
-    {
-      "epoch": 1.6901580748131582,
-      "grad_norm": 0.0010475862072780728,
-      "learning_rate": 0.00019999859158372572,
-      "loss": 46.0,
-      "step": 22106
-    },
-    {
-      "epoch": 1.690234531796548,
-      "grad_norm": 0.003309564897790551,
-      "learning_rate": 0.00019999859145624157,
-      "loss": 46.0,
-      "step": 22107
-    },
-    {
-      "epoch": 1.6903109887799377,
-      "grad_norm": 0.0010428582318127155,
-      "learning_rate": 0.00019999859132875168,
-      "loss": 46.0,
-      "step": 22108
-    },
-    {
-      "epoch": 1.6903874457633274,
-      "grad_norm": 0.0013270392082631588,
-      "learning_rate": 0.00019999859120125598,
-      "loss": 46.0,
-      "step": 22109
-    },
-    {
-      "epoch": 1.6904639027467172,
-      "grad_norm": 0.0012736058561131358,
-      "learning_rate": 0.00019999859107375454,
-      "loss": 46.0,
-      "step": 22110
-    },
-    {
-      "epoch": 1.690540359730107,
-      "grad_norm": 0.0010910237906500697,
-      "learning_rate": 0.00019999859094624732,
-      "loss": 46.0,
-      "step": 22111
-    },
-    {
-      "epoch": 1.6906168167134967,
-      "grad_norm": 0.00038344794302247465,
-      "learning_rate": 0.00019999859081873433,
-      "loss": 46.0,
-      "step": 22112
-    },
-    {
-      "epoch": 1.6906932736968863,
-      "grad_norm": 0.0010292945662513375,
-      "learning_rate": 0.00019999859069121556,
-      "loss": 46.0,
-      "step": 22113
-    },
-    {
-      "epoch": 1.690769730680276,
-      "grad_norm": 0.0012062484165653586,
-      "learning_rate": 0.00019999859056369103,
-      "loss": 46.0,
-      "step": 22114
-    },
-    {
-      "epoch": 1.6908461876636656,
-      "grad_norm": 0.0010345166083425283,
-      "learning_rate": 0.00019999859043616072,
-      "loss": 46.0,
-      "step": 22115
-    },
-    {
-      "epoch": 1.6909226446470553,
-      "grad_norm": 0.0007773747202008963,
-      "learning_rate": 0.00019999859030862466,
-      "loss": 46.0,
-      "step": 22116
-    },
-    {
-      "epoch": 1.690999101630445,
-      "grad_norm": 0.0009605976520106196,
-      "learning_rate": 0.00019999859018108283,
-      "loss": 46.0,
-      "step": 22117
-    },
-    {
-      "epoch": 1.6910755586138349,
-      "grad_norm": 0.002996504306793213,
-      "learning_rate": 0.00019999859005353523,
-      "loss": 46.0,
-      "step": 22118
-    },
-    {
-      "epoch": 1.6911520155972246,
-      "grad_norm": 0.002304126275703311,
-      "learning_rate": 0.00019999858992598185,
-      "loss": 46.0,
-      "step": 22119
-    },
-    {
-      "epoch": 1.6912284725806144,
-      "grad_norm": 0.0008009443408809602,
-      "learning_rate": 0.00019999858979842267,
-      "loss": 46.0,
-      "step": 22120
-    },
-    {
-      "epoch": 1.6913049295640041,
-      "grad_norm": 0.006163548678159714,
-      "learning_rate": 0.00019999858967085777,
-      "loss": 46.0,
-      "step": 22121
-    },
-    {
-      "epoch": 1.691381386547394,
-      "grad_norm": 0.0012128724483773112,
-      "learning_rate": 0.00019999858954328708,
-      "loss": 46.0,
-      "step": 22122
-    },
-    {
-      "epoch": 1.6914578435307837,
-      "grad_norm": 0.0010021511698141694,
-      "learning_rate": 0.00019999858941571063,
-      "loss": 46.0,
-      "step": 22123
-    },
-    {
-      "epoch": 1.6915343005141732,
-      "grad_norm": 0.0012031501391902566,
-      "learning_rate": 0.0001999985892881284,
-      "loss": 46.0,
-      "step": 22124
-    },
-    {
-      "epoch": 1.691610757497563,
-      "grad_norm": 0.003160058753564954,
-      "learning_rate": 0.00019999858916054043,
-      "loss": 46.0,
-      "step": 22125
-    },
-    {
-      "epoch": 1.6916872144809525,
-      "grad_norm": 0.0007245881133712828,
-      "learning_rate": 0.00019999858903294664,
-      "loss": 46.0,
-      "step": 22126
-    },
-    {
-      "epoch": 1.6917636714643423,
-      "grad_norm": 0.0022353974636644125,
-      "learning_rate": 0.0001999985889053471,
-      "loss": 46.0,
-      "step": 22127
-    },
-    {
-      "epoch": 1.691840128447732,
-      "grad_norm": 0.001276383874937892,
-      "learning_rate": 0.0001999985887777418,
-      "loss": 46.0,
-      "step": 22128
-    },
-    {
-      "epoch": 1.6919165854311218,
-      "grad_norm": 0.0005574924871325493,
-      "learning_rate": 0.00019999858865013076,
-      "loss": 46.0,
-      "step": 22129
-    },
-    {
-      "epoch": 1.6919930424145115,
-      "grad_norm": 0.0007169734453782439,
-      "learning_rate": 0.0001999985885225139,
-      "loss": 46.0,
-      "step": 22130
-    },
-    {
-      "epoch": 1.6920694993979013,
-      "grad_norm": 0.003125410992652178,
-      "learning_rate": 0.0001999985883948913,
-      "loss": 46.0,
-      "step": 22131
-    },
-    {
-      "epoch": 1.692145956381291,
-      "grad_norm": 0.001720004016533494,
-      "learning_rate": 0.0001999985882672629,
-      "loss": 46.0,
-      "step": 22132
-    },
-    {
-      "epoch": 1.6922224133646808,
-      "grad_norm": 0.002073258627206087,
-      "learning_rate": 0.00019999858813962877,
-      "loss": 46.0,
-      "step": 22133
-    },
-    {
-      "epoch": 1.6922988703480706,
-      "grad_norm": 0.0008305757655762136,
-      "learning_rate": 0.00019999858801198882,
-      "loss": 46.0,
-      "step": 22134
-    },
-    {
-      "epoch": 1.6923753273314601,
-      "grad_norm": 0.000664488528855145,
-      "learning_rate": 0.00019999858788434315,
-      "loss": 46.0,
-      "step": 22135
-    },
-    {
-      "epoch": 1.6924517843148499,
-      "grad_norm": 0.0013135633198544383,
-      "learning_rate": 0.0001999985877566917,
-      "loss": 46.0,
-      "step": 22136
-    },
-    {
-      "epoch": 1.6925282412982394,
-      "grad_norm": 0.0017636786215007305,
-      "learning_rate": 0.00019999858762903447,
-      "loss": 46.0,
-      "step": 22137
-    },
-    {
-      "epoch": 1.6926046982816292,
-      "grad_norm": 0.0005673085106536746,
-      "learning_rate": 0.00019999858750137145,
-      "loss": 46.0,
-      "step": 22138
-    },
-    {
-      "epoch": 1.692681155265019,
-      "grad_norm": 0.0012310820166021585,
-      "learning_rate": 0.0001999985873737027,
-      "loss": 46.0,
-      "step": 22139
-    },
-    {
-      "epoch": 1.6927576122484087,
-      "grad_norm": 0.0011759101180359721,
-      "learning_rate": 0.00019999858724602818,
-      "loss": 46.0,
-      "step": 22140
-    },
-    {
-      "epoch": 1.6928340692317985,
-      "grad_norm": 0.000560302403755486,
-      "learning_rate": 0.00019999858711834787,
-      "loss": 46.0,
-      "step": 22141
-    },
-    {
-      "epoch": 1.6929105262151882,
-      "grad_norm": 0.0007434495491907,
-      "learning_rate": 0.0001999985869906618,
-      "loss": 46.0,
-      "step": 22142
-    },
-    {
-      "epoch": 1.692986983198578,
-      "grad_norm": 0.0011841882951557636,
-      "learning_rate": 0.00019999858686296995,
-      "loss": 46.0,
-      "step": 22143
-    },
-    {
-      "epoch": 1.6930634401819677,
-      "grad_norm": 0.00042658360325731337,
-      "learning_rate": 0.00019999858673527235,
-      "loss": 46.0,
-      "step": 22144
-    },
-    {
-      "epoch": 1.6931398971653573,
-      "grad_norm": 0.0007580728852190077,
-      "learning_rate": 0.00019999858660756895,
-      "loss": 46.0,
-      "step": 22145
-    },
-    {
-      "epoch": 1.693216354148747,
-      "grad_norm": 0.0013402904151007533,
-      "learning_rate": 0.0001999985864798598,
-      "loss": 46.0,
-      "step": 22146
-    },
-    {
-      "epoch": 1.6932928111321368,
-      "grad_norm": 0.000532217905856669,
-      "learning_rate": 0.00019999858635214488,
-      "loss": 46.0,
-      "step": 22147
-    },
-    {
-      "epoch": 1.6933692681155263,
-      "grad_norm": 0.0008148018969222903,
-      "learning_rate": 0.0001999985862244242,
-      "loss": 46.0,
-      "step": 22148
-    },
-    {
-      "epoch": 1.693445725098916,
-      "grad_norm": 0.0010160935344174504,
-      "learning_rate": 0.00019999858609669774,
-      "loss": 46.0,
-      "step": 22149
-    },
-    {
-      "epoch": 1.6935221820823059,
-      "grad_norm": 0.0063598970882594585,
-      "learning_rate": 0.00019999858596896552,
-      "loss": 46.0,
-      "step": 22150
-    },
-    {
-      "epoch": 1.6935986390656956,
-      "grad_norm": 0.000679582473821938,
-      "learning_rate": 0.00019999858584122753,
-      "loss": 46.0,
-      "step": 22151
-    },
-    {
-      "epoch": 1.6936750960490854,
-      "grad_norm": 0.003162949113175273,
-      "learning_rate": 0.00019999858571348377,
-      "loss": 46.0,
-      "step": 22152
-    },
-    {
-      "epoch": 1.6937515530324752,
-      "grad_norm": 0.001294243847951293,
-      "learning_rate": 0.0001999985855857342,
-      "loss": 46.0,
-      "step": 22153
-    },
-    {
-      "epoch": 1.693828010015865,
-      "grad_norm": 0.0006573147838935256,
-      "learning_rate": 0.0001999985854579789,
-      "loss": 46.0,
-      "step": 22154
-    },
-    {
-      "epoch": 1.6939044669992547,
-      "grad_norm": 0.0013010881375521421,
-      "learning_rate": 0.00019999858533021784,
-      "loss": 46.0,
-      "step": 22155
-    },
-    {
-      "epoch": 1.6939809239826442,
-      "grad_norm": 0.0011456274660304189,
-      "learning_rate": 0.00019999858520245098,
-      "loss": 46.0,
-      "step": 22156
-    },
-    {
-      "epoch": 1.694057380966034,
-      "grad_norm": 0.002017498714849353,
-      "learning_rate": 0.00019999858507467838,
-      "loss": 46.0,
-      "step": 22157
-    },
-    {
-      "epoch": 1.6941338379494237,
-      "grad_norm": 0.001456520170904696,
-      "learning_rate": 0.00019999858494689998,
-      "loss": 46.0,
-      "step": 22158
-    },
-    {
-      "epoch": 1.6942102949328133,
-      "grad_norm": 0.0016963450470939279,
-      "learning_rate": 0.00019999858481911586,
-      "loss": 46.0,
-      "step": 22159
-    },
-    {
-      "epoch": 1.694286751916203,
-      "grad_norm": 0.0007291499641723931,
-      "learning_rate": 0.00019999858469132593,
-      "loss": 46.0,
-      "step": 22160
-    },
-    {
-      "epoch": 1.6943632088995928,
-      "grad_norm": 0.003190933493897319,
-      "learning_rate": 0.0001999985845635302,
-      "loss": 46.0,
-      "step": 22161
-    },
-    {
-      "epoch": 1.6944396658829826,
-      "grad_norm": 0.002851868513971567,
-      "learning_rate": 0.00019999858443572877,
-      "loss": 46.0,
-      "step": 22162
-    },
-    {
-      "epoch": 1.6945161228663723,
-      "grad_norm": 0.0009559214813634753,
-      "learning_rate": 0.00019999858430792152,
-      "loss": 46.0,
-      "step": 22163
-    },
-    {
-      "epoch": 1.694592579849762,
-      "grad_norm": 0.0032951198518276215,
-      "learning_rate": 0.00019999858418010853,
-      "loss": 46.0,
-      "step": 22164
-    },
-    {
-      "epoch": 1.6946690368331518,
-      "grad_norm": 0.0005475416546687484,
-      "learning_rate": 0.00019999858405228977,
-      "loss": 46.0,
-      "step": 22165
-    },
-    {
-      "epoch": 1.6947454938165416,
-      "grad_norm": 0.00342411152087152,
-      "learning_rate": 0.0001999985839244652,
-      "loss": 46.0,
-      "step": 22166
-    },
-    {
-      "epoch": 1.6948219507999311,
-      "grad_norm": 0.0008787123369984329,
-      "learning_rate": 0.00019999858379663493,
-      "loss": 46.0,
-      "step": 22167
-    },
-    {
-      "epoch": 1.694898407783321,
-      "grad_norm": 0.0025145593099296093,
-      "learning_rate": 0.00019999858366879884,
-      "loss": 46.0,
-      "step": 22168
-    },
-    {
-      "epoch": 1.6949748647667107,
-      "grad_norm": 0.0024322818499058485,
-      "learning_rate": 0.000199998583540957,
-      "loss": 46.0,
-      "step": 22169
-    },
-    {
-      "epoch": 1.6950513217501002,
-      "grad_norm": 0.0006192693253979087,
-      "learning_rate": 0.00019999858341310939,
-      "loss": 46.0,
-      "step": 22170
-    },
-    {
-      "epoch": 1.69512777873349,
-      "grad_norm": 0.0007469177944585681,
-      "learning_rate": 0.000199998583285256,
-      "loss": 46.0,
-      "step": 22171
-    },
-    {
-      "epoch": 1.6952042357168797,
-      "grad_norm": 0.0005055687506683171,
-      "learning_rate": 0.00019999858315739683,
-      "loss": 46.0,
-      "step": 22172
-    },
-    {
-      "epoch": 1.6952806927002695,
-      "grad_norm": 0.0008832396124489605,
-      "learning_rate": 0.0001999985830295319,
-      "loss": 46.0,
-      "step": 22173
-    },
-    {
-      "epoch": 1.6953571496836592,
-      "grad_norm": 0.0012023644521832466,
-      "learning_rate": 0.00019999858290166122,
-      "loss": 46.0,
-      "step": 22174
-    },
-    {
-      "epoch": 1.695433606667049,
-      "grad_norm": 0.0012463612947613,
-      "learning_rate": 0.00019999858277378475,
-      "loss": 46.0,
-      "step": 22175
-    },
-    {
-      "epoch": 1.6955100636504388,
-      "grad_norm": 0.0054426779970526695,
-      "learning_rate": 0.00019999858264590253,
-      "loss": 46.0,
-      "step": 22176
-    },
-    {
-      "epoch": 1.6955865206338285,
-      "grad_norm": 0.000678927346598357,
-      "learning_rate": 0.00019999858251801454,
-      "loss": 46.0,
-      "step": 22177
-    },
-    {
-      "epoch": 1.695662977617218,
-      "grad_norm": 0.0011926769511774182,
-      "learning_rate": 0.00019999858239012075,
-      "loss": 46.0,
-      "step": 22178
-    },
-    {
-      "epoch": 1.6957394346006078,
-      "grad_norm": 0.0005148768541403115,
-      "learning_rate": 0.00019999858226222122,
-      "loss": 46.0,
-      "step": 22179
-    },
-    {
-      "epoch": 1.6958158915839976,
-      "grad_norm": 0.0008881281246431172,
-      "learning_rate": 0.0001999985821343159,
-      "loss": 46.0,
-      "step": 22180
-    },
-    {
-      "epoch": 1.6958923485673871,
-      "grad_norm": 0.0043052202090620995,
-      "learning_rate": 0.00019999858200640483,
-      "loss": 46.0,
-      "step": 22181
-    },
-    {
-      "epoch": 1.695968805550777,
-      "grad_norm": 0.0009644617093726993,
-      "learning_rate": 0.00019999858187848797,
-      "loss": 46.0,
-      "step": 22182
-    },
-    {
-      "epoch": 1.6960452625341667,
-      "grad_norm": 0.00042988162022084,
-      "learning_rate": 0.00019999858175056537,
-      "loss": 46.0,
-      "step": 22183
-    },
-    {
-      "epoch": 1.6961217195175564,
-      "grad_norm": 0.0006817795219831169,
-      "learning_rate": 0.000199998581622637,
-      "loss": 46.0,
-      "step": 22184
-    },
-    {
-      "epoch": 1.6961981765009462,
-      "grad_norm": 0.0010689935879781842,
-      "learning_rate": 0.00019999858149470282,
-      "loss": 46.0,
-      "step": 22185
-    },
-    {
-      "epoch": 1.696274633484336,
-      "grad_norm": 0.0011991215869784355,
-      "learning_rate": 0.0001999985813667629,
-      "loss": 46.0,
-      "step": 22186
-    },
-    {
-      "epoch": 1.6963510904677257,
-      "grad_norm": 0.0006758029339835048,
-      "learning_rate": 0.0001999985812388172,
-      "loss": 46.0,
-      "step": 22187
-    },
-    {
-      "epoch": 1.6964275474511155,
-      "grad_norm": 0.0007255064556375146,
-      "learning_rate": 0.00019999858111086573,
-      "loss": 46.0,
-      "step": 22188
-    },
-    {
-      "epoch": 1.696504004434505,
-      "grad_norm": 0.004217163659632206,
-      "learning_rate": 0.0001999985809829085,
-      "loss": 46.0,
-      "step": 22189
-    },
-    {
-      "epoch": 1.6965804614178948,
-      "grad_norm": 0.0005545195308513939,
-      "learning_rate": 0.0001999985808549455,
-      "loss": 46.0,
-      "step": 22190
-    },
-    {
-      "epoch": 1.6966569184012845,
-      "grad_norm": 0.0020643165335059166,
-      "learning_rate": 0.00019999858072697674,
-      "loss": 46.0,
-      "step": 22191
-    },
-    {
-      "epoch": 1.696733375384674,
-      "grad_norm": 0.0017001588130369782,
-      "learning_rate": 0.00019999858059900215,
-      "loss": 46.0,
-      "step": 22192
-    },
-    {
-      "epoch": 1.6968098323680638,
-      "grad_norm": 0.0016963101224973798,
-      "learning_rate": 0.00019999858047102187,
-      "loss": 46.0,
-      "step": 22193
-    },
-    {
-      "epoch": 1.6968862893514536,
-      "grad_norm": 0.0013595379423350096,
-      "learning_rate": 0.00019999858034303579,
-      "loss": 46.0,
-      "step": 22194
-    },
-    {
-      "epoch": 1.6969627463348433,
-      "grad_norm": 0.004765457473695278,
-      "learning_rate": 0.00019999858021504393,
-      "loss": 46.0,
-      "step": 22195
-    },
-    {
-      "epoch": 1.697039203318233,
-      "grad_norm": 0.004340450279414654,
-      "learning_rate": 0.0001999985800870463,
-      "loss": 46.0,
-      "step": 22196
-    },
-    {
-      "epoch": 1.6971156603016229,
-      "grad_norm": 0.0005983118899166584,
-      "learning_rate": 0.00019999857995904293,
-      "loss": 46.0,
-      "step": 22197
-    },
-    {
-      "epoch": 1.6971921172850126,
-      "grad_norm": 0.0011501519475132227,
-      "learning_rate": 0.00019999857983103375,
-      "loss": 46.0,
-      "step": 22198
-    },
-    {
-      "epoch": 1.6972685742684024,
-      "grad_norm": 0.004356314428150654,
-      "learning_rate": 0.00019999857970301883,
-      "loss": 46.0,
-      "step": 22199
-    },
-    {
-      "epoch": 1.697345031251792,
-      "grad_norm": 0.001399780623614788,
-      "learning_rate": 0.00019999857957499813,
-      "loss": 46.0,
-      "step": 22200
-    },
-    {
-      "epoch": 1.6974214882351817,
-      "grad_norm": 0.0131297642365098,
-      "learning_rate": 0.0001999985794469717,
-      "loss": 46.0,
-      "step": 22201
-    },
-    {
-      "epoch": 1.6974979452185714,
-      "grad_norm": 0.0019582435488700867,
-      "learning_rate": 0.00019999857931893945,
-      "loss": 46.0,
-      "step": 22202
-    },
-    {
-      "epoch": 1.697574402201961,
-      "grad_norm": 0.0003096163272857666,
-      "learning_rate": 0.00019999857919090144,
-      "loss": 46.0,
-      "step": 22203
-    },
-    {
-      "epoch": 1.6976508591853507,
-      "grad_norm": 0.0008582757436670363,
-      "learning_rate": 0.00019999857906285767,
-      "loss": 46.0,
-      "step": 22204
-    },
-    {
-      "epoch": 1.6977273161687405,
-      "grad_norm": 0.0016848635859787464,
-      "learning_rate": 0.0001999985789348081,
-      "loss": 46.0,
-      "step": 22205
-    },
-    {
-      "epoch": 1.6978037731521303,
-      "grad_norm": 0.0004925362882204354,
-      "learning_rate": 0.0001999985788067528,
-      "loss": 46.0,
-      "step": 22206
-    },
-    {
-      "epoch": 1.69788023013552,
-      "grad_norm": 0.0012169420951977372,
-      "learning_rate": 0.00019999857867869172,
-      "loss": 46.0,
-      "step": 22207
-    },
-    {
-      "epoch": 1.6979566871189098,
-      "grad_norm": 0.0005178020219318569,
-      "learning_rate": 0.00019999857855062487,
-      "loss": 46.0,
-      "step": 22208
-    },
-    {
-      "epoch": 1.6980331441022996,
-      "grad_norm": 0.0008111104252748191,
-      "learning_rate": 0.00019999857842255227,
-      "loss": 46.0,
-      "step": 22209
-    },
-    {
-      "epoch": 1.6981096010856893,
-      "grad_norm": 0.002278172643855214,
-      "learning_rate": 0.00019999857829447386,
-      "loss": 46.0,
-      "step": 22210
-    },
-    {
-      "epoch": 1.6981860580690789,
-      "grad_norm": 0.004765786230564117,
-      "learning_rate": 0.0001999985781663897,
-      "loss": 46.0,
-      "step": 22211
-    },
-    {
-      "epoch": 1.6982625150524686,
-      "grad_norm": 0.0017840036889538169,
-      "learning_rate": 0.00019999857803829977,
-      "loss": 46.0,
-      "step": 22212
-    },
-    {
-      "epoch": 1.6983389720358584,
-      "grad_norm": 0.0006655256147496402,
-      "learning_rate": 0.00019999857791020407,
-      "loss": 46.0,
-      "step": 22213
-    },
-    {
-      "epoch": 1.698415429019248,
-      "grad_norm": 0.000860373314935714,
-      "learning_rate": 0.0001999985777821026,
-      "loss": 46.0,
-      "step": 22214
-    },
-    {
-      "epoch": 1.6984918860026377,
-      "grad_norm": 0.0009574061841703951,
-      "learning_rate": 0.00019999857765399536,
-      "loss": 46.0,
-      "step": 22215
-    },
-    {
-      "epoch": 1.6985683429860274,
-      "grad_norm": 0.0011977136600762606,
-      "learning_rate": 0.00019999857752588235,
-      "loss": 46.0,
-      "step": 22216
-    },
-    {
-      "epoch": 1.6986447999694172,
-      "grad_norm": 0.0012669811258092523,
-      "learning_rate": 0.00019999857739776359,
-      "loss": 46.0,
-      "step": 22217
-    },
-    {
-      "epoch": 1.698721256952807,
-      "grad_norm": 0.00323835713788867,
-      "learning_rate": 0.00019999857726963902,
-      "loss": 46.0,
-      "step": 22218
-    },
-    {
-      "epoch": 1.6987977139361967,
-      "grad_norm": 0.0010458544129505754,
-      "learning_rate": 0.00019999857714150872,
-      "loss": 46.0,
-      "step": 22219
-    },
-    {
-      "epoch": 1.6988741709195865,
-      "grad_norm": 0.0008137304685078561,
-      "learning_rate": 0.00019999857701337263,
-      "loss": 46.0,
-      "step": 22220
-    },
-    {
-      "epoch": 1.6989506279029762,
-      "grad_norm": 0.0007417897577397525,
-      "learning_rate": 0.00019999857688523078,
-      "loss": 46.0,
-      "step": 22221
-    },
-    {
-      "epoch": 1.6990270848863658,
-      "grad_norm": 0.0031227373983711004,
-      "learning_rate": 0.00019999857675708315,
-      "loss": 46.0,
-      "step": 22222
-    },
-    {
-      "epoch": 1.6991035418697555,
-      "grad_norm": 0.0012534618144854903,
-      "learning_rate": 0.00019999857662892975,
-      "loss": 46.0,
-      "step": 22223
-    },
-    {
-      "epoch": 1.6991799988531453,
-      "grad_norm": 0.00120448914822191,
-      "learning_rate": 0.0001999985765007706,
-      "loss": 46.0,
-      "step": 22224
-    },
-    {
-      "epoch": 1.6992564558365348,
-      "grad_norm": 0.0006296585779637098,
-      "learning_rate": 0.00019999857637260568,
-      "loss": 46.0,
-      "step": 22225
-    },
-    {
-      "epoch": 1.6993329128199246,
-      "grad_norm": 0.0029547917656600475,
-      "learning_rate": 0.00019999857624443496,
-      "loss": 46.0,
-      "step": 22226
-    },
-    {
-      "epoch": 1.6994093698033144,
-      "grad_norm": 0.0024203951470553875,
-      "learning_rate": 0.0001999985761162585,
-      "loss": 46.0,
-      "step": 22227
-    },
-    {
-      "epoch": 1.6994858267867041,
-      "grad_norm": 0.0005265744403004646,
-      "learning_rate": 0.00019999857598807625,
-      "loss": 46.0,
-      "step": 22228
-    },
-    {
-      "epoch": 1.6995622837700939,
-      "grad_norm": 0.0023372878786176443,
-      "learning_rate": 0.00019999857585988826,
-      "loss": 46.0,
-      "step": 22229
-    },
-    {
-      "epoch": 1.6996387407534836,
-      "grad_norm": 0.0041009425185620785,
-      "learning_rate": 0.00019999857573169447,
-      "loss": 46.0,
-      "step": 22230
-    },
-    {
-      "epoch": 1.6997151977368734,
-      "grad_norm": 0.00406024930998683,
-      "learning_rate": 0.00019999857560349494,
-      "loss": 46.0,
-      "step": 22231
-    },
-    {
-      "epoch": 1.6997916547202632,
-      "grad_norm": 0.0027244891971349716,
-      "learning_rate": 0.0001999985754752896,
-      "loss": 46.0,
-      "step": 22232
-    },
-    {
-      "epoch": 1.6998681117036527,
-      "grad_norm": 0.0015552338445559144,
-      "learning_rate": 0.0001999985753470785,
-      "loss": 46.0,
-      "step": 22233
-    },
-    {
-      "epoch": 1.6999445686870425,
-      "grad_norm": 0.0034095821902155876,
-      "learning_rate": 0.00019999857521886167,
-      "loss": 46.0,
-      "step": 22234
-    },
-    {
-      "epoch": 1.700021025670432,
-      "grad_norm": 0.0010631275363266468,
-      "learning_rate": 0.00019999857509063904,
-      "loss": 46.0,
-      "step": 22235
-    },
-    {
-      "epoch": 1.7000974826538218,
-      "grad_norm": 0.00038365149521268904,
-      "learning_rate": 0.00019999857496241067,
-      "loss": 46.0,
-      "step": 22236
-    },
-    {
-      "epoch": 1.7001739396372115,
-      "grad_norm": 0.0010809325613081455,
-      "learning_rate": 0.00019999857483417647,
-      "loss": 46.0,
-      "step": 22237
-    },
-    {
-      "epoch": 1.7002503966206013,
-      "grad_norm": 0.006074589677155018,
-      "learning_rate": 0.00019999857470593655,
-      "loss": 46.0,
-      "step": 22238
-    },
-    {
-      "epoch": 1.700326853603991,
-      "grad_norm": 0.0013358741998672485,
-      "learning_rate": 0.00019999857457769085,
-      "loss": 46.0,
-      "step": 22239
-    },
-    {
-      "epoch": 1.7004033105873808,
-      "grad_norm": 0.002121866913512349,
-      "learning_rate": 0.00019999857444943938,
-      "loss": 46.0,
-      "step": 22240
-    },
-    {
-      "epoch": 1.7004797675707706,
-      "grad_norm": 0.0013715768000110984,
-      "learning_rate": 0.00019999857432118212,
-      "loss": 46.0,
-      "step": 22241
-    },
-    {
-      "epoch": 1.7005562245541603,
-      "grad_norm": 0.0008254814893007278,
-      "learning_rate": 0.00019999857419291913,
-      "loss": 46.0,
-      "step": 22242
-    },
-    {
-      "epoch": 1.70063268153755,
-      "grad_norm": 0.0009830788476392627,
-      "learning_rate": 0.00019999857406465034,
-      "loss": 46.0,
-      "step": 22243
-    },
-    {
-      "epoch": 1.7007091385209396,
-      "grad_norm": 0.000562444212846458,
-      "learning_rate": 0.0001999985739363758,
-      "loss": 46.0,
-      "step": 22244
-    },
-    {
-      "epoch": 1.7007855955043294,
-      "grad_norm": 0.009231217205524445,
-      "learning_rate": 0.00019999857380809547,
-      "loss": 46.0,
-      "step": 22245
-    },
-    {
-      "epoch": 1.700862052487719,
-      "grad_norm": 0.0022698829416185617,
-      "learning_rate": 0.0001999985736798094,
-      "loss": 46.0,
-      "step": 22246
-    },
-    {
-      "epoch": 1.7009385094711087,
-      "grad_norm": 0.00686991261318326,
-      "learning_rate": 0.0001999985735515175,
-      "loss": 46.0,
-      "step": 22247
-    },
-    {
-      "epoch": 1.7010149664544985,
-      "grad_norm": 0.0010533536551520228,
-      "learning_rate": 0.0001999985734232199,
-      "loss": 46.0,
-      "step": 22248
-    },
-    {
-      "epoch": 1.7010914234378882,
-      "grad_norm": 0.0006744272541254759,
-      "learning_rate": 0.0001999985732949165,
-      "loss": 46.0,
-      "step": 22249
-    },
-    {
-      "epoch": 1.701167880421278,
-      "grad_norm": 0.0031580820214003325,
-      "learning_rate": 0.00019999857316660733,
-      "loss": 46.0,
-      "step": 22250
-    },
-    {
-      "epoch": 1.7012443374046677,
-      "grad_norm": 0.0035408653784543276,
-      "learning_rate": 0.00019999857303829241,
-      "loss": 46.0,
-      "step": 22251
-    },
-    {
-      "epoch": 1.7013207943880575,
-      "grad_norm": 0.0020211224909871817,
-      "learning_rate": 0.0001999985729099717,
-      "loss": 46.0,
-      "step": 22252
-    },
-    {
-      "epoch": 1.7013972513714473,
-      "grad_norm": 0.0011255443096160889,
-      "learning_rate": 0.00019999857278164523,
-      "loss": 46.0,
-      "step": 22253
-    },
-    {
-      "epoch": 1.701473708354837,
-      "grad_norm": 0.001478796824812889,
-      "learning_rate": 0.00019999857265331298,
-      "loss": 46.0,
-      "step": 22254
-    },
-    {
-      "epoch": 1.7015501653382266,
-      "grad_norm": 0.0010935028549283743,
-      "learning_rate": 0.00019999857252497497,
-      "loss": 46.0,
-      "step": 22255
-    },
-    {
-      "epoch": 1.7016266223216163,
-      "grad_norm": 0.000634104828350246,
-      "learning_rate": 0.0001999985723966312,
-      "loss": 46.0,
-      "step": 22256
-    },
-    {
-      "epoch": 1.7017030793050059,
-      "grad_norm": 0.0015172429848462343,
-      "learning_rate": 0.00019999857226828165,
-      "loss": 46.0,
-      "step": 22257
-    },
-    {
-      "epoch": 1.7017795362883956,
-      "grad_norm": 0.002083627739921212,
-      "learning_rate": 0.00019999857213992632,
-      "loss": 46.0,
-      "step": 22258
-    },
-    {
-      "epoch": 1.7018559932717854,
-      "grad_norm": 0.00191004597581923,
-      "learning_rate": 0.00019999857201156524,
-      "loss": 46.0,
-      "step": 22259
-    },
-    {
-      "epoch": 1.7019324502551751,
-      "grad_norm": 0.0006558081950061023,
-      "learning_rate": 0.00019999857188319838,
-      "loss": 46.0,
-      "step": 22260
-    },
-    {
-      "epoch": 1.702008907238565,
-      "grad_norm": 0.0027500188443809748,
-      "learning_rate": 0.00019999857175482575,
-      "loss": 46.0,
-      "step": 22261
-    },
-    {
-      "epoch": 1.7020853642219547,
-      "grad_norm": 0.0010153194889426231,
-      "learning_rate": 0.00019999857162644733,
-      "loss": 46.0,
-      "step": 22262
-    },
-    {
-      "epoch": 1.7021618212053444,
-      "grad_norm": 0.002050751820206642,
-      "learning_rate": 0.00019999857149806318,
-      "loss": 46.0,
-      "step": 22263
-    },
-    {
-      "epoch": 1.7022382781887342,
-      "grad_norm": 0.002942193066701293,
-      "learning_rate": 0.00019999857136967323,
-      "loss": 46.0,
-      "step": 22264
-    },
-    {
-      "epoch": 1.702314735172124,
-      "grad_norm": 0.0015713314060121775,
-      "learning_rate": 0.0001999985712412775,
-      "loss": 46.0,
-      "step": 22265
-    },
-    {
-      "epoch": 1.7023911921555135,
-      "grad_norm": 0.001802604179829359,
-      "learning_rate": 0.00019999857111287604,
-      "loss": 46.0,
-      "step": 22266
-    },
-    {
-      "epoch": 1.7024676491389032,
-      "grad_norm": 0.002200249582529068,
-      "learning_rate": 0.0001999985709844688,
-      "loss": 46.0,
-      "step": 22267
-    },
-    {
-      "epoch": 1.7025441061222928,
-      "grad_norm": 0.0011776078026741743,
-      "learning_rate": 0.0001999985708560558,
-      "loss": 46.0,
-      "step": 22268
-    },
-    {
-      "epoch": 1.7026205631056825,
-      "grad_norm": 0.000524084665812552,
-      "learning_rate": 0.000199998570727637,
-      "loss": 46.0,
-      "step": 22269
-    },
-    {
-      "epoch": 1.7026970200890723,
-      "grad_norm": 0.004391106776893139,
-      "learning_rate": 0.00019999857059921244,
-      "loss": 46.0,
-      "step": 22270
-    },
-    {
-      "epoch": 1.702773477072462,
-      "grad_norm": 0.000983295845799148,
-      "learning_rate": 0.00019999857047078214,
-      "loss": 46.0,
-      "step": 22271
-    },
-    {
-      "epoch": 1.7028499340558518,
-      "grad_norm": 0.0051885670982301235,
-      "learning_rate": 0.00019999857034234606,
-      "loss": 46.0,
-      "step": 22272
-    },
-    {
-      "epoch": 1.7029263910392416,
-      "grad_norm": 0.0005018426454626024,
-      "learning_rate": 0.00019999857021390418,
-      "loss": 46.0,
-      "step": 22273
-    },
-    {
-      "epoch": 1.7030028480226314,
-      "grad_norm": 0.004570841323584318,
-      "learning_rate": 0.00019999857008545658,
-      "loss": 46.0,
-      "step": 22274
-    },
-    {
-      "epoch": 1.7030793050060211,
-      "grad_norm": 0.0004142904363106936,
-      "learning_rate": 0.00019999856995700315,
-      "loss": 46.0,
-      "step": 22275
-    },
-    {
-      "epoch": 1.7031557619894107,
-      "grad_norm": 0.000696543138474226,
-      "learning_rate": 0.000199998569828544,
-      "loss": 46.0,
-      "step": 22276
-    },
-    {
-      "epoch": 1.7032322189728004,
-      "grad_norm": 0.007419273257255554,
-      "learning_rate": 0.00019999856970007905,
-      "loss": 46.0,
-      "step": 22277
-    },
-    {
-      "epoch": 1.7033086759561902,
-      "grad_norm": 0.00047700945287942886,
-      "learning_rate": 0.00019999856957160833,
-      "loss": 46.0,
-      "step": 22278
-    },
-    {
-      "epoch": 1.7033851329395797,
-      "grad_norm": 0.002744599711149931,
-      "learning_rate": 0.00019999856944313187,
-      "loss": 46.0,
-      "step": 22279
-    },
-    {
-      "epoch": 1.7034615899229695,
-      "grad_norm": 0.0004348990914877504,
-      "learning_rate": 0.00019999856931464963,
-      "loss": 46.0,
-      "step": 22280
-    },
-    {
-      "epoch": 1.7035380469063592,
-      "grad_norm": 0.0020320408511906862,
-      "learning_rate": 0.0001999985691861616,
-      "loss": 46.0,
-      "step": 22281
-    },
-    {
-      "epoch": 1.703614503889749,
-      "grad_norm": 0.00048128562048077583,
-      "learning_rate": 0.00019999856905766783,
-      "loss": 46.0,
-      "step": 22282
-    },
-    {
-      "epoch": 1.7036909608731388,
-      "grad_norm": 0.00269178650341928,
-      "learning_rate": 0.00019999856892916827,
-      "loss": 46.0,
-      "step": 22283
-    },
-    {
-      "epoch": 1.7037674178565285,
-      "grad_norm": 0.0007060584030114114,
-      "learning_rate": 0.00019999856880066293,
-      "loss": 46.0,
-      "step": 22284
-    },
-    {
-      "epoch": 1.7038438748399183,
-      "grad_norm": 0.0010880966437980533,
-      "learning_rate": 0.00019999856867215185,
-      "loss": 46.0,
-      "step": 22285
-    },
-    {
-      "epoch": 1.703920331823308,
-      "grad_norm": 0.0014449474401772022,
-      "learning_rate": 0.00019999856854363497,
-      "loss": 46.0,
-      "step": 22286
-    },
-    {
-      "epoch": 1.7039967888066976,
-      "grad_norm": 0.0018279115902259946,
-      "learning_rate": 0.00019999856841511235,
-      "loss": 46.0,
-      "step": 22287
-    },
-    {
-      "epoch": 1.7040732457900873,
-      "grad_norm": 0.0013416545698419213,
-      "learning_rate": 0.00019999856828658395,
-      "loss": 46.0,
-      "step": 22288
-    },
-    {
-      "epoch": 1.704149702773477,
-      "grad_norm": 0.0017435051267966628,
-      "learning_rate": 0.00019999856815804975,
-      "loss": 46.0,
-      "step": 22289
-    },
-    {
-      "epoch": 1.7042261597568666,
-      "grad_norm": 0.0012131447438150644,
-      "learning_rate": 0.00019999856802950983,
-      "loss": 46.0,
-      "step": 22290
-    },
-    {
-      "epoch": 1.7043026167402564,
-      "grad_norm": 0.002309046685695648,
-      "learning_rate": 0.0001999985679009641,
-      "loss": 46.0,
-      "step": 22291
-    },
-    {
-      "epoch": 1.7043790737236462,
-      "grad_norm": 0.002016992075368762,
-      "learning_rate": 0.00019999856777241264,
-      "loss": 46.0,
-      "step": 22292
-    },
-    {
-      "epoch": 1.704455530707036,
-      "grad_norm": 0.000857964507304132,
-      "learning_rate": 0.00019999856764385537,
-      "loss": 46.0,
-      "step": 22293
-    },
-    {
-      "epoch": 1.7045319876904257,
-      "grad_norm": 0.001404703943990171,
-      "learning_rate": 0.0001999985675152924,
-      "loss": 46.0,
-      "step": 22294
-    },
-    {
-      "epoch": 1.7046084446738154,
-      "grad_norm": 0.002127214567735791,
-      "learning_rate": 0.0001999985673867236,
-      "loss": 46.0,
-      "step": 22295
-    },
-    {
-      "epoch": 1.7046849016572052,
-      "grad_norm": 0.0007519699865952134,
-      "learning_rate": 0.00019999856725814901,
-      "loss": 46.0,
-      "step": 22296
-    },
-    {
-      "epoch": 1.704761358640595,
-      "grad_norm": 0.0020818302873522043,
-      "learning_rate": 0.0001999985671295687,
-      "loss": 46.0,
-      "step": 22297
-    },
-    {
-      "epoch": 1.7048378156239845,
-      "grad_norm": 0.0007663040305487812,
-      "learning_rate": 0.0001999985670009826,
-      "loss": 46.0,
-      "step": 22298
-    },
-    {
-      "epoch": 1.7049142726073743,
-      "grad_norm": 0.0031487327069044113,
-      "learning_rate": 0.00019999856687239075,
-      "loss": 46.0,
-      "step": 22299
-    },
-    {
-      "epoch": 1.704990729590764,
-      "grad_norm": 0.0017989104380831122,
-      "learning_rate": 0.00019999856674379312,
-      "loss": 46.0,
-      "step": 22300
-    },
-    {
-      "epoch": 1.7050671865741536,
-      "grad_norm": 0.0006492551183328032,
-      "learning_rate": 0.0001999985666151897,
-      "loss": 46.0,
-      "step": 22301
-    },
-    {
-      "epoch": 1.7051436435575433,
-      "grad_norm": 0.0030262882355600595,
-      "learning_rate": 0.00019999856648658052,
-      "loss": 46.0,
-      "step": 22302
-    },
-    {
-      "epoch": 1.705220100540933,
-      "grad_norm": 0.0006862126756459475,
-      "learning_rate": 0.00019999856635796558,
-      "loss": 46.0,
-      "step": 22303
-    },
-    {
-      "epoch": 1.7052965575243229,
-      "grad_norm": 0.00170803046785295,
-      "learning_rate": 0.00019999856622934488,
-      "loss": 46.0,
-      "step": 22304
-    },
-    {
-      "epoch": 1.7053730145077126,
-      "grad_norm": 0.0024671442806720734,
-      "learning_rate": 0.0001999985661007184,
-      "loss": 46.0,
-      "step": 22305
-    },
-    {
-      "epoch": 1.7054494714911024,
-      "grad_norm": 0.0012373417848721147,
-      "learning_rate": 0.00019999856597208615,
-      "loss": 46.0,
-      "step": 22306
-    },
-    {
-      "epoch": 1.7055259284744921,
-      "grad_norm": 0.000414506095694378,
-      "learning_rate": 0.00019999856584344814,
-      "loss": 46.0,
-      "step": 22307
-    },
-    {
-      "epoch": 1.705602385457882,
-      "grad_norm": 0.0029115143697708845,
-      "learning_rate": 0.00019999856571480433,
-      "loss": 46.0,
-      "step": 22308
-    },
-    {
-      "epoch": 1.7056788424412714,
-      "grad_norm": 0.0013761200243607163,
-      "learning_rate": 0.0001999985655861548,
-      "loss": 46.0,
-      "step": 22309
-    },
-    {
-      "epoch": 1.7057552994246612,
-      "grad_norm": 0.0019353954121470451,
-      "learning_rate": 0.00019999856545749943,
-      "loss": 46.0,
-      "step": 22310
-    },
-    {
-      "epoch": 1.705831756408051,
-      "grad_norm": 0.0014452057657763362,
-      "learning_rate": 0.00019999856532883835,
-      "loss": 46.0,
-      "step": 22311
-    },
-    {
-      "epoch": 1.7059082133914405,
-      "grad_norm": 0.0011837605852633715,
-      "learning_rate": 0.0001999985652001715,
-      "loss": 46.0,
-      "step": 22312
-    },
-    {
-      "epoch": 1.7059846703748303,
-      "grad_norm": 0.00034283779677934945,
-      "learning_rate": 0.00019999856507149885,
-      "loss": 46.0,
-      "step": 22313
-    },
-    {
-      "epoch": 1.70606112735822,
-      "grad_norm": 0.002047367161139846,
-      "learning_rate": 0.00019999856494282042,
-      "loss": 46.0,
-      "step": 22314
-    },
-    {
-      "epoch": 1.7061375843416098,
-      "grad_norm": 0.005829261615872383,
-      "learning_rate": 0.00019999856481413625,
-      "loss": 46.0,
-      "step": 22315
-    },
-    {
-      "epoch": 1.7062140413249995,
-      "grad_norm": 0.006298776715993881,
-      "learning_rate": 0.00019999856468544633,
-      "loss": 46.0,
-      "step": 22316
-    },
-    {
-      "epoch": 1.7062904983083893,
-      "grad_norm": 0.0007811666582711041,
-      "learning_rate": 0.0001999985645567506,
-      "loss": 46.0,
-      "step": 22317
-    },
-    {
-      "epoch": 1.706366955291779,
-      "grad_norm": 0.0005377213819883764,
-      "learning_rate": 0.00019999856442804914,
-      "loss": 46.0,
-      "step": 22318
-    },
-    {
-      "epoch": 1.7064434122751688,
-      "grad_norm": 0.0057342019863426685,
-      "learning_rate": 0.00019999856429934188,
-      "loss": 46.0,
-      "step": 22319
-    },
-    {
-      "epoch": 1.7065198692585584,
-      "grad_norm": 0.0014235972194001079,
-      "learning_rate": 0.00019999856417062887,
-      "loss": 46.0,
-      "step": 22320
-    },
-    {
-      "epoch": 1.7065963262419481,
-      "grad_norm": 0.00045009213499724865,
-      "learning_rate": 0.00019999856404191005,
-      "loss": 46.0,
-      "step": 22321
-    },
-    {
-      "epoch": 1.7066727832253379,
-      "grad_norm": 0.003260469762608409,
-      "learning_rate": 0.00019999856391318552,
-      "loss": 46.0,
-      "step": 22322
-    },
-    {
-      "epoch": 1.7067492402087274,
-      "grad_norm": 0.0010507296537980437,
-      "learning_rate": 0.00019999856378445516,
-      "loss": 46.0,
-      "step": 22323
-    },
-    {
-      "epoch": 1.7068256971921172,
-      "grad_norm": 0.00036351699964143336,
-      "learning_rate": 0.00019999856365571908,
-      "loss": 46.0,
-      "step": 22324
-    },
-    {
-      "epoch": 1.706902154175507,
-      "grad_norm": 0.005515695083886385,
-      "learning_rate": 0.0001999985635269772,
-      "loss": 46.0,
-      "step": 22325
-    },
-    {
-      "epoch": 1.7069786111588967,
-      "grad_norm": 0.0007527195266447961,
-      "learning_rate": 0.00019999856339822958,
-      "loss": 46.0,
-      "step": 22326
-    },
-    {
-      "epoch": 1.7070550681422865,
-      "grad_norm": 0.0006638483610004187,
-      "learning_rate": 0.00019999856326947618,
-      "loss": 46.0,
-      "step": 22327
-    },
-    {
-      "epoch": 1.7071315251256762,
-      "grad_norm": 0.003606505459174514,
-      "learning_rate": 0.00019999856314071698,
-      "loss": 46.0,
-      "step": 22328
-    },
-    {
-      "epoch": 1.707207982109066,
-      "grad_norm": 0.0006367659079842269,
-      "learning_rate": 0.00019999856301195203,
-      "loss": 46.0,
-      "step": 22329
-    },
-    {
-      "epoch": 1.7072844390924558,
-      "grad_norm": 0.001677622552961111,
-      "learning_rate": 0.00019999856288318134,
-      "loss": 46.0,
-      "step": 22330
-    },
-    {
-      "epoch": 1.7073608960758453,
-      "grad_norm": 0.0004196741501800716,
-      "learning_rate": 0.00019999856275440485,
-      "loss": 46.0,
-      "step": 22331
-    },
-    {
-      "epoch": 1.707437353059235,
-      "grad_norm": 0.0005129675264470279,
-      "learning_rate": 0.00019999856262562258,
-      "loss": 46.0,
-      "step": 22332
-    },
-    {
-      "epoch": 1.7075138100426248,
-      "grad_norm": 0.0010110505390912294,
-      "learning_rate": 0.00019999856249683457,
-      "loss": 46.0,
-      "step": 22333
-    },
-    {
-      "epoch": 1.7075902670260144,
-      "grad_norm": 0.004729064647108316,
-      "learning_rate": 0.00019999856236804079,
-      "loss": 46.0,
-      "step": 22334
-    },
-    {
-      "epoch": 1.7076667240094041,
-      "grad_norm": 0.00293494644574821,
-      "learning_rate": 0.0001999985622392412,
-      "loss": 46.0,
-      "step": 22335
-    },
-    {
-      "epoch": 1.7077431809927939,
-      "grad_norm": 0.012141870334744453,
-      "learning_rate": 0.0001999985621104359,
-      "loss": 46.0,
-      "step": 22336
-    },
-    {
-      "epoch": 1.7078196379761836,
-      "grad_norm": 0.0005728612304665148,
-      "learning_rate": 0.0001999985619816248,
-      "loss": 46.0,
-      "step": 22337
-    },
-    {
-      "epoch": 1.7078960949595734,
-      "grad_norm": 0.002871981356292963,
-      "learning_rate": 0.0001999985618528079,
-      "loss": 46.0,
-      "step": 22338
-    },
-    {
-      "epoch": 1.7079725519429632,
-      "grad_norm": 0.0005553775699809194,
-      "learning_rate": 0.00019999856172398528,
-      "loss": 46.0,
-      "step": 22339
-    },
-    {
-      "epoch": 1.708049008926353,
-      "grad_norm": 0.0020619346760213375,
-      "learning_rate": 0.00019999856159515686,
-      "loss": 46.0,
-      "step": 22340
-    },
-    {
-      "epoch": 1.7081254659097427,
-      "grad_norm": 0.0006550454418174922,
-      "learning_rate": 0.00019999856146632269,
-      "loss": 46.0,
-      "step": 22341
-    },
-    {
-      "epoch": 1.7082019228931322,
-      "grad_norm": 0.0030603879131376743,
-      "learning_rate": 0.00019999856133748274,
-      "loss": 46.0,
-      "step": 22342
-    },
-    {
-      "epoch": 1.708278379876522,
-      "grad_norm": 0.000612535048276186,
-      "learning_rate": 0.00019999856120863702,
-      "loss": 46.0,
-      "step": 22343
-    },
-    {
-      "epoch": 1.7083548368599117,
-      "grad_norm": 0.0009035649127326906,
-      "learning_rate": 0.00019999856107978553,
-      "loss": 46.0,
-      "step": 22344
-    },
-    {
-      "epoch": 1.7084312938433013,
-      "grad_norm": 0.0008447629516012967,
-      "learning_rate": 0.0001999985609509283,
-      "loss": 46.0,
-      "step": 22345
-    },
-    {
-      "epoch": 1.708507750826691,
-      "grad_norm": 0.0009394664084538817,
-      "learning_rate": 0.00019999856082206525,
-      "loss": 46.0,
-      "step": 22346
-    },
-    {
-      "epoch": 1.7085842078100808,
-      "grad_norm": 0.0056218672543764114,
-      "learning_rate": 0.00019999856069319644,
-      "loss": 46.0,
-      "step": 22347
-    },
-    {
-      "epoch": 1.7086606647934706,
-      "grad_norm": 0.0030335215851664543,
-      "learning_rate": 0.00019999856056432188,
-      "loss": 46.0,
-      "step": 22348
-    },
-    {
-      "epoch": 1.7087371217768603,
-      "grad_norm": 0.0012526758946478367,
-      "learning_rate": 0.00019999856043544155,
-      "loss": 46.0,
-      "step": 22349
-    },
-    {
-      "epoch": 1.70881357876025,
-      "grad_norm": 0.002431680681183934,
-      "learning_rate": 0.00019999856030655547,
-      "loss": 46.0,
-      "step": 22350
-    },
-    {
-      "epoch": 1.7088900357436398,
-      "grad_norm": 0.0017076322110369802,
-      "learning_rate": 0.0001999985601776636,
-      "loss": 46.0,
-      "step": 22351
-    },
-    {
-      "epoch": 1.7089664927270296,
-      "grad_norm": 0.0008788536651991308,
-      "learning_rate": 0.00019999856004876594,
-      "loss": 46.0,
-      "step": 22352
-    },
-    {
-      "epoch": 1.7090429497104191,
-      "grad_norm": 0.001363835297524929,
-      "learning_rate": 0.00019999855991986251,
-      "loss": 46.0,
-      "step": 22353
-    },
-    {
-      "epoch": 1.709119406693809,
-      "grad_norm": 0.003085857490077615,
-      "learning_rate": 0.00019999855979095334,
-      "loss": 46.0,
-      "step": 22354
-    },
-    {
-      "epoch": 1.7091958636771987,
-      "grad_norm": 0.0008214113768190145,
-      "learning_rate": 0.0001999985596620384,
-      "loss": 46.0,
-      "step": 22355
-    },
-    {
-      "epoch": 1.7092723206605882,
-      "grad_norm": 0.0009509060764685273,
-      "learning_rate": 0.00019999855953311768,
-      "loss": 46.0,
-      "step": 22356
-    },
-    {
-      "epoch": 1.709348777643978,
-      "grad_norm": 0.003197141457349062,
-      "learning_rate": 0.0001999985594041912,
-      "loss": 46.0,
-      "step": 22357
-    },
-    {
-      "epoch": 1.7094252346273677,
-      "grad_norm": 0.002035139361396432,
-      "learning_rate": 0.00019999855927525895,
-      "loss": 46.0,
-      "step": 22358
-    },
-    {
-      "epoch": 1.7095016916107575,
-      "grad_norm": 0.0012600065674632788,
-      "learning_rate": 0.0001999985591463209,
-      "loss": 46.0,
-      "step": 22359
-    },
-    {
-      "epoch": 1.7095781485941473,
-      "grad_norm": 0.0006923387991264462,
-      "learning_rate": 0.0001999985590173771,
-      "loss": 46.0,
-      "step": 22360
-    },
-    {
-      "epoch": 1.709654605577537,
-      "grad_norm": 0.0006955589633435011,
-      "learning_rate": 0.00019999855888842754,
-      "loss": 46.0,
-      "step": 22361
-    },
-    {
-      "epoch": 1.7097310625609268,
-      "grad_norm": 0.0016970790456980467,
-      "learning_rate": 0.0001999985587594722,
-      "loss": 46.0,
-      "step": 22362
-    },
-    {
-      "epoch": 1.7098075195443165,
-      "grad_norm": 0.018865732476115227,
-      "learning_rate": 0.0001999985586305111,
-      "loss": 46.0,
-      "step": 22363
-    },
-    {
-      "epoch": 1.709883976527706,
-      "grad_norm": 0.0024612094275653362,
-      "learning_rate": 0.00019999855850154423,
-      "loss": 46.0,
-      "step": 22364
-    },
-    {
-      "epoch": 1.7099604335110958,
-      "grad_norm": 0.0006160700577311218,
-      "learning_rate": 0.0001999985583725716,
-      "loss": 46.0,
-      "step": 22365
-    },
-    {
-      "epoch": 1.7100368904944856,
-      "grad_norm": 0.02170710824429989,
-      "learning_rate": 0.00019999855824359318,
-      "loss": 46.0,
-      "step": 22366
-    },
-    {
-      "epoch": 1.7101133474778751,
-      "grad_norm": 0.0005958536057732999,
-      "learning_rate": 0.000199998558114609,
-      "loss": 46.0,
-      "step": 22367
-    },
-    {
-      "epoch": 1.710189804461265,
-      "grad_norm": 0.0005198173457756639,
-      "learning_rate": 0.00019999855798561903,
-      "loss": 46.0,
-      "step": 22368
-    },
-    {
-      "epoch": 1.7102662614446547,
-      "grad_norm": 0.0020933523774147034,
-      "learning_rate": 0.00019999855785662331,
-      "loss": 46.0,
-      "step": 22369
-    },
-    {
-      "epoch": 1.7103427184280444,
-      "grad_norm": 0.0025250727776437998,
-      "learning_rate": 0.00019999855772762182,
-      "loss": 46.0,
-      "step": 22370
-    },
-    {
-      "epoch": 1.7104191754114342,
-      "grad_norm": 0.0009470495278947055,
-      "learning_rate": 0.00019999855759861456,
-      "loss": 46.0,
-      "step": 22371
-    },
-    {
-      "epoch": 1.710495632394824,
-      "grad_norm": 0.0021975236013531685,
-      "learning_rate": 0.00019999855746960155,
-      "loss": 46.0,
-      "step": 22372
-    },
-    {
-      "epoch": 1.7105720893782137,
-      "grad_norm": 0.00020594771194737405,
-      "learning_rate": 0.00019999855734058276,
-      "loss": 46.0,
-      "step": 22373
-    },
-    {
-      "epoch": 1.7106485463616035,
-      "grad_norm": 0.0026221165899187326,
-      "learning_rate": 0.00019999855721155818,
-      "loss": 46.0,
-      "step": 22374
-    },
-    {
-      "epoch": 1.710725003344993,
-      "grad_norm": 0.000708280480466783,
-      "learning_rate": 0.00019999855708252785,
-      "loss": 46.0,
-      "step": 22375
-    },
-    {
-      "epoch": 1.7108014603283828,
-      "grad_norm": 0.0017831235891208053,
-      "learning_rate": 0.0001999985569534917,
-      "loss": 46.0,
-      "step": 22376
-    },
-    {
-      "epoch": 1.7108779173117723,
-      "grad_norm": 0.002299598418176174,
-      "learning_rate": 0.00019999855682444986,
-      "loss": 46.0,
-      "step": 22377
-    },
-    {
-      "epoch": 1.710954374295162,
-      "grad_norm": 0.008643848821520805,
-      "learning_rate": 0.0001999985566954022,
-      "loss": 46.0,
-      "step": 22378
-    },
-    {
-      "epoch": 1.7110308312785518,
-      "grad_norm": 0.0027712073642760515,
-      "learning_rate": 0.00019999855656634879,
-      "loss": 46.0,
-      "step": 22379
-    },
-    {
-      "epoch": 1.7111072882619416,
-      "grad_norm": 0.0006109913811087608,
-      "learning_rate": 0.00019999855643728961,
-      "loss": 46.0,
-      "step": 22380
-    },
-    {
-      "epoch": 1.7111837452453313,
-      "grad_norm": 0.0009963305201381445,
-      "learning_rate": 0.00019999855630822467,
-      "loss": 46.0,
-      "step": 22381
-    },
-    {
-      "epoch": 1.711260202228721,
-      "grad_norm": 0.0012679974315688014,
-      "learning_rate": 0.00019999855617915393,
-      "loss": 46.0,
-      "step": 22382
-    },
-    {
-      "epoch": 1.7113366592121109,
-      "grad_norm": 0.001112680067308247,
-      "learning_rate": 0.00019999855605007743,
-      "loss": 46.0,
-      "step": 22383
-    },
-    {
-      "epoch": 1.7114131161955006,
-      "grad_norm": 0.0006381681305356324,
-      "learning_rate": 0.0001999985559209952,
-      "loss": 46.0,
-      "step": 22384
-    },
-    {
-      "epoch": 1.7114895731788904,
-      "grad_norm": 0.0017438214272260666,
-      "learning_rate": 0.00019999855579190716,
-      "loss": 46.0,
-      "step": 22385
-    },
-    {
-      "epoch": 1.71156603016228,
-      "grad_norm": 0.0004668562032748014,
-      "learning_rate": 0.00019999855566281335,
-      "loss": 46.0,
-      "step": 22386
-    },
-    {
-      "epoch": 1.7116424871456697,
-      "grad_norm": 0.0012550413375720382,
-      "learning_rate": 0.0001999985555337138,
-      "loss": 46.0,
-      "step": 22387
-    },
-    {
-      "epoch": 1.7117189441290592,
-      "grad_norm": 0.001102479174733162,
-      "learning_rate": 0.00019999855540460843,
-      "loss": 46.0,
-      "step": 22388
-    },
-    {
-      "epoch": 1.711795401112449,
-      "grad_norm": 0.0008673723787069321,
-      "learning_rate": 0.00019999855527549733,
-      "loss": 46.0,
-      "step": 22389
-    },
-    {
-      "epoch": 1.7118718580958387,
-      "grad_norm": 0.001250889734365046,
-      "learning_rate": 0.00019999855514638045,
-      "loss": 46.0,
-      "step": 22390
-    },
-    {
-      "epoch": 1.7119483150792285,
-      "grad_norm": 0.003054277505725622,
-      "learning_rate": 0.0001999985550172578,
-      "loss": 46.0,
-      "step": 22391
-    },
-    {
-      "epoch": 1.7120247720626183,
-      "grad_norm": 0.0006690003210678697,
-      "learning_rate": 0.0001999985548881294,
-      "loss": 46.0,
-      "step": 22392
-    },
-    {
-      "epoch": 1.712101229046008,
-      "grad_norm": 0.0006188425468280911,
-      "learning_rate": 0.0001999985547589952,
-      "loss": 46.0,
-      "step": 22393
-    },
-    {
-      "epoch": 1.7121776860293978,
-      "grad_norm": 0.00827113725244999,
-      "learning_rate": 0.00019999855462985523,
-      "loss": 46.0,
-      "step": 22394
-    },
-    {
-      "epoch": 1.7122541430127876,
-      "grad_norm": 0.008701412007212639,
-      "learning_rate": 0.00019999855450070951,
-      "loss": 46.0,
-      "step": 22395
-    },
-    {
-      "epoch": 1.7123305999961773,
-      "grad_norm": 0.0007379450835287571,
-      "learning_rate": 0.00019999855437155802,
-      "loss": 46.0,
-      "step": 22396
-    },
-    {
-      "epoch": 1.7124070569795669,
-      "grad_norm": 0.0005624516634270549,
-      "learning_rate": 0.00019999855424240076,
-      "loss": 46.0,
-      "step": 22397
-    },
-    {
-      "epoch": 1.7124835139629566,
-      "grad_norm": 0.0019218729576095939,
-      "learning_rate": 0.00019999855411323775,
-      "loss": 46.0,
-      "step": 22398
-    },
-    {
-      "epoch": 1.7125599709463462,
-      "grad_norm": 0.001076512853614986,
-      "learning_rate": 0.00019999855398406894,
-      "loss": 46.0,
-      "step": 22399
-    },
-    {
-      "epoch": 1.712636427929736,
-      "grad_norm": 0.0014896945795044303,
-      "learning_rate": 0.00019999855385489435,
-      "loss": 46.0,
-      "step": 22400
-    },
-    {
-      "epoch": 1.7127128849131257,
-      "grad_norm": 0.004581732675433159,
-      "learning_rate": 0.00019999855372571402,
-      "loss": 46.0,
-      "step": 22401
-    },
-    {
-      "epoch": 1.7127893418965154,
-      "grad_norm": 0.0032729562371969223,
-      "learning_rate": 0.00019999855359652792,
-      "loss": 46.0,
-      "step": 22402
-    },
-    {
-      "epoch": 1.7128657988799052,
-      "grad_norm": 0.003369504353031516,
-      "learning_rate": 0.00019999855346733604,
-      "loss": 46.0,
-      "step": 22403
-    },
-    {
-      "epoch": 1.712942255863295,
-      "grad_norm": 0.0024576676078140736,
-      "learning_rate": 0.00019999855333813836,
-      "loss": 46.0,
-      "step": 22404
-    },
-    {
-      "epoch": 1.7130187128466847,
-      "grad_norm": 0.001060210750438273,
-      "learning_rate": 0.00019999855320893497,
-      "loss": 46.0,
-      "step": 22405
-    },
-    {
-      "epoch": 1.7130951698300745,
-      "grad_norm": 0.0010517387418076396,
-      "learning_rate": 0.0001999985530797258,
-      "loss": 46.0,
-      "step": 22406
-    },
-    {
-      "epoch": 1.713171626813464,
-      "grad_norm": 0.0012878982815891504,
-      "learning_rate": 0.00019999855295051082,
-      "loss": 46.0,
-      "step": 22407
-    },
-    {
-      "epoch": 1.7132480837968538,
-      "grad_norm": 0.00787353701889515,
-      "learning_rate": 0.0001999985528212901,
-      "loss": 46.0,
-      "step": 22408
-    },
-    {
-      "epoch": 1.7133245407802435,
-      "grad_norm": 0.0009401249117217958,
-      "learning_rate": 0.0001999985526920636,
-      "loss": 46.0,
-      "step": 22409
-    },
-    {
-      "epoch": 1.713400997763633,
-      "grad_norm": 0.0017263785703107715,
-      "learning_rate": 0.00019999855256283132,
-      "loss": 46.0,
-      "step": 22410
-    },
-    {
-      "epoch": 1.7134774547470228,
-      "grad_norm": 0.0007351974491029978,
-      "learning_rate": 0.00019999855243359331,
-      "loss": 46.0,
-      "step": 22411
-    },
-    {
-      "epoch": 1.7135539117304126,
-      "grad_norm": 0.0007150332676246762,
-      "learning_rate": 0.0001999985523043495,
-      "loss": 46.0,
-      "step": 22412
-    },
-    {
-      "epoch": 1.7136303687138024,
-      "grad_norm": 0.007817660458385944,
-      "learning_rate": 0.00019999855217509992,
-      "loss": 46.0,
-      "step": 22413
-    },
-    {
-      "epoch": 1.7137068256971921,
-      "grad_norm": 0.0008801393560133874,
-      "learning_rate": 0.0001999985520458446,
-      "loss": 46.0,
-      "step": 22414
-    },
-    {
-      "epoch": 1.7137832826805819,
-      "grad_norm": 0.003122924827039242,
-      "learning_rate": 0.00019999855191658346,
-      "loss": 46.0,
-      "step": 22415
-    },
-    {
-      "epoch": 1.7138597396639716,
-      "grad_norm": 0.0023737852461636066,
-      "learning_rate": 0.00019999855178731658,
-      "loss": 46.0,
-      "step": 22416
-    },
-    {
-      "epoch": 1.7139361966473614,
-      "grad_norm": 0.001244217506609857,
-      "learning_rate": 0.00019999855165804396,
-      "loss": 46.0,
-      "step": 22417
-    },
-    {
-      "epoch": 1.714012653630751,
-      "grad_norm": 0.0006176520837470889,
-      "learning_rate": 0.00019999855152876553,
-      "loss": 46.0,
-      "step": 22418
-    },
-    {
-      "epoch": 1.7140891106141407,
-      "grad_norm": 0.00201747240498662,
-      "learning_rate": 0.0001999985513994813,
-      "loss": 46.0,
-      "step": 22419
-    },
-    {
-      "epoch": 1.7141655675975305,
-      "grad_norm": 0.001043471391312778,
-      "learning_rate": 0.00019999855127019137,
-      "loss": 46.0,
-      "step": 22420
-    },
-    {
-      "epoch": 1.71424202458092,
-      "grad_norm": 0.0007902960060164332,
-      "learning_rate": 0.00019999855114089562,
-      "loss": 46.0,
-      "step": 22421
-    },
-    {
-      "epoch": 1.7143184815643098,
-      "grad_norm": 0.0017413819441571832,
-      "learning_rate": 0.00019999855101159416,
-      "loss": 46.0,
-      "step": 22422
-    },
-    {
-      "epoch": 1.7143949385476995,
-      "grad_norm": 0.0011369165731593966,
-      "learning_rate": 0.00019999855088228687,
-      "loss": 46.0,
-      "step": 22423
-    },
-    {
-      "epoch": 1.7144713955310893,
-      "grad_norm": 0.003757861442863941,
-      "learning_rate": 0.00019999855075297386,
-      "loss": 46.0,
-      "step": 22424
-    },
-    {
-      "epoch": 1.714547852514479,
-      "grad_norm": 0.0028186722192913294,
-      "learning_rate": 0.00019999855062365502,
-      "loss": 46.0,
-      "step": 22425
-    },
-    {
-      "epoch": 1.7146243094978688,
-      "grad_norm": 0.0009562631603330374,
-      "learning_rate": 0.00019999855049433046,
-      "loss": 46.0,
-      "step": 22426
-    },
-    {
-      "epoch": 1.7147007664812586,
-      "grad_norm": 0.0009940763702616096,
-      "learning_rate": 0.0001999985503650001,
-      "loss": 46.0,
-      "step": 22427
-    },
-    {
-      "epoch": 1.7147772234646483,
-      "grad_norm": 0.0030138599686324596,
-      "learning_rate": 0.000199998550235664,
-      "loss": 46.0,
-      "step": 22428
-    },
-    {
-      "epoch": 1.7148536804480379,
-      "grad_norm": 0.0005366287077777088,
-      "learning_rate": 0.00019999855010632213,
-      "loss": 46.0,
-      "step": 22429
-    },
-    {
-      "epoch": 1.7149301374314276,
-      "grad_norm": 0.001073703751899302,
-      "learning_rate": 0.00019999854997697448,
-      "loss": 46.0,
-      "step": 22430
-    },
-    {
-      "epoch": 1.7150065944148174,
-      "grad_norm": 0.0009289165609516203,
-      "learning_rate": 0.00019999854984762103,
-      "loss": 46.0,
-      "step": 22431
-    },
-    {
-      "epoch": 1.715083051398207,
-      "grad_norm": 0.004284227266907692,
-      "learning_rate": 0.00019999854971826183,
-      "loss": 46.0,
-      "step": 22432
-    },
-    {
-      "epoch": 1.7151595083815967,
-      "grad_norm": 0.0010894276201725006,
-      "learning_rate": 0.00019999854958889689,
-      "loss": 46.0,
-      "step": 22433
-    },
-    {
-      "epoch": 1.7152359653649865,
-      "grad_norm": 0.004678845871239901,
-      "learning_rate": 0.00019999854945952617,
-      "loss": 46.0,
-      "step": 22434
-    },
-    {
-      "epoch": 1.7153124223483762,
-      "grad_norm": 0.0004139160446356982,
-      "learning_rate": 0.00019999854933014968,
-      "loss": 46.0,
-      "step": 22435
-    },
-    {
-      "epoch": 1.715388879331766,
-      "grad_norm": 0.000822081696242094,
-      "learning_rate": 0.0001999985492007674,
-      "loss": 46.0,
-      "step": 22436
-    },
-    {
-      "epoch": 1.7154653363151557,
-      "grad_norm": 0.0018326276913285255,
-      "learning_rate": 0.00019999854907137935,
-      "loss": 46.0,
-      "step": 22437
-    },
-    {
-      "epoch": 1.7155417932985455,
-      "grad_norm": 0.001037940033711493,
-      "learning_rate": 0.00019999854894198554,
-      "loss": 46.0,
-      "step": 22438
-    },
-    {
-      "epoch": 1.7156182502819353,
-      "grad_norm": 0.00408597057685256,
-      "learning_rate": 0.00019999854881258598,
-      "loss": 46.0,
-      "step": 22439
-    },
-    {
-      "epoch": 1.7156947072653248,
-      "grad_norm": 0.0005092289648018777,
-      "learning_rate": 0.00019999854868318063,
-      "loss": 46.0,
-      "step": 22440
-    },
-    {
-      "epoch": 1.7157711642487146,
-      "grad_norm": 0.0005906568840146065,
-      "learning_rate": 0.00019999854855376952,
-      "loss": 46.0,
-      "step": 22441
-    },
-    {
-      "epoch": 1.7158476212321043,
-      "grad_norm": 0.0013110678410157561,
-      "learning_rate": 0.00019999854842435265,
-      "loss": 46.0,
-      "step": 22442
-    },
-    {
-      "epoch": 1.7159240782154939,
-      "grad_norm": 0.0006217741756699979,
-      "learning_rate": 0.00019999854829493,
-      "loss": 46.0,
-      "step": 22443
-    },
-    {
-      "epoch": 1.7160005351988836,
-      "grad_norm": 0.0018422106513753533,
-      "learning_rate": 0.00019999854816550158,
-      "loss": 46.0,
-      "step": 22444
-    },
-    {
-      "epoch": 1.7160769921822734,
-      "grad_norm": 0.0005019055097363889,
-      "learning_rate": 0.00019999854803606738,
-      "loss": 46.0,
-      "step": 22445
-    },
-    {
-      "epoch": 1.7161534491656631,
-      "grad_norm": 0.0008745864033699036,
-      "learning_rate": 0.0001999985479066274,
-      "loss": 46.0,
-      "step": 22446
-    },
-    {
-      "epoch": 1.716229906149053,
-      "grad_norm": 0.0008612436940893531,
-      "learning_rate": 0.00019999854777718167,
-      "loss": 46.0,
-      "step": 22447
-    },
-    {
-      "epoch": 1.7163063631324427,
-      "grad_norm": 0.0018096229759976268,
-      "learning_rate": 0.00019999854764773015,
-      "loss": 46.0,
-      "step": 22448
-    },
-    {
-      "epoch": 1.7163828201158324,
-      "grad_norm": 0.005119806155562401,
-      "learning_rate": 0.00019999854751827289,
-      "loss": 46.0,
-      "step": 22449
-    },
-    {
-      "epoch": 1.7164592770992222,
-      "grad_norm": 0.0006359869730658829,
-      "learning_rate": 0.00019999854738880985,
-      "loss": 46.0,
-      "step": 22450
-    },
-    {
-      "epoch": 1.7165357340826117,
-      "grad_norm": 0.0008692768751643598,
-      "learning_rate": 0.00019999854725934104,
-      "loss": 46.0,
-      "step": 22451
-    },
-    {
-      "epoch": 1.7166121910660015,
-      "grad_norm": 0.0011077569797635078,
-      "learning_rate": 0.00019999854712986648,
-      "loss": 46.0,
-      "step": 22452
-    },
-    {
-      "epoch": 1.7166886480493913,
-      "grad_norm": 0.0009412099607288837,
-      "learning_rate": 0.00019999854700038613,
-      "loss": 46.0,
-      "step": 22453
-    },
-    {
-      "epoch": 1.7167651050327808,
-      "grad_norm": 0.003998979926109314,
-      "learning_rate": 0.0001999985468709,
-      "loss": 46.0,
-      "step": 22454
-    },
-    {
-      "epoch": 1.7168415620161706,
-      "grad_norm": 0.0013675636146217585,
-      "learning_rate": 0.0001999985467414081,
-      "loss": 46.0,
-      "step": 22455
-    },
-    {
-      "epoch": 1.7169180189995603,
-      "grad_norm": 0.0029458513017743826,
-      "learning_rate": 0.00019999854661191044,
-      "loss": 46.0,
-      "step": 22456
-    },
-    {
-      "epoch": 1.71699447598295,
-      "grad_norm": 0.0009958730079233646,
-      "learning_rate": 0.00019999854648240705,
-      "loss": 46.0,
-      "step": 22457
-    },
-    {
-      "epoch": 1.7170709329663398,
-      "grad_norm": 0.0008039625827223063,
-      "learning_rate": 0.00019999854635289782,
-      "loss": 46.0,
-      "step": 22458
-    },
-    {
-      "epoch": 1.7171473899497296,
-      "grad_norm": 0.0002651166287250817,
-      "learning_rate": 0.00019999854622338288,
-      "loss": 46.0,
-      "step": 22459
-    },
-    {
-      "epoch": 1.7172238469331194,
-      "grad_norm": 0.0006535813445225358,
-      "learning_rate": 0.00019999854609386214,
-      "loss": 46.0,
-      "step": 22460
-    },
-    {
-      "epoch": 1.7173003039165091,
-      "grad_norm": 0.002061316976323724,
-      "learning_rate": 0.00019999854596433565,
-      "loss": 46.0,
-      "step": 22461
-    },
-    {
-      "epoch": 1.7173767608998987,
-      "grad_norm": 0.001987624214962125,
-      "learning_rate": 0.00019999854583480336,
-      "loss": 46.0,
-      "step": 22462
-    },
-    {
-      "epoch": 1.7174532178832884,
-      "grad_norm": 0.0033599711023271084,
-      "learning_rate": 0.00019999854570526532,
-      "loss": 46.0,
-      "step": 22463
-    },
-    {
-      "epoch": 1.7175296748666782,
-      "grad_norm": 0.0007719573331996799,
-      "learning_rate": 0.00019999854557572151,
-      "loss": 46.0,
-      "step": 22464
-    },
-    {
-      "epoch": 1.7176061318500677,
-      "grad_norm": 0.004135304596275091,
-      "learning_rate": 0.0001999985454461719,
-      "loss": 46.0,
-      "step": 22465
-    },
-    {
-      "epoch": 1.7176825888334575,
-      "grad_norm": 0.004250828176736832,
-      "learning_rate": 0.00019999854531661657,
-      "loss": 46.0,
-      "step": 22466
-    },
-    {
-      "epoch": 1.7177590458168472,
-      "grad_norm": 0.0006882243906147778,
-      "learning_rate": 0.00019999854518705545,
-      "loss": 46.0,
-      "step": 22467
-    },
-    {
-      "epoch": 1.717835502800237,
-      "grad_norm": 0.004639922175556421,
-      "learning_rate": 0.00019999854505748857,
-      "loss": 46.0,
-      "step": 22468
-    },
-    {
-      "epoch": 1.7179119597836268,
-      "grad_norm": 0.0007389179663732648,
-      "learning_rate": 0.0001999985449279159,
-      "loss": 46.0,
-      "step": 22469
-    },
-    {
-      "epoch": 1.7179884167670165,
-      "grad_norm": 0.0032404428347945213,
-      "learning_rate": 0.0001999985447983375,
-      "loss": 46.0,
-      "step": 22470
-    },
-    {
-      "epoch": 1.7180648737504063,
-      "grad_norm": 0.0012873783707618713,
-      "learning_rate": 0.0001999985446687533,
-      "loss": 46.0,
-      "step": 22471
-    },
-    {
-      "epoch": 1.718141330733796,
-      "grad_norm": 0.003340303199365735,
-      "learning_rate": 0.0001999985445391633,
-      "loss": 46.0,
-      "step": 22472
-    },
-    {
-      "epoch": 1.7182177877171856,
-      "grad_norm": 0.0004160654207225889,
-      "learning_rate": 0.0001999985444095676,
-      "loss": 46.0,
-      "step": 22473
-    },
-    {
-      "epoch": 1.7182942447005753,
-      "grad_norm": 0.0007767430506646633,
-      "learning_rate": 0.00019999854427996607,
-      "loss": 46.0,
-      "step": 22474
-    },
-    {
-      "epoch": 1.718370701683965,
-      "grad_norm": 0.0007537707570008934,
-      "learning_rate": 0.0001999985441503588,
-      "loss": 46.0,
-      "step": 22475
-    },
-    {
-      "epoch": 1.7184471586673546,
-      "grad_norm": 0.003122064284980297,
-      "learning_rate": 0.00019999854402074577,
-      "loss": 46.0,
-      "step": 22476
-    },
-    {
-      "epoch": 1.7185236156507444,
-      "grad_norm": 0.003668226068839431,
-      "learning_rate": 0.00019999854389112697,
-      "loss": 46.0,
-      "step": 22477
-    },
-    {
-      "epoch": 1.7186000726341342,
-      "grad_norm": 0.0017019016668200493,
-      "learning_rate": 0.00019999854376150236,
-      "loss": 46.0,
-      "step": 22478
-    },
-    {
-      "epoch": 1.718676529617524,
-      "grad_norm": 0.0016353281680494547,
-      "learning_rate": 0.000199998543631872,
-      "loss": 46.0,
-      "step": 22479
-    },
-    {
-      "epoch": 1.7187529866009137,
-      "grad_norm": 0.00042351995944045484,
-      "learning_rate": 0.0001999985435022359,
-      "loss": 46.0,
-      "step": 22480
-    },
-    {
-      "epoch": 1.7188294435843035,
-      "grad_norm": 0.0013452437706291676,
-      "learning_rate": 0.000199998543372594,
-      "loss": 46.0,
-      "step": 22481
-    },
-    {
-      "epoch": 1.7189059005676932,
-      "grad_norm": 0.005218386184424162,
-      "learning_rate": 0.00019999854324294635,
-      "loss": 46.0,
-      "step": 22482
-    },
-    {
-      "epoch": 1.718982357551083,
-      "grad_norm": 0.001016703899949789,
-      "learning_rate": 0.00019999854311329292,
-      "loss": 46.0,
-      "step": 22483
-    },
-    {
-      "epoch": 1.7190588145344725,
-      "grad_norm": 0.0014405595138669014,
-      "learning_rate": 0.00019999854298363373,
-      "loss": 46.0,
-      "step": 22484
-    },
-    {
-      "epoch": 1.7191352715178623,
-      "grad_norm": 0.0009401018614880741,
-      "learning_rate": 0.00019999854285396876,
-      "loss": 46.0,
-      "step": 22485
-    },
-    {
-      "epoch": 1.719211728501252,
-      "grad_norm": 0.0015400574775412679,
-      "learning_rate": 0.00019999854272429802,
-      "loss": 46.0,
-      "step": 22486
-    },
-    {
-      "epoch": 1.7192881854846416,
-      "grad_norm": 0.0007076390320435166,
-      "learning_rate": 0.0001999985425946215,
-      "loss": 46.0,
-      "step": 22487
-    },
-    {
-      "epoch": 1.7193646424680313,
-      "grad_norm": 0.0006557370652444661,
-      "learning_rate": 0.00019999854246493924,
-      "loss": 46.0,
-      "step": 22488
-    },
-    {
-      "epoch": 1.719441099451421,
-      "grad_norm": 0.0005964445881545544,
-      "learning_rate": 0.0001999985423352512,
-      "loss": 46.0,
-      "step": 22489
-    },
-    {
-      "epoch": 1.7195175564348109,
-      "grad_norm": 0.0004475999448914081,
-      "learning_rate": 0.0001999985422055574,
-      "loss": 46.0,
-      "step": 22490
-    },
-    {
-      "epoch": 1.7195940134182006,
-      "grad_norm": 0.0014614162500947714,
-      "learning_rate": 0.00019999854207585778,
-      "loss": 46.0,
-      "step": 22491
-    },
-    {
-      "epoch": 1.7196704704015904,
-      "grad_norm": 0.0034353667870163918,
-      "learning_rate": 0.00019999854194615243,
-      "loss": 46.0,
-      "step": 22492
-    },
-    {
-      "epoch": 1.7197469273849801,
-      "grad_norm": 0.0009692535386420786,
-      "learning_rate": 0.0001999985418164413,
-      "loss": 46.0,
-      "step": 22493
-    },
-    {
-      "epoch": 1.71982338436837,
-      "grad_norm": 0.0014354020822793245,
-      "learning_rate": 0.0001999985416867244,
-      "loss": 46.0,
-      "step": 22494
-    },
-    {
-      "epoch": 1.7198998413517594,
-      "grad_norm": 0.0011246701469644904,
-      "learning_rate": 0.00019999854155700178,
-      "loss": 46.0,
-      "step": 22495
-    },
-    {
-      "epoch": 1.7199762983351492,
-      "grad_norm": 0.0016662514535710216,
-      "learning_rate": 0.00019999854142727333,
-      "loss": 46.0,
-      "step": 22496
-    },
-    {
-      "epoch": 1.720052755318539,
-      "grad_norm": 0.0009482745081186295,
-      "learning_rate": 0.00019999854129753913,
-      "loss": 46.0,
-      "step": 22497
-    },
-    {
-      "epoch": 1.7201292123019285,
-      "grad_norm": 0.0021958444267511368,
-      "learning_rate": 0.00019999854116779914,
-      "loss": 46.0,
-      "step": 22498
-    },
-    {
-      "epoch": 1.7202056692853183,
-      "grad_norm": 0.001114825252443552,
-      "learning_rate": 0.00019999854103805342,
-      "loss": 46.0,
-      "step": 22499
-    },
-    {
-      "epoch": 1.720282126268708,
-      "grad_norm": 0.0007832496194168925,
-      "learning_rate": 0.00019999854090830193,
-      "loss": 46.0,
-      "step": 22500
-    },
-    {
-      "epoch": 1.7203585832520978,
-      "grad_norm": 0.0022682598792016506,
-      "learning_rate": 0.00019999854077854462,
-      "loss": 46.0,
-      "step": 22501
-    },
-    {
-      "epoch": 1.7204350402354875,
-      "grad_norm": 0.0006361659034155309,
-      "learning_rate": 0.00019999854064878158,
-      "loss": 46.0,
-      "step": 22502
-    },
-    {
-      "epoch": 1.7205114972188773,
-      "grad_norm": 0.0004284860042389482,
-      "learning_rate": 0.00019999854051901277,
-      "loss": 46.0,
-      "step": 22503
-    },
-    {
-      "epoch": 1.720587954202267,
-      "grad_norm": 0.0005408914876170456,
-      "learning_rate": 0.0001999985403892382,
-      "loss": 46.0,
-      "step": 22504
-    },
-    {
-      "epoch": 1.7206644111856568,
-      "grad_norm": 0.0012192067224532366,
-      "learning_rate": 0.00019999854025945784,
-      "loss": 46.0,
-      "step": 22505
-    },
-    {
-      "epoch": 1.7207408681690464,
-      "grad_norm": 0.002179897390305996,
-      "learning_rate": 0.0001999985401296717,
-      "loss": 46.0,
-      "step": 22506
-    },
-    {
-      "epoch": 1.7208173251524361,
-      "grad_norm": 0.0016331454971805215,
-      "learning_rate": 0.0001999985399998798,
-      "loss": 46.0,
-      "step": 22507
-    },
-    {
-      "epoch": 1.7208937821358257,
-      "grad_norm": 0.0018505053594708443,
-      "learning_rate": 0.00019999853987008216,
-      "loss": 46.0,
-      "step": 22508
-    },
-    {
-      "epoch": 1.7209702391192154,
-      "grad_norm": 0.0017334234435111284,
-      "learning_rate": 0.0001999985397402787,
-      "loss": 46.0,
-      "step": 22509
-    },
-    {
-      "epoch": 1.7210466961026052,
-      "grad_norm": 0.004260999150574207,
-      "learning_rate": 0.0001999985396104695,
-      "loss": 46.0,
-      "step": 22510
-    },
-    {
-      "epoch": 1.721123153085995,
-      "grad_norm": 0.0005032224580645561,
-      "learning_rate": 0.00019999853948065454,
-      "loss": 46.0,
-      "step": 22511
-    },
-    {
-      "epoch": 1.7211996100693847,
-      "grad_norm": 0.0007490611169487238,
-      "learning_rate": 0.0001999985393508338,
-      "loss": 46.0,
-      "step": 22512
-    },
-    {
-      "epoch": 1.7212760670527745,
-      "grad_norm": 0.0002332647272851318,
-      "learning_rate": 0.0001999985392210073,
-      "loss": 46.0,
-      "step": 22513
-    },
-    {
-      "epoch": 1.7213525240361642,
-      "grad_norm": 0.008897039107978344,
-      "learning_rate": 0.000199998539091175,
-      "loss": 46.0,
-      "step": 22514
-    },
-    {
-      "epoch": 1.721428981019554,
-      "grad_norm": 0.002618935890495777,
-      "learning_rate": 0.00019999853896133696,
-      "loss": 46.0,
-      "step": 22515
-    },
-    {
-      "epoch": 1.7215054380029438,
-      "grad_norm": 0.000531247875187546,
-      "learning_rate": 0.00019999853883149316,
-      "loss": 46.0,
-      "step": 22516
-    },
-    {
-      "epoch": 1.7215818949863333,
-      "grad_norm": 0.001366589916869998,
-      "learning_rate": 0.00019999853870164355,
-      "loss": 46.0,
-      "step": 22517
-    },
-    {
-      "epoch": 1.721658351969723,
-      "grad_norm": 0.001164983375929296,
-      "learning_rate": 0.0001999985385717882,
-      "loss": 46.0,
-      "step": 22518
-    },
-    {
-      "epoch": 1.7217348089531126,
-      "grad_norm": 0.002358474303036928,
-      "learning_rate": 0.0001999985384419271,
-      "loss": 46.0,
-      "step": 22519
-    },
-    {
-      "epoch": 1.7218112659365024,
-      "grad_norm": 0.0009585806983523071,
-      "learning_rate": 0.00019999853831206016,
-      "loss": 46.0,
-      "step": 22520
-    },
-    {
-      "epoch": 1.7218877229198921,
-      "grad_norm": 0.00181785156019032,
-      "learning_rate": 0.00019999853818218751,
-      "loss": 46.0,
-      "step": 22521
-    },
-    {
-      "epoch": 1.7219641799032819,
-      "grad_norm": 0.0008936924859881401,
-      "learning_rate": 0.0001999985380523091,
-      "loss": 46.0,
-      "step": 22522
-    },
-    {
-      "epoch": 1.7220406368866716,
-      "grad_norm": 0.004332865122705698,
-      "learning_rate": 0.00019999853792242487,
-      "loss": 46.0,
-      "step": 22523
-    },
-    {
-      "epoch": 1.7221170938700614,
-      "grad_norm": 0.0008371897274628282,
-      "learning_rate": 0.0001999985377925349,
-      "loss": 46.0,
-      "step": 22524
-    },
-    {
-      "epoch": 1.7221935508534512,
-      "grad_norm": 0.0009357577073387802,
-      "learning_rate": 0.00019999853766263916,
-      "loss": 46.0,
-      "step": 22525
-    },
-    {
-      "epoch": 1.722270007836841,
-      "grad_norm": 0.001828839653171599,
-      "learning_rate": 0.00019999853753273765,
-      "loss": 46.0,
-      "step": 22526
-    },
-    {
-      "epoch": 1.7223464648202307,
-      "grad_norm": 0.0008495564106851816,
-      "learning_rate": 0.00019999853740283036,
-      "loss": 46.0,
-      "step": 22527
-    },
-    {
-      "epoch": 1.7224229218036202,
-      "grad_norm": 0.0005137830739840865,
-      "learning_rate": 0.0001999985372729173,
-      "loss": 46.0,
-      "step": 22528
-    },
-    {
-      "epoch": 1.72249937878701,
-      "grad_norm": 0.0014874048065394163,
-      "learning_rate": 0.0001999985371429985,
-      "loss": 46.0,
-      "step": 22529
-    },
-    {
-      "epoch": 1.7225758357703995,
-      "grad_norm": 0.0021972861140966415,
-      "learning_rate": 0.0001999985370130739,
-      "loss": 46.0,
-      "step": 22530
-    },
-    {
-      "epoch": 1.7226522927537893,
-      "grad_norm": 0.0007032732246443629,
-      "learning_rate": 0.00019999853688314355,
-      "loss": 46.0,
-      "step": 22531
-    },
-    {
-      "epoch": 1.722728749737179,
-      "grad_norm": 0.0021237845066934824,
-      "learning_rate": 0.0001999985367532074,
-      "loss": 46.0,
-      "step": 22532
-    },
-    {
-      "epoch": 1.7228052067205688,
-      "grad_norm": 0.000899318780284375,
-      "learning_rate": 0.0001999985366232655,
-      "loss": 46.0,
-      "step": 22533
-    },
-    {
-      "epoch": 1.7228816637039586,
-      "grad_norm": 0.0013699724804610014,
-      "learning_rate": 0.00019999853649331785,
-      "loss": 46.0,
-      "step": 22534
-    },
-    {
-      "epoch": 1.7229581206873483,
-      "grad_norm": 0.000825609196908772,
-      "learning_rate": 0.00019999853636336443,
-      "loss": 46.0,
-      "step": 22535
-    },
-    {
-      "epoch": 1.723034577670738,
-      "grad_norm": 0.0011562325526028872,
-      "learning_rate": 0.0001999985362334052,
-      "loss": 46.0,
-      "step": 22536
-    },
-    {
-      "epoch": 1.7231110346541278,
-      "grad_norm": 0.0005326572572812438,
-      "learning_rate": 0.00019999853610344024,
-      "loss": 46.0,
-      "step": 22537
-    },
-    {
-      "epoch": 1.7231874916375174,
-      "grad_norm": 0.0018527604406699538,
-      "learning_rate": 0.0001999985359734695,
-      "loss": 46.0,
-      "step": 22538
-    },
-    {
-      "epoch": 1.7232639486209071,
-      "grad_norm": 0.0006820294074714184,
-      "learning_rate": 0.00019999853584349298,
-      "loss": 46.0,
-      "step": 22539
-    },
-    {
-      "epoch": 1.723340405604297,
-      "grad_norm": 0.0011219180887565017,
-      "learning_rate": 0.0001999985357135107,
-      "loss": 46.0,
-      "step": 22540
-    },
-    {
-      "epoch": 1.7234168625876864,
-      "grad_norm": 0.001367273973301053,
-      "learning_rate": 0.00019999853558352263,
-      "loss": 46.0,
-      "step": 22541
-    },
-    {
-      "epoch": 1.7234933195710762,
-      "grad_norm": 0.0005039800889790058,
-      "learning_rate": 0.00019999853545352882,
-      "loss": 46.0,
-      "step": 22542
-    },
-    {
-      "epoch": 1.723569776554466,
-      "grad_norm": 0.0016077890759333968,
-      "learning_rate": 0.00019999853532352924,
-      "loss": 46.0,
-      "step": 22543
-    },
-    {
-      "epoch": 1.7236462335378557,
-      "grad_norm": 0.0010009424295276403,
-      "learning_rate": 0.00019999853519352386,
-      "loss": 46.0,
-      "step": 22544
-    },
-    {
-      "epoch": 1.7237226905212455,
-      "grad_norm": 0.0028042467311024666,
-      "learning_rate": 0.00019999853506351273,
-      "loss": 46.0,
-      "step": 22545
-    },
-    {
-      "epoch": 1.7237991475046353,
-      "grad_norm": 0.0027355896309018135,
-      "learning_rate": 0.00019999853493349586,
-      "loss": 46.0,
-      "step": 22546
-    },
-    {
-      "epoch": 1.723875604488025,
-      "grad_norm": 0.0010892339050769806,
-      "learning_rate": 0.00019999853480347316,
-      "loss": 46.0,
-      "step": 22547
-    },
-    {
-      "epoch": 1.7239520614714148,
-      "grad_norm": 0.007179956883192062,
-      "learning_rate": 0.0001999985346734447,
-      "loss": 46.0,
-      "step": 22548
-    },
-    {
-      "epoch": 1.7240285184548043,
-      "grad_norm": 0.0008726438391022384,
-      "learning_rate": 0.00019999853454341052,
-      "loss": 46.0,
-      "step": 22549
-    },
-    {
-      "epoch": 1.724104975438194,
-      "grad_norm": 0.001996300183236599,
-      "learning_rate": 0.00019999853441337052,
-      "loss": 46.0,
-      "step": 22550
-    },
-    {
-      "epoch": 1.7241814324215838,
-      "grad_norm": 0.00043070968240499496,
-      "learning_rate": 0.0001999985342833248,
-      "loss": 46.0,
-      "step": 22551
-    },
-    {
-      "epoch": 1.7242578894049734,
-      "grad_norm": 0.0023758583702147007,
-      "learning_rate": 0.0001999985341532733,
-      "loss": 46.0,
-      "step": 22552
-    },
-    {
-      "epoch": 1.7243343463883631,
-      "grad_norm": 0.0009823755826801062,
-      "learning_rate": 0.000199998534023216,
-      "loss": 46.0,
-      "step": 22553
-    },
-    {
-      "epoch": 1.724410803371753,
-      "grad_norm": 0.0009820658015087247,
-      "learning_rate": 0.00019999853389315295,
-      "loss": 46.0,
-      "step": 22554
-    },
-    {
-      "epoch": 1.7244872603551427,
-      "grad_norm": 0.0009053281391970813,
-      "learning_rate": 0.00019999853376308414,
-      "loss": 46.0,
-      "step": 22555
-    },
-    {
-      "epoch": 1.7245637173385324,
-      "grad_norm": 0.0004495630564633757,
-      "learning_rate": 0.00019999853363300953,
-      "loss": 46.0,
-      "step": 22556
-    },
-    {
-      "epoch": 1.7246401743219222,
-      "grad_norm": 0.00098418106790632,
-      "learning_rate": 0.00019999853350292918,
-      "loss": 46.0,
-      "step": 22557
-    },
-    {
-      "epoch": 1.724716631305312,
-      "grad_norm": 0.0007086388650350273,
-      "learning_rate": 0.00019999853337284305,
-      "loss": 46.0,
-      "step": 22558
-    },
-    {
-      "epoch": 1.7247930882887017,
-      "grad_norm": 0.0004983508842997253,
-      "learning_rate": 0.00019999853324275112,
-      "loss": 46.0,
-      "step": 22559
-    },
-    {
-      "epoch": 1.7248695452720912,
-      "grad_norm": 0.005471676122397184,
-      "learning_rate": 0.00019999853311265347,
-      "loss": 46.0,
-      "step": 22560
-    },
-    {
-      "epoch": 1.724946002255481,
-      "grad_norm": 0.005148309748619795,
-      "learning_rate": 0.00019999853298255003,
-      "loss": 46.0,
-      "step": 22561
-    },
-    {
-      "epoch": 1.7250224592388708,
-      "grad_norm": 0.002501945709809661,
-      "learning_rate": 0.0001999985328524408,
-      "loss": 46.0,
-      "step": 22562
-    },
-    {
-      "epoch": 1.7250989162222603,
-      "grad_norm": 0.0008293217397294939,
-      "learning_rate": 0.00019999853272232584,
-      "loss": 46.0,
-      "step": 22563
-    },
-    {
-      "epoch": 1.72517537320565,
-      "grad_norm": 0.0006360613624565303,
-      "learning_rate": 0.0001999985325922051,
-      "loss": 46.0,
-      "step": 22564
-    },
-    {
-      "epoch": 1.7252518301890398,
-      "grad_norm": 0.0017641987651586533,
-      "learning_rate": 0.00019999853246207856,
-      "loss": 46.0,
-      "step": 22565
-    },
-    {
-      "epoch": 1.7253282871724296,
-      "grad_norm": 0.0011632274836301804,
-      "learning_rate": 0.0001999985323319463,
-      "loss": 46.0,
-      "step": 22566
-    },
-    {
-      "epoch": 1.7254047441558193,
-      "grad_norm": 0.0006391158094629645,
-      "learning_rate": 0.0001999985322018082,
-      "loss": 46.0,
-      "step": 22567
-    },
-    {
-      "epoch": 1.725481201139209,
-      "grad_norm": 0.0006679279613308609,
-      "learning_rate": 0.0001999985320716644,
-      "loss": 46.0,
-      "step": 22568
-    },
-    {
-      "epoch": 1.7255576581225989,
-      "grad_norm": 0.0006939740851521492,
-      "learning_rate": 0.00019999853194151482,
-      "loss": 46.0,
-      "step": 22569
-    },
-    {
-      "epoch": 1.7256341151059886,
-      "grad_norm": 0.0010613654740154743,
-      "learning_rate": 0.00019999853181135944,
-      "loss": 46.0,
-      "step": 22570
-    },
-    {
-      "epoch": 1.7257105720893782,
-      "grad_norm": 0.0018004346638917923,
-      "learning_rate": 0.00019999853168119831,
-      "loss": 46.0,
-      "step": 22571
-    },
-    {
-      "epoch": 1.725787029072768,
-      "grad_norm": 0.0008225837373174727,
-      "learning_rate": 0.0001999985315510314,
-      "loss": 46.0,
-      "step": 22572
-    },
-    {
-      "epoch": 1.7258634860561577,
-      "grad_norm": 0.0013309657806530595,
-      "learning_rate": 0.00019999853142085874,
-      "loss": 46.0,
-      "step": 22573
-    },
-    {
-      "epoch": 1.7259399430395472,
-      "grad_norm": 0.0017102316487580538,
-      "learning_rate": 0.0001999985312906803,
-      "loss": 46.0,
-      "step": 22574
-    },
-    {
-      "epoch": 1.726016400022937,
-      "grad_norm": 0.0008002474787645042,
-      "learning_rate": 0.00019999853116049607,
-      "loss": 46.0,
-      "step": 22575
-    },
-    {
-      "epoch": 1.7260928570063268,
-      "grad_norm": 0.0006907505448907614,
-      "learning_rate": 0.0001999985310303061,
-      "loss": 46.0,
-      "step": 22576
-    },
-    {
-      "epoch": 1.7261693139897165,
-      "grad_norm": 0.0005228658556006849,
-      "learning_rate": 0.00019999853090011036,
-      "loss": 46.0,
-      "step": 22577
-    },
-    {
-      "epoch": 1.7262457709731063,
-      "grad_norm": 0.0005460300599224865,
-      "learning_rate": 0.00019999853076990882,
-      "loss": 46.0,
-      "step": 22578
-    },
-    {
-      "epoch": 1.726322227956496,
-      "grad_norm": 0.0007252333452925086,
-      "learning_rate": 0.00019999853063970156,
-      "loss": 46.0,
-      "step": 22579
-    },
-    {
-      "epoch": 1.7263986849398858,
-      "grad_norm": 0.000526864139828831,
-      "learning_rate": 0.00019999853050948848,
-      "loss": 46.0,
-      "step": 22580
-    },
-    {
-      "epoch": 1.7264751419232756,
-      "grad_norm": 0.0010521203512325883,
-      "learning_rate": 0.00019999853037926964,
-      "loss": 46.0,
-      "step": 22581
-    },
-    {
-      "epoch": 1.726551598906665,
-      "grad_norm": 0.0006113207200542092,
-      "learning_rate": 0.00019999853024904506,
-      "loss": 46.0,
-      "step": 22582
-    },
-    {
-      "epoch": 1.7266280558900549,
-      "grad_norm": 0.000968958018347621,
-      "learning_rate": 0.0001999985301188147,
-      "loss": 46.0,
-      "step": 22583
-    },
-    {
-      "epoch": 1.7267045128734446,
-      "grad_norm": 0.0004944971296936274,
-      "learning_rate": 0.00019999852998857856,
-      "loss": 46.0,
-      "step": 22584
-    },
-    {
-      "epoch": 1.7267809698568342,
-      "grad_norm": 0.0005346629186533391,
-      "learning_rate": 0.00019999852985833668,
-      "loss": 46.0,
-      "step": 22585
-    },
-    {
-      "epoch": 1.726857426840224,
-      "grad_norm": 0.0019707956817001104,
-      "learning_rate": 0.00019999852972808898,
-      "loss": 46.0,
-      "step": 22586
-    },
-    {
-      "epoch": 1.7269338838236137,
-      "grad_norm": 0.0025061715859919786,
-      "learning_rate": 0.00019999852959783553,
-      "loss": 46.0,
-      "step": 22587
-    },
-    {
-      "epoch": 1.7270103408070034,
-      "grad_norm": 0.000909914611838758,
-      "learning_rate": 0.00019999852946757631,
-      "loss": 46.0,
-      "step": 22588
-    },
-    {
-      "epoch": 1.7270867977903932,
-      "grad_norm": 0.0013463539071381092,
-      "learning_rate": 0.00019999852933731135,
-      "loss": 46.0,
-      "step": 22589
-    },
-    {
-      "epoch": 1.727163254773783,
-      "grad_norm": 0.0014716185396537185,
-      "learning_rate": 0.00019999852920704058,
-      "loss": 46.0,
-      "step": 22590
-    },
-    {
-      "epoch": 1.7272397117571727,
-      "grad_norm": 0.0021005552262067795,
-      "learning_rate": 0.00019999852907676407,
-      "loss": 46.0,
-      "step": 22591
-    },
-    {
-      "epoch": 1.7273161687405625,
-      "grad_norm": 0.000704847218003124,
-      "learning_rate": 0.00019999852894648175,
-      "loss": 46.0,
-      "step": 22592
-    },
-    {
-      "epoch": 1.727392625723952,
-      "grad_norm": 0.0011378333438187838,
-      "learning_rate": 0.00019999852881619372,
-      "loss": 46.0,
-      "step": 22593
-    },
-    {
-      "epoch": 1.7274690827073418,
-      "grad_norm": 0.0004219490510877222,
-      "learning_rate": 0.0001999985286858999,
-      "loss": 46.0,
-      "step": 22594
-    },
-    {
-      "epoch": 1.7275455396907315,
-      "grad_norm": 0.0025741581339389086,
-      "learning_rate": 0.0001999985285556003,
-      "loss": 46.0,
-      "step": 22595
-    },
-    {
-      "epoch": 1.727621996674121,
-      "grad_norm": 0.0034743815194815397,
-      "learning_rate": 0.00019999852842529493,
-      "loss": 46.0,
-      "step": 22596
-    },
-    {
-      "epoch": 1.7276984536575108,
-      "grad_norm": 0.0005063615390099585,
-      "learning_rate": 0.0001999985282949838,
-      "loss": 46.0,
-      "step": 22597
-    },
-    {
-      "epoch": 1.7277749106409006,
-      "grad_norm": 0.0009142651106230915,
-      "learning_rate": 0.0001999985281646669,
-      "loss": 46.0,
-      "step": 22598
-    },
-    {
-      "epoch": 1.7278513676242904,
-      "grad_norm": 0.0011641262099146843,
-      "learning_rate": 0.0001999985280343442,
-      "loss": 46.0,
-      "step": 22599
-    },
-    {
-      "epoch": 1.7279278246076801,
-      "grad_norm": 0.0010043210349977016,
-      "learning_rate": 0.00019999852790401575,
-      "loss": 46.0,
-      "step": 22600
-    },
-    {
-      "epoch": 1.72800428159107,
-      "grad_norm": 0.002094058785587549,
-      "learning_rate": 0.00019999852777368156,
-      "loss": 46.0,
-      "step": 22601
-    },
-    {
-      "epoch": 1.7280807385744597,
-      "grad_norm": 0.0008834440959617496,
-      "learning_rate": 0.00019999852764334154,
-      "loss": 46.0,
-      "step": 22602
-    },
-    {
-      "epoch": 1.7281571955578494,
-      "grad_norm": 0.011193652637302876,
-      "learning_rate": 0.00019999852751299583,
-      "loss": 46.0,
-      "step": 22603
-    },
-    {
-      "epoch": 1.728233652541239,
-      "grad_norm": 0.0003516439173836261,
-      "learning_rate": 0.0001999985273826443,
-      "loss": 46.0,
-      "step": 22604
-    },
-    {
-      "epoch": 1.7283101095246287,
-      "grad_norm": 0.0008961883140727878,
-      "learning_rate": 0.00019999852725228697,
-      "loss": 46.0,
-      "step": 22605
-    },
-    {
-      "epoch": 1.7283865665080185,
-      "grad_norm": 0.0012827515602111816,
-      "learning_rate": 0.00019999852712192394,
-      "loss": 46.0,
-      "step": 22606
-    },
-    {
-      "epoch": 1.728463023491408,
-      "grad_norm": 0.0034592922311276197,
-      "learning_rate": 0.0001999985269915551,
-      "loss": 46.0,
-      "step": 22607
-    },
-    {
-      "epoch": 1.7285394804747978,
-      "grad_norm": 0.000991699518635869,
-      "learning_rate": 0.0001999985268611805,
-      "loss": 46.0,
-      "step": 22608
-    },
-    {
-      "epoch": 1.7286159374581875,
-      "grad_norm": 0.0010638709645718336,
-      "learning_rate": 0.00019999852673080012,
-      "loss": 46.0,
-      "step": 22609
-    },
-    {
-      "epoch": 1.7286923944415773,
-      "grad_norm": 0.0017935744253918529,
-      "learning_rate": 0.00019999852660041397,
-      "loss": 46.0,
-      "step": 22610
-    },
-    {
-      "epoch": 1.728768851424967,
-      "grad_norm": 0.0016181034734472632,
-      "learning_rate": 0.0001999985264700221,
-      "loss": 46.0,
-      "step": 22611
-    },
-    {
-      "epoch": 1.7288453084083568,
-      "grad_norm": 0.00338179268874228,
-      "learning_rate": 0.00019999852633962442,
-      "loss": 46.0,
-      "step": 22612
-    },
-    {
-      "epoch": 1.7289217653917466,
-      "grad_norm": 0.002531736623495817,
-      "learning_rate": 0.00019999852620922095,
-      "loss": 46.0,
-      "step": 22613
-    },
-    {
-      "epoch": 1.7289982223751363,
-      "grad_norm": 0.00115268025547266,
-      "learning_rate": 0.00019999852607881176,
-      "loss": 46.0,
-      "step": 22614
-    },
-    {
-      "epoch": 1.7290746793585259,
-      "grad_norm": 0.0017992340726777911,
-      "learning_rate": 0.00019999852594839674,
-      "loss": 46.0,
-      "step": 22615
-    },
-    {
-      "epoch": 1.7291511363419156,
-      "grad_norm": 0.002800759393721819,
-      "learning_rate": 0.000199998525817976,
-      "loss": 46.0,
-      "step": 22616
-    },
-    {
-      "epoch": 1.7292275933253054,
-      "grad_norm": 0.0012787943705916405,
-      "learning_rate": 0.00019999852568754949,
-      "loss": 46.0,
-      "step": 22617
-    },
-    {
-      "epoch": 1.729304050308695,
-      "grad_norm": 0.0023138525430113077,
-      "learning_rate": 0.00019999852555711717,
-      "loss": 46.0,
-      "step": 22618
-    },
-    {
-      "epoch": 1.7293805072920847,
-      "grad_norm": 0.0033475991804152727,
-      "learning_rate": 0.0001999985254266791,
-      "loss": 46.0,
-      "step": 22619
-    },
-    {
-      "epoch": 1.7294569642754745,
-      "grad_norm": 0.001402844674885273,
-      "learning_rate": 0.00019999852529623528,
-      "loss": 46.0,
-      "step": 22620
-    },
-    {
-      "epoch": 1.7295334212588642,
-      "grad_norm": 0.0006659413920715451,
-      "learning_rate": 0.00019999852516578567,
-      "loss": 46.0,
-      "step": 22621
-    },
-    {
-      "epoch": 1.729609878242254,
-      "grad_norm": 0.00293273595161736,
-      "learning_rate": 0.00019999852503533032,
-      "loss": 46.0,
-      "step": 22622
-    },
-    {
-      "epoch": 1.7296863352256437,
-      "grad_norm": 0.0008309009135700762,
-      "learning_rate": 0.00019999852490486917,
-      "loss": 46.0,
-      "step": 22623
-    },
-    {
-      "epoch": 1.7297627922090335,
-      "grad_norm": 0.0005622395547106862,
-      "learning_rate": 0.00019999852477440227,
-      "loss": 46.0,
-      "step": 22624
-    },
-    {
-      "epoch": 1.7298392491924233,
-      "grad_norm": 0.0011021544924005866,
-      "learning_rate": 0.0001999985246439296,
-      "loss": 46.0,
-      "step": 22625
-    },
-    {
-      "epoch": 1.7299157061758128,
-      "grad_norm": 0.000948019849602133,
-      "learning_rate": 0.00019999852451345112,
-      "loss": 46.0,
-      "step": 22626
-    },
-    {
-      "epoch": 1.7299921631592026,
-      "grad_norm": 0.0019641867838799953,
-      "learning_rate": 0.0001999985243829669,
-      "loss": 46.0,
-      "step": 22627
-    },
-    {
-      "epoch": 1.7300686201425923,
-      "grad_norm": 0.0007797997677698731,
-      "learning_rate": 0.00019999852425247694,
-      "loss": 46.0,
-      "step": 22628
-    },
-    {
-      "epoch": 1.7301450771259819,
-      "grad_norm": 0.002256902866065502,
-      "learning_rate": 0.00019999852412198117,
-      "loss": 46.0,
-      "step": 22629
-    },
-    {
-      "epoch": 1.7302215341093716,
-      "grad_norm": 0.0006343925488181412,
-      "learning_rate": 0.00019999852399147963,
-      "loss": 46.0,
-      "step": 22630
-    },
-    {
-      "epoch": 1.7302979910927614,
-      "grad_norm": 0.0008882620604708791,
-      "learning_rate": 0.00019999852386097232,
-      "loss": 46.0,
-      "step": 22631
-    },
-    {
-      "epoch": 1.7303744480761512,
-      "grad_norm": 0.0006998606841079891,
-      "learning_rate": 0.00019999852373045926,
-      "loss": 46.0,
-      "step": 22632
-    },
-    {
-      "epoch": 1.730450905059541,
-      "grad_norm": 0.0008176497067324817,
-      "learning_rate": 0.00019999852359994046,
-      "loss": 46.0,
-      "step": 22633
-    },
-    {
-      "epoch": 1.7305273620429307,
-      "grad_norm": 0.000559588719625026,
-      "learning_rate": 0.00019999852346941585,
-      "loss": 46.0,
-      "step": 22634
-    },
-    {
-      "epoch": 1.7306038190263204,
-      "grad_norm": 0.0007625481812283397,
-      "learning_rate": 0.00019999852333888547,
-      "loss": 46.0,
-      "step": 22635
-    },
-    {
-      "epoch": 1.7306802760097102,
-      "grad_norm": 0.001884974422864616,
-      "learning_rate": 0.00019999852320834932,
-      "loss": 46.0,
-      "step": 22636
-    },
-    {
-      "epoch": 1.7307567329930997,
-      "grad_norm": 0.0005362986121326685,
-      "learning_rate": 0.00019999852307780742,
-      "loss": 46.0,
-      "step": 22637
-    },
-    {
-      "epoch": 1.7308331899764895,
-      "grad_norm": 0.0008835560292936862,
-      "learning_rate": 0.00019999852294725975,
-      "loss": 46.0,
-      "step": 22638
-    },
-    {
-      "epoch": 1.730909646959879,
-      "grad_norm": 0.0021581652108579874,
-      "learning_rate": 0.00019999852281670627,
-      "loss": 46.0,
-      "step": 22639
-    },
-    {
-      "epoch": 1.7309861039432688,
-      "grad_norm": 0.0010900796623900533,
-      "learning_rate": 0.00019999852268614708,
-      "loss": 46.0,
-      "step": 22640
-    },
-    {
-      "epoch": 1.7310625609266586,
-      "grad_norm": 0.0010500267380848527,
-      "learning_rate": 0.00019999852255558206,
-      "loss": 46.0,
-      "step": 22641
-    },
-    {
-      "epoch": 1.7311390179100483,
-      "grad_norm": 0.0011888942681252956,
-      "learning_rate": 0.00019999852242501132,
-      "loss": 46.0,
-      "step": 22642
-    },
-    {
-      "epoch": 1.731215474893438,
-      "grad_norm": 0.0010251005878672004,
-      "learning_rate": 0.00019999852229443478,
-      "loss": 46.0,
-      "step": 22643
-    },
-    {
-      "epoch": 1.7312919318768278,
-      "grad_norm": 0.0006767407176084816,
-      "learning_rate": 0.0001999985221638525,
-      "loss": 46.0,
-      "step": 22644
-    },
-    {
-      "epoch": 1.7313683888602176,
-      "grad_norm": 0.0017930514877662063,
-      "learning_rate": 0.00019999852203326444,
-      "loss": 46.0,
-      "step": 22645
-    },
-    {
-      "epoch": 1.7314448458436074,
-      "grad_norm": 0.0014192037051543593,
-      "learning_rate": 0.00019999852190267058,
-      "loss": 46.0,
-      "step": 22646
-    },
-    {
-      "epoch": 1.7315213028269971,
-      "grad_norm": 0.0012247315607964993,
-      "learning_rate": 0.00019999852177207098,
-      "loss": 46.0,
-      "step": 22647
-    },
-    {
-      "epoch": 1.7315977598103867,
-      "grad_norm": 0.002818105276674032,
-      "learning_rate": 0.00019999852164146562,
-      "loss": 46.0,
-      "step": 22648
-    },
-    {
-      "epoch": 1.7316742167937764,
-      "grad_norm": 0.0035968590527772903,
-      "learning_rate": 0.00019999852151085444,
-      "loss": 46.0,
-      "step": 22649
-    },
-    {
-      "epoch": 1.731750673777166,
-      "grad_norm": 0.0011192914098501205,
-      "learning_rate": 0.00019999852138023755,
-      "loss": 46.0,
-      "step": 22650
-    },
-    {
-      "epoch": 1.7318271307605557,
-      "grad_norm": 0.0007100970833562315,
-      "learning_rate": 0.00019999852124961487,
-      "loss": 46.0,
-      "step": 22651
-    },
-    {
-      "epoch": 1.7319035877439455,
-      "grad_norm": 0.0006707278080284595,
-      "learning_rate": 0.0001999985211189864,
-      "loss": 46.0,
-      "step": 22652
-    },
-    {
-      "epoch": 1.7319800447273352,
-      "grad_norm": 0.0016358091961592436,
-      "learning_rate": 0.00019999852098835218,
-      "loss": 46.0,
-      "step": 22653
-    },
-    {
-      "epoch": 1.732056501710725,
-      "grad_norm": 0.0017581508727744222,
-      "learning_rate": 0.0001999985208577122,
-      "loss": 46.0,
-      "step": 22654
-    },
-    {
-      "epoch": 1.7321329586941148,
-      "grad_norm": 0.002192989457398653,
-      "learning_rate": 0.00019999852072706645,
-      "loss": 46.0,
-      "step": 22655
-    },
-    {
-      "epoch": 1.7322094156775045,
-      "grad_norm": 0.005653319880366325,
-      "learning_rate": 0.0001999985205964149,
-      "loss": 46.0,
-      "step": 22656
-    },
-    {
-      "epoch": 1.7322858726608943,
-      "grad_norm": 0.0020329018589109182,
-      "learning_rate": 0.0001999985204657576,
-      "loss": 46.0,
-      "step": 22657
-    },
-    {
-      "epoch": 1.732362329644284,
-      "grad_norm": 0.01344548724591732,
-      "learning_rate": 0.00019999852033509454,
-      "loss": 46.0,
-      "step": 22658
-    },
-    {
-      "epoch": 1.7324387866276736,
-      "grad_norm": 0.002527801785618067,
-      "learning_rate": 0.0001999985202044257,
-      "loss": 46.0,
-      "step": 22659
-    },
-    {
-      "epoch": 1.7325152436110633,
-      "grad_norm": 0.000667993794195354,
-      "learning_rate": 0.00019999852007375108,
-      "loss": 46.0,
-      "step": 22660
-    },
-    {
-      "epoch": 1.7325917005944529,
-      "grad_norm": 0.0063514430075883865,
-      "learning_rate": 0.00019999851994307073,
-      "loss": 46.0,
-      "step": 22661
-    },
-    {
-      "epoch": 1.7326681575778426,
-      "grad_norm": 0.0037415847182273865,
-      "learning_rate": 0.00019999851981238455,
-      "loss": 46.0,
-      "step": 22662
-    },
-    {
-      "epoch": 1.7327446145612324,
-      "grad_norm": 0.0006423986051231623,
-      "learning_rate": 0.00019999851968169265,
-      "loss": 46.0,
-      "step": 22663
-    },
-    {
-      "epoch": 1.7328210715446222,
-      "grad_norm": 0.009163415059447289,
-      "learning_rate": 0.00019999851955099498,
-      "loss": 46.0,
-      "step": 22664
-    },
-    {
-      "epoch": 1.732897528528012,
-      "grad_norm": 0.000832437421195209,
-      "learning_rate": 0.00019999851942029153,
-      "loss": 46.0,
-      "step": 22665
-    },
-    {
-      "epoch": 1.7329739855114017,
-      "grad_norm": 0.004653178621083498,
-      "learning_rate": 0.0001999985192895823,
-      "loss": 46.0,
-      "step": 22666
-    },
-    {
-      "epoch": 1.7330504424947915,
-      "grad_norm": 0.002586974762380123,
-      "learning_rate": 0.0001999985191588673,
-      "loss": 46.0,
-      "step": 22667
-    },
-    {
-      "epoch": 1.7331268994781812,
-      "grad_norm": 0.0019834262784570456,
-      "learning_rate": 0.00019999851902814656,
-      "loss": 46.0,
-      "step": 22668
-    },
-    {
-      "epoch": 1.7332033564615708,
-      "grad_norm": 0.0011409041471779346,
-      "learning_rate": 0.00019999851889742002,
-      "loss": 46.0,
-      "step": 22669
-    },
-    {
-      "epoch": 1.7332798134449605,
-      "grad_norm": 0.0015969412634149194,
-      "learning_rate": 0.00019999851876668768,
-      "loss": 46.0,
-      "step": 22670
-    },
-    {
-      "epoch": 1.7333562704283503,
-      "grad_norm": 0.002990870736539364,
-      "learning_rate": 0.00019999851863594962,
-      "loss": 46.0,
-      "step": 22671
-    },
-    {
-      "epoch": 1.7334327274117398,
-      "grad_norm": 0.0015055029653012753,
-      "learning_rate": 0.0001999985185052058,
-      "loss": 46.0,
-      "step": 22672
-    },
-    {
-      "epoch": 1.7335091843951296,
-      "grad_norm": 0.0006832543876953423,
-      "learning_rate": 0.00019999851837445619,
-      "loss": 46.0,
-      "step": 22673
-    },
-    {
-      "epoch": 1.7335856413785193,
-      "grad_norm": 0.0020437815692275763,
-      "learning_rate": 0.0001999985182437008,
-      "loss": 46.0,
-      "step": 22674
-    },
-    {
-      "epoch": 1.733662098361909,
-      "grad_norm": 0.009566658176481724,
-      "learning_rate": 0.00019999851811293966,
-      "loss": 46.0,
-      "step": 22675
-    },
-    {
-      "epoch": 1.7337385553452989,
-      "grad_norm": 0.0012620704947039485,
-      "learning_rate": 0.00019999851798217276,
-      "loss": 46.0,
-      "step": 22676
-    },
-    {
-      "epoch": 1.7338150123286886,
-      "grad_norm": 0.0014410660369321704,
-      "learning_rate": 0.00019999851785140006,
-      "loss": 46.0,
-      "step": 22677
-    },
-    {
-      "epoch": 1.7338914693120784,
-      "grad_norm": 0.0013877763412892818,
-      "learning_rate": 0.0001999985177206216,
-      "loss": 46.0,
-      "step": 22678
-    },
-    {
-      "epoch": 1.7339679262954681,
-      "grad_norm": 0.0009364584111608565,
-      "learning_rate": 0.0001999985175898374,
-      "loss": 46.0,
-      "step": 22679
-    },
-    {
-      "epoch": 1.7340443832788577,
-      "grad_norm": 0.0032117620576173067,
-      "learning_rate": 0.00019999851745904738,
-      "loss": 46.0,
-      "step": 22680
-    },
-    {
-      "epoch": 1.7341208402622474,
-      "grad_norm": 0.0013126885751262307,
-      "learning_rate": 0.0001999985173282516,
-      "loss": 46.0,
-      "step": 22681
-    },
-    {
-      "epoch": 1.7341972972456372,
-      "grad_norm": 0.0009948459919542074,
-      "learning_rate": 0.00019999851719745007,
-      "loss": 46.0,
-      "step": 22682
-    },
-    {
-      "epoch": 1.7342737542290267,
-      "grad_norm": 0.001783974003046751,
-      "learning_rate": 0.0001999985170666428,
-      "loss": 46.0,
-      "step": 22683
-    },
-    {
-      "epoch": 1.7343502112124165,
-      "grad_norm": 0.0024737364146858454,
-      "learning_rate": 0.0001999985169358297,
-      "loss": 46.0,
-      "step": 22684
-    },
-    {
-      "epoch": 1.7344266681958063,
-      "grad_norm": 0.0004796576395165175,
-      "learning_rate": 0.00019999851680501087,
-      "loss": 46.0,
-      "step": 22685
-    },
-    {
-      "epoch": 1.734503125179196,
-      "grad_norm": 0.0012784480350092053,
-      "learning_rate": 0.00019999851667418624,
-      "loss": 46.0,
-      "step": 22686
-    },
-    {
-      "epoch": 1.7345795821625858,
-      "grad_norm": 0.000598890008404851,
-      "learning_rate": 0.00019999851654335586,
-      "loss": 46.0,
-      "step": 22687
-    },
-    {
-      "epoch": 1.7346560391459755,
-      "grad_norm": 0.0007786693749949336,
-      "learning_rate": 0.0001999985164125197,
-      "loss": 46.0,
-      "step": 22688
-    },
-    {
-      "epoch": 1.7347324961293653,
-      "grad_norm": 0.0007252361974678934,
-      "learning_rate": 0.00019999851628167781,
-      "loss": 46.0,
-      "step": 22689
-    },
-    {
-      "epoch": 1.734808953112755,
-      "grad_norm": 0.004248403944075108,
-      "learning_rate": 0.00019999851615083012,
-      "loss": 46.0,
-      "step": 22690
-    },
-    {
-      "epoch": 1.7348854100961446,
-      "grad_norm": 0.010970363393425941,
-      "learning_rate": 0.00019999851601997665,
-      "loss": 46.0,
-      "step": 22691
-    },
-    {
-      "epoch": 1.7349618670795344,
-      "grad_norm": 0.001021255855448544,
-      "learning_rate": 0.00019999851588911743,
-      "loss": 46.0,
-      "step": 22692
-    },
-    {
-      "epoch": 1.7350383240629241,
-      "grad_norm": 0.0009164549992419779,
-      "learning_rate": 0.0001999985157582524,
-      "loss": 46.0,
-      "step": 22693
-    },
-    {
-      "epoch": 1.7351147810463137,
-      "grad_norm": 0.0006815766100771725,
-      "learning_rate": 0.00019999851562738167,
-      "loss": 46.0,
-      "step": 22694
-    },
-    {
-      "epoch": 1.7351912380297034,
-      "grad_norm": 0.007553113624453545,
-      "learning_rate": 0.00019999851549650513,
-      "loss": 46.0,
-      "step": 22695
-    },
-    {
-      "epoch": 1.7352676950130932,
-      "grad_norm": 0.0015128928935155272,
-      "learning_rate": 0.00019999851536562285,
-      "loss": 46.0,
-      "step": 22696
-    },
-    {
-      "epoch": 1.735344151996483,
-      "grad_norm": 0.0016350175719708204,
-      "learning_rate": 0.00019999851523473474,
-      "loss": 46.0,
-      "step": 22697
-    },
-    {
-      "epoch": 1.7354206089798727,
-      "grad_norm": 0.006948954425752163,
-      "learning_rate": 0.0001999985151038409,
-      "loss": 46.0,
-      "step": 22698
-    },
-    {
-      "epoch": 1.7354970659632625,
-      "grad_norm": 0.0008353106677532196,
-      "learning_rate": 0.0001999985149729413,
-      "loss": 46.0,
-      "step": 22699
-    },
-    {
-      "epoch": 1.7355735229466522,
-      "grad_norm": 0.00039935094537213445,
-      "learning_rate": 0.0001999985148420359,
-      "loss": 46.0,
-      "step": 22700
-    },
-    {
-      "epoch": 1.735649979930042,
-      "grad_norm": 0.0009470641962252557,
-      "learning_rate": 0.00019999851471112477,
-      "loss": 46.0,
-      "step": 22701
-    },
-    {
-      "epoch": 1.7357264369134315,
-      "grad_norm": 0.003697985550388694,
-      "learning_rate": 0.00019999851458020785,
-      "loss": 46.0,
-      "step": 22702
-    },
-    {
-      "epoch": 1.7358028938968213,
-      "grad_norm": 0.0004675001255236566,
-      "learning_rate": 0.00019999851444928515,
-      "loss": 46.0,
-      "step": 22703
-    },
-    {
-      "epoch": 1.735879350880211,
-      "grad_norm": 0.0009979365859180689,
-      "learning_rate": 0.0001999985143183567,
-      "loss": 46.0,
-      "step": 22704
-    },
-    {
-      "epoch": 1.7359558078636006,
-      "grad_norm": 0.0005944758886471391,
-      "learning_rate": 0.00019999851418742246,
-      "loss": 46.0,
-      "step": 22705
-    },
-    {
-      "epoch": 1.7360322648469904,
-      "grad_norm": 0.0012176495511084795,
-      "learning_rate": 0.00019999851405648247,
-      "loss": 46.0,
-      "step": 22706
-    },
-    {
-      "epoch": 1.7361087218303801,
-      "grad_norm": 0.0009234109311364591,
-      "learning_rate": 0.0001999985139255367,
-      "loss": 46.0,
-      "step": 22707
-    },
-    {
-      "epoch": 1.7361851788137699,
-      "grad_norm": 0.0022533521987497807,
-      "learning_rate": 0.00019999851379458514,
-      "loss": 46.0,
-      "step": 22708
-    },
-    {
-      "epoch": 1.7362616357971596,
-      "grad_norm": 0.003385854884982109,
-      "learning_rate": 0.00019999851366362786,
-      "loss": 46.0,
-      "step": 22709
-    },
-    {
-      "epoch": 1.7363380927805494,
-      "grad_norm": 0.0017720299074426293,
-      "learning_rate": 0.00019999851353266478,
-      "loss": 46.0,
-      "step": 22710
-    },
-    {
-      "epoch": 1.7364145497639392,
-      "grad_norm": 0.001349471160210669,
-      "learning_rate": 0.00019999851340169592,
-      "loss": 46.0,
-      "step": 22711
-    },
-    {
-      "epoch": 1.736491006747329,
-      "grad_norm": 0.0009567879023961723,
-      "learning_rate": 0.00019999851327072129,
-      "loss": 46.0,
-      "step": 22712
-    },
-    {
-      "epoch": 1.7365674637307185,
-      "grad_norm": 0.002546332310885191,
-      "learning_rate": 0.0001999985131397409,
-      "loss": 46.0,
-      "step": 22713
-    },
-    {
-      "epoch": 1.7366439207141082,
-      "grad_norm": 0.0012086735805496573,
-      "learning_rate": 0.00019999851300875476,
-      "loss": 46.0,
-      "step": 22714
-    },
-    {
-      "epoch": 1.736720377697498,
-      "grad_norm": 0.0005902419798076153,
-      "learning_rate": 0.00019999851287776286,
-      "loss": 46.0,
-      "step": 22715
-    },
-    {
-      "epoch": 1.7367968346808875,
-      "grad_norm": 0.0007699631969444454,
-      "learning_rate": 0.00019999851274676516,
-      "loss": 46.0,
-      "step": 22716
-    },
-    {
-      "epoch": 1.7368732916642773,
-      "grad_norm": 0.004581313580274582,
-      "learning_rate": 0.0001999985126157617,
-      "loss": 46.0,
-      "step": 22717
-    },
-    {
-      "epoch": 1.736949748647667,
-      "grad_norm": 0.0004327432543504983,
-      "learning_rate": 0.00019999851248475245,
-      "loss": 46.0,
-      "step": 22718
-    },
-    {
-      "epoch": 1.7370262056310568,
-      "grad_norm": 0.0006114005809649825,
-      "learning_rate": 0.00019999851235373749,
-      "loss": 46.0,
-      "step": 22719
-    },
-    {
-      "epoch": 1.7371026626144466,
-      "grad_norm": 0.0005895322537980974,
-      "learning_rate": 0.0001999985122227167,
-      "loss": 46.0,
-      "step": 22720
-    },
-    {
-      "epoch": 1.7371791195978363,
-      "grad_norm": 0.0008027162402868271,
-      "learning_rate": 0.00019999851209169016,
-      "loss": 46.0,
-      "step": 22721
-    },
-    {
-      "epoch": 1.737255576581226,
-      "grad_norm": 0.0018236652249470353,
-      "learning_rate": 0.00019999851196065787,
-      "loss": 46.0,
-      "step": 22722
-    },
-    {
-      "epoch": 1.7373320335646159,
-      "grad_norm": 0.0018301403615623713,
-      "learning_rate": 0.00019999851182961976,
-      "loss": 46.0,
-      "step": 22723
-    },
-    {
-      "epoch": 1.7374084905480054,
-      "grad_norm": 0.0003246336127631366,
-      "learning_rate": 0.00019999851169857593,
-      "loss": 46.0,
-      "step": 22724
-    },
-    {
-      "epoch": 1.7374849475313952,
-      "grad_norm": 0.000664866587612778,
-      "learning_rate": 0.0001999985115675263,
-      "loss": 46.0,
-      "step": 22725
-    },
-    {
-      "epoch": 1.737561404514785,
-      "grad_norm": 0.0006422539008781314,
-      "learning_rate": 0.00019999851143647092,
-      "loss": 46.0,
-      "step": 22726
-    },
-    {
-      "epoch": 1.7376378614981745,
-      "grad_norm": 0.0055303629487752914,
-      "learning_rate": 0.00019999851130540977,
-      "loss": 46.0,
-      "step": 22727
-    },
-    {
-      "epoch": 1.7377143184815642,
-      "grad_norm": 0.00352671486325562,
-      "learning_rate": 0.00019999851117434285,
-      "loss": 46.0,
-      "step": 22728
-    },
-    {
-      "epoch": 1.737790775464954,
-      "grad_norm": 0.0010309522040188313,
-      "learning_rate": 0.00019999851104327015,
-      "loss": 46.0,
-      "step": 22729
-    },
-    {
-      "epoch": 1.7378672324483437,
-      "grad_norm": 0.0019963423255831003,
-      "learning_rate": 0.00019999851091219168,
-      "loss": 46.0,
-      "step": 22730
-    },
-    {
-      "epoch": 1.7379436894317335,
-      "grad_norm": 0.0005471197655424476,
-      "learning_rate": 0.00019999851078110744,
-      "loss": 46.0,
-      "step": 22731
-    },
-    {
-      "epoch": 1.7380201464151233,
-      "grad_norm": 0.0015014695236459374,
-      "learning_rate": 0.00019999851065001747,
-      "loss": 46.0,
-      "step": 22732
-    },
-    {
-      "epoch": 1.738096603398513,
-      "grad_norm": 0.0008950633346103132,
-      "learning_rate": 0.00019999851051892166,
-      "loss": 46.0,
-      "step": 22733
-    },
-    {
-      "epoch": 1.7381730603819028,
-      "grad_norm": 0.001532626454718411,
-      "learning_rate": 0.00019999851038782015,
-      "loss": 46.0,
-      "step": 22734
-    },
-    {
-      "epoch": 1.7382495173652923,
-      "grad_norm": 0.0011636392446234822,
-      "learning_rate": 0.00019999851025671284,
-      "loss": 46.0,
-      "step": 22735
-    },
-    {
-      "epoch": 1.738325974348682,
-      "grad_norm": 0.0015476715052500367,
-      "learning_rate": 0.00019999851012559975,
-      "loss": 46.0,
-      "step": 22736
-    },
-    {
-      "epoch": 1.7384024313320718,
-      "grad_norm": 0.0028576315380632877,
-      "learning_rate": 0.0001999985099944809,
-      "loss": 46.0,
-      "step": 22737
-    },
-    {
-      "epoch": 1.7384788883154614,
-      "grad_norm": 0.001266071922145784,
-      "learning_rate": 0.0001999985098633563,
-      "loss": 46.0,
-      "step": 22738
-    },
-    {
-      "epoch": 1.7385553452988511,
-      "grad_norm": 0.0009416180546395481,
-      "learning_rate": 0.0001999985097322259,
-      "loss": 46.0,
-      "step": 22739
-    },
-    {
-      "epoch": 1.738631802282241,
-      "grad_norm": 0.0014856200432404876,
-      "learning_rate": 0.00019999850960108974,
-      "loss": 46.0,
-      "step": 22740
-    },
-    {
-      "epoch": 1.7387082592656307,
-      "grad_norm": 0.0008043196867220104,
-      "learning_rate": 0.00019999850946994782,
-      "loss": 46.0,
-      "step": 22741
-    },
-    {
-      "epoch": 1.7387847162490204,
-      "grad_norm": 0.000951383204665035,
-      "learning_rate": 0.00019999850933880012,
-      "loss": 46.0,
-      "step": 22742
-    },
-    {
-      "epoch": 1.7388611732324102,
-      "grad_norm": 0.0018324218690395355,
-      "learning_rate": 0.00019999850920764668,
-      "loss": 46.0,
-      "step": 22743
-    },
-    {
-      "epoch": 1.7389376302158,
-      "grad_norm": 0.002452875254675746,
-      "learning_rate": 0.0001999985090764874,
-      "loss": 46.0,
-      "step": 22744
-    },
-    {
-      "epoch": 1.7390140871991897,
-      "grad_norm": 0.004986785817891359,
-      "learning_rate": 0.00019999850894532241,
-      "loss": 46.0,
-      "step": 22745
-    },
-    {
-      "epoch": 1.7390905441825792,
-      "grad_norm": 0.0012780589750036597,
-      "learning_rate": 0.00019999850881415165,
-      "loss": 46.0,
-      "step": 22746
-    },
-    {
-      "epoch": 1.739167001165969,
-      "grad_norm": 0.0015120753087103367,
-      "learning_rate": 0.00019999850868297511,
-      "loss": 46.0,
-      "step": 22747
-    },
-    {
-      "epoch": 1.7392434581493588,
-      "grad_norm": 0.0019318959675729275,
-      "learning_rate": 0.00019999850855179278,
-      "loss": 46.0,
-      "step": 22748
-    },
-    {
-      "epoch": 1.7393199151327483,
-      "grad_norm": 0.0015932531096041203,
-      "learning_rate": 0.0001999985084206047,
-      "loss": 46.0,
-      "step": 22749
-    },
-    {
-      "epoch": 1.739396372116138,
-      "grad_norm": 0.00028493956779129803,
-      "learning_rate": 0.00019999850828941086,
-      "loss": 46.0,
-      "step": 22750
-    },
-    {
-      "epoch": 1.7394728290995278,
-      "grad_norm": 0.0009779352694749832,
-      "learning_rate": 0.00019999850815821123,
-      "loss": 46.0,
-      "step": 22751
-    },
-    {
-      "epoch": 1.7395492860829176,
-      "grad_norm": 0.0015636475291103125,
-      "learning_rate": 0.00019999850802700583,
-      "loss": 46.0,
-      "step": 22752
-    },
-    {
-      "epoch": 1.7396257430663074,
-      "grad_norm": 0.002212472492828965,
-      "learning_rate": 0.00019999850789579468,
-      "loss": 46.0,
-      "step": 22753
-    },
-    {
-      "epoch": 1.7397022000496971,
-      "grad_norm": 0.001162045169621706,
-      "learning_rate": 0.00019999850776457776,
-      "loss": 46.0,
-      "step": 22754
-    },
-    {
-      "epoch": 1.7397786570330869,
-      "grad_norm": 0.013619589619338512,
-      "learning_rate": 0.00019999850763335506,
-      "loss": 46.0,
-      "step": 22755
-    },
-    {
-      "epoch": 1.7398551140164766,
-      "grad_norm": 0.003062523202970624,
-      "learning_rate": 0.0001999985075021266,
-      "loss": 46.0,
-      "step": 22756
-    },
-    {
-      "epoch": 1.7399315709998662,
-      "grad_norm": 0.0007701321155764163,
-      "learning_rate": 0.00019999850737089238,
-      "loss": 46.0,
-      "step": 22757
-    },
-    {
-      "epoch": 1.740008027983256,
-      "grad_norm": 0.009220766834914684,
-      "learning_rate": 0.00019999850723965236,
-      "loss": 46.0,
-      "step": 22758
-    },
-    {
-      "epoch": 1.7400844849666457,
-      "grad_norm": 0.0037550898268818855,
-      "learning_rate": 0.00019999850710840657,
-      "loss": 46.0,
-      "step": 22759
-    },
-    {
-      "epoch": 1.7401609419500352,
-      "grad_norm": 0.0007332914974540472,
-      "learning_rate": 0.00019999850697715503,
-      "loss": 46.0,
-      "step": 22760
-    },
-    {
-      "epoch": 1.740237398933425,
-      "grad_norm": 0.0005844080587849021,
-      "learning_rate": 0.00019999850684589775,
-      "loss": 46.0,
-      "step": 22761
-    },
-    {
-      "epoch": 1.7403138559168148,
-      "grad_norm": 0.0007936040055938065,
-      "learning_rate": 0.00019999850671463464,
-      "loss": 46.0,
-      "step": 22762
-    },
-    {
-      "epoch": 1.7403903129002045,
-      "grad_norm": 0.00065904128132388,
-      "learning_rate": 0.00019999850658336578,
-      "loss": 46.0,
-      "step": 22763
-    },
-    {
-      "epoch": 1.7404667698835943,
-      "grad_norm": 0.0003065250057261437,
-      "learning_rate": 0.00019999850645209115,
-      "loss": 46.0,
-      "step": 22764
-    },
-    {
-      "epoch": 1.740543226866984,
-      "grad_norm": 0.0008610678487457335,
-      "learning_rate": 0.00019999850632081078,
-      "loss": 46.0,
-      "step": 22765
-    },
-    {
-      "epoch": 1.7406196838503738,
-      "grad_norm": 0.0008119586855173111,
-      "learning_rate": 0.0001999985061895246,
-      "loss": 46.0,
-      "step": 22766
-    },
-    {
-      "epoch": 1.7406961408337636,
-      "grad_norm": 0.0013643893180415034,
-      "learning_rate": 0.0001999985060582327,
-      "loss": 46.0,
-      "step": 22767
-    },
-    {
-      "epoch": 1.740772597817153,
-      "grad_norm": 0.0010365574853494763,
-      "learning_rate": 0.00019999850592693498,
-      "loss": 46.0,
-      "step": 22768
-    },
-    {
-      "epoch": 1.7408490548005429,
-      "grad_norm": 0.0007099217618815601,
-      "learning_rate": 0.0001999985057956315,
-      "loss": 46.0,
-      "step": 22769
-    },
-    {
-      "epoch": 1.7409255117839324,
-      "grad_norm": 0.0008160346769727767,
-      "learning_rate": 0.00019999850566432227,
-      "loss": 46.0,
-      "step": 22770
-    },
-    {
-      "epoch": 1.7410019687673222,
-      "grad_norm": 0.002936852630227804,
-      "learning_rate": 0.00019999850553300728,
-      "loss": 46.0,
-      "step": 22771
-    },
-    {
-      "epoch": 1.741078425750712,
-      "grad_norm": 0.000620995881035924,
-      "learning_rate": 0.0001999985054016865,
-      "loss": 46.0,
-      "step": 22772
-    },
-    {
-      "epoch": 1.7411548827341017,
-      "grad_norm": 0.001320533687248826,
-      "learning_rate": 0.00019999850527035995,
-      "loss": 46.0,
-      "step": 22773
-    },
-    {
-      "epoch": 1.7412313397174914,
-      "grad_norm": 0.000999670708552003,
-      "learning_rate": 0.00019999850513902762,
-      "loss": 46.0,
-      "step": 22774
-    },
-    {
-      "epoch": 1.7413077967008812,
-      "grad_norm": 0.0026237950660288334,
-      "learning_rate": 0.00019999850500768953,
-      "loss": 46.0,
-      "step": 22775
-    },
-    {
-      "epoch": 1.741384253684271,
-      "grad_norm": 0.0025458999443799257,
-      "learning_rate": 0.0001999985048763457,
-      "loss": 46.0,
-      "step": 22776
-    },
-    {
-      "epoch": 1.7414607106676607,
-      "grad_norm": 0.0011643661418929696,
-      "learning_rate": 0.00019999850474499608,
-      "loss": 46.0,
-      "step": 22777
-    },
-    {
-      "epoch": 1.7415371676510505,
-      "grad_norm": 0.00045163059257902205,
-      "learning_rate": 0.0001999985046136407,
-      "loss": 46.0,
-      "step": 22778
-    },
-    {
-      "epoch": 1.74161362463444,
-      "grad_norm": 0.0009960890747606754,
-      "learning_rate": 0.0001999985044822795,
-      "loss": 46.0,
-      "step": 22779
-    },
-    {
-      "epoch": 1.7416900816178298,
-      "grad_norm": 0.001297186827287078,
-      "learning_rate": 0.00019999850435091257,
-      "loss": 46.0,
-      "step": 22780
-    },
-    {
-      "epoch": 1.7417665386012193,
-      "grad_norm": 0.0013079263735562563,
-      "learning_rate": 0.0001999985042195399,
-      "loss": 46.0,
-      "step": 22781
-    },
-    {
-      "epoch": 1.741842995584609,
-      "grad_norm": 0.0013282245490700006,
-      "learning_rate": 0.0001999985040881614,
-      "loss": 46.0,
-      "step": 22782
-    },
-    {
-      "epoch": 1.7419194525679988,
-      "grad_norm": 0.0006855812389403582,
-      "learning_rate": 0.00019999850395677717,
-      "loss": 46.0,
-      "step": 22783
-    },
-    {
-      "epoch": 1.7419959095513886,
-      "grad_norm": 0.013218983076512814,
-      "learning_rate": 0.00019999850382538715,
-      "loss": 46.0,
-      "step": 22784
-    },
-    {
-      "epoch": 1.7420723665347784,
-      "grad_norm": 0.0005991904763504863,
-      "learning_rate": 0.0001999985036939914,
-      "loss": 46.0,
-      "step": 22785
-    },
-    {
-      "epoch": 1.7421488235181681,
-      "grad_norm": 0.0018436353420838714,
-      "learning_rate": 0.00019999850356258985,
-      "loss": 46.0,
-      "step": 22786
-    },
-    {
-      "epoch": 1.742225280501558,
-      "grad_norm": 0.0019221490947529674,
-      "learning_rate": 0.00019999850343118252,
-      "loss": 46.0,
-      "step": 22787
-    },
-    {
-      "epoch": 1.7423017374849477,
-      "grad_norm": 0.001245797029696405,
-      "learning_rate": 0.00019999850329976943,
-      "loss": 46.0,
-      "step": 22788
-    },
-    {
-      "epoch": 1.7423781944683374,
-      "grad_norm": 0.002214134903624654,
-      "learning_rate": 0.00019999850316835058,
-      "loss": 46.0,
-      "step": 22789
-    },
-    {
-      "epoch": 1.742454651451727,
-      "grad_norm": 0.0006869768840260804,
-      "learning_rate": 0.00019999850303692595,
-      "loss": 46.0,
-      "step": 22790
-    },
-    {
-      "epoch": 1.7425311084351167,
-      "grad_norm": 0.0008450874011032283,
-      "learning_rate": 0.00019999850290549555,
-      "loss": 46.0,
-      "step": 22791
-    },
-    {
-      "epoch": 1.7426075654185063,
-      "grad_norm": 0.0008311583078466356,
-      "learning_rate": 0.0001999985027740594,
-      "loss": 46.0,
-      "step": 22792
-    },
-    {
-      "epoch": 1.742684022401896,
-      "grad_norm": 0.0007313786773011088,
-      "learning_rate": 0.00019999850264261745,
-      "loss": 46.0,
-      "step": 22793
-    },
-    {
-      "epoch": 1.7427604793852858,
-      "grad_norm": 0.0015601444756612182,
-      "learning_rate": 0.00019999850251116975,
-      "loss": 46.0,
-      "step": 22794
-    },
-    {
-      "epoch": 1.7428369363686755,
-      "grad_norm": 0.0024768984876573086,
-      "learning_rate": 0.00019999850237971626,
-      "loss": 46.0,
-      "step": 22795
-    },
-    {
-      "epoch": 1.7429133933520653,
-      "grad_norm": 0.000522041053045541,
-      "learning_rate": 0.00019999850224825701,
-      "loss": 46.0,
-      "step": 22796
-    },
-    {
-      "epoch": 1.742989850335455,
-      "grad_norm": 0.0008263241034001112,
-      "learning_rate": 0.00019999850211679202,
-      "loss": 46.0,
-      "step": 22797
-    },
-    {
-      "epoch": 1.7430663073188448,
-      "grad_norm": 0.001271361019462347,
-      "learning_rate": 0.00019999850198532126,
-      "loss": 46.0,
-      "step": 22798
-    },
-    {
-      "epoch": 1.7431427643022346,
-      "grad_norm": 0.001528938882984221,
-      "learning_rate": 0.0001999985018538447,
-      "loss": 46.0,
-      "step": 22799
-    },
-    {
-      "epoch": 1.7432192212856241,
-      "grad_norm": 0.000709356798324734,
-      "learning_rate": 0.0001999985017223624,
-      "loss": 46.0,
-      "step": 22800
-    },
-    {
-      "epoch": 1.7432956782690139,
-      "grad_norm": 0.002937842858955264,
-      "learning_rate": 0.0001999985015908743,
-      "loss": 46.0,
-      "step": 22801
-    },
-    {
-      "epoch": 1.7433721352524036,
-      "grad_norm": 0.0006658356287516654,
-      "learning_rate": 0.00019999850145938043,
-      "loss": 46.0,
-      "step": 22802
-    },
-    {
-      "epoch": 1.7434485922357932,
-      "grad_norm": 0.0014967272290959954,
-      "learning_rate": 0.0001999985013278808,
-      "loss": 46.0,
-      "step": 22803
-    },
-    {
-      "epoch": 1.743525049219183,
-      "grad_norm": 0.0020940289832651615,
-      "learning_rate": 0.00019999850119637542,
-      "loss": 46.0,
-      "step": 22804
-    },
-    {
-      "epoch": 1.7436015062025727,
-      "grad_norm": 0.0007779986481182277,
-      "learning_rate": 0.00019999850106486425,
-      "loss": 46.0,
-      "step": 22805
-    },
-    {
-      "epoch": 1.7436779631859625,
-      "grad_norm": 0.0008359465282410383,
-      "learning_rate": 0.0001999985009333473,
-      "loss": 46.0,
-      "step": 22806
-    },
-    {
-      "epoch": 1.7437544201693522,
-      "grad_norm": 0.0016202996484935284,
-      "learning_rate": 0.0001999985008018246,
-      "loss": 46.0,
-      "step": 22807
-    },
-    {
-      "epoch": 1.743830877152742,
-      "grad_norm": 0.000525411160197109,
-      "learning_rate": 0.00019999850067029613,
-      "loss": 46.0,
-      "step": 22808
-    },
-    {
-      "epoch": 1.7439073341361317,
-      "grad_norm": 0.0006692411843687296,
-      "learning_rate": 0.0001999985005387619,
-      "loss": 46.0,
-      "step": 22809
-    },
-    {
-      "epoch": 1.7439837911195215,
-      "grad_norm": 0.0014965205918997526,
-      "learning_rate": 0.00019999850040722188,
-      "loss": 46.0,
-      "step": 22810
-    },
-    {
-      "epoch": 1.744060248102911,
-      "grad_norm": 0.0017924440326169133,
-      "learning_rate": 0.0001999985002756761,
-      "loss": 46.0,
-      "step": 22811
-    },
-    {
-      "epoch": 1.7441367050863008,
-      "grad_norm": 0.0030297625344246626,
-      "learning_rate": 0.00019999850014412455,
-      "loss": 46.0,
-      "step": 22812
-    },
-    {
-      "epoch": 1.7442131620696906,
-      "grad_norm": 0.0009215819300152361,
-      "learning_rate": 0.00019999850001256722,
-      "loss": 46.0,
-      "step": 22813
-    },
-    {
-      "epoch": 1.74428961905308,
-      "grad_norm": 0.00822428334504366,
-      "learning_rate": 0.00019999849988100414,
-      "loss": 46.0,
-      "step": 22814
-    },
-    {
-      "epoch": 1.7443660760364699,
-      "grad_norm": 0.0009410441271029413,
-      "learning_rate": 0.00019999849974943528,
-      "loss": 46.0,
-      "step": 22815
-    },
-    {
-      "epoch": 1.7444425330198596,
-      "grad_norm": 0.0010528520215302706,
-      "learning_rate": 0.00019999849961786062,
-      "loss": 46.0,
-      "step": 22816
-    },
-    {
-      "epoch": 1.7445189900032494,
-      "grad_norm": 0.001200356287881732,
-      "learning_rate": 0.00019999849948628025,
-      "loss": 46.0,
-      "step": 22817
-    },
-    {
-      "epoch": 1.7445954469866392,
-      "grad_norm": 0.00629137409850955,
-      "learning_rate": 0.00019999849935469407,
-      "loss": 46.0,
-      "step": 22818
-    },
-    {
-      "epoch": 1.744671903970029,
-      "grad_norm": 0.0025854301638901234,
-      "learning_rate": 0.00019999849922310215,
-      "loss": 46.0,
-      "step": 22819
-    },
-    {
-      "epoch": 1.7447483609534187,
-      "grad_norm": 0.001349236466921866,
-      "learning_rate": 0.00019999849909150443,
-      "loss": 46.0,
-      "step": 22820
-    },
-    {
-      "epoch": 1.7448248179368084,
-      "grad_norm": 0.0007376992725767195,
-      "learning_rate": 0.00019999849895990094,
-      "loss": 46.0,
-      "step": 22821
-    },
-    {
-      "epoch": 1.744901274920198,
-      "grad_norm": 0.0024189534597098827,
-      "learning_rate": 0.00019999849882829172,
-      "loss": 46.0,
-      "step": 22822
-    },
-    {
-      "epoch": 1.7449777319035877,
-      "grad_norm": 0.001447937567718327,
-      "learning_rate": 0.0001999984986966767,
-      "loss": 46.0,
-      "step": 22823
-    },
-    {
-      "epoch": 1.7450541888869775,
-      "grad_norm": 0.001649067155085504,
-      "learning_rate": 0.0001999984985650559,
-      "loss": 46.0,
-      "step": 22824
-    },
-    {
-      "epoch": 1.745130645870367,
-      "grad_norm": 0.0008314444567076862,
-      "learning_rate": 0.00019999849843342938,
-      "loss": 46.0,
-      "step": 22825
-    },
-    {
-      "epoch": 1.7452071028537568,
-      "grad_norm": 0.0012076111743226647,
-      "learning_rate": 0.00019999849830179705,
-      "loss": 46.0,
-      "step": 22826
-    },
-    {
-      "epoch": 1.7452835598371466,
-      "grad_norm": 0.003345315344631672,
-      "learning_rate": 0.00019999849817015894,
-      "loss": 46.0,
-      "step": 22827
-    },
-    {
-      "epoch": 1.7453600168205363,
-      "grad_norm": 0.000997675582766533,
-      "learning_rate": 0.00019999849803851508,
-      "loss": 46.0,
-      "step": 22828
-    },
-    {
-      "epoch": 1.745436473803926,
-      "grad_norm": 0.0014490509638562799,
-      "learning_rate": 0.00019999849790686546,
-      "loss": 46.0,
-      "step": 22829
-    },
-    {
-      "epoch": 1.7455129307873158,
-      "grad_norm": 0.0012346961302682757,
-      "learning_rate": 0.00019999849777521008,
-      "loss": 46.0,
-      "step": 22830
-    },
-    {
-      "epoch": 1.7455893877707056,
-      "grad_norm": 0.0004516993649303913,
-      "learning_rate": 0.0001999984976435489,
-      "loss": 46.0,
-      "step": 22831
-    },
-    {
-      "epoch": 1.7456658447540954,
-      "grad_norm": 0.0026130510959774256,
-      "learning_rate": 0.00019999849751188196,
-      "loss": 46.0,
-      "step": 22832
-    },
-    {
-      "epoch": 1.745742301737485,
-      "grad_norm": 0.0012400804553180933,
-      "learning_rate": 0.00019999849738020924,
-      "loss": 46.0,
-      "step": 22833
-    },
-    {
-      "epoch": 1.7458187587208747,
-      "grad_norm": 0.0015285499393939972,
-      "learning_rate": 0.00019999849724853074,
-      "loss": 46.0,
-      "step": 22834
-    },
-    {
-      "epoch": 1.7458952157042644,
-      "grad_norm": 0.0011443976545706391,
-      "learning_rate": 0.00019999849711684653,
-      "loss": 46.0,
-      "step": 22835
-    },
-    {
-      "epoch": 1.745971672687654,
-      "grad_norm": 0.0007074170862324536,
-      "learning_rate": 0.00019999849698515648,
-      "loss": 46.0,
-      "step": 22836
-    },
-    {
-      "epoch": 1.7460481296710437,
-      "grad_norm": 0.0034424017649143934,
-      "learning_rate": 0.00019999849685346072,
-      "loss": 46.0,
-      "step": 22837
-    },
-    {
-      "epoch": 1.7461245866544335,
-      "grad_norm": 0.0018888873746618629,
-      "learning_rate": 0.00019999849672175913,
-      "loss": 46.0,
-      "step": 22838
-    },
-    {
-      "epoch": 1.7462010436378232,
-      "grad_norm": 0.0008307318203151226,
-      "learning_rate": 0.00019999849659005183,
-      "loss": 46.0,
-      "step": 22839
-    },
-    {
-      "epoch": 1.746277500621213,
-      "grad_norm": 0.0007107091369107366,
-      "learning_rate": 0.00019999849645833875,
-      "loss": 46.0,
-      "step": 22840
-    },
-    {
-      "epoch": 1.7463539576046028,
-      "grad_norm": 0.0010909700067713857,
-      "learning_rate": 0.0001999984963266199,
-      "loss": 46.0,
-      "step": 22841
-    },
-    {
-      "epoch": 1.7464304145879925,
-      "grad_norm": 0.001032295753248036,
-      "learning_rate": 0.00019999849619489526,
-      "loss": 46.0,
-      "step": 22842
-    },
-    {
-      "epoch": 1.7465068715713823,
-      "grad_norm": 0.0051668556407094,
-      "learning_rate": 0.00019999849606316486,
-      "loss": 46.0,
-      "step": 22843
-    },
-    {
-      "epoch": 1.7465833285547718,
-      "grad_norm": 0.0015473958337679505,
-      "learning_rate": 0.0001999984959314287,
-      "loss": 46.0,
-      "step": 22844
-    },
-    {
-      "epoch": 1.7466597855381616,
-      "grad_norm": 0.0013182054972276092,
-      "learning_rate": 0.0001999984957996867,
-      "loss": 46.0,
-      "step": 22845
-    },
-    {
-      "epoch": 1.7467362425215514,
-      "grad_norm": 0.0014130270574241877,
-      "learning_rate": 0.00019999849566793902,
-      "loss": 46.0,
-      "step": 22846
-    },
-    {
-      "epoch": 1.746812699504941,
-      "grad_norm": 0.0007180745014920831,
-      "learning_rate": 0.00019999849553618555,
-      "loss": 46.0,
-      "step": 22847
-    },
-    {
-      "epoch": 1.7468891564883307,
-      "grad_norm": 0.0026254537515342236,
-      "learning_rate": 0.00019999849540442628,
-      "loss": 46.0,
-      "step": 22848
-    },
-    {
-      "epoch": 1.7469656134717204,
-      "grad_norm": 0.0009421589784324169,
-      "learning_rate": 0.00019999849527266127,
-      "loss": 46.0,
-      "step": 22849
-    },
-    {
-      "epoch": 1.7470420704551102,
-      "grad_norm": 0.0006869077915325761,
-      "learning_rate": 0.00019999849514089048,
-      "loss": 46.0,
-      "step": 22850
-    },
-    {
-      "epoch": 1.7471185274385,
-      "grad_norm": 0.0007505064131692052,
-      "learning_rate": 0.00019999849500911392,
-      "loss": 46.0,
-      "step": 22851
-    },
-    {
-      "epoch": 1.7471949844218897,
-      "grad_norm": 0.0004789851373061538,
-      "learning_rate": 0.0001999984948773316,
-      "loss": 46.0,
-      "step": 22852
-    },
-    {
-      "epoch": 1.7472714414052795,
-      "grad_norm": 0.0008028162992559373,
-      "learning_rate": 0.0001999984947455435,
-      "loss": 46.0,
-      "step": 22853
-    },
-    {
-      "epoch": 1.7473478983886692,
-      "grad_norm": 0.0013775058323517442,
-      "learning_rate": 0.00019999849461374965,
-      "loss": 46.0,
-      "step": 22854
-    },
-    {
-      "epoch": 1.7474243553720588,
-      "grad_norm": 0.0004190747858956456,
-      "learning_rate": 0.00019999849448195,
-      "loss": 46.0,
-      "step": 22855
-    },
-    {
-      "epoch": 1.7475008123554485,
-      "grad_norm": 0.003468609880656004,
-      "learning_rate": 0.0001999984943501446,
-      "loss": 46.0,
-      "step": 22856
-    },
-    {
-      "epoch": 1.7475772693388383,
-      "grad_norm": 0.0013469042023643851,
-      "learning_rate": 0.00019999849421833345,
-      "loss": 46.0,
-      "step": 22857
-    },
-    {
-      "epoch": 1.7476537263222278,
-      "grad_norm": 0.0012016008840873837,
-      "learning_rate": 0.0001999984940865165,
-      "loss": 46.0,
-      "step": 22858
-    },
-    {
-      "epoch": 1.7477301833056176,
-      "grad_norm": 0.0007089224527589977,
-      "learning_rate": 0.00019999849395469378,
-      "loss": 46.0,
-      "step": 22859
-    },
-    {
-      "epoch": 1.7478066402890073,
-      "grad_norm": 0.007833764888346195,
-      "learning_rate": 0.0001999984938228653,
-      "loss": 46.0,
-      "step": 22860
-    },
-    {
-      "epoch": 1.747883097272397,
-      "grad_norm": 0.001163505483418703,
-      "learning_rate": 0.00019999849369103104,
-      "loss": 46.0,
-      "step": 22861
-    },
-    {
-      "epoch": 1.7479595542557869,
-      "grad_norm": 0.00446652015671134,
-      "learning_rate": 0.00019999849355919103,
-      "loss": 46.0,
-      "step": 22862
-    },
-    {
-      "epoch": 1.7480360112391766,
-      "grad_norm": 0.0031159506179392338,
-      "learning_rate": 0.00019999849342734524,
-      "loss": 46.0,
-      "step": 22863
-    },
-    {
-      "epoch": 1.7481124682225664,
-      "grad_norm": 0.0004764788900502026,
-      "learning_rate": 0.0001999984932954937,
-      "loss": 46.0,
-      "step": 22864
-    },
-    {
-      "epoch": 1.7481889252059561,
-      "grad_norm": 0.0005654712440446019,
-      "learning_rate": 0.00019999849316363634,
-      "loss": 46.0,
-      "step": 22865
-    },
-    {
-      "epoch": 1.7482653821893457,
-      "grad_norm": 0.0013280424755066633,
-      "learning_rate": 0.00019999849303177324,
-      "loss": 46.0,
-      "step": 22866
-    },
-    {
-      "epoch": 1.7483418391727354,
-      "grad_norm": 0.0020434551406651735,
-      "learning_rate": 0.0001999984928999044,
-      "loss": 46.0,
-      "step": 22867
-    },
-    {
-      "epoch": 1.7484182961561252,
-      "grad_norm": 0.0008360739448107779,
-      "learning_rate": 0.00019999849276802975,
-      "loss": 46.0,
-      "step": 22868
-    },
-    {
-      "epoch": 1.7484947531395147,
-      "grad_norm": 0.0031081035267561674,
-      "learning_rate": 0.00019999849263614933,
-      "loss": 46.0,
-      "step": 22869
-    },
-    {
-      "epoch": 1.7485712101229045,
-      "grad_norm": 0.0017827115952968597,
-      "learning_rate": 0.00019999849250426318,
-      "loss": 46.0,
-      "step": 22870
-    },
-    {
-      "epoch": 1.7486476671062943,
-      "grad_norm": 0.004078402183949947,
-      "learning_rate": 0.00019999849237237123,
-      "loss": 46.0,
-      "step": 22871
-    },
-    {
-      "epoch": 1.748724124089684,
-      "grad_norm": 0.0007813745178282261,
-      "learning_rate": 0.0001999984922404735,
-      "loss": 46.0,
-      "step": 22872
-    },
-    {
-      "epoch": 1.7488005810730738,
-      "grad_norm": 0.0005173644749447703,
-      "learning_rate": 0.00019999849210857002,
-      "loss": 46.0,
-      "step": 22873
-    },
-    {
-      "epoch": 1.7488770380564636,
-      "grad_norm": 0.0005492236814461648,
-      "learning_rate": 0.00019999849197666078,
-      "loss": 46.0,
-      "step": 22874
-    },
-    {
-      "epoch": 1.7489534950398533,
-      "grad_norm": 0.0014844798715785146,
-      "learning_rate": 0.00019999849184474576,
-      "loss": 46.0,
-      "step": 22875
-    },
-    {
-      "epoch": 1.749029952023243,
-      "grad_norm": 0.0006212327280081809,
-      "learning_rate": 0.00019999849171282498,
-      "loss": 46.0,
-      "step": 22876
-    },
-    {
-      "epoch": 1.7491064090066326,
-      "grad_norm": 0.00043857062701135874,
-      "learning_rate": 0.00019999849158089842,
-      "loss": 46.0,
-      "step": 22877
-    },
-    {
-      "epoch": 1.7491828659900224,
-      "grad_norm": 0.0007692112121731043,
-      "learning_rate": 0.00019999849144896608,
-      "loss": 46.0,
-      "step": 22878
-    },
-    {
-      "epoch": 1.7492593229734121,
-      "grad_norm": 0.0006290115416049957,
-      "learning_rate": 0.000199998491317028,
-      "loss": 46.0,
-      "step": 22879
-    },
-    {
-      "epoch": 1.7493357799568017,
-      "grad_norm": 0.0035379943437874317,
-      "learning_rate": 0.00019999849118508412,
-      "loss": 46.0,
-      "step": 22880
-    },
-    {
-      "epoch": 1.7494122369401914,
-      "grad_norm": 0.001420535147190094,
-      "learning_rate": 0.00019999849105313447,
-      "loss": 46.0,
-      "step": 22881
-    },
-    {
-      "epoch": 1.7494886939235812,
-      "grad_norm": 0.0011333078145980835,
-      "learning_rate": 0.00019999849092117907,
-      "loss": 46.0,
-      "step": 22882
-    },
-    {
-      "epoch": 1.749565150906971,
-      "grad_norm": 0.018663544207811356,
-      "learning_rate": 0.0001999984907892179,
-      "loss": 46.0,
-      "step": 22883
-    },
-    {
-      "epoch": 1.7496416078903607,
-      "grad_norm": 0.0029617047403007746,
-      "learning_rate": 0.00019999849065725095,
-      "loss": 46.0,
-      "step": 22884
-    },
-    {
-      "epoch": 1.7497180648737505,
-      "grad_norm": 0.0011189765064045787,
-      "learning_rate": 0.00019999849052527822,
-      "loss": 46.0,
-      "step": 22885
-    },
-    {
-      "epoch": 1.7497945218571402,
-      "grad_norm": 0.0007039718911983073,
-      "learning_rate": 0.00019999849039329973,
-      "loss": 46.0,
-      "step": 22886
-    },
-    {
-      "epoch": 1.74987097884053,
-      "grad_norm": 0.0008867841097526252,
-      "learning_rate": 0.0001999984902613155,
-      "loss": 46.0,
-      "step": 22887
-    },
-    {
-      "epoch": 1.7499474358239195,
-      "grad_norm": 0.0005496154772117734,
-      "learning_rate": 0.00019999849012932548,
-      "loss": 46.0,
-      "step": 22888
-    },
-    {
-      "epoch": 1.7500238928073093,
-      "grad_norm": 0.0009427402401342988,
-      "learning_rate": 0.0001999984899973297,
-      "loss": 46.0,
-      "step": 22889
-    },
-    {
-      "epoch": 1.750100349790699,
-      "grad_norm": 0.003354016225785017,
-      "learning_rate": 0.00019999848986532813,
-      "loss": 46.0,
-      "step": 22890
-    },
-    {
-      "epoch": 1.7501768067740886,
-      "grad_norm": 0.00044121965765953064,
-      "learning_rate": 0.00019999848973332077,
-      "loss": 46.0,
-      "step": 22891
-    },
-    {
-      "epoch": 1.7502532637574784,
-      "grad_norm": 0.0006626663380302489,
-      "learning_rate": 0.00019999848960130766,
-      "loss": 46.0,
-      "step": 22892
-    },
-    {
-      "epoch": 1.7503297207408681,
-      "grad_norm": 0.0046019949950277805,
-      "learning_rate": 0.0001999984894692888,
-      "loss": 46.0,
-      "step": 22893
-    },
-    {
-      "epoch": 1.7504061777242579,
-      "grad_norm": 0.005321100354194641,
-      "learning_rate": 0.00019999848933726418,
-      "loss": 46.0,
-      "step": 22894
-    },
-    {
-      "epoch": 1.7504826347076476,
-      "grad_norm": 0.0009239411447197199,
-      "learning_rate": 0.00019999848920523375,
-      "loss": 46.0,
-      "step": 22895
-    },
-    {
-      "epoch": 1.7505590916910374,
-      "grad_norm": 0.0020021754316985607,
-      "learning_rate": 0.00019999848907319758,
-      "loss": 46.0,
-      "step": 22896
-    },
-    {
-      "epoch": 1.7506355486744272,
-      "grad_norm": 0.005139068700373173,
-      "learning_rate": 0.00019999848894115566,
-      "loss": 46.0,
-      "step": 22897
-    },
-    {
-      "epoch": 1.750712005657817,
-      "grad_norm": 0.00038632916403003037,
-      "learning_rate": 0.00019999848880910791,
-      "loss": 46.0,
-      "step": 22898
-    },
-    {
-      "epoch": 1.7507884626412065,
-      "grad_norm": 0.000553794379811734,
-      "learning_rate": 0.00019999848867705442,
-      "loss": 46.0,
-      "step": 22899
-    },
-    {
-      "epoch": 1.7508649196245962,
-      "grad_norm": 0.00201168074272573,
-      "learning_rate": 0.00019999848854499515,
-      "loss": 46.0,
-      "step": 22900
-    },
-    {
-      "epoch": 1.7509413766079858,
-      "grad_norm": 0.0006868286873213947,
-      "learning_rate": 0.00019999848841293014,
-      "loss": 46.0,
-      "step": 22901
-    },
-    {
-      "epoch": 1.7510178335913755,
-      "grad_norm": 0.004044482950121164,
-      "learning_rate": 0.00019999848828085935,
-      "loss": 46.0,
-      "step": 22902
-    },
-    {
-      "epoch": 1.7510942905747653,
-      "grad_norm": 0.0012502634199336171,
-      "learning_rate": 0.0001999984881487828,
-      "loss": 46.0,
-      "step": 22903
-    },
-    {
-      "epoch": 1.751170747558155,
-      "grad_norm": 0.0013042453210800886,
-      "learning_rate": 0.00019999848801670046,
-      "loss": 46.0,
-      "step": 22904
-    },
-    {
-      "epoch": 1.7512472045415448,
-      "grad_norm": 0.0057823313400149345,
-      "learning_rate": 0.00019999848788461235,
-      "loss": 46.0,
-      "step": 22905
-    },
-    {
-      "epoch": 1.7513236615249346,
-      "grad_norm": 0.0007642376003786922,
-      "learning_rate": 0.00019999848775251847,
-      "loss": 46.0,
-      "step": 22906
-    },
-    {
-      "epoch": 1.7514001185083243,
-      "grad_norm": 0.0020913316402584314,
-      "learning_rate": 0.00019999848762041885,
-      "loss": 46.0,
-      "step": 22907
-    },
-    {
-      "epoch": 1.751476575491714,
-      "grad_norm": 0.001820555655285716,
-      "learning_rate": 0.00019999848748831345,
-      "loss": 46.0,
-      "step": 22908
-    },
-    {
-      "epoch": 1.7515530324751039,
-      "grad_norm": 0.004824264906346798,
-      "learning_rate": 0.00019999848735620225,
-      "loss": 46.0,
-      "step": 22909
-    },
-    {
-      "epoch": 1.7516294894584934,
-      "grad_norm": 0.0040700905956327915,
-      "learning_rate": 0.0001999984872240853,
-      "loss": 46.0,
-      "step": 22910
-    },
-    {
-      "epoch": 1.7517059464418832,
-      "grad_norm": 0.0009885140461847186,
-      "learning_rate": 0.00019999848709196258,
-      "loss": 46.0,
-      "step": 22911
-    },
-    {
-      "epoch": 1.7517824034252727,
-      "grad_norm": 0.0007716494728811085,
-      "learning_rate": 0.0001999984869598341,
-      "loss": 46.0,
-      "step": 22912
-    },
-    {
-      "epoch": 1.7518588604086625,
-      "grad_norm": 0.0009181308560073376,
-      "learning_rate": 0.00019999848682769985,
-      "loss": 46.0,
-      "step": 22913
-    },
-    {
-      "epoch": 1.7519353173920522,
-      "grad_norm": 0.0005727520911023021,
-      "learning_rate": 0.0001999984866955598,
-      "loss": 46.0,
-      "step": 22914
-    },
-    {
-      "epoch": 1.752011774375442,
-      "grad_norm": 0.000605431676376611,
-      "learning_rate": 0.000199998486563414,
-      "loss": 46.0,
-      "step": 22915
-    },
-    {
-      "epoch": 1.7520882313588317,
-      "grad_norm": 0.0036140515003353357,
-      "learning_rate": 0.00019999848643126244,
-      "loss": 46.0,
-      "step": 22916
-    },
-    {
-      "epoch": 1.7521646883422215,
-      "grad_norm": 0.0013606420252472162,
-      "learning_rate": 0.0001999984862991051,
-      "loss": 46.0,
-      "step": 22917
-    },
-    {
-      "epoch": 1.7522411453256113,
-      "grad_norm": 0.0018541462486609817,
-      "learning_rate": 0.00019999848616694202,
-      "loss": 46.0,
-      "step": 22918
-    },
-    {
-      "epoch": 1.752317602309001,
-      "grad_norm": 0.0006072652759030461,
-      "learning_rate": 0.00019999848603477314,
-      "loss": 46.0,
-      "step": 22919
-    },
-    {
-      "epoch": 1.7523940592923908,
-      "grad_norm": 0.0016855557914823294,
-      "learning_rate": 0.0001999984859025985,
-      "loss": 46.0,
-      "step": 22920
-    },
-    {
-      "epoch": 1.7524705162757803,
-      "grad_norm": 0.0008594723185524344,
-      "learning_rate": 0.00019999848577041806,
-      "loss": 46.0,
-      "step": 22921
-    },
-    {
-      "epoch": 1.75254697325917,
-      "grad_norm": 0.0010709441266953945,
-      "learning_rate": 0.0001999984856382319,
-      "loss": 46.0,
-      "step": 22922
-    },
-    {
-      "epoch": 1.7526234302425596,
-      "grad_norm": 0.00046520476462319493,
-      "learning_rate": 0.00019999848550603994,
-      "loss": 46.0,
-      "step": 22923
-    },
-    {
-      "epoch": 1.7526998872259494,
-      "grad_norm": 0.003397207008674741,
-      "learning_rate": 0.00019999848537384222,
-      "loss": 46.0,
-      "step": 22924
-    },
-    {
-      "epoch": 1.7527763442093391,
-      "grad_norm": 0.0015184527728706598,
-      "learning_rate": 0.00019999848524163873,
-      "loss": 46.0,
-      "step": 22925
-    },
-    {
-      "epoch": 1.752852801192729,
-      "grad_norm": 0.0005837680073454976,
-      "learning_rate": 0.00019999848510942947,
-      "loss": 46.0,
-      "step": 22926
-    },
-    {
-      "epoch": 1.7529292581761187,
-      "grad_norm": 0.0010285830358043313,
-      "learning_rate": 0.00019999848497721445,
-      "loss": 46.0,
-      "step": 22927
-    },
-    {
-      "epoch": 1.7530057151595084,
-      "grad_norm": 0.0011810037540271878,
-      "learning_rate": 0.00019999848484499367,
-      "loss": 46.0,
-      "step": 22928
-    },
-    {
-      "epoch": 1.7530821721428982,
-      "grad_norm": 0.0004259173874743283,
-      "learning_rate": 0.00019999848471276708,
-      "loss": 46.0,
-      "step": 22929
-    },
-    {
-      "epoch": 1.753158629126288,
-      "grad_norm": 0.0010500161442905664,
-      "learning_rate": 0.00019999848458053475,
-      "loss": 46.0,
-      "step": 22930
-    },
-    {
-      "epoch": 1.7532350861096775,
-      "grad_norm": 0.0011475651990622282,
-      "learning_rate": 0.00019999848444829664,
-      "loss": 46.0,
-      "step": 22931
-    },
-    {
-      "epoch": 1.7533115430930672,
-      "grad_norm": 0.00923268124461174,
-      "learning_rate": 0.00019999848431605276,
-      "loss": 46.0,
-      "step": 22932
-    },
-    {
-      "epoch": 1.753388000076457,
-      "grad_norm": 0.0013598434161394835,
-      "learning_rate": 0.0001999984841838031,
-      "loss": 46.0,
-      "step": 22933
-    },
-    {
-      "epoch": 1.7534644570598465,
-      "grad_norm": 0.0029720028396695852,
-      "learning_rate": 0.0001999984840515477,
-      "loss": 46.0,
-      "step": 22934
-    },
-    {
-      "epoch": 1.7535409140432363,
-      "grad_norm": 0.0008924399735406041,
-      "learning_rate": 0.0001999984839192865,
-      "loss": 46.0,
-      "step": 22935
-    },
-    {
-      "epoch": 1.753617371026626,
-      "grad_norm": 0.0014193335082381964,
-      "learning_rate": 0.00019999848378701956,
-      "loss": 46.0,
-      "step": 22936
-    },
-    {
-      "epoch": 1.7536938280100158,
-      "grad_norm": 0.00140817416831851,
-      "learning_rate": 0.00019999848365474685,
-      "loss": 46.0,
-      "step": 22937
-    },
-    {
-      "epoch": 1.7537702849934056,
-      "grad_norm": 0.0007353240507654846,
-      "learning_rate": 0.00019999848352246835,
-      "loss": 46.0,
-      "step": 22938
-    },
-    {
-      "epoch": 1.7538467419767954,
-      "grad_norm": 0.0014878481160849333,
-      "learning_rate": 0.0001999984833901841,
-      "loss": 46.0,
-      "step": 22939
-    },
-    {
-      "epoch": 1.7539231989601851,
-      "grad_norm": 0.0021783888805657625,
-      "learning_rate": 0.00019999848325789405,
-      "loss": 46.0,
-      "step": 22940
-    },
-    {
-      "epoch": 1.7539996559435749,
-      "grad_norm": 0.0012715599732473493,
-      "learning_rate": 0.00019999848312559826,
-      "loss": 46.0,
-      "step": 22941
-    },
-    {
-      "epoch": 1.7540761129269644,
-      "grad_norm": 0.0012889860663563013,
-      "learning_rate": 0.00019999848299329668,
-      "loss": 46.0,
-      "step": 22942
-    },
-    {
-      "epoch": 1.7541525699103542,
-      "grad_norm": 0.000898173835594207,
-      "learning_rate": 0.00019999848286098937,
-      "loss": 46.0,
-      "step": 22943
-    },
-    {
-      "epoch": 1.754229026893744,
-      "grad_norm": 0.0027461578138172626,
-      "learning_rate": 0.00019999848272867624,
-      "loss": 46.0,
-      "step": 22944
-    },
-    {
-      "epoch": 1.7543054838771335,
-      "grad_norm": 0.0007422103080898523,
-      "learning_rate": 0.00019999848259635736,
-      "loss": 46.0,
-      "step": 22945
-    },
-    {
-      "epoch": 1.7543819408605232,
-      "grad_norm": 0.00827016681432724,
-      "learning_rate": 0.00019999848246403273,
-      "loss": 46.0,
-      "step": 22946
-    },
-    {
-      "epoch": 1.754458397843913,
-      "grad_norm": 0.00832771323621273,
-      "learning_rate": 0.0001999984823317023,
-      "loss": 46.0,
-      "step": 22947
-    },
-    {
-      "epoch": 1.7545348548273028,
-      "grad_norm": 0.0016653466736897826,
-      "learning_rate": 0.00019999848219936613,
-      "loss": 46.0,
-      "step": 22948
-    },
-    {
-      "epoch": 1.7546113118106925,
-      "grad_norm": 0.0019959560595452785,
-      "learning_rate": 0.00019999848206702416,
-      "loss": 46.0,
-      "step": 22949
-    },
-    {
-      "epoch": 1.7546877687940823,
-      "grad_norm": 0.001784514868631959,
-      "learning_rate": 0.00019999848193467642,
-      "loss": 46.0,
-      "step": 22950
-    },
-    {
-      "epoch": 1.754764225777472,
-      "grad_norm": 0.00265725189819932,
-      "learning_rate": 0.00019999848180232295,
-      "loss": 46.0,
-      "step": 22951
-    },
-    {
-      "epoch": 1.7548406827608618,
-      "grad_norm": 0.0005549251800402999,
-      "learning_rate": 0.00019999848166996369,
-      "loss": 46.0,
-      "step": 22952
-    },
-    {
-      "epoch": 1.7549171397442513,
-      "grad_norm": 0.003805492538958788,
-      "learning_rate": 0.00019999848153759865,
-      "loss": 46.0,
-      "step": 22953
-    },
-    {
-      "epoch": 1.754993596727641,
-      "grad_norm": 0.001588365063071251,
-      "learning_rate": 0.00019999848140522786,
-      "loss": 46.0,
-      "step": 22954
-    },
-    {
-      "epoch": 1.7550700537110309,
-      "grad_norm": 0.0008523932774551213,
-      "learning_rate": 0.00019999848127285128,
-      "loss": 46.0,
-      "step": 22955
-    },
-    {
-      "epoch": 1.7551465106944204,
-      "grad_norm": 0.002328865695744753,
-      "learning_rate": 0.00019999848114046894,
-      "loss": 46.0,
-      "step": 22956
-    },
-    {
-      "epoch": 1.7552229676778102,
-      "grad_norm": 0.0005983063601888716,
-      "learning_rate": 0.00019999848100808084,
-      "loss": 46.0,
-      "step": 22957
-    },
-    {
-      "epoch": 1.7552994246612,
-      "grad_norm": 0.00479264697059989,
-      "learning_rate": 0.00019999848087568696,
-      "loss": 46.0,
-      "step": 22958
-    },
-    {
-      "epoch": 1.7553758816445897,
-      "grad_norm": 0.0005162210436537862,
-      "learning_rate": 0.0001999984807432873,
-      "loss": 46.0,
-      "step": 22959
-    },
-    {
-      "epoch": 1.7554523386279794,
-      "grad_norm": 0.0008861337555572391,
-      "learning_rate": 0.00019999848061088188,
-      "loss": 46.0,
-      "step": 22960
-    },
-    {
-      "epoch": 1.7555287956113692,
-      "grad_norm": 0.0009791561169549823,
-      "learning_rate": 0.0001999984804784707,
-      "loss": 46.0,
-      "step": 22961
-    },
-    {
-      "epoch": 1.755605252594759,
-      "grad_norm": 0.0018878286937251687,
-      "learning_rate": 0.00019999848034605374,
-      "loss": 46.0,
-      "step": 22962
-    },
-    {
-      "epoch": 1.7556817095781487,
-      "grad_norm": 0.0012427836190909147,
-      "learning_rate": 0.00019999848021363102,
-      "loss": 46.0,
-      "step": 22963
-    },
-    {
-      "epoch": 1.7557581665615383,
-      "grad_norm": 0.0010017188033089042,
-      "learning_rate": 0.00019999848008120253,
-      "loss": 46.0,
-      "step": 22964
-    },
-    {
-      "epoch": 1.755834623544928,
-      "grad_norm": 0.001413905993103981,
-      "learning_rate": 0.00019999847994876826,
-      "loss": 46.0,
-      "step": 22965
-    },
-    {
-      "epoch": 1.7559110805283178,
-      "grad_norm": 0.001142196706496179,
-      "learning_rate": 0.00019999847981632822,
-      "loss": 46.0,
-      "step": 22966
-    },
-    {
-      "epoch": 1.7559875375117073,
-      "grad_norm": 0.0004801061295438558,
-      "learning_rate": 0.00019999847968388244,
-      "loss": 46.0,
-      "step": 22967
-    },
-    {
-      "epoch": 1.756063994495097,
-      "grad_norm": 0.0002558489504735917,
-      "learning_rate": 0.00019999847955143085,
-      "loss": 46.0,
-      "step": 22968
-    },
-    {
-      "epoch": 1.7561404514784869,
-      "grad_norm": 0.0006692885654047132,
-      "learning_rate": 0.0001999984794189735,
-      "loss": 46.0,
-      "step": 22969
-    },
-    {
-      "epoch": 1.7562169084618766,
-      "grad_norm": 0.000777937937527895,
-      "learning_rate": 0.0001999984792865104,
-      "loss": 46.0,
-      "step": 22970
-    },
-    {
-      "epoch": 1.7562933654452664,
-      "grad_norm": 0.002101212739944458,
-      "learning_rate": 0.0001999984791540415,
-      "loss": 46.0,
-      "step": 22971
-    },
-    {
-      "epoch": 1.7563698224286561,
-      "grad_norm": 0.004076733253896236,
-      "learning_rate": 0.00019999847902156686,
-      "loss": 46.0,
-      "step": 22972
-    },
-    {
-      "epoch": 1.756446279412046,
-      "grad_norm": 0.00183341302908957,
-      "learning_rate": 0.00019999847888908643,
-      "loss": 46.0,
-      "step": 22973
-    },
-    {
-      "epoch": 1.7565227363954357,
-      "grad_norm": 0.0008881328976713121,
-      "learning_rate": 0.00019999847875660026,
-      "loss": 46.0,
-      "step": 22974
-    },
-    {
-      "epoch": 1.7565991933788252,
-      "grad_norm": 0.0024700553622096777,
-      "learning_rate": 0.0001999984786241083,
-      "loss": 46.0,
-      "step": 22975
-    },
-    {
-      "epoch": 1.756675650362215,
-      "grad_norm": 0.004311004187911749,
-      "learning_rate": 0.00019999847849161054,
-      "loss": 46.0,
-      "step": 22976
-    },
-    {
-      "epoch": 1.7567521073456047,
-      "grad_norm": 0.000527957221493125,
-      "learning_rate": 0.00019999847835910708,
-      "loss": 46.0,
-      "step": 22977
-    },
-    {
-      "epoch": 1.7568285643289943,
-      "grad_norm": 0.002179940929636359,
-      "learning_rate": 0.0001999984782265978,
-      "loss": 46.0,
-      "step": 22978
-    },
-    {
-      "epoch": 1.756905021312384,
-      "grad_norm": 0.001060448819771409,
-      "learning_rate": 0.00019999847809408277,
-      "loss": 46.0,
-      "step": 22979
-    },
-    {
-      "epoch": 1.7569814782957738,
-      "grad_norm": 0.0009896565461531281,
-      "learning_rate": 0.00019999847796156196,
-      "loss": 46.0,
-      "step": 22980
-    },
-    {
-      "epoch": 1.7570579352791635,
-      "grad_norm": 0.0007081157527863979,
-      "learning_rate": 0.0001999984778290354,
-      "loss": 46.0,
-      "step": 22981
-    },
-    {
-      "epoch": 1.7571343922625533,
-      "grad_norm": 0.0014334216248244047,
-      "learning_rate": 0.00019999847769650307,
-      "loss": 46.0,
-      "step": 22982
-    },
-    {
-      "epoch": 1.757210849245943,
-      "grad_norm": 0.0014801254728809,
-      "learning_rate": 0.00019999847756396494,
-      "loss": 46.0,
-      "step": 22983
-    },
-    {
-      "epoch": 1.7572873062293328,
-      "grad_norm": 0.001222072052769363,
-      "learning_rate": 0.00019999847743142106,
-      "loss": 46.0,
-      "step": 22984
-    },
-    {
-      "epoch": 1.7573637632127226,
-      "grad_norm": 0.0007329648360610008,
-      "learning_rate": 0.0001999984772988714,
-      "loss": 46.0,
-      "step": 22985
-    },
-    {
-      "epoch": 1.7574402201961121,
-      "grad_norm": 0.0017048392910510302,
-      "learning_rate": 0.00019999847716631596,
-      "loss": 46.0,
-      "step": 22986
-    },
-    {
-      "epoch": 1.7575166771795019,
-      "grad_norm": 0.0008884343551471829,
-      "learning_rate": 0.0001999984770337548,
-      "loss": 46.0,
-      "step": 22987
-    },
-    {
-      "epoch": 1.7575931341628916,
-      "grad_norm": 0.0016720156418159604,
-      "learning_rate": 0.00019999847690118782,
-      "loss": 46.0,
-      "step": 22988
-    },
-    {
-      "epoch": 1.7576695911462812,
-      "grad_norm": 0.004272850695997477,
-      "learning_rate": 0.0001999984767686151,
-      "loss": 46.0,
-      "step": 22989
-    },
-    {
-      "epoch": 1.757746048129671,
-      "grad_norm": 0.0010351334931328893,
-      "learning_rate": 0.00019999847663603658,
-      "loss": 46.0,
-      "step": 22990
-    },
-    {
-      "epoch": 1.7578225051130607,
-      "grad_norm": 0.001561406534165144,
-      "learning_rate": 0.00019999847650345234,
-      "loss": 46.0,
-      "step": 22991
-    },
-    {
-      "epoch": 1.7578989620964505,
-      "grad_norm": 0.0007286272011697292,
-      "learning_rate": 0.0001999984763708623,
-      "loss": 46.0,
-      "step": 22992
-    },
-    {
-      "epoch": 1.7579754190798402,
-      "grad_norm": 0.0009288830915465951,
-      "learning_rate": 0.0001999984762382665,
-      "loss": 46.0,
-      "step": 22993
-    },
-    {
-      "epoch": 1.75805187606323,
-      "grad_norm": 0.0020690688397735357,
-      "learning_rate": 0.0001999984761056649,
-      "loss": 46.0,
-      "step": 22994
-    },
-    {
-      "epoch": 1.7581283330466198,
-      "grad_norm": 0.0005922580603510141,
-      "learning_rate": 0.00019999847597305758,
-      "loss": 46.0,
-      "step": 22995
-    },
-    {
-      "epoch": 1.7582047900300095,
-      "grad_norm": 0.0017920979298651218,
-      "learning_rate": 0.00019999847584044444,
-      "loss": 46.0,
-      "step": 22996
-    },
-    {
-      "epoch": 1.758281247013399,
-      "grad_norm": 0.00025034809368662536,
-      "learning_rate": 0.00019999847570782557,
-      "loss": 46.0,
-      "step": 22997
-    },
-    {
-      "epoch": 1.7583577039967888,
-      "grad_norm": 0.0008215445559471846,
-      "learning_rate": 0.00019999847557520092,
-      "loss": 46.0,
-      "step": 22998
-    },
-    {
-      "epoch": 1.7584341609801786,
-      "grad_norm": 0.0008346502436324954,
-      "learning_rate": 0.0001999984754425705,
-      "loss": 46.0,
-      "step": 22999
-    },
-    {
-      "epoch": 1.7585106179635681,
-      "grad_norm": 0.0009983754716813564,
-      "learning_rate": 0.0001999984753099343,
-      "loss": 46.0,
-      "step": 23000
-    },
-    {
-      "epoch": 1.7585870749469579,
-      "grad_norm": 0.0016917455941438675,
-      "learning_rate": 0.00019999847517729232,
-      "loss": 46.0,
-      "step": 23001
-    },
-    {
-      "epoch": 1.7586635319303476,
-      "grad_norm": 0.0022480038460344076,
-      "learning_rate": 0.0001999984750446446,
-      "loss": 46.0,
-      "step": 23002
-    },
-    {
-      "epoch": 1.7587399889137374,
-      "grad_norm": 0.0006862612208351493,
-      "learning_rate": 0.0001999984749119911,
-      "loss": 46.0,
-      "step": 23003
-    },
-    {
-      "epoch": 1.7588164458971272,
-      "grad_norm": 0.0008875308558344841,
-      "learning_rate": 0.00019999847477933185,
-      "loss": 46.0,
-      "step": 23004
-    },
-    {
-      "epoch": 1.758892902880517,
-      "grad_norm": 0.003141864435747266,
-      "learning_rate": 0.00019999847464666678,
-      "loss": 46.0,
-      "step": 23005
-    },
-    {
-      "epoch": 1.7589693598639067,
-      "grad_norm": 0.0007503660162910819,
-      "learning_rate": 0.00019999847451399597,
-      "loss": 46.0,
-      "step": 23006
-    },
-    {
-      "epoch": 1.7590458168472964,
-      "grad_norm": 0.0006347611779347062,
-      "learning_rate": 0.00019999847438131941,
-      "loss": 46.0,
-      "step": 23007
-    },
-    {
-      "epoch": 1.759122273830686,
-      "grad_norm": 0.0020879756193608046,
-      "learning_rate": 0.00019999847424863706,
-      "loss": 46.0,
-      "step": 23008
-    },
-    {
-      "epoch": 1.7591987308140757,
-      "grad_norm": 0.0007142355316318572,
-      "learning_rate": 0.00019999847411594895,
-      "loss": 46.0,
-      "step": 23009
-    },
-    {
-      "epoch": 1.7592751877974655,
-      "grad_norm": 0.00042321221553720534,
-      "learning_rate": 0.00019999847398325505,
-      "loss": 46.0,
-      "step": 23010
-    },
-    {
-      "epoch": 1.759351644780855,
-      "grad_norm": 0.0007877916796132922,
-      "learning_rate": 0.00019999847385055542,
-      "loss": 46.0,
-      "step": 23011
-    },
-    {
-      "epoch": 1.7594281017642448,
-      "grad_norm": 0.0008601034642197192,
-      "learning_rate": 0.00019999847371784997,
-      "loss": 46.0,
-      "step": 23012
-    },
-    {
-      "epoch": 1.7595045587476346,
-      "grad_norm": 0.00041944821714423597,
-      "learning_rate": 0.0001999984735851388,
-      "loss": 46.0,
-      "step": 23013
-    },
-    {
-      "epoch": 1.7595810157310243,
-      "grad_norm": 0.0008222644100897014,
-      "learning_rate": 0.00019999847345242183,
-      "loss": 46.0,
-      "step": 23014
-    },
-    {
-      "epoch": 1.759657472714414,
-      "grad_norm": 0.0012930564116686583,
-      "learning_rate": 0.0001999984733196991,
-      "loss": 46.0,
-      "step": 23015
-    },
-    {
-      "epoch": 1.7597339296978038,
-      "grad_norm": 0.001170386909507215,
-      "learning_rate": 0.00019999847318697057,
-      "loss": 46.0,
-      "step": 23016
-    },
-    {
-      "epoch": 1.7598103866811936,
-      "grad_norm": 0.0008373709279112518,
-      "learning_rate": 0.0001999984730542363,
-      "loss": 46.0,
-      "step": 23017
-    },
-    {
-      "epoch": 1.7598868436645834,
-      "grad_norm": 0.0023464609403163195,
-      "learning_rate": 0.00019999847292149627,
-      "loss": 46.0,
-      "step": 23018
-    },
-    {
-      "epoch": 1.759963300647973,
-      "grad_norm": 0.000365028390660882,
-      "learning_rate": 0.00019999847278875046,
-      "loss": 46.0,
-      "step": 23019
-    },
-    {
-      "epoch": 1.7600397576313627,
-      "grad_norm": 0.004052398726344109,
-      "learning_rate": 0.00019999847265599887,
-      "loss": 46.0,
-      "step": 23020
-    },
-    {
-      "epoch": 1.7601162146147524,
-      "grad_norm": 0.0009370911866426468,
-      "learning_rate": 0.00019999847252324154,
-      "loss": 46.0,
-      "step": 23021
-    },
-    {
-      "epoch": 1.760192671598142,
-      "grad_norm": 0.001851333538070321,
-      "learning_rate": 0.0001999984723904784,
-      "loss": 46.0,
-      "step": 23022
-    },
-    {
-      "epoch": 1.7602691285815317,
-      "grad_norm": 0.0013390742242336273,
-      "learning_rate": 0.0001999984722577095,
-      "loss": 46.0,
-      "step": 23023
-    },
-    {
-      "epoch": 1.7603455855649215,
-      "grad_norm": 0.0013048751279711723,
-      "learning_rate": 0.00019999847212493488,
-      "loss": 46.0,
-      "step": 23024
-    },
-    {
-      "epoch": 1.7604220425483112,
-      "grad_norm": 0.001635401975363493,
-      "learning_rate": 0.00019999847199215446,
-      "loss": 46.0,
-      "step": 23025
-    },
-    {
-      "epoch": 1.760498499531701,
-      "grad_norm": 0.0007840105681680143,
-      "learning_rate": 0.00019999847185936823,
-      "loss": 46.0,
-      "step": 23026
-    },
-    {
-      "epoch": 1.7605749565150908,
-      "grad_norm": 0.00026450862060301006,
-      "learning_rate": 0.0001999984717265763,
-      "loss": 46.0,
-      "step": 23027
-    },
-    {
-      "epoch": 1.7606514134984805,
-      "grad_norm": 0.002210391918197274,
-      "learning_rate": 0.00019999847159377855,
-      "loss": 46.0,
-      "step": 23028
-    },
-    {
-      "epoch": 1.7607278704818703,
-      "grad_norm": 0.0007283276063390076,
-      "learning_rate": 0.00019999847146097506,
-      "loss": 46.0,
-      "step": 23029
-    },
-    {
-      "epoch": 1.7608043274652598,
-      "grad_norm": 0.003568722167983651,
-      "learning_rate": 0.0001999984713281658,
-      "loss": 46.0,
-      "step": 23030
-    },
-    {
-      "epoch": 1.7608807844486496,
-      "grad_norm": 0.0021393224596977234,
-      "learning_rate": 0.00019999847119535073,
-      "loss": 46.0,
-      "step": 23031
-    },
-    {
-      "epoch": 1.7609572414320391,
-      "grad_norm": 0.0008001381647773087,
-      "learning_rate": 0.00019999847106252992,
-      "loss": 46.0,
-      "step": 23032
-    },
-    {
-      "epoch": 1.761033698415429,
-      "grad_norm": 0.0010784866753965616,
-      "learning_rate": 0.0001999984709297033,
-      "loss": 46.0,
-      "step": 23033
-    },
-    {
-      "epoch": 1.7611101553988187,
-      "grad_norm": 0.0005454495549201965,
-      "learning_rate": 0.000199998470796871,
-      "loss": 46.0,
-      "step": 23034
-    },
-    {
-      "epoch": 1.7611866123822084,
-      "grad_norm": 0.00164863932877779,
-      "learning_rate": 0.00019999847066403285,
-      "loss": 46.0,
-      "step": 23035
-    },
-    {
-      "epoch": 1.7612630693655982,
-      "grad_norm": 0.0012681143125519156,
-      "learning_rate": 0.000199998470531189,
-      "loss": 46.0,
-      "step": 23036
-    },
-    {
-      "epoch": 1.761339526348988,
-      "grad_norm": 0.002939911326393485,
-      "learning_rate": 0.0001999984703983393,
-      "loss": 46.0,
-      "step": 23037
-    },
-    {
-      "epoch": 1.7614159833323777,
-      "grad_norm": 0.003414539387449622,
-      "learning_rate": 0.0001999984702654839,
-      "loss": 46.0,
-      "step": 23038
-    },
-    {
-      "epoch": 1.7614924403157675,
-      "grad_norm": 0.0007823705673217773,
-      "learning_rate": 0.0001999984701326227,
-      "loss": 46.0,
-      "step": 23039
-    },
-    {
-      "epoch": 1.7615688972991572,
-      "grad_norm": 0.0015217153122648597,
-      "learning_rate": 0.00019999846999975573,
-      "loss": 46.0,
-      "step": 23040
-    },
-    {
-      "epoch": 1.7616453542825468,
-      "grad_norm": 0.0008177928393706679,
-      "learning_rate": 0.000199998469866883,
-      "loss": 46.0,
-      "step": 23041
-    },
-    {
-      "epoch": 1.7617218112659365,
-      "grad_norm": 0.004731368273496628,
-      "learning_rate": 0.0001999984697340045,
-      "loss": 46.0,
-      "step": 23042
-    },
-    {
-      "epoch": 1.761798268249326,
-      "grad_norm": 0.00045001530088484287,
-      "learning_rate": 0.0001999984696011202,
-      "loss": 46.0,
-      "step": 23043
-    },
-    {
-      "epoch": 1.7618747252327158,
-      "grad_norm": 0.0010435177246108651,
-      "learning_rate": 0.00019999846946823017,
-      "loss": 46.0,
-      "step": 23044
-    },
-    {
-      "epoch": 1.7619511822161056,
-      "grad_norm": 0.0013058970216661692,
-      "learning_rate": 0.00019999846933533433,
-      "loss": 46.0,
-      "step": 23045
-    },
-    {
-      "epoch": 1.7620276391994953,
-      "grad_norm": 0.0031492572743445635,
-      "learning_rate": 0.00019999846920243277,
-      "loss": 46.0,
-      "step": 23046
-    },
-    {
-      "epoch": 1.762104096182885,
-      "grad_norm": 0.0010075252503156662,
-      "learning_rate": 0.0001999984690695254,
-      "loss": 46.0,
-      "step": 23047
-    },
-    {
-      "epoch": 1.7621805531662749,
-      "grad_norm": 0.0024334799963980913,
-      "learning_rate": 0.00019999846893661228,
-      "loss": 46.0,
-      "step": 23048
-    },
-    {
-      "epoch": 1.7622570101496646,
-      "grad_norm": 0.001103270798921585,
-      "learning_rate": 0.0001999984688036934,
-      "loss": 46.0,
-      "step": 23049
-    },
-    {
-      "epoch": 1.7623334671330544,
-      "grad_norm": 0.0024429045151919127,
-      "learning_rate": 0.00019999846867076873,
-      "loss": 46.0,
-      "step": 23050
-    },
-    {
-      "epoch": 1.7624099241164441,
-      "grad_norm": 0.0036698102485388517,
-      "learning_rate": 0.0001999984685378383,
-      "loss": 46.0,
-      "step": 23051
-    },
-    {
-      "epoch": 1.7624863810998337,
-      "grad_norm": 0.0009404487791471183,
-      "learning_rate": 0.0001999984684049021,
-      "loss": 46.0,
-      "step": 23052
-    },
-    {
-      "epoch": 1.7625628380832234,
-      "grad_norm": 0.0021824913565069437,
-      "learning_rate": 0.00019999846827196014,
-      "loss": 46.0,
-      "step": 23053
-    },
-    {
-      "epoch": 1.762639295066613,
-      "grad_norm": 0.0007494486053474247,
-      "learning_rate": 0.0001999984681390124,
-      "loss": 46.0,
-      "step": 23054
-    },
-    {
-      "epoch": 1.7627157520500027,
-      "grad_norm": 0.0014965999871492386,
-      "learning_rate": 0.00019999846800605888,
-      "loss": 46.0,
-      "step": 23055
-    },
-    {
-      "epoch": 1.7627922090333925,
-      "grad_norm": 0.0009514080011285841,
-      "learning_rate": 0.00019999846787309962,
-      "loss": 46.0,
-      "step": 23056
-    },
-    {
-      "epoch": 1.7628686660167823,
-      "grad_norm": 0.0007919525378383696,
-      "learning_rate": 0.00019999846774013458,
-      "loss": 46.0,
-      "step": 23057
-    },
-    {
-      "epoch": 1.762945123000172,
-      "grad_norm": 0.000772044702898711,
-      "learning_rate": 0.00019999846760716374,
-      "loss": 46.0,
-      "step": 23058
-    },
-    {
-      "epoch": 1.7630215799835618,
-      "grad_norm": 0.0012643402442336082,
-      "learning_rate": 0.0001999984674741872,
-      "loss": 46.0,
-      "step": 23059
-    },
-    {
-      "epoch": 1.7630980369669516,
-      "grad_norm": 0.0006897153798490763,
-      "learning_rate": 0.0001999984673412048,
-      "loss": 46.0,
-      "step": 23060
-    },
-    {
-      "epoch": 1.7631744939503413,
-      "grad_norm": 0.0012045793700963259,
-      "learning_rate": 0.0001999984672082167,
-      "loss": 46.0,
-      "step": 23061
-    },
-    {
-      "epoch": 1.763250950933731,
-      "grad_norm": 0.0007357140420936048,
-      "learning_rate": 0.0001999984670752228,
-      "loss": 46.0,
-      "step": 23062
-    },
-    {
-      "epoch": 1.7633274079171206,
-      "grad_norm": 0.003234150819480419,
-      "learning_rate": 0.00019999846694222312,
-      "loss": 46.0,
-      "step": 23063
-    },
-    {
-      "epoch": 1.7634038649005104,
-      "grad_norm": 0.0009918593568727374,
-      "learning_rate": 0.0001999984668092177,
-      "loss": 46.0,
-      "step": 23064
-    },
-    {
-      "epoch": 1.7634803218839,
-      "grad_norm": 0.0011889488669112325,
-      "learning_rate": 0.0001999984666762065,
-      "loss": 46.0,
-      "step": 23065
-    },
-    {
-      "epoch": 1.7635567788672897,
-      "grad_norm": 0.0027338166255503893,
-      "learning_rate": 0.00019999846654318956,
-      "loss": 46.0,
-      "step": 23066
-    },
-    {
-      "epoch": 1.7636332358506794,
-      "grad_norm": 0.0017383953090757132,
-      "learning_rate": 0.0001999984664101668,
-      "loss": 46.0,
-      "step": 23067
-    },
-    {
-      "epoch": 1.7637096928340692,
-      "grad_norm": 0.0018819868564605713,
-      "learning_rate": 0.00019999846627713827,
-      "loss": 46.0,
-      "step": 23068
-    },
-    {
-      "epoch": 1.763786149817459,
-      "grad_norm": 0.001236236421391368,
-      "learning_rate": 0.000199998466144104,
-      "loss": 46.0,
-      "step": 23069
-    },
-    {
-      "epoch": 1.7638626068008487,
-      "grad_norm": 0.00044397584861144423,
-      "learning_rate": 0.00019999846601106394,
-      "loss": 46.0,
-      "step": 23070
-    },
-    {
-      "epoch": 1.7639390637842385,
-      "grad_norm": 0.001205915934406221,
-      "learning_rate": 0.00019999846587801813,
-      "loss": 46.0,
-      "step": 23071
-    },
-    {
-      "epoch": 1.7640155207676282,
-      "grad_norm": 0.0011821012012660503,
-      "learning_rate": 0.00019999846574496655,
-      "loss": 46.0,
-      "step": 23072
-    },
-    {
-      "epoch": 1.7640919777510178,
-      "grad_norm": 0.002327796770259738,
-      "learning_rate": 0.0001999984656119092,
-      "loss": 46.0,
-      "step": 23073
-    },
-    {
-      "epoch": 1.7641684347344075,
-      "grad_norm": 0.0009161459747701883,
-      "learning_rate": 0.00019999846547884607,
-      "loss": 46.0,
-      "step": 23074
-    },
-    {
-      "epoch": 1.7642448917177973,
-      "grad_norm": 0.0005388643476180732,
-      "learning_rate": 0.00019999846534577716,
-      "loss": 46.0,
-      "step": 23075
-    },
-    {
-      "epoch": 1.7643213487011868,
-      "grad_norm": 0.0017050675814971328,
-      "learning_rate": 0.00019999846521270249,
-      "loss": 46.0,
-      "step": 23076
-    },
-    {
-      "epoch": 1.7643978056845766,
-      "grad_norm": 0.0002865987189579755,
-      "learning_rate": 0.0001999984650796221,
-      "loss": 46.0,
-      "step": 23077
-    },
-    {
-      "epoch": 1.7644742626679664,
-      "grad_norm": 0.000978047726675868,
-      "learning_rate": 0.00019999846494653587,
-      "loss": 46.0,
-      "step": 23078
-    },
-    {
-      "epoch": 1.7645507196513561,
-      "grad_norm": 0.0008743317448534071,
-      "learning_rate": 0.0001999984648134439,
-      "loss": 46.0,
-      "step": 23079
-    },
-    {
-      "epoch": 1.7646271766347459,
-      "grad_norm": 0.004247661214321852,
-      "learning_rate": 0.00019999846468034615,
-      "loss": 46.0,
-      "step": 23080
-    },
-    {
-      "epoch": 1.7647036336181356,
-      "grad_norm": 0.001804320840165019,
-      "learning_rate": 0.00019999846454724264,
-      "loss": 46.0,
-      "step": 23081
-    },
-    {
-      "epoch": 1.7647800906015254,
-      "grad_norm": 0.00039279242628253996,
-      "learning_rate": 0.00019999846441413338,
-      "loss": 46.0,
-      "step": 23082
-    },
-    {
-      "epoch": 1.7648565475849152,
-      "grad_norm": 0.0008443098631687462,
-      "learning_rate": 0.0001999984642810183,
-      "loss": 46.0,
-      "step": 23083
-    },
-    {
-      "epoch": 1.7649330045683047,
-      "grad_norm": 0.0042954012751579285,
-      "learning_rate": 0.0001999984641478975,
-      "loss": 46.0,
-      "step": 23084
-    },
-    {
-      "epoch": 1.7650094615516945,
-      "grad_norm": 0.0030831163749098778,
-      "learning_rate": 0.00019999846401477092,
-      "loss": 46.0,
-      "step": 23085
-    },
-    {
-      "epoch": 1.7650859185350842,
-      "grad_norm": 0.0035576969385147095,
-      "learning_rate": 0.00019999846388163856,
-      "loss": 46.0,
-      "step": 23086
-    },
-    {
-      "epoch": 1.7651623755184738,
-      "grad_norm": 0.0008713395218364894,
-      "learning_rate": 0.0001999984637485004,
-      "loss": 46.0,
-      "step": 23087
-    },
-    {
-      "epoch": 1.7652388325018635,
-      "grad_norm": 0.000810770783573389,
-      "learning_rate": 0.00019999846361535653,
-      "loss": 46.0,
-      "step": 23088
-    },
-    {
-      "epoch": 1.7653152894852533,
-      "grad_norm": 0.0007830265094526112,
-      "learning_rate": 0.00019999846348220686,
-      "loss": 46.0,
-      "step": 23089
-    },
-    {
-      "epoch": 1.765391746468643,
-      "grad_norm": 0.0017373106675222516,
-      "learning_rate": 0.00019999846334905143,
-      "loss": 46.0,
-      "step": 23090
-    },
-    {
-      "epoch": 1.7654682034520328,
-      "grad_norm": 0.0032556059304624796,
-      "learning_rate": 0.0001999984632158902,
-      "loss": 46.0,
-      "step": 23091
-    },
-    {
-      "epoch": 1.7655446604354226,
-      "grad_norm": 0.0028761804569512606,
-      "learning_rate": 0.00019999846308272324,
-      "loss": 46.0,
-      "step": 23092
-    },
-    {
-      "epoch": 1.7656211174188123,
-      "grad_norm": 0.001863745623268187,
-      "learning_rate": 0.00019999846294955053,
-      "loss": 46.0,
-      "step": 23093
-    },
-    {
-      "epoch": 1.765697574402202,
-      "grad_norm": 0.0008716261363588274,
-      "learning_rate": 0.00019999846281637198,
-      "loss": 46.0,
-      "step": 23094
-    },
-    {
-      "epoch": 1.7657740313855916,
-      "grad_norm": 0.0031519723124802113,
-      "learning_rate": 0.00019999846268318772,
-      "loss": 46.0,
-      "step": 23095
-    },
-    {
-      "epoch": 1.7658504883689814,
-      "grad_norm": 0.0037860311567783356,
-      "learning_rate": 0.00019999846254999766,
-      "loss": 46.0,
-      "step": 23096
-    },
-    {
-      "epoch": 1.7659269453523712,
-      "grad_norm": 0.002153954701498151,
-      "learning_rate": 0.00019999846241680182,
-      "loss": 46.0,
-      "step": 23097
-    },
-    {
-      "epoch": 1.7660034023357607,
-      "grad_norm": 0.0009732242324389517,
-      "learning_rate": 0.00019999846228360024,
-      "loss": 46.0,
-      "step": 23098
-    },
-    {
-      "epoch": 1.7660798593191505,
-      "grad_norm": 0.004467865917831659,
-      "learning_rate": 0.00019999846215039288,
-      "loss": 46.0,
-      "step": 23099
-    },
-    {
-      "epoch": 1.7661563163025402,
-      "grad_norm": 0.0008208779036067426,
-      "learning_rate": 0.00019999846201717975,
-      "loss": 46.0,
-      "step": 23100
-    },
-    {
-      "epoch": 1.76623277328593,
-      "grad_norm": 0.0012709691654890776,
-      "learning_rate": 0.00019999846188396085,
-      "loss": 46.0,
-      "step": 23101
-    },
-    {
-      "epoch": 1.7663092302693197,
-      "grad_norm": 0.0005803306121379137,
-      "learning_rate": 0.00019999846175073618,
-      "loss": 46.0,
-      "step": 23102
-    },
-    {
-      "epoch": 1.7663856872527095,
-      "grad_norm": 0.0005613188841380179,
-      "learning_rate": 0.00019999846161750575,
-      "loss": 46.0,
-      "step": 23103
-    },
-    {
-      "epoch": 1.7664621442360993,
-      "grad_norm": 0.0010739011922851205,
-      "learning_rate": 0.00019999846148426956,
-      "loss": 46.0,
-      "step": 23104
-    },
-    {
-      "epoch": 1.766538601219489,
-      "grad_norm": 0.003066170262172818,
-      "learning_rate": 0.00019999846135102756,
-      "loss": 46.0,
-      "step": 23105
-    },
-    {
-      "epoch": 1.7666150582028786,
-      "grad_norm": 0.0007236566161736846,
-      "learning_rate": 0.00019999846121777982,
-      "loss": 46.0,
-      "step": 23106
-    },
-    {
-      "epoch": 1.7666915151862683,
-      "grad_norm": 0.0012443193700164557,
-      "learning_rate": 0.0001999984610845263,
-      "loss": 46.0,
-      "step": 23107
-    },
-    {
-      "epoch": 1.766767972169658,
-      "grad_norm": 0.0005706559168174863,
-      "learning_rate": 0.00019999846095126702,
-      "loss": 46.0,
-      "step": 23108
-    },
-    {
-      "epoch": 1.7668444291530476,
-      "grad_norm": 0.0028388206847012043,
-      "learning_rate": 0.00019999846081800198,
-      "loss": 46.0,
-      "step": 23109
-    },
-    {
-      "epoch": 1.7669208861364374,
-      "grad_norm": 0.00118687329813838,
-      "learning_rate": 0.00019999846068473117,
-      "loss": 46.0,
-      "step": 23110
-    },
-    {
-      "epoch": 1.7669973431198271,
-      "grad_norm": 0.0007301591103896499,
-      "learning_rate": 0.00019999846055145456,
-      "loss": 46.0,
-      "step": 23111
-    },
-    {
-      "epoch": 1.767073800103217,
-      "grad_norm": 0.0006311266915872693,
-      "learning_rate": 0.0001999984604181722,
-      "loss": 46.0,
-      "step": 23112
-    },
-    {
-      "epoch": 1.7671502570866067,
-      "grad_norm": 0.0012225827667862177,
-      "learning_rate": 0.00019999846028488408,
-      "loss": 46.0,
-      "step": 23113
-    },
-    {
-      "epoch": 1.7672267140699964,
-      "grad_norm": 0.0006423905142582953,
-      "learning_rate": 0.00019999846015159015,
-      "loss": 46.0,
-      "step": 23114
-    },
-    {
-      "epoch": 1.7673031710533862,
-      "grad_norm": 0.0015107393264770508,
-      "learning_rate": 0.0001999984600182905,
-      "loss": 46.0,
-      "step": 23115
-    },
-    {
-      "epoch": 1.767379628036776,
-      "grad_norm": 0.0006345415604300797,
-      "learning_rate": 0.00019999845988498508,
-      "loss": 46.0,
-      "step": 23116
-    },
-    {
-      "epoch": 1.7674560850201655,
-      "grad_norm": 0.0009737061918713152,
-      "learning_rate": 0.00019999845975167383,
-      "loss": 46.0,
-      "step": 23117
-    },
-    {
-      "epoch": 1.7675325420035553,
-      "grad_norm": 0.0007959159556776285,
-      "learning_rate": 0.0001999984596183569,
-      "loss": 46.0,
-      "step": 23118
-    },
-    {
-      "epoch": 1.767608998986945,
-      "grad_norm": 0.009825431741774082,
-      "learning_rate": 0.00019999845948503412,
-      "loss": 46.0,
-      "step": 23119
-    },
-    {
-      "epoch": 1.7676854559703346,
-      "grad_norm": 0.0072423676028847694,
-      "learning_rate": 0.0001999984593517056,
-      "loss": 46.0,
-      "step": 23120
-    },
-    {
-      "epoch": 1.7677619129537243,
-      "grad_norm": 0.0006642839289270341,
-      "learning_rate": 0.00019999845921837132,
-      "loss": 46.0,
-      "step": 23121
-    },
-    {
-      "epoch": 1.767838369937114,
-      "grad_norm": 0.0011451351456344128,
-      "learning_rate": 0.00019999845908503128,
-      "loss": 46.0,
-      "step": 23122
-    },
-    {
-      "epoch": 1.7679148269205038,
-      "grad_norm": 0.001224048319272697,
-      "learning_rate": 0.00019999845895168544,
-      "loss": 46.0,
-      "step": 23123
-    },
-    {
-      "epoch": 1.7679912839038936,
-      "grad_norm": 0.0007788949878886342,
-      "learning_rate": 0.00019999845881833386,
-      "loss": 46.0,
-      "step": 23124
-    },
-    {
-      "epoch": 1.7680677408872834,
-      "grad_norm": 0.0035099743399769068,
-      "learning_rate": 0.00019999845868497648,
-      "loss": 46.0,
-      "step": 23125
-    },
-    {
-      "epoch": 1.7681441978706731,
-      "grad_norm": 0.0017941266996785998,
-      "learning_rate": 0.00019999845855161335,
-      "loss": 46.0,
-      "step": 23126
-    },
-    {
-      "epoch": 1.7682206548540629,
-      "grad_norm": 0.0011262808693572879,
-      "learning_rate": 0.00019999845841824445,
-      "loss": 46.0,
-      "step": 23127
-    },
-    {
-      "epoch": 1.7682971118374524,
-      "grad_norm": 0.0008886543801054358,
-      "learning_rate": 0.00019999845828486978,
-      "loss": 46.0,
-      "step": 23128
-    },
-    {
-      "epoch": 1.7683735688208422,
-      "grad_norm": 0.008047091774642467,
-      "learning_rate": 0.00019999845815148935,
-      "loss": 46.0,
-      "step": 23129
-    },
-    {
-      "epoch": 1.768450025804232,
-      "grad_norm": 0.0008445560815744102,
-      "learning_rate": 0.00019999845801810313,
-      "loss": 46.0,
-      "step": 23130
-    },
-    {
-      "epoch": 1.7685264827876215,
-      "grad_norm": 0.002351143630221486,
-      "learning_rate": 0.00019999845788471116,
-      "loss": 46.0,
-      "step": 23131
-    },
-    {
-      "epoch": 1.7686029397710112,
-      "grad_norm": 0.0011543554719537497,
-      "learning_rate": 0.0001999984577513134,
-      "loss": 46.0,
-      "step": 23132
-    },
-    {
-      "epoch": 1.768679396754401,
-      "grad_norm": 0.0013604448176920414,
-      "learning_rate": 0.00019999845761790988,
-      "loss": 46.0,
-      "step": 23133
-    },
-    {
-      "epoch": 1.7687558537377908,
-      "grad_norm": 0.0008139307610690594,
-      "learning_rate": 0.0001999984574845006,
-      "loss": 46.0,
-      "step": 23134
-    },
-    {
-      "epoch": 1.7688323107211805,
-      "grad_norm": 0.006762339733541012,
-      "learning_rate": 0.00019999845735108553,
-      "loss": 46.0,
-      "step": 23135
-    },
-    {
-      "epoch": 1.7689087677045703,
-      "grad_norm": 0.0014616911066696048,
-      "learning_rate": 0.0001999984572176647,
-      "loss": 46.0,
-      "step": 23136
-    },
-    {
-      "epoch": 1.76898522468796,
-      "grad_norm": 0.004571232013404369,
-      "learning_rate": 0.0001999984570842381,
-      "loss": 46.0,
-      "step": 23137
-    },
-    {
-      "epoch": 1.7690616816713498,
-      "grad_norm": 0.009523669257760048,
-      "learning_rate": 0.00019999845695080576,
-      "loss": 46.0,
-      "step": 23138
-    },
-    {
-      "epoch": 1.7691381386547393,
-      "grad_norm": 0.0015674973838031292,
-      "learning_rate": 0.00019999845681736763,
-      "loss": 46.0,
-      "step": 23139
-    },
-    {
-      "epoch": 1.769214595638129,
-      "grad_norm": 0.003483065403997898,
-      "learning_rate": 0.00019999845668392373,
-      "loss": 46.0,
-      "step": 23140
-    },
-    {
-      "epoch": 1.7692910526215189,
-      "grad_norm": 0.0011361864162608981,
-      "learning_rate": 0.00019999845655047405,
-      "loss": 46.0,
-      "step": 23141
-    },
-    {
-      "epoch": 1.7693675096049084,
-      "grad_norm": 0.0006945352652110159,
-      "learning_rate": 0.0001999984564170186,
-      "loss": 46.0,
-      "step": 23142
-    },
-    {
-      "epoch": 1.7694439665882982,
-      "grad_norm": 0.000498998211696744,
-      "learning_rate": 0.00019999845628355738,
-      "loss": 46.0,
-      "step": 23143
-    },
-    {
-      "epoch": 1.769520423571688,
-      "grad_norm": 0.0003876656701322645,
-      "learning_rate": 0.00019999845615009042,
-      "loss": 46.0,
-      "step": 23144
-    },
-    {
-      "epoch": 1.7695968805550777,
-      "grad_norm": 0.005981977563351393,
-      "learning_rate": 0.00019999845601661767,
-      "loss": 46.0,
-      "step": 23145
-    },
-    {
-      "epoch": 1.7696733375384675,
-      "grad_norm": 0.0029290628153830767,
-      "learning_rate": 0.00019999845588313916,
-      "loss": 46.0,
-      "step": 23146
-    },
-    {
-      "epoch": 1.7697497945218572,
-      "grad_norm": 0.0007858932367525995,
-      "learning_rate": 0.00019999845574965484,
-      "loss": 46.0,
-      "step": 23147
-    },
-    {
-      "epoch": 1.769826251505247,
-      "grad_norm": 0.002118223113939166,
-      "learning_rate": 0.00019999845561616478,
-      "loss": 46.0,
-      "step": 23148
-    },
-    {
-      "epoch": 1.7699027084886367,
-      "grad_norm": 0.000950882094912231,
-      "learning_rate": 0.00019999845548266895,
-      "loss": 46.0,
-      "step": 23149
-    },
-    {
-      "epoch": 1.7699791654720263,
-      "grad_norm": 0.0018941829912364483,
-      "learning_rate": 0.00019999845534916737,
-      "loss": 46.0,
-      "step": 23150
-    },
-    {
-      "epoch": 1.770055622455416,
-      "grad_norm": 0.0008296070736832917,
-      "learning_rate": 0.00019999845521566,
-      "loss": 46.0,
-      "step": 23151
-    },
-    {
-      "epoch": 1.7701320794388058,
-      "grad_norm": 0.0005480193067342043,
-      "learning_rate": 0.00019999845508214686,
-      "loss": 46.0,
-      "step": 23152
-    },
-    {
-      "epoch": 1.7702085364221953,
-      "grad_norm": 0.0007223273278214037,
-      "learning_rate": 0.00019999845494862796,
-      "loss": 46.0,
-      "step": 23153
-    },
-    {
-      "epoch": 1.770284993405585,
-      "grad_norm": 0.0009624581434763968,
-      "learning_rate": 0.00019999845481510328,
-      "loss": 46.0,
-      "step": 23154
-    },
-    {
-      "epoch": 1.7703614503889749,
-      "grad_norm": 0.0011638266732916236,
-      "learning_rate": 0.00019999845468157283,
-      "loss": 46.0,
-      "step": 23155
-    },
-    {
-      "epoch": 1.7704379073723646,
-      "grad_norm": 0.0037739588879048824,
-      "learning_rate": 0.00019999845454803664,
-      "loss": 46.0,
-      "step": 23156
-    },
-    {
-      "epoch": 1.7705143643557544,
-      "grad_norm": 0.001201388891786337,
-      "learning_rate": 0.00019999845441449464,
-      "loss": 46.0,
-      "step": 23157
-    },
-    {
-      "epoch": 1.7705908213391441,
-      "grad_norm": 0.0016899992479011416,
-      "learning_rate": 0.0001999984542809469,
-      "loss": 46.0,
-      "step": 23158
-    },
-    {
-      "epoch": 1.770667278322534,
-      "grad_norm": 0.012375622056424618,
-      "learning_rate": 0.0001999984541473934,
-      "loss": 46.0,
-      "step": 23159
-    },
-    {
-      "epoch": 1.7707437353059237,
-      "grad_norm": 0.0004891376011073589,
-      "learning_rate": 0.00019999845401383407,
-      "loss": 46.0,
-      "step": 23160
-    },
-    {
-      "epoch": 1.7708201922893132,
-      "grad_norm": 0.0014209039509296417,
-      "learning_rate": 0.000199998453880269,
-      "loss": 46.0,
-      "step": 23161
-    },
-    {
-      "epoch": 1.770896649272703,
-      "grad_norm": 0.0006470477674156427,
-      "learning_rate": 0.00019999845374669818,
-      "loss": 46.0,
-      "step": 23162
-    },
-    {
-      "epoch": 1.7709731062560925,
-      "grad_norm": 0.0029922465328127146,
-      "learning_rate": 0.0001999984536131216,
-      "loss": 46.0,
-      "step": 23163
-    },
-    {
-      "epoch": 1.7710495632394823,
-      "grad_norm": 0.012191470712423325,
-      "learning_rate": 0.00019999845347953922,
-      "loss": 46.0,
-      "step": 23164
-    },
-    {
-      "epoch": 1.771126020222872,
-      "grad_norm": 0.0024859150871634483,
-      "learning_rate": 0.0001999984533459511,
-      "loss": 46.0,
-      "step": 23165
-    },
-    {
-      "epoch": 1.7712024772062618,
-      "grad_norm": 0.0019891022238880396,
-      "learning_rate": 0.00019999845321235716,
-      "loss": 46.0,
-      "step": 23166
-    },
-    {
-      "epoch": 1.7712789341896515,
-      "grad_norm": 0.0013281208230182528,
-      "learning_rate": 0.0001999984530787575,
-      "loss": 46.0,
-      "step": 23167
-    },
-    {
-      "epoch": 1.7713553911730413,
-      "grad_norm": 0.00130205019377172,
-      "learning_rate": 0.00019999845294515204,
-      "loss": 46.0,
-      "step": 23168
-    },
-    {
-      "epoch": 1.771431848156431,
-      "grad_norm": 0.0005985440802760422,
-      "learning_rate": 0.00019999845281154085,
-      "loss": 46.0,
-      "step": 23169
-    },
-    {
-      "epoch": 1.7715083051398208,
-      "grad_norm": 0.002968721790239215,
-      "learning_rate": 0.00019999845267792385,
-      "loss": 46.0,
-      "step": 23170
-    },
-    {
-      "epoch": 1.7715847621232106,
-      "grad_norm": 0.0006349895847961307,
-      "learning_rate": 0.00019999845254430108,
-      "loss": 46.0,
-      "step": 23171
-    },
-    {
-      "epoch": 1.7716612191066001,
-      "grad_norm": 0.0006741798715665936,
-      "learning_rate": 0.00019999845241067257,
-      "loss": 46.0,
-      "step": 23172
-    },
-    {
-      "epoch": 1.7717376760899899,
-      "grad_norm": 0.0008166732732206583,
-      "learning_rate": 0.00019999845227703828,
-      "loss": 46.0,
-      "step": 23173
-    },
-    {
-      "epoch": 1.7718141330733794,
-      "grad_norm": 0.0023718588054180145,
-      "learning_rate": 0.00019999845214339822,
-      "loss": 46.0,
-      "step": 23174
-    },
-    {
-      "epoch": 1.7718905900567692,
-      "grad_norm": 0.001330870552919805,
-      "learning_rate": 0.0001999984520097524,
-      "loss": 46.0,
-      "step": 23175
-    },
-    {
-      "epoch": 1.771967047040159,
-      "grad_norm": 0.0010530907893553376,
-      "learning_rate": 0.0001999984518761008,
-      "loss": 46.0,
-      "step": 23176
-    },
-    {
-      "epoch": 1.7720435040235487,
-      "grad_norm": 0.005613513290882111,
-      "learning_rate": 0.0001999984517424434,
-      "loss": 46.0,
-      "step": 23177
-    },
-    {
-      "epoch": 1.7721199610069385,
-      "grad_norm": 0.0031455366406589746,
-      "learning_rate": 0.00019999845160878027,
-      "loss": 46.0,
-      "step": 23178
-    },
-    {
-      "epoch": 1.7721964179903282,
-      "grad_norm": 0.0010642630513757467,
-      "learning_rate": 0.00019999845147511137,
-      "loss": 46.0,
-      "step": 23179
-    },
-    {
-      "epoch": 1.772272874973718,
-      "grad_norm": 0.0014018104411661625,
-      "learning_rate": 0.0001999984513414367,
-      "loss": 46.0,
-      "step": 23180
-    },
-    {
-      "epoch": 1.7723493319571078,
-      "grad_norm": 0.0006225567194633186,
-      "learning_rate": 0.00019999845120775625,
-      "loss": 46.0,
-      "step": 23181
-    },
-    {
-      "epoch": 1.7724257889404975,
-      "grad_norm": 0.0005686302902176976,
-      "learning_rate": 0.00019999845107407003,
-      "loss": 46.0,
-      "step": 23182
-    },
-    {
-      "epoch": 1.772502245923887,
-      "grad_norm": 0.0025199546944350004,
-      "learning_rate": 0.00019999845094037806,
-      "loss": 46.0,
-      "step": 23183
-    },
-    {
-      "epoch": 1.7725787029072768,
-      "grad_norm": 0.0006324366549961269,
-      "learning_rate": 0.0001999984508066803,
-      "loss": 46.0,
-      "step": 23184
-    },
-    {
-      "epoch": 1.7726551598906664,
-      "grad_norm": 0.0022804951295256615,
-      "learning_rate": 0.00019999845067297675,
-      "loss": 46.0,
-      "step": 23185
-    },
-    {
-      "epoch": 1.7727316168740561,
-      "grad_norm": 0.0010371665703132749,
-      "learning_rate": 0.00019999845053926747,
-      "loss": 46.0,
-      "step": 23186
-    },
-    {
-      "epoch": 1.7728080738574459,
-      "grad_norm": 0.002356019103899598,
-      "learning_rate": 0.0001999984504055524,
-      "loss": 46.0,
-      "step": 23187
-    },
-    {
-      "epoch": 1.7728845308408356,
-      "grad_norm": 0.0010994799667969346,
-      "learning_rate": 0.00019999845027183157,
-      "loss": 46.0,
-      "step": 23188
-    },
-    {
-      "epoch": 1.7729609878242254,
-      "grad_norm": 0.0008478878880850971,
-      "learning_rate": 0.00019999845013810496,
-      "loss": 46.0,
-      "step": 23189
-    },
-    {
-      "epoch": 1.7730374448076152,
-      "grad_norm": 0.001982656307518482,
-      "learning_rate": 0.0001999984500043726,
-      "loss": 46.0,
-      "step": 23190
-    },
-    {
-      "epoch": 1.773113901791005,
-      "grad_norm": 0.0008776086033321917,
-      "learning_rate": 0.00019999844987063446,
-      "loss": 46.0,
-      "step": 23191
-    },
-    {
-      "epoch": 1.7731903587743947,
-      "grad_norm": 0.0026984878350049257,
-      "learning_rate": 0.00019999844973689056,
-      "loss": 46.0,
-      "step": 23192
-    },
-    {
-      "epoch": 1.7732668157577844,
-      "grad_norm": 0.0019534658640623093,
-      "learning_rate": 0.00019999844960314088,
-      "loss": 46.0,
-      "step": 23193
-    },
-    {
-      "epoch": 1.773343272741174,
-      "grad_norm": 0.0013197226217016578,
-      "learning_rate": 0.0001999984494693854,
-      "loss": 46.0,
-      "step": 23194
-    },
-    {
-      "epoch": 1.7734197297245637,
-      "grad_norm": 0.0026509107556194067,
-      "learning_rate": 0.0001999984493356242,
-      "loss": 46.0,
-      "step": 23195
-    },
-    {
-      "epoch": 1.7734961867079533,
-      "grad_norm": 0.0007069522398523986,
-      "learning_rate": 0.00019999844920185722,
-      "loss": 46.0,
-      "step": 23196
-    },
-    {
-      "epoch": 1.773572643691343,
-      "grad_norm": 0.002517035696655512,
-      "learning_rate": 0.00019999844906808445,
-      "loss": 46.0,
-      "step": 23197
-    },
-    {
-      "epoch": 1.7736491006747328,
-      "grad_norm": 0.0005949987680651248,
-      "learning_rate": 0.00019999844893430591,
-      "loss": 46.0,
-      "step": 23198
-    },
-    {
-      "epoch": 1.7737255576581226,
-      "grad_norm": 0.005007871892303228,
-      "learning_rate": 0.00019999844880052163,
-      "loss": 46.0,
-      "step": 23199
-    },
-    {
-      "epoch": 1.7738020146415123,
-      "grad_norm": 0.0006895462283864617,
-      "learning_rate": 0.00019999844866673157,
-      "loss": 46.0,
-      "step": 23200
-    },
-    {
-      "epoch": 1.773878471624902,
-      "grad_norm": 0.0015854892553761601,
-      "learning_rate": 0.00019999844853293573,
-      "loss": 46.0,
-      "step": 23201
-    },
-    {
-      "epoch": 1.7739549286082918,
-      "grad_norm": 0.002647966146469116,
-      "learning_rate": 0.00019999844839913413,
-      "loss": 46.0,
-      "step": 23202
-    },
-    {
-      "epoch": 1.7740313855916816,
-      "grad_norm": 0.001964349066838622,
-      "learning_rate": 0.00019999844826532677,
-      "loss": 46.0,
-      "step": 23203
-    },
-    {
-      "epoch": 1.7741078425750711,
-      "grad_norm": 0.0009689134894870222,
-      "learning_rate": 0.0001999984481315136,
-      "loss": 46.0,
-      "step": 23204
-    },
-    {
-      "epoch": 1.774184299558461,
-      "grad_norm": 0.0006538014858961105,
-      "learning_rate": 0.0001999984479976947,
-      "loss": 46.0,
-      "step": 23205
-    },
-    {
-      "epoch": 1.7742607565418507,
-      "grad_norm": 0.0004765463818330318,
-      "learning_rate": 0.00019999844786387002,
-      "loss": 46.0,
-      "step": 23206
-    },
-    {
-      "epoch": 1.7743372135252402,
-      "grad_norm": 0.0018598216120153666,
-      "learning_rate": 0.00019999844773003957,
-      "loss": 46.0,
-      "step": 23207
-    },
-    {
-      "epoch": 1.77441367050863,
-      "grad_norm": 0.0017507366137579083,
-      "learning_rate": 0.00019999844759620335,
-      "loss": 46.0,
-      "step": 23208
-    },
-    {
-      "epoch": 1.7744901274920197,
-      "grad_norm": 0.0015931391390040517,
-      "learning_rate": 0.00019999844746236133,
-      "loss": 46.0,
-      "step": 23209
-    },
-    {
-      "epoch": 1.7745665844754095,
-      "grad_norm": 0.0006228883867152035,
-      "learning_rate": 0.00019999844732851362,
-      "loss": 46.0,
-      "step": 23210
-    },
-    {
-      "epoch": 1.7746430414587993,
-      "grad_norm": 0.0010079315397888422,
-      "learning_rate": 0.00019999844719466005,
-      "loss": 46.0,
-      "step": 23211
-    },
-    {
-      "epoch": 1.774719498442189,
-      "grad_norm": 0.0026071411557495594,
-      "learning_rate": 0.00019999844706080076,
-      "loss": 46.0,
-      "step": 23212
-    },
-    {
-      "epoch": 1.7747959554255788,
-      "grad_norm": 0.0015334176132455468,
-      "learning_rate": 0.0001999984469269357,
-      "loss": 46.0,
-      "step": 23213
-    },
-    {
-      "epoch": 1.7748724124089685,
-      "grad_norm": 0.0020003607496619225,
-      "learning_rate": 0.00019999844679306484,
-      "loss": 46.0,
-      "step": 23214
-    },
-    {
-      "epoch": 1.774948869392358,
-      "grad_norm": 0.0009174122824333608,
-      "learning_rate": 0.00019999844665918827,
-      "loss": 46.0,
-      "step": 23215
-    },
-    {
-      "epoch": 1.7750253263757478,
-      "grad_norm": 0.0042031812481582165,
-      "learning_rate": 0.00019999844652530589,
-      "loss": 46.0,
-      "step": 23216
-    },
-    {
-      "epoch": 1.7751017833591376,
-      "grad_norm": 0.001706851995550096,
-      "learning_rate": 0.00019999844639141773,
-      "loss": 46.0,
-      "step": 23217
-    },
-    {
-      "epoch": 1.7751782403425271,
-      "grad_norm": 0.002087873872369528,
-      "learning_rate": 0.00019999844625752383,
-      "loss": 46.0,
-      "step": 23218
-    },
-    {
-      "epoch": 1.775254697325917,
-      "grad_norm": 0.0006772560300305486,
-      "learning_rate": 0.00019999844612362413,
-      "loss": 46.0,
-      "step": 23219
-    },
-    {
-      "epoch": 1.7753311543093067,
-      "grad_norm": 0.0020502344705164433,
-      "learning_rate": 0.00019999844598971869,
-      "loss": 46.0,
-      "step": 23220
-    },
-    {
-      "epoch": 1.7754076112926964,
-      "grad_norm": 0.0011054871138185263,
-      "learning_rate": 0.00019999844585580747,
-      "loss": 46.0,
-      "step": 23221
-    },
-    {
-      "epoch": 1.7754840682760862,
-      "grad_norm": 0.000559154199436307,
-      "learning_rate": 0.00019999844572189047,
-      "loss": 46.0,
-      "step": 23222
-    },
-    {
-      "epoch": 1.775560525259476,
-      "grad_norm": 0.003327616024762392,
-      "learning_rate": 0.00019999844558796773,
-      "loss": 46.0,
-      "step": 23223
-    },
-    {
-      "epoch": 1.7756369822428657,
-      "grad_norm": 0.002628596732392907,
-      "learning_rate": 0.0001999984454540392,
-      "loss": 46.0,
-      "step": 23224
-    },
-    {
-      "epoch": 1.7757134392262555,
-      "grad_norm": 0.0008326114621013403,
-      "learning_rate": 0.00019999844532010488,
-      "loss": 46.0,
-      "step": 23225
-    },
-    {
-      "epoch": 1.775789896209645,
-      "grad_norm": 0.0012390506453812122,
-      "learning_rate": 0.00019999844518616482,
-      "loss": 46.0,
-      "step": 23226
-    },
-    {
-      "epoch": 1.7758663531930348,
-      "grad_norm": 0.01298129465430975,
-      "learning_rate": 0.000199998445052219,
-      "loss": 46.0,
-      "step": 23227
-    },
-    {
-      "epoch": 1.7759428101764245,
-      "grad_norm": 0.00398150272667408,
-      "learning_rate": 0.00019999844491826738,
-      "loss": 46.0,
-      "step": 23228
-    },
-    {
-      "epoch": 1.776019267159814,
-      "grad_norm": 0.0006963791674934328,
-      "learning_rate": 0.00019999844478431,
-      "loss": 46.0,
-      "step": 23229
-    },
-    {
-      "epoch": 1.7760957241432038,
-      "grad_norm": 0.0018393455538898706,
-      "learning_rate": 0.00019999844465034685,
-      "loss": 46.0,
-      "step": 23230
-    },
-    {
-      "epoch": 1.7761721811265936,
-      "grad_norm": 0.0008311583660542965,
-      "learning_rate": 0.00019999844451637792,
-      "loss": 46.0,
-      "step": 23231
-    },
-    {
-      "epoch": 1.7762486381099833,
-      "grad_norm": 0.0007329022628255188,
-      "learning_rate": 0.00019999844438240322,
-      "loss": 46.0,
-      "step": 23232
-    },
-    {
-      "epoch": 1.776325095093373,
-      "grad_norm": 0.004611688200384378,
-      "learning_rate": 0.0001999984442484228,
-      "loss": 46.0,
-      "step": 23233
-    },
-    {
-      "epoch": 1.7764015520767629,
-      "grad_norm": 0.0015256782062351704,
-      "learning_rate": 0.00019999844411443656,
-      "loss": 46.0,
-      "step": 23234
-    },
-    {
-      "epoch": 1.7764780090601526,
-      "grad_norm": 0.0004698209522757679,
-      "learning_rate": 0.0001999984439804446,
-      "loss": 46.0,
-      "step": 23235
-    },
-    {
-      "epoch": 1.7765544660435424,
-      "grad_norm": 0.0007031442364677787,
-      "learning_rate": 0.0001999984438464468,
-      "loss": 46.0,
-      "step": 23236
-    },
-    {
-      "epoch": 1.776630923026932,
-      "grad_norm": 0.0022335692774504423,
-      "learning_rate": 0.00019999844371244328,
-      "loss": 46.0,
-      "step": 23237
-    },
-    {
-      "epoch": 1.7767073800103217,
-      "grad_norm": 0.012205605395138264,
-      "learning_rate": 0.00019999844357843397,
-      "loss": 46.0,
-      "step": 23238
-    },
-    {
-      "epoch": 1.7767838369937115,
-      "grad_norm": 0.0007610509055666625,
-      "learning_rate": 0.0001999984434444189,
-      "loss": 46.0,
-      "step": 23239
-    },
-    {
-      "epoch": 1.776860293977101,
-      "grad_norm": 0.0011236566351726651,
-      "learning_rate": 0.00019999844331039808,
-      "loss": 46.0,
-      "step": 23240
-    },
-    {
-      "epoch": 1.7769367509604908,
-      "grad_norm": 0.00048113096272572875,
-      "learning_rate": 0.00019999844317637147,
-      "loss": 46.0,
-      "step": 23241
-    },
-    {
-      "epoch": 1.7770132079438805,
-      "grad_norm": 0.0009932528482750058,
-      "learning_rate": 0.00019999844304233907,
-      "loss": 46.0,
-      "step": 23242
-    },
-    {
-      "epoch": 1.7770896649272703,
-      "grad_norm": 0.0040565053932368755,
-      "learning_rate": 0.00019999844290830091,
-      "loss": 46.0,
-      "step": 23243
-    },
-    {
-      "epoch": 1.77716612191066,
-      "grad_norm": 0.0005631509120576084,
-      "learning_rate": 0.00019999844277425702,
-      "loss": 46.0,
-      "step": 23244
-    },
-    {
-      "epoch": 1.7772425788940498,
-      "grad_norm": 0.0024882592260837555,
-      "learning_rate": 0.00019999844264020732,
-      "loss": 46.0,
-      "step": 23245
-    },
-    {
-      "epoch": 1.7773190358774396,
-      "grad_norm": 0.0007060154457576573,
-      "learning_rate": 0.00019999844250615184,
-      "loss": 46.0,
-      "step": 23246
-    },
-    {
-      "epoch": 1.7773954928608293,
-      "grad_norm": 0.002434036461636424,
-      "learning_rate": 0.00019999844237209065,
-      "loss": 46.0,
-      "step": 23247
-    },
-    {
-      "epoch": 1.7774719498442189,
-      "grad_norm": 0.00185084727127105,
-      "learning_rate": 0.00019999844223802363,
-      "loss": 46.0,
-      "step": 23248
-    },
-    {
-      "epoch": 1.7775484068276086,
-      "grad_norm": 0.000825387891381979,
-      "learning_rate": 0.00019999844210395087,
-      "loss": 46.0,
-      "step": 23249
-    },
-    {
-      "epoch": 1.7776248638109984,
-      "grad_norm": 0.0013457470340654254,
-      "learning_rate": 0.00019999844196987235,
-      "loss": 46.0,
-      "step": 23250
-    },
-    {
-      "epoch": 1.777701320794388,
-      "grad_norm": 0.0007637071539647877,
-      "learning_rate": 0.00019999844183578804,
-      "loss": 46.0,
-      "step": 23251
-    },
-    {
-      "epoch": 1.7777777777777777,
-      "grad_norm": 0.0006038442370481789,
-      "learning_rate": 0.00019999844170169798,
-      "loss": 46.0,
-      "step": 23252
-    },
-    {
-      "epoch": 1.7778542347611674,
-      "grad_norm": 0.001025978708639741,
-      "learning_rate": 0.00019999844156760212,
-      "loss": 46.0,
-      "step": 23253
-    },
-    {
-      "epoch": 1.7779306917445572,
-      "grad_norm": 0.0003737601509783417,
-      "learning_rate": 0.00019999844143350052,
-      "loss": 46.0,
-      "step": 23254
-    },
-    {
-      "epoch": 1.778007148727947,
-      "grad_norm": 0.0010456457966938615,
-      "learning_rate": 0.00019999844129939314,
-      "loss": 46.0,
-      "step": 23255
-    },
-    {
-      "epoch": 1.7780836057113367,
-      "grad_norm": 0.0019081287318840623,
-      "learning_rate": 0.00019999844116527999,
-      "loss": 46.0,
-      "step": 23256
-    },
-    {
-      "epoch": 1.7781600626947265,
-      "grad_norm": 0.0009576175943948328,
-      "learning_rate": 0.00019999844103116106,
-      "loss": 46.0,
-      "step": 23257
-    },
-    {
-      "epoch": 1.7782365196781162,
-      "grad_norm": 0.0008039544918574393,
-      "learning_rate": 0.0001999984408970364,
-      "loss": 46.0,
-      "step": 23258
-    },
-    {
-      "epoch": 1.7783129766615058,
-      "grad_norm": 0.002151490654796362,
-      "learning_rate": 0.00019999844076290591,
-      "loss": 46.0,
-      "step": 23259
-    },
-    {
-      "epoch": 1.7783894336448955,
-      "grad_norm": 0.0007029905100353062,
-      "learning_rate": 0.0001999984406287697,
-      "loss": 46.0,
-      "step": 23260
-    },
-    {
-      "epoch": 1.7784658906282853,
-      "grad_norm": 0.0008604549220763147,
-      "learning_rate": 0.0001999984404946277,
-      "loss": 46.0,
-      "step": 23261
-    },
-    {
-      "epoch": 1.7785423476116748,
-      "grad_norm": 0.0013133713509887457,
-      "learning_rate": 0.00019999844036047994,
-      "loss": 46.0,
-      "step": 23262
-    },
-    {
-      "epoch": 1.7786188045950646,
-      "grad_norm": 0.014470922760665417,
-      "learning_rate": 0.0001999984402263264,
-      "loss": 46.0,
-      "step": 23263
-    },
-    {
-      "epoch": 1.7786952615784544,
-      "grad_norm": 0.0008368071285076439,
-      "learning_rate": 0.00019999844009216711,
-      "loss": 46.0,
-      "step": 23264
-    },
-    {
-      "epoch": 1.7787717185618441,
-      "grad_norm": 0.0014800543431192636,
-      "learning_rate": 0.00019999843995800203,
-      "loss": 46.0,
-      "step": 23265
-    },
-    {
-      "epoch": 1.778848175545234,
-      "grad_norm": 0.002377198077738285,
-      "learning_rate": 0.00019999843982383117,
-      "loss": 46.0,
-      "step": 23266
-    },
-    {
-      "epoch": 1.7789246325286237,
-      "grad_norm": 0.0010161095997318625,
-      "learning_rate": 0.00019999843968965454,
-      "loss": 46.0,
-      "step": 23267
-    },
-    {
-      "epoch": 1.7790010895120134,
-      "grad_norm": 0.0011116121895611286,
-      "learning_rate": 0.00019999843955547219,
-      "loss": 46.0,
-      "step": 23268
-    },
-    {
-      "epoch": 1.7790775464954032,
-      "grad_norm": 0.0014691423857584596,
-      "learning_rate": 0.00019999843942128403,
-      "loss": 46.0,
-      "step": 23269
-    },
-    {
-      "epoch": 1.7791540034787927,
-      "grad_norm": 0.0012214378220960498,
-      "learning_rate": 0.00019999843928709013,
-      "loss": 46.0,
-      "step": 23270
-    },
-    {
-      "epoch": 1.7792304604621825,
-      "grad_norm": 0.0025345112662762403,
-      "learning_rate": 0.0001999984391528904,
-      "loss": 46.0,
-      "step": 23271
-    },
-    {
-      "epoch": 1.7793069174455722,
-      "grad_norm": 0.0007620805990882218,
-      "learning_rate": 0.00019999843901868496,
-      "loss": 46.0,
-      "step": 23272
-    },
-    {
-      "epoch": 1.7793833744289618,
-      "grad_norm": 0.0011552328942343593,
-      "learning_rate": 0.00019999843888447374,
-      "loss": 46.0,
-      "step": 23273
-    },
-    {
-      "epoch": 1.7794598314123515,
-      "grad_norm": 0.0011205559130758047,
-      "learning_rate": 0.00019999843875025675,
-      "loss": 46.0,
-      "step": 23274
-    },
-    {
-      "epoch": 1.7795362883957413,
-      "grad_norm": 0.0006173912552185357,
-      "learning_rate": 0.00019999843861603396,
-      "loss": 46.0,
-      "step": 23275
-    },
-    {
-      "epoch": 1.779612745379131,
-      "grad_norm": 0.0009429361089132726,
-      "learning_rate": 0.00019999843848180545,
-      "loss": 46.0,
-      "step": 23276
-    },
-    {
-      "epoch": 1.7796892023625208,
-      "grad_norm": 0.00042408870649524033,
-      "learning_rate": 0.00019999843834757114,
-      "loss": 46.0,
-      "step": 23277
-    },
-    {
-      "epoch": 1.7797656593459106,
-      "grad_norm": 0.0017061886610463262,
-      "learning_rate": 0.00019999843821333105,
-      "loss": 46.0,
-      "step": 23278
-    },
-    {
-      "epoch": 1.7798421163293003,
-      "grad_norm": 0.0010921771172434092,
-      "learning_rate": 0.0001999984380790852,
-      "loss": 46.0,
-      "step": 23279
-    },
-    {
-      "epoch": 1.77991857331269,
-      "grad_norm": 0.0016911420971155167,
-      "learning_rate": 0.0001999984379448336,
-      "loss": 46.0,
-      "step": 23280
-    },
-    {
-      "epoch": 1.7799950302960796,
-      "grad_norm": 0.0009914758848026395,
-      "learning_rate": 0.00019999843781057618,
-      "loss": 46.0,
-      "step": 23281
-    },
-    {
-      "epoch": 1.7800714872794694,
-      "grad_norm": 0.001459247781895101,
-      "learning_rate": 0.00019999843767631306,
-      "loss": 46.0,
-      "step": 23282
-    },
-    {
-      "epoch": 1.7801479442628592,
-      "grad_norm": 0.0006058759172447026,
-      "learning_rate": 0.00019999843754204413,
-      "loss": 46.0,
-      "step": 23283
-    },
-    {
-      "epoch": 1.7802244012462487,
-      "grad_norm": 0.0019957320764660835,
-      "learning_rate": 0.00019999843740776943,
-      "loss": 46.0,
-      "step": 23284
-    },
-    {
-      "epoch": 1.7803008582296385,
-      "grad_norm": 0.000458067370345816,
-      "learning_rate": 0.00019999843727348896,
-      "loss": 46.0,
-      "step": 23285
-    },
-    {
-      "epoch": 1.7803773152130282,
-      "grad_norm": 0.0007111165323294699,
-      "learning_rate": 0.00019999843713920274,
-      "loss": 46.0,
-      "step": 23286
-    },
-    {
-      "epoch": 1.780453772196418,
-      "grad_norm": 0.0012318477965891361,
-      "learning_rate": 0.00019999843700491075,
-      "loss": 46.0,
-      "step": 23287
-    },
-    {
-      "epoch": 1.7805302291798077,
-      "grad_norm": 0.005296248011291027,
-      "learning_rate": 0.00019999843687061296,
-      "loss": 46.0,
-      "step": 23288
-    },
-    {
-      "epoch": 1.7806066861631975,
-      "grad_norm": 0.005017813760787249,
-      "learning_rate": 0.00019999843673630942,
-      "loss": 46.0,
-      "step": 23289
-    },
-    {
-      "epoch": 1.7806831431465873,
-      "grad_norm": 0.0015848270850256085,
-      "learning_rate": 0.0001999984366020001,
-      "loss": 46.0,
-      "step": 23290
-    },
-    {
-      "epoch": 1.780759600129977,
-      "grad_norm": 0.0003489513765089214,
-      "learning_rate": 0.00019999843646768505,
-      "loss": 46.0,
-      "step": 23291
-    },
-    {
-      "epoch": 1.7808360571133666,
-      "grad_norm": 0.0017636175034567714,
-      "learning_rate": 0.0001999984363333642,
-      "loss": 46.0,
-      "step": 23292
-    },
-    {
-      "epoch": 1.7809125140967563,
-      "grad_norm": 0.0009710597805678844,
-      "learning_rate": 0.0001999984361990376,
-      "loss": 46.0,
-      "step": 23293
-    },
-    {
-      "epoch": 1.780988971080146,
-      "grad_norm": 0.0011906478321179748,
-      "learning_rate": 0.00019999843606470518,
-      "loss": 46.0,
-      "step": 23294
-    },
-    {
-      "epoch": 1.7810654280635356,
-      "grad_norm": 0.00041255855467170477,
-      "learning_rate": 0.00019999843593036703,
-      "loss": 46.0,
-      "step": 23295
-    },
-    {
-      "epoch": 1.7811418850469254,
-      "grad_norm": 0.0012901457957923412,
-      "learning_rate": 0.0001999984357960231,
-      "loss": 46.0,
-      "step": 23296
-    },
-    {
-      "epoch": 1.7812183420303151,
-      "grad_norm": 0.0015257798368111253,
-      "learning_rate": 0.0001999984356616734,
-      "loss": 46.0,
-      "step": 23297
-    },
-    {
-      "epoch": 1.781294799013705,
-      "grad_norm": 0.0012141375336796045,
-      "learning_rate": 0.00019999843552731794,
-      "loss": 46.0,
-      "step": 23298
-    },
-    {
-      "epoch": 1.7813712559970947,
-      "grad_norm": 0.001517360913567245,
-      "learning_rate": 0.00019999843539295672,
-      "loss": 46.0,
-      "step": 23299
-    },
-    {
-      "epoch": 1.7814477129804844,
-      "grad_norm": 0.0008944621658883989,
-      "learning_rate": 0.00019999843525858973,
-      "loss": 46.0,
-      "step": 23300
-    },
-    {
-      "epoch": 1.7815241699638742,
-      "grad_norm": 0.006637830287218094,
-      "learning_rate": 0.00019999843512421696,
-      "loss": 46.0,
-      "step": 23301
-    },
-    {
-      "epoch": 1.781600626947264,
-      "grad_norm": 0.0010529521387070417,
-      "learning_rate": 0.00019999843498983843,
-      "loss": 46.0,
-      "step": 23302
-    },
-    {
-      "epoch": 1.7816770839306535,
-      "grad_norm": 0.0012665257090702653,
-      "learning_rate": 0.0001999984348554541,
-      "loss": 46.0,
-      "step": 23303
-    },
-    {
-      "epoch": 1.7817535409140433,
-      "grad_norm": 0.0023495457135140896,
-      "learning_rate": 0.00019999843472106403,
-      "loss": 46.0,
-      "step": 23304
-    },
-    {
-      "epoch": 1.7818299978974328,
-      "grad_norm": 0.0011543849250301719,
-      "learning_rate": 0.00019999843458666817,
-      "loss": 46.0,
-      "step": 23305
-    },
-    {
-      "epoch": 1.7819064548808226,
-      "grad_norm": 0.005076871253550053,
-      "learning_rate": 0.00019999843445226657,
-      "loss": 46.0,
-      "step": 23306
-    },
-    {
-      "epoch": 1.7819829118642123,
-      "grad_norm": 0.0008754977607168257,
-      "learning_rate": 0.00019999843431785916,
-      "loss": 46.0,
-      "step": 23307
-    },
-    {
-      "epoch": 1.782059368847602,
-      "grad_norm": 0.00356099265627563,
-      "learning_rate": 0.00019999843418344598,
-      "loss": 46.0,
-      "step": 23308
-    },
-    {
-      "epoch": 1.7821358258309918,
-      "grad_norm": 0.0007451626588590443,
-      "learning_rate": 0.00019999843404902709,
-      "loss": 46.0,
-      "step": 23309
-    },
-    {
-      "epoch": 1.7822122828143816,
-      "grad_norm": 0.0005000513046979904,
-      "learning_rate": 0.0001999984339146024,
-      "loss": 46.0,
-      "step": 23310
-    },
-    {
-      "epoch": 1.7822887397977714,
-      "grad_norm": 0.000997535651549697,
-      "learning_rate": 0.00019999843378017192,
-      "loss": 46.0,
-      "step": 23311
-    },
-    {
-      "epoch": 1.7823651967811611,
-      "grad_norm": 0.0013332013040781021,
-      "learning_rate": 0.00019999843364573567,
-      "loss": 46.0,
-      "step": 23312
-    },
-    {
-      "epoch": 1.7824416537645509,
-      "grad_norm": 0.0013277566758915782,
-      "learning_rate": 0.00019999843351129368,
-      "loss": 46.0,
-      "step": 23313
-    },
-    {
-      "epoch": 1.7825181107479404,
-      "grad_norm": 0.003483612323179841,
-      "learning_rate": 0.00019999843337684592,
-      "loss": 46.0,
-      "step": 23314
-    },
-    {
-      "epoch": 1.7825945677313302,
-      "grad_norm": 0.001168434857390821,
-      "learning_rate": 0.00019999843324239235,
-      "loss": 46.0,
-      "step": 23315
-    },
-    {
-      "epoch": 1.7826710247147197,
-      "grad_norm": 0.004102638456970453,
-      "learning_rate": 0.00019999843310793307,
-      "loss": 46.0,
-      "step": 23316
-    },
-    {
-      "epoch": 1.7827474816981095,
-      "grad_norm": 0.0017078651580959558,
-      "learning_rate": 0.00019999843297346796,
-      "loss": 46.0,
-      "step": 23317
-    },
-    {
-      "epoch": 1.7828239386814992,
-      "grad_norm": 0.0015993935521692038,
-      "learning_rate": 0.00019999843283899713,
-      "loss": 46.0,
-      "step": 23318
-    },
-    {
-      "epoch": 1.782900395664889,
-      "grad_norm": 0.0006118700257502496,
-      "learning_rate": 0.00019999843270452052,
-      "loss": 46.0,
-      "step": 23319
-    },
-    {
-      "epoch": 1.7829768526482788,
-      "grad_norm": 0.0008508054888807237,
-      "learning_rate": 0.00019999843257003812,
-      "loss": 46.0,
-      "step": 23320
-    },
-    {
-      "epoch": 1.7830533096316685,
-      "grad_norm": 0.0012558602029457688,
-      "learning_rate": 0.00019999843243554994,
-      "loss": 46.0,
-      "step": 23321
-    },
-    {
-      "epoch": 1.7831297666150583,
-      "grad_norm": 0.0010096827754750848,
-      "learning_rate": 0.00019999843230105601,
-      "loss": 46.0,
-      "step": 23322
-    },
-    {
-      "epoch": 1.783206223598448,
-      "grad_norm": 0.0009758556261658669,
-      "learning_rate": 0.00019999843216655634,
-      "loss": 46.0,
-      "step": 23323
-    },
-    {
-      "epoch": 1.7832826805818378,
-      "grad_norm": 0.0032229137141257524,
-      "learning_rate": 0.00019999843203205087,
-      "loss": 46.0,
-      "step": 23324
-    },
-    {
-      "epoch": 1.7833591375652273,
-      "grad_norm": 0.0020327342208474874,
-      "learning_rate": 0.00019999843189753963,
-      "loss": 46.0,
-      "step": 23325
-    },
-    {
-      "epoch": 1.783435594548617,
-      "grad_norm": 0.0008022714173421264,
-      "learning_rate": 0.0001999984317630226,
-      "loss": 46.0,
-      "step": 23326
-    },
-    {
-      "epoch": 1.7835120515320066,
-      "grad_norm": 0.001887103426270187,
-      "learning_rate": 0.00019999843162849985,
-      "loss": 46.0,
-      "step": 23327
-    },
-    {
-      "epoch": 1.7835885085153964,
-      "grad_norm": 0.0009429011843167245,
-      "learning_rate": 0.0001999984314939713,
-      "loss": 46.0,
-      "step": 23328
-    },
-    {
-      "epoch": 1.7836649654987862,
-      "grad_norm": 0.0007383808260783553,
-      "learning_rate": 0.000199998431359437,
-      "loss": 46.0,
-      "step": 23329
-    },
-    {
-      "epoch": 1.783741422482176,
-      "grad_norm": 0.000742451346013695,
-      "learning_rate": 0.0001999984312248969,
-      "loss": 46.0,
-      "step": 23330
-    },
-    {
-      "epoch": 1.7838178794655657,
-      "grad_norm": 0.0013161482056602836,
-      "learning_rate": 0.00019999843109035106,
-      "loss": 46.0,
-      "step": 23331
-    },
-    {
-      "epoch": 1.7838943364489555,
-      "grad_norm": 0.0027356704231351614,
-      "learning_rate": 0.00019999843095579945,
-      "loss": 46.0,
-      "step": 23332
-    },
-    {
-      "epoch": 1.7839707934323452,
-      "grad_norm": 0.0010390293318778276,
-      "learning_rate": 0.00019999843082124205,
-      "loss": 46.0,
-      "step": 23333
-    },
-    {
-      "epoch": 1.784047250415735,
-      "grad_norm": 0.0035049952566623688,
-      "learning_rate": 0.0001999984306866789,
-      "loss": 46.0,
-      "step": 23334
-    },
-    {
-      "epoch": 1.7841237073991245,
-      "grad_norm": 0.0010374942794442177,
-      "learning_rate": 0.00019999843055210997,
-      "loss": 46.0,
-      "step": 23335
-    },
-    {
-      "epoch": 1.7842001643825143,
-      "grad_norm": 0.0009820755803957582,
-      "learning_rate": 0.00019999843041753525,
-      "loss": 46.0,
-      "step": 23336
-    },
-    {
-      "epoch": 1.784276621365904,
-      "grad_norm": 0.0035666734911501408,
-      "learning_rate": 0.00019999843028295478,
-      "loss": 46.0,
-      "step": 23337
-    },
-    {
-      "epoch": 1.7843530783492936,
-      "grad_norm": 0.0005721225752495229,
-      "learning_rate": 0.00019999843014836856,
-      "loss": 46.0,
-      "step": 23338
-    },
-    {
-      "epoch": 1.7844295353326833,
-      "grad_norm": 0.0029102619737386703,
-      "learning_rate": 0.00019999843001377655,
-      "loss": 46.0,
-      "step": 23339
-    },
-    {
-      "epoch": 1.784505992316073,
-      "grad_norm": 0.0008085412555374205,
-      "learning_rate": 0.00019999842987917875,
-      "loss": 46.0,
-      "step": 23340
-    },
-    {
-      "epoch": 1.7845824492994629,
-      "grad_norm": 0.0013353623216971755,
-      "learning_rate": 0.00019999842974457522,
-      "loss": 46.0,
-      "step": 23341
-    },
-    {
-      "epoch": 1.7846589062828526,
-      "grad_norm": 0.002384156221523881,
-      "learning_rate": 0.0001999984296099659,
-      "loss": 46.0,
-      "step": 23342
-    },
-    {
-      "epoch": 1.7847353632662424,
-      "grad_norm": 0.00034197737113572657,
-      "learning_rate": 0.00019999842947535082,
-      "loss": 46.0,
-      "step": 23343
-    },
-    {
-      "epoch": 1.7848118202496321,
-      "grad_norm": 0.005555443931370974,
-      "learning_rate": 0.00019999842934072997,
-      "loss": 46.0,
-      "step": 23344
-    },
-    {
-      "epoch": 1.784888277233022,
-      "grad_norm": 0.0022445947397500277,
-      "learning_rate": 0.00019999842920610334,
-      "loss": 46.0,
-      "step": 23345
-    },
-    {
-      "epoch": 1.7849647342164114,
-      "grad_norm": 0.0013989756116643548,
-      "learning_rate": 0.00019999842907147096,
-      "loss": 46.0,
-      "step": 23346
-    },
-    {
-      "epoch": 1.7850411911998012,
-      "grad_norm": 0.0013043450890108943,
-      "learning_rate": 0.0001999984289368328,
-      "loss": 46.0,
-      "step": 23347
-    },
-    {
-      "epoch": 1.785117648183191,
-      "grad_norm": 0.0011561455903574824,
-      "learning_rate": 0.00019999842880218886,
-      "loss": 46.0,
-      "step": 23348
-    },
-    {
-      "epoch": 1.7851941051665805,
-      "grad_norm": 0.0012594155268743634,
-      "learning_rate": 0.00019999842866753916,
-      "loss": 46.0,
-      "step": 23349
-    },
-    {
-      "epoch": 1.7852705621499703,
-      "grad_norm": 0.0005937193054705858,
-      "learning_rate": 0.0001999984285328837,
-      "loss": 46.0,
-      "step": 23350
-    },
-    {
-      "epoch": 1.78534701913336,
-      "grad_norm": 0.012542163021862507,
-      "learning_rate": 0.00019999842839822245,
-      "loss": 46.0,
-      "step": 23351
-    },
-    {
-      "epoch": 1.7854234761167498,
-      "grad_norm": 0.0011094792280346155,
-      "learning_rate": 0.00019999842826355543,
-      "loss": 46.0,
-      "step": 23352
-    },
-    {
-      "epoch": 1.7854999331001395,
-      "grad_norm": 0.0015730494633316994,
-      "learning_rate": 0.00019999842812888267,
-      "loss": 46.0,
-      "step": 23353
-    },
-    {
-      "epoch": 1.7855763900835293,
-      "grad_norm": 0.002759851748123765,
-      "learning_rate": 0.00019999842799420413,
-      "loss": 46.0,
-      "step": 23354
-    },
-    {
-      "epoch": 1.785652847066919,
-      "grad_norm": 0.0011959831463173032,
-      "learning_rate": 0.00019999842785951982,
-      "loss": 46.0,
-      "step": 23355
-    },
-    {
-      "epoch": 1.7857293040503088,
-      "grad_norm": 0.001111734309233725,
-      "learning_rate": 0.0001999984277248297,
-      "loss": 46.0,
-      "step": 23356
-    },
-    {
-      "epoch": 1.7858057610336984,
-      "grad_norm": 0.0008648911025375128,
-      "learning_rate": 0.00019999842759013385,
-      "loss": 46.0,
-      "step": 23357
-    },
-    {
-      "epoch": 1.7858822180170881,
-      "grad_norm": 0.0005462827393785119,
-      "learning_rate": 0.00019999842745543222,
-      "loss": 46.0,
-      "step": 23358
-    },
-    {
-      "epoch": 1.785958675000478,
-      "grad_norm": 0.0009702163515612483,
-      "learning_rate": 0.00019999842732072484,
-      "loss": 46.0,
-      "step": 23359
-    },
-    {
-      "epoch": 1.7860351319838674,
-      "grad_norm": 0.0010261706775054336,
-      "learning_rate": 0.00019999842718601167,
-      "loss": 46.0,
-      "step": 23360
-    },
-    {
-      "epoch": 1.7861115889672572,
-      "grad_norm": 0.0008456979412585497,
-      "learning_rate": 0.00019999842705129272,
-      "loss": 46.0,
-      "step": 23361
-    },
-    {
-      "epoch": 1.786188045950647,
-      "grad_norm": 0.0005397512577474117,
-      "learning_rate": 0.00019999842691656805,
-      "loss": 46.0,
-      "step": 23362
-    },
-    {
-      "epoch": 1.7862645029340367,
-      "grad_norm": 0.0019083087099716067,
-      "learning_rate": 0.00019999842678183755,
-      "loss": 46.0,
-      "step": 23363
-    },
-    {
-      "epoch": 1.7863409599174265,
-      "grad_norm": 0.0022420925088226795,
-      "learning_rate": 0.00019999842664710133,
-      "loss": 46.0,
-      "step": 23364
-    },
-    {
-      "epoch": 1.7864174169008162,
-      "grad_norm": 0.0007163463742472231,
-      "learning_rate": 0.00019999842651235932,
-      "loss": 46.0,
-      "step": 23365
-    },
-    {
-      "epoch": 1.786493873884206,
-      "grad_norm": 0.0005728755495510995,
-      "learning_rate": 0.00019999842637761153,
-      "loss": 46.0,
-      "step": 23366
-    },
-    {
-      "epoch": 1.7865703308675958,
-      "grad_norm": 0.0006644154782406986,
-      "learning_rate": 0.000199998426242858,
-      "loss": 46.0,
-      "step": 23367
-    },
-    {
-      "epoch": 1.7866467878509853,
-      "grad_norm": 0.000963221478741616,
-      "learning_rate": 0.00019999842610809868,
-      "loss": 46.0,
-      "step": 23368
-    },
-    {
-      "epoch": 1.786723244834375,
-      "grad_norm": 0.00035833221045322716,
-      "learning_rate": 0.00019999842597333357,
-      "loss": 46.0,
-      "step": 23369
-    },
-    {
-      "epoch": 1.7867997018177648,
-      "grad_norm": 0.0007129181176424026,
-      "learning_rate": 0.00019999842583856271,
-      "loss": 46.0,
-      "step": 23370
-    },
-    {
-      "epoch": 1.7868761588011544,
-      "grad_norm": 0.001610266393981874,
-      "learning_rate": 0.00019999842570378608,
-      "loss": 46.0,
-      "step": 23371
-    },
-    {
-      "epoch": 1.7869526157845441,
-      "grad_norm": 0.0009280890808440745,
-      "learning_rate": 0.0001999984255690037,
-      "loss": 46.0,
-      "step": 23372
-    },
-    {
-      "epoch": 1.7870290727679339,
-      "grad_norm": 0.004658679943531752,
-      "learning_rate": 0.00019999842543421553,
-      "loss": 46.0,
-      "step": 23373
-    },
-    {
-      "epoch": 1.7871055297513236,
-      "grad_norm": 0.007147517986595631,
-      "learning_rate": 0.00019999842529942158,
-      "loss": 46.0,
-      "step": 23374
-    },
-    {
-      "epoch": 1.7871819867347134,
-      "grad_norm": 0.001862075412645936,
-      "learning_rate": 0.0001999984251646219,
-      "loss": 46.0,
-      "step": 23375
-    },
-    {
-      "epoch": 1.7872584437181032,
-      "grad_norm": 0.0012501366436481476,
-      "learning_rate": 0.00019999842502981641,
-      "loss": 46.0,
-      "step": 23376
-    },
-    {
-      "epoch": 1.787334900701493,
-      "grad_norm": 0.0007985960692167282,
-      "learning_rate": 0.00019999842489500517,
-      "loss": 46.0,
-      "step": 23377
-    },
-    {
-      "epoch": 1.7874113576848827,
-      "grad_norm": 0.0022612749598920345,
-      "learning_rate": 0.00019999842476018818,
-      "loss": 46.0,
-      "step": 23378
-    },
-    {
-      "epoch": 1.7874878146682722,
-      "grad_norm": 0.001469761598855257,
-      "learning_rate": 0.00019999842462536537,
-      "loss": 46.0,
-      "step": 23379
-    },
-    {
-      "epoch": 1.787564271651662,
-      "grad_norm": 0.0005615563131868839,
-      "learning_rate": 0.00019999842449053683,
-      "loss": 46.0,
-      "step": 23380
-    },
-    {
-      "epoch": 1.7876407286350517,
-      "grad_norm": 0.004391221329569817,
-      "learning_rate": 0.00019999842435570252,
-      "loss": 46.0,
-      "step": 23381
-    },
-    {
-      "epoch": 1.7877171856184413,
-      "grad_norm": 0.0011570961214601994,
-      "learning_rate": 0.00019999842422086244,
-      "loss": 46.0,
-      "step": 23382
-    },
-    {
-      "epoch": 1.787793642601831,
-      "grad_norm": 0.0008647372596897185,
-      "learning_rate": 0.00019999842408601658,
-      "loss": 46.0,
-      "step": 23383
-    },
-    {
-      "epoch": 1.7878700995852208,
-      "grad_norm": 0.00046265454147942364,
-      "learning_rate": 0.00019999842395116492,
-      "loss": 46.0,
-      "step": 23384
-    },
-    {
-      "epoch": 1.7879465565686106,
-      "grad_norm": 0.0007150240708142519,
-      "learning_rate": 0.00019999842381630755,
-      "loss": 46.0,
-      "step": 23385
-    },
-    {
-      "epoch": 1.7880230135520003,
-      "grad_norm": 0.0008370680152438581,
-      "learning_rate": 0.00019999842368144437,
-      "loss": 46.0,
-      "step": 23386
-    },
-    {
-      "epoch": 1.78809947053539,
-      "grad_norm": 0.0006866789772175252,
-      "learning_rate": 0.00019999842354657542,
-      "loss": 46.0,
-      "step": 23387
-    },
-    {
-      "epoch": 1.7881759275187799,
-      "grad_norm": 0.0010378382867202163,
-      "learning_rate": 0.00019999842341170073,
-      "loss": 46.0,
-      "step": 23388
-    },
-    {
-      "epoch": 1.7882523845021696,
-      "grad_norm": 0.0016210711328312755,
-      "learning_rate": 0.00019999842327682026,
-      "loss": 46.0,
-      "step": 23389
-    },
-    {
-      "epoch": 1.7883288414855592,
-      "grad_norm": 0.0009405912132933736,
-      "learning_rate": 0.00019999842314193399,
-      "loss": 46.0,
-      "step": 23390
-    },
-    {
-      "epoch": 1.788405298468949,
-      "grad_norm": 0.0015011117793619633,
-      "learning_rate": 0.000199998423007042,
-      "loss": 46.0,
-      "step": 23391
-    },
-    {
-      "epoch": 1.7884817554523387,
-      "grad_norm": 0.0016328402562066913,
-      "learning_rate": 0.0001999984228721442,
-      "loss": 46.0,
-      "step": 23392
-    },
-    {
-      "epoch": 1.7885582124357282,
-      "grad_norm": 0.0017886487767100334,
-      "learning_rate": 0.00019999842273724064,
-      "loss": 46.0,
-      "step": 23393
-    },
-    {
-      "epoch": 1.788634669419118,
-      "grad_norm": 0.0006534748245030642,
-      "learning_rate": 0.00019999842260233134,
-      "loss": 46.0,
-      "step": 23394
-    },
-    {
-      "epoch": 1.7887111264025077,
-      "grad_norm": 0.0036050642374902964,
-      "learning_rate": 0.00019999842246741625,
-      "loss": 46.0,
-      "step": 23395
-    },
-    {
-      "epoch": 1.7887875833858975,
-      "grad_norm": 0.002600611187517643,
-      "learning_rate": 0.00019999842233249537,
-      "loss": 46.0,
-      "step": 23396
-    },
-    {
-      "epoch": 1.7888640403692873,
-      "grad_norm": 0.001363143790513277,
-      "learning_rate": 0.00019999842219756874,
-      "loss": 46.0,
-      "step": 23397
-    },
-    {
-      "epoch": 1.788940497352677,
-      "grad_norm": 0.0023228079080581665,
-      "learning_rate": 0.00019999842206263634,
-      "loss": 46.0,
-      "step": 23398
-    },
-    {
-      "epoch": 1.7890169543360668,
-      "grad_norm": 0.0024469674099236727,
-      "learning_rate": 0.0001999984219276982,
-      "loss": 46.0,
-      "step": 23399
-    },
-    {
-      "epoch": 1.7890934113194565,
-      "grad_norm": 0.0010389079106971622,
-      "learning_rate": 0.00019999842179275424,
-      "loss": 46.0,
-      "step": 23400
-    },
-    {
-      "epoch": 1.789169868302846,
-      "grad_norm": 0.0008803008240647614,
-      "learning_rate": 0.00019999842165780454,
-      "loss": 46.0,
-      "step": 23401
-    },
-    {
-      "epoch": 1.7892463252862358,
-      "grad_norm": 0.0020320320036262274,
-      "learning_rate": 0.00019999842152284907,
-      "loss": 46.0,
-      "step": 23402
-    },
-    {
-      "epoch": 1.7893227822696256,
-      "grad_norm": 0.0008957046666182578,
-      "learning_rate": 0.00019999842138788783,
-      "loss": 46.0,
-      "step": 23403
-    },
-    {
-      "epoch": 1.7893992392530151,
-      "grad_norm": 0.0007960941875353456,
-      "learning_rate": 0.00019999842125292082,
-      "loss": 46.0,
-      "step": 23404
-    },
-    {
-      "epoch": 1.789475696236405,
-      "grad_norm": 0.00265374593436718,
-      "learning_rate": 0.00019999842111794803,
-      "loss": 46.0,
-      "step": 23405
-    },
-    {
-      "epoch": 1.7895521532197947,
-      "grad_norm": 0.006574220489710569,
-      "learning_rate": 0.00019999842098296946,
-      "loss": 46.0,
-      "step": 23406
-    },
-    {
-      "epoch": 1.7896286102031844,
-      "grad_norm": 0.001125707058236003,
-      "learning_rate": 0.00019999842084798513,
-      "loss": 46.0,
-      "step": 23407
-    },
-    {
-      "epoch": 1.7897050671865742,
-      "grad_norm": 0.0005455115460790694,
-      "learning_rate": 0.00019999842071299505,
-      "loss": 46.0,
-      "step": 23408
-    },
-    {
-      "epoch": 1.789781524169964,
-      "grad_norm": 0.0018281489610671997,
-      "learning_rate": 0.0001999984205779992,
-      "loss": 46.0,
-      "step": 23409
-    },
-    {
-      "epoch": 1.7898579811533537,
-      "grad_norm": 0.0008425092091783881,
-      "learning_rate": 0.00019999842044299753,
-      "loss": 46.0,
-      "step": 23410
-    },
-    {
-      "epoch": 1.7899344381367435,
-      "grad_norm": 0.001513608731329441,
-      "learning_rate": 0.00019999842030799013,
-      "loss": 46.0,
-      "step": 23411
-    },
-    {
-      "epoch": 1.790010895120133,
-      "grad_norm": 0.0036630851682275534,
-      "learning_rate": 0.00019999842017297698,
-      "loss": 46.0,
-      "step": 23412
-    },
-    {
-      "epoch": 1.7900873521035228,
-      "grad_norm": 0.0008791261352598667,
-      "learning_rate": 0.00019999842003795803,
-      "loss": 46.0,
-      "step": 23413
-    },
-    {
-      "epoch": 1.7901638090869125,
-      "grad_norm": 0.0009894762188196182,
-      "learning_rate": 0.0001999984199029333,
-      "loss": 46.0,
-      "step": 23414
-    },
-    {
-      "epoch": 1.790240266070302,
-      "grad_norm": 0.0010719248093664646,
-      "learning_rate": 0.00019999841976790284,
-      "loss": 46.0,
-      "step": 23415
-    },
-    {
-      "epoch": 1.7903167230536918,
-      "grad_norm": 0.003220104146748781,
-      "learning_rate": 0.00019999841963286657,
-      "loss": 46.0,
-      "step": 23416
-    },
-    {
-      "epoch": 1.7903931800370816,
-      "grad_norm": 0.0007135960622690618,
-      "learning_rate": 0.00019999841949782456,
-      "loss": 46.0,
-      "step": 23417
-    },
-    {
-      "epoch": 1.7904696370204713,
-      "grad_norm": 0.0009460801375098526,
-      "learning_rate": 0.0001999984193627768,
-      "loss": 46.0,
-      "step": 23418
-    },
-    {
-      "epoch": 1.790546094003861,
-      "grad_norm": 0.004054906778037548,
-      "learning_rate": 0.00019999841922772323,
-      "loss": 46.0,
-      "step": 23419
-    },
-    {
-      "epoch": 1.7906225509872509,
-      "grad_norm": 0.00043670253944583237,
-      "learning_rate": 0.0001999984190926639,
-      "loss": 46.0,
-      "step": 23420
-    },
-    {
-      "epoch": 1.7906990079706406,
-      "grad_norm": 0.0010652946075424552,
-      "learning_rate": 0.00019999841895759884,
-      "loss": 46.0,
-      "step": 23421
-    },
-    {
-      "epoch": 1.7907754649540304,
-      "grad_norm": 0.0009400029084645212,
-      "learning_rate": 0.00019999841882252793,
-      "loss": 46.0,
-      "step": 23422
-    },
-    {
-      "epoch": 1.79085192193742,
-      "grad_norm": 0.0036266951356083155,
-      "learning_rate": 0.00019999841868745133,
-      "loss": 46.0,
-      "step": 23423
-    },
-    {
-      "epoch": 1.7909283789208097,
-      "grad_norm": 0.0013853814452886581,
-      "learning_rate": 0.0001999984185523689,
-      "loss": 46.0,
-      "step": 23424
-    },
-    {
-      "epoch": 1.7910048359041995,
-      "grad_norm": 0.002084429142996669,
-      "learning_rate": 0.00019999841841728073,
-      "loss": 46.0,
-      "step": 23425
-    },
-    {
-      "epoch": 1.791081292887589,
-      "grad_norm": 0.000407147133955732,
-      "learning_rate": 0.0001999984182821868,
-      "loss": 46.0,
-      "step": 23426
-    },
-    {
-      "epoch": 1.7911577498709788,
-      "grad_norm": 0.0008450127788819373,
-      "learning_rate": 0.00019999841814708708,
-      "loss": 46.0,
-      "step": 23427
-    },
-    {
-      "epoch": 1.7912342068543685,
-      "grad_norm": 0.0021378525998443365,
-      "learning_rate": 0.00019999841801198161,
-      "loss": 46.0,
-      "step": 23428
-    },
-    {
-      "epoch": 1.7913106638377583,
-      "grad_norm": 0.0005242145271040499,
-      "learning_rate": 0.00019999841787687035,
-      "loss": 46.0,
-      "step": 23429
-    },
-    {
-      "epoch": 1.791387120821148,
-      "grad_norm": 0.00328258378431201,
-      "learning_rate": 0.00019999841774175333,
-      "loss": 46.0,
-      "step": 23430
-    },
-    {
-      "epoch": 1.7914635778045378,
-      "grad_norm": 0.0004819945024792105,
-      "learning_rate": 0.00019999841760663057,
-      "loss": 46.0,
-      "step": 23431
-    },
-    {
-      "epoch": 1.7915400347879276,
-      "grad_norm": 0.001724945497699082,
-      "learning_rate": 0.00019999841747150198,
-      "loss": 46.0,
-      "step": 23432
-    },
-    {
-      "epoch": 1.7916164917713173,
-      "grad_norm": 0.0007490586722269654,
-      "learning_rate": 0.00019999841733636767,
-      "loss": 46.0,
-      "step": 23433
-    },
-    {
-      "epoch": 1.7916929487547069,
-      "grad_norm": 0.0010977743659168482,
-      "learning_rate": 0.00019999841720122756,
-      "loss": 46.0,
-      "step": 23434
-    },
-    {
-      "epoch": 1.7917694057380966,
-      "grad_norm": 0.0012437679106369615,
-      "learning_rate": 0.0001999984170660817,
-      "loss": 46.0,
-      "step": 23435
-    },
-    {
-      "epoch": 1.7918458627214862,
-      "grad_norm": 0.002973268274217844,
-      "learning_rate": 0.00019999841693093005,
-      "loss": 46.0,
-      "step": 23436
-    },
-    {
-      "epoch": 1.791922319704876,
-      "grad_norm": 0.003718478139489889,
-      "learning_rate": 0.00019999841679577268,
-      "loss": 46.0,
-      "step": 23437
-    },
-    {
-      "epoch": 1.7919987766882657,
-      "grad_norm": 0.0019366148626431823,
-      "learning_rate": 0.00019999841666060948,
-      "loss": 46.0,
-      "step": 23438
-    },
-    {
-      "epoch": 1.7920752336716554,
-      "grad_norm": 0.007893499918282032,
-      "learning_rate": 0.00019999841652544053,
-      "loss": 46.0,
-      "step": 23439
-    },
-    {
-      "epoch": 1.7921516906550452,
-      "grad_norm": 0.0017065148567780852,
-      "learning_rate": 0.00019999841639026583,
-      "loss": 46.0,
-      "step": 23440
-    },
-    {
-      "epoch": 1.792228147638435,
-      "grad_norm": 0.0012645057868212461,
-      "learning_rate": 0.00019999841625508536,
-      "loss": 46.0,
-      "step": 23441
-    },
-    {
-      "epoch": 1.7923046046218247,
-      "grad_norm": 0.001027383143082261,
-      "learning_rate": 0.00019999841611989912,
-      "loss": 46.0,
-      "step": 23442
-    },
-    {
-      "epoch": 1.7923810616052145,
-      "grad_norm": 0.0007045645616017282,
-      "learning_rate": 0.00019999841598470708,
-      "loss": 46.0,
-      "step": 23443
-    },
-    {
-      "epoch": 1.7924575185886042,
-      "grad_norm": 0.0009280061349272728,
-      "learning_rate": 0.0001999984158495093,
-      "loss": 46.0,
-      "step": 23444
-    },
-    {
-      "epoch": 1.7925339755719938,
-      "grad_norm": 0.0012330528115853667,
-      "learning_rate": 0.00019999841571430573,
-      "loss": 46.0,
-      "step": 23445
-    },
-    {
-      "epoch": 1.7926104325553835,
-      "grad_norm": 0.0005882453988306224,
-      "learning_rate": 0.0001999984155790964,
-      "loss": 46.0,
-      "step": 23446
-    },
-    {
-      "epoch": 1.792686889538773,
-      "grad_norm": 0.0041147139854729176,
-      "learning_rate": 0.0001999984154438813,
-      "loss": 46.0,
-      "step": 23447
-    },
-    {
-      "epoch": 1.7927633465221628,
-      "grad_norm": 0.0011442916002124548,
-      "learning_rate": 0.00019999841530866043,
-      "loss": 46.0,
-      "step": 23448
-    },
-    {
-      "epoch": 1.7928398035055526,
-      "grad_norm": 0.0006084247725084424,
-      "learning_rate": 0.0001999984151734338,
-      "loss": 46.0,
-      "step": 23449
-    },
-    {
-      "epoch": 1.7929162604889424,
-      "grad_norm": 0.004325206391513348,
-      "learning_rate": 0.0001999984150382014,
-      "loss": 46.0,
-      "step": 23450
-    },
-    {
-      "epoch": 1.7929927174723321,
-      "grad_norm": 0.0011547714238986373,
-      "learning_rate": 0.00019999841490296323,
-      "loss": 46.0,
-      "step": 23451
-    },
-    {
-      "epoch": 1.793069174455722,
-      "grad_norm": 0.0007342167082242668,
-      "learning_rate": 0.00019999841476771928,
-      "loss": 46.0,
-      "step": 23452
-    },
-    {
-      "epoch": 1.7931456314391117,
-      "grad_norm": 0.0003476165875326842,
-      "learning_rate": 0.00019999841463246956,
-      "loss": 46.0,
-      "step": 23453
-    },
-    {
-      "epoch": 1.7932220884225014,
-      "grad_norm": 0.002937949961051345,
-      "learning_rate": 0.00019999841449721406,
-      "loss": 46.0,
-      "step": 23454
-    },
-    {
-      "epoch": 1.7932985454058912,
-      "grad_norm": 0.0038411268033087254,
-      "learning_rate": 0.00019999841436195282,
-      "loss": 46.0,
-      "step": 23455
-    },
-    {
-      "epoch": 1.7933750023892807,
-      "grad_norm": 0.000800565758254379,
-      "learning_rate": 0.00019999841422668578,
-      "loss": 46.0,
-      "step": 23456
-    },
-    {
-      "epoch": 1.7934514593726705,
-      "grad_norm": 0.00106425816193223,
-      "learning_rate": 0.000199998414091413,
-      "loss": 46.0,
-      "step": 23457
-    },
-    {
-      "epoch": 1.79352791635606,
-      "grad_norm": 0.0009492255048826337,
-      "learning_rate": 0.00019999841395613443,
-      "loss": 46.0,
-      "step": 23458
-    },
-    {
-      "epoch": 1.7936043733394498,
-      "grad_norm": 0.00187025242485106,
-      "learning_rate": 0.00019999841382085012,
-      "loss": 46.0,
-      "step": 23459
-    },
-    {
-      "epoch": 1.7936808303228395,
-      "grad_norm": 0.0013016454176977277,
-      "learning_rate": 0.00019999841368556002,
-      "loss": 46.0,
-      "step": 23460
-    },
-    {
-      "epoch": 1.7937572873062293,
-      "grad_norm": 0.00206198962405324,
-      "learning_rate": 0.00019999841355026413,
-      "loss": 46.0,
-      "step": 23461
-    },
-    {
-      "epoch": 1.793833744289619,
-      "grad_norm": 0.004766110796481371,
-      "learning_rate": 0.0001999984134149625,
-      "loss": 46.0,
-      "step": 23462
-    },
-    {
-      "epoch": 1.7939102012730088,
-      "grad_norm": 0.0013337443815544248,
-      "learning_rate": 0.0001999984132796551,
-      "loss": 46.0,
-      "step": 23463
-    },
-    {
-      "epoch": 1.7939866582563986,
-      "grad_norm": 0.007464942522346973,
-      "learning_rate": 0.00019999841314434193,
-      "loss": 46.0,
-      "step": 23464
-    },
-    {
-      "epoch": 1.7940631152397883,
-      "grad_norm": 0.0017536778468638659,
-      "learning_rate": 0.00019999841300902298,
-      "loss": 46.0,
-      "step": 23465
-    },
-    {
-      "epoch": 1.7941395722231779,
-      "grad_norm": 0.0012567124795168638,
-      "learning_rate": 0.00019999841287369826,
-      "loss": 46.0,
-      "step": 23466
-    },
-    {
-      "epoch": 1.7942160292065676,
-      "grad_norm": 0.0011617616983130574,
-      "learning_rate": 0.00019999841273836777,
-      "loss": 46.0,
-      "step": 23467
-    },
-    {
-      "epoch": 1.7942924861899574,
-      "grad_norm": 0.0019567653071135283,
-      "learning_rate": 0.00019999841260303153,
-      "loss": 46.0,
-      "step": 23468
-    },
-    {
-      "epoch": 1.794368943173347,
-      "grad_norm": 0.0007796279969625175,
-      "learning_rate": 0.00019999841246768948,
-      "loss": 46.0,
-      "step": 23469
-    },
-    {
-      "epoch": 1.7944454001567367,
-      "grad_norm": 0.0009487696806900203,
-      "learning_rate": 0.0001999984123323417,
-      "loss": 46.0,
-      "step": 23470
-    },
-    {
-      "epoch": 1.7945218571401265,
-      "grad_norm": 0.0008429872686974704,
-      "learning_rate": 0.00019999841219698814,
-      "loss": 46.0,
-      "step": 23471
-    },
-    {
-      "epoch": 1.7945983141235162,
-      "grad_norm": 0.0013688217150047421,
-      "learning_rate": 0.0001999984120616288,
-      "loss": 46.0,
-      "step": 23472
-    },
-    {
-      "epoch": 1.794674771106906,
-      "grad_norm": 0.0008438929216936231,
-      "learning_rate": 0.0001999984119262637,
-      "loss": 46.0,
-      "step": 23473
-    },
-    {
-      "epoch": 1.7947512280902957,
-      "grad_norm": 0.002165168523788452,
-      "learning_rate": 0.0001999984117908928,
-      "loss": 46.0,
-      "step": 23474
-    },
-    {
-      "epoch": 1.7948276850736855,
-      "grad_norm": 0.0009734886698424816,
-      "learning_rate": 0.00019999841165551619,
-      "loss": 46.0,
-      "step": 23475
-    },
-    {
-      "epoch": 1.7949041420570753,
-      "grad_norm": 0.001065479009412229,
-      "learning_rate": 0.00019999841152013378,
-      "loss": 46.0,
-      "step": 23476
-    },
-    {
-      "epoch": 1.7949805990404648,
-      "grad_norm": 0.0009766994044184685,
-      "learning_rate": 0.0001999984113847456,
-      "loss": 46.0,
-      "step": 23477
-    },
-    {
-      "epoch": 1.7950570560238546,
-      "grad_norm": 0.0027736565098166466,
-      "learning_rate": 0.00019999841124935164,
-      "loss": 46.0,
-      "step": 23478
-    },
-    {
-      "epoch": 1.7951335130072443,
-      "grad_norm": 0.0007543741376139224,
-      "learning_rate": 0.00019999841111395194,
-      "loss": 46.0,
-      "step": 23479
-    },
-    {
-      "epoch": 1.7952099699906339,
-      "grad_norm": 0.0010395781137049198,
-      "learning_rate": 0.00019999841097854645,
-      "loss": 46.0,
-      "step": 23480
-    },
-    {
-      "epoch": 1.7952864269740236,
-      "grad_norm": 0.0012976148864254355,
-      "learning_rate": 0.0001999984108431352,
-      "loss": 46.0,
-      "step": 23481
-    },
-    {
-      "epoch": 1.7953628839574134,
-      "grad_norm": 0.0012276128400117159,
-      "learning_rate": 0.00019999841070771817,
-      "loss": 46.0,
-      "step": 23482
-    },
-    {
-      "epoch": 1.7954393409408032,
-      "grad_norm": 0.0030859080143272877,
-      "learning_rate": 0.00019999841057229535,
-      "loss": 46.0,
-      "step": 23483
-    },
-    {
-      "epoch": 1.795515797924193,
-      "grad_norm": 0.002846098504960537,
-      "learning_rate": 0.0001999984104368668,
-      "loss": 46.0,
-      "step": 23484
-    },
-    {
-      "epoch": 1.7955922549075827,
-      "grad_norm": 0.0029819656629115343,
-      "learning_rate": 0.00019999841030143248,
-      "loss": 46.0,
-      "step": 23485
-    },
-    {
-      "epoch": 1.7956687118909724,
-      "grad_norm": 0.0012364336289465427,
-      "learning_rate": 0.00019999841016599238,
-      "loss": 46.0,
-      "step": 23486
-    },
-    {
-      "epoch": 1.7957451688743622,
-      "grad_norm": 0.0013423376949504018,
-      "learning_rate": 0.0001999984100305465,
-      "loss": 46.0,
-      "step": 23487
-    },
-    {
-      "epoch": 1.7958216258577517,
-      "grad_norm": 0.0014464695705100894,
-      "learning_rate": 0.00019999840989509484,
-      "loss": 46.0,
-      "step": 23488
-    },
-    {
-      "epoch": 1.7958980828411415,
-      "grad_norm": 0.0015231097349897027,
-      "learning_rate": 0.00019999840975963744,
-      "loss": 46.0,
-      "step": 23489
-    },
-    {
-      "epoch": 1.7959745398245313,
-      "grad_norm": 0.0007667168392799795,
-      "learning_rate": 0.00019999840962417427,
-      "loss": 46.0,
-      "step": 23490
-    },
-    {
-      "epoch": 1.7960509968079208,
-      "grad_norm": 0.004429669119417667,
-      "learning_rate": 0.0001999984094887053,
-      "loss": 46.0,
-      "step": 23491
-    },
-    {
-      "epoch": 1.7961274537913106,
-      "grad_norm": 0.0008974652155302465,
-      "learning_rate": 0.0001999984093532306,
-      "loss": 46.0,
-      "step": 23492
-    },
-    {
-      "epoch": 1.7962039107747003,
-      "grad_norm": 0.0006273578619584441,
-      "learning_rate": 0.00019999840921775008,
-      "loss": 46.0,
-      "step": 23493
-    },
-    {
-      "epoch": 1.79628036775809,
-      "grad_norm": 0.000733337365090847,
-      "learning_rate": 0.00019999840908226384,
-      "loss": 46.0,
-      "step": 23494
-    },
-    {
-      "epoch": 1.7963568247414798,
-      "grad_norm": 0.0007398317102342844,
-      "learning_rate": 0.00019999840894677182,
-      "loss": 46.0,
-      "step": 23495
-    },
-    {
-      "epoch": 1.7964332817248696,
-      "grad_norm": 0.001640309114009142,
-      "learning_rate": 0.00019999840881127404,
-      "loss": 46.0,
-      "step": 23496
-    },
-    {
-      "epoch": 1.7965097387082594,
-      "grad_norm": 0.0008160446886904538,
-      "learning_rate": 0.00019999840867577045,
-      "loss": 46.0,
-      "step": 23497
-    },
-    {
-      "epoch": 1.7965861956916491,
-      "grad_norm": 0.0003719780361279845,
-      "learning_rate": 0.00019999840854026112,
-      "loss": 46.0,
-      "step": 23498
-    },
-    {
-      "epoch": 1.7966626526750387,
-      "grad_norm": 0.0009321917314082384,
-      "learning_rate": 0.000199998408404746,
-      "loss": 46.0,
-      "step": 23499
-    },
-    {
-      "epoch": 1.7967391096584284,
-      "grad_norm": 0.0004980378434993327,
-      "learning_rate": 0.00019999840826922513,
-      "loss": 46.0,
-      "step": 23500
-    },
-    {
-      "epoch": 1.7968155666418182,
-      "grad_norm": 0.006817714311182499,
-      "learning_rate": 0.00019999840813369847,
-      "loss": 46.0,
-      "step": 23501
-    },
-    {
-      "epoch": 1.7968920236252077,
-      "grad_norm": 0.001043003867380321,
-      "learning_rate": 0.00019999840799816607,
-      "loss": 46.0,
-      "step": 23502
-    },
-    {
-      "epoch": 1.7969684806085975,
-      "grad_norm": 0.0025603363756090403,
-      "learning_rate": 0.0001999984078626279,
-      "loss": 46.0,
-      "step": 23503
-    },
-    {
-      "epoch": 1.7970449375919872,
-      "grad_norm": 0.0005779876373708248,
-      "learning_rate": 0.00019999840772708393,
-      "loss": 46.0,
-      "step": 23504
-    },
-    {
-      "epoch": 1.797121394575377,
-      "grad_norm": 0.0028805567417293787,
-      "learning_rate": 0.0001999984075915342,
-      "loss": 46.0,
-      "step": 23505
-    },
-    {
-      "epoch": 1.7971978515587668,
-      "grad_norm": 0.001627816236577928,
-      "learning_rate": 0.00019999840745597874,
-      "loss": 46.0,
-      "step": 23506
-    },
-    {
-      "epoch": 1.7972743085421565,
-      "grad_norm": 0.015421341173350811,
-      "learning_rate": 0.00019999840732041747,
-      "loss": 46.0,
-      "step": 23507
-    },
-    {
-      "epoch": 1.7973507655255463,
-      "grad_norm": 0.00489919213578105,
-      "learning_rate": 0.00019999840718485043,
-      "loss": 46.0,
-      "step": 23508
-    },
-    {
-      "epoch": 1.797427222508936,
-      "grad_norm": 0.00043394602835178375,
-      "learning_rate": 0.00019999840704927764,
-      "loss": 46.0,
-      "step": 23509
-    },
-    {
-      "epoch": 1.7975036794923256,
-      "grad_norm": 0.010248390026390553,
-      "learning_rate": 0.00019999840691369908,
-      "loss": 46.0,
-      "step": 23510
-    },
-    {
-      "epoch": 1.7975801364757154,
-      "grad_norm": 0.00131643854547292,
-      "learning_rate": 0.00019999840677811475,
-      "loss": 46.0,
-      "step": 23511
-    },
-    {
-      "epoch": 1.7976565934591051,
-      "grad_norm": 0.002210026141256094,
-      "learning_rate": 0.00019999840664252462,
-      "loss": 46.0,
-      "step": 23512
-    },
-    {
-      "epoch": 1.7977330504424947,
-      "grad_norm": 0.005083800759166479,
-      "learning_rate": 0.00019999840650692874,
-      "loss": 46.0,
-      "step": 23513
-    },
-    {
-      "epoch": 1.7978095074258844,
-      "grad_norm": 0.0006176887545734644,
-      "learning_rate": 0.0001999984063713271,
-      "loss": 46.0,
-      "step": 23514
-    },
-    {
-      "epoch": 1.7978859644092742,
-      "grad_norm": 0.0029231118969619274,
-      "learning_rate": 0.00019999840623571968,
-      "loss": 46.0,
-      "step": 23515
-    },
-    {
-      "epoch": 1.797962421392664,
-      "grad_norm": 0.0007739661377854645,
-      "learning_rate": 0.0001999984061001065,
-      "loss": 46.0,
-      "step": 23516
-    },
-    {
-      "epoch": 1.7980388783760537,
-      "grad_norm": 0.0006957594305276871,
-      "learning_rate": 0.00019999840596448756,
-      "loss": 46.0,
-      "step": 23517
-    },
-    {
-      "epoch": 1.7981153353594435,
-      "grad_norm": 0.0016285991296172142,
-      "learning_rate": 0.00019999840582886282,
-      "loss": 46.0,
-      "step": 23518
-    },
-    {
-      "epoch": 1.7981917923428332,
-      "grad_norm": 0.0016668487805873156,
-      "learning_rate": 0.00019999840569323235,
-      "loss": 46.0,
-      "step": 23519
-    },
-    {
-      "epoch": 1.798268249326223,
-      "grad_norm": 0.0010149258887395263,
-      "learning_rate": 0.00019999840555759608,
-      "loss": 46.0,
-      "step": 23520
-    },
-    {
-      "epoch": 1.7983447063096125,
-      "grad_norm": 0.0011890220921486616,
-      "learning_rate": 0.00019999840542195404,
-      "loss": 46.0,
-      "step": 23521
-    },
-    {
-      "epoch": 1.7984211632930023,
-      "grad_norm": 0.0018648263067007065,
-      "learning_rate": 0.00019999840528630626,
-      "loss": 46.0,
-      "step": 23522
-    },
-    {
-      "epoch": 1.798497620276392,
-      "grad_norm": 0.0007314970134757459,
-      "learning_rate": 0.00019999840515065267,
-      "loss": 46.0,
-      "step": 23523
-    },
-    {
-      "epoch": 1.7985740772597816,
-      "grad_norm": 0.0007192611810751259,
-      "learning_rate": 0.00019999840501499334,
-      "loss": 46.0,
-      "step": 23524
-    },
-    {
-      "epoch": 1.7986505342431713,
-      "grad_norm": 0.0004841941990889609,
-      "learning_rate": 0.00019999840487932823,
-      "loss": 46.0,
-      "step": 23525
-    },
-    {
-      "epoch": 1.798726991226561,
-      "grad_norm": 0.001030182815156877,
-      "learning_rate": 0.00019999840474365735,
-      "loss": 46.0,
-      "step": 23526
-    },
-    {
-      "epoch": 1.7988034482099509,
-      "grad_norm": 0.000695998955052346,
-      "learning_rate": 0.00019999840460798072,
-      "loss": 46.0,
-      "step": 23527
-    },
-    {
-      "epoch": 1.7988799051933406,
-      "grad_norm": 0.0018407469615340233,
-      "learning_rate": 0.00019999840447229827,
-      "loss": 46.0,
-      "step": 23528
-    },
-    {
-      "epoch": 1.7989563621767304,
-      "grad_norm": 0.0009694445761851966,
-      "learning_rate": 0.0001999984043366101,
-      "loss": 46.0,
-      "step": 23529
-    },
-    {
-      "epoch": 1.7990328191601201,
-      "grad_norm": 0.0011146041797474027,
-      "learning_rate": 0.00019999840420091615,
-      "loss": 46.0,
-      "step": 23530
-    },
-    {
-      "epoch": 1.79910927614351,
-      "grad_norm": 0.0015319178346544504,
-      "learning_rate": 0.0001999984040652164,
-      "loss": 46.0,
-      "step": 23531
-    },
-    {
-      "epoch": 1.7991857331268994,
-      "grad_norm": 0.0006881053559482098,
-      "learning_rate": 0.0001999984039295109,
-      "loss": 46.0,
-      "step": 23532
-    },
-    {
-      "epoch": 1.7992621901102892,
-      "grad_norm": 0.010389677248895168,
-      "learning_rate": 0.00019999840379379967,
-      "loss": 46.0,
-      "step": 23533
-    },
-    {
-      "epoch": 1.799338647093679,
-      "grad_norm": 0.0007288535707630217,
-      "learning_rate": 0.00019999840365808263,
-      "loss": 46.0,
-      "step": 23534
-    },
-    {
-      "epoch": 1.7994151040770685,
-      "grad_norm": 0.0009432890801690519,
-      "learning_rate": 0.00019999840352235985,
-      "loss": 46.0,
-      "step": 23535
-    },
-    {
-      "epoch": 1.7994915610604583,
-      "grad_norm": 0.0015950363595038652,
-      "learning_rate": 0.00019999840338663126,
-      "loss": 46.0,
-      "step": 23536
-    },
-    {
-      "epoch": 1.799568018043848,
-      "grad_norm": 0.0016075170133262873,
-      "learning_rate": 0.00019999840325089693,
-      "loss": 46.0,
-      "step": 23537
-    },
-    {
-      "epoch": 1.7996444750272378,
-      "grad_norm": 0.0010066393297165632,
-      "learning_rate": 0.00019999840311515682,
-      "loss": 46.0,
-      "step": 23538
-    },
-    {
-      "epoch": 1.7997209320106275,
-      "grad_norm": 0.000552264042198658,
-      "learning_rate": 0.0001999984029794109,
-      "loss": 46.0,
-      "step": 23539
-    },
-    {
-      "epoch": 1.7997973889940173,
-      "grad_norm": 0.0019441158510744572,
-      "learning_rate": 0.00019999840284365926,
-      "loss": 46.0,
-      "step": 23540
-    },
-    {
-      "epoch": 1.799873845977407,
-      "grad_norm": 0.0015652806032449007,
-      "learning_rate": 0.00019999840270790186,
-      "loss": 46.0,
-      "step": 23541
-    },
-    {
-      "epoch": 1.7999503029607968,
-      "grad_norm": 0.002753129694610834,
-      "learning_rate": 0.00019999840257213866,
-      "loss": 46.0,
-      "step": 23542
-    },
-    {
-      "epoch": 1.8000267599441864,
-      "grad_norm": 0.0006505911005660892,
-      "learning_rate": 0.00019999840243636972,
-      "loss": 46.0,
-      "step": 23543
-    },
-    {
-      "epoch": 1.8001032169275761,
-      "grad_norm": 0.0009638468036428094,
-      "learning_rate": 0.00019999840230059497,
-      "loss": 46.0,
-      "step": 23544
-    },
-    {
-      "epoch": 1.800179673910966,
-      "grad_norm": 0.0031288028694689274,
-      "learning_rate": 0.0001999984021648145,
-      "loss": 46.0,
-      "step": 23545
-    },
-    {
-      "epoch": 1.8002561308943554,
-      "grad_norm": 0.0016086224932223558,
-      "learning_rate": 0.0001999984020290282,
-      "loss": 46.0,
-      "step": 23546
-    },
-    {
-      "epoch": 1.8003325878777452,
-      "grad_norm": 0.00047096493653953075,
-      "learning_rate": 0.0001999984018932362,
-      "loss": 46.0,
-      "step": 23547
-    },
-    {
-      "epoch": 1.800409044861135,
-      "grad_norm": 0.0031271015759557486,
-      "learning_rate": 0.00019999840175743838,
-      "loss": 46.0,
-      "step": 23548
-    },
-    {
-      "epoch": 1.8004855018445247,
-      "grad_norm": 0.0006806131568737328,
-      "learning_rate": 0.0001999984016216348,
-      "loss": 46.0,
-      "step": 23549
-    },
-    {
-      "epoch": 1.8005619588279145,
-      "grad_norm": 0.0012186251115053892,
-      "learning_rate": 0.00019999840148582547,
-      "loss": 46.0,
-      "step": 23550
-    },
-    {
-      "epoch": 1.8006384158113042,
-      "grad_norm": 0.0007826102082617581,
-      "learning_rate": 0.00019999840135001036,
-      "loss": 46.0,
-      "step": 23551
-    },
-    {
-      "epoch": 1.800714872794694,
-      "grad_norm": 0.0013373569818213582,
-      "learning_rate": 0.00019999840121418948,
-      "loss": 46.0,
-      "step": 23552
-    },
-    {
-      "epoch": 1.8007913297780838,
-      "grad_norm": 0.0004930038121528924,
-      "learning_rate": 0.00019999840107836283,
-      "loss": 46.0,
-      "step": 23553
-    },
-    {
-      "epoch": 1.8008677867614733,
-      "grad_norm": 0.0032909975852817297,
-      "learning_rate": 0.00019999840094253043,
-      "loss": 46.0,
-      "step": 23554
-    },
-    {
-      "epoch": 1.800944243744863,
-      "grad_norm": 0.00043303988059051335,
-      "learning_rate": 0.0001999984008066922,
-      "loss": 46.0,
-      "step": 23555
-    },
-    {
-      "epoch": 1.8010207007282528,
-      "grad_norm": 0.0006659660721197724,
-      "learning_rate": 0.00019999840067084826,
-      "loss": 46.0,
-      "step": 23556
-    },
-    {
-      "epoch": 1.8010971577116424,
-      "grad_norm": 0.0016180934617295861,
-      "learning_rate": 0.00019999840053499854,
-      "loss": 46.0,
-      "step": 23557
-    },
-    {
-      "epoch": 1.8011736146950321,
-      "grad_norm": 0.0027012594509869814,
-      "learning_rate": 0.00019999840039914305,
-      "loss": 46.0,
-      "step": 23558
-    },
-    {
-      "epoch": 1.8012500716784219,
-      "grad_norm": 0.003291347064077854,
-      "learning_rate": 0.00019999840026328175,
-      "loss": 46.0,
-      "step": 23559
-    },
-    {
-      "epoch": 1.8013265286618116,
-      "grad_norm": 0.0026221624575555325,
-      "learning_rate": 0.00019999840012741474,
-      "loss": 46.0,
-      "step": 23560
-    },
-    {
-      "epoch": 1.8014029856452014,
-      "grad_norm": 0.000646628555841744,
-      "learning_rate": 0.00019999839999154193,
-      "loss": 46.0,
-      "step": 23561
-    },
-    {
-      "epoch": 1.8014794426285912,
-      "grad_norm": 0.0007230503833852708,
-      "learning_rate": 0.00019999839985566334,
-      "loss": 46.0,
-      "step": 23562
-    },
-    {
-      "epoch": 1.801555899611981,
-      "grad_norm": 0.001434436533600092,
-      "learning_rate": 0.000199998399719779,
-      "loss": 46.0,
-      "step": 23563
-    },
-    {
-      "epoch": 1.8016323565953707,
-      "grad_norm": 0.0010613644262775779,
-      "learning_rate": 0.00019999839958388888,
-      "loss": 46.0,
-      "step": 23564
-    },
-    {
-      "epoch": 1.8017088135787602,
-      "grad_norm": 0.0015314940828830004,
-      "learning_rate": 0.000199998399447993,
-      "loss": 46.0,
-      "step": 23565
-    },
-    {
-      "epoch": 1.80178527056215,
-      "grad_norm": 0.0013053305447101593,
-      "learning_rate": 0.00019999839931209135,
-      "loss": 46.0,
-      "step": 23566
-    },
-    {
-      "epoch": 1.8018617275455395,
-      "grad_norm": 0.0005981647409498692,
-      "learning_rate": 0.00019999839917618392,
-      "loss": 46.0,
-      "step": 23567
-    },
-    {
-      "epoch": 1.8019381845289293,
-      "grad_norm": 0.0009032060042954981,
-      "learning_rate": 0.00019999839904027072,
-      "loss": 46.0,
-      "step": 23568
-    },
-    {
-      "epoch": 1.802014641512319,
-      "grad_norm": 0.0013273843796923757,
-      "learning_rate": 0.00019999839890435178,
-      "loss": 46.0,
-      "step": 23569
-    },
-    {
-      "epoch": 1.8020910984957088,
-      "grad_norm": 0.001075581181794405,
-      "learning_rate": 0.00019999839876842706,
-      "loss": 46.0,
-      "step": 23570
-    },
-    {
-      "epoch": 1.8021675554790986,
-      "grad_norm": 0.0008504153811372817,
-      "learning_rate": 0.00019999839863249656,
-      "loss": 46.0,
-      "step": 23571
-    },
-    {
-      "epoch": 1.8022440124624883,
-      "grad_norm": 0.0007186035509221256,
-      "learning_rate": 0.00019999839849656027,
-      "loss": 46.0,
-      "step": 23572
-    },
-    {
-      "epoch": 1.802320469445878,
-      "grad_norm": 0.0009708944126032293,
-      "learning_rate": 0.00019999839836061823,
-      "loss": 46.0,
-      "step": 23573
-    },
-    {
-      "epoch": 1.8023969264292679,
-      "grad_norm": 0.0009529021917842329,
-      "learning_rate": 0.00019999839822467045,
-      "loss": 46.0,
-      "step": 23574
-    },
-    {
-      "epoch": 1.8024733834126576,
-      "grad_norm": 0.001032594358548522,
-      "learning_rate": 0.00019999839808871686,
-      "loss": 46.0,
-      "step": 23575
-    },
-    {
-      "epoch": 1.8025498403960472,
-      "grad_norm": 0.0015712508466094732,
-      "learning_rate": 0.0001999983979527575,
-      "loss": 46.0,
-      "step": 23576
-    },
-    {
-      "epoch": 1.802626297379437,
-      "grad_norm": 0.004000730346888304,
-      "learning_rate": 0.0001999983978167924,
-      "loss": 46.0,
-      "step": 23577
-    },
-    {
-      "epoch": 1.8027027543628265,
-      "grad_norm": 0.0016209562309086323,
-      "learning_rate": 0.00019999839768082152,
-      "loss": 46.0,
-      "step": 23578
-    },
-    {
-      "epoch": 1.8027792113462162,
-      "grad_norm": 0.0018562778132036328,
-      "learning_rate": 0.00019999839754484487,
-      "loss": 46.0,
-      "step": 23579
-    },
-    {
-      "epoch": 1.802855668329606,
-      "grad_norm": 0.0016734784003347158,
-      "learning_rate": 0.00019999839740886244,
-      "loss": 46.0,
-      "step": 23580
-    },
-    {
-      "epoch": 1.8029321253129957,
-      "grad_norm": 0.0007380274473689497,
-      "learning_rate": 0.00019999839727287424,
-      "loss": 46.0,
-      "step": 23581
-    },
-    {
-      "epoch": 1.8030085822963855,
-      "grad_norm": 0.0011833934113383293,
-      "learning_rate": 0.00019999839713688027,
-      "loss": 46.0,
-      "step": 23582
-    },
-    {
-      "epoch": 1.8030850392797753,
-      "grad_norm": 0.0006618915940634906,
-      "learning_rate": 0.00019999839700088055,
-      "loss": 46.0,
-      "step": 23583
-    },
-    {
-      "epoch": 1.803161496263165,
-      "grad_norm": 0.0018449613125994802,
-      "learning_rate": 0.00019999839686487506,
-      "loss": 46.0,
-      "step": 23584
-    },
-    {
-      "epoch": 1.8032379532465548,
-      "grad_norm": 0.0027561953756958246,
-      "learning_rate": 0.00019999839672886377,
-      "loss": 46.0,
-      "step": 23585
-    },
-    {
-      "epoch": 1.8033144102299445,
-      "grad_norm": 0.006247633136808872,
-      "learning_rate": 0.00019999839659284673,
-      "loss": 46.0,
-      "step": 23586
-    },
-    {
-      "epoch": 1.803390867213334,
-      "grad_norm": 0.000633818272035569,
-      "learning_rate": 0.00019999839645682395,
-      "loss": 46.0,
-      "step": 23587
-    },
-    {
-      "epoch": 1.8034673241967238,
-      "grad_norm": 0.0013962165685370564,
-      "learning_rate": 0.00019999839632079533,
-      "loss": 46.0,
-      "step": 23588
-    },
-    {
-      "epoch": 1.8035437811801134,
-      "grad_norm": 0.0009096218855120242,
-      "learning_rate": 0.000199998396184761,
-      "loss": 46.0,
-      "step": 23589
-    },
-    {
-      "epoch": 1.8036202381635031,
-      "grad_norm": 0.0009115052525885403,
-      "learning_rate": 0.00019999839604872087,
-      "loss": 46.0,
-      "step": 23590
-    },
-    {
-      "epoch": 1.803696695146893,
-      "grad_norm": 0.003917634021490812,
-      "learning_rate": 0.000199998395912675,
-      "loss": 46.0,
-      "step": 23591
-    },
-    {
-      "epoch": 1.8037731521302827,
-      "grad_norm": 0.0016916001914069057,
-      "learning_rate": 0.00019999839577662334,
-      "loss": 46.0,
-      "step": 23592
-    },
-    {
-      "epoch": 1.8038496091136724,
-      "grad_norm": 0.0010094393510371447,
-      "learning_rate": 0.00019999839564056591,
-      "loss": 46.0,
-      "step": 23593
-    },
-    {
-      "epoch": 1.8039260660970622,
-      "grad_norm": 0.0007031704299151897,
-      "learning_rate": 0.00019999839550450272,
-      "loss": 46.0,
-      "step": 23594
-    },
-    {
-      "epoch": 1.804002523080452,
-      "grad_norm": 0.000712833134457469,
-      "learning_rate": 0.00019999839536843377,
-      "loss": 46.0,
-      "step": 23595
-    },
-    {
-      "epoch": 1.8040789800638417,
-      "grad_norm": 0.003031100146472454,
-      "learning_rate": 0.00019999839523235903,
-      "loss": 46.0,
-      "step": 23596
-    },
-    {
-      "epoch": 1.8041554370472312,
-      "grad_norm": 0.000821817375253886,
-      "learning_rate": 0.00019999839509627853,
-      "loss": 46.0,
-      "step": 23597
-    },
-    {
-      "epoch": 1.804231894030621,
-      "grad_norm": 0.0025261512491852045,
-      "learning_rate": 0.00019999839496019224,
-      "loss": 46.0,
-      "step": 23598
-    },
-    {
-      "epoch": 1.8043083510140108,
-      "grad_norm": 0.0014901439426466823,
-      "learning_rate": 0.0001999983948241002,
-      "loss": 46.0,
-      "step": 23599
-    },
-    {
-      "epoch": 1.8043848079974003,
-      "grad_norm": 0.0006671230075880885,
-      "learning_rate": 0.00019999839468800242,
-      "loss": 46.0,
-      "step": 23600
-    },
-    {
-      "epoch": 1.80446126498079,
-      "grad_norm": 0.0011447911383584142,
-      "learning_rate": 0.0001999983945518988,
-      "loss": 46.0,
-      "step": 23601
-    },
-    {
-      "epoch": 1.8045377219641798,
-      "grad_norm": 0.0018249230924993753,
-      "learning_rate": 0.00019999839441578948,
-      "loss": 46.0,
-      "step": 23602
-    },
-    {
-      "epoch": 1.8046141789475696,
-      "grad_norm": 0.006038696505129337,
-      "learning_rate": 0.00019999839427967434,
-      "loss": 46.0,
-      "step": 23603
-    },
-    {
-      "epoch": 1.8046906359309594,
-      "grad_norm": 0.0023688075598329306,
-      "learning_rate": 0.00019999839414355347,
-      "loss": 46.0,
-      "step": 23604
-    },
-    {
-      "epoch": 1.8047670929143491,
-      "grad_norm": 0.00037246185820549726,
-      "learning_rate": 0.00019999839400742681,
-      "loss": 46.0,
-      "step": 23605
-    },
-    {
-      "epoch": 1.8048435498977389,
-      "grad_norm": 0.0009016448748297989,
-      "learning_rate": 0.00019999839387129436,
-      "loss": 46.0,
-      "step": 23606
-    },
-    {
-      "epoch": 1.8049200068811286,
-      "grad_norm": 0.0030539450235664845,
-      "learning_rate": 0.0001999983937351562,
-      "loss": 46.0,
-      "step": 23607
-    },
-    {
-      "epoch": 1.8049964638645182,
-      "grad_norm": 0.0016823712503537536,
-      "learning_rate": 0.00019999839359901222,
-      "loss": 46.0,
-      "step": 23608
-    },
-    {
-      "epoch": 1.805072920847908,
-      "grad_norm": 0.007601631339639425,
-      "learning_rate": 0.00019999839346286248,
-      "loss": 46.0,
-      "step": 23609
-    },
-    {
-      "epoch": 1.8051493778312977,
-      "grad_norm": 0.0005124068120494485,
-      "learning_rate": 0.00019999839332670698,
-      "loss": 46.0,
-      "step": 23610
-    },
-    {
-      "epoch": 1.8052258348146872,
-      "grad_norm": 0.0030892896465957165,
-      "learning_rate": 0.00019999839319054572,
-      "loss": 46.0,
-      "step": 23611
-    },
-    {
-      "epoch": 1.805302291798077,
-      "grad_norm": 0.0022689029574394226,
-      "learning_rate": 0.00019999839305437865,
-      "loss": 46.0,
-      "step": 23612
-    },
-    {
-      "epoch": 1.8053787487814668,
-      "grad_norm": 0.0024960017763078213,
-      "learning_rate": 0.00019999839291820584,
-      "loss": 46.0,
-      "step": 23613
-    },
-    {
-      "epoch": 1.8054552057648565,
-      "grad_norm": 0.0002980822464451194,
-      "learning_rate": 0.00019999839278202726,
-      "loss": 46.0,
-      "step": 23614
-    },
-    {
-      "epoch": 1.8055316627482463,
-      "grad_norm": 0.000677394971717149,
-      "learning_rate": 0.0001999983926458429,
-      "loss": 46.0,
-      "step": 23615
-    },
-    {
-      "epoch": 1.805608119731636,
-      "grad_norm": 0.0016386767383664846,
-      "learning_rate": 0.0001999983925096528,
-      "loss": 46.0,
-      "step": 23616
-    },
-    {
-      "epoch": 1.8056845767150258,
-      "grad_norm": 0.0006867151241749525,
-      "learning_rate": 0.0001999983923734569,
-      "loss": 46.0,
-      "step": 23617
-    },
-    {
-      "epoch": 1.8057610336984156,
-      "grad_norm": 0.004006482660770416,
-      "learning_rate": 0.00019999839223725524,
-      "loss": 46.0,
-      "step": 23618
-    },
-    {
-      "epoch": 1.805837490681805,
-      "grad_norm": 0.0009820415871217847,
-      "learning_rate": 0.00019999839210104782,
-      "loss": 46.0,
-      "step": 23619
-    },
-    {
-      "epoch": 1.8059139476651949,
-      "grad_norm": 0.000442410382675007,
-      "learning_rate": 0.00019999839196483462,
-      "loss": 46.0,
-      "step": 23620
-    },
-    {
-      "epoch": 1.8059904046485846,
-      "grad_norm": 0.0007473917794413865,
-      "learning_rate": 0.00019999839182861565,
-      "loss": 46.0,
-      "step": 23621
-    },
-    {
-      "epoch": 1.8060668616319742,
-      "grad_norm": 0.003018592484295368,
-      "learning_rate": 0.00019999839169239093,
-      "loss": 46.0,
-      "step": 23622
-    },
-    {
-      "epoch": 1.806143318615364,
-      "grad_norm": 0.0009581079357303679,
-      "learning_rate": 0.0001999983915561604,
-      "loss": 46.0,
-      "step": 23623
-    },
-    {
-      "epoch": 1.8062197755987537,
-      "grad_norm": 0.0025521027855575085,
-      "learning_rate": 0.00019999839141992415,
-      "loss": 46.0,
-      "step": 23624
-    },
-    {
-      "epoch": 1.8062962325821434,
-      "grad_norm": 0.0013378465082496405,
-      "learning_rate": 0.00019999839128368208,
-      "loss": 46.0,
-      "step": 23625
-    },
-    {
-      "epoch": 1.8063726895655332,
-      "grad_norm": 0.011976220645010471,
-      "learning_rate": 0.00019999839114743427,
-      "loss": 46.0,
-      "step": 23626
-    },
-    {
-      "epoch": 1.806449146548923,
-      "grad_norm": 0.0006340249092318118,
-      "learning_rate": 0.0001999983910111807,
-      "loss": 46.0,
-      "step": 23627
-    },
-    {
-      "epoch": 1.8065256035323127,
-      "grad_norm": 0.000979077536612749,
-      "learning_rate": 0.00019999839087492133,
-      "loss": 46.0,
-      "step": 23628
-    },
-    {
-      "epoch": 1.8066020605157025,
-      "grad_norm": 0.0005895447684451938,
-      "learning_rate": 0.00019999839073865623,
-      "loss": 46.0,
-      "step": 23629
-    },
-    {
-      "epoch": 1.806678517499092,
-      "grad_norm": 0.00047078015631996095,
-      "learning_rate": 0.00019999839060238532,
-      "loss": 46.0,
-      "step": 23630
-    },
-    {
-      "epoch": 1.8067549744824818,
-      "grad_norm": 0.006959778722375631,
-      "learning_rate": 0.00019999839046610867,
-      "loss": 46.0,
-      "step": 23631
-    },
-    {
-      "epoch": 1.8068314314658716,
-      "grad_norm": 0.0007475718157365918,
-      "learning_rate": 0.00019999839032982625,
-      "loss": 46.0,
-      "step": 23632
-    },
-    {
-      "epoch": 1.806907888449261,
-      "grad_norm": 0.0008762756478972733,
-      "learning_rate": 0.00019999839019353802,
-      "loss": 46.0,
-      "step": 23633
-    },
-    {
-      "epoch": 1.8069843454326509,
-      "grad_norm": 0.0019413064001128078,
-      "learning_rate": 0.00019999839005724405,
-      "loss": 46.0,
-      "step": 23634
-    },
-    {
-      "epoch": 1.8070608024160406,
-      "grad_norm": 0.0014487196458503604,
-      "learning_rate": 0.00019999838992094433,
-      "loss": 46.0,
-      "step": 23635
-    },
-    {
-      "epoch": 1.8071372593994304,
-      "grad_norm": 0.0013770927907899022,
-      "learning_rate": 0.00019999838978463882,
-      "loss": 46.0,
-      "step": 23636
-    },
-    {
-      "epoch": 1.8072137163828201,
-      "grad_norm": 0.0015888100024312735,
-      "learning_rate": 0.00019999838964832755,
-      "loss": 46.0,
-      "step": 23637
-    },
-    {
-      "epoch": 1.80729017336621,
-      "grad_norm": 0.0005667250952683389,
-      "learning_rate": 0.00019999838951201051,
-      "loss": 46.0,
-      "step": 23638
-    },
-    {
-      "epoch": 1.8073666303495997,
-      "grad_norm": 0.0010357708670198917,
-      "learning_rate": 0.00019999838937568768,
-      "loss": 46.0,
-      "step": 23639
-    },
-    {
-      "epoch": 1.8074430873329894,
-      "grad_norm": 0.0009489455260336399,
-      "learning_rate": 0.0001999983892393591,
-      "loss": 46.0,
-      "step": 23640
-    },
-    {
-      "epoch": 1.807519544316379,
-      "grad_norm": 0.000291152362478897,
-      "learning_rate": 0.00019999838910302474,
-      "loss": 46.0,
-      "step": 23641
-    },
-    {
-      "epoch": 1.8075960012997687,
-      "grad_norm": 0.0016290185740217566,
-      "learning_rate": 0.0001999983889666846,
-      "loss": 46.0,
-      "step": 23642
-    },
-    {
-      "epoch": 1.8076724582831585,
-      "grad_norm": 0.0036848685704171658,
-      "learning_rate": 0.00019999838883033873,
-      "loss": 46.0,
-      "step": 23643
-    },
-    {
-      "epoch": 1.807748915266548,
-      "grad_norm": 0.0005374210304580629,
-      "learning_rate": 0.00019999838869398705,
-      "loss": 46.0,
-      "step": 23644
-    },
-    {
-      "epoch": 1.8078253722499378,
-      "grad_norm": 0.008087070658802986,
-      "learning_rate": 0.00019999838855762963,
-      "loss": 46.0,
-      "step": 23645
-    },
-    {
-      "epoch": 1.8079018292333275,
-      "grad_norm": 0.0008488206076435745,
-      "learning_rate": 0.00019999838842126643,
-      "loss": 46.0,
-      "step": 23646
-    },
-    {
-      "epoch": 1.8079782862167173,
-      "grad_norm": 0.0009354506619274616,
-      "learning_rate": 0.00019999838828489746,
-      "loss": 46.0,
-      "step": 23647
-    },
-    {
-      "epoch": 1.808054743200107,
-      "grad_norm": 0.002079383935779333,
-      "learning_rate": 0.0001999983881485227,
-      "loss": 46.0,
-      "step": 23648
-    },
-    {
-      "epoch": 1.8081312001834968,
-      "grad_norm": 0.0024932650849223137,
-      "learning_rate": 0.0001999983880121422,
-      "loss": 46.0,
-      "step": 23649
-    },
-    {
-      "epoch": 1.8082076571668866,
-      "grad_norm": 0.0020582699216902256,
-      "learning_rate": 0.0001999983878757559,
-      "loss": 46.0,
-      "step": 23650
-    },
-    {
-      "epoch": 1.8082841141502763,
-      "grad_norm": 0.0011367028346285224,
-      "learning_rate": 0.00019999838773936387,
-      "loss": 46.0,
-      "step": 23651
-    },
-    {
-      "epoch": 1.8083605711336659,
-      "grad_norm": 0.0027940308209508657,
-      "learning_rate": 0.00019999838760296609,
-      "loss": 46.0,
-      "step": 23652
-    },
-    {
-      "epoch": 1.8084370281170556,
-      "grad_norm": 0.002253140090033412,
-      "learning_rate": 0.00019999838746656247,
-      "loss": 46.0,
-      "step": 23653
-    },
-    {
-      "epoch": 1.8085134851004454,
-      "grad_norm": 0.0011912768241018057,
-      "learning_rate": 0.0001999983873301531,
-      "loss": 46.0,
-      "step": 23654
-    },
-    {
-      "epoch": 1.808589942083835,
-      "grad_norm": 0.003485002089291811,
-      "learning_rate": 0.00019999838719373801,
-      "loss": 46.0,
-      "step": 23655
-    },
-    {
-      "epoch": 1.8086663990672247,
-      "grad_norm": 0.001128022326156497,
-      "learning_rate": 0.00019999838705731708,
-      "loss": 46.0,
-      "step": 23656
-    },
-    {
-      "epoch": 1.8087428560506145,
-      "grad_norm": 0.0007546091801486909,
-      "learning_rate": 0.00019999838692089043,
-      "loss": 46.0,
-      "step": 23657
-    },
-    {
-      "epoch": 1.8088193130340042,
-      "grad_norm": 0.0011466775322332978,
-      "learning_rate": 0.000199998386784458,
-      "loss": 46.0,
-      "step": 23658
-    },
-    {
-      "epoch": 1.808895770017394,
-      "grad_norm": 0.001339720212854445,
-      "learning_rate": 0.00019999838664801979,
-      "loss": 46.0,
-      "step": 23659
-    },
-    {
-      "epoch": 1.8089722270007838,
-      "grad_norm": 0.0019825256895273924,
-      "learning_rate": 0.00019999838651157582,
-      "loss": 46.0,
-      "step": 23660
-    },
-    {
-      "epoch": 1.8090486839841735,
-      "grad_norm": 0.0009286838467232883,
-      "learning_rate": 0.00019999838637512607,
-      "loss": 46.0,
-      "step": 23661
-    },
-    {
-      "epoch": 1.8091251409675633,
-      "grad_norm": 0.0008940558764152229,
-      "learning_rate": 0.00019999838623867055,
-      "loss": 46.0,
-      "step": 23662
-    },
-    {
-      "epoch": 1.8092015979509528,
-      "grad_norm": 0.00195328495465219,
-      "learning_rate": 0.0001999983861022093,
-      "loss": 46.0,
-      "step": 23663
-    },
-    {
-      "epoch": 1.8092780549343426,
-      "grad_norm": 0.0006442503654398024,
-      "learning_rate": 0.00019999838596574225,
-      "loss": 46.0,
-      "step": 23664
-    },
-    {
-      "epoch": 1.8093545119177323,
-      "grad_norm": 0.0006528267404064536,
-      "learning_rate": 0.00019999838582926942,
-      "loss": 46.0,
-      "step": 23665
-    },
-    {
-      "epoch": 1.8094309689011219,
-      "grad_norm": 0.0006126930238679051,
-      "learning_rate": 0.00019999838569279083,
-      "loss": 46.0,
-      "step": 23666
-    },
-    {
-      "epoch": 1.8095074258845116,
-      "grad_norm": 0.00026850542053580284,
-      "learning_rate": 0.00019999838555630648,
-      "loss": 46.0,
-      "step": 23667
-    },
-    {
-      "epoch": 1.8095838828679014,
-      "grad_norm": 0.0020031880121678114,
-      "learning_rate": 0.00019999838541981635,
-      "loss": 46.0,
-      "step": 23668
-    },
-    {
-      "epoch": 1.8096603398512912,
-      "grad_norm": 0.001533976523205638,
-      "learning_rate": 0.00019999838528332044,
-      "loss": 46.0,
-      "step": 23669
-    },
-    {
-      "epoch": 1.809736796834681,
-      "grad_norm": 0.0004952155286446214,
-      "learning_rate": 0.0001999983851468188,
-      "loss": 46.0,
-      "step": 23670
-    },
-    {
-      "epoch": 1.8098132538180707,
-      "grad_norm": 0.0010664340807124972,
-      "learning_rate": 0.00019999838501031137,
-      "loss": 46.0,
-      "step": 23671
-    },
-    {
-      "epoch": 1.8098897108014604,
-      "grad_norm": 0.006680730730295181,
-      "learning_rate": 0.00019999838487379815,
-      "loss": 46.0,
-      "step": 23672
-    },
-    {
-      "epoch": 1.8099661677848502,
-      "grad_norm": 0.002702244557440281,
-      "learning_rate": 0.00019999838473727918,
-      "loss": 46.0,
-      "step": 23673
-    },
-    {
-      "epoch": 1.8100426247682397,
-      "grad_norm": 0.0005602518795058131,
-      "learning_rate": 0.00019999838460075443,
-      "loss": 46.0,
-      "step": 23674
-    },
-    {
-      "epoch": 1.8101190817516295,
-      "grad_norm": 0.0016347015043720603,
-      "learning_rate": 0.00019999838446422392,
-      "loss": 46.0,
-      "step": 23675
-    },
-    {
-      "epoch": 1.8101955387350193,
-      "grad_norm": 0.0038165065925568342,
-      "learning_rate": 0.00019999838432768763,
-      "loss": 46.0,
-      "step": 23676
-    },
-    {
-      "epoch": 1.8102719957184088,
-      "grad_norm": 0.0006928627844899893,
-      "learning_rate": 0.0001999983841911456,
-      "loss": 46.0,
-      "step": 23677
-    },
-    {
-      "epoch": 1.8103484527017986,
-      "grad_norm": 0.0009190915152430534,
-      "learning_rate": 0.00019999838405459775,
-      "loss": 46.0,
-      "step": 23678
-    },
-    {
-      "epoch": 1.8104249096851883,
-      "grad_norm": 0.0008778053452260792,
-      "learning_rate": 0.00019999838391804414,
-      "loss": 46.0,
-      "step": 23679
-    },
-    {
-      "epoch": 1.810501366668578,
-      "grad_norm": 0.0013824475463479757,
-      "learning_rate": 0.00019999838378148481,
-      "loss": 46.0,
-      "step": 23680
-    },
-    {
-      "epoch": 1.8105778236519678,
-      "grad_norm": 0.0032797250896692276,
-      "learning_rate": 0.00019999838364491968,
-      "loss": 46.0,
-      "step": 23681
-    },
-    {
-      "epoch": 1.8106542806353576,
-      "grad_norm": 0.0006235101027414203,
-      "learning_rate": 0.00019999838350834878,
-      "loss": 46.0,
-      "step": 23682
-    },
-    {
-      "epoch": 1.8107307376187474,
-      "grad_norm": 0.001360720139928162,
-      "learning_rate": 0.0001999983833717721,
-      "loss": 46.0,
-      "step": 23683
-    },
-    {
-      "epoch": 1.8108071946021371,
-      "grad_norm": 0.003522517392411828,
-      "learning_rate": 0.00019999838323518965,
-      "loss": 46.0,
-      "step": 23684
-    },
-    {
-      "epoch": 1.8108836515855267,
-      "grad_norm": 0.0017405670369043946,
-      "learning_rate": 0.00019999838309860146,
-      "loss": 46.0,
-      "step": 23685
-    },
-    {
-      "epoch": 1.8109601085689164,
-      "grad_norm": 0.0013798840809613466,
-      "learning_rate": 0.0001999983829620075,
-      "loss": 46.0,
-      "step": 23686
-    },
-    {
-      "epoch": 1.8110365655523062,
-      "grad_norm": 0.002594240475445986,
-      "learning_rate": 0.00019999838282540775,
-      "loss": 46.0,
-      "step": 23687
-    },
-    {
-      "epoch": 1.8111130225356957,
-      "grad_norm": 0.0012301730457693338,
-      "learning_rate": 0.00019999838268880226,
-      "loss": 46.0,
-      "step": 23688
-    },
-    {
-      "epoch": 1.8111894795190855,
-      "grad_norm": 0.0006875758990645409,
-      "learning_rate": 0.00019999838255219094,
-      "loss": 46.0,
-      "step": 23689
-    },
-    {
-      "epoch": 1.8112659365024752,
-      "grad_norm": 0.0044905696995556355,
-      "learning_rate": 0.0001999983824155739,
-      "loss": 46.0,
-      "step": 23690
-    },
-    {
-      "epoch": 1.811342393485865,
-      "grad_norm": 0.001636273693293333,
-      "learning_rate": 0.00019999838227895107,
-      "loss": 46.0,
-      "step": 23691
-    },
-    {
-      "epoch": 1.8114188504692548,
-      "grad_norm": 0.0005134460516273975,
-      "learning_rate": 0.00019999838214232248,
-      "loss": 46.0,
-      "step": 23692
-    },
-    {
-      "epoch": 1.8114953074526445,
-      "grad_norm": 0.0010170018067583442,
-      "learning_rate": 0.00019999838200568813,
-      "loss": 46.0,
-      "step": 23693
-    },
-    {
-      "epoch": 1.8115717644360343,
-      "grad_norm": 0.0015770571772009134,
-      "learning_rate": 0.00019999838186904797,
-      "loss": 46.0,
-      "step": 23694
-    },
-    {
-      "epoch": 1.811648221419424,
-      "grad_norm": 0.0012213855516165495,
-      "learning_rate": 0.0001999983817324021,
-      "loss": 46.0,
-      "step": 23695
-    },
-    {
-      "epoch": 1.8117246784028136,
-      "grad_norm": 0.0008021800895221531,
-      "learning_rate": 0.00019999838159575042,
-      "loss": 46.0,
-      "step": 23696
-    },
-    {
-      "epoch": 1.8118011353862034,
-      "grad_norm": 0.001608015038073063,
-      "learning_rate": 0.00019999838145909297,
-      "loss": 46.0,
-      "step": 23697
-    },
-    {
-      "epoch": 1.811877592369593,
-      "grad_norm": 0.0005111222271807492,
-      "learning_rate": 0.00019999838132242977,
-      "loss": 46.0,
-      "step": 23698
-    },
-    {
-      "epoch": 1.8119540493529827,
-      "grad_norm": 0.0006768002058379352,
-      "learning_rate": 0.0001999983811857608,
-      "loss": 46.0,
-      "step": 23699
-    },
-    {
-      "epoch": 1.8120305063363724,
-      "grad_norm": 0.002105721039697528,
-      "learning_rate": 0.00019999838104908606,
-      "loss": 46.0,
-      "step": 23700
-    },
-    {
-      "epoch": 1.8121069633197622,
-      "grad_norm": 0.0019101414363831282,
-      "learning_rate": 0.00019999838091240552,
-      "loss": 46.0,
-      "step": 23701
-    },
-    {
-      "epoch": 1.812183420303152,
-      "grad_norm": 0.001791138551197946,
-      "learning_rate": 0.00019999838077571926,
-      "loss": 46.0,
-      "step": 23702
-    },
-    {
-      "epoch": 1.8122598772865417,
-      "grad_norm": 0.0013671392807736993,
-      "learning_rate": 0.0001999983806390272,
-      "loss": 46.0,
-      "step": 23703
-    },
-    {
-      "epoch": 1.8123363342699315,
-      "grad_norm": 0.0011461729882284999,
-      "learning_rate": 0.00019999838050232938,
-      "loss": 46.0,
-      "step": 23704
-    },
-    {
-      "epoch": 1.8124127912533212,
-      "grad_norm": 0.000511440506670624,
-      "learning_rate": 0.00019999838036562577,
-      "loss": 46.0,
-      "step": 23705
-    },
-    {
-      "epoch": 1.812489248236711,
-      "grad_norm": 0.0007339512812905014,
-      "learning_rate": 0.00019999838022891642,
-      "loss": 46.0,
-      "step": 23706
-    },
-    {
-      "epoch": 1.8125657052201005,
-      "grad_norm": 0.001765978173352778,
-      "learning_rate": 0.00019999838009220126,
-      "loss": 46.0,
-      "step": 23707
-    },
-    {
-      "epoch": 1.8126421622034903,
-      "grad_norm": 0.00021233840379863977,
-      "learning_rate": 0.00019999837995548036,
-      "loss": 46.0,
-      "step": 23708
-    },
-    {
-      "epoch": 1.8127186191868798,
-      "grad_norm": 0.0006821598508395255,
-      "learning_rate": 0.0001999983798187537,
-      "loss": 46.0,
-      "step": 23709
-    },
-    {
-      "epoch": 1.8127950761702696,
-      "grad_norm": 0.0004592933692038059,
-      "learning_rate": 0.00019999837968202126,
-      "loss": 46.0,
-      "step": 23710
-    },
-    {
-      "epoch": 1.8128715331536593,
-      "grad_norm": 0.0021277684718370438,
-      "learning_rate": 0.00019999837954528307,
-      "loss": 46.0,
-      "step": 23711
-    },
-    {
-      "epoch": 1.812947990137049,
-      "grad_norm": 0.0005421715322881937,
-      "learning_rate": 0.00019999837940853907,
-      "loss": 46.0,
-      "step": 23712
-    },
-    {
-      "epoch": 1.8130244471204389,
-      "grad_norm": 0.0009878689888864756,
-      "learning_rate": 0.00019999837927178933,
-      "loss": 46.0,
-      "step": 23713
-    },
-    {
-      "epoch": 1.8131009041038286,
-      "grad_norm": 0.0016195743810385466,
-      "learning_rate": 0.0001999983791350338,
-      "loss": 46.0,
-      "step": 23714
-    },
-    {
-      "epoch": 1.8131773610872184,
-      "grad_norm": 0.0005191999371163547,
-      "learning_rate": 0.00019999837899827255,
-      "loss": 46.0,
-      "step": 23715
-    },
-    {
-      "epoch": 1.8132538180706081,
-      "grad_norm": 0.0020176684483885765,
-      "learning_rate": 0.00019999837886150546,
-      "loss": 46.0,
-      "step": 23716
-    },
-    {
-      "epoch": 1.813330275053998,
-      "grad_norm": 0.0008017811924219131,
-      "learning_rate": 0.00019999837872473265,
-      "loss": 46.0,
-      "step": 23717
-    },
-    {
-      "epoch": 1.8134067320373874,
-      "grad_norm": 0.0006969118840061128,
-      "learning_rate": 0.00019999837858795404,
-      "loss": 46.0,
-      "step": 23718
-    },
-    {
-      "epoch": 1.8134831890207772,
-      "grad_norm": 0.000578109291382134,
-      "learning_rate": 0.00019999837845116969,
-      "loss": 46.0,
-      "step": 23719
-    },
-    {
-      "epoch": 1.8135596460041667,
-      "grad_norm": 0.000796251988504082,
-      "learning_rate": 0.00019999837831437956,
-      "loss": 46.0,
-      "step": 23720
-    },
-    {
-      "epoch": 1.8136361029875565,
-      "grad_norm": 0.00046822612057439983,
-      "learning_rate": 0.00019999837817758365,
-      "loss": 46.0,
-      "step": 23721
-    },
-    {
-      "epoch": 1.8137125599709463,
-      "grad_norm": 0.001055878004990518,
-      "learning_rate": 0.00019999837804078195,
-      "loss": 46.0,
-      "step": 23722
-    },
-    {
-      "epoch": 1.813789016954336,
-      "grad_norm": 0.0005453979247249663,
-      "learning_rate": 0.00019999837790397453,
-      "loss": 46.0,
-      "step": 23723
-    },
-    {
-      "epoch": 1.8138654739377258,
-      "grad_norm": 0.0005892764893360436,
-      "learning_rate": 0.00019999837776716133,
-      "loss": 46.0,
-      "step": 23724
-    },
-    {
-      "epoch": 1.8139419309211156,
-      "grad_norm": 0.0007359874434769154,
-      "learning_rate": 0.00019999837763034234,
-      "loss": 46.0,
-      "step": 23725
-    },
-    {
-      "epoch": 1.8140183879045053,
-      "grad_norm": 0.003332388587296009,
-      "learning_rate": 0.0001999983774935176,
-      "loss": 46.0,
-      "step": 23726
-    },
-    {
-      "epoch": 1.814094844887895,
-      "grad_norm": 0.0012405895395204425,
-      "learning_rate": 0.00019999837735668705,
-      "loss": 46.0,
-      "step": 23727
-    },
-    {
-      "epoch": 1.8141713018712846,
-      "grad_norm": 0.0016399107407778502,
-      "learning_rate": 0.00019999837721985077,
-      "loss": 46.0,
-      "step": 23728
-    },
-    {
-      "epoch": 1.8142477588546744,
-      "grad_norm": 0.0019913967698812485,
-      "learning_rate": 0.00019999837708300873,
-      "loss": 46.0,
-      "step": 23729
-    },
-    {
-      "epoch": 1.8143242158380641,
-      "grad_norm": 0.0007014106377027929,
-      "learning_rate": 0.0001999983769461609,
-      "loss": 46.0,
-      "step": 23730
-    },
-    {
-      "epoch": 1.8144006728214537,
-      "grad_norm": 0.0024441899731755257,
-      "learning_rate": 0.0001999983768093073,
-      "loss": 46.0,
-      "step": 23731
-    },
-    {
-      "epoch": 1.8144771298048434,
-      "grad_norm": 0.001906158053316176,
-      "learning_rate": 0.00019999837667244793,
-      "loss": 46.0,
-      "step": 23732
-    },
-    {
-      "epoch": 1.8145535867882332,
-      "grad_norm": 0.0023151030763983727,
-      "learning_rate": 0.0001999983765355828,
-      "loss": 46.0,
-      "step": 23733
-    },
-    {
-      "epoch": 1.814630043771623,
-      "grad_norm": 0.0024479578714817762,
-      "learning_rate": 0.00019999837639871187,
-      "loss": 46.0,
-      "step": 23734
-    },
-    {
-      "epoch": 1.8147065007550127,
-      "grad_norm": 0.002640495542436838,
-      "learning_rate": 0.0001999983762618352,
-      "loss": 46.0,
-      "step": 23735
-    },
-    {
-      "epoch": 1.8147829577384025,
-      "grad_norm": 0.0013096173061057925,
-      "learning_rate": 0.00019999837612495278,
-      "loss": 46.0,
-      "step": 23736
-    },
-    {
-      "epoch": 1.8148594147217922,
-      "grad_norm": 0.0009974492713809013,
-      "learning_rate": 0.00019999837598806455,
-      "loss": 46.0,
-      "step": 23737
-    },
-    {
-      "epoch": 1.814935871705182,
-      "grad_norm": 0.0005888328305445611,
-      "learning_rate": 0.00019999837585117059,
-      "loss": 46.0,
-      "step": 23738
-    },
-    {
-      "epoch": 1.8150123286885715,
-      "grad_norm": 0.0010382991749793291,
-      "learning_rate": 0.00019999837571427082,
-      "loss": 46.0,
-      "step": 23739
-    },
-    {
-      "epoch": 1.8150887856719613,
-      "grad_norm": 0.0019064415246248245,
-      "learning_rate": 0.0001999983755773653,
-      "loss": 46.0,
-      "step": 23740
-    },
-    {
-      "epoch": 1.815165242655351,
-      "grad_norm": 0.0023607865441590548,
-      "learning_rate": 0.000199998375440454,
-      "loss": 46.0,
-      "step": 23741
-    },
-    {
-      "epoch": 1.8152416996387406,
-      "grad_norm": 0.001305295154452324,
-      "learning_rate": 0.00019999837530353695,
-      "loss": 46.0,
-      "step": 23742
-    },
-    {
-      "epoch": 1.8153181566221304,
-      "grad_norm": 0.002001626882702112,
-      "learning_rate": 0.00019999837516661412,
-      "loss": 46.0,
-      "step": 23743
-    },
-    {
-      "epoch": 1.8153946136055201,
-      "grad_norm": 0.00202653044834733,
-      "learning_rate": 0.00019999837502968553,
-      "loss": 46.0,
-      "step": 23744
-    },
-    {
-      "epoch": 1.8154710705889099,
-      "grad_norm": 0.0006119931931607425,
-      "learning_rate": 0.00019999837489275115,
-      "loss": 46.0,
-      "step": 23745
-    },
-    {
-      "epoch": 1.8155475275722996,
-      "grad_norm": 0.000539101311005652,
-      "learning_rate": 0.000199998374755811,
-      "loss": 46.0,
-      "step": 23746
-    },
-    {
-      "epoch": 1.8156239845556894,
-      "grad_norm": 0.00047082730452530086,
-      "learning_rate": 0.0001999983746188651,
-      "loss": 46.0,
-      "step": 23747
-    },
-    {
-      "epoch": 1.8157004415390792,
-      "grad_norm": 0.0008020388195291162,
-      "learning_rate": 0.00019999837448191342,
-      "loss": 46.0,
-      "step": 23748
-    },
-    {
-      "epoch": 1.815776898522469,
-      "grad_norm": 0.0007263301522471011,
-      "learning_rate": 0.000199998374344956,
-      "loss": 46.0,
-      "step": 23749
-    },
-    {
-      "epoch": 1.8158533555058585,
-      "grad_norm": 0.002049210947006941,
-      "learning_rate": 0.00019999837420799278,
-      "loss": 46.0,
-      "step": 23750
-    },
-    {
-      "epoch": 1.8159298124892482,
-      "grad_norm": 0.0015929215587675571,
-      "learning_rate": 0.0001999983740710238,
-      "loss": 46.0,
-      "step": 23751
-    },
-    {
-      "epoch": 1.816006269472638,
-      "grad_norm": 0.001869912026450038,
-      "learning_rate": 0.00019999837393404904,
-      "loss": 46.0,
-      "step": 23752
-    },
-    {
-      "epoch": 1.8160827264560275,
-      "grad_norm": 0.003016539616510272,
-      "learning_rate": 0.0001999983737970685,
-      "loss": 46.0,
-      "step": 23753
-    },
-    {
-      "epoch": 1.8161591834394173,
-      "grad_norm": 0.0009548344532959163,
-      "learning_rate": 0.0001999983736600822,
-      "loss": 46.0,
-      "step": 23754
-    },
-    {
-      "epoch": 1.816235640422807,
-      "grad_norm": 0.0027448381297290325,
-      "learning_rate": 0.00019999837352309015,
-      "loss": 46.0,
-      "step": 23755
-    },
-    {
-      "epoch": 1.8163120974061968,
-      "grad_norm": 0.0005462928093038499,
-      "learning_rate": 0.00019999837338609234,
-      "loss": 46.0,
-      "step": 23756
-    },
-    {
-      "epoch": 1.8163885543895866,
-      "grad_norm": 0.0006976991426199675,
-      "learning_rate": 0.00019999837324908873,
-      "loss": 46.0,
-      "step": 23757
-    },
-    {
-      "epoch": 1.8164650113729763,
-      "grad_norm": 0.0005639574374072254,
-      "learning_rate": 0.00019999837311207935,
-      "loss": 46.0,
-      "step": 23758
-    },
-    {
-      "epoch": 1.816541468356366,
-      "grad_norm": 0.0006333451601676643,
-      "learning_rate": 0.0001999983729750642,
-      "loss": 46.0,
-      "step": 23759
-    },
-    {
-      "epoch": 1.8166179253397559,
-      "grad_norm": 0.0009449411300010979,
-      "learning_rate": 0.00019999837283804332,
-      "loss": 46.0,
-      "step": 23760
-    },
-    {
-      "epoch": 1.8166943823231454,
-      "grad_norm": 0.0005404514377005398,
-      "learning_rate": 0.00019999837270101662,
-      "loss": 46.0,
-      "step": 23761
-    },
-    {
-      "epoch": 1.8167708393065352,
-      "grad_norm": 0.001246896805241704,
-      "learning_rate": 0.0001999983725639842,
-      "loss": 46.0,
-      "step": 23762
-    },
-    {
-      "epoch": 1.816847296289925,
-      "grad_norm": 0.0012557493755593896,
-      "learning_rate": 0.00019999837242694598,
-      "loss": 46.0,
-      "step": 23763
-    },
-    {
-      "epoch": 1.8169237532733145,
-      "grad_norm": 0.0015247507253661752,
-      "learning_rate": 0.00019999837228990198,
-      "loss": 46.0,
-      "step": 23764
-    },
-    {
-      "epoch": 1.8170002102567042,
-      "grad_norm": 0.0003656579356174916,
-      "learning_rate": 0.00019999837215285221,
-      "loss": 46.0,
-      "step": 23765
-    },
-    {
-      "epoch": 1.817076667240094,
-      "grad_norm": 0.0004148459993302822,
-      "learning_rate": 0.0001999983720157967,
-      "loss": 46.0,
-      "step": 23766
-    },
-    {
-      "epoch": 1.8171531242234837,
-      "grad_norm": 0.003125576302409172,
-      "learning_rate": 0.0001999983718787354,
-      "loss": 46.0,
-      "step": 23767
-    },
-    {
-      "epoch": 1.8172295812068735,
-      "grad_norm": 0.0005042135016992688,
-      "learning_rate": 0.00019999837174166832,
-      "loss": 46.0,
-      "step": 23768
-    },
-    {
-      "epoch": 1.8173060381902633,
-      "grad_norm": 0.0031226954888552427,
-      "learning_rate": 0.00019999837160459552,
-      "loss": 46.0,
-      "step": 23769
-    },
-    {
-      "epoch": 1.817382495173653,
-      "grad_norm": 0.0008610028307884932,
-      "learning_rate": 0.0001999983714675169,
-      "loss": 46.0,
-      "step": 23770
-    },
-    {
-      "epoch": 1.8174589521570428,
-      "grad_norm": 0.004178045317530632,
-      "learning_rate": 0.00019999837133043253,
-      "loss": 46.0,
-      "step": 23771
-    },
-    {
-      "epoch": 1.8175354091404323,
-      "grad_norm": 0.004781307652592659,
-      "learning_rate": 0.0001999983711933424,
-      "loss": 46.0,
-      "step": 23772
-    },
-    {
-      "epoch": 1.817611866123822,
-      "grad_norm": 0.0005874600610695779,
-      "learning_rate": 0.00019999837105624647,
-      "loss": 46.0,
-      "step": 23773
-    },
-    {
-      "epoch": 1.8176883231072118,
-      "grad_norm": 0.0006655587349087,
-      "learning_rate": 0.00019999837091914482,
-      "loss": 46.0,
-      "step": 23774
-    },
-    {
-      "epoch": 1.8177647800906014,
-      "grad_norm": 0.0015199481276795268,
-      "learning_rate": 0.00019999837078203735,
-      "loss": 46.0,
-      "step": 23775
-    },
-    {
-      "epoch": 1.8178412370739911,
-      "grad_norm": 0.0008756067254580557,
-      "learning_rate": 0.00019999837064492413,
-      "loss": 46.0,
-      "step": 23776
-    },
-    {
-      "epoch": 1.817917694057381,
-      "grad_norm": 0.0008462730911560357,
-      "learning_rate": 0.00019999837050780513,
-      "loss": 46.0,
-      "step": 23777
-    },
-    {
-      "epoch": 1.8179941510407707,
-      "grad_norm": 0.000778833928052336,
-      "learning_rate": 0.0001999983703706804,
-      "loss": 46.0,
-      "step": 23778
-    },
-    {
-      "epoch": 1.8180706080241604,
-      "grad_norm": 0.00037754737422801554,
-      "learning_rate": 0.00019999837023354985,
-      "loss": 46.0,
-      "step": 23779
-    },
-    {
-      "epoch": 1.8181470650075502,
-      "grad_norm": 0.0007783666951581836,
-      "learning_rate": 0.00019999837009641356,
-      "loss": 46.0,
-      "step": 23780
-    },
-    {
-      "epoch": 1.81822352199094,
-      "grad_norm": 0.004791202023625374,
-      "learning_rate": 0.0001999983699592715,
-      "loss": 46.0,
-      "step": 23781
-    },
-    {
-      "epoch": 1.8182999789743297,
-      "grad_norm": 0.0006258657667785883,
-      "learning_rate": 0.00019999836982212367,
-      "loss": 46.0,
-      "step": 23782
-    },
-    {
-      "epoch": 1.8183764359577193,
-      "grad_norm": 0.001412646030075848,
-      "learning_rate": 0.00019999836968497006,
-      "loss": 46.0,
-      "step": 23783
-    },
-    {
-      "epoch": 1.818452892941109,
-      "grad_norm": 0.0013155157212167978,
-      "learning_rate": 0.0001999983695478107,
-      "loss": 46.0,
-      "step": 23784
-    },
-    {
-      "epoch": 1.8185293499244988,
-      "grad_norm": 0.0030480362474918365,
-      "learning_rate": 0.00019999836941064555,
-      "loss": 46.0,
-      "step": 23785
-    },
-    {
-      "epoch": 1.8186058069078883,
-      "grad_norm": 0.001286276150494814,
-      "learning_rate": 0.00019999836927347465,
-      "loss": 46.0,
-      "step": 23786
-    },
-    {
-      "epoch": 1.818682263891278,
-      "grad_norm": 0.0009965029312297702,
-      "learning_rate": 0.00019999836913629798,
-      "loss": 46.0,
-      "step": 23787
-    },
-    {
-      "epoch": 1.8187587208746678,
-      "grad_norm": 0.00161589658819139,
-      "learning_rate": 0.00019999836899911553,
-      "loss": 46.0,
-      "step": 23788
-    },
-    {
-      "epoch": 1.8188351778580576,
-      "grad_norm": 0.000690197222866118,
-      "learning_rate": 0.00019999836886192728,
-      "loss": 46.0,
-      "step": 23789
-    },
-    {
-      "epoch": 1.8189116348414474,
-      "grad_norm": 0.0029922593384981155,
-      "learning_rate": 0.0001999983687247333,
-      "loss": 46.0,
-      "step": 23790
-    },
-    {
-      "epoch": 1.8189880918248371,
-      "grad_norm": 0.0005037240334786475,
-      "learning_rate": 0.00019999836858753355,
-      "loss": 46.0,
-      "step": 23791
-    },
-    {
-      "epoch": 1.8190645488082269,
-      "grad_norm": 0.0010681712301447988,
-      "learning_rate": 0.00019999836845032803,
-      "loss": 46.0,
-      "step": 23792
-    },
-    {
-      "epoch": 1.8191410057916166,
-      "grad_norm": 0.002176741138100624,
-      "learning_rate": 0.00019999836831311672,
-      "loss": 46.0,
-      "step": 23793
-    },
-    {
-      "epoch": 1.8192174627750062,
-      "grad_norm": 0.0037860942538827658,
-      "learning_rate": 0.00019999836817589966,
-      "loss": 46.0,
-      "step": 23794
-    },
-    {
-      "epoch": 1.819293919758396,
-      "grad_norm": 0.0011033108457922935,
-      "learning_rate": 0.00019999836803867682,
-      "loss": 46.0,
-      "step": 23795
-    },
-    {
-      "epoch": 1.8193703767417857,
-      "grad_norm": 0.0020454039331525564,
-      "learning_rate": 0.00019999836790144822,
-      "loss": 46.0,
-      "step": 23796
-    },
-    {
-      "epoch": 1.8194468337251752,
-      "grad_norm": 0.0002975747629534453,
-      "learning_rate": 0.00019999836776421384,
-      "loss": 46.0,
-      "step": 23797
-    },
-    {
-      "epoch": 1.819523290708565,
-      "grad_norm": 0.0005273958086036146,
-      "learning_rate": 0.0001999983676269737,
-      "loss": 46.0,
-      "step": 23798
-    },
-    {
-      "epoch": 1.8195997476919548,
-      "grad_norm": 0.0015240439679473639,
-      "learning_rate": 0.00019999836748972778,
-      "loss": 46.0,
-      "step": 23799
-    },
-    {
-      "epoch": 1.8196762046753445,
-      "grad_norm": 0.0007435697480104864,
-      "learning_rate": 0.00019999836735247608,
-      "loss": 46.0,
-      "step": 23800
-    },
-    {
-      "epoch": 1.8197526616587343,
-      "grad_norm": 0.0017040574457496405,
-      "learning_rate": 0.00019999836721521863,
-      "loss": 46.0,
-      "step": 23801
-    },
-    {
-      "epoch": 1.819829118642124,
-      "grad_norm": 0.0033538327552378178,
-      "learning_rate": 0.0001999983670779554,
-      "loss": 46.0,
-      "step": 23802
-    },
-    {
-      "epoch": 1.8199055756255138,
-      "grad_norm": 0.0022707749158143997,
-      "learning_rate": 0.00019999836694068645,
-      "loss": 46.0,
-      "step": 23803
-    },
-    {
-      "epoch": 1.8199820326089036,
-      "grad_norm": 0.001140329404734075,
-      "learning_rate": 0.00019999836680341165,
-      "loss": 46.0,
-      "step": 23804
-    },
-    {
-      "epoch": 1.820058489592293,
-      "grad_norm": 0.0039946772158145905,
-      "learning_rate": 0.00019999836666613114,
-      "loss": 46.0,
-      "step": 23805
-    },
-    {
-      "epoch": 1.8201349465756829,
-      "grad_norm": 0.0009491947712376714,
-      "learning_rate": 0.00019999836652884482,
-      "loss": 46.0,
-      "step": 23806
-    },
-    {
-      "epoch": 1.8202114035590726,
-      "grad_norm": 0.0009418984409421682,
-      "learning_rate": 0.0001999983663915528,
-      "loss": 46.0,
-      "step": 23807
-    },
-    {
-      "epoch": 1.8202878605424622,
-      "grad_norm": 0.0007848025416024029,
-      "learning_rate": 0.00019999836625425493,
-      "loss": 46.0,
-      "step": 23808
-    },
-    {
-      "epoch": 1.820364317525852,
-      "grad_norm": 0.0018619188340380788,
-      "learning_rate": 0.00019999836611695132,
-      "loss": 46.0,
-      "step": 23809
-    },
-    {
-      "epoch": 1.8204407745092417,
-      "grad_norm": 0.0036131732631474733,
-      "learning_rate": 0.00019999836597964194,
-      "loss": 46.0,
-      "step": 23810
-    },
-    {
-      "epoch": 1.8205172314926314,
-      "grad_norm": 0.0039264364168047905,
-      "learning_rate": 0.00019999836584232682,
-      "loss": 46.0,
-      "step": 23811
-    },
-    {
-      "epoch": 1.8205936884760212,
-      "grad_norm": 0.002276131883263588,
-      "learning_rate": 0.00019999836570500586,
-      "loss": 46.0,
-      "step": 23812
-    },
-    {
-      "epoch": 1.820670145459411,
-      "grad_norm": 0.004367159213870764,
-      "learning_rate": 0.0001999983655676792,
-      "loss": 46.0,
-      "step": 23813
-    },
-    {
-      "epoch": 1.8207466024428007,
-      "grad_norm": 0.0037661665119230747,
-      "learning_rate": 0.00019999836543034674,
-      "loss": 46.0,
-      "step": 23814
-    },
-    {
-      "epoch": 1.8208230594261905,
-      "grad_norm": 0.0037466511130332947,
-      "learning_rate": 0.00019999836529300852,
-      "loss": 46.0,
-      "step": 23815
-    },
-    {
-      "epoch": 1.82089951640958,
-      "grad_norm": 0.0008819775539450347,
-      "learning_rate": 0.00019999836515566455,
-      "loss": 46.0,
-      "step": 23816
-    },
-    {
-      "epoch": 1.8209759733929698,
-      "grad_norm": 0.0028808724600821733,
-      "learning_rate": 0.00019999836501831476,
-      "loss": 46.0,
-      "step": 23817
-    },
-    {
-      "epoch": 1.8210524303763596,
-      "grad_norm": 0.0007140140514820814,
-      "learning_rate": 0.00019999836488095925,
-      "loss": 46.0,
-      "step": 23818
-    },
-    {
-      "epoch": 1.821128887359749,
-      "grad_norm": 0.0004463373334147036,
-      "learning_rate": 0.00019999836474359793,
-      "loss": 46.0,
-      "step": 23819
-    },
-    {
-      "epoch": 1.8212053443431389,
-      "grad_norm": 0.0013685438316315413,
-      "learning_rate": 0.00019999836460623087,
-      "loss": 46.0,
-      "step": 23820
-    },
-    {
-      "epoch": 1.8212818013265286,
-      "grad_norm": 0.002619948238134384,
-      "learning_rate": 0.000199998364468858,
-      "loss": 46.0,
-      "step": 23821
-    },
-    {
-      "epoch": 1.8213582583099184,
-      "grad_norm": 0.001976925414055586,
-      "learning_rate": 0.00019999836433147943,
-      "loss": 46.0,
-      "step": 23822
-    },
-    {
-      "epoch": 1.8214347152933081,
-      "grad_norm": 0.0030305383261293173,
-      "learning_rate": 0.00019999836419409505,
-      "loss": 46.0,
-      "step": 23823
-    },
-    {
-      "epoch": 1.821511172276698,
-      "grad_norm": 0.0010502489749342203,
-      "learning_rate": 0.0001999983640567049,
-      "loss": 46.0,
-      "step": 23824
-    },
-    {
-      "epoch": 1.8215876292600877,
-      "grad_norm": 0.00035608242615126073,
-      "learning_rate": 0.000199998363919309,
-      "loss": 46.0,
-      "step": 23825
-    },
-    {
-      "epoch": 1.8216640862434774,
-      "grad_norm": 0.0006635666941292584,
-      "learning_rate": 0.0001999983637819073,
-      "loss": 46.0,
-      "step": 23826
-    },
-    {
-      "epoch": 1.821740543226867,
-      "grad_norm": 0.001828717882744968,
-      "learning_rate": 0.00019999836364449985,
-      "loss": 46.0,
-      "step": 23827
-    },
-    {
-      "epoch": 1.8218170002102567,
-      "grad_norm": 0.0007944001699797809,
-      "learning_rate": 0.0001999983635070866,
-      "loss": 46.0,
-      "step": 23828
-    },
-    {
-      "epoch": 1.8218934571936463,
-      "grad_norm": 0.003315124660730362,
-      "learning_rate": 0.0001999983633696676,
-      "loss": 46.0,
-      "step": 23829
-    },
-    {
-      "epoch": 1.821969914177036,
-      "grad_norm": 0.0069752843119204044,
-      "learning_rate": 0.00019999836323224285,
-      "loss": 46.0,
-      "step": 23830
-    },
-    {
-      "epoch": 1.8220463711604258,
-      "grad_norm": 0.0005215610144659877,
-      "learning_rate": 0.00019999836309481233,
-      "loss": 46.0,
-      "step": 23831
-    },
-    {
-      "epoch": 1.8221228281438155,
-      "grad_norm": 0.003613660577684641,
-      "learning_rate": 0.00019999836295737602,
-      "loss": 46.0,
-      "step": 23832
-    },
-    {
-      "epoch": 1.8221992851272053,
-      "grad_norm": 0.003564306069165468,
-      "learning_rate": 0.00019999836281993396,
-      "loss": 46.0,
-      "step": 23833
-    },
-    {
-      "epoch": 1.822275742110595,
-      "grad_norm": 0.0009329724125564098,
-      "learning_rate": 0.0001999983626824861,
-      "loss": 46.0,
-      "step": 23834
-    },
-    {
-      "epoch": 1.8223521990939848,
-      "grad_norm": 0.00513521209359169,
-      "learning_rate": 0.0001999983625450325,
-      "loss": 46.0,
-      "step": 23835
-    },
-    {
-      "epoch": 1.8224286560773746,
-      "grad_norm": 0.0009125597425736487,
-      "learning_rate": 0.0001999983624075731,
-      "loss": 46.0,
-      "step": 23836
-    },
-    {
-      "epoch": 1.8225051130607643,
-      "grad_norm": 0.0010099117644131184,
-      "learning_rate": 0.00019999836227010796,
-      "loss": 46.0,
-      "step": 23837
-    },
-    {
-      "epoch": 1.8225815700441539,
-      "grad_norm": 0.009658338502049446,
-      "learning_rate": 0.00019999836213263706,
-      "loss": 46.0,
-      "step": 23838
-    },
-    {
-      "epoch": 1.8226580270275436,
-      "grad_norm": 0.0008519923430867493,
-      "learning_rate": 0.00019999836199516036,
-      "loss": 46.0,
-      "step": 23839
-    },
-    {
-      "epoch": 1.8227344840109332,
-      "grad_norm": 0.0015861664433032274,
-      "learning_rate": 0.00019999836185767791,
-      "loss": 46.0,
-      "step": 23840
-    },
-    {
-      "epoch": 1.822810940994323,
-      "grad_norm": 0.0015659792115911841,
-      "learning_rate": 0.00019999836172018967,
-      "loss": 46.0,
-      "step": 23841
-    },
-    {
-      "epoch": 1.8228873979777127,
-      "grad_norm": 0.001510980655439198,
-      "learning_rate": 0.0001999983615826957,
-      "loss": 46.0,
-      "step": 23842
-    },
-    {
-      "epoch": 1.8229638549611025,
-      "grad_norm": 0.0022470734547823668,
-      "learning_rate": 0.0001999983614451959,
-      "loss": 46.0,
-      "step": 23843
-    },
-    {
-      "epoch": 1.8230403119444922,
-      "grad_norm": 0.0012548583326861262,
-      "learning_rate": 0.0001999983613076904,
-      "loss": 46.0,
-      "step": 23844
-    },
-    {
-      "epoch": 1.823116768927882,
-      "grad_norm": 0.001574006862938404,
-      "learning_rate": 0.00019999836117017908,
-      "loss": 46.0,
-      "step": 23845
-    },
-    {
-      "epoch": 1.8231932259112718,
-      "grad_norm": 0.0008820904768072069,
-      "learning_rate": 0.000199998361032662,
-      "loss": 46.0,
-      "step": 23846
-    },
-    {
-      "epoch": 1.8232696828946615,
-      "grad_norm": 0.0004603576671797782,
-      "learning_rate": 0.00019999836089513914,
-      "loss": 46.0,
-      "step": 23847
-    },
-    {
-      "epoch": 1.8233461398780513,
-      "grad_norm": 0.00055723573314026,
-      "learning_rate": 0.00019999836075761053,
-      "loss": 46.0,
-      "step": 23848
-    },
-    {
-      "epoch": 1.8234225968614408,
-      "grad_norm": 0.0015917620621621609,
-      "learning_rate": 0.00019999836062007615,
-      "loss": 46.0,
-      "step": 23849
-    },
-    {
-      "epoch": 1.8234990538448306,
-      "grad_norm": 0.000962583813816309,
-      "learning_rate": 0.00019999836048253603,
-      "loss": 46.0,
-      "step": 23850
-    },
-    {
-      "epoch": 1.8235755108282201,
-      "grad_norm": 0.0011149718193337321,
-      "learning_rate": 0.0001999983603449901,
-      "loss": 46.0,
-      "step": 23851
-    },
-    {
-      "epoch": 1.8236519678116099,
-      "grad_norm": 0.0024677978362888098,
-      "learning_rate": 0.0001999983602074384,
-      "loss": 46.0,
-      "step": 23852
-    },
-    {
-      "epoch": 1.8237284247949996,
-      "grad_norm": 0.0012178650358691812,
-      "learning_rate": 0.00019999836006988093,
-      "loss": 46.0,
-      "step": 23853
-    },
-    {
-      "epoch": 1.8238048817783894,
-      "grad_norm": 0.0010864818468689919,
-      "learning_rate": 0.00019999835993231773,
-      "loss": 46.0,
-      "step": 23854
-    },
-    {
-      "epoch": 1.8238813387617792,
-      "grad_norm": 0.005176450591534376,
-      "learning_rate": 0.00019999835979474871,
-      "loss": 46.0,
-      "step": 23855
-    },
-    {
-      "epoch": 1.823957795745169,
-      "grad_norm": 0.0017009952571243048,
-      "learning_rate": 0.00019999835965717395,
-      "loss": 46.0,
-      "step": 23856
-    },
-    {
-      "epoch": 1.8240342527285587,
-      "grad_norm": 0.0008226045756600797,
-      "learning_rate": 0.00019999835951959344,
-      "loss": 46.0,
-      "step": 23857
-    },
-    {
-      "epoch": 1.8241107097119484,
-      "grad_norm": 0.005510048009455204,
-      "learning_rate": 0.0001999983593820071,
-      "loss": 46.0,
-      "step": 23858
-    },
-    {
-      "epoch": 1.824187166695338,
-      "grad_norm": 0.0025144314859062433,
-      "learning_rate": 0.00019999835924441504,
-      "loss": 46.0,
-      "step": 23859
-    },
-    {
-      "epoch": 1.8242636236787277,
-      "grad_norm": 0.001036861096508801,
-      "learning_rate": 0.00019999835910681718,
-      "loss": 46.0,
-      "step": 23860
-    },
-    {
-      "epoch": 1.8243400806621175,
-      "grad_norm": 0.001475979806855321,
-      "learning_rate": 0.00019999835896921357,
-      "loss": 46.0,
-      "step": 23861
-    },
-    {
-      "epoch": 1.824416537645507,
-      "grad_norm": 0.0008923432324081659,
-      "learning_rate": 0.0001999983588316042,
-      "loss": 46.0,
-      "step": 23862
-    },
-    {
-      "epoch": 1.8244929946288968,
-      "grad_norm": 0.004577656276524067,
-      "learning_rate": 0.00019999835869398904,
-      "loss": 46.0,
-      "step": 23863
-    },
-    {
-      "epoch": 1.8245694516122866,
-      "grad_norm": 0.0021204662043601274,
-      "learning_rate": 0.00019999835855636811,
-      "loss": 46.0,
-      "step": 23864
-    },
-    {
-      "epoch": 1.8246459085956763,
-      "grad_norm": 0.000634261523373425,
-      "learning_rate": 0.00019999835841874144,
-      "loss": 46.0,
-      "step": 23865
-    },
-    {
-      "epoch": 1.824722365579066,
-      "grad_norm": 0.0014365861425176263,
-      "learning_rate": 0.00019999835828110894,
-      "loss": 46.0,
-      "step": 23866
-    },
-    {
-      "epoch": 1.8247988225624558,
-      "grad_norm": 0.0033843209967017174,
-      "learning_rate": 0.00019999835814347075,
-      "loss": 46.0,
-      "step": 23867
-    },
-    {
-      "epoch": 1.8248752795458456,
-      "grad_norm": 0.0025061562191694975,
-      "learning_rate": 0.00019999835800582673,
-      "loss": 46.0,
-      "step": 23868
-    },
-    {
-      "epoch": 1.8249517365292354,
-      "grad_norm": 0.0013637689407914877,
-      "learning_rate": 0.00019999835786817696,
-      "loss": 46.0,
-      "step": 23869
-    },
-    {
-      "epoch": 1.825028193512625,
-      "grad_norm": 0.0008603520691394806,
-      "learning_rate": 0.00019999835773052143,
-      "loss": 46.0,
-      "step": 23870
-    },
-    {
-      "epoch": 1.8251046504960147,
-      "grad_norm": 0.0006613720906898379,
-      "learning_rate": 0.0001999983575928601,
-      "loss": 46.0,
-      "step": 23871
-    },
-    {
-      "epoch": 1.8251811074794044,
-      "grad_norm": 0.0019336212426424026,
-      "learning_rate": 0.00019999835745519305,
-      "loss": 46.0,
-      "step": 23872
-    },
-    {
-      "epoch": 1.825257564462794,
-      "grad_norm": 0.014227689243853092,
-      "learning_rate": 0.0001999983573175202,
-      "loss": 46.0,
-      "step": 23873
-    },
-    {
-      "epoch": 1.8253340214461837,
-      "grad_norm": 0.0009199431515298784,
-      "learning_rate": 0.0001999983571798416,
-      "loss": 46.0,
-      "step": 23874
-    },
-    {
-      "epoch": 1.8254104784295735,
-      "grad_norm": 0.0006757903029210865,
-      "learning_rate": 0.0001999983570421572,
-      "loss": 46.0,
-      "step": 23875
-    },
-    {
-      "epoch": 1.8254869354129633,
-      "grad_norm": 0.0006481630261987448,
-      "learning_rate": 0.00019999835690446703,
-      "loss": 46.0,
-      "step": 23876
-    },
-    {
-      "epoch": 1.825563392396353,
-      "grad_norm": 0.0017809225246310234,
-      "learning_rate": 0.00019999835676677113,
-      "loss": 46.0,
-      "step": 23877
-    },
-    {
-      "epoch": 1.8256398493797428,
-      "grad_norm": 0.0004950346192345023,
-      "learning_rate": 0.00019999835662906943,
-      "loss": 46.0,
-      "step": 23878
-    },
-    {
-      "epoch": 1.8257163063631325,
-      "grad_norm": 0.0009703824762254953,
-      "learning_rate": 0.000199998356491362,
-      "loss": 46.0,
-      "step": 23879
-    },
-    {
-      "epoch": 1.8257927633465223,
-      "grad_norm": 0.0007006517262198031,
-      "learning_rate": 0.00019999835635364872,
-      "loss": 46.0,
-      "step": 23880
-    },
-    {
-      "epoch": 1.8258692203299118,
-      "grad_norm": 0.00054171122610569,
-      "learning_rate": 0.00019999835621592972,
-      "loss": 46.0,
-      "step": 23881
-    },
-    {
-      "epoch": 1.8259456773133016,
-      "grad_norm": 0.0021724977996200323,
-      "learning_rate": 0.00019999835607820496,
-      "loss": 46.0,
-      "step": 23882
-    },
-    {
-      "epoch": 1.8260221342966914,
-      "grad_norm": 0.002891576150432229,
-      "learning_rate": 0.00019999835594047442,
-      "loss": 46.0,
-      "step": 23883
-    },
-    {
-      "epoch": 1.826098591280081,
-      "grad_norm": 0.0015874544624239206,
-      "learning_rate": 0.0001999983558027381,
-      "loss": 46.0,
-      "step": 23884
-    },
-    {
-      "epoch": 1.8261750482634707,
-      "grad_norm": 0.0010487012332305312,
-      "learning_rate": 0.00019999835566499602,
-      "loss": 46.0,
-      "step": 23885
-    },
-    {
-      "epoch": 1.8262515052468604,
-      "grad_norm": 0.0005651236278936267,
-      "learning_rate": 0.0001999983555272482,
-      "loss": 46.0,
-      "step": 23886
-    },
-    {
-      "epoch": 1.8263279622302502,
-      "grad_norm": 0.0015597763704136014,
-      "learning_rate": 0.00019999835538949459,
-      "loss": 46.0,
-      "step": 23887
-    },
-    {
-      "epoch": 1.82640441921364,
-      "grad_norm": 0.000770699349232018,
-      "learning_rate": 0.00019999835525173518,
-      "loss": 46.0,
-      "step": 23888
-    },
-    {
-      "epoch": 1.8264808761970297,
-      "grad_norm": 0.0007349138031713665,
-      "learning_rate": 0.00019999835511397003,
-      "loss": 46.0,
-      "step": 23889
-    },
-    {
-      "epoch": 1.8265573331804195,
-      "grad_norm": 0.002255955943837762,
-      "learning_rate": 0.0001999983549761991,
-      "loss": 46.0,
-      "step": 23890
-    },
-    {
-      "epoch": 1.8266337901638092,
-      "grad_norm": 0.0017204832984134555,
-      "learning_rate": 0.00019999835483842243,
-      "loss": 46.0,
-      "step": 23891
-    },
-    {
-      "epoch": 1.8267102471471988,
-      "grad_norm": 0.0015940828016027808,
-      "learning_rate": 0.00019999835470063993,
-      "loss": 46.0,
-      "step": 23892
-    },
-    {
-      "epoch": 1.8267867041305885,
-      "grad_norm": 0.0005565414903685451,
-      "learning_rate": 0.00019999835456285169,
-      "loss": 46.0,
-      "step": 23893
-    },
-    {
-      "epoch": 1.8268631611139783,
-      "grad_norm": 0.000997836352325976,
-      "learning_rate": 0.0001999983544250577,
-      "loss": 46.0,
-      "step": 23894
-    },
-    {
-      "epoch": 1.8269396180973678,
-      "grad_norm": 0.0031602459494024515,
-      "learning_rate": 0.00019999835428725793,
-      "loss": 46.0,
-      "step": 23895
-    },
-    {
-      "epoch": 1.8270160750807576,
-      "grad_norm": 0.0021500152070075274,
-      "learning_rate": 0.0001999983541494524,
-      "loss": 46.0,
-      "step": 23896
-    },
-    {
-      "epoch": 1.8270925320641473,
-      "grad_norm": 0.0011405717814341187,
-      "learning_rate": 0.00019999835401164108,
-      "loss": 46.0,
-      "step": 23897
-    },
-    {
-      "epoch": 1.827168989047537,
-      "grad_norm": 0.0024421107955276966,
-      "learning_rate": 0.00019999835387382402,
-      "loss": 46.0,
-      "step": 23898
-    },
-    {
-      "epoch": 1.8272454460309269,
-      "grad_norm": 0.0010856501758098602,
-      "learning_rate": 0.00019999835373600116,
-      "loss": 46.0,
-      "step": 23899
-    },
-    {
-      "epoch": 1.8273219030143166,
-      "grad_norm": 0.0008228891529142857,
-      "learning_rate": 0.00019999835359817253,
-      "loss": 46.0,
-      "step": 23900
-    },
-    {
-      "epoch": 1.8273983599977064,
-      "grad_norm": 0.000771163497120142,
-      "learning_rate": 0.00019999835346033815,
-      "loss": 46.0,
-      "step": 23901
-    },
-    {
-      "epoch": 1.8274748169810962,
-      "grad_norm": 0.0006650617578998208,
-      "learning_rate": 0.000199998353322498,
-      "loss": 46.0,
-      "step": 23902
-    },
-    {
-      "epoch": 1.8275512739644857,
-      "grad_norm": 0.0006762298289686441,
-      "learning_rate": 0.00019999835318465205,
-      "loss": 46.0,
-      "step": 23903
-    },
-    {
-      "epoch": 1.8276277309478755,
-      "grad_norm": 0.001431665034033358,
-      "learning_rate": 0.00019999835304680038,
-      "loss": 46.0,
-      "step": 23904
-    },
-    {
-      "epoch": 1.8277041879312652,
-      "grad_norm": 0.003254946554079652,
-      "learning_rate": 0.0001999983529089429,
-      "loss": 46.0,
-      "step": 23905
-    },
-    {
-      "epoch": 1.8277806449146548,
-      "grad_norm": 0.0009096749126911163,
-      "learning_rate": 0.0001999983527710797,
-      "loss": 46.0,
-      "step": 23906
-    },
-    {
-      "epoch": 1.8278571018980445,
-      "grad_norm": 0.0009419335401616991,
-      "learning_rate": 0.00019999835263321067,
-      "loss": 46.0,
-      "step": 23907
-    },
-    {
-      "epoch": 1.8279335588814343,
-      "grad_norm": 0.004340171813964844,
-      "learning_rate": 0.0001999983524953359,
-      "loss": 46.0,
-      "step": 23908
-    },
-    {
-      "epoch": 1.828010015864824,
-      "grad_norm": 0.0004847559903282672,
-      "learning_rate": 0.00019999835235745534,
-      "loss": 46.0,
-      "step": 23909
-    },
-    {
-      "epoch": 1.8280864728482138,
-      "grad_norm": 0.0017887247959151864,
-      "learning_rate": 0.00019999835221956903,
-      "loss": 46.0,
-      "step": 23910
-    },
-    {
-      "epoch": 1.8281629298316036,
-      "grad_norm": 0.014102067798376083,
-      "learning_rate": 0.00019999835208167697,
-      "loss": 46.0,
-      "step": 23911
-    },
-    {
-      "epoch": 1.8282393868149933,
-      "grad_norm": 0.0011384808458387852,
-      "learning_rate": 0.00019999835194377908,
-      "loss": 46.0,
-      "step": 23912
-    },
-    {
-      "epoch": 1.828315843798383,
-      "grad_norm": 0.0021035217214375734,
-      "learning_rate": 0.0001999983518058755,
-      "loss": 46.0,
-      "step": 23913
-    },
-    {
-      "epoch": 1.8283923007817726,
-      "grad_norm": 0.0010395547142252326,
-      "learning_rate": 0.00019999835166796607,
-      "loss": 46.0,
-      "step": 23914
-    },
-    {
-      "epoch": 1.8284687577651624,
-      "grad_norm": 0.0013432534178718925,
-      "learning_rate": 0.00019999835153005092,
-      "loss": 46.0,
-      "step": 23915
-    },
-    {
-      "epoch": 1.8285452147485521,
-      "grad_norm": 0.005125477910041809,
-      "learning_rate": 0.00019999835139213003,
-      "loss": 46.0,
-      "step": 23916
-    },
-    {
-      "epoch": 1.8286216717319417,
-      "grad_norm": 0.0007548199500888586,
-      "learning_rate": 0.00019999835125420333,
-      "loss": 46.0,
-      "step": 23917
-    },
-    {
-      "epoch": 1.8286981287153314,
-      "grad_norm": 0.0029630232602357864,
-      "learning_rate": 0.00019999835111627083,
-      "loss": 46.0,
-      "step": 23918
-    },
-    {
-      "epoch": 1.8287745856987212,
-      "grad_norm": 0.0015760298119857907,
-      "learning_rate": 0.0001999983509783326,
-      "loss": 46.0,
-      "step": 23919
-    },
-    {
-      "epoch": 1.828851042682111,
-      "grad_norm": 0.001539901946671307,
-      "learning_rate": 0.0001999983508403886,
-      "loss": 46.0,
-      "step": 23920
-    },
-    {
-      "epoch": 1.8289274996655007,
-      "grad_norm": 0.007302130106836557,
-      "learning_rate": 0.00019999835070243883,
-      "loss": 46.0,
-      "step": 23921
-    },
-    {
-      "epoch": 1.8290039566488905,
-      "grad_norm": 0.0008780031930655241,
-      "learning_rate": 0.00019999835056448326,
-      "loss": 46.0,
-      "step": 23922
-    },
-    {
-      "epoch": 1.8290804136322802,
-      "grad_norm": 0.0008248342201113701,
-      "learning_rate": 0.00019999835042652198,
-      "loss": 46.0,
-      "step": 23923
-    },
-    {
-      "epoch": 1.82915687061567,
-      "grad_norm": 0.001435629790648818,
-      "learning_rate": 0.00019999835028855487,
-      "loss": 46.0,
-      "step": 23924
-    },
-    {
-      "epoch": 1.8292333275990595,
-      "grad_norm": 0.0005808002315461636,
-      "learning_rate": 0.00019999835015058204,
-      "loss": 46.0,
-      "step": 23925
-    },
-    {
-      "epoch": 1.8293097845824493,
-      "grad_norm": 0.000657212920486927,
-      "learning_rate": 0.00019999835001260338,
-      "loss": 46.0,
-      "step": 23926
-    },
-    {
-      "epoch": 1.829386241565839,
-      "grad_norm": 0.019714007154107094,
-      "learning_rate": 0.000199998349874619,
-      "loss": 46.0,
-      "step": 23927
-    },
-    {
-      "epoch": 1.8294626985492286,
-      "grad_norm": 0.00415431335568428,
-      "learning_rate": 0.00019999834973662885,
-      "loss": 46.0,
-      "step": 23928
-    },
-    {
-      "epoch": 1.8295391555326184,
-      "grad_norm": 0.004770858678966761,
-      "learning_rate": 0.00019999834959863293,
-      "loss": 46.0,
-      "step": 23929
-    },
-    {
-      "epoch": 1.8296156125160081,
-      "grad_norm": 0.002844860777258873,
-      "learning_rate": 0.00019999834946063123,
-      "loss": 46.0,
-      "step": 23930
-    },
-    {
-      "epoch": 1.8296920694993979,
-      "grad_norm": 0.0032237747218459845,
-      "learning_rate": 0.00019999834932262376,
-      "loss": 46.0,
-      "step": 23931
-    },
-    {
-      "epoch": 1.8297685264827876,
-      "grad_norm": 0.0010265747550874949,
-      "learning_rate": 0.0001999983491846105,
-      "loss": 46.0,
-      "step": 23932
-    },
-    {
-      "epoch": 1.8298449834661774,
-      "grad_norm": 0.001270824926905334,
-      "learning_rate": 0.0001999983490465915,
-      "loss": 46.0,
-      "step": 23933
-    },
-    {
-      "epoch": 1.8299214404495672,
-      "grad_norm": 0.0005657592555508018,
-      "learning_rate": 0.00019999834890856673,
-      "loss": 46.0,
-      "step": 23934
-    },
-    {
-      "epoch": 1.829997897432957,
-      "grad_norm": 0.0021599598694592714,
-      "learning_rate": 0.0001999983487705362,
-      "loss": 46.0,
-      "step": 23935
-    },
-    {
-      "epoch": 1.8300743544163465,
-      "grad_norm": 0.0010334642138332129,
-      "learning_rate": 0.00019999834863249985,
-      "loss": 46.0,
-      "step": 23936
-    },
-    {
-      "epoch": 1.8301508113997362,
-      "grad_norm": 0.0005314050358720124,
-      "learning_rate": 0.00019999834849445777,
-      "loss": 46.0,
-      "step": 23937
-    },
-    {
-      "epoch": 1.830227268383126,
-      "grad_norm": 0.0015683239325881004,
-      "learning_rate": 0.00019999834835640994,
-      "loss": 46.0,
-      "step": 23938
-    },
-    {
-      "epoch": 1.8303037253665155,
-      "grad_norm": 0.005588720552623272,
-      "learning_rate": 0.0001999983482183563,
-      "loss": 46.0,
-      "step": 23939
-    },
-    {
-      "epoch": 1.8303801823499053,
-      "grad_norm": 0.0012042330345138907,
-      "learning_rate": 0.0001999983480802969,
-      "loss": 46.0,
-      "step": 23940
-    },
-    {
-      "epoch": 1.830456639333295,
-      "grad_norm": 0.0019481373019516468,
-      "learning_rate": 0.00019999834794223173,
-      "loss": 46.0,
-      "step": 23941
-    },
-    {
-      "epoch": 1.8305330963166848,
-      "grad_norm": 0.0005353696760721505,
-      "learning_rate": 0.0001999983478041608,
-      "loss": 46.0,
-      "step": 23942
-    },
-    {
-      "epoch": 1.8306095533000746,
-      "grad_norm": 0.0012337693478912115,
-      "learning_rate": 0.0001999983476660841,
-      "loss": 46.0,
-      "step": 23943
-    },
-    {
-      "epoch": 1.8306860102834643,
-      "grad_norm": 0.0014952932251617312,
-      "learning_rate": 0.00019999834752800163,
-      "loss": 46.0,
-      "step": 23944
-    },
-    {
-      "epoch": 1.830762467266854,
-      "grad_norm": 0.0009090615203604102,
-      "learning_rate": 0.00019999834738991342,
-      "loss": 46.0,
-      "step": 23945
-    },
-    {
-      "epoch": 1.8308389242502439,
-      "grad_norm": 0.001103985938243568,
-      "learning_rate": 0.00019999834725181937,
-      "loss": 46.0,
-      "step": 23946
-    },
-    {
-      "epoch": 1.8309153812336334,
-      "grad_norm": 0.0006535687716677785,
-      "learning_rate": 0.0001999983471137196,
-      "loss": 46.0,
-      "step": 23947
-    },
-    {
-      "epoch": 1.8309918382170232,
-      "grad_norm": 0.0013266380410641432,
-      "learning_rate": 0.00019999834697561407,
-      "loss": 46.0,
-      "step": 23948
-    },
-    {
-      "epoch": 1.831068295200413,
-      "grad_norm": 0.00342384516261518,
-      "learning_rate": 0.00019999834683750276,
-      "loss": 46.0,
-      "step": 23949
-    },
-    {
-      "epoch": 1.8311447521838025,
-      "grad_norm": 0.0007869650144129992,
-      "learning_rate": 0.00019999834669938568,
-      "loss": 46.0,
-      "step": 23950
-    },
-    {
-      "epoch": 1.8312212091671922,
-      "grad_norm": 0.000919678364880383,
-      "learning_rate": 0.0001999983465612628,
-      "loss": 46.0,
-      "step": 23951
-    },
-    {
-      "epoch": 1.831297666150582,
-      "grad_norm": 0.0009309927700087428,
-      "learning_rate": 0.00019999834642313416,
-      "loss": 46.0,
-      "step": 23952
-    },
-    {
-      "epoch": 1.8313741231339717,
-      "grad_norm": 0.0029769206885248423,
-      "learning_rate": 0.00019999834628499978,
-      "loss": 46.0,
-      "step": 23953
-    },
-    {
-      "epoch": 1.8314505801173615,
-      "grad_norm": 0.003182024462148547,
-      "learning_rate": 0.0001999983461468596,
-      "loss": 46.0,
-      "step": 23954
-    },
-    {
-      "epoch": 1.8315270371007513,
-      "grad_norm": 0.005156370811164379,
-      "learning_rate": 0.00019999834600871368,
-      "loss": 46.0,
-      "step": 23955
-    },
-    {
-      "epoch": 1.831603494084141,
-      "grad_norm": 0.0028223861008882523,
-      "learning_rate": 0.000199998345870562,
-      "loss": 46.0,
-      "step": 23956
-    },
-    {
-      "epoch": 1.8316799510675308,
-      "grad_norm": 0.0010221661068499088,
-      "learning_rate": 0.00019999834573240452,
-      "loss": 46.0,
-      "step": 23957
-    },
-    {
-      "epoch": 1.8317564080509203,
-      "grad_norm": 0.000572660006582737,
-      "learning_rate": 0.00019999834559424127,
-      "loss": 46.0,
-      "step": 23958
-    },
-    {
-      "epoch": 1.83183286503431,
-      "grad_norm": 0.0006711501046083868,
-      "learning_rate": 0.00019999834545607225,
-      "loss": 46.0,
-      "step": 23959
-    },
-    {
-      "epoch": 1.8319093220176996,
-      "grad_norm": 0.0007115493062883615,
-      "learning_rate": 0.00019999834531789746,
-      "loss": 46.0,
-      "step": 23960
-    },
-    {
-      "epoch": 1.8319857790010894,
-      "grad_norm": 0.0006306515424512327,
-      "learning_rate": 0.00019999834517971693,
-      "loss": 46.0,
-      "step": 23961
-    },
-    {
-      "epoch": 1.8320622359844791,
-      "grad_norm": 0.004018108360469341,
-      "learning_rate": 0.00019999834504153062,
-      "loss": 46.0,
-      "step": 23962
-    },
-    {
-      "epoch": 1.832138692967869,
-      "grad_norm": 0.0013060973724350333,
-      "learning_rate": 0.0001999983449033385,
-      "loss": 46.0,
-      "step": 23963
-    },
-    {
-      "epoch": 1.8322151499512587,
-      "grad_norm": 0.001037942012771964,
-      "learning_rate": 0.00019999834476514065,
-      "loss": 46.0,
-      "step": 23964
-    },
-    {
-      "epoch": 1.8322916069346484,
-      "grad_norm": 0.0018497114069759846,
-      "learning_rate": 0.00019999834462693702,
-      "loss": 46.0,
-      "step": 23965
-    },
-    {
-      "epoch": 1.8323680639180382,
-      "grad_norm": 0.0012205595849081874,
-      "learning_rate": 0.00019999834448872764,
-      "loss": 46.0,
-      "step": 23966
-    },
-    {
-      "epoch": 1.832444520901428,
-      "grad_norm": 0.0009895114926621318,
-      "learning_rate": 0.0001999983443505125,
-      "loss": 46.0,
-      "step": 23967
-    },
-    {
-      "epoch": 1.8325209778848177,
-      "grad_norm": 0.0004822291084565222,
-      "learning_rate": 0.00019999834421229154,
-      "loss": 46.0,
-      "step": 23968
-    },
-    {
-      "epoch": 1.8325974348682073,
-      "grad_norm": 0.0015185363590717316,
-      "learning_rate": 0.00019999834407406484,
-      "loss": 46.0,
-      "step": 23969
-    },
-    {
-      "epoch": 1.832673891851597,
-      "grad_norm": 0.004448238294571638,
-      "learning_rate": 0.00019999834393583235,
-      "loss": 46.0,
-      "step": 23970
-    },
-    {
-      "epoch": 1.8327503488349866,
-      "grad_norm": 0.0028663533739745617,
-      "learning_rate": 0.00019999834379759413,
-      "loss": 46.0,
-      "step": 23971
-    },
-    {
-      "epoch": 1.8328268058183763,
-      "grad_norm": 0.0009537548758089542,
-      "learning_rate": 0.00019999834365935011,
-      "loss": 46.0,
-      "step": 23972
-    },
-    {
-      "epoch": 1.832903262801766,
-      "grad_norm": 0.0005530266207642853,
-      "learning_rate": 0.00019999834352110032,
-      "loss": 46.0,
-      "step": 23973
-    },
-    {
-      "epoch": 1.8329797197851558,
-      "grad_norm": 0.002375100739300251,
-      "learning_rate": 0.0001999983433828448,
-      "loss": 46.0,
-      "step": 23974
-    },
-    {
-      "epoch": 1.8330561767685456,
-      "grad_norm": 0.005531785078346729,
-      "learning_rate": 0.00019999834324458345,
-      "loss": 46.0,
-      "step": 23975
-    },
-    {
-      "epoch": 1.8331326337519354,
-      "grad_norm": 0.0005423836992122233,
-      "learning_rate": 0.00019999834310631637,
-      "loss": 46.0,
-      "step": 23976
-    },
-    {
-      "epoch": 1.8332090907353251,
-      "grad_norm": 0.001727782771922648,
-      "learning_rate": 0.00019999834296804348,
-      "loss": 46.0,
-      "step": 23977
-    },
-    {
-      "epoch": 1.8332855477187149,
-      "grad_norm": 0.000533060694579035,
-      "learning_rate": 0.00019999834282976488,
-      "loss": 46.0,
-      "step": 23978
-    },
-    {
-      "epoch": 1.8333620047021046,
-      "grad_norm": 0.0024386525619775057,
-      "learning_rate": 0.00019999834269148048,
-      "loss": 46.0,
-      "step": 23979
-    },
-    {
-      "epoch": 1.8334384616854942,
-      "grad_norm": 0.005179790314286947,
-      "learning_rate": 0.0001999983425531903,
-      "loss": 46.0,
-      "step": 23980
-    },
-    {
-      "epoch": 1.833514918668884,
-      "grad_norm": 0.0018118720036000013,
-      "learning_rate": 0.00019999834241489438,
-      "loss": 46.0,
-      "step": 23981
-    },
-    {
-      "epoch": 1.8335913756522735,
-      "grad_norm": 0.0005988035118207335,
-      "learning_rate": 0.00019999834227659265,
-      "loss": 46.0,
-      "step": 23982
-    },
-    {
-      "epoch": 1.8336678326356632,
-      "grad_norm": 0.0008260435424745083,
-      "learning_rate": 0.00019999834213828518,
-      "loss": 46.0,
-      "step": 23983
-    },
-    {
-      "epoch": 1.833744289619053,
-      "grad_norm": 0.0015572276897728443,
-      "learning_rate": 0.00019999834199997194,
-      "loss": 46.0,
-      "step": 23984
-    },
-    {
-      "epoch": 1.8338207466024428,
-      "grad_norm": 0.0012696572812274098,
-      "learning_rate": 0.00019999834186165295,
-      "loss": 46.0,
-      "step": 23985
-    },
-    {
-      "epoch": 1.8338972035858325,
-      "grad_norm": 0.0006861437577754259,
-      "learning_rate": 0.00019999834172332813,
-      "loss": 46.0,
-      "step": 23986
-    },
-    {
-      "epoch": 1.8339736605692223,
-      "grad_norm": 0.0014596692053601146,
-      "learning_rate": 0.0001999983415849976,
-      "loss": 46.0,
-      "step": 23987
-    },
-    {
-      "epoch": 1.834050117552612,
-      "grad_norm": 0.0033149810042232275,
-      "learning_rate": 0.00019999834144666126,
-      "loss": 46.0,
-      "step": 23988
-    },
-    {
-      "epoch": 1.8341265745360018,
-      "grad_norm": 0.00040873352554626763,
-      "learning_rate": 0.00019999834130831918,
-      "loss": 46.0,
-      "step": 23989
-    },
-    {
-      "epoch": 1.8342030315193913,
-      "grad_norm": 0.001727715483866632,
-      "learning_rate": 0.00019999834116997132,
-      "loss": 46.0,
-      "step": 23990
-    },
-    {
-      "epoch": 1.834279488502781,
-      "grad_norm": 0.0005510460468940437,
-      "learning_rate": 0.0001999983410316177,
-      "loss": 46.0,
-      "step": 23991
-    },
-    {
-      "epoch": 1.8343559454861709,
-      "grad_norm": 0.0008622353198006749,
-      "learning_rate": 0.0001999983408932583,
-      "loss": 46.0,
-      "step": 23992
-    },
-    {
-      "epoch": 1.8344324024695604,
-      "grad_norm": 0.0007293229573406279,
-      "learning_rate": 0.0001999983407548931,
-      "loss": 46.0,
-      "step": 23993
-    },
-    {
-      "epoch": 1.8345088594529502,
-      "grad_norm": 0.0008433896582573652,
-      "learning_rate": 0.0001999983406165222,
-      "loss": 46.0,
-      "step": 23994
-    },
-    {
-      "epoch": 1.83458531643634,
-      "grad_norm": 0.001933095045387745,
-      "learning_rate": 0.00019999834047814547,
-      "loss": 46.0,
-      "step": 23995
-    },
-    {
-      "epoch": 1.8346617734197297,
-      "grad_norm": 0.000515175866894424,
-      "learning_rate": 0.000199998340339763,
-      "loss": 46.0,
-      "step": 23996
-    },
-    {
-      "epoch": 1.8347382304031195,
-      "grad_norm": 0.002671354217454791,
-      "learning_rate": 0.00019999834020137475,
-      "loss": 46.0,
-      "step": 23997
-    },
-    {
-      "epoch": 1.8348146873865092,
-      "grad_norm": 0.0009272352326661348,
-      "learning_rate": 0.00019999834006298074,
-      "loss": 46.0,
-      "step": 23998
-    },
-    {
-      "epoch": 1.834891144369899,
-      "grad_norm": 0.001544341561384499,
-      "learning_rate": 0.00019999833992458095,
-      "loss": 46.0,
-      "step": 23999
-    },
-    {
-      "epoch": 1.8349676013532887,
-      "grad_norm": 0.0009590385016053915,
-      "learning_rate": 0.00019999833978617538,
-      "loss": 46.0,
-      "step": 24000
-    },
-    {
-      "epoch": 1.8350440583366783,
-      "grad_norm": 0.001049233484081924,
-      "learning_rate": 0.00019999833964776407,
-      "loss": 46.0,
-      "step": 24001
-    },
-    {
-      "epoch": 1.835120515320068,
-      "grad_norm": 0.0007759078871458769,
-      "learning_rate": 0.00019999833950934697,
-      "loss": 46.0,
-      "step": 24002
-    },
-    {
-      "epoch": 1.8351969723034578,
-      "grad_norm": 0.005377600435167551,
-      "learning_rate": 0.0001999983393709241,
-      "loss": 46.0,
-      "step": 24003
-    },
-    {
-      "epoch": 1.8352734292868473,
-      "grad_norm": 0.0015093449037522078,
-      "learning_rate": 0.00019999833923249548,
-      "loss": 46.0,
-      "step": 24004
-    },
-    {
-      "epoch": 1.835349886270237,
-      "grad_norm": 0.00214086240157485,
-      "learning_rate": 0.00019999833909406108,
-      "loss": 46.0,
-      "step": 24005
-    },
-    {
-      "epoch": 1.8354263432536269,
-      "grad_norm": 0.0038216100074350834,
-      "learning_rate": 0.0001999983389556209,
-      "loss": 46.0,
-      "step": 24006
-    },
-    {
-      "epoch": 1.8355028002370166,
-      "grad_norm": 0.0028912988491356373,
-      "learning_rate": 0.00019999833881717498,
-      "loss": 46.0,
-      "step": 24007
-    },
-    {
-      "epoch": 1.8355792572204064,
-      "grad_norm": 0.0007045595557428896,
-      "learning_rate": 0.00019999833867872323,
-      "loss": 46.0,
-      "step": 24008
-    },
-    {
-      "epoch": 1.8356557142037961,
-      "grad_norm": 0.0024463674053549767,
-      "learning_rate": 0.00019999833854026579,
-      "loss": 46.0,
-      "step": 24009
-    },
-    {
-      "epoch": 1.835732171187186,
-      "grad_norm": 0.00050332274986431,
-      "learning_rate": 0.00019999833840180254,
-      "loss": 46.0,
-      "step": 24010
-    },
-    {
-      "epoch": 1.8358086281705757,
-      "grad_norm": 0.0026274381671100855,
-      "learning_rate": 0.00019999833826333353,
-      "loss": 46.0,
-      "step": 24011
-    },
-    {
-      "epoch": 1.8358850851539652,
-      "grad_norm": 0.00470306072384119,
-      "learning_rate": 0.00019999833812485874,
-      "loss": 46.0,
-      "step": 24012
-    },
-    {
-      "epoch": 1.835961542137355,
-      "grad_norm": 0.001142130233347416,
-      "learning_rate": 0.00019999833798637818,
-      "loss": 46.0,
-      "step": 24013
-    },
-    {
-      "epoch": 1.8360379991207447,
-      "grad_norm": 0.0006553728599101305,
-      "learning_rate": 0.00019999833784789184,
-      "loss": 46.0,
-      "step": 24014
-    },
-    {
-      "epoch": 1.8361144561041343,
-      "grad_norm": 0.0010038039181381464,
-      "learning_rate": 0.00019999833770939973,
-      "loss": 46.0,
-      "step": 24015
-    },
-    {
-      "epoch": 1.836190913087524,
-      "grad_norm": 0.006854768842458725,
-      "learning_rate": 0.0001999983375709019,
-      "loss": 46.0,
-      "step": 24016
-    },
-    {
-      "epoch": 1.8362673700709138,
-      "grad_norm": 0.002887592650949955,
-      "learning_rate": 0.00019999833743239824,
-      "loss": 46.0,
-      "step": 24017
-    },
-    {
-      "epoch": 1.8363438270543035,
-      "grad_norm": 0.004634858574718237,
-      "learning_rate": 0.00019999833729388884,
-      "loss": 46.0,
-      "step": 24018
-    },
-    {
-      "epoch": 1.8364202840376933,
-      "grad_norm": 0.0022212162148207426,
-      "learning_rate": 0.00019999833715537367,
-      "loss": 46.0,
-      "step": 24019
-    },
-    {
-      "epoch": 1.836496741021083,
-      "grad_norm": 0.0008391932933591306,
-      "learning_rate": 0.00019999833701685274,
-      "loss": 46.0,
-      "step": 24020
-    },
-    {
-      "epoch": 1.8365731980044728,
-      "grad_norm": 0.001955055631697178,
-      "learning_rate": 0.00019999833687832602,
-      "loss": 46.0,
-      "step": 24021
-    },
-    {
-      "epoch": 1.8366496549878626,
-      "grad_norm": 0.001016722060739994,
-      "learning_rate": 0.00019999833673979353,
-      "loss": 46.0,
-      "step": 24022
-    },
-    {
-      "epoch": 1.8367261119712521,
-      "grad_norm": 0.0017843263922259212,
-      "learning_rate": 0.00019999833660125528,
-      "loss": 46.0,
-      "step": 24023
-    },
-    {
-      "epoch": 1.836802568954642,
-      "grad_norm": 0.0011481106048449874,
-      "learning_rate": 0.00019999833646271127,
-      "loss": 46.0,
-      "step": 24024
-    },
-    {
-      "epoch": 1.8368790259380317,
-      "grad_norm": 0.0020079100504517555,
-      "learning_rate": 0.0001999983363241615,
-      "loss": 46.0,
-      "step": 24025
-    },
-    {
-      "epoch": 1.8369554829214212,
-      "grad_norm": 0.001076492597348988,
-      "learning_rate": 0.00019999833618560592,
-      "loss": 46.0,
-      "step": 24026
-    },
-    {
-      "epoch": 1.837031939904811,
-      "grad_norm": 0.0018108839867636561,
-      "learning_rate": 0.00019999833604704458,
-      "loss": 46.0,
-      "step": 24027
-    },
-    {
-      "epoch": 1.8371083968882007,
-      "grad_norm": 0.003120560199022293,
-      "learning_rate": 0.0001999983359084775,
-      "loss": 46.0,
-      "step": 24028
-    },
-    {
-      "epoch": 1.8371848538715905,
-      "grad_norm": 0.0025472554843872786,
-      "learning_rate": 0.00019999833576990464,
-      "loss": 46.0,
-      "step": 24029
-    },
-    {
-      "epoch": 1.8372613108549802,
-      "grad_norm": 0.0004606877628248185,
-      "learning_rate": 0.000199998335631326,
-      "loss": 46.0,
-      "step": 24030
-    },
-    {
-      "epoch": 1.83733776783837,
-      "grad_norm": 0.0023512982297688723,
-      "learning_rate": 0.00019999833549274158,
-      "loss": 46.0,
-      "step": 24031
-    },
-    {
-      "epoch": 1.8374142248217598,
-      "grad_norm": 0.0003429347416386008,
-      "learning_rate": 0.00019999833535415144,
-      "loss": 46.0,
-      "step": 24032
-    },
-    {
-      "epoch": 1.8374906818051495,
-      "grad_norm": 0.0013568336144089699,
-      "learning_rate": 0.0001999983352155555,
-      "loss": 46.0,
-      "step": 24033
-    },
-    {
-      "epoch": 1.837567138788539,
-      "grad_norm": 0.00510236294940114,
-      "learning_rate": 0.00019999833507695376,
-      "loss": 46.0,
-      "step": 24034
-    },
-    {
-      "epoch": 1.8376435957719288,
-      "grad_norm": 0.0009480480221100152,
-      "learning_rate": 0.00019999833493834627,
-      "loss": 46.0,
-      "step": 24035
-    },
-    {
-      "epoch": 1.8377200527553186,
-      "grad_norm": 0.0010470944689586759,
-      "learning_rate": 0.00019999833479973303,
-      "loss": 46.0,
-      "step": 24036
-    },
-    {
-      "epoch": 1.8377965097387081,
-      "grad_norm": 0.0017882988322526217,
-      "learning_rate": 0.000199998334661114,
-      "loss": 46.0,
-      "step": 24037
-    },
-    {
-      "epoch": 1.8378729667220979,
-      "grad_norm": 0.003667086362838745,
-      "learning_rate": 0.00019999833452248922,
-      "loss": 46.0,
-      "step": 24038
-    },
-    {
-      "epoch": 1.8379494237054876,
-      "grad_norm": 0.0012558590387925506,
-      "learning_rate": 0.00019999833438385863,
-      "loss": 46.0,
-      "step": 24039
-    },
-    {
-      "epoch": 1.8380258806888774,
-      "grad_norm": 0.0007093364838510752,
-      "learning_rate": 0.00019999833424522235,
-      "loss": 46.0,
-      "step": 24040
-    },
-    {
-      "epoch": 1.8381023376722672,
-      "grad_norm": 0.001240081270225346,
-      "learning_rate": 0.00019999833410658022,
-      "loss": 46.0,
-      "step": 24041
-    },
-    {
-      "epoch": 1.838178794655657,
-      "grad_norm": 0.0033272383734583855,
-      "learning_rate": 0.00019999833396793236,
-      "loss": 46.0,
-      "step": 24042
-    },
-    {
-      "epoch": 1.8382552516390467,
-      "grad_norm": 0.0016868329839780927,
-      "learning_rate": 0.0001999983338292787,
-      "loss": 46.0,
-      "step": 24043
-    },
-    {
-      "epoch": 1.8383317086224364,
-      "grad_norm": 0.0010811745887622237,
-      "learning_rate": 0.0001999983336906193,
-      "loss": 46.0,
-      "step": 24044
-    },
-    {
-      "epoch": 1.838408165605826,
-      "grad_norm": 0.0007133649778552353,
-      "learning_rate": 0.00019999833355195416,
-      "loss": 46.0,
-      "step": 24045
-    },
-    {
-      "epoch": 1.8384846225892157,
-      "grad_norm": 0.0010707960464060307,
-      "learning_rate": 0.0001999983334132832,
-      "loss": 46.0,
-      "step": 24046
-    },
-    {
-      "epoch": 1.8385610795726055,
-      "grad_norm": 0.0031527355313301086,
-      "learning_rate": 0.00019999833327460649,
-      "loss": 46.0,
-      "step": 24047
-    },
-    {
-      "epoch": 1.838637536555995,
-      "grad_norm": 0.0007617617957293987,
-      "learning_rate": 0.000199998333135924,
-      "loss": 46.0,
-      "step": 24048
-    },
-    {
-      "epoch": 1.8387139935393848,
-      "grad_norm": 0.0016422446351498365,
-      "learning_rate": 0.00019999833299723575,
-      "loss": 46.0,
-      "step": 24049
-    },
-    {
-      "epoch": 1.8387904505227746,
-      "grad_norm": 0.004349039401859045,
-      "learning_rate": 0.0001999983328585417,
-      "loss": 46.0,
-      "step": 24050
-    },
-    {
-      "epoch": 1.8388669075061643,
-      "grad_norm": 0.002776196924969554,
-      "learning_rate": 0.00019999833271984194,
-      "loss": 46.0,
-      "step": 24051
-    },
-    {
-      "epoch": 1.838943364489554,
-      "grad_norm": 0.001398204592987895,
-      "learning_rate": 0.00019999833258113636,
-      "loss": 46.0,
-      "step": 24052
-    },
-    {
-      "epoch": 1.8390198214729438,
-      "grad_norm": 0.0006950525566935539,
-      "learning_rate": 0.00019999833244242505,
-      "loss": 46.0,
-      "step": 24053
-    },
-    {
-      "epoch": 1.8390962784563336,
-      "grad_norm": 0.004921953193843365,
-      "learning_rate": 0.0001999983323037079,
-      "loss": 46.0,
-      "step": 24054
-    },
-    {
-      "epoch": 1.8391727354397234,
-      "grad_norm": 0.00047293800162151456,
-      "learning_rate": 0.00019999833216498506,
-      "loss": 46.0,
-      "step": 24055
-    },
-    {
-      "epoch": 1.839249192423113,
-      "grad_norm": 0.0016401024768128991,
-      "learning_rate": 0.00019999833202625643,
-      "loss": 46.0,
-      "step": 24056
-    },
-    {
-      "epoch": 1.8393256494065027,
-      "grad_norm": 0.0003546375664882362,
-      "learning_rate": 0.00019999833188752203,
-      "loss": 46.0,
-      "step": 24057
-    },
-    {
-      "epoch": 1.8394021063898924,
-      "grad_norm": 0.009104827418923378,
-      "learning_rate": 0.00019999833174878185,
-      "loss": 46.0,
-      "step": 24058
-    },
-    {
-      "epoch": 1.839478563373282,
-      "grad_norm": 0.0014980998821556568,
-      "learning_rate": 0.0001999983316100359,
-      "loss": 46.0,
-      "step": 24059
-    },
-    {
-      "epoch": 1.8395550203566717,
-      "grad_norm": 0.0008139011915773153,
-      "learning_rate": 0.00019999833147128418,
-      "loss": 46.0,
-      "step": 24060
-    },
-    {
-      "epoch": 1.8396314773400615,
-      "grad_norm": 0.0004737124836537987,
-      "learning_rate": 0.0001999983313325267,
-      "loss": 46.0,
-      "step": 24061
-    },
-    {
-      "epoch": 1.8397079343234513,
-      "grad_norm": 0.002801937749609351,
-      "learning_rate": 0.00019999833119376345,
-      "loss": 46.0,
-      "step": 24062
-    },
-    {
-      "epoch": 1.839784391306841,
-      "grad_norm": 0.0006293219048529863,
-      "learning_rate": 0.0001999983310549944,
-      "loss": 46.0,
-      "step": 24063
-    },
-    {
-      "epoch": 1.8398608482902308,
-      "grad_norm": 0.0016475480515509844,
-      "learning_rate": 0.00019999833091621962,
-      "loss": 46.0,
-      "step": 24064
-    },
-    {
-      "epoch": 1.8399373052736205,
-      "grad_norm": 0.0005701237241737545,
-      "learning_rate": 0.00019999833077743905,
-      "loss": 46.0,
-      "step": 24065
-    },
-    {
-      "epoch": 1.8400137622570103,
-      "grad_norm": 0.0005094964290037751,
-      "learning_rate": 0.00019999833063865272,
-      "loss": 46.0,
-      "step": 24066
-    },
-    {
-      "epoch": 1.8400902192403998,
-      "grad_norm": 0.00043167563853785396,
-      "learning_rate": 0.00019999833049986064,
-      "loss": 46.0,
-      "step": 24067
-    },
-    {
-      "epoch": 1.8401666762237896,
-      "grad_norm": 0.001459711231291294,
-      "learning_rate": 0.00019999833036106276,
-      "loss": 46.0,
-      "step": 24068
-    },
-    {
-      "epoch": 1.8402431332071794,
-      "grad_norm": 0.0012314132181927562,
-      "learning_rate": 0.0001999983302222591,
-      "loss": 46.0,
-      "step": 24069
-    },
-    {
-      "epoch": 1.840319590190569,
-      "grad_norm": 0.00071003008633852,
-      "learning_rate": 0.0001999983300834497,
-      "loss": 46.0,
-      "step": 24070
-    },
-    {
-      "epoch": 1.8403960471739587,
-      "grad_norm": 0.0010922811925411224,
-      "learning_rate": 0.00019999832994463453,
-      "loss": 46.0,
-      "step": 24071
-    },
-    {
-      "epoch": 1.8404725041573484,
-      "grad_norm": 0.0020431228913366795,
-      "learning_rate": 0.00019999832980581358,
-      "loss": 46.0,
-      "step": 24072
-    },
-    {
-      "epoch": 1.8405489611407382,
-      "grad_norm": 0.00042537436820566654,
-      "learning_rate": 0.00019999832966698686,
-      "loss": 46.0,
-      "step": 24073
-    },
-    {
-      "epoch": 1.840625418124128,
-      "grad_norm": 0.002330636139959097,
-      "learning_rate": 0.00019999832952815436,
-      "loss": 46.0,
-      "step": 24074
-    },
-    {
-      "epoch": 1.8407018751075177,
-      "grad_norm": 0.002542466390877962,
-      "learning_rate": 0.00019999832938931612,
-      "loss": 46.0,
-      "step": 24075
-    },
-    {
-      "epoch": 1.8407783320909075,
-      "grad_norm": 0.0005889342282898724,
-      "learning_rate": 0.00019999832925047208,
-      "loss": 46.0,
-      "step": 24076
-    },
-    {
-      "epoch": 1.8408547890742972,
-      "grad_norm": 0.001610838226042688,
-      "learning_rate": 0.00019999832911162232,
-      "loss": 46.0,
-      "step": 24077
-    },
-    {
-      "epoch": 1.8409312460576868,
-      "grad_norm": 0.0012444778112694621,
-      "learning_rate": 0.0001999983289727667,
-      "loss": 46.0,
-      "step": 24078
-    },
-    {
-      "epoch": 1.8410077030410765,
-      "grad_norm": 0.0010654201032593846,
-      "learning_rate": 0.0001999983288339054,
-      "loss": 46.0,
-      "step": 24079
-    },
-    {
-      "epoch": 1.8410841600244663,
-      "grad_norm": 0.005882246419787407,
-      "learning_rate": 0.0001999983286950383,
-      "loss": 46.0,
-      "step": 24080
-    },
-    {
-      "epoch": 1.8411606170078558,
-      "grad_norm": 0.0012291306629776955,
-      "learning_rate": 0.00019999832855616543,
-      "loss": 46.0,
-      "step": 24081
-    },
-    {
-      "epoch": 1.8412370739912456,
-      "grad_norm": 0.0009694420150481164,
-      "learning_rate": 0.00019999832841728678,
-      "loss": 46.0,
-      "step": 24082
-    },
-    {
-      "epoch": 1.8413135309746353,
-      "grad_norm": 0.005421114154160023,
-      "learning_rate": 0.00019999832827840238,
-      "loss": 46.0,
-      "step": 24083
-    },
-    {
-      "epoch": 1.841389987958025,
-      "grad_norm": 0.005127171520143747,
-      "learning_rate": 0.00019999832813951218,
-      "loss": 46.0,
-      "step": 24084
-    },
-    {
-      "epoch": 1.8414664449414149,
-      "grad_norm": 0.0010110702132806182,
-      "learning_rate": 0.00019999832800061623,
-      "loss": 46.0,
-      "step": 24085
-    },
-    {
-      "epoch": 1.8415429019248046,
-      "grad_norm": 0.0014125773450359702,
-      "learning_rate": 0.0001999983278617145,
-      "loss": 46.0,
-      "step": 24086
-    },
-    {
-      "epoch": 1.8416193589081944,
-      "grad_norm": 0.0005949250189587474,
-      "learning_rate": 0.000199998327722807,
-      "loss": 46.0,
-      "step": 24087
-    },
-    {
-      "epoch": 1.8416958158915842,
-      "grad_norm": 0.0014252071268856525,
-      "learning_rate": 0.00019999832758389377,
-      "loss": 46.0,
-      "step": 24088
-    },
-    {
-      "epoch": 1.8417722728749737,
-      "grad_norm": 0.0009450973011553288,
-      "learning_rate": 0.00019999832744497473,
-      "loss": 46.0,
-      "step": 24089
-    },
-    {
-      "epoch": 1.8418487298583635,
-      "grad_norm": 0.0010257799876853824,
-      "learning_rate": 0.00019999832730604994,
-      "loss": 46.0,
-      "step": 24090
-    },
-    {
-      "epoch": 1.841925186841753,
-      "grad_norm": 0.0009541959734633565,
-      "learning_rate": 0.00019999832716711938,
-      "loss": 46.0,
-      "step": 24091
-    },
-    {
-      "epoch": 1.8420016438251428,
-      "grad_norm": 0.001316535985097289,
-      "learning_rate": 0.00019999832702818305,
-      "loss": 46.0,
-      "step": 24092
-    },
-    {
-      "epoch": 1.8420781008085325,
-      "grad_norm": 0.001131929107941687,
-      "learning_rate": 0.00019999832688924094,
-      "loss": 46.0,
-      "step": 24093
-    },
-    {
-      "epoch": 1.8421545577919223,
-      "grad_norm": 0.0019271766068413854,
-      "learning_rate": 0.00019999832675029306,
-      "loss": 46.0,
-      "step": 24094
-    },
-    {
-      "epoch": 1.842231014775312,
-      "grad_norm": 0.0007769920630380511,
-      "learning_rate": 0.0001999983266113394,
-      "loss": 46.0,
-      "step": 24095
-    },
-    {
-      "epoch": 1.8423074717587018,
-      "grad_norm": 0.0008586702169850469,
-      "learning_rate": 0.00019999832647238,
-      "loss": 46.0,
-      "step": 24096
-    },
-    {
-      "epoch": 1.8423839287420916,
-      "grad_norm": 0.0008281109621748328,
-      "learning_rate": 0.00019999832633341483,
-      "loss": 46.0,
-      "step": 24097
-    },
-    {
-      "epoch": 1.8424603857254813,
-      "grad_norm": 0.0011058085365220904,
-      "learning_rate": 0.00019999832619444386,
-      "loss": 46.0,
-      "step": 24098
-    },
-    {
-      "epoch": 1.842536842708871,
-      "grad_norm": 0.0008683353080414236,
-      "learning_rate": 0.00019999832605546716,
-      "loss": 46.0,
-      "step": 24099
-    },
-    {
-      "epoch": 1.8426132996922606,
-      "grad_norm": 0.000852260272949934,
-      "learning_rate": 0.00019999832591648464,
-      "loss": 46.0,
-      "step": 24100
-    },
-    {
-      "epoch": 1.8426897566756504,
-      "grad_norm": 0.0015756089705973864,
-      "learning_rate": 0.0001999983257774964,
-      "loss": 46.0,
-      "step": 24101
-    },
-    {
-      "epoch": 1.84276621365904,
-      "grad_norm": 0.0009335257345810533,
-      "learning_rate": 0.0001999983256385024,
-      "loss": 46.0,
-      "step": 24102
-    },
-    {
-      "epoch": 1.8428426706424297,
-      "grad_norm": 0.0011852050665766,
-      "learning_rate": 0.00019999832549950257,
-      "loss": 46.0,
-      "step": 24103
-    },
-    {
-      "epoch": 1.8429191276258194,
-      "grad_norm": 0.0047043063677847385,
-      "learning_rate": 0.00019999832536049699,
-      "loss": 46.0,
-      "step": 24104
-    },
-    {
-      "epoch": 1.8429955846092092,
-      "grad_norm": 0.005377822555601597,
-      "learning_rate": 0.00019999832522148565,
-      "loss": 46.0,
-      "step": 24105
-    },
-    {
-      "epoch": 1.843072041592599,
-      "grad_norm": 0.0017967717722058296,
-      "learning_rate": 0.00019999832508246854,
-      "loss": 46.0,
-      "step": 24106
-    },
-    {
-      "epoch": 1.8431484985759887,
-      "grad_norm": 0.0011472634505480528,
-      "learning_rate": 0.0001999983249434457,
-      "loss": 46.0,
-      "step": 24107
-    },
-    {
-      "epoch": 1.8432249555593785,
-      "grad_norm": 0.0007326845661737025,
-      "learning_rate": 0.00019999832480441704,
-      "loss": 46.0,
-      "step": 24108
-    },
-    {
-      "epoch": 1.8433014125427682,
-      "grad_norm": 0.0022598314099013805,
-      "learning_rate": 0.0001999983246653826,
-      "loss": 46.0,
-      "step": 24109
-    },
-    {
-      "epoch": 1.843377869526158,
-      "grad_norm": 0.0011644657934084535,
-      "learning_rate": 0.00019999832452634244,
-      "loss": 46.0,
-      "step": 24110
-    },
-    {
-      "epoch": 1.8434543265095475,
-      "grad_norm": 0.0004907556576654315,
-      "learning_rate": 0.0001999983243872965,
-      "loss": 46.0,
-      "step": 24111
-    },
-    {
-      "epoch": 1.8435307834929373,
-      "grad_norm": 0.0007266573375090957,
-      "learning_rate": 0.00019999832424824477,
-      "loss": 46.0,
-      "step": 24112
-    },
-    {
-      "epoch": 1.8436072404763268,
-      "grad_norm": 0.0006670369766652584,
-      "learning_rate": 0.00019999832410918727,
-      "loss": 46.0,
-      "step": 24113
-    },
-    {
-      "epoch": 1.8436836974597166,
-      "grad_norm": 0.0007401605253107846,
-      "learning_rate": 0.000199998323970124,
-      "loss": 46.0,
-      "step": 24114
-    },
-    {
-      "epoch": 1.8437601544431064,
-      "grad_norm": 0.0013634668430313468,
-      "learning_rate": 0.000199998323831055,
-      "loss": 46.0,
-      "step": 24115
-    },
-    {
-      "epoch": 1.8438366114264961,
-      "grad_norm": 0.002419000491499901,
-      "learning_rate": 0.00019999832369198018,
-      "loss": 46.0,
-      "step": 24116
-    },
-    {
-      "epoch": 1.843913068409886,
-      "grad_norm": 0.0018210975686088204,
-      "learning_rate": 0.0001999983235528996,
-      "loss": 46.0,
-      "step": 24117
-    },
-    {
-      "epoch": 1.8439895253932757,
-      "grad_norm": 0.0007208941970020533,
-      "learning_rate": 0.00019999832341381329,
-      "loss": 46.0,
-      "step": 24118
-    },
-    {
-      "epoch": 1.8440659823766654,
-      "grad_norm": 0.0020097719971090555,
-      "learning_rate": 0.00019999832327472118,
-      "loss": 46.0,
-      "step": 24119
-    },
-    {
-      "epoch": 1.8441424393600552,
-      "grad_norm": 0.0013149511069059372,
-      "learning_rate": 0.00019999832313562327,
-      "loss": 46.0,
-      "step": 24120
-    },
-    {
-      "epoch": 1.844218896343445,
-      "grad_norm": 0.0010369336232542992,
-      "learning_rate": 0.00019999832299651965,
-      "loss": 46.0,
-      "step": 24121
-    },
-    {
-      "epoch": 1.8442953533268345,
-      "grad_norm": 0.002007534261792898,
-      "learning_rate": 0.00019999832285741022,
-      "loss": 46.0,
-      "step": 24122
-    },
-    {
-      "epoch": 1.8443718103102242,
-      "grad_norm": 0.00027686450630426407,
-      "learning_rate": 0.00019999832271829502,
-      "loss": 46.0,
-      "step": 24123
-    },
-    {
-      "epoch": 1.8444482672936138,
-      "grad_norm": 0.0005134936654940248,
-      "learning_rate": 0.00019999832257917407,
-      "loss": 46.0,
-      "step": 24124
-    },
-    {
-      "epoch": 1.8445247242770035,
-      "grad_norm": 0.000567734008654952,
-      "learning_rate": 0.00019999832244004735,
-      "loss": 46.0,
-      "step": 24125
-    },
-    {
-      "epoch": 1.8446011812603933,
-      "grad_norm": 0.004502412863075733,
-      "learning_rate": 0.00019999832230091483,
-      "loss": 46.0,
-      "step": 24126
-    },
-    {
-      "epoch": 1.844677638243783,
-      "grad_norm": 0.0007243787404149771,
-      "learning_rate": 0.0001999983221617766,
-      "loss": 46.0,
-      "step": 24127
-    },
-    {
-      "epoch": 1.8447540952271728,
-      "grad_norm": 0.0008244475466199219,
-      "learning_rate": 0.00019999832202263255,
-      "loss": 46.0,
-      "step": 24128
-    },
-    {
-      "epoch": 1.8448305522105626,
-      "grad_norm": 0.002063054358586669,
-      "learning_rate": 0.00019999832188348276,
-      "loss": 46.0,
-      "step": 24129
-    },
-    {
-      "epoch": 1.8449070091939523,
-      "grad_norm": 0.0007605401915498078,
-      "learning_rate": 0.00019999832174432718,
-      "loss": 46.0,
-      "step": 24130
-    },
-    {
-      "epoch": 1.844983466177342,
-      "grad_norm": 0.0018615626031532884,
-      "learning_rate": 0.00019999832160516584,
-      "loss": 46.0,
-      "step": 24131
-    },
-    {
-      "epoch": 1.8450599231607316,
-      "grad_norm": 0.0009881944861263037,
-      "learning_rate": 0.00019999832146599876,
-      "loss": 46.0,
-      "step": 24132
-    },
-    {
-      "epoch": 1.8451363801441214,
-      "grad_norm": 0.000806662195827812,
-      "learning_rate": 0.00019999832132682586,
-      "loss": 46.0,
-      "step": 24133
-    },
-    {
-      "epoch": 1.8452128371275112,
-      "grad_norm": 0.004836688749492168,
-      "learning_rate": 0.0001999983211876472,
-      "loss": 46.0,
-      "step": 24134
-    },
-    {
-      "epoch": 1.8452892941109007,
-      "grad_norm": 0.0004650605551432818,
-      "learning_rate": 0.00019999832104846278,
-      "loss": 46.0,
-      "step": 24135
-    },
-    {
-      "epoch": 1.8453657510942905,
-      "grad_norm": 0.001476981327868998,
-      "learning_rate": 0.00019999832090927263,
-      "loss": 46.0,
-      "step": 24136
-    },
-    {
-      "epoch": 1.8454422080776802,
-      "grad_norm": 0.0005782695370726287,
-      "learning_rate": 0.00019999832077007666,
-      "loss": 46.0,
-      "step": 24137
-    },
-    {
-      "epoch": 1.84551866506107,
-      "grad_norm": 0.0011770921992138028,
-      "learning_rate": 0.0001999983206308749,
-      "loss": 46.0,
-      "step": 24138
-    },
-    {
-      "epoch": 1.8455951220444597,
-      "grad_norm": 0.003743873443454504,
-      "learning_rate": 0.00019999832049166742,
-      "loss": 46.0,
-      "step": 24139
-    },
-    {
-      "epoch": 1.8456715790278495,
-      "grad_norm": 0.0036658523604273796,
-      "learning_rate": 0.00019999832035245415,
-      "loss": 46.0,
-      "step": 24140
-    },
-    {
-      "epoch": 1.8457480360112393,
-      "grad_norm": 0.004497750662267208,
-      "learning_rate": 0.00019999832021323514,
-      "loss": 46.0,
-      "step": 24141
-    },
-    {
-      "epoch": 1.845824492994629,
-      "grad_norm": 0.000493558996822685,
-      "learning_rate": 0.00019999832007401032,
-      "loss": 46.0,
-      "step": 24142
-    },
-    {
-      "epoch": 1.8459009499780186,
-      "grad_norm": 0.0009167492389678955,
-      "learning_rate": 0.00019999831993477976,
-      "loss": 46.0,
-      "step": 24143
-    },
-    {
-      "epoch": 1.8459774069614083,
-      "grad_norm": 0.00026153927319683135,
-      "learning_rate": 0.00019999831979554343,
-      "loss": 46.0,
-      "step": 24144
-    },
-    {
-      "epoch": 1.846053863944798,
-      "grad_norm": 0.0036145898047834635,
-      "learning_rate": 0.0001999983196563013,
-      "loss": 46.0,
-      "step": 24145
-    },
-    {
-      "epoch": 1.8461303209281876,
-      "grad_norm": 0.0012017266126349568,
-      "learning_rate": 0.00019999831951705342,
-      "loss": 46.0,
-      "step": 24146
-    },
-    {
-      "epoch": 1.8462067779115774,
-      "grad_norm": 0.0007951234001666307,
-      "learning_rate": 0.00019999831937779977,
-      "loss": 46.0,
-      "step": 24147
-    },
-    {
-      "epoch": 1.8462832348949672,
-      "grad_norm": 0.0009669052669778466,
-      "learning_rate": 0.00019999831923854037,
-      "loss": 46.0,
-      "step": 24148
-    },
-    {
-      "epoch": 1.846359691878357,
-      "grad_norm": 0.002213160041719675,
-      "learning_rate": 0.00019999831909927517,
-      "loss": 46.0,
-      "step": 24149
-    },
-    {
-      "epoch": 1.8464361488617467,
-      "grad_norm": 0.0007168302545323968,
-      "learning_rate": 0.0001999983189600042,
-      "loss": 46.0,
-      "step": 24150
-    },
-    {
-      "epoch": 1.8465126058451364,
-      "grad_norm": 0.0014908612938597798,
-      "learning_rate": 0.00019999831882072745,
-      "loss": 46.0,
-      "step": 24151
-    },
-    {
-      "epoch": 1.8465890628285262,
-      "grad_norm": 0.0010554497130215168,
-      "learning_rate": 0.00019999831868144495,
-      "loss": 46.0,
-      "step": 24152
-    },
-    {
-      "epoch": 1.846665519811916,
-      "grad_norm": 0.00048272041021846235,
-      "learning_rate": 0.0001999983185421567,
-      "loss": 46.0,
-      "step": 24153
-    },
-    {
-      "epoch": 1.8467419767953055,
-      "grad_norm": 0.0024181148037314415,
-      "learning_rate": 0.00019999831840286267,
-      "loss": 46.0,
-      "step": 24154
-    },
-    {
-      "epoch": 1.8468184337786953,
-      "grad_norm": 0.0017503125127404928,
-      "learning_rate": 0.0001999983182635629,
-      "loss": 46.0,
-      "step": 24155
-    },
-    {
-      "epoch": 1.846894890762085,
-      "grad_norm": 0.0011440166272222996,
-      "learning_rate": 0.0001999983181242573,
-      "loss": 46.0,
-      "step": 24156
-    },
-    {
-      "epoch": 1.8469713477454746,
-      "grad_norm": 0.002768615260720253,
-      "learning_rate": 0.00019999831798494594,
-      "loss": 46.0,
-      "step": 24157
-    },
-    {
-      "epoch": 1.8470478047288643,
-      "grad_norm": 0.0012028460623696446,
-      "learning_rate": 0.00019999831784562884,
-      "loss": 46.0,
-      "step": 24158
-    },
-    {
-      "epoch": 1.847124261712254,
-      "grad_norm": 0.0015924079343676567,
-      "learning_rate": 0.00019999831770630596,
-      "loss": 46.0,
-      "step": 24159
-    },
-    {
-      "epoch": 1.8472007186956438,
-      "grad_norm": 0.0010614608181640506,
-      "learning_rate": 0.00019999831756697728,
-      "loss": 46.0,
-      "step": 24160
-    },
-    {
-      "epoch": 1.8472771756790336,
-      "grad_norm": 0.001744345761835575,
-      "learning_rate": 0.00019999831742764288,
-      "loss": 46.0,
-      "step": 24161
-    },
-    {
-      "epoch": 1.8473536326624234,
-      "grad_norm": 0.0016607313882559538,
-      "learning_rate": 0.0001999983172883027,
-      "loss": 46.0,
-      "step": 24162
-    },
-    {
-      "epoch": 1.8474300896458131,
-      "grad_norm": 0.0005190973752178252,
-      "learning_rate": 0.00019999831714895673,
-      "loss": 46.0,
-      "step": 24163
-    },
-    {
-      "epoch": 1.8475065466292029,
-      "grad_norm": 0.0010341916931793094,
-      "learning_rate": 0.00019999831700960501,
-      "loss": 46.0,
-      "step": 24164
-    },
-    {
-      "epoch": 1.8475830036125924,
-      "grad_norm": 0.0008348671835847199,
-      "learning_rate": 0.0001999983168702475,
-      "loss": 46.0,
-      "step": 24165
-    },
-    {
-      "epoch": 1.8476594605959822,
-      "grad_norm": 0.0004457558679860085,
-      "learning_rate": 0.00019999831673088423,
-      "loss": 46.0,
-      "step": 24166
-    },
-    {
-      "epoch": 1.847735917579372,
-      "grad_norm": 0.0006765160360373557,
-      "learning_rate": 0.0001999983165915152,
-      "loss": 46.0,
-      "step": 24167
-    },
-    {
-      "epoch": 1.8478123745627615,
-      "grad_norm": 0.0006265683914534748,
-      "learning_rate": 0.0001999983164521404,
-      "loss": 46.0,
-      "step": 24168
-    },
-    {
-      "epoch": 1.8478888315461512,
-      "grad_norm": 0.0017007633578032255,
-      "learning_rate": 0.00019999831631275982,
-      "loss": 46.0,
-      "step": 24169
-    },
-    {
-      "epoch": 1.847965288529541,
-      "grad_norm": 0.0015448606573045254,
-      "learning_rate": 0.00019999831617337348,
-      "loss": 46.0,
-      "step": 24170
-    },
-    {
-      "epoch": 1.8480417455129308,
-      "grad_norm": 0.003544918028637767,
-      "learning_rate": 0.00019999831603398132,
-      "loss": 46.0,
-      "step": 24171
-    },
-    {
-      "epoch": 1.8481182024963205,
-      "grad_norm": 0.0008507675374858081,
-      "learning_rate": 0.00019999831589458347,
-      "loss": 46.0,
-      "step": 24172
-    },
-    {
-      "epoch": 1.8481946594797103,
-      "grad_norm": 0.0016558998031541705,
-      "learning_rate": 0.00019999831575517982,
-      "loss": 46.0,
-      "step": 24173
-    },
-    {
-      "epoch": 1.8482711164631,
-      "grad_norm": 0.0016026265220716596,
-      "learning_rate": 0.0001999983156157704,
-      "loss": 46.0,
-      "step": 24174
-    },
-    {
-      "epoch": 1.8483475734464898,
-      "grad_norm": 0.0007121895905584097,
-      "learning_rate": 0.00019999831547635522,
-      "loss": 46.0,
-      "step": 24175
-    },
-    {
-      "epoch": 1.8484240304298794,
-      "grad_norm": 0.0003881090960931033,
-      "learning_rate": 0.00019999831533693425,
-      "loss": 46.0,
-      "step": 24176
-    },
-    {
-      "epoch": 1.8485004874132691,
-      "grad_norm": 0.0013854090357199311,
-      "learning_rate": 0.0001999983151975075,
-      "loss": 46.0,
-      "step": 24177
-    },
-    {
-      "epoch": 1.8485769443966589,
-      "grad_norm": 0.0011741119669750333,
-      "learning_rate": 0.000199998315058075,
-      "loss": 46.0,
-      "step": 24178
-    },
-    {
-      "epoch": 1.8486534013800484,
-      "grad_norm": 0.000811099773272872,
-      "learning_rate": 0.00019999831491863672,
-      "loss": 46.0,
-      "step": 24179
-    },
-    {
-      "epoch": 1.8487298583634382,
-      "grad_norm": 0.00379693484865129,
-      "learning_rate": 0.0001999983147791927,
-      "loss": 46.0,
-      "step": 24180
-    },
-    {
-      "epoch": 1.848806315346828,
-      "grad_norm": 0.0003310217580292374,
-      "learning_rate": 0.00019999831463974287,
-      "loss": 46.0,
-      "step": 24181
-    },
-    {
-      "epoch": 1.8488827723302177,
-      "grad_norm": 0.0012897496344521642,
-      "learning_rate": 0.0001999983145002873,
-      "loss": 46.0,
-      "step": 24182
-    },
-    {
-      "epoch": 1.8489592293136075,
-      "grad_norm": 0.0004151734756305814,
-      "learning_rate": 0.00019999831436082598,
-      "loss": 46.0,
-      "step": 24183
-    },
-    {
-      "epoch": 1.8490356862969972,
-      "grad_norm": 0.002177771646529436,
-      "learning_rate": 0.00019999831422135884,
-      "loss": 46.0,
-      "step": 24184
-    },
-    {
-      "epoch": 1.849112143280387,
-      "grad_norm": 0.0026943094562739134,
-      "learning_rate": 0.00019999831408188597,
-      "loss": 46.0,
-      "step": 24185
-    },
-    {
-      "epoch": 1.8491886002637767,
-      "grad_norm": 0.001724479952827096,
-      "learning_rate": 0.0001999983139424073,
-      "loss": 46.0,
-      "step": 24186
-    },
-    {
-      "epoch": 1.8492650572471663,
-      "grad_norm": 0.0006051899981684983,
-      "learning_rate": 0.0001999983138029229,
-      "loss": 46.0,
-      "step": 24187
-    },
-    {
-      "epoch": 1.849341514230556,
-      "grad_norm": 0.0022523547522723675,
-      "learning_rate": 0.0001999983136634327,
-      "loss": 46.0,
-      "step": 24188
-    },
-    {
-      "epoch": 1.8494179712139458,
-      "grad_norm": 0.0007815954741090536,
-      "learning_rate": 0.00019999831352393672,
-      "loss": 46.0,
-      "step": 24189
-    },
-    {
-      "epoch": 1.8494944281973353,
-      "grad_norm": 0.0035633379593491554,
-      "learning_rate": 0.00019999831338443497,
-      "loss": 46.0,
-      "step": 24190
-    },
-    {
-      "epoch": 1.849570885180725,
-      "grad_norm": 0.0012178485048934817,
-      "learning_rate": 0.0001999983132449275,
-      "loss": 46.0,
-      "step": 24191
-    },
-    {
-      "epoch": 1.8496473421641149,
-      "grad_norm": 0.002476731315255165,
-      "learning_rate": 0.00019999831310541421,
-      "loss": 46.0,
-      "step": 24192
-    },
-    {
-      "epoch": 1.8497237991475046,
-      "grad_norm": 0.0010207825107499957,
-      "learning_rate": 0.00019999831296589518,
-      "loss": 46.0,
-      "step": 24193
-    },
-    {
-      "epoch": 1.8498002561308944,
-      "grad_norm": 0.00046677826321683824,
-      "learning_rate": 0.00019999831282637036,
-      "loss": 46.0,
-      "step": 24194
-    },
-    {
-      "epoch": 1.8498767131142841,
-      "grad_norm": 0.004186040721833706,
-      "learning_rate": 0.0001999983126868398,
-      "loss": 46.0,
-      "step": 24195
-    },
-    {
-      "epoch": 1.849953170097674,
-      "grad_norm": 0.0006600570632144809,
-      "learning_rate": 0.00019999831254730345,
-      "loss": 46.0,
-      "step": 24196
-    },
-    {
-      "epoch": 1.8500296270810637,
-      "grad_norm": 0.0011380210053175688,
-      "learning_rate": 0.00019999831240776131,
-      "loss": 46.0,
-      "step": 24197
-    },
-    {
-      "epoch": 1.8501060840644532,
-      "grad_norm": 0.0014121267013251781,
-      "learning_rate": 0.0001999983122682134,
-      "loss": 46.0,
-      "step": 24198
-    },
-    {
-      "epoch": 1.850182541047843,
-      "grad_norm": 0.002492221537977457,
-      "learning_rate": 0.00019999831212865976,
-      "loss": 46.0,
-      "step": 24199
-    },
-    {
-      "epoch": 1.8502589980312327,
-      "grad_norm": 0.002731338609009981,
-      "learning_rate": 0.00019999831198910036,
-      "loss": 46.0,
-      "step": 24200
-    },
-    {
-      "epoch": 1.8503354550146223,
-      "grad_norm": 0.0006185830570757389,
-      "learning_rate": 0.00019999831184953516,
-      "loss": 46.0,
-      "step": 24201
-    },
-    {
-      "epoch": 1.850411911998012,
-      "grad_norm": 0.0008674424607306719,
-      "learning_rate": 0.0001999983117099642,
-      "loss": 46.0,
-      "step": 24202
-    },
-    {
-      "epoch": 1.8504883689814018,
-      "grad_norm": 0.0005722398054786026,
-      "learning_rate": 0.00019999831157038744,
-      "loss": 46.0,
-      "step": 24203
-    },
-    {
-      "epoch": 1.8505648259647915,
-      "grad_norm": 0.0009735375642776489,
-      "learning_rate": 0.00019999831143080495,
-      "loss": 46.0,
-      "step": 24204
-    },
-    {
-      "epoch": 1.8506412829481813,
-      "grad_norm": 0.002922134706750512,
-      "learning_rate": 0.00019999831129121666,
-      "loss": 46.0,
-      "step": 24205
-    },
-    {
-      "epoch": 1.850717739931571,
-      "grad_norm": 0.0005558782140724361,
-      "learning_rate": 0.00019999831115162265,
-      "loss": 46.0,
-      "step": 24206
-    },
-    {
-      "epoch": 1.8507941969149608,
-      "grad_norm": 0.0017395829781889915,
-      "learning_rate": 0.0001999983110120228,
-      "loss": 46.0,
-      "step": 24207
-    },
-    {
-      "epoch": 1.8508706538983506,
-      "grad_norm": 0.0012845478486269712,
-      "learning_rate": 0.00019999831087241722,
-      "loss": 46.0,
-      "step": 24208
-    },
-    {
-      "epoch": 1.8509471108817401,
-      "grad_norm": 0.0009314960334450006,
-      "learning_rate": 0.0001999983107328059,
-      "loss": 46.0,
-      "step": 24209
-    },
-    {
-      "epoch": 1.85102356786513,
-      "grad_norm": 0.0028973405715078115,
-      "learning_rate": 0.00019999831059318876,
-      "loss": 46.0,
-      "step": 24210
-    },
-    {
-      "epoch": 1.8511000248485197,
-      "grad_norm": 0.01054801419377327,
-      "learning_rate": 0.00019999831045356588,
-      "loss": 46.0,
-      "step": 24211
-    },
-    {
-      "epoch": 1.8511764818319092,
-      "grad_norm": 0.0014662856701761484,
-      "learning_rate": 0.0001999983103139372,
-      "loss": 46.0,
-      "step": 24212
-    },
-    {
-      "epoch": 1.851252938815299,
-      "grad_norm": 0.000963919737841934,
-      "learning_rate": 0.0001999983101743028,
-      "loss": 46.0,
-      "step": 24213
-    },
-    {
-      "epoch": 1.8513293957986887,
-      "grad_norm": 0.0008107098983600736,
-      "learning_rate": 0.0001999983100346626,
-      "loss": 46.0,
-      "step": 24214
-    },
-    {
-      "epoch": 1.8514058527820785,
-      "grad_norm": 0.008049333468079567,
-      "learning_rate": 0.00019999830989501664,
-      "loss": 46.0,
-      "step": 24215
-    },
-    {
-      "epoch": 1.8514823097654682,
-      "grad_norm": 0.0014261500909924507,
-      "learning_rate": 0.00019999830975536492,
-      "loss": 46.0,
-      "step": 24216
-    },
-    {
-      "epoch": 1.851558766748858,
-      "grad_norm": 0.0007155823986977339,
-      "learning_rate": 0.00019999830961570737,
-      "loss": 46.0,
-      "step": 24217
-    },
-    {
-      "epoch": 1.8516352237322478,
-      "grad_norm": 0.0010858505265787244,
-      "learning_rate": 0.00019999830947604413,
-      "loss": 46.0,
-      "step": 24218
-    },
-    {
-      "epoch": 1.8517116807156375,
-      "grad_norm": 0.0005700408946722746,
-      "learning_rate": 0.0001999983093363751,
-      "loss": 46.0,
-      "step": 24219
-    },
-    {
-      "epoch": 1.851788137699027,
-      "grad_norm": 0.0005103947478346527,
-      "learning_rate": 0.00019999830919670026,
-      "loss": 46.0,
-      "step": 24220
-    },
-    {
-      "epoch": 1.8518645946824168,
-      "grad_norm": 0.005461809225380421,
-      "learning_rate": 0.00019999830905701967,
-      "loss": 46.0,
-      "step": 24221
-    },
-    {
-      "epoch": 1.8519410516658064,
-      "grad_norm": 0.0006621723296120763,
-      "learning_rate": 0.00019999830891733332,
-      "loss": 46.0,
-      "step": 24222
-    },
-    {
-      "epoch": 1.8520175086491961,
-      "grad_norm": 0.0006473535322584212,
-      "learning_rate": 0.0001999983087776412,
-      "loss": 46.0,
-      "step": 24223
-    },
-    {
-      "epoch": 1.8520939656325859,
-      "grad_norm": 0.0009456057450734079,
-      "learning_rate": 0.0001999983086379433,
-      "loss": 46.0,
-      "step": 24224
-    },
-    {
-      "epoch": 1.8521704226159756,
-      "grad_norm": 0.0033528481144458055,
-      "learning_rate": 0.00019999830849823965,
-      "loss": 46.0,
-      "step": 24225
-    },
-    {
-      "epoch": 1.8522468795993654,
-      "grad_norm": 0.001613075379282236,
-      "learning_rate": 0.00019999830835853023,
-      "loss": 46.0,
-      "step": 24226
-    },
-    {
-      "epoch": 1.8523233365827552,
-      "grad_norm": 0.0012833367800340056,
-      "learning_rate": 0.00019999830821881503,
-      "loss": 46.0,
-      "step": 24227
-    },
-    {
-      "epoch": 1.852399793566145,
-      "grad_norm": 0.00898823980242014,
-      "learning_rate": 0.00019999830807909406,
-      "loss": 46.0,
-      "step": 24228
-    },
-    {
-      "epoch": 1.8524762505495347,
-      "grad_norm": 0.002006369410082698,
-      "learning_rate": 0.00019999830793936732,
-      "loss": 46.0,
-      "step": 24229
-    },
-    {
-      "epoch": 1.8525527075329244,
-      "grad_norm": 0.0016314416425302625,
-      "learning_rate": 0.0001999983077996348,
-      "loss": 46.0,
-      "step": 24230
-    },
-    {
-      "epoch": 1.852629164516314,
-      "grad_norm": 0.0009637699695304036,
-      "learning_rate": 0.00019999830765989653,
-      "loss": 46.0,
-      "step": 24231
-    },
-    {
-      "epoch": 1.8527056214997037,
-      "grad_norm": 0.0012076377170160413,
-      "learning_rate": 0.0001999983075201525,
-      "loss": 46.0,
-      "step": 24232
-    },
-    {
-      "epoch": 1.8527820784830933,
-      "grad_norm": 0.002940573263913393,
-      "learning_rate": 0.00019999830738040268,
-      "loss": 46.0,
-      "step": 24233
-    },
-    {
-      "epoch": 1.852858535466483,
-      "grad_norm": 0.0006985941436141729,
-      "learning_rate": 0.0001999983072406471,
-      "loss": 46.0,
-      "step": 24234
-    },
-    {
-      "epoch": 1.8529349924498728,
-      "grad_norm": 0.0002745035744737834,
-      "learning_rate": 0.00019999830710088574,
-      "loss": 46.0,
-      "step": 24235
-    },
-    {
-      "epoch": 1.8530114494332626,
-      "grad_norm": 0.004302772227674723,
-      "learning_rate": 0.0001999983069611186,
-      "loss": 46.0,
-      "step": 24236
-    },
-    {
-      "epoch": 1.8530879064166523,
-      "grad_norm": 0.0004975069314241409,
-      "learning_rate": 0.0001999983068213457,
-      "loss": 46.0,
-      "step": 24237
-    },
-    {
-      "epoch": 1.853164363400042,
-      "grad_norm": 0.004065149463713169,
-      "learning_rate": 0.00019999830668156708,
-      "loss": 46.0,
-      "step": 24238
-    },
-    {
-      "epoch": 1.8532408203834319,
-      "grad_norm": 0.0019213083432987332,
-      "learning_rate": 0.00019999830654178263,
-      "loss": 46.0,
-      "step": 24239
-    },
-    {
-      "epoch": 1.8533172773668216,
-      "grad_norm": 0.0006144684157334268,
-      "learning_rate": 0.00019999830640199243,
-      "loss": 46.0,
-      "step": 24240
-    },
-    {
-      "epoch": 1.8533937343502114,
-      "grad_norm": 0.001071637263521552,
-      "learning_rate": 0.00019999830626219644,
-      "loss": 46.0,
-      "step": 24241
-    },
-    {
-      "epoch": 1.853470191333601,
-      "grad_norm": 0.0008036333601921797,
-      "learning_rate": 0.00019999830612239472,
-      "loss": 46.0,
-      "step": 24242
-    },
-    {
-      "epoch": 1.8535466483169907,
-      "grad_norm": 0.0008139242418110371,
-      "learning_rate": 0.00019999830598258723,
-      "loss": 46.0,
-      "step": 24243
-    },
-    {
-      "epoch": 1.8536231053003802,
-      "grad_norm": 0.002035158919170499,
-      "learning_rate": 0.00019999830584277394,
-      "loss": 46.0,
-      "step": 24244
-    },
-    {
-      "epoch": 1.85369956228377,
-      "grad_norm": 0.0006918600993230939,
-      "learning_rate": 0.00019999830570295487,
-      "loss": 46.0,
-      "step": 24245
-    },
-    {
-      "epoch": 1.8537760192671597,
-      "grad_norm": 0.0010036255698651075,
-      "learning_rate": 0.00019999830556313006,
-      "loss": 46.0,
-      "step": 24246
-    },
-    {
-      "epoch": 1.8538524762505495,
-      "grad_norm": 0.005046769976615906,
-      "learning_rate": 0.0001999983054232995,
-      "loss": 46.0,
-      "step": 24247
-    },
-    {
-      "epoch": 1.8539289332339393,
-      "grad_norm": 0.0015095500275492668,
-      "learning_rate": 0.00019999830528346312,
-      "loss": 46.0,
-      "step": 24248
-    },
-    {
-      "epoch": 1.854005390217329,
-      "grad_norm": 0.00347004272043705,
-      "learning_rate": 0.000199998305143621,
-      "loss": 46.0,
-      "step": 24249
-    },
-    {
-      "epoch": 1.8540818472007188,
-      "grad_norm": 0.0014457611832767725,
-      "learning_rate": 0.0001999983050037731,
-      "loss": 46.0,
-      "step": 24250
-    },
-    {
-      "epoch": 1.8541583041841085,
-      "grad_norm": 0.0007619456155225635,
-      "learning_rate": 0.00019999830486391943,
-      "loss": 46.0,
-      "step": 24251
-    },
-    {
-      "epoch": 1.8542347611674983,
-      "grad_norm": 0.008653547614812851,
-      "learning_rate": 0.00019999830472406004,
-      "loss": 46.0,
-      "step": 24252
-    },
-    {
-      "epoch": 1.8543112181508878,
-      "grad_norm": 0.0010393777629360557,
-      "learning_rate": 0.00019999830458419479,
-      "loss": 46.0,
-      "step": 24253
-    },
-    {
-      "epoch": 1.8543876751342776,
-      "grad_norm": 0.0008056523511186242,
-      "learning_rate": 0.00019999830444432384,
-      "loss": 46.0,
-      "step": 24254
-    },
-    {
-      "epoch": 1.8544641321176671,
-      "grad_norm": 0.0012520093005150557,
-      "learning_rate": 0.0001999983043044471,
-      "loss": 46.0,
-      "step": 24255
-    },
-    {
-      "epoch": 1.854540589101057,
-      "grad_norm": 0.0006997457821853459,
-      "learning_rate": 0.0001999983041645646,
-      "loss": 46.0,
-      "step": 24256
-    },
-    {
-      "epoch": 1.8546170460844467,
-      "grad_norm": 0.004774757660925388,
-      "learning_rate": 0.00019999830402467632,
-      "loss": 46.0,
-      "step": 24257
-    },
-    {
-      "epoch": 1.8546935030678364,
-      "grad_norm": 0.0007005062070675194,
-      "learning_rate": 0.00019999830388478228,
-      "loss": 46.0,
-      "step": 24258
-    },
-    {
-      "epoch": 1.8547699600512262,
-      "grad_norm": 0.001984376460313797,
-      "learning_rate": 0.00019999830374488244,
-      "loss": 46.0,
-      "step": 24259
-    },
-    {
-      "epoch": 1.854846417034616,
-      "grad_norm": 0.0004810243262909353,
-      "learning_rate": 0.00019999830360497686,
-      "loss": 46.0,
-      "step": 24260
-    },
-    {
-      "epoch": 1.8549228740180057,
-      "grad_norm": 0.001636349712498486,
-      "learning_rate": 0.0001999983034650655,
-      "loss": 46.0,
-      "step": 24261
-    },
-    {
-      "epoch": 1.8549993310013955,
-      "grad_norm": 0.0006077042198739946,
-      "learning_rate": 0.00019999830332514837,
-      "loss": 46.0,
-      "step": 24262
-    },
-    {
-      "epoch": 1.855075787984785,
-      "grad_norm": 0.002515554428100586,
-      "learning_rate": 0.00019999830318522547,
-      "loss": 46.0,
-      "step": 24263
-    },
-    {
-      "epoch": 1.8551522449681748,
-      "grad_norm": 0.0008373632445000112,
-      "learning_rate": 0.00019999830304529682,
-      "loss": 46.0,
-      "step": 24264
-    },
-    {
-      "epoch": 1.8552287019515645,
-      "grad_norm": 0.0011075373040512204,
-      "learning_rate": 0.0001999983029053624,
-      "loss": 46.0,
-      "step": 24265
-    },
-    {
-      "epoch": 1.855305158934954,
-      "grad_norm": 0.026234447956085205,
-      "learning_rate": 0.00019999830276542217,
-      "loss": 46.0,
-      "step": 24266
-    },
-    {
-      "epoch": 1.8553816159183438,
-      "grad_norm": 0.005558467470109463,
-      "learning_rate": 0.0001999983026254762,
-      "loss": 46.0,
-      "step": 24267
-    },
-    {
-      "epoch": 1.8554580729017336,
-      "grad_norm": 0.0014849754516035318,
-      "learning_rate": 0.00019999830248552445,
-      "loss": 46.0,
-      "step": 24268
-    },
-    {
-      "epoch": 1.8555345298851234,
-      "grad_norm": 0.001033682725392282,
-      "learning_rate": 0.00019999830234556694,
-      "loss": 46.0,
-      "step": 24269
-    },
-    {
-      "epoch": 1.8556109868685131,
-      "grad_norm": 0.0011668866500258446,
-      "learning_rate": 0.00019999830220560365,
-      "loss": 46.0,
-      "step": 24270
-    },
-    {
-      "epoch": 1.8556874438519029,
-      "grad_norm": 0.002063876949250698,
-      "learning_rate": 0.0001999983020656346,
-      "loss": 46.0,
-      "step": 24271
-    },
-    {
-      "epoch": 1.8557639008352926,
-      "grad_norm": 0.0005701023619621992,
-      "learning_rate": 0.0001999983019256598,
-      "loss": 46.0,
-      "step": 24272
-    },
-    {
-      "epoch": 1.8558403578186824,
-      "grad_norm": 0.0007977632340043783,
-      "learning_rate": 0.0001999983017856792,
-      "loss": 46.0,
-      "step": 24273
-    },
-    {
-      "epoch": 1.855916814802072,
-      "grad_norm": 0.0007731782970950007,
-      "learning_rate": 0.00019999830164569283,
-      "loss": 46.0,
-      "step": 24274
-    },
-    {
-      "epoch": 1.8559932717854617,
-      "grad_norm": 0.0006289740558713675,
-      "learning_rate": 0.0001999983015057007,
-      "loss": 46.0,
-      "step": 24275
-    },
-    {
-      "epoch": 1.8560697287688515,
-      "grad_norm": 0.0010380460880696774,
-      "learning_rate": 0.00019999830136570282,
-      "loss": 46.0,
-      "step": 24276
-    },
-    {
-      "epoch": 1.856146185752241,
-      "grad_norm": 0.005266075022518635,
-      "learning_rate": 0.00019999830122569915,
-      "loss": 46.0,
-      "step": 24277
-    },
-    {
-      "epoch": 1.8562226427356308,
-      "grad_norm": 0.0026753107085824013,
-      "learning_rate": 0.0001999983010856897,
-      "loss": 46.0,
-      "step": 24278
-    },
-    {
-      "epoch": 1.8562990997190205,
-      "grad_norm": 0.0013278336264193058,
-      "learning_rate": 0.0001999983009456745,
-      "loss": 46.0,
-      "step": 24279
-    },
-    {
-      "epoch": 1.8563755567024103,
-      "grad_norm": 0.0006706651183776557,
-      "learning_rate": 0.00019999830080565353,
-      "loss": 46.0,
-      "step": 24280
-    },
-    {
-      "epoch": 1.8564520136858,
-      "grad_norm": 0.0008870938327163458,
-      "learning_rate": 0.0001999983006656268,
-      "loss": 46.0,
-      "step": 24281
-    },
-    {
-      "epoch": 1.8565284706691898,
-      "grad_norm": 0.0008552668150514364,
-      "learning_rate": 0.00019999830052559427,
-      "loss": 46.0,
-      "step": 24282
-    },
-    {
-      "epoch": 1.8566049276525796,
-      "grad_norm": 0.0005379029898904264,
-      "learning_rate": 0.00019999830038555598,
-      "loss": 46.0,
-      "step": 24283
-    },
-    {
-      "epoch": 1.8566813846359693,
-      "grad_norm": 0.0010993879986926913,
-      "learning_rate": 0.00019999830024551194,
-      "loss": 46.0,
-      "step": 24284
-    },
-    {
-      "epoch": 1.8567578416193589,
-      "grad_norm": 0.002867314266040921,
-      "learning_rate": 0.00019999830010546213,
-      "loss": 46.0,
-      "step": 24285
-    },
-    {
-      "epoch": 1.8568342986027486,
-      "grad_norm": 0.001601767842657864,
-      "learning_rate": 0.00019999829996540652,
-      "loss": 46.0,
-      "step": 24286
-    },
-    {
-      "epoch": 1.8569107555861384,
-      "grad_norm": 0.005336458329111338,
-      "learning_rate": 0.00019999829982534517,
-      "loss": 46.0,
-      "step": 24287
-    },
-    {
-      "epoch": 1.856987212569528,
-      "grad_norm": 0.0009401165298186243,
-      "learning_rate": 0.00019999829968527804,
-      "loss": 46.0,
-      "step": 24288
-    },
-    {
-      "epoch": 1.8570636695529177,
-      "grad_norm": 0.0017366111278533936,
-      "learning_rate": 0.00019999829954520513,
-      "loss": 46.0,
-      "step": 24289
-    },
-    {
-      "epoch": 1.8571401265363074,
-      "grad_norm": 0.000529967132024467,
-      "learning_rate": 0.00019999829940512646,
-      "loss": 46.0,
-      "step": 24290
-    },
-    {
-      "epoch": 1.8572165835196972,
-      "grad_norm": 0.0022200760431587696,
-      "learning_rate": 0.00019999829926504203,
-      "loss": 46.0,
-      "step": 24291
-    },
-    {
-      "epoch": 1.857293040503087,
-      "grad_norm": 0.0008585151517763734,
-      "learning_rate": 0.00019999829912495184,
-      "loss": 46.0,
-      "step": 24292
-    },
-    {
-      "epoch": 1.8573694974864767,
-      "grad_norm": 0.0011055023642256856,
-      "learning_rate": 0.00019999829898485584,
-      "loss": 46.0,
-      "step": 24293
-    },
-    {
-      "epoch": 1.8574459544698665,
-      "grad_norm": 0.001331742969341576,
-      "learning_rate": 0.0001999982988447541,
-      "loss": 46.0,
-      "step": 24294
-    },
-    {
-      "epoch": 1.8575224114532563,
-      "grad_norm": 0.001066754455678165,
-      "learning_rate": 0.00019999829870464658,
-      "loss": 46.0,
-      "step": 24295
-    },
-    {
-      "epoch": 1.8575988684366458,
-      "grad_norm": 0.0007064155070111156,
-      "learning_rate": 0.00019999829856453332,
-      "loss": 46.0,
-      "step": 24296
-    },
-    {
-      "epoch": 1.8576753254200356,
-      "grad_norm": 0.0017156120156869292,
-      "learning_rate": 0.00019999829842441425,
-      "loss": 46.0,
-      "step": 24297
-    },
-    {
-      "epoch": 1.8577517824034253,
-      "grad_norm": 0.0012110660318285227,
-      "learning_rate": 0.00019999829828428944,
-      "loss": 46.0,
-      "step": 24298
-    },
-    {
-      "epoch": 1.8578282393868149,
-      "grad_norm": 0.0020337607711553574,
-      "learning_rate": 0.00019999829814415883,
-      "loss": 46.0,
-      "step": 24299
-    },
-    {
-      "epoch": 1.8579046963702046,
-      "grad_norm": 0.0006884110043756664,
-      "learning_rate": 0.00019999829800402248,
-      "loss": 46.0,
-      "step": 24300
-    },
-    {
-      "epoch": 1.8579811533535944,
-      "grad_norm": 0.0013276750687509775,
-      "learning_rate": 0.00019999829786388035,
-      "loss": 46.0,
-      "step": 24301
-    },
-    {
-      "epoch": 1.8580576103369841,
-      "grad_norm": 0.0034233753103762865,
-      "learning_rate": 0.00019999829772373245,
-      "loss": 46.0,
-      "step": 24302
-    },
-    {
-      "epoch": 1.858134067320374,
-      "grad_norm": 0.0007153255864977837,
-      "learning_rate": 0.00019999829758357877,
-      "loss": 46.0,
-      "step": 24303
-    },
-    {
-      "epoch": 1.8582105243037637,
-      "grad_norm": 0.002965073101222515,
-      "learning_rate": 0.00019999829744341935,
-      "loss": 46.0,
-      "step": 24304
-    },
-    {
-      "epoch": 1.8582869812871534,
-      "grad_norm": 0.0028012837283313274,
-      "learning_rate": 0.00019999829730325412,
-      "loss": 46.0,
-      "step": 24305
-    },
-    {
-      "epoch": 1.8583634382705432,
-      "grad_norm": 0.00914770271629095,
-      "learning_rate": 0.00019999829716308315,
-      "loss": 46.0,
-      "step": 24306
-    },
-    {
-      "epoch": 1.8584398952539327,
-      "grad_norm": 0.0009376664529554546,
-      "learning_rate": 0.0001999982970229064,
-      "loss": 46.0,
-      "step": 24307
-    },
-    {
-      "epoch": 1.8585163522373225,
-      "grad_norm": 0.001472039963118732,
-      "learning_rate": 0.00019999829688272387,
-      "loss": 46.0,
-      "step": 24308
-    },
-    {
-      "epoch": 1.8585928092207122,
-      "grad_norm": 0.0006524782511405647,
-      "learning_rate": 0.0001999982967425356,
-      "loss": 46.0,
-      "step": 24309
-    },
-    {
-      "epoch": 1.8586692662041018,
-      "grad_norm": 0.000704131496604532,
-      "learning_rate": 0.00019999829660234152,
-      "loss": 46.0,
-      "step": 24310
-    },
-    {
-      "epoch": 1.8587457231874915,
-      "grad_norm": 0.000869977637194097,
-      "learning_rate": 0.00019999829646214173,
-      "loss": 46.0,
-      "step": 24311
-    },
-    {
-      "epoch": 1.8588221801708813,
-      "grad_norm": 0.002646066015586257,
-      "learning_rate": 0.00019999829632193612,
-      "loss": 46.0,
-      "step": 24312
-    },
-    {
-      "epoch": 1.858898637154271,
-      "grad_norm": 0.005296837538480759,
-      "learning_rate": 0.00019999829618172474,
-      "loss": 46.0,
-      "step": 24313
-    },
-    {
-      "epoch": 1.8589750941376608,
-      "grad_norm": 0.0011698000598698854,
-      "learning_rate": 0.0001999982960415076,
-      "loss": 46.0,
-      "step": 24314
-    },
-    {
-      "epoch": 1.8590515511210506,
-      "grad_norm": 0.0007332656532526016,
-      "learning_rate": 0.0001999982959012847,
-      "loss": 46.0,
-      "step": 24315
-    },
-    {
-      "epoch": 1.8591280081044403,
-      "grad_norm": 0.0012183219660073519,
-      "learning_rate": 0.00019999829576105603,
-      "loss": 46.0,
-      "step": 24316
-    },
-    {
-      "epoch": 1.85920446508783,
-      "grad_norm": 0.0010472637368366122,
-      "learning_rate": 0.0001999982956208216,
-      "loss": 46.0,
-      "step": 24317
-    },
-    {
-      "epoch": 1.8592809220712196,
-      "grad_norm": 0.0008507959428243339,
-      "learning_rate": 0.0001999982954805814,
-      "loss": 46.0,
-      "step": 24318
-    },
-    {
-      "epoch": 1.8593573790546094,
-      "grad_norm": 0.0007265129825100303,
-      "learning_rate": 0.00019999829534033542,
-      "loss": 46.0,
-      "step": 24319
-    },
-    {
-      "epoch": 1.8594338360379992,
-      "grad_norm": 0.0006136031006462872,
-      "learning_rate": 0.00019999829520008368,
-      "loss": 46.0,
-      "step": 24320
-    },
-    {
-      "epoch": 1.8595102930213887,
-      "grad_norm": 0.0007494620513170958,
-      "learning_rate": 0.00019999829505982616,
-      "loss": 46.0,
-      "step": 24321
-    },
-    {
-      "epoch": 1.8595867500047785,
-      "grad_norm": 0.0020406614057719707,
-      "learning_rate": 0.00019999829491956284,
-      "loss": 46.0,
-      "step": 24322
-    },
-    {
-      "epoch": 1.8596632069881682,
-      "grad_norm": 0.0006749049643985927,
-      "learning_rate": 0.0001999982947792938,
-      "loss": 46.0,
-      "step": 24323
-    },
-    {
-      "epoch": 1.859739663971558,
-      "grad_norm": 0.0007904137601144612,
-      "learning_rate": 0.00019999829463901897,
-      "loss": 46.0,
-      "step": 24324
-    },
-    {
-      "epoch": 1.8598161209549477,
-      "grad_norm": 0.00148884451482445,
-      "learning_rate": 0.00019999829449873836,
-      "loss": 46.0,
-      "step": 24325
-    },
-    {
-      "epoch": 1.8598925779383375,
-      "grad_norm": 0.0005614770343527198,
-      "learning_rate": 0.000199998294358452,
-      "loss": 46.0,
-      "step": 24326
-    },
-    {
-      "epoch": 1.8599690349217273,
-      "grad_norm": 0.000606716435868293,
-      "learning_rate": 0.00019999829421815988,
-      "loss": 46.0,
-      "step": 24327
-    },
-    {
-      "epoch": 1.860045491905117,
-      "grad_norm": 0.004627090413123369,
-      "learning_rate": 0.00019999829407786197,
-      "loss": 46.0,
-      "step": 24328
-    },
-    {
-      "epoch": 1.8601219488885066,
-      "grad_norm": 0.010523718781769276,
-      "learning_rate": 0.0001999982939375583,
-      "loss": 46.0,
-      "step": 24329
-    },
-    {
-      "epoch": 1.8601984058718963,
-      "grad_norm": 0.0008912984048947692,
-      "learning_rate": 0.00019999829379724888,
-      "loss": 46.0,
-      "step": 24330
-    },
-    {
-      "epoch": 1.860274862855286,
-      "grad_norm": 0.00428125960752368,
-      "learning_rate": 0.00019999829365693365,
-      "loss": 46.0,
-      "step": 24331
-    },
-    {
-      "epoch": 1.8603513198386756,
-      "grad_norm": 0.0007216015364974737,
-      "learning_rate": 0.00019999829351661269,
-      "loss": 46.0,
-      "step": 24332
-    },
-    {
-      "epoch": 1.8604277768220654,
-      "grad_norm": 0.0003870258806273341,
-      "learning_rate": 0.00019999829337628592,
-      "loss": 46.0,
-      "step": 24333
-    },
-    {
-      "epoch": 1.8605042338054552,
-      "grad_norm": 0.0009758063824847341,
-      "learning_rate": 0.0001999982932359534,
-      "loss": 46.0,
-      "step": 24334
-    },
-    {
-      "epoch": 1.860580690788845,
-      "grad_norm": 0.0008671103860251606,
-      "learning_rate": 0.0001999982930956151,
-      "loss": 46.0,
-      "step": 24335
-    },
-    {
-      "epoch": 1.8606571477722347,
-      "grad_norm": 0.0025011859834194183,
-      "learning_rate": 0.00019999829295527105,
-      "loss": 46.0,
-      "step": 24336
-    },
-    {
-      "epoch": 1.8607336047556244,
-      "grad_norm": 0.0011880442034453154,
-      "learning_rate": 0.00019999829281492124,
-      "loss": 46.0,
-      "step": 24337
-    },
-    {
-      "epoch": 1.8608100617390142,
-      "grad_norm": 0.0004758826980832964,
-      "learning_rate": 0.00019999829267456566,
-      "loss": 46.0,
-      "step": 24338
-    },
-    {
-      "epoch": 1.860886518722404,
-      "grad_norm": 0.003106341464444995,
-      "learning_rate": 0.00019999829253420428,
-      "loss": 46.0,
-      "step": 24339
-    },
-    {
-      "epoch": 1.8609629757057935,
-      "grad_norm": 0.0009115770226344466,
-      "learning_rate": 0.00019999829239383715,
-      "loss": 46.0,
-      "step": 24340
-    },
-    {
-      "epoch": 1.8610394326891833,
-      "grad_norm": 0.0018286345293745399,
-      "learning_rate": 0.00019999829225346422,
-      "loss": 46.0,
-      "step": 24341
-    },
-    {
-      "epoch": 1.861115889672573,
-      "grad_norm": 0.0003300034150015563,
-      "learning_rate": 0.00019999829211308554,
-      "loss": 46.0,
-      "step": 24342
-    },
-    {
-      "epoch": 1.8611923466559626,
-      "grad_norm": 0.0007188476738519967,
-      "learning_rate": 0.0001999982919727011,
-      "loss": 46.0,
-      "step": 24343
-    },
-    {
-      "epoch": 1.8612688036393523,
-      "grad_norm": 0.00057169608771801,
-      "learning_rate": 0.0001999982918323109,
-      "loss": 46.0,
-      "step": 24344
-    },
-    {
-      "epoch": 1.861345260622742,
-      "grad_norm": 0.0012310597812756896,
-      "learning_rate": 0.0001999982916919149,
-      "loss": 46.0,
-      "step": 24345
-    },
-    {
-      "epoch": 1.8614217176061318,
-      "grad_norm": 0.0005620619049295783,
-      "learning_rate": 0.00019999829155151316,
-      "loss": 46.0,
-      "step": 24346
-    },
-    {
-      "epoch": 1.8614981745895216,
-      "grad_norm": 0.0016944174421951175,
-      "learning_rate": 0.00019999829141110565,
-      "loss": 46.0,
-      "step": 24347
-    },
-    {
-      "epoch": 1.8615746315729114,
-      "grad_norm": 0.0011592715745791793,
-      "learning_rate": 0.00019999829127069233,
-      "loss": 46.0,
-      "step": 24348
-    },
-    {
-      "epoch": 1.8616510885563011,
-      "grad_norm": 0.000989419175311923,
-      "learning_rate": 0.0001999982911302733,
-      "loss": 46.0,
-      "step": 24349
-    },
-    {
-      "epoch": 1.8617275455396909,
-      "grad_norm": 0.0010174572234973311,
-      "learning_rate": 0.00019999829098984843,
-      "loss": 46.0,
-      "step": 24350
-    },
-    {
-      "epoch": 1.8618040025230804,
-      "grad_norm": 0.001392244128510356,
-      "learning_rate": 0.00019999829084941782,
-      "loss": 46.0,
-      "step": 24351
-    },
-    {
-      "epoch": 1.8618804595064702,
-      "grad_norm": 0.0012236990733072162,
-      "learning_rate": 0.0001999982907089815,
-      "loss": 46.0,
-      "step": 24352
-    },
-    {
-      "epoch": 1.86195691648986,
-      "grad_norm": 0.0009973953710868955,
-      "learning_rate": 0.00019999829056853937,
-      "loss": 46.0,
-      "step": 24353
-    },
-    {
-      "epoch": 1.8620333734732495,
-      "grad_norm": 0.005353773478418589,
-      "learning_rate": 0.00019999829042809144,
-      "loss": 46.0,
-      "step": 24354
-    },
-    {
-      "epoch": 1.8621098304566392,
-      "grad_norm": 0.0007683031726628542,
-      "learning_rate": 0.00019999829028763774,
-      "loss": 46.0,
-      "step": 24355
-    },
-    {
-      "epoch": 1.862186287440029,
-      "grad_norm": 0.0009289855370298028,
-      "learning_rate": 0.00019999829014717832,
-      "loss": 46.0,
-      "step": 24356
-    },
-    {
-      "epoch": 1.8622627444234188,
-      "grad_norm": 0.002057724166661501,
-      "learning_rate": 0.0001999982900067131,
-      "loss": 46.0,
-      "step": 24357
-    },
-    {
-      "epoch": 1.8623392014068085,
-      "grad_norm": 0.0004398580058477819,
-      "learning_rate": 0.00019999828986624213,
-      "loss": 46.0,
-      "step": 24358
-    },
-    {
-      "epoch": 1.8624156583901983,
-      "grad_norm": 0.0012263068929314613,
-      "learning_rate": 0.00019999828972576538,
-      "loss": 46.0,
-      "step": 24359
-    },
-    {
-      "epoch": 1.862492115373588,
-      "grad_norm": 0.0006371165509335697,
-      "learning_rate": 0.00019999828958528284,
-      "loss": 46.0,
-      "step": 24360
-    },
-    {
-      "epoch": 1.8625685723569778,
-      "grad_norm": 0.0007121982634998858,
-      "learning_rate": 0.00019999828944479455,
-      "loss": 46.0,
-      "step": 24361
-    },
-    {
-      "epoch": 1.8626450293403674,
-      "grad_norm": 0.0009636137983761728,
-      "learning_rate": 0.0001999982893043005,
-      "loss": 46.0,
-      "step": 24362
-    },
-    {
-      "epoch": 1.8627214863237571,
-      "grad_norm": 0.009043118916451931,
-      "learning_rate": 0.00019999828916380066,
-      "loss": 46.0,
-      "step": 24363
-    },
-    {
-      "epoch": 1.8627979433071467,
-      "grad_norm": 0.0006459135911427438,
-      "learning_rate": 0.00019999828902329505,
-      "loss": 46.0,
-      "step": 24364
-    },
-    {
-      "epoch": 1.8628744002905364,
-      "grad_norm": 0.0008643659530207515,
-      "learning_rate": 0.0001999982888827837,
-      "loss": 46.0,
-      "step": 24365
-    },
-    {
-      "epoch": 1.8629508572739262,
-      "grad_norm": 0.0010988058056682348,
-      "learning_rate": 0.00019999828874226654,
-      "loss": 46.0,
-      "step": 24366
-    },
-    {
-      "epoch": 1.863027314257316,
-      "grad_norm": 0.0008805750403553247,
-      "learning_rate": 0.00019999828860174364,
-      "loss": 46.0,
-      "step": 24367
-    },
-    {
-      "epoch": 1.8631037712407057,
-      "grad_norm": 0.0027690345887094736,
-      "learning_rate": 0.00019999828846121496,
-      "loss": 46.0,
-      "step": 24368
-    },
-    {
-      "epoch": 1.8631802282240955,
-      "grad_norm": 0.0021754212211817503,
-      "learning_rate": 0.00019999828832068051,
-      "loss": 46.0,
-      "step": 24369
-    },
-    {
-      "epoch": 1.8632566852074852,
-      "grad_norm": 0.0009094441775232553,
-      "learning_rate": 0.0001999982881801403,
-      "loss": 46.0,
-      "step": 24370
-    },
-    {
-      "epoch": 1.863333142190875,
-      "grad_norm": 0.0017631532391533256,
-      "learning_rate": 0.0001999982880395943,
-      "loss": 46.0,
-      "step": 24371
-    },
-    {
-      "epoch": 1.8634095991742647,
-      "grad_norm": 0.0023792963474988937,
-      "learning_rate": 0.00019999828789904256,
-      "loss": 46.0,
-      "step": 24372
-    },
-    {
-      "epoch": 1.8634860561576543,
-      "grad_norm": 0.0008597279665991664,
-      "learning_rate": 0.00019999828775848504,
-      "loss": 46.0,
-      "step": 24373
-    },
-    {
-      "epoch": 1.863562513141044,
-      "grad_norm": 0.0013290847418829799,
-      "learning_rate": 0.00019999828761792175,
-      "loss": 46.0,
-      "step": 24374
-    },
-    {
-      "epoch": 1.8636389701244336,
-      "grad_norm": 0.0005624671466648579,
-      "learning_rate": 0.00019999828747735266,
-      "loss": 46.0,
-      "step": 24375
-    },
-    {
-      "epoch": 1.8637154271078233,
-      "grad_norm": 0.014460711739957333,
-      "learning_rate": 0.00019999828733677783,
-      "loss": 46.0,
-      "step": 24376
-    },
-    {
-      "epoch": 1.863791884091213,
-      "grad_norm": 0.0007604233687743545,
-      "learning_rate": 0.00019999828719619722,
-      "loss": 46.0,
-      "step": 24377
-    },
-    {
-      "epoch": 1.8638683410746029,
-      "grad_norm": 0.0005617126007564366,
-      "learning_rate": 0.00019999828705561087,
-      "loss": 46.0,
-      "step": 24378
-    },
-    {
-      "epoch": 1.8639447980579926,
-      "grad_norm": 0.004724816884845495,
-      "learning_rate": 0.0001999982869150187,
-      "loss": 46.0,
-      "step": 24379
-    },
-    {
-      "epoch": 1.8640212550413824,
-      "grad_norm": 0.002206844510510564,
-      "learning_rate": 0.0001999982867744208,
-      "loss": 46.0,
-      "step": 24380
-    },
-    {
-      "epoch": 1.8640977120247721,
-      "grad_norm": 0.0007577717187814415,
-      "learning_rate": 0.0001999982866338171,
-      "loss": 46.0,
-      "step": 24381
-    },
-    {
-      "epoch": 1.864174169008162,
-      "grad_norm": 0.0020656753331422806,
-      "learning_rate": 0.0001999982864932077,
-      "loss": 46.0,
-      "step": 24382
-    },
-    {
-      "epoch": 1.8642506259915517,
-      "grad_norm": 0.002052735071629286,
-      "learning_rate": 0.00019999828635259247,
-      "loss": 46.0,
-      "step": 24383
-    },
-    {
-      "epoch": 1.8643270829749412,
-      "grad_norm": 0.0010528728598728776,
-      "learning_rate": 0.00019999828621197147,
-      "loss": 46.0,
-      "step": 24384
-    },
-    {
-      "epoch": 1.864403539958331,
-      "grad_norm": 0.0011211491655558348,
-      "learning_rate": 0.00019999828607134473,
-      "loss": 46.0,
-      "step": 24385
-    },
-    {
-      "epoch": 1.8644799969417205,
-      "grad_norm": 0.0009848263580352068,
-      "learning_rate": 0.0001999982859307122,
-      "loss": 46.0,
-      "step": 24386
-    },
-    {
-      "epoch": 1.8645564539251103,
-      "grad_norm": 0.0033407120499759912,
-      "learning_rate": 0.0001999982857900739,
-      "loss": 46.0,
-      "step": 24387
-    },
-    {
-      "epoch": 1.8646329109085,
-      "grad_norm": 0.0009989996906369925,
-      "learning_rate": 0.00019999828564942984,
-      "loss": 46.0,
-      "step": 24388
-    },
-    {
-      "epoch": 1.8647093678918898,
-      "grad_norm": 0.0010883756913244724,
-      "learning_rate": 0.00019999828550877998,
-      "loss": 46.0,
-      "step": 24389
-    },
-    {
-      "epoch": 1.8647858248752796,
-      "grad_norm": 0.004948021844029427,
-      "learning_rate": 0.0001999982853681244,
-      "loss": 46.0,
-      "step": 24390
-    },
-    {
-      "epoch": 1.8648622818586693,
-      "grad_norm": 0.0044440315105021,
-      "learning_rate": 0.00019999828522746302,
-      "loss": 46.0,
-      "step": 24391
-    },
-    {
-      "epoch": 1.864938738842059,
-      "grad_norm": 0.0008561125723645091,
-      "learning_rate": 0.0001999982850867959,
-      "loss": 46.0,
-      "step": 24392
-    },
-    {
-      "epoch": 1.8650151958254488,
-      "grad_norm": 0.0005297197494655848,
-      "learning_rate": 0.00019999828494612296,
-      "loss": 46.0,
-      "step": 24393
-    },
-    {
-      "epoch": 1.8650916528088384,
-      "grad_norm": 0.0011938869720324874,
-      "learning_rate": 0.0001999982848054443,
-      "loss": 46.0,
-      "step": 24394
-    },
-    {
-      "epoch": 1.8651681097922281,
-      "grad_norm": 0.0031873309053480625,
-      "learning_rate": 0.00019999828466475981,
-      "loss": 46.0,
-      "step": 24395
-    },
-    {
-      "epoch": 1.865244566775618,
-      "grad_norm": 0.006199810653924942,
-      "learning_rate": 0.00019999828452406962,
-      "loss": 46.0,
-      "step": 24396
-    },
-    {
-      "epoch": 1.8653210237590074,
-      "grad_norm": 0.0030445787124335766,
-      "learning_rate": 0.00019999828438337363,
-      "loss": 46.0,
-      "step": 24397
-    },
-    {
-      "epoch": 1.8653974807423972,
-      "grad_norm": 0.0008772070286795497,
-      "learning_rate": 0.00019999828424267186,
-      "loss": 46.0,
-      "step": 24398
-    },
-    {
-      "epoch": 1.865473937725787,
-      "grad_norm": 0.0004670654016081244,
-      "learning_rate": 0.00019999828410196434,
-      "loss": 46.0,
-      "step": 24399
-    },
-    {
-      "epoch": 1.8655503947091767,
-      "grad_norm": 0.0005748254479840398,
-      "learning_rate": 0.00019999828396125103,
-      "loss": 46.0,
-      "step": 24400
-    },
-    {
-      "epoch": 1.8656268516925665,
-      "grad_norm": 0.0012402923312038183,
-      "learning_rate": 0.00019999828382053197,
-      "loss": 46.0,
-      "step": 24401
-    },
-    {
-      "epoch": 1.8657033086759562,
-      "grad_norm": 0.002196715911850333,
-      "learning_rate": 0.00019999828367980716,
-      "loss": 46.0,
-      "step": 24402
-    },
-    {
-      "epoch": 1.865779765659346,
-      "grad_norm": 0.0008591005462221801,
-      "learning_rate": 0.00019999828353907653,
-      "loss": 46.0,
-      "step": 24403
-    },
-    {
-      "epoch": 1.8658562226427358,
-      "grad_norm": 0.00043613393791019917,
-      "learning_rate": 0.00019999828339834017,
-      "loss": 46.0,
-      "step": 24404
-    },
-    {
-      "epoch": 1.8659326796261253,
-      "grad_norm": 0.0006122419727034867,
-      "learning_rate": 0.000199998283257598,
-      "loss": 46.0,
-      "step": 24405
-    },
-    {
-      "epoch": 1.866009136609515,
-      "grad_norm": 0.002290830947458744,
-      "learning_rate": 0.0001999982831168501,
-      "loss": 46.0,
-      "step": 24406
-    },
-    {
-      "epoch": 1.8660855935929048,
-      "grad_norm": 0.0012005933094769716,
-      "learning_rate": 0.00019999828297609642,
-      "loss": 46.0,
-      "step": 24407
-    },
-    {
-      "epoch": 1.8661620505762944,
-      "grad_norm": 0.0013345555635169148,
-      "learning_rate": 0.00019999828283533697,
-      "loss": 46.0,
-      "step": 24408
-    },
-    {
-      "epoch": 1.8662385075596841,
-      "grad_norm": 0.0005205682828091085,
-      "learning_rate": 0.00019999828269457175,
-      "loss": 46.0,
-      "step": 24409
-    },
-    {
-      "epoch": 1.8663149645430739,
-      "grad_norm": 0.0007695188978686929,
-      "learning_rate": 0.00019999828255380076,
-      "loss": 46.0,
-      "step": 24410
-    },
-    {
-      "epoch": 1.8663914215264636,
-      "grad_norm": 0.0007895085145719349,
-      "learning_rate": 0.000199998282413024,
-      "loss": 46.0,
-      "step": 24411
-    },
-    {
-      "epoch": 1.8664678785098534,
-      "grad_norm": 0.0009795501828193665,
-      "learning_rate": 0.00019999828227224145,
-      "loss": 46.0,
-      "step": 24412
-    },
-    {
-      "epoch": 1.8665443354932432,
-      "grad_norm": 0.0012096166610717773,
-      "learning_rate": 0.00019999828213145316,
-      "loss": 46.0,
-      "step": 24413
-    },
-    {
-      "epoch": 1.866620792476633,
-      "grad_norm": 0.001128451549448073,
-      "learning_rate": 0.0001999982819906591,
-      "loss": 46.0,
-      "step": 24414
-    },
-    {
-      "epoch": 1.8666972494600227,
-      "grad_norm": 0.00040535206790082157,
-      "learning_rate": 0.00019999828184985927,
-      "loss": 46.0,
-      "step": 24415
-    },
-    {
-      "epoch": 1.8667737064434122,
-      "grad_norm": 0.002225355478003621,
-      "learning_rate": 0.00019999828170905363,
-      "loss": 46.0,
-      "step": 24416
-    },
-    {
-      "epoch": 1.866850163426802,
-      "grad_norm": 0.0013886241940781474,
-      "learning_rate": 0.00019999828156824228,
-      "loss": 46.0,
-      "step": 24417
-    },
-    {
-      "epoch": 1.8669266204101918,
-      "grad_norm": 0.0010280228452757,
-      "learning_rate": 0.00019999828142742515,
-      "loss": 46.0,
-      "step": 24418
-    },
-    {
-      "epoch": 1.8670030773935813,
-      "grad_norm": 0.0005310793640092015,
-      "learning_rate": 0.0001999982812866022,
-      "loss": 46.0,
-      "step": 24419
-    },
-    {
-      "epoch": 1.867079534376971,
-      "grad_norm": 0.0015119437593966722,
-      "learning_rate": 0.00019999828114577352,
-      "loss": 46.0,
-      "step": 24420
-    },
-    {
-      "epoch": 1.8671559913603608,
-      "grad_norm": 0.004198964685201645,
-      "learning_rate": 0.00019999828100493908,
-      "loss": 46.0,
-      "step": 24421
-    },
-    {
-      "epoch": 1.8672324483437506,
-      "grad_norm": 0.0007355962297879159,
-      "learning_rate": 0.00019999828086409886,
-      "loss": 46.0,
-      "step": 24422
-    },
-    {
-      "epoch": 1.8673089053271403,
-      "grad_norm": 0.006976368371397257,
-      "learning_rate": 0.00019999828072325286,
-      "loss": 46.0,
-      "step": 24423
-    },
-    {
-      "epoch": 1.86738536231053,
-      "grad_norm": 0.0012016210239380598,
-      "learning_rate": 0.0001999982805824011,
-      "loss": 46.0,
-      "step": 24424
-    },
-    {
-      "epoch": 1.8674618192939199,
-      "grad_norm": 0.001741177518852055,
-      "learning_rate": 0.00019999828044154356,
-      "loss": 46.0,
-      "step": 24425
-    },
-    {
-      "epoch": 1.8675382762773096,
-      "grad_norm": 0.002907464513555169,
-      "learning_rate": 0.00019999828030068027,
-      "loss": 46.0,
-      "step": 24426
-    },
-    {
-      "epoch": 1.8676147332606992,
-      "grad_norm": 0.0009455534745939076,
-      "learning_rate": 0.00019999828015981118,
-      "loss": 46.0,
-      "step": 24427
-    },
-    {
-      "epoch": 1.867691190244089,
-      "grad_norm": 0.0006432398222386837,
-      "learning_rate": 0.00019999828001893635,
-      "loss": 46.0,
-      "step": 24428
-    },
-    {
-      "epoch": 1.8677676472274787,
-      "grad_norm": 0.006211003288626671,
-      "learning_rate": 0.00019999827987805574,
-      "loss": 46.0,
-      "step": 24429
-    },
-    {
-      "epoch": 1.8678441042108682,
-      "grad_norm": 0.0007333803805522621,
-      "learning_rate": 0.0001999982797371694,
-      "loss": 46.0,
-      "step": 24430
-    },
-    {
-      "epoch": 1.867920561194258,
-      "grad_norm": 0.002615738194435835,
-      "learning_rate": 0.0001999982795962772,
-      "loss": 46.0,
-      "step": 24431
-    },
-    {
-      "epoch": 1.8679970181776477,
-      "grad_norm": 0.0016500019701197743,
-      "learning_rate": 0.0001999982794553793,
-      "loss": 46.0,
-      "step": 24432
-    },
-    {
-      "epoch": 1.8680734751610375,
-      "grad_norm": 0.00065139977959916,
-      "learning_rate": 0.0001999982793144756,
-      "loss": 46.0,
-      "step": 24433
-    },
-    {
-      "epoch": 1.8681499321444273,
-      "grad_norm": 0.0018682519439607859,
-      "learning_rate": 0.00019999827917356616,
-      "loss": 46.0,
-      "step": 24434
-    },
-    {
-      "epoch": 1.868226389127817,
-      "grad_norm": 0.0017565478337928653,
-      "learning_rate": 0.00019999827903265094,
-      "loss": 46.0,
-      "step": 24435
-    },
-    {
-      "epoch": 1.8683028461112068,
-      "grad_norm": 0.00086885632481426,
-      "learning_rate": 0.00019999827889172992,
-      "loss": 46.0,
-      "step": 24436
-    },
-    {
-      "epoch": 1.8683793030945965,
-      "grad_norm": 0.0026381935458630323,
-      "learning_rate": 0.00019999827875080318,
-      "loss": 46.0,
-      "step": 24437
-    },
-    {
-      "epoch": 1.868455760077986,
-      "grad_norm": 0.001720384112559259,
-      "learning_rate": 0.00019999827860987064,
-      "loss": 46.0,
-      "step": 24438
-    },
-    {
-      "epoch": 1.8685322170613758,
-      "grad_norm": 0.0009916112758219242,
-      "learning_rate": 0.00019999827846893235,
-      "loss": 46.0,
-      "step": 24439
-    },
-    {
-      "epoch": 1.8686086740447656,
-      "grad_norm": 0.001400334993377328,
-      "learning_rate": 0.00019999827832798827,
-      "loss": 46.0,
-      "step": 24440
-    },
-    {
-      "epoch": 1.8686851310281551,
-      "grad_norm": 0.00034276899532414973,
-      "learning_rate": 0.00019999827818703843,
-      "loss": 46.0,
-      "step": 24441
-    },
-    {
-      "epoch": 1.868761588011545,
-      "grad_norm": 0.0011658057337626815,
-      "learning_rate": 0.00019999827804608283,
-      "loss": 46.0,
-      "step": 24442
-    },
-    {
-      "epoch": 1.8688380449949347,
-      "grad_norm": 0.010067655704915524,
-      "learning_rate": 0.00019999827790512145,
-      "loss": 46.0,
-      "step": 24443
-    },
-    {
-      "epoch": 1.8689145019783244,
-      "grad_norm": 0.0008100769482553005,
-      "learning_rate": 0.0001999982777641543,
-      "loss": 46.0,
-      "step": 24444
-    },
-    {
-      "epoch": 1.8689909589617142,
-      "grad_norm": 0.0008821483352221549,
-      "learning_rate": 0.00019999827762318137,
-      "loss": 46.0,
-      "step": 24445
-    },
-    {
-      "epoch": 1.869067415945104,
-      "grad_norm": 0.0009331233450211585,
-      "learning_rate": 0.00019999827748220267,
-      "loss": 46.0,
-      "step": 24446
-    },
-    {
-      "epoch": 1.8691438729284937,
-      "grad_norm": 0.0002576448896434158,
-      "learning_rate": 0.00019999827734121822,
-      "loss": 46.0,
-      "step": 24447
-    },
-    {
-      "epoch": 1.8692203299118835,
-      "grad_norm": 0.00041902950033545494,
-      "learning_rate": 0.000199998277200228,
-      "loss": 46.0,
-      "step": 24448
-    },
-    {
-      "epoch": 1.869296786895273,
-      "grad_norm": 0.0009253700845874846,
-      "learning_rate": 0.000199998277059232,
-      "loss": 46.0,
-      "step": 24449
-    },
-    {
-      "epoch": 1.8693732438786628,
-      "grad_norm": 0.0013820132007822394,
-      "learning_rate": 0.00019999827691823024,
-      "loss": 46.0,
-      "step": 24450
-    },
-    {
-      "epoch": 1.8694497008620525,
-      "grad_norm": 0.006285613402724266,
-      "learning_rate": 0.0001999982767772227,
-      "loss": 46.0,
-      "step": 24451
-    },
-    {
-      "epoch": 1.869526157845442,
-      "grad_norm": 0.0009755613282322884,
-      "learning_rate": 0.00019999827663620941,
-      "loss": 46.0,
-      "step": 24452
-    },
-    {
-      "epoch": 1.8696026148288318,
-      "grad_norm": 0.0009945965139195323,
-      "learning_rate": 0.00019999827649519033,
-      "loss": 46.0,
-      "step": 24453
-    },
-    {
-      "epoch": 1.8696790718122216,
-      "grad_norm": 0.005110160913318396,
-      "learning_rate": 0.00019999827635416547,
-      "loss": 46.0,
-      "step": 24454
-    },
-    {
-      "epoch": 1.8697555287956114,
-      "grad_norm": 0.0016918452456593513,
-      "learning_rate": 0.00019999827621313486,
-      "loss": 46.0,
-      "step": 24455
-    },
-    {
-      "epoch": 1.8698319857790011,
-      "grad_norm": 0.0007825566572137177,
-      "learning_rate": 0.00019999827607209848,
-      "loss": 46.0,
-      "step": 24456
-    },
-    {
-      "epoch": 1.8699084427623909,
-      "grad_norm": 0.0004155226342845708,
-      "learning_rate": 0.00019999827593105636,
-      "loss": 46.0,
-      "step": 24457
-    },
-    {
-      "epoch": 1.8699848997457806,
-      "grad_norm": 0.0011150981299579144,
-      "learning_rate": 0.00019999827579000843,
-      "loss": 46.0,
-      "step": 24458
-    },
-    {
-      "epoch": 1.8700613567291704,
-      "grad_norm": 0.004238279070705175,
-      "learning_rate": 0.00019999827564895473,
-      "loss": 46.0,
-      "step": 24459
-    },
-    {
-      "epoch": 1.87013781371256,
-      "grad_norm": 0.0009951271349564195,
-      "learning_rate": 0.00019999827550789528,
-      "loss": 46.0,
-      "step": 24460
-    },
-    {
-      "epoch": 1.8702142706959497,
-      "grad_norm": 0.0006854430539533496,
-      "learning_rate": 0.00019999827536683004,
-      "loss": 46.0,
-      "step": 24461
-    },
-    {
-      "epoch": 1.8702907276793395,
-      "grad_norm": 0.009353235363960266,
-      "learning_rate": 0.00019999827522575905,
-      "loss": 46.0,
-      "step": 24462
-    },
-    {
-      "epoch": 1.870367184662729,
-      "grad_norm": 0.002244858071208,
-      "learning_rate": 0.00019999827508468228,
-      "loss": 46.0,
-      "step": 24463
-    },
-    {
-      "epoch": 1.8704436416461188,
-      "grad_norm": 0.0008229893282987177,
-      "learning_rate": 0.00019999827494359974,
-      "loss": 46.0,
-      "step": 24464
-    },
-    {
-      "epoch": 1.8705200986295085,
-      "grad_norm": 0.002377427415922284,
-      "learning_rate": 0.00019999827480251145,
-      "loss": 46.0,
-      "step": 24465
-    },
-    {
-      "epoch": 1.8705965556128983,
-      "grad_norm": 0.005021687597036362,
-      "learning_rate": 0.00019999827466141737,
-      "loss": 46.0,
-      "step": 24466
-    },
-    {
-      "epoch": 1.870673012596288,
-      "grad_norm": 0.0028901048935949802,
-      "learning_rate": 0.0001999982745203175,
-      "loss": 46.0,
-      "step": 24467
-    },
-    {
-      "epoch": 1.8707494695796778,
-      "grad_norm": 0.0014143723528832197,
-      "learning_rate": 0.0001999982743792119,
-      "loss": 46.0,
-      "step": 24468
-    },
-    {
-      "epoch": 1.8708259265630676,
-      "grad_norm": 0.005924125202000141,
-      "learning_rate": 0.00019999827423810052,
-      "loss": 46.0,
-      "step": 24469
-    },
-    {
-      "epoch": 1.8709023835464573,
-      "grad_norm": 0.0008604479371570051,
-      "learning_rate": 0.00019999827409698337,
-      "loss": 46.0,
-      "step": 24470
-    },
-    {
-      "epoch": 1.8709788405298469,
-      "grad_norm": 0.002774209249764681,
-      "learning_rate": 0.00019999827395586044,
-      "loss": 46.0,
-      "step": 24471
-    },
-    {
-      "epoch": 1.8710552975132366,
-      "grad_norm": 0.0010927956318482757,
-      "learning_rate": 0.00019999827381473177,
-      "loss": 46.0,
-      "step": 24472
-    },
-    {
-      "epoch": 1.8711317544966264,
-      "grad_norm": 0.006643348373472691,
-      "learning_rate": 0.0001999982736735973,
-      "loss": 46.0,
-      "step": 24473
-    },
-    {
-      "epoch": 1.871208211480016,
-      "grad_norm": 0.0009108408121392131,
-      "learning_rate": 0.00019999827353245708,
-      "loss": 46.0,
-      "step": 24474
-    },
-    {
-      "epoch": 1.8712846684634057,
-      "grad_norm": 0.0005059065879322588,
-      "learning_rate": 0.00019999827339131106,
-      "loss": 46.0,
-      "step": 24475
-    },
-    {
-      "epoch": 1.8713611254467954,
-      "grad_norm": 0.0016773603856563568,
-      "learning_rate": 0.0001999982732501593,
-      "loss": 46.0,
-      "step": 24476
-    },
-    {
-      "epoch": 1.8714375824301852,
-      "grad_norm": 0.0010655325604602695,
-      "learning_rate": 0.00019999827310900175,
-      "loss": 46.0,
-      "step": 24477
-    },
-    {
-      "epoch": 1.871514039413575,
-      "grad_norm": 0.002016855403780937,
-      "learning_rate": 0.00019999827296783847,
-      "loss": 46.0,
-      "step": 24478
-    },
-    {
-      "epoch": 1.8715904963969647,
-      "grad_norm": 0.001541712204925716,
-      "learning_rate": 0.00019999827282666938,
-      "loss": 46.0,
-      "step": 24479
-    },
-    {
-      "epoch": 1.8716669533803545,
-      "grad_norm": 0.0042297979816794395,
-      "learning_rate": 0.00019999827268549452,
-      "loss": 46.0,
-      "step": 24480
-    },
-    {
-      "epoch": 1.8717434103637443,
-      "grad_norm": 0.0007102172821760178,
-      "learning_rate": 0.00019999827254431392,
-      "loss": 46.0,
-      "step": 24481
-    },
-    {
-      "epoch": 1.8718198673471338,
-      "grad_norm": 0.0009684275719337165,
-      "learning_rate": 0.0001999982724031275,
-      "loss": 46.0,
-      "step": 24482
-    },
-    {
-      "epoch": 1.8718963243305236,
-      "grad_norm": 0.0009876557160168886,
-      "learning_rate": 0.0001999982722619354,
-      "loss": 46.0,
-      "step": 24483
-    },
-    {
-      "epoch": 1.8719727813139133,
-      "grad_norm": 0.0003435726393945515,
-      "learning_rate": 0.00019999827212073746,
-      "loss": 46.0,
-      "step": 24484
-    },
-    {
-      "epoch": 1.8720492382973029,
-      "grad_norm": 0.0016244461294263601,
-      "learning_rate": 0.0001999982719795338,
-      "loss": 46.0,
-      "step": 24485
-    },
-    {
-      "epoch": 1.8721256952806926,
-      "grad_norm": 0.0007226461893878877,
-      "learning_rate": 0.0001999982718383243,
-      "loss": 46.0,
-      "step": 24486
-    },
-    {
-      "epoch": 1.8722021522640824,
-      "grad_norm": 0.0019329085480421782,
-      "learning_rate": 0.00019999827169710907,
-      "loss": 46.0,
-      "step": 24487
-    },
-    {
-      "epoch": 1.8722786092474721,
-      "grad_norm": 0.0007299327407963574,
-      "learning_rate": 0.00019999827155588808,
-      "loss": 46.0,
-      "step": 24488
-    },
-    {
-      "epoch": 1.872355066230862,
-      "grad_norm": 0.00095442368183285,
-      "learning_rate": 0.00019999827141466129,
-      "loss": 46.0,
-      "step": 24489
-    },
-    {
-      "epoch": 1.8724315232142517,
-      "grad_norm": 0.001932155922986567,
-      "learning_rate": 0.00019999827127342877,
-      "loss": 46.0,
-      "step": 24490
-    },
-    {
-      "epoch": 1.8725079801976414,
-      "grad_norm": 0.0011166568147018552,
-      "learning_rate": 0.00019999827113219046,
-      "loss": 46.0,
-      "step": 24491
-    },
-    {
-      "epoch": 1.8725844371810312,
-      "grad_norm": 0.0005555366515181959,
-      "learning_rate": 0.0001999982709909464,
-      "loss": 46.0,
-      "step": 24492
-    },
-    {
-      "epoch": 1.8726608941644207,
-      "grad_norm": 0.0012485034530982375,
-      "learning_rate": 0.00019999827084969654,
-      "loss": 46.0,
-      "step": 24493
-    },
-    {
-      "epoch": 1.8727373511478105,
-      "grad_norm": 0.0007916319882497191,
-      "learning_rate": 0.0001999982707084409,
-      "loss": 46.0,
-      "step": 24494
-    },
-    {
-      "epoch": 1.8728138081312,
-      "grad_norm": 0.000595214543864131,
-      "learning_rate": 0.00019999827056717953,
-      "loss": 46.0,
-      "step": 24495
-    },
-    {
-      "epoch": 1.8728902651145898,
-      "grad_norm": 0.0010901688365265727,
-      "learning_rate": 0.00019999827042591238,
-      "loss": 46.0,
-      "step": 24496
-    },
-    {
-      "epoch": 1.8729667220979795,
-      "grad_norm": 0.0009604181395843625,
-      "learning_rate": 0.00019999827028463946,
-      "loss": 46.0,
-      "step": 24497
-    },
-    {
-      "epoch": 1.8730431790813693,
-      "grad_norm": 0.004549084696918726,
-      "learning_rate": 0.00019999827014336076,
-      "loss": 46.0,
-      "step": 24498
-    },
-    {
-      "epoch": 1.873119636064759,
-      "grad_norm": 0.0008366979891434312,
-      "learning_rate": 0.00019999827000207628,
-      "loss": 46.0,
-      "step": 24499
-    },
-    {
-      "epoch": 1.8731960930481488,
-      "grad_norm": 0.00029412517324090004,
-      "learning_rate": 0.00019999826986078604,
-      "loss": 46.0,
-      "step": 24500
-    },
-    {
-      "epoch": 1.8732725500315386,
-      "grad_norm": 0.000268264819169417,
-      "learning_rate": 0.00019999826971949005,
-      "loss": 46.0,
-      "step": 24501
-    },
-    {
-      "epoch": 1.8733490070149283,
-      "grad_norm": 0.0014895758358761668,
-      "learning_rate": 0.00019999826957818828,
-      "loss": 46.0,
-      "step": 24502
-    },
-    {
-      "epoch": 1.873425463998318,
-      "grad_norm": 0.00035663298331201077,
-      "learning_rate": 0.00019999826943688074,
-      "loss": 46.0,
-      "step": 24503
-    },
-    {
-      "epoch": 1.8735019209817076,
-      "grad_norm": 0.0011001632083207369,
-      "learning_rate": 0.00019999826929556743,
-      "loss": 46.0,
-      "step": 24504
-    },
-    {
-      "epoch": 1.8735783779650974,
-      "grad_norm": 0.0015001454157754779,
-      "learning_rate": 0.00019999826915424835,
-      "loss": 46.0,
-      "step": 24505
-    },
-    {
-      "epoch": 1.873654834948487,
-      "grad_norm": 0.0006530695827677846,
-      "learning_rate": 0.0001999982690129235,
-      "loss": 46.0,
-      "step": 24506
-    },
-    {
-      "epoch": 1.8737312919318767,
-      "grad_norm": 0.00045370415318757296,
-      "learning_rate": 0.00019999826887159288,
-      "loss": 46.0,
-      "step": 24507
-    },
-    {
-      "epoch": 1.8738077489152665,
-      "grad_norm": 0.00073546904604882,
-      "learning_rate": 0.00019999826873025648,
-      "loss": 46.0,
-      "step": 24508
-    },
-    {
-      "epoch": 1.8738842058986562,
-      "grad_norm": 0.00421450100839138,
-      "learning_rate": 0.00019999826858891435,
-      "loss": 46.0,
-      "step": 24509
-    },
-    {
-      "epoch": 1.873960662882046,
-      "grad_norm": 0.0016301005380228162,
-      "learning_rate": 0.0001999982684475664,
-      "loss": 46.0,
-      "step": 24510
-    },
-    {
-      "epoch": 1.8740371198654358,
-      "grad_norm": 0.0058384365402162075,
-      "learning_rate": 0.0001999982683062127,
-      "loss": 46.0,
-      "step": 24511
-    },
-    {
-      "epoch": 1.8741135768488255,
-      "grad_norm": 0.0005423123948276043,
-      "learning_rate": 0.00019999826816485326,
-      "loss": 46.0,
-      "step": 24512
-    },
-    {
-      "epoch": 1.8741900338322153,
-      "grad_norm": 0.0012862710282206535,
-      "learning_rate": 0.000199998268023488,
-      "loss": 46.0,
-      "step": 24513
-    },
-    {
-      "epoch": 1.874266490815605,
-      "grad_norm": 0.000843945425003767,
-      "learning_rate": 0.00019999826788211702,
-      "loss": 46.0,
-      "step": 24514
-    },
-    {
-      "epoch": 1.8743429477989946,
-      "grad_norm": 0.005360694136470556,
-      "learning_rate": 0.00019999826774074023,
-      "loss": 46.0,
-      "step": 24515
-    },
-    {
-      "epoch": 1.8744194047823843,
-      "grad_norm": 0.0008958461112342775,
-      "learning_rate": 0.0001999982675993577,
-      "loss": 46.0,
-      "step": 24516
-    },
-    {
-      "epoch": 1.8744958617657739,
-      "grad_norm": 0.0011793512385338545,
-      "learning_rate": 0.00019999826745796938,
-      "loss": 46.0,
-      "step": 24517
-    },
-    {
-      "epoch": 1.8745723187491636,
-      "grad_norm": 0.0011013824259862304,
-      "learning_rate": 0.00019999826731657532,
-      "loss": 46.0,
-      "step": 24518
-    },
-    {
-      "epoch": 1.8746487757325534,
-      "grad_norm": 0.0009038368589244783,
-      "learning_rate": 0.00019999826717517543,
-      "loss": 46.0,
-      "step": 24519
-    },
-    {
-      "epoch": 1.8747252327159432,
-      "grad_norm": 0.012815146706998348,
-      "learning_rate": 0.00019999826703376983,
-      "loss": 46.0,
-      "step": 24520
-    },
-    {
-      "epoch": 1.874801689699333,
-      "grad_norm": 0.0026535573415458202,
-      "learning_rate": 0.00019999826689235843,
-      "loss": 46.0,
-      "step": 24521
-    },
-    {
-      "epoch": 1.8748781466827227,
-      "grad_norm": 0.0013110985746607184,
-      "learning_rate": 0.0001999982667509413,
-      "loss": 46.0,
-      "step": 24522
-    },
-    {
-      "epoch": 1.8749546036661124,
-      "grad_norm": 0.0009825148154050112,
-      "learning_rate": 0.00019999826660951838,
-      "loss": 46.0,
-      "step": 24523
-    },
-    {
-      "epoch": 1.8750310606495022,
-      "grad_norm": 0.002428684150800109,
-      "learning_rate": 0.00019999826646808965,
-      "loss": 46.0,
-      "step": 24524
-    },
-    {
-      "epoch": 1.8751075176328917,
-      "grad_norm": 0.0005493476637639105,
-      "learning_rate": 0.00019999826632665518,
-      "loss": 46.0,
-      "step": 24525
-    },
-    {
-      "epoch": 1.8751839746162815,
-      "grad_norm": 0.0004378874145913869,
-      "learning_rate": 0.00019999826618521496,
-      "loss": 46.0,
-      "step": 24526
-    },
-    {
-      "epoch": 1.8752604315996713,
-      "grad_norm": 0.0024309037253260612,
-      "learning_rate": 0.00019999826604376894,
-      "loss": 46.0,
-      "step": 24527
-    },
-    {
-      "epoch": 1.8753368885830608,
-      "grad_norm": 0.009604115039110184,
-      "learning_rate": 0.00019999826590231718,
-      "loss": 46.0,
-      "step": 24528
-    },
-    {
-      "epoch": 1.8754133455664506,
-      "grad_norm": 0.0010836664587259293,
-      "learning_rate": 0.0001999982657608596,
-      "loss": 46.0,
-      "step": 24529
-    },
-    {
-      "epoch": 1.8754898025498403,
-      "grad_norm": 0.0012843436561524868,
-      "learning_rate": 0.00019999826561939633,
-      "loss": 46.0,
-      "step": 24530
-    },
-    {
-      "epoch": 1.87556625953323,
-      "grad_norm": 0.0002768023405224085,
-      "learning_rate": 0.00019999826547792724,
-      "loss": 46.0,
-      "step": 24531
-    },
-    {
-      "epoch": 1.8756427165166198,
-      "grad_norm": 0.0005219127633608878,
-      "learning_rate": 0.00019999826533645239,
-      "loss": 46.0,
-      "step": 24532
-    },
-    {
-      "epoch": 1.8757191735000096,
-      "grad_norm": 0.00045046856394037604,
-      "learning_rate": 0.00019999826519497175,
-      "loss": 46.0,
-      "step": 24533
-    },
-    {
-      "epoch": 1.8757956304833994,
-      "grad_norm": 0.002908071968704462,
-      "learning_rate": 0.00019999826505348538,
-      "loss": 46.0,
-      "step": 24534
-    },
-    {
-      "epoch": 1.8758720874667891,
-      "grad_norm": 0.0007360133458860219,
-      "learning_rate": 0.0001999982649119932,
-      "loss": 46.0,
-      "step": 24535
-    },
-    {
-      "epoch": 1.8759485444501787,
-      "grad_norm": 0.0009785944130271673,
-      "learning_rate": 0.00019999826477049527,
-      "loss": 46.0,
-      "step": 24536
-    },
-    {
-      "epoch": 1.8760250014335684,
-      "grad_norm": 0.0008189226500689983,
-      "learning_rate": 0.00019999826462899158,
-      "loss": 46.0,
-      "step": 24537
-    },
-    {
-      "epoch": 1.8761014584169582,
-      "grad_norm": 0.0006891925004310906,
-      "learning_rate": 0.0001999982644874821,
-      "loss": 46.0,
-      "step": 24538
-    },
-    {
-      "epoch": 1.8761779154003477,
-      "grad_norm": 0.0006073514232411981,
-      "learning_rate": 0.00019999826434596686,
-      "loss": 46.0,
-      "step": 24539
-    },
-    {
-      "epoch": 1.8762543723837375,
-      "grad_norm": 0.0030272388830780983,
-      "learning_rate": 0.00019999826420444587,
-      "loss": 46.0,
-      "step": 24540
-    },
-    {
-      "epoch": 1.8763308293671273,
-      "grad_norm": 0.002854966791346669,
-      "learning_rate": 0.00019999826406291908,
-      "loss": 46.0,
-      "step": 24541
-    },
-    {
-      "epoch": 1.876407286350517,
-      "grad_norm": 0.0007178594823926687,
-      "learning_rate": 0.00019999826392138654,
-      "loss": 46.0,
-      "step": 24542
-    },
-    {
-      "epoch": 1.8764837433339068,
-      "grad_norm": 0.0007962700328789651,
-      "learning_rate": 0.00019999826377984823,
-      "loss": 46.0,
-      "step": 24543
-    },
-    {
-      "epoch": 1.8765602003172965,
-      "grad_norm": 0.0013936894247308373,
-      "learning_rate": 0.00019999826363830415,
-      "loss": 46.0,
-      "step": 24544
-    },
-    {
-      "epoch": 1.8766366573006863,
-      "grad_norm": 0.0007003230857662857,
-      "learning_rate": 0.0001999982634967543,
-      "loss": 46.0,
-      "step": 24545
-    },
-    {
-      "epoch": 1.876713114284076,
-      "grad_norm": 0.000480629998492077,
-      "learning_rate": 0.00019999826335519866,
-      "loss": 46.0,
-      "step": 24546
-    },
-    {
-      "epoch": 1.8767895712674656,
-      "grad_norm": 0.0006380486884154379,
-      "learning_rate": 0.00019999826321363728,
-      "loss": 46.0,
-      "step": 24547
-    },
-    {
-      "epoch": 1.8768660282508554,
-      "grad_norm": 0.000577355211135,
-      "learning_rate": 0.00019999826307207013,
-      "loss": 46.0,
-      "step": 24548
-    },
-    {
-      "epoch": 1.8769424852342451,
-      "grad_norm": 0.005160320084542036,
-      "learning_rate": 0.0001999982629304972,
-      "loss": 46.0,
-      "step": 24549
-    },
-    {
-      "epoch": 1.8770189422176347,
-      "grad_norm": 0.0011860494269058108,
-      "learning_rate": 0.00019999826278891848,
-      "loss": 46.0,
-      "step": 24550
-    },
-    {
-      "epoch": 1.8770953992010244,
-      "grad_norm": 0.0023892447352409363,
-      "learning_rate": 0.000199998262647334,
-      "loss": 46.0,
-      "step": 24551
-    },
-    {
-      "epoch": 1.8771718561844142,
-      "grad_norm": 0.0014236947754397988,
-      "learning_rate": 0.00019999826250574376,
-      "loss": 46.0,
-      "step": 24552
-    },
-    {
-      "epoch": 1.877248313167804,
-      "grad_norm": 0.0006510316743515432,
-      "learning_rate": 0.00019999826236414775,
-      "loss": 46.0,
-      "step": 24553
-    },
-    {
-      "epoch": 1.8773247701511937,
-      "grad_norm": 0.0007313325186260045,
-      "learning_rate": 0.00019999826222254598,
-      "loss": 46.0,
-      "step": 24554
-    },
-    {
-      "epoch": 1.8774012271345835,
-      "grad_norm": 0.0008234083070419729,
-      "learning_rate": 0.00019999826208093845,
-      "loss": 46.0,
-      "step": 24555
-    },
-    {
-      "epoch": 1.8774776841179732,
-      "grad_norm": 0.001224440522491932,
-      "learning_rate": 0.0001999982619393251,
-      "loss": 46.0,
-      "step": 24556
-    },
-    {
-      "epoch": 1.877554141101363,
-      "grad_norm": 0.0005092094652354717,
-      "learning_rate": 0.00019999826179770605,
-      "loss": 46.0,
-      "step": 24557
-    },
-    {
-      "epoch": 1.8776305980847525,
-      "grad_norm": 0.001224069157615304,
-      "learning_rate": 0.00019999826165608117,
-      "loss": 46.0,
-      "step": 24558
-    },
-    {
-      "epoch": 1.8777070550681423,
-      "grad_norm": 0.0006747928564436734,
-      "learning_rate": 0.00019999826151445056,
-      "loss": 46.0,
-      "step": 24559
-    },
-    {
-      "epoch": 1.877783512051532,
-      "grad_norm": 0.0015635519521310925,
-      "learning_rate": 0.00019999826137281413,
-      "loss": 46.0,
-      "step": 24560
-    },
-    {
-      "epoch": 1.8778599690349216,
-      "grad_norm": 0.001057027606293559,
-      "learning_rate": 0.00019999826123117198,
-      "loss": 46.0,
-      "step": 24561
-    },
-    {
-      "epoch": 1.8779364260183113,
-      "grad_norm": 0.0012541123433038592,
-      "learning_rate": 0.00019999826108952406,
-      "loss": 46.0,
-      "step": 24562
-    },
-    {
-      "epoch": 1.878012883001701,
-      "grad_norm": 0.0019493142608553171,
-      "learning_rate": 0.00019999826094787039,
-      "loss": 46.0,
-      "step": 24563
-    },
-    {
-      "epoch": 1.8780893399850909,
-      "grad_norm": 0.010323435999453068,
-      "learning_rate": 0.0001999982608062109,
-      "loss": 46.0,
-      "step": 24564
-    },
-    {
-      "epoch": 1.8781657969684806,
-      "grad_norm": 0.005729845724999905,
-      "learning_rate": 0.00019999826066454564,
-      "loss": 46.0,
-      "step": 24565
-    },
-    {
-      "epoch": 1.8782422539518704,
-      "grad_norm": 0.0028460088651627302,
-      "learning_rate": 0.00019999826052287465,
-      "loss": 46.0,
-      "step": 24566
-    },
-    {
-      "epoch": 1.8783187109352601,
-      "grad_norm": 0.0009143223869614303,
-      "learning_rate": 0.00019999826038119784,
-      "loss": 46.0,
-      "step": 24567
-    },
-    {
-      "epoch": 1.87839516791865,
-      "grad_norm": 0.0013302953448146582,
-      "learning_rate": 0.0001999982602395153,
-      "loss": 46.0,
-      "step": 24568
-    },
-    {
-      "epoch": 1.8784716249020394,
-      "grad_norm": 0.00485475966706872,
-      "learning_rate": 0.00019999826009782702,
-      "loss": 46.0,
-      "step": 24569
-    },
-    {
-      "epoch": 1.8785480818854292,
-      "grad_norm": 0.001227396191097796,
-      "learning_rate": 0.0001999982599561329,
-      "loss": 46.0,
-      "step": 24570
-    },
-    {
-      "epoch": 1.878624538868819,
-      "grad_norm": 0.0036998779978603125,
-      "learning_rate": 0.00019999825981443302,
-      "loss": 46.0,
-      "step": 24571
-    },
-    {
-      "epoch": 1.8787009958522085,
-      "grad_norm": 0.0016569297295063734,
-      "learning_rate": 0.00019999825967272742,
-      "loss": 46.0,
-      "step": 24572
-    },
-    {
-      "epoch": 1.8787774528355983,
-      "grad_norm": 0.0006455841939896345,
-      "learning_rate": 0.00019999825953101601,
-      "loss": 46.0,
-      "step": 24573
-    },
-    {
-      "epoch": 1.878853909818988,
-      "grad_norm": 0.0013256885576993227,
-      "learning_rate": 0.00019999825938929886,
-      "loss": 46.0,
-      "step": 24574
-    },
-    {
-      "epoch": 1.8789303668023778,
-      "grad_norm": 0.0014563980512320995,
-      "learning_rate": 0.00019999825924757594,
-      "loss": 46.0,
-      "step": 24575
-    },
-    {
-      "epoch": 1.8790068237857676,
-      "grad_norm": 0.0011467279400676489,
-      "learning_rate": 0.00019999825910584722,
-      "loss": 46.0,
-      "step": 24576
-    },
-    {
-      "epoch": 1.8790832807691573,
-      "grad_norm": 0.0006850773352198303,
-      "learning_rate": 0.00019999825896411275,
-      "loss": 46.0,
-      "step": 24577
-    },
-    {
-      "epoch": 1.879159737752547,
-      "grad_norm": 0.0009535907302051783,
-      "learning_rate": 0.0001999982588223725,
-      "loss": 46.0,
-      "step": 24578
-    },
-    {
-      "epoch": 1.8792361947359368,
-      "grad_norm": 0.0009148971876129508,
-      "learning_rate": 0.00019999825868062649,
-      "loss": 46.0,
-      "step": 24579
-    },
-    {
-      "epoch": 1.8793126517193264,
-      "grad_norm": 0.001068438752554357,
-      "learning_rate": 0.0001999982585388747,
-      "loss": 46.0,
-      "step": 24580
-    },
-    {
-      "epoch": 1.8793891087027161,
-      "grad_norm": 0.0009263596730306745,
-      "learning_rate": 0.00019999825839711716,
-      "loss": 46.0,
-      "step": 24581
-    },
-    {
-      "epoch": 1.879465565686106,
-      "grad_norm": 0.0009450057987123728,
-      "learning_rate": 0.00019999825825535385,
-      "loss": 46.0,
-      "step": 24582
-    },
-    {
-      "epoch": 1.8795420226694954,
-      "grad_norm": 0.001825155457481742,
-      "learning_rate": 0.00019999825811358474,
-      "loss": 46.0,
-      "step": 24583
-    },
-    {
-      "epoch": 1.8796184796528852,
-      "grad_norm": 0.0005288769607432187,
-      "learning_rate": 0.00019999825797180988,
-      "loss": 46.0,
-      "step": 24584
-    },
-    {
-      "epoch": 1.879694936636275,
-      "grad_norm": 0.0012233048910275102,
-      "learning_rate": 0.00019999825783002928,
-      "loss": 46.0,
-      "step": 24585
-    },
-    {
-      "epoch": 1.8797713936196647,
-      "grad_norm": 0.0010230995249003172,
-      "learning_rate": 0.00019999825768824287,
-      "loss": 46.0,
-      "step": 24586
-    },
-    {
-      "epoch": 1.8798478506030545,
-      "grad_norm": 0.0020094281062483788,
-      "learning_rate": 0.0001999982575464507,
-      "loss": 46.0,
-      "step": 24587
-    },
-    {
-      "epoch": 1.8799243075864442,
-      "grad_norm": 0.0038012228906154633,
-      "learning_rate": 0.00019999825740465277,
-      "loss": 46.0,
-      "step": 24588
-    },
-    {
-      "epoch": 1.880000764569834,
-      "grad_norm": 0.0011561153223738074,
-      "learning_rate": 0.00019999825726284908,
-      "loss": 46.0,
-      "step": 24589
-    },
-    {
-      "epoch": 1.8800772215532238,
-      "grad_norm": 0.0010910233249887824,
-      "learning_rate": 0.00019999825712103958,
-      "loss": 46.0,
-      "step": 24590
-    },
-    {
-      "epoch": 1.8801536785366133,
-      "grad_norm": 0.0016625941498205066,
-      "learning_rate": 0.00019999825697922434,
-      "loss": 46.0,
-      "step": 24591
-    },
-    {
-      "epoch": 1.880230135520003,
-      "grad_norm": 0.0028500158805400133,
-      "learning_rate": 0.00019999825683740335,
-      "loss": 46.0,
-      "step": 24592
-    },
-    {
-      "epoch": 1.8803065925033928,
-      "grad_norm": 0.0029136845842003822,
-      "learning_rate": 0.00019999825669557656,
-      "loss": 46.0,
-      "step": 24593
-    },
-    {
-      "epoch": 1.8803830494867824,
-      "grad_norm": 0.0015776555519551039,
-      "learning_rate": 0.00019999825655374397,
-      "loss": 46.0,
-      "step": 24594
-    },
-    {
-      "epoch": 1.8804595064701721,
-      "grad_norm": 0.0006938427104614675,
-      "learning_rate": 0.00019999825641190566,
-      "loss": 46.0,
-      "step": 24595
-    },
-    {
-      "epoch": 1.8805359634535619,
-      "grad_norm": 0.0027691680006682873,
-      "learning_rate": 0.00019999825627006157,
-      "loss": 46.0,
-      "step": 24596
-    },
-    {
-      "epoch": 1.8806124204369516,
-      "grad_norm": 0.0011147845070809126,
-      "learning_rate": 0.0001999982561282117,
-      "loss": 46.0,
-      "step": 24597
-    },
-    {
-      "epoch": 1.8806888774203414,
-      "grad_norm": 0.001035679248161614,
-      "learning_rate": 0.0001999982559863561,
-      "loss": 46.0,
-      "step": 24598
-    },
-    {
-      "epoch": 1.8807653344037312,
-      "grad_norm": 0.0020652797538787127,
-      "learning_rate": 0.0001999982558444947,
-      "loss": 46.0,
-      "step": 24599
-    },
-    {
-      "epoch": 1.880841791387121,
-      "grad_norm": 0.0007283765589818358,
-      "learning_rate": 0.0001999982557026275,
-      "loss": 46.0,
-      "step": 24600
-    },
-    {
-      "epoch": 1.8809182483705107,
-      "grad_norm": 0.0011391901643946767,
-      "learning_rate": 0.0001999982555607546,
-      "loss": 46.0,
-      "step": 24601
-    },
-    {
-      "epoch": 1.8809947053539002,
-      "grad_norm": 0.0018763246480375528,
-      "learning_rate": 0.00019999825541887586,
-      "loss": 46.0,
-      "step": 24602
-    },
-    {
-      "epoch": 1.88107116233729,
-      "grad_norm": 0.0012015707325190306,
-      "learning_rate": 0.00019999825527699142,
-      "loss": 46.0,
-      "step": 24603
-    },
-    {
-      "epoch": 1.8811476193206798,
-      "grad_norm": 0.001744012231938541,
-      "learning_rate": 0.00019999825513510115,
-      "loss": 46.0,
-      "step": 24604
-    },
-    {
-      "epoch": 1.8812240763040693,
-      "grad_norm": 0.0005731271230615675,
-      "learning_rate": 0.00019999825499320513,
-      "loss": 46.0,
-      "step": 24605
-    },
-    {
-      "epoch": 1.881300533287459,
-      "grad_norm": 0.0017357015749439597,
-      "learning_rate": 0.00019999825485130334,
-      "loss": 46.0,
-      "step": 24606
-    },
-    {
-      "epoch": 1.8813769902708488,
-      "grad_norm": 0.0012829415500164032,
-      "learning_rate": 0.0001999982547093958,
-      "loss": 46.0,
-      "step": 24607
-    },
-    {
-      "epoch": 1.8814534472542386,
-      "grad_norm": 0.00039806694258004427,
-      "learning_rate": 0.00019999825456748247,
-      "loss": 46.0,
-      "step": 24608
-    },
-    {
-      "epoch": 1.8815299042376283,
-      "grad_norm": 0.0012564085191115737,
-      "learning_rate": 0.00019999825442556339,
-      "loss": 46.0,
-      "step": 24609
-    },
-    {
-      "epoch": 1.881606361221018,
-      "grad_norm": 0.0004147278086747974,
-      "learning_rate": 0.00019999825428363853,
-      "loss": 46.0,
-      "step": 24610
-    },
-    {
-      "epoch": 1.8816828182044079,
-      "grad_norm": 0.0023488050792366266,
-      "learning_rate": 0.0001999982541417079,
-      "loss": 46.0,
-      "step": 24611
-    },
-    {
-      "epoch": 1.8817592751877976,
-      "grad_norm": 0.0011952234199270606,
-      "learning_rate": 0.00019999825399977147,
-      "loss": 46.0,
-      "step": 24612
-    },
-    {
-      "epoch": 1.8818357321711872,
-      "grad_norm": 0.0028413129039108753,
-      "learning_rate": 0.00019999825385782932,
-      "loss": 46.0,
-      "step": 24613
-    },
-    {
-      "epoch": 1.881912189154577,
-      "grad_norm": 0.0017588648479431868,
-      "learning_rate": 0.00019999825371588137,
-      "loss": 46.0,
-      "step": 24614
-    },
-    {
-      "epoch": 1.8819886461379667,
-      "grad_norm": 0.0005073270876891911,
-      "learning_rate": 0.00019999825357392768,
-      "loss": 46.0,
-      "step": 24615
-    },
-    {
-      "epoch": 1.8820651031213562,
-      "grad_norm": 0.0005572976660914719,
-      "learning_rate": 0.0001999982534319682,
-      "loss": 46.0,
-      "step": 24616
-    },
-    {
-      "epoch": 1.882141560104746,
-      "grad_norm": 0.0005067386664450169,
-      "learning_rate": 0.00019999825329000294,
-      "loss": 46.0,
-      "step": 24617
-    },
-    {
-      "epoch": 1.8822180170881357,
-      "grad_norm": 0.0006053035031072795,
-      "learning_rate": 0.00019999825314803192,
-      "loss": 46.0,
-      "step": 24618
-    },
-    {
-      "epoch": 1.8822944740715255,
-      "grad_norm": 0.0004883640795014799,
-      "learning_rate": 0.00019999825300605516,
-      "loss": 46.0,
-      "step": 24619
-    },
-    {
-      "epoch": 1.8823709310549153,
-      "grad_norm": 0.0014096918748691678,
-      "learning_rate": 0.0001999982528640726,
-      "loss": 46.0,
-      "step": 24620
-    },
-    {
-      "epoch": 1.882447388038305,
-      "grad_norm": 0.0008274352876469493,
-      "learning_rate": 0.00019999825272208429,
-      "loss": 46.0,
-      "step": 24621
-    },
-    {
-      "epoch": 1.8825238450216948,
-      "grad_norm": 0.001104346476495266,
-      "learning_rate": 0.00019999825258009018,
-      "loss": 46.0,
-      "step": 24622
-    },
-    {
-      "epoch": 1.8826003020050845,
-      "grad_norm": 0.001666387775912881,
-      "learning_rate": 0.0001999982524380903,
-      "loss": 46.0,
-      "step": 24623
-    },
-    {
-      "epoch": 1.882676758988474,
-      "grad_norm": 0.009655337780714035,
-      "learning_rate": 0.0001999982522960847,
-      "loss": 46.0,
-      "step": 24624
-    },
-    {
-      "epoch": 1.8827532159718638,
-      "grad_norm": 0.0042439959943294525,
-      "learning_rate": 0.00019999825215407326,
-      "loss": 46.0,
-      "step": 24625
-    },
-    {
-      "epoch": 1.8828296729552534,
-      "grad_norm": 0.0009557383018545806,
-      "learning_rate": 0.0001999982520120561,
-      "loss": 46.0,
-      "step": 24626
-    },
-    {
-      "epoch": 1.8829061299386431,
-      "grad_norm": 0.0017679401207715273,
-      "learning_rate": 0.00019999825187003316,
-      "loss": 46.0,
-      "step": 24627
-    },
-    {
-      "epoch": 1.882982586922033,
-      "grad_norm": 0.0005287333624437451,
-      "learning_rate": 0.00019999825172800444,
-      "loss": 46.0,
-      "step": 24628
-    },
-    {
-      "epoch": 1.8830590439054227,
-      "grad_norm": 0.001273384434171021,
-      "learning_rate": 0.00019999825158596997,
-      "loss": 46.0,
-      "step": 24629
-    },
-    {
-      "epoch": 1.8831355008888124,
-      "grad_norm": 0.0006854168022982776,
-      "learning_rate": 0.00019999825144392973,
-      "loss": 46.0,
-      "step": 24630
-    },
-    {
-      "epoch": 1.8832119578722022,
-      "grad_norm": 0.0008531357161700726,
-      "learning_rate": 0.00019999825130188369,
-      "loss": 46.0,
-      "step": 24631
-    },
-    {
-      "epoch": 1.883288414855592,
-      "grad_norm": 0.0031230042222887278,
-      "learning_rate": 0.0001999982511598319,
-      "loss": 46.0,
-      "step": 24632
-    },
-    {
-      "epoch": 1.8833648718389817,
-      "grad_norm": 0.0006191385327838361,
-      "learning_rate": 0.00019999825101777436,
-      "loss": 46.0,
-      "step": 24633
-    },
-    {
-      "epoch": 1.8834413288223715,
-      "grad_norm": 0.000485429132822901,
-      "learning_rate": 0.00019999825087571105,
-      "loss": 46.0,
-      "step": 24634
-    },
-    {
-      "epoch": 1.883517785805761,
-      "grad_norm": 0.0030190711840987206,
-      "learning_rate": 0.00019999825073364194,
-      "loss": 46.0,
-      "step": 24635
-    },
-    {
-      "epoch": 1.8835942427891508,
-      "grad_norm": 0.002198781119659543,
-      "learning_rate": 0.00019999825059156706,
-      "loss": 46.0,
-      "step": 24636
-    },
-    {
-      "epoch": 1.8836706997725403,
-      "grad_norm": 0.0005672792322002351,
-      "learning_rate": 0.00019999825044948643,
-      "loss": 46.0,
-      "step": 24637
-    },
-    {
-      "epoch": 1.88374715675593,
-      "grad_norm": 0.001003442332148552,
-      "learning_rate": 0.00019999825030740006,
-      "loss": 46.0,
-      "step": 24638
-    },
-    {
-      "epoch": 1.8838236137393198,
-      "grad_norm": 0.0010664293076843023,
-      "learning_rate": 0.00019999825016530785,
-      "loss": 46.0,
-      "step": 24639
-    },
-    {
-      "epoch": 1.8839000707227096,
-      "grad_norm": 0.002378760138526559,
-      "learning_rate": 0.0001999982500232099,
-      "loss": 46.0,
-      "step": 24640
-    },
-    {
-      "epoch": 1.8839765277060994,
-      "grad_norm": 0.0005571945803239942,
-      "learning_rate": 0.0001999982498811062,
-      "loss": 46.0,
-      "step": 24641
-    },
-    {
-      "epoch": 1.8840529846894891,
-      "grad_norm": 0.0011518622050061822,
-      "learning_rate": 0.0001999982497389967,
-      "loss": 46.0,
-      "step": 24642
-    },
-    {
-      "epoch": 1.8841294416728789,
-      "grad_norm": 0.002100169425830245,
-      "learning_rate": 0.00019999824959688147,
-      "loss": 46.0,
-      "step": 24643
-    },
-    {
-      "epoch": 1.8842058986562686,
-      "grad_norm": 0.0033432957716286182,
-      "learning_rate": 0.00019999824945476046,
-      "loss": 46.0,
-      "step": 24644
-    },
-    {
-      "epoch": 1.8842823556396584,
-      "grad_norm": 0.014557997696101665,
-      "learning_rate": 0.00019999824931263367,
-      "loss": 46.0,
-      "step": 24645
-    },
-    {
-      "epoch": 1.884358812623048,
-      "grad_norm": 0.0005380836082622409,
-      "learning_rate": 0.0001999982491705011,
-      "loss": 46.0,
-      "step": 24646
-    },
-    {
-      "epoch": 1.8844352696064377,
-      "grad_norm": 0.0005122815491631627,
-      "learning_rate": 0.00019999824902836277,
-      "loss": 46.0,
-      "step": 24647
-    },
-    {
-      "epoch": 1.8845117265898272,
-      "grad_norm": 0.0013482756912708282,
-      "learning_rate": 0.0001999982488862187,
-      "loss": 46.0,
-      "step": 24648
-    },
-    {
-      "epoch": 1.884588183573217,
-      "grad_norm": 0.002124207792803645,
-      "learning_rate": 0.0001999982487440688,
-      "loss": 46.0,
-      "step": 24649
-    },
-    {
-      "epoch": 1.8846646405566068,
-      "grad_norm": 0.0017691556131467223,
-      "learning_rate": 0.00019999824860191318,
-      "loss": 46.0,
-      "step": 24650
-    },
-    {
-      "epoch": 1.8847410975399965,
-      "grad_norm": 0.0010154476622119546,
-      "learning_rate": 0.0001999982484597518,
-      "loss": 46.0,
-      "step": 24651
-    },
-    {
-      "epoch": 1.8848175545233863,
-      "grad_norm": 0.0008109635673463345,
-      "learning_rate": 0.0001999982483175846,
-      "loss": 46.0,
-      "step": 24652
-    },
-    {
-      "epoch": 1.884894011506776,
-      "grad_norm": 0.0010708397021517158,
-      "learning_rate": 0.00019999824817541165,
-      "loss": 46.0,
-      "step": 24653
-    },
-    {
-      "epoch": 1.8849704684901658,
-      "grad_norm": 0.007725434843450785,
-      "learning_rate": 0.00019999824803323296,
-      "loss": 46.0,
-      "step": 24654
-    },
-    {
-      "epoch": 1.8850469254735556,
-      "grad_norm": 0.0008287687669508159,
-      "learning_rate": 0.00019999824789104846,
-      "loss": 46.0,
-      "step": 24655
-    },
-    {
-      "epoch": 1.885123382456945,
-      "grad_norm": 0.0011678702430799603,
-      "learning_rate": 0.00019999824774885822,
-      "loss": 46.0,
-      "step": 24656
-    },
-    {
-      "epoch": 1.8851998394403349,
-      "grad_norm": 0.0008198132854886353,
-      "learning_rate": 0.0001999982476066622,
-      "loss": 46.0,
-      "step": 24657
-    },
-    {
-      "epoch": 1.8852762964237246,
-      "grad_norm": 0.004641393199563026,
-      "learning_rate": 0.0001999982474644604,
-      "loss": 46.0,
-      "step": 24658
-    },
-    {
-      "epoch": 1.8853527534071142,
-      "grad_norm": 0.0053079151548445225,
-      "learning_rate": 0.00019999824732225285,
-      "loss": 46.0,
-      "step": 24659
-    },
-    {
-      "epoch": 1.885429210390504,
-      "grad_norm": 0.000767974357586354,
-      "learning_rate": 0.0001999982471800395,
-      "loss": 46.0,
-      "step": 24660
-    },
-    {
-      "epoch": 1.8855056673738937,
-      "grad_norm": 0.0012130244867876172,
-      "learning_rate": 0.0001999982470378204,
-      "loss": 46.0,
-      "step": 24661
-    },
-    {
-      "epoch": 1.8855821243572835,
-      "grad_norm": 0.004133188631385565,
-      "learning_rate": 0.00019999824689559555,
-      "loss": 46.0,
-      "step": 24662
-    },
-    {
-      "epoch": 1.8856585813406732,
-      "grad_norm": 0.002551071112975478,
-      "learning_rate": 0.0001999982467533649,
-      "loss": 46.0,
-      "step": 24663
-    },
-    {
-      "epoch": 1.885735038324063,
-      "grad_norm": 0.001413128338754177,
-      "learning_rate": 0.0001999982466111285,
-      "loss": 46.0,
-      "step": 24664
-    },
-    {
-      "epoch": 1.8858114953074527,
-      "grad_norm": 0.003531447611749172,
-      "learning_rate": 0.00019999824646888632,
-      "loss": 46.0,
-      "step": 24665
-    },
-    {
-      "epoch": 1.8858879522908425,
-      "grad_norm": 0.0005414483603090048,
-      "learning_rate": 0.00019999824632663837,
-      "loss": 46.0,
-      "step": 24666
-    },
-    {
-      "epoch": 1.885964409274232,
-      "grad_norm": 0.0007398679736070335,
-      "learning_rate": 0.00019999824618438465,
-      "loss": 46.0,
-      "step": 24667
-    },
-    {
-      "epoch": 1.8860408662576218,
-      "grad_norm": 0.0009434225503355265,
-      "learning_rate": 0.00019999824604212516,
-      "loss": 46.0,
-      "step": 24668
-    },
-    {
-      "epoch": 1.8861173232410116,
-      "grad_norm": 0.0013627024600282311,
-      "learning_rate": 0.00019999824589985991,
-      "loss": 46.0,
-      "step": 24669
-    },
-    {
-      "epoch": 1.886193780224401,
-      "grad_norm": 0.0005271456320770085,
-      "learning_rate": 0.0001999982457575889,
-      "loss": 46.0,
-      "step": 24670
-    },
-    {
-      "epoch": 1.8862702372077909,
-      "grad_norm": 0.0014919457025825977,
-      "learning_rate": 0.0001999982456153121,
-      "loss": 46.0,
-      "step": 24671
-    },
-    {
-      "epoch": 1.8863466941911806,
-      "grad_norm": 0.0034128958359360695,
-      "learning_rate": 0.00019999824547302955,
-      "loss": 46.0,
-      "step": 24672
-    },
-    {
-      "epoch": 1.8864231511745704,
-      "grad_norm": 0.0010589085286483169,
-      "learning_rate": 0.00019999824533074121,
-      "loss": 46.0,
-      "step": 24673
-    },
-    {
-      "epoch": 1.8864996081579601,
-      "grad_norm": 0.0007168719894252717,
-      "learning_rate": 0.0001999982451884471,
-      "loss": 46.0,
-      "step": 24674
-    },
-    {
-      "epoch": 1.88657606514135,
-      "grad_norm": 0.0012177197495475411,
-      "learning_rate": 0.00019999824504614722,
-      "loss": 46.0,
-      "step": 24675
-    },
-    {
-      "epoch": 1.8866525221247397,
-      "grad_norm": 0.00045947369653731585,
-      "learning_rate": 0.0001999982449038416,
-      "loss": 46.0,
-      "step": 24676
-    },
-    {
-      "epoch": 1.8867289791081294,
-      "grad_norm": 0.0011018654331564903,
-      "learning_rate": 0.0001999982447615302,
-      "loss": 46.0,
-      "step": 24677
-    },
-    {
-      "epoch": 1.886805436091519,
-      "grad_norm": 0.003252594731748104,
-      "learning_rate": 0.00019999824461921302,
-      "loss": 46.0,
-      "step": 24678
-    },
-    {
-      "epoch": 1.8868818930749087,
-      "grad_norm": 0.0006382344872690737,
-      "learning_rate": 0.00019999824447689007,
-      "loss": 46.0,
-      "step": 24679
-    },
-    {
-      "epoch": 1.8869583500582985,
-      "grad_norm": 0.003114924533292651,
-      "learning_rate": 0.00019999824433456135,
-      "loss": 46.0,
-      "step": 24680
-    },
-    {
-      "epoch": 1.887034807041688,
-      "grad_norm": 0.005569116212427616,
-      "learning_rate": 0.00019999824419222688,
-      "loss": 46.0,
-      "step": 24681
-    },
-    {
-      "epoch": 1.8871112640250778,
-      "grad_norm": 0.0013368381187319756,
-      "learning_rate": 0.00019999824404988661,
-      "loss": 46.0,
-      "step": 24682
-    },
-    {
-      "epoch": 1.8871877210084675,
-      "grad_norm": 0.0011965315788984299,
-      "learning_rate": 0.00019999824390754057,
-      "loss": 46.0,
-      "step": 24683
-    },
-    {
-      "epoch": 1.8872641779918573,
-      "grad_norm": 0.0025736431125551462,
-      "learning_rate": 0.00019999824376518878,
-      "loss": 46.0,
-      "step": 24684
-    },
-    {
-      "epoch": 1.887340634975247,
-      "grad_norm": 0.0012781030964106321,
-      "learning_rate": 0.00019999824362283125,
-      "loss": 46.0,
-      "step": 24685
-    },
-    {
-      "epoch": 1.8874170919586368,
-      "grad_norm": 0.0010048523545265198,
-      "learning_rate": 0.0001999982434804679,
-      "loss": 46.0,
-      "step": 24686
-    },
-    {
-      "epoch": 1.8874935489420266,
-      "grad_norm": 0.0029230224899947643,
-      "learning_rate": 0.00019999824333809878,
-      "loss": 46.0,
-      "step": 24687
-    },
-    {
-      "epoch": 1.8875700059254163,
-      "grad_norm": 0.000682733254507184,
-      "learning_rate": 0.0001999982431957239,
-      "loss": 46.0,
-      "step": 24688
-    },
-    {
-      "epoch": 1.8876464629088059,
-      "grad_norm": 0.0018335724016651511,
-      "learning_rate": 0.00019999824305334327,
-      "loss": 46.0,
-      "step": 24689
-    },
-    {
-      "epoch": 1.8877229198921956,
-      "grad_norm": 0.0008607509662397206,
-      "learning_rate": 0.00019999824291095684,
-      "loss": 46.0,
-      "step": 24690
-    },
-    {
-      "epoch": 1.8877993768755854,
-      "grad_norm": 0.0016742546577006578,
-      "learning_rate": 0.0001999982427685647,
-      "loss": 46.0,
-      "step": 24691
-    },
-    {
-      "epoch": 1.887875833858975,
-      "grad_norm": 0.0028891898691654205,
-      "learning_rate": 0.00019999824262616675,
-      "loss": 46.0,
-      "step": 24692
-    },
-    {
-      "epoch": 1.8879522908423647,
-      "grad_norm": 0.002334264572709799,
-      "learning_rate": 0.00019999824248376303,
-      "loss": 46.0,
-      "step": 24693
-    },
-    {
-      "epoch": 1.8880287478257545,
-      "grad_norm": 0.0005821464001201093,
-      "learning_rate": 0.00019999824234135353,
-      "loss": 46.0,
-      "step": 24694
-    },
-    {
-      "epoch": 1.8881052048091442,
-      "grad_norm": 0.0017732614651322365,
-      "learning_rate": 0.0001999982421989383,
-      "loss": 46.0,
-      "step": 24695
-    },
-    {
-      "epoch": 1.888181661792534,
-      "grad_norm": 0.0034232784528285265,
-      "learning_rate": 0.00019999824205651728,
-      "loss": 46.0,
-      "step": 24696
-    },
-    {
-      "epoch": 1.8882581187759238,
-      "grad_norm": 0.0005851819296367466,
-      "learning_rate": 0.00019999824191409044,
-      "loss": 46.0,
-      "step": 24697
-    },
-    {
-      "epoch": 1.8883345757593135,
-      "grad_norm": 0.0005628893850371242,
-      "learning_rate": 0.0001999982417716579,
-      "loss": 46.0,
-      "step": 24698
-    },
-    {
-      "epoch": 1.8884110327427033,
-      "grad_norm": 0.0009058432769961655,
-      "learning_rate": 0.00019999824162921957,
-      "loss": 46.0,
-      "step": 24699
-    },
-    {
-      "epoch": 1.8884874897260928,
-      "grad_norm": 0.0008194043184630573,
-      "learning_rate": 0.00019999824148677546,
-      "loss": 46.0,
-      "step": 24700
-    },
-    {
-      "epoch": 1.8885639467094826,
-      "grad_norm": 0.0009048864594660699,
-      "learning_rate": 0.00019999824134432558,
-      "loss": 46.0,
-      "step": 24701
-    },
-    {
-      "epoch": 1.8886404036928723,
-      "grad_norm": 0.0005796966725029051,
-      "learning_rate": 0.00019999824120186995,
-      "loss": 46.0,
-      "step": 24702
-    },
-    {
-      "epoch": 1.8887168606762619,
-      "grad_norm": 0.002342530293390155,
-      "learning_rate": 0.00019999824105940852,
-      "loss": 46.0,
-      "step": 24703
-    },
-    {
-      "epoch": 1.8887933176596516,
-      "grad_norm": 0.0012555879075080156,
-      "learning_rate": 0.00019999824091694135,
-      "loss": 46.0,
-      "step": 24704
-    },
-    {
-      "epoch": 1.8888697746430414,
-      "grad_norm": 0.0007464020745828748,
-      "learning_rate": 0.0001999982407744684,
-      "loss": 46.0,
-      "step": 24705
-    },
-    {
-      "epoch": 1.8889462316264312,
-      "grad_norm": 0.00080097810132429,
-      "learning_rate": 0.00019999824063198965,
-      "loss": 46.0,
-      "step": 24706
-    },
-    {
-      "epoch": 1.889022688609821,
-      "grad_norm": 0.005033240187913179,
-      "learning_rate": 0.0001999982404895052,
-      "loss": 46.0,
-      "step": 24707
-    },
-    {
-      "epoch": 1.8890991455932107,
-      "grad_norm": 0.0006725105340592563,
-      "learning_rate": 0.00019999824034701492,
-      "loss": 46.0,
-      "step": 24708
-    },
-    {
-      "epoch": 1.8891756025766004,
-      "grad_norm": 0.0007440222543664277,
-      "learning_rate": 0.0001999982402045189,
-      "loss": 46.0,
-      "step": 24709
-    },
-    {
-      "epoch": 1.8892520595599902,
-      "grad_norm": 0.0007157246000133455,
-      "learning_rate": 0.00019999824006201712,
-      "loss": 46.0,
-      "step": 24710
-    },
-    {
-      "epoch": 1.8893285165433797,
-      "grad_norm": 0.0027019151020795107,
-      "learning_rate": 0.00019999823991950956,
-      "loss": 46.0,
-      "step": 24711
-    },
-    {
-      "epoch": 1.8894049735267695,
-      "grad_norm": 0.002709275344386697,
-      "learning_rate": 0.00019999823977699622,
-      "loss": 46.0,
-      "step": 24712
-    },
-    {
-      "epoch": 1.8894814305101593,
-      "grad_norm": 0.0031730926129966974,
-      "learning_rate": 0.00019999823963447712,
-      "loss": 46.0,
-      "step": 24713
-    },
-    {
-      "epoch": 1.8895578874935488,
-      "grad_norm": 0.0017869881121441722,
-      "learning_rate": 0.00019999823949195224,
-      "loss": 46.0,
-      "step": 24714
-    },
-    {
-      "epoch": 1.8896343444769386,
-      "grad_norm": 0.0026095665525645018,
-      "learning_rate": 0.00019999823934942158,
-      "loss": 46.0,
-      "step": 24715
-    },
-    {
-      "epoch": 1.8897108014603283,
-      "grad_norm": 0.0005542807630263269,
-      "learning_rate": 0.00019999823920688515,
-      "loss": 46.0,
-      "step": 24716
-    },
-    {
-      "epoch": 1.889787258443718,
-      "grad_norm": 0.0023922398686408997,
-      "learning_rate": 0.00019999823906434298,
-      "loss": 46.0,
-      "step": 24717
-    },
-    {
-      "epoch": 1.8898637154271078,
-      "grad_norm": 0.0002743404184002429,
-      "learning_rate": 0.00019999823892179503,
-      "loss": 46.0,
-      "step": 24718
-    },
-    {
-      "epoch": 1.8899401724104976,
-      "grad_norm": 0.0017191022634506226,
-      "learning_rate": 0.0001999982387792413,
-      "loss": 46.0,
-      "step": 24719
-    },
-    {
-      "epoch": 1.8900166293938874,
-      "grad_norm": 0.003249636385589838,
-      "learning_rate": 0.00019999823863668185,
-      "loss": 46.0,
-      "step": 24720
-    },
-    {
-      "epoch": 1.8900930863772771,
-      "grad_norm": 0.0009787646122276783,
-      "learning_rate": 0.00019999823849411655,
-      "loss": 46.0,
-      "step": 24721
-    },
-    {
-      "epoch": 1.8901695433606667,
-      "grad_norm": 0.0024778626393526793,
-      "learning_rate": 0.0001999982383515455,
-      "loss": 46.0,
-      "step": 24722
-    },
-    {
-      "epoch": 1.8902460003440564,
-      "grad_norm": 0.001309894840233028,
-      "learning_rate": 0.00019999823820896872,
-      "loss": 46.0,
-      "step": 24723
-    },
-    {
-      "epoch": 1.8903224573274462,
-      "grad_norm": 0.0016520543722435832,
-      "learning_rate": 0.00019999823806638616,
-      "loss": 46.0,
-      "step": 24724
-    },
-    {
-      "epoch": 1.8903989143108357,
-      "grad_norm": 0.001087581622414291,
-      "learning_rate": 0.00019999823792379783,
-      "loss": 46.0,
-      "step": 24725
-    },
-    {
-      "epoch": 1.8904753712942255,
-      "grad_norm": 0.0013746342156082392,
-      "learning_rate": 0.00019999823778120372,
-      "loss": 46.0,
-      "step": 24726
-    },
-    {
-      "epoch": 1.8905518282776153,
-      "grad_norm": 0.0037622456438839436,
-      "learning_rate": 0.00019999823763860387,
-      "loss": 46.0,
-      "step": 24727
-    },
-    {
-      "epoch": 1.890628285261005,
-      "grad_norm": 0.003025029320269823,
-      "learning_rate": 0.0001999982374959982,
-      "loss": 46.0,
-      "step": 24728
-    },
-    {
-      "epoch": 1.8907047422443948,
-      "grad_norm": 0.0031665421556681395,
-      "learning_rate": 0.0001999982373533868,
-      "loss": 46.0,
-      "step": 24729
-    },
-    {
-      "epoch": 1.8907811992277845,
-      "grad_norm": 0.009643260389566422,
-      "learning_rate": 0.00019999823721076961,
-      "loss": 46.0,
-      "step": 24730
-    },
-    {
-      "epoch": 1.8908576562111743,
-      "grad_norm": 0.0002975004317704588,
-      "learning_rate": 0.00019999823706814664,
-      "loss": 46.0,
-      "step": 24731
-    },
-    {
-      "epoch": 1.890934113194564,
-      "grad_norm": 0.026694513857364655,
-      "learning_rate": 0.00019999823692551792,
-      "loss": 46.0,
-      "step": 24732
-    },
-    {
-      "epoch": 1.8910105701779536,
-      "grad_norm": 0.0008916464284993708,
-      "learning_rate": 0.00019999823678288343,
-      "loss": 46.0,
-      "step": 24733
-    },
-    {
-      "epoch": 1.8910870271613434,
-      "grad_norm": 0.0030395861249417067,
-      "learning_rate": 0.00019999823664024319,
-      "loss": 46.0,
-      "step": 24734
-    },
-    {
-      "epoch": 1.8911634841447331,
-      "grad_norm": 0.0012122743064537644,
-      "learning_rate": 0.00019999823649759715,
-      "loss": 46.0,
-      "step": 24735
-    },
-    {
-      "epoch": 1.8912399411281227,
-      "grad_norm": 0.001330947270616889,
-      "learning_rate": 0.00019999823635494533,
-      "loss": 46.0,
-      "step": 24736
-    },
-    {
-      "epoch": 1.8913163981115124,
-      "grad_norm": 0.0003822512226179242,
-      "learning_rate": 0.00019999823621228777,
-      "loss": 46.0,
-      "step": 24737
-    },
-    {
-      "epoch": 1.8913928550949022,
-      "grad_norm": 0.0010991458548232913,
-      "learning_rate": 0.0001999982360696244,
-      "loss": 46.0,
-      "step": 24738
-    },
-    {
-      "epoch": 1.891469312078292,
-      "grad_norm": 0.0005206392961554229,
-      "learning_rate": 0.00019999823592695533,
-      "loss": 46.0,
-      "step": 24739
-    },
-    {
-      "epoch": 1.8915457690616817,
-      "grad_norm": 0.00035012204898521304,
-      "learning_rate": 0.00019999823578428045,
-      "loss": 46.0,
-      "step": 24740
-    },
-    {
-      "epoch": 1.8916222260450715,
-      "grad_norm": 0.001025093370117247,
-      "learning_rate": 0.0001999982356415998,
-      "loss": 46.0,
-      "step": 24741
-    },
-    {
-      "epoch": 1.8916986830284612,
-      "grad_norm": 0.00045775799662806094,
-      "learning_rate": 0.0001999982354989134,
-      "loss": 46.0,
-      "step": 24742
-    },
-    {
-      "epoch": 1.891775140011851,
-      "grad_norm": 0.0010018307948485017,
-      "learning_rate": 0.0001999982353562212,
-      "loss": 46.0,
-      "step": 24743
-    },
-    {
-      "epoch": 1.8918515969952405,
-      "grad_norm": 0.0011945945443585515,
-      "learning_rate": 0.00019999823521352322,
-      "loss": 46.0,
-      "step": 24744
-    },
-    {
-      "epoch": 1.8919280539786303,
-      "grad_norm": 0.0020748113747686148,
-      "learning_rate": 0.0001999982350708195,
-      "loss": 46.0,
-      "step": 24745
-    },
-    {
-      "epoch": 1.89200451096202,
-      "grad_norm": 0.0012052421225234866,
-      "learning_rate": 0.00019999823492811004,
-      "loss": 46.0,
-      "step": 24746
-    },
-    {
-      "epoch": 1.8920809679454096,
-      "grad_norm": 0.0009332199115306139,
-      "learning_rate": 0.00019999823478539477,
-      "loss": 46.0,
-      "step": 24747
-    },
-    {
-      "epoch": 1.8921574249287993,
-      "grad_norm": 0.0007355533307418227,
-      "learning_rate": 0.00019999823464267373,
-      "loss": 46.0,
-      "step": 24748
-    },
-    {
-      "epoch": 1.892233881912189,
-      "grad_norm": 0.0011991269420832396,
-      "learning_rate": 0.00019999823449994692,
-      "loss": 46.0,
-      "step": 24749
-    },
-    {
-      "epoch": 1.8923103388955789,
-      "grad_norm": 0.0010245504090562463,
-      "learning_rate": 0.00019999823435721433,
-      "loss": 46.0,
-      "step": 24750
-    },
-    {
-      "epoch": 1.8923867958789686,
-      "grad_norm": 0.0007938212365843356,
-      "learning_rate": 0.000199998234214476,
-      "loss": 46.0,
-      "step": 24751
-    },
-    {
-      "epoch": 1.8924632528623584,
-      "grad_norm": 0.0010722477454692125,
-      "learning_rate": 0.0001999982340717319,
-      "loss": 46.0,
-      "step": 24752
-    },
-    {
-      "epoch": 1.8925397098457482,
-      "grad_norm": 0.005164948757737875,
-      "learning_rate": 0.00019999823392898204,
-      "loss": 46.0,
-      "step": 24753
-    },
-    {
-      "epoch": 1.892616166829138,
-      "grad_norm": 0.0007275641546584666,
-      "learning_rate": 0.00019999823378622639,
-      "loss": 46.0,
-      "step": 24754
-    },
-    {
-      "epoch": 1.8926926238125275,
-      "grad_norm": 0.0010443564970046282,
-      "learning_rate": 0.00019999823364346496,
-      "loss": 46.0,
-      "step": 24755
-    },
-    {
-      "epoch": 1.8927690807959172,
-      "grad_norm": 0.003957248758524656,
-      "learning_rate": 0.00019999823350069776,
-      "loss": 46.0,
-      "step": 24756
-    },
-    {
-      "epoch": 1.8928455377793068,
-      "grad_norm": 0.0015251325676217675,
-      "learning_rate": 0.0001999982333579248,
-      "loss": 46.0,
-      "step": 24757
-    },
-    {
-      "epoch": 1.8929219947626965,
-      "grad_norm": 0.0004211643827147782,
-      "learning_rate": 0.0001999982332151461,
-      "loss": 46.0,
-      "step": 24758
-    },
-    {
-      "epoch": 1.8929984517460863,
-      "grad_norm": 0.0007466599345207214,
-      "learning_rate": 0.00019999823307236157,
-      "loss": 46.0,
-      "step": 24759
-    },
-    {
-      "epoch": 1.893074908729476,
-      "grad_norm": 0.004767555743455887,
-      "learning_rate": 0.0001999982329295713,
-      "loss": 46.0,
-      "step": 24760
-    },
-    {
-      "epoch": 1.8931513657128658,
-      "grad_norm": 0.010698717087507248,
-      "learning_rate": 0.00019999823278677527,
-      "loss": 46.0,
-      "step": 24761
-    },
-    {
-      "epoch": 1.8932278226962556,
-      "grad_norm": 0.006788797676563263,
-      "learning_rate": 0.00019999823264397348,
-      "loss": 46.0,
-      "step": 24762
-    },
-    {
-      "epoch": 1.8933042796796453,
-      "grad_norm": 0.0018999719759449363,
-      "learning_rate": 0.00019999823250116592,
-      "loss": 46.0,
-      "step": 24763
-    },
-    {
-      "epoch": 1.893380736663035,
-      "grad_norm": 0.0007776694837957621,
-      "learning_rate": 0.00019999823235835256,
-      "loss": 46.0,
-      "step": 24764
-    },
-    {
-      "epoch": 1.8934571936464248,
-      "grad_norm": 0.000996719696559012,
-      "learning_rate": 0.00019999823221553346,
-      "loss": 46.0,
-      "step": 24765
-    },
-    {
-      "epoch": 1.8935336506298144,
-      "grad_norm": 0.0007827046792954206,
-      "learning_rate": 0.00019999823207270858,
-      "loss": 46.0,
-      "step": 24766
-    },
-    {
-      "epoch": 1.8936101076132041,
-      "grad_norm": 0.0022264192812144756,
-      "learning_rate": 0.00019999823192987792,
-      "loss": 46.0,
-      "step": 24767
-    },
-    {
-      "epoch": 1.8936865645965937,
-      "grad_norm": 0.0022582896053791046,
-      "learning_rate": 0.00019999823178704152,
-      "loss": 46.0,
-      "step": 24768
-    },
-    {
-      "epoch": 1.8937630215799834,
-      "grad_norm": 0.002451102714985609,
-      "learning_rate": 0.00019999823164419932,
-      "loss": 46.0,
-      "step": 24769
-    },
-    {
-      "epoch": 1.8938394785633732,
-      "grad_norm": 0.0006675709155388176,
-      "learning_rate": 0.00019999823150135138,
-      "loss": 46.0,
-      "step": 24770
-    },
-    {
-      "epoch": 1.893915935546763,
-      "grad_norm": 0.0007069043931551278,
-      "learning_rate": 0.00019999823135849763,
-      "loss": 46.0,
-      "step": 24771
-    },
-    {
-      "epoch": 1.8939923925301527,
-      "grad_norm": 0.009171134792268276,
-      "learning_rate": 0.00019999823121563814,
-      "loss": 46.0,
-      "step": 24772
-    },
-    {
-      "epoch": 1.8940688495135425,
-      "grad_norm": 0.00439388258382678,
-      "learning_rate": 0.00019999823107277287,
-      "loss": 46.0,
-      "step": 24773
-    },
-    {
-      "epoch": 1.8941453064969322,
-      "grad_norm": 0.0012644353555515409,
-      "learning_rate": 0.0001999982309299018,
-      "loss": 46.0,
-      "step": 24774
-    },
-    {
-      "epoch": 1.894221763480322,
-      "grad_norm": 0.0014597425470128655,
-      "learning_rate": 0.00019999823078702505,
-      "loss": 46.0,
-      "step": 24775
-    },
-    {
-      "epoch": 1.8942982204637118,
-      "grad_norm": 0.0011703589698299766,
-      "learning_rate": 0.00019999823064414243,
-      "loss": 46.0,
-      "step": 24776
-    },
-    {
-      "epoch": 1.8943746774471013,
-      "grad_norm": 0.0012173581635579467,
-      "learning_rate": 0.00019999823050125413,
-      "loss": 46.0,
-      "step": 24777
-    },
-    {
-      "epoch": 1.894451134430491,
-      "grad_norm": 0.0006201710202731192,
-      "learning_rate": 0.00019999823035836,
-      "loss": 46.0,
-      "step": 24778
-    },
-    {
-      "epoch": 1.8945275914138806,
-      "grad_norm": 0.002437945455312729,
-      "learning_rate": 0.00019999823021546012,
-      "loss": 46.0,
-      "step": 24779
-    },
-    {
-      "epoch": 1.8946040483972704,
-      "grad_norm": 0.008216519840061665,
-      "learning_rate": 0.00019999823007255446,
-      "loss": 46.0,
-      "step": 24780
-    },
-    {
-      "epoch": 1.8946805053806601,
-      "grad_norm": 0.0022775973193347454,
-      "learning_rate": 0.00019999822992964307,
-      "loss": 46.0,
-      "step": 24781
-    },
-    {
-      "epoch": 1.89475696236405,
-      "grad_norm": 0.000508844677824527,
-      "learning_rate": 0.00019999822978672587,
-      "loss": 46.0,
-      "step": 24782
-    },
-    {
-      "epoch": 1.8948334193474397,
-      "grad_norm": 0.0006538983434438705,
-      "learning_rate": 0.0001999982296438029,
-      "loss": 46.0,
-      "step": 24783
-    },
-    {
-      "epoch": 1.8949098763308294,
-      "grad_norm": 0.0006354584475047886,
-      "learning_rate": 0.00019999822950087417,
-      "loss": 46.0,
-      "step": 24784
-    },
-    {
-      "epoch": 1.8949863333142192,
-      "grad_norm": 0.0010672869393602014,
-      "learning_rate": 0.00019999822935793966,
-      "loss": 46.0,
-      "step": 24785
-    },
-    {
-      "epoch": 1.895062790297609,
-      "grad_norm": 0.00047534782788716257,
-      "learning_rate": 0.0001999982292149994,
-      "loss": 46.0,
-      "step": 24786
-    },
-    {
-      "epoch": 1.8951392472809985,
-      "grad_norm": 0.002917158417403698,
-      "learning_rate": 0.00019999822907205338,
-      "loss": 46.0,
-      "step": 24787
-    },
-    {
-      "epoch": 1.8952157042643882,
-      "grad_norm": 0.002033337950706482,
-      "learning_rate": 0.00019999822892910154,
-      "loss": 46.0,
-      "step": 24788
-    },
-    {
-      "epoch": 1.895292161247778,
-      "grad_norm": 0.002093485789373517,
-      "learning_rate": 0.00019999822878614398,
-      "loss": 46.0,
-      "step": 24789
-    },
-    {
-      "epoch": 1.8953686182311675,
-      "grad_norm": 0.0015971862012520432,
-      "learning_rate": 0.00019999822864318062,
-      "loss": 46.0,
-      "step": 24790
-    },
-    {
-      "epoch": 1.8954450752145573,
-      "grad_norm": 0.0027375868521630764,
-      "learning_rate": 0.00019999822850021151,
-      "loss": 46.0,
-      "step": 24791
-    },
-    {
-      "epoch": 1.895521532197947,
-      "grad_norm": 0.0007920638890936971,
-      "learning_rate": 0.00019999822835723664,
-      "loss": 46.0,
-      "step": 24792
-    },
-    {
-      "epoch": 1.8955979891813368,
-      "grad_norm": 0.0016711171483621001,
-      "learning_rate": 0.00019999822821425598,
-      "loss": 46.0,
-      "step": 24793
-    },
-    {
-      "epoch": 1.8956744461647266,
-      "grad_norm": 0.0028647526632994413,
-      "learning_rate": 0.00019999822807126956,
-      "loss": 46.0,
-      "step": 24794
-    },
-    {
-      "epoch": 1.8957509031481163,
-      "grad_norm": 0.0002937439421657473,
-      "learning_rate": 0.00019999822792827736,
-      "loss": 46.0,
-      "step": 24795
-    },
-    {
-      "epoch": 1.895827360131506,
-      "grad_norm": 0.0005791856092400849,
-      "learning_rate": 0.00019999822778527939,
-      "loss": 46.0,
-      "step": 24796
-    },
-    {
-      "epoch": 1.8959038171148959,
-      "grad_norm": 0.0011031260946765542,
-      "learning_rate": 0.00019999822764227567,
-      "loss": 46.0,
-      "step": 24797
-    },
-    {
-      "epoch": 1.8959802740982854,
-      "grad_norm": 0.0009347690502181649,
-      "learning_rate": 0.00019999822749926618,
-      "loss": 46.0,
-      "step": 24798
-    },
-    {
-      "epoch": 1.8960567310816752,
-      "grad_norm": 0.00110166787635535,
-      "learning_rate": 0.00019999822735625088,
-      "loss": 46.0,
-      "step": 24799
-    },
-    {
-      "epoch": 1.896133188065065,
-      "grad_norm": 0.0014626196352764964,
-      "learning_rate": 0.00019999822721322987,
-      "loss": 46.0,
-      "step": 24800
-    },
-    {
-      "epoch": 1.8962096450484545,
-      "grad_norm": 0.00030768782016821206,
-      "learning_rate": 0.00019999822707020306,
-      "loss": 46.0,
-      "step": 24801
-    },
-    {
-      "epoch": 1.8962861020318442,
-      "grad_norm": 0.001377628417685628,
-      "learning_rate": 0.00019999822692717047,
-      "loss": 46.0,
-      "step": 24802
-    },
-    {
-      "epoch": 1.896362559015234,
-      "grad_norm": 0.0005198445869609714,
-      "learning_rate": 0.00019999822678413214,
-      "loss": 46.0,
-      "step": 24803
-    },
-    {
-      "epoch": 1.8964390159986237,
-      "grad_norm": 0.0016490245470777154,
-      "learning_rate": 0.00019999822664108804,
-      "loss": 46.0,
-      "step": 24804
-    },
-    {
-      "epoch": 1.8965154729820135,
-      "grad_norm": 0.0014037376968190074,
-      "learning_rate": 0.00019999822649803813,
-      "loss": 46.0,
-      "step": 24805
-    },
-    {
-      "epoch": 1.8965919299654033,
-      "grad_norm": 0.0014297586167231202,
-      "learning_rate": 0.00019999822635498248,
-      "loss": 46.0,
-      "step": 24806
-    },
-    {
-      "epoch": 1.896668386948793,
-      "grad_norm": 0.0008142938604578376,
-      "learning_rate": 0.00019999822621192105,
-      "loss": 46.0,
-      "step": 24807
-    },
-    {
-      "epoch": 1.8967448439321828,
-      "grad_norm": 0.005264353007078171,
-      "learning_rate": 0.00019999822606885385,
-      "loss": 46.0,
-      "step": 24808
-    },
-    {
-      "epoch": 1.8968213009155723,
-      "grad_norm": 0.0009005211759358644,
-      "learning_rate": 0.00019999822592578088,
-      "loss": 46.0,
-      "step": 24809
-    },
-    {
-      "epoch": 1.896897757898962,
-      "grad_norm": 0.0019074570154771209,
-      "learning_rate": 0.00019999822578270216,
-      "loss": 46.0,
-      "step": 24810
-    },
-    {
-      "epoch": 1.8969742148823519,
-      "grad_norm": 0.0023443959653377533,
-      "learning_rate": 0.00019999822563961765,
-      "loss": 46.0,
-      "step": 24811
-    },
-    {
-      "epoch": 1.8970506718657414,
-      "grad_norm": 0.0054117850959300995,
-      "learning_rate": 0.0001999982254965274,
-      "loss": 46.0,
-      "step": 24812
-    },
-    {
-      "epoch": 1.8971271288491312,
-      "grad_norm": 0.0011071175104007125,
-      "learning_rate": 0.00019999822535343134,
-      "loss": 46.0,
-      "step": 24813
-    },
-    {
-      "epoch": 1.897203585832521,
-      "grad_norm": 0.0024850473273545504,
-      "learning_rate": 0.00019999822521032953,
-      "loss": 46.0,
-      "step": 24814
-    },
-    {
-      "epoch": 1.8972800428159107,
-      "grad_norm": 0.0013730566715821624,
-      "learning_rate": 0.00019999822506722197,
-      "loss": 46.0,
-      "step": 24815
-    },
-    {
-      "epoch": 1.8973564997993004,
-      "grad_norm": 0.0004893143195658922,
-      "learning_rate": 0.0001999982249241086,
-      "loss": 46.0,
-      "step": 24816
-    },
-    {
-      "epoch": 1.8974329567826902,
-      "grad_norm": 0.00259625562466681,
-      "learning_rate": 0.0001999982247809895,
-      "loss": 46.0,
-      "step": 24817
-    },
-    {
-      "epoch": 1.89750941376608,
-      "grad_norm": 0.001510382047854364,
-      "learning_rate": 0.00019999822463786463,
-      "loss": 46.0,
-      "step": 24818
-    },
-    {
-      "epoch": 1.8975858707494697,
-      "grad_norm": 0.0018947870703414083,
-      "learning_rate": 0.00019999822449473395,
-      "loss": 46.0,
-      "step": 24819
-    },
-    {
-      "epoch": 1.8976623277328593,
-      "grad_norm": 0.013857456855475903,
-      "learning_rate": 0.00019999822435159753,
-      "loss": 46.0,
-      "step": 24820
-    },
-    {
-      "epoch": 1.897738784716249,
-      "grad_norm": 0.001011537853628397,
-      "learning_rate": 0.00019999822420845533,
-      "loss": 46.0,
-      "step": 24821
-    },
-    {
-      "epoch": 1.8978152416996388,
-      "grad_norm": 0.0015456390101462603,
-      "learning_rate": 0.00019999822406530738,
-      "loss": 46.0,
-      "step": 24822
-    },
-    {
-      "epoch": 1.8978916986830283,
-      "grad_norm": 0.0004182908160146326,
-      "learning_rate": 0.0001999982239221536,
-      "loss": 46.0,
-      "step": 24823
-    },
-    {
-      "epoch": 1.897968155666418,
-      "grad_norm": 0.0005073556094430387,
-      "learning_rate": 0.00019999822377899412,
-      "loss": 46.0,
-      "step": 24824
-    },
-    {
-      "epoch": 1.8980446126498078,
-      "grad_norm": 0.0011597982374951243,
-      "learning_rate": 0.00019999822363582885,
-      "loss": 46.0,
-      "step": 24825
-    },
-    {
-      "epoch": 1.8981210696331976,
-      "grad_norm": 0.0008622303139418364,
-      "learning_rate": 0.0001999982234926578,
-      "loss": 46.0,
-      "step": 24826
-    },
-    {
-      "epoch": 1.8981975266165874,
-      "grad_norm": 0.0008540943963453174,
-      "learning_rate": 0.000199998223349481,
-      "loss": 46.0,
-      "step": 24827
-    },
-    {
-      "epoch": 1.8982739835999771,
-      "grad_norm": 0.000912796298507601,
-      "learning_rate": 0.00019999822320629842,
-      "loss": 46.0,
-      "step": 24828
-    },
-    {
-      "epoch": 1.8983504405833669,
-      "grad_norm": 0.004008203744888306,
-      "learning_rate": 0.00019999822306311006,
-      "loss": 46.0,
-      "step": 24829
-    },
-    {
-      "epoch": 1.8984268975667566,
-      "grad_norm": 0.0013046471867710352,
-      "learning_rate": 0.00019999822291991596,
-      "loss": 46.0,
-      "step": 24830
-    },
-    {
-      "epoch": 1.8985033545501462,
-      "grad_norm": 0.0005455836653709412,
-      "learning_rate": 0.00019999822277671605,
-      "loss": 46.0,
-      "step": 24831
-    },
-    {
-      "epoch": 1.898579811533536,
-      "grad_norm": 0.0008767587132751942,
-      "learning_rate": 0.0001999982226335104,
-      "loss": 46.0,
-      "step": 24832
-    },
-    {
-      "epoch": 1.8986562685169257,
-      "grad_norm": 0.000601877982262522,
-      "learning_rate": 0.00019999822249029898,
-      "loss": 46.0,
-      "step": 24833
-    },
-    {
-      "epoch": 1.8987327255003152,
-      "grad_norm": 0.0016630530590191483,
-      "learning_rate": 0.00019999822234708178,
-      "loss": 46.0,
-      "step": 24834
-    },
-    {
-      "epoch": 1.898809182483705,
-      "grad_norm": 0.003101180773228407,
-      "learning_rate": 0.00019999822220385883,
-      "loss": 46.0,
-      "step": 24835
-    },
-    {
-      "epoch": 1.8988856394670948,
-      "grad_norm": 0.0003369378100614995,
-      "learning_rate": 0.0001999982220606301,
-      "loss": 46.0,
-      "step": 24836
-    },
-    {
-      "epoch": 1.8989620964504845,
-      "grad_norm": 0.0007574019255116582,
-      "learning_rate": 0.00019999822191739557,
-      "loss": 46.0,
-      "step": 24837
-    },
-    {
-      "epoch": 1.8990385534338743,
-      "grad_norm": 0.0008945493609644473,
-      "learning_rate": 0.0001999982217741553,
-      "loss": 46.0,
-      "step": 24838
-    },
-    {
-      "epoch": 1.899115010417264,
-      "grad_norm": 0.0010623873677104712,
-      "learning_rate": 0.00019999822163090927,
-      "loss": 46.0,
-      "step": 24839
-    },
-    {
-      "epoch": 1.8991914674006538,
-      "grad_norm": 0.0005498065147548914,
-      "learning_rate": 0.00019999822148765746,
-      "loss": 46.0,
-      "step": 24840
-    },
-    {
-      "epoch": 1.8992679243840436,
-      "grad_norm": 0.0017968276515603065,
-      "learning_rate": 0.00019999822134439987,
-      "loss": 46.0,
-      "step": 24841
-    },
-    {
-      "epoch": 1.899344381367433,
-      "grad_norm": 0.000995611073449254,
-      "learning_rate": 0.00019999822120113651,
-      "loss": 46.0,
-      "step": 24842
-    },
-    {
-      "epoch": 1.8994208383508229,
-      "grad_norm": 0.001208151807077229,
-      "learning_rate": 0.00019999822105786738,
-      "loss": 46.0,
-      "step": 24843
-    },
-    {
-      "epoch": 1.8994972953342126,
-      "grad_norm": 0.0004636195953935385,
-      "learning_rate": 0.00019999822091459248,
-      "loss": 46.0,
-      "step": 24844
-    },
-    {
-      "epoch": 1.8995737523176022,
-      "grad_norm": 0.0022059523034840822,
-      "learning_rate": 0.00019999822077131185,
-      "loss": 46.0,
-      "step": 24845
-    },
-    {
-      "epoch": 1.899650209300992,
-      "grad_norm": 0.0005472211632877588,
-      "learning_rate": 0.0001999982206280254,
-      "loss": 46.0,
-      "step": 24846
-    },
-    {
-      "epoch": 1.8997266662843817,
-      "grad_norm": 0.0030993276741355658,
-      "learning_rate": 0.00019999822048473323,
-      "loss": 46.0,
-      "step": 24847
-    },
-    {
-      "epoch": 1.8998031232677715,
-      "grad_norm": 0.0005370528087951243,
-      "learning_rate": 0.00019999822034143523,
-      "loss": 46.0,
-      "step": 24848
-    },
-    {
-      "epoch": 1.8998795802511612,
-      "grad_norm": 0.0009516928112134337,
-      "learning_rate": 0.0001999982201981315,
-      "loss": 46.0,
-      "step": 24849
-    },
-    {
-      "epoch": 1.899956037234551,
-      "grad_norm": 0.0011598489945754409,
-      "learning_rate": 0.000199998220054822,
-      "loss": 46.0,
-      "step": 24850
-    },
-    {
-      "epoch": 1.9000324942179407,
-      "grad_norm": 0.0034063542261719704,
-      "learning_rate": 0.00019999821991150673,
-      "loss": 46.0,
-      "step": 24851
-    },
-    {
-      "epoch": 1.9001089512013305,
-      "grad_norm": 0.0018962441245093942,
-      "learning_rate": 0.0001999982197681857,
-      "loss": 46.0,
-      "step": 24852
-    },
-    {
-      "epoch": 1.90018540818472,
-      "grad_norm": 0.0010442601051181555,
-      "learning_rate": 0.00019999821962485889,
-      "loss": 46.0,
-      "step": 24853
-    },
-    {
-      "epoch": 1.9002618651681098,
-      "grad_norm": 0.0017621165607124567,
-      "learning_rate": 0.00019999821948152627,
-      "loss": 46.0,
-      "step": 24854
-    },
-    {
-      "epoch": 1.9003383221514996,
-      "grad_norm": 0.0021643179934471846,
-      "learning_rate": 0.00019999821933818792,
-      "loss": 46.0,
-      "step": 24855
-    },
-    {
-      "epoch": 1.900414779134889,
-      "grad_norm": 0.0007741773733869195,
-      "learning_rate": 0.0001999982191948438,
-      "loss": 46.0,
-      "step": 24856
-    },
-    {
-      "epoch": 1.9004912361182789,
-      "grad_norm": 0.0004004188522230834,
-      "learning_rate": 0.0001999982190514939,
-      "loss": 46.0,
-      "step": 24857
-    },
-    {
-      "epoch": 1.9005676931016686,
-      "grad_norm": 0.00094946613535285,
-      "learning_rate": 0.00019999821890813823,
-      "loss": 46.0,
-      "step": 24858
-    },
-    {
-      "epoch": 1.9006441500850584,
-      "grad_norm": 0.0006203845259733498,
-      "learning_rate": 0.00019999821876477683,
-      "loss": 46.0,
-      "step": 24859
-    },
-    {
-      "epoch": 1.9007206070684481,
-      "grad_norm": 0.004983421880751848,
-      "learning_rate": 0.0001999982186214096,
-      "loss": 46.0,
-      "step": 24860
-    },
-    {
-      "epoch": 1.900797064051838,
-      "grad_norm": 0.0010092385346069932,
-      "learning_rate": 0.00019999821847803664,
-      "loss": 46.0,
-      "step": 24861
-    },
-    {
-      "epoch": 1.9008735210352277,
-      "grad_norm": 0.0009766437578946352,
-      "learning_rate": 0.0001999982183346579,
-      "loss": 46.0,
-      "step": 24862
-    },
-    {
-      "epoch": 1.9009499780186174,
-      "grad_norm": 0.0025472999550402164,
-      "learning_rate": 0.0001999982181912734,
-      "loss": 46.0,
-      "step": 24863
-    },
-    {
-      "epoch": 1.901026435002007,
-      "grad_norm": 0.0005673071136698127,
-      "learning_rate": 0.0001999982180478831,
-      "loss": 46.0,
-      "step": 24864
-    },
-    {
-      "epoch": 1.9011028919853967,
-      "grad_norm": 0.0073205516673624516,
-      "learning_rate": 0.00019999821790448708,
-      "loss": 46.0,
-      "step": 24865
-    },
-    {
-      "epoch": 1.9011793489687865,
-      "grad_norm": 0.0008029459859244525,
-      "learning_rate": 0.00019999821776108524,
-      "loss": 46.0,
-      "step": 24866
-    },
-    {
-      "epoch": 1.901255805952176,
-      "grad_norm": 0.0008055150974541903,
-      "learning_rate": 0.00019999821761767768,
-      "loss": 46.0,
-      "step": 24867
-    },
-    {
-      "epoch": 1.9013322629355658,
-      "grad_norm": 0.00026805861853063107,
-      "learning_rate": 0.00019999821747426432,
-      "loss": 46.0,
-      "step": 24868
-    },
-    {
-      "epoch": 1.9014087199189555,
-      "grad_norm": 0.0007115594926290214,
-      "learning_rate": 0.0001999982173308452,
-      "loss": 46.0,
-      "step": 24869
-    },
-    {
-      "epoch": 1.9014851769023453,
-      "grad_norm": 0.0013798220315948129,
-      "learning_rate": 0.00019999821718742032,
-      "loss": 46.0,
-      "step": 24870
-    },
-    {
-      "epoch": 1.901561633885735,
-      "grad_norm": 0.0007382621406577528,
-      "learning_rate": 0.00019999821704398964,
-      "loss": 46.0,
-      "step": 24871
-    },
-    {
-      "epoch": 1.9016380908691248,
-      "grad_norm": 0.0003800530394073576,
-      "learning_rate": 0.00019999821690055321,
-      "loss": 46.0,
-      "step": 24872
-    },
-    {
-      "epoch": 1.9017145478525146,
-      "grad_norm": 0.0007542367675341666,
-      "learning_rate": 0.00019999821675711102,
-      "loss": 46.0,
-      "step": 24873
-    },
-    {
-      "epoch": 1.9017910048359044,
-      "grad_norm": 0.0026554998476058245,
-      "learning_rate": 0.00019999821661366305,
-      "loss": 46.0,
-      "step": 24874
-    },
-    {
-      "epoch": 1.901867461819294,
-      "grad_norm": 0.000755745975766331,
-      "learning_rate": 0.0001999982164702093,
-      "loss": 46.0,
-      "step": 24875
-    },
-    {
-      "epoch": 1.9019439188026837,
-      "grad_norm": 0.0009823732543736696,
-      "learning_rate": 0.00019999821632674979,
-      "loss": 46.0,
-      "step": 24876
-    },
-    {
-      "epoch": 1.9020203757860734,
-      "grad_norm": 0.005809125956147909,
-      "learning_rate": 0.00019999821618328452,
-      "loss": 46.0,
-      "step": 24877
-    },
-    {
-      "epoch": 1.902096832769463,
-      "grad_norm": 0.0006884276517666876,
-      "learning_rate": 0.00019999821603981346,
-      "loss": 46.0,
-      "step": 24878
-    },
-    {
-      "epoch": 1.9021732897528527,
-      "grad_norm": 0.0004453766450751573,
-      "learning_rate": 0.00019999821589633665,
-      "loss": 46.0,
-      "step": 24879
-    },
-    {
-      "epoch": 1.9022497467362425,
-      "grad_norm": 0.0009323851554654539,
-      "learning_rate": 0.00019999821575285406,
-      "loss": 46.0,
-      "step": 24880
-    },
-    {
-      "epoch": 1.9023262037196322,
-      "grad_norm": 0.0009870842332020402,
-      "learning_rate": 0.00019999821560936568,
-      "loss": 46.0,
-      "step": 24881
-    },
-    {
-      "epoch": 1.902402660703022,
-      "grad_norm": 0.002390071516856551,
-      "learning_rate": 0.00019999821546587158,
-      "loss": 46.0,
-      "step": 24882
-    },
-    {
-      "epoch": 1.9024791176864118,
-      "grad_norm": 0.010729173198342323,
-      "learning_rate": 0.00019999821532237167,
-      "loss": 46.0,
-      "step": 24883
-    },
-    {
-      "epoch": 1.9025555746698015,
-      "grad_norm": 0.0007198489038273692,
-      "learning_rate": 0.00019999821517886602,
-      "loss": 46.0,
-      "step": 24884
-    },
-    {
-      "epoch": 1.9026320316531913,
-      "grad_norm": 0.0005699815810658038,
-      "learning_rate": 0.00019999821503535457,
-      "loss": 46.0,
-      "step": 24885
-    },
-    {
-      "epoch": 1.9027084886365808,
-      "grad_norm": 0.003231907030567527,
-      "learning_rate": 0.00019999821489183738,
-      "loss": 46.0,
-      "step": 24886
-    },
-    {
-      "epoch": 1.9027849456199706,
-      "grad_norm": 0.0007182261324487627,
-      "learning_rate": 0.0001999982147483144,
-      "loss": 46.0,
-      "step": 24887
-    },
-    {
-      "epoch": 1.9028614026033601,
-      "grad_norm": 0.0015613920986652374,
-      "learning_rate": 0.00019999821460478566,
-      "loss": 46.0,
-      "step": 24888
-    },
-    {
-      "epoch": 1.9029378595867499,
-      "grad_norm": 0.0016119254287332296,
-      "learning_rate": 0.00019999821446125114,
-      "loss": 46.0,
-      "step": 24889
-    },
-    {
-      "epoch": 1.9030143165701396,
-      "grad_norm": 0.0007292053196579218,
-      "learning_rate": 0.00019999821431771085,
-      "loss": 46.0,
-      "step": 24890
-    },
-    {
-      "epoch": 1.9030907735535294,
-      "grad_norm": 0.004297601990401745,
-      "learning_rate": 0.00019999821417416482,
-      "loss": 46.0,
-      "step": 24891
-    },
-    {
-      "epoch": 1.9031672305369192,
-      "grad_norm": 0.0006740153767168522,
-      "learning_rate": 0.000199998214030613,
-      "loss": 46.0,
-      "step": 24892
-    },
-    {
-      "epoch": 1.903243687520309,
-      "grad_norm": 0.0009984076023101807,
-      "learning_rate": 0.00019999821388705542,
-      "loss": 46.0,
-      "step": 24893
-    },
-    {
-      "epoch": 1.9033201445036987,
-      "grad_norm": 0.0015975938877090812,
-      "learning_rate": 0.00019999821374349207,
-      "loss": 46.0,
-      "step": 24894
-    },
-    {
-      "epoch": 1.9033966014870884,
-      "grad_norm": 0.001951965969055891,
-      "learning_rate": 0.00019999821359992294,
-      "loss": 46.0,
-      "step": 24895
-    },
-    {
-      "epoch": 1.9034730584704782,
-      "grad_norm": 0.004002948757261038,
-      "learning_rate": 0.00019999821345634803,
-      "loss": 46.0,
-      "step": 24896
-    },
-    {
-      "epoch": 1.9035495154538677,
-      "grad_norm": 0.0011364936362951994,
-      "learning_rate": 0.00019999821331276736,
-      "loss": 46.0,
-      "step": 24897
-    },
-    {
-      "epoch": 1.9036259724372575,
-      "grad_norm": 0.0010299276327714324,
-      "learning_rate": 0.00019999821316918093,
-      "loss": 46.0,
-      "step": 24898
-    },
-    {
-      "epoch": 1.903702429420647,
-      "grad_norm": 0.0002644795167725533,
-      "learning_rate": 0.0001999982130255887,
-      "loss": 46.0,
-      "step": 24899
-    },
-    {
-      "epoch": 1.9037788864040368,
-      "grad_norm": 0.008624094538390636,
-      "learning_rate": 0.00019999821288199074,
-      "loss": 46.0,
-      "step": 24900
-    },
-    {
-      "epoch": 1.9038553433874266,
-      "grad_norm": 0.0006809115293435752,
-      "learning_rate": 0.00019999821273838702,
-      "loss": 46.0,
-      "step": 24901
-    },
-    {
-      "epoch": 1.9039318003708163,
-      "grad_norm": 0.000670730834826827,
-      "learning_rate": 0.0001999982125947775,
-      "loss": 46.0,
-      "step": 24902
-    },
-    {
-      "epoch": 1.904008257354206,
-      "grad_norm": 0.0007364463526755571,
-      "learning_rate": 0.00019999821245116222,
-      "loss": 46.0,
-      "step": 24903
-    },
-    {
-      "epoch": 1.9040847143375959,
-      "grad_norm": 0.0006315343198366463,
-      "learning_rate": 0.00019999821230754115,
-      "loss": 46.0,
-      "step": 24904
-    },
-    {
-      "epoch": 1.9041611713209856,
-      "grad_norm": 0.0012751728063449264,
-      "learning_rate": 0.00019999821216391434,
-      "loss": 46.0,
-      "step": 24905
-    },
-    {
-      "epoch": 1.9042376283043754,
-      "grad_norm": 0.002455499954521656,
-      "learning_rate": 0.00019999821202028176,
-      "loss": 46.0,
-      "step": 24906
-    },
-    {
-      "epoch": 1.9043140852877651,
-      "grad_norm": 0.0006889952928759158,
-      "learning_rate": 0.00019999821187664338,
-      "loss": 46.0,
-      "step": 24907
-    },
-    {
-      "epoch": 1.9043905422711547,
-      "grad_norm": 0.0014112438075244427,
-      "learning_rate": 0.00019999821173299928,
-      "loss": 46.0,
-      "step": 24908
-    },
-    {
-      "epoch": 1.9044669992545444,
-      "grad_norm": 0.0009608845575712621,
-      "learning_rate": 0.00019999821158934937,
-      "loss": 46.0,
-      "step": 24909
-    },
-    {
-      "epoch": 1.904543456237934,
-      "grad_norm": 0.001845090533606708,
-      "learning_rate": 0.0001999982114456937,
-      "loss": 46.0,
-      "step": 24910
-    },
-    {
-      "epoch": 1.9046199132213237,
-      "grad_norm": 0.002541648456826806,
-      "learning_rate": 0.00019999821130203227,
-      "loss": 46.0,
-      "step": 24911
-    },
-    {
-      "epoch": 1.9046963702047135,
-      "grad_norm": 0.0008550227503292263,
-      "learning_rate": 0.00019999821115836505,
-      "loss": 46.0,
-      "step": 24912
-    },
-    {
-      "epoch": 1.9047728271881033,
-      "grad_norm": 0.0008201056625694036,
-      "learning_rate": 0.00019999821101469208,
-      "loss": 46.0,
-      "step": 24913
-    },
-    {
-      "epoch": 1.904849284171493,
-      "grad_norm": 0.001776192570105195,
-      "learning_rate": 0.00019999821087101334,
-      "loss": 46.0,
-      "step": 24914
-    },
-    {
-      "epoch": 1.9049257411548828,
-      "grad_norm": 0.00551563547924161,
-      "learning_rate": 0.00019999821072732882,
-      "loss": 46.0,
-      "step": 24915
-    },
-    {
-      "epoch": 1.9050021981382725,
-      "grad_norm": 0.0006977103766985238,
-      "learning_rate": 0.00019999821058363853,
-      "loss": 46.0,
-      "step": 24916
-    },
-    {
-      "epoch": 1.9050786551216623,
-      "grad_norm": 0.0027677600737661123,
-      "learning_rate": 0.0001999982104399425,
-      "loss": 46.0,
-      "step": 24917
-    },
-    {
-      "epoch": 1.9051551121050518,
-      "grad_norm": 0.0011622656602412462,
-      "learning_rate": 0.00019999821029624066,
-      "loss": 46.0,
-      "step": 24918
-    },
-    {
-      "epoch": 1.9052315690884416,
-      "grad_norm": 0.0006694824551232159,
-      "learning_rate": 0.00019999821015253305,
-      "loss": 46.0,
-      "step": 24919
-    },
-    {
-      "epoch": 1.9053080260718314,
-      "grad_norm": 0.007385769858956337,
-      "learning_rate": 0.00019999821000881972,
-      "loss": 46.0,
-      "step": 24920
-    },
-    {
-      "epoch": 1.905384483055221,
-      "grad_norm": 0.001618484384380281,
-      "learning_rate": 0.0001999982098651006,
-      "loss": 46.0,
-      "step": 24921
-    },
-    {
-      "epoch": 1.9054609400386107,
-      "grad_norm": 0.0033515943214297295,
-      "learning_rate": 0.0001999982097213757,
-      "loss": 46.0,
-      "step": 24922
-    },
-    {
-      "epoch": 1.9055373970220004,
-      "grad_norm": 0.001488825073465705,
-      "learning_rate": 0.000199998209577645,
-      "loss": 46.0,
-      "step": 24923
-    },
-    {
-      "epoch": 1.9056138540053902,
-      "grad_norm": 0.0005886798608116806,
-      "learning_rate": 0.00019999820943390856,
-      "loss": 46.0,
-      "step": 24924
-    },
-    {
-      "epoch": 1.90569031098878,
-      "grad_norm": 0.0011335702147334814,
-      "learning_rate": 0.00019999820929016634,
-      "loss": 46.0,
-      "step": 24925
-    },
-    {
-      "epoch": 1.9057667679721697,
-      "grad_norm": 0.0007771881064400077,
-      "learning_rate": 0.0001999982091464184,
-      "loss": 46.0,
-      "step": 24926
-    },
-    {
-      "epoch": 1.9058432249555595,
-      "grad_norm": 0.0014857404166832566,
-      "learning_rate": 0.00019999820900266463,
-      "loss": 46.0,
-      "step": 24927
-    },
-    {
-      "epoch": 1.9059196819389492,
-      "grad_norm": 0.0017242380417883396,
-      "learning_rate": 0.0001999982088589051,
-      "loss": 46.0,
-      "step": 24928
-    },
-    {
-      "epoch": 1.9059961389223388,
-      "grad_norm": 0.0013279590057209134,
-      "learning_rate": 0.00019999820871513982,
-      "loss": 46.0,
-      "step": 24929
-    },
-    {
-      "epoch": 1.9060725959057285,
-      "grad_norm": 0.0020307907834649086,
-      "learning_rate": 0.00019999820857136879,
-      "loss": 46.0,
-      "step": 24930
-    },
-    {
-      "epoch": 1.9061490528891183,
-      "grad_norm": 0.0022239654790610075,
-      "learning_rate": 0.00019999820842759195,
-      "loss": 46.0,
-      "step": 24931
-    },
-    {
-      "epoch": 1.9062255098725078,
-      "grad_norm": 0.0012013082159683108,
-      "learning_rate": 0.00019999820828380934,
-      "loss": 46.0,
-      "step": 24932
-    },
-    {
-      "epoch": 1.9063019668558976,
-      "grad_norm": 0.0005729928961955011,
-      "learning_rate": 0.00019999820814002098,
-      "loss": 46.0,
-      "step": 24933
-    },
-    {
-      "epoch": 1.9063784238392874,
-      "grad_norm": 0.000442824064521119,
-      "learning_rate": 0.00019999820799622686,
-      "loss": 46.0,
-      "step": 24934
-    },
-    {
-      "epoch": 1.9064548808226771,
-      "grad_norm": 0.00823877565562725,
-      "learning_rate": 0.00019999820785242698,
-      "loss": 46.0,
-      "step": 24935
-    },
-    {
-      "epoch": 1.9065313378060669,
-      "grad_norm": 0.002614240162074566,
-      "learning_rate": 0.0001999982077086213,
-      "loss": 46.0,
-      "step": 24936
-    },
-    {
-      "epoch": 1.9066077947894566,
-      "grad_norm": 0.0010755376424640417,
-      "learning_rate": 0.00019999820756480988,
-      "loss": 46.0,
-      "step": 24937
-    },
-    {
-      "epoch": 1.9066842517728464,
-      "grad_norm": 0.0008267826633527875,
-      "learning_rate": 0.00019999820742099263,
-      "loss": 46.0,
-      "step": 24938
-    },
-    {
-      "epoch": 1.9067607087562362,
-      "grad_norm": 0.001741600688546896,
-      "learning_rate": 0.00019999820727716966,
-      "loss": 46.0,
-      "step": 24939
-    },
-    {
-      "epoch": 1.9068371657396257,
-      "grad_norm": 0.0007187746232375503,
-      "learning_rate": 0.00019999820713334092,
-      "loss": 46.0,
-      "step": 24940
-    },
-    {
-      "epoch": 1.9069136227230155,
-      "grad_norm": 0.0011117152171209455,
-      "learning_rate": 0.00019999820698950643,
-      "loss": 46.0,
-      "step": 24941
-    },
-    {
-      "epoch": 1.9069900797064052,
-      "grad_norm": 0.003406405681744218,
-      "learning_rate": 0.00019999820684566612,
-      "loss": 46.0,
-      "step": 24942
-    },
-    {
-      "epoch": 1.9070665366897948,
-      "grad_norm": 0.004850083030760288,
-      "learning_rate": 0.00019999820670182005,
-      "loss": 46.0,
-      "step": 24943
-    },
-    {
-      "epoch": 1.9071429936731845,
-      "grad_norm": 0.0006139572360552847,
-      "learning_rate": 0.00019999820655796824,
-      "loss": 46.0,
-      "step": 24944
-    },
-    {
-      "epoch": 1.9072194506565743,
-      "grad_norm": 0.002164687728509307,
-      "learning_rate": 0.00019999820641411064,
-      "loss": 46.0,
-      "step": 24945
-    },
-    {
-      "epoch": 1.907295907639964,
-      "grad_norm": 0.0012175774900242686,
-      "learning_rate": 0.00019999820627024728,
-      "loss": 46.0,
-      "step": 24946
-    },
-    {
-      "epoch": 1.9073723646233538,
-      "grad_norm": 0.007076655514538288,
-      "learning_rate": 0.00019999820612637815,
-      "loss": 46.0,
-      "step": 24947
-    },
-    {
-      "epoch": 1.9074488216067436,
-      "grad_norm": 0.005158160347491503,
-      "learning_rate": 0.00019999820598250322,
-      "loss": 46.0,
-      "step": 24948
-    },
-    {
-      "epoch": 1.9075252785901333,
-      "grad_norm": 0.0009839009726420045,
-      "learning_rate": 0.0001999982058386226,
-      "loss": 46.0,
-      "step": 24949
-    },
-    {
-      "epoch": 1.907601735573523,
-      "grad_norm": 0.001031207968480885,
-      "learning_rate": 0.00019999820569473612,
-      "loss": 46.0,
-      "step": 24950
-    },
-    {
-      "epoch": 1.9076781925569126,
-      "grad_norm": 0.001690936041995883,
-      "learning_rate": 0.00019999820555084393,
-      "loss": 46.0,
-      "step": 24951
-    },
-    {
-      "epoch": 1.9077546495403024,
-      "grad_norm": 0.0012634992599487305,
-      "learning_rate": 0.00019999820540694593,
-      "loss": 46.0,
-      "step": 24952
-    },
-    {
-      "epoch": 1.9078311065236921,
-      "grad_norm": 0.000994466245174408,
-      "learning_rate": 0.0001999982052630422,
-      "loss": 46.0,
-      "step": 24953
-    },
-    {
-      "epoch": 1.9079075635070817,
-      "grad_norm": 0.0007749787764623761,
-      "learning_rate": 0.00019999820511913268,
-      "loss": 46.0,
-      "step": 24954
-    },
-    {
-      "epoch": 1.9079840204904714,
-      "grad_norm": 0.001949117286130786,
-      "learning_rate": 0.00019999820497521736,
-      "loss": 46.0,
-      "step": 24955
-    },
-    {
-      "epoch": 1.9080604774738612,
-      "grad_norm": 0.0007108062272891402,
-      "learning_rate": 0.00019999820483129633,
-      "loss": 46.0,
-      "step": 24956
-    },
-    {
-      "epoch": 1.908136934457251,
-      "grad_norm": 0.0007981735398061574,
-      "learning_rate": 0.0001999982046873695,
-      "loss": 46.0,
-      "step": 24957
-    },
-    {
-      "epoch": 1.9082133914406407,
-      "grad_norm": 0.0016814186237752438,
-      "learning_rate": 0.0001999982045434369,
-      "loss": 46.0,
-      "step": 24958
-    },
-    {
-      "epoch": 1.9082898484240305,
-      "grad_norm": 0.0019186069257557392,
-      "learning_rate": 0.00019999820439949853,
-      "loss": 46.0,
-      "step": 24959
-    },
-    {
-      "epoch": 1.9083663054074202,
-      "grad_norm": 0.0014015876222401857,
-      "learning_rate": 0.0001999982042555544,
-      "loss": 46.0,
-      "step": 24960
-    },
-    {
-      "epoch": 1.90844276239081,
-      "grad_norm": 0.0008946965681388974,
-      "learning_rate": 0.0001999982041116045,
-      "loss": 46.0,
-      "step": 24961
-    },
-    {
-      "epoch": 1.9085192193741995,
-      "grad_norm": 0.001257650787010789,
-      "learning_rate": 0.00019999820396764882,
-      "loss": 46.0,
-      "step": 24962
-    },
-    {
-      "epoch": 1.9085956763575893,
-      "grad_norm": 0.0005987566546536982,
-      "learning_rate": 0.0001999982038236874,
-      "loss": 46.0,
-      "step": 24963
-    },
-    {
-      "epoch": 1.908672133340979,
-      "grad_norm": 0.0006889775977469981,
-      "learning_rate": 0.00019999820367972015,
-      "loss": 46.0,
-      "step": 24964
-    },
-    {
-      "epoch": 1.9087485903243686,
-      "grad_norm": 0.010618265718221664,
-      "learning_rate": 0.00019999820353574718,
-      "loss": 46.0,
-      "step": 24965
-    },
-    {
-      "epoch": 1.9088250473077584,
-      "grad_norm": 0.0033406633883714676,
-      "learning_rate": 0.00019999820339176844,
-      "loss": 46.0,
-      "step": 24966
-    },
-    {
-      "epoch": 1.9089015042911481,
-      "grad_norm": 0.002791493898257613,
-      "learning_rate": 0.00019999820324778392,
-      "loss": 46.0,
-      "step": 24967
-    },
-    {
-      "epoch": 1.908977961274538,
-      "grad_norm": 0.0006676879711449146,
-      "learning_rate": 0.00019999820310379364,
-      "loss": 46.0,
-      "step": 24968
-    },
-    {
-      "epoch": 1.9090544182579277,
-      "grad_norm": 0.0012116502039134502,
-      "learning_rate": 0.00019999820295979757,
-      "loss": 46.0,
-      "step": 24969
-    },
-    {
-      "epoch": 1.9091308752413174,
-      "grad_norm": 0.0011264095082879066,
-      "learning_rate": 0.00019999820281579574,
-      "loss": 46.0,
-      "step": 24970
-    },
-    {
-      "epoch": 1.9092073322247072,
-      "grad_norm": 0.0005533446674235165,
-      "learning_rate": 0.00019999820267178816,
-      "loss": 46.0,
-      "step": 24971
-    },
-    {
-      "epoch": 1.909283789208097,
-      "grad_norm": 0.0004319793079048395,
-      "learning_rate": 0.00019999820252777478,
-      "loss": 46.0,
-      "step": 24972
-    },
-    {
-      "epoch": 1.9093602461914865,
-      "grad_norm": 0.0022293655201792717,
-      "learning_rate": 0.00019999820238375565,
-      "loss": 46.0,
-      "step": 24973
-    },
-    {
-      "epoch": 1.9094367031748762,
-      "grad_norm": 0.0009035140392370522,
-      "learning_rate": 0.00019999820223973075,
-      "loss": 46.0,
-      "step": 24974
-    },
-    {
-      "epoch": 1.909513160158266,
-      "grad_norm": 0.0008334499434567988,
-      "learning_rate": 0.00019999820209570004,
-      "loss": 46.0,
-      "step": 24975
-    },
-    {
-      "epoch": 1.9095896171416555,
-      "grad_norm": 0.0012106074718758464,
-      "learning_rate": 0.00019999820195166362,
-      "loss": 46.0,
-      "step": 24976
-    },
-    {
-      "epoch": 1.9096660741250453,
-      "grad_norm": 0.0011463393457233906,
-      "learning_rate": 0.0001999982018076214,
-      "loss": 46.0,
-      "step": 24977
-    },
-    {
-      "epoch": 1.909742531108435,
-      "grad_norm": 0.0005189673393033445,
-      "learning_rate": 0.0001999982016635734,
-      "loss": 46.0,
-      "step": 24978
-    },
-    {
-      "epoch": 1.9098189880918248,
-      "grad_norm": 0.0007190461037680507,
-      "learning_rate": 0.00019999820151951964,
-      "loss": 46.0,
-      "step": 24979
-    },
-    {
-      "epoch": 1.9098954450752146,
-      "grad_norm": 0.003450064454227686,
-      "learning_rate": 0.00019999820137546012,
-      "loss": 46.0,
-      "step": 24980
-    },
-    {
-      "epoch": 1.9099719020586043,
-      "grad_norm": 0.0008244753116741776,
-      "learning_rate": 0.00019999820123139483,
-      "loss": 46.0,
-      "step": 24981
-    },
-    {
-      "epoch": 1.910048359041994,
-      "grad_norm": 0.0006356416270136833,
-      "learning_rate": 0.0001999982010873238,
-      "loss": 46.0,
-      "step": 24982
-    },
-    {
-      "epoch": 1.9101248160253839,
-      "grad_norm": 0.0005068987375125289,
-      "learning_rate": 0.00019999820094324694,
-      "loss": 46.0,
-      "step": 24983
-    },
-    {
-      "epoch": 1.9102012730087734,
-      "grad_norm": 0.0022258213721215725,
-      "learning_rate": 0.00019999820079916436,
-      "loss": 46.0,
-      "step": 24984
-    },
-    {
-      "epoch": 1.9102777299921632,
-      "grad_norm": 0.003178951796144247,
-      "learning_rate": 0.00019999820065507597,
-      "loss": 46.0,
-      "step": 24985
-    },
-    {
-      "epoch": 1.910354186975553,
-      "grad_norm": 0.0010225767036899924,
-      "learning_rate": 0.00019999820051098185,
-      "loss": 46.0,
-      "step": 24986
-    },
-    {
-      "epoch": 1.9104306439589425,
-      "grad_norm": 0.005318465642631054,
-      "learning_rate": 0.00019999820036688192,
-      "loss": 46.0,
-      "step": 24987
-    },
-    {
-      "epoch": 1.9105071009423322,
-      "grad_norm": 0.0025950248818844557,
-      "learning_rate": 0.00019999820022277627,
-      "loss": 46.0,
-      "step": 24988
-    },
-    {
-      "epoch": 1.910583557925722,
-      "grad_norm": 0.006598325911909342,
-      "learning_rate": 0.00019999820007866482,
-      "loss": 46.0,
-      "step": 24989
-    },
-    {
-      "epoch": 1.9106600149091117,
-      "grad_norm": 0.0035286687780171633,
-      "learning_rate": 0.0001999981999345476,
-      "loss": 46.0,
-      "step": 24990
-    },
-    {
-      "epoch": 1.9107364718925015,
-      "grad_norm": 0.0010828207014128566,
-      "learning_rate": 0.0001999981997904246,
-      "loss": 46.0,
-      "step": 24991
-    },
-    {
-      "epoch": 1.9108129288758913,
-      "grad_norm": 0.0004757887218147516,
-      "learning_rate": 0.00019999819964629584,
-      "loss": 46.0,
-      "step": 24992
-    },
-    {
-      "epoch": 1.910889385859281,
-      "grad_norm": 0.003275728551670909,
-      "learning_rate": 0.00019999819950216135,
-      "loss": 46.0,
-      "step": 24993
-    },
-    {
-      "epoch": 1.9109658428426708,
-      "grad_norm": 0.0022997776977717876,
-      "learning_rate": 0.00019999819935802104,
-      "loss": 46.0,
-      "step": 24994
-    },
-    {
-      "epoch": 1.9110422998260603,
-      "grad_norm": 0.0006355869118124247,
-      "learning_rate": 0.000199998199213875,
-      "loss": 46.0,
-      "step": 24995
-    },
-    {
-      "epoch": 1.91111875680945,
-      "grad_norm": 0.000471688574180007,
-      "learning_rate": 0.00019999819906972314,
-      "loss": 46.0,
-      "step": 24996
-    },
-    {
-      "epoch": 1.9111952137928399,
-      "grad_norm": 0.0030428606551140547,
-      "learning_rate": 0.00019999819892556556,
-      "loss": 46.0,
-      "step": 24997
-    },
-    {
-      "epoch": 1.9112716707762294,
-      "grad_norm": 0.0025528762489557266,
-      "learning_rate": 0.00019999819878140218,
-      "loss": 46.0,
-      "step": 24998
-    },
-    {
-      "epoch": 1.9113481277596192,
-      "grad_norm": 0.0012705266708508134,
-      "learning_rate": 0.00019999819863723302,
-      "loss": 46.0,
-      "step": 24999
-    },
-    {
-      "epoch": 1.911424584743009,
-      "grad_norm": 0.004144930746406317,
-      "learning_rate": 0.00019999819849305812,
-      "loss": 46.0,
-      "step": 25000
-    },
-    {
-      "epoch": 1.9115010417263987,
-      "grad_norm": 0.001256758812814951,
-      "learning_rate": 0.00019999819834887745,
-      "loss": 46.0,
-      "step": 25001
-    },
-    {
-      "epoch": 1.9115774987097884,
-      "grad_norm": 0.0040659294463694096,
-      "learning_rate": 0.000199998198204691,
-      "loss": 46.0,
-      "step": 25002
-    },
-    {
-      "epoch": 1.9116539556931782,
-      "grad_norm": 0.0005735154845751822,
-      "learning_rate": 0.0001999981980604988,
-      "loss": 46.0,
-      "step": 25003
-    },
-    {
-      "epoch": 1.911730412676568,
-      "grad_norm": 0.0008928435272537172,
-      "learning_rate": 0.0001999981979163008,
-      "loss": 46.0,
-      "step": 25004
-    },
-    {
-      "epoch": 1.9118068696599577,
-      "grad_norm": 0.002758158603683114,
-      "learning_rate": 0.00019999819777209704,
-      "loss": 46.0,
-      "step": 25005
-    },
-    {
-      "epoch": 1.9118833266433473,
-      "grad_norm": 0.000967782165389508,
-      "learning_rate": 0.00019999819762788753,
-      "loss": 46.0,
-      "step": 25006
-    },
-    {
-      "epoch": 1.911959783626737,
-      "grad_norm": 0.011960156261920929,
-      "learning_rate": 0.00019999819748367222,
-      "loss": 46.0,
-      "step": 25007
-    },
-    {
-      "epoch": 1.9120362406101268,
-      "grad_norm": 0.0032648888882249594,
-      "learning_rate": 0.00019999819733945118,
-      "loss": 46.0,
-      "step": 25008
-    },
-    {
-      "epoch": 1.9121126975935163,
-      "grad_norm": 0.0015277514467015862,
-      "learning_rate": 0.00019999819719522432,
-      "loss": 46.0,
-      "step": 25009
-    },
-    {
-      "epoch": 1.912189154576906,
-      "grad_norm": 0.00117511919233948,
-      "learning_rate": 0.00019999819705099174,
-      "loss": 46.0,
-      "step": 25010
-    },
-    {
-      "epoch": 1.9122656115602958,
-      "grad_norm": 0.0008993755909614265,
-      "learning_rate": 0.00019999819690675333,
-      "loss": 46.0,
-      "step": 25011
-    },
-    {
-      "epoch": 1.9123420685436856,
-      "grad_norm": 0.003055133856832981,
-      "learning_rate": 0.0001999981967625092,
-      "loss": 46.0,
-      "step": 25012
-    },
-    {
-      "epoch": 1.9124185255270754,
-      "grad_norm": 0.006628992967307568,
-      "learning_rate": 0.0001999981966182593,
-      "loss": 46.0,
-      "step": 25013
-    },
-    {
-      "epoch": 1.9124949825104651,
-      "grad_norm": 0.004076141398400068,
-      "learning_rate": 0.00019999819647400363,
-      "loss": 46.0,
-      "step": 25014
-    },
-    {
-      "epoch": 1.9125714394938549,
-      "grad_norm": 0.0019407860236242414,
-      "learning_rate": 0.00019999819632974218,
-      "loss": 46.0,
-      "step": 25015
-    },
-    {
-      "epoch": 1.9126478964772446,
-      "grad_norm": 0.003689304692670703,
-      "learning_rate": 0.00019999819618547496,
-      "loss": 46.0,
-      "step": 25016
-    },
-    {
-      "epoch": 1.9127243534606342,
-      "grad_norm": 0.0009440301801078022,
-      "learning_rate": 0.00019999819604120197,
-      "loss": 46.0,
-      "step": 25017
-    },
-    {
-      "epoch": 1.912800810444024,
-      "grad_norm": 0.0006374557269737124,
-      "learning_rate": 0.0001999981958969232,
-      "loss": 46.0,
-      "step": 25018
-    },
-    {
-      "epoch": 1.9128772674274135,
-      "grad_norm": 0.0015290643787011504,
-      "learning_rate": 0.00019999819575263869,
-      "loss": 46.0,
-      "step": 25019
-    },
-    {
-      "epoch": 1.9129537244108032,
-      "grad_norm": 0.0005443200934678316,
-      "learning_rate": 0.00019999819560834837,
-      "loss": 46.0,
-      "step": 25020
-    },
-    {
-      "epoch": 1.913030181394193,
-      "grad_norm": 0.004492605570703745,
-      "learning_rate": 0.0001999981954640523,
-      "loss": 46.0,
-      "step": 25021
-    },
-    {
-      "epoch": 1.9131066383775828,
-      "grad_norm": 0.0010905354283750057,
-      "learning_rate": 0.0001999981953197505,
-      "loss": 46.0,
-      "step": 25022
-    },
-    {
-      "epoch": 1.9131830953609725,
-      "grad_norm": 0.0005812941235490143,
-      "learning_rate": 0.00019999819517544287,
-      "loss": 46.0,
-      "step": 25023
-    },
-    {
-      "epoch": 1.9132595523443623,
-      "grad_norm": 0.0013756934786215425,
-      "learning_rate": 0.00019999819503112952,
-      "loss": 46.0,
-      "step": 25024
-    },
-    {
-      "epoch": 1.913336009327752,
-      "grad_norm": 0.0013091216096654534,
-      "learning_rate": 0.00019999819488681036,
-      "loss": 46.0,
-      "step": 25025
-    },
-    {
-      "epoch": 1.9134124663111418,
-      "grad_norm": 0.013098536059260368,
-      "learning_rate": 0.00019999819474248546,
-      "loss": 46.0,
-      "step": 25026
-    },
-    {
-      "epoch": 1.9134889232945316,
-      "grad_norm": 0.0010521155782043934,
-      "learning_rate": 0.0001999981945981548,
-      "loss": 46.0,
-      "step": 25027
-    },
-    {
-      "epoch": 1.9135653802779211,
-      "grad_norm": 0.0008318438776768744,
-      "learning_rate": 0.00019999819445381834,
-      "loss": 46.0,
-      "step": 25028
-    },
-    {
-      "epoch": 1.9136418372613109,
-      "grad_norm": 0.0009173999424092472,
-      "learning_rate": 0.0001999981943094761,
-      "loss": 46.0,
-      "step": 25029
-    },
-    {
-      "epoch": 1.9137182942447004,
-      "grad_norm": 0.001803903141990304,
-      "learning_rate": 0.0001999981941651281,
-      "loss": 46.0,
-      "step": 25030
-    },
-    {
-      "epoch": 1.9137947512280902,
-      "grad_norm": 0.001230034395121038,
-      "learning_rate": 0.00019999819402077436,
-      "loss": 46.0,
-      "step": 25031
-    },
-    {
-      "epoch": 1.91387120821148,
-      "grad_norm": 0.002747897757217288,
-      "learning_rate": 0.00019999819387641482,
-      "loss": 46.0,
-      "step": 25032
-    },
-    {
-      "epoch": 1.9139476651948697,
-      "grad_norm": 0.0018156863516196609,
-      "learning_rate": 0.00019999819373204953,
-      "loss": 46.0,
-      "step": 25033
-    },
-    {
-      "epoch": 1.9140241221782595,
-      "grad_norm": 0.0005510213668458164,
-      "learning_rate": 0.00019999819358767847,
-      "loss": 46.0,
-      "step": 25034
-    },
-    {
-      "epoch": 1.9141005791616492,
-      "grad_norm": 0.003328816732391715,
-      "learning_rate": 0.00019999819344330164,
-      "loss": 46.0,
-      "step": 25035
-    },
-    {
-      "epoch": 1.914177036145039,
-      "grad_norm": 0.0007040451164357364,
-      "learning_rate": 0.00019999819329891903,
-      "loss": 46.0,
-      "step": 25036
-    },
-    {
-      "epoch": 1.9142534931284287,
-      "grad_norm": 0.0032834475859999657,
-      "learning_rate": 0.00019999819315453062,
-      "loss": 46.0,
-      "step": 25037
-    },
-    {
-      "epoch": 1.9143299501118185,
-      "grad_norm": 0.0008268989040516317,
-      "learning_rate": 0.0001999981930101365,
-      "loss": 46.0,
-      "step": 25038
-    },
-    {
-      "epoch": 1.914406407095208,
-      "grad_norm": 0.0016562157543376088,
-      "learning_rate": 0.00019999819286573657,
-      "loss": 46.0,
-      "step": 25039
-    },
-    {
-      "epoch": 1.9144828640785978,
-      "grad_norm": 0.0006790382321923971,
-      "learning_rate": 0.0001999981927213309,
-      "loss": 46.0,
-      "step": 25040
-    },
-    {
-      "epoch": 1.9145593210619873,
-      "grad_norm": 0.00041332311229780316,
-      "learning_rate": 0.00019999819257691945,
-      "loss": 46.0,
-      "step": 25041
-    },
-    {
-      "epoch": 1.914635778045377,
-      "grad_norm": 0.0013336250558495522,
-      "learning_rate": 0.00019999819243250223,
-      "loss": 46.0,
-      "step": 25042
-    },
-    {
-      "epoch": 1.9147122350287669,
-      "grad_norm": 0.0009243679814971983,
-      "learning_rate": 0.00019999819228807924,
-      "loss": 46.0,
-      "step": 25043
-    },
-    {
-      "epoch": 1.9147886920121566,
-      "grad_norm": 0.001260233111679554,
-      "learning_rate": 0.00019999819214365047,
-      "loss": 46.0,
-      "step": 25044
-    },
-    {
-      "epoch": 1.9148651489955464,
-      "grad_norm": 0.0013258402468636632,
-      "learning_rate": 0.00019999819199921596,
-      "loss": 46.0,
-      "step": 25045
-    },
-    {
-      "epoch": 1.9149416059789361,
-      "grad_norm": 0.005777197424322367,
-      "learning_rate": 0.00019999819185477567,
-      "loss": 46.0,
-      "step": 25046
-    },
-    {
-      "epoch": 1.915018062962326,
-      "grad_norm": 0.0008911778568290174,
-      "learning_rate": 0.00019999819171032956,
-      "loss": 46.0,
-      "step": 25047
-    },
-    {
-      "epoch": 1.9150945199457157,
-      "grad_norm": 0.0004140009405091405,
-      "learning_rate": 0.00019999819156587772,
-      "loss": 46.0,
-      "step": 25048
-    },
-    {
-      "epoch": 1.9151709769291054,
-      "grad_norm": 0.0008028841693885624,
-      "learning_rate": 0.00019999819142142014,
-      "loss": 46.0,
-      "step": 25049
-    },
-    {
-      "epoch": 1.915247433912495,
-      "grad_norm": 0.0010953544406220317,
-      "learning_rate": 0.00019999819127695676,
-      "loss": 46.0,
-      "step": 25050
-    },
-    {
-      "epoch": 1.9153238908958847,
-      "grad_norm": 0.0015279761282727122,
-      "learning_rate": 0.0001999981911324876,
-      "loss": 46.0,
-      "step": 25051
-    },
-    {
-      "epoch": 1.9154003478792743,
-      "grad_norm": 0.00328050390817225,
-      "learning_rate": 0.00019999819098801268,
-      "loss": 46.0,
-      "step": 25052
-    },
-    {
-      "epoch": 1.915476804862664,
-      "grad_norm": 0.0010097004706040025,
-      "learning_rate": 0.000199998190843532,
-      "loss": 46.0,
-      "step": 25053
-    },
-    {
-      "epoch": 1.9155532618460538,
-      "grad_norm": 0.0005634309491142631,
-      "learning_rate": 0.00019999819069904556,
-      "loss": 46.0,
-      "step": 25054
-    },
-    {
-      "epoch": 1.9156297188294436,
-      "grad_norm": 0.0018211569404229522,
-      "learning_rate": 0.00019999819055455334,
-      "loss": 46.0,
-      "step": 25055
-    },
-    {
-      "epoch": 1.9157061758128333,
-      "grad_norm": 0.0005940868286415935,
-      "learning_rate": 0.00019999819041005535,
-      "loss": 46.0,
-      "step": 25056
-    },
-    {
-      "epoch": 1.915782632796223,
-      "grad_norm": 0.0010370899690315127,
-      "learning_rate": 0.00019999819026555158,
-      "loss": 46.0,
-      "step": 25057
-    },
-    {
-      "epoch": 1.9158590897796128,
-      "grad_norm": 0.0007954982575029135,
-      "learning_rate": 0.00019999819012104207,
-      "loss": 46.0,
-      "step": 25058
-    },
-    {
-      "epoch": 1.9159355467630026,
-      "grad_norm": 0.001185583882033825,
-      "learning_rate": 0.00019999818997652676,
-      "loss": 46.0,
-      "step": 25059
-    },
-    {
-      "epoch": 1.9160120037463921,
-      "grad_norm": 0.0009369342587888241,
-      "learning_rate": 0.00019999818983200567,
-      "loss": 46.0,
-      "step": 25060
-    },
-    {
-      "epoch": 1.916088460729782,
-      "grad_norm": 0.001079972367733717,
-      "learning_rate": 0.00019999818968747884,
-      "loss": 46.0,
-      "step": 25061
-    },
-    {
-      "epoch": 1.9161649177131717,
-      "grad_norm": 0.00123128283303231,
-      "learning_rate": 0.00019999818954294623,
-      "loss": 46.0,
-      "step": 25062
-    },
-    {
-      "epoch": 1.9162413746965612,
-      "grad_norm": 0.0013428755337372422,
-      "learning_rate": 0.00019999818939840785,
-      "loss": 46.0,
-      "step": 25063
-    },
-    {
-      "epoch": 1.916317831679951,
-      "grad_norm": 0.0008612447418272495,
-      "learning_rate": 0.0001999981892538637,
-      "loss": 46.0,
-      "step": 25064
-    },
-    {
-      "epoch": 1.9163942886633407,
-      "grad_norm": 0.0007068425184115767,
-      "learning_rate": 0.00019999818910931377,
-      "loss": 46.0,
-      "step": 25065
-    },
-    {
-      "epoch": 1.9164707456467305,
-      "grad_norm": 0.004639300983399153,
-      "learning_rate": 0.0001999981889647581,
-      "loss": 46.0,
-      "step": 25066
-    },
-    {
-      "epoch": 1.9165472026301202,
-      "grad_norm": 0.003493841737508774,
-      "learning_rate": 0.00019999818882019663,
-      "loss": 46.0,
-      "step": 25067
-    },
-    {
-      "epoch": 1.91662365961351,
-      "grad_norm": 0.0006687366403639317,
-      "learning_rate": 0.00019999818867562943,
-      "loss": 46.0,
-      "step": 25068
-    },
-    {
-      "epoch": 1.9167001165968998,
-      "grad_norm": 0.0004994089249521494,
-      "learning_rate": 0.00019999818853105644,
-      "loss": 46.0,
-      "step": 25069
-    },
-    {
-      "epoch": 1.9167765735802895,
-      "grad_norm": 0.0008875160710886121,
-      "learning_rate": 0.00019999818838647765,
-      "loss": 46.0,
-      "step": 25070
-    },
-    {
-      "epoch": 1.916853030563679,
-      "grad_norm": 0.001617278321646154,
-      "learning_rate": 0.00019999818824189313,
-      "loss": 46.0,
-      "step": 25071
-    },
-    {
-      "epoch": 1.9169294875470688,
-      "grad_norm": 0.002318442100659013,
-      "learning_rate": 0.00019999818809730285,
-      "loss": 46.0,
-      "step": 25072
-    },
-    {
-      "epoch": 1.9170059445304586,
-      "grad_norm": 0.0011374695459380746,
-      "learning_rate": 0.00019999818795270676,
-      "loss": 46.0,
-      "step": 25073
-    },
-    {
-      "epoch": 1.9170824015138481,
-      "grad_norm": 0.0154255460947752,
-      "learning_rate": 0.00019999818780810493,
-      "loss": 46.0,
-      "step": 25074
-    },
-    {
-      "epoch": 1.9171588584972379,
-      "grad_norm": 0.0011985677992925048,
-      "learning_rate": 0.00019999818766349732,
-      "loss": 46.0,
-      "step": 25075
-    },
-    {
-      "epoch": 1.9172353154806276,
-      "grad_norm": 0.001337897963821888,
-      "learning_rate": 0.00019999818751888392,
-      "loss": 46.0,
-      "step": 25076
-    },
-    {
-      "epoch": 1.9173117724640174,
-      "grad_norm": 0.00039542795275337994,
-      "learning_rate": 0.00019999818737426476,
-      "loss": 46.0,
-      "step": 25077
-    },
-    {
-      "epoch": 1.9173882294474072,
-      "grad_norm": 0.00354016968049109,
-      "learning_rate": 0.00019999818722963986,
-      "loss": 46.0,
-      "step": 25078
-    },
-    {
-      "epoch": 1.917464686430797,
-      "grad_norm": 0.0007964569376781583,
-      "learning_rate": 0.00019999818708500917,
-      "loss": 46.0,
-      "step": 25079
-    },
-    {
-      "epoch": 1.9175411434141867,
-      "grad_norm": 0.0013906906824558973,
-      "learning_rate": 0.00019999818694037272,
-      "loss": 46.0,
-      "step": 25080
-    },
-    {
-      "epoch": 1.9176176003975764,
-      "grad_norm": 0.000562442815862596,
-      "learning_rate": 0.00019999818679573047,
-      "loss": 46.0,
-      "step": 25081
-    },
-    {
-      "epoch": 1.917694057380966,
-      "grad_norm": 0.0014158479170873761,
-      "learning_rate": 0.0001999981866510825,
-      "loss": 46.0,
-      "step": 25082
-    },
-    {
-      "epoch": 1.9177705143643557,
-      "grad_norm": 0.00161244987975806,
-      "learning_rate": 0.00019999818650642871,
-      "loss": 46.0,
-      "step": 25083
-    },
-    {
-      "epoch": 1.9178469713477455,
-      "grad_norm": 0.0005394053878262639,
-      "learning_rate": 0.0001999981863617692,
-      "loss": 46.0,
-      "step": 25084
-    },
-    {
-      "epoch": 1.917923428331135,
-      "grad_norm": 0.0004928641719743609,
-      "learning_rate": 0.0001999981862171039,
-      "loss": 46.0,
-      "step": 25085
-    },
-    {
-      "epoch": 1.9179998853145248,
-      "grad_norm": 0.0004640161932911724,
-      "learning_rate": 0.0001999981860724328,
-      "loss": 46.0,
-      "step": 25086
-    },
-    {
-      "epoch": 1.9180763422979146,
-      "grad_norm": 0.0010645025176927447,
-      "learning_rate": 0.000199998185927756,
-      "loss": 46.0,
-      "step": 25087
-    },
-    {
-      "epoch": 1.9181527992813043,
-      "grad_norm": 0.0006481001619249582,
-      "learning_rate": 0.00019999818578307337,
-      "loss": 46.0,
-      "step": 25088
-    },
-    {
-      "epoch": 1.918229256264694,
-      "grad_norm": 0.004494494758546352,
-      "learning_rate": 0.00019999818563838496,
-      "loss": 46.0,
-      "step": 25089
-    },
-    {
-      "epoch": 1.9183057132480839,
-      "grad_norm": 0.003421301953494549,
-      "learning_rate": 0.00019999818549369083,
-      "loss": 46.0,
-      "step": 25090
-    },
-    {
-      "epoch": 1.9183821702314736,
-      "grad_norm": 0.0030701770447194576,
-      "learning_rate": 0.0001999981853489909,
-      "loss": 46.0,
-      "step": 25091
-    },
-    {
-      "epoch": 1.9184586272148634,
-      "grad_norm": 0.0008201185846701264,
-      "learning_rate": 0.0001999981852042852,
-      "loss": 46.0,
-      "step": 25092
-    },
-    {
-      "epoch": 1.918535084198253,
-      "grad_norm": 0.000525744108017534,
-      "learning_rate": 0.00019999818505957376,
-      "loss": 46.0,
-      "step": 25093
-    },
-    {
-      "epoch": 1.9186115411816427,
-      "grad_norm": 0.0007714878884144127,
-      "learning_rate": 0.00019999818491485654,
-      "loss": 46.0,
-      "step": 25094
-    },
-    {
-      "epoch": 1.9186879981650324,
-      "grad_norm": 0.0017069801688194275,
-      "learning_rate": 0.00019999818477013355,
-      "loss": 46.0,
-      "step": 25095
-    },
-    {
-      "epoch": 1.918764455148422,
-      "grad_norm": 0.0008298956672661006,
-      "learning_rate": 0.00019999818462540479,
-      "loss": 46.0,
-      "step": 25096
-    },
-    {
-      "epoch": 1.9188409121318117,
-      "grad_norm": 0.0005693138227798045,
-      "learning_rate": 0.00019999818448067025,
-      "loss": 46.0,
-      "step": 25097
-    },
-    {
-      "epoch": 1.9189173691152015,
-      "grad_norm": 0.003054674481973052,
-      "learning_rate": 0.00019999818433592993,
-      "loss": 46.0,
-      "step": 25098
-    },
-    {
-      "epoch": 1.9189938260985913,
-      "grad_norm": 0.0007443517679348588,
-      "learning_rate": 0.00019999818419118385,
-      "loss": 46.0,
-      "step": 25099
-    },
-    {
-      "epoch": 1.919070283081981,
-      "grad_norm": 0.0006166899693198502,
-      "learning_rate": 0.000199998184046432,
-      "loss": 46.0,
-      "step": 25100
-    },
-    {
-      "epoch": 1.9191467400653708,
-      "grad_norm": 0.0007017873576842248,
-      "learning_rate": 0.00019999818390167438,
-      "loss": 46.0,
-      "step": 25101
-    },
-    {
-      "epoch": 1.9192231970487605,
-      "grad_norm": 0.000554573955014348,
-      "learning_rate": 0.000199998183756911,
-      "loss": 46.0,
-      "step": 25102
-    },
-    {
-      "epoch": 1.9192996540321503,
-      "grad_norm": 0.0008606591145507991,
-      "learning_rate": 0.00019999818361214185,
-      "loss": 46.0,
-      "step": 25103
-    },
-    {
-      "epoch": 1.9193761110155398,
-      "grad_norm": 0.0006248600548133254,
-      "learning_rate": 0.00019999818346736693,
-      "loss": 46.0,
-      "step": 25104
-    },
-    {
-      "epoch": 1.9194525679989296,
-      "grad_norm": 0.0015409360639750957,
-      "learning_rate": 0.00019999818332258626,
-      "loss": 46.0,
-      "step": 25105
-    },
-    {
-      "epoch": 1.9195290249823194,
-      "grad_norm": 0.0007284529856406152,
-      "learning_rate": 0.00019999818317779978,
-      "loss": 46.0,
-      "step": 25106
-    },
-    {
-      "epoch": 1.919605481965709,
-      "grad_norm": 0.0011467261938378215,
-      "learning_rate": 0.00019999818303300754,
-      "loss": 46.0,
-      "step": 25107
-    },
-    {
-      "epoch": 1.9196819389490987,
-      "grad_norm": 0.0019680545665323734,
-      "learning_rate": 0.00019999818288820955,
-      "loss": 46.0,
-      "step": 25108
-    },
-    {
-      "epoch": 1.9197583959324884,
-      "grad_norm": 0.0012068209471181035,
-      "learning_rate": 0.00019999818274340578,
-      "loss": 46.0,
-      "step": 25109
-    },
-    {
-      "epoch": 1.9198348529158782,
-      "grad_norm": 0.002536902204155922,
-      "learning_rate": 0.00019999818259859624,
-      "loss": 46.0,
-      "step": 25110
-    },
-    {
-      "epoch": 1.919911309899268,
-      "grad_norm": 0.0012823395663872361,
-      "learning_rate": 0.00019999818245378096,
-      "loss": 46.0,
-      "step": 25111
-    },
-    {
-      "epoch": 1.9199877668826577,
-      "grad_norm": 0.0014077186351642013,
-      "learning_rate": 0.00019999818230895987,
-      "loss": 46.0,
-      "step": 25112
-    },
-    {
-      "epoch": 1.9200642238660475,
-      "grad_norm": 0.0009194363374263048,
-      "learning_rate": 0.000199998182164133,
-      "loss": 46.0,
-      "step": 25113
-    },
-    {
-      "epoch": 1.9201406808494372,
-      "grad_norm": 0.0006999961915425956,
-      "learning_rate": 0.0001999981820193004,
-      "loss": 46.0,
-      "step": 25114
-    },
-    {
-      "epoch": 1.9202171378328268,
-      "grad_norm": 0.001703887595795095,
-      "learning_rate": 0.00019999818187446203,
-      "loss": 46.0,
-      "step": 25115
-    },
-    {
-      "epoch": 1.9202935948162165,
-      "grad_norm": 0.0011886197607964277,
-      "learning_rate": 0.00019999818172961788,
-      "loss": 46.0,
-      "step": 25116
-    },
-    {
-      "epoch": 1.9203700517996063,
-      "grad_norm": 0.0013008458772674203,
-      "learning_rate": 0.00019999818158476795,
-      "loss": 46.0,
-      "step": 25117
-    },
-    {
-      "epoch": 1.9204465087829958,
-      "grad_norm": 0.001087557291612029,
-      "learning_rate": 0.00019999818143991225,
-      "loss": 46.0,
-      "step": 25118
-    },
-    {
-      "epoch": 1.9205229657663856,
-      "grad_norm": 0.0013504237867891788,
-      "learning_rate": 0.00019999818129505078,
-      "loss": 46.0,
-      "step": 25119
-    },
-    {
-      "epoch": 1.9205994227497754,
-      "grad_norm": 0.007073573302477598,
-      "learning_rate": 0.00019999818115018356,
-      "loss": 46.0,
-      "step": 25120
-    },
-    {
-      "epoch": 1.9206758797331651,
-      "grad_norm": 0.003233939642086625,
-      "learning_rate": 0.00019999818100531057,
-      "loss": 46.0,
-      "step": 25121
-    },
-    {
-      "epoch": 1.9207523367165549,
-      "grad_norm": 0.0009536903235130012,
-      "learning_rate": 0.00019999818086043178,
-      "loss": 46.0,
-      "step": 25122
-    },
-    {
-      "epoch": 1.9208287936999446,
-      "grad_norm": 0.0018891231156885624,
-      "learning_rate": 0.00019999818071554727,
-      "loss": 46.0,
-      "step": 25123
-    },
-    {
-      "epoch": 1.9209052506833344,
-      "grad_norm": 0.0009204610832966864,
-      "learning_rate": 0.00019999818057065696,
-      "loss": 46.0,
-      "step": 25124
-    },
-    {
-      "epoch": 1.9209817076667242,
-      "grad_norm": 0.00030159568996168673,
-      "learning_rate": 0.00019999818042576087,
-      "loss": 46.0,
-      "step": 25125
-    },
-    {
-      "epoch": 1.9210581646501137,
-      "grad_norm": 0.0007492768345400691,
-      "learning_rate": 0.000199998180280859,
-      "loss": 46.0,
-      "step": 25126
-    },
-    {
-      "epoch": 1.9211346216335035,
-      "grad_norm": 0.0015603062929585576,
-      "learning_rate": 0.0001999981801359514,
-      "loss": 46.0,
-      "step": 25127
-    },
-    {
-      "epoch": 1.9212110786168932,
-      "grad_norm": 0.0006441359873861074,
-      "learning_rate": 0.00019999817999103803,
-      "loss": 46.0,
-      "step": 25128
-    },
-    {
-      "epoch": 1.9212875356002828,
-      "grad_norm": 0.0012904421892017126,
-      "learning_rate": 0.00019999817984611885,
-      "loss": 46.0,
-      "step": 25129
-    },
-    {
-      "epoch": 1.9213639925836725,
-      "grad_norm": 0.0012758687371388078,
-      "learning_rate": 0.00019999817970119395,
-      "loss": 46.0,
-      "step": 25130
-    },
-    {
-      "epoch": 1.9214404495670623,
-      "grad_norm": 0.007401332724839449,
-      "learning_rate": 0.00019999817955626323,
-      "loss": 46.0,
-      "step": 25131
-    },
-    {
-      "epoch": 1.921516906550452,
-      "grad_norm": 0.008546171709895134,
-      "learning_rate": 0.00019999817941132676,
-      "loss": 46.0,
-      "step": 25132
-    },
-    {
-      "epoch": 1.9215933635338418,
-      "grad_norm": 0.0012078093132004142,
-      "learning_rate": 0.00019999817926638454,
-      "loss": 46.0,
-      "step": 25133
-    },
-    {
-      "epoch": 1.9216698205172316,
-      "grad_norm": 0.0023466809652745724,
-      "learning_rate": 0.00019999817912143655,
-      "loss": 46.0,
-      "step": 25134
-    },
-    {
-      "epoch": 1.9217462775006213,
-      "grad_norm": 0.0008031139732338488,
-      "learning_rate": 0.00019999817897648278,
-      "loss": 46.0,
-      "step": 25135
-    },
-    {
-      "epoch": 1.921822734484011,
-      "grad_norm": 0.0016933862352743745,
-      "learning_rate": 0.00019999817883152324,
-      "loss": 46.0,
-      "step": 25136
-    },
-    {
-      "epoch": 1.9218991914674006,
-      "grad_norm": 0.0009341283002868295,
-      "learning_rate": 0.00019999817868655793,
-      "loss": 46.0,
-      "step": 25137
-    },
-    {
-      "epoch": 1.9219756484507904,
-      "grad_norm": 0.0019462230848148465,
-      "learning_rate": 0.00019999817854158685,
-      "loss": 46.0,
-      "step": 25138
-    },
-    {
-      "epoch": 1.9220521054341801,
-      "grad_norm": 0.0020560200791805983,
-      "learning_rate": 0.00019999817839661,
-      "loss": 46.0,
-      "step": 25139
-    },
-    {
-      "epoch": 1.9221285624175697,
-      "grad_norm": 0.0008402771200053394,
-      "learning_rate": 0.00019999817825162736,
-      "loss": 46.0,
-      "step": 25140
-    },
-    {
-      "epoch": 1.9222050194009594,
-      "grad_norm": 0.0008873976767063141,
-      "learning_rate": 0.00019999817810663898,
-      "loss": 46.0,
-      "step": 25141
-    },
-    {
-      "epoch": 1.9222814763843492,
-      "grad_norm": 0.0012485863408073783,
-      "learning_rate": 0.00019999817796164483,
-      "loss": 46.0,
-      "step": 25142
-    },
-    {
-      "epoch": 1.922357933367739,
-      "grad_norm": 0.0027492486406117678,
-      "learning_rate": 0.0001999981778166449,
-      "loss": 46.0,
-      "step": 25143
-    },
-    {
-      "epoch": 1.9224343903511287,
-      "grad_norm": 0.0022691525518894196,
-      "learning_rate": 0.0001999981776716392,
-      "loss": 46.0,
-      "step": 25144
-    },
-    {
-      "epoch": 1.9225108473345185,
-      "grad_norm": 0.0005444939597509801,
-      "learning_rate": 0.00019999817752662776,
-      "loss": 46.0,
-      "step": 25145
-    },
-    {
-      "epoch": 1.9225873043179083,
-      "grad_norm": 0.0011389129795134068,
-      "learning_rate": 0.00019999817738161052,
-      "loss": 46.0,
-      "step": 25146
-    },
-    {
-      "epoch": 1.922663761301298,
-      "grad_norm": 0.0013293647207319736,
-      "learning_rate": 0.00019999817723658753,
-      "loss": 46.0,
-      "step": 25147
-    },
-    {
-      "epoch": 1.9227402182846876,
-      "grad_norm": 0.005335854832082987,
-      "learning_rate": 0.00019999817709155873,
-      "loss": 46.0,
-      "step": 25148
-    },
-    {
-      "epoch": 1.9228166752680773,
-      "grad_norm": 0.0017869733273983002,
-      "learning_rate": 0.0001999981769465242,
-      "loss": 46.0,
-      "step": 25149
-    },
-    {
-      "epoch": 1.9228931322514669,
-      "grad_norm": 0.013788530603051186,
-      "learning_rate": 0.00019999817680148389,
-      "loss": 46.0,
-      "step": 25150
-    },
-    {
-      "epoch": 1.9229695892348566,
-      "grad_norm": 0.012004577554762363,
-      "learning_rate": 0.0001999981766564378,
-      "loss": 46.0,
-      "step": 25151
-    },
-    {
-      "epoch": 1.9230460462182464,
-      "grad_norm": 0.0015379858668893576,
-      "learning_rate": 0.00019999817651138594,
-      "loss": 46.0,
-      "step": 25152
-    },
-    {
-      "epoch": 1.9231225032016361,
-      "grad_norm": 0.0018959281733259559,
-      "learning_rate": 0.00019999817636632834,
-      "loss": 46.0,
-      "step": 25153
-    },
-    {
-      "epoch": 1.923198960185026,
-      "grad_norm": 0.0007205240544863045,
-      "learning_rate": 0.00019999817622126496,
-      "loss": 46.0,
-      "step": 25154
-    },
-    {
-      "epoch": 1.9232754171684157,
-      "grad_norm": 0.00045765176764689386,
-      "learning_rate": 0.0001999981760761958,
-      "loss": 46.0,
-      "step": 25155
-    },
-    {
-      "epoch": 1.9233518741518054,
-      "grad_norm": 0.0005802412633784115,
-      "learning_rate": 0.00019999817593112086,
-      "loss": 46.0,
-      "step": 25156
-    },
-    {
-      "epoch": 1.9234283311351952,
-      "grad_norm": 0.002795987529680133,
-      "learning_rate": 0.00019999817578604016,
-      "loss": 46.0,
-      "step": 25157
-    },
-    {
-      "epoch": 1.923504788118585,
-      "grad_norm": 0.0007610914763063192,
-      "learning_rate": 0.0001999981756409537,
-      "loss": 46.0,
-      "step": 25158
-    },
-    {
-      "epoch": 1.9235812451019745,
-      "grad_norm": 0.000974387745372951,
-      "learning_rate": 0.00019999817549586147,
-      "loss": 46.0,
-      "step": 25159
-    },
-    {
-      "epoch": 1.9236577020853642,
-      "grad_norm": 0.008340018801391125,
-      "learning_rate": 0.00019999817535076346,
-      "loss": 46.0,
-      "step": 25160
-    },
-    {
-      "epoch": 1.9237341590687538,
-      "grad_norm": 0.0003882547316607088,
-      "learning_rate": 0.00019999817520565966,
-      "loss": 46.0,
-      "step": 25161
-    },
-    {
-      "epoch": 1.9238106160521435,
-      "grad_norm": 0.0033759695943444967,
-      "learning_rate": 0.00019999817506055015,
-      "loss": 46.0,
-      "step": 25162
-    },
-    {
-      "epoch": 1.9238870730355333,
-      "grad_norm": 0.0009988202946260571,
-      "learning_rate": 0.00019999817491543482,
-      "loss": 46.0,
-      "step": 25163
-    },
-    {
-      "epoch": 1.923963530018923,
-      "grad_norm": 0.0021684356033802032,
-      "learning_rate": 0.00019999817477031373,
-      "loss": 46.0,
-      "step": 25164
-    },
-    {
-      "epoch": 1.9240399870023128,
-      "grad_norm": 0.0022273636423051357,
-      "learning_rate": 0.00019999817462518687,
-      "loss": 46.0,
-      "step": 25165
-    },
-    {
-      "epoch": 1.9241164439857026,
-      "grad_norm": 0.0047491323202848434,
-      "learning_rate": 0.00019999817448005427,
-      "loss": 46.0,
-      "step": 25166
-    },
-    {
-      "epoch": 1.9241929009690923,
-      "grad_norm": 0.0005047345766797662,
-      "learning_rate": 0.00019999817433491587,
-      "loss": 46.0,
-      "step": 25167
-    },
-    {
-      "epoch": 1.924269357952482,
-      "grad_norm": 0.0005172788514755666,
-      "learning_rate": 0.00019999817418977172,
-      "loss": 46.0,
-      "step": 25168
-    },
-    {
-      "epoch": 1.9243458149358719,
-      "grad_norm": 0.0019460206385701895,
-      "learning_rate": 0.00019999817404462176,
-      "loss": 46.0,
-      "step": 25169
-    },
-    {
-      "epoch": 1.9244222719192614,
-      "grad_norm": 0.0007274834788404405,
-      "learning_rate": 0.0001999981738994661,
-      "loss": 46.0,
-      "step": 25170
-    },
-    {
-      "epoch": 1.9244987289026512,
-      "grad_norm": 0.0011231759563088417,
-      "learning_rate": 0.00019999817375430462,
-      "loss": 46.0,
-      "step": 25171
-    },
-    {
-      "epoch": 1.9245751858860407,
-      "grad_norm": 0.0008739459444768727,
-      "learning_rate": 0.00019999817360913738,
-      "loss": 46.0,
-      "step": 25172
-    },
-    {
-      "epoch": 1.9246516428694305,
-      "grad_norm": 0.0007187564042396843,
-      "learning_rate": 0.0001999981734639644,
-      "loss": 46.0,
-      "step": 25173
-    },
-    {
-      "epoch": 1.9247280998528202,
-      "grad_norm": 0.0009323000558651984,
-      "learning_rate": 0.0001999981733187856,
-      "loss": 46.0,
-      "step": 25174
-    },
-    {
-      "epoch": 1.92480455683621,
-      "grad_norm": 0.001003554556518793,
-      "learning_rate": 0.00019999817317360106,
-      "loss": 46.0,
-      "step": 25175
-    },
-    {
-      "epoch": 1.9248810138195998,
-      "grad_norm": 0.004511723760515451,
-      "learning_rate": 0.00019999817302841075,
-      "loss": 46.0,
-      "step": 25176
-    },
-    {
-      "epoch": 1.9249574708029895,
-      "grad_norm": 0.000644420797470957,
-      "learning_rate": 0.00019999817288321467,
-      "loss": 46.0,
-      "step": 25177
-    },
-    {
-      "epoch": 1.9250339277863793,
-      "grad_norm": 0.0012313324259594083,
-      "learning_rate": 0.00019999817273801284,
-      "loss": 46.0,
-      "step": 25178
-    },
-    {
-      "epoch": 1.925110384769769,
-      "grad_norm": 0.0008592989179305732,
-      "learning_rate": 0.0001999981725928052,
-      "loss": 46.0,
-      "step": 25179
-    },
-    {
-      "epoch": 1.9251868417531588,
-      "grad_norm": 0.0007877004682086408,
-      "learning_rate": 0.0001999981724475918,
-      "loss": 46.0,
-      "step": 25180
-    },
-    {
-      "epoch": 1.9252632987365483,
-      "grad_norm": 0.0006632974254898727,
-      "learning_rate": 0.00019999817230237262,
-      "loss": 46.0,
-      "step": 25181
-    },
-    {
-      "epoch": 1.925339755719938,
-      "grad_norm": 0.0011624970939010382,
-      "learning_rate": 0.0001999981721571477,
-      "loss": 46.0,
-      "step": 25182
-    },
-    {
-      "epoch": 1.9254162127033276,
-      "grad_norm": 0.0034205024130642414,
-      "learning_rate": 0.000199998172011917,
-      "loss": 46.0,
-      "step": 25183
-    },
-    {
-      "epoch": 1.9254926696867174,
-      "grad_norm": 0.003955758176743984,
-      "learning_rate": 0.00019999817186668053,
-      "loss": 46.0,
-      "step": 25184
-    },
-    {
-      "epoch": 1.9255691266701072,
-      "grad_norm": 0.0015583012718707323,
-      "learning_rate": 0.0001999981717214383,
-      "loss": 46.0,
-      "step": 25185
-    },
-    {
-      "epoch": 1.925645583653497,
-      "grad_norm": 0.0016653392231091857,
-      "learning_rate": 0.00019999817157619027,
-      "loss": 46.0,
-      "step": 25186
-    },
-    {
-      "epoch": 1.9257220406368867,
-      "grad_norm": 0.0013273786753416061,
-      "learning_rate": 0.0001999981714309365,
-      "loss": 46.0,
-      "step": 25187
-    },
-    {
-      "epoch": 1.9257984976202764,
-      "grad_norm": 0.001043940195813775,
-      "learning_rate": 0.00019999817128567697,
-      "loss": 46.0,
-      "step": 25188
-    },
-    {
-      "epoch": 1.9258749546036662,
-      "grad_norm": 0.001366392825730145,
-      "learning_rate": 0.00019999817114041163,
-      "loss": 46.0,
-      "step": 25189
-    },
-    {
-      "epoch": 1.925951411587056,
-      "grad_norm": 0.001967476913705468,
-      "learning_rate": 0.00019999817099514055,
-      "loss": 46.0,
-      "step": 25190
-    },
-    {
-      "epoch": 1.9260278685704455,
-      "grad_norm": 0.0010032574646174908,
-      "learning_rate": 0.00019999817084986372,
-      "loss": 46.0,
-      "step": 25191
-    },
-    {
-      "epoch": 1.9261043255538353,
-      "grad_norm": 0.006519900634884834,
-      "learning_rate": 0.0001999981707045811,
-      "loss": 46.0,
-      "step": 25192
-    },
-    {
-      "epoch": 1.926180782537225,
-      "grad_norm": 0.00022109072597231716,
-      "learning_rate": 0.0001999981705592927,
-      "loss": 46.0,
-      "step": 25193
-    },
-    {
-      "epoch": 1.9262572395206146,
-      "grad_norm": 0.0020203767344355583,
-      "learning_rate": 0.0001999981704139985,
-      "loss": 46.0,
-      "step": 25194
-    },
-    {
-      "epoch": 1.9263336965040043,
-      "grad_norm": 0.0006868203636258841,
-      "learning_rate": 0.00019999817026869861,
-      "loss": 46.0,
-      "step": 25195
-    },
-    {
-      "epoch": 1.926410153487394,
-      "grad_norm": 0.0006123451748862863,
-      "learning_rate": 0.0001999981701233929,
-      "loss": 46.0,
-      "step": 25196
-    },
-    {
-      "epoch": 1.9264866104707838,
-      "grad_norm": 0.0004660837585106492,
-      "learning_rate": 0.00019999816997808142,
-      "loss": 46.0,
-      "step": 25197
-    },
-    {
-      "epoch": 1.9265630674541736,
-      "grad_norm": 0.0010515452595427632,
-      "learning_rate": 0.00019999816983276418,
-      "loss": 46.0,
-      "step": 25198
-    },
-    {
-      "epoch": 1.9266395244375634,
-      "grad_norm": 0.0011153577361255884,
-      "learning_rate": 0.00019999816968744116,
-      "loss": 46.0,
-      "step": 25199
-    },
-    {
-      "epoch": 1.9267159814209531,
-      "grad_norm": 0.004239494446665049,
-      "learning_rate": 0.0001999981695421124,
-      "loss": 46.0,
-      "step": 25200
-    },
-    {
-      "epoch": 1.9267924384043429,
-      "grad_norm": 0.001552197034470737,
-      "learning_rate": 0.00019999816939677783,
-      "loss": 46.0,
-      "step": 25201
-    },
-    {
-      "epoch": 1.9268688953877324,
-      "grad_norm": 0.0008156081312336028,
-      "learning_rate": 0.00019999816925143752,
-      "loss": 46.0,
-      "step": 25202
-    },
-    {
-      "epoch": 1.9269453523711222,
-      "grad_norm": 0.0005057839443907142,
-      "learning_rate": 0.00019999816910609144,
-      "loss": 46.0,
-      "step": 25203
-    },
-    {
-      "epoch": 1.927021809354512,
-      "grad_norm": 0.000987923238426447,
-      "learning_rate": 0.00019999816896073958,
-      "loss": 46.0,
-      "step": 25204
-    },
-    {
-      "epoch": 1.9270982663379015,
-      "grad_norm": 0.0010984322289004922,
-      "learning_rate": 0.00019999816881538195,
-      "loss": 46.0,
-      "step": 25205
-    },
-    {
-      "epoch": 1.9271747233212912,
-      "grad_norm": 0.0022356316912919283,
-      "learning_rate": 0.00019999816867001858,
-      "loss": 46.0,
-      "step": 25206
-    },
-    {
-      "epoch": 1.927251180304681,
-      "grad_norm": 0.004303027410060167,
-      "learning_rate": 0.0001999981685246494,
-      "loss": 46.0,
-      "step": 25207
-    },
-    {
-      "epoch": 1.9273276372880708,
-      "grad_norm": 0.0006896471022628248,
-      "learning_rate": 0.00019999816837927448,
-      "loss": 46.0,
-      "step": 25208
-    },
-    {
-      "epoch": 1.9274040942714605,
-      "grad_norm": 0.005964044947177172,
-      "learning_rate": 0.00019999816823389375,
-      "loss": 46.0,
-      "step": 25209
-    },
-    {
-      "epoch": 1.9274805512548503,
-      "grad_norm": 0.003415411338210106,
-      "learning_rate": 0.00019999816808850728,
-      "loss": 46.0,
-      "step": 25210
-    },
-    {
-      "epoch": 1.92755700823824,
-      "grad_norm": 0.002570156008005142,
-      "learning_rate": 0.00019999816794311507,
-      "loss": 46.0,
-      "step": 25211
-    },
-    {
-      "epoch": 1.9276334652216298,
-      "grad_norm": 0.0009039311553351581,
-      "learning_rate": 0.00019999816779771702,
-      "loss": 46.0,
-      "step": 25212
-    },
-    {
-      "epoch": 1.9277099222050194,
-      "grad_norm": 0.0007681567803956568,
-      "learning_rate": 0.00019999816765231326,
-      "loss": 46.0,
-      "step": 25213
-    },
-    {
-      "epoch": 1.9277863791884091,
-      "grad_norm": 0.0013136455090716481,
-      "learning_rate": 0.0001999981675069037,
-      "loss": 46.0,
-      "step": 25214
-    },
-    {
-      "epoch": 1.9278628361717989,
-      "grad_norm": 0.0033056868705898523,
-      "learning_rate": 0.0001999981673614884,
-      "loss": 46.0,
-      "step": 25215
-    },
-    {
-      "epoch": 1.9279392931551884,
-      "grad_norm": 0.0025418614968657494,
-      "learning_rate": 0.0001999981672160673,
-      "loss": 46.0,
-      "step": 25216
-    },
-    {
-      "epoch": 1.9280157501385782,
-      "grad_norm": 0.0019595385529100895,
-      "learning_rate": 0.00019999816707064042,
-      "loss": 46.0,
-      "step": 25217
-    },
-    {
-      "epoch": 1.928092207121968,
-      "grad_norm": 0.001870792475529015,
-      "learning_rate": 0.00019999816692520782,
-      "loss": 46.0,
-      "step": 25218
-    },
-    {
-      "epoch": 1.9281686641053577,
-      "grad_norm": 0.0021238909102976322,
-      "learning_rate": 0.00019999816677976942,
-      "loss": 46.0,
-      "step": 25219
-    },
-    {
-      "epoch": 1.9282451210887475,
-      "grad_norm": 0.0005851304158568382,
-      "learning_rate": 0.00019999816663432524,
-      "loss": 46.0,
-      "step": 25220
-    },
-    {
-      "epoch": 1.9283215780721372,
-      "grad_norm": 0.004765552934259176,
-      "learning_rate": 0.00019999816648887532,
-      "loss": 46.0,
-      "step": 25221
-    },
-    {
-      "epoch": 1.928398035055527,
-      "grad_norm": 0.0067898607812821865,
-      "learning_rate": 0.00019999816634341962,
-      "loss": 46.0,
-      "step": 25222
-    },
-    {
-      "epoch": 1.9284744920389167,
-      "grad_norm": 0.0009244895190931857,
-      "learning_rate": 0.00019999816619795815,
-      "loss": 46.0,
-      "step": 25223
-    },
-    {
-      "epoch": 1.9285509490223063,
-      "grad_norm": 0.0011424466501921415,
-      "learning_rate": 0.00019999816605249088,
-      "loss": 46.0,
-      "step": 25224
-    },
-    {
-      "epoch": 1.928627406005696,
-      "grad_norm": 0.0013715140521526337,
-      "learning_rate": 0.00019999816590701787,
-      "loss": 46.0,
-      "step": 25225
-    },
-    {
-      "epoch": 1.9287038629890858,
-      "grad_norm": 0.002194358967244625,
-      "learning_rate": 0.0001999981657615391,
-      "loss": 46.0,
-      "step": 25226
-    },
-    {
-      "epoch": 1.9287803199724753,
-      "grad_norm": 0.005988112650811672,
-      "learning_rate": 0.00019999816561605457,
-      "loss": 46.0,
-      "step": 25227
-    },
-    {
-      "epoch": 1.928856776955865,
-      "grad_norm": 0.0037977169267833233,
-      "learning_rate": 0.00019999816547056423,
-      "loss": 46.0,
-      "step": 25228
-    },
-    {
-      "epoch": 1.9289332339392549,
-      "grad_norm": 0.0004668365581892431,
-      "learning_rate": 0.00019999816532506812,
-      "loss": 46.0,
-      "step": 25229
-    },
-    {
-      "epoch": 1.9290096909226446,
-      "grad_norm": 0.0006393849616870284,
-      "learning_rate": 0.0001999981651795663,
-      "loss": 46.0,
-      "step": 25230
-    },
-    {
-      "epoch": 1.9290861479060344,
-      "grad_norm": 0.0017548591131344438,
-      "learning_rate": 0.00019999816503405864,
-      "loss": 46.0,
-      "step": 25231
-    },
-    {
-      "epoch": 1.9291626048894241,
-      "grad_norm": 0.0009567700326442719,
-      "learning_rate": 0.00019999816488854526,
-      "loss": 46.0,
-      "step": 25232
-    },
-    {
-      "epoch": 1.929239061872814,
-      "grad_norm": 0.0019936037715524435,
-      "learning_rate": 0.00019999816474302608,
-      "loss": 46.0,
-      "step": 25233
-    },
-    {
-      "epoch": 1.9293155188562037,
-      "grad_norm": 0.0008094929507933557,
-      "learning_rate": 0.00019999816459750113,
-      "loss": 46.0,
-      "step": 25234
-    },
-    {
-      "epoch": 1.9293919758395932,
-      "grad_norm": 0.0012513125548139215,
-      "learning_rate": 0.00019999816445197044,
-      "loss": 46.0,
-      "step": 25235
-    },
-    {
-      "epoch": 1.929468432822983,
-      "grad_norm": 0.0006199326599016786,
-      "learning_rate": 0.00019999816430643397,
-      "loss": 46.0,
-      "step": 25236
-    },
-    {
-      "epoch": 1.9295448898063727,
-      "grad_norm": 0.000641452323179692,
-      "learning_rate": 0.00019999816416089173,
-      "loss": 46.0,
-      "step": 25237
-    },
-    {
-      "epoch": 1.9296213467897623,
-      "grad_norm": 0.0013919707853347063,
-      "learning_rate": 0.0001999981640153437,
-      "loss": 46.0,
-      "step": 25238
-    },
-    {
-      "epoch": 1.929697803773152,
-      "grad_norm": 0.0026501931715756655,
-      "learning_rate": 0.00019999816386978992,
-      "loss": 46.0,
-      "step": 25239
-    },
-    {
-      "epoch": 1.9297742607565418,
-      "grad_norm": 0.0010532482992857695,
-      "learning_rate": 0.00019999816372423036,
-      "loss": 46.0,
-      "step": 25240
-    },
-    {
-      "epoch": 1.9298507177399316,
-      "grad_norm": 0.0013078592019155622,
-      "learning_rate": 0.00019999816357866505,
-      "loss": 46.0,
-      "step": 25241
-    },
-    {
-      "epoch": 1.9299271747233213,
-      "grad_norm": 0.008671666495501995,
-      "learning_rate": 0.00019999816343309394,
-      "loss": 46.0,
-      "step": 25242
-    },
-    {
-      "epoch": 1.930003631706711,
-      "grad_norm": 0.0004087663546670228,
-      "learning_rate": 0.0001999981632875171,
-      "loss": 46.0,
-      "step": 25243
-    },
-    {
-      "epoch": 1.9300800886901008,
-      "grad_norm": 0.00042735011084005237,
-      "learning_rate": 0.00019999816314193446,
-      "loss": 46.0,
-      "step": 25244
-    },
-    {
-      "epoch": 1.9301565456734906,
-      "grad_norm": 0.0017284011701121926,
-      "learning_rate": 0.00019999816299634605,
-      "loss": 46.0,
-      "step": 25245
-    },
-    {
-      "epoch": 1.9302330026568801,
-      "grad_norm": 0.0013545791152864695,
-      "learning_rate": 0.0001999981628507519,
-      "loss": 46.0,
-      "step": 25246
-    },
-    {
-      "epoch": 1.93030945964027,
-      "grad_norm": 0.000881209853105247,
-      "learning_rate": 0.00019999816270515198,
-      "loss": 46.0,
-      "step": 25247
-    },
-    {
-      "epoch": 1.9303859166236597,
-      "grad_norm": 0.0009371745400130749,
-      "learning_rate": 0.00019999816255954623,
-      "loss": 46.0,
-      "step": 25248
-    },
-    {
-      "epoch": 1.9304623736070492,
-      "grad_norm": 0.0008964454173110425,
-      "learning_rate": 0.00019999816241393476,
-      "loss": 46.0,
-      "step": 25249
-    },
-    {
-      "epoch": 1.930538830590439,
-      "grad_norm": 0.0014341735513880849,
-      "learning_rate": 0.00019999816226831752,
-      "loss": 46.0,
-      "step": 25250
-    },
-    {
-      "epoch": 1.9306152875738287,
-      "grad_norm": 0.0037198318168520927,
-      "learning_rate": 0.0001999981621226945,
-      "loss": 46.0,
-      "step": 25251
-    },
-    {
-      "epoch": 1.9306917445572185,
-      "grad_norm": 0.0015879302518442273,
-      "learning_rate": 0.00019999816197706572,
-      "loss": 46.0,
-      "step": 25252
-    },
-    {
-      "epoch": 1.9307682015406082,
-      "grad_norm": 0.0013303877785801888,
-      "learning_rate": 0.00019999816183143115,
-      "loss": 46.0,
-      "step": 25253
-    },
-    {
-      "epoch": 1.930844658523998,
-      "grad_norm": 0.0014115603407844901,
-      "learning_rate": 0.00019999816168579085,
-      "loss": 46.0,
-      "step": 25254
-    },
-    {
-      "epoch": 1.9309211155073878,
-      "grad_norm": 0.0012846238678321242,
-      "learning_rate": 0.00019999816154014474,
-      "loss": 46.0,
-      "step": 25255
-    },
-    {
-      "epoch": 1.9309975724907775,
-      "grad_norm": 0.001141722546890378,
-      "learning_rate": 0.00019999816139449288,
-      "loss": 46.0,
-      "step": 25256
-    },
-    {
-      "epoch": 1.931074029474167,
-      "grad_norm": 0.0007933792658150196,
-      "learning_rate": 0.00019999816124883525,
-      "loss": 46.0,
-      "step": 25257
-    },
-    {
-      "epoch": 1.9311504864575568,
-      "grad_norm": 0.0008432471659034491,
-      "learning_rate": 0.00019999816110317188,
-      "loss": 46.0,
-      "step": 25258
-    },
-    {
-      "epoch": 1.9312269434409466,
-      "grad_norm": 0.0018273926107212901,
-      "learning_rate": 0.0001999981609575027,
-      "loss": 46.0,
-      "step": 25259
-    },
-    {
-      "epoch": 1.9313034004243361,
-      "grad_norm": 0.0006823830190114677,
-      "learning_rate": 0.00019999816081182775,
-      "loss": 46.0,
-      "step": 25260
-    },
-    {
-      "epoch": 1.9313798574077259,
-      "grad_norm": 0.0030405069701373577,
-      "learning_rate": 0.00019999816066614706,
-      "loss": 46.0,
-      "step": 25261
-    },
-    {
-      "epoch": 1.9314563143911156,
-      "grad_norm": 0.0014062374830245972,
-      "learning_rate": 0.00019999816052046056,
-      "loss": 46.0,
-      "step": 25262
-    },
-    {
-      "epoch": 1.9315327713745054,
-      "grad_norm": 0.0016032709972932935,
-      "learning_rate": 0.0001999981603747683,
-      "loss": 46.0,
-      "step": 25263
-    },
-    {
-      "epoch": 1.9316092283578952,
-      "grad_norm": 0.00033817230723798275,
-      "learning_rate": 0.0001999981602290703,
-      "loss": 46.0,
-      "step": 25264
-    },
-    {
-      "epoch": 1.931685685341285,
-      "grad_norm": 0.001765438006259501,
-      "learning_rate": 0.00019999816008336652,
-      "loss": 46.0,
-      "step": 25265
-    },
-    {
-      "epoch": 1.9317621423246747,
-      "grad_norm": 0.0007238680846057832,
-      "learning_rate": 0.00019999815993765693,
-      "loss": 46.0,
-      "step": 25266
-    },
-    {
-      "epoch": 1.9318385993080645,
-      "grad_norm": 0.0012378082610666752,
-      "learning_rate": 0.00019999815979194162,
-      "loss": 46.0,
-      "step": 25267
-    },
-    {
-      "epoch": 1.931915056291454,
-      "grad_norm": 0.0016592685133218765,
-      "learning_rate": 0.00019999815964622054,
-      "loss": 46.0,
-      "step": 25268
-    },
-    {
-      "epoch": 1.9319915132748438,
-      "grad_norm": 0.00020512235641945153,
-      "learning_rate": 0.00019999815950049366,
-      "loss": 46.0,
-      "step": 25269
-    },
-    {
-      "epoch": 1.9320679702582335,
-      "grad_norm": 0.0008504390716552734,
-      "learning_rate": 0.00019999815935476103,
-      "loss": 46.0,
-      "step": 25270
-    },
-    {
-      "epoch": 1.932144427241623,
-      "grad_norm": 0.002243452239781618,
-      "learning_rate": 0.00019999815920902263,
-      "loss": 46.0,
-      "step": 25271
-    },
-    {
-      "epoch": 1.9322208842250128,
-      "grad_norm": 0.0007578788208775222,
-      "learning_rate": 0.00019999815906327845,
-      "loss": 46.0,
-      "step": 25272
-    },
-    {
-      "epoch": 1.9322973412084026,
-      "grad_norm": 0.0003603436052799225,
-      "learning_rate": 0.0001999981589175285,
-      "loss": 46.0,
-      "step": 25273
-    },
-    {
-      "epoch": 1.9323737981917923,
-      "grad_norm": 0.0021095096599310637,
-      "learning_rate": 0.00019999815877177278,
-      "loss": 46.0,
-      "step": 25274
-    },
-    {
-      "epoch": 1.932450255175182,
-      "grad_norm": 0.0007024008664302528,
-      "learning_rate": 0.0001999981586260113,
-      "loss": 46.0,
-      "step": 25275
-    },
-    {
-      "epoch": 1.9325267121585719,
-      "grad_norm": 0.001266497652977705,
-      "learning_rate": 0.00019999815848024407,
-      "loss": 46.0,
-      "step": 25276
-    },
-    {
-      "epoch": 1.9326031691419616,
-      "grad_norm": 0.0009625194361433387,
-      "learning_rate": 0.00019999815833447105,
-      "loss": 46.0,
-      "step": 25277
-    },
-    {
-      "epoch": 1.9326796261253514,
-      "grad_norm": 0.0005683636991307139,
-      "learning_rate": 0.00019999815818869227,
-      "loss": 46.0,
-      "step": 25278
-    },
-    {
-      "epoch": 1.932756083108741,
-      "grad_norm": 0.0009354943758808076,
-      "learning_rate": 0.0001999981580429077,
-      "loss": 46.0,
-      "step": 25279
-    },
-    {
-      "epoch": 1.9328325400921307,
-      "grad_norm": 0.0005600865697488189,
-      "learning_rate": 0.00019999815789711737,
-      "loss": 46.0,
-      "step": 25280
-    },
-    {
-      "epoch": 1.9329089970755204,
-      "grad_norm": 0.0012779291719198227,
-      "learning_rate": 0.0001999981577513213,
-      "loss": 46.0,
-      "step": 25281
-    },
-    {
-      "epoch": 1.93298545405891,
-      "grad_norm": 0.0018300687661394477,
-      "learning_rate": 0.00019999815760551943,
-      "loss": 46.0,
-      "step": 25282
-    },
-    {
-      "epoch": 1.9330619110422997,
-      "grad_norm": 0.0006032388191670179,
-      "learning_rate": 0.0001999981574597118,
-      "loss": 46.0,
-      "step": 25283
-    },
-    {
-      "epoch": 1.9331383680256895,
-      "grad_norm": 0.0010696015087887645,
-      "learning_rate": 0.0001999981573138984,
-      "loss": 46.0,
-      "step": 25284
-    },
-    {
-      "epoch": 1.9332148250090793,
-      "grad_norm": 0.0008619575528427958,
-      "learning_rate": 0.0001999981571680792,
-      "loss": 46.0,
-      "step": 25285
-    },
-    {
-      "epoch": 1.933291281992469,
-      "grad_norm": 0.0025029631797224283,
-      "learning_rate": 0.00019999815702225425,
-      "loss": 46.0,
-      "step": 25286
-    },
-    {
-      "epoch": 1.9333677389758588,
-      "grad_norm": 0.0016274285735562444,
-      "learning_rate": 0.00019999815687642356,
-      "loss": 46.0,
-      "step": 25287
-    },
-    {
-      "epoch": 1.9334441959592485,
-      "grad_norm": 0.0010962628293782473,
-      "learning_rate": 0.0001999981567305871,
-      "loss": 46.0,
-      "step": 25288
-    },
-    {
-      "epoch": 1.9335206529426383,
-      "grad_norm": 0.0013498312328010798,
-      "learning_rate": 0.00019999815658474482,
-      "loss": 46.0,
-      "step": 25289
-    },
-    {
-      "epoch": 1.9335971099260278,
-      "grad_norm": 0.0022390643134713173,
-      "learning_rate": 0.0001999981564388968,
-      "loss": 46.0,
-      "step": 25290
-    },
-    {
-      "epoch": 1.9336735669094176,
-      "grad_norm": 0.0022034659050405025,
-      "learning_rate": 0.00019999815629304302,
-      "loss": 46.0,
-      "step": 25291
-    },
-    {
-      "epoch": 1.9337500238928071,
-      "grad_norm": 0.008217648603022099,
-      "learning_rate": 0.00019999815614718346,
-      "loss": 46.0,
-      "step": 25292
-    },
-    {
-      "epoch": 1.933826480876197,
-      "grad_norm": 0.0010494983289390802,
-      "learning_rate": 0.00019999815600131812,
-      "loss": 46.0,
-      "step": 25293
-    },
-    {
-      "epoch": 1.9339029378595867,
-      "grad_norm": 0.00039278369513340294,
-      "learning_rate": 0.00019999815585544704,
-      "loss": 46.0,
-      "step": 25294
-    },
-    {
-      "epoch": 1.9339793948429764,
-      "grad_norm": 0.003571538021788001,
-      "learning_rate": 0.00019999815570957016,
-      "loss": 46.0,
-      "step": 25295
-    },
-    {
-      "epoch": 1.9340558518263662,
-      "grad_norm": 0.0031834174878895283,
-      "learning_rate": 0.00019999815556368753,
-      "loss": 46.0,
-      "step": 25296
-    },
-    {
-      "epoch": 1.934132308809756,
-      "grad_norm": 0.002159104449674487,
-      "learning_rate": 0.0001999981554177991,
-      "loss": 46.0,
-      "step": 25297
-    },
-    {
-      "epoch": 1.9342087657931457,
-      "grad_norm": 0.0008337416220456362,
-      "learning_rate": 0.00019999815527190493,
-      "loss": 46.0,
-      "step": 25298
-    },
-    {
-      "epoch": 1.9342852227765355,
-      "grad_norm": 0.0010350560769438744,
-      "learning_rate": 0.000199998155126005,
-      "loss": 46.0,
-      "step": 25299
-    },
-    {
-      "epoch": 1.9343616797599252,
-      "grad_norm": 0.0014388470444828272,
-      "learning_rate": 0.0001999981549800993,
-      "loss": 46.0,
-      "step": 25300
-    },
-    {
-      "epoch": 1.9344381367433148,
-      "grad_norm": 0.0008535103988833725,
-      "learning_rate": 0.0001999981548341878,
-      "loss": 46.0,
-      "step": 25301
-    },
-    {
-      "epoch": 1.9345145937267045,
-      "grad_norm": 0.0008447173167951405,
-      "learning_rate": 0.00019999815468827055,
-      "loss": 46.0,
-      "step": 25302
-    },
-    {
-      "epoch": 1.934591050710094,
-      "grad_norm": 0.0005892303888686001,
-      "learning_rate": 0.00019999815454234754,
-      "loss": 46.0,
-      "step": 25303
-    },
-    {
-      "epoch": 1.9346675076934838,
-      "grad_norm": 0.0025854837149381638,
-      "learning_rate": 0.00019999815439641872,
-      "loss": 46.0,
-      "step": 25304
-    },
-    {
-      "epoch": 1.9347439646768736,
-      "grad_norm": 0.002143304329365492,
-      "learning_rate": 0.0001999981542504842,
-      "loss": 46.0,
-      "step": 25305
-    },
-    {
-      "epoch": 1.9348204216602634,
-      "grad_norm": 0.0010795366251841187,
-      "learning_rate": 0.00019999815410454383,
-      "loss": 46.0,
-      "step": 25306
-    },
-    {
-      "epoch": 1.9348968786436531,
-      "grad_norm": 0.0010126135312020779,
-      "learning_rate": 0.00019999815395859772,
-      "loss": 46.0,
-      "step": 25307
-    },
-    {
-      "epoch": 1.9349733356270429,
-      "grad_norm": 0.0013496881583705544,
-      "learning_rate": 0.0001999981538126459,
-      "loss": 46.0,
-      "step": 25308
-    },
-    {
-      "epoch": 1.9350497926104326,
-      "grad_norm": 0.0009083639015443623,
-      "learning_rate": 0.00019999815366668824,
-      "loss": 46.0,
-      "step": 25309
-    },
-    {
-      "epoch": 1.9351262495938224,
-      "grad_norm": 0.0005320738418959081,
-      "learning_rate": 0.00019999815352072484,
-      "loss": 46.0,
-      "step": 25310
-    },
-    {
-      "epoch": 1.9352027065772122,
-      "grad_norm": 0.0005455210921354592,
-      "learning_rate": 0.00019999815337475564,
-      "loss": 46.0,
-      "step": 25311
-    },
-    {
-      "epoch": 1.9352791635606017,
-      "grad_norm": 0.0012280141236260533,
-      "learning_rate": 0.00019999815322878072,
-      "loss": 46.0,
-      "step": 25312
-    },
-    {
-      "epoch": 1.9353556205439915,
-      "grad_norm": 0.0026231713127344847,
-      "learning_rate": 0.0001999981530828,
-      "loss": 46.0,
-      "step": 25313
-    },
-    {
-      "epoch": 1.935432077527381,
-      "grad_norm": 0.0020567369647324085,
-      "learning_rate": 0.00019999815293681353,
-      "loss": 46.0,
-      "step": 25314
-    },
-    {
-      "epoch": 1.9355085345107708,
-      "grad_norm": 0.0010275328531861305,
-      "learning_rate": 0.00019999815279082126,
-      "loss": 46.0,
-      "step": 25315
-    },
-    {
-      "epoch": 1.9355849914941605,
-      "grad_norm": 0.0036778005305677652,
-      "learning_rate": 0.00019999815264482322,
-      "loss": 46.0,
-      "step": 25316
-    },
-    {
-      "epoch": 1.9356614484775503,
-      "grad_norm": 0.0015031290240585804,
-      "learning_rate": 0.00019999815249881943,
-      "loss": 46.0,
-      "step": 25317
-    },
-    {
-      "epoch": 1.93573790546094,
-      "grad_norm": 0.0015411186031997204,
-      "learning_rate": 0.00019999815235280987,
-      "loss": 46.0,
-      "step": 25318
-    },
-    {
-      "epoch": 1.9358143624443298,
-      "grad_norm": 0.001010417123325169,
-      "learning_rate": 0.00019999815220679454,
-      "loss": 46.0,
-      "step": 25319
-    },
-    {
-      "epoch": 1.9358908194277196,
-      "grad_norm": 0.0006620500353164971,
-      "learning_rate": 0.00019999815206077343,
-      "loss": 46.0,
-      "step": 25320
-    },
-    {
-      "epoch": 1.9359672764111093,
-      "grad_norm": 0.0012426540488377213,
-      "learning_rate": 0.00019999815191474658,
-      "loss": 46.0,
-      "step": 25321
-    },
-    {
-      "epoch": 1.9360437333944989,
-      "grad_norm": 0.0006940505118109286,
-      "learning_rate": 0.00019999815176871395,
-      "loss": 46.0,
-      "step": 25322
-    },
-    {
-      "epoch": 1.9361201903778886,
-      "grad_norm": 0.005281574092805386,
-      "learning_rate": 0.00019999815162267552,
-      "loss": 46.0,
-      "step": 25323
-    },
-    {
-      "epoch": 1.9361966473612784,
-      "grad_norm": 0.001288366038352251,
-      "learning_rate": 0.00019999815147663137,
-      "loss": 46.0,
-      "step": 25324
-    },
-    {
-      "epoch": 1.936273104344668,
-      "grad_norm": 0.0006116593722254038,
-      "learning_rate": 0.0001999981513305814,
-      "loss": 46.0,
-      "step": 25325
-    },
-    {
-      "epoch": 1.9363495613280577,
-      "grad_norm": 0.001009846106171608,
-      "learning_rate": 0.0001999981511845257,
-      "loss": 46.0,
-      "step": 25326
-    },
-    {
-      "epoch": 1.9364260183114475,
-      "grad_norm": 0.0007040408672764897,
-      "learning_rate": 0.0001999981510384642,
-      "loss": 46.0,
-      "step": 25327
-    },
-    {
-      "epoch": 1.9365024752948372,
-      "grad_norm": 0.006772223860025406,
-      "learning_rate": 0.00019999815089239692,
-      "loss": 46.0,
-      "step": 25328
-    },
-    {
-      "epoch": 1.936578932278227,
-      "grad_norm": 0.0005106626776978374,
-      "learning_rate": 0.0001999981507463239,
-      "loss": 46.0,
-      "step": 25329
-    },
-    {
-      "epoch": 1.9366553892616167,
-      "grad_norm": 0.0006273649632930756,
-      "learning_rate": 0.00019999815060024512,
-      "loss": 46.0,
-      "step": 25330
-    },
-    {
-      "epoch": 1.9367318462450065,
-      "grad_norm": 0.000693241716362536,
-      "learning_rate": 0.00019999815045416056,
-      "loss": 46.0,
-      "step": 25331
-    },
-    {
-      "epoch": 1.9368083032283963,
-      "grad_norm": 0.001070735976099968,
-      "learning_rate": 0.00019999815030807022,
-      "loss": 46.0,
-      "step": 25332
-    },
-    {
-      "epoch": 1.9368847602117858,
-      "grad_norm": 0.0012118174927309155,
-      "learning_rate": 0.00019999815016197414,
-      "loss": 46.0,
-      "step": 25333
-    },
-    {
-      "epoch": 1.9369612171951756,
-      "grad_norm": 0.008624296635389328,
-      "learning_rate": 0.00019999815001587226,
-      "loss": 46.0,
-      "step": 25334
-    },
-    {
-      "epoch": 1.9370376741785653,
-      "grad_norm": 0.0006888203206472099,
-      "learning_rate": 0.0001999981498697646,
-      "loss": 46.0,
-      "step": 25335
-    },
-    {
-      "epoch": 1.9371141311619549,
-      "grad_norm": 0.0005594720132648945,
-      "learning_rate": 0.0001999981497236512,
-      "loss": 46.0,
-      "step": 25336
-    },
-    {
-      "epoch": 1.9371905881453446,
-      "grad_norm": 0.0017651293892413378,
-      "learning_rate": 0.00019999814957753203,
-      "loss": 46.0,
-      "step": 25337
-    },
-    {
-      "epoch": 1.9372670451287344,
-      "grad_norm": 0.0004064141830895096,
-      "learning_rate": 0.00019999814943140706,
-      "loss": 46.0,
-      "step": 25338
-    },
-    {
-      "epoch": 1.9373435021121241,
-      "grad_norm": 0.0005659838207066059,
-      "learning_rate": 0.00019999814928527637,
-      "loss": 46.0,
-      "step": 25339
-    },
-    {
-      "epoch": 1.937419959095514,
-      "grad_norm": 0.0018348086159676313,
-      "learning_rate": 0.00019999814913913987,
-      "loss": 46.0,
-      "step": 25340
-    },
-    {
-      "epoch": 1.9374964160789037,
-      "grad_norm": 0.0003551933041308075,
-      "learning_rate": 0.00019999814899299763,
-      "loss": 46.0,
-      "step": 25341
-    },
-    {
-      "epoch": 1.9375728730622934,
-      "grad_norm": 0.0017288707895204425,
-      "learning_rate": 0.0001999981488468496,
-      "loss": 46.0,
-      "step": 25342
-    },
-    {
-      "epoch": 1.9376493300456832,
-      "grad_norm": 0.002727646380662918,
-      "learning_rate": 0.0001999981487006958,
-      "loss": 46.0,
-      "step": 25343
-    },
-    {
-      "epoch": 1.9377257870290727,
-      "grad_norm": 0.002783076139166951,
-      "learning_rate": 0.00019999814855453622,
-      "loss": 46.0,
-      "step": 25344
-    },
-    {
-      "epoch": 1.9378022440124625,
-      "grad_norm": 0.0007516831392422318,
-      "learning_rate": 0.00019999814840837088,
-      "loss": 46.0,
-      "step": 25345
-    },
-    {
-      "epoch": 1.9378787009958522,
-      "grad_norm": 0.0015455958200618625,
-      "learning_rate": 0.0001999981482621998,
-      "loss": 46.0,
-      "step": 25346
-    },
-    {
-      "epoch": 1.9379551579792418,
-      "grad_norm": 0.003510406706482172,
-      "learning_rate": 0.0001999981481160229,
-      "loss": 46.0,
-      "step": 25347
-    },
-    {
-      "epoch": 1.9380316149626315,
-      "grad_norm": 0.0008550632046535611,
-      "learning_rate": 0.0001999981479698403,
-      "loss": 46.0,
-      "step": 25348
-    },
-    {
-      "epoch": 1.9381080719460213,
-      "grad_norm": 0.0004901379579678178,
-      "learning_rate": 0.00019999814782365184,
-      "loss": 46.0,
-      "step": 25349
-    },
-    {
-      "epoch": 1.938184528929411,
-      "grad_norm": 0.0011447896249592304,
-      "learning_rate": 0.00019999814767745767,
-      "loss": 46.0,
-      "step": 25350
-    },
-    {
-      "epoch": 1.9382609859128008,
-      "grad_norm": 0.0068277595564723015,
-      "learning_rate": 0.00019999814753125775,
-      "loss": 46.0,
-      "step": 25351
-    },
-    {
-      "epoch": 1.9383374428961906,
-      "grad_norm": 0.002613328630104661,
-      "learning_rate": 0.00019999814738505203,
-      "loss": 46.0,
-      "step": 25352
-    },
-    {
-      "epoch": 1.9384138998795803,
-      "grad_norm": 0.001656971639022231,
-      "learning_rate": 0.0001999981472388405,
-      "loss": 46.0,
-      "step": 25353
-    },
-    {
-      "epoch": 1.93849035686297,
-      "grad_norm": 0.00023466994753107429,
-      "learning_rate": 0.00019999814709262327,
-      "loss": 46.0,
-      "step": 25354
-    },
-    {
-      "epoch": 1.9385668138463596,
-      "grad_norm": 0.001167806563898921,
-      "learning_rate": 0.00019999814694640026,
-      "loss": 46.0,
-      "step": 25355
-    },
-    {
-      "epoch": 1.9386432708297494,
-      "grad_norm": 0.001004403573460877,
-      "learning_rate": 0.00019999814680017144,
-      "loss": 46.0,
-      "step": 25356
-    },
-    {
-      "epoch": 1.9387197278131392,
-      "grad_norm": 0.001383994473144412,
-      "learning_rate": 0.00019999814665393686,
-      "loss": 46.0,
-      "step": 25357
-    },
-    {
-      "epoch": 1.9387961847965287,
-      "grad_norm": 0.0012152658309787512,
-      "learning_rate": 0.00019999814650769655,
-      "loss": 46.0,
-      "step": 25358
-    },
-    {
-      "epoch": 1.9388726417799185,
-      "grad_norm": 0.0019428218947723508,
-      "learning_rate": 0.00019999814636145042,
-      "loss": 46.0,
-      "step": 25359
-    },
-    {
-      "epoch": 1.9389490987633082,
-      "grad_norm": 0.0024508978240191936,
-      "learning_rate": 0.00019999814621519856,
-      "loss": 46.0,
-      "step": 25360
-    },
-    {
-      "epoch": 1.939025555746698,
-      "grad_norm": 0.001649106852710247,
-      "learning_rate": 0.00019999814606894088,
-      "loss": 46.0,
-      "step": 25361
-    },
-    {
-      "epoch": 1.9391020127300878,
-      "grad_norm": 0.0007924661040306091,
-      "learning_rate": 0.00019999814592267748,
-      "loss": 46.0,
-      "step": 25362
-    },
-    {
-      "epoch": 1.9391784697134775,
-      "grad_norm": 0.0015159822069108486,
-      "learning_rate": 0.0001999981457764083,
-      "loss": 46.0,
-      "step": 25363
-    },
-    {
-      "epoch": 1.9392549266968673,
-      "grad_norm": 0.00569218909367919,
-      "learning_rate": 0.00019999814563013336,
-      "loss": 46.0,
-      "step": 25364
-    },
-    {
-      "epoch": 1.939331383680257,
-      "grad_norm": 0.0020105738658457994,
-      "learning_rate": 0.00019999814548385264,
-      "loss": 46.0,
-      "step": 25365
-    },
-    {
-      "epoch": 1.9394078406636466,
-      "grad_norm": 0.0033240276388823986,
-      "learning_rate": 0.00019999814533756615,
-      "loss": 46.0,
-      "step": 25366
-    },
-    {
-      "epoch": 1.9394842976470363,
-      "grad_norm": 0.0016309726051986217,
-      "learning_rate": 0.00019999814519127389,
-      "loss": 46.0,
-      "step": 25367
-    },
-    {
-      "epoch": 1.939560754630426,
-      "grad_norm": 0.0010426866356283426,
-      "learning_rate": 0.00019999814504497585,
-      "loss": 46.0,
-      "step": 25368
-    },
-    {
-      "epoch": 1.9396372116138156,
-      "grad_norm": 0.0010331958765164018,
-      "learning_rate": 0.00019999814489867206,
-      "loss": 46.0,
-      "step": 25369
-    },
-    {
-      "epoch": 1.9397136685972054,
-      "grad_norm": 0.0015260276850312948,
-      "learning_rate": 0.0001999981447523625,
-      "loss": 46.0,
-      "step": 25370
-    },
-    {
-      "epoch": 1.9397901255805952,
-      "grad_norm": 0.005961915012449026,
-      "learning_rate": 0.00019999814460604717,
-      "loss": 46.0,
-      "step": 25371
-    },
-    {
-      "epoch": 1.939866582563985,
-      "grad_norm": 0.002002912340685725,
-      "learning_rate": 0.00019999814445972603,
-      "loss": 46.0,
-      "step": 25372
-    },
-    {
-      "epoch": 1.9399430395473747,
-      "grad_norm": 0.0018917162669822574,
-      "learning_rate": 0.00019999814431339918,
-      "loss": 46.0,
-      "step": 25373
-    },
-    {
-      "epoch": 1.9400194965307644,
-      "grad_norm": 0.0021304097026586533,
-      "learning_rate": 0.00019999814416706653,
-      "loss": 46.0,
-      "step": 25374
-    },
-    {
-      "epoch": 1.9400959535141542,
-      "grad_norm": 0.002490064362064004,
-      "learning_rate": 0.00019999814402072813,
-      "loss": 46.0,
-      "step": 25375
-    },
-    {
-      "epoch": 1.940172410497544,
-      "grad_norm": 0.009650619700551033,
-      "learning_rate": 0.0001999981438743839,
-      "loss": 46.0,
-      "step": 25376
-    },
-    {
-      "epoch": 1.9402488674809335,
-      "grad_norm": 0.001609961618669331,
-      "learning_rate": 0.00019999814372803396,
-      "loss": 46.0,
-      "step": 25377
-    },
-    {
-      "epoch": 1.9403253244643233,
-      "grad_norm": 0.0005743880174122751,
-      "learning_rate": 0.00019999814358167826,
-      "loss": 46.0,
-      "step": 25378
-    },
-    {
-      "epoch": 1.940401781447713,
-      "grad_norm": 0.001369268400594592,
-      "learning_rate": 0.00019999814343531677,
-      "loss": 46.0,
-      "step": 25379
-    },
-    {
-      "epoch": 1.9404782384311026,
-      "grad_norm": 0.0005733421421609819,
-      "learning_rate": 0.0001999981432889495,
-      "loss": 46.0,
-      "step": 25380
-    },
-    {
-      "epoch": 1.9405546954144923,
-      "grad_norm": 0.0019639874808490276,
-      "learning_rate": 0.00019999814314257646,
-      "loss": 46.0,
-      "step": 25381
-    },
-    {
-      "epoch": 1.940631152397882,
-      "grad_norm": 0.0008799354545772076,
-      "learning_rate": 0.00019999814299619768,
-      "loss": 46.0,
-      "step": 25382
-    },
-    {
-      "epoch": 1.9407076093812718,
-      "grad_norm": 0.00126199284568429,
-      "learning_rate": 0.0001999981428498131,
-      "loss": 46.0,
-      "step": 25383
-    },
-    {
-      "epoch": 1.9407840663646616,
-      "grad_norm": 0.0015734905609861016,
-      "learning_rate": 0.00019999814270342276,
-      "loss": 46.0,
-      "step": 25384
-    },
-    {
-      "epoch": 1.9408605233480514,
-      "grad_norm": 0.0005148928030394018,
-      "learning_rate": 0.00019999814255702665,
-      "loss": 46.0,
-      "step": 25385
-    },
-    {
-      "epoch": 1.9409369803314411,
-      "grad_norm": 0.0009339142707176507,
-      "learning_rate": 0.00019999814241062477,
-      "loss": 46.0,
-      "step": 25386
-    },
-    {
-      "epoch": 1.941013437314831,
-      "grad_norm": 0.004559422377496958,
-      "learning_rate": 0.00019999814226421712,
-      "loss": 46.0,
-      "step": 25387
-    },
-    {
-      "epoch": 1.9410898942982204,
-      "grad_norm": 0.0006120121688582003,
-      "learning_rate": 0.0001999981421178037,
-      "loss": 46.0,
-      "step": 25388
-    },
-    {
-      "epoch": 1.9411663512816102,
-      "grad_norm": 0.000705523241776973,
-      "learning_rate": 0.00019999814197138452,
-      "loss": 46.0,
-      "step": 25389
-    },
-    {
-      "epoch": 1.941242808265,
-      "grad_norm": 0.0014590610517188907,
-      "learning_rate": 0.00019999814182495958,
-      "loss": 46.0,
-      "step": 25390
-    },
-    {
-      "epoch": 1.9413192652483895,
-      "grad_norm": 0.004136715084314346,
-      "learning_rate": 0.00019999814167852886,
-      "loss": 46.0,
-      "step": 25391
-    },
-    {
-      "epoch": 1.9413957222317793,
-      "grad_norm": 0.0006367635796777904,
-      "learning_rate": 0.00019999814153209237,
-      "loss": 46.0,
-      "step": 25392
-    },
-    {
-      "epoch": 1.941472179215169,
-      "grad_norm": 0.002046600915491581,
-      "learning_rate": 0.00019999814138565007,
-      "loss": 46.0,
-      "step": 25393
-    },
-    {
-      "epoch": 1.9415486361985588,
-      "grad_norm": 0.001310759806074202,
-      "learning_rate": 0.00019999814123920206,
-      "loss": 46.0,
-      "step": 25394
-    },
-    {
-      "epoch": 1.9416250931819485,
-      "grad_norm": 0.0021803409326821566,
-      "learning_rate": 0.00019999814109274825,
-      "loss": 46.0,
-      "step": 25395
-    },
-    {
-      "epoch": 1.9417015501653383,
-      "grad_norm": 0.0022339874412864447,
-      "learning_rate": 0.0001999981409462887,
-      "loss": 46.0,
-      "step": 25396
-    },
-    {
-      "epoch": 1.941778007148728,
-      "grad_norm": 0.001016075606457889,
-      "learning_rate": 0.00019999814079982336,
-      "loss": 46.0,
-      "step": 25397
-    },
-    {
-      "epoch": 1.9418544641321178,
-      "grad_norm": 0.001984353642910719,
-      "learning_rate": 0.00019999814065335222,
-      "loss": 46.0,
-      "step": 25398
-    },
-    {
-      "epoch": 1.9419309211155074,
-      "grad_norm": 0.001188677386380732,
-      "learning_rate": 0.00019999814050687535,
-      "loss": 46.0,
-      "step": 25399
-    },
-    {
-      "epoch": 1.9420073780988971,
-      "grad_norm": 0.0018012761138379574,
-      "learning_rate": 0.0001999981403603927,
-      "loss": 46.0,
-      "step": 25400
-    },
-    {
-      "epoch": 1.9420838350822869,
-      "grad_norm": 0.0009258925565518439,
-      "learning_rate": 0.0001999981402139043,
-      "loss": 46.0,
-      "step": 25401
-    },
-    {
-      "epoch": 1.9421602920656764,
-      "grad_norm": 0.00034944259095937014,
-      "learning_rate": 0.00019999814006741012,
-      "loss": 46.0,
-      "step": 25402
-    },
-    {
-      "epoch": 1.9422367490490662,
-      "grad_norm": 0.0010209300089627504,
-      "learning_rate": 0.00019999813992091015,
-      "loss": 46.0,
-      "step": 25403
-    },
-    {
-      "epoch": 1.942313206032456,
-      "grad_norm": 0.0007083873497322202,
-      "learning_rate": 0.0001999981397744044,
-      "loss": 46.0,
-      "step": 25404
-    },
-    {
-      "epoch": 1.9423896630158457,
-      "grad_norm": 0.0010984893888235092,
-      "learning_rate": 0.00019999813962789294,
-      "loss": 46.0,
-      "step": 25405
-    },
-    {
-      "epoch": 1.9424661199992355,
-      "grad_norm": 0.0010761962039396167,
-      "learning_rate": 0.00019999813948137565,
-      "loss": 46.0,
-      "step": 25406
-    },
-    {
-      "epoch": 1.9425425769826252,
-      "grad_norm": 0.0017928986344486475,
-      "learning_rate": 0.00019999813933485263,
-      "loss": 46.0,
-      "step": 25407
-    },
-    {
-      "epoch": 1.942619033966015,
-      "grad_norm": 0.006016286555677652,
-      "learning_rate": 0.00019999813918832382,
-      "loss": 46.0,
-      "step": 25408
-    },
-    {
-      "epoch": 1.9426954909494047,
-      "grad_norm": 0.0014520544791594148,
-      "learning_rate": 0.00019999813904178924,
-      "loss": 46.0,
-      "step": 25409
-    },
-    {
-      "epoch": 1.9427719479327943,
-      "grad_norm": 0.0009543381165713072,
-      "learning_rate": 0.0001999981388952489,
-      "loss": 46.0,
-      "step": 25410
-    },
-    {
-      "epoch": 1.942848404916184,
-      "grad_norm": 0.0006197650800459087,
-      "learning_rate": 0.0001999981387487028,
-      "loss": 46.0,
-      "step": 25411
-    },
-    {
-      "epoch": 1.9429248618995738,
-      "grad_norm": 0.0014205684419721365,
-      "learning_rate": 0.00019999813860215092,
-      "loss": 46.0,
-      "step": 25412
-    },
-    {
-      "epoch": 1.9430013188829633,
-      "grad_norm": 0.0011642139870673418,
-      "learning_rate": 0.00019999813845559324,
-      "loss": 46.0,
-      "step": 25413
-    },
-    {
-      "epoch": 1.943077775866353,
-      "grad_norm": 0.0005908952443860471,
-      "learning_rate": 0.00019999813830902984,
-      "loss": 46.0,
-      "step": 25414
-    },
-    {
-      "epoch": 1.9431542328497429,
-      "grad_norm": 0.0007647752063348889,
-      "learning_rate": 0.00019999813816246067,
-      "loss": 46.0,
-      "step": 25415
-    },
-    {
-      "epoch": 1.9432306898331326,
-      "grad_norm": 0.0004617222584784031,
-      "learning_rate": 0.0001999981380158857,
-      "loss": 46.0,
-      "step": 25416
-    },
-    {
-      "epoch": 1.9433071468165224,
-      "grad_norm": 0.0007163193076848984,
-      "learning_rate": 0.00019999813786930498,
-      "loss": 46.0,
-      "step": 25417
-    },
-    {
-      "epoch": 1.9433836037999122,
-      "grad_norm": 0.0006509136292152107,
-      "learning_rate": 0.00019999813772271846,
-      "loss": 46.0,
-      "step": 25418
-    },
-    {
-      "epoch": 1.943460060783302,
-      "grad_norm": 0.0017256282735615969,
-      "learning_rate": 0.00019999813757612622,
-      "loss": 46.0,
-      "step": 25419
-    },
-    {
-      "epoch": 1.9435365177666917,
-      "grad_norm": 0.0009750222670845687,
-      "learning_rate": 0.00019999813742952818,
-      "loss": 46.0,
-      "step": 25420
-    },
-    {
-      "epoch": 1.9436129747500812,
-      "grad_norm": 0.0008193259709514678,
-      "learning_rate": 0.00019999813728292435,
-      "loss": 46.0,
-      "step": 25421
-    },
-    {
-      "epoch": 1.943689431733471,
-      "grad_norm": 0.0009368039900436997,
-      "learning_rate": 0.0001999981371363148,
-      "loss": 46.0,
-      "step": 25422
-    },
-    {
-      "epoch": 1.9437658887168605,
-      "grad_norm": 0.0009751998004503548,
-      "learning_rate": 0.00019999813698969946,
-      "loss": 46.0,
-      "step": 25423
-    },
-    {
-      "epoch": 1.9438423457002503,
-      "grad_norm": 0.003908711019903421,
-      "learning_rate": 0.00019999813684307832,
-      "loss": 46.0,
-      "step": 25424
-    },
-    {
-      "epoch": 1.94391880268364,
-      "grad_norm": 0.0008872703183442354,
-      "learning_rate": 0.00019999813669645145,
-      "loss": 46.0,
-      "step": 25425
-    },
-    {
-      "epoch": 1.9439952596670298,
-      "grad_norm": 0.003220543498173356,
-      "learning_rate": 0.0001999981365498188,
-      "loss": 46.0,
-      "step": 25426
-    },
-    {
-      "epoch": 1.9440717166504196,
-      "grad_norm": 0.002107213716953993,
-      "learning_rate": 0.0001999981364031804,
-      "loss": 46.0,
-      "step": 25427
-    },
-    {
-      "epoch": 1.9441481736338093,
-      "grad_norm": 0.0011610285146161914,
-      "learning_rate": 0.0001999981362565362,
-      "loss": 46.0,
-      "step": 25428
-    },
-    {
-      "epoch": 1.944224630617199,
-      "grad_norm": 0.0011864140396937728,
-      "learning_rate": 0.00019999813610988625,
-      "loss": 46.0,
-      "step": 25429
-    },
-    {
-      "epoch": 1.9443010876005888,
-      "grad_norm": 0.0018960388842970133,
-      "learning_rate": 0.00019999813596323048,
-      "loss": 46.0,
-      "step": 25430
-    },
-    {
-      "epoch": 1.9443775445839786,
-      "grad_norm": 0.0006463818717747927,
-      "learning_rate": 0.00019999813581656902,
-      "loss": 46.0,
-      "step": 25431
-    },
-    {
-      "epoch": 1.9444540015673681,
-      "grad_norm": 0.0008672216790728271,
-      "learning_rate": 0.00019999813566990175,
-      "loss": 46.0,
-      "step": 25432
-    },
-    {
-      "epoch": 1.944530458550758,
-      "grad_norm": 0.0021202166099101305,
-      "learning_rate": 0.0001999981355232287,
-      "loss": 46.0,
-      "step": 25433
-    },
-    {
-      "epoch": 1.9446069155341474,
-      "grad_norm": 0.001004306715913117,
-      "learning_rate": 0.0001999981353765499,
-      "loss": 46.0,
-      "step": 25434
-    },
-    {
-      "epoch": 1.9446833725175372,
-      "grad_norm": 0.0028675622306764126,
-      "learning_rate": 0.00019999813522986532,
-      "loss": 46.0,
-      "step": 25435
-    },
-    {
-      "epoch": 1.944759829500927,
-      "grad_norm": 0.000756549765355885,
-      "learning_rate": 0.00019999813508317498,
-      "loss": 46.0,
-      "step": 25436
-    },
-    {
-      "epoch": 1.9448362864843167,
-      "grad_norm": 0.0008775634923949838,
-      "learning_rate": 0.00019999813493647888,
-      "loss": 46.0,
-      "step": 25437
-    },
-    {
-      "epoch": 1.9449127434677065,
-      "grad_norm": 0.000666490406729281,
-      "learning_rate": 0.00019999813478977697,
-      "loss": 46.0,
-      "step": 25438
-    },
-    {
-      "epoch": 1.9449892004510962,
-      "grad_norm": 0.0005975561798550189,
-      "learning_rate": 0.00019999813464306932,
-      "loss": 46.0,
-      "step": 25439
-    },
-    {
-      "epoch": 1.945065657434486,
-      "grad_norm": 0.0012797012459486723,
-      "learning_rate": 0.00019999813449635593,
-      "loss": 46.0,
-      "step": 25440
-    },
-    {
-      "epoch": 1.9451421144178758,
-      "grad_norm": 0.0004020068736281246,
-      "learning_rate": 0.0001999981343496367,
-      "loss": 46.0,
-      "step": 25441
-    },
-    {
-      "epoch": 1.9452185714012655,
-      "grad_norm": 0.0005726832896471024,
-      "learning_rate": 0.00019999813420291176,
-      "loss": 46.0,
-      "step": 25442
-    },
-    {
-      "epoch": 1.945295028384655,
-      "grad_norm": 0.0012177075259387493,
-      "learning_rate": 0.000199998134056181,
-      "loss": 46.0,
-      "step": 25443
-    },
-    {
-      "epoch": 1.9453714853680448,
-      "grad_norm": 0.00281155901029706,
-      "learning_rate": 0.00019999813390944452,
-      "loss": 46.0,
-      "step": 25444
-    },
-    {
-      "epoch": 1.9454479423514344,
-      "grad_norm": 0.0035891542211174965,
-      "learning_rate": 0.00019999813376270225,
-      "loss": 46.0,
-      "step": 25445
-    },
-    {
-      "epoch": 1.9455243993348241,
-      "grad_norm": 0.0025906874798238277,
-      "learning_rate": 0.00019999813361595422,
-      "loss": 46.0,
-      "step": 25446
-    },
-    {
-      "epoch": 1.945600856318214,
-      "grad_norm": 0.000746597012039274,
-      "learning_rate": 0.0001999981334692004,
-      "loss": 46.0,
-      "step": 25447
-    },
-    {
-      "epoch": 1.9456773133016037,
-      "grad_norm": 0.0015209836419671774,
-      "learning_rate": 0.00019999813332244082,
-      "loss": 46.0,
-      "step": 25448
-    },
-    {
-      "epoch": 1.9457537702849934,
-      "grad_norm": 0.010853101499378681,
-      "learning_rate": 0.0001999981331756755,
-      "loss": 46.0,
-      "step": 25449
-    },
-    {
-      "epoch": 1.9458302272683832,
-      "grad_norm": 0.0025025890208780766,
-      "learning_rate": 0.00019999813302890438,
-      "loss": 46.0,
-      "step": 25450
-    },
-    {
-      "epoch": 1.945906684251773,
-      "grad_norm": 0.0013881336199119687,
-      "learning_rate": 0.00019999813288212748,
-      "loss": 46.0,
-      "step": 25451
-    },
-    {
-      "epoch": 1.9459831412351627,
-      "grad_norm": 0.0013710112543776631,
-      "learning_rate": 0.00019999813273534483,
-      "loss": 46.0,
-      "step": 25452
-    },
-    {
-      "epoch": 1.9460595982185522,
-      "grad_norm": 0.0022558008786290884,
-      "learning_rate": 0.0001999981325885564,
-      "loss": 46.0,
-      "step": 25453
-    },
-    {
-      "epoch": 1.946136055201942,
-      "grad_norm": 0.0016930341953411698,
-      "learning_rate": 0.00019999813244176223,
-      "loss": 46.0,
-      "step": 25454
-    },
-    {
-      "epoch": 1.9462125121853318,
-      "grad_norm": 0.004416623618453741,
-      "learning_rate": 0.00019999813229496224,
-      "loss": 46.0,
-      "step": 25455
-    },
-    {
-      "epoch": 1.9462889691687213,
-      "grad_norm": 0.0019515713211148977,
-      "learning_rate": 0.0001999981321481565,
-      "loss": 46.0,
-      "step": 25456
-    },
-    {
-      "epoch": 1.946365426152111,
-      "grad_norm": 0.0008018958615139127,
-      "learning_rate": 0.000199998132001345,
-      "loss": 46.0,
-      "step": 25457
-    },
-    {
-      "epoch": 1.9464418831355008,
-      "grad_norm": 0.0014892738545313478,
-      "learning_rate": 0.00019999813185452774,
-      "loss": 46.0,
-      "step": 25458
-    },
-    {
-      "epoch": 1.9465183401188906,
-      "grad_norm": 0.0015683931997045875,
-      "learning_rate": 0.0001999981317077047,
-      "loss": 46.0,
-      "step": 25459
-    },
-    {
-      "epoch": 1.9465947971022803,
-      "grad_norm": 0.0017586903413757682,
-      "learning_rate": 0.0001999981315608759,
-      "loss": 46.0,
-      "step": 25460
-    },
-    {
-      "epoch": 1.94667125408567,
-      "grad_norm": 0.003979188855737448,
-      "learning_rate": 0.0001999981314140413,
-      "loss": 46.0,
-      "step": 25461
-    },
-    {
-      "epoch": 1.9467477110690599,
-      "grad_norm": 0.00048734399024397135,
-      "learning_rate": 0.00019999813126720097,
-      "loss": 46.0,
-      "step": 25462
-    },
-    {
-      "epoch": 1.9468241680524496,
-      "grad_norm": 0.0008830137667246163,
-      "learning_rate": 0.00019999813112035484,
-      "loss": 46.0,
-      "step": 25463
-    },
-    {
-      "epoch": 1.9469006250358392,
-      "grad_norm": 0.0017410836881026626,
-      "learning_rate": 0.00019999813097350296,
-      "loss": 46.0,
-      "step": 25464
-    },
-    {
-      "epoch": 1.946977082019229,
-      "grad_norm": 0.0008175763650797307,
-      "learning_rate": 0.00019999813082664529,
-      "loss": 46.0,
-      "step": 25465
-    },
-    {
-      "epoch": 1.9470535390026187,
-      "grad_norm": 0.0017329795518890023,
-      "learning_rate": 0.0001999981306797819,
-      "loss": 46.0,
-      "step": 25466
-    },
-    {
-      "epoch": 1.9471299959860082,
-      "grad_norm": 0.001185034867376089,
-      "learning_rate": 0.0001999981305329127,
-      "loss": 46.0,
-      "step": 25467
-    },
-    {
-      "epoch": 1.947206452969398,
-      "grad_norm": 0.0005753164296038449,
-      "learning_rate": 0.00019999813038603772,
-      "loss": 46.0,
-      "step": 25468
-    },
-    {
-      "epoch": 1.9472829099527877,
-      "grad_norm": 0.0025529414415359497,
-      "learning_rate": 0.000199998130239157,
-      "loss": 46.0,
-      "step": 25469
-    },
-    {
-      "epoch": 1.9473593669361775,
-      "grad_norm": 0.0013759658904746175,
-      "learning_rate": 0.00019999813009227048,
-      "loss": 46.0,
-      "step": 25470
-    },
-    {
-      "epoch": 1.9474358239195673,
-      "grad_norm": 0.0007008545217104256,
-      "learning_rate": 0.00019999812994537822,
-      "loss": 46.0,
-      "step": 25471
-    },
-    {
-      "epoch": 1.947512280902957,
-      "grad_norm": 0.0006326701841317117,
-      "learning_rate": 0.00019999812979848018,
-      "loss": 46.0,
-      "step": 25472
-    },
-    {
-      "epoch": 1.9475887378863468,
-      "grad_norm": 0.0009554935386404395,
-      "learning_rate": 0.00019999812965157637,
-      "loss": 46.0,
-      "step": 25473
-    },
-    {
-      "epoch": 1.9476651948697365,
-      "grad_norm": 0.0008366239489987493,
-      "learning_rate": 0.00019999812950466676,
-      "loss": 46.0,
-      "step": 25474
-    },
-    {
-      "epoch": 1.947741651853126,
-      "grad_norm": 0.011264491826295853,
-      "learning_rate": 0.00019999812935775143,
-      "loss": 46.0,
-      "step": 25475
-    },
-    {
-      "epoch": 1.9478181088365158,
-      "grad_norm": 0.001632630592212081,
-      "learning_rate": 0.0001999981292108303,
-      "loss": 46.0,
-      "step": 25476
-    },
-    {
-      "epoch": 1.9478945658199056,
-      "grad_norm": 0.0018762430408969522,
-      "learning_rate": 0.00019999812906390342,
-      "loss": 46.0,
-      "step": 25477
-    },
-    {
-      "epoch": 1.9479710228032951,
-      "grad_norm": 0.0006108759553171694,
-      "learning_rate": 0.00019999812891697077,
-      "loss": 46.0,
-      "step": 25478
-    },
-    {
-      "epoch": 1.948047479786685,
-      "grad_norm": 0.006483633536845446,
-      "learning_rate": 0.00019999812877003235,
-      "loss": 46.0,
-      "step": 25479
-    },
-    {
-      "epoch": 1.9481239367700747,
-      "grad_norm": 0.001171313808299601,
-      "learning_rate": 0.00019999812862308812,
-      "loss": 46.0,
-      "step": 25480
-    },
-    {
-      "epoch": 1.9482003937534644,
-      "grad_norm": 0.00031468638917431235,
-      "learning_rate": 0.00019999812847613818,
-      "loss": 46.0,
-      "step": 25481
-    },
-    {
-      "epoch": 1.9482768507368542,
-      "grad_norm": 0.002096825046464801,
-      "learning_rate": 0.00019999812832918244,
-      "loss": 46.0,
-      "step": 25482
-    },
-    {
-      "epoch": 1.948353307720244,
-      "grad_norm": 0.0009315012721344829,
-      "learning_rate": 0.00019999812818222092,
-      "loss": 46.0,
-      "step": 25483
-    },
-    {
-      "epoch": 1.9484297647036337,
-      "grad_norm": 0.0004553587350528687,
-      "learning_rate": 0.00019999812803525366,
-      "loss": 46.0,
-      "step": 25484
-    },
-    {
-      "epoch": 1.9485062216870235,
-      "grad_norm": 0.0005571578512899578,
-      "learning_rate": 0.00019999812788828062,
-      "loss": 46.0,
-      "step": 25485
-    },
-    {
-      "epoch": 1.948582678670413,
-      "grad_norm": 0.0007103788084350526,
-      "learning_rate": 0.0001999981277413018,
-      "loss": 46.0,
-      "step": 25486
-    },
-    {
-      "epoch": 1.9486591356538028,
-      "grad_norm": 0.001896351808682084,
-      "learning_rate": 0.00019999812759431725,
-      "loss": 46.0,
-      "step": 25487
-    },
-    {
-      "epoch": 1.9487355926371925,
-      "grad_norm": 0.0005857925862073898,
-      "learning_rate": 0.00019999812744732687,
-      "loss": 46.0,
-      "step": 25488
-    },
-    {
-      "epoch": 1.948812049620582,
-      "grad_norm": 0.0006940364255569875,
-      "learning_rate": 0.00019999812730033076,
-      "loss": 46.0,
-      "step": 25489
-    },
-    {
-      "epoch": 1.9488885066039718,
-      "grad_norm": 0.0014010635204613209,
-      "learning_rate": 0.00019999812715332886,
-      "loss": 46.0,
-      "step": 25490
-    },
-    {
-      "epoch": 1.9489649635873616,
-      "grad_norm": 0.0005416049971245229,
-      "learning_rate": 0.0001999981270063212,
-      "loss": 46.0,
-      "step": 25491
-    },
-    {
-      "epoch": 1.9490414205707514,
-      "grad_norm": 0.003481562715023756,
-      "learning_rate": 0.00019999812685930776,
-      "loss": 46.0,
-      "step": 25492
-    },
-    {
-      "epoch": 1.9491178775541411,
-      "grad_norm": 0.0008785219397395849,
-      "learning_rate": 0.00019999812671228856,
-      "loss": 46.0,
-      "step": 25493
-    },
-    {
-      "epoch": 1.9491943345375309,
-      "grad_norm": 0.0005087285535410047,
-      "learning_rate": 0.0001999981265652636,
-      "loss": 46.0,
-      "step": 25494
-    },
-    {
-      "epoch": 1.9492707915209206,
-      "grad_norm": 0.0005706212832592428,
-      "learning_rate": 0.00019999812641823288,
-      "loss": 46.0,
-      "step": 25495
-    },
-    {
-      "epoch": 1.9493472485043104,
-      "grad_norm": 0.0011976402020081878,
-      "learning_rate": 0.00019999812627119639,
-      "loss": 46.0,
-      "step": 25496
-    },
-    {
-      "epoch": 1.9494237054877,
-      "grad_norm": 0.0029671434313058853,
-      "learning_rate": 0.0001999981261241541,
-      "loss": 46.0,
-      "step": 25497
-    },
-    {
-      "epoch": 1.9495001624710897,
-      "grad_norm": 0.000705836690030992,
-      "learning_rate": 0.00019999812597710606,
-      "loss": 46.0,
-      "step": 25498
-    },
-    {
-      "epoch": 1.9495766194544795,
-      "grad_norm": 0.0009436235413886607,
-      "learning_rate": 0.00019999812583005225,
-      "loss": 46.0,
-      "step": 25499
-    },
-    {
-      "epoch": 1.949653076437869,
-      "grad_norm": 0.001094362000003457,
-      "learning_rate": 0.00019999812568299266,
-      "loss": 46.0,
-      "step": 25500
-    },
-    {
-      "epoch": 1.9497295334212588,
-      "grad_norm": 0.0011653897818177938,
-      "learning_rate": 0.0001999981255359273,
-      "loss": 46.0,
-      "step": 25501
-    },
-    {
-      "epoch": 1.9498059904046485,
-      "grad_norm": 0.001219184254296124,
-      "learning_rate": 0.00019999812538885618,
-      "loss": 46.0,
-      "step": 25502
-    },
-    {
-      "epoch": 1.9498824473880383,
-      "grad_norm": 0.0019858512096107006,
-      "learning_rate": 0.0001999981252417793,
-      "loss": 46.0,
-      "step": 25503
-    },
-    {
-      "epoch": 1.949958904371428,
-      "grad_norm": 0.0009545744978822768,
-      "learning_rate": 0.00019999812509469665,
-      "loss": 46.0,
-      "step": 25504
-    },
-    {
-      "epoch": 1.9500353613548178,
-      "grad_norm": 0.0014235759153962135,
-      "learning_rate": 0.0001999981249476082,
-      "loss": 46.0,
-      "step": 25505
-    },
-    {
-      "epoch": 1.9501118183382076,
-      "grad_norm": 0.0008185433689504862,
-      "learning_rate": 0.000199998124800514,
-      "loss": 46.0,
-      "step": 25506
-    },
-    {
-      "epoch": 1.9501882753215973,
-      "grad_norm": 0.0014334002044051886,
-      "learning_rate": 0.00019999812465341403,
-      "loss": 46.0,
-      "step": 25507
-    },
-    {
-      "epoch": 1.9502647323049869,
-      "grad_norm": 0.0012235109461471438,
-      "learning_rate": 0.0001999981245063083,
-      "loss": 46.0,
-      "step": 25508
-    },
-    {
-      "epoch": 1.9503411892883766,
-      "grad_norm": 0.002971385605633259,
-      "learning_rate": 0.0001999981243591968,
-      "loss": 46.0,
-      "step": 25509
-    },
-    {
-      "epoch": 1.9504176462717664,
-      "grad_norm": 0.0035227371845394373,
-      "learning_rate": 0.0001999981242120795,
-      "loss": 46.0,
-      "step": 25510
-    },
-    {
-      "epoch": 1.950494103255156,
-      "grad_norm": 0.001821687095798552,
-      "learning_rate": 0.00019999812406495645,
-      "loss": 46.0,
-      "step": 25511
-    },
-    {
-      "epoch": 1.9505705602385457,
-      "grad_norm": 0.010380851104855537,
-      "learning_rate": 0.00019999812391782764,
-      "loss": 46.0,
-      "step": 25512
-    },
-    {
-      "epoch": 1.9506470172219355,
-      "grad_norm": 0.008459806442260742,
-      "learning_rate": 0.00019999812377069308,
-      "loss": 46.0,
-      "step": 25513
-    },
-    {
-      "epoch": 1.9507234742053252,
-      "grad_norm": 0.0032521216198801994,
-      "learning_rate": 0.00019999812362355272,
-      "loss": 46.0,
-      "step": 25514
-    },
-    {
-      "epoch": 1.950799931188715,
-      "grad_norm": 0.009646959602832794,
-      "learning_rate": 0.0001999981234764066,
-      "loss": 46.0,
-      "step": 25515
-    },
-    {
-      "epoch": 1.9508763881721047,
-      "grad_norm": 0.0008727413951419294,
-      "learning_rate": 0.0001999981233292547,
-      "loss": 46.0,
-      "step": 25516
-    },
-    {
-      "epoch": 1.9509528451554945,
-      "grad_norm": 0.0005606000777333975,
-      "learning_rate": 0.00019999812318209704,
-      "loss": 46.0,
-      "step": 25517
-    },
-    {
-      "epoch": 1.9510293021388843,
-      "grad_norm": 0.001337242778390646,
-      "learning_rate": 0.00019999812303493362,
-      "loss": 46.0,
-      "step": 25518
-    },
-    {
-      "epoch": 1.9511057591222738,
-      "grad_norm": 0.0008637694409117103,
-      "learning_rate": 0.0001999981228877644,
-      "loss": 46.0,
-      "step": 25519
-    },
-    {
-      "epoch": 1.9511822161056636,
-      "grad_norm": 0.0010775624541565776,
-      "learning_rate": 0.00019999812274058943,
-      "loss": 46.0,
-      "step": 25520
-    },
-    {
-      "epoch": 1.9512586730890533,
-      "grad_norm": 0.0016888573300093412,
-      "learning_rate": 0.00019999812259340868,
-      "loss": 46.0,
-      "step": 25521
-    },
-    {
-      "epoch": 1.9513351300724429,
-      "grad_norm": 0.0009854328818619251,
-      "learning_rate": 0.0001999981224462222,
-      "loss": 46.0,
-      "step": 25522
-    },
-    {
-      "epoch": 1.9514115870558326,
-      "grad_norm": 0.0014013720210641623,
-      "learning_rate": 0.0001999981222990299,
-      "loss": 46.0,
-      "step": 25523
-    },
-    {
-      "epoch": 1.9514880440392224,
-      "grad_norm": 0.0007134824991226196,
-      "learning_rate": 0.00019999812215183184,
-      "loss": 46.0,
-      "step": 25524
-    },
-    {
-      "epoch": 1.9515645010226121,
-      "grad_norm": 0.02319967746734619,
-      "learning_rate": 0.00019999812200462803,
-      "loss": 46.0,
-      "step": 25525
-    },
-    {
-      "epoch": 1.951640958006002,
-      "grad_norm": 0.001971699995920062,
-      "learning_rate": 0.00019999812185741848,
-      "loss": 46.0,
-      "step": 25526
-    },
-    {
-      "epoch": 1.9517174149893917,
-      "grad_norm": 0.0007795064593665302,
-      "learning_rate": 0.00019999812171020312,
-      "loss": 46.0,
-      "step": 25527
-    },
-    {
-      "epoch": 1.9517938719727814,
-      "grad_norm": 0.002493527950718999,
-      "learning_rate": 0.000199998121562982,
-      "loss": 46.0,
-      "step": 25528
-    },
-    {
-      "epoch": 1.9518703289561712,
-      "grad_norm": 0.0013603090774267912,
-      "learning_rate": 0.0001999981214157551,
-      "loss": 46.0,
-      "step": 25529
-    },
-    {
-      "epoch": 1.9519467859395607,
-      "grad_norm": 0.0010205727303400636,
-      "learning_rate": 0.0001999981212685224,
-      "loss": 46.0,
-      "step": 25530
-    },
-    {
-      "epoch": 1.9520232429229505,
-      "grad_norm": 0.0007125125266611576,
-      "learning_rate": 0.000199998121121284,
-      "loss": 46.0,
-      "step": 25531
-    },
-    {
-      "epoch": 1.9520996999063402,
-      "grad_norm": 0.0007476358441635966,
-      "learning_rate": 0.0001999981209740398,
-      "loss": 46.0,
-      "step": 25532
-    },
-    {
-      "epoch": 1.9521761568897298,
-      "grad_norm": 0.0009519326267763972,
-      "learning_rate": 0.00019999812082678982,
-      "loss": 46.0,
-      "step": 25533
-    },
-    {
-      "epoch": 1.9522526138731195,
-      "grad_norm": 0.0021328977309167385,
-      "learning_rate": 0.00019999812067953408,
-      "loss": 46.0,
-      "step": 25534
-    },
-    {
-      "epoch": 1.9523290708565093,
-      "grad_norm": 0.0036633401177823544,
-      "learning_rate": 0.00019999812053227256,
-      "loss": 46.0,
-      "step": 25535
-    },
-    {
-      "epoch": 1.952405527839899,
-      "grad_norm": 0.0006054287077859044,
-      "learning_rate": 0.0001999981203850053,
-      "loss": 46.0,
-      "step": 25536
-    },
-    {
-      "epoch": 1.9524819848232888,
-      "grad_norm": 0.0011634370312094688,
-      "learning_rate": 0.00019999812023773224,
-      "loss": 46.0,
-      "step": 25537
-    },
-    {
-      "epoch": 1.9525584418066786,
-      "grad_norm": 0.0015786882722750306,
-      "learning_rate": 0.0001999981200904534,
-      "loss": 46.0,
-      "step": 25538
-    },
-    {
-      "epoch": 1.9526348987900684,
-      "grad_norm": 0.002831559395417571,
-      "learning_rate": 0.00019999811994316882,
-      "loss": 46.0,
-      "step": 25539
-    },
-    {
-      "epoch": 1.9527113557734581,
-      "grad_norm": 0.0014576261164620519,
-      "learning_rate": 0.00019999811979587846,
-      "loss": 46.0,
-      "step": 25540
-    },
-    {
-      "epoch": 1.9527878127568477,
-      "grad_norm": 0.0006848961347714067,
-      "learning_rate": 0.00019999811964858233,
-      "loss": 46.0,
-      "step": 25541
-    },
-    {
-      "epoch": 1.9528642697402374,
-      "grad_norm": 0.0023620238061994314,
-      "learning_rate": 0.00019999811950128046,
-      "loss": 46.0,
-      "step": 25542
-    },
-    {
-      "epoch": 1.9529407267236272,
-      "grad_norm": 0.0006348474998958409,
-      "learning_rate": 0.00019999811935397278,
-      "loss": 46.0,
-      "step": 25543
-    },
-    {
-      "epoch": 1.9530171837070167,
-      "grad_norm": 0.007859431207180023,
-      "learning_rate": 0.00019999811920665933,
-      "loss": 46.0,
-      "step": 25544
-    },
-    {
-      "epoch": 1.9530936406904065,
-      "grad_norm": 0.0006576450541615486,
-      "learning_rate": 0.00019999811905934014,
-      "loss": 46.0,
-      "step": 25545
-    },
-    {
-      "epoch": 1.9531700976737962,
-      "grad_norm": 0.001759189646691084,
-      "learning_rate": 0.00019999811891201517,
-      "loss": 46.0,
-      "step": 25546
-    },
-    {
-      "epoch": 1.953246554657186,
-      "grad_norm": 0.0005286120576784015,
-      "learning_rate": 0.00019999811876468443,
-      "loss": 46.0,
-      "step": 25547
-    },
-    {
-      "epoch": 1.9533230116405758,
-      "grad_norm": 0.005120165646076202,
-      "learning_rate": 0.0001999981186173479,
-      "loss": 46.0,
-      "step": 25548
-    },
-    {
-      "epoch": 1.9533994686239655,
-      "grad_norm": 0.0009879935532808304,
-      "learning_rate": 0.00019999811847000562,
-      "loss": 46.0,
-      "step": 25549
-    },
-    {
-      "epoch": 1.9534759256073553,
-      "grad_norm": 0.0011777252657338977,
-      "learning_rate": 0.00019999811832265759,
-      "loss": 46.0,
-      "step": 25550
-    },
-    {
-      "epoch": 1.953552382590745,
-      "grad_norm": 0.000631751143373549,
-      "learning_rate": 0.00019999811817530375,
-      "loss": 46.0,
-      "step": 25551
-    },
-    {
-      "epoch": 1.9536288395741346,
-      "grad_norm": 0.0005040019750595093,
-      "learning_rate": 0.00019999811802794417,
-      "loss": 46.0,
-      "step": 25552
-    },
-    {
-      "epoch": 1.9537052965575243,
-      "grad_norm": 0.0011498959502205253,
-      "learning_rate": 0.0001999981178805788,
-      "loss": 46.0,
-      "step": 25553
-    },
-    {
-      "epoch": 1.9537817535409139,
-      "grad_norm": 0.0008133918745443225,
-      "learning_rate": 0.00019999811773320768,
-      "loss": 46.0,
-      "step": 25554
-    },
-    {
-      "epoch": 1.9538582105243036,
-      "grad_norm": 0.0009214473539032042,
-      "learning_rate": 0.0001999981175858308,
-      "loss": 46.0,
-      "step": 25555
-    },
-    {
-      "epoch": 1.9539346675076934,
-      "grad_norm": 0.000881068641319871,
-      "learning_rate": 0.00019999811743844813,
-      "loss": 46.0,
-      "step": 25556
-    },
-    {
-      "epoch": 1.9540111244910832,
-      "grad_norm": 0.0004997215000912547,
-      "learning_rate": 0.00019999811729105968,
-      "loss": 46.0,
-      "step": 25557
-    },
-    {
-      "epoch": 1.954087581474473,
-      "grad_norm": 0.01997225731611252,
-      "learning_rate": 0.0001999981171436655,
-      "loss": 46.0,
-      "step": 25558
-    },
-    {
-      "epoch": 1.9541640384578627,
-      "grad_norm": 0.001274542766623199,
-      "learning_rate": 0.00019999811699626552,
-      "loss": 46.0,
-      "step": 25559
-    },
-    {
-      "epoch": 1.9542404954412524,
-      "grad_norm": 0.009379551745951176,
-      "learning_rate": 0.00019999811684885975,
-      "loss": 46.0,
-      "step": 25560
-    },
-    {
-      "epoch": 1.9543169524246422,
-      "grad_norm": 0.0034876237623393536,
-      "learning_rate": 0.00019999811670144826,
-      "loss": 46.0,
-      "step": 25561
-    },
-    {
-      "epoch": 1.954393409408032,
-      "grad_norm": 0.000805043033324182,
-      "learning_rate": 0.00019999811655403094,
-      "loss": 46.0,
-      "step": 25562
-    },
-    {
-      "epoch": 1.9544698663914215,
-      "grad_norm": 0.0019923821091651917,
-      "learning_rate": 0.0001999981164066079,
-      "loss": 46.0,
-      "step": 25563
-    },
-    {
-      "epoch": 1.9545463233748113,
-      "grad_norm": 0.0010309857316315174,
-      "learning_rate": 0.00019999811625917907,
-      "loss": 46.0,
-      "step": 25564
-    },
-    {
-      "epoch": 1.9546227803582008,
-      "grad_norm": 0.01174786314368248,
-      "learning_rate": 0.0001999981161117445,
-      "loss": 46.0,
-      "step": 25565
-    },
-    {
-      "epoch": 1.9546992373415906,
-      "grad_norm": 0.0006500668241642416,
-      "learning_rate": 0.00019999811596430416,
-      "loss": 46.0,
-      "step": 25566
-    },
-    {
-      "epoch": 1.9547756943249803,
-      "grad_norm": 0.0038077477365732193,
-      "learning_rate": 0.000199998115816858,
-      "loss": 46.0,
-      "step": 25567
-    },
-    {
-      "epoch": 1.95485215130837,
-      "grad_norm": 0.0012108869850635529,
-      "learning_rate": 0.0001999981156694061,
-      "loss": 46.0,
-      "step": 25568
-    },
-    {
-      "epoch": 1.9549286082917599,
-      "grad_norm": 0.0035956581123173237,
-      "learning_rate": 0.00019999811552194846,
-      "loss": 46.0,
-      "step": 25569
-    },
-    {
-      "epoch": 1.9550050652751496,
-      "grad_norm": 0.0008804632816463709,
-      "learning_rate": 0.00019999811537448504,
-      "loss": 46.0,
-      "step": 25570
-    },
-    {
-      "epoch": 1.9550815222585394,
-      "grad_norm": 0.0004633062344510108,
-      "learning_rate": 0.0001999981152270158,
-      "loss": 46.0,
-      "step": 25571
-    },
-    {
-      "epoch": 1.9551579792419291,
-      "grad_norm": 0.0023061796091496944,
-      "learning_rate": 0.00019999811507954082,
-      "loss": 46.0,
-      "step": 25572
-    },
-    {
-      "epoch": 1.955234436225319,
-      "grad_norm": 0.001015381538309157,
-      "learning_rate": 0.00019999811493206008,
-      "loss": 46.0,
-      "step": 25573
-    },
-    {
-      "epoch": 1.9553108932087084,
-      "grad_norm": 0.0008870554156601429,
-      "learning_rate": 0.0001999981147845736,
-      "loss": 46.0,
-      "step": 25574
-    },
-    {
-      "epoch": 1.9553873501920982,
-      "grad_norm": 0.00047566034481860697,
-      "learning_rate": 0.0001999981146370813,
-      "loss": 46.0,
-      "step": 25575
-    },
-    {
-      "epoch": 1.9554638071754877,
-      "grad_norm": 0.0008879919187165797,
-      "learning_rate": 0.00019999811448958324,
-      "loss": 46.0,
-      "step": 25576
-    },
-    {
-      "epoch": 1.9555402641588775,
-      "grad_norm": 0.0009728758595883846,
-      "learning_rate": 0.0001999981143420794,
-      "loss": 46.0,
-      "step": 25577
-    },
-    {
-      "epoch": 1.9556167211422673,
-      "grad_norm": 0.000601473730057478,
-      "learning_rate": 0.00019999811419456982,
-      "loss": 46.0,
-      "step": 25578
-    },
-    {
-      "epoch": 1.955693178125657,
-      "grad_norm": 0.008816084824502468,
-      "learning_rate": 0.00019999811404705447,
-      "loss": 46.0,
-      "step": 25579
-    },
-    {
-      "epoch": 1.9557696351090468,
-      "grad_norm": 0.0013176584616303444,
-      "learning_rate": 0.0001999981138995333,
-      "loss": 46.0,
-      "step": 25580
-    },
-    {
-      "epoch": 1.9558460920924365,
-      "grad_norm": 0.0014934474602341652,
-      "learning_rate": 0.00019999811375200644,
-      "loss": 46.0,
-      "step": 25581
-    },
-    {
-      "epoch": 1.9559225490758263,
-      "grad_norm": 0.0017329715192317963,
-      "learning_rate": 0.00019999811360447376,
-      "loss": 46.0,
-      "step": 25582
-    },
-    {
-      "epoch": 1.955999006059216,
-      "grad_norm": 0.002628705697134137,
-      "learning_rate": 0.0001999981134569353,
-      "loss": 46.0,
-      "step": 25583
-    },
-    {
-      "epoch": 1.9560754630426056,
-      "grad_norm": 0.002184395445510745,
-      "learning_rate": 0.00019999811330939112,
-      "loss": 46.0,
-      "step": 25584
-    },
-    {
-      "epoch": 1.9561519200259954,
-      "grad_norm": 0.0011793995508924127,
-      "learning_rate": 0.00019999811316184115,
-      "loss": 46.0,
-      "step": 25585
-    },
-    {
-      "epoch": 1.9562283770093851,
-      "grad_norm": 0.003846992738544941,
-      "learning_rate": 0.00019999811301428538,
-      "loss": 46.0,
-      "step": 25586
-    },
-    {
-      "epoch": 1.9563048339927747,
-      "grad_norm": 0.0016753869131207466,
-      "learning_rate": 0.00019999811286672387,
-      "loss": 46.0,
-      "step": 25587
-    },
-    {
-      "epoch": 1.9563812909761644,
-      "grad_norm": 0.0011474102502688766,
-      "learning_rate": 0.00019999811271915658,
-      "loss": 46.0,
-      "step": 25588
-    },
-    {
-      "epoch": 1.9564577479595542,
-      "grad_norm": 0.003411001991480589,
-      "learning_rate": 0.00019999811257158352,
-      "loss": 46.0,
-      "step": 25589
-    },
-    {
-      "epoch": 1.956534204942944,
-      "grad_norm": 0.0010889185359701514,
-      "learning_rate": 0.0001999981124240047,
-      "loss": 46.0,
-      "step": 25590
-    },
-    {
-      "epoch": 1.9566106619263337,
-      "grad_norm": 0.0026897569186985493,
-      "learning_rate": 0.0001999981122764201,
-      "loss": 46.0,
-      "step": 25591
-    },
-    {
-      "epoch": 1.9566871189097235,
-      "grad_norm": 0.0013514735037460923,
-      "learning_rate": 0.00019999811212882975,
-      "loss": 46.0,
-      "step": 25592
-    },
-    {
-      "epoch": 1.9567635758931132,
-      "grad_norm": 0.006613563280552626,
-      "learning_rate": 0.00019999811198123362,
-      "loss": 46.0,
-      "step": 25593
-    },
-    {
-      "epoch": 1.956840032876503,
-      "grad_norm": 0.0008116142707876861,
-      "learning_rate": 0.00019999811183363174,
-      "loss": 46.0,
-      "step": 25594
-    },
-    {
-      "epoch": 1.9569164898598925,
-      "grad_norm": 0.0009808239992707968,
-      "learning_rate": 0.00019999811168602401,
-      "loss": 46.0,
-      "step": 25595
-    },
-    {
-      "epoch": 1.9569929468432823,
-      "grad_norm": 0.003019998548552394,
-      "learning_rate": 0.0001999981115384106,
-      "loss": 46.0,
-      "step": 25596
-    },
-    {
-      "epoch": 1.957069403826672,
-      "grad_norm": 0.0008172235684469342,
-      "learning_rate": 0.0001999981113907914,
-      "loss": 46.0,
-      "step": 25597
-    },
-    {
-      "epoch": 1.9571458608100616,
-      "grad_norm": 0.000792684790212661,
-      "learning_rate": 0.0001999981112431664,
-      "loss": 46.0,
-      "step": 25598
-    },
-    {
-      "epoch": 1.9572223177934513,
-      "grad_norm": 0.0031228649895638227,
-      "learning_rate": 0.00019999811109553566,
-      "loss": 46.0,
-      "step": 25599
-    },
-    {
-      "epoch": 1.957298774776841,
-      "grad_norm": 0.001094255829229951,
-      "learning_rate": 0.00019999811094789912,
-      "loss": 46.0,
-      "step": 25600
-    },
-    {
-      "epoch": 1.9573752317602309,
-      "grad_norm": 0.0005204302724450827,
-      "learning_rate": 0.00019999811080025686,
-      "loss": 46.0,
-      "step": 25601
-    },
-    {
-      "epoch": 1.9574516887436206,
-      "grad_norm": 0.0014343796065077186,
-      "learning_rate": 0.0001999981106526088,
-      "loss": 46.0,
-      "step": 25602
-    },
-    {
-      "epoch": 1.9575281457270104,
-      "grad_norm": 0.0008985139429569244,
-      "learning_rate": 0.000199998110504955,
-      "loss": 46.0,
-      "step": 25603
-    },
-    {
-      "epoch": 1.9576046027104002,
-      "grad_norm": 0.0005287400563247502,
-      "learning_rate": 0.00019999811035729538,
-      "loss": 46.0,
-      "step": 25604
-    },
-    {
-      "epoch": 1.95768105969379,
-      "grad_norm": 0.018086234107613564,
-      "learning_rate": 0.00019999811020963003,
-      "loss": 46.0,
-      "step": 25605
-    },
-    {
-      "epoch": 1.9577575166771795,
-      "grad_norm": 0.001463517313823104,
-      "learning_rate": 0.00019999811006195887,
-      "loss": 46.0,
-      "step": 25606
-    },
-    {
-      "epoch": 1.9578339736605692,
-      "grad_norm": 0.0005041114636696875,
-      "learning_rate": 0.00019999810991428197,
-      "loss": 46.0,
-      "step": 25607
-    },
-    {
-      "epoch": 1.957910430643959,
-      "grad_norm": 0.0006080283783376217,
-      "learning_rate": 0.00019999810976659933,
-      "loss": 46.0,
-      "step": 25608
-    },
-    {
-      "epoch": 1.9579868876273485,
-      "grad_norm": 0.0012026106705889106,
-      "learning_rate": 0.00019999810961891088,
-      "loss": 46.0,
-      "step": 25609
-    },
-    {
-      "epoch": 1.9580633446107383,
-      "grad_norm": 0.0009459232096560299,
-      "learning_rate": 0.00019999810947121666,
-      "loss": 46.0,
-      "step": 25610
-    },
-    {
-      "epoch": 1.958139801594128,
-      "grad_norm": 0.000572491786442697,
-      "learning_rate": 0.0001999981093235167,
-      "loss": 46.0,
-      "step": 25611
-    },
-    {
-      "epoch": 1.9582162585775178,
-      "grad_norm": 0.0014231883687898517,
-      "learning_rate": 0.00019999810917581092,
-      "loss": 46.0,
-      "step": 25612
-    },
-    {
-      "epoch": 1.9582927155609076,
-      "grad_norm": 0.0003486821660771966,
-      "learning_rate": 0.0001999981090280994,
-      "loss": 46.0,
-      "step": 25613
-    },
-    {
-      "epoch": 1.9583691725442973,
-      "grad_norm": 0.0021960726007819176,
-      "learning_rate": 0.00019999810888038212,
-      "loss": 46.0,
-      "step": 25614
-    },
-    {
-      "epoch": 1.958445629527687,
-      "grad_norm": 0.0007462743087671697,
-      "learning_rate": 0.00019999810873265906,
-      "loss": 46.0,
-      "step": 25615
-    },
-    {
-      "epoch": 1.9585220865110768,
-      "grad_norm": 0.0008867362048476934,
-      "learning_rate": 0.00019999810858493025,
-      "loss": 46.0,
-      "step": 25616
-    },
-    {
-      "epoch": 1.9585985434944664,
-      "grad_norm": 0.0010086533147841692,
-      "learning_rate": 0.00019999810843719564,
-      "loss": 46.0,
-      "step": 25617
-    },
-    {
-      "epoch": 1.9586750004778561,
-      "grad_norm": 0.0031216300558298826,
-      "learning_rate": 0.00019999810828945526,
-      "loss": 46.0,
-      "step": 25618
-    },
-    {
-      "epoch": 1.958751457461246,
-      "grad_norm": 0.001055656815879047,
-      "learning_rate": 0.00019999810814170913,
-      "loss": 46.0,
-      "step": 25619
-    },
-    {
-      "epoch": 1.9588279144446354,
-      "grad_norm": 0.002654892159625888,
-      "learning_rate": 0.00019999810799395723,
-      "loss": 46.0,
-      "step": 25620
-    },
-    {
-      "epoch": 1.9589043714280252,
-      "grad_norm": 0.0010736286640167236,
-      "learning_rate": 0.00019999810784619956,
-      "loss": 46.0,
-      "step": 25621
-    },
-    {
-      "epoch": 1.958980828411415,
-      "grad_norm": 0.000880467938259244,
-      "learning_rate": 0.0001999981076984361,
-      "loss": 46.0,
-      "step": 25622
-    },
-    {
-      "epoch": 1.9590572853948047,
-      "grad_norm": 0.0010455846786499023,
-      "learning_rate": 0.0001999981075506669,
-      "loss": 46.0,
-      "step": 25623
-    },
-    {
-      "epoch": 1.9591337423781945,
-      "grad_norm": 0.0007909988053143024,
-      "learning_rate": 0.0001999981074028919,
-      "loss": 46.0,
-      "step": 25624
-    },
-    {
-      "epoch": 1.9592101993615842,
-      "grad_norm": 0.0011371668661013246,
-      "learning_rate": 0.00019999810725511116,
-      "loss": 46.0,
-      "step": 25625
-    },
-    {
-      "epoch": 1.959286656344974,
-      "grad_norm": 0.0013723236043006182,
-      "learning_rate": 0.00019999810710732464,
-      "loss": 46.0,
-      "step": 25626
-    },
-    {
-      "epoch": 1.9593631133283638,
-      "grad_norm": 0.0006425781175494194,
-      "learning_rate": 0.00019999810695953235,
-      "loss": 46.0,
-      "step": 25627
-    },
-    {
-      "epoch": 1.9594395703117533,
-      "grad_norm": 0.0040087648667395115,
-      "learning_rate": 0.0001999981068117343,
-      "loss": 46.0,
-      "step": 25628
-    },
-    {
-      "epoch": 1.959516027295143,
-      "grad_norm": 0.0013753294479101896,
-      "learning_rate": 0.0001999981066639305,
-      "loss": 46.0,
-      "step": 25629
-    },
-    {
-      "epoch": 1.9595924842785328,
-      "grad_norm": 0.0009036906412802637,
-      "learning_rate": 0.00019999810651612088,
-      "loss": 46.0,
-      "step": 25630
-    },
-    {
-      "epoch": 1.9596689412619224,
-      "grad_norm": 0.0013251419877633452,
-      "learning_rate": 0.0001999981063683055,
-      "loss": 46.0,
-      "step": 25631
-    },
-    {
-      "epoch": 1.9597453982453121,
-      "grad_norm": 0.0032385559752583504,
-      "learning_rate": 0.00019999810622048437,
-      "loss": 46.0,
-      "step": 25632
-    },
-    {
-      "epoch": 1.959821855228702,
-      "grad_norm": 0.0002641786413732916,
-      "learning_rate": 0.00019999810607265744,
-      "loss": 46.0,
-      "step": 25633
-    },
-    {
-      "epoch": 1.9598983122120917,
-      "grad_norm": 0.001194876036606729,
-      "learning_rate": 0.00019999810592482477,
-      "loss": 46.0,
-      "step": 25634
-    },
-    {
-      "epoch": 1.9599747691954814,
-      "grad_norm": 0.004771309904754162,
-      "learning_rate": 0.00019999810577698635,
-      "loss": 46.0,
-      "step": 25635
-    },
-    {
-      "epoch": 1.9600512261788712,
-      "grad_norm": 0.0019493907457217574,
-      "learning_rate": 0.00019999810562914213,
-      "loss": 46.0,
-      "step": 25636
-    },
-    {
-      "epoch": 1.960127683162261,
-      "grad_norm": 0.0012141424231231213,
-      "learning_rate": 0.00019999810548129214,
-      "loss": 46.0,
-      "step": 25637
-    },
-    {
-      "epoch": 1.9602041401456507,
-      "grad_norm": 0.0014177298871800303,
-      "learning_rate": 0.0001999981053334364,
-      "loss": 46.0,
-      "step": 25638
-    },
-    {
-      "epoch": 1.9602805971290402,
-      "grad_norm": 0.0011091342894360423,
-      "learning_rate": 0.00019999810518557488,
-      "loss": 46.0,
-      "step": 25639
-    },
-    {
-      "epoch": 1.96035705411243,
-      "grad_norm": 0.001043921452946961,
-      "learning_rate": 0.00019999810503770757,
-      "loss": 46.0,
-      "step": 25640
-    },
-    {
-      "epoch": 1.9604335110958198,
-      "grad_norm": 0.0007383283227682114,
-      "learning_rate": 0.0001999981048898345,
-      "loss": 46.0,
-      "step": 25641
-    },
-    {
-      "epoch": 1.9605099680792093,
-      "grad_norm": 0.0012561003677546978,
-      "learning_rate": 0.00019999810474195567,
-      "loss": 46.0,
-      "step": 25642
-    },
-    {
-      "epoch": 1.960586425062599,
-      "grad_norm": 0.0009179412736557424,
-      "learning_rate": 0.0001999981045940711,
-      "loss": 46.0,
-      "step": 25643
-    },
-    {
-      "epoch": 1.9606628820459888,
-      "grad_norm": 0.0041306293569505215,
-      "learning_rate": 0.0001999981044461807,
-      "loss": 46.0,
-      "step": 25644
-    },
-    {
-      "epoch": 1.9607393390293786,
-      "grad_norm": 0.004089896101504564,
-      "learning_rate": 0.0001999981042982846,
-      "loss": 46.0,
-      "step": 25645
-    },
-    {
-      "epoch": 1.9608157960127683,
-      "grad_norm": 0.0046420409344136715,
-      "learning_rate": 0.00019999810415038266,
-      "loss": 46.0,
-      "step": 25646
-    },
-    {
-      "epoch": 1.960892252996158,
-      "grad_norm": 0.0018208050169050694,
-      "learning_rate": 0.00019999810400247499,
-      "loss": 46.0,
-      "step": 25647
-    },
-    {
-      "epoch": 1.9609687099795479,
-      "grad_norm": 0.0018417071551084518,
-      "learning_rate": 0.00019999810385456154,
-      "loss": 46.0,
-      "step": 25648
-    },
-    {
-      "epoch": 1.9610451669629376,
-      "grad_norm": 0.002569543430581689,
-      "learning_rate": 0.00019999810370664235,
-      "loss": 46.0,
-      "step": 25649
-    },
-    {
-      "epoch": 1.9611216239463272,
-      "grad_norm": 0.0021901200525462627,
-      "learning_rate": 0.00019999810355871732,
-      "loss": 46.0,
-      "step": 25650
-    },
-    {
-      "epoch": 1.961198080929717,
-      "grad_norm": 0.0031648639123886824,
-      "learning_rate": 0.00019999810341078659,
-      "loss": 46.0,
-      "step": 25651
-    },
-    {
-      "epoch": 1.9612745379131067,
-      "grad_norm": 0.0003894160327035934,
-      "learning_rate": 0.00019999810326285007,
-      "loss": 46.0,
-      "step": 25652
-    },
-    {
-      "epoch": 1.9613509948964962,
-      "grad_norm": 0.0010531754232943058,
-      "learning_rate": 0.00019999810311490776,
-      "loss": 46.0,
-      "step": 25653
-    },
-    {
-      "epoch": 1.961427451879886,
-      "grad_norm": 0.0009474078542552888,
-      "learning_rate": 0.0001999981029669597,
-      "loss": 46.0,
-      "step": 25654
-    },
-    {
-      "epoch": 1.9615039088632757,
-      "grad_norm": 0.0012509784428402781,
-      "learning_rate": 0.0001999981028190059,
-      "loss": 46.0,
-      "step": 25655
-    },
-    {
-      "epoch": 1.9615803658466655,
-      "grad_norm": 0.001959295244887471,
-      "learning_rate": 0.00019999810267104628,
-      "loss": 46.0,
-      "step": 25656
-    },
-    {
-      "epoch": 1.9616568228300553,
-      "grad_norm": 0.0005629324004985392,
-      "learning_rate": 0.0001999981025230809,
-      "loss": 46.0,
-      "step": 25657
-    },
-    {
-      "epoch": 1.961733279813445,
-      "grad_norm": 0.0011021456448361278,
-      "learning_rate": 0.00019999810237510975,
-      "loss": 46.0,
-      "step": 25658
-    },
-    {
-      "epoch": 1.9618097367968348,
-      "grad_norm": 0.0025703106075525284,
-      "learning_rate": 0.00019999810222713288,
-      "loss": 46.0,
-      "step": 25659
-    },
-    {
-      "epoch": 1.9618861937802246,
-      "grad_norm": 0.0005849493318237364,
-      "learning_rate": 0.0001999981020791502,
-      "loss": 46.0,
-      "step": 25660
-    },
-    {
-      "epoch": 1.961962650763614,
-      "grad_norm": 0.0006567466771230102,
-      "learning_rate": 0.00019999810193116173,
-      "loss": 46.0,
-      "step": 25661
-    },
-    {
-      "epoch": 1.9620391077470039,
-      "grad_norm": 0.0011481598485261202,
-      "learning_rate": 0.00019999810178316754,
-      "loss": 46.0,
-      "step": 25662
-    },
-    {
-      "epoch": 1.9621155647303936,
-      "grad_norm": 0.0009814308723434806,
-      "learning_rate": 0.00019999810163516752,
-      "loss": 46.0,
-      "step": 25663
-    },
-    {
-      "epoch": 1.9621920217137832,
-      "grad_norm": 0.0015112621476873755,
-      "learning_rate": 0.00019999810148716178,
-      "loss": 46.0,
-      "step": 25664
-    },
-    {
-      "epoch": 1.962268478697173,
-      "grad_norm": 0.0013320047874003649,
-      "learning_rate": 0.00019999810133915027,
-      "loss": 46.0,
-      "step": 25665
-    },
-    {
-      "epoch": 1.9623449356805627,
-      "grad_norm": 0.0009068603976629674,
-      "learning_rate": 0.00019999810119113295,
-      "loss": 46.0,
-      "step": 25666
-    },
-    {
-      "epoch": 1.9624213926639524,
-      "grad_norm": 0.00048425066052004695,
-      "learning_rate": 0.0001999981010431099,
-      "loss": 46.0,
-      "step": 25667
-    },
-    {
-      "epoch": 1.9624978496473422,
-      "grad_norm": 0.0020001623779535294,
-      "learning_rate": 0.00019999810089508106,
-      "loss": 46.0,
-      "step": 25668
-    },
-    {
-      "epoch": 1.962574306630732,
-      "grad_norm": 0.0005820230580866337,
-      "learning_rate": 0.00019999810074704645,
-      "loss": 46.0,
-      "step": 25669
-    },
-    {
-      "epoch": 1.9626507636141217,
-      "grad_norm": 0.002070485381409526,
-      "learning_rate": 0.0001999981005990061,
-      "loss": 46.0,
-      "step": 25670
-    },
-    {
-      "epoch": 1.9627272205975115,
-      "grad_norm": 0.0020292168483138084,
-      "learning_rate": 0.00019999810045095995,
-      "loss": 46.0,
-      "step": 25671
-    },
-    {
-      "epoch": 1.962803677580901,
-      "grad_norm": 0.0013971275184303522,
-      "learning_rate": 0.00019999810030290802,
-      "loss": 46.0,
-      "step": 25672
-    },
-    {
-      "epoch": 1.9628801345642908,
-      "grad_norm": 0.0020395535975694656,
-      "learning_rate": 0.00019999810015485037,
-      "loss": 46.0,
-      "step": 25673
-    },
-    {
-      "epoch": 1.9629565915476805,
-      "grad_norm": 0.0011849283473566175,
-      "learning_rate": 0.0001999981000067869,
-      "loss": 46.0,
-      "step": 25674
-    },
-    {
-      "epoch": 1.96303304853107,
-      "grad_norm": 0.0007762099266983569,
-      "learning_rate": 0.00019999809985871768,
-      "loss": 46.0,
-      "step": 25675
-    },
-    {
-      "epoch": 1.9631095055144598,
-      "grad_norm": 0.00038909068098291755,
-      "learning_rate": 0.0001999980997106427,
-      "loss": 46.0,
-      "step": 25676
-    },
-    {
-      "epoch": 1.9631859624978496,
-      "grad_norm": 0.0008692687260918319,
-      "learning_rate": 0.00019999809956256192,
-      "loss": 46.0,
-      "step": 25677
-    },
-    {
-      "epoch": 1.9632624194812394,
-      "grad_norm": 0.001047069556079805,
-      "learning_rate": 0.0001999980994144754,
-      "loss": 46.0,
-      "step": 25678
-    },
-    {
-      "epoch": 1.9633388764646291,
-      "grad_norm": 0.0035722774919122458,
-      "learning_rate": 0.00019999809926638312,
-      "loss": 46.0,
-      "step": 25679
-    },
-    {
-      "epoch": 1.9634153334480189,
-      "grad_norm": 0.001829311833716929,
-      "learning_rate": 0.00019999809911828504,
-      "loss": 46.0,
-      "step": 25680
-    },
-    {
-      "epoch": 1.9634917904314086,
-      "grad_norm": 0.0009578315657563508,
-      "learning_rate": 0.0001999980989701812,
-      "loss": 46.0,
-      "step": 25681
-    },
-    {
-      "epoch": 1.9635682474147984,
-      "grad_norm": 0.0008192209643311799,
-      "learning_rate": 0.0001999980988220716,
-      "loss": 46.0,
-      "step": 25682
-    },
-    {
-      "epoch": 1.963644704398188,
-      "grad_norm": 0.0016062733484432101,
-      "learning_rate": 0.00019999809867395625,
-      "loss": 46.0,
-      "step": 25683
-    },
-    {
-      "epoch": 1.9637211613815777,
-      "grad_norm": 0.0005128085031174123,
-      "learning_rate": 0.0001999980985258351,
-      "loss": 46.0,
-      "step": 25684
-    },
-    {
-      "epoch": 1.9637976183649672,
-      "grad_norm": 0.0018017567927017808,
-      "learning_rate": 0.0001999980983777082,
-      "loss": 46.0,
-      "step": 25685
-    },
-    {
-      "epoch": 1.963874075348357,
-      "grad_norm": 0.0028801364824175835,
-      "learning_rate": 0.0001999980982295755,
-      "loss": 46.0,
-      "step": 25686
-    },
-    {
-      "epoch": 1.9639505323317468,
-      "grad_norm": 0.0012952187098562717,
-      "learning_rate": 0.00019999809808143705,
-      "loss": 46.0,
-      "step": 25687
-    },
-    {
-      "epoch": 1.9640269893151365,
-      "grad_norm": 0.003607895690947771,
-      "learning_rate": 0.00019999809793329286,
-      "loss": 46.0,
-      "step": 25688
-    },
-    {
-      "epoch": 1.9641034462985263,
-      "grad_norm": 0.0013448529643937945,
-      "learning_rate": 0.00019999809778514284,
-      "loss": 46.0,
-      "step": 25689
-    },
-    {
-      "epoch": 1.964179903281916,
-      "grad_norm": 0.0013092899462208152,
-      "learning_rate": 0.0001999980976369871,
-      "loss": 46.0,
-      "step": 25690
-    },
-    {
-      "epoch": 1.9642563602653058,
-      "grad_norm": 0.000752260850276798,
-      "learning_rate": 0.00019999809748882553,
-      "loss": 46.0,
-      "step": 25691
-    },
-    {
-      "epoch": 1.9643328172486956,
-      "grad_norm": 0.0008124150335788727,
-      "learning_rate": 0.00019999809734065827,
-      "loss": 46.0,
-      "step": 25692
-    },
-    {
-      "epoch": 1.9644092742320853,
-      "grad_norm": 0.0008994932286441326,
-      "learning_rate": 0.00019999809719248519,
-      "loss": 46.0,
-      "step": 25693
-    },
-    {
-      "epoch": 1.9644857312154749,
-      "grad_norm": 0.0005908635212108493,
-      "learning_rate": 0.00019999809704430635,
-      "loss": 46.0,
-      "step": 25694
-    },
-    {
-      "epoch": 1.9645621881988646,
-      "grad_norm": 0.0008434213232249022,
-      "learning_rate": 0.00019999809689612175,
-      "loss": 46.0,
-      "step": 25695
-    },
-    {
-      "epoch": 1.9646386451822542,
-      "grad_norm": 0.00044164498103782535,
-      "learning_rate": 0.00019999809674793137,
-      "loss": 46.0,
-      "step": 25696
-    },
-    {
-      "epoch": 1.964715102165644,
-      "grad_norm": 0.0006314634811133146,
-      "learning_rate": 0.00019999809659973524,
-      "loss": 46.0,
-      "step": 25697
-    },
-    {
-      "epoch": 1.9647915591490337,
-      "grad_norm": 0.0005994917592033744,
-      "learning_rate": 0.00019999809645153334,
-      "loss": 46.0,
-      "step": 25698
-    },
-    {
-      "epoch": 1.9648680161324235,
-      "grad_norm": 0.000701598240993917,
-      "learning_rate": 0.00019999809630332562,
-      "loss": 46.0,
-      "step": 25699
-    },
-    {
-      "epoch": 1.9649444731158132,
-      "grad_norm": 0.00297808856703341,
-      "learning_rate": 0.00019999809615511217,
-      "loss": 46.0,
-      "step": 25700
-    },
-    {
-      "epoch": 1.965020930099203,
-      "grad_norm": 0.00306686176918447,
-      "learning_rate": 0.00019999809600689295,
-      "loss": 46.0,
-      "step": 25701
-    },
-    {
-      "epoch": 1.9650973870825927,
-      "grad_norm": 0.0013077683979645371,
-      "learning_rate": 0.000199998095858668,
-      "loss": 46.0,
-      "step": 25702
-    },
-    {
-      "epoch": 1.9651738440659825,
-      "grad_norm": 0.0027012452483177185,
-      "learning_rate": 0.00019999809571043722,
-      "loss": 46.0,
-      "step": 25703
-    },
-    {
-      "epoch": 1.9652503010493723,
-      "grad_norm": 0.0015396863454952836,
-      "learning_rate": 0.00019999809556220068,
-      "loss": 46.0,
-      "step": 25704
-    },
-    {
-      "epoch": 1.9653267580327618,
-      "grad_norm": 0.0005602787714451551,
-      "learning_rate": 0.00019999809541395837,
-      "loss": 46.0,
-      "step": 25705
-    },
-    {
-      "epoch": 1.9654032150161516,
-      "grad_norm": 0.00037825407343916595,
-      "learning_rate": 0.0001999980952657103,
-      "loss": 46.0,
-      "step": 25706
-    },
-    {
-      "epoch": 1.965479671999541,
-      "grad_norm": 0.0006492449319921434,
-      "learning_rate": 0.00019999809511745648,
-      "loss": 46.0,
-      "step": 25707
-    },
-    {
-      "epoch": 1.9655561289829309,
-      "grad_norm": 0.0038903469685465097,
-      "learning_rate": 0.00019999809496919685,
-      "loss": 46.0,
-      "step": 25708
-    },
-    {
-      "epoch": 1.9656325859663206,
-      "grad_norm": 0.0016727959737181664,
-      "learning_rate": 0.0001999980948209315,
-      "loss": 46.0,
-      "step": 25709
-    },
-    {
-      "epoch": 1.9657090429497104,
-      "grad_norm": 0.0006891209632158279,
-      "learning_rate": 0.00019999809467266034,
-      "loss": 46.0,
-      "step": 25710
-    },
-    {
-      "epoch": 1.9657854999331001,
-      "grad_norm": 0.003447185968980193,
-      "learning_rate": 0.00019999809452438342,
-      "loss": 46.0,
-      "step": 25711
-    },
-    {
-      "epoch": 1.96586195691649,
-      "grad_norm": 0.0005791171570308506,
-      "learning_rate": 0.00019999809437610074,
-      "loss": 46.0,
-      "step": 25712
-    },
-    {
-      "epoch": 1.9659384138998797,
-      "grad_norm": 0.0030269778799265623,
-      "learning_rate": 0.0001999980942278123,
-      "loss": 46.0,
-      "step": 25713
-    },
-    {
-      "epoch": 1.9660148708832694,
-      "grad_norm": 0.0016051674028858542,
-      "learning_rate": 0.00019999809407951808,
-      "loss": 46.0,
-      "step": 25714
-    },
-    {
-      "epoch": 1.966091327866659,
-      "grad_norm": 0.00045554747339338064,
-      "learning_rate": 0.0001999980939312181,
-      "loss": 46.0,
-      "step": 25715
-    },
-    {
-      "epoch": 1.9661677848500487,
-      "grad_norm": 0.0038114527706056833,
-      "learning_rate": 0.00019999809378291232,
-      "loss": 46.0,
-      "step": 25716
-    },
-    {
-      "epoch": 1.9662442418334385,
-      "grad_norm": 0.001503113890066743,
-      "learning_rate": 0.00019999809363460078,
-      "loss": 46.0,
-      "step": 25717
-    },
-    {
-      "epoch": 1.966320698816828,
-      "grad_norm": 0.0023027423303574324,
-      "learning_rate": 0.0001999980934862835,
-      "loss": 46.0,
-      "step": 25718
-    },
-    {
-      "epoch": 1.9663971558002178,
-      "grad_norm": 0.003781256265938282,
-      "learning_rate": 0.0001999980933379604,
-      "loss": 46.0,
-      "step": 25719
-    },
-    {
-      "epoch": 1.9664736127836075,
-      "grad_norm": 0.006362109910696745,
-      "learning_rate": 0.00019999809318963155,
-      "loss": 46.0,
-      "step": 25720
-    },
-    {
-      "epoch": 1.9665500697669973,
-      "grad_norm": 0.000668186810798943,
-      "learning_rate": 0.00019999809304129698,
-      "loss": 46.0,
-      "step": 25721
-    },
-    {
-      "epoch": 1.966626526750387,
-      "grad_norm": 0.0009403019212186337,
-      "learning_rate": 0.0001999980928929566,
-      "loss": 46.0,
-      "step": 25722
-    },
-    {
-      "epoch": 1.9667029837337768,
-      "grad_norm": 0.0006688606808893383,
-      "learning_rate": 0.00019999809274461045,
-      "loss": 46.0,
-      "step": 25723
-    },
-    {
-      "epoch": 1.9667794407171666,
-      "grad_norm": 0.0005047800950706005,
-      "learning_rate": 0.00019999809259625852,
-      "loss": 46.0,
-      "step": 25724
-    },
-    {
-      "epoch": 1.9668558977005564,
-      "grad_norm": 0.0005354401655495167,
-      "learning_rate": 0.00019999809244790085,
-      "loss": 46.0,
-      "step": 25725
-    },
-    {
-      "epoch": 1.966932354683946,
-      "grad_norm": 0.000846469949465245,
-      "learning_rate": 0.00019999809229953738,
-      "loss": 46.0,
-      "step": 25726
-    },
-    {
-      "epoch": 1.9670088116673357,
-      "grad_norm": 0.0009723450057208538,
-      "learning_rate": 0.00019999809215116816,
-      "loss": 46.0,
-      "step": 25727
-    },
-    {
-      "epoch": 1.9670852686507254,
-      "grad_norm": 0.001135780243203044,
-      "learning_rate": 0.00019999809200279317,
-      "loss": 46.0,
-      "step": 25728
-    },
-    {
-      "epoch": 1.967161725634115,
-      "grad_norm": 0.0013109707506373525,
-      "learning_rate": 0.0001999980918544124,
-      "loss": 46.0,
-      "step": 25729
-    },
-    {
-      "epoch": 1.9672381826175047,
-      "grad_norm": 0.00041831049020402133,
-      "learning_rate": 0.0001999980917060259,
-      "loss": 46.0,
-      "step": 25730
-    },
-    {
-      "epoch": 1.9673146396008945,
-      "grad_norm": 0.002529071643948555,
-      "learning_rate": 0.00019999809155763355,
-      "loss": 46.0,
-      "step": 25731
-    },
-    {
-      "epoch": 1.9673910965842842,
-      "grad_norm": 0.0008151988731697202,
-      "learning_rate": 0.00019999809140923552,
-      "loss": 46.0,
-      "step": 25732
-    },
-    {
-      "epoch": 1.967467553567674,
-      "grad_norm": 0.003350807586684823,
-      "learning_rate": 0.00019999809126083166,
-      "loss": 46.0,
-      "step": 25733
-    },
-    {
-      "epoch": 1.9675440105510638,
-      "grad_norm": 0.0011093176435679197,
-      "learning_rate": 0.00019999809111242206,
-      "loss": 46.0,
-      "step": 25734
-    },
-    {
-      "epoch": 1.9676204675344535,
-      "grad_norm": 0.0016163561958819628,
-      "learning_rate": 0.00019999809096400668,
-      "loss": 46.0,
-      "step": 25735
-    },
-    {
-      "epoch": 1.9676969245178433,
-      "grad_norm": 0.0030409004539251328,
-      "learning_rate": 0.00019999809081558553,
-      "loss": 46.0,
-      "step": 25736
-    },
-    {
-      "epoch": 1.9677733815012328,
-      "grad_norm": 0.0003842986188828945,
-      "learning_rate": 0.00019999809066715863,
-      "loss": 46.0,
-      "step": 25737
-    },
-    {
-      "epoch": 1.9678498384846226,
-      "grad_norm": 0.000913542287889868,
-      "learning_rate": 0.00019999809051872593,
-      "loss": 46.0,
-      "step": 25738
-    },
-    {
-      "epoch": 1.9679262954680123,
-      "grad_norm": 0.0011681531323119998,
-      "learning_rate": 0.00019999809037028748,
-      "loss": 46.0,
-      "step": 25739
-    },
-    {
-      "epoch": 1.9680027524514019,
-      "grad_norm": 0.0010768186766654253,
-      "learning_rate": 0.00019999809022184324,
-      "loss": 46.0,
-      "step": 25740
-    },
-    {
-      "epoch": 1.9680792094347916,
-      "grad_norm": 0.00041841319762170315,
-      "learning_rate": 0.00019999809007339322,
-      "loss": 46.0,
-      "step": 25741
-    },
-    {
-      "epoch": 1.9681556664181814,
-      "grad_norm": 0.005378073081374168,
-      "learning_rate": 0.00019999808992493748,
-      "loss": 46.0,
-      "step": 25742
-    },
-    {
-      "epoch": 1.9682321234015712,
-      "grad_norm": 0.0010246969759464264,
-      "learning_rate": 0.00019999808977647594,
-      "loss": 46.0,
-      "step": 25743
-    },
-    {
-      "epoch": 1.968308580384961,
-      "grad_norm": 0.0008413073373958468,
-      "learning_rate": 0.00019999808962800866,
-      "loss": 46.0,
-      "step": 25744
-    },
-    {
-      "epoch": 1.9683850373683507,
-      "grad_norm": 0.0009757535299286246,
-      "learning_rate": 0.00019999808947953555,
-      "loss": 46.0,
-      "step": 25745
-    },
-    {
-      "epoch": 1.9684614943517404,
-      "grad_norm": 0.0008605967741459608,
-      "learning_rate": 0.00019999808933105672,
-      "loss": 46.0,
-      "step": 25746
-    },
-    {
-      "epoch": 1.9685379513351302,
-      "grad_norm": 0.0011325961677357554,
-      "learning_rate": 0.0001999980891825721,
-      "loss": 46.0,
-      "step": 25747
-    },
-    {
-      "epoch": 1.9686144083185197,
-      "grad_norm": 0.0015184168005362153,
-      "learning_rate": 0.00019999808903408173,
-      "loss": 46.0,
-      "step": 25748
-    },
-    {
-      "epoch": 1.9686908653019095,
-      "grad_norm": 0.0010799546726047993,
-      "learning_rate": 0.00019999808888558558,
-      "loss": 46.0,
-      "step": 25749
-    },
-    {
-      "epoch": 1.9687673222852993,
-      "grad_norm": 0.0009057698771357536,
-      "learning_rate": 0.00019999808873708369,
-      "loss": 46.0,
-      "step": 25750
-    },
-    {
-      "epoch": 1.9688437792686888,
-      "grad_norm": 0.0006330139003694057,
-      "learning_rate": 0.000199998088588576,
-      "loss": 46.0,
-      "step": 25751
-    },
-    {
-      "epoch": 1.9689202362520786,
-      "grad_norm": 0.0008669638191349804,
-      "learning_rate": 0.00019999808844006252,
-      "loss": 46.0,
-      "step": 25752
-    },
-    {
-      "epoch": 1.9689966932354683,
-      "grad_norm": 0.0012531676329672337,
-      "learning_rate": 0.00019999808829154327,
-      "loss": 46.0,
-      "step": 25753
-    },
-    {
-      "epoch": 1.969073150218858,
-      "grad_norm": 0.0035813641734421253,
-      "learning_rate": 0.0001999980881430183,
-      "loss": 46.0,
-      "step": 25754
-    },
-    {
-      "epoch": 1.9691496072022479,
-      "grad_norm": 0.0034570202697068453,
-      "learning_rate": 0.00019999808799448752,
-      "loss": 46.0,
-      "step": 25755
-    },
-    {
-      "epoch": 1.9692260641856376,
-      "grad_norm": 0.0009613910806365311,
-      "learning_rate": 0.000199998087845951,
-      "loss": 46.0,
-      "step": 25756
-    },
-    {
-      "epoch": 1.9693025211690274,
-      "grad_norm": 0.00043255550554022193,
-      "learning_rate": 0.0001999980876974087,
-      "loss": 46.0,
-      "step": 25757
-    },
-    {
-      "epoch": 1.9693789781524171,
-      "grad_norm": 0.0005464975256472826,
-      "learning_rate": 0.0001999980875488606,
-      "loss": 46.0,
-      "step": 25758
-    },
-    {
-      "epoch": 1.9694554351358067,
-      "grad_norm": 0.0006330720498226583,
-      "learning_rate": 0.00019999808740030678,
-      "loss": 46.0,
-      "step": 25759
-    },
-    {
-      "epoch": 1.9695318921191964,
-      "grad_norm": 0.000557834398932755,
-      "learning_rate": 0.00019999808725174715,
-      "loss": 46.0,
-      "step": 25760
-    },
-    {
-      "epoch": 1.9696083491025862,
-      "grad_norm": 0.0012844385346397758,
-      "learning_rate": 0.00019999808710318177,
-      "loss": 46.0,
-      "step": 25761
-    },
-    {
-      "epoch": 1.9696848060859757,
-      "grad_norm": 0.0005189888179302216,
-      "learning_rate": 0.00019999808695461062,
-      "loss": 46.0,
-      "step": 25762
-    },
-    {
-      "epoch": 1.9697612630693655,
-      "grad_norm": 0.0010427827946841717,
-      "learning_rate": 0.0001999980868060337,
-      "loss": 46.0,
-      "step": 25763
-    },
-    {
-      "epoch": 1.9698377200527553,
-      "grad_norm": 0.0019316133111715317,
-      "learning_rate": 0.00019999808665745102,
-      "loss": 46.0,
-      "step": 25764
-    },
-    {
-      "epoch": 1.969914177036145,
-      "grad_norm": 0.0006965315551497042,
-      "learning_rate": 0.00019999808650886255,
-      "loss": 46.0,
-      "step": 25765
-    },
-    {
-      "epoch": 1.9699906340195348,
-      "grad_norm": 0.0007156726205721498,
-      "learning_rate": 0.0001999980863602683,
-      "loss": 46.0,
-      "step": 25766
-    },
-    {
-      "epoch": 1.9700670910029245,
-      "grad_norm": 0.0015348537126556039,
-      "learning_rate": 0.00019999808621166834,
-      "loss": 46.0,
-      "step": 25767
-    },
-    {
-      "epoch": 1.9701435479863143,
-      "grad_norm": 0.0017303547356277704,
-      "learning_rate": 0.00019999808606306255,
-      "loss": 46.0,
-      "step": 25768
-    },
-    {
-      "epoch": 1.970220004969704,
-      "grad_norm": 0.0033142436295747757,
-      "learning_rate": 0.00019999808591445101,
-      "loss": 46.0,
-      "step": 25769
-    },
-    {
-      "epoch": 1.9702964619530936,
-      "grad_norm": 0.0009024566388688982,
-      "learning_rate": 0.00019999808576583373,
-      "loss": 46.0,
-      "step": 25770
-    },
-    {
-      "epoch": 1.9703729189364834,
-      "grad_norm": 0.0009344329009763896,
-      "learning_rate": 0.00019999808561721065,
-      "loss": 46.0,
-      "step": 25771
-    },
-    {
-      "epoch": 1.9704493759198731,
-      "grad_norm": 0.0073862276040017605,
-      "learning_rate": 0.00019999808546858182,
-      "loss": 46.0,
-      "step": 25772
-    },
-    {
-      "epoch": 1.9705258329032627,
-      "grad_norm": 0.001017266302369535,
-      "learning_rate": 0.00019999808531994718,
-      "loss": 46.0,
-      "step": 25773
-    },
-    {
-      "epoch": 1.9706022898866524,
-      "grad_norm": 0.0009426310425624251,
-      "learning_rate": 0.0001999980851713068,
-      "loss": 46.0,
-      "step": 25774
-    },
-    {
-      "epoch": 1.9706787468700422,
-      "grad_norm": 0.0008773712906986475,
-      "learning_rate": 0.00019999808502266063,
-      "loss": 46.0,
-      "step": 25775
-    },
-    {
-      "epoch": 1.970755203853432,
-      "grad_norm": 0.0019621788524091244,
-      "learning_rate": 0.0001999980848740087,
-      "loss": 46.0,
-      "step": 25776
-    },
-    {
-      "epoch": 1.9708316608368217,
-      "grad_norm": 0.001762914820574224,
-      "learning_rate": 0.00019999808472535103,
-      "loss": 46.0,
-      "step": 25777
-    },
-    {
-      "epoch": 1.9709081178202115,
-      "grad_norm": 0.002573672216385603,
-      "learning_rate": 0.00019999808457668756,
-      "loss": 46.0,
-      "step": 25778
-    },
-    {
-      "epoch": 1.9709845748036012,
-      "grad_norm": 0.0005982061848044395,
-      "learning_rate": 0.00019999808442801835,
-      "loss": 46.0,
-      "step": 25779
-    },
-    {
-      "epoch": 1.971061031786991,
-      "grad_norm": 0.0006087124347686768,
-      "learning_rate": 0.00019999808427934333,
-      "loss": 46.0,
-      "step": 25780
-    },
-    {
-      "epoch": 1.9711374887703805,
-      "grad_norm": 0.001752177020534873,
-      "learning_rate": 0.00019999808413066257,
-      "loss": 46.0,
-      "step": 25781
-    },
-    {
-      "epoch": 1.9712139457537703,
-      "grad_norm": 0.0007067126571200788,
-      "learning_rate": 0.00019999808398197603,
-      "loss": 46.0,
-      "step": 25782
-    },
-    {
-      "epoch": 1.97129040273716,
-      "grad_norm": 0.0009051482775248587,
-      "learning_rate": 0.00019999808383328374,
-      "loss": 46.0,
-      "step": 25783
-    },
-    {
-      "epoch": 1.9713668597205496,
-      "grad_norm": 0.0013950142310932279,
-      "learning_rate": 0.00019999808368458566,
-      "loss": 46.0,
-      "step": 25784
-    },
-    {
-      "epoch": 1.9714433167039394,
-      "grad_norm": 0.0012019522255286574,
-      "learning_rate": 0.0001999980835358818,
-      "loss": 46.0,
-      "step": 25785
-    },
-    {
-      "epoch": 1.9715197736873291,
-      "grad_norm": 0.0009348238236270845,
-      "learning_rate": 0.00019999808338717217,
-      "loss": 46.0,
-      "step": 25786
-    },
-    {
-      "epoch": 1.9715962306707189,
-      "grad_norm": 0.0006031130906194448,
-      "learning_rate": 0.0001999980832384568,
-      "loss": 46.0,
-      "step": 25787
-    },
-    {
-      "epoch": 1.9716726876541086,
-      "grad_norm": 0.0023562414571642876,
-      "learning_rate": 0.00019999808308973565,
-      "loss": 46.0,
-      "step": 25788
-    },
-    {
-      "epoch": 1.9717491446374984,
-      "grad_norm": 0.00122918002307415,
-      "learning_rate": 0.00019999808294100872,
-      "loss": 46.0,
-      "step": 25789
-    },
-    {
-      "epoch": 1.9718256016208882,
-      "grad_norm": 0.0011660020099952817,
-      "learning_rate": 0.00019999808279227602,
-      "loss": 46.0,
-      "step": 25790
-    },
-    {
-      "epoch": 1.971902058604278,
-      "grad_norm": 0.0034926109947264194,
-      "learning_rate": 0.00019999808264353755,
-      "loss": 46.0,
-      "step": 25791
-    },
-    {
-      "epoch": 1.9719785155876675,
-      "grad_norm": 0.0026699271984398365,
-      "learning_rate": 0.00019999808249479334,
-      "loss": 46.0,
-      "step": 25792
-    },
-    {
-      "epoch": 1.9720549725710572,
-      "grad_norm": 0.01537258829921484,
-      "learning_rate": 0.00019999808234604335,
-      "loss": 46.0,
-      "step": 25793
-    },
-    {
-      "epoch": 1.972131429554447,
-      "grad_norm": 0.0008090611663646996,
-      "learning_rate": 0.00019999808219728756,
-      "loss": 46.0,
-      "step": 25794
-    },
-    {
-      "epoch": 1.9722078865378365,
-      "grad_norm": 0.001403270522132516,
-      "learning_rate": 0.00019999808204852602,
-      "loss": 46.0,
-      "step": 25795
-    },
-    {
-      "epoch": 1.9722843435212263,
-      "grad_norm": 0.00156811170745641,
-      "learning_rate": 0.0001999980818997587,
-      "loss": 46.0,
-      "step": 25796
-    },
-    {
-      "epoch": 1.972360800504616,
-      "grad_norm": 0.008453596383333206,
-      "learning_rate": 0.00019999808175098562,
-      "loss": 46.0,
-      "step": 25797
-    },
-    {
-      "epoch": 1.9724372574880058,
-      "grad_norm": 0.002186944941058755,
-      "learning_rate": 0.0001999980816022068,
-      "loss": 46.0,
-      "step": 25798
-    },
-    {
-      "epoch": 1.9725137144713956,
-      "grad_norm": 0.0006994829163886607,
-      "learning_rate": 0.0001999980814534222,
-      "loss": 46.0,
-      "step": 25799
-    },
-    {
-      "epoch": 1.9725901714547853,
-      "grad_norm": 0.0025082130450755358,
-      "learning_rate": 0.0001999980813046318,
-      "loss": 46.0,
-      "step": 25800
-    },
-    {
-      "epoch": 1.972666628438175,
-      "grad_norm": 0.0007570331799797714,
-      "learning_rate": 0.00019999808115583564,
-      "loss": 46.0,
-      "step": 25801
-    },
-    {
-      "epoch": 1.9727430854215648,
-      "grad_norm": 0.001030400162562728,
-      "learning_rate": 0.0001999980810070337,
-      "loss": 46.0,
-      "step": 25802
-    },
-    {
-      "epoch": 1.9728195424049544,
-      "grad_norm": 0.0006028040661476552,
-      "learning_rate": 0.000199998080858226,
-      "loss": 46.0,
-      "step": 25803
-    },
-    {
-      "epoch": 1.9728959993883441,
-      "grad_norm": 0.0006911895470693707,
-      "learning_rate": 0.00019999808070941252,
-      "loss": 46.0,
-      "step": 25804
-    },
-    {
-      "epoch": 1.972972456371734,
-      "grad_norm": 0.0027786351274698973,
-      "learning_rate": 0.0001999980805605933,
-      "loss": 46.0,
-      "step": 25805
-    },
-    {
-      "epoch": 1.9730489133551234,
-      "grad_norm": 0.00045096265967004,
-      "learning_rate": 0.00019999808041176831,
-      "loss": 46.0,
-      "step": 25806
-    },
-    {
-      "epoch": 1.9731253703385132,
-      "grad_norm": 0.00123696424998343,
-      "learning_rate": 0.00019999808026293755,
-      "loss": 46.0,
-      "step": 25807
-    },
-    {
-      "epoch": 1.973201827321903,
-      "grad_norm": 0.00045669314567930996,
-      "learning_rate": 0.000199998080114101,
-      "loss": 46.0,
-      "step": 25808
-    },
-    {
-      "epoch": 1.9732782843052927,
-      "grad_norm": 0.0012601757189258933,
-      "learning_rate": 0.00019999807996525868,
-      "loss": 46.0,
-      "step": 25809
-    },
-    {
-      "epoch": 1.9733547412886825,
-      "grad_norm": 0.0023155435919761658,
-      "learning_rate": 0.0001999980798164106,
-      "loss": 46.0,
-      "step": 25810
-    },
-    {
-      "epoch": 1.9734311982720723,
-      "grad_norm": 0.0015559191815555096,
-      "learning_rate": 0.00019999807966755674,
-      "loss": 46.0,
-      "step": 25811
-    },
-    {
-      "epoch": 1.973507655255462,
-      "grad_norm": 0.00292005087248981,
-      "learning_rate": 0.00019999807951869713,
-      "loss": 46.0,
-      "step": 25812
-    },
-    {
-      "epoch": 1.9735841122388518,
-      "grad_norm": 0.0007657215464860201,
-      "learning_rate": 0.00019999807936983176,
-      "loss": 46.0,
-      "step": 25813
-    },
-    {
-      "epoch": 1.9736605692222413,
-      "grad_norm": 0.0013588722795248032,
-      "learning_rate": 0.00019999807922096058,
-      "loss": 46.0,
-      "step": 25814
-    },
-    {
-      "epoch": 1.973737026205631,
-      "grad_norm": 0.0012404138687998056,
-      "learning_rate": 0.00019999807907208368,
-      "loss": 46.0,
-      "step": 25815
-    },
-    {
-      "epoch": 1.9738134831890206,
-      "grad_norm": 0.006692041177302599,
-      "learning_rate": 0.00019999807892320096,
-      "loss": 46.0,
-      "step": 25816
-    },
-    {
-      "epoch": 1.9738899401724104,
-      "grad_norm": 0.002543903887271881,
-      "learning_rate": 0.0001999980787743125,
-      "loss": 46.0,
-      "step": 25817
-    },
-    {
-      "epoch": 1.9739663971558001,
-      "grad_norm": 0.0012056382838636637,
-      "learning_rate": 0.00019999807862541827,
-      "loss": 46.0,
-      "step": 25818
-    },
-    {
-      "epoch": 1.97404285413919,
-      "grad_norm": 0.0008124654996208847,
-      "learning_rate": 0.00019999807847651826,
-      "loss": 46.0,
-      "step": 25819
-    },
-    {
-      "epoch": 1.9741193111225797,
-      "grad_norm": 0.01230429857969284,
-      "learning_rate": 0.0001999980783276125,
-      "loss": 46.0,
-      "step": 25820
-    },
-    {
-      "epoch": 1.9741957681059694,
-      "grad_norm": 0.0009111386025324464,
-      "learning_rate": 0.00019999807817870093,
-      "loss": 46.0,
-      "step": 25821
-    },
-    {
-      "epoch": 1.9742722250893592,
-      "grad_norm": 0.0012734156334772706,
-      "learning_rate": 0.00019999807802978362,
-      "loss": 46.0,
-      "step": 25822
-    },
-    {
-      "epoch": 1.974348682072749,
-      "grad_norm": 0.000698140705935657,
-      "learning_rate": 0.00019999807788086057,
-      "loss": 46.0,
-      "step": 25823
-    },
-    {
-      "epoch": 1.9744251390561387,
-      "grad_norm": 0.0029561358969658613,
-      "learning_rate": 0.00019999807773193168,
-      "loss": 46.0,
-      "step": 25824
-    },
-    {
-      "epoch": 1.9745015960395282,
-      "grad_norm": 0.001605145982466638,
-      "learning_rate": 0.00019999807758299708,
-      "loss": 46.0,
-      "step": 25825
-    },
-    {
-      "epoch": 1.974578053022918,
-      "grad_norm": 0.0010017509339377284,
-      "learning_rate": 0.00019999807743405668,
-      "loss": 46.0,
-      "step": 25826
-    },
-    {
-      "epoch": 1.9746545100063075,
-      "grad_norm": 0.0005134253297001123,
-      "learning_rate": 0.00019999807728511053,
-      "loss": 46.0,
-      "step": 25827
-    },
-    {
-      "epoch": 1.9747309669896973,
-      "grad_norm": 0.0010002549970522523,
-      "learning_rate": 0.0001999980771361586,
-      "loss": 46.0,
-      "step": 25828
-    },
-    {
-      "epoch": 1.974807423973087,
-      "grad_norm": 0.009470101445913315,
-      "learning_rate": 0.00019999807698720093,
-      "loss": 46.0,
-      "step": 25829
-    },
-    {
-      "epoch": 1.9748838809564768,
-      "grad_norm": 0.0013153848703950644,
-      "learning_rate": 0.00019999807683823744,
-      "loss": 46.0,
-      "step": 25830
-    },
-    {
-      "epoch": 1.9749603379398666,
-      "grad_norm": 0.0017125066369771957,
-      "learning_rate": 0.0001999980766892682,
-      "loss": 46.0,
-      "step": 25831
-    },
-    {
-      "epoch": 1.9750367949232563,
-      "grad_norm": 0.002290693111717701,
-      "learning_rate": 0.0001999980765402932,
-      "loss": 46.0,
-      "step": 25832
-    },
-    {
-      "epoch": 1.975113251906646,
-      "grad_norm": 0.002258292166516185,
-      "learning_rate": 0.00019999807639131242,
-      "loss": 46.0,
-      "step": 25833
-    },
-    {
-      "epoch": 1.9751897088900359,
-      "grad_norm": 0.0009066957863979042,
-      "learning_rate": 0.00019999807624232588,
-      "loss": 46.0,
-      "step": 25834
-    },
-    {
-      "epoch": 1.9752661658734256,
-      "grad_norm": 0.0003284249687567353,
-      "learning_rate": 0.00019999807609333357,
-      "loss": 46.0,
-      "step": 25835
-    },
-    {
-      "epoch": 1.9753426228568152,
-      "grad_norm": 0.0006234605680219829,
-      "learning_rate": 0.0001999980759443355,
-      "loss": 46.0,
-      "step": 25836
-    },
-    {
-      "epoch": 1.975419079840205,
-      "grad_norm": 0.002773670945316553,
-      "learning_rate": 0.00019999807579533163,
-      "loss": 46.0,
-      "step": 25837
-    },
-    {
-      "epoch": 1.9754955368235945,
-      "grad_norm": 0.0014588736230507493,
-      "learning_rate": 0.000199998075646322,
-      "loss": 46.0,
-      "step": 25838
-    },
-    {
-      "epoch": 1.9755719938069842,
-      "grad_norm": 0.0011562788859009743,
-      "learning_rate": 0.00019999807549730663,
-      "loss": 46.0,
-      "step": 25839
-    },
-    {
-      "epoch": 1.975648450790374,
-      "grad_norm": 0.0024272899609059095,
-      "learning_rate": 0.00019999807534828545,
-      "loss": 46.0,
-      "step": 25840
-    },
-    {
-      "epoch": 1.9757249077737637,
-      "grad_norm": 0.0009556411532685161,
-      "learning_rate": 0.00019999807519925853,
-      "loss": 46.0,
-      "step": 25841
-    },
-    {
-      "epoch": 1.9758013647571535,
-      "grad_norm": 0.0037731309421360493,
-      "learning_rate": 0.00019999807505022583,
-      "loss": 46.0,
-      "step": 25842
-    },
-    {
-      "epoch": 1.9758778217405433,
-      "grad_norm": 0.002451549516990781,
-      "learning_rate": 0.00019999807490118736,
-      "loss": 46.0,
-      "step": 25843
-    },
-    {
-      "epoch": 1.975954278723933,
-      "grad_norm": 0.0021754151675850153,
-      "learning_rate": 0.00019999807475214315,
-      "loss": 46.0,
-      "step": 25844
-    },
-    {
-      "epoch": 1.9760307357073228,
-      "grad_norm": 0.0030975828412920237,
-      "learning_rate": 0.00019999807460309313,
-      "loss": 46.0,
-      "step": 25845
-    },
-    {
-      "epoch": 1.9761071926907123,
-      "grad_norm": 0.008454015478491783,
-      "learning_rate": 0.00019999807445403734,
-      "loss": 46.0,
-      "step": 25846
-    },
-    {
-      "epoch": 1.976183649674102,
-      "grad_norm": 0.004546020179986954,
-      "learning_rate": 0.00019999807430497578,
-      "loss": 46.0,
-      "step": 25847
-    },
-    {
-      "epoch": 1.9762601066574919,
-      "grad_norm": 0.000900718558114022,
-      "learning_rate": 0.00019999807415590847,
-      "loss": 46.0,
-      "step": 25848
-    },
-    {
-      "epoch": 1.9763365636408814,
-      "grad_norm": 0.0016044018557295203,
-      "learning_rate": 0.0001999980740068354,
-      "loss": 46.0,
-      "step": 25849
-    },
-    {
-      "epoch": 1.9764130206242712,
-      "grad_norm": 0.0003852058725897223,
-      "learning_rate": 0.00019999807385775653,
-      "loss": 46.0,
-      "step": 25850
-    },
-    {
-      "epoch": 1.976489477607661,
-      "grad_norm": 0.003859543474391103,
-      "learning_rate": 0.00019999807370867193,
-      "loss": 46.0,
-      "step": 25851
-    },
-    {
-      "epoch": 1.9765659345910507,
-      "grad_norm": 0.0012387122260406613,
-      "learning_rate": 0.00019999807355958153,
-      "loss": 46.0,
-      "step": 25852
-    },
-    {
-      "epoch": 1.9766423915744404,
-      "grad_norm": 0.00046013304381631315,
-      "learning_rate": 0.00019999807341048538,
-      "loss": 46.0,
-      "step": 25853
-    },
-    {
-      "epoch": 1.9767188485578302,
-      "grad_norm": 0.001931617152877152,
-      "learning_rate": 0.00019999807326138346,
-      "loss": 46.0,
-      "step": 25854
-    },
-    {
-      "epoch": 1.97679530554122,
-      "grad_norm": 0.0021536045242100954,
-      "learning_rate": 0.00019999807311227573,
-      "loss": 46.0,
-      "step": 25855
-    },
-    {
-      "epoch": 1.9768717625246097,
-      "grad_norm": 0.0019783740863204002,
-      "learning_rate": 0.00019999807296316226,
-      "loss": 46.0,
-      "step": 25856
-    },
-    {
-      "epoch": 1.9769482195079993,
-      "grad_norm": 0.0029039422515779734,
-      "learning_rate": 0.00019999807281404302,
-      "loss": 46.0,
-      "step": 25857
-    },
-    {
-      "epoch": 1.977024676491389,
-      "grad_norm": 0.001038994058035314,
-      "learning_rate": 0.000199998072664918,
-      "loss": 46.0,
-      "step": 25858
-    },
-    {
-      "epoch": 1.9771011334747788,
-      "grad_norm": 0.0006801777053624392,
-      "learning_rate": 0.00019999807251578724,
-      "loss": 46.0,
-      "step": 25859
-    },
-    {
-      "epoch": 1.9771775904581683,
-      "grad_norm": 0.0006004204624332488,
-      "learning_rate": 0.00019999807236665068,
-      "loss": 46.0,
-      "step": 25860
-    },
-    {
-      "epoch": 1.977254047441558,
-      "grad_norm": 0.003392234444618225,
-      "learning_rate": 0.00019999807221750837,
-      "loss": 46.0,
-      "step": 25861
-    },
-    {
-      "epoch": 1.9773305044249478,
-      "grad_norm": 0.0020625144243240356,
-      "learning_rate": 0.0001999980720683603,
-      "loss": 46.0,
-      "step": 25862
-    },
-    {
-      "epoch": 1.9774069614083376,
-      "grad_norm": 0.0017339610494673252,
-      "learning_rate": 0.00019999807191920644,
-      "loss": 46.0,
-      "step": 25863
-    },
-    {
-      "epoch": 1.9774834183917274,
-      "grad_norm": 0.0004393297713249922,
-      "learning_rate": 0.00019999807177004678,
-      "loss": 46.0,
-      "step": 25864
-    },
-    {
-      "epoch": 1.9775598753751171,
-      "grad_norm": 0.0019282331923022866,
-      "learning_rate": 0.0001999980716208814,
-      "loss": 46.0,
-      "step": 25865
-    },
-    {
-      "epoch": 1.9776363323585069,
-      "grad_norm": 0.0029359781183302402,
-      "learning_rate": 0.00019999807147171023,
-      "loss": 46.0,
-      "step": 25866
-    },
-    {
-      "epoch": 1.9777127893418966,
-      "grad_norm": 0.000568628020118922,
-      "learning_rate": 0.00019999807132253333,
-      "loss": 46.0,
-      "step": 25867
-    },
-    {
-      "epoch": 1.9777892463252862,
-      "grad_norm": 0.001429157331585884,
-      "learning_rate": 0.00019999807117335064,
-      "loss": 46.0,
-      "step": 25868
-    },
-    {
-      "epoch": 1.977865703308676,
-      "grad_norm": 0.000368216511560604,
-      "learning_rate": 0.00019999807102416214,
-      "loss": 46.0,
-      "step": 25869
-    },
-    {
-      "epoch": 1.9779421602920657,
-      "grad_norm": 0.0009015920222736895,
-      "learning_rate": 0.0001999980708749679,
-      "loss": 46.0,
-      "step": 25870
-    },
-    {
-      "epoch": 1.9780186172754552,
-      "grad_norm": 0.008182347752153873,
-      "learning_rate": 0.0001999980707257679,
-      "loss": 46.0,
-      "step": 25871
-    },
-    {
-      "epoch": 1.978095074258845,
-      "grad_norm": 0.0006159769254736602,
-      "learning_rate": 0.0001999980705765621,
-      "loss": 46.0,
-      "step": 25872
-    },
-    {
-      "epoch": 1.9781715312422348,
-      "grad_norm": 0.0004768925136886537,
-      "learning_rate": 0.00019999807042735056,
-      "loss": 46.0,
-      "step": 25873
-    },
-    {
-      "epoch": 1.9782479882256245,
-      "grad_norm": 0.000916451623197645,
-      "learning_rate": 0.00019999807027813325,
-      "loss": 46.0,
-      "step": 25874
-    },
-    {
-      "epoch": 1.9783244452090143,
-      "grad_norm": 0.0027534232940524817,
-      "learning_rate": 0.00019999807012891017,
-      "loss": 46.0,
-      "step": 25875
-    },
-    {
-      "epoch": 1.978400902192404,
-      "grad_norm": 0.00047271931543946266,
-      "learning_rate": 0.0001999980699796813,
-      "loss": 46.0,
-      "step": 25876
-    },
-    {
-      "epoch": 1.9784773591757938,
-      "grad_norm": 0.000993673107586801,
-      "learning_rate": 0.00019999806983044666,
-      "loss": 46.0,
-      "step": 25877
-    },
-    {
-      "epoch": 1.9785538161591836,
-      "grad_norm": 0.0006514216074720025,
-      "learning_rate": 0.00019999806968120626,
-      "loss": 46.0,
-      "step": 25878
-    },
-    {
-      "epoch": 1.9786302731425731,
-      "grad_norm": 0.0008146577165462077,
-      "learning_rate": 0.0001999980695319601,
-      "loss": 46.0,
-      "step": 25879
-    },
-    {
-      "epoch": 1.9787067301259629,
-      "grad_norm": 0.0037862665485590696,
-      "learning_rate": 0.0001999980693827082,
-      "loss": 46.0,
-      "step": 25880
-    },
-    {
-      "epoch": 1.9787831871093526,
-      "grad_norm": 0.00133634137455374,
-      "learning_rate": 0.0001999980692334505,
-      "loss": 46.0,
-      "step": 25881
-    },
-    {
-      "epoch": 1.9788596440927422,
-      "grad_norm": 0.0007426688098348677,
-      "learning_rate": 0.00019999806908418703,
-      "loss": 46.0,
-      "step": 25882
-    },
-    {
-      "epoch": 1.978936101076132,
-      "grad_norm": 0.0014057334046810865,
-      "learning_rate": 0.00019999806893491778,
-      "loss": 46.0,
-      "step": 25883
-    },
-    {
-      "epoch": 1.9790125580595217,
-      "grad_norm": 0.001237798947840929,
-      "learning_rate": 0.00019999806878564277,
-      "loss": 46.0,
-      "step": 25884
-    },
-    {
-      "epoch": 1.9790890150429115,
-      "grad_norm": 0.0012995251454412937,
-      "learning_rate": 0.00019999806863636198,
-      "loss": 46.0,
-      "step": 25885
-    },
-    {
-      "epoch": 1.9791654720263012,
-      "grad_norm": 0.0014032222097739577,
-      "learning_rate": 0.00019999806848707542,
-      "loss": 46.0,
-      "step": 25886
-    },
-    {
-      "epoch": 1.979241929009691,
-      "grad_norm": 0.001716622500680387,
-      "learning_rate": 0.0001999980683377831,
-      "loss": 46.0,
-      "step": 25887
-    },
-    {
-      "epoch": 1.9793183859930807,
-      "grad_norm": 0.0006896345294080675,
-      "learning_rate": 0.000199998068188485,
-      "loss": 46.0,
-      "step": 25888
-    },
-    {
-      "epoch": 1.9793948429764705,
-      "grad_norm": 0.002268021460622549,
-      "learning_rate": 0.00019999806803918115,
-      "loss": 46.0,
-      "step": 25889
-    },
-    {
-      "epoch": 1.97947129995986,
-      "grad_norm": 0.0019452099222689867,
-      "learning_rate": 0.00019999806788987155,
-      "loss": 46.0,
-      "step": 25890
-    },
-    {
-      "epoch": 1.9795477569432498,
-      "grad_norm": 0.0024786870926618576,
-      "learning_rate": 0.00019999806774055614,
-      "loss": 46.0,
-      "step": 25891
-    },
-    {
-      "epoch": 1.9796242139266396,
-      "grad_norm": 0.0007155233179219067,
-      "learning_rate": 0.00019999806759123497,
-      "loss": 46.0,
-      "step": 25892
-    },
-    {
-      "epoch": 1.979700670910029,
-      "grad_norm": 0.002744987839832902,
-      "learning_rate": 0.00019999806744190802,
-      "loss": 46.0,
-      "step": 25893
-    },
-    {
-      "epoch": 1.9797771278934189,
-      "grad_norm": 0.0014073261991143227,
-      "learning_rate": 0.00019999806729257533,
-      "loss": 46.0,
-      "step": 25894
-    },
-    {
-      "epoch": 1.9798535848768086,
-      "grad_norm": 0.011025724932551384,
-      "learning_rate": 0.00019999806714323686,
-      "loss": 46.0,
-      "step": 25895
-    },
-    {
-      "epoch": 1.9799300418601984,
-      "grad_norm": 0.001961614703759551,
-      "learning_rate": 0.00019999806699389262,
-      "loss": 46.0,
-      "step": 25896
-    },
-    {
-      "epoch": 1.9800064988435881,
-      "grad_norm": 0.000389292516047135,
-      "learning_rate": 0.0001999980668445426,
-      "loss": 46.0,
-      "step": 25897
-    },
-    {
-      "epoch": 1.980082955826978,
-      "grad_norm": 0.0014290747931227088,
-      "learning_rate": 0.0001999980666951868,
-      "loss": 46.0,
-      "step": 25898
-    },
-    {
-      "epoch": 1.9801594128103677,
-      "grad_norm": 0.0033309024292975664,
-      "learning_rate": 0.00019999806654582525,
-      "loss": 46.0,
-      "step": 25899
-    },
-    {
-      "epoch": 1.9802358697937574,
-      "grad_norm": 0.002173813059926033,
-      "learning_rate": 0.00019999806639645792,
-      "loss": 46.0,
-      "step": 25900
-    },
-    {
-      "epoch": 1.980312326777147,
-      "grad_norm": 0.001985164824873209,
-      "learning_rate": 0.00019999806624708484,
-      "loss": 46.0,
-      "step": 25901
-    },
-    {
-      "epoch": 1.9803887837605367,
-      "grad_norm": 0.0027034750673919916,
-      "learning_rate": 0.00019999806609770598,
-      "loss": 46.0,
-      "step": 25902
-    },
-    {
-      "epoch": 1.9804652407439265,
-      "grad_norm": 0.0010117364581674337,
-      "learning_rate": 0.00019999806594832138,
-      "loss": 46.0,
-      "step": 25903
-    },
-    {
-      "epoch": 1.980541697727316,
-      "grad_norm": 0.0013257297687232494,
-      "learning_rate": 0.00019999806579893098,
-      "loss": 46.0,
-      "step": 25904
-    },
-    {
-      "epoch": 1.9806181547107058,
-      "grad_norm": 0.0014383801026269794,
-      "learning_rate": 0.00019999806564953478,
-      "loss": 46.0,
-      "step": 25905
-    },
-    {
-      "epoch": 1.9806946116940956,
-      "grad_norm": 0.0007488276460207999,
-      "learning_rate": 0.00019999806550013286,
-      "loss": 46.0,
-      "step": 25906
-    },
-    {
-      "epoch": 1.9807710686774853,
-      "grad_norm": 0.0008196709677577019,
-      "learning_rate": 0.00019999806535072513,
-      "loss": 46.0,
-      "step": 25907
-    },
-    {
-      "epoch": 1.980847525660875,
-      "grad_norm": 0.0027127836365252733,
-      "learning_rate": 0.00019999806520131167,
-      "loss": 46.0,
-      "step": 25908
-    },
-    {
-      "epoch": 1.9809239826442648,
-      "grad_norm": 0.0007803915650583804,
-      "learning_rate": 0.00019999806505189245,
-      "loss": 46.0,
-      "step": 25909
-    },
-    {
-      "epoch": 1.9810004396276546,
-      "grad_norm": 0.000966378313023597,
-      "learning_rate": 0.0001999980649024674,
-      "loss": 46.0,
-      "step": 25910
-    },
-    {
-      "epoch": 1.9810768966110444,
-      "grad_norm": 0.002839515684172511,
-      "learning_rate": 0.00019999806475303665,
-      "loss": 46.0,
-      "step": 25911
-    },
-    {
-      "epoch": 1.981153353594434,
-      "grad_norm": 0.004682973027229309,
-      "learning_rate": 0.00019999806460360006,
-      "loss": 46.0,
-      "step": 25912
-    },
-    {
-      "epoch": 1.9812298105778237,
-      "grad_norm": 0.0009504939080215991,
-      "learning_rate": 0.00019999806445415776,
-      "loss": 46.0,
-      "step": 25913
-    },
-    {
-      "epoch": 1.9813062675612134,
-      "grad_norm": 0.0019033750286325812,
-      "learning_rate": 0.00019999806430470967,
-      "loss": 46.0,
-      "step": 25914
-    },
-    {
-      "epoch": 1.981382724544603,
-      "grad_norm": 0.0028831204399466515,
-      "learning_rate": 0.00019999806415525582,
-      "loss": 46.0,
-      "step": 25915
-    },
-    {
-      "epoch": 1.9814591815279927,
-      "grad_norm": 0.0007136219064705074,
-      "learning_rate": 0.00019999806400579617,
-      "loss": 46.0,
-      "step": 25916
-    },
-    {
-      "epoch": 1.9815356385113825,
-      "grad_norm": 0.0005213957047089934,
-      "learning_rate": 0.00019999806385633076,
-      "loss": 46.0,
-      "step": 25917
-    },
-    {
-      "epoch": 1.9816120954947722,
-      "grad_norm": 0.001165029825642705,
-      "learning_rate": 0.00019999806370685962,
-      "loss": 46.0,
-      "step": 25918
-    },
-    {
-      "epoch": 1.981688552478162,
-      "grad_norm": 0.0018076051492244005,
-      "learning_rate": 0.0001999980635573827,
-      "loss": 46.0,
-      "step": 25919
-    },
-    {
-      "epoch": 1.9817650094615518,
-      "grad_norm": 0.004414466675370932,
-      "learning_rate": 0.00019999806340789995,
-      "loss": 46.0,
-      "step": 25920
-    },
-    {
-      "epoch": 1.9818414664449415,
-      "grad_norm": 0.0005241346079856157,
-      "learning_rate": 0.00019999806325841148,
-      "loss": 46.0,
-      "step": 25921
-    },
-    {
-      "epoch": 1.9819179234283313,
-      "grad_norm": 0.005071109160780907,
-      "learning_rate": 0.00019999806310891724,
-      "loss": 46.0,
-      "step": 25922
-    },
-    {
-      "epoch": 1.9819943804117208,
-      "grad_norm": 0.0006443731253966689,
-      "learning_rate": 0.00019999806295941723,
-      "loss": 46.0,
-      "step": 25923
-    },
-    {
-      "epoch": 1.9820708373951106,
-      "grad_norm": 0.0004258525150362402,
-      "learning_rate": 0.00019999806280991144,
-      "loss": 46.0,
-      "step": 25924
-    },
-    {
-      "epoch": 1.9821472943785003,
-      "grad_norm": 0.0004597913066390902,
-      "learning_rate": 0.00019999806266039988,
-      "loss": 46.0,
-      "step": 25925
-    },
-    {
-      "epoch": 1.9822237513618899,
-      "grad_norm": 0.0006636075559072196,
-      "learning_rate": 0.00019999806251088254,
-      "loss": 46.0,
-      "step": 25926
-    },
-    {
-      "epoch": 1.9823002083452796,
-      "grad_norm": 0.001049064681865275,
-      "learning_rate": 0.00019999806236135946,
-      "loss": 46.0,
-      "step": 25927
-    },
-    {
-      "epoch": 1.9823766653286694,
-      "grad_norm": 0.0014680858002975583,
-      "learning_rate": 0.0001999980622118306,
-      "loss": 46.0,
-      "step": 25928
-    },
-    {
-      "epoch": 1.9824531223120592,
-      "grad_norm": 0.0011501642875373363,
-      "learning_rate": 0.00019999806206229595,
-      "loss": 46.0,
-      "step": 25929
-    },
-    {
-      "epoch": 1.982529579295449,
-      "grad_norm": 0.0003905289340764284,
-      "learning_rate": 0.00019999806191275558,
-      "loss": 46.0,
-      "step": 25930
-    },
-    {
-      "epoch": 1.9826060362788387,
-      "grad_norm": 0.0021350185852497816,
-      "learning_rate": 0.00019999806176320938,
-      "loss": 46.0,
-      "step": 25931
-    },
-    {
-      "epoch": 1.9826824932622285,
-      "grad_norm": 0.0004402836493682116,
-      "learning_rate": 0.00019999806161365746,
-      "loss": 46.0,
-      "step": 25932
-    },
-    {
-      "epoch": 1.9827589502456182,
-      "grad_norm": 0.0006026846822351217,
-      "learning_rate": 0.00019999806146409977,
-      "loss": 46.0,
-      "step": 25933
-    },
-    {
-      "epoch": 1.9828354072290078,
-      "grad_norm": 0.0007048918050713837,
-      "learning_rate": 0.0001999980613145363,
-      "loss": 46.0,
-      "step": 25934
-    },
-    {
-      "epoch": 1.9829118642123975,
-      "grad_norm": 0.001930861035361886,
-      "learning_rate": 0.00019999806116496703,
-      "loss": 46.0,
-      "step": 25935
-    },
-    {
-      "epoch": 1.9829883211957873,
-      "grad_norm": 0.0017339850310236216,
-      "learning_rate": 0.00019999806101539202,
-      "loss": 46.0,
-      "step": 25936
-    },
-    {
-      "epoch": 1.9830647781791768,
-      "grad_norm": 0.004155580885708332,
-      "learning_rate": 0.00019999806086581123,
-      "loss": 46.0,
-      "step": 25937
-    },
-    {
-      "epoch": 1.9831412351625666,
-      "grad_norm": 0.0012208541156724095,
-      "learning_rate": 0.00019999806071622467,
-      "loss": 46.0,
-      "step": 25938
-    },
-    {
-      "epoch": 1.9832176921459563,
-      "grad_norm": 0.0019351525697857141,
-      "learning_rate": 0.00019999806056663234,
-      "loss": 46.0,
-      "step": 25939
-    },
-    {
-      "epoch": 1.983294149129346,
-      "grad_norm": 0.0014785719104111195,
-      "learning_rate": 0.00019999806041703426,
-      "loss": 46.0,
-      "step": 25940
-    },
-    {
-      "epoch": 1.9833706061127359,
-      "grad_norm": 0.0014351492282003164,
-      "learning_rate": 0.00019999806026743037,
-      "loss": 46.0,
-      "step": 25941
-    },
-    {
-      "epoch": 1.9834470630961256,
-      "grad_norm": 0.0004911868018098176,
-      "learning_rate": 0.00019999806011782077,
-      "loss": 46.0,
-      "step": 25942
-    },
-    {
-      "epoch": 1.9835235200795154,
-      "grad_norm": 0.00128077098634094,
-      "learning_rate": 0.00019999805996820532,
-      "loss": 46.0,
-      "step": 25943
-    },
-    {
-      "epoch": 1.9835999770629051,
-      "grad_norm": 0.0004483874363359064,
-      "learning_rate": 0.00019999805981858417,
-      "loss": 46.0,
-      "step": 25944
-    },
-    {
-      "epoch": 1.9836764340462947,
-      "grad_norm": 0.0012766679283231497,
-      "learning_rate": 0.00019999805966895723,
-      "loss": 46.0,
-      "step": 25945
-    },
-    {
-      "epoch": 1.9837528910296844,
-      "grad_norm": 0.0021675911266356707,
-      "learning_rate": 0.00019999805951932453,
-      "loss": 46.0,
-      "step": 25946
-    },
-    {
-      "epoch": 1.983829348013074,
-      "grad_norm": 0.0005191454547457397,
-      "learning_rate": 0.00019999805936968607,
-      "loss": 46.0,
-      "step": 25947
-    },
-    {
-      "epoch": 1.9839058049964637,
-      "grad_norm": 0.0016540009528398514,
-      "learning_rate": 0.0001999980592200418,
-      "loss": 46.0,
-      "step": 25948
-    },
-    {
-      "epoch": 1.9839822619798535,
-      "grad_norm": 0.0003261689271312207,
-      "learning_rate": 0.00019999805907039179,
-      "loss": 46.0,
-      "step": 25949
-    },
-    {
-      "epoch": 1.9840587189632433,
-      "grad_norm": 0.0013586319983005524,
-      "learning_rate": 0.00019999805892073597,
-      "loss": 46.0,
-      "step": 25950
-    },
-    {
-      "epoch": 1.984135175946633,
-      "grad_norm": 0.0015663780504837632,
-      "learning_rate": 0.00019999805877107444,
-      "loss": 46.0,
-      "step": 25951
-    },
-    {
-      "epoch": 1.9842116329300228,
-      "grad_norm": 0.0019280379638075829,
-      "learning_rate": 0.0001999980586214071,
-      "loss": 46.0,
-      "step": 25952
-    },
-    {
-      "epoch": 1.9842880899134125,
-      "grad_norm": 0.0028812652453780174,
-      "learning_rate": 0.000199998058471734,
-      "loss": 46.0,
-      "step": 25953
-    },
-    {
-      "epoch": 1.9843645468968023,
-      "grad_norm": 0.0006614577723667026,
-      "learning_rate": 0.00019999805832205514,
-      "loss": 46.0,
-      "step": 25954
-    },
-    {
-      "epoch": 1.984441003880192,
-      "grad_norm": 0.002217573346570134,
-      "learning_rate": 0.00019999805817237052,
-      "loss": 46.0,
-      "step": 25955
-    },
-    {
-      "epoch": 1.9845174608635816,
-      "grad_norm": 0.0008722925558686256,
-      "learning_rate": 0.0001999980580226801,
-      "loss": 46.0,
-      "step": 25956
-    },
-    {
-      "epoch": 1.9845939178469714,
-      "grad_norm": 0.0016122815432026982,
-      "learning_rate": 0.00019999805787298392,
-      "loss": 46.0,
-      "step": 25957
-    },
-    {
-      "epoch": 1.984670374830361,
-      "grad_norm": 0.0028325149323791265,
-      "learning_rate": 0.00019999805772328197,
-      "loss": 46.0,
-      "step": 25958
-    },
-    {
-      "epoch": 1.9847468318137507,
-      "grad_norm": 0.0017200781730934978,
-      "learning_rate": 0.00019999805757357425,
-      "loss": 46.0,
-      "step": 25959
-    },
-    {
-      "epoch": 1.9848232887971404,
-      "grad_norm": 0.001014393987134099,
-      "learning_rate": 0.00019999805742386078,
-      "loss": 46.0,
-      "step": 25960
-    },
-    {
-      "epoch": 1.9848997457805302,
-      "grad_norm": 0.0007479116902686656,
-      "learning_rate": 0.00019999805727414154,
-      "loss": 46.0,
-      "step": 25961
-    },
-    {
-      "epoch": 1.98497620276392,
-      "grad_norm": 0.0013300023274496198,
-      "learning_rate": 0.00019999805712441653,
-      "loss": 46.0,
-      "step": 25962
-    },
-    {
-      "epoch": 1.9850526597473097,
-      "grad_norm": 0.00508840149268508,
-      "learning_rate": 0.00019999805697468572,
-      "loss": 46.0,
-      "step": 25963
-    },
-    {
-      "epoch": 1.9851291167306995,
-      "grad_norm": 0.002394547453150153,
-      "learning_rate": 0.00019999805682494918,
-      "loss": 46.0,
-      "step": 25964
-    },
-    {
-      "epoch": 1.9852055737140892,
-      "grad_norm": 0.0009620438795536757,
-      "learning_rate": 0.00019999805667520682,
-      "loss": 46.0,
-      "step": 25965
-    },
-    {
-      "epoch": 1.985282030697479,
-      "grad_norm": 0.004395087715238333,
-      "learning_rate": 0.00019999805652545874,
-      "loss": 46.0,
-      "step": 25966
-    },
-    {
-      "epoch": 1.9853584876808685,
-      "grad_norm": 0.0007532541640102863,
-      "learning_rate": 0.0001999980563757049,
-      "loss": 46.0,
-      "step": 25967
-    },
-    {
-      "epoch": 1.9854349446642583,
-      "grad_norm": 0.0007761105662211776,
-      "learning_rate": 0.00019999805622594524,
-      "loss": 46.0,
-      "step": 25968
-    },
-    {
-      "epoch": 1.9855114016476478,
-      "grad_norm": 0.005108152516186237,
-      "learning_rate": 0.00019999805607617984,
-      "loss": 46.0,
-      "step": 25969
-    },
-    {
-      "epoch": 1.9855878586310376,
-      "grad_norm": 0.011084684170782566,
-      "learning_rate": 0.00019999805592640866,
-      "loss": 46.0,
-      "step": 25970
-    },
-    {
-      "epoch": 1.9856643156144274,
-      "grad_norm": 0.0010184103157371283,
-      "learning_rate": 0.00019999805577663172,
-      "loss": 46.0,
-      "step": 25971
-    },
-    {
-      "epoch": 1.9857407725978171,
-      "grad_norm": 0.0014622407034039497,
-      "learning_rate": 0.00019999805562684903,
-      "loss": 46.0,
-      "step": 25972
-    },
-    {
-      "epoch": 1.9858172295812069,
-      "grad_norm": 0.0006349731120280921,
-      "learning_rate": 0.00019999805547706053,
-      "loss": 46.0,
-      "step": 25973
-    },
-    {
-      "epoch": 1.9858936865645966,
-      "grad_norm": 0.001209433190524578,
-      "learning_rate": 0.00019999805532726627,
-      "loss": 46.0,
-      "step": 25974
-    },
-    {
-      "epoch": 1.9859701435479864,
-      "grad_norm": 0.001606301637366414,
-      "learning_rate": 0.00019999805517746623,
-      "loss": 46.0,
-      "step": 25975
-    },
-    {
-      "epoch": 1.9860466005313762,
-      "grad_norm": 0.0009062322205863893,
-      "learning_rate": 0.00019999805502766047,
-      "loss": 46.0,
-      "step": 25976
-    },
-    {
-      "epoch": 1.9861230575147657,
-      "grad_norm": 0.0009599516051821411,
-      "learning_rate": 0.00019999805487784888,
-      "loss": 46.0,
-      "step": 25977
-    },
-    {
-      "epoch": 1.9861995144981555,
-      "grad_norm": 0.0011032337788492441,
-      "learning_rate": 0.00019999805472803157,
-      "loss": 46.0,
-      "step": 25978
-    },
-    {
-      "epoch": 1.9862759714815452,
-      "grad_norm": 0.0007067614351399243,
-      "learning_rate": 0.0001999980545782085,
-      "loss": 46.0,
-      "step": 25979
-    },
-    {
-      "epoch": 1.9863524284649348,
-      "grad_norm": 0.0005981251015327871,
-      "learning_rate": 0.0001999980544283796,
-      "loss": 46.0,
-      "step": 25980
-    },
-    {
-      "epoch": 1.9864288854483245,
-      "grad_norm": 0.001199584687128663,
-      "learning_rate": 0.00019999805427854496,
-      "loss": 46.0,
-      "step": 25981
-    },
-    {
-      "epoch": 1.9865053424317143,
-      "grad_norm": 0.0008631689706817269,
-      "learning_rate": 0.00019999805412870456,
-      "loss": 46.0,
-      "step": 25982
-    },
-    {
-      "epoch": 1.986581799415104,
-      "grad_norm": 0.0036176980938762426,
-      "learning_rate": 0.0001999980539788584,
-      "loss": 46.0,
-      "step": 25983
-    },
-    {
-      "epoch": 1.9866582563984938,
-      "grad_norm": 0.0029676465783268213,
-      "learning_rate": 0.00019999805382900644,
-      "loss": 46.0,
-      "step": 25984
-    },
-    {
-      "epoch": 1.9867347133818836,
-      "grad_norm": 0.0018274352187290788,
-      "learning_rate": 0.00019999805367914872,
-      "loss": 46.0,
-      "step": 25985
-    },
-    {
-      "epoch": 1.9868111703652733,
-      "grad_norm": 0.0011082688579335809,
-      "learning_rate": 0.00019999805352928523,
-      "loss": 46.0,
-      "step": 25986
-    },
-    {
-      "epoch": 1.986887627348663,
-      "grad_norm": 0.001649734447710216,
-      "learning_rate": 0.000199998053379416,
-      "loss": 46.0,
-      "step": 25987
-    },
-    {
-      "epoch": 1.9869640843320526,
-      "grad_norm": 0.0005887331790290773,
-      "learning_rate": 0.00019999805322954095,
-      "loss": 46.0,
-      "step": 25988
-    },
-    {
-      "epoch": 1.9870405413154424,
-      "grad_norm": 0.0006896872655488551,
-      "learning_rate": 0.00019999805307966017,
-      "loss": 46.0,
-      "step": 25989
-    },
-    {
-      "epoch": 1.9871169982988321,
-      "grad_norm": 0.0018458407139405608,
-      "learning_rate": 0.0001999980529297736,
-      "loss": 46.0,
-      "step": 25990
-    },
-    {
-      "epoch": 1.9871934552822217,
-      "grad_norm": 0.0009387212339788675,
-      "learning_rate": 0.00019999805277988127,
-      "loss": 46.0,
-      "step": 25991
-    },
-    {
-      "epoch": 1.9872699122656114,
-      "grad_norm": 0.0010361146414652467,
-      "learning_rate": 0.00019999805262998317,
-      "loss": 46.0,
-      "step": 25992
-    },
-    {
-      "epoch": 1.9873463692490012,
-      "grad_norm": 0.0016938091721385717,
-      "learning_rate": 0.0001999980524800793,
-      "loss": 46.0,
-      "step": 25993
-    },
-    {
-      "epoch": 1.987422826232391,
-      "grad_norm": 0.0009540242026560009,
-      "learning_rate": 0.00019999805233016966,
-      "loss": 46.0,
-      "step": 25994
-    },
-    {
-      "epoch": 1.9874992832157807,
-      "grad_norm": 0.007173227611929178,
-      "learning_rate": 0.00019999805218025424,
-      "loss": 46.0,
-      "step": 25995
-    },
-    {
-      "epoch": 1.9875757401991705,
-      "grad_norm": 0.0010351529344916344,
-      "learning_rate": 0.0001999980520303331,
-      "loss": 46.0,
-      "step": 25996
-    },
-    {
-      "epoch": 1.9876521971825603,
-      "grad_norm": 0.0008168282802216709,
-      "learning_rate": 0.00019999805188040614,
-      "loss": 46.0,
-      "step": 25997
-    },
-    {
-      "epoch": 1.98772865416595,
-      "grad_norm": 0.001473135664127767,
-      "learning_rate": 0.00019999805173047343,
-      "loss": 46.0,
-      "step": 25998
-    },
-    {
-      "epoch": 1.9878051111493396,
-      "grad_norm": 0.0025913547724485397,
-      "learning_rate": 0.00019999805158053493,
-      "loss": 46.0,
-      "step": 25999
-    },
-    {
-      "epoch": 1.9878815681327293,
-      "grad_norm": 0.002686369465664029,
-      "learning_rate": 0.00019999805143059067,
-      "loss": 46.0,
-      "step": 26000
-    },
-    {
-      "epoch": 1.987958025116119,
-      "grad_norm": 0.0029672659002244473,
-      "learning_rate": 0.00019999805128064065,
-      "loss": 46.0,
-      "step": 26001
-    },
-    {
-      "epoch": 1.9880344820995086,
-      "grad_norm": 0.0016667507588863373,
-      "learning_rate": 0.00019999805113068487,
-      "loss": 46.0,
-      "step": 26002
-    },
-    {
-      "epoch": 1.9881109390828984,
-      "grad_norm": 0.0020327751990407705,
-      "learning_rate": 0.00019999805098072328,
-      "loss": 46.0,
-      "step": 26003
-    },
-    {
-      "epoch": 1.9881873960662881,
-      "grad_norm": 0.002505532931536436,
-      "learning_rate": 0.00019999805083075595,
-      "loss": 46.0,
-      "step": 26004
-    },
-    {
-      "epoch": 1.988263853049678,
-      "grad_norm": 0.002215274376794696,
-      "learning_rate": 0.00019999805068078284,
-      "loss": 46.0,
-      "step": 26005
-    },
-    {
-      "epoch": 1.9883403100330677,
-      "grad_norm": 0.0014072777703404427,
-      "learning_rate": 0.000199998050530804,
-      "loss": 46.0,
-      "step": 26006
-    },
-    {
-      "epoch": 1.9884167670164574,
-      "grad_norm": 0.0011279814643785357,
-      "learning_rate": 0.00019999805038081934,
-      "loss": 46.0,
-      "step": 26007
-    },
-    {
-      "epoch": 1.9884932239998472,
-      "grad_norm": 0.00045551860239356756,
-      "learning_rate": 0.00019999805023082894,
-      "loss": 46.0,
-      "step": 26008
-    },
-    {
-      "epoch": 1.988569680983237,
-      "grad_norm": 0.0014087604358792305,
-      "learning_rate": 0.00019999805008083277,
-      "loss": 46.0,
-      "step": 26009
-    },
-    {
-      "epoch": 1.9886461379666265,
-      "grad_norm": 0.0024430318735539913,
-      "learning_rate": 0.0001999980499308308,
-      "loss": 46.0,
-      "step": 26010
-    },
-    {
-      "epoch": 1.9887225949500162,
-      "grad_norm": 0.0013536142650991678,
-      "learning_rate": 0.0001999980497808231,
-      "loss": 46.0,
-      "step": 26011
-    },
-    {
-      "epoch": 1.988799051933406,
-      "grad_norm": 0.0013808598741889,
-      "learning_rate": 0.00019999804963080959,
-      "loss": 46.0,
-      "step": 26012
-    },
-    {
-      "epoch": 1.9888755089167955,
-      "grad_norm": 0.0016814154805615544,
-      "learning_rate": 0.00019999804948079035,
-      "loss": 46.0,
-      "step": 26013
-    },
-    {
-      "epoch": 1.9889519659001853,
-      "grad_norm": 0.0030319502111524343,
-      "learning_rate": 0.00019999804933076534,
-      "loss": 46.0,
-      "step": 26014
-    },
-    {
-      "epoch": 1.989028422883575,
-      "grad_norm": 0.0020242927130311728,
-      "learning_rate": 0.00019999804918073452,
-      "loss": 46.0,
-      "step": 26015
-    },
-    {
-      "epoch": 1.9891048798669648,
-      "grad_norm": 0.005415746476501226,
-      "learning_rate": 0.00019999804903069796,
-      "loss": 46.0,
-      "step": 26016
-    },
-    {
-      "epoch": 1.9891813368503546,
-      "grad_norm": 0.001192018506117165,
-      "learning_rate": 0.00019999804888065563,
-      "loss": 46.0,
-      "step": 26017
-    },
-    {
-      "epoch": 1.9892577938337443,
-      "grad_norm": 0.004027003422379494,
-      "learning_rate": 0.00019999804873060753,
-      "loss": 46.0,
-      "step": 26018
-    },
-    {
-      "epoch": 1.989334250817134,
-      "grad_norm": 0.0010035043815150857,
-      "learning_rate": 0.00019999804858055367,
-      "loss": 46.0,
-      "step": 26019
-    },
-    {
-      "epoch": 1.9894107078005239,
-      "grad_norm": 0.005634360946714878,
-      "learning_rate": 0.00019999804843049402,
-      "loss": 46.0,
-      "step": 26020
-    },
-    {
-      "epoch": 1.9894871647839134,
-      "grad_norm": 0.008986933156847954,
-      "learning_rate": 0.0001999980482804286,
-      "loss": 46.0,
-      "step": 26021
-    },
-    {
-      "epoch": 1.9895636217673032,
-      "grad_norm": 0.0012102812761440873,
-      "learning_rate": 0.00019999804813035743,
-      "loss": 46.0,
-      "step": 26022
-    },
-    {
-      "epoch": 1.989640078750693,
-      "grad_norm": 0.0015416034730151296,
-      "learning_rate": 0.00019999804798028045,
-      "loss": 46.0,
-      "step": 26023
-    },
-    {
-      "epoch": 1.9897165357340825,
-      "grad_norm": 0.0008822264499031007,
-      "learning_rate": 0.00019999804783019776,
-      "loss": 46.0,
-      "step": 26024
-    },
-    {
-      "epoch": 1.9897929927174722,
-      "grad_norm": 0.0008111972128972411,
-      "learning_rate": 0.00019999804768010927,
-      "loss": 46.0,
-      "step": 26025
-    },
-    {
-      "epoch": 1.989869449700862,
-      "grad_norm": 0.004551817663013935,
-      "learning_rate": 0.000199998047530015,
-      "loss": 46.0,
-      "step": 26026
-    },
-    {
-      "epoch": 1.9899459066842518,
-      "grad_norm": 0.0011565666645765305,
-      "learning_rate": 0.00019999804737991497,
-      "loss": 46.0,
-      "step": 26027
-    },
-    {
-      "epoch": 1.9900223636676415,
-      "grad_norm": 0.0027729745488613844,
-      "learning_rate": 0.00019999804722980918,
-      "loss": 46.0,
-      "step": 26028
-    },
-    {
-      "epoch": 1.9900988206510313,
-      "grad_norm": 0.00193033239338547,
-      "learning_rate": 0.0001999980470796976,
-      "loss": 46.0,
-      "step": 26029
-    },
-    {
-      "epoch": 1.990175277634421,
-      "grad_norm": 0.0007491422584280372,
-      "learning_rate": 0.0001999980469295803,
-      "loss": 46.0,
-      "step": 26030
-    },
-    {
-      "epoch": 1.9902517346178108,
-      "grad_norm": 0.004865617025643587,
-      "learning_rate": 0.00019999804677945716,
-      "loss": 46.0,
-      "step": 26031
-    },
-    {
-      "epoch": 1.9903281916012003,
-      "grad_norm": 0.008173923939466476,
-      "learning_rate": 0.0001999980466293283,
-      "loss": 46.0,
-      "step": 26032
-    },
-    {
-      "epoch": 1.99040464858459,
-      "grad_norm": 0.001339426962658763,
-      "learning_rate": 0.00019999804647919365,
-      "loss": 46.0,
-      "step": 26033
-    },
-    {
-      "epoch": 1.9904811055679799,
-      "grad_norm": 0.0027886289171874523,
-      "learning_rate": 0.00019999804632905326,
-      "loss": 46.0,
-      "step": 26034
-    },
-    {
-      "epoch": 1.9905575625513694,
-      "grad_norm": 0.01195152010768652,
-      "learning_rate": 0.00019999804617890706,
-      "loss": 46.0,
-      "step": 26035
-    },
-    {
-      "epoch": 1.9906340195347592,
-      "grad_norm": 0.0025208652950823307,
-      "learning_rate": 0.0001999980460287551,
-      "loss": 46.0,
-      "step": 26036
-    },
-    {
-      "epoch": 1.990710476518149,
-      "grad_norm": 0.000533510756213218,
-      "learning_rate": 0.0001999980458785974,
-      "loss": 46.0,
-      "step": 26037
-    },
-    {
-      "epoch": 1.9907869335015387,
-      "grad_norm": 0.005462232977151871,
-      "learning_rate": 0.0001999980457284339,
-      "loss": 46.0,
-      "step": 26038
-    },
-    {
-      "epoch": 1.9908633904849284,
-      "grad_norm": 0.00044619044638238847,
-      "learning_rate": 0.00019999804557826464,
-      "loss": 46.0,
-      "step": 26039
-    },
-    {
-      "epoch": 1.9909398474683182,
-      "grad_norm": 0.0014948435127735138,
-      "learning_rate": 0.0001999980454280896,
-      "loss": 46.0,
-      "step": 26040
-    },
-    {
-      "epoch": 1.991016304451708,
-      "grad_norm": 0.0008152392692863941,
-      "learning_rate": 0.00019999804527790881,
-      "loss": 46.0,
-      "step": 26041
-    },
-    {
-      "epoch": 1.9910927614350977,
-      "grad_norm": 0.0005629527731798589,
-      "learning_rate": 0.00019999804512772226,
-      "loss": 46.0,
-      "step": 26042
-    },
-    {
-      "epoch": 1.9911692184184873,
-      "grad_norm": 0.0007600269163958728,
-      "learning_rate": 0.00019999804497752993,
-      "loss": 46.0,
-      "step": 26043
-    },
-    {
-      "epoch": 1.991245675401877,
-      "grad_norm": 0.0010236066300421953,
-      "learning_rate": 0.0001999980448273318,
-      "loss": 46.0,
-      "step": 26044
-    },
-    {
-      "epoch": 1.9913221323852668,
-      "grad_norm": 0.0009042013552971184,
-      "learning_rate": 0.00019999804467712791,
-      "loss": 46.0,
-      "step": 26045
-    },
-    {
-      "epoch": 1.9913985893686563,
-      "grad_norm": 0.0013252947246655822,
-      "learning_rate": 0.0001999980445269183,
-      "loss": 46.0,
-      "step": 26046
-    },
-    {
-      "epoch": 1.991475046352046,
-      "grad_norm": 0.001456466969102621,
-      "learning_rate": 0.0001999980443767029,
-      "loss": 46.0,
-      "step": 26047
-    },
-    {
-      "epoch": 1.9915515033354358,
-      "grad_norm": 0.0011430311715230346,
-      "learning_rate": 0.00019999804422648167,
-      "loss": 46.0,
-      "step": 26048
-    },
-    {
-      "epoch": 1.9916279603188256,
-      "grad_norm": 0.0011806755792349577,
-      "learning_rate": 0.00019999804407625472,
-      "loss": 46.0,
-      "step": 26049
-    },
-    {
-      "epoch": 1.9917044173022154,
-      "grad_norm": 0.0009642531513236463,
-      "learning_rate": 0.000199998043926022,
-      "loss": 46.0,
-      "step": 26050
-    },
-    {
-      "epoch": 1.9917808742856051,
-      "grad_norm": 0.0027697968762367964,
-      "learning_rate": 0.00019999804377578354,
-      "loss": 46.0,
-      "step": 26051
-    },
-    {
-      "epoch": 1.991857331268995,
-      "grad_norm": 0.000881535466760397,
-      "learning_rate": 0.00019999804362553928,
-      "loss": 46.0,
-      "step": 26052
-    },
-    {
-      "epoch": 1.9919337882523847,
-      "grad_norm": 0.003264032071456313,
-      "learning_rate": 0.00019999804347528924,
-      "loss": 46.0,
-      "step": 26053
-    },
-    {
-      "epoch": 1.9920102452357742,
-      "grad_norm": 0.004112382885068655,
-      "learning_rate": 0.00019999804332503343,
-      "loss": 46.0,
-      "step": 26054
-    },
-    {
-      "epoch": 1.992086702219164,
-      "grad_norm": 0.0008862063987180591,
-      "learning_rate": 0.00019999804317477187,
-      "loss": 46.0,
-      "step": 26055
-    },
-    {
-      "epoch": 1.9921631592025537,
-      "grad_norm": 0.0017171716317534447,
-      "learning_rate": 0.00019999804302450454,
-      "loss": 46.0,
-      "step": 26056
-    },
-    {
-      "epoch": 1.9922396161859433,
-      "grad_norm": 0.001282471464946866,
-      "learning_rate": 0.0001999980428742314,
-      "loss": 46.0,
-      "step": 26057
-    },
-    {
-      "epoch": 1.992316073169333,
-      "grad_norm": 0.0034951879642903805,
-      "learning_rate": 0.00019999804272395255,
-      "loss": 46.0,
-      "step": 26058
-    },
-    {
-      "epoch": 1.9923925301527228,
-      "grad_norm": 0.0031374499667435884,
-      "learning_rate": 0.00019999804257366788,
-      "loss": 46.0,
-      "step": 26059
-    },
-    {
-      "epoch": 1.9924689871361125,
-      "grad_norm": 0.0011678824666887522,
-      "learning_rate": 0.00019999804242337748,
-      "loss": 46.0,
-      "step": 26060
-    },
-    {
-      "epoch": 1.9925454441195023,
-      "grad_norm": 0.0010851527331396937,
-      "learning_rate": 0.0001999980422730813,
-      "loss": 46.0,
-      "step": 26061
-    },
-    {
-      "epoch": 1.992621901102892,
-      "grad_norm": 0.0011215993436053395,
-      "learning_rate": 0.00019999804212277934,
-      "loss": 46.0,
-      "step": 26062
-    },
-    {
-      "epoch": 1.9926983580862818,
-      "grad_norm": 0.0023272421676665545,
-      "learning_rate": 0.00019999804197247162,
-      "loss": 46.0,
-      "step": 26063
-    },
-    {
-      "epoch": 1.9927748150696716,
-      "grad_norm": 0.0009803898865357041,
-      "learning_rate": 0.00019999804182215813,
-      "loss": 46.0,
-      "step": 26064
-    },
-    {
-      "epoch": 1.9928512720530611,
-      "grad_norm": 0.0008112943614833057,
-      "learning_rate": 0.00019999804167183886,
-      "loss": 46.0,
-      "step": 26065
-    },
-    {
-      "epoch": 1.9929277290364509,
-      "grad_norm": 0.00735689839348197,
-      "learning_rate": 0.00019999804152151383,
-      "loss": 46.0,
-      "step": 26066
-    },
-    {
-      "epoch": 1.9930041860198406,
-      "grad_norm": 0.003033566987141967,
-      "learning_rate": 0.00019999804137118301,
-      "loss": 46.0,
-      "step": 26067
-    },
-    {
-      "epoch": 1.9930806430032302,
-      "grad_norm": 0.0007896330207586288,
-      "learning_rate": 0.00019999804122084646,
-      "loss": 46.0,
-      "step": 26068
-    },
-    {
-      "epoch": 1.99315709998662,
-      "grad_norm": 0.0023039579391479492,
-      "learning_rate": 0.0001999980410705041,
-      "loss": 46.0,
-      "step": 26069
-    },
-    {
-      "epoch": 1.9932335569700097,
-      "grad_norm": 0.0030734494794160128,
-      "learning_rate": 0.00019999804092015602,
-      "loss": 46.0,
-      "step": 26070
-    },
-    {
-      "epoch": 1.9933100139533995,
-      "grad_norm": 0.0020111759658902884,
-      "learning_rate": 0.00019999804076980212,
-      "loss": 46.0,
-      "step": 26071
-    },
-    {
-      "epoch": 1.9933864709367892,
-      "grad_norm": 0.0006385320448316634,
-      "learning_rate": 0.00019999804061944247,
-      "loss": 46.0,
-      "step": 26072
-    },
-    {
-      "epoch": 1.993462927920179,
-      "grad_norm": 0.0008567767799831927,
-      "learning_rate": 0.00019999804046907707,
-      "loss": 46.0,
-      "step": 26073
-    },
-    {
-      "epoch": 1.9935393849035687,
-      "grad_norm": 0.0026293655391782522,
-      "learning_rate": 0.00019999804031870587,
-      "loss": 46.0,
-      "step": 26074
-    },
-    {
-      "epoch": 1.9936158418869585,
-      "grad_norm": 0.0008773842710070312,
-      "learning_rate": 0.00019999804016832893,
-      "loss": 46.0,
-      "step": 26075
-    },
-    {
-      "epoch": 1.993692298870348,
-      "grad_norm": 0.0010073985904455185,
-      "learning_rate": 0.0001999980400179462,
-      "loss": 46.0,
-      "step": 26076
-    },
-    {
-      "epoch": 1.9937687558537378,
-      "grad_norm": 0.0017932179616764188,
-      "learning_rate": 0.0001999980398675577,
-      "loss": 46.0,
-      "step": 26077
-    },
-    {
-      "epoch": 1.9938452128371273,
-      "grad_norm": 0.0030114867258816957,
-      "learning_rate": 0.00019999803971716346,
-      "loss": 46.0,
-      "step": 26078
-    },
-    {
-      "epoch": 1.993921669820517,
-      "grad_norm": 0.002553084399551153,
-      "learning_rate": 0.00019999803956676342,
-      "loss": 46.0,
-      "step": 26079
-    },
-    {
-      "epoch": 1.9939981268039069,
-      "grad_norm": 0.0013618904631584883,
-      "learning_rate": 0.0001999980394163576,
-      "loss": 46.0,
-      "step": 26080
-    },
-    {
-      "epoch": 1.9940745837872966,
-      "grad_norm": 0.0010776977287605405,
-      "learning_rate": 0.00019999803926594605,
-      "loss": 46.0,
-      "step": 26081
-    },
-    {
-      "epoch": 1.9941510407706864,
-      "grad_norm": 0.0017622618470340967,
-      "learning_rate": 0.0001999980391155287,
-      "loss": 46.0,
-      "step": 26082
-    },
-    {
-      "epoch": 1.9942274977540762,
-      "grad_norm": 0.0008909834432415664,
-      "learning_rate": 0.0001999980389651056,
-      "loss": 46.0,
-      "step": 26083
-    },
-    {
-      "epoch": 1.994303954737466,
-      "grad_norm": 0.0008263926720246673,
-      "learning_rate": 0.0001999980388146767,
-      "loss": 46.0,
-      "step": 26084
-    },
-    {
-      "epoch": 1.9943804117208557,
-      "grad_norm": 0.0010461872443556786,
-      "learning_rate": 0.00019999803866424206,
-      "loss": 46.0,
-      "step": 26085
-    },
-    {
-      "epoch": 1.9944568687042454,
-      "grad_norm": 0.0010621914407238364,
-      "learning_rate": 0.00019999803851380164,
-      "loss": 46.0,
-      "step": 26086
-    },
-    {
-      "epoch": 1.994533325687635,
-      "grad_norm": 0.00033374049235135317,
-      "learning_rate": 0.00019999803836335544,
-      "loss": 46.0,
-      "step": 26087
-    },
-    {
-      "epoch": 1.9946097826710247,
-      "grad_norm": 0.0009514071280136704,
-      "learning_rate": 0.0001999980382129035,
-      "loss": 46.0,
-      "step": 26088
-    },
-    {
-      "epoch": 1.9946862396544143,
-      "grad_norm": 0.0009135097498074174,
-      "learning_rate": 0.00019999803806244578,
-      "loss": 46.0,
-      "step": 26089
-    },
-    {
-      "epoch": 1.994762696637804,
-      "grad_norm": 0.0034728844184428453,
-      "learning_rate": 0.00019999803791198226,
-      "loss": 46.0,
-      "step": 26090
-    },
-    {
-      "epoch": 1.9948391536211938,
-      "grad_norm": 0.0013102206867188215,
-      "learning_rate": 0.000199998037761513,
-      "loss": 46.0,
-      "step": 26091
-    },
-    {
-      "epoch": 1.9949156106045836,
-      "grad_norm": 0.0035122144035995007,
-      "learning_rate": 0.00019999803761103796,
-      "loss": 46.0,
-      "step": 26092
-    },
-    {
-      "epoch": 1.9949920675879733,
-      "grad_norm": 0.0007056621252559125,
-      "learning_rate": 0.00019999803746055718,
-      "loss": 46.0,
-      "step": 26093
-    },
-    {
-      "epoch": 1.995068524571363,
-      "grad_norm": 0.0007049194537103176,
-      "learning_rate": 0.0001999980373100706,
-      "loss": 46.0,
-      "step": 26094
-    },
-    {
-      "epoch": 1.9951449815547528,
-      "grad_norm": 0.0006460077711381018,
-      "learning_rate": 0.00019999803715957826,
-      "loss": 46.0,
-      "step": 26095
-    },
-    {
-      "epoch": 1.9952214385381426,
-      "grad_norm": 0.0004916912293992937,
-      "learning_rate": 0.00019999803700908013,
-      "loss": 46.0,
-      "step": 26096
-    },
-    {
-      "epoch": 1.9952978955215324,
-      "grad_norm": 0.0008945545996539295,
-      "learning_rate": 0.00019999803685857625,
-      "loss": 46.0,
-      "step": 26097
-    },
-    {
-      "epoch": 1.995374352504922,
-      "grad_norm": 0.0024941160809248686,
-      "learning_rate": 0.0001999980367080666,
-      "loss": 46.0,
-      "step": 26098
-    },
-    {
-      "epoch": 1.9954508094883117,
-      "grad_norm": 0.0013370742090046406,
-      "learning_rate": 0.00019999803655755118,
-      "loss": 46.0,
-      "step": 26099
-    },
-    {
-      "epoch": 1.9955272664717012,
-      "grad_norm": 0.0011255971621721983,
-      "learning_rate": 0.00019999803640703,
-      "loss": 46.0,
-      "step": 26100
-    },
-    {
-      "epoch": 1.995603723455091,
-      "grad_norm": 0.00518334424123168,
-      "learning_rate": 0.00019999803625650304,
-      "loss": 46.0,
-      "step": 26101
-    },
-    {
-      "epoch": 1.9956801804384807,
-      "grad_norm": 0.0027693563606590033,
-      "learning_rate": 0.0001999980361059703,
-      "loss": 46.0,
-      "step": 26102
-    },
-    {
-      "epoch": 1.9957566374218705,
-      "grad_norm": 0.0012155806180089712,
-      "learning_rate": 0.0001999980359554318,
-      "loss": 46.0,
-      "step": 26103
-    },
-    {
-      "epoch": 1.9958330944052602,
-      "grad_norm": 0.0009681333904154599,
-      "learning_rate": 0.00019999803580488754,
-      "loss": 46.0,
-      "step": 26104
-    },
-    {
-      "epoch": 1.99590955138865,
-      "grad_norm": 0.0007021358469501138,
-      "learning_rate": 0.00019999803565433753,
-      "loss": 46.0,
-      "step": 26105
-    },
-    {
-      "epoch": 1.9959860083720398,
-      "grad_norm": 0.000937016389798373,
-      "learning_rate": 0.0001999980355037817,
-      "loss": 46.0,
-      "step": 26106
-    },
-    {
-      "epoch": 1.9960624653554295,
-      "grad_norm": 0.000996751245111227,
-      "learning_rate": 0.0001999980353532201,
-      "loss": 46.0,
-      "step": 26107
-    },
-    {
-      "epoch": 1.9961389223388193,
-      "grad_norm": 0.0006522854091599584,
-      "learning_rate": 0.00019999803520265278,
-      "loss": 46.0,
-      "step": 26108
-    },
-    {
-      "epoch": 1.9962153793222088,
-      "grad_norm": 0.001614025910384953,
-      "learning_rate": 0.00019999803505207968,
-      "loss": 46.0,
-      "step": 26109
-    },
-    {
-      "epoch": 1.9962918363055986,
-      "grad_norm": 0.0014393326127901673,
-      "learning_rate": 0.0001999980349015008,
-      "loss": 46.0,
-      "step": 26110
-    },
-    {
-      "epoch": 1.9963682932889881,
-      "grad_norm": 0.0014180174330249429,
-      "learning_rate": 0.00019999803475091615,
-      "loss": 46.0,
-      "step": 26111
-    },
-    {
-      "epoch": 1.9964447502723779,
-      "grad_norm": 0.0010683961445465684,
-      "learning_rate": 0.0001999980346003257,
-      "loss": 46.0,
-      "step": 26112
-    },
-    {
-      "epoch": 1.9965212072557676,
-      "grad_norm": 0.0029336409643292427,
-      "learning_rate": 0.0001999980344497295,
-      "loss": 46.0,
-      "step": 26113
-    },
-    {
-      "epoch": 1.9965976642391574,
-      "grad_norm": 0.0004483310040086508,
-      "learning_rate": 0.0001999980342991276,
-      "loss": 46.0,
-      "step": 26114
-    },
-    {
-      "epoch": 1.9966741212225472,
-      "grad_norm": 0.0025066423695534468,
-      "learning_rate": 0.00019999803414851982,
-      "loss": 46.0,
-      "step": 26115
-    },
-    {
-      "epoch": 1.996750578205937,
-      "grad_norm": 0.000863031717017293,
-      "learning_rate": 0.00019999803399790633,
-      "loss": 46.0,
-      "step": 26116
-    },
-    {
-      "epoch": 1.9968270351893267,
-      "grad_norm": 0.00048315984895452857,
-      "learning_rate": 0.00019999803384728707,
-      "loss": 46.0,
-      "step": 26117
-    },
-    {
-      "epoch": 1.9969034921727165,
-      "grad_norm": 0.0008483135025016963,
-      "learning_rate": 0.00019999803369666203,
-      "loss": 46.0,
-      "step": 26118
-    },
-    {
-      "epoch": 1.996979949156106,
-      "grad_norm": 0.0008925742004066706,
-      "learning_rate": 0.00019999803354603125,
-      "loss": 46.0,
-      "step": 26119
-    },
-    {
-      "epoch": 1.9970564061394958,
-      "grad_norm": 0.0008175320108421147,
-      "learning_rate": 0.00019999803339539464,
-      "loss": 46.0,
-      "step": 26120
-    },
-    {
-      "epoch": 1.9971328631228855,
-      "grad_norm": 0.002876169281080365,
-      "learning_rate": 0.0001999980332447523,
-      "loss": 46.0,
-      "step": 26121
-    },
-    {
-      "epoch": 1.997209320106275,
-      "grad_norm": 0.0006611901917494833,
-      "learning_rate": 0.00019999803309410418,
-      "loss": 46.0,
-      "step": 26122
-    },
-    {
-      "epoch": 1.9972857770896648,
-      "grad_norm": 0.0029720300808548927,
-      "learning_rate": 0.00019999803294345033,
-      "loss": 46.0,
-      "step": 26123
-    },
-    {
-      "epoch": 1.9973622340730546,
-      "grad_norm": 0.0038217692635953426,
-      "learning_rate": 0.00019999803279279065,
-      "loss": 46.0,
-      "step": 26124
-    },
-    {
-      "epoch": 1.9974386910564443,
-      "grad_norm": 0.0035990856122225523,
-      "learning_rate": 0.00019999803264212523,
-      "loss": 46.0,
-      "step": 26125
-    },
-    {
-      "epoch": 1.997515148039834,
-      "grad_norm": 0.0006057422142475843,
-      "learning_rate": 0.00019999803249145403,
-      "loss": 46.0,
-      "step": 26126
-    },
-    {
-      "epoch": 1.9975916050232239,
-      "grad_norm": 0.0006656552432104945,
-      "learning_rate": 0.0001999980323407771,
-      "loss": 46.0,
-      "step": 26127
-    },
-    {
-      "epoch": 1.9976680620066136,
-      "grad_norm": 0.0007760857697576284,
-      "learning_rate": 0.00019999803219009435,
-      "loss": 46.0,
-      "step": 26128
-    },
-    {
-      "epoch": 1.9977445189900034,
-      "grad_norm": 0.0007318463176488876,
-      "learning_rate": 0.00019999803203940583,
-      "loss": 46.0,
-      "step": 26129
-    },
-    {
-      "epoch": 1.997820975973393,
-      "grad_norm": 0.0005209353403188288,
-      "learning_rate": 0.00019999803188871157,
-      "loss": 46.0,
-      "step": 26130
-    },
-    {
-      "epoch": 1.9978974329567827,
-      "grad_norm": 0.0005699151079170406,
-      "learning_rate": 0.00019999803173801156,
-      "loss": 46.0,
-      "step": 26131
-    },
-    {
-      "epoch": 1.9979738899401724,
-      "grad_norm": 0.0031449166126549244,
-      "learning_rate": 0.00019999803158730572,
-      "loss": 46.0,
-      "step": 26132
-    },
-    {
-      "epoch": 1.998050346923562,
-      "grad_norm": 0.0007490452844649553,
-      "learning_rate": 0.00019999803143659414,
-      "loss": 46.0,
-      "step": 26133
-    },
-    {
-      "epoch": 1.9981268039069517,
-      "grad_norm": 0.0006352420896291733,
-      "learning_rate": 0.0001999980312858768,
-      "loss": 46.0,
-      "step": 26134
-    },
-    {
-      "epoch": 1.9982032608903415,
-      "grad_norm": 0.0009486954077146947,
-      "learning_rate": 0.0001999980311351537,
-      "loss": 46.0,
-      "step": 26135
-    },
-    {
-      "epoch": 1.9982797178737313,
-      "grad_norm": 0.0007956647896207869,
-      "learning_rate": 0.0001999980309844248,
-      "loss": 46.0,
-      "step": 26136
-    },
-    {
-      "epoch": 1.998356174857121,
-      "grad_norm": 0.0016845494974404573,
-      "learning_rate": 0.00019999803083369013,
-      "loss": 46.0,
-      "step": 26137
-    },
-    {
-      "epoch": 1.9984326318405108,
-      "grad_norm": 0.0012404262088239193,
-      "learning_rate": 0.00019999803068294973,
-      "loss": 46.0,
-      "step": 26138
-    },
-    {
-      "epoch": 1.9985090888239005,
-      "grad_norm": 0.0012078429572284222,
-      "learning_rate": 0.0001999980305322035,
-      "loss": 46.0,
-      "step": 26139
-    },
-    {
-      "epoch": 1.9985855458072903,
-      "grad_norm": 0.004205978475511074,
-      "learning_rate": 0.00019999803038145156,
-      "loss": 46.0,
-      "step": 26140
-    },
-    {
-      "epoch": 1.9986620027906798,
-      "grad_norm": 0.0024527739733457565,
-      "learning_rate": 0.00019999803023069382,
-      "loss": 46.0,
-      "step": 26141
-    },
-    {
-      "epoch": 1.9987384597740696,
-      "grad_norm": 0.002919460879638791,
-      "learning_rate": 0.00019999803007993033,
-      "loss": 46.0,
-      "step": 26142
-    },
-    {
-      "epoch": 1.9988149167574594,
-      "grad_norm": 0.0013226374285295606,
-      "learning_rate": 0.00019999802992916107,
-      "loss": 46.0,
-      "step": 26143
-    },
-    {
-      "epoch": 1.998891373740849,
-      "grad_norm": 0.0030463864095509052,
-      "learning_rate": 0.000199998029778386,
-      "loss": 46.0,
-      "step": 26144
-    },
-    {
-      "epoch": 1.9989678307242387,
-      "grad_norm": 0.0004927762784063816,
-      "learning_rate": 0.0001999980296276052,
-      "loss": 46.0,
-      "step": 26145
-    },
-    {
-      "epoch": 1.9990442877076284,
-      "grad_norm": 0.0010199008975178003,
-      "learning_rate": 0.00019999802947681861,
-      "loss": 46.0,
-      "step": 26146
-    },
-    {
-      "epoch": 1.9991207446910182,
-      "grad_norm": 0.0008828681311570108,
-      "learning_rate": 0.00019999802932602626,
-      "loss": 46.0,
-      "step": 26147
-    },
-    {
-      "epoch": 1.999197201674408,
-      "grad_norm": 0.0007143750553950667,
-      "learning_rate": 0.00019999802917522816,
-      "loss": 46.0,
-      "step": 26148
-    },
-    {
-      "epoch": 1.9992736586577977,
-      "grad_norm": 0.0005300296470522881,
-      "learning_rate": 0.00019999802902442428,
-      "loss": 46.0,
-      "step": 26149
-    },
-    {
-      "epoch": 1.9993501156411875,
-      "grad_norm": 0.0016304715536534786,
-      "learning_rate": 0.00019999802887361463,
-      "loss": 46.0,
-      "step": 26150
-    },
-    {
-      "epoch": 1.9994265726245772,
-      "grad_norm": 0.001515907235443592,
-      "learning_rate": 0.0001999980287227992,
-      "loss": 46.0,
-      "step": 26151
-    },
-    {
-      "epoch": 1.9995030296079668,
-      "grad_norm": 0.0008195392438210547,
-      "learning_rate": 0.00019999802857197801,
-      "loss": 46.0,
-      "step": 26152
-    },
-    {
-      "epoch": 1.9995794865913565,
-      "grad_norm": 0.0004347838694229722,
-      "learning_rate": 0.00019999802842115102,
-      "loss": 46.0,
-      "step": 26153
-    },
-    {
-      "epoch": 1.9996559435747463,
-      "grad_norm": 0.0007506748079322278,
-      "learning_rate": 0.0001999980282703183,
-      "loss": 46.0,
-      "step": 26154
-    },
-    {
-      "epoch": 1.9997324005581358,
-      "grad_norm": 0.0005916275549679995,
-      "learning_rate": 0.0001999980281194798,
-      "loss": 46.0,
-      "step": 26155
-    },
-    {
-      "epoch": 1.9998088575415256,
-      "grad_norm": 0.0009836844401434064,
-      "learning_rate": 0.00019999802796863555,
-      "loss": 46.0,
-      "step": 26156
-    },
-    {
-      "epoch": 1.9998853145249154,
-      "grad_norm": 0.0007093893946148455,
-      "learning_rate": 0.0001999980278177855,
-      "loss": 46.0,
-      "step": 26157
-    },
-    {
-      "epoch": 1.9999617715083051,
-      "grad_norm": 0.0006633367738686502,
-      "learning_rate": 0.00019999802766692968,
-      "loss": 46.0,
-      "step": 26158
-    },
-    {
-      "epoch": 1.9999617715083051,
-      "eval_loss": 11.5,
-      "eval_runtime": 31.9233,
-      "eval_samples_per_second": 172.538,
-      "eval_steps_per_second": 86.269,
-      "step": 26158
-    },
-    {
-      "epoch": 2.000038228491695,
-      "grad_norm": 0.004662771243602037,
-      "learning_rate": 0.0001999980275160681,
-      "loss": 46.0,
-      "step": 26159
-    },
-    {
-      "epoch": 2.0001146854750846,
-      "grad_norm": 0.0018271158915013075,
-      "learning_rate": 0.00019999802736520074,
-      "loss": 46.0,
-      "step": 26160
-    },
-    {
-      "epoch": 2.0001911424584744,
-      "grad_norm": 0.0010419727768748999,
-      "learning_rate": 0.00019999802721432764,
-      "loss": 46.0,
-      "step": 26161
-    },
-    {
-      "epoch": 2.000267599441864,
-      "grad_norm": 0.0009499929146841168,
-      "learning_rate": 0.00019999802706344874,
-      "loss": 46.0,
-      "step": 26162
-    },
-    {
-      "epoch": 2.000344056425254,
-      "grad_norm": 0.0011963375145569444,
-      "learning_rate": 0.0001999980269125641,
-      "loss": 46.0,
-      "step": 26163
-    },
-    {
-      "epoch": 2.0004205134086432,
-      "grad_norm": 0.0011605339823290706,
-      "learning_rate": 0.00019999802676167364,
-      "loss": 46.0,
-      "step": 26164
-    },
-    {
-      "epoch": 2.000496970392033,
-      "grad_norm": 0.0014711207477375865,
-      "learning_rate": 0.00019999802661077744,
-      "loss": 46.0,
-      "step": 26165
-    },
-    {
-      "epoch": 2.0005734273754228,
-      "grad_norm": 0.0037214034236967564,
-      "learning_rate": 0.00019999802645987547,
-      "loss": 46.0,
-      "step": 26166
-    },
-    {
-      "epoch": 2.0006498843588125,
-      "grad_norm": 0.0004070333670824766,
-      "learning_rate": 0.00019999802630896776,
-      "loss": 46.0,
-      "step": 26167
-    },
-    {
-      "epoch": 2.0007263413422023,
-      "grad_norm": 0.0008006632560864091,
-      "learning_rate": 0.00019999802615805424,
-      "loss": 46.0,
-      "step": 26168
-    },
-    {
-      "epoch": 2.000802798325592,
-      "grad_norm": 0.004580541513860226,
-      "learning_rate": 0.00019999802600713498,
-      "loss": 46.0,
-      "step": 26169
-    },
-    {
-      "epoch": 2.000879255308982,
-      "grad_norm": 0.0006492201937362552,
-      "learning_rate": 0.00019999802585620992,
-      "loss": 46.0,
-      "step": 26170
-    },
-    {
-      "epoch": 2.0009557122923716,
-      "grad_norm": 0.0020484321285039186,
-      "learning_rate": 0.0001999980257052791,
-      "loss": 46.0,
-      "step": 26171
-    },
-    {
-      "epoch": 2.0010321692757613,
-      "grad_norm": 0.0006456835544668138,
-      "learning_rate": 0.00019999802555434256,
-      "loss": 46.0,
-      "step": 26172
-    },
-    {
-      "epoch": 2.001108626259151,
-      "grad_norm": 0.0004316331760492176,
-      "learning_rate": 0.00019999802540340017,
-      "loss": 46.0,
-      "step": 26173
-    },
-    {
-      "epoch": 2.001185083242541,
-      "grad_norm": 0.0008028432494029403,
-      "learning_rate": 0.00019999802525245207,
-      "loss": 46.0,
-      "step": 26174
-    },
-    {
-      "epoch": 2.00126154022593,
-      "grad_norm": 0.0027233120054006577,
-      "learning_rate": 0.00019999802510149817,
-      "loss": 46.0,
-      "step": 26175
-    },
-    {
-      "epoch": 2.00133799720932,
-      "grad_norm": 0.0010254610097035766,
-      "learning_rate": 0.00019999802495053852,
-      "loss": 46.0,
-      "step": 26176
-    },
-    {
-      "epoch": 2.0014144541927097,
-      "grad_norm": 0.003934929613023996,
-      "learning_rate": 0.0001999980247995731,
-      "loss": 46.0,
-      "step": 26177
-    },
-    {
-      "epoch": 2.0014909111760995,
-      "grad_norm": 0.0006774758221581578,
-      "learning_rate": 0.0001999980246486019,
-      "loss": 46.0,
-      "step": 26178
-    },
-    {
-      "epoch": 2.001567368159489,
-      "grad_norm": 0.00043291132897138596,
-      "learning_rate": 0.00019999802449762494,
-      "loss": 46.0,
-      "step": 26179
-    },
-    {
-      "epoch": 2.001643825142879,
-      "grad_norm": 0.0008927814196795225,
-      "learning_rate": 0.0001999980243466422,
-      "loss": 46.0,
-      "step": 26180
-    },
-    {
-      "epoch": 2.0017202821262687,
-      "grad_norm": 0.0011041504330933094,
-      "learning_rate": 0.0001999980241956537,
-      "loss": 46.0,
-      "step": 26181
-    },
-    {
-      "epoch": 2.0017967391096585,
-      "grad_norm": 0.0015459402929991484,
-      "learning_rate": 0.00019999802404465942,
-      "loss": 46.0,
-      "step": 26182
-    },
-    {
-      "epoch": 2.0018731960930483,
-      "grad_norm": 0.0025369576178491116,
-      "learning_rate": 0.00019999802389365935,
-      "loss": 46.0,
-      "step": 26183
-    },
-    {
-      "epoch": 2.001949653076438,
-      "grad_norm": 0.001021634554490447,
-      "learning_rate": 0.00019999802374265355,
-      "loss": 46.0,
-      "step": 26184
-    },
-    {
-      "epoch": 2.002026110059828,
-      "grad_norm": 0.0010396811412647367,
-      "learning_rate": 0.00019999802359164196,
-      "loss": 46.0,
-      "step": 26185
-    },
-    {
-      "epoch": 2.002102567043217,
-      "grad_norm": 0.00226794951595366,
-      "learning_rate": 0.0001999980234406246,
-      "loss": 46.0,
-      "step": 26186
-    },
-    {
-      "epoch": 2.002179024026607,
-      "grad_norm": 0.002138368086889386,
-      "learning_rate": 0.00019999802328960148,
-      "loss": 46.0,
-      "step": 26187
-    },
-    {
-      "epoch": 2.0022554810099966,
-      "grad_norm": 0.0016221149126067758,
-      "learning_rate": 0.0001999980231385726,
-      "loss": 46.0,
-      "step": 26188
-    },
-    {
-      "epoch": 2.0023319379933864,
-      "grad_norm": 0.005963224917650223,
-      "learning_rate": 0.00019999802298753796,
-      "loss": 46.0,
-      "step": 26189
-    },
-    {
-      "epoch": 2.002408394976776,
-      "grad_norm": 0.0004824300413019955,
-      "learning_rate": 0.0001999980228364975,
-      "loss": 46.0,
-      "step": 26190
-    },
-    {
-      "epoch": 2.002484851960166,
-      "grad_norm": 0.002030147472396493,
-      "learning_rate": 0.00019999802268545131,
-      "loss": 46.0,
-      "step": 26191
-    },
-    {
-      "epoch": 2.0025613089435557,
-      "grad_norm": 0.005240280646830797,
-      "learning_rate": 0.00019999802253439935,
-      "loss": 46.0,
-      "step": 26192
-    },
-    {
-      "epoch": 2.0026377659269454,
-      "grad_norm": 0.0013183457776904106,
-      "learning_rate": 0.00019999802238334163,
-      "loss": 46.0,
-      "step": 26193
-    },
-    {
-      "epoch": 2.002714222910335,
-      "grad_norm": 0.0038624336011707783,
-      "learning_rate": 0.00019999802223227812,
-      "loss": 46.0,
-      "step": 26194
-    },
-    {
-      "epoch": 2.002790679893725,
-      "grad_norm": 0.001587224774993956,
-      "learning_rate": 0.0001999980220812088,
-      "loss": 46.0,
-      "step": 26195
-    },
-    {
-      "epoch": 2.0028671368771147,
-      "grad_norm": 0.0009375744266435504,
-      "learning_rate": 0.0001999980219301338,
-      "loss": 46.0,
-      "step": 26196
-    },
-    {
-      "epoch": 2.002943593860504,
-      "grad_norm": 0.006520427297800779,
-      "learning_rate": 0.00019999802177905296,
-      "loss": 46.0,
-      "step": 26197
-    },
-    {
-      "epoch": 2.003020050843894,
-      "grad_norm": 0.004415924660861492,
-      "learning_rate": 0.0001999980216279664,
-      "loss": 46.0,
-      "step": 26198
-    },
-    {
-      "epoch": 2.0030965078272835,
-      "grad_norm": 0.009158768691122532,
-      "learning_rate": 0.00019999802147687402,
-      "loss": 46.0,
-      "step": 26199
-    },
-    {
-      "epoch": 2.0031729648106733,
-      "grad_norm": 0.0010773178655654192,
-      "learning_rate": 0.0001999980213257759,
-      "loss": 46.0,
-      "step": 26200
-    },
-    {
-      "epoch": 2.003249421794063,
-      "grad_norm": 0.0032553726341575384,
-      "learning_rate": 0.00019999802117467202,
-      "loss": 46.0,
-      "step": 26201
-    },
-    {
-      "epoch": 2.003325878777453,
-      "grad_norm": 0.0009913387475535274,
-      "learning_rate": 0.00019999802102356235,
-      "loss": 46.0,
-      "step": 26202
-    },
-    {
-      "epoch": 2.0034023357608426,
-      "grad_norm": 0.000741445692256093,
-      "learning_rate": 0.00019999802087244692,
-      "loss": 46.0,
-      "step": 26203
-    },
-    {
-      "epoch": 2.0034787927442324,
-      "grad_norm": 0.0018948677461594343,
-      "learning_rate": 0.00019999802072132573,
-      "loss": 46.0,
-      "step": 26204
-    },
-    {
-      "epoch": 2.003555249727622,
-      "grad_norm": 0.002056448021903634,
-      "learning_rate": 0.00019999802057019873,
-      "loss": 46.0,
-      "step": 26205
-    },
-    {
-      "epoch": 2.003631706711012,
-      "grad_norm": 0.0007232400821521878,
-      "learning_rate": 0.000199998020419066,
-      "loss": 46.0,
-      "step": 26206
-    },
-    {
-      "epoch": 2.0037081636944016,
-      "grad_norm": 0.002205180935561657,
-      "learning_rate": 0.0001999980202679275,
-      "loss": 46.0,
-      "step": 26207
-    },
-    {
-      "epoch": 2.003784620677791,
-      "grad_norm": 0.0011409829603508115,
-      "learning_rate": 0.00019999802011678322,
-      "loss": 46.0,
-      "step": 26208
-    },
-    {
-      "epoch": 2.0038610776611807,
-      "grad_norm": 0.001275534275919199,
-      "learning_rate": 0.00019999801996563318,
-      "loss": 46.0,
-      "step": 26209
-    },
-    {
-      "epoch": 2.0039375346445705,
-      "grad_norm": 0.0022096901666373014,
-      "learning_rate": 0.00019999801981447735,
-      "loss": 46.0,
-      "step": 26210
-    },
-    {
-      "epoch": 2.0040139916279602,
-      "grad_norm": 0.0006243409006856382,
-      "learning_rate": 0.00019999801966331577,
-      "loss": 46.0,
-      "step": 26211
-    },
-    {
-      "epoch": 2.00409044861135,
-      "grad_norm": 0.0035669677890837193,
-      "learning_rate": 0.00019999801951214844,
-      "loss": 46.0,
-      "step": 26212
-    },
-    {
-      "epoch": 2.0041669055947398,
-      "grad_norm": 0.000551888020709157,
-      "learning_rate": 0.00019999801936097528,
-      "loss": 46.0,
-      "step": 26213
-    },
-    {
-      "epoch": 2.0042433625781295,
-      "grad_norm": 0.0009982648771256208,
-      "learning_rate": 0.00019999801920979638,
-      "loss": 46.0,
-      "step": 26214
-    },
-    {
-      "epoch": 2.0043198195615193,
-      "grad_norm": 0.0031910568941384554,
-      "learning_rate": 0.00019999801905861174,
-      "loss": 46.0,
-      "step": 26215
-    },
-    {
-      "epoch": 2.004396276544909,
-      "grad_norm": 0.0011441687820479274,
-      "learning_rate": 0.00019999801890742132,
-      "loss": 46.0,
-      "step": 26216
-    },
-    {
-      "epoch": 2.004472733528299,
-      "grad_norm": 0.0019939281046390533,
-      "learning_rate": 0.0001999980187562251,
-      "loss": 46.0,
-      "step": 26217
-    },
-    {
-      "epoch": 2.0045491905116886,
-      "grad_norm": 0.0009862144943326712,
-      "learning_rate": 0.00019999801860502313,
-      "loss": 46.0,
-      "step": 26218
-    },
-    {
-      "epoch": 2.004625647495078,
-      "grad_norm": 0.0004974405746906996,
-      "learning_rate": 0.00019999801845381539,
-      "loss": 46.0,
-      "step": 26219
-    },
-    {
-      "epoch": 2.0047021044784676,
-      "grad_norm": 0.0012085522757843137,
-      "learning_rate": 0.00019999801830260187,
-      "loss": 46.0,
-      "step": 26220
-    },
-    {
-      "epoch": 2.0047785614618574,
-      "grad_norm": 0.0007428917451761663,
-      "learning_rate": 0.00019999801815138258,
-      "loss": 46.0,
-      "step": 26221
-    },
-    {
-      "epoch": 2.004855018445247,
-      "grad_norm": 0.0011536243837326765,
-      "learning_rate": 0.00019999801800015755,
-      "loss": 46.0,
-      "step": 26222
-    },
-    {
-      "epoch": 2.004931475428637,
-      "grad_norm": 0.0007541900267824531,
-      "learning_rate": 0.00019999801784892674,
-      "loss": 46.0,
-      "step": 26223
-    },
-    {
-      "epoch": 2.0050079324120267,
-      "grad_norm": 0.0025192727334797382,
-      "learning_rate": 0.00019999801769769016,
-      "loss": 46.0,
-      "step": 26224
-    },
-    {
-      "epoch": 2.0050843893954164,
-      "grad_norm": 0.0008683687192387879,
-      "learning_rate": 0.00019999801754644778,
-      "loss": 46.0,
-      "step": 26225
-    },
-    {
-      "epoch": 2.005160846378806,
-      "grad_norm": 0.0012463143793866038,
-      "learning_rate": 0.00019999801739519965,
-      "loss": 46.0,
-      "step": 26226
-    },
-    {
-      "epoch": 2.005237303362196,
-      "grad_norm": 0.0021338139194995165,
-      "learning_rate": 0.00019999801724394578,
-      "loss": 46.0,
-      "step": 26227
-    },
-    {
-      "epoch": 2.0053137603455857,
-      "grad_norm": 0.0013347516069188714,
-      "learning_rate": 0.0001999980170926861,
-      "loss": 46.0,
-      "step": 26228
-    },
-    {
-      "epoch": 2.0053902173289755,
-      "grad_norm": 0.0004869582480750978,
-      "learning_rate": 0.00019999801694142066,
-      "loss": 46.0,
-      "step": 26229
-    },
-    {
-      "epoch": 2.005466674312365,
-      "grad_norm": 0.003374520456418395,
-      "learning_rate": 0.00019999801679014944,
-      "loss": 46.0,
-      "step": 26230
-    },
-    {
-      "epoch": 2.0055431312957546,
-      "grad_norm": 0.0006843886221759021,
-      "learning_rate": 0.0001999980166388725,
-      "loss": 46.0,
-      "step": 26231
-    },
-    {
-      "epoch": 2.0056195882791443,
-      "grad_norm": 0.0005083077703602612,
-      "learning_rate": 0.00019999801648758973,
-      "loss": 46.0,
-      "step": 26232
-    },
-    {
-      "epoch": 2.005696045262534,
-      "grad_norm": 0.0023535622749477625,
-      "learning_rate": 0.00019999801633630124,
-      "loss": 46.0,
-      "step": 26233
-    },
-    {
-      "epoch": 2.005772502245924,
-      "grad_norm": 0.00102468382101506,
-      "learning_rate": 0.00019999801618500695,
-      "loss": 46.0,
-      "step": 26234
-    },
-    {
-      "epoch": 2.0058489592293136,
-      "grad_norm": 0.0005148566560819745,
-      "learning_rate": 0.0001999980160337069,
-      "loss": 46.0,
-      "step": 26235
-    },
-    {
-      "epoch": 2.0059254162127034,
-      "grad_norm": 0.0004898388287983835,
-      "learning_rate": 0.00019999801588240109,
-      "loss": 46.0,
-      "step": 26236
-    },
-    {
-      "epoch": 2.006001873196093,
-      "grad_norm": 0.0011947706807404757,
-      "learning_rate": 0.0001999980157310895,
-      "loss": 46.0,
-      "step": 26237
-    },
-    {
-      "epoch": 2.006078330179483,
-      "grad_norm": 0.0011407856363803148,
-      "learning_rate": 0.00019999801557977212,
-      "loss": 46.0,
-      "step": 26238
-    },
-    {
-      "epoch": 2.0061547871628727,
-      "grad_norm": 0.0009961758041754365,
-      "learning_rate": 0.000199998015428449,
-      "loss": 46.0,
-      "step": 26239
-    },
-    {
-      "epoch": 2.0062312441462624,
-      "grad_norm": 0.0006893862737342715,
-      "learning_rate": 0.0001999980152771201,
-      "loss": 46.0,
-      "step": 26240
-    },
-    {
-      "epoch": 2.0063077011296517,
-      "grad_norm": 0.0017040036618709564,
-      "learning_rate": 0.00019999801512578545,
-      "loss": 46.0,
-      "step": 26241
-    },
-    {
-      "epoch": 2.0063841581130415,
-      "grad_norm": 0.003879678901284933,
-      "learning_rate": 0.000199998014974445,
-      "loss": 46.0,
-      "step": 26242
-    },
-    {
-      "epoch": 2.0064606150964313,
-      "grad_norm": 0.0016147332498803735,
-      "learning_rate": 0.0001999980148230988,
-      "loss": 46.0,
-      "step": 26243
-    },
-    {
-      "epoch": 2.006537072079821,
-      "grad_norm": 0.000772004306782037,
-      "learning_rate": 0.00019999801467174684,
-      "loss": 46.0,
-      "step": 26244
-    },
-    {
-      "epoch": 2.0066135290632108,
-      "grad_norm": 0.0011211322853341699,
-      "learning_rate": 0.0001999980145203891,
-      "loss": 46.0,
-      "step": 26245
-    },
-    {
-      "epoch": 2.0066899860466005,
-      "grad_norm": 0.00042306879186071455,
-      "learning_rate": 0.00019999801436902556,
-      "loss": 46.0,
-      "step": 26246
-    },
-    {
-      "epoch": 2.0067664430299903,
-      "grad_norm": 0.0012402822030708194,
-      "learning_rate": 0.0001999980142176563,
-      "loss": 46.0,
-      "step": 26247
-    },
-    {
-      "epoch": 2.00684290001338,
-      "grad_norm": 0.002053857082501054,
-      "learning_rate": 0.00019999801406628124,
-      "loss": 46.0,
-      "step": 26248
-    },
-    {
-      "epoch": 2.00691935699677,
-      "grad_norm": 0.0006379009573720396,
-      "learning_rate": 0.00019999801391490043,
-      "loss": 46.0,
-      "step": 26249
-    },
-    {
-      "epoch": 2.0069958139801596,
-      "grad_norm": 0.00084595667431131,
-      "learning_rate": 0.00019999801376351383,
-      "loss": 46.0,
-      "step": 26250
-    },
-    {
-      "epoch": 2.0070722709635493,
-      "grad_norm": 0.0027609895914793015,
-      "learning_rate": 0.00019999801361212147,
-      "loss": 46.0,
-      "step": 26251
-    },
-    {
-      "epoch": 2.0071487279469387,
-      "grad_norm": 0.0017518082167953253,
-      "learning_rate": 0.00019999801346072335,
-      "loss": 46.0,
-      "step": 26252
-    },
-    {
-      "epoch": 2.0072251849303284,
-      "grad_norm": 0.00341012142598629,
-      "learning_rate": 0.00019999801330931944,
-      "loss": 46.0,
-      "step": 26253
-    },
-    {
-      "epoch": 2.007301641913718,
-      "grad_norm": 0.0009974631248041987,
-      "learning_rate": 0.00019999801315790977,
-      "loss": 46.0,
-      "step": 26254
-    },
-    {
-      "epoch": 2.007378098897108,
-      "grad_norm": 0.0005263067432679236,
-      "learning_rate": 0.00019999801300649435,
-      "loss": 46.0,
-      "step": 26255
-    },
-    {
-      "epoch": 2.0074545558804977,
-      "grad_norm": 0.0030940650030970573,
-      "learning_rate": 0.00019999801285507313,
-      "loss": 46.0,
-      "step": 26256
-    },
-    {
-      "epoch": 2.0075310128638875,
-      "grad_norm": 0.0013548798160627484,
-      "learning_rate": 0.00019999801270364616,
-      "loss": 46.0,
-      "step": 26257
-    },
-    {
-      "epoch": 2.0076074698472772,
-      "grad_norm": 0.001492160721682012,
-      "learning_rate": 0.0001999980125522134,
-      "loss": 46.0,
-      "step": 26258
-    },
-    {
-      "epoch": 2.007683926830667,
-      "grad_norm": 0.0017141959397122264,
-      "learning_rate": 0.00019999801240077488,
-      "loss": 46.0,
-      "step": 26259
-    },
-    {
-      "epoch": 2.0077603838140567,
-      "grad_norm": 0.0003981981717515737,
-      "learning_rate": 0.00019999801224933062,
-      "loss": 46.0,
-      "step": 26260
-    },
-    {
-      "epoch": 2.0078368407974465,
-      "grad_norm": 0.0012443915475159883,
-      "learning_rate": 0.00019999801209788056,
-      "loss": 46.0,
-      "step": 26261
-    },
-    {
-      "epoch": 2.0079132977808363,
-      "grad_norm": 0.0016075114253908396,
-      "learning_rate": 0.00019999801194642473,
-      "loss": 46.0,
-      "step": 26262
-    },
-    {
-      "epoch": 2.0079897547642256,
-      "grad_norm": 0.002303424524143338,
-      "learning_rate": 0.00019999801179496312,
-      "loss": 46.0,
-      "step": 26263
-    },
-    {
-      "epoch": 2.0080662117476153,
-      "grad_norm": 0.0007340922602452338,
-      "learning_rate": 0.0001999980116434958,
-      "loss": 46.0,
-      "step": 26264
-    },
-    {
-      "epoch": 2.008142668731005,
-      "grad_norm": 0.0010445526568219066,
-      "learning_rate": 0.00019999801149202267,
-      "loss": 46.0,
-      "step": 26265
-    },
-    {
-      "epoch": 2.008219125714395,
-      "grad_norm": 0.003046773374080658,
-      "learning_rate": 0.00019999801134054374,
-      "loss": 46.0,
-      "step": 26266
-    },
-    {
-      "epoch": 2.0082955826977846,
-      "grad_norm": 0.0005591484368778765,
-      "learning_rate": 0.0001999980111890591,
-      "loss": 46.0,
-      "step": 26267
-    },
-    {
-      "epoch": 2.0083720396811744,
-      "grad_norm": 0.0007764684269204736,
-      "learning_rate": 0.00019999801103756862,
-      "loss": 46.0,
-      "step": 26268
-    },
-    {
-      "epoch": 2.008448496664564,
-      "grad_norm": 0.0018513009417802095,
-      "learning_rate": 0.00019999801088607243,
-      "loss": 46.0,
-      "step": 26269
-    },
-    {
-      "epoch": 2.008524953647954,
-      "grad_norm": 0.0007164459675550461,
-      "learning_rate": 0.00019999801073457046,
-      "loss": 46.0,
-      "step": 26270
-    },
-    {
-      "epoch": 2.0086014106313437,
-      "grad_norm": 0.0006123896455392241,
-      "learning_rate": 0.00019999801058306272,
-      "loss": 46.0,
-      "step": 26271
-    },
-    {
-      "epoch": 2.0086778676147334,
-      "grad_norm": 0.006820992566645145,
-      "learning_rate": 0.0001999980104315492,
-      "loss": 46.0,
-      "step": 26272
-    },
-    {
-      "epoch": 2.008754324598123,
-      "grad_norm": 0.0008229804807342589,
-      "learning_rate": 0.00019999801028002992,
-      "loss": 46.0,
-      "step": 26273
-    },
-    {
-      "epoch": 2.0088307815815125,
-      "grad_norm": 0.0026525594294071198,
-      "learning_rate": 0.00019999801012850487,
-      "loss": 46.0,
-      "step": 26274
-    },
-    {
-      "epoch": 2.0089072385649023,
-      "grad_norm": 0.0006436111871153116,
-      "learning_rate": 0.00019999800997697403,
-      "loss": 46.0,
-      "step": 26275
-    },
-    {
-      "epoch": 2.008983695548292,
-      "grad_norm": 0.005770350340753794,
-      "learning_rate": 0.00019999800982543743,
-      "loss": 46.0,
-      "step": 26276
-    },
-    {
-      "epoch": 2.009060152531682,
-      "grad_norm": 0.0012650461867451668,
-      "learning_rate": 0.00019999800967389507,
-      "loss": 46.0,
-      "step": 26277
-    },
-    {
-      "epoch": 2.0091366095150716,
-      "grad_norm": 0.002966303611174226,
-      "learning_rate": 0.00019999800952234695,
-      "loss": 46.0,
-      "step": 26278
-    },
-    {
-      "epoch": 2.0092130664984613,
-      "grad_norm": 0.002709926338866353,
-      "learning_rate": 0.00019999800937079305,
-      "loss": 46.0,
-      "step": 26279
-    },
-    {
-      "epoch": 2.009289523481851,
-      "grad_norm": 0.0008663387270644307,
-      "learning_rate": 0.00019999800921923337,
-      "loss": 46.0,
-      "step": 26280
-    },
-    {
-      "epoch": 2.009365980465241,
-      "grad_norm": 0.0016286926111206412,
-      "learning_rate": 0.00019999800906766795,
-      "loss": 46.0,
-      "step": 26281
-    },
-    {
-      "epoch": 2.0094424374486306,
-      "grad_norm": 0.00324411247856915,
-      "learning_rate": 0.00019999800891609673,
-      "loss": 46.0,
-      "step": 26282
-    },
-    {
-      "epoch": 2.0095188944320204,
-      "grad_norm": 0.0010447860695421696,
-      "learning_rate": 0.00019999800876451974,
-      "loss": 46.0,
-      "step": 26283
-    },
-    {
-      "epoch": 2.00959535141541,
-      "grad_norm": 0.0010546088451519608,
-      "learning_rate": 0.000199998008612937,
-      "loss": 46.0,
-      "step": 26284
-    },
-    {
-      "epoch": 2.0096718083987994,
-      "grad_norm": 0.0019695681985467672,
-      "learning_rate": 0.0001999980084613485,
-      "loss": 46.0,
-      "step": 26285
-    },
-    {
-      "epoch": 2.009748265382189,
-      "grad_norm": 0.00726426811888814,
-      "learning_rate": 0.0001999980083097542,
-      "loss": 46.0,
-      "step": 26286
-    },
-    {
-      "epoch": 2.009824722365579,
-      "grad_norm": 0.00047941997763700783,
-      "learning_rate": 0.00019999800815815414,
-      "loss": 46.0,
-      "step": 26287
-    },
-    {
-      "epoch": 2.0099011793489687,
-      "grad_norm": 0.0021181961055845022,
-      "learning_rate": 0.0001999980080065483,
-      "loss": 46.0,
-      "step": 26288
-    },
-    {
-      "epoch": 2.0099776363323585,
-      "grad_norm": 0.0018221414647996426,
-      "learning_rate": 0.00019999800785493673,
-      "loss": 46.0,
-      "step": 26289
-    },
-    {
-      "epoch": 2.0100540933157482,
-      "grad_norm": 0.0009569289977662265,
-      "learning_rate": 0.00019999800770331935,
-      "loss": 46.0,
-      "step": 26290
-    },
-    {
-      "epoch": 2.010130550299138,
-      "grad_norm": 0.0007281259750016034,
-      "learning_rate": 0.00019999800755169625,
-      "loss": 46.0,
-      "step": 26291
-    },
-    {
-      "epoch": 2.0102070072825278,
-      "grad_norm": 0.0034563683439046144,
-      "learning_rate": 0.0001999980074000673,
-      "loss": 46.0,
-      "step": 26292
-    },
-    {
-      "epoch": 2.0102834642659175,
-      "grad_norm": 0.0012553880224004388,
-      "learning_rate": 0.00019999800724843265,
-      "loss": 46.0,
-      "step": 26293
-    },
-    {
-      "epoch": 2.0103599212493073,
-      "grad_norm": 0.0018042916199192405,
-      "learning_rate": 0.0001999980070967922,
-      "loss": 46.0,
-      "step": 26294
-    },
-    {
-      "epoch": 2.0104363782326966,
-      "grad_norm": 0.0007203936693258584,
-      "learning_rate": 0.00019999800694514602,
-      "loss": 46.0,
-      "step": 26295
-    },
-    {
-      "epoch": 2.0105128352160864,
-      "grad_norm": 0.0013191327452659607,
-      "learning_rate": 0.00019999800679349402,
-      "loss": 46.0,
-      "step": 26296
-    },
-    {
-      "epoch": 2.010589292199476,
-      "grad_norm": 0.0009801177075132728,
-      "learning_rate": 0.00019999800664183628,
-      "loss": 46.0,
-      "step": 26297
-    },
-    {
-      "epoch": 2.010665749182866,
-      "grad_norm": 0.0008313339785672724,
-      "learning_rate": 0.00019999800649017274,
-      "loss": 46.0,
-      "step": 26298
-    },
-    {
-      "epoch": 2.0107422061662557,
-      "grad_norm": 0.003581333439797163,
-      "learning_rate": 0.00019999800633850346,
-      "loss": 46.0,
-      "step": 26299
-    },
-    {
-      "epoch": 2.0108186631496454,
-      "grad_norm": 0.0013816790888085961,
-      "learning_rate": 0.00019999800618682843,
-      "loss": 46.0,
-      "step": 26300
-    },
-    {
-      "epoch": 2.010895120133035,
-      "grad_norm": 0.0006482881144620478,
-      "learning_rate": 0.00019999800603514757,
-      "loss": 46.0,
-      "step": 26301
-    },
-    {
-      "epoch": 2.010971577116425,
-      "grad_norm": 0.0012095622951164842,
-      "learning_rate": 0.00019999800588346101,
-      "loss": 46.0,
-      "step": 26302
-    },
-    {
-      "epoch": 2.0110480340998147,
-      "grad_norm": 0.0011970933992415667,
-      "learning_rate": 0.0001999980057317686,
-      "loss": 46.0,
-      "step": 26303
-    },
-    {
-      "epoch": 2.0111244910832045,
-      "grad_norm": 0.0069392891600728035,
-      "learning_rate": 0.00019999800558007048,
-      "loss": 46.0,
-      "step": 26304
-    },
-    {
-      "epoch": 2.011200948066594,
-      "grad_norm": 0.0006005058530718088,
-      "learning_rate": 0.0001999980054283666,
-      "loss": 46.0,
-      "step": 26305
-    },
-    {
-      "epoch": 2.0112774050499835,
-      "grad_norm": 0.0013811110984534025,
-      "learning_rate": 0.0001999980052766569,
-      "loss": 46.0,
-      "step": 26306
-    },
-    {
-      "epoch": 2.0113538620333733,
-      "grad_norm": 0.0005704708746634424,
-      "learning_rate": 0.00019999800512494146,
-      "loss": 46.0,
-      "step": 26307
-    },
-    {
-      "epoch": 2.011430319016763,
-      "grad_norm": 0.0005890579777769744,
-      "learning_rate": 0.00019999800497322025,
-      "loss": 46.0,
-      "step": 26308
-    },
-    {
-      "epoch": 2.011506776000153,
-      "grad_norm": 0.002463872078806162,
-      "learning_rate": 0.00019999800482149328,
-      "loss": 46.0,
-      "step": 26309
-    },
-    {
-      "epoch": 2.0115832329835426,
-      "grad_norm": 0.0007687838515266776,
-      "learning_rate": 0.00019999800466976054,
-      "loss": 46.0,
-      "step": 26310
-    },
-    {
-      "epoch": 2.0116596899669323,
-      "grad_norm": 0.00044294408871792257,
-      "learning_rate": 0.000199998004518022,
-      "loss": 46.0,
-      "step": 26311
-    },
-    {
-      "epoch": 2.011736146950322,
-      "grad_norm": 0.0006065298803150654,
-      "learning_rate": 0.00019999800436627772,
-      "loss": 46.0,
-      "step": 26312
-    },
-    {
-      "epoch": 2.011812603933712,
-      "grad_norm": 0.0008302808273583651,
-      "learning_rate": 0.00019999800421452766,
-      "loss": 46.0,
-      "step": 26313
-    },
-    {
-      "epoch": 2.0118890609171016,
-      "grad_norm": 0.0008995538810268044,
-      "learning_rate": 0.00019999800406277183,
-      "loss": 46.0,
-      "step": 26314
-    },
-    {
-      "epoch": 2.0119655179004914,
-      "grad_norm": 0.0010585618438199162,
-      "learning_rate": 0.00019999800391101022,
-      "loss": 46.0,
-      "step": 26315
-    },
-    {
-      "epoch": 2.012041974883881,
-      "grad_norm": 0.00061994040152058,
-      "learning_rate": 0.00019999800375924287,
-      "loss": 46.0,
-      "step": 26316
-    },
-    {
-      "epoch": 2.0121184318672705,
-      "grad_norm": 0.000894365191925317,
-      "learning_rate": 0.00019999800360746974,
-      "loss": 46.0,
-      "step": 26317
-    },
-    {
-      "epoch": 2.0121948888506602,
-      "grad_norm": 0.0006692753522656858,
-      "learning_rate": 0.00019999800345569082,
-      "loss": 46.0,
-      "step": 26318
-    },
-    {
-      "epoch": 2.01227134583405,
-      "grad_norm": 0.0005519141559489071,
-      "learning_rate": 0.00019999800330390614,
-      "loss": 46.0,
-      "step": 26319
-    },
-    {
-      "epoch": 2.0123478028174397,
-      "grad_norm": 0.0019849473610520363,
-      "learning_rate": 0.00019999800315211573,
-      "loss": 46.0,
-      "step": 26320
-    },
-    {
-      "epoch": 2.0124242598008295,
-      "grad_norm": 0.0006093309493735433,
-      "learning_rate": 0.0001999980030003195,
-      "loss": 46.0,
-      "step": 26321
-    },
-    {
-      "epoch": 2.0125007167842193,
-      "grad_norm": 0.0018989529926329851,
-      "learning_rate": 0.00019999800284851752,
-      "loss": 46.0,
-      "step": 26322
-    },
-    {
-      "epoch": 2.012577173767609,
-      "grad_norm": 0.007206793874502182,
-      "learning_rate": 0.00019999800269670975,
-      "loss": 46.0,
-      "step": 26323
-    },
-    {
-      "epoch": 2.012653630750999,
-      "grad_norm": 0.0021626390516757965,
-      "learning_rate": 0.00019999800254489624,
-      "loss": 46.0,
-      "step": 26324
-    },
-    {
-      "epoch": 2.0127300877343886,
-      "grad_norm": 0.00046044550254009664,
-      "learning_rate": 0.00019999800239307693,
-      "loss": 46.0,
-      "step": 26325
-    },
-    {
-      "epoch": 2.0128065447177783,
-      "grad_norm": 0.001414434053003788,
-      "learning_rate": 0.0001999980022412519,
-      "loss": 46.0,
-      "step": 26326
-    },
-    {
-      "epoch": 2.012883001701168,
-      "grad_norm": 0.0021342714317142963,
-      "learning_rate": 0.00019999800208942106,
-      "loss": 46.0,
-      "step": 26327
-    },
-    {
-      "epoch": 2.0129594586845574,
-      "grad_norm": 0.001481067156419158,
-      "learning_rate": 0.00019999800193758446,
-      "loss": 46.0,
-      "step": 26328
-    },
-    {
-      "epoch": 2.013035915667947,
-      "grad_norm": 0.000655050971545279,
-      "learning_rate": 0.00019999800178574208,
-      "loss": 46.0,
-      "step": 26329
-    },
-    {
-      "epoch": 2.013112372651337,
-      "grad_norm": 0.015324120409786701,
-      "learning_rate": 0.00019999800163389395,
-      "loss": 46.0,
-      "step": 26330
-    },
-    {
-      "epoch": 2.0131888296347267,
-      "grad_norm": 0.015295472927391529,
-      "learning_rate": 0.00019999800148204005,
-      "loss": 46.0,
-      "step": 26331
-    },
-    {
-      "epoch": 2.0132652866181164,
-      "grad_norm": 0.0008215457201004028,
-      "learning_rate": 0.00019999800133018036,
-      "loss": 46.0,
-      "step": 26332
-    },
-    {
-      "epoch": 2.013341743601506,
-      "grad_norm": 0.0003924024640582502,
-      "learning_rate": 0.0001999980011783149,
-      "loss": 46.0,
-      "step": 26333
-    },
-    {
-      "epoch": 2.013418200584896,
-      "grad_norm": 0.0003850834327749908,
-      "learning_rate": 0.0001999980010264437,
-      "loss": 46.0,
-      "step": 26334
-    },
-    {
-      "epoch": 2.0134946575682857,
-      "grad_norm": 0.007263628300279379,
-      "learning_rate": 0.00019999800087456673,
-      "loss": 46.0,
-      "step": 26335
-    },
-    {
-      "epoch": 2.0135711145516755,
-      "grad_norm": 0.0025417886208742857,
-      "learning_rate": 0.000199998000722684,
-      "loss": 46.0,
-      "step": 26336
-    },
-    {
-      "epoch": 2.0136475715350652,
-      "grad_norm": 0.0022703923750668764,
-      "learning_rate": 0.00019999800057079545,
-      "loss": 46.0,
-      "step": 26337
-    },
-    {
-      "epoch": 2.013724028518455,
-      "grad_norm": 0.0010218846146017313,
-      "learning_rate": 0.00019999800041890117,
-      "loss": 46.0,
-      "step": 26338
-    },
-    {
-      "epoch": 2.0138004855018443,
-      "grad_norm": 0.0004849161487072706,
-      "learning_rate": 0.00019999800026700108,
-      "loss": 46.0,
-      "step": 26339
-    },
-    {
-      "epoch": 2.013876942485234,
-      "grad_norm": 0.001087265438400209,
-      "learning_rate": 0.00019999800011509528,
-      "loss": 46.0,
-      "step": 26340
-    },
-    {
-      "epoch": 2.013953399468624,
-      "grad_norm": 0.001947965007275343,
-      "learning_rate": 0.00019999799996318367,
-      "loss": 46.0,
-      "step": 26341
-    },
-    {
-      "epoch": 2.0140298564520136,
-      "grad_norm": 0.0014058618107810616,
-      "learning_rate": 0.0001999979998112663,
-      "loss": 46.0,
-      "step": 26342
-    },
-    {
-      "epoch": 2.0141063134354034,
-      "grad_norm": 0.0003884157631546259,
-      "learning_rate": 0.00019999799965934317,
-      "loss": 46.0,
-      "step": 26343
-    },
-    {
-      "epoch": 2.014182770418793,
-      "grad_norm": 0.0003111445694230497,
-      "learning_rate": 0.00019999799950741424,
-      "loss": 46.0,
-      "step": 26344
-    },
-    {
-      "epoch": 2.014259227402183,
-      "grad_norm": 0.0011368457926437259,
-      "learning_rate": 0.00019999799935547957,
-      "loss": 46.0,
-      "step": 26345
-    },
-    {
-      "epoch": 2.0143356843855726,
-      "grad_norm": 0.000923772226087749,
-      "learning_rate": 0.00019999799920353913,
-      "loss": 46.0,
-      "step": 26346
-    },
-    {
-      "epoch": 2.0144121413689624,
-      "grad_norm": 0.009267128072679043,
-      "learning_rate": 0.0001999979990515929,
-      "loss": 46.0,
-      "step": 26347
-    },
-    {
-      "epoch": 2.014488598352352,
-      "grad_norm": 0.0014039318775758147,
-      "learning_rate": 0.00019999799889964092,
-      "loss": 46.0,
-      "step": 26348
-    },
-    {
-      "epoch": 2.014565055335742,
-      "grad_norm": 0.0015578106977045536,
-      "learning_rate": 0.00019999799874768318,
-      "loss": 46.0,
-      "step": 26349
-    },
-    {
-      "epoch": 2.0146415123191312,
-      "grad_norm": 0.0006281944224610925,
-      "learning_rate": 0.00019999799859571964,
-      "loss": 46.0,
-      "step": 26350
-    },
-    {
-      "epoch": 2.014717969302521,
-      "grad_norm": 0.002535541309043765,
-      "learning_rate": 0.00019999799844375033,
-      "loss": 46.0,
-      "step": 26351
-    },
-    {
-      "epoch": 2.0147944262859108,
-      "grad_norm": 0.0014830816071480513,
-      "learning_rate": 0.00019999799829177527,
-      "loss": 46.0,
-      "step": 26352
-    },
-    {
-      "epoch": 2.0148708832693005,
-      "grad_norm": 0.0006594632286578417,
-      "learning_rate": 0.00019999799813979444,
-      "loss": 46.0,
-      "step": 26353
-    },
-    {
-      "epoch": 2.0149473402526903,
-      "grad_norm": 0.0027152851689606905,
-      "learning_rate": 0.00019999799798780786,
-      "loss": 46.0,
-      "step": 26354
-    },
-    {
-      "epoch": 2.01502379723608,
-      "grad_norm": 0.0011149500496685505,
-      "learning_rate": 0.00019999799783581548,
-      "loss": 46.0,
-      "step": 26355
-    },
-    {
-      "epoch": 2.01510025421947,
-      "grad_norm": 0.0028852594550698996,
-      "learning_rate": 0.00019999799768381733,
-      "loss": 46.0,
-      "step": 26356
-    },
-    {
-      "epoch": 2.0151767112028596,
-      "grad_norm": 0.002378114964812994,
-      "learning_rate": 0.00019999799753181343,
-      "loss": 46.0,
-      "step": 26357
-    },
-    {
-      "epoch": 2.0152531681862493,
-      "grad_norm": 0.0008286259835585952,
-      "learning_rate": 0.00019999799737980376,
-      "loss": 46.0,
-      "step": 26358
-    },
-    {
-      "epoch": 2.015329625169639,
-      "grad_norm": 0.0006026119226589799,
-      "learning_rate": 0.0001999979972277883,
-      "loss": 46.0,
-      "step": 26359
-    },
-    {
-      "epoch": 2.015406082153029,
-      "grad_norm": 0.0023870649747550488,
-      "learning_rate": 0.0001999979970757671,
-      "loss": 46.0,
-      "step": 26360
-    },
-    {
-      "epoch": 2.015482539136418,
-      "grad_norm": 0.0007586884312331676,
-      "learning_rate": 0.0001999979969237401,
-      "loss": 46.0,
-      "step": 26361
-    },
-    {
-      "epoch": 2.015558996119808,
-      "grad_norm": 0.0011644542682915926,
-      "learning_rate": 0.00019999799677170734,
-      "loss": 46.0,
-      "step": 26362
-    },
-    {
-      "epoch": 2.0156354531031977,
-      "grad_norm": 0.0014677379513159394,
-      "learning_rate": 0.00019999799661966883,
-      "loss": 46.0,
-      "step": 26363
-    },
-    {
-      "epoch": 2.0157119100865875,
-      "grad_norm": 0.0029217274859547615,
-      "learning_rate": 0.00019999799646762455,
-      "loss": 46.0,
-      "step": 26364
-    },
-    {
-      "epoch": 2.015788367069977,
-      "grad_norm": 0.002096845768392086,
-      "learning_rate": 0.00019999799631557446,
-      "loss": 46.0,
-      "step": 26365
-    },
-    {
-      "epoch": 2.015864824053367,
-      "grad_norm": 0.0031227206345647573,
-      "learning_rate": 0.00019999799616351863,
-      "loss": 46.0,
-      "step": 26366
-    },
-    {
-      "epoch": 2.0159412810367567,
-      "grad_norm": 0.0006880405708216131,
-      "learning_rate": 0.00019999799601145703,
-      "loss": 46.0,
-      "step": 26367
-    },
-    {
-      "epoch": 2.0160177380201465,
-      "grad_norm": 0.0011815892066806555,
-      "learning_rate": 0.00019999799585938965,
-      "loss": 46.0,
-      "step": 26368
-    },
-    {
-      "epoch": 2.0160941950035363,
-      "grad_norm": 0.0073180003091692924,
-      "learning_rate": 0.0001999979957073165,
-      "loss": 46.0,
-      "step": 26369
-    },
-    {
-      "epoch": 2.016170651986926,
-      "grad_norm": 0.002724348334595561,
-      "learning_rate": 0.0001999979955552376,
-      "loss": 46.0,
-      "step": 26370
-    },
-    {
-      "epoch": 2.016247108970316,
-      "grad_norm": 0.0005904012941755354,
-      "learning_rate": 0.0001999979954031529,
-      "loss": 46.0,
-      "step": 26371
-    },
-    {
-      "epoch": 2.016323565953705,
-      "grad_norm": 0.007443905342370272,
-      "learning_rate": 0.00019999799525106246,
-      "loss": 46.0,
-      "step": 26372
-    },
-    {
-      "epoch": 2.016400022937095,
-      "grad_norm": 0.0016939763445407152,
-      "learning_rate": 0.00019999799509896624,
-      "loss": 46.0,
-      "step": 26373
-    },
-    {
-      "epoch": 2.0164764799204846,
-      "grad_norm": 0.000984145444817841,
-      "learning_rate": 0.00019999799494686425,
-      "loss": 46.0,
-      "step": 26374
-    },
-    {
-      "epoch": 2.0165529369038744,
-      "grad_norm": 0.0018851279746741056,
-      "learning_rate": 0.00019999799479475652,
-      "loss": 46.0,
-      "step": 26375
-    },
-    {
-      "epoch": 2.016629393887264,
-      "grad_norm": 0.0027249157428741455,
-      "learning_rate": 0.00019999799464264298,
-      "loss": 46.0,
-      "step": 26376
-    },
-    {
-      "epoch": 2.016705850870654,
-      "grad_norm": 0.00255312561057508,
-      "learning_rate": 0.00019999799449052367,
-      "loss": 46.0,
-      "step": 26377
-    },
-    {
-      "epoch": 2.0167823078540437,
-      "grad_norm": 0.003908507991582155,
-      "learning_rate": 0.00019999799433839864,
-      "loss": 46.0,
-      "step": 26378
-    },
-    {
-      "epoch": 2.0168587648374334,
-      "grad_norm": 0.0006590376724489033,
-      "learning_rate": 0.00019999799418626778,
-      "loss": 46.0,
-      "step": 26379
-    },
-    {
-      "epoch": 2.016935221820823,
-      "grad_norm": 0.001372669474221766,
-      "learning_rate": 0.00019999799403413117,
-      "loss": 46.0,
-      "step": 26380
-    },
-    {
-      "epoch": 2.017011678804213,
-      "grad_norm": 0.0022291734348982573,
-      "learning_rate": 0.0001999979938819888,
-      "loss": 46.0,
-      "step": 26381
-    },
-    {
-      "epoch": 2.0170881357876027,
-      "grad_norm": 0.0011557965772226453,
-      "learning_rate": 0.00019999799372984067,
-      "loss": 46.0,
-      "step": 26382
-    },
-    {
-      "epoch": 2.017164592770992,
-      "grad_norm": 0.0014570147031918168,
-      "learning_rate": 0.00019999799357768675,
-      "loss": 46.0,
-      "step": 26383
-    },
-    {
-      "epoch": 2.017241049754382,
-      "grad_norm": 0.0031361987348645926,
-      "learning_rate": 0.00019999799342552708,
-      "loss": 46.0,
-      "step": 26384
-    },
-    {
-      "epoch": 2.0173175067377715,
-      "grad_norm": 0.0007653756183572114,
-      "learning_rate": 0.0001999979932733616,
-      "loss": 46.0,
-      "step": 26385
-    },
-    {
-      "epoch": 2.0173939637211613,
-      "grad_norm": 0.0031632056925445795,
-      "learning_rate": 0.0001999979931211904,
-      "loss": 46.0,
-      "step": 26386
-    },
-    {
-      "epoch": 2.017470420704551,
-      "grad_norm": 0.002237035194411874,
-      "learning_rate": 0.0001999979929690134,
-      "loss": 46.0,
-      "step": 26387
-    },
-    {
-      "epoch": 2.017546877687941,
-      "grad_norm": 0.0011668553343042731,
-      "learning_rate": 0.00019999799281683064,
-      "loss": 46.0,
-      "step": 26388
-    },
-    {
-      "epoch": 2.0176233346713306,
-      "grad_norm": 0.0007532284944318235,
-      "learning_rate": 0.00019999799266464213,
-      "loss": 46.0,
-      "step": 26389
-    },
-    {
-      "epoch": 2.0176997916547204,
-      "grad_norm": 0.0006863339222036302,
-      "learning_rate": 0.00019999799251244781,
-      "loss": 46.0,
-      "step": 26390
-    },
-    {
-      "epoch": 2.01777624863811,
-      "grad_norm": 0.0011143706506118178,
-      "learning_rate": 0.00019999799236024773,
-      "loss": 46.0,
-      "step": 26391
-    },
-    {
-      "epoch": 2.0178527056215,
-      "grad_norm": 0.0007110777078196406,
-      "learning_rate": 0.00019999799220804193,
-      "loss": 46.0,
-      "step": 26392
-    },
-    {
-      "epoch": 2.0179291626048896,
-      "grad_norm": 0.0007005173829384148,
-      "learning_rate": 0.00019999799205583032,
-      "loss": 46.0,
-      "step": 26393
-    },
-    {
-      "epoch": 2.018005619588279,
-      "grad_norm": 0.003621959825977683,
-      "learning_rate": 0.00019999799190361292,
-      "loss": 46.0,
-      "step": 26394
-    },
-    {
-      "epoch": 2.0180820765716687,
-      "grad_norm": 0.000936551543418318,
-      "learning_rate": 0.0001999979917513898,
-      "loss": 46.0,
-      "step": 26395
-    },
-    {
-      "epoch": 2.0181585335550585,
-      "grad_norm": 0.0008430216112174094,
-      "learning_rate": 0.00019999799159916087,
-      "loss": 46.0,
-      "step": 26396
-    },
-    {
-      "epoch": 2.0182349905384482,
-      "grad_norm": 0.0008361685322597623,
-      "learning_rate": 0.0001999979914469262,
-      "loss": 46.0,
-      "step": 26397
-    },
-    {
-      "epoch": 2.018311447521838,
-      "grad_norm": 0.0008514817454852164,
-      "learning_rate": 0.00019999799129468576,
-      "loss": 46.0,
-      "step": 26398
-    },
-    {
-      "epoch": 2.0183879045052278,
-      "grad_norm": 0.004284955561161041,
-      "learning_rate": 0.00019999799114243954,
-      "loss": 46.0,
-      "step": 26399
-    },
-    {
-      "epoch": 2.0184643614886175,
-      "grad_norm": 0.0005054807988926768,
-      "learning_rate": 0.00019999799099018752,
-      "loss": 46.0,
-      "step": 26400
-    },
-    {
-      "epoch": 2.0185408184720073,
-      "grad_norm": 0.0011327427346259356,
-      "learning_rate": 0.0001999979908379298,
-      "loss": 46.0,
-      "step": 26401
-    },
-    {
-      "epoch": 2.018617275455397,
-      "grad_norm": 0.0006181719363667071,
-      "learning_rate": 0.00019999799068566625,
-      "loss": 46.0,
-      "step": 26402
-    },
-    {
-      "epoch": 2.018693732438787,
-      "grad_norm": 0.002397105796262622,
-      "learning_rate": 0.00019999799053339694,
-      "loss": 46.0,
-      "step": 26403
-    },
-    {
-      "epoch": 2.0187701894221766,
-      "grad_norm": 0.002497076755389571,
-      "learning_rate": 0.00019999799038112186,
-      "loss": 46.0,
-      "step": 26404
-    },
-    {
-      "epoch": 2.018846646405566,
-      "grad_norm": 0.0006338916718959808,
-      "learning_rate": 0.00019999799022884105,
-      "loss": 46.0,
-      "step": 26405
-    },
-    {
-      "epoch": 2.0189231033889556,
-      "grad_norm": 0.0025392379611730576,
-      "learning_rate": 0.00019999799007655442,
-      "loss": 46.0,
-      "step": 26406
-    },
-    {
-      "epoch": 2.0189995603723454,
-      "grad_norm": 0.0035979480016976595,
-      "learning_rate": 0.00019999798992426207,
-      "loss": 46.0,
-      "step": 26407
-    },
-    {
-      "epoch": 2.019076017355735,
-      "grad_norm": 0.000816471700090915,
-      "learning_rate": 0.0001999979897719639,
-      "loss": 46.0,
-      "step": 26408
-    },
-    {
-      "epoch": 2.019152474339125,
-      "grad_norm": 0.005113798659294844,
-      "learning_rate": 0.00019999798961966,
-      "loss": 46.0,
-      "step": 26409
-    },
-    {
-      "epoch": 2.0192289313225147,
-      "grad_norm": 0.0008697230950929224,
-      "learning_rate": 0.0001999979894673503,
-      "loss": 46.0,
-      "step": 26410
-    },
-    {
-      "epoch": 2.0193053883059044,
-      "grad_norm": 0.0015793988713994622,
-      "learning_rate": 0.00019999798931503486,
-      "loss": 46.0,
-      "step": 26411
-    },
-    {
-      "epoch": 2.019381845289294,
-      "grad_norm": 0.0028395401313900948,
-      "learning_rate": 0.00019999798916271364,
-      "loss": 46.0,
-      "step": 26412
-    },
-    {
-      "epoch": 2.019458302272684,
-      "grad_norm": 0.000814637343864888,
-      "learning_rate": 0.00019999798901038665,
-      "loss": 46.0,
-      "step": 26413
-    },
-    {
-      "epoch": 2.0195347592560737,
-      "grad_norm": 0.0012641982175409794,
-      "learning_rate": 0.0001999979888580539,
-      "loss": 46.0,
-      "step": 26414
-    },
-    {
-      "epoch": 2.0196112162394635,
-      "grad_norm": 0.002331336960196495,
-      "learning_rate": 0.00019999798870571532,
-      "loss": 46.0,
-      "step": 26415
-    },
-    {
-      "epoch": 2.019687673222853,
-      "grad_norm": 0.0005343082593753934,
-      "learning_rate": 0.00019999798855337104,
-      "loss": 46.0,
-      "step": 26416
-    },
-    {
-      "epoch": 2.0197641302062426,
-      "grad_norm": 0.00045682990457862616,
-      "learning_rate": 0.00019999798840102099,
-      "loss": 46.0,
-      "step": 26417
-    },
-    {
-      "epoch": 2.0198405871896323,
-      "grad_norm": 0.0036687185056507587,
-      "learning_rate": 0.00019999798824866513,
-      "loss": 46.0,
-      "step": 26418
-    },
-    {
-      "epoch": 2.019917044173022,
-      "grad_norm": 0.0012728706933557987,
-      "learning_rate": 0.00019999798809630353,
-      "loss": 46.0,
-      "step": 26419
-    },
-    {
-      "epoch": 2.019993501156412,
-      "grad_norm": 0.001992689212784171,
-      "learning_rate": 0.00019999798794393615,
-      "loss": 46.0,
-      "step": 26420
-    },
-    {
-      "epoch": 2.0200699581398016,
-      "grad_norm": 0.0006950357928872108,
-      "learning_rate": 0.000199997987791563,
-      "loss": 46.0,
-      "step": 26421
-    },
-    {
-      "epoch": 2.0201464151231914,
-      "grad_norm": 0.003074574051424861,
-      "learning_rate": 0.00019999798763918408,
-      "loss": 46.0,
-      "step": 26422
-    },
-    {
-      "epoch": 2.020222872106581,
-      "grad_norm": 0.008224640972912312,
-      "learning_rate": 0.00019999798748679938,
-      "loss": 46.0,
-      "step": 26423
-    },
-    {
-      "epoch": 2.020299329089971,
-      "grad_norm": 0.000960544275585562,
-      "learning_rate": 0.00019999798733440894,
-      "loss": 46.0,
-      "step": 26424
-    },
-    {
-      "epoch": 2.0203757860733607,
-      "grad_norm": 0.005144823808223009,
-      "learning_rate": 0.0001999979871820127,
-      "loss": 46.0,
-      "step": 26425
-    },
-    {
-      "epoch": 2.0204522430567504,
-      "grad_norm": 0.0009282471728511155,
-      "learning_rate": 0.00019999798702961073,
-      "loss": 46.0,
-      "step": 26426
-    },
-    {
-      "epoch": 2.0205287000401397,
-      "grad_norm": 0.0011337038595229387,
-      "learning_rate": 0.00019999798687720297,
-      "loss": 46.0,
-      "step": 26427
-    },
-    {
-      "epoch": 2.0206051570235295,
-      "grad_norm": 0.0007164134876802564,
-      "learning_rate": 0.00019999798672478943,
-      "loss": 46.0,
-      "step": 26428
-    },
-    {
-      "epoch": 2.0206816140069193,
-      "grad_norm": 0.000708968669641763,
-      "learning_rate": 0.00019999798657237012,
-      "loss": 46.0,
-      "step": 26429
-    },
-    {
-      "epoch": 2.020758070990309,
-      "grad_norm": 0.0016046749660745263,
-      "learning_rate": 0.00019999798641994504,
-      "loss": 46.0,
-      "step": 26430
-    },
-    {
-      "epoch": 2.020834527973699,
-      "grad_norm": 0.0017448560101911426,
-      "learning_rate": 0.00019999798626751423,
-      "loss": 46.0,
-      "step": 26431
-    },
-    {
-      "epoch": 2.0209109849570885,
-      "grad_norm": 0.00240304390899837,
-      "learning_rate": 0.0001999979861150776,
-      "loss": 46.0,
-      "step": 26432
-    },
-    {
-      "epoch": 2.0209874419404783,
-      "grad_norm": 0.0013527818955481052,
-      "learning_rate": 0.0001999979859626352,
-      "loss": 46.0,
-      "step": 26433
-    },
-    {
-      "epoch": 2.021063898923868,
-      "grad_norm": 0.0011102689895778894,
-      "learning_rate": 0.00019999798581018708,
-      "loss": 46.0,
-      "step": 26434
-    },
-    {
-      "epoch": 2.021140355907258,
-      "grad_norm": 0.0008913030615076423,
-      "learning_rate": 0.00019999798565773316,
-      "loss": 46.0,
-      "step": 26435
-    },
-    {
-      "epoch": 2.0212168128906476,
-      "grad_norm": 0.0023256440181285143,
-      "learning_rate": 0.00019999798550527346,
-      "loss": 46.0,
-      "step": 26436
-    },
-    {
-      "epoch": 2.021293269874037,
-      "grad_norm": 0.0020580971613526344,
-      "learning_rate": 0.00019999798535280802,
-      "loss": 46.0,
-      "step": 26437
-    },
-    {
-      "epoch": 2.0213697268574267,
-      "grad_norm": 0.007729441858828068,
-      "learning_rate": 0.0001999979852003368,
-      "loss": 46.0,
-      "step": 26438
-    },
-    {
-      "epoch": 2.0214461838408164,
-      "grad_norm": 0.003236080752685666,
-      "learning_rate": 0.0001999979850478598,
-      "loss": 46.0,
-      "step": 26439
-    },
-    {
-      "epoch": 2.021522640824206,
-      "grad_norm": 0.0015236795879900455,
-      "learning_rate": 0.00019999798489537702,
-      "loss": 46.0,
-      "step": 26440
-    },
-    {
-      "epoch": 2.021599097807596,
-      "grad_norm": 0.0006803652504459023,
-      "learning_rate": 0.00019999798474288849,
-      "loss": 46.0,
-      "step": 26441
-    },
-    {
-      "epoch": 2.0216755547909857,
-      "grad_norm": 0.004668922629207373,
-      "learning_rate": 0.00019999798459039418,
-      "loss": 46.0,
-      "step": 26442
-    },
-    {
-      "epoch": 2.0217520117743755,
-      "grad_norm": 0.0008509972831234336,
-      "learning_rate": 0.00019999798443789412,
-      "loss": 46.0,
-      "step": 26443
-    },
-    {
-      "epoch": 2.0218284687577652,
-      "grad_norm": 0.0009522740147076547,
-      "learning_rate": 0.00019999798428538826,
-      "loss": 46.0,
-      "step": 26444
-    },
-    {
-      "epoch": 2.021904925741155,
-      "grad_norm": 0.000667439540848136,
-      "learning_rate": 0.00019999798413287663,
-      "loss": 46.0,
-      "step": 26445
-    },
-    {
-      "epoch": 2.0219813827245448,
-      "grad_norm": 0.0010644018184393644,
-      "learning_rate": 0.00019999798398035926,
-      "loss": 46.0,
-      "step": 26446
-    },
-    {
-      "epoch": 2.0220578397079345,
-      "grad_norm": 0.002345553133636713,
-      "learning_rate": 0.0001999979838278361,
-      "loss": 46.0,
-      "step": 26447
-    },
-    {
-      "epoch": 2.022134296691324,
-      "grad_norm": 0.000988130341283977,
-      "learning_rate": 0.0001999979836753072,
-      "loss": 46.0,
-      "step": 26448
-    },
-    {
-      "epoch": 2.0222107536747136,
-      "grad_norm": 0.0022990063298493624,
-      "learning_rate": 0.00019999798352277252,
-      "loss": 46.0,
-      "step": 26449
-    },
-    {
-      "epoch": 2.0222872106581034,
-      "grad_norm": 0.00030240416526794434,
-      "learning_rate": 0.00019999798337023208,
-      "loss": 46.0,
-      "step": 26450
-    },
-    {
-      "epoch": 2.022363667641493,
-      "grad_norm": 0.001114911399781704,
-      "learning_rate": 0.00019999798321768583,
-      "loss": 46.0,
-      "step": 26451
-    },
-    {
-      "epoch": 2.022440124624883,
-      "grad_norm": 0.002209915081039071,
-      "learning_rate": 0.00019999798306513384,
-      "loss": 46.0,
-      "step": 26452
-    },
-    {
-      "epoch": 2.0225165816082726,
-      "grad_norm": 0.0006973057752475142,
-      "learning_rate": 0.00019999798291257608,
-      "loss": 46.0,
-      "step": 26453
-    },
-    {
-      "epoch": 2.0225930385916624,
-      "grad_norm": 0.0023166206665337086,
-      "learning_rate": 0.00019999798276001255,
-      "loss": 46.0,
-      "step": 26454
-    },
-    {
-      "epoch": 2.022669495575052,
-      "grad_norm": 0.0005569932982325554,
-      "learning_rate": 0.00019999798260744324,
-      "loss": 46.0,
-      "step": 26455
-    },
-    {
-      "epoch": 2.022745952558442,
-      "grad_norm": 0.0023603159934282303,
-      "learning_rate": 0.00019999798245486815,
-      "loss": 46.0,
-      "step": 26456
-    },
-    {
-      "epoch": 2.0228224095418317,
-      "grad_norm": 0.0009010782232508063,
-      "learning_rate": 0.00019999798230228733,
-      "loss": 46.0,
-      "step": 26457
-    },
-    {
-      "epoch": 2.0228988665252214,
-      "grad_norm": 0.0006541848997585475,
-      "learning_rate": 0.00019999798214970072,
-      "loss": 46.0,
-      "step": 26458
-    },
-    {
-      "epoch": 2.0229753235086108,
-      "grad_norm": 0.005989361554384232,
-      "learning_rate": 0.00019999798199710832,
-      "loss": 46.0,
-      "step": 26459
-    },
-    {
-      "epoch": 2.0230517804920005,
-      "grad_norm": 0.0009209500858560205,
-      "learning_rate": 0.00019999798184451017,
-      "loss": 46.0,
-      "step": 26460
-    },
-    {
-      "epoch": 2.0231282374753903,
-      "grad_norm": 0.0010172826005145907,
-      "learning_rate": 0.00019999798169190625,
-      "loss": 46.0,
-      "step": 26461
-    },
-    {
-      "epoch": 2.02320469445878,
-      "grad_norm": 0.0018289603758603334,
-      "learning_rate": 0.00019999798153929655,
-      "loss": 46.0,
-      "step": 26462
-    },
-    {
-      "epoch": 2.02328115144217,
-      "grad_norm": 0.0005168657517060637,
-      "learning_rate": 0.0001999979813866811,
-      "loss": 46.0,
-      "step": 26463
-    },
-    {
-      "epoch": 2.0233576084255596,
-      "grad_norm": 0.0009984978241845965,
-      "learning_rate": 0.00019999798123405987,
-      "loss": 46.0,
-      "step": 26464
-    },
-    {
-      "epoch": 2.0234340654089493,
-      "grad_norm": 0.002276997547596693,
-      "learning_rate": 0.00019999798108143288,
-      "loss": 46.0,
-      "step": 26465
-    },
-    {
-      "epoch": 2.023510522392339,
-      "grad_norm": 0.0005757012986578047,
-      "learning_rate": 0.00019999798092880012,
-      "loss": 46.0,
-      "step": 26466
-    },
-    {
-      "epoch": 2.023586979375729,
-      "grad_norm": 0.0009865592001006007,
-      "learning_rate": 0.00019999798077616156,
-      "loss": 46.0,
-      "step": 26467
-    },
-    {
-      "epoch": 2.0236634363591186,
-      "grad_norm": 0.0007731883670203388,
-      "learning_rate": 0.00019999798062351727,
-      "loss": 46.0,
-      "step": 26468
-    },
-    {
-      "epoch": 2.0237398933425084,
-      "grad_norm": 0.0033960232976824045,
-      "learning_rate": 0.0001999979804708672,
-      "loss": 46.0,
-      "step": 26469
-    },
-    {
-      "epoch": 2.0238163503258977,
-      "grad_norm": 0.001232285751029849,
-      "learning_rate": 0.00019999798031821136,
-      "loss": 46.0,
-      "step": 26470
-    },
-    {
-      "epoch": 2.0238928073092874,
-      "grad_norm": 0.0022852083202451468,
-      "learning_rate": 0.00019999798016554973,
-      "loss": 46.0,
-      "step": 26471
-    },
-    {
-      "epoch": 2.023969264292677,
-      "grad_norm": 0.0045961979776620865,
-      "learning_rate": 0.00019999798001288236,
-      "loss": 46.0,
-      "step": 26472
-    },
-    {
-      "epoch": 2.024045721276067,
-      "grad_norm": 0.0006669955910183489,
-      "learning_rate": 0.00019999797986020918,
-      "loss": 46.0,
-      "step": 26473
-    },
-    {
-      "epoch": 2.0241221782594567,
-      "grad_norm": 0.002898002043366432,
-      "learning_rate": 0.00019999797970753026,
-      "loss": 46.0,
-      "step": 26474
-    },
-    {
-      "epoch": 2.0241986352428465,
-      "grad_norm": 0.0019249245524406433,
-      "learning_rate": 0.0001999979795548456,
-      "loss": 46.0,
-      "step": 26475
-    },
-    {
-      "epoch": 2.0242750922262363,
-      "grad_norm": 0.0028629314620047808,
-      "learning_rate": 0.00019999797940215512,
-      "loss": 46.0,
-      "step": 26476
-    },
-    {
-      "epoch": 2.024351549209626,
-      "grad_norm": 0.0016173362964764237,
-      "learning_rate": 0.00019999797924945888,
-      "loss": 46.0,
-      "step": 26477
-    },
-    {
-      "epoch": 2.0244280061930158,
-      "grad_norm": 0.0057194409891963005,
-      "learning_rate": 0.0001999979790967569,
-      "loss": 46.0,
-      "step": 26478
-    },
-    {
-      "epoch": 2.0245044631764055,
-      "grad_norm": 0.0009702262468636036,
-      "learning_rate": 0.00019999797894404913,
-      "loss": 46.0,
-      "step": 26479
-    },
-    {
-      "epoch": 2.0245809201597953,
-      "grad_norm": 0.002342892112210393,
-      "learning_rate": 0.00019999797879133557,
-      "loss": 46.0,
-      "step": 26480
-    },
-    {
-      "epoch": 2.0246573771431846,
-      "grad_norm": 0.0010883017675951123,
-      "learning_rate": 0.00019999797863861626,
-      "loss": 46.0,
-      "step": 26481
-    },
-    {
-      "epoch": 2.0247338341265744,
-      "grad_norm": 0.0021422302816063166,
-      "learning_rate": 0.00019999797848589118,
-      "loss": 46.0,
-      "step": 26482
-    },
-    {
-      "epoch": 2.024810291109964,
-      "grad_norm": 0.0022096848115324974,
-      "learning_rate": 0.00019999797833316035,
-      "loss": 46.0,
-      "step": 26483
-    },
-    {
-      "epoch": 2.024886748093354,
-      "grad_norm": 0.00527407368645072,
-      "learning_rate": 0.00019999797818042375,
-      "loss": 46.0,
-      "step": 26484
-    },
-    {
-      "epoch": 2.0249632050767437,
-      "grad_norm": 0.0013935323804616928,
-      "learning_rate": 0.00019999797802768137,
-      "loss": 46.0,
-      "step": 26485
-    },
-    {
-      "epoch": 2.0250396620601334,
-      "grad_norm": 0.0009756692452356219,
-      "learning_rate": 0.0001999979778749332,
-      "loss": 46.0,
-      "step": 26486
-    },
-    {
-      "epoch": 2.025116119043523,
-      "grad_norm": 0.0007487956318072975,
-      "learning_rate": 0.00019999797772217928,
-      "loss": 46.0,
-      "step": 26487
-    },
-    {
-      "epoch": 2.025192576026913,
-      "grad_norm": 0.0005695989821106195,
-      "learning_rate": 0.00019999797756941958,
-      "loss": 46.0,
-      "step": 26488
-    },
-    {
-      "epoch": 2.0252690330103027,
-      "grad_norm": 0.0005050969775766134,
-      "learning_rate": 0.00019999797741665414,
-      "loss": 46.0,
-      "step": 26489
-    },
-    {
-      "epoch": 2.0253454899936925,
-      "grad_norm": 0.0018592077540233731,
-      "learning_rate": 0.0001999979772638829,
-      "loss": 46.0,
-      "step": 26490
-    },
-    {
-      "epoch": 2.025421946977082,
-      "grad_norm": 0.0006684542750008404,
-      "learning_rate": 0.00019999797711110588,
-      "loss": 46.0,
-      "step": 26491
-    },
-    {
-      "epoch": 2.0254984039604715,
-      "grad_norm": 0.0018544172635301948,
-      "learning_rate": 0.00019999797695832312,
-      "loss": 46.0,
-      "step": 26492
-    },
-    {
-      "epoch": 2.0255748609438613,
-      "grad_norm": 0.01081087812781334,
-      "learning_rate": 0.0001999979768055346,
-      "loss": 46.0,
-      "step": 26493
-    },
-    {
-      "epoch": 2.025651317927251,
-      "grad_norm": 0.0010818091686815023,
-      "learning_rate": 0.00019999797665274028,
-      "loss": 46.0,
-      "step": 26494
-    },
-    {
-      "epoch": 2.025727774910641,
-      "grad_norm": 0.001547107007354498,
-      "learning_rate": 0.0001999979764999402,
-      "loss": 46.0,
-      "step": 26495
-    },
-    {
-      "epoch": 2.0258042318940306,
-      "grad_norm": 0.002483338350430131,
-      "learning_rate": 0.00019999797634713434,
-      "loss": 46.0,
-      "step": 26496
-    },
-    {
-      "epoch": 2.0258806888774203,
-      "grad_norm": 0.0010727066546678543,
-      "learning_rate": 0.00019999797619432274,
-      "loss": 46.0,
-      "step": 26497
-    },
-    {
-      "epoch": 2.02595714586081,
-      "grad_norm": 0.0017662536120042205,
-      "learning_rate": 0.00019999797604150534,
-      "loss": 46.0,
-      "step": 26498
-    },
-    {
-      "epoch": 2.0260336028442,
-      "grad_norm": 0.00038949030567891896,
-      "learning_rate": 0.0001999979758886822,
-      "loss": 46.0,
-      "step": 26499
-    },
-    {
-      "epoch": 2.0261100598275896,
-      "grad_norm": 0.0010763240279629827,
-      "learning_rate": 0.00019999797573585327,
-      "loss": 46.0,
-      "step": 26500
-    },
-    {
-      "epoch": 2.0261865168109794,
-      "grad_norm": 0.0010031070560216904,
-      "learning_rate": 0.00019999797558301857,
-      "loss": 46.0,
-      "step": 26501
-    },
-    {
-      "epoch": 2.026262973794369,
-      "grad_norm": 0.0021999813616275787,
-      "learning_rate": 0.0001999979754301781,
-      "loss": 46.0,
-      "step": 26502
-    },
-    {
-      "epoch": 2.0263394307777585,
-      "grad_norm": 0.0031785566825419664,
-      "learning_rate": 0.00019999797527733187,
-      "loss": 46.0,
-      "step": 26503
-    },
-    {
-      "epoch": 2.0264158877611482,
-      "grad_norm": 0.0025844541378319263,
-      "learning_rate": 0.00019999797512447988,
-      "loss": 46.0,
-      "step": 26504
-    },
-    {
-      "epoch": 2.026492344744538,
-      "grad_norm": 0.005652190186083317,
-      "learning_rate": 0.0001999979749716221,
-      "loss": 46.0,
-      "step": 26505
-    },
-    {
-      "epoch": 2.0265688017279277,
-      "grad_norm": 0.0024238815531134605,
-      "learning_rate": 0.00019999797481875855,
-      "loss": 46.0,
-      "step": 26506
-    },
-    {
-      "epoch": 2.0266452587113175,
-      "grad_norm": 0.0011899323435500264,
-      "learning_rate": 0.00019999797466588922,
-      "loss": 46.0,
-      "step": 26507
-    },
-    {
-      "epoch": 2.0267217156947073,
-      "grad_norm": 0.0012539083836600184,
-      "learning_rate": 0.00019999797451301417,
-      "loss": 46.0,
-      "step": 26508
-    },
-    {
-      "epoch": 2.026798172678097,
-      "grad_norm": 0.0009194174199365079,
-      "learning_rate": 0.0001999979743601333,
-      "loss": 46.0,
-      "step": 26509
-    },
-    {
-      "epoch": 2.026874629661487,
-      "grad_norm": 0.0006174949812702835,
-      "learning_rate": 0.00019999797420724668,
-      "loss": 46.0,
-      "step": 26510
-    },
-    {
-      "epoch": 2.0269510866448766,
-      "grad_norm": 0.0014386842958629131,
-      "learning_rate": 0.0001999979740543543,
-      "loss": 46.0,
-      "step": 26511
-    },
-    {
-      "epoch": 2.0270275436282663,
-      "grad_norm": 0.001159730483777821,
-      "learning_rate": 0.00019999797390145616,
-      "loss": 46.0,
-      "step": 26512
-    },
-    {
-      "epoch": 2.027104000611656,
-      "grad_norm": 0.0007140013622120023,
-      "learning_rate": 0.0001999979737485522,
-      "loss": 46.0,
-      "step": 26513
-    },
-    {
-      "epoch": 2.0271804575950454,
-      "grad_norm": 0.0007913279114291072,
-      "learning_rate": 0.00019999797359564254,
-      "loss": 46.0,
-      "step": 26514
-    },
-    {
-      "epoch": 2.027256914578435,
-      "grad_norm": 0.004414248745888472,
-      "learning_rate": 0.00019999797344272708,
-      "loss": 46.0,
-      "step": 26515
-    },
-    {
-      "epoch": 2.027333371561825,
-      "grad_norm": 0.0015020071296021342,
-      "learning_rate": 0.00019999797328980586,
-      "loss": 46.0,
-      "step": 26516
-    },
-    {
-      "epoch": 2.0274098285452147,
-      "grad_norm": 0.0007777907885611057,
-      "learning_rate": 0.00019999797313687885,
-      "loss": 46.0,
-      "step": 26517
-    },
-    {
-      "epoch": 2.0274862855286044,
-      "grad_norm": 0.0012671292060986161,
-      "learning_rate": 0.00019999797298394606,
-      "loss": 46.0,
-      "step": 26518
-    },
-    {
-      "epoch": 2.027562742511994,
-      "grad_norm": 0.0007127164863049984,
-      "learning_rate": 0.00019999797283100753,
-      "loss": 46.0,
-      "step": 26519
-    },
-    {
-      "epoch": 2.027639199495384,
-      "grad_norm": 0.0019728608895093203,
-      "learning_rate": 0.0001999979726780632,
-      "loss": 46.0,
-      "step": 26520
-    },
-    {
-      "epoch": 2.0277156564787737,
-      "grad_norm": 0.0012132716365158558,
-      "learning_rate": 0.0001999979725251131,
-      "loss": 46.0,
-      "step": 26521
-    },
-    {
-      "epoch": 2.0277921134621635,
-      "grad_norm": 0.0011979705886915326,
-      "learning_rate": 0.00019999797237215728,
-      "loss": 46.0,
-      "step": 26522
-    },
-    {
-      "epoch": 2.0278685704455532,
-      "grad_norm": 0.0006301268003880978,
-      "learning_rate": 0.00019999797221919566,
-      "loss": 46.0,
-      "step": 26523
-    },
-    {
-      "epoch": 2.027945027428943,
-      "grad_norm": 0.0036981524899601936,
-      "learning_rate": 0.00019999797206622825,
-      "loss": 46.0,
-      "step": 26524
-    },
-    {
-      "epoch": 2.0280214844123323,
-      "grad_norm": 0.0004456181777641177,
-      "learning_rate": 0.0001999979719132551,
-      "loss": 46.0,
-      "step": 26525
-    },
-    {
-      "epoch": 2.028097941395722,
-      "grad_norm": 0.0014028712175786495,
-      "learning_rate": 0.00019999797176027619,
-      "loss": 46.0,
-      "step": 26526
-    },
-    {
-      "epoch": 2.028174398379112,
-      "grad_norm": 0.0008825690601952374,
-      "learning_rate": 0.0001999979716072915,
-      "loss": 46.0,
-      "step": 26527
-    },
-    {
-      "epoch": 2.0282508553625016,
-      "grad_norm": 0.0006085058557800949,
-      "learning_rate": 0.00019999797145430102,
-      "loss": 46.0,
-      "step": 26528
-    },
-    {
-      "epoch": 2.0283273123458914,
-      "grad_norm": 0.0014870521845296025,
-      "learning_rate": 0.00019999797130130478,
-      "loss": 46.0,
-      "step": 26529
-    },
-    {
-      "epoch": 2.028403769329281,
-      "grad_norm": 0.0011964990990236402,
-      "learning_rate": 0.0001999979711483028,
-      "loss": 46.0,
-      "step": 26530
-    },
-    {
-      "epoch": 2.028480226312671,
-      "grad_norm": 0.0019460866460576653,
-      "learning_rate": 0.000199997970995295,
-      "loss": 46.0,
-      "step": 26531
-    },
-    {
-      "epoch": 2.0285566832960606,
-      "grad_norm": 0.0005429726443253458,
-      "learning_rate": 0.00019999797084228145,
-      "loss": 46.0,
-      "step": 26532
-    },
-    {
-      "epoch": 2.0286331402794504,
-      "grad_norm": 0.0006899991421960294,
-      "learning_rate": 0.00019999797068926217,
-      "loss": 46.0,
-      "step": 26533
-    },
-    {
-      "epoch": 2.02870959726284,
-      "grad_norm": 0.0005117070395499468,
-      "learning_rate": 0.00019999797053623706,
-      "loss": 46.0,
-      "step": 26534
-    },
-    {
-      "epoch": 2.02878605424623,
-      "grad_norm": 0.0006403716397471726,
-      "learning_rate": 0.00019999797038320623,
-      "loss": 46.0,
-      "step": 26535
-    },
-    {
-      "epoch": 2.0288625112296192,
-      "grad_norm": 0.0005592163070105016,
-      "learning_rate": 0.00019999797023016958,
-      "loss": 46.0,
-      "step": 26536
-    },
-    {
-      "epoch": 2.028938968213009,
-      "grad_norm": 0.001833010814152658,
-      "learning_rate": 0.0001999979700771272,
-      "loss": 46.0,
-      "step": 26537
-    },
-    {
-      "epoch": 2.0290154251963988,
-      "grad_norm": 0.00384948099963367,
-      "learning_rate": 0.00019999796992407903,
-      "loss": 46.0,
-      "step": 26538
-    },
-    {
-      "epoch": 2.0290918821797885,
-      "grad_norm": 0.0007120004738681018,
-      "learning_rate": 0.0001999979697710251,
-      "loss": 46.0,
-      "step": 26539
-    },
-    {
-      "epoch": 2.0291683391631783,
-      "grad_norm": 0.0012704110704362392,
-      "learning_rate": 0.00019999796961796542,
-      "loss": 46.0,
-      "step": 26540
-    },
-    {
-      "epoch": 2.029244796146568,
-      "grad_norm": 0.0012613125145435333,
-      "learning_rate": 0.00019999796946489995,
-      "loss": 46.0,
-      "step": 26541
-    },
-    {
-      "epoch": 2.029321253129958,
-      "grad_norm": 0.00263027916662395,
-      "learning_rate": 0.0001999979693118287,
-      "loss": 46.0,
-      "step": 26542
-    },
-    {
-      "epoch": 2.0293977101133476,
-      "grad_norm": 0.0010949325514957309,
-      "learning_rate": 0.0001999979691587517,
-      "loss": 46.0,
-      "step": 26543
-    },
-    {
-      "epoch": 2.0294741670967373,
-      "grad_norm": 0.0010598624357953668,
-      "learning_rate": 0.00019999796900566893,
-      "loss": 46.0,
-      "step": 26544
-    },
-    {
-      "epoch": 2.029550624080127,
-      "grad_norm": 0.0013497535837814212,
-      "learning_rate": 0.00019999796885258037,
-      "loss": 46.0,
-      "step": 26545
-    },
-    {
-      "epoch": 2.029627081063517,
-      "grad_norm": 0.0016930188285186887,
-      "learning_rate": 0.00019999796869948607,
-      "loss": 46.0,
-      "step": 26546
-    },
-    {
-      "epoch": 2.029703538046906,
-      "grad_norm": 0.0024236850440502167,
-      "learning_rate": 0.00019999796854638596,
-      "loss": 46.0,
-      "step": 26547
-    },
-    {
-      "epoch": 2.029779995030296,
-      "grad_norm": 0.0011239380110055208,
-      "learning_rate": 0.00019999796839328013,
-      "loss": 46.0,
-      "step": 26548
-    },
-    {
-      "epoch": 2.0298564520136857,
-      "grad_norm": 0.000773589825257659,
-      "learning_rate": 0.00019999796824016853,
-      "loss": 46.0,
-      "step": 26549
-    },
-    {
-      "epoch": 2.0299329089970755,
-      "grad_norm": 0.0020657656714320183,
-      "learning_rate": 0.0001999979680870511,
-      "loss": 46.0,
-      "step": 26550
-    },
-    {
-      "epoch": 2.030009365980465,
-      "grad_norm": 0.0006149280816316605,
-      "learning_rate": 0.00019999796793392796,
-      "loss": 46.0,
-      "step": 26551
-    },
-    {
-      "epoch": 2.030085822963855,
-      "grad_norm": 0.0023668883368372917,
-      "learning_rate": 0.000199997967780799,
-      "loss": 46.0,
-      "step": 26552
-    },
-    {
-      "epoch": 2.0301622799472447,
-      "grad_norm": 0.0012319053057581186,
-      "learning_rate": 0.00019999796762766432,
-      "loss": 46.0,
-      "step": 26553
-    },
-    {
-      "epoch": 2.0302387369306345,
-      "grad_norm": 0.0009415640961378813,
-      "learning_rate": 0.00019999796747452385,
-      "loss": 46.0,
-      "step": 26554
-    },
-    {
-      "epoch": 2.0303151939140243,
-      "grad_norm": 0.001029097125865519,
-      "learning_rate": 0.0001999979673213776,
-      "loss": 46.0,
-      "step": 26555
-    },
-    {
-      "epoch": 2.030391650897414,
-      "grad_norm": 0.0007834908901713789,
-      "learning_rate": 0.0001999979671682256,
-      "loss": 46.0,
-      "step": 26556
-    },
-    {
-      "epoch": 2.030468107880804,
-      "grad_norm": 0.0011065511498600245,
-      "learning_rate": 0.00019999796701506784,
-      "loss": 46.0,
-      "step": 26557
-    },
-    {
-      "epoch": 2.030544564864193,
-      "grad_norm": 0.0022991320583969355,
-      "learning_rate": 0.00019999796686190428,
-      "loss": 46.0,
-      "step": 26558
-    },
-    {
-      "epoch": 2.030621021847583,
-      "grad_norm": 0.0019270549528300762,
-      "learning_rate": 0.00019999796670873497,
-      "loss": 46.0,
-      "step": 26559
-    },
-    {
-      "epoch": 2.0306974788309726,
-      "grad_norm": 0.0003749753232114017,
-      "learning_rate": 0.0001999979665555599,
-      "loss": 46.0,
-      "step": 26560
-    },
-    {
-      "epoch": 2.0307739358143624,
-      "grad_norm": 0.0007511851144954562,
-      "learning_rate": 0.00019999796640237904,
-      "loss": 46.0,
-      "step": 26561
-    },
-    {
-      "epoch": 2.030850392797752,
-      "grad_norm": 0.0013289545895531774,
-      "learning_rate": 0.0001999979662491924,
-      "loss": 46.0,
-      "step": 26562
-    },
-    {
-      "epoch": 2.030926849781142,
-      "grad_norm": 0.00038059064536355436,
-      "learning_rate": 0.000199997966096,
-      "loss": 46.0,
-      "step": 26563
-    },
-    {
-      "epoch": 2.0310033067645317,
-      "grad_norm": 0.0013287510955706239,
-      "learning_rate": 0.00019999796594280184,
-      "loss": 46.0,
-      "step": 26564
-    },
-    {
-      "epoch": 2.0310797637479214,
-      "grad_norm": 0.000757434987463057,
-      "learning_rate": 0.00019999796578959794,
-      "loss": 46.0,
-      "step": 26565
-    },
-    {
-      "epoch": 2.031156220731311,
-      "grad_norm": 0.0007932082517072558,
-      "learning_rate": 0.00019999796563638822,
-      "loss": 46.0,
-      "step": 26566
-    },
-    {
-      "epoch": 2.031232677714701,
-      "grad_norm": 0.0010606384603306651,
-      "learning_rate": 0.00019999796548317273,
-      "loss": 46.0,
-      "step": 26567
-    },
-    {
-      "epoch": 2.0313091346980903,
-      "grad_norm": 0.0013175703352317214,
-      "learning_rate": 0.00019999796532995152,
-      "loss": 46.0,
-      "step": 26568
-    },
-    {
-      "epoch": 2.03138559168148,
-      "grad_norm": 0.001010414445772767,
-      "learning_rate": 0.00019999796517672448,
-      "loss": 46.0,
-      "step": 26569
-    },
-    {
-      "epoch": 2.03146204866487,
-      "grad_norm": 0.0046905288472771645,
-      "learning_rate": 0.00019999796502349172,
-      "loss": 46.0,
-      "step": 26570
-    },
-    {
-      "epoch": 2.0315385056482596,
-      "grad_norm": 0.0006144028156995773,
-      "learning_rate": 0.00019999796487025318,
-      "loss": 46.0,
-      "step": 26571
-    },
-    {
-      "epoch": 2.0316149626316493,
-      "grad_norm": 0.0019732704386115074,
-      "learning_rate": 0.00019999796471700885,
-      "loss": 46.0,
-      "step": 26572
-    },
-    {
-      "epoch": 2.031691419615039,
-      "grad_norm": 0.001210829708725214,
-      "learning_rate": 0.00019999796456375877,
-      "loss": 46.0,
-      "step": 26573
-    },
-    {
-      "epoch": 2.031767876598429,
-      "grad_norm": 0.0018193984869867563,
-      "learning_rate": 0.00019999796441050292,
-      "loss": 46.0,
-      "step": 26574
-    },
-    {
-      "epoch": 2.0318443335818186,
-      "grad_norm": 0.0006971064722165465,
-      "learning_rate": 0.0001999979642572413,
-      "loss": 46.0,
-      "step": 26575
-    },
-    {
-      "epoch": 2.0319207905652084,
-      "grad_norm": 0.00032437010668218136,
-      "learning_rate": 0.0001999979641039739,
-      "loss": 46.0,
-      "step": 26576
-    },
-    {
-      "epoch": 2.031997247548598,
-      "grad_norm": 0.001501981751061976,
-      "learning_rate": 0.00019999796395070072,
-      "loss": 46.0,
-      "step": 26577
-    },
-    {
-      "epoch": 2.032073704531988,
-      "grad_norm": 0.00476000877097249,
-      "learning_rate": 0.0001999979637974218,
-      "loss": 46.0,
-      "step": 26578
-    },
-    {
-      "epoch": 2.032150161515377,
-      "grad_norm": 0.001412849291227758,
-      "learning_rate": 0.0001999979636441371,
-      "loss": 46.0,
-      "step": 26579
-    },
-    {
-      "epoch": 2.032226618498767,
-      "grad_norm": 0.0006279490771703422,
-      "learning_rate": 0.0001999979634908466,
-      "loss": 46.0,
-      "step": 26580
-    },
-    {
-      "epoch": 2.0323030754821567,
-      "grad_norm": 0.00045324102393351495,
-      "learning_rate": 0.0001999979633375504,
-      "loss": 46.0,
-      "step": 26581
-    },
-    {
-      "epoch": 2.0323795324655465,
-      "grad_norm": 0.0009774421341717243,
-      "learning_rate": 0.00019999796318424836,
-      "loss": 46.0,
-      "step": 26582
-    },
-    {
-      "epoch": 2.0324559894489362,
-      "grad_norm": 0.00042095532990060747,
-      "learning_rate": 0.00019999796303094057,
-      "loss": 46.0,
-      "step": 26583
-    },
-    {
-      "epoch": 2.032532446432326,
-      "grad_norm": 0.0012581833871081471,
-      "learning_rate": 0.00019999796287762704,
-      "loss": 46.0,
-      "step": 26584
-    },
-    {
-      "epoch": 2.0326089034157158,
-      "grad_norm": 0.0007525114924646914,
-      "learning_rate": 0.00019999796272430773,
-      "loss": 46.0,
-      "step": 26585
-    },
-    {
-      "epoch": 2.0326853603991055,
-      "grad_norm": 0.0008561808499507606,
-      "learning_rate": 0.00019999796257098265,
-      "loss": 46.0,
-      "step": 26586
-    },
-    {
-      "epoch": 2.0327618173824953,
-      "grad_norm": 0.0007320060394704342,
-      "learning_rate": 0.00019999796241765177,
-      "loss": 46.0,
-      "step": 26587
-    },
-    {
-      "epoch": 2.032838274365885,
-      "grad_norm": 0.0029086056165397167,
-      "learning_rate": 0.00019999796226431515,
-      "loss": 46.0,
-      "step": 26588
-    },
-    {
-      "epoch": 2.032914731349275,
-      "grad_norm": 0.002113952534273267,
-      "learning_rate": 0.00019999796211097275,
-      "loss": 46.0,
-      "step": 26589
-    },
-    {
-      "epoch": 2.032991188332664,
-      "grad_norm": 0.000662376347463578,
-      "learning_rate": 0.0001999979619576246,
-      "loss": 46.0,
-      "step": 26590
-    },
-    {
-      "epoch": 2.033067645316054,
-      "grad_norm": 0.00044667781912721694,
-      "learning_rate": 0.00019999796180427066,
-      "loss": 46.0,
-      "step": 26591
-    },
-    {
-      "epoch": 2.0331441022994436,
-      "grad_norm": 0.0007176919025368989,
-      "learning_rate": 0.00019999796165091094,
-      "loss": 46.0,
-      "step": 26592
-    },
-    {
-      "epoch": 2.0332205592828334,
-      "grad_norm": 0.00146351323928684,
-      "learning_rate": 0.00019999796149754547,
-      "loss": 46.0,
-      "step": 26593
-    },
-    {
-      "epoch": 2.033297016266223,
-      "grad_norm": 0.0025684009306132793,
-      "learning_rate": 0.00019999796134417423,
-      "loss": 46.0,
-      "step": 26594
-    },
-    {
-      "epoch": 2.033373473249613,
-      "grad_norm": 0.00047959969379007816,
-      "learning_rate": 0.00019999796119079722,
-      "loss": 46.0,
-      "step": 26595
-    },
-    {
-      "epoch": 2.0334499302330027,
-      "grad_norm": 0.0003634125750977546,
-      "learning_rate": 0.00019999796103741443,
-      "loss": 46.0,
-      "step": 26596
-    },
-    {
-      "epoch": 2.0335263872163925,
-      "grad_norm": 0.0009037366835400462,
-      "learning_rate": 0.0001999979608840259,
-      "loss": 46.0,
-      "step": 26597
-    },
-    {
-      "epoch": 2.033602844199782,
-      "grad_norm": 0.0010114137548953295,
-      "learning_rate": 0.00019999796073063157,
-      "loss": 46.0,
-      "step": 26598
-    },
-    {
-      "epoch": 2.033679301183172,
-      "grad_norm": 0.001232185517437756,
-      "learning_rate": 0.0001999979605772315,
-      "loss": 46.0,
-      "step": 26599
-    },
-    {
-      "epoch": 2.0337557581665617,
-      "grad_norm": 0.008454413153231144,
-      "learning_rate": 0.0001999979604238256,
-      "loss": 46.0,
-      "step": 26600
-    },
-    {
-      "epoch": 2.033832215149951,
-      "grad_norm": 0.0003661074733827263,
-      "learning_rate": 0.000199997960270414,
-      "loss": 46.0,
-      "step": 26601
-    },
-    {
-      "epoch": 2.033908672133341,
-      "grad_norm": 0.0030759908258914948,
-      "learning_rate": 0.0001999979601169966,
-      "loss": 46.0,
-      "step": 26602
-    },
-    {
-      "epoch": 2.0339851291167306,
-      "grad_norm": 0.0008564048912376165,
-      "learning_rate": 0.0001999979599635734,
-      "loss": 46.0,
-      "step": 26603
-    },
-    {
-      "epoch": 2.0340615861001203,
-      "grad_norm": 0.009459891356527805,
-      "learning_rate": 0.00019999795981014452,
-      "loss": 46.0,
-      "step": 26604
-    },
-    {
-      "epoch": 2.03413804308351,
-      "grad_norm": 0.001192210242152214,
-      "learning_rate": 0.00019999795965670977,
-      "loss": 46.0,
-      "step": 26605
-    },
-    {
-      "epoch": 2.0342145000669,
-      "grad_norm": 0.001824631355702877,
-      "learning_rate": 0.0001999979595032693,
-      "loss": 46.0,
-      "step": 26606
-    },
-    {
-      "epoch": 2.0342909570502896,
-      "grad_norm": 0.0004505351826082915,
-      "learning_rate": 0.0001999979593498231,
-      "loss": 46.0,
-      "step": 26607
-    },
-    {
-      "epoch": 2.0343674140336794,
-      "grad_norm": 0.0012984368950128555,
-      "learning_rate": 0.00019999795919637106,
-      "loss": 46.0,
-      "step": 26608
-    },
-    {
-      "epoch": 2.034443871017069,
-      "grad_norm": 0.0022749153431504965,
-      "learning_rate": 0.00019999795904291327,
-      "loss": 46.0,
-      "step": 26609
-    },
-    {
-      "epoch": 2.034520328000459,
-      "grad_norm": 0.0014079343527555466,
-      "learning_rate": 0.0001999979588894497,
-      "loss": 46.0,
-      "step": 26610
-    },
-    {
-      "epoch": 2.0345967849838487,
-      "grad_norm": 0.00256090285256505,
-      "learning_rate": 0.0001999979587359804,
-      "loss": 46.0,
-      "step": 26611
-    },
-    {
-      "epoch": 2.034673241967238,
-      "grad_norm": 0.00040165524114854634,
-      "learning_rate": 0.00019999795858250533,
-      "loss": 46.0,
-      "step": 26612
-    },
-    {
-      "epoch": 2.0347496989506277,
-      "grad_norm": 0.002713286317884922,
-      "learning_rate": 0.00019999795842902445,
-      "loss": 46.0,
-      "step": 26613
-    },
-    {
-      "epoch": 2.0348261559340175,
-      "grad_norm": 0.0014820380602031946,
-      "learning_rate": 0.00019999795827553782,
-      "loss": 46.0,
-      "step": 26614
-    },
-    {
-      "epoch": 2.0349026129174073,
-      "grad_norm": 0.00043587872642092407,
-      "learning_rate": 0.00019999795812204542,
-      "loss": 46.0,
-      "step": 26615
-    },
-    {
-      "epoch": 2.034979069900797,
-      "grad_norm": 0.0004651909985113889,
-      "learning_rate": 0.00019999795796854725,
-      "loss": 46.0,
-      "step": 26616
-    },
-    {
-      "epoch": 2.035055526884187,
-      "grad_norm": 0.002295827027410269,
-      "learning_rate": 0.0001999979578150433,
-      "loss": 46.0,
-      "step": 26617
-    },
-    {
-      "epoch": 2.0351319838675765,
-      "grad_norm": 0.0010976034682244062,
-      "learning_rate": 0.00019999795766153361,
-      "loss": 46.0,
-      "step": 26618
-    },
-    {
-      "epoch": 2.0352084408509663,
-      "grad_norm": 0.002214741660282016,
-      "learning_rate": 0.00019999795750801815,
-      "loss": 46.0,
-      "step": 26619
-    },
-    {
-      "epoch": 2.035284897834356,
-      "grad_norm": 0.0007417284068651497,
-      "learning_rate": 0.00019999795735449688,
-      "loss": 46.0,
-      "step": 26620
-    },
-    {
-      "epoch": 2.035361354817746,
-      "grad_norm": 0.0004341478052083403,
-      "learning_rate": 0.00019999795720096987,
-      "loss": 46.0,
-      "step": 26621
-    },
-    {
-      "epoch": 2.0354378118011356,
-      "grad_norm": 0.0005533367511816323,
-      "learning_rate": 0.00019999795704743708,
-      "loss": 46.0,
-      "step": 26622
-    },
-    {
-      "epoch": 2.035514268784525,
-      "grad_norm": 0.0007011029520072043,
-      "learning_rate": 0.00019999795689389855,
-      "loss": 46.0,
-      "step": 26623
-    },
-    {
-      "epoch": 2.0355907257679147,
-      "grad_norm": 0.0004201651900075376,
-      "learning_rate": 0.00019999795674035422,
-      "loss": 46.0,
-      "step": 26624
-    },
-    {
-      "epoch": 2.0356671827513044,
-      "grad_norm": 0.0004260960849933326,
-      "learning_rate": 0.00019999795658680412,
-      "loss": 46.0,
-      "step": 26625
-    },
-    {
-      "epoch": 2.035743639734694,
-      "grad_norm": 0.0026603899896144867,
-      "learning_rate": 0.0001999979564332483,
-      "loss": 46.0,
-      "step": 26626
-    },
-    {
-      "epoch": 2.035820096718084,
-      "grad_norm": 0.0056581273674964905,
-      "learning_rate": 0.00019999795627968664,
-      "loss": 46.0,
-      "step": 26627
-    },
-    {
-      "epoch": 2.0358965537014737,
-      "grad_norm": 0.0002495713997632265,
-      "learning_rate": 0.00019999795612611924,
-      "loss": 46.0,
-      "step": 26628
-    },
-    {
-      "epoch": 2.0359730106848635,
-      "grad_norm": 0.00373982940800488,
-      "learning_rate": 0.00019999795597254607,
-      "loss": 46.0,
-      "step": 26629
-    },
-    {
-      "epoch": 2.0360494676682532,
-      "grad_norm": 0.0006167826941236854,
-      "learning_rate": 0.00019999795581896712,
-      "loss": 46.0,
-      "step": 26630
-    },
-    {
-      "epoch": 2.036125924651643,
-      "grad_norm": 0.0005011427565477788,
-      "learning_rate": 0.00019999795566538243,
-      "loss": 46.0,
-      "step": 26631
-    },
-    {
-      "epoch": 2.0362023816350328,
-      "grad_norm": 0.0005400745430961251,
-      "learning_rate": 0.00019999795551179194,
-      "loss": 46.0,
-      "step": 26632
-    },
-    {
-      "epoch": 2.0362788386184225,
-      "grad_norm": 0.0008155073737725616,
-      "learning_rate": 0.0001999979553581957,
-      "loss": 46.0,
-      "step": 26633
-    },
-    {
-      "epoch": 2.036355295601812,
-      "grad_norm": 0.000484047137433663,
-      "learning_rate": 0.0001999979552045937,
-      "loss": 46.0,
-      "step": 26634
-    },
-    {
-      "epoch": 2.0364317525852016,
-      "grad_norm": 0.0004688260378316045,
-      "learning_rate": 0.00019999795505098588,
-      "loss": 46.0,
-      "step": 26635
-    },
-    {
-      "epoch": 2.0365082095685914,
-      "grad_norm": 0.0009017328848131001,
-      "learning_rate": 0.00019999795489737234,
-      "loss": 46.0,
-      "step": 26636
-    },
-    {
-      "epoch": 2.036584666551981,
-      "grad_norm": 0.001145393936894834,
-      "learning_rate": 0.000199997954743753,
-      "loss": 46.0,
-      "step": 26637
-    },
-    {
-      "epoch": 2.036661123535371,
-      "grad_norm": 0.0014703063061460853,
-      "learning_rate": 0.0001999979545901279,
-      "loss": 46.0,
-      "step": 26638
-    },
-    {
-      "epoch": 2.0367375805187606,
-      "grad_norm": 0.0009786536684259772,
-      "learning_rate": 0.00019999795443649708,
-      "loss": 46.0,
-      "step": 26639
-    },
-    {
-      "epoch": 2.0368140375021504,
-      "grad_norm": 0.0030558928847312927,
-      "learning_rate": 0.0001999979542828604,
-      "loss": 46.0,
-      "step": 26640
-    },
-    {
-      "epoch": 2.03689049448554,
-      "grad_norm": 0.0013135335175320506,
-      "learning_rate": 0.00019999795412921803,
-      "loss": 46.0,
-      "step": 26641
-    },
-    {
-      "epoch": 2.03696695146893,
-      "grad_norm": 0.0017956553492695093,
-      "learning_rate": 0.00019999795397556986,
-      "loss": 46.0,
-      "step": 26642
-    },
-    {
-      "epoch": 2.0370434084523197,
-      "grad_norm": 0.0018976174760609865,
-      "learning_rate": 0.00019999795382191592,
-      "loss": 46.0,
-      "step": 26643
-    },
-    {
-      "epoch": 2.0371198654357094,
-      "grad_norm": 0.0019661334808915854,
-      "learning_rate": 0.0001999979536682562,
-      "loss": 46.0,
-      "step": 26644
-    },
-    {
-      "epoch": 2.0371963224190988,
-      "grad_norm": 0.0005052143824286759,
-      "learning_rate": 0.00019999795351459073,
-      "loss": 46.0,
-      "step": 26645
-    },
-    {
-      "epoch": 2.0372727794024885,
-      "grad_norm": 0.00033888386678881943,
-      "learning_rate": 0.0001999979533609195,
-      "loss": 46.0,
-      "step": 26646
-    },
-    {
-      "epoch": 2.0373492363858783,
-      "grad_norm": 0.006462957710027695,
-      "learning_rate": 0.00019999795320724246,
-      "loss": 46.0,
-      "step": 26647
-    },
-    {
-      "epoch": 2.037425693369268,
-      "grad_norm": 0.001122289919294417,
-      "learning_rate": 0.00019999795305355967,
-      "loss": 46.0,
-      "step": 26648
-    },
-    {
-      "epoch": 2.037502150352658,
-      "grad_norm": 0.0006262047681957483,
-      "learning_rate": 0.00019999795289987114,
-      "loss": 46.0,
-      "step": 26649
-    },
-    {
-      "epoch": 2.0375786073360476,
-      "grad_norm": 0.005542945582419634,
-      "learning_rate": 0.00019999795274617678,
-      "loss": 46.0,
-      "step": 26650
-    },
-    {
-      "epoch": 2.0376550643194373,
-      "grad_norm": 0.005652573890984058,
-      "learning_rate": 0.0001999979525924767,
-      "loss": 46.0,
-      "step": 26651
-    },
-    {
-      "epoch": 2.037731521302827,
-      "grad_norm": 0.004041920881718397,
-      "learning_rate": 0.00019999795243877083,
-      "loss": 46.0,
-      "step": 26652
-    },
-    {
-      "epoch": 2.037807978286217,
-      "grad_norm": 0.0023524577263742685,
-      "learning_rate": 0.0001999979522850592,
-      "loss": 46.0,
-      "step": 26653
-    },
-    {
-      "epoch": 2.0378844352696066,
-      "grad_norm": 0.004859734792262316,
-      "learning_rate": 0.0001999979521313418,
-      "loss": 46.0,
-      "step": 26654
-    },
-    {
-      "epoch": 2.0379608922529964,
-      "grad_norm": 0.0008747361716814339,
-      "learning_rate": 0.00019999795197761863,
-      "loss": 46.0,
-      "step": 26655
-    },
-    {
-      "epoch": 2.0380373492363857,
-      "grad_norm": 0.0009978747693821788,
-      "learning_rate": 0.0001999979518238897,
-      "loss": 46.0,
-      "step": 26656
-    },
-    {
-      "epoch": 2.0381138062197754,
-      "grad_norm": 0.0006871107034385204,
-      "learning_rate": 0.00019999795167015497,
-      "loss": 46.0,
-      "step": 26657
-    },
-    {
-      "epoch": 2.038190263203165,
-      "grad_norm": 0.0004938003839924932,
-      "learning_rate": 0.0001999979515164145,
-      "loss": 46.0,
-      "step": 26658
-    },
-    {
-      "epoch": 2.038266720186555,
-      "grad_norm": 0.0009424231830053031,
-      "learning_rate": 0.00019999795136266824,
-      "loss": 46.0,
-      "step": 26659
-    },
-    {
-      "epoch": 2.0383431771699447,
-      "grad_norm": 0.0007297934498637915,
-      "learning_rate": 0.00019999795120891623,
-      "loss": 46.0,
-      "step": 26660
-    },
-    {
-      "epoch": 2.0384196341533345,
-      "grad_norm": 0.0016007061349228024,
-      "learning_rate": 0.00019999795105515844,
-      "loss": 46.0,
-      "step": 26661
-    },
-    {
-      "epoch": 2.0384960911367243,
-      "grad_norm": 0.0010401232866570354,
-      "learning_rate": 0.0001999979509013949,
-      "loss": 46.0,
-      "step": 26662
-    },
-    {
-      "epoch": 2.038572548120114,
-      "grad_norm": 0.0026519561652094126,
-      "learning_rate": 0.00019999795074762556,
-      "loss": 46.0,
-      "step": 26663
-    },
-    {
-      "epoch": 2.0386490051035038,
-      "grad_norm": 0.001126496703363955,
-      "learning_rate": 0.00019999795059385048,
-      "loss": 46.0,
-      "step": 26664
-    },
-    {
-      "epoch": 2.0387254620868935,
-      "grad_norm": 0.001208712114021182,
-      "learning_rate": 0.0001999979504400696,
-      "loss": 46.0,
-      "step": 26665
-    },
-    {
-      "epoch": 2.0388019190702833,
-      "grad_norm": 0.0011292635463178158,
-      "learning_rate": 0.00019999795028628295,
-      "loss": 46.0,
-      "step": 26666
-    },
-    {
-      "epoch": 2.0388783760536726,
-      "grad_norm": 0.0005419857916422188,
-      "learning_rate": 0.00019999795013249055,
-      "loss": 46.0,
-      "step": 26667
-    },
-    {
-      "epoch": 2.0389548330370624,
-      "grad_norm": 0.0007096214685589075,
-      "learning_rate": 0.00019999794997869238,
-      "loss": 46.0,
-      "step": 26668
-    },
-    {
-      "epoch": 2.039031290020452,
-      "grad_norm": 0.00209606159478426,
-      "learning_rate": 0.0001999979498248884,
-      "loss": 46.0,
-      "step": 26669
-    },
-    {
-      "epoch": 2.039107747003842,
-      "grad_norm": 0.0011689257808029652,
-      "learning_rate": 0.00019999794967107872,
-      "loss": 46.0,
-      "step": 26670
-    },
-    {
-      "epoch": 2.0391842039872317,
-      "grad_norm": 0.0006484262412413955,
-      "learning_rate": 0.00019999794951726325,
-      "loss": 46.0,
-      "step": 26671
-    },
-    {
-      "epoch": 2.0392606609706214,
-      "grad_norm": 0.0007338208379223943,
-      "learning_rate": 0.000199997949363442,
-      "loss": 46.0,
-      "step": 26672
-    },
-    {
-      "epoch": 2.039337117954011,
-      "grad_norm": 0.004225425887852907,
-      "learning_rate": 0.00019999794920961498,
-      "loss": 46.0,
-      "step": 26673
-    },
-    {
-      "epoch": 2.039413574937401,
-      "grad_norm": 0.0009118360467255116,
-      "learning_rate": 0.00019999794905578217,
-      "loss": 46.0,
-      "step": 26674
-    },
-    {
-      "epoch": 2.0394900319207907,
-      "grad_norm": 0.0022862653713673353,
-      "learning_rate": 0.00019999794890194364,
-      "loss": 46.0,
-      "step": 26675
-    },
-    {
-      "epoch": 2.0395664889041805,
-      "grad_norm": 0.000911420735064894,
-      "learning_rate": 0.0001999979487480993,
-      "loss": 46.0,
-      "step": 26676
-    },
-    {
-      "epoch": 2.03964294588757,
-      "grad_norm": 0.0006587571115233004,
-      "learning_rate": 0.00019999794859424917,
-      "loss": 46.0,
-      "step": 26677
-    },
-    {
-      "epoch": 2.0397194028709595,
-      "grad_norm": 0.0024143436457961798,
-      "learning_rate": 0.00019999794844039332,
-      "loss": 46.0,
-      "step": 26678
-    },
-    {
-      "epoch": 2.0397958598543493,
-      "grad_norm": 0.0011215313570573926,
-      "learning_rate": 0.0001999979482865317,
-      "loss": 46.0,
-      "step": 26679
-    },
-    {
-      "epoch": 2.039872316837739,
-      "grad_norm": 0.002950820606201887,
-      "learning_rate": 0.00019999794813266428,
-      "loss": 46.0,
-      "step": 26680
-    },
-    {
-      "epoch": 2.039948773821129,
-      "grad_norm": 0.0012147912057116628,
-      "learning_rate": 0.0001999979479787911,
-      "loss": 46.0,
-      "step": 26681
-    },
-    {
-      "epoch": 2.0400252308045186,
-      "grad_norm": 0.0009096599533222616,
-      "learning_rate": 0.00019999794782491216,
-      "loss": 46.0,
-      "step": 26682
-    },
-    {
-      "epoch": 2.0401016877879083,
-      "grad_norm": 0.004348806571215391,
-      "learning_rate": 0.00019999794767102747,
-      "loss": 46.0,
-      "step": 26683
-    },
-    {
-      "epoch": 2.040178144771298,
-      "grad_norm": 0.0007725459872744977,
-      "learning_rate": 0.00019999794751713695,
-      "loss": 46.0,
-      "step": 26684
-    },
-    {
-      "epoch": 2.040254601754688,
-      "grad_norm": 0.0008875320781953633,
-      "learning_rate": 0.00019999794736324074,
-      "loss": 46.0,
-      "step": 26685
-    },
-    {
-      "epoch": 2.0403310587380776,
-      "grad_norm": 0.0004521760274656117,
-      "learning_rate": 0.00019999794720933868,
-      "loss": 46.0,
-      "step": 26686
-    },
-    {
-      "epoch": 2.0404075157214674,
-      "grad_norm": 0.0004650182672776282,
-      "learning_rate": 0.00019999794705543092,
-      "loss": 46.0,
-      "step": 26687
-    },
-    {
-      "epoch": 2.040483972704857,
-      "grad_norm": 0.003060281975194812,
-      "learning_rate": 0.00019999794690151736,
-      "loss": 46.0,
-      "step": 26688
-    },
-    {
-      "epoch": 2.0405604296882465,
-      "grad_norm": 0.0014394642785191536,
-      "learning_rate": 0.00019999794674759803,
-      "loss": 46.0,
-      "step": 26689
-    },
-    {
-      "epoch": 2.0406368866716362,
-      "grad_norm": 0.0010106581030413508,
-      "learning_rate": 0.00019999794659367293,
-      "loss": 46.0,
-      "step": 26690
-    },
-    {
-      "epoch": 2.040713343655026,
-      "grad_norm": 0.0006022258894518018,
-      "learning_rate": 0.00019999794643974205,
-      "loss": 46.0,
-      "step": 26691
-    },
-    {
-      "epoch": 2.0407898006384158,
-      "grad_norm": 0.002769224112853408,
-      "learning_rate": 0.00019999794628580543,
-      "loss": 46.0,
-      "step": 26692
-    },
-    {
-      "epoch": 2.0408662576218055,
-      "grad_norm": 0.0011104485020041466,
-      "learning_rate": 0.00019999794613186303,
-      "loss": 46.0,
-      "step": 26693
-    },
-    {
-      "epoch": 2.0409427146051953,
-      "grad_norm": 0.0010501760989427567,
-      "learning_rate": 0.00019999794597791483,
-      "loss": 46.0,
-      "step": 26694
-    },
-    {
-      "epoch": 2.041019171588585,
-      "grad_norm": 0.0013500908389687538,
-      "learning_rate": 0.00019999794582396086,
-      "loss": 46.0,
-      "step": 26695
-    },
-    {
-      "epoch": 2.041095628571975,
-      "grad_norm": 0.013040841557085514,
-      "learning_rate": 0.00019999794567000117,
-      "loss": 46.0,
-      "step": 26696
-    },
-    {
-      "epoch": 2.0411720855553646,
-      "grad_norm": 0.003713401732966304,
-      "learning_rate": 0.00019999794551603568,
-      "loss": 46.0,
-      "step": 26697
-    },
-    {
-      "epoch": 2.0412485425387543,
-      "grad_norm": 0.0008403608808293939,
-      "learning_rate": 0.00019999794536206444,
-      "loss": 46.0,
-      "step": 26698
-    },
-    {
-      "epoch": 2.0413249995221436,
-      "grad_norm": 0.0003959627647418529,
-      "learning_rate": 0.0001999979452080874,
-      "loss": 46.0,
-      "step": 26699
-    },
-    {
-      "epoch": 2.0414014565055334,
-      "grad_norm": 0.0018761565443128347,
-      "learning_rate": 0.0001999979450541046,
-      "loss": 46.0,
-      "step": 26700
-    },
-    {
-      "epoch": 2.041477913488923,
-      "grad_norm": 0.0012406965252012014,
-      "learning_rate": 0.00019999794490011604,
-      "loss": 46.0,
-      "step": 26701
-    },
-    {
-      "epoch": 2.041554370472313,
-      "grad_norm": 0.0005860960809513927,
-      "learning_rate": 0.00019999794474612174,
-      "loss": 46.0,
-      "step": 26702
-    },
-    {
-      "epoch": 2.0416308274557027,
-      "grad_norm": 0.0024115475825965405,
-      "learning_rate": 0.00019999794459212163,
-      "loss": 46.0,
-      "step": 26703
-    },
-    {
-      "epoch": 2.0417072844390924,
-      "grad_norm": 0.0008953747455962002,
-      "learning_rate": 0.00019999794443811576,
-      "loss": 46.0,
-      "step": 26704
-    },
-    {
-      "epoch": 2.041783741422482,
-      "grad_norm": 0.0035076530184596777,
-      "learning_rate": 0.00019999794428410413,
-      "loss": 46.0,
-      "step": 26705
-    },
-    {
-      "epoch": 2.041860198405872,
-      "grad_norm": 0.0011120779672637582,
-      "learning_rate": 0.00019999794413008674,
-      "loss": 46.0,
-      "step": 26706
-    },
-    {
-      "epoch": 2.0419366553892617,
-      "grad_norm": 0.009908235631883144,
-      "learning_rate": 0.00019999794397606354,
-      "loss": 46.0,
-      "step": 26707
-    },
-    {
-      "epoch": 2.0420131123726515,
-      "grad_norm": 0.004595461301505566,
-      "learning_rate": 0.0001999979438220346,
-      "loss": 46.0,
-      "step": 26708
-    },
-    {
-      "epoch": 2.0420895693560412,
-      "grad_norm": 0.005109483376145363,
-      "learning_rate": 0.00019999794366799988,
-      "loss": 46.0,
-      "step": 26709
-    },
-    {
-      "epoch": 2.0421660263394306,
-      "grad_norm": 0.0023741349577903748,
-      "learning_rate": 0.0001999979435139594,
-      "loss": 46.0,
-      "step": 26710
-    },
-    {
-      "epoch": 2.0422424833228203,
-      "grad_norm": 0.0020227041095495224,
-      "learning_rate": 0.00019999794335991312,
-      "loss": 46.0,
-      "step": 26711
-    },
-    {
-      "epoch": 2.04231894030621,
-      "grad_norm": 0.0009598316974006593,
-      "learning_rate": 0.0001999979432058611,
-      "loss": 46.0,
-      "step": 26712
-    },
-    {
-      "epoch": 2.0423953972896,
-      "grad_norm": 0.0011091948254033923,
-      "learning_rate": 0.0001999979430518033,
-      "loss": 46.0,
-      "step": 26713
-    },
-    {
-      "epoch": 2.0424718542729896,
-      "grad_norm": 0.0007525385008193552,
-      "learning_rate": 0.00019999794289773975,
-      "loss": 46.0,
-      "step": 26714
-    },
-    {
-      "epoch": 2.0425483112563794,
-      "grad_norm": 0.0006028954521752894,
-      "learning_rate": 0.00019999794274367042,
-      "loss": 46.0,
-      "step": 26715
-    },
-    {
-      "epoch": 2.042624768239769,
-      "grad_norm": 0.0011807328555732965,
-      "learning_rate": 0.0001999979425895953,
-      "loss": 46.0,
-      "step": 26716
-    },
-    {
-      "epoch": 2.042701225223159,
-      "grad_norm": 0.0008310778066515923,
-      "learning_rate": 0.00019999794243551444,
-      "loss": 46.0,
-      "step": 26717
-    },
-    {
-      "epoch": 2.0427776822065487,
-      "grad_norm": 0.000946046260651201,
-      "learning_rate": 0.00019999794228142781,
-      "loss": 46.0,
-      "step": 26718
-    },
-    {
-      "epoch": 2.0428541391899384,
-      "grad_norm": 0.0010109165450558066,
-      "learning_rate": 0.0001999979421273354,
-      "loss": 46.0,
-      "step": 26719
-    },
-    {
-      "epoch": 2.042930596173328,
-      "grad_norm": 0.001046556979417801,
-      "learning_rate": 0.00019999794197323722,
-      "loss": 46.0,
-      "step": 26720
-    },
-    {
-      "epoch": 2.0430070531567175,
-      "grad_norm": 0.002121644327417016,
-      "learning_rate": 0.00019999794181913328,
-      "loss": 46.0,
-      "step": 26721
-    },
-    {
-      "epoch": 2.0430835101401073,
-      "grad_norm": 0.0036012011114507914,
-      "learning_rate": 0.00019999794166502353,
-      "loss": 46.0,
-      "step": 26722
-    },
-    {
-      "epoch": 2.043159967123497,
-      "grad_norm": 0.0028198163490742445,
-      "learning_rate": 0.00019999794151090807,
-      "loss": 46.0,
-      "step": 26723
-    },
-    {
-      "epoch": 2.0432364241068868,
-      "grad_norm": 0.0006680210353806615,
-      "learning_rate": 0.00019999794135678684,
-      "loss": 46.0,
-      "step": 26724
-    },
-    {
-      "epoch": 2.0433128810902765,
-      "grad_norm": 0.0026276058051735163,
-      "learning_rate": 0.00019999794120265977,
-      "loss": 46.0,
-      "step": 26725
-    },
-    {
-      "epoch": 2.0433893380736663,
-      "grad_norm": 0.0033334163017570972,
-      "learning_rate": 0.000199997941048527,
-      "loss": 46.0,
-      "step": 26726
-    },
-    {
-      "epoch": 2.043465795057056,
-      "grad_norm": 0.00047325206105597317,
-      "learning_rate": 0.00019999794089438843,
-      "loss": 46.0,
-      "step": 26727
-    },
-    {
-      "epoch": 2.043542252040446,
-      "grad_norm": 0.0005689062527380884,
-      "learning_rate": 0.0001999979407402441,
-      "loss": 46.0,
-      "step": 26728
-    },
-    {
-      "epoch": 2.0436187090238356,
-      "grad_norm": 0.0008415696211159229,
-      "learning_rate": 0.00019999794058609397,
-      "loss": 46.0,
-      "step": 26729
-    },
-    {
-      "epoch": 2.0436951660072253,
-      "grad_norm": 0.0009475558181293309,
-      "learning_rate": 0.00019999794043193812,
-      "loss": 46.0,
-      "step": 26730
-    },
-    {
-      "epoch": 2.043771622990615,
-      "grad_norm": 0.000825090566650033,
-      "learning_rate": 0.00019999794027777647,
-      "loss": 46.0,
-      "step": 26731
-    },
-    {
-      "epoch": 2.0438480799740044,
-      "grad_norm": 0.005847132299095392,
-      "learning_rate": 0.00019999794012360905,
-      "loss": 46.0,
-      "step": 26732
-    },
-    {
-      "epoch": 2.043924536957394,
-      "grad_norm": 0.004025956150144339,
-      "learning_rate": 0.00019999793996943588,
-      "loss": 46.0,
-      "step": 26733
-    },
-    {
-      "epoch": 2.044000993940784,
-      "grad_norm": 0.000576301827095449,
-      "learning_rate": 0.00019999793981525694,
-      "loss": 46.0,
-      "step": 26734
-    },
-    {
-      "epoch": 2.0440774509241737,
-      "grad_norm": 0.0006756465299986303,
-      "learning_rate": 0.00019999793966107222,
-      "loss": 46.0,
-      "step": 26735
-    },
-    {
-      "epoch": 2.0441539079075635,
-      "grad_norm": 0.0028010420501232147,
-      "learning_rate": 0.00019999793950688173,
-      "loss": 46.0,
-      "step": 26736
-    },
-    {
-      "epoch": 2.0442303648909532,
-      "grad_norm": 0.001140445121563971,
-      "learning_rate": 0.00019999793935268547,
-      "loss": 46.0,
-      "step": 26737
-    },
-    {
-      "epoch": 2.044306821874343,
-      "grad_norm": 0.0012476647971197963,
-      "learning_rate": 0.00019999793919848343,
-      "loss": 46.0,
-      "step": 26738
-    },
-    {
-      "epoch": 2.0443832788577327,
-      "grad_norm": 0.005917396396398544,
-      "learning_rate": 0.00019999793904427565,
-      "loss": 46.0,
-      "step": 26739
-    },
-    {
-      "epoch": 2.0444597358411225,
-      "grad_norm": 0.0010810368694365025,
-      "learning_rate": 0.0001999979388900621,
-      "loss": 46.0,
-      "step": 26740
-    },
-    {
-      "epoch": 2.0445361928245123,
-      "grad_norm": 0.0007767686038278043,
-      "learning_rate": 0.00019999793873584274,
-      "loss": 46.0,
-      "step": 26741
-    },
-    {
-      "epoch": 2.044612649807902,
-      "grad_norm": 0.0006513047264888883,
-      "learning_rate": 0.00019999793858161764,
-      "loss": 46.0,
-      "step": 26742
-    },
-    {
-      "epoch": 2.0446891067912913,
-      "grad_norm": 0.00045319818309508264,
-      "learning_rate": 0.00019999793842738676,
-      "loss": 46.0,
-      "step": 26743
-    },
-    {
-      "epoch": 2.044765563774681,
-      "grad_norm": 0.004594389349222183,
-      "learning_rate": 0.0001999979382731501,
-      "loss": 46.0,
-      "step": 26744
-    },
-    {
-      "epoch": 2.044842020758071,
-      "grad_norm": 0.01075744442641735,
-      "learning_rate": 0.00019999793811890771,
-      "loss": 46.0,
-      "step": 26745
-    },
-    {
-      "epoch": 2.0449184777414606,
-      "grad_norm": 0.0008175843977369368,
-      "learning_rate": 0.00019999793796465954,
-      "loss": 46.0,
-      "step": 26746
-    },
-    {
-      "epoch": 2.0449949347248504,
-      "grad_norm": 0.0009589049150235951,
-      "learning_rate": 0.0001999979378104056,
-      "loss": 46.0,
-      "step": 26747
-    },
-    {
-      "epoch": 2.04507139170824,
-      "grad_norm": 0.0006627555703744292,
-      "learning_rate": 0.00019999793765614589,
-      "loss": 46.0,
-      "step": 26748
-    },
-    {
-      "epoch": 2.04514784869163,
-      "grad_norm": 0.0018860474228858948,
-      "learning_rate": 0.0001999979375018804,
-      "loss": 46.0,
-      "step": 26749
-    },
-    {
-      "epoch": 2.0452243056750197,
-      "grad_norm": 0.0005355579778552055,
-      "learning_rate": 0.00019999793734760913,
-      "loss": 46.0,
-      "step": 26750
-    },
-    {
-      "epoch": 2.0453007626584094,
-      "grad_norm": 0.001369601464830339,
-      "learning_rate": 0.0001999979371933321,
-      "loss": 46.0,
-      "step": 26751
-    },
-    {
-      "epoch": 2.045377219641799,
-      "grad_norm": 0.0010594576597213745,
-      "learning_rate": 0.0001999979370390493,
-      "loss": 46.0,
-      "step": 26752
-    },
-    {
-      "epoch": 2.045453676625189,
-      "grad_norm": 0.0010313743259757757,
-      "learning_rate": 0.0001999979368847607,
-      "loss": 46.0,
-      "step": 26753
-    },
-    {
-      "epoch": 2.0455301336085783,
-      "grad_norm": 0.00456597562879324,
-      "learning_rate": 0.00019999793673046638,
-      "loss": 46.0,
-      "step": 26754
-    },
-    {
-      "epoch": 2.045606590591968,
-      "grad_norm": 0.0008752791327424347,
-      "learning_rate": 0.00019999793657616627,
-      "loss": 46.0,
-      "step": 26755
-    },
-    {
-      "epoch": 2.045683047575358,
-      "grad_norm": 0.0017183005111292005,
-      "learning_rate": 0.0001999979364218604,
-      "loss": 46.0,
-      "step": 26756
-    },
-    {
-      "epoch": 2.0457595045587476,
-      "grad_norm": 0.0005333643639460206,
-      "learning_rate": 0.00019999793626754878,
-      "loss": 46.0,
-      "step": 26757
-    },
-    {
-      "epoch": 2.0458359615421373,
-      "grad_norm": 0.0005705517833121121,
-      "learning_rate": 0.00019999793611323135,
-      "loss": 46.0,
-      "step": 26758
-    },
-    {
-      "epoch": 2.045912418525527,
-      "grad_norm": 0.0025102447252720594,
-      "learning_rate": 0.00019999793595890813,
-      "loss": 46.0,
-      "step": 26759
-    },
-    {
-      "epoch": 2.045988875508917,
-      "grad_norm": 0.0011639269068837166,
-      "learning_rate": 0.0001999979358045792,
-      "loss": 46.0,
-      "step": 26760
-    },
-    {
-      "epoch": 2.0460653324923066,
-      "grad_norm": 0.0006512021063826978,
-      "learning_rate": 0.00019999793565024447,
-      "loss": 46.0,
-      "step": 26761
-    },
-    {
-      "epoch": 2.0461417894756964,
-      "grad_norm": 0.0011443783296272159,
-      "learning_rate": 0.000199997935495904,
-      "loss": 46.0,
-      "step": 26762
-    },
-    {
-      "epoch": 2.046218246459086,
-      "grad_norm": 0.01710735820233822,
-      "learning_rate": 0.00019999793534155772,
-      "loss": 46.0,
-      "step": 26763
-    },
-    {
-      "epoch": 2.046294703442476,
-      "grad_norm": 0.0020361868664622307,
-      "learning_rate": 0.0001999979351872057,
-      "loss": 46.0,
-      "step": 26764
-    },
-    {
-      "epoch": 2.046371160425865,
-      "grad_norm": 0.001960717374458909,
-      "learning_rate": 0.0001999979350328479,
-      "loss": 46.0,
-      "step": 26765
-    },
-    {
-      "epoch": 2.046447617409255,
-      "grad_norm": 0.0013600252568721771,
-      "learning_rate": 0.00019999793487848435,
-      "loss": 46.0,
-      "step": 26766
-    },
-    {
-      "epoch": 2.0465240743926447,
-      "grad_norm": 0.00043563719373196363,
-      "learning_rate": 0.000199997934724115,
-      "loss": 46.0,
-      "step": 26767
-    },
-    {
-      "epoch": 2.0466005313760345,
-      "grad_norm": 0.0008381299558095634,
-      "learning_rate": 0.0001999979345697399,
-      "loss": 46.0,
-      "step": 26768
-    },
-    {
-      "epoch": 2.0466769883594242,
-      "grad_norm": 0.0010570789454504848,
-      "learning_rate": 0.00019999793441535902,
-      "loss": 46.0,
-      "step": 26769
-    },
-    {
-      "epoch": 2.046753445342814,
-      "grad_norm": 0.00193371984641999,
-      "learning_rate": 0.00019999793426097237,
-      "loss": 46.0,
-      "step": 26770
-    },
-    {
-      "epoch": 2.0468299023262038,
-      "grad_norm": 0.0004712466325145215,
-      "learning_rate": 0.00019999793410657994,
-      "loss": 46.0,
-      "step": 26771
-    },
-    {
-      "epoch": 2.0469063593095935,
-      "grad_norm": 0.01838589459657669,
-      "learning_rate": 0.00019999793395218175,
-      "loss": 46.0,
-      "step": 26772
-    },
-    {
-      "epoch": 2.0469828162929833,
-      "grad_norm": 0.002186743775382638,
-      "learning_rate": 0.0001999979337977778,
-      "loss": 46.0,
-      "step": 26773
-    },
-    {
-      "epoch": 2.047059273276373,
-      "grad_norm": 0.000854240613989532,
-      "learning_rate": 0.0001999979336433681,
-      "loss": 46.0,
-      "step": 26774
-    },
-    {
-      "epoch": 2.047135730259763,
-      "grad_norm": 0.0005020123207941651,
-      "learning_rate": 0.0001999979334889526,
-      "loss": 46.0,
-      "step": 26775
-    },
-    {
-      "epoch": 2.047212187243152,
-      "grad_norm": 0.0009910749504342675,
-      "learning_rate": 0.00019999793333453134,
-      "loss": 46.0,
-      "step": 26776
-    },
-    {
-      "epoch": 2.047288644226542,
-      "grad_norm": 0.0016292175278067589,
-      "learning_rate": 0.0001999979331801043,
-      "loss": 46.0,
-      "step": 26777
-    },
-    {
-      "epoch": 2.0473651012099316,
-      "grad_norm": 0.0026920894160866737,
-      "learning_rate": 0.00019999793302567152,
-      "loss": 46.0,
-      "step": 26778
-    },
-    {
-      "epoch": 2.0474415581933214,
-      "grad_norm": 0.0006900280131958425,
-      "learning_rate": 0.00019999793287123294,
-      "loss": 46.0,
-      "step": 26779
-    },
-    {
-      "epoch": 2.047518015176711,
-      "grad_norm": 0.00046262776595540345,
-      "learning_rate": 0.00019999793271678859,
-      "loss": 46.0,
-      "step": 26780
-    },
-    {
-      "epoch": 2.047594472160101,
-      "grad_norm": 0.0025357462000101805,
-      "learning_rate": 0.00019999793256233848,
-      "loss": 46.0,
-      "step": 26781
-    },
-    {
-      "epoch": 2.0476709291434907,
-      "grad_norm": 0.0013350378721952438,
-      "learning_rate": 0.00019999793240788264,
-      "loss": 46.0,
-      "step": 26782
-    },
-    {
-      "epoch": 2.0477473861268805,
-      "grad_norm": 0.0007526238914579153,
-      "learning_rate": 0.000199997932253421,
-      "loss": 46.0,
-      "step": 26783
-    },
-    {
-      "epoch": 2.04782384311027,
-      "grad_norm": 0.0009977235458791256,
-      "learning_rate": 0.00019999793209895357,
-      "loss": 46.0,
-      "step": 26784
-    },
-    {
-      "epoch": 2.04790030009366,
-      "grad_norm": 0.000511665886733681,
-      "learning_rate": 0.00019999793194448037,
-      "loss": 46.0,
-      "step": 26785
-    },
-    {
-      "epoch": 2.0479767570770497,
-      "grad_norm": 0.0018593334825709462,
-      "learning_rate": 0.0001999979317900014,
-      "loss": 46.0,
-      "step": 26786
-    },
-    {
-      "epoch": 2.048053214060439,
-      "grad_norm": 0.0023004096001386642,
-      "learning_rate": 0.0001999979316355167,
-      "loss": 46.0,
-      "step": 26787
-    },
-    {
-      "epoch": 2.048129671043829,
-      "grad_norm": 0.0007995264022611082,
-      "learning_rate": 0.0001999979314810262,
-      "loss": 46.0,
-      "step": 26788
-    },
-    {
-      "epoch": 2.0482061280272186,
-      "grad_norm": 0.0011476317886263132,
-      "learning_rate": 0.00019999793132652994,
-      "loss": 46.0,
-      "step": 26789
-    },
-    {
-      "epoch": 2.0482825850106083,
-      "grad_norm": 0.0014823433011770248,
-      "learning_rate": 0.0001999979311720279,
-      "loss": 46.0,
-      "step": 26790
-    },
-    {
-      "epoch": 2.048359041993998,
-      "grad_norm": 0.000912494317162782,
-      "learning_rate": 0.0001999979310175201,
-      "loss": 46.0,
-      "step": 26791
-    },
-    {
-      "epoch": 2.048435498977388,
-      "grad_norm": 0.0040137311443686485,
-      "learning_rate": 0.00019999793086300654,
-      "loss": 46.0,
-      "step": 26792
-    },
-    {
-      "epoch": 2.0485119559607776,
-      "grad_norm": 0.0028996344190090895,
-      "learning_rate": 0.00019999793070848718,
-      "loss": 46.0,
-      "step": 26793
-    },
-    {
-      "epoch": 2.0485884129441674,
-      "grad_norm": 0.0004226558667141944,
-      "learning_rate": 0.00019999793055396208,
-      "loss": 46.0,
-      "step": 26794
-    },
-    {
-      "epoch": 2.048664869927557,
-      "grad_norm": 0.0003900914452970028,
-      "learning_rate": 0.0001999979303994312,
-      "loss": 46.0,
-      "step": 26795
-    },
-    {
-      "epoch": 2.048741326910947,
-      "grad_norm": 0.005706121679395437,
-      "learning_rate": 0.00019999793024489456,
-      "loss": 46.0,
-      "step": 26796
-    },
-    {
-      "epoch": 2.0488177838943367,
-      "grad_norm": 0.0014819680945947766,
-      "learning_rate": 0.00019999793009035214,
-      "loss": 46.0,
-      "step": 26797
-    },
-    {
-      "epoch": 2.048894240877726,
-      "grad_norm": 0.0008054747595451772,
-      "learning_rate": 0.00019999792993580394,
-      "loss": 46.0,
-      "step": 26798
-    },
-    {
-      "epoch": 2.0489706978611157,
-      "grad_norm": 0.0024097184650599957,
-      "learning_rate": 0.00019999792978124997,
-      "loss": 46.0,
-      "step": 26799
-    },
-    {
-      "epoch": 2.0490471548445055,
-      "grad_norm": 0.0009792178170755506,
-      "learning_rate": 0.00019999792962669026,
-      "loss": 46.0,
-      "step": 26800
-    },
-    {
-      "epoch": 2.0491236118278953,
-      "grad_norm": 0.0010091076837852597,
-      "learning_rate": 0.00019999792947212477,
-      "loss": 46.0,
-      "step": 26801
-    },
-    {
-      "epoch": 2.049200068811285,
-      "grad_norm": 0.0008666398352943361,
-      "learning_rate": 0.0001999979293175535,
-      "loss": 46.0,
-      "step": 26802
-    },
-    {
-      "epoch": 2.049276525794675,
-      "grad_norm": 0.005858314223587513,
-      "learning_rate": 0.00019999792916297645,
-      "loss": 46.0,
-      "step": 26803
-    },
-    {
-      "epoch": 2.0493529827780645,
-      "grad_norm": 0.0013598664663732052,
-      "learning_rate": 0.00019999792900839367,
-      "loss": 46.0,
-      "step": 26804
-    },
-    {
-      "epoch": 2.0494294397614543,
-      "grad_norm": 0.0007814023410901427,
-      "learning_rate": 0.0001999979288538051,
-      "loss": 46.0,
-      "step": 26805
-    },
-    {
-      "epoch": 2.049505896744844,
-      "grad_norm": 0.002629644935950637,
-      "learning_rate": 0.00019999792869921076,
-      "loss": 46.0,
-      "step": 26806
-    },
-    {
-      "epoch": 2.049582353728234,
-      "grad_norm": 0.00041831322596408427,
-      "learning_rate": 0.00019999792854461063,
-      "loss": 46.0,
-      "step": 26807
-    },
-    {
-      "epoch": 2.0496588107116236,
-      "grad_norm": 0.005221270490437746,
-      "learning_rate": 0.00019999792839000478,
-      "loss": 46.0,
-      "step": 26808
-    },
-    {
-      "epoch": 2.049735267695013,
-      "grad_norm": 0.002187157981097698,
-      "learning_rate": 0.0001999979282353931,
-      "loss": 46.0,
-      "step": 26809
-    },
-    {
-      "epoch": 2.0498117246784027,
-      "grad_norm": 0.0015697296475991607,
-      "learning_rate": 0.00019999792808077569,
-      "loss": 46.0,
-      "step": 26810
-    },
-    {
-      "epoch": 2.0498881816617924,
-      "grad_norm": 0.0010987631976604462,
-      "learning_rate": 0.00019999792792615252,
-      "loss": 46.0,
-      "step": 26811
-    },
-    {
-      "epoch": 2.049964638645182,
-      "grad_norm": 0.0005440912209451199,
-      "learning_rate": 0.00019999792777152355,
-      "loss": 46.0,
-      "step": 26812
-    },
-    {
-      "epoch": 2.050041095628572,
-      "grad_norm": 0.0017145854653790593,
-      "learning_rate": 0.0001999979276168888,
-      "loss": 46.0,
-      "step": 26813
-    },
-    {
-      "epoch": 2.0501175526119617,
-      "grad_norm": 0.003489426104351878,
-      "learning_rate": 0.00019999792746224832,
-      "loss": 46.0,
-      "step": 26814
-    },
-    {
-      "epoch": 2.0501940095953515,
-      "grad_norm": 0.0069366986863315105,
-      "learning_rate": 0.00019999792730760206,
-      "loss": 46.0,
-      "step": 26815
-    },
-    {
-      "epoch": 2.0502704665787412,
-      "grad_norm": 0.0020684825722128153,
-      "learning_rate": 0.00019999792715295,
-      "loss": 46.0,
-      "step": 26816
-    },
-    {
-      "epoch": 2.050346923562131,
-      "grad_norm": 0.0003260245139244944,
-      "learning_rate": 0.00019999792699829222,
-      "loss": 46.0,
-      "step": 26817
-    },
-    {
-      "epoch": 2.0504233805455208,
-      "grad_norm": 0.000811265257652849,
-      "learning_rate": 0.00019999792684362864,
-      "loss": 46.0,
-      "step": 26818
-    },
-    {
-      "epoch": 2.0504998375289105,
-      "grad_norm": 0.004917887505143881,
-      "learning_rate": 0.00019999792668895928,
-      "loss": 46.0,
-      "step": 26819
-    },
-    {
-      "epoch": 2.0505762945123,
-      "grad_norm": 0.0010727765038609505,
-      "learning_rate": 0.00019999792653428418,
-      "loss": 46.0,
-      "step": 26820
-    },
-    {
-      "epoch": 2.0506527514956896,
-      "grad_norm": 0.0016743084415793419,
-      "learning_rate": 0.0001999979263796033,
-      "loss": 46.0,
-      "step": 26821
-    },
-    {
-      "epoch": 2.0507292084790794,
-      "grad_norm": 0.012910761870443821,
-      "learning_rate": 0.00019999792622491666,
-      "loss": 46.0,
-      "step": 26822
-    },
-    {
-      "epoch": 2.050805665462469,
-      "grad_norm": 0.0011691940017044544,
-      "learning_rate": 0.00019999792607022424,
-      "loss": 46.0,
-      "step": 26823
-    },
-    {
-      "epoch": 2.050882122445859,
-      "grad_norm": 0.00044540679664351046,
-      "learning_rate": 0.00019999792591552604,
-      "loss": 46.0,
-      "step": 26824
-    },
-    {
-      "epoch": 2.0509585794292486,
-      "grad_norm": 0.000470242666779086,
-      "learning_rate": 0.00019999792576082208,
-      "loss": 46.0,
-      "step": 26825
-    },
-    {
-      "epoch": 2.0510350364126384,
-      "grad_norm": 0.001184387132525444,
-      "learning_rate": 0.00019999792560611236,
-      "loss": 46.0,
-      "step": 26826
-    },
-    {
-      "epoch": 2.051111493396028,
-      "grad_norm": 0.0026979998219758272,
-      "learning_rate": 0.00019999792545139687,
-      "loss": 46.0,
-      "step": 26827
-    },
-    {
-      "epoch": 2.051187950379418,
-      "grad_norm": 0.0015030615031719208,
-      "learning_rate": 0.0001999979252966756,
-      "loss": 46.0,
-      "step": 26828
-    },
-    {
-      "epoch": 2.0512644073628077,
-      "grad_norm": 0.008087236434221268,
-      "learning_rate": 0.00019999792514194855,
-      "loss": 46.0,
-      "step": 26829
-    },
-    {
-      "epoch": 2.051340864346197,
-      "grad_norm": 0.0014332369901239872,
-      "learning_rate": 0.00019999792498721575,
-      "loss": 46.0,
-      "step": 26830
-    },
-    {
-      "epoch": 2.0514173213295868,
-      "grad_norm": 0.0005862611578777432,
-      "learning_rate": 0.00019999792483247716,
-      "loss": 46.0,
-      "step": 26831
-    },
-    {
-      "epoch": 2.0514937783129765,
-      "grad_norm": 0.00040685906424187124,
-      "learning_rate": 0.0001999979246777328,
-      "loss": 46.0,
-      "step": 26832
-    },
-    {
-      "epoch": 2.0515702352963663,
-      "grad_norm": 0.0007863244391046464,
-      "learning_rate": 0.0001999979245229827,
-      "loss": 46.0,
-      "step": 26833
-    },
-    {
-      "epoch": 2.051646692279756,
-      "grad_norm": 0.0012992905685678124,
-      "learning_rate": 0.00019999792436822684,
-      "loss": 46.0,
-      "step": 26834
-    },
-    {
-      "epoch": 2.051723149263146,
-      "grad_norm": 0.0007701580179855227,
-      "learning_rate": 0.00019999792421346516,
-      "loss": 46.0,
-      "step": 26835
-    },
-    {
-      "epoch": 2.0517996062465356,
-      "grad_norm": 0.0012683391105383635,
-      "learning_rate": 0.00019999792405869777,
-      "loss": 46.0,
-      "step": 26836
-    },
-    {
-      "epoch": 2.0518760632299253,
-      "grad_norm": 0.0012288318248465657,
-      "learning_rate": 0.00019999792390392455,
-      "loss": 46.0,
-      "step": 26837
-    },
-    {
-      "epoch": 2.051952520213315,
-      "grad_norm": 0.0022708745673298836,
-      "learning_rate": 0.0001999979237491456,
-      "loss": 46.0,
-      "step": 26838
-    },
-    {
-      "epoch": 2.052028977196705,
-      "grad_norm": 0.0011641462333500385,
-      "learning_rate": 0.0001999979235943609,
-      "loss": 46.0,
-      "step": 26839
-    },
-    {
-      "epoch": 2.0521054341800946,
-      "grad_norm": 0.0005502661224454641,
-      "learning_rate": 0.00019999792343957038,
-      "loss": 46.0,
-      "step": 26840
-    },
-    {
-      "epoch": 2.052181891163484,
-      "grad_norm": 0.0011832385789602995,
-      "learning_rate": 0.00019999792328477412,
-      "loss": 46.0,
-      "step": 26841
-    },
-    {
-      "epoch": 2.0522583481468737,
-      "grad_norm": 0.002460158197209239,
-      "learning_rate": 0.00019999792312997208,
-      "loss": 46.0,
-      "step": 26842
-    },
-    {
-      "epoch": 2.0523348051302635,
-      "grad_norm": 0.0019206901779398322,
-      "learning_rate": 0.00019999792297516428,
-      "loss": 46.0,
-      "step": 26843
-    },
-    {
-      "epoch": 2.052411262113653,
-      "grad_norm": 0.0009876237018033862,
-      "learning_rate": 0.0001999979228203507,
-      "loss": 46.0,
-      "step": 26844
-    },
-    {
-      "epoch": 2.052487719097043,
-      "grad_norm": 0.0011415391927585006,
-      "learning_rate": 0.00019999792266553134,
-      "loss": 46.0,
-      "step": 26845
-    },
-    {
-      "epoch": 2.0525641760804327,
-      "grad_norm": 0.0009885606123134494,
-      "learning_rate": 0.00019999792251070622,
-      "loss": 46.0,
-      "step": 26846
-    },
-    {
-      "epoch": 2.0526406330638225,
-      "grad_norm": 0.000393895257730037,
-      "learning_rate": 0.00019999792235587537,
-      "loss": 46.0,
-      "step": 26847
-    },
-    {
-      "epoch": 2.0527170900472123,
-      "grad_norm": 0.0018497485434636474,
-      "learning_rate": 0.00019999792220103867,
-      "loss": 46.0,
-      "step": 26848
-    },
-    {
-      "epoch": 2.052793547030602,
-      "grad_norm": 0.0012509553926065564,
-      "learning_rate": 0.00019999792204619627,
-      "loss": 46.0,
-      "step": 26849
-    },
-    {
-      "epoch": 2.052870004013992,
-      "grad_norm": 0.0013123966054990888,
-      "learning_rate": 0.00019999792189134808,
-      "loss": 46.0,
-      "step": 26850
-    },
-    {
-      "epoch": 2.0529464609973815,
-      "grad_norm": 0.0008878799853846431,
-      "learning_rate": 0.00019999792173649409,
-      "loss": 46.0,
-      "step": 26851
-    },
-    {
-      "epoch": 2.053022917980771,
-      "grad_norm": 0.0007017144234851003,
-      "learning_rate": 0.00019999792158163437,
-      "loss": 46.0,
-      "step": 26852
-    },
-    {
-      "epoch": 2.0530993749641606,
-      "grad_norm": 0.0009450880461372435,
-      "learning_rate": 0.00019999792142676889,
-      "loss": 46.0,
-      "step": 26853
-    },
-    {
-      "epoch": 2.0531758319475504,
-      "grad_norm": 0.0024252634029835463,
-      "learning_rate": 0.0001999979212718976,
-      "loss": 46.0,
-      "step": 26854
-    },
-    {
-      "epoch": 2.05325228893094,
-      "grad_norm": 0.00022634721244685352,
-      "learning_rate": 0.00019999792111702057,
-      "loss": 46.0,
-      "step": 26855
-    },
-    {
-      "epoch": 2.05332874591433,
-      "grad_norm": 0.001143791014328599,
-      "learning_rate": 0.00019999792096213776,
-      "loss": 46.0,
-      "step": 26856
-    },
-    {
-      "epoch": 2.0534052028977197,
-      "grad_norm": 0.0012385756708681583,
-      "learning_rate": 0.0001999979208072492,
-      "loss": 46.0,
-      "step": 26857
-    },
-    {
-      "epoch": 2.0534816598811094,
-      "grad_norm": 0.0010120879160240293,
-      "learning_rate": 0.00019999792065235482,
-      "loss": 46.0,
-      "step": 26858
-    },
-    {
-      "epoch": 2.053558116864499,
-      "grad_norm": 0.0013031127164140344,
-      "learning_rate": 0.00019999792049745475,
-      "loss": 46.0,
-      "step": 26859
-    },
-    {
-      "epoch": 2.053634573847889,
-      "grad_norm": 0.0009092942927964032,
-      "learning_rate": 0.00019999792034254882,
-      "loss": 46.0,
-      "step": 26860
-    },
-    {
-      "epoch": 2.0537110308312787,
-      "grad_norm": 0.0010208553867414594,
-      "learning_rate": 0.00019999792018763718,
-      "loss": 46.0,
-      "step": 26861
-    },
-    {
-      "epoch": 2.0537874878146685,
-      "grad_norm": 0.004471789114177227,
-      "learning_rate": 0.00019999792003271976,
-      "loss": 46.0,
-      "step": 26862
-    },
-    {
-      "epoch": 2.053863944798058,
-      "grad_norm": 0.0008084700675681233,
-      "learning_rate": 0.00019999791987779654,
-      "loss": 46.0,
-      "step": 26863
-    },
-    {
-      "epoch": 2.0539404017814475,
-      "grad_norm": 0.008358980529010296,
-      "learning_rate": 0.0001999979197228676,
-      "loss": 46.0,
-      "step": 26864
-    },
-    {
-      "epoch": 2.0540168587648373,
-      "grad_norm": 0.0011034226045012474,
-      "learning_rate": 0.00019999791956793286,
-      "loss": 46.0,
-      "step": 26865
-    },
-    {
-      "epoch": 2.054093315748227,
-      "grad_norm": 0.004931580275297165,
-      "learning_rate": 0.00019999791941299234,
-      "loss": 46.0,
-      "step": 26866
-    },
-    {
-      "epoch": 2.054169772731617,
-      "grad_norm": 0.0019216890214011073,
-      "learning_rate": 0.00019999791925804608,
-      "loss": 46.0,
-      "step": 26867
-    },
-    {
-      "epoch": 2.0542462297150066,
-      "grad_norm": 0.0012013522209599614,
-      "learning_rate": 0.00019999791910309408,
-      "loss": 46.0,
-      "step": 26868
-    },
-    {
-      "epoch": 2.0543226866983963,
-      "grad_norm": 0.0013623957056552172,
-      "learning_rate": 0.00019999791894813624,
-      "loss": 46.0,
-      "step": 26869
-    },
-    {
-      "epoch": 2.054399143681786,
-      "grad_norm": 0.0010508917039260268,
-      "learning_rate": 0.00019999791879317266,
-      "loss": 46.0,
-      "step": 26870
-    },
-    {
-      "epoch": 2.054475600665176,
-      "grad_norm": 0.0007067056722007692,
-      "learning_rate": 0.0001999979186382033,
-      "loss": 46.0,
-      "step": 26871
-    },
-    {
-      "epoch": 2.0545520576485656,
-      "grad_norm": 0.0008177933050319552,
-      "learning_rate": 0.00019999791848322818,
-      "loss": 46.0,
-      "step": 26872
-    },
-    {
-      "epoch": 2.0546285146319554,
-      "grad_norm": 0.009712925180792809,
-      "learning_rate": 0.0001999979183282473,
-      "loss": 46.0,
-      "step": 26873
-    },
-    {
-      "epoch": 2.0547049716153447,
-      "grad_norm": 0.000902174215298146,
-      "learning_rate": 0.00019999791817326066,
-      "loss": 46.0,
-      "step": 26874
-    },
-    {
-      "epoch": 2.0547814285987345,
-      "grad_norm": 0.0019085563253611326,
-      "learning_rate": 0.00019999791801826824,
-      "loss": 46.0,
-      "step": 26875
-    },
-    {
-      "epoch": 2.0548578855821242,
-      "grad_norm": 0.0025579489301890135,
-      "learning_rate": 0.00019999791786327005,
-      "loss": 46.0,
-      "step": 26876
-    },
-    {
-      "epoch": 2.054934342565514,
-      "grad_norm": 0.0005497814272530377,
-      "learning_rate": 0.00019999791770826609,
-      "loss": 46.0,
-      "step": 26877
-    },
-    {
-      "epoch": 2.0550107995489038,
-      "grad_norm": 0.0005591757362708449,
-      "learning_rate": 0.00019999791755325635,
-      "loss": 46.0,
-      "step": 26878
-    },
-    {
-      "epoch": 2.0550872565322935,
-      "grad_norm": 0.0006346144364215434,
-      "learning_rate": 0.00019999791739824083,
-      "loss": 46.0,
-      "step": 26879
-    },
-    {
-      "epoch": 2.0551637135156833,
-      "grad_norm": 0.0009246206027455628,
-      "learning_rate": 0.00019999791724321955,
-      "loss": 46.0,
-      "step": 26880
-    },
-    {
-      "epoch": 2.055240170499073,
-      "grad_norm": 0.0008889971068128943,
-      "learning_rate": 0.00019999791708819254,
-      "loss": 46.0,
-      "step": 26881
-    },
-    {
-      "epoch": 2.055316627482463,
-      "grad_norm": 0.0010018158936873078,
-      "learning_rate": 0.0001999979169331597,
-      "loss": 46.0,
-      "step": 26882
-    },
-    {
-      "epoch": 2.0553930844658526,
-      "grad_norm": 0.0008579955901950598,
-      "learning_rate": 0.00019999791677812115,
-      "loss": 46.0,
-      "step": 26883
-    },
-    {
-      "epoch": 2.0554695414492423,
-      "grad_norm": 0.0028697752859443426,
-      "learning_rate": 0.00019999791662307677,
-      "loss": 46.0,
-      "step": 26884
-    },
-    {
-      "epoch": 2.0555459984326316,
-      "grad_norm": 0.0020914908964186907,
-      "learning_rate": 0.00019999791646802665,
-      "loss": 46.0,
-      "step": 26885
-    },
-    {
-      "epoch": 2.0556224554160214,
-      "grad_norm": 0.0008215868147090077,
-      "learning_rate": 0.00019999791631297077,
-      "loss": 46.0,
-      "step": 26886
-    },
-    {
-      "epoch": 2.055698912399411,
-      "grad_norm": 0.0004442485223989934,
-      "learning_rate": 0.00019999791615790913,
-      "loss": 46.0,
-      "step": 26887
-    },
-    {
-      "epoch": 2.055775369382801,
-      "grad_norm": 0.006440585479140282,
-      "learning_rate": 0.0001999979160028417,
-      "loss": 46.0,
-      "step": 26888
-    },
-    {
-      "epoch": 2.0558518263661907,
-      "grad_norm": 0.0020210817456245422,
-      "learning_rate": 0.0001999979158477685,
-      "loss": 46.0,
-      "step": 26889
-    },
-    {
-      "epoch": 2.0559282833495804,
-      "grad_norm": 0.0007858136668801308,
-      "learning_rate": 0.00019999791569268952,
-      "loss": 46.0,
-      "step": 26890
-    },
-    {
-      "epoch": 2.05600474033297,
-      "grad_norm": 0.0006385630113072693,
-      "learning_rate": 0.00019999791553760478,
-      "loss": 46.0,
-      "step": 26891
-    },
-    {
-      "epoch": 2.05608119731636,
-      "grad_norm": 0.0007511741714552045,
-      "learning_rate": 0.0001999979153825143,
-      "loss": 46.0,
-      "step": 26892
-    },
-    {
-      "epoch": 2.0561576542997497,
-      "grad_norm": 0.0018868870101869106,
-      "learning_rate": 0.000199997915227418,
-      "loss": 46.0,
-      "step": 26893
-    },
-    {
-      "epoch": 2.0562341112831395,
-      "grad_norm": 0.0011995711829513311,
-      "learning_rate": 0.00019999791507231595,
-      "loss": 46.0,
-      "step": 26894
-    },
-    {
-      "epoch": 2.0563105682665292,
-      "grad_norm": 0.0007007892127148807,
-      "learning_rate": 0.00019999791491720817,
-      "loss": 46.0,
-      "step": 26895
-    },
-    {
-      "epoch": 2.0563870252499186,
-      "grad_norm": 0.00204948871396482,
-      "learning_rate": 0.00019999791476209457,
-      "loss": 46.0,
-      "step": 26896
-    },
-    {
-      "epoch": 2.0564634822333083,
-      "grad_norm": 0.0017381446668878198,
-      "learning_rate": 0.00019999791460697521,
-      "loss": 46.0,
-      "step": 26897
-    },
-    {
-      "epoch": 2.056539939216698,
-      "grad_norm": 0.0002926558372564614,
-      "learning_rate": 0.0001999979144518501,
-      "loss": 46.0,
-      "step": 26898
-    },
-    {
-      "epoch": 2.056616396200088,
-      "grad_norm": 0.0029764098580926657,
-      "learning_rate": 0.00019999791429671922,
-      "loss": 46.0,
-      "step": 26899
-    },
-    {
-      "epoch": 2.0566928531834776,
-      "grad_norm": 0.0014221267774701118,
-      "learning_rate": 0.00019999791414158254,
-      "loss": 46.0,
-      "step": 26900
-    },
-    {
-      "epoch": 2.0567693101668674,
-      "grad_norm": 0.0029510289896279573,
-      "learning_rate": 0.00019999791398644012,
-      "loss": 46.0,
-      "step": 26901
-    },
-    {
-      "epoch": 2.056845767150257,
-      "grad_norm": 0.0007957772468216717,
-      "learning_rate": 0.0001999979138312919,
-      "loss": 46.0,
-      "step": 26902
-    },
-    {
-      "epoch": 2.056922224133647,
-      "grad_norm": 0.0019415791612118483,
-      "learning_rate": 0.00019999791367613797,
-      "loss": 46.0,
-      "step": 26903
-    },
-    {
-      "epoch": 2.0569986811170367,
-      "grad_norm": 0.004954686388373375,
-      "learning_rate": 0.00019999791352097823,
-      "loss": 46.0,
-      "step": 26904
-    },
-    {
-      "epoch": 2.0570751381004264,
-      "grad_norm": 0.0011634246911853552,
-      "learning_rate": 0.00019999791336581271,
-      "loss": 46.0,
-      "step": 26905
-    },
-    {
-      "epoch": 2.057151595083816,
-      "grad_norm": 0.0009029583306983113,
-      "learning_rate": 0.00019999791321064145,
-      "loss": 46.0,
-      "step": 26906
-    },
-    {
-      "epoch": 2.0572280520672055,
-      "grad_norm": 0.00045587020576931536,
-      "learning_rate": 0.00019999791305546437,
-      "loss": 46.0,
-      "step": 26907
-    },
-    {
-      "epoch": 2.0573045090505953,
-      "grad_norm": 0.0006289492012001574,
-      "learning_rate": 0.0001999979129002816,
-      "loss": 46.0,
-      "step": 26908
-    },
-    {
-      "epoch": 2.057380966033985,
-      "grad_norm": 0.0009635791648179293,
-      "learning_rate": 0.000199997912745093,
-      "loss": 46.0,
-      "step": 26909
-    },
-    {
-      "epoch": 2.0574574230173748,
-      "grad_norm": 0.0024948331993073225,
-      "learning_rate": 0.00019999791258989866,
-      "loss": 46.0,
-      "step": 26910
-    },
-    {
-      "epoch": 2.0575338800007645,
-      "grad_norm": 0.0032892015296965837,
-      "learning_rate": 0.00019999791243469853,
-      "loss": 46.0,
-      "step": 26911
-    },
-    {
-      "epoch": 2.0576103369841543,
-      "grad_norm": 0.0022792054805904627,
-      "learning_rate": 0.00019999791227949266,
-      "loss": 46.0,
-      "step": 26912
-    },
-    {
-      "epoch": 2.057686793967544,
-      "grad_norm": 0.0016760226571932435,
-      "learning_rate": 0.000199997912124281,
-      "loss": 46.0,
-      "step": 26913
-    },
-    {
-      "epoch": 2.057763250950934,
-      "grad_norm": 0.0016270535998046398,
-      "learning_rate": 0.00019999791196906354,
-      "loss": 46.0,
-      "step": 26914
-    },
-    {
-      "epoch": 2.0578397079343236,
-      "grad_norm": 0.0012395388912409544,
-      "learning_rate": 0.00019999791181384032,
-      "loss": 46.0,
-      "step": 26915
-    },
-    {
-      "epoch": 2.0579161649177133,
-      "grad_norm": 0.0005843533435836434,
-      "learning_rate": 0.00019999791165861138,
-      "loss": 46.0,
-      "step": 26916
-    },
-    {
-      "epoch": 2.057992621901103,
-      "grad_norm": 0.000997414579614997,
-      "learning_rate": 0.00019999791150337665,
-      "loss": 46.0,
-      "step": 26917
-    },
-    {
-      "epoch": 2.0580690788844924,
-      "grad_norm": 0.001252759713679552,
-      "learning_rate": 0.00019999791134813613,
-      "loss": 46.0,
-      "step": 26918
-    },
-    {
-      "epoch": 2.058145535867882,
-      "grad_norm": 0.0006714845658279955,
-      "learning_rate": 0.00019999791119288985,
-      "loss": 46.0,
-      "step": 26919
-    },
-    {
-      "epoch": 2.058221992851272,
-      "grad_norm": 0.002894036937505007,
-      "learning_rate": 0.00019999791103763782,
-      "loss": 46.0,
-      "step": 26920
-    },
-    {
-      "epoch": 2.0582984498346617,
-      "grad_norm": 0.00043962011113762856,
-      "learning_rate": 0.00019999791088238,
-      "loss": 46.0,
-      "step": 26921
-    },
-    {
-      "epoch": 2.0583749068180515,
-      "grad_norm": 0.0007253745570778847,
-      "learning_rate": 0.00019999791072711643,
-      "loss": 46.0,
-      "step": 26922
-    },
-    {
-      "epoch": 2.0584513638014412,
-      "grad_norm": 0.0009464692557230592,
-      "learning_rate": 0.00019999791057184708,
-      "loss": 46.0,
-      "step": 26923
-    },
-    {
-      "epoch": 2.058527820784831,
-      "grad_norm": 0.0027029828634113073,
-      "learning_rate": 0.00019999791041657195,
-      "loss": 46.0,
-      "step": 26924
-    },
-    {
-      "epoch": 2.0586042777682207,
-      "grad_norm": 0.0006274234619922936,
-      "learning_rate": 0.00019999791026129106,
-      "loss": 46.0,
-      "step": 26925
-    },
-    {
-      "epoch": 2.0586807347516105,
-      "grad_norm": 0.01396503672003746,
-      "learning_rate": 0.00019999791010600438,
-      "loss": 46.0,
-      "step": 26926
-    },
-    {
-      "epoch": 2.0587571917350003,
-      "grad_norm": 0.012848246842622757,
-      "learning_rate": 0.00019999790995071194,
-      "loss": 46.0,
-      "step": 26927
-    },
-    {
-      "epoch": 2.05883364871839,
-      "grad_norm": 0.00488661602139473,
-      "learning_rate": 0.00019999790979541375,
-      "loss": 46.0,
-      "step": 26928
-    },
-    {
-      "epoch": 2.0589101057017793,
-      "grad_norm": 0.0011477720690891147,
-      "learning_rate": 0.00019999790964010978,
-      "loss": 46.0,
-      "step": 26929
-    },
-    {
-      "epoch": 2.058986562685169,
-      "grad_norm": 0.0011884690029546618,
-      "learning_rate": 0.00019999790948480004,
-      "loss": 46.0,
-      "step": 26930
-    },
-    {
-      "epoch": 2.059063019668559,
-      "grad_norm": 0.0014746342785656452,
-      "learning_rate": 0.00019999790932948456,
-      "loss": 46.0,
-      "step": 26931
-    },
-    {
-      "epoch": 2.0591394766519486,
-      "grad_norm": 0.0021771288011223078,
-      "learning_rate": 0.00019999790917416324,
-      "loss": 46.0,
-      "step": 26932
-    },
-    {
-      "epoch": 2.0592159336353384,
-      "grad_norm": 0.0027020389679819345,
-      "learning_rate": 0.00019999790901883619,
-      "loss": 46.0,
-      "step": 26933
-    },
-    {
-      "epoch": 2.059292390618728,
-      "grad_norm": 0.0006614235462620854,
-      "learning_rate": 0.00019999790886350338,
-      "loss": 46.0,
-      "step": 26934
-    },
-    {
-      "epoch": 2.059368847602118,
-      "grad_norm": 0.0030041332356631756,
-      "learning_rate": 0.0001999979087081648,
-      "loss": 46.0,
-      "step": 26935
-    },
-    {
-      "epoch": 2.0594453045855077,
-      "grad_norm": 0.0033881342969834805,
-      "learning_rate": 0.00019999790855282042,
-      "loss": 46.0,
-      "step": 26936
-    },
-    {
-      "epoch": 2.0595217615688974,
-      "grad_norm": 0.0007130822050385177,
-      "learning_rate": 0.00019999790839747033,
-      "loss": 46.0,
-      "step": 26937
-    },
-    {
-      "epoch": 2.059598218552287,
-      "grad_norm": 0.0012841123389080167,
-      "learning_rate": 0.00019999790824211443,
-      "loss": 46.0,
-      "step": 26938
-    },
-    {
-      "epoch": 2.059674675535677,
-      "grad_norm": 0.00044165831059217453,
-      "learning_rate": 0.00019999790808675276,
-      "loss": 46.0,
-      "step": 26939
-    },
-    {
-      "epoch": 2.0597511325190663,
-      "grad_norm": 0.000506731157656759,
-      "learning_rate": 0.00019999790793138534,
-      "loss": 46.0,
-      "step": 26940
-    },
-    {
-      "epoch": 2.059827589502456,
-      "grad_norm": 0.0023624610621482134,
-      "learning_rate": 0.00019999790777601212,
-      "loss": 46.0,
-      "step": 26941
-    },
-    {
-      "epoch": 2.059904046485846,
-      "grad_norm": 0.0014661407331004739,
-      "learning_rate": 0.00019999790762063318,
-      "loss": 46.0,
-      "step": 26942
-    },
-    {
-      "epoch": 2.0599805034692356,
-      "grad_norm": 0.0004016837337985635,
-      "learning_rate": 0.00019999790746524842,
-      "loss": 46.0,
-      "step": 26943
-    },
-    {
-      "epoch": 2.0600569604526253,
-      "grad_norm": 0.0009443744784221053,
-      "learning_rate": 0.0001999979073098579,
-      "loss": 46.0,
-      "step": 26944
-    },
-    {
-      "epoch": 2.060133417436015,
-      "grad_norm": 0.0008529173210263252,
-      "learning_rate": 0.00019999790715446162,
-      "loss": 46.0,
-      "step": 26945
-    },
-    {
-      "epoch": 2.060209874419405,
-      "grad_norm": 0.0005631794920191169,
-      "learning_rate": 0.0001999979069990596,
-      "loss": 46.0,
-      "step": 26946
-    },
-    {
-      "epoch": 2.0602863314027946,
-      "grad_norm": 0.00044998229714110494,
-      "learning_rate": 0.00019999790684365176,
-      "loss": 46.0,
-      "step": 26947
-    },
-    {
-      "epoch": 2.0603627883861844,
-      "grad_norm": 0.0022053855936974287,
-      "learning_rate": 0.00019999790668823818,
-      "loss": 46.0,
-      "step": 26948
-    },
-    {
-      "epoch": 2.060439245369574,
-      "grad_norm": 0.0008079966064542532,
-      "learning_rate": 0.00019999790653281883,
-      "loss": 46.0,
-      "step": 26949
-    },
-    {
-      "epoch": 2.060515702352964,
-      "grad_norm": 0.0005781477666459978,
-      "learning_rate": 0.0001999979063773937,
-      "loss": 46.0,
-      "step": 26950
-    },
-    {
-      "epoch": 2.060592159336353,
-      "grad_norm": 0.002677736571058631,
-      "learning_rate": 0.0001999979062219628,
-      "loss": 46.0,
-      "step": 26951
-    },
-    {
-      "epoch": 2.060668616319743,
-      "grad_norm": 0.0018682688241824508,
-      "learning_rate": 0.00019999790606652616,
-      "loss": 46.0,
-      "step": 26952
-    },
-    {
-      "epoch": 2.0607450733031327,
-      "grad_norm": 0.007014586590230465,
-      "learning_rate": 0.0001999979059110837,
-      "loss": 46.0,
-      "step": 26953
-    },
-    {
-      "epoch": 2.0608215302865225,
-      "grad_norm": 0.000522536167409271,
-      "learning_rate": 0.0001999979057556355,
-      "loss": 46.0,
-      "step": 26954
-    },
-    {
-      "epoch": 2.0608979872699122,
-      "grad_norm": 0.0007256384706124663,
-      "learning_rate": 0.00019999790560018153,
-      "loss": 46.0,
-      "step": 26955
-    },
-    {
-      "epoch": 2.060974444253302,
-      "grad_norm": 0.0012019777204841375,
-      "learning_rate": 0.0001999979054447218,
-      "loss": 46.0,
-      "step": 26956
-    },
-    {
-      "epoch": 2.0610509012366918,
-      "grad_norm": 0.001009106868878007,
-      "learning_rate": 0.00019999790528925625,
-      "loss": 46.0,
-      "step": 26957
-    },
-    {
-      "epoch": 2.0611273582200815,
-      "grad_norm": 0.000709208135958761,
-      "learning_rate": 0.00019999790513378497,
-      "loss": 46.0,
-      "step": 26958
-    },
-    {
-      "epoch": 2.0612038152034713,
-      "grad_norm": 0.0006043802131898701,
-      "learning_rate": 0.00019999790497830794,
-      "loss": 46.0,
-      "step": 26959
-    },
-    {
-      "epoch": 2.061280272186861,
-      "grad_norm": 0.0005293880822136998,
-      "learning_rate": 0.00019999790482282513,
-      "loss": 46.0,
-      "step": 26960
-    },
-    {
-      "epoch": 2.0613567291702504,
-      "grad_norm": 0.0008860472589731216,
-      "learning_rate": 0.00019999790466733656,
-      "loss": 46.0,
-      "step": 26961
-    },
-    {
-      "epoch": 2.06143318615364,
-      "grad_norm": 0.004374388605356216,
-      "learning_rate": 0.00019999790451184218,
-      "loss": 46.0,
-      "step": 26962
-    },
-    {
-      "epoch": 2.06150964313703,
-      "grad_norm": 0.0013512099394574761,
-      "learning_rate": 0.00019999790435634205,
-      "loss": 46.0,
-      "step": 26963
-    },
-    {
-      "epoch": 2.0615861001204197,
-      "grad_norm": 0.0013697885442525148,
-      "learning_rate": 0.00019999790420083616,
-      "loss": 46.0,
-      "step": 26964
-    },
-    {
-      "epoch": 2.0616625571038094,
-      "grad_norm": 0.0012314216000959277,
-      "learning_rate": 0.0001999979040453245,
-      "loss": 46.0,
-      "step": 26965
-    },
-    {
-      "epoch": 2.061739014087199,
-      "grad_norm": 0.010410306043922901,
-      "learning_rate": 0.00019999790388980707,
-      "loss": 46.0,
-      "step": 26966
-    },
-    {
-      "epoch": 2.061815471070589,
-      "grad_norm": 0.00029443763196468353,
-      "learning_rate": 0.00019999790373428385,
-      "loss": 46.0,
-      "step": 26967
-    },
-    {
-      "epoch": 2.0618919280539787,
-      "grad_norm": 0.0008221578318625689,
-      "learning_rate": 0.00019999790357875489,
-      "loss": 46.0,
-      "step": 26968
-    },
-    {
-      "epoch": 2.0619683850373685,
-      "grad_norm": 0.00038452661829069257,
-      "learning_rate": 0.00019999790342322015,
-      "loss": 46.0,
-      "step": 26969
-    },
-    {
-      "epoch": 2.062044842020758,
-      "grad_norm": 0.0008576579275541008,
-      "learning_rate": 0.0001999979032676796,
-      "loss": 46.0,
-      "step": 26970
-    },
-    {
-      "epoch": 2.062121299004148,
-      "grad_norm": 0.0021967049688100815,
-      "learning_rate": 0.00019999790311213332,
-      "loss": 46.0,
-      "step": 26971
-    },
-    {
-      "epoch": 2.0621977559875373,
-      "grad_norm": 0.0037471975665539503,
-      "learning_rate": 0.00019999790295658127,
-      "loss": 46.0,
-      "step": 26972
-    },
-    {
-      "epoch": 2.062274212970927,
-      "grad_norm": 0.001114453887566924,
-      "learning_rate": 0.00019999790280102346,
-      "loss": 46.0,
-      "step": 26973
-    },
-    {
-      "epoch": 2.062350669954317,
-      "grad_norm": 0.01322798803448677,
-      "learning_rate": 0.00019999790264545989,
-      "loss": 46.0,
-      "step": 26974
-    },
-    {
-      "epoch": 2.0624271269377066,
-      "grad_norm": 0.00037531653651967645,
-      "learning_rate": 0.0001999979024898905,
-      "loss": 46.0,
-      "step": 26975
-    },
-    {
-      "epoch": 2.0625035839210963,
-      "grad_norm": 0.002553671132773161,
-      "learning_rate": 0.00019999790233431538,
-      "loss": 46.0,
-      "step": 26976
-    },
-    {
-      "epoch": 2.062580040904486,
-      "grad_norm": 0.002734871581196785,
-      "learning_rate": 0.0001999979021787345,
-      "loss": 46.0,
-      "step": 26977
-    },
-    {
-      "epoch": 2.062656497887876,
-      "grad_norm": 0.0004395974974613637,
-      "learning_rate": 0.00019999790202314782,
-      "loss": 46.0,
-      "step": 26978
-    },
-    {
-      "epoch": 2.0627329548712656,
-      "grad_norm": 0.0006870915531180799,
-      "learning_rate": 0.00019999790186755537,
-      "loss": 46.0,
-      "step": 26979
-    },
-    {
-      "epoch": 2.0628094118546554,
-      "grad_norm": 0.001701031462289393,
-      "learning_rate": 0.00019999790171195718,
-      "loss": 46.0,
-      "step": 26980
-    },
-    {
-      "epoch": 2.062885868838045,
-      "grad_norm": 0.0014159544371068478,
-      "learning_rate": 0.00019999790155635322,
-      "loss": 46.0,
-      "step": 26981
-    },
-    {
-      "epoch": 2.062962325821435,
-      "grad_norm": 0.008852141909301281,
-      "learning_rate": 0.00019999790140074345,
-      "loss": 46.0,
-      "step": 26982
-    },
-    {
-      "epoch": 2.0630387828048242,
-      "grad_norm": 0.0014562713913619518,
-      "learning_rate": 0.00019999790124512794,
-      "loss": 46.0,
-      "step": 26983
-    },
-    {
-      "epoch": 2.063115239788214,
-      "grad_norm": 0.0004403952043503523,
-      "learning_rate": 0.00019999790108950666,
-      "loss": 46.0,
-      "step": 26984
-    },
-    {
-      "epoch": 2.0631916967716037,
-      "grad_norm": 0.00234870333224535,
-      "learning_rate": 0.00019999790093387963,
-      "loss": 46.0,
-      "step": 26985
-    },
-    {
-      "epoch": 2.0632681537549935,
-      "grad_norm": 0.0035048872232437134,
-      "learning_rate": 0.0001999979007782468,
-      "loss": 46.0,
-      "step": 26986
-    },
-    {
-      "epoch": 2.0633446107383833,
-      "grad_norm": 0.0006169637781567872,
-      "learning_rate": 0.00019999790062260822,
-      "loss": 46.0,
-      "step": 26987
-    },
-    {
-      "epoch": 2.063421067721773,
-      "grad_norm": 0.001449449686333537,
-      "learning_rate": 0.00019999790046696384,
-      "loss": 46.0,
-      "step": 26988
-    },
-    {
-      "epoch": 2.063497524705163,
-      "grad_norm": 0.0023134795483201742,
-      "learning_rate": 0.00019999790031131372,
-      "loss": 46.0,
-      "step": 26989
-    },
-    {
-      "epoch": 2.0635739816885525,
-      "grad_norm": 0.0005178030696697533,
-      "learning_rate": 0.00019999790015565782,
-      "loss": 46.0,
-      "step": 26990
-    },
-    {
-      "epoch": 2.0636504386719423,
-      "grad_norm": 0.001162930391728878,
-      "learning_rate": 0.00019999789999999615,
-      "loss": 46.0,
-      "step": 26991
-    },
-    {
-      "epoch": 2.063726895655332,
-      "grad_norm": 0.005006987601518631,
-      "learning_rate": 0.0001999978998443287,
-      "loss": 46.0,
-      "step": 26992
-    },
-    {
-      "epoch": 2.063803352638722,
-      "grad_norm": 0.0010016504675149918,
-      "learning_rate": 0.0001999978996886555,
-      "loss": 46.0,
-      "step": 26993
-    },
-    {
-      "epoch": 2.063879809622111,
-      "grad_norm": 0.002906529698520899,
-      "learning_rate": 0.00019999789953297652,
-      "loss": 46.0,
-      "step": 26994
-    },
-    {
-      "epoch": 2.063956266605501,
-      "grad_norm": 0.0005867416621185839,
-      "learning_rate": 0.0001999978993772918,
-      "loss": 46.0,
-      "step": 26995
-    },
-    {
-      "epoch": 2.0640327235888907,
-      "grad_norm": 0.00039372959872707725,
-      "learning_rate": 0.00019999789922160125,
-      "loss": 46.0,
-      "step": 26996
-    },
-    {
-      "epoch": 2.0641091805722804,
-      "grad_norm": 0.001543025253340602,
-      "learning_rate": 0.000199997899065905,
-      "loss": 46.0,
-      "step": 26997
-    },
-    {
-      "epoch": 2.06418563755567,
-      "grad_norm": 0.0013029432157054543,
-      "learning_rate": 0.00019999789891020294,
-      "loss": 46.0,
-      "step": 26998
-    },
-    {
-      "epoch": 2.06426209453906,
-      "grad_norm": 0.0007536399061791599,
-      "learning_rate": 0.00019999789875449508,
-      "loss": 46.0,
-      "step": 26999
-    },
-    {
-      "epoch": 2.0643385515224497,
-      "grad_norm": 0.0012177103199064732,
-      "learning_rate": 0.0001999978985987815,
-      "loss": 46.0,
-      "step": 27000
-    },
-    {
-      "epoch": 2.0644150085058395,
-      "grad_norm": 0.0004639303660951555,
-      "learning_rate": 0.00019999789844306215,
-      "loss": 46.0,
-      "step": 27001
-    },
-    {
-      "epoch": 2.0644914654892292,
-      "grad_norm": 0.0006161812343634665,
-      "learning_rate": 0.00019999789828733703,
-      "loss": 46.0,
-      "step": 27002
-    },
-    {
-      "epoch": 2.064567922472619,
-      "grad_norm": 0.0008269344107247889,
-      "learning_rate": 0.0001999978981316061,
-      "loss": 46.0,
-      "step": 27003
-    },
-    {
-      "epoch": 2.0646443794560088,
-      "grad_norm": 0.0009487119968980551,
-      "learning_rate": 0.00019999789797586943,
-      "loss": 46.0,
-      "step": 27004
-    },
-    {
-      "epoch": 2.064720836439398,
-      "grad_norm": 0.0010664633009582758,
-      "learning_rate": 0.000199997897820127,
-      "loss": 46.0,
-      "step": 27005
-    },
-    {
-      "epoch": 2.064797293422788,
-      "grad_norm": 0.002439675386995077,
-      "learning_rate": 0.0001999978976643788,
-      "loss": 46.0,
-      "step": 27006
-    },
-    {
-      "epoch": 2.0648737504061776,
-      "grad_norm": 0.001499276957474649,
-      "learning_rate": 0.00019999789750862484,
-      "loss": 46.0,
-      "step": 27007
-    },
-    {
-      "epoch": 2.0649502073895674,
-      "grad_norm": 0.0008093096548691392,
-      "learning_rate": 0.00019999789735286507,
-      "loss": 46.0,
-      "step": 27008
-    },
-    {
-      "epoch": 2.065026664372957,
-      "grad_norm": 0.001459756982512772,
-      "learning_rate": 0.00019999789719709956,
-      "loss": 46.0,
-      "step": 27009
-    },
-    {
-      "epoch": 2.065103121356347,
-      "grad_norm": 0.0008315164595842361,
-      "learning_rate": 0.00019999789704132828,
-      "loss": 46.0,
-      "step": 27010
-    },
-    {
-      "epoch": 2.0651795783397366,
-      "grad_norm": 0.000978629570454359,
-      "learning_rate": 0.00019999789688555122,
-      "loss": 46.0,
-      "step": 27011
-    },
-    {
-      "epoch": 2.0652560353231264,
-      "grad_norm": 0.0014649737859144807,
-      "learning_rate": 0.0001999978967297684,
-      "loss": 46.0,
-      "step": 27012
-    },
-    {
-      "epoch": 2.065332492306516,
-      "grad_norm": 0.0019375067204236984,
-      "learning_rate": 0.00019999789657397982,
-      "loss": 46.0,
-      "step": 27013
-    },
-    {
-      "epoch": 2.065408949289906,
-      "grad_norm": 0.0020096905063837767,
-      "learning_rate": 0.00019999789641818544,
-      "loss": 46.0,
-      "step": 27014
-    },
-    {
-      "epoch": 2.0654854062732957,
-      "grad_norm": 0.0005466511356644332,
-      "learning_rate": 0.0001999978962623853,
-      "loss": 46.0,
-      "step": 27015
-    },
-    {
-      "epoch": 2.065561863256685,
-      "grad_norm": 0.0015550549142062664,
-      "learning_rate": 0.0001999978961065794,
-      "loss": 46.0,
-      "step": 27016
-    },
-    {
-      "epoch": 2.0656383202400748,
-      "grad_norm": 0.0024740565568208694,
-      "learning_rate": 0.00019999789595076775,
-      "loss": 46.0,
-      "step": 27017
-    },
-    {
-      "epoch": 2.0657147772234645,
-      "grad_norm": 0.0009245170513167977,
-      "learning_rate": 0.0001999978957949503,
-      "loss": 46.0,
-      "step": 27018
-    },
-    {
-      "epoch": 2.0657912342068543,
-      "grad_norm": 0.004177370108664036,
-      "learning_rate": 0.0001999978956391271,
-      "loss": 46.0,
-      "step": 27019
-    },
-    {
-      "epoch": 2.065867691190244,
-      "grad_norm": 0.0007365798228420317,
-      "learning_rate": 0.0001999978954832981,
-      "loss": 46.0,
-      "step": 27020
-    },
-    {
-      "epoch": 2.065944148173634,
-      "grad_norm": 0.0013109984574839473,
-      "learning_rate": 0.00019999789532746336,
-      "loss": 46.0,
-      "step": 27021
-    },
-    {
-      "epoch": 2.0660206051570236,
-      "grad_norm": 0.0008615059196017683,
-      "learning_rate": 0.00019999789517162283,
-      "loss": 46.0,
-      "step": 27022
-    },
-    {
-      "epoch": 2.0660970621404133,
-      "grad_norm": 0.0005882181576453149,
-      "learning_rate": 0.00019999789501577654,
-      "loss": 46.0,
-      "step": 27023
-    },
-    {
-      "epoch": 2.066173519123803,
-      "grad_norm": 0.0008883198606781662,
-      "learning_rate": 0.00019999789485992449,
-      "loss": 46.0,
-      "step": 27024
-    },
-    {
-      "epoch": 2.066249976107193,
-      "grad_norm": 0.0009025599574670196,
-      "learning_rate": 0.00019999789470406666,
-      "loss": 46.0,
-      "step": 27025
-    },
-    {
-      "epoch": 2.0663264330905826,
-      "grad_norm": 0.0005582623416557908,
-      "learning_rate": 0.00019999789454820305,
-      "loss": 46.0,
-      "step": 27026
-    },
-    {
-      "epoch": 2.066402890073972,
-      "grad_norm": 0.003043276024982333,
-      "learning_rate": 0.0001999978943923337,
-      "loss": 46.0,
-      "step": 27027
-    },
-    {
-      "epoch": 2.0664793470573617,
-      "grad_norm": 0.0008435086929239333,
-      "learning_rate": 0.00019999789423645855,
-      "loss": 46.0,
-      "step": 27028
-    },
-    {
-      "epoch": 2.0665558040407515,
-      "grad_norm": 0.0020478428341448307,
-      "learning_rate": 0.00019999789408057766,
-      "loss": 46.0,
-      "step": 27029
-    },
-    {
-      "epoch": 2.066632261024141,
-      "grad_norm": 0.0008808975107967854,
-      "learning_rate": 0.000199997893924691,
-      "loss": 46.0,
-      "step": 27030
-    },
-    {
-      "epoch": 2.066708718007531,
-      "grad_norm": 0.0030391705222427845,
-      "learning_rate": 0.00019999789376879857,
-      "loss": 46.0,
-      "step": 27031
-    },
-    {
-      "epoch": 2.0667851749909207,
-      "grad_norm": 0.0026454650796949863,
-      "learning_rate": 0.00019999789361290033,
-      "loss": 46.0,
-      "step": 27032
-    },
-    {
-      "epoch": 2.0668616319743105,
-      "grad_norm": 0.0004930328577756882,
-      "learning_rate": 0.00019999789345699636,
-      "loss": 46.0,
-      "step": 27033
-    },
-    {
-      "epoch": 2.0669380889577003,
-      "grad_norm": 0.0008861568639986217,
-      "learning_rate": 0.0001999978933010866,
-      "loss": 46.0,
-      "step": 27034
-    },
-    {
-      "epoch": 2.06701454594109,
-      "grad_norm": 0.0018538491567596793,
-      "learning_rate": 0.0001999978931451711,
-      "loss": 46.0,
-      "step": 27035
-    },
-    {
-      "epoch": 2.06709100292448,
-      "grad_norm": 0.0005827159038744867,
-      "learning_rate": 0.0001999978929892498,
-      "loss": 46.0,
-      "step": 27036
-    },
-    {
-      "epoch": 2.0671674599078695,
-      "grad_norm": 0.0014680143212899566,
-      "learning_rate": 0.00019999789283332273,
-      "loss": 46.0,
-      "step": 27037
-    },
-    {
-      "epoch": 2.067243916891259,
-      "grad_norm": 0.0009790032636374235,
-      "learning_rate": 0.0001999978926773899,
-      "loss": 46.0,
-      "step": 27038
-    },
-    {
-      "epoch": 2.0673203738746486,
-      "grad_norm": 0.0013785073533654213,
-      "learning_rate": 0.00019999789252145132,
-      "loss": 46.0,
-      "step": 27039
-    },
-    {
-      "epoch": 2.0673968308580384,
-      "grad_norm": 0.0017776767490431666,
-      "learning_rate": 0.00019999789236550694,
-      "loss": 46.0,
-      "step": 27040
-    },
-    {
-      "epoch": 2.067473287841428,
-      "grad_norm": 0.00031912574195303023,
-      "learning_rate": 0.00019999789220955682,
-      "loss": 46.0,
-      "step": 27041
-    },
-    {
-      "epoch": 2.067549744824818,
-      "grad_norm": 0.0017054181080311537,
-      "learning_rate": 0.00019999789205360093,
-      "loss": 46.0,
-      "step": 27042
-    },
-    {
-      "epoch": 2.0676262018082077,
-      "grad_norm": 0.0004103182291146368,
-      "learning_rate": 0.00019999789189763926,
-      "loss": 46.0,
-      "step": 27043
-    },
-    {
-      "epoch": 2.0677026587915974,
-      "grad_norm": 0.002302873879671097,
-      "learning_rate": 0.0001999978917416718,
-      "loss": 46.0,
-      "step": 27044
-    },
-    {
-      "epoch": 2.067779115774987,
-      "grad_norm": 0.0009202692890539765,
-      "learning_rate": 0.0001999978915856986,
-      "loss": 46.0,
-      "step": 27045
-    },
-    {
-      "epoch": 2.067855572758377,
-      "grad_norm": 0.0006078872247599065,
-      "learning_rate": 0.0001999978914297196,
-      "loss": 46.0,
-      "step": 27046
-    },
-    {
-      "epoch": 2.0679320297417667,
-      "grad_norm": 0.0006707943975925446,
-      "learning_rate": 0.00019999789127373485,
-      "loss": 46.0,
-      "step": 27047
-    },
-    {
-      "epoch": 2.0680084867251565,
-      "grad_norm": 0.001298778923228383,
-      "learning_rate": 0.00019999789111774434,
-      "loss": 46.0,
-      "step": 27048
-    },
-    {
-      "epoch": 2.068084943708546,
-      "grad_norm": 0.0007975633488968015,
-      "learning_rate": 0.00019999789096174805,
-      "loss": 46.0,
-      "step": 27049
-    },
-    {
-      "epoch": 2.0681614006919355,
-      "grad_norm": 0.0030256900936365128,
-      "learning_rate": 0.000199997890805746,
-      "loss": 46.0,
-      "step": 27050
-    },
-    {
-      "epoch": 2.0682378576753253,
-      "grad_norm": 0.0021144249476492405,
-      "learning_rate": 0.00019999789064973817,
-      "loss": 46.0,
-      "step": 27051
-    },
-    {
-      "epoch": 2.068314314658715,
-      "grad_norm": 0.0005385276162996888,
-      "learning_rate": 0.00019999789049372457,
-      "loss": 46.0,
-      "step": 27052
-    },
-    {
-      "epoch": 2.068390771642105,
-      "grad_norm": 0.0002385624684393406,
-      "learning_rate": 0.0001999978903377052,
-      "loss": 46.0,
-      "step": 27053
-    },
-    {
-      "epoch": 2.0684672286254946,
-      "grad_norm": 0.0005000167293474078,
-      "learning_rate": 0.00019999789018168004,
-      "loss": 46.0,
-      "step": 27054
-    },
-    {
-      "epoch": 2.0685436856088844,
-      "grad_norm": 0.0010154960909858346,
-      "learning_rate": 0.00019999789002564914,
-      "loss": 46.0,
-      "step": 27055
-    },
-    {
-      "epoch": 2.068620142592274,
-      "grad_norm": 0.002301678992807865,
-      "learning_rate": 0.00019999788986961248,
-      "loss": 46.0,
-      "step": 27056
-    },
-    {
-      "epoch": 2.068696599575664,
-      "grad_norm": 0.0034767414908856153,
-      "learning_rate": 0.00019999788971357003,
-      "loss": 46.0,
-      "step": 27057
-    },
-    {
-      "epoch": 2.0687730565590536,
-      "grad_norm": 0.00034108993713743985,
-      "learning_rate": 0.00019999788955752182,
-      "loss": 46.0,
-      "step": 27058
-    },
-    {
-      "epoch": 2.0688495135424434,
-      "grad_norm": 0.0010133845498785377,
-      "learning_rate": 0.00019999788940146786,
-      "loss": 46.0,
-      "step": 27059
-    },
-    {
-      "epoch": 2.0689259705258327,
-      "grad_norm": 0.0009166516829282045,
-      "learning_rate": 0.00019999788924540807,
-      "loss": 46.0,
-      "step": 27060
-    },
-    {
-      "epoch": 2.0690024275092225,
-      "grad_norm": 0.0005939473048783839,
-      "learning_rate": 0.00019999788908934256,
-      "loss": 46.0,
-      "step": 27061
-    },
-    {
-      "epoch": 2.0690788844926122,
-      "grad_norm": 0.0009271614253520966,
-      "learning_rate": 0.00019999788893327127,
-      "loss": 46.0,
-      "step": 27062
-    },
-    {
-      "epoch": 2.069155341476002,
-      "grad_norm": 0.0013534929603338242,
-      "learning_rate": 0.00019999788877719422,
-      "loss": 46.0,
-      "step": 27063
-    },
-    {
-      "epoch": 2.0692317984593918,
-      "grad_norm": 0.0010275986278429627,
-      "learning_rate": 0.00019999788862111136,
-      "loss": 46.0,
-      "step": 27064
-    },
-    {
-      "epoch": 2.0693082554427815,
-      "grad_norm": 0.0010461921337991953,
-      "learning_rate": 0.0001999978884650228,
-      "loss": 46.0,
-      "step": 27065
-    },
-    {
-      "epoch": 2.0693847124261713,
-      "grad_norm": 0.0014470914611592889,
-      "learning_rate": 0.0001999978883089284,
-      "loss": 46.0,
-      "step": 27066
-    },
-    {
-      "epoch": 2.069461169409561,
-      "grad_norm": 0.0006521940813399851,
-      "learning_rate": 0.00019999788815282826,
-      "loss": 46.0,
-      "step": 27067
-    },
-    {
-      "epoch": 2.069537626392951,
-      "grad_norm": 0.0011575151002034545,
-      "learning_rate": 0.00019999788799672237,
-      "loss": 46.0,
-      "step": 27068
-    },
-    {
-      "epoch": 2.0696140833763406,
-      "grad_norm": 0.0009787343442440033,
-      "learning_rate": 0.0001999978878406107,
-      "loss": 46.0,
-      "step": 27069
-    },
-    {
-      "epoch": 2.06969054035973,
-      "grad_norm": 0.0006229884456843138,
-      "learning_rate": 0.00019999788768449323,
-      "loss": 46.0,
-      "step": 27070
-    },
-    {
-      "epoch": 2.0697669973431196,
-      "grad_norm": 0.005218871869146824,
-      "learning_rate": 0.00019999788752837004,
-      "loss": 46.0,
-      "step": 27071
-    },
-    {
-      "epoch": 2.0698434543265094,
-      "grad_norm": 0.011910800822079182,
-      "learning_rate": 0.00019999788737224105,
-      "loss": 46.0,
-      "step": 27072
-    },
-    {
-      "epoch": 2.069919911309899,
-      "grad_norm": 0.0005000763921998441,
-      "learning_rate": 0.00019999788721610632,
-      "loss": 46.0,
-      "step": 27073
-    },
-    {
-      "epoch": 2.069996368293289,
-      "grad_norm": 0.0006613932782784104,
-      "learning_rate": 0.00019999788705996578,
-      "loss": 46.0,
-      "step": 27074
-    },
-    {
-      "epoch": 2.0700728252766787,
-      "grad_norm": 0.0008409673464484513,
-      "learning_rate": 0.0001999978869038195,
-      "loss": 46.0,
-      "step": 27075
-    },
-    {
-      "epoch": 2.0701492822600684,
-      "grad_norm": 0.00089884246699512,
-      "learning_rate": 0.00019999788674766742,
-      "loss": 46.0,
-      "step": 27076
-    },
-    {
-      "epoch": 2.070225739243458,
-      "grad_norm": 0.002012914279475808,
-      "learning_rate": 0.0001999978865915096,
-      "loss": 46.0,
-      "step": 27077
-    },
-    {
-      "epoch": 2.070302196226848,
-      "grad_norm": 0.001167898066341877,
-      "learning_rate": 0.000199997886435346,
-      "loss": 46.0,
-      "step": 27078
-    },
-    {
-      "epoch": 2.0703786532102377,
-      "grad_norm": 0.0015673909801989794,
-      "learning_rate": 0.0001999978862791766,
-      "loss": 46.0,
-      "step": 27079
-    },
-    {
-      "epoch": 2.0704551101936275,
-      "grad_norm": 0.023026851937174797,
-      "learning_rate": 0.0001999978861230015,
-      "loss": 46.0,
-      "step": 27080
-    },
-    {
-      "epoch": 2.0705315671770173,
-      "grad_norm": 0.00261838105507195,
-      "learning_rate": 0.00019999788596682057,
-      "loss": 46.0,
-      "step": 27081
-    },
-    {
-      "epoch": 2.0706080241604066,
-      "grad_norm": 0.0005020395037718117,
-      "learning_rate": 0.0001999978858106339,
-      "loss": 46.0,
-      "step": 27082
-    },
-    {
-      "epoch": 2.0706844811437963,
-      "grad_norm": 0.002186959143728018,
-      "learning_rate": 0.00019999788565444143,
-      "loss": 46.0,
-      "step": 27083
-    },
-    {
-      "epoch": 2.070760938127186,
-      "grad_norm": 0.000831065874081105,
-      "learning_rate": 0.00019999788549824321,
-      "loss": 46.0,
-      "step": 27084
-    },
-    {
-      "epoch": 2.070837395110576,
-      "grad_norm": 0.000870771415065974,
-      "learning_rate": 0.00019999788534203923,
-      "loss": 46.0,
-      "step": 27085
-    },
-    {
-      "epoch": 2.0709138520939656,
-      "grad_norm": 0.00061557482695207,
-      "learning_rate": 0.00019999788518582946,
-      "loss": 46.0,
-      "step": 27086
-    },
-    {
-      "epoch": 2.0709903090773554,
-      "grad_norm": 0.0012566782534122467,
-      "learning_rate": 0.00019999788502961396,
-      "loss": 46.0,
-      "step": 27087
-    },
-    {
-      "epoch": 2.071066766060745,
-      "grad_norm": 0.0012375015066936612,
-      "learning_rate": 0.00019999788487339265,
-      "loss": 46.0,
-      "step": 27088
-    },
-    {
-      "epoch": 2.071143223044135,
-      "grad_norm": 0.0006211168947629631,
-      "learning_rate": 0.00019999788471716562,
-      "loss": 46.0,
-      "step": 27089
-    },
-    {
-      "epoch": 2.0712196800275247,
-      "grad_norm": 0.0004771646053995937,
-      "learning_rate": 0.00019999788456093276,
-      "loss": 46.0,
-      "step": 27090
-    },
-    {
-      "epoch": 2.0712961370109144,
-      "grad_norm": 0.002767386846244335,
-      "learning_rate": 0.0001999978844046942,
-      "loss": 46.0,
-      "step": 27091
-    },
-    {
-      "epoch": 2.0713725939943037,
-      "grad_norm": 0.003315454814583063,
-      "learning_rate": 0.0001999978842484498,
-      "loss": 46.0,
-      "step": 27092
-    },
-    {
-      "epoch": 2.0714490509776935,
-      "grad_norm": 0.0018309837905690074,
-      "learning_rate": 0.00019999788409219967,
-      "loss": 46.0,
-      "step": 27093
-    },
-    {
-      "epoch": 2.0715255079610833,
-      "grad_norm": 0.0011916599469259381,
-      "learning_rate": 0.00019999788393594374,
-      "loss": 46.0,
-      "step": 27094
-    },
-    {
-      "epoch": 2.071601964944473,
-      "grad_norm": 0.000992544344626367,
-      "learning_rate": 0.00019999788377968208,
-      "loss": 46.0,
-      "step": 27095
-    },
-    {
-      "epoch": 2.071678421927863,
-      "grad_norm": 0.0011875139316543937,
-      "learning_rate": 0.0001999978836234146,
-      "loss": 46.0,
-      "step": 27096
-    },
-    {
-      "epoch": 2.0717548789112525,
-      "grad_norm": 0.0010922644287347794,
-      "learning_rate": 0.0001999978834671414,
-      "loss": 46.0,
-      "step": 27097
-    },
-    {
-      "epoch": 2.0718313358946423,
-      "grad_norm": 0.002796332584694028,
-      "learning_rate": 0.0001999978833108624,
-      "loss": 46.0,
-      "step": 27098
-    },
-    {
-      "epoch": 2.071907792878032,
-      "grad_norm": 0.0022023599594831467,
-      "learning_rate": 0.00019999788315457767,
-      "loss": 46.0,
-      "step": 27099
-    },
-    {
-      "epoch": 2.071984249861422,
-      "grad_norm": 0.0005100789712741971,
-      "learning_rate": 0.00019999788299828714,
-      "loss": 46.0,
-      "step": 27100
-    },
-    {
-      "epoch": 2.0720607068448116,
-      "grad_norm": 0.0017492339247837663,
-      "learning_rate": 0.00019999788284199083,
-      "loss": 46.0,
-      "step": 27101
-    },
-    {
-      "epoch": 2.0721371638282013,
-      "grad_norm": 0.0030357837677001953,
-      "learning_rate": 0.00019999788268568877,
-      "loss": 46.0,
-      "step": 27102
-    },
-    {
-      "epoch": 2.072213620811591,
-      "grad_norm": 0.0005458348896354437,
-      "learning_rate": 0.00019999788252938094,
-      "loss": 46.0,
-      "step": 27103
-    },
-    {
-      "epoch": 2.0722900777949804,
-      "grad_norm": 0.001348218065686524,
-      "learning_rate": 0.00019999788237306734,
-      "loss": 46.0,
-      "step": 27104
-    },
-    {
-      "epoch": 2.07236653477837,
-      "grad_norm": 0.0007464727968908846,
-      "learning_rate": 0.00019999788221674797,
-      "loss": 46.0,
-      "step": 27105
-    },
-    {
-      "epoch": 2.07244299176176,
-      "grad_norm": 0.00191450584679842,
-      "learning_rate": 0.00019999788206042282,
-      "loss": 46.0,
-      "step": 27106
-    },
-    {
-      "epoch": 2.0725194487451497,
-      "grad_norm": 0.0019547617994248867,
-      "learning_rate": 0.00019999788190409193,
-      "loss": 46.0,
-      "step": 27107
-    },
-    {
-      "epoch": 2.0725959057285395,
-      "grad_norm": 0.000526728224940598,
-      "learning_rate": 0.00019999788174775526,
-      "loss": 46.0,
-      "step": 27108
-    },
-    {
-      "epoch": 2.0726723627119292,
-      "grad_norm": 0.000797897286247462,
-      "learning_rate": 0.0001999978815914128,
-      "loss": 46.0,
-      "step": 27109
-    },
-    {
-      "epoch": 2.072748819695319,
-      "grad_norm": 0.001858215662650764,
-      "learning_rate": 0.00019999788143506457,
-      "loss": 46.0,
-      "step": 27110
-    },
-    {
-      "epoch": 2.0728252766787088,
-      "grad_norm": 0.003177137114107609,
-      "learning_rate": 0.0001999978812787106,
-      "loss": 46.0,
-      "step": 27111
-    },
-    {
-      "epoch": 2.0729017336620985,
-      "grad_norm": 0.0003472694952506572,
-      "learning_rate": 0.0001999978811223508,
-      "loss": 46.0,
-      "step": 27112
-    },
-    {
-      "epoch": 2.0729781906454883,
-      "grad_norm": 0.0010311231017112732,
-      "learning_rate": 0.0001999978809659853,
-      "loss": 46.0,
-      "step": 27113
-    },
-    {
-      "epoch": 2.0730546476288776,
-      "grad_norm": 0.0006290294695645571,
-      "learning_rate": 0.00019999788080961398,
-      "loss": 46.0,
-      "step": 27114
-    },
-    {
-      "epoch": 2.0731311046122674,
-      "grad_norm": 0.0008104062872007489,
-      "learning_rate": 0.00019999788065323693,
-      "loss": 46.0,
-      "step": 27115
-    },
-    {
-      "epoch": 2.073207561595657,
-      "grad_norm": 0.000582873763050884,
-      "learning_rate": 0.00019999788049685407,
-      "loss": 46.0,
-      "step": 27116
-    },
-    {
-      "epoch": 2.073284018579047,
-      "grad_norm": 0.0006381890270859003,
-      "learning_rate": 0.0001999978803404655,
-      "loss": 46.0,
-      "step": 27117
-    },
-    {
-      "epoch": 2.0733604755624366,
-      "grad_norm": 0.0008024231647141278,
-      "learning_rate": 0.00019999788018407113,
-      "loss": 46.0,
-      "step": 27118
-    },
-    {
-      "epoch": 2.0734369325458264,
-      "grad_norm": 0.0009202121291309595,
-      "learning_rate": 0.00019999788002767098,
-      "loss": 46.0,
-      "step": 27119
-    },
-    {
-      "epoch": 2.073513389529216,
-      "grad_norm": 0.000793017097748816,
-      "learning_rate": 0.00019999787987126506,
-      "loss": 46.0,
-      "step": 27120
-    },
-    {
-      "epoch": 2.073589846512606,
-      "grad_norm": 0.0007888094405643642,
-      "learning_rate": 0.0001999978797148534,
-      "loss": 46.0,
-      "step": 27121
-    },
-    {
-      "epoch": 2.0736663034959957,
-      "grad_norm": 0.00149096071254462,
-      "learning_rate": 0.00019999787955843595,
-      "loss": 46.0,
-      "step": 27122
-    },
-    {
-      "epoch": 2.0737427604793854,
-      "grad_norm": 0.0005638212896883488,
-      "learning_rate": 0.0001999978794020127,
-      "loss": 46.0,
-      "step": 27123
-    },
-    {
-      "epoch": 2.073819217462775,
-      "grad_norm": 0.0013729806523770094,
-      "learning_rate": 0.00019999787924558372,
-      "loss": 46.0,
-      "step": 27124
-    },
-    {
-      "epoch": 2.0738956744461645,
-      "grad_norm": 0.00045726681128144264,
-      "learning_rate": 0.00019999787908914896,
-      "loss": 46.0,
-      "step": 27125
-    },
-    {
-      "epoch": 2.0739721314295543,
-      "grad_norm": 0.0016534518217667937,
-      "learning_rate": 0.00019999787893270842,
-      "loss": 46.0,
-      "step": 27126
-    },
-    {
-      "epoch": 2.074048588412944,
-      "grad_norm": 0.0011107750469818711,
-      "learning_rate": 0.00019999787877626212,
-      "loss": 46.0,
-      "step": 27127
-    },
-    {
-      "epoch": 2.074125045396334,
-      "grad_norm": 0.0006969883688725531,
-      "learning_rate": 0.00019999787861981006,
-      "loss": 46.0,
-      "step": 27128
-    },
-    {
-      "epoch": 2.0742015023797236,
-      "grad_norm": 0.0012100323801860213,
-      "learning_rate": 0.00019999787846335223,
-      "loss": 46.0,
-      "step": 27129
-    },
-    {
-      "epoch": 2.0742779593631133,
-      "grad_norm": 0.0011009123409166932,
-      "learning_rate": 0.00019999787830688863,
-      "loss": 46.0,
-      "step": 27130
-    },
-    {
-      "epoch": 2.074354416346503,
-      "grad_norm": 0.003460992593318224,
-      "learning_rate": 0.00019999787815041926,
-      "loss": 46.0,
-      "step": 27131
-    },
-    {
-      "epoch": 2.074430873329893,
-      "grad_norm": 0.0018930698279291391,
-      "learning_rate": 0.0001999978779939441,
-      "loss": 46.0,
-      "step": 27132
-    },
-    {
-      "epoch": 2.0745073303132826,
-      "grad_norm": 0.0025584097020328045,
-      "learning_rate": 0.0001999978778374632,
-      "loss": 46.0,
-      "step": 27133
-    },
-    {
-      "epoch": 2.0745837872966724,
-      "grad_norm": 0.0021529991645365953,
-      "learning_rate": 0.0001999978776809765,
-      "loss": 46.0,
-      "step": 27134
-    },
-    {
-      "epoch": 2.074660244280062,
-      "grad_norm": 0.001232546172104776,
-      "learning_rate": 0.00019999787752448406,
-      "loss": 46.0,
-      "step": 27135
-    },
-    {
-      "epoch": 2.0747367012634514,
-      "grad_norm": 0.0013623720733448863,
-      "learning_rate": 0.00019999787736798584,
-      "loss": 46.0,
-      "step": 27136
-    },
-    {
-      "epoch": 2.074813158246841,
-      "grad_norm": 0.001318274182267487,
-      "learning_rate": 0.00019999787721148186,
-      "loss": 46.0,
-      "step": 27137
-    },
-    {
-      "epoch": 2.074889615230231,
-      "grad_norm": 0.002330901101231575,
-      "learning_rate": 0.0001999978770549721,
-      "loss": 46.0,
-      "step": 27138
-    },
-    {
-      "epoch": 2.0749660722136207,
-      "grad_norm": 0.002453487366437912,
-      "learning_rate": 0.00019999787689845656,
-      "loss": 46.0,
-      "step": 27139
-    },
-    {
-      "epoch": 2.0750425291970105,
-      "grad_norm": 0.0031309437472373247,
-      "learning_rate": 0.00019999787674193525,
-      "loss": 46.0,
-      "step": 27140
-    },
-    {
-      "epoch": 2.0751189861804002,
-      "grad_norm": 0.001266809063963592,
-      "learning_rate": 0.00019999787658540817,
-      "loss": 46.0,
-      "step": 27141
-    },
-    {
-      "epoch": 2.07519544316379,
-      "grad_norm": 0.0007018725736998022,
-      "learning_rate": 0.00019999787642887532,
-      "loss": 46.0,
-      "step": 27142
-    },
-    {
-      "epoch": 2.0752719001471798,
-      "grad_norm": 0.0005419120425358415,
-      "learning_rate": 0.00019999787627233672,
-      "loss": 46.0,
-      "step": 27143
-    },
-    {
-      "epoch": 2.0753483571305695,
-      "grad_norm": 0.03149529919028282,
-      "learning_rate": 0.00019999787611579234,
-      "loss": 46.0,
-      "step": 27144
-    },
-    {
-      "epoch": 2.0754248141139593,
-      "grad_norm": 0.000695561699103564,
-      "learning_rate": 0.00019999787595924222,
-      "loss": 46.0,
-      "step": 27145
-    },
-    {
-      "epoch": 2.075501271097349,
-      "grad_norm": 0.00033221839112229645,
-      "learning_rate": 0.00019999787580268633,
-      "loss": 46.0,
-      "step": 27146
-    },
-    {
-      "epoch": 2.0755777280807384,
-      "grad_norm": 0.0008184429025277495,
-      "learning_rate": 0.0001999978756461246,
-      "loss": 46.0,
-      "step": 27147
-    },
-    {
-      "epoch": 2.075654185064128,
-      "grad_norm": 0.0005841483362019062,
-      "learning_rate": 0.00019999787548955714,
-      "loss": 46.0,
-      "step": 27148
-    },
-    {
-      "epoch": 2.075730642047518,
-      "grad_norm": 0.0008841967792250216,
-      "learning_rate": 0.00019999787533298396,
-      "loss": 46.0,
-      "step": 27149
-    },
-    {
-      "epoch": 2.0758070990309077,
-      "grad_norm": 0.0010597413638606668,
-      "learning_rate": 0.00019999787517640494,
-      "loss": 46.0,
-      "step": 27150
-    },
-    {
-      "epoch": 2.0758835560142974,
-      "grad_norm": 0.0016691063065081835,
-      "learning_rate": 0.00019999787501982018,
-      "loss": 46.0,
-      "step": 27151
-    },
-    {
-      "epoch": 2.075960012997687,
-      "grad_norm": 0.003026694757863879,
-      "learning_rate": 0.00019999787486322967,
-      "loss": 46.0,
-      "step": 27152
-    },
-    {
-      "epoch": 2.076036469981077,
-      "grad_norm": 0.0014178812270984054,
-      "learning_rate": 0.00019999787470663337,
-      "loss": 46.0,
-      "step": 27153
-    },
-    {
-      "epoch": 2.0761129269644667,
-      "grad_norm": 0.004958159755915403,
-      "learning_rate": 0.00019999787455003126,
-      "loss": 46.0,
-      "step": 27154
-    },
-    {
-      "epoch": 2.0761893839478565,
-      "grad_norm": 0.0016772547969594598,
-      "learning_rate": 0.00019999787439342343,
-      "loss": 46.0,
-      "step": 27155
-    },
-    {
-      "epoch": 2.076265840931246,
-      "grad_norm": 0.0008967339526861906,
-      "learning_rate": 0.00019999787423680983,
-      "loss": 46.0,
-      "step": 27156
-    },
-    {
-      "epoch": 2.076342297914636,
-      "grad_norm": 0.0015132429543882608,
-      "learning_rate": 0.00019999787408019046,
-      "loss": 46.0,
-      "step": 27157
-    },
-    {
-      "epoch": 2.0764187548980253,
-      "grad_norm": 0.003755326149985194,
-      "learning_rate": 0.00019999787392356534,
-      "loss": 46.0,
-      "step": 27158
-    },
-    {
-      "epoch": 2.076495211881415,
-      "grad_norm": 0.0023535825312137604,
-      "learning_rate": 0.0001999978737669344,
-      "loss": 46.0,
-      "step": 27159
-    },
-    {
-      "epoch": 2.076571668864805,
-      "grad_norm": 0.0011980890994891524,
-      "learning_rate": 0.00019999787361029773,
-      "loss": 46.0,
-      "step": 27160
-    },
-    {
-      "epoch": 2.0766481258481946,
-      "grad_norm": 0.0011827265843749046,
-      "learning_rate": 0.00019999787345365526,
-      "loss": 46.0,
-      "step": 27161
-    },
-    {
-      "epoch": 2.0767245828315843,
-      "grad_norm": 0.00028965191449970007,
-      "learning_rate": 0.00019999787329700705,
-      "loss": 46.0,
-      "step": 27162
-    },
-    {
-      "epoch": 2.076801039814974,
-      "grad_norm": 0.0016712151700630784,
-      "learning_rate": 0.00019999787314035303,
-      "loss": 46.0,
-      "step": 27163
-    },
-    {
-      "epoch": 2.076877496798364,
-      "grad_norm": 0.0021500373259186745,
-      "learning_rate": 0.00019999787298369327,
-      "loss": 46.0,
-      "step": 27164
-    },
-    {
-      "epoch": 2.0769539537817536,
-      "grad_norm": 0.0015568399103358388,
-      "learning_rate": 0.00019999787282702777,
-      "loss": 46.0,
-      "step": 27165
-    },
-    {
-      "epoch": 2.0770304107651434,
-      "grad_norm": 0.0004950611037202179,
-      "learning_rate": 0.00019999787267035646,
-      "loss": 46.0,
-      "step": 27166
-    },
-    {
-      "epoch": 2.077106867748533,
-      "grad_norm": 0.0026304114144295454,
-      "learning_rate": 0.00019999787251367938,
-      "loss": 46.0,
-      "step": 27167
-    },
-    {
-      "epoch": 2.077183324731923,
-      "grad_norm": 0.0010290222708135843,
-      "learning_rate": 0.00019999787235699655,
-      "loss": 46.0,
-      "step": 27168
-    },
-    {
-      "epoch": 2.0772597817153122,
-      "grad_norm": 0.0009328462765552104,
-      "learning_rate": 0.00019999787220030793,
-      "loss": 46.0,
-      "step": 27169
-    },
-    {
-      "epoch": 2.077336238698702,
-      "grad_norm": 0.0008161712903529406,
-      "learning_rate": 0.00019999787204361352,
-      "loss": 46.0,
-      "step": 27170
-    },
-    {
-      "epoch": 2.0774126956820917,
-      "grad_norm": 0.0015659695491194725,
-      "learning_rate": 0.0001999978718869134,
-      "loss": 46.0,
-      "step": 27171
-    },
-    {
-      "epoch": 2.0774891526654815,
-      "grad_norm": 0.0008939250255934894,
-      "learning_rate": 0.00019999787173020749,
-      "loss": 46.0,
-      "step": 27172
-    },
-    {
-      "epoch": 2.0775656096488713,
-      "grad_norm": 0.0009759397944435477,
-      "learning_rate": 0.0001999978715734958,
-      "loss": 46.0,
-      "step": 27173
-    },
-    {
-      "epoch": 2.077642066632261,
-      "grad_norm": 0.0008491888293065131,
-      "learning_rate": 0.00019999787141677833,
-      "loss": 46.0,
-      "step": 27174
-    },
-    {
-      "epoch": 2.077718523615651,
-      "grad_norm": 0.0013619356323033571,
-      "learning_rate": 0.0001999978712600551,
-      "loss": 46.0,
-      "step": 27175
-    },
-    {
-      "epoch": 2.0777949805990406,
-      "grad_norm": 0.000696946750395,
-      "learning_rate": 0.00019999787110332613,
-      "loss": 46.0,
-      "step": 27176
-    },
-    {
-      "epoch": 2.0778714375824303,
-      "grad_norm": 0.0015801778063178062,
-      "learning_rate": 0.00019999787094659134,
-      "loss": 46.0,
-      "step": 27177
-    },
-    {
-      "epoch": 2.07794789456582,
-      "grad_norm": 0.0004969353904016316,
-      "learning_rate": 0.0001999978707898508,
-      "loss": 46.0,
-      "step": 27178
-    },
-    {
-      "epoch": 2.07802435154921,
-      "grad_norm": 0.001622940762899816,
-      "learning_rate": 0.0001999978706331045,
-      "loss": 46.0,
-      "step": 27179
-    },
-    {
-      "epoch": 2.078100808532599,
-      "grad_norm": 0.0004925756948068738,
-      "learning_rate": 0.00019999787047635245,
-      "loss": 46.0,
-      "step": 27180
-    },
-    {
-      "epoch": 2.078177265515989,
-      "grad_norm": 0.001371089369058609,
-      "learning_rate": 0.0001999978703195946,
-      "loss": 46.0,
-      "step": 27181
-    },
-    {
-      "epoch": 2.0782537224993787,
-      "grad_norm": 0.0013874656287953258,
-      "learning_rate": 0.00019999787016283097,
-      "loss": 46.0,
-      "step": 27182
-    },
-    {
-      "epoch": 2.0783301794827684,
-      "grad_norm": 0.0007348840590566397,
-      "learning_rate": 0.0001999978700060616,
-      "loss": 46.0,
-      "step": 27183
-    },
-    {
-      "epoch": 2.078406636466158,
-      "grad_norm": 0.0009697225177660584,
-      "learning_rate": 0.00019999786984928645,
-      "loss": 46.0,
-      "step": 27184
-    },
-    {
-      "epoch": 2.078483093449548,
-      "grad_norm": 0.0007268933695740998,
-      "learning_rate": 0.00019999786969250556,
-      "loss": 46.0,
-      "step": 27185
-    },
-    {
-      "epoch": 2.0785595504329377,
-      "grad_norm": 0.0007455849554389715,
-      "learning_rate": 0.00019999786953571884,
-      "loss": 46.0,
-      "step": 27186
-    },
-    {
-      "epoch": 2.0786360074163275,
-      "grad_norm": 0.0007582865655422211,
-      "learning_rate": 0.0001999978693789264,
-      "loss": 46.0,
-      "step": 27187
-    },
-    {
-      "epoch": 2.0787124643997172,
-      "grad_norm": 0.0033378128428012133,
-      "learning_rate": 0.00019999786922212818,
-      "loss": 46.0,
-      "step": 27188
-    },
-    {
-      "epoch": 2.078788921383107,
-      "grad_norm": 0.0017556612147018313,
-      "learning_rate": 0.00019999786906532417,
-      "loss": 46.0,
-      "step": 27189
-    },
-    {
-      "epoch": 2.0788653783664968,
-      "grad_norm": 0.0025836050044745207,
-      "learning_rate": 0.0001999978689085144,
-      "loss": 46.0,
-      "step": 27190
-    },
-    {
-      "epoch": 2.078941835349886,
-      "grad_norm": 0.0014807026600465178,
-      "learning_rate": 0.00019999786875169888,
-      "loss": 46.0,
-      "step": 27191
-    },
-    {
-      "epoch": 2.079018292333276,
-      "grad_norm": 0.0008699865429662168,
-      "learning_rate": 0.00019999786859487757,
-      "loss": 46.0,
-      "step": 27192
-    },
-    {
-      "epoch": 2.0790947493166656,
-      "grad_norm": 0.005615171976387501,
-      "learning_rate": 0.0001999978684380505,
-      "loss": 46.0,
-      "step": 27193
-    },
-    {
-      "epoch": 2.0791712063000554,
-      "grad_norm": 0.005202721804380417,
-      "learning_rate": 0.00019999786828121764,
-      "loss": 46.0,
-      "step": 27194
-    },
-    {
-      "epoch": 2.079247663283445,
-      "grad_norm": 0.00414247065782547,
-      "learning_rate": 0.00019999786812437904,
-      "loss": 46.0,
-      "step": 27195
-    },
-    {
-      "epoch": 2.079324120266835,
-      "grad_norm": 0.0024178854655474424,
-      "learning_rate": 0.00019999786796753467,
-      "loss": 46.0,
-      "step": 27196
-    },
-    {
-      "epoch": 2.0794005772502246,
-      "grad_norm": 0.0006391534116119146,
-      "learning_rate": 0.0001999978678106845,
-      "loss": 46.0,
-      "step": 27197
-    },
-    {
-      "epoch": 2.0794770342336144,
-      "grad_norm": 0.0013488776748999953,
-      "learning_rate": 0.0001999978676538286,
-      "loss": 46.0,
-      "step": 27198
-    },
-    {
-      "epoch": 2.079553491217004,
-      "grad_norm": 0.0010994477197527885,
-      "learning_rate": 0.00019999786749696688,
-      "loss": 46.0,
-      "step": 27199
-    },
-    {
-      "epoch": 2.079629948200394,
-      "grad_norm": 0.0019452725537121296,
-      "learning_rate": 0.00019999786734009944,
-      "loss": 46.0,
-      "step": 27200
-    },
-    {
-      "epoch": 2.0797064051837832,
-      "grad_norm": 0.002070052782073617,
-      "learning_rate": 0.0001999978671832262,
-      "loss": 46.0,
-      "step": 27201
-    },
-    {
-      "epoch": 2.079782862167173,
-      "grad_norm": 0.0009187899995595217,
-      "learning_rate": 0.0001999978670263472,
-      "loss": 46.0,
-      "step": 27202
-    },
-    {
-      "epoch": 2.0798593191505628,
-      "grad_norm": 0.0007883846410550177,
-      "learning_rate": 0.00019999786686946246,
-      "loss": 46.0,
-      "step": 27203
-    },
-    {
-      "epoch": 2.0799357761339525,
-      "grad_norm": 0.0005143506568856537,
-      "learning_rate": 0.00019999786671257193,
-      "loss": 46.0,
-      "step": 27204
-    },
-    {
-      "epoch": 2.0800122331173423,
-      "grad_norm": 0.0021083930041640997,
-      "learning_rate": 0.0001999978665556756,
-      "loss": 46.0,
-      "step": 27205
-    },
-    {
-      "epoch": 2.080088690100732,
-      "grad_norm": 0.000846718205139041,
-      "learning_rate": 0.00019999786639877351,
-      "loss": 46.0,
-      "step": 27206
-    },
-    {
-      "epoch": 2.080165147084122,
-      "grad_norm": 0.0009303530096076429,
-      "learning_rate": 0.0001999978662418657,
-      "loss": 46.0,
-      "step": 27207
-    },
-    {
-      "epoch": 2.0802416040675116,
-      "grad_norm": 0.001431467360816896,
-      "learning_rate": 0.00019999786608495206,
-      "loss": 46.0,
-      "step": 27208
-    },
-    {
-      "epoch": 2.0803180610509013,
-      "grad_norm": 0.000938993995077908,
-      "learning_rate": 0.0001999978659280327,
-      "loss": 46.0,
-      "step": 27209
-    },
-    {
-      "epoch": 2.080394518034291,
-      "grad_norm": 0.0013045774539932609,
-      "learning_rate": 0.00019999786577110755,
-      "loss": 46.0,
-      "step": 27210
-    },
-    {
-      "epoch": 2.080470975017681,
-      "grad_norm": 0.00042622911860235035,
-      "learning_rate": 0.00019999786561417663,
-      "loss": 46.0,
-      "step": 27211
-    },
-    {
-      "epoch": 2.0805474320010706,
-      "grad_norm": 0.0006025473121553659,
-      "learning_rate": 0.0001999978654572399,
-      "loss": 46.0,
-      "step": 27212
-    },
-    {
-      "epoch": 2.08062388898446,
-      "grad_norm": 0.00041579693788662553,
-      "learning_rate": 0.00019999786530029744,
-      "loss": 46.0,
-      "step": 27213
-    },
-    {
-      "epoch": 2.0807003459678497,
-      "grad_norm": 0.0020275823771953583,
-      "learning_rate": 0.00019999786514334923,
-      "loss": 46.0,
-      "step": 27214
-    },
-    {
-      "epoch": 2.0807768029512395,
-      "grad_norm": 0.0013582655228674412,
-      "learning_rate": 0.00019999786498639524,
-      "loss": 46.0,
-      "step": 27215
-    },
-    {
-      "epoch": 2.080853259934629,
-      "grad_norm": 0.0008413522155024111,
-      "learning_rate": 0.00019999786482943549,
-      "loss": 46.0,
-      "step": 27216
-    },
-    {
-      "epoch": 2.080929716918019,
-      "grad_norm": 0.0008687196532264352,
-      "learning_rate": 0.00019999786467246993,
-      "loss": 46.0,
-      "step": 27217
-    },
-    {
-      "epoch": 2.0810061739014087,
-      "grad_norm": 0.0009470437071286142,
-      "learning_rate": 0.00019999786451549862,
-      "loss": 46.0,
-      "step": 27218
-    },
-    {
-      "epoch": 2.0810826308847985,
-      "grad_norm": 0.0009760914836078882,
-      "learning_rate": 0.00019999786435852154,
-      "loss": 46.0,
-      "step": 27219
-    },
-    {
-      "epoch": 2.0811590878681883,
-      "grad_norm": 0.0015747162979096174,
-      "learning_rate": 0.0001999978642015387,
-      "loss": 46.0,
-      "step": 27220
-    },
-    {
-      "epoch": 2.081235544851578,
-      "grad_norm": 0.0014986732276156545,
-      "learning_rate": 0.00019999786404455006,
-      "loss": 46.0,
-      "step": 27221
-    },
-    {
-      "epoch": 2.081312001834968,
-      "grad_norm": 0.002020235639065504,
-      "learning_rate": 0.0001999978638875557,
-      "loss": 46.0,
-      "step": 27222
-    },
-    {
-      "epoch": 2.081388458818357,
-      "grad_norm": 0.011122310534119606,
-      "learning_rate": 0.00019999786373055555,
-      "loss": 46.0,
-      "step": 27223
-    },
-    {
-      "epoch": 2.081464915801747,
-      "grad_norm": 0.0009557961602695286,
-      "learning_rate": 0.00019999786357354963,
-      "loss": 46.0,
-      "step": 27224
-    },
-    {
-      "epoch": 2.0815413727851366,
-      "grad_norm": 0.0012390066403895617,
-      "learning_rate": 0.00019999786341653794,
-      "loss": 46.0,
-      "step": 27225
-    },
-    {
-      "epoch": 2.0816178297685264,
-      "grad_norm": 0.003337208181619644,
-      "learning_rate": 0.00019999786325952047,
-      "loss": 46.0,
-      "step": 27226
-    },
-    {
-      "epoch": 2.081694286751916,
-      "grad_norm": 0.0016678700922057033,
-      "learning_rate": 0.00019999786310249723,
-      "loss": 46.0,
-      "step": 27227
-    },
-    {
-      "epoch": 2.081770743735306,
-      "grad_norm": 0.005351881962269545,
-      "learning_rate": 0.00019999786294546822,
-      "loss": 46.0,
-      "step": 27228
-    },
-    {
-      "epoch": 2.0818472007186957,
-      "grad_norm": 0.0004929532879032195,
-      "learning_rate": 0.0001999978627884335,
-      "loss": 46.0,
-      "step": 27229
-    },
-    {
-      "epoch": 2.0819236577020854,
-      "grad_norm": 0.0010795847047120333,
-      "learning_rate": 0.00019999786263139293,
-      "loss": 46.0,
-      "step": 27230
-    },
-    {
-      "epoch": 2.082000114685475,
-      "grad_norm": 0.0016849403036758304,
-      "learning_rate": 0.00019999786247434662,
-      "loss": 46.0,
-      "step": 27231
-    },
-    {
-      "epoch": 2.082076571668865,
-      "grad_norm": 0.002279861830174923,
-      "learning_rate": 0.00019999786231729455,
-      "loss": 46.0,
-      "step": 27232
-    },
-    {
-      "epoch": 2.0821530286522547,
-      "grad_norm": 0.0023578687105327845,
-      "learning_rate": 0.0001999978621602367,
-      "loss": 46.0,
-      "step": 27233
-    },
-    {
-      "epoch": 2.0822294856356445,
-      "grad_norm": 0.0023254305124282837,
-      "learning_rate": 0.0001999978620031731,
-      "loss": 46.0,
-      "step": 27234
-    },
-    {
-      "epoch": 2.082305942619034,
-      "grad_norm": 0.0010756539413705468,
-      "learning_rate": 0.0001999978618461037,
-      "loss": 46.0,
-      "step": 27235
-    },
-    {
-      "epoch": 2.0823823996024236,
-      "grad_norm": 0.0006331177428364754,
-      "learning_rate": 0.00019999786168902855,
-      "loss": 46.0,
-      "step": 27236
-    },
-    {
-      "epoch": 2.0824588565858133,
-      "grad_norm": 0.001233407179825008,
-      "learning_rate": 0.0001999978615319476,
-      "loss": 46.0,
-      "step": 27237
-    },
-    {
-      "epoch": 2.082535313569203,
-      "grad_norm": 0.0005494873621501029,
-      "learning_rate": 0.00019999786137486094,
-      "loss": 46.0,
-      "step": 27238
-    },
-    {
-      "epoch": 2.082611770552593,
-      "grad_norm": 0.000868365284986794,
-      "learning_rate": 0.00019999786121776848,
-      "loss": 46.0,
-      "step": 27239
-    },
-    {
-      "epoch": 2.0826882275359826,
-      "grad_norm": 0.0011070248438045382,
-      "learning_rate": 0.0001999978610606702,
-      "loss": 46.0,
-      "step": 27240
-    },
-    {
-      "epoch": 2.0827646845193724,
-      "grad_norm": 0.001520760590210557,
-      "learning_rate": 0.00019999786090356625,
-      "loss": 46.0,
-      "step": 27241
-    },
-    {
-      "epoch": 2.082841141502762,
-      "grad_norm": 0.0016706987516954541,
-      "learning_rate": 0.00019999786074645647,
-      "loss": 46.0,
-      "step": 27242
-    },
-    {
-      "epoch": 2.082917598486152,
-      "grad_norm": 0.0006322450935840607,
-      "learning_rate": 0.00019999786058934094,
-      "loss": 46.0,
-      "step": 27243
-    },
-    {
-      "epoch": 2.0829940554695416,
-      "grad_norm": 0.003286790568381548,
-      "learning_rate": 0.0001999978604322196,
-      "loss": 46.0,
-      "step": 27244
-    },
-    {
-      "epoch": 2.083070512452931,
-      "grad_norm": 0.000545597868040204,
-      "learning_rate": 0.00019999786027509253,
-      "loss": 46.0,
-      "step": 27245
-    },
-    {
-      "epoch": 2.0831469694363207,
-      "grad_norm": 0.0008285172516480088,
-      "learning_rate": 0.0001999978601179597,
-      "loss": 46.0,
-      "step": 27246
-    },
-    {
-      "epoch": 2.0832234264197105,
-      "grad_norm": 0.004713290371000767,
-      "learning_rate": 0.00019999785996082108,
-      "loss": 46.0,
-      "step": 27247
-    },
-    {
-      "epoch": 2.0832998834031002,
-      "grad_norm": 0.0027769531589001417,
-      "learning_rate": 0.00019999785980367668,
-      "loss": 46.0,
-      "step": 27248
-    },
-    {
-      "epoch": 2.08337634038649,
-      "grad_norm": 0.00078834273153916,
-      "learning_rate": 0.00019999785964652653,
-      "loss": 46.0,
-      "step": 27249
-    },
-    {
-      "epoch": 2.0834527973698798,
-      "grad_norm": 0.0006405291496776044,
-      "learning_rate": 0.0001999978594893706,
-      "loss": 46.0,
-      "step": 27250
-    },
-    {
-      "epoch": 2.0835292543532695,
-      "grad_norm": 0.013986635021865368,
-      "learning_rate": 0.0001999978593322089,
-      "loss": 46.0,
-      "step": 27251
-    },
-    {
-      "epoch": 2.0836057113366593,
-      "grad_norm": 0.003889655228704214,
-      "learning_rate": 0.00019999785917504143,
-      "loss": 46.0,
-      "step": 27252
-    },
-    {
-      "epoch": 2.083682168320049,
-      "grad_norm": 0.0031468758825212717,
-      "learning_rate": 0.0001999978590178682,
-      "loss": 46.0,
-      "step": 27253
-    },
-    {
-      "epoch": 2.083758625303439,
-      "grad_norm": 0.0005284941871650517,
-      "learning_rate": 0.0001999978588606892,
-      "loss": 46.0,
-      "step": 27254
-    },
-    {
-      "epoch": 2.0838350822868286,
-      "grad_norm": 0.0021087052300572395,
-      "learning_rate": 0.00019999785870350442,
-      "loss": 46.0,
-      "step": 27255
-    },
-    {
-      "epoch": 2.083911539270218,
-      "grad_norm": 0.0007566118729300797,
-      "learning_rate": 0.0001999978585463139,
-      "loss": 46.0,
-      "step": 27256
-    },
-    {
-      "epoch": 2.0839879962536076,
-      "grad_norm": 0.0006520853494293988,
-      "learning_rate": 0.0001999978583891176,
-      "loss": 46.0,
-      "step": 27257
-    },
-    {
-      "epoch": 2.0840644532369974,
-      "grad_norm": 0.0004206908051855862,
-      "learning_rate": 0.0001999978582319155,
-      "loss": 46.0,
-      "step": 27258
-    },
-    {
-      "epoch": 2.084140910220387,
-      "grad_norm": 0.0018530928064137697,
-      "learning_rate": 0.00019999785807470763,
-      "loss": 46.0,
-      "step": 27259
-    },
-    {
-      "epoch": 2.084217367203777,
-      "grad_norm": 0.0024774286430329084,
-      "learning_rate": 0.00019999785791749403,
-      "loss": 46.0,
-      "step": 27260
-    },
-    {
-      "epoch": 2.0842938241871667,
-      "grad_norm": 0.0017375156749039888,
-      "learning_rate": 0.00019999785776027464,
-      "loss": 46.0,
-      "step": 27261
-    },
-    {
-      "epoch": 2.0843702811705564,
-      "grad_norm": 0.0033782843966037035,
-      "learning_rate": 0.0001999978576030495,
-      "loss": 46.0,
-      "step": 27262
-    },
-    {
-      "epoch": 2.084446738153946,
-      "grad_norm": 0.0007590474560856819,
-      "learning_rate": 0.00019999785744581855,
-      "loss": 46.0,
-      "step": 27263
-    },
-    {
-      "epoch": 2.084523195137336,
-      "grad_norm": 0.012711228802800179,
-      "learning_rate": 0.00019999785728858186,
-      "loss": 46.0,
-      "step": 27264
-    },
-    {
-      "epoch": 2.0845996521207257,
-      "grad_norm": 0.0006222252268344164,
-      "learning_rate": 0.00019999785713133942,
-      "loss": 46.0,
-      "step": 27265
-    },
-    {
-      "epoch": 2.0846761091041155,
-      "grad_norm": 0.002791031962260604,
-      "learning_rate": 0.00019999785697409115,
-      "loss": 46.0,
-      "step": 27266
-    },
-    {
-      "epoch": 2.084752566087505,
-      "grad_norm": 0.0009851750219240785,
-      "learning_rate": 0.00019999785681683717,
-      "loss": 46.0,
-      "step": 27267
-    },
-    {
-      "epoch": 2.0848290230708946,
-      "grad_norm": 0.0007425334770232439,
-      "learning_rate": 0.0001999978566595774,
-      "loss": 46.0,
-      "step": 27268
-    },
-    {
-      "epoch": 2.0849054800542843,
-      "grad_norm": 0.0029472720343619585,
-      "learning_rate": 0.00019999785650231185,
-      "loss": 46.0,
-      "step": 27269
-    },
-    {
-      "epoch": 2.084981937037674,
-      "grad_norm": 0.0027116015553474426,
-      "learning_rate": 0.00019999785634504052,
-      "loss": 46.0,
-      "step": 27270
-    },
-    {
-      "epoch": 2.085058394021064,
-      "grad_norm": 0.0015690447762608528,
-      "learning_rate": 0.00019999785618776344,
-      "loss": 46.0,
-      "step": 27271
-    },
-    {
-      "epoch": 2.0851348510044536,
-      "grad_norm": 0.001972117694094777,
-      "learning_rate": 0.0001999978560304806,
-      "loss": 46.0,
-      "step": 27272
-    },
-    {
-      "epoch": 2.0852113079878434,
-      "grad_norm": 0.0009207900729961693,
-      "learning_rate": 0.00019999785587319197,
-      "loss": 46.0,
-      "step": 27273
-    },
-    {
-      "epoch": 2.085287764971233,
-      "grad_norm": 0.000603479624260217,
-      "learning_rate": 0.0001999978557158976,
-      "loss": 46.0,
-      "step": 27274
-    },
-    {
-      "epoch": 2.085364221954623,
-      "grad_norm": 0.0006268219440244138,
-      "learning_rate": 0.00019999785555859743,
-      "loss": 46.0,
-      "step": 27275
-    },
-    {
-      "epoch": 2.0854406789380127,
-      "grad_norm": 0.0023255066480487585,
-      "learning_rate": 0.0001999978554012915,
-      "loss": 46.0,
-      "step": 27276
-    },
-    {
-      "epoch": 2.0855171359214024,
-      "grad_norm": 0.0021955957636237144,
-      "learning_rate": 0.0001999978552439798,
-      "loss": 46.0,
-      "step": 27277
-    },
-    {
-      "epoch": 2.0855935929047917,
-      "grad_norm": 0.00640465784817934,
-      "learning_rate": 0.00019999785508666233,
-      "loss": 46.0,
-      "step": 27278
-    },
-    {
-      "epoch": 2.0856700498881815,
-      "grad_norm": 0.001450004754588008,
-      "learning_rate": 0.0001999978549293391,
-      "loss": 46.0,
-      "step": 27279
-    },
-    {
-      "epoch": 2.0857465068715713,
-      "grad_norm": 0.002285077003762126,
-      "learning_rate": 0.0001999978547720101,
-      "loss": 46.0,
-      "step": 27280
-    },
-    {
-      "epoch": 2.085822963854961,
-      "grad_norm": 0.0024938893038779497,
-      "learning_rate": 0.00019999785461467532,
-      "loss": 46.0,
-      "step": 27281
-    },
-    {
-      "epoch": 2.085899420838351,
-      "grad_norm": 0.0012369889300316572,
-      "learning_rate": 0.0001999978544573348,
-      "loss": 46.0,
-      "step": 27282
-    },
-    {
-      "epoch": 2.0859758778217405,
-      "grad_norm": 0.005502518266439438,
-      "learning_rate": 0.00019999785429998846,
-      "loss": 46.0,
-      "step": 27283
-    },
-    {
-      "epoch": 2.0860523348051303,
-      "grad_norm": 0.00030840406543575227,
-      "learning_rate": 0.00019999785414263638,
-      "loss": 46.0,
-      "step": 27284
-    },
-    {
-      "epoch": 2.08612879178852,
-      "grad_norm": 0.0007111412123776972,
-      "learning_rate": 0.00019999785398527853,
-      "loss": 46.0,
-      "step": 27285
-    },
-    {
-      "epoch": 2.08620524877191,
-      "grad_norm": 0.0018708074931055307,
-      "learning_rate": 0.00019999785382791488,
-      "loss": 46.0,
-      "step": 27286
-    },
-    {
-      "epoch": 2.0862817057552996,
-      "grad_norm": 0.00042130364454351366,
-      "learning_rate": 0.00019999785367054554,
-      "loss": 46.0,
-      "step": 27287
-    },
-    {
-      "epoch": 2.0863581627386893,
-      "grad_norm": 0.005188572220504284,
-      "learning_rate": 0.00019999785351317034,
-      "loss": 46.0,
-      "step": 27288
-    },
-    {
-      "epoch": 2.0864346197220787,
-      "grad_norm": 0.00044203459401614964,
-      "learning_rate": 0.00019999785335578945,
-      "loss": 46.0,
-      "step": 27289
-    },
-    {
-      "epoch": 2.0865110767054684,
-      "grad_norm": 0.0005602473393082619,
-      "learning_rate": 0.0001999978531984027,
-      "loss": 46.0,
-      "step": 27290
-    },
-    {
-      "epoch": 2.086587533688858,
-      "grad_norm": 0.0008185104816220701,
-      "learning_rate": 0.00019999785304101027,
-      "loss": 46.0,
-      "step": 27291
-    },
-    {
-      "epoch": 2.086663990672248,
-      "grad_norm": 0.00395533861592412,
-      "learning_rate": 0.00019999785288361203,
-      "loss": 46.0,
-      "step": 27292
-    },
-    {
-      "epoch": 2.0867404476556377,
-      "grad_norm": 0.002584238536655903,
-      "learning_rate": 0.00019999785272620805,
-      "loss": 46.0,
-      "step": 27293
-    },
-    {
-      "epoch": 2.0868169046390275,
-      "grad_norm": 0.0016792984679341316,
-      "learning_rate": 0.00019999785256879824,
-      "loss": 46.0,
-      "step": 27294
-    },
-    {
-      "epoch": 2.0868933616224172,
-      "grad_norm": 0.0009061884484253824,
-      "learning_rate": 0.0001999978524113827,
-      "loss": 46.0,
-      "step": 27295
-    },
-    {
-      "epoch": 2.086969818605807,
-      "grad_norm": 0.000508524535689503,
-      "learning_rate": 0.0001999978522539614,
-      "loss": 46.0,
-      "step": 27296
-    },
-    {
-      "epoch": 2.0870462755891968,
-      "grad_norm": 0.0006886657210998237,
-      "learning_rate": 0.0001999978520965343,
-      "loss": 46.0,
-      "step": 27297
-    },
-    {
-      "epoch": 2.0871227325725865,
-      "grad_norm": 0.011873285286128521,
-      "learning_rate": 0.00019999785193910142,
-      "loss": 46.0,
-      "step": 27298
-    },
-    {
-      "epoch": 2.0871991895559763,
-      "grad_norm": 0.002773172687739134,
-      "learning_rate": 0.00019999785178166282,
-      "loss": 46.0,
-      "step": 27299
-    },
-    {
-      "epoch": 2.0872756465393656,
-      "grad_norm": 0.005620944779366255,
-      "learning_rate": 0.00019999785162421843,
-      "loss": 46.0,
-      "step": 27300
-    },
-    {
-      "epoch": 2.0873521035227554,
-      "grad_norm": 0.0004794311826117337,
-      "learning_rate": 0.00019999785146676828,
-      "loss": 46.0,
-      "step": 27301
-    },
-    {
-      "epoch": 2.087428560506145,
-      "grad_norm": 0.0017960167024284601,
-      "learning_rate": 0.00019999785130931234,
-      "loss": 46.0,
-      "step": 27302
-    },
-    {
-      "epoch": 2.087505017489535,
-      "grad_norm": 0.0016684964066371322,
-      "learning_rate": 0.00019999785115185062,
-      "loss": 46.0,
-      "step": 27303
-    },
-    {
-      "epoch": 2.0875814744729246,
-      "grad_norm": 0.00072661874582991,
-      "learning_rate": 0.0001999978509943832,
-      "loss": 46.0,
-      "step": 27304
-    },
-    {
-      "epoch": 2.0876579314563144,
-      "grad_norm": 0.00037797962431795895,
-      "learning_rate": 0.00019999785083690995,
-      "loss": 46.0,
-      "step": 27305
-    },
-    {
-      "epoch": 2.087734388439704,
-      "grad_norm": 0.002528723794966936,
-      "learning_rate": 0.00019999785067943094,
-      "loss": 46.0,
-      "step": 27306
-    },
-    {
-      "epoch": 2.087810845423094,
-      "grad_norm": 0.00035425880923867226,
-      "learning_rate": 0.00019999785052194616,
-      "loss": 46.0,
-      "step": 27307
-    },
-    {
-      "epoch": 2.0878873024064837,
-      "grad_norm": 0.0007851978298276663,
-      "learning_rate": 0.0001999978503644556,
-      "loss": 46.0,
-      "step": 27308
-    },
-    {
-      "epoch": 2.0879637593898734,
-      "grad_norm": 0.0005193176330067217,
-      "learning_rate": 0.00019999785020695927,
-      "loss": 46.0,
-      "step": 27309
-    },
-    {
-      "epoch": 2.088040216373263,
-      "grad_norm": 0.0006123346392996609,
-      "learning_rate": 0.00019999785004945722,
-      "loss": 46.0,
-      "step": 27310
-    },
-    {
-      "epoch": 2.0881166733566525,
-      "grad_norm": 0.0011612738016992807,
-      "learning_rate": 0.00019999784989194934,
-      "loss": 46.0,
-      "step": 27311
-    },
-    {
-      "epoch": 2.0881931303400423,
-      "grad_norm": 0.0014939234824851155,
-      "learning_rate": 0.00019999784973443572,
-      "loss": 46.0,
-      "step": 27312
-    },
-    {
-      "epoch": 2.088269587323432,
-      "grad_norm": 0.006138111464679241,
-      "learning_rate": 0.00019999784957691632,
-      "loss": 46.0,
-      "step": 27313
-    },
-    {
-      "epoch": 2.088346044306822,
-      "grad_norm": 0.0014626990305259824,
-      "learning_rate": 0.00019999784941939118,
-      "loss": 46.0,
-      "step": 27314
-    },
-    {
-      "epoch": 2.0884225012902116,
-      "grad_norm": 0.0016980463406071067,
-      "learning_rate": 0.00019999784926186026,
-      "loss": 46.0,
-      "step": 27315
-    },
-    {
-      "epoch": 2.0884989582736013,
-      "grad_norm": 0.0005950651247985661,
-      "learning_rate": 0.00019999784910432354,
-      "loss": 46.0,
-      "step": 27316
-    },
-    {
-      "epoch": 2.088575415256991,
-      "grad_norm": 0.0050689708441495895,
-      "learning_rate": 0.00019999784894678105,
-      "loss": 46.0,
-      "step": 27317
-    },
-    {
-      "epoch": 2.088651872240381,
-      "grad_norm": 0.01390141248703003,
-      "learning_rate": 0.00019999784878923282,
-      "loss": 46.0,
-      "step": 27318
-    },
-    {
-      "epoch": 2.0887283292237706,
-      "grad_norm": 0.0019448251696303487,
-      "learning_rate": 0.0001999978486316788,
-      "loss": 46.0,
-      "step": 27319
-    },
-    {
-      "epoch": 2.0888047862071604,
-      "grad_norm": 0.0018287294078618288,
-      "learning_rate": 0.00019999784847411905,
-      "loss": 46.0,
-      "step": 27320
-    },
-    {
-      "epoch": 2.08888124319055,
-      "grad_norm": 0.0015632157446816564,
-      "learning_rate": 0.0001999978483165535,
-      "loss": 46.0,
-      "step": 27321
-    },
-    {
-      "epoch": 2.0889577001739394,
-      "grad_norm": 0.0005896036163903773,
-      "learning_rate": 0.00019999784815898216,
-      "loss": 46.0,
-      "step": 27322
-    },
-    {
-      "epoch": 2.089034157157329,
-      "grad_norm": 0.0005592514644376934,
-      "learning_rate": 0.0001999978480014051,
-      "loss": 46.0,
-      "step": 27323
-    },
-    {
-      "epoch": 2.089110614140719,
-      "grad_norm": 0.0008798418566584587,
-      "learning_rate": 0.00019999784784382224,
-      "loss": 46.0,
-      "step": 27324
-    },
-    {
-      "epoch": 2.0891870711241087,
-      "grad_norm": 0.001165579305961728,
-      "learning_rate": 0.0001999978476862336,
-      "loss": 46.0,
-      "step": 27325
-    },
-    {
-      "epoch": 2.0892635281074985,
-      "grad_norm": 0.0003788323374465108,
-      "learning_rate": 0.00019999784752863922,
-      "loss": 46.0,
-      "step": 27326
-    },
-    {
-      "epoch": 2.0893399850908883,
-      "grad_norm": 0.001779798767529428,
-      "learning_rate": 0.00019999784737103905,
-      "loss": 46.0,
-      "step": 27327
-    },
-    {
-      "epoch": 2.089416442074278,
-      "grad_norm": 0.001192900468595326,
-      "learning_rate": 0.0001999978472134331,
-      "loss": 46.0,
-      "step": 27328
-    },
-    {
-      "epoch": 2.0894928990576678,
-      "grad_norm": 0.0009551694965921342,
-      "learning_rate": 0.00019999784705582142,
-      "loss": 46.0,
-      "step": 27329
-    },
-    {
-      "epoch": 2.0895693560410575,
-      "grad_norm": 0.0009529570234008133,
-      "learning_rate": 0.00019999784689820393,
-      "loss": 46.0,
-      "step": 27330
-    },
-    {
-      "epoch": 2.0896458130244473,
-      "grad_norm": 0.0014148111222311854,
-      "learning_rate": 0.00019999784674058072,
-      "loss": 46.0,
-      "step": 27331
-    },
-    {
-      "epoch": 2.0897222700078366,
-      "grad_norm": 0.0022507377434521914,
-      "learning_rate": 0.00019999784658295168,
-      "loss": 46.0,
-      "step": 27332
-    },
-    {
-      "epoch": 2.0897987269912264,
-      "grad_norm": 0.0004598718660417944,
-      "learning_rate": 0.0001999978464253169,
-      "loss": 46.0,
-      "step": 27333
-    },
-    {
-      "epoch": 2.089875183974616,
-      "grad_norm": 0.0006176439928822219,
-      "learning_rate": 0.00019999784626767637,
-      "loss": 46.0,
-      "step": 27334
-    },
-    {
-      "epoch": 2.089951640958006,
-      "grad_norm": 0.0007755841943435371,
-      "learning_rate": 0.00019999784611003004,
-      "loss": 46.0,
-      "step": 27335
-    },
-    {
-      "epoch": 2.0900280979413957,
-      "grad_norm": 0.017024213448166847,
-      "learning_rate": 0.00019999784595237796,
-      "loss": 46.0,
-      "step": 27336
-    },
-    {
-      "epoch": 2.0901045549247854,
-      "grad_norm": 0.001571474364027381,
-      "learning_rate": 0.00019999784579472009,
-      "loss": 46.0,
-      "step": 27337
-    },
-    {
-      "epoch": 2.090181011908175,
-      "grad_norm": 0.004843530245125294,
-      "learning_rate": 0.00019999784563705646,
-      "loss": 46.0,
-      "step": 27338
-    },
-    {
-      "epoch": 2.090257468891565,
-      "grad_norm": 0.0032593111973255873,
-      "learning_rate": 0.00019999784547938707,
-      "loss": 46.0,
-      "step": 27339
-    },
-    {
-      "epoch": 2.0903339258749547,
-      "grad_norm": 0.0005252194823697209,
-      "learning_rate": 0.0001999978453217119,
-      "loss": 46.0,
-      "step": 27340
-    },
-    {
-      "epoch": 2.0904103828583445,
-      "grad_norm": 0.0029758221935480833,
-      "learning_rate": 0.00019999784516403098,
-      "loss": 46.0,
-      "step": 27341
-    },
-    {
-      "epoch": 2.0904868398417342,
-      "grad_norm": 0.0007039497722871602,
-      "learning_rate": 0.00019999784500634426,
-      "loss": 46.0,
-      "step": 27342
-    },
-    {
-      "epoch": 2.090563296825124,
-      "grad_norm": 0.004084882326424122,
-      "learning_rate": 0.00019999784484865177,
-      "loss": 46.0,
-      "step": 27343
-    },
-    {
-      "epoch": 2.0906397538085133,
-      "grad_norm": 0.0009567401721142232,
-      "learning_rate": 0.00019999784469095354,
-      "loss": 46.0,
-      "step": 27344
-    },
-    {
-      "epoch": 2.090716210791903,
-      "grad_norm": 0.0023373793810606003,
-      "learning_rate": 0.00019999784453324956,
-      "loss": 46.0,
-      "step": 27345
-    },
-    {
-      "epoch": 2.090792667775293,
-      "grad_norm": 0.0024321535602211952,
-      "learning_rate": 0.00019999784437553975,
-      "loss": 46.0,
-      "step": 27346
-    },
-    {
-      "epoch": 2.0908691247586826,
-      "grad_norm": 0.0017820857465267181,
-      "learning_rate": 0.00019999784421782422,
-      "loss": 46.0,
-      "step": 27347
-    },
-    {
-      "epoch": 2.0909455817420723,
-      "grad_norm": 0.000932272698264569,
-      "learning_rate": 0.0001999978440601029,
-      "loss": 46.0,
-      "step": 27348
-    },
-    {
-      "epoch": 2.091022038725462,
-      "grad_norm": 0.0011680368334054947,
-      "learning_rate": 0.00019999784390237579,
-      "loss": 46.0,
-      "step": 27349
-    },
-    {
-      "epoch": 2.091098495708852,
-      "grad_norm": 0.0011892203474417329,
-      "learning_rate": 0.00019999784374464294,
-      "loss": 46.0,
-      "step": 27350
-    },
-    {
-      "epoch": 2.0911749526922416,
-      "grad_norm": 0.0010113869793713093,
-      "learning_rate": 0.00019999784358690431,
-      "loss": 46.0,
-      "step": 27351
-    },
-    {
-      "epoch": 2.0912514096756314,
-      "grad_norm": 0.0007817111327312887,
-      "learning_rate": 0.00019999784342915992,
-      "loss": 46.0,
-      "step": 27352
-    },
-    {
-      "epoch": 2.091327866659021,
-      "grad_norm": 0.0012770001776516438,
-      "learning_rate": 0.00019999784327140975,
-      "loss": 46.0,
-      "step": 27353
-    },
-    {
-      "epoch": 2.0914043236424105,
-      "grad_norm": 0.0009075301932170987,
-      "learning_rate": 0.0001999978431136538,
-      "loss": 46.0,
-      "step": 27354
-    },
-    {
-      "epoch": 2.0914807806258002,
-      "grad_norm": 0.005038647912442684,
-      "learning_rate": 0.0001999978429558921,
-      "loss": 46.0,
-      "step": 27355
-    },
-    {
-      "epoch": 2.09155723760919,
-      "grad_norm": 0.0027564982883632183,
-      "learning_rate": 0.00019999784279812463,
-      "loss": 46.0,
-      "step": 27356
-    },
-    {
-      "epoch": 2.0916336945925798,
-      "grad_norm": 0.00037014728877693415,
-      "learning_rate": 0.0001999978426403514,
-      "loss": 46.0,
-      "step": 27357
-    },
-    {
-      "epoch": 2.0917101515759695,
-      "grad_norm": 0.0010014689760282636,
-      "learning_rate": 0.00019999784248257236,
-      "loss": 46.0,
-      "step": 27358
-    },
-    {
-      "epoch": 2.0917866085593593,
-      "grad_norm": 0.0021338353399187326,
-      "learning_rate": 0.0001999978423247876,
-      "loss": 46.0,
-      "step": 27359
-    },
-    {
-      "epoch": 2.091863065542749,
-      "grad_norm": 0.0012704251566901803,
-      "learning_rate": 0.00019999784216699704,
-      "loss": 46.0,
-      "step": 27360
-    },
-    {
-      "epoch": 2.091939522526139,
-      "grad_norm": 0.0008454549824818969,
-      "learning_rate": 0.00019999784200920072,
-      "loss": 46.0,
-      "step": 27361
-    },
-    {
-      "epoch": 2.0920159795095286,
-      "grad_norm": 0.0013233498902991414,
-      "learning_rate": 0.0001999978418513986,
-      "loss": 46.0,
-      "step": 27362
-    },
-    {
-      "epoch": 2.0920924364929183,
-      "grad_norm": 0.000730192638002336,
-      "learning_rate": 0.00019999784169359076,
-      "loss": 46.0,
-      "step": 27363
-    },
-    {
-      "epoch": 2.092168893476308,
-      "grad_norm": 0.003717245766893029,
-      "learning_rate": 0.00019999784153577714,
-      "loss": 46.0,
-      "step": 27364
-    },
-    {
-      "epoch": 2.092245350459698,
-      "grad_norm": 0.002286767354235053,
-      "learning_rate": 0.00019999784137795775,
-      "loss": 46.0,
-      "step": 27365
-    },
-    {
-      "epoch": 2.092321807443087,
-      "grad_norm": 0.000832725374493748,
-      "learning_rate": 0.00019999784122013258,
-      "loss": 46.0,
-      "step": 27366
-    },
-    {
-      "epoch": 2.092398264426477,
-      "grad_norm": 0.0010958941420540214,
-      "learning_rate": 0.00019999784106230163,
-      "loss": 46.0,
-      "step": 27367
-    },
-    {
-      "epoch": 2.0924747214098667,
-      "grad_norm": 0.0005153437377884984,
-      "learning_rate": 0.00019999784090446495,
-      "loss": 46.0,
-      "step": 27368
-    },
-    {
-      "epoch": 2.0925511783932564,
-      "grad_norm": 0.0010657175444066525,
-      "learning_rate": 0.00019999784074662246,
-      "loss": 46.0,
-      "step": 27369
-    },
-    {
-      "epoch": 2.092627635376646,
-      "grad_norm": 0.0005108549376018345,
-      "learning_rate": 0.00019999784058877422,
-      "loss": 46.0,
-      "step": 27370
-    },
-    {
-      "epoch": 2.092704092360036,
-      "grad_norm": 0.0005293979193083942,
-      "learning_rate": 0.0001999978404309202,
-      "loss": 46.0,
-      "step": 27371
-    },
-    {
-      "epoch": 2.0927805493434257,
-      "grad_norm": 0.0012036376865580678,
-      "learning_rate": 0.0001999978402730604,
-      "loss": 46.0,
-      "step": 27372
-    },
-    {
-      "epoch": 2.0928570063268155,
-      "grad_norm": 0.0007132362807169557,
-      "learning_rate": 0.00019999784011519488,
-      "loss": 46.0,
-      "step": 27373
-    },
-    {
-      "epoch": 2.0929334633102052,
-      "grad_norm": 0.0008389372960664332,
-      "learning_rate": 0.00019999783995732355,
-      "loss": 46.0,
-      "step": 27374
-    },
-    {
-      "epoch": 2.093009920293595,
-      "grad_norm": 0.001961604692041874,
-      "learning_rate": 0.00019999783979944647,
-      "loss": 46.0,
-      "step": 27375
-    },
-    {
-      "epoch": 2.0930863772769843,
-      "grad_norm": 0.0014159164857119322,
-      "learning_rate": 0.0001999978396415636,
-      "loss": 46.0,
-      "step": 27376
-    },
-    {
-      "epoch": 2.093162834260374,
-      "grad_norm": 0.0018381391419097781,
-      "learning_rate": 0.00019999783948367497,
-      "loss": 46.0,
-      "step": 27377
-    },
-    {
-      "epoch": 2.093239291243764,
-      "grad_norm": 0.0009232028387486935,
-      "learning_rate": 0.00019999783932578055,
-      "loss": 46.0,
-      "step": 27378
-    },
-    {
-      "epoch": 2.0933157482271536,
-      "grad_norm": 0.0007812344119884074,
-      "learning_rate": 0.00019999783916788038,
-      "loss": 46.0,
-      "step": 27379
-    },
-    {
-      "epoch": 2.0933922052105434,
-      "grad_norm": 0.0005640701856464148,
-      "learning_rate": 0.00019999783900997447,
-      "loss": 46.0,
-      "step": 27380
-    },
-    {
-      "epoch": 2.093468662193933,
-      "grad_norm": 0.02172602340579033,
-      "learning_rate": 0.00019999783885206273,
-      "loss": 46.0,
-      "step": 27381
-    },
-    {
-      "epoch": 2.093545119177323,
-      "grad_norm": 0.0021992987021803856,
-      "learning_rate": 0.0001999978386941453,
-      "loss": 46.0,
-      "step": 27382
-    },
-    {
-      "epoch": 2.0936215761607126,
-      "grad_norm": 0.0011721898335963488,
-      "learning_rate": 0.000199997838536222,
-      "loss": 46.0,
-      "step": 27383
-    },
-    {
-      "epoch": 2.0936980331441024,
-      "grad_norm": 0.0007042097277007997,
-      "learning_rate": 0.000199997838378293,
-      "loss": 46.0,
-      "step": 27384
-    },
-    {
-      "epoch": 2.093774490127492,
-      "grad_norm": 0.0006267200224101543,
-      "learning_rate": 0.0001999978382203582,
-      "loss": 46.0,
-      "step": 27385
-    },
-    {
-      "epoch": 2.093850947110882,
-      "grad_norm": 0.0007163263508118689,
-      "learning_rate": 0.00019999783806241768,
-      "loss": 46.0,
-      "step": 27386
-    },
-    {
-      "epoch": 2.0939274040942712,
-      "grad_norm": 0.010717884637415409,
-      "learning_rate": 0.00019999783790447136,
-      "loss": 46.0,
-      "step": 27387
-    },
-    {
-      "epoch": 2.094003861077661,
-      "grad_norm": 0.00101911136880517,
-      "learning_rate": 0.00019999783774651925,
-      "loss": 46.0,
-      "step": 27388
-    },
-    {
-      "epoch": 2.0940803180610508,
-      "grad_norm": 0.0009173436556011438,
-      "learning_rate": 0.00019999783758856138,
-      "loss": 46.0,
-      "step": 27389
-    },
-    {
-      "epoch": 2.0941567750444405,
-      "grad_norm": 0.0005543183069676161,
-      "learning_rate": 0.00019999783743059776,
-      "loss": 46.0,
-      "step": 27390
-    },
-    {
-      "epoch": 2.0942332320278303,
-      "grad_norm": 0.00109396583866328,
-      "learning_rate": 0.00019999783727262836,
-      "loss": 46.0,
-      "step": 27391
-    },
-    {
-      "epoch": 2.09430968901122,
-      "grad_norm": 0.0013153445906937122,
-      "learning_rate": 0.0001999978371146532,
-      "loss": 46.0,
-      "step": 27392
-    },
-    {
-      "epoch": 2.09438614599461,
-      "grad_norm": 0.0007452231948263943,
-      "learning_rate": 0.00019999783695667225,
-      "loss": 46.0,
-      "step": 27393
-    },
-    {
-      "epoch": 2.0944626029779996,
-      "grad_norm": 0.0012657151091843843,
-      "learning_rate": 0.00019999783679868554,
-      "loss": 46.0,
-      "step": 27394
-    },
-    {
-      "epoch": 2.0945390599613893,
-      "grad_norm": 0.0019953420851379633,
-      "learning_rate": 0.00019999783664069307,
-      "loss": 46.0,
-      "step": 27395
-    },
-    {
-      "epoch": 2.094615516944779,
-      "grad_norm": 0.000672145513817668,
-      "learning_rate": 0.0001999978364826948,
-      "loss": 46.0,
-      "step": 27396
-    },
-    {
-      "epoch": 2.094691973928169,
-      "grad_norm": 0.0007700035930611193,
-      "learning_rate": 0.00019999783632469078,
-      "loss": 46.0,
-      "step": 27397
-    },
-    {
-      "epoch": 2.094768430911558,
-      "grad_norm": 0.005130458157509565,
-      "learning_rate": 0.000199997836166681,
-      "loss": 46.0,
-      "step": 27398
-    },
-    {
-      "epoch": 2.094844887894948,
-      "grad_norm": 0.0005214900593273342,
-      "learning_rate": 0.00019999783600866544,
-      "loss": 46.0,
-      "step": 27399
-    },
-    {
-      "epoch": 2.0949213448783377,
-      "grad_norm": 0.0025057243183255196,
-      "learning_rate": 0.0001999978358506441,
-      "loss": 46.0,
-      "step": 27400
-    },
-    {
-      "epoch": 2.0949978018617275,
-      "grad_norm": 0.0017635300755500793,
-      "learning_rate": 0.00019999783569261704,
-      "loss": 46.0,
-      "step": 27401
-    },
-    {
-      "epoch": 2.095074258845117,
-      "grad_norm": 0.001078938483260572,
-      "learning_rate": 0.00019999783553458416,
-      "loss": 46.0,
-      "step": 27402
-    },
-    {
-      "epoch": 2.095150715828507,
-      "grad_norm": 0.0021621212363243103,
-      "learning_rate": 0.00019999783537654552,
-      "loss": 46.0,
-      "step": 27403
-    },
-    {
-      "epoch": 2.0952271728118967,
-      "grad_norm": 0.0026818534824997187,
-      "learning_rate": 0.00019999783521850112,
-      "loss": 46.0,
-      "step": 27404
-    },
-    {
-      "epoch": 2.0953036297952865,
-      "grad_norm": 0.0007877573953010142,
-      "learning_rate": 0.00019999783506045095,
-      "loss": 46.0,
-      "step": 27405
-    },
-    {
-      "epoch": 2.0953800867786763,
-      "grad_norm": 0.0025736417155712843,
-      "learning_rate": 0.00019999783490239504,
-      "loss": 46.0,
-      "step": 27406
-    },
-    {
-      "epoch": 2.095456543762066,
-      "grad_norm": 0.0005176902050152421,
-      "learning_rate": 0.00019999783474433332,
-      "loss": 46.0,
-      "step": 27407
-    },
-    {
-      "epoch": 2.095533000745456,
-      "grad_norm": 0.003798508085310459,
-      "learning_rate": 0.0001999978345862658,
-      "loss": 46.0,
-      "step": 27408
-    },
-    {
-      "epoch": 2.095609457728845,
-      "grad_norm": 0.005138091742992401,
-      "learning_rate": 0.00019999783442819257,
-      "loss": 46.0,
-      "step": 27409
-    },
-    {
-      "epoch": 2.095685914712235,
-      "grad_norm": 0.0003937718574889004,
-      "learning_rate": 0.00019999783427011357,
-      "loss": 46.0,
-      "step": 27410
-    },
-    {
-      "epoch": 2.0957623716956246,
-      "grad_norm": 0.002517223358154297,
-      "learning_rate": 0.00019999783411202876,
-      "loss": 46.0,
-      "step": 27411
-    },
-    {
-      "epoch": 2.0958388286790144,
-      "grad_norm": 0.0006115194410085678,
-      "learning_rate": 0.0001999978339539382,
-      "loss": 46.0,
-      "step": 27412
-    },
-    {
-      "epoch": 2.095915285662404,
-      "grad_norm": 0.002805655123665929,
-      "learning_rate": 0.0001999978337958419,
-      "loss": 46.0,
-      "step": 27413
-    },
-    {
-      "epoch": 2.095991742645794,
-      "grad_norm": 0.0009261516388505697,
-      "learning_rate": 0.0001999978336377398,
-      "loss": 46.0,
-      "step": 27414
-    },
-    {
-      "epoch": 2.0960681996291837,
-      "grad_norm": 0.001984564121812582,
-      "learning_rate": 0.00019999783347963193,
-      "loss": 46.0,
-      "step": 27415
-    },
-    {
-      "epoch": 2.0961446566125734,
-      "grad_norm": 0.0007494467427022755,
-      "learning_rate": 0.0001999978333215183,
-      "loss": 46.0,
-      "step": 27416
-    },
-    {
-      "epoch": 2.096221113595963,
-      "grad_norm": 0.009969832375645638,
-      "learning_rate": 0.00019999783316339888,
-      "loss": 46.0,
-      "step": 27417
-    },
-    {
-      "epoch": 2.096297570579353,
-      "grad_norm": 0.0009585009538568556,
-      "learning_rate": 0.00019999783300527372,
-      "loss": 46.0,
-      "step": 27418
-    },
-    {
-      "epoch": 2.0963740275627427,
-      "grad_norm": 0.0005163499736227095,
-      "learning_rate": 0.00019999783284714278,
-      "loss": 46.0,
-      "step": 27419
-    },
-    {
-      "epoch": 2.096450484546132,
-      "grad_norm": 0.004340521991252899,
-      "learning_rate": 0.00019999783268900606,
-      "loss": 46.0,
-      "step": 27420
-    },
-    {
-      "epoch": 2.096526941529522,
-      "grad_norm": 0.0008606056217104197,
-      "learning_rate": 0.00019999783253086357,
-      "loss": 46.0,
-      "step": 27421
-    },
-    {
-      "epoch": 2.0966033985129116,
-      "grad_norm": 0.002138251205906272,
-      "learning_rate": 0.0001999978323727153,
-      "loss": 46.0,
-      "step": 27422
-    },
-    {
-      "epoch": 2.0966798554963013,
-      "grad_norm": 0.0006655004690401256,
-      "learning_rate": 0.00019999783221456133,
-      "loss": 46.0,
-      "step": 27423
-    },
-    {
-      "epoch": 2.096756312479691,
-      "grad_norm": 0.0015485911862924695,
-      "learning_rate": 0.00019999783205640152,
-      "loss": 46.0,
-      "step": 27424
-    },
-    {
-      "epoch": 2.096832769463081,
-      "grad_norm": 0.002647699322551489,
-      "learning_rate": 0.00019999783189823597,
-      "loss": 46.0,
-      "step": 27425
-    },
-    {
-      "epoch": 2.0969092264464706,
-      "grad_norm": 0.0024343065451830626,
-      "learning_rate": 0.00019999783174006464,
-      "loss": 46.0,
-      "step": 27426
-    },
-    {
-      "epoch": 2.0969856834298604,
-      "grad_norm": 0.005338119808584452,
-      "learning_rate": 0.00019999783158188754,
-      "loss": 46.0,
-      "step": 27427
-    },
-    {
-      "epoch": 2.09706214041325,
-      "grad_norm": 0.018196184188127518,
-      "learning_rate": 0.00019999783142370467,
-      "loss": 46.0,
-      "step": 27428
-    },
-    {
-      "epoch": 2.09713859739664,
-      "grad_norm": 0.0008322685025632381,
-      "learning_rate": 0.00019999783126551605,
-      "loss": 46.0,
-      "step": 27429
-    },
-    {
-      "epoch": 2.0972150543800296,
-      "grad_norm": 0.0009649796411395073,
-      "learning_rate": 0.00019999783110732165,
-      "loss": 46.0,
-      "step": 27430
-    },
-    {
-      "epoch": 2.097291511363419,
-      "grad_norm": 0.0008222355972975492,
-      "learning_rate": 0.00019999783094912146,
-      "loss": 46.0,
-      "step": 27431
-    },
-    {
-      "epoch": 2.0973679683468087,
-      "grad_norm": 0.0009169584955088794,
-      "learning_rate": 0.00019999783079091552,
-      "loss": 46.0,
-      "step": 27432
-    },
-    {
-      "epoch": 2.0974444253301985,
-      "grad_norm": 0.0006126546650193632,
-      "learning_rate": 0.0001999978306327038,
-      "loss": 46.0,
-      "step": 27433
-    },
-    {
-      "epoch": 2.0975208823135882,
-      "grad_norm": 0.0005853792536072433,
-      "learning_rate": 0.00019999783047448632,
-      "loss": 46.0,
-      "step": 27434
-    },
-    {
-      "epoch": 2.097597339296978,
-      "grad_norm": 0.0007943316013552248,
-      "learning_rate": 0.00019999783031626308,
-      "loss": 46.0,
-      "step": 27435
-    },
-    {
-      "epoch": 2.0976737962803678,
-      "grad_norm": 0.0010237219976261258,
-      "learning_rate": 0.00019999783015803405,
-      "loss": 46.0,
-      "step": 27436
-    },
-    {
-      "epoch": 2.0977502532637575,
-      "grad_norm": 0.0032180820126086473,
-      "learning_rate": 0.00019999782999979927,
-      "loss": 46.0,
-      "step": 27437
-    },
-    {
-      "epoch": 2.0978267102471473,
-      "grad_norm": 0.0026606442406773567,
-      "learning_rate": 0.0001999978298415587,
-      "loss": 46.0,
-      "step": 27438
-    },
-    {
-      "epoch": 2.097903167230537,
-      "grad_norm": 0.005456660874187946,
-      "learning_rate": 0.00019999782968331239,
-      "loss": 46.0,
-      "step": 27439
-    },
-    {
-      "epoch": 2.097979624213927,
-      "grad_norm": 0.0027833774220198393,
-      "learning_rate": 0.00019999782952506028,
-      "loss": 46.0,
-      "step": 27440
-    },
-    {
-      "epoch": 2.0980560811973166,
-      "grad_norm": 0.0013499169144779444,
-      "learning_rate": 0.00019999782936680244,
-      "loss": 46.0,
-      "step": 27441
-    },
-    {
-      "epoch": 2.098132538180706,
-      "grad_norm": 0.004375061485916376,
-      "learning_rate": 0.0001999978292085388,
-      "loss": 46.0,
-      "step": 27442
-    },
-    {
-      "epoch": 2.0982089951640956,
-      "grad_norm": 0.0013925025705248117,
-      "learning_rate": 0.00019999782905026937,
-      "loss": 46.0,
-      "step": 27443
-    },
-    {
-      "epoch": 2.0982854521474854,
-      "grad_norm": 0.0011829192517325282,
-      "learning_rate": 0.00019999782889199418,
-      "loss": 46.0,
-      "step": 27444
-    },
-    {
-      "epoch": 2.098361909130875,
-      "grad_norm": 0.0022398512810468674,
-      "learning_rate": 0.00019999782873371324,
-      "loss": 46.0,
-      "step": 27445
-    },
-    {
-      "epoch": 2.098438366114265,
-      "grad_norm": 0.004042470827698708,
-      "learning_rate": 0.00019999782857542652,
-      "loss": 46.0,
-      "step": 27446
-    },
-    {
-      "epoch": 2.0985148230976547,
-      "grad_norm": 0.002573939273133874,
-      "learning_rate": 0.00019999782841713403,
-      "loss": 46.0,
-      "step": 27447
-    },
-    {
-      "epoch": 2.0985912800810445,
-      "grad_norm": 0.0006682987441308796,
-      "learning_rate": 0.0001999978282588358,
-      "loss": 46.0,
-      "step": 27448
-    },
-    {
-      "epoch": 2.098667737064434,
-      "grad_norm": 0.0019486340461298823,
-      "learning_rate": 0.0001999978281005318,
-      "loss": 46.0,
-      "step": 27449
-    },
-    {
-      "epoch": 2.098744194047824,
-      "grad_norm": 0.0011153511004522443,
-      "learning_rate": 0.00019999782794222196,
-      "loss": 46.0,
-      "step": 27450
-    },
-    {
-      "epoch": 2.0988206510312137,
-      "grad_norm": 0.001907104509882629,
-      "learning_rate": 0.00019999782778390643,
-      "loss": 46.0,
-      "step": 27451
-    },
-    {
-      "epoch": 2.0988971080146035,
-      "grad_norm": 0.0010149977169930935,
-      "learning_rate": 0.0001999978276255851,
-      "loss": 46.0,
-      "step": 27452
-    },
-    {
-      "epoch": 2.098973564997993,
-      "grad_norm": 0.00487825833261013,
-      "learning_rate": 0.00019999782746725798,
-      "loss": 46.0,
-      "step": 27453
-    },
-    {
-      "epoch": 2.0990500219813826,
-      "grad_norm": 0.0009998572058975697,
-      "learning_rate": 0.00019999782730892513,
-      "loss": 46.0,
-      "step": 27454
-    },
-    {
-      "epoch": 2.0991264789647723,
-      "grad_norm": 0.00228071678429842,
-      "learning_rate": 0.0001999978271505865,
-      "loss": 46.0,
-      "step": 27455
-    },
-    {
-      "epoch": 2.099202935948162,
-      "grad_norm": 0.0023862337693572044,
-      "learning_rate": 0.00019999782699224206,
-      "loss": 46.0,
-      "step": 27456
-    },
-    {
-      "epoch": 2.099279392931552,
-      "grad_norm": 0.0014261300675570965,
-      "learning_rate": 0.0001999978268338919,
-      "loss": 46.0,
-      "step": 27457
-    },
-    {
-      "epoch": 2.0993558499149416,
-      "grad_norm": 0.0011226380011066794,
-      "learning_rate": 0.00019999782667553596,
-      "loss": 46.0,
-      "step": 27458
-    },
-    {
-      "epoch": 2.0994323068983314,
-      "grad_norm": 0.001110877376049757,
-      "learning_rate": 0.00019999782651717424,
-      "loss": 46.0,
-      "step": 27459
-    },
-    {
-      "epoch": 2.099508763881721,
-      "grad_norm": 0.0034316410310566425,
-      "learning_rate": 0.00019999782635880676,
-      "loss": 46.0,
-      "step": 27460
-    },
-    {
-      "epoch": 2.099585220865111,
-      "grad_norm": 0.0014385415706783533,
-      "learning_rate": 0.0001999978262004335,
-      "loss": 46.0,
-      "step": 27461
-    },
-    {
-      "epoch": 2.0996616778485007,
-      "grad_norm": 0.006219201255589724,
-      "learning_rate": 0.0001999978260420545,
-      "loss": 46.0,
-      "step": 27462
-    },
-    {
-      "epoch": 2.09973813483189,
-      "grad_norm": 0.0006632667500525713,
-      "learning_rate": 0.00019999782588366968,
-      "loss": 46.0,
-      "step": 27463
-    },
-    {
-      "epoch": 2.0998145918152797,
-      "grad_norm": 0.00036832300247624516,
-      "learning_rate": 0.00019999782572527913,
-      "loss": 46.0,
-      "step": 27464
-    },
-    {
-      "epoch": 2.0998910487986695,
-      "grad_norm": 0.0010592605685815215,
-      "learning_rate": 0.0001999978255668828,
-      "loss": 46.0,
-      "step": 27465
-    },
-    {
-      "epoch": 2.0999675057820593,
-      "grad_norm": 0.00135167280677706,
-      "learning_rate": 0.0001999978254084807,
-      "loss": 46.0,
-      "step": 27466
-    },
-    {
-      "epoch": 2.100043962765449,
-      "grad_norm": 0.0005404517869465053,
-      "learning_rate": 0.00019999782525007283,
-      "loss": 46.0,
-      "step": 27467
-    },
-    {
-      "epoch": 2.100120419748839,
-      "grad_norm": 0.0008791881846264005,
-      "learning_rate": 0.00019999782509165918,
-      "loss": 46.0,
-      "step": 27468
-    },
-    {
-      "epoch": 2.1001968767322285,
-      "grad_norm": 0.0017522121779620647,
-      "learning_rate": 0.00019999782493323976,
-      "loss": 46.0,
-      "step": 27469
-    },
-    {
-      "epoch": 2.1002733337156183,
-      "grad_norm": 0.0006317515508271754,
-      "learning_rate": 0.00019999782477481462,
-      "loss": 46.0,
-      "step": 27470
-    },
-    {
-      "epoch": 2.100349790699008,
-      "grad_norm": 0.0007047151448205113,
-      "learning_rate": 0.00019999782461638366,
-      "loss": 46.0,
-      "step": 27471
-    },
-    {
-      "epoch": 2.100426247682398,
-      "grad_norm": 0.0013135392218828201,
-      "learning_rate": 0.00019999782445794692,
-      "loss": 46.0,
-      "step": 27472
-    },
-    {
-      "epoch": 2.1005027046657876,
-      "grad_norm": 0.002107311738654971,
-      "learning_rate": 0.00019999782429950446,
-      "loss": 46.0,
-      "step": 27473
-    },
-    {
-      "epoch": 2.1005791616491774,
-      "grad_norm": 0.0007449019467458129,
-      "learning_rate": 0.0001999978241410562,
-      "loss": 46.0,
-      "step": 27474
-    },
-    {
-      "epoch": 2.1006556186325667,
-      "grad_norm": 0.0005109795019961894,
-      "learning_rate": 0.0001999978239826022,
-      "loss": 46.0,
-      "step": 27475
-    },
-    {
-      "epoch": 2.1007320756159564,
-      "grad_norm": 0.0011709368554875255,
-      "learning_rate": 0.00019999782382414236,
-      "loss": 46.0,
-      "step": 27476
-    },
-    {
-      "epoch": 2.100808532599346,
-      "grad_norm": 0.0006436691619455814,
-      "learning_rate": 0.0001999978236656768,
-      "loss": 46.0,
-      "step": 27477
-    },
-    {
-      "epoch": 2.100884989582736,
-      "grad_norm": 0.0023157019168138504,
-      "learning_rate": 0.00019999782350720548,
-      "loss": 46.0,
-      "step": 27478
-    },
-    {
-      "epoch": 2.1009614465661257,
-      "grad_norm": 0.0009241512161679566,
-      "learning_rate": 0.00019999782334872838,
-      "loss": 46.0,
-      "step": 27479
-    },
-    {
-      "epoch": 2.1010379035495155,
-      "grad_norm": 0.004700734280049801,
-      "learning_rate": 0.0001999978231902455,
-      "loss": 46.0,
-      "step": 27480
-    },
-    {
-      "epoch": 2.1011143605329052,
-      "grad_norm": 0.0006546802469529212,
-      "learning_rate": 0.00019999782303175686,
-      "loss": 46.0,
-      "step": 27481
-    },
-    {
-      "epoch": 2.101190817516295,
-      "grad_norm": 0.0028919437900185585,
-      "learning_rate": 0.00019999782287326247,
-      "loss": 46.0,
-      "step": 27482
-    },
-    {
-      "epoch": 2.1012672744996848,
-      "grad_norm": 0.00415746821090579,
-      "learning_rate": 0.00019999782271476227,
-      "loss": 46.0,
-      "step": 27483
-    },
-    {
-      "epoch": 2.1013437314830745,
-      "grad_norm": 0.0029722987674176693,
-      "learning_rate": 0.00019999782255625633,
-      "loss": 46.0,
-      "step": 27484
-    },
-    {
-      "epoch": 2.101420188466464,
-      "grad_norm": 0.0011102004209533334,
-      "learning_rate": 0.0001999978223977446,
-      "loss": 46.0,
-      "step": 27485
-    },
-    {
-      "epoch": 2.1014966454498536,
-      "grad_norm": 0.00449033547192812,
-      "learning_rate": 0.0001999978222392271,
-      "loss": 46.0,
-      "step": 27486
-    },
-    {
-      "epoch": 2.1015731024332434,
-      "grad_norm": 0.00145477254409343,
-      "learning_rate": 0.00019999782208070385,
-      "loss": 46.0,
-      "step": 27487
-    },
-    {
-      "epoch": 2.101649559416633,
-      "grad_norm": 0.0008745861705392599,
-      "learning_rate": 0.00019999782192217484,
-      "loss": 46.0,
-      "step": 27488
-    },
-    {
-      "epoch": 2.101726016400023,
-      "grad_norm": 0.003108362201601267,
-      "learning_rate": 0.00019999782176364006,
-      "loss": 46.0,
-      "step": 27489
-    },
-    {
-      "epoch": 2.1018024733834126,
-      "grad_norm": 0.000680555880535394,
-      "learning_rate": 0.00019999782160509948,
-      "loss": 46.0,
-      "step": 27490
-    },
-    {
-      "epoch": 2.1018789303668024,
-      "grad_norm": 0.0017704834463074803,
-      "learning_rate": 0.00019999782144655316,
-      "loss": 46.0,
-      "step": 27491
-    },
-    {
-      "epoch": 2.101955387350192,
-      "grad_norm": 0.0005636248970404267,
-      "learning_rate": 0.00019999782128800103,
-      "loss": 46.0,
-      "step": 27492
-    },
-    {
-      "epoch": 2.102031844333582,
-      "grad_norm": 0.0006046387716196477,
-      "learning_rate": 0.00019999782112944318,
-      "loss": 46.0,
-      "step": 27493
-    },
-    {
-      "epoch": 2.1021083013169717,
-      "grad_norm": 0.001230064663104713,
-      "learning_rate": 0.00019999782097087954,
-      "loss": 46.0,
-      "step": 27494
-    },
-    {
-      "epoch": 2.1021847583003614,
-      "grad_norm": 0.0015667584957554936,
-      "learning_rate": 0.00019999782081231012,
-      "loss": 46.0,
-      "step": 27495
-    },
-    {
-      "epoch": 2.102261215283751,
-      "grad_norm": 0.0008459329837933183,
-      "learning_rate": 0.00019999782065373493,
-      "loss": 46.0,
-      "step": 27496
-    },
-    {
-      "epoch": 2.1023376722671405,
-      "grad_norm": 0.0007667461177334189,
-      "learning_rate": 0.00019999782049515399,
-      "loss": 46.0,
-      "step": 27497
-    },
-    {
-      "epoch": 2.1024141292505303,
-      "grad_norm": 0.0020758130121976137,
-      "learning_rate": 0.00019999782033656727,
-      "loss": 46.0,
-      "step": 27498
-    },
-    {
-      "epoch": 2.10249058623392,
-      "grad_norm": 0.0034676725044846535,
-      "learning_rate": 0.00019999782017797476,
-      "loss": 46.0,
-      "step": 27499
-    },
-    {
-      "epoch": 2.10256704321731,
-      "grad_norm": 0.00042554130777716637,
-      "learning_rate": 0.00019999782001937653,
-      "loss": 46.0,
-      "step": 27500
-    },
-    {
-      "epoch": 2.1026435002006996,
-      "grad_norm": 0.0011001051170751452,
-      "learning_rate": 0.0001999978198607725,
-      "loss": 46.0,
-      "step": 27501
-    },
-    {
-      "epoch": 2.1027199571840893,
-      "grad_norm": 0.0012970971874892712,
-      "learning_rate": 0.00019999781970216272,
-      "loss": 46.0,
-      "step": 27502
-    },
-    {
-      "epoch": 2.102796414167479,
-      "grad_norm": 0.002104746177792549,
-      "learning_rate": 0.00019999781954354714,
-      "loss": 46.0,
-      "step": 27503
-    },
-    {
-      "epoch": 2.102872871150869,
-      "grad_norm": 0.0006146163213998079,
-      "learning_rate": 0.0001999978193849258,
-      "loss": 46.0,
-      "step": 27504
-    },
-    {
-      "epoch": 2.1029493281342586,
-      "grad_norm": 0.00045064990990795195,
-      "learning_rate": 0.0001999978192262987,
-      "loss": 46.0,
-      "step": 27505
-    },
-    {
-      "epoch": 2.1030257851176484,
-      "grad_norm": 0.0005981334252282977,
-      "learning_rate": 0.0001999978190676658,
-      "loss": 46.0,
-      "step": 27506
-    },
-    {
-      "epoch": 2.1031022421010377,
-      "grad_norm": 0.0009634100715629756,
-      "learning_rate": 0.00019999781890902717,
-      "loss": 46.0,
-      "step": 27507
-    },
-    {
-      "epoch": 2.1031786990844274,
-      "grad_norm": 0.0005045508733019233,
-      "learning_rate": 0.00019999781875038278,
-      "loss": 46.0,
-      "step": 27508
-    },
-    {
-      "epoch": 2.103255156067817,
-      "grad_norm": 0.006769275292754173,
-      "learning_rate": 0.00019999781859173256,
-      "loss": 46.0,
-      "step": 27509
-    },
-    {
-      "epoch": 2.103331613051207,
-      "grad_norm": 0.0006632230360992253,
-      "learning_rate": 0.00019999781843307664,
-      "loss": 46.0,
-      "step": 27510
-    },
-    {
-      "epoch": 2.1034080700345967,
-      "grad_norm": 0.0021935333497822285,
-      "learning_rate": 0.00019999781827441493,
-      "loss": 46.0,
-      "step": 27511
-    },
-    {
-      "epoch": 2.1034845270179865,
-      "grad_norm": 0.0008420352824032307,
-      "learning_rate": 0.00019999781811574742,
-      "loss": 46.0,
-      "step": 27512
-    },
-    {
-      "epoch": 2.1035609840013763,
-      "grad_norm": 0.0014697188744321465,
-      "learning_rate": 0.00019999781795707416,
-      "loss": 46.0,
-      "step": 27513
-    },
-    {
-      "epoch": 2.103637440984766,
-      "grad_norm": 0.0006531400722451508,
-      "learning_rate": 0.00019999781779839513,
-      "loss": 46.0,
-      "step": 27514
-    },
-    {
-      "epoch": 2.1037138979681558,
-      "grad_norm": 0.005551649257540703,
-      "learning_rate": 0.00019999781763971032,
-      "loss": 46.0,
-      "step": 27515
-    },
-    {
-      "epoch": 2.1037903549515455,
-      "grad_norm": 0.0033420671243220568,
-      "learning_rate": 0.00019999781748101974,
-      "loss": 46.0,
-      "step": 27516
-    },
-    {
-      "epoch": 2.1038668119349353,
-      "grad_norm": 0.0027969086077064276,
-      "learning_rate": 0.00019999781732232344,
-      "loss": 46.0,
-      "step": 27517
-    },
-    {
-      "epoch": 2.1039432689183246,
-      "grad_norm": 0.00125506438780576,
-      "learning_rate": 0.00019999781716362135,
-      "loss": 46.0,
-      "step": 27518
-    },
-    {
-      "epoch": 2.1040197259017144,
-      "grad_norm": 0.0017572084907442331,
-      "learning_rate": 0.00019999781700491345,
-      "loss": 46.0,
-      "step": 27519
-    },
-    {
-      "epoch": 2.104096182885104,
-      "grad_norm": 0.0008902869885787368,
-      "learning_rate": 0.0001999978168461998,
-      "loss": 46.0,
-      "step": 27520
-    },
-    {
-      "epoch": 2.104172639868494,
-      "grad_norm": 0.0009279965306632221,
-      "learning_rate": 0.00019999781668748038,
-      "loss": 46.0,
-      "step": 27521
-    },
-    {
-      "epoch": 2.1042490968518837,
-      "grad_norm": 0.00045384641271084547,
-      "learning_rate": 0.00019999781652875522,
-      "loss": 46.0,
-      "step": 27522
-    },
-    {
-      "epoch": 2.1043255538352734,
-      "grad_norm": 0.0006511066458187997,
-      "learning_rate": 0.00019999781637002425,
-      "loss": 46.0,
-      "step": 27523
-    },
-    {
-      "epoch": 2.104402010818663,
-      "grad_norm": 0.0017435166519135237,
-      "learning_rate": 0.0001999978162112875,
-      "loss": 46.0,
-      "step": 27524
-    },
-    {
-      "epoch": 2.104478467802053,
-      "grad_norm": 0.0006301420507952571,
-      "learning_rate": 0.00019999781605254503,
-      "loss": 46.0,
-      "step": 27525
-    },
-    {
-      "epoch": 2.1045549247854427,
-      "grad_norm": 0.0015568938106298447,
-      "learning_rate": 0.00019999781589379677,
-      "loss": 46.0,
-      "step": 27526
-    },
-    {
-      "epoch": 2.1046313817688325,
-      "grad_norm": 0.0011748585384339094,
-      "learning_rate": 0.00019999781573504274,
-      "loss": 46.0,
-      "step": 27527
-    },
-    {
-      "epoch": 2.1047078387522222,
-      "grad_norm": 0.0005240196478553116,
-      "learning_rate": 0.00019999781557628293,
-      "loss": 46.0,
-      "step": 27528
-    },
-    {
-      "epoch": 2.1047842957356115,
-      "grad_norm": 0.00037409065407700837,
-      "learning_rate": 0.00019999781541751738,
-      "loss": 46.0,
-      "step": 27529
-    },
-    {
-      "epoch": 2.1048607527190013,
-      "grad_norm": 0.0031022352632135153,
-      "learning_rate": 0.00019999781525874605,
-      "loss": 46.0,
-      "step": 27530
-    },
-    {
-      "epoch": 2.104937209702391,
-      "grad_norm": 0.0019445521757006645,
-      "learning_rate": 0.00019999781509996893,
-      "loss": 46.0,
-      "step": 27531
-    },
-    {
-      "epoch": 2.105013666685781,
-      "grad_norm": 0.001530882902443409,
-      "learning_rate": 0.00019999781494118608,
-      "loss": 46.0,
-      "step": 27532
-    },
-    {
-      "epoch": 2.1050901236691706,
-      "grad_norm": 0.0040167272090911865,
-      "learning_rate": 0.0001999978147823974,
-      "loss": 46.0,
-      "step": 27533
-    },
-    {
-      "epoch": 2.1051665806525603,
-      "grad_norm": 0.0012179792392998934,
-      "learning_rate": 0.00019999781462360302,
-      "loss": 46.0,
-      "step": 27534
-    },
-    {
-      "epoch": 2.10524303763595,
-      "grad_norm": 0.0013874232536181808,
-      "learning_rate": 0.00019999781446480283,
-      "loss": 46.0,
-      "step": 27535
-    },
-    {
-      "epoch": 2.10531949461934,
-      "grad_norm": 0.004843584261834621,
-      "learning_rate": 0.00019999781430599686,
-      "loss": 46.0,
-      "step": 27536
-    },
-    {
-      "epoch": 2.1053959516027296,
-      "grad_norm": 0.003989243879914284,
-      "learning_rate": 0.00019999781414718515,
-      "loss": 46.0,
-      "step": 27537
-    },
-    {
-      "epoch": 2.1054724085861194,
-      "grad_norm": 0.0006604177760891616,
-      "learning_rate": 0.00019999781398836767,
-      "loss": 46.0,
-      "step": 27538
-    },
-    {
-      "epoch": 2.105548865569509,
-      "grad_norm": 0.0009026466286741197,
-      "learning_rate": 0.00019999781382954438,
-      "loss": 46.0,
-      "step": 27539
-    },
-    {
-      "epoch": 2.1056253225528985,
-      "grad_norm": 0.0026330859400331974,
-      "learning_rate": 0.00019999781367071538,
-      "loss": 46.0,
-      "step": 27540
-    },
-    {
-      "epoch": 2.1057017795362882,
-      "grad_norm": 0.001139921136200428,
-      "learning_rate": 0.00019999781351188057,
-      "loss": 46.0,
-      "step": 27541
-    },
-    {
-      "epoch": 2.105778236519678,
-      "grad_norm": 0.0007302314625121653,
-      "learning_rate": 0.00019999781335304,
-      "loss": 46.0,
-      "step": 27542
-    },
-    {
-      "epoch": 2.1058546935030678,
-      "grad_norm": 0.001985919428989291,
-      "learning_rate": 0.00019999781319419367,
-      "loss": 46.0,
-      "step": 27543
-    },
-    {
-      "epoch": 2.1059311504864575,
-      "grad_norm": 0.0006481681484729052,
-      "learning_rate": 0.00019999781303534154,
-      "loss": 46.0,
-      "step": 27544
-    },
-    {
-      "epoch": 2.1060076074698473,
-      "grad_norm": 0.0011985978344455361,
-      "learning_rate": 0.00019999781287648367,
-      "loss": 46.0,
-      "step": 27545
-    },
-    {
-      "epoch": 2.106084064453237,
-      "grad_norm": 0.0006719738594256341,
-      "learning_rate": 0.00019999781271762003,
-      "loss": 46.0,
-      "step": 27546
-    },
-    {
-      "epoch": 2.106160521436627,
-      "grad_norm": 0.0028443941846489906,
-      "learning_rate": 0.0001999978125587506,
-      "loss": 46.0,
-      "step": 27547
-    },
-    {
-      "epoch": 2.1062369784200166,
-      "grad_norm": 0.0007125627598725259,
-      "learning_rate": 0.00019999781239987542,
-      "loss": 46.0,
-      "step": 27548
-    },
-    {
-      "epoch": 2.1063134354034063,
-      "grad_norm": 0.000829347234684974,
-      "learning_rate": 0.00019999781224099445,
-      "loss": 46.0,
-      "step": 27549
-    },
-    {
-      "epoch": 2.106389892386796,
-      "grad_norm": 0.003277435200288892,
-      "learning_rate": 0.00019999781208210774,
-      "loss": 46.0,
-      "step": 27550
-    },
-    {
-      "epoch": 2.1064663493701854,
-      "grad_norm": 0.0051928856410086155,
-      "learning_rate": 0.00019999781192321525,
-      "loss": 46.0,
-      "step": 27551
-    },
-    {
-      "epoch": 2.106542806353575,
-      "grad_norm": 0.0010621713008731604,
-      "learning_rate": 0.000199997811764317,
-      "loss": 46.0,
-      "step": 27552
-    },
-    {
-      "epoch": 2.106619263336965,
-      "grad_norm": 0.0022010935936123133,
-      "learning_rate": 0.00019999781160541297,
-      "loss": 46.0,
-      "step": 27553
-    },
-    {
-      "epoch": 2.1066957203203547,
-      "grad_norm": 0.010605143383145332,
-      "learning_rate": 0.00019999781144650313,
-      "loss": 46.0,
-      "step": 27554
-    },
-    {
-      "epoch": 2.1067721773037444,
-      "grad_norm": 0.001072065089829266,
-      "learning_rate": 0.00019999781128758758,
-      "loss": 46.0,
-      "step": 27555
-    },
-    {
-      "epoch": 2.106848634287134,
-      "grad_norm": 0.002340103732421994,
-      "learning_rate": 0.00019999781112866623,
-      "loss": 46.0,
-      "step": 27556
-    },
-    {
-      "epoch": 2.106925091270524,
-      "grad_norm": 0.0019150434527546167,
-      "learning_rate": 0.00019999781096973913,
-      "loss": 46.0,
-      "step": 27557
-    },
-    {
-      "epoch": 2.1070015482539137,
-      "grad_norm": 0.0013690509367734194,
-      "learning_rate": 0.00019999781081080626,
-      "loss": 46.0,
-      "step": 27558
-    },
-    {
-      "epoch": 2.1070780052373035,
-      "grad_norm": 0.000352658040355891,
-      "learning_rate": 0.0001999978106518676,
-      "loss": 46.0,
-      "step": 27559
-    },
-    {
-      "epoch": 2.1071544622206932,
-      "grad_norm": 0.0019908156245946884,
-      "learning_rate": 0.0001999978104929232,
-      "loss": 46.0,
-      "step": 27560
-    },
-    {
-      "epoch": 2.107230919204083,
-      "grad_norm": 0.005864780396223068,
-      "learning_rate": 0.000199997810333973,
-      "loss": 46.0,
-      "step": 27561
-    },
-    {
-      "epoch": 2.1073073761874723,
-      "grad_norm": 0.000879982253536582,
-      "learning_rate": 0.00019999781017501704,
-      "loss": 46.0,
-      "step": 27562
-    },
-    {
-      "epoch": 2.107383833170862,
-      "grad_norm": 0.0015050513902679086,
-      "learning_rate": 0.00019999781001605533,
-      "loss": 46.0,
-      "step": 27563
-    },
-    {
-      "epoch": 2.107460290154252,
-      "grad_norm": 0.0012313692132011056,
-      "learning_rate": 0.00019999780985708782,
-      "loss": 46.0,
-      "step": 27564
-    },
-    {
-      "epoch": 2.1075367471376416,
-      "grad_norm": 0.00040437025018036366,
-      "learning_rate": 0.00019999780969811454,
-      "loss": 46.0,
-      "step": 27565
-    },
-    {
-      "epoch": 2.1076132041210314,
-      "grad_norm": 0.0006390642956830561,
-      "learning_rate": 0.00019999780953913553,
-      "loss": 46.0,
-      "step": 27566
-    },
-    {
-      "epoch": 2.107689661104421,
-      "grad_norm": 0.000998621340841055,
-      "learning_rate": 0.0001999978093801507,
-      "loss": 46.0,
-      "step": 27567
-    },
-    {
-      "epoch": 2.107766118087811,
-      "grad_norm": 0.0014330863486975431,
-      "learning_rate": 0.00019999780922116015,
-      "loss": 46.0,
-      "step": 27568
-    },
-    {
-      "epoch": 2.1078425750712007,
-      "grad_norm": 0.0013231082120910287,
-      "learning_rate": 0.0001999978090621638,
-      "loss": 46.0,
-      "step": 27569
-    },
-    {
-      "epoch": 2.1079190320545904,
-      "grad_norm": 0.000830391189083457,
-      "learning_rate": 0.0001999978089031617,
-      "loss": 46.0,
-      "step": 27570
-    },
-    {
-      "epoch": 2.10799548903798,
-      "grad_norm": 0.002778175286948681,
-      "learning_rate": 0.0001999978087441538,
-      "loss": 46.0,
-      "step": 27571
-    },
-    {
-      "epoch": 2.10807194602137,
-      "grad_norm": 0.0007058788323774934,
-      "learning_rate": 0.00019999780858514016,
-      "loss": 46.0,
-      "step": 27572
-    },
-    {
-      "epoch": 2.1081484030047593,
-      "grad_norm": 0.0005704647046513855,
-      "learning_rate": 0.00019999780842612074,
-      "loss": 46.0,
-      "step": 27573
-    },
-    {
-      "epoch": 2.108224859988149,
-      "grad_norm": 0.0006529319216497242,
-      "learning_rate": 0.00019999780826709555,
-      "loss": 46.0,
-      "step": 27574
-    },
-    {
-      "epoch": 2.1083013169715388,
-      "grad_norm": 0.010321738198399544,
-      "learning_rate": 0.00019999780810806459,
-      "loss": 46.0,
-      "step": 27575
-    },
-    {
-      "epoch": 2.1083777739549285,
-      "grad_norm": 0.0006926292553544044,
-      "learning_rate": 0.0001999978079490279,
-      "loss": 46.0,
-      "step": 27576
-    },
-    {
-      "epoch": 2.1084542309383183,
-      "grad_norm": 0.0008111309725791216,
-      "learning_rate": 0.0001999978077899854,
-      "loss": 46.0,
-      "step": 27577
-    },
-    {
-      "epoch": 2.108530687921708,
-      "grad_norm": 0.0013768266653642058,
-      "learning_rate": 0.0001999978076309371,
-      "loss": 46.0,
-      "step": 27578
-    },
-    {
-      "epoch": 2.108607144905098,
-      "grad_norm": 0.002121939091011882,
-      "learning_rate": 0.00019999780747188308,
-      "loss": 46.0,
-      "step": 27579
-    },
-    {
-      "epoch": 2.1086836018884876,
-      "grad_norm": 0.0009338922682218254,
-      "learning_rate": 0.0001999978073128233,
-      "loss": 46.0,
-      "step": 27580
-    },
-    {
-      "epoch": 2.1087600588718773,
-      "grad_norm": 0.0012911935336887836,
-      "learning_rate": 0.0001999978071537577,
-      "loss": 46.0,
-      "step": 27581
-    },
-    {
-      "epoch": 2.108836515855267,
-      "grad_norm": 0.0007539521902799606,
-      "learning_rate": 0.00019999780699468634,
-      "loss": 46.0,
-      "step": 27582
-    },
-    {
-      "epoch": 2.108912972838657,
-      "grad_norm": 0.0004395316354930401,
-      "learning_rate": 0.00019999780683560925,
-      "loss": 46.0,
-      "step": 27583
-    },
-    {
-      "epoch": 2.108989429822046,
-      "grad_norm": 0.0007877614698372781,
-      "learning_rate": 0.00019999780667652637,
-      "loss": 46.0,
-      "step": 27584
-    },
-    {
-      "epoch": 2.109065886805436,
-      "grad_norm": 0.00099780666641891,
-      "learning_rate": 0.0001999978065174377,
-      "loss": 46.0,
-      "step": 27585
-    },
-    {
-      "epoch": 2.1091423437888257,
-      "grad_norm": 0.0016647278098389506,
-      "learning_rate": 0.00019999780635834329,
-      "loss": 46.0,
-      "step": 27586
-    },
-    {
-      "epoch": 2.1092188007722155,
-      "grad_norm": 0.004062425810843706,
-      "learning_rate": 0.00019999780619924312,
-      "loss": 46.0,
-      "step": 27587
-    },
-    {
-      "epoch": 2.1092952577556052,
-      "grad_norm": 0.0009334741625934839,
-      "learning_rate": 0.00019999780604013716,
-      "loss": 46.0,
-      "step": 27588
-    },
-    {
-      "epoch": 2.109371714738995,
-      "grad_norm": 0.0007265450549311936,
-      "learning_rate": 0.00019999780588102542,
-      "loss": 46.0,
-      "step": 27589
-    },
-    {
-      "epoch": 2.1094481717223847,
-      "grad_norm": 0.0015771707985550165,
-      "learning_rate": 0.00019999780572190794,
-      "loss": 46.0,
-      "step": 27590
-    },
-    {
-      "epoch": 2.1095246287057745,
-      "grad_norm": 0.0020960504189133644,
-      "learning_rate": 0.00019999780556278465,
-      "loss": 46.0,
-      "step": 27591
-    },
-    {
-      "epoch": 2.1096010856891643,
-      "grad_norm": 0.0007886430830694735,
-      "learning_rate": 0.00019999780540365562,
-      "loss": 46.0,
-      "step": 27592
-    },
-    {
-      "epoch": 2.109677542672554,
-      "grad_norm": 0.0016777862329035997,
-      "learning_rate": 0.00019999780524452082,
-      "loss": 46.0,
-      "step": 27593
-    },
-    {
-      "epoch": 2.1097539996559433,
-      "grad_norm": 0.0007015528390184045,
-      "learning_rate": 0.00019999780508538024,
-      "loss": 46.0,
-      "step": 27594
-    },
-    {
-      "epoch": 2.109830456639333,
-      "grad_norm": 0.0005086369928903878,
-      "learning_rate": 0.0001999978049262339,
-      "loss": 46.0,
-      "step": 27595
-    },
-    {
-      "epoch": 2.109906913622723,
-      "grad_norm": 0.0009788682218641043,
-      "learning_rate": 0.0001999978047670818,
-      "loss": 46.0,
-      "step": 27596
-    },
-    {
-      "epoch": 2.1099833706061126,
-      "grad_norm": 0.0019625723361968994,
-      "learning_rate": 0.0001999978046079239,
-      "loss": 46.0,
-      "step": 27597
-    },
-    {
-      "epoch": 2.1100598275895024,
-      "grad_norm": 0.0022894656285643578,
-      "learning_rate": 0.00019999780444876025,
-      "loss": 46.0,
-      "step": 27598
-    },
-    {
-      "epoch": 2.110136284572892,
-      "grad_norm": 0.003526546061038971,
-      "learning_rate": 0.00019999780428959084,
-      "loss": 46.0,
-      "step": 27599
-    },
-    {
-      "epoch": 2.110212741556282,
-      "grad_norm": 0.004093230236321688,
-      "learning_rate": 0.00019999780413041562,
-      "loss": 46.0,
-      "step": 27600
-    },
-    {
-      "epoch": 2.1102891985396717,
-      "grad_norm": 0.0010514372261241078,
-      "learning_rate": 0.00019999780397123468,
-      "loss": 46.0,
-      "step": 27601
-    },
-    {
-      "epoch": 2.1103656555230614,
-      "grad_norm": 0.0004950147122144699,
-      "learning_rate": 0.00019999780381204794,
-      "loss": 46.0,
-      "step": 27602
-    },
-    {
-      "epoch": 2.110442112506451,
-      "grad_norm": 0.0010895001469179988,
-      "learning_rate": 0.00019999780365285546,
-      "loss": 46.0,
-      "step": 27603
-    },
-    {
-      "epoch": 2.110518569489841,
-      "grad_norm": 0.003022734308615327,
-      "learning_rate": 0.0001999978034936572,
-      "loss": 46.0,
-      "step": 27604
-    },
-    {
-      "epoch": 2.1105950264732307,
-      "grad_norm": 0.002471886109560728,
-      "learning_rate": 0.00019999780333445315,
-      "loss": 46.0,
-      "step": 27605
-    },
-    {
-      "epoch": 2.11067148345662,
-      "grad_norm": 0.002287641167640686,
-      "learning_rate": 0.00019999780317524332,
-      "loss": 46.0,
-      "step": 27606
-    },
-    {
-      "epoch": 2.11074794044001,
-      "grad_norm": 0.002804981777444482,
-      "learning_rate": 0.00019999780301602774,
-      "loss": 46.0,
-      "step": 27607
-    },
-    {
-      "epoch": 2.1108243974233996,
-      "grad_norm": 0.001787775312550366,
-      "learning_rate": 0.00019999780285680642,
-      "loss": 46.0,
-      "step": 27608
-    },
-    {
-      "epoch": 2.1109008544067893,
-      "grad_norm": 0.0010987324640154839,
-      "learning_rate": 0.0001999978026975793,
-      "loss": 46.0,
-      "step": 27609
-    },
-    {
-      "epoch": 2.110977311390179,
-      "grad_norm": 0.001675209030508995,
-      "learning_rate": 0.00019999780253834642,
-      "loss": 46.0,
-      "step": 27610
-    },
-    {
-      "epoch": 2.111053768373569,
-      "grad_norm": 0.00035454094177111983,
-      "learning_rate": 0.00019999780237910778,
-      "loss": 46.0,
-      "step": 27611
-    },
-    {
-      "epoch": 2.1111302253569586,
-      "grad_norm": 0.0007373957778327167,
-      "learning_rate": 0.00019999780221986334,
-      "loss": 46.0,
-      "step": 27612
-    },
-    {
-      "epoch": 2.1112066823403484,
-      "grad_norm": 0.0004193242348264903,
-      "learning_rate": 0.00019999780206061317,
-      "loss": 46.0,
-      "step": 27613
-    },
-    {
-      "epoch": 2.111283139323738,
-      "grad_norm": 0.002905827248468995,
-      "learning_rate": 0.00019999780190135718,
-      "loss": 46.0,
-      "step": 27614
-    },
-    {
-      "epoch": 2.111359596307128,
-      "grad_norm": 0.0011484738206490874,
-      "learning_rate": 0.00019999780174209545,
-      "loss": 46.0,
-      "step": 27615
-    },
-    {
-      "epoch": 2.111436053290517,
-      "grad_norm": 0.0032124840654432774,
-      "learning_rate": 0.00019999780158282796,
-      "loss": 46.0,
-      "step": 27616
-    },
-    {
-      "epoch": 2.111512510273907,
-      "grad_norm": 0.0008766617393121123,
-      "learning_rate": 0.00019999780142355468,
-      "loss": 46.0,
-      "step": 27617
-    },
-    {
-      "epoch": 2.1115889672572967,
-      "grad_norm": 0.0007554818294011056,
-      "learning_rate": 0.00019999780126427565,
-      "loss": 46.0,
-      "step": 27618
-    },
-    {
-      "epoch": 2.1116654242406865,
-      "grad_norm": 0.000889607414137572,
-      "learning_rate": 0.00019999780110499085,
-      "loss": 46.0,
-      "step": 27619
-    },
-    {
-      "epoch": 2.1117418812240762,
-      "grad_norm": 0.0013870007824152708,
-      "learning_rate": 0.00019999780094570027,
-      "loss": 46.0,
-      "step": 27620
-    },
-    {
-      "epoch": 2.111818338207466,
-      "grad_norm": 0.0020611132495105267,
-      "learning_rate": 0.00019999780078640392,
-      "loss": 46.0,
-      "step": 27621
-    },
-    {
-      "epoch": 2.1118947951908558,
-      "grad_norm": 0.006050414405763149,
-      "learning_rate": 0.0001999978006271018,
-      "loss": 46.0,
-      "step": 27622
-    },
-    {
-      "epoch": 2.1119712521742455,
-      "grad_norm": 0.0011397033231332898,
-      "learning_rate": 0.0001999978004677939,
-      "loss": 46.0,
-      "step": 27623
-    },
-    {
-      "epoch": 2.1120477091576353,
-      "grad_norm": 0.007902242243289948,
-      "learning_rate": 0.00019999780030848028,
-      "loss": 46.0,
-      "step": 27624
-    },
-    {
-      "epoch": 2.112124166141025,
-      "grad_norm": 0.0010233998764306307,
-      "learning_rate": 0.00019999780014916084,
-      "loss": 46.0,
-      "step": 27625
-    },
-    {
-      "epoch": 2.112200623124415,
-      "grad_norm": 0.002083076862618327,
-      "learning_rate": 0.00019999779998983565,
-      "loss": 46.0,
-      "step": 27626
-    },
-    {
-      "epoch": 2.1122770801078046,
-      "grad_norm": 0.006393703632056713,
-      "learning_rate": 0.0001999977998305047,
-      "loss": 46.0,
-      "step": 27627
-    },
-    {
-      "epoch": 2.112353537091194,
-      "grad_norm": 0.0009162043570540845,
-      "learning_rate": 0.00019999779967116795,
-      "loss": 46.0,
-      "step": 27628
-    },
-    {
-      "epoch": 2.1124299940745837,
-      "grad_norm": 0.000548450043424964,
-      "learning_rate": 0.00019999779951182544,
-      "loss": 46.0,
-      "step": 27629
-    },
-    {
-      "epoch": 2.1125064510579734,
-      "grad_norm": 0.0003302285913378,
-      "learning_rate": 0.00019999779935247718,
-      "loss": 46.0,
-      "step": 27630
-    },
-    {
-      "epoch": 2.112582908041363,
-      "grad_norm": 0.002078081015497446,
-      "learning_rate": 0.00019999779919312313,
-      "loss": 46.0,
-      "step": 27631
-    },
-    {
-      "epoch": 2.112659365024753,
-      "grad_norm": 0.0022325501777231693,
-      "learning_rate": 0.00019999779903376332,
-      "loss": 46.0,
-      "step": 27632
-    },
-    {
-      "epoch": 2.1127358220081427,
-      "grad_norm": 0.0017688550287857652,
-      "learning_rate": 0.00019999779887439775,
-      "loss": 46.0,
-      "step": 27633
-    },
-    {
-      "epoch": 2.1128122789915325,
-      "grad_norm": 0.0003088069788645953,
-      "learning_rate": 0.00019999779871502643,
-      "loss": 46.0,
-      "step": 27634
-    },
-    {
-      "epoch": 2.112888735974922,
-      "grad_norm": 0.0013076105387881398,
-      "learning_rate": 0.00019999779855564928,
-      "loss": 46.0,
-      "step": 27635
-    },
-    {
-      "epoch": 2.112965192958312,
-      "grad_norm": 0.0015641204081475735,
-      "learning_rate": 0.0001999977983962664,
-      "loss": 46.0,
-      "step": 27636
-    },
-    {
-      "epoch": 2.1130416499417017,
-      "grad_norm": 0.008307659067213535,
-      "learning_rate": 0.00019999779823687774,
-      "loss": 46.0,
-      "step": 27637
-    },
-    {
-      "epoch": 2.113118106925091,
-      "grad_norm": 0.013898754492402077,
-      "learning_rate": 0.00019999779807748332,
-      "loss": 46.0,
-      "step": 27638
-    },
-    {
-      "epoch": 2.113194563908481,
-      "grad_norm": 0.0004593391786329448,
-      "learning_rate": 0.00019999779791808313,
-      "loss": 46.0,
-      "step": 27639
-    },
-    {
-      "epoch": 2.1132710208918706,
-      "grad_norm": 0.0010538318892940879,
-      "learning_rate": 0.00019999779775867714,
-      "loss": 46.0,
-      "step": 27640
-    },
-    {
-      "epoch": 2.1133474778752603,
-      "grad_norm": 0.0004920208011753857,
-      "learning_rate": 0.00019999779759926543,
-      "loss": 46.0,
-      "step": 27641
-    },
-    {
-      "epoch": 2.11342393485865,
-      "grad_norm": 0.002719099400565028,
-      "learning_rate": 0.00019999779743984792,
-      "loss": 46.0,
-      "step": 27642
-    },
-    {
-      "epoch": 2.11350039184204,
-      "grad_norm": 0.001338088302873075,
-      "learning_rate": 0.00019999779728042464,
-      "loss": 46.0,
-      "step": 27643
-    },
-    {
-      "epoch": 2.1135768488254296,
-      "grad_norm": 0.001461145468056202,
-      "learning_rate": 0.0001999977971209956,
-      "loss": 46.0,
-      "step": 27644
-    },
-    {
-      "epoch": 2.1136533058088194,
-      "grad_norm": 0.0006191398133523762,
-      "learning_rate": 0.00019999779696156083,
-      "loss": 46.0,
-      "step": 27645
-    },
-    {
-      "epoch": 2.113729762792209,
-      "grad_norm": 0.0007986398413777351,
-      "learning_rate": 0.00019999779680212023,
-      "loss": 46.0,
-      "step": 27646
-    },
-    {
-      "epoch": 2.113806219775599,
-      "grad_norm": 0.0018254269380122423,
-      "learning_rate": 0.00019999779664267388,
-      "loss": 46.0,
-      "step": 27647
-    },
-    {
-      "epoch": 2.1138826767589887,
-      "grad_norm": 0.0023519706446677446,
-      "learning_rate": 0.00019999779648322176,
-      "loss": 46.0,
-      "step": 27648
-    },
-    {
-      "epoch": 2.113959133742378,
-      "grad_norm": 0.0011385384714230895,
-      "learning_rate": 0.00019999779632376386,
-      "loss": 46.0,
-      "step": 27649
-    },
-    {
-      "epoch": 2.1140355907257677,
-      "grad_norm": 0.0014540132833644748,
-      "learning_rate": 0.00019999779616430022,
-      "loss": 46.0,
-      "step": 27650
-    },
-    {
-      "epoch": 2.1141120477091575,
-      "grad_norm": 0.0033372421748936176,
-      "learning_rate": 0.0001999977960048308,
-      "loss": 46.0,
-      "step": 27651
-    },
-    {
-      "epoch": 2.1141885046925473,
-      "grad_norm": 0.00061256461776793,
-      "learning_rate": 0.00019999779584535562,
-      "loss": 46.0,
-      "step": 27652
-    },
-    {
-      "epoch": 2.114264961675937,
-      "grad_norm": 0.0029663462191820145,
-      "learning_rate": 0.00019999779568587465,
-      "loss": 46.0,
-      "step": 27653
-    },
-    {
-      "epoch": 2.114341418659327,
-      "grad_norm": 0.004290277138352394,
-      "learning_rate": 0.00019999779552638792,
-      "loss": 46.0,
-      "step": 27654
-    },
-    {
-      "epoch": 2.1144178756427165,
-      "grad_norm": 0.0018869865452870727,
-      "learning_rate": 0.0001999977953668954,
-      "loss": 46.0,
-      "step": 27655
-    },
-    {
-      "epoch": 2.1144943326261063,
-      "grad_norm": 0.0005279462784528732,
-      "learning_rate": 0.00019999779520739713,
-      "loss": 46.0,
-      "step": 27656
-    },
-    {
-      "epoch": 2.114570789609496,
-      "grad_norm": 0.0009307373547926545,
-      "learning_rate": 0.00019999779504789307,
-      "loss": 46.0,
-      "step": 27657
-    },
-    {
-      "epoch": 2.114647246592886,
-      "grad_norm": 0.0015755825443193316,
-      "learning_rate": 0.00019999779488838327,
-      "loss": 46.0,
-      "step": 27658
-    },
-    {
-      "epoch": 2.1147237035762756,
-      "grad_norm": 0.0016170954331755638,
-      "learning_rate": 0.0001999977947288677,
-      "loss": 46.0,
-      "step": 27659
-    },
-    {
-      "epoch": 2.114800160559665,
-      "grad_norm": 0.0008384857792407274,
-      "learning_rate": 0.00019999779456934634,
-      "loss": 46.0,
-      "step": 27660
-    },
-    {
-      "epoch": 2.1148766175430547,
-      "grad_norm": 0.0005755027523264289,
-      "learning_rate": 0.0001999977944098192,
-      "loss": 46.0,
-      "step": 27661
-    },
-    {
-      "epoch": 2.1149530745264444,
-      "grad_norm": 0.0016407616203650832,
-      "learning_rate": 0.00019999779425028632,
-      "loss": 46.0,
-      "step": 27662
-    },
-    {
-      "epoch": 2.115029531509834,
-      "grad_norm": 0.0009915753034874797,
-      "learning_rate": 0.00019999779409074768,
-      "loss": 46.0,
-      "step": 27663
-    },
-    {
-      "epoch": 2.115105988493224,
-      "grad_norm": 0.000602319254539907,
-      "learning_rate": 0.00019999779393120324,
-      "loss": 46.0,
-      "step": 27664
-    },
-    {
-      "epoch": 2.1151824454766137,
-      "grad_norm": 0.0016069767298176885,
-      "learning_rate": 0.00019999779377165305,
-      "loss": 46.0,
-      "step": 27665
-    },
-    {
-      "epoch": 2.1152589024600035,
-      "grad_norm": 0.004144642502069473,
-      "learning_rate": 0.00019999779361209706,
-      "loss": 46.0,
-      "step": 27666
-    },
-    {
-      "epoch": 2.1153353594433932,
-      "grad_norm": 0.0019507389515638351,
-      "learning_rate": 0.00019999779345253535,
-      "loss": 46.0,
-      "step": 27667
-    },
-    {
-      "epoch": 2.115411816426783,
-      "grad_norm": 0.0011293807765468955,
-      "learning_rate": 0.00019999779329296784,
-      "loss": 46.0,
-      "step": 27668
-    },
-    {
-      "epoch": 2.1154882734101728,
-      "grad_norm": 0.001420993241481483,
-      "learning_rate": 0.00019999779313339456,
-      "loss": 46.0,
-      "step": 27669
-    },
-    {
-      "epoch": 2.1155647303935625,
-      "grad_norm": 0.0008896767976693809,
-      "learning_rate": 0.00019999779297381553,
-      "loss": 46.0,
-      "step": 27670
-    },
-    {
-      "epoch": 2.115641187376952,
-      "grad_norm": 0.0018266815459355712,
-      "learning_rate": 0.0001999977928142307,
-      "loss": 46.0,
-      "step": 27671
-    },
-    {
-      "epoch": 2.1157176443603416,
-      "grad_norm": 0.0008133428054861724,
-      "learning_rate": 0.00019999779265464013,
-      "loss": 46.0,
-      "step": 27672
-    },
-    {
-      "epoch": 2.1157941013437314,
-      "grad_norm": 0.00037843038444407284,
-      "learning_rate": 0.00019999779249504375,
-      "loss": 46.0,
-      "step": 27673
-    },
-    {
-      "epoch": 2.115870558327121,
-      "grad_norm": 0.0010723500745370984,
-      "learning_rate": 0.00019999779233544166,
-      "loss": 46.0,
-      "step": 27674
-    },
-    {
-      "epoch": 2.115947015310511,
-      "grad_norm": 0.000778036832343787,
-      "learning_rate": 0.00019999779217583376,
-      "loss": 46.0,
-      "step": 27675
-    },
-    {
-      "epoch": 2.1160234722939006,
-      "grad_norm": 0.0008204008336178958,
-      "learning_rate": 0.00019999779201622012,
-      "loss": 46.0,
-      "step": 27676
-    },
-    {
-      "epoch": 2.1160999292772904,
-      "grad_norm": 0.004079892765730619,
-      "learning_rate": 0.00019999779185660068,
-      "loss": 46.0,
-      "step": 27677
-    },
-    {
-      "epoch": 2.11617638626068,
-      "grad_norm": 0.0034579394850879908,
-      "learning_rate": 0.0001999977916969755,
-      "loss": 46.0,
-      "step": 27678
-    },
-    {
-      "epoch": 2.11625284324407,
-      "grad_norm": 0.0021807660814374685,
-      "learning_rate": 0.0001999977915373445,
-      "loss": 46.0,
-      "step": 27679
-    },
-    {
-      "epoch": 2.1163293002274597,
-      "grad_norm": 0.0017785833915695548,
-      "learning_rate": 0.00019999779137770777,
-      "loss": 46.0,
-      "step": 27680
-    },
-    {
-      "epoch": 2.1164057572108494,
-      "grad_norm": 0.001374895335175097,
-      "learning_rate": 0.00019999779121806528,
-      "loss": 46.0,
-      "step": 27681
-    },
-    {
-      "epoch": 2.1164822141942388,
-      "grad_norm": 0.0007049769628793001,
-      "learning_rate": 0.00019999779105841698,
-      "loss": 46.0,
-      "step": 27682
-    },
-    {
-      "epoch": 2.1165586711776285,
-      "grad_norm": 0.002382354810833931,
-      "learning_rate": 0.00019999779089876295,
-      "loss": 46.0,
-      "step": 27683
-    },
-    {
-      "epoch": 2.1166351281610183,
-      "grad_norm": 0.0006863875314593315,
-      "learning_rate": 0.00019999779073910312,
-      "loss": 46.0,
-      "step": 27684
-    },
-    {
-      "epoch": 2.116711585144408,
-      "grad_norm": 0.0005150326760485768,
-      "learning_rate": 0.00019999779057943754,
-      "loss": 46.0,
-      "step": 27685
-    },
-    {
-      "epoch": 2.116788042127798,
-      "grad_norm": 0.0013440201291814446,
-      "learning_rate": 0.0001999977904197662,
-      "loss": 46.0,
-      "step": 27686
-    },
-    {
-      "epoch": 2.1168644991111876,
-      "grad_norm": 0.0017601497238501906,
-      "learning_rate": 0.00019999779026008907,
-      "loss": 46.0,
-      "step": 27687
-    },
-    {
-      "epoch": 2.1169409560945773,
-      "grad_norm": 0.0005629589431919158,
-      "learning_rate": 0.00019999779010040618,
-      "loss": 46.0,
-      "step": 27688
-    },
-    {
-      "epoch": 2.117017413077967,
-      "grad_norm": 0.000687082763761282,
-      "learning_rate": 0.0001999977899407175,
-      "loss": 46.0,
-      "step": 27689
-    },
-    {
-      "epoch": 2.117093870061357,
-      "grad_norm": 0.0006971581024117768,
-      "learning_rate": 0.0001999977897810231,
-      "loss": 46.0,
-      "step": 27690
-    },
-    {
-      "epoch": 2.1171703270447466,
-      "grad_norm": 0.0018247404368594289,
-      "learning_rate": 0.0001999977896213229,
-      "loss": 46.0,
-      "step": 27691
-    },
-    {
-      "epoch": 2.1172467840281364,
-      "grad_norm": 0.002912582363933325,
-      "learning_rate": 0.00019999778946161692,
-      "loss": 46.0,
-      "step": 27692
-    },
-    {
-      "epoch": 2.1173232410115257,
-      "grad_norm": 0.0009077168069779873,
-      "learning_rate": 0.00019999778930190518,
-      "loss": 46.0,
-      "step": 27693
-    },
-    {
-      "epoch": 2.1173996979949155,
-      "grad_norm": 0.00042187661165371537,
-      "learning_rate": 0.00019999778914218767,
-      "loss": 46.0,
-      "step": 27694
-    },
-    {
-      "epoch": 2.117476154978305,
-      "grad_norm": 0.0013393854023888707,
-      "learning_rate": 0.0001999977889824644,
-      "loss": 46.0,
-      "step": 27695
-    },
-    {
-      "epoch": 2.117552611961695,
-      "grad_norm": 0.0012687164125964046,
-      "learning_rate": 0.00019999778882273534,
-      "loss": 46.0,
-      "step": 27696
-    },
-    {
-      "epoch": 2.1176290689450847,
-      "grad_norm": 0.0006825268501415849,
-      "learning_rate": 0.00019999778866300054,
-      "loss": 46.0,
-      "step": 27697
-    },
-    {
-      "epoch": 2.1177055259284745,
-      "grad_norm": 0.0008617136627435684,
-      "learning_rate": 0.00019999778850325996,
-      "loss": 46.0,
-      "step": 27698
-    },
-    {
-      "epoch": 2.1177819829118643,
-      "grad_norm": 0.0017924109706655145,
-      "learning_rate": 0.00019999778834351361,
-      "loss": 46.0,
-      "step": 27699
-    },
-    {
-      "epoch": 2.117858439895254,
-      "grad_norm": 0.00128801423124969,
-      "learning_rate": 0.0001999977881837615,
-      "loss": 46.0,
-      "step": 27700
-    },
-    {
-      "epoch": 2.117934896878644,
-      "grad_norm": 0.0009495370904915035,
-      "learning_rate": 0.0001999977880240036,
-      "loss": 46.0,
-      "step": 27701
-    },
-    {
-      "epoch": 2.1180113538620335,
-      "grad_norm": 0.003119442379102111,
-      "learning_rate": 0.00019999778786423993,
-      "loss": 46.0,
-      "step": 27702
-    },
-    {
-      "epoch": 2.1180878108454233,
-      "grad_norm": 0.0013680238043889403,
-      "learning_rate": 0.0001999977877044705,
-      "loss": 46.0,
-      "step": 27703
-    },
-    {
-      "epoch": 2.1181642678288126,
-      "grad_norm": 0.0006327729788608849,
-      "learning_rate": 0.0001999977875446953,
-      "loss": 46.0,
-      "step": 27704
-    },
-    {
-      "epoch": 2.1182407248122024,
-      "grad_norm": 0.0007384635391645133,
-      "learning_rate": 0.00019999778738491434,
-      "loss": 46.0,
-      "step": 27705
-    },
-    {
-      "epoch": 2.118317181795592,
-      "grad_norm": 0.0009168931283056736,
-      "learning_rate": 0.00019999778722512758,
-      "loss": 46.0,
-      "step": 27706
-    },
-    {
-      "epoch": 2.118393638778982,
-      "grad_norm": 0.001216028118506074,
-      "learning_rate": 0.00019999778706533507,
-      "loss": 46.0,
-      "step": 27707
-    },
-    {
-      "epoch": 2.1184700957623717,
-      "grad_norm": 0.001020344439893961,
-      "learning_rate": 0.00019999778690553681,
-      "loss": 46.0,
-      "step": 27708
-    },
-    {
-      "epoch": 2.1185465527457614,
-      "grad_norm": 0.0010994928888976574,
-      "learning_rate": 0.00019999778674573276,
-      "loss": 46.0,
-      "step": 27709
-    },
-    {
-      "epoch": 2.118623009729151,
-      "grad_norm": 0.0005078541580587626,
-      "learning_rate": 0.00019999778658592293,
-      "loss": 46.0,
-      "step": 27710
-    },
-    {
-      "epoch": 2.118699466712541,
-      "grad_norm": 0.0018513964023441076,
-      "learning_rate": 0.00019999778642610733,
-      "loss": 46.0,
-      "step": 27711
-    },
-    {
-      "epoch": 2.1187759236959307,
-      "grad_norm": 0.0009828828042373061,
-      "learning_rate": 0.00019999778626628598,
-      "loss": 46.0,
-      "step": 27712
-    },
-    {
-      "epoch": 2.1188523806793205,
-      "grad_norm": 0.0010659564286470413,
-      "learning_rate": 0.00019999778610645886,
-      "loss": 46.0,
-      "step": 27713
-    },
-    {
-      "epoch": 2.1189288376627102,
-      "grad_norm": 0.002268376061692834,
-      "learning_rate": 0.00019999778594662597,
-      "loss": 46.0,
-      "step": 27714
-    },
-    {
-      "epoch": 2.1190052946460995,
-      "grad_norm": 0.002673167735338211,
-      "learning_rate": 0.00019999778578678733,
-      "loss": 46.0,
-      "step": 27715
-    },
-    {
-      "epoch": 2.1190817516294893,
-      "grad_norm": 0.0028183767572045326,
-      "learning_rate": 0.00019999778562694286,
-      "loss": 46.0,
-      "step": 27716
-    },
-    {
-      "epoch": 2.119158208612879,
-      "grad_norm": 0.0013293303782120347,
-      "learning_rate": 0.0001999977854670927,
-      "loss": 46.0,
-      "step": 27717
-    },
-    {
-      "epoch": 2.119234665596269,
-      "grad_norm": 0.0037902083713561296,
-      "learning_rate": 0.0001999977853072367,
-      "loss": 46.0,
-      "step": 27718
-    },
-    {
-      "epoch": 2.1193111225796586,
-      "grad_norm": 0.001904433243907988,
-      "learning_rate": 0.00019999778514737497,
-      "loss": 46.0,
-      "step": 27719
-    },
-    {
-      "epoch": 2.1193875795630484,
-      "grad_norm": 0.0004010601551271975,
-      "learning_rate": 0.00019999778498750744,
-      "loss": 46.0,
-      "step": 27720
-    },
-    {
-      "epoch": 2.119464036546438,
-      "grad_norm": 0.0017100354889407754,
-      "learning_rate": 0.00019999778482763416,
-      "loss": 46.0,
-      "step": 27721
-    },
-    {
-      "epoch": 2.119540493529828,
-      "grad_norm": 0.0010594935156404972,
-      "learning_rate": 0.0001999977846677551,
-      "loss": 46.0,
-      "step": 27722
-    },
-    {
-      "epoch": 2.1196169505132176,
-      "grad_norm": 0.0007256284006871283,
-      "learning_rate": 0.0001999977845078703,
-      "loss": 46.0,
-      "step": 27723
-    },
-    {
-      "epoch": 2.1196934074966074,
-      "grad_norm": 0.002242007292807102,
-      "learning_rate": 0.0001999977843479797,
-      "loss": 46.0,
-      "step": 27724
-    },
-    {
-      "epoch": 2.119769864479997,
-      "grad_norm": 0.0021430812776088715,
-      "learning_rate": 0.00019999778418808338,
-      "loss": 46.0,
-      "step": 27725
-    },
-    {
-      "epoch": 2.1198463214633865,
-      "grad_norm": 0.0014775723684579134,
-      "learning_rate": 0.00019999778402818123,
-      "loss": 46.0,
-      "step": 27726
-    },
-    {
-      "epoch": 2.1199227784467762,
-      "grad_norm": 0.0010617939988151193,
-      "learning_rate": 0.00019999778386827334,
-      "loss": 46.0,
-      "step": 27727
-    },
-    {
-      "epoch": 2.119999235430166,
-      "grad_norm": 0.0015301000094041228,
-      "learning_rate": 0.00019999778370835967,
-      "loss": 46.0,
-      "step": 27728
-    },
-    {
-      "epoch": 2.1200756924135558,
-      "grad_norm": 0.0012853695079684258,
-      "learning_rate": 0.00019999778354844026,
-      "loss": 46.0,
-      "step": 27729
-    },
-    {
-      "epoch": 2.1201521493969455,
-      "grad_norm": 0.0005558666889555752,
-      "learning_rate": 0.00019999778338851504,
-      "loss": 46.0,
-      "step": 27730
-    },
-    {
-      "epoch": 2.1202286063803353,
-      "grad_norm": 0.0022962314542382956,
-      "learning_rate": 0.00019999778322858406,
-      "loss": 46.0,
-      "step": 27731
-    },
-    {
-      "epoch": 2.120305063363725,
-      "grad_norm": 0.0018421991262584925,
-      "learning_rate": 0.00019999778306864735,
-      "loss": 46.0,
-      "step": 27732
-    },
-    {
-      "epoch": 2.120381520347115,
-      "grad_norm": 0.0019613695330917835,
-      "learning_rate": 0.00019999778290870484,
-      "loss": 46.0,
-      "step": 27733
-    },
-    {
-      "epoch": 2.1204579773305046,
-      "grad_norm": 0.0026510327588766813,
-      "learning_rate": 0.00019999778274875653,
-      "loss": 46.0,
-      "step": 27734
-    },
-    {
-      "epoch": 2.1205344343138943,
-      "grad_norm": 0.020627731457352638,
-      "learning_rate": 0.00019999778258880248,
-      "loss": 46.0,
-      "step": 27735
-    },
-    {
-      "epoch": 2.120610891297284,
-      "grad_norm": 0.007026944309473038,
-      "learning_rate": 0.00019999778242884268,
-      "loss": 46.0,
-      "step": 27736
-    },
-    {
-      "epoch": 2.1206873482806734,
-      "grad_norm": 0.0020663077011704445,
-      "learning_rate": 0.00019999778226887708,
-      "loss": 46.0,
-      "step": 27737
-    },
-    {
-      "epoch": 2.120763805264063,
-      "grad_norm": 0.0009447532356716692,
-      "learning_rate": 0.0001999977821089057,
-      "loss": 46.0,
-      "step": 27738
-    },
-    {
-      "epoch": 2.120840262247453,
-      "grad_norm": 0.0006349161267280579,
-      "learning_rate": 0.00019999778194892858,
-      "loss": 46.0,
-      "step": 27739
-    },
-    {
-      "epoch": 2.1209167192308427,
-      "grad_norm": 0.0012721461243927479,
-      "learning_rate": 0.00019999778178894572,
-      "loss": 46.0,
-      "step": 27740
-    },
-    {
-      "epoch": 2.1209931762142324,
-      "grad_norm": 0.0034698189701884985,
-      "learning_rate": 0.00019999778162895705,
-      "loss": 46.0,
-      "step": 27741
-    },
-    {
-      "epoch": 2.121069633197622,
-      "grad_norm": 0.0009711988968774676,
-      "learning_rate": 0.0001999977814689626,
-      "loss": 46.0,
-      "step": 27742
-    },
-    {
-      "epoch": 2.121146090181012,
-      "grad_norm": 0.0008573653176426888,
-      "learning_rate": 0.0001999977813089624,
-      "loss": 46.0,
-      "step": 27743
-    },
-    {
-      "epoch": 2.1212225471644017,
-      "grad_norm": 0.0014904525596648455,
-      "learning_rate": 0.0001999977811489564,
-      "loss": 46.0,
-      "step": 27744
-    },
-    {
-      "epoch": 2.1212990041477915,
-      "grad_norm": 0.002192295854911208,
-      "learning_rate": 0.00019999778098894467,
-      "loss": 46.0,
-      "step": 27745
-    },
-    {
-      "epoch": 2.1213754611311813,
-      "grad_norm": 0.00025476867449469864,
-      "learning_rate": 0.00019999778082892714,
-      "loss": 46.0,
-      "step": 27746
-    },
-    {
-      "epoch": 2.1214519181145706,
-      "grad_norm": 0.0009082362521439791,
-      "learning_rate": 0.0001999977806689039,
-      "loss": 46.0,
-      "step": 27747
-    },
-    {
-      "epoch": 2.1215283750979603,
-      "grad_norm": 0.0007706684991717339,
-      "learning_rate": 0.00019999778050887483,
-      "loss": 46.0,
-      "step": 27748
-    },
-    {
-      "epoch": 2.12160483208135,
-      "grad_norm": 0.0011777285253629088,
-      "learning_rate": 0.00019999778034884003,
-      "loss": 46.0,
-      "step": 27749
-    },
-    {
-      "epoch": 2.12168128906474,
-      "grad_norm": 0.0007248474285006523,
-      "learning_rate": 0.00019999778018879943,
-      "loss": 46.0,
-      "step": 27750
-    },
-    {
-      "epoch": 2.1217577460481296,
-      "grad_norm": 0.0019676508381962776,
-      "learning_rate": 0.00019999778002875306,
-      "loss": 46.0,
-      "step": 27751
-    },
-    {
-      "epoch": 2.1218342030315194,
-      "grad_norm": 0.0009099239250645041,
-      "learning_rate": 0.00019999777986870094,
-      "loss": 46.0,
-      "step": 27752
-    },
-    {
-      "epoch": 2.121910660014909,
-      "grad_norm": 0.0005753159057348967,
-      "learning_rate": 0.00019999777970864304,
-      "loss": 46.0,
-      "step": 27753
-    },
-    {
-      "epoch": 2.121987116998299,
-      "grad_norm": 0.001207655994221568,
-      "learning_rate": 0.00019999777954857938,
-      "loss": 46.0,
-      "step": 27754
-    },
-    {
-      "epoch": 2.1220635739816887,
-      "grad_norm": 0.0003887069469783455,
-      "learning_rate": 0.00019999777938850994,
-      "loss": 46.0,
-      "step": 27755
-    },
-    {
-      "epoch": 2.1221400309650784,
-      "grad_norm": 0.007342228665947914,
-      "learning_rate": 0.00019999777922843475,
-      "loss": 46.0,
-      "step": 27756
-    },
-    {
-      "epoch": 2.122216487948468,
-      "grad_norm": 0.0007428711396642029,
-      "learning_rate": 0.00019999777906835376,
-      "loss": 46.0,
-      "step": 27757
-    },
-    {
-      "epoch": 2.122292944931858,
-      "grad_norm": 0.0020686613861471415,
-      "learning_rate": 0.000199997778908267,
-      "loss": 46.0,
-      "step": 27758
-    },
-    {
-      "epoch": 2.1223694019152473,
-      "grad_norm": 0.0007537698256783187,
-      "learning_rate": 0.0001999977787481745,
-      "loss": 46.0,
-      "step": 27759
-    },
-    {
-      "epoch": 2.122445858898637,
-      "grad_norm": 0.0020727405790239573,
-      "learning_rate": 0.00019999777858807622,
-      "loss": 46.0,
-      "step": 27760
-    },
-    {
-      "epoch": 2.1225223158820268,
-      "grad_norm": 0.0022943192161619663,
-      "learning_rate": 0.00019999777842797216,
-      "loss": 46.0,
-      "step": 27761
-    },
-    {
-      "epoch": 2.1225987728654165,
-      "grad_norm": 0.001227724482305348,
-      "learning_rate": 0.00019999777826786234,
-      "loss": 46.0,
-      "step": 27762
-    },
-    {
-      "epoch": 2.1226752298488063,
-      "grad_norm": 0.00046310090692713857,
-      "learning_rate": 0.00019999777810774676,
-      "loss": 46.0,
-      "step": 27763
-    },
-    {
-      "epoch": 2.122751686832196,
-      "grad_norm": 0.00034540367778390646,
-      "learning_rate": 0.00019999777794762536,
-      "loss": 46.0,
-      "step": 27764
-    },
-    {
-      "epoch": 2.122828143815586,
-      "grad_norm": 0.0015350074972957373,
-      "learning_rate": 0.00019999777778749827,
-      "loss": 46.0,
-      "step": 27765
-    },
-    {
-      "epoch": 2.1229046007989756,
-      "grad_norm": 0.00046227211714722216,
-      "learning_rate": 0.00019999777762736535,
-      "loss": 46.0,
-      "step": 27766
-    },
-    {
-      "epoch": 2.1229810577823653,
-      "grad_norm": 0.0007318872376345098,
-      "learning_rate": 0.00019999777746722668,
-      "loss": 46.0,
-      "step": 27767
-    },
-    {
-      "epoch": 2.123057514765755,
-      "grad_norm": 0.005794634576886892,
-      "learning_rate": 0.00019999777730708224,
-      "loss": 46.0,
-      "step": 27768
-    },
-    {
-      "epoch": 2.1231339717491444,
-      "grad_norm": 0.001248258980922401,
-      "learning_rate": 0.00019999777714693203,
-      "loss": 46.0,
-      "step": 27769
-    },
-    {
-      "epoch": 2.123210428732534,
-      "grad_norm": 0.0023213066160678864,
-      "learning_rate": 0.00019999777698677607,
-      "loss": 46.0,
-      "step": 27770
-    },
-    {
-      "epoch": 2.123286885715924,
-      "grad_norm": 0.0003884672769345343,
-      "learning_rate": 0.0001999977768266143,
-      "loss": 46.0,
-      "step": 27771
-    },
-    {
-      "epoch": 2.1233633426993137,
-      "grad_norm": 0.0006161300116218626,
-      "learning_rate": 0.0001999977766664468,
-      "loss": 46.0,
-      "step": 27772
-    },
-    {
-      "epoch": 2.1234397996827035,
-      "grad_norm": 0.0007281790021806955,
-      "learning_rate": 0.00019999777650627355,
-      "loss": 46.0,
-      "step": 27773
-    },
-    {
-      "epoch": 2.1235162566660932,
-      "grad_norm": 0.0017733452841639519,
-      "learning_rate": 0.00019999777634609447,
-      "loss": 46.0,
-      "step": 27774
-    },
-    {
-      "epoch": 2.123592713649483,
-      "grad_norm": 0.0006097950972616673,
-      "learning_rate": 0.00019999777618590964,
-      "loss": 46.0,
-      "step": 27775
-    },
-    {
-      "epoch": 2.1236691706328727,
-      "grad_norm": 0.0017350472044199705,
-      "learning_rate": 0.00019999777602571904,
-      "loss": 46.0,
-      "step": 27776
-    },
-    {
-      "epoch": 2.1237456276162625,
-      "grad_norm": 0.0023962666746228933,
-      "learning_rate": 0.0001999977758655227,
-      "loss": 46.0,
-      "step": 27777
-    },
-    {
-      "epoch": 2.1238220845996523,
-      "grad_norm": 0.001431509735994041,
-      "learning_rate": 0.00019999777570532058,
-      "loss": 46.0,
-      "step": 27778
-    },
-    {
-      "epoch": 2.123898541583042,
-      "grad_norm": 0.0013171338941901922,
-      "learning_rate": 0.00019999777554511266,
-      "loss": 46.0,
-      "step": 27779
-    },
-    {
-      "epoch": 2.1239749985664313,
-      "grad_norm": 0.0022890930995345116,
-      "learning_rate": 0.000199997775384899,
-      "loss": 46.0,
-      "step": 27780
-    },
-    {
-      "epoch": 2.124051455549821,
-      "grad_norm": 0.0018152977572754025,
-      "learning_rate": 0.00019999777522467955,
-      "loss": 46.0,
-      "step": 27781
-    },
-    {
-      "epoch": 2.124127912533211,
-      "grad_norm": 0.0008151123765856028,
-      "learning_rate": 0.00019999777506445434,
-      "loss": 46.0,
-      "step": 27782
-    },
-    {
-      "epoch": 2.1242043695166006,
-      "grad_norm": 0.0008217376307584345,
-      "learning_rate": 0.00019999777490422338,
-      "loss": 46.0,
-      "step": 27783
-    },
-    {
-      "epoch": 2.1242808264999904,
-      "grad_norm": 0.0020327442325651646,
-      "learning_rate": 0.00019999777474398665,
-      "loss": 46.0,
-      "step": 27784
-    },
-    {
-      "epoch": 2.12435728348338,
-      "grad_norm": 0.0020759673789143562,
-      "learning_rate": 0.0001999977745837441,
-      "loss": 46.0,
-      "step": 27785
-    },
-    {
-      "epoch": 2.12443374046677,
-      "grad_norm": 0.0007270768983289599,
-      "learning_rate": 0.00019999777442349583,
-      "loss": 46.0,
-      "step": 27786
-    },
-    {
-      "epoch": 2.1245101974501597,
-      "grad_norm": 0.0008436166681349277,
-      "learning_rate": 0.00019999777426324178,
-      "loss": 46.0,
-      "step": 27787
-    },
-    {
-      "epoch": 2.1245866544335494,
-      "grad_norm": 0.0007066066027618945,
-      "learning_rate": 0.00019999777410298195,
-      "loss": 46.0,
-      "step": 27788
-    },
-    {
-      "epoch": 2.124663111416939,
-      "grad_norm": 0.002340213628485799,
-      "learning_rate": 0.00019999777394271635,
-      "loss": 46.0,
-      "step": 27789
-    },
-    {
-      "epoch": 2.124739568400329,
-      "grad_norm": 0.0012920309090986848,
-      "learning_rate": 0.000199997773782445,
-      "loss": 46.0,
-      "step": 27790
-    },
-    {
-      "epoch": 2.1248160253837183,
-      "grad_norm": 0.002201956696808338,
-      "learning_rate": 0.00019999777362216786,
-      "loss": 46.0,
-      "step": 27791
-    },
-    {
-      "epoch": 2.124892482367108,
-      "grad_norm": 0.0011296358425170183,
-      "learning_rate": 0.00019999777346188497,
-      "loss": 46.0,
-      "step": 27792
-    },
-    {
-      "epoch": 2.124968939350498,
-      "grad_norm": 0.0017244359478354454,
-      "learning_rate": 0.0001999977733015963,
-      "loss": 46.0,
-      "step": 27793
-    },
-    {
-      "epoch": 2.1250453963338876,
-      "grad_norm": 0.001204316969960928,
-      "learning_rate": 0.00019999777314130186,
-      "loss": 46.0,
-      "step": 27794
-    },
-    {
-      "epoch": 2.1251218533172773,
-      "grad_norm": 0.0012584576616063714,
-      "learning_rate": 0.00019999777298100165,
-      "loss": 46.0,
-      "step": 27795
-    },
-    {
-      "epoch": 2.125198310300667,
-      "grad_norm": 0.003922926727682352,
-      "learning_rate": 0.00019999777282069564,
-      "loss": 46.0,
-      "step": 27796
-    },
-    {
-      "epoch": 2.125274767284057,
-      "grad_norm": 0.0008311591227538884,
-      "learning_rate": 0.0001999977726603839,
-      "loss": 46.0,
-      "step": 27797
-    },
-    {
-      "epoch": 2.1253512242674466,
-      "grad_norm": 0.0012705157278105617,
-      "learning_rate": 0.0001999977725000664,
-      "loss": 46.0,
-      "step": 27798
-    },
-    {
-      "epoch": 2.1254276812508364,
-      "grad_norm": 0.0012697366764768958,
-      "learning_rate": 0.0001999977723397431,
-      "loss": 46.0,
-      "step": 27799
-    },
-    {
-      "epoch": 2.125504138234226,
-      "grad_norm": 0.002598720835521817,
-      "learning_rate": 0.00019999777217941404,
-      "loss": 46.0,
-      "step": 27800
-    },
-    {
-      "epoch": 2.125580595217616,
-      "grad_norm": 0.0006474059773609042,
-      "learning_rate": 0.00019999777201907924,
-      "loss": 46.0,
-      "step": 27801
-    },
-    {
-      "epoch": 2.125657052201005,
-      "grad_norm": 0.002168721053749323,
-      "learning_rate": 0.00019999777185873864,
-      "loss": 46.0,
-      "step": 27802
-    },
-    {
-      "epoch": 2.125733509184395,
-      "grad_norm": 0.0018245772225782275,
-      "learning_rate": 0.00019999777169839227,
-      "loss": 46.0,
-      "step": 27803
-    },
-    {
-      "epoch": 2.1258099661677847,
-      "grad_norm": 0.0015790582401677966,
-      "learning_rate": 0.00019999777153804012,
-      "loss": 46.0,
-      "step": 27804
-    },
-    {
-      "epoch": 2.1258864231511745,
-      "grad_norm": 0.0008621436427347362,
-      "learning_rate": 0.00019999777137768223,
-      "loss": 46.0,
-      "step": 27805
-    },
-    {
-      "epoch": 2.1259628801345642,
-      "grad_norm": 0.0017095734365284443,
-      "learning_rate": 0.00019999777121731857,
-      "loss": 46.0,
-      "step": 27806
-    },
-    {
-      "epoch": 2.126039337117954,
-      "grad_norm": 0.0005265677464194596,
-      "learning_rate": 0.00019999777105694913,
-      "loss": 46.0,
-      "step": 27807
-    },
-    {
-      "epoch": 2.1261157941013438,
-      "grad_norm": 0.0013383154291659594,
-      "learning_rate": 0.0001999977708965739,
-      "loss": 46.0,
-      "step": 27808
-    },
-    {
-      "epoch": 2.1261922510847335,
-      "grad_norm": 0.0014789337292313576,
-      "learning_rate": 0.00019999777073619296,
-      "loss": 46.0,
-      "step": 27809
-    },
-    {
-      "epoch": 2.1262687080681233,
-      "grad_norm": 0.001086424570530653,
-      "learning_rate": 0.00019999777057580617,
-      "loss": 46.0,
-      "step": 27810
-    },
-    {
-      "epoch": 2.126345165051513,
-      "grad_norm": 0.0009949494851753116,
-      "learning_rate": 0.00019999777041541366,
-      "loss": 46.0,
-      "step": 27811
-    },
-    {
-      "epoch": 2.126421622034903,
-      "grad_norm": 0.0010510762222111225,
-      "learning_rate": 0.00019999777025501536,
-      "loss": 46.0,
-      "step": 27812
-    },
-    {
-      "epoch": 2.126498079018292,
-      "grad_norm": 0.0009002738515846431,
-      "learning_rate": 0.0001999977700946113,
-      "loss": 46.0,
-      "step": 27813
-    },
-    {
-      "epoch": 2.126574536001682,
-      "grad_norm": 0.0006245758268050849,
-      "learning_rate": 0.0001999977699342015,
-      "loss": 46.0,
-      "step": 27814
-    },
-    {
-      "epoch": 2.1266509929850717,
-      "grad_norm": 0.0006332048214972019,
-      "learning_rate": 0.0001999977697737859,
-      "loss": 46.0,
-      "step": 27815
-    },
-    {
-      "epoch": 2.1267274499684614,
-      "grad_norm": 0.0009491315577179193,
-      "learning_rate": 0.00019999776961336454,
-      "loss": 46.0,
-      "step": 27816
-    },
-    {
-      "epoch": 2.126803906951851,
-      "grad_norm": 0.0008021583780646324,
-      "learning_rate": 0.0001999977694529374,
-      "loss": 46.0,
-      "step": 27817
-    },
-    {
-      "epoch": 2.126880363935241,
-      "grad_norm": 0.008494246751070023,
-      "learning_rate": 0.0001999977692925045,
-      "loss": 46.0,
-      "step": 27818
-    },
-    {
-      "epoch": 2.1269568209186307,
-      "grad_norm": 0.0004216600500512868,
-      "learning_rate": 0.0001999977691320658,
-      "loss": 46.0,
-      "step": 27819
-    },
-    {
-      "epoch": 2.1270332779020205,
-      "grad_norm": 0.0015819082036614418,
-      "learning_rate": 0.0001999977689716214,
-      "loss": 46.0,
-      "step": 27820
-    },
-    {
-      "epoch": 2.12710973488541,
-      "grad_norm": 0.0010109863942489028,
-      "learning_rate": 0.00019999776881117116,
-      "loss": 46.0,
-      "step": 27821
-    },
-    {
-      "epoch": 2.1271861918688,
-      "grad_norm": 0.000561410328373313,
-      "learning_rate": 0.00019999776865071517,
-      "loss": 46.0,
-      "step": 27822
-    },
-    {
-      "epoch": 2.1272626488521897,
-      "grad_norm": 0.0018921643495559692,
-      "learning_rate": 0.0001999977684902534,
-      "loss": 46.0,
-      "step": 27823
-    },
-    {
-      "epoch": 2.127339105835579,
-      "grad_norm": 0.0008970113121904433,
-      "learning_rate": 0.0001999977683297859,
-      "loss": 46.0,
-      "step": 27824
-    },
-    {
-      "epoch": 2.127415562818969,
-      "grad_norm": 0.0017257422441616654,
-      "learning_rate": 0.00019999776816931263,
-      "loss": 46.0,
-      "step": 27825
-    },
-    {
-      "epoch": 2.1274920198023586,
-      "grad_norm": 0.002155852038413286,
-      "learning_rate": 0.00019999776800883357,
-      "loss": 46.0,
-      "step": 27826
-    },
-    {
-      "epoch": 2.1275684767857483,
-      "grad_norm": 0.0019253874197602272,
-      "learning_rate": 0.00019999776784834875,
-      "loss": 46.0,
-      "step": 27827
-    },
-    {
-      "epoch": 2.127644933769138,
-      "grad_norm": 0.001639417139813304,
-      "learning_rate": 0.00019999776768785815,
-      "loss": 46.0,
-      "step": 27828
-    },
-    {
-      "epoch": 2.127721390752528,
-      "grad_norm": 0.0009411929058842361,
-      "learning_rate": 0.00019999776752736178,
-      "loss": 46.0,
-      "step": 27829
-    },
-    {
-      "epoch": 2.1277978477359176,
-      "grad_norm": 0.0006266910932026803,
-      "learning_rate": 0.00019999776736685963,
-      "loss": 46.0,
-      "step": 27830
-    },
-    {
-      "epoch": 2.1278743047193074,
-      "grad_norm": 0.000506626209244132,
-      "learning_rate": 0.00019999776720635171,
-      "loss": 46.0,
-      "step": 27831
-    },
-    {
-      "epoch": 2.127950761702697,
-      "grad_norm": 0.0017454660264775157,
-      "learning_rate": 0.00019999776704583805,
-      "loss": 46.0,
-      "step": 27832
-    },
-    {
-      "epoch": 2.128027218686087,
-      "grad_norm": 0.0007652774220332503,
-      "learning_rate": 0.0001999977668853186,
-      "loss": 46.0,
-      "step": 27833
-    },
-    {
-      "epoch": 2.1281036756694767,
-      "grad_norm": 0.00030271473224274814,
-      "learning_rate": 0.0001999977667247934,
-      "loss": 46.0,
-      "step": 27834
-    },
-    {
-      "epoch": 2.128180132652866,
-      "grad_norm": 0.0005601273733191192,
-      "learning_rate": 0.00019999776656426241,
-      "loss": 46.0,
-      "step": 27835
-    },
-    {
-      "epoch": 2.1282565896362557,
-      "grad_norm": 0.001075144624337554,
-      "learning_rate": 0.00019999776640372568,
-      "loss": 46.0,
-      "step": 27836
-    },
-    {
-      "epoch": 2.1283330466196455,
-      "grad_norm": 0.0016779261641204357,
-      "learning_rate": 0.00019999776624318315,
-      "loss": 46.0,
-      "step": 27837
-    },
-    {
-      "epoch": 2.1284095036030353,
-      "grad_norm": 0.0013826611684635282,
-      "learning_rate": 0.00019999776608263487,
-      "loss": 46.0,
-      "step": 27838
-    },
-    {
-      "epoch": 2.128485960586425,
-      "grad_norm": 0.002510193968191743,
-      "learning_rate": 0.00019999776592208082,
-      "loss": 46.0,
-      "step": 27839
-    },
-    {
-      "epoch": 2.128562417569815,
-      "grad_norm": 0.0008974681841209531,
-      "learning_rate": 0.000199997765761521,
-      "loss": 46.0,
-      "step": 27840
-    },
-    {
-      "epoch": 2.1286388745532046,
-      "grad_norm": 0.0010345522314310074,
-      "learning_rate": 0.00019999776560095537,
-      "loss": 46.0,
-      "step": 27841
-    },
-    {
-      "epoch": 2.1287153315365943,
-      "grad_norm": 0.0006212533335201442,
-      "learning_rate": 0.000199997765440384,
-      "loss": 46.0,
-      "step": 27842
-    },
-    {
-      "epoch": 2.128791788519984,
-      "grad_norm": 0.0011552287032827735,
-      "learning_rate": 0.00019999776527980688,
-      "loss": 46.0,
-      "step": 27843
-    },
-    {
-      "epoch": 2.128868245503374,
-      "grad_norm": 0.0013556129997596145,
-      "learning_rate": 0.00019999776511922396,
-      "loss": 46.0,
-      "step": 27844
-    },
-    {
-      "epoch": 2.1289447024867636,
-      "grad_norm": 0.0008895316859707236,
-      "learning_rate": 0.00019999776495863527,
-      "loss": 46.0,
-      "step": 27845
-    },
-    {
-      "epoch": 2.129021159470153,
-      "grad_norm": 0.003528506960719824,
-      "learning_rate": 0.00019999776479804083,
-      "loss": 46.0,
-      "step": 27846
-    },
-    {
-      "epoch": 2.1290976164535427,
-      "grad_norm": 0.0007977734203450382,
-      "learning_rate": 0.00019999776463744062,
-      "loss": 46.0,
-      "step": 27847
-    },
-    {
-      "epoch": 2.1291740734369324,
-      "grad_norm": 0.0010544314282014966,
-      "learning_rate": 0.00019999776447683464,
-      "loss": 46.0,
-      "step": 27848
-    },
-    {
-      "epoch": 2.129250530420322,
-      "grad_norm": 0.0011289088288322091,
-      "learning_rate": 0.00019999776431622288,
-      "loss": 46.0,
-      "step": 27849
-    },
-    {
-      "epoch": 2.129326987403712,
-      "grad_norm": 0.001412665005773306,
-      "learning_rate": 0.00019999776415560537,
-      "loss": 46.0,
-      "step": 27850
-    },
-    {
-      "epoch": 2.1294034443871017,
-      "grad_norm": 0.0014999600825831294,
-      "learning_rate": 0.00019999776399498207,
-      "loss": 46.0,
-      "step": 27851
-    },
-    {
-      "epoch": 2.1294799013704915,
-      "grad_norm": 0.0008316526073031127,
-      "learning_rate": 0.000199997763834353,
-      "loss": 46.0,
-      "step": 27852
-    },
-    {
-      "epoch": 2.1295563583538812,
-      "grad_norm": 0.0014714428689330816,
-      "learning_rate": 0.0001999977636737182,
-      "loss": 46.0,
-      "step": 27853
-    },
-    {
-      "epoch": 2.129632815337271,
-      "grad_norm": 0.0010015941224992275,
-      "learning_rate": 0.0001999977635130776,
-      "loss": 46.0,
-      "step": 27854
-    },
-    {
-      "epoch": 2.1297092723206608,
-      "grad_norm": 0.0007377064903266728,
-      "learning_rate": 0.00019999776335243122,
-      "loss": 46.0,
-      "step": 27855
-    },
-    {
-      "epoch": 2.12978572930405,
-      "grad_norm": 0.00147099862806499,
-      "learning_rate": 0.00019999776319177908,
-      "loss": 46.0,
-      "step": 27856
-    },
-    {
-      "epoch": 2.12986218628744,
-      "grad_norm": 0.0007974232430569828,
-      "learning_rate": 0.00019999776303112116,
-      "loss": 46.0,
-      "step": 27857
-    },
-    {
-      "epoch": 2.1299386432708296,
-      "grad_norm": 0.001324589946307242,
-      "learning_rate": 0.0001999977628704575,
-      "loss": 46.0,
-      "step": 27858
-    },
-    {
-      "epoch": 2.1300151002542194,
-      "grad_norm": 0.001063237781636417,
-      "learning_rate": 0.00019999776270978806,
-      "loss": 46.0,
-      "step": 27859
-    },
-    {
-      "epoch": 2.130091557237609,
-      "grad_norm": 0.0014845584519207478,
-      "learning_rate": 0.00019999776254911282,
-      "loss": 46.0,
-      "step": 27860
-    },
-    {
-      "epoch": 2.130168014220999,
-      "grad_norm": 0.0006285892450250685,
-      "learning_rate": 0.00019999776238843183,
-      "loss": 46.0,
-      "step": 27861
-    },
-    {
-      "epoch": 2.1302444712043886,
-      "grad_norm": 0.0018885077442973852,
-      "learning_rate": 0.00019999776222774508,
-      "loss": 46.0,
-      "step": 27862
-    },
-    {
-      "epoch": 2.1303209281877784,
-      "grad_norm": 0.001775530632585287,
-      "learning_rate": 0.00019999776206705257,
-      "loss": 46.0,
-      "step": 27863
-    },
-    {
-      "epoch": 2.130397385171168,
-      "grad_norm": 0.003006625920534134,
-      "learning_rate": 0.00019999776190635427,
-      "loss": 46.0,
-      "step": 27864
-    },
-    {
-      "epoch": 2.130473842154558,
-      "grad_norm": 0.0011765756644308567,
-      "learning_rate": 0.00019999776174565022,
-      "loss": 46.0,
-      "step": 27865
-    },
-    {
-      "epoch": 2.1305502991379477,
-      "grad_norm": 0.0007958024507388473,
-      "learning_rate": 0.0001999977615849404,
-      "loss": 46.0,
-      "step": 27866
-    },
-    {
-      "epoch": 2.1306267561213375,
-      "grad_norm": 0.0006438548443838954,
-      "learning_rate": 0.00019999776142422477,
-      "loss": 46.0,
-      "step": 27867
-    },
-    {
-      "epoch": 2.1307032131047268,
-      "grad_norm": 0.0017352061113342643,
-      "learning_rate": 0.00019999776126350342,
-      "loss": 46.0,
-      "step": 27868
-    },
-    {
-      "epoch": 2.1307796700881165,
-      "grad_norm": 0.0014211511006578803,
-      "learning_rate": 0.00019999776110277625,
-      "loss": 46.0,
-      "step": 27869
-    },
-    {
-      "epoch": 2.1308561270715063,
-      "grad_norm": 0.0027469047345221043,
-      "learning_rate": 0.00019999776094204336,
-      "loss": 46.0,
-      "step": 27870
-    },
-    {
-      "epoch": 2.130932584054896,
-      "grad_norm": 0.0008085627341642976,
-      "learning_rate": 0.0001999977607813047,
-      "loss": 46.0,
-      "step": 27871
-    },
-    {
-      "epoch": 2.131009041038286,
-      "grad_norm": 0.0008705726359039545,
-      "learning_rate": 0.00019999776062056023,
-      "loss": 46.0,
-      "step": 27872
-    },
-    {
-      "epoch": 2.1310854980216756,
-      "grad_norm": 0.0011628699721768498,
-      "learning_rate": 0.00019999776045981002,
-      "loss": 46.0,
-      "step": 27873
-    },
-    {
-      "epoch": 2.1311619550050653,
-      "grad_norm": 0.0017635178519412875,
-      "learning_rate": 0.00019999776029905404,
-      "loss": 46.0,
-      "step": 27874
-    },
-    {
-      "epoch": 2.131238411988455,
-      "grad_norm": 0.0007235931698232889,
-      "learning_rate": 0.00019999776013829228,
-      "loss": 46.0,
-      "step": 27875
-    },
-    {
-      "epoch": 2.131314868971845,
-      "grad_norm": 0.0007826391374692321,
-      "learning_rate": 0.00019999775997752478,
-      "loss": 46.0,
-      "step": 27876
-    },
-    {
-      "epoch": 2.1313913259552346,
-      "grad_norm": 0.0008739493787288666,
-      "learning_rate": 0.00019999775981675147,
-      "loss": 46.0,
-      "step": 27877
-    },
-    {
-      "epoch": 2.131467782938624,
-      "grad_norm": 0.0006460209260694683,
-      "learning_rate": 0.0001999977596559724,
-      "loss": 46.0,
-      "step": 27878
-    },
-    {
-      "epoch": 2.1315442399220137,
-      "grad_norm": 0.0008441006066277623,
-      "learning_rate": 0.00019999775949518757,
-      "loss": 46.0,
-      "step": 27879
-    },
-    {
-      "epoch": 2.1316206969054035,
-      "grad_norm": 0.002067247871309519,
-      "learning_rate": 0.00019999775933439697,
-      "loss": 46.0,
-      "step": 27880
-    },
-    {
-      "epoch": 2.131697153888793,
-      "grad_norm": 0.0008495355723425746,
-      "learning_rate": 0.0001999977591736006,
-      "loss": 46.0,
-      "step": 27881
-    },
-    {
-      "epoch": 2.131773610872183,
-      "grad_norm": 0.0007296452531591058,
-      "learning_rate": 0.00019999775901279846,
-      "loss": 46.0,
-      "step": 27882
-    },
-    {
-      "epoch": 2.1318500678555727,
-      "grad_norm": 0.0008365765097551048,
-      "learning_rate": 0.00019999775885199054,
-      "loss": 46.0,
-      "step": 27883
-    },
-    {
-      "epoch": 2.1319265248389625,
-      "grad_norm": 0.00044905408867634833,
-      "learning_rate": 0.00019999775869117685,
-      "loss": 46.0,
-      "step": 27884
-    },
-    {
-      "epoch": 2.1320029818223523,
-      "grad_norm": 0.0013494088780134916,
-      "learning_rate": 0.0001999977585303574,
-      "loss": 46.0,
-      "step": 27885
-    },
-    {
-      "epoch": 2.132079438805742,
-      "grad_norm": 0.0014080164255574346,
-      "learning_rate": 0.00019999775836953217,
-      "loss": 46.0,
-      "step": 27886
-    },
-    {
-      "epoch": 2.132155895789132,
-      "grad_norm": 0.0013624605489894748,
-      "learning_rate": 0.00019999775820870122,
-      "loss": 46.0,
-      "step": 27887
-    },
-    {
-      "epoch": 2.1322323527725215,
-      "grad_norm": 0.0010071718133985996,
-      "learning_rate": 0.00019999775804786446,
-      "loss": 46.0,
-      "step": 27888
-    },
-    {
-      "epoch": 2.1323088097559113,
-      "grad_norm": 0.001538605079986155,
-      "learning_rate": 0.00019999775788702193,
-      "loss": 46.0,
-      "step": 27889
-    },
-    {
-      "epoch": 2.1323852667393006,
-      "grad_norm": 0.0038290575612336397,
-      "learning_rate": 0.00019999775772617365,
-      "loss": 46.0,
-      "step": 27890
-    },
-    {
-      "epoch": 2.1324617237226904,
-      "grad_norm": 0.00046507318620570004,
-      "learning_rate": 0.00019999775756531957,
-      "loss": 46.0,
-      "step": 27891
-    },
-    {
-      "epoch": 2.13253818070608,
-      "grad_norm": 0.0015731814783066511,
-      "learning_rate": 0.00019999775740445975,
-      "loss": 46.0,
-      "step": 27892
-    },
-    {
-      "epoch": 2.13261463768947,
-      "grad_norm": 0.0016571552259847522,
-      "learning_rate": 0.00019999775724359413,
-      "loss": 46.0,
-      "step": 27893
-    },
-    {
-      "epoch": 2.1326910946728597,
-      "grad_norm": 0.001173077616840601,
-      "learning_rate": 0.00019999775708272278,
-      "loss": 46.0,
-      "step": 27894
-    },
-    {
-      "epoch": 2.1327675516562494,
-      "grad_norm": 0.0004726675688289106,
-      "learning_rate": 0.00019999775692184564,
-      "loss": 46.0,
-      "step": 27895
-    },
-    {
-      "epoch": 2.132844008639639,
-      "grad_norm": 0.0013790634693577886,
-      "learning_rate": 0.0001999977567609627,
-      "loss": 46.0,
-      "step": 27896
-    },
-    {
-      "epoch": 2.132920465623029,
-      "grad_norm": 0.0020232738461345434,
-      "learning_rate": 0.00019999775660007403,
-      "loss": 46.0,
-      "step": 27897
-    },
-    {
-      "epoch": 2.1329969226064187,
-      "grad_norm": 0.0009763463749550283,
-      "learning_rate": 0.00019999775643917957,
-      "loss": 46.0,
-      "step": 27898
-    },
-    {
-      "epoch": 2.1330733795898085,
-      "grad_norm": 0.0006836850079707801,
-      "learning_rate": 0.00019999775627827936,
-      "loss": 46.0,
-      "step": 27899
-    },
-    {
-      "epoch": 2.133149836573198,
-      "grad_norm": 0.0038131983019411564,
-      "learning_rate": 0.00019999775611737337,
-      "loss": 46.0,
-      "step": 27900
-    },
-    {
-      "epoch": 2.1332262935565875,
-      "grad_norm": 0.0021967804059386253,
-      "learning_rate": 0.0001999977559564616,
-      "loss": 46.0,
-      "step": 27901
-    },
-    {
-      "epoch": 2.1333027505399773,
-      "grad_norm": 0.004178035072982311,
-      "learning_rate": 0.00019999775579554409,
-      "loss": 46.0,
-      "step": 27902
-    },
-    {
-      "epoch": 2.133379207523367,
-      "grad_norm": 0.001492820680141449,
-      "learning_rate": 0.00019999775563462078,
-      "loss": 46.0,
-      "step": 27903
-    },
-    {
-      "epoch": 2.133455664506757,
-      "grad_norm": 0.0011182373855262995,
-      "learning_rate": 0.00019999775547369173,
-      "loss": 46.0,
-      "step": 27904
-    },
-    {
-      "epoch": 2.1335321214901466,
-      "grad_norm": 0.00037381317815743387,
-      "learning_rate": 0.00019999775531275688,
-      "loss": 46.0,
-      "step": 27905
-    },
-    {
-      "epoch": 2.1336085784735364,
-      "grad_norm": 0.0016465885564684868,
-      "learning_rate": 0.00019999775515181626,
-      "loss": 46.0,
-      "step": 27906
-    },
-    {
-      "epoch": 2.133685035456926,
-      "grad_norm": 0.0010482021607458591,
-      "learning_rate": 0.0001999977549908699,
-      "loss": 46.0,
-      "step": 27907
-    },
-    {
-      "epoch": 2.133761492440316,
-      "grad_norm": 0.001215518917888403,
-      "learning_rate": 0.00019999775482991777,
-      "loss": 46.0,
-      "step": 27908
-    },
-    {
-      "epoch": 2.1338379494237056,
-      "grad_norm": 0.010715700685977936,
-      "learning_rate": 0.00019999775466895985,
-      "loss": 46.0,
-      "step": 27909
-    },
-    {
-      "epoch": 2.1339144064070954,
-      "grad_norm": 0.0006658357451669872,
-      "learning_rate": 0.00019999775450799616,
-      "loss": 46.0,
-      "step": 27910
-    },
-    {
-      "epoch": 2.133990863390485,
-      "grad_norm": 0.0007772714598104358,
-      "learning_rate": 0.00019999775434702673,
-      "loss": 46.0,
-      "step": 27911
-    },
-    {
-      "epoch": 2.1340673203738745,
-      "grad_norm": 0.0009552838164381683,
-      "learning_rate": 0.00019999775418605152,
-      "loss": 46.0,
-      "step": 27912
-    },
-    {
-      "epoch": 2.1341437773572642,
-      "grad_norm": 0.0040089404210448265,
-      "learning_rate": 0.0001999977540250705,
-      "loss": 46.0,
-      "step": 27913
-    },
-    {
-      "epoch": 2.134220234340654,
-      "grad_norm": 0.0014650429366156459,
-      "learning_rate": 0.00019999775386408378,
-      "loss": 46.0,
-      "step": 27914
-    },
-    {
-      "epoch": 2.1342966913240438,
-      "grad_norm": 0.0007803139160387218,
-      "learning_rate": 0.00019999775370309122,
-      "loss": 46.0,
-      "step": 27915
-    },
-    {
-      "epoch": 2.1343731483074335,
-      "grad_norm": 0.0016963358502835035,
-      "learning_rate": 0.00019999775354209294,
-      "loss": 46.0,
-      "step": 27916
-    },
-    {
-      "epoch": 2.1344496052908233,
-      "grad_norm": 0.0009868739871308208,
-      "learning_rate": 0.00019999775338108887,
-      "loss": 46.0,
-      "step": 27917
-    },
-    {
-      "epoch": 2.134526062274213,
-      "grad_norm": 0.0005647193756885827,
-      "learning_rate": 0.00019999775322007902,
-      "loss": 46.0,
-      "step": 27918
-    },
-    {
-      "epoch": 2.134602519257603,
-      "grad_norm": 0.0013062083162367344,
-      "learning_rate": 0.00019999775305906342,
-      "loss": 46.0,
-      "step": 27919
-    },
-    {
-      "epoch": 2.1346789762409926,
-      "grad_norm": 0.0013398105511441827,
-      "learning_rate": 0.00019999775289804202,
-      "loss": 46.0,
-      "step": 27920
-    },
-    {
-      "epoch": 2.1347554332243823,
-      "grad_norm": 0.007981443777680397,
-      "learning_rate": 0.00019999775273701488,
-      "loss": 46.0,
-      "step": 27921
-    },
-    {
-      "epoch": 2.1348318902077716,
-      "grad_norm": 0.00045654072891920805,
-      "learning_rate": 0.000199997752575982,
-      "loss": 46.0,
-      "step": 27922
-    },
-    {
-      "epoch": 2.1349083471911614,
-      "grad_norm": 0.0006403446313925087,
-      "learning_rate": 0.00019999775241494333,
-      "loss": 46.0,
-      "step": 27923
-    },
-    {
-      "epoch": 2.134984804174551,
-      "grad_norm": 0.0004985718405805528,
-      "learning_rate": 0.00019999775225389886,
-      "loss": 46.0,
-      "step": 27924
-    },
-    {
-      "epoch": 2.135061261157941,
-      "grad_norm": 0.0005125298630446196,
-      "learning_rate": 0.00019999775209284863,
-      "loss": 46.0,
-      "step": 27925
-    },
-    {
-      "epoch": 2.1351377181413307,
-      "grad_norm": 0.000692663190420717,
-      "learning_rate": 0.00019999775193179264,
-      "loss": 46.0,
-      "step": 27926
-    },
-    {
-      "epoch": 2.1352141751247204,
-      "grad_norm": 0.0012574001448228955,
-      "learning_rate": 0.0001999977517707309,
-      "loss": 46.0,
-      "step": 27927
-    },
-    {
-      "epoch": 2.13529063210811,
-      "grad_norm": 0.0015778085216879845,
-      "learning_rate": 0.00019999775160966333,
-      "loss": 46.0,
-      "step": 27928
-    },
-    {
-      "epoch": 2.1353670890915,
-      "grad_norm": 0.0018670731224119663,
-      "learning_rate": 0.00019999775144859006,
-      "loss": 46.0,
-      "step": 27929
-    },
-    {
-      "epoch": 2.1354435460748897,
-      "grad_norm": 0.001294909161515534,
-      "learning_rate": 0.000199997751287511,
-      "loss": 46.0,
-      "step": 27930
-    },
-    {
-      "epoch": 2.1355200030582795,
-      "grad_norm": 0.008364435285329819,
-      "learning_rate": 0.00019999775112642616,
-      "loss": 46.0,
-      "step": 27931
-    },
-    {
-      "epoch": 2.1355964600416693,
-      "grad_norm": 0.0007175628561526537,
-      "learning_rate": 0.00019999775096533556,
-      "loss": 46.0,
-      "step": 27932
-    },
-    {
-      "epoch": 2.1356729170250586,
-      "grad_norm": 0.0007800923776812851,
-      "learning_rate": 0.00019999775080423916,
-      "loss": 46.0,
-      "step": 27933
-    },
-    {
-      "epoch": 2.1357493740084483,
-      "grad_norm": 0.0006643543601967394,
-      "learning_rate": 0.00019999775064313702,
-      "loss": 46.0,
-      "step": 27934
-    },
-    {
-      "epoch": 2.135825830991838,
-      "grad_norm": 0.0008329321863129735,
-      "learning_rate": 0.00019999775048202908,
-      "loss": 46.0,
-      "step": 27935
-    },
-    {
-      "epoch": 2.135902287975228,
-      "grad_norm": 0.0005805922555737197,
-      "learning_rate": 0.00019999775032091544,
-      "loss": 46.0,
-      "step": 27936
-    },
-    {
-      "epoch": 2.1359787449586176,
-      "grad_norm": 0.0008534258813597262,
-      "learning_rate": 0.00019999775015979595,
-      "loss": 46.0,
-      "step": 27937
-    },
-    {
-      "epoch": 2.1360552019420074,
-      "grad_norm": 0.004057834390550852,
-      "learning_rate": 0.00019999774999867077,
-      "loss": 46.0,
-      "step": 27938
-    },
-    {
-      "epoch": 2.136131658925397,
-      "grad_norm": 0.0008537324029020965,
-      "learning_rate": 0.00019999774983753973,
-      "loss": 46.0,
-      "step": 27939
-    },
-    {
-      "epoch": 2.136208115908787,
-      "grad_norm": 0.0014889901503920555,
-      "learning_rate": 0.000199997749676403,
-      "loss": 46.0,
-      "step": 27940
-    },
-    {
-      "epoch": 2.1362845728921767,
-      "grad_norm": 0.001632365514524281,
-      "learning_rate": 0.00019999774951526047,
-      "loss": 46.0,
-      "step": 27941
-    },
-    {
-      "epoch": 2.1363610298755664,
-      "grad_norm": 0.0012648382689803839,
-      "learning_rate": 0.00019999774935411217,
-      "loss": 46.0,
-      "step": 27942
-    },
-    {
-      "epoch": 2.136437486858956,
-      "grad_norm": 0.000894739234354347,
-      "learning_rate": 0.0001999977491929581,
-      "loss": 46.0,
-      "step": 27943
-    },
-    {
-      "epoch": 2.1365139438423455,
-      "grad_norm": 0.006497282534837723,
-      "learning_rate": 0.00019999774903179827,
-      "loss": 46.0,
-      "step": 27944
-    },
-    {
-      "epoch": 2.1365904008257353,
-      "grad_norm": 0.001762912841513753,
-      "learning_rate": 0.00019999774887063265,
-      "loss": 46.0,
-      "step": 27945
-    },
-    {
-      "epoch": 2.136666857809125,
-      "grad_norm": 0.0006510132807306945,
-      "learning_rate": 0.00019999774870946125,
-      "loss": 46.0,
-      "step": 27946
-    },
-    {
-      "epoch": 2.136743314792515,
-      "grad_norm": 0.0003557926102075726,
-      "learning_rate": 0.0001999977485482841,
-      "loss": 46.0,
-      "step": 27947
-    },
-    {
-      "epoch": 2.1368197717759045,
-      "grad_norm": 0.000626552093308419,
-      "learning_rate": 0.00019999774838710122,
-      "loss": 46.0,
-      "step": 27948
-    },
-    {
-      "epoch": 2.1368962287592943,
-      "grad_norm": 0.001279408810660243,
-      "learning_rate": 0.0001999977482259125,
-      "loss": 46.0,
-      "step": 27949
-    },
-    {
-      "epoch": 2.136972685742684,
-      "grad_norm": 0.0012648004340007901,
-      "learning_rate": 0.00019999774806471807,
-      "loss": 46.0,
-      "step": 27950
-    },
-    {
-      "epoch": 2.137049142726074,
-      "grad_norm": 0.0004089291614945978,
-      "learning_rate": 0.00019999774790351783,
-      "loss": 46.0,
-      "step": 27951
-    },
-    {
-      "epoch": 2.1371255997094636,
-      "grad_norm": 0.000834234815556556,
-      "learning_rate": 0.00019999774774231182,
-      "loss": 46.0,
-      "step": 27952
-    },
-    {
-      "epoch": 2.1372020566928533,
-      "grad_norm": 0.003803718602284789,
-      "learning_rate": 0.00019999774758110007,
-      "loss": 46.0,
-      "step": 27953
-    },
-    {
-      "epoch": 2.137278513676243,
-      "grad_norm": 0.0020154034718871117,
-      "learning_rate": 0.00019999774741988257,
-      "loss": 46.0,
-      "step": 27954
-    },
-    {
-      "epoch": 2.1373549706596324,
-      "grad_norm": 0.006825027987360954,
-      "learning_rate": 0.00019999774725865926,
-      "loss": 46.0,
-      "step": 27955
-    },
-    {
-      "epoch": 2.137431427643022,
-      "grad_norm": 0.0014343821676447988,
-      "learning_rate": 0.0001999977470974302,
-      "loss": 46.0,
-      "step": 27956
-    },
-    {
-      "epoch": 2.137507884626412,
-      "grad_norm": 0.004281510133296251,
-      "learning_rate": 0.00019999774693619534,
-      "loss": 46.0,
-      "step": 27957
-    },
-    {
-      "epoch": 2.1375843416098017,
-      "grad_norm": 0.001521748723462224,
-      "learning_rate": 0.00019999774677495474,
-      "loss": 46.0,
-      "step": 27958
-    },
-    {
-      "epoch": 2.1376607985931915,
-      "grad_norm": 0.0005761711508966982,
-      "learning_rate": 0.00019999774661370835,
-      "loss": 46.0,
-      "step": 27959
-    },
-    {
-      "epoch": 2.1377372555765812,
-      "grad_norm": 0.0015655792085453868,
-      "learning_rate": 0.0001999977464524562,
-      "loss": 46.0,
-      "step": 27960
-    },
-    {
-      "epoch": 2.137813712559971,
-      "grad_norm": 0.016118254512548447,
-      "learning_rate": 0.0001999977462911983,
-      "loss": 46.0,
-      "step": 27961
-    },
-    {
-      "epoch": 2.1378901695433608,
-      "grad_norm": 0.0006259109941311181,
-      "learning_rate": 0.0001999977461299346,
-      "loss": 46.0,
-      "step": 27962
-    },
-    {
-      "epoch": 2.1379666265267505,
-      "grad_norm": 0.0008122969884425402,
-      "learning_rate": 0.00019999774596866514,
-      "loss": 46.0,
-      "step": 27963
-    },
-    {
-      "epoch": 2.1380430835101403,
-      "grad_norm": 0.0012238132767379284,
-      "learning_rate": 0.00019999774580738993,
-      "loss": 46.0,
-      "step": 27964
-    },
-    {
-      "epoch": 2.13811954049353,
-      "grad_norm": 0.0007204969879239798,
-      "learning_rate": 0.00019999774564610895,
-      "loss": 46.0,
-      "step": 27965
-    },
-    {
-      "epoch": 2.1381959974769194,
-      "grad_norm": 0.000909200927708298,
-      "learning_rate": 0.00019999774548482216,
-      "loss": 46.0,
-      "step": 27966
-    },
-    {
-      "epoch": 2.138272454460309,
-      "grad_norm": 0.003151579527184367,
-      "learning_rate": 0.00019999774532352964,
-      "loss": 46.0,
-      "step": 27967
-    },
-    {
-      "epoch": 2.138348911443699,
-      "grad_norm": 0.002466138917952776,
-      "learning_rate": 0.00019999774516223133,
-      "loss": 46.0,
-      "step": 27968
-    },
-    {
-      "epoch": 2.1384253684270886,
-      "grad_norm": 0.0014224511105567217,
-      "learning_rate": 0.00019999774500092726,
-      "loss": 46.0,
-      "step": 27969
-    },
-    {
-      "epoch": 2.1385018254104784,
-      "grad_norm": 0.0007494601304642856,
-      "learning_rate": 0.0001999977448396174,
-      "loss": 46.0,
-      "step": 27970
-    },
-    {
-      "epoch": 2.138578282393868,
-      "grad_norm": 0.0014879771042615175,
-      "learning_rate": 0.00019999774467830181,
-      "loss": 46.0,
-      "step": 27971
-    },
-    {
-      "epoch": 2.138654739377258,
-      "grad_norm": 0.00064283941173926,
-      "learning_rate": 0.00019999774451698042,
-      "loss": 46.0,
-      "step": 27972
-    },
-    {
-      "epoch": 2.1387311963606477,
-      "grad_norm": 0.033043231815099716,
-      "learning_rate": 0.00019999774435565328,
-      "loss": 46.0,
-      "step": 27973
-    },
-    {
-      "epoch": 2.1388076533440374,
-      "grad_norm": 0.0009749619639478624,
-      "learning_rate": 0.00019999774419432036,
-      "loss": 46.0,
-      "step": 27974
-    },
-    {
-      "epoch": 2.138884110327427,
-      "grad_norm": 0.0006747089792042971,
-      "learning_rate": 0.00019999774403298167,
-      "loss": 46.0,
-      "step": 27975
-    },
-    {
-      "epoch": 2.138960567310817,
-      "grad_norm": 0.010180780664086342,
-      "learning_rate": 0.0001999977438716372,
-      "loss": 46.0,
-      "step": 27976
-    },
-    {
-      "epoch": 2.1390370242942063,
-      "grad_norm": 0.006115399766713381,
-      "learning_rate": 0.000199997743710287,
-      "loss": 46.0,
-      "step": 27977
-    },
-    {
-      "epoch": 2.139113481277596,
-      "grad_norm": 0.0011515038786455989,
-      "learning_rate": 0.000199997743548931,
-      "loss": 46.0,
-      "step": 27978
-    },
-    {
-      "epoch": 2.139189938260986,
-      "grad_norm": 0.0011794361053034663,
-      "learning_rate": 0.00019999774338756924,
-      "loss": 46.0,
-      "step": 27979
-    },
-    {
-      "epoch": 2.1392663952443756,
-      "grad_norm": 0.007787879090756178,
-      "learning_rate": 0.00019999774322620168,
-      "loss": 46.0,
-      "step": 27980
-    },
-    {
-      "epoch": 2.1393428522277653,
-      "grad_norm": 0.0015810028417035937,
-      "learning_rate": 0.00019999774306482838,
-      "loss": 46.0,
-      "step": 27981
-    },
-    {
-      "epoch": 2.139419309211155,
-      "grad_norm": 0.0005167325725778937,
-      "learning_rate": 0.00019999774290344933,
-      "loss": 46.0,
-      "step": 27982
-    },
-    {
-      "epoch": 2.139495766194545,
-      "grad_norm": 0.0005183730972930789,
-      "learning_rate": 0.00019999774274206446,
-      "loss": 46.0,
-      "step": 27983
-    },
-    {
-      "epoch": 2.1395722231779346,
-      "grad_norm": 0.0016227837186306715,
-      "learning_rate": 0.00019999774258067386,
-      "loss": 46.0,
-      "step": 27984
-    },
-    {
-      "epoch": 2.1396486801613244,
-      "grad_norm": 0.0008283095667138696,
-      "learning_rate": 0.00019999774241927747,
-      "loss": 46.0,
-      "step": 27985
-    },
-    {
-      "epoch": 2.139725137144714,
-      "grad_norm": 0.0029107846785336733,
-      "learning_rate": 0.00019999774225787532,
-      "loss": 46.0,
-      "step": 27986
-    },
-    {
-      "epoch": 2.1398015941281034,
-      "grad_norm": 0.0008983362349681556,
-      "learning_rate": 0.0001999977420964674,
-      "loss": 46.0,
-      "step": 27987
-    },
-    {
-      "epoch": 2.139878051111493,
-      "grad_norm": 0.002306839916855097,
-      "learning_rate": 0.00019999774193505372,
-      "loss": 46.0,
-      "step": 27988
-    },
-    {
-      "epoch": 2.139954508094883,
-      "grad_norm": 0.0008934852667152882,
-      "learning_rate": 0.00019999774177363426,
-      "loss": 46.0,
-      "step": 27989
-    },
-    {
-      "epoch": 2.1400309650782727,
-      "grad_norm": 0.0009797102538868785,
-      "learning_rate": 0.00019999774161220902,
-      "loss": 46.0,
-      "step": 27990
-    },
-    {
-      "epoch": 2.1401074220616625,
-      "grad_norm": 0.006482865661382675,
-      "learning_rate": 0.00019999774145077804,
-      "loss": 46.0,
-      "step": 27991
-    },
-    {
-      "epoch": 2.1401838790450523,
-      "grad_norm": 0.003567473264411092,
-      "learning_rate": 0.0001999977412893413,
-      "loss": 46.0,
-      "step": 27992
-    },
-    {
-      "epoch": 2.140260336028442,
-      "grad_norm": 0.0028472740668803453,
-      "learning_rate": 0.00019999774112789873,
-      "loss": 46.0,
-      "step": 27993
-    },
-    {
-      "epoch": 2.1403367930118318,
-      "grad_norm": 0.0009431094513274729,
-      "learning_rate": 0.00019999774096645043,
-      "loss": 46.0,
-      "step": 27994
-    },
-    {
-      "epoch": 2.1404132499952215,
-      "grad_norm": 0.003307847073301673,
-      "learning_rate": 0.00019999774080499636,
-      "loss": 46.0,
-      "step": 27995
-    },
-    {
-      "epoch": 2.1404897069786113,
-      "grad_norm": 0.00043861649464815855,
-      "learning_rate": 0.0001999977406435365,
-      "loss": 46.0,
-      "step": 27996
-    },
-    {
-      "epoch": 2.140566163962001,
-      "grad_norm": 0.0011678007431328297,
-      "learning_rate": 0.00019999774048207089,
-      "loss": 46.0,
-      "step": 27997
-    },
-    {
-      "epoch": 2.140642620945391,
-      "grad_norm": 0.0013952864101156592,
-      "learning_rate": 0.00019999774032059952,
-      "loss": 46.0,
-      "step": 27998
-    },
-    {
-      "epoch": 2.14071907792878,
-      "grad_norm": 0.00679080281406641,
-      "learning_rate": 0.00019999774015912235,
-      "loss": 46.0,
-      "step": 27999
-    },
-    {
-      "epoch": 2.14079553491217,
-      "grad_norm": 0.001105370232835412,
-      "learning_rate": 0.00019999773999763943,
-      "loss": 46.0,
-      "step": 28000
-    },
-    {
-      "epoch": 2.1408719918955597,
-      "grad_norm": 0.0013363149482756853,
-      "learning_rate": 0.00019999773983615075,
-      "loss": 46.0,
-      "step": 28001
-    },
-    {
-      "epoch": 2.1409484488789494,
-      "grad_norm": 0.0008694211137481034,
-      "learning_rate": 0.00019999773967465629,
-      "loss": 46.0,
-      "step": 28002
-    },
-    {
-      "epoch": 2.141024905862339,
-      "grad_norm": 0.0006769180181436241,
-      "learning_rate": 0.00019999773951315608,
-      "loss": 46.0,
-      "step": 28003
-    },
-    {
-      "epoch": 2.141101362845729,
-      "grad_norm": 0.0027024210430681705,
-      "learning_rate": 0.00019999773935165007,
-      "loss": 46.0,
-      "step": 28004
-    },
-    {
-      "epoch": 2.1411778198291187,
-      "grad_norm": 0.0007253543008118868,
-      "learning_rate": 0.0001999977391901383,
-      "loss": 46.0,
-      "step": 28005
-    },
-    {
-      "epoch": 2.1412542768125085,
-      "grad_norm": 0.0038207811303436756,
-      "learning_rate": 0.00019999773902862076,
-      "loss": 46.0,
-      "step": 28006
-    },
-    {
-      "epoch": 2.1413307337958982,
-      "grad_norm": 0.0024014287628233433,
-      "learning_rate": 0.00019999773886709746,
-      "loss": 46.0,
-      "step": 28007
-    },
-    {
-      "epoch": 2.141407190779288,
-      "grad_norm": 0.0011907346779480577,
-      "learning_rate": 0.00019999773870556838,
-      "loss": 46.0,
-      "step": 28008
-    },
-    {
-      "epoch": 2.1414836477626773,
-      "grad_norm": 0.0012580914190039039,
-      "learning_rate": 0.0001999977385440335,
-      "loss": 46.0,
-      "step": 28009
-    },
-    {
-      "epoch": 2.141560104746067,
-      "grad_norm": 0.0008508822065778077,
-      "learning_rate": 0.00019999773838249291,
-      "loss": 46.0,
-      "step": 28010
-    },
-    {
-      "epoch": 2.141636561729457,
-      "grad_norm": 0.0010860462207347155,
-      "learning_rate": 0.00019999773822094655,
-      "loss": 46.0,
-      "step": 28011
-    },
-    {
-      "epoch": 2.1417130187128466,
-      "grad_norm": 0.0021899437997490168,
-      "learning_rate": 0.00019999773805939438,
-      "loss": 46.0,
-      "step": 28012
-    },
-    {
-      "epoch": 2.1417894756962363,
-      "grad_norm": 0.0044187684543430805,
-      "learning_rate": 0.00019999773789783646,
-      "loss": 46.0,
-      "step": 28013
-    },
-    {
-      "epoch": 2.141865932679626,
-      "grad_norm": 0.0007306295447051525,
-      "learning_rate": 0.00019999773773627278,
-      "loss": 46.0,
-      "step": 28014
-    },
-    {
-      "epoch": 2.141942389663016,
-      "grad_norm": 0.0011351504363119602,
-      "learning_rate": 0.00019999773757470332,
-      "loss": 46.0,
-      "step": 28015
-    },
-    {
-      "epoch": 2.1420188466464056,
-      "grad_norm": 0.0031075968872755766,
-      "learning_rate": 0.00019999773741312808,
-      "loss": 46.0,
-      "step": 28016
-    },
-    {
-      "epoch": 2.1420953036297954,
-      "grad_norm": 0.0008862606482580304,
-      "learning_rate": 0.0001999977372515471,
-      "loss": 46.0,
-      "step": 28017
-    },
-    {
-      "epoch": 2.142171760613185,
-      "grad_norm": 0.008570067584514618,
-      "learning_rate": 0.0001999977370899603,
-      "loss": 46.0,
-      "step": 28018
-    },
-    {
-      "epoch": 2.142248217596575,
-      "grad_norm": 0.001295009278692305,
-      "learning_rate": 0.0001999977369283678,
-      "loss": 46.0,
-      "step": 28019
-    },
-    {
-      "epoch": 2.1423246745799647,
-      "grad_norm": 0.0007180798565968871,
-      "learning_rate": 0.00019999773676676946,
-      "loss": 46.0,
-      "step": 28020
-    },
-    {
-      "epoch": 2.142401131563354,
-      "grad_norm": 0.001086206641048193,
-      "learning_rate": 0.0001999977366051654,
-      "loss": 46.0,
-      "step": 28021
-    },
-    {
-      "epoch": 2.1424775885467437,
-      "grad_norm": 0.0005852627218700945,
-      "learning_rate": 0.00019999773644355554,
-      "loss": 46.0,
-      "step": 28022
-    },
-    {
-      "epoch": 2.1425540455301335,
-      "grad_norm": 0.0008816480403766036,
-      "learning_rate": 0.00019999773628193995,
-      "loss": 46.0,
-      "step": 28023
-    },
-    {
-      "epoch": 2.1426305025135233,
-      "grad_norm": 0.0017591941868886352,
-      "learning_rate": 0.00019999773612031855,
-      "loss": 46.0,
-      "step": 28024
-    },
-    {
-      "epoch": 2.142706959496913,
-      "grad_norm": 0.0010571813909336925,
-      "learning_rate": 0.0001999977359586914,
-      "loss": 46.0,
-      "step": 28025
-    },
-    {
-      "epoch": 2.142783416480303,
-      "grad_norm": 0.0016324438620358706,
-      "learning_rate": 0.00019999773579705847,
-      "loss": 46.0,
-      "step": 28026
-    },
-    {
-      "epoch": 2.1428598734636926,
-      "grad_norm": 0.0010612218175083399,
-      "learning_rate": 0.00019999773563541976,
-      "loss": 46.0,
-      "step": 28027
-    },
-    {
-      "epoch": 2.1429363304470823,
-      "grad_norm": 0.0007551426533609629,
-      "learning_rate": 0.00019999773547377532,
-      "loss": 46.0,
-      "step": 28028
-    },
-    {
-      "epoch": 2.143012787430472,
-      "grad_norm": 0.0011655011912807822,
-      "learning_rate": 0.0001999977353121251,
-      "loss": 46.0,
-      "step": 28029
-    },
-    {
-      "epoch": 2.143089244413862,
-      "grad_norm": 0.000542212335858494,
-      "learning_rate": 0.00019999773515046908,
-      "loss": 46.0,
-      "step": 28030
-    },
-    {
-      "epoch": 2.143165701397251,
-      "grad_norm": 0.0006167124374769628,
-      "learning_rate": 0.00019999773498880733,
-      "loss": 46.0,
-      "step": 28031
-    },
-    {
-      "epoch": 2.143242158380641,
-      "grad_norm": 0.0026519224047660828,
-      "learning_rate": 0.00019999773482713977,
-      "loss": 46.0,
-      "step": 28032
-    },
-    {
-      "epoch": 2.1433186153640307,
-      "grad_norm": 0.001772786257788539,
-      "learning_rate": 0.00019999773466546645,
-      "loss": 46.0,
-      "step": 28033
-    },
-    {
-      "epoch": 2.1433950723474204,
-      "grad_norm": 0.0009836714016273618,
-      "learning_rate": 0.00019999773450378737,
-      "loss": 46.0,
-      "step": 28034
-    },
-    {
-      "epoch": 2.14347152933081,
-      "grad_norm": 0.005014047026634216,
-      "learning_rate": 0.00019999773434210255,
-      "loss": 46.0,
-      "step": 28035
-    },
-    {
-      "epoch": 2.1435479863142,
-      "grad_norm": 0.007744866423308849,
-      "learning_rate": 0.00019999773418041193,
-      "loss": 46.0,
-      "step": 28036
-    },
-    {
-      "epoch": 2.1436244432975897,
-      "grad_norm": 0.0008506677113473415,
-      "learning_rate": 0.00019999773401871554,
-      "loss": 46.0,
-      "step": 28037
-    },
-    {
-      "epoch": 2.1437009002809795,
-      "grad_norm": 0.0007242302526719868,
-      "learning_rate": 0.00019999773385701337,
-      "loss": 46.0,
-      "step": 28038
-    },
-    {
-      "epoch": 2.1437773572643692,
-      "grad_norm": 0.0009783850982785225,
-      "learning_rate": 0.00019999773369530545,
-      "loss": 46.0,
-      "step": 28039
-    },
-    {
-      "epoch": 2.143853814247759,
-      "grad_norm": 0.0013943061931058764,
-      "learning_rate": 0.00019999773353359177,
-      "loss": 46.0,
-      "step": 28040
-    },
-    {
-      "epoch": 2.1439302712311488,
-      "grad_norm": 0.0013013249263167381,
-      "learning_rate": 0.00019999773337187228,
-      "loss": 46.0,
-      "step": 28041
-    },
-    {
-      "epoch": 2.1440067282145385,
-      "grad_norm": 0.00034890859387815,
-      "learning_rate": 0.00019999773321014707,
-      "loss": 46.0,
-      "step": 28042
-    },
-    {
-      "epoch": 2.144083185197928,
-      "grad_norm": 0.0010609349701553583,
-      "learning_rate": 0.00019999773304841607,
-      "loss": 46.0,
-      "step": 28043
-    },
-    {
-      "epoch": 2.1441596421813176,
-      "grad_norm": 0.0006238010246306658,
-      "learning_rate": 0.00019999773288667929,
-      "loss": 46.0,
-      "step": 28044
-    },
-    {
-      "epoch": 2.1442360991647074,
-      "grad_norm": 0.0022068372927606106,
-      "learning_rate": 0.00019999773272493676,
-      "loss": 46.0,
-      "step": 28045
-    },
-    {
-      "epoch": 2.144312556148097,
-      "grad_norm": 0.000992662156932056,
-      "learning_rate": 0.00019999773256318846,
-      "loss": 46.0,
-      "step": 28046
-    },
-    {
-      "epoch": 2.144389013131487,
-      "grad_norm": 0.003069251775741577,
-      "learning_rate": 0.00019999773240143438,
-      "loss": 46.0,
-      "step": 28047
-    },
-    {
-      "epoch": 2.1444654701148766,
-      "grad_norm": 0.0011823850218206644,
-      "learning_rate": 0.00019999773223967454,
-      "loss": 46.0,
-      "step": 28048
-    },
-    {
-      "epoch": 2.1445419270982664,
-      "grad_norm": 0.0007566019776277244,
-      "learning_rate": 0.00019999773207790892,
-      "loss": 46.0,
-      "step": 28049
-    },
-    {
-      "epoch": 2.144618384081656,
-      "grad_norm": 0.0004884752561338246,
-      "learning_rate": 0.00019999773191613752,
-      "loss": 46.0,
-      "step": 28050
-    },
-    {
-      "epoch": 2.144694841065046,
-      "grad_norm": 0.0006829904159530997,
-      "learning_rate": 0.00019999773175436033,
-      "loss": 46.0,
-      "step": 28051
-    },
-    {
-      "epoch": 2.1447712980484357,
-      "grad_norm": 0.001857191789895296,
-      "learning_rate": 0.00019999773159257744,
-      "loss": 46.0,
-      "step": 28052
-    },
-    {
-      "epoch": 2.144847755031825,
-      "grad_norm": 0.000579994055442512,
-      "learning_rate": 0.00019999773143078873,
-      "loss": 46.0,
-      "step": 28053
-    },
-    {
-      "epoch": 2.1449242120152148,
-      "grad_norm": 0.0011155155953019857,
-      "learning_rate": 0.00019999773126899427,
-      "loss": 46.0,
-      "step": 28054
-    },
-    {
-      "epoch": 2.1450006689986045,
-      "grad_norm": 0.0009850916685536504,
-      "learning_rate": 0.00019999773110719403,
-      "loss": 46.0,
-      "step": 28055
-    },
-    {
-      "epoch": 2.1450771259819943,
-      "grad_norm": 0.0007776428828947246,
-      "learning_rate": 0.00019999773094538803,
-      "loss": 46.0,
-      "step": 28056
-    },
-    {
-      "epoch": 2.145153582965384,
-      "grad_norm": 0.0023823254741728306,
-      "learning_rate": 0.00019999773078357627,
-      "loss": 46.0,
-      "step": 28057
-    },
-    {
-      "epoch": 2.145230039948774,
-      "grad_norm": 0.0007415247382596135,
-      "learning_rate": 0.00019999773062175872,
-      "loss": 46.0,
-      "step": 28058
-    },
-    {
-      "epoch": 2.1453064969321636,
-      "grad_norm": 0.0011347433319315314,
-      "learning_rate": 0.00019999773045993542,
-      "loss": 46.0,
-      "step": 28059
-    },
-    {
-      "epoch": 2.1453829539155533,
-      "grad_norm": 0.00033677570172585547,
-      "learning_rate": 0.00019999773029810635,
-      "loss": 46.0,
-      "step": 28060
-    },
-    {
-      "epoch": 2.145459410898943,
-      "grad_norm": 0.0017952433554455638,
-      "learning_rate": 0.00019999773013627147,
-      "loss": 46.0,
-      "step": 28061
-    },
-    {
-      "epoch": 2.145535867882333,
-      "grad_norm": 0.0005006867577321827,
-      "learning_rate": 0.00019999772997443085,
-      "loss": 46.0,
-      "step": 28062
-    },
-    {
-      "epoch": 2.1456123248657226,
-      "grad_norm": 0.003177864011377096,
-      "learning_rate": 0.0001999977298125845,
-      "loss": 46.0,
-      "step": 28063
-    },
-    {
-      "epoch": 2.145688781849112,
-      "grad_norm": 0.0009311573230661452,
-      "learning_rate": 0.00019999772965073232,
-      "loss": 46.0,
-      "step": 28064
-    },
-    {
-      "epoch": 2.1457652388325017,
-      "grad_norm": 0.001280890661291778,
-      "learning_rate": 0.00019999772948887438,
-      "loss": 46.0,
-      "step": 28065
-    },
-    {
-      "epoch": 2.1458416958158915,
-      "grad_norm": 0.0011088397586718202,
-      "learning_rate": 0.0001999977293270107,
-      "loss": 46.0,
-      "step": 28066
-    },
-    {
-      "epoch": 2.145918152799281,
-      "grad_norm": 0.0009297026554122567,
-      "learning_rate": 0.00019999772916514123,
-      "loss": 46.0,
-      "step": 28067
-    },
-    {
-      "epoch": 2.145994609782671,
-      "grad_norm": 0.0015929682413116097,
-      "learning_rate": 0.000199997729003266,
-      "loss": 46.0,
-      "step": 28068
-    },
-    {
-      "epoch": 2.1460710667660607,
-      "grad_norm": 0.000654333911370486,
-      "learning_rate": 0.000199997728841385,
-      "loss": 46.0,
-      "step": 28069
-    },
-    {
-      "epoch": 2.1461475237494505,
-      "grad_norm": 0.0017585826572030783,
-      "learning_rate": 0.00019999772867949821,
-      "loss": 46.0,
-      "step": 28070
-    },
-    {
-      "epoch": 2.1462239807328403,
-      "grad_norm": 0.0013783980393782258,
-      "learning_rate": 0.0001999977285176057,
-      "loss": 46.0,
-      "step": 28071
-    },
-    {
-      "epoch": 2.14630043771623,
-      "grad_norm": 0.0007270924979820848,
-      "learning_rate": 0.00019999772835570736,
-      "loss": 46.0,
-      "step": 28072
-    },
-    {
-      "epoch": 2.14637689469962,
-      "grad_norm": 0.0006456988048739731,
-      "learning_rate": 0.0001999977281938033,
-      "loss": 46.0,
-      "step": 28073
-    },
-    {
-      "epoch": 2.1464533516830095,
-      "grad_norm": 0.001488837180659175,
-      "learning_rate": 0.00019999772803189344,
-      "loss": 46.0,
-      "step": 28074
-    },
-    {
-      "epoch": 2.146529808666399,
-      "grad_norm": 0.0008940851548686624,
-      "learning_rate": 0.00019999772786997782,
-      "loss": 46.0,
-      "step": 28075
-    },
-    {
-      "epoch": 2.1466062656497886,
-      "grad_norm": 0.0016282738652080297,
-      "learning_rate": 0.00019999772770805643,
-      "loss": 46.0,
-      "step": 28076
-    },
-    {
-      "epoch": 2.1466827226331784,
-      "grad_norm": 0.0012227408587932587,
-      "learning_rate": 0.00019999772754612923,
-      "loss": 46.0,
-      "step": 28077
-    },
-    {
-      "epoch": 2.146759179616568,
-      "grad_norm": 0.001382127869874239,
-      "learning_rate": 0.00019999772738419632,
-      "loss": 46.0,
-      "step": 28078
-    },
-    {
-      "epoch": 2.146835636599958,
-      "grad_norm": 0.0005174451507627964,
-      "learning_rate": 0.00019999772722225764,
-      "loss": 46.0,
-      "step": 28079
-    },
-    {
-      "epoch": 2.1469120935833477,
-      "grad_norm": 0.0015492273960262537,
-      "learning_rate": 0.00019999772706031318,
-      "loss": 46.0,
-      "step": 28080
-    },
-    {
-      "epoch": 2.1469885505667374,
-      "grad_norm": 0.000808349228464067,
-      "learning_rate": 0.00019999772689836294,
-      "loss": 46.0,
-      "step": 28081
-    },
-    {
-      "epoch": 2.147065007550127,
-      "grad_norm": 0.0008739153272472322,
-      "learning_rate": 0.00019999772673640694,
-      "loss": 46.0,
-      "step": 28082
-    },
-    {
-      "epoch": 2.147141464533517,
-      "grad_norm": 0.004506739787757397,
-      "learning_rate": 0.00019999772657444516,
-      "loss": 46.0,
-      "step": 28083
-    },
-    {
-      "epoch": 2.1472179215169067,
-      "grad_norm": 0.0005172499804757535,
-      "learning_rate": 0.0001999977264124776,
-      "loss": 46.0,
-      "step": 28084
-    },
-    {
-      "epoch": 2.1472943785002965,
-      "grad_norm": 0.0008176789269782603,
-      "learning_rate": 0.0001999977262505043,
-      "loss": 46.0,
-      "step": 28085
-    },
-    {
-      "epoch": 2.147370835483686,
-      "grad_norm": 0.0012802501441910863,
-      "learning_rate": 0.0001999977260885252,
-      "loss": 46.0,
-      "step": 28086
-    },
-    {
-      "epoch": 2.1474472924670756,
-      "grad_norm": 0.0011771807912737131,
-      "learning_rate": 0.00019999772592654036,
-      "loss": 46.0,
-      "step": 28087
-    },
-    {
-      "epoch": 2.1475237494504653,
-      "grad_norm": 0.0016017784364521503,
-      "learning_rate": 0.00019999772576454974,
-      "loss": 46.0,
-      "step": 28088
-    },
-    {
-      "epoch": 2.147600206433855,
-      "grad_norm": 0.0008139270939864218,
-      "learning_rate": 0.00019999772560255335,
-      "loss": 46.0,
-      "step": 28089
-    },
-    {
-      "epoch": 2.147676663417245,
-      "grad_norm": 0.001272616907954216,
-      "learning_rate": 0.00019999772544055118,
-      "loss": 46.0,
-      "step": 28090
-    },
-    {
-      "epoch": 2.1477531204006346,
-      "grad_norm": 0.002162357559427619,
-      "learning_rate": 0.00019999772527854327,
-      "loss": 46.0,
-      "step": 28091
-    },
-    {
-      "epoch": 2.1478295773840244,
-      "grad_norm": 0.0013778364518657327,
-      "learning_rate": 0.00019999772511652955,
-      "loss": 46.0,
-      "step": 28092
-    },
-    {
-      "epoch": 2.147906034367414,
-      "grad_norm": 0.0012272101594135165,
-      "learning_rate": 0.0001999977249545101,
-      "loss": 46.0,
-      "step": 28093
-    },
-    {
-      "epoch": 2.147982491350804,
-      "grad_norm": 0.003676793770864606,
-      "learning_rate": 0.00019999772479248486,
-      "loss": 46.0,
-      "step": 28094
-    },
-    {
-      "epoch": 2.1480589483341936,
-      "grad_norm": 0.0011014972114935517,
-      "learning_rate": 0.00019999772463045386,
-      "loss": 46.0,
-      "step": 28095
-    },
-    {
-      "epoch": 2.1481354053175834,
-      "grad_norm": 0.0036270723212510347,
-      "learning_rate": 0.00019999772446841708,
-      "loss": 46.0,
-      "step": 28096
-    },
-    {
-      "epoch": 2.1482118623009727,
-      "grad_norm": 0.0010349606163799763,
-      "learning_rate": 0.00019999772430637452,
-      "loss": 46.0,
-      "step": 28097
-    },
-    {
-      "epoch": 2.1482883192843625,
-      "grad_norm": 0.000926801934838295,
-      "learning_rate": 0.00019999772414432623,
-      "loss": 46.0,
-      "step": 28098
-    },
-    {
-      "epoch": 2.1483647762677522,
-      "grad_norm": 0.0006331445183604956,
-      "learning_rate": 0.00019999772398227213,
-      "loss": 46.0,
-      "step": 28099
-    },
-    {
-      "epoch": 2.148441233251142,
-      "grad_norm": 0.005006878636777401,
-      "learning_rate": 0.00019999772382021228,
-      "loss": 46.0,
-      "step": 28100
-    },
-    {
-      "epoch": 2.1485176902345318,
-      "grad_norm": 0.001148741808719933,
-      "learning_rate": 0.00019999772365814663,
-      "loss": 46.0,
-      "step": 28101
-    },
-    {
-      "epoch": 2.1485941472179215,
-      "grad_norm": 0.00098808400798589,
-      "learning_rate": 0.00019999772349607527,
-      "loss": 46.0,
-      "step": 28102
-    },
-    {
-      "epoch": 2.1486706042013113,
-      "grad_norm": 0.0012053870595991611,
-      "learning_rate": 0.0001999977233339981,
-      "loss": 46.0,
-      "step": 28103
-    },
-    {
-      "epoch": 2.148747061184701,
-      "grad_norm": 0.0019479095935821533,
-      "learning_rate": 0.0001999977231719152,
-      "loss": 46.0,
-      "step": 28104
-    },
-    {
-      "epoch": 2.148823518168091,
-      "grad_norm": 0.0005970261408947408,
-      "learning_rate": 0.00019999772300982648,
-      "loss": 46.0,
-      "step": 28105
-    },
-    {
-      "epoch": 2.1488999751514806,
-      "grad_norm": 0.0034951060079038143,
-      "learning_rate": 0.00019999772284773202,
-      "loss": 46.0,
-      "step": 28106
-    },
-    {
-      "epoch": 2.1489764321348703,
-      "grad_norm": 0.0025081115309149027,
-      "learning_rate": 0.00019999772268563176,
-      "loss": 46.0,
-      "step": 28107
-    },
-    {
-      "epoch": 2.1490528891182596,
-      "grad_norm": 0.0004526465490926057,
-      "learning_rate": 0.00019999772252352575,
-      "loss": 46.0,
-      "step": 28108
-    },
-    {
-      "epoch": 2.1491293461016494,
-      "grad_norm": 0.0007097624475136399,
-      "learning_rate": 0.00019999772236141397,
-      "loss": 46.0,
-      "step": 28109
-    },
-    {
-      "epoch": 2.149205803085039,
-      "grad_norm": 0.0007064482779242098,
-      "learning_rate": 0.00019999772219929645,
-      "loss": 46.0,
-      "step": 28110
-    },
-    {
-      "epoch": 2.149282260068429,
-      "grad_norm": 0.0005932310596108437,
-      "learning_rate": 0.0001999977220371731,
-      "loss": 46.0,
-      "step": 28111
-    },
-    {
-      "epoch": 2.1493587170518187,
-      "grad_norm": 0.0026616195682436228,
-      "learning_rate": 0.00019999772187504405,
-      "loss": 46.0,
-      "step": 28112
-    },
-    {
-      "epoch": 2.1494351740352085,
-      "grad_norm": 0.0016878944588825107,
-      "learning_rate": 0.00019999772171290918,
-      "loss": 46.0,
-      "step": 28113
-    },
-    {
-      "epoch": 2.149511631018598,
-      "grad_norm": 0.0006121466867625713,
-      "learning_rate": 0.00019999772155076856,
-      "loss": 46.0,
-      "step": 28114
-    },
-    {
-      "epoch": 2.149588088001988,
-      "grad_norm": 0.0008225361234508455,
-      "learning_rate": 0.00019999772138862217,
-      "loss": 46.0,
-      "step": 28115
-    },
-    {
-      "epoch": 2.1496645449853777,
-      "grad_norm": 0.0010969842551276088,
-      "learning_rate": 0.00019999772122647,
-      "loss": 46.0,
-      "step": 28116
-    },
-    {
-      "epoch": 2.1497410019687675,
-      "grad_norm": 0.0011599557474255562,
-      "learning_rate": 0.00019999772106431206,
-      "loss": 46.0,
-      "step": 28117
-    },
-    {
-      "epoch": 2.149817458952157,
-      "grad_norm": 0.0009778441162779927,
-      "learning_rate": 0.00019999772090214838,
-      "loss": 46.0,
-      "step": 28118
-    },
-    {
-      "epoch": 2.1498939159355466,
-      "grad_norm": 0.002477919915691018,
-      "learning_rate": 0.0001999977207399789,
-      "loss": 46.0,
-      "step": 28119
-    },
-    {
-      "epoch": 2.1499703729189363,
-      "grad_norm": 0.000579033512622118,
-      "learning_rate": 0.00019999772057780366,
-      "loss": 46.0,
-      "step": 28120
-    },
-    {
-      "epoch": 2.150046829902326,
-      "grad_norm": 0.0004702067526523024,
-      "learning_rate": 0.00019999772041562265,
-      "loss": 46.0,
-      "step": 28121
-    },
-    {
-      "epoch": 2.150123286885716,
-      "grad_norm": 0.0004753461398649961,
-      "learning_rate": 0.00019999772025343585,
-      "loss": 46.0,
-      "step": 28122
-    },
-    {
-      "epoch": 2.1501997438691056,
-      "grad_norm": 0.000491965445689857,
-      "learning_rate": 0.00019999772009124332,
-      "loss": 46.0,
-      "step": 28123
-    },
-    {
-      "epoch": 2.1502762008524954,
-      "grad_norm": 0.0008067890885286033,
-      "learning_rate": 0.000199997719929045,
-      "loss": 46.0,
-      "step": 28124
-    },
-    {
-      "epoch": 2.150352657835885,
-      "grad_norm": 0.0021548105869442225,
-      "learning_rate": 0.00019999771976684093,
-      "loss": 46.0,
-      "step": 28125
-    },
-    {
-      "epoch": 2.150429114819275,
-      "grad_norm": 0.0007861893973313272,
-      "learning_rate": 0.00019999771960463108,
-      "loss": 46.0,
-      "step": 28126
-    },
-    {
-      "epoch": 2.1505055718026647,
-      "grad_norm": 0.0009069948573596776,
-      "learning_rate": 0.00019999771944241544,
-      "loss": 46.0,
-      "step": 28127
-    },
-    {
-      "epoch": 2.1505820287860544,
-      "grad_norm": 0.001138515304774046,
-      "learning_rate": 0.00019999771928019404,
-      "loss": 46.0,
-      "step": 28128
-    },
-    {
-      "epoch": 2.150658485769444,
-      "grad_norm": 0.0009136605658568442,
-      "learning_rate": 0.00019999771911796688,
-      "loss": 46.0,
-      "step": 28129
-    },
-    {
-      "epoch": 2.1507349427528335,
-      "grad_norm": 0.003166587557643652,
-      "learning_rate": 0.00019999771895573397,
-      "loss": 46.0,
-      "step": 28130
-    },
-    {
-      "epoch": 2.1508113997362233,
-      "grad_norm": 0.0015549215022474527,
-      "learning_rate": 0.00019999771879349525,
-      "loss": 46.0,
-      "step": 28131
-    },
-    {
-      "epoch": 2.150887856719613,
-      "grad_norm": 0.0008785352692939341,
-      "learning_rate": 0.00019999771863125077,
-      "loss": 46.0,
-      "step": 28132
-    },
-    {
-      "epoch": 2.150964313703003,
-      "grad_norm": 0.0017593599623069167,
-      "learning_rate": 0.00019999771846900054,
-      "loss": 46.0,
-      "step": 28133
-    },
-    {
-      "epoch": 2.1510407706863925,
-      "grad_norm": 0.0013651741901412606,
-      "learning_rate": 0.00019999771830674453,
-      "loss": 46.0,
-      "step": 28134
-    },
-    {
-      "epoch": 2.1511172276697823,
-      "grad_norm": 0.0009699792135506868,
-      "learning_rate": 0.00019999771814448273,
-      "loss": 46.0,
-      "step": 28135
-    },
-    {
-      "epoch": 2.151193684653172,
-      "grad_norm": 0.000791808997746557,
-      "learning_rate": 0.00019999771798221517,
-      "loss": 46.0,
-      "step": 28136
-    },
-    {
-      "epoch": 2.151270141636562,
-      "grad_norm": 0.0008816052577458322,
-      "learning_rate": 0.00019999771781994188,
-      "loss": 46.0,
-      "step": 28137
-    },
-    {
-      "epoch": 2.1513465986199516,
-      "grad_norm": 0.0015526552451774478,
-      "learning_rate": 0.0001999977176576628,
-      "loss": 46.0,
-      "step": 28138
-    },
-    {
-      "epoch": 2.1514230556033413,
-      "grad_norm": 0.0015143491327762604,
-      "learning_rate": 0.00019999771749537793,
-      "loss": 46.0,
-      "step": 28139
-    },
-    {
-      "epoch": 2.1514995125867307,
-      "grad_norm": 0.0009356113150715828,
-      "learning_rate": 0.00019999771733308732,
-      "loss": 46.0,
-      "step": 28140
-    },
-    {
-      "epoch": 2.1515759695701204,
-      "grad_norm": 0.0009212291915901005,
-      "learning_rate": 0.00019999771717079092,
-      "loss": 46.0,
-      "step": 28141
-    },
-    {
-      "epoch": 2.15165242655351,
-      "grad_norm": 0.0007717022090218961,
-      "learning_rate": 0.00019999771700848876,
-      "loss": 46.0,
-      "step": 28142
-    },
-    {
-      "epoch": 2.1517288835369,
-      "grad_norm": 0.0004468267725314945,
-      "learning_rate": 0.0001999977168461808,
-      "loss": 46.0,
-      "step": 28143
-    },
-    {
-      "epoch": 2.1518053405202897,
-      "grad_norm": 0.003542429767549038,
-      "learning_rate": 0.0001999977166838671,
-      "loss": 46.0,
-      "step": 28144
-    },
-    {
-      "epoch": 2.1518817975036795,
-      "grad_norm": 0.0007410286925733089,
-      "learning_rate": 0.00019999771652154765,
-      "loss": 46.0,
-      "step": 28145
-    },
-    {
-      "epoch": 2.1519582544870692,
-      "grad_norm": 0.000999771524220705,
-      "learning_rate": 0.0001999977163592224,
-      "loss": 46.0,
-      "step": 28146
-    },
-    {
-      "epoch": 2.152034711470459,
-      "grad_norm": 0.0009272146853618324,
-      "learning_rate": 0.0001999977161968914,
-      "loss": 46.0,
-      "step": 28147
-    },
-    {
-      "epoch": 2.1521111684538488,
-      "grad_norm": 0.0006438328418880701,
-      "learning_rate": 0.0001999977160345546,
-      "loss": 46.0,
-      "step": 28148
-    },
-    {
-      "epoch": 2.1521876254372385,
-      "grad_norm": 0.002356426790356636,
-      "learning_rate": 0.00019999771587221206,
-      "loss": 46.0,
-      "step": 28149
-    },
-    {
-      "epoch": 2.1522640824206283,
-      "grad_norm": 0.0007545550470240414,
-      "learning_rate": 0.00019999771570986373,
-      "loss": 46.0,
-      "step": 28150
-    },
-    {
-      "epoch": 2.152340539404018,
-      "grad_norm": 0.0020118916872888803,
-      "learning_rate": 0.00019999771554750963,
-      "loss": 46.0,
-      "step": 28151
-    },
-    {
-      "epoch": 2.1524169963874074,
-      "grad_norm": 0.0011059088865295053,
-      "learning_rate": 0.00019999771538514976,
-      "loss": 46.0,
-      "step": 28152
-    },
-    {
-      "epoch": 2.152493453370797,
-      "grad_norm": 0.006322795990854502,
-      "learning_rate": 0.00019999771522278414,
-      "loss": 46.0,
-      "step": 28153
-    },
-    {
-      "epoch": 2.152569910354187,
-      "grad_norm": 0.0011724461801350117,
-      "learning_rate": 0.00019999771506041275,
-      "loss": 46.0,
-      "step": 28154
-    },
-    {
-      "epoch": 2.1526463673375766,
-      "grad_norm": 0.0015976238064467907,
-      "learning_rate": 0.00019999771489803562,
-      "loss": 46.0,
-      "step": 28155
-    },
-    {
-      "epoch": 2.1527228243209664,
-      "grad_norm": 0.0013990487204864621,
-      "learning_rate": 0.00019999771473565265,
-      "loss": 46.0,
-      "step": 28156
-    },
-    {
-      "epoch": 2.152799281304356,
-      "grad_norm": 0.0005238635931164026,
-      "learning_rate": 0.00019999771457326394,
-      "loss": 46.0,
-      "step": 28157
-    },
-    {
-      "epoch": 2.152875738287746,
-      "grad_norm": 0.006004489026963711,
-      "learning_rate": 0.00019999771441086948,
-      "loss": 46.0,
-      "step": 28158
-    },
-    {
-      "epoch": 2.1529521952711357,
-      "grad_norm": 0.0007889384869486094,
-      "learning_rate": 0.00019999771424846925,
-      "loss": 46.0,
-      "step": 28159
-    },
-    {
-      "epoch": 2.1530286522545254,
-      "grad_norm": 0.0021235414315015078,
-      "learning_rate": 0.00019999771408606322,
-      "loss": 46.0,
-      "step": 28160
-    },
-    {
-      "epoch": 2.153105109237915,
-      "grad_norm": 0.005783233791589737,
-      "learning_rate": 0.00019999771392365144,
-      "loss": 46.0,
-      "step": 28161
-    },
-    {
-      "epoch": 2.1531815662213045,
-      "grad_norm": 0.0008668258087709546,
-      "learning_rate": 0.0001999977137612339,
-      "loss": 46.0,
-      "step": 28162
-    },
-    {
-      "epoch": 2.1532580232046943,
-      "grad_norm": 0.0010577006032690406,
-      "learning_rate": 0.00019999771359881057,
-      "loss": 46.0,
-      "step": 28163
-    },
-    {
-      "epoch": 2.153334480188084,
-      "grad_norm": 0.001113516860641539,
-      "learning_rate": 0.00019999771343638147,
-      "loss": 46.0,
-      "step": 28164
-    },
-    {
-      "epoch": 2.153410937171474,
-      "grad_norm": 0.002328047063201666,
-      "learning_rate": 0.0001999977132739466,
-      "loss": 46.0,
-      "step": 28165
-    },
-    {
-      "epoch": 2.1534873941548636,
-      "grad_norm": 0.0016389908269047737,
-      "learning_rate": 0.00019999771311150598,
-      "loss": 46.0,
-      "step": 28166
-    },
-    {
-      "epoch": 2.1535638511382533,
-      "grad_norm": 0.0022578593343496323,
-      "learning_rate": 0.00019999771294905961,
-      "loss": 46.0,
-      "step": 28167
-    },
-    {
-      "epoch": 2.153640308121643,
-      "grad_norm": 0.0009417764958925545,
-      "learning_rate": 0.00019999771278660742,
-      "loss": 46.0,
-      "step": 28168
-    },
-    {
-      "epoch": 2.153716765105033,
-      "grad_norm": 0.0007314240210689604,
-      "learning_rate": 0.0001999977126241495,
-      "loss": 46.0,
-      "step": 28169
-    },
-    {
-      "epoch": 2.1537932220884226,
-      "grad_norm": 0.005837566219270229,
-      "learning_rate": 0.0001999977124616858,
-      "loss": 46.0,
-      "step": 28170
-    },
-    {
-      "epoch": 2.1538696790718124,
-      "grad_norm": 0.0014265247154980898,
-      "learning_rate": 0.00019999771229921632,
-      "loss": 46.0,
-      "step": 28171
-    },
-    {
-      "epoch": 2.153946136055202,
-      "grad_norm": 0.002703160047531128,
-      "learning_rate": 0.00019999771213674109,
-      "loss": 46.0,
-      "step": 28172
-    },
-    {
-      "epoch": 2.154022593038592,
-      "grad_norm": 0.0010520732030272484,
-      "learning_rate": 0.00019999771197426008,
-      "loss": 46.0,
-      "step": 28173
-    },
-    {
-      "epoch": 2.154099050021981,
-      "grad_norm": 0.001062014140188694,
-      "learning_rate": 0.00019999771181177328,
-      "loss": 46.0,
-      "step": 28174
-    },
-    {
-      "epoch": 2.154175507005371,
-      "grad_norm": 0.0011998831760138273,
-      "learning_rate": 0.00019999771164928073,
-      "loss": 46.0,
-      "step": 28175
-    },
-    {
-      "epoch": 2.1542519639887607,
-      "grad_norm": 0.007597212214022875,
-      "learning_rate": 0.0001999977114867824,
-      "loss": 46.0,
-      "step": 28176
-    },
-    {
-      "epoch": 2.1543284209721505,
-      "grad_norm": 0.0011730515398085117,
-      "learning_rate": 0.0001999977113242783,
-      "loss": 46.0,
-      "step": 28177
-    },
-    {
-      "epoch": 2.1544048779555403,
-      "grad_norm": 0.0007938652997836471,
-      "learning_rate": 0.00019999771116176843,
-      "loss": 46.0,
-      "step": 28178
-    },
-    {
-      "epoch": 2.15448133493893,
-      "grad_norm": 0.002646838780492544,
-      "learning_rate": 0.00019999771099925282,
-      "loss": 46.0,
-      "step": 28179
-    },
-    {
-      "epoch": 2.1545577919223198,
-      "grad_norm": 0.001280545606277883,
-      "learning_rate": 0.00019999771083673143,
-      "loss": 46.0,
-      "step": 28180
-    },
-    {
-      "epoch": 2.1546342489057095,
-      "grad_norm": 0.0018157809972763062,
-      "learning_rate": 0.00019999771067420426,
-      "loss": 46.0,
-      "step": 28181
-    },
-    {
-      "epoch": 2.1547107058890993,
-      "grad_norm": 0.0014921803958714008,
-      "learning_rate": 0.0001999977105116713,
-      "loss": 46.0,
-      "step": 28182
-    },
-    {
-      "epoch": 2.154787162872489,
-      "grad_norm": 0.0003949670644942671,
-      "learning_rate": 0.00019999771034913261,
-      "loss": 46.0,
-      "step": 28183
-    },
-    {
-      "epoch": 2.1548636198558784,
-      "grad_norm": 0.0010988599387928843,
-      "learning_rate": 0.00019999771018658813,
-      "loss": 46.0,
-      "step": 28184
-    },
-    {
-      "epoch": 2.154940076839268,
-      "grad_norm": 0.001070576487109065,
-      "learning_rate": 0.0001999977100240379,
-      "loss": 46.0,
-      "step": 28185
-    },
-    {
-      "epoch": 2.155016533822658,
-      "grad_norm": 0.000707069702912122,
-      "learning_rate": 0.00019999770986148184,
-      "loss": 46.0,
-      "step": 28186
-    },
-    {
-      "epoch": 2.1550929908060477,
-      "grad_norm": 0.0013154692715033889,
-      "learning_rate": 0.0001999977096989201,
-      "loss": 46.0,
-      "step": 28187
-    },
-    {
-      "epoch": 2.1551694477894374,
-      "grad_norm": 0.0007828108500689268,
-      "learning_rate": 0.00019999770953635254,
-      "loss": 46.0,
-      "step": 28188
-    },
-    {
-      "epoch": 2.155245904772827,
-      "grad_norm": 0.0027716923505067825,
-      "learning_rate": 0.00019999770937377922,
-      "loss": 46.0,
-      "step": 28189
-    },
-    {
-      "epoch": 2.155322361756217,
-      "grad_norm": 0.0010585255222395062,
-      "learning_rate": 0.00019999770921120012,
-      "loss": 46.0,
-      "step": 28190
-    },
-    {
-      "epoch": 2.1553988187396067,
-      "grad_norm": 0.0016748480265960097,
-      "learning_rate": 0.00019999770904861525,
-      "loss": 46.0,
-      "step": 28191
-    },
-    {
-      "epoch": 2.1554752757229965,
-      "grad_norm": 0.0008144259918481112,
-      "learning_rate": 0.00019999770888602463,
-      "loss": 46.0,
-      "step": 28192
-    },
-    {
-      "epoch": 2.1555517327063862,
-      "grad_norm": 0.0015611799899488688,
-      "learning_rate": 0.00019999770872342821,
-      "loss": 46.0,
-      "step": 28193
-    },
-    {
-      "epoch": 2.155628189689776,
-      "grad_norm": 0.0008937178063206375,
-      "learning_rate": 0.00019999770856082605,
-      "loss": 46.0,
-      "step": 28194
-    },
-    {
-      "epoch": 2.1557046466731653,
-      "grad_norm": 0.003062192350625992,
-      "learning_rate": 0.0001999977083982181,
-      "loss": 46.0,
-      "step": 28195
-    },
-    {
-      "epoch": 2.155781103656555,
-      "grad_norm": 0.0013641052646562457,
-      "learning_rate": 0.00019999770823560443,
-      "loss": 46.0,
-      "step": 28196
-    },
-    {
-      "epoch": 2.155857560639945,
-      "grad_norm": 0.002345721935853362,
-      "learning_rate": 0.00019999770807298492,
-      "loss": 46.0,
-      "step": 28197
-    },
-    {
-      "epoch": 2.1559340176233346,
-      "grad_norm": 0.0016041523776948452,
-      "learning_rate": 0.0001999977079103597,
-      "loss": 46.0,
-      "step": 28198
-    },
-    {
-      "epoch": 2.1560104746067243,
-      "grad_norm": 0.002357837976887822,
-      "learning_rate": 0.00019999770774772869,
-      "loss": 46.0,
-      "step": 28199
-    },
-    {
-      "epoch": 2.156086931590114,
-      "grad_norm": 0.00382078904658556,
-      "learning_rate": 0.00019999770758509188,
-      "loss": 46.0,
-      "step": 28200
-    },
-    {
-      "epoch": 2.156163388573504,
-      "grad_norm": 0.0013290635542944074,
-      "learning_rate": 0.00019999770742244933,
-      "loss": 46.0,
-      "step": 28201
-    },
-    {
-      "epoch": 2.1562398455568936,
-      "grad_norm": 0.0005938831018283963,
-      "learning_rate": 0.000199997707259801,
-      "loss": 46.0,
-      "step": 28202
-    },
-    {
-      "epoch": 2.1563163025402834,
-      "grad_norm": 0.0008766081882640719,
-      "learning_rate": 0.0001999977070971469,
-      "loss": 46.0,
-      "step": 28203
-    },
-    {
-      "epoch": 2.156392759523673,
-      "grad_norm": 0.0023452858440577984,
-      "learning_rate": 0.00019999770693448704,
-      "loss": 46.0,
-      "step": 28204
-    },
-    {
-      "epoch": 2.156469216507063,
-      "grad_norm": 0.00046968087553977966,
-      "learning_rate": 0.00019999770677182142,
-      "loss": 46.0,
-      "step": 28205
-    },
-    {
-      "epoch": 2.1565456734904522,
-      "grad_norm": 0.0012829380575567484,
-      "learning_rate": 0.00019999770660915003,
-      "loss": 46.0,
-      "step": 28206
-    },
-    {
-      "epoch": 2.156622130473842,
-      "grad_norm": 0.001255856710486114,
-      "learning_rate": 0.00019999770644647284,
-      "loss": 46.0,
-      "step": 28207
-    },
-    {
-      "epoch": 2.1566985874572318,
-      "grad_norm": 0.0016059057088568807,
-      "learning_rate": 0.00019999770628378988,
-      "loss": 46.0,
-      "step": 28208
-    },
-    {
-      "epoch": 2.1567750444406215,
-      "grad_norm": 0.00527126295492053,
-      "learning_rate": 0.00019999770612110117,
-      "loss": 46.0,
-      "step": 28209
-    },
-    {
-      "epoch": 2.1568515014240113,
-      "grad_norm": 0.0012687172275036573,
-      "learning_rate": 0.0001999977059584067,
-      "loss": 46.0,
-      "step": 28210
-    },
-    {
-      "epoch": 2.156927958407401,
-      "grad_norm": 0.001291608321480453,
-      "learning_rate": 0.00019999770579570648,
-      "loss": 46.0,
-      "step": 28211
-    },
-    {
-      "epoch": 2.157004415390791,
-      "grad_norm": 0.0018162453779950738,
-      "learning_rate": 0.00019999770563300045,
-      "loss": 46.0,
-      "step": 28212
-    },
-    {
-      "epoch": 2.1570808723741806,
-      "grad_norm": 0.0008445921703241765,
-      "learning_rate": 0.00019999770547028868,
-      "loss": 46.0,
-      "step": 28213
-    },
-    {
-      "epoch": 2.1571573293575703,
-      "grad_norm": 0.001552355708554387,
-      "learning_rate": 0.0001999977053075711,
-      "loss": 46.0,
-      "step": 28214
-    },
-    {
-      "epoch": 2.15723378634096,
-      "grad_norm": 0.0003529755340423435,
-      "learning_rate": 0.0001999977051448478,
-      "loss": 46.0,
-      "step": 28215
-    },
-    {
-      "epoch": 2.15731024332435,
-      "grad_norm": 0.0006778242532163858,
-      "learning_rate": 0.00019999770498211868,
-      "loss": 46.0,
-      "step": 28216
-    },
-    {
-      "epoch": 2.157386700307739,
-      "grad_norm": 0.001351646613329649,
-      "learning_rate": 0.00019999770481938384,
-      "loss": 46.0,
-      "step": 28217
-    },
-    {
-      "epoch": 2.157463157291129,
-      "grad_norm": 0.0038238682318478823,
-      "learning_rate": 0.0001999977046566432,
-      "loss": 46.0,
-      "step": 28218
-    },
-    {
-      "epoch": 2.1575396142745187,
-      "grad_norm": 0.0014866123674437404,
-      "learning_rate": 0.00019999770449389678,
-      "loss": 46.0,
-      "step": 28219
-    },
-    {
-      "epoch": 2.1576160712579084,
-      "grad_norm": 0.0018138274317607284,
-      "learning_rate": 0.0001999977043311446,
-      "loss": 46.0,
-      "step": 28220
-    },
-    {
-      "epoch": 2.157692528241298,
-      "grad_norm": 0.0011176791740581393,
-      "learning_rate": 0.00019999770416838668,
-      "loss": 46.0,
-      "step": 28221
-    },
-    {
-      "epoch": 2.157768985224688,
-      "grad_norm": 0.0014468267327174544,
-      "learning_rate": 0.00019999770400562297,
-      "loss": 46.0,
-      "step": 28222
-    },
-    {
-      "epoch": 2.1578454422080777,
-      "grad_norm": 0.0005226039793342352,
-      "learning_rate": 0.00019999770384285348,
-      "loss": 46.0,
-      "step": 28223
-    },
-    {
-      "epoch": 2.1579218991914675,
-      "grad_norm": 0.0016708774492144585,
-      "learning_rate": 0.00019999770368007825,
-      "loss": 46.0,
-      "step": 28224
-    },
-    {
-      "epoch": 2.1579983561748572,
-      "grad_norm": 0.000577491067815572,
-      "learning_rate": 0.0001999977035172972,
-      "loss": 46.0,
-      "step": 28225
-    },
-    {
-      "epoch": 2.158074813158247,
-      "grad_norm": 0.0012306488351896405,
-      "learning_rate": 0.00019999770335451042,
-      "loss": 46.0,
-      "step": 28226
-    },
-    {
-      "epoch": 2.1581512701416368,
-      "grad_norm": 0.0006994212744757533,
-      "learning_rate": 0.00019999770319171787,
-      "loss": 46.0,
-      "step": 28227
-    },
-    {
-      "epoch": 2.158227727125026,
-      "grad_norm": 0.0019818840082734823,
-      "learning_rate": 0.00019999770302891955,
-      "loss": 46.0,
-      "step": 28228
-    },
-    {
-      "epoch": 2.158304184108416,
-      "grad_norm": 0.0004970363224856555,
-      "learning_rate": 0.00019999770286611542,
-      "loss": 46.0,
-      "step": 28229
-    },
-    {
-      "epoch": 2.1583806410918056,
-      "grad_norm": 0.0019437340088188648,
-      "learning_rate": 0.00019999770270330558,
-      "loss": 46.0,
-      "step": 28230
-    },
-    {
-      "epoch": 2.1584570980751954,
-      "grad_norm": 0.0007514884928241372,
-      "learning_rate": 0.00019999770254048994,
-      "loss": 46.0,
-      "step": 28231
-    },
-    {
-      "epoch": 2.158533555058585,
-      "grad_norm": 0.001444541267119348,
-      "learning_rate": 0.00019999770237766855,
-      "loss": 46.0,
-      "step": 28232
-    },
-    {
-      "epoch": 2.158610012041975,
-      "grad_norm": 0.0014975356170907617,
-      "learning_rate": 0.00019999770221484139,
-      "loss": 46.0,
-      "step": 28233
-    },
-    {
-      "epoch": 2.1586864690253647,
-      "grad_norm": 0.0008805756224319339,
-      "learning_rate": 0.00019999770205200842,
-      "loss": 46.0,
-      "step": 28234
-    },
-    {
-      "epoch": 2.1587629260087544,
-      "grad_norm": 0.0008572934893891215,
-      "learning_rate": 0.0001999977018891697,
-      "loss": 46.0,
-      "step": 28235
-    },
-    {
-      "epoch": 2.158839382992144,
-      "grad_norm": 0.0011078552342951298,
-      "learning_rate": 0.00019999770172632523,
-      "loss": 46.0,
-      "step": 28236
-    },
-    {
-      "epoch": 2.158915839975534,
-      "grad_norm": 0.0005424771225079894,
-      "learning_rate": 0.000199997701563475,
-      "loss": 46.0,
-      "step": 28237
-    },
-    {
-      "epoch": 2.1589922969589237,
-      "grad_norm": 0.0015068185748532414,
-      "learning_rate": 0.00019999770140061894,
-      "loss": 46.0,
-      "step": 28238
-    },
-    {
-      "epoch": 2.159068753942313,
-      "grad_norm": 0.002984751481562853,
-      "learning_rate": 0.0001999977012377572,
-      "loss": 46.0,
-      "step": 28239
-    },
-    {
-      "epoch": 2.1591452109257028,
-      "grad_norm": 0.0026946093421429396,
-      "learning_rate": 0.00019999770107488962,
-      "loss": 46.0,
-      "step": 28240
-    },
-    {
-      "epoch": 2.1592216679090925,
-      "grad_norm": 0.0006708583678118885,
-      "learning_rate": 0.0001999977009120163,
-      "loss": 46.0,
-      "step": 28241
-    },
-    {
-      "epoch": 2.1592981248924823,
-      "grad_norm": 0.0010932415025308728,
-      "learning_rate": 0.0001999977007491372,
-      "loss": 46.0,
-      "step": 28242
-    },
-    {
-      "epoch": 2.159374581875872,
-      "grad_norm": 0.0019411728717386723,
-      "learning_rate": 0.00019999770058625233,
-      "loss": 46.0,
-      "step": 28243
-    },
-    {
-      "epoch": 2.159451038859262,
-      "grad_norm": 0.0012245491379871964,
-      "learning_rate": 0.0001999977004233617,
-      "loss": 46.0,
-      "step": 28244
-    },
-    {
-      "epoch": 2.1595274958426516,
-      "grad_norm": 0.0008670325041748583,
-      "learning_rate": 0.0001999977002604653,
-      "loss": 46.0,
-      "step": 28245
-    },
-    {
-      "epoch": 2.1596039528260413,
-      "grad_norm": 0.0008080180268734694,
-      "learning_rate": 0.0001999977000975631,
-      "loss": 46.0,
-      "step": 28246
-    },
-    {
-      "epoch": 2.159680409809431,
-      "grad_norm": 0.0008524100994691253,
-      "learning_rate": 0.00019999769993465515,
-      "loss": 46.0,
-      "step": 28247
-    },
-    {
-      "epoch": 2.159756866792821,
-      "grad_norm": 0.001763631938956678,
-      "learning_rate": 0.00019999769977174146,
-      "loss": 46.0,
-      "step": 28248
-    },
-    {
-      "epoch": 2.15983332377621,
-      "grad_norm": 0.0016128664137795568,
-      "learning_rate": 0.00019999769960882198,
-      "loss": 46.0,
-      "step": 28249
-    },
-    {
-      "epoch": 2.1599097807596,
-      "grad_norm": 0.0004104067920707166,
-      "learning_rate": 0.00019999769944589673,
-      "loss": 46.0,
-      "step": 28250
-    },
-    {
-      "epoch": 2.1599862377429897,
-      "grad_norm": 0.00031212990870699286,
-      "learning_rate": 0.0001999976992829657,
-      "loss": 46.0,
-      "step": 28251
-    },
-    {
-      "epoch": 2.1600626947263795,
-      "grad_norm": 0.0012686926638707519,
-      "learning_rate": 0.00019999769912002892,
-      "loss": 46.0,
-      "step": 28252
-    },
-    {
-      "epoch": 2.1601391517097692,
-      "grad_norm": 0.0009924668120220304,
-      "learning_rate": 0.00019999769895708634,
-      "loss": 46.0,
-      "step": 28253
-    },
-    {
-      "epoch": 2.160215608693159,
-      "grad_norm": 0.000501585251186043,
-      "learning_rate": 0.00019999769879413802,
-      "loss": 46.0,
-      "step": 28254
-    },
-    {
-      "epoch": 2.1602920656765487,
-      "grad_norm": 0.0016998882638290524,
-      "learning_rate": 0.00019999769863118393,
-      "loss": 46.0,
-      "step": 28255
-    },
-    {
-      "epoch": 2.1603685226599385,
-      "grad_norm": 0.0015705376863479614,
-      "learning_rate": 0.00019999769846822406,
-      "loss": 46.0,
-      "step": 28256
-    },
-    {
-      "epoch": 2.1604449796433283,
-      "grad_norm": 0.0014292121632024646,
-      "learning_rate": 0.00019999769830525844,
-      "loss": 46.0,
-      "step": 28257
-    },
-    {
-      "epoch": 2.160521436626718,
-      "grad_norm": 0.0019807922653853893,
-      "learning_rate": 0.00019999769814228703,
-      "loss": 46.0,
-      "step": 28258
-    },
-    {
-      "epoch": 2.160597893610108,
-      "grad_norm": 0.0015184932854026556,
-      "learning_rate": 0.00019999769797930984,
-      "loss": 46.0,
-      "step": 28259
-    },
-    {
-      "epoch": 2.1606743505934976,
-      "grad_norm": 0.0026066794525831938,
-      "learning_rate": 0.0001999976978163269,
-      "loss": 46.0,
-      "step": 28260
-    },
-    {
-      "epoch": 2.160750807576887,
-      "grad_norm": 0.004086014349013567,
-      "learning_rate": 0.0001999976976533382,
-      "loss": 46.0,
-      "step": 28261
-    },
-    {
-      "epoch": 2.1608272645602766,
-      "grad_norm": 0.0009970151586458087,
-      "learning_rate": 0.00019999769749034368,
-      "loss": 46.0,
-      "step": 28262
-    },
-    {
-      "epoch": 2.1609037215436664,
-      "grad_norm": 0.002410683548077941,
-      "learning_rate": 0.00019999769732734345,
-      "loss": 46.0,
-      "step": 28263
-    },
-    {
-      "epoch": 2.160980178527056,
-      "grad_norm": 0.0016978831263259053,
-      "learning_rate": 0.00019999769716433743,
-      "loss": 46.0,
-      "step": 28264
-    },
-    {
-      "epoch": 2.161056635510446,
-      "grad_norm": 0.0009385807788930833,
-      "learning_rate": 0.00019999769700132562,
-      "loss": 46.0,
-      "step": 28265
-    },
-    {
-      "epoch": 2.1611330924938357,
-      "grad_norm": 0.000730443571228534,
-      "learning_rate": 0.00019999769683830807,
-      "loss": 46.0,
-      "step": 28266
-    },
-    {
-      "epoch": 2.1612095494772254,
-      "grad_norm": 0.0020312333945184946,
-      "learning_rate": 0.00019999769667528475,
-      "loss": 46.0,
-      "step": 28267
-    },
-    {
-      "epoch": 2.161286006460615,
-      "grad_norm": 0.000674801180139184,
-      "learning_rate": 0.00019999769651225563,
-      "loss": 46.0,
-      "step": 28268
-    },
-    {
-      "epoch": 2.161362463444005,
-      "grad_norm": 0.001050399150699377,
-      "learning_rate": 0.00019999769634922076,
-      "loss": 46.0,
-      "step": 28269
-    },
-    {
-      "epoch": 2.1614389204273947,
-      "grad_norm": 0.0018419305561110377,
-      "learning_rate": 0.00019999769618618015,
-      "loss": 46.0,
-      "step": 28270
-    },
-    {
-      "epoch": 2.161515377410784,
-      "grad_norm": 0.0010177392978221178,
-      "learning_rate": 0.00019999769602313373,
-      "loss": 46.0,
-      "step": 28271
-    },
-    {
-      "epoch": 2.161591834394174,
-      "grad_norm": 0.0005326177342794836,
-      "learning_rate": 0.00019999769586008154,
-      "loss": 46.0,
-      "step": 28272
-    },
-    {
-      "epoch": 2.1616682913775636,
-      "grad_norm": 0.0018263114616274834,
-      "learning_rate": 0.00019999769569702363,
-      "loss": 46.0,
-      "step": 28273
-    },
-    {
-      "epoch": 2.1617447483609533,
-      "grad_norm": 0.001399523694999516,
-      "learning_rate": 0.00019999769553395987,
-      "loss": 46.0,
-      "step": 28274
-    },
-    {
-      "epoch": 2.161821205344343,
-      "grad_norm": 0.002858817344531417,
-      "learning_rate": 0.00019999769537089042,
-      "loss": 46.0,
-      "step": 28275
-    },
-    {
-      "epoch": 2.161897662327733,
-      "grad_norm": 0.002031118143349886,
-      "learning_rate": 0.00019999769520781516,
-      "loss": 46.0,
-      "step": 28276
-    },
-    {
-      "epoch": 2.1619741193111226,
-      "grad_norm": 0.0006862554582767189,
-      "learning_rate": 0.00019999769504473413,
-      "loss": 46.0,
-      "step": 28277
-    },
-    {
-      "epoch": 2.1620505762945124,
-      "grad_norm": 0.0006814321968704462,
-      "learning_rate": 0.00019999769488164733,
-      "loss": 46.0,
-      "step": 28278
-    },
-    {
-      "epoch": 2.162127033277902,
-      "grad_norm": 0.00224114628508687,
-      "learning_rate": 0.00019999769471855475,
-      "loss": 46.0,
-      "step": 28279
-    },
-    {
-      "epoch": 2.162203490261292,
-      "grad_norm": 0.00044544244883581996,
-      "learning_rate": 0.00019999769455545646,
-      "loss": 46.0,
-      "step": 28280
-    },
-    {
-      "epoch": 2.1622799472446816,
-      "grad_norm": 0.0017573442310094833,
-      "learning_rate": 0.00019999769439235234,
-      "loss": 46.0,
-      "step": 28281
-    },
-    {
-      "epoch": 2.1623564042280714,
-      "grad_norm": 0.00034093158319592476,
-      "learning_rate": 0.0001999976942292425,
-      "loss": 46.0,
-      "step": 28282
-    },
-    {
-      "epoch": 2.1624328612114607,
-      "grad_norm": 0.0009124795324169099,
-      "learning_rate": 0.00019999769406612683,
-      "loss": 46.0,
-      "step": 28283
-    },
-    {
-      "epoch": 2.1625093181948505,
-      "grad_norm": 0.0010847555240616202,
-      "learning_rate": 0.00019999769390300544,
-      "loss": 46.0,
-      "step": 28284
-    },
-    {
-      "epoch": 2.1625857751782402,
-      "grad_norm": 0.0007826763321645558,
-      "learning_rate": 0.00019999769373987825,
-      "loss": 46.0,
-      "step": 28285
-    },
-    {
-      "epoch": 2.16266223216163,
-      "grad_norm": 0.0010230930056422949,
-      "learning_rate": 0.00019999769357674532,
-      "loss": 46.0,
-      "step": 28286
-    },
-    {
-      "epoch": 2.1627386891450198,
-      "grad_norm": 0.0009571608970873058,
-      "learning_rate": 0.00019999769341360658,
-      "loss": 46.0,
-      "step": 28287
-    },
-    {
-      "epoch": 2.1628151461284095,
-      "grad_norm": 0.000638592173345387,
-      "learning_rate": 0.0001999976932504621,
-      "loss": 46.0,
-      "step": 28288
-    },
-    {
-      "epoch": 2.1628916031117993,
-      "grad_norm": 0.00393934641033411,
-      "learning_rate": 0.00019999769308731184,
-      "loss": 46.0,
-      "step": 28289
-    },
-    {
-      "epoch": 2.162968060095189,
-      "grad_norm": 0.0010481675853952765,
-      "learning_rate": 0.00019999769292415581,
-      "loss": 46.0,
-      "step": 28290
-    },
-    {
-      "epoch": 2.163044517078579,
-      "grad_norm": 0.0017305562505498528,
-      "learning_rate": 0.00019999769276099404,
-      "loss": 46.0,
-      "step": 28291
-    },
-    {
-      "epoch": 2.1631209740619686,
-      "grad_norm": 0.001234453055076301,
-      "learning_rate": 0.00019999769259782646,
-      "loss": 46.0,
-      "step": 28292
-    },
-    {
-      "epoch": 2.163197431045358,
-      "grad_norm": 0.001165780471637845,
-      "learning_rate": 0.00019999769243465314,
-      "loss": 46.0,
-      "step": 28293
-    },
-    {
-      "epoch": 2.1632738880287476,
-      "grad_norm": 0.0007124131661839783,
-      "learning_rate": 0.00019999769227147405,
-      "loss": 46.0,
-      "step": 28294
-    },
-    {
-      "epoch": 2.1633503450121374,
-      "grad_norm": 0.0036853512283414602,
-      "learning_rate": 0.00019999769210828915,
-      "loss": 46.0,
-      "step": 28295
-    },
-    {
-      "epoch": 2.163426801995527,
-      "grad_norm": 0.0039886184968054295,
-      "learning_rate": 0.0001999976919450985,
-      "loss": 46.0,
-      "step": 28296
-    },
-    {
-      "epoch": 2.163503258978917,
-      "grad_norm": 0.004854655358940363,
-      "learning_rate": 0.0001999976917819021,
-      "loss": 46.0,
-      "step": 28297
-    },
-    {
-      "epoch": 2.1635797159623067,
-      "grad_norm": 0.0009602952050045133,
-      "learning_rate": 0.00019999769161869994,
-      "loss": 46.0,
-      "step": 28298
-    },
-    {
-      "epoch": 2.1636561729456965,
-      "grad_norm": 0.0008612944511696696,
-      "learning_rate": 0.00019999769145549197,
-      "loss": 46.0,
-      "step": 28299
-    },
-    {
-      "epoch": 2.163732629929086,
-      "grad_norm": 0.001840696088038385,
-      "learning_rate": 0.00019999769129227827,
-      "loss": 46.0,
-      "step": 28300
-    },
-    {
-      "epoch": 2.163809086912476,
-      "grad_norm": 0.0011351652210578322,
-      "learning_rate": 0.00019999769112905878,
-      "loss": 46.0,
-      "step": 28301
-    },
-    {
-      "epoch": 2.1638855438958657,
-      "grad_norm": 0.0006566941738128662,
-      "learning_rate": 0.00019999769096583353,
-      "loss": 46.0,
-      "step": 28302
-    },
-    {
-      "epoch": 2.1639620008792555,
-      "grad_norm": 0.0008486962760798633,
-      "learning_rate": 0.0001999976908026025,
-      "loss": 46.0,
-      "step": 28303
-    },
-    {
-      "epoch": 2.1640384578626453,
-      "grad_norm": 0.001236417330801487,
-      "learning_rate": 0.0001999976906393657,
-      "loss": 46.0,
-      "step": 28304
-    },
-    {
-      "epoch": 2.1641149148460346,
-      "grad_norm": 0.0004845599760301411,
-      "learning_rate": 0.00019999769047612315,
-      "loss": 46.0,
-      "step": 28305
-    },
-    {
-      "epoch": 2.1641913718294243,
-      "grad_norm": 0.0007713251980021596,
-      "learning_rate": 0.0001999976903128748,
-      "loss": 46.0,
-      "step": 28306
-    },
-    {
-      "epoch": 2.164267828812814,
-      "grad_norm": 0.0032170279882848263,
-      "learning_rate": 0.0001999976901496207,
-      "loss": 46.0,
-      "step": 28307
-    },
-    {
-      "epoch": 2.164344285796204,
-      "grad_norm": 0.001863210811279714,
-      "learning_rate": 0.00019999768998636084,
-      "loss": 46.0,
-      "step": 28308
-    },
-    {
-      "epoch": 2.1644207427795936,
-      "grad_norm": 0.0010804088087752461,
-      "learning_rate": 0.0001999976898230952,
-      "loss": 46.0,
-      "step": 28309
-    },
-    {
-      "epoch": 2.1644971997629834,
-      "grad_norm": 0.001021967502310872,
-      "learning_rate": 0.00019999768965982376,
-      "loss": 46.0,
-      "step": 28310
-    },
-    {
-      "epoch": 2.164573656746373,
-      "grad_norm": 0.0004115740885026753,
-      "learning_rate": 0.0001999976894965466,
-      "loss": 46.0,
-      "step": 28311
-    },
-    {
-      "epoch": 2.164650113729763,
-      "grad_norm": 0.0034954624716192484,
-      "learning_rate": 0.00019999768933326366,
-      "loss": 46.0,
-      "step": 28312
-    },
-    {
-      "epoch": 2.1647265707131527,
-      "grad_norm": 0.0009458708227612078,
-      "learning_rate": 0.00019999768916997496,
-      "loss": 46.0,
-      "step": 28313
-    },
-    {
-      "epoch": 2.1648030276965424,
-      "grad_norm": 0.001612994121387601,
-      "learning_rate": 0.00019999768900668045,
-      "loss": 46.0,
-      "step": 28314
-    },
-    {
-      "epoch": 2.1648794846799317,
-      "grad_norm": 0.0025195502676069736,
-      "learning_rate": 0.0001999976888433802,
-      "loss": 46.0,
-      "step": 28315
-    },
-    {
-      "epoch": 2.1649559416633215,
-      "grad_norm": 0.0018592727137729526,
-      "learning_rate": 0.00019999768868007417,
-      "loss": 46.0,
-      "step": 28316
-    },
-    {
-      "epoch": 2.1650323986467113,
-      "grad_norm": 0.0014857631176710129,
-      "learning_rate": 0.00019999768851676234,
-      "loss": 46.0,
-      "step": 28317
-    },
-    {
-      "epoch": 2.165108855630101,
-      "grad_norm": 0.0009768782183527946,
-      "learning_rate": 0.00019999768835344482,
-      "loss": 46.0,
-      "step": 28318
-    },
-    {
-      "epoch": 2.165185312613491,
-      "grad_norm": 0.0011763478396460414,
-      "learning_rate": 0.00019999768819012147,
-      "loss": 46.0,
-      "step": 28319
-    },
-    {
-      "epoch": 2.1652617695968805,
-      "grad_norm": 0.0009184136870317161,
-      "learning_rate": 0.00019999768802679238,
-      "loss": 46.0,
-      "step": 28320
-    },
-    {
-      "epoch": 2.1653382265802703,
-      "grad_norm": 0.0015678530326113105,
-      "learning_rate": 0.00019999768786345748,
-      "loss": 46.0,
-      "step": 28321
-    },
-    {
-      "epoch": 2.16541468356366,
-      "grad_norm": 0.0006505833007395267,
-      "learning_rate": 0.00019999768770011684,
-      "loss": 46.0,
-      "step": 28322
-    },
-    {
-      "epoch": 2.16549114054705,
-      "grad_norm": 0.00388232059776783,
-      "learning_rate": 0.00019999768753677045,
-      "loss": 46.0,
-      "step": 28323
-    },
-    {
-      "epoch": 2.1655675975304396,
-      "grad_norm": 0.0018786248983815312,
-      "learning_rate": 0.00019999768737341827,
-      "loss": 46.0,
-      "step": 28324
-    },
-    {
-      "epoch": 2.1656440545138294,
-      "grad_norm": 0.0014186324551701546,
-      "learning_rate": 0.0001999976872100603,
-      "loss": 46.0,
-      "step": 28325
-    },
-    {
-      "epoch": 2.1657205114972187,
-      "grad_norm": 0.0012227930128574371,
-      "learning_rate": 0.0001999976870466966,
-      "loss": 46.0,
-      "step": 28326
-    },
-    {
-      "epoch": 2.1657969684806084,
-      "grad_norm": 0.0009038235293701291,
-      "learning_rate": 0.0001999976868833271,
-      "loss": 46.0,
-      "step": 28327
-    },
-    {
-      "epoch": 2.165873425463998,
-      "grad_norm": 0.0005395268672145903,
-      "learning_rate": 0.00019999768671995183,
-      "loss": 46.0,
-      "step": 28328
-    },
-    {
-      "epoch": 2.165949882447388,
-      "grad_norm": 0.0006006858311593533,
-      "learning_rate": 0.0001999976865565708,
-      "loss": 46.0,
-      "step": 28329
-    },
-    {
-      "epoch": 2.1660263394307777,
-      "grad_norm": 0.0006018558633513749,
-      "learning_rate": 0.000199997686393184,
-      "loss": 46.0,
-      "step": 28330
-    },
-    {
-      "epoch": 2.1661027964141675,
-      "grad_norm": 0.001946346485055983,
-      "learning_rate": 0.00019999768622979143,
-      "loss": 46.0,
-      "step": 28331
-    },
-    {
-      "epoch": 2.1661792533975572,
-      "grad_norm": 0.0005670227692462504,
-      "learning_rate": 0.0001999976860663931,
-      "loss": 46.0,
-      "step": 28332
-    },
-    {
-      "epoch": 2.166255710380947,
-      "grad_norm": 0.002674239221960306,
-      "learning_rate": 0.000199997685902989,
-      "loss": 46.0,
-      "step": 28333
-    },
-    {
-      "epoch": 2.1663321673643368,
-      "grad_norm": 0.0036192391999065876,
-      "learning_rate": 0.00019999768573957912,
-      "loss": 46.0,
-      "step": 28334
-    },
-    {
-      "epoch": 2.1664086243477265,
-      "grad_norm": 0.0012197496835142374,
-      "learning_rate": 0.00019999768557616348,
-      "loss": 46.0,
-      "step": 28335
-    },
-    {
-      "epoch": 2.1664850813311163,
-      "grad_norm": 0.0017973009962588549,
-      "learning_rate": 0.00019999768541274207,
-      "loss": 46.0,
-      "step": 28336
-    },
-    {
-      "epoch": 2.1665615383145056,
-      "grad_norm": 0.0008222817559726536,
-      "learning_rate": 0.00019999768524931488,
-      "loss": 46.0,
-      "step": 28337
-    },
-    {
-      "epoch": 2.1666379952978954,
-      "grad_norm": 0.0007087428239174187,
-      "learning_rate": 0.00019999768508588192,
-      "loss": 46.0,
-      "step": 28338
-    },
-    {
-      "epoch": 2.166714452281285,
-      "grad_norm": 0.0010377614526078105,
-      "learning_rate": 0.00019999768492244321,
-      "loss": 46.0,
-      "step": 28339
-    },
-    {
-      "epoch": 2.166790909264675,
-      "grad_norm": 0.00105758267454803,
-      "learning_rate": 0.00019999768475899873,
-      "loss": 46.0,
-      "step": 28340
-    },
-    {
-      "epoch": 2.1668673662480646,
-      "grad_norm": 0.001613438711501658,
-      "learning_rate": 0.00019999768459554845,
-      "loss": 46.0,
-      "step": 28341
-    },
-    {
-      "epoch": 2.1669438232314544,
-      "grad_norm": 0.0007211464690044522,
-      "learning_rate": 0.00019999768443209243,
-      "loss": 46.0,
-      "step": 28342
-    },
-    {
-      "epoch": 2.167020280214844,
-      "grad_norm": 0.0007866926025599241,
-      "learning_rate": 0.00019999768426863062,
-      "loss": 46.0,
-      "step": 28343
-    },
-    {
-      "epoch": 2.167096737198234,
-      "grad_norm": 0.0007860049372538924,
-      "learning_rate": 0.00019999768410516305,
-      "loss": 46.0,
-      "step": 28344
-    },
-    {
-      "epoch": 2.1671731941816237,
-      "grad_norm": 0.013147357851266861,
-      "learning_rate": 0.00019999768394168973,
-      "loss": 46.0,
-      "step": 28345
-    },
-    {
-      "epoch": 2.1672496511650134,
-      "grad_norm": 0.0018320149974897504,
-      "learning_rate": 0.0001999976837782106,
-      "loss": 46.0,
-      "step": 28346
-    },
-    {
-      "epoch": 2.167326108148403,
-      "grad_norm": 0.0005081528215669096,
-      "learning_rate": 0.00019999768361472574,
-      "loss": 46.0,
-      "step": 28347
-    },
-    {
-      "epoch": 2.1674025651317925,
-      "grad_norm": 0.0013384188059717417,
-      "learning_rate": 0.0001999976834512351,
-      "loss": 46.0,
-      "step": 28348
-    },
-    {
-      "epoch": 2.1674790221151823,
-      "grad_norm": 0.001732338685542345,
-      "learning_rate": 0.0001999976832877387,
-      "loss": 46.0,
-      "step": 28349
-    },
-    {
-      "epoch": 2.167555479098572,
-      "grad_norm": 0.011148832738399506,
-      "learning_rate": 0.0001999976831242365,
-      "loss": 46.0,
-      "step": 28350
-    },
-    {
-      "epoch": 2.167631936081962,
-      "grad_norm": 0.0025218259543180466,
-      "learning_rate": 0.00019999768296072854,
-      "loss": 46.0,
-      "step": 28351
-    },
-    {
-      "epoch": 2.1677083930653516,
-      "grad_norm": 0.0008667050278745592,
-      "learning_rate": 0.00019999768279721484,
-      "loss": 46.0,
-      "step": 28352
-    },
-    {
-      "epoch": 2.1677848500487413,
-      "grad_norm": 0.0012461276492103934,
-      "learning_rate": 0.00019999768263369533,
-      "loss": 46.0,
-      "step": 28353
-    },
-    {
-      "epoch": 2.167861307032131,
-      "grad_norm": 0.0017693970585241914,
-      "learning_rate": 0.00019999768247017007,
-      "loss": 46.0,
-      "step": 28354
-    },
-    {
-      "epoch": 2.167937764015521,
-      "grad_norm": 0.0010632730554789305,
-      "learning_rate": 0.00019999768230663905,
-      "loss": 46.0,
-      "step": 28355
-    },
-    {
-      "epoch": 2.1680142209989106,
-      "grad_norm": 0.00891291256994009,
-      "learning_rate": 0.00019999768214310225,
-      "loss": 46.0,
-      "step": 28356
-    },
-    {
-      "epoch": 2.1680906779823004,
-      "grad_norm": 0.0016254974761977792,
-      "learning_rate": 0.00019999768197955967,
-      "loss": 46.0,
-      "step": 28357
-    },
-    {
-      "epoch": 2.16816713496569,
-      "grad_norm": 0.0012648567790165544,
-      "learning_rate": 0.00019999768181601133,
-      "loss": 46.0,
-      "step": 28358
-    },
-    {
-      "epoch": 2.1682435919490795,
-      "grad_norm": 0.0013248417526483536,
-      "learning_rate": 0.00019999768165245723,
-      "loss": 46.0,
-      "step": 28359
-    },
-    {
-      "epoch": 2.168320048932469,
-      "grad_norm": 0.0006009713397361338,
-      "learning_rate": 0.00019999768148889734,
-      "loss": 46.0,
-      "step": 28360
-    },
-    {
-      "epoch": 2.168396505915859,
-      "grad_norm": 0.0029473367612808943,
-      "learning_rate": 0.0001999976813253317,
-      "loss": 46.0,
-      "step": 28361
-    },
-    {
-      "epoch": 2.1684729628992487,
-      "grad_norm": 0.00048392993630841374,
-      "learning_rate": 0.0001999976811617603,
-      "loss": 46.0,
-      "step": 28362
-    },
-    {
-      "epoch": 2.1685494198826385,
-      "grad_norm": 0.0031633959151804447,
-      "learning_rate": 0.00019999768099818313,
-      "loss": 46.0,
-      "step": 28363
-    },
-    {
-      "epoch": 2.1686258768660283,
-      "grad_norm": 0.000784774252679199,
-      "learning_rate": 0.00019999768083460017,
-      "loss": 46.0,
-      "step": 28364
-    },
-    {
-      "epoch": 2.168702333849418,
-      "grad_norm": 0.00042832139297388494,
-      "learning_rate": 0.00019999768067101143,
-      "loss": 46.0,
-      "step": 28365
-    },
-    {
-      "epoch": 2.168778790832808,
-      "grad_norm": 0.0007260804995894432,
-      "learning_rate": 0.00019999768050741695,
-      "loss": 46.0,
-      "step": 28366
-    },
-    {
-      "epoch": 2.1688552478161975,
-      "grad_norm": 0.0040352558717131615,
-      "learning_rate": 0.0001999976803438167,
-      "loss": 46.0,
-      "step": 28367
-    },
-    {
-      "epoch": 2.1689317047995873,
-      "grad_norm": 0.0009794466895982623,
-      "learning_rate": 0.00019999768018021065,
-      "loss": 46.0,
-      "step": 28368
-    },
-    {
-      "epoch": 2.169008161782977,
-      "grad_norm": 0.0014764280058443546,
-      "learning_rate": 0.00019999768001659885,
-      "loss": 46.0,
-      "step": 28369
-    },
-    {
-      "epoch": 2.1690846187663664,
-      "grad_norm": 0.0005682627088390291,
-      "learning_rate": 0.00019999767985298127,
-      "loss": 46.0,
-      "step": 28370
-    },
-    {
-      "epoch": 2.169161075749756,
-      "grad_norm": 0.0026647301856428385,
-      "learning_rate": 0.00019999767968935795,
-      "loss": 46.0,
-      "step": 28371
-    },
-    {
-      "epoch": 2.169237532733146,
-      "grad_norm": 0.0009426974575035274,
-      "learning_rate": 0.00019999767952572883,
-      "loss": 46.0,
-      "step": 28372
-    },
-    {
-      "epoch": 2.1693139897165357,
-      "grad_norm": 0.00038993547786958516,
-      "learning_rate": 0.00019999767936209394,
-      "loss": 46.0,
-      "step": 28373
-    },
-    {
-      "epoch": 2.1693904466999254,
-      "grad_norm": 0.0008578879642300308,
-      "learning_rate": 0.0001999976791984533,
-      "loss": 46.0,
-      "step": 28374
-    },
-    {
-      "epoch": 2.169466903683315,
-      "grad_norm": 0.001049996237270534,
-      "learning_rate": 0.0001999976790348069,
-      "loss": 46.0,
-      "step": 28375
-    },
-    {
-      "epoch": 2.169543360666705,
-      "grad_norm": 0.0007479878258891404,
-      "learning_rate": 0.0001999976788711547,
-      "loss": 46.0,
-      "step": 28376
-    },
-    {
-      "epoch": 2.1696198176500947,
-      "grad_norm": 0.001156297861598432,
-      "learning_rate": 0.00019999767870749674,
-      "loss": 46.0,
-      "step": 28377
-    },
-    {
-      "epoch": 2.1696962746334845,
-      "grad_norm": 0.0023184586316347122,
-      "learning_rate": 0.00019999767854383304,
-      "loss": 46.0,
-      "step": 28378
-    },
-    {
-      "epoch": 2.1697727316168742,
-      "grad_norm": 0.0008291049744002521,
-      "learning_rate": 0.00019999767838016356,
-      "loss": 46.0,
-      "step": 28379
-    },
-    {
-      "epoch": 2.1698491886002635,
-      "grad_norm": 0.001846388797275722,
-      "learning_rate": 0.00019999767821648825,
-      "loss": 46.0,
-      "step": 28380
-    },
-    {
-      "epoch": 2.1699256455836533,
-      "grad_norm": 0.0034819217398762703,
-      "learning_rate": 0.00019999767805280722,
-      "loss": 46.0,
-      "step": 28381
-    },
-    {
-      "epoch": 2.170002102567043,
-      "grad_norm": 0.000577624246943742,
-      "learning_rate": 0.00019999767788912042,
-      "loss": 46.0,
-      "step": 28382
-    },
-    {
-      "epoch": 2.170078559550433,
-      "grad_norm": 0.0009315473726019263,
-      "learning_rate": 0.00019999767772542785,
-      "loss": 46.0,
-      "step": 28383
-    },
-    {
-      "epoch": 2.1701550165338226,
-      "grad_norm": 0.0022571387235075235,
-      "learning_rate": 0.0001999976775617295,
-      "loss": 46.0,
-      "step": 28384
-    },
-    {
-      "epoch": 2.1702314735172124,
-      "grad_norm": 0.0006864149472676218,
-      "learning_rate": 0.00019999767739802538,
-      "loss": 46.0,
-      "step": 28385
-    },
-    {
-      "epoch": 2.170307930500602,
-      "grad_norm": 0.000870219839271158,
-      "learning_rate": 0.00019999767723431552,
-      "loss": 46.0,
-      "step": 28386
-    },
-    {
-      "epoch": 2.170384387483992,
-      "grad_norm": 0.002720368094742298,
-      "learning_rate": 0.00019999767707059988,
-      "loss": 46.0,
-      "step": 28387
-    },
-    {
-      "epoch": 2.1704608444673816,
-      "grad_norm": 0.0020148123148828745,
-      "learning_rate": 0.00019999767690687844,
-      "loss": 46.0,
-      "step": 28388
-    },
-    {
-      "epoch": 2.1705373014507714,
-      "grad_norm": 0.0012028336059302092,
-      "learning_rate": 0.00019999767674315128,
-      "loss": 46.0,
-      "step": 28389
-    },
-    {
-      "epoch": 2.170613758434161,
-      "grad_norm": 0.0016353317769244313,
-      "learning_rate": 0.00019999767657941832,
-      "loss": 46.0,
-      "step": 28390
-    },
-    {
-      "epoch": 2.170690215417551,
-      "grad_norm": 0.0004611596232280135,
-      "learning_rate": 0.0001999976764156796,
-      "loss": 46.0,
-      "step": 28391
-    },
-    {
-      "epoch": 2.1707666724009402,
-      "grad_norm": 0.004697082564234734,
-      "learning_rate": 0.0001999976762519351,
-      "loss": 46.0,
-      "step": 28392
-    },
-    {
-      "epoch": 2.17084312938433,
-      "grad_norm": 0.002459216397255659,
-      "learning_rate": 0.00019999767608818486,
-      "loss": 46.0,
-      "step": 28393
-    },
-    {
-      "epoch": 2.1709195863677198,
-      "grad_norm": 0.0006335514481179416,
-      "learning_rate": 0.00019999767592442883,
-      "loss": 46.0,
-      "step": 28394
-    },
-    {
-      "epoch": 2.1709960433511095,
-      "grad_norm": 0.0007775943377055228,
-      "learning_rate": 0.000199997675760667,
-      "loss": 46.0,
-      "step": 28395
-    },
-    {
-      "epoch": 2.1710725003344993,
-      "grad_norm": 0.0010786314960569143,
-      "learning_rate": 0.00019999767559689943,
-      "loss": 46.0,
-      "step": 28396
-    },
-    {
-      "epoch": 2.171148957317889,
-      "grad_norm": 0.0008986023021861911,
-      "learning_rate": 0.00019999767543312609,
-      "loss": 46.0,
-      "step": 28397
-    },
-    {
-      "epoch": 2.171225414301279,
-      "grad_norm": 0.0007970093283802271,
-      "learning_rate": 0.00019999767526934697,
-      "loss": 46.0,
-      "step": 28398
-    },
-    {
-      "epoch": 2.1713018712846686,
-      "grad_norm": 0.0006517599103972316,
-      "learning_rate": 0.0001999976751055621,
-      "loss": 46.0,
-      "step": 28399
-    },
-    {
-      "epoch": 2.1713783282680583,
-      "grad_norm": 0.0005501593695953488,
-      "learning_rate": 0.00019999767494177143,
-      "loss": 46.0,
-      "step": 28400
-    },
-    {
-      "epoch": 2.171454785251448,
-      "grad_norm": 0.0023823336232453585,
-      "learning_rate": 0.00019999767477797502,
-      "loss": 46.0,
-      "step": 28401
-    },
-    {
-      "epoch": 2.1715312422348374,
-      "grad_norm": 0.0012514422414824367,
-      "learning_rate": 0.00019999767461417284,
-      "loss": 46.0,
-      "step": 28402
-    },
-    {
-      "epoch": 2.171607699218227,
-      "grad_norm": 0.0027574875857681036,
-      "learning_rate": 0.00019999767445036488,
-      "loss": 46.0,
-      "step": 28403
-    },
-    {
-      "epoch": 2.171684156201617,
-      "grad_norm": 0.0008573058294132352,
-      "learning_rate": 0.00019999767428655117,
-      "loss": 46.0,
-      "step": 28404
-    },
-    {
-      "epoch": 2.1717606131850067,
-      "grad_norm": 0.0004672650247812271,
-      "learning_rate": 0.00019999767412273167,
-      "loss": 46.0,
-      "step": 28405
-    },
-    {
-      "epoch": 2.1718370701683964,
-      "grad_norm": 0.014130059629678726,
-      "learning_rate": 0.0001999976739589064,
-      "loss": 46.0,
-      "step": 28406
-    },
-    {
-      "epoch": 2.171913527151786,
-      "grad_norm": 0.0010685843881219625,
-      "learning_rate": 0.00019999767379507536,
-      "loss": 46.0,
-      "step": 28407
-    },
-    {
-      "epoch": 2.171989984135176,
-      "grad_norm": 0.0009260507067665458,
-      "learning_rate": 0.00019999767363123856,
-      "loss": 46.0,
-      "step": 28408
-    },
-    {
-      "epoch": 2.1720664411185657,
-      "grad_norm": 0.005498750600963831,
-      "learning_rate": 0.00019999767346739596,
-      "loss": 46.0,
-      "step": 28409
-    },
-    {
-      "epoch": 2.1721428981019555,
-      "grad_norm": 0.0010454320581629872,
-      "learning_rate": 0.00019999767330354764,
-      "loss": 46.0,
-      "step": 28410
-    },
-    {
-      "epoch": 2.1722193550853452,
-      "grad_norm": 0.0003078437002841383,
-      "learning_rate": 0.00019999767313969352,
-      "loss": 46.0,
-      "step": 28411
-    },
-    {
-      "epoch": 2.172295812068735,
-      "grad_norm": 0.0008880148525349796,
-      "learning_rate": 0.00019999767297583363,
-      "loss": 46.0,
-      "step": 28412
-    },
-    {
-      "epoch": 2.1723722690521248,
-      "grad_norm": 0.0009869198547676206,
-      "learning_rate": 0.000199997672811968,
-      "loss": 46.0,
-      "step": 28413
-    },
-    {
-      "epoch": 2.172448726035514,
-      "grad_norm": 0.0009513136465102434,
-      "learning_rate": 0.00019999767264809658,
-      "loss": 46.0,
-      "step": 28414
-    },
-    {
-      "epoch": 2.172525183018904,
-      "grad_norm": 0.0027022380381822586,
-      "learning_rate": 0.00019999767248421937,
-      "loss": 46.0,
-      "step": 28415
-    },
-    {
-      "epoch": 2.1726016400022936,
-      "grad_norm": 0.0009560069884173572,
-      "learning_rate": 0.0001999976723203364,
-      "loss": 46.0,
-      "step": 28416
-    },
-    {
-      "epoch": 2.1726780969856834,
-      "grad_norm": 0.002306233858689666,
-      "learning_rate": 0.00019999767215644768,
-      "loss": 46.0,
-      "step": 28417
-    },
-    {
-      "epoch": 2.172754553969073,
-      "grad_norm": 0.0020294920541346073,
-      "learning_rate": 0.0001999976719925532,
-      "loss": 46.0,
-      "step": 28418
-    },
-    {
-      "epoch": 2.172831010952463,
-      "grad_norm": 0.015760544687509537,
-      "learning_rate": 0.00019999767182865292,
-      "loss": 46.0,
-      "step": 28419
-    },
-    {
-      "epoch": 2.1729074679358527,
-      "grad_norm": 0.0006957454606890678,
-      "learning_rate": 0.0001999976716647469,
-      "loss": 46.0,
-      "step": 28420
-    },
-    {
-      "epoch": 2.1729839249192424,
-      "grad_norm": 0.00290909456089139,
-      "learning_rate": 0.0001999976715008351,
-      "loss": 46.0,
-      "step": 28421
-    },
-    {
-      "epoch": 2.173060381902632,
-      "grad_norm": 0.001131995115429163,
-      "learning_rate": 0.00019999767133691752,
-      "loss": 46.0,
-      "step": 28422
-    },
-    {
-      "epoch": 2.173136838886022,
-      "grad_norm": 0.0005287793464958668,
-      "learning_rate": 0.00019999767117299418,
-      "loss": 46.0,
-      "step": 28423
-    },
-    {
-      "epoch": 2.1732132958694113,
-      "grad_norm": 0.0012256193440407515,
-      "learning_rate": 0.00019999767100906506,
-      "loss": 46.0,
-      "step": 28424
-    },
-    {
-      "epoch": 2.173289752852801,
-      "grad_norm": 0.000694221118465066,
-      "learning_rate": 0.00019999767084513017,
-      "loss": 46.0,
-      "step": 28425
-    },
-    {
-      "epoch": 2.1733662098361908,
-      "grad_norm": 0.000910363916773349,
-      "learning_rate": 0.0001999976706811895,
-      "loss": 46.0,
-      "step": 28426
-    },
-    {
-      "epoch": 2.1734426668195805,
-      "grad_norm": 0.004417065065354109,
-      "learning_rate": 0.0001999976705172431,
-      "loss": 46.0,
-      "step": 28427
-    },
-    {
-      "epoch": 2.1735191238029703,
-      "grad_norm": 0.0019813564140349627,
-      "learning_rate": 0.0001999976703532909,
-      "loss": 46.0,
-      "step": 28428
-    },
-    {
-      "epoch": 2.17359558078636,
-      "grad_norm": 0.0007425614167004824,
-      "learning_rate": 0.00019999767018933295,
-      "loss": 46.0,
-      "step": 28429
-    },
-    {
-      "epoch": 2.17367203776975,
-      "grad_norm": 0.0013123922981321812,
-      "learning_rate": 0.00019999767002536922,
-      "loss": 46.0,
-      "step": 28430
-    },
-    {
-      "epoch": 2.1737484947531396,
-      "grad_norm": 0.0014108201721683145,
-      "learning_rate": 0.0001999976698613997,
-      "loss": 46.0,
-      "step": 28431
-    },
-    {
-      "epoch": 2.1738249517365293,
-      "grad_norm": 0.0011263181222602725,
-      "learning_rate": 0.00019999766969742446,
-      "loss": 46.0,
-      "step": 28432
-    },
-    {
-      "epoch": 2.173901408719919,
-      "grad_norm": 0.0019500887719914317,
-      "learning_rate": 0.00019999766953344343,
-      "loss": 46.0,
-      "step": 28433
-    },
-    {
-      "epoch": 2.173977865703309,
-      "grad_norm": 0.0007986238924786448,
-      "learning_rate": 0.0001999976693694566,
-      "loss": 46.0,
-      "step": 28434
-    },
-    {
-      "epoch": 2.1740543226866986,
-      "grad_norm": 0.003296899376437068,
-      "learning_rate": 0.00019999766920546404,
-      "loss": 46.0,
-      "step": 28435
-    },
-    {
-      "epoch": 2.174130779670088,
-      "grad_norm": 0.0010163268307223916,
-      "learning_rate": 0.0001999976690414657,
-      "loss": 46.0,
-      "step": 28436
-    },
-    {
-      "epoch": 2.1742072366534777,
-      "grad_norm": 0.000838159816339612,
-      "learning_rate": 0.00019999766887746157,
-      "loss": 46.0,
-      "step": 28437
-    },
-    {
-      "epoch": 2.1742836936368675,
-      "grad_norm": 0.0005727846873924136,
-      "learning_rate": 0.00019999766871345168,
-      "loss": 46.0,
-      "step": 28438
-    },
-    {
-      "epoch": 2.1743601506202572,
-      "grad_norm": 0.0019931441638618708,
-      "learning_rate": 0.00019999766854943602,
-      "loss": 46.0,
-      "step": 28439
-    },
-    {
-      "epoch": 2.174436607603647,
-      "grad_norm": 0.0007575023337267339,
-      "learning_rate": 0.0001999976683854146,
-      "loss": 46.0,
-      "step": 28440
-    },
-    {
-      "epoch": 2.1745130645870367,
-      "grad_norm": 0.0005464000278152525,
-      "learning_rate": 0.00019999766822138742,
-      "loss": 46.0,
-      "step": 28441
-    },
-    {
-      "epoch": 2.1745895215704265,
-      "grad_norm": 0.0010065704118460417,
-      "learning_rate": 0.00019999766805735446,
-      "loss": 46.0,
-      "step": 28442
-    },
-    {
-      "epoch": 2.1746659785538163,
-      "grad_norm": 0.0011554675875231624,
-      "learning_rate": 0.00019999766789331573,
-      "loss": 46.0,
-      "step": 28443
-    },
-    {
-      "epoch": 2.174742435537206,
-      "grad_norm": 0.0018307155696675181,
-      "learning_rate": 0.00019999766772927125,
-      "loss": 46.0,
-      "step": 28444
-    },
-    {
-      "epoch": 2.174818892520596,
-      "grad_norm": 0.0019894202705472708,
-      "learning_rate": 0.00019999766756522095,
-      "loss": 46.0,
-      "step": 28445
-    },
-    {
-      "epoch": 2.174895349503985,
-      "grad_norm": 0.0007236563833430409,
-      "learning_rate": 0.00019999766740116492,
-      "loss": 46.0,
-      "step": 28446
-    },
-    {
-      "epoch": 2.174971806487375,
-      "grad_norm": 0.0011490915203467011,
-      "learning_rate": 0.00019999766723710312,
-      "loss": 46.0,
-      "step": 28447
-    },
-    {
-      "epoch": 2.1750482634707646,
-      "grad_norm": 0.0015396373346447945,
-      "learning_rate": 0.00019999766707303555,
-      "loss": 46.0,
-      "step": 28448
-    },
-    {
-      "epoch": 2.1751247204541544,
-      "grad_norm": 0.0019520844798535109,
-      "learning_rate": 0.0001999976669089622,
-      "loss": 46.0,
-      "step": 28449
-    },
-    {
-      "epoch": 2.175201177437544,
-      "grad_norm": 0.0009827141184359789,
-      "learning_rate": 0.00019999766674488306,
-      "loss": 46.0,
-      "step": 28450
-    },
-    {
-      "epoch": 2.175277634420934,
-      "grad_norm": 0.001469287439249456,
-      "learning_rate": 0.0001999976665807982,
-      "loss": 46.0,
-      "step": 28451
-    },
-    {
-      "epoch": 2.1753540914043237,
-      "grad_norm": 0.0007992509636096656,
-      "learning_rate": 0.00019999766641670753,
-      "loss": 46.0,
-      "step": 28452
-    },
-    {
-      "epoch": 2.1754305483877134,
-      "grad_norm": 0.0008464643033221364,
-      "learning_rate": 0.00019999766625261112,
-      "loss": 46.0,
-      "step": 28453
-    },
-    {
-      "epoch": 2.175507005371103,
-      "grad_norm": 0.0010568229481577873,
-      "learning_rate": 0.0001999976660885089,
-      "loss": 46.0,
-      "step": 28454
-    },
-    {
-      "epoch": 2.175583462354493,
-      "grad_norm": 0.001394504914060235,
-      "learning_rate": 0.00019999766592440098,
-      "loss": 46.0,
-      "step": 28455
-    },
-    {
-      "epoch": 2.1756599193378827,
-      "grad_norm": 0.0011778074549511075,
-      "learning_rate": 0.00019999766576028725,
-      "loss": 46.0,
-      "step": 28456
-    },
-    {
-      "epoch": 2.175736376321272,
-      "grad_norm": 0.001161605236120522,
-      "learning_rate": 0.00019999766559616772,
-      "loss": 46.0,
-      "step": 28457
-    },
-    {
-      "epoch": 2.175812833304662,
-      "grad_norm": 0.0010076920734718442,
-      "learning_rate": 0.00019999766543204247,
-      "loss": 46.0,
-      "step": 28458
-    },
-    {
-      "epoch": 2.1758892902880516,
-      "grad_norm": 0.0009201422799378633,
-      "learning_rate": 0.00019999766526791141,
-      "loss": 46.0,
-      "step": 28459
-    },
-    {
-      "epoch": 2.1759657472714413,
-      "grad_norm": 0.0005596308619715273,
-      "learning_rate": 0.00019999766510377462,
-      "loss": 46.0,
-      "step": 28460
-    },
-    {
-      "epoch": 2.176042204254831,
-      "grad_norm": 0.0011208391515538096,
-      "learning_rate": 0.00019999766493963205,
-      "loss": 46.0,
-      "step": 28461
-    },
-    {
-      "epoch": 2.176118661238221,
-      "grad_norm": 0.0027786337304860353,
-      "learning_rate": 0.00019999766477548367,
-      "loss": 46.0,
-      "step": 28462
-    },
-    {
-      "epoch": 2.1761951182216106,
-      "grad_norm": 0.0011292733252048492,
-      "learning_rate": 0.00019999766461132956,
-      "loss": 46.0,
-      "step": 28463
-    },
-    {
-      "epoch": 2.1762715752050004,
-      "grad_norm": 0.0004198331735096872,
-      "learning_rate": 0.00019999766444716966,
-      "loss": 46.0,
-      "step": 28464
-    },
-    {
-      "epoch": 2.17634803218839,
-      "grad_norm": 0.0017612830270081758,
-      "learning_rate": 0.00019999766428300403,
-      "loss": 46.0,
-      "step": 28465
-    },
-    {
-      "epoch": 2.17642448917178,
-      "grad_norm": 0.00228533404879272,
-      "learning_rate": 0.00019999766411883262,
-      "loss": 46.0,
-      "step": 28466
-    },
-    {
-      "epoch": 2.1765009461551696,
-      "grad_norm": 0.005738257430493832,
-      "learning_rate": 0.00019999766395465543,
-      "loss": 46.0,
-      "step": 28467
-    },
-    {
-      "epoch": 2.176577403138559,
-      "grad_norm": 0.0017074800562113523,
-      "learning_rate": 0.00019999766379047245,
-      "loss": 46.0,
-      "step": 28468
-    },
-    {
-      "epoch": 2.1766538601219487,
-      "grad_norm": 0.0009029096108861268,
-      "learning_rate": 0.00019999766362628372,
-      "loss": 46.0,
-      "step": 28469
-    },
-    {
-      "epoch": 2.1767303171053385,
-      "grad_norm": 0.006646163761615753,
-      "learning_rate": 0.0001999976634620892,
-      "loss": 46.0,
-      "step": 28470
-    },
-    {
-      "epoch": 2.1768067740887282,
-      "grad_norm": 0.0008539999253116548,
-      "learning_rate": 0.00019999766329788893,
-      "loss": 46.0,
-      "step": 28471
-    },
-    {
-      "epoch": 2.176883231072118,
-      "grad_norm": 0.0013579073129221797,
-      "learning_rate": 0.00019999766313368288,
-      "loss": 46.0,
-      "step": 28472
-    },
-    {
-      "epoch": 2.1769596880555078,
-      "grad_norm": 0.0014794875169172883,
-      "learning_rate": 0.00019999766296947109,
-      "loss": 46.0,
-      "step": 28473
-    },
-    {
-      "epoch": 2.1770361450388975,
-      "grad_norm": 0.006017064675688744,
-      "learning_rate": 0.00019999766280525352,
-      "loss": 46.0,
-      "step": 28474
-    },
-    {
-      "epoch": 2.1771126020222873,
-      "grad_norm": 0.0022261340636759996,
-      "learning_rate": 0.00019999766264103017,
-      "loss": 46.0,
-      "step": 28475
-    },
-    {
-      "epoch": 2.177189059005677,
-      "grad_norm": 0.00023276920546777546,
-      "learning_rate": 0.00019999766247680105,
-      "loss": 46.0,
-      "step": 28476
-    },
-    {
-      "epoch": 2.177265515989067,
-      "grad_norm": 0.0015872431686148047,
-      "learning_rate": 0.00019999766231256616,
-      "loss": 46.0,
-      "step": 28477
-    },
-    {
-      "epoch": 2.1773419729724566,
-      "grad_norm": 0.0007817122968845069,
-      "learning_rate": 0.0001999976621483255,
-      "loss": 46.0,
-      "step": 28478
-    },
-    {
-      "epoch": 2.177418429955846,
-      "grad_norm": 0.00046454579569399357,
-      "learning_rate": 0.0001999976619840791,
-      "loss": 46.0,
-      "step": 28479
-    },
-    {
-      "epoch": 2.1774948869392357,
-      "grad_norm": 0.0019957246258854866,
-      "learning_rate": 0.00019999766181982688,
-      "loss": 46.0,
-      "step": 28480
-    },
-    {
-      "epoch": 2.1775713439226254,
-      "grad_norm": 0.0014977352693676949,
-      "learning_rate": 0.0001999976616555689,
-      "loss": 46.0,
-      "step": 28481
-    },
-    {
-      "epoch": 2.177647800906015,
-      "grad_norm": 0.000837024359498173,
-      "learning_rate": 0.0001999976614913052,
-      "loss": 46.0,
-      "step": 28482
-    },
-    {
-      "epoch": 2.177724257889405,
-      "grad_norm": 0.0029853328596800566,
-      "learning_rate": 0.00019999766132703568,
-      "loss": 46.0,
-      "step": 28483
-    },
-    {
-      "epoch": 2.1778007148727947,
-      "grad_norm": 0.001866753795184195,
-      "learning_rate": 0.0001999976611627604,
-      "loss": 46.0,
-      "step": 28484
-    },
-    {
-      "epoch": 2.1778771718561845,
-      "grad_norm": 0.0015747282886877656,
-      "learning_rate": 0.00019999766099847936,
-      "loss": 46.0,
-      "step": 28485
-    },
-    {
-      "epoch": 2.177953628839574,
-      "grad_norm": 0.0005321996286511421,
-      "learning_rate": 0.00019999766083419256,
-      "loss": 46.0,
-      "step": 28486
-    },
-    {
-      "epoch": 2.178030085822964,
-      "grad_norm": 0.0024297041818499565,
-      "learning_rate": 0.0001999976606699,
-      "loss": 46.0,
-      "step": 28487
-    },
-    {
-      "epoch": 2.1781065428063537,
-      "grad_norm": 0.0009199777850881219,
-      "learning_rate": 0.00019999766050560162,
-      "loss": 46.0,
-      "step": 28488
-    },
-    {
-      "epoch": 2.1781829997897435,
-      "grad_norm": 0.0015354490606114268,
-      "learning_rate": 0.0001999976603412975,
-      "loss": 46.0,
-      "step": 28489
-    },
-    {
-      "epoch": 2.178259456773133,
-      "grad_norm": 0.0014288178645074368,
-      "learning_rate": 0.0001999976601769876,
-      "loss": 46.0,
-      "step": 28490
-    },
-    {
-      "epoch": 2.1783359137565226,
-      "grad_norm": 0.0006210981518961489,
-      "learning_rate": 0.00019999766001267197,
-      "loss": 46.0,
-      "step": 28491
-    },
-    {
-      "epoch": 2.1784123707399123,
-      "grad_norm": 0.0014171134680509567,
-      "learning_rate": 0.00019999765984835054,
-      "loss": 46.0,
-      "step": 28492
-    },
-    {
-      "epoch": 2.178488827723302,
-      "grad_norm": 0.0013052221620455384,
-      "learning_rate": 0.00019999765968402333,
-      "loss": 46.0,
-      "step": 28493
-    },
-    {
-      "epoch": 2.178565284706692,
-      "grad_norm": 0.001749410410411656,
-      "learning_rate": 0.00019999765951969037,
-      "loss": 46.0,
-      "step": 28494
-    },
-    {
-      "epoch": 2.1786417416900816,
-      "grad_norm": 0.0008194524561986327,
-      "learning_rate": 0.00019999765935535164,
-      "loss": 46.0,
-      "step": 28495
-    },
-    {
-      "epoch": 2.1787181986734714,
-      "grad_norm": 0.0043754540383815765,
-      "learning_rate": 0.00019999765919100713,
-      "loss": 46.0,
-      "step": 28496
-    },
-    {
-      "epoch": 2.178794655656861,
-      "grad_norm": 0.0006392080103978515,
-      "learning_rate": 0.00019999765902665686,
-      "loss": 46.0,
-      "step": 28497
-    },
-    {
-      "epoch": 2.178871112640251,
-      "grad_norm": 0.001968333264812827,
-      "learning_rate": 0.0001999976588623008,
-      "loss": 46.0,
-      "step": 28498
-    },
-    {
-      "epoch": 2.1789475696236407,
-      "grad_norm": 0.0005172672681510448,
-      "learning_rate": 0.000199997658697939,
-      "loss": 46.0,
-      "step": 28499
-    },
-    {
-      "epoch": 2.1790240266070304,
-      "grad_norm": 0.0004186072910670191,
-      "learning_rate": 0.0001999976585335714,
-      "loss": 46.0,
-      "step": 28500
-    },
-    {
-      "epoch": 2.1791004835904197,
-      "grad_norm": 0.001641149865463376,
-      "learning_rate": 0.00019999765836919807,
-      "loss": 46.0,
-      "step": 28501
-    },
-    {
-      "epoch": 2.1791769405738095,
-      "grad_norm": 0.0024131264071911573,
-      "learning_rate": 0.00019999765820481895,
-      "loss": 46.0,
-      "step": 28502
-    },
-    {
-      "epoch": 2.1792533975571993,
-      "grad_norm": 0.0015256662154570222,
-      "learning_rate": 0.00019999765804043404,
-      "loss": 46.0,
-      "step": 28503
-    },
-    {
-      "epoch": 2.179329854540589,
-      "grad_norm": 0.0029041327070444822,
-      "learning_rate": 0.0001999976578760434,
-      "loss": 46.0,
-      "step": 28504
-    },
-    {
-      "epoch": 2.179406311523979,
-      "grad_norm": 0.0011511488119140267,
-      "learning_rate": 0.00019999765771164696,
-      "loss": 46.0,
-      "step": 28505
-    },
-    {
-      "epoch": 2.1794827685073686,
-      "grad_norm": 0.0019296801183372736,
-      "learning_rate": 0.00019999765754724478,
-      "loss": 46.0,
-      "step": 28506
-    },
-    {
-      "epoch": 2.1795592254907583,
-      "grad_norm": 0.001309265848249197,
-      "learning_rate": 0.00019999765738283682,
-      "loss": 46.0,
-      "step": 28507
-    },
-    {
-      "epoch": 2.179635682474148,
-      "grad_norm": 0.0017570970812812448,
-      "learning_rate": 0.00019999765721842306,
-      "loss": 46.0,
-      "step": 28508
-    },
-    {
-      "epoch": 2.179712139457538,
-      "grad_norm": 0.0012124072527512908,
-      "learning_rate": 0.00019999765705400356,
-      "loss": 46.0,
-      "step": 28509
-    },
-    {
-      "epoch": 2.1797885964409276,
-      "grad_norm": 0.0014030297752469778,
-      "learning_rate": 0.00019999765688957828,
-      "loss": 46.0,
-      "step": 28510
-    },
-    {
-      "epoch": 2.179865053424317,
-      "grad_norm": 0.0018287461716681719,
-      "learning_rate": 0.00019999765672514726,
-      "loss": 46.0,
-      "step": 28511
-    },
-    {
-      "epoch": 2.1799415104077067,
-      "grad_norm": 0.0009697952191345394,
-      "learning_rate": 0.00019999765656071044,
-      "loss": 46.0,
-      "step": 28512
-    },
-    {
-      "epoch": 2.1800179673910964,
-      "grad_norm": 0.0011770287528634071,
-      "learning_rate": 0.00019999765639626787,
-      "loss": 46.0,
-      "step": 28513
-    },
-    {
-      "epoch": 2.180094424374486,
-      "grad_norm": 0.0009654753957875073,
-      "learning_rate": 0.0001999976562318195,
-      "loss": 46.0,
-      "step": 28514
-    },
-    {
-      "epoch": 2.180170881357876,
-      "grad_norm": 0.00402452889829874,
-      "learning_rate": 0.00019999765606736538,
-      "loss": 46.0,
-      "step": 28515
-    },
-    {
-      "epoch": 2.1802473383412657,
-      "grad_norm": 0.0029834695160388947,
-      "learning_rate": 0.0001999976559029055,
-      "loss": 46.0,
-      "step": 28516
-    },
-    {
-      "epoch": 2.1803237953246555,
-      "grad_norm": 0.003193502314388752,
-      "learning_rate": 0.00019999765573843983,
-      "loss": 46.0,
-      "step": 28517
-    },
-    {
-      "epoch": 2.1804002523080452,
-      "grad_norm": 0.001085283001884818,
-      "learning_rate": 0.0001999976555739684,
-      "loss": 46.0,
-      "step": 28518
-    },
-    {
-      "epoch": 2.180476709291435,
-      "grad_norm": 0.0005285979132167995,
-      "learning_rate": 0.0001999976554094912,
-      "loss": 46.0,
-      "step": 28519
-    },
-    {
-      "epoch": 2.1805531662748248,
-      "grad_norm": 0.005591289605945349,
-      "learning_rate": 0.00019999765524500825,
-      "loss": 46.0,
-      "step": 28520
-    },
-    {
-      "epoch": 2.1806296232582145,
-      "grad_norm": 0.0013330476358532906,
-      "learning_rate": 0.0001999976550805195,
-      "loss": 46.0,
-      "step": 28521
-    },
-    {
-      "epoch": 2.1807060802416043,
-      "grad_norm": 0.0007984755211509764,
-      "learning_rate": 0.000199997654916025,
-      "loss": 46.0,
-      "step": 28522
-    },
-    {
-      "epoch": 2.1807825372249936,
-      "grad_norm": 0.0009099752642214298,
-      "learning_rate": 0.00019999765475152472,
-      "loss": 46.0,
-      "step": 28523
-    },
-    {
-      "epoch": 2.1808589942083834,
-      "grad_norm": 0.0015721693634986877,
-      "learning_rate": 0.00019999765458701867,
-      "loss": 46.0,
-      "step": 28524
-    },
-    {
-      "epoch": 2.180935451191773,
-      "grad_norm": 0.0005699247121810913,
-      "learning_rate": 0.00019999765442250684,
-      "loss": 46.0,
-      "step": 28525
-    },
-    {
-      "epoch": 2.181011908175163,
-      "grad_norm": 0.0009928439976647496,
-      "learning_rate": 0.00019999765425798927,
-      "loss": 46.0,
-      "step": 28526
-    },
-    {
-      "epoch": 2.1810883651585526,
-      "grad_norm": 0.0010869466932490468,
-      "learning_rate": 0.0001999976540934659,
-      "loss": 46.0,
-      "step": 28527
-    },
-    {
-      "epoch": 2.1811648221419424,
-      "grad_norm": 0.0005256158183328807,
-      "learning_rate": 0.0001999976539289368,
-      "loss": 46.0,
-      "step": 28528
-    },
-    {
-      "epoch": 2.181241279125332,
-      "grad_norm": 0.0007434640428982675,
-      "learning_rate": 0.0001999976537644019,
-      "loss": 46.0,
-      "step": 28529
-    },
-    {
-      "epoch": 2.181317736108722,
-      "grad_norm": 0.0010680626146495342,
-      "learning_rate": 0.00019999765359986123,
-      "loss": 46.0,
-      "step": 28530
-    },
-    {
-      "epoch": 2.1813941930921117,
-      "grad_norm": 0.0009314455091953278,
-      "learning_rate": 0.00019999765343531483,
-      "loss": 46.0,
-      "step": 28531
-    },
-    {
-      "epoch": 2.1814706500755014,
-      "grad_norm": 0.005388634745031595,
-      "learning_rate": 0.00019999765327076262,
-      "loss": 46.0,
-      "step": 28532
-    },
-    {
-      "epoch": 2.1815471070588908,
-      "grad_norm": 0.0010014433646574616,
-      "learning_rate": 0.00019999765310620463,
-      "loss": 46.0,
-      "step": 28533
-    },
-    {
-      "epoch": 2.1816235640422805,
-      "grad_norm": 0.0009149159304797649,
-      "learning_rate": 0.0001999976529416409,
-      "loss": 46.0,
-      "step": 28534
-    },
-    {
-      "epoch": 2.1817000210256703,
-      "grad_norm": 0.000616028846707195,
-      "learning_rate": 0.0001999976527770714,
-      "loss": 46.0,
-      "step": 28535
-    },
-    {
-      "epoch": 2.18177647800906,
-      "grad_norm": 0.0008973000221885741,
-      "learning_rate": 0.00019999765261249612,
-      "loss": 46.0,
-      "step": 28536
-    },
-    {
-      "epoch": 2.18185293499245,
-      "grad_norm": 0.001988670090213418,
-      "learning_rate": 0.00019999765244791507,
-      "loss": 46.0,
-      "step": 28537
-    },
-    {
-      "epoch": 2.1819293919758396,
-      "grad_norm": 0.0006010684883221984,
-      "learning_rate": 0.00019999765228332825,
-      "loss": 46.0,
-      "step": 28538
-    },
-    {
-      "epoch": 2.1820058489592293,
-      "grad_norm": 0.0014233929105103016,
-      "learning_rate": 0.00019999765211873566,
-      "loss": 46.0,
-      "step": 28539
-    },
-    {
-      "epoch": 2.182082305942619,
-      "grad_norm": 0.001085110241547227,
-      "learning_rate": 0.00019999765195413734,
-      "loss": 46.0,
-      "step": 28540
-    },
-    {
-      "epoch": 2.182158762926009,
-      "grad_norm": 0.0005333789740689099,
-      "learning_rate": 0.00019999765178953317,
-      "loss": 46.0,
-      "step": 28541
-    },
-    {
-      "epoch": 2.1822352199093986,
-      "grad_norm": 0.0011887053260579705,
-      "learning_rate": 0.0001999976516249233,
-      "loss": 46.0,
-      "step": 28542
-    },
-    {
-      "epoch": 2.1823116768927884,
-      "grad_norm": 0.0009352117776870728,
-      "learning_rate": 0.00019999765146030764,
-      "loss": 46.0,
-      "step": 28543
-    },
-    {
-      "epoch": 2.182388133876178,
-      "grad_norm": 0.0012499524746090174,
-      "learning_rate": 0.0001999976512956862,
-      "loss": 46.0,
-      "step": 28544
-    },
-    {
-      "epoch": 2.1824645908595675,
-      "grad_norm": 0.0006399460835382342,
-      "learning_rate": 0.00019999765113105903,
-      "loss": 46.0,
-      "step": 28545
-    },
-    {
-      "epoch": 2.182541047842957,
-      "grad_norm": 0.0006755090435035527,
-      "learning_rate": 0.00019999765096642604,
-      "loss": 46.0,
-      "step": 28546
-    },
-    {
-      "epoch": 2.182617504826347,
-      "grad_norm": 0.0009542009211145341,
-      "learning_rate": 0.0001999976508017873,
-      "loss": 46.0,
-      "step": 28547
-    },
-    {
-      "epoch": 2.1826939618097367,
-      "grad_norm": 0.0008284849463962018,
-      "learning_rate": 0.00019999765063714279,
-      "loss": 46.0,
-      "step": 28548
-    },
-    {
-      "epoch": 2.1827704187931265,
-      "grad_norm": 0.0009998140158131719,
-      "learning_rate": 0.0001999976504724925,
-      "loss": 46.0,
-      "step": 28549
-    },
-    {
-      "epoch": 2.1828468757765163,
-      "grad_norm": 0.0027649146504700184,
-      "learning_rate": 0.00019999765030783646,
-      "loss": 46.0,
-      "step": 28550
-    },
-    {
-      "epoch": 2.182923332759906,
-      "grad_norm": 0.003018118441104889,
-      "learning_rate": 0.00019999765014317464,
-      "loss": 46.0,
-      "step": 28551
-    },
-    {
-      "epoch": 2.182999789743296,
-      "grad_norm": 0.0004947337438352406,
-      "learning_rate": 0.00019999764997850704,
-      "loss": 46.0,
-      "step": 28552
-    },
-    {
-      "epoch": 2.1830762467266855,
-      "grad_norm": 0.002311462303623557,
-      "learning_rate": 0.0001999976498138337,
-      "loss": 46.0,
-      "step": 28553
-    },
-    {
-      "epoch": 2.1831527037100753,
-      "grad_norm": 0.0008055961225181818,
-      "learning_rate": 0.00019999764964915458,
-      "loss": 46.0,
-      "step": 28554
-    },
-    {
-      "epoch": 2.1832291606934646,
-      "grad_norm": 0.0026294426061213017,
-      "learning_rate": 0.0001999976494844697,
-      "loss": 46.0,
-      "step": 28555
-    },
-    {
-      "epoch": 2.1833056176768544,
-      "grad_norm": 0.0018904200987890363,
-      "learning_rate": 0.00019999764931977903,
-      "loss": 46.0,
-      "step": 28556
-    },
-    {
-      "epoch": 2.183382074660244,
-      "grad_norm": 0.0013703533913940191,
-      "learning_rate": 0.0001999976491550826,
-      "loss": 46.0,
-      "step": 28557
-    },
-    {
-      "epoch": 2.183458531643634,
-      "grad_norm": 0.0007844749488867819,
-      "learning_rate": 0.0001999976489903804,
-      "loss": 46.0,
-      "step": 28558
-    },
-    {
-      "epoch": 2.1835349886270237,
-      "grad_norm": 0.0003083696065004915,
-      "learning_rate": 0.00019999764882567243,
-      "loss": 46.0,
-      "step": 28559
-    },
-    {
-      "epoch": 2.1836114456104134,
-      "grad_norm": 0.0007838538731448352,
-      "learning_rate": 0.00019999764866095868,
-      "loss": 46.0,
-      "step": 28560
-    },
-    {
-      "epoch": 2.183687902593803,
-      "grad_norm": 0.0008964691660366952,
-      "learning_rate": 0.00019999764849623917,
-      "loss": 46.0,
-      "step": 28561
-    },
-    {
-      "epoch": 2.183764359577193,
-      "grad_norm": 0.0011745627271011472,
-      "learning_rate": 0.0001999976483315139,
-      "loss": 46.0,
-      "step": 28562
-    },
-    {
-      "epoch": 2.1838408165605827,
-      "grad_norm": 0.0009995668660849333,
-      "learning_rate": 0.00019999764816678285,
-      "loss": 46.0,
-      "step": 28563
-    },
-    {
-      "epoch": 2.1839172735439725,
-      "grad_norm": 0.0007942367228679359,
-      "learning_rate": 0.000199997648002046,
-      "loss": 46.0,
-      "step": 28564
-    },
-    {
-      "epoch": 2.1839937305273622,
-      "grad_norm": 0.0010753748938441277,
-      "learning_rate": 0.00019999764783730343,
-      "loss": 46.0,
-      "step": 28565
-    },
-    {
-      "epoch": 2.184070187510752,
-      "grad_norm": 0.00035657055559568107,
-      "learning_rate": 0.0001999976476725551,
-      "loss": 46.0,
-      "step": 28566
-    },
-    {
-      "epoch": 2.1841466444941413,
-      "grad_norm": 0.0006955236895009875,
-      "learning_rate": 0.00019999764750780095,
-      "loss": 46.0,
-      "step": 28567
-    },
-    {
-      "epoch": 2.184223101477531,
-      "grad_norm": 0.0009154097642749548,
-      "learning_rate": 0.00019999764734304106,
-      "loss": 46.0,
-      "step": 28568
-    },
-    {
-      "epoch": 2.184299558460921,
-      "grad_norm": 0.0010483013466000557,
-      "learning_rate": 0.0001999976471782754,
-      "loss": 46.0,
-      "step": 28569
-    },
-    {
-      "epoch": 2.1843760154443106,
-      "grad_norm": 0.0007674329681321979,
-      "learning_rate": 0.00019999764701350396,
-      "loss": 46.0,
-      "step": 28570
-    },
-    {
-      "epoch": 2.1844524724277004,
-      "grad_norm": 0.0013115676119923592,
-      "learning_rate": 0.00019999764684872675,
-      "loss": 46.0,
-      "step": 28571
-    },
-    {
-      "epoch": 2.18452892941109,
-      "grad_norm": 0.0006922197644598782,
-      "learning_rate": 0.00019999764668394377,
-      "loss": 46.0,
-      "step": 28572
-    },
-    {
-      "epoch": 2.18460538639448,
-      "grad_norm": 0.0006658604834228754,
-      "learning_rate": 0.00019999764651915504,
-      "loss": 46.0,
-      "step": 28573
-    },
-    {
-      "epoch": 2.1846818433778696,
-      "grad_norm": 0.0010801381431519985,
-      "learning_rate": 0.0001999976463543605,
-      "loss": 46.0,
-      "step": 28574
-    },
-    {
-      "epoch": 2.1847583003612594,
-      "grad_norm": 0.0011674728011712432,
-      "learning_rate": 0.00019999764618956024,
-      "loss": 46.0,
-      "step": 28575
-    },
-    {
-      "epoch": 2.184834757344649,
-      "grad_norm": 0.0015218290500342846,
-      "learning_rate": 0.0001999976460247542,
-      "loss": 46.0,
-      "step": 28576
-    },
-    {
-      "epoch": 2.1849112143280385,
-      "grad_norm": 0.00114711734931916,
-      "learning_rate": 0.00019999764585994237,
-      "loss": 46.0,
-      "step": 28577
-    },
-    {
-      "epoch": 2.1849876713114282,
-      "grad_norm": 0.0013971454463899136,
-      "learning_rate": 0.00019999764569512477,
-      "loss": 46.0,
-      "step": 28578
-    },
-    {
-      "epoch": 2.185064128294818,
-      "grad_norm": 0.0008302445057779551,
-      "learning_rate": 0.00019999764553030143,
-      "loss": 46.0,
-      "step": 28579
-    },
-    {
-      "epoch": 2.1851405852782078,
-      "grad_norm": 0.004947067238390446,
-      "learning_rate": 0.0001999976453654723,
-      "loss": 46.0,
-      "step": 28580
-    },
-    {
-      "epoch": 2.1852170422615975,
-      "grad_norm": 0.0009899416472762823,
-      "learning_rate": 0.0001999976452006374,
-      "loss": 46.0,
-      "step": 28581
-    },
-    {
-      "epoch": 2.1852934992449873,
-      "grad_norm": 0.0005146818002685905,
-      "learning_rate": 0.00019999764503579674,
-      "loss": 46.0,
-      "step": 28582
-    },
-    {
-      "epoch": 2.185369956228377,
-      "grad_norm": 0.003303535282611847,
-      "learning_rate": 0.00019999764487095033,
-      "loss": 46.0,
-      "step": 28583
-    },
-    {
-      "epoch": 2.185446413211767,
-      "grad_norm": 0.0007727764314040542,
-      "learning_rate": 0.0001999976447060981,
-      "loss": 46.0,
-      "step": 28584
-    },
-    {
-      "epoch": 2.1855228701951566,
-      "grad_norm": 0.0003680314985103905,
-      "learning_rate": 0.0001999976445412401,
-      "loss": 46.0,
-      "step": 28585
-    },
-    {
-      "epoch": 2.1855993271785463,
-      "grad_norm": 0.0013785776682198048,
-      "learning_rate": 0.00019999764437637638,
-      "loss": 46.0,
-      "step": 28586
-    },
-    {
-      "epoch": 2.185675784161936,
-      "grad_norm": 0.0007073109154589474,
-      "learning_rate": 0.00019999764421150688,
-      "loss": 46.0,
-      "step": 28587
-    },
-    {
-      "epoch": 2.185752241145326,
-      "grad_norm": 0.0008490776526741683,
-      "learning_rate": 0.00019999764404663158,
-      "loss": 46.0,
-      "step": 28588
-    },
-    {
-      "epoch": 2.185828698128715,
-      "grad_norm": 0.0005670891259796917,
-      "learning_rate": 0.00019999764388175053,
-      "loss": 46.0,
-      "step": 28589
-    },
-    {
-      "epoch": 2.185905155112105,
-      "grad_norm": 0.0008611117373220623,
-      "learning_rate": 0.0001999976437168637,
-      "loss": 46.0,
-      "step": 28590
-    },
-    {
-      "epoch": 2.1859816120954947,
-      "grad_norm": 0.0037224180996418,
-      "learning_rate": 0.00019999764355197112,
-      "loss": 46.0,
-      "step": 28591
-    },
-    {
-      "epoch": 2.1860580690788844,
-      "grad_norm": 0.0009443138842470944,
-      "learning_rate": 0.00019999764338707277,
-      "loss": 46.0,
-      "step": 28592
-    },
-    {
-      "epoch": 2.186134526062274,
-      "grad_norm": 0.0009225825197063386,
-      "learning_rate": 0.00019999764322216863,
-      "loss": 46.0,
-      "step": 28593
-    },
-    {
-      "epoch": 2.186210983045664,
-      "grad_norm": 0.0017345858504995704,
-      "learning_rate": 0.00019999764305725874,
-      "loss": 46.0,
-      "step": 28594
-    },
-    {
-      "epoch": 2.1862874400290537,
-      "grad_norm": 0.0009863072773441672,
-      "learning_rate": 0.00019999764289234308,
-      "loss": 46.0,
-      "step": 28595
-    },
-    {
-      "epoch": 2.1863638970124435,
-      "grad_norm": 0.0015016990946605802,
-      "learning_rate": 0.00019999764272742162,
-      "loss": 46.0,
-      "step": 28596
-    },
-    {
-      "epoch": 2.1864403539958333,
-      "grad_norm": 0.0006458009593188763,
-      "learning_rate": 0.00019999764256249444,
-      "loss": 46.0,
-      "step": 28597
-    },
-    {
-      "epoch": 2.186516810979223,
-      "grad_norm": 0.0031094634905457497,
-      "learning_rate": 0.00019999764239756146,
-      "loss": 46.0,
-      "step": 28598
-    },
-    {
-      "epoch": 2.1865932679626123,
-      "grad_norm": 0.001075044390745461,
-      "learning_rate": 0.0001999976422326227,
-      "loss": 46.0,
-      "step": 28599
-    },
-    {
-      "epoch": 2.186669724946002,
-      "grad_norm": 0.002897440455853939,
-      "learning_rate": 0.0001999976420676782,
-      "loss": 46.0,
-      "step": 28600
-    },
-    {
-      "epoch": 2.186746181929392,
-      "grad_norm": 0.0005038680392317474,
-      "learning_rate": 0.00019999764190272793,
-      "loss": 46.0,
-      "step": 28601
-    },
-    {
-      "epoch": 2.1868226389127816,
-      "grad_norm": 0.0014954719226807356,
-      "learning_rate": 0.00019999764173777188,
-      "loss": 46.0,
-      "step": 28602
-    },
-    {
-      "epoch": 2.1868990958961714,
-      "grad_norm": 0.0006701614474877715,
-      "learning_rate": 0.00019999764157281006,
-      "loss": 46.0,
-      "step": 28603
-    },
-    {
-      "epoch": 2.186975552879561,
-      "grad_norm": 0.0012092175893485546,
-      "learning_rate": 0.00019999764140784246,
-      "loss": 46.0,
-      "step": 28604
-    },
-    {
-      "epoch": 2.187052009862951,
-      "grad_norm": 0.0021120295859873295,
-      "learning_rate": 0.0001999976412428691,
-      "loss": 46.0,
-      "step": 28605
-    },
-    {
-      "epoch": 2.1871284668463407,
-      "grad_norm": 0.0011620563454926014,
-      "learning_rate": 0.00019999764107788995,
-      "loss": 46.0,
-      "step": 28606
-    },
-    {
-      "epoch": 2.1872049238297304,
-      "grad_norm": 0.0010778659489005804,
-      "learning_rate": 0.00019999764091290507,
-      "loss": 46.0,
-      "step": 28607
-    },
-    {
-      "epoch": 2.18728138081312,
-      "grad_norm": 0.0008230854291468859,
-      "learning_rate": 0.0001999976407479144,
-      "loss": 46.0,
-      "step": 28608
-    },
-    {
-      "epoch": 2.18735783779651,
-      "grad_norm": 0.003926537930965424,
-      "learning_rate": 0.00019999764058291797,
-      "loss": 46.0,
-      "step": 28609
-    },
-    {
-      "epoch": 2.1874342947798993,
-      "grad_norm": 0.0014673509867861867,
-      "learning_rate": 0.00019999764041791576,
-      "loss": 46.0,
-      "step": 28610
-    },
-    {
-      "epoch": 2.187510751763289,
-      "grad_norm": 0.0023358173202723265,
-      "learning_rate": 0.00019999764025290778,
-      "loss": 46.0,
-      "step": 28611
-    },
-    {
-      "epoch": 2.187587208746679,
-      "grad_norm": 0.0008677556761540473,
-      "learning_rate": 0.00019999764008789403,
-      "loss": 46.0,
-      "step": 28612
-    },
-    {
-      "epoch": 2.1876636657300685,
-      "grad_norm": 0.002461050171405077,
-      "learning_rate": 0.00019999763992287453,
-      "loss": 46.0,
-      "step": 28613
-    },
-    {
-      "epoch": 2.1877401227134583,
-      "grad_norm": 0.0013192816404625773,
-      "learning_rate": 0.00019999763975784925,
-      "loss": 46.0,
-      "step": 28614
-    },
-    {
-      "epoch": 2.187816579696848,
-      "grad_norm": 0.000389408553019166,
-      "learning_rate": 0.00019999763959281818,
-      "loss": 46.0,
-      "step": 28615
-    },
-    {
-      "epoch": 2.187893036680238,
-      "grad_norm": 0.00046909047523513436,
-      "learning_rate": 0.00019999763942778138,
-      "loss": 46.0,
-      "step": 28616
-    },
-    {
-      "epoch": 2.1879694936636276,
-      "grad_norm": 0.0029299843590706587,
-      "learning_rate": 0.0001999976392627388,
-      "loss": 46.0,
-      "step": 28617
-    },
-    {
-      "epoch": 2.1880459506470173,
-      "grad_norm": 0.0012349144089967012,
-      "learning_rate": 0.0001999976390976904,
-      "loss": 46.0,
-      "step": 28618
-    },
-    {
-      "epoch": 2.188122407630407,
-      "grad_norm": 0.0012928008800372481,
-      "learning_rate": 0.00019999763893263628,
-      "loss": 46.0,
-      "step": 28619
-    },
-    {
-      "epoch": 2.188198864613797,
-      "grad_norm": 0.0014898685039952397,
-      "learning_rate": 0.00019999763876757637,
-      "loss": 46.0,
-      "step": 28620
-    },
-    {
-      "epoch": 2.188275321597186,
-      "grad_norm": 0.0006685965345241129,
-      "learning_rate": 0.0001999976386025107,
-      "loss": 46.0,
-      "step": 28621
-    },
-    {
-      "epoch": 2.188351778580576,
-      "grad_norm": 0.0011451664613559842,
-      "learning_rate": 0.00019999763843743927,
-      "loss": 46.0,
-      "step": 28622
-    },
-    {
-      "epoch": 2.1884282355639657,
-      "grad_norm": 0.0012365257134661078,
-      "learning_rate": 0.00019999763827236206,
-      "loss": 46.0,
-      "step": 28623
-    },
-    {
-      "epoch": 2.1885046925473555,
-      "grad_norm": 0.0016647852025926113,
-      "learning_rate": 0.00019999763810727908,
-      "loss": 46.0,
-      "step": 28624
-    },
-    {
-      "epoch": 2.1885811495307452,
-      "grad_norm": 0.0007989972946234047,
-      "learning_rate": 0.00019999763794219036,
-      "loss": 46.0,
-      "step": 28625
-    },
-    {
-      "epoch": 2.188657606514135,
-      "grad_norm": 0.0036217966116964817,
-      "learning_rate": 0.00019999763777709583,
-      "loss": 46.0,
-      "step": 28626
-    },
-    {
-      "epoch": 2.1887340634975248,
-      "grad_norm": 0.0012161037884652615,
-      "learning_rate": 0.00019999763761199555,
-      "loss": 46.0,
-      "step": 28627
-    },
-    {
-      "epoch": 2.1888105204809145,
-      "grad_norm": 0.0006489575025625527,
-      "learning_rate": 0.00019999763744688948,
-      "loss": 46.0,
-      "step": 28628
-    },
-    {
-      "epoch": 2.1888869774643043,
-      "grad_norm": 0.0014411701122298837,
-      "learning_rate": 0.00019999763728177766,
-      "loss": 46.0,
-      "step": 28629
-    },
-    {
-      "epoch": 2.188963434447694,
-      "grad_norm": 0.0010443803621456027,
-      "learning_rate": 0.0001999976371166601,
-      "loss": 46.0,
-      "step": 28630
-    },
-    {
-      "epoch": 2.189039891431084,
-      "grad_norm": 0.015116575174033642,
-      "learning_rate": 0.0001999976369515367,
-      "loss": 46.0,
-      "step": 28631
-    },
-    {
-      "epoch": 2.189116348414473,
-      "grad_norm": 0.0015881045255810022,
-      "learning_rate": 0.00019999763678640758,
-      "loss": 46.0,
-      "step": 28632
-    },
-    {
-      "epoch": 2.189192805397863,
-      "grad_norm": 0.00038838779437355697,
-      "learning_rate": 0.00019999763662127267,
-      "loss": 46.0,
-      "step": 28633
-    },
-    {
-      "epoch": 2.1892692623812526,
-      "grad_norm": 0.0012234424939379096,
-      "learning_rate": 0.000199997636456132,
-      "loss": 46.0,
-      "step": 28634
-    },
-    {
-      "epoch": 2.1893457193646424,
-      "grad_norm": 0.004736605565994978,
-      "learning_rate": 0.00019999763629098557,
-      "loss": 46.0,
-      "step": 28635
-    },
-    {
-      "epoch": 2.189422176348032,
-      "grad_norm": 0.0003342684358358383,
-      "learning_rate": 0.00019999763612583337,
-      "loss": 46.0,
-      "step": 28636
-    },
-    {
-      "epoch": 2.189498633331422,
-      "grad_norm": 0.0005885021528229117,
-      "learning_rate": 0.0001999976359606754,
-      "loss": 46.0,
-      "step": 28637
-    },
-    {
-      "epoch": 2.1895750903148117,
-      "grad_norm": 0.0011080928379669785,
-      "learning_rate": 0.00019999763579551163,
-      "loss": 46.0,
-      "step": 28638
-    },
-    {
-      "epoch": 2.1896515472982014,
-      "grad_norm": 0.0019398547010496259,
-      "learning_rate": 0.00019999763563034213,
-      "loss": 46.0,
-      "step": 28639
-    },
-    {
-      "epoch": 2.189728004281591,
-      "grad_norm": 0.0007282826700247824,
-      "learning_rate": 0.00019999763546516683,
-      "loss": 46.0,
-      "step": 28640
-    },
-    {
-      "epoch": 2.189804461264981,
-      "grad_norm": 0.0006332109333015978,
-      "learning_rate": 0.00019999763529998576,
-      "loss": 46.0,
-      "step": 28641
-    },
-    {
-      "epoch": 2.1898809182483703,
-      "grad_norm": 0.0010016696760430932,
-      "learning_rate": 0.00019999763513479894,
-      "loss": 46.0,
-      "step": 28642
-    },
-    {
-      "epoch": 2.18995737523176,
-      "grad_norm": 0.002773863961920142,
-      "learning_rate": 0.00019999763496960634,
-      "loss": 46.0,
-      "step": 28643
-    },
-    {
-      "epoch": 2.19003383221515,
-      "grad_norm": 0.0007875406881794333,
-      "learning_rate": 0.000199997634804408,
-      "loss": 46.0,
-      "step": 28644
-    },
-    {
-      "epoch": 2.1901102891985396,
-      "grad_norm": 0.0017082927515730262,
-      "learning_rate": 0.00019999763463920384,
-      "loss": 46.0,
-      "step": 28645
-    },
-    {
-      "epoch": 2.1901867461819293,
-      "grad_norm": 0.0016531051369383931,
-      "learning_rate": 0.00019999763447399395,
-      "loss": 46.0,
-      "step": 28646
-    },
-    {
-      "epoch": 2.190263203165319,
-      "grad_norm": 0.0011235586134716868,
-      "learning_rate": 0.0001999976343087783,
-      "loss": 46.0,
-      "step": 28647
-    },
-    {
-      "epoch": 2.190339660148709,
-      "grad_norm": 0.0009194666054099798,
-      "learning_rate": 0.00019999763414355686,
-      "loss": 46.0,
-      "step": 28648
-    },
-    {
-      "epoch": 2.1904161171320986,
-      "grad_norm": 0.0006754841306246817,
-      "learning_rate": 0.00019999763397832965,
-      "loss": 46.0,
-      "step": 28649
-    },
-    {
-      "epoch": 2.1904925741154884,
-      "grad_norm": 0.0021153809502720833,
-      "learning_rate": 0.00019999763381309667,
-      "loss": 46.0,
-      "step": 28650
-    },
-    {
-      "epoch": 2.190569031098878,
-      "grad_norm": 0.0015513041289523244,
-      "learning_rate": 0.00019999763364785791,
-      "loss": 46.0,
-      "step": 28651
-    },
-    {
-      "epoch": 2.190645488082268,
-      "grad_norm": 0.0005015857750549912,
-      "learning_rate": 0.0001999976334826134,
-      "loss": 46.0,
-      "step": 28652
-    },
-    {
-      "epoch": 2.1907219450656576,
-      "grad_norm": 0.0014262673212215304,
-      "learning_rate": 0.0001999976333173631,
-      "loss": 46.0,
-      "step": 28653
-    },
-    {
-      "epoch": 2.190798402049047,
-      "grad_norm": 0.0010433244751766324,
-      "learning_rate": 0.00019999763315210704,
-      "loss": 46.0,
-      "step": 28654
-    },
-    {
-      "epoch": 2.1908748590324367,
-      "grad_norm": 0.0009178290492855012,
-      "learning_rate": 0.00019999763298684522,
-      "loss": 46.0,
-      "step": 28655
-    },
-    {
-      "epoch": 2.1909513160158265,
-      "grad_norm": 0.0038189892657101154,
-      "learning_rate": 0.00019999763282157763,
-      "loss": 46.0,
-      "step": 28656
-    },
-    {
-      "epoch": 2.1910277729992162,
-      "grad_norm": 0.013782191090285778,
-      "learning_rate": 0.00019999763265630426,
-      "loss": 46.0,
-      "step": 28657
-    },
-    {
-      "epoch": 2.191104229982606,
-      "grad_norm": 0.0006278374930843711,
-      "learning_rate": 0.00019999763249102512,
-      "loss": 46.0,
-      "step": 28658
-    },
-    {
-      "epoch": 2.1911806869659958,
-      "grad_norm": 0.0021108014043420553,
-      "learning_rate": 0.00019999763232574023,
-      "loss": 46.0,
-      "step": 28659
-    },
-    {
-      "epoch": 2.1912571439493855,
-      "grad_norm": 0.006868427619338036,
-      "learning_rate": 0.00019999763216044952,
-      "loss": 46.0,
-      "step": 28660
-    },
-    {
-      "epoch": 2.1913336009327753,
-      "grad_norm": 0.0005290281842462718,
-      "learning_rate": 0.0001999976319951531,
-      "loss": 46.0,
-      "step": 28661
-    },
-    {
-      "epoch": 2.191410057916165,
-      "grad_norm": 0.002222876762971282,
-      "learning_rate": 0.0001999976318298509,
-      "loss": 46.0,
-      "step": 28662
-    },
-    {
-      "epoch": 2.191486514899555,
-      "grad_norm": 0.0008497189846821129,
-      "learning_rate": 0.00019999763166454293,
-      "loss": 46.0,
-      "step": 28663
-    },
-    {
-      "epoch": 2.191562971882944,
-      "grad_norm": 0.0006918266881257296,
-      "learning_rate": 0.00019999763149922917,
-      "loss": 46.0,
-      "step": 28664
-    },
-    {
-      "epoch": 2.191639428866334,
-      "grad_norm": 0.0006093996344134212,
-      "learning_rate": 0.00019999763133390965,
-      "loss": 46.0,
-      "step": 28665
-    },
-    {
-      "epoch": 2.1917158858497237,
-      "grad_norm": 0.0003278770309407264,
-      "learning_rate": 0.00019999763116858437,
-      "loss": 46.0,
-      "step": 28666
-    },
-    {
-      "epoch": 2.1917923428331134,
-      "grad_norm": 0.0008522662683390081,
-      "learning_rate": 0.0001999976310032533,
-      "loss": 46.0,
-      "step": 28667
-    },
-    {
-      "epoch": 2.191868799816503,
-      "grad_norm": 0.0012082374887540936,
-      "learning_rate": 0.00019999763083791648,
-      "loss": 46.0,
-      "step": 28668
-    },
-    {
-      "epoch": 2.191945256799893,
-      "grad_norm": 0.0005280483746901155,
-      "learning_rate": 0.0001999976306725739,
-      "loss": 46.0,
-      "step": 28669
-    },
-    {
-      "epoch": 2.1920217137832827,
-      "grad_norm": 0.0009032447705976665,
-      "learning_rate": 0.00019999763050722552,
-      "loss": 46.0,
-      "step": 28670
-    },
-    {
-      "epoch": 2.1920981707666725,
-      "grad_norm": 0.0006261604139581323,
-      "learning_rate": 0.00019999763034187138,
-      "loss": 46.0,
-      "step": 28671
-    },
-    {
-      "epoch": 2.192174627750062,
-      "grad_norm": 0.0012686256086453795,
-      "learning_rate": 0.00019999763017651147,
-      "loss": 46.0,
-      "step": 28672
-    },
-    {
-      "epoch": 2.192251084733452,
-      "grad_norm": 0.0008066724985837936,
-      "learning_rate": 0.0001999976300111458,
-      "loss": 46.0,
-      "step": 28673
-    },
-    {
-      "epoch": 2.1923275417168417,
-      "grad_norm": 0.0007947124540805817,
-      "learning_rate": 0.00019999762984577438,
-      "loss": 46.0,
-      "step": 28674
-    },
-    {
-      "epoch": 2.1924039987002315,
-      "grad_norm": 0.002620997140184045,
-      "learning_rate": 0.00019999762968039714,
-      "loss": 46.0,
-      "step": 28675
-    },
-    {
-      "epoch": 2.192480455683621,
-      "grad_norm": 0.0011536844540387392,
-      "learning_rate": 0.00019999762951501416,
-      "loss": 46.0,
-      "step": 28676
-    },
-    {
-      "epoch": 2.1925569126670106,
-      "grad_norm": 0.0018091228557750583,
-      "learning_rate": 0.00019999762934962544,
-      "loss": 46.0,
-      "step": 28677
-    },
-    {
-      "epoch": 2.1926333696504003,
-      "grad_norm": 0.002807512879371643,
-      "learning_rate": 0.00019999762918423088,
-      "loss": 46.0,
-      "step": 28678
-    },
-    {
-      "epoch": 2.19270982663379,
-      "grad_norm": 0.0012814572546631098,
-      "learning_rate": 0.0001999976290188306,
-      "loss": 46.0,
-      "step": 28679
-    },
-    {
-      "epoch": 2.19278628361718,
-      "grad_norm": 0.0015938919968903065,
-      "learning_rate": 0.00019999762885342454,
-      "loss": 46.0,
-      "step": 28680
-    },
-    {
-      "epoch": 2.1928627406005696,
-      "grad_norm": 0.0003120073233731091,
-      "learning_rate": 0.00019999762868801272,
-      "loss": 46.0,
-      "step": 28681
-    },
-    {
-      "epoch": 2.1929391975839594,
-      "grad_norm": 0.0014784664381295443,
-      "learning_rate": 0.00019999762852259512,
-      "loss": 46.0,
-      "step": 28682
-    },
-    {
-      "epoch": 2.193015654567349,
-      "grad_norm": 0.00027768261497840285,
-      "learning_rate": 0.00019999762835717176,
-      "loss": 46.0,
-      "step": 28683
-    },
-    {
-      "epoch": 2.193092111550739,
-      "grad_norm": 0.0013586970744654536,
-      "learning_rate": 0.00019999762819174262,
-      "loss": 46.0,
-      "step": 28684
-    },
-    {
-      "epoch": 2.1931685685341287,
-      "grad_norm": 0.0025398428551852703,
-      "learning_rate": 0.0001999976280263077,
-      "loss": 46.0,
-      "step": 28685
-    },
-    {
-      "epoch": 2.193245025517518,
-      "grad_norm": 0.0019675600342452526,
-      "learning_rate": 0.00019999762786086702,
-      "loss": 46.0,
-      "step": 28686
-    },
-    {
-      "epoch": 2.1933214825009077,
-      "grad_norm": 0.0013736619148403406,
-      "learning_rate": 0.0001999976276954206,
-      "loss": 46.0,
-      "step": 28687
-    },
-    {
-      "epoch": 2.1933979394842975,
-      "grad_norm": 0.0007126835407689214,
-      "learning_rate": 0.00019999762752996838,
-      "loss": 46.0,
-      "step": 28688
-    },
-    {
-      "epoch": 2.1934743964676873,
-      "grad_norm": 0.0012680370127782226,
-      "learning_rate": 0.0001999976273645104,
-      "loss": 46.0,
-      "step": 28689
-    },
-    {
-      "epoch": 2.193550853451077,
-      "grad_norm": 0.0031738809775561094,
-      "learning_rate": 0.00019999762719904665,
-      "loss": 46.0,
-      "step": 28690
-    },
-    {
-      "epoch": 2.193627310434467,
-      "grad_norm": 0.0006710441666655242,
-      "learning_rate": 0.00019999762703357712,
-      "loss": 46.0,
-      "step": 28691
-    },
-    {
-      "epoch": 2.1937037674178566,
-      "grad_norm": 0.0011365828104317188,
-      "learning_rate": 0.00019999762686810182,
-      "loss": 46.0,
-      "step": 28692
-    },
-    {
-      "epoch": 2.1937802244012463,
-      "grad_norm": 0.008483314886689186,
-      "learning_rate": 0.00019999762670262078,
-      "loss": 46.0,
-      "step": 28693
-    },
-    {
-      "epoch": 2.193856681384636,
-      "grad_norm": 0.0013568695867434144,
-      "learning_rate": 0.00019999762653713393,
-      "loss": 46.0,
-      "step": 28694
-    },
-    {
-      "epoch": 2.193933138368026,
-      "grad_norm": 0.0020197569392621517,
-      "learning_rate": 0.00019999762637164134,
-      "loss": 46.0,
-      "step": 28695
-    },
-    {
-      "epoch": 2.1940095953514156,
-      "grad_norm": 0.0010159629164263606,
-      "learning_rate": 0.00019999762620614297,
-      "loss": 46.0,
-      "step": 28696
-    },
-    {
-      "epoch": 2.1940860523348054,
-      "grad_norm": 0.0005594915128313005,
-      "learning_rate": 0.00019999762604063883,
-      "loss": 46.0,
-      "step": 28697
-    },
-    {
-      "epoch": 2.1941625093181947,
-      "grad_norm": 0.0014561618445441127,
-      "learning_rate": 0.00019999762587512895,
-      "loss": 46.0,
-      "step": 28698
-    },
-    {
-      "epoch": 2.1942389663015844,
-      "grad_norm": 0.0015211966820061207,
-      "learning_rate": 0.00019999762570961326,
-      "loss": 46.0,
-      "step": 28699
-    },
-    {
-      "epoch": 2.194315423284974,
-      "grad_norm": 0.0023055304773151875,
-      "learning_rate": 0.0001999976255440918,
-      "loss": 46.0,
-      "step": 28700
-    },
-    {
-      "epoch": 2.194391880268364,
-      "grad_norm": 0.0005494029610417783,
-      "learning_rate": 0.0001999976253785646,
-      "loss": 46.0,
-      "step": 28701
-    },
-    {
-      "epoch": 2.1944683372517537,
-      "grad_norm": 0.0007750372169539332,
-      "learning_rate": 0.0001999976252130316,
-      "loss": 46.0,
-      "step": 28702
-    },
-    {
-      "epoch": 2.1945447942351435,
-      "grad_norm": 0.0037118406035006046,
-      "learning_rate": 0.00019999762504749287,
-      "loss": 46.0,
-      "step": 28703
-    },
-    {
-      "epoch": 2.1946212512185332,
-      "grad_norm": 0.00036844678106717765,
-      "learning_rate": 0.00019999762488194834,
-      "loss": 46.0,
-      "step": 28704
-    },
-    {
-      "epoch": 2.194697708201923,
-      "grad_norm": 0.0020455995108932257,
-      "learning_rate": 0.00019999762471639804,
-      "loss": 46.0,
-      "step": 28705
-    },
-    {
-      "epoch": 2.1947741651853128,
-      "grad_norm": 0.0008475796785205603,
-      "learning_rate": 0.000199997624550842,
-      "loss": 46.0,
-      "step": 28706
-    },
-    {
-      "epoch": 2.1948506221687025,
-      "grad_norm": 0.0011455765925347805,
-      "learning_rate": 0.00019999762438528015,
-      "loss": 46.0,
-      "step": 28707
-    },
-    {
-      "epoch": 2.194927079152092,
-      "grad_norm": 0.0022386317141354084,
-      "learning_rate": 0.00019999762421971256,
-      "loss": 46.0,
-      "step": 28708
-    },
-    {
-      "epoch": 2.1950035361354816,
-      "grad_norm": 0.0007858487660996616,
-      "learning_rate": 0.0001999976240541392,
-      "loss": 46.0,
-      "step": 28709
-    },
-    {
-      "epoch": 2.1950799931188714,
-      "grad_norm": 0.0014285441720858216,
-      "learning_rate": 0.00019999762388856005,
-      "loss": 46.0,
-      "step": 28710
-    },
-    {
-      "epoch": 2.195156450102261,
-      "grad_norm": 0.0006505746277980506,
-      "learning_rate": 0.0001999976237229751,
-      "loss": 46.0,
-      "step": 28711
-    },
-    {
-      "epoch": 2.195232907085651,
-      "grad_norm": 0.0007188899908214808,
-      "learning_rate": 0.00019999762355738448,
-      "loss": 46.0,
-      "step": 28712
-    },
-    {
-      "epoch": 2.1953093640690406,
-      "grad_norm": 0.002463469048961997,
-      "learning_rate": 0.000199997623391788,
-      "loss": 46.0,
-      "step": 28713
-    },
-    {
-      "epoch": 2.1953858210524304,
-      "grad_norm": 0.0006235355394892395,
-      "learning_rate": 0.00019999762322618582,
-      "loss": 46.0,
-      "step": 28714
-    },
-    {
-      "epoch": 2.19546227803582,
-      "grad_norm": 0.004243816249072552,
-      "learning_rate": 0.0001999976230605778,
-      "loss": 46.0,
-      "step": 28715
-    },
-    {
-      "epoch": 2.19553873501921,
-      "grad_norm": 0.0006481556920334697,
-      "learning_rate": 0.00019999762289496406,
-      "loss": 46.0,
-      "step": 28716
-    },
-    {
-      "epoch": 2.1956151920025997,
-      "grad_norm": 0.0009381757699884474,
-      "learning_rate": 0.00019999762272934453,
-      "loss": 46.0,
-      "step": 28717
-    },
-    {
-      "epoch": 2.1956916489859895,
-      "grad_norm": 0.0008142460137605667,
-      "learning_rate": 0.00019999762256371924,
-      "loss": 46.0,
-      "step": 28718
-    },
-    {
-      "epoch": 2.195768105969379,
-      "grad_norm": 0.008301754482090473,
-      "learning_rate": 0.00019999762239808816,
-      "loss": 46.0,
-      "step": 28719
-    },
-    {
-      "epoch": 2.1958445629527685,
-      "grad_norm": 0.0015477570705115795,
-      "learning_rate": 0.00019999762223245134,
-      "loss": 46.0,
-      "step": 28720
-    },
-    {
-      "epoch": 2.1959210199361583,
-      "grad_norm": 0.0016006340738385916,
-      "learning_rate": 0.00019999762206680875,
-      "loss": 46.0,
-      "step": 28721
-    },
-    {
-      "epoch": 2.195997476919548,
-      "grad_norm": 0.0005535492091439664,
-      "learning_rate": 0.0001999976219011604,
-      "loss": 46.0,
-      "step": 28722
-    },
-    {
-      "epoch": 2.196073933902938,
-      "grad_norm": 0.004451784770935774,
-      "learning_rate": 0.00019999762173550622,
-      "loss": 46.0,
-      "step": 28723
-    },
-    {
-      "epoch": 2.1961503908863276,
-      "grad_norm": 0.0018984171329066157,
-      "learning_rate": 0.00019999762156984634,
-      "loss": 46.0,
-      "step": 28724
-    },
-    {
-      "epoch": 2.1962268478697173,
-      "grad_norm": 0.0003904541954398155,
-      "learning_rate": 0.00019999762140418065,
-      "loss": 46.0,
-      "step": 28725
-    },
-    {
-      "epoch": 2.196303304853107,
-      "grad_norm": 0.0025313321966677904,
-      "learning_rate": 0.0001999976212385092,
-      "loss": 46.0,
-      "step": 28726
-    },
-    {
-      "epoch": 2.196379761836497,
-      "grad_norm": 0.00069714046549052,
-      "learning_rate": 0.000199997621072832,
-      "loss": 46.0,
-      "step": 28727
-    },
-    {
-      "epoch": 2.1964562188198866,
-      "grad_norm": 0.0022827403154224157,
-      "learning_rate": 0.000199997620907149,
-      "loss": 46.0,
-      "step": 28728
-    },
-    {
-      "epoch": 2.1965326758032764,
-      "grad_norm": 0.0012537987204268575,
-      "learning_rate": 0.00019999762074146026,
-      "loss": 46.0,
-      "step": 28729
-    },
-    {
-      "epoch": 2.1966091327866657,
-      "grad_norm": 0.00026411254657432437,
-      "learning_rate": 0.00019999762057576573,
-      "loss": 46.0,
-      "step": 28730
-    },
-    {
-      "epoch": 2.1966855897700555,
-      "grad_norm": 0.0024505213368684053,
-      "learning_rate": 0.00019999762041006543,
-      "loss": 46.0,
-      "step": 28731
-    },
-    {
-      "epoch": 2.196762046753445,
-      "grad_norm": 0.004176511894911528,
-      "learning_rate": 0.00019999762024435936,
-      "loss": 46.0,
-      "step": 28732
-    },
-    {
-      "epoch": 2.196838503736835,
-      "grad_norm": 0.0022932779975235462,
-      "learning_rate": 0.00019999762007864752,
-      "loss": 46.0,
-      "step": 28733
-    },
-    {
-      "epoch": 2.1969149607202247,
-      "grad_norm": 0.0028575395699590445,
-      "learning_rate": 0.00019999761991292992,
-      "loss": 46.0,
-      "step": 28734
-    },
-    {
-      "epoch": 2.1969914177036145,
-      "grad_norm": 0.0026604263111948967,
-      "learning_rate": 0.00019999761974720656,
-      "loss": 46.0,
-      "step": 28735
-    },
-    {
-      "epoch": 2.1970678746870043,
-      "grad_norm": 0.0012218424817547202,
-      "learning_rate": 0.0001999976195814774,
-      "loss": 46.0,
-      "step": 28736
-    },
-    {
-      "epoch": 2.197144331670394,
-      "grad_norm": 0.00360929686576128,
-      "learning_rate": 0.0001999976194157425,
-      "loss": 46.0,
-      "step": 28737
-    },
-    {
-      "epoch": 2.197220788653784,
-      "grad_norm": 0.001043299213051796,
-      "learning_rate": 0.00019999761925000185,
-      "loss": 46.0,
-      "step": 28738
-    },
-    {
-      "epoch": 2.1972972456371735,
-      "grad_norm": 0.00041641900315880775,
-      "learning_rate": 0.0001999976190842554,
-      "loss": 46.0,
-      "step": 28739
-    },
-    {
-      "epoch": 2.1973737026205633,
-      "grad_norm": 0.0025470878463238478,
-      "learning_rate": 0.00019999761891850316,
-      "loss": 46.0,
-      "step": 28740
-    },
-    {
-      "epoch": 2.1974501596039526,
-      "grad_norm": 0.002362113678827882,
-      "learning_rate": 0.00019999761875274516,
-      "loss": 46.0,
-      "step": 28741
-    },
-    {
-      "epoch": 2.1975266165873424,
-      "grad_norm": 0.004090269096195698,
-      "learning_rate": 0.0001999976185869814,
-      "loss": 46.0,
-      "step": 28742
-    },
-    {
-      "epoch": 2.197603073570732,
-      "grad_norm": 0.002057581441476941,
-      "learning_rate": 0.0001999976184212119,
-      "loss": 46.0,
-      "step": 28743
-    },
-    {
-      "epoch": 2.197679530554122,
-      "grad_norm": 0.0011703399941325188,
-      "learning_rate": 0.0001999976182554366,
-      "loss": 46.0,
-      "step": 28744
-    },
-    {
-      "epoch": 2.1977559875375117,
-      "grad_norm": 0.002911994932219386,
-      "learning_rate": 0.00019999761808965554,
-      "loss": 46.0,
-      "step": 28745
-    },
-    {
-      "epoch": 2.1978324445209014,
-      "grad_norm": 0.0007452918798662722,
-      "learning_rate": 0.0001999976179238687,
-      "loss": 46.0,
-      "step": 28746
-    },
-    {
-      "epoch": 2.197908901504291,
-      "grad_norm": 0.0007457157480530441,
-      "learning_rate": 0.0001999976177580761,
-      "loss": 46.0,
-      "step": 28747
-    },
-    {
-      "epoch": 2.197985358487681,
-      "grad_norm": 0.0027052657678723335,
-      "learning_rate": 0.0001999976175922777,
-      "loss": 46.0,
-      "step": 28748
-    },
-    {
-      "epoch": 2.1980618154710707,
-      "grad_norm": 0.0007662579300813377,
-      "learning_rate": 0.00019999761742647357,
-      "loss": 46.0,
-      "step": 28749
-    },
-    {
-      "epoch": 2.1981382724544605,
-      "grad_norm": 0.001998987514525652,
-      "learning_rate": 0.0001999976172606637,
-      "loss": 46.0,
-      "step": 28750
-    },
-    {
-      "epoch": 2.1982147294378502,
-      "grad_norm": 0.0020735764410346746,
-      "learning_rate": 0.00019999761709484798,
-      "loss": 46.0,
-      "step": 28751
-    },
-    {
-      "epoch": 2.1982911864212396,
-      "grad_norm": 0.0017626654589548707,
-      "learning_rate": 0.00019999761692902654,
-      "loss": 46.0,
-      "step": 28752
-    },
-    {
-      "epoch": 2.1983676434046293,
-      "grad_norm": 0.001942440983839333,
-      "learning_rate": 0.0001999976167631993,
-      "loss": 46.0,
-      "step": 28753
-    },
-    {
-      "epoch": 2.198444100388019,
-      "grad_norm": 0.0012866543838754296,
-      "learning_rate": 0.00019999761659736634,
-      "loss": 46.0,
-      "step": 28754
-    },
-    {
-      "epoch": 2.198520557371409,
-      "grad_norm": 0.001886275364086032,
-      "learning_rate": 0.00019999761643152758,
-      "loss": 46.0,
-      "step": 28755
-    },
-    {
-      "epoch": 2.1985970143547986,
-      "grad_norm": 0.002110939472913742,
-      "learning_rate": 0.00019999761626568306,
-      "loss": 46.0,
-      "step": 28756
-    },
-    {
-      "epoch": 2.1986734713381884,
-      "grad_norm": 0.00474103819578886,
-      "learning_rate": 0.00019999761609983273,
-      "loss": 46.0,
-      "step": 28757
-    },
-    {
-      "epoch": 2.198749928321578,
-      "grad_norm": 0.0020059135276824236,
-      "learning_rate": 0.0001999976159339767,
-      "loss": 46.0,
-      "step": 28758
-    },
-    {
-      "epoch": 2.198826385304968,
-      "grad_norm": 0.007026014383882284,
-      "learning_rate": 0.00019999761576811485,
-      "loss": 46.0,
-      "step": 28759
-    },
-    {
-      "epoch": 2.1989028422883576,
-      "grad_norm": 0.000674641109071672,
-      "learning_rate": 0.00019999761560224725,
-      "loss": 46.0,
-      "step": 28760
-    },
-    {
-      "epoch": 2.1989792992717474,
-      "grad_norm": 0.00117670523468405,
-      "learning_rate": 0.00019999761543637386,
-      "loss": 46.0,
-      "step": 28761
-    },
-    {
-      "epoch": 2.199055756255137,
-      "grad_norm": 0.00044244524906389415,
-      "learning_rate": 0.00019999761527049473,
-      "loss": 46.0,
-      "step": 28762
-    },
-    {
-      "epoch": 2.1991322132385265,
-      "grad_norm": 0.00045616025454364717,
-      "learning_rate": 0.00019999761510460981,
-      "loss": 46.0,
-      "step": 28763
-    },
-    {
-      "epoch": 2.1992086702219162,
-      "grad_norm": 0.001300415606237948,
-      "learning_rate": 0.00019999761493871913,
-      "loss": 46.0,
-      "step": 28764
-    },
-    {
-      "epoch": 2.199285127205306,
-      "grad_norm": 0.0009729649173095822,
-      "learning_rate": 0.00019999761477282267,
-      "loss": 46.0,
-      "step": 28765
-    },
-    {
-      "epoch": 2.1993615841886958,
-      "grad_norm": 0.010112939402461052,
-      "learning_rate": 0.00019999761460692047,
-      "loss": 46.0,
-      "step": 28766
-    },
-    {
-      "epoch": 2.1994380411720855,
-      "grad_norm": 0.0027120918966829777,
-      "learning_rate": 0.0001999976144410125,
-      "loss": 46.0,
-      "step": 28767
-    },
-    {
-      "epoch": 2.1995144981554753,
-      "grad_norm": 0.0008452770416624844,
-      "learning_rate": 0.00019999761427509874,
-      "loss": 46.0,
-      "step": 28768
-    },
-    {
-      "epoch": 2.199590955138865,
-      "grad_norm": 0.0013440067414194345,
-      "learning_rate": 0.0001999976141091792,
-      "loss": 46.0,
-      "step": 28769
-    },
-    {
-      "epoch": 2.199667412122255,
-      "grad_norm": 0.0008116482640616596,
-      "learning_rate": 0.0001999976139432539,
-      "loss": 46.0,
-      "step": 28770
-    },
-    {
-      "epoch": 2.1997438691056446,
-      "grad_norm": 0.002448539948090911,
-      "learning_rate": 0.00019999761377732282,
-      "loss": 46.0,
-      "step": 28771
-    },
-    {
-      "epoch": 2.1998203260890343,
-      "grad_norm": 0.0006727167638018727,
-      "learning_rate": 0.000199997613611386,
-      "loss": 46.0,
-      "step": 28772
-    },
-    {
-      "epoch": 2.1998967830724236,
-      "grad_norm": 0.002355123171582818,
-      "learning_rate": 0.00019999761344544338,
-      "loss": 46.0,
-      "step": 28773
-    },
-    {
-      "epoch": 2.1999732400558134,
-      "grad_norm": 0.0008522709831595421,
-      "learning_rate": 0.000199997613279495,
-      "loss": 46.0,
-      "step": 28774
-    },
-    {
-      "epoch": 2.200049697039203,
-      "grad_norm": 0.0040439129807055,
-      "learning_rate": 0.00019999761311354086,
-      "loss": 46.0,
-      "step": 28775
-    },
-    {
-      "epoch": 2.200126154022593,
-      "grad_norm": 0.0015502751339226961,
-      "learning_rate": 0.00019999761294758095,
-      "loss": 46.0,
-      "step": 28776
-    },
-    {
-      "epoch": 2.2002026110059827,
-      "grad_norm": 0.0007321261800825596,
-      "learning_rate": 0.00019999761278161526,
-      "loss": 46.0,
-      "step": 28777
-    },
-    {
-      "epoch": 2.2002790679893725,
-      "grad_norm": 0.002785300835967064,
-      "learning_rate": 0.00019999761261564383,
-      "loss": 46.0,
-      "step": 28778
-    },
-    {
-      "epoch": 2.200355524972762,
-      "grad_norm": 0.0008571309153921902,
-      "learning_rate": 0.00019999761244966663,
-      "loss": 46.0,
-      "step": 28779
-    },
-    {
-      "epoch": 2.200431981956152,
-      "grad_norm": 0.005570200737565756,
-      "learning_rate": 0.00019999761228368362,
-      "loss": 46.0,
-      "step": 28780
-    },
-    {
-      "epoch": 2.2005084389395417,
-      "grad_norm": 0.0031015167478471994,
-      "learning_rate": 0.00019999761211769485,
-      "loss": 46.0,
-      "step": 28781
-    },
-    {
-      "epoch": 2.2005848959229315,
-      "grad_norm": 0.0013768210774287581,
-      "learning_rate": 0.00019999761195170032,
-      "loss": 46.0,
-      "step": 28782
-    },
-    {
-      "epoch": 2.2006613529063213,
-      "grad_norm": 0.000893348129466176,
-      "learning_rate": 0.00019999761178570002,
-      "loss": 46.0,
-      "step": 28783
-    },
-    {
-      "epoch": 2.200737809889711,
-      "grad_norm": 0.0005879162927158177,
-      "learning_rate": 0.00019999761161969395,
-      "loss": 46.0,
-      "step": 28784
-    },
-    {
-      "epoch": 2.2008142668731003,
-      "grad_norm": 0.0035728267394006252,
-      "learning_rate": 0.0001999976114536821,
-      "loss": 46.0,
-      "step": 28785
-    },
-    {
-      "epoch": 2.20089072385649,
-      "grad_norm": 0.0007705323514528573,
-      "learning_rate": 0.00019999761128766452,
-      "loss": 46.0,
-      "step": 28786
-    },
-    {
-      "epoch": 2.20096718083988,
-      "grad_norm": 0.0020545287989079952,
-      "learning_rate": 0.00019999761112164116,
-      "loss": 46.0,
-      "step": 28787
-    },
-    {
-      "epoch": 2.2010436378232696,
-      "grad_norm": 0.0018090162193402648,
-      "learning_rate": 0.000199997610955612,
-      "loss": 46.0,
-      "step": 28788
-    },
-    {
-      "epoch": 2.2011200948066594,
-      "grad_norm": 0.005108046345412731,
-      "learning_rate": 0.00019999761078957708,
-      "loss": 46.0,
-      "step": 28789
-    },
-    {
-      "epoch": 2.201196551790049,
-      "grad_norm": 0.0011435793712735176,
-      "learning_rate": 0.0001999976106235364,
-      "loss": 46.0,
-      "step": 28790
-    },
-    {
-      "epoch": 2.201273008773439,
-      "grad_norm": 0.001082267495803535,
-      "learning_rate": 0.0001999976104574899,
-      "loss": 46.0,
-      "step": 28791
-    },
-    {
-      "epoch": 2.2013494657568287,
-      "grad_norm": 0.004285803530365229,
-      "learning_rate": 0.00019999761029143774,
-      "loss": 46.0,
-      "step": 28792
-    },
-    {
-      "epoch": 2.2014259227402184,
-      "grad_norm": 0.0010516910115256906,
-      "learning_rate": 0.00019999761012537973,
-      "loss": 46.0,
-      "step": 28793
-    },
-    {
-      "epoch": 2.201502379723608,
-      "grad_norm": 0.005702539347112179,
-      "learning_rate": 0.00019999760995931598,
-      "loss": 46.0,
-      "step": 28794
-    },
-    {
-      "epoch": 2.2015788367069975,
-      "grad_norm": 0.006272371392697096,
-      "learning_rate": 0.00019999760979324643,
-      "loss": 46.0,
-      "step": 28795
-    },
-    {
-      "epoch": 2.2016552936903873,
-      "grad_norm": 0.00568053312599659,
-      "learning_rate": 0.00019999760962717113,
-      "loss": 46.0,
-      "step": 28796
-    },
-    {
-      "epoch": 2.201731750673777,
-      "grad_norm": 0.0018560357857495546,
-      "learning_rate": 0.00019999760946109006,
-      "loss": 46.0,
-      "step": 28797
-    },
-    {
-      "epoch": 2.201808207657167,
-      "grad_norm": 0.0004826985823456198,
-      "learning_rate": 0.00019999760929500325,
-      "loss": 46.0,
-      "step": 28798
-    },
-    {
-      "epoch": 2.2018846646405565,
-      "grad_norm": 0.00208675442263484,
-      "learning_rate": 0.00019999760912891063,
-      "loss": 46.0,
-      "step": 28799
-    },
-    {
-      "epoch": 2.2019611216239463,
-      "grad_norm": 0.00037602230440825224,
-      "learning_rate": 0.00019999760896281227,
-      "loss": 46.0,
-      "step": 28800
-    },
-    {
-      "epoch": 2.202037578607336,
-      "grad_norm": 0.0011239483719691634,
-      "learning_rate": 0.0001999976087967081,
-      "loss": 46.0,
-      "step": 28801
-    },
-    {
-      "epoch": 2.202114035590726,
-      "grad_norm": 0.0009544836357235909,
-      "learning_rate": 0.00019999760863059817,
-      "loss": 46.0,
-      "step": 28802
-    },
-    {
-      "epoch": 2.2021904925741156,
-      "grad_norm": 0.0008614493417553604,
-      "learning_rate": 0.0001999976084644825,
-      "loss": 46.0,
-      "step": 28803
-    },
-    {
-      "epoch": 2.2022669495575053,
-      "grad_norm": 0.0009722508257254958,
-      "learning_rate": 0.00019999760829836105,
-      "loss": 46.0,
-      "step": 28804
-    },
-    {
-      "epoch": 2.202343406540895,
-      "grad_norm": 0.0010264882585033774,
-      "learning_rate": 0.0001999976081322338,
-      "loss": 46.0,
-      "step": 28805
-    },
-    {
-      "epoch": 2.202419863524285,
-      "grad_norm": 0.005427997559309006,
-      "learning_rate": 0.00019999760796610082,
-      "loss": 46.0,
-      "step": 28806
-    },
-    {
-      "epoch": 2.202496320507674,
-      "grad_norm": 0.0008990340866148472,
-      "learning_rate": 0.00019999760779996207,
-      "loss": 46.0,
-      "step": 28807
-    },
-    {
-      "epoch": 2.202572777491064,
-      "grad_norm": 0.001176820253022015,
-      "learning_rate": 0.00019999760763381752,
-      "loss": 46.0,
-      "step": 28808
-    },
-    {
-      "epoch": 2.2026492344744537,
-      "grad_norm": 0.0008145632455125451,
-      "learning_rate": 0.00019999760746766722,
-      "loss": 46.0,
-      "step": 28809
-    },
-    {
-      "epoch": 2.2027256914578435,
-      "grad_norm": 0.0014016643399372697,
-      "learning_rate": 0.00019999760730151115,
-      "loss": 46.0,
-      "step": 28810
-    },
-    {
-      "epoch": 2.2028021484412332,
-      "grad_norm": 0.0002928642206825316,
-      "learning_rate": 0.0001999976071353493,
-      "loss": 46.0,
-      "step": 28811
-    },
-    {
-      "epoch": 2.202878605424623,
-      "grad_norm": 0.0014874918852001429,
-      "learning_rate": 0.0001999976069691817,
-      "loss": 46.0,
-      "step": 28812
-    },
-    {
-      "epoch": 2.2029550624080128,
-      "grad_norm": 0.0008088365430012345,
-      "learning_rate": 0.00019999760680300833,
-      "loss": 46.0,
-      "step": 28813
-    },
-    {
-      "epoch": 2.2030315193914025,
-      "grad_norm": 0.0005311599816195667,
-      "learning_rate": 0.0001999976066368292,
-      "loss": 46.0,
-      "step": 28814
-    },
-    {
-      "epoch": 2.2031079763747923,
-      "grad_norm": 0.002118383301422,
-      "learning_rate": 0.00019999760647064428,
-      "loss": 46.0,
-      "step": 28815
-    },
-    {
-      "epoch": 2.203184433358182,
-      "grad_norm": 0.0006980802863836288,
-      "learning_rate": 0.0001999976063044536,
-      "loss": 46.0,
-      "step": 28816
-    },
-    {
-      "epoch": 2.2032608903415714,
-      "grad_norm": 0.0014909004094079137,
-      "learning_rate": 0.00019999760613825714,
-      "loss": 46.0,
-      "step": 28817
-    },
-    {
-      "epoch": 2.203337347324961,
-      "grad_norm": 0.0007738404674455523,
-      "learning_rate": 0.0001999976059720549,
-      "loss": 46.0,
-      "step": 28818
-    },
-    {
-      "epoch": 2.203413804308351,
-      "grad_norm": 0.0008588539203628898,
-      "learning_rate": 0.0001999976058058469,
-      "loss": 46.0,
-      "step": 28819
-    },
-    {
-      "epoch": 2.2034902612917406,
-      "grad_norm": 0.0014501033583655953,
-      "learning_rate": 0.00019999760563963313,
-      "loss": 46.0,
-      "step": 28820
-    },
-    {
-      "epoch": 2.2035667182751304,
-      "grad_norm": 0.0010203676065430045,
-      "learning_rate": 0.0001999976054734136,
-      "loss": 46.0,
-      "step": 28821
-    },
-    {
-      "epoch": 2.20364317525852,
-      "grad_norm": 0.0009154072031378746,
-      "learning_rate": 0.00019999760530718831,
-      "loss": 46.0,
-      "step": 28822
-    },
-    {
-      "epoch": 2.20371963224191,
-      "grad_norm": 0.001322239637374878,
-      "learning_rate": 0.00019999760514095724,
-      "loss": 46.0,
-      "step": 28823
-    },
-    {
-      "epoch": 2.2037960892252997,
-      "grad_norm": 0.001761665684171021,
-      "learning_rate": 0.0001999976049747204,
-      "loss": 46.0,
-      "step": 28824
-    },
-    {
-      "epoch": 2.2038725462086894,
-      "grad_norm": 0.0014802577206864953,
-      "learning_rate": 0.00019999760480847778,
-      "loss": 46.0,
-      "step": 28825
-    },
-    {
-      "epoch": 2.203949003192079,
-      "grad_norm": 0.0011306633241474628,
-      "learning_rate": 0.00019999760464222942,
-      "loss": 46.0,
-      "step": 28826
-    },
-    {
-      "epoch": 2.204025460175469,
-      "grad_norm": 0.0028315153904259205,
-      "learning_rate": 0.00019999760447597526,
-      "loss": 46.0,
-      "step": 28827
-    },
-    {
-      "epoch": 2.2041019171588587,
-      "grad_norm": 0.00026699324371293187,
-      "learning_rate": 0.00019999760430971535,
-      "loss": 46.0,
-      "step": 28828
-    },
-    {
-      "epoch": 2.204178374142248,
-      "grad_norm": 0.0008715919102542102,
-      "learning_rate": 0.00019999760414344967,
-      "loss": 46.0,
-      "step": 28829
-    },
-    {
-      "epoch": 2.204254831125638,
-      "grad_norm": 0.0008248778758570552,
-      "learning_rate": 0.0001999976039771782,
-      "loss": 46.0,
-      "step": 28830
-    },
-    {
-      "epoch": 2.2043312881090276,
-      "grad_norm": 0.0011730906553566456,
-      "learning_rate": 0.00019999760381090098,
-      "loss": 46.0,
-      "step": 28831
-    },
-    {
-      "epoch": 2.2044077450924173,
-      "grad_norm": 0.0004987865104340017,
-      "learning_rate": 0.00019999760364461798,
-      "loss": 46.0,
-      "step": 28832
-    },
-    {
-      "epoch": 2.204484202075807,
-      "grad_norm": 0.0008527519530616701,
-      "learning_rate": 0.0001999976034783292,
-      "loss": 46.0,
-      "step": 28833
-    },
-    {
-      "epoch": 2.204560659059197,
-      "grad_norm": 0.001204493921250105,
-      "learning_rate": 0.00019999760331203468,
-      "loss": 46.0,
-      "step": 28834
-    },
-    {
-      "epoch": 2.2046371160425866,
-      "grad_norm": 0.000427691062213853,
-      "learning_rate": 0.00019999760314573438,
-      "loss": 46.0,
-      "step": 28835
-    },
-    {
-      "epoch": 2.2047135730259764,
-      "grad_norm": 0.001846226747147739,
-      "learning_rate": 0.0001999976029794283,
-      "loss": 46.0,
-      "step": 28836
-    },
-    {
-      "epoch": 2.204790030009366,
-      "grad_norm": 0.002269761636853218,
-      "learning_rate": 0.00019999760281311647,
-      "loss": 46.0,
-      "step": 28837
-    },
-    {
-      "epoch": 2.204866486992756,
-      "grad_norm": 0.0025457690935581923,
-      "learning_rate": 0.00019999760264679885,
-      "loss": 46.0,
-      "step": 28838
-    },
-    {
-      "epoch": 2.204942943976145,
-      "grad_norm": 0.0004558280052151531,
-      "learning_rate": 0.00019999760248047546,
-      "loss": 46.0,
-      "step": 28839
-    },
-    {
-      "epoch": 2.205019400959535,
-      "grad_norm": 0.0012941573513671756,
-      "learning_rate": 0.00019999760231414633,
-      "loss": 46.0,
-      "step": 28840
-    },
-    {
-      "epoch": 2.2050958579429247,
-      "grad_norm": 0.00124286487698555,
-      "learning_rate": 0.00019999760214781142,
-      "loss": 46.0,
-      "step": 28841
-    },
-    {
-      "epoch": 2.2051723149263145,
-      "grad_norm": 0.0010324915638193488,
-      "learning_rate": 0.00019999760198147074,
-      "loss": 46.0,
-      "step": 28842
-    },
-    {
-      "epoch": 2.2052487719097043,
-      "grad_norm": 0.0014269310049712658,
-      "learning_rate": 0.00019999760181512425,
-      "loss": 46.0,
-      "step": 28843
-    },
-    {
-      "epoch": 2.205325228893094,
-      "grad_norm": 0.002782462863251567,
-      "learning_rate": 0.00019999760164877202,
-      "loss": 46.0,
-      "step": 28844
-    },
-    {
-      "epoch": 2.2054016858764838,
-      "grad_norm": 0.0006852689548395574,
-      "learning_rate": 0.00019999760148241405,
-      "loss": 46.0,
-      "step": 28845
-    },
-    {
-      "epoch": 2.2054781428598735,
-      "grad_norm": 0.0009855644311755896,
-      "learning_rate": 0.00019999760131605027,
-      "loss": 46.0,
-      "step": 28846
-    },
-    {
-      "epoch": 2.2055545998432633,
-      "grad_norm": 0.0015077366260811687,
-      "learning_rate": 0.00019999760114968072,
-      "loss": 46.0,
-      "step": 28847
-    },
-    {
-      "epoch": 2.205631056826653,
-      "grad_norm": 0.0008704654173925519,
-      "learning_rate": 0.00019999760098330543,
-      "loss": 46.0,
-      "step": 28848
-    },
-    {
-      "epoch": 2.205707513810043,
-      "grad_norm": 0.0008795936009846628,
-      "learning_rate": 0.00019999760081692439,
-      "loss": 46.0,
-      "step": 28849
-    },
-    {
-      "epoch": 2.2057839707934326,
-      "grad_norm": 0.0019369556102901697,
-      "learning_rate": 0.00019999760065053752,
-      "loss": 46.0,
-      "step": 28850
-    },
-    {
-      "epoch": 2.205860427776822,
-      "grad_norm": 0.0008396393968723714,
-      "learning_rate": 0.0001999976004841449,
-      "loss": 46.0,
-      "step": 28851
-    },
-    {
-      "epoch": 2.2059368847602117,
-      "grad_norm": 0.0009077504510059953,
-      "learning_rate": 0.00019999760031774654,
-      "loss": 46.0,
-      "step": 28852
-    },
-    {
-      "epoch": 2.2060133417436014,
-      "grad_norm": 0.00030257628532126546,
-      "learning_rate": 0.0001999976001513424,
-      "loss": 46.0,
-      "step": 28853
-    },
-    {
-      "epoch": 2.206089798726991,
-      "grad_norm": 0.0011218987638130784,
-      "learning_rate": 0.00019999759998493247,
-      "loss": 46.0,
-      "step": 28854
-    },
-    {
-      "epoch": 2.206166255710381,
-      "grad_norm": 0.0005522216088138521,
-      "learning_rate": 0.00019999759981851676,
-      "loss": 46.0,
-      "step": 28855
-    },
-    {
-      "epoch": 2.2062427126937707,
-      "grad_norm": 0.0035300019662827253,
-      "learning_rate": 0.0001999975996520953,
-      "loss": 46.0,
-      "step": 28856
-    },
-    {
-      "epoch": 2.2063191696771605,
-      "grad_norm": 0.0007878244505263865,
-      "learning_rate": 0.00019999759948566807,
-      "loss": 46.0,
-      "step": 28857
-    },
-    {
-      "epoch": 2.2063956266605502,
-      "grad_norm": 0.00852157361805439,
-      "learning_rate": 0.00019999759931923507,
-      "loss": 46.0,
-      "step": 28858
-    },
-    {
-      "epoch": 2.20647208364394,
-      "grad_norm": 0.0022826530039310455,
-      "learning_rate": 0.00019999759915279632,
-      "loss": 46.0,
-      "step": 28859
-    },
-    {
-      "epoch": 2.2065485406273297,
-      "grad_norm": 0.0010222750715911388,
-      "learning_rate": 0.0001999975989863518,
-      "loss": 46.0,
-      "step": 28860
-    },
-    {
-      "epoch": 2.206624997610719,
-      "grad_norm": 0.0007364195771515369,
-      "learning_rate": 0.00019999759881990148,
-      "loss": 46.0,
-      "step": 28861
-    },
-    {
-      "epoch": 2.206701454594109,
-      "grad_norm": 0.0007557598291896284,
-      "learning_rate": 0.0001999975986534454,
-      "loss": 46.0,
-      "step": 28862
-    },
-    {
-      "epoch": 2.2067779115774986,
-      "grad_norm": 0.0010475916787981987,
-      "learning_rate": 0.00019999759848698357,
-      "loss": 46.0,
-      "step": 28863
-    },
-    {
-      "epoch": 2.2068543685608883,
-      "grad_norm": 0.0005715079605579376,
-      "learning_rate": 0.00019999759832051595,
-      "loss": 46.0,
-      "step": 28864
-    },
-    {
-      "epoch": 2.206930825544278,
-      "grad_norm": 0.0008115984383039176,
-      "learning_rate": 0.00019999759815404256,
-      "loss": 46.0,
-      "step": 28865
-    },
-    {
-      "epoch": 2.207007282527668,
-      "grad_norm": 0.0007341348100453615,
-      "learning_rate": 0.0001999975979875634,
-      "loss": 46.0,
-      "step": 28866
-    },
-    {
-      "epoch": 2.2070837395110576,
-      "grad_norm": 0.0010737431002780795,
-      "learning_rate": 0.0001999975978210785,
-      "loss": 46.0,
-      "step": 28867
-    },
-    {
-      "epoch": 2.2071601964944474,
-      "grad_norm": 0.0015940858284011483,
-      "learning_rate": 0.0001999975976545878,
-      "loss": 46.0,
-      "step": 28868
-    },
-    {
-      "epoch": 2.207236653477837,
-      "grad_norm": 0.0010333394166082144,
-      "learning_rate": 0.00019999759748809135,
-      "loss": 46.0,
-      "step": 28869
-    },
-    {
-      "epoch": 2.207313110461227,
-      "grad_norm": 0.0005897369119338691,
-      "learning_rate": 0.00019999759732158913,
-      "loss": 46.0,
-      "step": 28870
-    },
-    {
-      "epoch": 2.2073895674446167,
-      "grad_norm": 0.0035926199052482843,
-      "learning_rate": 0.0001999975971550811,
-      "loss": 46.0,
-      "step": 28871
-    },
-    {
-      "epoch": 2.207466024428006,
-      "grad_norm": 0.0009241283987648785,
-      "learning_rate": 0.00019999759698856735,
-      "loss": 46.0,
-      "step": 28872
-    },
-    {
-      "epoch": 2.2075424814113958,
-      "grad_norm": 0.0006126143271103501,
-      "learning_rate": 0.00019999759682204783,
-      "loss": 46.0,
-      "step": 28873
-    },
-    {
-      "epoch": 2.2076189383947855,
-      "grad_norm": 0.00040630382136441767,
-      "learning_rate": 0.00019999759665552248,
-      "loss": 46.0,
-      "step": 28874
-    },
-    {
-      "epoch": 2.2076953953781753,
-      "grad_norm": 0.0035690891090780497,
-      "learning_rate": 0.00019999759648899144,
-      "loss": 46.0,
-      "step": 28875
-    },
-    {
-      "epoch": 2.207771852361565,
-      "grad_norm": 0.0008592213271185756,
-      "learning_rate": 0.0001999975963224546,
-      "loss": 46.0,
-      "step": 28876
-    },
-    {
-      "epoch": 2.207848309344955,
-      "grad_norm": 0.0007553239120170474,
-      "learning_rate": 0.00019999759615591195,
-      "loss": 46.0,
-      "step": 28877
-    },
-    {
-      "epoch": 2.2079247663283446,
-      "grad_norm": 0.0004961151862516999,
-      "learning_rate": 0.0001999975959893636,
-      "loss": 46.0,
-      "step": 28878
-    },
-    {
-      "epoch": 2.2080012233117343,
-      "grad_norm": 0.0014502119738608599,
-      "learning_rate": 0.00019999759582280943,
-      "loss": 46.0,
-      "step": 28879
-    },
-    {
-      "epoch": 2.208077680295124,
-      "grad_norm": 0.0011711064726114273,
-      "learning_rate": 0.00019999759565624952,
-      "loss": 46.0,
-      "step": 28880
-    },
-    {
-      "epoch": 2.208154137278514,
-      "grad_norm": 0.00160564377438277,
-      "learning_rate": 0.0001999975954896838,
-      "loss": 46.0,
-      "step": 28881
-    },
-    {
-      "epoch": 2.2082305942619036,
-      "grad_norm": 0.001099162152968347,
-      "learning_rate": 0.00019999759532311236,
-      "loss": 46.0,
-      "step": 28882
-    },
-    {
-      "epoch": 2.208307051245293,
-      "grad_norm": 0.004697258118540049,
-      "learning_rate": 0.00019999759515653513,
-      "loss": 46.0,
-      "step": 28883
-    },
-    {
-      "epoch": 2.2083835082286827,
-      "grad_norm": 0.0012617643224075437,
-      "learning_rate": 0.00019999759498995215,
-      "loss": 46.0,
-      "step": 28884
-    },
-    {
-      "epoch": 2.2084599652120724,
-      "grad_norm": 0.0015428911428898573,
-      "learning_rate": 0.00019999759482336335,
-      "loss": 46.0,
-      "step": 28885
-    },
-    {
-      "epoch": 2.208536422195462,
-      "grad_norm": 0.00045823512482456863,
-      "learning_rate": 0.00019999759465676883,
-      "loss": 46.0,
-      "step": 28886
-    },
-    {
-      "epoch": 2.208612879178852,
-      "grad_norm": 0.0012080523883923888,
-      "learning_rate": 0.0001999975944901685,
-      "loss": 46.0,
-      "step": 28887
-    },
-    {
-      "epoch": 2.2086893361622417,
-      "grad_norm": 0.0011397481430321932,
-      "learning_rate": 0.00019999759432356244,
-      "loss": 46.0,
-      "step": 28888
-    },
-    {
-      "epoch": 2.2087657931456315,
-      "grad_norm": 0.0012885655742138624,
-      "learning_rate": 0.0001999975941569506,
-      "loss": 46.0,
-      "step": 28889
-    },
-    {
-      "epoch": 2.2088422501290212,
-      "grad_norm": 0.00044631215860135853,
-      "learning_rate": 0.00019999759399033298,
-      "loss": 46.0,
-      "step": 28890
-    },
-    {
-      "epoch": 2.208918707112411,
-      "grad_norm": 0.0004790983221028,
-      "learning_rate": 0.0001999975938237096,
-      "loss": 46.0,
-      "step": 28891
-    },
-    {
-      "epoch": 2.2089951640958008,
-      "grad_norm": 0.0013729401398450136,
-      "learning_rate": 0.00019999759365708043,
-      "loss": 46.0,
-      "step": 28892
-    },
-    {
-      "epoch": 2.2090716210791905,
-      "grad_norm": 0.0013881829800084233,
-      "learning_rate": 0.0001999975934904455,
-      "loss": 46.0,
-      "step": 28893
-    },
-    {
-      "epoch": 2.20914807806258,
-      "grad_norm": 0.00031469701207242906,
-      "learning_rate": 0.00019999759332380484,
-      "loss": 46.0,
-      "step": 28894
-    },
-    {
-      "epoch": 2.2092245350459696,
-      "grad_norm": 0.0004291992518119514,
-      "learning_rate": 0.00019999759315715836,
-      "loss": 46.0,
-      "step": 28895
-    },
-    {
-      "epoch": 2.2093009920293594,
-      "grad_norm": 0.0068331751972436905,
-      "learning_rate": 0.00019999759299050613,
-      "loss": 46.0,
-      "step": 28896
-    },
-    {
-      "epoch": 2.209377449012749,
-      "grad_norm": 0.000925276952330023,
-      "learning_rate": 0.00019999759282384813,
-      "loss": 46.0,
-      "step": 28897
-    },
-    {
-      "epoch": 2.209453905996139,
-      "grad_norm": 0.00033266397076658905,
-      "learning_rate": 0.00019999759265718436,
-      "loss": 46.0,
-      "step": 28898
-    },
-    {
-      "epoch": 2.2095303629795287,
-      "grad_norm": 0.001193828647956252,
-      "learning_rate": 0.0001999975924905148,
-      "loss": 46.0,
-      "step": 28899
-    },
-    {
-      "epoch": 2.2096068199629184,
-      "grad_norm": 0.0005277470336295664,
-      "learning_rate": 0.00019999759232383952,
-      "loss": 46.0,
-      "step": 28900
-    },
-    {
-      "epoch": 2.209683276946308,
-      "grad_norm": 0.0014322582865133882,
-      "learning_rate": 0.00019999759215715842,
-      "loss": 46.0,
-      "step": 28901
-    },
-    {
-      "epoch": 2.209759733929698,
-      "grad_norm": 0.0012266046833246946,
-      "learning_rate": 0.00019999759199047158,
-      "loss": 46.0,
-      "step": 28902
-    },
-    {
-      "epoch": 2.2098361909130877,
-      "grad_norm": 0.002536962041631341,
-      "learning_rate": 0.00019999759182377897,
-      "loss": 46.0,
-      "step": 28903
-    },
-    {
-      "epoch": 2.209912647896477,
-      "grad_norm": 0.0010576860513538122,
-      "learning_rate": 0.00019999759165708058,
-      "loss": 46.0,
-      "step": 28904
-    },
-    {
-      "epoch": 2.2099891048798668,
-      "grad_norm": 0.0016458620084449649,
-      "learning_rate": 0.00019999759149037644,
-      "loss": 46.0,
-      "step": 28905
-    },
-    {
-      "epoch": 2.2100655618632565,
-      "grad_norm": 0.0017055054195225239,
-      "learning_rate": 0.0001999975913236665,
-      "loss": 46.0,
-      "step": 28906
-    },
-    {
-      "epoch": 2.2101420188466463,
-      "grad_norm": 0.0030669074039906263,
-      "learning_rate": 0.0001999975911569508,
-      "loss": 46.0,
-      "step": 28907
-    },
-    {
-      "epoch": 2.210218475830036,
-      "grad_norm": 0.0017010248266160488,
-      "learning_rate": 0.00019999759099022937,
-      "loss": 46.0,
-      "step": 28908
-    },
-    {
-      "epoch": 2.210294932813426,
-      "grad_norm": 0.0011826349655166268,
-      "learning_rate": 0.00019999759082350212,
-      "loss": 46.0,
-      "step": 28909
-    },
-    {
-      "epoch": 2.2103713897968156,
-      "grad_norm": 0.0007568614673800766,
-      "learning_rate": 0.00019999759065676912,
-      "loss": 46.0,
-      "step": 28910
-    },
-    {
-      "epoch": 2.2104478467802053,
-      "grad_norm": 0.0037644265685230494,
-      "learning_rate": 0.00019999759049003034,
-      "loss": 46.0,
-      "step": 28911
-    },
-    {
-      "epoch": 2.210524303763595,
-      "grad_norm": 0.0008171837544068694,
-      "learning_rate": 0.0001999975903232858,
-      "loss": 46.0,
-      "step": 28912
-    },
-    {
-      "epoch": 2.210600760746985,
-      "grad_norm": 0.0005448766751214862,
-      "learning_rate": 0.0001999975901565355,
-      "loss": 46.0,
-      "step": 28913
-    },
-    {
-      "epoch": 2.2106772177303746,
-      "grad_norm": 0.0006229626596905291,
-      "learning_rate": 0.00019999758998977943,
-      "loss": 46.0,
-      "step": 28914
-    },
-    {
-      "epoch": 2.2107536747137644,
-      "grad_norm": 0.0005808335263282061,
-      "learning_rate": 0.00019999758982301757,
-      "loss": 46.0,
-      "step": 28915
-    },
-    {
-      "epoch": 2.2108301316971537,
-      "grad_norm": 0.0008625047048553824,
-      "learning_rate": 0.00019999758965624995,
-      "loss": 46.0,
-      "step": 28916
-    },
-    {
-      "epoch": 2.2109065886805435,
-      "grad_norm": 0.0015434701927006245,
-      "learning_rate": 0.00019999758948947656,
-      "loss": 46.0,
-      "step": 28917
-    },
-    {
-      "epoch": 2.210983045663933,
-      "grad_norm": 0.0006651518633589149,
-      "learning_rate": 0.0001999975893226974,
-      "loss": 46.0,
-      "step": 28918
-    },
-    {
-      "epoch": 2.211059502647323,
-      "grad_norm": 0.0009781579719856381,
-      "learning_rate": 0.0001999975891559125,
-      "loss": 46.0,
-      "step": 28919
-    },
-    {
-      "epoch": 2.2111359596307127,
-      "grad_norm": 0.0038510675076395273,
-      "learning_rate": 0.0001999975889891218,
-      "loss": 46.0,
-      "step": 28920
-    },
-    {
-      "epoch": 2.2112124166141025,
-      "grad_norm": 0.0006015028338879347,
-      "learning_rate": 0.00019999758882232533,
-      "loss": 46.0,
-      "step": 28921
-    },
-    {
-      "epoch": 2.2112888735974923,
-      "grad_norm": 0.00036297718179412186,
-      "learning_rate": 0.0001999975886555231,
-      "loss": 46.0,
-      "step": 28922
-    },
-    {
-      "epoch": 2.211365330580882,
-      "grad_norm": 0.0009397089597769082,
-      "learning_rate": 0.00019999758848871508,
-      "loss": 46.0,
-      "step": 28923
-    },
-    {
-      "epoch": 2.211441787564272,
-      "grad_norm": 0.0017561400309205055,
-      "learning_rate": 0.00019999758832190133,
-      "loss": 46.0,
-      "step": 28924
-    },
-    {
-      "epoch": 2.2115182445476615,
-      "grad_norm": 0.0021246513351798058,
-      "learning_rate": 0.00019999758815508178,
-      "loss": 46.0,
-      "step": 28925
-    },
-    {
-      "epoch": 2.211594701531051,
-      "grad_norm": 0.0005786185502074659,
-      "learning_rate": 0.00019999758798825646,
-      "loss": 46.0,
-      "step": 28926
-    },
-    {
-      "epoch": 2.2116711585144406,
-      "grad_norm": 0.0005074609653092921,
-      "learning_rate": 0.0001999975878214254,
-      "loss": 46.0,
-      "step": 28927
-    },
-    {
-      "epoch": 2.2117476154978304,
-      "grad_norm": 0.0008231866522692144,
-      "learning_rate": 0.00019999758765458853,
-      "loss": 46.0,
-      "step": 28928
-    },
-    {
-      "epoch": 2.21182407248122,
-      "grad_norm": 0.00392834423109889,
-      "learning_rate": 0.0001999975874877459,
-      "loss": 46.0,
-      "step": 28929
-    },
-    {
-      "epoch": 2.21190052946461,
-      "grad_norm": 0.0005317357718013227,
-      "learning_rate": 0.00019999758732089753,
-      "loss": 46.0,
-      "step": 28930
-    },
-    {
-      "epoch": 2.2119769864479997,
-      "grad_norm": 0.0017007219139486551,
-      "learning_rate": 0.00019999758715404337,
-      "loss": 46.0,
-      "step": 28931
-    },
-    {
-      "epoch": 2.2120534434313894,
-      "grad_norm": 0.002958127995952964,
-      "learning_rate": 0.00019999758698718346,
-      "loss": 46.0,
-      "step": 28932
-    },
-    {
-      "epoch": 2.212129900414779,
-      "grad_norm": 0.00095657620113343,
-      "learning_rate": 0.00019999758682031775,
-      "loss": 46.0,
-      "step": 28933
-    },
-    {
-      "epoch": 2.212206357398169,
-      "grad_norm": 0.0016081796493381262,
-      "learning_rate": 0.0001999975866534463,
-      "loss": 46.0,
-      "step": 28934
-    },
-    {
-      "epoch": 2.2122828143815587,
-      "grad_norm": 0.0012840284034609795,
-      "learning_rate": 0.00019999758648656904,
-      "loss": 46.0,
-      "step": 28935
-    },
-    {
-      "epoch": 2.2123592713649485,
-      "grad_norm": 0.0019641744438558817,
-      "learning_rate": 0.00019999758631968604,
-      "loss": 46.0,
-      "step": 28936
-    },
-    {
-      "epoch": 2.2124357283483382,
-      "grad_norm": 0.0012826048769056797,
-      "learning_rate": 0.00019999758615279727,
-      "loss": 46.0,
-      "step": 28937
-    },
-    {
-      "epoch": 2.2125121853317276,
-      "grad_norm": 0.0023787023965269327,
-      "learning_rate": 0.00019999758598590272,
-      "loss": 46.0,
-      "step": 28938
-    },
-    {
-      "epoch": 2.2125886423151173,
-      "grad_norm": 0.0016007553786039352,
-      "learning_rate": 0.0001999975858190024,
-      "loss": 46.0,
-      "step": 28939
-    },
-    {
-      "epoch": 2.212665099298507,
-      "grad_norm": 0.0034340412821620703,
-      "learning_rate": 0.00019999758565209633,
-      "loss": 46.0,
-      "step": 28940
-    },
-    {
-      "epoch": 2.212741556281897,
-      "grad_norm": 0.0003411919460631907,
-      "learning_rate": 0.0001999975854851845,
-      "loss": 46.0,
-      "step": 28941
-    },
-    {
-      "epoch": 2.2128180132652866,
-      "grad_norm": 0.0008884069975465536,
-      "learning_rate": 0.00019999758531826688,
-      "loss": 46.0,
-      "step": 28942
-    },
-    {
-      "epoch": 2.2128944702486764,
-      "grad_norm": 0.0008395428885705769,
-      "learning_rate": 0.0001999975851513435,
-      "loss": 46.0,
-      "step": 28943
-    },
-    {
-      "epoch": 2.212970927232066,
-      "grad_norm": 0.0028906133957207203,
-      "learning_rate": 0.00019999758498441433,
-      "loss": 46.0,
-      "step": 28944
-    },
-    {
-      "epoch": 2.213047384215456,
-      "grad_norm": 0.0013134530745446682,
-      "learning_rate": 0.0001999975848174794,
-      "loss": 46.0,
-      "step": 28945
-    },
-    {
-      "epoch": 2.2131238411988456,
-      "grad_norm": 0.001592721207998693,
-      "learning_rate": 0.0001999975846505387,
-      "loss": 46.0,
-      "step": 28946
-    },
-    {
-      "epoch": 2.2132002981822354,
-      "grad_norm": 0.0007467641262337565,
-      "learning_rate": 0.00019999758448359224,
-      "loss": 46.0,
-      "step": 28947
-    },
-    {
-      "epoch": 2.2132767551656247,
-      "grad_norm": 0.00161384092643857,
-      "learning_rate": 0.00019999758431663998,
-      "loss": 46.0,
-      "step": 28948
-    },
-    {
-      "epoch": 2.2133532121490145,
-      "grad_norm": 0.0009858588455244899,
-      "learning_rate": 0.00019999758414968198,
-      "loss": 46.0,
-      "step": 28949
-    },
-    {
-      "epoch": 2.2134296691324042,
-      "grad_norm": 0.00833367183804512,
-      "learning_rate": 0.0001999975839827182,
-      "loss": 46.0,
-      "step": 28950
-    },
-    {
-      "epoch": 2.213506126115794,
-      "grad_norm": 0.0013271848438307643,
-      "learning_rate": 0.00019999758381574866,
-      "loss": 46.0,
-      "step": 28951
-    },
-    {
-      "epoch": 2.2135825830991838,
-      "grad_norm": 0.0022770448122173548,
-      "learning_rate": 0.00019999758364877334,
-      "loss": 46.0,
-      "step": 28952
-    },
-    {
-      "epoch": 2.2136590400825735,
-      "grad_norm": 0.002240579342469573,
-      "learning_rate": 0.00019999758348179228,
-      "loss": 46.0,
-      "step": 28953
-    },
-    {
-      "epoch": 2.2137354970659633,
-      "grad_norm": 0.0006508915685117245,
-      "learning_rate": 0.00019999758331480544,
-      "loss": 46.0,
-      "step": 28954
-    },
-    {
-      "epoch": 2.213811954049353,
-      "grad_norm": 0.0009769409662112594,
-      "learning_rate": 0.0001999975831478128,
-      "loss": 46.0,
-      "step": 28955
-    },
-    {
-      "epoch": 2.213888411032743,
-      "grad_norm": 0.0014440763043239713,
-      "learning_rate": 0.0001999975829808144,
-      "loss": 46.0,
-      "step": 28956
-    },
-    {
-      "epoch": 2.2139648680161326,
-      "grad_norm": 0.0008397704223170877,
-      "learning_rate": 0.00019999758281381028,
-      "loss": 46.0,
-      "step": 28957
-    },
-    {
-      "epoch": 2.2140413249995223,
-      "grad_norm": 0.0013478489127010107,
-      "learning_rate": 0.00019999758264680032,
-      "loss": 46.0,
-      "step": 28958
-    },
-    {
-      "epoch": 2.214117781982912,
-      "grad_norm": 0.001117086154408753,
-      "learning_rate": 0.0001999975824797846,
-      "loss": 46.0,
-      "step": 28959
-    },
-    {
-      "epoch": 2.2141942389663014,
-      "grad_norm": 0.0009121336042881012,
-      "learning_rate": 0.00019999758231276316,
-      "loss": 46.0,
-      "step": 28960
-    },
-    {
-      "epoch": 2.214270695949691,
-      "grad_norm": 0.001009747269563377,
-      "learning_rate": 0.00019999758214573593,
-      "loss": 46.0,
-      "step": 28961
-    },
-    {
-      "epoch": 2.214347152933081,
-      "grad_norm": 0.0005775009631179273,
-      "learning_rate": 0.0001999975819787029,
-      "loss": 46.0,
-      "step": 28962
-    },
-    {
-      "epoch": 2.2144236099164707,
-      "grad_norm": 0.004599041771143675,
-      "learning_rate": 0.00019999758181166416,
-      "loss": 46.0,
-      "step": 28963
-    },
-    {
-      "epoch": 2.2145000668998605,
-      "grad_norm": 0.0018458062550053,
-      "learning_rate": 0.0001999975816446196,
-      "loss": 46.0,
-      "step": 28964
-    },
-    {
-      "epoch": 2.21457652388325,
-      "grad_norm": 0.0008384262910112739,
-      "learning_rate": 0.0001999975814775693,
-      "loss": 46.0,
-      "step": 28965
-    },
-    {
-      "epoch": 2.21465298086664,
-      "grad_norm": 0.0021188645623624325,
-      "learning_rate": 0.0001999975813105132,
-      "loss": 46.0,
-      "step": 28966
-    },
-    {
-      "epoch": 2.2147294378500297,
-      "grad_norm": 0.0008339123451150954,
-      "learning_rate": 0.00019999758114345136,
-      "loss": 46.0,
-      "step": 28967
-    },
-    {
-      "epoch": 2.2148058948334195,
-      "grad_norm": 0.001448485883884132,
-      "learning_rate": 0.00019999758097638372,
-      "loss": 46.0,
-      "step": 28968
-    },
-    {
-      "epoch": 2.2148823518168093,
-      "grad_norm": 0.0024356681387871504,
-      "learning_rate": 0.00019999758080931033,
-      "loss": 46.0,
-      "step": 28969
-    },
-    {
-      "epoch": 2.2149588088001986,
-      "grad_norm": 0.0020377894397825003,
-      "learning_rate": 0.00019999758064223117,
-      "loss": 46.0,
-      "step": 28970
-    },
-    {
-      "epoch": 2.2150352657835883,
-      "grad_norm": 0.0015785840805619955,
-      "learning_rate": 0.00019999758047514624,
-      "loss": 46.0,
-      "step": 28971
-    },
-    {
-      "epoch": 2.215111722766978,
-      "grad_norm": 0.0015167194651439786,
-      "learning_rate": 0.00019999758030805556,
-      "loss": 46.0,
-      "step": 28972
-    },
-    {
-      "epoch": 2.215188179750368,
-      "grad_norm": 0.0036254897713661194,
-      "learning_rate": 0.00019999758014095908,
-      "loss": 46.0,
-      "step": 28973
-    },
-    {
-      "epoch": 2.2152646367337576,
-      "grad_norm": 0.0017698084702715278,
-      "learning_rate": 0.00019999757997385685,
-      "loss": 46.0,
-      "step": 28974
-    },
-    {
-      "epoch": 2.2153410937171474,
-      "grad_norm": 0.0010463435901328921,
-      "learning_rate": 0.00019999757980674883,
-      "loss": 46.0,
-      "step": 28975
-    },
-    {
-      "epoch": 2.215417550700537,
-      "grad_norm": 0.001757300109602511,
-      "learning_rate": 0.00019999757963963505,
-      "loss": 46.0,
-      "step": 28976
-    },
-    {
-      "epoch": 2.215494007683927,
-      "grad_norm": 0.0006821682909503579,
-      "learning_rate": 0.0001999975794725155,
-      "loss": 46.0,
-      "step": 28977
-    },
-    {
-      "epoch": 2.2155704646673167,
-      "grad_norm": 0.005596705246716738,
-      "learning_rate": 0.0001999975793053902,
-      "loss": 46.0,
-      "step": 28978
-    },
-    {
-      "epoch": 2.2156469216507064,
-      "grad_norm": 0.0012318710796535015,
-      "learning_rate": 0.00019999757913825912,
-      "loss": 46.0,
-      "step": 28979
-    },
-    {
-      "epoch": 2.215723378634096,
-      "grad_norm": 0.0012017672415822744,
-      "learning_rate": 0.00019999757897112225,
-      "loss": 46.0,
-      "step": 28980
-    },
-    {
-      "epoch": 2.215799835617486,
-      "grad_norm": 0.0015774633502587676,
-      "learning_rate": 0.00019999757880397962,
-      "loss": 46.0,
-      "step": 28981
-    },
-    {
-      "epoch": 2.2158762926008753,
-      "grad_norm": 0.002926749177277088,
-      "learning_rate": 0.00019999757863683123,
-      "loss": 46.0,
-      "step": 28982
-    },
-    {
-      "epoch": 2.215952749584265,
-      "grad_norm": 0.0011569702764973044,
-      "learning_rate": 0.00019999757846967707,
-      "loss": 46.0,
-      "step": 28983
-    },
-    {
-      "epoch": 2.216029206567655,
-      "grad_norm": 0.0007350613013841212,
-      "learning_rate": 0.00019999757830251716,
-      "loss": 46.0,
-      "step": 28984
-    },
-    {
-      "epoch": 2.2161056635510445,
-      "grad_norm": 0.0005084079457446933,
-      "learning_rate": 0.00019999757813535146,
-      "loss": 46.0,
-      "step": 28985
-    },
-    {
-      "epoch": 2.2161821205344343,
-      "grad_norm": 0.0005607547936961055,
-      "learning_rate": 0.00019999757796817998,
-      "loss": 46.0,
-      "step": 28986
-    },
-    {
-      "epoch": 2.216258577517824,
-      "grad_norm": 0.0009682882810011506,
-      "learning_rate": 0.00019999757780100275,
-      "loss": 46.0,
-      "step": 28987
-    },
-    {
-      "epoch": 2.216335034501214,
-      "grad_norm": 0.00064097746508196,
-      "learning_rate": 0.00019999757763381975,
-      "loss": 46.0,
-      "step": 28988
-    },
-    {
-      "epoch": 2.2164114914846036,
-      "grad_norm": 0.002359879668802023,
-      "learning_rate": 0.00019999757746663095,
-      "loss": 46.0,
-      "step": 28989
-    },
-    {
-      "epoch": 2.2164879484679934,
-      "grad_norm": 0.0016890749102458358,
-      "learning_rate": 0.00019999757729943643,
-      "loss": 46.0,
-      "step": 28990
-    },
-    {
-      "epoch": 2.216564405451383,
-      "grad_norm": 0.0005565067403949797,
-      "learning_rate": 0.0001999975771322361,
-      "loss": 46.0,
-      "step": 28991
-    },
-    {
-      "epoch": 2.2166408624347724,
-      "grad_norm": 0.0008554342784918845,
-      "learning_rate": 0.00019999757696503002,
-      "loss": 46.0,
-      "step": 28992
-    },
-    {
-      "epoch": 2.216717319418162,
-      "grad_norm": 0.002095424570143223,
-      "learning_rate": 0.00019999757679781816,
-      "loss": 46.0,
-      "step": 28993
-    },
-    {
-      "epoch": 2.216793776401552,
-      "grad_norm": 0.0017710519023239613,
-      "learning_rate": 0.00019999757663060054,
-      "loss": 46.0,
-      "step": 28994
-    },
-    {
-      "epoch": 2.2168702333849417,
-      "grad_norm": 0.003226671600714326,
-      "learning_rate": 0.00019999757646337713,
-      "loss": 46.0,
-      "step": 28995
-    },
-    {
-      "epoch": 2.2169466903683315,
-      "grad_norm": 0.0006709518493153155,
-      "learning_rate": 0.00019999757629614797,
-      "loss": 46.0,
-      "step": 28996
-    },
-    {
-      "epoch": 2.2170231473517212,
-      "grad_norm": 0.0033667893148958683,
-      "learning_rate": 0.00019999757612891307,
-      "loss": 46.0,
-      "step": 28997
-    },
-    {
-      "epoch": 2.217099604335111,
-      "grad_norm": 0.0008563104784116149,
-      "learning_rate": 0.00019999757596167233,
-      "loss": 46.0,
-      "step": 28998
-    },
-    {
-      "epoch": 2.2171760613185008,
-      "grad_norm": 0.0011014026822522283,
-      "learning_rate": 0.00019999757579442585,
-      "loss": 46.0,
-      "step": 28999
-    },
-    {
-      "epoch": 2.2172525183018905,
-      "grad_norm": 0.0014451269526034594,
-      "learning_rate": 0.00019999757562717363,
-      "loss": 46.0,
-      "step": 29000
-    },
-    {
-      "epoch": 2.2173289752852803,
-      "grad_norm": 0.0015442516887560487,
-      "learning_rate": 0.00019999757545991563,
-      "loss": 46.0,
-      "step": 29001
-    },
-    {
-      "epoch": 2.21740543226867,
-      "grad_norm": 0.0016162347747012973,
-      "learning_rate": 0.00019999757529265183,
-      "loss": 46.0,
-      "step": 29002
-    },
-    {
-      "epoch": 2.2174818892520594,
-      "grad_norm": 0.00045074548688717186,
-      "learning_rate": 0.00019999757512538228,
-      "loss": 46.0,
-      "step": 29003
-    },
-    {
-      "epoch": 2.217558346235449,
-      "grad_norm": 0.001895284396596253,
-      "learning_rate": 0.000199997574958107,
-      "loss": 46.0,
-      "step": 29004
-    },
-    {
-      "epoch": 2.217634803218839,
-      "grad_norm": 0.0005610543303191662,
-      "learning_rate": 0.0001999975747908259,
-      "loss": 46.0,
-      "step": 29005
-    },
-    {
-      "epoch": 2.2177112602022286,
-      "grad_norm": 0.0018726419657468796,
-      "learning_rate": 0.00019999757462353906,
-      "loss": 46.0,
-      "step": 29006
-    },
-    {
-      "epoch": 2.2177877171856184,
-      "grad_norm": 0.0012618129840120673,
-      "learning_rate": 0.0001999975744562464,
-      "loss": 46.0,
-      "step": 29007
-    },
-    {
-      "epoch": 2.217864174169008,
-      "grad_norm": 0.002981960540637374,
-      "learning_rate": 0.00019999757428894804,
-      "loss": 46.0,
-      "step": 29008
-    },
-    {
-      "epoch": 2.217940631152398,
-      "grad_norm": 0.005760539788752794,
-      "learning_rate": 0.00019999757412164385,
-      "loss": 46.0,
-      "step": 29009
-    },
-    {
-      "epoch": 2.2180170881357877,
-      "grad_norm": 0.0030491300858557224,
-      "learning_rate": 0.00019999757395433392,
-      "loss": 46.0,
-      "step": 29010
-    },
-    {
-      "epoch": 2.2180935451191774,
-      "grad_norm": 0.0012788968160748482,
-      "learning_rate": 0.00019999757378701821,
-      "loss": 46.0,
-      "step": 29011
-    },
-    {
-      "epoch": 2.218170002102567,
-      "grad_norm": 0.0007759974105283618,
-      "learning_rate": 0.00019999757361969676,
-      "loss": 46.0,
-      "step": 29012
-    },
-    {
-      "epoch": 2.218246459085957,
-      "grad_norm": 0.00045431137550622225,
-      "learning_rate": 0.00019999757345236948,
-      "loss": 46.0,
-      "step": 29013
-    },
-    {
-      "epoch": 2.2183229160693463,
-      "grad_norm": 0.0020434728357940912,
-      "learning_rate": 0.0001999975732850365,
-      "loss": 46.0,
-      "step": 29014
-    },
-    {
-      "epoch": 2.218399373052736,
-      "grad_norm": 0.0011778234038501978,
-      "learning_rate": 0.0001999975731176977,
-      "loss": 46.0,
-      "step": 29015
-    },
-    {
-      "epoch": 2.218475830036126,
-      "grad_norm": 0.0006551499245688319,
-      "learning_rate": 0.00019999757295035314,
-      "loss": 46.0,
-      "step": 29016
-    },
-    {
-      "epoch": 2.2185522870195156,
-      "grad_norm": 0.0012449778150767088,
-      "learning_rate": 0.00019999757278300285,
-      "loss": 46.0,
-      "step": 29017
-    },
-    {
-      "epoch": 2.2186287440029053,
-      "grad_norm": 0.0014794125454500318,
-      "learning_rate": 0.00019999757261564676,
-      "loss": 46.0,
-      "step": 29018
-    },
-    {
-      "epoch": 2.218705200986295,
-      "grad_norm": 0.001708697876892984,
-      "learning_rate": 0.0001999975724482849,
-      "loss": 46.0,
-      "step": 29019
-    },
-    {
-      "epoch": 2.218781657969685,
-      "grad_norm": 0.0008790239808149636,
-      "learning_rate": 0.00019999757228091728,
-      "loss": 46.0,
-      "step": 29020
-    },
-    {
-      "epoch": 2.2188581149530746,
-      "grad_norm": 0.0005551482900045812,
-      "learning_rate": 0.0001999975721135439,
-      "loss": 46.0,
-      "step": 29021
-    },
-    {
-      "epoch": 2.2189345719364644,
-      "grad_norm": 0.0006564059294760227,
-      "learning_rate": 0.00019999757194616473,
-      "loss": 46.0,
-      "step": 29022
-    },
-    {
-      "epoch": 2.219011028919854,
-      "grad_norm": 0.0008292549755424261,
-      "learning_rate": 0.00019999757177877978,
-      "loss": 46.0,
-      "step": 29023
-    },
-    {
-      "epoch": 2.219087485903244,
-      "grad_norm": 0.0009482022142037749,
-      "learning_rate": 0.00019999757161138907,
-      "loss": 46.0,
-      "step": 29024
-    },
-    {
-      "epoch": 2.219163942886633,
-      "grad_norm": 0.0020089088939130306,
-      "learning_rate": 0.0001999975714439926,
-      "loss": 46.0,
-      "step": 29025
-    },
-    {
-      "epoch": 2.219240399870023,
-      "grad_norm": 0.00059704011073336,
-      "learning_rate": 0.00019999757127659037,
-      "loss": 46.0,
-      "step": 29026
-    },
-    {
-      "epoch": 2.2193168568534127,
-      "grad_norm": 0.0024245691020041704,
-      "learning_rate": 0.00019999757110918237,
-      "loss": 46.0,
-      "step": 29027
-    },
-    {
-      "epoch": 2.2193933138368025,
-      "grad_norm": 0.0011015565833076835,
-      "learning_rate": 0.00019999757094176857,
-      "loss": 46.0,
-      "step": 29028
-    },
-    {
-      "epoch": 2.2194697708201923,
-      "grad_norm": 0.0008510947227478027,
-      "learning_rate": 0.00019999757077434903,
-      "loss": 46.0,
-      "step": 29029
-    },
-    {
-      "epoch": 2.219546227803582,
-      "grad_norm": 0.0020030790474265814,
-      "learning_rate": 0.0001999975706069237,
-      "loss": 46.0,
-      "step": 29030
-    },
-    {
-      "epoch": 2.219622684786972,
-      "grad_norm": 0.0009097789879888296,
-      "learning_rate": 0.00019999757043949261,
-      "loss": 46.0,
-      "step": 29031
-    },
-    {
-      "epoch": 2.2196991417703615,
-      "grad_norm": 0.0007160550449043512,
-      "learning_rate": 0.00019999757027205575,
-      "loss": 46.0,
-      "step": 29032
-    },
-    {
-      "epoch": 2.2197755987537513,
-      "grad_norm": 0.0007419795147143304,
-      "learning_rate": 0.00019999757010461314,
-      "loss": 46.0,
-      "step": 29033
-    },
-    {
-      "epoch": 2.219852055737141,
-      "grad_norm": 0.001492612063884735,
-      "learning_rate": 0.00019999756993716475,
-      "loss": 46.0,
-      "step": 29034
-    },
-    {
-      "epoch": 2.2199285127205304,
-      "grad_norm": 0.0010232396889477968,
-      "learning_rate": 0.00019999756976971057,
-      "loss": 46.0,
-      "step": 29035
-    },
-    {
-      "epoch": 2.22000496970392,
-      "grad_norm": 0.0006374738295562565,
-      "learning_rate": 0.00019999756960225064,
-      "loss": 46.0,
-      "step": 29036
-    },
-    {
-      "epoch": 2.22008142668731,
-      "grad_norm": 0.0018267721170559525,
-      "learning_rate": 0.0001999975694347849,
-      "loss": 46.0,
-      "step": 29037
-    },
-    {
-      "epoch": 2.2201578836706997,
-      "grad_norm": 0.001536606578156352,
-      "learning_rate": 0.00019999756926731345,
-      "loss": 46.0,
-      "step": 29038
-    },
-    {
-      "epoch": 2.2202343406540894,
-      "grad_norm": 0.000566108908969909,
-      "learning_rate": 0.00019999756909983623,
-      "loss": 46.0,
-      "step": 29039
-    },
-    {
-      "epoch": 2.220310797637479,
-      "grad_norm": 0.002099972916767001,
-      "learning_rate": 0.0001999975689323532,
-      "loss": 46.0,
-      "step": 29040
-    },
-    {
-      "epoch": 2.220387254620869,
-      "grad_norm": 0.0007780368905514479,
-      "learning_rate": 0.0001999975687648644,
-      "loss": 46.0,
-      "step": 29041
-    },
-    {
-      "epoch": 2.2204637116042587,
-      "grad_norm": 0.01432411465793848,
-      "learning_rate": 0.00019999756859736986,
-      "loss": 46.0,
-      "step": 29042
-    },
-    {
-      "epoch": 2.2205401685876485,
-      "grad_norm": 0.002702204743400216,
-      "learning_rate": 0.00019999756842986954,
-      "loss": 46.0,
-      "step": 29043
-    },
-    {
-      "epoch": 2.2206166255710382,
-      "grad_norm": 0.001272039138711989,
-      "learning_rate": 0.00019999756826236348,
-      "loss": 46.0,
-      "step": 29044
-    },
-    {
-      "epoch": 2.220693082554428,
-      "grad_norm": 0.000633472518529743,
-      "learning_rate": 0.00019999756809485159,
-      "loss": 46.0,
-      "step": 29045
-    },
-    {
-      "epoch": 2.2207695395378177,
-      "grad_norm": 0.0009532385738566518,
-      "learning_rate": 0.00019999756792733398,
-      "loss": 46.0,
-      "step": 29046
-    },
-    {
-      "epoch": 2.220845996521207,
-      "grad_norm": 0.0037663697730749846,
-      "learning_rate": 0.0001999975677598106,
-      "loss": 46.0,
-      "step": 29047
-    },
-    {
-      "epoch": 2.220922453504597,
-      "grad_norm": 0.001251509878784418,
-      "learning_rate": 0.00019999756759228138,
-      "loss": 46.0,
-      "step": 29048
-    },
-    {
-      "epoch": 2.2209989104879866,
-      "grad_norm": 0.002210004720836878,
-      "learning_rate": 0.00019999756742474647,
-      "loss": 46.0,
-      "step": 29049
-    },
-    {
-      "epoch": 2.2210753674713763,
-      "grad_norm": 0.003042635740712285,
-      "learning_rate": 0.00019999756725720577,
-      "loss": 46.0,
-      "step": 29050
-    },
-    {
-      "epoch": 2.221151824454766,
-      "grad_norm": 0.0010396906873211265,
-      "learning_rate": 0.0001999975670896593,
-      "loss": 46.0,
-      "step": 29051
-    },
-    {
-      "epoch": 2.221228281438156,
-      "grad_norm": 0.0018654249142855406,
-      "learning_rate": 0.000199997566922107,
-      "loss": 46.0,
-      "step": 29052
-    },
-    {
-      "epoch": 2.2213047384215456,
-      "grad_norm": 0.0015053300885483623,
-      "learning_rate": 0.00019999756675454902,
-      "loss": 46.0,
-      "step": 29053
-    },
-    {
-      "epoch": 2.2213811954049354,
-      "grad_norm": 0.0016937172040343285,
-      "learning_rate": 0.00019999756658698522,
-      "loss": 46.0,
-      "step": 29054
-    },
-    {
-      "epoch": 2.221457652388325,
-      "grad_norm": 0.004003510810434818,
-      "learning_rate": 0.0001999975664194157,
-      "loss": 46.0,
-      "step": 29055
-    },
-    {
-      "epoch": 2.221534109371715,
-      "grad_norm": 0.0014084666036069393,
-      "learning_rate": 0.00019999756625184038,
-      "loss": 46.0,
-      "step": 29056
-    },
-    {
-      "epoch": 2.2216105663551042,
-      "grad_norm": 0.0011506223818287253,
-      "learning_rate": 0.0001999975660842593,
-      "loss": 46.0,
-      "step": 29057
-    },
-    {
-      "epoch": 2.221687023338494,
-      "grad_norm": 0.0011702999472618103,
-      "learning_rate": 0.00019999756591667243,
-      "loss": 46.0,
-      "step": 29058
-    },
-    {
-      "epoch": 2.2217634803218838,
-      "grad_norm": 0.0009672439773567021,
-      "learning_rate": 0.0001999975657490798,
-      "loss": 46.0,
-      "step": 29059
-    },
-    {
-      "epoch": 2.2218399373052735,
-      "grad_norm": 0.0016482521314173937,
-      "learning_rate": 0.0001999975655814814,
-      "loss": 46.0,
-      "step": 29060
-    },
-    {
-      "epoch": 2.2219163942886633,
-      "grad_norm": 0.002098256256431341,
-      "learning_rate": 0.00019999756541387722,
-      "loss": 46.0,
-      "step": 29061
-    },
-    {
-      "epoch": 2.221992851272053,
-      "grad_norm": 0.0007959646754898131,
-      "learning_rate": 0.0001999975652462673,
-      "loss": 46.0,
-      "step": 29062
-    },
-    {
-      "epoch": 2.222069308255443,
-      "grad_norm": 0.0004244661540724337,
-      "learning_rate": 0.00019999756507865159,
-      "loss": 46.0,
-      "step": 29063
-    },
-    {
-      "epoch": 2.2221457652388326,
-      "grad_norm": 0.011796372942626476,
-      "learning_rate": 0.0001999975649110301,
-      "loss": 46.0,
-      "step": 29064
-    },
-    {
-      "epoch": 2.2222222222222223,
-      "grad_norm": 0.0017134278314188123,
-      "learning_rate": 0.00019999756474340286,
-      "loss": 46.0,
-      "step": 29065
-    },
-    {
-      "epoch": 2.222298679205612,
-      "grad_norm": 0.0015598268946632743,
-      "learning_rate": 0.00019999756457576986,
-      "loss": 46.0,
-      "step": 29066
-    },
-    {
-      "epoch": 2.222375136189002,
-      "grad_norm": 0.0009513128316029906,
-      "learning_rate": 0.00019999756440813106,
-      "loss": 46.0,
-      "step": 29067
-    },
-    {
-      "epoch": 2.2224515931723916,
-      "grad_norm": 0.0017385525861755013,
-      "learning_rate": 0.0001999975642404865,
-      "loss": 46.0,
-      "step": 29068
-    },
-    {
-      "epoch": 2.222528050155781,
-      "grad_norm": 0.0072024790570139885,
-      "learning_rate": 0.00019999756407283617,
-      "loss": 46.0,
-      "step": 29069
-    },
-    {
-      "epoch": 2.2226045071391707,
-      "grad_norm": 0.0007386407232843339,
-      "learning_rate": 0.0001999975639051801,
-      "loss": 46.0,
-      "step": 29070
-    },
-    {
-      "epoch": 2.2226809641225604,
-      "grad_norm": 0.0009203071822412312,
-      "learning_rate": 0.00019999756373751824,
-      "loss": 46.0,
-      "step": 29071
-    },
-    {
-      "epoch": 2.22275742110595,
-      "grad_norm": 0.0005235082353465259,
-      "learning_rate": 0.0001999975635698506,
-      "loss": 46.0,
-      "step": 29072
-    },
-    {
-      "epoch": 2.22283387808934,
-      "grad_norm": 0.0011627032654359937,
-      "learning_rate": 0.00019999756340217722,
-      "loss": 46.0,
-      "step": 29073
-    },
-    {
-      "epoch": 2.2229103350727297,
-      "grad_norm": 0.005454846657812595,
-      "learning_rate": 0.00019999756323449804,
-      "loss": 46.0,
-      "step": 29074
-    },
-    {
-      "epoch": 2.2229867920561195,
-      "grad_norm": 0.0014561025891453028,
-      "learning_rate": 0.0001999975630668131,
-      "loss": 46.0,
-      "step": 29075
-    },
-    {
-      "epoch": 2.2230632490395092,
-      "grad_norm": 0.0006992297130636871,
-      "learning_rate": 0.00019999756289912238,
-      "loss": 46.0,
-      "step": 29076
-    },
-    {
-      "epoch": 2.223139706022899,
-      "grad_norm": 0.000693952664732933,
-      "learning_rate": 0.00019999756273142593,
-      "loss": 46.0,
-      "step": 29077
-    },
-    {
-      "epoch": 2.2232161630062888,
-      "grad_norm": 0.0017759924521669745,
-      "learning_rate": 0.00019999756256372365,
-      "loss": 46.0,
-      "step": 29078
-    },
-    {
-      "epoch": 2.223292619989678,
-      "grad_norm": 0.0012003574520349503,
-      "learning_rate": 0.00019999756239601565,
-      "loss": 46.0,
-      "step": 29079
-    },
-    {
-      "epoch": 2.223369076973068,
-      "grad_norm": 0.0021588949020951986,
-      "learning_rate": 0.00019999756222830185,
-      "loss": 46.0,
-      "step": 29080
-    },
-    {
-      "epoch": 2.2234455339564576,
-      "grad_norm": 0.0017198417335748672,
-      "learning_rate": 0.0001999975620605823,
-      "loss": 46.0,
-      "step": 29081
-    },
-    {
-      "epoch": 2.2235219909398474,
-      "grad_norm": 0.0013162444811314344,
-      "learning_rate": 0.000199997561892857,
-      "loss": 46.0,
-      "step": 29082
-    },
-    {
-      "epoch": 2.223598447923237,
-      "grad_norm": 0.0019963777158409357,
-      "learning_rate": 0.0001999975617251259,
-      "loss": 46.0,
-      "step": 29083
-    },
-    {
-      "epoch": 2.223674904906627,
-      "grad_norm": 0.0010876356391236186,
-      "learning_rate": 0.000199997561557389,
-      "loss": 46.0,
-      "step": 29084
-    },
-    {
-      "epoch": 2.2237513618900167,
-      "grad_norm": 0.0005942288553342223,
-      "learning_rate": 0.0001999975613896464,
-      "loss": 46.0,
-      "step": 29085
-    },
-    {
-      "epoch": 2.2238278188734064,
-      "grad_norm": 0.0017909345915541053,
-      "learning_rate": 0.00019999756122189802,
-      "loss": 46.0,
-      "step": 29086
-    },
-    {
-      "epoch": 2.223904275856796,
-      "grad_norm": 0.0018780564423650503,
-      "learning_rate": 0.00019999756105414383,
-      "loss": 46.0,
-      "step": 29087
-    },
-    {
-      "epoch": 2.223980732840186,
-      "grad_norm": 0.001375261228531599,
-      "learning_rate": 0.00019999756088638388,
-      "loss": 46.0,
-      "step": 29088
-    },
-    {
-      "epoch": 2.2240571898235757,
-      "grad_norm": 0.0004752881359308958,
-      "learning_rate": 0.0001999975607186182,
-      "loss": 46.0,
-      "step": 29089
-    },
-    {
-      "epoch": 2.2241336468069655,
-      "grad_norm": 0.0005805412074550986,
-      "learning_rate": 0.0001999975605508467,
-      "loss": 46.0,
-      "step": 29090
-    },
-    {
-      "epoch": 2.2242101037903548,
-      "grad_norm": 0.00048358997446484864,
-      "learning_rate": 0.00019999756038306945,
-      "loss": 46.0,
-      "step": 29091
-    },
-    {
-      "epoch": 2.2242865607737445,
-      "grad_norm": 0.0012624766677618027,
-      "learning_rate": 0.00019999756021528645,
-      "loss": 46.0,
-      "step": 29092
-    },
-    {
-      "epoch": 2.2243630177571343,
-      "grad_norm": 0.001781711122021079,
-      "learning_rate": 0.00019999756004749765,
-      "loss": 46.0,
-      "step": 29093
-    },
-    {
-      "epoch": 2.224439474740524,
-      "grad_norm": 0.0008308349642902613,
-      "learning_rate": 0.00019999755987970308,
-      "loss": 46.0,
-      "step": 29094
-    },
-    {
-      "epoch": 2.224515931723914,
-      "grad_norm": 0.0009148496319539845,
-      "learning_rate": 0.00019999755971190276,
-      "loss": 46.0,
-      "step": 29095
-    },
-    {
-      "epoch": 2.2245923887073036,
-      "grad_norm": 0.0010498544434085488,
-      "learning_rate": 0.0001999975595440967,
-      "loss": 46.0,
-      "step": 29096
-    },
-    {
-      "epoch": 2.2246688456906933,
-      "grad_norm": 0.0008421143284067512,
-      "learning_rate": 0.0001999975593762848,
-      "loss": 46.0,
-      "step": 29097
-    },
-    {
-      "epoch": 2.224745302674083,
-      "grad_norm": 0.0019818521104753017,
-      "learning_rate": 0.00019999755920846717,
-      "loss": 46.0,
-      "step": 29098
-    },
-    {
-      "epoch": 2.224821759657473,
-      "grad_norm": 0.0006081559113226831,
-      "learning_rate": 0.00019999755904064376,
-      "loss": 46.0,
-      "step": 29099
-    },
-    {
-      "epoch": 2.2248982166408626,
-      "grad_norm": 0.000737370690330863,
-      "learning_rate": 0.0001999975588728146,
-      "loss": 46.0,
-      "step": 29100
-    },
-    {
-      "epoch": 2.224974673624252,
-      "grad_norm": 0.0006839355337433517,
-      "learning_rate": 0.00019999755870497968,
-      "loss": 46.0,
-      "step": 29101
-    },
-    {
-      "epoch": 2.2250511306076417,
-      "grad_norm": 0.0008955879602581263,
-      "learning_rate": 0.00019999755853713895,
-      "loss": 46.0,
-      "step": 29102
-    },
-    {
-      "epoch": 2.2251275875910315,
-      "grad_norm": 0.0004602540866471827,
-      "learning_rate": 0.00019999755836929247,
-      "loss": 46.0,
-      "step": 29103
-    },
-    {
-      "epoch": 2.2252040445744212,
-      "grad_norm": 0.0018893477972596884,
-      "learning_rate": 0.00019999755820144022,
-      "loss": 46.0,
-      "step": 29104
-    },
-    {
-      "epoch": 2.225280501557811,
-      "grad_norm": 0.0012615799205377698,
-      "learning_rate": 0.00019999755803358222,
-      "loss": 46.0,
-      "step": 29105
-    },
-    {
-      "epoch": 2.2253569585412007,
-      "grad_norm": 0.002261328510940075,
-      "learning_rate": 0.0001999975578657184,
-      "loss": 46.0,
-      "step": 29106
-    },
-    {
-      "epoch": 2.2254334155245905,
-      "grad_norm": 0.000585398986004293,
-      "learning_rate": 0.00019999755769784888,
-      "loss": 46.0,
-      "step": 29107
-    },
-    {
-      "epoch": 2.2255098725079803,
-      "grad_norm": 0.003862884361296892,
-      "learning_rate": 0.00019999755752997357,
-      "loss": 46.0,
-      "step": 29108
-    },
-    {
-      "epoch": 2.22558632949137,
-      "grad_norm": 0.0008283432689495385,
-      "learning_rate": 0.00019999755736209245,
-      "loss": 46.0,
-      "step": 29109
-    },
-    {
-      "epoch": 2.22566278647476,
-      "grad_norm": 0.0007984499097801745,
-      "learning_rate": 0.00019999755719420556,
-      "loss": 46.0,
-      "step": 29110
-    },
-    {
-      "epoch": 2.2257392434581496,
-      "grad_norm": 0.0005845974083058536,
-      "learning_rate": 0.00019999755702631292,
-      "loss": 46.0,
-      "step": 29111
-    },
-    {
-      "epoch": 2.2258157004415393,
-      "grad_norm": 0.0010624047135934234,
-      "learning_rate": 0.00019999755685841454,
-      "loss": 46.0,
-      "step": 29112
-    },
-    {
-      "epoch": 2.2258921574249286,
-      "grad_norm": 0.0006716938805766404,
-      "learning_rate": 0.00019999755669051036,
-      "loss": 46.0,
-      "step": 29113
-    },
-    {
-      "epoch": 2.2259686144083184,
-      "grad_norm": 0.00025363240274600685,
-      "learning_rate": 0.00019999755652260043,
-      "loss": 46.0,
-      "step": 29114
-    },
-    {
-      "epoch": 2.226045071391708,
-      "grad_norm": 0.0028277297969907522,
-      "learning_rate": 0.00019999755635468472,
-      "loss": 46.0,
-      "step": 29115
-    },
-    {
-      "epoch": 2.226121528375098,
-      "grad_norm": 0.0010492609580978751,
-      "learning_rate": 0.00019999755618676325,
-      "loss": 46.0,
-      "step": 29116
-    },
-    {
-      "epoch": 2.2261979853584877,
-      "grad_norm": 0.0014806295512244105,
-      "learning_rate": 0.000199997556018836,
-      "loss": 46.0,
-      "step": 29117
-    },
-    {
-      "epoch": 2.2262744423418774,
-      "grad_norm": 0.0009304140694439411,
-      "learning_rate": 0.00019999755585090297,
-      "loss": 46.0,
-      "step": 29118
-    },
-    {
-      "epoch": 2.226350899325267,
-      "grad_norm": 0.0007762485765852034,
-      "learning_rate": 0.00019999755568296418,
-      "loss": 46.0,
-      "step": 29119
-    },
-    {
-      "epoch": 2.226427356308657,
-      "grad_norm": 0.0019231531769037247,
-      "learning_rate": 0.0001999975555150196,
-      "loss": 46.0,
-      "step": 29120
-    },
-    {
-      "epoch": 2.2265038132920467,
-      "grad_norm": 0.0030659798067063093,
-      "learning_rate": 0.0001999975553470693,
-      "loss": 46.0,
-      "step": 29121
-    },
-    {
-      "epoch": 2.2265802702754365,
-      "grad_norm": 0.0014600735157728195,
-      "learning_rate": 0.0001999975551791132,
-      "loss": 46.0,
-      "step": 29122
-    },
-    {
-      "epoch": 2.226656727258826,
-      "grad_norm": 0.0010399551829323173,
-      "learning_rate": 0.00019999755501115134,
-      "loss": 46.0,
-      "step": 29123
-    },
-    {
-      "epoch": 2.2267331842422156,
-      "grad_norm": 0.00101527851074934,
-      "learning_rate": 0.0001999975548431837,
-      "loss": 46.0,
-      "step": 29124
-    },
-    {
-      "epoch": 2.2268096412256053,
-      "grad_norm": 0.0012097390135750175,
-      "learning_rate": 0.0001999975546752103,
-      "loss": 46.0,
-      "step": 29125
-    },
-    {
-      "epoch": 2.226886098208995,
-      "grad_norm": 0.0029957378283143044,
-      "learning_rate": 0.0001999975545072311,
-      "loss": 46.0,
-      "step": 29126
-    },
-    {
-      "epoch": 2.226962555192385,
-      "grad_norm": 0.003678811015561223,
-      "learning_rate": 0.00019999755433924618,
-      "loss": 46.0,
-      "step": 29127
-    },
-    {
-      "epoch": 2.2270390121757746,
-      "grad_norm": 0.0009390611667186022,
-      "learning_rate": 0.00019999755417125548,
-      "loss": 46.0,
-      "step": 29128
-    },
-    {
-      "epoch": 2.2271154691591644,
-      "grad_norm": 0.0006514430278912187,
-      "learning_rate": 0.000199997554003259,
-      "loss": 46.0,
-      "step": 29129
-    },
-    {
-      "epoch": 2.227191926142554,
-      "grad_norm": 0.0013060710625723004,
-      "learning_rate": 0.00019999755383525673,
-      "loss": 46.0,
-      "step": 29130
-    },
-    {
-      "epoch": 2.227268383125944,
-      "grad_norm": 0.0007545781554654241,
-      "learning_rate": 0.0001999975536672487,
-      "loss": 46.0,
-      "step": 29131
-    },
-    {
-      "epoch": 2.2273448401093336,
-      "grad_norm": 0.0017160180723294616,
-      "learning_rate": 0.00019999755349923493,
-      "loss": 46.0,
-      "step": 29132
-    },
-    {
-      "epoch": 2.2274212970927234,
-      "grad_norm": 0.0004708092601504177,
-      "learning_rate": 0.00019999755333121536,
-      "loss": 46.0,
-      "step": 29133
-    },
-    {
-      "epoch": 2.2274977540761127,
-      "grad_norm": 0.00128403480630368,
-      "learning_rate": 0.00019999755316319005,
-      "loss": 46.0,
-      "step": 29134
-    },
-    {
-      "epoch": 2.2275742110595025,
-      "grad_norm": 0.0048139686696231365,
-      "learning_rate": 0.00019999755299515893,
-      "loss": 46.0,
-      "step": 29135
-    },
-    {
-      "epoch": 2.2276506680428922,
-      "grad_norm": 0.0017914820928126574,
-      "learning_rate": 0.00019999755282712207,
-      "loss": 46.0,
-      "step": 29136
-    },
-    {
-      "epoch": 2.227727125026282,
-      "grad_norm": 0.0016948640113696456,
-      "learning_rate": 0.00019999755265907943,
-      "loss": 46.0,
-      "step": 29137
-    },
-    {
-      "epoch": 2.2278035820096718,
-      "grad_norm": 0.0007627106970176101,
-      "learning_rate": 0.00019999755249103105,
-      "loss": 46.0,
-      "step": 29138
-    },
-    {
-      "epoch": 2.2278800389930615,
-      "grad_norm": 0.0066308495588600636,
-      "learning_rate": 0.00019999755232297687,
-      "loss": 46.0,
-      "step": 29139
-    },
-    {
-      "epoch": 2.2279564959764513,
-      "grad_norm": 0.0007879217155277729,
-      "learning_rate": 0.00019999755215491694,
-      "loss": 46.0,
-      "step": 29140
-    },
-    {
-      "epoch": 2.228032952959841,
-      "grad_norm": 0.0016604687552899122,
-      "learning_rate": 0.00019999755198685118,
-      "loss": 46.0,
-      "step": 29141
-    },
-    {
-      "epoch": 2.228109409943231,
-      "grad_norm": 0.005728233605623245,
-      "learning_rate": 0.00019999755181877973,
-      "loss": 46.0,
-      "step": 29142
-    },
-    {
-      "epoch": 2.2281858669266206,
-      "grad_norm": 0.0014835888287052512,
-      "learning_rate": 0.00019999755165070248,
-      "loss": 46.0,
-      "step": 29143
-    },
-    {
-      "epoch": 2.2282623239100103,
-      "grad_norm": 0.0017513331258669496,
-      "learning_rate": 0.00019999755148261946,
-      "loss": 46.0,
-      "step": 29144
-    },
-    {
-      "epoch": 2.2283387808933997,
-      "grad_norm": 0.0019309240160509944,
-      "learning_rate": 0.00019999755131453067,
-      "loss": 46.0,
-      "step": 29145
-    },
-    {
-      "epoch": 2.2284152378767894,
-      "grad_norm": 0.0015076907584443688,
-      "learning_rate": 0.0001999975511464361,
-      "loss": 46.0,
-      "step": 29146
-    },
-    {
-      "epoch": 2.228491694860179,
-      "grad_norm": 0.002651678863912821,
-      "learning_rate": 0.00019999755097833578,
-      "loss": 46.0,
-      "step": 29147
-    },
-    {
-      "epoch": 2.228568151843569,
-      "grad_norm": 0.00253850850276649,
-      "learning_rate": 0.0001999975508102297,
-      "loss": 46.0,
-      "step": 29148
-    },
-    {
-      "epoch": 2.2286446088269587,
-      "grad_norm": 0.001062647090293467,
-      "learning_rate": 0.0001999975506421178,
-      "loss": 46.0,
-      "step": 29149
-    },
-    {
-      "epoch": 2.2287210658103485,
-      "grad_norm": 0.0008598386775702238,
-      "learning_rate": 0.00019999755047400017,
-      "loss": 46.0,
-      "step": 29150
-    },
-    {
-      "epoch": 2.228797522793738,
-      "grad_norm": 0.0016063852235674858,
-      "learning_rate": 0.00019999755030587676,
-      "loss": 46.0,
-      "step": 29151
-    },
-    {
-      "epoch": 2.228873979777128,
-      "grad_norm": 0.0008654667180962861,
-      "learning_rate": 0.0001999975501377476,
-      "loss": 46.0,
-      "step": 29152
-    },
-    {
-      "epoch": 2.2289504367605177,
-      "grad_norm": 0.0009056190610863268,
-      "learning_rate": 0.00019999754996961265,
-      "loss": 46.0,
-      "step": 29153
-    },
-    {
-      "epoch": 2.2290268937439075,
-      "grad_norm": 0.0034059497993439436,
-      "learning_rate": 0.00019999754980147194,
-      "loss": 46.0,
-      "step": 29154
-    },
-    {
-      "epoch": 2.2291033507272973,
-      "grad_norm": 0.0016089368145912886,
-      "learning_rate": 0.00019999754963332544,
-      "loss": 46.0,
-      "step": 29155
-    },
-    {
-      "epoch": 2.2291798077106866,
-      "grad_norm": 0.0010404606582596898,
-      "learning_rate": 0.0001999975494651732,
-      "loss": 46.0,
-      "step": 29156
-    },
-    {
-      "epoch": 2.2292562646940763,
-      "grad_norm": 0.0008481297409161925,
-      "learning_rate": 0.00019999754929701517,
-      "loss": 46.0,
-      "step": 29157
-    },
-    {
-      "epoch": 2.229332721677466,
-      "grad_norm": 0.0007183749112300575,
-      "learning_rate": 0.00019999754912885138,
-      "loss": 46.0,
-      "step": 29158
-    },
-    {
-      "epoch": 2.229409178660856,
-      "grad_norm": 0.0015572463162243366,
-      "learning_rate": 0.0001999975489606818,
-      "loss": 46.0,
-      "step": 29159
-    },
-    {
-      "epoch": 2.2294856356442456,
-      "grad_norm": 0.001245118328370154,
-      "learning_rate": 0.0001999975487925065,
-      "loss": 46.0,
-      "step": 29160
-    },
-    {
-      "epoch": 2.2295620926276354,
-      "grad_norm": 0.001970109064131975,
-      "learning_rate": 0.00019999754862432538,
-      "loss": 46.0,
-      "step": 29161
-    },
-    {
-      "epoch": 2.229638549611025,
-      "grad_norm": 0.0010958451312035322,
-      "learning_rate": 0.0001999975484561385,
-      "loss": 46.0,
-      "step": 29162
-    },
-    {
-      "epoch": 2.229715006594415,
-      "grad_norm": 0.0012278254143893719,
-      "learning_rate": 0.00019999754828794588,
-      "loss": 46.0,
-      "step": 29163
-    },
-    {
-      "epoch": 2.2297914635778047,
-      "grad_norm": 0.0021249623969197273,
-      "learning_rate": 0.00019999754811974747,
-      "loss": 46.0,
-      "step": 29164
-    },
-    {
-      "epoch": 2.2298679205611944,
-      "grad_norm": 0.0006810056511312723,
-      "learning_rate": 0.0001999975479515433,
-      "loss": 46.0,
-      "step": 29165
-    },
-    {
-      "epoch": 2.2299443775445837,
-      "grad_norm": 0.0012682306114584208,
-      "learning_rate": 0.00019999754778333336,
-      "loss": 46.0,
-      "step": 29166
-    },
-    {
-      "epoch": 2.2300208345279735,
-      "grad_norm": 0.001094619045034051,
-      "learning_rate": 0.00019999754761511763,
-      "loss": 46.0,
-      "step": 29167
-    },
-    {
-      "epoch": 2.2300972915113633,
-      "grad_norm": 0.0009834646480157971,
-      "learning_rate": 0.00019999754744689615,
-      "loss": 46.0,
-      "step": 29168
-    },
-    {
-      "epoch": 2.230173748494753,
-      "grad_norm": 0.007009634282439947,
-      "learning_rate": 0.0001999975472786689,
-      "loss": 46.0,
-      "step": 29169
-    },
-    {
-      "epoch": 2.230250205478143,
-      "grad_norm": 0.0008895605569705367,
-      "learning_rate": 0.00019999754711043588,
-      "loss": 46.0,
-      "step": 29170
-    },
-    {
-      "epoch": 2.2303266624615325,
-      "grad_norm": 0.003990990109741688,
-      "learning_rate": 0.0001999975469421971,
-      "loss": 46.0,
-      "step": 29171
-    },
-    {
-      "epoch": 2.2304031194449223,
-      "grad_norm": 0.0022045692894607782,
-      "learning_rate": 0.0001999975467739525,
-      "loss": 46.0,
-      "step": 29172
-    },
-    {
-      "epoch": 2.230479576428312,
-      "grad_norm": 0.0006977233570069075,
-      "learning_rate": 0.0001999975466057022,
-      "loss": 46.0,
-      "step": 29173
-    },
-    {
-      "epoch": 2.230556033411702,
-      "grad_norm": 0.000602979795075953,
-      "learning_rate": 0.0001999975464374461,
-      "loss": 46.0,
-      "step": 29174
-    },
-    {
-      "epoch": 2.2306324903950916,
-      "grad_norm": 0.0009595765150152147,
-      "learning_rate": 0.00019999754626918423,
-      "loss": 46.0,
-      "step": 29175
-    },
-    {
-      "epoch": 2.2307089473784814,
-      "grad_norm": 0.0008591891964897513,
-      "learning_rate": 0.00019999754610091657,
-      "loss": 46.0,
-      "step": 29176
-    },
-    {
-      "epoch": 2.230785404361871,
-      "grad_norm": 0.0005220207385718822,
-      "learning_rate": 0.00019999754593264318,
-      "loss": 46.0,
-      "step": 29177
-    },
-    {
-      "epoch": 2.2308618613452604,
-      "grad_norm": 0.001243693521246314,
-      "learning_rate": 0.000199997545764364,
-      "loss": 46.0,
-      "step": 29178
-    },
-    {
-      "epoch": 2.23093831832865,
-      "grad_norm": 0.002552089747041464,
-      "learning_rate": 0.00019999754559607907,
-      "loss": 46.0,
-      "step": 29179
-    },
-    {
-      "epoch": 2.23101477531204,
-      "grad_norm": 0.0009439396671950817,
-      "learning_rate": 0.00019999754542778835,
-      "loss": 46.0,
-      "step": 29180
-    },
-    {
-      "epoch": 2.2310912322954297,
-      "grad_norm": 0.0016397950239479542,
-      "learning_rate": 0.00019999754525949187,
-      "loss": 46.0,
-      "step": 29181
-    },
-    {
-      "epoch": 2.2311676892788195,
-      "grad_norm": 0.00040711203473620117,
-      "learning_rate": 0.0001999975450911896,
-      "loss": 46.0,
-      "step": 29182
-    },
-    {
-      "epoch": 2.2312441462622092,
-      "grad_norm": 0.0015877274563536048,
-      "learning_rate": 0.00019999754492288157,
-      "loss": 46.0,
-      "step": 29183
-    },
-    {
-      "epoch": 2.231320603245599,
-      "grad_norm": 0.0008642931352369487,
-      "learning_rate": 0.00019999754475456775,
-      "loss": 46.0,
-      "step": 29184
-    },
-    {
-      "epoch": 2.2313970602289888,
-      "grad_norm": 0.0043951366096735,
-      "learning_rate": 0.0001999975445862482,
-      "loss": 46.0,
-      "step": 29185
-    },
-    {
-      "epoch": 2.2314735172123785,
-      "grad_norm": 0.0015020916471257806,
-      "learning_rate": 0.00019999754441792287,
-      "loss": 46.0,
-      "step": 29186
-    },
-    {
-      "epoch": 2.2315499741957683,
-      "grad_norm": 0.00326914363540709,
-      "learning_rate": 0.00019999754424959178,
-      "loss": 46.0,
-      "step": 29187
-    },
-    {
-      "epoch": 2.2316264311791576,
-      "grad_norm": 0.0157815869897604,
-      "learning_rate": 0.00019999754408125492,
-      "loss": 46.0,
-      "step": 29188
-    },
-    {
-      "epoch": 2.2317028881625474,
-      "grad_norm": 0.0005216411082074046,
-      "learning_rate": 0.00019999754391291226,
-      "loss": 46.0,
-      "step": 29189
-    },
-    {
-      "epoch": 2.231779345145937,
-      "grad_norm": 0.0023198905400931835,
-      "learning_rate": 0.00019999754374456385,
-      "loss": 46.0,
-      "step": 29190
-    },
-    {
-      "epoch": 2.231855802129327,
-      "grad_norm": 0.0006171483546495438,
-      "learning_rate": 0.00019999754357620967,
-      "loss": 46.0,
-      "step": 29191
-    },
-    {
-      "epoch": 2.2319322591127166,
-      "grad_norm": 0.001198531361296773,
-      "learning_rate": 0.00019999754340784974,
-      "loss": 46.0,
-      "step": 29192
-    },
-    {
-      "epoch": 2.2320087160961064,
-      "grad_norm": 0.0006718302611261606,
-      "learning_rate": 0.000199997543239484,
-      "loss": 46.0,
-      "step": 29193
-    },
-    {
-      "epoch": 2.232085173079496,
-      "grad_norm": 0.0009702850366011262,
-      "learning_rate": 0.0001999975430711125,
-      "loss": 46.0,
-      "step": 29194
-    },
-    {
-      "epoch": 2.232161630062886,
-      "grad_norm": 0.0006308432784862816,
-      "learning_rate": 0.00019999754290273526,
-      "loss": 46.0,
-      "step": 29195
-    },
-    {
-      "epoch": 2.2322380870462757,
-      "grad_norm": 0.0014636690029874444,
-      "learning_rate": 0.00019999754273435224,
-      "loss": 46.0,
-      "step": 29196
-    },
-    {
-      "epoch": 2.2323145440296654,
-      "grad_norm": 0.0006831357604824007,
-      "learning_rate": 0.00019999754256596345,
-      "loss": 46.0,
-      "step": 29197
-    },
-    {
-      "epoch": 2.232391001013055,
-      "grad_norm": 0.0012689635623246431,
-      "learning_rate": 0.0001999975423975689,
-      "loss": 46.0,
-      "step": 29198
-    },
-    {
-      "epoch": 2.232467457996445,
-      "grad_norm": 0.0027095209807157516,
-      "learning_rate": 0.00019999754222916856,
-      "loss": 46.0,
-      "step": 29199
-    },
-    {
-      "epoch": 2.2325439149798343,
-      "grad_norm": 0.0010933097219094634,
-      "learning_rate": 0.00019999754206076245,
-      "loss": 46.0,
-      "step": 29200
-    },
-    {
-      "epoch": 2.232620371963224,
-      "grad_norm": 0.003304373938590288,
-      "learning_rate": 0.0001999975418923506,
-      "loss": 46.0,
-      "step": 29201
-    },
-    {
-      "epoch": 2.232696828946614,
-      "grad_norm": 0.004600857384502888,
-      "learning_rate": 0.00019999754172393295,
-      "loss": 46.0,
-      "step": 29202
-    },
-    {
-      "epoch": 2.2327732859300036,
-      "grad_norm": 0.0017135116504505277,
-      "learning_rate": 0.00019999754155550955,
-      "loss": 46.0,
-      "step": 29203
-    },
-    {
-      "epoch": 2.2328497429133933,
-      "grad_norm": 0.0021816326770931482,
-      "learning_rate": 0.00019999754138708036,
-      "loss": 46.0,
-      "step": 29204
-    },
-    {
-      "epoch": 2.232926199896783,
-      "grad_norm": 0.0009967110818251967,
-      "learning_rate": 0.0001999975412186454,
-      "loss": 46.0,
-      "step": 29205
-    },
-    {
-      "epoch": 2.233002656880173,
-      "grad_norm": 0.0008666834910400212,
-      "learning_rate": 0.00019999754105020468,
-      "loss": 46.0,
-      "step": 29206
-    },
-    {
-      "epoch": 2.2330791138635626,
-      "grad_norm": 0.0010114125907421112,
-      "learning_rate": 0.00019999754088175818,
-      "loss": 46.0,
-      "step": 29207
-    },
-    {
-      "epoch": 2.2331555708469524,
-      "grad_norm": 0.001203992054797709,
-      "learning_rate": 0.00019999754071330593,
-      "loss": 46.0,
-      "step": 29208
-    },
-    {
-      "epoch": 2.233232027830342,
-      "grad_norm": 0.0034374005626887083,
-      "learning_rate": 0.00019999754054484789,
-      "loss": 46.0,
-      "step": 29209
-    },
-    {
-      "epoch": 2.2333084848137315,
-      "grad_norm": 0.0025958274491131306,
-      "learning_rate": 0.00019999754037638412,
-      "loss": 46.0,
-      "step": 29210
-    },
-    {
-      "epoch": 2.233384941797121,
-      "grad_norm": 0.00039328786078840494,
-      "learning_rate": 0.00019999754020791455,
-      "loss": 46.0,
-      "step": 29211
-    },
-    {
-      "epoch": 2.233461398780511,
-      "grad_norm": 0.0014856280758976936,
-      "learning_rate": 0.0001999975400394392,
-      "loss": 46.0,
-      "step": 29212
-    },
-    {
-      "epoch": 2.2335378557639007,
-      "grad_norm": 0.005227528512477875,
-      "learning_rate": 0.0001999975398709581,
-      "loss": 46.0,
-      "step": 29213
-    },
-    {
-      "epoch": 2.2336143127472905,
-      "grad_norm": 0.000613532611168921,
-      "learning_rate": 0.00019999753970247123,
-      "loss": 46.0,
-      "step": 29214
-    },
-    {
-      "epoch": 2.2336907697306803,
-      "grad_norm": 0.0010660778498277068,
-      "learning_rate": 0.0001999975395339786,
-      "loss": 46.0,
-      "step": 29215
-    },
-    {
-      "epoch": 2.23376722671407,
-      "grad_norm": 0.0008584194001741707,
-      "learning_rate": 0.0001999975393654802,
-      "loss": 46.0,
-      "step": 29216
-    },
-    {
-      "epoch": 2.23384368369746,
-      "grad_norm": 0.0007386516663245857,
-      "learning_rate": 0.00019999753919697598,
-      "loss": 46.0,
-      "step": 29217
-    },
-    {
-      "epoch": 2.2339201406808495,
-      "grad_norm": 0.001097518834285438,
-      "learning_rate": 0.00019999753902846606,
-      "loss": 46.0,
-      "step": 29218
-    },
-    {
-      "epoch": 2.2339965976642393,
-      "grad_norm": 0.0004880745545960963,
-      "learning_rate": 0.00019999753885995033,
-      "loss": 46.0,
-      "step": 29219
-    },
-    {
-      "epoch": 2.234073054647629,
-      "grad_norm": 0.0006976108998060226,
-      "learning_rate": 0.00019999753869142883,
-      "loss": 46.0,
-      "step": 29220
-    },
-    {
-      "epoch": 2.234149511631019,
-      "grad_norm": 0.0009242414962500334,
-      "learning_rate": 0.0001999975385229016,
-      "loss": 46.0,
-      "step": 29221
-    },
-    {
-      "epoch": 2.234225968614408,
-      "grad_norm": 0.0010113274911418557,
-      "learning_rate": 0.00019999753835436853,
-      "loss": 46.0,
-      "step": 29222
-    },
-    {
-      "epoch": 2.234302425597798,
-      "grad_norm": 0.0007292842492461205,
-      "learning_rate": 0.00019999753818582977,
-      "loss": 46.0,
-      "step": 29223
-    },
-    {
-      "epoch": 2.2343788825811877,
-      "grad_norm": 0.0007697870023548603,
-      "learning_rate": 0.0001999975380172852,
-      "loss": 46.0,
-      "step": 29224
-    },
-    {
-      "epoch": 2.2344553395645774,
-      "grad_norm": 0.0007101548253558576,
-      "learning_rate": 0.00019999753784873489,
-      "loss": 46.0,
-      "step": 29225
-    },
-    {
-      "epoch": 2.234531796547967,
-      "grad_norm": 0.002976537449285388,
-      "learning_rate": 0.00019999753768017874,
-      "loss": 46.0,
-      "step": 29226
-    },
-    {
-      "epoch": 2.234608253531357,
-      "grad_norm": 0.0007559265941381454,
-      "learning_rate": 0.00019999753751161688,
-      "loss": 46.0,
-      "step": 29227
-    },
-    {
-      "epoch": 2.2346847105147467,
-      "grad_norm": 0.0008199620060622692,
-      "learning_rate": 0.00019999753734304925,
-      "loss": 46.0,
-      "step": 29228
-    },
-    {
-      "epoch": 2.2347611674981365,
-      "grad_norm": 0.0014298021560534835,
-      "learning_rate": 0.00019999753717447582,
-      "loss": 46.0,
-      "step": 29229
-    },
-    {
-      "epoch": 2.2348376244815262,
-      "grad_norm": 0.004141592886298895,
-      "learning_rate": 0.00019999753700589664,
-      "loss": 46.0,
-      "step": 29230
-    },
-    {
-      "epoch": 2.234914081464916,
-      "grad_norm": 0.003073490224778652,
-      "learning_rate": 0.0001999975368373117,
-      "loss": 46.0,
-      "step": 29231
-    },
-    {
-      "epoch": 2.2349905384483053,
-      "grad_norm": 0.0032897889614105225,
-      "learning_rate": 0.00019999753666872095,
-      "loss": 46.0,
-      "step": 29232
-    },
-    {
-      "epoch": 2.235066995431695,
-      "grad_norm": 0.0016638361848890781,
-      "learning_rate": 0.00019999753650012448,
-      "loss": 46.0,
-      "step": 29233
-    },
-    {
-      "epoch": 2.235143452415085,
-      "grad_norm": 0.00111988908611238,
-      "learning_rate": 0.0001999975363315222,
-      "loss": 46.0,
-      "step": 29234
-    },
-    {
-      "epoch": 2.2352199093984746,
-      "grad_norm": 0.00046848802594468,
-      "learning_rate": 0.00019999753616291419,
-      "loss": 46.0,
-      "step": 29235
-    },
-    {
-      "epoch": 2.2352963663818644,
-      "grad_norm": 0.0004590329772327095,
-      "learning_rate": 0.0001999975359943004,
-      "loss": 46.0,
-      "step": 29236
-    },
-    {
-      "epoch": 2.235372823365254,
-      "grad_norm": 0.0006543307099491358,
-      "learning_rate": 0.00019999753582568083,
-      "loss": 46.0,
-      "step": 29237
-    },
-    {
-      "epoch": 2.235449280348644,
-      "grad_norm": 0.0008996034739539027,
-      "learning_rate": 0.0001999975356570555,
-      "loss": 46.0,
-      "step": 29238
-    },
-    {
-      "epoch": 2.2355257373320336,
-      "grad_norm": 0.00228469748981297,
-      "learning_rate": 0.0001999975354884244,
-      "loss": 46.0,
-      "step": 29239
-    },
-    {
-      "epoch": 2.2356021943154234,
-      "grad_norm": 0.00202162005007267,
-      "learning_rate": 0.0001999975353197875,
-      "loss": 46.0,
-      "step": 29240
-    },
-    {
-      "epoch": 2.235678651298813,
-      "grad_norm": 0.000999077339656651,
-      "learning_rate": 0.00019999753515114485,
-      "loss": 46.0,
-      "step": 29241
-    },
-    {
-      "epoch": 2.235755108282203,
-      "grad_norm": 0.0009139621397480369,
-      "learning_rate": 0.00019999753498249644,
-      "loss": 46.0,
-      "step": 29242
-    },
-    {
-      "epoch": 2.2358315652655927,
-      "grad_norm": 0.0007274738745763898,
-      "learning_rate": 0.00019999753481384226,
-      "loss": 46.0,
-      "step": 29243
-    },
-    {
-      "epoch": 2.235908022248982,
-      "grad_norm": 0.0013336505508050323,
-      "learning_rate": 0.0001999975346451823,
-      "loss": 46.0,
-      "step": 29244
-    },
-    {
-      "epoch": 2.2359844792323718,
-      "grad_norm": 0.0009994828142225742,
-      "learning_rate": 0.00019999753447651658,
-      "loss": 46.0,
-      "step": 29245
-    },
-    {
-      "epoch": 2.2360609362157615,
-      "grad_norm": 0.00039548170752823353,
-      "learning_rate": 0.0001999975343078451,
-      "loss": 46.0,
-      "step": 29246
-    },
-    {
-      "epoch": 2.2361373931991513,
-      "grad_norm": 0.0018002608558163047,
-      "learning_rate": 0.00019999753413916784,
-      "loss": 46.0,
-      "step": 29247
-    },
-    {
-      "epoch": 2.236213850182541,
-      "grad_norm": 0.0008931010379455984,
-      "learning_rate": 0.00019999753397048482,
-      "loss": 46.0,
-      "step": 29248
-    },
-    {
-      "epoch": 2.236290307165931,
-      "grad_norm": 0.0009544584318064153,
-      "learning_rate": 0.000199997533801796,
-      "loss": 46.0,
-      "step": 29249
-    },
-    {
-      "epoch": 2.2363667641493206,
-      "grad_norm": 0.0008065239526331425,
-      "learning_rate": 0.00019999753363310146,
-      "loss": 46.0,
-      "step": 29250
-    },
-    {
-      "epoch": 2.2364432211327103,
-      "grad_norm": 0.0009010698995552957,
-      "learning_rate": 0.00019999753346440112,
-      "loss": 46.0,
-      "step": 29251
-    },
-    {
-      "epoch": 2.2365196781161,
-      "grad_norm": 0.002153923502191901,
-      "learning_rate": 0.00019999753329569503,
-      "loss": 46.0,
-      "step": 29252
-    },
-    {
-      "epoch": 2.23659613509949,
-      "grad_norm": 0.0018601641058921814,
-      "learning_rate": 0.00019999753312698311,
-      "loss": 46.0,
-      "step": 29253
-    },
-    {
-      "epoch": 2.236672592082879,
-      "grad_norm": 0.0007505971589125693,
-      "learning_rate": 0.00019999753295826545,
-      "loss": 46.0,
-      "step": 29254
-    },
-    {
-      "epoch": 2.236749049066269,
-      "grad_norm": 0.0007172297919169068,
-      "learning_rate": 0.00019999753278954207,
-      "loss": 46.0,
-      "step": 29255
-    },
-    {
-      "epoch": 2.2368255060496587,
-      "grad_norm": 0.0005753936129622161,
-      "learning_rate": 0.00019999753262081287,
-      "loss": 46.0,
-      "step": 29256
-    },
-    {
-      "epoch": 2.2369019630330484,
-      "grad_norm": 0.000982632045634091,
-      "learning_rate": 0.00019999753245207794,
-      "loss": 46.0,
-      "step": 29257
-    },
-    {
-      "epoch": 2.236978420016438,
-      "grad_norm": 0.0021230040583759546,
-      "learning_rate": 0.00019999753228333722,
-      "loss": 46.0,
-      "step": 29258
-    },
-    {
-      "epoch": 2.237054876999828,
-      "grad_norm": 0.0005564973107539117,
-      "learning_rate": 0.00019999753211459071,
-      "loss": 46.0,
-      "step": 29259
-    },
-    {
-      "epoch": 2.2371313339832177,
-      "grad_norm": 0.0007678851834498346,
-      "learning_rate": 0.00019999753194583847,
-      "loss": 46.0,
-      "step": 29260
-    },
-    {
-      "epoch": 2.2372077909666075,
-      "grad_norm": 0.0006714258925057948,
-      "learning_rate": 0.00019999753177708042,
-      "loss": 46.0,
-      "step": 29261
-    },
-    {
-      "epoch": 2.2372842479499973,
-      "grad_norm": 0.0012710849987342954,
-      "learning_rate": 0.00019999753160831663,
-      "loss": 46.0,
-      "step": 29262
-    },
-    {
-      "epoch": 2.237360704933387,
-      "grad_norm": 0.0015369082102552056,
-      "learning_rate": 0.00019999753143954706,
-      "loss": 46.0,
-      "step": 29263
-    },
-    {
-      "epoch": 2.2374371619167768,
-      "grad_norm": 0.0006091243121773005,
-      "learning_rate": 0.00019999753127077172,
-      "loss": 46.0,
-      "step": 29264
-    },
-    {
-      "epoch": 2.237513618900166,
-      "grad_norm": 0.0011071724584326148,
-      "learning_rate": 0.0001999975311019906,
-      "loss": 46.0,
-      "step": 29265
-    },
-    {
-      "epoch": 2.237590075883556,
-      "grad_norm": 0.0009000373538583517,
-      "learning_rate": 0.00019999753093320372,
-      "loss": 46.0,
-      "step": 29266
-    },
-    {
-      "epoch": 2.2376665328669456,
-      "grad_norm": 0.0009503897745162249,
-      "learning_rate": 0.0001999975307644111,
-      "loss": 46.0,
-      "step": 29267
-    },
-    {
-      "epoch": 2.2377429898503354,
-      "grad_norm": 0.0017044448759406805,
-      "learning_rate": 0.00019999753059561268,
-      "loss": 46.0,
-      "step": 29268
-    },
-    {
-      "epoch": 2.237819446833725,
-      "grad_norm": 0.003015250666067004,
-      "learning_rate": 0.00019999753042680848,
-      "loss": 46.0,
-      "step": 29269
-    },
-    {
-      "epoch": 2.237895903817115,
-      "grad_norm": 0.0005690613761544228,
-      "learning_rate": 0.00019999753025799852,
-      "loss": 46.0,
-      "step": 29270
-    },
-    {
-      "epoch": 2.2379723608005047,
-      "grad_norm": 0.0005843983963131905,
-      "learning_rate": 0.0001999975300891828,
-      "loss": 46.0,
-      "step": 29271
-    },
-    {
-      "epoch": 2.2380488177838944,
-      "grad_norm": 0.00123932387214154,
-      "learning_rate": 0.00019999752992036132,
-      "loss": 46.0,
-      "step": 29272
-    },
-    {
-      "epoch": 2.238125274767284,
-      "grad_norm": 0.0008801360381767154,
-      "learning_rate": 0.00019999752975153405,
-      "loss": 46.0,
-      "step": 29273
-    },
-    {
-      "epoch": 2.238201731750674,
-      "grad_norm": 0.005004067439585924,
-      "learning_rate": 0.000199997529582701,
-      "loss": 46.0,
-      "step": 29274
-    },
-    {
-      "epoch": 2.2382781887340637,
-      "grad_norm": 0.0057562850415706635,
-      "learning_rate": 0.0001999975294138622,
-      "loss": 46.0,
-      "step": 29275
-    },
-    {
-      "epoch": 2.238354645717453,
-      "grad_norm": 0.0020179220009595156,
-      "learning_rate": 0.00019999752924501765,
-      "loss": 46.0,
-      "step": 29276
-    },
-    {
-      "epoch": 2.238431102700843,
-      "grad_norm": 0.0020850191358476877,
-      "learning_rate": 0.0001999975290761673,
-      "loss": 46.0,
-      "step": 29277
-    },
-    {
-      "epoch": 2.2385075596842325,
-      "grad_norm": 0.0006886178161948919,
-      "learning_rate": 0.0001999975289073112,
-      "loss": 46.0,
-      "step": 29278
-    },
-    {
-      "epoch": 2.2385840166676223,
-      "grad_norm": 0.0004178480012342334,
-      "learning_rate": 0.00019999752873844933,
-      "loss": 46.0,
-      "step": 29279
-    },
-    {
-      "epoch": 2.238660473651012,
-      "grad_norm": 0.0009546743240207434,
-      "learning_rate": 0.00019999752856958167,
-      "loss": 46.0,
-      "step": 29280
-    },
-    {
-      "epoch": 2.238736930634402,
-      "grad_norm": 0.004087148234248161,
-      "learning_rate": 0.00019999752840070824,
-      "loss": 46.0,
-      "step": 29281
-    },
-    {
-      "epoch": 2.2388133876177916,
-      "grad_norm": 0.0008135377429425716,
-      "learning_rate": 0.00019999752823182906,
-      "loss": 46.0,
-      "step": 29282
-    },
-    {
-      "epoch": 2.2388898446011813,
-      "grad_norm": 0.0008384595275856555,
-      "learning_rate": 0.0001999975280629441,
-      "loss": 46.0,
-      "step": 29283
-    },
-    {
-      "epoch": 2.238966301584571,
-      "grad_norm": 0.0008722853963263333,
-      "learning_rate": 0.00019999752789405338,
-      "loss": 46.0,
-      "step": 29284
-    },
-    {
-      "epoch": 2.239042758567961,
-      "grad_norm": 0.0008590986835770309,
-      "learning_rate": 0.00019999752772515688,
-      "loss": 46.0,
-      "step": 29285
-    },
-    {
-      "epoch": 2.2391192155513506,
-      "grad_norm": 0.0006694166222587228,
-      "learning_rate": 0.00019999752755625464,
-      "loss": 46.0,
-      "step": 29286
-    },
-    {
-      "epoch": 2.23919567253474,
-      "grad_norm": 0.0012403350556269288,
-      "learning_rate": 0.0001999975273873466,
-      "loss": 46.0,
-      "step": 29287
-    },
-    {
-      "epoch": 2.2392721295181297,
-      "grad_norm": 0.0006181875360198319,
-      "learning_rate": 0.0001999975272184328,
-      "loss": 46.0,
-      "step": 29288
-    },
-    {
-      "epoch": 2.2393485865015195,
-      "grad_norm": 0.0008266624645330012,
-      "learning_rate": 0.00019999752704951323,
-      "loss": 46.0,
-      "step": 29289
-    },
-    {
-      "epoch": 2.2394250434849092,
-      "grad_norm": 0.002642162377014756,
-      "learning_rate": 0.0001999975268805879,
-      "loss": 46.0,
-      "step": 29290
-    },
-    {
-      "epoch": 2.239501500468299,
-      "grad_norm": 0.0012388363247737288,
-      "learning_rate": 0.00019999752671165676,
-      "loss": 46.0,
-      "step": 29291
-    },
-    {
-      "epoch": 2.2395779574516888,
-      "grad_norm": 0.0012843264266848564,
-      "learning_rate": 0.0001999975265427199,
-      "loss": 46.0,
-      "step": 29292
-    },
-    {
-      "epoch": 2.2396544144350785,
-      "grad_norm": 0.0009894461836665869,
-      "learning_rate": 0.00019999752637377724,
-      "loss": 46.0,
-      "step": 29293
-    },
-    {
-      "epoch": 2.2397308714184683,
-      "grad_norm": 0.0005327691906131804,
-      "learning_rate": 0.00019999752620482883,
-      "loss": 46.0,
-      "step": 29294
-    },
-    {
-      "epoch": 2.239807328401858,
-      "grad_norm": 0.000824386312160641,
-      "learning_rate": 0.00019999752603587463,
-      "loss": 46.0,
-      "step": 29295
-    },
-    {
-      "epoch": 2.239883785385248,
-      "grad_norm": 0.001715023652650416,
-      "learning_rate": 0.00019999752586691467,
-      "loss": 46.0,
-      "step": 29296
-    },
-    {
-      "epoch": 2.239960242368637,
-      "grad_norm": 0.0012950344244018197,
-      "learning_rate": 0.00019999752569794897,
-      "loss": 46.0,
-      "step": 29297
-    },
-    {
-      "epoch": 2.240036699352027,
-      "grad_norm": 0.0022716624662280083,
-      "learning_rate": 0.00019999752552897748,
-      "loss": 46.0,
-      "step": 29298
-    },
-    {
-      "epoch": 2.2401131563354166,
-      "grad_norm": 0.0012147424276918173,
-      "learning_rate": 0.0001999975253600002,
-      "loss": 46.0,
-      "step": 29299
-    },
-    {
-      "epoch": 2.2401896133188064,
-      "grad_norm": 0.0008777669863775373,
-      "learning_rate": 0.00019999752519101716,
-      "loss": 46.0,
-      "step": 29300
-    },
-    {
-      "epoch": 2.240266070302196,
-      "grad_norm": 0.000599419348873198,
-      "learning_rate": 0.00019999752502202836,
-      "loss": 46.0,
-      "step": 29301
-    },
-    {
-      "epoch": 2.240342527285586,
-      "grad_norm": 0.0009029298089444637,
-      "learning_rate": 0.00019999752485303377,
-      "loss": 46.0,
-      "step": 29302
-    },
-    {
-      "epoch": 2.2404189842689757,
-      "grad_norm": 0.001753746997565031,
-      "learning_rate": 0.00019999752468403346,
-      "loss": 46.0,
-      "step": 29303
-    },
-    {
-      "epoch": 2.2404954412523654,
-      "grad_norm": 0.0020846514962613583,
-      "learning_rate": 0.00019999752451502735,
-      "loss": 46.0,
-      "step": 29304
-    },
-    {
-      "epoch": 2.240571898235755,
-      "grad_norm": 0.0007304241880774498,
-      "learning_rate": 0.00019999752434601546,
-      "loss": 46.0,
-      "step": 29305
-    },
-    {
-      "epoch": 2.240648355219145,
-      "grad_norm": 0.003506737295538187,
-      "learning_rate": 0.0001999975241769978,
-      "loss": 46.0,
-      "step": 29306
-    },
-    {
-      "epoch": 2.2407248122025347,
-      "grad_norm": 0.001534824026748538,
-      "learning_rate": 0.00019999752400797437,
-      "loss": 46.0,
-      "step": 29307
-    },
-    {
-      "epoch": 2.2408012691859245,
-      "grad_norm": 0.0008726065279915929,
-      "learning_rate": 0.0001999975238389452,
-      "loss": 46.0,
-      "step": 29308
-    },
-    {
-      "epoch": 2.240877726169314,
-      "grad_norm": 0.0010125860571861267,
-      "learning_rate": 0.00019999752366991024,
-      "loss": 46.0,
-      "step": 29309
-    },
-    {
-      "epoch": 2.2409541831527036,
-      "grad_norm": 0.004232482519000769,
-      "learning_rate": 0.00019999752350086954,
-      "loss": 46.0,
-      "step": 29310
-    },
-    {
-      "epoch": 2.2410306401360933,
-      "grad_norm": 0.002225664909929037,
-      "learning_rate": 0.00019999752333182302,
-      "loss": 46.0,
-      "step": 29311
-    },
-    {
-      "epoch": 2.241107097119483,
-      "grad_norm": 0.0006565770599991083,
-      "learning_rate": 0.00019999752316277074,
-      "loss": 46.0,
-      "step": 29312
-    },
-    {
-      "epoch": 2.241183554102873,
-      "grad_norm": 0.0008778393967077136,
-      "learning_rate": 0.00019999752299371273,
-      "loss": 46.0,
-      "step": 29313
-    },
-    {
-      "epoch": 2.2412600110862626,
-      "grad_norm": 0.0012631692225113511,
-      "learning_rate": 0.0001999975228246489,
-      "loss": 46.0,
-      "step": 29314
-    },
-    {
-      "epoch": 2.2413364680696524,
-      "grad_norm": 0.000767264689784497,
-      "learning_rate": 0.00019999752265557934,
-      "loss": 46.0,
-      "step": 29315
-    },
-    {
-      "epoch": 2.241412925053042,
-      "grad_norm": 0.0009724089177325368,
-      "learning_rate": 0.000199997522486504,
-      "loss": 46.0,
-      "step": 29316
-    },
-    {
-      "epoch": 2.241489382036432,
-      "grad_norm": 0.0004554348415695131,
-      "learning_rate": 0.0001999975223174229,
-      "loss": 46.0,
-      "step": 29317
-    },
-    {
-      "epoch": 2.2415658390198216,
-      "grad_norm": 0.0011803876841440797,
-      "learning_rate": 0.000199997522148336,
-      "loss": 46.0,
-      "step": 29318
-    },
-    {
-      "epoch": 2.241642296003211,
-      "grad_norm": 0.0009750897879712284,
-      "learning_rate": 0.00019999752197924335,
-      "loss": 46.0,
-      "step": 29319
-    },
-    {
-      "epoch": 2.2417187529866007,
-      "grad_norm": 0.00057305145310238,
-      "learning_rate": 0.00019999752181014494,
-      "loss": 46.0,
-      "step": 29320
-    },
-    {
-      "epoch": 2.2417952099699905,
-      "grad_norm": 0.0006997028249315917,
-      "learning_rate": 0.00019999752164104076,
-      "loss": 46.0,
-      "step": 29321
-    },
-    {
-      "epoch": 2.2418716669533802,
-      "grad_norm": 0.0007159091182984412,
-      "learning_rate": 0.00019999752147193078,
-      "loss": 46.0,
-      "step": 29322
-    },
-    {
-      "epoch": 2.24194812393677,
-      "grad_norm": 0.0004743227909784764,
-      "learning_rate": 0.00019999752130281506,
-      "loss": 46.0,
-      "step": 29323
-    },
-    {
-      "epoch": 2.2420245809201598,
-      "grad_norm": 0.0013039861805737019,
-      "learning_rate": 0.00019999752113369356,
-      "loss": 46.0,
-      "step": 29324
-    },
-    {
-      "epoch": 2.2421010379035495,
-      "grad_norm": 0.0015581886982545257,
-      "learning_rate": 0.0001999975209645663,
-      "loss": 46.0,
-      "step": 29325
-    },
-    {
-      "epoch": 2.2421774948869393,
-      "grad_norm": 0.0008446983410976827,
-      "learning_rate": 0.00019999752079543327,
-      "loss": 46.0,
-      "step": 29326
-    },
-    {
-      "epoch": 2.242253951870329,
-      "grad_norm": 0.001050122664310038,
-      "learning_rate": 0.00019999752062629445,
-      "loss": 46.0,
-      "step": 29327
-    },
-    {
-      "epoch": 2.242330408853719,
-      "grad_norm": 0.0005471666227094829,
-      "learning_rate": 0.0001999975204571499,
-      "loss": 46.0,
-      "step": 29328
-    },
-    {
-      "epoch": 2.2424068658371086,
-      "grad_norm": 0.003242134116590023,
-      "learning_rate": 0.00019999752028799955,
-      "loss": 46.0,
-      "step": 29329
-    },
-    {
-      "epoch": 2.2424833228204983,
-      "grad_norm": 0.001610635663382709,
-      "learning_rate": 0.00019999752011884344,
-      "loss": 46.0,
-      "step": 29330
-    },
-    {
-      "epoch": 2.2425597798038877,
-      "grad_norm": 0.009515034966170788,
-      "learning_rate": 0.00019999751994968155,
-      "loss": 46.0,
-      "step": 29331
-    },
-    {
-      "epoch": 2.2426362367872774,
-      "grad_norm": 0.0009892922826111317,
-      "learning_rate": 0.0001999975197805139,
-      "loss": 46.0,
-      "step": 29332
-    },
-    {
-      "epoch": 2.242712693770667,
-      "grad_norm": 0.0009540737955830991,
-      "learning_rate": 0.0001999975196113405,
-      "loss": 46.0,
-      "step": 29333
-    },
-    {
-      "epoch": 2.242789150754057,
-      "grad_norm": 0.0009146829834207892,
-      "learning_rate": 0.00019999751944216128,
-      "loss": 46.0,
-      "step": 29334
-    },
-    {
-      "epoch": 2.2428656077374467,
-      "grad_norm": 0.0017535848310217261,
-      "learning_rate": 0.0001999975192729763,
-      "loss": 46.0,
-      "step": 29335
-    },
-    {
-      "epoch": 2.2429420647208365,
-      "grad_norm": 0.003172148484736681,
-      "learning_rate": 0.00019999751910378558,
-      "loss": 46.0,
-      "step": 29336
-    },
-    {
-      "epoch": 2.243018521704226,
-      "grad_norm": 0.0013302986044436693,
-      "learning_rate": 0.00019999751893458908,
-      "loss": 46.0,
-      "step": 29337
-    },
-    {
-      "epoch": 2.243094978687616,
-      "grad_norm": 0.0008674630662426353,
-      "learning_rate": 0.0001999975187653868,
-      "loss": 46.0,
-      "step": 29338
-    },
-    {
-      "epoch": 2.2431714356710057,
-      "grad_norm": 0.0010425063082948327,
-      "learning_rate": 0.0001999975185961788,
-      "loss": 46.0,
-      "step": 29339
-    },
-    {
-      "epoch": 2.2432478926543955,
-      "grad_norm": 0.0006177789182402194,
-      "learning_rate": 0.000199997518426965,
-      "loss": 46.0,
-      "step": 29340
-    },
-    {
-      "epoch": 2.243324349637785,
-      "grad_norm": 0.0006449943757615983,
-      "learning_rate": 0.0001999975182577454,
-      "loss": 46.0,
-      "step": 29341
-    },
-    {
-      "epoch": 2.2434008066211746,
-      "grad_norm": 0.0010228544706478715,
-      "learning_rate": 0.00019999751808852007,
-      "loss": 46.0,
-      "step": 29342
-    },
-    {
-      "epoch": 2.2434772636045643,
-      "grad_norm": 0.0019454704597592354,
-      "learning_rate": 0.00019999751791928896,
-      "loss": 46.0,
-      "step": 29343
-    },
-    {
-      "epoch": 2.243553720587954,
-      "grad_norm": 0.0006619360647164285,
-      "learning_rate": 0.00019999751775005208,
-      "loss": 46.0,
-      "step": 29344
-    },
-    {
-      "epoch": 2.243630177571344,
-      "grad_norm": 0.0011464472627267241,
-      "learning_rate": 0.0001999975175808094,
-      "loss": 46.0,
-      "step": 29345
-    },
-    {
-      "epoch": 2.2437066345547336,
-      "grad_norm": 0.0009071673848666251,
-      "learning_rate": 0.000199997517411561,
-      "loss": 46.0,
-      "step": 29346
-    },
-    {
-      "epoch": 2.2437830915381234,
-      "grad_norm": 0.0016402152832597494,
-      "learning_rate": 0.0001999975172423068,
-      "loss": 46.0,
-      "step": 29347
-    },
-    {
-      "epoch": 2.243859548521513,
-      "grad_norm": 0.0016108324052765965,
-      "learning_rate": 0.00019999751707304683,
-      "loss": 46.0,
-      "step": 29348
-    },
-    {
-      "epoch": 2.243936005504903,
-      "grad_norm": 0.0009354056091979146,
-      "learning_rate": 0.0001999975169037811,
-      "loss": 46.0,
-      "step": 29349
-    },
-    {
-      "epoch": 2.2440124624882927,
-      "grad_norm": 0.0011478504166007042,
-      "learning_rate": 0.0001999975167345096,
-      "loss": 46.0,
-      "step": 29350
-    },
-    {
-      "epoch": 2.2440889194716824,
-      "grad_norm": 0.0016720503335818648,
-      "learning_rate": 0.00019999751656523234,
-      "loss": 46.0,
-      "step": 29351
-    },
-    {
-      "epoch": 2.244165376455072,
-      "grad_norm": 0.0006281096138991416,
-      "learning_rate": 0.0001999975163959493,
-      "loss": 46.0,
-      "step": 29352
-    },
-    {
-      "epoch": 2.2442418334384615,
-      "grad_norm": 0.00035824498627334833,
-      "learning_rate": 0.0001999975162266605,
-      "loss": 46.0,
-      "step": 29353
-    },
-    {
-      "epoch": 2.2443182904218513,
-      "grad_norm": 0.0005668998346664011,
-      "learning_rate": 0.0001999975160573659,
-      "loss": 46.0,
-      "step": 29354
-    },
-    {
-      "epoch": 2.244394747405241,
-      "grad_norm": 0.0008568732882849872,
-      "learning_rate": 0.00019999751588806557,
-      "loss": 46.0,
-      "step": 29355
-    },
-    {
-      "epoch": 2.244471204388631,
-      "grad_norm": 0.0013889207039028406,
-      "learning_rate": 0.00019999751571875946,
-      "loss": 46.0,
-      "step": 29356
-    },
-    {
-      "epoch": 2.2445476613720206,
-      "grad_norm": 0.001104099559597671,
-      "learning_rate": 0.00019999751554944758,
-      "loss": 46.0,
-      "step": 29357
-    },
-    {
-      "epoch": 2.2446241183554103,
-      "grad_norm": 0.0008470212342217565,
-      "learning_rate": 0.00019999751538012992,
-      "loss": 46.0,
-      "step": 29358
-    },
-    {
-      "epoch": 2.2447005753388,
-      "grad_norm": 0.001955494051799178,
-      "learning_rate": 0.0001999975152108065,
-      "loss": 46.0,
-      "step": 29359
-    },
-    {
-      "epoch": 2.24477703232219,
-      "grad_norm": 0.001402291003614664,
-      "learning_rate": 0.0001999975150414773,
-      "loss": 46.0,
-      "step": 29360
-    },
-    {
-      "epoch": 2.2448534893055796,
-      "grad_norm": 0.0017629220383241773,
-      "learning_rate": 0.00019999751487214233,
-      "loss": 46.0,
-      "step": 29361
-    },
-    {
-      "epoch": 2.2449299462889694,
-      "grad_norm": 0.0005411452730186284,
-      "learning_rate": 0.0001999975147028016,
-      "loss": 46.0,
-      "step": 29362
-    },
-    {
-      "epoch": 2.2450064032723587,
-      "grad_norm": 0.0006768542807549238,
-      "learning_rate": 0.0001999975145334551,
-      "loss": 46.0,
-      "step": 29363
-    },
-    {
-      "epoch": 2.2450828602557484,
-      "grad_norm": 0.0007872622227296233,
-      "learning_rate": 0.00019999751436410287,
-      "loss": 46.0,
-      "step": 29364
-    },
-    {
-      "epoch": 2.245159317239138,
-      "grad_norm": 0.0003998048778157681,
-      "learning_rate": 0.00019999751419474482,
-      "loss": 46.0,
-      "step": 29365
-    },
-    {
-      "epoch": 2.245235774222528,
-      "grad_norm": 0.0008432682952843606,
-      "learning_rate": 0.000199997514025381,
-      "loss": 46.0,
-      "step": 29366
-    },
-    {
-      "epoch": 2.2453122312059177,
-      "grad_norm": 0.0018499908037483692,
-      "learning_rate": 0.00019999751385601141,
-      "loss": 46.0,
-      "step": 29367
-    },
-    {
-      "epoch": 2.2453886881893075,
-      "grad_norm": 0.0022874511778354645,
-      "learning_rate": 0.00019999751368663605,
-      "loss": 46.0,
-      "step": 29368
-    },
-    {
-      "epoch": 2.2454651451726972,
-      "grad_norm": 0.0006427046610042453,
-      "learning_rate": 0.00019999751351725494,
-      "loss": 46.0,
-      "step": 29369
-    },
-    {
-      "epoch": 2.245541602156087,
-      "grad_norm": 0.0009616664028726518,
-      "learning_rate": 0.00019999751334786808,
-      "loss": 46.0,
-      "step": 29370
-    },
-    {
-      "epoch": 2.2456180591394768,
-      "grad_norm": 0.0009112671250477433,
-      "learning_rate": 0.00019999751317847543,
-      "loss": 46.0,
-      "step": 29371
-    },
-    {
-      "epoch": 2.2456945161228665,
-      "grad_norm": 0.00039321009535342455,
-      "learning_rate": 0.000199997513009077,
-      "loss": 46.0,
-      "step": 29372
-    },
-    {
-      "epoch": 2.2457709731062563,
-      "grad_norm": 0.0014207576168701053,
-      "learning_rate": 0.00019999751283967282,
-      "loss": 46.0,
-      "step": 29373
-    },
-    {
-      "epoch": 2.245847430089646,
-      "grad_norm": 0.0010622048284858465,
-      "learning_rate": 0.00019999751267026284,
-      "loss": 46.0,
-      "step": 29374
-    },
-    {
-      "epoch": 2.2459238870730354,
-      "grad_norm": 0.0010640205582603812,
-      "learning_rate": 0.0001999975125008471,
-      "loss": 46.0,
-      "step": 29375
-    },
-    {
-      "epoch": 2.246000344056425,
-      "grad_norm": 0.000943817023653537,
-      "learning_rate": 0.0001999975123314256,
-      "loss": 46.0,
-      "step": 29376
-    },
-    {
-      "epoch": 2.246076801039815,
-      "grad_norm": 0.0022350195795297623,
-      "learning_rate": 0.00019999751216199832,
-      "loss": 46.0,
-      "step": 29377
-    },
-    {
-      "epoch": 2.2461532580232046,
-      "grad_norm": 0.0010578795336186886,
-      "learning_rate": 0.00019999751199256528,
-      "loss": 46.0,
-      "step": 29378
-    },
-    {
-      "epoch": 2.2462297150065944,
-      "grad_norm": 0.0008161651203408837,
-      "learning_rate": 0.0001999975118231265,
-      "loss": 46.0,
-      "step": 29379
-    },
-    {
-      "epoch": 2.246306171989984,
-      "grad_norm": 0.0018506267806515098,
-      "learning_rate": 0.0001999975116536819,
-      "loss": 46.0,
-      "step": 29380
-    },
-    {
-      "epoch": 2.246382628973374,
-      "grad_norm": 0.00091931241331622,
-      "learning_rate": 0.00019999751148423153,
-      "loss": 46.0,
-      "step": 29381
-    },
-    {
-      "epoch": 2.2464590859567637,
-      "grad_norm": 0.0011082744458690286,
-      "learning_rate": 0.00019999751131477545,
-      "loss": 46.0,
-      "step": 29382
-    },
-    {
-      "epoch": 2.2465355429401535,
-      "grad_norm": 0.0008622603490948677,
-      "learning_rate": 0.00019999751114531354,
-      "loss": 46.0,
-      "step": 29383
-    },
-    {
-      "epoch": 2.246611999923543,
-      "grad_norm": 0.0023591257631778717,
-      "learning_rate": 0.00019999751097584588,
-      "loss": 46.0,
-      "step": 29384
-    },
-    {
-      "epoch": 2.2466884569069325,
-      "grad_norm": 0.0021558767184615135,
-      "learning_rate": 0.00019999751080637248,
-      "loss": 46.0,
-      "step": 29385
-    },
-    {
-      "epoch": 2.2467649138903223,
-      "grad_norm": 0.004051904659718275,
-      "learning_rate": 0.00019999751063689327,
-      "loss": 46.0,
-      "step": 29386
-    },
-    {
-      "epoch": 2.246841370873712,
-      "grad_norm": 0.00044367031659930944,
-      "learning_rate": 0.00019999751046740832,
-      "loss": 46.0,
-      "step": 29387
-    },
-    {
-      "epoch": 2.246917827857102,
-      "grad_norm": 0.002209751633927226,
-      "learning_rate": 0.00019999751029791757,
-      "loss": 46.0,
-      "step": 29388
-    },
-    {
-      "epoch": 2.2469942848404916,
-      "grad_norm": 0.00037307501770555973,
-      "learning_rate": 0.00019999751012842108,
-      "loss": 46.0,
-      "step": 29389
-    },
-    {
-      "epoch": 2.2470707418238813,
-      "grad_norm": 0.0007075180765241385,
-      "learning_rate": 0.0001999975099589188,
-      "loss": 46.0,
-      "step": 29390
-    },
-    {
-      "epoch": 2.247147198807271,
-      "grad_norm": 0.0005778814665973186,
-      "learning_rate": 0.00019999750978941076,
-      "loss": 46.0,
-      "step": 29391
-    },
-    {
-      "epoch": 2.247223655790661,
-      "grad_norm": 0.001449332688935101,
-      "learning_rate": 0.00019999750961989694,
-      "loss": 46.0,
-      "step": 29392
-    },
-    {
-      "epoch": 2.2473001127740506,
-      "grad_norm": 0.0008784688543528318,
-      "learning_rate": 0.00019999750945037738,
-      "loss": 46.0,
-      "step": 29393
-    },
-    {
-      "epoch": 2.2473765697574404,
-      "grad_norm": 0.0005454125930555165,
-      "learning_rate": 0.00019999750928085202,
-      "loss": 46.0,
-      "step": 29394
-    },
-    {
-      "epoch": 2.24745302674083,
-      "grad_norm": 0.0008241348550654948,
-      "learning_rate": 0.0001999975091113209,
-      "loss": 46.0,
-      "step": 29395
-    },
-    {
-      "epoch": 2.2475294837242195,
-      "grad_norm": 0.0003697400097735226,
-      "learning_rate": 0.00019999750894178402,
-      "loss": 46.0,
-      "step": 29396
-    },
-    {
-      "epoch": 2.247605940707609,
-      "grad_norm": 0.0016068790573626757,
-      "learning_rate": 0.00019999750877224134,
-      "loss": 46.0,
-      "step": 29397
-    },
-    {
-      "epoch": 2.247682397690999,
-      "grad_norm": 0.0020260708406567574,
-      "learning_rate": 0.00019999750860269294,
-      "loss": 46.0,
-      "step": 29398
-    },
-    {
-      "epoch": 2.2477588546743887,
-      "grad_norm": 0.0008248375379480422,
-      "learning_rate": 0.00019999750843313873,
-      "loss": 46.0,
-      "step": 29399
-    },
-    {
-      "epoch": 2.2478353116577785,
-      "grad_norm": 0.00041719956789165735,
-      "learning_rate": 0.00019999750826357878,
-      "loss": 46.0,
-      "step": 29400
-    },
-    {
-      "epoch": 2.2479117686411683,
-      "grad_norm": 0.003310014959424734,
-      "learning_rate": 0.00019999750809401303,
-      "loss": 46.0,
-      "step": 29401
-    },
-    {
-      "epoch": 2.247988225624558,
-      "grad_norm": 0.0011237786384299397,
-      "learning_rate": 0.0001999975079244415,
-      "loss": 46.0,
-      "step": 29402
-    },
-    {
-      "epoch": 2.248064682607948,
-      "grad_norm": 0.0049889967776834965,
-      "learning_rate": 0.00019999750775486427,
-      "loss": 46.0,
-      "step": 29403
-    },
-    {
-      "epoch": 2.2481411395913375,
-      "grad_norm": 0.0010788104264065623,
-      "learning_rate": 0.00019999750758528122,
-      "loss": 46.0,
-      "step": 29404
-    },
-    {
-      "epoch": 2.2482175965747273,
-      "grad_norm": 0.0011255888966843486,
-      "learning_rate": 0.0001999975074156924,
-      "loss": 46.0,
-      "step": 29405
-    },
-    {
-      "epoch": 2.248294053558117,
-      "grad_norm": 0.0010205833241343498,
-      "learning_rate": 0.00019999750724609782,
-      "loss": 46.0,
-      "step": 29406
-    },
-    {
-      "epoch": 2.2483705105415064,
-      "grad_norm": 0.0005901691038161516,
-      "learning_rate": 0.00019999750707649748,
-      "loss": 46.0,
-      "step": 29407
-    },
-    {
-      "epoch": 2.248446967524896,
-      "grad_norm": 0.0007039479096420109,
-      "learning_rate": 0.00019999750690689134,
-      "loss": 46.0,
-      "step": 29408
-    },
-    {
-      "epoch": 2.248523424508286,
-      "grad_norm": 0.0026410738937556744,
-      "learning_rate": 0.00019999750673727946,
-      "loss": 46.0,
-      "step": 29409
-    },
-    {
-      "epoch": 2.2485998814916757,
-      "grad_norm": 0.0024696753825992346,
-      "learning_rate": 0.0001999975065676618,
-      "loss": 46.0,
-      "step": 29410
-    },
-    {
-      "epoch": 2.2486763384750654,
-      "grad_norm": 0.000584301189519465,
-      "learning_rate": 0.00019999750639803837,
-      "loss": 46.0,
-      "step": 29411
-    },
-    {
-      "epoch": 2.248752795458455,
-      "grad_norm": 0.0004448372928891331,
-      "learning_rate": 0.00019999750622840917,
-      "loss": 46.0,
-      "step": 29412
-    },
-    {
-      "epoch": 2.248829252441845,
-      "grad_norm": 0.004667339380830526,
-      "learning_rate": 0.00019999750605877422,
-      "loss": 46.0,
-      "step": 29413
-    },
-    {
-      "epoch": 2.2489057094252347,
-      "grad_norm": 0.0008403750834986567,
-      "learning_rate": 0.0001999975058891335,
-      "loss": 46.0,
-      "step": 29414
-    },
-    {
-      "epoch": 2.2489821664086245,
-      "grad_norm": 0.0008512880303896964,
-      "learning_rate": 0.000199997505719487,
-      "loss": 46.0,
-      "step": 29415
-    },
-    {
-      "epoch": 2.2490586233920142,
-      "grad_norm": 0.0006080466555431485,
-      "learning_rate": 0.0001999975055498347,
-      "loss": 46.0,
-      "step": 29416
-    },
-    {
-      "epoch": 2.249135080375404,
-      "grad_norm": 0.001602043048478663,
-      "learning_rate": 0.00019999750538017666,
-      "loss": 46.0,
-      "step": 29417
-    },
-    {
-      "epoch": 2.2492115373587933,
-      "grad_norm": 0.0003134894941467792,
-      "learning_rate": 0.00019999750521051285,
-      "loss": 46.0,
-      "step": 29418
-    },
-    {
-      "epoch": 2.249287994342183,
-      "grad_norm": 0.000761925708502531,
-      "learning_rate": 0.00019999750504084328,
-      "loss": 46.0,
-      "step": 29419
-    },
-    {
-      "epoch": 2.249364451325573,
-      "grad_norm": 0.0009567033848725259,
-      "learning_rate": 0.00019999750487116792,
-      "loss": 46.0,
-      "step": 29420
-    },
-    {
-      "epoch": 2.2494409083089626,
-      "grad_norm": 0.0008765193051658571,
-      "learning_rate": 0.0001999975047014868,
-      "loss": 46.0,
-      "step": 29421
-    },
-    {
-      "epoch": 2.2495173652923524,
-      "grad_norm": 0.0009923727484419942,
-      "learning_rate": 0.0001999975045317999,
-      "loss": 46.0,
-      "step": 29422
-    },
-    {
-      "epoch": 2.249593822275742,
-      "grad_norm": 0.003081100294366479,
-      "learning_rate": 0.00019999750436210724,
-      "loss": 46.0,
-      "step": 29423
-    },
-    {
-      "epoch": 2.249670279259132,
-      "grad_norm": 0.0005716641899198294,
-      "learning_rate": 0.00019999750419240884,
-      "loss": 46.0,
-      "step": 29424
-    },
-    {
-      "epoch": 2.2497467362425216,
-      "grad_norm": 0.001090982579626143,
-      "learning_rate": 0.00019999750402270464,
-      "loss": 46.0,
-      "step": 29425
-    },
-    {
-      "epoch": 2.2498231932259114,
-      "grad_norm": 0.0011637975694611669,
-      "learning_rate": 0.00019999750385299466,
-      "loss": 46.0,
-      "step": 29426
-    },
-    {
-      "epoch": 2.249899650209301,
-      "grad_norm": 0.0017158568371087313,
-      "learning_rate": 0.0001999975036832789,
-      "loss": 46.0,
-      "step": 29427
-    },
-    {
-      "epoch": 2.2499761071926905,
-      "grad_norm": 0.0006134028662927449,
-      "learning_rate": 0.00019999750351355742,
-      "loss": 46.0,
-      "step": 29428
-    },
-    {
-      "epoch": 2.2500525641760802,
-      "grad_norm": 0.0015880957944318652,
-      "learning_rate": 0.00019999750334383015,
-      "loss": 46.0,
-      "step": 29429
-    },
-    {
-      "epoch": 2.25012902115947,
-      "grad_norm": 0.0008633694960735738,
-      "learning_rate": 0.00019999750317409708,
-      "loss": 46.0,
-      "step": 29430
-    },
-    {
-      "epoch": 2.2502054781428598,
-      "grad_norm": 0.001385657349601388,
-      "learning_rate": 0.0001999975030043583,
-      "loss": 46.0,
-      "step": 29431
-    },
-    {
-      "epoch": 2.2502819351262495,
-      "grad_norm": 0.0020005065016448498,
-      "learning_rate": 0.00019999750283461373,
-      "loss": 46.0,
-      "step": 29432
-    },
-    {
-      "epoch": 2.2503583921096393,
-      "grad_norm": 0.0009481015149503946,
-      "learning_rate": 0.00019999750266486334,
-      "loss": 46.0,
-      "step": 29433
-    },
-    {
-      "epoch": 2.250434849093029,
-      "grad_norm": 0.0007318708812817931,
-      "learning_rate": 0.00019999750249510723,
-      "loss": 46.0,
-      "step": 29434
-    },
-    {
-      "epoch": 2.250511306076419,
-      "grad_norm": 0.0009338194504380226,
-      "learning_rate": 0.00019999750232534534,
-      "loss": 46.0,
-      "step": 29435
-    },
-    {
-      "epoch": 2.2505877630598086,
-      "grad_norm": 0.003099459456279874,
-      "learning_rate": 0.0001999975021555777,
-      "loss": 46.0,
-      "step": 29436
-    },
-    {
-      "epoch": 2.2506642200431983,
-      "grad_norm": 0.0013390317326411605,
-      "learning_rate": 0.00019999750198580426,
-      "loss": 46.0,
-      "step": 29437
-    },
-    {
-      "epoch": 2.250740677026588,
-      "grad_norm": 0.0012756186770275235,
-      "learning_rate": 0.00019999750181602503,
-      "loss": 46.0,
-      "step": 29438
-    },
-    {
-      "epoch": 2.250817134009978,
-      "grad_norm": 0.0015720747178420424,
-      "learning_rate": 0.00019999750164624008,
-      "loss": 46.0,
-      "step": 29439
-    },
-    {
-      "epoch": 2.250893590993367,
-      "grad_norm": 0.012928523123264313,
-      "learning_rate": 0.00019999750147644933,
-      "loss": 46.0,
-      "step": 29440
-    },
-    {
-      "epoch": 2.250970047976757,
-      "grad_norm": 0.0005817175842821598,
-      "learning_rate": 0.00019999750130665284,
-      "loss": 46.0,
-      "step": 29441
-    },
-    {
-      "epoch": 2.2510465049601467,
-      "grad_norm": 0.0013524278765544295,
-      "learning_rate": 0.00019999750113685057,
-      "loss": 46.0,
-      "step": 29442
-    },
-    {
-      "epoch": 2.2511229619435364,
-      "grad_norm": 0.0006919489824213088,
-      "learning_rate": 0.0001999975009670425,
-      "loss": 46.0,
-      "step": 29443
-    },
-    {
-      "epoch": 2.251199418926926,
-      "grad_norm": 0.0015825749142095447,
-      "learning_rate": 0.0001999975007972287,
-      "loss": 46.0,
-      "step": 29444
-    },
-    {
-      "epoch": 2.251275875910316,
-      "grad_norm": 0.0010053508449345827,
-      "learning_rate": 0.00019999750062740912,
-      "loss": 46.0,
-      "step": 29445
-    },
-    {
-      "epoch": 2.2513523328937057,
-      "grad_norm": 0.0017596713732928038,
-      "learning_rate": 0.00019999750045758376,
-      "loss": 46.0,
-      "step": 29446
-    },
-    {
-      "epoch": 2.2514287898770955,
-      "grad_norm": 0.001610561739653349,
-      "learning_rate": 0.00019999750028775265,
-      "loss": 46.0,
-      "step": 29447
-    },
-    {
-      "epoch": 2.2515052468604853,
-      "grad_norm": 0.001406658673658967,
-      "learning_rate": 0.00019999750011791574,
-      "loss": 46.0,
-      "step": 29448
-    },
-    {
-      "epoch": 2.251581703843875,
-      "grad_norm": 0.0011525223962962627,
-      "learning_rate": 0.00019999749994807308,
-      "loss": 46.0,
-      "step": 29449
-    },
-    {
-      "epoch": 2.2516581608272643,
-      "grad_norm": 0.000749699305742979,
-      "learning_rate": 0.00019999749977822465,
-      "loss": 46.0,
-      "step": 29450
-    },
-    {
-      "epoch": 2.251734617810654,
-      "grad_norm": 0.001271737739443779,
-      "learning_rate": 0.00019999749960837045,
-      "loss": 46.0,
-      "step": 29451
-    },
-    {
-      "epoch": 2.251811074794044,
-      "grad_norm": 0.0007768102223053575,
-      "learning_rate": 0.0001999974994385105,
-      "loss": 46.0,
-      "step": 29452
-    },
-    {
-      "epoch": 2.2518875317774336,
-      "grad_norm": 0.0006906852358952165,
-      "learning_rate": 0.00019999749926864475,
-      "loss": 46.0,
-      "step": 29453
-    },
-    {
-      "epoch": 2.2519639887608234,
-      "grad_norm": 0.0012180572375655174,
-      "learning_rate": 0.00019999749909877323,
-      "loss": 46.0,
-      "step": 29454
-    },
-    {
-      "epoch": 2.252040445744213,
-      "grad_norm": 0.0012188206892460585,
-      "learning_rate": 0.00019999749892889596,
-      "loss": 46.0,
-      "step": 29455
-    },
-    {
-      "epoch": 2.252116902727603,
-      "grad_norm": 0.0030461689457297325,
-      "learning_rate": 0.00019999749875901292,
-      "loss": 46.0,
-      "step": 29456
-    },
-    {
-      "epoch": 2.2521933597109927,
-      "grad_norm": 0.0017065851716324687,
-      "learning_rate": 0.0001999974985891241,
-      "loss": 46.0,
-      "step": 29457
-    },
-    {
-      "epoch": 2.2522698166943824,
-      "grad_norm": 0.001509082387201488,
-      "learning_rate": 0.00019999749841922952,
-      "loss": 46.0,
-      "step": 29458
-    },
-    {
-      "epoch": 2.252346273677772,
-      "grad_norm": 0.000760240713134408,
-      "learning_rate": 0.00019999749824932915,
-      "loss": 46.0,
-      "step": 29459
-    },
-    {
-      "epoch": 2.252422730661162,
-      "grad_norm": 0.0006394116207957268,
-      "learning_rate": 0.00019999749807942302,
-      "loss": 46.0,
-      "step": 29460
-    },
-    {
-      "epoch": 2.2524991876445517,
-      "grad_norm": 0.0008832362364046276,
-      "learning_rate": 0.00019999749790951116,
-      "loss": 46.0,
-      "step": 29461
-    },
-    {
-      "epoch": 2.252575644627941,
-      "grad_norm": 0.0010208101011812687,
-      "learning_rate": 0.00019999749773959348,
-      "loss": 46.0,
-      "step": 29462
-    },
-    {
-      "epoch": 2.252652101611331,
-      "grad_norm": 0.000825251976493746,
-      "learning_rate": 0.00019999749756967005,
-      "loss": 46.0,
-      "step": 29463
-    },
-    {
-      "epoch": 2.2527285585947205,
-      "grad_norm": 0.0006938123260624707,
-      "learning_rate": 0.00019999749739974085,
-      "loss": 46.0,
-      "step": 29464
-    },
-    {
-      "epoch": 2.2528050155781103,
-      "grad_norm": 0.004282000940293074,
-      "learning_rate": 0.0001999974972298059,
-      "loss": 46.0,
-      "step": 29465
-    },
-    {
-      "epoch": 2.2528814725615,
-      "grad_norm": 0.0006187624530866742,
-      "learning_rate": 0.00019999749705986515,
-      "loss": 46.0,
-      "step": 29466
-    },
-    {
-      "epoch": 2.25295792954489,
-      "grad_norm": 0.0004884095978923142,
-      "learning_rate": 0.00019999749688991863,
-      "loss": 46.0,
-      "step": 29467
-    },
-    {
-      "epoch": 2.2530343865282796,
-      "grad_norm": 0.0007862441707402468,
-      "learning_rate": 0.00019999749671996634,
-      "loss": 46.0,
-      "step": 29468
-    },
-    {
-      "epoch": 2.2531108435116693,
-      "grad_norm": 0.002405275823548436,
-      "learning_rate": 0.0001999974965500083,
-      "loss": 46.0,
-      "step": 29469
-    },
-    {
-      "epoch": 2.253187300495059,
-      "grad_norm": 0.0012418583501130342,
-      "learning_rate": 0.0001999974963800445,
-      "loss": 46.0,
-      "step": 29470
-    },
-    {
-      "epoch": 2.253263757478449,
-      "grad_norm": 0.0008651570533402264,
-      "learning_rate": 0.00019999749621007492,
-      "loss": 46.0,
-      "step": 29471
-    },
-    {
-      "epoch": 2.253340214461838,
-      "grad_norm": 0.003094903426244855,
-      "learning_rate": 0.00019999749604009956,
-      "loss": 46.0,
-      "step": 29472
-    },
-    {
-      "epoch": 2.253416671445228,
-      "grad_norm": 0.0023649903014302254,
-      "learning_rate": 0.00019999749587011842,
-      "loss": 46.0,
-      "step": 29473
-    },
-    {
-      "epoch": 2.2534931284286177,
-      "grad_norm": 0.0015670970315113664,
-      "learning_rate": 0.00019999749570013157,
-      "loss": 46.0,
-      "step": 29474
-    },
-    {
-      "epoch": 2.2535695854120075,
-      "grad_norm": 0.0008354040328413248,
-      "learning_rate": 0.00019999749553013886,
-      "loss": 46.0,
-      "step": 29475
-    },
-    {
-      "epoch": 2.2536460423953972,
-      "grad_norm": 0.0007118411595001817,
-      "learning_rate": 0.00019999749536014046,
-      "loss": 46.0,
-      "step": 29476
-    },
-    {
-      "epoch": 2.253722499378787,
-      "grad_norm": 0.0009362486889585853,
-      "learning_rate": 0.00019999749519013625,
-      "loss": 46.0,
-      "step": 29477
-    },
-    {
-      "epoch": 2.2537989563621768,
-      "grad_norm": 0.0008002620888873935,
-      "learning_rate": 0.00019999749502012625,
-      "loss": 46.0,
-      "step": 29478
-    },
-    {
-      "epoch": 2.2538754133455665,
-      "grad_norm": 0.000947941851336509,
-      "learning_rate": 0.00019999749485011053,
-      "loss": 46.0,
-      "step": 29479
-    },
-    {
-      "epoch": 2.2539518703289563,
-      "grad_norm": 0.0005985077586956322,
-      "learning_rate": 0.000199997494680089,
-      "loss": 46.0,
-      "step": 29480
-    },
-    {
-      "epoch": 2.254028327312346,
-      "grad_norm": 0.0011883239494636655,
-      "learning_rate": 0.00019999749451006174,
-      "loss": 46.0,
-      "step": 29481
-    },
-    {
-      "epoch": 2.254104784295736,
-      "grad_norm": 0.001346498727798462,
-      "learning_rate": 0.00019999749434002867,
-      "loss": 46.0,
-      "step": 29482
-    },
-    {
-      "epoch": 2.2541812412791256,
-      "grad_norm": 0.0012821832206100225,
-      "learning_rate": 0.00019999749416998988,
-      "loss": 46.0,
-      "step": 29483
-    },
-    {
-      "epoch": 2.254257698262515,
-      "grad_norm": 0.0011586984619498253,
-      "learning_rate": 0.0001999974939999453,
-      "loss": 46.0,
-      "step": 29484
-    },
-    {
-      "epoch": 2.2543341552459046,
-      "grad_norm": 0.0006998831522651017,
-      "learning_rate": 0.00019999749382989493,
-      "loss": 46.0,
-      "step": 29485
-    },
-    {
-      "epoch": 2.2544106122292944,
-      "grad_norm": 0.0012054620310664177,
-      "learning_rate": 0.00019999749365983883,
-      "loss": 46.0,
-      "step": 29486
-    },
-    {
-      "epoch": 2.254487069212684,
-      "grad_norm": 0.0005119956331327558,
-      "learning_rate": 0.0001999974934897769,
-      "loss": 46.0,
-      "step": 29487
-    },
-    {
-      "epoch": 2.254563526196074,
-      "grad_norm": 0.006121808663010597,
-      "learning_rate": 0.00019999749331970926,
-      "loss": 46.0,
-      "step": 29488
-    },
-    {
-      "epoch": 2.2546399831794637,
-      "grad_norm": 0.002351575531065464,
-      "learning_rate": 0.00019999749314963584,
-      "loss": 46.0,
-      "step": 29489
-    },
-    {
-      "epoch": 2.2547164401628534,
-      "grad_norm": 0.0008991486392915249,
-      "learning_rate": 0.0001999974929795566,
-      "loss": 46.0,
-      "step": 29490
-    },
-    {
-      "epoch": 2.254792897146243,
-      "grad_norm": 0.001591112231835723,
-      "learning_rate": 0.00019999749280947163,
-      "loss": 46.0,
-      "step": 29491
-    },
-    {
-      "epoch": 2.254869354129633,
-      "grad_norm": 0.0016514176968485117,
-      "learning_rate": 0.0001999974926393809,
-      "loss": 46.0,
-      "step": 29492
-    },
-    {
-      "epoch": 2.2549458111130227,
-      "grad_norm": 0.0009082417236641049,
-      "learning_rate": 0.0001999974924692844,
-      "loss": 46.0,
-      "step": 29493
-    },
-    {
-      "epoch": 2.255022268096412,
-      "grad_norm": 0.0008060195250436664,
-      "learning_rate": 0.00019999749229918212,
-      "loss": 46.0,
-      "step": 29494
-    },
-    {
-      "epoch": 2.255098725079802,
-      "grad_norm": 0.0004528650897555053,
-      "learning_rate": 0.00019999749212907405,
-      "loss": 46.0,
-      "step": 29495
-    },
-    {
-      "epoch": 2.2551751820631916,
-      "grad_norm": 0.0017089232569560409,
-      "learning_rate": 0.0001999974919589602,
-      "loss": 46.0,
-      "step": 29496
-    },
-    {
-      "epoch": 2.2552516390465813,
-      "grad_norm": 0.0009664823883213103,
-      "learning_rate": 0.00019999749178884065,
-      "loss": 46.0,
-      "step": 29497
-    },
-    {
-      "epoch": 2.255328096029971,
-      "grad_norm": 0.005389222875237465,
-      "learning_rate": 0.0001999974916187153,
-      "loss": 46.0,
-      "step": 29498
-    },
-    {
-      "epoch": 2.255404553013361,
-      "grad_norm": 0.0006029255455359817,
-      "learning_rate": 0.00019999749144858418,
-      "loss": 46.0,
-      "step": 29499
-    },
-    {
-      "epoch": 2.2554810099967506,
-      "grad_norm": 0.0005662412731908262,
-      "learning_rate": 0.00019999749127844727,
-      "loss": 46.0,
-      "step": 29500
-    },
-    {
-      "epoch": 2.2555574669801404,
-      "grad_norm": 0.0008485973230563104,
-      "learning_rate": 0.0001999974911083046,
-      "loss": 46.0,
-      "step": 29501
-    },
-    {
-      "epoch": 2.25563392396353,
-      "grad_norm": 0.00564415380358696,
-      "learning_rate": 0.00019999749093815617,
-      "loss": 46.0,
-      "step": 29502
-    },
-    {
-      "epoch": 2.25571038094692,
-      "grad_norm": 0.0012757977237924933,
-      "learning_rate": 0.00019999749076800196,
-      "loss": 46.0,
-      "step": 29503
-    },
-    {
-      "epoch": 2.2557868379303097,
-      "grad_norm": 0.0005799197242595255,
-      "learning_rate": 0.000199997490597842,
-      "loss": 46.0,
-      "step": 29504
-    },
-    {
-      "epoch": 2.2558632949136994,
-      "grad_norm": 0.0013380179880186915,
-      "learning_rate": 0.00019999749042767624,
-      "loss": 46.0,
-      "step": 29505
-    },
-    {
-      "epoch": 2.2559397518970887,
-      "grad_norm": 0.0011360276257619262,
-      "learning_rate": 0.00019999749025750475,
-      "loss": 46.0,
-      "step": 29506
-    },
-    {
-      "epoch": 2.2560162088804785,
-      "grad_norm": 0.0008341424399986863,
-      "learning_rate": 0.00019999749008732745,
-      "loss": 46.0,
-      "step": 29507
-    },
-    {
-      "epoch": 2.2560926658638683,
-      "grad_norm": 0.0003634248860180378,
-      "learning_rate": 0.00019999748991714439,
-      "loss": 46.0,
-      "step": 29508
-    },
-    {
-      "epoch": 2.256169122847258,
-      "grad_norm": 0.00027449720073491335,
-      "learning_rate": 0.0001999974897469556,
-      "loss": 46.0,
-      "step": 29509
-    },
-    {
-      "epoch": 2.2562455798306478,
-      "grad_norm": 0.0008891826728358865,
-      "learning_rate": 0.00019999748957676098,
-      "loss": 46.0,
-      "step": 29510
-    },
-    {
-      "epoch": 2.2563220368140375,
-      "grad_norm": 0.002426140708848834,
-      "learning_rate": 0.00019999748940656065,
-      "loss": 46.0,
-      "step": 29511
-    },
-    {
-      "epoch": 2.2563984937974273,
-      "grad_norm": 0.0028874347917735577,
-      "learning_rate": 0.00019999748923635454,
-      "loss": 46.0,
-      "step": 29512
-    },
-    {
-      "epoch": 2.256474950780817,
-      "grad_norm": 0.0008574674720875919,
-      "learning_rate": 0.0001999974890661426,
-      "loss": 46.0,
-      "step": 29513
-    },
-    {
-      "epoch": 2.256551407764207,
-      "grad_norm": 0.0010328348726034164,
-      "learning_rate": 0.00019999748889592495,
-      "loss": 46.0,
-      "step": 29514
-    },
-    {
-      "epoch": 2.256627864747596,
-      "grad_norm": 0.0008847320568747818,
-      "learning_rate": 0.00019999748872570153,
-      "loss": 46.0,
-      "step": 29515
-    },
-    {
-      "epoch": 2.256704321730986,
-      "grad_norm": 0.0007886847597546875,
-      "learning_rate": 0.0001999974885554723,
-      "loss": 46.0,
-      "step": 29516
-    },
-    {
-      "epoch": 2.2567807787143757,
-      "grad_norm": 0.0005528398905880749,
-      "learning_rate": 0.00019999748838523735,
-      "loss": 46.0,
-      "step": 29517
-    },
-    {
-      "epoch": 2.2568572356977654,
-      "grad_norm": 0.001189409988000989,
-      "learning_rate": 0.0001999974882149966,
-      "loss": 46.0,
-      "step": 29518
-    },
-    {
-      "epoch": 2.256933692681155,
-      "grad_norm": 0.0007394845597445965,
-      "learning_rate": 0.00019999748804475008,
-      "loss": 46.0,
-      "step": 29519
-    },
-    {
-      "epoch": 2.257010149664545,
-      "grad_norm": 0.004855920094996691,
-      "learning_rate": 0.00019999748787449782,
-      "loss": 46.0,
-      "step": 29520
-    },
-    {
-      "epoch": 2.2570866066479347,
-      "grad_norm": 0.0017416721675544977,
-      "learning_rate": 0.00019999748770423975,
-      "loss": 46.0,
-      "step": 29521
-    },
-    {
-      "epoch": 2.2571630636313245,
-      "grad_norm": 0.001229771296493709,
-      "learning_rate": 0.00019999748753397593,
-      "loss": 46.0,
-      "step": 29522
-    },
-    {
-      "epoch": 2.2572395206147142,
-      "grad_norm": 0.0015130923129618168,
-      "learning_rate": 0.00019999748736370635,
-      "loss": 46.0,
-      "step": 29523
-    },
-    {
-      "epoch": 2.257315977598104,
-      "grad_norm": 0.0007134611369110644,
-      "learning_rate": 0.00019999748719343096,
-      "loss": 46.0,
-      "step": 29524
-    },
-    {
-      "epoch": 2.2573924345814937,
-      "grad_norm": 0.0020269188098609447,
-      "learning_rate": 0.00019999748702314985,
-      "loss": 46.0,
-      "step": 29525
-    },
-    {
-      "epoch": 2.2574688915648835,
-      "grad_norm": 0.001218091114424169,
-      "learning_rate": 0.00019999748685286295,
-      "loss": 46.0,
-      "step": 29526
-    },
-    {
-      "epoch": 2.2575453485482733,
-      "grad_norm": 0.0012776233488693833,
-      "learning_rate": 0.0001999974866825703,
-      "loss": 46.0,
-      "step": 29527
-    },
-    {
-      "epoch": 2.2576218055316626,
-      "grad_norm": 0.0025559107307344675,
-      "learning_rate": 0.00019999748651227184,
-      "loss": 46.0,
-      "step": 29528
-    },
-    {
-      "epoch": 2.2576982625150523,
-      "grad_norm": 0.0003450391814112663,
-      "learning_rate": 0.00019999748634196764,
-      "loss": 46.0,
-      "step": 29529
-    },
-    {
-      "epoch": 2.257774719498442,
-      "grad_norm": 0.0008357316837646067,
-      "learning_rate": 0.0001999974861716577,
-      "loss": 46.0,
-      "step": 29530
-    },
-    {
-      "epoch": 2.257851176481832,
-      "grad_norm": 0.0022005855571478605,
-      "learning_rate": 0.00019999748600134192,
-      "loss": 46.0,
-      "step": 29531
-    },
-    {
-      "epoch": 2.2579276334652216,
-      "grad_norm": 0.003092424478381872,
-      "learning_rate": 0.00019999748583102042,
-      "loss": 46.0,
-      "step": 29532
-    },
-    {
-      "epoch": 2.2580040904486114,
-      "grad_norm": 0.0006877164705656469,
-      "learning_rate": 0.00019999748566069313,
-      "loss": 46.0,
-      "step": 29533
-    },
-    {
-      "epoch": 2.258080547432001,
-      "grad_norm": 0.0008767228573560715,
-      "learning_rate": 0.0001999974854903601,
-      "loss": 46.0,
-      "step": 29534
-    },
-    {
-      "epoch": 2.258157004415391,
-      "grad_norm": 0.002584712579846382,
-      "learning_rate": 0.00019999748532002125,
-      "loss": 46.0,
-      "step": 29535
-    },
-    {
-      "epoch": 2.2582334613987807,
-      "grad_norm": 0.002083809347823262,
-      "learning_rate": 0.00019999748514967666,
-      "loss": 46.0,
-      "step": 29536
-    },
-    {
-      "epoch": 2.25830991838217,
-      "grad_norm": 0.0006496751448139548,
-      "learning_rate": 0.0001999974849793263,
-      "loss": 46.0,
-      "step": 29537
-    },
-    {
-      "epoch": 2.2583863753655598,
-      "grad_norm": 0.0010729285422712564,
-      "learning_rate": 0.00019999748480897017,
-      "loss": 46.0,
-      "step": 29538
-    },
-    {
-      "epoch": 2.2584628323489495,
-      "grad_norm": 0.0014099609106779099,
-      "learning_rate": 0.00019999748463860826,
-      "loss": 46.0,
-      "step": 29539
-    },
-    {
-      "epoch": 2.2585392893323393,
-      "grad_norm": 0.007729626260697842,
-      "learning_rate": 0.0001999974844682406,
-      "loss": 46.0,
-      "step": 29540
-    },
-    {
-      "epoch": 2.258615746315729,
-      "grad_norm": 0.002664318773895502,
-      "learning_rate": 0.00019999748429786718,
-      "loss": 46.0,
-      "step": 29541
-    },
-    {
-      "epoch": 2.258692203299119,
-      "grad_norm": 0.0011940925614908338,
-      "learning_rate": 0.00019999748412748795,
-      "loss": 46.0,
-      "step": 29542
-    },
-    {
-      "epoch": 2.2587686602825086,
-      "grad_norm": 0.0060840873047709465,
-      "learning_rate": 0.00019999748395710298,
-      "loss": 46.0,
-      "step": 29543
-    },
-    {
-      "epoch": 2.2588451172658983,
-      "grad_norm": 0.0009045965853147209,
-      "learning_rate": 0.00019999748378671223,
-      "loss": 46.0,
-      "step": 29544
-    },
-    {
-      "epoch": 2.258921574249288,
-      "grad_norm": 0.0004799450689461082,
-      "learning_rate": 0.00019999748361631574,
-      "loss": 46.0,
-      "step": 29545
-    },
-    {
-      "epoch": 2.258998031232678,
-      "grad_norm": 0.0007900987402535975,
-      "learning_rate": 0.00019999748344591342,
-      "loss": 46.0,
-      "step": 29546
-    },
-    {
-      "epoch": 2.2590744882160676,
-      "grad_norm": 0.0010051662102341652,
-      "learning_rate": 0.00019999748327550538,
-      "loss": 46.0,
-      "step": 29547
-    },
-    {
-      "epoch": 2.2591509451994574,
-      "grad_norm": 0.0010583508992567658,
-      "learning_rate": 0.00019999748310509156,
-      "loss": 46.0,
-      "step": 29548
-    },
-    {
-      "epoch": 2.259227402182847,
-      "grad_norm": 0.001418265514075756,
-      "learning_rate": 0.00019999748293467198,
-      "loss": 46.0,
-      "step": 29549
-    },
-    {
-      "epoch": 2.2593038591662364,
-      "grad_norm": 0.0005542614962905645,
-      "learning_rate": 0.00019999748276424662,
-      "loss": 46.0,
-      "step": 29550
-    },
-    {
-      "epoch": 2.259380316149626,
-      "grad_norm": 0.003746501635760069,
-      "learning_rate": 0.00019999748259381548,
-      "loss": 46.0,
-      "step": 29551
-    },
-    {
-      "epoch": 2.259456773133016,
-      "grad_norm": 0.0011258306913077831,
-      "learning_rate": 0.00019999748242337858,
-      "loss": 46.0,
-      "step": 29552
-    },
-    {
-      "epoch": 2.2595332301164057,
-      "grad_norm": 0.0018026726320385933,
-      "learning_rate": 0.0001999974822529359,
-      "loss": 46.0,
-      "step": 29553
-    },
-    {
-      "epoch": 2.2596096870997955,
-      "grad_norm": 0.0005133642116561532,
-      "learning_rate": 0.00019999748208248747,
-      "loss": 46.0,
-      "step": 29554
-    },
-    {
-      "epoch": 2.2596861440831852,
-      "grad_norm": 0.0009562474442645907,
-      "learning_rate": 0.00019999748191203327,
-      "loss": 46.0,
-      "step": 29555
-    },
-    {
-      "epoch": 2.259762601066575,
-      "grad_norm": 0.0010139537043869495,
-      "learning_rate": 0.0001999974817415733,
-      "loss": 46.0,
-      "step": 29556
-    },
-    {
-      "epoch": 2.2598390580499648,
-      "grad_norm": 0.0009557768935337663,
-      "learning_rate": 0.00019999748157110752,
-      "loss": 46.0,
-      "step": 29557
-    },
-    {
-      "epoch": 2.2599155150333545,
-      "grad_norm": 0.00045353241148404777,
-      "learning_rate": 0.000199997481400636,
-      "loss": 46.0,
-      "step": 29558
-    },
-    {
-      "epoch": 2.259991972016744,
-      "grad_norm": 0.006289840210229158,
-      "learning_rate": 0.00019999748123015874,
-      "loss": 46.0,
-      "step": 29559
-    },
-    {
-      "epoch": 2.2600684290001336,
-      "grad_norm": 0.0006640045321546495,
-      "learning_rate": 0.00019999748105967567,
-      "loss": 46.0,
-      "step": 29560
-    },
-    {
-      "epoch": 2.2601448859835234,
-      "grad_norm": 0.0006483257166109979,
-      "learning_rate": 0.00019999748088918686,
-      "loss": 46.0,
-      "step": 29561
-    },
-    {
-      "epoch": 2.260221342966913,
-      "grad_norm": 0.0006216985639184713,
-      "learning_rate": 0.00019999748071869227,
-      "loss": 46.0,
-      "step": 29562
-    },
-    {
-      "epoch": 2.260297799950303,
-      "grad_norm": 0.0009260012884624302,
-      "learning_rate": 0.00019999748054819189,
-      "loss": 46.0,
-      "step": 29563
-    },
-    {
-      "epoch": 2.2603742569336926,
-      "grad_norm": 0.019574521109461784,
-      "learning_rate": 0.00019999748037768578,
-      "loss": 46.0,
-      "step": 29564
-    },
-    {
-      "epoch": 2.2604507139170824,
-      "grad_norm": 0.000898151658475399,
-      "learning_rate": 0.00019999748020717387,
-      "loss": 46.0,
-      "step": 29565
-    },
-    {
-      "epoch": 2.260527170900472,
-      "grad_norm": 0.00130720273591578,
-      "learning_rate": 0.0001999974800366562,
-      "loss": 46.0,
-      "step": 29566
-    },
-    {
-      "epoch": 2.260603627883862,
-      "grad_norm": 0.0009814626537263393,
-      "learning_rate": 0.00019999747986613277,
-      "loss": 46.0,
-      "step": 29567
-    },
-    {
-      "epoch": 2.2606800848672517,
-      "grad_norm": 0.001772459247149527,
-      "learning_rate": 0.00019999747969560354,
-      "loss": 46.0,
-      "step": 29568
-    },
-    {
-      "epoch": 2.2607565418506415,
-      "grad_norm": 0.0010695073287934065,
-      "learning_rate": 0.00019999747952506857,
-      "loss": 46.0,
-      "step": 29569
-    },
-    {
-      "epoch": 2.260832998834031,
-      "grad_norm": 0.005123097449541092,
-      "learning_rate": 0.00019999747935452785,
-      "loss": 46.0,
-      "step": 29570
-    },
-    {
-      "epoch": 2.2609094558174205,
-      "grad_norm": 0.003967008087784052,
-      "learning_rate": 0.00019999747918398133,
-      "loss": 46.0,
-      "step": 29571
-    },
-    {
-      "epoch": 2.2609859128008103,
-      "grad_norm": 0.0028595540206879377,
-      "learning_rate": 0.00019999747901342904,
-      "loss": 46.0,
-      "step": 29572
-    },
-    {
-      "epoch": 2.2610623697842,
-      "grad_norm": 0.0009488065261393785,
-      "learning_rate": 0.00019999747884287097,
-      "loss": 46.0,
-      "step": 29573
-    },
-    {
-      "epoch": 2.26113882676759,
-      "grad_norm": 0.001119833905249834,
-      "learning_rate": 0.00019999747867230716,
-      "loss": 46.0,
-      "step": 29574
-    },
-    {
-      "epoch": 2.2612152837509796,
-      "grad_norm": 0.0013063232181593776,
-      "learning_rate": 0.00019999747850173754,
-      "loss": 46.0,
-      "step": 29575
-    },
-    {
-      "epoch": 2.2612917407343693,
-      "grad_norm": 0.0007169015007093549,
-      "learning_rate": 0.00019999747833116218,
-      "loss": 46.0,
-      "step": 29576
-    },
-    {
-      "epoch": 2.261368197717759,
-      "grad_norm": 0.0009960820898413658,
-      "learning_rate": 0.00019999747816058105,
-      "loss": 46.0,
-      "step": 29577
-    },
-    {
-      "epoch": 2.261444654701149,
-      "grad_norm": 0.0028823399916291237,
-      "learning_rate": 0.00019999747798999415,
-      "loss": 46.0,
-      "step": 29578
-    },
-    {
-      "epoch": 2.2615211116845386,
-      "grad_norm": 0.0024333784822374582,
-      "learning_rate": 0.00019999747781940147,
-      "loss": 46.0,
-      "step": 29579
-    },
-    {
-      "epoch": 2.2615975686679284,
-      "grad_norm": 0.0026083190459758043,
-      "learning_rate": 0.00019999747764880304,
-      "loss": 46.0,
-      "step": 29580
-    },
-    {
-      "epoch": 2.2616740256513177,
-      "grad_norm": 0.002803124487400055,
-      "learning_rate": 0.00019999747747819884,
-      "loss": 46.0,
-      "step": 29581
-    },
-    {
-      "epoch": 2.2617504826347075,
-      "grad_norm": 0.0011537878308445215,
-      "learning_rate": 0.00019999747730758887,
-      "loss": 46.0,
-      "step": 29582
-    },
-    {
-      "epoch": 2.261826939618097,
-      "grad_norm": 0.0011096703819930553,
-      "learning_rate": 0.00019999747713697312,
-      "loss": 46.0,
-      "step": 29583
-    },
-    {
-      "epoch": 2.261903396601487,
-      "grad_norm": 0.0015065460465848446,
-      "learning_rate": 0.0001999974769663516,
-      "loss": 46.0,
-      "step": 29584
-    },
-    {
-      "epoch": 2.2619798535848767,
-      "grad_norm": 0.0005231752875261009,
-      "learning_rate": 0.0001999974767957243,
-      "loss": 46.0,
-      "step": 29585
-    },
-    {
-      "epoch": 2.2620563105682665,
-      "grad_norm": 0.002661655889824033,
-      "learning_rate": 0.00019999747662509125,
-      "loss": 46.0,
-      "step": 29586
-    },
-    {
-      "epoch": 2.2621327675516563,
-      "grad_norm": 0.0010993279283866286,
-      "learning_rate": 0.0001999974764544524,
-      "loss": 46.0,
-      "step": 29587
-    },
-    {
-      "epoch": 2.262209224535046,
-      "grad_norm": 0.0008182556484825909,
-      "learning_rate": 0.00019999747628380782,
-      "loss": 46.0,
-      "step": 29588
-    },
-    {
-      "epoch": 2.262285681518436,
-      "grad_norm": 0.0008605673792771995,
-      "learning_rate": 0.00019999747611315746,
-      "loss": 46.0,
-      "step": 29589
-    },
-    {
-      "epoch": 2.2623621385018255,
-      "grad_norm": 0.0006910401862114668,
-      "learning_rate": 0.0001999974759425013,
-      "loss": 46.0,
-      "step": 29590
-    },
-    {
-      "epoch": 2.2624385954852153,
-      "grad_norm": 0.0030075605027377605,
-      "learning_rate": 0.00019999747577183942,
-      "loss": 46.0,
-      "step": 29591
-    },
-    {
-      "epoch": 2.262515052468605,
-      "grad_norm": 0.000518039392773062,
-      "learning_rate": 0.00019999747560117174,
-      "loss": 46.0,
-      "step": 29592
-    },
-    {
-      "epoch": 2.2625915094519944,
-      "grad_norm": 0.0009006580803543329,
-      "learning_rate": 0.0001999974754304983,
-      "loss": 46.0,
-      "step": 29593
-    },
-    {
-      "epoch": 2.262667966435384,
-      "grad_norm": 0.0022540846839547157,
-      "learning_rate": 0.0001999974752598191,
-      "loss": 46.0,
-      "step": 29594
-    },
-    {
-      "epoch": 2.262744423418774,
-      "grad_norm": 0.0008996726246550679,
-      "learning_rate": 0.00019999747508913412,
-      "loss": 46.0,
-      "step": 29595
-    },
-    {
-      "epoch": 2.2628208804021637,
-      "grad_norm": 0.0013292315416038036,
-      "learning_rate": 0.00019999747491844337,
-      "loss": 46.0,
-      "step": 29596
-    },
-    {
-      "epoch": 2.2628973373855534,
-      "grad_norm": 0.0029233135282993317,
-      "learning_rate": 0.00019999747474774683,
-      "loss": 46.0,
-      "step": 29597
-    },
-    {
-      "epoch": 2.262973794368943,
-      "grad_norm": 0.00048096824320964515,
-      "learning_rate": 0.00019999747457704456,
-      "loss": 46.0,
-      "step": 29598
-    },
-    {
-      "epoch": 2.263050251352333,
-      "grad_norm": 0.0011139829875901341,
-      "learning_rate": 0.00019999747440633652,
-      "loss": 46.0,
-      "step": 29599
-    },
-    {
-      "epoch": 2.2631267083357227,
-      "grad_norm": 0.0020469920709729195,
-      "learning_rate": 0.00019999747423562268,
-      "loss": 46.0,
-      "step": 29600
-    },
-    {
-      "epoch": 2.2632031653191125,
-      "grad_norm": 0.0010408778907731175,
-      "learning_rate": 0.00019999747406490307,
-      "loss": 46.0,
-      "step": 29601
-    },
-    {
-      "epoch": 2.2632796223025022,
-      "grad_norm": 0.0006072634132578969,
-      "learning_rate": 0.00019999747389417771,
-      "loss": 46.0,
-      "step": 29602
-    },
-    {
-      "epoch": 2.2633560792858916,
-      "grad_norm": 0.0003907477075699717,
-      "learning_rate": 0.00019999747372344658,
-      "loss": 46.0,
-      "step": 29603
-    },
-    {
-      "epoch": 2.2634325362692813,
-      "grad_norm": 0.001082484144717455,
-      "learning_rate": 0.00019999747355270968,
-      "loss": 46.0,
-      "step": 29604
-    },
-    {
-      "epoch": 2.263508993252671,
-      "grad_norm": 0.0016665211878716946,
-      "learning_rate": 0.000199997473381967,
-      "loss": 46.0,
-      "step": 29605
-    },
-    {
-      "epoch": 2.263585450236061,
-      "grad_norm": 0.0007992112077772617,
-      "learning_rate": 0.00019999747321121854,
-      "loss": 46.0,
-      "step": 29606
-    },
-    {
-      "epoch": 2.2636619072194506,
-      "grad_norm": 0.0005896664806641638,
-      "learning_rate": 0.00019999747304046435,
-      "loss": 46.0,
-      "step": 29607
-    },
-    {
-      "epoch": 2.2637383642028404,
-      "grad_norm": 0.0008752181311137974,
-      "learning_rate": 0.00019999747286970437,
-      "loss": 46.0,
-      "step": 29608
-    },
-    {
-      "epoch": 2.26381482118623,
-      "grad_norm": 0.0006907433271408081,
-      "learning_rate": 0.0001999974726989386,
-      "loss": 46.0,
-      "step": 29609
-    },
-    {
-      "epoch": 2.26389127816962,
-      "grad_norm": 0.011877243407070637,
-      "learning_rate": 0.00019999747252816708,
-      "loss": 46.0,
-      "step": 29610
-    },
-    {
-      "epoch": 2.2639677351530096,
-      "grad_norm": 0.0005057640955783427,
-      "learning_rate": 0.00019999747235738982,
-      "loss": 46.0,
-      "step": 29611
-    },
-    {
-      "epoch": 2.2640441921363994,
-      "grad_norm": 0.0006477448623627424,
-      "learning_rate": 0.00019999747218660673,
-      "loss": 46.0,
-      "step": 29612
-    },
-    {
-      "epoch": 2.264120649119789,
-      "grad_norm": 0.0009521179017610848,
-      "learning_rate": 0.00019999747201581794,
-      "loss": 46.0,
-      "step": 29613
-    },
-    {
-      "epoch": 2.264197106103179,
-      "grad_norm": 0.012359128333628178,
-      "learning_rate": 0.00019999747184502333,
-      "loss": 46.0,
-      "step": 29614
-    },
-    {
-      "epoch": 2.2642735630865682,
-      "grad_norm": 0.005482463166117668,
-      "learning_rate": 0.00019999747167422297,
-      "loss": 46.0,
-      "step": 29615
-    },
-    {
-      "epoch": 2.264350020069958,
-      "grad_norm": 0.0012245887191966176,
-      "learning_rate": 0.00019999747150341684,
-      "loss": 46.0,
-      "step": 29616
-    },
-    {
-      "epoch": 2.2644264770533478,
-      "grad_norm": 0.0010517871705815196,
-      "learning_rate": 0.00019999747133260493,
-      "loss": 46.0,
-      "step": 29617
-    },
-    {
-      "epoch": 2.2645029340367375,
-      "grad_norm": 0.0007861704798415303,
-      "learning_rate": 0.00019999747116178726,
-      "loss": 46.0,
-      "step": 29618
-    },
-    {
-      "epoch": 2.2645793910201273,
-      "grad_norm": 0.0008534854860045016,
-      "learning_rate": 0.0001999974709909638,
-      "loss": 46.0,
-      "step": 29619
-    },
-    {
-      "epoch": 2.264655848003517,
-      "grad_norm": 0.0005096105160191655,
-      "learning_rate": 0.0001999974708201346,
-      "loss": 46.0,
-      "step": 29620
-    },
-    {
-      "epoch": 2.264732304986907,
-      "grad_norm": 0.0011729886755347252,
-      "learning_rate": 0.0001999974706492996,
-      "loss": 46.0,
-      "step": 29621
-    },
-    {
-      "epoch": 2.2648087619702966,
-      "grad_norm": 0.0016736198449507356,
-      "learning_rate": 0.00019999747047845886,
-      "loss": 46.0,
-      "step": 29622
-    },
-    {
-      "epoch": 2.2648852189536863,
-      "grad_norm": 0.0012622796930372715,
-      "learning_rate": 0.00019999747030761234,
-      "loss": 46.0,
-      "step": 29623
-    },
-    {
-      "epoch": 2.264961675937076,
-      "grad_norm": 0.0012547977967187762,
-      "learning_rate": 0.00019999747013676005,
-      "loss": 46.0,
-      "step": 29624
-    },
-    {
-      "epoch": 2.2650381329204654,
-      "grad_norm": 0.0004887098912149668,
-      "learning_rate": 0.00019999746996590199,
-      "loss": 46.0,
-      "step": 29625
-    },
-    {
-      "epoch": 2.265114589903855,
-      "grad_norm": 0.0005050490726716816,
-      "learning_rate": 0.00019999746979503815,
-      "loss": 46.0,
-      "step": 29626
-    },
-    {
-      "epoch": 2.265191046887245,
-      "grad_norm": 0.0016263237921521068,
-      "learning_rate": 0.00019999746962416856,
-      "loss": 46.0,
-      "step": 29627
-    },
-    {
-      "epoch": 2.2652675038706347,
-      "grad_norm": 0.002395712537690997,
-      "learning_rate": 0.0001999974694532932,
-      "loss": 46.0,
-      "step": 29628
-    },
-    {
-      "epoch": 2.2653439608540245,
-      "grad_norm": 0.0006180467898957431,
-      "learning_rate": 0.00019999746928241205,
-      "loss": 46.0,
-      "step": 29629
-    },
-    {
-      "epoch": 2.265420417837414,
-      "grad_norm": 0.0007490097777917981,
-      "learning_rate": 0.00019999746911152514,
-      "loss": 46.0,
-      "step": 29630
-    },
-    {
-      "epoch": 2.265496874820804,
-      "grad_norm": 0.002582067623734474,
-      "learning_rate": 0.00019999746894063246,
-      "loss": 46.0,
-      "step": 29631
-    },
-    {
-      "epoch": 2.2655733318041937,
-      "grad_norm": 0.0018171604024246335,
-      "learning_rate": 0.00019999746876973404,
-      "loss": 46.0,
-      "step": 29632
-    },
-    {
-      "epoch": 2.2656497887875835,
-      "grad_norm": 0.0010933029698207974,
-      "learning_rate": 0.0001999974685988298,
-      "loss": 46.0,
-      "step": 29633
-    },
-    {
-      "epoch": 2.2657262457709733,
-      "grad_norm": 0.0016335133695974946,
-      "learning_rate": 0.00019999746842791981,
-      "loss": 46.0,
-      "step": 29634
-    },
-    {
-      "epoch": 2.265802702754363,
-      "grad_norm": 0.0020243460312485695,
-      "learning_rate": 0.0001999974682570041,
-      "loss": 46.0,
-      "step": 29635
-    },
-    {
-      "epoch": 2.265879159737753,
-      "grad_norm": 0.0008935952791944146,
-      "learning_rate": 0.00019999746808608255,
-      "loss": 46.0,
-      "step": 29636
-    },
-    {
-      "epoch": 2.265955616721142,
-      "grad_norm": 0.0014319374458864331,
-      "learning_rate": 0.0001999974679151553,
-      "loss": 46.0,
-      "step": 29637
-    },
-    {
-      "epoch": 2.266032073704532,
-      "grad_norm": 0.000753024301957339,
-      "learning_rate": 0.0001999974677442222,
-      "loss": 46.0,
-      "step": 29638
-    },
-    {
-      "epoch": 2.2661085306879216,
-      "grad_norm": 0.0008614014368504286,
-      "learning_rate": 0.00019999746757328336,
-      "loss": 46.0,
-      "step": 29639
-    },
-    {
-      "epoch": 2.2661849876713114,
-      "grad_norm": 0.0008059954270720482,
-      "learning_rate": 0.00019999746740233877,
-      "loss": 46.0,
-      "step": 29640
-    },
-    {
-      "epoch": 2.266261444654701,
-      "grad_norm": 0.0008873986080288887,
-      "learning_rate": 0.00019999746723138842,
-      "loss": 46.0,
-      "step": 29641
-    },
-    {
-      "epoch": 2.266337901638091,
-      "grad_norm": 0.0017346162348985672,
-      "learning_rate": 0.00019999746706043228,
-      "loss": 46.0,
-      "step": 29642
-    },
-    {
-      "epoch": 2.2664143586214807,
-      "grad_norm": 0.0013345525367185473,
-      "learning_rate": 0.00019999746688947035,
-      "loss": 46.0,
-      "step": 29643
-    },
-    {
-      "epoch": 2.2664908156048704,
-      "grad_norm": 0.0009429907659068704,
-      "learning_rate": 0.00019999746671850267,
-      "loss": 46.0,
-      "step": 29644
-    },
-    {
-      "epoch": 2.26656727258826,
-      "grad_norm": 0.0003901784948538989,
-      "learning_rate": 0.00019999746654752925,
-      "loss": 46.0,
-      "step": 29645
-    },
-    {
-      "epoch": 2.2666437295716495,
-      "grad_norm": 0.0008872750331647694,
-      "learning_rate": 0.00019999746637655003,
-      "loss": 46.0,
-      "step": 29646
-    },
-    {
-      "epoch": 2.2667201865550393,
-      "grad_norm": 0.010518618859350681,
-      "learning_rate": 0.00019999746620556503,
-      "loss": 46.0,
-      "step": 29647
-    },
-    {
-      "epoch": 2.266796643538429,
-      "grad_norm": 0.0004517077759373933,
-      "learning_rate": 0.0001999974660345743,
-      "loss": 46.0,
-      "step": 29648
-    },
-    {
-      "epoch": 2.266873100521819,
-      "grad_norm": 0.0004741284647025168,
-      "learning_rate": 0.00019999746586357777,
-      "loss": 46.0,
-      "step": 29649
-    },
-    {
-      "epoch": 2.2669495575052085,
-      "grad_norm": 0.0011188186472281814,
-      "learning_rate": 0.00019999746569257547,
-      "loss": 46.0,
-      "step": 29650
-    },
-    {
-      "epoch": 2.2670260144885983,
-      "grad_norm": 0.0005691573023796082,
-      "learning_rate": 0.0001999974655215674,
-      "loss": 46.0,
-      "step": 29651
-    },
-    {
-      "epoch": 2.267102471471988,
-      "grad_norm": 0.0015496093546971679,
-      "learning_rate": 0.0001999974653505536,
-      "loss": 46.0,
-      "step": 29652
-    },
-    {
-      "epoch": 2.267178928455378,
-      "grad_norm": 0.0010420939652249217,
-      "learning_rate": 0.000199997465179534,
-      "loss": 46.0,
-      "step": 29653
-    },
-    {
-      "epoch": 2.2672553854387676,
-      "grad_norm": 0.0010665485169738531,
-      "learning_rate": 0.00019999746500850863,
-      "loss": 46.0,
-      "step": 29654
-    },
-    {
-      "epoch": 2.2673318424221574,
-      "grad_norm": 0.0026481342501938343,
-      "learning_rate": 0.00019999746483747747,
-      "loss": 46.0,
-      "step": 29655
-    },
-    {
-      "epoch": 2.267408299405547,
-      "grad_norm": 0.0008155222167260945,
-      "learning_rate": 0.00019999746466644057,
-      "loss": 46.0,
-      "step": 29656
-    },
-    {
-      "epoch": 2.267484756388937,
-      "grad_norm": 0.0023945928551256657,
-      "learning_rate": 0.0001999974644953979,
-      "loss": 46.0,
-      "step": 29657
-    },
-    {
-      "epoch": 2.2675612133723266,
-      "grad_norm": 0.0009923020843416452,
-      "learning_rate": 0.00019999746432434944,
-      "loss": 46.0,
-      "step": 29658
-    },
-    {
-      "epoch": 2.267637670355716,
-      "grad_norm": 0.0012029190547764301,
-      "learning_rate": 0.00019999746415329522,
-      "loss": 46.0,
-      "step": 29659
-    },
-    {
-      "epoch": 2.2677141273391057,
-      "grad_norm": 0.0007509376155212522,
-      "learning_rate": 0.00019999746398223524,
-      "loss": 46.0,
-      "step": 29660
-    },
-    {
-      "epoch": 2.2677905843224955,
-      "grad_norm": 0.0009993338026106358,
-      "learning_rate": 0.00019999746381116947,
-      "loss": 46.0,
-      "step": 29661
-    },
-    {
-      "epoch": 2.2678670413058852,
-      "grad_norm": 0.0031778153497725725,
-      "learning_rate": 0.00019999746364009796,
-      "loss": 46.0,
-      "step": 29662
-    },
-    {
-      "epoch": 2.267943498289275,
-      "grad_norm": 0.0007198613602668047,
-      "learning_rate": 0.00019999746346902067,
-      "loss": 46.0,
-      "step": 29663
-    },
-    {
-      "epoch": 2.2680199552726648,
-      "grad_norm": 0.0012843615841120481,
-      "learning_rate": 0.0001999974632979376,
-      "loss": 46.0,
-      "step": 29664
-    },
-    {
-      "epoch": 2.2680964122560545,
-      "grad_norm": 0.0006143125938251615,
-      "learning_rate": 0.00019999746312684876,
-      "loss": 46.0,
-      "step": 29665
-    },
-    {
-      "epoch": 2.2681728692394443,
-      "grad_norm": 0.003900025272741914,
-      "learning_rate": 0.00019999746295575418,
-      "loss": 46.0,
-      "step": 29666
-    },
-    {
-      "epoch": 2.268249326222834,
-      "grad_norm": 0.0013759088469669223,
-      "learning_rate": 0.0001999974627846538,
-      "loss": 46.0,
-      "step": 29667
-    },
-    {
-      "epoch": 2.2683257832062234,
-      "grad_norm": 0.0015312323812395334,
-      "learning_rate": 0.00019999746261354766,
-      "loss": 46.0,
-      "step": 29668
-    },
-    {
-      "epoch": 2.268402240189613,
-      "grad_norm": 0.0011515525402501225,
-      "learning_rate": 0.00019999746244243576,
-      "loss": 46.0,
-      "step": 29669
-    },
-    {
-      "epoch": 2.268478697173003,
-      "grad_norm": 0.0018562345067039132,
-      "learning_rate": 0.00019999746227131808,
-      "loss": 46.0,
-      "step": 29670
-    },
-    {
-      "epoch": 2.2685551541563926,
-      "grad_norm": 0.0018017544643953443,
-      "learning_rate": 0.00019999746210019463,
-      "loss": 46.0,
-      "step": 29671
-    },
-    {
-      "epoch": 2.2686316111397824,
-      "grad_norm": 0.005346236284822226,
-      "learning_rate": 0.0001999974619290654,
-      "loss": 46.0,
-      "step": 29672
-    },
-    {
-      "epoch": 2.268708068123172,
-      "grad_norm": 0.0018701773369684815,
-      "learning_rate": 0.0001999974617579304,
-      "loss": 46.0,
-      "step": 29673
-    },
-    {
-      "epoch": 2.268784525106562,
-      "grad_norm": 0.0005911273183301091,
-      "learning_rate": 0.00019999746158678967,
-      "loss": 46.0,
-      "step": 29674
-    },
-    {
-      "epoch": 2.2688609820899517,
-      "grad_norm": 0.001537785166874528,
-      "learning_rate": 0.00019999746141564315,
-      "loss": 46.0,
-      "step": 29675
-    },
-    {
-      "epoch": 2.2689374390733414,
-      "grad_norm": 0.0005459311651065946,
-      "learning_rate": 0.00019999746124449086,
-      "loss": 46.0,
-      "step": 29676
-    },
-    {
-      "epoch": 2.269013896056731,
-      "grad_norm": 0.0008583724847994745,
-      "learning_rate": 0.0001999974610733328,
-      "loss": 46.0,
-      "step": 29677
-    },
-    {
-      "epoch": 2.269090353040121,
-      "grad_norm": 0.0016862861812114716,
-      "learning_rate": 0.00019999746090216896,
-      "loss": 46.0,
-      "step": 29678
-    },
-    {
-      "epoch": 2.2691668100235107,
-      "grad_norm": 0.0015586403897032142,
-      "learning_rate": 0.00019999746073099935,
-      "loss": 46.0,
-      "step": 29679
-    },
-    {
-      "epoch": 2.2692432670069005,
-      "grad_norm": 0.009733383543789387,
-      "learning_rate": 0.000199997460559824,
-      "loss": 46.0,
-      "step": 29680
-    },
-    {
-      "epoch": 2.26931972399029,
-      "grad_norm": 0.0011432974133640528,
-      "learning_rate": 0.00019999746038864283,
-      "loss": 46.0,
-      "step": 29681
-    },
-    {
-      "epoch": 2.2693961809736796,
-      "grad_norm": 0.0027637737803161144,
-      "learning_rate": 0.00019999746021745593,
-      "loss": 46.0,
-      "step": 29682
-    },
-    {
-      "epoch": 2.2694726379570693,
-      "grad_norm": 0.0008523364667780697,
-      "learning_rate": 0.00019999746004626325,
-      "loss": 46.0,
-      "step": 29683
-    },
-    {
-      "epoch": 2.269549094940459,
-      "grad_norm": 0.005892177112400532,
-      "learning_rate": 0.00019999745987506483,
-      "loss": 46.0,
-      "step": 29684
-    },
-    {
-      "epoch": 2.269625551923849,
-      "grad_norm": 0.00045390776358544827,
-      "learning_rate": 0.00019999745970386058,
-      "loss": 46.0,
-      "step": 29685
-    },
-    {
-      "epoch": 2.2697020089072386,
-      "grad_norm": 0.0009760102839209139,
-      "learning_rate": 0.00019999745953265058,
-      "loss": 46.0,
-      "step": 29686
-    },
-    {
-      "epoch": 2.2697784658906284,
-      "grad_norm": 0.0030714794993400574,
-      "learning_rate": 0.00019999745936143484,
-      "loss": 46.0,
-      "step": 29687
-    },
-    {
-      "epoch": 2.269854922874018,
-      "grad_norm": 0.0006153856520541012,
-      "learning_rate": 0.00019999745919021332,
-      "loss": 46.0,
-      "step": 29688
-    },
-    {
-      "epoch": 2.269931379857408,
-      "grad_norm": 0.0010291390353813767,
-      "learning_rate": 0.000199997459018986,
-      "loss": 46.0,
-      "step": 29689
-    },
-    {
-      "epoch": 2.270007836840797,
-      "grad_norm": 0.0010182630503550172,
-      "learning_rate": 0.00019999745884775297,
-      "loss": 46.0,
-      "step": 29690
-    },
-    {
-      "epoch": 2.270084293824187,
-      "grad_norm": 0.001276559429243207,
-      "learning_rate": 0.00019999745867651413,
-      "loss": 46.0,
-      "step": 29691
-    },
-    {
-      "epoch": 2.2701607508075767,
-      "grad_norm": 0.0018659690394997597,
-      "learning_rate": 0.00019999745850526952,
-      "loss": 46.0,
-      "step": 29692
-    },
-    {
-      "epoch": 2.2702372077909665,
-      "grad_norm": 0.002090269234031439,
-      "learning_rate": 0.00019999745833401914,
-      "loss": 46.0,
-      "step": 29693
-    },
-    {
-      "epoch": 2.2703136647743563,
-      "grad_norm": 0.0008636604179628193,
-      "learning_rate": 0.000199997458162763,
-      "loss": 46.0,
-      "step": 29694
-    },
-    {
-      "epoch": 2.270390121757746,
-      "grad_norm": 0.000776358472649008,
-      "learning_rate": 0.0001999974579915011,
-      "loss": 46.0,
-      "step": 29695
-    },
-    {
-      "epoch": 2.2704665787411358,
-      "grad_norm": 0.0010542470263317227,
-      "learning_rate": 0.0001999974578202334,
-      "loss": 46.0,
-      "step": 29696
-    },
-    {
-      "epoch": 2.2705430357245255,
-      "grad_norm": 0.0009724535630084574,
-      "learning_rate": 0.00019999745764895995,
-      "loss": 46.0,
-      "step": 29697
-    },
-    {
-      "epoch": 2.2706194927079153,
-      "grad_norm": 0.0012763264821842313,
-      "learning_rate": 0.00019999745747768075,
-      "loss": 46.0,
-      "step": 29698
-    },
-    {
-      "epoch": 2.270695949691305,
-      "grad_norm": 0.003610215848311782,
-      "learning_rate": 0.00019999745730639576,
-      "loss": 46.0,
-      "step": 29699
-    },
-    {
-      "epoch": 2.270772406674695,
-      "grad_norm": 0.0007914112065918744,
-      "learning_rate": 0.00019999745713510501,
-      "loss": 46.0,
-      "step": 29700
-    },
-    {
-      "epoch": 2.2708488636580846,
-      "grad_norm": 0.00998303759843111,
-      "learning_rate": 0.00019999745696380847,
-      "loss": 46.0,
-      "step": 29701
-    },
-    {
-      "epoch": 2.270925320641474,
-      "grad_norm": 0.0024524356704205275,
-      "learning_rate": 0.00019999745679250615,
-      "loss": 46.0,
-      "step": 29702
-    },
-    {
-      "epoch": 2.2710017776248637,
-      "grad_norm": 0.0009396827081218362,
-      "learning_rate": 0.00019999745662119812,
-      "loss": 46.0,
-      "step": 29703
-    },
-    {
-      "epoch": 2.2710782346082534,
-      "grad_norm": 0.0023927337024360895,
-      "learning_rate": 0.00019999745644988428,
-      "loss": 46.0,
-      "step": 29704
-    },
-    {
-      "epoch": 2.271154691591643,
-      "grad_norm": 0.0009745231363922358,
-      "learning_rate": 0.00019999745627856467,
-      "loss": 46.0,
-      "step": 29705
-    },
-    {
-      "epoch": 2.271231148575033,
-      "grad_norm": 0.0026453144382685423,
-      "learning_rate": 0.0001999974561072393,
-      "loss": 46.0,
-      "step": 29706
-    },
-    {
-      "epoch": 2.2713076055584227,
-      "grad_norm": 0.002038449514657259,
-      "learning_rate": 0.00019999745593590816,
-      "loss": 46.0,
-      "step": 29707
-    },
-    {
-      "epoch": 2.2713840625418125,
-      "grad_norm": 0.0005880210665054619,
-      "learning_rate": 0.00019999745576457123,
-      "loss": 46.0,
-      "step": 29708
-    },
-    {
-      "epoch": 2.2714605195252022,
-      "grad_norm": 0.001248908112756908,
-      "learning_rate": 0.00019999745559322855,
-      "loss": 46.0,
-      "step": 29709
-    },
-    {
-      "epoch": 2.271536976508592,
-      "grad_norm": 0.0015942917671054602,
-      "learning_rate": 0.0001999974554218801,
-      "loss": 46.0,
-      "step": 29710
-    },
-    {
-      "epoch": 2.2716134334919817,
-      "grad_norm": 0.0035309777595102787,
-      "learning_rate": 0.00019999745525052588,
-      "loss": 46.0,
-      "step": 29711
-    },
-    {
-      "epoch": 2.271689890475371,
-      "grad_norm": 0.00039428056334145367,
-      "learning_rate": 0.0001999974550791659,
-      "loss": 46.0,
-      "step": 29712
-    },
-    {
-      "epoch": 2.271766347458761,
-      "grad_norm": 0.0013230836484581232,
-      "learning_rate": 0.00019999745490780014,
-      "loss": 46.0,
-      "step": 29713
-    },
-    {
-      "epoch": 2.2718428044421506,
-      "grad_norm": 0.0016719972481951118,
-      "learning_rate": 0.00019999745473642862,
-      "loss": 46.0,
-      "step": 29714
-    },
-    {
-      "epoch": 2.2719192614255403,
-      "grad_norm": 0.0009867926128208637,
-      "learning_rate": 0.00019999745456505133,
-      "loss": 46.0,
-      "step": 29715
-    },
-    {
-      "epoch": 2.27199571840893,
-      "grad_norm": 0.0016819201409816742,
-      "learning_rate": 0.00019999745439366824,
-      "loss": 46.0,
-      "step": 29716
-    },
-    {
-      "epoch": 2.27207217539232,
-      "grad_norm": 0.003248332068324089,
-      "learning_rate": 0.00019999745422227943,
-      "loss": 46.0,
-      "step": 29717
-    },
-    {
-      "epoch": 2.2721486323757096,
-      "grad_norm": 0.002380089135840535,
-      "learning_rate": 0.00019999745405088482,
-      "loss": 46.0,
-      "step": 29718
-    },
-    {
-      "epoch": 2.2722250893590994,
-      "grad_norm": 0.0023042226675897837,
-      "learning_rate": 0.00019999745387948444,
-      "loss": 46.0,
-      "step": 29719
-    },
-    {
-      "epoch": 2.272301546342489,
-      "grad_norm": 0.0011556106619536877,
-      "learning_rate": 0.00019999745370807829,
-      "loss": 46.0,
-      "step": 29720
-    },
-    {
-      "epoch": 2.272378003325879,
-      "grad_norm": 0.0018479221034795046,
-      "learning_rate": 0.00019999745353666638,
-      "loss": 46.0,
-      "step": 29721
-    },
-    {
-      "epoch": 2.2724544603092687,
-      "grad_norm": 0.0017148521728813648,
-      "learning_rate": 0.0001999974533652487,
-      "loss": 46.0,
-      "step": 29722
-    },
-    {
-      "epoch": 2.2725309172926584,
-      "grad_norm": 0.0037179149221628904,
-      "learning_rate": 0.00019999745319382526,
-      "loss": 46.0,
-      "step": 29723
-    },
-    {
-      "epoch": 2.2726073742760478,
-      "grad_norm": 0.0006437920383177698,
-      "learning_rate": 0.00019999745302239603,
-      "loss": 46.0,
-      "step": 29724
-    },
-    {
-      "epoch": 2.2726838312594375,
-      "grad_norm": 0.0008031605393625796,
-      "learning_rate": 0.00019999745285096104,
-      "loss": 46.0,
-      "step": 29725
-    },
-    {
-      "epoch": 2.2727602882428273,
-      "grad_norm": 0.001416450017131865,
-      "learning_rate": 0.00019999745267952027,
-      "loss": 46.0,
-      "step": 29726
-    },
-    {
-      "epoch": 2.272836745226217,
-      "grad_norm": 0.0010032163700088859,
-      "learning_rate": 0.00019999745250807375,
-      "loss": 46.0,
-      "step": 29727
-    },
-    {
-      "epoch": 2.272913202209607,
-      "grad_norm": 0.000730924482923001,
-      "learning_rate": 0.00019999745233662144,
-      "loss": 46.0,
-      "step": 29728
-    },
-    {
-      "epoch": 2.2729896591929966,
-      "grad_norm": 0.003745750989764929,
-      "learning_rate": 0.0001999974521651634,
-      "loss": 46.0,
-      "step": 29729
-    },
-    {
-      "epoch": 2.2730661161763863,
-      "grad_norm": 0.0022143807727843523,
-      "learning_rate": 0.00019999745199369954,
-      "loss": 46.0,
-      "step": 29730
-    },
-    {
-      "epoch": 2.273142573159776,
-      "grad_norm": 0.0009323981939814985,
-      "learning_rate": 0.00019999745182222993,
-      "loss": 46.0,
-      "step": 29731
-    },
-    {
-      "epoch": 2.273219030143166,
-      "grad_norm": 0.0009212127770297229,
-      "learning_rate": 0.00019999745165075457,
-      "loss": 46.0,
-      "step": 29732
-    },
-    {
-      "epoch": 2.2732954871265556,
-      "grad_norm": 0.000590279814787209,
-      "learning_rate": 0.00019999745147927342,
-      "loss": 46.0,
-      "step": 29733
-    },
-    {
-      "epoch": 2.273371944109945,
-      "grad_norm": 0.0015829089097678661,
-      "learning_rate": 0.00019999745130778651,
-      "loss": 46.0,
-      "step": 29734
-    },
-    {
-      "epoch": 2.2734484010933347,
-      "grad_norm": 0.0006811293424107134,
-      "learning_rate": 0.0001999974511362938,
-      "loss": 46.0,
-      "step": 29735
-    },
-    {
-      "epoch": 2.2735248580767244,
-      "grad_norm": 0.0013754547107964754,
-      "learning_rate": 0.00019999745096479536,
-      "loss": 46.0,
-      "step": 29736
-    },
-    {
-      "epoch": 2.273601315060114,
-      "grad_norm": 0.0018407799070701003,
-      "learning_rate": 0.00019999745079329114,
-      "loss": 46.0,
-      "step": 29737
-    },
-    {
-      "epoch": 2.273677772043504,
-      "grad_norm": 0.001791534828953445,
-      "learning_rate": 0.00019999745062178117,
-      "loss": 46.0,
-      "step": 29738
-    },
-    {
-      "epoch": 2.2737542290268937,
-      "grad_norm": 0.0009379560360684991,
-      "learning_rate": 0.0001999974504502654,
-      "loss": 46.0,
-      "step": 29739
-    },
-    {
-      "epoch": 2.2738306860102835,
-      "grad_norm": 0.0031715580262243748,
-      "learning_rate": 0.0001999974502787439,
-      "loss": 46.0,
-      "step": 29740
-    },
-    {
-      "epoch": 2.2739071429936732,
-      "grad_norm": 0.001070274505764246,
-      "learning_rate": 0.00019999745010721654,
-      "loss": 46.0,
-      "step": 29741
-    },
-    {
-      "epoch": 2.273983599977063,
-      "grad_norm": 0.0007721201982349157,
-      "learning_rate": 0.0001999974499356835,
-      "loss": 46.0,
-      "step": 29742
-    },
-    {
-      "epoch": 2.2740600569604528,
-      "grad_norm": 0.0006021598819643259,
-      "learning_rate": 0.00019999744976414467,
-      "loss": 46.0,
-      "step": 29743
-    },
-    {
-      "epoch": 2.2741365139438425,
-      "grad_norm": 0.0005464857094921172,
-      "learning_rate": 0.00019999744959260006,
-      "loss": 46.0,
-      "step": 29744
-    },
-    {
-      "epoch": 2.2742129709272323,
-      "grad_norm": 0.0006201923824846745,
-      "learning_rate": 0.00019999744942104968,
-      "loss": 46.0,
-      "step": 29745
-    },
-    {
-      "epoch": 2.2742894279106216,
-      "grad_norm": 0.000782308226916939,
-      "learning_rate": 0.00019999744924949353,
-      "loss": 46.0,
-      "step": 29746
-    },
-    {
-      "epoch": 2.2743658848940114,
-      "grad_norm": 0.0025512855499982834,
-      "learning_rate": 0.00019999744907793162,
-      "loss": 46.0,
-      "step": 29747
-    },
-    {
-      "epoch": 2.274442341877401,
-      "grad_norm": 0.0008598458371125162,
-      "learning_rate": 0.00019999744890636395,
-      "loss": 46.0,
-      "step": 29748
-    },
-    {
-      "epoch": 2.274518798860791,
-      "grad_norm": 0.0007506250403821468,
-      "learning_rate": 0.00019999744873479047,
-      "loss": 46.0,
-      "step": 29749
-    },
-    {
-      "epoch": 2.2745952558441807,
-      "grad_norm": 0.0023013518657535315,
-      "learning_rate": 0.00019999744856321125,
-      "loss": 46.0,
-      "step": 29750
-    },
-    {
-      "epoch": 2.2746717128275704,
-      "grad_norm": 0.0005325338570401073,
-      "learning_rate": 0.00019999744839162625,
-      "loss": 46.0,
-      "step": 29751
-    },
-    {
-      "epoch": 2.27474816981096,
-      "grad_norm": 0.0015212559374049306,
-      "learning_rate": 0.0001999974482200355,
-      "loss": 46.0,
-      "step": 29752
-    },
-    {
-      "epoch": 2.27482462679435,
-      "grad_norm": 0.000996324117295444,
-      "learning_rate": 0.00019999744804843897,
-      "loss": 46.0,
-      "step": 29753
-    },
-    {
-      "epoch": 2.2749010837777397,
-      "grad_norm": 0.0012367131421342492,
-      "learning_rate": 0.00019999744787683668,
-      "loss": 46.0,
-      "step": 29754
-    },
-    {
-      "epoch": 2.2749775407611295,
-      "grad_norm": 0.0016539925709366798,
-      "learning_rate": 0.00019999744770522862,
-      "loss": 46.0,
-      "step": 29755
-    },
-    {
-      "epoch": 2.2750539977445188,
-      "grad_norm": 0.0005481173284351826,
-      "learning_rate": 0.00019999744753361478,
-      "loss": 46.0,
-      "step": 29756
-    },
-    {
-      "epoch": 2.2751304547279085,
-      "grad_norm": 0.0004594842903316021,
-      "learning_rate": 0.00019999744736199518,
-      "loss": 46.0,
-      "step": 29757
-    },
-    {
-      "epoch": 2.2752069117112983,
-      "grad_norm": 0.0009141605696640909,
-      "learning_rate": 0.0001999974471903698,
-      "loss": 46.0,
-      "step": 29758
-    },
-    {
-      "epoch": 2.275283368694688,
-      "grad_norm": 0.0008808863931335509,
-      "learning_rate": 0.00019999744701873864,
-      "loss": 46.0,
-      "step": 29759
-    },
-    {
-      "epoch": 2.275359825678078,
-      "grad_norm": 0.003569790394976735,
-      "learning_rate": 0.0001999974468471017,
-      "loss": 46.0,
-      "step": 29760
-    },
-    {
-      "epoch": 2.2754362826614676,
-      "grad_norm": 0.0015856667887419462,
-      "learning_rate": 0.00019999744667545903,
-      "loss": 46.0,
-      "step": 29761
-    },
-    {
-      "epoch": 2.2755127396448573,
-      "grad_norm": 0.0010833556298166513,
-      "learning_rate": 0.00019999744650381059,
-      "loss": 46.0,
-      "step": 29762
-    },
-    {
-      "epoch": 2.275589196628247,
-      "grad_norm": 0.0008403917890973389,
-      "learning_rate": 0.00019999744633215636,
-      "loss": 46.0,
-      "step": 29763
-    },
-    {
-      "epoch": 2.275665653611637,
-      "grad_norm": 0.0012686635600402951,
-      "learning_rate": 0.00019999744616049637,
-      "loss": 46.0,
-      "step": 29764
-    },
-    {
-      "epoch": 2.2757421105950266,
-      "grad_norm": 0.0013573673786595464,
-      "learning_rate": 0.0001999974459888306,
-      "loss": 46.0,
-      "step": 29765
-    },
-    {
-      "epoch": 2.2758185675784164,
-      "grad_norm": 0.006290150340646505,
-      "learning_rate": 0.00019999744581715906,
-      "loss": 46.0,
-      "step": 29766
-    },
-    {
-      "epoch": 2.275895024561806,
-      "grad_norm": 0.0023067155852913857,
-      "learning_rate": 0.00019999744564548174,
-      "loss": 46.0,
-      "step": 29767
-    },
-    {
-      "epoch": 2.2759714815451955,
-      "grad_norm": 0.000956893723923713,
-      "learning_rate": 0.00019999744547379868,
-      "loss": 46.0,
-      "step": 29768
-    },
-    {
-      "epoch": 2.2760479385285852,
-      "grad_norm": 0.0006701195961795747,
-      "learning_rate": 0.00019999744530210985,
-      "loss": 46.0,
-      "step": 29769
-    },
-    {
-      "epoch": 2.276124395511975,
-      "grad_norm": 0.0016100106295198202,
-      "learning_rate": 0.00019999744513041524,
-      "loss": 46.0,
-      "step": 29770
-    },
-    {
-      "epoch": 2.2762008524953647,
-      "grad_norm": 0.000908153597265482,
-      "learning_rate": 0.00019999744495871486,
-      "loss": 46.0,
-      "step": 29771
-    },
-    {
-      "epoch": 2.2762773094787545,
-      "grad_norm": 0.0010683792643249035,
-      "learning_rate": 0.00019999744478700873,
-      "loss": 46.0,
-      "step": 29772
-    },
-    {
-      "epoch": 2.2763537664621443,
-      "grad_norm": 0.0010068045230582356,
-      "learning_rate": 0.0001999974446152968,
-      "loss": 46.0,
-      "step": 29773
-    },
-    {
-      "epoch": 2.276430223445534,
-      "grad_norm": 0.0007091158186085522,
-      "learning_rate": 0.00019999744444357912,
-      "loss": 46.0,
-      "step": 29774
-    },
-    {
-      "epoch": 2.276506680428924,
-      "grad_norm": 0.0006944980123080313,
-      "learning_rate": 0.00019999744427185565,
-      "loss": 46.0,
-      "step": 29775
-    },
-    {
-      "epoch": 2.2765831374123136,
-      "grad_norm": 0.0012299803784117103,
-      "learning_rate": 0.00019999744410012645,
-      "loss": 46.0,
-      "step": 29776
-    },
-    {
-      "epoch": 2.276659594395703,
-      "grad_norm": 0.0010052475845441222,
-      "learning_rate": 0.00019999744392839143,
-      "loss": 46.0,
-      "step": 29777
-    },
-    {
-      "epoch": 2.2767360513790926,
-      "grad_norm": 0.000735848443582654,
-      "learning_rate": 0.0001999974437566507,
-      "loss": 46.0,
-      "step": 29778
-    },
-    {
-      "epoch": 2.2768125083624824,
-      "grad_norm": 0.0009324339334852993,
-      "learning_rate": 0.00019999744358490412,
-      "loss": 46.0,
-      "step": 29779
-    },
-    {
-      "epoch": 2.276888965345872,
-      "grad_norm": 0.0026561864651739597,
-      "learning_rate": 0.00019999744341315183,
-      "loss": 46.0,
-      "step": 29780
-    },
-    {
-      "epoch": 2.276965422329262,
-      "grad_norm": 0.0018196672899648547,
-      "learning_rate": 0.00019999744324139375,
-      "loss": 46.0,
-      "step": 29781
-    },
-    {
-      "epoch": 2.2770418793126517,
-      "grad_norm": 0.0010685429442673922,
-      "learning_rate": 0.00019999744306962994,
-      "loss": 46.0,
-      "step": 29782
-    },
-    {
-      "epoch": 2.2771183362960414,
-      "grad_norm": 0.0009594897856004536,
-      "learning_rate": 0.00019999744289786033,
-      "loss": 46.0,
-      "step": 29783
-    },
-    {
-      "epoch": 2.277194793279431,
-      "grad_norm": 0.0013892094139009714,
-      "learning_rate": 0.00019999744272608495,
-      "loss": 46.0,
-      "step": 29784
-    },
-    {
-      "epoch": 2.277271250262821,
-      "grad_norm": 0.0011564657324925065,
-      "learning_rate": 0.0001999974425543038,
-      "loss": 46.0,
-      "step": 29785
-    },
-    {
-      "epoch": 2.2773477072462107,
-      "grad_norm": 0.0018852401990443468,
-      "learning_rate": 0.00019999744238251687,
-      "loss": 46.0,
-      "step": 29786
-    },
-    {
-      "epoch": 2.2774241642296005,
-      "grad_norm": 0.0007022134959697723,
-      "learning_rate": 0.0001999974422107242,
-      "loss": 46.0,
-      "step": 29787
-    },
-    {
-      "epoch": 2.2775006212129902,
-      "grad_norm": 0.0003709019219968468,
-      "learning_rate": 0.00019999744203892572,
-      "loss": 46.0,
-      "step": 29788
-    },
-    {
-      "epoch": 2.27757707819638,
-      "grad_norm": 0.0013929371489211917,
-      "learning_rate": 0.0001999974418671215,
-      "loss": 46.0,
-      "step": 29789
-    },
-    {
-      "epoch": 2.2776535351797693,
-      "grad_norm": 0.007136743050068617,
-      "learning_rate": 0.00019999744169531153,
-      "loss": 46.0,
-      "step": 29790
-    },
-    {
-      "epoch": 2.277729992163159,
-      "grad_norm": 0.003079803427681327,
-      "learning_rate": 0.00019999744152349576,
-      "loss": 46.0,
-      "step": 29791
-    },
-    {
-      "epoch": 2.277806449146549,
-      "grad_norm": 0.00047877951874397695,
-      "learning_rate": 0.00019999744135167422,
-      "loss": 46.0,
-      "step": 29792
-    },
-    {
-      "epoch": 2.2778829061299386,
-      "grad_norm": 0.0004885787493549287,
-      "learning_rate": 0.0001999974411798469,
-      "loss": 46.0,
-      "step": 29793
-    },
-    {
-      "epoch": 2.2779593631133284,
-      "grad_norm": 0.001778580597601831,
-      "learning_rate": 0.00019999744100801384,
-      "loss": 46.0,
-      "step": 29794
-    },
-    {
-      "epoch": 2.278035820096718,
-      "grad_norm": 0.0012993629788979888,
-      "learning_rate": 0.000199997440836175,
-      "loss": 46.0,
-      "step": 29795
-    },
-    {
-      "epoch": 2.278112277080108,
-      "grad_norm": 0.0024956779088824987,
-      "learning_rate": 0.0001999974406643304,
-      "loss": 46.0,
-      "step": 29796
-    },
-    {
-      "epoch": 2.2781887340634976,
-      "grad_norm": 0.0010920710628852248,
-      "learning_rate": 0.00019999744049248,
-      "loss": 46.0,
-      "step": 29797
-    },
-    {
-      "epoch": 2.2782651910468874,
-      "grad_norm": 0.002022413071244955,
-      "learning_rate": 0.00019999744032062386,
-      "loss": 46.0,
-      "step": 29798
-    },
-    {
-      "epoch": 2.2783416480302767,
-      "grad_norm": 0.002443901728838682,
-      "learning_rate": 0.00019999744014876194,
-      "loss": 46.0,
-      "step": 29799
-    },
-    {
-      "epoch": 2.2784181050136665,
-      "grad_norm": 0.003246761392802,
-      "learning_rate": 0.00019999743997689424,
-      "loss": 46.0,
-      "step": 29800
-    },
-    {
-      "epoch": 2.2784945619970562,
-      "grad_norm": 0.0005772531148977578,
-      "learning_rate": 0.0001999974398050208,
-      "loss": 46.0,
-      "step": 29801
-    },
-    {
-      "epoch": 2.278571018980446,
-      "grad_norm": 0.0007120491936802864,
-      "learning_rate": 0.00019999743963314157,
-      "loss": 46.0,
-      "step": 29802
-    },
-    {
-      "epoch": 2.2786474759638358,
-      "grad_norm": 0.0008428823784925044,
-      "learning_rate": 0.00019999743946125657,
-      "loss": 46.0,
-      "step": 29803
-    },
-    {
-      "epoch": 2.2787239329472255,
-      "grad_norm": 0.00045496365055441856,
-      "learning_rate": 0.0001999974392893658,
-      "loss": 46.0,
-      "step": 29804
-    },
-    {
-      "epoch": 2.2788003899306153,
-      "grad_norm": 0.0004526723932940513,
-      "learning_rate": 0.00019999743911746926,
-      "loss": 46.0,
-      "step": 29805
-    },
-    {
-      "epoch": 2.278876846914005,
-      "grad_norm": 0.000582497101277113,
-      "learning_rate": 0.00019999743894556695,
-      "loss": 46.0,
-      "step": 29806
-    },
-    {
-      "epoch": 2.278953303897395,
-      "grad_norm": 0.0008856842177920043,
-      "learning_rate": 0.0001999974387736589,
-      "loss": 46.0,
-      "step": 29807
-    },
-    {
-      "epoch": 2.2790297608807846,
-      "grad_norm": 0.0016849333187565207,
-      "learning_rate": 0.00019999743860174505,
-      "loss": 46.0,
-      "step": 29808
-    },
-    {
-      "epoch": 2.2791062178641743,
-      "grad_norm": 0.0005244106287136674,
-      "learning_rate": 0.00019999743842982545,
-      "loss": 46.0,
-      "step": 29809
-    },
-    {
-      "epoch": 2.279182674847564,
-      "grad_norm": 0.0010295978281646967,
-      "learning_rate": 0.00019999743825790004,
-      "loss": 46.0,
-      "step": 29810
-    },
-    {
-      "epoch": 2.279259131830954,
-      "grad_norm": 0.004000594839453697,
-      "learning_rate": 0.0001999974380859689,
-      "loss": 46.0,
-      "step": 29811
-    },
-    {
-      "epoch": 2.279335588814343,
-      "grad_norm": 0.0008389134891331196,
-      "learning_rate": 0.00019999743791403198,
-      "loss": 46.0,
-      "step": 29812
-    },
-    {
-      "epoch": 2.279412045797733,
-      "grad_norm": 0.00045328133273869753,
-      "learning_rate": 0.00019999743774208928,
-      "loss": 46.0,
-      "step": 29813
-    },
-    {
-      "epoch": 2.2794885027811227,
-      "grad_norm": 0.0017156020039692521,
-      "learning_rate": 0.0001999974375701408,
-      "loss": 46.0,
-      "step": 29814
-    },
-    {
-      "epoch": 2.2795649597645125,
-      "grad_norm": 0.0011768981348723173,
-      "learning_rate": 0.0001999974373981866,
-      "loss": 46.0,
-      "step": 29815
-    },
-    {
-      "epoch": 2.279641416747902,
-      "grad_norm": 0.0009523766930215061,
-      "learning_rate": 0.00019999743722622662,
-      "loss": 46.0,
-      "step": 29816
-    },
-    {
-      "epoch": 2.279717873731292,
-      "grad_norm": 0.0010894424049183726,
-      "learning_rate": 0.00019999743705426083,
-      "loss": 46.0,
-      "step": 29817
-    },
-    {
-      "epoch": 2.2797943307146817,
-      "grad_norm": 0.001336152316071093,
-      "learning_rate": 0.00019999743688228931,
-      "loss": 46.0,
-      "step": 29818
-    },
-    {
-      "epoch": 2.2798707876980715,
-      "grad_norm": 0.0010350003140047193,
-      "learning_rate": 0.000199997436710312,
-      "loss": 46.0,
-      "step": 29819
-    },
-    {
-      "epoch": 2.2799472446814613,
-      "grad_norm": 0.0022319364361464977,
-      "learning_rate": 0.00019999743653832894,
-      "loss": 46.0,
-      "step": 29820
-    },
-    {
-      "epoch": 2.2800237016648506,
-      "grad_norm": 0.0007240134291350842,
-      "learning_rate": 0.00019999743636634008,
-      "loss": 46.0,
-      "step": 29821
-    },
-    {
-      "epoch": 2.2801001586482403,
-      "grad_norm": 0.0012203151127323508,
-      "learning_rate": 0.00019999743619434547,
-      "loss": 46.0,
-      "step": 29822
-    },
-    {
-      "epoch": 2.28017661563163,
-      "grad_norm": 0.0013332647504284978,
-      "learning_rate": 0.0001999974360223451,
-      "loss": 46.0,
-      "step": 29823
-    },
-    {
-      "epoch": 2.28025307261502,
-      "grad_norm": 0.003884687786921859,
-      "learning_rate": 0.00019999743585033894,
-      "loss": 46.0,
-      "step": 29824
-    },
-    {
-      "epoch": 2.2803295295984096,
-      "grad_norm": 0.0010043522343039513,
-      "learning_rate": 0.000199997435678327,
-      "loss": 46.0,
-      "step": 29825
-    },
-    {
-      "epoch": 2.2804059865817994,
-      "grad_norm": 0.0025026064831763506,
-      "learning_rate": 0.0001999974355063093,
-      "loss": 46.0,
-      "step": 29826
-    },
-    {
-      "epoch": 2.280482443565189,
-      "grad_norm": 0.000705933605786413,
-      "learning_rate": 0.00019999743533428586,
-      "loss": 46.0,
-      "step": 29827
-    },
-    {
-      "epoch": 2.280558900548579,
-      "grad_norm": 0.0018747919239103794,
-      "learning_rate": 0.00019999743516225664,
-      "loss": 46.0,
-      "step": 29828
-    },
-    {
-      "epoch": 2.2806353575319687,
-      "grad_norm": 0.00440255319699645,
-      "learning_rate": 0.00019999743499022165,
-      "loss": 46.0,
-      "step": 29829
-    },
-    {
-      "epoch": 2.2807118145153584,
-      "grad_norm": 0.0008105953456833959,
-      "learning_rate": 0.00019999743481818088,
-      "loss": 46.0,
-      "step": 29830
-    },
-    {
-      "epoch": 2.280788271498748,
-      "grad_norm": 0.0013962170341983438,
-      "learning_rate": 0.00019999743464613434,
-      "loss": 46.0,
-      "step": 29831
-    },
-    {
-      "epoch": 2.280864728482138,
-      "grad_norm": 0.000740594114176929,
-      "learning_rate": 0.00019999743447408203,
-      "loss": 46.0,
-      "step": 29832
-    },
-    {
-      "epoch": 2.2809411854655273,
-      "grad_norm": 0.0007576592615805566,
-      "learning_rate": 0.00019999743430202397,
-      "loss": 46.0,
-      "step": 29833
-    },
-    {
-      "epoch": 2.281017642448917,
-      "grad_norm": 0.0009967153891921043,
-      "learning_rate": 0.0001999974341299601,
-      "loss": 46.0,
-      "step": 29834
-    },
-    {
-      "epoch": 2.281094099432307,
-      "grad_norm": 0.0013766296906396747,
-      "learning_rate": 0.0001999974339578905,
-      "loss": 46.0,
-      "step": 29835
-    },
-    {
-      "epoch": 2.2811705564156965,
-      "grad_norm": 0.001410242053680122,
-      "learning_rate": 0.00019999743378581512,
-      "loss": 46.0,
-      "step": 29836
-    },
-    {
-      "epoch": 2.2812470133990863,
-      "grad_norm": 0.0006855498650111258,
-      "learning_rate": 0.00019999743361373397,
-      "loss": 46.0,
-      "step": 29837
-    },
-    {
-      "epoch": 2.281323470382476,
-      "grad_norm": 0.00066139898262918,
-      "learning_rate": 0.00019999743344164707,
-      "loss": 46.0,
-      "step": 29838
-    },
-    {
-      "epoch": 2.281399927365866,
-      "grad_norm": 0.0007244086591526866,
-      "learning_rate": 0.00019999743326955437,
-      "loss": 46.0,
-      "step": 29839
-    },
-    {
-      "epoch": 2.2814763843492556,
-      "grad_norm": 0.0006686818669550121,
-      "learning_rate": 0.0001999974330974559,
-      "loss": 46.0,
-      "step": 29840
-    },
-    {
-      "epoch": 2.2815528413326454,
-      "grad_norm": 0.00043235751218162477,
-      "learning_rate": 0.00019999743292535164,
-      "loss": 46.0,
-      "step": 29841
-    },
-    {
-      "epoch": 2.281629298316035,
-      "grad_norm": 0.0006182743818499148,
-      "learning_rate": 0.00019999743275324165,
-      "loss": 46.0,
-      "step": 29842
-    },
-    {
-      "epoch": 2.2817057552994244,
-      "grad_norm": 0.001644460717216134,
-      "learning_rate": 0.00019999743258112588,
-      "loss": 46.0,
-      "step": 29843
-    },
-    {
-      "epoch": 2.281782212282814,
-      "grad_norm": 0.002482478739693761,
-      "learning_rate": 0.00019999743240900434,
-      "loss": 46.0,
-      "step": 29844
-    },
-    {
-      "epoch": 2.281858669266204,
-      "grad_norm": 0.002156356815248728,
-      "learning_rate": 0.00019999743223687703,
-      "loss": 46.0,
-      "step": 29845
-    },
-    {
-      "epoch": 2.2819351262495937,
-      "grad_norm": 0.0010309579083696008,
-      "learning_rate": 0.00019999743206474397,
-      "loss": 46.0,
-      "step": 29846
-    },
-    {
-      "epoch": 2.2820115832329835,
-      "grad_norm": 0.0010222181444987655,
-      "learning_rate": 0.00019999743189260514,
-      "loss": 46.0,
-      "step": 29847
-    },
-    {
-      "epoch": 2.2820880402163732,
-      "grad_norm": 0.0021566657815128565,
-      "learning_rate": 0.0001999974317204605,
-      "loss": 46.0,
-      "step": 29848
-    },
-    {
-      "epoch": 2.282164497199763,
-      "grad_norm": 0.002492171246558428,
-      "learning_rate": 0.00019999743154831012,
-      "loss": 46.0,
-      "step": 29849
-    },
-    {
-      "epoch": 2.2822409541831528,
-      "grad_norm": 0.0013407922815531492,
-      "learning_rate": 0.000199997431376154,
-      "loss": 46.0,
-      "step": 29850
-    },
-    {
-      "epoch": 2.2823174111665425,
-      "grad_norm": 0.0008965315064415336,
-      "learning_rate": 0.00019999743120399207,
-      "loss": 46.0,
-      "step": 29851
-    },
-    {
-      "epoch": 2.2823938681499323,
-      "grad_norm": 0.002933842595666647,
-      "learning_rate": 0.00019999743103182434,
-      "loss": 46.0,
-      "step": 29852
-    },
-    {
-      "epoch": 2.282470325133322,
-      "grad_norm": 0.0007745855255052447,
-      "learning_rate": 0.00019999743085965087,
-      "loss": 46.0,
-      "step": 29853
-    },
-    {
-      "epoch": 2.282546782116712,
-      "grad_norm": 0.00517926923930645,
-      "learning_rate": 0.00019999743068747168,
-      "loss": 46.0,
-      "step": 29854
-    },
-    {
-      "epoch": 2.282623239100101,
-      "grad_norm": 0.0013732551597058773,
-      "learning_rate": 0.00019999743051528666,
-      "loss": 46.0,
-      "step": 29855
-    },
-    {
-      "epoch": 2.282699696083491,
-      "grad_norm": 0.0012999350437894464,
-      "learning_rate": 0.0001999974303430959,
-      "loss": 46.0,
-      "step": 29856
-    },
-    {
-      "epoch": 2.2827761530668806,
-      "grad_norm": 0.0008606440969742835,
-      "learning_rate": 0.00019999743017089935,
-      "loss": 46.0,
-      "step": 29857
-    },
-    {
-      "epoch": 2.2828526100502704,
-      "grad_norm": 0.0005423149559646845,
-      "learning_rate": 0.00019999742999869704,
-      "loss": 46.0,
-      "step": 29858
-    },
-    {
-      "epoch": 2.28292906703366,
-      "grad_norm": 0.0017095900839194655,
-      "learning_rate": 0.00019999742982648898,
-      "loss": 46.0,
-      "step": 29859
-    },
-    {
-      "epoch": 2.28300552401705,
-      "grad_norm": 0.0012269478756934404,
-      "learning_rate": 0.00019999742965427512,
-      "loss": 46.0,
-      "step": 29860
-    },
-    {
-      "epoch": 2.2830819810004397,
-      "grad_norm": 0.0009281620150431991,
-      "learning_rate": 0.0001999974294820555,
-      "loss": 46.0,
-      "step": 29861
-    },
-    {
-      "epoch": 2.2831584379838294,
-      "grad_norm": 0.000761642586439848,
-      "learning_rate": 0.0001999974293098301,
-      "loss": 46.0,
-      "step": 29862
-    },
-    {
-      "epoch": 2.283234894967219,
-      "grad_norm": 0.0007000572513788939,
-      "learning_rate": 0.00019999742913759895,
-      "loss": 46.0,
-      "step": 29863
-    },
-    {
-      "epoch": 2.283311351950609,
-      "grad_norm": 0.0002810922160279006,
-      "learning_rate": 0.00019999742896536205,
-      "loss": 46.0,
-      "step": 29864
-    },
-    {
-      "epoch": 2.2833878089339983,
-      "grad_norm": 0.0009073476539924741,
-      "learning_rate": 0.00019999742879311935,
-      "loss": 46.0,
-      "step": 29865
-    },
-    {
-      "epoch": 2.283464265917388,
-      "grad_norm": 0.0023306068032979965,
-      "learning_rate": 0.00019999742862087088,
-      "loss": 46.0,
-      "step": 29866
-    },
-    {
-      "epoch": 2.283540722900778,
-      "grad_norm": 0.0026636889670044184,
-      "learning_rate": 0.00019999742844861666,
-      "loss": 46.0,
-      "step": 29867
-    },
-    {
-      "epoch": 2.2836171798841676,
-      "grad_norm": 0.0037422257009893656,
-      "learning_rate": 0.00019999742827635667,
-      "loss": 46.0,
-      "step": 29868
-    },
-    {
-      "epoch": 2.2836936368675573,
-      "grad_norm": 0.0008414930780418217,
-      "learning_rate": 0.00019999742810409088,
-      "loss": 46.0,
-      "step": 29869
-    },
-    {
-      "epoch": 2.283770093850947,
-      "grad_norm": 0.0005929945036768913,
-      "learning_rate": 0.00019999742793181934,
-      "loss": 46.0,
-      "step": 29870
-    },
-    {
-      "epoch": 2.283846550834337,
-      "grad_norm": 0.0012171786511316895,
-      "learning_rate": 0.00019999742775954205,
-      "loss": 46.0,
-      "step": 29871
-    },
-    {
-      "epoch": 2.2839230078177266,
-      "grad_norm": 0.0004863725043833256,
-      "learning_rate": 0.00019999742758725896,
-      "loss": 46.0,
-      "step": 29872
-    },
-    {
-      "epoch": 2.2839994648011164,
-      "grad_norm": 0.0014412231976166368,
-      "learning_rate": 0.0001999974274149701,
-      "loss": 46.0,
-      "step": 29873
-    },
-    {
-      "epoch": 2.284075921784506,
-      "grad_norm": 0.001976309111341834,
-      "learning_rate": 0.0001999974272426755,
-      "loss": 46.0,
-      "step": 29874
-    },
-    {
-      "epoch": 2.284152378767896,
-      "grad_norm": 0.0014832784654572606,
-      "learning_rate": 0.0001999974270703751,
-      "loss": 46.0,
-      "step": 29875
-    },
-    {
-      "epoch": 2.2842288357512857,
-      "grad_norm": 0.0011276789009571075,
-      "learning_rate": 0.00019999742689806894,
-      "loss": 46.0,
-      "step": 29876
-    },
-    {
-      "epoch": 2.284305292734675,
-      "grad_norm": 0.0005336818867363036,
-      "learning_rate": 0.00019999742672575704,
-      "loss": 46.0,
-      "step": 29877
-    },
-    {
-      "epoch": 2.2843817497180647,
-      "grad_norm": 0.0009416636312380433,
-      "learning_rate": 0.00019999742655343934,
-      "loss": 46.0,
-      "step": 29878
-    },
-    {
-      "epoch": 2.2844582067014545,
-      "grad_norm": 0.0009630148415453732,
-      "learning_rate": 0.00019999742638111587,
-      "loss": 46.0,
-      "step": 29879
-    },
-    {
-      "epoch": 2.2845346636848443,
-      "grad_norm": 0.0008959217229858041,
-      "learning_rate": 0.00019999742620878665,
-      "loss": 46.0,
-      "step": 29880
-    },
-    {
-      "epoch": 2.284611120668234,
-      "grad_norm": 0.0006849021883681417,
-      "learning_rate": 0.00019999742603645163,
-      "loss": 46.0,
-      "step": 29881
-    },
-    {
-      "epoch": 2.284687577651624,
-      "grad_norm": 0.0010131739545613527,
-      "learning_rate": 0.00019999742586411087,
-      "loss": 46.0,
-      "step": 29882
-    },
-    {
-      "epoch": 2.2847640346350135,
-      "grad_norm": 0.001503929146565497,
-      "learning_rate": 0.00019999742569176433,
-      "loss": 46.0,
-      "step": 29883
-    },
-    {
-      "epoch": 2.2848404916184033,
-      "grad_norm": 0.0009931280510500073,
-      "learning_rate": 0.000199997425519412,
-      "loss": 46.0,
-      "step": 29884
-    },
-    {
-      "epoch": 2.284916948601793,
-      "grad_norm": 0.0009015887626446784,
-      "learning_rate": 0.00019999742534705393,
-      "loss": 46.0,
-      "step": 29885
-    },
-    {
-      "epoch": 2.284993405585183,
-      "grad_norm": 0.0009522573673166335,
-      "learning_rate": 0.0001999974251746901,
-      "loss": 46.0,
-      "step": 29886
-    },
-    {
-      "epoch": 2.285069862568572,
-      "grad_norm": 0.0006824164884164929,
-      "learning_rate": 0.0001999974250023205,
-      "loss": 46.0,
-      "step": 29887
-    },
-    {
-      "epoch": 2.285146319551962,
-      "grad_norm": 0.0007158354856073856,
-      "learning_rate": 0.00019999742482994508,
-      "loss": 46.0,
-      "step": 29888
-    },
-    {
-      "epoch": 2.2852227765353517,
-      "grad_norm": 0.0004895995953120291,
-      "learning_rate": 0.00019999742465756393,
-      "loss": 46.0,
-      "step": 29889
-    },
-    {
-      "epoch": 2.2852992335187414,
-      "grad_norm": 0.008395830169320107,
-      "learning_rate": 0.000199997424485177,
-      "loss": 46.0,
-      "step": 29890
-    },
-    {
-      "epoch": 2.285375690502131,
-      "grad_norm": 0.000889509916305542,
-      "learning_rate": 0.00019999742431278433,
-      "loss": 46.0,
-      "step": 29891
-    },
-    {
-      "epoch": 2.285452147485521,
-      "grad_norm": 0.0013936301693320274,
-      "learning_rate": 0.00019999742414038583,
-      "loss": 46.0,
-      "step": 29892
-    },
-    {
-      "epoch": 2.2855286044689107,
-      "grad_norm": 0.0006106882356107235,
-      "learning_rate": 0.0001999974239679816,
-      "loss": 46.0,
-      "step": 29893
-    },
-    {
-      "epoch": 2.2856050614523005,
-      "grad_norm": 0.006154028233140707,
-      "learning_rate": 0.0001999974237955716,
-      "loss": 46.0,
-      "step": 29894
-    },
-    {
-      "epoch": 2.2856815184356902,
-      "grad_norm": 0.0005006272112950683,
-      "learning_rate": 0.00019999742362315583,
-      "loss": 46.0,
-      "step": 29895
-    },
-    {
-      "epoch": 2.28575797541908,
-      "grad_norm": 0.0012299169320613146,
-      "learning_rate": 0.0001999974234507343,
-      "loss": 46.0,
-      "step": 29896
-    },
-    {
-      "epoch": 2.2858344324024698,
-      "grad_norm": 0.001086096977815032,
-      "learning_rate": 0.00019999742327830698,
-      "loss": 46.0,
-      "step": 29897
-    },
-    {
-      "epoch": 2.2859108893858595,
-      "grad_norm": 0.0012332922779023647,
-      "learning_rate": 0.0001999974231058739,
-      "loss": 46.0,
-      "step": 29898
-    },
-    {
-      "epoch": 2.285987346369249,
-      "grad_norm": 0.0012582505587488413,
-      "learning_rate": 0.00019999742293343506,
-      "loss": 46.0,
-      "step": 29899
-    },
-    {
-      "epoch": 2.2860638033526386,
-      "grad_norm": 0.0005635573179461062,
-      "learning_rate": 0.00019999742276099043,
-      "loss": 46.0,
-      "step": 29900
-    },
-    {
-      "epoch": 2.2861402603360284,
-      "grad_norm": 0.0013460407499223948,
-      "learning_rate": 0.00019999742258854005,
-      "loss": 46.0,
-      "step": 29901
-    },
-    {
-      "epoch": 2.286216717319418,
-      "grad_norm": 0.00185186299495399,
-      "learning_rate": 0.00019999742241608387,
-      "loss": 46.0,
-      "step": 29902
-    },
-    {
-      "epoch": 2.286293174302808,
-      "grad_norm": 0.0011935721850022674,
-      "learning_rate": 0.00019999742224362197,
-      "loss": 46.0,
-      "step": 29903
-    },
-    {
-      "epoch": 2.2863696312861976,
-      "grad_norm": 0.0013290493516251445,
-      "learning_rate": 0.00019999742207115425,
-      "loss": 46.0,
-      "step": 29904
-    },
-    {
-      "epoch": 2.2864460882695874,
-      "grad_norm": 0.0009953428525477648,
-      "learning_rate": 0.00019999742189868078,
-      "loss": 46.0,
-      "step": 29905
-    },
-    {
-      "epoch": 2.286522545252977,
-      "grad_norm": 0.0010958951897919178,
-      "learning_rate": 0.00019999742172620156,
-      "loss": 46.0,
-      "step": 29906
-    },
-    {
-      "epoch": 2.286599002236367,
-      "grad_norm": 0.00038658553967252374,
-      "learning_rate": 0.00019999742155371654,
-      "loss": 46.0,
-      "step": 29907
-    },
-    {
-      "epoch": 2.2866754592197567,
-      "grad_norm": 0.003159634303301573,
-      "learning_rate": 0.00019999742138122577,
-      "loss": 46.0,
-      "step": 29908
-    },
-    {
-      "epoch": 2.286751916203146,
-      "grad_norm": 0.0006177702452987432,
-      "learning_rate": 0.00019999742120872923,
-      "loss": 46.0,
-      "step": 29909
-    },
-    {
-      "epoch": 2.2868283731865358,
-      "grad_norm": 0.000541712564881891,
-      "learning_rate": 0.00019999742103622695,
-      "loss": 46.0,
-      "step": 29910
-    },
-    {
-      "epoch": 2.2869048301699255,
-      "grad_norm": 0.0014085343573242426,
-      "learning_rate": 0.00019999742086371884,
-      "loss": 46.0,
-      "step": 29911
-    },
-    {
-      "epoch": 2.2869812871533153,
-      "grad_norm": 0.0007358420989476144,
-      "learning_rate": 0.000199997420691205,
-      "loss": 46.0,
-      "step": 29912
-    },
-    {
-      "epoch": 2.287057744136705,
-      "grad_norm": 0.005819522775709629,
-      "learning_rate": 0.00019999742051868537,
-      "loss": 46.0,
-      "step": 29913
-    },
-    {
-      "epoch": 2.287134201120095,
-      "grad_norm": 0.00030125753255560994,
-      "learning_rate": 0.00019999742034616,
-      "loss": 46.0,
-      "step": 29914
-    },
-    {
-      "epoch": 2.2872106581034846,
-      "grad_norm": 0.002479326678439975,
-      "learning_rate": 0.00019999742017362884,
-      "loss": 46.0,
-      "step": 29915
-    },
-    {
-      "epoch": 2.2872871150868743,
-      "grad_norm": 0.0004563711991067976,
-      "learning_rate": 0.0001999974200010919,
-      "loss": 46.0,
-      "step": 29916
-    },
-    {
-      "epoch": 2.287363572070264,
-      "grad_norm": 0.0006198675255291164,
-      "learning_rate": 0.0001999974198285492,
-      "loss": 46.0,
-      "step": 29917
-    },
-    {
-      "epoch": 2.287440029053654,
-      "grad_norm": 0.0020556203089654446,
-      "learning_rate": 0.00019999741965600075,
-      "loss": 46.0,
-      "step": 29918
-    },
-    {
-      "epoch": 2.2875164860370436,
-      "grad_norm": 0.004435187671333551,
-      "learning_rate": 0.0001999974194834465,
-      "loss": 46.0,
-      "step": 29919
-    },
-    {
-      "epoch": 2.2875929430204334,
-      "grad_norm": 0.0008035156643018126,
-      "learning_rate": 0.00019999741931088648,
-      "loss": 46.0,
-      "step": 29920
-    },
-    {
-      "epoch": 2.2876694000038227,
-      "grad_norm": 0.0015783481067046523,
-      "learning_rate": 0.00019999741913832072,
-      "loss": 46.0,
-      "step": 29921
-    },
-    {
-      "epoch": 2.2877458569872124,
-      "grad_norm": 0.0007616088259965181,
-      "learning_rate": 0.00019999741896574918,
-      "loss": 46.0,
-      "step": 29922
-    },
-    {
-      "epoch": 2.287822313970602,
-      "grad_norm": 0.0054249633103609085,
-      "learning_rate": 0.00019999741879317187,
-      "loss": 46.0,
-      "step": 29923
-    },
-    {
-      "epoch": 2.287898770953992,
-      "grad_norm": 0.0009115795837715268,
-      "learning_rate": 0.00019999741862058879,
-      "loss": 46.0,
-      "step": 29924
-    },
-    {
-      "epoch": 2.2879752279373817,
-      "grad_norm": 0.0009051746455952525,
-      "learning_rate": 0.00019999741844799993,
-      "loss": 46.0,
-      "step": 29925
-    },
-    {
-      "epoch": 2.2880516849207715,
-      "grad_norm": 0.0028160931542515755,
-      "learning_rate": 0.00019999741827540532,
-      "loss": 46.0,
-      "step": 29926
-    },
-    {
-      "epoch": 2.2881281419041613,
-      "grad_norm": 0.0008904538699425757,
-      "learning_rate": 0.00019999741810280492,
-      "loss": 46.0,
-      "step": 29927
-    },
-    {
-      "epoch": 2.288204598887551,
-      "grad_norm": 0.0006686669075861573,
-      "learning_rate": 0.00019999741793019874,
-      "loss": 46.0,
-      "step": 29928
-    },
-    {
-      "epoch": 2.2882810558709408,
-      "grad_norm": 0.0006663257372565567,
-      "learning_rate": 0.00019999741775758681,
-      "loss": 46.0,
-      "step": 29929
-    },
-    {
-      "epoch": 2.28835751285433,
-      "grad_norm": 0.0008810656727291644,
-      "learning_rate": 0.00019999741758496912,
-      "loss": 46.0,
-      "step": 29930
-    },
-    {
-      "epoch": 2.28843396983772,
-      "grad_norm": 0.0005971870268695056,
-      "learning_rate": 0.00019999741741234564,
-      "loss": 46.0,
-      "step": 29931
-    },
-    {
-      "epoch": 2.2885104268211096,
-      "grad_norm": 0.0008884619455784559,
-      "learning_rate": 0.00019999741723971643,
-      "loss": 46.0,
-      "step": 29932
-    },
-    {
-      "epoch": 2.2885868838044994,
-      "grad_norm": 0.0011491224868223071,
-      "learning_rate": 0.00019999741706708144,
-      "loss": 46.0,
-      "step": 29933
-    },
-    {
-      "epoch": 2.288663340787889,
-      "grad_norm": 0.001505630207248032,
-      "learning_rate": 0.00019999741689444064,
-      "loss": 46.0,
-      "step": 29934
-    },
-    {
-      "epoch": 2.288739797771279,
-      "grad_norm": 0.0007792623364366591,
-      "learning_rate": 0.0001999974167217941,
-      "loss": 46.0,
-      "step": 29935
-    },
-    {
-      "epoch": 2.2888162547546687,
-      "grad_norm": 0.0008154707029461861,
-      "learning_rate": 0.0001999974165491418,
-      "loss": 46.0,
-      "step": 29936
-    },
-    {
-      "epoch": 2.2888927117380584,
-      "grad_norm": 0.000983793055638671,
-      "learning_rate": 0.00019999741637648374,
-      "loss": 46.0,
-      "step": 29937
-    },
-    {
-      "epoch": 2.288969168721448,
-      "grad_norm": 0.0006218686467036605,
-      "learning_rate": 0.00019999741620381985,
-      "loss": 46.0,
-      "step": 29938
-    },
-    {
-      "epoch": 2.289045625704838,
-      "grad_norm": 0.0004598811501637101,
-      "learning_rate": 0.00019999741603115022,
-      "loss": 46.0,
-      "step": 29939
-    },
-    {
-      "epoch": 2.2891220826882277,
-      "grad_norm": 0.0019994915928691626,
-      "learning_rate": 0.00019999741585847484,
-      "loss": 46.0,
-      "step": 29940
-    },
-    {
-      "epoch": 2.2891985396716175,
-      "grad_norm": 0.0004396119329612702,
-      "learning_rate": 0.0001999974156857937,
-      "loss": 46.0,
-      "step": 29941
-    },
-    {
-      "epoch": 2.289274996655007,
-      "grad_norm": 0.0009876680560410023,
-      "learning_rate": 0.00019999741551310677,
-      "loss": 46.0,
-      "step": 29942
-    },
-    {
-      "epoch": 2.2893514536383965,
-      "grad_norm": 0.0024848051834851503,
-      "learning_rate": 0.00019999741534041404,
-      "loss": 46.0,
-      "step": 29943
-    },
-    {
-      "epoch": 2.2894279106217863,
-      "grad_norm": 0.0008179487776942551,
-      "learning_rate": 0.00019999741516771557,
-      "loss": 46.0,
-      "step": 29944
-    },
-    {
-      "epoch": 2.289504367605176,
-      "grad_norm": 0.001497873105108738,
-      "learning_rate": 0.00019999741499501138,
-      "loss": 46.0,
-      "step": 29945
-    },
-    {
-      "epoch": 2.289580824588566,
-      "grad_norm": 0.0014988635666668415,
-      "learning_rate": 0.00019999741482230134,
-      "loss": 46.0,
-      "step": 29946
-    },
-    {
-      "epoch": 2.2896572815719556,
-      "grad_norm": 0.0013089224230498075,
-      "learning_rate": 0.00019999741464958557,
-      "loss": 46.0,
-      "step": 29947
-    },
-    {
-      "epoch": 2.2897337385553453,
-      "grad_norm": 0.0023359693586826324,
-      "learning_rate": 0.000199997414476864,
-      "loss": 46.0,
-      "step": 29948
-    },
-    {
-      "epoch": 2.289810195538735,
-      "grad_norm": 0.002481769537553191,
-      "learning_rate": 0.00019999741430413672,
-      "loss": 46.0,
-      "step": 29949
-    },
-    {
-      "epoch": 2.289886652522125,
-      "grad_norm": 0.0012910236837342381,
-      "learning_rate": 0.00019999741413140364,
-      "loss": 46.0,
-      "step": 29950
-    },
-    {
-      "epoch": 2.2899631095055146,
-      "grad_norm": 0.0023895101621747017,
-      "learning_rate": 0.00019999741395866475,
-      "loss": 46.0,
-      "step": 29951
-    },
-    {
-      "epoch": 2.290039566488904,
-      "grad_norm": 0.0013047148240730166,
-      "learning_rate": 0.00019999741378592015,
-      "loss": 46.0,
-      "step": 29952
-    },
-    {
-      "epoch": 2.2901160234722937,
-      "grad_norm": 0.002602041931822896,
-      "learning_rate": 0.00019999741361316975,
-      "loss": 46.0,
-      "step": 29953
-    },
-    {
-      "epoch": 2.2901924804556835,
-      "grad_norm": 0.0013167491415515542,
-      "learning_rate": 0.0001999974134404136,
-      "loss": 46.0,
-      "step": 29954
-    },
-    {
-      "epoch": 2.2902689374390732,
-      "grad_norm": 0.003955171909183264,
-      "learning_rate": 0.00019999741326765167,
-      "loss": 46.0,
-      "step": 29955
-    },
-    {
-      "epoch": 2.290345394422463,
-      "grad_norm": 0.0011904009152203798,
-      "learning_rate": 0.00019999741309488397,
-      "loss": 46.0,
-      "step": 29956
-    },
-    {
-      "epoch": 2.2904218514058527,
-      "grad_norm": 0.00259603769518435,
-      "learning_rate": 0.0001999974129221105,
-      "loss": 46.0,
-      "step": 29957
-    },
-    {
-      "epoch": 2.2904983083892425,
-      "grad_norm": 0.0013402254553511739,
-      "learning_rate": 0.00019999741274933126,
-      "loss": 46.0,
-      "step": 29958
-    },
-    {
-      "epoch": 2.2905747653726323,
-      "grad_norm": 0.0010348311625421047,
-      "learning_rate": 0.00019999741257654624,
-      "loss": 46.0,
-      "step": 29959
-    },
-    {
-      "epoch": 2.290651222356022,
-      "grad_norm": 0.0031335444655269384,
-      "learning_rate": 0.00019999741240375548,
-      "loss": 46.0,
-      "step": 29960
-    },
-    {
-      "epoch": 2.290727679339412,
-      "grad_norm": 0.0008222724427469075,
-      "learning_rate": 0.0001999974122309589,
-      "loss": 46.0,
-      "step": 29961
-    },
-    {
-      "epoch": 2.2908041363228016,
-      "grad_norm": 0.0014036670327186584,
-      "learning_rate": 0.0001999974120581566,
-      "loss": 46.0,
-      "step": 29962
-    },
-    {
-      "epoch": 2.2908805933061913,
-      "grad_norm": 0.0012424660380929708,
-      "learning_rate": 0.00019999741188534852,
-      "loss": 46.0,
-      "step": 29963
-    },
-    {
-      "epoch": 2.2909570502895806,
-      "grad_norm": 0.0011386418482288718,
-      "learning_rate": 0.0001999974117125347,
-      "loss": 46.0,
-      "step": 29964
-    },
-    {
-      "epoch": 2.2910335072729704,
-      "grad_norm": 0.0006336446385830641,
-      "learning_rate": 0.00019999741153971506,
-      "loss": 46.0,
-      "step": 29965
-    },
-    {
-      "epoch": 2.29110996425636,
-      "grad_norm": 0.0011672864202409983,
-      "learning_rate": 0.00019999741136688965,
-      "loss": 46.0,
-      "step": 29966
-    },
-    {
-      "epoch": 2.29118642123975,
-      "grad_norm": 0.00043369957711547613,
-      "learning_rate": 0.00019999741119405847,
-      "loss": 46.0,
-      "step": 29967
-    },
-    {
-      "epoch": 2.2912628782231397,
-      "grad_norm": 0.0011581830913200974,
-      "learning_rate": 0.00019999741102122155,
-      "loss": 46.0,
-      "step": 29968
-    },
-    {
-      "epoch": 2.2913393352065294,
-      "grad_norm": 0.0007346579222939909,
-      "learning_rate": 0.00019999741084837885,
-      "loss": 46.0,
-      "step": 29969
-    },
-    {
-      "epoch": 2.291415792189919,
-      "grad_norm": 0.0009424564777873456,
-      "learning_rate": 0.00019999741067553038,
-      "loss": 46.0,
-      "step": 29970
-    },
-    {
-      "epoch": 2.291492249173309,
-      "grad_norm": 0.0012498552678152919,
-      "learning_rate": 0.00019999741050267617,
-      "loss": 46.0,
-      "step": 29971
-    },
-    {
-      "epoch": 2.2915687061566987,
-      "grad_norm": 0.0029973573982715607,
-      "learning_rate": 0.00019999741032981615,
-      "loss": 46.0,
-      "step": 29972
-    },
-    {
-      "epoch": 2.2916451631400885,
-      "grad_norm": 0.0024053265806287527,
-      "learning_rate": 0.00019999741015695036,
-      "loss": 46.0,
-      "step": 29973
-    },
-    {
-      "epoch": 2.291721620123478,
-      "grad_norm": 0.0004428261308930814,
-      "learning_rate": 0.00019999740998407882,
-      "loss": 46.0,
-      "step": 29974
-    },
-    {
-      "epoch": 2.2917980771068676,
-      "grad_norm": 0.0013902232749387622,
-      "learning_rate": 0.0001999974098112015,
-      "loss": 46.0,
-      "step": 29975
-    },
-    {
-      "epoch": 2.2918745340902573,
-      "grad_norm": 0.002455491106957197,
-      "learning_rate": 0.0001999974096383184,
-      "loss": 46.0,
-      "step": 29976
-    },
-    {
-      "epoch": 2.291950991073647,
-      "grad_norm": 0.0016400108579546213,
-      "learning_rate": 0.00019999740946542954,
-      "loss": 46.0,
-      "step": 29977
-    },
-    {
-      "epoch": 2.292027448057037,
-      "grad_norm": 0.0013364371843636036,
-      "learning_rate": 0.0001999974092925349,
-      "loss": 46.0,
-      "step": 29978
-    },
-    {
-      "epoch": 2.2921039050404266,
-      "grad_norm": 0.0009853814262896776,
-      "learning_rate": 0.00019999740911963454,
-      "loss": 46.0,
-      "step": 29979
-    },
-    {
-      "epoch": 2.2921803620238164,
-      "grad_norm": 0.002387967659160495,
-      "learning_rate": 0.00019999740894672839,
-      "loss": 46.0,
-      "step": 29980
-    },
-    {
-      "epoch": 2.292256819007206,
-      "grad_norm": 0.0014814838068559766,
-      "learning_rate": 0.00019999740877381643,
-      "loss": 46.0,
-      "step": 29981
-    },
-    {
-      "epoch": 2.292333275990596,
-      "grad_norm": 0.008097806014120579,
-      "learning_rate": 0.00019999740860089874,
-      "loss": 46.0,
-      "step": 29982
-    },
-    {
-      "epoch": 2.2924097329739856,
-      "grad_norm": 0.0017949416069313884,
-      "learning_rate": 0.00019999740842797527,
-      "loss": 46.0,
-      "step": 29983
-    },
-    {
-      "epoch": 2.2924861899573754,
-      "grad_norm": 0.0006521407631225884,
-      "learning_rate": 0.00019999740825504602,
-      "loss": 46.0,
-      "step": 29984
-    },
-    {
-      "epoch": 2.292562646940765,
-      "grad_norm": 0.0005579882999882102,
-      "learning_rate": 0.00019999740808211103,
-      "loss": 46.0,
-      "step": 29985
-    },
-    {
-      "epoch": 2.2926391039241545,
-      "grad_norm": 0.0008918361854739487,
-      "learning_rate": 0.00019999740790917024,
-      "loss": 46.0,
-      "step": 29986
-    },
-    {
-      "epoch": 2.2927155609075442,
-      "grad_norm": 0.0007091266452334821,
-      "learning_rate": 0.0001999974077362237,
-      "loss": 46.0,
-      "step": 29987
-    },
-    {
-      "epoch": 2.292792017890934,
-      "grad_norm": 0.0009522182517684996,
-      "learning_rate": 0.0001999974075632714,
-      "loss": 46.0,
-      "step": 29988
-    },
-    {
-      "epoch": 2.2928684748743238,
-      "grad_norm": 0.00140350719448179,
-      "learning_rate": 0.0001999974073903133,
-      "loss": 46.0,
-      "step": 29989
-    },
-    {
-      "epoch": 2.2929449318577135,
-      "grad_norm": 0.0009052487439475954,
-      "learning_rate": 0.00019999740721734943,
-      "loss": 46.0,
-      "step": 29990
-    },
-    {
-      "epoch": 2.2930213888411033,
-      "grad_norm": 0.00048189322114922106,
-      "learning_rate": 0.0001999974070443798,
-      "loss": 46.0,
-      "step": 29991
-    },
-    {
-      "epoch": 2.293097845824493,
-      "grad_norm": 0.0009244322427548468,
-      "learning_rate": 0.00019999740687140442,
-      "loss": 46.0,
-      "step": 29992
-    },
-    {
-      "epoch": 2.293174302807883,
-      "grad_norm": 0.0008392860763706267,
-      "learning_rate": 0.00019999740669842325,
-      "loss": 46.0,
-      "step": 29993
-    },
-    {
-      "epoch": 2.2932507597912726,
-      "grad_norm": 0.0004440682241693139,
-      "learning_rate": 0.00019999740652543632,
-      "loss": 46.0,
-      "step": 29994
-    },
-    {
-      "epoch": 2.2933272167746623,
-      "grad_norm": 0.0014964850852265954,
-      "learning_rate": 0.0001999974063524436,
-      "loss": 46.0,
-      "step": 29995
-    },
-    {
-      "epoch": 2.2934036737580517,
-      "grad_norm": 0.0008686116780154407,
-      "learning_rate": 0.00019999740617944513,
-      "loss": 46.0,
-      "step": 29996
-    },
-    {
-      "epoch": 2.2934801307414414,
-      "grad_norm": 0.0015177861787378788,
-      "learning_rate": 0.0001999974060064409,
-      "loss": 46.0,
-      "step": 29997
-    },
-    {
-      "epoch": 2.293556587724831,
-      "grad_norm": 0.0005570792127400637,
-      "learning_rate": 0.0001999974058334309,
-      "loss": 46.0,
-      "step": 29998
-    },
-    {
-      "epoch": 2.293633044708221,
-      "grad_norm": 0.0007198775419965386,
-      "learning_rate": 0.0001999974056604151,
-      "loss": 46.0,
-      "step": 29999
-    },
-    {
-      "epoch": 2.2937095016916107,
-      "grad_norm": 0.000669146073050797,
-      "learning_rate": 0.00019999740548739357,
-      "loss": 46.0,
-      "step": 30000
-    },
-    {
-      "epoch": 2.2937859586750005,
-      "grad_norm": 0.0008395511540584266,
-      "learning_rate": 0.00019999740531436623,
-      "loss": 46.0,
-      "step": 30001
-    },
-    {
-      "epoch": 2.29386241565839,
-      "grad_norm": 0.0005476570804603398,
-      "learning_rate": 0.00019999740514133315,
-      "loss": 46.0,
-      "step": 30002
-    },
-    {
-      "epoch": 2.29393887264178,
-      "grad_norm": 0.0007812397088855505,
-      "learning_rate": 0.00019999740496829432,
-      "loss": 46.0,
-      "step": 30003
-    },
-    {
-      "epoch": 2.2940153296251697,
-      "grad_norm": 0.001501706661656499,
-      "learning_rate": 0.00019999740479524966,
-      "loss": 46.0,
-      "step": 30004
-    },
-    {
-      "epoch": 2.2940917866085595,
-      "grad_norm": 0.0033552004024386406,
-      "learning_rate": 0.00019999740462219926,
-      "loss": 46.0,
-      "step": 30005
-    },
-    {
-      "epoch": 2.2941682435919493,
-      "grad_norm": 0.0019420883618295193,
-      "learning_rate": 0.0001999974044491431,
-      "loss": 46.0,
-      "step": 30006
-    },
-    {
-      "epoch": 2.294244700575339,
-      "grad_norm": 0.000641855294816196,
-      "learning_rate": 0.0001999974042760812,
-      "loss": 46.0,
-      "step": 30007
-    },
-    {
-      "epoch": 2.2943211575587283,
-      "grad_norm": 0.0012845391174778342,
-      "learning_rate": 0.00019999740410301347,
-      "loss": 46.0,
-      "step": 30008
-    },
-    {
-      "epoch": 2.294397614542118,
-      "grad_norm": 0.000756037246901542,
-      "learning_rate": 0.00019999740392994,
-      "loss": 46.0,
-      "step": 30009
-    },
-    {
-      "epoch": 2.294474071525508,
-      "grad_norm": 0.0076445769518613815,
-      "learning_rate": 0.00019999740375686075,
-      "loss": 46.0,
-      "step": 30010
-    },
-    {
-      "epoch": 2.2945505285088976,
-      "grad_norm": 0.000882270629517734,
-      "learning_rate": 0.00019999740358377574,
-      "loss": 46.0,
-      "step": 30011
-    },
-    {
-      "epoch": 2.2946269854922874,
-      "grad_norm": 0.003852242836728692,
-      "learning_rate": 0.00019999740341068495,
-      "loss": 46.0,
-      "step": 30012
-    },
-    {
-      "epoch": 2.294703442475677,
-      "grad_norm": 0.0011800319189205766,
-      "learning_rate": 0.0001999974032375884,
-      "loss": 46.0,
-      "step": 30013
-    },
-    {
-      "epoch": 2.294779899459067,
-      "grad_norm": 0.005899304524064064,
-      "learning_rate": 0.0001999974030644861,
-      "loss": 46.0,
-      "step": 30014
-    },
-    {
-      "epoch": 2.2948563564424567,
-      "grad_norm": 0.005453599151223898,
-      "learning_rate": 0.00019999740289137802,
-      "loss": 46.0,
-      "step": 30015
-    },
-    {
-      "epoch": 2.2949328134258464,
-      "grad_norm": 0.0016312829684466124,
-      "learning_rate": 0.00019999740271826416,
-      "loss": 46.0,
-      "step": 30016
-    },
-    {
-      "epoch": 2.295009270409236,
-      "grad_norm": 0.0029654079116880894,
-      "learning_rate": 0.0001999974025451445,
-      "loss": 46.0,
-      "step": 30017
-    },
-    {
-      "epoch": 2.2950857273926255,
-      "grad_norm": 0.0005732237477786839,
-      "learning_rate": 0.0001999974023720191,
-      "loss": 46.0,
-      "step": 30018
-    },
-    {
-      "epoch": 2.2951621843760153,
-      "grad_norm": 0.0008385609253309667,
-      "learning_rate": 0.00019999740219888795,
-      "loss": 46.0,
-      "step": 30019
-    },
-    {
-      "epoch": 2.295238641359405,
-      "grad_norm": 0.0013872620183974504,
-      "learning_rate": 0.000199997402025751,
-      "loss": 46.0,
-      "step": 30020
-    },
-    {
-      "epoch": 2.295315098342795,
-      "grad_norm": 0.0018239635974168777,
-      "learning_rate": 0.0001999974018526083,
-      "loss": 46.0,
-      "step": 30021
-    },
-    {
-      "epoch": 2.2953915553261846,
-      "grad_norm": 0.001474417862482369,
-      "learning_rate": 0.00019999740167945984,
-      "loss": 46.0,
-      "step": 30022
-    },
-    {
-      "epoch": 2.2954680123095743,
-      "grad_norm": 0.0007369461818598211,
-      "learning_rate": 0.0001999974015063056,
-      "loss": 46.0,
-      "step": 30023
-    },
-    {
-      "epoch": 2.295544469292964,
-      "grad_norm": 0.002494187094271183,
-      "learning_rate": 0.0001999974013331456,
-      "loss": 46.0,
-      "step": 30024
-    },
-    {
-      "epoch": 2.295620926276354,
-      "grad_norm": 0.0020532524213194847,
-      "learning_rate": 0.0001999974011599798,
-      "loss": 46.0,
-      "step": 30025
-    },
-    {
-      "epoch": 2.2956973832597436,
-      "grad_norm": 0.001050082384608686,
-      "learning_rate": 0.00019999740098680823,
-      "loss": 46.0,
-      "step": 30026
-    },
-    {
-      "epoch": 2.2957738402431334,
-      "grad_norm": 0.0012227309634909034,
-      "learning_rate": 0.00019999740081363095,
-      "loss": 46.0,
-      "step": 30027
-    },
-    {
-      "epoch": 2.295850297226523,
-      "grad_norm": 0.0008813955937512219,
-      "learning_rate": 0.00019999740064044784,
-      "loss": 46.0,
-      "step": 30028
-    },
-    {
-      "epoch": 2.295926754209913,
-      "grad_norm": 0.0013336681295186281,
-      "learning_rate": 0.00019999740046725898,
-      "loss": 46.0,
-      "step": 30029
-    },
-    {
-      "epoch": 2.296003211193302,
-      "grad_norm": 0.00067457917612046,
-      "learning_rate": 0.00019999740029406435,
-      "loss": 46.0,
-      "step": 30030
-    },
-    {
-      "epoch": 2.296079668176692,
-      "grad_norm": 0.000672497961204499,
-      "learning_rate": 0.00019999740012086395,
-      "loss": 46.0,
-      "step": 30031
-    },
-    {
-      "epoch": 2.2961561251600817,
-      "grad_norm": 0.001501060905866325,
-      "learning_rate": 0.00019999739994765777,
-      "loss": 46.0,
-      "step": 30032
-    },
-    {
-      "epoch": 2.2962325821434715,
-      "grad_norm": 0.0044987439177930355,
-      "learning_rate": 0.00019999739977444585,
-      "loss": 46.0,
-      "step": 30033
-    },
-    {
-      "epoch": 2.2963090391268612,
-      "grad_norm": 0.01607019267976284,
-      "learning_rate": 0.00019999739960122815,
-      "loss": 46.0,
-      "step": 30034
-    },
-    {
-      "epoch": 2.296385496110251,
-      "grad_norm": 0.0008642894681543112,
-      "learning_rate": 0.00019999739942800466,
-      "loss": 46.0,
-      "step": 30035
-    },
-    {
-      "epoch": 2.2964619530936408,
-      "grad_norm": 0.0005976841785013676,
-      "learning_rate": 0.00019999739925477544,
-      "loss": 46.0,
-      "step": 30036
-    },
-    {
-      "epoch": 2.2965384100770305,
-      "grad_norm": 0.0015923274913802743,
-      "learning_rate": 0.00019999739908154043,
-      "loss": 46.0,
-      "step": 30037
-    },
-    {
-      "epoch": 2.2966148670604203,
-      "grad_norm": 0.0007830019458197057,
-      "learning_rate": 0.00019999739890829964,
-      "loss": 46.0,
-      "step": 30038
-    },
-    {
-      "epoch": 2.29669132404381,
-      "grad_norm": 0.0009042581077665091,
-      "learning_rate": 0.00019999739873505308,
-      "loss": 46.0,
-      "step": 30039
-    },
-    {
-      "epoch": 2.2967677810271994,
-      "grad_norm": 0.000887769041582942,
-      "learning_rate": 0.00019999739856180074,
-      "loss": 46.0,
-      "step": 30040
-    },
-    {
-      "epoch": 2.296844238010589,
-      "grad_norm": 0.000370532157830894,
-      "learning_rate": 0.00019999739838854268,
-      "loss": 46.0,
-      "step": 30041
-    },
-    {
-      "epoch": 2.296920694993979,
-      "grad_norm": 0.003807772183790803,
-      "learning_rate": 0.0001999973982152788,
-      "loss": 46.0,
-      "step": 30042
-    },
-    {
-      "epoch": 2.2969971519773686,
-      "grad_norm": 0.002727210521697998,
-      "learning_rate": 0.00019999739804200917,
-      "loss": 46.0,
-      "step": 30043
-    },
-    {
-      "epoch": 2.2970736089607584,
-      "grad_norm": 0.0009425010648556054,
-      "learning_rate": 0.00019999739786873377,
-      "loss": 46.0,
-      "step": 30044
-    },
-    {
-      "epoch": 2.297150065944148,
-      "grad_norm": 0.00043616388575173914,
-      "learning_rate": 0.0001999973976954526,
-      "loss": 46.0,
-      "step": 30045
-    },
-    {
-      "epoch": 2.297226522927538,
-      "grad_norm": 0.0006302734836935997,
-      "learning_rate": 0.00019999739752216567,
-      "loss": 46.0,
-      "step": 30046
-    },
-    {
-      "epoch": 2.2973029799109277,
-      "grad_norm": 0.0007072797161526978,
-      "learning_rate": 0.00019999739734887295,
-      "loss": 46.0,
-      "step": 30047
-    },
-    {
-      "epoch": 2.2973794368943175,
-      "grad_norm": 0.0006940692546777427,
-      "learning_rate": 0.00019999739717557448,
-      "loss": 46.0,
-      "step": 30048
-    },
-    {
-      "epoch": 2.297455893877707,
-      "grad_norm": 0.008013441227376461,
-      "learning_rate": 0.00019999739700227024,
-      "loss": 46.0,
-      "step": 30049
-    },
-    {
-      "epoch": 2.297532350861097,
-      "grad_norm": 0.00067564332857728,
-      "learning_rate": 0.00019999739682896022,
-      "loss": 46.0,
-      "step": 30050
-    },
-    {
-      "epoch": 2.2976088078444867,
-      "grad_norm": 0.0023563748691231012,
-      "learning_rate": 0.00019999739665564446,
-      "loss": 46.0,
-      "step": 30051
-    },
-    {
-      "epoch": 2.297685264827876,
-      "grad_norm": 0.001248229993507266,
-      "learning_rate": 0.0001999973964823229,
-      "loss": 46.0,
-      "step": 30052
-    },
-    {
-      "epoch": 2.297761721811266,
-      "grad_norm": 0.0003808919573202729,
-      "learning_rate": 0.00019999739630899557,
-      "loss": 46.0,
-      "step": 30053
-    },
-    {
-      "epoch": 2.2978381787946556,
-      "grad_norm": 0.006725207436829805,
-      "learning_rate": 0.00019999739613566248,
-      "loss": 46.0,
-      "step": 30054
-    },
-    {
-      "epoch": 2.2979146357780453,
-      "grad_norm": 0.00045433282502926886,
-      "learning_rate": 0.00019999739596232363,
-      "loss": 46.0,
-      "step": 30055
-    },
-    {
-      "epoch": 2.297991092761435,
-      "grad_norm": 0.0012498340802267194,
-      "learning_rate": 0.000199997395788979,
-      "loss": 46.0,
-      "step": 30056
-    },
-    {
-      "epoch": 2.298067549744825,
-      "grad_norm": 0.0007556340424343944,
-      "learning_rate": 0.0001999973956156286,
-      "loss": 46.0,
-      "step": 30057
-    },
-    {
-      "epoch": 2.2981440067282146,
-      "grad_norm": 0.003670429578050971,
-      "learning_rate": 0.00019999739544227242,
-      "loss": 46.0,
-      "step": 30058
-    },
-    {
-      "epoch": 2.2982204637116044,
-      "grad_norm": 0.0035471406299620867,
-      "learning_rate": 0.00019999739526891047,
-      "loss": 46.0,
-      "step": 30059
-    },
-    {
-      "epoch": 2.298296920694994,
-      "grad_norm": 0.0005399485817179084,
-      "learning_rate": 0.00019999739509554275,
-      "loss": 46.0,
-      "step": 30060
-    },
-    {
-      "epoch": 2.2983733776783835,
-      "grad_norm": 0.003899995470419526,
-      "learning_rate": 0.0001999973949221693,
-      "loss": 46.0,
-      "step": 30061
-    },
-    {
-      "epoch": 2.298449834661773,
-      "grad_norm": 0.0012264401884749532,
-      "learning_rate": 0.00019999739474879004,
-      "loss": 46.0,
-      "step": 30062
-    },
-    {
-      "epoch": 2.298526291645163,
-      "grad_norm": 0.000918879231903702,
-      "learning_rate": 0.00019999739457540505,
-      "loss": 46.0,
-      "step": 30063
-    },
-    {
-      "epoch": 2.2986027486285527,
-      "grad_norm": 0.0006566160591319203,
-      "learning_rate": 0.00019999739440201424,
-      "loss": 46.0,
-      "step": 30064
-    },
-    {
-      "epoch": 2.2986792056119425,
-      "grad_norm": 0.001869792933575809,
-      "learning_rate": 0.0001999973942286177,
-      "loss": 46.0,
-      "step": 30065
-    },
-    {
-      "epoch": 2.2987556625953323,
-      "grad_norm": 0.0034287371672689915,
-      "learning_rate": 0.00019999739405521537,
-      "loss": 46.0,
-      "step": 30066
-    },
-    {
-      "epoch": 2.298832119578722,
-      "grad_norm": 0.004114525858312845,
-      "learning_rate": 0.0001999973938818073,
-      "loss": 46.0,
-      "step": 30067
-    },
-    {
-      "epoch": 2.298908576562112,
-      "grad_norm": 0.0008481629774905741,
-      "learning_rate": 0.00019999739370839343,
-      "loss": 46.0,
-      "step": 30068
-    },
-    {
-      "epoch": 2.2989850335455015,
-      "grad_norm": 0.000555143051315099,
-      "learning_rate": 0.0001999973935349738,
-      "loss": 46.0,
-      "step": 30069
-    },
-    {
-      "epoch": 2.2990614905288913,
-      "grad_norm": 0.0005480466643348336,
-      "learning_rate": 0.0001999973933615484,
-      "loss": 46.0,
-      "step": 30070
-    },
-    {
-      "epoch": 2.299137947512281,
-      "grad_norm": 0.0005450470489449799,
-      "learning_rate": 0.0001999973931881172,
-      "loss": 46.0,
-      "step": 30071
-    },
-    {
-      "epoch": 2.299214404495671,
-      "grad_norm": 0.0013974080793559551,
-      "learning_rate": 0.00019999739301468028,
-      "loss": 46.0,
-      "step": 30072
-    },
-    {
-      "epoch": 2.2992908614790606,
-      "grad_norm": 0.001103778020478785,
-      "learning_rate": 0.00019999739284123758,
-      "loss": 46.0,
-      "step": 30073
-    },
-    {
-      "epoch": 2.29936731846245,
-      "grad_norm": 0.000586771231610328,
-      "learning_rate": 0.00019999739266778906,
-      "loss": 46.0,
-      "step": 30074
-    },
-    {
-      "epoch": 2.2994437754458397,
-      "grad_norm": 0.005902382545173168,
-      "learning_rate": 0.00019999739249433485,
-      "loss": 46.0,
-      "step": 30075
-    },
-    {
-      "epoch": 2.2995202324292294,
-      "grad_norm": 0.00202909461222589,
-      "learning_rate": 0.00019999739232087483,
-      "loss": 46.0,
-      "step": 30076
-    },
-    {
-      "epoch": 2.299596689412619,
-      "grad_norm": 0.0013913969742134213,
-      "learning_rate": 0.00019999739214740904,
-      "loss": 46.0,
-      "step": 30077
-    },
-    {
-      "epoch": 2.299673146396009,
-      "grad_norm": 0.0008264286443591118,
-      "learning_rate": 0.00019999739197393748,
-      "loss": 46.0,
-      "step": 30078
-    },
-    {
-      "epoch": 2.2997496033793987,
-      "grad_norm": 0.0007550848531536758,
-      "learning_rate": 0.00019999739180046015,
-      "loss": 46.0,
-      "step": 30079
-    },
-    {
-      "epoch": 2.2998260603627885,
-      "grad_norm": 0.0009506526403129101,
-      "learning_rate": 0.00019999739162697707,
-      "loss": 46.0,
-      "step": 30080
-    },
-    {
-      "epoch": 2.2999025173461782,
-      "grad_norm": 0.0008553520892746747,
-      "learning_rate": 0.00019999739145348818,
-      "loss": 46.0,
-      "step": 30081
-    },
-    {
-      "epoch": 2.299978974329568,
-      "grad_norm": 0.0010694566881284118,
-      "learning_rate": 0.00019999739127999358,
-      "loss": 46.0,
-      "step": 30082
-    },
-    {
-      "epoch": 2.3000554313129573,
-      "grad_norm": 0.0004899316118098795,
-      "learning_rate": 0.00019999739110649318,
-      "loss": 46.0,
-      "step": 30083
-    },
-    {
-      "epoch": 2.300131888296347,
-      "grad_norm": 0.0008950239280238748,
-      "learning_rate": 0.000199997390932987,
-      "loss": 46.0,
-      "step": 30084
-    },
-    {
-      "epoch": 2.300208345279737,
-      "grad_norm": 0.001813948038034141,
-      "learning_rate": 0.00019999739075947506,
-      "loss": 46.0,
-      "step": 30085
-    },
-    {
-      "epoch": 2.3002848022631266,
-      "grad_norm": 0.0028806107584387064,
-      "learning_rate": 0.00019999739058595734,
-      "loss": 46.0,
-      "step": 30086
-    },
-    {
-      "epoch": 2.3003612592465164,
-      "grad_norm": 0.0012276624329388142,
-      "learning_rate": 0.00019999739041243387,
-      "loss": 46.0,
-      "step": 30087
-    },
-    {
-      "epoch": 2.300437716229906,
-      "grad_norm": 0.0006821006536483765,
-      "learning_rate": 0.00019999739023890463,
-      "loss": 46.0,
-      "step": 30088
-    },
-    {
-      "epoch": 2.300514173213296,
-      "grad_norm": 0.0035636688116937876,
-      "learning_rate": 0.0001999973900653696,
-      "loss": 46.0,
-      "step": 30089
-    },
-    {
-      "epoch": 2.3005906301966856,
-      "grad_norm": 0.0020399398636072874,
-      "learning_rate": 0.0001999973898918288,
-      "loss": 46.0,
-      "step": 30090
-    },
-    {
-      "epoch": 2.3006670871800754,
-      "grad_norm": 0.001916010514833033,
-      "learning_rate": 0.00019999738971828227,
-      "loss": 46.0,
-      "step": 30091
-    },
-    {
-      "epoch": 2.300743544163465,
-      "grad_norm": 0.0010494361631572247,
-      "learning_rate": 0.00019999738954472993,
-      "loss": 46.0,
-      "step": 30092
-    },
-    {
-      "epoch": 2.300820001146855,
-      "grad_norm": 0.0022276821546256542,
-      "learning_rate": 0.00019999738937117182,
-      "loss": 46.0,
-      "step": 30093
-    },
-    {
-      "epoch": 2.3008964581302447,
-      "grad_norm": 0.00485273590311408,
-      "learning_rate": 0.00019999738919760797,
-      "loss": 46.0,
-      "step": 30094
-    },
-    {
-      "epoch": 2.300972915113634,
-      "grad_norm": 0.002339188475161791,
-      "learning_rate": 0.00019999738902403834,
-      "loss": 46.0,
-      "step": 30095
-    },
-    {
-      "epoch": 2.3010493720970238,
-      "grad_norm": 0.00916279386729002,
-      "learning_rate": 0.00019999738885046294,
-      "loss": 46.0,
-      "step": 30096
-    },
-    {
-      "epoch": 2.3011258290804135,
-      "grad_norm": 0.001643507624976337,
-      "learning_rate": 0.0001999973886768818,
-      "loss": 46.0,
-      "step": 30097
-    },
-    {
-      "epoch": 2.3012022860638033,
-      "grad_norm": 0.00224167387932539,
-      "learning_rate": 0.00019999738850329485,
-      "loss": 46.0,
-      "step": 30098
-    },
-    {
-      "epoch": 2.301278743047193,
-      "grad_norm": 0.0005773654556833208,
-      "learning_rate": 0.00019999738832970212,
-      "loss": 46.0,
-      "step": 30099
-    },
-    {
-      "epoch": 2.301355200030583,
-      "grad_norm": 0.0008862800314091146,
-      "learning_rate": 0.00019999738815610366,
-      "loss": 46.0,
-      "step": 30100
-    },
-    {
-      "epoch": 2.3014316570139726,
-      "grad_norm": 0.000557686376851052,
-      "learning_rate": 0.00019999738798249942,
-      "loss": 46.0,
-      "step": 30101
-    },
-    {
-      "epoch": 2.3015081139973623,
-      "grad_norm": 0.005363051779568195,
-      "learning_rate": 0.00019999738780888937,
-      "loss": 46.0,
-      "step": 30102
-    },
-    {
-      "epoch": 2.301584570980752,
-      "grad_norm": 0.0007143226102925837,
-      "learning_rate": 0.0001999973876352736,
-      "loss": 46.0,
-      "step": 30103
-    },
-    {
-      "epoch": 2.301661027964142,
-      "grad_norm": 0.0008837385685183108,
-      "learning_rate": 0.00019999738746165205,
-      "loss": 46.0,
-      "step": 30104
-    },
-    {
-      "epoch": 2.301737484947531,
-      "grad_norm": 0.0028602525126188993,
-      "learning_rate": 0.00019999738728802472,
-      "loss": 46.0,
-      "step": 30105
-    },
-    {
-      "epoch": 2.301813941930921,
-      "grad_norm": 0.001361931674182415,
-      "learning_rate": 0.0001999973871143916,
-      "loss": 46.0,
-      "step": 30106
-    },
-    {
-      "epoch": 2.3018903989143107,
-      "grad_norm": 0.0005813190946355462,
-      "learning_rate": 0.00019999738694075273,
-      "loss": 46.0,
-      "step": 30107
-    },
-    {
-      "epoch": 2.3019668558977004,
-      "grad_norm": 0.0005225312779657543,
-      "learning_rate": 0.0001999973867671081,
-      "loss": 46.0,
-      "step": 30108
-    },
-    {
-      "epoch": 2.30204331288109,
-      "grad_norm": 0.0018029806669801474,
-      "learning_rate": 0.0001999973865934577,
-      "loss": 46.0,
-      "step": 30109
-    },
-    {
-      "epoch": 2.30211976986448,
-      "grad_norm": 0.00046599714551120996,
-      "learning_rate": 0.00019999738641980153,
-      "loss": 46.0,
-      "step": 30110
-    },
-    {
-      "epoch": 2.3021962268478697,
-      "grad_norm": 0.000968014937825501,
-      "learning_rate": 0.0001999973862461396,
-      "loss": 46.0,
-      "step": 30111
-    },
-    {
-      "epoch": 2.3022726838312595,
-      "grad_norm": 0.0008470361935906112,
-      "learning_rate": 0.0001999973860724719,
-      "loss": 46.0,
-      "step": 30112
-    },
-    {
-      "epoch": 2.3023491408146493,
-      "grad_norm": 0.001534096198156476,
-      "learning_rate": 0.0001999973858987984,
-      "loss": 46.0,
-      "step": 30113
-    },
-    {
-      "epoch": 2.302425597798039,
-      "grad_norm": 0.0013971490552648902,
-      "learning_rate": 0.00019999738572511915,
-      "loss": 46.0,
-      "step": 30114
-    },
-    {
-      "epoch": 2.3025020547814288,
-      "grad_norm": 0.0011443597031757236,
-      "learning_rate": 0.0001999973855514341,
-      "loss": 46.0,
-      "step": 30115
-    },
-    {
-      "epoch": 2.3025785117648185,
-      "grad_norm": 0.0007007725071161985,
-      "learning_rate": 0.00019999738537774335,
-      "loss": 46.0,
-      "step": 30116
-    },
-    {
-      "epoch": 2.302654968748208,
-      "grad_norm": 0.0005706321680918336,
-      "learning_rate": 0.0001999973852040468,
-      "loss": 46.0,
-      "step": 30117
-    },
-    {
-      "epoch": 2.3027314257315976,
-      "grad_norm": 0.0025433956179767847,
-      "learning_rate": 0.00019999738503034443,
-      "loss": 46.0,
-      "step": 30118
-    },
-    {
-      "epoch": 2.3028078827149874,
-      "grad_norm": 0.0020307658705860376,
-      "learning_rate": 0.00019999738485663638,
-      "loss": 46.0,
-      "step": 30119
-    },
-    {
-      "epoch": 2.302884339698377,
-      "grad_norm": 0.0009990239050239325,
-      "learning_rate": 0.0001999973846829225,
-      "loss": 46.0,
-      "step": 30120
-    },
-    {
-      "epoch": 2.302960796681767,
-      "grad_norm": 0.0009324668790213764,
-      "learning_rate": 0.00019999738450920287,
-      "loss": 46.0,
-      "step": 30121
-    },
-    {
-      "epoch": 2.3030372536651567,
-      "grad_norm": 0.0004060252394992858,
-      "learning_rate": 0.00019999738433547747,
-      "loss": 46.0,
-      "step": 30122
-    },
-    {
-      "epoch": 2.3031137106485464,
-      "grad_norm": 0.0008862410904839635,
-      "learning_rate": 0.00019999738416174627,
-      "loss": 46.0,
-      "step": 30123
-    },
-    {
-      "epoch": 2.303190167631936,
-      "grad_norm": 0.0026099178940057755,
-      "learning_rate": 0.00019999738398800935,
-      "loss": 46.0,
-      "step": 30124
-    },
-    {
-      "epoch": 2.303266624615326,
-      "grad_norm": 0.011006048880517483,
-      "learning_rate": 0.00019999738381426663,
-      "loss": 46.0,
-      "step": 30125
-    },
-    {
-      "epoch": 2.3033430815987157,
-      "grad_norm": 0.0055942474864423275,
-      "learning_rate": 0.00019999738364051813,
-      "loss": 46.0,
-      "step": 30126
-    },
-    {
-      "epoch": 2.303419538582105,
-      "grad_norm": 0.0008677778532728553,
-      "learning_rate": 0.0001999973834667639,
-      "loss": 46.0,
-      "step": 30127
-    },
-    {
-      "epoch": 2.303495995565495,
-      "grad_norm": 0.0008999879937618971,
-      "learning_rate": 0.00019999738329300385,
-      "loss": 46.0,
-      "step": 30128
-    },
-    {
-      "epoch": 2.3035724525488845,
-      "grad_norm": 0.0031719410326331854,
-      "learning_rate": 0.0001999973831192381,
-      "loss": 46.0,
-      "step": 30129
-    },
-    {
-      "epoch": 2.3036489095322743,
-      "grad_norm": 0.0007733224774710834,
-      "learning_rate": 0.00019999738294546653,
-      "loss": 46.0,
-      "step": 30130
-    },
-    {
-      "epoch": 2.303725366515664,
-      "grad_norm": 0.0013872962445020676,
-      "learning_rate": 0.0001999973827716892,
-      "loss": 46.0,
-      "step": 30131
-    },
-    {
-      "epoch": 2.303801823499054,
-      "grad_norm": 0.001186657347716391,
-      "learning_rate": 0.0001999973825979061,
-      "loss": 46.0,
-      "step": 30132
-    },
-    {
-      "epoch": 2.3038782804824436,
-      "grad_norm": 0.0025625890120863914,
-      "learning_rate": 0.00019999738242411724,
-      "loss": 46.0,
-      "step": 30133
-    },
-    {
-      "epoch": 2.3039547374658333,
-      "grad_norm": 0.0007266376633197069,
-      "learning_rate": 0.00019999738225032258,
-      "loss": 46.0,
-      "step": 30134
-    },
-    {
-      "epoch": 2.304031194449223,
-      "grad_norm": 0.00029289996018633246,
-      "learning_rate": 0.00019999738207652218,
-      "loss": 46.0,
-      "step": 30135
-    },
-    {
-      "epoch": 2.304107651432613,
-      "grad_norm": 0.0008302509668283165,
-      "learning_rate": 0.000199997381902716,
-      "loss": 46.0,
-      "step": 30136
-    },
-    {
-      "epoch": 2.3041841084160026,
-      "grad_norm": 0.0012541902251541615,
-      "learning_rate": 0.00019999738172890406,
-      "loss": 46.0,
-      "step": 30137
-    },
-    {
-      "epoch": 2.3042605653993924,
-      "grad_norm": 0.001695367624051869,
-      "learning_rate": 0.00019999738155508637,
-      "loss": 46.0,
-      "step": 30138
-    },
-    {
-      "epoch": 2.3043370223827817,
-      "grad_norm": 0.0007267678156495094,
-      "learning_rate": 0.00019999738138126288,
-      "loss": 46.0,
-      "step": 30139
-    },
-    {
-      "epoch": 2.3044134793661715,
-      "grad_norm": 0.0007607286097481847,
-      "learning_rate": 0.0001999973812074336,
-      "loss": 46.0,
-      "step": 30140
-    },
-    {
-      "epoch": 2.3044899363495612,
-      "grad_norm": 0.0027221804484725,
-      "learning_rate": 0.00019999738103359862,
-      "loss": 46.0,
-      "step": 30141
-    },
-    {
-      "epoch": 2.304566393332951,
-      "grad_norm": 0.0008816447807475924,
-      "learning_rate": 0.0001999973808597578,
-      "loss": 46.0,
-      "step": 30142
-    },
-    {
-      "epoch": 2.3046428503163408,
-      "grad_norm": 0.00118237582501024,
-      "learning_rate": 0.00019999738068591125,
-      "loss": 46.0,
-      "step": 30143
-    },
-    {
-      "epoch": 2.3047193072997305,
-      "grad_norm": 0.002199642825871706,
-      "learning_rate": 0.00019999738051205892,
-      "loss": 46.0,
-      "step": 30144
-    },
-    {
-      "epoch": 2.3047957642831203,
-      "grad_norm": 0.0005205171182751656,
-      "learning_rate": 0.00019999738033820084,
-      "loss": 46.0,
-      "step": 30145
-    },
-    {
-      "epoch": 2.30487222126651,
-      "grad_norm": 0.0010612175101414323,
-      "learning_rate": 0.00019999738016433696,
-      "loss": 46.0,
-      "step": 30146
-    },
-    {
-      "epoch": 2.3049486782499,
-      "grad_norm": 0.0012619533808901906,
-      "learning_rate": 0.00019999737999046733,
-      "loss": 46.0,
-      "step": 30147
-    },
-    {
-      "epoch": 2.3050251352332896,
-      "grad_norm": 0.0014017487410455942,
-      "learning_rate": 0.0001999973798165919,
-      "loss": 46.0,
-      "step": 30148
-    },
-    {
-      "epoch": 2.305101592216679,
-      "grad_norm": 0.0009200768545269966,
-      "learning_rate": 0.00019999737964271073,
-      "loss": 46.0,
-      "step": 30149
-    },
-    {
-      "epoch": 2.3051780492000686,
-      "grad_norm": 0.0013504025992006063,
-      "learning_rate": 0.00019999737946882378,
-      "loss": 46.0,
-      "step": 30150
-    },
-    {
-      "epoch": 2.3052545061834584,
-      "grad_norm": 0.0005463776760734618,
-      "learning_rate": 0.00019999737929493106,
-      "loss": 46.0,
-      "step": 30151
-    },
-    {
-      "epoch": 2.305330963166848,
-      "grad_norm": 0.0007936308393254876,
-      "learning_rate": 0.0001999973791210326,
-      "loss": 46.0,
-      "step": 30152
-    },
-    {
-      "epoch": 2.305407420150238,
-      "grad_norm": 0.00023521768162027001,
-      "learning_rate": 0.00019999737894712833,
-      "loss": 46.0,
-      "step": 30153
-    },
-    {
-      "epoch": 2.3054838771336277,
-      "grad_norm": 0.0022964964155107737,
-      "learning_rate": 0.00019999737877321835,
-      "loss": 46.0,
-      "step": 30154
-    },
-    {
-      "epoch": 2.3055603341170174,
-      "grad_norm": 0.0014007611898705363,
-      "learning_rate": 0.00019999737859930253,
-      "loss": 46.0,
-      "step": 30155
-    },
-    {
-      "epoch": 2.305636791100407,
-      "grad_norm": 0.011091357097029686,
-      "learning_rate": 0.00019999737842538097,
-      "loss": 46.0,
-      "step": 30156
-    },
-    {
-      "epoch": 2.305713248083797,
-      "grad_norm": 0.002516264794394374,
-      "learning_rate": 0.00019999737825145364,
-      "loss": 46.0,
-      "step": 30157
-    },
-    {
-      "epoch": 2.3057897050671867,
-      "grad_norm": 0.0006208026316016912,
-      "learning_rate": 0.00019999737807752053,
-      "loss": 46.0,
-      "step": 30158
-    },
-    {
-      "epoch": 2.3058661620505765,
-      "grad_norm": 0.001432916265912354,
-      "learning_rate": 0.00019999737790358165,
-      "loss": 46.0,
-      "step": 30159
-    },
-    {
-      "epoch": 2.3059426190339662,
-      "grad_norm": 0.00058855174575001,
-      "learning_rate": 0.00019999737772963703,
-      "loss": 46.0,
-      "step": 30160
-    },
-    {
-      "epoch": 2.3060190760173556,
-      "grad_norm": 0.0009267079294659197,
-      "learning_rate": 0.00019999737755568663,
-      "loss": 46.0,
-      "step": 30161
-    },
-    {
-      "epoch": 2.3060955330007453,
-      "grad_norm": 0.0009682028321549296,
-      "learning_rate": 0.00019999737738173046,
-      "loss": 46.0,
-      "step": 30162
-    },
-    {
-      "epoch": 2.306171989984135,
-      "grad_norm": 0.002434520749375224,
-      "learning_rate": 0.0001999973772077685,
-      "loss": 46.0,
-      "step": 30163
-    },
-    {
-      "epoch": 2.306248446967525,
-      "grad_norm": 0.0008917953819036484,
-      "learning_rate": 0.0001999973770338008,
-      "loss": 46.0,
-      "step": 30164
-    },
-    {
-      "epoch": 2.3063249039509146,
-      "grad_norm": 0.0004295255639590323,
-      "learning_rate": 0.0001999973768598273,
-      "loss": 46.0,
-      "step": 30165
-    },
-    {
-      "epoch": 2.3064013609343044,
-      "grad_norm": 0.0009161210618913174,
-      "learning_rate": 0.00019999737668584806,
-      "loss": 46.0,
-      "step": 30166
-    },
-    {
-      "epoch": 2.306477817917694,
-      "grad_norm": 0.010465409606695175,
-      "learning_rate": 0.00019999737651186304,
-      "loss": 46.0,
-      "step": 30167
-    },
-    {
-      "epoch": 2.306554274901084,
-      "grad_norm": 0.0020112800411880016,
-      "learning_rate": 0.00019999737633787223,
-      "loss": 46.0,
-      "step": 30168
-    },
-    {
-      "epoch": 2.3066307318844737,
-      "grad_norm": 0.0007780776941217482,
-      "learning_rate": 0.00019999737616387567,
-      "loss": 46.0,
-      "step": 30169
-    },
-    {
-      "epoch": 2.3067071888678634,
-      "grad_norm": 0.0005456142826005816,
-      "learning_rate": 0.00019999737598987337,
-      "loss": 46.0,
-      "step": 30170
-    },
-    {
-      "epoch": 2.3067836458512527,
-      "grad_norm": 0.0008832073071971536,
-      "learning_rate": 0.00019999737581586523,
-      "loss": 46.0,
-      "step": 30171
-    },
-    {
-      "epoch": 2.3068601028346425,
-      "grad_norm": 0.0006058014114387333,
-      "learning_rate": 0.00019999737564185138,
-      "loss": 46.0,
-      "step": 30172
-    },
-    {
-      "epoch": 2.3069365598180323,
-      "grad_norm": 0.001385160954669118,
-      "learning_rate": 0.00019999737546783176,
-      "loss": 46.0,
-      "step": 30173
-    },
-    {
-      "epoch": 2.307013016801422,
-      "grad_norm": 0.002277539111673832,
-      "learning_rate": 0.00019999737529380633,
-      "loss": 46.0,
-      "step": 30174
-    },
-    {
-      "epoch": 2.3070894737848118,
-      "grad_norm": 0.0012423406587913632,
-      "learning_rate": 0.00019999737511977516,
-      "loss": 46.0,
-      "step": 30175
-    },
-    {
-      "epoch": 2.3071659307682015,
-      "grad_norm": 0.00147822720464319,
-      "learning_rate": 0.0001999973749457382,
-      "loss": 46.0,
-      "step": 30176
-    },
-    {
-      "epoch": 2.3072423877515913,
-      "grad_norm": 0.0013297604164108634,
-      "learning_rate": 0.0001999973747716955,
-      "loss": 46.0,
-      "step": 30177
-    },
-    {
-      "epoch": 2.307318844734981,
-      "grad_norm": 0.0011833177413791418,
-      "learning_rate": 0.000199997374597647,
-      "loss": 46.0,
-      "step": 30178
-    },
-    {
-      "epoch": 2.307395301718371,
-      "grad_norm": 0.0004478892660699785,
-      "learning_rate": 0.00019999737442359273,
-      "loss": 46.0,
-      "step": 30179
-    },
-    {
-      "epoch": 2.3074717587017606,
-      "grad_norm": 0.0011697242734953761,
-      "learning_rate": 0.00019999737424953272,
-      "loss": 46.0,
-      "step": 30180
-    },
-    {
-      "epoch": 2.3075482156851503,
-      "grad_norm": 0.00387997692450881,
-      "learning_rate": 0.00019999737407546694,
-      "loss": 46.0,
-      "step": 30181
-    },
-    {
-      "epoch": 2.30762467266854,
-      "grad_norm": 0.0017362211365252733,
-      "learning_rate": 0.00019999737390139538,
-      "loss": 46.0,
-      "step": 30182
-    },
-    {
-      "epoch": 2.3077011296519294,
-      "grad_norm": 0.001383223687298596,
-      "learning_rate": 0.00019999737372731804,
-      "loss": 46.0,
-      "step": 30183
-    },
-    {
-      "epoch": 2.307777586635319,
-      "grad_norm": 0.001065898803062737,
-      "learning_rate": 0.00019999737355323494,
-      "loss": 46.0,
-      "step": 30184
-    },
-    {
-      "epoch": 2.307854043618709,
-      "grad_norm": 0.002347080735489726,
-      "learning_rate": 0.00019999737337914606,
-      "loss": 46.0,
-      "step": 30185
-    },
-    {
-      "epoch": 2.3079305006020987,
-      "grad_norm": 0.0007108405698090792,
-      "learning_rate": 0.00019999737320505143,
-      "loss": 46.0,
-      "step": 30186
-    },
-    {
-      "epoch": 2.3080069575854885,
-      "grad_norm": 0.0005572242080233991,
-      "learning_rate": 0.000199997373030951,
-      "loss": 46.0,
-      "step": 30187
-    },
-    {
-      "epoch": 2.3080834145688782,
-      "grad_norm": 0.0014329697005450726,
-      "learning_rate": 0.00019999737285684484,
-      "loss": 46.0,
-      "step": 30188
-    },
-    {
-      "epoch": 2.308159871552268,
-      "grad_norm": 0.008112660609185696,
-      "learning_rate": 0.0001999973726827329,
-      "loss": 46.0,
-      "step": 30189
-    },
-    {
-      "epoch": 2.3082363285356577,
-      "grad_norm": 0.0018196408636868,
-      "learning_rate": 0.00019999737250861517,
-      "loss": 46.0,
-      "step": 30190
-    },
-    {
-      "epoch": 2.3083127855190475,
-      "grad_norm": 0.002917099976912141,
-      "learning_rate": 0.00019999737233449168,
-      "loss": 46.0,
-      "step": 30191
-    },
-    {
-      "epoch": 2.308389242502437,
-      "grad_norm": 0.00043530986295081675,
-      "learning_rate": 0.00019999737216036244,
-      "loss": 46.0,
-      "step": 30192
-    },
-    {
-      "epoch": 2.3084656994858266,
-      "grad_norm": 0.000667903630528599,
-      "learning_rate": 0.0001999973719862274,
-      "loss": 46.0,
-      "step": 30193
-    },
-    {
-      "epoch": 2.3085421564692163,
-      "grad_norm": 0.0007341289892792702,
-      "learning_rate": 0.00019999737181208662,
-      "loss": 46.0,
-      "step": 30194
-    },
-    {
-      "epoch": 2.308618613452606,
-      "grad_norm": 0.004593354184180498,
-      "learning_rate": 0.00019999737163794006,
-      "loss": 46.0,
-      "step": 30195
-    },
-    {
-      "epoch": 2.308695070435996,
-      "grad_norm": 0.004649362992495298,
-      "learning_rate": 0.00019999737146378772,
-      "loss": 46.0,
-      "step": 30196
-    },
-    {
-      "epoch": 2.3087715274193856,
-      "grad_norm": 0.0009565045475028455,
-      "learning_rate": 0.00019999737128962962,
-      "loss": 46.0,
-      "step": 30197
-    },
-    {
-      "epoch": 2.3088479844027754,
-      "grad_norm": 0.00112698704469949,
-      "learning_rate": 0.0001999973711154657,
-      "loss": 46.0,
-      "step": 30198
-    },
-    {
-      "epoch": 2.308924441386165,
-      "grad_norm": 0.0011483555426821113,
-      "learning_rate": 0.00019999737094129612,
-      "loss": 46.0,
-      "step": 30199
-    },
-    {
-      "epoch": 2.309000898369555,
-      "grad_norm": 0.0005525632295757532,
-      "learning_rate": 0.0001999973707671207,
-      "loss": 46.0,
-      "step": 30200
-    },
-    {
-      "epoch": 2.3090773553529447,
-      "grad_norm": 0.0005164967733435333,
-      "learning_rate": 0.00019999737059293952,
-      "loss": 46.0,
-      "step": 30201
-    },
-    {
-      "epoch": 2.3091538123363344,
-      "grad_norm": 0.0015221787616610527,
-      "learning_rate": 0.00019999737041875255,
-      "loss": 46.0,
-      "step": 30202
-    },
-    {
-      "epoch": 2.309230269319724,
-      "grad_norm": 0.0009435153915546834,
-      "learning_rate": 0.00019999737024455985,
-      "loss": 46.0,
-      "step": 30203
-    },
-    {
-      "epoch": 2.309306726303114,
-      "grad_norm": 0.0009867754997685552,
-      "learning_rate": 0.00019999737007036136,
-      "loss": 46.0,
-      "step": 30204
-    },
-    {
-      "epoch": 2.3093831832865033,
-      "grad_norm": 0.004971676971763372,
-      "learning_rate": 0.00019999736989615712,
-      "loss": 46.0,
-      "step": 30205
-    },
-    {
-      "epoch": 2.309459640269893,
-      "grad_norm": 0.0011034803465008736,
-      "learning_rate": 0.00019999736972194706,
-      "loss": 46.0,
-      "step": 30206
-    },
-    {
-      "epoch": 2.309536097253283,
-      "grad_norm": 0.0004129599255975336,
-      "learning_rate": 0.0001999973695477313,
-      "loss": 46.0,
-      "step": 30207
-    },
-    {
-      "epoch": 2.3096125542366726,
-      "grad_norm": 0.001159068662673235,
-      "learning_rate": 0.00019999736937350971,
-      "loss": 46.0,
-      "step": 30208
-    },
-    {
-      "epoch": 2.3096890112200623,
-      "grad_norm": 0.0007293835515156388,
-      "learning_rate": 0.00019999736919928238,
-      "loss": 46.0,
-      "step": 30209
-    },
-    {
-      "epoch": 2.309765468203452,
-      "grad_norm": 0.001638144487515092,
-      "learning_rate": 0.00019999736902504928,
-      "loss": 46.0,
-      "step": 30210
-    },
-    {
-      "epoch": 2.309841925186842,
-      "grad_norm": 0.0009650388965383172,
-      "learning_rate": 0.00019999736885081043,
-      "loss": 46.0,
-      "step": 30211
-    },
-    {
-      "epoch": 2.3099183821702316,
-      "grad_norm": 0.0005872561596333981,
-      "learning_rate": 0.00019999736867656575,
-      "loss": 46.0,
-      "step": 30212
-    },
-    {
-      "epoch": 2.3099948391536214,
-      "grad_norm": 0.0011244539637118578,
-      "learning_rate": 0.00019999736850231538,
-      "loss": 46.0,
-      "step": 30213
-    },
-    {
-      "epoch": 2.3100712961370107,
-      "grad_norm": 0.0007432249840348959,
-      "learning_rate": 0.00019999736832805918,
-      "loss": 46.0,
-      "step": 30214
-    },
-    {
-      "epoch": 2.3101477531204004,
-      "grad_norm": 0.0015721962554380298,
-      "learning_rate": 0.0001999973681537972,
-      "loss": 46.0,
-      "step": 30215
-    },
-    {
-      "epoch": 2.31022421010379,
-      "grad_norm": 0.0021371503826230764,
-      "learning_rate": 0.00019999736797952951,
-      "loss": 46.0,
-      "step": 30216
-    },
-    {
-      "epoch": 2.31030066708718,
-      "grad_norm": 0.0011967411264777184,
-      "learning_rate": 0.00019999736780525602,
-      "loss": 46.0,
-      "step": 30217
-    },
-    {
-      "epoch": 2.3103771240705697,
-      "grad_norm": 0.0012975335121154785,
-      "learning_rate": 0.00019999736763097676,
-      "loss": 46.0,
-      "step": 30218
-    },
-    {
-      "epoch": 2.3104535810539595,
-      "grad_norm": 0.0028240846004337072,
-      "learning_rate": 0.00019999736745669172,
-      "loss": 46.0,
-      "step": 30219
-    },
-    {
-      "epoch": 2.3105300380373492,
-      "grad_norm": 0.003750396892428398,
-      "learning_rate": 0.00019999736728240093,
-      "loss": 46.0,
-      "step": 30220
-    },
-    {
-      "epoch": 2.310606495020739,
-      "grad_norm": 0.0035218799021095037,
-      "learning_rate": 0.00019999736710810438,
-      "loss": 46.0,
-      "step": 30221
-    },
-    {
-      "epoch": 2.3106829520041288,
-      "grad_norm": 0.0011214811820536852,
-      "learning_rate": 0.00019999736693380204,
-      "loss": 46.0,
-      "step": 30222
-    },
-    {
-      "epoch": 2.3107594089875185,
-      "grad_norm": 0.0009723555995151401,
-      "learning_rate": 0.0001999973667594939,
-      "loss": 46.0,
-      "step": 30223
-    },
-    {
-      "epoch": 2.3108358659709083,
-      "grad_norm": 0.0012814803048968315,
-      "learning_rate": 0.00019999736658518006,
-      "loss": 46.0,
-      "step": 30224
-    },
-    {
-      "epoch": 2.310912322954298,
-      "grad_norm": 0.0007300298311747611,
-      "learning_rate": 0.0001999973664108604,
-      "loss": 46.0,
-      "step": 30225
-    },
-    {
-      "epoch": 2.3109887799376874,
-      "grad_norm": 0.0017820047214627266,
-      "learning_rate": 0.000199997366236535,
-      "loss": 46.0,
-      "step": 30226
-    },
-    {
-      "epoch": 2.311065236921077,
-      "grad_norm": 0.0025835237465798855,
-      "learning_rate": 0.00019999736606220384,
-      "loss": 46.0,
-      "step": 30227
-    },
-    {
-      "epoch": 2.311141693904467,
-      "grad_norm": 0.0013711490901187062,
-      "learning_rate": 0.0001999973658878669,
-      "loss": 46.0,
-      "step": 30228
-    },
-    {
-      "epoch": 2.3112181508878566,
-      "grad_norm": 0.0012380421394482255,
-      "learning_rate": 0.00019999736571352418,
-      "loss": 46.0,
-      "step": 30229
-    },
-    {
-      "epoch": 2.3112946078712464,
-      "grad_norm": 0.0014767007669433951,
-      "learning_rate": 0.0001999973655391757,
-      "loss": 46.0,
-      "step": 30230
-    },
-    {
-      "epoch": 2.311371064854636,
-      "grad_norm": 0.00156877760309726,
-      "learning_rate": 0.00019999736536482142,
-      "loss": 46.0,
-      "step": 30231
-    },
-    {
-      "epoch": 2.311447521838026,
-      "grad_norm": 0.00189289974514395,
-      "learning_rate": 0.00019999736519046138,
-      "loss": 46.0,
-      "step": 30232
-    },
-    {
-      "epoch": 2.3115239788214157,
-      "grad_norm": 0.0013805158669129014,
-      "learning_rate": 0.00019999736501609557,
-      "loss": 46.0,
-      "step": 30233
-    },
-    {
-      "epoch": 2.3116004358048055,
-      "grad_norm": 0.0008209001389332116,
-      "learning_rate": 0.00019999736484172401,
-      "loss": 46.0,
-      "step": 30234
-    },
-    {
-      "epoch": 2.311676892788195,
-      "grad_norm": 0.002141595119610429,
-      "learning_rate": 0.00019999736466734666,
-      "loss": 46.0,
-      "step": 30235
-    },
-    {
-      "epoch": 2.3117533497715845,
-      "grad_norm": 0.0018898362759500742,
-      "learning_rate": 0.00019999736449296358,
-      "loss": 46.0,
-      "step": 30236
-    },
-    {
-      "epoch": 2.3118298067549743,
-      "grad_norm": 0.0008827374549582601,
-      "learning_rate": 0.0001999973643185747,
-      "loss": 46.0,
-      "step": 30237
-    },
-    {
-      "epoch": 2.311906263738364,
-      "grad_norm": 0.0011403128737583756,
-      "learning_rate": 0.00019999736414418005,
-      "loss": 46.0,
-      "step": 30238
-    },
-    {
-      "epoch": 2.311982720721754,
-      "grad_norm": 0.0005950057529844344,
-      "learning_rate": 0.00019999736396977965,
-      "loss": 46.0,
-      "step": 30239
-    },
-    {
-      "epoch": 2.3120591777051436,
-      "grad_norm": 0.0008201288874261081,
-      "learning_rate": 0.00019999736379537345,
-      "loss": 46.0,
-      "step": 30240
-    },
-    {
-      "epoch": 2.3121356346885333,
-      "grad_norm": 0.0006285528652369976,
-      "learning_rate": 0.0001999973636209615,
-      "loss": 46.0,
-      "step": 30241
-    },
-    {
-      "epoch": 2.312212091671923,
-      "grad_norm": 0.0007488488918170333,
-      "learning_rate": 0.0001999973634465438,
-      "loss": 46.0,
-      "step": 30242
-    },
-    {
-      "epoch": 2.312288548655313,
-      "grad_norm": 0.001049170969054103,
-      "learning_rate": 0.00019999736327212027,
-      "loss": 46.0,
-      "step": 30243
-    },
-    {
-      "epoch": 2.3123650056387026,
-      "grad_norm": 0.001010696403682232,
-      "learning_rate": 0.00019999736309769104,
-      "loss": 46.0,
-      "step": 30244
-    },
-    {
-      "epoch": 2.3124414626220924,
-      "grad_norm": 0.0008575752144679427,
-      "learning_rate": 0.00019999736292325603,
-      "loss": 46.0,
-      "step": 30245
-    },
-    {
-      "epoch": 2.312517919605482,
-      "grad_norm": 0.001815955387428403,
-      "learning_rate": 0.0001999973627488152,
-      "loss": 46.0,
-      "step": 30246
-    },
-    {
-      "epoch": 2.312594376588872,
-      "grad_norm": 0.0006540436297655106,
-      "learning_rate": 0.00019999736257436866,
-      "loss": 46.0,
-      "step": 30247
-    },
-    {
-      "epoch": 2.312670833572261,
-      "grad_norm": 0.0007758033461868763,
-      "learning_rate": 0.0001999973623999163,
-      "loss": 46.0,
-      "step": 30248
-    },
-    {
-      "epoch": 2.312747290555651,
-      "grad_norm": 0.001631412305869162,
-      "learning_rate": 0.0001999973622254582,
-      "loss": 46.0,
-      "step": 30249
-    },
-    {
-      "epoch": 2.3128237475390407,
-      "grad_norm": 0.001589313498698175,
-      "learning_rate": 0.00019999736205099432,
-      "loss": 46.0,
-      "step": 30250
-    },
-    {
-      "epoch": 2.3129002045224305,
-      "grad_norm": 0.0008404984255321324,
-      "learning_rate": 0.0001999973618765247,
-      "loss": 46.0,
-      "step": 30251
-    },
-    {
-      "epoch": 2.3129766615058203,
-      "grad_norm": 0.0006540753529407084,
-      "learning_rate": 0.00019999736170204927,
-      "loss": 46.0,
-      "step": 30252
-    },
-    {
-      "epoch": 2.31305311848921,
-      "grad_norm": 0.006734836380928755,
-      "learning_rate": 0.0001999973615275681,
-      "loss": 46.0,
-      "step": 30253
-    },
-    {
-      "epoch": 2.3131295754726,
-      "grad_norm": 0.0005335891037248075,
-      "learning_rate": 0.00019999736135308113,
-      "loss": 46.0,
-      "step": 30254
-    },
-    {
-      "epoch": 2.3132060324559895,
-      "grad_norm": 0.008970017544925213,
-      "learning_rate": 0.0001999973611785884,
-      "loss": 46.0,
-      "step": 30255
-    },
-    {
-      "epoch": 2.3132824894393793,
-      "grad_norm": 0.0017543798312544823,
-      "learning_rate": 0.00019999736100408992,
-      "loss": 46.0,
-      "step": 30256
-    },
-    {
-      "epoch": 2.313358946422769,
-      "grad_norm": 0.0038674629759043455,
-      "learning_rate": 0.00019999736082958566,
-      "loss": 46.0,
-      "step": 30257
-    },
-    {
-      "epoch": 2.3134354034061584,
-      "grad_norm": 0.0016310069477185607,
-      "learning_rate": 0.00019999736065507564,
-      "loss": 46.0,
-      "step": 30258
-    },
-    {
-      "epoch": 2.313511860389548,
-      "grad_norm": 0.0008364637033082545,
-      "learning_rate": 0.00019999736048055983,
-      "loss": 46.0,
-      "step": 30259
-    },
-    {
-      "epoch": 2.313588317372938,
-      "grad_norm": 0.0006872367230243981,
-      "learning_rate": 0.00019999736030603825,
-      "loss": 46.0,
-      "step": 30260
-    },
-    {
-      "epoch": 2.3136647743563277,
-      "grad_norm": 0.0005224893102422357,
-      "learning_rate": 0.00019999736013151092,
-      "loss": 46.0,
-      "step": 30261
-    },
-    {
-      "epoch": 2.3137412313397174,
-      "grad_norm": 0.001109803793951869,
-      "learning_rate": 0.00019999735995697782,
-      "loss": 46.0,
-      "step": 30262
-    },
-    {
-      "epoch": 2.313817688323107,
-      "grad_norm": 0.000571669836062938,
-      "learning_rate": 0.00019999735978243894,
-      "loss": 46.0,
-      "step": 30263
-    },
-    {
-      "epoch": 2.313894145306497,
-      "grad_norm": 0.0007968081044964492,
-      "learning_rate": 0.0001999973596078943,
-      "loss": 46.0,
-      "step": 30264
-    },
-    {
-      "epoch": 2.3139706022898867,
-      "grad_norm": 0.002043564571067691,
-      "learning_rate": 0.0001999973594333439,
-      "loss": 46.0,
-      "step": 30265
-    },
-    {
-      "epoch": 2.3140470592732765,
-      "grad_norm": 0.0011303542414680123,
-      "learning_rate": 0.0001999973592587877,
-      "loss": 46.0,
-      "step": 30266
-    },
-    {
-      "epoch": 2.3141235162566662,
-      "grad_norm": 0.0017548424657434225,
-      "learning_rate": 0.00019999735908422575,
-      "loss": 46.0,
-      "step": 30267
-    },
-    {
-      "epoch": 2.314199973240056,
-      "grad_norm": 0.002858856227248907,
-      "learning_rate": 0.00019999735890965803,
-      "loss": 46.0,
-      "step": 30268
-    },
-    {
-      "epoch": 2.3142764302234458,
-      "grad_norm": 0.0018452861113473773,
-      "learning_rate": 0.00019999735873508454,
-      "loss": 46.0,
-      "step": 30269
-    },
-    {
-      "epoch": 2.314352887206835,
-      "grad_norm": 0.0005100212874822319,
-      "learning_rate": 0.00019999735856050528,
-      "loss": 46.0,
-      "step": 30270
-    },
-    {
-      "epoch": 2.314429344190225,
-      "grad_norm": 0.00043524097418412566,
-      "learning_rate": 0.00019999735838592024,
-      "loss": 46.0,
-      "step": 30271
-    },
-    {
-      "epoch": 2.3145058011736146,
-      "grad_norm": 0.003048122860491276,
-      "learning_rate": 0.00019999735821132943,
-      "loss": 46.0,
-      "step": 30272
-    },
-    {
-      "epoch": 2.3145822581570044,
-      "grad_norm": 0.0009016020921990275,
-      "learning_rate": 0.00019999735803673287,
-      "loss": 46.0,
-      "step": 30273
-    },
-    {
-      "epoch": 2.314658715140394,
-      "grad_norm": 0.0008780946373008192,
-      "learning_rate": 0.00019999735786213054,
-      "loss": 46.0,
-      "step": 30274
-    },
-    {
-      "epoch": 2.314735172123784,
-      "grad_norm": 0.0017649485962465405,
-      "learning_rate": 0.00019999735768752244,
-      "loss": 46.0,
-      "step": 30275
-    },
-    {
-      "epoch": 2.3148116291071736,
-      "grad_norm": 0.0006517200963571668,
-      "learning_rate": 0.00019999735751290856,
-      "loss": 46.0,
-      "step": 30276
-    },
-    {
-      "epoch": 2.3148880860905634,
-      "grad_norm": 0.0012541132746264338,
-      "learning_rate": 0.0001999973573382889,
-      "loss": 46.0,
-      "step": 30277
-    },
-    {
-      "epoch": 2.314964543073953,
-      "grad_norm": 0.0021966828498989344,
-      "learning_rate": 0.0001999973571636635,
-      "loss": 46.0,
-      "step": 30278
-    },
-    {
-      "epoch": 2.315041000057343,
-      "grad_norm": 0.0016574910841882229,
-      "learning_rate": 0.0001999973569890323,
-      "loss": 46.0,
-      "step": 30279
-    },
-    {
-      "epoch": 2.3151174570407322,
-      "grad_norm": 0.0009617339237593114,
-      "learning_rate": 0.00019999735681439535,
-      "loss": 46.0,
-      "step": 30280
-    },
-    {
-      "epoch": 2.315193914024122,
-      "grad_norm": 0.0018893566448241472,
-      "learning_rate": 0.00019999735663975263,
-      "loss": 46.0,
-      "step": 30281
-    },
-    {
-      "epoch": 2.3152703710075118,
-      "grad_norm": 0.0006024938193149865,
-      "learning_rate": 0.00019999735646510414,
-      "loss": 46.0,
-      "step": 30282
-    },
-    {
-      "epoch": 2.3153468279909015,
-      "grad_norm": 0.002040133811533451,
-      "learning_rate": 0.00019999735629044988,
-      "loss": 46.0,
-      "step": 30283
-    },
-    {
-      "epoch": 2.3154232849742913,
-      "grad_norm": 0.0017384991515427828,
-      "learning_rate": 0.00019999735611578981,
-      "loss": 46.0,
-      "step": 30284
-    },
-    {
-      "epoch": 2.315499741957681,
-      "grad_norm": 0.0019039979670196772,
-      "learning_rate": 0.00019999735594112403,
-      "loss": 46.0,
-      "step": 30285
-    },
-    {
-      "epoch": 2.315576198941071,
-      "grad_norm": 0.001144712558016181,
-      "learning_rate": 0.00019999735576645247,
-      "loss": 46.0,
-      "step": 30286
-    },
-    {
-      "epoch": 2.3156526559244606,
-      "grad_norm": 0.0006676096236333251,
-      "learning_rate": 0.00019999735559177512,
-      "loss": 46.0,
-      "step": 30287
-    },
-    {
-      "epoch": 2.3157291129078503,
-      "grad_norm": 0.000968686887063086,
-      "learning_rate": 0.00019999735541709201,
-      "loss": 46.0,
-      "step": 30288
-    },
-    {
-      "epoch": 2.31580556989124,
-      "grad_norm": 0.002427425468340516,
-      "learning_rate": 0.00019999735524240314,
-      "loss": 46.0,
-      "step": 30289
-    },
-    {
-      "epoch": 2.31588202687463,
-      "grad_norm": 0.0005142023437656462,
-      "learning_rate": 0.0001999973550677085,
-      "loss": 46.0,
-      "step": 30290
-    },
-    {
-      "epoch": 2.3159584838580196,
-      "grad_norm": 0.000832767691463232,
-      "learning_rate": 0.0001999973548930081,
-      "loss": 46.0,
-      "step": 30291
-    },
-    {
-      "epoch": 2.316034940841409,
-      "grad_norm": 0.0015518913278356194,
-      "learning_rate": 0.0001999973547183019,
-      "loss": 46.0,
-      "step": 30292
-    },
-    {
-      "epoch": 2.3161113978247987,
-      "grad_norm": 0.0012677236227318645,
-      "learning_rate": 0.00019999735454358995,
-      "loss": 46.0,
-      "step": 30293
-    },
-    {
-      "epoch": 2.3161878548081885,
-      "grad_norm": 0.003694751998409629,
-      "learning_rate": 0.0001999973543688722,
-      "loss": 46.0,
-      "step": 30294
-    },
-    {
-      "epoch": 2.316264311791578,
-      "grad_norm": 0.0014075635699555278,
-      "learning_rate": 0.00019999735419414872,
-      "loss": 46.0,
-      "step": 30295
-    },
-    {
-      "epoch": 2.316340768774968,
-      "grad_norm": 0.0017439830116927624,
-      "learning_rate": 0.00019999735401941945,
-      "loss": 46.0,
-      "step": 30296
-    },
-    {
-      "epoch": 2.3164172257583577,
-      "grad_norm": 0.0006086484063416719,
-      "learning_rate": 0.00019999735384468445,
-      "loss": 46.0,
-      "step": 30297
-    },
-    {
-      "epoch": 2.3164936827417475,
-      "grad_norm": 0.0037569128908216953,
-      "learning_rate": 0.00019999735366994364,
-      "loss": 46.0,
-      "step": 30298
-    },
-    {
-      "epoch": 2.3165701397251373,
-      "grad_norm": 0.0025564441457390785,
-      "learning_rate": 0.00019999735349519705,
-      "loss": 46.0,
-      "step": 30299
-    },
-    {
-      "epoch": 2.316646596708527,
-      "grad_norm": 0.0008942708373069763,
-      "learning_rate": 0.00019999735332044472,
-      "loss": 46.0,
-      "step": 30300
-    },
-    {
-      "epoch": 2.316723053691917,
-      "grad_norm": 0.0012030428042635322,
-      "learning_rate": 0.0001999973531456866,
-      "loss": 46.0,
-      "step": 30301
-    },
-    {
-      "epoch": 2.316799510675306,
-      "grad_norm": 0.002030459465458989,
-      "learning_rate": 0.00019999735297092272,
-      "loss": 46.0,
-      "step": 30302
-    },
-    {
-      "epoch": 2.316875967658696,
-      "grad_norm": 0.0016904083313420415,
-      "learning_rate": 0.0001999973527961531,
-      "loss": 46.0,
-      "step": 30303
-    },
-    {
-      "epoch": 2.3169524246420856,
-      "grad_norm": 0.004592102952301502,
-      "learning_rate": 0.00019999735262137767,
-      "loss": 46.0,
-      "step": 30304
-    },
-    {
-      "epoch": 2.3170288816254754,
-      "grad_norm": 0.0010771737433969975,
-      "learning_rate": 0.00019999735244659647,
-      "loss": 46.0,
-      "step": 30305
-    },
-    {
-      "epoch": 2.317105338608865,
-      "grad_norm": 0.0024316797498613596,
-      "learning_rate": 0.0001999973522718095,
-      "loss": 46.0,
-      "step": 30306
-    },
-    {
-      "epoch": 2.317181795592255,
-      "grad_norm": 0.0016414186684414744,
-      "learning_rate": 0.0001999973520970168,
-      "loss": 46.0,
-      "step": 30307
-    },
-    {
-      "epoch": 2.3172582525756447,
-      "grad_norm": 0.0006019505672156811,
-      "learning_rate": 0.0001999973519222183,
-      "loss": 46.0,
-      "step": 30308
-    },
-    {
-      "epoch": 2.3173347095590344,
-      "grad_norm": 0.0006140986224636436,
-      "learning_rate": 0.00019999735174741404,
-      "loss": 46.0,
-      "step": 30309
-    },
-    {
-      "epoch": 2.317411166542424,
-      "grad_norm": 0.0009695172775536776,
-      "learning_rate": 0.000199997351572604,
-      "loss": 46.0,
-      "step": 30310
-    },
-    {
-      "epoch": 2.317487623525814,
-      "grad_norm": 0.0035742600448429585,
-      "learning_rate": 0.0001999973513977882,
-      "loss": 46.0,
-      "step": 30311
-    },
-    {
-      "epoch": 2.3175640805092037,
-      "grad_norm": 0.004137310199439526,
-      "learning_rate": 0.0001999973512229666,
-      "loss": 46.0,
-      "step": 30312
-    },
-    {
-      "epoch": 2.3176405374925935,
-      "grad_norm": 0.0005126777687110007,
-      "learning_rate": 0.00019999735104813928,
-      "loss": 46.0,
-      "step": 30313
-    },
-    {
-      "epoch": 2.317716994475983,
-      "grad_norm": 0.003727610455825925,
-      "learning_rate": 0.00019999735087330617,
-      "loss": 46.0,
-      "step": 30314
-    },
-    {
-      "epoch": 2.3177934514593725,
-      "grad_norm": 0.0030770886223763227,
-      "learning_rate": 0.00019999735069846727,
-      "loss": 46.0,
-      "step": 30315
-    },
-    {
-      "epoch": 2.3178699084427623,
-      "grad_norm": 0.0015235097380355,
-      "learning_rate": 0.00019999735052362265,
-      "loss": 46.0,
-      "step": 30316
-    },
-    {
-      "epoch": 2.317946365426152,
-      "grad_norm": 0.0004589337622746825,
-      "learning_rate": 0.00019999735034877223,
-      "loss": 46.0,
-      "step": 30317
-    },
-    {
-      "epoch": 2.318022822409542,
-      "grad_norm": 0.0015530851669609547,
-      "learning_rate": 0.00019999735017391603,
-      "loss": 46.0,
-      "step": 30318
-    },
-    {
-      "epoch": 2.3180992793929316,
-      "grad_norm": 0.0016325052129104733,
-      "learning_rate": 0.0001999973499990541,
-      "loss": 46.0,
-      "step": 30319
-    },
-    {
-      "epoch": 2.3181757363763213,
-      "grad_norm": 0.0013463397044688463,
-      "learning_rate": 0.00019999734982418637,
-      "loss": 46.0,
-      "step": 30320
-    },
-    {
-      "epoch": 2.318252193359711,
-      "grad_norm": 0.0008545163436792791,
-      "learning_rate": 0.00019999734964931288,
-      "loss": 46.0,
-      "step": 30321
-    },
-    {
-      "epoch": 2.318328650343101,
-      "grad_norm": 0.0005678696907125413,
-      "learning_rate": 0.0001999973494744336,
-      "loss": 46.0,
-      "step": 30322
-    },
-    {
-      "epoch": 2.31840510732649,
-      "grad_norm": 0.0007805890636518598,
-      "learning_rate": 0.00019999734929954856,
-      "loss": 46.0,
-      "step": 30323
-    },
-    {
-      "epoch": 2.31848156430988,
-      "grad_norm": 0.0005209120572544634,
-      "learning_rate": 0.00019999734912465775,
-      "loss": 46.0,
-      "step": 30324
-    },
-    {
-      "epoch": 2.3185580212932697,
-      "grad_norm": 0.0008189770742319524,
-      "learning_rate": 0.0001999973489497612,
-      "loss": 46.0,
-      "step": 30325
-    },
-    {
-      "epoch": 2.3186344782766595,
-      "grad_norm": 0.0010674988152459264,
-      "learning_rate": 0.00019999734877485886,
-      "loss": 46.0,
-      "step": 30326
-    },
-    {
-      "epoch": 2.3187109352600492,
-      "grad_norm": 0.0014612135710194707,
-      "learning_rate": 0.0001999973485999507,
-      "loss": 46.0,
-      "step": 30327
-    },
-    {
-      "epoch": 2.318787392243439,
-      "grad_norm": 0.001121841254644096,
-      "learning_rate": 0.00019999734842503686,
-      "loss": 46.0,
-      "step": 30328
-    },
-    {
-      "epoch": 2.3188638492268288,
-      "grad_norm": 0.001231951406225562,
-      "learning_rate": 0.0001999973482501172,
-      "loss": 46.0,
-      "step": 30329
-    },
-    {
-      "epoch": 2.3189403062102185,
-      "grad_norm": 0.006180086638778448,
-      "learning_rate": 0.00019999734807519176,
-      "loss": 46.0,
-      "step": 30330
-    },
-    {
-      "epoch": 2.3190167631936083,
-      "grad_norm": 0.001452418277040124,
-      "learning_rate": 0.0001999973479002606,
-      "loss": 46.0,
-      "step": 30331
-    },
-    {
-      "epoch": 2.319093220176998,
-      "grad_norm": 0.0003985106013715267,
-      "learning_rate": 0.00019999734772532365,
-      "loss": 46.0,
-      "step": 30332
-    },
-    {
-      "epoch": 2.319169677160388,
-      "grad_norm": 0.0009089506347663701,
-      "learning_rate": 0.0001999973475503809,
-      "loss": 46.0,
-      "step": 30333
-    },
-    {
-      "epoch": 2.3192461341437776,
-      "grad_norm": 0.0010232562199234962,
-      "learning_rate": 0.00019999734737543242,
-      "loss": 46.0,
-      "step": 30334
-    },
-    {
-      "epoch": 2.3193225911271673,
-      "grad_norm": 0.0010310267098248005,
-      "learning_rate": 0.00019999734720047813,
-      "loss": 46.0,
-      "step": 30335
-    },
-    {
-      "epoch": 2.3193990481105566,
-      "grad_norm": 0.0007289510103873909,
-      "learning_rate": 0.00019999734702551812,
-      "loss": 46.0,
-      "step": 30336
-    },
-    {
-      "epoch": 2.3194755050939464,
-      "grad_norm": 0.000700947770383209,
-      "learning_rate": 0.0001999973468505523,
-      "loss": 46.0,
-      "step": 30337
-    },
-    {
-      "epoch": 2.319551962077336,
-      "grad_norm": 0.0006614306475967169,
-      "learning_rate": 0.00019999734667558073,
-      "loss": 46.0,
-      "step": 30338
-    },
-    {
-      "epoch": 2.319628419060726,
-      "grad_norm": 0.0013760696165263653,
-      "learning_rate": 0.0001999973465006034,
-      "loss": 46.0,
-      "step": 30339
-    },
-    {
-      "epoch": 2.3197048760441157,
-      "grad_norm": 0.001355270855128765,
-      "learning_rate": 0.00019999734632562027,
-      "loss": 46.0,
-      "step": 30340
-    },
-    {
-      "epoch": 2.3197813330275054,
-      "grad_norm": 0.006319751963019371,
-      "learning_rate": 0.0001999973461506314,
-      "loss": 46.0,
-      "step": 30341
-    },
-    {
-      "epoch": 2.319857790010895,
-      "grad_norm": 0.0006171645945869386,
-      "learning_rate": 0.00019999734597563675,
-      "loss": 46.0,
-      "step": 30342
-    },
-    {
-      "epoch": 2.319934246994285,
-      "grad_norm": 0.00040689727757126093,
-      "learning_rate": 0.00019999734580063632,
-      "loss": 46.0,
-      "step": 30343
-    },
-    {
-      "epoch": 2.3200107039776747,
-      "grad_norm": 0.0015397823881357908,
-      "learning_rate": 0.00019999734562563016,
-      "loss": 46.0,
-      "step": 30344
-    },
-    {
-      "epoch": 2.320087160961064,
-      "grad_norm": 0.00041277610580436885,
-      "learning_rate": 0.0001999973454506182,
-      "loss": 46.0,
-      "step": 30345
-    },
-    {
-      "epoch": 2.320163617944454,
-      "grad_norm": 0.003013300010934472,
-      "learning_rate": 0.00019999734527560045,
-      "loss": 46.0,
-      "step": 30346
-    },
-    {
-      "epoch": 2.3202400749278436,
-      "grad_norm": 0.003307922510430217,
-      "learning_rate": 0.00019999734510057696,
-      "loss": 46.0,
-      "step": 30347
-    },
-    {
-      "epoch": 2.3203165319112333,
-      "grad_norm": 0.003326849080622196,
-      "learning_rate": 0.00019999734492554767,
-      "loss": 46.0,
-      "step": 30348
-    },
-    {
-      "epoch": 2.320392988894623,
-      "grad_norm": 0.002616100711748004,
-      "learning_rate": 0.00019999734475051266,
-      "loss": 46.0,
-      "step": 30349
-    },
-    {
-      "epoch": 2.320469445878013,
-      "grad_norm": 0.0008141118451021612,
-      "learning_rate": 0.00019999734457547182,
-      "loss": 46.0,
-      "step": 30350
-    },
-    {
-      "epoch": 2.3205459028614026,
-      "grad_norm": 0.003728854237124324,
-      "learning_rate": 0.00019999734440042527,
-      "loss": 46.0,
-      "step": 30351
-    },
-    {
-      "epoch": 2.3206223598447924,
-      "grad_norm": 0.0004859961336478591,
-      "learning_rate": 0.00019999734422537294,
-      "loss": 46.0,
-      "step": 30352
-    },
-    {
-      "epoch": 2.320698816828182,
-      "grad_norm": 0.0008103257860057056,
-      "learning_rate": 0.0001999973440503148,
-      "loss": 46.0,
-      "step": 30353
-    },
-    {
-      "epoch": 2.320775273811572,
-      "grad_norm": 0.0008832439198158681,
-      "learning_rate": 0.00019999734387525094,
-      "loss": 46.0,
-      "step": 30354
-    },
-    {
-      "epoch": 2.3208517307949617,
-      "grad_norm": 0.0009649523999541998,
-      "learning_rate": 0.0001999973437001813,
-      "loss": 46.0,
-      "step": 30355
-    },
-    {
-      "epoch": 2.3209281877783514,
-      "grad_norm": 0.0007761421729810536,
-      "learning_rate": 0.00019999734352510587,
-      "loss": 46.0,
-      "step": 30356
-    },
-    {
-      "epoch": 2.3210046447617407,
-      "grad_norm": 0.0006364434957504272,
-      "learning_rate": 0.00019999734335002467,
-      "loss": 46.0,
-      "step": 30357
-    },
-    {
-      "epoch": 2.3210811017451305,
-      "grad_norm": 0.0007939972565509379,
-      "learning_rate": 0.0001999973431749377,
-      "loss": 46.0,
-      "step": 30358
-    },
-    {
-      "epoch": 2.3211575587285203,
-      "grad_norm": 0.0008202563039958477,
-      "learning_rate": 0.000199997342999845,
-      "loss": 46.0,
-      "step": 30359
-    },
-    {
-      "epoch": 2.32123401571191,
-      "grad_norm": 0.0007119790534488857,
-      "learning_rate": 0.0001999973428247465,
-      "loss": 46.0,
-      "step": 30360
-    },
-    {
-      "epoch": 2.3213104726952998,
-      "grad_norm": 0.001329778111539781,
-      "learning_rate": 0.0001999973426496422,
-      "loss": 46.0,
-      "step": 30361
-    },
-    {
-      "epoch": 2.3213869296786895,
-      "grad_norm": 0.001179746468551457,
-      "learning_rate": 0.0001999973424745322,
-      "loss": 46.0,
-      "step": 30362
-    },
-    {
-      "epoch": 2.3214633866620793,
-      "grad_norm": 0.000469493301352486,
-      "learning_rate": 0.00019999734229941637,
-      "loss": 46.0,
-      "step": 30363
-    },
-    {
-      "epoch": 2.321539843645469,
-      "grad_norm": 0.000981647172011435,
-      "learning_rate": 0.00019999734212429479,
-      "loss": 46.0,
-      "step": 30364
-    },
-    {
-      "epoch": 2.321616300628859,
-      "grad_norm": 0.001795116811990738,
-      "learning_rate": 0.00019999734194916746,
-      "loss": 46.0,
-      "step": 30365
-    },
-    {
-      "epoch": 2.3216927576122486,
-      "grad_norm": 0.0005095938104204834,
-      "learning_rate": 0.00019999734177403433,
-      "loss": 46.0,
-      "step": 30366
-    },
-    {
-      "epoch": 2.321769214595638,
-      "grad_norm": 0.0017778949113562703,
-      "learning_rate": 0.00019999734159889545,
-      "loss": 46.0,
-      "step": 30367
-    },
-    {
-      "epoch": 2.3218456715790277,
-      "grad_norm": 0.0010609703604131937,
-      "learning_rate": 0.0001999973414237508,
-      "loss": 46.0,
-      "step": 30368
-    },
-    {
-      "epoch": 2.3219221285624174,
-      "grad_norm": 0.0007412207778543234,
-      "learning_rate": 0.00019999734124860038,
-      "loss": 46.0,
-      "step": 30369
-    },
-    {
-      "epoch": 2.321998585545807,
-      "grad_norm": 0.008994394913315773,
-      "learning_rate": 0.0001999973410734442,
-      "loss": 46.0,
-      "step": 30370
-    },
-    {
-      "epoch": 2.322075042529197,
-      "grad_norm": 0.0006412541260942817,
-      "learning_rate": 0.00019999734089828222,
-      "loss": 46.0,
-      "step": 30371
-    },
-    {
-      "epoch": 2.3221514995125867,
-      "grad_norm": 0.0005678889574483037,
-      "learning_rate": 0.00019999734072311448,
-      "loss": 46.0,
-      "step": 30372
-    },
-    {
-      "epoch": 2.3222279564959765,
-      "grad_norm": 0.0014511931221932173,
-      "learning_rate": 0.000199997340547941,
-      "loss": 46.0,
-      "step": 30373
-    },
-    {
-      "epoch": 2.3223044134793662,
-      "grad_norm": 0.0009684332180768251,
-      "learning_rate": 0.00019999734037276173,
-      "loss": 46.0,
-      "step": 30374
-    },
-    {
-      "epoch": 2.322380870462756,
-      "grad_norm": 0.005044231191277504,
-      "learning_rate": 0.00019999734019757667,
-      "loss": 46.0,
-      "step": 30375
-    },
-    {
-      "epoch": 2.3224573274461457,
-      "grad_norm": 0.0008116128738038242,
-      "learning_rate": 0.0001999973400223859,
-      "loss": 46.0,
-      "step": 30376
-    },
-    {
-      "epoch": 2.3225337844295355,
-      "grad_norm": 0.0010314596584066749,
-      "learning_rate": 0.0001999973398471893,
-      "loss": 46.0,
-      "step": 30377
-    },
-    {
-      "epoch": 2.3226102414129253,
-      "grad_norm": 0.00035514350747689605,
-      "learning_rate": 0.00019999733967198698,
-      "loss": 46.0,
-      "step": 30378
-    },
-    {
-      "epoch": 2.3226866983963146,
-      "grad_norm": 0.012497498653829098,
-      "learning_rate": 0.00019999733949677885,
-      "loss": 46.0,
-      "step": 30379
-    },
-    {
-      "epoch": 2.3227631553797043,
-      "grad_norm": 0.0017736826557666063,
-      "learning_rate": 0.00019999733932156498,
-      "loss": 46.0,
-      "step": 30380
-    },
-    {
-      "epoch": 2.322839612363094,
-      "grad_norm": 0.0007666827295906842,
-      "learning_rate": 0.0001999973391463453,
-      "loss": 46.0,
-      "step": 30381
-    },
-    {
-      "epoch": 2.322916069346484,
-      "grad_norm": 0.00315473391674459,
-      "learning_rate": 0.00019999733897111988,
-      "loss": 46.0,
-      "step": 30382
-    },
-    {
-      "epoch": 2.3229925263298736,
-      "grad_norm": 0.005299501121044159,
-      "learning_rate": 0.00019999733879588869,
-      "loss": 46.0,
-      "step": 30383
-    },
-    {
-      "epoch": 2.3230689833132634,
-      "grad_norm": 0.0009433921659365296,
-      "learning_rate": 0.00019999733862065172,
-      "loss": 46.0,
-      "step": 30384
-    },
-    {
-      "epoch": 2.323145440296653,
-      "grad_norm": 0.002504811156541109,
-      "learning_rate": 0.000199997338445409,
-      "loss": 46.0,
-      "step": 30385
-    },
-    {
-      "epoch": 2.323221897280043,
-      "grad_norm": 0.001130260294303298,
-      "learning_rate": 0.0001999973382701605,
-      "loss": 46.0,
-      "step": 30386
-    },
-    {
-      "epoch": 2.3232983542634327,
-      "grad_norm": 0.0006547889206558466,
-      "learning_rate": 0.00019999733809490623,
-      "loss": 46.0,
-      "step": 30387
-    },
-    {
-      "epoch": 2.3233748112468224,
-      "grad_norm": 0.0007459830376319587,
-      "learning_rate": 0.0001999973379196462,
-      "loss": 46.0,
-      "step": 30388
-    },
-    {
-      "epoch": 2.3234512682302118,
-      "grad_norm": 0.0011226703645661473,
-      "learning_rate": 0.00019999733774438039,
-      "loss": 46.0,
-      "step": 30389
-    },
-    {
-      "epoch": 2.3235277252136015,
-      "grad_norm": 0.0005145056638866663,
-      "learning_rate": 0.0001999973375691088,
-      "loss": 46.0,
-      "step": 30390
-    },
-    {
-      "epoch": 2.3236041821969913,
-      "grad_norm": 0.0028296527452766895,
-      "learning_rate": 0.00019999733739383145,
-      "loss": 46.0,
-      "step": 30391
-    },
-    {
-      "epoch": 2.323680639180381,
-      "grad_norm": 0.0005512085044756532,
-      "learning_rate": 0.00019999733721854835,
-      "loss": 46.0,
-      "step": 30392
-    },
-    {
-      "epoch": 2.323757096163771,
-      "grad_norm": 0.0010922551155090332,
-      "learning_rate": 0.00019999733704325948,
-      "loss": 46.0,
-      "step": 30393
-    },
-    {
-      "epoch": 2.3238335531471606,
-      "grad_norm": 0.0026275126729160547,
-      "learning_rate": 0.0001999973368679648,
-      "loss": 46.0,
-      "step": 30394
-    },
-    {
-      "epoch": 2.3239100101305503,
-      "grad_norm": 0.0015638680197298527,
-      "learning_rate": 0.00019999733669266438,
-      "loss": 46.0,
-      "step": 30395
-    },
-    {
-      "epoch": 2.32398646711394,
-      "grad_norm": 0.0012028798228129745,
-      "learning_rate": 0.00019999733651735819,
-      "loss": 46.0,
-      "step": 30396
-    },
-    {
-      "epoch": 2.32406292409733,
-      "grad_norm": 0.0008055296493694186,
-      "learning_rate": 0.00019999733634204622,
-      "loss": 46.0,
-      "step": 30397
-    },
-    {
-      "epoch": 2.3241393810807196,
-      "grad_norm": 0.0015386966988444328,
-      "learning_rate": 0.00019999733616672848,
-      "loss": 46.0,
-      "step": 30398
-    },
-    {
-      "epoch": 2.3242158380641094,
-      "grad_norm": 0.00240634148940444,
-      "learning_rate": 0.000199997335991405,
-      "loss": 46.0,
-      "step": 30399
-    },
-    {
-      "epoch": 2.324292295047499,
-      "grad_norm": 0.0007274376112036407,
-      "learning_rate": 0.00019999733581607573,
-      "loss": 46.0,
-      "step": 30400
-    },
-    {
-      "epoch": 2.3243687520308884,
-      "grad_norm": 0.0027278985362499952,
-      "learning_rate": 0.00019999733564074067,
-      "loss": 46.0,
-      "step": 30401
-    },
-    {
-      "epoch": 2.324445209014278,
-      "grad_norm": 0.00022958245244808495,
-      "learning_rate": 0.00019999733546539986,
-      "loss": 46.0,
-      "step": 30402
-    },
-    {
-      "epoch": 2.324521665997668,
-      "grad_norm": 0.0007272228831425309,
-      "learning_rate": 0.0001999973352900533,
-      "loss": 46.0,
-      "step": 30403
-    },
-    {
-      "epoch": 2.3245981229810577,
-      "grad_norm": 0.002035551704466343,
-      "learning_rate": 0.00019999733511470095,
-      "loss": 46.0,
-      "step": 30404
-    },
-    {
-      "epoch": 2.3246745799644475,
-      "grad_norm": 0.00048469213652424514,
-      "learning_rate": 0.00019999733493934282,
-      "loss": 46.0,
-      "step": 30405
-    },
-    {
-      "epoch": 2.3247510369478372,
-      "grad_norm": 0.0007801530882716179,
-      "learning_rate": 0.00019999733476397895,
-      "loss": 46.0,
-      "step": 30406
-    },
-    {
-      "epoch": 2.324827493931227,
-      "grad_norm": 0.0012340234825387597,
-      "learning_rate": 0.0001999973345886093,
-      "loss": 46.0,
-      "step": 30407
-    },
-    {
-      "epoch": 2.3249039509146168,
-      "grad_norm": 0.003101398004218936,
-      "learning_rate": 0.00019999733441323386,
-      "loss": 46.0,
-      "step": 30408
-    },
-    {
-      "epoch": 2.3249804078980065,
-      "grad_norm": 0.0011867587454617023,
-      "learning_rate": 0.00019999733423785266,
-      "loss": 46.0,
-      "step": 30409
-    },
-    {
-      "epoch": 2.3250568648813963,
-      "grad_norm": 0.0005943225114606321,
-      "learning_rate": 0.00019999733406246572,
-      "loss": 46.0,
-      "step": 30410
-    },
-    {
-      "epoch": 2.3251333218647856,
-      "grad_norm": 0.0003382488212082535,
-      "learning_rate": 0.00019999733388707298,
-      "loss": 46.0,
-      "step": 30411
-    },
-    {
-      "epoch": 2.3252097788481754,
-      "grad_norm": 0.001074868137948215,
-      "learning_rate": 0.00019999733371167447,
-      "loss": 46.0,
-      "step": 30412
-    },
-    {
-      "epoch": 2.325286235831565,
-      "grad_norm": 0.0004787916550412774,
-      "learning_rate": 0.00019999733353627018,
-      "loss": 46.0,
-      "step": 30413
-    },
-    {
-      "epoch": 2.325362692814955,
-      "grad_norm": 0.0014202031306922436,
-      "learning_rate": 0.00019999733336086015,
-      "loss": 46.0,
-      "step": 30414
-    },
-    {
-      "epoch": 2.3254391497983447,
-      "grad_norm": 0.0015058716526255012,
-      "learning_rate": 0.00019999733318544437,
-      "loss": 46.0,
-      "step": 30415
-    },
-    {
-      "epoch": 2.3255156067817344,
-      "grad_norm": 0.0023006699047982693,
-      "learning_rate": 0.00019999733301002276,
-      "loss": 46.0,
-      "step": 30416
-    },
-    {
-      "epoch": 2.325592063765124,
-      "grad_norm": 0.0005122700240463018,
-      "learning_rate": 0.00019999733283459543,
-      "loss": 46.0,
-      "step": 30417
-    },
-    {
-      "epoch": 2.325668520748514,
-      "grad_norm": 0.0009394968510605395,
-      "learning_rate": 0.0001999973326591623,
-      "loss": 46.0,
-      "step": 30418
-    },
-    {
-      "epoch": 2.3257449777319037,
-      "grad_norm": 0.0005133896484039724,
-      "learning_rate": 0.00019999733248372343,
-      "loss": 46.0,
-      "step": 30419
-    },
-    {
-      "epoch": 2.3258214347152935,
-      "grad_norm": 0.001428685849532485,
-      "learning_rate": 0.00019999733230827876,
-      "loss": 46.0,
-      "step": 30420
-    },
-    {
-      "epoch": 2.325897891698683,
-      "grad_norm": 0.0014269306557253003,
-      "learning_rate": 0.00019999733213282834,
-      "loss": 46.0,
-      "step": 30421
-    },
-    {
-      "epoch": 2.325974348682073,
-      "grad_norm": 0.001305045560002327,
-      "learning_rate": 0.00019999733195737217,
-      "loss": 46.0,
-      "step": 30422
-    },
-    {
-      "epoch": 2.3260508056654623,
-      "grad_norm": 0.0014530970947816968,
-      "learning_rate": 0.00019999733178191018,
-      "loss": 46.0,
-      "step": 30423
-    },
-    {
-      "epoch": 2.326127262648852,
-      "grad_norm": 0.0005421048845164478,
-      "learning_rate": 0.00019999733160644246,
-      "loss": 46.0,
-      "step": 30424
-    },
-    {
-      "epoch": 2.326203719632242,
-      "grad_norm": 0.0009188229450955987,
-      "learning_rate": 0.00019999733143096892,
-      "loss": 46.0,
-      "step": 30425
-    },
-    {
-      "epoch": 2.3262801766156316,
-      "grad_norm": 0.0007503820233978331,
-      "learning_rate": 0.00019999733125548966,
-      "loss": 46.0,
-      "step": 30426
-    },
-    {
-      "epoch": 2.3263566335990213,
-      "grad_norm": 0.0008856317726895213,
-      "learning_rate": 0.00019999733108000463,
-      "loss": 46.0,
-      "step": 30427
-    },
-    {
-      "epoch": 2.326433090582411,
-      "grad_norm": 0.002229111734777689,
-      "learning_rate": 0.00019999733090451382,
-      "loss": 46.0,
-      "step": 30428
-    },
-    {
-      "epoch": 2.326509547565801,
-      "grad_norm": 0.0005680381436832249,
-      "learning_rate": 0.00019999733072901724,
-      "loss": 46.0,
-      "step": 30429
-    },
-    {
-      "epoch": 2.3265860045491906,
-      "grad_norm": 0.0021651163697242737,
-      "learning_rate": 0.00019999733055351492,
-      "loss": 46.0,
-      "step": 30430
-    },
-    {
-      "epoch": 2.3266624615325804,
-      "grad_norm": 0.0011214156402274966,
-      "learning_rate": 0.00019999733037800676,
-      "loss": 46.0,
-      "step": 30431
-    },
-    {
-      "epoch": 2.32673891851597,
-      "grad_norm": 0.001414934522472322,
-      "learning_rate": 0.0001999973302024929,
-      "loss": 46.0,
-      "step": 30432
-    },
-    {
-      "epoch": 2.3268153754993595,
-      "grad_norm": 0.0006146717933006585,
-      "learning_rate": 0.00019999733002697324,
-      "loss": 46.0,
-      "step": 30433
-    },
-    {
-      "epoch": 2.3268918324827492,
-      "grad_norm": 0.003162570297718048,
-      "learning_rate": 0.0001999973298514478,
-      "loss": 46.0,
-      "step": 30434
-    },
-    {
-      "epoch": 2.326968289466139,
-      "grad_norm": 0.0012074160622432828,
-      "learning_rate": 0.0001999973296759166,
-      "loss": 46.0,
-      "step": 30435
-    },
-    {
-      "epoch": 2.3270447464495287,
-      "grad_norm": 0.001848422223702073,
-      "learning_rate": 0.00019999732950037963,
-      "loss": 46.0,
-      "step": 30436
-    },
-    {
-      "epoch": 2.3271212034329185,
-      "grad_norm": 0.0008912977646104991,
-      "learning_rate": 0.0001999973293248369,
-      "loss": 46.0,
-      "step": 30437
-    },
-    {
-      "epoch": 2.3271976604163083,
-      "grad_norm": 0.0007331767701543868,
-      "learning_rate": 0.00019999732914928838,
-      "loss": 46.0,
-      "step": 30438
-    },
-    {
-      "epoch": 2.327274117399698,
-      "grad_norm": 0.0008977054385468364,
-      "learning_rate": 0.00019999732897373412,
-      "loss": 46.0,
-      "step": 30439
-    },
-    {
-      "epoch": 2.327350574383088,
-      "grad_norm": 0.0007164858980104327,
-      "learning_rate": 0.0001999973287981741,
-      "loss": 46.0,
-      "step": 30440
-    },
-    {
-      "epoch": 2.3274270313664776,
-      "grad_norm": 0.0018716392805799842,
-      "learning_rate": 0.00019999732862260828,
-      "loss": 46.0,
-      "step": 30441
-    },
-    {
-      "epoch": 2.3275034883498673,
-      "grad_norm": 0.0040169828571379185,
-      "learning_rate": 0.0001999973284470367,
-      "loss": 46.0,
-      "step": 30442
-    },
-    {
-      "epoch": 2.327579945333257,
-      "grad_norm": 0.0009843767620623112,
-      "learning_rate": 0.00019999732827145935,
-      "loss": 46.0,
-      "step": 30443
-    },
-    {
-      "epoch": 2.327656402316647,
-      "grad_norm": 0.0008170006331056356,
-      "learning_rate": 0.00019999732809587622,
-      "loss": 46.0,
-      "step": 30444
-    },
-    {
-      "epoch": 2.327732859300036,
-      "grad_norm": 0.0007758899009786546,
-      "learning_rate": 0.00019999732792028735,
-      "loss": 46.0,
-      "step": 30445
-    },
-    {
-      "epoch": 2.327809316283426,
-      "grad_norm": 0.005486639216542244,
-      "learning_rate": 0.00019999732774469268,
-      "loss": 46.0,
-      "step": 30446
-    },
-    {
-      "epoch": 2.3278857732668157,
-      "grad_norm": 0.0006798380054533482,
-      "learning_rate": 0.00019999732756909223,
-      "loss": 46.0,
-      "step": 30447
-    },
-    {
-      "epoch": 2.3279622302502054,
-      "grad_norm": 0.0021661140490323305,
-      "learning_rate": 0.00019999732739348604,
-      "loss": 46.0,
-      "step": 30448
-    },
-    {
-      "epoch": 2.328038687233595,
-      "grad_norm": 0.00602521700784564,
-      "learning_rate": 0.00019999732721787407,
-      "loss": 46.0,
-      "step": 30449
-    },
-    {
-      "epoch": 2.328115144216985,
-      "grad_norm": 0.0005183838657103479,
-      "learning_rate": 0.00019999732704225636,
-      "loss": 46.0,
-      "step": 30450
-    },
-    {
-      "epoch": 2.3281916012003747,
-      "grad_norm": 0.0010052935685962439,
-      "learning_rate": 0.00019999732686663287,
-      "loss": 46.0,
-      "step": 30451
-    },
-    {
-      "epoch": 2.3282680581837645,
-      "grad_norm": 0.0010794798145070672,
-      "learning_rate": 0.00019999732669100358,
-      "loss": 46.0,
-      "step": 30452
-    },
-    {
-      "epoch": 2.3283445151671542,
-      "grad_norm": 0.0012344920542091131,
-      "learning_rate": 0.00019999732651536852,
-      "loss": 46.0,
-      "step": 30453
-    },
-    {
-      "epoch": 2.3284209721505436,
-      "grad_norm": 0.017593327909708023,
-      "learning_rate": 0.00019999732633972772,
-      "loss": 46.0,
-      "step": 30454
-    },
-    {
-      "epoch": 2.3284974291339333,
-      "grad_norm": 0.0006167201208882034,
-      "learning_rate": 0.00019999732616408114,
-      "loss": 46.0,
-      "step": 30455
-    },
-    {
-      "epoch": 2.328573886117323,
-      "grad_norm": 0.004931359086185694,
-      "learning_rate": 0.00019999732598842879,
-      "loss": 46.0,
-      "step": 30456
-    },
-    {
-      "epoch": 2.328650343100713,
-      "grad_norm": 0.007100431248545647,
-      "learning_rate": 0.00019999732581277066,
-      "loss": 46.0,
-      "step": 30457
-    },
-    {
-      "epoch": 2.3287268000841026,
-      "grad_norm": 0.0004416010924614966,
-      "learning_rate": 0.0001999973256371068,
-      "loss": 46.0,
-      "step": 30458
-    },
-    {
-      "epoch": 2.3288032570674924,
-      "grad_norm": 0.00205968483351171,
-      "learning_rate": 0.00019999732546143711,
-      "loss": 46.0,
-      "step": 30459
-    },
-    {
-      "epoch": 2.328879714050882,
-      "grad_norm": 0.0005541814607568085,
-      "learning_rate": 0.0001999973252857617,
-      "loss": 46.0,
-      "step": 30460
-    },
-    {
-      "epoch": 2.328956171034272,
-      "grad_norm": 0.0046439687721431255,
-      "learning_rate": 0.0001999973251100805,
-      "loss": 46.0,
-      "step": 30461
-    },
-    {
-      "epoch": 2.3290326280176616,
-      "grad_norm": 0.0010102344676852226,
-      "learning_rate": 0.0001999973249343935,
-      "loss": 46.0,
-      "step": 30462
-    },
-    {
-      "epoch": 2.3291090850010514,
-      "grad_norm": 0.0005489391623996198,
-      "learning_rate": 0.0001999973247587008,
-      "loss": 46.0,
-      "step": 30463
-    },
-    {
-      "epoch": 2.329185541984441,
-      "grad_norm": 0.0008963418076746166,
-      "learning_rate": 0.00019999732458300228,
-      "loss": 46.0,
-      "step": 30464
-    },
-    {
-      "epoch": 2.329261998967831,
-      "grad_norm": 0.0007536340854130685,
-      "learning_rate": 0.000199997324407298,
-      "loss": 46.0,
-      "step": 30465
-    },
-    {
-      "epoch": 2.3293384559512207,
-      "grad_norm": 0.0007768994546495378,
-      "learning_rate": 0.00019999732423158797,
-      "loss": 46.0,
-      "step": 30466
-    },
-    {
-      "epoch": 2.32941491293461,
-      "grad_norm": 0.000318201316986233,
-      "learning_rate": 0.00019999732405587216,
-      "loss": 46.0,
-      "step": 30467
-    },
-    {
-      "epoch": 2.3294913699179998,
-      "grad_norm": 0.0025257545057684183,
-      "learning_rate": 0.00019999732388015055,
-      "loss": 46.0,
-      "step": 30468
-    },
-    {
-      "epoch": 2.3295678269013895,
-      "grad_norm": 0.0012237076880410314,
-      "learning_rate": 0.0001999973237044232,
-      "loss": 46.0,
-      "step": 30469
-    },
-    {
-      "epoch": 2.3296442838847793,
-      "grad_norm": 0.0021016024984419346,
-      "learning_rate": 0.0001999973235286901,
-      "loss": 46.0,
-      "step": 30470
-    },
-    {
-      "epoch": 2.329720740868169,
-      "grad_norm": 0.00078096758807078,
-      "learning_rate": 0.0001999973233529512,
-      "loss": 46.0,
-      "step": 30471
-    },
-    {
-      "epoch": 2.329797197851559,
-      "grad_norm": 0.0003566005907487124,
-      "learning_rate": 0.00019999732317720653,
-      "loss": 46.0,
-      "step": 30472
-    },
-    {
-      "epoch": 2.3298736548349486,
-      "grad_norm": 0.0019797333516180515,
-      "learning_rate": 0.0001999973230014561,
-      "loss": 46.0,
-      "step": 30473
-    },
-    {
-      "epoch": 2.3299501118183383,
-      "grad_norm": 0.002198479138314724,
-      "learning_rate": 0.00019999732282569992,
-      "loss": 46.0,
-      "step": 30474
-    },
-    {
-      "epoch": 2.330026568801728,
-      "grad_norm": 0.001873291446827352,
-      "learning_rate": 0.00019999732264993795,
-      "loss": 46.0,
-      "step": 30475
-    },
-    {
-      "epoch": 2.3301030257851174,
-      "grad_norm": 0.0012714068870991468,
-      "learning_rate": 0.0001999973224741702,
-      "loss": 46.0,
-      "step": 30476
-    },
-    {
-      "epoch": 2.330179482768507,
-      "grad_norm": 0.0009153428836725652,
-      "learning_rate": 0.00019999732229839673,
-      "loss": 46.0,
-      "step": 30477
-    },
-    {
-      "epoch": 2.330255939751897,
-      "grad_norm": 0.0014175878604874015,
-      "learning_rate": 0.00019999732212261741,
-      "loss": 46.0,
-      "step": 30478
-    },
-    {
-      "epoch": 2.3303323967352867,
-      "grad_norm": 0.0028817395213991404,
-      "learning_rate": 0.00019999732194683238,
-      "loss": 46.0,
-      "step": 30479
-    },
-    {
-      "epoch": 2.3304088537186765,
-      "grad_norm": 0.0017851786687970161,
-      "learning_rate": 0.00019999732177104158,
-      "loss": 46.0,
-      "step": 30480
-    },
-    {
-      "epoch": 2.330485310702066,
-      "grad_norm": 0.0026716215070337057,
-      "learning_rate": 0.00019999732159524497,
-      "loss": 46.0,
-      "step": 30481
-    },
-    {
-      "epoch": 2.330561767685456,
-      "grad_norm": 0.0009459813009016216,
-      "learning_rate": 0.00019999732141944262,
-      "loss": 46.0,
-      "step": 30482
-    },
-    {
-      "epoch": 2.3306382246688457,
-      "grad_norm": 0.0034814453683793545,
-      "learning_rate": 0.0001999973212436345,
-      "loss": 46.0,
-      "step": 30483
-    },
-    {
-      "epoch": 2.3307146816522355,
-      "grad_norm": 0.0011281903134658933,
-      "learning_rate": 0.00019999732106782062,
-      "loss": 46.0,
-      "step": 30484
-    },
-    {
-      "epoch": 2.3307911386356253,
-      "grad_norm": 0.0005879094824194908,
-      "learning_rate": 0.00019999732089200095,
-      "loss": 46.0,
-      "step": 30485
-    },
-    {
-      "epoch": 2.330867595619015,
-      "grad_norm": 0.002650426933541894,
-      "learning_rate": 0.00019999732071617553,
-      "loss": 46.0,
-      "step": 30486
-    },
-    {
-      "epoch": 2.330944052602405,
-      "grad_norm": 0.0069006881676614285,
-      "learning_rate": 0.0001999973205403443,
-      "loss": 46.0,
-      "step": 30487
-    },
-    {
-      "epoch": 2.331020509585794,
-      "grad_norm": 0.0012131606927141547,
-      "learning_rate": 0.00019999732036450735,
-      "loss": 46.0,
-      "step": 30488
-    },
-    {
-      "epoch": 2.331096966569184,
-      "grad_norm": 0.0006248114514164627,
-      "learning_rate": 0.0001999973201886646,
-      "loss": 46.0,
-      "step": 30489
-    },
-    {
-      "epoch": 2.3311734235525736,
-      "grad_norm": 0.0011029633460566401,
-      "learning_rate": 0.0001999973200128161,
-      "loss": 46.0,
-      "step": 30490
-    },
-    {
-      "epoch": 2.3312498805359634,
-      "grad_norm": 0.0009441353613510728,
-      "learning_rate": 0.00019999731983696184,
-      "loss": 46.0,
-      "step": 30491
-    },
-    {
-      "epoch": 2.331326337519353,
-      "grad_norm": 0.0024268615525215864,
-      "learning_rate": 0.0001999973196611018,
-      "loss": 46.0,
-      "step": 30492
-    },
-    {
-      "epoch": 2.331402794502743,
-      "grad_norm": 0.00297020236030221,
-      "learning_rate": 0.000199997319485236,
-      "loss": 46.0,
-      "step": 30493
-    },
-    {
-      "epoch": 2.3314792514861327,
-      "grad_norm": 0.0012593739666044712,
-      "learning_rate": 0.0001999973193093644,
-      "loss": 46.0,
-      "step": 30494
-    },
-    {
-      "epoch": 2.3315557084695224,
-      "grad_norm": 0.0007747412309981883,
-      "learning_rate": 0.00019999731913348704,
-      "loss": 46.0,
-      "step": 30495
-    },
-    {
-      "epoch": 2.331632165452912,
-      "grad_norm": 0.0008016584906727076,
-      "learning_rate": 0.00019999731895760392,
-      "loss": 46.0,
-      "step": 30496
-    },
-    {
-      "epoch": 2.331708622436302,
-      "grad_norm": 0.00738663412630558,
-      "learning_rate": 0.00019999731878171502,
-      "loss": 46.0,
-      "step": 30497
-    },
-    {
-      "epoch": 2.3317850794196913,
-      "grad_norm": 0.000793817569501698,
-      "learning_rate": 0.00019999731860582037,
-      "loss": 46.0,
-      "step": 30498
-    },
-    {
-      "epoch": 2.331861536403081,
-      "grad_norm": 0.0009280915837734938,
-      "learning_rate": 0.00019999731842991993,
-      "loss": 46.0,
-      "step": 30499
-    },
-    {
-      "epoch": 2.331937993386471,
-      "grad_norm": 0.0012025247560814023,
-      "learning_rate": 0.00019999731825401374,
-      "loss": 46.0,
-      "step": 30500
-    },
-    {
-      "epoch": 2.3320144503698605,
-      "grad_norm": 0.0007714900420978665,
-      "learning_rate": 0.00019999731807810177,
-      "loss": 46.0,
-      "step": 30501
-    },
-    {
-      "epoch": 2.3320909073532503,
-      "grad_norm": 0.0005416963831521571,
-      "learning_rate": 0.00019999731790218403,
-      "loss": 46.0,
-      "step": 30502
-    },
-    {
-      "epoch": 2.33216736433664,
-      "grad_norm": 0.001500443322584033,
-      "learning_rate": 0.00019999731772626052,
-      "loss": 46.0,
-      "step": 30503
-    },
-    {
-      "epoch": 2.33224382132003,
-      "grad_norm": 0.0024475310929119587,
-      "learning_rate": 0.0001999973175503312,
-      "loss": 46.0,
-      "step": 30504
-    },
-    {
-      "epoch": 2.3323202783034196,
-      "grad_norm": 0.000529143784660846,
-      "learning_rate": 0.00019999731737439618,
-      "loss": 46.0,
-      "step": 30505
-    },
-    {
-      "epoch": 2.3323967352868094,
-      "grad_norm": 0.0009378319373354316,
-      "learning_rate": 0.00019999731719845537,
-      "loss": 46.0,
-      "step": 30506
-    },
-    {
-      "epoch": 2.332473192270199,
-      "grad_norm": 0.0016075520543381572,
-      "learning_rate": 0.0001999973170225088,
-      "loss": 46.0,
-      "step": 30507
-    },
-    {
-      "epoch": 2.332549649253589,
-      "grad_norm": 0.0009560227626934648,
-      "learning_rate": 0.00019999731684655641,
-      "loss": 46.0,
-      "step": 30508
-    },
-    {
-      "epoch": 2.3326261062369786,
-      "grad_norm": 0.002297154162079096,
-      "learning_rate": 0.00019999731667059832,
-      "loss": 46.0,
-      "step": 30509
-    },
-    {
-      "epoch": 2.332702563220368,
-      "grad_norm": 0.0015520333545282483,
-      "learning_rate": 0.0001999973164946344,
-      "loss": 46.0,
-      "step": 30510
-    },
-    {
-      "epoch": 2.3327790202037577,
-      "grad_norm": 0.0014315292937681079,
-      "learning_rate": 0.00019999731631866477,
-      "loss": 46.0,
-      "step": 30511
-    },
-    {
-      "epoch": 2.3328554771871475,
-      "grad_norm": 0.0010838517919182777,
-      "learning_rate": 0.0001999973161426893,
-      "loss": 46.0,
-      "step": 30512
-    },
-    {
-      "epoch": 2.3329319341705372,
-      "grad_norm": 0.0006422838196158409,
-      "learning_rate": 0.0001999973159667081,
-      "loss": 46.0,
-      "step": 30513
-    },
-    {
-      "epoch": 2.333008391153927,
-      "grad_norm": 0.0005534583469852805,
-      "learning_rate": 0.00019999731579072117,
-      "loss": 46.0,
-      "step": 30514
-    },
-    {
-      "epoch": 2.3330848481373168,
-      "grad_norm": 0.0018934594700112939,
-      "learning_rate": 0.0001999973156147284,
-      "loss": 46.0,
-      "step": 30515
-    },
-    {
-      "epoch": 2.3331613051207065,
-      "grad_norm": 0.0005745832459069788,
-      "learning_rate": 0.0001999973154387299,
-      "loss": 46.0,
-      "step": 30516
-    },
-    {
-      "epoch": 2.3332377621040963,
-      "grad_norm": 0.000632535433396697,
-      "learning_rate": 0.0001999973152627256,
-      "loss": 46.0,
-      "step": 30517
-    },
-    {
-      "epoch": 2.333314219087486,
-      "grad_norm": 0.0062083289958536625,
-      "learning_rate": 0.00019999731508671558,
-      "loss": 46.0,
-      "step": 30518
-    },
-    {
-      "epoch": 2.333390676070876,
-      "grad_norm": 0.00106734037399292,
-      "learning_rate": 0.00019999731491069975,
-      "loss": 46.0,
-      "step": 30519
-    },
-    {
-      "epoch": 2.333467133054265,
-      "grad_norm": 0.0006080340826883912,
-      "learning_rate": 0.00019999731473467817,
-      "loss": 46.0,
-      "step": 30520
-    },
-    {
-      "epoch": 2.333543590037655,
-      "grad_norm": 0.00070382677949965,
-      "learning_rate": 0.00019999731455865082,
-      "loss": 46.0,
-      "step": 30521
-    },
-    {
-      "epoch": 2.3336200470210446,
-      "grad_norm": 0.0015601797495037317,
-      "learning_rate": 0.0001999973143826177,
-      "loss": 46.0,
-      "step": 30522
-    },
-    {
-      "epoch": 2.3336965040044344,
-      "grad_norm": 0.0027359300293028355,
-      "learning_rate": 0.0001999973142065788,
-      "loss": 46.0,
-      "step": 30523
-    },
-    {
-      "epoch": 2.333772960987824,
-      "grad_norm": 0.000908554473426193,
-      "learning_rate": 0.00019999731403053415,
-      "loss": 46.0,
-      "step": 30524
-    },
-    {
-      "epoch": 2.333849417971214,
-      "grad_norm": 0.001058922614902258,
-      "learning_rate": 0.0001999973138544837,
-      "loss": 46.0,
-      "step": 30525
-    },
-    {
-      "epoch": 2.3339258749546037,
-      "grad_norm": 0.0007167409639805555,
-      "learning_rate": 0.0001999973136784275,
-      "loss": 46.0,
-      "step": 30526
-    },
-    {
-      "epoch": 2.3340023319379934,
-      "grad_norm": 0.0019888204988092184,
-      "learning_rate": 0.00019999731350236552,
-      "loss": 46.0,
-      "step": 30527
-    },
-    {
-      "epoch": 2.334078788921383,
-      "grad_norm": 0.00045928670442663133,
-      "learning_rate": 0.00019999731332629779,
-      "loss": 46.0,
-      "step": 30528
-    },
-    {
-      "epoch": 2.334155245904773,
-      "grad_norm": 0.0015401450218632817,
-      "learning_rate": 0.00019999731315022427,
-      "loss": 46.0,
-      "step": 30529
-    },
-    {
-      "epoch": 2.3342317028881627,
-      "grad_norm": 0.001193465432152152,
-      "learning_rate": 0.000199997312974145,
-      "loss": 46.0,
-      "step": 30530
-    },
-    {
-      "epoch": 2.3343081598715525,
-      "grad_norm": 0.0012732665054500103,
-      "learning_rate": 0.00019999731279805993,
-      "loss": 46.0,
-      "step": 30531
-    },
-    {
-      "epoch": 2.334384616854942,
-      "grad_norm": 0.0010085650719702244,
-      "learning_rate": 0.00019999731262196913,
-      "loss": 46.0,
-      "step": 30532
-    },
-    {
-      "epoch": 2.3344610738383316,
-      "grad_norm": 0.0021177898161113262,
-      "learning_rate": 0.00019999731244587255,
-      "loss": 46.0,
-      "step": 30533
-    },
-    {
-      "epoch": 2.3345375308217213,
-      "grad_norm": 0.003935427404940128,
-      "learning_rate": 0.0001999973122697702,
-      "loss": 46.0,
-      "step": 30534
-    },
-    {
-      "epoch": 2.334613987805111,
-      "grad_norm": 0.0009917697170749307,
-      "learning_rate": 0.00019999731209366205,
-      "loss": 46.0,
-      "step": 30535
-    },
-    {
-      "epoch": 2.334690444788501,
-      "grad_norm": 0.002426746068522334,
-      "learning_rate": 0.00019999731191754815,
-      "loss": 46.0,
-      "step": 30536
-    },
-    {
-      "epoch": 2.3347669017718906,
-      "grad_norm": 0.0019559075590223074,
-      "learning_rate": 0.00019999731174142848,
-      "loss": 46.0,
-      "step": 30537
-    },
-    {
-      "epoch": 2.3348433587552804,
-      "grad_norm": 0.0004813242703676224,
-      "learning_rate": 0.00019999731156530306,
-      "loss": 46.0,
-      "step": 30538
-    },
-    {
-      "epoch": 2.33491981573867,
-      "grad_norm": 0.0017152774380519986,
-      "learning_rate": 0.00019999731138917187,
-      "loss": 46.0,
-      "step": 30539
-    },
-    {
-      "epoch": 2.33499627272206,
-      "grad_norm": 0.0010211910121142864,
-      "learning_rate": 0.00019999731121303488,
-      "loss": 46.0,
-      "step": 30540
-    },
-    {
-      "epoch": 2.3350727297054497,
-      "grad_norm": 0.002255661878734827,
-      "learning_rate": 0.00019999731103689214,
-      "loss": 46.0,
-      "step": 30541
-    },
-    {
-      "epoch": 2.335149186688839,
-      "grad_norm": 0.0010235972004011273,
-      "learning_rate": 0.00019999731086074363,
-      "loss": 46.0,
-      "step": 30542
-    },
-    {
-      "epoch": 2.3352256436722287,
-      "grad_norm": 0.0012261184165254235,
-      "learning_rate": 0.00019999731068458935,
-      "loss": 46.0,
-      "step": 30543
-    },
-    {
-      "epoch": 2.3353021006556185,
-      "grad_norm": 0.001387918833643198,
-      "learning_rate": 0.00019999731050842932,
-      "loss": 46.0,
-      "step": 30544
-    },
-    {
-      "epoch": 2.3353785576390083,
-      "grad_norm": 0.0010396098950877786,
-      "learning_rate": 0.00019999731033226348,
-      "loss": 46.0,
-      "step": 30545
-    },
-    {
-      "epoch": 2.335455014622398,
-      "grad_norm": 0.0004188910825178027,
-      "learning_rate": 0.0001999973101560919,
-      "loss": 46.0,
-      "step": 30546
-    },
-    {
-      "epoch": 2.335531471605788,
-      "grad_norm": 0.00046145892702043056,
-      "learning_rate": 0.00019999730997991456,
-      "loss": 46.0,
-      "step": 30547
-    },
-    {
-      "epoch": 2.3356079285891775,
-      "grad_norm": 0.0006948307272978127,
-      "learning_rate": 0.0001999973098037314,
-      "loss": 46.0,
-      "step": 30548
-    },
-    {
-      "epoch": 2.3356843855725673,
-      "grad_norm": 0.002568683587014675,
-      "learning_rate": 0.00019999730962754253,
-      "loss": 46.0,
-      "step": 30549
-    },
-    {
-      "epoch": 2.335760842555957,
-      "grad_norm": 0.0007333027315326035,
-      "learning_rate": 0.00019999730945134786,
-      "loss": 46.0,
-      "step": 30550
-    },
-    {
-      "epoch": 2.335837299539347,
-      "grad_norm": 0.0020218323916196823,
-      "learning_rate": 0.00019999730927514745,
-      "loss": 46.0,
-      "step": 30551
-    },
-    {
-      "epoch": 2.3359137565227366,
-      "grad_norm": 0.00773219857364893,
-      "learning_rate": 0.00019999730909894123,
-      "loss": 46.0,
-      "step": 30552
-    },
-    {
-      "epoch": 2.3359902135061263,
-      "grad_norm": 0.0006330051692202687,
-      "learning_rate": 0.00019999730892272927,
-      "loss": 46.0,
-      "step": 30553
-    },
-    {
-      "epoch": 2.3360666704895157,
-      "grad_norm": 0.001696947729215026,
-      "learning_rate": 0.0001999973087465115,
-      "loss": 46.0,
-      "step": 30554
-    },
-    {
-      "epoch": 2.3361431274729054,
-      "grad_norm": 0.004732252564281225,
-      "learning_rate": 0.000199997308570288,
-      "loss": 46.0,
-      "step": 30555
-    },
-    {
-      "epoch": 2.336219584456295,
-      "grad_norm": 0.0006153184804134071,
-      "learning_rate": 0.00019999730839405873,
-      "loss": 46.0,
-      "step": 30556
-    },
-    {
-      "epoch": 2.336296041439685,
-      "grad_norm": 0.0025214930064976215,
-      "learning_rate": 0.00019999730821782365,
-      "loss": 46.0,
-      "step": 30557
-    },
-    {
-      "epoch": 2.3363724984230747,
-      "grad_norm": 0.0012366330483928323,
-      "learning_rate": 0.00019999730804158285,
-      "loss": 46.0,
-      "step": 30558
-    },
-    {
-      "epoch": 2.3364489554064645,
-      "grad_norm": 0.0008409314905293286,
-      "learning_rate": 0.00019999730786533624,
-      "loss": 46.0,
-      "step": 30559
-    },
-    {
-      "epoch": 2.3365254123898542,
-      "grad_norm": 0.0023845331743359566,
-      "learning_rate": 0.0001999973076890839,
-      "loss": 46.0,
-      "step": 30560
-    },
-    {
-      "epoch": 2.336601869373244,
-      "grad_norm": 0.0009311073808930814,
-      "learning_rate": 0.00019999730751282577,
-      "loss": 46.0,
-      "step": 30561
-    },
-    {
-      "epoch": 2.3366783263566338,
-      "grad_norm": 0.0021973440889269114,
-      "learning_rate": 0.00019999730733656187,
-      "loss": 46.0,
-      "step": 30562
-    },
-    {
-      "epoch": 2.3367547833400235,
-      "grad_norm": 0.0005608109058812261,
-      "learning_rate": 0.00019999730716029223,
-      "loss": 46.0,
-      "step": 30563
-    },
-    {
-      "epoch": 2.336831240323413,
-      "grad_norm": 0.0018422291614115238,
-      "learning_rate": 0.00019999730698401678,
-      "loss": 46.0,
-      "step": 30564
-    },
-    {
-      "epoch": 2.3369076973068026,
-      "grad_norm": 0.0034021951723843813,
-      "learning_rate": 0.00019999730680773557,
-      "loss": 46.0,
-      "step": 30565
-    },
-    {
-      "epoch": 2.3369841542901924,
-      "grad_norm": 0.001579405041411519,
-      "learning_rate": 0.00019999730663144857,
-      "loss": 46.0,
-      "step": 30566
-    },
-    {
-      "epoch": 2.337060611273582,
-      "grad_norm": 0.000303540175082162,
-      "learning_rate": 0.00019999730645515584,
-      "loss": 46.0,
-      "step": 30567
-    },
-    {
-      "epoch": 2.337137068256972,
-      "grad_norm": 0.0019256646046414971,
-      "learning_rate": 0.00019999730627885733,
-      "loss": 46.0,
-      "step": 30568
-    },
-    {
-      "epoch": 2.3372135252403616,
-      "grad_norm": 0.0006757780211046338,
-      "learning_rate": 0.00019999730610255304,
-      "loss": 46.0,
-      "step": 30569
-    },
-    {
-      "epoch": 2.3372899822237514,
-      "grad_norm": 0.0009310618625022471,
-      "learning_rate": 0.000199997305926243,
-      "loss": 46.0,
-      "step": 30570
-    },
-    {
-      "epoch": 2.337366439207141,
-      "grad_norm": 0.0010758353164419532,
-      "learning_rate": 0.00019999730574992718,
-      "loss": 46.0,
-      "step": 30571
-    },
-    {
-      "epoch": 2.337442896190531,
-      "grad_norm": 0.0011591686634346843,
-      "learning_rate": 0.00019999730557360558,
-      "loss": 46.0,
-      "step": 30572
-    },
-    {
-      "epoch": 2.3375193531739207,
-      "grad_norm": 0.002370295813307166,
-      "learning_rate": 0.00019999730539727823,
-      "loss": 46.0,
-      "step": 30573
-    },
-    {
-      "epoch": 2.3375958101573104,
-      "grad_norm": 0.0007798952865414321,
-      "learning_rate": 0.00019999730522094508,
-      "loss": 46.0,
-      "step": 30574
-    },
-    {
-      "epoch": 2.3376722671407,
-      "grad_norm": 0.001364949275739491,
-      "learning_rate": 0.00019999730504460618,
-      "loss": 46.0,
-      "step": 30575
-    },
-    {
-      "epoch": 2.3377487241240895,
-      "grad_norm": 0.0006470885127782822,
-      "learning_rate": 0.00019999730486826154,
-      "loss": 46.0,
-      "step": 30576
-    },
-    {
-      "epoch": 2.3378251811074793,
-      "grad_norm": 0.005174380727112293,
-      "learning_rate": 0.00019999730469191112,
-      "loss": 46.0,
-      "step": 30577
-    },
-    {
-      "epoch": 2.337901638090869,
-      "grad_norm": 0.0017438264330849051,
-      "learning_rate": 0.0001999973045155549,
-      "loss": 46.0,
-      "step": 30578
-    },
-    {
-      "epoch": 2.337978095074259,
-      "grad_norm": 0.0007920488133095205,
-      "learning_rate": 0.00019999730433919292,
-      "loss": 46.0,
-      "step": 30579
-    },
-    {
-      "epoch": 2.3380545520576486,
-      "grad_norm": 0.0006459720898419619,
-      "learning_rate": 0.00019999730416282518,
-      "loss": 46.0,
-      "step": 30580
-    },
-    {
-      "epoch": 2.3381310090410383,
-      "grad_norm": 0.001057083485648036,
-      "learning_rate": 0.00019999730398645164,
-      "loss": 46.0,
-      "step": 30581
-    },
-    {
-      "epoch": 2.338207466024428,
-      "grad_norm": 0.0011769833508878946,
-      "learning_rate": 0.00019999730381007239,
-      "loss": 46.0,
-      "step": 30582
-    },
-    {
-      "epoch": 2.338283923007818,
-      "grad_norm": 0.000645259628072381,
-      "learning_rate": 0.00019999730363368733,
-      "loss": 46.0,
-      "step": 30583
-    },
-    {
-      "epoch": 2.3383603799912076,
-      "grad_norm": 0.0006173406145535409,
-      "learning_rate": 0.0001999973034572965,
-      "loss": 46.0,
-      "step": 30584
-    },
-    {
-      "epoch": 2.338436836974597,
-      "grad_norm": 0.0008263772469945252,
-      "learning_rate": 0.00019999730328089992,
-      "loss": 46.0,
-      "step": 30585
-    },
-    {
-      "epoch": 2.3385132939579867,
-      "grad_norm": 0.0009204766829498112,
-      "learning_rate": 0.00019999730310449755,
-      "loss": 46.0,
-      "step": 30586
-    },
-    {
-      "epoch": 2.3385897509413764,
-      "grad_norm": 0.0016777734272181988,
-      "learning_rate": 0.00019999730292808942,
-      "loss": 46.0,
-      "step": 30587
-    },
-    {
-      "epoch": 2.338666207924766,
-      "grad_norm": 0.0012596884043887258,
-      "learning_rate": 0.00019999730275167555,
-      "loss": 46.0,
-      "step": 30588
-    },
-    {
-      "epoch": 2.338742664908156,
-      "grad_norm": 0.0017181789735332131,
-      "learning_rate": 0.00019999730257525586,
-      "loss": 46.0,
-      "step": 30589
-    },
-    {
-      "epoch": 2.3388191218915457,
-      "grad_norm": 0.0006443418678827584,
-      "learning_rate": 0.0001999973023988304,
-      "loss": 46.0,
-      "step": 30590
-    },
-    {
-      "epoch": 2.3388955788749355,
-      "grad_norm": 0.0016996599733829498,
-      "learning_rate": 0.00019999730222239922,
-      "loss": 46.0,
-      "step": 30591
-    },
-    {
-      "epoch": 2.3389720358583252,
-      "grad_norm": 0.0005018014926463366,
-      "learning_rate": 0.00019999730204596223,
-      "loss": 46.0,
-      "step": 30592
-    },
-    {
-      "epoch": 2.339048492841715,
-      "grad_norm": 0.0034854840487241745,
-      "learning_rate": 0.0001999973018695195,
-      "loss": 46.0,
-      "step": 30593
-    },
-    {
-      "epoch": 2.3391249498251048,
-      "grad_norm": 0.004720806144177914,
-      "learning_rate": 0.000199997301693071,
-      "loss": 46.0,
-      "step": 30594
-    },
-    {
-      "epoch": 2.3392014068084945,
-      "grad_norm": 0.0018677798798307776,
-      "learning_rate": 0.00019999730151661668,
-      "loss": 46.0,
-      "step": 30595
-    },
-    {
-      "epoch": 2.3392778637918843,
-      "grad_norm": 0.001091250916942954,
-      "learning_rate": 0.00019999730134015665,
-      "loss": 46.0,
-      "step": 30596
-    },
-    {
-      "epoch": 2.339354320775274,
-      "grad_norm": 0.0006587326060980558,
-      "learning_rate": 0.00019999730116369082,
-      "loss": 46.0,
-      "step": 30597
-    },
-    {
-      "epoch": 2.3394307777586634,
-      "grad_norm": 0.00306577212177217,
-      "learning_rate": 0.00019999730098721921,
-      "loss": 46.0,
-      "step": 30598
-    },
-    {
-      "epoch": 2.339507234742053,
-      "grad_norm": 0.0012429801281541586,
-      "learning_rate": 0.00019999730081074186,
-      "loss": 46.0,
-      "step": 30599
-    },
-    {
-      "epoch": 2.339583691725443,
-      "grad_norm": 0.0006678019417449832,
-      "learning_rate": 0.00019999730063425872,
-      "loss": 46.0,
-      "step": 30600
-    },
-    {
-      "epoch": 2.3396601487088327,
-      "grad_norm": 0.0007223449647426605,
-      "learning_rate": 0.00019999730045776985,
-      "loss": 46.0,
-      "step": 30601
-    },
-    {
-      "epoch": 2.3397366056922224,
-      "grad_norm": 0.001051132334396243,
-      "learning_rate": 0.00019999730028127518,
-      "loss": 46.0,
-      "step": 30602
-    },
-    {
-      "epoch": 2.339813062675612,
-      "grad_norm": 0.0008924936992116272,
-      "learning_rate": 0.00019999730010477473,
-      "loss": 46.0,
-      "step": 30603
-    },
-    {
-      "epoch": 2.339889519659002,
-      "grad_norm": 0.0009778476087376475,
-      "learning_rate": 0.00019999729992826852,
-      "loss": 46.0,
-      "step": 30604
-    },
-    {
-      "epoch": 2.3399659766423917,
-      "grad_norm": 0.0033746790140867233,
-      "learning_rate": 0.00019999729975175655,
-      "loss": 46.0,
-      "step": 30605
-    },
-    {
-      "epoch": 2.3400424336257815,
-      "grad_norm": 0.0012777198571711779,
-      "learning_rate": 0.00019999729957523882,
-      "loss": 46.0,
-      "step": 30606
-    },
-    {
-      "epoch": 2.3401188906091708,
-      "grad_norm": 0.0009875097312033176,
-      "learning_rate": 0.00019999729939871528,
-      "loss": 46.0,
-      "step": 30607
-    },
-    {
-      "epoch": 2.3401953475925605,
-      "grad_norm": 0.0009483841131441295,
-      "learning_rate": 0.000199997299222186,
-      "loss": 46.0,
-      "step": 30608
-    },
-    {
-      "epoch": 2.3402718045759503,
-      "grad_norm": 0.000711943837814033,
-      "learning_rate": 0.00019999729904565094,
-      "loss": 46.0,
-      "step": 30609
-    },
-    {
-      "epoch": 2.34034826155934,
-      "grad_norm": 0.00043770045158453286,
-      "learning_rate": 0.00019999729886911011,
-      "loss": 46.0,
-      "step": 30610
-    },
-    {
-      "epoch": 2.34042471854273,
-      "grad_norm": 0.0006217218469828367,
-      "learning_rate": 0.0001999972986925635,
-      "loss": 46.0,
-      "step": 30611
-    },
-    {
-      "epoch": 2.3405011755261196,
-      "grad_norm": 0.0036873382050544024,
-      "learning_rate": 0.00019999729851601116,
-      "loss": 46.0,
-      "step": 30612
-    },
-    {
-      "epoch": 2.3405776325095093,
-      "grad_norm": 0.011502964422106743,
-      "learning_rate": 0.00019999729833945304,
-      "loss": 46.0,
-      "step": 30613
-    },
-    {
-      "epoch": 2.340654089492899,
-      "grad_norm": 0.0013821851462125778,
-      "learning_rate": 0.00019999729816288912,
-      "loss": 46.0,
-      "step": 30614
-    },
-    {
-      "epoch": 2.340730546476289,
-      "grad_norm": 0.002367577748373151,
-      "learning_rate": 0.00019999729798631947,
-      "loss": 46.0,
-      "step": 30615
-    },
-    {
-      "epoch": 2.3408070034596786,
-      "grad_norm": 0.001113171805627644,
-      "learning_rate": 0.00019999729780974403,
-      "loss": 46.0,
-      "step": 30616
-    },
-    {
-      "epoch": 2.3408834604430684,
-      "grad_norm": 0.0007057064794935286,
-      "learning_rate": 0.00019999729763316282,
-      "loss": 46.0,
-      "step": 30617
-    },
-    {
-      "epoch": 2.340959917426458,
-      "grad_norm": 0.0005223373300395906,
-      "learning_rate": 0.00019999729745657585,
-      "loss": 46.0,
-      "step": 30618
-    },
-    {
-      "epoch": 2.341036374409848,
-      "grad_norm": 0.0008474922506138682,
-      "learning_rate": 0.00019999729727998312,
-      "loss": 46.0,
-      "step": 30619
-    },
-    {
-      "epoch": 2.3411128313932372,
-      "grad_norm": 0.0010565832490101457,
-      "learning_rate": 0.0001999972971033846,
-      "loss": 46.0,
-      "step": 30620
-    },
-    {
-      "epoch": 2.341189288376627,
-      "grad_norm": 0.001985379261896014,
-      "learning_rate": 0.0001999972969267803,
-      "loss": 46.0,
-      "step": 30621
-    },
-    {
-      "epoch": 2.3412657453600167,
-      "grad_norm": 0.0007389218080788851,
-      "learning_rate": 0.00019999729675017024,
-      "loss": 46.0,
-      "step": 30622
-    },
-    {
-      "epoch": 2.3413422023434065,
-      "grad_norm": 0.0008347560069523752,
-      "learning_rate": 0.00019999729657355441,
-      "loss": 46.0,
-      "step": 30623
-    },
-    {
-      "epoch": 2.3414186593267963,
-      "grad_norm": 0.0028940513730049133,
-      "learning_rate": 0.0001999972963969328,
-      "loss": 46.0,
-      "step": 30624
-    },
-    {
-      "epoch": 2.341495116310186,
-      "grad_norm": 0.0005485435831360519,
-      "learning_rate": 0.00019999729622030546,
-      "loss": 46.0,
-      "step": 30625
-    },
-    {
-      "epoch": 2.341571573293576,
-      "grad_norm": 0.0010249732295051217,
-      "learning_rate": 0.0001999972960436723,
-      "loss": 46.0,
-      "step": 30626
-    },
-    {
-      "epoch": 2.3416480302769656,
-      "grad_norm": 0.0017250453820452094,
-      "learning_rate": 0.00019999729586703344,
-      "loss": 46.0,
-      "step": 30627
-    },
-    {
-      "epoch": 2.3417244872603553,
-      "grad_norm": 0.00275959400460124,
-      "learning_rate": 0.00019999729569038875,
-      "loss": 46.0,
-      "step": 30628
-    },
-    {
-      "epoch": 2.3418009442437446,
-      "grad_norm": 0.0008972191135399044,
-      "learning_rate": 0.0001999972955137383,
-      "loss": 46.0,
-      "step": 30629
-    },
-    {
-      "epoch": 2.3418774012271344,
-      "grad_norm": 0.005577577743679285,
-      "learning_rate": 0.00019999729533708212,
-      "loss": 46.0,
-      "step": 30630
-    },
-    {
-      "epoch": 2.341953858210524,
-      "grad_norm": 0.006267010234296322,
-      "learning_rate": 0.00019999729516042013,
-      "loss": 46.0,
-      "step": 30631
-    },
-    {
-      "epoch": 2.342030315193914,
-      "grad_norm": 0.003595035057514906,
-      "learning_rate": 0.0001999972949837524,
-      "loss": 46.0,
-      "step": 30632
-    },
-    {
-      "epoch": 2.3421067721773037,
-      "grad_norm": 0.005955283064395189,
-      "learning_rate": 0.00019999729480707888,
-      "loss": 46.0,
-      "step": 30633
-    },
-    {
-      "epoch": 2.3421832291606934,
-      "grad_norm": 0.0007644465076737106,
-      "learning_rate": 0.00019999729463039958,
-      "loss": 46.0,
-      "step": 30634
-    },
-    {
-      "epoch": 2.342259686144083,
-      "grad_norm": 0.0017080981051549315,
-      "learning_rate": 0.00019999729445371452,
-      "loss": 46.0,
-      "step": 30635
-    },
-    {
-      "epoch": 2.342336143127473,
-      "grad_norm": 0.0015398006653413177,
-      "learning_rate": 0.0001999972942770237,
-      "loss": 46.0,
-      "step": 30636
-    },
-    {
-      "epoch": 2.3424126001108627,
-      "grad_norm": 0.0013983638491481543,
-      "learning_rate": 0.00019999729410032712,
-      "loss": 46.0,
-      "step": 30637
-    },
-    {
-      "epoch": 2.3424890570942525,
-      "grad_norm": 0.0028409857768565416,
-      "learning_rate": 0.00019999729392362474,
-      "loss": 46.0,
-      "step": 30638
-    },
-    {
-      "epoch": 2.3425655140776422,
-      "grad_norm": 0.002657357370480895,
-      "learning_rate": 0.0001999972937469166,
-      "loss": 46.0,
-      "step": 30639
-    },
-    {
-      "epoch": 2.342641971061032,
-      "grad_norm": 0.0007117083296179771,
-      "learning_rate": 0.00019999729357020272,
-      "loss": 46.0,
-      "step": 30640
-    },
-    {
-      "epoch": 2.3427184280444213,
-      "grad_norm": 0.0010999025544151664,
-      "learning_rate": 0.00019999729339348303,
-      "loss": 46.0,
-      "step": 30641
-    },
-    {
-      "epoch": 2.342794885027811,
-      "grad_norm": 0.0006137785385362804,
-      "learning_rate": 0.00019999729321675759,
-      "loss": 46.0,
-      "step": 30642
-    },
-    {
-      "epoch": 2.342871342011201,
-      "grad_norm": 0.00030042778234928846,
-      "learning_rate": 0.00019999729304002637,
-      "loss": 46.0,
-      "step": 30643
-    },
-    {
-      "epoch": 2.3429477989945906,
-      "grad_norm": 0.001172129064798355,
-      "learning_rate": 0.0001999972928632894,
-      "loss": 46.0,
-      "step": 30644
-    },
-    {
-      "epoch": 2.3430242559779804,
-      "grad_norm": 0.0009331195615231991,
-      "learning_rate": 0.00019999729268654665,
-      "loss": 46.0,
-      "step": 30645
-    },
-    {
-      "epoch": 2.34310071296137,
-      "grad_norm": 0.0014410910662263632,
-      "learning_rate": 0.00019999729250979814,
-      "loss": 46.0,
-      "step": 30646
-    },
-    {
-      "epoch": 2.34317716994476,
-      "grad_norm": 0.0015979704912751913,
-      "learning_rate": 0.00019999729233304386,
-      "loss": 46.0,
-      "step": 30647
-    },
-    {
-      "epoch": 2.3432536269281496,
-      "grad_norm": 0.0007211298798210919,
-      "learning_rate": 0.00019999729215628377,
-      "loss": 46.0,
-      "step": 30648
-    },
-    {
-      "epoch": 2.3433300839115394,
-      "grad_norm": 0.0012526677455753088,
-      "learning_rate": 0.00019999729197951797,
-      "loss": 46.0,
-      "step": 30649
-    },
-    {
-      "epoch": 2.343406540894929,
-      "grad_norm": 0.0009590137051418424,
-      "learning_rate": 0.00019999729180274637,
-      "loss": 46.0,
-      "step": 30650
-    },
-    {
-      "epoch": 2.3434829978783185,
-      "grad_norm": 0.0009515249985270202,
-      "learning_rate": 0.000199997291625969,
-      "loss": 46.0,
-      "step": 30651
-    },
-    {
-      "epoch": 2.3435594548617082,
-      "grad_norm": 0.0008460519602522254,
-      "learning_rate": 0.00019999729144918587,
-      "loss": 46.0,
-      "step": 30652
-    },
-    {
-      "epoch": 2.343635911845098,
-      "grad_norm": 0.0012833202490583062,
-      "learning_rate": 0.00019999729127239698,
-      "loss": 46.0,
-      "step": 30653
-    },
-    {
-      "epoch": 2.3437123688284878,
-      "grad_norm": 0.0005600783042609692,
-      "learning_rate": 0.0001999972910956023,
-      "loss": 46.0,
-      "step": 30654
-    },
-    {
-      "epoch": 2.3437888258118775,
-      "grad_norm": 0.0024408174213021994,
-      "learning_rate": 0.00019999729091880187,
-      "loss": 46.0,
-      "step": 30655
-    },
-    {
-      "epoch": 2.3438652827952673,
-      "grad_norm": 0.001440030406229198,
-      "learning_rate": 0.00019999729074199565,
-      "loss": 46.0,
-      "step": 30656
-    },
-    {
-      "epoch": 2.343941739778657,
-      "grad_norm": 0.0006131033296696842,
-      "learning_rate": 0.0001999972905651837,
-      "loss": 46.0,
-      "step": 30657
-    },
-    {
-      "epoch": 2.344018196762047,
-      "grad_norm": 0.0018768934532999992,
-      "learning_rate": 0.00019999729038836593,
-      "loss": 46.0,
-      "step": 30658
-    },
-    {
-      "epoch": 2.3440946537454366,
-      "grad_norm": 0.0006309091695584357,
-      "learning_rate": 0.0001999972902115424,
-      "loss": 46.0,
-      "step": 30659
-    },
-    {
-      "epoch": 2.3441711107288263,
-      "grad_norm": 0.001732588279992342,
-      "learning_rate": 0.00019999729003471311,
-      "loss": 46.0,
-      "step": 30660
-    },
-    {
-      "epoch": 2.344247567712216,
-      "grad_norm": 0.0013158574001863599,
-      "learning_rate": 0.00019999728985787803,
-      "loss": 46.0,
-      "step": 30661
-    },
-    {
-      "epoch": 2.344324024695606,
-      "grad_norm": 0.000650020781904459,
-      "learning_rate": 0.00019999728968103723,
-      "loss": 46.0,
-      "step": 30662
-    },
-    {
-      "epoch": 2.344400481678995,
-      "grad_norm": 0.0009020527359098196,
-      "learning_rate": 0.00019999728950419063,
-      "loss": 46.0,
-      "step": 30663
-    },
-    {
-      "epoch": 2.344476938662385,
-      "grad_norm": 0.0011322556529194117,
-      "learning_rate": 0.00019999728932733825,
-      "loss": 46.0,
-      "step": 30664
-    },
-    {
-      "epoch": 2.3445533956457747,
-      "grad_norm": 0.0035704541951417923,
-      "learning_rate": 0.00019999728915048013,
-      "loss": 46.0,
-      "step": 30665
-    },
-    {
-      "epoch": 2.3446298526291645,
-      "grad_norm": 0.004053534008562565,
-      "learning_rate": 0.00019999728897361624,
-      "loss": 46.0,
-      "step": 30666
-    },
-    {
-      "epoch": 2.344706309612554,
-      "grad_norm": 0.0008521229028701782,
-      "learning_rate": 0.00019999728879674654,
-      "loss": 46.0,
-      "step": 30667
-    },
-    {
-      "epoch": 2.344782766595944,
-      "grad_norm": 0.00043917031143791974,
-      "learning_rate": 0.0001999972886198711,
-      "loss": 46.0,
-      "step": 30668
-    },
-    {
-      "epoch": 2.3448592235793337,
-      "grad_norm": 0.0017025087727233768,
-      "learning_rate": 0.0001999972884429899,
-      "loss": 46.0,
-      "step": 30669
-    },
-    {
-      "epoch": 2.3449356805627235,
-      "grad_norm": 0.0006187063409015536,
-      "learning_rate": 0.00019999728826610293,
-      "loss": 46.0,
-      "step": 30670
-    },
-    {
-      "epoch": 2.3450121375461133,
-      "grad_norm": 0.0034894514828920364,
-      "learning_rate": 0.00019999728808921016,
-      "loss": 46.0,
-      "step": 30671
-    },
-    {
-      "epoch": 2.345088594529503,
-      "grad_norm": 0.0006457815179601312,
-      "learning_rate": 0.00019999728791231166,
-      "loss": 46.0,
-      "step": 30672
-    },
-    {
-      "epoch": 2.3451650515128923,
-      "grad_norm": 0.0014341780915856361,
-      "learning_rate": 0.00019999728773540735,
-      "loss": 46.0,
-      "step": 30673
-    },
-    {
-      "epoch": 2.345241508496282,
-      "grad_norm": 0.003676380729302764,
-      "learning_rate": 0.00019999728755849732,
-      "loss": 46.0,
-      "step": 30674
-    },
-    {
-      "epoch": 2.345317965479672,
-      "grad_norm": 0.0005658547743223608,
-      "learning_rate": 0.00019999728738158147,
-      "loss": 46.0,
-      "step": 30675
-    },
-    {
-      "epoch": 2.3453944224630616,
-      "grad_norm": 0.0010032716672867537,
-      "learning_rate": 0.00019999728720465986,
-      "loss": 46.0,
-      "step": 30676
-    },
-    {
-      "epoch": 2.3454708794464514,
-      "grad_norm": 0.0024425850715488195,
-      "learning_rate": 0.00019999728702773252,
-      "loss": 46.0,
-      "step": 30677
-    },
-    {
-      "epoch": 2.345547336429841,
-      "grad_norm": 0.00108078692574054,
-      "learning_rate": 0.00019999728685079937,
-      "loss": 46.0,
-      "step": 30678
-    },
-    {
-      "epoch": 2.345623793413231,
-      "grad_norm": 0.0026095963548868895,
-      "learning_rate": 0.00019999728667386048,
-      "loss": 46.0,
-      "step": 30679
-    },
-    {
-      "epoch": 2.3457002503966207,
-      "grad_norm": 0.0005333885783329606,
-      "learning_rate": 0.00019999728649691578,
-      "loss": 46.0,
-      "step": 30680
-    },
-    {
-      "epoch": 2.3457767073800104,
-      "grad_norm": 0.006590796168893576,
-      "learning_rate": 0.00019999728631996534,
-      "loss": 46.0,
-      "step": 30681
-    },
-    {
-      "epoch": 2.3458531643634,
-      "grad_norm": 0.0011731353588402271,
-      "learning_rate": 0.00019999728614300915,
-      "loss": 46.0,
-      "step": 30682
-    },
-    {
-      "epoch": 2.34592962134679,
-      "grad_norm": 0.001045377692207694,
-      "learning_rate": 0.00019999728596604714,
-      "loss": 46.0,
-      "step": 30683
-    },
-    {
-      "epoch": 2.3460060783301797,
-      "grad_norm": 0.0006777209928259254,
-      "learning_rate": 0.0001999972857890794,
-      "loss": 46.0,
-      "step": 30684
-    },
-    {
-      "epoch": 2.346082535313569,
-      "grad_norm": 0.0013066051760688424,
-      "learning_rate": 0.0001999972856121059,
-      "loss": 46.0,
-      "step": 30685
-    },
-    {
-      "epoch": 2.346158992296959,
-      "grad_norm": 0.0022873030975461006,
-      "learning_rate": 0.0001999972854351266,
-      "loss": 46.0,
-      "step": 30686
-    },
-    {
-      "epoch": 2.3462354492803486,
-      "grad_norm": 0.000904332147911191,
-      "learning_rate": 0.00019999728525814153,
-      "loss": 46.0,
-      "step": 30687
-    },
-    {
-      "epoch": 2.3463119062637383,
-      "grad_norm": 0.0005471340846270323,
-      "learning_rate": 0.00019999728508115073,
-      "loss": 46.0,
-      "step": 30688
-    },
-    {
-      "epoch": 2.346388363247128,
-      "grad_norm": 0.000711886037606746,
-      "learning_rate": 0.00019999728490415413,
-      "loss": 46.0,
-      "step": 30689
-    },
-    {
-      "epoch": 2.346464820230518,
-      "grad_norm": 0.001956287072971463,
-      "learning_rate": 0.00019999728472715176,
-      "loss": 46.0,
-      "step": 30690
-    },
-    {
-      "epoch": 2.3465412772139076,
-      "grad_norm": 0.001109390170313418,
-      "learning_rate": 0.0001999972845501436,
-      "loss": 46.0,
-      "step": 30691
-    },
-    {
-      "epoch": 2.3466177341972974,
-      "grad_norm": 0.0011955811642110348,
-      "learning_rate": 0.0001999972843731297,
-      "loss": 46.0,
-      "step": 30692
-    },
-    {
-      "epoch": 2.346694191180687,
-      "grad_norm": 0.0011027324944734573,
-      "learning_rate": 0.00019999728419611002,
-      "loss": 46.0,
-      "step": 30693
-    },
-    {
-      "epoch": 2.346770648164077,
-      "grad_norm": 0.0012277988716959953,
-      "learning_rate": 0.0001999972840190846,
-      "loss": 46.0,
-      "step": 30694
-    },
-    {
-      "epoch": 2.346847105147466,
-      "grad_norm": 0.001902602263726294,
-      "learning_rate": 0.00019999728384205337,
-      "loss": 46.0,
-      "step": 30695
-    },
-    {
-      "epoch": 2.346923562130856,
-      "grad_norm": 0.0032249605283141136,
-      "learning_rate": 0.0001999972836650164,
-      "loss": 46.0,
-      "step": 30696
-    },
-    {
-      "epoch": 2.3470000191142457,
-      "grad_norm": 0.0006361831328831613,
-      "learning_rate": 0.00019999728348797365,
-      "loss": 46.0,
-      "step": 30697
-    },
-    {
-      "epoch": 2.3470764760976355,
-      "grad_norm": 0.001081528957001865,
-      "learning_rate": 0.0001999972833109251,
-      "loss": 46.0,
-      "step": 30698
-    },
-    {
-      "epoch": 2.3471529330810252,
-      "grad_norm": 0.0009883008897304535,
-      "learning_rate": 0.00019999728313387083,
-      "loss": 46.0,
-      "step": 30699
-    },
-    {
-      "epoch": 2.347229390064415,
-      "grad_norm": 0.0011008074507117271,
-      "learning_rate": 0.00019999728295681078,
-      "loss": 46.0,
-      "step": 30700
-    },
-    {
-      "epoch": 2.3473058470478048,
-      "grad_norm": 0.0003863166202791035,
-      "learning_rate": 0.00019999728277974495,
-      "loss": 46.0,
-      "step": 30701
-    },
-    {
-      "epoch": 2.3473823040311945,
-      "grad_norm": 0.0012528032530099154,
-      "learning_rate": 0.00019999728260267335,
-      "loss": 46.0,
-      "step": 30702
-    },
-    {
-      "epoch": 2.3474587610145843,
-      "grad_norm": 0.0021434552036225796,
-      "learning_rate": 0.00019999728242559597,
-      "loss": 46.0,
-      "step": 30703
-    },
-    {
-      "epoch": 2.347535217997974,
-      "grad_norm": 0.0015423644799739122,
-      "learning_rate": 0.00019999728224851283,
-      "loss": 46.0,
-      "step": 30704
-    },
-    {
-      "epoch": 2.347611674981364,
-      "grad_norm": 0.0014932374469935894,
-      "learning_rate": 0.0001999972820714239,
-      "loss": 46.0,
-      "step": 30705
-    },
-    {
-      "epoch": 2.3476881319647536,
-      "grad_norm": 0.0009682325180619955,
-      "learning_rate": 0.00019999728189432924,
-      "loss": 46.0,
-      "step": 30706
-    },
-    {
-      "epoch": 2.347764588948143,
-      "grad_norm": 0.0006935708224773407,
-      "learning_rate": 0.0001999972817172288,
-      "loss": 46.0,
-      "step": 30707
-    },
-    {
-      "epoch": 2.3478410459315326,
-      "grad_norm": 0.0032118032686412334,
-      "learning_rate": 0.00019999728154012258,
-      "loss": 46.0,
-      "step": 30708
-    },
-    {
-      "epoch": 2.3479175029149224,
-      "grad_norm": 0.005844541825354099,
-      "learning_rate": 0.00019999728136301062,
-      "loss": 46.0,
-      "step": 30709
-    },
-    {
-      "epoch": 2.347993959898312,
-      "grad_norm": 0.0003876216651406139,
-      "learning_rate": 0.00019999728118589284,
-      "loss": 46.0,
-      "step": 30710
-    },
-    {
-      "epoch": 2.348070416881702,
-      "grad_norm": 0.0008173697860911489,
-      "learning_rate": 0.00019999728100876933,
-      "loss": 46.0,
-      "step": 30711
-    },
-    {
-      "epoch": 2.3481468738650917,
-      "grad_norm": 0.0008290798286907375,
-      "learning_rate": 0.00019999728083164005,
-      "loss": 46.0,
-      "step": 30712
-    },
-    {
-      "epoch": 2.3482233308484814,
-      "grad_norm": 0.0008743872749619186,
-      "learning_rate": 0.00019999728065450497,
-      "loss": 46.0,
-      "step": 30713
-    },
-    {
-      "epoch": 2.348299787831871,
-      "grad_norm": 0.0017585615860298276,
-      "learning_rate": 0.00019999728047736414,
-      "loss": 46.0,
-      "step": 30714
-    },
-    {
-      "epoch": 2.348376244815261,
-      "grad_norm": 0.0006434962269850075,
-      "learning_rate": 0.00019999728030021757,
-      "loss": 46.0,
-      "step": 30715
-    },
-    {
-      "epoch": 2.3484527017986503,
-      "grad_norm": 0.0011547869071364403,
-      "learning_rate": 0.00019999728012306517,
-      "loss": 46.0,
-      "step": 30716
-    },
-    {
-      "epoch": 2.34852915878204,
-      "grad_norm": 0.004181582015007734,
-      "learning_rate": 0.00019999727994590705,
-      "loss": 46.0,
-      "step": 30717
-    },
-    {
-      "epoch": 2.34860561576543,
-      "grad_norm": 0.002044013002887368,
-      "learning_rate": 0.00019999727976874313,
-      "loss": 46.0,
-      "step": 30718
-    },
-    {
-      "epoch": 2.3486820727488196,
-      "grad_norm": 0.0012599320616573095,
-      "learning_rate": 0.00019999727959157346,
-      "loss": 46.0,
-      "step": 30719
-    },
-    {
-      "epoch": 2.3487585297322093,
-      "grad_norm": 0.0017522042617201805,
-      "learning_rate": 0.00019999727941439802,
-      "loss": 46.0,
-      "step": 30720
-    },
-    {
-      "epoch": 2.348834986715599,
-      "grad_norm": 0.0011873355833813548,
-      "learning_rate": 0.0001999972792372168,
-      "loss": 46.0,
-      "step": 30721
-    },
-    {
-      "epoch": 2.348911443698989,
-      "grad_norm": 0.002941085258498788,
-      "learning_rate": 0.00019999727906002982,
-      "loss": 46.0,
-      "step": 30722
-    },
-    {
-      "epoch": 2.3489879006823786,
-      "grad_norm": 0.006120025645941496,
-      "learning_rate": 0.00019999727888283706,
-      "loss": 46.0,
-      "step": 30723
-    },
-    {
-      "epoch": 2.3490643576657684,
-      "grad_norm": 0.0012597902677953243,
-      "learning_rate": 0.00019999727870563853,
-      "loss": 46.0,
-      "step": 30724
-    },
-    {
-      "epoch": 2.349140814649158,
-      "grad_norm": 0.0005356183391995728,
-      "learning_rate": 0.00019999727852843425,
-      "loss": 46.0,
-      "step": 30725
-    },
-    {
-      "epoch": 2.349217271632548,
-      "grad_norm": 0.000579779502004385,
-      "learning_rate": 0.0001999972783512242,
-      "loss": 46.0,
-      "step": 30726
-    },
-    {
-      "epoch": 2.3492937286159377,
-      "grad_norm": 0.0021668465342372656,
-      "learning_rate": 0.00019999727817400834,
-      "loss": 46.0,
-      "step": 30727
-    },
-    {
-      "epoch": 2.3493701855993274,
-      "grad_norm": 0.00046065717469900846,
-      "learning_rate": 0.00019999727799678674,
-      "loss": 46.0,
-      "step": 30728
-    },
-    {
-      "epoch": 2.3494466425827167,
-      "grad_norm": 0.0023027784191071987,
-      "learning_rate": 0.0001999972778195594,
-      "loss": 46.0,
-      "step": 30729
-    },
-    {
-      "epoch": 2.3495230995661065,
-      "grad_norm": 0.0004900216590613127,
-      "learning_rate": 0.00019999727764232622,
-      "loss": 46.0,
-      "step": 30730
-    },
-    {
-      "epoch": 2.3495995565494963,
-      "grad_norm": 0.0006098304875195026,
-      "learning_rate": 0.00019999727746508735,
-      "loss": 46.0,
-      "step": 30731
-    },
-    {
-      "epoch": 2.349676013532886,
-      "grad_norm": 0.0008547514444217086,
-      "learning_rate": 0.00019999727728784266,
-      "loss": 46.0,
-      "step": 30732
-    },
-    {
-      "epoch": 2.349752470516276,
-      "grad_norm": 0.0005831779562868178,
-      "learning_rate": 0.0001999972771105922,
-      "loss": 46.0,
-      "step": 30733
-    },
-    {
-      "epoch": 2.3498289274996655,
-      "grad_norm": 0.0009844928281381726,
-      "learning_rate": 0.00019999727693333598,
-      "loss": 46.0,
-      "step": 30734
-    },
-    {
-      "epoch": 2.3499053844830553,
-      "grad_norm": 0.0020535937510430813,
-      "learning_rate": 0.000199997276756074,
-      "loss": 46.0,
-      "step": 30735
-    },
-    {
-      "epoch": 2.349981841466445,
-      "grad_norm": 0.0007150209858082235,
-      "learning_rate": 0.00019999727657880626,
-      "loss": 46.0,
-      "step": 30736
-    },
-    {
-      "epoch": 2.350058298449835,
-      "grad_norm": 0.001040019909851253,
-      "learning_rate": 0.00019999727640153273,
-      "loss": 46.0,
-      "step": 30737
-    },
-    {
-      "epoch": 2.350134755433224,
-      "grad_norm": 0.003776006633415818,
-      "learning_rate": 0.00019999727622425345,
-      "loss": 46.0,
-      "step": 30738
-    },
-    {
-      "epoch": 2.350211212416614,
-      "grad_norm": 0.0020182861480861902,
-      "learning_rate": 0.0001999972760469684,
-      "loss": 46.0,
-      "step": 30739
-    },
-    {
-      "epoch": 2.3502876694000037,
-      "grad_norm": 0.0010680132545530796,
-      "learning_rate": 0.00019999727586967754,
-      "loss": 46.0,
-      "step": 30740
-    },
-    {
-      "epoch": 2.3503641263833934,
-      "grad_norm": 0.0028938716277480125,
-      "learning_rate": 0.00019999727569238094,
-      "loss": 46.0,
-      "step": 30741
-    },
-    {
-      "epoch": 2.350440583366783,
-      "grad_norm": 0.009856638498604298,
-      "learning_rate": 0.00019999727551507857,
-      "loss": 46.0,
-      "step": 30742
-    },
-    {
-      "epoch": 2.350517040350173,
-      "grad_norm": 0.0007910510757938027,
-      "learning_rate": 0.00019999727533777045,
-      "loss": 46.0,
-      "step": 30743
-    },
-    {
-      "epoch": 2.3505934973335627,
-      "grad_norm": 0.0017557626124471426,
-      "learning_rate": 0.00019999727516045653,
-      "loss": 46.0,
-      "step": 30744
-    },
-    {
-      "epoch": 2.3506699543169525,
-      "grad_norm": 0.0006075821002013981,
-      "learning_rate": 0.00019999727498313684,
-      "loss": 46.0,
-      "step": 30745
-    },
-    {
-      "epoch": 2.3507464113003422,
-      "grad_norm": 0.0007803900516591966,
-      "learning_rate": 0.0001999972748058114,
-      "loss": 46.0,
-      "step": 30746
-    },
-    {
-      "epoch": 2.350822868283732,
-      "grad_norm": 0.0017525721341371536,
-      "learning_rate": 0.00019999727462848016,
-      "loss": 46.0,
-      "step": 30747
-    },
-    {
-      "epoch": 2.3508993252671218,
-      "grad_norm": 0.002435565460473299,
-      "learning_rate": 0.00019999727445114317,
-      "loss": 46.0,
-      "step": 30748
-    },
-    {
-      "epoch": 2.3509757822505115,
-      "grad_norm": 0.000667648680973798,
-      "learning_rate": 0.00019999727427380044,
-      "loss": 46.0,
-      "step": 30749
-    },
-    {
-      "epoch": 2.3510522392339013,
-      "grad_norm": 0.0017121753189712763,
-      "learning_rate": 0.00019999727409645193,
-      "loss": 46.0,
-      "step": 30750
-    },
-    {
-      "epoch": 2.3511286962172906,
-      "grad_norm": 0.0009407647303305566,
-      "learning_rate": 0.00019999727391909763,
-      "loss": 46.0,
-      "step": 30751
-    },
-    {
-      "epoch": 2.3512051532006804,
-      "grad_norm": 0.0024833460338413715,
-      "learning_rate": 0.00019999727374173755,
-      "loss": 46.0,
-      "step": 30752
-    },
-    {
-      "epoch": 2.35128161018407,
-      "grad_norm": 0.001971642719581723,
-      "learning_rate": 0.00019999727356437172,
-      "loss": 46.0,
-      "step": 30753
-    },
-    {
-      "epoch": 2.35135806716746,
-      "grad_norm": 0.0005710643017664552,
-      "learning_rate": 0.00019999727338700012,
-      "loss": 46.0,
-      "step": 30754
-    },
-    {
-      "epoch": 2.3514345241508496,
-      "grad_norm": 0.0008880415698513389,
-      "learning_rate": 0.00019999727320962275,
-      "loss": 46.0,
-      "step": 30755
-    },
-    {
-      "epoch": 2.3515109811342394,
-      "grad_norm": 0.0020124532748013735,
-      "learning_rate": 0.0001999972730322396,
-      "loss": 46.0,
-      "step": 30756
-    },
-    {
-      "epoch": 2.351587438117629,
-      "grad_norm": 0.001457576989196241,
-      "learning_rate": 0.0001999972728548507,
-      "loss": 46.0,
-      "step": 30757
-    },
-    {
-      "epoch": 2.351663895101019,
-      "grad_norm": 0.001222244231030345,
-      "learning_rate": 0.00019999727267745602,
-      "loss": 46.0,
-      "step": 30758
-    },
-    {
-      "epoch": 2.3517403520844087,
-      "grad_norm": 0.00220334529876709,
-      "learning_rate": 0.00019999727250005558,
-      "loss": 46.0,
-      "step": 30759
-    },
-    {
-      "epoch": 2.351816809067798,
-      "grad_norm": 0.0007857642485760152,
-      "learning_rate": 0.00019999727232264936,
-      "loss": 46.0,
-      "step": 30760
-    },
-    {
-      "epoch": 2.3518932660511878,
-      "grad_norm": 0.0033274961169809103,
-      "learning_rate": 0.00019999727214523735,
-      "loss": 46.0,
-      "step": 30761
-    },
-    {
-      "epoch": 2.3519697230345775,
-      "grad_norm": 0.0010401670588180423,
-      "learning_rate": 0.00019999727196781962,
-      "loss": 46.0,
-      "step": 30762
-    },
-    {
-      "epoch": 2.3520461800179673,
-      "grad_norm": 0.0006546966033056378,
-      "learning_rate": 0.0001999972717903961,
-      "loss": 46.0,
-      "step": 30763
-    },
-    {
-      "epoch": 2.352122637001357,
-      "grad_norm": 0.001543274731375277,
-      "learning_rate": 0.00019999727161296678,
-      "loss": 46.0,
-      "step": 30764
-    },
-    {
-      "epoch": 2.352199093984747,
-      "grad_norm": 0.0011219036532565951,
-      "learning_rate": 0.00019999727143553173,
-      "loss": 46.0,
-      "step": 30765
-    },
-    {
-      "epoch": 2.3522755509681366,
-      "grad_norm": 0.0004796836001332849,
-      "learning_rate": 0.0001999972712580909,
-      "loss": 46.0,
-      "step": 30766
-    },
-    {
-      "epoch": 2.3523520079515263,
-      "grad_norm": 0.0017966593150049448,
-      "learning_rate": 0.00019999727108064428,
-      "loss": 46.0,
-      "step": 30767
-    },
-    {
-      "epoch": 2.352428464934916,
-      "grad_norm": 0.0005488636088557541,
-      "learning_rate": 0.0001999972709031919,
-      "loss": 46.0,
-      "step": 30768
-    },
-    {
-      "epoch": 2.352504921918306,
-      "grad_norm": 0.002760576317086816,
-      "learning_rate": 0.00019999727072573379,
-      "loss": 46.0,
-      "step": 30769
-    },
-    {
-      "epoch": 2.3525813789016956,
-      "grad_norm": 0.001007545622996986,
-      "learning_rate": 0.00019999727054826987,
-      "loss": 46.0,
-      "step": 30770
-    },
-    {
-      "epoch": 2.3526578358850854,
-      "grad_norm": 0.0010352828539907932,
-      "learning_rate": 0.00019999727037080017,
-      "loss": 46.0,
-      "step": 30771
-    },
-    {
-      "epoch": 2.3527342928684747,
-      "grad_norm": 0.0012859228299930692,
-      "learning_rate": 0.00019999727019332473,
-      "loss": 46.0,
-      "step": 30772
-    },
-    {
-      "epoch": 2.3528107498518644,
-      "grad_norm": 0.000690045184455812,
-      "learning_rate": 0.00019999727001584352,
-      "loss": 46.0,
-      "step": 30773
-    },
-    {
-      "epoch": 2.352887206835254,
-      "grad_norm": 0.002257169224321842,
-      "learning_rate": 0.00019999726983835654,
-      "loss": 46.0,
-      "step": 30774
-    },
-    {
-      "epoch": 2.352963663818644,
-      "grad_norm": 0.0005929278559051454,
-      "learning_rate": 0.00019999726966086375,
-      "loss": 46.0,
-      "step": 30775
-    },
-    {
-      "epoch": 2.3530401208020337,
-      "grad_norm": 0.0006345452857203782,
-      "learning_rate": 0.00019999726948336525,
-      "loss": 46.0,
-      "step": 30776
-    },
-    {
-      "epoch": 2.3531165777854235,
-      "grad_norm": 0.0011484071146696806,
-      "learning_rate": 0.00019999726930586094,
-      "loss": 46.0,
-      "step": 30777
-    },
-    {
-      "epoch": 2.3531930347688133,
-      "grad_norm": 0.00044060597429051995,
-      "learning_rate": 0.0001999972691283509,
-      "loss": 46.0,
-      "step": 30778
-    },
-    {
-      "epoch": 2.353269491752203,
-      "grad_norm": 0.0034412231761962175,
-      "learning_rate": 0.00019999726895083506,
-      "loss": 46.0,
-      "step": 30779
-    },
-    {
-      "epoch": 2.3533459487355928,
-      "grad_norm": 0.0036489868070930243,
-      "learning_rate": 0.00019999726877331344,
-      "loss": 46.0,
-      "step": 30780
-    },
-    {
-      "epoch": 2.3534224057189825,
-      "grad_norm": 0.000718778814189136,
-      "learning_rate": 0.00019999726859578606,
-      "loss": 46.0,
-      "step": 30781
-    },
-    {
-      "epoch": 2.353498862702372,
-      "grad_norm": 0.0014931521145626903,
-      "learning_rate": 0.00019999726841825292,
-      "loss": 46.0,
-      "step": 30782
-    },
-    {
-      "epoch": 2.3535753196857616,
-      "grad_norm": 0.001067790319211781,
-      "learning_rate": 0.00019999726824071403,
-      "loss": 46.0,
-      "step": 30783
-    },
-    {
-      "epoch": 2.3536517766691514,
-      "grad_norm": 0.0009619550546631217,
-      "learning_rate": 0.00019999726806316933,
-      "loss": 46.0,
-      "step": 30784
-    },
-    {
-      "epoch": 2.353728233652541,
-      "grad_norm": 0.0009539675083942711,
-      "learning_rate": 0.00019999726788561887,
-      "loss": 46.0,
-      "step": 30785
-    },
-    {
-      "epoch": 2.353804690635931,
-      "grad_norm": 0.0015302995452657342,
-      "learning_rate": 0.00019999726770806266,
-      "loss": 46.0,
-      "step": 30786
-    },
-    {
-      "epoch": 2.3538811476193207,
-      "grad_norm": 0.0007947874255478382,
-      "learning_rate": 0.00019999726753050067,
-      "loss": 46.0,
-      "step": 30787
-    },
-    {
-      "epoch": 2.3539576046027104,
-      "grad_norm": 0.0013851206749677658,
-      "learning_rate": 0.0001999972673529329,
-      "loss": 46.0,
-      "step": 30788
-    },
-    {
-      "epoch": 2.3540340615861,
-      "grad_norm": 0.003003871301189065,
-      "learning_rate": 0.00019999726717535938,
-      "loss": 46.0,
-      "step": 30789
-    },
-    {
-      "epoch": 2.35411051856949,
-      "grad_norm": 0.0026445782277733088,
-      "learning_rate": 0.0001999972669977801,
-      "loss": 46.0,
-      "step": 30790
-    },
-    {
-      "epoch": 2.3541869755528797,
-      "grad_norm": 0.0009551442926749587,
-      "learning_rate": 0.00019999726682019505,
-      "loss": 46.0,
-      "step": 30791
-    },
-    {
-      "epoch": 2.3542634325362695,
-      "grad_norm": 0.0006128668901510537,
-      "learning_rate": 0.0001999972666426042,
-      "loss": 46.0,
-      "step": 30792
-    },
-    {
-      "epoch": 2.3543398895196592,
-      "grad_norm": 0.011979698203504086,
-      "learning_rate": 0.0001999972664650076,
-      "loss": 46.0,
-      "step": 30793
-    },
-    {
-      "epoch": 2.3544163465030485,
-      "grad_norm": 0.0013513767626136541,
-      "learning_rate": 0.00019999726628740523,
-      "loss": 46.0,
-      "step": 30794
-    },
-    {
-      "epoch": 2.3544928034864383,
-      "grad_norm": 0.0025088328402489424,
-      "learning_rate": 0.00019999726610979708,
-      "loss": 46.0,
-      "step": 30795
-    },
-    {
-      "epoch": 2.354569260469828,
-      "grad_norm": 0.0006789755425415933,
-      "learning_rate": 0.00019999726593218316,
-      "loss": 46.0,
-      "step": 30796
-    },
-    {
-      "epoch": 2.354645717453218,
-      "grad_norm": 0.000706896185874939,
-      "learning_rate": 0.00019999726575456347,
-      "loss": 46.0,
-      "step": 30797
-    },
-    {
-      "epoch": 2.3547221744366076,
-      "grad_norm": 0.0007117788773030043,
-      "learning_rate": 0.00019999726557693803,
-      "loss": 46.0,
-      "step": 30798
-    },
-    {
-      "epoch": 2.3547986314199973,
-      "grad_norm": 0.0007525378605350852,
-      "learning_rate": 0.00019999726539930682,
-      "loss": 46.0,
-      "step": 30799
-    },
-    {
-      "epoch": 2.354875088403387,
-      "grad_norm": 0.0009500780142843723,
-      "learning_rate": 0.0001999972652216698,
-      "loss": 46.0,
-      "step": 30800
-    },
-    {
-      "epoch": 2.354951545386777,
-      "grad_norm": 0.004723402205854654,
-      "learning_rate": 0.00019999726504402708,
-      "loss": 46.0,
-      "step": 30801
-    },
-    {
-      "epoch": 2.3550280023701666,
-      "grad_norm": 0.0009589608525857329,
-      "learning_rate": 0.00019999726486637852,
-      "loss": 46.0,
-      "step": 30802
-    },
-    {
-      "epoch": 2.3551044593535564,
-      "grad_norm": 0.0011295626172795892,
-      "learning_rate": 0.00019999726468872422,
-      "loss": 46.0,
-      "step": 30803
-    },
-    {
-      "epoch": 2.3551809163369457,
-      "grad_norm": 0.0011377278715372086,
-      "learning_rate": 0.00019999726451106416,
-      "loss": 46.0,
-      "step": 30804
-    },
-    {
-      "epoch": 2.3552573733203355,
-      "grad_norm": 0.007689039688557386,
-      "learning_rate": 0.0001999972643333983,
-      "loss": 46.0,
-      "step": 30805
-    },
-    {
-      "epoch": 2.3553338303037252,
-      "grad_norm": 0.0011764847440645099,
-      "learning_rate": 0.00019999726415572671,
-      "loss": 46.0,
-      "step": 30806
-    },
-    {
-      "epoch": 2.355410287287115,
-      "grad_norm": 0.0006372727802954614,
-      "learning_rate": 0.00019999726397804934,
-      "loss": 46.0,
-      "step": 30807
-    },
-    {
-      "epoch": 2.3554867442705048,
-      "grad_norm": 0.004856351763010025,
-      "learning_rate": 0.0001999972638003662,
-      "loss": 46.0,
-      "step": 30808
-    },
-    {
-      "epoch": 2.3555632012538945,
-      "grad_norm": 0.001070099649950862,
-      "learning_rate": 0.00019999726362267728,
-      "loss": 46.0,
-      "step": 30809
-    },
-    {
-      "epoch": 2.3556396582372843,
-      "grad_norm": 0.0009453344391658902,
-      "learning_rate": 0.00019999726344498262,
-      "loss": 46.0,
-      "step": 30810
-    },
-    {
-      "epoch": 2.355716115220674,
-      "grad_norm": 0.0009171488345600665,
-      "learning_rate": 0.00019999726326728215,
-      "loss": 46.0,
-      "step": 30811
-    },
-    {
-      "epoch": 2.355792572204064,
-      "grad_norm": 0.003414018778130412,
-      "learning_rate": 0.00019999726308957594,
-      "loss": 46.0,
-      "step": 30812
-    },
-    {
-      "epoch": 2.3558690291874536,
-      "grad_norm": 0.0006449449574574828,
-      "learning_rate": 0.00019999726291186395,
-      "loss": 46.0,
-      "step": 30813
-    },
-    {
-      "epoch": 2.3559454861708433,
-      "grad_norm": 0.0008448041626252234,
-      "learning_rate": 0.0001999972627341462,
-      "loss": 46.0,
-      "step": 30814
-    },
-    {
-      "epoch": 2.356021943154233,
-      "grad_norm": 0.0008702003979124129,
-      "learning_rate": 0.00019999726255642266,
-      "loss": 46.0,
-      "step": 30815
-    },
-    {
-      "epoch": 2.3560984001376224,
-      "grad_norm": 0.0020965328440070152,
-      "learning_rate": 0.00019999726237869336,
-      "loss": 46.0,
-      "step": 30816
-    },
-    {
-      "epoch": 2.356174857121012,
-      "grad_norm": 0.0018821465782821178,
-      "learning_rate": 0.00019999726220095828,
-      "loss": 46.0,
-      "step": 30817
-    },
-    {
-      "epoch": 2.356251314104402,
-      "grad_norm": 0.002290338510647416,
-      "learning_rate": 0.00019999726202321746,
-      "loss": 46.0,
-      "step": 30818
-    },
-    {
-      "epoch": 2.3563277710877917,
-      "grad_norm": 0.0006560833426192403,
-      "learning_rate": 0.00019999726184547086,
-      "loss": 46.0,
-      "step": 30819
-    },
-    {
-      "epoch": 2.3564042280711814,
-      "grad_norm": 0.0010972610907629132,
-      "learning_rate": 0.00019999726166771846,
-      "loss": 46.0,
-      "step": 30820
-    },
-    {
-      "epoch": 2.356480685054571,
-      "grad_norm": 0.0018406495219096541,
-      "learning_rate": 0.00019999726148996034,
-      "loss": 46.0,
-      "step": 30821
-    },
-    {
-      "epoch": 2.356557142037961,
-      "grad_norm": 0.0008733987342566252,
-      "learning_rate": 0.0001999972613121964,
-      "loss": 46.0,
-      "step": 30822
-    },
-    {
-      "epoch": 2.3566335990213507,
-      "grad_norm": 0.0029574641957879066,
-      "learning_rate": 0.00019999726113442673,
-      "loss": 46.0,
-      "step": 30823
-    },
-    {
-      "epoch": 2.3567100560047405,
-      "grad_norm": 0.0005557197728194296,
-      "learning_rate": 0.00019999726095665127,
-      "loss": 46.0,
-      "step": 30824
-    },
-    {
-      "epoch": 2.3567865129881302,
-      "grad_norm": 0.0033350992016494274,
-      "learning_rate": 0.00019999726077887003,
-      "loss": 46.0,
-      "step": 30825
-    },
-    {
-      "epoch": 2.3568629699715196,
-      "grad_norm": 0.0005353637970983982,
-      "learning_rate": 0.00019999726060108305,
-      "loss": 46.0,
-      "step": 30826
-    },
-    {
-      "epoch": 2.3569394269549093,
-      "grad_norm": 0.000887516827788204,
-      "learning_rate": 0.0001999972604232903,
-      "loss": 46.0,
-      "step": 30827
-    },
-    {
-      "epoch": 2.357015883938299,
-      "grad_norm": 0.0009481757879257202,
-      "learning_rate": 0.00019999726024549176,
-      "loss": 46.0,
-      "step": 30828
-    },
-    {
-      "epoch": 2.357092340921689,
-      "grad_norm": 0.005767423659563065,
-      "learning_rate": 0.00019999726006768745,
-      "loss": 46.0,
-      "step": 30829
-    },
-    {
-      "epoch": 2.3571687979050786,
-      "grad_norm": 0.0010970557341352105,
-      "learning_rate": 0.00019999725988987738,
-      "loss": 46.0,
-      "step": 30830
-    },
-    {
-      "epoch": 2.3572452548884684,
-      "grad_norm": 0.007884407415986061,
-      "learning_rate": 0.00019999725971206158,
-      "loss": 46.0,
-      "step": 30831
-    },
-    {
-      "epoch": 2.357321711871858,
-      "grad_norm": 0.0010746211046352983,
-      "learning_rate": 0.00019999725953423995,
-      "loss": 46.0,
-      "step": 30832
-    },
-    {
-      "epoch": 2.357398168855248,
-      "grad_norm": 0.0007399875321425498,
-      "learning_rate": 0.00019999725935641258,
-      "loss": 46.0,
-      "step": 30833
-    },
-    {
-      "epoch": 2.3574746258386376,
-      "grad_norm": 0.0010197452502325177,
-      "learning_rate": 0.00019999725917857944,
-      "loss": 46.0,
-      "step": 30834
-    },
-    {
-      "epoch": 2.3575510828220274,
-      "grad_norm": 0.0006012238445691764,
-      "learning_rate": 0.00019999725900074052,
-      "loss": 46.0,
-      "step": 30835
-    },
-    {
-      "epoch": 2.357627539805417,
-      "grad_norm": 0.0015155320288613439,
-      "learning_rate": 0.00019999725882289583,
-      "loss": 46.0,
-      "step": 30836
-    },
-    {
-      "epoch": 2.357703996788807,
-      "grad_norm": 0.0023198137059807777,
-      "learning_rate": 0.00019999725864504537,
-      "loss": 46.0,
-      "step": 30837
-    },
-    {
-      "epoch": 2.3577804537721962,
-      "grad_norm": 0.0015870064962655306,
-      "learning_rate": 0.00019999725846718916,
-      "loss": 46.0,
-      "step": 30838
-    },
-    {
-      "epoch": 2.357856910755586,
-      "grad_norm": 0.000814481230918318,
-      "learning_rate": 0.00019999725828932717,
-      "loss": 46.0,
-      "step": 30839
-    },
-    {
-      "epoch": 2.3579333677389758,
-      "grad_norm": 0.0036859349347651005,
-      "learning_rate": 0.0001999972581114594,
-      "loss": 46.0,
-      "step": 30840
-    },
-    {
-      "epoch": 2.3580098247223655,
-      "grad_norm": 0.0014893336920067668,
-      "learning_rate": 0.00019999725793358588,
-      "loss": 46.0,
-      "step": 30841
-    },
-    {
-      "epoch": 2.3580862817057553,
-      "grad_norm": 0.0007407942903228104,
-      "learning_rate": 0.00019999725775570658,
-      "loss": 46.0,
-      "step": 30842
-    },
-    {
-      "epoch": 2.358162738689145,
-      "grad_norm": 0.0024777348153293133,
-      "learning_rate": 0.0001999972575778215,
-      "loss": 46.0,
-      "step": 30843
-    },
-    {
-      "epoch": 2.358239195672535,
-      "grad_norm": 0.0009335758513770998,
-      "learning_rate": 0.00019999725739993068,
-      "loss": 46.0,
-      "step": 30844
-    },
-    {
-      "epoch": 2.3583156526559246,
-      "grad_norm": 0.0009884078754112124,
-      "learning_rate": 0.00019999725722203405,
-      "loss": 46.0,
-      "step": 30845
-    },
-    {
-      "epoch": 2.3583921096393143,
-      "grad_norm": 0.002203387673944235,
-      "learning_rate": 0.00019999725704413168,
-      "loss": 46.0,
-      "step": 30846
-    },
-    {
-      "epoch": 2.3584685666227037,
-      "grad_norm": 0.0014249550877138972,
-      "learning_rate": 0.00019999725686622354,
-      "loss": 46.0,
-      "step": 30847
-    },
-    {
-      "epoch": 2.3585450236060934,
-      "grad_norm": 0.0006944107008166611,
-      "learning_rate": 0.00019999725668830962,
-      "loss": 46.0,
-      "step": 30848
-    },
-    {
-      "epoch": 2.358621480589483,
-      "grad_norm": 0.0008494915091432631,
-      "learning_rate": 0.00019999725651038993,
-      "loss": 46.0,
-      "step": 30849
-    },
-    {
-      "epoch": 2.358697937572873,
-      "grad_norm": 0.001505842199549079,
-      "learning_rate": 0.0001999972563324645,
-      "loss": 46.0,
-      "step": 30850
-    },
-    {
-      "epoch": 2.3587743945562627,
-      "grad_norm": 0.0015603555366396904,
-      "learning_rate": 0.00019999725615453326,
-      "loss": 46.0,
-      "step": 30851
-    },
-    {
-      "epoch": 2.3588508515396525,
-      "grad_norm": 0.0012774834176525474,
-      "learning_rate": 0.00019999725597659627,
-      "loss": 46.0,
-      "step": 30852
-    },
-    {
-      "epoch": 2.358927308523042,
-      "grad_norm": 0.006070606876164675,
-      "learning_rate": 0.0001999972557986535,
-      "loss": 46.0,
-      "step": 30853
-    },
-    {
-      "epoch": 2.359003765506432,
-      "grad_norm": 0.0010035528102889657,
-      "learning_rate": 0.00019999725562070496,
-      "loss": 46.0,
-      "step": 30854
-    },
-    {
-      "epoch": 2.3590802224898217,
-      "grad_norm": 0.0014161423314362764,
-      "learning_rate": 0.00019999725544275068,
-      "loss": 46.0,
-      "step": 30855
-    },
-    {
-      "epoch": 2.3591566794732115,
-      "grad_norm": 0.0005423652473837137,
-      "learning_rate": 0.00019999725526479058,
-      "loss": 46.0,
-      "step": 30856
-    },
-    {
-      "epoch": 2.3592331364566013,
-      "grad_norm": 0.0007155163330025971,
-      "learning_rate": 0.00019999725508682475,
-      "loss": 46.0,
-      "step": 30857
-    },
-    {
-      "epoch": 2.359309593439991,
-      "grad_norm": 0.001379706314764917,
-      "learning_rate": 0.00019999725490885316,
-      "loss": 46.0,
-      "step": 30858
-    },
-    {
-      "epoch": 2.359386050423381,
-      "grad_norm": 0.0008177123381756246,
-      "learning_rate": 0.00019999725473087576,
-      "loss": 46.0,
-      "step": 30859
-    },
-    {
-      "epoch": 2.35946250740677,
-      "grad_norm": 0.0007992482278496027,
-      "learning_rate": 0.00019999725455289262,
-      "loss": 46.0,
-      "step": 30860
-    },
-    {
-      "epoch": 2.35953896439016,
-      "grad_norm": 0.0012482131132856011,
-      "learning_rate": 0.0001999972543749037,
-      "loss": 46.0,
-      "step": 30861
-    },
-    {
-      "epoch": 2.3596154213735496,
-      "grad_norm": 0.0006626953254453838,
-      "learning_rate": 0.00019999725419690904,
-      "loss": 46.0,
-      "step": 30862
-    },
-    {
-      "epoch": 2.3596918783569394,
-      "grad_norm": 0.0007058997871354222,
-      "learning_rate": 0.00019999725401890857,
-      "loss": 46.0,
-      "step": 30863
-    },
-    {
-      "epoch": 2.359768335340329,
-      "grad_norm": 0.0005448069423437119,
-      "learning_rate": 0.00019999725384090233,
-      "loss": 46.0,
-      "step": 30864
-    },
-    {
-      "epoch": 2.359844792323719,
-      "grad_norm": 0.002846070798113942,
-      "learning_rate": 0.00019999725366289035,
-      "loss": 46.0,
-      "step": 30865
-    },
-    {
-      "epoch": 2.3599212493071087,
-      "grad_norm": 0.0016631190665066242,
-      "learning_rate": 0.0001999972534848726,
-      "loss": 46.0,
-      "step": 30866
-    },
-    {
-      "epoch": 2.3599977062904984,
-      "grad_norm": 0.0006251556915231049,
-      "learning_rate": 0.00019999725330684904,
-      "loss": 46.0,
-      "step": 30867
-    },
-    {
-      "epoch": 2.360074163273888,
-      "grad_norm": 0.0035454677417874336,
-      "learning_rate": 0.00019999725312881976,
-      "loss": 46.0,
-      "step": 30868
-    },
-    {
-      "epoch": 2.3601506202572775,
-      "grad_norm": 0.0012839263072237372,
-      "learning_rate": 0.00019999725295078466,
-      "loss": 46.0,
-      "step": 30869
-    },
-    {
-      "epoch": 2.3602270772406673,
-      "grad_norm": 0.0020389000419527292,
-      "learning_rate": 0.00019999725277274383,
-      "loss": 46.0,
-      "step": 30870
-    },
-    {
-      "epoch": 2.360303534224057,
-      "grad_norm": 0.0011894642375409603,
-      "learning_rate": 0.00019999725259469724,
-      "loss": 46.0,
-      "step": 30871
-    },
-    {
-      "epoch": 2.360379991207447,
-      "grad_norm": 0.002931153168901801,
-      "learning_rate": 0.00019999725241664487,
-      "loss": 46.0,
-      "step": 30872
-    },
-    {
-      "epoch": 2.3604564481908366,
-      "grad_norm": 0.0009152429993264377,
-      "learning_rate": 0.0001999972522385867,
-      "loss": 46.0,
-      "step": 30873
-    },
-    {
-      "epoch": 2.3605329051742263,
-      "grad_norm": 0.003977891523391008,
-      "learning_rate": 0.0001999972520605228,
-      "loss": 46.0,
-      "step": 30874
-    },
-    {
-      "epoch": 2.360609362157616,
-      "grad_norm": 0.0004825768992304802,
-      "learning_rate": 0.00019999725188245312,
-      "loss": 46.0,
-      "step": 30875
-    },
-    {
-      "epoch": 2.360685819141006,
-      "grad_norm": 0.0015788174932822585,
-      "learning_rate": 0.00019999725170437765,
-      "loss": 46.0,
-      "step": 30876
-    },
-    {
-      "epoch": 2.3607622761243956,
-      "grad_norm": 0.006124661769717932,
-      "learning_rate": 0.0001999972515262964,
-      "loss": 46.0,
-      "step": 30877
-    },
-    {
-      "epoch": 2.3608387331077854,
-      "grad_norm": 0.004751675296574831,
-      "learning_rate": 0.00019999725134820943,
-      "loss": 46.0,
-      "step": 30878
-    },
-    {
-      "epoch": 2.360915190091175,
-      "grad_norm": 0.0005537885590456426,
-      "learning_rate": 0.00019999725117011665,
-      "loss": 46.0,
-      "step": 30879
-    },
-    {
-      "epoch": 2.360991647074565,
-      "grad_norm": 0.002137881936505437,
-      "learning_rate": 0.00019999725099201815,
-      "loss": 46.0,
-      "step": 30880
-    },
-    {
-      "epoch": 2.3610681040579546,
-      "grad_norm": 0.0030518837738782167,
-      "learning_rate": 0.00019999725081391382,
-      "loss": 46.0,
-      "step": 30881
-    },
-    {
-      "epoch": 2.361144561041344,
-      "grad_norm": 0.0020048003643751144,
-      "learning_rate": 0.00019999725063580374,
-      "loss": 46.0,
-      "step": 30882
-    },
-    {
-      "epoch": 2.3612210180247337,
-      "grad_norm": 0.0003562805359251797,
-      "learning_rate": 0.00019999725045768792,
-      "loss": 46.0,
-      "step": 30883
-    },
-    {
-      "epoch": 2.3612974750081235,
-      "grad_norm": 0.0008817478665150702,
-      "learning_rate": 0.0001999972502795663,
-      "loss": 46.0,
-      "step": 30884
-    },
-    {
-      "epoch": 2.3613739319915132,
-      "grad_norm": 0.001112551661208272,
-      "learning_rate": 0.00019999725010143892,
-      "loss": 46.0,
-      "step": 30885
-    },
-    {
-      "epoch": 2.361450388974903,
-      "grad_norm": 0.002130249049514532,
-      "learning_rate": 0.00019999724992330578,
-      "loss": 46.0,
-      "step": 30886
-    },
-    {
-      "epoch": 2.3615268459582928,
-      "grad_norm": 0.002317998791113496,
-      "learning_rate": 0.00019999724974516684,
-      "loss": 46.0,
-      "step": 30887
-    },
-    {
-      "epoch": 2.3616033029416825,
-      "grad_norm": 0.0006509704398922622,
-      "learning_rate": 0.00019999724956702215,
-      "loss": 46.0,
-      "step": 30888
-    },
-    {
-      "epoch": 2.3616797599250723,
-      "grad_norm": 0.0009209293057210743,
-      "learning_rate": 0.0001999972493888717,
-      "loss": 46.0,
-      "step": 30889
-    },
-    {
-      "epoch": 2.361756216908462,
-      "grad_norm": 0.003643839852884412,
-      "learning_rate": 0.00019999724921071548,
-      "loss": 46.0,
-      "step": 30890
-    },
-    {
-      "epoch": 2.3618326738918514,
-      "grad_norm": 0.002169376937672496,
-      "learning_rate": 0.0001999972490325535,
-      "loss": 46.0,
-      "step": 30891
-    },
-    {
-      "epoch": 2.361909130875241,
-      "grad_norm": 0.000851595017593354,
-      "learning_rate": 0.0001999972488543857,
-      "loss": 46.0,
-      "step": 30892
-    },
-    {
-      "epoch": 2.361985587858631,
-      "grad_norm": 0.003256110940128565,
-      "learning_rate": 0.00019999724867621218,
-      "loss": 46.0,
-      "step": 30893
-    },
-    {
-      "epoch": 2.3620620448420206,
-      "grad_norm": 0.002151092980057001,
-      "learning_rate": 0.0001999972484980329,
-      "loss": 46.0,
-      "step": 30894
-    },
-    {
-      "epoch": 2.3621385018254104,
-      "grad_norm": 0.0015356120420619845,
-      "learning_rate": 0.0001999972483198478,
-      "loss": 46.0,
-      "step": 30895
-    },
-    {
-      "epoch": 2.3622149588088,
-      "grad_norm": 0.0007646828889846802,
-      "learning_rate": 0.00019999724814165698,
-      "loss": 46.0,
-      "step": 30896
-    },
-    {
-      "epoch": 2.36229141579219,
-      "grad_norm": 0.0017656392883509398,
-      "learning_rate": 0.00019999724796346036,
-      "loss": 46.0,
-      "step": 30897
-    },
-    {
-      "epoch": 2.3623678727755797,
-      "grad_norm": 0.0050153667107224464,
-      "learning_rate": 0.00019999724778525796,
-      "loss": 46.0,
-      "step": 30898
-    },
-    {
-      "epoch": 2.3624443297589695,
-      "grad_norm": 0.0005469491006806493,
-      "learning_rate": 0.00019999724760704984,
-      "loss": 46.0,
-      "step": 30899
-    },
-    {
-      "epoch": 2.362520786742359,
-      "grad_norm": 0.0013224628055468202,
-      "learning_rate": 0.0001999972474288359,
-      "loss": 46.0,
-      "step": 30900
-    },
-    {
-      "epoch": 2.362597243725749,
-      "grad_norm": 0.0009792111814022064,
-      "learning_rate": 0.0001999972472506162,
-      "loss": 46.0,
-      "step": 30901
-    },
-    {
-      "epoch": 2.3626737007091387,
-      "grad_norm": 0.004378196317702532,
-      "learning_rate": 0.00019999724707239075,
-      "loss": 46.0,
-      "step": 30902
-    },
-    {
-      "epoch": 2.362750157692528,
-      "grad_norm": 0.00042918085819110274,
-      "learning_rate": 0.00019999724689415954,
-      "loss": 46.0,
-      "step": 30903
-    },
-    {
-      "epoch": 2.362826614675918,
-      "grad_norm": 0.0004521795781329274,
-      "learning_rate": 0.00019999724671592253,
-      "loss": 46.0,
-      "step": 30904
-    },
-    {
-      "epoch": 2.3629030716593076,
-      "grad_norm": 0.002102494239807129,
-      "learning_rate": 0.00019999724653767977,
-      "loss": 46.0,
-      "step": 30905
-    },
-    {
-      "epoch": 2.3629795286426973,
-      "grad_norm": 0.006573577877134085,
-      "learning_rate": 0.00019999724635943124,
-      "loss": 46.0,
-      "step": 30906
-    },
-    {
-      "epoch": 2.363055985626087,
-      "grad_norm": 0.001122870366089046,
-      "learning_rate": 0.00019999724618117694,
-      "loss": 46.0,
-      "step": 30907
-    },
-    {
-      "epoch": 2.363132442609477,
-      "grad_norm": 0.0008915234357118607,
-      "learning_rate": 0.00019999724600291687,
-      "loss": 46.0,
-      "step": 30908
-    },
-    {
-      "epoch": 2.3632088995928666,
-      "grad_norm": 0.004498935304582119,
-      "learning_rate": 0.00019999724582465102,
-      "loss": 46.0,
-      "step": 30909
-    },
-    {
-      "epoch": 2.3632853565762564,
-      "grad_norm": 0.0027440243866294622,
-      "learning_rate": 0.0001999972456463794,
-      "loss": 46.0,
-      "step": 30910
-    },
-    {
-      "epoch": 2.363361813559646,
-      "grad_norm": 0.005329929757863283,
-      "learning_rate": 0.00019999724546810202,
-      "loss": 46.0,
-      "step": 30911
-    },
-    {
-      "epoch": 2.363438270543036,
-      "grad_norm": 0.002080376725643873,
-      "learning_rate": 0.00019999724528981888,
-      "loss": 46.0,
-      "step": 30912
-    },
-    {
-      "epoch": 2.363514727526425,
-      "grad_norm": 0.0010124585824087262,
-      "learning_rate": 0.00019999724511152994,
-      "loss": 46.0,
-      "step": 30913
-    },
-    {
-      "epoch": 2.363591184509815,
-      "grad_norm": 0.0005254840361885726,
-      "learning_rate": 0.00019999724493323525,
-      "loss": 46.0,
-      "step": 30914
-    },
-    {
-      "epoch": 2.3636676414932047,
-      "grad_norm": 0.0017286852234974504,
-      "learning_rate": 0.0001999972447549348,
-      "loss": 46.0,
-      "step": 30915
-    },
-    {
-      "epoch": 2.3637440984765945,
-      "grad_norm": 0.0022400347515940666,
-      "learning_rate": 0.00019999724457662858,
-      "loss": 46.0,
-      "step": 30916
-    },
-    {
-      "epoch": 2.3638205554599843,
-      "grad_norm": 0.0013412097468972206,
-      "learning_rate": 0.00019999724439831657,
-      "loss": 46.0,
-      "step": 30917
-    },
-    {
-      "epoch": 2.363897012443374,
-      "grad_norm": 0.001063137548044324,
-      "learning_rate": 0.0001999972442199988,
-      "loss": 46.0,
-      "step": 30918
-    },
-    {
-      "epoch": 2.363973469426764,
-      "grad_norm": 0.0012534894049167633,
-      "learning_rate": 0.00019999724404167526,
-      "loss": 46.0,
-      "step": 30919
-    },
-    {
-      "epoch": 2.3640499264101535,
-      "grad_norm": 0.001384580391459167,
-      "learning_rate": 0.00019999724386334596,
-      "loss": 46.0,
-      "step": 30920
-    },
-    {
-      "epoch": 2.3641263833935433,
-      "grad_norm": 0.0021318020299077034,
-      "learning_rate": 0.00019999724368501088,
-      "loss": 46.0,
-      "step": 30921
-    },
-    {
-      "epoch": 2.364202840376933,
-      "grad_norm": 0.0035450602881610394,
-      "learning_rate": 0.00019999724350667006,
-      "loss": 46.0,
-      "step": 30922
-    },
-    {
-      "epoch": 2.364279297360323,
-      "grad_norm": 0.0009228177950717509,
-      "learning_rate": 0.00019999724332832344,
-      "loss": 46.0,
-      "step": 30923
-    },
-    {
-      "epoch": 2.3643557543437126,
-      "grad_norm": 0.0019502609502524137,
-      "learning_rate": 0.00019999724314997104,
-      "loss": 46.0,
-      "step": 30924
-    },
-    {
-      "epoch": 2.364432211327102,
-      "grad_norm": 0.0012009326601400971,
-      "learning_rate": 0.0001999972429716129,
-      "loss": 46.0,
-      "step": 30925
-    },
-    {
-      "epoch": 2.3645086683104917,
-      "grad_norm": 0.0007818982703611255,
-      "learning_rate": 0.00019999724279324898,
-      "loss": 46.0,
-      "step": 30926
-    },
-    {
-      "epoch": 2.3645851252938814,
-      "grad_norm": 0.0011191205121576786,
-      "learning_rate": 0.0001999972426148793,
-      "loss": 46.0,
-      "step": 30927
-    },
-    {
-      "epoch": 2.364661582277271,
-      "grad_norm": 0.00046191579895094037,
-      "learning_rate": 0.00019999724243650383,
-      "loss": 46.0,
-      "step": 30928
-    },
-    {
-      "epoch": 2.364738039260661,
-      "grad_norm": 0.0007101739174686372,
-      "learning_rate": 0.0001999972422581226,
-      "loss": 46.0,
-      "step": 30929
-    },
-    {
-      "epoch": 2.3648144962440507,
-      "grad_norm": 0.0007685082382522523,
-      "learning_rate": 0.00019999724207973561,
-      "loss": 46.0,
-      "step": 30930
-    },
-    {
-      "epoch": 2.3648909532274405,
-      "grad_norm": 0.0011201307643204927,
-      "learning_rate": 0.00019999724190134283,
-      "loss": 46.0,
-      "step": 30931
-    },
-    {
-      "epoch": 2.3649674102108302,
-      "grad_norm": 0.0009113825508393347,
-      "learning_rate": 0.0001999972417229443,
-      "loss": 46.0,
-      "step": 30932
-    },
-    {
-      "epoch": 2.36504386719422,
-      "grad_norm": 0.007421477697789669,
-      "learning_rate": 0.00019999724154454,
-      "loss": 46.0,
-      "step": 30933
-    },
-    {
-      "epoch": 2.3651203241776098,
-      "grad_norm": 0.0011129294289276004,
-      "learning_rate": 0.00019999724136612993,
-      "loss": 46.0,
-      "step": 30934
-    },
-    {
-      "epoch": 2.365196781160999,
-      "grad_norm": 0.0013804876944050193,
-      "learning_rate": 0.00019999724118771405,
-      "loss": 46.0,
-      "step": 30935
-    },
-    {
-      "epoch": 2.365273238144389,
-      "grad_norm": 0.000568172603379935,
-      "learning_rate": 0.00019999724100929246,
-      "loss": 46.0,
-      "step": 30936
-    },
-    {
-      "epoch": 2.3653496951277786,
-      "grad_norm": 0.0005792875890620053,
-      "learning_rate": 0.00019999724083086506,
-      "loss": 46.0,
-      "step": 30937
-    },
-    {
-      "epoch": 2.3654261521111684,
-      "grad_norm": 0.0008729451801627874,
-      "learning_rate": 0.00019999724065243192,
-      "loss": 46.0,
-      "step": 30938
-    },
-    {
-      "epoch": 2.365502609094558,
-      "grad_norm": 0.0023423947859555483,
-      "learning_rate": 0.000199997240473993,
-      "loss": 46.0,
-      "step": 30939
-    },
-    {
-      "epoch": 2.365579066077948,
-      "grad_norm": 0.0012942733010277152,
-      "learning_rate": 0.00019999724029554831,
-      "loss": 46.0,
-      "step": 30940
-    },
-    {
-      "epoch": 2.3656555230613376,
-      "grad_norm": 0.0011621961602941155,
-      "learning_rate": 0.00019999724011709785,
-      "loss": 46.0,
-      "step": 30941
-    },
-    {
-      "epoch": 2.3657319800447274,
-      "grad_norm": 0.0010152551112696528,
-      "learning_rate": 0.00019999723993864162,
-      "loss": 46.0,
-      "step": 30942
-    },
-    {
-      "epoch": 2.365808437028117,
-      "grad_norm": 0.0011088368482887745,
-      "learning_rate": 0.0001999972397601796,
-      "loss": 46.0,
-      "step": 30943
-    },
-    {
-      "epoch": 2.365884894011507,
-      "grad_norm": 0.001341847237199545,
-      "learning_rate": 0.00019999723958171185,
-      "loss": 46.0,
-      "step": 30944
-    },
-    {
-      "epoch": 2.3659613509948967,
-      "grad_norm": 0.0016284238081425428,
-      "learning_rate": 0.00019999723940323833,
-      "loss": 46.0,
-      "step": 30945
-    },
-    {
-      "epoch": 2.3660378079782864,
-      "grad_norm": 0.0007567990687675774,
-      "learning_rate": 0.00019999723922475902,
-      "loss": 46.0,
-      "step": 30946
-    },
-    {
-      "epoch": 2.3661142649616758,
-      "grad_norm": 0.002202435629442334,
-      "learning_rate": 0.00019999723904627392,
-      "loss": 46.0,
-      "step": 30947
-    },
-    {
-      "epoch": 2.3661907219450655,
-      "grad_norm": 0.0023028121795505285,
-      "learning_rate": 0.0001999972388677831,
-      "loss": 46.0,
-      "step": 30948
-    },
-    {
-      "epoch": 2.3662671789284553,
-      "grad_norm": 0.004321018699556589,
-      "learning_rate": 0.00019999723868928645,
-      "loss": 46.0,
-      "step": 30949
-    },
-    {
-      "epoch": 2.366343635911845,
-      "grad_norm": 0.0028072157874703407,
-      "learning_rate": 0.00019999723851078408,
-      "loss": 46.0,
-      "step": 30950
-    },
-    {
-      "epoch": 2.366420092895235,
-      "grad_norm": 0.0012792878551408648,
-      "learning_rate": 0.00019999723833227594,
-      "loss": 46.0,
-      "step": 30951
-    },
-    {
-      "epoch": 2.3664965498786246,
-      "grad_norm": 0.0009372485219500959,
-      "learning_rate": 0.000199997238153762,
-      "loss": 46.0,
-      "step": 30952
-    },
-    {
-      "epoch": 2.3665730068620143,
-      "grad_norm": 0.0024277525953948498,
-      "learning_rate": 0.0001999972379752423,
-      "loss": 46.0,
-      "step": 30953
-    },
-    {
-      "epoch": 2.366649463845404,
-      "grad_norm": 0.0006208621198311448,
-      "learning_rate": 0.00019999723779671685,
-      "loss": 46.0,
-      "step": 30954
-    },
-    {
-      "epoch": 2.366725920828794,
-      "grad_norm": 0.002180324401706457,
-      "learning_rate": 0.00019999723761818562,
-      "loss": 46.0,
-      "step": 30955
-    },
-    {
-      "epoch": 2.3668023778121836,
-      "grad_norm": 0.0007477523759007454,
-      "learning_rate": 0.00019999723743964863,
-      "loss": 46.0,
-      "step": 30956
-    },
-    {
-      "epoch": 2.366878834795573,
-      "grad_norm": 0.0010377010330557823,
-      "learning_rate": 0.00019999723726110585,
-      "loss": 46.0,
-      "step": 30957
-    },
-    {
-      "epoch": 2.3669552917789627,
-      "grad_norm": 0.0016980805667117238,
-      "learning_rate": 0.0001999972370825573,
-      "loss": 46.0,
-      "step": 30958
-    },
-    {
-      "epoch": 2.3670317487623524,
-      "grad_norm": 0.000663291197270155,
-      "learning_rate": 0.00019999723690400302,
-      "loss": 46.0,
-      "step": 30959
-    },
-    {
-      "epoch": 2.367108205745742,
-      "grad_norm": 0.003171132877469063,
-      "learning_rate": 0.00019999723672544292,
-      "loss": 46.0,
-      "step": 30960
-    },
-    {
-      "epoch": 2.367184662729132,
-      "grad_norm": 0.0006985990912653506,
-      "learning_rate": 0.0001999972365468771,
-      "loss": 46.0,
-      "step": 30961
-    },
-    {
-      "epoch": 2.3672611197125217,
-      "grad_norm": 0.00045496723032556474,
-      "learning_rate": 0.00019999723636830548,
-      "loss": 46.0,
-      "step": 30962
-    },
-    {
-      "epoch": 2.3673375766959115,
-      "grad_norm": 0.004115582909435034,
-      "learning_rate": 0.00019999723618972808,
-      "loss": 46.0,
-      "step": 30963
-    },
-    {
-      "epoch": 2.3674140336793013,
-      "grad_norm": 0.0023013551253825426,
-      "learning_rate": 0.00019999723601114492,
-      "loss": 46.0,
-      "step": 30964
-    },
-    {
-      "epoch": 2.367490490662691,
-      "grad_norm": 0.0006369391339831054,
-      "learning_rate": 0.000199997235832556,
-      "loss": 46.0,
-      "step": 30965
-    },
-    {
-      "epoch": 2.3675669476460808,
-      "grad_norm": 0.0007399871246889234,
-      "learning_rate": 0.0001999972356539613,
-      "loss": 46.0,
-      "step": 30966
-    },
-    {
-      "epoch": 2.3676434046294705,
-      "grad_norm": 0.0016781376907601953,
-      "learning_rate": 0.00019999723547536085,
-      "loss": 46.0,
-      "step": 30967
-    },
-    {
-      "epoch": 2.3677198616128603,
-      "grad_norm": 0.0019182696705684066,
-      "learning_rate": 0.00019999723529675462,
-      "loss": 46.0,
-      "step": 30968
-    },
-    {
-      "epoch": 2.3677963185962496,
-      "grad_norm": 0.0011726620141416788,
-      "learning_rate": 0.00019999723511814264,
-      "loss": 46.0,
-      "step": 30969
-    },
-    {
-      "epoch": 2.3678727755796394,
-      "grad_norm": 0.0006101656472310424,
-      "learning_rate": 0.00019999723493952486,
-      "loss": 46.0,
-      "step": 30970
-    },
-    {
-      "epoch": 2.367949232563029,
-      "grad_norm": 0.0014591108774766326,
-      "learning_rate": 0.00019999723476090133,
-      "loss": 46.0,
-      "step": 30971
-    },
-    {
-      "epoch": 2.368025689546419,
-      "grad_norm": 0.0007382669136859477,
-      "learning_rate": 0.000199997234582272,
-      "loss": 46.0,
-      "step": 30972
-    },
-    {
-      "epoch": 2.3681021465298087,
-      "grad_norm": 0.0012290574377402663,
-      "learning_rate": 0.00019999723440363693,
-      "loss": 46.0,
-      "step": 30973
-    },
-    {
-      "epoch": 2.3681786035131984,
-      "grad_norm": 0.00162209733389318,
-      "learning_rate": 0.0001999972342249961,
-      "loss": 46.0,
-      "step": 30974
-    },
-    {
-      "epoch": 2.368255060496588,
-      "grad_norm": 0.0007517147460021079,
-      "learning_rate": 0.00019999723404634946,
-      "loss": 46.0,
-      "step": 30975
-    },
-    {
-      "epoch": 2.368331517479978,
-      "grad_norm": 0.006431076675653458,
-      "learning_rate": 0.0001999972338676971,
-      "loss": 46.0,
-      "step": 30976
-    },
-    {
-      "epoch": 2.3684079744633677,
-      "grad_norm": 0.0002466006553731859,
-      "learning_rate": 0.00019999723368903892,
-      "loss": 46.0,
-      "step": 30977
-    },
-    {
-      "epoch": 2.368484431446757,
-      "grad_norm": 0.0009315300267189741,
-      "learning_rate": 0.000199997233510375,
-      "loss": 46.0,
-      "step": 30978
-    },
-    {
-      "epoch": 2.368560888430147,
-      "grad_norm": 0.002660072175785899,
-      "learning_rate": 0.00019999723333170532,
-      "loss": 46.0,
-      "step": 30979
-    },
-    {
-      "epoch": 2.3686373454135365,
-      "grad_norm": 0.0045874943025410175,
-      "learning_rate": 0.00019999723315302986,
-      "loss": 46.0,
-      "step": 30980
-    },
-    {
-      "epoch": 2.3687138023969263,
-      "grad_norm": 0.000573175901081413,
-      "learning_rate": 0.0001999972329743486,
-      "loss": 46.0,
-      "step": 30981
-    },
-    {
-      "epoch": 2.368790259380316,
-      "grad_norm": 0.0015418661059811711,
-      "learning_rate": 0.00019999723279566162,
-      "loss": 46.0,
-      "step": 30982
-    },
-    {
-      "epoch": 2.368866716363706,
-      "grad_norm": 0.0032072793692350388,
-      "learning_rate": 0.00019999723261696884,
-      "loss": 46.0,
-      "step": 30983
-    },
-    {
-      "epoch": 2.3689431733470956,
-      "grad_norm": 0.002618050202727318,
-      "learning_rate": 0.0001999972324382703,
-      "loss": 46.0,
-      "step": 30984
-    },
-    {
-      "epoch": 2.3690196303304853,
-      "grad_norm": 0.0007513040909543633,
-      "learning_rate": 0.00019999723225956598,
-      "loss": 46.0,
-      "step": 30985
-    },
-    {
-      "epoch": 2.369096087313875,
-      "grad_norm": 0.0012579533504322171,
-      "learning_rate": 0.0001999972320808559,
-      "loss": 46.0,
-      "step": 30986
-    },
-    {
-      "epoch": 2.369172544297265,
-      "grad_norm": 0.0029055820778012276,
-      "learning_rate": 0.00019999723190214006,
-      "loss": 46.0,
-      "step": 30987
-    },
-    {
-      "epoch": 2.3692490012806546,
-      "grad_norm": 0.0009218064951710403,
-      "learning_rate": 0.00019999723172341844,
-      "loss": 46.0,
-      "step": 30988
-    },
-    {
-      "epoch": 2.3693254582640444,
-      "grad_norm": 0.002410757588222623,
-      "learning_rate": 0.00019999723154469104,
-      "loss": 46.0,
-      "step": 30989
-    },
-    {
-      "epoch": 2.369401915247434,
-      "grad_norm": 0.0009782094275578856,
-      "learning_rate": 0.0001999972313659579,
-      "loss": 46.0,
-      "step": 30990
-    },
-    {
-      "epoch": 2.3694783722308235,
-      "grad_norm": 0.0027987691573798656,
-      "learning_rate": 0.000199997231187219,
-      "loss": 46.0,
-      "step": 30991
-    },
-    {
-      "epoch": 2.3695548292142132,
-      "grad_norm": 0.0019433513516560197,
-      "learning_rate": 0.0001999972310084743,
-      "loss": 46.0,
-      "step": 30992
-    },
-    {
-      "epoch": 2.369631286197603,
-      "grad_norm": 0.002832096768543124,
-      "learning_rate": 0.00019999723082972381,
-      "loss": 46.0,
-      "step": 30993
-    },
-    {
-      "epoch": 2.3697077431809928,
-      "grad_norm": 0.0005691216210834682,
-      "learning_rate": 0.0001999972306509676,
-      "loss": 46.0,
-      "step": 30994
-    },
-    {
-      "epoch": 2.3697842001643825,
-      "grad_norm": 0.0015089535154402256,
-      "learning_rate": 0.00019999723047220557,
-      "loss": 46.0,
-      "step": 30995
-    },
-    {
-      "epoch": 2.3698606571477723,
-      "grad_norm": 0.0014779954217374325,
-      "learning_rate": 0.00019999723029343782,
-      "loss": 46.0,
-      "step": 30996
-    },
-    {
-      "epoch": 2.369937114131162,
-      "grad_norm": 0.0012628003023564816,
-      "learning_rate": 0.0001999972301146643,
-      "loss": 46.0,
-      "step": 30997
-    },
-    {
-      "epoch": 2.370013571114552,
-      "grad_norm": 0.0011182072339579463,
-      "learning_rate": 0.00019999722993588496,
-      "loss": 46.0,
-      "step": 30998
-    },
-    {
-      "epoch": 2.3700900280979416,
-      "grad_norm": 0.0027189922984689474,
-      "learning_rate": 0.0001999972297570999,
-      "loss": 46.0,
-      "step": 30999
-    },
-    {
-      "epoch": 2.370166485081331,
-      "grad_norm": 0.0006680736551061273,
-      "learning_rate": 0.00019999722957830904,
-      "loss": 46.0,
-      "step": 31000
-    },
-    {
-      "epoch": 2.3702429420647206,
-      "grad_norm": 0.0004944147076457739,
-      "learning_rate": 0.00019999722939951242,
-      "loss": 46.0,
-      "step": 31001
-    },
-    {
-      "epoch": 2.3703193990481104,
-      "grad_norm": 0.0015551430406048894,
-      "learning_rate": 0.00019999722922071006,
-      "loss": 46.0,
-      "step": 31002
-    },
-    {
-      "epoch": 2.3703958560315,
-      "grad_norm": 0.000798713939730078,
-      "learning_rate": 0.0001999972290419019,
-      "loss": 46.0,
-      "step": 31003
-    },
-    {
-      "epoch": 2.37047231301489,
-      "grad_norm": 0.0010620767716318369,
-      "learning_rate": 0.00019999722886308795,
-      "loss": 46.0,
-      "step": 31004
-    },
-    {
-      "epoch": 2.3705487699982797,
-      "grad_norm": 0.0005952194333076477,
-      "learning_rate": 0.00019999722868426826,
-      "loss": 46.0,
-      "step": 31005
-    },
-    {
-      "epoch": 2.3706252269816694,
-      "grad_norm": 0.002582385204732418,
-      "learning_rate": 0.00019999722850544277,
-      "loss": 46.0,
-      "step": 31006
-    },
-    {
-      "epoch": 2.370701683965059,
-      "grad_norm": 0.0011500270338729024,
-      "learning_rate": 0.00019999722832661157,
-      "loss": 46.0,
-      "step": 31007
-    },
-    {
-      "epoch": 2.370778140948449,
-      "grad_norm": 0.0010102280648425221,
-      "learning_rate": 0.00019999722814777456,
-      "loss": 46.0,
-      "step": 31008
-    },
-    {
-      "epoch": 2.3708545979318387,
-      "grad_norm": 0.0017857919447124004,
-      "learning_rate": 0.00019999722796893175,
-      "loss": 46.0,
-      "step": 31009
-    },
-    {
-      "epoch": 2.3709310549152285,
-      "grad_norm": 0.0013549618888646364,
-      "learning_rate": 0.00019999722779008323,
-      "loss": 46.0,
-      "step": 31010
-    },
-    {
-      "epoch": 2.3710075118986182,
-      "grad_norm": 0.0007024784572422504,
-      "learning_rate": 0.00019999722761122893,
-      "loss": 46.0,
-      "step": 31011
-    },
-    {
-      "epoch": 2.371083968882008,
-      "grad_norm": 0.001392704900354147,
-      "learning_rate": 0.00019999722743236883,
-      "loss": 46.0,
-      "step": 31012
-    },
-    {
-      "epoch": 2.3711604258653973,
-      "grad_norm": 0.000905881286598742,
-      "learning_rate": 0.00019999722725350298,
-      "loss": 46.0,
-      "step": 31013
-    },
-    {
-      "epoch": 2.371236882848787,
-      "grad_norm": 0.0008513322682119906,
-      "learning_rate": 0.00019999722707463139,
-      "loss": 46.0,
-      "step": 31014
-    },
-    {
-      "epoch": 2.371313339832177,
-      "grad_norm": 0.0006747301667928696,
-      "learning_rate": 0.000199997226895754,
-      "loss": 46.0,
-      "step": 31015
-    },
-    {
-      "epoch": 2.3713897968155666,
-      "grad_norm": 0.0037178483325988054,
-      "learning_rate": 0.00019999722671687083,
-      "loss": 46.0,
-      "step": 31016
-    },
-    {
-      "epoch": 2.3714662537989564,
-      "grad_norm": 0.000788997916970402,
-      "learning_rate": 0.0001999972265379819,
-      "loss": 46.0,
-      "step": 31017
-    },
-    {
-      "epoch": 2.371542710782346,
-      "grad_norm": 0.0034271508920937777,
-      "learning_rate": 0.0001999972263590872,
-      "loss": 46.0,
-      "step": 31018
-    },
-    {
-      "epoch": 2.371619167765736,
-      "grad_norm": 0.005774281453341246,
-      "learning_rate": 0.00019999722618018674,
-      "loss": 46.0,
-      "step": 31019
-    },
-    {
-      "epoch": 2.3716956247491257,
-      "grad_norm": 0.0017899870872497559,
-      "learning_rate": 0.00019999722600128053,
-      "loss": 46.0,
-      "step": 31020
-    },
-    {
-      "epoch": 2.3717720817325154,
-      "grad_norm": 0.0028651640750467777,
-      "learning_rate": 0.0001999972258223685,
-      "loss": 46.0,
-      "step": 31021
-    },
-    {
-      "epoch": 2.3718485387159047,
-      "grad_norm": 0.007358357310295105,
-      "learning_rate": 0.00019999722564345075,
-      "loss": 46.0,
-      "step": 31022
-    },
-    {
-      "epoch": 2.3719249956992945,
-      "grad_norm": 0.0011318784672766924,
-      "learning_rate": 0.0001999972254645272,
-      "loss": 46.0,
-      "step": 31023
-    },
-    {
-      "epoch": 2.3720014526826843,
-      "grad_norm": 0.001023461576551199,
-      "learning_rate": 0.00019999722528559786,
-      "loss": 46.0,
-      "step": 31024
-    },
-    {
-      "epoch": 2.372077909666074,
-      "grad_norm": 0.0014502619160339236,
-      "learning_rate": 0.00019999722510666282,
-      "loss": 46.0,
-      "step": 31025
-    },
-    {
-      "epoch": 2.3721543666494638,
-      "grad_norm": 0.002782789058983326,
-      "learning_rate": 0.00019999722492772195,
-      "loss": 46.0,
-      "step": 31026
-    },
-    {
-      "epoch": 2.3722308236328535,
-      "grad_norm": 0.0050197262316942215,
-      "learning_rate": 0.00019999722474877533,
-      "loss": 46.0,
-      "step": 31027
-    },
-    {
-      "epoch": 2.3723072806162433,
-      "grad_norm": 0.00089174514869228,
-      "learning_rate": 0.00019999722456982296,
-      "loss": 46.0,
-      "step": 31028
-    },
-    {
-      "epoch": 2.372383737599633,
-      "grad_norm": 0.0007411700207740068,
-      "learning_rate": 0.00019999722439086477,
-      "loss": 46.0,
-      "step": 31029
-    },
-    {
-      "epoch": 2.372460194583023,
-      "grad_norm": 0.0029685255140066147,
-      "learning_rate": 0.00019999722421190085,
-      "loss": 46.0,
-      "step": 31030
-    },
-    {
-      "epoch": 2.3725366515664126,
-      "grad_norm": 0.0009744045091792941,
-      "learning_rate": 0.00019999722403293114,
-      "loss": 46.0,
-      "step": 31031
-    },
-    {
-      "epoch": 2.3726131085498023,
-      "grad_norm": 0.0021773381158709526,
-      "learning_rate": 0.0001999972238539557,
-      "loss": 46.0,
-      "step": 31032
-    },
-    {
-      "epoch": 2.372689565533192,
-      "grad_norm": 0.0009067789651453495,
-      "learning_rate": 0.00019999722367497445,
-      "loss": 46.0,
-      "step": 31033
-    },
-    {
-      "epoch": 2.3727660225165814,
-      "grad_norm": 0.0006509035592898726,
-      "learning_rate": 0.00019999722349598744,
-      "loss": 46.0,
-      "step": 31034
-    },
-    {
-      "epoch": 2.372842479499971,
-      "grad_norm": 0.001086916890926659,
-      "learning_rate": 0.00019999722331699466,
-      "loss": 46.0,
-      "step": 31035
-    },
-    {
-      "epoch": 2.372918936483361,
-      "grad_norm": 0.0010047780815511942,
-      "learning_rate": 0.00019999722313799613,
-      "loss": 46.0,
-      "step": 31036
-    },
-    {
-      "epoch": 2.3729953934667507,
-      "grad_norm": 0.0004870621196459979,
-      "learning_rate": 0.00019999722295899183,
-      "loss": 46.0,
-      "step": 31037
-    },
-    {
-      "epoch": 2.3730718504501405,
-      "grad_norm": 0.0014111370546743274,
-      "learning_rate": 0.00019999722277998173,
-      "loss": 46.0,
-      "step": 31038
-    },
-    {
-      "epoch": 2.3731483074335302,
-      "grad_norm": 0.000663279031869024,
-      "learning_rate": 0.00019999722260096586,
-      "loss": 46.0,
-      "step": 31039
-    },
-    {
-      "epoch": 2.37322476441692,
-      "grad_norm": 0.0018719497602432966,
-      "learning_rate": 0.00019999722242194427,
-      "loss": 46.0,
-      "step": 31040
-    },
-    {
-      "epoch": 2.3733012214003097,
-      "grad_norm": 0.0011070228647440672,
-      "learning_rate": 0.00019999722224291688,
-      "loss": 46.0,
-      "step": 31041
-    },
-    {
-      "epoch": 2.3733776783836995,
-      "grad_norm": 0.0011250664247199893,
-      "learning_rate": 0.0001999972220638837,
-      "loss": 46.0,
-      "step": 31042
-    },
-    {
-      "epoch": 2.3734541353670893,
-      "grad_norm": 0.0011409572325646877,
-      "learning_rate": 0.0001999972218848448,
-      "loss": 46.0,
-      "step": 31043
-    },
-    {
-      "epoch": 2.3735305923504786,
-      "grad_norm": 0.0025951277930289507,
-      "learning_rate": 0.0001999972217058001,
-      "loss": 46.0,
-      "step": 31044
-    },
-    {
-      "epoch": 2.3736070493338683,
-      "grad_norm": 0.0009983801282942295,
-      "learning_rate": 0.00019999722152674962,
-      "loss": 46.0,
-      "step": 31045
-    },
-    {
-      "epoch": 2.373683506317258,
-      "grad_norm": 0.001446805545128882,
-      "learning_rate": 0.0001999972213476934,
-      "loss": 46.0,
-      "step": 31046
-    },
-    {
-      "epoch": 2.373759963300648,
-      "grad_norm": 0.0011241381289437413,
-      "learning_rate": 0.00019999722116863139,
-      "loss": 46.0,
-      "step": 31047
-    },
-    {
-      "epoch": 2.3738364202840376,
-      "grad_norm": 0.0019013514975085855,
-      "learning_rate": 0.0001999972209895636,
-      "loss": 46.0,
-      "step": 31048
-    },
-    {
-      "epoch": 2.3739128772674274,
-      "grad_norm": 0.005932185333222151,
-      "learning_rate": 0.00019999722081049005,
-      "loss": 46.0,
-      "step": 31049
-    },
-    {
-      "epoch": 2.373989334250817,
-      "grad_norm": 0.0008484215359203517,
-      "learning_rate": 0.00019999722063141075,
-      "loss": 46.0,
-      "step": 31050
-    },
-    {
-      "epoch": 2.374065791234207,
-      "grad_norm": 0.0006371770286932588,
-      "learning_rate": 0.00019999722045232568,
-      "loss": 46.0,
-      "step": 31051
-    },
-    {
-      "epoch": 2.3741422482175967,
-      "grad_norm": 0.0008268074016086757,
-      "learning_rate": 0.0001999972202732348,
-      "loss": 46.0,
-      "step": 31052
-    },
-    {
-      "epoch": 2.3742187052009864,
-      "grad_norm": 0.001894623157568276,
-      "learning_rate": 0.0001999972200941382,
-      "loss": 46.0,
-      "step": 31053
-    },
-    {
-      "epoch": 2.374295162184376,
-      "grad_norm": 0.0017251019598916173,
-      "learning_rate": 0.00019999721991503582,
-      "loss": 46.0,
-      "step": 31054
-    },
-    {
-      "epoch": 2.374371619167766,
-      "grad_norm": 0.0010152189061045647,
-      "learning_rate": 0.00019999721973592763,
-      "loss": 46.0,
-      "step": 31055
-    },
-    {
-      "epoch": 2.3744480761511553,
-      "grad_norm": 0.0009373384527862072,
-      "learning_rate": 0.00019999721955681372,
-      "loss": 46.0,
-      "step": 31056
-    },
-    {
-      "epoch": 2.374524533134545,
-      "grad_norm": 0.0018818433163687587,
-      "learning_rate": 0.000199997219377694,
-      "loss": 46.0,
-      "step": 31057
-    },
-    {
-      "epoch": 2.374600990117935,
-      "grad_norm": 0.0017436678754165769,
-      "learning_rate": 0.00019999721919856855,
-      "loss": 46.0,
-      "step": 31058
-    },
-    {
-      "epoch": 2.3746774471013246,
-      "grad_norm": 0.0021920257713645697,
-      "learning_rate": 0.00019999721901943731,
-      "loss": 46.0,
-      "step": 31059
-    },
-    {
-      "epoch": 2.3747539040847143,
-      "grad_norm": 0.0013635688228532672,
-      "learning_rate": 0.0001999972188403003,
-      "loss": 46.0,
-      "step": 31060
-    },
-    {
-      "epoch": 2.374830361068104,
-      "grad_norm": 0.005182044580578804,
-      "learning_rate": 0.00019999721866115753,
-      "loss": 46.0,
-      "step": 31061
-    },
-    {
-      "epoch": 2.374906818051494,
-      "grad_norm": 0.001008194638416171,
-      "learning_rate": 0.00019999721848200898,
-      "loss": 46.0,
-      "step": 31062
-    },
-    {
-      "epoch": 2.3749832750348836,
-      "grad_norm": 0.0006955301505513489,
-      "learning_rate": 0.00019999721830285468,
-      "loss": 46.0,
-      "step": 31063
-    },
-    {
-      "epoch": 2.3750597320182734,
-      "grad_norm": 0.001902335905469954,
-      "learning_rate": 0.00019999721812369458,
-      "loss": 46.0,
-      "step": 31064
-    },
-    {
-      "epoch": 2.375136189001663,
-      "grad_norm": 0.001501190010458231,
-      "learning_rate": 0.00019999721794452873,
-      "loss": 46.0,
-      "step": 31065
-    },
-    {
-      "epoch": 2.3752126459850524,
-      "grad_norm": 0.0017575160600245,
-      "learning_rate": 0.00019999721776535711,
-      "loss": 46.0,
-      "step": 31066
-    },
-    {
-      "epoch": 2.375289102968442,
-      "grad_norm": 0.0008912758203223348,
-      "learning_rate": 0.00019999721758617972,
-      "loss": 46.0,
-      "step": 31067
-    },
-    {
-      "epoch": 2.375365559951832,
-      "grad_norm": 0.0009623733349144459,
-      "learning_rate": 0.00019999721740699658,
-      "loss": 46.0,
-      "step": 31068
-    },
-    {
-      "epoch": 2.3754420169352217,
-      "grad_norm": 0.0036720200441777706,
-      "learning_rate": 0.00019999721722780764,
-      "loss": 46.0,
-      "step": 31069
-    },
-    {
-      "epoch": 2.3755184739186115,
-      "grad_norm": 0.0041097113862633705,
-      "learning_rate": 0.00019999721704861293,
-      "loss": 46.0,
-      "step": 31070
-    },
-    {
-      "epoch": 2.3755949309020012,
-      "grad_norm": 0.000994296045973897,
-      "learning_rate": 0.00019999721686941247,
-      "loss": 46.0,
-      "step": 31071
-    },
-    {
-      "epoch": 2.375671387885391,
-      "grad_norm": 0.0012296505738049746,
-      "learning_rate": 0.00019999721669020621,
-      "loss": 46.0,
-      "step": 31072
-    },
-    {
-      "epoch": 2.3757478448687808,
-      "grad_norm": 0.001984257949516177,
-      "learning_rate": 0.0001999972165109942,
-      "loss": 46.0,
-      "step": 31073
-    },
-    {
-      "epoch": 2.3758243018521705,
-      "grad_norm": 0.000604613742325455,
-      "learning_rate": 0.00019999721633177643,
-      "loss": 46.0,
-      "step": 31074
-    },
-    {
-      "epoch": 2.3759007588355603,
-      "grad_norm": 0.00041230194619856775,
-      "learning_rate": 0.0001999972161525529,
-      "loss": 46.0,
-      "step": 31075
-    },
-    {
-      "epoch": 2.37597721581895,
-      "grad_norm": 0.0019305472960695624,
-      "learning_rate": 0.00019999721597332358,
-      "loss": 46.0,
-      "step": 31076
-    },
-    {
-      "epoch": 2.37605367280234,
-      "grad_norm": 0.0006674590986222029,
-      "learning_rate": 0.0001999972157940885,
-      "loss": 46.0,
-      "step": 31077
-    },
-    {
-      "epoch": 2.376130129785729,
-      "grad_norm": 0.0015203512739390135,
-      "learning_rate": 0.00019999721561484766,
-      "loss": 46.0,
-      "step": 31078
-    },
-    {
-      "epoch": 2.376206586769119,
-      "grad_norm": 0.0005426773568615317,
-      "learning_rate": 0.00019999721543560102,
-      "loss": 46.0,
-      "step": 31079
-    },
-    {
-      "epoch": 2.3762830437525087,
-      "grad_norm": 0.0009335912181995809,
-      "learning_rate": 0.00019999721525634862,
-      "loss": 46.0,
-      "step": 31080
-    },
-    {
-      "epoch": 2.3763595007358984,
-      "grad_norm": 0.0007931062136776745,
-      "learning_rate": 0.00019999721507709046,
-      "loss": 46.0,
-      "step": 31081
-    },
-    {
-      "epoch": 2.376435957719288,
-      "grad_norm": 0.0015477658016607165,
-      "learning_rate": 0.00019999721489782655,
-      "loss": 46.0,
-      "step": 31082
-    },
-    {
-      "epoch": 2.376512414702678,
-      "grad_norm": 0.0005699508474208415,
-      "learning_rate": 0.00019999721471855683,
-      "loss": 46.0,
-      "step": 31083
-    },
-    {
-      "epoch": 2.3765888716860677,
-      "grad_norm": 0.0011459199013188481,
-      "learning_rate": 0.00019999721453928135,
-      "loss": 46.0,
-      "step": 31084
-    },
-    {
-      "epoch": 2.3766653286694575,
-      "grad_norm": 0.002297770930454135,
-      "learning_rate": 0.00019999721436000014,
-      "loss": 46.0,
-      "step": 31085
-    },
-    {
-      "epoch": 2.376741785652847,
-      "grad_norm": 0.000576640188228339,
-      "learning_rate": 0.0001999972141807131,
-      "loss": 46.0,
-      "step": 31086
-    },
-    {
-      "epoch": 2.376818242636237,
-      "grad_norm": 0.0005016205250285566,
-      "learning_rate": 0.00019999721400142036,
-      "loss": 46.0,
-      "step": 31087
-    },
-    {
-      "epoch": 2.3768946996196263,
-      "grad_norm": 0.0010746658081188798,
-      "learning_rate": 0.0001999972138221218,
-      "loss": 46.0,
-      "step": 31088
-    },
-    {
-      "epoch": 2.376971156603016,
-      "grad_norm": 0.0003642332158051431,
-      "learning_rate": 0.00019999721364281748,
-      "loss": 46.0,
-      "step": 31089
-    },
-    {
-      "epoch": 2.377047613586406,
-      "grad_norm": 0.0005703719216398895,
-      "learning_rate": 0.00019999721346350738,
-      "loss": 46.0,
-      "step": 31090
-    },
-    {
-      "epoch": 2.3771240705697956,
-      "grad_norm": 0.0012409646296873689,
-      "learning_rate": 0.00019999721328419154,
-      "loss": 46.0,
-      "step": 31091
-    },
-    {
-      "epoch": 2.3772005275531853,
-      "grad_norm": 0.0014029587619006634,
-      "learning_rate": 0.00019999721310486992,
-      "loss": 46.0,
-      "step": 31092
-    },
-    {
-      "epoch": 2.377276984536575,
-      "grad_norm": 0.0009414395317435265,
-      "learning_rate": 0.00019999721292554253,
-      "loss": 46.0,
-      "step": 31093
-    },
-    {
-      "epoch": 2.377353441519965,
-      "grad_norm": 0.0009052905370481312,
-      "learning_rate": 0.00019999721274620934,
-      "loss": 46.0,
-      "step": 31094
-    },
-    {
-      "epoch": 2.3774298985033546,
-      "grad_norm": 0.0005171510856598616,
-      "learning_rate": 0.00019999721256687043,
-      "loss": 46.0,
-      "step": 31095
-    },
-    {
-      "epoch": 2.3775063554867444,
-      "grad_norm": 0.000886720372363925,
-      "learning_rate": 0.00019999721238752574,
-      "loss": 46.0,
-      "step": 31096
-    },
-    {
-      "epoch": 2.377582812470134,
-      "grad_norm": 0.0028995713219046593,
-      "learning_rate": 0.00019999721220817526,
-      "loss": 46.0,
-      "step": 31097
-    },
-    {
-      "epoch": 2.377659269453524,
-      "grad_norm": 0.006990660913288593,
-      "learning_rate": 0.00019999721202881903,
-      "loss": 46.0,
-      "step": 31098
-    },
-    {
-      "epoch": 2.3777357264369137,
-      "grad_norm": 0.0026172089856117964,
-      "learning_rate": 0.00019999721184945702,
-      "loss": 46.0,
-      "step": 31099
-    },
-    {
-      "epoch": 2.377812183420303,
-      "grad_norm": 0.0028872620314359665,
-      "learning_rate": 0.00019999721167008921,
-      "loss": 46.0,
-      "step": 31100
-    },
-    {
-      "epoch": 2.3778886404036927,
-      "grad_norm": 0.00242223241366446,
-      "learning_rate": 0.0001999972114907157,
-      "loss": 46.0,
-      "step": 31101
-    },
-    {
-      "epoch": 2.3779650973870825,
-      "grad_norm": 0.0010536222252994776,
-      "learning_rate": 0.00019999721131133637,
-      "loss": 46.0,
-      "step": 31102
-    },
-    {
-      "epoch": 2.3780415543704723,
-      "grad_norm": 0.0017761014169082046,
-      "learning_rate": 0.0001999972111319513,
-      "loss": 46.0,
-      "step": 31103
-    },
-    {
-      "epoch": 2.378118011353862,
-      "grad_norm": 0.0007009514956735075,
-      "learning_rate": 0.00019999721095256042,
-      "loss": 46.0,
-      "step": 31104
-    },
-    {
-      "epoch": 2.378194468337252,
-      "grad_norm": 0.0007399828173220158,
-      "learning_rate": 0.0001999972107731638,
-      "loss": 46.0,
-      "step": 31105
-    },
-    {
-      "epoch": 2.3782709253206415,
-      "grad_norm": 0.0005016362993046641,
-      "learning_rate": 0.00019999721059376139,
-      "loss": 46.0,
-      "step": 31106
-    },
-    {
-      "epoch": 2.3783473823040313,
-      "grad_norm": 0.0009125939104706049,
-      "learning_rate": 0.00019999721041435325,
-      "loss": 46.0,
-      "step": 31107
-    },
-    {
-      "epoch": 2.378423839287421,
-      "grad_norm": 0.0016421690816059709,
-      "learning_rate": 0.0001999972102349393,
-      "loss": 46.0,
-      "step": 31108
-    },
-    {
-      "epoch": 2.3785002962708104,
-      "grad_norm": 0.0006779235554859042,
-      "learning_rate": 0.00019999721005551963,
-      "loss": 46.0,
-      "step": 31109
-    },
-    {
-      "epoch": 2.3785767532542,
-      "grad_norm": 0.0011737538734450936,
-      "learning_rate": 0.00019999720987609414,
-      "loss": 46.0,
-      "step": 31110
-    },
-    {
-      "epoch": 2.37865321023759,
-      "grad_norm": 0.000546792522072792,
-      "learning_rate": 0.0001999972096966629,
-      "loss": 46.0,
-      "step": 31111
-    },
-    {
-      "epoch": 2.3787296672209797,
-      "grad_norm": 0.0015708922874182463,
-      "learning_rate": 0.0001999972095172259,
-      "loss": 46.0,
-      "step": 31112
-    },
-    {
-      "epoch": 2.3788061242043694,
-      "grad_norm": 0.0027509615756571293,
-      "learning_rate": 0.0001999972093377831,
-      "loss": 46.0,
-      "step": 31113
-    },
-    {
-      "epoch": 2.378882581187759,
-      "grad_norm": 0.0005737630417570472,
-      "learning_rate": 0.00019999720915833455,
-      "loss": 46.0,
-      "step": 31114
-    },
-    {
-      "epoch": 2.378959038171149,
-      "grad_norm": 0.003620326053351164,
-      "learning_rate": 0.00019999720897888022,
-      "loss": 46.0,
-      "step": 31115
-    },
-    {
-      "epoch": 2.3790354951545387,
-      "grad_norm": 0.0005674313870258629,
-      "learning_rate": 0.00019999720879942015,
-      "loss": 46.0,
-      "step": 31116
-    },
-    {
-      "epoch": 2.3791119521379285,
-      "grad_norm": 0.00027439510449767113,
-      "learning_rate": 0.0001999972086199543,
-      "loss": 46.0,
-      "step": 31117
-    },
-    {
-      "epoch": 2.3791884091213182,
-      "grad_norm": 0.001495744800195098,
-      "learning_rate": 0.0001999972084404827,
-      "loss": 46.0,
-      "step": 31118
-    },
-    {
-      "epoch": 2.379264866104708,
-      "grad_norm": 0.0009910111548379064,
-      "learning_rate": 0.00019999720826100527,
-      "loss": 46.0,
-      "step": 31119
-    },
-    {
-      "epoch": 2.3793413230880978,
-      "grad_norm": 0.0007201422704383731,
-      "learning_rate": 0.0001999972080815221,
-      "loss": 46.0,
-      "step": 31120
-    },
-    {
-      "epoch": 2.3794177800714875,
-      "grad_norm": 0.0007935430039651692,
-      "learning_rate": 0.00019999720790203317,
-      "loss": 46.0,
-      "step": 31121
-    },
-    {
-      "epoch": 2.379494237054877,
-      "grad_norm": 0.0013482357608154416,
-      "learning_rate": 0.0001999972077225385,
-      "loss": 46.0,
-      "step": 31122
-    },
-    {
-      "epoch": 2.3795706940382666,
-      "grad_norm": 0.0011176994303241372,
-      "learning_rate": 0.000199997207543038,
-      "loss": 46.0,
-      "step": 31123
-    },
-    {
-      "epoch": 2.3796471510216564,
-      "grad_norm": 0.0024398467503488064,
-      "learning_rate": 0.00019999720736353177,
-      "loss": 46.0,
-      "step": 31124
-    },
-    {
-      "epoch": 2.379723608005046,
-      "grad_norm": 0.002355023752897978,
-      "learning_rate": 0.00019999720718401974,
-      "loss": 46.0,
-      "step": 31125
-    },
-    {
-      "epoch": 2.379800064988436,
-      "grad_norm": 0.0017233600374311209,
-      "learning_rate": 0.00019999720700450196,
-      "loss": 46.0,
-      "step": 31126
-    },
-    {
-      "epoch": 2.3798765219718256,
-      "grad_norm": 0.0014562989817932248,
-      "learning_rate": 0.0001999972068249784,
-      "loss": 46.0,
-      "step": 31127
-    },
-    {
-      "epoch": 2.3799529789552154,
-      "grad_norm": 0.0007559538935311139,
-      "learning_rate": 0.0001999972066454491,
-      "loss": 46.0,
-      "step": 31128
-    },
-    {
-      "epoch": 2.380029435938605,
-      "grad_norm": 0.0008023447007872164,
-      "learning_rate": 0.00019999720646591402,
-      "loss": 46.0,
-      "step": 31129
-    },
-    {
-      "epoch": 2.380105892921995,
-      "grad_norm": 0.0030296442564576864,
-      "learning_rate": 0.00019999720628637317,
-      "loss": 46.0,
-      "step": 31130
-    },
-    {
-      "epoch": 2.3801823499053842,
-      "grad_norm": 0.0007195067591965199,
-      "learning_rate": 0.00019999720610682653,
-      "loss": 46.0,
-      "step": 31131
-    },
-    {
-      "epoch": 2.380258806888774,
-      "grad_norm": 0.00210877344943583,
-      "learning_rate": 0.00019999720592727416,
-      "loss": 46.0,
-      "step": 31132
-    },
-    {
-      "epoch": 2.3803352638721638,
-      "grad_norm": 0.0011491795303300023,
-      "learning_rate": 0.00019999720574771597,
-      "loss": 46.0,
-      "step": 31133
-    },
-    {
-      "epoch": 2.3804117208555535,
-      "grad_norm": 0.0038050892762839794,
-      "learning_rate": 0.00019999720556815204,
-      "loss": 46.0,
-      "step": 31134
-    },
-    {
-      "epoch": 2.3804881778389433,
-      "grad_norm": 0.0013043897924944758,
-      "learning_rate": 0.00019999720538858235,
-      "loss": 46.0,
-      "step": 31135
-    },
-    {
-      "epoch": 2.380564634822333,
-      "grad_norm": 0.0022914758883416653,
-      "learning_rate": 0.00019999720520900687,
-      "loss": 46.0,
-      "step": 31136
-    },
-    {
-      "epoch": 2.380641091805723,
-      "grad_norm": 0.0007555970805697143,
-      "learning_rate": 0.00019999720502942564,
-      "loss": 46.0,
-      "step": 31137
-    },
-    {
-      "epoch": 2.3807175487891126,
-      "grad_norm": 0.0010308842174708843,
-      "learning_rate": 0.0001999972048498386,
-      "loss": 46.0,
-      "step": 31138
-    },
-    {
-      "epoch": 2.3807940057725023,
-      "grad_norm": 0.001108479336835444,
-      "learning_rate": 0.00019999720467024583,
-      "loss": 46.0,
-      "step": 31139
-    },
-    {
-      "epoch": 2.380870462755892,
-      "grad_norm": 0.0010350996162742376,
-      "learning_rate": 0.0001999972044906473,
-      "loss": 46.0,
-      "step": 31140
-    },
-    {
-      "epoch": 2.380946919739282,
-      "grad_norm": 0.0009307586005888879,
-      "learning_rate": 0.00019999720431104295,
-      "loss": 46.0,
-      "step": 31141
-    },
-    {
-      "epoch": 2.3810233767226716,
-      "grad_norm": 0.0016070916317403316,
-      "learning_rate": 0.00019999720413143288,
-      "loss": 46.0,
-      "step": 31142
-    },
-    {
-      "epoch": 2.3810998337060614,
-      "grad_norm": 0.0019747591577470303,
-      "learning_rate": 0.000199997203951817,
-      "loss": 46.0,
-      "step": 31143
-    },
-    {
-      "epoch": 2.3811762906894507,
-      "grad_norm": 0.0006214067107066512,
-      "learning_rate": 0.0001999972037721954,
-      "loss": 46.0,
-      "step": 31144
-    },
-    {
-      "epoch": 2.3812527476728405,
-      "grad_norm": 0.001856733695603907,
-      "learning_rate": 0.000199997203592568,
-      "loss": 46.0,
-      "step": 31145
-    },
-    {
-      "epoch": 2.38132920465623,
-      "grad_norm": 0.0006737332441844046,
-      "learning_rate": 0.00019999720341293484,
-      "loss": 46.0,
-      "step": 31146
-    },
-    {
-      "epoch": 2.38140566163962,
-      "grad_norm": 0.001253923517651856,
-      "learning_rate": 0.0001999972032332959,
-      "loss": 46.0,
-      "step": 31147
-    },
-    {
-      "epoch": 2.3814821186230097,
-      "grad_norm": 0.0016853242414072156,
-      "learning_rate": 0.0001999972030536512,
-      "loss": 46.0,
-      "step": 31148
-    },
-    {
-      "epoch": 2.3815585756063995,
-      "grad_norm": 0.0011951141059398651,
-      "learning_rate": 0.0001999972028740007,
-      "loss": 46.0,
-      "step": 31149
-    },
-    {
-      "epoch": 2.3816350325897893,
-      "grad_norm": 0.00036048179026693106,
-      "learning_rate": 0.00019999720269434448,
-      "loss": 46.0,
-      "step": 31150
-    },
-    {
-      "epoch": 2.381711489573179,
-      "grad_norm": 0.000941413571126759,
-      "learning_rate": 0.00019999720251468245,
-      "loss": 46.0,
-      "step": 31151
-    },
-    {
-      "epoch": 2.381787946556569,
-      "grad_norm": 0.0020912436302751303,
-      "learning_rate": 0.0001999972023350147,
-      "loss": 46.0,
-      "step": 31152
-    },
-    {
-      "epoch": 2.381864403539958,
-      "grad_norm": 0.005619356874376535,
-      "learning_rate": 0.00019999720215534112,
-      "loss": 46.0,
-      "step": 31153
-    },
-    {
-      "epoch": 2.381940860523348,
-      "grad_norm": 0.0009027887717820704,
-      "learning_rate": 0.0001999972019756618,
-      "loss": 46.0,
-      "step": 31154
-    },
-    {
-      "epoch": 2.3820173175067376,
-      "grad_norm": 0.0007508570561185479,
-      "learning_rate": 0.00019999720179597673,
-      "loss": 46.0,
-      "step": 31155
-    },
-    {
-      "epoch": 2.3820937744901274,
-      "grad_norm": 0.0010588339064270258,
-      "learning_rate": 0.00019999720161628588,
-      "loss": 46.0,
-      "step": 31156
-    },
-    {
-      "epoch": 2.382170231473517,
-      "grad_norm": 0.006090869195759296,
-      "learning_rate": 0.00019999720143658921,
-      "loss": 46.0,
-      "step": 31157
-    },
-    {
-      "epoch": 2.382246688456907,
-      "grad_norm": 0.0016812618123367429,
-      "learning_rate": 0.00019999720125688682,
-      "loss": 46.0,
-      "step": 31158
-    },
-    {
-      "epoch": 2.3823231454402967,
-      "grad_norm": 0.0006664101383648813,
-      "learning_rate": 0.0001999972010771787,
-      "loss": 46.0,
-      "step": 31159
-    },
-    {
-      "epoch": 2.3823996024236864,
-      "grad_norm": 0.0016779318684712052,
-      "learning_rate": 0.00019999720089746475,
-      "loss": 46.0,
-      "step": 31160
-    },
-    {
-      "epoch": 2.382476059407076,
-      "grad_norm": 0.0023215990513563156,
-      "learning_rate": 0.00019999720071774504,
-      "loss": 46.0,
-      "step": 31161
-    },
-    {
-      "epoch": 2.382552516390466,
-      "grad_norm": 0.0011817141203209758,
-      "learning_rate": 0.00019999720053801956,
-      "loss": 46.0,
-      "step": 31162
-    },
-    {
-      "epoch": 2.3826289733738557,
-      "grad_norm": 0.0021164584904909134,
-      "learning_rate": 0.0001999972003582883,
-      "loss": 46.0,
-      "step": 31163
-    },
-    {
-      "epoch": 2.3827054303572455,
-      "grad_norm": 0.00023572304053232074,
-      "learning_rate": 0.0001999972001785513,
-      "loss": 46.0,
-      "step": 31164
-    },
-    {
-      "epoch": 2.382781887340635,
-      "grad_norm": 0.001020329655148089,
-      "learning_rate": 0.00019999719999880852,
-      "loss": 46.0,
-      "step": 31165
-    },
-    {
-      "epoch": 2.3828583443240245,
-      "grad_norm": 0.0002638470323290676,
-      "learning_rate": 0.00019999719981905994,
-      "loss": 46.0,
-      "step": 31166
-    },
-    {
-      "epoch": 2.3829348013074143,
-      "grad_norm": 0.0009640008211135864,
-      "learning_rate": 0.00019999719963930565,
-      "loss": 46.0,
-      "step": 31167
-    },
-    {
-      "epoch": 2.383011258290804,
-      "grad_norm": 0.0006028738571330905,
-      "learning_rate": 0.00019999719945954555,
-      "loss": 46.0,
-      "step": 31168
-    },
-    {
-      "epoch": 2.383087715274194,
-      "grad_norm": 0.004470419138669968,
-      "learning_rate": 0.00019999719927977968,
-      "loss": 46.0,
-      "step": 31169
-    },
-    {
-      "epoch": 2.3831641722575836,
-      "grad_norm": 0.0014518130337819457,
-      "learning_rate": 0.00019999719910000806,
-      "loss": 46.0,
-      "step": 31170
-    },
-    {
-      "epoch": 2.3832406292409734,
-      "grad_norm": 0.004461311735212803,
-      "learning_rate": 0.00019999719892023067,
-      "loss": 46.0,
-      "step": 31171
-    },
-    {
-      "epoch": 2.383317086224363,
-      "grad_norm": 0.0008494371431879699,
-      "learning_rate": 0.0001999971987404475,
-      "loss": 46.0,
-      "step": 31172
-    },
-    {
-      "epoch": 2.383393543207753,
-      "grad_norm": 0.0008391728042624891,
-      "learning_rate": 0.00019999719856065857,
-      "loss": 46.0,
-      "step": 31173
-    },
-    {
-      "epoch": 2.3834700001911426,
-      "grad_norm": 0.0011764371301978827,
-      "learning_rate": 0.00019999719838086386,
-      "loss": 46.0,
-      "step": 31174
-    },
-    {
-      "epoch": 2.383546457174532,
-      "grad_norm": 0.0017578359693288803,
-      "learning_rate": 0.00019999719820106338,
-      "loss": 46.0,
-      "step": 31175
-    },
-    {
-      "epoch": 2.3836229141579217,
-      "grad_norm": 0.0004938683705404401,
-      "learning_rate": 0.00019999719802125715,
-      "loss": 46.0,
-      "step": 31176
-    },
-    {
-      "epoch": 2.3836993711413115,
-      "grad_norm": 0.000595057150349021,
-      "learning_rate": 0.00019999719784144512,
-      "loss": 46.0,
-      "step": 31177
-    },
-    {
-      "epoch": 2.3837758281247012,
-      "grad_norm": 0.0013430517865344882,
-      "learning_rate": 0.00019999719766162734,
-      "loss": 46.0,
-      "step": 31178
-    },
-    {
-      "epoch": 2.383852285108091,
-      "grad_norm": 0.0007416472653858364,
-      "learning_rate": 0.0001999971974818038,
-      "loss": 46.0,
-      "step": 31179
-    },
-    {
-      "epoch": 2.3839287420914808,
-      "grad_norm": 0.0010170770110562444,
-      "learning_rate": 0.00019999719730197447,
-      "loss": 46.0,
-      "step": 31180
-    },
-    {
-      "epoch": 2.3840051990748705,
-      "grad_norm": 0.0012651252327486873,
-      "learning_rate": 0.00019999719712213937,
-      "loss": 46.0,
-      "step": 31181
-    },
-    {
-      "epoch": 2.3840816560582603,
-      "grad_norm": 0.0011170330690219998,
-      "learning_rate": 0.0001999971969422985,
-      "loss": 46.0,
-      "step": 31182
-    },
-    {
-      "epoch": 2.38415811304165,
-      "grad_norm": 0.0007763446774333715,
-      "learning_rate": 0.0001999971967624519,
-      "loss": 46.0,
-      "step": 31183
-    },
-    {
-      "epoch": 2.38423457002504,
-      "grad_norm": 0.002614803845062852,
-      "learning_rate": 0.00019999719658259947,
-      "loss": 46.0,
-      "step": 31184
-    },
-    {
-      "epoch": 2.3843110270084296,
-      "grad_norm": 0.001550038461573422,
-      "learning_rate": 0.00019999719640274134,
-      "loss": 46.0,
-      "step": 31185
-    },
-    {
-      "epoch": 2.3843874839918193,
-      "grad_norm": 0.0008787338738329709,
-      "learning_rate": 0.00019999719622287737,
-      "loss": 46.0,
-      "step": 31186
-    },
-    {
-      "epoch": 2.3844639409752086,
-      "grad_norm": 0.0003545415820553899,
-      "learning_rate": 0.0001999971960430077,
-      "loss": 46.0,
-      "step": 31187
-    },
-    {
-      "epoch": 2.3845403979585984,
-      "grad_norm": 0.0014985463349148631,
-      "learning_rate": 0.00019999719586313218,
-      "loss": 46.0,
-      "step": 31188
-    },
-    {
-      "epoch": 2.384616854941988,
-      "grad_norm": 0.0008877535001374781,
-      "learning_rate": 0.00019999719568325098,
-      "loss": 46.0,
-      "step": 31189
-    },
-    {
-      "epoch": 2.384693311925378,
-      "grad_norm": 0.0011304514482617378,
-      "learning_rate": 0.00019999719550336392,
-      "loss": 46.0,
-      "step": 31190
-    },
-    {
-      "epoch": 2.3847697689087677,
-      "grad_norm": 0.0015305752167478204,
-      "learning_rate": 0.00019999719532347117,
-      "loss": 46.0,
-      "step": 31191
-    },
-    {
-      "epoch": 2.3848462258921574,
-      "grad_norm": 0.0006367014721035957,
-      "learning_rate": 0.0001999971951435726,
-      "loss": 46.0,
-      "step": 31192
-    },
-    {
-      "epoch": 2.384922682875547,
-      "grad_norm": 0.0004974103649146855,
-      "learning_rate": 0.0001999971949636683,
-      "loss": 46.0,
-      "step": 31193
-    },
-    {
-      "epoch": 2.384999139858937,
-      "grad_norm": 0.004650414921343327,
-      "learning_rate": 0.00019999719478375818,
-      "loss": 46.0,
-      "step": 31194
-    },
-    {
-      "epoch": 2.3850755968423267,
-      "grad_norm": 0.0012259904760867357,
-      "learning_rate": 0.0001999971946038423,
-      "loss": 46.0,
-      "step": 31195
-    },
-    {
-      "epoch": 2.3851520538257165,
-      "grad_norm": 0.0017547912430018187,
-      "learning_rate": 0.0001999971944239207,
-      "loss": 46.0,
-      "step": 31196
-    },
-    {
-      "epoch": 2.385228510809106,
-      "grad_norm": 0.000748968857806176,
-      "learning_rate": 0.0001999971942439933,
-      "loss": 46.0,
-      "step": 31197
-    },
-    {
-      "epoch": 2.3853049677924956,
-      "grad_norm": 0.001265846542082727,
-      "learning_rate": 0.00019999719406406014,
-      "loss": 46.0,
-      "step": 31198
-    },
-    {
-      "epoch": 2.3853814247758853,
-      "grad_norm": 0.0012564935022965074,
-      "learning_rate": 0.0001999971938841212,
-      "loss": 46.0,
-      "step": 31199
-    },
-    {
-      "epoch": 2.385457881759275,
-      "grad_norm": 0.0007321368320845068,
-      "learning_rate": 0.0001999971937041765,
-      "loss": 46.0,
-      "step": 31200
-    },
-    {
-      "epoch": 2.385534338742665,
-      "grad_norm": 0.001971021294593811,
-      "learning_rate": 0.000199997193524226,
-      "loss": 46.0,
-      "step": 31201
-    },
-    {
-      "epoch": 2.3856107957260546,
-      "grad_norm": 0.0019034226424992085,
-      "learning_rate": 0.00019999719334426976,
-      "loss": 46.0,
-      "step": 31202
-    },
-    {
-      "epoch": 2.3856872527094444,
-      "grad_norm": 0.0008287046803161502,
-      "learning_rate": 0.00019999719316430773,
-      "loss": 46.0,
-      "step": 31203
-    },
-    {
-      "epoch": 2.385763709692834,
-      "grad_norm": 0.0008830353617668152,
-      "learning_rate": 0.00019999719298433998,
-      "loss": 46.0,
-      "step": 31204
-    },
-    {
-      "epoch": 2.385840166676224,
-      "grad_norm": 0.0006927649956196547,
-      "learning_rate": 0.00019999719280436643,
-      "loss": 46.0,
-      "step": 31205
-    },
-    {
-      "epoch": 2.3859166236596137,
-      "grad_norm": 0.0009765359573066235,
-      "learning_rate": 0.0001999971926243871,
-      "loss": 46.0,
-      "step": 31206
-    },
-    {
-      "epoch": 2.3859930806430034,
-      "grad_norm": 0.0032678490970283747,
-      "learning_rate": 0.000199997192444402,
-      "loss": 46.0,
-      "step": 31207
-    },
-    {
-      "epoch": 2.386069537626393,
-      "grad_norm": 0.000645723775960505,
-      "learning_rate": 0.00019999719226441114,
-      "loss": 46.0,
-      "step": 31208
-    },
-    {
-      "epoch": 2.3861459946097825,
-      "grad_norm": 0.0046941922046244144,
-      "learning_rate": 0.0001999971920844145,
-      "loss": 46.0,
-      "step": 31209
-    },
-    {
-      "epoch": 2.3862224515931723,
-      "grad_norm": 0.0006700173835270107,
-      "learning_rate": 0.0001999971919044121,
-      "loss": 46.0,
-      "step": 31210
-    },
-    {
-      "epoch": 2.386298908576562,
-      "grad_norm": 0.0007795327110216022,
-      "learning_rate": 0.00019999719172440395,
-      "loss": 46.0,
-      "step": 31211
-    },
-    {
-      "epoch": 2.3863753655599518,
-      "grad_norm": 0.001132732955738902,
-      "learning_rate": 0.00019999719154438998,
-      "loss": 46.0,
-      "step": 31212
-    },
-    {
-      "epoch": 2.3864518225433415,
-      "grad_norm": 0.0007726813782937825,
-      "learning_rate": 0.0001999971913643703,
-      "loss": 46.0,
-      "step": 31213
-    },
-    {
-      "epoch": 2.3865282795267313,
-      "grad_norm": 0.000650379981379956,
-      "learning_rate": 0.00019999719118434482,
-      "loss": 46.0,
-      "step": 31214
-    },
-    {
-      "epoch": 2.386604736510121,
-      "grad_norm": 0.0025822732131928205,
-      "learning_rate": 0.00019999719100431356,
-      "loss": 46.0,
-      "step": 31215
-    },
-    {
-      "epoch": 2.386681193493511,
-      "grad_norm": 0.0006342643173411489,
-      "learning_rate": 0.00019999719082427654,
-      "loss": 46.0,
-      "step": 31216
-    },
-    {
-      "epoch": 2.3867576504769006,
-      "grad_norm": 0.0015285132685676217,
-      "learning_rate": 0.00019999719064423376,
-      "loss": 46.0,
-      "step": 31217
-    },
-    {
-      "epoch": 2.3868341074602903,
-      "grad_norm": 0.0009955320274457335,
-      "learning_rate": 0.0001999971904641852,
-      "loss": 46.0,
-      "step": 31218
-    },
-    {
-      "epoch": 2.3869105644436797,
-      "grad_norm": 0.0007325439946725965,
-      "learning_rate": 0.0001999971902841309,
-      "loss": 46.0,
-      "step": 31219
-    },
-    {
-      "epoch": 2.3869870214270694,
-      "grad_norm": 0.0027244549710303545,
-      "learning_rate": 0.0001999971901040708,
-      "loss": 46.0,
-      "step": 31220
-    },
-    {
-      "epoch": 2.387063478410459,
-      "grad_norm": 0.0036848834715783596,
-      "learning_rate": 0.00019999718992400493,
-      "loss": 46.0,
-      "step": 31221
-    },
-    {
-      "epoch": 2.387139935393849,
-      "grad_norm": 0.002210303209722042,
-      "learning_rate": 0.0001999971897439333,
-      "loss": 46.0,
-      "step": 31222
-    },
-    {
-      "epoch": 2.3872163923772387,
-      "grad_norm": 0.002620500046759844,
-      "learning_rate": 0.0001999971895638559,
-      "loss": 46.0,
-      "step": 31223
-    },
-    {
-      "epoch": 2.3872928493606285,
-      "grad_norm": 0.0015748973237350583,
-      "learning_rate": 0.00019999718938377273,
-      "loss": 46.0,
-      "step": 31224
-    },
-    {
-      "epoch": 2.3873693063440182,
-      "grad_norm": 0.0005100901471450925,
-      "learning_rate": 0.00019999718920368377,
-      "loss": 46.0,
-      "step": 31225
-    },
-    {
-      "epoch": 2.387445763327408,
-      "grad_norm": 0.0011751210549846292,
-      "learning_rate": 0.0001999971890235891,
-      "loss": 46.0,
-      "step": 31226
-    },
-    {
-      "epoch": 2.3875222203107977,
-      "grad_norm": 0.0005510689807124436,
-      "learning_rate": 0.00019999718884348858,
-      "loss": 46.0,
-      "step": 31227
-    },
-    {
-      "epoch": 2.3875986772941875,
-      "grad_norm": 0.0006459723808802664,
-      "learning_rate": 0.00019999718866338235,
-      "loss": 46.0,
-      "step": 31228
-    },
-    {
-      "epoch": 2.3876751342775773,
-      "grad_norm": 0.00062156020430848,
-      "learning_rate": 0.00019999718848327035,
-      "loss": 46.0,
-      "step": 31229
-    },
-    {
-      "epoch": 2.387751591260967,
-      "grad_norm": 0.0003964526695199311,
-      "learning_rate": 0.00019999718830315255,
-      "loss": 46.0,
-      "step": 31230
-    },
-    {
-      "epoch": 2.3878280482443563,
-      "grad_norm": 0.0015967487124726176,
-      "learning_rate": 0.000199997188123029,
-      "loss": 46.0,
-      "step": 31231
-    },
-    {
-      "epoch": 2.387904505227746,
-      "grad_norm": 0.0016473009018227458,
-      "learning_rate": 0.00019999718794289968,
-      "loss": 46.0,
-      "step": 31232
-    },
-    {
-      "epoch": 2.387980962211136,
-      "grad_norm": 0.000670640030875802,
-      "learning_rate": 0.00019999718776276458,
-      "loss": 46.0,
-      "step": 31233
-    },
-    {
-      "epoch": 2.3880574191945256,
-      "grad_norm": 0.0016202578553929925,
-      "learning_rate": 0.0001999971875826237,
-      "loss": 46.0,
-      "step": 31234
-    },
-    {
-      "epoch": 2.3881338761779154,
-      "grad_norm": 0.0014540303964167833,
-      "learning_rate": 0.00019999718740247707,
-      "loss": 46.0,
-      "step": 31235
-    },
-    {
-      "epoch": 2.388210333161305,
-      "grad_norm": 0.009749515913426876,
-      "learning_rate": 0.00019999718722232468,
-      "loss": 46.0,
-      "step": 31236
-    },
-    {
-      "epoch": 2.388286790144695,
-      "grad_norm": 0.0029620288405567408,
-      "learning_rate": 0.0001999971870421665,
-      "loss": 46.0,
-      "step": 31237
-    },
-    {
-      "epoch": 2.3883632471280847,
-      "grad_norm": 0.0009739661472849548,
-      "learning_rate": 0.00019999718686200256,
-      "loss": 46.0,
-      "step": 31238
-    },
-    {
-      "epoch": 2.3884397041114744,
-      "grad_norm": 0.0006199827766977251,
-      "learning_rate": 0.00019999718668183285,
-      "loss": 46.0,
-      "step": 31239
-    },
-    {
-      "epoch": 2.3885161610948638,
-      "grad_norm": 0.000636140292044729,
-      "learning_rate": 0.00019999718650165737,
-      "loss": 46.0,
-      "step": 31240
-    },
-    {
-      "epoch": 2.3885926180782535,
-      "grad_norm": 0.00253242626786232,
-      "learning_rate": 0.00019999718632147611,
-      "loss": 46.0,
-      "step": 31241
-    },
-    {
-      "epoch": 2.3886690750616433,
-      "grad_norm": 0.0029478955548256636,
-      "learning_rate": 0.0001999971861412891,
-      "loss": 46.0,
-      "step": 31242
-    },
-    {
-      "epoch": 2.388745532045033,
-      "grad_norm": 0.001852401182986796,
-      "learning_rate": 0.00019999718596109634,
-      "loss": 46.0,
-      "step": 31243
-    },
-    {
-      "epoch": 2.388821989028423,
-      "grad_norm": 0.000414897920563817,
-      "learning_rate": 0.00019999718578089776,
-      "loss": 46.0,
-      "step": 31244
-    },
-    {
-      "epoch": 2.3888984460118126,
-      "grad_norm": 0.0010077353799715638,
-      "learning_rate": 0.00019999718560069341,
-      "loss": 46.0,
-      "step": 31245
-    },
-    {
-      "epoch": 2.3889749029952023,
-      "grad_norm": 0.0010178263764828444,
-      "learning_rate": 0.00019999718542048335,
-      "loss": 46.0,
-      "step": 31246
-    },
-    {
-      "epoch": 2.389051359978592,
-      "grad_norm": 0.000524612027220428,
-      "learning_rate": 0.00019999718524026748,
-      "loss": 46.0,
-      "step": 31247
-    },
-    {
-      "epoch": 2.389127816961982,
-      "grad_norm": 0.001703913207165897,
-      "learning_rate": 0.00019999718506004586,
-      "loss": 46.0,
-      "step": 31248
-    },
-    {
-      "epoch": 2.3892042739453716,
-      "grad_norm": 0.0011087754974141717,
-      "learning_rate": 0.00019999718487981845,
-      "loss": 46.0,
-      "step": 31249
-    },
-    {
-      "epoch": 2.3892807309287614,
-      "grad_norm": 0.0014830634463578463,
-      "learning_rate": 0.00019999718469958526,
-      "loss": 46.0,
-      "step": 31250
-    },
-    {
-      "epoch": 2.389357187912151,
-      "grad_norm": 0.0029669131617993116,
-      "learning_rate": 0.00019999718451934633,
-      "loss": 46.0,
-      "step": 31251
-    },
-    {
-      "epoch": 2.389433644895541,
-      "grad_norm": 0.0004268599150236696,
-      "learning_rate": 0.00019999718433910162,
-      "loss": 46.0,
-      "step": 31252
-    },
-    {
-      "epoch": 2.38951010187893,
-      "grad_norm": 0.000671941670589149,
-      "learning_rate": 0.00019999718415885114,
-      "loss": 46.0,
-      "step": 31253
-    },
-    {
-      "epoch": 2.38958655886232,
-      "grad_norm": 0.0008306749514304101,
-      "learning_rate": 0.00019999718397859488,
-      "loss": 46.0,
-      "step": 31254
-    },
-    {
-      "epoch": 2.3896630158457097,
-      "grad_norm": 0.0003019349533133209,
-      "learning_rate": 0.00019999718379833288,
-      "loss": 46.0,
-      "step": 31255
-    },
-    {
-      "epoch": 2.3897394728290995,
-      "grad_norm": 0.0019008255330845714,
-      "learning_rate": 0.0001999971836180651,
-      "loss": 46.0,
-      "step": 31256
-    },
-    {
-      "epoch": 2.3898159298124892,
-      "grad_norm": 0.001142167137004435,
-      "learning_rate": 0.00019999718343779153,
-      "loss": 46.0,
-      "step": 31257
-    },
-    {
-      "epoch": 2.389892386795879,
-      "grad_norm": 0.0017596747493371367,
-      "learning_rate": 0.0001999971832575122,
-      "loss": 46.0,
-      "step": 31258
-    },
-    {
-      "epoch": 2.3899688437792688,
-      "grad_norm": 0.0011836143676191568,
-      "learning_rate": 0.00019999718307722711,
-      "loss": 46.0,
-      "step": 31259
-    },
-    {
-      "epoch": 2.3900453007626585,
-      "grad_norm": 0.0012896500993520021,
-      "learning_rate": 0.00019999718289693625,
-      "loss": 46.0,
-      "step": 31260
-    },
-    {
-      "epoch": 2.3901217577460483,
-      "grad_norm": 0.0018199884798377752,
-      "learning_rate": 0.0001999971827166396,
-      "loss": 46.0,
-      "step": 31261
-    },
-    {
-      "epoch": 2.3901982147294376,
-      "grad_norm": 0.0008378308266401291,
-      "learning_rate": 0.0001999971825363372,
-      "loss": 46.0,
-      "step": 31262
-    },
-    {
-      "epoch": 2.3902746717128274,
-      "grad_norm": 0.0016890474362298846,
-      "learning_rate": 0.00019999718235602903,
-      "loss": 46.0,
-      "step": 31263
-    },
-    {
-      "epoch": 2.390351128696217,
-      "grad_norm": 0.004787810146808624,
-      "learning_rate": 0.00019999718217571507,
-      "loss": 46.0,
-      "step": 31264
-    },
-    {
-      "epoch": 2.390427585679607,
-      "grad_norm": 0.00042374138138256967,
-      "learning_rate": 0.00019999718199539536,
-      "loss": 46.0,
-      "step": 31265
-    },
-    {
-      "epoch": 2.3905040426629967,
-      "grad_norm": 0.0007421452319249511,
-      "learning_rate": 0.00019999718181506988,
-      "loss": 46.0,
-      "step": 31266
-    },
-    {
-      "epoch": 2.3905804996463864,
-      "grad_norm": 0.0014471496688202024,
-      "learning_rate": 0.00019999718163473865,
-      "loss": 46.0,
-      "step": 31267
-    },
-    {
-      "epoch": 2.390656956629776,
-      "grad_norm": 0.000660435005556792,
-      "learning_rate": 0.00019999718145440162,
-      "loss": 46.0,
-      "step": 31268
-    },
-    {
-      "epoch": 2.390733413613166,
-      "grad_norm": 0.00048395595513284206,
-      "learning_rate": 0.00019999718127405885,
-      "loss": 46.0,
-      "step": 31269
-    },
-    {
-      "epoch": 2.3908098705965557,
-      "grad_norm": 0.0012016019318252802,
-      "learning_rate": 0.00019999718109371028,
-      "loss": 46.0,
-      "step": 31270
-    },
-    {
-      "epoch": 2.3908863275799455,
-      "grad_norm": 0.0011572001967579126,
-      "learning_rate": 0.00019999718091335595,
-      "loss": 46.0,
-      "step": 31271
-    },
-    {
-      "epoch": 2.390962784563335,
-      "grad_norm": 0.0034466516226530075,
-      "learning_rate": 0.00019999718073299586,
-      "loss": 46.0,
-      "step": 31272
-    },
-    {
-      "epoch": 2.391039241546725,
-      "grad_norm": 0.0009810482151806355,
-      "learning_rate": 0.00019999718055263,
-      "loss": 46.0,
-      "step": 31273
-    },
-    {
-      "epoch": 2.3911156985301147,
-      "grad_norm": 0.0008015355560928583,
-      "learning_rate": 0.00019999718037225835,
-      "loss": 46.0,
-      "step": 31274
-    },
-    {
-      "epoch": 2.391192155513504,
-      "grad_norm": 0.0008540457347407937,
-      "learning_rate": 0.00019999718019188096,
-      "loss": 46.0,
-      "step": 31275
-    },
-    {
-      "epoch": 2.391268612496894,
-      "grad_norm": 0.0015199249610304832,
-      "learning_rate": 0.00019999718001149778,
-      "loss": 46.0,
-      "step": 31276
-    },
-    {
-      "epoch": 2.3913450694802836,
-      "grad_norm": 0.001088008750230074,
-      "learning_rate": 0.00019999717983110884,
-      "loss": 46.0,
-      "step": 31277
-    },
-    {
-      "epoch": 2.3914215264636733,
-      "grad_norm": 0.0005309219122864306,
-      "learning_rate": 0.0001999971796507141,
-      "loss": 46.0,
-      "step": 31278
-    },
-    {
-      "epoch": 2.391497983447063,
-      "grad_norm": 0.0019616049248725176,
-      "learning_rate": 0.00019999717947031365,
-      "loss": 46.0,
-      "step": 31279
-    },
-    {
-      "epoch": 2.391574440430453,
-      "grad_norm": 0.007848215289413929,
-      "learning_rate": 0.0001999971792899074,
-      "loss": 46.0,
-      "step": 31280
-    },
-    {
-      "epoch": 2.3916508974138426,
-      "grad_norm": 0.0005991210928186774,
-      "learning_rate": 0.00019999717910949535,
-      "loss": 46.0,
-      "step": 31281
-    },
-    {
-      "epoch": 2.3917273543972324,
-      "grad_norm": 0.0008754156297072768,
-      "learning_rate": 0.0001999971789290776,
-      "loss": 46.0,
-      "step": 31282
-    },
-    {
-      "epoch": 2.391803811380622,
-      "grad_norm": 0.0019264626316726208,
-      "learning_rate": 0.00019999717874865402,
-      "loss": 46.0,
-      "step": 31283
-    },
-    {
-      "epoch": 2.3918802683640115,
-      "grad_norm": 0.0007541723316535354,
-      "learning_rate": 0.00019999717856822468,
-      "loss": 46.0,
-      "step": 31284
-    },
-    {
-      "epoch": 2.3919567253474012,
-      "grad_norm": 0.0006770867039449513,
-      "learning_rate": 0.0001999971783877896,
-      "loss": 46.0,
-      "step": 31285
-    },
-    {
-      "epoch": 2.392033182330791,
-      "grad_norm": 0.001182405510917306,
-      "learning_rate": 0.00019999717820734874,
-      "loss": 46.0,
-      "step": 31286
-    },
-    {
-      "epoch": 2.3921096393141807,
-      "grad_norm": 0.0008292681304737926,
-      "learning_rate": 0.0001999971780269021,
-      "loss": 46.0,
-      "step": 31287
-    },
-    {
-      "epoch": 2.3921860962975705,
-      "grad_norm": 0.0008599006687290967,
-      "learning_rate": 0.0001999971778464497,
-      "loss": 46.0,
-      "step": 31288
-    },
-    {
-      "epoch": 2.3922625532809603,
-      "grad_norm": 0.0004576443461701274,
-      "learning_rate": 0.0001999971776659915,
-      "loss": 46.0,
-      "step": 31289
-    },
-    {
-      "epoch": 2.39233901026435,
-      "grad_norm": 0.001664156443439424,
-      "learning_rate": 0.00019999717748552757,
-      "loss": 46.0,
-      "step": 31290
-    },
-    {
-      "epoch": 2.39241546724774,
-      "grad_norm": 0.0005999609711579978,
-      "learning_rate": 0.00019999717730505786,
-      "loss": 46.0,
-      "step": 31291
-    },
-    {
-      "epoch": 2.3924919242311296,
-      "grad_norm": 0.0020037388894706964,
-      "learning_rate": 0.00019999717712458238,
-      "loss": 46.0,
-      "step": 31292
-    },
-    {
-      "epoch": 2.3925683812145193,
-      "grad_norm": 0.0004422641941346228,
-      "learning_rate": 0.00019999717694410115,
-      "loss": 46.0,
-      "step": 31293
-    },
-    {
-      "epoch": 2.392644838197909,
-      "grad_norm": 0.0010603275150060654,
-      "learning_rate": 0.0001999971767636141,
-      "loss": 46.0,
-      "step": 31294
-    },
-    {
-      "epoch": 2.392721295181299,
-      "grad_norm": 0.001079891575500369,
-      "learning_rate": 0.00019999717658312132,
-      "loss": 46.0,
-      "step": 31295
-    },
-    {
-      "epoch": 2.392797752164688,
-      "grad_norm": 0.004955118987709284,
-      "learning_rate": 0.00019999717640262278,
-      "loss": 46.0,
-      "step": 31296
-    },
-    {
-      "epoch": 2.392874209148078,
-      "grad_norm": 0.0013935695169493556,
-      "learning_rate": 0.00019999717622211843,
-      "loss": 46.0,
-      "step": 31297
-    },
-    {
-      "epoch": 2.3929506661314677,
-      "grad_norm": 0.0029185789171606302,
-      "learning_rate": 0.00019999717604160834,
-      "loss": 46.0,
-      "step": 31298
-    },
-    {
-      "epoch": 2.3930271231148574,
-      "grad_norm": 0.0003364873700775206,
-      "learning_rate": 0.00019999717586109247,
-      "loss": 46.0,
-      "step": 31299
-    },
-    {
-      "epoch": 2.393103580098247,
-      "grad_norm": 0.002857153071090579,
-      "learning_rate": 0.00019999717568057083,
-      "loss": 46.0,
-      "step": 31300
-    },
-    {
-      "epoch": 2.393180037081637,
-      "grad_norm": 0.0005324921803548932,
-      "learning_rate": 0.00019999717550004342,
-      "loss": 46.0,
-      "step": 31301
-    },
-    {
-      "epoch": 2.3932564940650267,
-      "grad_norm": 0.003195501398295164,
-      "learning_rate": 0.00019999717531951023,
-      "loss": 46.0,
-      "step": 31302
-    },
-    {
-      "epoch": 2.3933329510484165,
-      "grad_norm": 0.0012877669651061296,
-      "learning_rate": 0.0001999971751389713,
-      "loss": 46.0,
-      "step": 31303
-    },
-    {
-      "epoch": 2.3934094080318062,
-      "grad_norm": 0.0008315282175317407,
-      "learning_rate": 0.0001999971749584266,
-      "loss": 46.0,
-      "step": 31304
-    },
-    {
-      "epoch": 2.393485865015196,
-      "grad_norm": 0.0009507263312116265,
-      "learning_rate": 0.0001999971747778761,
-      "loss": 46.0,
-      "step": 31305
-    },
-    {
-      "epoch": 2.3935623219985853,
-      "grad_norm": 0.002657354576513171,
-      "learning_rate": 0.00019999717459731983,
-      "loss": 46.0,
-      "step": 31306
-    },
-    {
-      "epoch": 2.393638778981975,
-      "grad_norm": 0.002530923578888178,
-      "learning_rate": 0.00019999717441675783,
-      "loss": 46.0,
-      "step": 31307
-    },
-    {
-      "epoch": 2.393715235965365,
-      "grad_norm": 0.0006451071822084486,
-      "learning_rate": 0.00019999717423619005,
-      "loss": 46.0,
-      "step": 31308
-    },
-    {
-      "epoch": 2.3937916929487546,
-      "grad_norm": 0.004086668603122234,
-      "learning_rate": 0.00019999717405561648,
-      "loss": 46.0,
-      "step": 31309
-    },
-    {
-      "epoch": 2.3938681499321444,
-      "grad_norm": 0.0005853670299984515,
-      "learning_rate": 0.00019999717387503716,
-      "loss": 46.0,
-      "step": 31310
-    },
-    {
-      "epoch": 2.393944606915534,
-      "grad_norm": 0.001546251936815679,
-      "learning_rate": 0.00019999717369445207,
-      "loss": 46.0,
-      "step": 31311
-    },
-    {
-      "epoch": 2.394021063898924,
-      "grad_norm": 0.00194163725245744,
-      "learning_rate": 0.00019999717351386117,
-      "loss": 46.0,
-      "step": 31312
-    },
-    {
-      "epoch": 2.3940975208823136,
-      "grad_norm": 0.0006438237614929676,
-      "learning_rate": 0.00019999717333326453,
-      "loss": 46.0,
-      "step": 31313
-    },
-    {
-      "epoch": 2.3941739778657034,
-      "grad_norm": 0.0033377467188984156,
-      "learning_rate": 0.00019999717315266212,
-      "loss": 46.0,
-      "step": 31314
-    },
-    {
-      "epoch": 2.394250434849093,
-      "grad_norm": 0.0006141770863905549,
-      "learning_rate": 0.00019999717297205396,
-      "loss": 46.0,
-      "step": 31315
-    },
-    {
-      "epoch": 2.394326891832483,
-      "grad_norm": 0.0017294621793553233,
-      "learning_rate": 0.00019999717279144,
-      "loss": 46.0,
-      "step": 31316
-    },
-    {
-      "epoch": 2.3944033488158727,
-      "grad_norm": 0.00232842774130404,
-      "learning_rate": 0.0001999971726108203,
-      "loss": 46.0,
-      "step": 31317
-    },
-    {
-      "epoch": 2.394479805799262,
-      "grad_norm": 0.000787562457844615,
-      "learning_rate": 0.0001999971724301948,
-      "loss": 46.0,
-      "step": 31318
-    },
-    {
-      "epoch": 2.3945562627826518,
-      "grad_norm": 0.002118197735399008,
-      "learning_rate": 0.0001999971722495636,
-      "loss": 46.0,
-      "step": 31319
-    },
-    {
-      "epoch": 2.3946327197660415,
-      "grad_norm": 0.0007089497521519661,
-      "learning_rate": 0.00019999717206892653,
-      "loss": 46.0,
-      "step": 31320
-    },
-    {
-      "epoch": 2.3947091767494313,
-      "grad_norm": 0.0035323509946465492,
-      "learning_rate": 0.00019999717188828376,
-      "loss": 46.0,
-      "step": 31321
-    },
-    {
-      "epoch": 2.394785633732821,
-      "grad_norm": 0.0014900483656674623,
-      "learning_rate": 0.0001999971717076352,
-      "loss": 46.0,
-      "step": 31322
-    },
-    {
-      "epoch": 2.394862090716211,
-      "grad_norm": 0.00037516685551963747,
-      "learning_rate": 0.00019999717152698084,
-      "loss": 46.0,
-      "step": 31323
-    },
-    {
-      "epoch": 2.3949385476996006,
-      "grad_norm": 0.001524752820841968,
-      "learning_rate": 0.00019999717134632077,
-      "loss": 46.0,
-      "step": 31324
-    },
-    {
-      "epoch": 2.3950150046829903,
-      "grad_norm": 0.001080443151295185,
-      "learning_rate": 0.00019999717116565488,
-      "loss": 46.0,
-      "step": 31325
-    },
-    {
-      "epoch": 2.39509146166638,
-      "grad_norm": 0.0011418763315305114,
-      "learning_rate": 0.00019999717098498324,
-      "loss": 46.0,
-      "step": 31326
-    },
-    {
-      "epoch": 2.39516791864977,
-      "grad_norm": 0.0006253799074329436,
-      "learning_rate": 0.00019999717080430586,
-      "loss": 46.0,
-      "step": 31327
-    },
-    {
-      "epoch": 2.395244375633159,
-      "grad_norm": 0.0028666167054325342,
-      "learning_rate": 0.00019999717062362267,
-      "loss": 46.0,
-      "step": 31328
-    },
-    {
-      "epoch": 2.395320832616549,
-      "grad_norm": 0.0007449330296367407,
-      "learning_rate": 0.0001999971704429337,
-      "loss": 46.0,
-      "step": 31329
-    },
-    {
-      "epoch": 2.3953972895999387,
-      "grad_norm": 0.001074870233424008,
-      "learning_rate": 0.000199997170262239,
-      "loss": 46.0,
-      "step": 31330
-    },
-    {
-      "epoch": 2.3954737465833285,
-      "grad_norm": 0.0013929917477071285,
-      "learning_rate": 0.00019999717008153852,
-      "loss": 46.0,
-      "step": 31331
-    },
-    {
-      "epoch": 2.395550203566718,
-      "grad_norm": 0.0007614702917635441,
-      "learning_rate": 0.00019999716990083227,
-      "loss": 46.0,
-      "step": 31332
-    },
-    {
-      "epoch": 2.395626660550108,
-      "grad_norm": 0.0010670649353414774,
-      "learning_rate": 0.00019999716972012027,
-      "loss": 46.0,
-      "step": 31333
-    },
-    {
-      "epoch": 2.3957031175334977,
-      "grad_norm": 0.0053705996833741665,
-      "learning_rate": 0.00019999716953940247,
-      "loss": 46.0,
-      "step": 31334
-    },
-    {
-      "epoch": 2.3957795745168875,
-      "grad_norm": 0.001210033311508596,
-      "learning_rate": 0.0001999971693586789,
-      "loss": 46.0,
-      "step": 31335
-    },
-    {
-      "epoch": 2.3958560315002773,
-      "grad_norm": 0.0006040573352947831,
-      "learning_rate": 0.00019999716917794955,
-      "loss": 46.0,
-      "step": 31336
-    },
-    {
-      "epoch": 2.395932488483667,
-      "grad_norm": 0.0007626016740687191,
-      "learning_rate": 0.00019999716899721446,
-      "loss": 46.0,
-      "step": 31337
-    },
-    {
-      "epoch": 2.396008945467057,
-      "grad_norm": 0.0016253925859928131,
-      "learning_rate": 0.0001999971688164736,
-      "loss": 46.0,
-      "step": 31338
-    },
-    {
-      "epoch": 2.3960854024504465,
-      "grad_norm": 0.0010328862117603421,
-      "learning_rate": 0.00019999716863572695,
-      "loss": 46.0,
-      "step": 31339
-    },
-    {
-      "epoch": 2.396161859433836,
-      "grad_norm": 0.000635054602753371,
-      "learning_rate": 0.00019999716845497454,
-      "loss": 46.0,
-      "step": 31340
-    },
-    {
-      "epoch": 2.3962383164172256,
-      "grad_norm": 0.0006187382969073951,
-      "learning_rate": 0.00019999716827421638,
-      "loss": 46.0,
-      "step": 31341
-    },
-    {
-      "epoch": 2.3963147734006154,
-      "grad_norm": 0.0009499857551418245,
-      "learning_rate": 0.00019999716809345242,
-      "loss": 46.0,
-      "step": 31342
-    },
-    {
-      "epoch": 2.396391230384005,
-      "grad_norm": 0.002331551630049944,
-      "learning_rate": 0.00019999716791268272,
-      "loss": 46.0,
-      "step": 31343
-    },
-    {
-      "epoch": 2.396467687367395,
-      "grad_norm": 0.0027765617705881596,
-      "learning_rate": 0.0001999971677319072,
-      "loss": 46.0,
-      "step": 31344
-    },
-    {
-      "epoch": 2.3965441443507847,
-      "grad_norm": 0.0008973247604444623,
-      "learning_rate": 0.00019999716755112596,
-      "loss": 46.0,
-      "step": 31345
-    },
-    {
-      "epoch": 2.3966206013341744,
-      "grad_norm": 0.0007501794607378542,
-      "learning_rate": 0.00019999716737033893,
-      "loss": 46.0,
-      "step": 31346
-    },
-    {
-      "epoch": 2.396697058317564,
-      "grad_norm": 0.0018190309638157487,
-      "learning_rate": 0.00019999716718954616,
-      "loss": 46.0,
-      "step": 31347
-    },
-    {
-      "epoch": 2.396773515300954,
-      "grad_norm": 0.00607098825275898,
-      "learning_rate": 0.0001999971670087476,
-      "loss": 46.0,
-      "step": 31348
-    },
-    {
-      "epoch": 2.3968499722843437,
-      "grad_norm": 0.0010250707855448127,
-      "learning_rate": 0.00019999716682794324,
-      "loss": 46.0,
-      "step": 31349
-    },
-    {
-      "epoch": 2.396926429267733,
-      "grad_norm": 0.001407513045705855,
-      "learning_rate": 0.00019999716664713315,
-      "loss": 46.0,
-      "step": 31350
-    },
-    {
-      "epoch": 2.397002886251123,
-      "grad_norm": 0.001078450120985508,
-      "learning_rate": 0.00019999716646631728,
-      "loss": 46.0,
-      "step": 31351
-    },
-    {
-      "epoch": 2.3970793432345125,
-      "grad_norm": 0.0011363965459167957,
-      "learning_rate": 0.00019999716628549562,
-      "loss": 46.0,
-      "step": 31352
-    },
-    {
-      "epoch": 2.3971558002179023,
-      "grad_norm": 0.0013813456753268838,
-      "learning_rate": 0.00019999716610466823,
-      "loss": 46.0,
-      "step": 31353
-    },
-    {
-      "epoch": 2.397232257201292,
-      "grad_norm": 0.0007420259062200785,
-      "learning_rate": 0.00019999716592383507,
-      "loss": 46.0,
-      "step": 31354
-    },
-    {
-      "epoch": 2.397308714184682,
-      "grad_norm": 0.0005797110497951508,
-      "learning_rate": 0.0001999971657429961,
-      "loss": 46.0,
-      "step": 31355
-    },
-    {
-      "epoch": 2.3973851711680716,
-      "grad_norm": 0.0020147799514234066,
-      "learning_rate": 0.0001999971655621514,
-      "loss": 46.0,
-      "step": 31356
-    },
-    {
-      "epoch": 2.3974616281514614,
-      "grad_norm": 0.0009422108996659517,
-      "learning_rate": 0.0001999971653813009,
-      "loss": 46.0,
-      "step": 31357
-    },
-    {
-      "epoch": 2.397538085134851,
-      "grad_norm": 0.0028956353198736906,
-      "learning_rate": 0.00019999716520044465,
-      "loss": 46.0,
-      "step": 31358
-    },
-    {
-      "epoch": 2.397614542118241,
-      "grad_norm": 0.0019463779171928763,
-      "learning_rate": 0.00019999716501958262,
-      "loss": 46.0,
-      "step": 31359
-    },
-    {
-      "epoch": 2.3976909991016306,
-      "grad_norm": 0.0006795019144192338,
-      "learning_rate": 0.00019999716483871482,
-      "loss": 46.0,
-      "step": 31360
-    },
-    {
-      "epoch": 2.3977674560850204,
-      "grad_norm": 0.0010647271992638707,
-      "learning_rate": 0.00019999716465784125,
-      "loss": 46.0,
-      "step": 31361
-    },
-    {
-      "epoch": 2.3978439130684097,
-      "grad_norm": 0.004848768934607506,
-      "learning_rate": 0.00019999716447696193,
-      "loss": 46.0,
-      "step": 31362
-    },
-    {
-      "epoch": 2.3979203700517995,
-      "grad_norm": 0.0045928750187158585,
-      "learning_rate": 0.00019999716429607681,
-      "loss": 46.0,
-      "step": 31363
-    },
-    {
-      "epoch": 2.3979968270351892,
-      "grad_norm": 0.0007254926022142172,
-      "learning_rate": 0.00019999716411518598,
-      "loss": 46.0,
-      "step": 31364
-    },
-    {
-      "epoch": 2.398073284018579,
-      "grad_norm": 0.0014428653521463275,
-      "learning_rate": 0.00019999716393428934,
-      "loss": 46.0,
-      "step": 31365
-    },
-    {
-      "epoch": 2.3981497410019688,
-      "grad_norm": 0.002059647347778082,
-      "learning_rate": 0.0001999971637533869,
-      "loss": 46.0,
-      "step": 31366
-    },
-    {
-      "epoch": 2.3982261979853585,
-      "grad_norm": 0.0028466309886425734,
-      "learning_rate": 0.00019999716357247874,
-      "loss": 46.0,
-      "step": 31367
-    },
-    {
-      "epoch": 2.3983026549687483,
-      "grad_norm": 0.0023598556872457266,
-      "learning_rate": 0.00019999716339156478,
-      "loss": 46.0,
-      "step": 31368
-    },
-    {
-      "epoch": 2.398379111952138,
-      "grad_norm": 0.0007187400478869677,
-      "learning_rate": 0.00019999716321064508,
-      "loss": 46.0,
-      "step": 31369
-    },
-    {
-      "epoch": 2.398455568935528,
-      "grad_norm": 0.003509592032060027,
-      "learning_rate": 0.00019999716302971957,
-      "loss": 46.0,
-      "step": 31370
-    },
-    {
-      "epoch": 2.398532025918917,
-      "grad_norm": 0.0012699612416327,
-      "learning_rate": 0.00019999716284878832,
-      "loss": 46.0,
-      "step": 31371
-    },
-    {
-      "epoch": 2.398608482902307,
-      "grad_norm": 0.0013038552133366466,
-      "learning_rate": 0.0001999971626678513,
-      "loss": 46.0,
-      "step": 31372
-    },
-    {
-      "epoch": 2.3986849398856966,
-      "grad_norm": 0.002157720737159252,
-      "learning_rate": 0.0001999971624869085,
-      "loss": 46.0,
-      "step": 31373
-    },
-    {
-      "epoch": 2.3987613968690864,
-      "grad_norm": 0.0009187221294268966,
-      "learning_rate": 0.00019999716230595995,
-      "loss": 46.0,
-      "step": 31374
-    },
-    {
-      "epoch": 2.398837853852476,
-      "grad_norm": 0.0021000653505325317,
-      "learning_rate": 0.00019999716212500558,
-      "loss": 46.0,
-      "step": 31375
-    },
-    {
-      "epoch": 2.398914310835866,
-      "grad_norm": 0.0013093932066112757,
-      "learning_rate": 0.00019999716194404548,
-      "loss": 46.0,
-      "step": 31376
-    },
-    {
-      "epoch": 2.3989907678192557,
-      "grad_norm": 0.0006359801045618951,
-      "learning_rate": 0.00019999716176307962,
-      "loss": 46.0,
-      "step": 31377
-    },
-    {
-      "epoch": 2.3990672248026454,
-      "grad_norm": 0.0007734121754765511,
-      "learning_rate": 0.000199997161582108,
-      "loss": 46.0,
-      "step": 31378
-    },
-    {
-      "epoch": 2.399143681786035,
-      "grad_norm": 0.0008498116512782872,
-      "learning_rate": 0.0001999971614011306,
-      "loss": 46.0,
-      "step": 31379
-    },
-    {
-      "epoch": 2.399220138769425,
-      "grad_norm": 0.004063235595822334,
-      "learning_rate": 0.0001999971612201474,
-      "loss": 46.0,
-      "step": 31380
-    },
-    {
-      "epoch": 2.3992965957528147,
-      "grad_norm": 0.004265876021236181,
-      "learning_rate": 0.00019999716103915845,
-      "loss": 46.0,
-      "step": 31381
-    },
-    {
-      "epoch": 2.3993730527362045,
-      "grad_norm": 0.0007822285988368094,
-      "learning_rate": 0.00019999716085816372,
-      "loss": 46.0,
-      "step": 31382
-    },
-    {
-      "epoch": 2.3994495097195943,
-      "grad_norm": 0.0044850800186395645,
-      "learning_rate": 0.00019999716067716324,
-      "loss": 46.0,
-      "step": 31383
-    },
-    {
-      "epoch": 2.3995259667029836,
-      "grad_norm": 0.0004107136046513915,
-      "learning_rate": 0.000199997160496157,
-      "loss": 46.0,
-      "step": 31384
-    },
-    {
-      "epoch": 2.3996024236863733,
-      "grad_norm": 0.0012389696203172207,
-      "learning_rate": 0.00019999716031514496,
-      "loss": 46.0,
-      "step": 31385
-    },
-    {
-      "epoch": 2.399678880669763,
-      "grad_norm": 0.0011143486481159925,
-      "learning_rate": 0.00019999716013412714,
-      "loss": 46.0,
-      "step": 31386
-    },
-    {
-      "epoch": 2.399755337653153,
-      "grad_norm": 0.001441480708308518,
-      "learning_rate": 0.0001999971599531036,
-      "loss": 46.0,
-      "step": 31387
-    },
-    {
-      "epoch": 2.3998317946365426,
-      "grad_norm": 0.0011616770643740892,
-      "learning_rate": 0.00019999715977207428,
-      "loss": 46.0,
-      "step": 31388
-    },
-    {
-      "epoch": 2.3999082516199324,
-      "grad_norm": 0.0009899999713525176,
-      "learning_rate": 0.00019999715959103919,
-      "loss": 46.0,
-      "step": 31389
-    },
-    {
-      "epoch": 2.399984708603322,
-      "grad_norm": 0.000577994214836508,
-      "learning_rate": 0.0001999971594099983,
-      "loss": 46.0,
-      "step": 31390
-    },
-    {
-      "epoch": 2.400061165586712,
-      "grad_norm": 0.0006987660308368504,
-      "learning_rate": 0.00019999715922895166,
-      "loss": 46.0,
-      "step": 31391
-    },
-    {
-      "epoch": 2.4001376225701017,
-      "grad_norm": 0.0009973723208531737,
-      "learning_rate": 0.00019999715904789924,
-      "loss": 46.0,
-      "step": 31392
-    },
-    {
-      "epoch": 2.400214079553491,
-      "grad_norm": 0.000840356049593538,
-      "learning_rate": 0.00019999715886684106,
-      "loss": 46.0,
-      "step": 31393
-    },
-    {
-      "epoch": 2.4002905365368807,
-      "grad_norm": 0.001468954375013709,
-      "learning_rate": 0.00019999715868577707,
-      "loss": 46.0,
-      "step": 31394
-    },
-    {
-      "epoch": 2.4003669935202705,
-      "grad_norm": 0.0015069774817675352,
-      "learning_rate": 0.0001999971585047074,
-      "loss": 46.0,
-      "step": 31395
-    },
-    {
-      "epoch": 2.4004434505036603,
-      "grad_norm": 0.0023700129240751266,
-      "learning_rate": 0.0001999971583236319,
-      "loss": 46.0,
-      "step": 31396
-    },
-    {
-      "epoch": 2.40051990748705,
-      "grad_norm": 0.005073704291135073,
-      "learning_rate": 0.00019999715814255064,
-      "loss": 46.0,
-      "step": 31397
-    },
-    {
-      "epoch": 2.40059636447044,
-      "grad_norm": 0.00040106047526933253,
-      "learning_rate": 0.00019999715796146361,
-      "loss": 46.0,
-      "step": 31398
-    },
-    {
-      "epoch": 2.4006728214538295,
-      "grad_norm": 0.0010780381271615624,
-      "learning_rate": 0.00019999715778037082,
-      "loss": 46.0,
-      "step": 31399
-    },
-    {
-      "epoch": 2.4007492784372193,
-      "grad_norm": 0.0011490309843793511,
-      "learning_rate": 0.00019999715759927224,
-      "loss": 46.0,
-      "step": 31400
-    },
-    {
-      "epoch": 2.400825735420609,
-      "grad_norm": 0.002880827756598592,
-      "learning_rate": 0.0001999971574181679,
-      "loss": 46.0,
-      "step": 31401
-    },
-    {
-      "epoch": 2.400902192403999,
-      "grad_norm": 0.0007725796313025057,
-      "learning_rate": 0.0001999971572370578,
-      "loss": 46.0,
-      "step": 31402
-    },
-    {
-      "epoch": 2.4009786493873886,
-      "grad_norm": 0.0008708566310815513,
-      "learning_rate": 0.00019999715705594192,
-      "loss": 46.0,
-      "step": 31403
-    },
-    {
-      "epoch": 2.4010551063707783,
-      "grad_norm": 0.0014673697296530008,
-      "learning_rate": 0.00019999715687482028,
-      "loss": 46.0,
-      "step": 31404
-    },
-    {
-      "epoch": 2.401131563354168,
-      "grad_norm": 0.0008105520391836762,
-      "learning_rate": 0.00019999715669369287,
-      "loss": 46.0,
-      "step": 31405
-    },
-    {
-      "epoch": 2.4012080203375574,
-      "grad_norm": 0.0025470424443483353,
-      "learning_rate": 0.0001999971565125597,
-      "loss": 46.0,
-      "step": 31406
-    },
-    {
-      "epoch": 2.401284477320947,
-      "grad_norm": 0.0012722652172669768,
-      "learning_rate": 0.00019999715633142073,
-      "loss": 46.0,
-      "step": 31407
-    },
-    {
-      "epoch": 2.401360934304337,
-      "grad_norm": 0.0007586120627820492,
-      "learning_rate": 0.00019999715615027602,
-      "loss": 46.0,
-      "step": 31408
-    },
-    {
-      "epoch": 2.4014373912877267,
-      "grad_norm": 0.0020619595889002085,
-      "learning_rate": 0.00019999715596912554,
-      "loss": 46.0,
-      "step": 31409
-    },
-    {
-      "epoch": 2.4015138482711165,
-      "grad_norm": 0.0012643811060115695,
-      "learning_rate": 0.00019999715578796927,
-      "loss": 46.0,
-      "step": 31410
-    },
-    {
-      "epoch": 2.4015903052545062,
-      "grad_norm": 0.0006271416204981506,
-      "learning_rate": 0.00019999715560680724,
-      "loss": 46.0,
-      "step": 31411
-    },
-    {
-      "epoch": 2.401666762237896,
-      "grad_norm": 0.0019694161601364613,
-      "learning_rate": 0.00019999715542563947,
-      "loss": 46.0,
-      "step": 31412
-    },
-    {
-      "epoch": 2.4017432192212858,
-      "grad_norm": 0.002762384945526719,
-      "learning_rate": 0.0001999971552444659,
-      "loss": 46.0,
-      "step": 31413
-    },
-    {
-      "epoch": 2.4018196762046755,
-      "grad_norm": 0.002677842741832137,
-      "learning_rate": 0.00019999715506328655,
-      "loss": 46.0,
-      "step": 31414
-    },
-    {
-      "epoch": 2.401896133188065,
-      "grad_norm": 0.004689258988946676,
-      "learning_rate": 0.00019999715488210144,
-      "loss": 46.0,
-      "step": 31415
-    },
-    {
-      "epoch": 2.4019725901714546,
-      "grad_norm": 0.000655218493193388,
-      "learning_rate": 0.00019999715470091057,
-      "loss": 46.0,
-      "step": 31416
-    },
-    {
-      "epoch": 2.4020490471548444,
-      "grad_norm": 0.0021223174408078194,
-      "learning_rate": 0.00019999715451971394,
-      "loss": 46.0,
-      "step": 31417
-    },
-    {
-      "epoch": 2.402125504138234,
-      "grad_norm": 0.020028213039040565,
-      "learning_rate": 0.00019999715433851152,
-      "loss": 46.0,
-      "step": 31418
-    },
-    {
-      "epoch": 2.402201961121624,
-      "grad_norm": 0.0004493211454246193,
-      "learning_rate": 0.00019999715415730334,
-      "loss": 46.0,
-      "step": 31419
-    },
-    {
-      "epoch": 2.4022784181050136,
-      "grad_norm": 0.005673614796251059,
-      "learning_rate": 0.00019999715397608938,
-      "loss": 46.0,
-      "step": 31420
-    },
-    {
-      "epoch": 2.4023548750884034,
-      "grad_norm": 0.0017824191600084305,
-      "learning_rate": 0.00019999715379486968,
-      "loss": 46.0,
-      "step": 31421
-    },
-    {
-      "epoch": 2.402431332071793,
-      "grad_norm": 0.000902598025277257,
-      "learning_rate": 0.00019999715361364417,
-      "loss": 46.0,
-      "step": 31422
-    },
-    {
-      "epoch": 2.402507789055183,
-      "grad_norm": 0.0007216076483018696,
-      "learning_rate": 0.0001999971534324129,
-      "loss": 46.0,
-      "step": 31423
-    },
-    {
-      "epoch": 2.4025842460385727,
-      "grad_norm": 0.004235380329191685,
-      "learning_rate": 0.00019999715325117587,
-      "loss": 46.0,
-      "step": 31424
-    },
-    {
-      "epoch": 2.4026607030219624,
-      "grad_norm": 0.003838630858808756,
-      "learning_rate": 0.00019999715306993307,
-      "loss": 46.0,
-      "step": 31425
-    },
-    {
-      "epoch": 2.402737160005352,
-      "grad_norm": 0.0007197781815193594,
-      "learning_rate": 0.00019999715288868453,
-      "loss": 46.0,
-      "step": 31426
-    },
-    {
-      "epoch": 2.4028136169887415,
-      "grad_norm": 0.0013190425233915448,
-      "learning_rate": 0.0001999971527074302,
-      "loss": 46.0,
-      "step": 31427
-    },
-    {
-      "epoch": 2.4028900739721313,
-      "grad_norm": 0.00269205286167562,
-      "learning_rate": 0.0001999971525261701,
-      "loss": 46.0,
-      "step": 31428
-    },
-    {
-      "epoch": 2.402966530955521,
-      "grad_norm": 0.0007927425322122872,
-      "learning_rate": 0.0001999971523449042,
-      "loss": 46.0,
-      "step": 31429
-    },
-    {
-      "epoch": 2.403042987938911,
-      "grad_norm": 0.0002986110339406878,
-      "learning_rate": 0.00019999715216363257,
-      "loss": 46.0,
-      "step": 31430
-    },
-    {
-      "epoch": 2.4031194449223006,
-      "grad_norm": 0.0008843218674883246,
-      "learning_rate": 0.00019999715198235513,
-      "loss": 46.0,
-      "step": 31431
-    },
-    {
-      "epoch": 2.4031959019056903,
-      "grad_norm": 0.0009437150438316166,
-      "learning_rate": 0.00019999715180107197,
-      "loss": 46.0,
-      "step": 31432
-    },
-    {
-      "epoch": 2.40327235888908,
-      "grad_norm": 0.00034062284976243973,
-      "learning_rate": 0.00019999715161978302,
-      "loss": 46.0,
-      "step": 31433
-    },
-    {
-      "epoch": 2.40334881587247,
-      "grad_norm": 0.0017825865652412176,
-      "learning_rate": 0.00019999715143848829,
-      "loss": 46.0,
-      "step": 31434
-    },
-    {
-      "epoch": 2.4034252728558596,
-      "grad_norm": 0.002644026419147849,
-      "learning_rate": 0.00019999715125718778,
-      "loss": 46.0,
-      "step": 31435
-    },
-    {
-      "epoch": 2.4035017298392494,
-      "grad_norm": 0.0011010251473635435,
-      "learning_rate": 0.00019999715107588153,
-      "loss": 46.0,
-      "step": 31436
-    },
-    {
-      "epoch": 2.4035781868226387,
-      "grad_norm": 0.0040678055956959724,
-      "learning_rate": 0.0001999971508945695,
-      "loss": 46.0,
-      "step": 31437
-    },
-    {
-      "epoch": 2.4036546438060284,
-      "grad_norm": 0.0007749042706564069,
-      "learning_rate": 0.0001999971507132517,
-      "loss": 46.0,
-      "step": 31438
-    },
-    {
-      "epoch": 2.403731100789418,
-      "grad_norm": 0.005407374817878008,
-      "learning_rate": 0.00019999715053192817,
-      "loss": 46.0,
-      "step": 31439
-    },
-    {
-      "epoch": 2.403807557772808,
-      "grad_norm": 0.0012303779367357492,
-      "learning_rate": 0.0001999971503505988,
-      "loss": 46.0,
-      "step": 31440
-    },
-    {
-      "epoch": 2.4038840147561977,
-      "grad_norm": 0.0011702467454597354,
-      "learning_rate": 0.0001999971501692637,
-      "loss": 46.0,
-      "step": 31441
-    },
-    {
-      "epoch": 2.4039604717395875,
-      "grad_norm": 0.0011009284062311053,
-      "learning_rate": 0.00019999714998792281,
-      "loss": 46.0,
-      "step": 31442
-    },
-    {
-      "epoch": 2.4040369287229773,
-      "grad_norm": 0.0022076440509408712,
-      "learning_rate": 0.00019999714980657618,
-      "loss": 46.0,
-      "step": 31443
-    },
-    {
-      "epoch": 2.404113385706367,
-      "grad_norm": 0.0006162030040286481,
-      "learning_rate": 0.00019999714962522377,
-      "loss": 46.0,
-      "step": 31444
-    },
-    {
-      "epoch": 2.4041898426897568,
-      "grad_norm": 0.0013319840654730797,
-      "learning_rate": 0.00019999714944386558,
-      "loss": 46.0,
-      "step": 31445
-    },
-    {
-      "epoch": 2.4042662996731465,
-      "grad_norm": 0.0016527213156223297,
-      "learning_rate": 0.00019999714926250163,
-      "loss": 46.0,
-      "step": 31446
-    },
-    {
-      "epoch": 2.4043427566565363,
-      "grad_norm": 0.00266616465523839,
-      "learning_rate": 0.00019999714908113192,
-      "loss": 46.0,
-      "step": 31447
-    },
-    {
-      "epoch": 2.404419213639926,
-      "grad_norm": 0.000726548838429153,
-      "learning_rate": 0.00019999714889975642,
-      "loss": 46.0,
-      "step": 31448
-    },
-    {
-      "epoch": 2.4044956706233154,
-      "grad_norm": 0.0028805367182940245,
-      "learning_rate": 0.00019999714871837517,
-      "loss": 46.0,
-      "step": 31449
-    },
-    {
-      "epoch": 2.404572127606705,
-      "grad_norm": 0.0006409891066141427,
-      "learning_rate": 0.00019999714853698812,
-      "loss": 46.0,
-      "step": 31450
-    },
-    {
-      "epoch": 2.404648584590095,
-      "grad_norm": 0.0016447784146293998,
-      "learning_rate": 0.00019999714835559532,
-      "loss": 46.0,
-      "step": 31451
-    },
-    {
-      "epoch": 2.4047250415734847,
-      "grad_norm": 0.0009670992731116712,
-      "learning_rate": 0.00019999714817419678,
-      "loss": 46.0,
-      "step": 31452
-    },
-    {
-      "epoch": 2.4048014985568744,
-      "grad_norm": 0.0010341479210183024,
-      "learning_rate": 0.0001999971479927924,
-      "loss": 46.0,
-      "step": 31453
-    },
-    {
-      "epoch": 2.404877955540264,
-      "grad_norm": 0.0007617168012075126,
-      "learning_rate": 0.00019999714781138232,
-      "loss": 46.0,
-      "step": 31454
-    },
-    {
-      "epoch": 2.404954412523654,
-      "grad_norm": 0.0035740381572395563,
-      "learning_rate": 0.00019999714762996643,
-      "loss": 46.0,
-      "step": 31455
-    },
-    {
-      "epoch": 2.4050308695070437,
-      "grad_norm": 0.0007605867576785386,
-      "learning_rate": 0.0001999971474485448,
-      "loss": 46.0,
-      "step": 31456
-    },
-    {
-      "epoch": 2.4051073264904335,
-      "grad_norm": 0.00036915618693456054,
-      "learning_rate": 0.00019999714726711738,
-      "loss": 46.0,
-      "step": 31457
-    },
-    {
-      "epoch": 2.4051837834738232,
-      "grad_norm": 0.0008888214360922575,
-      "learning_rate": 0.0001999971470856842,
-      "loss": 46.0,
-      "step": 31458
-    },
-    {
-      "epoch": 2.4052602404572125,
-      "grad_norm": 0.0006965924403630197,
-      "learning_rate": 0.00019999714690424524,
-      "loss": 46.0,
-      "step": 31459
-    },
-    {
-      "epoch": 2.4053366974406023,
-      "grad_norm": 0.0020298114977777004,
-      "learning_rate": 0.0001999971467228005,
-      "loss": 46.0,
-      "step": 31460
-    },
-    {
-      "epoch": 2.405413154423992,
-      "grad_norm": 0.0011782522778958082,
-      "learning_rate": 0.00019999714654135003,
-      "loss": 46.0,
-      "step": 31461
-    },
-    {
-      "epoch": 2.405489611407382,
-      "grad_norm": 0.0005383405950851738,
-      "learning_rate": 0.00019999714635989376,
-      "loss": 46.0,
-      "step": 31462
-    },
-    {
-      "epoch": 2.4055660683907716,
-      "grad_norm": 0.0018920161528512836,
-      "learning_rate": 0.00019999714617843173,
-      "loss": 46.0,
-      "step": 31463
-    },
-    {
-      "epoch": 2.4056425253741613,
-      "grad_norm": 0.0005845411214977503,
-      "learning_rate": 0.00019999714599696394,
-      "loss": 46.0,
-      "step": 31464
-    },
-    {
-      "epoch": 2.405718982357551,
-      "grad_norm": 0.0007085256511345506,
-      "learning_rate": 0.00019999714581549034,
-      "loss": 46.0,
-      "step": 31465
-    },
-    {
-      "epoch": 2.405795439340941,
-      "grad_norm": 0.0014195661060512066,
-      "learning_rate": 0.00019999714563401102,
-      "loss": 46.0,
-      "step": 31466
-    },
-    {
-      "epoch": 2.4058718963243306,
-      "grad_norm": 0.004898398648947477,
-      "learning_rate": 0.0001999971454525259,
-      "loss": 46.0,
-      "step": 31467
-    },
-    {
-      "epoch": 2.4059483533077204,
-      "grad_norm": 0.0015270456206053495,
-      "learning_rate": 0.00019999714527103502,
-      "loss": 46.0,
-      "step": 31468
-    },
-    {
-      "epoch": 2.40602481029111,
-      "grad_norm": 0.0007033307920210063,
-      "learning_rate": 0.00019999714508953838,
-      "loss": 46.0,
-      "step": 31469
-    },
-    {
-      "epoch": 2.4061012672745,
-      "grad_norm": 0.00536193186417222,
-      "learning_rate": 0.00019999714490803597,
-      "loss": 46.0,
-      "step": 31470
-    },
-    {
-      "epoch": 2.4061777242578892,
-      "grad_norm": 0.0017196249682456255,
-      "learning_rate": 0.00019999714472652776,
-      "loss": 46.0,
-      "step": 31471
-    },
-    {
-      "epoch": 2.406254181241279,
-      "grad_norm": 0.0007414100109599531,
-      "learning_rate": 0.0001999971445450138,
-      "loss": 46.0,
-      "step": 31472
-    },
-    {
-      "epoch": 2.4063306382246687,
-      "grad_norm": 0.0031941295601427555,
-      "learning_rate": 0.0001999971443634941,
-      "loss": 46.0,
-      "step": 31473
-    },
-    {
-      "epoch": 2.4064070952080585,
-      "grad_norm": 0.0009738372173160315,
-      "learning_rate": 0.00019999714418196863,
-      "loss": 46.0,
-      "step": 31474
-    },
-    {
-      "epoch": 2.4064835521914483,
-      "grad_norm": 0.0012634678278118372,
-      "learning_rate": 0.00019999714400043735,
-      "loss": 46.0,
-      "step": 31475
-    },
-    {
-      "epoch": 2.406560009174838,
-      "grad_norm": 0.0009300952078774571,
-      "learning_rate": 0.0001999971438189003,
-      "loss": 46.0,
-      "step": 31476
-    },
-    {
-      "epoch": 2.406636466158228,
-      "grad_norm": 0.001526662497781217,
-      "learning_rate": 0.0001999971436373575,
-      "loss": 46.0,
-      "step": 31477
-    },
-    {
-      "epoch": 2.4067129231416176,
-      "grad_norm": 0.0007352812099270523,
-      "learning_rate": 0.00019999714345580896,
-      "loss": 46.0,
-      "step": 31478
-    },
-    {
-      "epoch": 2.4067893801250073,
-      "grad_norm": 0.001667393953539431,
-      "learning_rate": 0.00019999714327425462,
-      "loss": 46.0,
-      "step": 31479
-    },
-    {
-      "epoch": 2.406865837108397,
-      "grad_norm": 0.0012034173123538494,
-      "learning_rate": 0.00019999714309269448,
-      "loss": 46.0,
-      "step": 31480
-    },
-    {
-      "epoch": 2.4069422940917864,
-      "grad_norm": 0.0015440764836966991,
-      "learning_rate": 0.0001999971429111286,
-      "loss": 46.0,
-      "step": 31481
-    },
-    {
-      "epoch": 2.407018751075176,
-      "grad_norm": 0.0015276235062628984,
-      "learning_rate": 0.00019999714272955698,
-      "loss": 46.0,
-      "step": 31482
-    },
-    {
-      "epoch": 2.407095208058566,
-      "grad_norm": 0.0011639675358310342,
-      "learning_rate": 0.00019999714254797954,
-      "loss": 46.0,
-      "step": 31483
-    },
-    {
-      "epoch": 2.4071716650419557,
-      "grad_norm": 0.0016474663279950619,
-      "learning_rate": 0.00019999714236639638,
-      "loss": 46.0,
-      "step": 31484
-    },
-    {
-      "epoch": 2.4072481220253454,
-      "grad_norm": 0.001146140624769032,
-      "learning_rate": 0.00019999714218480743,
-      "loss": 46.0,
-      "step": 31485
-    },
-    {
-      "epoch": 2.407324579008735,
-      "grad_norm": 0.0010244814911857247,
-      "learning_rate": 0.0001999971420032127,
-      "loss": 46.0,
-      "step": 31486
-    },
-    {
-      "epoch": 2.407401035992125,
-      "grad_norm": 0.0009179611806757748,
-      "learning_rate": 0.0001999971418216122,
-      "loss": 46.0,
-      "step": 31487
-    },
-    {
-      "epoch": 2.4074774929755147,
-      "grad_norm": 0.0006006889743730426,
-      "learning_rate": 0.00019999714164000595,
-      "loss": 46.0,
-      "step": 31488
-    },
-    {
-      "epoch": 2.4075539499589045,
-      "grad_norm": 0.0005202093161642551,
-      "learning_rate": 0.0001999971414583939,
-      "loss": 46.0,
-      "step": 31489
-    },
-    {
-      "epoch": 2.4076304069422942,
-      "grad_norm": 0.0145442895591259,
-      "learning_rate": 0.0001999971412767761,
-      "loss": 46.0,
-      "step": 31490
-    },
-    {
-      "epoch": 2.407706863925684,
-      "grad_norm": 0.0003983795759268105,
-      "learning_rate": 0.00019999714109515253,
-      "loss": 46.0,
-      "step": 31491
-    },
-    {
-      "epoch": 2.4077833209090738,
-      "grad_norm": 0.0032133013010025024,
-      "learning_rate": 0.0001999971409135232,
-      "loss": 46.0,
-      "step": 31492
-    },
-    {
-      "epoch": 2.407859777892463,
-      "grad_norm": 0.0011245921486988664,
-      "learning_rate": 0.0001999971407318881,
-      "loss": 46.0,
-      "step": 31493
-    },
-    {
-      "epoch": 2.407936234875853,
-      "grad_norm": 0.0017752742860466242,
-      "learning_rate": 0.0001999971405502472,
-      "loss": 46.0,
-      "step": 31494
-    },
-    {
-      "epoch": 2.4080126918592426,
-      "grad_norm": 0.00025600663502700627,
-      "learning_rate": 0.00019999714036860055,
-      "loss": 46.0,
-      "step": 31495
-    },
-    {
-      "epoch": 2.4080891488426324,
-      "grad_norm": 0.004592161625623703,
-      "learning_rate": 0.00019999714018694814,
-      "loss": 46.0,
-      "step": 31496
-    },
-    {
-      "epoch": 2.408165605826022,
-      "grad_norm": 0.0018526840722188354,
-      "learning_rate": 0.00019999714000528993,
-      "loss": 46.0,
-      "step": 31497
-    },
-    {
-      "epoch": 2.408242062809412,
-      "grad_norm": 0.0015600936021655798,
-      "learning_rate": 0.000199997139823626,
-      "loss": 46.0,
-      "step": 31498
-    },
-    {
-      "epoch": 2.4083185197928016,
-      "grad_norm": 0.0012808283790946007,
-      "learning_rate": 0.00019999713964195625,
-      "loss": 46.0,
-      "step": 31499
-    },
-    {
-      "epoch": 2.4083949767761914,
-      "grad_norm": 0.00048348886775784194,
-      "learning_rate": 0.00019999713946028077,
-      "loss": 46.0,
-      "step": 31500
-    },
-    {
-      "epoch": 2.408471433759581,
-      "grad_norm": 0.0006938119768165052,
-      "learning_rate": 0.00019999713927859952,
-      "loss": 46.0,
-      "step": 31501
-    },
-    {
-      "epoch": 2.4085478907429705,
-      "grad_norm": 0.0008967588655650616,
-      "learning_rate": 0.00019999713909691247,
-      "loss": 46.0,
-      "step": 31502
-    },
-    {
-      "epoch": 2.4086243477263602,
-      "grad_norm": 0.0019483311334624887,
-      "learning_rate": 0.00019999713891521968,
-      "loss": 46.0,
-      "step": 31503
-    },
-    {
-      "epoch": 2.40870080470975,
-      "grad_norm": 0.000997532275505364,
-      "learning_rate": 0.0001999971387335211,
-      "loss": 46.0,
-      "step": 31504
-    },
-    {
-      "epoch": 2.4087772616931398,
-      "grad_norm": 0.0006950461538508534,
-      "learning_rate": 0.00019999713855181676,
-      "loss": 46.0,
-      "step": 31505
-    },
-    {
-      "epoch": 2.4088537186765295,
-      "grad_norm": 0.0004208044265396893,
-      "learning_rate": 0.00019999713837010665,
-      "loss": 46.0,
-      "step": 31506
-    },
-    {
-      "epoch": 2.4089301756599193,
-      "grad_norm": 0.0014009278966113925,
-      "learning_rate": 0.00019999713818839076,
-      "loss": 46.0,
-      "step": 31507
-    },
-    {
-      "epoch": 2.409006632643309,
-      "grad_norm": 0.002767933765426278,
-      "learning_rate": 0.0001999971380066691,
-      "loss": 46.0,
-      "step": 31508
-    },
-    {
-      "epoch": 2.409083089626699,
-      "grad_norm": 0.0007349203224293888,
-      "learning_rate": 0.00019999713782494172,
-      "loss": 46.0,
-      "step": 31509
-    },
-    {
-      "epoch": 2.4091595466100886,
-      "grad_norm": 0.0011516865342855453,
-      "learning_rate": 0.0001999971376432085,
-      "loss": 46.0,
-      "step": 31510
-    },
-    {
-      "epoch": 2.4092360035934783,
-      "grad_norm": 0.002653359202668071,
-      "learning_rate": 0.00019999713746146955,
-      "loss": 46.0,
-      "step": 31511
-    },
-    {
-      "epoch": 2.409312460576868,
-      "grad_norm": 0.0011696862056851387,
-      "learning_rate": 0.00019999713727972482,
-      "loss": 46.0,
-      "step": 31512
-    },
-    {
-      "epoch": 2.409388917560258,
-      "grad_norm": 0.0016497414326295257,
-      "learning_rate": 0.00019999713709797435,
-      "loss": 46.0,
-      "step": 31513
-    },
-    {
-      "epoch": 2.4094653745436476,
-      "grad_norm": 0.0007007354288361967,
-      "learning_rate": 0.0001999971369162181,
-      "loss": 46.0,
-      "step": 31514
-    },
-    {
-      "epoch": 2.409541831527037,
-      "grad_norm": 0.0008845037664286792,
-      "learning_rate": 0.00019999713673445605,
-      "loss": 46.0,
-      "step": 31515
-    },
-    {
-      "epoch": 2.4096182885104267,
-      "grad_norm": 0.0003855802060570568,
-      "learning_rate": 0.00019999713655268823,
-      "loss": 46.0,
-      "step": 31516
-    },
-    {
-      "epoch": 2.4096947454938165,
-      "grad_norm": 0.001040634699165821,
-      "learning_rate": 0.00019999713637091466,
-      "loss": 46.0,
-      "step": 31517
-    },
-    {
-      "epoch": 2.409771202477206,
-      "grad_norm": 0.0004562884569168091,
-      "learning_rate": 0.00019999713618913532,
-      "loss": 46.0,
-      "step": 31518
-    },
-    {
-      "epoch": 2.409847659460596,
-      "grad_norm": 0.00244263862259686,
-      "learning_rate": 0.00019999713600735023,
-      "loss": 46.0,
-      "step": 31519
-    },
-    {
-      "epoch": 2.4099241164439857,
-      "grad_norm": 0.000925388652831316,
-      "learning_rate": 0.0001999971358255593,
-      "loss": 46.0,
-      "step": 31520
-    },
-    {
-      "epoch": 2.4100005734273755,
-      "grad_norm": 0.0005213592085056007,
-      "learning_rate": 0.00019999713564376268,
-      "loss": 46.0,
-      "step": 31521
-    },
-    {
-      "epoch": 2.4100770304107653,
-      "grad_norm": 0.0021028933115303516,
-      "learning_rate": 0.00019999713546196027,
-      "loss": 46.0,
-      "step": 31522
-    },
-    {
-      "epoch": 2.410153487394155,
-      "grad_norm": 0.0018643634393811226,
-      "learning_rate": 0.00019999713528015206,
-      "loss": 46.0,
-      "step": 31523
-    },
-    {
-      "epoch": 2.4102299443775443,
-      "grad_norm": 0.00029715648270212114,
-      "learning_rate": 0.0001999971350983381,
-      "loss": 46.0,
-      "step": 31524
-    },
-    {
-      "epoch": 2.410306401360934,
-      "grad_norm": 0.0009367327438667417,
-      "learning_rate": 0.0001999971349165184,
-      "loss": 46.0,
-      "step": 31525
-    },
-    {
-      "epoch": 2.410382858344324,
-      "grad_norm": 0.0008120790007524192,
-      "learning_rate": 0.0001999971347346929,
-      "loss": 46.0,
-      "step": 31526
-    },
-    {
-      "epoch": 2.4104593153277136,
-      "grad_norm": 0.004285277798771858,
-      "learning_rate": 0.00019999713455286163,
-      "loss": 46.0,
-      "step": 31527
-    },
-    {
-      "epoch": 2.4105357723111034,
-      "grad_norm": 0.004256781190633774,
-      "learning_rate": 0.00019999713437102458,
-      "loss": 46.0,
-      "step": 31528
-    },
-    {
-      "epoch": 2.410612229294493,
-      "grad_norm": 0.0008301177294924855,
-      "learning_rate": 0.0001999971341891818,
-      "loss": 46.0,
-      "step": 31529
-    },
-    {
-      "epoch": 2.410688686277883,
-      "grad_norm": 0.0032910851296037436,
-      "learning_rate": 0.00019999713400733321,
-      "loss": 46.0,
-      "step": 31530
-    },
-    {
-      "epoch": 2.4107651432612727,
-      "grad_norm": 0.0005402807146310806,
-      "learning_rate": 0.00019999713382547887,
-      "loss": 46.0,
-      "step": 31531
-    },
-    {
-      "epoch": 2.4108416002446624,
-      "grad_norm": 0.004251751117408276,
-      "learning_rate": 0.00019999713364361876,
-      "loss": 46.0,
-      "step": 31532
-    },
-    {
-      "epoch": 2.410918057228052,
-      "grad_norm": 0.0009280891972593963,
-      "learning_rate": 0.0001999971334617529,
-      "loss": 46.0,
-      "step": 31533
-    },
-    {
-      "epoch": 2.410994514211442,
-      "grad_norm": 0.000361142068868503,
-      "learning_rate": 0.0001999971332798812,
-      "loss": 46.0,
-      "step": 31534
-    },
-    {
-      "epoch": 2.4110709711948317,
-      "grad_norm": 0.0009967322694137692,
-      "learning_rate": 0.0001999971330980038,
-      "loss": 46.0,
-      "step": 31535
-    },
-    {
-      "epoch": 2.4111474281782215,
-      "grad_norm": 0.0011202595196664333,
-      "learning_rate": 0.00019999713291612064,
-      "loss": 46.0,
-      "step": 31536
-    },
-    {
-      "epoch": 2.411223885161611,
-      "grad_norm": 0.0006760488613508642,
-      "learning_rate": 0.00019999713273423166,
-      "loss": 46.0,
-      "step": 31537
-    },
-    {
-      "epoch": 2.4113003421450006,
-      "grad_norm": 0.0012034527026116848,
-      "learning_rate": 0.0001999971325523369,
-      "loss": 46.0,
-      "step": 31538
-    },
-    {
-      "epoch": 2.4113767991283903,
-      "grad_norm": 0.0009123107302002609,
-      "learning_rate": 0.00019999713237043643,
-      "loss": 46.0,
-      "step": 31539
-    },
-    {
-      "epoch": 2.41145325611178,
-      "grad_norm": 0.0012663532979786396,
-      "learning_rate": 0.00019999713218853016,
-      "loss": 46.0,
-      "step": 31540
-    },
-    {
-      "epoch": 2.41152971309517,
-      "grad_norm": 0.002751997672021389,
-      "learning_rate": 0.00019999713200661814,
-      "loss": 46.0,
-      "step": 31541
-    },
-    {
-      "epoch": 2.4116061700785596,
-      "grad_norm": 0.0032665308099240065,
-      "learning_rate": 0.00019999713182470032,
-      "loss": 46.0,
-      "step": 31542
-    },
-    {
-      "epoch": 2.4116826270619494,
-      "grad_norm": 0.0006853691302239895,
-      "learning_rate": 0.00019999713164277675,
-      "loss": 46.0,
-      "step": 31543
-    },
-    {
-      "epoch": 2.411759084045339,
-      "grad_norm": 0.001948764780536294,
-      "learning_rate": 0.0001999971314608474,
-      "loss": 46.0,
-      "step": 31544
-    },
-    {
-      "epoch": 2.411835541028729,
-      "grad_norm": 0.0012749229790642858,
-      "learning_rate": 0.00019999713127891232,
-      "loss": 46.0,
-      "step": 31545
-    },
-    {
-      "epoch": 2.411911998012118,
-      "grad_norm": 0.0024692656006664038,
-      "learning_rate": 0.0001999971310969714,
-      "loss": 46.0,
-      "step": 31546
-    },
-    {
-      "epoch": 2.411988454995508,
-      "grad_norm": 0.0035269109066575766,
-      "learning_rate": 0.00019999713091502477,
-      "loss": 46.0,
-      "step": 31547
-    },
-    {
-      "epoch": 2.4120649119788977,
-      "grad_norm": 0.0007178937084972858,
-      "learning_rate": 0.00019999713073307236,
-      "loss": 46.0,
-      "step": 31548
-    },
-    {
-      "epoch": 2.4121413689622875,
-      "grad_norm": 0.0016796267591416836,
-      "learning_rate": 0.00019999713055111415,
-      "loss": 46.0,
-      "step": 31549
-    },
-    {
-      "epoch": 2.4122178259456772,
-      "grad_norm": 0.00041162510751746595,
-      "learning_rate": 0.0001999971303691502,
-      "loss": 46.0,
-      "step": 31550
-    },
-    {
-      "epoch": 2.412294282929067,
-      "grad_norm": 0.001185059198178351,
-      "learning_rate": 0.00019999713018718047,
-      "loss": 46.0,
-      "step": 31551
-    },
-    {
-      "epoch": 2.4123707399124568,
-      "grad_norm": 0.0041820150800049305,
-      "learning_rate": 0.00019999713000520497,
-      "loss": 46.0,
-      "step": 31552
-    },
-    {
-      "epoch": 2.4124471968958465,
-      "grad_norm": 0.0005849146982654929,
-      "learning_rate": 0.00019999712982322372,
-      "loss": 46.0,
-      "step": 31553
-    },
-    {
-      "epoch": 2.4125236538792363,
-      "grad_norm": 0.0012389356270432472,
-      "learning_rate": 0.00019999712964123667,
-      "loss": 46.0,
-      "step": 31554
-    },
-    {
-      "epoch": 2.412600110862626,
-      "grad_norm": 0.0004523008537944406,
-      "learning_rate": 0.00019999712945924385,
-      "loss": 46.0,
-      "step": 31555
-    },
-    {
-      "epoch": 2.412676567846016,
-      "grad_norm": 0.0015283380635082722,
-      "learning_rate": 0.0001999971292772453,
-      "loss": 46.0,
-      "step": 31556
-    },
-    {
-      "epoch": 2.4127530248294056,
-      "grad_norm": 0.0006729109445586801,
-      "learning_rate": 0.00019999712909524094,
-      "loss": 46.0,
-      "step": 31557
-    },
-    {
-      "epoch": 2.412829481812795,
-      "grad_norm": 0.004310677759349346,
-      "learning_rate": 0.00019999712891323083,
-      "loss": 46.0,
-      "step": 31558
-    },
-    {
-      "epoch": 2.4129059387961846,
-      "grad_norm": 0.003804714884608984,
-      "learning_rate": 0.00019999712873121494,
-      "loss": 46.0,
-      "step": 31559
-    },
-    {
-      "epoch": 2.4129823957795744,
-      "grad_norm": 0.002762932563200593,
-      "learning_rate": 0.00019999712854919328,
-      "loss": 46.0,
-      "step": 31560
-    },
-    {
-      "epoch": 2.413058852762964,
-      "grad_norm": 0.004486938007175922,
-      "learning_rate": 0.00019999712836716587,
-      "loss": 46.0,
-      "step": 31561
-    },
-    {
-      "epoch": 2.413135309746354,
-      "grad_norm": 0.0009690411970950663,
-      "learning_rate": 0.00019999712818513266,
-      "loss": 46.0,
-      "step": 31562
-    },
-    {
-      "epoch": 2.4132117667297437,
-      "grad_norm": 0.002592853968963027,
-      "learning_rate": 0.0001999971280030937,
-      "loss": 46.0,
-      "step": 31563
-    },
-    {
-      "epoch": 2.4132882237131335,
-      "grad_norm": 0.0028453718405216932,
-      "learning_rate": 0.00019999712782104898,
-      "loss": 46.0,
-      "step": 31564
-    },
-    {
-      "epoch": 2.413364680696523,
-      "grad_norm": 0.002826140960678458,
-      "learning_rate": 0.00019999712763899848,
-      "loss": 46.0,
-      "step": 31565
-    },
-    {
-      "epoch": 2.413441137679913,
-      "grad_norm": 0.0014487982261925936,
-      "learning_rate": 0.00019999712745694223,
-      "loss": 46.0,
-      "step": 31566
-    },
-    {
-      "epoch": 2.4135175946633027,
-      "grad_norm": 0.0006863312446512282,
-      "learning_rate": 0.00019999712727488019,
-      "loss": 46.0,
-      "step": 31567
-    },
-    {
-      "epoch": 2.413594051646692,
-      "grad_norm": 0.0010164452251046896,
-      "learning_rate": 0.0001999971270928124,
-      "loss": 46.0,
-      "step": 31568
-    },
-    {
-      "epoch": 2.413670508630082,
-      "grad_norm": 0.0006221099756658077,
-      "learning_rate": 0.0001999971269107388,
-      "loss": 46.0,
-      "step": 31569
-    },
-    {
-      "epoch": 2.4137469656134716,
-      "grad_norm": 0.0015117890434339643,
-      "learning_rate": 0.00019999712672865946,
-      "loss": 46.0,
-      "step": 31570
-    },
-    {
-      "epoch": 2.4138234225968613,
-      "grad_norm": 0.0005253482377156615,
-      "learning_rate": 0.00019999712654657437,
-      "loss": 46.0,
-      "step": 31571
-    },
-    {
-      "epoch": 2.413899879580251,
-      "grad_norm": 0.001459247781895101,
-      "learning_rate": 0.00019999712636448346,
-      "loss": 46.0,
-      "step": 31572
-    },
-    {
-      "epoch": 2.413976336563641,
-      "grad_norm": 0.0011814262252300978,
-      "learning_rate": 0.00019999712618238682,
-      "loss": 46.0,
-      "step": 31573
-    },
-    {
-      "epoch": 2.4140527935470306,
-      "grad_norm": 0.0007734876126050949,
-      "learning_rate": 0.0001999971260002844,
-      "loss": 46.0,
-      "step": 31574
-    },
-    {
-      "epoch": 2.4141292505304204,
-      "grad_norm": 0.0020652583334594965,
-      "learning_rate": 0.0001999971258181762,
-      "loss": 46.0,
-      "step": 31575
-    },
-    {
-      "epoch": 2.41420570751381,
-      "grad_norm": 0.0021665054373443127,
-      "learning_rate": 0.00019999712563606225,
-      "loss": 46.0,
-      "step": 31576
-    },
-    {
-      "epoch": 2.4142821644972,
-      "grad_norm": 0.0009689588914625347,
-      "learning_rate": 0.00019999712545394253,
-      "loss": 46.0,
-      "step": 31577
-    },
-    {
-      "epoch": 2.4143586214805897,
-      "grad_norm": 0.0005599790019914508,
-      "learning_rate": 0.000199997125271817,
-      "loss": 46.0,
-      "step": 31578
-    },
-    {
-      "epoch": 2.4144350784639794,
-      "grad_norm": 0.0007454440928995609,
-      "learning_rate": 0.00019999712508968575,
-      "loss": 46.0,
-      "step": 31579
-    },
-    {
-      "epoch": 2.4145115354473687,
-      "grad_norm": 0.0014415301848202944,
-      "learning_rate": 0.0001999971249075487,
-      "loss": 46.0,
-      "step": 31580
-    },
-    {
-      "epoch": 2.4145879924307585,
-      "grad_norm": 0.0008488707826472819,
-      "learning_rate": 0.0001999971247254059,
-      "loss": 46.0,
-      "step": 31581
-    },
-    {
-      "epoch": 2.4146644494141483,
-      "grad_norm": 0.0015603253850713372,
-      "learning_rate": 0.00019999712454325732,
-      "loss": 46.0,
-      "step": 31582
-    },
-    {
-      "epoch": 2.414740906397538,
-      "grad_norm": 0.0016115342732518911,
-      "learning_rate": 0.00019999712436110295,
-      "loss": 46.0,
-      "step": 31583
-    },
-    {
-      "epoch": 2.414817363380928,
-      "grad_norm": 0.0008736657910048962,
-      "learning_rate": 0.00019999712417894286,
-      "loss": 46.0,
-      "step": 31584
-    },
-    {
-      "epoch": 2.4148938203643175,
-      "grad_norm": 0.0008387601701542735,
-      "learning_rate": 0.00019999712399677697,
-      "loss": 46.0,
-      "step": 31585
-    },
-    {
-      "epoch": 2.4149702773477073,
-      "grad_norm": 0.0005933020147494972,
-      "learning_rate": 0.0001999971238146053,
-      "loss": 46.0,
-      "step": 31586
-    },
-    {
-      "epoch": 2.415046734331097,
-      "grad_norm": 0.0011197460116818547,
-      "learning_rate": 0.00019999712363242788,
-      "loss": 46.0,
-      "step": 31587
-    },
-    {
-      "epoch": 2.415123191314487,
-      "grad_norm": 0.0008185844053514302,
-      "learning_rate": 0.0001999971234502447,
-      "loss": 46.0,
-      "step": 31588
-    },
-    {
-      "epoch": 2.4151996482978766,
-      "grad_norm": 0.0007765372283756733,
-      "learning_rate": 0.00019999712326805575,
-      "loss": 46.0,
-      "step": 31589
-    },
-    {
-      "epoch": 2.415276105281266,
-      "grad_norm": 0.003065679920837283,
-      "learning_rate": 0.000199997123085861,
-      "loss": 46.0,
-      "step": 31590
-    },
-    {
-      "epoch": 2.4153525622646557,
-      "grad_norm": 0.0013270934578031301,
-      "learning_rate": 0.00019999712290366052,
-      "loss": 46.0,
-      "step": 31591
-    },
-    {
-      "epoch": 2.4154290192480454,
-      "grad_norm": 0.0016827181680127978,
-      "learning_rate": 0.00019999712272145424,
-      "loss": 46.0,
-      "step": 31592
-    },
-    {
-      "epoch": 2.415505476231435,
-      "grad_norm": 0.0006743364501744509,
-      "learning_rate": 0.0001999971225392422,
-      "loss": 46.0,
-      "step": 31593
-    },
-    {
-      "epoch": 2.415581933214825,
-      "grad_norm": 0.0009680672665126622,
-      "learning_rate": 0.0001999971223570244,
-      "loss": 46.0,
-      "step": 31594
-    },
-    {
-      "epoch": 2.4156583901982147,
-      "grad_norm": 0.0009353538043797016,
-      "learning_rate": 0.0001999971221748008,
-      "loss": 46.0,
-      "step": 31595
-    },
-    {
-      "epoch": 2.4157348471816045,
-      "grad_norm": 0.001025684643536806,
-      "learning_rate": 0.00019999712199257147,
-      "loss": 46.0,
-      "step": 31596
-    },
-    {
-      "epoch": 2.4158113041649942,
-      "grad_norm": 0.003344607073813677,
-      "learning_rate": 0.00019999712181033636,
-      "loss": 46.0,
-      "step": 31597
-    },
-    {
-      "epoch": 2.415887761148384,
-      "grad_norm": 0.0011846135603263974,
-      "learning_rate": 0.00019999712162809547,
-      "loss": 46.0,
-      "step": 31598
-    },
-    {
-      "epoch": 2.4159642181317738,
-      "grad_norm": 0.00117380334995687,
-      "learning_rate": 0.00019999712144584884,
-      "loss": 46.0,
-      "step": 31599
-    },
-    {
-      "epoch": 2.4160406751151635,
-      "grad_norm": 0.0008556417305953801,
-      "learning_rate": 0.0001999971212635964,
-      "loss": 46.0,
-      "step": 31600
-    },
-    {
-      "epoch": 2.4161171320985533,
-      "grad_norm": 0.0029806990642100573,
-      "learning_rate": 0.00019999712108133822,
-      "loss": 46.0,
-      "step": 31601
-    },
-    {
-      "epoch": 2.4161935890819426,
-      "grad_norm": 0.0006079701706767082,
-      "learning_rate": 0.00019999712089907424,
-      "loss": 46.0,
-      "step": 31602
-    },
-    {
-      "epoch": 2.4162700460653324,
-      "grad_norm": 0.0009088863735087216,
-      "learning_rate": 0.0001999971207168045,
-      "loss": 46.0,
-      "step": 31603
-    },
-    {
-      "epoch": 2.416346503048722,
-      "grad_norm": 0.001433251309208572,
-      "learning_rate": 0.00019999712053452902,
-      "loss": 46.0,
-      "step": 31604
-    },
-    {
-      "epoch": 2.416422960032112,
-      "grad_norm": 0.0022963511291891336,
-      "learning_rate": 0.00019999712035224777,
-      "loss": 46.0,
-      "step": 31605
-    },
-    {
-      "epoch": 2.4164994170155016,
-      "grad_norm": 0.0035245544277131557,
-      "learning_rate": 0.0001999971201699607,
-      "loss": 46.0,
-      "step": 31606
-    },
-    {
-      "epoch": 2.4165758739988914,
-      "grad_norm": 0.0010022609494626522,
-      "learning_rate": 0.00019999711998766788,
-      "loss": 46.0,
-      "step": 31607
-    },
-    {
-      "epoch": 2.416652330982281,
-      "grad_norm": 0.0012003815500065684,
-      "learning_rate": 0.0001999971198053693,
-      "loss": 46.0,
-      "step": 31608
-    },
-    {
-      "epoch": 2.416728787965671,
-      "grad_norm": 0.0019965777173638344,
-      "learning_rate": 0.00019999711962306494,
-      "loss": 46.0,
-      "step": 31609
-    },
-    {
-      "epoch": 2.4168052449490607,
-      "grad_norm": 0.00037816326948814094,
-      "learning_rate": 0.00019999711944075486,
-      "loss": 46.0,
-      "step": 31610
-    },
-    {
-      "epoch": 2.4168817019324504,
-      "grad_norm": 0.0005169878713786602,
-      "learning_rate": 0.00019999711925843897,
-      "loss": 46.0,
-      "step": 31611
-    },
-    {
-      "epoch": 2.4169581589158398,
-      "grad_norm": 0.0006264549447223544,
-      "learning_rate": 0.0001999971190761173,
-      "loss": 46.0,
-      "step": 31612
-    },
-    {
-      "epoch": 2.4170346158992295,
-      "grad_norm": 0.001726500573568046,
-      "learning_rate": 0.0001999971188937899,
-      "loss": 46.0,
-      "step": 31613
-    },
-    {
-      "epoch": 2.4171110728826193,
-      "grad_norm": 0.0012484944891184568,
-      "learning_rate": 0.0001999971187114567,
-      "loss": 46.0,
-      "step": 31614
-    },
-    {
-      "epoch": 2.417187529866009,
-      "grad_norm": 0.00048276971210725605,
-      "learning_rate": 0.00019999711852911774,
-      "loss": 46.0,
-      "step": 31615
-    },
-    {
-      "epoch": 2.417263986849399,
-      "grad_norm": 0.00045784362009726465,
-      "learning_rate": 0.000199997118346773,
-      "loss": 46.0,
-      "step": 31616
-    },
-    {
-      "epoch": 2.4173404438327886,
-      "grad_norm": 0.001244532992132008,
-      "learning_rate": 0.0001999971181644225,
-      "loss": 46.0,
-      "step": 31617
-    },
-    {
-      "epoch": 2.4174169008161783,
-      "grad_norm": 0.0007591007743030787,
-      "learning_rate": 0.00019999711798206624,
-      "loss": 46.0,
-      "step": 31618
-    },
-    {
-      "epoch": 2.417493357799568,
-      "grad_norm": 0.000927189365029335,
-      "learning_rate": 0.0001999971177997042,
-      "loss": 46.0,
-      "step": 31619
-    },
-    {
-      "epoch": 2.417569814782958,
-      "grad_norm": 0.0026697805151343346,
-      "learning_rate": 0.0001999971176173364,
-      "loss": 46.0,
-      "step": 31620
-    },
-    {
-      "epoch": 2.4176462717663476,
-      "grad_norm": 0.0009276054333895445,
-      "learning_rate": 0.0001999971174349628,
-      "loss": 46.0,
-      "step": 31621
-    },
-    {
-      "epoch": 2.4177227287497374,
-      "grad_norm": 0.0027608326636254787,
-      "learning_rate": 0.00019999711725258345,
-      "loss": 46.0,
-      "step": 31622
-    },
-    {
-      "epoch": 2.417799185733127,
-      "grad_norm": 0.002007543807849288,
-      "learning_rate": 0.00019999711707019833,
-      "loss": 46.0,
-      "step": 31623
-    },
-    {
-      "epoch": 2.4178756427165164,
-      "grad_norm": 0.0006276968051679432,
-      "learning_rate": 0.00019999711688780745,
-      "loss": 46.0,
-      "step": 31624
-    },
-    {
-      "epoch": 2.417952099699906,
-      "grad_norm": 0.0011470308527350426,
-      "learning_rate": 0.0001999971167054108,
-      "loss": 46.0,
-      "step": 31625
-    },
-    {
-      "epoch": 2.418028556683296,
-      "grad_norm": 0.00213835877366364,
-      "learning_rate": 0.00019999711652300835,
-      "loss": 46.0,
-      "step": 31626
-    },
-    {
-      "epoch": 2.4181050136666857,
-      "grad_norm": 0.0007017204770818353,
-      "learning_rate": 0.00019999711634060017,
-      "loss": 46.0,
-      "step": 31627
-    },
-    {
-      "epoch": 2.4181814706500755,
-      "grad_norm": 0.0052144769579172134,
-      "learning_rate": 0.00019999711615818622,
-      "loss": 46.0,
-      "step": 31628
-    },
-    {
-      "epoch": 2.4182579276334653,
-      "grad_norm": 0.0019888682290911674,
-      "learning_rate": 0.0001999971159757665,
-      "loss": 46.0,
-      "step": 31629
-    },
-    {
-      "epoch": 2.418334384616855,
-      "grad_norm": 0.00046531064435839653,
-      "learning_rate": 0.00019999711579334097,
-      "loss": 46.0,
-      "step": 31630
-    },
-    {
-      "epoch": 2.4184108416002448,
-      "grad_norm": 0.0008483295678161085,
-      "learning_rate": 0.0001999971156109097,
-      "loss": 46.0,
-      "step": 31631
-    },
-    {
-      "epoch": 2.4184872985836345,
-      "grad_norm": 0.004493400454521179,
-      "learning_rate": 0.00019999711542847268,
-      "loss": 46.0,
-      "step": 31632
-    },
-    {
-      "epoch": 2.418563755567024,
-      "grad_norm": 0.0008407704881392419,
-      "learning_rate": 0.00019999711524602986,
-      "loss": 46.0,
-      "step": 31633
-    },
-    {
-      "epoch": 2.4186402125504136,
-      "grad_norm": 0.0018366235308349133,
-      "learning_rate": 0.00019999711506358127,
-      "loss": 46.0,
-      "step": 31634
-    },
-    {
-      "epoch": 2.4187166695338034,
-      "grad_norm": 0.0008160322322510183,
-      "learning_rate": 0.0001999971148811269,
-      "loss": 46.0,
-      "step": 31635
-    },
-    {
-      "epoch": 2.418793126517193,
-      "grad_norm": 0.0033823465928435326,
-      "learning_rate": 0.00019999711469866681,
-      "loss": 46.0,
-      "step": 31636
-    },
-    {
-      "epoch": 2.418869583500583,
-      "grad_norm": 0.0017869259463623166,
-      "learning_rate": 0.0001999971145162009,
-      "loss": 46.0,
-      "step": 31637
-    },
-    {
-      "epoch": 2.4189460404839727,
-      "grad_norm": 0.0011285413056612015,
-      "learning_rate": 0.00019999711433372927,
-      "loss": 46.0,
-      "step": 31638
-    },
-    {
-      "epoch": 2.4190224974673624,
-      "grad_norm": 0.0016802398022264242,
-      "learning_rate": 0.00019999711415125184,
-      "loss": 46.0,
-      "step": 31639
-    },
-    {
-      "epoch": 2.419098954450752,
-      "grad_norm": 0.001447243383154273,
-      "learning_rate": 0.00019999711396876863,
-      "loss": 46.0,
-      "step": 31640
-    },
-    {
-      "epoch": 2.419175411434142,
-      "grad_norm": 0.0010327561758458614,
-      "learning_rate": 0.00019999711378627968,
-      "loss": 46.0,
-      "step": 31641
-    },
-    {
-      "epoch": 2.4192518684175317,
-      "grad_norm": 0.0011783323716372252,
-      "learning_rate": 0.00019999711360378495,
-      "loss": 46.0,
-      "step": 31642
-    },
-    {
-      "epoch": 2.4193283254009215,
-      "grad_norm": 0.0013162004761397839,
-      "learning_rate": 0.00019999711342128445,
-      "loss": 46.0,
-      "step": 31643
-    },
-    {
-      "epoch": 2.4194047823843112,
-      "grad_norm": 0.0014117922401055694,
-      "learning_rate": 0.00019999711323877818,
-      "loss": 46.0,
-      "step": 31644
-    },
-    {
-      "epoch": 2.419481239367701,
-      "grad_norm": 0.0011919975513592362,
-      "learning_rate": 0.0001999971130562661,
-      "loss": 46.0,
-      "step": 31645
-    },
-    {
-      "epoch": 2.4195576963510903,
-      "grad_norm": 0.0003853417874779552,
-      "learning_rate": 0.00019999711287374831,
-      "loss": 46.0,
-      "step": 31646
-    },
-    {
-      "epoch": 2.41963415333448,
-      "grad_norm": 0.002028072252869606,
-      "learning_rate": 0.00019999711269122475,
-      "loss": 46.0,
-      "step": 31647
-    },
-    {
-      "epoch": 2.41971061031787,
-      "grad_norm": 0.0008285519434139132,
-      "learning_rate": 0.00019999711250869538,
-      "loss": 46.0,
-      "step": 31648
-    },
-    {
-      "epoch": 2.4197870673012596,
-      "grad_norm": 0.002793597988784313,
-      "learning_rate": 0.00019999711232616027,
-      "loss": 46.0,
-      "step": 31649
-    },
-    {
-      "epoch": 2.4198635242846493,
-      "grad_norm": 0.003838328877463937,
-      "learning_rate": 0.00019999711214361939,
-      "loss": 46.0,
-      "step": 31650
-    },
-    {
-      "epoch": 2.419939981268039,
-      "grad_norm": 0.0007785335765220225,
-      "learning_rate": 0.00019999711196107273,
-      "loss": 46.0,
-      "step": 31651
-    },
-    {
-      "epoch": 2.420016438251429,
-      "grad_norm": 0.0006616014870814979,
-      "learning_rate": 0.0001999971117785203,
-      "loss": 46.0,
-      "step": 31652
-    },
-    {
-      "epoch": 2.4200928952348186,
-      "grad_norm": 0.0004947613342665136,
-      "learning_rate": 0.00019999711159596212,
-      "loss": 46.0,
-      "step": 31653
-    },
-    {
-      "epoch": 2.4201693522182084,
-      "grad_norm": 0.0010030142730101943,
-      "learning_rate": 0.00019999711141339814,
-      "loss": 46.0,
-      "step": 31654
-    },
-    {
-      "epoch": 2.4202458092015977,
-      "grad_norm": 0.00097366125555709,
-      "learning_rate": 0.0001999971112308284,
-      "loss": 46.0,
-      "step": 31655
-    },
-    {
-      "epoch": 2.4203222661849875,
-      "grad_norm": 0.00045676305308006704,
-      "learning_rate": 0.00019999711104825289,
-      "loss": 46.0,
-      "step": 31656
-    },
-    {
-      "epoch": 2.4203987231683772,
-      "grad_norm": 0.001204445376060903,
-      "learning_rate": 0.00019999711086567164,
-      "loss": 46.0,
-      "step": 31657
-    },
-    {
-      "epoch": 2.420475180151767,
-      "grad_norm": 0.005395912565290928,
-      "learning_rate": 0.0001999971106830846,
-      "loss": 46.0,
-      "step": 31658
-    },
-    {
-      "epoch": 2.4205516371351568,
-      "grad_norm": 0.001471759402193129,
-      "learning_rate": 0.00019999711050049178,
-      "loss": 46.0,
-      "step": 31659
-    },
-    {
-      "epoch": 2.4206280941185465,
-      "grad_norm": 0.0007064960664138198,
-      "learning_rate": 0.0001999971103178932,
-      "loss": 46.0,
-      "step": 31660
-    },
-    {
-      "epoch": 2.4207045511019363,
-      "grad_norm": 0.0021935170516371727,
-      "learning_rate": 0.00019999711013528884,
-      "loss": 46.0,
-      "step": 31661
-    },
-    {
-      "epoch": 2.420781008085326,
-      "grad_norm": 0.003960418980568647,
-      "learning_rate": 0.00019999710995267873,
-      "loss": 46.0,
-      "step": 31662
-    },
-    {
-      "epoch": 2.420857465068716,
-      "grad_norm": 0.0005817844066768885,
-      "learning_rate": 0.00019999710977006285,
-      "loss": 46.0,
-      "step": 31663
-    },
-    {
-      "epoch": 2.4209339220521056,
-      "grad_norm": 0.0022377900313585997,
-      "learning_rate": 0.0001999971095874412,
-      "loss": 46.0,
-      "step": 31664
-    },
-    {
-      "epoch": 2.4210103790354953,
-      "grad_norm": 0.0008167008636519313,
-      "learning_rate": 0.00019999710940481376,
-      "loss": 46.0,
-      "step": 31665
-    },
-    {
-      "epoch": 2.421086836018885,
-      "grad_norm": 0.0031989584676921368,
-      "learning_rate": 0.00019999710922218055,
-      "loss": 46.0,
-      "step": 31666
-    },
-    {
-      "epoch": 2.421163293002275,
-      "grad_norm": 0.00047056740731932223,
-      "learning_rate": 0.00019999710903954157,
-      "loss": 46.0,
-      "step": 31667
-    },
-    {
-      "epoch": 2.421239749985664,
-      "grad_norm": 0.0010123159736394882,
-      "learning_rate": 0.00019999710885689687,
-      "loss": 46.0,
-      "step": 31668
-    },
-    {
-      "epoch": 2.421316206969054,
-      "grad_norm": 0.0006796225789003074,
-      "learning_rate": 0.00019999710867424635,
-      "loss": 46.0,
-      "step": 31669
-    },
-    {
-      "epoch": 2.4213926639524437,
-      "grad_norm": 0.0011805481044575572,
-      "learning_rate": 0.00019999710849159008,
-      "loss": 46.0,
-      "step": 31670
-    },
-    {
-      "epoch": 2.4214691209358334,
-      "grad_norm": 0.0012977055739611387,
-      "learning_rate": 0.00019999710830892803,
-      "loss": 46.0,
-      "step": 31671
-    },
-    {
-      "epoch": 2.421545577919223,
-      "grad_norm": 0.0007809889502823353,
-      "learning_rate": 0.0001999971081262602,
-      "loss": 46.0,
-      "step": 31672
-    },
-    {
-      "epoch": 2.421622034902613,
-      "grad_norm": 0.0019299163250252604,
-      "learning_rate": 0.00019999710794358665,
-      "loss": 46.0,
-      "step": 31673
-    },
-    {
-      "epoch": 2.4216984918860027,
-      "grad_norm": 0.0006584340590052307,
-      "learning_rate": 0.0001999971077609073,
-      "loss": 46.0,
-      "step": 31674
-    },
-    {
-      "epoch": 2.4217749488693925,
-      "grad_norm": 0.0007480731001123786,
-      "learning_rate": 0.00019999710757822217,
-      "loss": 46.0,
-      "step": 31675
-    },
-    {
-      "epoch": 2.4218514058527822,
-      "grad_norm": 0.002124340273439884,
-      "learning_rate": 0.0001999971073955313,
-      "loss": 46.0,
-      "step": 31676
-    },
-    {
-      "epoch": 2.4219278628361716,
-      "grad_norm": 0.004308842122554779,
-      "learning_rate": 0.00019999710721283463,
-      "loss": 46.0,
-      "step": 31677
-    },
-    {
-      "epoch": 2.4220043198195613,
-      "grad_norm": 0.0022425497882068157,
-      "learning_rate": 0.0001999971070301322,
-      "loss": 46.0,
-      "step": 31678
-    },
-    {
-      "epoch": 2.422080776802951,
-      "grad_norm": 0.0007315670954994857,
-      "learning_rate": 0.00019999710684742402,
-      "loss": 46.0,
-      "step": 31679
-    },
-    {
-      "epoch": 2.422157233786341,
-      "grad_norm": 0.0017121689161285758,
-      "learning_rate": 0.00019999710666471004,
-      "loss": 46.0,
-      "step": 31680
-    },
-    {
-      "epoch": 2.4222336907697306,
-      "grad_norm": 0.0007860715268179774,
-      "learning_rate": 0.0001999971064819903,
-      "loss": 46.0,
-      "step": 31681
-    },
-    {
-      "epoch": 2.4223101477531204,
-      "grad_norm": 0.00105209369212389,
-      "learning_rate": 0.0001999971062992648,
-      "loss": 46.0,
-      "step": 31682
-    },
-    {
-      "epoch": 2.42238660473651,
-      "grad_norm": 0.0011122459545731544,
-      "learning_rate": 0.00019999710611653352,
-      "loss": 46.0,
-      "step": 31683
-    },
-    {
-      "epoch": 2.4224630617199,
-      "grad_norm": 0.0007105217082425952,
-      "learning_rate": 0.00019999710593379647,
-      "loss": 46.0,
-      "step": 31684
-    },
-    {
-      "epoch": 2.4225395187032897,
-      "grad_norm": 0.0014697462320327759,
-      "learning_rate": 0.00019999710575105366,
-      "loss": 46.0,
-      "step": 31685
-    },
-    {
-      "epoch": 2.4226159756866794,
-      "grad_norm": 0.003164944937452674,
-      "learning_rate": 0.00019999710556830506,
-      "loss": 46.0,
-      "step": 31686
-    },
-    {
-      "epoch": 2.422692432670069,
-      "grad_norm": 0.0006941268220543861,
-      "learning_rate": 0.00019999710538555073,
-      "loss": 46.0,
-      "step": 31687
-    },
-    {
-      "epoch": 2.422768889653459,
-      "grad_norm": 0.0008146260515786707,
-      "learning_rate": 0.00019999710520279061,
-      "loss": 46.0,
-      "step": 31688
-    },
-    {
-      "epoch": 2.4228453466368483,
-      "grad_norm": 0.0024288722779601812,
-      "learning_rate": 0.0001999971050200247,
-      "loss": 46.0,
-      "step": 31689
-    },
-    {
-      "epoch": 2.422921803620238,
-      "grad_norm": 0.0005013248883187771,
-      "learning_rate": 0.00019999710483725304,
-      "loss": 46.0,
-      "step": 31690
-    },
-    {
-      "epoch": 2.4229982606036278,
-      "grad_norm": 0.0020023786928504705,
-      "learning_rate": 0.00019999710465447564,
-      "loss": 46.0,
-      "step": 31691
-    },
-    {
-      "epoch": 2.4230747175870175,
-      "grad_norm": 0.0013715730747208,
-      "learning_rate": 0.00019999710447169244,
-      "loss": 46.0,
-      "step": 31692
-    },
-    {
-      "epoch": 2.4231511745704073,
-      "grad_norm": 0.03180105984210968,
-      "learning_rate": 0.00019999710428890346,
-      "loss": 46.0,
-      "step": 31693
-    },
-    {
-      "epoch": 2.423227631553797,
-      "grad_norm": 0.0007216735975816846,
-      "learning_rate": 0.00019999710410610873,
-      "loss": 46.0,
-      "step": 31694
-    },
-    {
-      "epoch": 2.423304088537187,
-      "grad_norm": 0.0008931442280299962,
-      "learning_rate": 0.0001999971039233082,
-      "loss": 46.0,
-      "step": 31695
-    },
-    {
-      "epoch": 2.4233805455205766,
-      "grad_norm": 0.002911068731918931,
-      "learning_rate": 0.00019999710374050196,
-      "loss": 46.0,
-      "step": 31696
-    },
-    {
-      "epoch": 2.4234570025039663,
-      "grad_norm": 0.0006665420951321721,
-      "learning_rate": 0.0001999971035576899,
-      "loss": 46.0,
-      "step": 31697
-    },
-    {
-      "epoch": 2.423533459487356,
-      "grad_norm": 0.0019252111669629812,
-      "learning_rate": 0.00019999710337487207,
-      "loss": 46.0,
-      "step": 31698
-    },
-    {
-      "epoch": 2.4236099164707454,
-      "grad_norm": 0.001246387604624033,
-      "learning_rate": 0.00019999710319204848,
-      "loss": 46.0,
-      "step": 31699
-    },
-    {
-      "epoch": 2.423686373454135,
-      "grad_norm": 0.0019187004072591662,
-      "learning_rate": 0.00019999710300921915,
-      "loss": 46.0,
-      "step": 31700
-    },
-    {
-      "epoch": 2.423762830437525,
-      "grad_norm": 0.0011649333173409104,
-      "learning_rate": 0.000199997102826384,
-      "loss": 46.0,
-      "step": 31701
-    },
-    {
-      "epoch": 2.4238392874209147,
-      "grad_norm": 0.005542614497244358,
-      "learning_rate": 0.00019999710264354312,
-      "loss": 46.0,
-      "step": 31702
-    },
-    {
-      "epoch": 2.4239157444043045,
-      "grad_norm": 0.0004525418335106224,
-      "learning_rate": 0.00019999710246069647,
-      "loss": 46.0,
-      "step": 31703
-    },
-    {
-      "epoch": 2.4239922013876942,
-      "grad_norm": 0.0006581030320376158,
-      "learning_rate": 0.00019999710227784406,
-      "loss": 46.0,
-      "step": 31704
-    },
-    {
-      "epoch": 2.424068658371084,
-      "grad_norm": 0.0010272247018292546,
-      "learning_rate": 0.00019999710209498586,
-      "loss": 46.0,
-      "step": 31705
-    },
-    {
-      "epoch": 2.4241451153544737,
-      "grad_norm": 0.001188663300126791,
-      "learning_rate": 0.00019999710191212188,
-      "loss": 46.0,
-      "step": 31706
-    },
-    {
-      "epoch": 2.4242215723378635,
-      "grad_norm": 0.0005571183864958584,
-      "learning_rate": 0.00019999710172925215,
-      "loss": 46.0,
-      "step": 31707
-    },
-    {
-      "epoch": 2.4242980293212533,
-      "grad_norm": 0.0007985879783518612,
-      "learning_rate": 0.00019999710154637663,
-      "loss": 46.0,
-      "step": 31708
-    },
-    {
-      "epoch": 2.424374486304643,
-      "grad_norm": 0.0010718220146372914,
-      "learning_rate": 0.00019999710136349536,
-      "loss": 46.0,
-      "step": 31709
-    },
-    {
-      "epoch": 2.424450943288033,
-      "grad_norm": 0.0004937925841659307,
-      "learning_rate": 0.00019999710118060831,
-      "loss": 46.0,
-      "step": 31710
-    },
-    {
-      "epoch": 2.424527400271422,
-      "grad_norm": 0.0007205493748188019,
-      "learning_rate": 0.00019999710099771552,
-      "loss": 46.0,
-      "step": 31711
-    },
-    {
-      "epoch": 2.424603857254812,
-      "grad_norm": 0.0011248005321249366,
-      "learning_rate": 0.00019999710081481693,
-      "loss": 46.0,
-      "step": 31712
-    },
-    {
-      "epoch": 2.4246803142382016,
-      "grad_norm": 0.0012535803252831101,
-      "learning_rate": 0.00019999710063191257,
-      "loss": 46.0,
-      "step": 31713
-    },
-    {
-      "epoch": 2.4247567712215914,
-      "grad_norm": 0.001577549846842885,
-      "learning_rate": 0.00019999710044900243,
-      "loss": 46.0,
-      "step": 31714
-    },
-    {
-      "epoch": 2.424833228204981,
-      "grad_norm": 0.002583804540336132,
-      "learning_rate": 0.00019999710026608655,
-      "loss": 46.0,
-      "step": 31715
-    },
-    {
-      "epoch": 2.424909685188371,
-      "grad_norm": 0.0013151037273928523,
-      "learning_rate": 0.0001999971000831649,
-      "loss": 46.0,
-      "step": 31716
-    },
-    {
-      "epoch": 2.4249861421717607,
-      "grad_norm": 0.0008083076681941748,
-      "learning_rate": 0.00019999709990023746,
-      "loss": 46.0,
-      "step": 31717
-    },
-    {
-      "epoch": 2.4250625991551504,
-      "grad_norm": 0.0055492958053946495,
-      "learning_rate": 0.00019999709971730428,
-      "loss": 46.0,
-      "step": 31718
-    },
-    {
-      "epoch": 2.42513905613854,
-      "grad_norm": 0.0009085440542548895,
-      "learning_rate": 0.0001999970995343653,
-      "loss": 46.0,
-      "step": 31719
-    },
-    {
-      "epoch": 2.42521551312193,
-      "grad_norm": 0.0007682053255848587,
-      "learning_rate": 0.00019999709935142055,
-      "loss": 46.0,
-      "step": 31720
-    },
-    {
-      "epoch": 2.4252919701053193,
-      "grad_norm": 0.0011408563004806638,
-      "learning_rate": 0.00019999709916847006,
-      "loss": 46.0,
-      "step": 31721
-    },
-    {
-      "epoch": 2.425368427088709,
-      "grad_norm": 0.003419312182813883,
-      "learning_rate": 0.00019999709898551379,
-      "loss": 46.0,
-      "step": 31722
-    },
-    {
-      "epoch": 2.425444884072099,
-      "grad_norm": 0.0007048809784464538,
-      "learning_rate": 0.00019999709880255174,
-      "loss": 46.0,
-      "step": 31723
-    },
-    {
-      "epoch": 2.4255213410554886,
-      "grad_norm": 0.001151367207057774,
-      "learning_rate": 0.00019999709861958392,
-      "loss": 46.0,
-      "step": 31724
-    },
-    {
-      "epoch": 2.4255977980388783,
-      "grad_norm": 0.0008718199096620083,
-      "learning_rate": 0.0001999970984366103,
-      "loss": 46.0,
-      "step": 31725
-    },
-    {
-      "epoch": 2.425674255022268,
-      "grad_norm": 0.001035741763189435,
-      "learning_rate": 0.00019999709825363097,
-      "loss": 46.0,
-      "step": 31726
-    },
-    {
-      "epoch": 2.425750712005658,
-      "grad_norm": 0.0016902287025004625,
-      "learning_rate": 0.00019999709807064586,
-      "loss": 46.0,
-      "step": 31727
-    },
-    {
-      "epoch": 2.4258271689890476,
-      "grad_norm": 0.007216343656182289,
-      "learning_rate": 0.00019999709788765495,
-      "loss": 46.0,
-      "step": 31728
-    },
-    {
-      "epoch": 2.4259036259724374,
-      "grad_norm": 0.0051716906018555164,
-      "learning_rate": 0.00019999709770465832,
-      "loss": 46.0,
-      "step": 31729
-    },
-    {
-      "epoch": 2.425980082955827,
-      "grad_norm": 0.0012299424270167947,
-      "learning_rate": 0.00019999709752165586,
-      "loss": 46.0,
-      "step": 31730
-    },
-    {
-      "epoch": 2.426056539939217,
-      "grad_norm": 0.0009762330446392298,
-      "learning_rate": 0.00019999709733864766,
-      "loss": 46.0,
-      "step": 31731
-    },
-    {
-      "epoch": 2.4261329969226066,
-      "grad_norm": 0.000712329987436533,
-      "learning_rate": 0.0001999970971556337,
-      "loss": 46.0,
-      "step": 31732
-    },
-    {
-      "epoch": 2.426209453905996,
-      "grad_norm": 0.0041829003021121025,
-      "learning_rate": 0.00019999709697261398,
-      "loss": 46.0,
-      "step": 31733
-    },
-    {
-      "epoch": 2.4262859108893857,
-      "grad_norm": 0.0012989091919735074,
-      "learning_rate": 0.00019999709678958843,
-      "loss": 46.0,
-      "step": 31734
-    },
-    {
-      "epoch": 2.4263623678727755,
-      "grad_norm": 0.001802045269869268,
-      "learning_rate": 0.0001999970966065572,
-      "loss": 46.0,
-      "step": 31735
-    },
-    {
-      "epoch": 2.4264388248561652,
-      "grad_norm": 0.004469194449484348,
-      "learning_rate": 0.00019999709642352015,
-      "loss": 46.0,
-      "step": 31736
-    },
-    {
-      "epoch": 2.426515281839555,
-      "grad_norm": 0.0012777147348970175,
-      "learning_rate": 0.0001999970962404773,
-      "loss": 46.0,
-      "step": 31737
-    },
-    {
-      "epoch": 2.4265917388229448,
-      "grad_norm": 0.0007960747461766005,
-      "learning_rate": 0.0001999970960574287,
-      "loss": 46.0,
-      "step": 31738
-    },
-    {
-      "epoch": 2.4266681958063345,
-      "grad_norm": 0.0008298384491354227,
-      "learning_rate": 0.00019999709587437437,
-      "loss": 46.0,
-      "step": 31739
-    },
-    {
-      "epoch": 2.4267446527897243,
-      "grad_norm": 0.0012634429149329662,
-      "learning_rate": 0.00019999709569131424,
-      "loss": 46.0,
-      "step": 31740
-    },
-    {
-      "epoch": 2.426821109773114,
-      "grad_norm": 0.0007753687677904963,
-      "learning_rate": 0.00019999709550824835,
-      "loss": 46.0,
-      "step": 31741
-    },
-    {
-      "epoch": 2.426897566756504,
-      "grad_norm": 0.0021268592681735754,
-      "learning_rate": 0.00019999709532517667,
-      "loss": 46.0,
-      "step": 31742
-    },
-    {
-      "epoch": 2.426974023739893,
-      "grad_norm": 0.004711328539997339,
-      "learning_rate": 0.00019999709514209927,
-      "loss": 46.0,
-      "step": 31743
-    },
-    {
-      "epoch": 2.427050480723283,
-      "grad_norm": 0.0008751682471483946,
-      "learning_rate": 0.00019999709495901606,
-      "loss": 46.0,
-      "step": 31744
-    },
-    {
-      "epoch": 2.4271269377066726,
-      "grad_norm": 0.001242566853761673,
-      "learning_rate": 0.00019999709477592709,
-      "loss": 46.0,
-      "step": 31745
-    },
-    {
-      "epoch": 2.4272033946900624,
-      "grad_norm": 0.0014638626016676426,
-      "learning_rate": 0.00019999709459283236,
-      "loss": 46.0,
-      "step": 31746
-    },
-    {
-      "epoch": 2.427279851673452,
-      "grad_norm": 0.0010472440626472235,
-      "learning_rate": 0.00019999709440973184,
-      "loss": 46.0,
-      "step": 31747
-    },
-    {
-      "epoch": 2.427356308656842,
-      "grad_norm": 0.0005147071206010878,
-      "learning_rate": 0.00019999709422662557,
-      "loss": 46.0,
-      "step": 31748
-    },
-    {
-      "epoch": 2.4274327656402317,
-      "grad_norm": 0.0014069130411371589,
-      "learning_rate": 0.0001999970940435135,
-      "loss": 46.0,
-      "step": 31749
-    },
-    {
-      "epoch": 2.4275092226236215,
-      "grad_norm": 0.0008542793802917004,
-      "learning_rate": 0.0001999970938603957,
-      "loss": 46.0,
-      "step": 31750
-    },
-    {
-      "epoch": 2.427585679607011,
-      "grad_norm": 0.001664007198996842,
-      "learning_rate": 0.00019999709367727212,
-      "loss": 46.0,
-      "step": 31751
-    },
-    {
-      "epoch": 2.427662136590401,
-      "grad_norm": 0.0010284207528457046,
-      "learning_rate": 0.00019999709349414278,
-      "loss": 46.0,
-      "step": 31752
-    },
-    {
-      "epoch": 2.4277385935737907,
-      "grad_norm": 0.0014334612060338259,
-      "learning_rate": 0.00019999709331100762,
-      "loss": 46.0,
-      "step": 31753
-    },
-    {
-      "epoch": 2.4278150505571805,
-      "grad_norm": 0.0014943047426640987,
-      "learning_rate": 0.00019999709312786674,
-      "loss": 46.0,
-      "step": 31754
-    },
-    {
-      "epoch": 2.42789150754057,
-      "grad_norm": 0.0010804033372551203,
-      "learning_rate": 0.00019999709294472008,
-      "loss": 46.0,
-      "step": 31755
-    },
-    {
-      "epoch": 2.4279679645239596,
-      "grad_norm": 0.0009538778685964644,
-      "learning_rate": 0.00019999709276156765,
-      "loss": 46.0,
-      "step": 31756
-    },
-    {
-      "epoch": 2.4280444215073493,
-      "grad_norm": 0.0008481864933855832,
-      "learning_rate": 0.00019999709257840945,
-      "loss": 46.0,
-      "step": 31757
-    },
-    {
-      "epoch": 2.428120878490739,
-      "grad_norm": 0.005737654864788055,
-      "learning_rate": 0.00019999709239524547,
-      "loss": 46.0,
-      "step": 31758
-    },
-    {
-      "epoch": 2.428197335474129,
-      "grad_norm": 0.003212561132386327,
-      "learning_rate": 0.00019999709221207575,
-      "loss": 46.0,
-      "step": 31759
-    },
-    {
-      "epoch": 2.4282737924575186,
-      "grad_norm": 0.004905946087092161,
-      "learning_rate": 0.00019999709202890022,
-      "loss": 46.0,
-      "step": 31760
-    },
-    {
-      "epoch": 2.4283502494409084,
-      "grad_norm": 0.0010980576043948531,
-      "learning_rate": 0.00019999709184571893,
-      "loss": 46.0,
-      "step": 31761
-    },
-    {
-      "epoch": 2.428426706424298,
-      "grad_norm": 0.0005742415669374168,
-      "learning_rate": 0.0001999970916625319,
-      "loss": 46.0,
-      "step": 31762
-    },
-    {
-      "epoch": 2.428503163407688,
-      "grad_norm": 0.0009037034469656646,
-      "learning_rate": 0.00019999709147933907,
-      "loss": 46.0,
-      "step": 31763
-    },
-    {
-      "epoch": 2.428579620391077,
-      "grad_norm": 0.0034793277736753225,
-      "learning_rate": 0.0001999970912961405,
-      "loss": 46.0,
-      "step": 31764
-    },
-    {
-      "epoch": 2.428656077374467,
-      "grad_norm": 0.0029805137310177088,
-      "learning_rate": 0.00019999709111293611,
-      "loss": 46.0,
-      "step": 31765
-    },
-    {
-      "epoch": 2.4287325343578567,
-      "grad_norm": 0.0028320078272372484,
-      "learning_rate": 0.000199997090929726,
-      "loss": 46.0,
-      "step": 31766
-    },
-    {
-      "epoch": 2.4288089913412465,
-      "grad_norm": 0.0004579621599987149,
-      "learning_rate": 0.00019999709074651012,
-      "loss": 46.0,
-      "step": 31767
-    },
-    {
-      "epoch": 2.4288854483246363,
-      "grad_norm": 0.002480153925716877,
-      "learning_rate": 0.00019999709056328847,
-      "loss": 46.0,
-      "step": 31768
-    },
-    {
-      "epoch": 2.428961905308026,
-      "grad_norm": 0.0014389334246516228,
-      "learning_rate": 0.00019999709038006104,
-      "loss": 46.0,
-      "step": 31769
-    },
-    {
-      "epoch": 2.429038362291416,
-      "grad_norm": 0.0009162189671769738,
-      "learning_rate": 0.0001999970901968278,
-      "loss": 46.0,
-      "step": 31770
-    },
-    {
-      "epoch": 2.4291148192748055,
-      "grad_norm": 0.001437582541257143,
-      "learning_rate": 0.00019999709001358883,
-      "loss": 46.0,
-      "step": 31771
-    },
-    {
-      "epoch": 2.4291912762581953,
-      "grad_norm": 0.0005465935682877898,
-      "learning_rate": 0.0001999970898303441,
-      "loss": 46.0,
-      "step": 31772
-    },
-    {
-      "epoch": 2.429267733241585,
-      "grad_norm": 0.0018559836316853762,
-      "learning_rate": 0.0001999970896470936,
-      "loss": 46.0,
-      "step": 31773
-    },
-    {
-      "epoch": 2.429344190224975,
-      "grad_norm": 0.0007934711175039411,
-      "learning_rate": 0.00019999708946383732,
-      "loss": 46.0,
-      "step": 31774
-    },
-    {
-      "epoch": 2.4294206472083646,
-      "grad_norm": 0.0009905733168125153,
-      "learning_rate": 0.00019999708928057527,
-      "loss": 46.0,
-      "step": 31775
-    },
-    {
-      "epoch": 2.4294971041917544,
-      "grad_norm": 0.00230024172924459,
-      "learning_rate": 0.00019999708909730743,
-      "loss": 46.0,
-      "step": 31776
-    },
-    {
-      "epoch": 2.4295735611751437,
-      "grad_norm": 0.0014973955694586039,
-      "learning_rate": 0.00019999708891403387,
-      "loss": 46.0,
-      "step": 31777
-    },
-    {
-      "epoch": 2.4296500181585334,
-      "grad_norm": 0.0024959947913885117,
-      "learning_rate": 0.0001999970887307545,
-      "loss": 46.0,
-      "step": 31778
-    },
-    {
-      "epoch": 2.429726475141923,
-      "grad_norm": 0.012260590679943562,
-      "learning_rate": 0.0001999970885474694,
-      "loss": 46.0,
-      "step": 31779
-    },
-    {
-      "epoch": 2.429802932125313,
-      "grad_norm": 0.0025973531883209944,
-      "learning_rate": 0.00019999708836417849,
-      "loss": 46.0,
-      "step": 31780
-    },
-    {
-      "epoch": 2.4298793891087027,
-      "grad_norm": 0.001155383768491447,
-      "learning_rate": 0.00019999708818088183,
-      "loss": 46.0,
-      "step": 31781
-    },
-    {
-      "epoch": 2.4299558460920925,
-      "grad_norm": 0.0009758244268596172,
-      "learning_rate": 0.0001999970879975794,
-      "loss": 46.0,
-      "step": 31782
-    },
-    {
-      "epoch": 2.4300323030754822,
-      "grad_norm": 0.0007594484486617148,
-      "learning_rate": 0.00019999708781427117,
-      "loss": 46.0,
-      "step": 31783
-    },
-    {
-      "epoch": 2.430108760058872,
-      "grad_norm": 0.0009277135832235217,
-      "learning_rate": 0.00019999708763095722,
-      "loss": 46.0,
-      "step": 31784
-    },
-    {
-      "epoch": 2.4301852170422618,
-      "grad_norm": 0.003238823963329196,
-      "learning_rate": 0.00019999708744763747,
-      "loss": 46.0,
-      "step": 31785
-    },
-    {
-      "epoch": 2.430261674025651,
-      "grad_norm": 0.0010456267045810819,
-      "learning_rate": 0.00019999708726431195,
-      "loss": 46.0,
-      "step": 31786
-    },
-    {
-      "epoch": 2.430338131009041,
-      "grad_norm": 0.0018340791575610638,
-      "learning_rate": 0.00019999708708098068,
-      "loss": 46.0,
-      "step": 31787
-    },
-    {
-      "epoch": 2.4304145879924306,
-      "grad_norm": 0.006502361968159676,
-      "learning_rate": 0.00019999708689764364,
-      "loss": 46.0,
-      "step": 31788
-    },
-    {
-      "epoch": 2.4304910449758204,
-      "grad_norm": 0.0012338829692453146,
-      "learning_rate": 0.0001999970867143008,
-      "loss": 46.0,
-      "step": 31789
-    },
-    {
-      "epoch": 2.43056750195921,
-      "grad_norm": 0.0012245511170476675,
-      "learning_rate": 0.0001999970865309522,
-      "loss": 46.0,
-      "step": 31790
-    },
-    {
-      "epoch": 2.4306439589426,
-      "grad_norm": 0.0005599466385319829,
-      "learning_rate": 0.00019999708634759787,
-      "loss": 46.0,
-      "step": 31791
-    },
-    {
-      "epoch": 2.4307204159259896,
-      "grad_norm": 0.001547261024825275,
-      "learning_rate": 0.00019999708616423774,
-      "loss": 46.0,
-      "step": 31792
-    },
-    {
-      "epoch": 2.4307968729093794,
-      "grad_norm": 0.002378972480073571,
-      "learning_rate": 0.00019999708598087183,
-      "loss": 46.0,
-      "step": 31793
-    },
-    {
-      "epoch": 2.430873329892769,
-      "grad_norm": 0.0019048320828005672,
-      "learning_rate": 0.00019999708579750017,
-      "loss": 46.0,
-      "step": 31794
-    },
-    {
-      "epoch": 2.430949786876159,
-      "grad_norm": 0.0014762452337890863,
-      "learning_rate": 0.00019999708561412274,
-      "loss": 46.0,
-      "step": 31795
-    },
-    {
-      "epoch": 2.4310262438595487,
-      "grad_norm": 0.0006561970221810043,
-      "learning_rate": 0.00019999708543073954,
-      "loss": 46.0,
-      "step": 31796
-    },
-    {
-      "epoch": 2.4311027008429384,
-      "grad_norm": 0.008589284494519234,
-      "learning_rate": 0.00019999708524735056,
-      "loss": 46.0,
-      "step": 31797
-    },
-    {
-      "epoch": 2.431179157826328,
-      "grad_norm": 0.001867231447249651,
-      "learning_rate": 0.00019999708506395581,
-      "loss": 46.0,
-      "step": 31798
-    },
-    {
-      "epoch": 2.4312556148097175,
-      "grad_norm": 0.0009940499439835548,
-      "learning_rate": 0.00019999708488055532,
-      "loss": 46.0,
-      "step": 31799
-    },
-    {
-      "epoch": 2.4313320717931073,
-      "grad_norm": 0.0003697279025800526,
-      "learning_rate": 0.00019999708469714902,
-      "loss": 46.0,
-      "step": 31800
-    },
-    {
-      "epoch": 2.431408528776497,
-      "grad_norm": 0.000913189840503037,
-      "learning_rate": 0.00019999708451373698,
-      "loss": 46.0,
-      "step": 31801
-    },
-    {
-      "epoch": 2.431484985759887,
-      "grad_norm": 0.00109881442040205,
-      "learning_rate": 0.00019999708433031917,
-      "loss": 46.0,
-      "step": 31802
-    },
-    {
-      "epoch": 2.4315614427432766,
-      "grad_norm": 0.0009672069572843611,
-      "learning_rate": 0.00019999708414689558,
-      "loss": 46.0,
-      "step": 31803
-    },
-    {
-      "epoch": 2.4316378997266663,
-      "grad_norm": 0.0010592021280899644,
-      "learning_rate": 0.00019999708396346621,
-      "loss": 46.0,
-      "step": 31804
-    },
-    {
-      "epoch": 2.431714356710056,
-      "grad_norm": 0.0005268522654660046,
-      "learning_rate": 0.00019999708378003108,
-      "loss": 46.0,
-      "step": 31805
-    },
-    {
-      "epoch": 2.431790813693446,
-      "grad_norm": 0.00042245833901688457,
-      "learning_rate": 0.00019999708359659017,
-      "loss": 46.0,
-      "step": 31806
-    },
-    {
-      "epoch": 2.4318672706768356,
-      "grad_norm": 0.003212742740288377,
-      "learning_rate": 0.00019999708341314354,
-      "loss": 46.0,
-      "step": 31807
-    },
-    {
-      "epoch": 2.431943727660225,
-      "grad_norm": 0.00044132917537353933,
-      "learning_rate": 0.00019999708322969109,
-      "loss": 46.0,
-      "step": 31808
-    },
-    {
-      "epoch": 2.4320201846436147,
-      "grad_norm": 0.0009283123072236776,
-      "learning_rate": 0.00019999708304623288,
-      "loss": 46.0,
-      "step": 31809
-    },
-    {
-      "epoch": 2.4320966416270045,
-      "grad_norm": 0.0013121285010129213,
-      "learning_rate": 0.0001999970828627689,
-      "loss": 46.0,
-      "step": 31810
-    },
-    {
-      "epoch": 2.432173098610394,
-      "grad_norm": 0.0009874447714537382,
-      "learning_rate": 0.0001999970826792992,
-      "loss": 46.0,
-      "step": 31811
-    },
-    {
-      "epoch": 2.432249555593784,
-      "grad_norm": 0.0009543704800307751,
-      "learning_rate": 0.00019999708249582364,
-      "loss": 46.0,
-      "step": 31812
-    },
-    {
-      "epoch": 2.4323260125771737,
-      "grad_norm": 0.0035185085143893957,
-      "learning_rate": 0.00019999708231234237,
-      "loss": 46.0,
-      "step": 31813
-    },
-    {
-      "epoch": 2.4324024695605635,
-      "grad_norm": 0.0032497625797986984,
-      "learning_rate": 0.0001999970821288553,
-      "loss": 46.0,
-      "step": 31814
-    },
-    {
-      "epoch": 2.4324789265439533,
-      "grad_norm": 0.0007659861003048718,
-      "learning_rate": 0.0001999970819453625,
-      "loss": 46.0,
-      "step": 31815
-    },
-    {
-      "epoch": 2.432555383527343,
-      "grad_norm": 0.0014378949999809265,
-      "learning_rate": 0.00019999708176186392,
-      "loss": 46.0,
-      "step": 31816
-    },
-    {
-      "epoch": 2.432631840510733,
-      "grad_norm": 0.008012588135898113,
-      "learning_rate": 0.00019999708157835956,
-      "loss": 46.0,
-      "step": 31817
-    },
-    {
-      "epoch": 2.4327082974941225,
-      "grad_norm": 0.001989668468013406,
-      "learning_rate": 0.00019999708139484943,
-      "loss": 46.0,
-      "step": 31818
-    },
-    {
-      "epoch": 2.4327847544775123,
-      "grad_norm": 0.0018083906034007668,
-      "learning_rate": 0.00019999708121133352,
-      "loss": 46.0,
-      "step": 31819
-    },
-    {
-      "epoch": 2.4328612114609016,
-      "grad_norm": 0.0019533506128937006,
-      "learning_rate": 0.00019999708102781186,
-      "loss": 46.0,
-      "step": 31820
-    },
-    {
-      "epoch": 2.4329376684442914,
-      "grad_norm": 0.0029634260572493076,
-      "learning_rate": 0.0001999970808442844,
-      "loss": 46.0,
-      "step": 31821
-    },
-    {
-      "epoch": 2.433014125427681,
-      "grad_norm": 0.0013277167454361916,
-      "learning_rate": 0.00019999708066075123,
-      "loss": 46.0,
-      "step": 31822
-    },
-    {
-      "epoch": 2.433090582411071,
-      "grad_norm": 0.0011007209541276097,
-      "learning_rate": 0.00019999708047721226,
-      "loss": 46.0,
-      "step": 31823
-    },
-    {
-      "epoch": 2.4331670393944607,
-      "grad_norm": 0.0014727492816746235,
-      "learning_rate": 0.0001999970802936675,
-      "loss": 46.0,
-      "step": 31824
-    },
-    {
-      "epoch": 2.4332434963778504,
-      "grad_norm": 0.0006560039473697543,
-      "learning_rate": 0.00019999708011011696,
-      "loss": 46.0,
-      "step": 31825
-    },
-    {
-      "epoch": 2.43331995336124,
-      "grad_norm": 0.0007435241132043302,
-      "learning_rate": 0.0001999970799265607,
-      "loss": 46.0,
-      "step": 31826
-    },
-    {
-      "epoch": 2.43339641034463,
-      "grad_norm": 0.0004581456014420837,
-      "learning_rate": 0.00019999707974299865,
-      "loss": 46.0,
-      "step": 31827
-    },
-    {
-      "epoch": 2.4334728673280197,
-      "grad_norm": 0.0006350258481688797,
-      "learning_rate": 0.00019999707955943084,
-      "loss": 46.0,
-      "step": 31828
-    },
-    {
-      "epoch": 2.4335493243114095,
-      "grad_norm": 0.0023034026380628347,
-      "learning_rate": 0.00019999707937585725,
-      "loss": 46.0,
-      "step": 31829
-    },
-    {
-      "epoch": 2.433625781294799,
-      "grad_norm": 0.0016607922734692693,
-      "learning_rate": 0.0001999970791922779,
-      "loss": 46.0,
-      "step": 31830
-    },
-    {
-      "epoch": 2.4337022382781885,
-      "grad_norm": 0.001769884373061359,
-      "learning_rate": 0.00019999707900869275,
-      "loss": 46.0,
-      "step": 31831
-    },
-    {
-      "epoch": 2.4337786952615783,
-      "grad_norm": 0.0013325974578037858,
-      "learning_rate": 0.00019999707882510187,
-      "loss": 46.0,
-      "step": 31832
-    },
-    {
-      "epoch": 2.433855152244968,
-      "grad_norm": 0.0011648869840428233,
-      "learning_rate": 0.0001999970786415052,
-      "loss": 46.0,
-      "step": 31833
-    },
-    {
-      "epoch": 2.433931609228358,
-      "grad_norm": 0.000647606560960412,
-      "learning_rate": 0.00019999707845790273,
-      "loss": 46.0,
-      "step": 31834
-    },
-    {
-      "epoch": 2.4340080662117476,
-      "grad_norm": 0.0010726067703217268,
-      "learning_rate": 0.00019999707827429456,
-      "loss": 46.0,
-      "step": 31835
-    },
-    {
-      "epoch": 2.4340845231951374,
-      "grad_norm": 0.0014563523000106215,
-      "learning_rate": 0.00019999707809068059,
-      "loss": 46.0,
-      "step": 31836
-    },
-    {
-      "epoch": 2.434160980178527,
-      "grad_norm": 0.0017075929790735245,
-      "learning_rate": 0.00019999707790706084,
-      "loss": 46.0,
-      "step": 31837
-    },
-    {
-      "epoch": 2.434237437161917,
-      "grad_norm": 0.0021645165979862213,
-      "learning_rate": 0.00019999707772343532,
-      "loss": 46.0,
-      "step": 31838
-    },
-    {
-      "epoch": 2.4343138941453066,
-      "grad_norm": 0.0009498960571363568,
-      "learning_rate": 0.00019999707753980402,
-      "loss": 46.0,
-      "step": 31839
-    },
-    {
-      "epoch": 2.4343903511286964,
-      "grad_norm": 0.002539821667596698,
-      "learning_rate": 0.00019999707735616698,
-      "loss": 46.0,
-      "step": 31840
-    },
-    {
-      "epoch": 2.434466808112086,
-      "grad_norm": 0.0008213865221478045,
-      "learning_rate": 0.00019999707717252417,
-      "loss": 46.0,
-      "step": 31841
-    },
-    {
-      "epoch": 2.4345432650954755,
-      "grad_norm": 0.003667982295155525,
-      "learning_rate": 0.00019999707698887558,
-      "loss": 46.0,
-      "step": 31842
-    },
-    {
-      "epoch": 2.4346197220788652,
-      "grad_norm": 0.0010683451546356082,
-      "learning_rate": 0.0001999970768052212,
-      "loss": 46.0,
-      "step": 31843
-    },
-    {
-      "epoch": 2.434696179062255,
-      "grad_norm": 0.0005103599978610873,
-      "learning_rate": 0.00019999707662156108,
-      "loss": 46.0,
-      "step": 31844
-    },
-    {
-      "epoch": 2.4347726360456448,
-      "grad_norm": 0.0010878884932026267,
-      "learning_rate": 0.00019999707643789517,
-      "loss": 46.0,
-      "step": 31845
-    },
-    {
-      "epoch": 2.4348490930290345,
-      "grad_norm": 0.002139989286661148,
-      "learning_rate": 0.0001999970762542235,
-      "loss": 46.0,
-      "step": 31846
-    },
-    {
-      "epoch": 2.4349255500124243,
-      "grad_norm": 0.0013321478618308902,
-      "learning_rate": 0.00019999707607054606,
-      "loss": 46.0,
-      "step": 31847
-    },
-    {
-      "epoch": 2.435002006995814,
-      "grad_norm": 0.0012961842585355043,
-      "learning_rate": 0.00019999707588686286,
-      "loss": 46.0,
-      "step": 31848
-    },
-    {
-      "epoch": 2.435078463979204,
-      "grad_norm": 0.0013146473793312907,
-      "learning_rate": 0.0001999970757031739,
-      "loss": 46.0,
-      "step": 31849
-    },
-    {
-      "epoch": 2.4351549209625936,
-      "grad_norm": 0.0008391342125833035,
-      "learning_rate": 0.00019999707551947914,
-      "loss": 46.0,
-      "step": 31850
-    },
-    {
-      "epoch": 2.4352313779459833,
-      "grad_norm": 0.0009723032708279788,
-      "learning_rate": 0.00019999707533577862,
-      "loss": 46.0,
-      "step": 31851
-    },
-    {
-      "epoch": 2.4353078349293726,
-      "grad_norm": 0.0024366267025470734,
-      "learning_rate": 0.00019999707515207233,
-      "loss": 46.0,
-      "step": 31852
-    },
-    {
-      "epoch": 2.4353842919127624,
-      "grad_norm": 0.0009454222745262086,
-      "learning_rate": 0.0001999970749683603,
-      "loss": 46.0,
-      "step": 31853
-    },
-    {
-      "epoch": 2.435460748896152,
-      "grad_norm": 0.0012927548959851265,
-      "learning_rate": 0.00019999707478464247,
-      "loss": 46.0,
-      "step": 31854
-    },
-    {
-      "epoch": 2.435537205879542,
-      "grad_norm": 0.0013167349388822913,
-      "learning_rate": 0.00019999707460091888,
-      "loss": 46.0,
-      "step": 31855
-    },
-    {
-      "epoch": 2.4356136628629317,
-      "grad_norm": 0.00020622643933165818,
-      "learning_rate": 0.00019999707441718952,
-      "loss": 46.0,
-      "step": 31856
-    },
-    {
-      "epoch": 2.4356901198463214,
-      "grad_norm": 0.000532140547875315,
-      "learning_rate": 0.0001999970742334544,
-      "loss": 46.0,
-      "step": 31857
-    },
-    {
-      "epoch": 2.435766576829711,
-      "grad_norm": 0.0008245834615081549,
-      "learning_rate": 0.0001999970740497135,
-      "loss": 46.0,
-      "step": 31858
-    },
-    {
-      "epoch": 2.435843033813101,
-      "grad_norm": 0.00235170335508883,
-      "learning_rate": 0.00019999707386596683,
-      "loss": 46.0,
-      "step": 31859
-    },
-    {
-      "epoch": 2.4359194907964907,
-      "grad_norm": 0.0011770181590691209,
-      "learning_rate": 0.0001999970736822144,
-      "loss": 46.0,
-      "step": 31860
-    },
-    {
-      "epoch": 2.4359959477798805,
-      "grad_norm": 0.0007052259752526879,
-      "learning_rate": 0.00019999707349845617,
-      "loss": 46.0,
-      "step": 31861
-    },
-    {
-      "epoch": 2.4360724047632702,
-      "grad_norm": 0.003323481883853674,
-      "learning_rate": 0.0001999970733146922,
-      "loss": 46.0,
-      "step": 31862
-    },
-    {
-      "epoch": 2.43614886174666,
-      "grad_norm": 0.0009671826846897602,
-      "learning_rate": 0.00019999707313092245,
-      "loss": 46.0,
-      "step": 31863
-    },
-    {
-      "epoch": 2.4362253187300493,
-      "grad_norm": 0.0006834216765128076,
-      "learning_rate": 0.00019999707294714693,
-      "loss": 46.0,
-      "step": 31864
-    },
-    {
-      "epoch": 2.436301775713439,
-      "grad_norm": 0.0007340940064750612,
-      "learning_rate": 0.00019999707276336566,
-      "loss": 46.0,
-      "step": 31865
-    },
-    {
-      "epoch": 2.436378232696829,
-      "grad_norm": 0.0011766324751079082,
-      "learning_rate": 0.0001999970725795786,
-      "loss": 46.0,
-      "step": 31866
-    },
-    {
-      "epoch": 2.4364546896802186,
-      "grad_norm": 0.00044746213825419545,
-      "learning_rate": 0.00019999707239578578,
-      "loss": 46.0,
-      "step": 31867
-    },
-    {
-      "epoch": 2.4365311466636084,
-      "grad_norm": 0.0004813310515601188,
-      "learning_rate": 0.0001999970722119872,
-      "loss": 46.0,
-      "step": 31868
-    },
-    {
-      "epoch": 2.436607603646998,
-      "grad_norm": 0.0016076306346803904,
-      "learning_rate": 0.00019999707202818283,
-      "loss": 46.0,
-      "step": 31869
-    },
-    {
-      "epoch": 2.436684060630388,
-      "grad_norm": 0.0016828869702294469,
-      "learning_rate": 0.0001999970718443727,
-      "loss": 46.0,
-      "step": 31870
-    },
-    {
-      "epoch": 2.4367605176137777,
-      "grad_norm": 0.003962460905313492,
-      "learning_rate": 0.0001999970716605568,
-      "loss": 46.0,
-      "step": 31871
-    },
-    {
-      "epoch": 2.4368369745971674,
-      "grad_norm": 0.0010938819032162428,
-      "learning_rate": 0.0001999970714767351,
-      "loss": 46.0,
-      "step": 31872
-    },
-    {
-      "epoch": 2.436913431580557,
-      "grad_norm": 0.0016832476248964667,
-      "learning_rate": 0.0001999970712929077,
-      "loss": 46.0,
-      "step": 31873
-    },
-    {
-      "epoch": 2.4369898885639465,
-      "grad_norm": 0.0013671672204509377,
-      "learning_rate": 0.00019999707110907448,
-      "loss": 46.0,
-      "step": 31874
-    },
-    {
-      "epoch": 2.4370663455473363,
-      "grad_norm": 0.001036601490341127,
-      "learning_rate": 0.0001999970709252355,
-      "loss": 46.0,
-      "step": 31875
-    },
-    {
-      "epoch": 2.437142802530726,
-      "grad_norm": 0.0007080077775754035,
-      "learning_rate": 0.00019999707074139076,
-      "loss": 46.0,
-      "step": 31876
-    },
-    {
-      "epoch": 2.4372192595141158,
-      "grad_norm": 0.0007337298011407256,
-      "learning_rate": 0.00019999707055754024,
-      "loss": 46.0,
-      "step": 31877
-    },
-    {
-      "epoch": 2.4372957164975055,
-      "grad_norm": 0.0007968803984113038,
-      "learning_rate": 0.00019999707037368395,
-      "loss": 46.0,
-      "step": 31878
-    },
-    {
-      "epoch": 2.4373721734808953,
-      "grad_norm": 0.0016598189249634743,
-      "learning_rate": 0.0001999970701898219,
-      "loss": 46.0,
-      "step": 31879
-    },
-    {
-      "epoch": 2.437448630464285,
-      "grad_norm": 0.0009609736152924597,
-      "learning_rate": 0.0001999970700059541,
-      "loss": 46.0,
-      "step": 31880
-    },
-    {
-      "epoch": 2.437525087447675,
-      "grad_norm": 0.0015193286817520857,
-      "learning_rate": 0.00019999706982208048,
-      "loss": 46.0,
-      "step": 31881
-    },
-    {
-      "epoch": 2.4376015444310646,
-      "grad_norm": 0.0011460300302132964,
-      "learning_rate": 0.00019999706963820112,
-      "loss": 46.0,
-      "step": 31882
-    },
-    {
-      "epoch": 2.4376780014144543,
-      "grad_norm": 0.0017795903841033578,
-      "learning_rate": 0.00019999706945431599,
-      "loss": 46.0,
-      "step": 31883
-    },
-    {
-      "epoch": 2.437754458397844,
-      "grad_norm": 0.0004253470688126981,
-      "learning_rate": 0.00019999706927042508,
-      "loss": 46.0,
-      "step": 31884
-    },
-    {
-      "epoch": 2.437830915381234,
-      "grad_norm": 0.0007202128763310611,
-      "learning_rate": 0.00019999706908652843,
-      "loss": 46.0,
-      "step": 31885
-    },
-    {
-      "epoch": 2.437907372364623,
-      "grad_norm": 0.0009381232084706426,
-      "learning_rate": 0.00019999706890262597,
-      "loss": 46.0,
-      "step": 31886
-    },
-    {
-      "epoch": 2.437983829348013,
-      "grad_norm": 0.0008243191405199468,
-      "learning_rate": 0.00019999706871871777,
-      "loss": 46.0,
-      "step": 31887
-    },
-    {
-      "epoch": 2.4380602863314027,
-      "grad_norm": 0.0009223247179761529,
-      "learning_rate": 0.00019999706853480377,
-      "loss": 46.0,
-      "step": 31888
-    },
-    {
-      "epoch": 2.4381367433147925,
-      "grad_norm": 0.0010832598200067878,
-      "learning_rate": 0.00019999706835088405,
-      "loss": 46.0,
-      "step": 31889
-    },
-    {
-      "epoch": 2.4382132002981822,
-      "grad_norm": 0.0009400818380527198,
-      "learning_rate": 0.0001999970681669585,
-      "loss": 46.0,
-      "step": 31890
-    },
-    {
-      "epoch": 2.438289657281572,
-      "grad_norm": 0.001397757907398045,
-      "learning_rate": 0.00019999706798302724,
-      "loss": 46.0,
-      "step": 31891
-    },
-    {
-      "epoch": 2.4383661142649617,
-      "grad_norm": 0.0007155849016271532,
-      "learning_rate": 0.0001999970677990902,
-      "loss": 46.0,
-      "step": 31892
-    },
-    {
-      "epoch": 2.4384425712483515,
-      "grad_norm": 0.0024621887132525444,
-      "learning_rate": 0.00019999706761514736,
-      "loss": 46.0,
-      "step": 31893
-    },
-    {
-      "epoch": 2.4385190282317413,
-      "grad_norm": 0.002815013052895665,
-      "learning_rate": 0.00019999706743119877,
-      "loss": 46.0,
-      "step": 31894
-    },
-    {
-      "epoch": 2.438595485215131,
-      "grad_norm": 0.0024654981680214405,
-      "learning_rate": 0.0001999970672472444,
-      "loss": 46.0,
-      "step": 31895
-    },
-    {
-      "epoch": 2.4386719421985203,
-      "grad_norm": 0.0006057376158423722,
-      "learning_rate": 0.00019999706706328428,
-      "loss": 46.0,
-      "step": 31896
-    },
-    {
-      "epoch": 2.43874839918191,
-      "grad_norm": 0.000602005748078227,
-      "learning_rate": 0.0001999970668793184,
-      "loss": 46.0,
-      "step": 31897
-    },
-    {
-      "epoch": 2.4388248561653,
-      "grad_norm": 0.0019210334867238998,
-      "learning_rate": 0.0001999970666953467,
-      "loss": 46.0,
-      "step": 31898
-    },
-    {
-      "epoch": 2.4389013131486896,
-      "grad_norm": 0.0010675546946004033,
-      "learning_rate": 0.00019999706651136924,
-      "loss": 46.0,
-      "step": 31899
-    },
-    {
-      "epoch": 2.4389777701320794,
-      "grad_norm": 0.0012831983622163534,
-      "learning_rate": 0.00019999706632738606,
-      "loss": 46.0,
-      "step": 31900
-    },
-    {
-      "epoch": 2.439054227115469,
-      "grad_norm": 0.0009948968654498458,
-      "learning_rate": 0.00019999706614339706,
-      "loss": 46.0,
-      "step": 31901
-    },
-    {
-      "epoch": 2.439130684098859,
-      "grad_norm": 0.0006226995028555393,
-      "learning_rate": 0.00019999706595940232,
-      "loss": 46.0,
-      "step": 31902
-    },
-    {
-      "epoch": 2.4392071410822487,
-      "grad_norm": 0.0006625099340453744,
-      "learning_rate": 0.0001999970657754018,
-      "loss": 46.0,
-      "step": 31903
-    },
-    {
-      "epoch": 2.4392835980656384,
-      "grad_norm": 0.0006312231998890638,
-      "learning_rate": 0.0001999970655913955,
-      "loss": 46.0,
-      "step": 31904
-    },
-    {
-      "epoch": 2.439360055049028,
-      "grad_norm": 0.0005924067227169871,
-      "learning_rate": 0.00019999706540738346,
-      "loss": 46.0,
-      "step": 31905
-    },
-    {
-      "epoch": 2.439436512032418,
-      "grad_norm": 0.0010023250943049788,
-      "learning_rate": 0.00019999706522336562,
-      "loss": 46.0,
-      "step": 31906
-    },
-    {
-      "epoch": 2.4395129690158077,
-      "grad_norm": 0.002129745204001665,
-      "learning_rate": 0.00019999706503934204,
-      "loss": 46.0,
-      "step": 31907
-    },
-    {
-      "epoch": 2.439589425999197,
-      "grad_norm": 0.0006177141331136227,
-      "learning_rate": 0.00019999706485531268,
-      "loss": 46.0,
-      "step": 31908
-    },
-    {
-      "epoch": 2.439665882982587,
-      "grad_norm": 0.0008443251717835665,
-      "learning_rate": 0.00019999706467127755,
-      "loss": 46.0,
-      "step": 31909
-    },
-    {
-      "epoch": 2.4397423399659766,
-      "grad_norm": 0.002235991880297661,
-      "learning_rate": 0.00019999706448723664,
-      "loss": 46.0,
-      "step": 31910
-    },
-    {
-      "epoch": 2.4398187969493663,
-      "grad_norm": 0.0009709433652460575,
-      "learning_rate": 0.00019999706430318996,
-      "loss": 46.0,
-      "step": 31911
-    },
-    {
-      "epoch": 2.439895253932756,
-      "grad_norm": 0.0016714951489120722,
-      "learning_rate": 0.00019999706411913753,
-      "loss": 46.0,
-      "step": 31912
-    },
-    {
-      "epoch": 2.439971710916146,
-      "grad_norm": 0.0006552743725478649,
-      "learning_rate": 0.0001999970639350793,
-      "loss": 46.0,
-      "step": 31913
-    },
-    {
-      "epoch": 2.4400481678995356,
-      "grad_norm": 0.0012795705115422606,
-      "learning_rate": 0.00019999706375101533,
-      "loss": 46.0,
-      "step": 31914
-    },
-    {
-      "epoch": 2.4401246248829254,
-      "grad_norm": 0.0016064875526353717,
-      "learning_rate": 0.00019999706356694561,
-      "loss": 46.0,
-      "step": 31915
-    },
-    {
-      "epoch": 2.440201081866315,
-      "grad_norm": 0.002970742527395487,
-      "learning_rate": 0.00019999706338287007,
-      "loss": 46.0,
-      "step": 31916
-    },
-    {
-      "epoch": 2.4402775388497044,
-      "grad_norm": 0.0008429615991190076,
-      "learning_rate": 0.0001999970631987888,
-      "loss": 46.0,
-      "step": 31917
-    },
-    {
-      "epoch": 2.440353995833094,
-      "grad_norm": 0.014287672936916351,
-      "learning_rate": 0.00019999706301470174,
-      "loss": 46.0,
-      "step": 31918
-    },
-    {
-      "epoch": 2.440430452816484,
-      "grad_norm": 0.0012704402906820178,
-      "learning_rate": 0.0001999970628306089,
-      "loss": 46.0,
-      "step": 31919
-    },
-    {
-      "epoch": 2.4405069097998737,
-      "grad_norm": 0.0016309116035699844,
-      "learning_rate": 0.00019999706264651034,
-      "loss": 46.0,
-      "step": 31920
-    },
-    {
-      "epoch": 2.4405833667832635,
-      "grad_norm": 0.0007852125563658774,
-      "learning_rate": 0.00019999706246240595,
-      "loss": 46.0,
-      "step": 31921
-    },
-    {
-      "epoch": 2.4406598237666532,
-      "grad_norm": 0.0008507322636432946,
-      "learning_rate": 0.0001999970622782958,
-      "loss": 46.0,
-      "step": 31922
-    },
-    {
-      "epoch": 2.440736280750043,
-      "grad_norm": 0.0010519071947783232,
-      "learning_rate": 0.0001999970620941799,
-      "loss": 46.0,
-      "step": 31923
-    },
-    {
-      "epoch": 2.4408127377334328,
-      "grad_norm": 0.0009500946616753936,
-      "learning_rate": 0.00019999706191005823,
-      "loss": 46.0,
-      "step": 31924
-    },
-    {
-      "epoch": 2.4408891947168225,
-      "grad_norm": 0.0014651718083769083,
-      "learning_rate": 0.0001999970617259308,
-      "loss": 46.0,
-      "step": 31925
-    },
-    {
-      "epoch": 2.4409656517002123,
-      "grad_norm": 0.0013047963147982955,
-      "learning_rate": 0.00019999706154179758,
-      "loss": 46.0,
-      "step": 31926
-    },
-    {
-      "epoch": 2.441042108683602,
-      "grad_norm": 0.0008419928490184247,
-      "learning_rate": 0.0001999970613576586,
-      "loss": 46.0,
-      "step": 31927
-    },
-    {
-      "epoch": 2.441118565666992,
-      "grad_norm": 0.0018451045034453273,
-      "learning_rate": 0.00019999706117351384,
-      "loss": 46.0,
-      "step": 31928
-    },
-    {
-      "epoch": 2.4411950226503816,
-      "grad_norm": 0.0006400843267329037,
-      "learning_rate": 0.00019999706098936332,
-      "loss": 46.0,
-      "step": 31929
-    },
-    {
-      "epoch": 2.441271479633771,
-      "grad_norm": 0.0015969921369105577,
-      "learning_rate": 0.00019999706080520705,
-      "loss": 46.0,
-      "step": 31930
-    },
-    {
-      "epoch": 2.4413479366171607,
-      "grad_norm": 0.000968107720836997,
-      "learning_rate": 0.00019999706062104498,
-      "loss": 46.0,
-      "step": 31931
-    },
-    {
-      "epoch": 2.4414243936005504,
-      "grad_norm": 0.0012496842537075281,
-      "learning_rate": 0.00019999706043687717,
-      "loss": 46.0,
-      "step": 31932
-    },
-    {
-      "epoch": 2.44150085058394,
-      "grad_norm": 0.0011838324135169387,
-      "learning_rate": 0.00019999706025270356,
-      "loss": 46.0,
-      "step": 31933
-    },
-    {
-      "epoch": 2.44157730756733,
-      "grad_norm": 0.0006128470413386822,
-      "learning_rate": 0.0001999970600685242,
-      "loss": 46.0,
-      "step": 31934
-    },
-    {
-      "epoch": 2.4416537645507197,
-      "grad_norm": 0.0011169547215104103,
-      "learning_rate": 0.00019999705988433907,
-      "loss": 46.0,
-      "step": 31935
-    },
-    {
-      "epoch": 2.4417302215341095,
-      "grad_norm": 0.0014762532664462924,
-      "learning_rate": 0.00019999705970014816,
-      "loss": 46.0,
-      "step": 31936
-    },
-    {
-      "epoch": 2.441806678517499,
-      "grad_norm": 0.0013938657939434052,
-      "learning_rate": 0.00019999705951595148,
-      "loss": 46.0,
-      "step": 31937
-    },
-    {
-      "epoch": 2.441883135500889,
-      "grad_norm": 0.0011981924762949347,
-      "learning_rate": 0.00019999705933174906,
-      "loss": 46.0,
-      "step": 31938
-    },
-    {
-      "epoch": 2.4419595924842783,
-      "grad_norm": 0.0009762713452801108,
-      "learning_rate": 0.00019999705914754086,
-      "loss": 46.0,
-      "step": 31939
-    },
-    {
-      "epoch": 2.442036049467668,
-      "grad_norm": 0.0007055072346702218,
-      "learning_rate": 0.00019999705896332686,
-      "loss": 46.0,
-      "step": 31940
-    },
-    {
-      "epoch": 2.442112506451058,
-      "grad_norm": 0.00048238105955533683,
-      "learning_rate": 0.00019999705877910711,
-      "loss": 46.0,
-      "step": 31941
-    },
-    {
-      "epoch": 2.4421889634344476,
-      "grad_norm": 0.0010008387034758925,
-      "learning_rate": 0.0001999970585948816,
-      "loss": 46.0,
-      "step": 31942
-    },
-    {
-      "epoch": 2.4422654204178373,
-      "grad_norm": 0.0013541887747123837,
-      "learning_rate": 0.0001999970584106503,
-      "loss": 46.0,
-      "step": 31943
-    },
-    {
-      "epoch": 2.442341877401227,
-      "grad_norm": 0.000608786242082715,
-      "learning_rate": 0.00019999705822641324,
-      "loss": 46.0,
-      "step": 31944
-    },
-    {
-      "epoch": 2.442418334384617,
-      "grad_norm": 0.001000485965050757,
-      "learning_rate": 0.00019999705804217042,
-      "loss": 46.0,
-      "step": 31945
-    },
-    {
-      "epoch": 2.4424947913680066,
-      "grad_norm": 0.0013240780681371689,
-      "learning_rate": 0.0001999970578579218,
-      "loss": 46.0,
-      "step": 31946
-    },
-    {
-      "epoch": 2.4425712483513964,
-      "grad_norm": 0.00046530243707820773,
-      "learning_rate": 0.00019999705767366745,
-      "loss": 46.0,
-      "step": 31947
-    },
-    {
-      "epoch": 2.442647705334786,
-      "grad_norm": 0.0012926728231832385,
-      "learning_rate": 0.00019999705748940732,
-      "loss": 46.0,
-      "step": 31948
-    },
-    {
-      "epoch": 2.442724162318176,
-      "grad_norm": 0.0011619871947914362,
-      "learning_rate": 0.00019999705730514142,
-      "loss": 46.0,
-      "step": 31949
-    },
-    {
-      "epoch": 2.4428006193015657,
-      "grad_norm": 0.0077623589895665646,
-      "learning_rate": 0.00019999705712086976,
-      "loss": 46.0,
-      "step": 31950
-    },
-    {
-      "epoch": 2.442877076284955,
-      "grad_norm": 0.0014879575464874506,
-      "learning_rate": 0.00019999705693659228,
-      "loss": 46.0,
-      "step": 31951
-    },
-    {
-      "epoch": 2.4429535332683447,
-      "grad_norm": 0.0017783952644094825,
-      "learning_rate": 0.00019999705675230909,
-      "loss": 46.0,
-      "step": 31952
-    },
-    {
-      "epoch": 2.4430299902517345,
-      "grad_norm": 0.0004658265970647335,
-      "learning_rate": 0.00019999705656802011,
-      "loss": 46.0,
-      "step": 31953
-    },
-    {
-      "epoch": 2.4431064472351243,
-      "grad_norm": 0.01797652244567871,
-      "learning_rate": 0.00019999705638372537,
-      "loss": 46.0,
-      "step": 31954
-    },
-    {
-      "epoch": 2.443182904218514,
-      "grad_norm": 0.0008855718770064414,
-      "learning_rate": 0.00019999705619942485,
-      "loss": 46.0,
-      "step": 31955
-    },
-    {
-      "epoch": 2.443259361201904,
-      "grad_norm": 0.001002999721094966,
-      "learning_rate": 0.00019999705601511853,
-      "loss": 46.0,
-      "step": 31956
-    },
-    {
-      "epoch": 2.4433358181852936,
-      "grad_norm": 0.0015826683957129717,
-      "learning_rate": 0.0001999970558308065,
-      "loss": 46.0,
-      "step": 31957
-    },
-    {
-      "epoch": 2.4434122751686833,
-      "grad_norm": 0.009470977820456028,
-      "learning_rate": 0.00019999705564648865,
-      "loss": 46.0,
-      "step": 31958
-    },
-    {
-      "epoch": 2.443488732152073,
-      "grad_norm": 0.003207790432497859,
-      "learning_rate": 0.00019999705546216507,
-      "loss": 46.0,
-      "step": 31959
-    },
-    {
-      "epoch": 2.443565189135463,
-      "grad_norm": 0.002510436112061143,
-      "learning_rate": 0.0001999970552778357,
-      "loss": 46.0,
-      "step": 31960
-    },
-    {
-      "epoch": 2.443641646118852,
-      "grad_norm": 0.0008366541005671024,
-      "learning_rate": 0.00019999705509350055,
-      "loss": 46.0,
-      "step": 31961
-    },
-    {
-      "epoch": 2.443718103102242,
-      "grad_norm": 0.001121058245189488,
-      "learning_rate": 0.00019999705490915965,
-      "loss": 46.0,
-      "step": 31962
-    },
-    {
-      "epoch": 2.4437945600856317,
-      "grad_norm": 0.003746144473552704,
-      "learning_rate": 0.000199997054724813,
-      "loss": 46.0,
-      "step": 31963
-    },
-    {
-      "epoch": 2.4438710170690214,
-      "grad_norm": 0.0009716642089188099,
-      "learning_rate": 0.00019999705454046054,
-      "loss": 46.0,
-      "step": 31964
-    },
-    {
-      "epoch": 2.443947474052411,
-      "grad_norm": 0.0008370826835744083,
-      "learning_rate": 0.00019999705435610232,
-      "loss": 46.0,
-      "step": 31965
-    },
-    {
-      "epoch": 2.444023931035801,
-      "grad_norm": 0.001802515471354127,
-      "learning_rate": 0.00019999705417173832,
-      "loss": 46.0,
-      "step": 31966
-    },
-    {
-      "epoch": 2.4441003880191907,
-      "grad_norm": 0.0009477935382165015,
-      "learning_rate": 0.00019999705398736857,
-      "loss": 46.0,
-      "step": 31967
-    },
-    {
-      "epoch": 2.4441768450025805,
-      "grad_norm": 0.00043646173435263336,
-      "learning_rate": 0.00019999705380299308,
-      "loss": 46.0,
-      "step": 31968
-    },
-    {
-      "epoch": 2.4442533019859702,
-      "grad_norm": 0.007541113067418337,
-      "learning_rate": 0.00019999705361861176,
-      "loss": 46.0,
-      "step": 31969
-    },
-    {
-      "epoch": 2.44432975896936,
-      "grad_norm": 0.003198739606887102,
-      "learning_rate": 0.00019999705343422473,
-      "loss": 46.0,
-      "step": 31970
-    },
-    {
-      "epoch": 2.4444062159527498,
-      "grad_norm": 0.0025840254966169596,
-      "learning_rate": 0.0001999970532498319,
-      "loss": 46.0,
-      "step": 31971
-    },
-    {
-      "epoch": 2.4444826729361395,
-      "grad_norm": 0.0018801280530169606,
-      "learning_rate": 0.00019999705306543328,
-      "loss": 46.0,
-      "step": 31972
-    },
-    {
-      "epoch": 2.444559129919529,
-      "grad_norm": 0.015308624133467674,
-      "learning_rate": 0.00019999705288102892,
-      "loss": 46.0,
-      "step": 31973
-    },
-    {
-      "epoch": 2.4446355869029186,
-      "grad_norm": 0.0012719205114990473,
-      "learning_rate": 0.0001999970526966188,
-      "loss": 46.0,
-      "step": 31974
-    },
-    {
-      "epoch": 2.4447120438863084,
-      "grad_norm": 0.0006378758698701859,
-      "learning_rate": 0.00019999705251220288,
-      "loss": 46.0,
-      "step": 31975
-    },
-    {
-      "epoch": 2.444788500869698,
-      "grad_norm": 0.0011293942807242274,
-      "learning_rate": 0.0001999970523277812,
-      "loss": 46.0,
-      "step": 31976
-    },
-    {
-      "epoch": 2.444864957853088,
-      "grad_norm": 0.002601534128189087,
-      "learning_rate": 0.00019999705214335375,
-      "loss": 46.0,
-      "step": 31977
-    },
-    {
-      "epoch": 2.4449414148364776,
-      "grad_norm": 0.00045958292321301997,
-      "learning_rate": 0.00019999705195892053,
-      "loss": 46.0,
-      "step": 31978
-    },
-    {
-      "epoch": 2.4450178718198674,
-      "grad_norm": 0.0007823935593478382,
-      "learning_rate": 0.00019999705177448158,
-      "loss": 46.0,
-      "step": 31979
-    },
-    {
-      "epoch": 2.445094328803257,
-      "grad_norm": 0.0014765918022021651,
-      "learning_rate": 0.0001999970515900368,
-      "loss": 46.0,
-      "step": 31980
-    },
-    {
-      "epoch": 2.445170785786647,
-      "grad_norm": 0.0008068651659414172,
-      "learning_rate": 0.0001999970514055863,
-      "loss": 46.0,
-      "step": 31981
-    },
-    {
-      "epoch": 2.4452472427700367,
-      "grad_norm": 0.0008751056739129126,
-      "learning_rate": 0.00019999705122113,
-      "loss": 46.0,
-      "step": 31982
-    },
-    {
-      "epoch": 2.445323699753426,
-      "grad_norm": 0.00212300568819046,
-      "learning_rate": 0.00019999705103666794,
-      "loss": 46.0,
-      "step": 31983
-    },
-    {
-      "epoch": 2.4454001567368158,
-      "grad_norm": 0.0007084000390022993,
-      "learning_rate": 0.00019999705085220013,
-      "loss": 46.0,
-      "step": 31984
-    },
-    {
-      "epoch": 2.4454766137202055,
-      "grad_norm": 0.001924376585520804,
-      "learning_rate": 0.00019999705066772651,
-      "loss": 46.0,
-      "step": 31985
-    },
-    {
-      "epoch": 2.4455530707035953,
-      "grad_norm": 0.0019496403401717544,
-      "learning_rate": 0.00019999705048324716,
-      "loss": 46.0,
-      "step": 31986
-    },
-    {
-      "epoch": 2.445629527686985,
-      "grad_norm": 0.001012491062283516,
-      "learning_rate": 0.00019999705029876203,
-      "loss": 46.0,
-      "step": 31987
-    },
-    {
-      "epoch": 2.445705984670375,
-      "grad_norm": 0.0011541333515197039,
-      "learning_rate": 0.0001999970501142711,
-      "loss": 46.0,
-      "step": 31988
-    },
-    {
-      "epoch": 2.4457824416537646,
-      "grad_norm": 0.0011622386518865824,
-      "learning_rate": 0.00019999704992977444,
-      "loss": 46.0,
-      "step": 31989
-    },
-    {
-      "epoch": 2.4458588986371543,
-      "grad_norm": 0.00046954042045399547,
-      "learning_rate": 0.000199997049745272,
-      "loss": 46.0,
-      "step": 31990
-    },
-    {
-      "epoch": 2.445935355620544,
-      "grad_norm": 0.0008914373465813696,
-      "learning_rate": 0.00019999704956076377,
-      "loss": 46.0,
-      "step": 31991
-    },
-    {
-      "epoch": 2.446011812603934,
-      "grad_norm": 0.0016433949349448085,
-      "learning_rate": 0.0001999970493762498,
-      "loss": 46.0,
-      "step": 31992
-    },
-    {
-      "epoch": 2.4460882695873236,
-      "grad_norm": 0.0010150804882869124,
-      "learning_rate": 0.00019999704919173002,
-      "loss": 46.0,
-      "step": 31993
-    },
-    {
-      "epoch": 2.4461647265707134,
-      "grad_norm": 0.001598947448655963,
-      "learning_rate": 0.0001999970490072045,
-      "loss": 46.0,
-      "step": 31994
-    },
-    {
-      "epoch": 2.4462411835541027,
-      "grad_norm": 0.001388934557326138,
-      "learning_rate": 0.00019999704882267322,
-      "loss": 46.0,
-      "step": 31995
-    },
-    {
-      "epoch": 2.4463176405374925,
-      "grad_norm": 0.0008281638729386032,
-      "learning_rate": 0.00019999704863813615,
-      "loss": 46.0,
-      "step": 31996
-    },
-    {
-      "epoch": 2.446394097520882,
-      "grad_norm": 0.0008215660927817225,
-      "learning_rate": 0.00019999704845359334,
-      "loss": 46.0,
-      "step": 31997
-    },
-    {
-      "epoch": 2.446470554504272,
-      "grad_norm": 0.009521810337901115,
-      "learning_rate": 0.00019999704826904473,
-      "loss": 46.0,
-      "step": 31998
-    },
-    {
-      "epoch": 2.4465470114876617,
-      "grad_norm": 0.0006725895800627768,
-      "learning_rate": 0.00019999704808449034,
-      "loss": 46.0,
-      "step": 31999
-    },
-    {
-      "epoch": 2.4466234684710515,
-      "grad_norm": 0.001011132262647152,
-      "learning_rate": 0.00019999704789993024,
-      "loss": 46.0,
-      "step": 32000
-    },
-    {
-      "epoch": 2.4466999254544413,
-      "grad_norm": 0.0009333459311164916,
-      "learning_rate": 0.0001999970477153643,
-      "loss": 46.0,
-      "step": 32001
-    },
-    {
-      "epoch": 2.446776382437831,
-      "grad_norm": 0.0007062758086249232,
-      "learning_rate": 0.00019999704753079263,
-      "loss": 46.0,
-      "step": 32002
-    },
-    {
-      "epoch": 2.446852839421221,
-      "grad_norm": 0.0009965809294953942,
-      "learning_rate": 0.0001999970473462152,
-      "loss": 46.0,
-      "step": 32003
-    },
-    {
-      "epoch": 2.4469292964046105,
-      "grad_norm": 0.0012880159774795175,
-      "learning_rate": 0.00019999704716163196,
-      "loss": 46.0,
-      "step": 32004
-    },
-    {
-      "epoch": 2.447005753388,
-      "grad_norm": 0.0006319022504612803,
-      "learning_rate": 0.000199997046977043,
-      "loss": 46.0,
-      "step": 32005
-    },
-    {
-      "epoch": 2.4470822103713896,
-      "grad_norm": 0.0012123904889449477,
-      "learning_rate": 0.00019999704679244824,
-      "loss": 46.0,
-      "step": 32006
-    },
-    {
-      "epoch": 2.4471586673547794,
-      "grad_norm": 0.0016980203799903393,
-      "learning_rate": 0.0001999970466078477,
-      "loss": 46.0,
-      "step": 32007
-    },
-    {
-      "epoch": 2.447235124338169,
-      "grad_norm": 0.0016341886948794127,
-      "learning_rate": 0.0001999970464232414,
-      "loss": 46.0,
-      "step": 32008
-    },
-    {
-      "epoch": 2.447311581321559,
-      "grad_norm": 0.0005806373665109277,
-      "learning_rate": 0.00019999704623862937,
-      "loss": 46.0,
-      "step": 32009
-    },
-    {
-      "epoch": 2.4473880383049487,
-      "grad_norm": 0.0018561262404546142,
-      "learning_rate": 0.00019999704605401153,
-      "loss": 46.0,
-      "step": 32010
-    },
-    {
-      "epoch": 2.4474644952883384,
-      "grad_norm": 0.001861725701019168,
-      "learning_rate": 0.00019999704586938792,
-      "loss": 46.0,
-      "step": 32011
-    },
-    {
-      "epoch": 2.447540952271728,
-      "grad_norm": 0.0013776309788227081,
-      "learning_rate": 0.00019999704568475856,
-      "loss": 46.0,
-      "step": 32012
-    },
-    {
-      "epoch": 2.447617409255118,
-      "grad_norm": 0.012885675765573978,
-      "learning_rate": 0.00019999704550012343,
-      "loss": 46.0,
-      "step": 32013
-    },
-    {
-      "epoch": 2.4476938662385077,
-      "grad_norm": 0.0007430001278407872,
-      "learning_rate": 0.0001999970453154825,
-      "loss": 46.0,
-      "step": 32014
-    },
-    {
-      "epoch": 2.4477703232218975,
-      "grad_norm": 0.0009364989236928523,
-      "learning_rate": 0.00019999704513083583,
-      "loss": 46.0,
-      "step": 32015
-    },
-    {
-      "epoch": 2.4478467802052872,
-      "grad_norm": 0.00456291763111949,
-      "learning_rate": 0.00019999704494618337,
-      "loss": 46.0,
-      "step": 32016
-    },
-    {
-      "epoch": 2.4479232371886765,
-      "grad_norm": 0.0014080549590289593,
-      "learning_rate": 0.00019999704476152518,
-      "loss": 46.0,
-      "step": 32017
-    },
-    {
-      "epoch": 2.4479996941720663,
-      "grad_norm": 0.0025584816467016935,
-      "learning_rate": 0.00019999704457686118,
-      "loss": 46.0,
-      "step": 32018
-    },
-    {
-      "epoch": 2.448076151155456,
-      "grad_norm": 0.002959862118586898,
-      "learning_rate": 0.00019999704439219144,
-      "loss": 46.0,
-      "step": 32019
-    },
-    {
-      "epoch": 2.448152608138846,
-      "grad_norm": 0.0004960898659192026,
-      "learning_rate": 0.0001999970442075159,
-      "loss": 46.0,
-      "step": 32020
-    },
-    {
-      "epoch": 2.4482290651222356,
-      "grad_norm": 0.004734597634524107,
-      "learning_rate": 0.0001999970440228346,
-      "loss": 46.0,
-      "step": 32021
-    },
-    {
-      "epoch": 2.4483055221056254,
-      "grad_norm": 0.0010306901531293988,
-      "learning_rate": 0.00019999704383814754,
-      "loss": 46.0,
-      "step": 32022
-    },
-    {
-      "epoch": 2.448381979089015,
-      "grad_norm": 0.00044556776992976665,
-      "learning_rate": 0.00019999704365345473,
-      "loss": 46.0,
-      "step": 32023
-    },
-    {
-      "epoch": 2.448458436072405,
-      "grad_norm": 0.0014852783642709255,
-      "learning_rate": 0.00019999704346875614,
-      "loss": 46.0,
-      "step": 32024
-    },
-    {
-      "epoch": 2.4485348930557946,
-      "grad_norm": 0.001416479586623609,
-      "learning_rate": 0.00019999704328405176,
-      "loss": 46.0,
-      "step": 32025
-    },
-    {
-      "epoch": 2.4486113500391844,
-      "grad_norm": 0.003507385728880763,
-      "learning_rate": 0.00019999704309934163,
-      "loss": 46.0,
-      "step": 32026
-    },
-    {
-      "epoch": 2.4486878070225737,
-      "grad_norm": 0.00270731165073812,
-      "learning_rate": 0.0001999970429146257,
-      "loss": 46.0,
-      "step": 32027
-    },
-    {
-      "epoch": 2.4487642640059635,
-      "grad_norm": 0.0009261043160222471,
-      "learning_rate": 0.00019999704272990402,
-      "loss": 46.0,
-      "step": 32028
-    },
-    {
-      "epoch": 2.4488407209893532,
-      "grad_norm": 0.0010700608836486936,
-      "learning_rate": 0.00019999704254517657,
-      "loss": 46.0,
-      "step": 32029
-    },
-    {
-      "epoch": 2.448917177972743,
-      "grad_norm": 0.001064253505319357,
-      "learning_rate": 0.00019999704236044338,
-      "loss": 46.0,
-      "step": 32030
-    },
-    {
-      "epoch": 2.4489936349561328,
-      "grad_norm": 0.00033997936407104135,
-      "learning_rate": 0.00019999704217570438,
-      "loss": 46.0,
-      "step": 32031
-    },
-    {
-      "epoch": 2.4490700919395225,
-      "grad_norm": 0.001571160857565701,
-      "learning_rate": 0.00019999704199095964,
-      "loss": 46.0,
-      "step": 32032
-    },
-    {
-      "epoch": 2.4491465489229123,
-      "grad_norm": 0.0011864836560562253,
-      "learning_rate": 0.00019999704180620912,
-      "loss": 46.0,
-      "step": 32033
-    },
-    {
-      "epoch": 2.449223005906302,
-      "grad_norm": 0.0008409509900957346,
-      "learning_rate": 0.0001999970416214528,
-      "loss": 46.0,
-      "step": 32034
-    },
-    {
-      "epoch": 2.449299462889692,
-      "grad_norm": 0.000771131890360266,
-      "learning_rate": 0.00019999704143669074,
-      "loss": 46.0,
-      "step": 32035
-    },
-    {
-      "epoch": 2.4493759198730816,
-      "grad_norm": 0.004988211207091808,
-      "learning_rate": 0.0001999970412519229,
-      "loss": 46.0,
-      "step": 32036
-    },
-    {
-      "epoch": 2.4494523768564713,
-      "grad_norm": 0.0006533483974635601,
-      "learning_rate": 0.00019999704106714932,
-      "loss": 46.0,
-      "step": 32037
-    },
-    {
-      "epoch": 2.449528833839861,
-      "grad_norm": 0.0023822709918022156,
-      "learning_rate": 0.00019999704088236993,
-      "loss": 46.0,
-      "step": 32038
-    },
-    {
-      "epoch": 2.4496052908232504,
-      "grad_norm": 0.0010483047226443887,
-      "learning_rate": 0.0001999970406975848,
-      "loss": 46.0,
-      "step": 32039
-    },
-    {
-      "epoch": 2.44968174780664,
-      "grad_norm": 0.0023576875682920218,
-      "learning_rate": 0.0001999970405127939,
-      "loss": 46.0,
-      "step": 32040
-    },
-    {
-      "epoch": 2.44975820479003,
-      "grad_norm": 0.0004939362406730652,
-      "learning_rate": 0.00019999704032799722,
-      "loss": 46.0,
-      "step": 32041
-    },
-    {
-      "epoch": 2.4498346617734197,
-      "grad_norm": 0.0010357382707297802,
-      "learning_rate": 0.00019999704014319475,
-      "loss": 46.0,
-      "step": 32042
-    },
-    {
-      "epoch": 2.4499111187568094,
-      "grad_norm": 0.0008340787026099861,
-      "learning_rate": 0.00019999703995838655,
-      "loss": 46.0,
-      "step": 32043
-    },
-    {
-      "epoch": 2.449987575740199,
-      "grad_norm": 0.0011341251665726304,
-      "learning_rate": 0.00019999703977357255,
-      "loss": 46.0,
-      "step": 32044
-    },
-    {
-      "epoch": 2.450064032723589,
-      "grad_norm": 0.0005242135957814753,
-      "learning_rate": 0.0001999970395887528,
-      "loss": 46.0,
-      "step": 32045
-    },
-    {
-      "epoch": 2.4501404897069787,
-      "grad_norm": 0.003730755066499114,
-      "learning_rate": 0.0001999970394039273,
-      "loss": 46.0,
-      "step": 32046
-    },
-    {
-      "epoch": 2.4502169466903685,
-      "grad_norm": 0.0014987685717642307,
-      "learning_rate": 0.000199997039219096,
-      "loss": 46.0,
-      "step": 32047
-    },
-    {
-      "epoch": 2.450293403673758,
-      "grad_norm": 0.006126203574240208,
-      "learning_rate": 0.00019999703903425891,
-      "loss": 46.0,
-      "step": 32048
-    },
-    {
-      "epoch": 2.4503698606571476,
-      "grad_norm": 0.0012287484714761376,
-      "learning_rate": 0.0001999970388494161,
-      "loss": 46.0,
-      "step": 32049
-    },
-    {
-      "epoch": 2.4504463176405373,
-      "grad_norm": 0.002660485915839672,
-      "learning_rate": 0.0001999970386645675,
-      "loss": 46.0,
-      "step": 32050
-    },
-    {
-      "epoch": 2.450522774623927,
-      "grad_norm": 0.0015183178475126624,
-      "learning_rate": 0.0001999970384797131,
-      "loss": 46.0,
-      "step": 32051
-    },
-    {
-      "epoch": 2.450599231607317,
-      "grad_norm": 0.0007381152245216072,
-      "learning_rate": 0.00019999703829485296,
-      "loss": 46.0,
-      "step": 32052
-    },
-    {
-      "epoch": 2.4506756885907066,
-      "grad_norm": 0.0008989167399704456,
-      "learning_rate": 0.00019999703810998708,
-      "loss": 46.0,
-      "step": 32053
-    },
-    {
-      "epoch": 2.4507521455740964,
-      "grad_norm": 0.002915708813816309,
-      "learning_rate": 0.0001999970379251154,
-      "loss": 46.0,
-      "step": 32054
-    },
-    {
-      "epoch": 2.450828602557486,
-      "grad_norm": 0.0012920403387397528,
-      "learning_rate": 0.00019999703774023795,
-      "loss": 46.0,
-      "step": 32055
-    },
-    {
-      "epoch": 2.450905059540876,
-      "grad_norm": 0.0013374312547966838,
-      "learning_rate": 0.00019999703755535473,
-      "loss": 46.0,
-      "step": 32056
-    },
-    {
-      "epoch": 2.4509815165242657,
-      "grad_norm": 0.0007244454463943839,
-      "learning_rate": 0.00019999703737046573,
-      "loss": 46.0,
-      "step": 32057
-    },
-    {
-      "epoch": 2.4510579735076554,
-      "grad_norm": 0.00081466423580423,
-      "learning_rate": 0.00019999703718557096,
-      "loss": 46.0,
-      "step": 32058
-    },
-    {
-      "epoch": 2.451134430491045,
-      "grad_norm": 0.0006333947530947626,
-      "learning_rate": 0.00019999703700067045,
-      "loss": 46.0,
-      "step": 32059
-    },
-    {
-      "epoch": 2.451210887474435,
-      "grad_norm": 0.0007041457574814558,
-      "learning_rate": 0.00019999703681576416,
-      "loss": 46.0,
-      "step": 32060
-    },
-    {
-      "epoch": 2.4512873444578243,
-      "grad_norm": 0.0004056443285662681,
-      "learning_rate": 0.0001999970366308521,
-      "loss": 46.0,
-      "step": 32061
-    },
-    {
-      "epoch": 2.451363801441214,
-      "grad_norm": 0.0014375589089468122,
-      "learning_rate": 0.00019999703644593426,
-      "loss": 46.0,
-      "step": 32062
-    },
-    {
-      "epoch": 2.451440258424604,
-      "grad_norm": 0.0017896929057314992,
-      "learning_rate": 0.00019999703626101068,
-      "loss": 46.0,
-      "step": 32063
-    },
-    {
-      "epoch": 2.4515167154079935,
-      "grad_norm": 0.004768447019159794,
-      "learning_rate": 0.0001999970360760813,
-      "loss": 46.0,
-      "step": 32064
-    },
-    {
-      "epoch": 2.4515931723913833,
-      "grad_norm": 0.0019432223634794354,
-      "learning_rate": 0.00019999703589114614,
-      "loss": 46.0,
-      "step": 32065
-    },
-    {
-      "epoch": 2.451669629374773,
-      "grad_norm": 0.0008068630704656243,
-      "learning_rate": 0.00019999703570620523,
-      "loss": 46.0,
-      "step": 32066
-    },
-    {
-      "epoch": 2.451746086358163,
-      "grad_norm": 0.0009804224828258157,
-      "learning_rate": 0.00019999703552125856,
-      "loss": 46.0,
-      "step": 32067
-    },
-    {
-      "epoch": 2.4518225433415526,
-      "grad_norm": 0.0006549140089191496,
-      "learning_rate": 0.00019999703533630608,
-      "loss": 46.0,
-      "step": 32068
-    },
-    {
-      "epoch": 2.4518990003249423,
-      "grad_norm": 0.0006274455226957798,
-      "learning_rate": 0.00019999703515134789,
-      "loss": 46.0,
-      "step": 32069
-    },
-    {
-      "epoch": 2.4519754573083317,
-      "grad_norm": 0.0029337084852159023,
-      "learning_rate": 0.00019999703496638392,
-      "loss": 46.0,
-      "step": 32070
-    },
-    {
-      "epoch": 2.4520519142917214,
-      "grad_norm": 0.0013975371839478612,
-      "learning_rate": 0.00019999703478141415,
-      "loss": 46.0,
-      "step": 32071
-    },
-    {
-      "epoch": 2.452128371275111,
-      "grad_norm": 0.001947292941622436,
-      "learning_rate": 0.0001999970345964386,
-      "loss": 46.0,
-      "step": 32072
-    },
-    {
-      "epoch": 2.452204828258501,
-      "grad_norm": 0.001765129272826016,
-      "learning_rate": 0.00019999703441145732,
-      "loss": 46.0,
-      "step": 32073
-    },
-    {
-      "epoch": 2.4522812852418907,
-      "grad_norm": 0.0006401055143214762,
-      "learning_rate": 0.00019999703422647028,
-      "loss": 46.0,
-      "step": 32074
-    },
-    {
-      "epoch": 2.4523577422252805,
-      "grad_norm": 0.001285024918615818,
-      "learning_rate": 0.00019999703404147742,
-      "loss": 46.0,
-      "step": 32075
-    },
-    {
-      "epoch": 2.4524341992086702,
-      "grad_norm": 0.001259715761989355,
-      "learning_rate": 0.0001999970338564788,
-      "loss": 46.0,
-      "step": 32076
-    },
-    {
-      "epoch": 2.45251065619206,
-      "grad_norm": 0.0008648103685118258,
-      "learning_rate": 0.00019999703367147445,
-      "loss": 46.0,
-      "step": 32077
-    },
-    {
-      "epoch": 2.4525871131754498,
-      "grad_norm": 0.0014680068707093596,
-      "learning_rate": 0.0001999970334864643,
-      "loss": 46.0,
-      "step": 32078
-    },
-    {
-      "epoch": 2.4526635701588395,
-      "grad_norm": 0.00038193483487702906,
-      "learning_rate": 0.0001999970333014484,
-      "loss": 46.0,
-      "step": 32079
-    },
-    {
-      "epoch": 2.4527400271422293,
-      "grad_norm": 0.0006948375958018005,
-      "learning_rate": 0.0001999970331164267,
-      "loss": 46.0,
-      "step": 32080
-    },
-    {
-      "epoch": 2.452816484125619,
-      "grad_norm": 0.0009676854824647307,
-      "learning_rate": 0.00019999703293139927,
-      "loss": 46.0,
-      "step": 32081
-    },
-    {
-      "epoch": 2.4528929411090084,
-      "grad_norm": 0.0006656266050413251,
-      "learning_rate": 0.00019999703274636605,
-      "loss": 46.0,
-      "step": 32082
-    },
-    {
-      "epoch": 2.452969398092398,
-      "grad_norm": 0.004371342249214649,
-      "learning_rate": 0.00019999703256132705,
-      "loss": 46.0,
-      "step": 32083
-    },
-    {
-      "epoch": 2.453045855075788,
-      "grad_norm": 0.0012889942154288292,
-      "learning_rate": 0.0001999970323762823,
-      "loss": 46.0,
-      "step": 32084
-    },
-    {
-      "epoch": 2.4531223120591776,
-      "grad_norm": 0.0029644868336617947,
-      "learning_rate": 0.00019999703219123177,
-      "loss": 46.0,
-      "step": 32085
-    },
-    {
-      "epoch": 2.4531987690425674,
-      "grad_norm": 0.0009314757189713418,
-      "learning_rate": 0.00019999703200617548,
-      "loss": 46.0,
-      "step": 32086
-    },
-    {
-      "epoch": 2.453275226025957,
-      "grad_norm": 0.0008678769809193909,
-      "learning_rate": 0.00019999703182111341,
-      "loss": 46.0,
-      "step": 32087
-    },
-    {
-      "epoch": 2.453351683009347,
-      "grad_norm": 0.0007079067872837186,
-      "learning_rate": 0.00019999703163604558,
-      "loss": 46.0,
-      "step": 32088
-    },
-    {
-      "epoch": 2.4534281399927367,
-      "grad_norm": 0.0004546970303636044,
-      "learning_rate": 0.00019999703145097197,
-      "loss": 46.0,
-      "step": 32089
-    },
-    {
-      "epoch": 2.4535045969761264,
-      "grad_norm": 0.0015264780959114432,
-      "learning_rate": 0.0001999970312658926,
-      "loss": 46.0,
-      "step": 32090
-    },
-    {
-      "epoch": 2.453581053959516,
-      "grad_norm": 0.0009667367557995021,
-      "learning_rate": 0.00019999703108080746,
-      "loss": 46.0,
-      "step": 32091
-    },
-    {
-      "epoch": 2.4536575109429055,
-      "grad_norm": 0.0011149076744914055,
-      "learning_rate": 0.00019999703089571656,
-      "loss": 46.0,
-      "step": 32092
-    },
-    {
-      "epoch": 2.4537339679262953,
-      "grad_norm": 0.0019979476928710938,
-      "learning_rate": 0.00019999703071061985,
-      "loss": 46.0,
-      "step": 32093
-    },
-    {
-      "epoch": 2.453810424909685,
-      "grad_norm": 0.0008110285270959139,
-      "learning_rate": 0.0001999970305255174,
-      "loss": 46.0,
-      "step": 32094
-    },
-    {
-      "epoch": 2.453886881893075,
-      "grad_norm": 0.0009752092300914228,
-      "learning_rate": 0.00019999703034040918,
-      "loss": 46.0,
-      "step": 32095
-    },
-    {
-      "epoch": 2.4539633388764646,
-      "grad_norm": 0.0010237341048195958,
-      "learning_rate": 0.00019999703015529522,
-      "loss": 46.0,
-      "step": 32096
-    },
-    {
-      "epoch": 2.4540397958598543,
-      "grad_norm": 0.0045866151340305805,
-      "learning_rate": 0.00019999702997017545,
-      "loss": 46.0,
-      "step": 32097
-    },
-    {
-      "epoch": 2.454116252843244,
-      "grad_norm": 0.0014634841354563832,
-      "learning_rate": 0.00019999702978504993,
-      "loss": 46.0,
-      "step": 32098
-    },
-    {
-      "epoch": 2.454192709826634,
-      "grad_norm": 0.0004835967265535146,
-      "learning_rate": 0.00019999702959991862,
-      "loss": 46.0,
-      "step": 32099
-    },
-    {
-      "epoch": 2.4542691668100236,
-      "grad_norm": 0.0005293546710163355,
-      "learning_rate": 0.00019999702941478155,
-      "loss": 46.0,
-      "step": 32100
-    },
-    {
-      "epoch": 2.4543456237934134,
-      "grad_norm": 0.00044838787289336324,
-      "learning_rate": 0.00019999702922963872,
-      "loss": 46.0,
-      "step": 32101
-    },
-    {
-      "epoch": 2.454422080776803,
-      "grad_norm": 0.0010153127368539572,
-      "learning_rate": 0.0001999970290444901,
-      "loss": 46.0,
-      "step": 32102
-    },
-    {
-      "epoch": 2.454498537760193,
-      "grad_norm": 0.0007873608265072107,
-      "learning_rate": 0.00019999702885933575,
-      "loss": 46.0,
-      "step": 32103
-    },
-    {
-      "epoch": 2.454574994743582,
-      "grad_norm": 0.002770470455288887,
-      "learning_rate": 0.0001999970286741756,
-      "loss": 46.0,
-      "step": 32104
-    },
-    {
-      "epoch": 2.454651451726972,
-      "grad_norm": 0.00040425301995128393,
-      "learning_rate": 0.00019999702848900967,
-      "loss": 46.0,
-      "step": 32105
-    },
-    {
-      "epoch": 2.4547279087103617,
-      "grad_norm": 0.0010005495278164744,
-      "learning_rate": 0.000199997028303838,
-      "loss": 46.0,
-      "step": 32106
-    },
-    {
-      "epoch": 2.4548043656937515,
-      "grad_norm": 0.0009729396551847458,
-      "learning_rate": 0.00019999702811866055,
-      "loss": 46.0,
-      "step": 32107
-    },
-    {
-      "epoch": 2.4548808226771412,
-      "grad_norm": 0.0011714320862665772,
-      "learning_rate": 0.00019999702793347735,
-      "loss": 46.0,
-      "step": 32108
-    },
-    {
-      "epoch": 2.454957279660531,
-      "grad_norm": 0.0004402663034852594,
-      "learning_rate": 0.00019999702774828836,
-      "loss": 46.0,
-      "step": 32109
-    },
-    {
-      "epoch": 2.4550337366439208,
-      "grad_norm": 0.0009628739207983017,
-      "learning_rate": 0.0001999970275630936,
-      "loss": 46.0,
-      "step": 32110
-    },
-    {
-      "epoch": 2.4551101936273105,
-      "grad_norm": 0.0004306513292249292,
-      "learning_rate": 0.00019999702737789305,
-      "loss": 46.0,
-      "step": 32111
-    },
-    {
-      "epoch": 2.4551866506107003,
-      "grad_norm": 0.0023378280457109213,
-      "learning_rate": 0.00019999702719268676,
-      "loss": 46.0,
-      "step": 32112
-    },
-    {
-      "epoch": 2.45526310759409,
-      "grad_norm": 0.0012850608909502625,
-      "learning_rate": 0.0001999970270074747,
-      "loss": 46.0,
-      "step": 32113
-    },
-    {
-      "epoch": 2.4553395645774794,
-      "grad_norm": 0.001222437946125865,
-      "learning_rate": 0.00019999702682225686,
-      "loss": 46.0,
-      "step": 32114
-    },
-    {
-      "epoch": 2.455416021560869,
-      "grad_norm": 0.0012058448046445847,
-      "learning_rate": 0.00019999702663703323,
-      "loss": 46.0,
-      "step": 32115
-    },
-    {
-      "epoch": 2.455492478544259,
-      "grad_norm": 0.001976849278435111,
-      "learning_rate": 0.00019999702645180387,
-      "loss": 46.0,
-      "step": 32116
-    },
-    {
-      "epoch": 2.4555689355276487,
-      "grad_norm": 0.0006149174296297133,
-      "learning_rate": 0.00019999702626656874,
-      "loss": 46.0,
-      "step": 32117
-    },
-    {
-      "epoch": 2.4556453925110384,
-      "grad_norm": 0.00037595003959722817,
-      "learning_rate": 0.00019999702608132781,
-      "loss": 46.0,
-      "step": 32118
-    },
-    {
-      "epoch": 2.455721849494428,
-      "grad_norm": 0.0008315707091242075,
-      "learning_rate": 0.00019999702589608114,
-      "loss": 46.0,
-      "step": 32119
-    },
-    {
-      "epoch": 2.455798306477818,
-      "grad_norm": 0.0006714237970300019,
-      "learning_rate": 0.00019999702571082866,
-      "loss": 46.0,
-      "step": 32120
-    },
-    {
-      "epoch": 2.4558747634612077,
-      "grad_norm": 0.001386646879836917,
-      "learning_rate": 0.00019999702552557044,
-      "loss": 46.0,
-      "step": 32121
-    },
-    {
-      "epoch": 2.4559512204445975,
-      "grad_norm": 0.0036568334326148033,
-      "learning_rate": 0.00019999702534030648,
-      "loss": 46.0,
-      "step": 32122
-    },
-    {
-      "epoch": 2.456027677427987,
-      "grad_norm": 0.00029667068156413734,
-      "learning_rate": 0.0001999970251550367,
-      "loss": 46.0,
-      "step": 32123
-    },
-    {
-      "epoch": 2.456104134411377,
-      "grad_norm": 0.0011132522486150265,
-      "learning_rate": 0.0001999970249697612,
-      "loss": 46.0,
-      "step": 32124
-    },
-    {
-      "epoch": 2.4561805913947667,
-      "grad_norm": 0.0009052717359736562,
-      "learning_rate": 0.00019999702478447988,
-      "loss": 46.0,
-      "step": 32125
-    },
-    {
-      "epoch": 2.456257048378156,
-      "grad_norm": 0.0037274565547704697,
-      "learning_rate": 0.00019999702459919282,
-      "loss": 46.0,
-      "step": 32126
-    },
-    {
-      "epoch": 2.456333505361546,
-      "grad_norm": 0.0019299163250252604,
-      "learning_rate": 0.00019999702441389998,
-      "loss": 46.0,
-      "step": 32127
-    },
-    {
-      "epoch": 2.4564099623449356,
-      "grad_norm": 0.0007135948399081826,
-      "learning_rate": 0.00019999702422860137,
-      "loss": 46.0,
-      "step": 32128
-    },
-    {
-      "epoch": 2.4564864193283253,
-      "grad_norm": 0.0008471620385535061,
-      "learning_rate": 0.000199997024043297,
-      "loss": 46.0,
-      "step": 32129
-    },
-    {
-      "epoch": 2.456562876311715,
-      "grad_norm": 0.0017200012225657701,
-      "learning_rate": 0.00019999702385798686,
-      "loss": 46.0,
-      "step": 32130
-    },
-    {
-      "epoch": 2.456639333295105,
-      "grad_norm": 0.010277921333909035,
-      "learning_rate": 0.00019999702367267094,
-      "loss": 46.0,
-      "step": 32131
-    },
-    {
-      "epoch": 2.4567157902784946,
-      "grad_norm": 0.0011906875297427177,
-      "learning_rate": 0.00019999702348734926,
-      "loss": 46.0,
-      "step": 32132
-    },
-    {
-      "epoch": 2.4567922472618844,
-      "grad_norm": 0.0069778901524841785,
-      "learning_rate": 0.00019999702330202181,
-      "loss": 46.0,
-      "step": 32133
-    },
-    {
-      "epoch": 2.456868704245274,
-      "grad_norm": 0.001418792293407023,
-      "learning_rate": 0.00019999702311668857,
-      "loss": 46.0,
-      "step": 32134
-    },
-    {
-      "epoch": 2.456945161228664,
-      "grad_norm": 0.002820760477334261,
-      "learning_rate": 0.0001999970229313496,
-      "loss": 46.0,
-      "step": 32135
-    },
-    {
-      "epoch": 2.4570216182120532,
-      "grad_norm": 0.0032284879125654697,
-      "learning_rate": 0.00019999702274600483,
-      "loss": 46.0,
-      "step": 32136
-    },
-    {
-      "epoch": 2.457098075195443,
-      "grad_norm": 0.0012881056172773242,
-      "learning_rate": 0.00019999702256065432,
-      "loss": 46.0,
-      "step": 32137
-    },
-    {
-      "epoch": 2.4571745321788327,
-      "grad_norm": 0.0013070888817310333,
-      "learning_rate": 0.000199997022375298,
-      "loss": 46.0,
-      "step": 32138
-    },
-    {
-      "epoch": 2.4572509891622225,
-      "grad_norm": 0.0015307943103834987,
-      "learning_rate": 0.0001999970221899359,
-      "loss": 46.0,
-      "step": 32139
-    },
-    {
-      "epoch": 2.4573274461456123,
-      "grad_norm": 0.000834343780297786,
-      "learning_rate": 0.00019999702200456808,
-      "loss": 46.0,
-      "step": 32140
-    },
-    {
-      "epoch": 2.457403903129002,
-      "grad_norm": 0.0012775223003700376,
-      "learning_rate": 0.0001999970218191945,
-      "loss": 46.0,
-      "step": 32141
-    },
-    {
-      "epoch": 2.457480360112392,
-      "grad_norm": 0.0007632126798853278,
-      "learning_rate": 0.00019999702163381512,
-      "loss": 46.0,
-      "step": 32142
-    },
-    {
-      "epoch": 2.4575568170957816,
-      "grad_norm": 0.003969459794461727,
-      "learning_rate": 0.00019999702144842996,
-      "loss": 46.0,
-      "step": 32143
-    },
-    {
-      "epoch": 2.4576332740791713,
-      "grad_norm": 0.0008371869334951043,
-      "learning_rate": 0.00019999702126303906,
-      "loss": 46.0,
-      "step": 32144
-    },
-    {
-      "epoch": 2.457709731062561,
-      "grad_norm": 0.0017724273493513465,
-      "learning_rate": 0.0001999970210776424,
-      "loss": 46.0,
-      "step": 32145
-    },
-    {
-      "epoch": 2.457786188045951,
-      "grad_norm": 0.001332386047579348,
-      "learning_rate": 0.00019999702089223994,
-      "loss": 46.0,
-      "step": 32146
-    },
-    {
-      "epoch": 2.4578626450293406,
-      "grad_norm": 0.0004408008244354278,
-      "learning_rate": 0.00019999702070683172,
-      "loss": 46.0,
-      "step": 32147
-    },
-    {
-      "epoch": 2.45793910201273,
-      "grad_norm": 0.0024061615113168955,
-      "learning_rate": 0.0001999970205214177,
-      "loss": 46.0,
-      "step": 32148
-    },
-    {
-      "epoch": 2.4580155589961197,
-      "grad_norm": 0.0027121463790535927,
-      "learning_rate": 0.00019999702033599793,
-      "loss": 46.0,
-      "step": 32149
-    },
-    {
-      "epoch": 2.4580920159795094,
-      "grad_norm": 0.004833335522562265,
-      "learning_rate": 0.0001999970201505724,
-      "loss": 46.0,
-      "step": 32150
-    },
-    {
-      "epoch": 2.458168472962899,
-      "grad_norm": 0.0013423613272607327,
-      "learning_rate": 0.0001999970199651411,
-      "loss": 46.0,
-      "step": 32151
-    },
-    {
-      "epoch": 2.458244929946289,
-      "grad_norm": 0.0016987406415864825,
-      "learning_rate": 0.00019999701977970404,
-      "loss": 46.0,
-      "step": 32152
-    },
-    {
-      "epoch": 2.4583213869296787,
-      "grad_norm": 0.0005409973673522472,
-      "learning_rate": 0.0001999970195942612,
-      "loss": 46.0,
-      "step": 32153
-    },
-    {
-      "epoch": 2.4583978439130685,
-      "grad_norm": 0.0005198657163418829,
-      "learning_rate": 0.0001999970194088126,
-      "loss": 46.0,
-      "step": 32154
-    },
-    {
-      "epoch": 2.4584743008964582,
-      "grad_norm": 0.0006227836711332202,
-      "learning_rate": 0.00019999701922335822,
-      "loss": 46.0,
-      "step": 32155
-    },
-    {
-      "epoch": 2.458550757879848,
-      "grad_norm": 0.004347793757915497,
-      "learning_rate": 0.00019999701903789806,
-      "loss": 46.0,
-      "step": 32156
-    },
-    {
-      "epoch": 2.4586272148632378,
-      "grad_norm": 0.0011652481043711305,
-      "learning_rate": 0.00019999701885243216,
-      "loss": 46.0,
-      "step": 32157
-    },
-    {
-      "epoch": 2.458703671846627,
-      "grad_norm": 0.0013047493994235992,
-      "learning_rate": 0.00019999701866696046,
-      "loss": 46.0,
-      "step": 32158
-    },
-    {
-      "epoch": 2.458780128830017,
-      "grad_norm": 0.0013005641521885991,
-      "learning_rate": 0.00019999701848148302,
-      "loss": 46.0,
-      "step": 32159
-    },
-    {
-      "epoch": 2.4588565858134066,
-      "grad_norm": 0.0007573722396045923,
-      "learning_rate": 0.0001999970182959998,
-      "loss": 46.0,
-      "step": 32160
-    },
-    {
-      "epoch": 2.4589330427967964,
-      "grad_norm": 0.0008397951605729759,
-      "learning_rate": 0.0001999970181105108,
-      "loss": 46.0,
-      "step": 32161
-    },
-    {
-      "epoch": 2.459009499780186,
-      "grad_norm": 0.0004184510326012969,
-      "learning_rate": 0.00019999701792501603,
-      "loss": 46.0,
-      "step": 32162
-    },
-    {
-      "epoch": 2.459085956763576,
-      "grad_norm": 0.0016234355280175805,
-      "learning_rate": 0.00019999701773951552,
-      "loss": 46.0,
-      "step": 32163
-    },
-    {
-      "epoch": 2.4591624137469656,
-      "grad_norm": 0.0015076295239850879,
-      "learning_rate": 0.0001999970175540092,
-      "loss": 46.0,
-      "step": 32164
-    },
-    {
-      "epoch": 2.4592388707303554,
-      "grad_norm": 0.0006178086623549461,
-      "learning_rate": 0.00019999701736849715,
-      "loss": 46.0,
-      "step": 32165
-    },
-    {
-      "epoch": 2.459315327713745,
-      "grad_norm": 0.001316750654950738,
-      "learning_rate": 0.00019999701718297929,
-      "loss": 46.0,
-      "step": 32166
-    },
-    {
-      "epoch": 2.459391784697135,
-      "grad_norm": 0.0028804135508835316,
-      "learning_rate": 0.0001999970169974557,
-      "loss": 46.0,
-      "step": 32167
-    },
-    {
-      "epoch": 2.4594682416805247,
-      "grad_norm": 0.0008769866544753313,
-      "learning_rate": 0.00019999701681192633,
-      "loss": 46.0,
-      "step": 32168
-    },
-    {
-      "epoch": 2.4595446986639145,
-      "grad_norm": 0.0012958385050296783,
-      "learning_rate": 0.00019999701662639117,
-      "loss": 46.0,
-      "step": 32169
-    },
-    {
-      "epoch": 2.4596211556473038,
-      "grad_norm": 0.0011356809409335256,
-      "learning_rate": 0.00019999701644085024,
-      "loss": 46.0,
-      "step": 32170
-    },
-    {
-      "epoch": 2.4596976126306935,
-      "grad_norm": 0.00117954658344388,
-      "learning_rate": 0.00019999701625530357,
-      "loss": 46.0,
-      "step": 32171
-    },
-    {
-      "epoch": 2.4597740696140833,
-      "grad_norm": 0.007854013703763485,
-      "learning_rate": 0.00019999701606975112,
-      "loss": 46.0,
-      "step": 32172
-    },
-    {
-      "epoch": 2.459850526597473,
-      "grad_norm": 0.0016878879396244884,
-      "learning_rate": 0.00019999701588419288,
-      "loss": 46.0,
-      "step": 32173
-    },
-    {
-      "epoch": 2.459926983580863,
-      "grad_norm": 0.001979435095563531,
-      "learning_rate": 0.0001999970156986289,
-      "loss": 46.0,
-      "step": 32174
-    },
-    {
-      "epoch": 2.4600034405642526,
-      "grad_norm": 0.002631532261148095,
-      "learning_rate": 0.00019999701551305914,
-      "loss": 46.0,
-      "step": 32175
-    },
-    {
-      "epoch": 2.4600798975476423,
-      "grad_norm": 0.003942139912396669,
-      "learning_rate": 0.0001999970153274836,
-      "loss": 46.0,
-      "step": 32176
-    },
-    {
-      "epoch": 2.460156354531032,
-      "grad_norm": 0.0004847529053222388,
-      "learning_rate": 0.0001999970151419023,
-      "loss": 46.0,
-      "step": 32177
-    },
-    {
-      "epoch": 2.460232811514422,
-      "grad_norm": 0.0006172974244691432,
-      "learning_rate": 0.00019999701495631526,
-      "loss": 46.0,
-      "step": 32178
-    },
-    {
-      "epoch": 2.460309268497811,
-      "grad_norm": 0.0007586546125821769,
-      "learning_rate": 0.0001999970147707224,
-      "loss": 46.0,
-      "step": 32179
-    },
-    {
-      "epoch": 2.460385725481201,
-      "grad_norm": 0.0016714805969968438,
-      "learning_rate": 0.0001999970145851238,
-      "loss": 46.0,
-      "step": 32180
-    },
-    {
-      "epoch": 2.4604621824645907,
-      "grad_norm": 0.0025886481162160635,
-      "learning_rate": 0.00019999701439951938,
-      "loss": 46.0,
-      "step": 32181
-    },
-    {
-      "epoch": 2.4605386394479805,
-      "grad_norm": 0.0012383094290271401,
-      "learning_rate": 0.00019999701421390925,
-      "loss": 46.0,
-      "step": 32182
-    },
-    {
-      "epoch": 2.46061509643137,
-      "grad_norm": 0.0012315765488892794,
-      "learning_rate": 0.00019999701402829333,
-      "loss": 46.0,
-      "step": 32183
-    },
-    {
-      "epoch": 2.46069155341476,
-      "grad_norm": 0.0004620544204954058,
-      "learning_rate": 0.00019999701384267165,
-      "loss": 46.0,
-      "step": 32184
-    },
-    {
-      "epoch": 2.4607680103981497,
-      "grad_norm": 0.0013298518024384975,
-      "learning_rate": 0.0001999970136570442,
-      "loss": 46.0,
-      "step": 32185
-    },
-    {
-      "epoch": 2.4608444673815395,
-      "grad_norm": 0.002070858608931303,
-      "learning_rate": 0.000199997013471411,
-      "loss": 46.0,
-      "step": 32186
-    },
-    {
-      "epoch": 2.4609209243649293,
-      "grad_norm": 0.0007270520436577499,
-      "learning_rate": 0.000199997013285772,
-      "loss": 46.0,
-      "step": 32187
-    },
-    {
-      "epoch": 2.460997381348319,
-      "grad_norm": 0.0015808738535270095,
-      "learning_rate": 0.00019999701310012723,
-      "loss": 46.0,
-      "step": 32188
-    },
-    {
-      "epoch": 2.461073838331709,
-      "grad_norm": 0.000589851348195225,
-      "learning_rate": 0.0001999970129144767,
-      "loss": 46.0,
-      "step": 32189
-    },
-    {
-      "epoch": 2.4611502953150985,
-      "grad_norm": 0.0008861060487106442,
-      "learning_rate": 0.00019999701272882037,
-      "loss": 46.0,
-      "step": 32190
-    },
-    {
-      "epoch": 2.4612267522984883,
-      "grad_norm": 0.0009151807171292603,
-      "learning_rate": 0.00019999701254315831,
-      "loss": 46.0,
-      "step": 32191
-    },
-    {
-      "epoch": 2.4613032092818776,
-      "grad_norm": 0.0012511750683188438,
-      "learning_rate": 0.00019999701235749048,
-      "loss": 46.0,
-      "step": 32192
-    },
-    {
-      "epoch": 2.4613796662652674,
-      "grad_norm": 0.0006788653554394841,
-      "learning_rate": 0.00019999701217181688,
-      "loss": 46.0,
-      "step": 32193
-    },
-    {
-      "epoch": 2.461456123248657,
-      "grad_norm": 0.0006716499337926507,
-      "learning_rate": 0.0001999970119861375,
-      "loss": 46.0,
-      "step": 32194
-    },
-    {
-      "epoch": 2.461532580232047,
-      "grad_norm": 0.0012442311272025108,
-      "learning_rate": 0.00019999701180045234,
-      "loss": 46.0,
-      "step": 32195
-    },
-    {
-      "epoch": 2.4616090372154367,
-      "grad_norm": 0.0007843896746635437,
-      "learning_rate": 0.00019999701161476142,
-      "loss": 46.0,
-      "step": 32196
-    },
-    {
-      "epoch": 2.4616854941988264,
-      "grad_norm": 0.0032779022585600615,
-      "learning_rate": 0.00019999701142906474,
-      "loss": 46.0,
-      "step": 32197
-    },
-    {
-      "epoch": 2.461761951182216,
-      "grad_norm": 0.0005988853517919779,
-      "learning_rate": 0.00019999701124336227,
-      "loss": 46.0,
-      "step": 32198
-    },
-    {
-      "epoch": 2.461838408165606,
-      "grad_norm": 0.0013059888733550906,
-      "learning_rate": 0.00019999701105765405,
-      "loss": 46.0,
-      "step": 32199
-    },
-    {
-      "epoch": 2.4619148651489957,
-      "grad_norm": 0.001087350188754499,
-      "learning_rate": 0.00019999701087194006,
-      "loss": 46.0,
-      "step": 32200
-    },
-    {
-      "epoch": 2.461991322132385,
-      "grad_norm": 0.0027107377536594868,
-      "learning_rate": 0.0001999970106862203,
-      "loss": 46.0,
-      "step": 32201
-    },
-    {
-      "epoch": 2.462067779115775,
-      "grad_norm": 0.0014383131638169289,
-      "learning_rate": 0.00019999701050049478,
-      "loss": 46.0,
-      "step": 32202
-    },
-    {
-      "epoch": 2.4621442360991646,
-      "grad_norm": 0.0004719188145827502,
-      "learning_rate": 0.00019999701031476346,
-      "loss": 46.0,
-      "step": 32203
-    },
-    {
-      "epoch": 2.4622206930825543,
-      "grad_norm": 0.0006269064615480602,
-      "learning_rate": 0.00019999701012902638,
-      "loss": 46.0,
-      "step": 32204
-    },
-    {
-      "epoch": 2.462297150065944,
-      "grad_norm": 0.0017841834342107177,
-      "learning_rate": 0.00019999700994328354,
-      "loss": 46.0,
-      "step": 32205
-    },
-    {
-      "epoch": 2.462373607049334,
-      "grad_norm": 0.000889175571501255,
-      "learning_rate": 0.00019999700975753494,
-      "loss": 46.0,
-      "step": 32206
-    },
-    {
-      "epoch": 2.4624500640327236,
-      "grad_norm": 0.0005687976372428238,
-      "learning_rate": 0.00019999700957178056,
-      "loss": 46.0,
-      "step": 32207
-    },
-    {
-      "epoch": 2.4625265210161134,
-      "grad_norm": 0.0019288725452497602,
-      "learning_rate": 0.0001999970093860204,
-      "loss": 46.0,
-      "step": 32208
-    },
-    {
-      "epoch": 2.462602977999503,
-      "grad_norm": 0.0008934037177823484,
-      "learning_rate": 0.0001999970092002545,
-      "loss": 46.0,
-      "step": 32209
-    },
-    {
-      "epoch": 2.462679434982893,
-      "grad_norm": 0.0010685424786061049,
-      "learning_rate": 0.0001999970090144828,
-      "loss": 46.0,
-      "step": 32210
-    },
-    {
-      "epoch": 2.4627558919662826,
-      "grad_norm": 0.0014368095435202122,
-      "learning_rate": 0.00019999700882870536,
-      "loss": 46.0,
-      "step": 32211
-    },
-    {
-      "epoch": 2.4628323489496724,
-      "grad_norm": 0.0029335259459912777,
-      "learning_rate": 0.00019999700864292212,
-      "loss": 46.0,
-      "step": 32212
-    },
-    {
-      "epoch": 2.4629088059330617,
-      "grad_norm": 0.0017619087593629956,
-      "learning_rate": 0.00019999700845713312,
-      "loss": 46.0,
-      "step": 32213
-    },
-    {
-      "epoch": 2.4629852629164515,
-      "grad_norm": 0.0010946551337838173,
-      "learning_rate": 0.00019999700827133836,
-      "loss": 46.0,
-      "step": 32214
-    },
-    {
-      "epoch": 2.4630617198998412,
-      "grad_norm": 0.002092518378049135,
-      "learning_rate": 0.00019999700808553782,
-      "loss": 46.0,
-      "step": 32215
-    },
-    {
-      "epoch": 2.463138176883231,
-      "grad_norm": 0.0003836005926132202,
-      "learning_rate": 0.00019999700789973153,
-      "loss": 46.0,
-      "step": 32216
-    },
-    {
-      "epoch": 2.4632146338666208,
-      "grad_norm": 0.0011468147858977318,
-      "learning_rate": 0.00019999700771391944,
-      "loss": 46.0,
-      "step": 32217
-    },
-    {
-      "epoch": 2.4632910908500105,
-      "grad_norm": 0.0011520790867507458,
-      "learning_rate": 0.00019999700752810159,
-      "loss": 46.0,
-      "step": 32218
-    },
-    {
-      "epoch": 2.4633675478334003,
-      "grad_norm": 0.0026980803813785315,
-      "learning_rate": 0.000199997007342278,
-      "loss": 46.0,
-      "step": 32219
-    },
-    {
-      "epoch": 2.46344400481679,
-      "grad_norm": 0.002163398778066039,
-      "learning_rate": 0.00019999700715644863,
-      "loss": 46.0,
-      "step": 32220
-    },
-    {
-      "epoch": 2.46352046180018,
-      "grad_norm": 0.002358921803534031,
-      "learning_rate": 0.00019999700697061347,
-      "loss": 46.0,
-      "step": 32221
-    },
-    {
-      "epoch": 2.4635969187835696,
-      "grad_norm": 0.0031822770833969116,
-      "learning_rate": 0.00019999700678477255,
-      "loss": 46.0,
-      "step": 32222
-    },
-    {
-      "epoch": 2.463673375766959,
-      "grad_norm": 0.0005781924701295793,
-      "learning_rate": 0.00019999700659892585,
-      "loss": 46.0,
-      "step": 32223
-    },
-    {
-      "epoch": 2.4637498327503486,
-      "grad_norm": 0.001497967285104096,
-      "learning_rate": 0.0001999970064130734,
-      "loss": 46.0,
-      "step": 32224
-    },
-    {
-      "epoch": 2.4638262897337384,
-      "grad_norm": 0.0009857508121058345,
-      "learning_rate": 0.00019999700622721518,
-      "loss": 46.0,
-      "step": 32225
-    },
-    {
-      "epoch": 2.463902746717128,
-      "grad_norm": 0.0014936871593818069,
-      "learning_rate": 0.0001999970060413512,
-      "loss": 46.0,
-      "step": 32226
-    },
-    {
-      "epoch": 2.463979203700518,
-      "grad_norm": 0.0029334549326449633,
-      "learning_rate": 0.00019999700585548143,
-      "loss": 46.0,
-      "step": 32227
-    },
-    {
-      "epoch": 2.4640556606839077,
-      "grad_norm": 0.0009575043222866952,
-      "learning_rate": 0.0001999970056696059,
-      "loss": 46.0,
-      "step": 32228
-    },
-    {
-      "epoch": 2.4641321176672975,
-      "grad_norm": 0.0009826376335695386,
-      "learning_rate": 0.00019999700548372457,
-      "loss": 46.0,
-      "step": 32229
-    },
-    {
-      "epoch": 2.464208574650687,
-      "grad_norm": 0.0008980318089015782,
-      "learning_rate": 0.00019999700529783752,
-      "loss": 46.0,
-      "step": 32230
-    },
-    {
-      "epoch": 2.464285031634077,
-      "grad_norm": 0.0011363662779331207,
-      "learning_rate": 0.00019999700511194468,
-      "loss": 46.0,
-      "step": 32231
-    },
-    {
-      "epoch": 2.4643614886174667,
-      "grad_norm": 0.000897760153748095,
-      "learning_rate": 0.00019999700492604605,
-      "loss": 46.0,
-      "step": 32232
-    },
-    {
-      "epoch": 2.4644379456008565,
-      "grad_norm": 0.001164128421805799,
-      "learning_rate": 0.0001999970047401417,
-      "loss": 46.0,
-      "step": 32233
-    },
-    {
-      "epoch": 2.4645144025842463,
-      "grad_norm": 0.0026480199303478003,
-      "learning_rate": 0.00019999700455423152,
-      "loss": 46.0,
-      "step": 32234
-    },
-    {
-      "epoch": 2.4645908595676356,
-      "grad_norm": 0.0019794886466115713,
-      "learning_rate": 0.00019999700436831562,
-      "loss": 46.0,
-      "step": 32235
-    },
-    {
-      "epoch": 2.4646673165510253,
-      "grad_norm": 0.001332507119514048,
-      "learning_rate": 0.00019999700418239392,
-      "loss": 46.0,
-      "step": 32236
-    },
-    {
-      "epoch": 2.464743773534415,
-      "grad_norm": 0.001517377095296979,
-      "learning_rate": 0.00019999700399646645,
-      "loss": 46.0,
-      "step": 32237
-    },
-    {
-      "epoch": 2.464820230517805,
-      "grad_norm": 0.0037037094589322805,
-      "learning_rate": 0.00019999700381053323,
-      "loss": 46.0,
-      "step": 32238
-    },
-    {
-      "epoch": 2.4648966875011946,
-      "grad_norm": 0.0012010071659460664,
-      "learning_rate": 0.00019999700362459424,
-      "loss": 46.0,
-      "step": 32239
-    },
-    {
-      "epoch": 2.4649731444845844,
-      "grad_norm": 0.005148399155586958,
-      "learning_rate": 0.00019999700343864947,
-      "loss": 46.0,
-      "step": 32240
-    },
-    {
-      "epoch": 2.465049601467974,
-      "grad_norm": 0.0004465868987608701,
-      "learning_rate": 0.00019999700325269894,
-      "loss": 46.0,
-      "step": 32241
-    },
-    {
-      "epoch": 2.465126058451364,
-      "grad_norm": 0.0009997423039749265,
-      "learning_rate": 0.00019999700306674262,
-      "loss": 46.0,
-      "step": 32242
-    },
-    {
-      "epoch": 2.4652025154347537,
-      "grad_norm": 0.001208460540510714,
-      "learning_rate": 0.00019999700288078056,
-      "loss": 46.0,
-      "step": 32243
-    },
-    {
-      "epoch": 2.4652789724181434,
-      "grad_norm": 0.0018073919927701354,
-      "learning_rate": 0.0001999970026948127,
-      "loss": 46.0,
-      "step": 32244
-    },
-    {
-      "epoch": 2.4653554294015327,
-      "grad_norm": 0.0026139733381569386,
-      "learning_rate": 0.0001999970025088391,
-      "loss": 46.0,
-      "step": 32245
-    },
-    {
-      "epoch": 2.4654318863849225,
-      "grad_norm": 0.0010302651207894087,
-      "learning_rate": 0.00019999700232285972,
-      "loss": 46.0,
-      "step": 32246
-    },
-    {
-      "epoch": 2.4655083433683123,
-      "grad_norm": 0.0005748834810219705,
-      "learning_rate": 0.00019999700213687457,
-      "loss": 46.0,
-      "step": 32247
-    },
-    {
-      "epoch": 2.465584800351702,
-      "grad_norm": 0.0008782865479588509,
-      "learning_rate": 0.00019999700195088367,
-      "loss": 46.0,
-      "step": 32248
-    },
-    {
-      "epoch": 2.465661257335092,
-      "grad_norm": 0.0010622115805745125,
-      "learning_rate": 0.00019999700176488697,
-      "loss": 46.0,
-      "step": 32249
-    },
-    {
-      "epoch": 2.4657377143184815,
-      "grad_norm": 0.0006868946366012096,
-      "learning_rate": 0.0001999970015788845,
-      "loss": 46.0,
-      "step": 32250
-    },
-    {
-      "epoch": 2.4658141713018713,
-      "grad_norm": 0.0005327576072886586,
-      "learning_rate": 0.00019999700139287628,
-      "loss": 46.0,
-      "step": 32251
-    },
-    {
-      "epoch": 2.465890628285261,
-      "grad_norm": 0.0008218061411753297,
-      "learning_rate": 0.0001999970012068623,
-      "loss": 46.0,
-      "step": 32252
-    },
-    {
-      "epoch": 2.465967085268651,
-      "grad_norm": 0.0008427886059507728,
-      "learning_rate": 0.0001999970010208425,
-      "loss": 46.0,
-      "step": 32253
-    },
-    {
-      "epoch": 2.4660435422520406,
-      "grad_norm": 0.0016122242668643594,
-      "learning_rate": 0.00019999700083481696,
-      "loss": 46.0,
-      "step": 32254
-    },
-    {
-      "epoch": 2.4661199992354303,
-      "grad_norm": 0.0029113860800862312,
-      "learning_rate": 0.00019999700064878568,
-      "loss": 46.0,
-      "step": 32255
-    },
-    {
-      "epoch": 2.46619645621882,
-      "grad_norm": 0.0021183029748499393,
-      "learning_rate": 0.0001999970004627486,
-      "loss": 46.0,
-      "step": 32256
-    },
-    {
-      "epoch": 2.4662729132022094,
-      "grad_norm": 0.0018390774494037032,
-      "learning_rate": 0.00019999700027670576,
-      "loss": 46.0,
-      "step": 32257
-    },
-    {
-      "epoch": 2.466349370185599,
-      "grad_norm": 0.002160990610718727,
-      "learning_rate": 0.00019999700009065713,
-      "loss": 46.0,
-      "step": 32258
-    },
-    {
-      "epoch": 2.466425827168989,
-      "grad_norm": 0.0007847305969335139,
-      "learning_rate": 0.00019999699990460275,
-      "loss": 46.0,
-      "step": 32259
-    },
-    {
-      "epoch": 2.4665022841523787,
-      "grad_norm": 0.0011799189960584044,
-      "learning_rate": 0.0001999969997185426,
-      "loss": 46.0,
-      "step": 32260
-    },
-    {
-      "epoch": 2.4665787411357685,
-      "grad_norm": 0.0003658886707853526,
-      "learning_rate": 0.00019999699953247667,
-      "loss": 46.0,
-      "step": 32261
-    },
-    {
-      "epoch": 2.4666551981191582,
-      "grad_norm": 0.000489134166855365,
-      "learning_rate": 0.000199996999346405,
-      "loss": 46.0,
-      "step": 32262
-    },
-    {
-      "epoch": 2.466731655102548,
-      "grad_norm": 0.004200296476483345,
-      "learning_rate": 0.00019999699916032753,
-      "loss": 46.0,
-      "step": 32263
-    },
-    {
-      "epoch": 2.4668081120859378,
-      "grad_norm": 0.0016868250677362084,
-      "learning_rate": 0.0001999969989742443,
-      "loss": 46.0,
-      "step": 32264
-    },
-    {
-      "epoch": 2.4668845690693275,
-      "grad_norm": 0.0009139256435446441,
-      "learning_rate": 0.00019999699878815532,
-      "loss": 46.0,
-      "step": 32265
-    },
-    {
-      "epoch": 2.4669610260527173,
-      "grad_norm": 0.0009249647846445441,
-      "learning_rate": 0.00019999699860206053,
-      "loss": 46.0,
-      "step": 32266
-    },
-    {
-      "epoch": 2.4670374830361066,
-      "grad_norm": 0.0003955393913201988,
-      "learning_rate": 0.00019999699841596002,
-      "loss": 46.0,
-      "step": 32267
-    },
-    {
-      "epoch": 2.4671139400194964,
-      "grad_norm": 0.0006288831355050206,
-      "learning_rate": 0.0001999969982298537,
-      "loss": 46.0,
-      "step": 32268
-    },
-    {
-      "epoch": 2.467190397002886,
-      "grad_norm": 0.00339855276979506,
-      "learning_rate": 0.00019999699804374165,
-      "loss": 46.0,
-      "step": 32269
-    },
-    {
-      "epoch": 2.467266853986276,
-      "grad_norm": 0.0021211400162428617,
-      "learning_rate": 0.0001999969978576238,
-      "loss": 46.0,
-      "step": 32270
-    },
-    {
-      "epoch": 2.4673433109696656,
-      "grad_norm": 0.00104645942337811,
-      "learning_rate": 0.00019999699767150018,
-      "loss": 46.0,
-      "step": 32271
-    },
-    {
-      "epoch": 2.4674197679530554,
-      "grad_norm": 0.0008200169540941715,
-      "learning_rate": 0.00019999699748537078,
-      "loss": 46.0,
-      "step": 32272
-    },
-    {
-      "epoch": 2.467496224936445,
-      "grad_norm": 0.0027899937704205513,
-      "learning_rate": 0.00019999699729923563,
-      "loss": 46.0,
-      "step": 32273
-    },
-    {
-      "epoch": 2.467572681919835,
-      "grad_norm": 0.003018836723640561,
-      "learning_rate": 0.00019999699711309473,
-      "loss": 46.0,
-      "step": 32274
-    },
-    {
-      "epoch": 2.4676491389032247,
-      "grad_norm": 0.002071422291919589,
-      "learning_rate": 0.00019999699692694803,
-      "loss": 46.0,
-      "step": 32275
-    },
-    {
-      "epoch": 2.4677255958866144,
-      "grad_norm": 0.0008218807051889598,
-      "learning_rate": 0.00019999699674079556,
-      "loss": 46.0,
-      "step": 32276
-    },
-    {
-      "epoch": 2.467802052870004,
-      "grad_norm": 0.0008420305675826967,
-      "learning_rate": 0.00019999699655463731,
-      "loss": 46.0,
-      "step": 32277
-    },
-    {
-      "epoch": 2.467878509853394,
-      "grad_norm": 0.002019083360210061,
-      "learning_rate": 0.00019999699636847335,
-      "loss": 46.0,
-      "step": 32278
-    },
-    {
-      "epoch": 2.4679549668367833,
-      "grad_norm": 0.0021862878929823637,
-      "learning_rate": 0.00019999699618230356,
-      "loss": 46.0,
-      "step": 32279
-    },
-    {
-      "epoch": 2.468031423820173,
-      "grad_norm": 0.004516915418207645,
-      "learning_rate": 0.00019999699599612805,
-      "loss": 46.0,
-      "step": 32280
-    },
-    {
-      "epoch": 2.468107880803563,
-      "grad_norm": 0.0007316782139241695,
-      "learning_rate": 0.00019999699580994674,
-      "loss": 46.0,
-      "step": 32281
-    },
-    {
-      "epoch": 2.4681843377869526,
-      "grad_norm": 0.0011412501335144043,
-      "learning_rate": 0.00019999699562375965,
-      "loss": 46.0,
-      "step": 32282
-    },
-    {
-      "epoch": 2.4682607947703423,
-      "grad_norm": 0.0019098496995866299,
-      "learning_rate": 0.0001999969954375668,
-      "loss": 46.0,
-      "step": 32283
-    },
-    {
-      "epoch": 2.468337251753732,
-      "grad_norm": 0.0008229966042563319,
-      "learning_rate": 0.0001999969952513682,
-      "loss": 46.0,
-      "step": 32284
-    },
-    {
-      "epoch": 2.468413708737122,
-      "grad_norm": 0.0008402031962759793,
-      "learning_rate": 0.0001999969950651638,
-      "loss": 46.0,
-      "step": 32285
-    },
-    {
-      "epoch": 2.4684901657205116,
-      "grad_norm": 0.001198038924485445,
-      "learning_rate": 0.00019999699487895363,
-      "loss": 46.0,
-      "step": 32286
-    },
-    {
-      "epoch": 2.4685666227039014,
-      "grad_norm": 0.002581742126494646,
-      "learning_rate": 0.00019999699469273774,
-      "loss": 46.0,
-      "step": 32287
-    },
-    {
-      "epoch": 2.468643079687291,
-      "grad_norm": 0.0008640863816253841,
-      "learning_rate": 0.00019999699450651607,
-      "loss": 46.0,
-      "step": 32288
-    },
-    {
-      "epoch": 2.4687195366706804,
-      "grad_norm": 0.0006658991915173829,
-      "learning_rate": 0.00019999699432028857,
-      "loss": 46.0,
-      "step": 32289
-    },
-    {
-      "epoch": 2.46879599365407,
-      "grad_norm": 0.0011746136005967855,
-      "learning_rate": 0.00019999699413405535,
-      "loss": 46.0,
-      "step": 32290
-    },
-    {
-      "epoch": 2.46887245063746,
-      "grad_norm": 0.0010299982968717813,
-      "learning_rate": 0.00019999699394781636,
-      "loss": 46.0,
-      "step": 32291
-    },
-    {
-      "epoch": 2.4689489076208497,
-      "grad_norm": 0.0011975272791460156,
-      "learning_rate": 0.0001999969937615716,
-      "loss": 46.0,
-      "step": 32292
-    },
-    {
-      "epoch": 2.4690253646042395,
-      "grad_norm": 0.0013336162082850933,
-      "learning_rate": 0.00019999699357532106,
-      "loss": 46.0,
-      "step": 32293
-    },
-    {
-      "epoch": 2.4691018215876293,
-      "grad_norm": 0.0005860582459717989,
-      "learning_rate": 0.00019999699338906475,
-      "loss": 46.0,
-      "step": 32294
-    },
-    {
-      "epoch": 2.469178278571019,
-      "grad_norm": 0.0009031447698362172,
-      "learning_rate": 0.00019999699320280266,
-      "loss": 46.0,
-      "step": 32295
-    },
-    {
-      "epoch": 2.4692547355544088,
-      "grad_norm": 0.0010756520787253976,
-      "learning_rate": 0.0001999969930165348,
-      "loss": 46.0,
-      "step": 32296
-    },
-    {
-      "epoch": 2.4693311925377985,
-      "grad_norm": 0.0005827349959872663,
-      "learning_rate": 0.0001999969928302612,
-      "loss": 46.0,
-      "step": 32297
-    },
-    {
-      "epoch": 2.4694076495211883,
-      "grad_norm": 0.0012503070756793022,
-      "learning_rate": 0.00019999699264398182,
-      "loss": 46.0,
-      "step": 32298
-    },
-    {
-      "epoch": 2.469484106504578,
-      "grad_norm": 0.0012438365956768394,
-      "learning_rate": 0.00019999699245769667,
-      "loss": 46.0,
-      "step": 32299
-    },
-    {
-      "epoch": 2.469560563487968,
-      "grad_norm": 0.0010974142933264375,
-      "learning_rate": 0.00019999699227140575,
-      "loss": 46.0,
-      "step": 32300
-    },
-    {
-      "epoch": 2.469637020471357,
-      "grad_norm": 0.0007490361458621919,
-      "learning_rate": 0.00019999699208510908,
-      "loss": 46.0,
-      "step": 32301
-    },
-    {
-      "epoch": 2.469713477454747,
-      "grad_norm": 0.0009076794958673418,
-      "learning_rate": 0.00019999699189880658,
-      "loss": 46.0,
-      "step": 32302
-    },
-    {
-      "epoch": 2.4697899344381367,
-      "grad_norm": 0.004136491566896439,
-      "learning_rate": 0.00019999699171249836,
-      "loss": 46.0,
-      "step": 32303
-    },
-    {
-      "epoch": 2.4698663914215264,
-      "grad_norm": 0.0006584388902410865,
-      "learning_rate": 0.00019999699152618437,
-      "loss": 46.0,
-      "step": 32304
-    },
-    {
-      "epoch": 2.469942848404916,
-      "grad_norm": 0.005605303682386875,
-      "learning_rate": 0.00019999699133986458,
-      "loss": 46.0,
-      "step": 32305
-    },
-    {
-      "epoch": 2.470019305388306,
-      "grad_norm": 0.0013109392020851374,
-      "learning_rate": 0.00019999699115353905,
-      "loss": 46.0,
-      "step": 32306
-    },
-    {
-      "epoch": 2.4700957623716957,
-      "grad_norm": 0.0028625130653381348,
-      "learning_rate": 0.00019999699096720773,
-      "loss": 46.0,
-      "step": 32307
-    },
-    {
-      "epoch": 2.4701722193550855,
-      "grad_norm": 0.0007773717516101897,
-      "learning_rate": 0.00019999699078087065,
-      "loss": 46.0,
-      "step": 32308
-    },
-    {
-      "epoch": 2.4702486763384752,
-      "grad_norm": 0.0008473945199511945,
-      "learning_rate": 0.00019999699059452782,
-      "loss": 46.0,
-      "step": 32309
-    },
-    {
-      "epoch": 2.4703251333218645,
-      "grad_norm": 0.0017666032072156668,
-      "learning_rate": 0.00019999699040817922,
-      "loss": 46.0,
-      "step": 32310
-    },
-    {
-      "epoch": 2.4704015903052543,
-      "grad_norm": 0.0006012325175106525,
-      "learning_rate": 0.00019999699022182484,
-      "loss": 46.0,
-      "step": 32311
-    },
-    {
-      "epoch": 2.470478047288644,
-      "grad_norm": 0.0017275470308959484,
-      "learning_rate": 0.00019999699003546466,
-      "loss": 46.0,
-      "step": 32312
-    },
-    {
-      "epoch": 2.470554504272034,
-      "grad_norm": 0.0014314352301880717,
-      "learning_rate": 0.00019999698984909877,
-      "loss": 46.0,
-      "step": 32313
-    },
-    {
-      "epoch": 2.4706309612554236,
-      "grad_norm": 0.0008748232503421605,
-      "learning_rate": 0.00019999698966272704,
-      "loss": 46.0,
-      "step": 32314
-    },
-    {
-      "epoch": 2.4707074182388133,
-      "grad_norm": 0.0006330252508632839,
-      "learning_rate": 0.0001999969894763496,
-      "loss": 46.0,
-      "step": 32315
-    },
-    {
-      "epoch": 2.470783875222203,
-      "grad_norm": 0.0024749613367021084,
-      "learning_rate": 0.00019999698928996638,
-      "loss": 46.0,
-      "step": 32316
-    },
-    {
-      "epoch": 2.470860332205593,
-      "grad_norm": 0.0007952415617182851,
-      "learning_rate": 0.00019999698910357736,
-      "loss": 46.0,
-      "step": 32317
-    },
-    {
-      "epoch": 2.4709367891889826,
-      "grad_norm": 0.0006256563356146216,
-      "learning_rate": 0.0001999969889171826,
-      "loss": 46.0,
-      "step": 32318
-    },
-    {
-      "epoch": 2.4710132461723724,
-      "grad_norm": 0.0018135980935767293,
-      "learning_rate": 0.00019999698873078206,
-      "loss": 46.0,
-      "step": 32319
-    },
-    {
-      "epoch": 2.471089703155762,
-      "grad_norm": 0.0026998789981007576,
-      "learning_rate": 0.00019999698854437575,
-      "loss": 46.0,
-      "step": 32320
-    },
-    {
-      "epoch": 2.471166160139152,
-      "grad_norm": 0.0007628350285813212,
-      "learning_rate": 0.00019999698835796367,
-      "loss": 46.0,
-      "step": 32321
-    },
-    {
-      "epoch": 2.4712426171225417,
-      "grad_norm": 0.0013211965560913086,
-      "learning_rate": 0.0001999969881715458,
-      "loss": 46.0,
-      "step": 32322
-    },
-    {
-      "epoch": 2.471319074105931,
-      "grad_norm": 0.0022015448193997145,
-      "learning_rate": 0.0001999969879851222,
-      "loss": 46.0,
-      "step": 32323
-    },
-    {
-      "epoch": 2.4713955310893208,
-      "grad_norm": 0.0012848052429035306,
-      "learning_rate": 0.00019999698779869283,
-      "loss": 46.0,
-      "step": 32324
-    },
-    {
-      "epoch": 2.4714719880727105,
-      "grad_norm": 0.0006338043604046106,
-      "learning_rate": 0.00019999698761225768,
-      "loss": 46.0,
-      "step": 32325
-    },
-    {
-      "epoch": 2.4715484450561003,
-      "grad_norm": 0.0014433825854212046,
-      "learning_rate": 0.00019999698742581673,
-      "loss": 46.0,
-      "step": 32326
-    },
-    {
-      "epoch": 2.47162490203949,
-      "grad_norm": 0.001359913614578545,
-      "learning_rate": 0.00019999698723937006,
-      "loss": 46.0,
-      "step": 32327
-    },
-    {
-      "epoch": 2.47170135902288,
-      "grad_norm": 0.0005980661371722817,
-      "learning_rate": 0.0001999969870529176,
-      "loss": 46.0,
-      "step": 32328
-    },
-    {
-      "epoch": 2.4717778160062696,
-      "grad_norm": 0.004556672647595406,
-      "learning_rate": 0.00019999698686645935,
-      "loss": 46.0,
-      "step": 32329
-    },
-    {
-      "epoch": 2.4718542729896593,
-      "grad_norm": 0.0006020395085215569,
-      "learning_rate": 0.00019999698667999536,
-      "loss": 46.0,
-      "step": 32330
-    },
-    {
-      "epoch": 2.471930729973049,
-      "grad_norm": 0.0012320332461968064,
-      "learning_rate": 0.0001999969864935256,
-      "loss": 46.0,
-      "step": 32331
-    },
-    {
-      "epoch": 2.4720071869564384,
-      "grad_norm": 0.0012373302597552538,
-      "learning_rate": 0.00019999698630705006,
-      "loss": 46.0,
-      "step": 32332
-    },
-    {
-      "epoch": 2.472083643939828,
-      "grad_norm": 0.0025837216526269913,
-      "learning_rate": 0.00019999698612056875,
-      "loss": 46.0,
-      "step": 32333
-    },
-    {
-      "epoch": 2.472160100923218,
-      "grad_norm": 0.0013923693913966417,
-      "learning_rate": 0.00019999698593408164,
-      "loss": 46.0,
-      "step": 32334
-    },
-    {
-      "epoch": 2.4722365579066077,
-      "grad_norm": 0.0004421780467964709,
-      "learning_rate": 0.00019999698574758883,
-      "loss": 46.0,
-      "step": 32335
-    },
-    {
-      "epoch": 2.4723130148899974,
-      "grad_norm": 0.001366337644867599,
-      "learning_rate": 0.00019999698556109018,
-      "loss": 46.0,
-      "step": 32336
-    },
-    {
-      "epoch": 2.472389471873387,
-      "grad_norm": 0.0003482487518340349,
-      "learning_rate": 0.00019999698537458583,
-      "loss": 46.0,
-      "step": 32337
-    },
-    {
-      "epoch": 2.472465928856777,
-      "grad_norm": 0.004511781502515078,
-      "learning_rate": 0.00019999698518807565,
-      "loss": 46.0,
-      "step": 32338
-    },
-    {
-      "epoch": 2.4725423858401667,
-      "grad_norm": 0.0009244129760190845,
-      "learning_rate": 0.00019999698500155973,
-      "loss": 46.0,
-      "step": 32339
-    },
-    {
-      "epoch": 2.4726188428235565,
-      "grad_norm": 0.0006865226314403117,
-      "learning_rate": 0.00019999698481503803,
-      "loss": 46.0,
-      "step": 32340
-    },
-    {
-      "epoch": 2.4726952998069462,
-      "grad_norm": 0.0011483271373435855,
-      "learning_rate": 0.00019999698462851056,
-      "loss": 46.0,
-      "step": 32341
-    },
-    {
-      "epoch": 2.472771756790336,
-      "grad_norm": 0.001501023885793984,
-      "learning_rate": 0.00019999698444197735,
-      "loss": 46.0,
-      "step": 32342
-    },
-    {
-      "epoch": 2.4728482137737258,
-      "grad_norm": 0.0008674286073073745,
-      "learning_rate": 0.00019999698425543833,
-      "loss": 46.0,
-      "step": 32343
-    },
-    {
-      "epoch": 2.472924670757115,
-      "grad_norm": 0.0015875682001933455,
-      "learning_rate": 0.00019999698406889357,
-      "loss": 46.0,
-      "step": 32344
-    },
-    {
-      "epoch": 2.473001127740505,
-      "grad_norm": 0.0008716161246411502,
-      "learning_rate": 0.00019999698388234303,
-      "loss": 46.0,
-      "step": 32345
-    },
-    {
-      "epoch": 2.4730775847238946,
-      "grad_norm": 0.009208821691572666,
-      "learning_rate": 0.00019999698369578672,
-      "loss": 46.0,
-      "step": 32346
-    },
-    {
-      "epoch": 2.4731540417072844,
-      "grad_norm": 0.002032272983342409,
-      "learning_rate": 0.00019999698350922464,
-      "loss": 46.0,
-      "step": 32347
-    },
-    {
-      "epoch": 2.473230498690674,
-      "grad_norm": 0.0009493536781519651,
-      "learning_rate": 0.0001999969833226568,
-      "loss": 46.0,
-      "step": 32348
-    },
-    {
-      "epoch": 2.473306955674064,
-      "grad_norm": 0.0016980364453047514,
-      "learning_rate": 0.00019999698313608318,
-      "loss": 46.0,
-      "step": 32349
-    },
-    {
-      "epoch": 2.4733834126574537,
-      "grad_norm": 0.003632499137893319,
-      "learning_rate": 0.00019999698294950377,
-      "loss": 46.0,
-      "step": 32350
-    },
-    {
-      "epoch": 2.4734598696408434,
-      "grad_norm": 0.0018604836659505963,
-      "learning_rate": 0.00019999698276291863,
-      "loss": 46.0,
-      "step": 32351
-    },
-    {
-      "epoch": 2.473536326624233,
-      "grad_norm": 0.0009248324204236269,
-      "learning_rate": 0.0001999969825763277,
-      "loss": 46.0,
-      "step": 32352
-    },
-    {
-      "epoch": 2.473612783607623,
-      "grad_norm": 0.0007362568285316229,
-      "learning_rate": 0.000199996982389731,
-      "loss": 46.0,
-      "step": 32353
-    },
-    {
-      "epoch": 2.4736892405910123,
-      "grad_norm": 0.0016131887678056955,
-      "learning_rate": 0.00019999698220312854,
-      "loss": 46.0,
-      "step": 32354
-    },
-    {
-      "epoch": 2.473765697574402,
-      "grad_norm": 0.002820591675117612,
-      "learning_rate": 0.0001999969820165203,
-      "loss": 46.0,
-      "step": 32355
-    },
-    {
-      "epoch": 2.4738421545577918,
-      "grad_norm": 0.002857802202925086,
-      "learning_rate": 0.0001999969818299063,
-      "loss": 46.0,
-      "step": 32356
-    },
-    {
-      "epoch": 2.4739186115411815,
-      "grad_norm": 0.0010428531095385551,
-      "learning_rate": 0.00019999698164328654,
-      "loss": 46.0,
-      "step": 32357
-    },
-    {
-      "epoch": 2.4739950685245713,
-      "grad_norm": 0.001030318788252771,
-      "learning_rate": 0.00019999698145666098,
-      "loss": 46.0,
-      "step": 32358
-    },
-    {
-      "epoch": 2.474071525507961,
-      "grad_norm": 0.016375398263335228,
-      "learning_rate": 0.0001999969812700297,
-      "loss": 46.0,
-      "step": 32359
-    },
-    {
-      "epoch": 2.474147982491351,
-      "grad_norm": 0.0018805437721312046,
-      "learning_rate": 0.0001999969810833926,
-      "loss": 46.0,
-      "step": 32360
-    },
-    {
-      "epoch": 2.4742244394747406,
-      "grad_norm": 0.001070512575097382,
-      "learning_rate": 0.00019999698089674976,
-      "loss": 46.0,
-      "step": 32361
-    },
-    {
-      "epoch": 2.4743008964581303,
-      "grad_norm": 0.002054716693237424,
-      "learning_rate": 0.00019999698071010116,
-      "loss": 46.0,
-      "step": 32362
-    },
-    {
-      "epoch": 2.47437735344152,
-      "grad_norm": 0.0006869182689115405,
-      "learning_rate": 0.00019999698052344675,
-      "loss": 46.0,
-      "step": 32363
-    },
-    {
-      "epoch": 2.47445381042491,
-      "grad_norm": 0.0006861745496280491,
-      "learning_rate": 0.0001999969803367866,
-      "loss": 46.0,
-      "step": 32364
-    },
-    {
-      "epoch": 2.4745302674082996,
-      "grad_norm": 0.0037119807675480843,
-      "learning_rate": 0.00019999698015012068,
-      "loss": 46.0,
-      "step": 32365
-    },
-    {
-      "epoch": 2.474606724391689,
-      "grad_norm": 0.001989027950912714,
-      "learning_rate": 0.00019999697996344898,
-      "loss": 46.0,
-      "step": 32366
-    },
-    {
-      "epoch": 2.4746831813750787,
-      "grad_norm": 0.001872391439974308,
-      "learning_rate": 0.00019999697977677152,
-      "loss": 46.0,
-      "step": 32367
-    },
-    {
-      "epoch": 2.4747596383584685,
-      "grad_norm": 0.0007466906099580228,
-      "learning_rate": 0.00019999697959008827,
-      "loss": 46.0,
-      "step": 32368
-    },
-    {
-      "epoch": 2.474836095341858,
-      "grad_norm": 0.002406191546469927,
-      "learning_rate": 0.00019999697940339928,
-      "loss": 46.0,
-      "step": 32369
-    },
-    {
-      "epoch": 2.474912552325248,
-      "grad_norm": 0.0009625056991353631,
-      "learning_rate": 0.0001999969792167045,
-      "loss": 46.0,
-      "step": 32370
-    },
-    {
-      "epoch": 2.4749890093086377,
-      "grad_norm": 0.0009684687829576433,
-      "learning_rate": 0.00019999697903000396,
-      "loss": 46.0,
-      "step": 32371
-    },
-    {
-      "epoch": 2.4750654662920275,
-      "grad_norm": 0.0006879267748445272,
-      "learning_rate": 0.00019999697884329765,
-      "loss": 46.0,
-      "step": 32372
-    },
-    {
-      "epoch": 2.4751419232754173,
-      "grad_norm": 0.0007508319686166942,
-      "learning_rate": 0.00019999697865658557,
-      "loss": 46.0,
-      "step": 32373
-    },
-    {
-      "epoch": 2.475218380258807,
-      "grad_norm": 0.0012423944426700473,
-      "learning_rate": 0.00019999697846986774,
-      "loss": 46.0,
-      "step": 32374
-    },
-    {
-      "epoch": 2.475294837242197,
-      "grad_norm": 0.001323615200817585,
-      "learning_rate": 0.0001999969782831441,
-      "loss": 46.0,
-      "step": 32375
-    },
-    {
-      "epoch": 2.475371294225586,
-      "grad_norm": 0.0038680562283843756,
-      "learning_rate": 0.00019999697809641473,
-      "loss": 46.0,
-      "step": 32376
-    },
-    {
-      "epoch": 2.475447751208976,
-      "grad_norm": 0.000774978194385767,
-      "learning_rate": 0.00019999697790967959,
-      "loss": 46.0,
-      "step": 32377
-    },
-    {
-      "epoch": 2.4755242081923656,
-      "grad_norm": 0.0005598383140750229,
-      "learning_rate": 0.00019999697772293864,
-      "loss": 46.0,
-      "step": 32378
-    },
-    {
-      "epoch": 2.4756006651757554,
-      "grad_norm": 0.0015458286507055163,
-      "learning_rate": 0.00019999697753619194,
-      "loss": 46.0,
-      "step": 32379
-    },
-    {
-      "epoch": 2.475677122159145,
-      "grad_norm": 0.0006524888449348509,
-      "learning_rate": 0.00019999697734943947,
-      "loss": 46.0,
-      "step": 32380
-    },
-    {
-      "epoch": 2.475753579142535,
-      "grad_norm": 0.00337987975217402,
-      "learning_rate": 0.00019999697716268123,
-      "loss": 46.0,
-      "step": 32381
-    },
-    {
-      "epoch": 2.4758300361259247,
-      "grad_norm": 0.0012290806043893099,
-      "learning_rate": 0.00019999697697591724,
-      "loss": 46.0,
-      "step": 32382
-    },
-    {
-      "epoch": 2.4759064931093144,
-      "grad_norm": 0.0008876342908479273,
-      "learning_rate": 0.00019999697678914745,
-      "loss": 46.0,
-      "step": 32383
-    },
-    {
-      "epoch": 2.475982950092704,
-      "grad_norm": 0.0016609154408797622,
-      "learning_rate": 0.00019999697660237194,
-      "loss": 46.0,
-      "step": 32384
-    },
-    {
-      "epoch": 2.476059407076094,
-      "grad_norm": 0.005838238634169102,
-      "learning_rate": 0.0001999969764155906,
-      "loss": 46.0,
-      "step": 32385
-    },
-    {
-      "epoch": 2.4761358640594837,
-      "grad_norm": 0.004494909197092056,
-      "learning_rate": 0.00019999697622880353,
-      "loss": 46.0,
-      "step": 32386
-    },
-    {
-      "epoch": 2.4762123210428735,
-      "grad_norm": 0.0010775546543300152,
-      "learning_rate": 0.0001999969760420107,
-      "loss": 46.0,
-      "step": 32387
-    },
-    {
-      "epoch": 2.476288778026263,
-      "grad_norm": 0.0007386936922557652,
-      "learning_rate": 0.00019999697585521207,
-      "loss": 46.0,
-      "step": 32388
-    },
-    {
-      "epoch": 2.4763652350096526,
-      "grad_norm": 0.0006629854906350374,
-      "learning_rate": 0.00019999697566840767,
-      "loss": 46.0,
-      "step": 32389
-    },
-    {
-      "epoch": 2.4764416919930423,
-      "grad_norm": 0.0007553491741418839,
-      "learning_rate": 0.00019999697548159752,
-      "loss": 46.0,
-      "step": 32390
-    },
-    {
-      "epoch": 2.476518148976432,
-      "grad_norm": 0.0013514121528714895,
-      "learning_rate": 0.0001999969752947816,
-      "loss": 46.0,
-      "step": 32391
-    },
-    {
-      "epoch": 2.476594605959822,
-      "grad_norm": 0.000956163217779249,
-      "learning_rate": 0.0001999969751079599,
-      "loss": 46.0,
-      "step": 32392
-    },
-    {
-      "epoch": 2.4766710629432116,
-      "grad_norm": 0.0015494560357183218,
-      "learning_rate": 0.00019999697492113243,
-      "loss": 46.0,
-      "step": 32393
-    },
-    {
-      "epoch": 2.4767475199266014,
-      "grad_norm": 0.0021596322767436504,
-      "learning_rate": 0.0001999969747342992,
-      "loss": 46.0,
-      "step": 32394
-    },
-    {
-      "epoch": 2.476823976909991,
-      "grad_norm": 0.003033431014046073,
-      "learning_rate": 0.00019999697454746017,
-      "loss": 46.0,
-      "step": 32395
-    },
-    {
-      "epoch": 2.476900433893381,
-      "grad_norm": 0.0135241923853755,
-      "learning_rate": 0.0001999969743606154,
-      "loss": 46.0,
-      "step": 32396
-    },
-    {
-      "epoch": 2.4769768908767706,
-      "grad_norm": 0.0014106598682701588,
-      "learning_rate": 0.00019999697417376488,
-      "loss": 46.0,
-      "step": 32397
-    },
-    {
-      "epoch": 2.47705334786016,
-      "grad_norm": 0.0006915717967785895,
-      "learning_rate": 0.00019999697398690857,
-      "loss": 46.0,
-      "step": 32398
-    },
-    {
-      "epoch": 2.4771298048435497,
-      "grad_norm": 0.0007915030582807958,
-      "learning_rate": 0.0001999969738000465,
-      "loss": 46.0,
-      "step": 32399
-    },
-    {
-      "epoch": 2.4772062618269395,
-      "grad_norm": 0.0016740431310608983,
-      "learning_rate": 0.00019999697361317863,
-      "loss": 46.0,
-      "step": 32400
-    },
-    {
-      "epoch": 2.4772827188103292,
-      "grad_norm": 0.0011863181134685874,
-      "learning_rate": 0.000199996973426305,
-      "loss": 46.0,
-      "step": 32401
-    },
-    {
-      "epoch": 2.477359175793719,
-      "grad_norm": 0.0049256556667387486,
-      "learning_rate": 0.0001999969732394256,
-      "loss": 46.0,
-      "step": 32402
-    },
-    {
-      "epoch": 2.4774356327771088,
-      "grad_norm": 0.008531459607183933,
-      "learning_rate": 0.00019999697305254045,
-      "loss": 46.0,
-      "step": 32403
-    },
-    {
-      "epoch": 2.4775120897604985,
-      "grad_norm": 0.0011433891486376524,
-      "learning_rate": 0.00019999697286564953,
-      "loss": 46.0,
-      "step": 32404
-    },
-    {
-      "epoch": 2.4775885467438883,
-      "grad_norm": 0.0006535463035106659,
-      "learning_rate": 0.00019999697267875284,
-      "loss": 46.0,
-      "step": 32405
-    },
-    {
-      "epoch": 2.477665003727278,
-      "grad_norm": 0.0004138807416893542,
-      "learning_rate": 0.00019999697249185037,
-      "loss": 46.0,
-      "step": 32406
-    },
-    {
-      "epoch": 2.477741460710668,
-      "grad_norm": 0.0014902279945090413,
-      "learning_rate": 0.00019999697230494215,
-      "loss": 46.0,
-      "step": 32407
-    },
-    {
-      "epoch": 2.4778179176940576,
-      "grad_norm": 0.0012376923114061356,
-      "learning_rate": 0.00019999697211802814,
-      "loss": 46.0,
-      "step": 32408
-    },
-    {
-      "epoch": 2.4778943746774473,
-      "grad_norm": 0.0007314681424759328,
-      "learning_rate": 0.00019999697193110835,
-      "loss": 46.0,
-      "step": 32409
-    },
-    {
-      "epoch": 2.4779708316608366,
-      "grad_norm": 0.00043874586117453873,
-      "learning_rate": 0.00019999697174418282,
-      "loss": 46.0,
-      "step": 32410
-    },
-    {
-      "epoch": 2.4780472886442264,
-      "grad_norm": 0.0008751305867917836,
-      "learning_rate": 0.0001999969715572515,
-      "loss": 46.0,
-      "step": 32411
-    },
-    {
-      "epoch": 2.478123745627616,
-      "grad_norm": 0.0021014243829995394,
-      "learning_rate": 0.00019999697137031443,
-      "loss": 46.0,
-      "step": 32412
-    },
-    {
-      "epoch": 2.478200202611006,
-      "grad_norm": 0.0017999489791691303,
-      "learning_rate": 0.00019999697118337157,
-      "loss": 46.0,
-      "step": 32413
-    },
-    {
-      "epoch": 2.4782766595943957,
-      "grad_norm": 0.0012681579682976007,
-      "learning_rate": 0.00019999697099642294,
-      "loss": 46.0,
-      "step": 32414
-    },
-    {
-      "epoch": 2.4783531165777855,
-      "grad_norm": 0.005125392694026232,
-      "learning_rate": 0.00019999697080946854,
-      "loss": 46.0,
-      "step": 32415
-    },
-    {
-      "epoch": 2.478429573561175,
-      "grad_norm": 0.0006486070342361927,
-      "learning_rate": 0.0001999969706225084,
-      "loss": 46.0,
-      "step": 32416
-    },
-    {
-      "epoch": 2.478506030544565,
-      "grad_norm": 0.0009953194530680776,
-      "learning_rate": 0.00019999697043554247,
-      "loss": 46.0,
-      "step": 32417
-    },
-    {
-      "epoch": 2.4785824875279547,
-      "grad_norm": 0.0013166008284315467,
-      "learning_rate": 0.00019999697024857075,
-      "loss": 46.0,
-      "step": 32418
-    },
-    {
-      "epoch": 2.4786589445113445,
-      "grad_norm": 0.0005559337441809475,
-      "learning_rate": 0.0001999969700615933,
-      "loss": 46.0,
-      "step": 32419
-    },
-    {
-      "epoch": 2.478735401494734,
-      "grad_norm": 0.001008525025099516,
-      "learning_rate": 0.00019999696987461007,
-      "loss": 46.0,
-      "step": 32420
-    },
-    {
-      "epoch": 2.4788118584781236,
-      "grad_norm": 0.0012607964454218745,
-      "learning_rate": 0.00019999696968762108,
-      "loss": 46.0,
-      "step": 32421
-    },
-    {
-      "epoch": 2.4788883154615133,
-      "grad_norm": 0.0005901282420381904,
-      "learning_rate": 0.0001999969695006263,
-      "loss": 46.0,
-      "step": 32422
-    },
-    {
-      "epoch": 2.478964772444903,
-      "grad_norm": 0.0042494540102779865,
-      "learning_rate": 0.00019999696931362576,
-      "loss": 46.0,
-      "step": 32423
-    },
-    {
-      "epoch": 2.479041229428293,
-      "grad_norm": 0.0010255294619128108,
-      "learning_rate": 0.00019999696912661942,
-      "loss": 46.0,
-      "step": 32424
-    },
-    {
-      "epoch": 2.4791176864116826,
-      "grad_norm": 0.0007286560721695423,
-      "learning_rate": 0.00019999696893960734,
-      "loss": 46.0,
-      "step": 32425
-    },
-    {
-      "epoch": 2.4791941433950724,
-      "grad_norm": 0.0015877713449299335,
-      "learning_rate": 0.00019999696875258951,
-      "loss": 46.0,
-      "step": 32426
-    },
-    {
-      "epoch": 2.479270600378462,
-      "grad_norm": 0.001166089903563261,
-      "learning_rate": 0.00019999696856556586,
-      "loss": 46.0,
-      "step": 32427
-    },
-    {
-      "epoch": 2.479347057361852,
-      "grad_norm": 0.0010971451411023736,
-      "learning_rate": 0.00019999696837853649,
-      "loss": 46.0,
-      "step": 32428
-    },
-    {
-      "epoch": 2.4794235143452417,
-      "grad_norm": 0.0034140772186219692,
-      "learning_rate": 0.0001999969681915013,
-      "loss": 46.0,
-      "step": 32429
-    },
-    {
-      "epoch": 2.4794999713286314,
-      "grad_norm": 0.00035172211937606335,
-      "learning_rate": 0.0001999969680044604,
-      "loss": 46.0,
-      "step": 32430
-    },
-    {
-      "epoch": 2.479576428312021,
-      "grad_norm": 0.005393309984356165,
-      "learning_rate": 0.0001999969678174137,
-      "loss": 46.0,
-      "step": 32431
-    },
-    {
-      "epoch": 2.4796528852954105,
-      "grad_norm": 0.0007292910595424473,
-      "learning_rate": 0.00019999696763036125,
-      "loss": 46.0,
-      "step": 32432
-    },
-    {
-      "epoch": 2.4797293422788003,
-      "grad_norm": 0.0033124315086752176,
-      "learning_rate": 0.000199996967443303,
-      "loss": 46.0,
-      "step": 32433
-    },
-    {
-      "epoch": 2.47980579926219,
-      "grad_norm": 0.0006346868467517197,
-      "learning_rate": 0.000199996967256239,
-      "loss": 46.0,
-      "step": 32434
-    },
-    {
-      "epoch": 2.47988225624558,
-      "grad_norm": 0.00042091519571840763,
-      "learning_rate": 0.0001999969670691692,
-      "loss": 46.0,
-      "step": 32435
-    },
-    {
-      "epoch": 2.4799587132289695,
-      "grad_norm": 0.0014532859204337,
-      "learning_rate": 0.00019999696688209365,
-      "loss": 46.0,
-      "step": 32436
-    },
-    {
-      "epoch": 2.4800351702123593,
-      "grad_norm": 0.0007465852540917695,
-      "learning_rate": 0.00019999696669501234,
-      "loss": 46.0,
-      "step": 32437
-    },
-    {
-      "epoch": 2.480111627195749,
-      "grad_norm": 0.0013091983273625374,
-      "learning_rate": 0.0001999969665079253,
-      "loss": 46.0,
-      "step": 32438
-    },
-    {
-      "epoch": 2.480188084179139,
-      "grad_norm": 0.00368438265286386,
-      "learning_rate": 0.00019999696632083243,
-      "loss": 46.0,
-      "step": 32439
-    },
-    {
-      "epoch": 2.4802645411625286,
-      "grad_norm": 0.0008188743959181011,
-      "learning_rate": 0.0001999969661337338,
-      "loss": 46.0,
-      "step": 32440
-    },
-    {
-      "epoch": 2.480340998145918,
-      "grad_norm": 0.0009064456098712981,
-      "learning_rate": 0.00019999696594662943,
-      "loss": 46.0,
-      "step": 32441
-    },
-    {
-      "epoch": 2.4804174551293077,
-      "grad_norm": 0.0004914067685604095,
-      "learning_rate": 0.00019999696575951926,
-      "loss": 46.0,
-      "step": 32442
-    },
-    {
-      "epoch": 2.4804939121126974,
-      "grad_norm": 0.001155688427388668,
-      "learning_rate": 0.0001999969655724033,
-      "loss": 46.0,
-      "step": 32443
-    },
-    {
-      "epoch": 2.480570369096087,
-      "grad_norm": 0.0013556908816099167,
-      "learning_rate": 0.00019999696538528162,
-      "loss": 46.0,
-      "step": 32444
-    },
-    {
-      "epoch": 2.480646826079477,
-      "grad_norm": 0.0025988768320530653,
-      "learning_rate": 0.00019999696519815415,
-      "loss": 46.0,
-      "step": 32445
-    },
-    {
-      "epoch": 2.4807232830628667,
-      "grad_norm": 0.0007351776584982872,
-      "learning_rate": 0.0001999969650110209,
-      "loss": 46.0,
-      "step": 32446
-    },
-    {
-      "epoch": 2.4807997400462565,
-      "grad_norm": 0.001170598203316331,
-      "learning_rate": 0.0001999969648238819,
-      "loss": 46.0,
-      "step": 32447
-    },
-    {
-      "epoch": 2.4808761970296462,
-      "grad_norm": 0.0006654290482401848,
-      "learning_rate": 0.00019999696463673713,
-      "loss": 46.0,
-      "step": 32448
-    },
-    {
-      "epoch": 2.480952654013036,
-      "grad_norm": 0.0014254014240577817,
-      "learning_rate": 0.00019999696444958657,
-      "loss": 46.0,
-      "step": 32449
-    },
-    {
-      "epoch": 2.4810291109964258,
-      "grad_norm": 0.0009064446203410625,
-      "learning_rate": 0.0001999969642624303,
-      "loss": 46.0,
-      "step": 32450
-    },
-    {
-      "epoch": 2.4811055679798155,
-      "grad_norm": 0.0004398701130412519,
-      "learning_rate": 0.00019999696407526819,
-      "loss": 46.0,
-      "step": 32451
-    },
-    {
-      "epoch": 2.4811820249632053,
-      "grad_norm": 0.0022492967545986176,
-      "learning_rate": 0.00019999696388810033,
-      "loss": 46.0,
-      "step": 32452
-    },
-    {
-      "epoch": 2.481258481946595,
-      "grad_norm": 0.004112998489290476,
-      "learning_rate": 0.0001999969637009267,
-      "loss": 46.0,
-      "step": 32453
-    },
-    {
-      "epoch": 2.4813349389299844,
-      "grad_norm": 0.000455722794868052,
-      "learning_rate": 0.0001999969635137473,
-      "loss": 46.0,
-      "step": 32454
-    },
-    {
-      "epoch": 2.481411395913374,
-      "grad_norm": 0.0008955828379839659,
-      "learning_rate": 0.00019999696332656216,
-      "loss": 46.0,
-      "step": 32455
-    },
-    {
-      "epoch": 2.481487852896764,
-      "grad_norm": 0.0007259807898662984,
-      "learning_rate": 0.0001999969631393712,
-      "loss": 46.0,
-      "step": 32456
-    },
-    {
-      "epoch": 2.4815643098801536,
-      "grad_norm": 0.0006770557956770062,
-      "learning_rate": 0.00019999696295217452,
-      "loss": 46.0,
-      "step": 32457
-    },
-    {
-      "epoch": 2.4816407668635434,
-      "grad_norm": 0.005172649398446083,
-      "learning_rate": 0.00019999696276497208,
-      "loss": 46.0,
-      "step": 32458
-    },
-    {
-      "epoch": 2.481717223846933,
-      "grad_norm": 0.0006946736248210073,
-      "learning_rate": 0.00019999696257776384,
-      "loss": 46.0,
-      "step": 32459
-    },
-    {
-      "epoch": 2.481793680830323,
-      "grad_norm": 0.0003239812212996185,
-      "learning_rate": 0.00019999696239054982,
-      "loss": 46.0,
-      "step": 32460
-    },
-    {
-      "epoch": 2.4818701378137127,
-      "grad_norm": 0.0033833985216915607,
-      "learning_rate": 0.00019999696220333003,
-      "loss": 46.0,
-      "step": 32461
-    },
-    {
-      "epoch": 2.4819465947971024,
-      "grad_norm": 0.0011916621588170528,
-      "learning_rate": 0.0001999969620161045,
-      "loss": 46.0,
-      "step": 32462
-    },
-    {
-      "epoch": 2.4820230517804918,
-      "grad_norm": 0.0005481817643158138,
-      "learning_rate": 0.00019999696182887317,
-      "loss": 46.0,
-      "step": 32463
-    },
-    {
-      "epoch": 2.4820995087638815,
-      "grad_norm": 0.0004768413200508803,
-      "learning_rate": 0.0001999969616416361,
-      "loss": 46.0,
-      "step": 32464
-    },
-    {
-      "epoch": 2.4821759657472713,
-      "grad_norm": 0.0014096852391958237,
-      "learning_rate": 0.00019999696145439323,
-      "loss": 46.0,
-      "step": 32465
-    },
-    {
-      "epoch": 2.482252422730661,
-      "grad_norm": 0.005327824503183365,
-      "learning_rate": 0.00019999696126714463,
-      "loss": 46.0,
-      "step": 32466
-    },
-    {
-      "epoch": 2.482328879714051,
-      "grad_norm": 0.0014724304201081395,
-      "learning_rate": 0.00019999696107989023,
-      "loss": 46.0,
-      "step": 32467
-    },
-    {
-      "epoch": 2.4824053366974406,
-      "grad_norm": 0.0009439281420782208,
-      "learning_rate": 0.00019999696089263006,
-      "loss": 46.0,
-      "step": 32468
-    },
-    {
-      "epoch": 2.4824817936808303,
-      "grad_norm": 0.0021287312265485525,
-      "learning_rate": 0.00019999696070536414,
-      "loss": 46.0,
-      "step": 32469
-    },
-    {
-      "epoch": 2.48255825066422,
-      "grad_norm": 0.0019370531663298607,
-      "learning_rate": 0.00019999696051809242,
-      "loss": 46.0,
-      "step": 32470
-    },
-    {
-      "epoch": 2.48263470764761,
-      "grad_norm": 0.000743091746699065,
-      "learning_rate": 0.00019999696033081495,
-      "loss": 46.0,
-      "step": 32471
-    },
-    {
-      "epoch": 2.4827111646309996,
-      "grad_norm": 0.0014310280093923211,
-      "learning_rate": 0.00019999696014353174,
-      "loss": 46.0,
-      "step": 32472
-    },
-    {
-      "epoch": 2.4827876216143894,
-      "grad_norm": 0.0010180834215134382,
-      "learning_rate": 0.0001999969599562427,
-      "loss": 46.0,
-      "step": 32473
-    },
-    {
-      "epoch": 2.482864078597779,
-      "grad_norm": 0.0030324929393827915,
-      "learning_rate": 0.00019999695976894794,
-      "loss": 46.0,
-      "step": 32474
-    },
-    {
-      "epoch": 2.4829405355811685,
-      "grad_norm": 0.0007764458423480392,
-      "learning_rate": 0.00019999695958164738,
-      "loss": 46.0,
-      "step": 32475
-    },
-    {
-      "epoch": 2.483016992564558,
-      "grad_norm": 0.0027090120129287243,
-      "learning_rate": 0.00019999695939434107,
-      "loss": 46.0,
-      "step": 32476
-    },
-    {
-      "epoch": 2.483093449547948,
-      "grad_norm": 0.0004728749336209148,
-      "learning_rate": 0.000199996959207029,
-      "loss": 46.0,
-      "step": 32477
-    },
-    {
-      "epoch": 2.4831699065313377,
-      "grad_norm": 0.0012124466011300683,
-      "learning_rate": 0.00019999695901971114,
-      "loss": 46.0,
-      "step": 32478
-    },
-    {
-      "epoch": 2.4832463635147275,
-      "grad_norm": 0.012036803178489208,
-      "learning_rate": 0.0001999969588323875,
-      "loss": 46.0,
-      "step": 32479
-    },
-    {
-      "epoch": 2.4833228204981173,
-      "grad_norm": 0.0007612515473738313,
-      "learning_rate": 0.0001999969586450581,
-      "loss": 46.0,
-      "step": 32480
-    },
-    {
-      "epoch": 2.483399277481507,
-      "grad_norm": 0.0007346108322963119,
-      "learning_rate": 0.00019999695845772294,
-      "loss": 46.0,
-      "step": 32481
-    },
-    {
-      "epoch": 2.483475734464897,
-      "grad_norm": 0.0005843139952048659,
-      "learning_rate": 0.00019999695827038202,
-      "loss": 46.0,
-      "step": 32482
-    },
-    {
-      "epoch": 2.4835521914482865,
-      "grad_norm": 0.0017749707913026214,
-      "learning_rate": 0.00019999695808303533,
-      "loss": 46.0,
-      "step": 32483
-    },
-    {
-      "epoch": 2.4836286484316763,
-      "grad_norm": 0.013531950302422047,
-      "learning_rate": 0.00019999695789568286,
-      "loss": 46.0,
-      "step": 32484
-    },
-    {
-      "epoch": 2.4837051054150656,
-      "grad_norm": 0.0015886970795691013,
-      "learning_rate": 0.00019999695770832462,
-      "loss": 46.0,
-      "step": 32485
-    },
-    {
-      "epoch": 2.4837815623984554,
-      "grad_norm": 0.0021449862979352474,
-      "learning_rate": 0.0001999969575209606,
-      "loss": 46.0,
-      "step": 32486
-    },
-    {
-      "epoch": 2.483858019381845,
-      "grad_norm": 0.000996781513094902,
-      "learning_rate": 0.00019999695733359082,
-      "loss": 46.0,
-      "step": 32487
-    },
-    {
-      "epoch": 2.483934476365235,
-      "grad_norm": 0.0012609054101631045,
-      "learning_rate": 0.0001999969571462153,
-      "loss": 46.0,
-      "step": 32488
-    },
-    {
-      "epoch": 2.4840109333486247,
-      "grad_norm": 0.0006104173371568322,
-      "learning_rate": 0.00019999695695883398,
-      "loss": 46.0,
-      "step": 32489
-    },
-    {
-      "epoch": 2.4840873903320144,
-      "grad_norm": 0.0028106090612709522,
-      "learning_rate": 0.0001999969567714469,
-      "loss": 46.0,
-      "step": 32490
-    },
-    {
-      "epoch": 2.484163847315404,
-      "grad_norm": 0.0007372854161076248,
-      "learning_rate": 0.00019999695658405405,
-      "loss": 46.0,
-      "step": 32491
-    },
-    {
-      "epoch": 2.484240304298794,
-      "grad_norm": 0.001557126292027533,
-      "learning_rate": 0.00019999695639665542,
-      "loss": 46.0,
-      "step": 32492
-    },
-    {
-      "epoch": 2.4843167612821837,
-      "grad_norm": 0.000501113070640713,
-      "learning_rate": 0.000199996956209251,
-      "loss": 46.0,
-      "step": 32493
-    },
-    {
-      "epoch": 2.4843932182655735,
-      "grad_norm": 0.0021587435621768236,
-      "learning_rate": 0.00019999695602184082,
-      "loss": 46.0,
-      "step": 32494
-    },
-    {
-      "epoch": 2.4844696752489632,
-      "grad_norm": 0.0008666223147884011,
-      "learning_rate": 0.00019999695583442493,
-      "loss": 46.0,
-      "step": 32495
-    },
-    {
-      "epoch": 2.484546132232353,
-      "grad_norm": 0.000774035113863647,
-      "learning_rate": 0.0001999969556470032,
-      "loss": 46.0,
-      "step": 32496
-    },
-    {
-      "epoch": 2.4846225892157423,
-      "grad_norm": 0.00021022098371759057,
-      "learning_rate": 0.00019999695545957574,
-      "loss": 46.0,
-      "step": 32497
-    },
-    {
-      "epoch": 2.484699046199132,
-      "grad_norm": 0.0007480334024876356,
-      "learning_rate": 0.0001999969552721425,
-      "loss": 46.0,
-      "step": 32498
-    },
-    {
-      "epoch": 2.484775503182522,
-      "grad_norm": 0.000816861167550087,
-      "learning_rate": 0.0001999969550847035,
-      "loss": 46.0,
-      "step": 32499
-    },
-    {
-      "epoch": 2.4848519601659116,
-      "grad_norm": 0.0007491225842386484,
-      "learning_rate": 0.0001999969548972587,
-      "loss": 46.0,
-      "step": 32500
-    },
-    {
-      "epoch": 2.4849284171493013,
-      "grad_norm": 0.000617528916336596,
-      "learning_rate": 0.00019999695470980817,
-      "loss": 46.0,
-      "step": 32501
-    },
-    {
-      "epoch": 2.485004874132691,
-      "grad_norm": 0.008154521696269512,
-      "learning_rate": 0.00019999695452235187,
-      "loss": 46.0,
-      "step": 32502
-    },
-    {
-      "epoch": 2.485081331116081,
-      "grad_norm": 0.0012328510638326406,
-      "learning_rate": 0.00019999695433488976,
-      "loss": 46.0,
-      "step": 32503
-    },
-    {
-      "epoch": 2.4851577880994706,
-      "grad_norm": 0.0009552455157972872,
-      "learning_rate": 0.0001999969541474219,
-      "loss": 46.0,
-      "step": 32504
-    },
-    {
-      "epoch": 2.4852342450828604,
-      "grad_norm": 0.0008059277315624058,
-      "learning_rate": 0.00019999695395994828,
-      "loss": 46.0,
-      "step": 32505
-    },
-    {
-      "epoch": 2.48531070206625,
-      "grad_norm": 0.0009990391554310918,
-      "learning_rate": 0.0001999969537724689,
-      "loss": 46.0,
-      "step": 32506
-    },
-    {
-      "epoch": 2.4853871590496395,
-      "grad_norm": 0.004519835114479065,
-      "learning_rate": 0.0001999969535849837,
-      "loss": 46.0,
-      "step": 32507
-    },
-    {
-      "epoch": 2.4854636160330292,
-      "grad_norm": 0.0011185647454112768,
-      "learning_rate": 0.0001999969533974928,
-      "loss": 46.0,
-      "step": 32508
-    },
-    {
-      "epoch": 2.485540073016419,
-      "grad_norm": 0.005013096146285534,
-      "learning_rate": 0.00019999695320999607,
-      "loss": 46.0,
-      "step": 32509
-    },
-    {
-      "epoch": 2.4856165299998088,
-      "grad_norm": 0.006383952219039202,
-      "learning_rate": 0.0001999969530224936,
-      "loss": 46.0,
-      "step": 32510
-    },
-    {
-      "epoch": 2.4856929869831985,
-      "grad_norm": 0.0010234022047370672,
-      "learning_rate": 0.0001999969528349854,
-      "loss": 46.0,
-      "step": 32511
-    },
-    {
-      "epoch": 2.4857694439665883,
-      "grad_norm": 0.0013509561540558934,
-      "learning_rate": 0.00019999695264747138,
-      "loss": 46.0,
-      "step": 32512
-    },
-    {
-      "epoch": 2.485845900949978,
-      "grad_norm": 0.0020701014436781406,
-      "learning_rate": 0.0001999969524599516,
-      "loss": 46.0,
-      "step": 32513
-    },
-    {
-      "epoch": 2.485922357933368,
-      "grad_norm": 0.0007009868859313428,
-      "learning_rate": 0.00019999695227242606,
-      "loss": 46.0,
-      "step": 32514
-    },
-    {
-      "epoch": 2.4859988149167576,
-      "grad_norm": 0.0017209473298862576,
-      "learning_rate": 0.00019999695208489473,
-      "loss": 46.0,
-      "step": 32515
-    },
-    {
-      "epoch": 2.4860752719001473,
-      "grad_norm": 0.0010355832055211067,
-      "learning_rate": 0.00019999695189735762,
-      "loss": 46.0,
-      "step": 32516
-    },
-    {
-      "epoch": 2.486151728883537,
-      "grad_norm": 0.0010259883711114526,
-      "learning_rate": 0.00019999695170981477,
-      "loss": 46.0,
-      "step": 32517
-    },
-    {
-      "epoch": 2.486228185866927,
-      "grad_norm": 0.0019412415567785501,
-      "learning_rate": 0.00019999695152226617,
-      "loss": 46.0,
-      "step": 32518
-    },
-    {
-      "epoch": 2.486304642850316,
-      "grad_norm": 0.0009146252414211631,
-      "learning_rate": 0.00019999695133471175,
-      "loss": 46.0,
-      "step": 32519
-    },
-    {
-      "epoch": 2.486381099833706,
-      "grad_norm": 0.003369443817064166,
-      "learning_rate": 0.0001999969511471516,
-      "loss": 46.0,
-      "step": 32520
-    },
-    {
-      "epoch": 2.4864575568170957,
-      "grad_norm": 0.0010440386831760406,
-      "learning_rate": 0.00019999695095958565,
-      "loss": 46.0,
-      "step": 32521
-    },
-    {
-      "epoch": 2.4865340138004854,
-      "grad_norm": 0.0016332159284502268,
-      "learning_rate": 0.00019999695077201394,
-      "loss": 46.0,
-      "step": 32522
-    },
-    {
-      "epoch": 2.486610470783875,
-      "grad_norm": 0.0008343071676790714,
-      "learning_rate": 0.00019999695058443647,
-      "loss": 46.0,
-      "step": 32523
-    },
-    {
-      "epoch": 2.486686927767265,
-      "grad_norm": 0.0004725205944851041,
-      "learning_rate": 0.00019999695039685323,
-      "loss": 46.0,
-      "step": 32524
-    },
-    {
-      "epoch": 2.4867633847506547,
-      "grad_norm": 0.0013226979644969106,
-      "learning_rate": 0.00019999695020926425,
-      "loss": 46.0,
-      "step": 32525
-    },
-    {
-      "epoch": 2.4868398417340445,
-      "grad_norm": 0.00045004760613664985,
-      "learning_rate": 0.00019999695002166946,
-      "loss": 46.0,
-      "step": 32526
-    },
-    {
-      "epoch": 2.4869162987174342,
-      "grad_norm": 0.0012962721521034837,
-      "learning_rate": 0.0001999969498340689,
-      "loss": 46.0,
-      "step": 32527
-    },
-    {
-      "epoch": 2.486992755700824,
-      "grad_norm": 0.001267138752155006,
-      "learning_rate": 0.0001999969496464626,
-      "loss": 46.0,
-      "step": 32528
-    },
-    {
-      "epoch": 2.4870692126842133,
-      "grad_norm": 0.0006998016033321619,
-      "learning_rate": 0.0001999969494588505,
-      "loss": 46.0,
-      "step": 32529
-    },
-    {
-      "epoch": 2.487145669667603,
-      "grad_norm": 0.0007355526322498918,
-      "learning_rate": 0.00019999694927123264,
-      "loss": 46.0,
-      "step": 32530
-    },
-    {
-      "epoch": 2.487222126650993,
-      "grad_norm": 0.0008501976262778044,
-      "learning_rate": 0.000199996949083609,
-      "loss": 46.0,
-      "step": 32531
-    },
-    {
-      "epoch": 2.4872985836343826,
-      "grad_norm": 0.0025960856582969427,
-      "learning_rate": 0.00019999694889597964,
-      "loss": 46.0,
-      "step": 32532
-    },
-    {
-      "epoch": 2.4873750406177724,
-      "grad_norm": 0.003075724234804511,
-      "learning_rate": 0.00019999694870834444,
-      "loss": 46.0,
-      "step": 32533
-    },
-    {
-      "epoch": 2.487451497601162,
-      "grad_norm": 0.0011970649939030409,
-      "learning_rate": 0.00019999694852070352,
-      "loss": 46.0,
-      "step": 32534
-    },
-    {
-      "epoch": 2.487527954584552,
-      "grad_norm": 0.0022439074236899614,
-      "learning_rate": 0.00019999694833305683,
-      "loss": 46.0,
-      "step": 32535
-    },
-    {
-      "epoch": 2.4876044115679417,
-      "grad_norm": 0.0016282013384625316,
-      "learning_rate": 0.00019999694814540434,
-      "loss": 46.0,
-      "step": 32536
-    },
-    {
-      "epoch": 2.4876808685513314,
-      "grad_norm": 0.0013592778705060482,
-      "learning_rate": 0.0001999969479577461,
-      "loss": 46.0,
-      "step": 32537
-    },
-    {
-      "epoch": 2.487757325534721,
-      "grad_norm": 0.0005730963894166052,
-      "learning_rate": 0.0001999969477700821,
-      "loss": 46.0,
-      "step": 32538
-    },
-    {
-      "epoch": 2.487833782518111,
-      "grad_norm": 0.0007011077832430601,
-      "learning_rate": 0.00019999694758241233,
-      "loss": 46.0,
-      "step": 32539
-    },
-    {
-      "epoch": 2.4879102395015007,
-      "grad_norm": 0.0015100507298484445,
-      "learning_rate": 0.00019999694739473677,
-      "loss": 46.0,
-      "step": 32540
-    },
-    {
-      "epoch": 2.48798669648489,
-      "grad_norm": 0.0009014597744680941,
-      "learning_rate": 0.00019999694720705544,
-      "loss": 46.0,
-      "step": 32541
-    },
-    {
-      "epoch": 2.4880631534682798,
-      "grad_norm": 0.001240005251020193,
-      "learning_rate": 0.00019999694701936836,
-      "loss": 46.0,
-      "step": 32542
-    },
-    {
-      "epoch": 2.4881396104516695,
-      "grad_norm": 0.001145404065027833,
-      "learning_rate": 0.00019999694683167548,
-      "loss": 46.0,
-      "step": 32543
-    },
-    {
-      "epoch": 2.4882160674350593,
-      "grad_norm": 0.0012710869777947664,
-      "learning_rate": 0.00019999694664397688,
-      "loss": 46.0,
-      "step": 32544
-    },
-    {
-      "epoch": 2.488292524418449,
-      "grad_norm": 0.0004915580502711236,
-      "learning_rate": 0.0001999969464562725,
-      "loss": 46.0,
-      "step": 32545
-    },
-    {
-      "epoch": 2.488368981401839,
-      "grad_norm": 0.0012591304257512093,
-      "learning_rate": 0.00019999694626856231,
-      "loss": 46.0,
-      "step": 32546
-    },
-    {
-      "epoch": 2.4884454383852286,
-      "grad_norm": 0.0006453980458900332,
-      "learning_rate": 0.00019999694608084637,
-      "loss": 46.0,
-      "step": 32547
-    },
-    {
-      "epoch": 2.4885218953686183,
-      "grad_norm": 0.0010155484778806567,
-      "learning_rate": 0.00019999694589312468,
-      "loss": 46.0,
-      "step": 32548
-    },
-    {
-      "epoch": 2.488598352352008,
-      "grad_norm": 0.0007393939304165542,
-      "learning_rate": 0.0001999969457053972,
-      "loss": 46.0,
-      "step": 32549
-    },
-    {
-      "epoch": 2.488674809335398,
-      "grad_norm": 0.0004635357763618231,
-      "learning_rate": 0.00019999694551766397,
-      "loss": 46.0,
-      "step": 32550
-    },
-    {
-      "epoch": 2.488751266318787,
-      "grad_norm": 0.002516088541597128,
-      "learning_rate": 0.00019999694532992494,
-      "loss": 46.0,
-      "step": 32551
-    },
-    {
-      "epoch": 2.488827723302177,
-      "grad_norm": 0.0008175940602086484,
-      "learning_rate": 0.00019999694514218018,
-      "loss": 46.0,
-      "step": 32552
-    },
-    {
-      "epoch": 2.4889041802855667,
-      "grad_norm": 0.0004407418891787529,
-      "learning_rate": 0.00019999694495442962,
-      "loss": 46.0,
-      "step": 32553
-    },
-    {
-      "epoch": 2.4889806372689565,
-      "grad_norm": 0.00406007980927825,
-      "learning_rate": 0.0001999969447666733,
-      "loss": 46.0,
-      "step": 32554
-    },
-    {
-      "epoch": 2.4890570942523462,
-      "grad_norm": 0.0010890637058764696,
-      "learning_rate": 0.0001999969445789112,
-      "loss": 46.0,
-      "step": 32555
-    },
-    {
-      "epoch": 2.489133551235736,
-      "grad_norm": 0.0007092013256624341,
-      "learning_rate": 0.00019999694439114336,
-      "loss": 46.0,
-      "step": 32556
-    },
-    {
-      "epoch": 2.4892100082191257,
-      "grad_norm": 0.0005649402737617493,
-      "learning_rate": 0.00019999694420336973,
-      "loss": 46.0,
-      "step": 32557
-    },
-    {
-      "epoch": 2.4892864652025155,
-      "grad_norm": 0.0011889153392985463,
-      "learning_rate": 0.00019999694401559034,
-      "loss": 46.0,
-      "step": 32558
-    },
-    {
-      "epoch": 2.4893629221859053,
-      "grad_norm": 0.00298025063239038,
-      "learning_rate": 0.00019999694382780516,
-      "loss": 46.0,
-      "step": 32559
-    },
-    {
-      "epoch": 2.489439379169295,
-      "grad_norm": 0.0008088881731964648,
-      "learning_rate": 0.00019999694364001422,
-      "loss": 46.0,
-      "step": 32560
-    },
-    {
-      "epoch": 2.489515836152685,
-      "grad_norm": 0.0007073302404023707,
-      "learning_rate": 0.00019999694345221753,
-      "loss": 46.0,
-      "step": 32561
-    },
-    {
-      "epoch": 2.4895922931360746,
-      "grad_norm": 0.001739811385050416,
-      "learning_rate": 0.00019999694326441504,
-      "loss": 46.0,
-      "step": 32562
-    },
-    {
-      "epoch": 2.489668750119464,
-      "grad_norm": 0.0009057508432306349,
-      "learning_rate": 0.0001999969430766068,
-      "loss": 46.0,
-      "step": 32563
-    },
-    {
-      "epoch": 2.4897452071028536,
-      "grad_norm": 0.00041170476470142603,
-      "learning_rate": 0.00019999694288879281,
-      "loss": 46.0,
-      "step": 32564
-    },
-    {
-      "epoch": 2.4898216640862434,
-      "grad_norm": 0.001177989412099123,
-      "learning_rate": 0.000199996942700973,
-      "loss": 46.0,
-      "step": 32565
-    },
-    {
-      "epoch": 2.489898121069633,
-      "grad_norm": 0.0005341498763300478,
-      "learning_rate": 0.00019999694251314747,
-      "loss": 46.0,
-      "step": 32566
-    },
-    {
-      "epoch": 2.489974578053023,
-      "grad_norm": 0.0008835952612571418,
-      "learning_rate": 0.00019999694232531614,
-      "loss": 46.0,
-      "step": 32567
-    },
-    {
-      "epoch": 2.4900510350364127,
-      "grad_norm": 0.004304313100874424,
-      "learning_rate": 0.00019999694213747904,
-      "loss": 46.0,
-      "step": 32568
-    },
-    {
-      "epoch": 2.4901274920198024,
-      "grad_norm": 0.0006942129693925381,
-      "learning_rate": 0.00019999694194963618,
-      "loss": 46.0,
-      "step": 32569
-    },
-    {
-      "epoch": 2.490203949003192,
-      "grad_norm": 0.0007404296775348485,
-      "learning_rate": 0.00019999694176178756,
-      "loss": 46.0,
-      "step": 32570
-    },
-    {
-      "epoch": 2.490280405986582,
-      "grad_norm": 0.001132192905060947,
-      "learning_rate": 0.00019999694157393316,
-      "loss": 46.0,
-      "step": 32571
-    },
-    {
-      "epoch": 2.4903568629699713,
-      "grad_norm": 0.0007190343458205462,
-      "learning_rate": 0.000199996941386073,
-      "loss": 46.0,
-      "step": 32572
-    },
-    {
-      "epoch": 2.490433319953361,
-      "grad_norm": 0.0016687260940670967,
-      "learning_rate": 0.00019999694119820707,
-      "loss": 46.0,
-      "step": 32573
-    },
-    {
-      "epoch": 2.490509776936751,
-      "grad_norm": 0.001053826417773962,
-      "learning_rate": 0.00019999694101033538,
-      "loss": 46.0,
-      "step": 32574
-    },
-    {
-      "epoch": 2.4905862339201406,
-      "grad_norm": 0.004366938024759293,
-      "learning_rate": 0.0001999969408224579,
-      "loss": 46.0,
-      "step": 32575
-    },
-    {
-      "epoch": 2.4906626909035303,
-      "grad_norm": 0.0012544711353257298,
-      "learning_rate": 0.00019999694063457465,
-      "loss": 46.0,
-      "step": 32576
-    },
-    {
-      "epoch": 2.49073914788692,
-      "grad_norm": 0.00032960408134385943,
-      "learning_rate": 0.00019999694044668562,
-      "loss": 46.0,
-      "step": 32577
-    },
-    {
-      "epoch": 2.49081560487031,
-      "grad_norm": 0.001531166024506092,
-      "learning_rate": 0.00019999694025879086,
-      "loss": 46.0,
-      "step": 32578
-    },
-    {
-      "epoch": 2.4908920618536996,
-      "grad_norm": 0.000752673193346709,
-      "learning_rate": 0.0001999969400708903,
-      "loss": 46.0,
-      "step": 32579
-    },
-    {
-      "epoch": 2.4909685188370894,
-      "grad_norm": 0.0005099460249766707,
-      "learning_rate": 0.00019999693988298397,
-      "loss": 46.0,
-      "step": 32580
-    },
-    {
-      "epoch": 2.491044975820479,
-      "grad_norm": 0.0019086366519331932,
-      "learning_rate": 0.00019999693969507186,
-      "loss": 46.0,
-      "step": 32581
-    },
-    {
-      "epoch": 2.491121432803869,
-      "grad_norm": 0.004595336504280567,
-      "learning_rate": 0.00019999693950715401,
-      "loss": 46.0,
-      "step": 32582
-    },
-    {
-      "epoch": 2.4911978897872586,
-      "grad_norm": 0.0014849234139546752,
-      "learning_rate": 0.0001999969393192304,
-      "loss": 46.0,
-      "step": 32583
-    },
-    {
-      "epoch": 2.4912743467706484,
-      "grad_norm": 0.0008845271659083664,
-      "learning_rate": 0.000199996939131301,
-      "loss": 46.0,
-      "step": 32584
-    },
-    {
-      "epoch": 2.4913508037540377,
-      "grad_norm": 0.0021731401793658733,
-      "learning_rate": 0.00019999693894336585,
-      "loss": 46.0,
-      "step": 32585
-    },
-    {
-      "epoch": 2.4914272607374275,
-      "grad_norm": 0.0014632310485467315,
-      "learning_rate": 0.00019999693875542488,
-      "loss": 46.0,
-      "step": 32586
-    },
-    {
-      "epoch": 2.4915037177208172,
-      "grad_norm": 0.001202008337713778,
-      "learning_rate": 0.0001999969385674782,
-      "loss": 46.0,
-      "step": 32587
-    },
-    {
-      "epoch": 2.491580174704207,
-      "grad_norm": 0.0009073092369362712,
-      "learning_rate": 0.00019999693837952572,
-      "loss": 46.0,
-      "step": 32588
-    },
-    {
-      "epoch": 2.4916566316875968,
-      "grad_norm": 0.0008802267257124186,
-      "learning_rate": 0.00019999693819156749,
-      "loss": 46.0,
-      "step": 32589
-    },
-    {
-      "epoch": 2.4917330886709865,
-      "grad_norm": 0.0012836562236770988,
-      "learning_rate": 0.00019999693800360347,
-      "loss": 46.0,
-      "step": 32590
-    },
-    {
-      "epoch": 2.4918095456543763,
-      "grad_norm": 0.001667350996285677,
-      "learning_rate": 0.00019999693781563366,
-      "loss": 46.0,
-      "step": 32591
-    },
-    {
-      "epoch": 2.491886002637766,
-      "grad_norm": 0.0005499919061549008,
-      "learning_rate": 0.00019999693762765813,
-      "loss": 46.0,
-      "step": 32592
-    },
-    {
-      "epoch": 2.491962459621156,
-      "grad_norm": 0.0009524726192466915,
-      "learning_rate": 0.0001999969374396768,
-      "loss": 46.0,
-      "step": 32593
-    },
-    {
-      "epoch": 2.492038916604545,
-      "grad_norm": 0.0006169872940517962,
-      "learning_rate": 0.0001999969372516897,
-      "loss": 46.0,
-      "step": 32594
-    },
-    {
-      "epoch": 2.492115373587935,
-      "grad_norm": 0.0004173332708887756,
-      "learning_rate": 0.00019999693706369685,
-      "loss": 46.0,
-      "step": 32595
-    },
-    {
-      "epoch": 2.4921918305713247,
-      "grad_norm": 0.0013197311200201511,
-      "learning_rate": 0.00019999693687569822,
-      "loss": 46.0,
-      "step": 32596
-    },
-    {
-      "epoch": 2.4922682875547144,
-      "grad_norm": 0.0007887695683166385,
-      "learning_rate": 0.00019999693668769383,
-      "loss": 46.0,
-      "step": 32597
-    },
-    {
-      "epoch": 2.492344744538104,
-      "grad_norm": 0.0005454887286759913,
-      "learning_rate": 0.00019999693649968366,
-      "loss": 46.0,
-      "step": 32598
-    },
-    {
-      "epoch": 2.492421201521494,
-      "grad_norm": 0.0009504213812761009,
-      "learning_rate": 0.00019999693631166774,
-      "loss": 46.0,
-      "step": 32599
-    },
-    {
-      "epoch": 2.4924976585048837,
-      "grad_norm": 0.0014154567616060376,
-      "learning_rate": 0.000199996936123646,
-      "loss": 46.0,
-      "step": 32600
-    },
-    {
-      "epoch": 2.4925741154882735,
-      "grad_norm": 0.0009172839345410466,
-      "learning_rate": 0.00019999693593561856,
-      "loss": 46.0,
-      "step": 32601
-    },
-    {
-      "epoch": 2.492650572471663,
-      "grad_norm": 0.0005338436458259821,
-      "learning_rate": 0.0001999969357475853,
-      "loss": 46.0,
-      "step": 32602
-    },
-    {
-      "epoch": 2.492727029455053,
-      "grad_norm": 0.00151747465133667,
-      "learning_rate": 0.00019999693555954628,
-      "loss": 46.0,
-      "step": 32603
-    },
-    {
-      "epoch": 2.4928034864384427,
-      "grad_norm": 0.0008318493491970003,
-      "learning_rate": 0.00019999693537150147,
-      "loss": 46.0,
-      "step": 32604
-    },
-    {
-      "epoch": 2.4928799434218325,
-      "grad_norm": 0.0011670246021822095,
-      "learning_rate": 0.00019999693518345092,
-      "loss": 46.0,
-      "step": 32605
-    },
-    {
-      "epoch": 2.4929564004052223,
-      "grad_norm": 0.0017090964829549193,
-      "learning_rate": 0.0001999969349953946,
-      "loss": 46.0,
-      "step": 32606
-    },
-    {
-      "epoch": 2.4930328573886116,
-      "grad_norm": 0.0006046075141057372,
-      "learning_rate": 0.00019999693480733254,
-      "loss": 46.0,
-      "step": 32607
-    },
-    {
-      "epoch": 2.4931093143720013,
-      "grad_norm": 0.00048030464677140117,
-      "learning_rate": 0.00019999693461926469,
-      "loss": 46.0,
-      "step": 32608
-    },
-    {
-      "epoch": 2.493185771355391,
-      "grad_norm": 0.0007360557792708278,
-      "learning_rate": 0.00019999693443119104,
-      "loss": 46.0,
-      "step": 32609
-    },
-    {
-      "epoch": 2.493262228338781,
-      "grad_norm": 0.0014631547965109348,
-      "learning_rate": 0.00019999693424311167,
-      "loss": 46.0,
-      "step": 32610
-    },
-    {
-      "epoch": 2.4933386853221706,
-      "grad_norm": 0.0015953963156789541,
-      "learning_rate": 0.00019999693405502647,
-      "loss": 46.0,
-      "step": 32611
-    },
-    {
-      "epoch": 2.4934151423055604,
-      "grad_norm": 0.002187523990869522,
-      "learning_rate": 0.00019999693386693555,
-      "loss": 46.0,
-      "step": 32612
-    },
-    {
-      "epoch": 2.49349159928895,
-      "grad_norm": 0.002458705101162195,
-      "learning_rate": 0.00019999693367883883,
-      "loss": 46.0,
-      "step": 32613
-    },
-    {
-      "epoch": 2.49356805627234,
-      "grad_norm": 0.0013839320745319128,
-      "learning_rate": 0.00019999693349073634,
-      "loss": 46.0,
-      "step": 32614
-    },
-    {
-      "epoch": 2.4936445132557297,
-      "grad_norm": 0.0014073910424485803,
-      "learning_rate": 0.0001999969333026281,
-      "loss": 46.0,
-      "step": 32615
-    },
-    {
-      "epoch": 2.493720970239119,
-      "grad_norm": 0.0007907609106041491,
-      "learning_rate": 0.0001999969331145141,
-      "loss": 46.0,
-      "step": 32616
-    },
-    {
-      "epoch": 2.4937974272225087,
-      "grad_norm": 0.0008132322109304368,
-      "learning_rate": 0.0001999969329263943,
-      "loss": 46.0,
-      "step": 32617
-    },
-    {
-      "epoch": 2.4938738842058985,
-      "grad_norm": 0.0012784870341420174,
-      "learning_rate": 0.00019999693273826876,
-      "loss": 46.0,
-      "step": 32618
-    },
-    {
-      "epoch": 2.4939503411892883,
-      "grad_norm": 0.0011030759196728468,
-      "learning_rate": 0.00019999693255013745,
-      "loss": 46.0,
-      "step": 32619
-    },
-    {
-      "epoch": 2.494026798172678,
-      "grad_norm": 0.0020263739861547947,
-      "learning_rate": 0.00019999693236200035,
-      "loss": 46.0,
-      "step": 32620
-    },
-    {
-      "epoch": 2.494103255156068,
-      "grad_norm": 0.000965129816904664,
-      "learning_rate": 0.00019999693217385747,
-      "loss": 46.0,
-      "step": 32621
-    },
-    {
-      "epoch": 2.4941797121394575,
-      "grad_norm": 0.0012597328750416636,
-      "learning_rate": 0.00019999693198570885,
-      "loss": 46.0,
-      "step": 32622
-    },
-    {
-      "epoch": 2.4942561691228473,
-      "grad_norm": 0.0016392285469919443,
-      "learning_rate": 0.00019999693179755445,
-      "loss": 46.0,
-      "step": 32623
-    },
-    {
-      "epoch": 2.494332626106237,
-      "grad_norm": 0.001721691689454019,
-      "learning_rate": 0.00019999693160939428,
-      "loss": 46.0,
-      "step": 32624
-    },
-    {
-      "epoch": 2.494409083089627,
-      "grad_norm": 0.004060937557369471,
-      "learning_rate": 0.00019999693142122834,
-      "loss": 46.0,
-      "step": 32625
-    },
-    {
-      "epoch": 2.4944855400730166,
-      "grad_norm": 0.001772783463820815,
-      "learning_rate": 0.00019999693123305665,
-      "loss": 46.0,
-      "step": 32626
-    },
-    {
-      "epoch": 2.4945619970564064,
-      "grad_norm": 0.000491643266286701,
-      "learning_rate": 0.00019999693104487916,
-      "loss": 46.0,
-      "step": 32627
-    },
-    {
-      "epoch": 2.4946384540397957,
-      "grad_norm": 0.0010192712070420384,
-      "learning_rate": 0.00019999693085669592,
-      "loss": 46.0,
-      "step": 32628
-    },
-    {
-      "epoch": 2.4947149110231854,
-      "grad_norm": 0.0007798741571605206,
-      "learning_rate": 0.00019999693066850691,
-      "loss": 46.0,
-      "step": 32629
-    },
-    {
-      "epoch": 2.494791368006575,
-      "grad_norm": 0.0015332680195569992,
-      "learning_rate": 0.0001999969304803121,
-      "loss": 46.0,
-      "step": 32630
-    },
-    {
-      "epoch": 2.494867824989965,
-      "grad_norm": 0.0057282885536551476,
-      "learning_rate": 0.00019999693029211155,
-      "loss": 46.0,
-      "step": 32631
-    },
-    {
-      "epoch": 2.4949442819733547,
-      "grad_norm": 0.0012404846493154764,
-      "learning_rate": 0.00019999693010390525,
-      "loss": 46.0,
-      "step": 32632
-    },
-    {
-      "epoch": 2.4950207389567445,
-      "grad_norm": 0.0008654487901367247,
-      "learning_rate": 0.00019999692991569314,
-      "loss": 46.0,
-      "step": 32633
-    },
-    {
-      "epoch": 2.4950971959401342,
-      "grad_norm": 0.0031990830320864916,
-      "learning_rate": 0.0001999969297274753,
-      "loss": 46.0,
-      "step": 32634
-    },
-    {
-      "epoch": 2.495173652923524,
-      "grad_norm": 0.0012202790239825845,
-      "learning_rate": 0.00019999692953925164,
-      "loss": 46.0,
-      "step": 32635
-    },
-    {
-      "epoch": 2.4952501099069138,
-      "grad_norm": 0.0010098565835505724,
-      "learning_rate": 0.00019999692935102225,
-      "loss": 46.0,
-      "step": 32636
-    },
-    {
-      "epoch": 2.4953265668903035,
-      "grad_norm": 0.0014236140996217728,
-      "learning_rate": 0.0001999969291627871,
-      "loss": 46.0,
-      "step": 32637
-    },
-    {
-      "epoch": 2.495403023873693,
-      "grad_norm": 0.003221197286620736,
-      "learning_rate": 0.00019999692897454613,
-      "loss": 46.0,
-      "step": 32638
-    },
-    {
-      "epoch": 2.4954794808570826,
-      "grad_norm": 0.0005447784205898643,
-      "learning_rate": 0.00019999692878629944,
-      "loss": 46.0,
-      "step": 32639
-    },
-    {
-      "epoch": 2.4955559378404724,
-      "grad_norm": 0.004033575765788555,
-      "learning_rate": 0.00019999692859804698,
-      "loss": 46.0,
-      "step": 32640
-    },
-    {
-      "epoch": 2.495632394823862,
-      "grad_norm": 0.0007976160850375891,
-      "learning_rate": 0.00019999692840978872,
-      "loss": 46.0,
-      "step": 32641
-    },
-    {
-      "epoch": 2.495708851807252,
-      "grad_norm": 0.0008567469194531441,
-      "learning_rate": 0.00019999692822152468,
-      "loss": 46.0,
-      "step": 32642
-    },
-    {
-      "epoch": 2.4957853087906416,
-      "grad_norm": 0.0055723777040839195,
-      "learning_rate": 0.00019999692803325493,
-      "loss": 46.0,
-      "step": 32643
-    },
-    {
-      "epoch": 2.4958617657740314,
-      "grad_norm": 0.0006186652462929487,
-      "learning_rate": 0.00019999692784497937,
-      "loss": 46.0,
-      "step": 32644
-    },
-    {
-      "epoch": 2.495938222757421,
-      "grad_norm": 0.001370106590911746,
-      "learning_rate": 0.000199996927656698,
-      "loss": 46.0,
-      "step": 32645
-    },
-    {
-      "epoch": 2.496014679740811,
-      "grad_norm": 0.0014947515446692705,
-      "learning_rate": 0.00019999692746841094,
-      "loss": 46.0,
-      "step": 32646
-    },
-    {
-      "epoch": 2.4960911367242007,
-      "grad_norm": 0.0015226037940010428,
-      "learning_rate": 0.00019999692728011806,
-      "loss": 46.0,
-      "step": 32647
-    },
-    {
-      "epoch": 2.4961675937075904,
-      "grad_norm": 0.0005861383397132158,
-      "learning_rate": 0.00019999692709181944,
-      "loss": 46.0,
-      "step": 32648
-    },
-    {
-      "epoch": 2.49624405069098,
-      "grad_norm": 0.0015473462408408523,
-      "learning_rate": 0.00019999692690351504,
-      "loss": 46.0,
-      "step": 32649
-    },
-    {
-      "epoch": 2.4963205076743695,
-      "grad_norm": 0.0011263686465099454,
-      "learning_rate": 0.00019999692671520487,
-      "loss": 46.0,
-      "step": 32650
-    },
-    {
-      "epoch": 2.4963969646577593,
-      "grad_norm": 0.0026451812591403723,
-      "learning_rate": 0.00019999692652688893,
-      "loss": 46.0,
-      "step": 32651
-    },
-    {
-      "epoch": 2.496473421641149,
-      "grad_norm": 0.001112994970753789,
-      "learning_rate": 0.00019999692633856721,
-      "loss": 46.0,
-      "step": 32652
-    },
-    {
-      "epoch": 2.496549878624539,
-      "grad_norm": 0.0016106105176731944,
-      "learning_rate": 0.00019999692615023975,
-      "loss": 46.0,
-      "step": 32653
-    },
-    {
-      "epoch": 2.4966263356079286,
-      "grad_norm": 0.0025343927554786205,
-      "learning_rate": 0.00019999692596190652,
-      "loss": 46.0,
-      "step": 32654
-    },
-    {
-      "epoch": 2.4967027925913183,
-      "grad_norm": 0.0008390550501644611,
-      "learning_rate": 0.00019999692577356748,
-      "loss": 46.0,
-      "step": 32655
-    },
-    {
-      "epoch": 2.496779249574708,
-      "grad_norm": 0.0007332876557484269,
-      "learning_rate": 0.0001999969255852227,
-      "loss": 46.0,
-      "step": 32656
-    },
-    {
-      "epoch": 2.496855706558098,
-      "grad_norm": 0.0015012803487479687,
-      "learning_rate": 0.00019999692539687214,
-      "loss": 46.0,
-      "step": 32657
-    },
-    {
-      "epoch": 2.4969321635414876,
-      "grad_norm": 0.0027170462999492884,
-      "learning_rate": 0.00019999692520851584,
-      "loss": 46.0,
-      "step": 32658
-    },
-    {
-      "epoch": 2.4970086205248774,
-      "grad_norm": 0.0007027161191217601,
-      "learning_rate": 0.00019999692502015374,
-      "loss": 46.0,
-      "step": 32659
-    },
-    {
-      "epoch": 2.4970850775082667,
-      "grad_norm": 0.0015824147267267108,
-      "learning_rate": 0.00019999692483178586,
-      "loss": 46.0,
-      "step": 32660
-    },
-    {
-      "epoch": 2.4971615344916565,
-      "grad_norm": 0.0011959223775193095,
-      "learning_rate": 0.0001999969246434122,
-      "loss": 46.0,
-      "step": 32661
-    },
-    {
-      "epoch": 2.497237991475046,
-      "grad_norm": 0.002277828985825181,
-      "learning_rate": 0.00019999692445503282,
-      "loss": 46.0,
-      "step": 32662
-    },
-    {
-      "epoch": 2.497314448458436,
-      "grad_norm": 0.00031184719409793615,
-      "learning_rate": 0.00019999692426664765,
-      "loss": 46.0,
-      "step": 32663
-    },
-    {
-      "epoch": 2.4973909054418257,
-      "grad_norm": 0.0008964571752585471,
-      "learning_rate": 0.00019999692407825673,
-      "loss": 46.0,
-      "step": 32664
-    },
-    {
-      "epoch": 2.4974673624252155,
-      "grad_norm": 0.0010842683259397745,
-      "learning_rate": 0.00019999692388986002,
-      "loss": 46.0,
-      "step": 32665
-    },
-    {
-      "epoch": 2.4975438194086053,
-      "grad_norm": 0.0008160679135471582,
-      "learning_rate": 0.00019999692370145753,
-      "loss": 46.0,
-      "step": 32666
-    },
-    {
-      "epoch": 2.497620276391995,
-      "grad_norm": 0.0011588375782594085,
-      "learning_rate": 0.0001999969235130493,
-      "loss": 46.0,
-      "step": 32667
-    },
-    {
-      "epoch": 2.497696733375385,
-      "grad_norm": 0.002093835035338998,
-      "learning_rate": 0.00019999692332463526,
-      "loss": 46.0,
-      "step": 32668
-    },
-    {
-      "epoch": 2.4977731903587745,
-      "grad_norm": 0.0007744574686512351,
-      "learning_rate": 0.00019999692313621547,
-      "loss": 46.0,
-      "step": 32669
-    },
-    {
-      "epoch": 2.4978496473421643,
-      "grad_norm": 0.0005367117119021714,
-      "learning_rate": 0.00019999692294778992,
-      "loss": 46.0,
-      "step": 32670
-    },
-    {
-      "epoch": 2.497926104325554,
-      "grad_norm": 0.0008444421691820025,
-      "learning_rate": 0.0001999969227593586,
-      "loss": 46.0,
-      "step": 32671
-    },
-    {
-      "epoch": 2.4980025613089434,
-      "grad_norm": 0.0024097352288663387,
-      "learning_rate": 0.00019999692257092151,
-      "loss": 46.0,
-      "step": 32672
-    },
-    {
-      "epoch": 2.498079018292333,
-      "grad_norm": 0.0011364661622792482,
-      "learning_rate": 0.00019999692238247864,
-      "loss": 46.0,
-      "step": 32673
-    },
-    {
-      "epoch": 2.498155475275723,
-      "grad_norm": 0.000714063469786197,
-      "learning_rate": 0.00019999692219403002,
-      "loss": 46.0,
-      "step": 32674
-    },
-    {
-      "epoch": 2.4982319322591127,
-      "grad_norm": 0.0016102782683447003,
-      "learning_rate": 0.0001999969220055756,
-      "loss": 46.0,
-      "step": 32675
-    },
-    {
-      "epoch": 2.4983083892425024,
-      "grad_norm": 0.0009185677045024931,
-      "learning_rate": 0.00019999692181711543,
-      "loss": 46.0,
-      "step": 32676
-    },
-    {
-      "epoch": 2.498384846225892,
-      "grad_norm": 0.0014042760012671351,
-      "learning_rate": 0.00019999692162864948,
-      "loss": 46.0,
-      "step": 32677
-    },
-    {
-      "epoch": 2.498461303209282,
-      "grad_norm": 0.0019407031359151006,
-      "learning_rate": 0.0001999969214401778,
-      "loss": 46.0,
-      "step": 32678
-    },
-    {
-      "epoch": 2.4985377601926717,
-      "grad_norm": 0.0013697692193090916,
-      "learning_rate": 0.0001999969212517003,
-      "loss": 46.0,
-      "step": 32679
-    },
-    {
-      "epoch": 2.4986142171760615,
-      "grad_norm": 0.0006164151127450168,
-      "learning_rate": 0.00019999692106321707,
-      "loss": 46.0,
-      "step": 32680
-    },
-    {
-      "epoch": 2.4986906741594512,
-      "grad_norm": 0.000803289411123842,
-      "learning_rate": 0.00019999692087472806,
-      "loss": 46.0,
-      "step": 32681
-    },
-    {
-      "epoch": 2.4987671311428405,
-      "grad_norm": 0.004597480408847332,
-      "learning_rate": 0.00019999692068623325,
-      "loss": 46.0,
-      "step": 32682
-    },
-    {
-      "epoch": 2.4988435881262303,
-      "grad_norm": 0.000799823144916445,
-      "learning_rate": 0.0001999969204977327,
-      "loss": 46.0,
-      "step": 32683
-    },
-    {
-      "epoch": 2.49892004510962,
-      "grad_norm": 0.0007131488528102636,
-      "learning_rate": 0.00019999692030922637,
-      "loss": 46.0,
-      "step": 32684
-    },
-    {
-      "epoch": 2.49899650209301,
-      "grad_norm": 0.001024338649585843,
-      "learning_rate": 0.00019999692012071427,
-      "loss": 46.0,
-      "step": 32685
-    },
-    {
-      "epoch": 2.4990729590763996,
-      "grad_norm": 0.0010273678926751018,
-      "learning_rate": 0.0001999969199321964,
-      "loss": 46.0,
-      "step": 32686
-    },
-    {
-      "epoch": 2.4991494160597894,
-      "grad_norm": 0.0014682613546028733,
-      "learning_rate": 0.00019999691974367277,
-      "loss": 46.0,
-      "step": 32687
-    },
-    {
-      "epoch": 2.499225873043179,
-      "grad_norm": 0.002199523849412799,
-      "learning_rate": 0.0001999969195551434,
-      "loss": 46.0,
-      "step": 32688
-    },
-    {
-      "epoch": 2.499302330026569,
-      "grad_norm": 0.0007699707639403641,
-      "learning_rate": 0.0001999969193666082,
-      "loss": 46.0,
-      "step": 32689
-    },
-    {
-      "epoch": 2.4993787870099586,
-      "grad_norm": 0.0017689234809949994,
-      "learning_rate": 0.00019999691917806727,
-      "loss": 46.0,
-      "step": 32690
-    },
-    {
-      "epoch": 2.4994552439933484,
-      "grad_norm": 0.002967545296996832,
-      "learning_rate": 0.00019999691898952055,
-      "loss": 46.0,
-      "step": 32691
-    },
-    {
-      "epoch": 2.499531700976738,
-      "grad_norm": 0.005779649131000042,
-      "learning_rate": 0.00019999691880096806,
-      "loss": 46.0,
-      "step": 32692
-    },
-    {
-      "epoch": 2.499608157960128,
-      "grad_norm": 0.0022865410428494215,
-      "learning_rate": 0.0001999969186124098,
-      "loss": 46.0,
-      "step": 32693
-    },
-    {
-      "epoch": 2.4996846149435172,
-      "grad_norm": 0.00222966936416924,
-      "learning_rate": 0.0001999969184238458,
-      "loss": 46.0,
-      "step": 32694
-    },
-    {
-      "epoch": 2.499761071926907,
-      "grad_norm": 0.0006263068644329906,
-      "learning_rate": 0.000199996918235276,
-      "loss": 46.0,
-      "step": 32695
-    },
-    {
-      "epoch": 2.4998375289102968,
-      "grad_norm": 0.0014327605022117496,
-      "learning_rate": 0.00019999691804670046,
-      "loss": 46.0,
-      "step": 32696
-    },
-    {
-      "epoch": 2.4999139858936865,
-      "grad_norm": 0.00161940383259207,
-      "learning_rate": 0.00019999691785811913,
-      "loss": 46.0,
-      "step": 32697
-    },
-    {
-      "epoch": 2.4999904428770763,
-      "grad_norm": 0.0005888966261409223,
-      "learning_rate": 0.00019999691766953203,
-      "loss": 46.0,
-      "step": 32698
-    },
-    {
-      "epoch": 2.500066899860466,
-      "grad_norm": 0.0008472443441860378,
-      "learning_rate": 0.00019999691748093915,
-      "loss": 46.0,
-      "step": 32699
-    },
-    {
-      "epoch": 2.500143356843856,
-      "grad_norm": 0.0009773269994184375,
-      "learning_rate": 0.00019999691729234056,
-      "loss": 46.0,
-      "step": 32700
-    },
-    {
-      "epoch": 2.5002198138272456,
-      "grad_norm": 0.0011044393759220839,
-      "learning_rate": 0.00019999691710373613,
-      "loss": 46.0,
-      "step": 32701
-    },
-    {
-      "epoch": 2.5002962708106353,
-      "grad_norm": 0.007905993610620499,
-      "learning_rate": 0.00019999691691512597,
-      "loss": 46.0,
-      "step": 32702
-    },
-    {
-      "epoch": 2.5003727277940246,
-      "grad_norm": 0.0008292117854580283,
-      "learning_rate": 0.00019999691672651002,
-      "loss": 46.0,
-      "step": 32703
-    },
-    {
-      "epoch": 2.5004491847774144,
-      "grad_norm": 0.0011388702550902963,
-      "learning_rate": 0.0001999969165378883,
-      "loss": 46.0,
-      "step": 32704
-    },
-    {
-      "epoch": 2.500525641760804,
-      "grad_norm": 0.0009121305192820728,
-      "learning_rate": 0.00019999691634926082,
-      "loss": 46.0,
-      "step": 32705
-    },
-    {
-      "epoch": 2.500602098744194,
-      "grad_norm": 0.0013917204923927784,
-      "learning_rate": 0.00019999691616062756,
-      "loss": 46.0,
-      "step": 32706
-    },
-    {
-      "epoch": 2.5006785557275837,
-      "grad_norm": 0.0006432253867387772,
-      "learning_rate": 0.00019999691597198855,
-      "loss": 46.0,
-      "step": 32707
-    },
-    {
-      "epoch": 2.5007550127109734,
-      "grad_norm": 0.0015068117063492537,
-      "learning_rate": 0.00019999691578334377,
-      "loss": 46.0,
-      "step": 32708
-    },
-    {
-      "epoch": 2.500831469694363,
-      "grad_norm": 0.0011101285926997662,
-      "learning_rate": 0.00019999691559469322,
-      "loss": 46.0,
-      "step": 32709
-    },
-    {
-      "epoch": 2.500907926677753,
-      "grad_norm": 0.0012562782503664494,
-      "learning_rate": 0.0001999969154060369,
-      "loss": 46.0,
-      "step": 32710
-    },
-    {
-      "epoch": 2.5009843836611427,
-      "grad_norm": 0.0008389346767216921,
-      "learning_rate": 0.00019999691521737481,
-      "loss": 46.0,
-      "step": 32711
-    },
-    {
-      "epoch": 2.5010608406445325,
-      "grad_norm": 0.0009495936101302505,
-      "learning_rate": 0.0001999969150287069,
-      "loss": 46.0,
-      "step": 32712
-    },
-    {
-      "epoch": 2.5011372976279223,
-      "grad_norm": 0.0013499533524736762,
-      "learning_rate": 0.00019999691484003332,
-      "loss": 46.0,
-      "step": 32713
-    },
-    {
-      "epoch": 2.501213754611312,
-      "grad_norm": 0.0017085482832044363,
-      "learning_rate": 0.0001999969146513539,
-      "loss": 46.0,
-      "step": 32714
-    },
-    {
-      "epoch": 2.5012902115947018,
-      "grad_norm": 0.0028621528763324022,
-      "learning_rate": 0.00019999691446266873,
-      "loss": 46.0,
-      "step": 32715
-    },
-    {
-      "epoch": 2.501366668578091,
-      "grad_norm": 0.0010731735965237021,
-      "learning_rate": 0.0001999969142739778,
-      "loss": 46.0,
-      "step": 32716
-    },
-    {
-      "epoch": 2.501443125561481,
-      "grad_norm": 0.0013197783846408129,
-      "learning_rate": 0.00019999691408528105,
-      "loss": 46.0,
-      "step": 32717
-    },
-    {
-      "epoch": 2.5015195825448706,
-      "grad_norm": 0.004666207358241081,
-      "learning_rate": 0.00019999691389657859,
-      "loss": 46.0,
-      "step": 32718
-    },
-    {
-      "epoch": 2.5015960395282604,
-      "grad_norm": 0.00262489658780396,
-      "learning_rate": 0.00019999691370787032,
-      "loss": 46.0,
-      "step": 32719
-    },
-    {
-      "epoch": 2.50167249651165,
-      "grad_norm": 0.002234694315120578,
-      "learning_rate": 0.00019999691351915632,
-      "loss": 46.0,
-      "step": 32720
-    },
-    {
-      "epoch": 2.50174895349504,
-      "grad_norm": 0.0006830698112025857,
-      "learning_rate": 0.00019999691333043654,
-      "loss": 46.0,
-      "step": 32721
-    },
-    {
-      "epoch": 2.5018254104784297,
-      "grad_norm": 0.0013971508014947176,
-      "learning_rate": 0.00019999691314171098,
-      "loss": 46.0,
-      "step": 32722
-    },
-    {
-      "epoch": 2.5019018674618194,
-      "grad_norm": 0.0012611669953912497,
-      "learning_rate": 0.00019999691295297965,
-      "loss": 46.0,
-      "step": 32723
-    },
-    {
-      "epoch": 2.501978324445209,
-      "grad_norm": 0.0026633082889020443,
-      "learning_rate": 0.00019999691276424255,
-      "loss": 46.0,
-      "step": 32724
-    },
-    {
-      "epoch": 2.5020547814285985,
-      "grad_norm": 0.0008328151889145374,
-      "learning_rate": 0.00019999691257549968,
-      "loss": 46.0,
-      "step": 32725
-    },
-    {
-      "epoch": 2.5021312384119883,
-      "grad_norm": 0.000989908934570849,
-      "learning_rate": 0.00019999691238675106,
-      "loss": 46.0,
-      "step": 32726
-    },
-    {
-      "epoch": 2.502207695395378,
-      "grad_norm": 0.0012619729386642575,
-      "learning_rate": 0.00019999691219799664,
-      "loss": 46.0,
-      "step": 32727
-    },
-    {
-      "epoch": 2.502284152378768,
-      "grad_norm": 0.0007630998734384775,
-      "learning_rate": 0.00019999691200923647,
-      "loss": 46.0,
-      "step": 32728
-    },
-    {
-      "epoch": 2.5023606093621575,
-      "grad_norm": 0.0009400617564097047,
-      "learning_rate": 0.00019999691182047053,
-      "loss": 46.0,
-      "step": 32729
-    },
-    {
-      "epoch": 2.5024370663455473,
-      "grad_norm": 0.0006512033869512379,
-      "learning_rate": 0.00019999691163169881,
-      "loss": 46.0,
-      "step": 32730
-    },
-    {
-      "epoch": 2.502513523328937,
-      "grad_norm": 0.0071093435399234295,
-      "learning_rate": 0.00019999691144292135,
-      "loss": 46.0,
-      "step": 32731
-    },
-    {
-      "epoch": 2.502589980312327,
-      "grad_norm": 0.0006232858868315816,
-      "learning_rate": 0.0001999969112541381,
-      "loss": 46.0,
-      "step": 32732
-    },
-    {
-      "epoch": 2.5026664372957166,
-      "grad_norm": 0.0005956319510005414,
-      "learning_rate": 0.00019999691106534906,
-      "loss": 46.0,
-      "step": 32733
-    },
-    {
-      "epoch": 2.5027428942791063,
-      "grad_norm": 0.0031116027384996414,
-      "learning_rate": 0.00019999691087655425,
-      "loss": 46.0,
-      "step": 32734
-    },
-    {
-      "epoch": 2.502819351262496,
-      "grad_norm": 0.0009393964428454638,
-      "learning_rate": 0.00019999691068775372,
-      "loss": 46.0,
-      "step": 32735
-    },
-    {
-      "epoch": 2.502895808245886,
-      "grad_norm": 0.0032348413951694965,
-      "learning_rate": 0.00019999691049894737,
-      "loss": 46.0,
-      "step": 32736
-    },
-    {
-      "epoch": 2.5029722652292756,
-      "grad_norm": 0.0026711774989962578,
-      "learning_rate": 0.0001999969103101353,
-      "loss": 46.0,
-      "step": 32737
-    },
-    {
-      "epoch": 2.503048722212665,
-      "grad_norm": 0.0005530026974156499,
-      "learning_rate": 0.0001999969101213174,
-      "loss": 46.0,
-      "step": 32738
-    },
-    {
-      "epoch": 2.5031251791960547,
-      "grad_norm": 0.004038301762193441,
-      "learning_rate": 0.00019999690993249377,
-      "loss": 46.0,
-      "step": 32739
-    },
-    {
-      "epoch": 2.5032016361794445,
-      "grad_norm": 0.0007219588151201606,
-      "learning_rate": 0.00019999690974366438,
-      "loss": 46.0,
-      "step": 32740
-    },
-    {
-      "epoch": 2.5032780931628342,
-      "grad_norm": 0.0010287873446941376,
-      "learning_rate": 0.00019999690955482921,
-      "loss": 46.0,
-      "step": 32741
-    },
-    {
-      "epoch": 2.503354550146224,
-      "grad_norm": 0.0012785352300852537,
-      "learning_rate": 0.00019999690936598827,
-      "loss": 46.0,
-      "step": 32742
-    },
-    {
-      "epoch": 2.5034310071296138,
-      "grad_norm": 0.001279329531826079,
-      "learning_rate": 0.00019999690917714156,
-      "loss": 46.0,
-      "step": 32743
-    },
-    {
-      "epoch": 2.5035074641130035,
-      "grad_norm": 0.0006859968416392803,
-      "learning_rate": 0.00019999690898828907,
-      "loss": 46.0,
-      "step": 32744
-    },
-    {
-      "epoch": 2.5035839210963933,
-      "grad_norm": 0.0009535603458061814,
-      "learning_rate": 0.0001999969087994308,
-      "loss": 46.0,
-      "step": 32745
-    },
-    {
-      "epoch": 2.5036603780797826,
-      "grad_norm": 0.0008990428759716451,
-      "learning_rate": 0.00019999690861056678,
-      "loss": 46.0,
-      "step": 32746
-    },
-    {
-      "epoch": 2.5037368350631724,
-      "grad_norm": 0.0030643814243376255,
-      "learning_rate": 0.000199996908421697,
-      "loss": 46.0,
-      "step": 32747
-    },
-    {
-      "epoch": 2.503813292046562,
-      "grad_norm": 0.0015444988384842873,
-      "learning_rate": 0.00019999690823282144,
-      "loss": 46.0,
-      "step": 32748
-    },
-    {
-      "epoch": 2.503889749029952,
-      "grad_norm": 0.001566438120789826,
-      "learning_rate": 0.00019999690804394012,
-      "loss": 46.0,
-      "step": 32749
-    },
-    {
-      "epoch": 2.5039662060133416,
-      "grad_norm": 0.0034345362801104784,
-      "learning_rate": 0.00019999690785505302,
-      "loss": 46.0,
-      "step": 32750
-    },
-    {
-      "epoch": 2.5040426629967314,
-      "grad_norm": 0.001355501706711948,
-      "learning_rate": 0.00019999690766616014,
-      "loss": 46.0,
-      "step": 32751
-    },
-    {
-      "epoch": 2.504119119980121,
-      "grad_norm": 0.0006056506535969675,
-      "learning_rate": 0.00019999690747726152,
-      "loss": 46.0,
-      "step": 32752
-    },
-    {
-      "epoch": 2.504195576963511,
-      "grad_norm": 0.0035692111123353243,
-      "learning_rate": 0.0001999969072883571,
-      "loss": 46.0,
-      "step": 32753
-    },
-    {
-      "epoch": 2.5042720339469007,
-      "grad_norm": 0.00087547063594684,
-      "learning_rate": 0.00019999690709944694,
-      "loss": 46.0,
-      "step": 32754
-    },
-    {
-      "epoch": 2.5043484909302904,
-      "grad_norm": 0.0014224721817299724,
-      "learning_rate": 0.000199996906910531,
-      "loss": 46.0,
-      "step": 32755
-    },
-    {
-      "epoch": 2.50442494791368,
-      "grad_norm": 0.0005808738642372191,
-      "learning_rate": 0.00019999690672160928,
-      "loss": 46.0,
-      "step": 32756
-    },
-    {
-      "epoch": 2.50450140489707,
-      "grad_norm": 0.0012215047609061003,
-      "learning_rate": 0.0001999969065326818,
-      "loss": 46.0,
-      "step": 32757
-    },
-    {
-      "epoch": 2.5045778618804597,
-      "grad_norm": 0.00176725210621953,
-      "learning_rate": 0.00019999690634374853,
-      "loss": 46.0,
-      "step": 32758
-    },
-    {
-      "epoch": 2.5046543188638495,
-      "grad_norm": 0.0017734557623043656,
-      "learning_rate": 0.00019999690615480953,
-      "loss": 46.0,
-      "step": 32759
-    },
-    {
-      "epoch": 2.504730775847239,
-      "grad_norm": 0.0005930705810897052,
-      "learning_rate": 0.00019999690596586472,
-      "loss": 46.0,
-      "step": 32760
-    },
-    {
-      "epoch": 2.5048072328306286,
-      "grad_norm": 0.002758664544671774,
-      "learning_rate": 0.00019999690577691417,
-      "loss": 46.0,
-      "step": 32761
-    },
-    {
-      "epoch": 2.5048836898140183,
-      "grad_norm": 0.0008418118814006448,
-      "learning_rate": 0.00019999690558795784,
-      "loss": 46.0,
-      "step": 32762
-    },
-    {
-      "epoch": 2.504960146797408,
-      "grad_norm": 0.002822203328832984,
-      "learning_rate": 0.00019999690539899577,
-      "loss": 46.0,
-      "step": 32763
-    },
-    {
-      "epoch": 2.505036603780798,
-      "grad_norm": 0.0014254904817789793,
-      "learning_rate": 0.00019999690521002787,
-      "loss": 46.0,
-      "step": 32764
-    },
-    {
-      "epoch": 2.5051130607641876,
-      "grad_norm": 0.0014400569489225745,
-      "learning_rate": 0.00019999690502105425,
-      "loss": 46.0,
-      "step": 32765
-    },
-    {
-      "epoch": 2.5051895177475774,
-      "grad_norm": 0.0014554229564964771,
-      "learning_rate": 0.00019999690483207485,
-      "loss": 46.0,
-      "step": 32766
-    },
-    {
-      "epoch": 2.505265974730967,
-      "grad_norm": 0.0006737028597854078,
-      "learning_rate": 0.00019999690464308966,
-      "loss": 46.0,
-      "step": 32767
-    },
-    {
-      "epoch": 2.5053424317143564,
-      "grad_norm": 0.000716458132956177,
-      "learning_rate": 0.00019999690445409872,
-      "loss": 46.0,
-      "step": 32768
-    },
-    {
-      "epoch": 2.505418888697746,
-      "grad_norm": 0.000928579771425575,
-      "learning_rate": 0.000199996904265102,
-      "loss": 46.0,
-      "step": 32769
-    },
-    {
-      "epoch": 2.505495345681136,
-      "grad_norm": 0.0034551192075014114,
-      "learning_rate": 0.00019999690407609952,
-      "loss": 46.0,
-      "step": 32770
-    },
-    {
-      "epoch": 2.5055718026645257,
-      "grad_norm": 0.0009016767144203186,
-      "learning_rate": 0.00019999690388709126,
-      "loss": 46.0,
-      "step": 32771
-    },
-    {
-      "epoch": 2.5056482596479155,
-      "grad_norm": 0.0013303476152941585,
-      "learning_rate": 0.00019999690369807725,
-      "loss": 46.0,
-      "step": 32772
-    },
-    {
-      "epoch": 2.5057247166313052,
-      "grad_norm": 0.0006002801819704473,
-      "learning_rate": 0.00019999690350905747,
-      "loss": 46.0,
-      "step": 32773
-    },
-    {
-      "epoch": 2.505801173614695,
-      "grad_norm": 0.0036748128477483988,
-      "learning_rate": 0.0001999969033200319,
-      "loss": 46.0,
-      "step": 32774
-    },
-    {
-      "epoch": 2.5058776305980848,
-      "grad_norm": 0.0011790312128141522,
-      "learning_rate": 0.00019999690313100054,
-      "loss": 46.0,
-      "step": 32775
-    },
-    {
-      "epoch": 2.5059540875814745,
-      "grad_norm": 0.0011653387919068336,
-      "learning_rate": 0.00019999690294196344,
-      "loss": 46.0,
-      "step": 32776
-    },
-    {
-      "epoch": 2.5060305445648643,
-      "grad_norm": 0.0011330617126077414,
-      "learning_rate": 0.00019999690275292057,
-      "loss": 46.0,
-      "step": 32777
-    },
-    {
-      "epoch": 2.506107001548254,
-      "grad_norm": 0.0009041562443599105,
-      "learning_rate": 0.00019999690256387195,
-      "loss": 46.0,
-      "step": 32778
-    },
-    {
-      "epoch": 2.506183458531644,
-      "grad_norm": 0.0019856621511280537,
-      "learning_rate": 0.00019999690237481756,
-      "loss": 46.0,
-      "step": 32779
-    },
-    {
-      "epoch": 2.5062599155150336,
-      "grad_norm": 0.0006077262223698199,
-      "learning_rate": 0.00019999690218575736,
-      "loss": 46.0,
-      "step": 32780
-    },
-    {
-      "epoch": 2.5063363724984233,
-      "grad_norm": 0.0012735208729282022,
-      "learning_rate": 0.00019999690199669142,
-      "loss": 46.0,
-      "step": 32781
-    },
-    {
-      "epoch": 2.5064128294818127,
-      "grad_norm": 0.005194342229515314,
-      "learning_rate": 0.0001999969018076197,
-      "loss": 46.0,
-      "step": 32782
-    },
-    {
-      "epoch": 2.5064892864652024,
-      "grad_norm": 0.0016863117925822735,
-      "learning_rate": 0.00019999690161854222,
-      "loss": 46.0,
-      "step": 32783
-    },
-    {
-      "epoch": 2.506565743448592,
-      "grad_norm": 0.005246866028755903,
-      "learning_rate": 0.00019999690142945896,
-      "loss": 46.0,
-      "step": 32784
-    },
-    {
-      "epoch": 2.506642200431982,
-      "grad_norm": 0.0007456231978721917,
-      "learning_rate": 0.00019999690124036996,
-      "loss": 46.0,
-      "step": 32785
-    },
-    {
-      "epoch": 2.5067186574153717,
-      "grad_norm": 0.0008065645815804601,
-      "learning_rate": 0.00019999690105127518,
-      "loss": 46.0,
-      "step": 32786
-    },
-    {
-      "epoch": 2.5067951143987615,
-      "grad_norm": 0.0010334714315831661,
-      "learning_rate": 0.00019999690086217463,
-      "loss": 46.0,
-      "step": 32787
-    },
-    {
-      "epoch": 2.506871571382151,
-      "grad_norm": 0.0019579287618398666,
-      "learning_rate": 0.0001999969006730683,
-      "loss": 46.0,
-      "step": 32788
-    },
-    {
-      "epoch": 2.506948028365541,
-      "grad_norm": 0.001493968302384019,
-      "learning_rate": 0.00019999690048395617,
-      "loss": 46.0,
-      "step": 32789
-    },
-    {
-      "epoch": 2.5070244853489303,
-      "grad_norm": 0.003920863848179579,
-      "learning_rate": 0.0001999969002948383,
-      "loss": 46.0,
-      "step": 32790
-    },
-    {
-      "epoch": 2.50710094233232,
-      "grad_norm": 0.0006900986190885305,
-      "learning_rate": 0.00019999690010571465,
-      "loss": 46.0,
-      "step": 32791
-    },
-    {
-      "epoch": 2.50717739931571,
-      "grad_norm": 0.003912812098860741,
-      "learning_rate": 0.00019999689991658526,
-      "loss": 46.0,
-      "step": 32792
-    },
-    {
-      "epoch": 2.5072538562990996,
-      "grad_norm": 0.0009005999891087413,
-      "learning_rate": 0.00019999689972745007,
-      "loss": 46.0,
-      "step": 32793
-    },
-    {
-      "epoch": 2.5073303132824893,
-      "grad_norm": 0.0008407142595387995,
-      "learning_rate": 0.00019999689953830913,
-      "loss": 46.0,
-      "step": 32794
-    },
-    {
-      "epoch": 2.507406770265879,
-      "grad_norm": 0.0013869135873392224,
-      "learning_rate": 0.00019999689934916242,
-      "loss": 46.0,
-      "step": 32795
-    },
-    {
-      "epoch": 2.507483227249269,
-      "grad_norm": 0.0012822578428313136,
-      "learning_rate": 0.00019999689916000993,
-      "loss": 46.0,
-      "step": 32796
-    },
-    {
-      "epoch": 2.5075596842326586,
-      "grad_norm": 0.0007940813666209579,
-      "learning_rate": 0.0001999968989708517,
-      "loss": 46.0,
-      "step": 32797
-    },
-    {
-      "epoch": 2.5076361412160484,
-      "grad_norm": 0.001456239726394415,
-      "learning_rate": 0.00019999689878168767,
-      "loss": 46.0,
-      "step": 32798
-    },
-    {
-      "epoch": 2.507712598199438,
-      "grad_norm": 0.0014756184536963701,
-      "learning_rate": 0.00019999689859251786,
-      "loss": 46.0,
-      "step": 32799
-    },
-    {
-      "epoch": 2.507789055182828,
-      "grad_norm": 0.0007517407066188753,
-      "learning_rate": 0.0001999968984033423,
-      "loss": 46.0,
-      "step": 32800
-    },
-    {
-      "epoch": 2.5078655121662177,
-      "grad_norm": 0.0006073712138459086,
-      "learning_rate": 0.00019999689821416098,
-      "loss": 46.0,
-      "step": 32801
-    },
-    {
-      "epoch": 2.5079419691496074,
-      "grad_norm": 0.001129685202613473,
-      "learning_rate": 0.00019999689802497388,
-      "loss": 46.0,
-      "step": 32802
-    },
-    {
-      "epoch": 2.508018426132997,
-      "grad_norm": 0.0008915666840039194,
-      "learning_rate": 0.00019999689783578098,
-      "loss": 46.0,
-      "step": 32803
-    },
-    {
-      "epoch": 2.5080948831163865,
-      "grad_norm": 0.002187796402722597,
-      "learning_rate": 0.0001999968976465824,
-      "loss": 46.0,
-      "step": 32804
-    },
-    {
-      "epoch": 2.5081713400997763,
-      "grad_norm": 0.0031287139281630516,
-      "learning_rate": 0.00019999689745737795,
-      "loss": 46.0,
-      "step": 32805
-    },
-    {
-      "epoch": 2.508247797083166,
-      "grad_norm": 0.0019955255556851625,
-      "learning_rate": 0.0001999968972681678,
-      "loss": 46.0,
-      "step": 32806
-    },
-    {
-      "epoch": 2.508324254066556,
-      "grad_norm": 0.000831499753985554,
-      "learning_rate": 0.00019999689707895184,
-      "loss": 46.0,
-      "step": 32807
-    },
-    {
-      "epoch": 2.5084007110499456,
-      "grad_norm": 0.003255280200392008,
-      "learning_rate": 0.0001999968968897301,
-      "loss": 46.0,
-      "step": 32808
-    },
-    {
-      "epoch": 2.5084771680333353,
-      "grad_norm": 0.0008602631860412657,
-      "learning_rate": 0.00019999689670050264,
-      "loss": 46.0,
-      "step": 32809
-    },
-    {
-      "epoch": 2.508553625016725,
-      "grad_norm": 0.0027964110486209393,
-      "learning_rate": 0.00019999689651126938,
-      "loss": 46.0,
-      "step": 32810
-    },
-    {
-      "epoch": 2.508630082000115,
-      "grad_norm": 0.007330305874347687,
-      "learning_rate": 0.00019999689632203035,
-      "loss": 46.0,
-      "step": 32811
-    },
-    {
-      "epoch": 2.508706538983504,
-      "grad_norm": 0.000710338877979666,
-      "learning_rate": 0.00019999689613278557,
-      "loss": 46.0,
-      "step": 32812
-    },
-    {
-      "epoch": 2.508782995966894,
-      "grad_norm": 0.0018319184891879559,
-      "learning_rate": 0.000199996895943535,
-      "loss": 46.0,
-      "step": 32813
-    },
-    {
-      "epoch": 2.5088594529502837,
-      "grad_norm": 0.0028413746040314436,
-      "learning_rate": 0.00019999689575427867,
-      "loss": 46.0,
-      "step": 32814
-    },
-    {
-      "epoch": 2.5089359099336734,
-      "grad_norm": 0.0018154596909880638,
-      "learning_rate": 0.00019999689556501657,
-      "loss": 46.0,
-      "step": 32815
-    },
-    {
-      "epoch": 2.509012366917063,
-      "grad_norm": 0.0008049430907703936,
-      "learning_rate": 0.0001999968953757487,
-      "loss": 46.0,
-      "step": 32816
-    },
-    {
-      "epoch": 2.509088823900453,
-      "grad_norm": 0.0011200510198250413,
-      "learning_rate": 0.00019999689518647505,
-      "loss": 46.0,
-      "step": 32817
-    },
-    {
-      "epoch": 2.5091652808838427,
-      "grad_norm": 0.0011329521657899022,
-      "learning_rate": 0.00019999689499719566,
-      "loss": 46.0,
-      "step": 32818
-    },
-    {
-      "epoch": 2.5092417378672325,
-      "grad_norm": 0.0006639102357439697,
-      "learning_rate": 0.00019999689480791047,
-      "loss": 46.0,
-      "step": 32819
-    },
-    {
-      "epoch": 2.5093181948506222,
-      "grad_norm": 0.0013859100872650743,
-      "learning_rate": 0.00019999689461861953,
-      "loss": 46.0,
-      "step": 32820
-    },
-    {
-      "epoch": 2.509394651834012,
-      "grad_norm": 0.0009265526896342635,
-      "learning_rate": 0.00019999689442932282,
-      "loss": 46.0,
-      "step": 32821
-    },
-    {
-      "epoch": 2.5094711088174018,
-      "grad_norm": 0.0005473807686939836,
-      "learning_rate": 0.00019999689424002033,
-      "loss": 46.0,
-      "step": 32822
-    },
-    {
-      "epoch": 2.5095475658007915,
-      "grad_norm": 0.0008722553611733019,
-      "learning_rate": 0.00019999689405071207,
-      "loss": 46.0,
-      "step": 32823
-    },
-    {
-      "epoch": 2.5096240227841813,
-      "grad_norm": 0.00034943880746141076,
-      "learning_rate": 0.00019999689386139804,
-      "loss": 46.0,
-      "step": 32824
-    },
-    {
-      "epoch": 2.509700479767571,
-      "grad_norm": 0.0010079960338771343,
-      "learning_rate": 0.00019999689367207826,
-      "loss": 46.0,
-      "step": 32825
-    },
-    {
-      "epoch": 2.5097769367509604,
-      "grad_norm": 0.0007239811238832772,
-      "learning_rate": 0.0001999968934827527,
-      "loss": 46.0,
-      "step": 32826
-    },
-    {
-      "epoch": 2.50985339373435,
-      "grad_norm": 0.000974804803263396,
-      "learning_rate": 0.00019999689329342136,
-      "loss": 46.0,
-      "step": 32827
-    },
-    {
-      "epoch": 2.50992985071774,
-      "grad_norm": 0.00154615449719131,
-      "learning_rate": 0.00019999689310408426,
-      "loss": 46.0,
-      "step": 32828
-    },
-    {
-      "epoch": 2.5100063077011296,
-      "grad_norm": 0.0010489854030311108,
-      "learning_rate": 0.0001999968929147414,
-      "loss": 46.0,
-      "step": 32829
-    },
-    {
-      "epoch": 2.5100827646845194,
-      "grad_norm": 0.0009069915977306664,
-      "learning_rate": 0.00019999689272539274,
-      "loss": 46.0,
-      "step": 32830
-    },
-    {
-      "epoch": 2.510159221667909,
-      "grad_norm": 0.0017528278985992074,
-      "learning_rate": 0.00019999689253603835,
-      "loss": 46.0,
-      "step": 32831
-    },
-    {
-      "epoch": 2.510235678651299,
-      "grad_norm": 0.001177777536213398,
-      "learning_rate": 0.00019999689234667819,
-      "loss": 46.0,
-      "step": 32832
-    },
-    {
-      "epoch": 2.5103121356346887,
-      "grad_norm": 0.0024528945796191692,
-      "learning_rate": 0.00019999689215731222,
-      "loss": 46.0,
-      "step": 32833
-    },
-    {
-      "epoch": 2.510388592618078,
-      "grad_norm": 0.0008305267547257245,
-      "learning_rate": 0.00019999689196794048,
-      "loss": 46.0,
-      "step": 32834
-    },
-    {
-      "epoch": 2.5104650496014678,
-      "grad_norm": 0.001171586220152676,
-      "learning_rate": 0.00019999689177856302,
-      "loss": 46.0,
-      "step": 32835
-    },
-    {
-      "epoch": 2.5105415065848575,
-      "grad_norm": 0.0007078165654093027,
-      "learning_rate": 0.00019999689158917977,
-      "loss": 46.0,
-      "step": 32836
-    },
-    {
-      "epoch": 2.5106179635682473,
-      "grad_norm": 0.0019236498046666384,
-      "learning_rate": 0.00019999689139979073,
-      "loss": 46.0,
-      "step": 32837
-    },
-    {
-      "epoch": 2.510694420551637,
-      "grad_norm": 0.0007407772354781628,
-      "learning_rate": 0.00019999689121039593,
-      "loss": 46.0,
-      "step": 32838
-    },
-    {
-      "epoch": 2.510770877535027,
-      "grad_norm": 0.0016303022857755423,
-      "learning_rate": 0.00019999689102099538,
-      "loss": 46.0,
-      "step": 32839
-    },
-    {
-      "epoch": 2.5108473345184166,
-      "grad_norm": 0.0019795165862888098,
-      "learning_rate": 0.00019999689083158905,
-      "loss": 46.0,
-      "step": 32840
-    },
-    {
-      "epoch": 2.5109237915018063,
-      "grad_norm": 0.001191888004541397,
-      "learning_rate": 0.00019999689064217693,
-      "loss": 46.0,
-      "step": 32841
-    },
-    {
-      "epoch": 2.511000248485196,
-      "grad_norm": 0.0008496284135617316,
-      "learning_rate": 0.00019999689045275905,
-      "loss": 46.0,
-      "step": 32842
-    },
-    {
-      "epoch": 2.511076705468586,
-      "grad_norm": 0.0013293935917317867,
-      "learning_rate": 0.00019999689026333544,
-      "loss": 46.0,
-      "step": 32843
-    },
-    {
-      "epoch": 2.5111531624519756,
-      "grad_norm": 0.010424631647765636,
-      "learning_rate": 0.00019999689007390602,
-      "loss": 46.0,
-      "step": 32844
-    },
-    {
-      "epoch": 2.5112296194353654,
-      "grad_norm": 0.0011201835004612803,
-      "learning_rate": 0.00019999688988447085,
-      "loss": 46.0,
-      "step": 32845
-    },
-    {
-      "epoch": 2.511306076418755,
-      "grad_norm": 0.00044287124183028936,
-      "learning_rate": 0.0001999968896950299,
-      "loss": 46.0,
-      "step": 32846
-    },
-    {
-      "epoch": 2.5113825334021445,
-      "grad_norm": 0.0008032558253034949,
-      "learning_rate": 0.00019999688950558318,
-      "loss": 46.0,
-      "step": 32847
-    },
-    {
-      "epoch": 2.511458990385534,
-      "grad_norm": 0.000719800591468811,
-      "learning_rate": 0.00019999688931613067,
-      "loss": 46.0,
-      "step": 32848
-    },
-    {
-      "epoch": 2.511535447368924,
-      "grad_norm": 0.0009950101375579834,
-      "learning_rate": 0.00019999688912667243,
-      "loss": 46.0,
-      "step": 32849
-    },
-    {
-      "epoch": 2.5116119043523137,
-      "grad_norm": 0.0013402453623712063,
-      "learning_rate": 0.0001999968889372084,
-      "loss": 46.0,
-      "step": 32850
-    },
-    {
-      "epoch": 2.5116883613357035,
-      "grad_norm": 0.001264795777387917,
-      "learning_rate": 0.0001999968887477386,
-      "loss": 46.0,
-      "step": 32851
-    },
-    {
-      "epoch": 2.5117648183190933,
-      "grad_norm": 0.0008573863306082785,
-      "learning_rate": 0.00019999688855826305,
-      "loss": 46.0,
-      "step": 32852
-    },
-    {
-      "epoch": 2.511841275302483,
-      "grad_norm": 0.005295548588037491,
-      "learning_rate": 0.00019999688836878172,
-      "loss": 46.0,
-      "step": 32853
-    },
-    {
-      "epoch": 2.511917732285873,
-      "grad_norm": 0.0005234805284999311,
-      "learning_rate": 0.00019999688817929462,
-      "loss": 46.0,
-      "step": 32854
-    },
-    {
-      "epoch": 2.5119941892692625,
-      "grad_norm": 0.002573030535131693,
-      "learning_rate": 0.00019999688798980173,
-      "loss": 46.0,
-      "step": 32855
-    },
-    {
-      "epoch": 2.512070646252652,
-      "grad_norm": 0.0009861671132966876,
-      "learning_rate": 0.0001999968878003031,
-      "loss": 46.0,
-      "step": 32856
-    },
-    {
-      "epoch": 2.5121471032360416,
-      "grad_norm": 0.0071407221257686615,
-      "learning_rate": 0.0001999968876107987,
-      "loss": 46.0,
-      "step": 32857
-    },
-    {
-      "epoch": 2.5122235602194314,
-      "grad_norm": 0.0008289107936434448,
-      "learning_rate": 0.00019999688742128853,
-      "loss": 46.0,
-      "step": 32858
-    },
-    {
-      "epoch": 2.512300017202821,
-      "grad_norm": 0.0008245704229921103,
-      "learning_rate": 0.00019999688723177256,
-      "loss": 46.0,
-      "step": 32859
-    },
-    {
-      "epoch": 2.512376474186211,
-      "grad_norm": 0.001350258244201541,
-      "learning_rate": 0.00019999688704225082,
-      "loss": 46.0,
-      "step": 32860
-    },
-    {
-      "epoch": 2.5124529311696007,
-      "grad_norm": 0.0007459091139025986,
-      "learning_rate": 0.00019999688685272337,
-      "loss": 46.0,
-      "step": 32861
-    },
-    {
-      "epoch": 2.5125293881529904,
-      "grad_norm": 0.0016938502667471766,
-      "learning_rate": 0.0001999968866631901,
-      "loss": 46.0,
-      "step": 32862
-    },
-    {
-      "epoch": 2.51260584513638,
-      "grad_norm": 0.0015764908166602254,
-      "learning_rate": 0.00019999688647365108,
-      "loss": 46.0,
-      "step": 32863
-    },
-    {
-      "epoch": 2.51268230211977,
-      "grad_norm": 0.0005885313148610294,
-      "learning_rate": 0.00019999688628410627,
-      "loss": 46.0,
-      "step": 32864
-    },
-    {
-      "epoch": 2.5127587591031597,
-      "grad_norm": 0.0015417566755786538,
-      "learning_rate": 0.00019999688609455572,
-      "loss": 46.0,
-      "step": 32865
-    },
-    {
-      "epoch": 2.5128352160865495,
-      "grad_norm": 0.0005860993987880647,
-      "learning_rate": 0.0001999968859049994,
-      "loss": 46.0,
-      "step": 32866
-    },
-    {
-      "epoch": 2.5129116730699392,
-      "grad_norm": 0.0011281523620709777,
-      "learning_rate": 0.0001999968857154373,
-      "loss": 46.0,
-      "step": 32867
-    },
-    {
-      "epoch": 2.512988130053329,
-      "grad_norm": 0.0008245331700891256,
-      "learning_rate": 0.0001999968855258694,
-      "loss": 46.0,
-      "step": 32868
-    },
-    {
-      "epoch": 2.5130645870367183,
-      "grad_norm": 0.0006819698610343039,
-      "learning_rate": 0.00019999688533629576,
-      "loss": 46.0,
-      "step": 32869
-    },
-    {
-      "epoch": 2.513141044020108,
-      "grad_norm": 0.0015945348422974348,
-      "learning_rate": 0.00019999688514671634,
-      "loss": 46.0,
-      "step": 32870
-    },
-    {
-      "epoch": 2.513217501003498,
-      "grad_norm": 0.0012751793256029487,
-      "learning_rate": 0.00019999688495713115,
-      "loss": 46.0,
-      "step": 32871
-    },
-    {
-      "epoch": 2.5132939579868876,
-      "grad_norm": 0.003421960398554802,
-      "learning_rate": 0.00019999688476754024,
-      "loss": 46.0,
-      "step": 32872
-    },
-    {
-      "epoch": 2.5133704149702774,
-      "grad_norm": 0.0015391081105917692,
-      "learning_rate": 0.0001999968845779435,
-      "loss": 46.0,
-      "step": 32873
-    },
-    {
-      "epoch": 2.513446871953667,
-      "grad_norm": 0.0010220227995887399,
-      "learning_rate": 0.00019999688438834104,
-      "loss": 46.0,
-      "step": 32874
-    },
-    {
-      "epoch": 2.513523328937057,
-      "grad_norm": 0.0011553529184311628,
-      "learning_rate": 0.00019999688419873276,
-      "loss": 46.0,
-      "step": 32875
-    },
-    {
-      "epoch": 2.5135997859204466,
-      "grad_norm": 0.0015240157954394817,
-      "learning_rate": 0.00019999688400911873,
-      "loss": 46.0,
-      "step": 32876
-    },
-    {
-      "epoch": 2.513676242903836,
-      "grad_norm": 0.0022138566710054874,
-      "learning_rate": 0.00019999688381949895,
-      "loss": 46.0,
-      "step": 32877
-    },
-    {
-      "epoch": 2.5137526998872257,
-      "grad_norm": 0.0008458830998279154,
-      "learning_rate": 0.0001999968836298734,
-      "loss": 46.0,
-      "step": 32878
-    },
-    {
-      "epoch": 2.5138291568706155,
-      "grad_norm": 0.0003650127036962658,
-      "learning_rate": 0.00019999688344024205,
-      "loss": 46.0,
-      "step": 32879
-    },
-    {
-      "epoch": 2.5139056138540052,
-      "grad_norm": 0.0014310518745332956,
-      "learning_rate": 0.00019999688325060495,
-      "loss": 46.0,
-      "step": 32880
-    },
-    {
-      "epoch": 2.513982070837395,
-      "grad_norm": 0.0009572127601131797,
-      "learning_rate": 0.00019999688306096208,
-      "loss": 46.0,
-      "step": 32881
-    },
-    {
-      "epoch": 2.5140585278207848,
-      "grad_norm": 0.0021009778138250113,
-      "learning_rate": 0.0001999968828713134,
-      "loss": 46.0,
-      "step": 32882
-    },
-    {
-      "epoch": 2.5141349848041745,
-      "grad_norm": 0.0013248994946479797,
-      "learning_rate": 0.00019999688268165902,
-      "loss": 46.0,
-      "step": 32883
-    },
-    {
-      "epoch": 2.5142114417875643,
-      "grad_norm": 0.001694239559583366,
-      "learning_rate": 0.00019999688249199883,
-      "loss": 46.0,
-      "step": 32884
-    },
-    {
-      "epoch": 2.514287898770954,
-      "grad_norm": 0.00220459490083158,
-      "learning_rate": 0.00019999688230233287,
-      "loss": 46.0,
-      "step": 32885
-    },
-    {
-      "epoch": 2.514364355754344,
-      "grad_norm": 0.001286650775000453,
-      "learning_rate": 0.00019999688211266116,
-      "loss": 46.0,
-      "step": 32886
-    },
-    {
-      "epoch": 2.5144408127377336,
-      "grad_norm": 0.0016371086239814758,
-      "learning_rate": 0.00019999688192298367,
-      "loss": 46.0,
-      "step": 32887
-    },
-    {
-      "epoch": 2.5145172697211233,
-      "grad_norm": 0.0019026239169761539,
-      "learning_rate": 0.0001999968817333004,
-      "loss": 46.0,
-      "step": 32888
-    },
-    {
-      "epoch": 2.514593726704513,
-      "grad_norm": 0.0012701016385108232,
-      "learning_rate": 0.00019999688154361138,
-      "loss": 46.0,
-      "step": 32889
-    },
-    {
-      "epoch": 2.514670183687903,
-      "grad_norm": 0.004471160005778074,
-      "learning_rate": 0.00019999688135391658,
-      "loss": 46.0,
-      "step": 32890
-    },
-    {
-      "epoch": 2.514746640671292,
-      "grad_norm": 0.0031129554845392704,
-      "learning_rate": 0.00019999688116421603,
-      "loss": 46.0,
-      "step": 32891
-    },
-    {
-      "epoch": 2.514823097654682,
-      "grad_norm": 0.0012290870072320104,
-      "learning_rate": 0.00019999688097450968,
-      "loss": 46.0,
-      "step": 32892
-    },
-    {
-      "epoch": 2.5148995546380717,
-      "grad_norm": 0.0015509986551478505,
-      "learning_rate": 0.00019999688078479758,
-      "loss": 46.0,
-      "step": 32893
-    },
-    {
-      "epoch": 2.5149760116214614,
-      "grad_norm": 0.0010045382659882307,
-      "learning_rate": 0.00019999688059507974,
-      "loss": 46.0,
-      "step": 32894
-    },
-    {
-      "epoch": 2.515052468604851,
-      "grad_norm": 0.0011538172839209437,
-      "learning_rate": 0.00019999688040535607,
-      "loss": 46.0,
-      "step": 32895
-    },
-    {
-      "epoch": 2.515128925588241,
-      "grad_norm": 0.0009462969610467553,
-      "learning_rate": 0.00019999688021562665,
-      "loss": 46.0,
-      "step": 32896
-    },
-    {
-      "epoch": 2.5152053825716307,
-      "grad_norm": 0.0011038155062124133,
-      "learning_rate": 0.0001999968800258915,
-      "loss": 46.0,
-      "step": 32897
-    },
-    {
-      "epoch": 2.5152818395550205,
-      "grad_norm": 0.000499115907587111,
-      "learning_rate": 0.00019999687983615055,
-      "loss": 46.0,
-      "step": 32898
-    },
-    {
-      "epoch": 2.51535829653841,
-      "grad_norm": 0.001995911356061697,
-      "learning_rate": 0.00019999687964640381,
-      "loss": 46.0,
-      "step": 32899
-    },
-    {
-      "epoch": 2.5154347535217996,
-      "grad_norm": 0.0008694082498550415,
-      "learning_rate": 0.0001999968794566513,
-      "loss": 46.0,
-      "step": 32900
-    },
-    {
-      "epoch": 2.5155112105051893,
-      "grad_norm": 0.0004751631640829146,
-      "learning_rate": 0.00019999687926689305,
-      "loss": 46.0,
-      "step": 32901
-    },
-    {
-      "epoch": 2.515587667488579,
-      "grad_norm": 0.0011406440753489733,
-      "learning_rate": 0.00019999687907712902,
-      "loss": 46.0,
-      "step": 32902
-    },
-    {
-      "epoch": 2.515664124471969,
-      "grad_norm": 0.002283995971083641,
-      "learning_rate": 0.00019999687888735924,
-      "loss": 46.0,
-      "step": 32903
-    },
-    {
-      "epoch": 2.5157405814553586,
-      "grad_norm": 0.0010508904233574867,
-      "learning_rate": 0.00019999687869758366,
-      "loss": 46.0,
-      "step": 32904
-    },
-    {
-      "epoch": 2.5158170384387484,
-      "grad_norm": 0.0009684675605967641,
-      "learning_rate": 0.00019999687850780234,
-      "loss": 46.0,
-      "step": 32905
-    },
-    {
-      "epoch": 2.515893495422138,
-      "grad_norm": 0.0034374427050352097,
-      "learning_rate": 0.00019999687831801524,
-      "loss": 46.0,
-      "step": 32906
-    },
-    {
-      "epoch": 2.515969952405528,
-      "grad_norm": 0.0010148673318326473,
-      "learning_rate": 0.00019999687812822235,
-      "loss": 46.0,
-      "step": 32907
-    },
-    {
-      "epoch": 2.5160464093889177,
-      "grad_norm": 0.0016581009840592742,
-      "learning_rate": 0.00019999687793842373,
-      "loss": 46.0,
-      "step": 32908
-    },
-    {
-      "epoch": 2.5161228663723074,
-      "grad_norm": 0.0014887274010106921,
-      "learning_rate": 0.0001999968777486193,
-      "loss": 46.0,
-      "step": 32909
-    },
-    {
-      "epoch": 2.516199323355697,
-      "grad_norm": 0.0026363113429397345,
-      "learning_rate": 0.00019999687755880912,
-      "loss": 46.0,
-      "step": 32910
-    },
-    {
-      "epoch": 2.516275780339087,
-      "grad_norm": 0.0006123919156379998,
-      "learning_rate": 0.00019999687736899319,
-      "loss": 46.0,
-      "step": 32911
-    },
-    {
-      "epoch": 2.5163522373224767,
-      "grad_norm": 0.0008107998874038458,
-      "learning_rate": 0.00019999687717917145,
-      "loss": 46.0,
-      "step": 32912
-    },
-    {
-      "epoch": 2.516428694305866,
-      "grad_norm": 0.0015831629280000925,
-      "learning_rate": 0.00019999687698934397,
-      "loss": 46.0,
-      "step": 32913
-    },
-    {
-      "epoch": 2.516505151289256,
-      "grad_norm": 0.001789933885447681,
-      "learning_rate": 0.0001999968767995107,
-      "loss": 46.0,
-      "step": 32914
-    },
-    {
-      "epoch": 2.5165816082726455,
-      "grad_norm": 0.0007072822190821171,
-      "learning_rate": 0.00019999687660967168,
-      "loss": 46.0,
-      "step": 32915
-    },
-    {
-      "epoch": 2.5166580652560353,
-      "grad_norm": 0.0008345975656993687,
-      "learning_rate": 0.00019999687641982688,
-      "loss": 46.0,
-      "step": 32916
-    },
-    {
-      "epoch": 2.516734522239425,
-      "grad_norm": 0.0014782397774979472,
-      "learning_rate": 0.0001999968762299763,
-      "loss": 46.0,
-      "step": 32917
-    },
-    {
-      "epoch": 2.516810979222815,
-      "grad_norm": 0.0030025290325284004,
-      "learning_rate": 0.00019999687604011998,
-      "loss": 46.0,
-      "step": 32918
-    },
-    {
-      "epoch": 2.5168874362062046,
-      "grad_norm": 0.00047987012658268213,
-      "learning_rate": 0.00019999687585025785,
-      "loss": 46.0,
-      "step": 32919
-    },
-    {
-      "epoch": 2.5169638931895943,
-      "grad_norm": 0.0013627614825963974,
-      "learning_rate": 0.00019999687566039,
-      "loss": 46.0,
-      "step": 32920
-    },
-    {
-      "epoch": 2.5170403501729837,
-      "grad_norm": 0.0011875573545694351,
-      "learning_rate": 0.00019999687547051634,
-      "loss": 46.0,
-      "step": 32921
-    },
-    {
-      "epoch": 2.5171168071563734,
-      "grad_norm": 0.0007395835127681494,
-      "learning_rate": 0.00019999687528063692,
-      "loss": 46.0,
-      "step": 32922
-    },
-    {
-      "epoch": 2.517193264139763,
-      "grad_norm": 0.0012649290729314089,
-      "learning_rate": 0.00019999687509075176,
-      "loss": 46.0,
-      "step": 32923
-    },
-    {
-      "epoch": 2.517269721123153,
-      "grad_norm": 0.0015310108428820968,
-      "learning_rate": 0.0001999968749008608,
-      "loss": 46.0,
-      "step": 32924
-    },
-    {
-      "epoch": 2.5173461781065427,
-      "grad_norm": 0.001608656020835042,
-      "learning_rate": 0.0001999968747109641,
-      "loss": 46.0,
-      "step": 32925
-    },
-    {
-      "epoch": 2.5174226350899325,
-      "grad_norm": 0.0010443860664963722,
-      "learning_rate": 0.0001999968745210616,
-      "loss": 46.0,
-      "step": 32926
-    },
-    {
-      "epoch": 2.5174990920733222,
-      "grad_norm": 0.0007467984687536955,
-      "learning_rate": 0.00019999687433115332,
-      "loss": 46.0,
-      "step": 32927
-    },
-    {
-      "epoch": 2.517575549056712,
-      "grad_norm": 0.0017736062873154879,
-      "learning_rate": 0.0001999968741412393,
-      "loss": 46.0,
-      "step": 32928
-    },
-    {
-      "epoch": 2.5176520060401018,
-      "grad_norm": 0.0007338566356338561,
-      "learning_rate": 0.00019999687395131952,
-      "loss": 46.0,
-      "step": 32929
-    },
-    {
-      "epoch": 2.5177284630234915,
-      "grad_norm": 0.0023565846495330334,
-      "learning_rate": 0.00019999687376139394,
-      "loss": 46.0,
-      "step": 32930
-    },
-    {
-      "epoch": 2.5178049200068813,
-      "grad_norm": 0.0006290346500463784,
-      "learning_rate": 0.0001999968735714626,
-      "loss": 46.0,
-      "step": 32931
-    },
-    {
-      "epoch": 2.517881376990271,
-      "grad_norm": 0.0005563198938034475,
-      "learning_rate": 0.0001999968733815255,
-      "loss": 46.0,
-      "step": 32932
-    },
-    {
-      "epoch": 2.517957833973661,
-      "grad_norm": 0.0013433565618470311,
-      "learning_rate": 0.00019999687319158263,
-      "loss": 46.0,
-      "step": 32933
-    },
-    {
-      "epoch": 2.5180342909570506,
-      "grad_norm": 0.0016104771057143807,
-      "learning_rate": 0.00019999687300163398,
-      "loss": 46.0,
-      "step": 32934
-    },
-    {
-      "epoch": 2.51811074794044,
-      "grad_norm": 0.0004023423243779689,
-      "learning_rate": 0.00019999687281167957,
-      "loss": 46.0,
-      "step": 32935
-    },
-    {
-      "epoch": 2.5181872049238296,
-      "grad_norm": 0.0013931648572906852,
-      "learning_rate": 0.00019999687262171938,
-      "loss": 46.0,
-      "step": 32936
-    },
-    {
-      "epoch": 2.5182636619072194,
-      "grad_norm": 0.001002289354801178,
-      "learning_rate": 0.00019999687243175341,
-      "loss": 46.0,
-      "step": 32937
-    },
-    {
-      "epoch": 2.518340118890609,
-      "grad_norm": 0.0022146524861454964,
-      "learning_rate": 0.0001999968722417817,
-      "loss": 46.0,
-      "step": 32938
-    },
-    {
-      "epoch": 2.518416575873999,
-      "grad_norm": 0.0013069614069536328,
-      "learning_rate": 0.00019999687205180422,
-      "loss": 46.0,
-      "step": 32939
-    },
-    {
-      "epoch": 2.5184930328573887,
-      "grad_norm": 0.003414312144741416,
-      "learning_rate": 0.00019999687186182094,
-      "loss": 46.0,
-      "step": 32940
-    },
-    {
-      "epoch": 2.5185694898407784,
-      "grad_norm": 0.0006989172543399036,
-      "learning_rate": 0.0001999968716718319,
-      "loss": 46.0,
-      "step": 32941
-    },
-    {
-      "epoch": 2.518645946824168,
-      "grad_norm": 0.0010113092139363289,
-      "learning_rate": 0.00019999687148183713,
-      "loss": 46.0,
-      "step": 32942
-    },
-    {
-      "epoch": 2.5187224038075575,
-      "grad_norm": 0.0007022622157819569,
-      "learning_rate": 0.00019999687129183659,
-      "loss": 46.0,
-      "step": 32943
-    },
-    {
-      "epoch": 2.5187988607909473,
-      "grad_norm": 0.000595089397393167,
-      "learning_rate": 0.0001999968711018302,
-      "loss": 46.0,
-      "step": 32944
-    },
-    {
-      "epoch": 2.518875317774337,
-      "grad_norm": 0.0014249193482100964,
-      "learning_rate": 0.00019999687091181811,
-      "loss": 46.0,
-      "step": 32945
-    },
-    {
-      "epoch": 2.518951774757727,
-      "grad_norm": 0.0017278653103858232,
-      "learning_rate": 0.00019999687072180024,
-      "loss": 46.0,
-      "step": 32946
-    },
-    {
-      "epoch": 2.5190282317411166,
-      "grad_norm": 0.0006695809424854815,
-      "learning_rate": 0.0001999968705317766,
-      "loss": 46.0,
-      "step": 32947
-    },
-    {
-      "epoch": 2.5191046887245063,
-      "grad_norm": 0.003549568820744753,
-      "learning_rate": 0.0001999968703417472,
-      "loss": 46.0,
-      "step": 32948
-    },
-    {
-      "epoch": 2.519181145707896,
-      "grad_norm": 0.0015840554842725396,
-      "learning_rate": 0.000199996870151712,
-      "loss": 46.0,
-      "step": 32949
-    },
-    {
-      "epoch": 2.519257602691286,
-      "grad_norm": 0.0017992067150771618,
-      "learning_rate": 0.00019999686996167103,
-      "loss": 46.0,
-      "step": 32950
-    },
-    {
-      "epoch": 2.5193340596746756,
-      "grad_norm": 0.0005472741322591901,
-      "learning_rate": 0.00019999686977162433,
-      "loss": 46.0,
-      "step": 32951
-    },
-    {
-      "epoch": 2.5194105166580654,
-      "grad_norm": 0.0027882393915206194,
-      "learning_rate": 0.00019999686958157184,
-      "loss": 46.0,
-      "step": 32952
-    },
-    {
-      "epoch": 2.519486973641455,
-      "grad_norm": 0.003878476098179817,
-      "learning_rate": 0.00019999686939151356,
-      "loss": 46.0,
-      "step": 32953
-    },
-    {
-      "epoch": 2.519563430624845,
-      "grad_norm": 0.0015768310986459255,
-      "learning_rate": 0.00019999686920144956,
-      "loss": 46.0,
-      "step": 32954
-    },
-    {
-      "epoch": 2.5196398876082347,
-      "grad_norm": 0.0009607269312255085,
-      "learning_rate": 0.00019999686901137973,
-      "loss": 46.0,
-      "step": 32955
-    },
-    {
-      "epoch": 2.5197163445916244,
-      "grad_norm": 0.0030656075105071068,
-      "learning_rate": 0.00019999686882130418,
-      "loss": 46.0,
-      "step": 32956
-    },
-    {
-      "epoch": 2.5197928015750137,
-      "grad_norm": 0.0005757392500527203,
-      "learning_rate": 0.00019999686863122283,
-      "loss": 46.0,
-      "step": 32957
-    },
-    {
-      "epoch": 2.5198692585584035,
-      "grad_norm": 0.0006189522100612521,
-      "learning_rate": 0.00019999686844113574,
-      "loss": 46.0,
-      "step": 32958
-    },
-    {
-      "epoch": 2.5199457155417933,
-      "grad_norm": 0.001912234234623611,
-      "learning_rate": 0.00019999686825104287,
-      "loss": 46.0,
-      "step": 32959
-    },
-    {
-      "epoch": 2.520022172525183,
-      "grad_norm": 0.003201256040483713,
-      "learning_rate": 0.00019999686806094423,
-      "loss": 46.0,
-      "step": 32960
-    },
-    {
-      "epoch": 2.5200986295085728,
-      "grad_norm": 0.0025721059646457434,
-      "learning_rate": 0.0001999968678708398,
-      "loss": 46.0,
-      "step": 32961
-    },
-    {
-      "epoch": 2.5201750864919625,
-      "grad_norm": 0.0004677971010096371,
-      "learning_rate": 0.0001999968676807296,
-      "loss": 46.0,
-      "step": 32962
-    },
-    {
-      "epoch": 2.5202515434753523,
-      "grad_norm": 0.0006928106886334717,
-      "learning_rate": 0.00019999686749061366,
-      "loss": 46.0,
-      "step": 32963
-    },
-    {
-      "epoch": 2.520328000458742,
-      "grad_norm": 0.0036883829161524773,
-      "learning_rate": 0.00019999686730049195,
-      "loss": 46.0,
-      "step": 32964
-    },
-    {
-      "epoch": 2.5204044574421314,
-      "grad_norm": 0.0023962792474776506,
-      "learning_rate": 0.00019999686711036441,
-      "loss": 46.0,
-      "step": 32965
-    },
-    {
-      "epoch": 2.520480914425521,
-      "grad_norm": 0.0009388107573613524,
-      "learning_rate": 0.00019999686692023119,
-      "loss": 46.0,
-      "step": 32966
-    },
-    {
-      "epoch": 2.520557371408911,
-      "grad_norm": 0.0010167048312723637,
-      "learning_rate": 0.00019999686673009216,
-      "loss": 46.0,
-      "step": 32967
-    },
-    {
-      "epoch": 2.5206338283923007,
-      "grad_norm": 0.0011221676832064986,
-      "learning_rate": 0.00019999686653994733,
-      "loss": 46.0,
-      "step": 32968
-    },
-    {
-      "epoch": 2.5207102853756904,
-      "grad_norm": 0.00042411015601828694,
-      "learning_rate": 0.00019999686634979678,
-      "loss": 46.0,
-      "step": 32969
-    },
-    {
-      "epoch": 2.52078674235908,
-      "grad_norm": 0.0015562771586701274,
-      "learning_rate": 0.00019999686615964043,
-      "loss": 46.0,
-      "step": 32970
-    },
-    {
-      "epoch": 2.52086319934247,
-      "grad_norm": 0.0009822442661970854,
-      "learning_rate": 0.00019999686596947834,
-      "loss": 46.0,
-      "step": 32971
-    },
-    {
-      "epoch": 2.5209396563258597,
-      "grad_norm": 0.0010432639392092824,
-      "learning_rate": 0.00019999686577931044,
-      "loss": 46.0,
-      "step": 32972
-    },
-    {
-      "epoch": 2.5210161133092495,
-      "grad_norm": 0.002705514198169112,
-      "learning_rate": 0.00019999686558913683,
-      "loss": 46.0,
-      "step": 32973
-    },
-    {
-      "epoch": 2.5210925702926392,
-      "grad_norm": 0.0012569775572046638,
-      "learning_rate": 0.0001999968653989574,
-      "loss": 46.0,
-      "step": 32974
-    },
-    {
-      "epoch": 2.521169027276029,
-      "grad_norm": 0.004672076087445021,
-      "learning_rate": 0.00019999686520877222,
-      "loss": 46.0,
-      "step": 32975
-    },
-    {
-      "epoch": 2.5212454842594187,
-      "grad_norm": 0.0038774562999606133,
-      "learning_rate": 0.00019999686501858126,
-      "loss": 46.0,
-      "step": 32976
-    },
-    {
-      "epoch": 2.5213219412428085,
-      "grad_norm": 0.002517918823286891,
-      "learning_rate": 0.00019999686482838453,
-      "loss": 46.0,
-      "step": 32977
-    },
-    {
-      "epoch": 2.521398398226198,
-      "grad_norm": 0.0015174434520304203,
-      "learning_rate": 0.00019999686463818207,
-      "loss": 46.0,
-      "step": 32978
-    },
-    {
-      "epoch": 2.5214748552095876,
-      "grad_norm": 0.0007237580721266568,
-      "learning_rate": 0.0001999968644479738,
-      "loss": 46.0,
-      "step": 32979
-    },
-    {
-      "epoch": 2.5215513121929773,
-      "grad_norm": 0.0016403637127950788,
-      "learning_rate": 0.00019999686425775973,
-      "loss": 46.0,
-      "step": 32980
-    },
-    {
-      "epoch": 2.521627769176367,
-      "grad_norm": 0.0016672651981934905,
-      "learning_rate": 0.00019999686406753996,
-      "loss": 46.0,
-      "step": 32981
-    },
-    {
-      "epoch": 2.521704226159757,
-      "grad_norm": 0.0010922338115051389,
-      "learning_rate": 0.00019999686387731438,
-      "loss": 46.0,
-      "step": 32982
-    },
-    {
-      "epoch": 2.5217806831431466,
-      "grad_norm": 0.001425395836122334,
-      "learning_rate": 0.00019999686368708303,
-      "loss": 46.0,
-      "step": 32983
-    },
-    {
-      "epoch": 2.5218571401265364,
-      "grad_norm": 0.0012705454137176275,
-      "learning_rate": 0.0001999968634968459,
-      "loss": 46.0,
-      "step": 32984
-    },
-    {
-      "epoch": 2.521933597109926,
-      "grad_norm": 0.0016701509011909366,
-      "learning_rate": 0.00019999686330660305,
-      "loss": 46.0,
-      "step": 32985
-    },
-    {
-      "epoch": 2.522010054093316,
-      "grad_norm": 0.0021274802275002003,
-      "learning_rate": 0.0001999968631163544,
-      "loss": 46.0,
-      "step": 32986
-    },
-    {
-      "epoch": 2.5220865110767052,
-      "grad_norm": 0.0012934100814163685,
-      "learning_rate": 0.0001999968629261,
-      "loss": 46.0,
-      "step": 32987
-    },
-    {
-      "epoch": 2.522162968060095,
-      "grad_norm": 0.0013350382214412093,
-      "learning_rate": 0.00019999686273583983,
-      "loss": 46.0,
-      "step": 32988
-    },
-    {
-      "epoch": 2.5222394250434848,
-      "grad_norm": 0.004998221527785063,
-      "learning_rate": 0.00019999686254557387,
-      "loss": 46.0,
-      "step": 32989
-    },
-    {
-      "epoch": 2.5223158820268745,
-      "grad_norm": 0.0007904151570983231,
-      "learning_rate": 0.00019999686235530213,
-      "loss": 46.0,
-      "step": 32990
-    },
-    {
-      "epoch": 2.5223923390102643,
-      "grad_norm": 0.0021653117146342993,
-      "learning_rate": 0.00019999686216502465,
-      "loss": 46.0,
-      "step": 32991
-    },
-    {
-      "epoch": 2.522468795993654,
-      "grad_norm": 0.0019591893069446087,
-      "learning_rate": 0.0001999968619747414,
-      "loss": 46.0,
-      "step": 32992
-    },
-    {
-      "epoch": 2.522545252977044,
-      "grad_norm": 0.0014295496512204409,
-      "learning_rate": 0.00019999686178445237,
-      "loss": 46.0,
-      "step": 32993
-    },
-    {
-      "epoch": 2.5226217099604336,
-      "grad_norm": 0.003106633434072137,
-      "learning_rate": 0.00019999686159415757,
-      "loss": 46.0,
-      "step": 32994
-    },
-    {
-      "epoch": 2.5226981669438233,
-      "grad_norm": 0.0009059727308340371,
-      "learning_rate": 0.000199996861403857,
-      "loss": 46.0,
-      "step": 32995
-    },
-    {
-      "epoch": 2.522774623927213,
-      "grad_norm": 0.0010645522270351648,
-      "learning_rate": 0.00019999686121355064,
-      "loss": 46.0,
-      "step": 32996
-    },
-    {
-      "epoch": 2.522851080910603,
-      "grad_norm": 0.0027523839380592108,
-      "learning_rate": 0.00019999686102323855,
-      "loss": 46.0,
-      "step": 32997
-    },
-    {
-      "epoch": 2.5229275378939926,
-      "grad_norm": 0.002353076823055744,
-      "learning_rate": 0.00019999686083292065,
-      "loss": 46.0,
-      "step": 32998
-    },
-    {
-      "epoch": 2.5230039948773824,
-      "grad_norm": 0.0017124420264735818,
-      "learning_rate": 0.00019999686064259698,
-      "loss": 46.0,
-      "step": 32999
-    },
-    {
-      "epoch": 2.5230804518607717,
-      "grad_norm": 0.0005272567505016923,
-      "learning_rate": 0.0001999968604522676,
-      "loss": 46.0,
-      "step": 33000
-    },
-    {
-      "epoch": 2.5231569088441614,
-      "grad_norm": 0.001233465038239956,
-      "learning_rate": 0.0001999968602619324,
-      "loss": 46.0,
-      "step": 33001
-    },
-    {
-      "epoch": 2.523233365827551,
-      "grad_norm": 0.001403333037160337,
-      "learning_rate": 0.00019999686007159145,
-      "loss": 46.0,
-      "step": 33002
-    },
-    {
-      "epoch": 2.523309822810941,
-      "grad_norm": 0.0008262894698418677,
-      "learning_rate": 0.00019999685988124471,
-      "loss": 46.0,
-      "step": 33003
-    },
-    {
-      "epoch": 2.5233862797943307,
-      "grad_norm": 0.0021805153228342533,
-      "learning_rate": 0.00019999685969089223,
-      "loss": 46.0,
-      "step": 33004
-    },
-    {
-      "epoch": 2.5234627367777205,
-      "grad_norm": 0.000986619503237307,
-      "learning_rate": 0.00019999685950053398,
-      "loss": 46.0,
-      "step": 33005
-    },
-    {
-      "epoch": 2.5235391937611102,
-      "grad_norm": 0.00140160892624408,
-      "learning_rate": 0.00019999685931016995,
-      "loss": 46.0,
-      "step": 33006
-    },
-    {
-      "epoch": 2.5236156507445,
-      "grad_norm": 0.0011755131417885423,
-      "learning_rate": 0.00019999685911980012,
-      "loss": 46.0,
-      "step": 33007
-    },
-    {
-      "epoch": 2.5236921077278893,
-      "grad_norm": 0.0016934167360886931,
-      "learning_rate": 0.00019999685892942458,
-      "loss": 46.0,
-      "step": 33008
-    },
-    {
-      "epoch": 2.523768564711279,
-      "grad_norm": 0.0009053974063135684,
-      "learning_rate": 0.00019999685873904323,
-      "loss": 46.0,
-      "step": 33009
-    },
-    {
-      "epoch": 2.523845021694669,
-      "grad_norm": 0.002372286980971694,
-      "learning_rate": 0.0001999968585486561,
-      "loss": 46.0,
-      "step": 33010
-    },
-    {
-      "epoch": 2.5239214786780586,
-      "grad_norm": 0.0016073455335572362,
-      "learning_rate": 0.00019999685835826324,
-      "loss": 46.0,
-      "step": 33011
-    },
-    {
-      "epoch": 2.5239979356614484,
-      "grad_norm": 0.0011299309553578496,
-      "learning_rate": 0.00019999685816786457,
-      "loss": 46.0,
-      "step": 33012
-    },
-    {
-      "epoch": 2.524074392644838,
-      "grad_norm": 0.0006076949648559093,
-      "learning_rate": 0.00019999685797746018,
-      "loss": 46.0,
-      "step": 33013
-    },
-    {
-      "epoch": 2.524150849628228,
-      "grad_norm": 0.0010196286020800471,
-      "learning_rate": 0.00019999685778705,
-      "loss": 46.0,
-      "step": 33014
-    },
-    {
-      "epoch": 2.5242273066116176,
-      "grad_norm": 0.0008471967303194106,
-      "learning_rate": 0.00019999685759663403,
-      "loss": 46.0,
-      "step": 33015
-    },
-    {
-      "epoch": 2.5243037635950074,
-      "grad_norm": 0.0006085417116992176,
-      "learning_rate": 0.0001999968574062123,
-      "loss": 46.0,
-      "step": 33016
-    },
-    {
-      "epoch": 2.524380220578397,
-      "grad_norm": 0.0013407089281827211,
-      "learning_rate": 0.00019999685721578482,
-      "loss": 46.0,
-      "step": 33017
-    },
-    {
-      "epoch": 2.524456677561787,
-      "grad_norm": 0.0007719437126070261,
-      "learning_rate": 0.00019999685702535157,
-      "loss": 46.0,
-      "step": 33018
-    },
-    {
-      "epoch": 2.5245331345451767,
-      "grad_norm": 0.0005961884162388742,
-      "learning_rate": 0.0001999968568349125,
-      "loss": 46.0,
-      "step": 33019
-    },
-    {
-      "epoch": 2.5246095915285665,
-      "grad_norm": 0.003857127856463194,
-      "learning_rate": 0.0001999968566444677,
-      "loss": 46.0,
-      "step": 33020
-    },
-    {
-      "epoch": 2.524686048511956,
-      "grad_norm": 0.0009103150805458426,
-      "learning_rate": 0.00019999685645401716,
-      "loss": 46.0,
-      "step": 33021
-    },
-    {
-      "epoch": 2.5247625054953455,
-      "grad_norm": 0.001532554510049522,
-      "learning_rate": 0.0001999968562635608,
-      "loss": 46.0,
-      "step": 33022
-    },
-    {
-      "epoch": 2.5248389624787353,
-      "grad_norm": 0.004269981291145086,
-      "learning_rate": 0.0001999968560730987,
-      "loss": 46.0,
-      "step": 33023
-    },
-    {
-      "epoch": 2.524915419462125,
-      "grad_norm": 0.0023879276122897863,
-      "learning_rate": 0.00019999685588263083,
-      "loss": 46.0,
-      "step": 33024
-    },
-    {
-      "epoch": 2.524991876445515,
-      "grad_norm": 0.0017792272847145796,
-      "learning_rate": 0.00019999685569215716,
-      "loss": 46.0,
-      "step": 33025
-    },
-    {
-      "epoch": 2.5250683334289046,
-      "grad_norm": 0.0006454586982727051,
-      "learning_rate": 0.00019999685550167775,
-      "loss": 46.0,
-      "step": 33026
-    },
-    {
-      "epoch": 2.5251447904122943,
-      "grad_norm": 0.0007813402335159481,
-      "learning_rate": 0.00019999685531119256,
-      "loss": 46.0,
-      "step": 33027
-    },
-    {
-      "epoch": 2.525221247395684,
-      "grad_norm": 0.0016382804606109858,
-      "learning_rate": 0.0001999968551207016,
-      "loss": 46.0,
-      "step": 33028
-    },
-    {
-      "epoch": 2.525297704379074,
-      "grad_norm": 0.004850442986935377,
-      "learning_rate": 0.0001999968549302049,
-      "loss": 46.0,
-      "step": 33029
-    },
-    {
-      "epoch": 2.525374161362463,
-      "grad_norm": 0.0013013448333367705,
-      "learning_rate": 0.00019999685473970238,
-      "loss": 46.0,
-      "step": 33030
-    },
-    {
-      "epoch": 2.525450618345853,
-      "grad_norm": 0.0010715138632804155,
-      "learning_rate": 0.00019999685454919413,
-      "loss": 46.0,
-      "step": 33031
-    },
-    {
-      "epoch": 2.5255270753292427,
-      "grad_norm": 0.0010582986287772655,
-      "learning_rate": 0.0001999968543586801,
-      "loss": 46.0,
-      "step": 33032
-    },
-    {
-      "epoch": 2.5256035323126325,
-      "grad_norm": 0.0021574136335402727,
-      "learning_rate": 0.00019999685416816028,
-      "loss": 46.0,
-      "step": 33033
-    },
-    {
-      "epoch": 2.525679989296022,
-      "grad_norm": 0.0013527146074920893,
-      "learning_rate": 0.00019999685397763473,
-      "loss": 46.0,
-      "step": 33034
-    },
-    {
-      "epoch": 2.525756446279412,
-      "grad_norm": 0.0011443831026554108,
-      "learning_rate": 0.00019999685378710338,
-      "loss": 46.0,
-      "step": 33035
-    },
-    {
-      "epoch": 2.5258329032628017,
-      "grad_norm": 0.0012167183449491858,
-      "learning_rate": 0.00019999685359656626,
-      "loss": 46.0,
-      "step": 33036
-    },
-    {
-      "epoch": 2.5259093602461915,
-      "grad_norm": 0.0017361333593726158,
-      "learning_rate": 0.0001999968534060234,
-      "loss": 46.0,
-      "step": 33037
-    },
-    {
-      "epoch": 2.5259858172295813,
-      "grad_norm": 0.0005349047714844346,
-      "learning_rate": 0.00019999685321547475,
-      "loss": 46.0,
-      "step": 33038
-    },
-    {
-      "epoch": 2.526062274212971,
-      "grad_norm": 0.0010023682843893766,
-      "learning_rate": 0.00019999685302492034,
-      "loss": 46.0,
-      "step": 33039
-    },
-    {
-      "epoch": 2.526138731196361,
-      "grad_norm": 0.001555194379761815,
-      "learning_rate": 0.00019999685283436013,
-      "loss": 46.0,
-      "step": 33040
-    },
-    {
-      "epoch": 2.5262151881797505,
-      "grad_norm": 0.0010983027750626206,
-      "learning_rate": 0.00019999685264379417,
-      "loss": 46.0,
-      "step": 33041
-    },
-    {
-      "epoch": 2.5262916451631403,
-      "grad_norm": 0.004526956006884575,
-      "learning_rate": 0.00019999685245322243,
-      "loss": 46.0,
-      "step": 33042
-    },
-    {
-      "epoch": 2.52636810214653,
-      "grad_norm": 0.0010569618316367269,
-      "learning_rate": 0.00019999685226264495,
-      "loss": 46.0,
-      "step": 33043
-    },
-    {
-      "epoch": 2.5264445591299194,
-      "grad_norm": 0.001752821379341185,
-      "learning_rate": 0.0001999968520720617,
-      "loss": 46.0,
-      "step": 33044
-    },
-    {
-      "epoch": 2.526521016113309,
-      "grad_norm": 0.0019070666749030352,
-      "learning_rate": 0.00019999685188147264,
-      "loss": 46.0,
-      "step": 33045
-    },
-    {
-      "epoch": 2.526597473096699,
-      "grad_norm": 0.004493042826652527,
-      "learning_rate": 0.00019999685169087784,
-      "loss": 46.0,
-      "step": 33046
-    },
-    {
-      "epoch": 2.5266739300800887,
-      "grad_norm": 0.0007848155801184475,
-      "learning_rate": 0.0001999968515002773,
-      "loss": 46.0,
-      "step": 33047
-    },
-    {
-      "epoch": 2.5267503870634784,
-      "grad_norm": 0.00039175813435576856,
-      "learning_rate": 0.00019999685130967095,
-      "loss": 46.0,
-      "step": 33048
-    },
-    {
-      "epoch": 2.526826844046868,
-      "grad_norm": 0.0006324544665403664,
-      "learning_rate": 0.0001999968511190588,
-      "loss": 46.0,
-      "step": 33049
-    },
-    {
-      "epoch": 2.526903301030258,
-      "grad_norm": 0.0021999473683536053,
-      "learning_rate": 0.00019999685092844094,
-      "loss": 46.0,
-      "step": 33050
-    },
-    {
-      "epoch": 2.5269797580136477,
-      "grad_norm": 0.0011834505712613463,
-      "learning_rate": 0.0001999968507378173,
-      "loss": 46.0,
-      "step": 33051
-    },
-    {
-      "epoch": 2.527056214997037,
-      "grad_norm": 0.00431178929284215,
-      "learning_rate": 0.00019999685054718788,
-      "loss": 46.0,
-      "step": 33052
-    },
-    {
-      "epoch": 2.527132671980427,
-      "grad_norm": 0.002885617082938552,
-      "learning_rate": 0.0001999968503565527,
-      "loss": 46.0,
-      "step": 33053
-    },
-    {
-      "epoch": 2.5272091289638166,
-      "grad_norm": 0.001554681919515133,
-      "learning_rate": 0.00019999685016591174,
-      "loss": 46.0,
-      "step": 33054
-    },
-    {
-      "epoch": 2.5272855859472063,
-      "grad_norm": 0.011085913516581059,
-      "learning_rate": 0.000199996849975265,
-      "loss": 46.0,
-      "step": 33055
-    },
-    {
-      "epoch": 2.527362042930596,
-      "grad_norm": 0.0007977046770974994,
-      "learning_rate": 0.00019999684978461252,
-      "loss": 46.0,
-      "step": 33056
-    },
-    {
-      "epoch": 2.527438499913986,
-      "grad_norm": 0.00047069592983461916,
-      "learning_rate": 0.00019999684959395424,
-      "loss": 46.0,
-      "step": 33057
-    },
-    {
-      "epoch": 2.5275149568973756,
-      "grad_norm": 0.0007736576371826231,
-      "learning_rate": 0.00019999684940329022,
-      "loss": 46.0,
-      "step": 33058
-    },
-    {
-      "epoch": 2.5275914138807654,
-      "grad_norm": 0.004061607643961906,
-      "learning_rate": 0.00019999684921262042,
-      "loss": 46.0,
-      "step": 33059
-    },
-    {
-      "epoch": 2.527667870864155,
-      "grad_norm": 0.0016254741931334138,
-      "learning_rate": 0.00019999684902194484,
-      "loss": 46.0,
-      "step": 33060
-    },
-    {
-      "epoch": 2.527744327847545,
-      "grad_norm": 0.0014038145309314132,
-      "learning_rate": 0.00019999684883126347,
-      "loss": 46.0,
-      "step": 33061
-    },
-    {
-      "epoch": 2.5278207848309346,
-      "grad_norm": 0.002375418320298195,
-      "learning_rate": 0.00019999684864057638,
-      "loss": 46.0,
-      "step": 33062
-    },
-    {
-      "epoch": 2.5278972418143244,
-      "grad_norm": 0.0010355737758800387,
-      "learning_rate": 0.0001999968484498835,
-      "loss": 46.0,
-      "step": 33063
-    },
-    {
-      "epoch": 2.527973698797714,
-      "grad_norm": 0.0009131869301199913,
-      "learning_rate": 0.00019999684825918485,
-      "loss": 46.0,
-      "step": 33064
-    },
-    {
-      "epoch": 2.528050155781104,
-      "grad_norm": 0.0016305450117215514,
-      "learning_rate": 0.00019999684806848046,
-      "loss": 46.0,
-      "step": 33065
-    },
-    {
-      "epoch": 2.5281266127644932,
-      "grad_norm": 0.0015315579948946834,
-      "learning_rate": 0.00019999684787777025,
-      "loss": 46.0,
-      "step": 33066
-    },
-    {
-      "epoch": 2.528203069747883,
-      "grad_norm": 0.0011016825446859002,
-      "learning_rate": 0.00019999684768705429,
-      "loss": 46.0,
-      "step": 33067
-    },
-    {
-      "epoch": 2.5282795267312728,
-      "grad_norm": 0.0016450200928375125,
-      "learning_rate": 0.00019999684749633255,
-      "loss": 46.0,
-      "step": 33068
-    },
-    {
-      "epoch": 2.5283559837146625,
-      "grad_norm": 0.001494886470027268,
-      "learning_rate": 0.00019999684730560507,
-      "loss": 46.0,
-      "step": 33069
-    },
-    {
-      "epoch": 2.5284324406980523,
-      "grad_norm": 0.0009504807530902326,
-      "learning_rate": 0.0001999968471148718,
-      "loss": 46.0,
-      "step": 33070
-    },
-    {
-      "epoch": 2.528508897681442,
-      "grad_norm": 0.0010892717400565743,
-      "learning_rate": 0.00019999684692413277,
-      "loss": 46.0,
-      "step": 33071
-    },
-    {
-      "epoch": 2.528585354664832,
-      "grad_norm": 0.0015142413321882486,
-      "learning_rate": 0.00019999684673338797,
-      "loss": 46.0,
-      "step": 33072
-    },
-    {
-      "epoch": 2.5286618116482216,
-      "grad_norm": 0.0008207831415347755,
-      "learning_rate": 0.0001999968465426374,
-      "loss": 46.0,
-      "step": 33073
-    },
-    {
-      "epoch": 2.528738268631611,
-      "grad_norm": 0.0041599883697927,
-      "learning_rate": 0.00019999684635188105,
-      "loss": 46.0,
-      "step": 33074
-    },
-    {
-      "epoch": 2.5288147256150006,
-      "grad_norm": 0.0008629719377495348,
-      "learning_rate": 0.00019999684616111893,
-      "loss": 46.0,
-      "step": 33075
-    },
-    {
-      "epoch": 2.5288911825983904,
-      "grad_norm": 0.0015698411734774709,
-      "learning_rate": 0.00019999684597035104,
-      "loss": 46.0,
-      "step": 33076
-    },
-    {
-      "epoch": 2.52896763958178,
-      "grad_norm": 0.0048864311538636684,
-      "learning_rate": 0.0001999968457795774,
-      "loss": 46.0,
-      "step": 33077
-    },
-    {
-      "epoch": 2.52904409656517,
-      "grad_norm": 0.0007527322159148753,
-      "learning_rate": 0.00019999684558879798,
-      "loss": 46.0,
-      "step": 33078
-    },
-    {
-      "epoch": 2.5291205535485597,
-      "grad_norm": 0.0012936079874634743,
-      "learning_rate": 0.0001999968453980128,
-      "loss": 46.0,
-      "step": 33079
-    },
-    {
-      "epoch": 2.5291970105319495,
-      "grad_norm": 0.003917722962796688,
-      "learning_rate": 0.00019999684520722184,
-      "loss": 46.0,
-      "step": 33080
-    },
-    {
-      "epoch": 2.529273467515339,
-      "grad_norm": 0.00046918168663978577,
-      "learning_rate": 0.00019999684501642513,
-      "loss": 46.0,
-      "step": 33081
-    },
-    {
-      "epoch": 2.529349924498729,
-      "grad_norm": 0.0012198842596262693,
-      "learning_rate": 0.00019999684482562263,
-      "loss": 46.0,
-      "step": 33082
-    },
-    {
-      "epoch": 2.5294263814821187,
-      "grad_norm": 0.0036734065506607294,
-      "learning_rate": 0.00019999684463481435,
-      "loss": 46.0,
-      "step": 33083
-    },
-    {
-      "epoch": 2.5295028384655085,
-      "grad_norm": 0.0011560390703380108,
-      "learning_rate": 0.00019999684444400032,
-      "loss": 46.0,
-      "step": 33084
-    },
-    {
-      "epoch": 2.5295792954488983,
-      "grad_norm": 0.0005636009736917913,
-      "learning_rate": 0.0001999968442531805,
-      "loss": 46.0,
-      "step": 33085
-    },
-    {
-      "epoch": 2.529655752432288,
-      "grad_norm": 0.0014248858205974102,
-      "learning_rate": 0.00019999684406235492,
-      "loss": 46.0,
-      "step": 33086
-    },
-    {
-      "epoch": 2.529732209415678,
-      "grad_norm": 0.001160150277428329,
-      "learning_rate": 0.0001999968438715236,
-      "loss": 46.0,
-      "step": 33087
-    },
-    {
-      "epoch": 2.529808666399067,
-      "grad_norm": 0.006105609238147736,
-      "learning_rate": 0.00019999684368068648,
-      "loss": 46.0,
-      "step": 33088
-    },
-    {
-      "epoch": 2.529885123382457,
-      "grad_norm": 0.0021130649838596582,
-      "learning_rate": 0.0001999968434898436,
-      "loss": 46.0,
-      "step": 33089
-    },
-    {
-      "epoch": 2.5299615803658466,
-      "grad_norm": 0.002258037682622671,
-      "learning_rate": 0.00019999684329899495,
-      "loss": 46.0,
-      "step": 33090
-    },
-    {
-      "epoch": 2.5300380373492364,
-      "grad_norm": 0.0029479372315108776,
-      "learning_rate": 0.00019999684310814054,
-      "loss": 46.0,
-      "step": 33091
-    },
-    {
-      "epoch": 2.530114494332626,
-      "grad_norm": 0.0010797957656905055,
-      "learning_rate": 0.00019999684291728033,
-      "loss": 46.0,
-      "step": 33092
-    },
-    {
-      "epoch": 2.530190951316016,
-      "grad_norm": 0.0007504408131353557,
-      "learning_rate": 0.00019999684272641437,
-      "loss": 46.0,
-      "step": 33093
-    },
-    {
-      "epoch": 2.5302674082994057,
-      "grad_norm": 0.001633386593312025,
-      "learning_rate": 0.00019999684253554266,
-      "loss": 46.0,
-      "step": 33094
-    },
-    {
-      "epoch": 2.5303438652827954,
-      "grad_norm": 0.00040354696102440357,
-      "learning_rate": 0.00019999684234466516,
-      "loss": 46.0,
-      "step": 33095
-    },
-    {
-      "epoch": 2.5304203222661847,
-      "grad_norm": 0.0015485252952203155,
-      "learning_rate": 0.0001999968421537819,
-      "loss": 46.0,
-      "step": 33096
-    },
-    {
-      "epoch": 2.5304967792495745,
-      "grad_norm": 0.0013022536877542734,
-      "learning_rate": 0.00019999684196289285,
-      "loss": 46.0,
-      "step": 33097
-    },
-    {
-      "epoch": 2.5305732362329643,
-      "grad_norm": 0.0011216049315407872,
-      "learning_rate": 0.00019999684177199805,
-      "loss": 46.0,
-      "step": 33098
-    },
-    {
-      "epoch": 2.530649693216354,
-      "grad_norm": 0.0005394744803197682,
-      "learning_rate": 0.00019999684158109745,
-      "loss": 46.0,
-      "step": 33099
-    },
-    {
-      "epoch": 2.530726150199744,
-      "grad_norm": 0.0030475840903818607,
-      "learning_rate": 0.00019999684139019114,
-      "loss": 46.0,
-      "step": 33100
-    },
-    {
-      "epoch": 2.5308026071831335,
-      "grad_norm": 0.001269277185201645,
-      "learning_rate": 0.00019999684119927902,
-      "loss": 46.0,
-      "step": 33101
-    },
-    {
-      "epoch": 2.5308790641665233,
-      "grad_norm": 0.0021002446301281452,
-      "learning_rate": 0.00019999684100836112,
-      "loss": 46.0,
-      "step": 33102
-    },
-    {
-      "epoch": 2.530955521149913,
-      "grad_norm": 0.0009012917871586978,
-      "learning_rate": 0.00019999684081743749,
-      "loss": 46.0,
-      "step": 33103
-    },
-    {
-      "epoch": 2.531031978133303,
-      "grad_norm": 0.0014785972889512777,
-      "learning_rate": 0.00019999684062650807,
-      "loss": 46.0,
-      "step": 33104
-    },
-    {
-      "epoch": 2.5311084351166926,
-      "grad_norm": 0.0034333476796746254,
-      "learning_rate": 0.00019999684043557286,
-      "loss": 46.0,
-      "step": 33105
-    },
-    {
-      "epoch": 2.5311848921000824,
-      "grad_norm": 0.0017998154507949948,
-      "learning_rate": 0.0001999968402446319,
-      "loss": 46.0,
-      "step": 33106
-    },
-    {
-      "epoch": 2.531261349083472,
-      "grad_norm": 0.0005658913287334144,
-      "learning_rate": 0.00019999684005368517,
-      "loss": 46.0,
-      "step": 33107
-    },
-    {
-      "epoch": 2.531337806066862,
-      "grad_norm": 0.004063844680786133,
-      "learning_rate": 0.00019999683986273266,
-      "loss": 46.0,
-      "step": 33108
-    },
-    {
-      "epoch": 2.531414263050251,
-      "grad_norm": 0.011886722408235073,
-      "learning_rate": 0.0001999968396717744,
-      "loss": 46.0,
-      "step": 33109
-    },
-    {
-      "epoch": 2.531490720033641,
-      "grad_norm": 0.0013547228882089257,
-      "learning_rate": 0.0001999968394808104,
-      "loss": 46.0,
-      "step": 33110
-    },
-    {
-      "epoch": 2.5315671770170307,
-      "grad_norm": 0.0014104462461546063,
-      "learning_rate": 0.0001999968392898406,
-      "loss": 46.0,
-      "step": 33111
-    },
-    {
-      "epoch": 2.5316436340004205,
-      "grad_norm": 0.0008720263722352684,
-      "learning_rate": 0.000199996839098865,
-      "loss": 46.0,
-      "step": 33112
-    },
-    {
-      "epoch": 2.5317200909838102,
-      "grad_norm": 0.001358148641884327,
-      "learning_rate": 0.00019999683890788364,
-      "loss": 46.0,
-      "step": 33113
-    },
-    {
-      "epoch": 2.5317965479672,
-      "grad_norm": 0.0017387621337547898,
-      "learning_rate": 0.00019999683871689653,
-      "loss": 46.0,
-      "step": 33114
-    },
-    {
-      "epoch": 2.5318730049505898,
-      "grad_norm": 0.001016968977637589,
-      "learning_rate": 0.00019999683852590366,
-      "loss": 46.0,
-      "step": 33115
-    },
-    {
-      "epoch": 2.5319494619339795,
-      "grad_norm": 0.002298773266375065,
-      "learning_rate": 0.00019999683833490502,
-      "loss": 46.0,
-      "step": 33116
-    },
-    {
-      "epoch": 2.5320259189173693,
-      "grad_norm": 0.0013734218664467335,
-      "learning_rate": 0.00019999683814390058,
-      "loss": 46.0,
-      "step": 33117
-    },
-    {
-      "epoch": 2.5321023759007586,
-      "grad_norm": 0.002011583186686039,
-      "learning_rate": 0.0001999968379528904,
-      "loss": 46.0,
-      "step": 33118
-    },
-    {
-      "epoch": 2.5321788328841484,
-      "grad_norm": 0.001407851930707693,
-      "learning_rate": 0.00019999683776187444,
-      "loss": 46.0,
-      "step": 33119
-    },
-    {
-      "epoch": 2.532255289867538,
-      "grad_norm": 0.0011419655056670308,
-      "learning_rate": 0.0001999968375708527,
-      "loss": 46.0,
-      "step": 33120
-    },
-    {
-      "epoch": 2.532331746850928,
-      "grad_norm": 0.0007425840012729168,
-      "learning_rate": 0.0001999968373798252,
-      "loss": 46.0,
-      "step": 33121
-    },
-    {
-      "epoch": 2.5324082038343176,
-      "grad_norm": 0.008075503632426262,
-      "learning_rate": 0.00019999683718879195,
-      "loss": 46.0,
-      "step": 33122
-    },
-    {
-      "epoch": 2.5324846608177074,
-      "grad_norm": 0.0005412347964011133,
-      "learning_rate": 0.0001999968369977529,
-      "loss": 46.0,
-      "step": 33123
-    },
-    {
-      "epoch": 2.532561117801097,
-      "grad_norm": 0.0007657576934434474,
-      "learning_rate": 0.00019999683680670807,
-      "loss": 46.0,
-      "step": 33124
-    },
-    {
-      "epoch": 2.532637574784487,
-      "grad_norm": 0.001452778815291822,
-      "learning_rate": 0.00019999683661565753,
-      "loss": 46.0,
-      "step": 33125
-    },
-    {
-      "epoch": 2.5327140317678767,
-      "grad_norm": 0.0012236229376867414,
-      "learning_rate": 0.00019999683642460119,
-      "loss": 46.0,
-      "step": 33126
-    },
-    {
-      "epoch": 2.5327904887512664,
-      "grad_norm": 0.0012881732545793056,
-      "learning_rate": 0.00019999683623353907,
-      "loss": 46.0,
-      "step": 33127
-    },
-    {
-      "epoch": 2.532866945734656,
-      "grad_norm": 0.0012932823738083243,
-      "learning_rate": 0.0001999968360424712,
-      "loss": 46.0,
-      "step": 33128
-    },
-    {
-      "epoch": 2.532943402718046,
-      "grad_norm": 0.002646426670253277,
-      "learning_rate": 0.00019999683585139754,
-      "loss": 46.0,
-      "step": 33129
-    },
-    {
-      "epoch": 2.5330198597014357,
-      "grad_norm": 0.0012315618805587292,
-      "learning_rate": 0.00019999683566031813,
-      "loss": 46.0,
-      "step": 33130
-    },
-    {
-      "epoch": 2.533096316684825,
-      "grad_norm": 0.0010689053451642394,
-      "learning_rate": 0.00019999683546923294,
-      "loss": 46.0,
-      "step": 33131
-    },
-    {
-      "epoch": 2.533172773668215,
-      "grad_norm": 0.0018512161914259195,
-      "learning_rate": 0.00019999683527814198,
-      "loss": 46.0,
-      "step": 33132
-    },
-    {
-      "epoch": 2.5332492306516046,
-      "grad_norm": 0.003827217733487487,
-      "learning_rate": 0.00019999683508704522,
-      "loss": 46.0,
-      "step": 33133
-    },
-    {
-      "epoch": 2.5333256876349943,
-      "grad_norm": 0.004490390885621309,
-      "learning_rate": 0.00019999683489594272,
-      "loss": 46.0,
-      "step": 33134
-    },
-    {
-      "epoch": 2.533402144618384,
-      "grad_norm": 0.0033484799787402153,
-      "learning_rate": 0.00019999683470483447,
-      "loss": 46.0,
-      "step": 33135
-    },
-    {
-      "epoch": 2.533478601601774,
-      "grad_norm": 0.0030469768680632114,
-      "learning_rate": 0.00019999683451372042,
-      "loss": 46.0,
-      "step": 33136
-    },
-    {
-      "epoch": 2.5335550585851636,
-      "grad_norm": 0.0019159261137247086,
-      "learning_rate": 0.00019999683432260065,
-      "loss": 46.0,
-      "step": 33137
-    },
-    {
-      "epoch": 2.5336315155685534,
-      "grad_norm": 0.0017138720722869039,
-      "learning_rate": 0.00019999683413147502,
-      "loss": 46.0,
-      "step": 33138
-    },
-    {
-      "epoch": 2.5337079725519427,
-      "grad_norm": 0.00629083439707756,
-      "learning_rate": 0.0001999968339403437,
-      "loss": 46.0,
-      "step": 33139
-    },
-    {
-      "epoch": 2.5337844295353324,
-      "grad_norm": 0.0014840165385976434,
-      "learning_rate": 0.00019999683374920658,
-      "loss": 46.0,
-      "step": 33140
-    },
-    {
-      "epoch": 2.533860886518722,
-      "grad_norm": 0.001617331407032907,
-      "learning_rate": 0.00019999683355806372,
-      "loss": 46.0,
-      "step": 33141
-    },
-    {
-      "epoch": 2.533937343502112,
-      "grad_norm": 0.005992131307721138,
-      "learning_rate": 0.00019999683336691505,
-      "loss": 46.0,
-      "step": 33142
-    },
-    {
-      "epoch": 2.5340138004855017,
-      "grad_norm": 0.0008454010239802301,
-      "learning_rate": 0.00019999683317576064,
-      "loss": 46.0,
-      "step": 33143
-    },
-    {
-      "epoch": 2.5340902574688915,
-      "grad_norm": 0.002283011330291629,
-      "learning_rate": 0.00019999683298460043,
-      "loss": 46.0,
-      "step": 33144
-    },
-    {
-      "epoch": 2.5341667144522813,
-      "grad_norm": 0.002449593273922801,
-      "learning_rate": 0.00019999683279343447,
-      "loss": 46.0,
-      "step": 33145
-    },
-    {
-      "epoch": 2.534243171435671,
-      "grad_norm": 0.0007709492347203195,
-      "learning_rate": 0.00019999683260226274,
-      "loss": 46.0,
-      "step": 33146
-    },
-    {
-      "epoch": 2.5343196284190608,
-      "grad_norm": 0.0013838186860084534,
-      "learning_rate": 0.00019999683241108524,
-      "loss": 46.0,
-      "step": 33147
-    },
-    {
-      "epoch": 2.5343960854024505,
-      "grad_norm": 0.0021905521862208843,
-      "learning_rate": 0.000199996832219902,
-      "loss": 46.0,
-      "step": 33148
-    },
-    {
-      "epoch": 2.5344725423858403,
-      "grad_norm": 0.0016100776847451925,
-      "learning_rate": 0.00019999683202871294,
-      "loss": 46.0,
-      "step": 33149
-    },
-    {
-      "epoch": 2.53454899936923,
-      "grad_norm": 0.0025662125553935766,
-      "learning_rate": 0.00019999683183751814,
-      "loss": 46.0,
-      "step": 33150
-    },
-    {
-      "epoch": 2.53462545635262,
-      "grad_norm": 0.004624331835657358,
-      "learning_rate": 0.00019999683164631754,
-      "loss": 46.0,
-      "step": 33151
-    },
-    {
-      "epoch": 2.5347019133360096,
-      "grad_norm": 0.0040935128927230835,
-      "learning_rate": 0.00019999683145511122,
-      "loss": 46.0,
-      "step": 33152
-    },
-    {
-      "epoch": 2.534778370319399,
-      "grad_norm": 0.0006664943648502231,
-      "learning_rate": 0.00019999683126389908,
-      "loss": 46.0,
-      "step": 33153
-    },
-    {
-      "epoch": 2.5348548273027887,
-      "grad_norm": 0.0034209024161100388,
-      "learning_rate": 0.00019999683107268121,
-      "loss": 46.0,
-      "step": 33154
-    },
-    {
-      "epoch": 2.5349312842861784,
-      "grad_norm": 0.0006979820318520069,
-      "learning_rate": 0.00019999683088145755,
-      "loss": 46.0,
-      "step": 33155
-    },
-    {
-      "epoch": 2.535007741269568,
-      "grad_norm": 0.0009852969087660313,
-      "learning_rate": 0.00019999683069022814,
-      "loss": 46.0,
-      "step": 33156
-    },
-    {
-      "epoch": 2.535084198252958,
-      "grad_norm": 0.0010831133695319295,
-      "learning_rate": 0.00019999683049899293,
-      "loss": 46.0,
-      "step": 33157
-    },
-    {
-      "epoch": 2.5351606552363477,
-      "grad_norm": 0.001308755250647664,
-      "learning_rate": 0.00019999683030775197,
-      "loss": 46.0,
-      "step": 33158
-    },
-    {
-      "epoch": 2.5352371122197375,
-      "grad_norm": 0.0013171881437301636,
-      "learning_rate": 0.00019999683011650524,
-      "loss": 46.0,
-      "step": 33159
-    },
-    {
-      "epoch": 2.5353135692031272,
-      "grad_norm": 0.0004779967712238431,
-      "learning_rate": 0.00019999682992525276,
-      "loss": 46.0,
-      "step": 33160
-    },
-    {
-      "epoch": 2.5353900261865165,
-      "grad_norm": 0.0014062876580283046,
-      "learning_rate": 0.00019999682973399448,
-      "loss": 46.0,
-      "step": 33161
-    },
-    {
-      "epoch": 2.5354664831699063,
-      "grad_norm": 0.00204377225600183,
-      "learning_rate": 0.00019999682954273043,
-      "loss": 46.0,
-      "step": 33162
-    },
-    {
-      "epoch": 2.535542940153296,
-      "grad_norm": 0.0034058387391269207,
-      "learning_rate": 0.00019999682935146064,
-      "loss": 46.0,
-      "step": 33163
-    },
-    {
-      "epoch": 2.535619397136686,
-      "grad_norm": 0.0014958331594243646,
-      "learning_rate": 0.00019999682916018504,
-      "loss": 46.0,
-      "step": 33164
-    },
-    {
-      "epoch": 2.5356958541200756,
-      "grad_norm": 0.001771115930750966,
-      "learning_rate": 0.0001999968289689037,
-      "loss": 46.0,
-      "step": 33165
-    },
-    {
-      "epoch": 2.5357723111034653,
-      "grad_norm": 0.0006459006108343601,
-      "learning_rate": 0.0001999968287776166,
-      "loss": 46.0,
-      "step": 33166
-    },
-    {
-      "epoch": 2.535848768086855,
-      "grad_norm": 0.0008857916691340506,
-      "learning_rate": 0.0001999968285863237,
-      "loss": 46.0,
-      "step": 33167
-    },
-    {
-      "epoch": 2.535925225070245,
-      "grad_norm": 0.0011233666446059942,
-      "learning_rate": 0.00019999682839502505,
-      "loss": 46.0,
-      "step": 33168
-    },
-    {
-      "epoch": 2.5360016820536346,
-      "grad_norm": 0.0017704321071505547,
-      "learning_rate": 0.00019999682820372064,
-      "loss": 46.0,
-      "step": 33169
-    },
-    {
-      "epoch": 2.5360781390370244,
-      "grad_norm": 0.0009050744702108204,
-      "learning_rate": 0.00019999682801241045,
-      "loss": 46.0,
-      "step": 33170
-    },
-    {
-      "epoch": 2.536154596020414,
-      "grad_norm": 0.0032069936860352755,
-      "learning_rate": 0.00019999682782109447,
-      "loss": 46.0,
-      "step": 33171
-    },
-    {
-      "epoch": 2.536231053003804,
-      "grad_norm": 0.0012016250984743237,
-      "learning_rate": 0.00019999682762977274,
-      "loss": 46.0,
-      "step": 33172
-    },
-    {
-      "epoch": 2.5363075099871937,
-      "grad_norm": 0.0016288973856717348,
-      "learning_rate": 0.00019999682743844526,
-      "loss": 46.0,
-      "step": 33173
-    },
-    {
-      "epoch": 2.5363839669705834,
-      "grad_norm": 0.004902354441583157,
-      "learning_rate": 0.00019999682724711199,
-      "loss": 46.0,
-      "step": 33174
-    },
-    {
-      "epoch": 2.5364604239539728,
-      "grad_norm": 0.004460535012185574,
-      "learning_rate": 0.00019999682705577293,
-      "loss": 46.0,
-      "step": 33175
-    },
-    {
-      "epoch": 2.5365368809373625,
-      "grad_norm": 0.0013700368581339717,
-      "learning_rate": 0.00019999682686442814,
-      "loss": 46.0,
-      "step": 33176
-    },
-    {
-      "epoch": 2.5366133379207523,
-      "grad_norm": 0.0005593557725660503,
-      "learning_rate": 0.00019999682667307754,
-      "loss": 46.0,
-      "step": 33177
-    },
-    {
-      "epoch": 2.536689794904142,
-      "grad_norm": 0.0013128576101735234,
-      "learning_rate": 0.0001999968264817212,
-      "loss": 46.0,
-      "step": 33178
-    },
-    {
-      "epoch": 2.536766251887532,
-      "grad_norm": 0.0026080291718244553,
-      "learning_rate": 0.0001999968262903591,
-      "loss": 46.0,
-      "step": 33179
-    },
-    {
-      "epoch": 2.5368427088709216,
-      "grad_norm": 0.003099835244938731,
-      "learning_rate": 0.0001999968260989912,
-      "loss": 46.0,
-      "step": 33180
-    },
-    {
-      "epoch": 2.5369191658543113,
-      "grad_norm": 0.001067670644260943,
-      "learning_rate": 0.00019999682590761755,
-      "loss": 46.0,
-      "step": 33181
-    },
-    {
-      "epoch": 2.536995622837701,
-      "grad_norm": 0.0036657836753875017,
-      "learning_rate": 0.00019999682571623812,
-      "loss": 46.0,
-      "step": 33182
-    },
-    {
-      "epoch": 2.5370720798210904,
-      "grad_norm": 0.0008162794983945787,
-      "learning_rate": 0.00019999682552485293,
-      "loss": 46.0,
-      "step": 33183
-    },
-    {
-      "epoch": 2.53714853680448,
-      "grad_norm": 0.0023726962972432375,
-      "learning_rate": 0.00019999682533346195,
-      "loss": 46.0,
-      "step": 33184
-    },
-    {
-      "epoch": 2.53722499378787,
-      "grad_norm": 0.004566030576825142,
-      "learning_rate": 0.00019999682514206524,
-      "loss": 46.0,
-      "step": 33185
-    },
-    {
-      "epoch": 2.5373014507712597,
-      "grad_norm": 0.001993596088141203,
-      "learning_rate": 0.00019999682495066274,
-      "loss": 46.0,
-      "step": 33186
-    },
-    {
-      "epoch": 2.5373779077546494,
-      "grad_norm": 0.0012082206085324287,
-      "learning_rate": 0.00019999682475925446,
-      "loss": 46.0,
-      "step": 33187
-    },
-    {
-      "epoch": 2.537454364738039,
-      "grad_norm": 0.0015303262043744326,
-      "learning_rate": 0.00019999682456784044,
-      "loss": 46.0,
-      "step": 33188
-    },
-    {
-      "epoch": 2.537530821721429,
-      "grad_norm": 0.0009081342723220587,
-      "learning_rate": 0.00019999682437642062,
-      "loss": 46.0,
-      "step": 33189
-    },
-    {
-      "epoch": 2.5376072787048187,
-      "grad_norm": 0.0024489767383784056,
-      "learning_rate": 0.00019999682418499505,
-      "loss": 46.0,
-      "step": 33190
-    },
-    {
-      "epoch": 2.5376837356882085,
-      "grad_norm": 0.0011608997592702508,
-      "learning_rate": 0.0001999968239935637,
-      "loss": 46.0,
-      "step": 33191
-    },
-    {
-      "epoch": 2.5377601926715982,
-      "grad_norm": 0.00101863918825984,
-      "learning_rate": 0.00019999682380212659,
-      "loss": 46.0,
-      "step": 33192
-    },
-    {
-      "epoch": 2.537836649654988,
-      "grad_norm": 0.0020845935214310884,
-      "learning_rate": 0.00019999682361068367,
-      "loss": 46.0,
-      "step": 33193
-    },
-    {
-      "epoch": 2.5379131066383778,
-      "grad_norm": 0.0013141219969838858,
-      "learning_rate": 0.00019999682341923506,
-      "loss": 46.0,
-      "step": 33194
-    },
-    {
-      "epoch": 2.5379895636217675,
-      "grad_norm": 0.0008195218397304416,
-      "learning_rate": 0.00019999682322778062,
-      "loss": 46.0,
-      "step": 33195
-    },
-    {
-      "epoch": 2.5380660206051573,
-      "grad_norm": 0.0008450259338133037,
-      "learning_rate": 0.0001999968230363204,
-      "loss": 46.0,
-      "step": 33196
-    },
-    {
-      "epoch": 2.5381424775885466,
-      "grad_norm": 0.0016064007068052888,
-      "learning_rate": 0.00019999682284485446,
-      "loss": 46.0,
-      "step": 33197
-    },
-    {
-      "epoch": 2.5382189345719364,
-      "grad_norm": 0.0030375681817531586,
-      "learning_rate": 0.00019999682265338273,
-      "loss": 46.0,
-      "step": 33198
-    },
-    {
-      "epoch": 2.538295391555326,
-      "grad_norm": 0.00444938987493515,
-      "learning_rate": 0.00019999682246190522,
-      "loss": 46.0,
-      "step": 33199
-    },
-    {
-      "epoch": 2.538371848538716,
-      "grad_norm": 0.0016077859327197075,
-      "learning_rate": 0.00019999682227042195,
-      "loss": 46.0,
-      "step": 33200
-    },
-    {
-      "epoch": 2.5384483055221057,
-      "grad_norm": 0.0022703332360833883,
-      "learning_rate": 0.00019999682207893292,
-      "loss": 46.0,
-      "step": 33201
-    },
-    {
-      "epoch": 2.5385247625054954,
-      "grad_norm": 0.006222869269549847,
-      "learning_rate": 0.0001999968218874381,
-      "loss": 46.0,
-      "step": 33202
-    },
-    {
-      "epoch": 2.538601219488885,
-      "grad_norm": 0.001716992468573153,
-      "learning_rate": 0.00019999682169593753,
-      "loss": 46.0,
-      "step": 33203
-    },
-    {
-      "epoch": 2.538677676472275,
-      "grad_norm": 0.004201965872198343,
-      "learning_rate": 0.0001999968215044312,
-      "loss": 46.0,
-      "step": 33204
-    },
-    {
-      "epoch": 2.5387541334556643,
-      "grad_norm": 0.003738280152902007,
-      "learning_rate": 0.00019999682131291907,
-      "loss": 46.0,
-      "step": 33205
-    },
-    {
-      "epoch": 2.538830590439054,
-      "grad_norm": 0.001802522805519402,
-      "learning_rate": 0.00019999682112140118,
-      "loss": 46.0,
-      "step": 33206
-    },
-    {
-      "epoch": 2.5389070474224438,
-      "grad_norm": 0.0011579536367207766,
-      "learning_rate": 0.00019999682092987752,
-      "loss": 46.0,
-      "step": 33207
-    },
-    {
-      "epoch": 2.5389835044058335,
-      "grad_norm": 0.002159234369173646,
-      "learning_rate": 0.00019999682073834808,
-      "loss": 46.0,
-      "step": 33208
-    },
-    {
-      "epoch": 2.5390599613892233,
-      "grad_norm": 0.0009778555249795318,
-      "learning_rate": 0.0001999968205468129,
-      "loss": 46.0,
-      "step": 33209
-    },
-    {
-      "epoch": 2.539136418372613,
-      "grad_norm": 0.0015894186217337847,
-      "learning_rate": 0.00019999682035527191,
-      "loss": 46.0,
-      "step": 33210
-    },
-    {
-      "epoch": 2.539212875356003,
-      "grad_norm": 0.002017409075051546,
-      "learning_rate": 0.00019999682016372518,
-      "loss": 46.0,
-      "step": 33211
-    },
-    {
-      "epoch": 2.5392893323393926,
-      "grad_norm": 0.0012309554731473327,
-      "learning_rate": 0.0001999968199721727,
-      "loss": 46.0,
-      "step": 33212
-    },
-    {
-      "epoch": 2.5393657893227823,
-      "grad_norm": 0.0009768244344741106,
-      "learning_rate": 0.00019999681978061443,
-      "loss": 46.0,
-      "step": 33213
-    },
-    {
-      "epoch": 2.539442246306172,
-      "grad_norm": 0.001832338748499751,
-      "learning_rate": 0.00019999681958905038,
-      "loss": 46.0,
-      "step": 33214
-    },
-    {
-      "epoch": 2.539518703289562,
-      "grad_norm": 0.0016539524076506495,
-      "learning_rate": 0.00019999681939748056,
-      "loss": 46.0,
-      "step": 33215
-    },
-    {
-      "epoch": 2.5395951602729516,
-      "grad_norm": 0.005241594277322292,
-      "learning_rate": 0.00019999681920590502,
-      "loss": 46.0,
-      "step": 33216
-    },
-    {
-      "epoch": 2.5396716172563414,
-      "grad_norm": 0.0011519135441631079,
-      "learning_rate": 0.00019999681901432365,
-      "loss": 46.0,
-      "step": 33217
-    },
-    {
-      "epoch": 2.539748074239731,
-      "grad_norm": 0.00048249386600218713,
-      "learning_rate": 0.00019999681882273653,
-      "loss": 46.0,
-      "step": 33218
-    },
-    {
-      "epoch": 2.5398245312231205,
-      "grad_norm": 0.0012649719137698412,
-      "learning_rate": 0.00019999681863114367,
-      "loss": 46.0,
-      "step": 33219
-    },
-    {
-      "epoch": 2.5399009882065102,
-      "grad_norm": 0.0008692637202329934,
-      "learning_rate": 0.00019999681843954498,
-      "loss": 46.0,
-      "step": 33220
-    },
-    {
-      "epoch": 2.5399774451899,
-      "grad_norm": 0.0014745030784979463,
-      "learning_rate": 0.00019999681824794057,
-      "loss": 46.0,
-      "step": 33221
-    },
-    {
-      "epoch": 2.5400539021732897,
-      "grad_norm": 0.0013729128986597061,
-      "learning_rate": 0.00019999681805633036,
-      "loss": 46.0,
-      "step": 33222
-    },
-    {
-      "epoch": 2.5401303591566795,
-      "grad_norm": 0.0007544734980911016,
-      "learning_rate": 0.0001999968178647144,
-      "loss": 46.0,
-      "step": 33223
-    },
-    {
-      "epoch": 2.5402068161400693,
-      "grad_norm": 0.0025739565026015043,
-      "learning_rate": 0.00019999681767309267,
-      "loss": 46.0,
-      "step": 33224
-    },
-    {
-      "epoch": 2.540283273123459,
-      "grad_norm": 0.000791975820902735,
-      "learning_rate": 0.00019999681748146517,
-      "loss": 46.0,
-      "step": 33225
-    },
-    {
-      "epoch": 2.540359730106849,
-      "grad_norm": 0.00694398395717144,
-      "learning_rate": 0.0001999968172898319,
-      "loss": 46.0,
-      "step": 33226
-    },
-    {
-      "epoch": 2.540436187090238,
-      "grad_norm": 0.0014294743305072188,
-      "learning_rate": 0.00019999681709819287,
-      "loss": 46.0,
-      "step": 33227
-    },
-    {
-      "epoch": 2.540512644073628,
-      "grad_norm": 0.0010745887411758304,
-      "learning_rate": 0.00019999681690654802,
-      "loss": 46.0,
-      "step": 33228
-    },
-    {
-      "epoch": 2.5405891010570176,
-      "grad_norm": 0.009002290666103363,
-      "learning_rate": 0.00019999681671489748,
-      "loss": 46.0,
-      "step": 33229
-    },
-    {
-      "epoch": 2.5406655580404074,
-      "grad_norm": 0.0010493833106011152,
-      "learning_rate": 0.0001999968165232411,
-      "loss": 46.0,
-      "step": 33230
-    },
-    {
-      "epoch": 2.540742015023797,
-      "grad_norm": 0.0008426719577983022,
-      "learning_rate": 0.000199996816331579,
-      "loss": 46.0,
-      "step": 33231
-    },
-    {
-      "epoch": 2.540818472007187,
-      "grad_norm": 0.0007171841571107507,
-      "learning_rate": 0.0001999968161399111,
-      "loss": 46.0,
-      "step": 33232
-    },
-    {
-      "epoch": 2.5408949289905767,
-      "grad_norm": 0.002814331790432334,
-      "learning_rate": 0.00019999681594823744,
-      "loss": 46.0,
-      "step": 33233
-    },
-    {
-      "epoch": 2.5409713859739664,
-      "grad_norm": 0.00034515763400122523,
-      "learning_rate": 0.000199996815756558,
-      "loss": 46.0,
-      "step": 33234
-    },
-    {
-      "epoch": 2.541047842957356,
-      "grad_norm": 0.0007364174234680831,
-      "learning_rate": 0.00019999681556487283,
-      "loss": 46.0,
-      "step": 33235
-    },
-    {
-      "epoch": 2.541124299940746,
-      "grad_norm": 0.0016923134680837393,
-      "learning_rate": 0.00019999681537318187,
-      "loss": 46.0,
-      "step": 33236
-    },
-    {
-      "epoch": 2.5412007569241357,
-      "grad_norm": 0.001314935740083456,
-      "learning_rate": 0.00019999681518148514,
-      "loss": 46.0,
-      "step": 33237
-    },
-    {
-      "epoch": 2.5412772139075255,
-      "grad_norm": 0.0013194195926189423,
-      "learning_rate": 0.0001999968149897826,
-      "loss": 46.0,
-      "step": 33238
-    },
-    {
-      "epoch": 2.5413536708909152,
-      "grad_norm": 0.0010526138357818127,
-      "learning_rate": 0.00019999681479807436,
-      "loss": 46.0,
-      "step": 33239
-    },
-    {
-      "epoch": 2.5414301278743046,
-      "grad_norm": 0.0029739351011812687,
-      "learning_rate": 0.0001999968146063603,
-      "loss": 46.0,
-      "step": 33240
-    },
-    {
-      "epoch": 2.5415065848576943,
-      "grad_norm": 0.001488157780840993,
-      "learning_rate": 0.00019999681441464052,
-      "loss": 46.0,
-      "step": 33241
-    },
-    {
-      "epoch": 2.541583041841084,
-      "grad_norm": 0.005616537760943174,
-      "learning_rate": 0.00019999681422291495,
-      "loss": 46.0,
-      "step": 33242
-    },
-    {
-      "epoch": 2.541659498824474,
-      "grad_norm": 0.0008851112797856331,
-      "learning_rate": 0.00019999681403118358,
-      "loss": 46.0,
-      "step": 33243
-    },
-    {
-      "epoch": 2.5417359558078636,
-      "grad_norm": 0.0014336224412545562,
-      "learning_rate": 0.00019999681383944646,
-      "loss": 46.0,
-      "step": 33244
-    },
-    {
-      "epoch": 2.5418124127912534,
-      "grad_norm": 0.0009643150260671973,
-      "learning_rate": 0.00019999681364770358,
-      "loss": 46.0,
-      "step": 33245
-    },
-    {
-      "epoch": 2.541888869774643,
-      "grad_norm": 0.0013688254402950406,
-      "learning_rate": 0.0001999968134559549,
-      "loss": 46.0,
-      "step": 33246
-    },
-    {
-      "epoch": 2.541965326758033,
-      "grad_norm": 0.00523220282047987,
-      "learning_rate": 0.0001999968132642005,
-      "loss": 46.0,
-      "step": 33247
-    },
-    {
-      "epoch": 2.5420417837414226,
-      "grad_norm": 0.0007325896294787526,
-      "learning_rate": 0.0001999968130724403,
-      "loss": 46.0,
-      "step": 33248
-    },
-    {
-      "epoch": 2.542118240724812,
-      "grad_norm": 0.0009506585774943233,
-      "learning_rate": 0.00019999681288067434,
-      "loss": 46.0,
-      "step": 33249
-    },
-    {
-      "epoch": 2.5421946977082017,
-      "grad_norm": 0.0008434379706159234,
-      "learning_rate": 0.0001999968126889026,
-      "loss": 46.0,
-      "step": 33250
-    },
-    {
-      "epoch": 2.5422711546915915,
-      "grad_norm": 0.0008841908420436084,
-      "learning_rate": 0.0001999968124971251,
-      "loss": 46.0,
-      "step": 33251
-    },
-    {
-      "epoch": 2.5423476116749812,
-      "grad_norm": 0.0014496170915663242,
-      "learning_rate": 0.0001999968123053418,
-      "loss": 46.0,
-      "step": 33252
-    },
-    {
-      "epoch": 2.542424068658371,
-      "grad_norm": 0.0011465571587905288,
-      "learning_rate": 0.00019999681211355278,
-      "loss": 46.0,
-      "step": 33253
-    },
-    {
-      "epoch": 2.5425005256417608,
-      "grad_norm": 0.00113820587284863,
-      "learning_rate": 0.00019999681192175796,
-      "loss": 46.0,
-      "step": 33254
-    },
-    {
-      "epoch": 2.5425769826251505,
-      "grad_norm": 0.000366052845492959,
-      "learning_rate": 0.0001999968117299574,
-      "loss": 46.0,
-      "step": 33255
-    },
-    {
-      "epoch": 2.5426534396085403,
-      "grad_norm": 0.0012464741012081504,
-      "learning_rate": 0.00019999681153815103,
-      "loss": 46.0,
-      "step": 33256
-    },
-    {
-      "epoch": 2.54272989659193,
-      "grad_norm": 0.0011187423951923847,
-      "learning_rate": 0.0001999968113463389,
-      "loss": 46.0,
-      "step": 33257
-    },
-    {
-      "epoch": 2.54280635357532,
-      "grad_norm": 0.0029635403770953417,
-      "learning_rate": 0.00019999681115452102,
-      "loss": 46.0,
-      "step": 33258
-    },
-    {
-      "epoch": 2.5428828105587096,
-      "grad_norm": 0.0016510817222297192,
-      "learning_rate": 0.0001999968109626974,
-      "loss": 46.0,
-      "step": 33259
-    },
-    {
-      "epoch": 2.5429592675420993,
-      "grad_norm": 0.0010473427828401327,
-      "learning_rate": 0.00019999681077086792,
-      "loss": 46.0,
-      "step": 33260
-    },
-    {
-      "epoch": 2.543035724525489,
-      "grad_norm": 0.0004295198596082628,
-      "learning_rate": 0.00019999681057903274,
-      "loss": 46.0,
-      "step": 33261
-    },
-    {
-      "epoch": 2.5431121815088784,
-      "grad_norm": 0.0024259027559310198,
-      "learning_rate": 0.0001999968103871918,
-      "loss": 46.0,
-      "step": 33262
-    },
-    {
-      "epoch": 2.543188638492268,
-      "grad_norm": 0.0007955812616273761,
-      "learning_rate": 0.00019999681019534503,
-      "loss": 46.0,
-      "step": 33263
-    },
-    {
-      "epoch": 2.543265095475658,
-      "grad_norm": 0.0005930929910391569,
-      "learning_rate": 0.00019999681000349253,
-      "loss": 46.0,
-      "step": 33264
-    },
-    {
-      "epoch": 2.5433415524590477,
-      "grad_norm": 0.0008387153502553701,
-      "learning_rate": 0.00019999680981163426,
-      "loss": 46.0,
-      "step": 33265
-    },
-    {
-      "epoch": 2.5434180094424375,
-      "grad_norm": 0.0006535985157825053,
-      "learning_rate": 0.00019999680961977023,
-      "loss": 46.0,
-      "step": 33266
-    },
-    {
-      "epoch": 2.543494466425827,
-      "grad_norm": 0.0007571714813821018,
-      "learning_rate": 0.0001999968094279004,
-      "loss": 46.0,
-      "step": 33267
-    },
-    {
-      "epoch": 2.543570923409217,
-      "grad_norm": 0.000826736562885344,
-      "learning_rate": 0.00019999680923602484,
-      "loss": 46.0,
-      "step": 33268
-    },
-    {
-      "epoch": 2.5436473803926067,
-      "grad_norm": 0.010101170279085636,
-      "learning_rate": 0.00019999680904414347,
-      "loss": 46.0,
-      "step": 33269
-    },
-    {
-      "epoch": 2.543723837375996,
-      "grad_norm": 0.0013747261837124825,
-      "learning_rate": 0.00019999680885225636,
-      "loss": 46.0,
-      "step": 33270
-    },
-    {
-      "epoch": 2.543800294359386,
-      "grad_norm": 0.0010341558372601867,
-      "learning_rate": 0.00019999680866036347,
-      "loss": 46.0,
-      "step": 33271
-    },
-    {
-      "epoch": 2.5438767513427756,
-      "grad_norm": 0.0012722710380330682,
-      "learning_rate": 0.00019999680846846484,
-      "loss": 46.0,
-      "step": 33272
-    },
-    {
-      "epoch": 2.5439532083261653,
-      "grad_norm": 0.00043182953959330916,
-      "learning_rate": 0.00019999680827656037,
-      "loss": 46.0,
-      "step": 33273
-    },
-    {
-      "epoch": 2.544029665309555,
-      "grad_norm": 0.0030942584853619337,
-      "learning_rate": 0.0001999968080846502,
-      "loss": 46.0,
-      "step": 33274
-    },
-    {
-      "epoch": 2.544106122292945,
-      "grad_norm": 0.002040538005530834,
-      "learning_rate": 0.00019999680789273424,
-      "loss": 46.0,
-      "step": 33275
-    },
-    {
-      "epoch": 2.5441825792763346,
-      "grad_norm": 0.001795622636564076,
-      "learning_rate": 0.00019999680770081248,
-      "loss": 46.0,
-      "step": 33276
-    },
-    {
-      "epoch": 2.5442590362597244,
-      "grad_norm": 0.0005658180452883244,
-      "learning_rate": 0.00019999680750888498,
-      "loss": 46.0,
-      "step": 33277
-    },
-    {
-      "epoch": 2.544335493243114,
-      "grad_norm": 0.00115725037176162,
-      "learning_rate": 0.0001999968073169517,
-      "loss": 46.0,
-      "step": 33278
-    },
-    {
-      "epoch": 2.544411950226504,
-      "grad_norm": 0.0005812536110170186,
-      "learning_rate": 0.00019999680712501269,
-      "loss": 46.0,
-      "step": 33279
-    },
-    {
-      "epoch": 2.5444884072098937,
-      "grad_norm": 0.002618224360048771,
-      "learning_rate": 0.00019999680693306786,
-      "loss": 46.0,
-      "step": 33280
-    },
-    {
-      "epoch": 2.5445648641932834,
-      "grad_norm": 0.0012272645253688097,
-      "learning_rate": 0.00019999680674111727,
-      "loss": 46.0,
-      "step": 33281
-    },
-    {
-      "epoch": 2.544641321176673,
-      "grad_norm": 0.0010032330173999071,
-      "learning_rate": 0.0001999968065491609,
-      "loss": 46.0,
-      "step": 33282
-    },
-    {
-      "epoch": 2.544717778160063,
-      "grad_norm": 0.0012789410538971424,
-      "learning_rate": 0.00019999680635719881,
-      "loss": 46.0,
-      "step": 33283
-    },
-    {
-      "epoch": 2.5447942351434523,
-      "grad_norm": 0.0009242161177098751,
-      "learning_rate": 0.0001999968061652309,
-      "loss": 46.0,
-      "step": 33284
-    },
-    {
-      "epoch": 2.544870692126842,
-      "grad_norm": 0.0009049131767824292,
-      "learning_rate": 0.00019999680597325724,
-      "loss": 46.0,
-      "step": 33285
-    },
-    {
-      "epoch": 2.544947149110232,
-      "grad_norm": 0.0017617036355659366,
-      "learning_rate": 0.00019999680578127783,
-      "loss": 46.0,
-      "step": 33286
-    },
-    {
-      "epoch": 2.5450236060936215,
-      "grad_norm": 0.0009699770016595721,
-      "learning_rate": 0.00019999680558929262,
-      "loss": 46.0,
-      "step": 33287
-    },
-    {
-      "epoch": 2.5451000630770113,
-      "grad_norm": 0.0015616221353411674,
-      "learning_rate": 0.00019999680539730167,
-      "loss": 46.0,
-      "step": 33288
-    },
-    {
-      "epoch": 2.545176520060401,
-      "grad_norm": 0.0021542140748351812,
-      "learning_rate": 0.00019999680520530494,
-      "loss": 46.0,
-      "step": 33289
-    },
-    {
-      "epoch": 2.545252977043791,
-      "grad_norm": 0.000639402074739337,
-      "learning_rate": 0.0001999968050133024,
-      "loss": 46.0,
-      "step": 33290
-    },
-    {
-      "epoch": 2.5453294340271806,
-      "grad_norm": 0.0028817288111895323,
-      "learning_rate": 0.00019999680482129414,
-      "loss": 46.0,
-      "step": 33291
-    },
-    {
-      "epoch": 2.54540589101057,
-      "grad_norm": 0.002486302051693201,
-      "learning_rate": 0.00019999680462928012,
-      "loss": 46.0,
-      "step": 33292
-    },
-    {
-      "epoch": 2.5454823479939597,
-      "grad_norm": 0.0012124533532187343,
-      "learning_rate": 0.0001999968044372603,
-      "loss": 46.0,
-      "step": 33293
-    },
-    {
-      "epoch": 2.5455588049773494,
-      "grad_norm": 0.0017838348867371678,
-      "learning_rate": 0.0001999968042452347,
-      "loss": 46.0,
-      "step": 33294
-    },
-    {
-      "epoch": 2.545635261960739,
-      "grad_norm": 0.001625219825655222,
-      "learning_rate": 0.00019999680405320336,
-      "loss": 46.0,
-      "step": 33295
-    },
-    {
-      "epoch": 2.545711718944129,
-      "grad_norm": 0.0009912997484207153,
-      "learning_rate": 0.00019999680386116624,
-      "loss": 46.0,
-      "step": 33296
-    },
-    {
-      "epoch": 2.5457881759275187,
-      "grad_norm": 0.0007726296898908913,
-      "learning_rate": 0.00019999680366912336,
-      "loss": 46.0,
-      "step": 33297
-    },
-    {
-      "epoch": 2.5458646329109085,
-      "grad_norm": 0.0005682088667526841,
-      "learning_rate": 0.00019999680347707467,
-      "loss": 46.0,
-      "step": 33298
-    },
-    {
-      "epoch": 2.5459410898942982,
-      "grad_norm": 0.0009149821707978845,
-      "learning_rate": 0.00019999680328502026,
-      "loss": 46.0,
-      "step": 33299
-    },
-    {
-      "epoch": 2.546017546877688,
-      "grad_norm": 0.0012556836009025574,
-      "learning_rate": 0.00019999680309296008,
-      "loss": 46.0,
-      "step": 33300
-    },
-    {
-      "epoch": 2.5460940038610778,
-      "grad_norm": 0.0017113303765654564,
-      "learning_rate": 0.0001999968029008941,
-      "loss": 46.0,
-      "step": 33301
-    },
-    {
-      "epoch": 2.5461704608444675,
-      "grad_norm": 0.0011956775560975075,
-      "learning_rate": 0.00019999680270882237,
-      "loss": 46.0,
-      "step": 33302
-    },
-    {
-      "epoch": 2.5462469178278573,
-      "grad_norm": 0.005448531825095415,
-      "learning_rate": 0.00019999680251674487,
-      "loss": 46.0,
-      "step": 33303
-    },
-    {
-      "epoch": 2.546323374811247,
-      "grad_norm": 0.001171428244560957,
-      "learning_rate": 0.00019999680232466157,
-      "loss": 46.0,
-      "step": 33304
-    },
-    {
-      "epoch": 2.546399831794637,
-      "grad_norm": 0.0004494576423894614,
-      "learning_rate": 0.00019999680213257252,
-      "loss": 46.0,
-      "step": 33305
-    },
-    {
-      "epoch": 2.546476288778026,
-      "grad_norm": 0.0011833661701530218,
-      "learning_rate": 0.00019999680194047773,
-      "loss": 46.0,
-      "step": 33306
-    },
-    {
-      "epoch": 2.546552745761416,
-      "grad_norm": 0.0019324348540976644,
-      "learning_rate": 0.00019999680174837713,
-      "loss": 46.0,
-      "step": 33307
-    },
-    {
-      "epoch": 2.5466292027448056,
-      "grad_norm": 0.002247591968625784,
-      "learning_rate": 0.0001999968015562708,
-      "loss": 46.0,
-      "step": 33308
-    },
-    {
-      "epoch": 2.5467056597281954,
-      "grad_norm": 0.0015211805002763867,
-      "learning_rate": 0.00019999680136415865,
-      "loss": 46.0,
-      "step": 33309
-    },
-    {
-      "epoch": 2.546782116711585,
-      "grad_norm": 0.0016391740646213293,
-      "learning_rate": 0.0001999968011720408,
-      "loss": 46.0,
-      "step": 33310
-    },
-    {
-      "epoch": 2.546858573694975,
-      "grad_norm": 0.0014959746040403843,
-      "learning_rate": 0.00019999680097991713,
-      "loss": 46.0,
-      "step": 33311
-    },
-    {
-      "epoch": 2.5469350306783647,
-      "grad_norm": 0.0007031409186311066,
-      "learning_rate": 0.0001999968007877877,
-      "loss": 46.0,
-      "step": 33312
-    },
-    {
-      "epoch": 2.5470114876617544,
-      "grad_norm": 0.0019786416087299585,
-      "learning_rate": 0.0001999968005956525,
-      "loss": 46.0,
-      "step": 33313
-    },
-    {
-      "epoch": 2.5470879446451438,
-      "grad_norm": 0.0008429271983914077,
-      "learning_rate": 0.0001999968004035115,
-      "loss": 46.0,
-      "step": 33314
-    },
-    {
-      "epoch": 2.5471644016285335,
-      "grad_norm": 0.0015400771517306566,
-      "learning_rate": 0.0001999968002113648,
-      "loss": 46.0,
-      "step": 33315
-    },
-    {
-      "epoch": 2.5472408586119233,
-      "grad_norm": 0.0009245475521311164,
-      "learning_rate": 0.00019999680001921228,
-      "loss": 46.0,
-      "step": 33316
-    },
-    {
-      "epoch": 2.547317315595313,
-      "grad_norm": 0.0016756549011915922,
-      "learning_rate": 0.00019999679982705398,
-      "loss": 46.0,
-      "step": 33317
-    },
-    {
-      "epoch": 2.547393772578703,
-      "grad_norm": 0.0011326465755701065,
-      "learning_rate": 0.00019999679963489,
-      "loss": 46.0,
-      "step": 33318
-    },
-    {
-      "epoch": 2.5474702295620926,
-      "grad_norm": 0.0005454855272546411,
-      "learning_rate": 0.00019999679944272014,
-      "loss": 46.0,
-      "step": 33319
-    },
-    {
-      "epoch": 2.5475466865454823,
-      "grad_norm": 0.0008574057137593627,
-      "learning_rate": 0.00019999679925054457,
-      "loss": 46.0,
-      "step": 33320
-    },
-    {
-      "epoch": 2.547623143528872,
-      "grad_norm": 0.0014961649430915713,
-      "learning_rate": 0.0001999967990583632,
-      "loss": 46.0,
-      "step": 33321
-    },
-    {
-      "epoch": 2.547699600512262,
-      "grad_norm": 0.0017506371950730681,
-      "learning_rate": 0.0001999967988661761,
-      "loss": 46.0,
-      "step": 33322
-    },
-    {
-      "epoch": 2.5477760574956516,
-      "grad_norm": 0.0020219087600708008,
-      "learning_rate": 0.0001999967986739832,
-      "loss": 46.0,
-      "step": 33323
-    },
-    {
-      "epoch": 2.5478525144790414,
-      "grad_norm": 0.004890563897788525,
-      "learning_rate": 0.00019999679848178454,
-      "loss": 46.0,
-      "step": 33324
-    },
-    {
-      "epoch": 2.547928971462431,
-      "grad_norm": 0.0014269209932535887,
-      "learning_rate": 0.0001999967982895801,
-      "loss": 46.0,
-      "step": 33325
-    },
-    {
-      "epoch": 2.548005428445821,
-      "grad_norm": 0.0008529675542376935,
-      "learning_rate": 0.0001999967980973699,
-      "loss": 46.0,
-      "step": 33326
-    },
-    {
-      "epoch": 2.5480818854292107,
-      "grad_norm": 0.001352623919956386,
-      "learning_rate": 0.00019999679790515392,
-      "loss": 46.0,
-      "step": 33327
-    },
-    {
-      "epoch": 2.5481583424126,
-      "grad_norm": 0.0012432413641363382,
-      "learning_rate": 0.0001999967977129322,
-      "loss": 46.0,
-      "step": 33328
-    },
-    {
-      "epoch": 2.5482347993959897,
-      "grad_norm": 0.0006741942488588393,
-      "learning_rate": 0.0001999967975207047,
-      "loss": 46.0,
-      "step": 33329
-    },
-    {
-      "epoch": 2.5483112563793795,
-      "grad_norm": 0.0006733816699124873,
-      "learning_rate": 0.0001999967973284714,
-      "loss": 46.0,
-      "step": 33330
-    },
-    {
-      "epoch": 2.5483877133627693,
-      "grad_norm": 0.0005597376730293036,
-      "learning_rate": 0.00019999679713623238,
-      "loss": 46.0,
-      "step": 33331
-    },
-    {
-      "epoch": 2.548464170346159,
-      "grad_norm": 0.0027153403498232365,
-      "learning_rate": 0.00019999679694398756,
-      "loss": 46.0,
-      "step": 33332
-    },
-    {
-      "epoch": 2.548540627329549,
-      "grad_norm": 0.0020021172240376472,
-      "learning_rate": 0.00019999679675173696,
-      "loss": 46.0,
-      "step": 33333
-    },
-    {
-      "epoch": 2.5486170843129385,
-      "grad_norm": 0.0025056200101971626,
-      "learning_rate": 0.00019999679655948062,
-      "loss": 46.0,
-      "step": 33334
-    },
-    {
-      "epoch": 2.5486935412963283,
-      "grad_norm": 0.0010182602563872933,
-      "learning_rate": 0.0001999967963672185,
-      "loss": 46.0,
-      "step": 33335
-    },
-    {
-      "epoch": 2.5487699982797176,
-      "grad_norm": 0.0009912733221426606,
-      "learning_rate": 0.00019999679617495062,
-      "loss": 46.0,
-      "step": 33336
-    },
-    {
-      "epoch": 2.5488464552631074,
-      "grad_norm": 0.0009709055302664638,
-      "learning_rate": 0.00019999679598267696,
-      "loss": 46.0,
-      "step": 33337
-    },
-    {
-      "epoch": 2.548922912246497,
-      "grad_norm": 0.001121595036238432,
-      "learning_rate": 0.00019999679579039753,
-      "loss": 46.0,
-      "step": 33338
-    },
-    {
-      "epoch": 2.548999369229887,
-      "grad_norm": 0.0010202991543337703,
-      "learning_rate": 0.0001999967955981123,
-      "loss": 46.0,
-      "step": 33339
-    },
-    {
-      "epoch": 2.5490758262132767,
-      "grad_norm": 0.0019579462241381407,
-      "learning_rate": 0.00019999679540582137,
-      "loss": 46.0,
-      "step": 33340
-    },
-    {
-      "epoch": 2.5491522831966664,
-      "grad_norm": 0.0018069371581077576,
-      "learning_rate": 0.00019999679521352461,
-      "loss": 46.0,
-      "step": 33341
-    },
-    {
-      "epoch": 2.549228740180056,
-      "grad_norm": 0.00036687328247353435,
-      "learning_rate": 0.00019999679502122211,
-      "loss": 46.0,
-      "step": 33342
-    },
-    {
-      "epoch": 2.549305197163446,
-      "grad_norm": 0.0003666691482067108,
-      "learning_rate": 0.00019999679482891381,
-      "loss": 46.0,
-      "step": 33343
-    },
-    {
-      "epoch": 2.5493816541468357,
-      "grad_norm": 0.0014866949059069157,
-      "learning_rate": 0.0001999967946365998,
-      "loss": 46.0,
-      "step": 33344
-    },
-    {
-      "epoch": 2.5494581111302255,
-      "grad_norm": 0.000989401014521718,
-      "learning_rate": 0.00019999679444427997,
-      "loss": 46.0,
-      "step": 33345
-    },
-    {
-      "epoch": 2.5495345681136152,
-      "grad_norm": 0.0017552630743011832,
-      "learning_rate": 0.00019999679425195438,
-      "loss": 46.0,
-      "step": 33346
-    },
-    {
-      "epoch": 2.549611025097005,
-      "grad_norm": 0.0023222470190376043,
-      "learning_rate": 0.00019999679405962304,
-      "loss": 46.0,
-      "step": 33347
-    },
-    {
-      "epoch": 2.5496874820803948,
-      "grad_norm": 0.0011187938507646322,
-      "learning_rate": 0.0001999967938672859,
-      "loss": 46.0,
-      "step": 33348
-    },
-    {
-      "epoch": 2.5497639390637845,
-      "grad_norm": 0.0005822560051456094,
-      "learning_rate": 0.00019999679367494302,
-      "loss": 46.0,
-      "step": 33349
-    },
-    {
-      "epoch": 2.549840396047174,
-      "grad_norm": 0.0005749605479650199,
-      "learning_rate": 0.00019999679348259438,
-      "loss": 46.0,
-      "step": 33350
-    },
-    {
-      "epoch": 2.5499168530305636,
-      "grad_norm": 0.0005238081212155521,
-      "learning_rate": 0.00019999679329023992,
-      "loss": 46.0,
-      "step": 33351
-    },
-    {
-      "epoch": 2.5499933100139534,
-      "grad_norm": 0.0006781385745853186,
-      "learning_rate": 0.00019999679309787974,
-      "loss": 46.0,
-      "step": 33352
-    },
-    {
-      "epoch": 2.550069766997343,
-      "grad_norm": 0.0012105421628803015,
-      "learning_rate": 0.00019999679290551376,
-      "loss": 46.0,
-      "step": 33353
-    },
-    {
-      "epoch": 2.550146223980733,
-      "grad_norm": 0.002104924526065588,
-      "learning_rate": 0.000199996792713142,
-      "loss": 46.0,
-      "step": 33354
-    },
-    {
-      "epoch": 2.5502226809641226,
-      "grad_norm": 0.0015381406992673874,
-      "learning_rate": 0.0001999967925207645,
-      "loss": 46.0,
-      "step": 33355
-    },
-    {
-      "epoch": 2.5502991379475124,
-      "grad_norm": 0.0014545018784701824,
-      "learning_rate": 0.00019999679232838124,
-      "loss": 46.0,
-      "step": 33356
-    },
-    {
-      "epoch": 2.550375594930902,
-      "grad_norm": 0.0015850639902055264,
-      "learning_rate": 0.00019999679213599222,
-      "loss": 46.0,
-      "step": 33357
-    },
-    {
-      "epoch": 2.5504520519142915,
-      "grad_norm": 0.0022465726360678673,
-      "learning_rate": 0.00019999679194359737,
-      "loss": 46.0,
-      "step": 33358
-    },
-    {
-      "epoch": 2.5505285088976812,
-      "grad_norm": 0.0015107401413843036,
-      "learning_rate": 0.00019999679175119678,
-      "loss": 46.0,
-      "step": 33359
-    },
-    {
-      "epoch": 2.550604965881071,
-      "grad_norm": 0.0005488633178174496,
-      "learning_rate": 0.00019999679155879044,
-      "loss": 46.0,
-      "step": 33360
-    },
-    {
-      "epoch": 2.5506814228644608,
-      "grad_norm": 0.0007519600912928581,
-      "learning_rate": 0.00019999679136637833,
-      "loss": 46.0,
-      "step": 33361
-    },
-    {
-      "epoch": 2.5507578798478505,
-      "grad_norm": 0.0017860318766906857,
-      "learning_rate": 0.0001999967911739604,
-      "loss": 46.0,
-      "step": 33362
-    },
-    {
-      "epoch": 2.5508343368312403,
-      "grad_norm": 0.0011648437939584255,
-      "learning_rate": 0.00019999679098153675,
-      "loss": 46.0,
-      "step": 33363
-    },
-    {
-      "epoch": 2.55091079381463,
-      "grad_norm": 0.002433040412142873,
-      "learning_rate": 0.00019999679078910735,
-      "loss": 46.0,
-      "step": 33364
-    },
-    {
-      "epoch": 2.55098725079802,
-      "grad_norm": 0.0032220236025750637,
-      "learning_rate": 0.00019999679059667211,
-      "loss": 46.0,
-      "step": 33365
-    },
-    {
-      "epoch": 2.5510637077814096,
-      "grad_norm": 0.0004233763611409813,
-      "learning_rate": 0.00019999679040423116,
-      "loss": 46.0,
-      "step": 33366
-    },
-    {
-      "epoch": 2.5511401647647993,
-      "grad_norm": 0.0004299703286960721,
-      "learning_rate": 0.00019999679021178444,
-      "loss": 46.0,
-      "step": 33367
-    },
-    {
-      "epoch": 2.551216621748189,
-      "grad_norm": 0.001925118500366807,
-      "learning_rate": 0.0001999967900193319,
-      "loss": 46.0,
-      "step": 33368
-    },
-    {
-      "epoch": 2.551293078731579,
-      "grad_norm": 0.0017158965347334743,
-      "learning_rate": 0.0001999967898268736,
-      "loss": 46.0,
-      "step": 33369
-    },
-    {
-      "epoch": 2.5513695357149686,
-      "grad_norm": 0.00152087421156466,
-      "learning_rate": 0.00019999678963440956,
-      "loss": 46.0,
-      "step": 33370
-    },
-    {
-      "epoch": 2.551445992698358,
-      "grad_norm": 0.0013696595560759306,
-      "learning_rate": 0.00019999678944193977,
-      "loss": 46.0,
-      "step": 33371
-    },
-    {
-      "epoch": 2.5515224496817477,
-      "grad_norm": 0.001246079453267157,
-      "learning_rate": 0.00019999678924946418,
-      "loss": 46.0,
-      "step": 33372
-    },
-    {
-      "epoch": 2.5515989066651374,
-      "grad_norm": 0.0010100255021825433,
-      "learning_rate": 0.00019999678905698284,
-      "loss": 46.0,
-      "step": 33373
-    },
-    {
-      "epoch": 2.551675363648527,
-      "grad_norm": 0.0011573780793696642,
-      "learning_rate": 0.0001999967888644957,
-      "loss": 46.0,
-      "step": 33374
-    },
-    {
-      "epoch": 2.551751820631917,
-      "grad_norm": 0.0016226014122366905,
-      "learning_rate": 0.0001999967886720028,
-      "loss": 46.0,
-      "step": 33375
-    },
-    {
-      "epoch": 2.5518282776153067,
-      "grad_norm": 0.0026063635013997555,
-      "learning_rate": 0.00019999678847950416,
-      "loss": 46.0,
-      "step": 33376
-    },
-    {
-      "epoch": 2.5519047345986965,
-      "grad_norm": 0.0005384384421631694,
-      "learning_rate": 0.00019999678828699972,
-      "loss": 46.0,
-      "step": 33377
-    },
-    {
-      "epoch": 2.5519811915820863,
-      "grad_norm": 0.0006929469527676702,
-      "learning_rate": 0.00019999678809448952,
-      "loss": 46.0,
-      "step": 33378
-    },
-    {
-      "epoch": 2.552057648565476,
-      "grad_norm": 0.00575612485408783,
-      "learning_rate": 0.00019999678790197356,
-      "loss": 46.0,
-      "step": 33379
-    },
-    {
-      "epoch": 2.5521341055488653,
-      "grad_norm": 0.0017424135003238916,
-      "learning_rate": 0.00019999678770945184,
-      "loss": 46.0,
-      "step": 33380
-    },
-    {
-      "epoch": 2.552210562532255,
-      "grad_norm": 0.0011673882836475968,
-      "learning_rate": 0.0001999967875169243,
-      "loss": 46.0,
-      "step": 33381
-    },
-    {
-      "epoch": 2.552287019515645,
-      "grad_norm": 0.0013589125592261553,
-      "learning_rate": 0.000199996787324391,
-      "loss": 46.0,
-      "step": 33382
-    },
-    {
-      "epoch": 2.5523634764990346,
-      "grad_norm": 0.0007842128979973495,
-      "learning_rate": 0.000199996787131852,
-      "loss": 46.0,
-      "step": 33383
-    },
-    {
-      "epoch": 2.5524399334824244,
-      "grad_norm": 0.0025036949664354324,
-      "learning_rate": 0.00019999678693930718,
-      "loss": 46.0,
-      "step": 33384
-    },
-    {
-      "epoch": 2.552516390465814,
-      "grad_norm": 0.0018862350843846798,
-      "learning_rate": 0.00019999678674675658,
-      "loss": 46.0,
-      "step": 33385
-    },
-    {
-      "epoch": 2.552592847449204,
-      "grad_norm": 0.006162185687571764,
-      "learning_rate": 0.00019999678655420022,
-      "loss": 46.0,
-      "step": 33386
-    },
-    {
-      "epoch": 2.5526693044325937,
-      "grad_norm": 0.005736600141972303,
-      "learning_rate": 0.0001999967863616381,
-      "loss": 46.0,
-      "step": 33387
-    },
-    {
-      "epoch": 2.5527457614159834,
-      "grad_norm": 0.0017516075167804956,
-      "learning_rate": 0.0001999967861690702,
-      "loss": 46.0,
-      "step": 33388
-    },
-    {
-      "epoch": 2.552822218399373,
-      "grad_norm": 0.0022579121869057417,
-      "learning_rate": 0.00019999678597649653,
-      "loss": 46.0,
-      "step": 33389
-    },
-    {
-      "epoch": 2.552898675382763,
-      "grad_norm": 0.0016126515110954642,
-      "learning_rate": 0.00019999678578391713,
-      "loss": 46.0,
-      "step": 33390
-    },
-    {
-      "epoch": 2.5529751323661527,
-      "grad_norm": 0.001691891928203404,
-      "learning_rate": 0.0001999967855913319,
-      "loss": 46.0,
-      "step": 33391
-    },
-    {
-      "epoch": 2.5530515893495425,
-      "grad_norm": 0.0013627305161207914,
-      "learning_rate": 0.00019999678539874094,
-      "loss": 46.0,
-      "step": 33392
-    },
-    {
-      "epoch": 2.5531280463329318,
-      "grad_norm": 0.00231366278603673,
-      "learning_rate": 0.0001999967852061442,
-      "loss": 46.0,
-      "step": 33393
-    },
-    {
-      "epoch": 2.5532045033163215,
-      "grad_norm": 0.00300302030518651,
-      "learning_rate": 0.0001999967850135417,
-      "loss": 46.0,
-      "step": 33394
-    },
-    {
-      "epoch": 2.5532809602997113,
-      "grad_norm": 0.001603800104930997,
-      "learning_rate": 0.00019999678482093342,
-      "loss": 46.0,
-      "step": 33395
-    },
-    {
-      "epoch": 2.553357417283101,
-      "grad_norm": 0.0015295346966013312,
-      "learning_rate": 0.00019999678462831938,
-      "loss": 46.0,
-      "step": 33396
-    },
-    {
-      "epoch": 2.553433874266491,
-      "grad_norm": 0.001510647707618773,
-      "learning_rate": 0.00019999678443569953,
-      "loss": 46.0,
-      "step": 33397
-    },
-    {
-      "epoch": 2.5535103312498806,
-      "grad_norm": 0.0020446134731173515,
-      "learning_rate": 0.00019999678424307397,
-      "loss": 46.0,
-      "step": 33398
-    },
-    {
-      "epoch": 2.5535867882332703,
-      "grad_norm": 0.0016776991542428732,
-      "learning_rate": 0.0001999967840504426,
-      "loss": 46.0,
-      "step": 33399
-    },
-    {
-      "epoch": 2.55366324521666,
-      "grad_norm": 0.0006239119102247059,
-      "learning_rate": 0.0001999967838578055,
-      "loss": 46.0,
-      "step": 33400
-    },
-    {
-      "epoch": 2.5537397022000494,
-      "grad_norm": 0.002045438624918461,
-      "learning_rate": 0.0001999967836651626,
-      "loss": 46.0,
-      "step": 33401
-    },
-    {
-      "epoch": 2.553816159183439,
-      "grad_norm": 0.008139276877045631,
-      "learning_rate": 0.00019999678347251392,
-      "loss": 46.0,
-      "step": 33402
-    },
-    {
-      "epoch": 2.553892616166829,
-      "grad_norm": 0.002613227814435959,
-      "learning_rate": 0.00019999678327985949,
-      "loss": 46.0,
-      "step": 33403
-    },
-    {
-      "epoch": 2.5539690731502187,
-      "grad_norm": 0.001017614733427763,
-      "learning_rate": 0.00019999678308719928,
-      "loss": 46.0,
-      "step": 33404
-    },
-    {
-      "epoch": 2.5540455301336085,
-      "grad_norm": 0.005765401292592287,
-      "learning_rate": 0.00019999678289453333,
-      "loss": 46.0,
-      "step": 33405
-    },
-    {
-      "epoch": 2.5541219871169982,
-      "grad_norm": 0.0026276176795363426,
-      "learning_rate": 0.00019999678270186158,
-      "loss": 46.0,
-      "step": 33406
-    },
-    {
-      "epoch": 2.554198444100388,
-      "grad_norm": 0.0007947334670461714,
-      "learning_rate": 0.00019999678250918405,
-      "loss": 46.0,
-      "step": 33407
-    },
-    {
-      "epoch": 2.5542749010837777,
-      "grad_norm": 0.0008668103255331516,
-      "learning_rate": 0.00019999678231650078,
-      "loss": 46.0,
-      "step": 33408
-    },
-    {
-      "epoch": 2.5543513580671675,
-      "grad_norm": 0.0006482668686658144,
-      "learning_rate": 0.00019999678212381173,
-      "loss": 46.0,
-      "step": 33409
-    },
-    {
-      "epoch": 2.5544278150505573,
-      "grad_norm": 0.0008261186303570867,
-      "learning_rate": 0.00019999678193111692,
-      "loss": 46.0,
-      "step": 33410
-    },
-    {
-      "epoch": 2.554504272033947,
-      "grad_norm": 0.002974668750539422,
-      "learning_rate": 0.00019999678173841635,
-      "loss": 46.0,
-      "step": 33411
-    },
-    {
-      "epoch": 2.554580729017337,
-      "grad_norm": 0.0009730941383168101,
-      "learning_rate": 0.00019999678154570999,
-      "loss": 46.0,
-      "step": 33412
-    },
-    {
-      "epoch": 2.5546571860007266,
-      "grad_norm": 0.001791209913790226,
-      "learning_rate": 0.00019999678135299785,
-      "loss": 46.0,
-      "step": 33413
-    },
-    {
-      "epoch": 2.5547336429841163,
-      "grad_norm": 0.0011224858462810516,
-      "learning_rate": 0.00019999678116027996,
-      "loss": 46.0,
-      "step": 33414
-    },
-    {
-      "epoch": 2.5548100999675056,
-      "grad_norm": 0.0017287597293034196,
-      "learning_rate": 0.0001999967809675563,
-      "loss": 46.0,
-      "step": 33415
-    },
-    {
-      "epoch": 2.5548865569508954,
-      "grad_norm": 0.001097777159884572,
-      "learning_rate": 0.00019999678077482687,
-      "loss": 46.0,
-      "step": 33416
-    },
-    {
-      "epoch": 2.554963013934285,
-      "grad_norm": 0.001997461309656501,
-      "learning_rate": 0.00019999678058209167,
-      "loss": 46.0,
-      "step": 33417
-    },
-    {
-      "epoch": 2.555039470917675,
-      "grad_norm": 0.0033196588046848774,
-      "learning_rate": 0.0001999967803893507,
-      "loss": 46.0,
-      "step": 33418
-    },
-    {
-      "epoch": 2.5551159279010647,
-      "grad_norm": 0.003946193493902683,
-      "learning_rate": 0.00019999678019660394,
-      "loss": 46.0,
-      "step": 33419
-    },
-    {
-      "epoch": 2.5551923848844544,
-      "grad_norm": 0.0007623904966749251,
-      "learning_rate": 0.00019999678000385144,
-      "loss": 46.0,
-      "step": 33420
-    },
-    {
-      "epoch": 2.555268841867844,
-      "grad_norm": 0.002262078458443284,
-      "learning_rate": 0.00019999677981109314,
-      "loss": 46.0,
-      "step": 33421
-    },
-    {
-      "epoch": 2.555345298851234,
-      "grad_norm": 0.0016731873620301485,
-      "learning_rate": 0.00019999677961832912,
-      "loss": 46.0,
-      "step": 33422
-    },
-    {
-      "epoch": 2.5554217558346233,
-      "grad_norm": 0.0014001419767737389,
-      "learning_rate": 0.00019999677942555928,
-      "loss": 46.0,
-      "step": 33423
-    },
-    {
-      "epoch": 2.555498212818013,
-      "grad_norm": 0.0010974627221003175,
-      "learning_rate": 0.00019999677923278371,
-      "loss": 46.0,
-      "step": 33424
-    },
-    {
-      "epoch": 2.555574669801403,
-      "grad_norm": 0.0031964555382728577,
-      "learning_rate": 0.00019999677904000235,
-      "loss": 46.0,
-      "step": 33425
-    },
-    {
-      "epoch": 2.5556511267847926,
-      "grad_norm": 0.002898680279031396,
-      "learning_rate": 0.00019999677884721524,
-      "loss": 46.0,
-      "step": 33426
-    },
-    {
-      "epoch": 2.5557275837681823,
-      "grad_norm": 0.0013742714654654264,
-      "learning_rate": 0.00019999677865442235,
-      "loss": 46.0,
-      "step": 33427
-    },
-    {
-      "epoch": 2.555804040751572,
-      "grad_norm": 0.0030296859331429005,
-      "learning_rate": 0.00019999677846162367,
-      "loss": 46.0,
-      "step": 33428
-    },
-    {
-      "epoch": 2.555880497734962,
-      "grad_norm": 0.001304943929426372,
-      "learning_rate": 0.00019999677826881924,
-      "loss": 46.0,
-      "step": 33429
-    },
-    {
-      "epoch": 2.5559569547183516,
-      "grad_norm": 0.0012732669711112976,
-      "learning_rate": 0.00019999677807600903,
-      "loss": 46.0,
-      "step": 33430
-    },
-    {
-      "epoch": 2.5560334117017414,
-      "grad_norm": 0.005901511758565903,
-      "learning_rate": 0.00019999677788319305,
-      "loss": 46.0,
-      "step": 33431
-    },
-    {
-      "epoch": 2.556109868685131,
-      "grad_norm": 0.001071861363016069,
-      "learning_rate": 0.00019999677769037133,
-      "loss": 46.0,
-      "step": 33432
-    },
-    {
-      "epoch": 2.556186325668521,
-      "grad_norm": 0.014260247349739075,
-      "learning_rate": 0.0001999967774975438,
-      "loss": 46.0,
-      "step": 33433
-    },
-    {
-      "epoch": 2.5562627826519106,
-      "grad_norm": 0.0006738616502843797,
-      "learning_rate": 0.00019999677730471053,
-      "loss": 46.0,
-      "step": 33434
-    },
-    {
-      "epoch": 2.5563392396353004,
-      "grad_norm": 0.0005618222639895976,
-      "learning_rate": 0.0001999967771118715,
-      "loss": 46.0,
-      "step": 33435
-    },
-    {
-      "epoch": 2.55641569661869,
-      "grad_norm": 0.0009934313129633665,
-      "learning_rate": 0.00019999677691902664,
-      "loss": 46.0,
-      "step": 33436
-    },
-    {
-      "epoch": 2.5564921536020795,
-      "grad_norm": 0.001667368458583951,
-      "learning_rate": 0.00019999677672617608,
-      "loss": 46.0,
-      "step": 33437
-    },
-    {
-      "epoch": 2.5565686105854692,
-      "grad_norm": 0.0036817453801631927,
-      "learning_rate": 0.00019999677653331972,
-      "loss": 46.0,
-      "step": 33438
-    },
-    {
-      "epoch": 2.556645067568859,
-      "grad_norm": 0.001768269226886332,
-      "learning_rate": 0.0001999967763404576,
-      "loss": 46.0,
-      "step": 33439
-    },
-    {
-      "epoch": 2.5567215245522488,
-      "grad_norm": 0.0015852535143494606,
-      "learning_rate": 0.0001999967761475897,
-      "loss": 46.0,
-      "step": 33440
-    },
-    {
-      "epoch": 2.5567979815356385,
-      "grad_norm": 0.0037774841766804457,
-      "learning_rate": 0.000199996775954716,
-      "loss": 46.0,
-      "step": 33441
-    },
-    {
-      "epoch": 2.5568744385190283,
-      "grad_norm": 0.0019883010536432266,
-      "learning_rate": 0.0001999967757618366,
-      "loss": 46.0,
-      "step": 33442
-    },
-    {
-      "epoch": 2.556950895502418,
-      "grad_norm": 0.0014235106064006686,
-      "learning_rate": 0.0001999967755689514,
-      "loss": 46.0,
-      "step": 33443
-    },
-    {
-      "epoch": 2.557027352485808,
-      "grad_norm": 0.0007633913191966712,
-      "learning_rate": 0.0001999967753760604,
-      "loss": 46.0,
-      "step": 33444
-    },
-    {
-      "epoch": 2.557103809469197,
-      "grad_norm": 0.0004396808799356222,
-      "learning_rate": 0.0001999967751831637,
-      "loss": 46.0,
-      "step": 33445
-    },
-    {
-      "epoch": 2.557180266452587,
-      "grad_norm": 0.0007296592812053859,
-      "learning_rate": 0.00019999677499026118,
-      "loss": 46.0,
-      "step": 33446
-    },
-    {
-      "epoch": 2.5572567234359767,
-      "grad_norm": 0.0022788874339312315,
-      "learning_rate": 0.00019999677479735288,
-      "loss": 46.0,
-      "step": 33447
-    },
-    {
-      "epoch": 2.5573331804193664,
-      "grad_norm": 0.0008734318544156849,
-      "learning_rate": 0.00019999677460443883,
-      "loss": 46.0,
-      "step": 33448
-    },
-    {
-      "epoch": 2.557409637402756,
-      "grad_norm": 0.0006941691972315311,
-      "learning_rate": 0.00019999677441151902,
-      "loss": 46.0,
-      "step": 33449
-    },
-    {
-      "epoch": 2.557486094386146,
-      "grad_norm": 0.004113301634788513,
-      "learning_rate": 0.00019999677421859345,
-      "loss": 46.0,
-      "step": 33450
-    },
-    {
-      "epoch": 2.5575625513695357,
-      "grad_norm": 0.0012209225678816438,
-      "learning_rate": 0.0001999967740256621,
-      "loss": 46.0,
-      "step": 33451
-    },
-    {
-      "epoch": 2.5576390083529255,
-      "grad_norm": 0.0018497559940442443,
-      "learning_rate": 0.00019999677383272495,
-      "loss": 46.0,
-      "step": 33452
-    },
-    {
-      "epoch": 2.557715465336315,
-      "grad_norm": 0.0012195002054795623,
-      "learning_rate": 0.00019999677363978204,
-      "loss": 46.0,
-      "step": 33453
-    },
-    {
-      "epoch": 2.557791922319705,
-      "grad_norm": 0.0019482439383864403,
-      "learning_rate": 0.00019999677344683338,
-      "loss": 46.0,
-      "step": 33454
-    },
-    {
-      "epoch": 2.5578683793030947,
-      "grad_norm": 0.0013766209594905376,
-      "learning_rate": 0.00019999677325387895,
-      "loss": 46.0,
-      "step": 33455
-    },
-    {
-      "epoch": 2.5579448362864845,
-      "grad_norm": 0.0021581826731562614,
-      "learning_rate": 0.00019999677306091872,
-      "loss": 46.0,
-      "step": 33456
-    },
-    {
-      "epoch": 2.5580212932698743,
-      "grad_norm": 0.003526849439367652,
-      "learning_rate": 0.00019999677286795277,
-      "loss": 46.0,
-      "step": 33457
-    },
-    {
-      "epoch": 2.558097750253264,
-      "grad_norm": 0.0009136443841271102,
-      "learning_rate": 0.00019999677267498102,
-      "loss": 46.0,
-      "step": 33458
-    },
-    {
-      "epoch": 2.5581742072366533,
-      "grad_norm": 0.0012041646987199783,
-      "learning_rate": 0.0001999967724820035,
-      "loss": 46.0,
-      "step": 33459
-    },
-    {
-      "epoch": 2.558250664220043,
-      "grad_norm": 0.0011009437730535865,
-      "learning_rate": 0.00019999677228902022,
-      "loss": 46.0,
-      "step": 33460
-    },
-    {
-      "epoch": 2.558327121203433,
-      "grad_norm": 0.0035847376566380262,
-      "learning_rate": 0.00019999677209603118,
-      "loss": 46.0,
-      "step": 33461
-    },
-    {
-      "epoch": 2.5584035781868226,
-      "grad_norm": 0.003409388242289424,
-      "learning_rate": 0.00019999677190303636,
-      "loss": 46.0,
-      "step": 33462
-    },
-    {
-      "epoch": 2.5584800351702124,
-      "grad_norm": 0.000990078435279429,
-      "learning_rate": 0.00019999677171003577,
-      "loss": 46.0,
-      "step": 33463
-    },
-    {
-      "epoch": 2.558556492153602,
-      "grad_norm": 0.0005495616933330894,
-      "learning_rate": 0.00019999677151702944,
-      "loss": 46.0,
-      "step": 33464
-    },
-    {
-      "epoch": 2.558632949136992,
-      "grad_norm": 0.0006910604424774647,
-      "learning_rate": 0.0001999967713240173,
-      "loss": 46.0,
-      "step": 33465
-    },
-    {
-      "epoch": 2.5587094061203817,
-      "grad_norm": 0.0028846152126789093,
-      "learning_rate": 0.0001999967711309994,
-      "loss": 46.0,
-      "step": 33466
-    },
-    {
-      "epoch": 2.558785863103771,
-      "grad_norm": 0.0016454203287139535,
-      "learning_rate": 0.00019999677093797573,
-      "loss": 46.0,
-      "step": 33467
-    },
-    {
-      "epoch": 2.5588623200871607,
-      "grad_norm": 0.0008280647452920675,
-      "learning_rate": 0.0001999967707449463,
-      "loss": 46.0,
-      "step": 33468
-    },
-    {
-      "epoch": 2.5589387770705505,
-      "grad_norm": 0.0014637556159868836,
-      "learning_rate": 0.0001999967705519111,
-      "loss": 46.0,
-      "step": 33469
-    },
-    {
-      "epoch": 2.5590152340539403,
-      "grad_norm": 0.001353220664896071,
-      "learning_rate": 0.00019999677035887012,
-      "loss": 46.0,
-      "step": 33470
-    },
-    {
-      "epoch": 2.55909169103733,
-      "grad_norm": 0.001629892736673355,
-      "learning_rate": 0.00019999677016582337,
-      "loss": 46.0,
-      "step": 33471
-    },
-    {
-      "epoch": 2.55916814802072,
-      "grad_norm": 0.003064531832933426,
-      "learning_rate": 0.00019999676997277085,
-      "loss": 46.0,
-      "step": 33472
-    },
-    {
-      "epoch": 2.5592446050041096,
-      "grad_norm": 0.005234165582805872,
-      "learning_rate": 0.00019999676977971258,
-      "loss": 46.0,
-      "step": 33473
-    },
-    {
-      "epoch": 2.5593210619874993,
-      "grad_norm": 0.0006723539554513991,
-      "learning_rate": 0.00019999676958664853,
-      "loss": 46.0,
-      "step": 33474
-    },
-    {
-      "epoch": 2.559397518970889,
-      "grad_norm": 0.0031113573350012302,
-      "learning_rate": 0.00019999676939357872,
-      "loss": 46.0,
-      "step": 33475
-    },
-    {
-      "epoch": 2.559473975954279,
-      "grad_norm": 0.00047030122368596494,
-      "learning_rate": 0.00019999676920050315,
-      "loss": 46.0,
-      "step": 33476
-    },
-    {
-      "epoch": 2.5595504329376686,
-      "grad_norm": 0.0019747463520616293,
-      "learning_rate": 0.00019999676900742176,
-      "loss": 46.0,
-      "step": 33477
-    },
-    {
-      "epoch": 2.5596268899210584,
-      "grad_norm": 0.0023429852444678545,
-      "learning_rate": 0.00019999676881433465,
-      "loss": 46.0,
-      "step": 33478
-    },
-    {
-      "epoch": 2.559703346904448,
-      "grad_norm": 0.005712544545531273,
-      "learning_rate": 0.00019999676862124174,
-      "loss": 46.0,
-      "step": 33479
-    },
-    {
-      "epoch": 2.559779803887838,
-      "grad_norm": 0.001561876735650003,
-      "learning_rate": 0.00019999676842814308,
-      "loss": 46.0,
-      "step": 33480
-    },
-    {
-      "epoch": 2.559856260871227,
-      "grad_norm": 0.00130911183077842,
-      "learning_rate": 0.00019999676823503863,
-      "loss": 46.0,
-      "step": 33481
-    },
-    {
-      "epoch": 2.559932717854617,
-      "grad_norm": 0.0027656827587634325,
-      "learning_rate": 0.00019999676804192845,
-      "loss": 46.0,
-      "step": 33482
-    },
-    {
-      "epoch": 2.5600091748380067,
-      "grad_norm": 0.0010164499981328845,
-      "learning_rate": 0.00019999676784881247,
-      "loss": 46.0,
-      "step": 33483
-    },
-    {
-      "epoch": 2.5600856318213965,
-      "grad_norm": 0.0018488691421225667,
-      "learning_rate": 0.00019999676765569072,
-      "loss": 46.0,
-      "step": 33484
-    },
-    {
-      "epoch": 2.5601620888047862,
-      "grad_norm": 0.0022583925165235996,
-      "learning_rate": 0.00019999676746256323,
-      "loss": 46.0,
-      "step": 33485
-    },
-    {
-      "epoch": 2.560238545788176,
-      "grad_norm": 0.001279329531826079,
-      "learning_rate": 0.00019999676726942993,
-      "loss": 46.0,
-      "step": 33486
-    },
-    {
-      "epoch": 2.5603150027715658,
-      "grad_norm": 0.002781086368486285,
-      "learning_rate": 0.00019999676707629086,
-      "loss": 46.0,
-      "step": 33487
-    },
-    {
-      "epoch": 2.5603914597549555,
-      "grad_norm": 0.0018222532235085964,
-      "learning_rate": 0.00019999676688314607,
-      "loss": 46.0,
-      "step": 33488
-    },
-    {
-      "epoch": 2.560467916738345,
-      "grad_norm": 0.00404335418716073,
-      "learning_rate": 0.00019999676668999548,
-      "loss": 46.0,
-      "step": 33489
-    },
-    {
-      "epoch": 2.5605443737217346,
-      "grad_norm": 0.0010960038052871823,
-      "learning_rate": 0.0001999967664968391,
-      "loss": 46.0,
-      "step": 33490
-    },
-    {
-      "epoch": 2.5606208307051244,
-      "grad_norm": 0.0065406016074121,
-      "learning_rate": 0.00019999676630367698,
-      "loss": 46.0,
-      "step": 33491
-    },
-    {
-      "epoch": 2.560697287688514,
-      "grad_norm": 0.0009528365335427225,
-      "learning_rate": 0.00019999676611050907,
-      "loss": 46.0,
-      "step": 33492
-    },
-    {
-      "epoch": 2.560773744671904,
-      "grad_norm": 0.0016528575215488672,
-      "learning_rate": 0.00019999676591733541,
-      "loss": 46.0,
-      "step": 33493
-    },
-    {
-      "epoch": 2.5608502016552936,
-      "grad_norm": 0.0014290288090705872,
-      "learning_rate": 0.00019999676572415599,
-      "loss": 46.0,
-      "step": 33494
-    },
-    {
-      "epoch": 2.5609266586386834,
-      "grad_norm": 0.0037852406967431307,
-      "learning_rate": 0.00019999676553097078,
-      "loss": 46.0,
-      "step": 33495
-    },
-    {
-      "epoch": 2.561003115622073,
-      "grad_norm": 0.0017256696010008454,
-      "learning_rate": 0.0001999967653377798,
-      "loss": 46.0,
-      "step": 33496
-    },
-    {
-      "epoch": 2.561079572605463,
-      "grad_norm": 0.0007360866293311119,
-      "learning_rate": 0.00019999676514458306,
-      "loss": 46.0,
-      "step": 33497
-    },
-    {
-      "epoch": 2.5611560295888527,
-      "grad_norm": 0.0005949679180048406,
-      "learning_rate": 0.00019999676495138053,
-      "loss": 46.0,
-      "step": 33498
-    },
-    {
-      "epoch": 2.5612324865722425,
-      "grad_norm": 0.0014350616838783026,
-      "learning_rate": 0.00019999676475817226,
-      "loss": 46.0,
-      "step": 33499
-    },
-    {
-      "epoch": 2.561308943555632,
-      "grad_norm": 0.0010082963854074478,
-      "learning_rate": 0.00019999676456495822,
-      "loss": 46.0,
-      "step": 33500
-    },
-    {
-      "epoch": 2.561385400539022,
-      "grad_norm": 0.0010203431593254209,
-      "learning_rate": 0.00019999676437173838,
-      "loss": 46.0,
-      "step": 33501
-    },
-    {
-      "epoch": 2.5614618575224113,
-      "grad_norm": 0.001922943745739758,
-      "learning_rate": 0.00019999676417851281,
-      "loss": 46.0,
-      "step": 33502
-    },
-    {
-      "epoch": 2.561538314505801,
-      "grad_norm": 0.0004550886806100607,
-      "learning_rate": 0.00019999676398528145,
-      "loss": 46.0,
-      "step": 33503
-    },
-    {
-      "epoch": 2.561614771489191,
-      "grad_norm": 0.0019321753643453121,
-      "learning_rate": 0.00019999676379204432,
-      "loss": 46.0,
-      "step": 33504
-    },
-    {
-      "epoch": 2.5616912284725806,
-      "grad_norm": 0.0015047990018501878,
-      "learning_rate": 0.0001999967635988014,
-      "loss": 46.0,
-      "step": 33505
-    },
-    {
-      "epoch": 2.5617676854559703,
-      "grad_norm": 0.0038688669446855783,
-      "learning_rate": 0.00019999676340555275,
-      "loss": 46.0,
-      "step": 33506
-    },
-    {
-      "epoch": 2.56184414243936,
-      "grad_norm": 0.001031518098898232,
-      "learning_rate": 0.00019999676321229832,
-      "loss": 46.0,
-      "step": 33507
-    },
-    {
-      "epoch": 2.56192059942275,
-      "grad_norm": 0.001505220658145845,
-      "learning_rate": 0.00019999676301903812,
-      "loss": 46.0,
-      "step": 33508
-    },
-    {
-      "epoch": 2.5619970564061396,
-      "grad_norm": 0.0029162021819502115,
-      "learning_rate": 0.00019999676282577214,
-      "loss": 46.0,
-      "step": 33509
-    },
-    {
-      "epoch": 2.5620735133895294,
-      "grad_norm": 0.013235975056886673,
-      "learning_rate": 0.0001999967626325004,
-      "loss": 46.0,
-      "step": 33510
-    },
-    {
-      "epoch": 2.5621499703729187,
-      "grad_norm": 0.0014610809739679098,
-      "learning_rate": 0.00019999676243922287,
-      "loss": 46.0,
-      "step": 33511
-    },
-    {
-      "epoch": 2.5622264273563085,
-      "grad_norm": 0.0014320885529741645,
-      "learning_rate": 0.0001999967622459396,
-      "loss": 46.0,
-      "step": 33512
-    },
-    {
-      "epoch": 2.562302884339698,
-      "grad_norm": 0.0010532052256166935,
-      "learning_rate": 0.00019999676205265053,
-      "loss": 46.0,
-      "step": 33513
-    },
-    {
-      "epoch": 2.562379341323088,
-      "grad_norm": 0.0010041113710030913,
-      "learning_rate": 0.00019999676185935571,
-      "loss": 46.0,
-      "step": 33514
-    },
-    {
-      "epoch": 2.5624557983064777,
-      "grad_norm": 0.0019522263901308179,
-      "learning_rate": 0.00019999676166605512,
-      "loss": 46.0,
-      "step": 33515
-    },
-    {
-      "epoch": 2.5625322552898675,
-      "grad_norm": 0.001968939322978258,
-      "learning_rate": 0.00019999676147274876,
-      "loss": 46.0,
-      "step": 33516
-    },
-    {
-      "epoch": 2.5626087122732573,
-      "grad_norm": 0.0012359711108729243,
-      "learning_rate": 0.00019999676127943663,
-      "loss": 46.0,
-      "step": 33517
-    },
-    {
-      "epoch": 2.562685169256647,
-      "grad_norm": 0.0009915182599797845,
-      "learning_rate": 0.00019999676108611874,
-      "loss": 46.0,
-      "step": 33518
-    },
-    {
-      "epoch": 2.562761626240037,
-      "grad_norm": 0.005109227262437344,
-      "learning_rate": 0.00019999676089279506,
-      "loss": 46.0,
-      "step": 33519
-    },
-    {
-      "epoch": 2.5628380832234265,
-      "grad_norm": 0.0030386755242943764,
-      "learning_rate": 0.00019999676069946563,
-      "loss": 46.0,
-      "step": 33520
-    },
-    {
-      "epoch": 2.5629145402068163,
-      "grad_norm": 0.003647509263828397,
-      "learning_rate": 0.00019999676050613043,
-      "loss": 46.0,
-      "step": 33521
-    },
-    {
-      "epoch": 2.562990997190206,
-      "grad_norm": 0.001174344215542078,
-      "learning_rate": 0.00019999676031278945,
-      "loss": 46.0,
-      "step": 33522
-    },
-    {
-      "epoch": 2.563067454173596,
-      "grad_norm": 0.0018427810864523053,
-      "learning_rate": 0.0001999967601194427,
-      "loss": 46.0,
-      "step": 33523
-    },
-    {
-      "epoch": 2.563143911156985,
-      "grad_norm": 0.002110244706273079,
-      "learning_rate": 0.00019999675992609018,
-      "loss": 46.0,
-      "step": 33524
-    },
-    {
-      "epoch": 2.563220368140375,
-      "grad_norm": 0.002172976965084672,
-      "learning_rate": 0.0001999967597327319,
-      "loss": 46.0,
-      "step": 33525
-    },
-    {
-      "epoch": 2.5632968251237647,
-      "grad_norm": 0.002311317715793848,
-      "learning_rate": 0.00019999675953936787,
-      "loss": 46.0,
-      "step": 33526
-    },
-    {
-      "epoch": 2.5633732821071544,
-      "grad_norm": 0.0005780644714832306,
-      "learning_rate": 0.00019999675934599803,
-      "loss": 46.0,
-      "step": 33527
-    },
-    {
-      "epoch": 2.563449739090544,
-      "grad_norm": 0.002176650334149599,
-      "learning_rate": 0.00019999675915262244,
-      "loss": 46.0,
-      "step": 33528
-    },
-    {
-      "epoch": 2.563526196073934,
-      "grad_norm": 0.000832623743917793,
-      "learning_rate": 0.00019999675895924108,
-      "loss": 46.0,
-      "step": 33529
-    },
-    {
-      "epoch": 2.5636026530573237,
-      "grad_norm": 0.0008746240637265146,
-      "learning_rate": 0.00019999675876585394,
-      "loss": 46.0,
-      "step": 33530
-    },
-    {
-      "epoch": 2.5636791100407135,
-      "grad_norm": 0.0012328110169619322,
-      "learning_rate": 0.00019999675857246106,
-      "loss": 46.0,
-      "step": 33531
-    },
-    {
-      "epoch": 2.563755567024103,
-      "grad_norm": 0.0006522799958474934,
-      "learning_rate": 0.00019999675837906238,
-      "loss": 46.0,
-      "step": 33532
-    },
-    {
-      "epoch": 2.5638320240074925,
-      "grad_norm": 0.0017877287464216352,
-      "learning_rate": 0.00019999675818565795,
-      "loss": 46.0,
-      "step": 33533
-    },
-    {
-      "epoch": 2.5639084809908823,
-      "grad_norm": 0.0022729779593646526,
-      "learning_rate": 0.00019999675799224775,
-      "loss": 46.0,
-      "step": 33534
-    },
-    {
-      "epoch": 2.563984937974272,
-      "grad_norm": 0.0014852328458800912,
-      "learning_rate": 0.00019999675779883177,
-      "loss": 46.0,
-      "step": 33535
-    },
-    {
-      "epoch": 2.564061394957662,
-      "grad_norm": 0.0029160999692976475,
-      "learning_rate": 0.00019999675760541002,
-      "loss": 46.0,
-      "step": 33536
-    },
-    {
-      "epoch": 2.5641378519410516,
-      "grad_norm": 0.0014590134378522635,
-      "learning_rate": 0.0001999967574119825,
-      "loss": 46.0,
-      "step": 33537
-    },
-    {
-      "epoch": 2.5642143089244414,
-      "grad_norm": 0.0007296906551346183,
-      "learning_rate": 0.00019999675721854923,
-      "loss": 46.0,
-      "step": 33538
-    },
-    {
-      "epoch": 2.564290765907831,
-      "grad_norm": 0.008215557783842087,
-      "learning_rate": 0.0001999967570251102,
-      "loss": 46.0,
-      "step": 33539
-    },
-    {
-      "epoch": 2.564367222891221,
-      "grad_norm": 0.004027934279292822,
-      "learning_rate": 0.00019999675683166537,
-      "loss": 46.0,
-      "step": 33540
-    },
-    {
-      "epoch": 2.5644436798746106,
-      "grad_norm": 0.007147032767534256,
-      "learning_rate": 0.00019999675663821476,
-      "loss": 46.0,
-      "step": 33541
-    },
-    {
-      "epoch": 2.5645201368580004,
-      "grad_norm": 0.0016328798374161124,
-      "learning_rate": 0.00019999675644475842,
-      "loss": 46.0,
-      "step": 33542
-    },
-    {
-      "epoch": 2.56459659384139,
-      "grad_norm": 0.001315317116677761,
-      "learning_rate": 0.0001999967562512963,
-      "loss": 46.0,
-      "step": 33543
-    },
-    {
-      "epoch": 2.56467305082478,
-      "grad_norm": 0.004822699353098869,
-      "learning_rate": 0.00019999675605782838,
-      "loss": 46.0,
-      "step": 33544
-    },
-    {
-      "epoch": 2.5647495078081697,
-      "grad_norm": 0.0032904769759625196,
-      "learning_rate": 0.0001999967558643547,
-      "loss": 46.0,
-      "step": 33545
-    },
-    {
-      "epoch": 2.564825964791559,
-      "grad_norm": 0.0007563504041172564,
-      "learning_rate": 0.00019999675567087527,
-      "loss": 46.0,
-      "step": 33546
-    },
-    {
-      "epoch": 2.5649024217749488,
-      "grad_norm": 0.0038826102390885353,
-      "learning_rate": 0.00019999675547739004,
-      "loss": 46.0,
-      "step": 33547
-    },
-    {
-      "epoch": 2.5649788787583385,
-      "grad_norm": 0.0017097495729103684,
-      "learning_rate": 0.0001999967552838991,
-      "loss": 46.0,
-      "step": 33548
-    },
-    {
-      "epoch": 2.5650553357417283,
-      "grad_norm": 0.0016232189955189824,
-      "learning_rate": 0.00019999675509040231,
-      "loss": 46.0,
-      "step": 33549
-    },
-    {
-      "epoch": 2.565131792725118,
-      "grad_norm": 0.0006448444910347462,
-      "learning_rate": 0.00019999675489689982,
-      "loss": 46.0,
-      "step": 33550
-    },
-    {
-      "epoch": 2.565208249708508,
-      "grad_norm": 0.001872544176876545,
-      "learning_rate": 0.00019999675470339155,
-      "loss": 46.0,
-      "step": 33551
-    },
-    {
-      "epoch": 2.5652847066918976,
-      "grad_norm": 0.0015749622834846377,
-      "learning_rate": 0.00019999675450987748,
-      "loss": 46.0,
-      "step": 33552
-    },
-    {
-      "epoch": 2.5653611636752873,
-      "grad_norm": 0.0016301916912198067,
-      "learning_rate": 0.00019999675431635767,
-      "loss": 46.0,
-      "step": 33553
-    },
-    {
-      "epoch": 2.5654376206586766,
-      "grad_norm": 0.003610240062698722,
-      "learning_rate": 0.00019999675412283208,
-      "loss": 46.0,
-      "step": 33554
-    },
-    {
-      "epoch": 2.5655140776420664,
-      "grad_norm": 0.000785507436376065,
-      "learning_rate": 0.00019999675392930072,
-      "loss": 46.0,
-      "step": 33555
-    },
-    {
-      "epoch": 2.565590534625456,
-      "grad_norm": 0.0005011885077692568,
-      "learning_rate": 0.00019999675373576356,
-      "loss": 46.0,
-      "step": 33556
-    },
-    {
-      "epoch": 2.565666991608846,
-      "grad_norm": 0.0007357841241173446,
-      "learning_rate": 0.00019999675354222067,
-      "loss": 46.0,
-      "step": 33557
-    },
-    {
-      "epoch": 2.5657434485922357,
-      "grad_norm": 0.0011004769476130605,
-      "learning_rate": 0.000199996753348672,
-      "loss": 46.0,
-      "step": 33558
-    },
-    {
-      "epoch": 2.5658199055756254,
-      "grad_norm": 0.0021856778766959906,
-      "learning_rate": 0.00019999675315511756,
-      "loss": 46.0,
-      "step": 33559
-    },
-    {
-      "epoch": 2.565896362559015,
-      "grad_norm": 0.0007466661627404392,
-      "learning_rate": 0.00019999675296155736,
-      "loss": 46.0,
-      "step": 33560
-    },
-    {
-      "epoch": 2.565972819542405,
-      "grad_norm": 0.0012138142483308911,
-      "learning_rate": 0.0001999967527679914,
-      "loss": 46.0,
-      "step": 33561
-    },
-    {
-      "epoch": 2.5660492765257947,
-      "grad_norm": 0.0024261625949293375,
-      "learning_rate": 0.00019999675257441967,
-      "loss": 46.0,
-      "step": 33562
-    },
-    {
-      "epoch": 2.5661257335091845,
-      "grad_norm": 0.003449970157817006,
-      "learning_rate": 0.00019999675238084212,
-      "loss": 46.0,
-      "step": 33563
-    },
-    {
-      "epoch": 2.5662021904925743,
-      "grad_norm": 0.005892294924706221,
-      "learning_rate": 0.00019999675218725885,
-      "loss": 46.0,
-      "step": 33564
-    },
-    {
-      "epoch": 2.566278647475964,
-      "grad_norm": 0.004068222362548113,
-      "learning_rate": 0.00019999675199366978,
-      "loss": 46.0,
-      "step": 33565
-    },
-    {
-      "epoch": 2.5663551044593538,
-      "grad_norm": 0.003267345717176795,
-      "learning_rate": 0.00019999675180007497,
-      "loss": 46.0,
-      "step": 33566
-    },
-    {
-      "epoch": 2.5664315614427435,
-      "grad_norm": 0.0005713776336051524,
-      "learning_rate": 0.00019999675160647438,
-      "loss": 46.0,
-      "step": 33567
-    },
-    {
-      "epoch": 2.566508018426133,
-      "grad_norm": 0.000813572492916137,
-      "learning_rate": 0.000199996751412868,
-      "loss": 46.0,
-      "step": 33568
-    },
-    {
-      "epoch": 2.5665844754095226,
-      "grad_norm": 0.0006860420107841492,
-      "learning_rate": 0.00019999675121925588,
-      "loss": 46.0,
-      "step": 33569
-    },
-    {
-      "epoch": 2.5666609323929124,
-      "grad_norm": 0.0026067919097840786,
-      "learning_rate": 0.00019999675102563797,
-      "loss": 46.0,
-      "step": 33570
-    },
-    {
-      "epoch": 2.566737389376302,
-      "grad_norm": 0.0018609222024679184,
-      "learning_rate": 0.00019999675083201432,
-      "loss": 46.0,
-      "step": 33571
-    },
-    {
-      "epoch": 2.566813846359692,
-      "grad_norm": 0.0013453407445922494,
-      "learning_rate": 0.00019999675063838487,
-      "loss": 46.0,
-      "step": 33572
-    },
-    {
-      "epoch": 2.5668903033430817,
-      "grad_norm": 0.0034865285269916058,
-      "learning_rate": 0.00019999675044474966,
-      "loss": 46.0,
-      "step": 33573
-    },
-    {
-      "epoch": 2.5669667603264714,
-      "grad_norm": 0.0015838018152862787,
-      "learning_rate": 0.0001999967502511087,
-      "loss": 46.0,
-      "step": 33574
-    },
-    {
-      "epoch": 2.567043217309861,
-      "grad_norm": 0.000810201745480299,
-      "learning_rate": 0.00019999675005746194,
-      "loss": 46.0,
-      "step": 33575
-    },
-    {
-      "epoch": 2.5671196742932505,
-      "grad_norm": 0.0010030876146629453,
-      "learning_rate": 0.00019999674986380942,
-      "loss": 46.0,
-      "step": 33576
-    },
-    {
-      "epoch": 2.5671961312766403,
-      "grad_norm": 0.001643717521801591,
-      "learning_rate": 0.00019999674967015115,
-      "loss": 46.0,
-      "step": 33577
-    },
-    {
-      "epoch": 2.56727258826003,
-      "grad_norm": 0.0008654296980239451,
-      "learning_rate": 0.00019999674947648708,
-      "loss": 46.0,
-      "step": 33578
-    },
-    {
-      "epoch": 2.56734904524342,
-      "grad_norm": 0.005815157666802406,
-      "learning_rate": 0.00019999674928281727,
-      "loss": 46.0,
-      "step": 33579
-    },
-    {
-      "epoch": 2.5674255022268095,
-      "grad_norm": 0.0018154950812458992,
-      "learning_rate": 0.00019999674908914168,
-      "loss": 46.0,
-      "step": 33580
-    },
-    {
-      "epoch": 2.5675019592101993,
-      "grad_norm": 0.0009040693985298276,
-      "learning_rate": 0.0001999967488954603,
-      "loss": 46.0,
-      "step": 33581
-    },
-    {
-      "epoch": 2.567578416193589,
-      "grad_norm": 0.0006755805225111544,
-      "learning_rate": 0.0001999967487017732,
-      "loss": 46.0,
-      "step": 33582
-    },
-    {
-      "epoch": 2.567654873176979,
-      "grad_norm": 0.0009333838243037462,
-      "learning_rate": 0.00019999674850808025,
-      "loss": 46.0,
-      "step": 33583
-    },
-    {
-      "epoch": 2.5677313301603686,
-      "grad_norm": 0.003079622983932495,
-      "learning_rate": 0.0001999967483143816,
-      "loss": 46.0,
-      "step": 33584
-    },
-    {
-      "epoch": 2.5678077871437583,
-      "grad_norm": 0.00397895323112607,
-      "learning_rate": 0.00019999674812067717,
-      "loss": 46.0,
-      "step": 33585
-    },
-    {
-      "epoch": 2.567884244127148,
-      "grad_norm": 0.008154308423399925,
-      "learning_rate": 0.00019999674792696697,
-      "loss": 46.0,
-      "step": 33586
-    },
-    {
-      "epoch": 2.567960701110538,
-      "grad_norm": 0.006857165135443211,
-      "learning_rate": 0.000199996747733251,
-      "loss": 46.0,
-      "step": 33587
-    },
-    {
-      "epoch": 2.5680371580939276,
-      "grad_norm": 0.00257407920435071,
-      "learning_rate": 0.00019999674753952925,
-      "loss": 46.0,
-      "step": 33588
-    },
-    {
-      "epoch": 2.5681136150773174,
-      "grad_norm": 0.0014389710268005729,
-      "learning_rate": 0.00019999674734580173,
-      "loss": 46.0,
-      "step": 33589
-    },
-    {
-      "epoch": 2.5681900720607067,
-      "grad_norm": 0.0016827452927827835,
-      "learning_rate": 0.00019999674715206843,
-      "loss": 46.0,
-      "step": 33590
-    },
-    {
-      "epoch": 2.5682665290440965,
-      "grad_norm": 0.0016161476960405707,
-      "learning_rate": 0.00019999674695832936,
-      "loss": 46.0,
-      "step": 33591
-    },
-    {
-      "epoch": 2.5683429860274862,
-      "grad_norm": 0.003877685870975256,
-      "learning_rate": 0.00019999674676458455,
-      "loss": 46.0,
-      "step": 33592
-    },
-    {
-      "epoch": 2.568419443010876,
-      "grad_norm": 0.0011132400250062346,
-      "learning_rate": 0.00019999674657083396,
-      "loss": 46.0,
-      "step": 33593
-    },
-    {
-      "epoch": 2.5684958999942658,
-      "grad_norm": 0.0015398639952763915,
-      "learning_rate": 0.0001999967463770776,
-      "loss": 46.0,
-      "step": 33594
-    },
-    {
-      "epoch": 2.5685723569776555,
-      "grad_norm": 0.0035009966231882572,
-      "learning_rate": 0.00019999674618331547,
-      "loss": 46.0,
-      "step": 33595
-    },
-    {
-      "epoch": 2.5686488139610453,
-      "grad_norm": 0.0007522220839746296,
-      "learning_rate": 0.00019999674598954756,
-      "loss": 46.0,
-      "step": 33596
-    },
-    {
-      "epoch": 2.568725270944435,
-      "grad_norm": 0.0007269795169122517,
-      "learning_rate": 0.00019999674579577388,
-      "loss": 46.0,
-      "step": 33597
-    },
-    {
-      "epoch": 2.5688017279278244,
-      "grad_norm": 0.0006011362420395017,
-      "learning_rate": 0.00019999674560199445,
-      "loss": 46.0,
-      "step": 33598
-    },
-    {
-      "epoch": 2.568878184911214,
-      "grad_norm": 0.0009947501821443439,
-      "learning_rate": 0.00019999674540820922,
-      "loss": 46.0,
-      "step": 33599
-    },
-    {
-      "epoch": 2.568954641894604,
-      "grad_norm": 0.0006983648636378348,
-      "learning_rate": 0.00019999674521441828,
-      "loss": 46.0,
-      "step": 33600
-    },
-    {
-      "epoch": 2.5690310988779936,
-      "grad_norm": 0.0011247338261455297,
-      "learning_rate": 0.0001999967450206215,
-      "loss": 46.0,
-      "step": 33601
-    },
-    {
-      "epoch": 2.5691075558613834,
-      "grad_norm": 0.0011075049405917525,
-      "learning_rate": 0.000199996744826819,
-      "loss": 46.0,
-      "step": 33602
-    },
-    {
-      "epoch": 2.569184012844773,
-      "grad_norm": 0.0007380927563644946,
-      "learning_rate": 0.0001999967446330107,
-      "loss": 46.0,
-      "step": 33603
-    },
-    {
-      "epoch": 2.569260469828163,
-      "grad_norm": 0.0014994388911873102,
-      "learning_rate": 0.00019999674443919668,
-      "loss": 46.0,
-      "step": 33604
-    },
-    {
-      "epoch": 2.5693369268115527,
-      "grad_norm": 0.0019309587078168988,
-      "learning_rate": 0.00019999674424537683,
-      "loss": 46.0,
-      "step": 33605
-    },
-    {
-      "epoch": 2.5694133837949424,
-      "grad_norm": 0.00415783142670989,
-      "learning_rate": 0.00019999674405155125,
-      "loss": 46.0,
-      "step": 33606
-    },
-    {
-      "epoch": 2.569489840778332,
-      "grad_norm": 0.0017408544663339853,
-      "learning_rate": 0.0001999967438577199,
-      "loss": 46.0,
-      "step": 33607
-    },
-    {
-      "epoch": 2.569566297761722,
-      "grad_norm": 0.0021758307702839375,
-      "learning_rate": 0.00019999674366388275,
-      "loss": 46.0,
-      "step": 33608
-    },
-    {
-      "epoch": 2.5696427547451117,
-      "grad_norm": 0.0010166955180466175,
-      "learning_rate": 0.00019999674347003982,
-      "loss": 46.0,
-      "step": 33609
-    },
-    {
-      "epoch": 2.5697192117285015,
-      "grad_norm": 0.0029189339838922024,
-      "learning_rate": 0.00019999674327619117,
-      "loss": 46.0,
-      "step": 33610
-    },
-    {
-      "epoch": 2.5697956687118912,
-      "grad_norm": 0.0016484318766742945,
-      "learning_rate": 0.00019999674308233674,
-      "loss": 46.0,
-      "step": 33611
-    },
-    {
-      "epoch": 2.5698721256952806,
-      "grad_norm": 0.003079600166529417,
-      "learning_rate": 0.0001999967428884765,
-      "loss": 46.0,
-      "step": 33612
-    },
-    {
-      "epoch": 2.5699485826786703,
-      "grad_norm": 0.003807327477261424,
-      "learning_rate": 0.00019999674269461054,
-      "loss": 46.0,
-      "step": 33613
-    },
-    {
-      "epoch": 2.57002503966206,
-      "grad_norm": 0.001914694090373814,
-      "learning_rate": 0.0001999967425007388,
-      "loss": 46.0,
-      "step": 33614
-    },
-    {
-      "epoch": 2.57010149664545,
-      "grad_norm": 0.0032240008004009724,
-      "learning_rate": 0.00019999674230686127,
-      "loss": 46.0,
-      "step": 33615
-    },
-    {
-      "epoch": 2.5701779536288396,
-      "grad_norm": 0.0016452540876343846,
-      "learning_rate": 0.00019999674211297798,
-      "loss": 46.0,
-      "step": 33616
-    },
-    {
-      "epoch": 2.5702544106122294,
-      "grad_norm": 0.0035483958199620247,
-      "learning_rate": 0.00019999674191908894,
-      "loss": 46.0,
-      "step": 33617
-    },
-    {
-      "epoch": 2.570330867595619,
-      "grad_norm": 0.0014126044698059559,
-      "learning_rate": 0.00019999674172519412,
-      "loss": 46.0,
-      "step": 33618
-    },
-    {
-      "epoch": 2.570407324579009,
-      "grad_norm": 0.0008452071342617273,
-      "learning_rate": 0.0001999967415312935,
-      "loss": 46.0,
-      "step": 33619
-    },
-    {
-      "epoch": 2.570483781562398,
-      "grad_norm": 0.0029648886993527412,
-      "learning_rate": 0.00019999674133738715,
-      "loss": 46.0,
-      "step": 33620
-    },
-    {
-      "epoch": 2.570560238545788,
-      "grad_norm": 0.0016587642021477222,
-      "learning_rate": 0.00019999674114347502,
-      "loss": 46.0,
-      "step": 33621
-    },
-    {
-      "epoch": 2.5706366955291777,
-      "grad_norm": 0.000729656545445323,
-      "learning_rate": 0.0001999967409495571,
-      "loss": 46.0,
-      "step": 33622
-    },
-    {
-      "epoch": 2.5707131525125675,
-      "grad_norm": 0.001706088543869555,
-      "learning_rate": 0.00019999674075563343,
-      "loss": 46.0,
-      "step": 33623
-    },
-    {
-      "epoch": 2.5707896094959573,
-      "grad_norm": 0.0004884802037850022,
-      "learning_rate": 0.000199996740561704,
-      "loss": 46.0,
-      "step": 33624
-    },
-    {
-      "epoch": 2.570866066479347,
-      "grad_norm": 0.002172946697100997,
-      "learning_rate": 0.0001999967403677688,
-      "loss": 46.0,
-      "step": 33625
-    },
-    {
-      "epoch": 2.5709425234627368,
-      "grad_norm": 0.0008171516237780452,
-      "learning_rate": 0.0001999967401738278,
-      "loss": 46.0,
-      "step": 33626
-    },
-    {
-      "epoch": 2.5710189804461265,
-      "grad_norm": 0.006667159032076597,
-      "learning_rate": 0.00019999673997988105,
-      "loss": 46.0,
-      "step": 33627
-    },
-    {
-      "epoch": 2.5710954374295163,
-      "grad_norm": 0.006327591836452484,
-      "learning_rate": 0.00019999673978592853,
-      "loss": 46.0,
-      "step": 33628
-    },
-    {
-      "epoch": 2.571171894412906,
-      "grad_norm": 0.0012598424218595028,
-      "learning_rate": 0.00019999673959197027,
-      "loss": 46.0,
-      "step": 33629
-    },
-    {
-      "epoch": 2.571248351396296,
-      "grad_norm": 0.001248611486516893,
-      "learning_rate": 0.0001999967393980062,
-      "loss": 46.0,
-      "step": 33630
-    },
-    {
-      "epoch": 2.5713248083796856,
-      "grad_norm": 0.002740581287071109,
-      "learning_rate": 0.0001999967392040364,
-      "loss": 46.0,
-      "step": 33631
-    },
-    {
-      "epoch": 2.5714012653630753,
-      "grad_norm": 0.00342528335750103,
-      "learning_rate": 0.00019999673901006078,
-      "loss": 46.0,
-      "step": 33632
-    },
-    {
-      "epoch": 2.5714777223464647,
-      "grad_norm": 0.0024696860928088427,
-      "learning_rate": 0.00019999673881607942,
-      "loss": 46.0,
-      "step": 33633
-    },
-    {
-      "epoch": 2.5715541793298544,
-      "grad_norm": 0.0038649081252515316,
-      "learning_rate": 0.00019999673862209228,
-      "loss": 46.0,
-      "step": 33634
-    },
-    {
-      "epoch": 2.571630636313244,
-      "grad_norm": 0.0012987615773454309,
-      "learning_rate": 0.0001999967384280994,
-      "loss": 46.0,
-      "step": 33635
-    },
-    {
-      "epoch": 2.571707093296634,
-      "grad_norm": 0.003049397375434637,
-      "learning_rate": 0.00019999673823410072,
-      "loss": 46.0,
-      "step": 33636
-    },
-    {
-      "epoch": 2.5717835502800237,
-      "grad_norm": 0.0021371340844780207,
-      "learning_rate": 0.00019999673804009627,
-      "loss": 46.0,
-      "step": 33637
-    },
-    {
-      "epoch": 2.5718600072634135,
-      "grad_norm": 0.0009890544461086392,
-      "learning_rate": 0.00019999673784608607,
-      "loss": 46.0,
-      "step": 33638
-    },
-    {
-      "epoch": 2.5719364642468032,
-      "grad_norm": 0.0012890303041785955,
-      "learning_rate": 0.00019999673765207007,
-      "loss": 46.0,
-      "step": 33639
-    },
-    {
-      "epoch": 2.572012921230193,
-      "grad_norm": 0.0018265777034685016,
-      "learning_rate": 0.00019999673745804835,
-      "loss": 46.0,
-      "step": 33640
-    },
-    {
-      "epoch": 2.5720893782135827,
-      "grad_norm": 0.0012183840153738856,
-      "learning_rate": 0.00019999673726402083,
-      "loss": 46.0,
-      "step": 33641
-    },
-    {
-      "epoch": 2.572165835196972,
-      "grad_norm": 0.0014565408928319812,
-      "learning_rate": 0.0001999967370699875,
-      "loss": 46.0,
-      "step": 33642
-    },
-    {
-      "epoch": 2.572242292180362,
-      "grad_norm": 0.0016076304018497467,
-      "learning_rate": 0.00019999673687594847,
-      "loss": 46.0,
-      "step": 33643
-    },
-    {
-      "epoch": 2.5723187491637516,
-      "grad_norm": 0.003128182841464877,
-      "learning_rate": 0.00019999673668190363,
-      "loss": 46.0,
-      "step": 33644
-    },
-    {
-      "epoch": 2.5723952061471413,
-      "grad_norm": 0.0024909370113164186,
-      "learning_rate": 0.00019999673648785304,
-      "loss": 46.0,
-      "step": 33645
-    },
-    {
-      "epoch": 2.572471663130531,
-      "grad_norm": 0.003303632140159607,
-      "learning_rate": 0.00019999673629379669,
-      "loss": 46.0,
-      "step": 33646
-    },
-    {
-      "epoch": 2.572548120113921,
-      "grad_norm": 0.0010045719100162387,
-      "learning_rate": 0.00019999673609973458,
-      "loss": 46.0,
-      "step": 33647
-    },
-    {
-      "epoch": 2.5726245770973106,
-      "grad_norm": 0.0004578182997647673,
-      "learning_rate": 0.00019999673590566665,
-      "loss": 46.0,
-      "step": 33648
-    },
-    {
-      "epoch": 2.5727010340807004,
-      "grad_norm": 0.0010105711407959461,
-      "learning_rate": 0.000199996735711593,
-      "loss": 46.0,
-      "step": 33649
-    },
-    {
-      "epoch": 2.57277749106409,
-      "grad_norm": 0.0014336196472868323,
-      "learning_rate": 0.00019999673551751354,
-      "loss": 46.0,
-      "step": 33650
-    },
-    {
-      "epoch": 2.57285394804748,
-      "grad_norm": 0.003676900640130043,
-      "learning_rate": 0.00019999673532342834,
-      "loss": 46.0,
-      "step": 33651
-    },
-    {
-      "epoch": 2.5729304050308697,
-      "grad_norm": 0.0015876519028097391,
-      "learning_rate": 0.00019999673512933734,
-      "loss": 46.0,
-      "step": 33652
-    },
-    {
-      "epoch": 2.5730068620142594,
-      "grad_norm": 0.0017812944715842605,
-      "learning_rate": 0.0001999967349352406,
-      "loss": 46.0,
-      "step": 33653
-    },
-    {
-      "epoch": 2.573083318997649,
-      "grad_norm": 0.0015645832754671574,
-      "learning_rate": 0.0001999967347411381,
-      "loss": 46.0,
-      "step": 33654
-    },
-    {
-      "epoch": 2.5731597759810385,
-      "grad_norm": 0.0014201278099790215,
-      "learning_rate": 0.0001999967345470298,
-      "loss": 46.0,
-      "step": 33655
-    },
-    {
-      "epoch": 2.5732362329644283,
-      "grad_norm": 0.003005894599482417,
-      "learning_rate": 0.00019999673435291574,
-      "loss": 46.0,
-      "step": 33656
-    },
-    {
-      "epoch": 2.573312689947818,
-      "grad_norm": 0.0018718059873208404,
-      "learning_rate": 0.0001999967341587959,
-      "loss": 46.0,
-      "step": 33657
-    },
-    {
-      "epoch": 2.573389146931208,
-      "grad_norm": 0.001077447785064578,
-      "learning_rate": 0.00019999673396467032,
-      "loss": 46.0,
-      "step": 33658
-    },
-    {
-      "epoch": 2.5734656039145976,
-      "grad_norm": 0.0024721743538975716,
-      "learning_rate": 0.00019999673377053893,
-      "loss": 46.0,
-      "step": 33659
-    },
-    {
-      "epoch": 2.5735420608979873,
-      "grad_norm": 0.0018371124751865864,
-      "learning_rate": 0.00019999673357640183,
-      "loss": 46.0,
-      "step": 33660
-    },
-    {
-      "epoch": 2.573618517881377,
-      "grad_norm": 0.0035809038672596216,
-      "learning_rate": 0.0001999967333822589,
-      "loss": 46.0,
-      "step": 33661
-    },
-    {
-      "epoch": 2.573694974864767,
-      "grad_norm": 0.0020826654508709908,
-      "learning_rate": 0.00019999673318811024,
-      "loss": 46.0,
-      "step": 33662
-    },
-    {
-      "epoch": 2.573771431848156,
-      "grad_norm": 0.0010911946883425117,
-      "learning_rate": 0.0001999967329939558,
-      "loss": 46.0,
-      "step": 33663
-    },
-    {
-      "epoch": 2.573847888831546,
-      "grad_norm": 0.0034325115848332644,
-      "learning_rate": 0.0001999967327997956,
-      "loss": 46.0,
-      "step": 33664
-    },
-    {
-      "epoch": 2.5739243458149357,
-      "grad_norm": 0.0009272743482142687,
-      "learning_rate": 0.00019999673260562962,
-      "loss": 46.0,
-      "step": 33665
-    },
-    {
-      "epoch": 2.5740008027983254,
-      "grad_norm": 0.002528467681258917,
-      "learning_rate": 0.00019999673241145787,
-      "loss": 46.0,
-      "step": 33666
-    },
-    {
-      "epoch": 2.574077259781715,
-      "grad_norm": 0.003457816783338785,
-      "learning_rate": 0.00019999673221728035,
-      "loss": 46.0,
-      "step": 33667
-    },
-    {
-      "epoch": 2.574153716765105,
-      "grad_norm": 0.0015440174611285329,
-      "learning_rate": 0.00019999673202309706,
-      "loss": 46.0,
-      "step": 33668
-    },
-    {
-      "epoch": 2.5742301737484947,
-      "grad_norm": 0.0019255020888522267,
-      "learning_rate": 0.000199996731828908,
-      "loss": 46.0,
-      "step": 33669
-    },
-    {
-      "epoch": 2.5743066307318845,
-      "grad_norm": 0.002855937462300062,
-      "learning_rate": 0.00019999673163471316,
-      "loss": 46.0,
-      "step": 33670
-    },
-    {
-      "epoch": 2.5743830877152742,
-      "grad_norm": 0.0024788465816527605,
-      "learning_rate": 0.0001999967314405126,
-      "loss": 46.0,
-      "step": 33671
-    },
-    {
-      "epoch": 2.574459544698664,
-      "grad_norm": 0.0024461187422275543,
-      "learning_rate": 0.0001999967312463062,
-      "loss": 46.0,
-      "step": 33672
-    },
-    {
-      "epoch": 2.5745360016820538,
-      "grad_norm": 0.0008722626371309161,
-      "learning_rate": 0.00019999673105209408,
-      "loss": 46.0,
-      "step": 33673
-    },
-    {
-      "epoch": 2.5746124586654435,
-      "grad_norm": 0.001374467508867383,
-      "learning_rate": 0.00019999673085787615,
-      "loss": 46.0,
-      "step": 33674
-    },
-    {
-      "epoch": 2.5746889156488333,
-      "grad_norm": 0.0013584342086687684,
-      "learning_rate": 0.0001999967306636525,
-      "loss": 46.0,
-      "step": 33675
-    },
-    {
-      "epoch": 2.574765372632223,
-      "grad_norm": 0.0022337569389492273,
-      "learning_rate": 0.00019999673046942307,
-      "loss": 46.0,
-      "step": 33676
-    },
-    {
-      "epoch": 2.5748418296156124,
-      "grad_norm": 0.0015917246928438544,
-      "learning_rate": 0.00019999673027518784,
-      "loss": 46.0,
-      "step": 33677
-    },
-    {
-      "epoch": 2.574918286599002,
-      "grad_norm": 0.004520599730312824,
-      "learning_rate": 0.00019999673008094687,
-      "loss": 46.0,
-      "step": 33678
-    },
-    {
-      "epoch": 2.574994743582392,
-      "grad_norm": 0.003556224051862955,
-      "learning_rate": 0.0001999967298867001,
-      "loss": 46.0,
-      "step": 33679
-    },
-    {
-      "epoch": 2.5750712005657816,
-      "grad_norm": 0.002192916814237833,
-      "learning_rate": 0.0001999967296924476,
-      "loss": 46.0,
-      "step": 33680
-    },
-    {
-      "epoch": 2.5751476575491714,
-      "grad_norm": 0.0011125840246677399,
-      "learning_rate": 0.00019999672949818931,
-      "loss": 46.0,
-      "step": 33681
-    },
-    {
-      "epoch": 2.575224114532561,
-      "grad_norm": 0.0029193125665187836,
-      "learning_rate": 0.00019999672930392522,
-      "loss": 46.0,
-      "step": 33682
-    },
-    {
-      "epoch": 2.575300571515951,
-      "grad_norm": 0.0005085760494694114,
-      "learning_rate": 0.0001999967291096554,
-      "loss": 46.0,
-      "step": 33683
-    },
-    {
-      "epoch": 2.5753770284993407,
-      "grad_norm": 0.0013059613993391395,
-      "learning_rate": 0.00019999672891537982,
-      "loss": 46.0,
-      "step": 33684
-    },
-    {
-      "epoch": 2.57545348548273,
-      "grad_norm": 0.0014554500812664628,
-      "learning_rate": 0.00019999672872109844,
-      "loss": 46.0,
-      "step": 33685
-    },
-    {
-      "epoch": 2.5755299424661198,
-      "grad_norm": 0.0008898632368072867,
-      "learning_rate": 0.0001999967285268113,
-      "loss": 46.0,
-      "step": 33686
-    },
-    {
-      "epoch": 2.5756063994495095,
-      "grad_norm": 0.0008985575404949486,
-      "learning_rate": 0.0001999967283325184,
-      "loss": 46.0,
-      "step": 33687
-    },
-    {
-      "epoch": 2.5756828564328993,
-      "grad_norm": 0.0007783126784488559,
-      "learning_rate": 0.00019999672813821975,
-      "loss": 46.0,
-      "step": 33688
-    },
-    {
-      "epoch": 2.575759313416289,
-      "grad_norm": 0.00416223518550396,
-      "learning_rate": 0.0001999967279439153,
-      "loss": 46.0,
-      "step": 33689
-    },
-    {
-      "epoch": 2.575835770399679,
-      "grad_norm": 0.00104646070394665,
-      "learning_rate": 0.0001999967277496051,
-      "loss": 46.0,
-      "step": 33690
-    },
-    {
-      "epoch": 2.5759122273830686,
-      "grad_norm": 0.0002899757237173617,
-      "learning_rate": 0.0001999967275552891,
-      "loss": 46.0,
-      "step": 33691
-    },
-    {
-      "epoch": 2.5759886843664583,
-      "grad_norm": 0.0018683980451896787,
-      "learning_rate": 0.00019999672736096736,
-      "loss": 46.0,
-      "step": 33692
-    },
-    {
-      "epoch": 2.576065141349848,
-      "grad_norm": 0.003739645704627037,
-      "learning_rate": 0.0001999967271666398,
-      "loss": 46.0,
-      "step": 33693
-    },
-    {
-      "epoch": 2.576141598333238,
-      "grad_norm": 0.0026869659777730703,
-      "learning_rate": 0.00019999672697230655,
-      "loss": 46.0,
-      "step": 33694
-    },
-    {
-      "epoch": 2.5762180553166276,
-      "grad_norm": 0.0011040762765333056,
-      "learning_rate": 0.00019999672677796748,
-      "loss": 46.0,
-      "step": 33695
-    },
-    {
-      "epoch": 2.5762945123000174,
-      "grad_norm": 0.002665249863639474,
-      "learning_rate": 0.00019999672658362267,
-      "loss": 46.0,
-      "step": 33696
-    },
-    {
-      "epoch": 2.576370969283407,
-      "grad_norm": 0.0019124011741951108,
-      "learning_rate": 0.00019999672638927206,
-      "loss": 46.0,
-      "step": 33697
-    },
-    {
-      "epoch": 2.576447426266797,
-      "grad_norm": 0.00514562800526619,
-      "learning_rate": 0.0001999967261949157,
-      "loss": 46.0,
-      "step": 33698
-    },
-    {
-      "epoch": 2.576523883250186,
-      "grad_norm": 0.0018402902642264962,
-      "learning_rate": 0.00019999672600055357,
-      "loss": 46.0,
-      "step": 33699
-    },
-    {
-      "epoch": 2.576600340233576,
-      "grad_norm": 0.0008133978117257357,
-      "learning_rate": 0.00019999672580618566,
-      "loss": 46.0,
-      "step": 33700
-    },
-    {
-      "epoch": 2.5766767972169657,
-      "grad_norm": 0.0008695360156707466,
-      "learning_rate": 0.00019999672561181198,
-      "loss": 46.0,
-      "step": 33701
-    },
-    {
-      "epoch": 2.5767532542003555,
-      "grad_norm": 0.0003591017739381641,
-      "learning_rate": 0.00019999672541743256,
-      "loss": 46.0,
-      "step": 33702
-    },
-    {
-      "epoch": 2.5768297111837453,
-      "grad_norm": 0.0021886201575398445,
-      "learning_rate": 0.00019999672522304733,
-      "loss": 46.0,
-      "step": 33703
-    },
-    {
-      "epoch": 2.576906168167135,
-      "grad_norm": 0.0007825404172763228,
-      "learning_rate": 0.00019999672502865633,
-      "loss": 46.0,
-      "step": 33704
-    },
-    {
-      "epoch": 2.576982625150525,
-      "grad_norm": 0.0009387555182911456,
-      "learning_rate": 0.0001999967248342596,
-      "loss": 46.0,
-      "step": 33705
-    },
-    {
-      "epoch": 2.5770590821339145,
-      "grad_norm": 0.00335617084056139,
-      "learning_rate": 0.00019999672463985707,
-      "loss": 46.0,
-      "step": 33706
-    },
-    {
-      "epoch": 2.577135539117304,
-      "grad_norm": 0.0018673943122848868,
-      "learning_rate": 0.00019999672444544878,
-      "loss": 46.0,
-      "step": 33707
-    },
-    {
-      "epoch": 2.5772119961006936,
-      "grad_norm": 0.0021284364629536867,
-      "learning_rate": 0.00019999672425103472,
-      "loss": 46.0,
-      "step": 33708
-    },
-    {
-      "epoch": 2.5772884530840834,
-      "grad_norm": 0.0014881722163408995,
-      "learning_rate": 0.0001999967240566149,
-      "loss": 46.0,
-      "step": 33709
-    },
-    {
-      "epoch": 2.577364910067473,
-      "grad_norm": 0.0010011779377236962,
-      "learning_rate": 0.0001999967238621893,
-      "loss": 46.0,
-      "step": 33710
-    },
-    {
-      "epoch": 2.577441367050863,
-      "grad_norm": 0.0006519795279018581,
-      "learning_rate": 0.00019999672366775793,
-      "loss": 46.0,
-      "step": 33711
-    },
-    {
-      "epoch": 2.5775178240342527,
-      "grad_norm": 0.010698611848056316,
-      "learning_rate": 0.0001999967234733208,
-      "loss": 46.0,
-      "step": 33712
-    },
-    {
-      "epoch": 2.5775942810176424,
-      "grad_norm": 0.001863163779489696,
-      "learning_rate": 0.00019999672327887787,
-      "loss": 46.0,
-      "step": 33713
-    },
-    {
-      "epoch": 2.577670738001032,
-      "grad_norm": 0.0022500120103359222,
-      "learning_rate": 0.0001999967230844292,
-      "loss": 46.0,
-      "step": 33714
-    },
-    {
-      "epoch": 2.577747194984422,
-      "grad_norm": 0.0022092899307608604,
-      "learning_rate": 0.00019999672288997477,
-      "loss": 46.0,
-      "step": 33715
-    },
-    {
-      "epoch": 2.5778236519678117,
-      "grad_norm": 0.0008713945280760527,
-      "learning_rate": 0.00019999672269551457,
-      "loss": 46.0,
-      "step": 33716
-    },
-    {
-      "epoch": 2.5779001089512015,
-      "grad_norm": 0.0024856626987457275,
-      "learning_rate": 0.00019999672250104857,
-      "loss": 46.0,
-      "step": 33717
-    },
-    {
-      "epoch": 2.5779765659345912,
-      "grad_norm": 0.002778966911137104,
-      "learning_rate": 0.00019999672230657683,
-      "loss": 46.0,
-      "step": 33718
-    },
-    {
-      "epoch": 2.578053022917981,
-      "grad_norm": 0.0017632093513384461,
-      "learning_rate": 0.0001999967221120993,
-      "loss": 46.0,
-      "step": 33719
-    },
-    {
-      "epoch": 2.5781294799013708,
-      "grad_norm": 0.001748785492964089,
-      "learning_rate": 0.00019999672191761602,
-      "loss": 46.0,
-      "step": 33720
-    },
-    {
-      "epoch": 2.57820593688476,
-      "grad_norm": 0.0016892984276637435,
-      "learning_rate": 0.00019999672172312695,
-      "loss": 46.0,
-      "step": 33721
-    },
-    {
-      "epoch": 2.57828239386815,
-      "grad_norm": 0.0011058708187192678,
-      "learning_rate": 0.00019999672152863212,
-      "loss": 46.0,
-      "step": 33722
-    },
-    {
-      "epoch": 2.5783588508515396,
-      "grad_norm": 0.00383828766644001,
-      "learning_rate": 0.00019999672133413153,
-      "loss": 46.0,
-      "step": 33723
-    },
-    {
-      "epoch": 2.5784353078349294,
-      "grad_norm": 0.0030533461831510067,
-      "learning_rate": 0.00019999672113962515,
-      "loss": 46.0,
-      "step": 33724
-    },
-    {
-      "epoch": 2.578511764818319,
-      "grad_norm": 0.0014342984650284052,
-      "learning_rate": 0.00019999672094511302,
-      "loss": 46.0,
-      "step": 33725
-    },
-    {
-      "epoch": 2.578588221801709,
-      "grad_norm": 0.001511389040388167,
-      "learning_rate": 0.0001999967207505951,
-      "loss": 46.0,
-      "step": 33726
-    },
-    {
-      "epoch": 2.5786646787850986,
-      "grad_norm": 0.001929558697156608,
-      "learning_rate": 0.00019999672055607146,
-      "loss": 46.0,
-      "step": 33727
-    },
-    {
-      "epoch": 2.5787411357684884,
-      "grad_norm": 0.0009858261328190565,
-      "learning_rate": 0.000199996720361542,
-      "loss": 46.0,
-      "step": 33728
-    },
-    {
-      "epoch": 2.5788175927518777,
-      "grad_norm": 0.001606738893315196,
-      "learning_rate": 0.00019999672016700679,
-      "loss": 46.0,
-      "step": 33729
-    },
-    {
-      "epoch": 2.5788940497352675,
-      "grad_norm": 0.0007273387163877487,
-      "learning_rate": 0.00019999671997246582,
-      "loss": 46.0,
-      "step": 33730
-    },
-    {
-      "epoch": 2.5789705067186572,
-      "grad_norm": 0.0014322358183562756,
-      "learning_rate": 0.00019999671977791904,
-      "loss": 46.0,
-      "step": 33731
-    },
-    {
-      "epoch": 2.579046963702047,
-      "grad_norm": 0.0007518419879488647,
-      "learning_rate": 0.00019999671958336653,
-      "loss": 46.0,
-      "step": 33732
-    },
-    {
-      "epoch": 2.5791234206854368,
-      "grad_norm": 0.0021305095870047808,
-      "learning_rate": 0.00019999671938880824,
-      "loss": 46.0,
-      "step": 33733
-    },
-    {
-      "epoch": 2.5791998776688265,
-      "grad_norm": 0.0024050401989370584,
-      "learning_rate": 0.0001999967191942442,
-      "loss": 46.0,
-      "step": 33734
-    },
-    {
-      "epoch": 2.5792763346522163,
-      "grad_norm": 0.0021108253858983517,
-      "learning_rate": 0.00019999671899967436,
-      "loss": 46.0,
-      "step": 33735
-    },
-    {
-      "epoch": 2.579352791635606,
-      "grad_norm": 0.00048105738824233413,
-      "learning_rate": 0.00019999671880509875,
-      "loss": 46.0,
-      "step": 33736
-    },
-    {
-      "epoch": 2.579429248618996,
-      "grad_norm": 0.0007690681959502399,
-      "learning_rate": 0.0001999967186105174,
-      "loss": 46.0,
-      "step": 33737
-    },
-    {
-      "epoch": 2.5795057056023856,
-      "grad_norm": 0.00043410947546362877,
-      "learning_rate": 0.00019999671841593023,
-      "loss": 46.0,
-      "step": 33738
-    },
-    {
-      "epoch": 2.5795821625857753,
-      "grad_norm": 0.0018299379153177142,
-      "learning_rate": 0.00019999671822133733,
-      "loss": 46.0,
-      "step": 33739
-    },
-    {
-      "epoch": 2.579658619569165,
-      "grad_norm": 0.0031106076203286648,
-      "learning_rate": 0.00019999671802673868,
-      "loss": 46.0,
-      "step": 33740
-    },
-    {
-      "epoch": 2.579735076552555,
-      "grad_norm": 0.00333911064080894,
-      "learning_rate": 0.0001999967178321342,
-      "loss": 46.0,
-      "step": 33741
-    },
-    {
-      "epoch": 2.5798115335359446,
-      "grad_norm": 0.0020811359863728285,
-      "learning_rate": 0.000199996717637524,
-      "loss": 46.0,
-      "step": 33742
-    },
-    {
-      "epoch": 2.579887990519334,
-      "grad_norm": 0.0011897641234099865,
-      "learning_rate": 0.00019999671744290803,
-      "loss": 46.0,
-      "step": 33743
-    },
-    {
-      "epoch": 2.5799644475027237,
-      "grad_norm": 0.002339713741093874,
-      "learning_rate": 0.00019999671724828626,
-      "loss": 46.0,
-      "step": 33744
-    },
-    {
-      "epoch": 2.5800409044861135,
-      "grad_norm": 0.00048502953723073006,
-      "learning_rate": 0.00019999671705365875,
-      "loss": 46.0,
-      "step": 33745
-    },
-    {
-      "epoch": 2.580117361469503,
-      "grad_norm": 0.003254833398386836,
-      "learning_rate": 0.00019999671685902546,
-      "loss": 46.0,
-      "step": 33746
-    },
-    {
-      "epoch": 2.580193818452893,
-      "grad_norm": 0.0021795781794935465,
-      "learning_rate": 0.0001999967166643864,
-      "loss": 46.0,
-      "step": 33747
-    },
-    {
-      "epoch": 2.5802702754362827,
-      "grad_norm": 0.004946145229041576,
-      "learning_rate": 0.00019999671646974155,
-      "loss": 46.0,
-      "step": 33748
-    },
-    {
-      "epoch": 2.5803467324196725,
-      "grad_norm": 0.0015211660647764802,
-      "learning_rate": 0.00019999671627509094,
-      "loss": 46.0,
-      "step": 33749
-    },
-    {
-      "epoch": 2.5804231894030623,
-      "grad_norm": 0.0014608881901949644,
-      "learning_rate": 0.0001999967160804346,
-      "loss": 46.0,
-      "step": 33750
-    },
-    {
-      "epoch": 2.5804996463864516,
-      "grad_norm": 0.0010390982497483492,
-      "learning_rate": 0.00019999671588577246,
-      "loss": 46.0,
-      "step": 33751
-    },
-    {
-      "epoch": 2.5805761033698413,
-      "grad_norm": 0.0014662137255072594,
-      "learning_rate": 0.00019999671569110455,
-      "loss": 46.0,
-      "step": 33752
-    },
-    {
-      "epoch": 2.580652560353231,
-      "grad_norm": 0.0027485128957778215,
-      "learning_rate": 0.00019999671549643088,
-      "loss": 46.0,
-      "step": 33753
-    },
-    {
-      "epoch": 2.580729017336621,
-      "grad_norm": 0.002101237652823329,
-      "learning_rate": 0.00019999671530175143,
-      "loss": 46.0,
-      "step": 33754
-    },
-    {
-      "epoch": 2.5808054743200106,
-      "grad_norm": 0.0024111801758408546,
-      "learning_rate": 0.0001999967151070662,
-      "loss": 46.0,
-      "step": 33755
-    },
-    {
-      "epoch": 2.5808819313034004,
-      "grad_norm": 0.0012782351113855839,
-      "learning_rate": 0.00019999671491237523,
-      "loss": 46.0,
-      "step": 33756
-    },
-    {
-      "epoch": 2.58095838828679,
-      "grad_norm": 0.0021875794045627117,
-      "learning_rate": 0.0001999967147176785,
-      "loss": 46.0,
-      "step": 33757
-    },
-    {
-      "epoch": 2.58103484527018,
-      "grad_norm": 0.001406670780852437,
-      "learning_rate": 0.00019999671452297597,
-      "loss": 46.0,
-      "step": 33758
-    },
-    {
-      "epoch": 2.5811113022535697,
-      "grad_norm": 0.0009112459956668317,
-      "learning_rate": 0.00019999671432826768,
-      "loss": 46.0,
-      "step": 33759
-    },
-    {
-      "epoch": 2.5811877592369594,
-      "grad_norm": 0.0012908895732834935,
-      "learning_rate": 0.00019999671413355362,
-      "loss": 46.0,
-      "step": 33760
-    },
-    {
-      "epoch": 2.581264216220349,
-      "grad_norm": 0.0015117382863536477,
-      "learning_rate": 0.00019999671393883378,
-      "loss": 46.0,
-      "step": 33761
-    },
-    {
-      "epoch": 2.581340673203739,
-      "grad_norm": 0.0008893916965462267,
-      "learning_rate": 0.0001999967137441082,
-      "loss": 46.0,
-      "step": 33762
-    },
-    {
-      "epoch": 2.5814171301871287,
-      "grad_norm": 0.0013658194802701473,
-      "learning_rate": 0.0001999967135493768,
-      "loss": 46.0,
-      "step": 33763
-    },
-    {
-      "epoch": 2.581493587170518,
-      "grad_norm": 0.0013998374342918396,
-      "learning_rate": 0.00019999671335463968,
-      "loss": 46.0,
-      "step": 33764
-    },
-    {
-      "epoch": 2.581570044153908,
-      "grad_norm": 0.0026385600212961435,
-      "learning_rate": 0.00019999671315989675,
-      "loss": 46.0,
-      "step": 33765
-    },
-    {
-      "epoch": 2.5816465011372975,
-      "grad_norm": 0.0012058269931003451,
-      "learning_rate": 0.0001999967129651481,
-      "loss": 46.0,
-      "step": 33766
-    },
-    {
-      "epoch": 2.5817229581206873,
-      "grad_norm": 0.000891146541107446,
-      "learning_rate": 0.00019999671277039365,
-      "loss": 46.0,
-      "step": 33767
-    },
-    {
-      "epoch": 2.581799415104077,
-      "grad_norm": 0.002193536376580596,
-      "learning_rate": 0.00019999671257563343,
-      "loss": 46.0,
-      "step": 33768
-    },
-    {
-      "epoch": 2.581875872087467,
-      "grad_norm": 0.0008351202704943717,
-      "learning_rate": 0.00019999671238086746,
-      "loss": 46.0,
-      "step": 33769
-    },
-    {
-      "epoch": 2.5819523290708566,
-      "grad_norm": 0.0015749818412587047,
-      "learning_rate": 0.00019999671218609572,
-      "loss": 46.0,
-      "step": 33770
-    },
-    {
-      "epoch": 2.5820287860542463,
-      "grad_norm": 0.002329234965145588,
-      "learning_rate": 0.00019999671199131817,
-      "loss": 46.0,
-      "step": 33771
-    },
-    {
-      "epoch": 2.582105243037636,
-      "grad_norm": 0.003988371230661869,
-      "learning_rate": 0.0001999967117965349,
-      "loss": 46.0,
-      "step": 33772
-    },
-    {
-      "epoch": 2.5821817000210254,
-      "grad_norm": 0.0020972026977688074,
-      "learning_rate": 0.00019999671160174582,
-      "loss": 46.0,
-      "step": 33773
-    },
-    {
-      "epoch": 2.582258157004415,
-      "grad_norm": 0.0029230595100671053,
-      "learning_rate": 0.00019999671140695098,
-      "loss": 46.0,
-      "step": 33774
-    },
-    {
-      "epoch": 2.582334613987805,
-      "grad_norm": 0.0022066854871809483,
-      "learning_rate": 0.0001999967112121504,
-      "loss": 46.0,
-      "step": 33775
-    },
-    {
-      "epoch": 2.5824110709711947,
-      "grad_norm": 0.0011470660101622343,
-      "learning_rate": 0.00019999671101734404,
-      "loss": 46.0,
-      "step": 33776
-    },
-    {
-      "epoch": 2.5824875279545845,
-      "grad_norm": 0.002847378607839346,
-      "learning_rate": 0.00019999671082253189,
-      "loss": 46.0,
-      "step": 33777
-    },
-    {
-      "epoch": 2.5825639849379742,
-      "grad_norm": 0.00179480726365,
-      "learning_rate": 0.00019999671062771398,
-      "loss": 46.0,
-      "step": 33778
-    },
-    {
-      "epoch": 2.582640441921364,
-      "grad_norm": 0.00517666433006525,
-      "learning_rate": 0.0001999967104328903,
-      "loss": 46.0,
-      "step": 33779
-    },
-    {
-      "epoch": 2.5827168989047538,
-      "grad_norm": 0.0033719746861606836,
-      "learning_rate": 0.00019999671023806086,
-      "loss": 46.0,
-      "step": 33780
-    },
-    {
-      "epoch": 2.5827933558881435,
-      "grad_norm": 0.000728595710825175,
-      "learning_rate": 0.00019999671004322563,
-      "loss": 46.0,
-      "step": 33781
-    },
-    {
-      "epoch": 2.5828698128715333,
-      "grad_norm": 0.0019043530337512493,
-      "learning_rate": 0.00019999670984838466,
-      "loss": 46.0,
-      "step": 33782
-    },
-    {
-      "epoch": 2.582946269854923,
-      "grad_norm": 0.001992369070649147,
-      "learning_rate": 0.00019999670965353792,
-      "loss": 46.0,
-      "step": 33783
-    },
-    {
-      "epoch": 2.583022726838313,
-      "grad_norm": 0.002411763882264495,
-      "learning_rate": 0.00019999670945868538,
-      "loss": 46.0,
-      "step": 33784
-    },
-    {
-      "epoch": 2.5830991838217026,
-      "grad_norm": 0.0005923790158703923,
-      "learning_rate": 0.0001999967092638271,
-      "loss": 46.0,
-      "step": 33785
-    },
-    {
-      "epoch": 2.583175640805092,
-      "grad_norm": 0.0019952126313000917,
-      "learning_rate": 0.00019999670906896305,
-      "loss": 46.0,
-      "step": 33786
-    },
-    {
-      "epoch": 2.5832520977884816,
-      "grad_norm": 0.002692426787689328,
-      "learning_rate": 0.0001999967088740932,
-      "loss": 46.0,
-      "step": 33787
-    },
-    {
-      "epoch": 2.5833285547718714,
-      "grad_norm": 0.0026989937759935856,
-      "learning_rate": 0.0001999967086792176,
-      "loss": 46.0,
-      "step": 33788
-    },
-    {
-      "epoch": 2.583405011755261,
-      "grad_norm": 0.006905751768499613,
-      "learning_rate": 0.00019999670848433622,
-      "loss": 46.0,
-      "step": 33789
-    },
-    {
-      "epoch": 2.583481468738651,
-      "grad_norm": 0.0012205137172713876,
-      "learning_rate": 0.00019999670828944907,
-      "loss": 46.0,
-      "step": 33790
-    },
-    {
-      "epoch": 2.5835579257220407,
-      "grad_norm": 0.0014203093014657497,
-      "learning_rate": 0.00019999670809455616,
-      "loss": 46.0,
-      "step": 33791
-    },
-    {
-      "epoch": 2.5836343827054304,
-      "grad_norm": 0.0013054253067821264,
-      "learning_rate": 0.00019999670789965749,
-      "loss": 46.0,
-      "step": 33792
-    },
-    {
-      "epoch": 2.58371083968882,
-      "grad_norm": 0.0023591769859194756,
-      "learning_rate": 0.00019999670770475304,
-      "loss": 46.0,
-      "step": 33793
-    },
-    {
-      "epoch": 2.5837872966722095,
-      "grad_norm": 0.0023021188098937273,
-      "learning_rate": 0.00019999670750984284,
-      "loss": 46.0,
-      "step": 33794
-    },
-    {
-      "epoch": 2.5838637536555993,
-      "grad_norm": 0.0019056061282753944,
-      "learning_rate": 0.00019999670731492684,
-      "loss": 46.0,
-      "step": 33795
-    },
-    {
-      "epoch": 2.583940210638989,
-      "grad_norm": 0.0023336634039878845,
-      "learning_rate": 0.0001999967071200051,
-      "loss": 46.0,
-      "step": 33796
-    },
-    {
-      "epoch": 2.584016667622379,
-      "grad_norm": 0.0038977316580712795,
-      "learning_rate": 0.00019999670692507756,
-      "loss": 46.0,
-      "step": 33797
-    },
-    {
-      "epoch": 2.5840931246057686,
-      "grad_norm": 0.0013777086278423667,
-      "learning_rate": 0.0001999967067301443,
-      "loss": 46.0,
-      "step": 33798
-    },
-    {
-      "epoch": 2.5841695815891583,
-      "grad_norm": 0.002245998941361904,
-      "learning_rate": 0.0001999967065352052,
-      "loss": 46.0,
-      "step": 33799
-    },
-    {
-      "epoch": 2.584246038572548,
-      "grad_norm": 0.007543109357357025,
-      "learning_rate": 0.0001999967063402604,
-      "loss": 46.0,
-      "step": 33800
-    },
-    {
-      "epoch": 2.584322495555938,
-      "grad_norm": 0.001238432014361024,
-      "learning_rate": 0.0001999967061453098,
-      "loss": 46.0,
-      "step": 33801
-    },
-    {
-      "epoch": 2.5843989525393276,
-      "grad_norm": 0.0006391782080754638,
-      "learning_rate": 0.00019999670595035343,
-      "loss": 46.0,
-      "step": 33802
-    },
-    {
-      "epoch": 2.5844754095227174,
-      "grad_norm": 0.0020712323021143675,
-      "learning_rate": 0.00019999670575539128,
-      "loss": 46.0,
-      "step": 33803
-    },
-    {
-      "epoch": 2.584551866506107,
-      "grad_norm": 0.00156506453640759,
-      "learning_rate": 0.00019999670556042335,
-      "loss": 46.0,
-      "step": 33804
-    },
-    {
-      "epoch": 2.584628323489497,
-      "grad_norm": 0.0009177823667414486,
-      "learning_rate": 0.0001999967053654497,
-      "loss": 46.0,
-      "step": 33805
-    },
-    {
-      "epoch": 2.5847047804728867,
-      "grad_norm": 0.0037315429653972387,
-      "learning_rate": 0.00019999670517047028,
-      "loss": 46.0,
-      "step": 33806
-    },
-    {
-      "epoch": 2.5847812374562764,
-      "grad_norm": 0.0017961516277864575,
-      "learning_rate": 0.00019999670497548503,
-      "loss": 46.0,
-      "step": 33807
-    },
-    {
-      "epoch": 2.5848576944396657,
-      "grad_norm": 0.0024719026405364275,
-      "learning_rate": 0.00019999670478049403,
-      "loss": 46.0,
-      "step": 33808
-    },
-    {
-      "epoch": 2.5849341514230555,
-      "grad_norm": 0.0015128189697861671,
-      "learning_rate": 0.00019999670458549731,
-      "loss": 46.0,
-      "step": 33809
-    },
-    {
-      "epoch": 2.5850106084064453,
-      "grad_norm": 0.001197112025693059,
-      "learning_rate": 0.00019999670439049477,
-      "loss": 46.0,
-      "step": 33810
-    },
-    {
-      "epoch": 2.585087065389835,
-      "grad_norm": 0.001631994266062975,
-      "learning_rate": 0.00019999670419548648,
-      "loss": 46.0,
-      "step": 33811
-    },
-    {
-      "epoch": 2.5851635223732248,
-      "grad_norm": 0.0007233674987219274,
-      "learning_rate": 0.00019999670400047242,
-      "loss": 46.0,
-      "step": 33812
-    },
-    {
-      "epoch": 2.5852399793566145,
-      "grad_norm": 0.000873743963893503,
-      "learning_rate": 0.00019999670380545256,
-      "loss": 46.0,
-      "step": 33813
-    },
-    {
-      "epoch": 2.5853164363400043,
-      "grad_norm": 0.0009753777994774282,
-      "learning_rate": 0.00019999670361042698,
-      "loss": 46.0,
-      "step": 33814
-    },
-    {
-      "epoch": 2.585392893323394,
-      "grad_norm": 0.0008296352461911738,
-      "learning_rate": 0.0001999967034153956,
-      "loss": 46.0,
-      "step": 33815
-    },
-    {
-      "epoch": 2.5854693503067834,
-      "grad_norm": 0.002044152468442917,
-      "learning_rate": 0.00019999670322035846,
-      "loss": 46.0,
-      "step": 33816
-    },
-    {
-      "epoch": 2.585545807290173,
-      "grad_norm": 0.001382470945827663,
-      "learning_rate": 0.00019999670302531556,
-      "loss": 46.0,
-      "step": 33817
-    },
-    {
-      "epoch": 2.585622264273563,
-      "grad_norm": 0.0017183058662340045,
-      "learning_rate": 0.00019999670283026691,
-      "loss": 46.0,
-      "step": 33818
-    },
-    {
-      "epoch": 2.5856987212569527,
-      "grad_norm": 0.0010268158512189984,
-      "learning_rate": 0.00019999670263521244,
-      "loss": 46.0,
-      "step": 33819
-    },
-    {
-      "epoch": 2.5857751782403424,
-      "grad_norm": 0.002678610384464264,
-      "learning_rate": 0.00019999670244015222,
-      "loss": 46.0,
-      "step": 33820
-    },
-    {
-      "epoch": 2.585851635223732,
-      "grad_norm": 0.000850468291901052,
-      "learning_rate": 0.00019999670224508625,
-      "loss": 46.0,
-      "step": 33821
-    },
-    {
-      "epoch": 2.585928092207122,
-      "grad_norm": 0.001257752301171422,
-      "learning_rate": 0.00019999670205001448,
-      "loss": 46.0,
-      "step": 33822
-    },
-    {
-      "epoch": 2.5860045491905117,
-      "grad_norm": 0.0014282895717769861,
-      "learning_rate": 0.00019999670185493696,
-      "loss": 46.0,
-      "step": 33823
-    },
-    {
-      "epoch": 2.5860810061739015,
-      "grad_norm": 0.0011255107820034027,
-      "learning_rate": 0.00019999670165985367,
-      "loss": 46.0,
-      "step": 33824
-    },
-    {
-      "epoch": 2.5861574631572912,
-      "grad_norm": 0.0014628826174885035,
-      "learning_rate": 0.0001999967014647646,
-      "loss": 46.0,
-      "step": 33825
-    },
-    {
-      "epoch": 2.586233920140681,
-      "grad_norm": 0.0013565486297011375,
-      "learning_rate": 0.00019999670126966978,
-      "loss": 46.0,
-      "step": 33826
-    },
-    {
-      "epoch": 2.5863103771240707,
-      "grad_norm": 0.001358826644718647,
-      "learning_rate": 0.0001999967010745692,
-      "loss": 46.0,
-      "step": 33827
-    },
-    {
-      "epoch": 2.5863868341074605,
-      "grad_norm": 0.0026653208769857883,
-      "learning_rate": 0.0001999967008794628,
-      "loss": 46.0,
-      "step": 33828
-    },
-    {
-      "epoch": 2.5864632910908503,
-      "grad_norm": 0.0028843041509389877,
-      "learning_rate": 0.00019999670068435066,
-      "loss": 46.0,
-      "step": 33829
-    },
-    {
-      "epoch": 2.5865397480742396,
-      "grad_norm": 0.0011480532120913267,
-      "learning_rate": 0.00019999670048923273,
-      "loss": 46.0,
-      "step": 33830
-    },
-    {
-      "epoch": 2.5866162050576293,
-      "grad_norm": 0.0025333103258162737,
-      "learning_rate": 0.00019999670029410908,
-      "loss": 46.0,
-      "step": 33831
-    },
-    {
-      "epoch": 2.586692662041019,
-      "grad_norm": 0.001028066617436707,
-      "learning_rate": 0.0001999967000989796,
-      "loss": 46.0,
-      "step": 33832
-    },
-    {
-      "epoch": 2.586769119024409,
-      "grad_norm": 0.0008120053098537028,
-      "learning_rate": 0.0001999966999038444,
-      "loss": 46.0,
-      "step": 33833
-    },
-    {
-      "epoch": 2.5868455760077986,
-      "grad_norm": 0.005214180797338486,
-      "learning_rate": 0.0001999966997087034,
-      "loss": 46.0,
-      "step": 33834
-    },
-    {
-      "epoch": 2.5869220329911884,
-      "grad_norm": 0.0017119479598477483,
-      "learning_rate": 0.00019999669951355667,
-      "loss": 46.0,
-      "step": 33835
-    },
-    {
-      "epoch": 2.586998489974578,
-      "grad_norm": 0.0007648076862096786,
-      "learning_rate": 0.00019999669931840416,
-      "loss": 46.0,
-      "step": 33836
-    },
-    {
-      "epoch": 2.587074946957968,
-      "grad_norm": 0.0007501570507884026,
-      "learning_rate": 0.00019999669912324587,
-      "loss": 46.0,
-      "step": 33837
-    },
-    {
-      "epoch": 2.5871514039413572,
-      "grad_norm": 0.0037393071688711643,
-      "learning_rate": 0.0001999966989280818,
-      "loss": 46.0,
-      "step": 33838
-    },
-    {
-      "epoch": 2.587227860924747,
-      "grad_norm": 0.004476012662053108,
-      "learning_rate": 0.00019999669873291197,
-      "loss": 46.0,
-      "step": 33839
-    },
-    {
-      "epoch": 2.5873043179081368,
-      "grad_norm": 0.0015364629216492176,
-      "learning_rate": 0.00019999669853773634,
-      "loss": 46.0,
-      "step": 33840
-    },
-    {
-      "epoch": 2.5873807748915265,
-      "grad_norm": 0.0007751963566988707,
-      "learning_rate": 0.00019999669834255495,
-      "loss": 46.0,
-      "step": 33841
-    },
-    {
-      "epoch": 2.5874572318749163,
-      "grad_norm": 0.0038668352644890547,
-      "learning_rate": 0.00019999669814736783,
-      "loss": 46.0,
-      "step": 33842
-    },
-    {
-      "epoch": 2.587533688858306,
-      "grad_norm": 0.0022029713727533817,
-      "learning_rate": 0.00019999669795217492,
-      "loss": 46.0,
-      "step": 33843
-    },
-    {
-      "epoch": 2.587610145841696,
-      "grad_norm": 0.0037909240927547216,
-      "learning_rate": 0.00019999669775697625,
-      "loss": 46.0,
-      "step": 33844
-    },
-    {
-      "epoch": 2.5876866028250856,
-      "grad_norm": 0.002746768994256854,
-      "learning_rate": 0.0001999966975617718,
-      "loss": 46.0,
-      "step": 33845
-    },
-    {
-      "epoch": 2.5877630598084753,
-      "grad_norm": 0.0005620379815809429,
-      "learning_rate": 0.00019999669736656158,
-      "loss": 46.0,
-      "step": 33846
-    },
-    {
-      "epoch": 2.587839516791865,
-      "grad_norm": 0.00043551131966523826,
-      "learning_rate": 0.00019999669717134558,
-      "loss": 46.0,
-      "step": 33847
-    },
-    {
-      "epoch": 2.587915973775255,
-      "grad_norm": 0.0010091025615110993,
-      "learning_rate": 0.00019999669697612384,
-      "loss": 46.0,
-      "step": 33848
-    },
-    {
-      "epoch": 2.5879924307586446,
-      "grad_norm": 0.0017041726969182491,
-      "learning_rate": 0.00019999669678089633,
-      "loss": 46.0,
-      "step": 33849
-    },
-    {
-      "epoch": 2.5880688877420344,
-      "grad_norm": 0.001076510176062584,
-      "learning_rate": 0.000199996696585663,
-      "loss": 46.0,
-      "step": 33850
-    },
-    {
-      "epoch": 2.588145344725424,
-      "grad_norm": 0.0009676502668298781,
-      "learning_rate": 0.00019999669639042395,
-      "loss": 46.0,
-      "step": 33851
-    },
-    {
-      "epoch": 2.5882218017088134,
-      "grad_norm": 0.0008860592497512698,
-      "learning_rate": 0.00019999669619517914,
-      "loss": 46.0,
-      "step": 33852
-    },
-    {
-      "epoch": 2.588298258692203,
-      "grad_norm": 0.0005833989707753062,
-      "learning_rate": 0.0001999966959999285,
-      "loss": 46.0,
-      "step": 33853
-    },
-    {
-      "epoch": 2.588374715675593,
-      "grad_norm": 0.001482087536714971,
-      "learning_rate": 0.00019999669580467215,
-      "loss": 46.0,
-      "step": 33854
-    },
-    {
-      "epoch": 2.5884511726589827,
-      "grad_norm": 0.0020538996905088425,
-      "learning_rate": 0.00019999669560941,
-      "loss": 46.0,
-      "step": 33855
-    },
-    {
-      "epoch": 2.5885276296423725,
-      "grad_norm": 0.0028575328178703785,
-      "learning_rate": 0.0001999966954141421,
-      "loss": 46.0,
-      "step": 33856
-    },
-    {
-      "epoch": 2.5886040866257622,
-      "grad_norm": 0.0011522978311404586,
-      "learning_rate": 0.00019999669521886842,
-      "loss": 46.0,
-      "step": 33857
-    },
-    {
-      "epoch": 2.588680543609152,
-      "grad_norm": 0.0020093759521842003,
-      "learning_rate": 0.00019999669502358898,
-      "loss": 46.0,
-      "step": 33858
-    },
-    {
-      "epoch": 2.5887570005925418,
-      "grad_norm": 0.0018720257794484496,
-      "learning_rate": 0.00019999669482830375,
-      "loss": 46.0,
-      "step": 33859
-    },
-    {
-      "epoch": 2.588833457575931,
-      "grad_norm": 0.0005863170954398811,
-      "learning_rate": 0.00019999669463301279,
-      "loss": 46.0,
-      "step": 33860
-    },
-    {
-      "epoch": 2.588909914559321,
-      "grad_norm": 0.0005447547882795334,
-      "learning_rate": 0.00019999669443771602,
-      "loss": 46.0,
-      "step": 33861
-    },
-    {
-      "epoch": 2.5889863715427106,
-      "grad_norm": 0.0011578646954149008,
-      "learning_rate": 0.0001999966942424135,
-      "loss": 46.0,
-      "step": 33862
-    },
-    {
-      "epoch": 2.5890628285261004,
-      "grad_norm": 0.0027566482312977314,
-      "learning_rate": 0.0001999966940471052,
-      "loss": 46.0,
-      "step": 33863
-    },
-    {
-      "epoch": 2.58913928550949,
-      "grad_norm": 0.0008959327824413776,
-      "learning_rate": 0.00019999669385179113,
-      "loss": 46.0,
-      "step": 33864
-    },
-    {
-      "epoch": 2.58921574249288,
-      "grad_norm": 0.0005726159433834255,
-      "learning_rate": 0.0001999966936564713,
-      "loss": 46.0,
-      "step": 33865
-    },
-    {
-      "epoch": 2.5892921994762697,
-      "grad_norm": 0.0006823652074672282,
-      "learning_rate": 0.00019999669346114569,
-      "loss": 46.0,
-      "step": 33866
-    },
-    {
-      "epoch": 2.5893686564596594,
-      "grad_norm": 0.0037108648102730513,
-      "learning_rate": 0.0001999966932658143,
-      "loss": 46.0,
-      "step": 33867
-    },
-    {
-      "epoch": 2.589445113443049,
-      "grad_norm": 0.003020165953785181,
-      "learning_rate": 0.00019999669307047718,
-      "loss": 46.0,
-      "step": 33868
-    },
-    {
-      "epoch": 2.589521570426439,
-      "grad_norm": 0.005767005030065775,
-      "learning_rate": 0.00019999669287513428,
-      "loss": 46.0,
-      "step": 33869
-    },
-    {
-      "epoch": 2.5895980274098287,
-      "grad_norm": 0.0068554868921637535,
-      "learning_rate": 0.0001999966926797856,
-      "loss": 46.0,
-      "step": 33870
-    },
-    {
-      "epoch": 2.5896744843932185,
-      "grad_norm": 0.000507902936078608,
-      "learning_rate": 0.00019999669248443113,
-      "loss": 46.0,
-      "step": 33871
-    },
-    {
-      "epoch": 2.589750941376608,
-      "grad_norm": 0.000857727078255266,
-      "learning_rate": 0.00019999669228907093,
-      "loss": 46.0,
-      "step": 33872
-    },
-    {
-      "epoch": 2.589827398359998,
-      "grad_norm": 0.0010478412732481956,
-      "learning_rate": 0.00019999669209370494,
-      "loss": 46.0,
-      "step": 33873
-    },
-    {
-      "epoch": 2.5899038553433873,
-      "grad_norm": 0.0006254063337109983,
-      "learning_rate": 0.0001999966918983332,
-      "loss": 46.0,
-      "step": 33874
-    },
-    {
-      "epoch": 2.589980312326777,
-      "grad_norm": 0.0029933182522654533,
-      "learning_rate": 0.00019999669170295566,
-      "loss": 46.0,
-      "step": 33875
-    },
-    {
-      "epoch": 2.590056769310167,
-      "grad_norm": 0.0027603567577898502,
-      "learning_rate": 0.00019999669150757234,
-      "loss": 46.0,
-      "step": 33876
-    },
-    {
-      "epoch": 2.5901332262935566,
-      "grad_norm": 0.004642440937459469,
-      "learning_rate": 0.0001999966913121833,
-      "loss": 46.0,
-      "step": 33877
-    },
-    {
-      "epoch": 2.5902096832769463,
-      "grad_norm": 0.000852566328831017,
-      "learning_rate": 0.00019999669111678845,
-      "loss": 46.0,
-      "step": 33878
-    },
-    {
-      "epoch": 2.590286140260336,
-      "grad_norm": 0.002421814249828458,
-      "learning_rate": 0.00019999669092138787,
-      "loss": 46.0,
-      "step": 33879
-    },
-    {
-      "epoch": 2.590362597243726,
-      "grad_norm": 0.0021579849999397993,
-      "learning_rate": 0.00019999669072598148,
-      "loss": 46.0,
-      "step": 33880
-    },
-    {
-      "epoch": 2.5904390542271156,
-      "grad_norm": 0.001946039847098291,
-      "learning_rate": 0.00019999669053056933,
-      "loss": 46.0,
-      "step": 33881
-    },
-    {
-      "epoch": 2.590515511210505,
-      "grad_norm": 0.0004008824471384287,
-      "learning_rate": 0.00019999669033515143,
-      "loss": 46.0,
-      "step": 33882
-    },
-    {
-      "epoch": 2.5905919681938947,
-      "grad_norm": 0.0016343913739547133,
-      "learning_rate": 0.00019999669013972776,
-      "loss": 46.0,
-      "step": 33883
-    },
-    {
-      "epoch": 2.5906684251772845,
-      "grad_norm": 0.0013753061648458242,
-      "learning_rate": 0.0001999966899442983,
-      "loss": 46.0,
-      "step": 33884
-    },
-    {
-      "epoch": 2.5907448821606742,
-      "grad_norm": 0.0008443729602731764,
-      "learning_rate": 0.0001999966897488631,
-      "loss": 46.0,
-      "step": 33885
-    },
-    {
-      "epoch": 2.590821339144064,
-      "grad_norm": 0.0020147201139479876,
-      "learning_rate": 0.00019999668955342207,
-      "loss": 46.0,
-      "step": 33886
-    },
-    {
-      "epoch": 2.5908977961274537,
-      "grad_norm": 0.001294368994422257,
-      "learning_rate": 0.00019999668935797533,
-      "loss": 46.0,
-      "step": 33887
-    },
-    {
-      "epoch": 2.5909742531108435,
-      "grad_norm": 0.000877061509527266,
-      "learning_rate": 0.0001999966891625228,
-      "loss": 46.0,
-      "step": 33888
-    },
-    {
-      "epoch": 2.5910507100942333,
-      "grad_norm": 0.0011326199164614081,
-      "learning_rate": 0.00019999668896706453,
-      "loss": 46.0,
-      "step": 33889
-    },
-    {
-      "epoch": 2.591127167077623,
-      "grad_norm": 0.0012657109182327986,
-      "learning_rate": 0.00019999668877160044,
-      "loss": 46.0,
-      "step": 33890
-    },
-    {
-      "epoch": 2.591203624061013,
-      "grad_norm": 0.0007595903007313609,
-      "learning_rate": 0.0001999966885761306,
-      "loss": 46.0,
-      "step": 33891
-    },
-    {
-      "epoch": 2.5912800810444026,
-      "grad_norm": 0.0013786893105134368,
-      "learning_rate": 0.000199996688380655,
-      "loss": 46.0,
-      "step": 33892
-    },
-    {
-      "epoch": 2.5913565380277923,
-      "grad_norm": 0.0011777592590078712,
-      "learning_rate": 0.00019999668818517364,
-      "loss": 46.0,
-      "step": 33893
-    },
-    {
-      "epoch": 2.591432995011182,
-      "grad_norm": 0.0014845497207716107,
-      "learning_rate": 0.0001999966879896865,
-      "loss": 46.0,
-      "step": 33894
-    },
-    {
-      "epoch": 2.5915094519945714,
-      "grad_norm": 0.001108934753574431,
-      "learning_rate": 0.0001999966877941936,
-      "loss": 46.0,
-      "step": 33895
-    },
-    {
-      "epoch": 2.591585908977961,
-      "grad_norm": 0.0019122897647321224,
-      "learning_rate": 0.0001999966875986949,
-      "loss": 46.0,
-      "step": 33896
-    },
-    {
-      "epoch": 2.591662365961351,
-      "grad_norm": 0.0029160024132579565,
-      "learning_rate": 0.00019999668740319044,
-      "loss": 46.0,
-      "step": 33897
-    },
-    {
-      "epoch": 2.5917388229447407,
-      "grad_norm": 0.003632324282079935,
-      "learning_rate": 0.00019999668720768022,
-      "loss": 46.0,
-      "step": 33898
-    },
-    {
-      "epoch": 2.5918152799281304,
-      "grad_norm": 0.003607020480558276,
-      "learning_rate": 0.00019999668701216425,
-      "loss": 46.0,
-      "step": 33899
-    },
-    {
-      "epoch": 2.59189173691152,
-      "grad_norm": 0.00321124866604805,
-      "learning_rate": 0.0001999966868166425,
-      "loss": 46.0,
-      "step": 33900
-    },
-    {
-      "epoch": 2.59196819389491,
-      "grad_norm": 0.0006294617196545005,
-      "learning_rate": 0.00019999668662111497,
-      "loss": 46.0,
-      "step": 33901
-    },
-    {
-      "epoch": 2.5920446508782997,
-      "grad_norm": 0.0020169615745544434,
-      "learning_rate": 0.00019999668642558166,
-      "loss": 46.0,
-      "step": 33902
-    },
-    {
-      "epoch": 2.5921211078616895,
-      "grad_norm": 0.001106072450056672,
-      "learning_rate": 0.0001999966862300426,
-      "loss": 46.0,
-      "step": 33903
-    },
-    {
-      "epoch": 2.592197564845079,
-      "grad_norm": 0.0036720754578709602,
-      "learning_rate": 0.00019999668603449777,
-      "loss": 46.0,
-      "step": 33904
-    },
-    {
-      "epoch": 2.5922740218284686,
-      "grad_norm": 0.0009459693101234734,
-      "learning_rate": 0.00019999668583894719,
-      "loss": 46.0,
-      "step": 33905
-    },
-    {
-      "epoch": 2.5923504788118583,
-      "grad_norm": 0.0033843584824353456,
-      "learning_rate": 0.0001999966856433908,
-      "loss": 46.0,
-      "step": 33906
-    },
-    {
-      "epoch": 2.592426935795248,
-      "grad_norm": 0.0013822912005707622,
-      "learning_rate": 0.00019999668544782865,
-      "loss": 46.0,
-      "step": 33907
-    },
-    {
-      "epoch": 2.592503392778638,
-      "grad_norm": 0.0008619792060926557,
-      "learning_rate": 0.00019999668525226075,
-      "loss": 46.0,
-      "step": 33908
-    },
-    {
-      "epoch": 2.5925798497620276,
-      "grad_norm": 0.0005219600861892104,
-      "learning_rate": 0.00019999668505668705,
-      "loss": 46.0,
-      "step": 33909
-    },
-    {
-      "epoch": 2.5926563067454174,
-      "grad_norm": 0.0026846961118280888,
-      "learning_rate": 0.0001999966848611076,
-      "loss": 46.0,
-      "step": 33910
-    },
-    {
-      "epoch": 2.592732763728807,
-      "grad_norm": 0.0019448735984042287,
-      "learning_rate": 0.00019999668466552238,
-      "loss": 46.0,
-      "step": 33911
-    },
-    {
-      "epoch": 2.592809220712197,
-      "grad_norm": 0.005883508827537298,
-      "learning_rate": 0.0001999966844699314,
-      "loss": 46.0,
-      "step": 33912
-    },
-    {
-      "epoch": 2.5928856776955866,
-      "grad_norm": 0.0012319523375481367,
-      "learning_rate": 0.00019999668427433465,
-      "loss": 46.0,
-      "step": 33913
-    },
-    {
-      "epoch": 2.5929621346789764,
-      "grad_norm": 0.0009735599160194397,
-      "learning_rate": 0.0001999966840787321,
-      "loss": 46.0,
-      "step": 33914
-    },
-    {
-      "epoch": 2.593038591662366,
-      "grad_norm": 0.0007737409323453903,
-      "learning_rate": 0.0001999966838831238,
-      "loss": 46.0,
-      "step": 33915
-    },
-    {
-      "epoch": 2.593115048645756,
-      "grad_norm": 0.001422416535206139,
-      "learning_rate": 0.00019999668368750974,
-      "loss": 46.0,
-      "step": 33916
-    },
-    {
-      "epoch": 2.5931915056291452,
-      "grad_norm": 0.0009661955991759896,
-      "learning_rate": 0.0001999966834918899,
-      "loss": 46.0,
-      "step": 33917
-    },
-    {
-      "epoch": 2.593267962612535,
-      "grad_norm": 0.0022469093091785908,
-      "learning_rate": 0.0001999966832962643,
-      "loss": 46.0,
-      "step": 33918
-    },
-    {
-      "epoch": 2.5933444195959248,
-      "grad_norm": 0.0009600328048691154,
-      "learning_rate": 0.00019999668310063294,
-      "loss": 46.0,
-      "step": 33919
-    },
-    {
-      "epoch": 2.5934208765793145,
-      "grad_norm": 0.0019555389881134033,
-      "learning_rate": 0.0001999966829049958,
-      "loss": 46.0,
-      "step": 33920
-    },
-    {
-      "epoch": 2.5934973335627043,
-      "grad_norm": 0.0006724236882291734,
-      "learning_rate": 0.0001999966827093529,
-      "loss": 46.0,
-      "step": 33921
-    },
-    {
-      "epoch": 2.593573790546094,
-      "grad_norm": 0.0019694152288138866,
-      "learning_rate": 0.0001999966825137042,
-      "loss": 46.0,
-      "step": 33922
-    },
-    {
-      "epoch": 2.593650247529484,
-      "grad_norm": 0.002382457023486495,
-      "learning_rate": 0.00019999668231804974,
-      "loss": 46.0,
-      "step": 33923
-    },
-    {
-      "epoch": 2.5937267045128736,
-      "grad_norm": 0.0046404823660850525,
-      "learning_rate": 0.00019999668212238953,
-      "loss": 46.0,
-      "step": 33924
-    },
-    {
-      "epoch": 2.593803161496263,
-      "grad_norm": 0.001402257476001978,
-      "learning_rate": 0.00019999668192672353,
-      "loss": 46.0,
-      "step": 33925
-    },
-    {
-      "epoch": 2.5938796184796526,
-      "grad_norm": 0.0011917828815057874,
-      "learning_rate": 0.00019999668173105177,
-      "loss": 46.0,
-      "step": 33926
-    },
-    {
-      "epoch": 2.5939560754630424,
-      "grad_norm": 0.003565862076357007,
-      "learning_rate": 0.00019999668153537425,
-      "loss": 46.0,
-      "step": 33927
-    },
-    {
-      "epoch": 2.594032532446432,
-      "grad_norm": 0.0021697422489523888,
-      "learning_rate": 0.00019999668133969097,
-      "loss": 46.0,
-      "step": 33928
-    },
-    {
-      "epoch": 2.594108989429822,
-      "grad_norm": 0.002213774248957634,
-      "learning_rate": 0.0001999966811440019,
-      "loss": 46.0,
-      "step": 33929
-    },
-    {
-      "epoch": 2.5941854464132117,
-      "grad_norm": 0.0020207951311022043,
-      "learning_rate": 0.00019999668094830705,
-      "loss": 46.0,
-      "step": 33930
-    },
-    {
-      "epoch": 2.5942619033966015,
-      "grad_norm": 0.0025760820135474205,
-      "learning_rate": 0.00019999668075260644,
-      "loss": 46.0,
-      "step": 33931
-    },
-    {
-      "epoch": 2.594338360379991,
-      "grad_norm": 0.0029874928295612335,
-      "learning_rate": 0.00019999668055690006,
-      "loss": 46.0,
-      "step": 33932
-    },
-    {
-      "epoch": 2.594414817363381,
-      "grad_norm": 0.0019100755453109741,
-      "learning_rate": 0.00019999668036118796,
-      "loss": 46.0,
-      "step": 33933
-    },
-    {
-      "epoch": 2.5944912743467707,
-      "grad_norm": 0.0020764644723385572,
-      "learning_rate": 0.00019999668016547,
-      "loss": 46.0,
-      "step": 33934
-    },
-    {
-      "epoch": 2.5945677313301605,
-      "grad_norm": 0.0012909704819321632,
-      "learning_rate": 0.00019999667996974634,
-      "loss": 46.0,
-      "step": 33935
-    },
-    {
-      "epoch": 2.5946441883135503,
-      "grad_norm": 0.0033717325422912836,
-      "learning_rate": 0.0001999966797740169,
-      "loss": 46.0,
-      "step": 33936
-    },
-    {
-      "epoch": 2.59472064529694,
-      "grad_norm": 0.0011126762256026268,
-      "learning_rate": 0.00019999667957828167,
-      "loss": 46.0,
-      "step": 33937
-    },
-    {
-      "epoch": 2.59479710228033,
-      "grad_norm": 0.0026965197175741196,
-      "learning_rate": 0.00019999667938254068,
-      "loss": 46.0,
-      "step": 33938
-    },
-    {
-      "epoch": 2.594873559263719,
-      "grad_norm": 0.0013524913229048252,
-      "learning_rate": 0.0001999966791867939,
-      "loss": 46.0,
-      "step": 33939
-    },
-    {
-      "epoch": 2.594950016247109,
-      "grad_norm": 0.0035900238435715437,
-      "learning_rate": 0.00019999667899104137,
-      "loss": 46.0,
-      "step": 33940
-    },
-    {
-      "epoch": 2.5950264732304986,
-      "grad_norm": 0.0008633043034933507,
-      "learning_rate": 0.0001999966787952831,
-      "loss": 46.0,
-      "step": 33941
-    },
-    {
-      "epoch": 2.5951029302138884,
-      "grad_norm": 0.0011326251551508904,
-      "learning_rate": 0.00019999667859951903,
-      "loss": 46.0,
-      "step": 33942
-    },
-    {
-      "epoch": 2.595179387197278,
-      "grad_norm": 0.0024322012905031443,
-      "learning_rate": 0.0001999966784037492,
-      "loss": 46.0,
-      "step": 33943
-    },
-    {
-      "epoch": 2.595255844180668,
-      "grad_norm": 0.0012551966356113553,
-      "learning_rate": 0.0001999966782079736,
-      "loss": 46.0,
-      "step": 33944
-    },
-    {
-      "epoch": 2.5953323011640577,
-      "grad_norm": 0.0033246458042412996,
-      "learning_rate": 0.0001999966780121922,
-      "loss": 46.0,
-      "step": 33945
-    },
-    {
-      "epoch": 2.5954087581474474,
-      "grad_norm": 0.0011773412115871906,
-      "learning_rate": 0.00019999667781640508,
-      "loss": 46.0,
-      "step": 33946
-    },
-    {
-      "epoch": 2.5954852151308367,
-      "grad_norm": 0.0021568629890680313,
-      "learning_rate": 0.00019999667762061213,
-      "loss": 46.0,
-      "step": 33947
-    },
-    {
-      "epoch": 2.5955616721142265,
-      "grad_norm": 0.0020415030885487795,
-      "learning_rate": 0.00019999667742481348,
-      "loss": 46.0,
-      "step": 33948
-    },
-    {
-      "epoch": 2.5956381290976163,
-      "grad_norm": 0.0015277519123628736,
-      "learning_rate": 0.00019999667722900904,
-      "loss": 46.0,
-      "step": 33949
-    },
-    {
-      "epoch": 2.595714586081006,
-      "grad_norm": 0.0013656937517225742,
-      "learning_rate": 0.0001999966770331988,
-      "loss": 46.0,
-      "step": 33950
-    },
-    {
-      "epoch": 2.595791043064396,
-      "grad_norm": 0.001709373202174902,
-      "learning_rate": 0.0001999966768373828,
-      "loss": 46.0,
-      "step": 33951
-    },
-    {
-      "epoch": 2.5958675000477855,
-      "grad_norm": 0.0029922376852482557,
-      "learning_rate": 0.00019999667664156106,
-      "loss": 46.0,
-      "step": 33952
-    },
-    {
-      "epoch": 2.5959439570311753,
-      "grad_norm": 0.0014170288341119885,
-      "learning_rate": 0.00019999667644573352,
-      "loss": 46.0,
-      "step": 33953
-    },
-    {
-      "epoch": 2.596020414014565,
-      "grad_norm": 0.0010529591236263514,
-      "learning_rate": 0.00019999667624990024,
-      "loss": 46.0,
-      "step": 33954
-    },
-    {
-      "epoch": 2.596096870997955,
-      "grad_norm": 0.0014678220031782985,
-      "learning_rate": 0.00019999667605406115,
-      "loss": 46.0,
-      "step": 33955
-    },
-    {
-      "epoch": 2.5961733279813446,
-      "grad_norm": 0.0017287025693804026,
-      "learning_rate": 0.00019999667585821632,
-      "loss": 46.0,
-      "step": 33956
-    },
-    {
-      "epoch": 2.5962497849647344,
-      "grad_norm": 0.0024102828465402126,
-      "learning_rate": 0.00019999667566236574,
-      "loss": 46.0,
-      "step": 33957
-    },
-    {
-      "epoch": 2.596326241948124,
-      "grad_norm": 0.006789756938815117,
-      "learning_rate": 0.00019999667546650933,
-      "loss": 46.0,
-      "step": 33958
-    },
-    {
-      "epoch": 2.596402698931514,
-      "grad_norm": 0.0015074698021635413,
-      "learning_rate": 0.0001999966752706472,
-      "loss": 46.0,
-      "step": 33959
-    },
-    {
-      "epoch": 2.5964791559149036,
-      "grad_norm": 0.0037692601326853037,
-      "learning_rate": 0.00019999667507477928,
-      "loss": 46.0,
-      "step": 33960
-    },
-    {
-      "epoch": 2.596555612898293,
-      "grad_norm": 0.0010296533582732081,
-      "learning_rate": 0.00019999667487890558,
-      "loss": 46.0,
-      "step": 33961
-    },
-    {
-      "epoch": 2.5966320698816827,
-      "grad_norm": 0.0011312720598652959,
-      "learning_rate": 0.00019999667468302614,
-      "loss": 46.0,
-      "step": 33962
-    },
-    {
-      "epoch": 2.5967085268650725,
-      "grad_norm": 0.004012007731944323,
-      "learning_rate": 0.00019999667448714092,
-      "loss": 46.0,
-      "step": 33963
-    },
-    {
-      "epoch": 2.5967849838484622,
-      "grad_norm": 0.002886988688260317,
-      "learning_rate": 0.00019999667429124993,
-      "loss": 46.0,
-      "step": 33964
-    },
-    {
-      "epoch": 2.596861440831852,
-      "grad_norm": 0.0023925844579935074,
-      "learning_rate": 0.00019999667409535316,
-      "loss": 46.0,
-      "step": 33965
-    },
-    {
-      "epoch": 2.5969378978152418,
-      "grad_norm": 0.0028296641539782286,
-      "learning_rate": 0.00019999667389945065,
-      "loss": 46.0,
-      "step": 33966
-    },
-    {
-      "epoch": 2.5970143547986315,
-      "grad_norm": 0.0014572347281500697,
-      "learning_rate": 0.00019999667370354236,
-      "loss": 46.0,
-      "step": 33967
-    },
-    {
-      "epoch": 2.5970908117820213,
-      "grad_norm": 0.002130937995389104,
-      "learning_rate": 0.00019999667350762828,
-      "loss": 46.0,
-      "step": 33968
-    },
-    {
-      "epoch": 2.5971672687654106,
-      "grad_norm": 0.0013775993138551712,
-      "learning_rate": 0.00019999667331170847,
-      "loss": 46.0,
-      "step": 33969
-    },
-    {
-      "epoch": 2.5972437257488004,
-      "grad_norm": 0.002277441555634141,
-      "learning_rate": 0.00019999667311578284,
-      "loss": 46.0,
-      "step": 33970
-    },
-    {
-      "epoch": 2.59732018273219,
-      "grad_norm": 0.0018924273317679763,
-      "learning_rate": 0.00019999667291985146,
-      "loss": 46.0,
-      "step": 33971
-    },
-    {
-      "epoch": 2.59739663971558,
-      "grad_norm": 0.005080229137092829,
-      "learning_rate": 0.00019999667272391434,
-      "loss": 46.0,
-      "step": 33972
-    },
-    {
-      "epoch": 2.5974730966989696,
-      "grad_norm": 0.0018999003805220127,
-      "learning_rate": 0.0001999966725279714,
-      "loss": 46.0,
-      "step": 33973
-    },
-    {
-      "epoch": 2.5975495536823594,
-      "grad_norm": 0.0007585359853692353,
-      "learning_rate": 0.00019999667233202274,
-      "loss": 46.0,
-      "step": 33974
-    },
-    {
-      "epoch": 2.597626010665749,
-      "grad_norm": 0.0010942388325929642,
-      "learning_rate": 0.00019999667213606827,
-      "loss": 46.0,
-      "step": 33975
-    },
-    {
-      "epoch": 2.597702467649139,
-      "grad_norm": 0.0010093394666910172,
-      "learning_rate": 0.00019999667194010805,
-      "loss": 46.0,
-      "step": 33976
-    },
-    {
-      "epoch": 2.5977789246325287,
-      "grad_norm": 0.0016063363291323185,
-      "learning_rate": 0.00019999667174414208,
-      "loss": 46.0,
-      "step": 33977
-    },
-    {
-      "epoch": 2.5978553816159184,
-      "grad_norm": 0.004621224012225866,
-      "learning_rate": 0.0001999966715481703,
-      "loss": 46.0,
-      "step": 33978
-    },
-    {
-      "epoch": 2.597931838599308,
-      "grad_norm": 0.0021016260143369436,
-      "learning_rate": 0.00019999667135219275,
-      "loss": 46.0,
-      "step": 33979
-    },
-    {
-      "epoch": 2.598008295582698,
-      "grad_norm": 0.0015274283941835165,
-      "learning_rate": 0.0001999966711562095,
-      "loss": 46.0,
-      "step": 33980
-    },
-    {
-      "epoch": 2.5980847525660877,
-      "grad_norm": 0.0007747607305645943,
-      "learning_rate": 0.00019999667096022038,
-      "loss": 46.0,
-      "step": 33981
-    },
-    {
-      "epoch": 2.5981612095494775,
-      "grad_norm": 0.002963385544717312,
-      "learning_rate": 0.00019999667076422558,
-      "loss": 46.0,
-      "step": 33982
-    },
-    {
-      "epoch": 2.598237666532867,
-      "grad_norm": 0.0015990951796993613,
-      "learning_rate": 0.00019999667056822497,
-      "loss": 46.0,
-      "step": 33983
-    },
-    {
-      "epoch": 2.5983141235162566,
-      "grad_norm": 0.0021712330635637045,
-      "learning_rate": 0.0001999966703722186,
-      "loss": 46.0,
-      "step": 33984
-    },
-    {
-      "epoch": 2.5983905804996463,
-      "grad_norm": 0.0018673304002732038,
-      "learning_rate": 0.00019999667017620641,
-      "loss": 46.0,
-      "step": 33985
-    },
-    {
-      "epoch": 2.598467037483036,
-      "grad_norm": 0.003089440520852804,
-      "learning_rate": 0.00019999666998018852,
-      "loss": 46.0,
-      "step": 33986
-    },
-    {
-      "epoch": 2.598543494466426,
-      "grad_norm": 0.0027108315844088793,
-      "learning_rate": 0.00019999666978416484,
-      "loss": 46.0,
-      "step": 33987
-    },
-    {
-      "epoch": 2.5986199514498156,
-      "grad_norm": 0.0011818775674328208,
-      "learning_rate": 0.0001999966695881354,
-      "loss": 46.0,
-      "step": 33988
-    },
-    {
-      "epoch": 2.5986964084332054,
-      "grad_norm": 0.004194196779280901,
-      "learning_rate": 0.00019999666939210018,
-      "loss": 46.0,
-      "step": 33989
-    },
-    {
-      "epoch": 2.598772865416595,
-      "grad_norm": 0.0020329616963863373,
-      "learning_rate": 0.00019999666919605916,
-      "loss": 46.0,
-      "step": 33990
-    },
-    {
-      "epoch": 2.5988493223999845,
-      "grad_norm": 0.0012380784610286355,
-      "learning_rate": 0.00019999666900001243,
-      "loss": 46.0,
-      "step": 33991
-    },
-    {
-      "epoch": 2.598925779383374,
-      "grad_norm": 0.0014517760137096047,
-      "learning_rate": 0.0001999966688039599,
-      "loss": 46.0,
-      "step": 33992
-    },
-    {
-      "epoch": 2.599002236366764,
-      "grad_norm": 0.0012300026137381792,
-      "learning_rate": 0.00019999666860790158,
-      "loss": 46.0,
-      "step": 33993
-    },
-    {
-      "epoch": 2.5990786933501537,
-      "grad_norm": 0.0010018979664891958,
-      "learning_rate": 0.00019999666841183752,
-      "loss": 46.0,
-      "step": 33994
-    },
-    {
-      "epoch": 2.5991551503335435,
-      "grad_norm": 0.0036307142581790686,
-      "learning_rate": 0.00019999666821576769,
-      "loss": 46.0,
-      "step": 33995
-    },
-    {
-      "epoch": 2.5992316073169333,
-      "grad_norm": 0.0016579385846853256,
-      "learning_rate": 0.00019999666801969208,
-      "loss": 46.0,
-      "step": 33996
-    },
-    {
-      "epoch": 2.599308064300323,
-      "grad_norm": 0.0024304785765707493,
-      "learning_rate": 0.0001999966678236107,
-      "loss": 46.0,
-      "step": 33997
-    },
-    {
-      "epoch": 2.599384521283713,
-      "grad_norm": 0.0007768187206238508,
-      "learning_rate": 0.00019999666762752355,
-      "loss": 46.0,
-      "step": 33998
-    },
-    {
-      "epoch": 2.5994609782671025,
-      "grad_norm": 0.0021852378267794847,
-      "learning_rate": 0.00019999666743143063,
-      "loss": 46.0,
-      "step": 33999
-    },
-    {
-      "epoch": 2.5995374352504923,
-      "grad_norm": 0.0020817629992961884,
-      "learning_rate": 0.00019999666723533196,
-      "loss": 46.0,
-      "step": 34000
-    },
-    {
-      "epoch": 2.599613892233882,
-      "grad_norm": 0.0021746279671788216,
-      "learning_rate": 0.0001999966670392275,
-      "loss": 46.0,
-      "step": 34001
-    },
-    {
-      "epoch": 2.599690349217272,
-      "grad_norm": 0.003117659827694297,
-      "learning_rate": 0.0001999966668431173,
-      "loss": 46.0,
-      "step": 34002
-    },
-    {
-      "epoch": 2.5997668062006616,
-      "grad_norm": 0.004377887584269047,
-      "learning_rate": 0.00019999666664700127,
-      "loss": 46.0,
-      "step": 34003
-    },
-    {
-      "epoch": 2.5998432631840513,
-      "grad_norm": 0.0021769730374217033,
-      "learning_rate": 0.00019999666645087954,
-      "loss": 46.0,
-      "step": 34004
-    },
-    {
-      "epoch": 2.5999197201674407,
-      "grad_norm": 0.0030063041485846043,
-      "learning_rate": 0.000199996666254752,
-      "loss": 46.0,
-      "step": 34005
-    },
-    {
-      "epoch": 2.5999961771508304,
-      "grad_norm": 0.0011040999088436365,
-      "learning_rate": 0.0001999966660586187,
-      "loss": 46.0,
-      "step": 34006
-    },
-    {
-      "epoch": 2.60007263413422,
-      "grad_norm": 0.0009236817131750286,
-      "learning_rate": 0.00019999666586247963,
-      "loss": 46.0,
-      "step": 34007
-    },
-    {
-      "epoch": 2.60014909111761,
-      "grad_norm": 0.0016864055069163442,
-      "learning_rate": 0.0001999966656663348,
-      "loss": 46.0,
-      "step": 34008
-    },
-    {
-      "epoch": 2.6002255481009997,
-      "grad_norm": 0.0010945064714178443,
-      "learning_rate": 0.0001999966654701842,
-      "loss": 46.0,
-      "step": 34009
-    },
-    {
-      "epoch": 2.6003020050843895,
-      "grad_norm": 0.00214216741733253,
-      "learning_rate": 0.00019999666527402782,
-      "loss": 46.0,
-      "step": 34010
-    },
-    {
-      "epoch": 2.6003784620677792,
-      "grad_norm": 0.003336291527375579,
-      "learning_rate": 0.00019999666507786567,
-      "loss": 46.0,
-      "step": 34011
-    },
-    {
-      "epoch": 2.600454919051169,
-      "grad_norm": 0.0020719943568110466,
-      "learning_rate": 0.00019999666488169774,
-      "loss": 46.0,
-      "step": 34012
-    },
-    {
-      "epoch": 2.6005313760345583,
-      "grad_norm": 0.004970516078174114,
-      "learning_rate": 0.00019999666468552407,
-      "loss": 46.0,
-      "step": 34013
-    },
-    {
-      "epoch": 2.600607833017948,
-      "grad_norm": 0.006673033349215984,
-      "learning_rate": 0.0001999966644893446,
-      "loss": 46.0,
-      "step": 34014
-    },
-    {
-      "epoch": 2.600684290001338,
-      "grad_norm": 0.0027091046795248985,
-      "learning_rate": 0.0001999966642931594,
-      "loss": 46.0,
-      "step": 34015
-    },
-    {
-      "epoch": 2.6007607469847276,
-      "grad_norm": 0.0013261794811114669,
-      "learning_rate": 0.0001999966640969684,
-      "loss": 46.0,
-      "step": 34016
-    },
-    {
-      "epoch": 2.6008372039681174,
-      "grad_norm": 0.0004412404086906463,
-      "learning_rate": 0.00019999666390077163,
-      "loss": 46.0,
-      "step": 34017
-    },
-    {
-      "epoch": 2.600913660951507,
-      "grad_norm": 0.0014693073462694883,
-      "learning_rate": 0.00019999666370456911,
-      "loss": 46.0,
-      "step": 34018
-    },
-    {
-      "epoch": 2.600990117934897,
-      "grad_norm": 0.0025201099924743176,
-      "learning_rate": 0.0001999966635083608,
-      "loss": 46.0,
-      "step": 34019
-    },
-    {
-      "epoch": 2.6010665749182866,
-      "grad_norm": 0.002558163134381175,
-      "learning_rate": 0.00019999666331214675,
-      "loss": 46.0,
-      "step": 34020
-    },
-    {
-      "epoch": 2.6011430319016764,
-      "grad_norm": 0.0022936142049729824,
-      "learning_rate": 0.00019999666311592692,
-      "loss": 46.0,
-      "step": 34021
-    },
-    {
-      "epoch": 2.601219488885066,
-      "grad_norm": 0.0013978230999782681,
-      "learning_rate": 0.00019999666291970128,
-      "loss": 46.0,
-      "step": 34022
-    },
-    {
-      "epoch": 2.601295945868456,
-      "grad_norm": 0.0016516291070729494,
-      "learning_rate": 0.0001999966627234699,
-      "loss": 46.0,
-      "step": 34023
-    },
-    {
-      "epoch": 2.6013724028518457,
-      "grad_norm": 0.001629309612326324,
-      "learning_rate": 0.00019999666252723278,
-      "loss": 46.0,
-      "step": 34024
-    },
-    {
-      "epoch": 2.6014488598352354,
-      "grad_norm": 0.0021412211935967207,
-      "learning_rate": 0.00019999666233098986,
-      "loss": 46.0,
-      "step": 34025
-    },
-    {
-      "epoch": 2.6015253168186248,
-      "grad_norm": 0.0010138375218957663,
-      "learning_rate": 0.00019999666213474116,
-      "loss": 46.0,
-      "step": 34026
-    },
-    {
-      "epoch": 2.6016017738020145,
-      "grad_norm": 0.004015614278614521,
-      "learning_rate": 0.00019999666193848672,
-      "loss": 46.0,
-      "step": 34027
-    },
-    {
-      "epoch": 2.6016782307854043,
-      "grad_norm": 0.0007702329312451184,
-      "learning_rate": 0.0001999966617422265,
-      "loss": 46.0,
-      "step": 34028
-    },
-    {
-      "epoch": 2.601754687768794,
-      "grad_norm": 0.002063112799078226,
-      "learning_rate": 0.00019999666154596048,
-      "loss": 46.0,
-      "step": 34029
-    },
-    {
-      "epoch": 2.601831144752184,
-      "grad_norm": 0.001537511358037591,
-      "learning_rate": 0.00019999666134968872,
-      "loss": 46.0,
-      "step": 34030
-    },
-    {
-      "epoch": 2.6019076017355736,
-      "grad_norm": 0.001073982915841043,
-      "learning_rate": 0.0001999966611534112,
-      "loss": 46.0,
-      "step": 34031
-    },
-    {
-      "epoch": 2.6019840587189633,
-      "grad_norm": 0.0009681865922175348,
-      "learning_rate": 0.0001999966609571279,
-      "loss": 46.0,
-      "step": 34032
-    },
-    {
-      "epoch": 2.602060515702353,
-      "grad_norm": 0.001937026041559875,
-      "learning_rate": 0.00019999666076083884,
-      "loss": 46.0,
-      "step": 34033
-    },
-    {
-      "epoch": 2.602136972685743,
-      "grad_norm": 0.0020606957841664553,
-      "learning_rate": 0.00019999666056454398,
-      "loss": 46.0,
-      "step": 34034
-    },
-    {
-      "epoch": 2.602213429669132,
-      "grad_norm": 0.0016524831298738718,
-      "learning_rate": 0.00019999666036824338,
-      "loss": 46.0,
-      "step": 34035
-    },
-    {
-      "epoch": 2.602289886652522,
-      "grad_norm": 0.0014185130130499601,
-      "learning_rate": 0.000199996660171937,
-      "loss": 46.0,
-      "step": 34036
-    },
-    {
-      "epoch": 2.6023663436359117,
-      "grad_norm": 0.002785137854516506,
-      "learning_rate": 0.00019999665997562488,
-      "loss": 46.0,
-      "step": 34037
-    },
-    {
-      "epoch": 2.6024428006193014,
-      "grad_norm": 0.0029016488697379827,
-      "learning_rate": 0.00019999665977930695,
-      "loss": 46.0,
-      "step": 34038
-    },
-    {
-      "epoch": 2.602519257602691,
-      "grad_norm": 0.0011367335682734847,
-      "learning_rate": 0.00019999665958298326,
-      "loss": 46.0,
-      "step": 34039
-    },
-    {
-      "epoch": 2.602595714586081,
-      "grad_norm": 0.001681118505075574,
-      "learning_rate": 0.00019999665938665379,
-      "loss": 46.0,
-      "step": 34040
-    },
-    {
-      "epoch": 2.6026721715694707,
-      "grad_norm": 0.0014663425972685218,
-      "learning_rate": 0.00019999665919031857,
-      "loss": 46.0,
-      "step": 34041
-    },
-    {
-      "epoch": 2.6027486285528605,
-      "grad_norm": 0.0017261416651308537,
-      "learning_rate": 0.0001999966589939776,
-      "loss": 46.0,
-      "step": 34042
-    },
-    {
-      "epoch": 2.6028250855362502,
-      "grad_norm": 0.0016541347140446305,
-      "learning_rate": 0.00019999665879763084,
-      "loss": 46.0,
-      "step": 34043
-    },
-    {
-      "epoch": 2.60290154251964,
-      "grad_norm": 0.0010272046783939004,
-      "learning_rate": 0.0001999966586012783,
-      "loss": 46.0,
-      "step": 34044
-    },
-    {
-      "epoch": 2.6029779995030298,
-      "grad_norm": 0.0005189761868678033,
-      "learning_rate": 0.00019999665840492,
-      "loss": 46.0,
-      "step": 34045
-    },
-    {
-      "epoch": 2.6030544564864195,
-      "grad_norm": 0.0015348091255873442,
-      "learning_rate": 0.00019999665820855594,
-      "loss": 46.0,
-      "step": 34046
-    },
-    {
-      "epoch": 2.6031309134698093,
-      "grad_norm": 0.001924305222928524,
-      "learning_rate": 0.0001999966580121861,
-      "loss": 46.0,
-      "step": 34047
-    },
-    {
-      "epoch": 2.6032073704531986,
-      "grad_norm": 0.0012631723657250404,
-      "learning_rate": 0.00019999665781581048,
-      "loss": 46.0,
-      "step": 34048
-    },
-    {
-      "epoch": 2.6032838274365884,
-      "grad_norm": 0.0016728206537663937,
-      "learning_rate": 0.00019999665761942913,
-      "loss": 46.0,
-      "step": 34049
-    },
-    {
-      "epoch": 2.603360284419978,
-      "grad_norm": 0.00212831050157547,
-      "learning_rate": 0.00019999665742304195,
-      "loss": 46.0,
-      "step": 34050
-    },
-    {
-      "epoch": 2.603436741403368,
-      "grad_norm": 0.0016493863658979535,
-      "learning_rate": 0.00019999665722664905,
-      "loss": 46.0,
-      "step": 34051
-    },
-    {
-      "epoch": 2.6035131983867577,
-      "grad_norm": 0.0008678300655446947,
-      "learning_rate": 0.00019999665703025036,
-      "loss": 46.0,
-      "step": 34052
-    },
-    {
-      "epoch": 2.6035896553701474,
-      "grad_norm": 0.0013137904461473227,
-      "learning_rate": 0.00019999665683384589,
-      "loss": 46.0,
-      "step": 34053
-    },
-    {
-      "epoch": 2.603666112353537,
-      "grad_norm": 0.0011362849036231637,
-      "learning_rate": 0.00019999665663743567,
-      "loss": 46.0,
-      "step": 34054
-    },
-    {
-      "epoch": 2.603742569336927,
-      "grad_norm": 0.004044428933411837,
-      "learning_rate": 0.00019999665644101968,
-      "loss": 46.0,
-      "step": 34055
-    },
-    {
-      "epoch": 2.6038190263203163,
-      "grad_norm": 0.0015811502235010266,
-      "learning_rate": 0.00019999665624459794,
-      "loss": 46.0,
-      "step": 34056
-    },
-    {
-      "epoch": 2.603895483303706,
-      "grad_norm": 0.0016064064111560583,
-      "learning_rate": 0.0001999966560481704,
-      "loss": 46.0,
-      "step": 34057
-    },
-    {
-      "epoch": 2.6039719402870958,
-      "grad_norm": 0.0035136446822434664,
-      "learning_rate": 0.00019999665585173707,
-      "loss": 46.0,
-      "step": 34058
-    },
-    {
-      "epoch": 2.6040483972704855,
-      "grad_norm": 0.0018162113847211003,
-      "learning_rate": 0.000199996655655298,
-      "loss": 46.0,
-      "step": 34059
-    },
-    {
-      "epoch": 2.6041248542538753,
-      "grad_norm": 0.00223892112262547,
-      "learning_rate": 0.00019999665545885318,
-      "loss": 46.0,
-      "step": 34060
-    },
-    {
-      "epoch": 2.604201311237265,
-      "grad_norm": 0.0017682722536846995,
-      "learning_rate": 0.00019999665526240255,
-      "loss": 46.0,
-      "step": 34061
-    },
-    {
-      "epoch": 2.604277768220655,
-      "grad_norm": 0.0018040291033685207,
-      "learning_rate": 0.00019999665506594618,
-      "loss": 46.0,
-      "step": 34062
-    },
-    {
-      "epoch": 2.6043542252040446,
-      "grad_norm": 0.0020374213345348835,
-      "learning_rate": 0.00019999665486948405,
-      "loss": 46.0,
-      "step": 34063
-    },
-    {
-      "epoch": 2.6044306821874343,
-      "grad_norm": 0.007367412094026804,
-      "learning_rate": 0.00019999665467301613,
-      "loss": 46.0,
-      "step": 34064
-    },
-    {
-      "epoch": 2.604507139170824,
-      "grad_norm": 0.0012284088879823685,
-      "learning_rate": 0.00019999665447654246,
-      "loss": 46.0,
-      "step": 34065
-    },
-    {
-      "epoch": 2.604583596154214,
-      "grad_norm": 0.001375415246002376,
-      "learning_rate": 0.000199996654280063,
-      "loss": 46.0,
-      "step": 34066
-    },
-    {
-      "epoch": 2.6046600531376036,
-      "grad_norm": 0.0028562911320477724,
-      "learning_rate": 0.00019999665408357777,
-      "loss": 46.0,
-      "step": 34067
-    },
-    {
-      "epoch": 2.6047365101209934,
-      "grad_norm": 0.0009460800793021917,
-      "learning_rate": 0.00019999665388708678,
-      "loss": 46.0,
-      "step": 34068
-    },
-    {
-      "epoch": 2.604812967104383,
-      "grad_norm": 0.0012787526939064264,
-      "learning_rate": 0.00019999665369059,
-      "loss": 46.0,
-      "step": 34069
-    },
-    {
-      "epoch": 2.6048894240877725,
-      "grad_norm": 0.0011006894055753946,
-      "learning_rate": 0.00019999665349408748,
-      "loss": 46.0,
-      "step": 34070
-    },
-    {
-      "epoch": 2.6049658810711622,
-      "grad_norm": 0.0007063025841489434,
-      "learning_rate": 0.00019999665329757917,
-      "loss": 46.0,
-      "step": 34071
-    },
-    {
-      "epoch": 2.605042338054552,
-      "grad_norm": 0.0017444679979234934,
-      "learning_rate": 0.0001999966531010651,
-      "loss": 46.0,
-      "step": 34072
-    },
-    {
-      "epoch": 2.6051187950379417,
-      "grad_norm": 0.00123389414511621,
-      "learning_rate": 0.00019999665290454526,
-      "loss": 46.0,
-      "step": 34073
-    },
-    {
-      "epoch": 2.6051952520213315,
-      "grad_norm": 0.004127791617065668,
-      "learning_rate": 0.00019999665270801966,
-      "loss": 46.0,
-      "step": 34074
-    },
-    {
-      "epoch": 2.6052717090047213,
-      "grad_norm": 0.0009403133881278336,
-      "learning_rate": 0.00019999665251148826,
-      "loss": 46.0,
-      "step": 34075
-    },
-    {
-      "epoch": 2.605348165988111,
-      "grad_norm": 0.002468319144099951,
-      "learning_rate": 0.00019999665231495113,
-      "loss": 46.0,
-      "step": 34076
-    },
-    {
-      "epoch": 2.605424622971501,
-      "grad_norm": 0.001703559304587543,
-      "learning_rate": 0.0001999966521184082,
-      "loss": 46.0,
-      "step": 34077
-    },
-    {
-      "epoch": 2.60550107995489,
-      "grad_norm": 0.0014593893429264426,
-      "learning_rate": 0.00019999665192185951,
-      "loss": 46.0,
-      "step": 34078
-    },
-    {
-      "epoch": 2.60557753693828,
-      "grad_norm": 0.002438412746414542,
-      "learning_rate": 0.00019999665172530507,
-      "loss": 46.0,
-      "step": 34079
-    },
-    {
-      "epoch": 2.6056539939216696,
-      "grad_norm": 0.004381807986646891,
-      "learning_rate": 0.00019999665152874486,
-      "loss": 46.0,
-      "step": 34080
-    },
-    {
-      "epoch": 2.6057304509050594,
-      "grad_norm": 0.0005928900209255517,
-      "learning_rate": 0.00019999665133217884,
-      "loss": 46.0,
-      "step": 34081
-    },
-    {
-      "epoch": 2.605806907888449,
-      "grad_norm": 0.0010792040266096592,
-      "learning_rate": 0.00019999665113560708,
-      "loss": 46.0,
-      "step": 34082
-    },
-    {
-      "epoch": 2.605883364871839,
-      "grad_norm": 0.0020960774272680283,
-      "learning_rate": 0.00019999665093902954,
-      "loss": 46.0,
-      "step": 34083
-    },
-    {
-      "epoch": 2.6059598218552287,
-      "grad_norm": 0.0011029322631657124,
-      "learning_rate": 0.00019999665074244626,
-      "loss": 46.0,
-      "step": 34084
-    },
-    {
-      "epoch": 2.6060362788386184,
-      "grad_norm": 0.001569297513924539,
-      "learning_rate": 0.00019999665054585717,
-      "loss": 46.0,
-      "step": 34085
-    },
-    {
-      "epoch": 2.606112735822008,
-      "grad_norm": 0.002657995093613863,
-      "learning_rate": 0.00019999665034926235,
-      "loss": 46.0,
-      "step": 34086
-    },
-    {
-      "epoch": 2.606189192805398,
-      "grad_norm": 0.0014559200499206781,
-      "learning_rate": 0.00019999665015266174,
-      "loss": 46.0,
-      "step": 34087
-    },
-    {
-      "epoch": 2.6062656497887877,
-      "grad_norm": 0.0014072259655222297,
-      "learning_rate": 0.00019999664995605537,
-      "loss": 46.0,
-      "step": 34088
-    },
-    {
-      "epoch": 2.6063421067721775,
-      "grad_norm": 0.0015695304609835148,
-      "learning_rate": 0.00019999664975944322,
-      "loss": 46.0,
-      "step": 34089
-    },
-    {
-      "epoch": 2.6064185637555672,
-      "grad_norm": 0.0016932080034166574,
-      "learning_rate": 0.0001999966495628253,
-      "loss": 46.0,
-      "step": 34090
-    },
-    {
-      "epoch": 2.606495020738957,
-      "grad_norm": 0.0007436703890562057,
-      "learning_rate": 0.0001999966493662016,
-      "loss": 46.0,
-      "step": 34091
-    },
-    {
-      "epoch": 2.6065714777223463,
-      "grad_norm": 0.005224339663982391,
-      "learning_rate": 0.00019999664916957216,
-      "loss": 46.0,
-      "step": 34092
-    },
-    {
-      "epoch": 2.606647934705736,
-      "grad_norm": 0.0010078026680275798,
-      "learning_rate": 0.00019999664897293694,
-      "loss": 46.0,
-      "step": 34093
-    },
-    {
-      "epoch": 2.606724391689126,
-      "grad_norm": 0.0017173726810142398,
-      "learning_rate": 0.00019999664877629592,
-      "loss": 46.0,
-      "step": 34094
-    },
-    {
-      "epoch": 2.6068008486725156,
-      "grad_norm": 0.0013045882806181908,
-      "learning_rate": 0.00019999664857964916,
-      "loss": 46.0,
-      "step": 34095
-    },
-    {
-      "epoch": 2.6068773056559054,
-      "grad_norm": 0.004429432563483715,
-      "learning_rate": 0.00019999664838299663,
-      "loss": 46.0,
-      "step": 34096
-    },
-    {
-      "epoch": 2.606953762639295,
-      "grad_norm": 0.0021900886204093695,
-      "learning_rate": 0.00019999664818633832,
-      "loss": 46.0,
-      "step": 34097
-    },
-    {
-      "epoch": 2.607030219622685,
-      "grad_norm": 0.0018946188502013683,
-      "learning_rate": 0.00019999664798967426,
-      "loss": 46.0,
-      "step": 34098
-    },
-    {
-      "epoch": 2.6071066766060746,
-      "grad_norm": 0.0026240062434226274,
-      "learning_rate": 0.00019999664779300443,
-      "loss": 46.0,
-      "step": 34099
-    },
-    {
-      "epoch": 2.607183133589464,
-      "grad_norm": 0.002428285777568817,
-      "learning_rate": 0.0001999966475963288,
-      "loss": 46.0,
-      "step": 34100
-    },
-    {
-      "epoch": 2.6072595905728537,
-      "grad_norm": 0.0018335528438910842,
-      "learning_rate": 0.00019999664739964743,
-      "loss": 46.0,
-      "step": 34101
-    },
-    {
-      "epoch": 2.6073360475562435,
-      "grad_norm": 0.0008534009684808552,
-      "learning_rate": 0.00019999664720296028,
-      "loss": 46.0,
-      "step": 34102
-    },
-    {
-      "epoch": 2.6074125045396332,
-      "grad_norm": 0.0008628587238490582,
-      "learning_rate": 0.00019999664700626736,
-      "loss": 46.0,
-      "step": 34103
-    },
-    {
-      "epoch": 2.607488961523023,
-      "grad_norm": 0.001418377156369388,
-      "learning_rate": 0.00019999664680956866,
-      "loss": 46.0,
-      "step": 34104
-    },
-    {
-      "epoch": 2.6075654185064128,
-      "grad_norm": 0.004012479446828365,
-      "learning_rate": 0.00019999664661286422,
-      "loss": 46.0,
-      "step": 34105
-    },
-    {
-      "epoch": 2.6076418754898025,
-      "grad_norm": 0.001176545862108469,
-      "learning_rate": 0.00019999664641615398,
-      "loss": 46.0,
-      "step": 34106
-    },
-    {
-      "epoch": 2.6077183324731923,
-      "grad_norm": 0.0008842108072713017,
-      "learning_rate": 0.000199996646219438,
-      "loss": 46.0,
-      "step": 34107
-    },
-    {
-      "epoch": 2.607794789456582,
-      "grad_norm": 0.0022622086107730865,
-      "learning_rate": 0.00019999664602271623,
-      "loss": 46.0,
-      "step": 34108
-    },
-    {
-      "epoch": 2.607871246439972,
-      "grad_norm": 0.0013154157204553485,
-      "learning_rate": 0.0001999966458259887,
-      "loss": 46.0,
-      "step": 34109
-    },
-    {
-      "epoch": 2.6079477034233616,
-      "grad_norm": 0.002860425738617778,
-      "learning_rate": 0.0001999966456292554,
-      "loss": 46.0,
-      "step": 34110
-    },
-    {
-      "epoch": 2.6080241604067513,
-      "grad_norm": 0.0016381718451157212,
-      "learning_rate": 0.00019999664543251633,
-      "loss": 46.0,
-      "step": 34111
-    },
-    {
-      "epoch": 2.608100617390141,
-      "grad_norm": 0.0011745529482141137,
-      "learning_rate": 0.0001999966452357715,
-      "loss": 46.0,
-      "step": 34112
-    },
-    {
-      "epoch": 2.608177074373531,
-      "grad_norm": 0.001570095424540341,
-      "learning_rate": 0.00019999664503902087,
-      "loss": 46.0,
-      "step": 34113
-    },
-    {
-      "epoch": 2.60825353135692,
-      "grad_norm": 0.0018143949564546347,
-      "learning_rate": 0.0001999966448422645,
-      "loss": 46.0,
-      "step": 34114
-    },
-    {
-      "epoch": 2.60832998834031,
-      "grad_norm": 0.0024891297798603773,
-      "learning_rate": 0.00019999664464550234,
-      "loss": 46.0,
-      "step": 34115
-    },
-    {
-      "epoch": 2.6084064453236997,
-      "grad_norm": 0.002164674922823906,
-      "learning_rate": 0.00019999664444873442,
-      "loss": 46.0,
-      "step": 34116
-    },
-    {
-      "epoch": 2.6084829023070895,
-      "grad_norm": 0.0016388414660468698,
-      "learning_rate": 0.00019999664425196075,
-      "loss": 46.0,
-      "step": 34117
-    },
-    {
-      "epoch": 2.608559359290479,
-      "grad_norm": 0.001399126835167408,
-      "learning_rate": 0.00019999664405518129,
-      "loss": 46.0,
-      "step": 34118
-    },
-    {
-      "epoch": 2.608635816273869,
-      "grad_norm": 0.00407001469284296,
-      "learning_rate": 0.00019999664385839604,
-      "loss": 46.0,
-      "step": 34119
-    },
-    {
-      "epoch": 2.6087122732572587,
-      "grad_norm": 0.0023797585163265467,
-      "learning_rate": 0.00019999664366160505,
-      "loss": 46.0,
-      "step": 34120
-    },
-    {
-      "epoch": 2.6087887302406485,
-      "grad_norm": 0.002845124341547489,
-      "learning_rate": 0.00019999664346480832,
-      "loss": 46.0,
-      "step": 34121
-    },
-    {
-      "epoch": 2.608865187224038,
-      "grad_norm": 0.0032964125275611877,
-      "learning_rate": 0.00019999664326800576,
-      "loss": 46.0,
-      "step": 34122
-    },
-    {
-      "epoch": 2.6089416442074276,
-      "grad_norm": 0.0012169118272140622,
-      "learning_rate": 0.00019999664307119745,
-      "loss": 46.0,
-      "step": 34123
-    },
-    {
-      "epoch": 2.6090181011908173,
-      "grad_norm": 0.0029273962136358023,
-      "learning_rate": 0.0001999966428743834,
-      "loss": 46.0,
-      "step": 34124
-    },
-    {
-      "epoch": 2.609094558174207,
-      "grad_norm": 0.0015583352651447058,
-      "learning_rate": 0.00019999664267756354,
-      "loss": 46.0,
-      "step": 34125
-    },
-    {
-      "epoch": 2.609171015157597,
-      "grad_norm": 0.000859149091411382,
-      "learning_rate": 0.00019999664248073794,
-      "loss": 46.0,
-      "step": 34126
-    },
-    {
-      "epoch": 2.6092474721409866,
-      "grad_norm": 0.0019579019863158464,
-      "learning_rate": 0.00019999664228390656,
-      "loss": 46.0,
-      "step": 34127
-    },
-    {
-      "epoch": 2.6093239291243764,
-      "grad_norm": 0.002120393794029951,
-      "learning_rate": 0.00019999664208706941,
-      "loss": 46.0,
-      "step": 34128
-    },
-    {
-      "epoch": 2.609400386107766,
-      "grad_norm": 0.002389185829088092,
-      "learning_rate": 0.0001999966418902265,
-      "loss": 46.0,
-      "step": 34129
-    },
-    {
-      "epoch": 2.609476843091156,
-      "grad_norm": 0.003060850314795971,
-      "learning_rate": 0.0001999966416933778,
-      "loss": 46.0,
-      "step": 34130
-    },
-    {
-      "epoch": 2.6095533000745457,
-      "grad_norm": 0.000556423154193908,
-      "learning_rate": 0.00019999664149652336,
-      "loss": 46.0,
-      "step": 34131
-    },
-    {
-      "epoch": 2.6096297570579354,
-      "grad_norm": 0.001494173426181078,
-      "learning_rate": 0.00019999664129966311,
-      "loss": 46.0,
-      "step": 34132
-    },
-    {
-      "epoch": 2.609706214041325,
-      "grad_norm": 0.0013685955200344324,
-      "learning_rate": 0.00019999664110279713,
-      "loss": 46.0,
-      "step": 34133
-    },
-    {
-      "epoch": 2.609782671024715,
-      "grad_norm": 0.0028649617452174425,
-      "learning_rate": 0.00019999664090592536,
-      "loss": 46.0,
-      "step": 34134
-    },
-    {
-      "epoch": 2.6098591280081047,
-      "grad_norm": 0.0011249084491282701,
-      "learning_rate": 0.00019999664070904783,
-      "loss": 46.0,
-      "step": 34135
-    },
-    {
-      "epoch": 2.609935584991494,
-      "grad_norm": 0.0007842201739549637,
-      "learning_rate": 0.00019999664051216452,
-      "loss": 46.0,
-      "step": 34136
-    },
-    {
-      "epoch": 2.610012041974884,
-      "grad_norm": 0.01031413022428751,
-      "learning_rate": 0.00019999664031527544,
-      "loss": 46.0,
-      "step": 34137
-    },
-    {
-      "epoch": 2.6100884989582736,
-      "grad_norm": 0.001108855358324945,
-      "learning_rate": 0.00019999664011838058,
-      "loss": 46.0,
-      "step": 34138
-    },
-    {
-      "epoch": 2.6101649559416633,
-      "grad_norm": 0.0007446836098097265,
-      "learning_rate": 0.00019999663992147998,
-      "loss": 46.0,
-      "step": 34139
-    },
-    {
-      "epoch": 2.610241412925053,
-      "grad_norm": 0.001520541263744235,
-      "learning_rate": 0.0001999966397245736,
-      "loss": 46.0,
-      "step": 34140
-    },
-    {
-      "epoch": 2.610317869908443,
-      "grad_norm": 0.0014953704085201025,
-      "learning_rate": 0.0001999966395276615,
-      "loss": 46.0,
-      "step": 34141
-    },
-    {
-      "epoch": 2.6103943268918326,
-      "grad_norm": 0.0013079511700198054,
-      "learning_rate": 0.00019999663933074354,
-      "loss": 46.0,
-      "step": 34142
-    },
-    {
-      "epoch": 2.6104707838752224,
-      "grad_norm": 0.0025454566348344088,
-      "learning_rate": 0.00019999663913381984,
-      "loss": 46.0,
-      "step": 34143
-    },
-    {
-      "epoch": 2.6105472408586117,
-      "grad_norm": 0.003493966767564416,
-      "learning_rate": 0.0001999966389368904,
-      "loss": 46.0,
-      "step": 34144
-    },
-    {
-      "epoch": 2.6106236978420014,
-      "grad_norm": 0.007100183051079512,
-      "learning_rate": 0.0001999966387399552,
-      "loss": 46.0,
-      "step": 34145
-    },
-    {
-      "epoch": 2.610700154825391,
-      "grad_norm": 0.000790994381532073,
-      "learning_rate": 0.00019999663854301417,
-      "loss": 46.0,
-      "step": 34146
-    },
-    {
-      "epoch": 2.610776611808781,
-      "grad_norm": 0.0023144404403865337,
-      "learning_rate": 0.0001999966383460674,
-      "loss": 46.0,
-      "step": 34147
-    },
-    {
-      "epoch": 2.6108530687921707,
-      "grad_norm": 0.0013884566724300385,
-      "learning_rate": 0.00019999663814911488,
-      "loss": 46.0,
-      "step": 34148
-    },
-    {
-      "epoch": 2.6109295257755605,
-      "grad_norm": 0.001954339211806655,
-      "learning_rate": 0.00019999663795215657,
-      "loss": 46.0,
-      "step": 34149
-    },
-    {
-      "epoch": 2.6110059827589502,
-      "grad_norm": 0.0010637211380526423,
-      "learning_rate": 0.00019999663775519252,
-      "loss": 46.0,
-      "step": 34150
-    },
-    {
-      "epoch": 2.61108243974234,
-      "grad_norm": 0.0017372184665873647,
-      "learning_rate": 0.00019999663755822263,
-      "loss": 46.0,
-      "step": 34151
-    },
-    {
-      "epoch": 2.6111588967257298,
-      "grad_norm": 0.0013366000493988395,
-      "learning_rate": 0.00019999663736124706,
-      "loss": 46.0,
-      "step": 34152
-    },
-    {
-      "epoch": 2.6112353537091195,
-      "grad_norm": 0.0007023279322311282,
-      "learning_rate": 0.00019999663716426566,
-      "loss": 46.0,
-      "step": 34153
-    },
-    {
-      "epoch": 2.6113118106925093,
-      "grad_norm": 0.004130127374082804,
-      "learning_rate": 0.0001999966369672785,
-      "loss": 46.0,
-      "step": 34154
-    },
-    {
-      "epoch": 2.611388267675899,
-      "grad_norm": 0.000591421325225383,
-      "learning_rate": 0.0001999966367702856,
-      "loss": 46.0,
-      "step": 34155
-    },
-    {
-      "epoch": 2.611464724659289,
-      "grad_norm": 0.0014904154231771827,
-      "learning_rate": 0.00019999663657328692,
-      "loss": 46.0,
-      "step": 34156
-    },
-    {
-      "epoch": 2.611541181642678,
-      "grad_norm": 0.0037304868455976248,
-      "learning_rate": 0.00019999663637628245,
-      "loss": 46.0,
-      "step": 34157
-    },
-    {
-      "epoch": 2.611617638626068,
-      "grad_norm": 0.003107240190729499,
-      "learning_rate": 0.0001999966361792722,
-      "loss": 46.0,
-      "step": 34158
-    },
-    {
-      "epoch": 2.6116940956094576,
-      "grad_norm": 0.00446449825540185,
-      "learning_rate": 0.00019999663598225623,
-      "loss": 46.0,
-      "step": 34159
-    },
-    {
-      "epoch": 2.6117705525928474,
-      "grad_norm": 0.0037031034007668495,
-      "learning_rate": 0.00019999663578523446,
-      "loss": 46.0,
-      "step": 34160
-    },
-    {
-      "epoch": 2.611847009576237,
-      "grad_norm": 0.0011913626221939921,
-      "learning_rate": 0.00019999663558820693,
-      "loss": 46.0,
-      "step": 34161
-    },
-    {
-      "epoch": 2.611923466559627,
-      "grad_norm": 0.0018605723744258285,
-      "learning_rate": 0.00019999663539117362,
-      "loss": 46.0,
-      "step": 34162
-    },
-    {
-      "epoch": 2.6119999235430167,
-      "grad_norm": 0.00363340531475842,
-      "learning_rate": 0.00019999663519413457,
-      "loss": 46.0,
-      "step": 34163
-    },
-    {
-      "epoch": 2.6120763805264064,
-      "grad_norm": 0.0017527599120512605,
-      "learning_rate": 0.0001999966349970897,
-      "loss": 46.0,
-      "step": 34164
-    },
-    {
-      "epoch": 2.612152837509796,
-      "grad_norm": 0.00210291869007051,
-      "learning_rate": 0.0001999966348000391,
-      "loss": 46.0,
-      "step": 34165
-    },
-    {
-      "epoch": 2.6122292944931855,
-      "grad_norm": 0.002765426877886057,
-      "learning_rate": 0.0001999966346029827,
-      "loss": 46.0,
-      "step": 34166
-    },
-    {
-      "epoch": 2.6123057514765753,
-      "grad_norm": 0.001917311572469771,
-      "learning_rate": 0.00019999663440592057,
-      "loss": 46.0,
-      "step": 34167
-    },
-    {
-      "epoch": 2.612382208459965,
-      "grad_norm": 0.0011326767271384597,
-      "learning_rate": 0.00019999663420885264,
-      "loss": 46.0,
-      "step": 34168
-    },
-    {
-      "epoch": 2.612458665443355,
-      "grad_norm": 0.0016144639812409878,
-      "learning_rate": 0.00019999663401177895,
-      "loss": 46.0,
-      "step": 34169
-    },
-    {
-      "epoch": 2.6125351224267446,
-      "grad_norm": 0.0007132149185054004,
-      "learning_rate": 0.0001999966338146995,
-      "loss": 46.0,
-      "step": 34170
-    },
-    {
-      "epoch": 2.6126115794101343,
-      "grad_norm": 0.0034199159126728773,
-      "learning_rate": 0.00019999663361761427,
-      "loss": 46.0,
-      "step": 34171
-    },
-    {
-      "epoch": 2.612688036393524,
-      "grad_norm": 0.00129833840765059,
-      "learning_rate": 0.00019999663342052325,
-      "loss": 46.0,
-      "step": 34172
-    },
-    {
-      "epoch": 2.612764493376914,
-      "grad_norm": 0.0021075336262583733,
-      "learning_rate": 0.00019999663322342652,
-      "loss": 46.0,
-      "step": 34173
-    },
-    {
-      "epoch": 2.6128409503603036,
-      "grad_norm": 0.004253344144672155,
-      "learning_rate": 0.00019999663302632396,
-      "loss": 46.0,
-      "step": 34174
-    },
-    {
-      "epoch": 2.6129174073436934,
-      "grad_norm": 0.0020897623617202044,
-      "learning_rate": 0.00019999663282921568,
-      "loss": 46.0,
-      "step": 34175
-    },
-    {
-      "epoch": 2.612993864327083,
-      "grad_norm": 0.0017245084745809436,
-      "learning_rate": 0.00019999663263210157,
-      "loss": 46.0,
-      "step": 34176
-    },
-    {
-      "epoch": 2.613070321310473,
-      "grad_norm": 0.0014214001130312681,
-      "learning_rate": 0.00019999663243498175,
-      "loss": 46.0,
-      "step": 34177
-    },
-    {
-      "epoch": 2.6131467782938627,
-      "grad_norm": 0.001643747091293335,
-      "learning_rate": 0.00019999663223785614,
-      "loss": 46.0,
-      "step": 34178
-    },
-    {
-      "epoch": 2.613223235277252,
-      "grad_norm": 0.0024216067977249622,
-      "learning_rate": 0.00019999663204072477,
-      "loss": 46.0,
-      "step": 34179
-    },
-    {
-      "epoch": 2.6132996922606417,
-      "grad_norm": 0.0014718774473294616,
-      "learning_rate": 0.00019999663184358762,
-      "loss": 46.0,
-      "step": 34180
-    },
-    {
-      "epoch": 2.6133761492440315,
-      "grad_norm": 0.0007778847357258201,
-      "learning_rate": 0.00019999663164644468,
-      "loss": 46.0,
-      "step": 34181
-    },
-    {
-      "epoch": 2.6134526062274213,
-      "grad_norm": 0.0027990483213216066,
-      "learning_rate": 0.00019999663144929598,
-      "loss": 46.0,
-      "step": 34182
-    },
-    {
-      "epoch": 2.613529063210811,
-      "grad_norm": 0.0014099058462306857,
-      "learning_rate": 0.00019999663125214152,
-      "loss": 46.0,
-      "step": 34183
-    },
-    {
-      "epoch": 2.613605520194201,
-      "grad_norm": 0.0008423132821917534,
-      "learning_rate": 0.00019999663105498133,
-      "loss": 46.0,
-      "step": 34184
-    },
-    {
-      "epoch": 2.6136819771775905,
-      "grad_norm": 0.0017143214354291558,
-      "learning_rate": 0.00019999663085781531,
-      "loss": 46.0,
-      "step": 34185
-    },
-    {
-      "epoch": 2.6137584341609803,
-      "grad_norm": 0.001519199926406145,
-      "learning_rate": 0.00019999663066064355,
-      "loss": 46.0,
-      "step": 34186
-    },
-    {
-      "epoch": 2.61383489114437,
-      "grad_norm": 0.00189464190043509,
-      "learning_rate": 0.000199996630463466,
-      "loss": 46.0,
-      "step": 34187
-    },
-    {
-      "epoch": 2.6139113481277594,
-      "grad_norm": 0.0021376097574830055,
-      "learning_rate": 0.00019999663026628271,
-      "loss": 46.0,
-      "step": 34188
-    },
-    {
-      "epoch": 2.613987805111149,
-      "grad_norm": 0.0017821603687480092,
-      "learning_rate": 0.00019999663006909363,
-      "loss": 46.0,
-      "step": 34189
-    },
-    {
-      "epoch": 2.614064262094539,
-      "grad_norm": 0.005011957138776779,
-      "learning_rate": 0.0001999966298718988,
-      "loss": 46.0,
-      "step": 34190
-    },
-    {
-      "epoch": 2.6141407190779287,
-      "grad_norm": 0.0014494098722934723,
-      "learning_rate": 0.00019999662967469818,
-      "loss": 46.0,
-      "step": 34191
-    },
-    {
-      "epoch": 2.6142171760613184,
-      "grad_norm": 0.0032881658989936113,
-      "learning_rate": 0.0001999966294774918,
-      "loss": 46.0,
-      "step": 34192
-    },
-    {
-      "epoch": 2.614293633044708,
-      "grad_norm": 0.0013622696278616786,
-      "learning_rate": 0.00019999662928027963,
-      "loss": 46.0,
-      "step": 34193
-    },
-    {
-      "epoch": 2.614370090028098,
-      "grad_norm": 0.0026153349317610264,
-      "learning_rate": 0.0001999966290830617,
-      "loss": 46.0,
-      "step": 34194
-    },
-    {
-      "epoch": 2.6144465470114877,
-      "grad_norm": 0.001099519431591034,
-      "learning_rate": 0.00019999662888583805,
-      "loss": 46.0,
-      "step": 34195
-    },
-    {
-      "epoch": 2.6145230039948775,
-      "grad_norm": 0.002636526944115758,
-      "learning_rate": 0.00019999662868860858,
-      "loss": 46.0,
-      "step": 34196
-    },
-    {
-      "epoch": 2.6145994609782672,
-      "grad_norm": 0.0010001073824241757,
-      "learning_rate": 0.00019999662849137334,
-      "loss": 46.0,
-      "step": 34197
-    },
-    {
-      "epoch": 2.614675917961657,
-      "grad_norm": 0.0005777434562332928,
-      "learning_rate": 0.00019999662829413235,
-      "loss": 46.0,
-      "step": 34198
-    },
-    {
-      "epoch": 2.6147523749450468,
-      "grad_norm": 0.001196935772895813,
-      "learning_rate": 0.0001999966280968856,
-      "loss": 46.0,
-      "step": 34199
-    },
-    {
-      "epoch": 2.6148288319284365,
-      "grad_norm": 0.0016990507720038295,
-      "learning_rate": 0.00019999662789963306,
-      "loss": 46.0,
-      "step": 34200
-    },
-    {
-      "epoch": 2.614905288911826,
-      "grad_norm": 0.00142157101072371,
-      "learning_rate": 0.00019999662770237473,
-      "loss": 46.0,
-      "step": 34201
-    },
-    {
-      "epoch": 2.6149817458952156,
-      "grad_norm": 0.0005694128340110183,
-      "learning_rate": 0.00019999662750511065,
-      "loss": 46.0,
-      "step": 34202
-    },
-    {
-      "epoch": 2.6150582028786054,
-      "grad_norm": 0.0015122366603463888,
-      "learning_rate": 0.00019999662730784082,
-      "loss": 46.0,
-      "step": 34203
-    },
-    {
-      "epoch": 2.615134659861995,
-      "grad_norm": 0.0010198099771514535,
-      "learning_rate": 0.0001999966271105652,
-      "loss": 46.0,
-      "step": 34204
-    },
-    {
-      "epoch": 2.615211116845385,
-      "grad_norm": 0.0018776233773678541,
-      "learning_rate": 0.00019999662691328382,
-      "loss": 46.0,
-      "step": 34205
-    },
-    {
-      "epoch": 2.6152875738287746,
-      "grad_norm": 0.0016221227124333382,
-      "learning_rate": 0.00019999662671599667,
-      "loss": 46.0,
-      "step": 34206
-    },
-    {
-      "epoch": 2.6153640308121644,
-      "grad_norm": 0.0007513632299378514,
-      "learning_rate": 0.00019999662651870375,
-      "loss": 46.0,
-      "step": 34207
-    },
-    {
-      "epoch": 2.615440487795554,
-      "grad_norm": 0.004027173854410648,
-      "learning_rate": 0.00019999662632140509,
-      "loss": 46.0,
-      "step": 34208
-    },
-    {
-      "epoch": 2.6155169447789435,
-      "grad_norm": 0.0006107151857577264,
-      "learning_rate": 0.0001999966261241006,
-      "loss": 46.0,
-      "step": 34209
-    },
-    {
-      "epoch": 2.6155934017623332,
-      "grad_norm": 0.0035889095161110163,
-      "learning_rate": 0.00019999662592679038,
-      "loss": 46.0,
-      "step": 34210
-    },
-    {
-      "epoch": 2.615669858745723,
-      "grad_norm": 0.0011578392004594207,
-      "learning_rate": 0.00019999662572947437,
-      "loss": 46.0,
-      "step": 34211
-    },
-    {
-      "epoch": 2.6157463157291128,
-      "grad_norm": 0.0013914982555434108,
-      "learning_rate": 0.0001999966255321526,
-      "loss": 46.0,
-      "step": 34212
-    },
-    {
-      "epoch": 2.6158227727125025,
-      "grad_norm": 0.0019806597847491503,
-      "learning_rate": 0.00019999662533482507,
-      "loss": 46.0,
-      "step": 34213
-    },
-    {
-      "epoch": 2.6158992296958923,
-      "grad_norm": 0.0034138422925025225,
-      "learning_rate": 0.00019999662513749177,
-      "loss": 46.0,
-      "step": 34214
-    },
-    {
-      "epoch": 2.615975686679282,
-      "grad_norm": 0.003941693343222141,
-      "learning_rate": 0.0001999966249401527,
-      "loss": 46.0,
-      "step": 34215
-    },
-    {
-      "epoch": 2.616052143662672,
-      "grad_norm": 0.002259183442220092,
-      "learning_rate": 0.00019999662474280783,
-      "loss": 46.0,
-      "step": 34216
-    },
-    {
-      "epoch": 2.6161286006460616,
-      "grad_norm": 0.0019805864430963993,
-      "learning_rate": 0.00019999662454545724,
-      "loss": 46.0,
-      "step": 34217
-    },
-    {
-      "epoch": 2.6162050576294513,
-      "grad_norm": 0.0004943533567711711,
-      "learning_rate": 0.00019999662434810084,
-      "loss": 46.0,
-      "step": 34218
-    },
-    {
-      "epoch": 2.616281514612841,
-      "grad_norm": 0.0006937652360647917,
-      "learning_rate": 0.0001999966241507387,
-      "loss": 46.0,
-      "step": 34219
-    },
-    {
-      "epoch": 2.616357971596231,
-      "grad_norm": 0.002335480647161603,
-      "learning_rate": 0.00019999662395337077,
-      "loss": 46.0,
-      "step": 34220
-    },
-    {
-      "epoch": 2.6164344285796206,
-      "grad_norm": 0.0006180853815749288,
-      "learning_rate": 0.0001999966237559971,
-      "loss": 46.0,
-      "step": 34221
-    },
-    {
-      "epoch": 2.6165108855630104,
-      "grad_norm": 0.0008049941970966756,
-      "learning_rate": 0.00019999662355861764,
-      "loss": 46.0,
-      "step": 34222
-    },
-    {
-      "epoch": 2.6165873425463997,
-      "grad_norm": 0.003091156482696533,
-      "learning_rate": 0.0001999966233612324,
-      "loss": 46.0,
-      "step": 34223
-    },
-    {
-      "epoch": 2.6166637995297894,
-      "grad_norm": 0.008240773342549801,
-      "learning_rate": 0.0001999966231638414,
-      "loss": 46.0,
-      "step": 34224
-    },
-    {
-      "epoch": 2.616740256513179,
-      "grad_norm": 0.0018525755731388927,
-      "learning_rate": 0.00019999662296644463,
-      "loss": 46.0,
-      "step": 34225
-    },
-    {
-      "epoch": 2.616816713496569,
-      "grad_norm": 0.0033004239667207003,
-      "learning_rate": 0.0001999966227690421,
-      "loss": 46.0,
-      "step": 34226
-    },
-    {
-      "epoch": 2.6168931704799587,
-      "grad_norm": 0.0016091905999928713,
-      "learning_rate": 0.0001999966225716338,
-      "loss": 46.0,
-      "step": 34227
-    },
-    {
-      "epoch": 2.6169696274633485,
-      "grad_norm": 0.001077565597370267,
-      "learning_rate": 0.00019999662237421973,
-      "loss": 46.0,
-      "step": 34228
-    },
-    {
-      "epoch": 2.6170460844467383,
-      "grad_norm": 0.0013370043598115444,
-      "learning_rate": 0.00019999662217679988,
-      "loss": 46.0,
-      "step": 34229
-    },
-    {
-      "epoch": 2.617122541430128,
-      "grad_norm": 0.007584643084555864,
-      "learning_rate": 0.00019999662197937426,
-      "loss": 46.0,
-      "step": 34230
-    },
-    {
-      "epoch": 2.6171989984135173,
-      "grad_norm": 0.0016999304061755538,
-      "learning_rate": 0.00019999662178194288,
-      "loss": 46.0,
-      "step": 34231
-    },
-    {
-      "epoch": 2.617275455396907,
-      "grad_norm": 0.0012719600927084684,
-      "learning_rate": 0.0001999966215845057,
-      "loss": 46.0,
-      "step": 34232
-    },
-    {
-      "epoch": 2.617351912380297,
-      "grad_norm": 0.0020562242716550827,
-      "learning_rate": 0.00019999662138706282,
-      "loss": 46.0,
-      "step": 34233
-    },
-    {
-      "epoch": 2.6174283693636866,
-      "grad_norm": 0.004436023533344269,
-      "learning_rate": 0.0001999966211896141,
-      "loss": 46.0,
-      "step": 34234
-    },
-    {
-      "epoch": 2.6175048263470764,
-      "grad_norm": 0.0015971716493368149,
-      "learning_rate": 0.00019999662099215966,
-      "loss": 46.0,
-      "step": 34235
-    },
-    {
-      "epoch": 2.617581283330466,
-      "grad_norm": 0.0008590783108957112,
-      "learning_rate": 0.00019999662079469942,
-      "loss": 46.0,
-      "step": 34236
-    },
-    {
-      "epoch": 2.617657740313856,
-      "grad_norm": 0.0024100837763398886,
-      "learning_rate": 0.0001999966205972334,
-      "loss": 46.0,
-      "step": 34237
-    },
-    {
-      "epoch": 2.6177341972972457,
-      "grad_norm": 0.001463544089347124,
-      "learning_rate": 0.00019999662039976165,
-      "loss": 46.0,
-      "step": 34238
-    },
-    {
-      "epoch": 2.6178106542806354,
-      "grad_norm": 0.0020168256014585495,
-      "learning_rate": 0.00019999662020228414,
-      "loss": 46.0,
-      "step": 34239
-    },
-    {
-      "epoch": 2.617887111264025,
-      "grad_norm": 0.0016613135812804103,
-      "learning_rate": 0.0001999966200048008,
-      "loss": 46.0,
-      "step": 34240
-    },
-    {
-      "epoch": 2.617963568247415,
-      "grad_norm": 0.001150017138570547,
-      "learning_rate": 0.00019999661980731173,
-      "loss": 46.0,
-      "step": 34241
-    },
-    {
-      "epoch": 2.6180400252308047,
-      "grad_norm": 0.0012975874124094844,
-      "learning_rate": 0.0001999966196098169,
-      "loss": 46.0,
-      "step": 34242
-    },
-    {
-      "epoch": 2.6181164822141945,
-      "grad_norm": 0.0014641644665971398,
-      "learning_rate": 0.00019999661941231628,
-      "loss": 46.0,
-      "step": 34243
-    },
-    {
-      "epoch": 2.6181929391975842,
-      "grad_norm": 0.0006540699396282434,
-      "learning_rate": 0.0001999966192148099,
-      "loss": 46.0,
-      "step": 34244
-    },
-    {
-      "epoch": 2.6182693961809735,
-      "grad_norm": 0.0019124718382954597,
-      "learning_rate": 0.00019999661901729773,
-      "loss": 46.0,
-      "step": 34245
-    },
-    {
-      "epoch": 2.6183458531643633,
-      "grad_norm": 0.002147477585822344,
-      "learning_rate": 0.00019999661881977982,
-      "loss": 46.0,
-      "step": 34246
-    },
-    {
-      "epoch": 2.618422310147753,
-      "grad_norm": 0.0020005377009510994,
-      "learning_rate": 0.00019999661862225612,
-      "loss": 46.0,
-      "step": 34247
-    },
-    {
-      "epoch": 2.618498767131143,
-      "grad_norm": 0.001586235361173749,
-      "learning_rate": 0.00019999661842472666,
-      "loss": 46.0,
-      "step": 34248
-    },
-    {
-      "epoch": 2.6185752241145326,
-      "grad_norm": 0.0025970041751861572,
-      "learning_rate": 0.00019999661822719142,
-      "loss": 46.0,
-      "step": 34249
-    },
-    {
-      "epoch": 2.6186516810979223,
-      "grad_norm": 0.00046401366125792265,
-      "learning_rate": 0.00019999661802965043,
-      "loss": 46.0,
-      "step": 34250
-    },
-    {
-      "epoch": 2.618728138081312,
-      "grad_norm": 0.0013988347491249442,
-      "learning_rate": 0.00019999661783210368,
-      "loss": 46.0,
-      "step": 34251
-    },
-    {
-      "epoch": 2.618804595064702,
-      "grad_norm": 0.0010600073728710413,
-      "learning_rate": 0.00019999661763455114,
-      "loss": 46.0,
-      "step": 34252
-    },
-    {
-      "epoch": 2.618881052048091,
-      "grad_norm": 0.003847547108307481,
-      "learning_rate": 0.00019999661743699284,
-      "loss": 46.0,
-      "step": 34253
-    },
-    {
-      "epoch": 2.618957509031481,
-      "grad_norm": 0.0008074132492765784,
-      "learning_rate": 0.00019999661723942876,
-      "loss": 46.0,
-      "step": 34254
-    },
-    {
-      "epoch": 2.6190339660148707,
-      "grad_norm": 0.0013632638147100806,
-      "learning_rate": 0.0001999966170418589,
-      "loss": 46.0,
-      "step": 34255
-    },
-    {
-      "epoch": 2.6191104229982605,
-      "grad_norm": 0.0018430332420393825,
-      "learning_rate": 0.00019999661684428328,
-      "loss": 46.0,
-      "step": 34256
-    },
-    {
-      "epoch": 2.6191868799816502,
-      "grad_norm": 0.0004163621924817562,
-      "learning_rate": 0.0001999966166467019,
-      "loss": 46.0,
-      "step": 34257
-    },
-    {
-      "epoch": 2.61926333696504,
-      "grad_norm": 0.0022576674818992615,
-      "learning_rate": 0.00019999661644911476,
-      "loss": 46.0,
-      "step": 34258
-    },
-    {
-      "epoch": 2.6193397939484298,
-      "grad_norm": 0.002548081800341606,
-      "learning_rate": 0.00019999661625152185,
-      "loss": 46.0,
-      "step": 34259
-    },
-    {
-      "epoch": 2.6194162509318195,
-      "grad_norm": 0.0010557289933785796,
-      "learning_rate": 0.00019999661605392315,
-      "loss": 46.0,
-      "step": 34260
-    },
-    {
-      "epoch": 2.6194927079152093,
-      "grad_norm": 0.0024785371497273445,
-      "learning_rate": 0.0001999966158563187,
-      "loss": 46.0,
-      "step": 34261
-    },
-    {
-      "epoch": 2.619569164898599,
-      "grad_norm": 0.0010202809935435653,
-      "learning_rate": 0.00019999661565870845,
-      "loss": 46.0,
-      "step": 34262
-    },
-    {
-      "epoch": 2.619645621881989,
-      "grad_norm": 0.004763559903949499,
-      "learning_rate": 0.00019999661546109246,
-      "loss": 46.0,
-      "step": 34263
-    },
-    {
-      "epoch": 2.6197220788653786,
-      "grad_norm": 0.0013072541914880276,
-      "learning_rate": 0.00019999661526347068,
-      "loss": 46.0,
-      "step": 34264
-    },
-    {
-      "epoch": 2.6197985358487683,
-      "grad_norm": 0.0020412427838891745,
-      "learning_rate": 0.00019999661506584312,
-      "loss": 46.0,
-      "step": 34265
-    },
-    {
-      "epoch": 2.619874992832158,
-      "grad_norm": 0.0008841636590659618,
-      "learning_rate": 0.00019999661486820984,
-      "loss": 46.0,
-      "step": 34266
-    },
-    {
-      "epoch": 2.6199514498155474,
-      "grad_norm": 0.0012474969262257218,
-      "learning_rate": 0.00019999661467057076,
-      "loss": 46.0,
-      "step": 34267
-    },
-    {
-      "epoch": 2.620027906798937,
-      "grad_norm": 0.002453595632687211,
-      "learning_rate": 0.0001999966144729259,
-      "loss": 46.0,
-      "step": 34268
-    },
-    {
-      "epoch": 2.620104363782327,
-      "grad_norm": 0.0005820005317218602,
-      "learning_rate": 0.00019999661427527529,
-      "loss": 46.0,
-      "step": 34269
-    },
-    {
-      "epoch": 2.6201808207657167,
-      "grad_norm": 0.0011149175697937608,
-      "learning_rate": 0.00019999661407761891,
-      "loss": 46.0,
-      "step": 34270
-    },
-    {
-      "epoch": 2.6202572777491064,
-      "grad_norm": 0.0014274739660322666,
-      "learning_rate": 0.00019999661387995677,
-      "loss": 46.0,
-      "step": 34271
-    },
-    {
-      "epoch": 2.620333734732496,
-      "grad_norm": 0.00154223816934973,
-      "learning_rate": 0.00019999661368228882,
-      "loss": 46.0,
-      "step": 34272
-    },
-    {
-      "epoch": 2.620410191715886,
-      "grad_norm": 0.00814348366111517,
-      "learning_rate": 0.00019999661348461513,
-      "loss": 46.0,
-      "step": 34273
-    },
-    {
-      "epoch": 2.6204866486992757,
-      "grad_norm": 0.0024103494361042976,
-      "learning_rate": 0.0001999966132869357,
-      "loss": 46.0,
-      "step": 34274
-    },
-    {
-      "epoch": 2.620563105682665,
-      "grad_norm": 0.001020262949168682,
-      "learning_rate": 0.00019999661308925046,
-      "loss": 46.0,
-      "step": 34275
-    },
-    {
-      "epoch": 2.620639562666055,
-      "grad_norm": 0.0028688423335552216,
-      "learning_rate": 0.00019999661289155944,
-      "loss": 46.0,
-      "step": 34276
-    },
-    {
-      "epoch": 2.6207160196494446,
-      "grad_norm": 0.002358399098739028,
-      "learning_rate": 0.0001999966126938627,
-      "loss": 46.0,
-      "step": 34277
-    },
-    {
-      "epoch": 2.6207924766328343,
-      "grad_norm": 0.0029547843150794506,
-      "learning_rate": 0.00019999661249616016,
-      "loss": 46.0,
-      "step": 34278
-    },
-    {
-      "epoch": 2.620868933616224,
-      "grad_norm": 0.0013284212909638882,
-      "learning_rate": 0.00019999661229845182,
-      "loss": 46.0,
-      "step": 34279
-    },
-    {
-      "epoch": 2.620945390599614,
-      "grad_norm": 0.0023126585874706507,
-      "learning_rate": 0.00019999661210073777,
-      "loss": 46.0,
-      "step": 34280
-    },
-    {
-      "epoch": 2.6210218475830036,
-      "grad_norm": 0.0021834131330251694,
-      "learning_rate": 0.00019999661190301792,
-      "loss": 46.0,
-      "step": 34281
-    },
-    {
-      "epoch": 2.6210983045663934,
-      "grad_norm": 0.001145324669778347,
-      "learning_rate": 0.00019999661170529232,
-      "loss": 46.0,
-      "step": 34282
-    },
-    {
-      "epoch": 2.621174761549783,
-      "grad_norm": 0.0011937310919165611,
-      "learning_rate": 0.00019999661150756092,
-      "loss": 46.0,
-      "step": 34283
-    },
-    {
-      "epoch": 2.621251218533173,
-      "grad_norm": 0.0054331193678081036,
-      "learning_rate": 0.00019999661130982378,
-      "loss": 46.0,
-      "step": 34284
-    },
-    {
-      "epoch": 2.6213276755165626,
-      "grad_norm": 0.002767684403806925,
-      "learning_rate": 0.00019999661111208083,
-      "loss": 46.0,
-      "step": 34285
-    },
-    {
-      "epoch": 2.6214041324999524,
-      "grad_norm": 0.008863287046551704,
-      "learning_rate": 0.00019999661091433214,
-      "loss": 46.0,
-      "step": 34286
-    },
-    {
-      "epoch": 2.621480589483342,
-      "grad_norm": 0.001534686191007495,
-      "learning_rate": 0.0001999966107165777,
-      "loss": 46.0,
-      "step": 34287
-    },
-    {
-      "epoch": 2.6215570464667315,
-      "grad_norm": 0.000786025368142873,
-      "learning_rate": 0.00019999661051881747,
-      "loss": 46.0,
-      "step": 34288
-    },
-    {
-      "epoch": 2.6216335034501212,
-      "grad_norm": 0.0008636950515210629,
-      "learning_rate": 0.00019999661032105146,
-      "loss": 46.0,
-      "step": 34289
-    },
-    {
-      "epoch": 2.621709960433511,
-      "grad_norm": 0.0013416169676929712,
-      "learning_rate": 0.0001999966101232797,
-      "loss": 46.0,
-      "step": 34290
-    },
-    {
-      "epoch": 2.6217864174169008,
-      "grad_norm": 0.0021538599394261837,
-      "learning_rate": 0.00019999660992550217,
-      "loss": 46.0,
-      "step": 34291
-    },
-    {
-      "epoch": 2.6218628744002905,
-      "grad_norm": 0.001304124714806676,
-      "learning_rate": 0.00019999660972771884,
-      "loss": 46.0,
-      "step": 34292
-    },
-    {
-      "epoch": 2.6219393313836803,
-      "grad_norm": 0.0011879676021635532,
-      "learning_rate": 0.00019999660952992976,
-      "loss": 46.0,
-      "step": 34293
-    },
-    {
-      "epoch": 2.62201578836707,
-      "grad_norm": 0.0019534036982804537,
-      "learning_rate": 0.0001999966093321349,
-      "loss": 46.0,
-      "step": 34294
-    },
-    {
-      "epoch": 2.62209224535046,
-      "grad_norm": 0.003510749898850918,
-      "learning_rate": 0.00019999660913433428,
-      "loss": 46.0,
-      "step": 34295
-    },
-    {
-      "epoch": 2.6221687023338496,
-      "grad_norm": 0.005587299820035696,
-      "learning_rate": 0.0001999966089365279,
-      "loss": 46.0,
-      "step": 34296
-    },
-    {
-      "epoch": 2.622245159317239,
-      "grad_norm": 0.000609685608651489,
-      "learning_rate": 0.00019999660873871577,
-      "loss": 46.0,
-      "step": 34297
-    },
-    {
-      "epoch": 2.6223216163006287,
-      "grad_norm": 0.002166262362152338,
-      "learning_rate": 0.00019999660854089782,
-      "loss": 46.0,
-      "step": 34298
-    },
-    {
-      "epoch": 2.6223980732840184,
-      "grad_norm": 0.0036390682216733694,
-      "learning_rate": 0.00019999660834307416,
-      "loss": 46.0,
-      "step": 34299
-    },
-    {
-      "epoch": 2.622474530267408,
-      "grad_norm": 0.0044851237908005714,
-      "learning_rate": 0.00019999660814524467,
-      "loss": 46.0,
-      "step": 34300
-    },
-    {
-      "epoch": 2.622550987250798,
-      "grad_norm": 0.0018594893626868725,
-      "learning_rate": 0.00019999660794740946,
-      "loss": 46.0,
-      "step": 34301
-    },
-    {
-      "epoch": 2.6226274442341877,
-      "grad_norm": 0.0031945365481078625,
-      "learning_rate": 0.00019999660774956845,
-      "loss": 46.0,
-      "step": 34302
-    },
-    {
-      "epoch": 2.6227039012175775,
-      "grad_norm": 0.001575936796143651,
-      "learning_rate": 0.00019999660755172166,
-      "loss": 46.0,
-      "step": 34303
-    },
-    {
-      "epoch": 2.622780358200967,
-      "grad_norm": 0.0022734710946679115,
-      "learning_rate": 0.00019999660735386913,
-      "loss": 46.0,
-      "step": 34304
-    },
-    {
-      "epoch": 2.622856815184357,
-      "grad_norm": 0.0025252013001590967,
-      "learning_rate": 0.00019999660715601083,
-      "loss": 46.0,
-      "step": 34305
-    },
-    {
-      "epoch": 2.6229332721677467,
-      "grad_norm": 0.001038175425492227,
-      "learning_rate": 0.00019999660695814675,
-      "loss": 46.0,
-      "step": 34306
-    },
-    {
-      "epoch": 2.6230097291511365,
-      "grad_norm": 0.0008089214097708464,
-      "learning_rate": 0.00019999660676027692,
-      "loss": 46.0,
-      "step": 34307
-    },
-    {
-      "epoch": 2.6230861861345263,
-      "grad_norm": 0.004474613815546036,
-      "learning_rate": 0.0001999966065624013,
-      "loss": 46.0,
-      "step": 34308
-    },
-    {
-      "epoch": 2.623162643117916,
-      "grad_norm": 0.002928811125457287,
-      "learning_rate": 0.0001999966063645199,
-      "loss": 46.0,
-      "step": 34309
-    },
-    {
-      "epoch": 2.6232391001013053,
-      "grad_norm": 0.0070663499645888805,
-      "learning_rate": 0.00019999660616663276,
-      "loss": 46.0,
-      "step": 34310
-    },
-    {
-      "epoch": 2.623315557084695,
-      "grad_norm": 0.0021506063640117645,
-      "learning_rate": 0.00019999660596873984,
-      "loss": 46.0,
-      "step": 34311
-    },
-    {
-      "epoch": 2.623392014068085,
-      "grad_norm": 0.0022853894624859095,
-      "learning_rate": 0.00019999660577084115,
-      "loss": 46.0,
-      "step": 34312
-    },
-    {
-      "epoch": 2.6234684710514746,
-      "grad_norm": 0.000793284852989018,
-      "learning_rate": 0.00019999660557293666,
-      "loss": 46.0,
-      "step": 34313
-    },
-    {
-      "epoch": 2.6235449280348644,
-      "grad_norm": 0.0008688968373462558,
-      "learning_rate": 0.00019999660537502645,
-      "loss": 46.0,
-      "step": 34314
-    },
-    {
-      "epoch": 2.623621385018254,
-      "grad_norm": 0.001831280766054988,
-      "learning_rate": 0.00019999660517711047,
-      "loss": 46.0,
-      "step": 34315
-    },
-    {
-      "epoch": 2.623697842001644,
-      "grad_norm": 0.0012517948634922504,
-      "learning_rate": 0.00019999660497918868,
-      "loss": 46.0,
-      "step": 34316
-    },
-    {
-      "epoch": 2.6237742989850337,
-      "grad_norm": 0.0013124850811436772,
-      "learning_rate": 0.00019999660478126115,
-      "loss": 46.0,
-      "step": 34317
-    },
-    {
-      "epoch": 2.6238507559684234,
-      "grad_norm": 0.0013771505327895284,
-      "learning_rate": 0.00019999660458332782,
-      "loss": 46.0,
-      "step": 34318
-    },
-    {
-      "epoch": 2.6239272129518127,
-      "grad_norm": 0.004698462318629026,
-      "learning_rate": 0.00019999660438538874,
-      "loss": 46.0,
-      "step": 34319
-    },
-    {
-      "epoch": 2.6240036699352025,
-      "grad_norm": 0.0013371316017583013,
-      "learning_rate": 0.0001999966041874439,
-      "loss": 46.0,
-      "step": 34320
-    },
-    {
-      "epoch": 2.6240801269185923,
-      "grad_norm": 0.002094406634569168,
-      "learning_rate": 0.0001999966039894933,
-      "loss": 46.0,
-      "step": 34321
-    },
-    {
-      "epoch": 2.624156583901982,
-      "grad_norm": 0.0022246220614761114,
-      "learning_rate": 0.0001999966037915369,
-      "loss": 46.0,
-      "step": 34322
-    },
-    {
-      "epoch": 2.624233040885372,
-      "grad_norm": 0.0017451178282499313,
-      "learning_rate": 0.00019999660359357475,
-      "loss": 46.0,
-      "step": 34323
-    },
-    {
-      "epoch": 2.6243094978687616,
-      "grad_norm": 0.001889218227006495,
-      "learning_rate": 0.00019999660339560684,
-      "loss": 46.0,
-      "step": 34324
-    },
-    {
-      "epoch": 2.6243859548521513,
-      "grad_norm": 0.0023120564874261618,
-      "learning_rate": 0.00019999660319763315,
-      "loss": 46.0,
-      "step": 34325
-    },
-    {
-      "epoch": 2.624462411835541,
-      "grad_norm": 0.0010360432788729668,
-      "learning_rate": 0.00019999660299965366,
-      "loss": 46.0,
-      "step": 34326
-    },
-    {
-      "epoch": 2.624538868818931,
-      "grad_norm": 0.0015052318340167403,
-      "learning_rate": 0.00019999660280166842,
-      "loss": 46.0,
-      "step": 34327
-    },
-    {
-      "epoch": 2.6246153258023206,
-      "grad_norm": 0.0027069970965385437,
-      "learning_rate": 0.00019999660260367743,
-      "loss": 46.0,
-      "step": 34328
-    },
-    {
-      "epoch": 2.6246917827857104,
-      "grad_norm": 0.003471989184617996,
-      "learning_rate": 0.00019999660240568065,
-      "loss": 46.0,
-      "step": 34329
-    },
-    {
-      "epoch": 2.6247682397691,
-      "grad_norm": 0.002597326645627618,
-      "learning_rate": 0.00019999660220767812,
-      "loss": 46.0,
-      "step": 34330
-    },
-    {
-      "epoch": 2.62484469675249,
-      "grad_norm": 0.0014209537766873837,
-      "learning_rate": 0.00019999660200966984,
-      "loss": 46.0,
-      "step": 34331
-    },
-    {
-      "epoch": 2.624921153735879,
-      "grad_norm": 0.0005744562949985266,
-      "learning_rate": 0.00019999660181165574,
-      "loss": 46.0,
-      "step": 34332
-    },
-    {
-      "epoch": 2.624997610719269,
-      "grad_norm": 0.0027566375210881233,
-      "learning_rate": 0.0001999966016136359,
-      "loss": 46.0,
-      "step": 34333
-    },
-    {
-      "epoch": 2.6250740677026587,
-      "grad_norm": 0.002971071982756257,
-      "learning_rate": 0.00019999660141561027,
-      "loss": 46.0,
-      "step": 34334
-    },
-    {
-      "epoch": 2.6251505246860485,
-      "grad_norm": 0.0012065370101481676,
-      "learning_rate": 0.0001999966012175789,
-      "loss": 46.0,
-      "step": 34335
-    },
-    {
-      "epoch": 2.6252269816694382,
-      "grad_norm": 0.00252229324541986,
-      "learning_rate": 0.00019999660101954175,
-      "loss": 46.0,
-      "step": 34336
-    },
-    {
-      "epoch": 2.625303438652828,
-      "grad_norm": 0.0016391382087022066,
-      "learning_rate": 0.0001999966008214988,
-      "loss": 46.0,
-      "step": 34337
-    },
-    {
-      "epoch": 2.6253798956362178,
-      "grad_norm": 0.008321761153638363,
-      "learning_rate": 0.00019999660062345012,
-      "loss": 46.0,
-      "step": 34338
-    },
-    {
-      "epoch": 2.6254563526196075,
-      "grad_norm": 0.0021071250084787607,
-      "learning_rate": 0.00019999660042539566,
-      "loss": 46.0,
-      "step": 34339
-    },
-    {
-      "epoch": 2.625532809602997,
-      "grad_norm": 0.00257648597471416,
-      "learning_rate": 0.00019999660022733542,
-      "loss": 46.0,
-      "step": 34340
-    },
-    {
-      "epoch": 2.6256092665863866,
-      "grad_norm": 0.005129849538207054,
-      "learning_rate": 0.0001999966000292694,
-      "loss": 46.0,
-      "step": 34341
-    },
-    {
-      "epoch": 2.6256857235697764,
-      "grad_norm": 0.004966387525200844,
-      "learning_rate": 0.00019999659983119765,
-      "loss": 46.0,
-      "step": 34342
-    },
-    {
-      "epoch": 2.625762180553166,
-      "grad_norm": 0.0020260175224393606,
-      "learning_rate": 0.00019999659963312012,
-      "loss": 46.0,
-      "step": 34343
-    },
-    {
-      "epoch": 2.625838637536556,
-      "grad_norm": 0.0017710226820781827,
-      "learning_rate": 0.0001999965994350368,
-      "loss": 46.0,
-      "step": 34344
-    },
-    {
-      "epoch": 2.6259150945199456,
-      "grad_norm": 0.0019671388436108828,
-      "learning_rate": 0.00019999659923694774,
-      "loss": 46.0,
-      "step": 34345
-    },
-    {
-      "epoch": 2.6259915515033354,
-      "grad_norm": 0.0016586111159995198,
-      "learning_rate": 0.00019999659903885287,
-      "loss": 46.0,
-      "step": 34346
-    },
-    {
-      "epoch": 2.626068008486725,
-      "grad_norm": 0.0020401792135089636,
-      "learning_rate": 0.00019999659884075227,
-      "loss": 46.0,
-      "step": 34347
-    },
-    {
-      "epoch": 2.626144465470115,
-      "grad_norm": 0.0009087237995117903,
-      "learning_rate": 0.00019999659864264587,
-      "loss": 46.0,
-      "step": 34348
-    },
-    {
-      "epoch": 2.6262209224535047,
-      "grad_norm": 0.001577352057211101,
-      "learning_rate": 0.0001999965984445337,
-      "loss": 46.0,
-      "step": 34349
-    },
-    {
-      "epoch": 2.6262973794368945,
-      "grad_norm": 0.002824016846716404,
-      "learning_rate": 0.00019999659824641579,
-      "loss": 46.0,
-      "step": 34350
-    },
-    {
-      "epoch": 2.626373836420284,
-      "grad_norm": 0.001687657437287271,
-      "learning_rate": 0.0001999965980482921,
-      "loss": 46.0,
-      "step": 34351
-    },
-    {
-      "epoch": 2.626450293403674,
-      "grad_norm": 0.0012981118634343147,
-      "learning_rate": 0.00019999659785016266,
-      "loss": 46.0,
-      "step": 34352
-    },
-    {
-      "epoch": 2.6265267503870637,
-      "grad_norm": 0.0025908120442181826,
-      "learning_rate": 0.0001999965976520274,
-      "loss": 46.0,
-      "step": 34353
-    },
-    {
-      "epoch": 2.626603207370453,
-      "grad_norm": 0.001559897675178945,
-      "learning_rate": 0.0001999965974538864,
-      "loss": 46.0,
-      "step": 34354
-    },
-    {
-      "epoch": 2.626679664353843,
-      "grad_norm": 0.0015025973552837968,
-      "learning_rate": 0.00019999659725573963,
-      "loss": 46.0,
-      "step": 34355
-    },
-    {
-      "epoch": 2.6267561213372326,
-      "grad_norm": 0.0030372445471584797,
-      "learning_rate": 0.0001999965970575871,
-      "loss": 46.0,
-      "step": 34356
-    },
-    {
-      "epoch": 2.6268325783206223,
-      "grad_norm": 0.0015179236652329564,
-      "learning_rate": 0.00019999659685942877,
-      "loss": 46.0,
-      "step": 34357
-    },
-    {
-      "epoch": 2.626909035304012,
-      "grad_norm": 0.0029208441264927387,
-      "learning_rate": 0.0001999965966612647,
-      "loss": 46.0,
-      "step": 34358
-    },
-    {
-      "epoch": 2.626985492287402,
-      "grad_norm": 0.001754774246364832,
-      "learning_rate": 0.00019999659646309484,
-      "loss": 46.0,
-      "step": 34359
-    },
-    {
-      "epoch": 2.6270619492707916,
-      "grad_norm": 0.0019002397311851382,
-      "learning_rate": 0.00019999659626491925,
-      "loss": 46.0,
-      "step": 34360
-    },
-    {
-      "epoch": 2.6271384062541814,
-      "grad_norm": 0.0021670605055987835,
-      "learning_rate": 0.00019999659606673785,
-      "loss": 46.0,
-      "step": 34361
-    },
-    {
-      "epoch": 2.6272148632375707,
-      "grad_norm": 0.0015996188158169389,
-      "learning_rate": 0.0001999965958685507,
-      "loss": 46.0,
-      "step": 34362
-    },
-    {
-      "epoch": 2.6272913202209605,
-      "grad_norm": 0.0019499182235449553,
-      "learning_rate": 0.00019999659567035777,
-      "loss": 46.0,
-      "step": 34363
-    },
-    {
-      "epoch": 2.62736777720435,
-      "grad_norm": 0.002738367998972535,
-      "learning_rate": 0.00019999659547215908,
-      "loss": 46.0,
-      "step": 34364
-    },
-    {
-      "epoch": 2.62744423418774,
-      "grad_norm": 0.0008483403944410384,
-      "learning_rate": 0.00019999659527395461,
-      "loss": 46.0,
-      "step": 34365
-    },
-    {
-      "epoch": 2.6275206911711297,
-      "grad_norm": 0.0020275218412280083,
-      "learning_rate": 0.00019999659507574438,
-      "loss": 46.0,
-      "step": 34366
-    },
-    {
-      "epoch": 2.6275971481545195,
-      "grad_norm": 0.0024220733903348446,
-      "learning_rate": 0.00019999659487752837,
-      "loss": 46.0,
-      "step": 34367
-    },
-    {
-      "epoch": 2.6276736051379093,
-      "grad_norm": 0.0010920250788331032,
-      "learning_rate": 0.0001999965946793066,
-      "loss": 46.0,
-      "step": 34368
-    },
-    {
-      "epoch": 2.627750062121299,
-      "grad_norm": 0.0010820974130183458,
-      "learning_rate": 0.00019999659448107906,
-      "loss": 46.0,
-      "step": 34369
-    },
-    {
-      "epoch": 2.627826519104689,
-      "grad_norm": 0.0027158481534570456,
-      "learning_rate": 0.00019999659428284575,
-      "loss": 46.0,
-      "step": 34370
-    },
-    {
-      "epoch": 2.6279029760880785,
-      "grad_norm": 0.002140153432264924,
-      "learning_rate": 0.00019999659408460668,
-      "loss": 46.0,
-      "step": 34371
-    },
-    {
-      "epoch": 2.6279794330714683,
-      "grad_norm": 0.0016918446635827422,
-      "learning_rate": 0.00019999659388636183,
-      "loss": 46.0,
-      "step": 34372
-    },
-    {
-      "epoch": 2.628055890054858,
-      "grad_norm": 0.0025619249790906906,
-      "learning_rate": 0.00019999659368811118,
-      "loss": 46.0,
-      "step": 34373
-    },
-    {
-      "epoch": 2.628132347038248,
-      "grad_norm": 0.0029054523911327124,
-      "learning_rate": 0.0001999965934898548,
-      "loss": 46.0,
-      "step": 34374
-    },
-    {
-      "epoch": 2.6282088040216376,
-      "grad_norm": 0.001972772879526019,
-      "learning_rate": 0.00019999659329159267,
-      "loss": 46.0,
-      "step": 34375
-    },
-    {
-      "epoch": 2.628285261005027,
-      "grad_norm": 0.0021854841616004705,
-      "learning_rate": 0.00019999659309332475,
-      "loss": 46.0,
-      "step": 34376
-    },
-    {
-      "epoch": 2.6283617179884167,
-      "grad_norm": 0.000992395682260394,
-      "learning_rate": 0.00019999659289505104,
-      "loss": 46.0,
-      "step": 34377
-    },
-    {
-      "epoch": 2.6284381749718064,
-      "grad_norm": 0.0008169599459506571,
-      "learning_rate": 0.00019999659269677157,
-      "loss": 46.0,
-      "step": 34378
-    },
-    {
-      "epoch": 2.628514631955196,
-      "grad_norm": 0.0013462668284773827,
-      "learning_rate": 0.00019999659249848634,
-      "loss": 46.0,
-      "step": 34379
-    },
-    {
-      "epoch": 2.628591088938586,
-      "grad_norm": 0.0016212827758863568,
-      "learning_rate": 0.00019999659230019533,
-      "loss": 46.0,
-      "step": 34380
-    },
-    {
-      "epoch": 2.6286675459219757,
-      "grad_norm": 0.0017279828898608685,
-      "learning_rate": 0.00019999659210189857,
-      "loss": 46.0,
-      "step": 34381
-    },
-    {
-      "epoch": 2.6287440029053655,
-      "grad_norm": 0.0017663320759311318,
-      "learning_rate": 0.00019999659190359604,
-      "loss": 46.0,
-      "step": 34382
-    },
-    {
-      "epoch": 2.6288204598887552,
-      "grad_norm": 0.0015197661705315113,
-      "learning_rate": 0.00019999659170528774,
-      "loss": 46.0,
-      "step": 34383
-    },
-    {
-      "epoch": 2.6288969168721446,
-      "grad_norm": 0.00364927900955081,
-      "learning_rate": 0.00019999659150697367,
-      "loss": 46.0,
-      "step": 34384
-    },
-    {
-      "epoch": 2.6289733738555343,
-      "grad_norm": 0.004130763933062553,
-      "learning_rate": 0.0001999965913086538,
-      "loss": 46.0,
-      "step": 34385
-    },
-    {
-      "epoch": 2.629049830838924,
-      "grad_norm": 0.0013872553827241063,
-      "learning_rate": 0.00019999659111032817,
-      "loss": 46.0,
-      "step": 34386
-    },
-    {
-      "epoch": 2.629126287822314,
-      "grad_norm": 0.0010498985648155212,
-      "learning_rate": 0.00019999659091199677,
-      "loss": 46.0,
-      "step": 34387
-    },
-    {
-      "epoch": 2.6292027448057036,
-      "grad_norm": 0.0013104953104630113,
-      "learning_rate": 0.0001999965907136596,
-      "loss": 46.0,
-      "step": 34388
-    },
-    {
-      "epoch": 2.6292792017890934,
-      "grad_norm": 0.0033699474297463894,
-      "learning_rate": 0.0001999965905153167,
-      "loss": 46.0,
-      "step": 34389
-    },
-    {
-      "epoch": 2.629355658772483,
-      "grad_norm": 0.007030009757727385,
-      "learning_rate": 0.000199996590316968,
-      "loss": 46.0,
-      "step": 34390
-    },
-    {
-      "epoch": 2.629432115755873,
-      "grad_norm": 0.002016279147937894,
-      "learning_rate": 0.00019999659011861356,
-      "loss": 46.0,
-      "step": 34391
-    },
-    {
-      "epoch": 2.6295085727392626,
-      "grad_norm": 0.0018494086107239127,
-      "learning_rate": 0.0001999965899202533,
-      "loss": 46.0,
-      "step": 34392
-    },
-    {
-      "epoch": 2.6295850297226524,
-      "grad_norm": 0.0013845224166288972,
-      "learning_rate": 0.00019999658972188732,
-      "loss": 46.0,
-      "step": 34393
-    },
-    {
-      "epoch": 2.629661486706042,
-      "grad_norm": 0.0006367545574903488,
-      "learning_rate": 0.00019999658952351554,
-      "loss": 46.0,
-      "step": 34394
-    },
-    {
-      "epoch": 2.629737943689432,
-      "grad_norm": 0.0026914917398244143,
-      "learning_rate": 0.000199996589325138,
-      "loss": 46.0,
-      "step": 34395
-    },
-    {
-      "epoch": 2.6298144006728217,
-      "grad_norm": 0.000510679034050554,
-      "learning_rate": 0.0001999965891267547,
-      "loss": 46.0,
-      "step": 34396
-    },
-    {
-      "epoch": 2.6298908576562114,
-      "grad_norm": 0.002147011226043105,
-      "learning_rate": 0.0001999965889283656,
-      "loss": 46.0,
-      "step": 34397
-    },
-    {
-      "epoch": 2.6299673146396008,
-      "grad_norm": 0.0006715363706462085,
-      "learning_rate": 0.00019999658872997075,
-      "loss": 46.0,
-      "step": 34398
-    },
-    {
-      "epoch": 2.6300437716229905,
-      "grad_norm": 0.0020176381804049015,
-      "learning_rate": 0.00019999658853157013,
-      "loss": 46.0,
-      "step": 34399
-    },
-    {
-      "epoch": 2.6301202286063803,
-      "grad_norm": 0.0038959882222115993,
-      "learning_rate": 0.00019999658833316374,
-      "loss": 46.0,
-      "step": 34400
-    },
-    {
-      "epoch": 2.63019668558977,
-      "grad_norm": 0.0014475034549832344,
-      "learning_rate": 0.0001999965881347516,
-      "loss": 46.0,
-      "step": 34401
-    },
-    {
-      "epoch": 2.63027314257316,
-      "grad_norm": 0.0027651104610413313,
-      "learning_rate": 0.00019999658793633368,
-      "loss": 46.0,
-      "step": 34402
-    },
-    {
-      "epoch": 2.6303495995565496,
-      "grad_norm": 0.0041971090249717236,
-      "learning_rate": 0.00019999658773790996,
-      "loss": 46.0,
-      "step": 34403
-    },
-    {
-      "epoch": 2.6304260565399393,
-      "grad_norm": 0.0018077807035297155,
-      "learning_rate": 0.00019999658753948053,
-      "loss": 46.0,
-      "step": 34404
-    },
-    {
-      "epoch": 2.630502513523329,
-      "grad_norm": 0.00109379505738616,
-      "learning_rate": 0.00019999658734104527,
-      "loss": 46.0,
-      "step": 34405
-    },
-    {
-      "epoch": 2.6305789705067184,
-      "grad_norm": 0.003543097758665681,
-      "learning_rate": 0.00019999658714260426,
-      "loss": 46.0,
-      "step": 34406
-    },
-    {
-      "epoch": 2.630655427490108,
-      "grad_norm": 0.001951010199263692,
-      "learning_rate": 0.00019999658694415748,
-      "loss": 46.0,
-      "step": 34407
-    },
-    {
-      "epoch": 2.630731884473498,
-      "grad_norm": 0.0016046827659010887,
-      "learning_rate": 0.00019999658674570495,
-      "loss": 46.0,
-      "step": 34408
-    },
-    {
-      "epoch": 2.6308083414568877,
-      "grad_norm": 0.0026712275575846434,
-      "learning_rate": 0.00019999658654724665,
-      "loss": 46.0,
-      "step": 34409
-    },
-    {
-      "epoch": 2.6308847984402775,
-      "grad_norm": 0.0020372001454234123,
-      "learning_rate": 0.00019999658634878257,
-      "loss": 46.0,
-      "step": 34410
-    },
-    {
-      "epoch": 2.630961255423667,
-      "grad_norm": 0.007009337190538645,
-      "learning_rate": 0.00019999658615031272,
-      "loss": 46.0,
-      "step": 34411
-    },
-    {
-      "epoch": 2.631037712407057,
-      "grad_norm": 0.0014605749165639281,
-      "learning_rate": 0.00019999658595183707,
-      "loss": 46.0,
-      "step": 34412
-    },
-    {
-      "epoch": 2.6311141693904467,
-      "grad_norm": 0.0013487362302839756,
-      "learning_rate": 0.0001999965857533557,
-      "loss": 46.0,
-      "step": 34413
-    },
-    {
-      "epoch": 2.6311906263738365,
-      "grad_norm": 0.0034657728392630816,
-      "learning_rate": 0.00019999658555486856,
-      "loss": 46.0,
-      "step": 34414
-    },
-    {
-      "epoch": 2.6312670833572263,
-      "grad_norm": 0.004519772715866566,
-      "learning_rate": 0.00019999658535637562,
-      "loss": 46.0,
-      "step": 34415
-    },
-    {
-      "epoch": 2.631343540340616,
-      "grad_norm": 0.0028691054321825504,
-      "learning_rate": 0.00019999658515787693,
-      "loss": 46.0,
-      "step": 34416
-    },
-    {
-      "epoch": 2.6314199973240058,
-      "grad_norm": 0.004534099251031876,
-      "learning_rate": 0.00019999658495937245,
-      "loss": 46.0,
-      "step": 34417
-    },
-    {
-      "epoch": 2.6314964543073955,
-      "grad_norm": 0.0007312087691389024,
-      "learning_rate": 0.00019999658476086224,
-      "loss": 46.0,
-      "step": 34418
-    },
-    {
-      "epoch": 2.6315729112907853,
-      "grad_norm": 0.000681075151078403,
-      "learning_rate": 0.00019999658456234623,
-      "loss": 46.0,
-      "step": 34419
-    },
-    {
-      "epoch": 2.6316493682741746,
-      "grad_norm": 0.0007741820299997926,
-      "learning_rate": 0.00019999658436382445,
-      "loss": 46.0,
-      "step": 34420
-    },
-    {
-      "epoch": 2.6317258252575644,
-      "grad_norm": 0.0019792323000729084,
-      "learning_rate": 0.00019999658416529692,
-      "loss": 46.0,
-      "step": 34421
-    },
-    {
-      "epoch": 2.631802282240954,
-      "grad_norm": 0.001056221197359264,
-      "learning_rate": 0.0001999965839667636,
-      "loss": 46.0,
-      "step": 34422
-    },
-    {
-      "epoch": 2.631878739224344,
-      "grad_norm": 0.0012056275736540556,
-      "learning_rate": 0.00019999658376822452,
-      "loss": 46.0,
-      "step": 34423
-    },
-    {
-      "epoch": 2.6319551962077337,
-      "grad_norm": 0.002178872236981988,
-      "learning_rate": 0.00019999658356967967,
-      "loss": 46.0,
-      "step": 34424
-    },
-    {
-      "epoch": 2.6320316531911234,
-      "grad_norm": 0.004085775464773178,
-      "learning_rate": 0.00019999658337112902,
-      "loss": 46.0,
-      "step": 34425
-    },
-    {
-      "epoch": 2.632108110174513,
-      "grad_norm": 0.0010769017972052097,
-      "learning_rate": 0.00019999658317257265,
-      "loss": 46.0,
-      "step": 34426
-    },
-    {
-      "epoch": 2.632184567157903,
-      "grad_norm": 0.006025294773280621,
-      "learning_rate": 0.0001999965829740105,
-      "loss": 46.0,
-      "step": 34427
-    },
-    {
-      "epoch": 2.6322610241412923,
-      "grad_norm": 0.0007110938895493746,
-      "learning_rate": 0.00019999658277544257,
-      "loss": 46.0,
-      "step": 34428
-    },
-    {
-      "epoch": 2.632337481124682,
-      "grad_norm": 0.005359198898077011,
-      "learning_rate": 0.00019999658257686885,
-      "loss": 46.0,
-      "step": 34429
-    },
-    {
-      "epoch": 2.632413938108072,
-      "grad_norm": 0.001609212253242731,
-      "learning_rate": 0.0001999965823782894,
-      "loss": 46.0,
-      "step": 34430
-    },
-    {
-      "epoch": 2.6324903950914615,
-      "grad_norm": 0.0019025879446417093,
-      "learning_rate": 0.00019999658217970419,
-      "loss": 46.0,
-      "step": 34431
-    },
-    {
-      "epoch": 2.6325668520748513,
-      "grad_norm": 0.001044404343701899,
-      "learning_rate": 0.00019999658198111318,
-      "loss": 46.0,
-      "step": 34432
-    },
-    {
-      "epoch": 2.632643309058241,
-      "grad_norm": 0.002205275697633624,
-      "learning_rate": 0.00019999658178251642,
-      "loss": 46.0,
-      "step": 34433
-    },
-    {
-      "epoch": 2.632719766041631,
-      "grad_norm": 0.001529672066681087,
-      "learning_rate": 0.00019999658158391387,
-      "loss": 46.0,
-      "step": 34434
-    },
-    {
-      "epoch": 2.6327962230250206,
-      "grad_norm": 0.0007609836757183075,
-      "learning_rate": 0.00019999658138530554,
-      "loss": 46.0,
-      "step": 34435
-    },
-    {
-      "epoch": 2.6328726800084103,
-      "grad_norm": 0.01854216866195202,
-      "learning_rate": 0.00019999658118669147,
-      "loss": 46.0,
-      "step": 34436
-    },
-    {
-      "epoch": 2.6329491369918,
-      "grad_norm": 0.0010859571630135179,
-      "learning_rate": 0.00019999658098807162,
-      "loss": 46.0,
-      "step": 34437
-    },
-    {
-      "epoch": 2.63302559397519,
-      "grad_norm": 0.0020236559212207794,
-      "learning_rate": 0.000199996580789446,
-      "loss": 46.0,
-      "step": 34438
-    },
-    {
-      "epoch": 2.6331020509585796,
-      "grad_norm": 0.0007272697403095663,
-      "learning_rate": 0.0001999965805908146,
-      "loss": 46.0,
-      "step": 34439
-    },
-    {
-      "epoch": 2.6331785079419694,
-      "grad_norm": 0.0018953965045511723,
-      "learning_rate": 0.00019999658039217746,
-      "loss": 46.0,
-      "step": 34440
-    },
-    {
-      "epoch": 2.6332549649253587,
-      "grad_norm": 0.0022129833232611418,
-      "learning_rate": 0.00019999658019353452,
-      "loss": 46.0,
-      "step": 34441
-    },
-    {
-      "epoch": 2.6333314219087485,
-      "grad_norm": 0.0027869415935128927,
-      "learning_rate": 0.00019999657999488583,
-      "loss": 46.0,
-      "step": 34442
-    },
-    {
-      "epoch": 2.6334078788921382,
-      "grad_norm": 0.0012360612163320184,
-      "learning_rate": 0.00019999657979623137,
-      "loss": 46.0,
-      "step": 34443
-    },
-    {
-      "epoch": 2.633484335875528,
-      "grad_norm": 0.0013782597379758954,
-      "learning_rate": 0.0001999965795975711,
-      "loss": 46.0,
-      "step": 34444
-    },
-    {
-      "epoch": 2.6335607928589178,
-      "grad_norm": 0.0070426990278065205,
-      "learning_rate": 0.00019999657939890513,
-      "loss": 46.0,
-      "step": 34445
-    },
-    {
-      "epoch": 2.6336372498423075,
-      "grad_norm": 0.002031357027590275,
-      "learning_rate": 0.00019999657920023335,
-      "loss": 46.0,
-      "step": 34446
-    },
-    {
-      "epoch": 2.6337137068256973,
-      "grad_norm": 0.003125097369775176,
-      "learning_rate": 0.0001999965790015558,
-      "loss": 46.0,
-      "step": 34447
-    },
-    {
-      "epoch": 2.633790163809087,
-      "grad_norm": 0.002093460876494646,
-      "learning_rate": 0.0001999965788028725,
-      "loss": 46.0,
-      "step": 34448
-    },
-    {
-      "epoch": 2.633866620792477,
-      "grad_norm": 0.0041023981757462025,
-      "learning_rate": 0.00019999657860418342,
-      "loss": 46.0,
-      "step": 34449
-    },
-    {
-      "epoch": 2.633943077775866,
-      "grad_norm": 0.0030651772394776344,
-      "learning_rate": 0.00019999657840548854,
-      "loss": 46.0,
-      "step": 34450
-    },
-    {
-      "epoch": 2.634019534759256,
-      "grad_norm": 0.0016462085768580437,
-      "learning_rate": 0.00019999657820678795,
-      "loss": 46.0,
-      "step": 34451
-    },
-    {
-      "epoch": 2.6340959917426456,
-      "grad_norm": 0.003537294454872608,
-      "learning_rate": 0.00019999657800808156,
-      "loss": 46.0,
-      "step": 34452
-    },
-    {
-      "epoch": 2.6341724487260354,
-      "grad_norm": 0.0009102692711167037,
-      "learning_rate": 0.0001999965778093694,
-      "loss": 46.0,
-      "step": 34453
-    },
-    {
-      "epoch": 2.634248905709425,
-      "grad_norm": 0.0050512878224253654,
-      "learning_rate": 0.00019999657761065148,
-      "loss": 46.0,
-      "step": 34454
-    },
-    {
-      "epoch": 2.634325362692815,
-      "grad_norm": 0.0009945003548637033,
-      "learning_rate": 0.00019999657741192776,
-      "loss": 46.0,
-      "step": 34455
-    },
-    {
-      "epoch": 2.6344018196762047,
-      "grad_norm": 0.001974952407181263,
-      "learning_rate": 0.0001999965772131983,
-      "loss": 46.0,
-      "step": 34456
-    },
-    {
-      "epoch": 2.6344782766595944,
-      "grad_norm": 0.003743717446923256,
-      "learning_rate": 0.0001999965770144631,
-      "loss": 46.0,
-      "step": 34457
-    },
-    {
-      "epoch": 2.634554733642984,
-      "grad_norm": 0.004385035485029221,
-      "learning_rate": 0.00019999657681572206,
-      "loss": 46.0,
-      "step": 34458
-    },
-    {
-      "epoch": 2.634631190626374,
-      "grad_norm": 0.0026215077377855778,
-      "learning_rate": 0.0001999965766169753,
-      "loss": 46.0,
-      "step": 34459
-    },
-    {
-      "epoch": 2.6347076476097637,
-      "grad_norm": 0.0020132833160459995,
-      "learning_rate": 0.00019999657641822272,
-      "loss": 46.0,
-      "step": 34460
-    },
-    {
-      "epoch": 2.6347841045931535,
-      "grad_norm": 0.002285352209582925,
-      "learning_rate": 0.00019999657621946445,
-      "loss": 46.0,
-      "step": 34461
-    },
-    {
-      "epoch": 2.6348605615765432,
-      "grad_norm": 0.001081867259927094,
-      "learning_rate": 0.00019999657602070038,
-      "loss": 46.0,
-      "step": 34462
-    },
-    {
-      "epoch": 2.6349370185599326,
-      "grad_norm": 0.0041771335527300835,
-      "learning_rate": 0.0001999965758219305,
-      "loss": 46.0,
-      "step": 34463
-    },
-    {
-      "epoch": 2.6350134755433223,
-      "grad_norm": 0.00625573517754674,
-      "learning_rate": 0.00019999657562315488,
-      "loss": 46.0,
-      "step": 34464
-    },
-    {
-      "epoch": 2.635089932526712,
-      "grad_norm": 0.0029771230183541775,
-      "learning_rate": 0.0001999965754243735,
-      "loss": 46.0,
-      "step": 34465
-    },
-    {
-      "epoch": 2.635166389510102,
-      "grad_norm": 0.0036802447866648436,
-      "learning_rate": 0.00019999657522558635,
-      "loss": 46.0,
-      "step": 34466
-    },
-    {
-      "epoch": 2.6352428464934916,
-      "grad_norm": 0.002327551832422614,
-      "learning_rate": 0.0001999965750267934,
-      "loss": 46.0,
-      "step": 34467
-    },
-    {
-      "epoch": 2.6353193034768814,
-      "grad_norm": 0.002293108031153679,
-      "learning_rate": 0.0001999965748279947,
-      "loss": 46.0,
-      "step": 34468
-    },
-    {
-      "epoch": 2.635395760460271,
-      "grad_norm": 0.002166707767173648,
-      "learning_rate": 0.00019999657462919026,
-      "loss": 46.0,
-      "step": 34469
-    },
-    {
-      "epoch": 2.635472217443661,
-      "grad_norm": 0.0009519589366391301,
-      "learning_rate": 0.00019999657443038,
-      "loss": 46.0,
-      "step": 34470
-    },
-    {
-      "epoch": 2.63554867442705,
-      "grad_norm": 0.0011486202711239457,
-      "learning_rate": 0.00019999657423156402,
-      "loss": 46.0,
-      "step": 34471
-    },
-    {
-      "epoch": 2.63562513141044,
-      "grad_norm": 0.0026585632003843784,
-      "learning_rate": 0.00019999657403274224,
-      "loss": 46.0,
-      "step": 34472
-    },
-    {
-      "epoch": 2.6357015883938297,
-      "grad_norm": 0.0011468218872323632,
-      "learning_rate": 0.00019999657383391469,
-      "loss": 46.0,
-      "step": 34473
-    },
-    {
-      "epoch": 2.6357780453772195,
-      "grad_norm": 0.001684209331870079,
-      "learning_rate": 0.00019999657363508139,
-      "loss": 46.0,
-      "step": 34474
-    },
-    {
-      "epoch": 2.6358545023606093,
-      "grad_norm": 0.0006070074741728604,
-      "learning_rate": 0.00019999657343624228,
-      "loss": 46.0,
-      "step": 34475
-    },
-    {
-      "epoch": 2.635930959343999,
-      "grad_norm": 0.0072415233589708805,
-      "learning_rate": 0.00019999657323739744,
-      "loss": 46.0,
-      "step": 34476
-    },
-    {
-      "epoch": 2.6360074163273888,
-      "grad_norm": 0.0012715159682556987,
-      "learning_rate": 0.00019999657303854682,
-      "loss": 46.0,
-      "step": 34477
-    },
-    {
-      "epoch": 2.6360838733107785,
-      "grad_norm": 0.0023942769039422274,
-      "learning_rate": 0.00019999657283969042,
-      "loss": 46.0,
-      "step": 34478
-    },
-    {
-      "epoch": 2.6361603302941683,
-      "grad_norm": 0.0010972925228998065,
-      "learning_rate": 0.00019999657264082828,
-      "loss": 46.0,
-      "step": 34479
-    },
-    {
-      "epoch": 2.636236787277558,
-      "grad_norm": 0.003284666920080781,
-      "learning_rate": 0.00019999657244196034,
-      "loss": 46.0,
-      "step": 34480
-    },
-    {
-      "epoch": 2.636313244260948,
-      "grad_norm": 0.004400253761559725,
-      "learning_rate": 0.00019999657224308663,
-      "loss": 46.0,
-      "step": 34481
-    },
-    {
-      "epoch": 2.6363897012443376,
-      "grad_norm": 0.0010560192167758942,
-      "learning_rate": 0.0001999965720442072,
-      "loss": 46.0,
-      "step": 34482
-    },
-    {
-      "epoch": 2.6364661582277273,
-      "grad_norm": 0.0020533977076411247,
-      "learning_rate": 0.00019999657184532194,
-      "loss": 46.0,
-      "step": 34483
-    },
-    {
-      "epoch": 2.636542615211117,
-      "grad_norm": 0.007894976064562798,
-      "learning_rate": 0.00019999657164643093,
-      "loss": 46.0,
-      "step": 34484
-    },
-    {
-      "epoch": 2.6366190721945064,
-      "grad_norm": 0.0035504489205777645,
-      "learning_rate": 0.00019999657144753418,
-      "loss": 46.0,
-      "step": 34485
-    },
-    {
-      "epoch": 2.636695529177896,
-      "grad_norm": 0.0021106658969074488,
-      "learning_rate": 0.00019999657124863162,
-      "loss": 46.0,
-      "step": 34486
-    },
-    {
-      "epoch": 2.636771986161286,
-      "grad_norm": 0.0019726473838090897,
-      "learning_rate": 0.00019999657104972332,
-      "loss": 46.0,
-      "step": 34487
-    },
-    {
-      "epoch": 2.6368484431446757,
-      "grad_norm": 0.0006805500597693026,
-      "learning_rate": 0.00019999657085080922,
-      "loss": 46.0,
-      "step": 34488
-    },
-    {
-      "epoch": 2.6369249001280655,
-      "grad_norm": 0.001005430007353425,
-      "learning_rate": 0.00019999657065188938,
-      "loss": 46.0,
-      "step": 34489
-    },
-    {
-      "epoch": 2.6370013571114552,
-      "grad_norm": 0.0013657687231898308,
-      "learning_rate": 0.00019999657045296373,
-      "loss": 46.0,
-      "step": 34490
-    },
-    {
-      "epoch": 2.637077814094845,
-      "grad_norm": 0.0024143001064658165,
-      "learning_rate": 0.00019999657025403236,
-      "loss": 46.0,
-      "step": 34491
-    },
-    {
-      "epoch": 2.6371542710782347,
-      "grad_norm": 0.0012777001829817891,
-      "learning_rate": 0.0001999965700550952,
-      "loss": 46.0,
-      "step": 34492
-    },
-    {
-      "epoch": 2.637230728061624,
-      "grad_norm": 0.0016745832981541753,
-      "learning_rate": 0.00019999656985615228,
-      "loss": 46.0,
-      "step": 34493
-    },
-    {
-      "epoch": 2.637307185045014,
-      "grad_norm": 0.0018149771494790912,
-      "learning_rate": 0.00019999656965720357,
-      "loss": 46.0,
-      "step": 34494
-    },
-    {
-      "epoch": 2.6373836420284036,
-      "grad_norm": 0.0011310898698866367,
-      "learning_rate": 0.0001999965694582491,
-      "loss": 46.0,
-      "step": 34495
-    },
-    {
-      "epoch": 2.6374600990117933,
-      "grad_norm": 0.0013143346877768636,
-      "learning_rate": 0.00019999656925928888,
-      "loss": 46.0,
-      "step": 34496
-    },
-    {
-      "epoch": 2.637536555995183,
-      "grad_norm": 0.0025121895596385,
-      "learning_rate": 0.00019999656906032287,
-      "loss": 46.0,
-      "step": 34497
-    },
-    {
-      "epoch": 2.637613012978573,
-      "grad_norm": 0.0021099471487104893,
-      "learning_rate": 0.0001999965688613511,
-      "loss": 46.0,
-      "step": 34498
-    },
-    {
-      "epoch": 2.6376894699619626,
-      "grad_norm": 0.001771169132553041,
-      "learning_rate": 0.00019999656866237354,
-      "loss": 46.0,
-      "step": 34499
-    },
-    {
-      "epoch": 2.6377659269453524,
-      "grad_norm": 0.005528313107788563,
-      "learning_rate": 0.00019999656846339024,
-      "loss": 46.0,
-      "step": 34500
-    },
-    {
-      "epoch": 2.637842383928742,
-      "grad_norm": 0.0035174351651221514,
-      "learning_rate": 0.00019999656826440117,
-      "loss": 46.0,
-      "step": 34501
-    },
-    {
-      "epoch": 2.637918840912132,
-      "grad_norm": 0.001308982609771192,
-      "learning_rate": 0.0001999965680654063,
-      "loss": 46.0,
-      "step": 34502
-    },
-    {
-      "epoch": 2.6379952978955217,
-      "grad_norm": 0.00360392895527184,
-      "learning_rate": 0.00019999656786640567,
-      "loss": 46.0,
-      "step": 34503
-    },
-    {
-      "epoch": 2.6380717548789114,
-      "grad_norm": 0.001978496555238962,
-      "learning_rate": 0.00019999656766739928,
-      "loss": 46.0,
-      "step": 34504
-    },
-    {
-      "epoch": 2.638148211862301,
-      "grad_norm": 0.0020974003709852695,
-      "learning_rate": 0.00019999656746838714,
-      "loss": 46.0,
-      "step": 34505
-    },
-    {
-      "epoch": 2.638224668845691,
-      "grad_norm": 0.0007851683767512441,
-      "learning_rate": 0.0001999965672693692,
-      "loss": 46.0,
-      "step": 34506
-    },
-    {
-      "epoch": 2.6383011258290803,
-      "grad_norm": 0.0012631692225113511,
-      "learning_rate": 0.00019999656707034552,
-      "loss": 46.0,
-      "step": 34507
-    },
-    {
-      "epoch": 2.63837758281247,
-      "grad_norm": 0.0027504574973136187,
-      "learning_rate": 0.00019999656687131606,
-      "loss": 46.0,
-      "step": 34508
-    },
-    {
-      "epoch": 2.63845403979586,
-      "grad_norm": 0.0007870674598962069,
-      "learning_rate": 0.0001999965666722808,
-      "loss": 46.0,
-      "step": 34509
-    },
-    {
-      "epoch": 2.6385304967792496,
-      "grad_norm": 0.0020757061429321766,
-      "learning_rate": 0.0001999965664732398,
-      "loss": 46.0,
-      "step": 34510
-    },
-    {
-      "epoch": 2.6386069537626393,
-      "grad_norm": 0.0027561241295188665,
-      "learning_rate": 0.00019999656627419304,
-      "loss": 46.0,
-      "step": 34511
-    },
-    {
-      "epoch": 2.638683410746029,
-      "grad_norm": 0.007109851110726595,
-      "learning_rate": 0.00019999656607514048,
-      "loss": 46.0,
-      "step": 34512
-    },
-    {
-      "epoch": 2.638759867729419,
-      "grad_norm": 0.002134371316060424,
-      "learning_rate": 0.00019999656587608216,
-      "loss": 46.0,
-      "step": 34513
-    },
-    {
-      "epoch": 2.6388363247128086,
-      "grad_norm": 0.0007126507116481662,
-      "learning_rate": 0.00019999656567701809,
-      "loss": 46.0,
-      "step": 34514
-    },
-    {
-      "epoch": 2.638912781696198,
-      "grad_norm": 0.004940371494740248,
-      "learning_rate": 0.00019999656547794824,
-      "loss": 46.0,
-      "step": 34515
-    },
-    {
-      "epoch": 2.6389892386795877,
-      "grad_norm": 0.0021288772113621235,
-      "learning_rate": 0.0001999965652788726,
-      "loss": 46.0,
-      "step": 34516
-    },
-    {
-      "epoch": 2.6390656956629774,
-      "grad_norm": 0.0030443675350397825,
-      "learning_rate": 0.0001999965650797912,
-      "loss": 46.0,
-      "step": 34517
-    },
-    {
-      "epoch": 2.639142152646367,
-      "grad_norm": 0.00041789733222685754,
-      "learning_rate": 0.00019999656488070406,
-      "loss": 46.0,
-      "step": 34518
-    },
-    {
-      "epoch": 2.639218609629757,
-      "grad_norm": 0.0013990928418934345,
-      "learning_rate": 0.00019999656468161112,
-      "loss": 46.0,
-      "step": 34519
-    },
-    {
-      "epoch": 2.6392950666131467,
-      "grad_norm": 0.004180360585451126,
-      "learning_rate": 0.00019999656448251244,
-      "loss": 46.0,
-      "step": 34520
-    },
-    {
-      "epoch": 2.6393715235965365,
-      "grad_norm": 0.003556195180863142,
-      "learning_rate": 0.00019999656428340795,
-      "loss": 46.0,
-      "step": 34521
-    },
-    {
-      "epoch": 2.6394479805799262,
-      "grad_norm": 0.0025172897148877382,
-      "learning_rate": 0.00019999656408429772,
-      "loss": 46.0,
-      "step": 34522
-    },
-    {
-      "epoch": 2.639524437563316,
-      "grad_norm": 0.007746831979602575,
-      "learning_rate": 0.0001999965638851817,
-      "loss": 46.0,
-      "step": 34523
-    },
-    {
-      "epoch": 2.6396008945467058,
-      "grad_norm": 0.0015589401591569185,
-      "learning_rate": 0.00019999656368605993,
-      "loss": 46.0,
-      "step": 34524
-    },
-    {
-      "epoch": 2.6396773515300955,
-      "grad_norm": 0.0030775293707847595,
-      "learning_rate": 0.00019999656348693238,
-      "loss": 46.0,
-      "step": 34525
-    },
-    {
-      "epoch": 2.6397538085134853,
-      "grad_norm": 0.0007467251853086054,
-      "learning_rate": 0.00019999656328779908,
-      "loss": 46.0,
-      "step": 34526
-    },
-    {
-      "epoch": 2.639830265496875,
-      "grad_norm": 0.0019878584425896406,
-      "learning_rate": 0.00019999656308865998,
-      "loss": 46.0,
-      "step": 34527
-    },
-    {
-      "epoch": 2.639906722480265,
-      "grad_norm": 0.003020251402631402,
-      "learning_rate": 0.00019999656288951514,
-      "loss": 46.0,
-      "step": 34528
-    },
-    {
-      "epoch": 2.639983179463654,
-      "grad_norm": 0.0054749296978116035,
-      "learning_rate": 0.00019999656269036452,
-      "loss": 46.0,
-      "step": 34529
-    },
-    {
-      "epoch": 2.640059636447044,
-      "grad_norm": 0.0028721713460981846,
-      "learning_rate": 0.00019999656249120813,
-      "loss": 46.0,
-      "step": 34530
-    },
-    {
-      "epoch": 2.6401360934304337,
-      "grad_norm": 0.0021708596032112837,
-      "learning_rate": 0.000199996562292046,
-      "loss": 46.0,
-      "step": 34531
-    },
-    {
-      "epoch": 2.6402125504138234,
-      "grad_norm": 0.0022460289765149355,
-      "learning_rate": 0.00019999656209287805,
-      "loss": 46.0,
-      "step": 34532
-    },
-    {
-      "epoch": 2.640289007397213,
-      "grad_norm": 0.0009714094339869916,
-      "learning_rate": 0.00019999656189370434,
-      "loss": 46.0,
-      "step": 34533
-    },
-    {
-      "epoch": 2.640365464380603,
-      "grad_norm": 0.0015169760445132852,
-      "learning_rate": 0.00019999656169452488,
-      "loss": 46.0,
-      "step": 34534
-    },
-    {
-      "epoch": 2.6404419213639927,
-      "grad_norm": 0.0034266652073711157,
-      "learning_rate": 0.00019999656149533962,
-      "loss": 46.0,
-      "step": 34535
-    },
-    {
-      "epoch": 2.6405183783473825,
-      "grad_norm": 0.0005773219163529575,
-      "learning_rate": 0.00019999656129614864,
-      "loss": 46.0,
-      "step": 34536
-    },
-    {
-      "epoch": 2.6405948353307718,
-      "grad_norm": 0.0011426068376749754,
-      "learning_rate": 0.00019999656109695186,
-      "loss": 46.0,
-      "step": 34537
-    },
-    {
-      "epoch": 2.6406712923141615,
-      "grad_norm": 0.001394849386997521,
-      "learning_rate": 0.0001999965608977493,
-      "loss": 46.0,
-      "step": 34538
-    },
-    {
-      "epoch": 2.6407477492975513,
-      "grad_norm": 0.0030240879859775305,
-      "learning_rate": 0.000199996560698541,
-      "loss": 46.0,
-      "step": 34539
-    },
-    {
-      "epoch": 2.640824206280941,
-      "grad_norm": 0.0009540761238895357,
-      "learning_rate": 0.0001999965604993269,
-      "loss": 46.0,
-      "step": 34540
-    },
-    {
-      "epoch": 2.640900663264331,
-      "grad_norm": 0.0013019067700952291,
-      "learning_rate": 0.00019999656030010704,
-      "loss": 46.0,
-      "step": 34541
-    },
-    {
-      "epoch": 2.6409771202477206,
-      "grad_norm": 0.001934366999194026,
-      "learning_rate": 0.00019999656010088145,
-      "loss": 46.0,
-      "step": 34542
-    },
-    {
-      "epoch": 2.6410535772311103,
-      "grad_norm": 0.006734516471624374,
-      "learning_rate": 0.00019999655990165006,
-      "loss": 46.0,
-      "step": 34543
-    },
-    {
-      "epoch": 2.6411300342145,
-      "grad_norm": 0.0030596167780458927,
-      "learning_rate": 0.0001999965597024129,
-      "loss": 46.0,
-      "step": 34544
-    },
-    {
-      "epoch": 2.64120649119789,
-      "grad_norm": 0.0014932332560420036,
-      "learning_rate": 0.00019999655950316995,
-      "loss": 46.0,
-      "step": 34545
-    },
-    {
-      "epoch": 2.6412829481812796,
-      "grad_norm": 0.002858963096514344,
-      "learning_rate": 0.00019999655930392124,
-      "loss": 46.0,
-      "step": 34546
-    },
-    {
-      "epoch": 2.6413594051646694,
-      "grad_norm": 0.0037363243754953146,
-      "learning_rate": 0.00019999655910466678,
-      "loss": 46.0,
-      "step": 34547
-    },
-    {
-      "epoch": 2.641435862148059,
-      "grad_norm": 0.0030468739569187164,
-      "learning_rate": 0.00019999655890540655,
-      "loss": 46.0,
-      "step": 34548
-    },
-    {
-      "epoch": 2.641512319131449,
-      "grad_norm": 0.0016986342379823327,
-      "learning_rate": 0.00019999655870614054,
-      "loss": 46.0,
-      "step": 34549
-    },
-    {
-      "epoch": 2.6415887761148387,
-      "grad_norm": 0.002146833809092641,
-      "learning_rate": 0.0001999965585068688,
-      "loss": 46.0,
-      "step": 34550
-    },
-    {
-      "epoch": 2.641665233098228,
-      "grad_norm": 0.0015977276489138603,
-      "learning_rate": 0.0001999965583075912,
-      "loss": 46.0,
-      "step": 34551
-    },
-    {
-      "epoch": 2.6417416900816177,
-      "grad_norm": 0.0032917342614382505,
-      "learning_rate": 0.00019999655810830791,
-      "loss": 46.0,
-      "step": 34552
-    },
-    {
-      "epoch": 2.6418181470650075,
-      "grad_norm": 0.0008477296214550734,
-      "learning_rate": 0.00019999655790901884,
-      "loss": 46.0,
-      "step": 34553
-    },
-    {
-      "epoch": 2.6418946040483973,
-      "grad_norm": 0.0011758451582863927,
-      "learning_rate": 0.00019999655770972397,
-      "loss": 46.0,
-      "step": 34554
-    },
-    {
-      "epoch": 2.641971061031787,
-      "grad_norm": 0.008312544785439968,
-      "learning_rate": 0.00019999655751042335,
-      "loss": 46.0,
-      "step": 34555
-    },
-    {
-      "epoch": 2.642047518015177,
-      "grad_norm": 0.0012131272815167904,
-      "learning_rate": 0.00019999655731111696,
-      "loss": 46.0,
-      "step": 34556
-    },
-    {
-      "epoch": 2.6421239749985665,
-      "grad_norm": 0.0031499280594289303,
-      "learning_rate": 0.0001999965571118048,
-      "loss": 46.0,
-      "step": 34557
-    },
-    {
-      "epoch": 2.6422004319819563,
-      "grad_norm": 0.0006568707758560777,
-      "learning_rate": 0.00019999655691248686,
-      "loss": 46.0,
-      "step": 34558
-    },
-    {
-      "epoch": 2.6422768889653456,
-      "grad_norm": 0.0007174179772846401,
-      "learning_rate": 0.00019999655671316315,
-      "loss": 46.0,
-      "step": 34559
-    },
-    {
-      "epoch": 2.6423533459487354,
-      "grad_norm": 0.0008868909208104014,
-      "learning_rate": 0.00019999655651383371,
-      "loss": 46.0,
-      "step": 34560
-    },
-    {
-      "epoch": 2.642429802932125,
-      "grad_norm": 0.0007538720965385437,
-      "learning_rate": 0.00019999655631449846,
-      "loss": 46.0,
-      "step": 34561
-    },
-    {
-      "epoch": 2.642506259915515,
-      "grad_norm": 0.0009438938577659428,
-      "learning_rate": 0.00019999655611515745,
-      "loss": 46.0,
-      "step": 34562
-    },
-    {
-      "epoch": 2.6425827168989047,
-      "grad_norm": 0.0023029108997434378,
-      "learning_rate": 0.00019999655591581067,
-      "loss": 46.0,
-      "step": 34563
-    },
-    {
-      "epoch": 2.6426591738822944,
-      "grad_norm": 0.0010392619296908379,
-      "learning_rate": 0.00019999655571645812,
-      "loss": 46.0,
-      "step": 34564
-    },
-    {
-      "epoch": 2.642735630865684,
-      "grad_norm": 0.0024848429020494223,
-      "learning_rate": 0.00019999655551709982,
-      "loss": 46.0,
-      "step": 34565
-    },
-    {
-      "epoch": 2.642812087849074,
-      "grad_norm": 0.0009105200879275799,
-      "learning_rate": 0.00019999655531773573,
-      "loss": 46.0,
-      "step": 34566
-    },
-    {
-      "epoch": 2.6428885448324637,
-      "grad_norm": 0.0010038192849606276,
-      "learning_rate": 0.00019999655511836588,
-      "loss": 46.0,
-      "step": 34567
-    },
-    {
-      "epoch": 2.6429650018158535,
-      "grad_norm": 0.0014624958857893944,
-      "learning_rate": 0.00019999655491899026,
-      "loss": 46.0,
-      "step": 34568
-    },
-    {
-      "epoch": 2.6430414587992432,
-      "grad_norm": 0.00266293459571898,
-      "learning_rate": 0.00019999655471960884,
-      "loss": 46.0,
-      "step": 34569
-    },
-    {
-      "epoch": 2.643117915782633,
-      "grad_norm": 0.0011956710368394852,
-      "learning_rate": 0.0001999965545202217,
-      "loss": 46.0,
-      "step": 34570
-    },
-    {
-      "epoch": 2.6431943727660228,
-      "grad_norm": 0.001659775385633111,
-      "learning_rate": 0.00019999655432082877,
-      "loss": 46.0,
-      "step": 34571
-    },
-    {
-      "epoch": 2.643270829749412,
-      "grad_norm": 0.0020263863261789083,
-      "learning_rate": 0.00019999655412143006,
-      "loss": 46.0,
-      "step": 34572
-    },
-    {
-      "epoch": 2.643347286732802,
-      "grad_norm": 0.0014143850421532989,
-      "learning_rate": 0.0001999965539220256,
-      "loss": 46.0,
-      "step": 34573
-    },
-    {
-      "epoch": 2.6434237437161916,
-      "grad_norm": 0.000806483905762434,
-      "learning_rate": 0.00019999655372261537,
-      "loss": 46.0,
-      "step": 34574
-    },
-    {
-      "epoch": 2.6435002006995814,
-      "grad_norm": 0.002073373179882765,
-      "learning_rate": 0.00019999655352319934,
-      "loss": 46.0,
-      "step": 34575
-    },
-    {
-      "epoch": 2.643576657682971,
-      "grad_norm": 0.0022042703349143267,
-      "learning_rate": 0.00019999655332377759,
-      "loss": 46.0,
-      "step": 34576
-    },
-    {
-      "epoch": 2.643653114666361,
-      "grad_norm": 0.0010752958478406072,
-      "learning_rate": 0.00019999655312435003,
-      "loss": 46.0,
-      "step": 34577
-    },
-    {
-      "epoch": 2.6437295716497506,
-      "grad_norm": 0.0013884485233575106,
-      "learning_rate": 0.00019999655292491674,
-      "loss": 46.0,
-      "step": 34578
-    },
-    {
-      "epoch": 2.6438060286331404,
-      "grad_norm": 0.0021038968116045,
-      "learning_rate": 0.00019999655272547764,
-      "loss": 46.0,
-      "step": 34579
-    },
-    {
-      "epoch": 2.64388248561653,
-      "grad_norm": 0.004786514677107334,
-      "learning_rate": 0.00019999655252603277,
-      "loss": 46.0,
-      "step": 34580
-    },
-    {
-      "epoch": 2.6439589425999195,
-      "grad_norm": 0.002001231536269188,
-      "learning_rate": 0.00019999655232658215,
-      "loss": 46.0,
-      "step": 34581
-    },
-    {
-      "epoch": 2.6440353995833092,
-      "grad_norm": 0.007867136038839817,
-      "learning_rate": 0.00019999655212712576,
-      "loss": 46.0,
-      "step": 34582
-    },
-    {
-      "epoch": 2.644111856566699,
-      "grad_norm": 0.0010004050564020872,
-      "learning_rate": 0.0001999965519276636,
-      "loss": 46.0,
-      "step": 34583
-    },
-    {
-      "epoch": 2.6441883135500888,
-      "grad_norm": 0.003883182071149349,
-      "learning_rate": 0.00019999655172819566,
-      "loss": 46.0,
-      "step": 34584
-    },
-    {
-      "epoch": 2.6442647705334785,
-      "grad_norm": 0.0037523575592786074,
-      "learning_rate": 0.00019999655152872197,
-      "loss": 46.0,
-      "step": 34585
-    },
-    {
-      "epoch": 2.6443412275168683,
-      "grad_norm": 0.0014262836193665862,
-      "learning_rate": 0.00019999655132924251,
-      "loss": 46.0,
-      "step": 34586
-    },
-    {
-      "epoch": 2.644417684500258,
-      "grad_norm": 0.0019873965065926313,
-      "learning_rate": 0.00019999655112975726,
-      "loss": 46.0,
-      "step": 34587
-    },
-    {
-      "epoch": 2.644494141483648,
-      "grad_norm": 0.0050540766678750515,
-      "learning_rate": 0.00019999655093026625,
-      "loss": 46.0,
-      "step": 34588
-    },
-    {
-      "epoch": 2.6445705984670376,
-      "grad_norm": 0.0022524537052959204,
-      "learning_rate": 0.00019999655073076947,
-      "loss": 46.0,
-      "step": 34589
-    },
-    {
-      "epoch": 2.6446470554504273,
-      "grad_norm": 0.013481991365551949,
-      "learning_rate": 0.00019999655053126692,
-      "loss": 46.0,
-      "step": 34590
-    },
-    {
-      "epoch": 2.644723512433817,
-      "grad_norm": 0.0007073396118357778,
-      "learning_rate": 0.00019999655033175863,
-      "loss": 46.0,
-      "step": 34591
-    },
-    {
-      "epoch": 2.644799969417207,
-      "grad_norm": 0.0007290412322618067,
-      "learning_rate": 0.00019999655013224456,
-      "loss": 46.0,
-      "step": 34592
-    },
-    {
-      "epoch": 2.6448764264005966,
-      "grad_norm": 0.0034683244302868843,
-      "learning_rate": 0.00019999654993272468,
-      "loss": 46.0,
-      "step": 34593
-    },
-    {
-      "epoch": 2.644952883383986,
-      "grad_norm": 0.00046749861212447286,
-      "learning_rate": 0.00019999654973319907,
-      "loss": 46.0,
-      "step": 34594
-    },
-    {
-      "epoch": 2.6450293403673757,
-      "grad_norm": 0.0016555659240111709,
-      "learning_rate": 0.00019999654953366768,
-      "loss": 46.0,
-      "step": 34595
-    },
-    {
-      "epoch": 2.6451057973507655,
-      "grad_norm": 0.0027320387307554483,
-      "learning_rate": 0.0001999965493341305,
-      "loss": 46.0,
-      "step": 34596
-    },
-    {
-      "epoch": 2.645182254334155,
-      "grad_norm": 0.02012760005891323,
-      "learning_rate": 0.00019999654913458757,
-      "loss": 46.0,
-      "step": 34597
-    },
-    {
-      "epoch": 2.645258711317545,
-      "grad_norm": 0.001547559048049152,
-      "learning_rate": 0.00019999654893503886,
-      "loss": 46.0,
-      "step": 34598
-    },
-    {
-      "epoch": 2.6453351683009347,
-      "grad_norm": 0.001451008371077478,
-      "learning_rate": 0.0001999965487354844,
-      "loss": 46.0,
-      "step": 34599
-    },
-    {
-      "epoch": 2.6454116252843245,
-      "grad_norm": 0.0030293797608464956,
-      "learning_rate": 0.00019999654853592415,
-      "loss": 46.0,
-      "step": 34600
-    },
-    {
-      "epoch": 2.6454880822677143,
-      "grad_norm": 0.0010853284038603306,
-      "learning_rate": 0.00019999654833635815,
-      "loss": 46.0,
-      "step": 34601
-    },
-    {
-      "epoch": 2.6455645392511036,
-      "grad_norm": 0.0025759569834917784,
-      "learning_rate": 0.0001999965481367864,
-      "loss": 46.0,
-      "step": 34602
-    },
-    {
-      "epoch": 2.6456409962344933,
-      "grad_norm": 0.0020716486033052206,
-      "learning_rate": 0.00019999654793720884,
-      "loss": 46.0,
-      "step": 34603
-    },
-    {
-      "epoch": 2.645717453217883,
-      "grad_norm": 0.001761682447977364,
-      "learning_rate": 0.0001999965477376255,
-      "loss": 46.0,
-      "step": 34604
-    },
-    {
-      "epoch": 2.645793910201273,
-      "grad_norm": 0.002164250472560525,
-      "learning_rate": 0.00019999654753803642,
-      "loss": 46.0,
-      "step": 34605
-    },
-    {
-      "epoch": 2.6458703671846626,
-      "grad_norm": 0.0022773926611989737,
-      "learning_rate": 0.00019999654733844158,
-      "loss": 46.0,
-      "step": 34606
-    },
-    {
-      "epoch": 2.6459468241680524,
-      "grad_norm": 0.0038353088311851025,
-      "learning_rate": 0.00019999654713884096,
-      "loss": 46.0,
-      "step": 34607
-    },
-    {
-      "epoch": 2.646023281151442,
-      "grad_norm": 0.0009662198135629296,
-      "learning_rate": 0.00019999654693923454,
-      "loss": 46.0,
-      "step": 34608
-    },
-    {
-      "epoch": 2.646099738134832,
-      "grad_norm": 0.0008317287429235876,
-      "learning_rate": 0.0001999965467396224,
-      "loss": 46.0,
-      "step": 34609
-    },
-    {
-      "epoch": 2.6461761951182217,
-      "grad_norm": 0.0014157487312331796,
-      "learning_rate": 0.00019999654654000447,
-      "loss": 46.0,
-      "step": 34610
-    },
-    {
-      "epoch": 2.6462526521016114,
-      "grad_norm": 0.00263407826423645,
-      "learning_rate": 0.00019999654634038076,
-      "loss": 46.0,
-      "step": 34611
-    },
-    {
-      "epoch": 2.646329109085001,
-      "grad_norm": 0.001302023883908987,
-      "learning_rate": 0.0001999965461407513,
-      "loss": 46.0,
-      "step": 34612
-    },
-    {
-      "epoch": 2.646405566068391,
-      "grad_norm": 0.0025027331430464983,
-      "learning_rate": 0.00019999654594111605,
-      "loss": 46.0,
-      "step": 34613
-    },
-    {
-      "epoch": 2.6464820230517807,
-      "grad_norm": 0.0037913252599537373,
-      "learning_rate": 0.00019999654574147504,
-      "loss": 46.0,
-      "step": 34614
-    },
-    {
-      "epoch": 2.6465584800351705,
-      "grad_norm": 0.0040735891088843346,
-      "learning_rate": 0.00019999654554182827,
-      "loss": 46.0,
-      "step": 34615
-    },
-    {
-      "epoch": 2.64663493701856,
-      "grad_norm": 0.0013186606811359525,
-      "learning_rate": 0.00019999654534217572,
-      "loss": 46.0,
-      "step": 34616
-    },
-    {
-      "epoch": 2.6467113940019495,
-      "grad_norm": 0.005385016091167927,
-      "learning_rate": 0.0001999965451425174,
-      "loss": 46.0,
-      "step": 34617
-    },
-    {
-      "epoch": 2.6467878509853393,
-      "grad_norm": 0.001892649452202022,
-      "learning_rate": 0.00019999654494285332,
-      "loss": 46.0,
-      "step": 34618
-    },
-    {
-      "epoch": 2.646864307968729,
-      "grad_norm": 0.0024929847568273544,
-      "learning_rate": 0.00019999654474318348,
-      "loss": 46.0,
-      "step": 34619
-    },
-    {
-      "epoch": 2.646940764952119,
-      "grad_norm": 0.001314322929829359,
-      "learning_rate": 0.00019999654454350783,
-      "loss": 46.0,
-      "step": 34620
-    },
-    {
-      "epoch": 2.6470172219355086,
-      "grad_norm": 0.001464893575757742,
-      "learning_rate": 0.00019999654434382644,
-      "loss": 46.0,
-      "step": 34621
-    },
-    {
-      "epoch": 2.6470936789188984,
-      "grad_norm": 0.0019550009164959192,
-      "learning_rate": 0.00019999654414413928,
-      "loss": 46.0,
-      "step": 34622
-    },
-    {
-      "epoch": 2.647170135902288,
-      "grad_norm": 0.004785238299518824,
-      "learning_rate": 0.00019999654394444637,
-      "loss": 46.0,
-      "step": 34623
-    },
-    {
-      "epoch": 2.6472465928856774,
-      "grad_norm": 0.0021834736689925194,
-      "learning_rate": 0.00019999654374474766,
-      "loss": 46.0,
-      "step": 34624
-    },
-    {
-      "epoch": 2.647323049869067,
-      "grad_norm": 0.002081941580399871,
-      "learning_rate": 0.00019999654354504318,
-      "loss": 46.0,
-      "step": 34625
-    },
-    {
-      "epoch": 2.647399506852457,
-      "grad_norm": 0.0018951010424643755,
-      "learning_rate": 0.00019999654334533292,
-      "loss": 46.0,
-      "step": 34626
-    },
-    {
-      "epoch": 2.6474759638358467,
-      "grad_norm": 0.0009608858381398022,
-      "learning_rate": 0.00019999654314561694,
-      "loss": 46.0,
-      "step": 34627
-    },
-    {
-      "epoch": 2.6475524208192365,
-      "grad_norm": 0.0017959446413442492,
-      "learning_rate": 0.00019999654294589517,
-      "loss": 46.0,
-      "step": 34628
-    },
-    {
-      "epoch": 2.6476288778026262,
-      "grad_norm": 0.0013378075091168284,
-      "learning_rate": 0.0001999965427461676,
-      "loss": 46.0,
-      "step": 34629
-    },
-    {
-      "epoch": 2.647705334786016,
-      "grad_norm": 0.0006940686726011336,
-      "learning_rate": 0.0001999965425464343,
-      "loss": 46.0,
-      "step": 34630
-    },
-    {
-      "epoch": 2.6477817917694058,
-      "grad_norm": 0.0012696638004854321,
-      "learning_rate": 0.0001999965423466952,
-      "loss": 46.0,
-      "step": 34631
-    },
-    {
-      "epoch": 2.6478582487527955,
-      "grad_norm": 0.0016210098983719945,
-      "learning_rate": 0.00019999654214695033,
-      "loss": 46.0,
-      "step": 34632
-    },
-    {
-      "epoch": 2.6479347057361853,
-      "grad_norm": 0.003102767514064908,
-      "learning_rate": 0.0001999965419471997,
-      "loss": 46.0,
-      "step": 34633
-    },
-    {
-      "epoch": 2.648011162719575,
-      "grad_norm": 0.0026267266366630793,
-      "learning_rate": 0.00019999654174744332,
-      "loss": 46.0,
-      "step": 34634
-    },
-    {
-      "epoch": 2.648087619702965,
-      "grad_norm": 0.0010340509470552206,
-      "learning_rate": 0.00019999654154768118,
-      "loss": 46.0,
-      "step": 34635
-    },
-    {
-      "epoch": 2.6481640766863546,
-      "grad_norm": 0.0023651120718568563,
-      "learning_rate": 0.00019999654134791325,
-      "loss": 46.0,
-      "step": 34636
-    },
-    {
-      "epoch": 2.6482405336697443,
-      "grad_norm": 0.0007886143866926432,
-      "learning_rate": 0.00019999654114813954,
-      "loss": 46.0,
-      "step": 34637
-    },
-    {
-      "epoch": 2.6483169906531336,
-      "grad_norm": 0.004997018724679947,
-      "learning_rate": 0.00019999654094836005,
-      "loss": 46.0,
-      "step": 34638
-    },
-    {
-      "epoch": 2.6483934476365234,
-      "grad_norm": 0.002043304732069373,
-      "learning_rate": 0.00019999654074857482,
-      "loss": 46.0,
-      "step": 34639
-    },
-    {
-      "epoch": 2.648469904619913,
-      "grad_norm": 0.003641656832769513,
-      "learning_rate": 0.0001999965405487838,
-      "loss": 46.0,
-      "step": 34640
-    },
-    {
-      "epoch": 2.648546361603303,
-      "grad_norm": 0.00275347288697958,
-      "learning_rate": 0.00019999654034898705,
-      "loss": 46.0,
-      "step": 34641
-    },
-    {
-      "epoch": 2.6486228185866927,
-      "grad_norm": 0.0019136063056066632,
-      "learning_rate": 0.00019999654014918447,
-      "loss": 46.0,
-      "step": 34642
-    },
-    {
-      "epoch": 2.6486992755700824,
-      "grad_norm": 0.0008162970771081746,
-      "learning_rate": 0.00019999653994937617,
-      "loss": 46.0,
-      "step": 34643
-    },
-    {
-      "epoch": 2.648775732553472,
-      "grad_norm": 0.0010963745880872011,
-      "learning_rate": 0.0001999965397495621,
-      "loss": 46.0,
-      "step": 34644
-    },
-    {
-      "epoch": 2.648852189536862,
-      "grad_norm": 0.0033200655598193407,
-      "learning_rate": 0.00019999653954974223,
-      "loss": 46.0,
-      "step": 34645
-    },
-    {
-      "epoch": 2.6489286465202513,
-      "grad_norm": 0.0012055494589731097,
-      "learning_rate": 0.0001999965393499166,
-      "loss": 46.0,
-      "step": 34646
-    },
-    {
-      "epoch": 2.649005103503641,
-      "grad_norm": 0.0018360440153628588,
-      "learning_rate": 0.00019999653915008523,
-      "loss": 46.0,
-      "step": 34647
-    },
-    {
-      "epoch": 2.649081560487031,
-      "grad_norm": 0.005940403789281845,
-      "learning_rate": 0.00019999653895024806,
-      "loss": 46.0,
-      "step": 34648
-    },
-    {
-      "epoch": 2.6491580174704206,
-      "grad_norm": 0.001664777984842658,
-      "learning_rate": 0.00019999653875040513,
-      "loss": 46.0,
-      "step": 34649
-    },
-    {
-      "epoch": 2.6492344744538103,
-      "grad_norm": 0.00216398062184453,
-      "learning_rate": 0.0001999965385505564,
-      "loss": 46.0,
-      "step": 34650
-    },
-    {
-      "epoch": 2.6493109314372,
-      "grad_norm": 0.0035663945600390434,
-      "learning_rate": 0.00019999653835070196,
-      "loss": 46.0,
-      "step": 34651
-    },
-    {
-      "epoch": 2.64938738842059,
-      "grad_norm": 0.001355396001599729,
-      "learning_rate": 0.0001999965381508417,
-      "loss": 46.0,
-      "step": 34652
-    },
-    {
-      "epoch": 2.6494638454039796,
-      "grad_norm": 0.001657457323744893,
-      "learning_rate": 0.0001999965379509757,
-      "loss": 46.0,
-      "step": 34653
-    },
-    {
-      "epoch": 2.6495403023873694,
-      "grad_norm": 0.004674006253480911,
-      "learning_rate": 0.00019999653775110393,
-      "loss": 46.0,
-      "step": 34654
-    },
-    {
-      "epoch": 2.649616759370759,
-      "grad_norm": 0.0028455324936658144,
-      "learning_rate": 0.00019999653755122635,
-      "loss": 46.0,
-      "step": 34655
-    },
-    {
-      "epoch": 2.649693216354149,
-      "grad_norm": 0.0024973968975245953,
-      "learning_rate": 0.00019999653735134306,
-      "loss": 46.0,
-      "step": 34656
-    },
-    {
-      "epoch": 2.6497696733375387,
-      "grad_norm": 0.00198940047994256,
-      "learning_rate": 0.00019999653715145396,
-      "loss": 46.0,
-      "step": 34657
-    },
-    {
-      "epoch": 2.6498461303209284,
-      "grad_norm": 0.0021798640955239534,
-      "learning_rate": 0.00019999653695155912,
-      "loss": 46.0,
-      "step": 34658
-    },
-    {
-      "epoch": 2.649922587304318,
-      "grad_norm": 0.0021034269593656063,
-      "learning_rate": 0.00019999653675165847,
-      "loss": 46.0,
-      "step": 34659
-    },
-    {
-      "epoch": 2.6499990442877075,
-      "grad_norm": 0.004055732395499945,
-      "learning_rate": 0.00019999653655175209,
-      "loss": 46.0,
-      "step": 34660
-    },
-    {
-      "epoch": 2.6500755012710973,
-      "grad_norm": 0.0025905249640345573,
-      "learning_rate": 0.00019999653635183992,
-      "loss": 46.0,
-      "step": 34661
-    },
-    {
-      "epoch": 2.650151958254487,
-      "grad_norm": 0.0036613019183278084,
-      "learning_rate": 0.000199996536151922,
-      "loss": 46.0,
-      "step": 34662
-    },
-    {
-      "epoch": 2.6502284152378768,
-      "grad_norm": 0.0006996187730692327,
-      "learning_rate": 0.00019999653595199828,
-      "loss": 46.0,
-      "step": 34663
-    },
-    {
-      "epoch": 2.6503048722212665,
-      "grad_norm": 0.0028038721065968275,
-      "learning_rate": 0.00019999653575206882,
-      "loss": 46.0,
-      "step": 34664
-    },
-    {
-      "epoch": 2.6503813292046563,
-      "grad_norm": 0.0009371620253659785,
-      "learning_rate": 0.0001999965355521336,
-      "loss": 46.0,
-      "step": 34665
-    },
-    {
-      "epoch": 2.650457786188046,
-      "grad_norm": 0.0011505559086799622,
-      "learning_rate": 0.00019999653535219256,
-      "loss": 46.0,
-      "step": 34666
-    },
-    {
-      "epoch": 2.650534243171436,
-      "grad_norm": 0.0015809711767360568,
-      "learning_rate": 0.0001999965351522458,
-      "loss": 46.0,
-      "step": 34667
-    },
-    {
-      "epoch": 2.650610700154825,
-      "grad_norm": 0.0022850402165204287,
-      "learning_rate": 0.00019999653495229324,
-      "loss": 46.0,
-      "step": 34668
-    },
-    {
-      "epoch": 2.650687157138215,
-      "grad_norm": 0.004635780584067106,
-      "learning_rate": 0.00019999653475233494,
-      "loss": 46.0,
-      "step": 34669
-    },
-    {
-      "epoch": 2.6507636141216047,
-      "grad_norm": 0.0013846515212208033,
-      "learning_rate": 0.00019999653455237085,
-      "loss": 46.0,
-      "step": 34670
-    },
-    {
-      "epoch": 2.6508400711049944,
-      "grad_norm": 0.0009411721839569509,
-      "learning_rate": 0.00019999653435240098,
-      "loss": 46.0,
-      "step": 34671
-    },
-    {
-      "epoch": 2.650916528088384,
-      "grad_norm": 0.0021812478080391884,
-      "learning_rate": 0.00019999653415242534,
-      "loss": 46.0,
-      "step": 34672
-    },
-    {
-      "epoch": 2.650992985071774,
-      "grad_norm": 0.006250261329114437,
-      "learning_rate": 0.00019999653395244397,
-      "loss": 46.0,
-      "step": 34673
-    },
-    {
-      "epoch": 2.6510694420551637,
-      "grad_norm": 0.0016414192505180836,
-      "learning_rate": 0.00019999653375245678,
-      "loss": 46.0,
-      "step": 34674
-    },
-    {
-      "epoch": 2.6511458990385535,
-      "grad_norm": 0.0011620079167187214,
-      "learning_rate": 0.00019999653355246385,
-      "loss": 46.0,
-      "step": 34675
-    },
-    {
-      "epoch": 2.6512223560219432,
-      "grad_norm": 0.0025420573074370623,
-      "learning_rate": 0.00019999653335246517,
-      "loss": 46.0,
-      "step": 34676
-    },
-    {
-      "epoch": 2.651298813005333,
-      "grad_norm": 0.0022182592656463385,
-      "learning_rate": 0.00019999653315246068,
-      "loss": 46.0,
-      "step": 34677
-    },
-    {
-      "epoch": 2.6513752699887227,
-      "grad_norm": 0.0012899975990876555,
-      "learning_rate": 0.00019999653295245046,
-      "loss": 46.0,
-      "step": 34678
-    },
-    {
-      "epoch": 2.6514517269721125,
-      "grad_norm": 0.0014714538119733334,
-      "learning_rate": 0.00019999653275243443,
-      "loss": 46.0,
-      "step": 34679
-    },
-    {
-      "epoch": 2.6515281839555023,
-      "grad_norm": 0.0020265597850084305,
-      "learning_rate": 0.00019999653255241265,
-      "loss": 46.0,
-      "step": 34680
-    },
-    {
-      "epoch": 2.651604640938892,
-      "grad_norm": 0.0014766695676371455,
-      "learning_rate": 0.0001999965323523851,
-      "loss": 46.0,
-      "step": 34681
-    },
-    {
-      "epoch": 2.6516810979222813,
-      "grad_norm": 0.0008644903427921236,
-      "learning_rate": 0.0001999965321523518,
-      "loss": 46.0,
-      "step": 34682
-    },
-    {
-      "epoch": 2.651757554905671,
-      "grad_norm": 0.0012086700880900025,
-      "learning_rate": 0.0001999965319523127,
-      "loss": 46.0,
-      "step": 34683
-    },
-    {
-      "epoch": 2.651834011889061,
-      "grad_norm": 0.0030023485887795687,
-      "learning_rate": 0.00019999653175226784,
-      "loss": 46.0,
-      "step": 34684
-    },
-    {
-      "epoch": 2.6519104688724506,
-      "grad_norm": 0.004265329800546169,
-      "learning_rate": 0.00019999653155221723,
-      "loss": 46.0,
-      "step": 34685
-    },
-    {
-      "epoch": 2.6519869258558404,
-      "grad_norm": 0.0036564667243510485,
-      "learning_rate": 0.00019999653135216084,
-      "loss": 46.0,
-      "step": 34686
-    },
-    {
-      "epoch": 2.65206338283923,
-      "grad_norm": 0.0026469770818948746,
-      "learning_rate": 0.00019999653115209868,
-      "loss": 46.0,
-      "step": 34687
-    },
-    {
-      "epoch": 2.65213983982262,
-      "grad_norm": 0.0012685690307989717,
-      "learning_rate": 0.00019999653095203071,
-      "loss": 46.0,
-      "step": 34688
-    },
-    {
-      "epoch": 2.6522162968060097,
-      "grad_norm": 0.0019143827958032489,
-      "learning_rate": 0.00019999653075195703,
-      "loss": 46.0,
-      "step": 34689
-    },
-    {
-      "epoch": 2.652292753789399,
-      "grad_norm": 0.002410567831248045,
-      "learning_rate": 0.00019999653055187758,
-      "loss": 46.0,
-      "step": 34690
-    },
-    {
-      "epoch": 2.6523692107727888,
-      "grad_norm": 0.0011470012832432985,
-      "learning_rate": 0.00019999653035179232,
-      "loss": 46.0,
-      "step": 34691
-    },
-    {
-      "epoch": 2.6524456677561785,
-      "grad_norm": 0.0011017803335562348,
-      "learning_rate": 0.00019999653015170132,
-      "loss": 46.0,
-      "step": 34692
-    },
-    {
-      "epoch": 2.6525221247395683,
-      "grad_norm": 0.004031628370285034,
-      "learning_rate": 0.00019999652995160455,
-      "loss": 46.0,
-      "step": 34693
-    },
-    {
-      "epoch": 2.652598581722958,
-      "grad_norm": 0.004409824963659048,
-      "learning_rate": 0.00019999652975150197,
-      "loss": 46.0,
-      "step": 34694
-    },
-    {
-      "epoch": 2.652675038706348,
-      "grad_norm": 0.0014532955829054117,
-      "learning_rate": 0.00019999652955139365,
-      "loss": 46.0,
-      "step": 34695
-    },
-    {
-      "epoch": 2.6527514956897376,
-      "grad_norm": 0.0015394381480291486,
-      "learning_rate": 0.00019999652935127958,
-      "loss": 46.0,
-      "step": 34696
-    },
-    {
-      "epoch": 2.6528279526731273,
-      "grad_norm": 0.0015390662010759115,
-      "learning_rate": 0.0001999965291511597,
-      "loss": 46.0,
-      "step": 34697
-    },
-    {
-      "epoch": 2.652904409656517,
-      "grad_norm": 0.0018092122627422214,
-      "learning_rate": 0.0001999965289510341,
-      "loss": 46.0,
-      "step": 34698
-    },
-    {
-      "epoch": 2.652980866639907,
-      "grad_norm": 0.001475409371778369,
-      "learning_rate": 0.00019999652875090268,
-      "loss": 46.0,
-      "step": 34699
-    },
-    {
-      "epoch": 2.6530573236232966,
-      "grad_norm": 0.001401489251293242,
-      "learning_rate": 0.00019999652855076555,
-      "loss": 46.0,
-      "step": 34700
-    },
-    {
-      "epoch": 2.6531337806066864,
-      "grad_norm": 0.0014195594703778625,
-      "learning_rate": 0.0001999965283506226,
-      "loss": 46.0,
-      "step": 34701
-    },
-    {
-      "epoch": 2.653210237590076,
-      "grad_norm": 0.0015073702670633793,
-      "learning_rate": 0.0001999965281504739,
-      "loss": 46.0,
-      "step": 34702
-    },
-    {
-      "epoch": 2.6532866945734654,
-      "grad_norm": 0.015037347562611103,
-      "learning_rate": 0.00019999652795031942,
-      "loss": 46.0,
-      "step": 34703
-    },
-    {
-      "epoch": 2.653363151556855,
-      "grad_norm": 0.0013826737413182855,
-      "learning_rate": 0.00019999652775015917,
-      "loss": 46.0,
-      "step": 34704
-    },
-    {
-      "epoch": 2.653439608540245,
-      "grad_norm": 0.0038377426099032164,
-      "learning_rate": 0.0001999965275499932,
-      "loss": 46.0,
-      "step": 34705
-    },
-    {
-      "epoch": 2.6535160655236347,
-      "grad_norm": 0.0030768071301281452,
-      "learning_rate": 0.0001999965273498214,
-      "loss": 46.0,
-      "step": 34706
-    },
-    {
-      "epoch": 2.6535925225070245,
-      "grad_norm": 0.0010840905597433448,
-      "learning_rate": 0.00019999652714964384,
-      "loss": 46.0,
-      "step": 34707
-    },
-    {
-      "epoch": 2.6536689794904142,
-      "grad_norm": 0.0008362262160517275,
-      "learning_rate": 0.00019999652694946055,
-      "loss": 46.0,
-      "step": 34708
-    },
-    {
-      "epoch": 2.653745436473804,
-      "grad_norm": 0.0028353342786431313,
-      "learning_rate": 0.00019999652674927143,
-      "loss": 46.0,
-      "step": 34709
-    },
-    {
-      "epoch": 2.6538218934571938,
-      "grad_norm": 0.0010955880861729383,
-      "learning_rate": 0.00019999652654907659,
-      "loss": 46.0,
-      "step": 34710
-    },
-    {
-      "epoch": 2.6538983504405835,
-      "grad_norm": 0.002578646643087268,
-      "learning_rate": 0.00019999652634887594,
-      "loss": 46.0,
-      "step": 34711
-    },
-    {
-      "epoch": 2.653974807423973,
-      "grad_norm": 0.0019248604075983167,
-      "learning_rate": 0.00019999652614866956,
-      "loss": 46.0,
-      "step": 34712
-    },
-    {
-      "epoch": 2.6540512644073626,
-      "grad_norm": 0.0011777599574998021,
-      "learning_rate": 0.0001999965259484574,
-      "loss": 46.0,
-      "step": 34713
-    },
-    {
-      "epoch": 2.6541277213907524,
-      "grad_norm": 0.0010027263779193163,
-      "learning_rate": 0.00019999652574823946,
-      "loss": 46.0,
-      "step": 34714
-    },
-    {
-      "epoch": 2.654204178374142,
-      "grad_norm": 0.0019008881645277143,
-      "learning_rate": 0.00019999652554801575,
-      "loss": 46.0,
-      "step": 34715
-    },
-    {
-      "epoch": 2.654280635357532,
-      "grad_norm": 0.001758950762450695,
-      "learning_rate": 0.00019999652534778627,
-      "loss": 46.0,
-      "step": 34716
-    },
-    {
-      "epoch": 2.6543570923409217,
-      "grad_norm": 0.002618011087179184,
-      "learning_rate": 0.00019999652514755102,
-      "loss": 46.0,
-      "step": 34717
-    },
-    {
-      "epoch": 2.6544335493243114,
-      "grad_norm": 0.0020383228547871113,
-      "learning_rate": 0.00019999652494731004,
-      "loss": 46.0,
-      "step": 34718
-    },
-    {
-      "epoch": 2.654510006307701,
-      "grad_norm": 0.003801194950938225,
-      "learning_rate": 0.00019999652474706324,
-      "loss": 46.0,
-      "step": 34719
-    },
-    {
-      "epoch": 2.654586463291091,
-      "grad_norm": 0.00209213700145483,
-      "learning_rate": 0.0001999965245468107,
-      "loss": 46.0,
-      "step": 34720
-    },
-    {
-      "epoch": 2.6546629202744807,
-      "grad_norm": 0.0017593438969925046,
-      "learning_rate": 0.00019999652434655237,
-      "loss": 46.0,
-      "step": 34721
-    },
-    {
-      "epoch": 2.6547393772578705,
-      "grad_norm": 0.001032951520755887,
-      "learning_rate": 0.0001999965241462883,
-      "loss": 46.0,
-      "step": 34722
-    },
-    {
-      "epoch": 2.65481583424126,
-      "grad_norm": 0.005916541907936335,
-      "learning_rate": 0.00019999652394601844,
-      "loss": 46.0,
-      "step": 34723
-    },
-    {
-      "epoch": 2.65489229122465,
-      "grad_norm": 0.0013478639302775264,
-      "learning_rate": 0.0001999965237457428,
-      "loss": 46.0,
-      "step": 34724
-    },
-    {
-      "epoch": 2.6549687482080393,
-      "grad_norm": 0.002013017423450947,
-      "learning_rate": 0.0001999965235454614,
-      "loss": 46.0,
-      "step": 34725
-    },
-    {
-      "epoch": 2.655045205191429,
-      "grad_norm": 0.0032584015280008316,
-      "learning_rate": 0.00019999652334517425,
-      "loss": 46.0,
-      "step": 34726
-    },
-    {
-      "epoch": 2.655121662174819,
-      "grad_norm": 0.003934497945010662,
-      "learning_rate": 0.0001999965231448813,
-      "loss": 46.0,
-      "step": 34727
-    },
-    {
-      "epoch": 2.6551981191582086,
-      "grad_norm": 0.0013849419774487615,
-      "learning_rate": 0.0001999965229445826,
-      "loss": 46.0,
-      "step": 34728
-    },
-    {
-      "epoch": 2.6552745761415983,
-      "grad_norm": 0.0014424193650484085,
-      "learning_rate": 0.00019999652274427815,
-      "loss": 46.0,
-      "step": 34729
-    },
-    {
-      "epoch": 2.655351033124988,
-      "grad_norm": 0.0008990371134132147,
-      "learning_rate": 0.0001999965225439679,
-      "loss": 46.0,
-      "step": 34730
-    },
-    {
-      "epoch": 2.655427490108378,
-      "grad_norm": 0.0011558630503714085,
-      "learning_rate": 0.0001999965223436519,
-      "loss": 46.0,
-      "step": 34731
-    },
-    {
-      "epoch": 2.6555039470917676,
-      "grad_norm": 0.001537567120976746,
-      "learning_rate": 0.00019999652214333012,
-      "loss": 46.0,
-      "step": 34732
-    },
-    {
-      "epoch": 2.655580404075157,
-      "grad_norm": 0.000707258703187108,
-      "learning_rate": 0.00019999652194300255,
-      "loss": 46.0,
-      "step": 34733
-    },
-    {
-      "epoch": 2.6556568610585467,
-      "grad_norm": 0.005646726116538048,
-      "learning_rate": 0.00019999652174266922,
-      "loss": 46.0,
-      "step": 34734
-    },
-    {
-      "epoch": 2.6557333180419365,
-      "grad_norm": 0.003111784579232335,
-      "learning_rate": 0.00019999652154233016,
-      "loss": 46.0,
-      "step": 34735
-    },
-    {
-      "epoch": 2.6558097750253262,
-      "grad_norm": 0.0026187580078840256,
-      "learning_rate": 0.0001999965213419853,
-      "loss": 46.0,
-      "step": 34736
-    },
-    {
-      "epoch": 2.655886232008716,
-      "grad_norm": 0.0016155462944880128,
-      "learning_rate": 0.00019999652114163468,
-      "loss": 46.0,
-      "step": 34737
-    },
-    {
-      "epoch": 2.6559626889921057,
-      "grad_norm": 0.001129030715674162,
-      "learning_rate": 0.0001999965209412783,
-      "loss": 46.0,
-      "step": 34738
-    },
-    {
-      "epoch": 2.6560391459754955,
-      "grad_norm": 0.001172374584712088,
-      "learning_rate": 0.0001999965207409161,
-      "loss": 46.0,
-      "step": 34739
-    },
-    {
-      "epoch": 2.6561156029588853,
-      "grad_norm": 0.003757399506866932,
-      "learning_rate": 0.00019999652054054817,
-      "loss": 46.0,
-      "step": 34740
-    },
-    {
-      "epoch": 2.656192059942275,
-      "grad_norm": 0.005926027428358793,
-      "learning_rate": 0.00019999652034017446,
-      "loss": 46.0,
-      "step": 34741
-    },
-    {
-      "epoch": 2.656268516925665,
-      "grad_norm": 0.005261502228677273,
-      "learning_rate": 0.000199996520139795,
-      "loss": 46.0,
-      "step": 34742
-    },
-    {
-      "epoch": 2.6563449739090546,
-      "grad_norm": 0.0016953968442976475,
-      "learning_rate": 0.00019999651993940975,
-      "loss": 46.0,
-      "step": 34743
-    },
-    {
-      "epoch": 2.6564214308924443,
-      "grad_norm": 0.003016469534486532,
-      "learning_rate": 0.00019999651973901875,
-      "loss": 46.0,
-      "step": 34744
-    },
-    {
-      "epoch": 2.656497887875834,
-      "grad_norm": 0.0041351779364049435,
-      "learning_rate": 0.00019999651953862198,
-      "loss": 46.0,
-      "step": 34745
-    },
-    {
-      "epoch": 2.656574344859224,
-      "grad_norm": 0.001842982484959066,
-      "learning_rate": 0.0001999965193382194,
-      "loss": 46.0,
-      "step": 34746
-    },
-    {
-      "epoch": 2.656650801842613,
-      "grad_norm": 0.0019013939891010523,
-      "learning_rate": 0.00019999651913781108,
-      "loss": 46.0,
-      "step": 34747
-    },
-    {
-      "epoch": 2.656727258826003,
-      "grad_norm": 0.0015365495346486568,
-      "learning_rate": 0.000199996518937397,
-      "loss": 46.0,
-      "step": 34748
-    },
-    {
-      "epoch": 2.6568037158093927,
-      "grad_norm": 0.002518998458981514,
-      "learning_rate": 0.00019999651873697715,
-      "loss": 46.0,
-      "step": 34749
-    },
-    {
-      "epoch": 2.6568801727927824,
-      "grad_norm": 0.0017262675100937486,
-      "learning_rate": 0.0001999965185365515,
-      "loss": 46.0,
-      "step": 34750
-    },
-    {
-      "epoch": 2.656956629776172,
-      "grad_norm": 0.0008960299892351031,
-      "learning_rate": 0.00019999651833612012,
-      "loss": 46.0,
-      "step": 34751
-    },
-    {
-      "epoch": 2.657033086759562,
-      "grad_norm": 0.002901954110711813,
-      "learning_rate": 0.00019999651813568296,
-      "loss": 46.0,
-      "step": 34752
-    },
-    {
-      "epoch": 2.6571095437429517,
-      "grad_norm": 0.001309744082391262,
-      "learning_rate": 0.00019999651793524003,
-      "loss": 46.0,
-      "step": 34753
-    },
-    {
-      "epoch": 2.6571860007263415,
-      "grad_norm": 0.0017822058871388435,
-      "learning_rate": 0.00019999651773479135,
-      "loss": 46.0,
-      "step": 34754
-    },
-    {
-      "epoch": 2.657262457709731,
-      "grad_norm": 0.001950585749000311,
-      "learning_rate": 0.00019999651753433684,
-      "loss": 46.0,
-      "step": 34755
-    },
-    {
-      "epoch": 2.6573389146931206,
-      "grad_norm": 0.0018254952738061547,
-      "learning_rate": 0.0001999965173338766,
-      "loss": 46.0,
-      "step": 34756
-    },
-    {
-      "epoch": 2.6574153716765103,
-      "grad_norm": 0.0036601268220692873,
-      "learning_rate": 0.00019999651713341058,
-      "loss": 46.0,
-      "step": 34757
-    },
-    {
-      "epoch": 2.6574918286599,
-      "grad_norm": 0.0006742942496202886,
-      "learning_rate": 0.00019999651693293878,
-      "loss": 46.0,
-      "step": 34758
-    },
-    {
-      "epoch": 2.65756828564329,
-      "grad_norm": 0.004033296834677458,
-      "learning_rate": 0.00019999651673246124,
-      "loss": 46.0,
-      "step": 34759
-    },
-    {
-      "epoch": 2.6576447426266796,
-      "grad_norm": 0.003381472546607256,
-      "learning_rate": 0.00019999651653197792,
-      "loss": 46.0,
-      "step": 34760
-    },
-    {
-      "epoch": 2.6577211996100694,
-      "grad_norm": 0.0013182959519326687,
-      "learning_rate": 0.00019999651633148885,
-      "loss": 46.0,
-      "step": 34761
-    },
-    {
-      "epoch": 2.657797656593459,
-      "grad_norm": 0.001889455015771091,
-      "learning_rate": 0.00019999651613099398,
-      "loss": 46.0,
-      "step": 34762
-    },
-    {
-      "epoch": 2.657874113576849,
-      "grad_norm": 0.0028760316781699657,
-      "learning_rate": 0.00019999651593049337,
-      "loss": 46.0,
-      "step": 34763
-    },
-    {
-      "epoch": 2.6579505705602386,
-      "grad_norm": 0.001243208535015583,
-      "learning_rate": 0.00019999651572998696,
-      "loss": 46.0,
-      "step": 34764
-    },
-    {
-      "epoch": 2.6580270275436284,
-      "grad_norm": 0.0011337959440425038,
-      "learning_rate": 0.0001999965155294748,
-      "loss": 46.0,
-      "step": 34765
-    },
-    {
-      "epoch": 2.658103484527018,
-      "grad_norm": 0.001040091854520142,
-      "learning_rate": 0.00019999651532895686,
-      "loss": 46.0,
-      "step": 34766
-    },
-    {
-      "epoch": 2.658179941510408,
-      "grad_norm": 0.0016802186146378517,
-      "learning_rate": 0.00019999651512843316,
-      "loss": 46.0,
-      "step": 34767
-    },
-    {
-      "epoch": 2.6582563984937977,
-      "grad_norm": 0.002288917312398553,
-      "learning_rate": 0.00019999651492790368,
-      "loss": 46.0,
-      "step": 34768
-    },
-    {
-      "epoch": 2.658332855477187,
-      "grad_norm": 0.0032778973691165447,
-      "learning_rate": 0.00019999651472736845,
-      "loss": 46.0,
-      "step": 34769
-    },
-    {
-      "epoch": 2.6584093124605768,
-      "grad_norm": 0.00397077389061451,
-      "learning_rate": 0.00019999651452682745,
-      "loss": 46.0,
-      "step": 34770
-    },
-    {
-      "epoch": 2.6584857694439665,
-      "grad_norm": 0.0044480194337666035,
-      "learning_rate": 0.00019999651432628065,
-      "loss": 46.0,
-      "step": 34771
-    },
-    {
-      "epoch": 2.6585622264273563,
-      "grad_norm": 0.0025676286313682795,
-      "learning_rate": 0.0001999965141257281,
-      "loss": 46.0,
-      "step": 34772
-    },
-    {
-      "epoch": 2.658638683410746,
-      "grad_norm": 0.0006629474228248,
-      "learning_rate": 0.00019999651392516978,
-      "loss": 46.0,
-      "step": 34773
-    },
-    {
-      "epoch": 2.658715140394136,
-      "grad_norm": 0.0013394899433478713,
-      "learning_rate": 0.0001999965137246057,
-      "loss": 46.0,
-      "step": 34774
-    },
-    {
-      "epoch": 2.6587915973775256,
-      "grad_norm": 0.0027176544535905123,
-      "learning_rate": 0.00019999651352403582,
-      "loss": 46.0,
-      "step": 34775
-    },
-    {
-      "epoch": 2.6588680543609153,
-      "grad_norm": 0.0010863530915230513,
-      "learning_rate": 0.0001999965133234602,
-      "loss": 46.0,
-      "step": 34776
-    },
-    {
-      "epoch": 2.6589445113443047,
-      "grad_norm": 0.005446822382509708,
-      "learning_rate": 0.00019999651312287882,
-      "loss": 46.0,
-      "step": 34777
-    },
-    {
-      "epoch": 2.6590209683276944,
-      "grad_norm": 0.002435858128592372,
-      "learning_rate": 0.00019999651292229164,
-      "loss": 46.0,
-      "step": 34778
-    },
-    {
-      "epoch": 2.659097425311084,
-      "grad_norm": 0.008534245193004608,
-      "learning_rate": 0.0001999965127216987,
-      "loss": 46.0,
-      "step": 34779
-    },
-    {
-      "epoch": 2.659173882294474,
-      "grad_norm": 0.002253937069326639,
-      "learning_rate": 0.0001999965125211,
-      "loss": 46.0,
-      "step": 34780
-    },
-    {
-      "epoch": 2.6592503392778637,
-      "grad_norm": 0.0034409069921821356,
-      "learning_rate": 0.00019999651232049552,
-      "loss": 46.0,
-      "step": 34781
-    },
-    {
-      "epoch": 2.6593267962612535,
-      "grad_norm": 0.0028056346345692873,
-      "learning_rate": 0.0001999965121198853,
-      "loss": 46.0,
-      "step": 34782
-    },
-    {
-      "epoch": 2.659403253244643,
-      "grad_norm": 0.0018720593070611358,
-      "learning_rate": 0.00019999651191926926,
-      "loss": 46.0,
-      "step": 34783
-    },
-    {
-      "epoch": 2.659479710228033,
-      "grad_norm": 0.005867813713848591,
-      "learning_rate": 0.0001999965117186475,
-      "loss": 46.0,
-      "step": 34784
-    },
-    {
-      "epoch": 2.6595561672114227,
-      "grad_norm": 0.000925446511246264,
-      "learning_rate": 0.00019999651151801994,
-      "loss": 46.0,
-      "step": 34785
-    },
-    {
-      "epoch": 2.6596326241948125,
-      "grad_norm": 0.003714770544320345,
-      "learning_rate": 0.00019999651131738663,
-      "loss": 46.0,
-      "step": 34786
-    },
-    {
-      "epoch": 2.6597090811782023,
-      "grad_norm": 0.0031848398502916098,
-      "learning_rate": 0.00019999651111674753,
-      "loss": 46.0,
-      "step": 34787
-    },
-    {
-      "epoch": 2.659785538161592,
-      "grad_norm": 0.0030491426587104797,
-      "learning_rate": 0.00019999651091610267,
-      "loss": 46.0,
-      "step": 34788
-    },
-    {
-      "epoch": 2.659861995144982,
-      "grad_norm": 0.0005798854399472475,
-      "learning_rate": 0.00019999651071545205,
-      "loss": 46.0,
-      "step": 34789
-    },
-    {
-      "epoch": 2.6599384521283715,
-      "grad_norm": 0.0024995480198413134,
-      "learning_rate": 0.00019999651051479567,
-      "loss": 46.0,
-      "step": 34790
-    },
-    {
-      "epoch": 2.660014909111761,
-      "grad_norm": 0.001593344029970467,
-      "learning_rate": 0.00019999651031413348,
-      "loss": 46.0,
-      "step": 34791
-    },
-    {
-      "epoch": 2.6600913660951506,
-      "grad_norm": 0.0030092853121459484,
-      "learning_rate": 0.00019999651011346555,
-      "loss": 46.0,
-      "step": 34792
-    },
-    {
-      "epoch": 2.6601678230785404,
-      "grad_norm": 0.001418916042894125,
-      "learning_rate": 0.00019999650991279184,
-      "loss": 46.0,
-      "step": 34793
-    },
-    {
-      "epoch": 2.66024428006193,
-      "grad_norm": 0.0015205746749415994,
-      "learning_rate": 0.00019999650971211236,
-      "loss": 46.0,
-      "step": 34794
-    },
-    {
-      "epoch": 2.66032073704532,
-      "grad_norm": 0.0015802271664142609,
-      "learning_rate": 0.0001999965095114271,
-      "loss": 46.0,
-      "step": 34795
-    },
-    {
-      "epoch": 2.6603971940287097,
-      "grad_norm": 0.0026045767590403557,
-      "learning_rate": 0.0001999965093107361,
-      "loss": 46.0,
-      "step": 34796
-    },
-    {
-      "epoch": 2.6604736510120994,
-      "grad_norm": 0.0038425433449447155,
-      "learning_rate": 0.00019999650911003934,
-      "loss": 46.0,
-      "step": 34797
-    },
-    {
-      "epoch": 2.660550107995489,
-      "grad_norm": 0.005135990213602781,
-      "learning_rate": 0.0001999965089093368,
-      "loss": 46.0,
-      "step": 34798
-    },
-    {
-      "epoch": 2.6606265649788785,
-      "grad_norm": 0.0011006761342287064,
-      "learning_rate": 0.00019999650870862845,
-      "loss": 46.0,
-      "step": 34799
-    },
-    {
-      "epoch": 2.6607030219622683,
-      "grad_norm": 0.0017809598939493299,
-      "learning_rate": 0.00019999650850791438,
-      "loss": 46.0,
-      "step": 34800
-    },
-    {
-      "epoch": 2.660779478945658,
-      "grad_norm": 0.0019728532060980797,
-      "learning_rate": 0.00019999650830719449,
-      "loss": 46.0,
-      "step": 34801
-    },
-    {
-      "epoch": 2.660855935929048,
-      "grad_norm": 0.004906200338155031,
-      "learning_rate": 0.00019999650810646887,
-      "loss": 46.0,
-      "step": 34802
-    },
-    {
-      "epoch": 2.6609323929124375,
-      "grad_norm": 0.0007619601092301309,
-      "learning_rate": 0.0001999965079057375,
-      "loss": 46.0,
-      "step": 34803
-    },
-    {
-      "epoch": 2.6610088498958273,
-      "grad_norm": 0.0022770592477172613,
-      "learning_rate": 0.0001999965077050003,
-      "loss": 46.0,
-      "step": 34804
-    },
-    {
-      "epoch": 2.661085306879217,
-      "grad_norm": 0.0015122673939913511,
-      "learning_rate": 0.00019999650750425737,
-      "loss": 46.0,
-      "step": 34805
-    },
-    {
-      "epoch": 2.661161763862607,
-      "grad_norm": 0.0011271742405369878,
-      "learning_rate": 0.00019999650730350866,
-      "loss": 46.0,
-      "step": 34806
-    },
-    {
-      "epoch": 2.6612382208459966,
-      "grad_norm": 0.0006319869426079094,
-      "learning_rate": 0.0001999965071027542,
-      "loss": 46.0,
-      "step": 34807
-    },
-    {
-      "epoch": 2.6613146778293864,
-      "grad_norm": 0.0006759246462024748,
-      "learning_rate": 0.00019999650690199396,
-      "loss": 46.0,
-      "step": 34808
-    },
-    {
-      "epoch": 2.661391134812776,
-      "grad_norm": 0.0028826105408370495,
-      "learning_rate": 0.00019999650670122793,
-      "loss": 46.0,
-      "step": 34809
-    },
-    {
-      "epoch": 2.661467591796166,
-      "grad_norm": 0.002271621488034725,
-      "learning_rate": 0.00019999650650045616,
-      "loss": 46.0,
-      "step": 34810
-    },
-    {
-      "epoch": 2.6615440487795556,
-      "grad_norm": 0.003738553263247013,
-      "learning_rate": 0.00019999650629967862,
-      "loss": 46.0,
-      "step": 34811
-    },
-    {
-      "epoch": 2.6616205057629454,
-      "grad_norm": 0.0016260184347629547,
-      "learning_rate": 0.0001999965060988953,
-      "loss": 46.0,
-      "step": 34812
-    },
-    {
-      "epoch": 2.6616969627463347,
-      "grad_norm": 0.0011628199135884643,
-      "learning_rate": 0.00019999650589810618,
-      "loss": 46.0,
-      "step": 34813
-    },
-    {
-      "epoch": 2.6617734197297245,
-      "grad_norm": 0.0026137125678360462,
-      "learning_rate": 0.00019999650569731134,
-      "loss": 46.0,
-      "step": 34814
-    },
-    {
-      "epoch": 2.6618498767131142,
-      "grad_norm": 0.001854165457189083,
-      "learning_rate": 0.00019999650549651073,
-      "loss": 46.0,
-      "step": 34815
-    },
-    {
-      "epoch": 2.661926333696504,
-      "grad_norm": 0.006665118504315615,
-      "learning_rate": 0.00019999650529570434,
-      "loss": 46.0,
-      "step": 34816
-    },
-    {
-      "epoch": 2.6620027906798938,
-      "grad_norm": 0.00467985263094306,
-      "learning_rate": 0.00019999650509489215,
-      "loss": 46.0,
-      "step": 34817
-    },
-    {
-      "epoch": 2.6620792476632835,
-      "grad_norm": 0.0038444646634161472,
-      "learning_rate": 0.00019999650489407422,
-      "loss": 46.0,
-      "step": 34818
-    },
-    {
-      "epoch": 2.6621557046466733,
-      "grad_norm": 0.004331445321440697,
-      "learning_rate": 0.00019999650469325052,
-      "loss": 46.0,
-      "step": 34819
-    },
-    {
-      "epoch": 2.662232161630063,
-      "grad_norm": 0.0020280415192246437,
-      "learning_rate": 0.000199996504492421,
-      "loss": 46.0,
-      "step": 34820
-    },
-    {
-      "epoch": 2.6623086186134524,
-      "grad_norm": 0.0009757000952959061,
-      "learning_rate": 0.00019999650429158579,
-      "loss": 46.0,
-      "step": 34821
-    },
-    {
-      "epoch": 2.662385075596842,
-      "grad_norm": 0.0019797596614807844,
-      "learning_rate": 0.0001999965040907448,
-      "loss": 46.0,
-      "step": 34822
-    },
-    {
-      "epoch": 2.662461532580232,
-      "grad_norm": 0.004333363380283117,
-      "learning_rate": 0.000199996503889898,
-      "loss": 46.0,
-      "step": 34823
-    },
-    {
-      "epoch": 2.6625379895636216,
-      "grad_norm": 0.00207108655013144,
-      "learning_rate": 0.00019999650368904544,
-      "loss": 46.0,
-      "step": 34824
-    },
-    {
-      "epoch": 2.6626144465470114,
-      "grad_norm": 0.0006575423758476973,
-      "learning_rate": 0.00019999650348818712,
-      "loss": 46.0,
-      "step": 34825
-    },
-    {
-      "epoch": 2.662690903530401,
-      "grad_norm": 0.0009055683040060103,
-      "learning_rate": 0.00019999650328732303,
-      "loss": 46.0,
-      "step": 34826
-    },
-    {
-      "epoch": 2.662767360513791,
-      "grad_norm": 0.0012490932131186128,
-      "learning_rate": 0.00019999650308645317,
-      "loss": 46.0,
-      "step": 34827
-    },
-    {
-      "epoch": 2.6628438174971807,
-      "grad_norm": 0.00224423804320395,
-      "learning_rate": 0.00019999650288557753,
-      "loss": 46.0,
-      "step": 34828
-    },
-    {
-      "epoch": 2.6629202744805704,
-      "grad_norm": 0.0010030536213889718,
-      "learning_rate": 0.00019999650268469614,
-      "loss": 46.0,
-      "step": 34829
-    },
-    {
-      "epoch": 2.66299673146396,
-      "grad_norm": 0.0027507049962878227,
-      "learning_rate": 0.00019999650248380898,
-      "loss": 46.0,
-      "step": 34830
-    },
-    {
-      "epoch": 2.66307318844735,
-      "grad_norm": 0.003532292554154992,
-      "learning_rate": 0.00019999650228291605,
-      "loss": 46.0,
-      "step": 34831
-    },
-    {
-      "epoch": 2.6631496454307397,
-      "grad_norm": 0.0010683690197765827,
-      "learning_rate": 0.00019999650208201732,
-      "loss": 46.0,
-      "step": 34832
-    },
-    {
-      "epoch": 2.6632261024141295,
-      "grad_norm": 0.0010906958486884832,
-      "learning_rate": 0.00019999650188111287,
-      "loss": 46.0,
-      "step": 34833
-    },
-    {
-      "epoch": 2.663302559397519,
-      "grad_norm": 0.0026041092351078987,
-      "learning_rate": 0.00019999650168020262,
-      "loss": 46.0,
-      "step": 34834
-    },
-    {
-      "epoch": 2.6633790163809086,
-      "grad_norm": 0.0015963248442858458,
-      "learning_rate": 0.00019999650147928662,
-      "loss": 46.0,
-      "step": 34835
-    },
-    {
-      "epoch": 2.6634554733642983,
-      "grad_norm": 0.001647595432586968,
-      "learning_rate": 0.00019999650127836482,
-      "loss": 46.0,
-      "step": 34836
-    },
-    {
-      "epoch": 2.663531930347688,
-      "grad_norm": 0.0028493814170360565,
-      "learning_rate": 0.00019999650107743725,
-      "loss": 46.0,
-      "step": 34837
-    },
-    {
-      "epoch": 2.663608387331078,
-      "grad_norm": 0.0014009576989337802,
-      "learning_rate": 0.00019999650087650396,
-      "loss": 46.0,
-      "step": 34838
-    },
-    {
-      "epoch": 2.6636848443144676,
-      "grad_norm": 0.0016593976179137826,
-      "learning_rate": 0.00019999650067556486,
-      "loss": 46.0,
-      "step": 34839
-    },
-    {
-      "epoch": 2.6637613012978574,
-      "grad_norm": 0.0015545575879514217,
-      "learning_rate": 0.00019999650047462,
-      "loss": 46.0,
-      "step": 34840
-    },
-    {
-      "epoch": 2.663837758281247,
-      "grad_norm": 0.00259993108920753,
-      "learning_rate": 0.00019999650027366936,
-      "loss": 46.0,
-      "step": 34841
-    },
-    {
-      "epoch": 2.663914215264637,
-      "grad_norm": 0.0014673768309876323,
-      "learning_rate": 0.00019999650007271295,
-      "loss": 46.0,
-      "step": 34842
-    },
-    {
-      "epoch": 2.663990672248026,
-      "grad_norm": 0.0011018171207979321,
-      "learning_rate": 0.0001999964998717508,
-      "loss": 46.0,
-      "step": 34843
-    },
-    {
-      "epoch": 2.664067129231416,
-      "grad_norm": 0.0022359294816851616,
-      "learning_rate": 0.00019999649967078286,
-      "loss": 46.0,
-      "step": 34844
-    },
-    {
-      "epoch": 2.6641435862148057,
-      "grad_norm": 0.0011683318298310041,
-      "learning_rate": 0.00019999649946980915,
-      "loss": 46.0,
-      "step": 34845
-    },
-    {
-      "epoch": 2.6642200431981955,
-      "grad_norm": 0.0017478878144174814,
-      "learning_rate": 0.00019999649926882968,
-      "loss": 46.0,
-      "step": 34846
-    },
-    {
-      "epoch": 2.6642965001815853,
-      "grad_norm": 0.001801581820473075,
-      "learning_rate": 0.00019999649906784442,
-      "loss": 46.0,
-      "step": 34847
-    },
-    {
-      "epoch": 2.664372957164975,
-      "grad_norm": 0.0030484963208436966,
-      "learning_rate": 0.0001999964988668534,
-      "loss": 46.0,
-      "step": 34848
-    },
-    {
-      "epoch": 2.664449414148365,
-      "grad_norm": 0.0020317472517490387,
-      "learning_rate": 0.00019999649866585663,
-      "loss": 46.0,
-      "step": 34849
-    },
-    {
-      "epoch": 2.6645258711317545,
-      "grad_norm": 0.001673631020821631,
-      "learning_rate": 0.00019999649846485408,
-      "loss": 46.0,
-      "step": 34850
-    },
-    {
-      "epoch": 2.6646023281151443,
-      "grad_norm": 0.0023410418070852757,
-      "learning_rate": 0.00019999649826384577,
-      "loss": 46.0,
-      "step": 34851
-    },
-    {
-      "epoch": 2.664678785098534,
-      "grad_norm": 0.006933731026947498,
-      "learning_rate": 0.00019999649806283167,
-      "loss": 46.0,
-      "step": 34852
-    },
-    {
-      "epoch": 2.664755242081924,
-      "grad_norm": 0.0024782984983175993,
-      "learning_rate": 0.0001999964978618118,
-      "loss": 46.0,
-      "step": 34853
-    },
-    {
-      "epoch": 2.6648316990653136,
-      "grad_norm": 0.006291090976446867,
-      "learning_rate": 0.0001999964976607862,
-      "loss": 46.0,
-      "step": 34854
-    },
-    {
-      "epoch": 2.6649081560487033,
-      "grad_norm": 0.0015015807002782822,
-      "learning_rate": 0.00019999649745975479,
-      "loss": 46.0,
-      "step": 34855
-    },
-    {
-      "epoch": 2.6649846130320927,
-      "grad_norm": 0.0023913918994367123,
-      "learning_rate": 0.0001999964972587176,
-      "loss": 46.0,
-      "step": 34856
-    },
-    {
-      "epoch": 2.6650610700154824,
-      "grad_norm": 0.004275403451174498,
-      "learning_rate": 0.0001999964970576747,
-      "loss": 46.0,
-      "step": 34857
-    },
-    {
-      "epoch": 2.665137526998872,
-      "grad_norm": 0.0036315189208835363,
-      "learning_rate": 0.000199996496856626,
-      "loss": 46.0,
-      "step": 34858
-    },
-    {
-      "epoch": 2.665213983982262,
-      "grad_norm": 0.0021812699269503355,
-      "learning_rate": 0.0001999964966555715,
-      "loss": 46.0,
-      "step": 34859
-    },
-    {
-      "epoch": 2.6652904409656517,
-      "grad_norm": 0.0026718545705080032,
-      "learning_rate": 0.00019999649645451126,
-      "loss": 46.0,
-      "step": 34860
-    },
-    {
-      "epoch": 2.6653668979490415,
-      "grad_norm": 0.004206669516861439,
-      "learning_rate": 0.00019999649625344527,
-      "loss": 46.0,
-      "step": 34861
-    },
-    {
-      "epoch": 2.6654433549324312,
-      "grad_norm": 0.0021670302376151085,
-      "learning_rate": 0.00019999649605237347,
-      "loss": 46.0,
-      "step": 34862
-    },
-    {
-      "epoch": 2.665519811915821,
-      "grad_norm": 0.0020318003371357918,
-      "learning_rate": 0.00019999649585129592,
-      "loss": 46.0,
-      "step": 34863
-    },
-    {
-      "epoch": 2.6655962688992103,
-      "grad_norm": 0.0037286237347871065,
-      "learning_rate": 0.00019999649565021258,
-      "loss": 46.0,
-      "step": 34864
-    },
-    {
-      "epoch": 2.6656727258826,
-      "grad_norm": 0.0020174169912934303,
-      "learning_rate": 0.0001999964954491235,
-      "loss": 46.0,
-      "step": 34865
-    },
-    {
-      "epoch": 2.66574918286599,
-      "grad_norm": 0.001935688080266118,
-      "learning_rate": 0.00019999649524802865,
-      "loss": 46.0,
-      "step": 34866
-    },
-    {
-      "epoch": 2.6658256398493796,
-      "grad_norm": 0.0025332774966955185,
-      "learning_rate": 0.000199996495046928,
-      "loss": 46.0,
-      "step": 34867
-    },
-    {
-      "epoch": 2.6659020968327694,
-      "grad_norm": 0.0013528376584872603,
-      "learning_rate": 0.0001999964948458216,
-      "loss": 46.0,
-      "step": 34868
-    },
-    {
-      "epoch": 2.665978553816159,
-      "grad_norm": 0.001148971146903932,
-      "learning_rate": 0.00019999649464470944,
-      "loss": 46.0,
-      "step": 34869
-    },
-    {
-      "epoch": 2.666055010799549,
-      "grad_norm": 0.0028508210089057684,
-      "learning_rate": 0.00019999649444359149,
-      "loss": 46.0,
-      "step": 34870
-    },
-    {
-      "epoch": 2.6661314677829386,
-      "grad_norm": 0.002312930068001151,
-      "learning_rate": 0.00019999649424246778,
-      "loss": 46.0,
-      "step": 34871
-    },
-    {
-      "epoch": 2.6662079247663284,
-      "grad_norm": 0.003013030393049121,
-      "learning_rate": 0.0001999964940413383,
-      "loss": 46.0,
-      "step": 34872
-    },
-    {
-      "epoch": 2.666284381749718,
-      "grad_norm": 0.002281905384734273,
-      "learning_rate": 0.00019999649384020308,
-      "loss": 46.0,
-      "step": 34873
-    },
-    {
-      "epoch": 2.666360838733108,
-      "grad_norm": 0.0022345969919115305,
-      "learning_rate": 0.00019999649363906206,
-      "loss": 46.0,
-      "step": 34874
-    },
-    {
-      "epoch": 2.6664372957164977,
-      "grad_norm": 0.0032459436915814877,
-      "learning_rate": 0.00019999649343791526,
-      "loss": 46.0,
-      "step": 34875
-    },
-    {
-      "epoch": 2.6665137526998874,
-      "grad_norm": 0.0013581333914771676,
-      "learning_rate": 0.0001999964932367627,
-      "loss": 46.0,
-      "step": 34876
-    },
-    {
-      "epoch": 2.666590209683277,
-      "grad_norm": 0.0010926889954134822,
-      "learning_rate": 0.00019999649303560442,
-      "loss": 46.0,
-      "step": 34877
-    },
-    {
-      "epoch": 2.6666666666666665,
-      "grad_norm": 0.006870744749903679,
-      "learning_rate": 0.0001999964928344403,
-      "loss": 46.0,
-      "step": 34878
-    },
-    {
-      "epoch": 2.6667431236500563,
-      "grad_norm": 0.001762778265401721,
-      "learning_rate": 0.00019999649263327044,
-      "loss": 46.0,
-      "step": 34879
-    },
-    {
-      "epoch": 2.666819580633446,
-      "grad_norm": 0.0024501406587660313,
-      "learning_rate": 0.00019999649243209483,
-      "loss": 46.0,
-      "step": 34880
-    },
-    {
-      "epoch": 2.666896037616836,
-      "grad_norm": 0.0017349875997751951,
-      "learning_rate": 0.00019999649223091342,
-      "loss": 46.0,
-      "step": 34881
-    },
-    {
-      "epoch": 2.6669724946002256,
-      "grad_norm": 0.004817704204469919,
-      "learning_rate": 0.00019999649202972624,
-      "loss": 46.0,
-      "step": 34882
-    },
-    {
-      "epoch": 2.6670489515836153,
-      "grad_norm": 0.002187000121921301,
-      "learning_rate": 0.0001999964918285333,
-      "loss": 46.0,
-      "step": 34883
-    },
-    {
-      "epoch": 2.667125408567005,
-      "grad_norm": 0.0023201985750347376,
-      "learning_rate": 0.0001999964916273346,
-      "loss": 46.0,
-      "step": 34884
-    },
-    {
-      "epoch": 2.667201865550395,
-      "grad_norm": 0.0014903944684192538,
-      "learning_rate": 0.00019999649142613012,
-      "loss": 46.0,
-      "step": 34885
-    },
-    {
-      "epoch": 2.667278322533784,
-      "grad_norm": 0.0037571508437395096,
-      "learning_rate": 0.00019999649122491987,
-      "loss": 46.0,
-      "step": 34886
-    },
-    {
-      "epoch": 2.667354779517174,
-      "grad_norm": 0.0024626918602734804,
-      "learning_rate": 0.00019999649102370385,
-      "loss": 46.0,
-      "step": 34887
-    },
-    {
-      "epoch": 2.6674312365005637,
-      "grad_norm": 0.0016515154857188463,
-      "learning_rate": 0.00019999649082248208,
-      "loss": 46.0,
-      "step": 34888
-    },
-    {
-      "epoch": 2.6675076934839534,
-      "grad_norm": 0.0016520003555342555,
-      "learning_rate": 0.00019999649062125453,
-      "loss": 46.0,
-      "step": 34889
-    },
-    {
-      "epoch": 2.667584150467343,
-      "grad_norm": 0.0006765213911421597,
-      "learning_rate": 0.0001999964904200212,
-      "loss": 46.0,
-      "step": 34890
-    },
-    {
-      "epoch": 2.667660607450733,
-      "grad_norm": 0.001740854699164629,
-      "learning_rate": 0.00019999649021878213,
-      "loss": 46.0,
-      "step": 34891
-    },
-    {
-      "epoch": 2.6677370644341227,
-      "grad_norm": 0.0018777099903672934,
-      "learning_rate": 0.00019999649001753724,
-      "loss": 46.0,
-      "step": 34892
-    },
-    {
-      "epoch": 2.6678135214175125,
-      "grad_norm": 0.0012235044268891215,
-      "learning_rate": 0.00019999648981628663,
-      "loss": 46.0,
-      "step": 34893
-    },
-    {
-      "epoch": 2.6678899784009023,
-      "grad_norm": 0.0076610553078353405,
-      "learning_rate": 0.00019999648961503024,
-      "loss": 46.0,
-      "step": 34894
-    },
-    {
-      "epoch": 2.667966435384292,
-      "grad_norm": 0.002160225994884968,
-      "learning_rate": 0.00019999648941376809,
-      "loss": 46.0,
-      "step": 34895
-    },
-    {
-      "epoch": 2.6680428923676818,
-      "grad_norm": 0.004819528199732304,
-      "learning_rate": 0.00019999648921250013,
-      "loss": 46.0,
-      "step": 34896
-    },
-    {
-      "epoch": 2.6681193493510715,
-      "grad_norm": 0.0034319048281759024,
-      "learning_rate": 0.00019999648901122642,
-      "loss": 46.0,
-      "step": 34897
-    },
-    {
-      "epoch": 2.6681958063344613,
-      "grad_norm": 0.002473304281011224,
-      "learning_rate": 0.00019999648880994695,
-      "loss": 46.0,
-      "step": 34898
-    },
-    {
-      "epoch": 2.668272263317851,
-      "grad_norm": 0.0007150843739509583,
-      "learning_rate": 0.0001999964886086617,
-      "loss": 46.0,
-      "step": 34899
-    },
-    {
-      "epoch": 2.6683487203012404,
-      "grad_norm": 0.002251900965347886,
-      "learning_rate": 0.00019999648840737067,
-      "loss": 46.0,
-      "step": 34900
-    },
-    {
-      "epoch": 2.66842517728463,
-      "grad_norm": 0.0034669458400458097,
-      "learning_rate": 0.0001999964882060739,
-      "loss": 46.0,
-      "step": 34901
-    },
-    {
-      "epoch": 2.66850163426802,
-      "grad_norm": 0.0013607167638838291,
-      "learning_rate": 0.00019999648800477133,
-      "loss": 46.0,
-      "step": 34902
-    },
-    {
-      "epoch": 2.6685780912514097,
-      "grad_norm": 0.002192132640630007,
-      "learning_rate": 0.00019999648780346302,
-      "loss": 46.0,
-      "step": 34903
-    },
-    {
-      "epoch": 2.6686545482347994,
-      "grad_norm": 0.004075739067047834,
-      "learning_rate": 0.00019999648760214893,
-      "loss": 46.0,
-      "step": 34904
-    },
-    {
-      "epoch": 2.668731005218189,
-      "grad_norm": 0.0013467296957969666,
-      "learning_rate": 0.00019999648740082906,
-      "loss": 46.0,
-      "step": 34905
-    },
-    {
-      "epoch": 2.668807462201579,
-      "grad_norm": 0.0016778097487986088,
-      "learning_rate": 0.00019999648719950342,
-      "loss": 46.0,
-      "step": 34906
-    },
-    {
-      "epoch": 2.6688839191849687,
-      "grad_norm": 0.0029106067959219217,
-      "learning_rate": 0.00019999648699817204,
-      "loss": 46.0,
-      "step": 34907
-    },
-    {
-      "epoch": 2.668960376168358,
-      "grad_norm": 0.0023601623252034187,
-      "learning_rate": 0.00019999648679683486,
-      "loss": 46.0,
-      "step": 34908
-    },
-    {
-      "epoch": 2.669036833151748,
-      "grad_norm": 0.0019676717929542065,
-      "learning_rate": 0.00019999648659549193,
-      "loss": 46.0,
-      "step": 34909
-    },
-    {
-      "epoch": 2.6691132901351375,
-      "grad_norm": 0.0014156763209030032,
-      "learning_rate": 0.00019999648639414322,
-      "loss": 46.0,
-      "step": 34910
-    },
-    {
-      "epoch": 2.6691897471185273,
-      "grad_norm": 0.0007549666916020215,
-      "learning_rate": 0.00019999648619278875,
-      "loss": 46.0,
-      "step": 34911
-    },
-    {
-      "epoch": 2.669266204101917,
-      "grad_norm": 0.0026200178544968367,
-      "learning_rate": 0.0001999964859914285,
-      "loss": 46.0,
-      "step": 34912
-    },
-    {
-      "epoch": 2.669342661085307,
-      "grad_norm": 0.002217887667939067,
-      "learning_rate": 0.00019999648579006247,
-      "loss": 46.0,
-      "step": 34913
-    },
-    {
-      "epoch": 2.6694191180686966,
-      "grad_norm": 0.004524400923401117,
-      "learning_rate": 0.0001999964855886907,
-      "loss": 46.0,
-      "step": 34914
-    },
-    {
-      "epoch": 2.6694955750520863,
-      "grad_norm": 0.003973620478063822,
-      "learning_rate": 0.00019999648538731316,
-      "loss": 46.0,
-      "step": 34915
-    },
-    {
-      "epoch": 2.669572032035476,
-      "grad_norm": 0.001967710442841053,
-      "learning_rate": 0.00019999648518592982,
-      "loss": 46.0,
-      "step": 34916
-    },
-    {
-      "epoch": 2.669648489018866,
-      "grad_norm": 0.001152605633251369,
-      "learning_rate": 0.00019999648498454073,
-      "loss": 46.0,
-      "step": 34917
-    },
-    {
-      "epoch": 2.6697249460022556,
-      "grad_norm": 0.0014740233309566975,
-      "learning_rate": 0.00019999648478314586,
-      "loss": 46.0,
-      "step": 34918
-    },
-    {
-      "epoch": 2.6698014029856454,
-      "grad_norm": 0.0013740730937570333,
-      "learning_rate": 0.00019999648458174526,
-      "loss": 46.0,
-      "step": 34919
-    },
-    {
-      "epoch": 2.669877859969035,
-      "grad_norm": 0.0022710999473929405,
-      "learning_rate": 0.00019999648438033885,
-      "loss": 46.0,
-      "step": 34920
-    },
-    {
-      "epoch": 2.669954316952425,
-      "grad_norm": 0.0019967190455645323,
-      "learning_rate": 0.00019999648417892666,
-      "loss": 46.0,
-      "step": 34921
-    },
-    {
-      "epoch": 2.6700307739358142,
-      "grad_norm": 0.0017750104889273643,
-      "learning_rate": 0.00019999648397750873,
-      "loss": 46.0,
-      "step": 34922
-    },
-    {
-      "epoch": 2.670107230919204,
-      "grad_norm": 0.0023550225887447596,
-      "learning_rate": 0.00019999648377608503,
-      "loss": 46.0,
-      "step": 34923
-    },
-    {
-      "epoch": 2.6701836879025937,
-      "grad_norm": 0.0017071013571694493,
-      "learning_rate": 0.00019999648357465555,
-      "loss": 46.0,
-      "step": 34924
-    },
-    {
-      "epoch": 2.6702601448859835,
-      "grad_norm": 0.002447047270834446,
-      "learning_rate": 0.0001999964833732203,
-      "loss": 46.0,
-      "step": 34925
-    },
-    {
-      "epoch": 2.6703366018693733,
-      "grad_norm": 0.0017103665741160512,
-      "learning_rate": 0.00019999648317177928,
-      "loss": 46.0,
-      "step": 34926
-    },
-    {
-      "epoch": 2.670413058852763,
-      "grad_norm": 0.003538508666679263,
-      "learning_rate": 0.0001999964829703325,
-      "loss": 46.0,
-      "step": 34927
-    },
-    {
-      "epoch": 2.670489515836153,
-      "grad_norm": 0.0041619157418608665,
-      "learning_rate": 0.00019999648276887994,
-      "loss": 46.0,
-      "step": 34928
-    },
-    {
-      "epoch": 2.6705659728195426,
-      "grad_norm": 0.0019612619653344154,
-      "learning_rate": 0.0001999964825674216,
-      "loss": 46.0,
-      "step": 34929
-    },
-    {
-      "epoch": 2.670642429802932,
-      "grad_norm": 0.0009307538857683539,
-      "learning_rate": 0.0001999964823659575,
-      "loss": 46.0,
-      "step": 34930
-    },
-    {
-      "epoch": 2.6707188867863216,
-      "grad_norm": 0.0013061750214546919,
-      "learning_rate": 0.00019999648216448764,
-      "loss": 46.0,
-      "step": 34931
-    },
-    {
-      "epoch": 2.6707953437697114,
-      "grad_norm": 0.0037661294918507338,
-      "learning_rate": 0.00019999648196301204,
-      "loss": 46.0,
-      "step": 34932
-    },
-    {
-      "epoch": 2.670871800753101,
-      "grad_norm": 0.011401028372347355,
-      "learning_rate": 0.00019999648176153065,
-      "loss": 46.0,
-      "step": 34933
-    },
-    {
-      "epoch": 2.670948257736491,
-      "grad_norm": 0.0036667597014456987,
-      "learning_rate": 0.00019999648156004347,
-      "loss": 46.0,
-      "step": 34934
-    },
-    {
-      "epoch": 2.6710247147198807,
-      "grad_norm": 0.0012479517608880997,
-      "learning_rate": 0.0001999964813585505,
-      "loss": 46.0,
-      "step": 34935
-    },
-    {
-      "epoch": 2.6711011717032704,
-      "grad_norm": 0.00181959371548146,
-      "learning_rate": 0.00019999648115705184,
-      "loss": 46.0,
-      "step": 34936
-    },
-    {
-      "epoch": 2.67117762868666,
-      "grad_norm": 0.0012184573570266366,
-      "learning_rate": 0.00019999648095554733,
-      "loss": 46.0,
-      "step": 34937
-    },
-    {
-      "epoch": 2.67125408567005,
-      "grad_norm": 0.0032569922041147947,
-      "learning_rate": 0.00019999648075403709,
-      "loss": 46.0,
-      "step": 34938
-    },
-    {
-      "epoch": 2.6713305426534397,
-      "grad_norm": 0.0022691187914460897,
-      "learning_rate": 0.0001999964805525211,
-      "loss": 46.0,
-      "step": 34939
-    },
-    {
-      "epoch": 2.6714069996368295,
-      "grad_norm": 0.001322792493738234,
-      "learning_rate": 0.0001999964803509993,
-      "loss": 46.0,
-      "step": 34940
-    },
-    {
-      "epoch": 2.6714834566202192,
-      "grad_norm": 0.003794808639213443,
-      "learning_rate": 0.00019999648014947175,
-      "loss": 46.0,
-      "step": 34941
-    },
-    {
-      "epoch": 2.671559913603609,
-      "grad_norm": 0.002194920089095831,
-      "learning_rate": 0.0001999964799479384,
-      "loss": 46.0,
-      "step": 34942
-    },
-    {
-      "epoch": 2.6716363705869988,
-      "grad_norm": 0.002413443522527814,
-      "learning_rate": 0.00019999647974639932,
-      "loss": 46.0,
-      "step": 34943
-    },
-    {
-      "epoch": 2.671712827570388,
-      "grad_norm": 0.0017411314183846116,
-      "learning_rate": 0.00019999647954485448,
-      "loss": 46.0,
-      "step": 34944
-    },
-    {
-      "epoch": 2.671789284553778,
-      "grad_norm": 0.001352843944914639,
-      "learning_rate": 0.00019999647934330382,
-      "loss": 46.0,
-      "step": 34945
-    },
-    {
-      "epoch": 2.6718657415371676,
-      "grad_norm": 0.0018036641413345933,
-      "learning_rate": 0.0001999964791417474,
-      "loss": 46.0,
-      "step": 34946
-    },
-    {
-      "epoch": 2.6719421985205574,
-      "grad_norm": 0.0032898527570068836,
-      "learning_rate": 0.00019999647894018528,
-      "loss": 46.0,
-      "step": 34947
-    },
-    {
-      "epoch": 2.672018655503947,
-      "grad_norm": 0.005762800574302673,
-      "learning_rate": 0.0001999964787386173,
-      "loss": 46.0,
-      "step": 34948
-    },
-    {
-      "epoch": 2.672095112487337,
-      "grad_norm": 0.00338967377319932,
-      "learning_rate": 0.00019999647853704362,
-      "loss": 46.0,
-      "step": 34949
-    },
-    {
-      "epoch": 2.6721715694707266,
-      "grad_norm": 0.002865397837013006,
-      "learning_rate": 0.00019999647833546412,
-      "loss": 46.0,
-      "step": 34950
-    },
-    {
-      "epoch": 2.6722480264541164,
-      "grad_norm": 0.0030455742962658405,
-      "learning_rate": 0.0001999964781338789,
-      "loss": 46.0,
-      "step": 34951
-    },
-    {
-      "epoch": 2.6723244834375057,
-      "grad_norm": 0.0018135480349883437,
-      "learning_rate": 0.00019999647793228785,
-      "loss": 46.0,
-      "step": 34952
-    },
-    {
-      "epoch": 2.6724009404208955,
-      "grad_norm": 0.0018838674295693636,
-      "learning_rate": 0.00019999647773069108,
-      "loss": 46.0,
-      "step": 34953
-    },
-    {
-      "epoch": 2.6724773974042852,
-      "grad_norm": 0.009196566417813301,
-      "learning_rate": 0.0001999964775290885,
-      "loss": 46.0,
-      "step": 34954
-    },
-    {
-      "epoch": 2.672553854387675,
-      "grad_norm": 0.0010365053312852979,
-      "learning_rate": 0.00019999647732748022,
-      "loss": 46.0,
-      "step": 34955
-    },
-    {
-      "epoch": 2.6726303113710648,
-      "grad_norm": 0.0014754708390682936,
-      "learning_rate": 0.00019999647712586613,
-      "loss": 46.0,
-      "step": 34956
-    },
-    {
-      "epoch": 2.6727067683544545,
-      "grad_norm": 0.0023555499501526356,
-      "learning_rate": 0.00019999647692424624,
-      "loss": 46.0,
-      "step": 34957
-    },
-    {
-      "epoch": 2.6727832253378443,
-      "grad_norm": 0.004240604117512703,
-      "learning_rate": 0.00019999647672262064,
-      "loss": 46.0,
-      "step": 34958
-    },
-    {
-      "epoch": 2.672859682321234,
-      "grad_norm": 0.0013119049835950136,
-      "learning_rate": 0.00019999647652098923,
-      "loss": 46.0,
-      "step": 34959
-    },
-    {
-      "epoch": 2.672936139304624,
-      "grad_norm": 0.007284971885383129,
-      "learning_rate": 0.00019999647631935204,
-      "loss": 46.0,
-      "step": 34960
-    },
-    {
-      "epoch": 2.6730125962880136,
-      "grad_norm": 0.001796469441615045,
-      "learning_rate": 0.00019999647611770912,
-      "loss": 46.0,
-      "step": 34961
-    },
-    {
-      "epoch": 2.6730890532714033,
-      "grad_norm": 0.004304004367440939,
-      "learning_rate": 0.0001999964759160604,
-      "loss": 46.0,
-      "step": 34962
-    },
-    {
-      "epoch": 2.673165510254793,
-      "grad_norm": 0.0023026950657367706,
-      "learning_rate": 0.00019999647571440594,
-      "loss": 46.0,
-      "step": 34963
-    },
-    {
-      "epoch": 2.673241967238183,
-      "grad_norm": 0.001401193207129836,
-      "learning_rate": 0.0001999964755127457,
-      "loss": 46.0,
-      "step": 34964
-    },
-    {
-      "epoch": 2.673318424221572,
-      "grad_norm": 0.0014134080847725272,
-      "learning_rate": 0.00019999647531107967,
-      "loss": 46.0,
-      "step": 34965
-    },
-    {
-      "epoch": 2.673394881204962,
-      "grad_norm": 0.008156070485711098,
-      "learning_rate": 0.0001999964751094079,
-      "loss": 46.0,
-      "step": 34966
-    },
-    {
-      "epoch": 2.6734713381883517,
-      "grad_norm": 0.0025683222338557243,
-      "learning_rate": 0.00019999647490773033,
-      "loss": 46.0,
-      "step": 34967
-    },
-    {
-      "epoch": 2.6735477951717415,
-      "grad_norm": 0.004338476341217756,
-      "learning_rate": 0.000199996474706047,
-      "loss": 46.0,
-      "step": 34968
-    },
-    {
-      "epoch": 2.673624252155131,
-      "grad_norm": 0.0038774064742028713,
-      "learning_rate": 0.00019999647450435793,
-      "loss": 46.0,
-      "step": 34969
-    },
-    {
-      "epoch": 2.673700709138521,
-      "grad_norm": 0.0011908930027857423,
-      "learning_rate": 0.00019999647430266304,
-      "loss": 46.0,
-      "step": 34970
-    },
-    {
-      "epoch": 2.6737771661219107,
-      "grad_norm": 0.001355922082439065,
-      "learning_rate": 0.00019999647410096243,
-      "loss": 46.0,
-      "step": 34971
-    },
-    {
-      "epoch": 2.6738536231053005,
-      "grad_norm": 0.0017022473039105535,
-      "learning_rate": 0.000199996473899256,
-      "loss": 46.0,
-      "step": 34972
-    },
-    {
-      "epoch": 2.6739300800886903,
-      "grad_norm": 0.0013512865407392383,
-      "learning_rate": 0.00019999647369754384,
-      "loss": 46.0,
-      "step": 34973
-    },
-    {
-      "epoch": 2.6740065370720796,
-      "grad_norm": 0.0008856017957441509,
-      "learning_rate": 0.0001999964734958259,
-      "loss": 46.0,
-      "step": 34974
-    },
-    {
-      "epoch": 2.6740829940554693,
-      "grad_norm": 0.003687289310619235,
-      "learning_rate": 0.0001999964732941022,
-      "loss": 46.0,
-      "step": 34975
-    },
-    {
-      "epoch": 2.674159451038859,
-      "grad_norm": 0.0016867963131517172,
-      "learning_rate": 0.00019999647309237273,
-      "loss": 46.0,
-      "step": 34976
-    },
-    {
-      "epoch": 2.674235908022249,
-      "grad_norm": 0.0017421615775674582,
-      "learning_rate": 0.00019999647289063748,
-      "loss": 46.0,
-      "step": 34977
-    },
-    {
-      "epoch": 2.6743123650056386,
-      "grad_norm": 0.0016528844134882092,
-      "learning_rate": 0.00019999647268889646,
-      "loss": 46.0,
-      "step": 34978
-    },
-    {
-      "epoch": 2.6743888219890284,
-      "grad_norm": 0.0010495816823095083,
-      "learning_rate": 0.00019999647248714967,
-      "loss": 46.0,
-      "step": 34979
-    },
-    {
-      "epoch": 2.674465278972418,
-      "grad_norm": 0.005541035905480385,
-      "learning_rate": 0.0001999964722853971,
-      "loss": 46.0,
-      "step": 34980
-    },
-    {
-      "epoch": 2.674541735955808,
-      "grad_norm": 0.0022761805448681116,
-      "learning_rate": 0.00019999647208363878,
-      "loss": 46.0,
-      "step": 34981
-    },
-    {
-      "epoch": 2.6746181929391977,
-      "grad_norm": 0.0017221964662894607,
-      "learning_rate": 0.0001999964718818747,
-      "loss": 46.0,
-      "step": 34982
-    },
-    {
-      "epoch": 2.6746946499225874,
-      "grad_norm": 0.001439619343727827,
-      "learning_rate": 0.00019999647168010483,
-      "loss": 46.0,
-      "step": 34983
-    },
-    {
-      "epoch": 2.674771106905977,
-      "grad_norm": 0.0028904296923428774,
-      "learning_rate": 0.0001999964714783292,
-      "loss": 46.0,
-      "step": 34984
-    },
-    {
-      "epoch": 2.674847563889367,
-      "grad_norm": 0.001968087861314416,
-      "learning_rate": 0.00019999647127654782,
-      "loss": 46.0,
-      "step": 34985
-    },
-    {
-      "epoch": 2.6749240208727567,
-      "grad_norm": 0.0012570518301799893,
-      "learning_rate": 0.00019999647107476063,
-      "loss": 46.0,
-      "step": 34986
-    },
-    {
-      "epoch": 2.675000477856146,
-      "grad_norm": 0.0020639647264033556,
-      "learning_rate": 0.00019999647087296768,
-      "loss": 46.0,
-      "step": 34987
-    },
-    {
-      "epoch": 2.675076934839536,
-      "grad_norm": 0.002605196787044406,
-      "learning_rate": 0.00019999647067116898,
-      "loss": 46.0,
-      "step": 34988
-    },
-    {
-      "epoch": 2.6751533918229256,
-      "grad_norm": 0.000987744890153408,
-      "learning_rate": 0.0001999964704693645,
-      "loss": 46.0,
-      "step": 34989
-    },
-    {
-      "epoch": 2.6752298488063153,
-      "grad_norm": 0.00182677514385432,
-      "learning_rate": 0.00019999647026755426,
-      "loss": 46.0,
-      "step": 34990
-    },
-    {
-      "epoch": 2.675306305789705,
-      "grad_norm": 0.003008483210578561,
-      "learning_rate": 0.00019999647006573823,
-      "loss": 46.0,
-      "step": 34991
-    },
-    {
-      "epoch": 2.675382762773095,
-      "grad_norm": 0.0008762613288126886,
-      "learning_rate": 0.00019999646986391644,
-      "loss": 46.0,
-      "step": 34992
-    },
-    {
-      "epoch": 2.6754592197564846,
-      "grad_norm": 0.002440882846713066,
-      "learning_rate": 0.0001999964696620889,
-      "loss": 46.0,
-      "step": 34993
-    },
-    {
-      "epoch": 2.6755356767398744,
-      "grad_norm": 0.003429112955927849,
-      "learning_rate": 0.00019999646946025558,
-      "loss": 46.0,
-      "step": 34994
-    },
-    {
-      "epoch": 2.6756121337232637,
-      "grad_norm": 0.001544648315757513,
-      "learning_rate": 0.00019999646925841647,
-      "loss": 46.0,
-      "step": 34995
-    },
-    {
-      "epoch": 2.6756885907066534,
-      "grad_norm": 0.002543485490605235,
-      "learning_rate": 0.0001999964690565716,
-      "loss": 46.0,
-      "step": 34996
-    },
-    {
-      "epoch": 2.675765047690043,
-      "grad_norm": 0.0014046822907403111,
-      "learning_rate": 0.000199996468854721,
-      "loss": 46.0,
-      "step": 34997
-    },
-    {
-      "epoch": 2.675841504673433,
-      "grad_norm": 0.001448543625883758,
-      "learning_rate": 0.0001999964686528646,
-      "loss": 46.0,
-      "step": 34998
-    },
-    {
-      "epoch": 2.6759179616568227,
-      "grad_norm": 0.002085628919303417,
-      "learning_rate": 0.0001999964684510024,
-      "loss": 46.0,
-      "step": 34999
-    },
-    {
-      "epoch": 2.6759944186402125,
-      "grad_norm": 0.0050635188817977905,
-      "learning_rate": 0.00019999646824913448,
-      "loss": 46.0,
-      "step": 35000
-    },
-    {
-      "epoch": 2.6760708756236022,
-      "grad_norm": 0.00619416544213891,
-      "learning_rate": 0.00019999646804726075,
-      "loss": 46.0,
-      "step": 35001
-    },
-    {
-      "epoch": 2.676147332606992,
-      "grad_norm": 0.0022781158331781626,
-      "learning_rate": 0.00019999646784538128,
-      "loss": 46.0,
-      "step": 35002
-    },
-    {
-      "epoch": 2.6762237895903818,
-      "grad_norm": 0.0011711466358974576,
-      "learning_rate": 0.00019999646764349603,
-      "loss": 46.0,
-      "step": 35003
-    },
-    {
-      "epoch": 2.6763002465737715,
-      "grad_norm": 0.0022175174672156572,
-      "learning_rate": 0.000199996467441605,
-      "loss": 46.0,
-      "step": 35004
-    },
-    {
-      "epoch": 2.6763767035571613,
-      "grad_norm": 0.0016163773834705353,
-      "learning_rate": 0.00019999646723970824,
-      "loss": 46.0,
-      "step": 35005
-    },
-    {
-      "epoch": 2.676453160540551,
-      "grad_norm": 0.0019132839515805244,
-      "learning_rate": 0.00019999646703780567,
-      "loss": 46.0,
-      "step": 35006
-    },
-    {
-      "epoch": 2.676529617523941,
-      "grad_norm": 0.0010947369737550616,
-      "learning_rate": 0.00019999646683589733,
-      "loss": 46.0,
-      "step": 35007
-    },
-    {
-      "epoch": 2.6766060745073306,
-      "grad_norm": 0.0025650332681834698,
-      "learning_rate": 0.00019999646663398327,
-      "loss": 46.0,
-      "step": 35008
-    },
-    {
-      "epoch": 2.67668253149072,
-      "grad_norm": 0.0014202925376594067,
-      "learning_rate": 0.0001999964664320634,
-      "loss": 46.0,
-      "step": 35009
-    },
-    {
-      "epoch": 2.6767589884741096,
-      "grad_norm": 0.0020427873823791742,
-      "learning_rate": 0.00019999646623013778,
-      "loss": 46.0,
-      "step": 35010
-    },
-    {
-      "epoch": 2.6768354454574994,
-      "grad_norm": 0.008256982080638409,
-      "learning_rate": 0.00019999646602820637,
-      "loss": 46.0,
-      "step": 35011
-    },
-    {
-      "epoch": 2.676911902440889,
-      "grad_norm": 0.002671420807018876,
-      "learning_rate": 0.0001999964658262692,
-      "loss": 46.0,
-      "step": 35012
-    },
-    {
-      "epoch": 2.676988359424279,
-      "grad_norm": 0.001022005220875144,
-      "learning_rate": 0.00019999646562432623,
-      "loss": 46.0,
-      "step": 35013
-    },
-    {
-      "epoch": 2.6770648164076687,
-      "grad_norm": 0.0018111125100404024,
-      "learning_rate": 0.00019999646542237756,
-      "loss": 46.0,
-      "step": 35014
-    },
-    {
-      "epoch": 2.6771412733910585,
-      "grad_norm": 0.0015490794321522117,
-      "learning_rate": 0.00019999646522042306,
-      "loss": 46.0,
-      "step": 35015
-    },
-    {
-      "epoch": 2.677217730374448,
-      "grad_norm": 0.0018617082387208939,
-      "learning_rate": 0.0001999964650184628,
-      "loss": 46.0,
-      "step": 35016
-    },
-    {
-      "epoch": 2.6772941873578375,
-      "grad_norm": 0.0005698470631614327,
-      "learning_rate": 0.0001999964648164968,
-      "loss": 46.0,
-      "step": 35017
-    },
-    {
-      "epoch": 2.6773706443412273,
-      "grad_norm": 0.0016014582943171263,
-      "learning_rate": 0.00019999646461452502,
-      "loss": 46.0,
-      "step": 35018
-    },
-    {
-      "epoch": 2.677447101324617,
-      "grad_norm": 0.0014512023190036416,
-      "learning_rate": 0.00019999646441254748,
-      "loss": 46.0,
-      "step": 35019
-    },
-    {
-      "epoch": 2.677523558308007,
-      "grad_norm": 0.000892001437023282,
-      "learning_rate": 0.00019999646421056414,
-      "loss": 46.0,
-      "step": 35020
-    },
-    {
-      "epoch": 2.6776000152913966,
-      "grad_norm": 0.0026293147820979357,
-      "learning_rate": 0.00019999646400857503,
-      "loss": 46.0,
-      "step": 35021
-    },
-    {
-      "epoch": 2.6776764722747863,
-      "grad_norm": 0.0035947246942669153,
-      "learning_rate": 0.00019999646380658017,
-      "loss": 46.0,
-      "step": 35022
-    },
-    {
-      "epoch": 2.677752929258176,
-      "grad_norm": 0.0032561246771365404,
-      "learning_rate": 0.00019999646360457953,
-      "loss": 46.0,
-      "step": 35023
-    },
-    {
-      "epoch": 2.677829386241566,
-      "grad_norm": 0.005203768610954285,
-      "learning_rate": 0.00019999646340257315,
-      "loss": 46.0,
-      "step": 35024
-    },
-    {
-      "epoch": 2.6779058432249556,
-      "grad_norm": 0.001112635713070631,
-      "learning_rate": 0.00019999646320056097,
-      "loss": 46.0,
-      "step": 35025
-    },
-    {
-      "epoch": 2.6779823002083454,
-      "grad_norm": 0.0029776745941489935,
-      "learning_rate": 0.00019999646299854302,
-      "loss": 46.0,
-      "step": 35026
-    },
-    {
-      "epoch": 2.678058757191735,
-      "grad_norm": 0.0018760369857773185,
-      "learning_rate": 0.00019999646279651932,
-      "loss": 46.0,
-      "step": 35027
-    },
-    {
-      "epoch": 2.678135214175125,
-      "grad_norm": 0.005247129593044519,
-      "learning_rate": 0.00019999646259448984,
-      "loss": 46.0,
-      "step": 35028
-    },
-    {
-      "epoch": 2.6782116711585147,
-      "grad_norm": 0.0019117380725219846,
-      "learning_rate": 0.00019999646239245457,
-      "loss": 46.0,
-      "step": 35029
-    },
-    {
-      "epoch": 2.6782881281419044,
-      "grad_norm": 0.00510834576562047,
-      "learning_rate": 0.00019999646219041357,
-      "loss": 46.0,
-      "step": 35030
-    },
-    {
-      "epoch": 2.6783645851252937,
-      "grad_norm": 0.001243475009687245,
-      "learning_rate": 0.00019999646198836678,
-      "loss": 46.0,
-      "step": 35031
-    },
-    {
-      "epoch": 2.6784410421086835,
-      "grad_norm": 0.00366071704775095,
-      "learning_rate": 0.00019999646178631424,
-      "loss": 46.0,
-      "step": 35032
-    },
-    {
-      "epoch": 2.6785174990920733,
-      "grad_norm": 0.003371445694938302,
-      "learning_rate": 0.0001999964615842559,
-      "loss": 46.0,
-      "step": 35033
-    },
-    {
-      "epoch": 2.678593956075463,
-      "grad_norm": 0.0020993282087147236,
-      "learning_rate": 0.0001999964613821918,
-      "loss": 46.0,
-      "step": 35034
-    },
-    {
-      "epoch": 2.678670413058853,
-      "grad_norm": 0.0020369626581668854,
-      "learning_rate": 0.00019999646118012195,
-      "loss": 46.0,
-      "step": 35035
-    },
-    {
-      "epoch": 2.6787468700422425,
-      "grad_norm": 0.0016634553903713822,
-      "learning_rate": 0.0001999964609780463,
-      "loss": 46.0,
-      "step": 35036
-    },
-    {
-      "epoch": 2.6788233270256323,
-      "grad_norm": 0.0015325468266382813,
-      "learning_rate": 0.0001999964607759649,
-      "loss": 46.0,
-      "step": 35037
-    },
-    {
-      "epoch": 2.678899784009022,
-      "grad_norm": 0.005376342684030533,
-      "learning_rate": 0.00019999646057387773,
-      "loss": 46.0,
-      "step": 35038
-    },
-    {
-      "epoch": 2.6789762409924114,
-      "grad_norm": 0.001149569870904088,
-      "learning_rate": 0.0001999964603717848,
-      "loss": 46.0,
-      "step": 35039
-    },
-    {
-      "epoch": 2.679052697975801,
-      "grad_norm": 0.0020430092699825764,
-      "learning_rate": 0.00019999646016968608,
-      "loss": 46.0,
-      "step": 35040
-    },
-    {
-      "epoch": 2.679129154959191,
-      "grad_norm": 0.0006907646893523633,
-      "learning_rate": 0.0001999964599675816,
-      "loss": 46.0,
-      "step": 35041
-    },
-    {
-      "epoch": 2.6792056119425807,
-      "grad_norm": 0.003037497866898775,
-      "learning_rate": 0.00019999645976547136,
-      "loss": 46.0,
-      "step": 35042
-    },
-    {
-      "epoch": 2.6792820689259704,
-      "grad_norm": 0.0015470362268388271,
-      "learning_rate": 0.00019999645956335536,
-      "loss": 46.0,
-      "step": 35043
-    },
-    {
-      "epoch": 2.67935852590936,
-      "grad_norm": 0.0027011220809072256,
-      "learning_rate": 0.00019999645936123357,
-      "loss": 46.0,
-      "step": 35044
-    },
-    {
-      "epoch": 2.67943498289275,
-      "grad_norm": 0.005984518676996231,
-      "learning_rate": 0.000199996459159106,
-      "loss": 46.0,
-      "step": 35045
-    },
-    {
-      "epoch": 2.6795114398761397,
-      "grad_norm": 0.003937001805752516,
-      "learning_rate": 0.0001999964589569727,
-      "loss": 46.0,
-      "step": 35046
-    },
-    {
-      "epoch": 2.6795878968595295,
-      "grad_norm": 0.002172737615182996,
-      "learning_rate": 0.00019999645875483357,
-      "loss": 46.0,
-      "step": 35047
-    },
-    {
-      "epoch": 2.6796643538429192,
-      "grad_norm": 0.003583714831620455,
-      "learning_rate": 0.00019999645855268871,
-      "loss": 46.0,
-      "step": 35048
-    },
-    {
-      "epoch": 2.679740810826309,
-      "grad_norm": 0.002341628773137927,
-      "learning_rate": 0.00019999645835053808,
-      "loss": 46.0,
-      "step": 35049
-    },
-    {
-      "epoch": 2.6798172678096988,
-      "grad_norm": 0.0005359657225199044,
-      "learning_rate": 0.00019999645814838167,
-      "loss": 46.0,
-      "step": 35050
-    },
-    {
-      "epoch": 2.6798937247930885,
-      "grad_norm": 0.002184459939599037,
-      "learning_rate": 0.0001999964579462195,
-      "loss": 46.0,
-      "step": 35051
-    },
-    {
-      "epoch": 2.6799701817764783,
-      "grad_norm": 0.0036105432081967592,
-      "learning_rate": 0.00019999645774405157,
-      "loss": 46.0,
-      "step": 35052
-    },
-    {
-      "epoch": 2.6800466387598676,
-      "grad_norm": 0.002817615633830428,
-      "learning_rate": 0.00019999645754187787,
-      "loss": 46.0,
-      "step": 35053
-    },
-    {
-      "epoch": 2.6801230957432574,
-      "grad_norm": 0.0012330783065408468,
-      "learning_rate": 0.00019999645733969837,
-      "loss": 46.0,
-      "step": 35054
-    },
-    {
-      "epoch": 2.680199552726647,
-      "grad_norm": 0.0019340860890224576,
-      "learning_rate": 0.00019999645713751312,
-      "loss": 46.0,
-      "step": 35055
-    },
-    {
-      "epoch": 2.680276009710037,
-      "grad_norm": 0.009502885863184929,
-      "learning_rate": 0.0001999964569353221,
-      "loss": 46.0,
-      "step": 35056
-    },
-    {
-      "epoch": 2.6803524666934266,
-      "grad_norm": 0.0009414135129190981,
-      "learning_rate": 0.0001999964567331253,
-      "loss": 46.0,
-      "step": 35057
-    },
-    {
-      "epoch": 2.6804289236768164,
-      "grad_norm": 0.002125833183526993,
-      "learning_rate": 0.00019999645653092277,
-      "loss": 46.0,
-      "step": 35058
-    },
-    {
-      "epoch": 2.680505380660206,
-      "grad_norm": 0.003920309711247683,
-      "learning_rate": 0.00019999645632871445,
-      "loss": 46.0,
-      "step": 35059
-    },
-    {
-      "epoch": 2.680581837643596,
-      "grad_norm": 0.0010468428954482079,
-      "learning_rate": 0.00019999645612650034,
-      "loss": 46.0,
-      "step": 35060
-    },
-    {
-      "epoch": 2.6806582946269852,
-      "grad_norm": 0.0008408360299654305,
-      "learning_rate": 0.00019999645592428048,
-      "loss": 46.0,
-      "step": 35061
-    },
-    {
-      "epoch": 2.680734751610375,
-      "grad_norm": 0.001694794395007193,
-      "learning_rate": 0.00019999645572205482,
-      "loss": 46.0,
-      "step": 35062
-    },
-    {
-      "epoch": 2.6808112085937648,
-      "grad_norm": 0.004552492871880531,
-      "learning_rate": 0.00019999645551982344,
-      "loss": 46.0,
-      "step": 35063
-    },
-    {
-      "epoch": 2.6808876655771545,
-      "grad_norm": 0.0019799016881734133,
-      "learning_rate": 0.0001999964553175863,
-      "loss": 46.0,
-      "step": 35064
-    },
-    {
-      "epoch": 2.6809641225605443,
-      "grad_norm": 0.0021943531464785337,
-      "learning_rate": 0.00019999645511534333,
-      "loss": 46.0,
-      "step": 35065
-    },
-    {
-      "epoch": 2.681040579543934,
-      "grad_norm": 0.002820164430886507,
-      "learning_rate": 0.00019999645491309464,
-      "loss": 46.0,
-      "step": 35066
-    },
-    {
-      "epoch": 2.681117036527324,
-      "grad_norm": 0.00448134820908308,
-      "learning_rate": 0.00019999645471084014,
-      "loss": 46.0,
-      "step": 35067
-    },
-    {
-      "epoch": 2.6811934935107136,
-      "grad_norm": 0.0018064838368445635,
-      "learning_rate": 0.0001999964545085799,
-      "loss": 46.0,
-      "step": 35068
-    },
-    {
-      "epoch": 2.6812699504941033,
-      "grad_norm": 0.0037594132591038942,
-      "learning_rate": 0.0001999964543063139,
-      "loss": 46.0,
-      "step": 35069
-    },
-    {
-      "epoch": 2.681346407477493,
-      "grad_norm": 0.001437538187019527,
-      "learning_rate": 0.00019999645410404208,
-      "loss": 46.0,
-      "step": 35070
-    },
-    {
-      "epoch": 2.681422864460883,
-      "grad_norm": 0.0025133390445262194,
-      "learning_rate": 0.0001999964539017645,
-      "loss": 46.0,
-      "step": 35071
-    },
-    {
-      "epoch": 2.6814993214442726,
-      "grad_norm": 0.0035058173816651106,
-      "learning_rate": 0.00019999645369948122,
-      "loss": 46.0,
-      "step": 35072
-    },
-    {
-      "epoch": 2.6815757784276624,
-      "grad_norm": 0.0013061017962172627,
-      "learning_rate": 0.00019999645349719208,
-      "loss": 46.0,
-      "step": 35073
-    },
-    {
-      "epoch": 2.681652235411052,
-      "grad_norm": 0.0017626304179430008,
-      "learning_rate": 0.00019999645329489725,
-      "loss": 46.0,
-      "step": 35074
-    },
-    {
-      "epoch": 2.6817286923944414,
-      "grad_norm": 0.0030665218364447355,
-      "learning_rate": 0.0001999964530925966,
-      "loss": 46.0,
-      "step": 35075
-    },
-    {
-      "epoch": 2.681805149377831,
-      "grad_norm": 0.0013497495092451572,
-      "learning_rate": 0.0001999964528902902,
-      "loss": 46.0,
-      "step": 35076
-    },
-    {
-      "epoch": 2.681881606361221,
-      "grad_norm": 0.006613880395889282,
-      "learning_rate": 0.00019999645268797803,
-      "loss": 46.0,
-      "step": 35077
-    },
-    {
-      "epoch": 2.6819580633446107,
-      "grad_norm": 0.0018535639392212033,
-      "learning_rate": 0.00019999645248566008,
-      "loss": 46.0,
-      "step": 35078
-    },
-    {
-      "epoch": 2.6820345203280005,
-      "grad_norm": 0.0017786029493436217,
-      "learning_rate": 0.00019999645228333635,
-      "loss": 46.0,
-      "step": 35079
-    },
-    {
-      "epoch": 2.6821109773113903,
-      "grad_norm": 0.0045642368495464325,
-      "learning_rate": 0.0001999964520810069,
-      "loss": 46.0,
-      "step": 35080
-    },
-    {
-      "epoch": 2.68218743429478,
-      "grad_norm": 0.003160534193739295,
-      "learning_rate": 0.00019999645187867166,
-      "loss": 46.0,
-      "step": 35081
-    },
-    {
-      "epoch": 2.6822638912781698,
-      "grad_norm": 0.0013153674080967903,
-      "learning_rate": 0.00019999645167633064,
-      "loss": 46.0,
-      "step": 35082
-    },
-    {
-      "epoch": 2.682340348261559,
-      "grad_norm": 0.001571283326484263,
-      "learning_rate": 0.00019999645147398385,
-      "loss": 46.0,
-      "step": 35083
-    },
-    {
-      "epoch": 2.682416805244949,
-      "grad_norm": 0.002594613702967763,
-      "learning_rate": 0.00019999645127163128,
-      "loss": 46.0,
-      "step": 35084
-    },
-    {
-      "epoch": 2.6824932622283386,
-      "grad_norm": 0.0020451454911381006,
-      "learning_rate": 0.00019999645106927297,
-      "loss": 46.0,
-      "step": 35085
-    },
-    {
-      "epoch": 2.6825697192117284,
-      "grad_norm": 0.004231937695294619,
-      "learning_rate": 0.00019999645086690886,
-      "loss": 46.0,
-      "step": 35086
-    },
-    {
-      "epoch": 2.682646176195118,
-      "grad_norm": 0.003918634261935949,
-      "learning_rate": 0.000199996450664539,
-      "loss": 46.0,
-      "step": 35087
-    },
-    {
-      "epoch": 2.682722633178508,
-      "grad_norm": 0.001403136528097093,
-      "learning_rate": 0.00019999645046216337,
-      "loss": 46.0,
-      "step": 35088
-    },
-    {
-      "epoch": 2.6827990901618977,
-      "grad_norm": 0.002732867607846856,
-      "learning_rate": 0.000199996450259782,
-      "loss": 46.0,
-      "step": 35089
-    },
-    {
-      "epoch": 2.6828755471452874,
-      "grad_norm": 0.0022099416237324476,
-      "learning_rate": 0.0001999964500573948,
-      "loss": 46.0,
-      "step": 35090
-    },
-    {
-      "epoch": 2.682952004128677,
-      "grad_norm": 0.0009359550895169377,
-      "learning_rate": 0.00019999644985500186,
-      "loss": 46.0,
-      "step": 35091
-    },
-    {
-      "epoch": 2.683028461112067,
-      "grad_norm": 0.0019647094886749983,
-      "learning_rate": 0.00019999644965260313,
-      "loss": 46.0,
-      "step": 35092
-    },
-    {
-      "epoch": 2.6831049180954567,
-      "grad_norm": 0.0013681192649528384,
-      "learning_rate": 0.00019999644945019866,
-      "loss": 46.0,
-      "step": 35093
-    },
-    {
-      "epoch": 2.6831813750788465,
-      "grad_norm": 0.0019265684532001615,
-      "learning_rate": 0.0001999964492477884,
-      "loss": 46.0,
-      "step": 35094
-    },
-    {
-      "epoch": 2.6832578320622362,
-      "grad_norm": 0.0015732484171167016,
-      "learning_rate": 0.0001999964490453724,
-      "loss": 46.0,
-      "step": 35095
-    },
-    {
-      "epoch": 2.6833342890456255,
-      "grad_norm": 0.0016988591523841023,
-      "learning_rate": 0.0001999964488429506,
-      "loss": 46.0,
-      "step": 35096
-    },
-    {
-      "epoch": 2.6834107460290153,
-      "grad_norm": 0.0039989701472222805,
-      "learning_rate": 0.00019999644864052306,
-      "loss": 46.0,
-      "step": 35097
-    },
-    {
-      "epoch": 2.683487203012405,
-      "grad_norm": 0.001421229331754148,
-      "learning_rate": 0.00019999644843808972,
-      "loss": 46.0,
-      "step": 35098
-    },
-    {
-      "epoch": 2.683563659995795,
-      "grad_norm": 0.0016833083936944604,
-      "learning_rate": 0.00019999644823565064,
-      "loss": 46.0,
-      "step": 35099
-    },
-    {
-      "epoch": 2.6836401169791846,
-      "grad_norm": 0.0023268675431609154,
-      "learning_rate": 0.00019999644803320575,
-      "loss": 46.0,
-      "step": 35100
-    },
-    {
-      "epoch": 2.6837165739625743,
-      "grad_norm": 0.0021617510356009007,
-      "learning_rate": 0.00019999644783075512,
-      "loss": 46.0,
-      "step": 35101
-    },
-    {
-      "epoch": 2.683793030945964,
-      "grad_norm": 0.003536701900884509,
-      "learning_rate": 0.0001999964476282987,
-      "loss": 46.0,
-      "step": 35102
-    },
-    {
-      "epoch": 2.683869487929354,
-      "grad_norm": 0.0004385752836242318,
-      "learning_rate": 0.00019999644742583653,
-      "loss": 46.0,
-      "step": 35103
-    },
-    {
-      "epoch": 2.6839459449127436,
-      "grad_norm": 0.0013113029999658465,
-      "learning_rate": 0.0001999964472233686,
-      "loss": 46.0,
-      "step": 35104
-    },
-    {
-      "epoch": 2.684022401896133,
-      "grad_norm": 0.0024137503933161497,
-      "learning_rate": 0.0001999964470208949,
-      "loss": 46.0,
-      "step": 35105
-    },
-    {
-      "epoch": 2.6840988588795227,
-      "grad_norm": 0.0028730938211083412,
-      "learning_rate": 0.0001999964468184154,
-      "loss": 46.0,
-      "step": 35106
-    },
-    {
-      "epoch": 2.6841753158629125,
-      "grad_norm": 0.0022919427137821913,
-      "learning_rate": 0.00019999644661593017,
-      "loss": 46.0,
-      "step": 35107
-    },
-    {
-      "epoch": 2.6842517728463022,
-      "grad_norm": 0.0031354199163615704,
-      "learning_rate": 0.00019999644641343912,
-      "loss": 46.0,
-      "step": 35108
-    },
-    {
-      "epoch": 2.684328229829692,
-      "grad_norm": 0.0026893760077655315,
-      "learning_rate": 0.00019999644621094236,
-      "loss": 46.0,
-      "step": 35109
-    },
-    {
-      "epoch": 2.6844046868130818,
-      "grad_norm": 0.00113083200994879,
-      "learning_rate": 0.00019999644600843982,
-      "loss": 46.0,
-      "step": 35110
-    },
-    {
-      "epoch": 2.6844811437964715,
-      "grad_norm": 0.0008158150012604892,
-      "learning_rate": 0.00019999644580593148,
-      "loss": 46.0,
-      "step": 35111
-    },
-    {
-      "epoch": 2.6845576007798613,
-      "grad_norm": 0.0029527395963668823,
-      "learning_rate": 0.0001999964456034174,
-      "loss": 46.0,
-      "step": 35112
-    },
-    {
-      "epoch": 2.684634057763251,
-      "grad_norm": 0.004799459595233202,
-      "learning_rate": 0.00019999644540089753,
-      "loss": 46.0,
-      "step": 35113
-    },
-    {
-      "epoch": 2.684710514746641,
-      "grad_norm": 0.0021030528005212545,
-      "learning_rate": 0.00019999644519837188,
-      "loss": 46.0,
-      "step": 35114
-    },
-    {
-      "epoch": 2.6847869717300306,
-      "grad_norm": 0.004186443984508514,
-      "learning_rate": 0.00019999644499584047,
-      "loss": 46.0,
-      "step": 35115
-    },
-    {
-      "epoch": 2.6848634287134203,
-      "grad_norm": 0.003666236298158765,
-      "learning_rate": 0.00019999644479330332,
-      "loss": 46.0,
-      "step": 35116
-    },
-    {
-      "epoch": 2.68493988569681,
-      "grad_norm": 0.0019482123898342252,
-      "learning_rate": 0.00019999644459076037,
-      "loss": 46.0,
-      "step": 35117
-    },
-    {
-      "epoch": 2.6850163426801994,
-      "grad_norm": 0.010220256634056568,
-      "learning_rate": 0.00019999644438821164,
-      "loss": 46.0,
-      "step": 35118
-    },
-    {
-      "epoch": 2.685092799663589,
-      "grad_norm": 0.0038749543018639088,
-      "learning_rate": 0.00019999644418565717,
-      "loss": 46.0,
-      "step": 35119
-    },
-    {
-      "epoch": 2.685169256646979,
-      "grad_norm": 0.0007671939674764872,
-      "learning_rate": 0.00019999644398309692,
-      "loss": 46.0,
-      "step": 35120
-    },
-    {
-      "epoch": 2.6852457136303687,
-      "grad_norm": 0.005448718089610338,
-      "learning_rate": 0.00019999644378053088,
-      "loss": 46.0,
-      "step": 35121
-    },
-    {
-      "epoch": 2.6853221706137584,
-      "grad_norm": 0.0010346717899665236,
-      "learning_rate": 0.00019999644357795911,
-      "loss": 46.0,
-      "step": 35122
-    },
-    {
-      "epoch": 2.685398627597148,
-      "grad_norm": 0.001677076448686421,
-      "learning_rate": 0.00019999644337538155,
-      "loss": 46.0,
-      "step": 35123
-    },
-    {
-      "epoch": 2.685475084580538,
-      "grad_norm": 0.004414484370499849,
-      "learning_rate": 0.0001999964431727982,
-      "loss": 46.0,
-      "step": 35124
-    },
-    {
-      "epoch": 2.6855515415639277,
-      "grad_norm": 0.0010584424016997218,
-      "learning_rate": 0.00019999644297020913,
-      "loss": 46.0,
-      "step": 35125
-    },
-    {
-      "epoch": 2.685627998547317,
-      "grad_norm": 0.0024583705235272646,
-      "learning_rate": 0.00019999644276761427,
-      "loss": 46.0,
-      "step": 35126
-    },
-    {
-      "epoch": 2.685704455530707,
-      "grad_norm": 0.0018608884420245886,
-      "learning_rate": 0.00019999644256501364,
-      "loss": 46.0,
-      "step": 35127
-    },
-    {
-      "epoch": 2.6857809125140966,
-      "grad_norm": 0.004331204108893871,
-      "learning_rate": 0.00019999644236240723,
-      "loss": 46.0,
-      "step": 35128
-    },
-    {
-      "epoch": 2.6858573694974863,
-      "grad_norm": 0.0006679832586087286,
-      "learning_rate": 0.00019999644215979505,
-      "loss": 46.0,
-      "step": 35129
-    },
-    {
-      "epoch": 2.685933826480876,
-      "grad_norm": 0.003459856379777193,
-      "learning_rate": 0.0001999964419571771,
-      "loss": 46.0,
-      "step": 35130
-    },
-    {
-      "epoch": 2.686010283464266,
-      "grad_norm": 0.0014105868758633733,
-      "learning_rate": 0.0001999964417545534,
-      "loss": 46.0,
-      "step": 35131
-    },
-    {
-      "epoch": 2.6860867404476556,
-      "grad_norm": 0.0034913215786218643,
-      "learning_rate": 0.00019999644155192393,
-      "loss": 46.0,
-      "step": 35132
-    },
-    {
-      "epoch": 2.6861631974310454,
-      "grad_norm": 0.0031154786702245474,
-      "learning_rate": 0.00019999644134928866,
-      "loss": 46.0,
-      "step": 35133
-    },
-    {
-      "epoch": 2.686239654414435,
-      "grad_norm": 0.003329252591356635,
-      "learning_rate": 0.00019999644114664764,
-      "loss": 46.0,
-      "step": 35134
-    },
-    {
-      "epoch": 2.686316111397825,
-      "grad_norm": 0.0034857955761253834,
-      "learning_rate": 0.00019999644094400085,
-      "loss": 46.0,
-      "step": 35135
-    },
-    {
-      "epoch": 2.6863925683812147,
-      "grad_norm": 0.0012331334874033928,
-      "learning_rate": 0.00019999644074134828,
-      "loss": 46.0,
-      "step": 35136
-    },
-    {
-      "epoch": 2.6864690253646044,
-      "grad_norm": 0.0033851582556962967,
-      "learning_rate": 0.00019999644053868995,
-      "loss": 46.0,
-      "step": 35137
-    },
-    {
-      "epoch": 2.686545482347994,
-      "grad_norm": 0.0023544603027403355,
-      "learning_rate": 0.0001999964403360259,
-      "loss": 46.0,
-      "step": 35138
-    },
-    {
-      "epoch": 2.686621939331384,
-      "grad_norm": 0.002336224541068077,
-      "learning_rate": 0.000199996440133356,
-      "loss": 46.0,
-      "step": 35139
-    },
-    {
-      "epoch": 2.6866983963147733,
-      "grad_norm": 0.004990197252482176,
-      "learning_rate": 0.00019999643993068037,
-      "loss": 46.0,
-      "step": 35140
-    },
-    {
-      "epoch": 2.686774853298163,
-      "grad_norm": 0.0011775752063840628,
-      "learning_rate": 0.00019999643972799897,
-      "loss": 46.0,
-      "step": 35141
-    },
-    {
-      "epoch": 2.6868513102815528,
-      "grad_norm": 0.004172973334789276,
-      "learning_rate": 0.0001999964395253118,
-      "loss": 46.0,
-      "step": 35142
-    },
-    {
-      "epoch": 2.6869277672649425,
-      "grad_norm": 0.0011100547853857279,
-      "learning_rate": 0.00019999643932261884,
-      "loss": 46.0,
-      "step": 35143
-    },
-    {
-      "epoch": 2.6870042242483323,
-      "grad_norm": 0.003154816571623087,
-      "learning_rate": 0.00019999643911992014,
-      "loss": 46.0,
-      "step": 35144
-    },
-    {
-      "epoch": 2.687080681231722,
-      "grad_norm": 0.0008737738244235516,
-      "learning_rate": 0.00019999643891721567,
-      "loss": 46.0,
-      "step": 35145
-    },
-    {
-      "epoch": 2.687157138215112,
-      "grad_norm": 0.0017895514611154795,
-      "learning_rate": 0.00019999643871450543,
-      "loss": 46.0,
-      "step": 35146
-    },
-    {
-      "epoch": 2.6872335951985016,
-      "grad_norm": 0.0014592085499316454,
-      "learning_rate": 0.0001999964385117894,
-      "loss": 46.0,
-      "step": 35147
-    },
-    {
-      "epoch": 2.687310052181891,
-      "grad_norm": 0.003521059639751911,
-      "learning_rate": 0.0001999964383090676,
-      "loss": 46.0,
-      "step": 35148
-    },
-    {
-      "epoch": 2.6873865091652807,
-      "grad_norm": 0.0019142618402838707,
-      "learning_rate": 0.00019999643810634003,
-      "loss": 46.0,
-      "step": 35149
-    },
-    {
-      "epoch": 2.6874629661486704,
-      "grad_norm": 0.0029140296392142773,
-      "learning_rate": 0.00019999643790360671,
-      "loss": 46.0,
-      "step": 35150
-    },
-    {
-      "epoch": 2.68753942313206,
-      "grad_norm": 0.001297891023568809,
-      "learning_rate": 0.00019999643770086763,
-      "loss": 46.0,
-      "step": 35151
-    },
-    {
-      "epoch": 2.68761588011545,
-      "grad_norm": 0.0015747452853247523,
-      "learning_rate": 0.00019999643749812275,
-      "loss": 46.0,
-      "step": 35152
-    },
-    {
-      "epoch": 2.6876923370988397,
-      "grad_norm": 0.0018668327247723937,
-      "learning_rate": 0.00019999643729537211,
-      "loss": 46.0,
-      "step": 35153
-    },
-    {
-      "epoch": 2.6877687940822295,
-      "grad_norm": 0.0010680914856493473,
-      "learning_rate": 0.00019999643709261574,
-      "loss": 46.0,
-      "step": 35154
-    },
-    {
-      "epoch": 2.6878452510656192,
-      "grad_norm": 0.002542471047490835,
-      "learning_rate": 0.00019999643688985353,
-      "loss": 46.0,
-      "step": 35155
-    },
-    {
-      "epoch": 2.687921708049009,
-      "grad_norm": 0.0011306102387607098,
-      "learning_rate": 0.00019999643668708558,
-      "loss": 46.0,
-      "step": 35156
-    },
-    {
-      "epoch": 2.6879981650323987,
-      "grad_norm": 0.0032563714776188135,
-      "learning_rate": 0.00019999643648431188,
-      "loss": 46.0,
-      "step": 35157
-    },
-    {
-      "epoch": 2.6880746220157885,
-      "grad_norm": 0.002381636993959546,
-      "learning_rate": 0.00019999643628153239,
-      "loss": 46.0,
-      "step": 35158
-    },
-    {
-      "epoch": 2.6881510789991783,
-      "grad_norm": 0.0018264978425577283,
-      "learning_rate": 0.00019999643607874714,
-      "loss": 46.0,
-      "step": 35159
-    },
-    {
-      "epoch": 2.688227535982568,
-      "grad_norm": 0.0035214636009186506,
-      "learning_rate": 0.00019999643587595612,
-      "loss": 46.0,
-      "step": 35160
-    },
-    {
-      "epoch": 2.688303992965958,
-      "grad_norm": 0.0015457180561497808,
-      "learning_rate": 0.00019999643567315936,
-      "loss": 46.0,
-      "step": 35161
-    },
-    {
-      "epoch": 2.688380449949347,
-      "grad_norm": 0.0030938242562115192,
-      "learning_rate": 0.00019999643547035677,
-      "loss": 46.0,
-      "step": 35162
-    },
-    {
-      "epoch": 2.688456906932737,
-      "grad_norm": 0.0020048683509230614,
-      "learning_rate": 0.00019999643526754846,
-      "loss": 46.0,
-      "step": 35163
-    },
-    {
-      "epoch": 2.6885333639161266,
-      "grad_norm": 0.0010409500682726502,
-      "learning_rate": 0.00019999643506473435,
-      "loss": 46.0,
-      "step": 35164
-    },
-    {
-      "epoch": 2.6886098208995164,
-      "grad_norm": 0.0026394836604595184,
-      "learning_rate": 0.00019999643486191452,
-      "loss": 46.0,
-      "step": 35165
-    },
-    {
-      "epoch": 2.688686277882906,
-      "grad_norm": 0.0027438250835984945,
-      "learning_rate": 0.00019999643465908886,
-      "loss": 46.0,
-      "step": 35166
-    },
-    {
-      "epoch": 2.688762734866296,
-      "grad_norm": 0.0018162140622735023,
-      "learning_rate": 0.00019999643445625743,
-      "loss": 46.0,
-      "step": 35167
-    },
-    {
-      "epoch": 2.6888391918496857,
-      "grad_norm": 0.004773767665028572,
-      "learning_rate": 0.00019999643425342028,
-      "loss": 46.0,
-      "step": 35168
-    },
-    {
-      "epoch": 2.6889156488330754,
-      "grad_norm": 0.002019051695242524,
-      "learning_rate": 0.00019999643405057733,
-      "loss": 46.0,
-      "step": 35169
-    },
-    {
-      "epoch": 2.6889921058164648,
-      "grad_norm": 0.001027501537464559,
-      "learning_rate": 0.0001999964338477286,
-      "loss": 46.0,
-      "step": 35170
-    },
-    {
-      "epoch": 2.6890685627998545,
-      "grad_norm": 0.001205040025524795,
-      "learning_rate": 0.00019999643364487416,
-      "loss": 46.0,
-      "step": 35171
-    },
-    {
-      "epoch": 2.6891450197832443,
-      "grad_norm": 0.0021995757706463337,
-      "learning_rate": 0.00019999643344201392,
-      "loss": 46.0,
-      "step": 35172
-    },
-    {
-      "epoch": 2.689221476766634,
-      "grad_norm": 0.0010647772578522563,
-      "learning_rate": 0.00019999643323914787,
-      "loss": 46.0,
-      "step": 35173
-    },
-    {
-      "epoch": 2.689297933750024,
-      "grad_norm": 0.0008045045542530715,
-      "learning_rate": 0.00019999643303627608,
-      "loss": 46.0,
-      "step": 35174
-    },
-    {
-      "epoch": 2.6893743907334136,
-      "grad_norm": 0.0009922184981405735,
-      "learning_rate": 0.00019999643283339852,
-      "loss": 46.0,
-      "step": 35175
-    },
-    {
-      "epoch": 2.6894508477168033,
-      "grad_norm": 0.0014489333843812346,
-      "learning_rate": 0.0001999964326305152,
-      "loss": 46.0,
-      "step": 35176
-    },
-    {
-      "epoch": 2.689527304700193,
-      "grad_norm": 0.0020227523054927588,
-      "learning_rate": 0.0001999964324276261,
-      "loss": 46.0,
-      "step": 35177
-    },
-    {
-      "epoch": 2.689603761683583,
-      "grad_norm": 0.001468850881792605,
-      "learning_rate": 0.00019999643222473124,
-      "loss": 46.0,
-      "step": 35178
-    },
-    {
-      "epoch": 2.6896802186669726,
-      "grad_norm": 0.003612266853451729,
-      "learning_rate": 0.0001999964320218306,
-      "loss": 46.0,
-      "step": 35179
-    },
-    {
-      "epoch": 2.6897566756503624,
-      "grad_norm": 0.0022340486757457256,
-      "learning_rate": 0.0001999964318189242,
-      "loss": 46.0,
-      "step": 35180
-    },
-    {
-      "epoch": 2.689833132633752,
-      "grad_norm": 0.004980416037142277,
-      "learning_rate": 0.00019999643161601203,
-      "loss": 46.0,
-      "step": 35181
-    },
-    {
-      "epoch": 2.689909589617142,
-      "grad_norm": 0.003069869475439191,
-      "learning_rate": 0.00019999643141309408,
-      "loss": 46.0,
-      "step": 35182
-    },
-    {
-      "epoch": 2.6899860466005316,
-      "grad_norm": 0.0022921806667000055,
-      "learning_rate": 0.00019999643121017035,
-      "loss": 46.0,
-      "step": 35183
-    },
-    {
-      "epoch": 2.690062503583921,
-      "grad_norm": 0.0012576646404340863,
-      "learning_rate": 0.00019999643100724088,
-      "loss": 46.0,
-      "step": 35184
-    },
-    {
-      "epoch": 2.6901389605673107,
-      "grad_norm": 0.0012805479345843196,
-      "learning_rate": 0.00019999643080430564,
-      "loss": 46.0,
-      "step": 35185
-    },
-    {
-      "epoch": 2.6902154175507005,
-      "grad_norm": 0.004034394398331642,
-      "learning_rate": 0.00019999643060136462,
-      "loss": 46.0,
-      "step": 35186
-    },
-    {
-      "epoch": 2.6902918745340902,
-      "grad_norm": 0.0019110506400465965,
-      "learning_rate": 0.00019999643039841783,
-      "loss": 46.0,
-      "step": 35187
-    },
-    {
-      "epoch": 2.69036833151748,
-      "grad_norm": 0.005394344218075275,
-      "learning_rate": 0.00019999643019546524,
-      "loss": 46.0,
-      "step": 35188
-    },
-    {
-      "epoch": 2.6904447885008698,
-      "grad_norm": 0.003047197125852108,
-      "learning_rate": 0.0001999964299925069,
-      "loss": 46.0,
-      "step": 35189
-    },
-    {
-      "epoch": 2.6905212454842595,
-      "grad_norm": 0.001858064904808998,
-      "learning_rate": 0.00019999642978954282,
-      "loss": 46.0,
-      "step": 35190
-    },
-    {
-      "epoch": 2.6905977024676493,
-      "grad_norm": 0.002759219380095601,
-      "learning_rate": 0.00019999642958657297,
-      "loss": 46.0,
-      "step": 35191
-    },
-    {
-      "epoch": 2.6906741594510386,
-      "grad_norm": 0.0016702038701623678,
-      "learning_rate": 0.00019999642938359734,
-      "loss": 46.0,
-      "step": 35192
-    },
-    {
-      "epoch": 2.6907506164344284,
-      "grad_norm": 0.005024959798902273,
-      "learning_rate": 0.0001999964291806159,
-      "loss": 46.0,
-      "step": 35193
-    },
-    {
-      "epoch": 2.690827073417818,
-      "grad_norm": 0.0030106569174677134,
-      "learning_rate": 0.00019999642897762876,
-      "loss": 46.0,
-      "step": 35194
-    },
-    {
-      "epoch": 2.690903530401208,
-      "grad_norm": 0.0036216035950928926,
-      "learning_rate": 0.00019999642877463583,
-      "loss": 46.0,
-      "step": 35195
-    },
-    {
-      "epoch": 2.6909799873845976,
-      "grad_norm": 0.0009093974367715418,
-      "learning_rate": 0.00019999642857163708,
-      "loss": 46.0,
-      "step": 35196
-    },
-    {
-      "epoch": 2.6910564443679874,
-      "grad_norm": 0.0031241197139024734,
-      "learning_rate": 0.0001999964283686326,
-      "loss": 46.0,
-      "step": 35197
-    },
-    {
-      "epoch": 2.691132901351377,
-      "grad_norm": 0.0017396444454789162,
-      "learning_rate": 0.00019999642816562237,
-      "loss": 46.0,
-      "step": 35198
-    },
-    {
-      "epoch": 2.691209358334767,
-      "grad_norm": 0.0011495795333757997,
-      "learning_rate": 0.00019999642796260635,
-      "loss": 46.0,
-      "step": 35199
-    },
-    {
-      "epoch": 2.6912858153181567,
-      "grad_norm": 0.002570424461737275,
-      "learning_rate": 0.00019999642775958456,
-      "loss": 46.0,
-      "step": 35200
-    },
-    {
-      "epoch": 2.6913622723015465,
-      "grad_norm": 0.002257262822240591,
-      "learning_rate": 0.000199996427556557,
-      "loss": 46.0,
-      "step": 35201
-    },
-    {
-      "epoch": 2.691438729284936,
-      "grad_norm": 0.0009384382283315063,
-      "learning_rate": 0.00019999642735352366,
-      "loss": 46.0,
-      "step": 35202
-    },
-    {
-      "epoch": 2.691515186268326,
-      "grad_norm": 0.0033902712166309357,
-      "learning_rate": 0.00019999642715048455,
-      "loss": 46.0,
-      "step": 35203
-    },
-    {
-      "epoch": 2.6915916432517157,
-      "grad_norm": 0.0037153740413486958,
-      "learning_rate": 0.0001999964269474397,
-      "loss": 46.0,
-      "step": 35204
-    },
-    {
-      "epoch": 2.6916681002351055,
-      "grad_norm": 0.0019718408584594727,
-      "learning_rate": 0.00019999642674438907,
-      "loss": 46.0,
-      "step": 35205
-    },
-    {
-      "epoch": 2.691744557218495,
-      "grad_norm": 0.0008516894886270165,
-      "learning_rate": 0.00019999642654133266,
-      "loss": 46.0,
-      "step": 35206
-    },
-    {
-      "epoch": 2.6918210142018846,
-      "grad_norm": 0.0014777202159166336,
-      "learning_rate": 0.0001999964263382705,
-      "loss": 46.0,
-      "step": 35207
-    },
-    {
-      "epoch": 2.6918974711852743,
-      "grad_norm": 0.0006840305286459625,
-      "learning_rate": 0.00019999642613520254,
-      "loss": 46.0,
-      "step": 35208
-    },
-    {
-      "epoch": 2.691973928168664,
-      "grad_norm": 0.003831393551081419,
-      "learning_rate": 0.00019999642593212884,
-      "loss": 46.0,
-      "step": 35209
-    },
-    {
-      "epoch": 2.692050385152054,
-      "grad_norm": 0.003159074578434229,
-      "learning_rate": 0.00019999642572904935,
-      "loss": 46.0,
-      "step": 35210
-    },
-    {
-      "epoch": 2.6921268421354436,
-      "grad_norm": 0.0026791805867105722,
-      "learning_rate": 0.0001999964255259641,
-      "loss": 46.0,
-      "step": 35211
-    },
-    {
-      "epoch": 2.6922032991188334,
-      "grad_norm": 0.006213878747075796,
-      "learning_rate": 0.00019999642532287309,
-      "loss": 46.0,
-      "step": 35212
-    },
-    {
-      "epoch": 2.692279756102223,
-      "grad_norm": 0.001594706904143095,
-      "learning_rate": 0.0001999964251197763,
-      "loss": 46.0,
-      "step": 35213
-    },
-    {
-      "epoch": 2.6923562130856125,
-      "grad_norm": 0.0015687403501942754,
-      "learning_rate": 0.00019999642491667373,
-      "loss": 46.0,
-      "step": 35214
-    },
-    {
-      "epoch": 2.692432670069002,
-      "grad_norm": 0.0021522280294448137,
-      "learning_rate": 0.00019999642471356542,
-      "loss": 46.0,
-      "step": 35215
-    },
-    {
-      "epoch": 2.692509127052392,
-      "grad_norm": 0.0032415541354566813,
-      "learning_rate": 0.00019999642451045131,
-      "loss": 46.0,
-      "step": 35216
-    },
-    {
-      "epoch": 2.6925855840357817,
-      "grad_norm": 0.007459606509655714,
-      "learning_rate": 0.00019999642430733143,
-      "loss": 46.0,
-      "step": 35217
-    },
-    {
-      "epoch": 2.6926620410191715,
-      "grad_norm": 0.0024852720089256763,
-      "learning_rate": 0.00019999642410420577,
-      "loss": 46.0,
-      "step": 35218
-    },
-    {
-      "epoch": 2.6927384980025613,
-      "grad_norm": 0.0013977159978821874,
-      "learning_rate": 0.0001999964239010744,
-      "loss": 46.0,
-      "step": 35219
-    },
-    {
-      "epoch": 2.692814954985951,
-      "grad_norm": 0.0006656849291175604,
-      "learning_rate": 0.0001999964236979372,
-      "loss": 46.0,
-      "step": 35220
-    },
-    {
-      "epoch": 2.692891411969341,
-      "grad_norm": 0.0030040808487683535,
-      "learning_rate": 0.00019999642349479427,
-      "loss": 46.0,
-      "step": 35221
-    },
-    {
-      "epoch": 2.6929678689527305,
-      "grad_norm": 0.0021106444764882326,
-      "learning_rate": 0.00019999642329164555,
-      "loss": 46.0,
-      "step": 35222
-    },
-    {
-      "epoch": 2.6930443259361203,
-      "grad_norm": 0.0013050640700384974,
-      "learning_rate": 0.00019999642308849108,
-      "loss": 46.0,
-      "step": 35223
-    },
-    {
-      "epoch": 2.69312078291951,
-      "grad_norm": 0.001634501852095127,
-      "learning_rate": 0.0001999964228853308,
-      "loss": 46.0,
-      "step": 35224
-    },
-    {
-      "epoch": 2.6931972399029,
-      "grad_norm": 0.005037304479628801,
-      "learning_rate": 0.0001999964226821648,
-      "loss": 46.0,
-      "step": 35225
-    },
-    {
-      "epoch": 2.6932736968862896,
-      "grad_norm": 0.0028133774176239967,
-      "learning_rate": 0.000199996422478993,
-      "loss": 46.0,
-      "step": 35226
-    },
-    {
-      "epoch": 2.693350153869679,
-      "grad_norm": 0.0013189658056944609,
-      "learning_rate": 0.00019999642227581544,
-      "loss": 46.0,
-      "step": 35227
-    },
-    {
-      "epoch": 2.6934266108530687,
-      "grad_norm": 0.003289174987003207,
-      "learning_rate": 0.0001999964220726321,
-      "loss": 46.0,
-      "step": 35228
-    },
-    {
-      "epoch": 2.6935030678364584,
-      "grad_norm": 0.0027267406694591045,
-      "learning_rate": 0.00019999642186944303,
-      "loss": 46.0,
-      "step": 35229
-    },
-    {
-      "epoch": 2.693579524819848,
-      "grad_norm": 0.0015421024290844798,
-      "learning_rate": 0.00019999642166624817,
-      "loss": 46.0,
-      "step": 35230
-    },
-    {
-      "epoch": 2.693655981803238,
-      "grad_norm": 0.00500150490552187,
-      "learning_rate": 0.00019999642146304751,
-      "loss": 46.0,
-      "step": 35231
-    },
-    {
-      "epoch": 2.6937324387866277,
-      "grad_norm": 0.0019818164873868227,
-      "learning_rate": 0.00019999642125984114,
-      "loss": 46.0,
-      "step": 35232
-    },
-    {
-      "epoch": 2.6938088957700175,
-      "grad_norm": 0.0008503650315105915,
-      "learning_rate": 0.00019999642105662894,
-      "loss": 46.0,
-      "step": 35233
-    },
-    {
-      "epoch": 2.6938853527534072,
-      "grad_norm": 0.0012651070719584823,
-      "learning_rate": 0.000199996420853411,
-      "loss": 46.0,
-      "step": 35234
-    },
-    {
-      "epoch": 2.693961809736797,
-      "grad_norm": 0.0013718416448682547,
-      "learning_rate": 0.0001999964206501873,
-      "loss": 46.0,
-      "step": 35235
-    },
-    {
-      "epoch": 2.6940382667201863,
-      "grad_norm": 0.0012118752347305417,
-      "learning_rate": 0.00019999642044695782,
-      "loss": 46.0,
-      "step": 35236
-    },
-    {
-      "epoch": 2.694114723703576,
-      "grad_norm": 0.0019576693885028362,
-      "learning_rate": 0.00019999642024372255,
-      "loss": 46.0,
-      "step": 35237
-    },
-    {
-      "epoch": 2.694191180686966,
-      "grad_norm": 0.0013692042557522655,
-      "learning_rate": 0.00019999642004048154,
-      "loss": 46.0,
-      "step": 35238
-    },
-    {
-      "epoch": 2.6942676376703556,
-      "grad_norm": 0.001424482325091958,
-      "learning_rate": 0.00019999641983723475,
-      "loss": 46.0,
-      "step": 35239
-    },
-    {
-      "epoch": 2.6943440946537454,
-      "grad_norm": 0.0012743906117975712,
-      "learning_rate": 0.00019999641963398216,
-      "loss": 46.0,
-      "step": 35240
-    },
-    {
-      "epoch": 2.694420551637135,
-      "grad_norm": 0.001204194501042366,
-      "learning_rate": 0.00019999641943072385,
-      "loss": 46.0,
-      "step": 35241
-    },
-    {
-      "epoch": 2.694497008620525,
-      "grad_norm": 0.003842654637992382,
-      "learning_rate": 0.00019999641922745974,
-      "loss": 46.0,
-      "step": 35242
-    },
-    {
-      "epoch": 2.6945734656039146,
-      "grad_norm": 0.004496491048485041,
-      "learning_rate": 0.00019999641902418989,
-      "loss": 46.0,
-      "step": 35243
-    },
-    {
-      "epoch": 2.6946499225873044,
-      "grad_norm": 0.006841009948402643,
-      "learning_rate": 0.00019999641882091426,
-      "loss": 46.0,
-      "step": 35244
-    },
-    {
-      "epoch": 2.694726379570694,
-      "grad_norm": 0.0011372884036973119,
-      "learning_rate": 0.00019999641861763285,
-      "loss": 46.0,
-      "step": 35245
-    },
-    {
-      "epoch": 2.694802836554084,
-      "grad_norm": 0.003786845598369837,
-      "learning_rate": 0.00019999641841434565,
-      "loss": 46.0,
-      "step": 35246
-    },
-    {
-      "epoch": 2.6948792935374737,
-      "grad_norm": 0.0032982053235173225,
-      "learning_rate": 0.0001999964182110527,
-      "loss": 46.0,
-      "step": 35247
-    },
-    {
-      "epoch": 2.6949557505208634,
-      "grad_norm": 0.002931008581072092,
-      "learning_rate": 0.00019999641800775398,
-      "loss": 46.0,
-      "step": 35248
-    },
-    {
-      "epoch": 2.6950322075042528,
-      "grad_norm": 0.001842274097725749,
-      "learning_rate": 0.0001999964178044495,
-      "loss": 46.0,
-      "step": 35249
-    },
-    {
-      "epoch": 2.6951086644876425,
-      "grad_norm": 0.0025896658189594746,
-      "learning_rate": 0.00019999641760113927,
-      "loss": 46.0,
-      "step": 35250
-    },
-    {
-      "epoch": 2.6951851214710323,
-      "grad_norm": 0.002406393177807331,
-      "learning_rate": 0.00019999641739782326,
-      "loss": 46.0,
-      "step": 35251
-    },
-    {
-      "epoch": 2.695261578454422,
-      "grad_norm": 0.0017267242074012756,
-      "learning_rate": 0.00019999641719450147,
-      "loss": 46.0,
-      "step": 35252
-    },
-    {
-      "epoch": 2.695338035437812,
-      "grad_norm": 0.005018622614443302,
-      "learning_rate": 0.00019999641699117388,
-      "loss": 46.0,
-      "step": 35253
-    },
-    {
-      "epoch": 2.6954144924212016,
-      "grad_norm": 0.0018275509355589747,
-      "learning_rate": 0.00019999641678784057,
-      "loss": 46.0,
-      "step": 35254
-    },
-    {
-      "epoch": 2.6954909494045913,
-      "grad_norm": 0.002327537862583995,
-      "learning_rate": 0.00019999641658450146,
-      "loss": 46.0,
-      "step": 35255
-    },
-    {
-      "epoch": 2.695567406387981,
-      "grad_norm": 0.0013509526615962386,
-      "learning_rate": 0.0001999964163811566,
-      "loss": 46.0,
-      "step": 35256
-    },
-    {
-      "epoch": 2.6956438633713704,
-      "grad_norm": 0.0010866483207792044,
-      "learning_rate": 0.00019999641617780598,
-      "loss": 46.0,
-      "step": 35257
-    },
-    {
-      "epoch": 2.69572032035476,
-      "grad_norm": 0.0009713412146084011,
-      "learning_rate": 0.00019999641597444955,
-      "loss": 46.0,
-      "step": 35258
-    },
-    {
-      "epoch": 2.69579677733815,
-      "grad_norm": 0.001560694188810885,
-      "learning_rate": 0.0001999964157710874,
-      "loss": 46.0,
-      "step": 35259
-    },
-    {
-      "epoch": 2.6958732343215397,
-      "grad_norm": 0.001534345792606473,
-      "learning_rate": 0.00019999641556771945,
-      "loss": 46.0,
-      "step": 35260
-    },
-    {
-      "epoch": 2.6959496913049295,
-      "grad_norm": 0.0016061970964074135,
-      "learning_rate": 0.00019999641536434573,
-      "loss": 46.0,
-      "step": 35261
-    },
-    {
-      "epoch": 2.696026148288319,
-      "grad_norm": 0.0008186472696252167,
-      "learning_rate": 0.00019999641516096626,
-      "loss": 46.0,
-      "step": 35262
-    },
-    {
-      "epoch": 2.696102605271709,
-      "grad_norm": 0.0018672121223062277,
-      "learning_rate": 0.000199996414957581,
-      "loss": 46.0,
-      "step": 35263
-    },
-    {
-      "epoch": 2.6961790622550987,
-      "grad_norm": 0.005111853592097759,
-      "learning_rate": 0.00019999641475418995,
-      "loss": 46.0,
-      "step": 35264
-    },
-    {
-      "epoch": 2.6962555192384885,
-      "grad_norm": 0.0013540758518502116,
-      "learning_rate": 0.0001999964145507932,
-      "loss": 46.0,
-      "step": 35265
-    },
-    {
-      "epoch": 2.6963319762218783,
-      "grad_norm": 0.0012928255600854754,
-      "learning_rate": 0.0001999964143473906,
-      "loss": 46.0,
-      "step": 35266
-    },
-    {
-      "epoch": 2.696408433205268,
-      "grad_norm": 0.000994555070064962,
-      "learning_rate": 0.0001999964141439823,
-      "loss": 46.0,
-      "step": 35267
-    },
-    {
-      "epoch": 2.696484890188658,
-      "grad_norm": 0.004056311212480068,
-      "learning_rate": 0.0001999964139405682,
-      "loss": 46.0,
-      "step": 35268
-    },
-    {
-      "epoch": 2.6965613471720475,
-      "grad_norm": 0.0022508034016937017,
-      "learning_rate": 0.00019999641373714833,
-      "loss": 46.0,
-      "step": 35269
-    },
-    {
-      "epoch": 2.6966378041554373,
-      "grad_norm": 0.0005544980522245169,
-      "learning_rate": 0.00019999641353372267,
-      "loss": 46.0,
-      "step": 35270
-    },
-    {
-      "epoch": 2.6967142611388266,
-      "grad_norm": 0.002987751504406333,
-      "learning_rate": 0.00019999641333029127,
-      "loss": 46.0,
-      "step": 35271
-    },
-    {
-      "epoch": 2.6967907181222164,
-      "grad_norm": 0.0006851975922472775,
-      "learning_rate": 0.0001999964131268541,
-      "loss": 46.0,
-      "step": 35272
-    },
-    {
-      "epoch": 2.696867175105606,
-      "grad_norm": 0.0015056090196594596,
-      "learning_rate": 0.00019999641292341115,
-      "loss": 46.0,
-      "step": 35273
-    },
-    {
-      "epoch": 2.696943632088996,
-      "grad_norm": 0.0022397409193217754,
-      "learning_rate": 0.00019999641271996245,
-      "loss": 46.0,
-      "step": 35274
-    },
-    {
-      "epoch": 2.6970200890723857,
-      "grad_norm": 0.007908019237220287,
-      "learning_rate": 0.00019999641251650796,
-      "loss": 46.0,
-      "step": 35275
-    },
-    {
-      "epoch": 2.6970965460557754,
-      "grad_norm": 0.002086890395730734,
-      "learning_rate": 0.00019999641231304772,
-      "loss": 46.0,
-      "step": 35276
-    },
-    {
-      "epoch": 2.697173003039165,
-      "grad_norm": 0.002060619881376624,
-      "learning_rate": 0.00019999641210958168,
-      "loss": 46.0,
-      "step": 35277
-    },
-    {
-      "epoch": 2.697249460022555,
-      "grad_norm": 0.0007522001978941262,
-      "learning_rate": 0.0001999964119061099,
-      "loss": 46.0,
-      "step": 35278
-    },
-    {
-      "epoch": 2.6973259170059443,
-      "grad_norm": 0.0033598621375858784,
-      "learning_rate": 0.00019999641170263233,
-      "loss": 46.0,
-      "step": 35279
-    },
-    {
-      "epoch": 2.697402373989334,
-      "grad_norm": 0.001232060370966792,
-      "learning_rate": 0.000199996411499149,
-      "loss": 46.0,
-      "step": 35280
-    },
-    {
-      "epoch": 2.697478830972724,
-      "grad_norm": 0.002128458581864834,
-      "learning_rate": 0.0001999964112956599,
-      "loss": 46.0,
-      "step": 35281
-    },
-    {
-      "epoch": 2.6975552879561135,
-      "grad_norm": 0.003697927575558424,
-      "learning_rate": 0.00019999641109216506,
-      "loss": 46.0,
-      "step": 35282
-    },
-    {
-      "epoch": 2.6976317449395033,
-      "grad_norm": 0.0032263384200632572,
-      "learning_rate": 0.0001999964108886644,
-      "loss": 46.0,
-      "step": 35283
-    },
-    {
-      "epoch": 2.697708201922893,
-      "grad_norm": 0.0013719000853598118,
-      "learning_rate": 0.000199996410685158,
-      "loss": 46.0,
-      "step": 35284
-    },
-    {
-      "epoch": 2.697784658906283,
-      "grad_norm": 0.0016746128676459193,
-      "learning_rate": 0.0001999964104816458,
-      "loss": 46.0,
-      "step": 35285
-    },
-    {
-      "epoch": 2.6978611158896726,
-      "grad_norm": 0.002153475768864155,
-      "learning_rate": 0.00019999641027812788,
-      "loss": 46.0,
-      "step": 35286
-    },
-    {
-      "epoch": 2.6979375728730624,
-      "grad_norm": 0.0036397355142980814,
-      "learning_rate": 0.00019999641007460416,
-      "loss": 46.0,
-      "step": 35287
-    },
-    {
-      "epoch": 2.698014029856452,
-      "grad_norm": 0.0024249032139778137,
-      "learning_rate": 0.00019999640987107466,
-      "loss": 46.0,
-      "step": 35288
-    },
-    {
-      "epoch": 2.698090486839842,
-      "grad_norm": 0.002020876156166196,
-      "learning_rate": 0.00019999640966753942,
-      "loss": 46.0,
-      "step": 35289
-    },
-    {
-      "epoch": 2.6981669438232316,
-      "grad_norm": 0.002728401217609644,
-      "learning_rate": 0.0001999964094639984,
-      "loss": 46.0,
-      "step": 35290
-    },
-    {
-      "epoch": 2.6982434008066214,
-      "grad_norm": 0.002699471777305007,
-      "learning_rate": 0.0001999964092604516,
-      "loss": 46.0,
-      "step": 35291
-    },
-    {
-      "epoch": 2.698319857790011,
-      "grad_norm": 0.0021462140139192343,
-      "learning_rate": 0.00019999640905689906,
-      "loss": 46.0,
-      "step": 35292
-    },
-    {
-      "epoch": 2.6983963147734005,
-      "grad_norm": 0.0015100553864613175,
-      "learning_rate": 0.00019999640885334072,
-      "loss": 46.0,
-      "step": 35293
-    },
-    {
-      "epoch": 2.6984727717567902,
-      "grad_norm": 0.0021135956048965454,
-      "learning_rate": 0.00019999640864977662,
-      "loss": 46.0,
-      "step": 35294
-    },
-    {
-      "epoch": 2.69854922874018,
-      "grad_norm": 0.0013445416698232293,
-      "learning_rate": 0.00019999640844620676,
-      "loss": 46.0,
-      "step": 35295
-    },
-    {
-      "epoch": 2.6986256857235698,
-      "grad_norm": 0.0022414170671254396,
-      "learning_rate": 0.0001999964082426311,
-      "loss": 46.0,
-      "step": 35296
-    },
-    {
-      "epoch": 2.6987021427069595,
-      "grad_norm": 0.011541006155312061,
-      "learning_rate": 0.0001999964080390497,
-      "loss": 46.0,
-      "step": 35297
-    },
-    {
-      "epoch": 2.6987785996903493,
-      "grad_norm": 0.0015841620042920113,
-      "learning_rate": 0.00019999640783546253,
-      "loss": 46.0,
-      "step": 35298
-    },
-    {
-      "epoch": 2.698855056673739,
-      "grad_norm": 0.002681355457752943,
-      "learning_rate": 0.00019999640763186958,
-      "loss": 46.0,
-      "step": 35299
-    },
-    {
-      "epoch": 2.698931513657129,
-      "grad_norm": 0.0009408853948116302,
-      "learning_rate": 0.0001999964074282709,
-      "loss": 46.0,
-      "step": 35300
-    },
-    {
-      "epoch": 2.699007970640518,
-      "grad_norm": 0.0065516140311956406,
-      "learning_rate": 0.00019999640722466637,
-      "loss": 46.0,
-      "step": 35301
-    },
-    {
-      "epoch": 2.699084427623908,
-      "grad_norm": 0.0010787565261125565,
-      "learning_rate": 0.00019999640702105613,
-      "loss": 46.0,
-      "step": 35302
-    },
-    {
-      "epoch": 2.6991608846072976,
-      "grad_norm": 0.0023870819713920355,
-      "learning_rate": 0.00019999640681744011,
-      "loss": 46.0,
-      "step": 35303
-    },
-    {
-      "epoch": 2.6992373415906874,
-      "grad_norm": 0.002004152163863182,
-      "learning_rate": 0.00019999640661381833,
-      "loss": 46.0,
-      "step": 35304
-    },
-    {
-      "epoch": 2.699313798574077,
-      "grad_norm": 0.0015158592723309994,
-      "learning_rate": 0.00019999640641019077,
-      "loss": 46.0,
-      "step": 35305
-    },
-    {
-      "epoch": 2.699390255557467,
-      "grad_norm": 0.0020799008198082447,
-      "learning_rate": 0.00019999640620655743,
-      "loss": 46.0,
-      "step": 35306
-    },
-    {
-      "epoch": 2.6994667125408567,
-      "grad_norm": 0.0017356773605570197,
-      "learning_rate": 0.00019999640600291832,
-      "loss": 46.0,
-      "step": 35307
-    },
-    {
-      "epoch": 2.6995431695242464,
-      "grad_norm": 0.0011968230828642845,
-      "learning_rate": 0.00019999640579927344,
-      "loss": 46.0,
-      "step": 35308
-    },
-    {
-      "epoch": 2.699619626507636,
-      "grad_norm": 0.0025514685548841953,
-      "learning_rate": 0.00019999640559562282,
-      "loss": 46.0,
-      "step": 35309
-    },
-    {
-      "epoch": 2.699696083491026,
-      "grad_norm": 0.005680154077708721,
-      "learning_rate": 0.00019999640539196642,
-      "loss": 46.0,
-      "step": 35310
-    },
-    {
-      "epoch": 2.6997725404744157,
-      "grad_norm": 0.0023623218294233084,
-      "learning_rate": 0.00019999640518830422,
-      "loss": 46.0,
-      "step": 35311
-    },
-    {
-      "epoch": 2.6998489974578055,
-      "grad_norm": 0.0020611362997442484,
-      "learning_rate": 0.0001999964049846363,
-      "loss": 46.0,
-      "step": 35312
-    },
-    {
-      "epoch": 2.6999254544411952,
-      "grad_norm": 0.001051096012815833,
-      "learning_rate": 0.00019999640478096257,
-      "loss": 46.0,
-      "step": 35313
-    },
-    {
-      "epoch": 2.700001911424585,
-      "grad_norm": 0.0034514509607106447,
-      "learning_rate": 0.00019999640457728308,
-      "loss": 46.0,
-      "step": 35314
-    },
-    {
-      "epoch": 2.7000783684079743,
-      "grad_norm": 0.0023732339031994343,
-      "learning_rate": 0.00019999640437359784,
-      "loss": 46.0,
-      "step": 35315
-    },
-    {
-      "epoch": 2.700154825391364,
-      "grad_norm": 0.00437965476885438,
-      "learning_rate": 0.00019999640416990683,
-      "loss": 46.0,
-      "step": 35316
-    },
-    {
-      "epoch": 2.700231282374754,
-      "grad_norm": 0.002494441345334053,
-      "learning_rate": 0.00019999640396621,
-      "loss": 46.0,
-      "step": 35317
-    },
-    {
-      "epoch": 2.7003077393581436,
-      "grad_norm": 0.001108883647248149,
-      "learning_rate": 0.00019999640376250745,
-      "loss": 46.0,
-      "step": 35318
-    },
-    {
-      "epoch": 2.7003841963415334,
-      "grad_norm": 0.0007003109785728157,
-      "learning_rate": 0.00019999640355879914,
-      "loss": 46.0,
-      "step": 35319
-    },
-    {
-      "epoch": 2.700460653324923,
-      "grad_norm": 0.002885231049731374,
-      "learning_rate": 0.00019999640335508504,
-      "loss": 46.0,
-      "step": 35320
-    },
-    {
-      "epoch": 2.700537110308313,
-      "grad_norm": 0.006922852247953415,
-      "learning_rate": 0.00019999640315136516,
-      "loss": 46.0,
-      "step": 35321
-    },
-    {
-      "epoch": 2.7006135672917027,
-      "grad_norm": 0.003610148560255766,
-      "learning_rate": 0.00019999640294763953,
-      "loss": 46.0,
-      "step": 35322
-    },
-    {
-      "epoch": 2.700690024275092,
-      "grad_norm": 0.00418445048853755,
-      "learning_rate": 0.00019999640274390813,
-      "loss": 46.0,
-      "step": 35323
-    },
-    {
-      "epoch": 2.7007664812584817,
-      "grad_norm": 0.0013093260349705815,
-      "learning_rate": 0.00019999640254017096,
-      "loss": 46.0,
-      "step": 35324
-    },
-    {
-      "epoch": 2.7008429382418715,
-      "grad_norm": 0.0038081479724496603,
-      "learning_rate": 0.000199996402336428,
-      "loss": 46.0,
-      "step": 35325
-    },
-    {
-      "epoch": 2.7009193952252613,
-      "grad_norm": 0.001398626365698874,
-      "learning_rate": 0.00019999640213267926,
-      "loss": 46.0,
-      "step": 35326
-    },
-    {
-      "epoch": 2.700995852208651,
-      "grad_norm": 0.0038214665837585926,
-      "learning_rate": 0.0001999964019289248,
-      "loss": 46.0,
-      "step": 35327
-    },
-    {
-      "epoch": 2.7010723091920408,
-      "grad_norm": 0.00497637502849102,
-      "learning_rate": 0.00019999640172516453,
-      "loss": 46.0,
-      "step": 35328
-    },
-    {
-      "epoch": 2.7011487661754305,
-      "grad_norm": 0.00842752680182457,
-      "learning_rate": 0.00019999640152139851,
-      "loss": 46.0,
-      "step": 35329
-    },
-    {
-      "epoch": 2.7012252231588203,
-      "grad_norm": 0.001993365352973342,
-      "learning_rate": 0.00019999640131762673,
-      "loss": 46.0,
-      "step": 35330
-    },
-    {
-      "epoch": 2.70130168014221,
-      "grad_norm": 0.00212588207796216,
-      "learning_rate": 0.0001999964011138492,
-      "loss": 46.0,
-      "step": 35331
-    },
-    {
-      "epoch": 2.7013781371256,
-      "grad_norm": 0.0052423616871237755,
-      "learning_rate": 0.00019999640091006583,
-      "loss": 46.0,
-      "step": 35332
-    },
-    {
-      "epoch": 2.7014545941089896,
-      "grad_norm": 0.0010455293813720345,
-      "learning_rate": 0.00019999640070627675,
-      "loss": 46.0,
-      "step": 35333
-    },
-    {
-      "epoch": 2.7015310510923793,
-      "grad_norm": 0.002992102177813649,
-      "learning_rate": 0.00019999640050248187,
-      "loss": 46.0,
-      "step": 35334
-    },
-    {
-      "epoch": 2.701607508075769,
-      "grad_norm": 0.006704271305352449,
-      "learning_rate": 0.00019999640029868122,
-      "loss": 46.0,
-      "step": 35335
-    },
-    {
-      "epoch": 2.701683965059159,
-      "grad_norm": 0.0024180531036108732,
-      "learning_rate": 0.00019999640009487482,
-      "loss": 46.0,
-      "step": 35336
-    },
-    {
-      "epoch": 2.701760422042548,
-      "grad_norm": 0.002983035985380411,
-      "learning_rate": 0.00019999639989106265,
-      "loss": 46.0,
-      "step": 35337
-    },
-    {
-      "epoch": 2.701836879025938,
-      "grad_norm": 0.0051630097441375256,
-      "learning_rate": 0.0001999963996872447,
-      "loss": 46.0,
-      "step": 35338
-    },
-    {
-      "epoch": 2.7019133360093277,
-      "grad_norm": 0.0015458324924111366,
-      "learning_rate": 0.00019999639948342098,
-      "loss": 46.0,
-      "step": 35339
-    },
-    {
-      "epoch": 2.7019897929927175,
-      "grad_norm": 0.003122902475297451,
-      "learning_rate": 0.00019999639927959151,
-      "loss": 46.0,
-      "step": 35340
-    },
-    {
-      "epoch": 2.7020662499761072,
-      "grad_norm": 0.0019318468403071165,
-      "learning_rate": 0.00019999639907575625,
-      "loss": 46.0,
-      "step": 35341
-    },
-    {
-      "epoch": 2.702142706959497,
-      "grad_norm": 0.005658584646880627,
-      "learning_rate": 0.0001999963988719152,
-      "loss": 46.0,
-      "step": 35342
-    },
-    {
-      "epoch": 2.7022191639428867,
-      "grad_norm": 0.002577787032350898,
-      "learning_rate": 0.00019999639866806842,
-      "loss": 46.0,
-      "step": 35343
-    },
-    {
-      "epoch": 2.7022956209262765,
-      "grad_norm": 0.002360620303079486,
-      "learning_rate": 0.00019999639846421586,
-      "loss": 46.0,
-      "step": 35344
-    },
-    {
-      "epoch": 2.702372077909666,
-      "grad_norm": 0.0013628533342853189,
-      "learning_rate": 0.00019999639826035755,
-      "loss": 46.0,
-      "step": 35345
-    },
-    {
-      "epoch": 2.7024485348930556,
-      "grad_norm": 0.0034952175337821245,
-      "learning_rate": 0.00019999639805649345,
-      "loss": 46.0,
-      "step": 35346
-    },
-    {
-      "epoch": 2.7025249918764453,
-      "grad_norm": 0.0005553431692533195,
-      "learning_rate": 0.0001999963978526236,
-      "loss": 46.0,
-      "step": 35347
-    },
-    {
-      "epoch": 2.702601448859835,
-      "grad_norm": 0.0011918711243197322,
-      "learning_rate": 0.00019999639764874794,
-      "loss": 46.0,
-      "step": 35348
-    },
-    {
-      "epoch": 2.702677905843225,
-      "grad_norm": 0.003305414691567421,
-      "learning_rate": 0.00019999639744486652,
-      "loss": 46.0,
-      "step": 35349
-    },
-    {
-      "epoch": 2.7027543628266146,
-      "grad_norm": 0.0020327558740973473,
-      "learning_rate": 0.00019999639724097934,
-      "loss": 46.0,
-      "step": 35350
-    },
-    {
-      "epoch": 2.7028308198100044,
-      "grad_norm": 0.001078551053069532,
-      "learning_rate": 0.0001999963970370864,
-      "loss": 46.0,
-      "step": 35351
-    },
-    {
-      "epoch": 2.702907276793394,
-      "grad_norm": 0.0008975128876045346,
-      "learning_rate": 0.0001999963968331877,
-      "loss": 46.0,
-      "step": 35352
-    },
-    {
-      "epoch": 2.702983733776784,
-      "grad_norm": 0.0022663024719804525,
-      "learning_rate": 0.0001999963966292832,
-      "loss": 46.0,
-      "step": 35353
-    },
-    {
-      "epoch": 2.7030601907601737,
-      "grad_norm": 0.0071113938465714455,
-      "learning_rate": 0.00019999639642537294,
-      "loss": 46.0,
-      "step": 35354
-    },
-    {
-      "epoch": 2.7031366477435634,
-      "grad_norm": 0.0020904135890305042,
-      "learning_rate": 0.0001999963962214569,
-      "loss": 46.0,
-      "step": 35355
-    },
-    {
-      "epoch": 2.703213104726953,
-      "grad_norm": 0.0027654962614178658,
-      "learning_rate": 0.00019999639601753514,
-      "loss": 46.0,
-      "step": 35356
-    },
-    {
-      "epoch": 2.703289561710343,
-      "grad_norm": 0.0020273367408663034,
-      "learning_rate": 0.00019999639581360756,
-      "loss": 46.0,
-      "step": 35357
-    },
-    {
-      "epoch": 2.7033660186937323,
-      "grad_norm": 0.0011409459402784705,
-      "learning_rate": 0.00019999639560967425,
-      "loss": 46.0,
-      "step": 35358
-    },
-    {
-      "epoch": 2.703442475677122,
-      "grad_norm": 0.0032199742272496223,
-      "learning_rate": 0.00019999639540573515,
-      "loss": 46.0,
-      "step": 35359
-    },
-    {
-      "epoch": 2.703518932660512,
-      "grad_norm": 0.002354648895561695,
-      "learning_rate": 0.00019999639520179027,
-      "loss": 46.0,
-      "step": 35360
-    },
-    {
-      "epoch": 2.7035953896439016,
-      "grad_norm": 0.004205424804240465,
-      "learning_rate": 0.0001999963949978396,
-      "loss": 46.0,
-      "step": 35361
-    },
-    {
-      "epoch": 2.7036718466272913,
-      "grad_norm": 0.0016587713034823537,
-      "learning_rate": 0.00019999639479388321,
-      "loss": 46.0,
-      "step": 35362
-    },
-    {
-      "epoch": 2.703748303610681,
-      "grad_norm": 0.0016274135559797287,
-      "learning_rate": 0.00019999639458992104,
-      "loss": 46.0,
-      "step": 35363
-    },
-    {
-      "epoch": 2.703824760594071,
-      "grad_norm": 0.0011680605821311474,
-      "learning_rate": 0.0001999963943859531,
-      "loss": 46.0,
-      "step": 35364
-    },
-    {
-      "epoch": 2.7039012175774606,
-      "grad_norm": 0.0032782619819045067,
-      "learning_rate": 0.0001999963941819794,
-      "loss": 46.0,
-      "step": 35365
-    },
-    {
-      "epoch": 2.7039776745608504,
-      "grad_norm": 0.0015273921890184283,
-      "learning_rate": 0.00019999639397799988,
-      "loss": 46.0,
-      "step": 35366
-    },
-    {
-      "epoch": 2.7040541315442397,
-      "grad_norm": 0.0031500523909926414,
-      "learning_rate": 0.00019999639377401464,
-      "loss": 46.0,
-      "step": 35367
-    },
-    {
-      "epoch": 2.7041305885276294,
-      "grad_norm": 0.0013141479576006532,
-      "learning_rate": 0.00019999639357002363,
-      "loss": 46.0,
-      "step": 35368
-    },
-    {
-      "epoch": 2.704207045511019,
-      "grad_norm": 0.0020598049741238356,
-      "learning_rate": 0.00019999639336602685,
-      "loss": 46.0,
-      "step": 35369
-    },
-    {
-      "epoch": 2.704283502494409,
-      "grad_norm": 0.002656572964042425,
-      "learning_rate": 0.0001999963931620243,
-      "loss": 46.0,
-      "step": 35370
-    },
-    {
-      "epoch": 2.7043599594777987,
-      "grad_norm": 0.0013831154210492969,
-      "learning_rate": 0.00019999639295801593,
-      "loss": 46.0,
-      "step": 35371
-    },
-    {
-      "epoch": 2.7044364164611885,
-      "grad_norm": 0.0015021731378510594,
-      "learning_rate": 0.00019999639275400185,
-      "loss": 46.0,
-      "step": 35372
-    },
-    {
-      "epoch": 2.7045128734445782,
-      "grad_norm": 0.0015439930139109492,
-      "learning_rate": 0.00019999639254998197,
-      "loss": 46.0,
-      "step": 35373
-    },
-    {
-      "epoch": 2.704589330427968,
-      "grad_norm": 0.001346655422821641,
-      "learning_rate": 0.00019999639234595632,
-      "loss": 46.0,
-      "step": 35374
-    },
-    {
-      "epoch": 2.7046657874113578,
-      "grad_norm": 0.002888454357162118,
-      "learning_rate": 0.00019999639214192492,
-      "loss": 46.0,
-      "step": 35375
-    },
-    {
-      "epoch": 2.7047422443947475,
-      "grad_norm": 0.004119241610169411,
-      "learning_rate": 0.00019999639193788774,
-      "loss": 46.0,
-      "step": 35376
-    },
-    {
-      "epoch": 2.7048187013781373,
-      "grad_norm": 0.0015408092876896262,
-      "learning_rate": 0.0001999963917338448,
-      "loss": 46.0,
-      "step": 35377
-    },
-    {
-      "epoch": 2.704895158361527,
-      "grad_norm": 0.0027071263175457716,
-      "learning_rate": 0.00019999639152979608,
-      "loss": 46.0,
-      "step": 35378
-    },
-    {
-      "epoch": 2.704971615344917,
-      "grad_norm": 0.0008009267039597034,
-      "learning_rate": 0.0001999963913257416,
-      "loss": 46.0,
-      "step": 35379
-    },
-    {
-      "epoch": 2.705048072328306,
-      "grad_norm": 0.001985331764444709,
-      "learning_rate": 0.00019999639112168135,
-      "loss": 46.0,
-      "step": 35380
-    },
-    {
-      "epoch": 2.705124529311696,
-      "grad_norm": 0.0011344733648002148,
-      "learning_rate": 0.00019999639091761534,
-      "loss": 46.0,
-      "step": 35381
-    },
-    {
-      "epoch": 2.7052009862950857,
-      "grad_norm": 0.001126070972532034,
-      "learning_rate": 0.00019999639071354352,
-      "loss": 46.0,
-      "step": 35382
-    },
-    {
-      "epoch": 2.7052774432784754,
-      "grad_norm": 0.0031002238392829895,
-      "learning_rate": 0.00019999639050946596,
-      "loss": 46.0,
-      "step": 35383
-    },
-    {
-      "epoch": 2.705353900261865,
-      "grad_norm": 0.0029096296057105064,
-      "learning_rate": 0.00019999639030538263,
-      "loss": 46.0,
-      "step": 35384
-    },
-    {
-      "epoch": 2.705430357245255,
-      "grad_norm": 0.0006352112395688891,
-      "learning_rate": 0.00019999639010129353,
-      "loss": 46.0,
-      "step": 35385
-    },
-    {
-      "epoch": 2.7055068142286447,
-      "grad_norm": 0.002712411340326071,
-      "learning_rate": 0.00019999638989719868,
-      "loss": 46.0,
-      "step": 35386
-    },
-    {
-      "epoch": 2.7055832712120345,
-      "grad_norm": 0.0032077289652079344,
-      "learning_rate": 0.00019999638969309805,
-      "loss": 46.0,
-      "step": 35387
-    },
-    {
-      "epoch": 2.7056597281954238,
-      "grad_norm": 0.00955118890851736,
-      "learning_rate": 0.00019999638948899162,
-      "loss": 46.0,
-      "step": 35388
-    },
-    {
-      "epoch": 2.7057361851788135,
-      "grad_norm": 0.005012406036257744,
-      "learning_rate": 0.00019999638928487942,
-      "loss": 46.0,
-      "step": 35389
-    },
-    {
-      "epoch": 2.7058126421622033,
-      "grad_norm": 0.0012305768905207515,
-      "learning_rate": 0.0001999963890807615,
-      "loss": 46.0,
-      "step": 35390
-    },
-    {
-      "epoch": 2.705889099145593,
-      "grad_norm": 0.0027132295072078705,
-      "learning_rate": 0.00019999638887663776,
-      "loss": 46.0,
-      "step": 35391
-    },
-    {
-      "epoch": 2.705965556128983,
-      "grad_norm": 0.004948517307639122,
-      "learning_rate": 0.0001999963886725083,
-      "loss": 46.0,
-      "step": 35392
-    },
-    {
-      "epoch": 2.7060420131123726,
-      "grad_norm": 0.002816193038597703,
-      "learning_rate": 0.00019999638846837306,
-      "loss": 46.0,
-      "step": 35393
-    },
-    {
-      "epoch": 2.7061184700957623,
-      "grad_norm": 0.0007753594545647502,
-      "learning_rate": 0.00019999638826423202,
-      "loss": 46.0,
-      "step": 35394
-    },
-    {
-      "epoch": 2.706194927079152,
-      "grad_norm": 0.0018347802106291056,
-      "learning_rate": 0.0001999963880600852,
-      "loss": 46.0,
-      "step": 35395
-    },
-    {
-      "epoch": 2.706271384062542,
-      "grad_norm": 0.0029829605482518673,
-      "learning_rate": 0.00019999638785593265,
-      "loss": 46.0,
-      "step": 35396
-    },
-    {
-      "epoch": 2.7063478410459316,
-      "grad_norm": 0.002420371398329735,
-      "learning_rate": 0.00019999638765177434,
-      "loss": 46.0,
-      "step": 35397
-    },
-    {
-      "epoch": 2.7064242980293214,
-      "grad_norm": 0.002120414050295949,
-      "learning_rate": 0.0001999963874476102,
-      "loss": 46.0,
-      "step": 35398
-    },
-    {
-      "epoch": 2.706500755012711,
-      "grad_norm": 0.0017406987026333809,
-      "learning_rate": 0.00019999638724344036,
-      "loss": 46.0,
-      "step": 35399
-    },
-    {
-      "epoch": 2.706577211996101,
-      "grad_norm": 0.0028157096821814775,
-      "learning_rate": 0.0001999963870392647,
-      "loss": 46.0,
-      "step": 35400
-    },
-    {
-      "epoch": 2.7066536689794907,
-      "grad_norm": 0.002767870668321848,
-      "learning_rate": 0.0001999963868350833,
-      "loss": 46.0,
-      "step": 35401
-    },
-    {
-      "epoch": 2.70673012596288,
-      "grad_norm": 0.002995781134814024,
-      "learning_rate": 0.00019999638663089614,
-      "loss": 46.0,
-      "step": 35402
-    },
-    {
-      "epoch": 2.7068065829462697,
-      "grad_norm": 0.0030997698195278645,
-      "learning_rate": 0.0001999963864267032,
-      "loss": 46.0,
-      "step": 35403
-    },
-    {
-      "epoch": 2.7068830399296595,
-      "grad_norm": 0.0011447790311649442,
-      "learning_rate": 0.00019999638622250447,
-      "loss": 46.0,
-      "step": 35404
-    },
-    {
-      "epoch": 2.7069594969130493,
-      "grad_norm": 0.00422257836908102,
-      "learning_rate": 0.00019999638601829998,
-      "loss": 46.0,
-      "step": 35405
-    },
-    {
-      "epoch": 2.707035953896439,
-      "grad_norm": 0.0011334065347909927,
-      "learning_rate": 0.00019999638581408972,
-      "loss": 46.0,
-      "step": 35406
-    },
-    {
-      "epoch": 2.707112410879829,
-      "grad_norm": 0.0016981714870780706,
-      "learning_rate": 0.0001999963856098737,
-      "loss": 46.0,
-      "step": 35407
-    },
-    {
-      "epoch": 2.7071888678632186,
-      "grad_norm": 0.0013773051323369145,
-      "learning_rate": 0.00019999638540565192,
-      "loss": 46.0,
-      "step": 35408
-    },
-    {
-      "epoch": 2.7072653248466083,
-      "grad_norm": 0.0024255216121673584,
-      "learning_rate": 0.00019999638520142433,
-      "loss": 46.0,
-      "step": 35409
-    },
-    {
-      "epoch": 2.7073417818299976,
-      "grad_norm": 0.0020831571891903877,
-      "learning_rate": 0.000199996384997191,
-      "loss": 46.0,
-      "step": 35410
-    },
-    {
-      "epoch": 2.7074182388133874,
-      "grad_norm": 0.003036827314645052,
-      "learning_rate": 0.00019999638479295192,
-      "loss": 46.0,
-      "step": 35411
-    },
-    {
-      "epoch": 2.707494695796777,
-      "grad_norm": 0.0010015048319473863,
-      "learning_rate": 0.00019999638458870705,
-      "loss": 46.0,
-      "step": 35412
-    },
-    {
-      "epoch": 2.707571152780167,
-      "grad_norm": 0.0033317063935101032,
-      "learning_rate": 0.0001999963843844564,
-      "loss": 46.0,
-      "step": 35413
-    },
-    {
-      "epoch": 2.7076476097635567,
-      "grad_norm": 0.000883603235706687,
-      "learning_rate": 0.0001999963841802,
-      "loss": 46.0,
-      "step": 35414
-    },
-    {
-      "epoch": 2.7077240667469464,
-      "grad_norm": 0.0031761000864207745,
-      "learning_rate": 0.00019999638397593782,
-      "loss": 46.0,
-      "step": 35415
-    },
-    {
-      "epoch": 2.707800523730336,
-      "grad_norm": 0.0026982135605067015,
-      "learning_rate": 0.00019999638377166988,
-      "loss": 46.0,
-      "step": 35416
-    },
-    {
-      "epoch": 2.707876980713726,
-      "grad_norm": 0.0013232608325779438,
-      "learning_rate": 0.00019999638356739613,
-      "loss": 46.0,
-      "step": 35417
-    },
-    {
-      "epoch": 2.7079534376971157,
-      "grad_norm": 0.0022729686461389065,
-      "learning_rate": 0.00019999638336311667,
-      "loss": 46.0,
-      "step": 35418
-    },
-    {
-      "epoch": 2.7080298946805055,
-      "grad_norm": 0.0015212746802717447,
-      "learning_rate": 0.0001999963831588314,
-      "loss": 46.0,
-      "step": 35419
-    },
-    {
-      "epoch": 2.7081063516638952,
-      "grad_norm": 0.002108345041051507,
-      "learning_rate": 0.0001999963829545404,
-      "loss": 46.0,
-      "step": 35420
-    },
-    {
-      "epoch": 2.708182808647285,
-      "grad_norm": 0.0010278718546032906,
-      "learning_rate": 0.00019999638275024358,
-      "loss": 46.0,
-      "step": 35421
-    },
-    {
-      "epoch": 2.7082592656306748,
-      "grad_norm": 0.0030459852423518896,
-      "learning_rate": 0.00019999638254594105,
-      "loss": 46.0,
-      "step": 35422
-    },
-    {
-      "epoch": 2.7083357226140645,
-      "grad_norm": 0.0027643898501992226,
-      "learning_rate": 0.0001999963823416327,
-      "loss": 46.0,
-      "step": 35423
-    },
-    {
-      "epoch": 2.708412179597454,
-      "grad_norm": 0.0010355358244851232,
-      "learning_rate": 0.00019999638213731862,
-      "loss": 46.0,
-      "step": 35424
-    },
-    {
-      "epoch": 2.7084886365808436,
-      "grad_norm": 0.0035593723878264427,
-      "learning_rate": 0.0001999963819329987,
-      "loss": 46.0,
-      "step": 35425
-    },
-    {
-      "epoch": 2.7085650935642334,
-      "grad_norm": 0.0013980477815493941,
-      "learning_rate": 0.00019999638172867309,
-      "loss": 46.0,
-      "step": 35426
-    },
-    {
-      "epoch": 2.708641550547623,
-      "grad_norm": 0.001255303737707436,
-      "learning_rate": 0.0001999963815243417,
-      "loss": 46.0,
-      "step": 35427
-    },
-    {
-      "epoch": 2.708718007531013,
-      "grad_norm": 0.001013144152238965,
-      "learning_rate": 0.0001999963813200045,
-      "loss": 46.0,
-      "step": 35428
-    },
-    {
-      "epoch": 2.7087944645144026,
-      "grad_norm": 0.0032639699056744576,
-      "learning_rate": 0.00019999638111566155,
-      "loss": 46.0,
-      "step": 35429
-    },
-    {
-      "epoch": 2.7088709214977924,
-      "grad_norm": 0.0023879779037088156,
-      "learning_rate": 0.00019999638091131283,
-      "loss": 46.0,
-      "step": 35430
-    },
-    {
-      "epoch": 2.708947378481182,
-      "grad_norm": 0.0034250114113092422,
-      "learning_rate": 0.00019999638070695834,
-      "loss": 46.0,
-      "step": 35431
-    },
-    {
-      "epoch": 2.7090238354645715,
-      "grad_norm": 0.0020324087236076593,
-      "learning_rate": 0.0001999963805025981,
-      "loss": 46.0,
-      "step": 35432
-    },
-    {
-      "epoch": 2.7091002924479612,
-      "grad_norm": 0.0038056583143770695,
-      "learning_rate": 0.0001999963802982321,
-      "loss": 46.0,
-      "step": 35433
-    },
-    {
-      "epoch": 2.709176749431351,
-      "grad_norm": 0.004083080217242241,
-      "learning_rate": 0.00019999638009386028,
-      "loss": 46.0,
-      "step": 35434
-    },
-    {
-      "epoch": 2.7092532064147408,
-      "grad_norm": 0.004164343234151602,
-      "learning_rate": 0.00019999637988948272,
-      "loss": 46.0,
-      "step": 35435
-    },
-    {
-      "epoch": 2.7093296633981305,
-      "grad_norm": 0.002578981453552842,
-      "learning_rate": 0.0001999963796850994,
-      "loss": 46.0,
-      "step": 35436
-    },
-    {
-      "epoch": 2.7094061203815203,
-      "grad_norm": 0.004260354209691286,
-      "learning_rate": 0.00019999637948071028,
-      "loss": 46.0,
-      "step": 35437
-    },
-    {
-      "epoch": 2.70948257736491,
-      "grad_norm": 0.00152291648555547,
-      "learning_rate": 0.0001999963792763154,
-      "loss": 46.0,
-      "step": 35438
-    },
-    {
-      "epoch": 2.7095590343483,
-      "grad_norm": 0.0012238899944350123,
-      "learning_rate": 0.00019999637907191478,
-      "loss": 46.0,
-      "step": 35439
-    },
-    {
-      "epoch": 2.7096354913316896,
-      "grad_norm": 0.0012467509368434548,
-      "learning_rate": 0.00019999637886750838,
-      "loss": 46.0,
-      "step": 35440
-    },
-    {
-      "epoch": 2.7097119483150793,
-      "grad_norm": 0.00599660025909543,
-      "learning_rate": 0.0001999963786630962,
-      "loss": 46.0,
-      "step": 35441
-    },
-    {
-      "epoch": 2.709788405298469,
-      "grad_norm": 0.0012749666348099709,
-      "learning_rate": 0.00019999637845867824,
-      "loss": 46.0,
-      "step": 35442
-    },
-    {
-      "epoch": 2.709864862281859,
-      "grad_norm": 0.0025537002366036177,
-      "learning_rate": 0.00019999637825425453,
-      "loss": 46.0,
-      "step": 35443
-    },
-    {
-      "epoch": 2.7099413192652486,
-      "grad_norm": 0.010889332741498947,
-      "learning_rate": 0.00019999637804982503,
-      "loss": 46.0,
-      "step": 35444
-    },
-    {
-      "epoch": 2.7100177762486384,
-      "grad_norm": 0.0038424122612923384,
-      "learning_rate": 0.00019999637784538977,
-      "loss": 46.0,
-      "step": 35445
-    },
-    {
-      "epoch": 2.7100942332320277,
-      "grad_norm": 0.002768601756542921,
-      "learning_rate": 0.00019999637764094876,
-      "loss": 46.0,
-      "step": 35446
-    },
-    {
-      "epoch": 2.7101706902154175,
-      "grad_norm": 0.0013511487049981952,
-      "learning_rate": 0.00019999637743650198,
-      "loss": 46.0,
-      "step": 35447
-    },
-    {
-      "epoch": 2.710247147198807,
-      "grad_norm": 0.004036264028400183,
-      "learning_rate": 0.0001999963772320494,
-      "loss": 46.0,
-      "step": 35448
-    },
-    {
-      "epoch": 2.710323604182197,
-      "grad_norm": 0.0021058148704469204,
-      "learning_rate": 0.00019999637702759106,
-      "loss": 46.0,
-      "step": 35449
-    },
-    {
-      "epoch": 2.7104000611655867,
-      "grad_norm": 0.0024338255170732737,
-      "learning_rate": 0.00019999637682312696,
-      "loss": 46.0,
-      "step": 35450
-    },
-    {
-      "epoch": 2.7104765181489765,
-      "grad_norm": 0.0029360386542975903,
-      "learning_rate": 0.00019999637661865708,
-      "loss": 46.0,
-      "step": 35451
-    },
-    {
-      "epoch": 2.7105529751323663,
-      "grad_norm": 0.0011251167161390185,
-      "learning_rate": 0.00019999637641418145,
-      "loss": 46.0,
-      "step": 35452
-    },
-    {
-      "epoch": 2.710629432115756,
-      "grad_norm": 0.0020247255451977253,
-      "learning_rate": 0.00019999637620970003,
-      "loss": 46.0,
-      "step": 35453
-    },
-    {
-      "epoch": 2.7107058890991453,
-      "grad_norm": 0.0014465537387877703,
-      "learning_rate": 0.00019999637600521286,
-      "loss": 46.0,
-      "step": 35454
-    },
-    {
-      "epoch": 2.710782346082535,
-      "grad_norm": 0.0016894011059775949,
-      "learning_rate": 0.00019999637580071992,
-      "loss": 46.0,
-      "step": 35455
-    },
-    {
-      "epoch": 2.710858803065925,
-      "grad_norm": 0.0010414608987048268,
-      "learning_rate": 0.00019999637559622123,
-      "loss": 46.0,
-      "step": 35456
-    },
-    {
-      "epoch": 2.7109352600493146,
-      "grad_norm": 0.003288756823167205,
-      "learning_rate": 0.0001999963753917167,
-      "loss": 46.0,
-      "step": 35457
-    },
-    {
-      "epoch": 2.7110117170327044,
-      "grad_norm": 0.0006927825161255896,
-      "learning_rate": 0.00019999637518720647,
-      "loss": 46.0,
-      "step": 35458
-    },
-    {
-      "epoch": 2.711088174016094,
-      "grad_norm": 0.0015430875355377793,
-      "learning_rate": 0.00019999637498269043,
-      "loss": 46.0,
-      "step": 35459
-    },
-    {
-      "epoch": 2.711164630999484,
-      "grad_norm": 0.002221212489530444,
-      "learning_rate": 0.00019999637477816865,
-      "loss": 46.0,
-      "step": 35460
-    },
-    {
-      "epoch": 2.7112410879828737,
-      "grad_norm": 0.006392726209014654,
-      "learning_rate": 0.00019999637457364107,
-      "loss": 46.0,
-      "step": 35461
-    },
-    {
-      "epoch": 2.7113175449662634,
-      "grad_norm": 0.002017437480390072,
-      "learning_rate": 0.00019999637436910774,
-      "loss": 46.0,
-      "step": 35462
-    },
-    {
-      "epoch": 2.711394001949653,
-      "grad_norm": 0.0013278081314638257,
-      "learning_rate": 0.00019999637416456863,
-      "loss": 46.0,
-      "step": 35463
-    },
-    {
-      "epoch": 2.711470458933043,
-      "grad_norm": 0.0023082876577973366,
-      "learning_rate": 0.00019999637396002378,
-      "loss": 46.0,
-      "step": 35464
-    },
-    {
-      "epoch": 2.7115469159164327,
-      "grad_norm": 0.003115879837423563,
-      "learning_rate": 0.00019999637375547313,
-      "loss": 46.0,
-      "step": 35465
-    },
-    {
-      "epoch": 2.7116233728998225,
-      "grad_norm": 0.002698844764381647,
-      "learning_rate": 0.0001999963735509167,
-      "loss": 46.0,
-      "step": 35466
-    },
-    {
-      "epoch": 2.7116998298832122,
-      "grad_norm": 0.0011624912731349468,
-      "learning_rate": 0.00019999637334635454,
-      "loss": 46.0,
-      "step": 35467
-    },
-    {
-      "epoch": 2.7117762868666015,
-      "grad_norm": 0.002155412919819355,
-      "learning_rate": 0.00019999637314178662,
-      "loss": 46.0,
-      "step": 35468
-    },
-    {
-      "epoch": 2.7118527438499913,
-      "grad_norm": 0.000701119948644191,
-      "learning_rate": 0.00019999637293721288,
-      "loss": 46.0,
-      "step": 35469
-    },
-    {
-      "epoch": 2.711929200833381,
-      "grad_norm": 0.0009733604383654892,
-      "learning_rate": 0.00019999637273263341,
-      "loss": 46.0,
-      "step": 35470
-    },
-    {
-      "epoch": 2.712005657816771,
-      "grad_norm": 0.004210932180285454,
-      "learning_rate": 0.00019999637252804812,
-      "loss": 46.0,
-      "step": 35471
-    },
-    {
-      "epoch": 2.7120821148001606,
-      "grad_norm": 0.0015431327046826482,
-      "learning_rate": 0.00019999637232345711,
-      "loss": 46.0,
-      "step": 35472
-    },
-    {
-      "epoch": 2.7121585717835504,
-      "grad_norm": 0.0025987043045461178,
-      "learning_rate": 0.00019999637211886033,
-      "loss": 46.0,
-      "step": 35473
-    },
-    {
-      "epoch": 2.71223502876694,
-      "grad_norm": 0.0026992056518793106,
-      "learning_rate": 0.00019999637191425777,
-      "loss": 46.0,
-      "step": 35474
-    },
-    {
-      "epoch": 2.71231148575033,
-      "grad_norm": 0.0030761456582695246,
-      "learning_rate": 0.00019999637170964942,
-      "loss": 46.0,
-      "step": 35475
-    },
-    {
-      "epoch": 2.712387942733719,
-      "grad_norm": 0.0014795601600781083,
-      "learning_rate": 0.00019999637150503534,
-      "loss": 46.0,
-      "step": 35476
-    },
-    {
-      "epoch": 2.712464399717109,
-      "grad_norm": 0.0015331594040617347,
-      "learning_rate": 0.00019999637130041546,
-      "loss": 46.0,
-      "step": 35477
-    },
-    {
-      "epoch": 2.7125408567004987,
-      "grad_norm": 0.0017951850313693285,
-      "learning_rate": 0.0001999963710957898,
-      "loss": 46.0,
-      "step": 35478
-    },
-    {
-      "epoch": 2.7126173136838885,
-      "grad_norm": 0.0012449405621737242,
-      "learning_rate": 0.0001999963708911584,
-      "loss": 46.0,
-      "step": 35479
-    },
-    {
-      "epoch": 2.7126937706672782,
-      "grad_norm": 0.001645549084059894,
-      "learning_rate": 0.00019999637068652125,
-      "loss": 46.0,
-      "step": 35480
-    },
-    {
-      "epoch": 2.712770227650668,
-      "grad_norm": 0.001400281791575253,
-      "learning_rate": 0.00019999637048187828,
-      "loss": 46.0,
-      "step": 35481
-    },
-    {
-      "epoch": 2.7128466846340578,
-      "grad_norm": 0.004993239417672157,
-      "learning_rate": 0.00019999637027722956,
-      "loss": 46.0,
-      "step": 35482
-    },
-    {
-      "epoch": 2.7129231416174475,
-      "grad_norm": 0.001269301981665194,
-      "learning_rate": 0.00019999637007257507,
-      "loss": 46.0,
-      "step": 35483
-    },
-    {
-      "epoch": 2.7129995986008373,
-      "grad_norm": 0.0011594791430979967,
-      "learning_rate": 0.0001999963698679148,
-      "loss": 46.0,
-      "step": 35484
-    },
-    {
-      "epoch": 2.713076055584227,
-      "grad_norm": 0.00197885325178504,
-      "learning_rate": 0.0001999963696632488,
-      "loss": 46.0,
-      "step": 35485
-    },
-    {
-      "epoch": 2.713152512567617,
-      "grad_norm": 0.0017935370560735464,
-      "learning_rate": 0.00019999636945857701,
-      "loss": 46.0,
-      "step": 35486
-    },
-    {
-      "epoch": 2.7132289695510066,
-      "grad_norm": 0.0006822082796134055,
-      "learning_rate": 0.00019999636925389943,
-      "loss": 46.0,
-      "step": 35487
-    },
-    {
-      "epoch": 2.7133054265343963,
-      "grad_norm": 0.0006907025235705078,
-      "learning_rate": 0.0001999963690492161,
-      "loss": 46.0,
-      "step": 35488
-    },
-    {
-      "epoch": 2.7133818835177856,
-      "grad_norm": 0.0016619703965261579,
-      "learning_rate": 0.000199996368844527,
-      "loss": 46.0,
-      "step": 35489
-    },
-    {
-      "epoch": 2.7134583405011754,
-      "grad_norm": 0.003065954428166151,
-      "learning_rate": 0.00019999636863983212,
-      "loss": 46.0,
-      "step": 35490
-    },
-    {
-      "epoch": 2.713534797484565,
-      "grad_norm": 0.002040877938270569,
-      "learning_rate": 0.0001999963684351315,
-      "loss": 46.0,
-      "step": 35491
-    },
-    {
-      "epoch": 2.713611254467955,
-      "grad_norm": 0.00245712255127728,
-      "learning_rate": 0.00019999636823042508,
-      "loss": 46.0,
-      "step": 35492
-    },
-    {
-      "epoch": 2.7136877114513447,
-      "grad_norm": 0.005380833987146616,
-      "learning_rate": 0.0001999963680257129,
-      "loss": 46.0,
-      "step": 35493
-    },
-    {
-      "epoch": 2.7137641684347344,
-      "grad_norm": 0.007019604090601206,
-      "learning_rate": 0.00019999636782099494,
-      "loss": 46.0,
-      "step": 35494
-    },
-    {
-      "epoch": 2.713840625418124,
-      "grad_norm": 0.002466310281306505,
-      "learning_rate": 0.00019999636761627122,
-      "loss": 46.0,
-      "step": 35495
-    },
-    {
-      "epoch": 2.713917082401514,
-      "grad_norm": 0.0022517023608088493,
-      "learning_rate": 0.00019999636741154176,
-      "loss": 46.0,
-      "step": 35496
-    },
-    {
-      "epoch": 2.7139935393849037,
-      "grad_norm": 0.004487857688218355,
-      "learning_rate": 0.0001999963672068065,
-      "loss": 46.0,
-      "step": 35497
-    },
-    {
-      "epoch": 2.714069996368293,
-      "grad_norm": 0.002486432669684291,
-      "learning_rate": 0.00019999636700206548,
-      "loss": 46.0,
-      "step": 35498
-    },
-    {
-      "epoch": 2.714146453351683,
-      "grad_norm": 0.0017079716781154275,
-      "learning_rate": 0.00019999636679731867,
-      "loss": 46.0,
-      "step": 35499
-    },
-    {
-      "epoch": 2.7142229103350726,
-      "grad_norm": 0.0009632365545257926,
-      "learning_rate": 0.00019999636659256612,
-      "loss": 46.0,
-      "step": 35500
-    },
-    {
-      "epoch": 2.7142993673184623,
-      "grad_norm": 0.001022086013108492,
-      "learning_rate": 0.00019999636638780776,
-      "loss": 46.0,
-      "step": 35501
-    },
-    {
-      "epoch": 2.714375824301852,
-      "grad_norm": 0.0017497050575911999,
-      "learning_rate": 0.00019999636618304366,
-      "loss": 46.0,
-      "step": 35502
-    },
-    {
-      "epoch": 2.714452281285242,
-      "grad_norm": 0.0019058617763221264,
-      "learning_rate": 0.0001999963659782738,
-      "loss": 46.0,
-      "step": 35503
-    },
-    {
-      "epoch": 2.7145287382686316,
-      "grad_norm": 0.001119736349210143,
-      "learning_rate": 0.00019999636577349816,
-      "loss": 46.0,
-      "step": 35504
-    },
-    {
-      "epoch": 2.7146051952520214,
-      "grad_norm": 0.0005977515247650445,
-      "learning_rate": 0.00019999636556871674,
-      "loss": 46.0,
-      "step": 35505
-    },
-    {
-      "epoch": 2.714681652235411,
-      "grad_norm": 0.0005228149238973856,
-      "learning_rate": 0.0001999963653639296,
-      "loss": 46.0,
-      "step": 35506
-    },
-    {
-      "epoch": 2.714758109218801,
-      "grad_norm": 0.0025001182220876217,
-      "learning_rate": 0.00019999636515913663,
-      "loss": 46.0,
-      "step": 35507
-    },
-    {
-      "epoch": 2.7148345662021907,
-      "grad_norm": 0.0024831844493746758,
-      "learning_rate": 0.00019999636495433788,
-      "loss": 46.0,
-      "step": 35508
-    },
-    {
-      "epoch": 2.7149110231855804,
-      "grad_norm": 0.0020419922657310963,
-      "learning_rate": 0.0001999963647495334,
-      "loss": 46.0,
-      "step": 35509
-    },
-    {
-      "epoch": 2.71498748016897,
-      "grad_norm": 0.0034890954848378897,
-      "learning_rate": 0.00019999636454472316,
-      "loss": 46.0,
-      "step": 35510
-    },
-    {
-      "epoch": 2.7150639371523595,
-      "grad_norm": 0.0022106829565018415,
-      "learning_rate": 0.00019999636433990715,
-      "loss": 46.0,
-      "step": 35511
-    },
-    {
-      "epoch": 2.7151403941357493,
-      "grad_norm": 0.0013709880877286196,
-      "learning_rate": 0.00019999636413508537,
-      "loss": 46.0,
-      "step": 35512
-    },
-    {
-      "epoch": 2.715216851119139,
-      "grad_norm": 0.001358649111352861,
-      "learning_rate": 0.00019999636393025778,
-      "loss": 46.0,
-      "step": 35513
-    },
-    {
-      "epoch": 2.715293308102529,
-      "grad_norm": 0.004970632493495941,
-      "learning_rate": 0.00019999636372542443,
-      "loss": 46.0,
-      "step": 35514
-    },
-    {
-      "epoch": 2.7153697650859185,
-      "grad_norm": 0.0006761665572412312,
-      "learning_rate": 0.00019999636352058535,
-      "loss": 46.0,
-      "step": 35515
-    },
-    {
-      "epoch": 2.7154462220693083,
-      "grad_norm": 0.002757573500275612,
-      "learning_rate": 0.00019999636331574045,
-      "loss": 46.0,
-      "step": 35516
-    },
-    {
-      "epoch": 2.715522679052698,
-      "grad_norm": 0.001515775453299284,
-      "learning_rate": 0.00019999636311088983,
-      "loss": 46.0,
-      "step": 35517
-    },
-    {
-      "epoch": 2.715599136036088,
-      "grad_norm": 0.0017731004627421498,
-      "learning_rate": 0.00019999636290603343,
-      "loss": 46.0,
-      "step": 35518
-    },
-    {
-      "epoch": 2.715675593019477,
-      "grad_norm": 0.0014121552230790257,
-      "learning_rate": 0.00019999636270117124,
-      "loss": 46.0,
-      "step": 35519
-    },
-    {
-      "epoch": 2.715752050002867,
-      "grad_norm": 0.0013069083215668797,
-      "learning_rate": 0.0001999963624963033,
-      "loss": 46.0,
-      "step": 35520
-    },
-    {
-      "epoch": 2.7158285069862567,
-      "grad_norm": 0.0019986287225037813,
-      "learning_rate": 0.00019999636229142958,
-      "loss": 46.0,
-      "step": 35521
-    },
-    {
-      "epoch": 2.7159049639696464,
-      "grad_norm": 0.000965594663284719,
-      "learning_rate": 0.0001999963620865501,
-      "loss": 46.0,
-      "step": 35522
-    },
-    {
-      "epoch": 2.715981420953036,
-      "grad_norm": 0.0017062557162716985,
-      "learning_rate": 0.00019999636188166485,
-      "loss": 46.0,
-      "step": 35523
-    },
-    {
-      "epoch": 2.716057877936426,
-      "grad_norm": 0.001205394510179758,
-      "learning_rate": 0.0001999963616767738,
-      "loss": 46.0,
-      "step": 35524
-    },
-    {
-      "epoch": 2.7161343349198157,
-      "grad_norm": 0.0023399549536406994,
-      "learning_rate": 0.00019999636147187698,
-      "loss": 46.0,
-      "step": 35525
-    },
-    {
-      "epoch": 2.7162107919032055,
-      "grad_norm": 0.0017552935751155019,
-      "learning_rate": 0.00019999636126697445,
-      "loss": 46.0,
-      "step": 35526
-    },
-    {
-      "epoch": 2.7162872488865952,
-      "grad_norm": 0.0023944140411913395,
-      "learning_rate": 0.0001999963610620661,
-      "loss": 46.0,
-      "step": 35527
-    },
-    {
-      "epoch": 2.716363705869985,
-      "grad_norm": 0.002572909463196993,
-      "learning_rate": 0.00019999636085715202,
-      "loss": 46.0,
-      "step": 35528
-    },
-    {
-      "epoch": 2.7164401628533748,
-      "grad_norm": 0.0019933246076107025,
-      "learning_rate": 0.00019999636065223215,
-      "loss": 46.0,
-      "step": 35529
-    },
-    {
-      "epoch": 2.7165166198367645,
-      "grad_norm": 0.0023498532827943563,
-      "learning_rate": 0.0001999963604473065,
-      "loss": 46.0,
-      "step": 35530
-    },
-    {
-      "epoch": 2.7165930768201543,
-      "grad_norm": 0.0037701076362282038,
-      "learning_rate": 0.00019999636024237508,
-      "loss": 46.0,
-      "step": 35531
-    },
-    {
-      "epoch": 2.716669533803544,
-      "grad_norm": 0.0011174597311764956,
-      "learning_rate": 0.00019999636003743788,
-      "loss": 46.0,
-      "step": 35532
-    },
-    {
-      "epoch": 2.7167459907869334,
-      "grad_norm": 0.0029042684473097324,
-      "learning_rate": 0.00019999635983249496,
-      "loss": 46.0,
-      "step": 35533
-    },
-    {
-      "epoch": 2.716822447770323,
-      "grad_norm": 0.003120404202491045,
-      "learning_rate": 0.00019999635962754625,
-      "loss": 46.0,
-      "step": 35534
-    },
-    {
-      "epoch": 2.716898904753713,
-      "grad_norm": 0.0019677034579217434,
-      "learning_rate": 0.00019999635942259176,
-      "loss": 46.0,
-      "step": 35535
-    },
-    {
-      "epoch": 2.7169753617371026,
-      "grad_norm": 0.002027340931817889,
-      "learning_rate": 0.0001999963592176315,
-      "loss": 46.0,
-      "step": 35536
-    },
-    {
-      "epoch": 2.7170518187204924,
-      "grad_norm": 0.0013542985543608665,
-      "learning_rate": 0.00019999635901266546,
-      "loss": 46.0,
-      "step": 35537
-    },
-    {
-      "epoch": 2.717128275703882,
-      "grad_norm": 0.002734426874667406,
-      "learning_rate": 0.00019999635880769368,
-      "loss": 46.0,
-      "step": 35538
-    },
-    {
-      "epoch": 2.717204732687272,
-      "grad_norm": 0.0033826292492449284,
-      "learning_rate": 0.0001999963586027161,
-      "loss": 46.0,
-      "step": 35539
-    },
-    {
-      "epoch": 2.7172811896706617,
-      "grad_norm": 0.001405687304213643,
-      "learning_rate": 0.00019999635839773277,
-      "loss": 46.0,
-      "step": 35540
-    },
-    {
-      "epoch": 2.717357646654051,
-      "grad_norm": 0.003008811268955469,
-      "learning_rate": 0.00019999635819274367,
-      "loss": 46.0,
-      "step": 35541
-    },
-    {
-      "epoch": 2.7174341036374408,
-      "grad_norm": 0.0008095461525954306,
-      "learning_rate": 0.0001999963579877488,
-      "loss": 46.0,
-      "step": 35542
-    },
-    {
-      "epoch": 2.7175105606208305,
-      "grad_norm": 0.003856959054246545,
-      "learning_rate": 0.00019999635778274817,
-      "loss": 46.0,
-      "step": 35543
-    },
-    {
-      "epoch": 2.7175870176042203,
-      "grad_norm": 0.0023096930235624313,
-      "learning_rate": 0.00019999635757774175,
-      "loss": 46.0,
-      "step": 35544
-    },
-    {
-      "epoch": 2.71766347458761,
-      "grad_norm": 0.0013384120538830757,
-      "learning_rate": 0.00019999635737272958,
-      "loss": 46.0,
-      "step": 35545
-    },
-    {
-      "epoch": 2.717739931571,
-      "grad_norm": 0.001882214448414743,
-      "learning_rate": 0.0001999963571677116,
-      "loss": 46.0,
-      "step": 35546
-    },
-    {
-      "epoch": 2.7178163885543896,
-      "grad_norm": 0.0015160755719989538,
-      "learning_rate": 0.0001999963569626879,
-      "loss": 46.0,
-      "step": 35547
-    },
-    {
-      "epoch": 2.7178928455377793,
-      "grad_norm": 0.0011680960888043046,
-      "learning_rate": 0.0001999963567576584,
-      "loss": 46.0,
-      "step": 35548
-    },
-    {
-      "epoch": 2.717969302521169,
-      "grad_norm": 0.002470925450325012,
-      "learning_rate": 0.00019999635655262315,
-      "loss": 46.0,
-      "step": 35549
-    },
-    {
-      "epoch": 2.718045759504559,
-      "grad_norm": 0.0013076576869934797,
-      "learning_rate": 0.00019999635634758214,
-      "loss": 46.0,
-      "step": 35550
-    },
-    {
-      "epoch": 2.7181222164879486,
-      "grad_norm": 0.003878205083310604,
-      "learning_rate": 0.00019999635614253533,
-      "loss": 46.0,
-      "step": 35551
-    },
-    {
-      "epoch": 2.7181986734713384,
-      "grad_norm": 0.006067465990781784,
-      "learning_rate": 0.00019999635593748275,
-      "loss": 46.0,
-      "step": 35552
-    },
-    {
-      "epoch": 2.718275130454728,
-      "grad_norm": 0.0033899592235684395,
-      "learning_rate": 0.00019999635573242442,
-      "loss": 46.0,
-      "step": 35553
-    },
-    {
-      "epoch": 2.718351587438118,
-      "grad_norm": 0.0023160893470048904,
-      "learning_rate": 0.00019999635552736035,
-      "loss": 46.0,
-      "step": 35554
-    },
-    {
-      "epoch": 2.718428044421507,
-      "grad_norm": 0.0024210549890995026,
-      "learning_rate": 0.00019999635532229047,
-      "loss": 46.0,
-      "step": 35555
-    },
-    {
-      "epoch": 2.718504501404897,
-      "grad_norm": 0.0027642895001918077,
-      "learning_rate": 0.00019999635511721482,
-      "loss": 46.0,
-      "step": 35556
-    },
-    {
-      "epoch": 2.7185809583882867,
-      "grad_norm": 0.002420208416879177,
-      "learning_rate": 0.0001999963549121334,
-      "loss": 46.0,
-      "step": 35557
-    },
-    {
-      "epoch": 2.7186574153716765,
-      "grad_norm": 0.0031396544072777033,
-      "learning_rate": 0.00019999635470704623,
-      "loss": 46.0,
-      "step": 35558
-    },
-    {
-      "epoch": 2.7187338723550663,
-      "grad_norm": 0.0014422963140532374,
-      "learning_rate": 0.0001999963545019533,
-      "loss": 46.0,
-      "step": 35559
-    },
-    {
-      "epoch": 2.718810329338456,
-      "grad_norm": 0.0010809110244736075,
-      "learning_rate": 0.00019999635429685455,
-      "loss": 46.0,
-      "step": 35560
-    },
-    {
-      "epoch": 2.7188867863218458,
-      "grad_norm": 0.007205615285784006,
-      "learning_rate": 0.0001999963540917501,
-      "loss": 46.0,
-      "step": 35561
-    },
-    {
-      "epoch": 2.7189632433052355,
-      "grad_norm": 0.0031048774253576994,
-      "learning_rate": 0.00019999635388663983,
-      "loss": 46.0,
-      "step": 35562
-    },
-    {
-      "epoch": 2.719039700288625,
-      "grad_norm": 0.001426363829523325,
-      "learning_rate": 0.0001999963536815238,
-      "loss": 46.0,
-      "step": 35563
-    },
-    {
-      "epoch": 2.7191161572720146,
-      "grad_norm": 0.006565196439623833,
-      "learning_rate": 0.00019999635347640199,
-      "loss": 46.0,
-      "step": 35564
-    },
-    {
-      "epoch": 2.7191926142554044,
-      "grad_norm": 0.0015313653275370598,
-      "learning_rate": 0.00019999635327127443,
-      "loss": 46.0,
-      "step": 35565
-    },
-    {
-      "epoch": 2.719269071238794,
-      "grad_norm": 0.004134522285312414,
-      "learning_rate": 0.0001999963530661411,
-      "loss": 46.0,
-      "step": 35566
-    },
-    {
-      "epoch": 2.719345528222184,
-      "grad_norm": 0.001232789596542716,
-      "learning_rate": 0.000199996352861002,
-      "loss": 46.0,
-      "step": 35567
-    },
-    {
-      "epoch": 2.7194219852055737,
-      "grad_norm": 0.0021813849452883005,
-      "learning_rate": 0.00019999635265585713,
-      "loss": 46.0,
-      "step": 35568
-    },
-    {
-      "epoch": 2.7194984421889634,
-      "grad_norm": 0.0008139744168147445,
-      "learning_rate": 0.00019999635245070648,
-      "loss": 46.0,
-      "step": 35569
-    },
-    {
-      "epoch": 2.719574899172353,
-      "grad_norm": 0.005653867963701487,
-      "learning_rate": 0.00019999635224555008,
-      "loss": 46.0,
-      "step": 35570
-    },
-    {
-      "epoch": 2.719651356155743,
-      "grad_norm": 0.0007541353115811944,
-      "learning_rate": 0.0001999963520403879,
-      "loss": 46.0,
-      "step": 35571
-    },
-    {
-      "epoch": 2.7197278131391327,
-      "grad_norm": 0.0018454411765560508,
-      "learning_rate": 0.00019999635183521995,
-      "loss": 46.0,
-      "step": 35572
-    },
-    {
-      "epoch": 2.7198042701225225,
-      "grad_norm": 0.002281837398186326,
-      "learning_rate": 0.0001999963516300462,
-      "loss": 46.0,
-      "step": 35573
-    },
-    {
-      "epoch": 2.719880727105912,
-      "grad_norm": 0.0009177671163342893,
-      "learning_rate": 0.00019999635142486675,
-      "loss": 46.0,
-      "step": 35574
-    },
-    {
-      "epoch": 2.719957184089302,
-      "grad_norm": 0.003966895863413811,
-      "learning_rate": 0.00019999635121968149,
-      "loss": 46.0,
-      "step": 35575
-    },
-    {
-      "epoch": 2.7200336410726917,
-      "grad_norm": 0.0015515388222411275,
-      "learning_rate": 0.00019999635101449045,
-      "loss": 46.0,
-      "step": 35576
-    },
-    {
-      "epoch": 2.720110098056081,
-      "grad_norm": 0.002380692632868886,
-      "learning_rate": 0.00019999635080929364,
-      "loss": 46.0,
-      "step": 35577
-    },
-    {
-      "epoch": 2.720186555039471,
-      "grad_norm": 0.0020098290406167507,
-      "learning_rate": 0.0001999963506040911,
-      "loss": 46.0,
-      "step": 35578
-    },
-    {
-      "epoch": 2.7202630120228606,
-      "grad_norm": 0.0036084617022424936,
-      "learning_rate": 0.00019999635039888276,
-      "loss": 46.0,
-      "step": 35579
-    },
-    {
-      "epoch": 2.7203394690062503,
-      "grad_norm": 0.001635113381780684,
-      "learning_rate": 0.00019999635019366866,
-      "loss": 46.0,
-      "step": 35580
-    },
-    {
-      "epoch": 2.72041592598964,
-      "grad_norm": 0.0009914236143231392,
-      "learning_rate": 0.00019999634998844879,
-      "loss": 46.0,
-      "step": 35581
-    },
-    {
-      "epoch": 2.72049238297303,
-      "grad_norm": 0.0019363588653504848,
-      "learning_rate": 0.00019999634978322314,
-      "loss": 46.0,
-      "step": 35582
-    },
-    {
-      "epoch": 2.7205688399564196,
-      "grad_norm": 0.005446561146527529,
-      "learning_rate": 0.00019999634957799174,
-      "loss": 46.0,
-      "step": 35583
-    },
-    {
-      "epoch": 2.7206452969398094,
-      "grad_norm": 0.001541463891044259,
-      "learning_rate": 0.00019999634937275455,
-      "loss": 46.0,
-      "step": 35584
-    },
-    {
-      "epoch": 2.7207217539231987,
-      "grad_norm": 0.004012261051684618,
-      "learning_rate": 0.0001999963491675116,
-      "loss": 46.0,
-      "step": 35585
-    },
-    {
-      "epoch": 2.7207982109065885,
-      "grad_norm": 0.0019922463688999414,
-      "learning_rate": 0.0001999963489622629,
-      "loss": 46.0,
-      "step": 35586
-    },
-    {
-      "epoch": 2.7208746678899782,
-      "grad_norm": 0.003240313148126006,
-      "learning_rate": 0.00019999634875700838,
-      "loss": 46.0,
-      "step": 35587
-    },
-    {
-      "epoch": 2.720951124873368,
-      "grad_norm": 0.0028779287822544575,
-      "learning_rate": 0.00019999634855174815,
-      "loss": 46.0,
-      "step": 35588
-    },
-    {
-      "epoch": 2.7210275818567577,
-      "grad_norm": 0.00312438840046525,
-      "learning_rate": 0.0001999963483464821,
-      "loss": 46.0,
-      "step": 35589
-    },
-    {
-      "epoch": 2.7211040388401475,
-      "grad_norm": 0.0009869623463600874,
-      "learning_rate": 0.0001999963481412103,
-      "loss": 46.0,
-      "step": 35590
-    },
-    {
-      "epoch": 2.7211804958235373,
-      "grad_norm": 0.0036377054639160633,
-      "learning_rate": 0.00019999634793593275,
-      "loss": 46.0,
-      "step": 35591
-    },
-    {
-      "epoch": 2.721256952806927,
-      "grad_norm": 0.0024209481198340654,
-      "learning_rate": 0.00019999634773064942,
-      "loss": 46.0,
-      "step": 35592
-    },
-    {
-      "epoch": 2.721333409790317,
-      "grad_norm": 0.0050981780514121056,
-      "learning_rate": 0.00019999634752536032,
-      "loss": 46.0,
-      "step": 35593
-    },
-    {
-      "epoch": 2.7214098667737066,
-      "grad_norm": 0.0013969967840239406,
-      "learning_rate": 0.00019999634732006545,
-      "loss": 46.0,
-      "step": 35594
-    },
-    {
-      "epoch": 2.7214863237570963,
-      "grad_norm": 0.004011660348623991,
-      "learning_rate": 0.0001999963471147648,
-      "loss": 46.0,
-      "step": 35595
-    },
-    {
-      "epoch": 2.721562780740486,
-      "grad_norm": 0.002396687399595976,
-      "learning_rate": 0.00019999634690945838,
-      "loss": 46.0,
-      "step": 35596
-    },
-    {
-      "epoch": 2.721639237723876,
-      "grad_norm": 0.0010593049228191376,
-      "learning_rate": 0.00019999634670414621,
-      "loss": 46.0,
-      "step": 35597
-    },
-    {
-      "epoch": 2.7217156947072656,
-      "grad_norm": 0.0007295301184058189,
-      "learning_rate": 0.00019999634649882825,
-      "loss": 46.0,
-      "step": 35598
-    },
-    {
-      "epoch": 2.721792151690655,
-      "grad_norm": 0.0017183547606691718,
-      "learning_rate": 0.00019999634629350453,
-      "loss": 46.0,
-      "step": 35599
-    },
-    {
-      "epoch": 2.7218686086740447,
-      "grad_norm": 0.002091650851070881,
-      "learning_rate": 0.00019999634608817505,
-      "loss": 46.0,
-      "step": 35600
-    },
-    {
-      "epoch": 2.7219450656574344,
-      "grad_norm": 0.0021779905073344707,
-      "learning_rate": 0.00019999634588283979,
-      "loss": 46.0,
-      "step": 35601
-    },
-    {
-      "epoch": 2.722021522640824,
-      "grad_norm": 0.003952302038669586,
-      "learning_rate": 0.00019999634567749875,
-      "loss": 46.0,
-      "step": 35602
-    },
-    {
-      "epoch": 2.722097979624214,
-      "grad_norm": 0.0035736628342419863,
-      "learning_rate": 0.00019999634547215197,
-      "loss": 46.0,
-      "step": 35603
-    },
-    {
-      "epoch": 2.7221744366076037,
-      "grad_norm": 0.014075192622840405,
-      "learning_rate": 0.0001999963452667994,
-      "loss": 46.0,
-      "step": 35604
-    },
-    {
-      "epoch": 2.7222508935909935,
-      "grad_norm": 0.003875596448779106,
-      "learning_rate": 0.00019999634506144106,
-      "loss": 46.0,
-      "step": 35605
-    },
-    {
-      "epoch": 2.7223273505743832,
-      "grad_norm": 0.012529782019555569,
-      "learning_rate": 0.00019999634485607696,
-      "loss": 46.0,
-      "step": 35606
-    },
-    {
-      "epoch": 2.7224038075577726,
-      "grad_norm": 0.0012497693533077836,
-      "learning_rate": 0.0001999963446507071,
-      "loss": 46.0,
-      "step": 35607
-    },
-    {
-      "epoch": 2.7224802645411623,
-      "grad_norm": 0.0019326459150761366,
-      "learning_rate": 0.00019999634444533144,
-      "loss": 46.0,
-      "step": 35608
-    },
-    {
-      "epoch": 2.722556721524552,
-      "grad_norm": 0.0034439931623637676,
-      "learning_rate": 0.00019999634423995005,
-      "loss": 46.0,
-      "step": 35609
-    },
-    {
-      "epoch": 2.722633178507942,
-      "grad_norm": 0.0005621527088806033,
-      "learning_rate": 0.00019999634403456286,
-      "loss": 46.0,
-      "step": 35610
-    },
-    {
-      "epoch": 2.7227096354913316,
-      "grad_norm": 0.0011200627777725458,
-      "learning_rate": 0.00019999634382916992,
-      "loss": 46.0,
-      "step": 35611
-    },
-    {
-      "epoch": 2.7227860924747214,
-      "grad_norm": 0.0025317443069070578,
-      "learning_rate": 0.0001999963436237712,
-      "loss": 46.0,
-      "step": 35612
-    },
-    {
-      "epoch": 2.722862549458111,
-      "grad_norm": 0.008220239542424679,
-      "learning_rate": 0.00019999634341836672,
-      "loss": 46.0,
-      "step": 35613
-    },
-    {
-      "epoch": 2.722939006441501,
-      "grad_norm": 0.0037902274634689093,
-      "learning_rate": 0.00019999634321295643,
-      "loss": 46.0,
-      "step": 35614
-    },
-    {
-      "epoch": 2.7230154634248906,
-      "grad_norm": 0.0019833194091916084,
-      "learning_rate": 0.00019999634300754042,
-      "loss": 46.0,
-      "step": 35615
-    },
-    {
-      "epoch": 2.7230919204082804,
-      "grad_norm": 0.0006837555556558073,
-      "learning_rate": 0.00019999634280211864,
-      "loss": 46.0,
-      "step": 35616
-    },
-    {
-      "epoch": 2.72316837739167,
-      "grad_norm": 0.004103212617337704,
-      "learning_rate": 0.00019999634259669103,
-      "loss": 46.0,
-      "step": 35617
-    },
-    {
-      "epoch": 2.72324483437506,
-      "grad_norm": 0.0024128654040396214,
-      "learning_rate": 0.00019999634239125774,
-      "loss": 46.0,
-      "step": 35618
-    },
-    {
-      "epoch": 2.7233212913584497,
-      "grad_norm": 0.001528444467112422,
-      "learning_rate": 0.0001999963421858186,
-      "loss": 46.0,
-      "step": 35619
-    },
-    {
-      "epoch": 2.723397748341839,
-      "grad_norm": 0.0012990866089239717,
-      "learning_rate": 0.00019999634198037373,
-      "loss": 46.0,
-      "step": 35620
-    },
-    {
-      "epoch": 2.7234742053252288,
-      "grad_norm": 0.0012682609958574176,
-      "learning_rate": 0.00019999634177492312,
-      "loss": 46.0,
-      "step": 35621
-    },
-    {
-      "epoch": 2.7235506623086185,
-      "grad_norm": 0.004261441994458437,
-      "learning_rate": 0.0001999963415694667,
-      "loss": 46.0,
-      "step": 35622
-    },
-    {
-      "epoch": 2.7236271192920083,
-      "grad_norm": 0.0017196964472532272,
-      "learning_rate": 0.0001999963413640045,
-      "loss": 46.0,
-      "step": 35623
-    },
-    {
-      "epoch": 2.723703576275398,
-      "grad_norm": 0.0011589573696255684,
-      "learning_rate": 0.00019999634115853656,
-      "loss": 46.0,
-      "step": 35624
-    },
-    {
-      "epoch": 2.723780033258788,
-      "grad_norm": 0.001352710765786469,
-      "learning_rate": 0.00019999634095306282,
-      "loss": 46.0,
-      "step": 35625
-    },
-    {
-      "epoch": 2.7238564902421776,
-      "grad_norm": 0.0031290026381611824,
-      "learning_rate": 0.00019999634074758336,
-      "loss": 46.0,
-      "step": 35626
-    },
-    {
-      "epoch": 2.7239329472255673,
-      "grad_norm": 0.001416188315488398,
-      "learning_rate": 0.0001999963405420981,
-      "loss": 46.0,
-      "step": 35627
-    },
-    {
-      "epoch": 2.724009404208957,
-      "grad_norm": 0.002246958203613758,
-      "learning_rate": 0.00019999634033660707,
-      "loss": 46.0,
-      "step": 35628
-    },
-    {
-      "epoch": 2.7240858611923464,
-      "grad_norm": 0.0038429428823292255,
-      "learning_rate": 0.0001999963401311103,
-      "loss": 46.0,
-      "step": 35629
-    },
-    {
-      "epoch": 2.724162318175736,
-      "grad_norm": 0.005536423996090889,
-      "learning_rate": 0.0001999963399256077,
-      "loss": 46.0,
-      "step": 35630
-    },
-    {
-      "epoch": 2.724238775159126,
-      "grad_norm": 0.003033737652003765,
-      "learning_rate": 0.00019999633972009936,
-      "loss": 46.0,
-      "step": 35631
-    },
-    {
-      "epoch": 2.7243152321425157,
-      "grad_norm": 0.0022738086991012096,
-      "learning_rate": 0.00019999633951458528,
-      "loss": 46.0,
-      "step": 35632
-    },
-    {
-      "epoch": 2.7243916891259055,
-      "grad_norm": 0.0040948837995529175,
-      "learning_rate": 0.0001999963393090654,
-      "loss": 46.0,
-      "step": 35633
-    },
-    {
-      "epoch": 2.724468146109295,
-      "grad_norm": 0.0018557875882834196,
-      "learning_rate": 0.00019999633910353976,
-      "loss": 46.0,
-      "step": 35634
-    },
-    {
-      "epoch": 2.724544603092685,
-      "grad_norm": 0.002667769556865096,
-      "learning_rate": 0.00019999633889800834,
-      "loss": 46.0,
-      "step": 35635
-    },
-    {
-      "epoch": 2.7246210600760747,
-      "grad_norm": 0.001529740751720965,
-      "learning_rate": 0.00019999633869247115,
-      "loss": 46.0,
-      "step": 35636
-    },
-    {
-      "epoch": 2.7246975170594645,
-      "grad_norm": 0.0015548748197034001,
-      "learning_rate": 0.0001999963384869282,
-      "loss": 46.0,
-      "step": 35637
-    },
-    {
-      "epoch": 2.7247739740428543,
-      "grad_norm": 0.0031659023370593786,
-      "learning_rate": 0.0001999963382813795,
-      "loss": 46.0,
-      "step": 35638
-    },
-    {
-      "epoch": 2.724850431026244,
-      "grad_norm": 0.005182690918445587,
-      "learning_rate": 0.000199996338075825,
-      "loss": 46.0,
-      "step": 35639
-    },
-    {
-      "epoch": 2.724926888009634,
-      "grad_norm": 0.0036581619642674923,
-      "learning_rate": 0.00019999633787026475,
-      "loss": 46.0,
-      "step": 35640
-    },
-    {
-      "epoch": 2.7250033449930235,
-      "grad_norm": 0.00366486725397408,
-      "learning_rate": 0.00019999633766469872,
-      "loss": 46.0,
-      "step": 35641
-    },
-    {
-      "epoch": 2.725079801976413,
-      "grad_norm": 0.0017780952621251345,
-      "learning_rate": 0.00019999633745912694,
-      "loss": 46.0,
-      "step": 35642
-    },
-    {
-      "epoch": 2.7251562589598026,
-      "grad_norm": 0.0021709902212023735,
-      "learning_rate": 0.00019999633725354936,
-      "loss": 46.0,
-      "step": 35643
-    },
-    {
-      "epoch": 2.7252327159431924,
-      "grad_norm": 0.0026098215021193027,
-      "learning_rate": 0.000199996337047966,
-      "loss": 46.0,
-      "step": 35644
-    },
-    {
-      "epoch": 2.725309172926582,
-      "grad_norm": 0.003806529100984335,
-      "learning_rate": 0.0001999963368423769,
-      "loss": 46.0,
-      "step": 35645
-    },
-    {
-      "epoch": 2.725385629909972,
-      "grad_norm": 0.004472440108656883,
-      "learning_rate": 0.00019999633663678204,
-      "loss": 46.0,
-      "step": 35646
-    },
-    {
-      "epoch": 2.7254620868933617,
-      "grad_norm": 0.0022441651672124863,
-      "learning_rate": 0.0001999963364311814,
-      "loss": 46.0,
-      "step": 35647
-    },
-    {
-      "epoch": 2.7255385438767514,
-      "grad_norm": 0.0036506294272840023,
-      "learning_rate": 0.000199996336225575,
-      "loss": 46.0,
-      "step": 35648
-    },
-    {
-      "epoch": 2.725615000860141,
-      "grad_norm": 0.001735470024868846,
-      "learning_rate": 0.0001999963360199628,
-      "loss": 46.0,
-      "step": 35649
-    },
-    {
-      "epoch": 2.7256914578435305,
-      "grad_norm": 0.0015078134601935744,
-      "learning_rate": 0.00019999633581434484,
-      "loss": 46.0,
-      "step": 35650
-    },
-    {
-      "epoch": 2.7257679148269203,
-      "grad_norm": 0.0022744243033230305,
-      "learning_rate": 0.00019999633560872115,
-      "loss": 46.0,
-      "step": 35651
-    },
-    {
-      "epoch": 2.72584437181031,
-      "grad_norm": 0.0013780840672552586,
-      "learning_rate": 0.00019999633540309164,
-      "loss": 46.0,
-      "step": 35652
-    },
-    {
-      "epoch": 2.7259208287937,
-      "grad_norm": 0.0014011393068358302,
-      "learning_rate": 0.0001999963351974564,
-      "loss": 46.0,
-      "step": 35653
-    },
-    {
-      "epoch": 2.7259972857770896,
-      "grad_norm": 0.0027522617019712925,
-      "learning_rate": 0.00019999633499181538,
-      "loss": 46.0,
-      "step": 35654
-    },
-    {
-      "epoch": 2.7260737427604793,
-      "grad_norm": 0.003042238298803568,
-      "learning_rate": 0.00019999633478616857,
-      "loss": 46.0,
-      "step": 35655
-    },
-    {
-      "epoch": 2.726150199743869,
-      "grad_norm": 0.0026717178989201784,
-      "learning_rate": 0.000199996334580516,
-      "loss": 46.0,
-      "step": 35656
-    },
-    {
-      "epoch": 2.726226656727259,
-      "grad_norm": 0.001669479301199317,
-      "learning_rate": 0.00019999633437485766,
-      "loss": 46.0,
-      "step": 35657
-    },
-    {
-      "epoch": 2.7263031137106486,
-      "grad_norm": 0.000637056480627507,
-      "learning_rate": 0.00019999633416919357,
-      "loss": 46.0,
-      "step": 35658
-    },
-    {
-      "epoch": 2.7263795706940384,
-      "grad_norm": 0.0011748549295589328,
-      "learning_rate": 0.0001999963339635237,
-      "loss": 46.0,
-      "step": 35659
-    },
-    {
-      "epoch": 2.726456027677428,
-      "grad_norm": 0.0032335759606212378,
-      "learning_rate": 0.00019999633375784805,
-      "loss": 46.0,
-      "step": 35660
-    },
-    {
-      "epoch": 2.726532484660818,
-      "grad_norm": 0.004441841039806604,
-      "learning_rate": 0.00019999633355216663,
-      "loss": 46.0,
-      "step": 35661
-    },
-    {
-      "epoch": 2.7266089416442076,
-      "grad_norm": 0.0007603495614603162,
-      "learning_rate": 0.00019999633334647946,
-      "loss": 46.0,
-      "step": 35662
-    },
-    {
-      "epoch": 2.7266853986275974,
-      "grad_norm": 0.002256447449326515,
-      "learning_rate": 0.0001999963331407865,
-      "loss": 46.0,
-      "step": 35663
-    },
-    {
-      "epoch": 2.7267618556109867,
-      "grad_norm": 0.0007049803389236331,
-      "learning_rate": 0.00019999633293508778,
-      "loss": 46.0,
-      "step": 35664
-    },
-    {
-      "epoch": 2.7268383125943765,
-      "grad_norm": 0.0009467173367738724,
-      "learning_rate": 0.0001999963327293833,
-      "loss": 46.0,
-      "step": 35665
-    },
-    {
-      "epoch": 2.7269147695777662,
-      "grad_norm": 0.0037316635716706514,
-      "learning_rate": 0.000199996332523673,
-      "loss": 46.0,
-      "step": 35666
-    },
-    {
-      "epoch": 2.726991226561156,
-      "grad_norm": 0.0007724087918177247,
-      "learning_rate": 0.000199996332317957,
-      "loss": 46.0,
-      "step": 35667
-    },
-    {
-      "epoch": 2.7270676835445458,
-      "grad_norm": 0.0012444661697372794,
-      "learning_rate": 0.0001999963321122352,
-      "loss": 46.0,
-      "step": 35668
-    },
-    {
-      "epoch": 2.7271441405279355,
-      "grad_norm": 0.002533516613766551,
-      "learning_rate": 0.00019999633190650765,
-      "loss": 46.0,
-      "step": 35669
-    },
-    {
-      "epoch": 2.7272205975113253,
-      "grad_norm": 0.003680545138195157,
-      "learning_rate": 0.0001999963317007743,
-      "loss": 46.0,
-      "step": 35670
-    },
-    {
-      "epoch": 2.727297054494715,
-      "grad_norm": 0.003232403425499797,
-      "learning_rate": 0.0001999963314950352,
-      "loss": 46.0,
-      "step": 35671
-    },
-    {
-      "epoch": 2.7273735114781044,
-      "grad_norm": 0.0034296144731342793,
-      "learning_rate": 0.00019999633128929033,
-      "loss": 46.0,
-      "step": 35672
-    },
-    {
-      "epoch": 2.727449968461494,
-      "grad_norm": 0.003555634757503867,
-      "learning_rate": 0.00019999633108353968,
-      "loss": 46.0,
-      "step": 35673
-    },
-    {
-      "epoch": 2.727526425444884,
-      "grad_norm": 0.00254733394831419,
-      "learning_rate": 0.0001999963308777833,
-      "loss": 46.0,
-      "step": 35674
-    },
-    {
-      "epoch": 2.7276028824282736,
-      "grad_norm": 0.00363207352347672,
-      "learning_rate": 0.00019999633067202107,
-      "loss": 46.0,
-      "step": 35675
-    },
-    {
-      "epoch": 2.7276793394116634,
-      "grad_norm": 0.0008738340111449361,
-      "learning_rate": 0.00019999633046625316,
-      "loss": 46.0,
-      "step": 35676
-    },
-    {
-      "epoch": 2.727755796395053,
-      "grad_norm": 0.0018527911743149161,
-      "learning_rate": 0.00019999633026047945,
-      "loss": 46.0,
-      "step": 35677
-    },
-    {
-      "epoch": 2.727832253378443,
-      "grad_norm": 0.0008666341891512275,
-      "learning_rate": 0.00019999633005469996,
-      "loss": 46.0,
-      "step": 35678
-    },
-    {
-      "epoch": 2.7279087103618327,
-      "grad_norm": 0.0027163878548890352,
-      "learning_rate": 0.00019999632984891468,
-      "loss": 46.0,
-      "step": 35679
-    },
-    {
-      "epoch": 2.7279851673452225,
-      "grad_norm": 0.001168629969470203,
-      "learning_rate": 0.00019999632964312364,
-      "loss": 46.0,
-      "step": 35680
-    },
-    {
-      "epoch": 2.728061624328612,
-      "grad_norm": 0.0011977845570072532,
-      "learning_rate": 0.00019999632943732687,
-      "loss": 46.0,
-      "step": 35681
-    },
-    {
-      "epoch": 2.728138081312002,
-      "grad_norm": 0.0009905926417559385,
-      "learning_rate": 0.0001999963292315243,
-      "loss": 46.0,
-      "step": 35682
-    },
-    {
-      "epoch": 2.7282145382953917,
-      "grad_norm": 0.0015684437239542603,
-      "learning_rate": 0.00019999632902571593,
-      "loss": 46.0,
-      "step": 35683
-    },
-    {
-      "epoch": 2.7282909952787815,
-      "grad_norm": 0.0016992834862321615,
-      "learning_rate": 0.00019999632881990184,
-      "loss": 46.0,
-      "step": 35684
-    },
-    {
-      "epoch": 2.7283674522621713,
-      "grad_norm": 0.0010718251578509808,
-      "learning_rate": 0.00019999632861408196,
-      "loss": 46.0,
-      "step": 35685
-    },
-    {
-      "epoch": 2.7284439092455606,
-      "grad_norm": 0.002412281697615981,
-      "learning_rate": 0.00019999632840825632,
-      "loss": 46.0,
-      "step": 35686
-    },
-    {
-      "epoch": 2.7285203662289503,
-      "grad_norm": 0.0036779099609702826,
-      "learning_rate": 0.0001999963282024249,
-      "loss": 46.0,
-      "step": 35687
-    },
-    {
-      "epoch": 2.72859682321234,
-      "grad_norm": 0.0032590029295533895,
-      "learning_rate": 0.00019999632799658774,
-      "loss": 46.0,
-      "step": 35688
-    },
-    {
-      "epoch": 2.72867328019573,
-      "grad_norm": 0.00938719604164362,
-      "learning_rate": 0.0001999963277907448,
-      "loss": 46.0,
-      "step": 35689
-    },
-    {
-      "epoch": 2.7287497371791196,
-      "grad_norm": 0.0025378817226737738,
-      "learning_rate": 0.00019999632758489609,
-      "loss": 46.0,
-      "step": 35690
-    },
-    {
-      "epoch": 2.7288261941625094,
-      "grad_norm": 0.0009176959283649921,
-      "learning_rate": 0.0001999963273790416,
-      "loss": 46.0,
-      "step": 35691
-    },
-    {
-      "epoch": 2.728902651145899,
-      "grad_norm": 0.007104227319359779,
-      "learning_rate": 0.00019999632717318134,
-      "loss": 46.0,
-      "step": 35692
-    },
-    {
-      "epoch": 2.728979108129289,
-      "grad_norm": 0.0013569480506703258,
-      "learning_rate": 0.00019999632696731528,
-      "loss": 46.0,
-      "step": 35693
-    },
-    {
-      "epoch": 2.729055565112678,
-      "grad_norm": 0.004530667793005705,
-      "learning_rate": 0.00019999632676144348,
-      "loss": 46.0,
-      "step": 35694
-    },
-    {
-      "epoch": 2.729132022096068,
-      "grad_norm": 0.0025318230036646128,
-      "learning_rate": 0.00019999632655556593,
-      "loss": 46.0,
-      "step": 35695
-    },
-    {
-      "epoch": 2.7292084790794577,
-      "grad_norm": 0.001357459113933146,
-      "learning_rate": 0.0001999963263496826,
-      "loss": 46.0,
-      "step": 35696
-    },
-    {
-      "epoch": 2.7292849360628475,
-      "grad_norm": 0.007906139828264713,
-      "learning_rate": 0.00019999632614379348,
-      "loss": 46.0,
-      "step": 35697
-    },
-    {
-      "epoch": 2.7293613930462373,
-      "grad_norm": 0.0061918217688798904,
-      "learning_rate": 0.0001999963259378986,
-      "loss": 46.0,
-      "step": 35698
-    },
-    {
-      "epoch": 2.729437850029627,
-      "grad_norm": 0.0012706361012533307,
-      "learning_rate": 0.000199996325731998,
-      "loss": 46.0,
-      "step": 35699
-    },
-    {
-      "epoch": 2.729514307013017,
-      "grad_norm": 0.0025306467432528734,
-      "learning_rate": 0.00019999632552609157,
-      "loss": 46.0,
-      "step": 35700
-    },
-    {
-      "epoch": 2.7295907639964065,
-      "grad_norm": 0.0007252030191011727,
-      "learning_rate": 0.00019999632532017938,
-      "loss": 46.0,
-      "step": 35701
-    },
-    {
-      "epoch": 2.7296672209797963,
-      "grad_norm": 0.0014977443497627974,
-      "learning_rate": 0.00019999632511426144,
-      "loss": 46.0,
-      "step": 35702
-    },
-    {
-      "epoch": 2.729743677963186,
-      "grad_norm": 0.010900103487074375,
-      "learning_rate": 0.0001999963249083377,
-      "loss": 46.0,
-      "step": 35703
-    },
-    {
-      "epoch": 2.729820134946576,
-      "grad_norm": 0.0011279555037617683,
-      "learning_rate": 0.00019999632470240825,
-      "loss": 46.0,
-      "step": 35704
-    },
-    {
-      "epoch": 2.7298965919299656,
-      "grad_norm": 0.0035279574804008007,
-      "learning_rate": 0.00019999632449647296,
-      "loss": 46.0,
-      "step": 35705
-    },
-    {
-      "epoch": 2.7299730489133553,
-      "grad_norm": 0.004021984990686178,
-      "learning_rate": 0.00019999632429053193,
-      "loss": 46.0,
-      "step": 35706
-    },
-    {
-      "epoch": 2.730049505896745,
-      "grad_norm": 0.0011964804725721478,
-      "learning_rate": 0.00019999632408458515,
-      "loss": 46.0,
-      "step": 35707
-    },
-    {
-      "epoch": 2.7301259628801344,
-      "grad_norm": 0.0013586652930825949,
-      "learning_rate": 0.00019999632387863257,
-      "loss": 46.0,
-      "step": 35708
-    },
-    {
-      "epoch": 2.730202419863524,
-      "grad_norm": 0.001162472297437489,
-      "learning_rate": 0.00019999632367267425,
-      "loss": 46.0,
-      "step": 35709
-    },
-    {
-      "epoch": 2.730278876846914,
-      "grad_norm": 0.001532992348074913,
-      "learning_rate": 0.00019999632346671012,
-      "loss": 46.0,
-      "step": 35710
-    },
-    {
-      "epoch": 2.7303553338303037,
-      "grad_norm": 0.0024481674190610647,
-      "learning_rate": 0.00019999632326074025,
-      "loss": 46.0,
-      "step": 35711
-    },
-    {
-      "epoch": 2.7304317908136935,
-      "grad_norm": 0.0020043975673615932,
-      "learning_rate": 0.0001999963230547646,
-      "loss": 46.0,
-      "step": 35712
-    },
-    {
-      "epoch": 2.7305082477970832,
-      "grad_norm": 0.0006835907697677612,
-      "learning_rate": 0.0001999963228487832,
-      "loss": 46.0,
-      "step": 35713
-    },
-    {
-      "epoch": 2.730584704780473,
-      "grad_norm": 0.0015180815244093537,
-      "learning_rate": 0.00019999632264279603,
-      "loss": 46.0,
-      "step": 35714
-    },
-    {
-      "epoch": 2.7306611617638628,
-      "grad_norm": 0.0008863538387231529,
-      "learning_rate": 0.00019999632243680306,
-      "loss": 46.0,
-      "step": 35715
-    },
-    {
-      "epoch": 2.730737618747252,
-      "grad_norm": 0.001259046490304172,
-      "learning_rate": 0.00019999632223080432,
-      "loss": 46.0,
-      "step": 35716
-    },
-    {
-      "epoch": 2.730814075730642,
-      "grad_norm": 0.0015540440799668431,
-      "learning_rate": 0.00019999632202479987,
-      "loss": 46.0,
-      "step": 35717
-    },
-    {
-      "epoch": 2.7308905327140316,
-      "grad_norm": 0.002205285709351301,
-      "learning_rate": 0.0001999963218187896,
-      "loss": 46.0,
-      "step": 35718
-    },
-    {
-      "epoch": 2.7309669896974214,
-      "grad_norm": 0.002824450610205531,
-      "learning_rate": 0.00019999632161277358,
-      "loss": 46.0,
-      "step": 35719
-    },
-    {
-      "epoch": 2.731043446680811,
-      "grad_norm": 0.002018020721152425,
-      "learning_rate": 0.00019999632140675177,
-      "loss": 46.0,
-      "step": 35720
-    },
-    {
-      "epoch": 2.731119903664201,
-      "grad_norm": 0.0010712331859394908,
-      "learning_rate": 0.0001999963212007242,
-      "loss": 46.0,
-      "step": 35721
-    },
-    {
-      "epoch": 2.7311963606475906,
-      "grad_norm": 0.0025074591394513845,
-      "learning_rate": 0.00019999632099469084,
-      "loss": 46.0,
-      "step": 35722
-    },
-    {
-      "epoch": 2.7312728176309804,
-      "grad_norm": 0.0017509774770587683,
-      "learning_rate": 0.00019999632078865175,
-      "loss": 46.0,
-      "step": 35723
-    },
-    {
-      "epoch": 2.73134927461437,
-      "grad_norm": 0.003326605772599578,
-      "learning_rate": 0.00019999632058260688,
-      "loss": 46.0,
-      "step": 35724
-    },
-    {
-      "epoch": 2.73142573159776,
-      "grad_norm": 0.003630370367318392,
-      "learning_rate": 0.00019999632037655626,
-      "loss": 46.0,
-      "step": 35725
-    },
-    {
-      "epoch": 2.7315021885811497,
-      "grad_norm": 0.0009780654218047857,
-      "learning_rate": 0.00019999632017049984,
-      "loss": 46.0,
-      "step": 35726
-    },
-    {
-      "epoch": 2.7315786455645394,
-      "grad_norm": 0.0018377938540652394,
-      "learning_rate": 0.00019999631996443765,
-      "loss": 46.0,
-      "step": 35727
-    },
-    {
-      "epoch": 2.731655102547929,
-      "grad_norm": 0.0017985814483836293,
-      "learning_rate": 0.00019999631975836969,
-      "loss": 46.0,
-      "step": 35728
-    },
-    {
-      "epoch": 2.731731559531319,
-      "grad_norm": 0.001997133484110236,
-      "learning_rate": 0.00019999631955229597,
-      "loss": 46.0,
-      "step": 35729
-    },
-    {
-      "epoch": 2.7318080165147083,
-      "grad_norm": 0.0013265230227261782,
-      "learning_rate": 0.00019999631934621646,
-      "loss": 46.0,
-      "step": 35730
-    },
-    {
-      "epoch": 2.731884473498098,
-      "grad_norm": 0.002895316341891885,
-      "learning_rate": 0.00019999631914013123,
-      "loss": 46.0,
-      "step": 35731
-    },
-    {
-      "epoch": 2.731960930481488,
-      "grad_norm": 0.001926775206811726,
-      "learning_rate": 0.0001999963189340402,
-      "loss": 46.0,
-      "step": 35732
-    },
-    {
-      "epoch": 2.7320373874648776,
-      "grad_norm": 0.0025911573320627213,
-      "learning_rate": 0.0001999963187279434,
-      "loss": 46.0,
-      "step": 35733
-    },
-    {
-      "epoch": 2.7321138444482673,
-      "grad_norm": 0.002382179256528616,
-      "learning_rate": 0.00019999631852184085,
-      "loss": 46.0,
-      "step": 35734
-    },
-    {
-      "epoch": 2.732190301431657,
-      "grad_norm": 0.0013867170782759786,
-      "learning_rate": 0.00019999631831573252,
-      "loss": 46.0,
-      "step": 35735
-    },
-    {
-      "epoch": 2.732266758415047,
-      "grad_norm": 0.0007950286963023245,
-      "learning_rate": 0.0001999963181096184,
-      "loss": 46.0,
-      "step": 35736
-    },
-    {
-      "epoch": 2.7323432153984366,
-      "grad_norm": 0.002047120127826929,
-      "learning_rate": 0.00019999631790349853,
-      "loss": 46.0,
-      "step": 35737
-    },
-    {
-      "epoch": 2.732419672381826,
-      "grad_norm": 0.00254746968857944,
-      "learning_rate": 0.00019999631769737289,
-      "loss": 46.0,
-      "step": 35738
-    },
-    {
-      "epoch": 2.7324961293652157,
-      "grad_norm": 0.0014278758317232132,
-      "learning_rate": 0.00019999631749124147,
-      "loss": 46.0,
-      "step": 35739
-    },
-    {
-      "epoch": 2.7325725863486054,
-      "grad_norm": 0.00474910531193018,
-      "learning_rate": 0.00019999631728510428,
-      "loss": 46.0,
-      "step": 35740
-    },
-    {
-      "epoch": 2.732649043331995,
-      "grad_norm": 0.002569854026660323,
-      "learning_rate": 0.00019999631707896134,
-      "loss": 46.0,
-      "step": 35741
-    },
-    {
-      "epoch": 2.732725500315385,
-      "grad_norm": 0.0026957355439662933,
-      "learning_rate": 0.0001999963168728126,
-      "loss": 46.0,
-      "step": 35742
-    },
-    {
-      "epoch": 2.7328019572987747,
-      "grad_norm": 0.002226717071607709,
-      "learning_rate": 0.00019999631666665812,
-      "loss": 46.0,
-      "step": 35743
-    },
-    {
-      "epoch": 2.7328784142821645,
-      "grad_norm": 0.0013536263722926378,
-      "learning_rate": 0.0001999963164604979,
-      "loss": 46.0,
-      "step": 35744
-    },
-    {
-      "epoch": 2.7329548712655543,
-      "grad_norm": 0.0018991974648088217,
-      "learning_rate": 0.00019999631625433183,
-      "loss": 46.0,
-      "step": 35745
-    },
-    {
-      "epoch": 2.733031328248944,
-      "grad_norm": 0.0013735548127442598,
-      "learning_rate": 0.00019999631604816003,
-      "loss": 46.0,
-      "step": 35746
-    },
-    {
-      "epoch": 2.7331077852323338,
-      "grad_norm": 0.0017134378431364894,
-      "learning_rate": 0.00019999631584198245,
-      "loss": 46.0,
-      "step": 35747
-    },
-    {
-      "epoch": 2.7331842422157235,
-      "grad_norm": 0.0024166374932974577,
-      "learning_rate": 0.00019999631563579913,
-      "loss": 46.0,
-      "step": 35748
-    },
-    {
-      "epoch": 2.7332606991991133,
-      "grad_norm": 0.0044545698910951614,
-      "learning_rate": 0.00019999631542961003,
-      "loss": 46.0,
-      "step": 35749
-    },
-    {
-      "epoch": 2.733337156182503,
-      "grad_norm": 0.0028468146920204163,
-      "learning_rate": 0.00019999631522341516,
-      "loss": 46.0,
-      "step": 35750
-    },
-    {
-      "epoch": 2.7334136131658924,
-      "grad_norm": 0.0013901748461648822,
-      "learning_rate": 0.00019999631501721452,
-      "loss": 46.0,
-      "step": 35751
-    },
-    {
-      "epoch": 2.733490070149282,
-      "grad_norm": 0.0034736236557364464,
-      "learning_rate": 0.0001999963148110081,
-      "loss": 46.0,
-      "step": 35752
-    },
-    {
-      "epoch": 2.733566527132672,
-      "grad_norm": 0.0017775128362700343,
-      "learning_rate": 0.00019999631460479594,
-      "loss": 46.0,
-      "step": 35753
-    },
-    {
-      "epoch": 2.7336429841160617,
-      "grad_norm": 0.003621581243351102,
-      "learning_rate": 0.00019999631439857795,
-      "loss": 46.0,
-      "step": 35754
-    },
-    {
-      "epoch": 2.7337194410994514,
-      "grad_norm": 0.0021192056592553854,
-      "learning_rate": 0.00019999631419235426,
-      "loss": 46.0,
-      "step": 35755
-    },
-    {
-      "epoch": 2.733795898082841,
-      "grad_norm": 0.0020271912217140198,
-      "learning_rate": 0.00019999631398612478,
-      "loss": 46.0,
-      "step": 35756
-    },
-    {
-      "epoch": 2.733872355066231,
-      "grad_norm": 0.0035420232452452183,
-      "learning_rate": 0.0001999963137798895,
-      "loss": 46.0,
-      "step": 35757
-    },
-    {
-      "epoch": 2.7339488120496207,
-      "grad_norm": 0.0020351996645331383,
-      "learning_rate": 0.00019999631357364847,
-      "loss": 46.0,
-      "step": 35758
-    },
-    {
-      "epoch": 2.7340252690330105,
-      "grad_norm": 0.003690814832225442,
-      "learning_rate": 0.00019999631336740166,
-      "loss": 46.0,
-      "step": 35759
-    },
-    {
-      "epoch": 2.7341017260164,
-      "grad_norm": 0.0050978162325918674,
-      "learning_rate": 0.0001999963131611491,
-      "loss": 46.0,
-      "step": 35760
-    },
-    {
-      "epoch": 2.7341781829997895,
-      "grad_norm": 0.0019029497634619474,
-      "learning_rate": 0.00019999631295489076,
-      "loss": 46.0,
-      "step": 35761
-    },
-    {
-      "epoch": 2.7342546399831793,
-      "grad_norm": 0.005722524598240852,
-      "learning_rate": 0.00019999631274862667,
-      "loss": 46.0,
-      "step": 35762
-    },
-    {
-      "epoch": 2.734331096966569,
-      "grad_norm": 0.0033656407613307238,
-      "learning_rate": 0.00019999631254235677,
-      "loss": 46.0,
-      "step": 35763
-    },
-    {
-      "epoch": 2.734407553949959,
-      "grad_norm": 0.0016470981063321233,
-      "learning_rate": 0.00019999631233608115,
-      "loss": 46.0,
-      "step": 35764
-    },
-    {
-      "epoch": 2.7344840109333486,
-      "grad_norm": 0.001653734128922224,
-      "learning_rate": 0.00019999631212979974,
-      "loss": 46.0,
-      "step": 35765
-    },
-    {
-      "epoch": 2.7345604679167383,
-      "grad_norm": 0.0025281873531639576,
-      "learning_rate": 0.00019999631192351255,
-      "loss": 46.0,
-      "step": 35766
-    },
-    {
-      "epoch": 2.734636924900128,
-      "grad_norm": 0.0008028540760278702,
-      "learning_rate": 0.0001999963117172196,
-      "loss": 46.0,
-      "step": 35767
-    },
-    {
-      "epoch": 2.734713381883518,
-      "grad_norm": 0.0012554824352264404,
-      "learning_rate": 0.00019999631151092087,
-      "loss": 46.0,
-      "step": 35768
-    },
-    {
-      "epoch": 2.7347898388669076,
-      "grad_norm": 0.0021011321805417538,
-      "learning_rate": 0.0001999963113046164,
-      "loss": 46.0,
-      "step": 35769
-    },
-    {
-      "epoch": 2.7348662958502974,
-      "grad_norm": 0.0006718956283293664,
-      "learning_rate": 0.00019999631109830613,
-      "loss": 46.0,
-      "step": 35770
-    },
-    {
-      "epoch": 2.734942752833687,
-      "grad_norm": 0.0012235450558364391,
-      "learning_rate": 0.0001999963108919901,
-      "loss": 46.0,
-      "step": 35771
-    },
-    {
-      "epoch": 2.735019209817077,
-      "grad_norm": 0.005615275353193283,
-      "learning_rate": 0.00019999631068566827,
-      "loss": 46.0,
-      "step": 35772
-    },
-    {
-      "epoch": 2.7350956668004662,
-      "grad_norm": 0.0024202633649110794,
-      "learning_rate": 0.00019999631047934072,
-      "loss": 46.0,
-      "step": 35773
-    },
-    {
-      "epoch": 2.735172123783856,
-      "grad_norm": 0.0009801515843719244,
-      "learning_rate": 0.0001999963102730074,
-      "loss": 46.0,
-      "step": 35774
-    },
-    {
-      "epoch": 2.7352485807672458,
-      "grad_norm": 0.0037219219375401735,
-      "learning_rate": 0.00019999631006666828,
-      "loss": 46.0,
-      "step": 35775
-    },
-    {
-      "epoch": 2.7353250377506355,
-      "grad_norm": 0.002658379962667823,
-      "learning_rate": 0.00019999630986032344,
-      "loss": 46.0,
-      "step": 35776
-    },
-    {
-      "epoch": 2.7354014947340253,
-      "grad_norm": 0.0014870022423565388,
-      "learning_rate": 0.00019999630965397277,
-      "loss": 46.0,
-      "step": 35777
-    },
-    {
-      "epoch": 2.735477951717415,
-      "grad_norm": 0.0005365445977076888,
-      "learning_rate": 0.00019999630944761635,
-      "loss": 46.0,
-      "step": 35778
-    },
-    {
-      "epoch": 2.735554408700805,
-      "grad_norm": 0.0019919397309422493,
-      "learning_rate": 0.0001999963092412542,
-      "loss": 46.0,
-      "step": 35779
-    },
-    {
-      "epoch": 2.7356308656841946,
-      "grad_norm": 0.004397986456751823,
-      "learning_rate": 0.00019999630903488622,
-      "loss": 46.0,
-      "step": 35780
-    },
-    {
-      "epoch": 2.735707322667584,
-      "grad_norm": 0.001296763657592237,
-      "learning_rate": 0.00019999630882851251,
-      "loss": 46.0,
-      "step": 35781
-    },
-    {
-      "epoch": 2.7357837796509736,
-      "grad_norm": 0.0016664611175656319,
-      "learning_rate": 0.000199996308622133,
-      "loss": 46.0,
-      "step": 35782
-    },
-    {
-      "epoch": 2.7358602366343634,
-      "grad_norm": 0.0016566052800044417,
-      "learning_rate": 0.00019999630841574778,
-      "loss": 46.0,
-      "step": 35783
-    },
-    {
-      "epoch": 2.735936693617753,
-      "grad_norm": 0.008721900172531605,
-      "learning_rate": 0.00019999630820935672,
-      "loss": 46.0,
-      "step": 35784
-    },
-    {
-      "epoch": 2.736013150601143,
-      "grad_norm": 0.0013912980211898685,
-      "learning_rate": 0.00019999630800295992,
-      "loss": 46.0,
-      "step": 35785
-    },
-    {
-      "epoch": 2.7360896075845327,
-      "grad_norm": 0.0019085319945588708,
-      "learning_rate": 0.00019999630779655737,
-      "loss": 46.0,
-      "step": 35786
-    },
-    {
-      "epoch": 2.7361660645679224,
-      "grad_norm": 0.002847634954378009,
-      "learning_rate": 0.00019999630759014902,
-      "loss": 46.0,
-      "step": 35787
-    },
-    {
-      "epoch": 2.736242521551312,
-      "grad_norm": 0.0035724679473787546,
-      "learning_rate": 0.00019999630738373492,
-      "loss": 46.0,
-      "step": 35788
-    },
-    {
-      "epoch": 2.736318978534702,
-      "grad_norm": 0.0010674335062503815,
-      "learning_rate": 0.00019999630717731502,
-      "loss": 46.0,
-      "step": 35789
-    },
-    {
-      "epoch": 2.7363954355180917,
-      "grad_norm": 0.002224615076556802,
-      "learning_rate": 0.0001999963069708894,
-      "loss": 46.0,
-      "step": 35790
-    },
-    {
-      "epoch": 2.7364718925014815,
-      "grad_norm": 0.0008803022210486233,
-      "learning_rate": 0.000199996306764458,
-      "loss": 46.0,
-      "step": 35791
-    },
-    {
-      "epoch": 2.7365483494848712,
-      "grad_norm": 0.0015376565279439092,
-      "learning_rate": 0.00019999630655802083,
-      "loss": 46.0,
-      "step": 35792
-    },
-    {
-      "epoch": 2.736624806468261,
-      "grad_norm": 0.0029344032518565655,
-      "learning_rate": 0.00019999630635157787,
-      "loss": 46.0,
-      "step": 35793
-    },
-    {
-      "epoch": 2.7367012634516508,
-      "grad_norm": 0.0033081460278481245,
-      "learning_rate": 0.00019999630614512913,
-      "loss": 46.0,
-      "step": 35794
-    },
-    {
-      "epoch": 2.73677772043504,
-      "grad_norm": 0.0015810366021469235,
-      "learning_rate": 0.00019999630593867465,
-      "loss": 46.0,
-      "step": 35795
-    },
-    {
-      "epoch": 2.73685417741843,
-      "grad_norm": 0.0069196452386677265,
-      "learning_rate": 0.00019999630573221437,
-      "loss": 46.0,
-      "step": 35796
-    },
-    {
-      "epoch": 2.7369306344018196,
-      "grad_norm": 0.003984841983765364,
-      "learning_rate": 0.00019999630552574834,
-      "loss": 46.0,
-      "step": 35797
-    },
-    {
-      "epoch": 2.7370070913852094,
-      "grad_norm": 0.0014060544781386852,
-      "learning_rate": 0.00019999630531927656,
-      "loss": 46.0,
-      "step": 35798
-    },
-    {
-      "epoch": 2.737083548368599,
-      "grad_norm": 0.006560854613780975,
-      "learning_rate": 0.00019999630511279899,
-      "loss": 46.0,
-      "step": 35799
-    },
-    {
-      "epoch": 2.737160005351989,
-      "grad_norm": 0.0018308509606868029,
-      "learning_rate": 0.00019999630490631564,
-      "loss": 46.0,
-      "step": 35800
-    },
-    {
-      "epoch": 2.7372364623353787,
-      "grad_norm": 0.007635123562067747,
-      "learning_rate": 0.00019999630469982654,
-      "loss": 46.0,
-      "step": 35801
-    },
-    {
-      "epoch": 2.7373129193187684,
-      "grad_norm": 0.0012488069478422403,
-      "learning_rate": 0.00019999630449333167,
-      "loss": 46.0,
-      "step": 35802
-    },
-    {
-      "epoch": 2.7373893763021577,
-      "grad_norm": 0.003487673122435808,
-      "learning_rate": 0.00019999630428683103,
-      "loss": 46.0,
-      "step": 35803
-    },
-    {
-      "epoch": 2.7374658332855475,
-      "grad_norm": 0.0012271342566236854,
-      "learning_rate": 0.00019999630408032461,
-      "loss": 46.0,
-      "step": 35804
-    },
-    {
-      "epoch": 2.7375422902689373,
-      "grad_norm": 0.0009661121875979006,
-      "learning_rate": 0.00019999630387381243,
-      "loss": 46.0,
-      "step": 35805
-    },
-    {
-      "epoch": 2.737618747252327,
-      "grad_norm": 0.0016262778080999851,
-      "learning_rate": 0.00019999630366729446,
-      "loss": 46.0,
-      "step": 35806
-    },
-    {
-      "epoch": 2.7376952042357168,
-      "grad_norm": 0.0008614902035333216,
-      "learning_rate": 0.00019999630346077075,
-      "loss": 46.0,
-      "step": 35807
-    },
-    {
-      "epoch": 2.7377716612191065,
-      "grad_norm": 0.0013567577116191387,
-      "learning_rate": 0.00019999630325424125,
-      "loss": 46.0,
-      "step": 35808
-    },
-    {
-      "epoch": 2.7378481182024963,
-      "grad_norm": 0.0010432898998260498,
-      "learning_rate": 0.000199996303047706,
-      "loss": 46.0,
-      "step": 35809
-    },
-    {
-      "epoch": 2.737924575185886,
-      "grad_norm": 0.005283687729388475,
-      "learning_rate": 0.00019999630284116496,
-      "loss": 46.0,
-      "step": 35810
-    },
-    {
-      "epoch": 2.738001032169276,
-      "grad_norm": 0.007014344446361065,
-      "learning_rate": 0.00019999630263461819,
-      "loss": 46.0,
-      "step": 35811
-    },
-    {
-      "epoch": 2.7380774891526656,
-      "grad_norm": 0.0013796740677207708,
-      "learning_rate": 0.00019999630242806558,
-      "loss": 46.0,
-      "step": 35812
-    },
-    {
-      "epoch": 2.7381539461360553,
-      "grad_norm": 0.0013678011018782854,
-      "learning_rate": 0.00019999630222150726,
-      "loss": 46.0,
-      "step": 35813
-    },
-    {
-      "epoch": 2.738230403119445,
-      "grad_norm": 0.0013025131775066257,
-      "learning_rate": 0.00019999630201494317,
-      "loss": 46.0,
-      "step": 35814
-    },
-    {
-      "epoch": 2.738306860102835,
-      "grad_norm": 0.0008969233022071421,
-      "learning_rate": 0.0001999963018083733,
-      "loss": 46.0,
-      "step": 35815
-    },
-    {
-      "epoch": 2.7383833170862246,
-      "grad_norm": 0.002112897578626871,
-      "learning_rate": 0.00019999630160179766,
-      "loss": 46.0,
-      "step": 35816
-    },
-    {
-      "epoch": 2.738459774069614,
-      "grad_norm": 0.0024056946858763695,
-      "learning_rate": 0.00019999630139521624,
-      "loss": 46.0,
-      "step": 35817
-    },
-    {
-      "epoch": 2.7385362310530037,
-      "grad_norm": 0.0019872456323355436,
-      "learning_rate": 0.00019999630118862905,
-      "loss": 46.0,
-      "step": 35818
-    },
-    {
-      "epoch": 2.7386126880363935,
-      "grad_norm": 0.002187522826716304,
-      "learning_rate": 0.0001999963009820361,
-      "loss": 46.0,
-      "step": 35819
-    },
-    {
-      "epoch": 2.7386891450197832,
-      "grad_norm": 0.0013820001622661948,
-      "learning_rate": 0.00019999630077543738,
-      "loss": 46.0,
-      "step": 35820
-    },
-    {
-      "epoch": 2.738765602003173,
-      "grad_norm": 0.003323413198813796,
-      "learning_rate": 0.00019999630056883287,
-      "loss": 46.0,
-      "step": 35821
-    },
-    {
-      "epoch": 2.7388420589865627,
-      "grad_norm": 0.0035491979215294123,
-      "learning_rate": 0.00019999630036222264,
-      "loss": 46.0,
-      "step": 35822
-    },
-    {
-      "epoch": 2.7389185159699525,
-      "grad_norm": 0.0008238217560574412,
-      "learning_rate": 0.0001999963001556066,
-      "loss": 46.0,
-      "step": 35823
-    },
-    {
-      "epoch": 2.7389949729533423,
-      "grad_norm": 0.002367644337937236,
-      "learning_rate": 0.00019999629994898479,
-      "loss": 46.0,
-      "step": 35824
-    },
-    {
-      "epoch": 2.7390714299367316,
-      "grad_norm": 0.002570437965914607,
-      "learning_rate": 0.00019999629974235724,
-      "loss": 46.0,
-      "step": 35825
-    },
-    {
-      "epoch": 2.7391478869201213,
-      "grad_norm": 0.0036188121885061264,
-      "learning_rate": 0.0001999962995357239,
-      "loss": 46.0,
-      "step": 35826
-    },
-    {
-      "epoch": 2.739224343903511,
-      "grad_norm": 0.0014184325700625777,
-      "learning_rate": 0.0001999962993290848,
-      "loss": 46.0,
-      "step": 35827
-    },
-    {
-      "epoch": 2.739300800886901,
-      "grad_norm": 0.0021532007958739996,
-      "learning_rate": 0.0001999962991224399,
-      "loss": 46.0,
-      "step": 35828
-    },
-    {
-      "epoch": 2.7393772578702906,
-      "grad_norm": 0.002907169982790947,
-      "learning_rate": 0.00019999629891578926,
-      "loss": 46.0,
-      "step": 35829
-    },
-    {
-      "epoch": 2.7394537148536804,
-      "grad_norm": 0.0019722990691661835,
-      "learning_rate": 0.00019999629870913287,
-      "loss": 46.0,
-      "step": 35830
-    },
-    {
-      "epoch": 2.73953017183707,
-      "grad_norm": 0.0025312057696282864,
-      "learning_rate": 0.00019999629850247068,
-      "loss": 46.0,
-      "step": 35831
-    },
-    {
-      "epoch": 2.73960662882046,
-      "grad_norm": 0.0027756057679653168,
-      "learning_rate": 0.00019999629829580272,
-      "loss": 46.0,
-      "step": 35832
-    },
-    {
-      "epoch": 2.7396830858038497,
-      "grad_norm": 0.0014693514676764607,
-      "learning_rate": 0.000199996298089129,
-      "loss": 46.0,
-      "step": 35833
-    },
-    {
-      "epoch": 2.7397595427872394,
-      "grad_norm": 0.002156902104616165,
-      "learning_rate": 0.0001999962978824495,
-      "loss": 46.0,
-      "step": 35834
-    },
-    {
-      "epoch": 2.739835999770629,
-      "grad_norm": 0.000835475861094892,
-      "learning_rate": 0.00019999629767576425,
-      "loss": 46.0,
-      "step": 35835
-    },
-    {
-      "epoch": 2.739912456754019,
-      "grad_norm": 0.00239433441311121,
-      "learning_rate": 0.00019999629746907322,
-      "loss": 46.0,
-      "step": 35836
-    },
-    {
-      "epoch": 2.7399889137374087,
-      "grad_norm": 0.003271217690780759,
-      "learning_rate": 0.00019999629726237642,
-      "loss": 46.0,
-      "step": 35837
-    },
-    {
-      "epoch": 2.7400653707207985,
-      "grad_norm": 0.0021858541294932365,
-      "learning_rate": 0.00019999629705567387,
-      "loss": 46.0,
-      "step": 35838
-    },
-    {
-      "epoch": 2.740141827704188,
-      "grad_norm": 0.0009301493410021067,
-      "learning_rate": 0.00019999629684896552,
-      "loss": 46.0,
-      "step": 35839
-    },
-    {
-      "epoch": 2.7402182846875776,
-      "grad_norm": 0.003283920930698514,
-      "learning_rate": 0.0001999962966422514,
-      "loss": 46.0,
-      "step": 35840
-    },
-    {
-      "epoch": 2.7402947416709673,
-      "grad_norm": 0.0007741856388747692,
-      "learning_rate": 0.00019999629643553153,
-      "loss": 46.0,
-      "step": 35841
-    },
-    {
-      "epoch": 2.740371198654357,
-      "grad_norm": 0.003037592861801386,
-      "learning_rate": 0.0001999962962288059,
-      "loss": 46.0,
-      "step": 35842
-    },
-    {
-      "epoch": 2.740447655637747,
-      "grad_norm": 0.0016240758122876287,
-      "learning_rate": 0.00019999629602207448,
-      "loss": 46.0,
-      "step": 35843
-    },
-    {
-      "epoch": 2.7405241126211366,
-      "grad_norm": 0.0018706904957070947,
-      "learning_rate": 0.0001999962958153373,
-      "loss": 46.0,
-      "step": 35844
-    },
-    {
-      "epoch": 2.7406005696045264,
-      "grad_norm": 0.0018460426945239305,
-      "learning_rate": 0.00019999629560859435,
-      "loss": 46.0,
-      "step": 35845
-    },
-    {
-      "epoch": 2.740677026587916,
-      "grad_norm": 0.0030594519339501858,
-      "learning_rate": 0.00019999629540184565,
-      "loss": 46.0,
-      "step": 35846
-    },
-    {
-      "epoch": 2.7407534835713054,
-      "grad_norm": 0.0038096415810287,
-      "learning_rate": 0.00019999629519509114,
-      "loss": 46.0,
-      "step": 35847
-    },
-    {
-      "epoch": 2.740829940554695,
-      "grad_norm": 0.002642523031681776,
-      "learning_rate": 0.00019999629498833086,
-      "loss": 46.0,
-      "step": 35848
-    },
-    {
-      "epoch": 2.740906397538085,
-      "grad_norm": 0.0019593259785324335,
-      "learning_rate": 0.00019999629478156483,
-      "loss": 46.0,
-      "step": 35849
-    },
-    {
-      "epoch": 2.7409828545214747,
-      "grad_norm": 0.0007764084148220718,
-      "learning_rate": 0.00019999629457479303,
-      "loss": 46.0,
-      "step": 35850
-    },
-    {
-      "epoch": 2.7410593115048645,
-      "grad_norm": 0.002579642226919532,
-      "learning_rate": 0.00019999629436801545,
-      "loss": 46.0,
-      "step": 35851
-    },
-    {
-      "epoch": 2.7411357684882542,
-      "grad_norm": 0.0026985201984643936,
-      "learning_rate": 0.00019999629416123213,
-      "loss": 46.0,
-      "step": 35852
-    },
-    {
-      "epoch": 2.741212225471644,
-      "grad_norm": 0.0014592364896088839,
-      "learning_rate": 0.00019999629395444304,
-      "loss": 46.0,
-      "step": 35853
-    },
-    {
-      "epoch": 2.7412886824550338,
-      "grad_norm": 0.0005410446901805699,
-      "learning_rate": 0.00019999629374764814,
-      "loss": 46.0,
-      "step": 35854
-    },
-    {
-      "epoch": 2.7413651394384235,
-      "grad_norm": 0.0033891620114445686,
-      "learning_rate": 0.00019999629354084753,
-      "loss": 46.0,
-      "step": 35855
-    },
-    {
-      "epoch": 2.7414415964218133,
-      "grad_norm": 0.008296642452478409,
-      "learning_rate": 0.00019999629333404109,
-      "loss": 46.0,
-      "step": 35856
-    },
-    {
-      "epoch": 2.741518053405203,
-      "grad_norm": 0.004144382197409868,
-      "learning_rate": 0.00019999629312722892,
-      "loss": 46.0,
-      "step": 35857
-    },
-    {
-      "epoch": 2.741594510388593,
-      "grad_norm": 0.0008417487842962146,
-      "learning_rate": 0.00019999629292041096,
-      "loss": 46.0,
-      "step": 35858
-    },
-    {
-      "epoch": 2.7416709673719826,
-      "grad_norm": 0.0037346736062318087,
-      "learning_rate": 0.00019999629271358723,
-      "loss": 46.0,
-      "step": 35859
-    },
-    {
-      "epoch": 2.7417474243553723,
-      "grad_norm": 0.0009178059990517795,
-      "learning_rate": 0.00019999629250675772,
-      "loss": 46.0,
-      "step": 35860
-    },
-    {
-      "epoch": 2.7418238813387616,
-      "grad_norm": 0.005763344932347536,
-      "learning_rate": 0.00019999629229992247,
-      "loss": 46.0,
-      "step": 35861
-    },
-    {
-      "epoch": 2.7419003383221514,
-      "grad_norm": 0.0063377609476447105,
-      "learning_rate": 0.00019999629209308147,
-      "loss": 46.0,
-      "step": 35862
-    },
-    {
-      "epoch": 2.741976795305541,
-      "grad_norm": 0.0018648783443495631,
-      "learning_rate": 0.00019999629188623467,
-      "loss": 46.0,
-      "step": 35863
-    },
-    {
-      "epoch": 2.742053252288931,
-      "grad_norm": 0.001431356300599873,
-      "learning_rate": 0.0001999962916793821,
-      "loss": 46.0,
-      "step": 35864
-    },
-    {
-      "epoch": 2.7421297092723207,
-      "grad_norm": 0.005612307693809271,
-      "learning_rate": 0.00019999629147252374,
-      "loss": 46.0,
-      "step": 35865
-    },
-    {
-      "epoch": 2.7422061662557105,
-      "grad_norm": 0.0014139526756480336,
-      "learning_rate": 0.00019999629126565965,
-      "loss": 46.0,
-      "step": 35866
-    },
-    {
-      "epoch": 2.7422826232391,
-      "grad_norm": 0.0016323266318067908,
-      "learning_rate": 0.00019999629105878976,
-      "loss": 46.0,
-      "step": 35867
-    },
-    {
-      "epoch": 2.74235908022249,
-      "grad_norm": 0.003030815627425909,
-      "learning_rate": 0.00019999629085191414,
-      "loss": 46.0,
-      "step": 35868
-    },
-    {
-      "epoch": 2.7424355372058793,
-      "grad_norm": 0.004062585532665253,
-      "learning_rate": 0.0001999962906450327,
-      "loss": 46.0,
-      "step": 35869
-    },
-    {
-      "epoch": 2.742511994189269,
-      "grad_norm": 0.0009527135989628732,
-      "learning_rate": 0.0001999962904381455,
-      "loss": 46.0,
-      "step": 35870
-    },
-    {
-      "epoch": 2.742588451172659,
-      "grad_norm": 0.0015048084314912558,
-      "learning_rate": 0.00019999629023125258,
-      "loss": 46.0,
-      "step": 35871
-    },
-    {
-      "epoch": 2.7426649081560486,
-      "grad_norm": 0.0022276260424405336,
-      "learning_rate": 0.00019999629002435384,
-      "loss": 46.0,
-      "step": 35872
-    },
-    {
-      "epoch": 2.7427413651394383,
-      "grad_norm": 0.0015374044887721539,
-      "learning_rate": 0.00019999628981744936,
-      "loss": 46.0,
-      "step": 35873
-    },
-    {
-      "epoch": 2.742817822122828,
-      "grad_norm": 0.0025957589969038963,
-      "learning_rate": 0.00019999628961053908,
-      "loss": 46.0,
-      "step": 35874
-    },
-    {
-      "epoch": 2.742894279106218,
-      "grad_norm": 0.0025320476852357388,
-      "learning_rate": 0.00019999628940362308,
-      "loss": 46.0,
-      "step": 35875
-    },
-    {
-      "epoch": 2.7429707360896076,
-      "grad_norm": 0.003992842510342598,
-      "learning_rate": 0.00019999628919670128,
-      "loss": 46.0,
-      "step": 35876
-    },
-    {
-      "epoch": 2.7430471930729974,
-      "grad_norm": 0.0008204219047911465,
-      "learning_rate": 0.00019999628898977368,
-      "loss": 46.0,
-      "step": 35877
-    },
-    {
-      "epoch": 2.743123650056387,
-      "grad_norm": 0.0045212795957922935,
-      "learning_rate": 0.0001999962887828404,
-      "loss": 46.0,
-      "step": 35878
-    },
-    {
-      "epoch": 2.743200107039777,
-      "grad_norm": 0.0022172306198626757,
-      "learning_rate": 0.00019999628857590124,
-      "loss": 46.0,
-      "step": 35879
-    },
-    {
-      "epoch": 2.7432765640231667,
-      "grad_norm": 0.0025309273041784763,
-      "learning_rate": 0.00019999628836895637,
-      "loss": 46.0,
-      "step": 35880
-    },
-    {
-      "epoch": 2.7433530210065564,
-      "grad_norm": 0.0032449280843138695,
-      "learning_rate": 0.00019999628816200576,
-      "loss": 46.0,
-      "step": 35881
-    },
-    {
-      "epoch": 2.7434294779899457,
-      "grad_norm": 0.0018204592633992434,
-      "learning_rate": 0.00019999628795504932,
-      "loss": 46.0,
-      "step": 35882
-    },
-    {
-      "epoch": 2.7435059349733355,
-      "grad_norm": 0.0017473356565460563,
-      "learning_rate": 0.00019999628774808713,
-      "loss": 46.0,
-      "step": 35883
-    },
-    {
-      "epoch": 2.7435823919567253,
-      "grad_norm": 0.0010288238991051912,
-      "learning_rate": 0.0001999962875411192,
-      "loss": 46.0,
-      "step": 35884
-    },
-    {
-      "epoch": 2.743658848940115,
-      "grad_norm": 0.0011149861384183168,
-      "learning_rate": 0.0001999962873341455,
-      "loss": 46.0,
-      "step": 35885
-    },
-    {
-      "epoch": 2.743735305923505,
-      "grad_norm": 0.0027040743734687567,
-      "learning_rate": 0.00019999628712716598,
-      "loss": 46.0,
-      "step": 35886
-    },
-    {
-      "epoch": 2.7438117629068945,
-      "grad_norm": 0.0010989997535943985,
-      "learning_rate": 0.0001999962869201807,
-      "loss": 46.0,
-      "step": 35887
-    },
-    {
-      "epoch": 2.7438882198902843,
-      "grad_norm": 0.003336984198540449,
-      "learning_rate": 0.00019999628671318968,
-      "loss": 46.0,
-      "step": 35888
-    },
-    {
-      "epoch": 2.743964676873674,
-      "grad_norm": 0.0012599349720403552,
-      "learning_rate": 0.00019999628650619288,
-      "loss": 46.0,
-      "step": 35889
-    },
-    {
-      "epoch": 2.744041133857064,
-      "grad_norm": 0.0048316605389118195,
-      "learning_rate": 0.00019999628629919033,
-      "loss": 46.0,
-      "step": 35890
-    },
-    {
-      "epoch": 2.744117590840453,
-      "grad_norm": 0.00418563699349761,
-      "learning_rate": 0.00019999628609218198,
-      "loss": 46.0,
-      "step": 35891
-    },
-    {
-      "epoch": 2.744194047823843,
-      "grad_norm": 0.0033980535808950663,
-      "learning_rate": 0.0001999962858851679,
-      "loss": 46.0,
-      "step": 35892
-    },
-    {
-      "epoch": 2.7442705048072327,
-      "grad_norm": 0.0039049030747264624,
-      "learning_rate": 0.00019999628567814802,
-      "loss": 46.0,
-      "step": 35893
-    },
-    {
-      "epoch": 2.7443469617906224,
-      "grad_norm": 0.002149237785488367,
-      "learning_rate": 0.00019999628547112238,
-      "loss": 46.0,
-      "step": 35894
-    },
-    {
-      "epoch": 2.744423418774012,
-      "grad_norm": 0.0016097250627353787,
-      "learning_rate": 0.00019999628526409094,
-      "loss": 46.0,
-      "step": 35895
-    },
-    {
-      "epoch": 2.744499875757402,
-      "grad_norm": 0.004439320880919695,
-      "learning_rate": 0.00019999628505705375,
-      "loss": 46.0,
-      "step": 35896
-    },
-    {
-      "epoch": 2.7445763327407917,
-      "grad_norm": 0.004211506340652704,
-      "learning_rate": 0.0001999962848500108,
-      "loss": 46.0,
-      "step": 35897
-    },
-    {
-      "epoch": 2.7446527897241815,
-      "grad_norm": 0.002718964358791709,
-      "learning_rate": 0.00019999628464296206,
-      "loss": 46.0,
-      "step": 35898
-    },
-    {
-      "epoch": 2.7447292467075712,
-      "grad_norm": 0.0012025551404803991,
-      "learning_rate": 0.00019999628443590758,
-      "loss": 46.0,
-      "step": 35899
-    },
-    {
-      "epoch": 2.744805703690961,
-      "grad_norm": 0.0016557564958930016,
-      "learning_rate": 0.00019999628422884733,
-      "loss": 46.0,
-      "step": 35900
-    },
-    {
-      "epoch": 2.7448821606743508,
-      "grad_norm": 0.0010153529001399875,
-      "learning_rate": 0.0001999962840217813,
-      "loss": 46.0,
-      "step": 35901
-    },
-    {
-      "epoch": 2.7449586176577405,
-      "grad_norm": 0.002281966619193554,
-      "learning_rate": 0.0001999962838147095,
-      "loss": 46.0,
-      "step": 35902
-    },
-    {
-      "epoch": 2.7450350746411303,
-      "grad_norm": 0.0016252151690423489,
-      "learning_rate": 0.00019999628360763193,
-      "loss": 46.0,
-      "step": 35903
-    },
-    {
-      "epoch": 2.7451115316245196,
-      "grad_norm": 0.0010124027030542493,
-      "learning_rate": 0.00019999628340054858,
-      "loss": 46.0,
-      "step": 35904
-    },
-    {
-      "epoch": 2.7451879886079094,
-      "grad_norm": 0.0011132683139294386,
-      "learning_rate": 0.0001999962831934595,
-      "loss": 46.0,
-      "step": 35905
-    },
-    {
-      "epoch": 2.745264445591299,
-      "grad_norm": 0.0018804400460794568,
-      "learning_rate": 0.00019999628298636462,
-      "loss": 46.0,
-      "step": 35906
-    },
-    {
-      "epoch": 2.745340902574689,
-      "grad_norm": 0.002009012270718813,
-      "learning_rate": 0.00019999628277926395,
-      "loss": 46.0,
-      "step": 35907
-    },
-    {
-      "epoch": 2.7454173595580786,
-      "grad_norm": 0.002859940053895116,
-      "learning_rate": 0.00019999628257215757,
-      "loss": 46.0,
-      "step": 35908
-    },
-    {
-      "epoch": 2.7454938165414684,
-      "grad_norm": 0.0019179665250703692,
-      "learning_rate": 0.00019999628236504538,
-      "loss": 46.0,
-      "step": 35909
-    },
-    {
-      "epoch": 2.745570273524858,
-      "grad_norm": 0.0022360177244991064,
-      "learning_rate": 0.0001999962821579274,
-      "loss": 46.0,
-      "step": 35910
-    },
-    {
-      "epoch": 2.745646730508248,
-      "grad_norm": 0.002752751810476184,
-      "learning_rate": 0.0001999962819508037,
-      "loss": 46.0,
-      "step": 35911
-    },
-    {
-      "epoch": 2.7457231874916372,
-      "grad_norm": 0.0011977370595559478,
-      "learning_rate": 0.0001999962817436742,
-      "loss": 46.0,
-      "step": 35912
-    },
-    {
-      "epoch": 2.745799644475027,
-      "grad_norm": 0.0015832443023100495,
-      "learning_rate": 0.00019999628153653896,
-      "loss": 46.0,
-      "step": 35913
-    },
-    {
-      "epoch": 2.7458761014584168,
-      "grad_norm": 0.004873890429735184,
-      "learning_rate": 0.0001999962813293979,
-      "loss": 46.0,
-      "step": 35914
-    },
-    {
-      "epoch": 2.7459525584418065,
-      "grad_norm": 0.003168850438669324,
-      "learning_rate": 0.0001999962811222511,
-      "loss": 46.0,
-      "step": 35915
-    },
-    {
-      "epoch": 2.7460290154251963,
-      "grad_norm": 0.003808521432802081,
-      "learning_rate": 0.00019999628091509856,
-      "loss": 46.0,
-      "step": 35916
-    },
-    {
-      "epoch": 2.746105472408586,
-      "grad_norm": 0.0008133513038046658,
-      "learning_rate": 0.0001999962807079402,
-      "loss": 46.0,
-      "step": 35917
-    },
-    {
-      "epoch": 2.746181929391976,
-      "grad_norm": 0.0031596627086400986,
-      "learning_rate": 0.0001999962805007761,
-      "loss": 46.0,
-      "step": 35918
-    },
-    {
-      "epoch": 2.7462583863753656,
-      "grad_norm": 0.0028175306506454945,
-      "learning_rate": 0.00019999628029360622,
-      "loss": 46.0,
-      "step": 35919
-    },
-    {
-      "epoch": 2.7463348433587553,
-      "grad_norm": 0.0016469416441395879,
-      "learning_rate": 0.00019999628008643056,
-      "loss": 46.0,
-      "step": 35920
-    },
-    {
-      "epoch": 2.746411300342145,
-      "grad_norm": 0.00614013010635972,
-      "learning_rate": 0.00019999627987924917,
-      "loss": 46.0,
-      "step": 35921
-    },
-    {
-      "epoch": 2.746487757325535,
-      "grad_norm": 0.0020270897075533867,
-      "learning_rate": 0.00019999627967206199,
-      "loss": 46.0,
-      "step": 35922
-    },
-    {
-      "epoch": 2.7465642143089246,
-      "grad_norm": 0.014481998980045319,
-      "learning_rate": 0.00019999627946486903,
-      "loss": 46.0,
-      "step": 35923
-    },
-    {
-      "epoch": 2.7466406712923144,
-      "grad_norm": 0.0012304041301831603,
-      "learning_rate": 0.0001999962792576703,
-      "loss": 46.0,
-      "step": 35924
-    },
-    {
-      "epoch": 2.746717128275704,
-      "grad_norm": 0.003599070245400071,
-      "learning_rate": 0.00019999627905046581,
-      "loss": 46.0,
-      "step": 35925
-    },
-    {
-      "epoch": 2.7467935852590935,
-      "grad_norm": 0.0024315344635397196,
-      "learning_rate": 0.00019999627884325553,
-      "loss": 46.0,
-      "step": 35926
-    },
-    {
-      "epoch": 2.746870042242483,
-      "grad_norm": 0.0017161985160782933,
-      "learning_rate": 0.0001999962786360395,
-      "loss": 46.0,
-      "step": 35927
-    },
-    {
-      "epoch": 2.746946499225873,
-      "grad_norm": 0.0007929989951662719,
-      "learning_rate": 0.0001999962784288177,
-      "loss": 46.0,
-      "step": 35928
-    },
-    {
-      "epoch": 2.7470229562092627,
-      "grad_norm": 0.005236527882516384,
-      "learning_rate": 0.00019999627822159016,
-      "loss": 46.0,
-      "step": 35929
-    },
-    {
-      "epoch": 2.7470994131926525,
-      "grad_norm": 0.00308925355784595,
-      "learning_rate": 0.00019999627801435682,
-      "loss": 46.0,
-      "step": 35930
-    },
-    {
-      "epoch": 2.7471758701760423,
-      "grad_norm": 0.004056284204125404,
-      "learning_rate": 0.0001999962778071177,
-      "loss": 46.0,
-      "step": 35931
-    },
-    {
-      "epoch": 2.747252327159432,
-      "grad_norm": 0.004575693979859352,
-      "learning_rate": 0.00019999627759987283,
-      "loss": 46.0,
-      "step": 35932
-    },
-    {
-      "epoch": 2.747328784142822,
-      "grad_norm": 0.0008140190038830042,
-      "learning_rate": 0.0001999962773926222,
-      "loss": 46.0,
-      "step": 35933
-    },
-    {
-      "epoch": 2.747405241126211,
-      "grad_norm": 0.0013864230131730437,
-      "learning_rate": 0.00019999627718536575,
-      "loss": 46.0,
-      "step": 35934
-    },
-    {
-      "epoch": 2.747481698109601,
-      "grad_norm": 0.0034038848243653774,
-      "learning_rate": 0.0001999962769781036,
-      "loss": 46.0,
-      "step": 35935
-    },
-    {
-      "epoch": 2.7475581550929906,
-      "grad_norm": 0.003007151186466217,
-      "learning_rate": 0.00019999627677083563,
-      "loss": 46.0,
-      "step": 35936
-    },
-    {
-      "epoch": 2.7476346120763804,
-      "grad_norm": 0.0012666697148233652,
-      "learning_rate": 0.0001999962765635619,
-      "loss": 46.0,
-      "step": 35937
-    },
-    {
-      "epoch": 2.74771106905977,
-      "grad_norm": 0.005064800847321749,
-      "learning_rate": 0.00019999627635628242,
-      "loss": 46.0,
-      "step": 35938
-    },
-    {
-      "epoch": 2.74778752604316,
-      "grad_norm": 0.002488100901246071,
-      "learning_rate": 0.00019999627614899717,
-      "loss": 46.0,
-      "step": 35939
-    },
-    {
-      "epoch": 2.7478639830265497,
-      "grad_norm": 0.004570317454636097,
-      "learning_rate": 0.00019999627594170614,
-      "loss": 46.0,
-      "step": 35940
-    },
-    {
-      "epoch": 2.7479404400099394,
-      "grad_norm": 0.00286275171674788,
-      "learning_rate": 0.00019999627573440935,
-      "loss": 46.0,
-      "step": 35941
-    },
-    {
-      "epoch": 2.748016896993329,
-      "grad_norm": 0.0007824230706319213,
-      "learning_rate": 0.00019999627552710675,
-      "loss": 46.0,
-      "step": 35942
-    },
-    {
-      "epoch": 2.748093353976719,
-      "grad_norm": 0.004667325410991907,
-      "learning_rate": 0.0001999962753197984,
-      "loss": 46.0,
-      "step": 35943
-    },
-    {
-      "epoch": 2.7481698109601087,
-      "grad_norm": 0.003174850717186928,
-      "learning_rate": 0.0001999962751124843,
-      "loss": 46.0,
-      "step": 35944
-    },
-    {
-      "epoch": 2.7482462679434985,
-      "grad_norm": 0.002118821721524,
-      "learning_rate": 0.00019999627490516444,
-      "loss": 46.0,
-      "step": 35945
-    },
-    {
-      "epoch": 2.7483227249268882,
-      "grad_norm": 0.00480992067605257,
-      "learning_rate": 0.0001999962746978388,
-      "loss": 46.0,
-      "step": 35946
-    },
-    {
-      "epoch": 2.748399181910278,
-      "grad_norm": 0.0017848351271823049,
-      "learning_rate": 0.00019999627449050736,
-      "loss": 46.0,
-      "step": 35947
-    },
-    {
-      "epoch": 2.7484756388936673,
-      "grad_norm": 0.0010206529404968023,
-      "learning_rate": 0.0001999962742831702,
-      "loss": 46.0,
-      "step": 35948
-    },
-    {
-      "epoch": 2.748552095877057,
-      "grad_norm": 0.007254617754369974,
-      "learning_rate": 0.00019999627407582725,
-      "loss": 46.0,
-      "step": 35949
-    },
-    {
-      "epoch": 2.748628552860447,
-      "grad_norm": 0.0021318329963833094,
-      "learning_rate": 0.00019999627386847851,
-      "loss": 46.0,
-      "step": 35950
-    },
-    {
-      "epoch": 2.7487050098438366,
-      "grad_norm": 0.003249607514590025,
-      "learning_rate": 0.000199996273661124,
-      "loss": 46.0,
-      "step": 35951
-    },
-    {
-      "epoch": 2.7487814668272263,
-      "grad_norm": 0.0022035473957657814,
-      "learning_rate": 0.00019999627345376376,
-      "loss": 46.0,
-      "step": 35952
-    },
-    {
-      "epoch": 2.748857923810616,
-      "grad_norm": 0.0011220560409128666,
-      "learning_rate": 0.00019999627324639773,
-      "loss": 46.0,
-      "step": 35953
-    },
-    {
-      "epoch": 2.748934380794006,
-      "grad_norm": 0.005362468305975199,
-      "learning_rate": 0.00019999627303902596,
-      "loss": 46.0,
-      "step": 35954
-    },
-    {
-      "epoch": 2.7490108377773956,
-      "grad_norm": 0.00683090602979064,
-      "learning_rate": 0.00019999627283164836,
-      "loss": 46.0,
-      "step": 35955
-    },
-    {
-      "epoch": 2.749087294760785,
-      "grad_norm": 0.004931860137730837,
-      "learning_rate": 0.00019999627262426502,
-      "loss": 46.0,
-      "step": 35956
-    },
-    {
-      "epoch": 2.7491637517441747,
-      "grad_norm": 0.0010500292992219329,
-      "learning_rate": 0.00019999627241687592,
-      "loss": 46.0,
-      "step": 35957
-    },
-    {
-      "epoch": 2.7492402087275645,
-      "grad_norm": 0.002588866977021098,
-      "learning_rate": 0.00019999627220948106,
-      "loss": 46.0,
-      "step": 35958
-    },
-    {
-      "epoch": 2.7493166657109542,
-      "grad_norm": 0.0017054423224180937,
-      "learning_rate": 0.0001999962720020804,
-      "loss": 46.0,
-      "step": 35959
-    },
-    {
-      "epoch": 2.749393122694344,
-      "grad_norm": 0.0036954416427761316,
-      "learning_rate": 0.000199996271794674,
-      "loss": 46.0,
-      "step": 35960
-    },
-    {
-      "epoch": 2.7494695796777338,
-      "grad_norm": 0.0021198117174208164,
-      "learning_rate": 0.0001999962715872618,
-      "loss": 46.0,
-      "step": 35961
-    },
-    {
-      "epoch": 2.7495460366611235,
-      "grad_norm": 0.00528778275474906,
-      "learning_rate": 0.00019999627137984384,
-      "loss": 46.0,
-      "step": 35962
-    },
-    {
-      "epoch": 2.7496224936445133,
-      "grad_norm": 0.0011940563563257456,
-      "learning_rate": 0.00019999627117242013,
-      "loss": 46.0,
-      "step": 35963
-    },
-    {
-      "epoch": 2.749698950627903,
-      "grad_norm": 0.0019459824543446302,
-      "learning_rate": 0.00019999627096499063,
-      "loss": 46.0,
-      "step": 35964
-    },
-    {
-      "epoch": 2.749775407611293,
-      "grad_norm": 0.001858714036643505,
-      "learning_rate": 0.00019999627075755537,
-      "loss": 46.0,
-      "step": 35965
-    },
-    {
-      "epoch": 2.7498518645946826,
-      "grad_norm": 0.005933698266744614,
-      "learning_rate": 0.00019999627055011435,
-      "loss": 46.0,
-      "step": 35966
-    },
-    {
-      "epoch": 2.7499283215780723,
-      "grad_norm": 0.00395730696618557,
-      "learning_rate": 0.00019999627034266755,
-      "loss": 46.0,
-      "step": 35967
-    },
-    {
-      "epoch": 2.750004778561462,
-      "grad_norm": 0.003128431737422943,
-      "learning_rate": 0.00019999627013521498,
-      "loss": 46.0,
-      "step": 35968
-    },
-    {
-      "epoch": 2.750081235544852,
-      "grad_norm": 0.0015686870319768786,
-      "learning_rate": 0.0001999962699277566,
-      "loss": 46.0,
-      "step": 35969
-    },
-    {
-      "epoch": 2.750157692528241,
-      "grad_norm": 0.0016519950004294515,
-      "learning_rate": 0.00019999626972029252,
-      "loss": 46.0,
-      "step": 35970
-    },
-    {
-      "epoch": 2.750234149511631,
-      "grad_norm": 0.0013959029456600547,
-      "learning_rate": 0.00019999626951282268,
-      "loss": 46.0,
-      "step": 35971
-    },
-    {
-      "epoch": 2.7503106064950207,
-      "grad_norm": 0.0047317915596067905,
-      "learning_rate": 0.000199996269305347,
-      "loss": 46.0,
-      "step": 35972
-    },
-    {
-      "epoch": 2.7503870634784104,
-      "grad_norm": 0.004043021705001593,
-      "learning_rate": 0.00019999626909786557,
-      "loss": 46.0,
-      "step": 35973
-    },
-    {
-      "epoch": 2.7504635204618,
-      "grad_norm": 0.0009831939823925495,
-      "learning_rate": 0.0001999962688903784,
-      "loss": 46.0,
-      "step": 35974
-    },
-    {
-      "epoch": 2.75053997744519,
-      "grad_norm": 0.001072244718670845,
-      "learning_rate": 0.00019999626868288546,
-      "loss": 46.0,
-      "step": 35975
-    },
-    {
-      "epoch": 2.7506164344285797,
-      "grad_norm": 0.008005915209650993,
-      "learning_rate": 0.00019999626847538673,
-      "loss": 46.0,
-      "step": 35976
-    },
-    {
-      "epoch": 2.7506928914119695,
-      "grad_norm": 0.0027740371879190207,
-      "learning_rate": 0.00019999626826788225,
-      "loss": 46.0,
-      "step": 35977
-    },
-    {
-      "epoch": 2.750769348395359,
-      "grad_norm": 0.0008578238193877041,
-      "learning_rate": 0.00019999626806037197,
-      "loss": 46.0,
-      "step": 35978
-    },
-    {
-      "epoch": 2.7508458053787486,
-      "grad_norm": 0.002180932555347681,
-      "learning_rate": 0.00019999626785285594,
-      "loss": 46.0,
-      "step": 35979
-    },
-    {
-      "epoch": 2.7509222623621383,
-      "grad_norm": 0.005434263963252306,
-      "learning_rate": 0.00019999626764533415,
-      "loss": 46.0,
-      "step": 35980
-    },
-    {
-      "epoch": 2.750998719345528,
-      "grad_norm": 0.005442031193524599,
-      "learning_rate": 0.00019999626743780657,
-      "loss": 46.0,
-      "step": 35981
-    },
-    {
-      "epoch": 2.751075176328918,
-      "grad_norm": 0.0021151516120880842,
-      "learning_rate": 0.00019999626723027323,
-      "loss": 46.0,
-      "step": 35982
-    },
-    {
-      "epoch": 2.7511516333123076,
-      "grad_norm": 0.0027981321327388287,
-      "learning_rate": 0.00019999626702273414,
-      "loss": 46.0,
-      "step": 35983
-    },
-    {
-      "epoch": 2.7512280902956974,
-      "grad_norm": 0.0025466277729719877,
-      "learning_rate": 0.00019999626681518925,
-      "loss": 46.0,
-      "step": 35984
-    },
-    {
-      "epoch": 2.751304547279087,
-      "grad_norm": 0.009051463566720486,
-      "learning_rate": 0.0001999962666076386,
-      "loss": 46.0,
-      "step": 35985
-    },
-    {
-      "epoch": 2.751381004262477,
-      "grad_norm": 0.0024369198363274336,
-      "learning_rate": 0.0001999962664000822,
-      "loss": 46.0,
-      "step": 35986
-    },
-    {
-      "epoch": 2.7514574612458667,
-      "grad_norm": 0.0011120055569335818,
-      "learning_rate": 0.00019999626619252,
-      "loss": 46.0,
-      "step": 35987
-    },
-    {
-      "epoch": 2.7515339182292564,
-      "grad_norm": 0.0014921552501618862,
-      "learning_rate": 0.00019999626598495205,
-      "loss": 46.0,
-      "step": 35988
-    },
-    {
-      "epoch": 2.751610375212646,
-      "grad_norm": 0.003287061583250761,
-      "learning_rate": 0.00019999626577737832,
-      "loss": 46.0,
-      "step": 35989
-    },
-    {
-      "epoch": 2.751686832196036,
-      "grad_norm": 0.000821066030766815,
-      "learning_rate": 0.00019999626556979885,
-      "loss": 46.0,
-      "step": 35990
-    },
-    {
-      "epoch": 2.7517632891794257,
-      "grad_norm": 0.004554067738354206,
-      "learning_rate": 0.00019999626536221357,
-      "loss": 46.0,
-      "step": 35991
-    },
-    {
-      "epoch": 2.751839746162815,
-      "grad_norm": 0.005245990585535765,
-      "learning_rate": 0.00019999626515462254,
-      "loss": 46.0,
-      "step": 35992
-    },
-    {
-      "epoch": 2.7519162031462048,
-      "grad_norm": 0.002175557194277644,
-      "learning_rate": 0.00019999626494702575,
-      "loss": 46.0,
-      "step": 35993
-    },
-    {
-      "epoch": 2.7519926601295945,
-      "grad_norm": 0.0015694230096414685,
-      "learning_rate": 0.00019999626473942317,
-      "loss": 46.0,
-      "step": 35994
-    },
-    {
-      "epoch": 2.7520691171129843,
-      "grad_norm": 0.0018481186125427485,
-      "learning_rate": 0.00019999626453181483,
-      "loss": 46.0,
-      "step": 35995
-    },
-    {
-      "epoch": 2.752145574096374,
-      "grad_norm": 0.0029249568469822407,
-      "learning_rate": 0.00019999626432420074,
-      "loss": 46.0,
-      "step": 35996
-    },
-    {
-      "epoch": 2.752222031079764,
-      "grad_norm": 0.002087234053760767,
-      "learning_rate": 0.00019999626411658085,
-      "loss": 46.0,
-      "step": 35997
-    },
-    {
-      "epoch": 2.7522984880631536,
-      "grad_norm": 0.0006739587406627834,
-      "learning_rate": 0.0001999962639089552,
-      "loss": 46.0,
-      "step": 35998
-    },
-    {
-      "epoch": 2.7523749450465433,
-      "grad_norm": 0.002084338106215,
-      "learning_rate": 0.00019999626370132377,
-      "loss": 46.0,
-      "step": 35999
-    },
-    {
-      "epoch": 2.7524514020299327,
-      "grad_norm": 0.0034885734785348177,
-      "learning_rate": 0.00019999626349368661,
-      "loss": 46.0,
-      "step": 36000
-    },
-    {
-      "epoch": 2.7525278590133224,
-      "grad_norm": 0.004073447547852993,
-      "learning_rate": 0.00019999626328604366,
-      "loss": 46.0,
-      "step": 36001
-    },
-    {
-      "epoch": 2.752604315996712,
-      "grad_norm": 0.008138821460306644,
-      "learning_rate": 0.00019999626307839493,
-      "loss": 46.0,
-      "step": 36002
-    },
-    {
-      "epoch": 2.752680772980102,
-      "grad_norm": 0.00457029277458787,
-      "learning_rate": 0.00019999626287074045,
-      "loss": 46.0,
-      "step": 36003
-    },
-    {
-      "epoch": 2.7527572299634917,
-      "grad_norm": 0.0012259484501555562,
-      "learning_rate": 0.00019999626266308017,
-      "loss": 46.0,
-      "step": 36004
-    },
-    {
-      "epoch": 2.7528336869468815,
-      "grad_norm": 0.003478058846667409,
-      "learning_rate": 0.00019999626245541415,
-      "loss": 46.0,
-      "step": 36005
-    },
-    {
-      "epoch": 2.7529101439302712,
-      "grad_norm": 0.0032879735808819532,
-      "learning_rate": 0.00019999626224774235,
-      "loss": 46.0,
-      "step": 36006
-    },
-    {
-      "epoch": 2.752986600913661,
-      "grad_norm": 0.0019854367710649967,
-      "learning_rate": 0.00019999626204006478,
-      "loss": 46.0,
-      "step": 36007
-    },
-    {
-      "epoch": 2.7530630578970507,
-      "grad_norm": 0.001890413579531014,
-      "learning_rate": 0.00019999626183238146,
-      "loss": 46.0,
-      "step": 36008
-    },
-    {
-      "epoch": 2.7531395148804405,
-      "grad_norm": 0.001008575432933867,
-      "learning_rate": 0.00019999626162469234,
-      "loss": 46.0,
-      "step": 36009
-    },
-    {
-      "epoch": 2.7532159718638303,
-      "grad_norm": 0.0013948059640824795,
-      "learning_rate": 0.00019999626141699745,
-      "loss": 46.0,
-      "step": 36010
-    },
-    {
-      "epoch": 2.75329242884722,
-      "grad_norm": 0.0015301278326660395,
-      "learning_rate": 0.0001999962612092968,
-      "loss": 46.0,
-      "step": 36011
-    },
-    {
-      "epoch": 2.75336888583061,
-      "grad_norm": 0.0025844676420092583,
-      "learning_rate": 0.00019999626100159038,
-      "loss": 46.0,
-      "step": 36012
-    },
-    {
-      "epoch": 2.753445342813999,
-      "grad_norm": 0.0061693922616541386,
-      "learning_rate": 0.00019999626079387822,
-      "loss": 46.0,
-      "step": 36013
-    },
-    {
-      "epoch": 2.753521799797389,
-      "grad_norm": 0.0053797210566699505,
-      "learning_rate": 0.00019999626058616026,
-      "loss": 46.0,
-      "step": 36014
-    },
-    {
-      "epoch": 2.7535982567807786,
-      "grad_norm": 0.0013666145969182253,
-      "learning_rate": 0.00019999626037843653,
-      "loss": 46.0,
-      "step": 36015
-    },
-    {
-      "epoch": 2.7536747137641684,
-      "grad_norm": 0.004994364455342293,
-      "learning_rate": 0.00019999626017070705,
-      "loss": 46.0,
-      "step": 36016
-    },
-    {
-      "epoch": 2.753751170747558,
-      "grad_norm": 0.0019044834189116955,
-      "learning_rate": 0.00019999625996297178,
-      "loss": 46.0,
-      "step": 36017
-    },
-    {
-      "epoch": 2.753827627730948,
-      "grad_norm": 0.0020784682128578424,
-      "learning_rate": 0.00019999625975523075,
-      "loss": 46.0,
-      "step": 36018
-    },
-    {
-      "epoch": 2.7539040847143377,
-      "grad_norm": 0.0018723491812124848,
-      "learning_rate": 0.00019999625954748393,
-      "loss": 46.0,
-      "step": 36019
-    },
-    {
-      "epoch": 2.7539805416977274,
-      "grad_norm": 0.0008518744725733995,
-      "learning_rate": 0.00019999625933973139,
-      "loss": 46.0,
-      "step": 36020
-    },
-    {
-      "epoch": 2.754056998681117,
-      "grad_norm": 0.0081937275826931,
-      "learning_rate": 0.00019999625913197304,
-      "loss": 46.0,
-      "step": 36021
-    },
-    {
-      "epoch": 2.7541334556645065,
-      "grad_norm": 0.0035514591727405787,
-      "learning_rate": 0.00019999625892420895,
-      "loss": 46.0,
-      "step": 36022
-    },
-    {
-      "epoch": 2.7542099126478963,
-      "grad_norm": 0.002853967482224107,
-      "learning_rate": 0.00019999625871643906,
-      "loss": 46.0,
-      "step": 36023
-    },
-    {
-      "epoch": 2.754286369631286,
-      "grad_norm": 0.002367012668401003,
-      "learning_rate": 0.00019999625850866342,
-      "loss": 46.0,
-      "step": 36024
-    },
-    {
-      "epoch": 2.754362826614676,
-      "grad_norm": 0.0019829736556857824,
-      "learning_rate": 0.000199996258300882,
-      "loss": 46.0,
-      "step": 36025
-    },
-    {
-      "epoch": 2.7544392835980656,
-      "grad_norm": 0.0019415569258853793,
-      "learning_rate": 0.0001999962580930948,
-      "loss": 46.0,
-      "step": 36026
-    },
-    {
-      "epoch": 2.7545157405814553,
-      "grad_norm": 0.0017016794299706817,
-      "learning_rate": 0.00019999625788530185,
-      "loss": 46.0,
-      "step": 36027
-    },
-    {
-      "epoch": 2.754592197564845,
-      "grad_norm": 0.0019193514017388225,
-      "learning_rate": 0.00019999625767750314,
-      "loss": 46.0,
-      "step": 36028
-    },
-    {
-      "epoch": 2.754668654548235,
-      "grad_norm": 0.0009293374023400247,
-      "learning_rate": 0.00019999625746969867,
-      "loss": 46.0,
-      "step": 36029
-    },
-    {
-      "epoch": 2.7547451115316246,
-      "grad_norm": 0.0015002108411863446,
-      "learning_rate": 0.0001999962572618884,
-      "loss": 46.0,
-      "step": 36030
-    },
-    {
-      "epoch": 2.7548215685150144,
-      "grad_norm": 0.001663388917222619,
-      "learning_rate": 0.00019999625705407234,
-      "loss": 46.0,
-      "step": 36031
-    },
-    {
-      "epoch": 2.754898025498404,
-      "grad_norm": 0.001732103293761611,
-      "learning_rate": 0.00019999625684625057,
-      "loss": 46.0,
-      "step": 36032
-    },
-    {
-      "epoch": 2.754974482481794,
-      "grad_norm": 0.0009997776942327619,
-      "learning_rate": 0.000199996256638423,
-      "loss": 46.0,
-      "step": 36033
-    },
-    {
-      "epoch": 2.7550509394651836,
-      "grad_norm": 0.0010868809185922146,
-      "learning_rate": 0.00019999625643058963,
-      "loss": 46.0,
-      "step": 36034
-    },
-    {
-      "epoch": 2.755127396448573,
-      "grad_norm": 0.006627646274864674,
-      "learning_rate": 0.00019999625622275054,
-      "loss": 46.0,
-      "step": 36035
-    },
-    {
-      "epoch": 2.7552038534319627,
-      "grad_norm": 0.006199042312800884,
-      "learning_rate": 0.00019999625601490565,
-      "loss": 46.0,
-      "step": 36036
-    },
-    {
-      "epoch": 2.7552803104153525,
-      "grad_norm": 0.001618643756955862,
-      "learning_rate": 0.000199996255807055,
-      "loss": 46.0,
-      "step": 36037
-    },
-    {
-      "epoch": 2.7553567673987422,
-      "grad_norm": 0.006239763461053371,
-      "learning_rate": 0.0001999962555991986,
-      "loss": 46.0,
-      "step": 36038
-    },
-    {
-      "epoch": 2.755433224382132,
-      "grad_norm": 0.00206790491938591,
-      "learning_rate": 0.00019999625539133642,
-      "loss": 46.0,
-      "step": 36039
-    },
-    {
-      "epoch": 2.7555096813655218,
-      "grad_norm": 0.002926650457084179,
-      "learning_rate": 0.00019999625518346846,
-      "loss": 46.0,
-      "step": 36040
-    },
-    {
-      "epoch": 2.7555861383489115,
-      "grad_norm": 0.0022546935360878706,
-      "learning_rate": 0.00019999625497559473,
-      "loss": 46.0,
-      "step": 36041
-    },
-    {
-      "epoch": 2.7556625953323013,
-      "grad_norm": 0.0048040603287518024,
-      "learning_rate": 0.00019999625476771525,
-      "loss": 46.0,
-      "step": 36042
-    },
-    {
-      "epoch": 2.7557390523156906,
-      "grad_norm": 0.011813667602837086,
-      "learning_rate": 0.00019999625455982998,
-      "loss": 46.0,
-      "step": 36043
-    },
-    {
-      "epoch": 2.7558155092990804,
-      "grad_norm": 0.0035084239207208157,
-      "learning_rate": 0.00019999625435193895,
-      "loss": 46.0,
-      "step": 36044
-    },
-    {
-      "epoch": 2.75589196628247,
-      "grad_norm": 0.0012613576836884022,
-      "learning_rate": 0.00019999625414404216,
-      "loss": 46.0,
-      "step": 36045
-    },
-    {
-      "epoch": 2.75596842326586,
-      "grad_norm": 0.0015386604936793447,
-      "learning_rate": 0.00019999625393613956,
-      "loss": 46.0,
-      "step": 36046
-    },
-    {
-      "epoch": 2.7560448802492497,
-      "grad_norm": 0.0007678747642785311,
-      "learning_rate": 0.00019999625372823124,
-      "loss": 46.0,
-      "step": 36047
-    },
-    {
-      "epoch": 2.7561213372326394,
-      "grad_norm": 0.003417386207729578,
-      "learning_rate": 0.00019999625352031713,
-      "loss": 46.0,
-      "step": 36048
-    },
-    {
-      "epoch": 2.756197794216029,
-      "grad_norm": 0.0015921785961836576,
-      "learning_rate": 0.00019999625331239726,
-      "loss": 46.0,
-      "step": 36049
-    },
-    {
-      "epoch": 2.756274251199419,
-      "grad_norm": 0.006827584933489561,
-      "learning_rate": 0.0001999962531044716,
-      "loss": 46.0,
-      "step": 36050
-    },
-    {
-      "epoch": 2.7563507081828087,
-      "grad_norm": 0.002582926768809557,
-      "learning_rate": 0.0001999962528965402,
-      "loss": 46.0,
-      "step": 36051
-    },
-    {
-      "epoch": 2.7564271651661985,
-      "grad_norm": 0.0012997756712138653,
-      "learning_rate": 0.000199996252688603,
-      "loss": 46.0,
-      "step": 36052
-    },
-    {
-      "epoch": 2.756503622149588,
-      "grad_norm": 0.0009591053240001202,
-      "learning_rate": 0.00019999625248066005,
-      "loss": 46.0,
-      "step": 36053
-    },
-    {
-      "epoch": 2.756580079132978,
-      "grad_norm": 0.002003337722271681,
-      "learning_rate": 0.00019999625227271132,
-      "loss": 46.0,
-      "step": 36054
-    },
-    {
-      "epoch": 2.7566565361163677,
-      "grad_norm": 0.0012922347523272038,
-      "learning_rate": 0.00019999625206475682,
-      "loss": 46.0,
-      "step": 36055
-    },
-    {
-      "epoch": 2.7567329930997575,
-      "grad_norm": 0.0029477658681571484,
-      "learning_rate": 0.00019999625185679657,
-      "loss": 46.0,
-      "step": 36056
-    },
-    {
-      "epoch": 2.756809450083147,
-      "grad_norm": 0.0017977464012801647,
-      "learning_rate": 0.00019999625164883055,
-      "loss": 46.0,
-      "step": 36057
-    },
-    {
-      "epoch": 2.7568859070665366,
-      "grad_norm": 0.004344752058386803,
-      "learning_rate": 0.00019999625144085875,
-      "loss": 46.0,
-      "step": 36058
-    },
-    {
-      "epoch": 2.7569623640499263,
-      "grad_norm": 0.0014876354252919555,
-      "learning_rate": 0.00019999625123288118,
-      "loss": 46.0,
-      "step": 36059
-    },
-    {
-      "epoch": 2.757038821033316,
-      "grad_norm": 0.004128091968595982,
-      "learning_rate": 0.00019999625102489784,
-      "loss": 46.0,
-      "step": 36060
-    },
-    {
-      "epoch": 2.757115278016706,
-      "grad_norm": 0.0014715323923155665,
-      "learning_rate": 0.00019999625081690875,
-      "loss": 46.0,
-      "step": 36061
-    },
-    {
-      "epoch": 2.7571917350000956,
-      "grad_norm": 0.0011906761210411787,
-      "learning_rate": 0.00019999625060891386,
-      "loss": 46.0,
-      "step": 36062
-    },
-    {
-      "epoch": 2.7572681919834854,
-      "grad_norm": 0.0031480940524488688,
-      "learning_rate": 0.0001999962504009132,
-      "loss": 46.0,
-      "step": 36063
-    },
-    {
-      "epoch": 2.757344648966875,
-      "grad_norm": 0.0014148110058158636,
-      "learning_rate": 0.00019999625019290678,
-      "loss": 46.0,
-      "step": 36064
-    },
-    {
-      "epoch": 2.7574211059502645,
-      "grad_norm": 0.0014582074945792556,
-      "learning_rate": 0.0001999962499848946,
-      "loss": 46.0,
-      "step": 36065
-    },
-    {
-      "epoch": 2.7574975629336542,
-      "grad_norm": 0.0007444196962751448,
-      "learning_rate": 0.00019999624977687662,
-      "loss": 46.0,
-      "step": 36066
-    },
-    {
-      "epoch": 2.757574019917044,
-      "grad_norm": 0.0017299795290455222,
-      "learning_rate": 0.00019999624956885292,
-      "loss": 46.0,
-      "step": 36067
-    },
-    {
-      "epoch": 2.7576504769004337,
-      "grad_norm": 0.0023293776903301477,
-      "learning_rate": 0.00019999624936082344,
-      "loss": 46.0,
-      "step": 36068
-    },
-    {
-      "epoch": 2.7577269338838235,
-      "grad_norm": 0.0019161797827109694,
-      "learning_rate": 0.00019999624915278816,
-      "loss": 46.0,
-      "step": 36069
-    },
-    {
-      "epoch": 2.7578033908672133,
-      "grad_norm": 0.002756755333393812,
-      "learning_rate": 0.00019999624894474714,
-      "loss": 46.0,
-      "step": 36070
-    },
-    {
-      "epoch": 2.757879847850603,
-      "grad_norm": 0.0016580346273258328,
-      "learning_rate": 0.00019999624873670032,
-      "loss": 46.0,
-      "step": 36071
-    },
-    {
-      "epoch": 2.757956304833993,
-      "grad_norm": 0.0033509107306599617,
-      "learning_rate": 0.00019999624852864778,
-      "loss": 46.0,
-      "step": 36072
-    },
-    {
-      "epoch": 2.7580327618173825,
-      "grad_norm": 0.0018110659439116716,
-      "learning_rate": 0.00019999624832058943,
-      "loss": 46.0,
-      "step": 36073
-    },
-    {
-      "epoch": 2.7581092188007723,
-      "grad_norm": 0.00214362726546824,
-      "learning_rate": 0.00019999624811252532,
-      "loss": 46.0,
-      "step": 36074
-    },
-    {
-      "epoch": 2.758185675784162,
-      "grad_norm": 0.0007678482215851545,
-      "learning_rate": 0.00019999624790445543,
-      "loss": 46.0,
-      "step": 36075
-    },
-    {
-      "epoch": 2.758262132767552,
-      "grad_norm": 0.00136837107129395,
-      "learning_rate": 0.0001999962476963798,
-      "loss": 46.0,
-      "step": 36076
-    },
-    {
-      "epoch": 2.7583385897509416,
-      "grad_norm": 0.0011882303515449166,
-      "learning_rate": 0.00019999624748829838,
-      "loss": 46.0,
-      "step": 36077
-    },
-    {
-      "epoch": 2.7584150467343314,
-      "grad_norm": 0.000984542421065271,
-      "learning_rate": 0.00019999624728021117,
-      "loss": 46.0,
-      "step": 36078
-    },
-    {
-      "epoch": 2.7584915037177207,
-      "grad_norm": 0.0015301520470529795,
-      "learning_rate": 0.00019999624707211825,
-      "loss": 46.0,
-      "step": 36079
-    },
-    {
-      "epoch": 2.7585679607011104,
-      "grad_norm": 0.01070896815508604,
-      "learning_rate": 0.00019999624686401952,
-      "loss": 46.0,
-      "step": 36080
-    },
-    {
-      "epoch": 2.7586444176845,
-      "grad_norm": 0.0012187910033389926,
-      "learning_rate": 0.000199996246655915,
-      "loss": 46.0,
-      "step": 36081
-    },
-    {
-      "epoch": 2.75872087466789,
-      "grad_norm": 0.0012138438178226352,
-      "learning_rate": 0.00019999624644780476,
-      "loss": 46.0,
-      "step": 36082
-    },
-    {
-      "epoch": 2.7587973316512797,
-      "grad_norm": 0.004818329121917486,
-      "learning_rate": 0.00019999624623968874,
-      "loss": 46.0,
-      "step": 36083
-    },
-    {
-      "epoch": 2.7588737886346695,
-      "grad_norm": 0.005547572858631611,
-      "learning_rate": 0.00019999624603156692,
-      "loss": 46.0,
-      "step": 36084
-    },
-    {
-      "epoch": 2.7589502456180592,
-      "grad_norm": 0.003371372353285551,
-      "learning_rate": 0.00019999624582343938,
-      "loss": 46.0,
-      "step": 36085
-    },
-    {
-      "epoch": 2.759026702601449,
-      "grad_norm": 0.0012276427587494254,
-      "learning_rate": 0.00019999624561530604,
-      "loss": 46.0,
-      "step": 36086
-    },
-    {
-      "epoch": 2.7591031595848383,
-      "grad_norm": 0.0015174514846876264,
-      "learning_rate": 0.00019999624540716692,
-      "loss": 46.0,
-      "step": 36087
-    },
-    {
-      "epoch": 2.759179616568228,
-      "grad_norm": 0.0019222641130909324,
-      "learning_rate": 0.00019999624519902206,
-      "loss": 46.0,
-      "step": 36088
-    },
-    {
-      "epoch": 2.759256073551618,
-      "grad_norm": 0.001449376461096108,
-      "learning_rate": 0.0001999962449908714,
-      "loss": 46.0,
-      "step": 36089
-    },
-    {
-      "epoch": 2.7593325305350076,
-      "grad_norm": 0.003077549859881401,
-      "learning_rate": 0.00019999624478271496,
-      "loss": 46.0,
-      "step": 36090
-    },
-    {
-      "epoch": 2.7594089875183974,
-      "grad_norm": 0.0014875874621793628,
-      "learning_rate": 0.0001999962445745528,
-      "loss": 46.0,
-      "step": 36091
-    },
-    {
-      "epoch": 2.759485444501787,
-      "grad_norm": 0.002305681584402919,
-      "learning_rate": 0.00019999624436638485,
-      "loss": 46.0,
-      "step": 36092
-    },
-    {
-      "epoch": 2.759561901485177,
-      "grad_norm": 0.00429685739800334,
-      "learning_rate": 0.0001999962441582111,
-      "loss": 46.0,
-      "step": 36093
-    },
-    {
-      "epoch": 2.7596383584685666,
-      "grad_norm": 0.002804601565003395,
-      "learning_rate": 0.00019999624395003162,
-      "loss": 46.0,
-      "step": 36094
-    },
-    {
-      "epoch": 2.7597148154519564,
-      "grad_norm": 0.007309092208743095,
-      "learning_rate": 0.00019999624374184637,
-      "loss": 46.0,
-      "step": 36095
-    },
-    {
-      "epoch": 2.759791272435346,
-      "grad_norm": 0.002028807532042265,
-      "learning_rate": 0.00019999624353365532,
-      "loss": 46.0,
-      "step": 36096
-    },
-    {
-      "epoch": 2.759867729418736,
-      "grad_norm": 0.0016862526535987854,
-      "learning_rate": 0.00019999624332545852,
-      "loss": 46.0,
-      "step": 36097
-    },
-    {
-      "epoch": 2.7599441864021257,
-      "grad_norm": 0.001222234801389277,
-      "learning_rate": 0.00019999624311725596,
-      "loss": 46.0,
-      "step": 36098
-    },
-    {
-      "epoch": 2.7600206433855154,
-      "grad_norm": 0.0014830922009423375,
-      "learning_rate": 0.0001999962429090476,
-      "loss": 46.0,
-      "step": 36099
-    },
-    {
-      "epoch": 2.760097100368905,
-      "grad_norm": 0.0014992174692451954,
-      "learning_rate": 0.0001999962427008335,
-      "loss": 46.0,
-      "step": 36100
-    },
-    {
-      "epoch": 2.7601735573522945,
-      "grad_norm": 0.0015757655492052436,
-      "learning_rate": 0.0001999962424926136,
-      "loss": 46.0,
-      "step": 36101
-    },
-    {
-      "epoch": 2.7602500143356843,
-      "grad_norm": 0.0027269606944173574,
-      "learning_rate": 0.00019999624228438797,
-      "loss": 46.0,
-      "step": 36102
-    },
-    {
-      "epoch": 2.760326471319074,
-      "grad_norm": 0.0009732483886182308,
-      "learning_rate": 0.00019999624207615657,
-      "loss": 46.0,
-      "step": 36103
-    },
-    {
-      "epoch": 2.760402928302464,
-      "grad_norm": 0.00180434447247535,
-      "learning_rate": 0.00019999624186791936,
-      "loss": 46.0,
-      "step": 36104
-    },
-    {
-      "epoch": 2.7604793852858536,
-      "grad_norm": 0.0014392690500244498,
-      "learning_rate": 0.00019999624165967643,
-      "loss": 46.0,
-      "step": 36105
-    },
-    {
-      "epoch": 2.7605558422692433,
-      "grad_norm": 0.006552328821271658,
-      "learning_rate": 0.0001999962414514277,
-      "loss": 46.0,
-      "step": 36106
-    },
-    {
-      "epoch": 2.760632299252633,
-      "grad_norm": 0.005886552855372429,
-      "learning_rate": 0.0001999962412431732,
-      "loss": 46.0,
-      "step": 36107
-    },
-    {
-      "epoch": 2.760708756236023,
-      "grad_norm": 0.0030877916142344475,
-      "learning_rate": 0.00019999624103491295,
-      "loss": 46.0,
-      "step": 36108
-    },
-    {
-      "epoch": 2.760785213219412,
-      "grad_norm": 0.002319512888789177,
-      "learning_rate": 0.00019999624082664693,
-      "loss": 46.0,
-      "step": 36109
-    },
-    {
-      "epoch": 2.760861670202802,
-      "grad_norm": 0.0023680205922573805,
-      "learning_rate": 0.0001999962406183751,
-      "loss": 46.0,
-      "step": 36110
-    },
-    {
-      "epoch": 2.7609381271861917,
-      "grad_norm": 0.006912963464856148,
-      "learning_rate": 0.00019999624041009754,
-      "loss": 46.0,
-      "step": 36111
-    },
-    {
-      "epoch": 2.7610145841695815,
-      "grad_norm": 0.0069908746518194675,
-      "learning_rate": 0.0001999962402018142,
-      "loss": 46.0,
-      "step": 36112
-    },
-    {
-      "epoch": 2.761091041152971,
-      "grad_norm": 0.0026829326525330544,
-      "learning_rate": 0.00019999623999352508,
-      "loss": 46.0,
-      "step": 36113
-    },
-    {
-      "epoch": 2.761167498136361,
-      "grad_norm": 0.004067771602421999,
-      "learning_rate": 0.00019999623978523022,
-      "loss": 46.0,
-      "step": 36114
-    },
-    {
-      "epoch": 2.7612439551197507,
-      "grad_norm": 0.0021445206366479397,
-      "learning_rate": 0.00019999623957692956,
-      "loss": 46.0,
-      "step": 36115
-    },
-    {
-      "epoch": 2.7613204121031405,
-      "grad_norm": 0.001799249555915594,
-      "learning_rate": 0.00019999623936862318,
-      "loss": 46.0,
-      "step": 36116
-    },
-    {
-      "epoch": 2.7613968690865303,
-      "grad_norm": 0.0019214011263102293,
-      "learning_rate": 0.00019999623916031097,
-      "loss": 46.0,
-      "step": 36117
-    },
-    {
-      "epoch": 2.76147332606992,
-      "grad_norm": 0.0016728892223909497,
-      "learning_rate": 0.000199996238951993,
-      "loss": 46.0,
-      "step": 36118
-    },
-    {
-      "epoch": 2.76154978305331,
-      "grad_norm": 0.0009442161535844207,
-      "learning_rate": 0.0001999962387436693,
-      "loss": 46.0,
-      "step": 36119
-    },
-    {
-      "epoch": 2.7616262400366995,
-      "grad_norm": 0.004111644346266985,
-      "learning_rate": 0.0001999962385353398,
-      "loss": 46.0,
-      "step": 36120
-    },
-    {
-      "epoch": 2.7617026970200893,
-      "grad_norm": 0.0012342092813923955,
-      "learning_rate": 0.00019999623832700454,
-      "loss": 46.0,
-      "step": 36121
-    },
-    {
-      "epoch": 2.761779154003479,
-      "grad_norm": 0.006362533662468195,
-      "learning_rate": 0.00019999623811866351,
-      "loss": 46.0,
-      "step": 36122
-    },
-    {
-      "epoch": 2.7618556109868684,
-      "grad_norm": 0.001338394358754158,
-      "learning_rate": 0.00019999623791031672,
-      "loss": 46.0,
-      "step": 36123
-    },
-    {
-      "epoch": 2.761932067970258,
-      "grad_norm": 0.004385354463011026,
-      "learning_rate": 0.00019999623770196415,
-      "loss": 46.0,
-      "step": 36124
-    },
-    {
-      "epoch": 2.762008524953648,
-      "grad_norm": 0.0026679078582674265,
-      "learning_rate": 0.00019999623749360578,
-      "loss": 46.0,
-      "step": 36125
-    },
-    {
-      "epoch": 2.7620849819370377,
-      "grad_norm": 0.0011312422575429082,
-      "learning_rate": 0.0001999962372852417,
-      "loss": 46.0,
-      "step": 36126
-    },
-    {
-      "epoch": 2.7621614389204274,
-      "grad_norm": 0.0019035037839785218,
-      "learning_rate": 0.0001999962370768718,
-      "loss": 46.0,
-      "step": 36127
-    },
-    {
-      "epoch": 2.762237895903817,
-      "grad_norm": 0.001605572528205812,
-      "learning_rate": 0.00019999623686849617,
-      "loss": 46.0,
-      "step": 36128
-    },
-    {
-      "epoch": 2.762314352887207,
-      "grad_norm": 0.0028997021727263927,
-      "learning_rate": 0.00019999623666011474,
-      "loss": 46.0,
-      "step": 36129
-    },
-    {
-      "epoch": 2.7623908098705967,
-      "grad_norm": 0.004105194937437773,
-      "learning_rate": 0.00019999623645172756,
-      "loss": 46.0,
-      "step": 36130
-    },
-    {
-      "epoch": 2.762467266853986,
-      "grad_norm": 0.008957820944488049,
-      "learning_rate": 0.0001999962362433346,
-      "loss": 46.0,
-      "step": 36131
-    },
-    {
-      "epoch": 2.762543723837376,
-      "grad_norm": 0.004277107771486044,
-      "learning_rate": 0.0001999962360349359,
-      "loss": 46.0,
-      "step": 36132
-    },
-    {
-      "epoch": 2.7626201808207655,
-      "grad_norm": 0.005884347017854452,
-      "learning_rate": 0.0001999962358265314,
-      "loss": 46.0,
-      "step": 36133
-    },
-    {
-      "epoch": 2.7626966378041553,
-      "grad_norm": 0.003550578374415636,
-      "learning_rate": 0.00019999623561812112,
-      "loss": 46.0,
-      "step": 36134
-    },
-    {
-      "epoch": 2.762773094787545,
-      "grad_norm": 0.002957143122330308,
-      "learning_rate": 0.0001999962354097051,
-      "loss": 46.0,
-      "step": 36135
-    },
-    {
-      "epoch": 2.762849551770935,
-      "grad_norm": 0.0036615384742617607,
-      "learning_rate": 0.0001999962352012833,
-      "loss": 46.0,
-      "step": 36136
-    },
-    {
-      "epoch": 2.7629260087543246,
-      "grad_norm": 0.003823390230536461,
-      "learning_rate": 0.00019999623499285574,
-      "loss": 46.0,
-      "step": 36137
-    },
-    {
-      "epoch": 2.7630024657377144,
-      "grad_norm": 0.0018904708558693528,
-      "learning_rate": 0.00019999623478442237,
-      "loss": 46.0,
-      "step": 36138
-    },
-    {
-      "epoch": 2.763078922721104,
-      "grad_norm": 0.003042828291654587,
-      "learning_rate": 0.00019999623457598329,
-      "loss": 46.0,
-      "step": 36139
-    },
-    {
-      "epoch": 2.763155379704494,
-      "grad_norm": 0.0005860620294697583,
-      "learning_rate": 0.00019999623436753837,
-      "loss": 46.0,
-      "step": 36140
-    },
-    {
-      "epoch": 2.7632318366878836,
-      "grad_norm": 0.00408286415040493,
-      "learning_rate": 0.00019999623415908774,
-      "loss": 46.0,
-      "step": 36141
-    },
-    {
-      "epoch": 2.7633082936712734,
-      "grad_norm": 0.001012353110127151,
-      "learning_rate": 0.00019999623395063133,
-      "loss": 46.0,
-      "step": 36142
-    },
-    {
-      "epoch": 2.763384750654663,
-      "grad_norm": 0.004084533080458641,
-      "learning_rate": 0.00019999623374216915,
-      "loss": 46.0,
-      "step": 36143
-    },
-    {
-      "epoch": 2.7634612076380525,
-      "grad_norm": 0.0017718251328915358,
-      "learning_rate": 0.0001999962335337012,
-      "loss": 46.0,
-      "step": 36144
-    },
-    {
-      "epoch": 2.7635376646214422,
-      "grad_norm": 0.0029293482657521963,
-      "learning_rate": 0.00019999623332522747,
-      "loss": 46.0,
-      "step": 36145
-    },
-    {
-      "epoch": 2.763614121604832,
-      "grad_norm": 0.00159519596491009,
-      "learning_rate": 0.00019999623311674797,
-      "loss": 46.0,
-      "step": 36146
-    },
-    {
-      "epoch": 2.7636905785882218,
-      "grad_norm": 0.001751039526425302,
-      "learning_rate": 0.0001999962329082627,
-      "loss": 46.0,
-      "step": 36147
-    },
-    {
-      "epoch": 2.7637670355716115,
-      "grad_norm": 0.003855740884318948,
-      "learning_rate": 0.00019999623269977167,
-      "loss": 46.0,
-      "step": 36148
-    },
-    {
-      "epoch": 2.7638434925550013,
-      "grad_norm": 0.001822623424232006,
-      "learning_rate": 0.00019999623249127488,
-      "loss": 46.0,
-      "step": 36149
-    },
-    {
-      "epoch": 2.763919949538391,
-      "grad_norm": 0.002561405999585986,
-      "learning_rate": 0.0001999962322827723,
-      "loss": 46.0,
-      "step": 36150
-    },
-    {
-      "epoch": 2.763996406521781,
-      "grad_norm": 0.001635123509913683,
-      "learning_rate": 0.00019999623207426397,
-      "loss": 46.0,
-      "step": 36151
-    },
-    {
-      "epoch": 2.7640728635051706,
-      "grad_norm": 0.0016717080725356936,
-      "learning_rate": 0.00019999623186574988,
-      "loss": 46.0,
-      "step": 36152
-    },
-    {
-      "epoch": 2.76414932048856,
-      "grad_norm": 0.0024649505503475666,
-      "learning_rate": 0.00019999623165723,
-      "loss": 46.0,
-      "step": 36153
-    },
-    {
-      "epoch": 2.7642257774719496,
-      "grad_norm": 0.0011624830076470971,
-      "learning_rate": 0.00019999623144870433,
-      "loss": 46.0,
-      "step": 36154
-    },
-    {
-      "epoch": 2.7643022344553394,
-      "grad_norm": 0.003462236374616623,
-      "learning_rate": 0.00019999623124017293,
-      "loss": 46.0,
-      "step": 36155
-    },
-    {
-      "epoch": 2.764378691438729,
-      "grad_norm": 0.0019129704451188445,
-      "learning_rate": 0.00019999623103163574,
-      "loss": 46.0,
-      "step": 36156
-    },
-    {
-      "epoch": 2.764455148422119,
-      "grad_norm": 0.0025555144529789686,
-      "learning_rate": 0.0001999962308230928,
-      "loss": 46.0,
-      "step": 36157
-    },
-    {
-      "epoch": 2.7645316054055087,
-      "grad_norm": 0.006339132320135832,
-      "learning_rate": 0.00019999623061454406,
-      "loss": 46.0,
-      "step": 36158
-    },
-    {
-      "epoch": 2.7646080623888984,
-      "grad_norm": 0.0020024816039949656,
-      "learning_rate": 0.00019999623040598956,
-      "loss": 46.0,
-      "step": 36159
-    },
-    {
-      "epoch": 2.764684519372288,
-      "grad_norm": 0.00119693367742002,
-      "learning_rate": 0.0001999962301974293,
-      "loss": 46.0,
-      "step": 36160
-    },
-    {
-      "epoch": 2.764760976355678,
-      "grad_norm": 0.00252747954800725,
-      "learning_rate": 0.00019999622998886327,
-      "loss": 46.0,
-      "step": 36161
-    },
-    {
-      "epoch": 2.7648374333390677,
-      "grad_norm": 0.001226059626787901,
-      "learning_rate": 0.00019999622978029147,
-      "loss": 46.0,
-      "step": 36162
-    },
-    {
-      "epoch": 2.7649138903224575,
-      "grad_norm": 0.008969691582024097,
-      "learning_rate": 0.00019999622957171388,
-      "loss": 46.0,
-      "step": 36163
-    },
-    {
-      "epoch": 2.7649903473058473,
-      "grad_norm": 0.0032498487271368504,
-      "learning_rate": 0.00019999622936313057,
-      "loss": 46.0,
-      "step": 36164
-    },
-    {
-      "epoch": 2.765066804289237,
-      "grad_norm": 0.0022688761819154024,
-      "learning_rate": 0.00019999622915454145,
-      "loss": 46.0,
-      "step": 36165
-    },
-    {
-      "epoch": 2.7651432612726263,
-      "grad_norm": 0.0025000951718539,
-      "learning_rate": 0.0001999962289459466,
-      "loss": 46.0,
-      "step": 36166
-    },
-    {
-      "epoch": 2.765219718256016,
-      "grad_norm": 0.0007303148158825934,
-      "learning_rate": 0.00019999622873734593,
-      "loss": 46.0,
-      "step": 36167
-    },
-    {
-      "epoch": 2.765296175239406,
-      "grad_norm": 0.007277305703610182,
-      "learning_rate": 0.00019999622852873953,
-      "loss": 46.0,
-      "step": 36168
-    },
-    {
-      "epoch": 2.7653726322227956,
-      "grad_norm": 0.0010990467853844166,
-      "learning_rate": 0.00019999622832012732,
-      "loss": 46.0,
-      "step": 36169
-    },
-    {
-      "epoch": 2.7654490892061854,
-      "grad_norm": 0.0014773707371205091,
-      "learning_rate": 0.00019999622811150936,
-      "loss": 46.0,
-      "step": 36170
-    },
-    {
-      "epoch": 2.765525546189575,
-      "grad_norm": 0.003702755318954587,
-      "learning_rate": 0.00019999622790288566,
-      "loss": 46.0,
-      "step": 36171
-    },
-    {
-      "epoch": 2.765602003172965,
-      "grad_norm": 0.0007949652499519289,
-      "learning_rate": 0.00019999622769425614,
-      "loss": 46.0,
-      "step": 36172
-    },
-    {
-      "epoch": 2.7656784601563547,
-      "grad_norm": 0.001810187241062522,
-      "learning_rate": 0.0001999962274856209,
-      "loss": 46.0,
-      "step": 36173
-    },
-    {
-      "epoch": 2.7657549171397444,
-      "grad_norm": 0.0023992531932890415,
-      "learning_rate": 0.00019999622727697984,
-      "loss": 46.0,
-      "step": 36174
-    },
-    {
-      "epoch": 2.7658313741231337,
-      "grad_norm": 0.0015166165539994836,
-      "learning_rate": 0.00019999622706833305,
-      "loss": 46.0,
-      "step": 36175
-    },
-    {
-      "epoch": 2.7659078311065235,
-      "grad_norm": 0.002091736765578389,
-      "learning_rate": 0.00019999622685968048,
-      "loss": 46.0,
-      "step": 36176
-    },
-    {
-      "epoch": 2.7659842880899133,
-      "grad_norm": 0.0018164057983085513,
-      "learning_rate": 0.00019999622665102214,
-      "loss": 46.0,
-      "step": 36177
-    },
-    {
-      "epoch": 2.766060745073303,
-      "grad_norm": 0.0012643246445804834,
-      "learning_rate": 0.00019999622644235806,
-      "loss": 46.0,
-      "step": 36178
-    },
-    {
-      "epoch": 2.766137202056693,
-      "grad_norm": 0.0014303921489045024,
-      "learning_rate": 0.00019999622623368817,
-      "loss": 46.0,
-      "step": 36179
-    },
-    {
-      "epoch": 2.7662136590400825,
-      "grad_norm": 0.0027956566773355007,
-      "learning_rate": 0.0001999962260250125,
-      "loss": 46.0,
-      "step": 36180
-    },
-    {
-      "epoch": 2.7662901160234723,
-      "grad_norm": 0.0030597124714404345,
-      "learning_rate": 0.00019999622581633108,
-      "loss": 46.0,
-      "step": 36181
-    },
-    {
-      "epoch": 2.766366573006862,
-      "grad_norm": 0.0010386410867795348,
-      "learning_rate": 0.0001999962256076439,
-      "loss": 46.0,
-      "step": 36182
-    },
-    {
-      "epoch": 2.766443029990252,
-      "grad_norm": 0.005176815204322338,
-      "learning_rate": 0.00019999622539895094,
-      "loss": 46.0,
-      "step": 36183
-    },
-    {
-      "epoch": 2.7665194869736416,
-      "grad_norm": 0.002462107688188553,
-      "learning_rate": 0.00019999622519025222,
-      "loss": 46.0,
-      "step": 36184
-    },
-    {
-      "epoch": 2.7665959439570313,
-      "grad_norm": 0.005378553178161383,
-      "learning_rate": 0.00019999622498154774,
-      "loss": 46.0,
-      "step": 36185
-    },
-    {
-      "epoch": 2.766672400940421,
-      "grad_norm": 0.0006650412105955184,
-      "learning_rate": 0.00019999622477283747,
-      "loss": 46.0,
-      "step": 36186
-    },
-    {
-      "epoch": 2.766748857923811,
-      "grad_norm": 0.0008463840349577367,
-      "learning_rate": 0.00019999622456412145,
-      "loss": 46.0,
-      "step": 36187
-    },
-    {
-      "epoch": 2.7668253149072,
-      "grad_norm": 0.0014053566846996546,
-      "learning_rate": 0.00019999622435539963,
-      "loss": 46.0,
-      "step": 36188
-    },
-    {
-      "epoch": 2.76690177189059,
-      "grad_norm": 0.0012803409481421113,
-      "learning_rate": 0.00019999622414667206,
-      "loss": 46.0,
-      "step": 36189
-    },
-    {
-      "epoch": 2.7669782288739797,
-      "grad_norm": 0.0019033501157537103,
-      "learning_rate": 0.00019999622393793872,
-      "loss": 46.0,
-      "step": 36190
-    },
-    {
-      "epoch": 2.7670546858573695,
-      "grad_norm": 0.000809476594440639,
-      "learning_rate": 0.00019999622372919964,
-      "loss": 46.0,
-      "step": 36191
-    },
-    {
-      "epoch": 2.7671311428407592,
-      "grad_norm": 0.0021041552536189556,
-      "learning_rate": 0.00019999622352045475,
-      "loss": 46.0,
-      "step": 36192
-    },
-    {
-      "epoch": 2.767207599824149,
-      "grad_norm": 0.0007359951268881559,
-      "learning_rate": 0.0001999962233117041,
-      "loss": 46.0,
-      "step": 36193
-    },
-    {
-      "epoch": 2.7672840568075388,
-      "grad_norm": 0.0015482243616133928,
-      "learning_rate": 0.00019999622310294768,
-      "loss": 46.0,
-      "step": 36194
-    },
-    {
-      "epoch": 2.7673605137909285,
-      "grad_norm": 0.002026343485340476,
-      "learning_rate": 0.0001999962228941855,
-      "loss": 46.0,
-      "step": 36195
-    },
-    {
-      "epoch": 2.767436970774318,
-      "grad_norm": 0.0033258041366934776,
-      "learning_rate": 0.00019999622268541755,
-      "loss": 46.0,
-      "step": 36196
-    },
-    {
-      "epoch": 2.7675134277577076,
-      "grad_norm": 0.0007042661891318858,
-      "learning_rate": 0.0001999962224766438,
-      "loss": 46.0,
-      "step": 36197
-    },
-    {
-      "epoch": 2.7675898847410974,
-      "grad_norm": 0.0012359219836071134,
-      "learning_rate": 0.00019999622226786433,
-      "loss": 46.0,
-      "step": 36198
-    },
-    {
-      "epoch": 2.767666341724487,
-      "grad_norm": 0.001761345542035997,
-      "learning_rate": 0.00019999622205907905,
-      "loss": 46.0,
-      "step": 36199
-    },
-    {
-      "epoch": 2.767742798707877,
-      "grad_norm": 0.004379366058856249,
-      "learning_rate": 0.00019999622185028803,
-      "loss": 46.0,
-      "step": 36200
-    },
-    {
-      "epoch": 2.7678192556912666,
-      "grad_norm": 0.004129890818148851,
-      "learning_rate": 0.00019999622164149124,
-      "loss": 46.0,
-      "step": 36201
-    },
-    {
-      "epoch": 2.7678957126746564,
-      "grad_norm": 0.0027833811473101377,
-      "learning_rate": 0.00019999622143268865,
-      "loss": 46.0,
-      "step": 36202
-    },
-    {
-      "epoch": 2.767972169658046,
-      "grad_norm": 0.005369459744542837,
-      "learning_rate": 0.0001999962212238803,
-      "loss": 46.0,
-      "step": 36203
-    },
-    {
-      "epoch": 2.768048626641436,
-      "grad_norm": 0.005488489288836718,
-      "learning_rate": 0.00019999622101506622,
-      "loss": 46.0,
-      "step": 36204
-    },
-    {
-      "epoch": 2.7681250836248257,
-      "grad_norm": 0.0015503311296924949,
-      "learning_rate": 0.00019999622080624633,
-      "loss": 46.0,
-      "step": 36205
-    },
-    {
-      "epoch": 2.7682015406082154,
-      "grad_norm": 0.0011316556483507156,
-      "learning_rate": 0.00019999622059742068,
-      "loss": 46.0,
-      "step": 36206
-    },
-    {
-      "epoch": 2.768277997591605,
-      "grad_norm": 0.002888135612010956,
-      "learning_rate": 0.00019999622038858927,
-      "loss": 46.0,
-      "step": 36207
-    },
-    {
-      "epoch": 2.768354454574995,
-      "grad_norm": 0.0037156047765165567,
-      "learning_rate": 0.0001999962201797521,
-      "loss": 46.0,
-      "step": 36208
-    },
-    {
-      "epoch": 2.7684309115583847,
-      "grad_norm": 0.0015997167211025953,
-      "learning_rate": 0.0001999962199709091,
-      "loss": 46.0,
-      "step": 36209
-    },
-    {
-      "epoch": 2.768507368541774,
-      "grad_norm": 0.0008313936414197087,
-      "learning_rate": 0.0001999962197620604,
-      "loss": 46.0,
-      "step": 36210
-    },
-    {
-      "epoch": 2.768583825525164,
-      "grad_norm": 0.001955775311216712,
-      "learning_rate": 0.0001999962195532059,
-      "loss": 46.0,
-      "step": 36211
-    },
-    {
-      "epoch": 2.7686602825085536,
-      "grad_norm": 0.0029694768600165844,
-      "learning_rate": 0.00019999621934434564,
-      "loss": 46.0,
-      "step": 36212
-    },
-    {
-      "epoch": 2.7687367394919433,
-      "grad_norm": 0.0021423816215246916,
-      "learning_rate": 0.00019999621913547962,
-      "loss": 46.0,
-      "step": 36213
-    },
-    {
-      "epoch": 2.768813196475333,
-      "grad_norm": 0.0011938296956941485,
-      "learning_rate": 0.0001999962189266078,
-      "loss": 46.0,
-      "step": 36214
-    },
-    {
-      "epoch": 2.768889653458723,
-      "grad_norm": 0.0014556162059307098,
-      "learning_rate": 0.00019999621871773023,
-      "loss": 46.0,
-      "step": 36215
-    },
-    {
-      "epoch": 2.7689661104421126,
-      "grad_norm": 0.00334395794197917,
-      "learning_rate": 0.00019999621850884692,
-      "loss": 46.0,
-      "step": 36216
-    },
-    {
-      "epoch": 2.7690425674255024,
-      "grad_norm": 0.0022246367298066616,
-      "learning_rate": 0.00019999621829995778,
-      "loss": 46.0,
-      "step": 36217
-    },
-    {
-      "epoch": 2.7691190244088917,
-      "grad_norm": 0.002446898492053151,
-      "learning_rate": 0.00019999621809106292,
-      "loss": 46.0,
-      "step": 36218
-    },
-    {
-      "epoch": 2.7691954813922814,
-      "grad_norm": 0.002417887095361948,
-      "learning_rate": 0.00019999621788216226,
-      "loss": 46.0,
-      "step": 36219
-    },
-    {
-      "epoch": 2.769271938375671,
-      "grad_norm": 0.0012067490024492145,
-      "learning_rate": 0.00019999621767325586,
-      "loss": 46.0,
-      "step": 36220
-    },
-    {
-      "epoch": 2.769348395359061,
-      "grad_norm": 0.003124658018350601,
-      "learning_rate": 0.00019999621746434368,
-      "loss": 46.0,
-      "step": 36221
-    },
-    {
-      "epoch": 2.7694248523424507,
-      "grad_norm": 0.006745269522070885,
-      "learning_rate": 0.00019999621725542573,
-      "loss": 46.0,
-      "step": 36222
-    },
-    {
-      "epoch": 2.7695013093258405,
-      "grad_norm": 0.0035063677933067083,
-      "learning_rate": 0.000199996217046502,
-      "loss": 46.0,
-      "step": 36223
-    },
-    {
-      "epoch": 2.7695777663092302,
-      "grad_norm": 0.003929018508642912,
-      "learning_rate": 0.0001999962168375725,
-      "loss": 46.0,
-      "step": 36224
-    },
-    {
-      "epoch": 2.76965422329262,
-      "grad_norm": 0.0029506676364690065,
-      "learning_rate": 0.00019999621662863723,
-      "loss": 46.0,
-      "step": 36225
-    },
-    {
-      "epoch": 2.7697306802760098,
-      "grad_norm": 0.0023165347520262003,
-      "learning_rate": 0.00019999621641969618,
-      "loss": 46.0,
-      "step": 36226
-    },
-    {
-      "epoch": 2.7698071372593995,
-      "grad_norm": 0.002396000549197197,
-      "learning_rate": 0.00019999621621074936,
-      "loss": 46.0,
-      "step": 36227
-    },
-    {
-      "epoch": 2.7698835942427893,
-      "grad_norm": 0.001416377373971045,
-      "learning_rate": 0.00019999621600179683,
-      "loss": 46.0,
-      "step": 36228
-    },
-    {
-      "epoch": 2.769960051226179,
-      "grad_norm": 0.0013749051140621305,
-      "learning_rate": 0.00019999621579283846,
-      "loss": 46.0,
-      "step": 36229
-    },
-    {
-      "epoch": 2.770036508209569,
-      "grad_norm": 0.002663565333932638,
-      "learning_rate": 0.00019999621558387438,
-      "loss": 46.0,
-      "step": 36230
-    },
-    {
-      "epoch": 2.7701129651929586,
-      "grad_norm": 0.0009916143026202917,
-      "learning_rate": 0.0001999962153749045,
-      "loss": 46.0,
-      "step": 36231
-    },
-    {
-      "epoch": 2.770189422176348,
-      "grad_norm": 0.0006273167091421783,
-      "learning_rate": 0.00019999621516592883,
-      "loss": 46.0,
-      "step": 36232
-    },
-    {
-      "epoch": 2.7702658791597377,
-      "grad_norm": 0.007817982695996761,
-      "learning_rate": 0.00019999621495694743,
-      "loss": 46.0,
-      "step": 36233
-    },
-    {
-      "epoch": 2.7703423361431274,
-      "grad_norm": 0.0025522843934595585,
-      "learning_rate": 0.00019999621474796025,
-      "loss": 46.0,
-      "step": 36234
-    },
-    {
-      "epoch": 2.770418793126517,
-      "grad_norm": 0.0009370943298563361,
-      "learning_rate": 0.0001999962145389673,
-      "loss": 46.0,
-      "step": 36235
-    },
-    {
-      "epoch": 2.770495250109907,
-      "grad_norm": 0.002330845221877098,
-      "learning_rate": 0.00019999621432996857,
-      "loss": 46.0,
-      "step": 36236
-    },
-    {
-      "epoch": 2.7705717070932967,
-      "grad_norm": 0.004805863369256258,
-      "learning_rate": 0.00019999621412096407,
-      "loss": 46.0,
-      "step": 36237
-    },
-    {
-      "epoch": 2.7706481640766865,
-      "grad_norm": 0.0023467671126127243,
-      "learning_rate": 0.00019999621391195382,
-      "loss": 46.0,
-      "step": 36238
-    },
-    {
-      "epoch": 2.770724621060076,
-      "grad_norm": 0.0021957475692033768,
-      "learning_rate": 0.00019999621370293778,
-      "loss": 46.0,
-      "step": 36239
-    },
-    {
-      "epoch": 2.7708010780434655,
-      "grad_norm": 0.004642194602638483,
-      "learning_rate": 0.00019999621349391596,
-      "loss": 46.0,
-      "step": 36240
-    },
-    {
-      "epoch": 2.7708775350268553,
-      "grad_norm": 0.0004261403810232878,
-      "learning_rate": 0.0001999962132848884,
-      "loss": 46.0,
-      "step": 36241
-    },
-    {
-      "epoch": 2.770953992010245,
-      "grad_norm": 0.005633929278701544,
-      "learning_rate": 0.00019999621307585508,
-      "loss": 46.0,
-      "step": 36242
-    },
-    {
-      "epoch": 2.771030448993635,
-      "grad_norm": 0.0016725407913327217,
-      "learning_rate": 0.00019999621286681594,
-      "loss": 46.0,
-      "step": 36243
-    },
-    {
-      "epoch": 2.7711069059770246,
-      "grad_norm": 0.002157253213226795,
-      "learning_rate": 0.00019999621265777106,
-      "loss": 46.0,
-      "step": 36244
-    },
-    {
-      "epoch": 2.7711833629604143,
-      "grad_norm": 0.002128210850059986,
-      "learning_rate": 0.00019999621244872043,
-      "loss": 46.0,
-      "step": 36245
-    },
-    {
-      "epoch": 2.771259819943804,
-      "grad_norm": 0.0013361454475671053,
-      "learning_rate": 0.000199996212239664,
-      "loss": 46.0,
-      "step": 36246
-    },
-    {
-      "epoch": 2.771336276927194,
-      "grad_norm": 0.002243607770651579,
-      "learning_rate": 0.00019999621203060182,
-      "loss": 46.0,
-      "step": 36247
-    },
-    {
-      "epoch": 2.7714127339105836,
-      "grad_norm": 0.0027922396548092365,
-      "learning_rate": 0.00019999621182153387,
-      "loss": 46.0,
-      "step": 36248
-    },
-    {
-      "epoch": 2.7714891908939734,
-      "grad_norm": 0.003739304607734084,
-      "learning_rate": 0.00019999621161246017,
-      "loss": 46.0,
-      "step": 36249
-    },
-    {
-      "epoch": 2.771565647877363,
-      "grad_norm": 0.0011978268157690763,
-      "learning_rate": 0.00019999621140338064,
-      "loss": 46.0,
-      "step": 36250
-    },
-    {
-      "epoch": 2.771642104860753,
-      "grad_norm": 0.002482454990968108,
-      "learning_rate": 0.0001999962111942954,
-      "loss": 46.0,
-      "step": 36251
-    },
-    {
-      "epoch": 2.7717185618441427,
-      "grad_norm": 0.0015692224260419607,
-      "learning_rate": 0.00019999621098520435,
-      "loss": 46.0,
-      "step": 36252
-    },
-    {
-      "epoch": 2.7717950188275324,
-      "grad_norm": 0.002782933646813035,
-      "learning_rate": 0.00019999621077610756,
-      "loss": 46.0,
-      "step": 36253
-    },
-    {
-      "epoch": 2.7718714758109217,
-      "grad_norm": 0.001485670218244195,
-      "learning_rate": 0.00019999621056700497,
-      "loss": 46.0,
-      "step": 36254
-    },
-    {
-      "epoch": 2.7719479327943115,
-      "grad_norm": 0.0019724643789231777,
-      "learning_rate": 0.00019999621035789663,
-      "loss": 46.0,
-      "step": 36255
-    },
-    {
-      "epoch": 2.7720243897777013,
-      "grad_norm": 0.005401813890784979,
-      "learning_rate": 0.00019999621014878255,
-      "loss": 46.0,
-      "step": 36256
-    },
-    {
-      "epoch": 2.772100846761091,
-      "grad_norm": 0.0036992630921304226,
-      "learning_rate": 0.00019999620993966266,
-      "loss": 46.0,
-      "step": 36257
-    },
-    {
-      "epoch": 2.772177303744481,
-      "grad_norm": 0.0020485948771238327,
-      "learning_rate": 0.000199996209730537,
-      "loss": 46.0,
-      "step": 36258
-    },
-    {
-      "epoch": 2.7722537607278706,
-      "grad_norm": 0.0007092243176884949,
-      "learning_rate": 0.0001999962095214056,
-      "loss": 46.0,
-      "step": 36259
-    },
-    {
-      "epoch": 2.7723302177112603,
-      "grad_norm": 0.0016095779137685895,
-      "learning_rate": 0.0001999962093122684,
-      "loss": 46.0,
-      "step": 36260
-    },
-    {
-      "epoch": 2.77240667469465,
-      "grad_norm": 0.0028140642680227757,
-      "learning_rate": 0.00019999620910312547,
-      "loss": 46.0,
-      "step": 36261
-    },
-    {
-      "epoch": 2.7724831316780394,
-      "grad_norm": 0.001295390771701932,
-      "learning_rate": 0.00019999620889397672,
-      "loss": 46.0,
-      "step": 36262
-    },
-    {
-      "epoch": 2.772559588661429,
-      "grad_norm": 0.007932894863188267,
-      "learning_rate": 0.00019999620868482222,
-      "loss": 46.0,
-      "step": 36263
-    },
-    {
-      "epoch": 2.772636045644819,
-      "grad_norm": 0.0014076223596930504,
-      "learning_rate": 0.00019999620847566197,
-      "loss": 46.0,
-      "step": 36264
-    },
-    {
-      "epoch": 2.7727125026282087,
-      "grad_norm": 0.0055397008545696735,
-      "learning_rate": 0.00019999620826649596,
-      "loss": 46.0,
-      "step": 36265
-    },
-    {
-      "epoch": 2.7727889596115984,
-      "grad_norm": 0.003996656276285648,
-      "learning_rate": 0.00019999620805732414,
-      "loss": 46.0,
-      "step": 36266
-    },
-    {
-      "epoch": 2.772865416594988,
-      "grad_norm": 0.004375852644443512,
-      "learning_rate": 0.00019999620784814657,
-      "loss": 46.0,
-      "step": 36267
-    },
-    {
-      "epoch": 2.772941873578378,
-      "grad_norm": 0.0020823837257921696,
-      "learning_rate": 0.00019999620763896323,
-      "loss": 46.0,
-      "step": 36268
-    },
-    {
-      "epoch": 2.7730183305617677,
-      "grad_norm": 0.001259038457646966,
-      "learning_rate": 0.00019999620742977412,
-      "loss": 46.0,
-      "step": 36269
-    },
-    {
-      "epoch": 2.7730947875451575,
-      "grad_norm": 0.001441759755834937,
-      "learning_rate": 0.00019999620722057924,
-      "loss": 46.0,
-      "step": 36270
-    },
-    {
-      "epoch": 2.7731712445285472,
-      "grad_norm": 0.00216677226126194,
-      "learning_rate": 0.0001999962070113786,
-      "loss": 46.0,
-      "step": 36271
-    },
-    {
-      "epoch": 2.773247701511937,
-      "grad_norm": 0.0010187334846705198,
-      "learning_rate": 0.00019999620680217218,
-      "loss": 46.0,
-      "step": 36272
-    },
-    {
-      "epoch": 2.7733241584953268,
-      "grad_norm": 0.0012521959142759442,
-      "learning_rate": 0.00019999620659296,
-      "loss": 46.0,
-      "step": 36273
-    },
-    {
-      "epoch": 2.7734006154787165,
-      "grad_norm": 0.007458158303052187,
-      "learning_rate": 0.00019999620638374205,
-      "loss": 46.0,
-      "step": 36274
-    },
-    {
-      "epoch": 2.773477072462106,
-      "grad_norm": 0.0010355290723964572,
-      "learning_rate": 0.0001999962061745183,
-      "loss": 46.0,
-      "step": 36275
-    },
-    {
-      "epoch": 2.7735535294454956,
-      "grad_norm": 0.002079756697639823,
-      "learning_rate": 0.0001999962059652888,
-      "loss": 46.0,
-      "step": 36276
-    },
-    {
-      "epoch": 2.7736299864288854,
-      "grad_norm": 0.003104130271822214,
-      "learning_rate": 0.00019999620575605355,
-      "loss": 46.0,
-      "step": 36277
-    },
-    {
-      "epoch": 2.773706443412275,
-      "grad_norm": 0.0018078532302752137,
-      "learning_rate": 0.00019999620554681254,
-      "loss": 46.0,
-      "step": 36278
-    },
-    {
-      "epoch": 2.773782900395665,
-      "grad_norm": 0.0025641294196248055,
-      "learning_rate": 0.00019999620533756572,
-      "loss": 46.0,
-      "step": 36279
-    },
-    {
-      "epoch": 2.7738593573790546,
-      "grad_norm": 0.004777280148118734,
-      "learning_rate": 0.00019999620512831315,
-      "loss": 46.0,
-      "step": 36280
-    },
-    {
-      "epoch": 2.7739358143624444,
-      "grad_norm": 0.0016596839996054769,
-      "learning_rate": 0.00019999620491905482,
-      "loss": 46.0,
-      "step": 36281
-    },
-    {
-      "epoch": 2.774012271345834,
-      "grad_norm": 0.0017896078061312437,
-      "learning_rate": 0.0001999962047097907,
-      "loss": 46.0,
-      "step": 36282
-    },
-    {
-      "epoch": 2.774088728329224,
-      "grad_norm": 0.0012772009940817952,
-      "learning_rate": 0.00019999620450052082,
-      "loss": 46.0,
-      "step": 36283
-    },
-    {
-      "epoch": 2.7741651853126132,
-      "grad_norm": 0.002763568190857768,
-      "learning_rate": 0.00019999620429124516,
-      "loss": 46.0,
-      "step": 36284
-    },
-    {
-      "epoch": 2.774241642296003,
-      "grad_norm": 0.003044476266950369,
-      "learning_rate": 0.00019999620408196376,
-      "loss": 46.0,
-      "step": 36285
-    },
-    {
-      "epoch": 2.7743180992793928,
-      "grad_norm": 0.0017673844704404473,
-      "learning_rate": 0.00019999620387267658,
-      "loss": 46.0,
-      "step": 36286
-    },
-    {
-      "epoch": 2.7743945562627825,
-      "grad_norm": 0.0011492582270875573,
-      "learning_rate": 0.0001999962036633836,
-      "loss": 46.0,
-      "step": 36287
-    },
-    {
-      "epoch": 2.7744710132461723,
-      "grad_norm": 0.0019949048291891813,
-      "learning_rate": 0.0001999962034540849,
-      "loss": 46.0,
-      "step": 36288
-    },
-    {
-      "epoch": 2.774547470229562,
-      "grad_norm": 0.003000054508447647,
-      "learning_rate": 0.00019999620324478038,
-      "loss": 46.0,
-      "step": 36289
-    },
-    {
-      "epoch": 2.774623927212952,
-      "grad_norm": 0.0009036542032845318,
-      "learning_rate": 0.00019999620303547014,
-      "loss": 46.0,
-      "step": 36290
-    },
-    {
-      "epoch": 2.7747003841963416,
-      "grad_norm": 0.003439062973484397,
-      "learning_rate": 0.0001999962028261541,
-      "loss": 46.0,
-      "step": 36291
-    },
-    {
-      "epoch": 2.7747768411797313,
-      "grad_norm": 0.0027141759637743235,
-      "learning_rate": 0.0001999962026168323,
-      "loss": 46.0,
-      "step": 36292
-    },
-    {
-      "epoch": 2.774853298163121,
-      "grad_norm": 0.001287938212044537,
-      "learning_rate": 0.0001999962024075047,
-      "loss": 46.0,
-      "step": 36293
-    },
-    {
-      "epoch": 2.774929755146511,
-      "grad_norm": 0.003848046064376831,
-      "learning_rate": 0.0001999962021981714,
-      "loss": 46.0,
-      "step": 36294
-    },
-    {
-      "epoch": 2.7750062121299006,
-      "grad_norm": 0.0013844300992786884,
-      "learning_rate": 0.00019999620198883226,
-      "loss": 46.0,
-      "step": 36295
-    },
-    {
-      "epoch": 2.7750826691132904,
-      "grad_norm": 0.0021978113800287247,
-      "learning_rate": 0.00019999620177948738,
-      "loss": 46.0,
-      "step": 36296
-    },
-    {
-      "epoch": 2.7751591260966797,
-      "grad_norm": 0.0028412803076207638,
-      "learning_rate": 0.00019999620157013675,
-      "loss": 46.0,
-      "step": 36297
-    },
-    {
-      "epoch": 2.7752355830800695,
-      "grad_norm": 0.0024475762620568275,
-      "learning_rate": 0.00019999620136078035,
-      "loss": 46.0,
-      "step": 36298
-    },
-    {
-      "epoch": 2.775312040063459,
-      "grad_norm": 0.003025034675374627,
-      "learning_rate": 0.00019999620115141814,
-      "loss": 46.0,
-      "step": 36299
-    },
-    {
-      "epoch": 2.775388497046849,
-      "grad_norm": 0.002512165578082204,
-      "learning_rate": 0.0001999962009420502,
-      "loss": 46.0,
-      "step": 36300
-    },
-    {
-      "epoch": 2.7754649540302387,
-      "grad_norm": 0.002844077069312334,
-      "learning_rate": 0.00019999620073267647,
-      "loss": 46.0,
-      "step": 36301
-    },
-    {
-      "epoch": 2.7755414110136285,
-      "grad_norm": 0.0025078949984163046,
-      "learning_rate": 0.00019999620052329697,
-      "loss": 46.0,
-      "step": 36302
-    },
-    {
-      "epoch": 2.7756178679970183,
-      "grad_norm": 0.0013857072917744517,
-      "learning_rate": 0.0001999962003139117,
-      "loss": 46.0,
-      "step": 36303
-    },
-    {
-      "epoch": 2.775694324980408,
-      "grad_norm": 0.0011191281955689192,
-      "learning_rate": 0.00019999620010452068,
-      "loss": 46.0,
-      "step": 36304
-    },
-    {
-      "epoch": 2.775770781963798,
-      "grad_norm": 0.002449200488626957,
-      "learning_rate": 0.00019999619989512386,
-      "loss": 46.0,
-      "step": 36305
-    },
-    {
-      "epoch": 2.775847238947187,
-      "grad_norm": 0.006454708520323038,
-      "learning_rate": 0.0001999961996857213,
-      "loss": 46.0,
-      "step": 36306
-    },
-    {
-      "epoch": 2.775923695930577,
-      "grad_norm": 0.0035495825577527285,
-      "learning_rate": 0.00019999619947631296,
-      "loss": 46.0,
-      "step": 36307
-    },
-    {
-      "epoch": 2.7760001529139666,
-      "grad_norm": 0.003855351125821471,
-      "learning_rate": 0.00019999619926689885,
-      "loss": 46.0,
-      "step": 36308
-    },
-    {
-      "epoch": 2.7760766098973564,
-      "grad_norm": 0.006759051699191332,
-      "learning_rate": 0.00019999619905747897,
-      "loss": 46.0,
-      "step": 36309
-    },
-    {
-      "epoch": 2.776153066880746,
-      "grad_norm": 0.0027375759091228247,
-      "learning_rate": 0.00019999619884805334,
-      "loss": 46.0,
-      "step": 36310
-    },
-    {
-      "epoch": 2.776229523864136,
-      "grad_norm": 0.0016428364906460047,
-      "learning_rate": 0.0001999961986386219,
-      "loss": 46.0,
-      "step": 36311
-    },
-    {
-      "epoch": 2.7763059808475257,
-      "grad_norm": 0.003984153736382723,
-      "learning_rate": 0.0001999961984291847,
-      "loss": 46.0,
-      "step": 36312
-    },
-    {
-      "epoch": 2.7763824378309154,
-      "grad_norm": 0.0033783495891839266,
-      "learning_rate": 0.00019999619821974176,
-      "loss": 46.0,
-      "step": 36313
-    },
-    {
-      "epoch": 2.776458894814305,
-      "grad_norm": 0.0026289864908903837,
-      "learning_rate": 0.00019999619801029303,
-      "loss": 46.0,
-      "step": 36314
-    },
-    {
-      "epoch": 2.776535351797695,
-      "grad_norm": 0.002158515388146043,
-      "learning_rate": 0.00019999619780083853,
-      "loss": 46.0,
-      "step": 36315
-    },
-    {
-      "epoch": 2.7766118087810847,
-      "grad_norm": 0.0044413902796804905,
-      "learning_rate": 0.0001999961975913783,
-      "loss": 46.0,
-      "step": 36316
-    },
-    {
-      "epoch": 2.7766882657644745,
-      "grad_norm": 0.0013291488867253065,
-      "learning_rate": 0.00019999619738191227,
-      "loss": 46.0,
-      "step": 36317
-    },
-    {
-      "epoch": 2.7767647227478642,
-      "grad_norm": 0.0017508446471765637,
-      "learning_rate": 0.00019999619717244046,
-      "loss": 46.0,
-      "step": 36318
-    },
-    {
-      "epoch": 2.7768411797312536,
-      "grad_norm": 0.0025974586606025696,
-      "learning_rate": 0.00019999619696296287,
-      "loss": 46.0,
-      "step": 36319
-    },
-    {
-      "epoch": 2.7769176367146433,
-      "grad_norm": 0.0022220220416784286,
-      "learning_rate": 0.00019999619675347953,
-      "loss": 46.0,
-      "step": 36320
-    },
-    {
-      "epoch": 2.776994093698033,
-      "grad_norm": 0.0022402596659958363,
-      "learning_rate": 0.00019999619654399045,
-      "loss": 46.0,
-      "step": 36321
-    },
-    {
-      "epoch": 2.777070550681423,
-      "grad_norm": 0.00602367939427495,
-      "learning_rate": 0.00019999619633449554,
-      "loss": 46.0,
-      "step": 36322
-    },
-    {
-      "epoch": 2.7771470076648126,
-      "grad_norm": 0.0013270715717226267,
-      "learning_rate": 0.0001999961961249949,
-      "loss": 46.0,
-      "step": 36323
-    },
-    {
-      "epoch": 2.7772234646482024,
-      "grad_norm": 0.0033610500395298004,
-      "learning_rate": 0.00019999619591548848,
-      "loss": 46.0,
-      "step": 36324
-    },
-    {
-      "epoch": 2.777299921631592,
-      "grad_norm": 0.0031422311440110207,
-      "learning_rate": 0.0001999961957059763,
-      "loss": 46.0,
-      "step": 36325
-    },
-    {
-      "epoch": 2.777376378614982,
-      "grad_norm": 0.002408050699159503,
-      "learning_rate": 0.00019999619549645835,
-      "loss": 46.0,
-      "step": 36326
-    },
-    {
-      "epoch": 2.777452835598371,
-      "grad_norm": 0.0022337448317557573,
-      "learning_rate": 0.0001999961952869346,
-      "loss": 46.0,
-      "step": 36327
-    },
-    {
-      "epoch": 2.777529292581761,
-      "grad_norm": 0.0023750371765345335,
-      "learning_rate": 0.00019999619507740513,
-      "loss": 46.0,
-      "step": 36328
-    },
-    {
-      "epoch": 2.7776057495651507,
-      "grad_norm": 0.0017480048118159175,
-      "learning_rate": 0.00019999619486786986,
-      "loss": 46.0,
-      "step": 36329
-    },
-    {
-      "epoch": 2.7776822065485405,
-      "grad_norm": 0.00305254478007555,
-      "learning_rate": 0.00019999619465832882,
-      "loss": 46.0,
-      "step": 36330
-    },
-    {
-      "epoch": 2.7777586635319302,
-      "grad_norm": 0.0008896057843230665,
-      "learning_rate": 0.00019999619444878203,
-      "loss": 46.0,
-      "step": 36331
-    },
-    {
-      "epoch": 2.77783512051532,
-      "grad_norm": 0.0029970339965075254,
-      "learning_rate": 0.00019999619423922944,
-      "loss": 46.0,
-      "step": 36332
-    },
-    {
-      "epoch": 2.7779115774987098,
-      "grad_norm": 0.002943855244666338,
-      "learning_rate": 0.0001999961940296711,
-      "loss": 46.0,
-      "step": 36333
-    },
-    {
-      "epoch": 2.7779880344820995,
-      "grad_norm": 0.002170650754123926,
-      "learning_rate": 0.00019999619382010702,
-      "loss": 46.0,
-      "step": 36334
-    },
-    {
-      "epoch": 2.7780644914654893,
-      "grad_norm": 0.000411079527111724,
-      "learning_rate": 0.00019999619361053714,
-      "loss": 46.0,
-      "step": 36335
-    },
-    {
-      "epoch": 2.778140948448879,
-      "grad_norm": 0.003213879419490695,
-      "learning_rate": 0.00019999619340096148,
-      "loss": 46.0,
-      "step": 36336
-    },
-    {
-      "epoch": 2.778217405432269,
-      "grad_norm": 0.004092518240213394,
-      "learning_rate": 0.00019999619319138005,
-      "loss": 46.0,
-      "step": 36337
-    },
-    {
-      "epoch": 2.7782938624156586,
-      "grad_norm": 0.0019328441703692079,
-      "learning_rate": 0.00019999619298179287,
-      "loss": 46.0,
-      "step": 36338
-    },
-    {
-      "epoch": 2.7783703193990483,
-      "grad_norm": 0.0021066172048449516,
-      "learning_rate": 0.00019999619277219992,
-      "loss": 46.0,
-      "step": 36339
-    },
-    {
-      "epoch": 2.778446776382438,
-      "grad_norm": 0.0015207099495455623,
-      "learning_rate": 0.00019999619256260117,
-      "loss": 46.0,
-      "step": 36340
-    },
-    {
-      "epoch": 2.7785232333658274,
-      "grad_norm": 0.0011930682230740786,
-      "learning_rate": 0.0001999961923529967,
-      "loss": 46.0,
-      "step": 36341
-    },
-    {
-      "epoch": 2.778599690349217,
-      "grad_norm": 0.0009223195957019925,
-      "learning_rate": 0.00019999619214338643,
-      "loss": 46.0,
-      "step": 36342
-    },
-    {
-      "epoch": 2.778676147332607,
-      "grad_norm": 0.0011909353779628873,
-      "learning_rate": 0.00019999619193377042,
-      "loss": 46.0,
-      "step": 36343
-    },
-    {
-      "epoch": 2.7787526043159967,
-      "grad_norm": 0.0025372852105647326,
-      "learning_rate": 0.00019999619172414858,
-      "loss": 46.0,
-      "step": 36344
-    },
-    {
-      "epoch": 2.7788290612993864,
-      "grad_norm": 0.0016447481466457248,
-      "learning_rate": 0.000199996191514521,
-      "loss": 46.0,
-      "step": 36345
-    },
-    {
-      "epoch": 2.778905518282776,
-      "grad_norm": 0.0015572209376841784,
-      "learning_rate": 0.0001999961913048877,
-      "loss": 46.0,
-      "step": 36346
-    },
-    {
-      "epoch": 2.778981975266166,
-      "grad_norm": 0.0010300199501216412,
-      "learning_rate": 0.0001999961910952486,
-      "loss": 46.0,
-      "step": 36347
-    },
-    {
-      "epoch": 2.7790584322495557,
-      "grad_norm": 0.0009643419180065393,
-      "learning_rate": 0.0001999961908856037,
-      "loss": 46.0,
-      "step": 36348
-    },
-    {
-      "epoch": 2.779134889232945,
-      "grad_norm": 0.0009546820656396449,
-      "learning_rate": 0.00019999619067595305,
-      "loss": 46.0,
-      "step": 36349
-    },
-    {
-      "epoch": 2.779211346216335,
-      "grad_norm": 0.0013749372446909547,
-      "learning_rate": 0.00019999619046629663,
-      "loss": 46.0,
-      "step": 36350
-    },
-    {
-      "epoch": 2.7792878031997246,
-      "grad_norm": 0.004427598789334297,
-      "learning_rate": 0.00019999619025663445,
-      "loss": 46.0,
-      "step": 36351
-    },
-    {
-      "epoch": 2.7793642601831143,
-      "grad_norm": 0.0016899374313652515,
-      "learning_rate": 0.0001999961900469665,
-      "loss": 46.0,
-      "step": 36352
-    },
-    {
-      "epoch": 2.779440717166504,
-      "grad_norm": 0.0013406631769612432,
-      "learning_rate": 0.00019999618983729275,
-      "loss": 46.0,
-      "step": 36353
-    },
-    {
-      "epoch": 2.779517174149894,
-      "grad_norm": 0.0010044530499726534,
-      "learning_rate": 0.00019999618962761328,
-      "loss": 46.0,
-      "step": 36354
-    },
-    {
-      "epoch": 2.7795936311332836,
-      "grad_norm": 0.0012139335740357637,
-      "learning_rate": 0.000199996189417928,
-      "loss": 46.0,
-      "step": 36355
-    },
-    {
-      "epoch": 2.7796700881166734,
-      "grad_norm": 0.0035704125184565783,
-      "learning_rate": 0.00019999618920823697,
-      "loss": 46.0,
-      "step": 36356
-    },
-    {
-      "epoch": 2.779746545100063,
-      "grad_norm": 0.008283907547593117,
-      "learning_rate": 0.00019999618899854018,
-      "loss": 46.0,
-      "step": 36357
-    },
-    {
-      "epoch": 2.779823002083453,
-      "grad_norm": 0.0021186447702348232,
-      "learning_rate": 0.00019999618878883762,
-      "loss": 46.0,
-      "step": 36358
-    },
-    {
-      "epoch": 2.7798994590668427,
-      "grad_norm": 0.006470333784818649,
-      "learning_rate": 0.00019999618857912928,
-      "loss": 46.0,
-      "step": 36359
-    },
-    {
-      "epoch": 2.7799759160502324,
-      "grad_norm": 0.006633661687374115,
-      "learning_rate": 0.00019999618836941517,
-      "loss": 46.0,
-      "step": 36360
-    },
-    {
-      "epoch": 2.780052373033622,
-      "grad_norm": 0.0023193766828626394,
-      "learning_rate": 0.00019999618815969526,
-      "loss": 46.0,
-      "step": 36361
-    },
-    {
-      "epoch": 2.780128830017012,
-      "grad_norm": 0.0008143925806507468,
-      "learning_rate": 0.0001999961879499696,
-      "loss": 46.0,
-      "step": 36362
-    },
-    {
-      "epoch": 2.7802052870004013,
-      "grad_norm": 0.00410993630066514,
-      "learning_rate": 0.0001999961877402382,
-      "loss": 46.0,
-      "step": 36363
-    },
-    {
-      "epoch": 2.780281743983791,
-      "grad_norm": 0.00489655788987875,
-      "learning_rate": 0.00019999618753050103,
-      "loss": 46.0,
-      "step": 36364
-    },
-    {
-      "epoch": 2.780358200967181,
-      "grad_norm": 0.003732833778485656,
-      "learning_rate": 0.00019999618732075805,
-      "loss": 46.0,
-      "step": 36365
-    },
-    {
-      "epoch": 2.7804346579505705,
-      "grad_norm": 0.0021146514918655157,
-      "learning_rate": 0.00019999618711100933,
-      "loss": 46.0,
-      "step": 36366
-    },
-    {
-      "epoch": 2.7805111149339603,
-      "grad_norm": 0.0016313346568495035,
-      "learning_rate": 0.00019999618690125486,
-      "loss": 46.0,
-      "step": 36367
-    },
-    {
-      "epoch": 2.78058757191735,
-      "grad_norm": 0.0056966692209243774,
-      "learning_rate": 0.00019999618669149457,
-      "loss": 46.0,
-      "step": 36368
-    },
-    {
-      "epoch": 2.78066402890074,
-      "grad_norm": 0.002248887438327074,
-      "learning_rate": 0.00019999618648172855,
-      "loss": 46.0,
-      "step": 36369
-    },
-    {
-      "epoch": 2.7807404858841296,
-      "grad_norm": 0.0024897693656384945,
-      "learning_rate": 0.00019999618627195676,
-      "loss": 46.0,
-      "step": 36370
-    },
-    {
-      "epoch": 2.780816942867519,
-      "grad_norm": 0.001384552102535963,
-      "learning_rate": 0.00019999618606217917,
-      "loss": 46.0,
-      "step": 36371
-    },
-    {
-      "epoch": 2.7808933998509087,
-      "grad_norm": 0.0013797096908092499,
-      "learning_rate": 0.00019999618585239584,
-      "loss": 46.0,
-      "step": 36372
-    },
-    {
-      "epoch": 2.7809698568342984,
-      "grad_norm": 0.0019975188188254833,
-      "learning_rate": 0.00019999618564260673,
-      "loss": 46.0,
-      "step": 36373
-    },
-    {
-      "epoch": 2.781046313817688,
-      "grad_norm": 0.003750791074708104,
-      "learning_rate": 0.00019999618543281185,
-      "loss": 46.0,
-      "step": 36374
-    },
-    {
-      "epoch": 2.781122770801078,
-      "grad_norm": 0.0025771965738385916,
-      "learning_rate": 0.0001999961852230112,
-      "loss": 46.0,
-      "step": 36375
-    },
-    {
-      "epoch": 2.7811992277844677,
-      "grad_norm": 0.0018562711775302887,
-      "learning_rate": 0.0001999961850132048,
-      "loss": 46.0,
-      "step": 36376
-    },
-    {
-      "epoch": 2.7812756847678575,
-      "grad_norm": 0.0020029630977660418,
-      "learning_rate": 0.0001999961848033926,
-      "loss": 46.0,
-      "step": 36377
-    },
-    {
-      "epoch": 2.7813521417512472,
-      "grad_norm": 0.001481458661146462,
-      "learning_rate": 0.00019999618459357464,
-      "loss": 46.0,
-      "step": 36378
-    },
-    {
-      "epoch": 2.781428598734637,
-      "grad_norm": 0.0021441939752548933,
-      "learning_rate": 0.00019999618438375092,
-      "loss": 46.0,
-      "step": 36379
-    },
-    {
-      "epoch": 2.7815050557180268,
-      "grad_norm": 0.0022772380616515875,
-      "learning_rate": 0.00019999618417392142,
-      "loss": 46.0,
-      "step": 36380
-    },
-    {
-      "epoch": 2.7815815127014165,
-      "grad_norm": 0.003487536683678627,
-      "learning_rate": 0.00019999618396408618,
-      "loss": 46.0,
-      "step": 36381
-    },
-    {
-      "epoch": 2.7816579696848063,
-      "grad_norm": 0.0007609502063132823,
-      "learning_rate": 0.00019999618375424514,
-      "loss": 46.0,
-      "step": 36382
-    },
-    {
-      "epoch": 2.781734426668196,
-      "grad_norm": 0.002699241740629077,
-      "learning_rate": 0.00019999618354439832,
-      "loss": 46.0,
-      "step": 36383
-    },
-    {
-      "epoch": 2.781810883651586,
-      "grad_norm": 0.0015332651091739535,
-      "learning_rate": 0.00019999618333454576,
-      "loss": 46.0,
-      "step": 36384
-    },
-    {
-      "epoch": 2.781887340634975,
-      "grad_norm": 0.0027881485875695944,
-      "learning_rate": 0.0001999961831246874,
-      "loss": 46.0,
-      "step": 36385
-    },
-    {
-      "epoch": 2.781963797618365,
-      "grad_norm": 0.000976945273578167,
-      "learning_rate": 0.00019999618291482332,
-      "loss": 46.0,
-      "step": 36386
-    },
-    {
-      "epoch": 2.7820402546017546,
-      "grad_norm": 0.002280328655615449,
-      "learning_rate": 0.0001999961827049534,
-      "loss": 46.0,
-      "step": 36387
-    },
-    {
-      "epoch": 2.7821167115851444,
-      "grad_norm": 0.0009409145568497479,
-      "learning_rate": 0.00019999618249507778,
-      "loss": 46.0,
-      "step": 36388
-    },
-    {
-      "epoch": 2.782193168568534,
-      "grad_norm": 0.0022007119841873646,
-      "learning_rate": 0.00019999618228519635,
-      "loss": 46.0,
-      "step": 36389
-    },
-    {
-      "epoch": 2.782269625551924,
-      "grad_norm": 0.0029440466314554214,
-      "learning_rate": 0.00019999618207530918,
-      "loss": 46.0,
-      "step": 36390
-    },
-    {
-      "epoch": 2.7823460825353137,
-      "grad_norm": 0.0009036447736434639,
-      "learning_rate": 0.00019999618186541623,
-      "loss": 46.0,
-      "step": 36391
-    },
-    {
-      "epoch": 2.7824225395187034,
-      "grad_norm": 0.0019512152066454291,
-      "learning_rate": 0.0001999961816555175,
-      "loss": 46.0,
-      "step": 36392
-    },
-    {
-      "epoch": 2.7824989965020928,
-      "grad_norm": 0.0009316488867625594,
-      "learning_rate": 0.000199996181445613,
-      "loss": 46.0,
-      "step": 36393
-    },
-    {
-      "epoch": 2.7825754534854825,
-      "grad_norm": 0.0027817185036838055,
-      "learning_rate": 0.00019999618123570274,
-      "loss": 46.0,
-      "step": 36394
-    },
-    {
-      "epoch": 2.7826519104688723,
-      "grad_norm": 0.001116726896725595,
-      "learning_rate": 0.0001999961810257867,
-      "loss": 46.0,
-      "step": 36395
-    },
-    {
-      "epoch": 2.782728367452262,
-      "grad_norm": 0.003959340043365955,
-      "learning_rate": 0.00019999618081586489,
-      "loss": 46.0,
-      "step": 36396
-    },
-    {
-      "epoch": 2.782804824435652,
-      "grad_norm": 0.002118962351232767,
-      "learning_rate": 0.00019999618060593732,
-      "loss": 46.0,
-      "step": 36397
-    },
-    {
-      "epoch": 2.7828812814190416,
-      "grad_norm": 0.0013942746445536613,
-      "learning_rate": 0.000199996180396004,
-      "loss": 46.0,
-      "step": 36398
-    },
-    {
-      "epoch": 2.7829577384024313,
-      "grad_norm": 0.0030810667667537928,
-      "learning_rate": 0.00019999618018606488,
-      "loss": 46.0,
-      "step": 36399
-    },
-    {
-      "epoch": 2.783034195385821,
-      "grad_norm": 0.007912750355899334,
-      "learning_rate": 0.00019999617997612,
-      "loss": 46.0,
-      "step": 36400
-    },
-    {
-      "epoch": 2.783110652369211,
-      "grad_norm": 0.009619493968784809,
-      "learning_rate": 0.00019999617976616934,
-      "loss": 46.0,
-      "step": 36401
-    },
-    {
-      "epoch": 2.7831871093526006,
-      "grad_norm": 0.0014583384618163109,
-      "learning_rate": 0.00019999617955621294,
-      "loss": 46.0,
-      "step": 36402
-    },
-    {
-      "epoch": 2.7832635663359904,
-      "grad_norm": 0.0037126613315194845,
-      "learning_rate": 0.00019999617934625074,
-      "loss": 46.0,
-      "step": 36403
-    },
-    {
-      "epoch": 2.78334002331938,
-      "grad_norm": 0.0033482497092336416,
-      "learning_rate": 0.0001999961791362828,
-      "loss": 46.0,
-      "step": 36404
-    },
-    {
-      "epoch": 2.78341648030277,
-      "grad_norm": 0.0044174217619001865,
-      "learning_rate": 0.00019999617892630907,
-      "loss": 46.0,
-      "step": 36405
-    },
-    {
-      "epoch": 2.7834929372861597,
-      "grad_norm": 0.0028654313646256924,
-      "learning_rate": 0.00019999617871632958,
-      "loss": 46.0,
-      "step": 36406
-    },
-    {
-      "epoch": 2.783569394269549,
-      "grad_norm": 0.003379259957000613,
-      "learning_rate": 0.0001999961785063443,
-      "loss": 46.0,
-      "step": 36407
-    },
-    {
-      "epoch": 2.7836458512529387,
-      "grad_norm": 0.0035564000718295574,
-      "learning_rate": 0.00019999617829635327,
-      "loss": 46.0,
-      "step": 36408
-    },
-    {
-      "epoch": 2.7837223082363285,
-      "grad_norm": 0.0028411243110895157,
-      "learning_rate": 0.00019999617808635648,
-      "loss": 46.0,
-      "step": 36409
-    },
-    {
-      "epoch": 2.7837987652197183,
-      "grad_norm": 0.0036967976484447718,
-      "learning_rate": 0.0001999961778763539,
-      "loss": 46.0,
-      "step": 36410
-    },
-    {
-      "epoch": 2.783875222203108,
-      "grad_norm": 0.006522329058498144,
-      "learning_rate": 0.00019999617766634558,
-      "loss": 46.0,
-      "step": 36411
-    },
-    {
-      "epoch": 2.7839516791864978,
-      "grad_norm": 0.0036786424461752176,
-      "learning_rate": 0.00019999617745633142,
-      "loss": 46.0,
-      "step": 36412
-    },
-    {
-      "epoch": 2.7840281361698875,
-      "grad_norm": 0.0025744715239852667,
-      "learning_rate": 0.00019999617724631157,
-      "loss": 46.0,
-      "step": 36413
-    },
-    {
-      "epoch": 2.7841045931532773,
-      "grad_norm": 0.001146244234405458,
-      "learning_rate": 0.00019999617703628591,
-      "loss": 46.0,
-      "step": 36414
-    },
-    {
-      "epoch": 2.7841810501366666,
-      "grad_norm": 0.0016148522263392806,
-      "learning_rate": 0.0001999961768262545,
-      "loss": 46.0,
-      "step": 36415
-    },
-    {
-      "epoch": 2.7842575071200564,
-      "grad_norm": 0.005639056209474802,
-      "learning_rate": 0.0001999961766162173,
-      "loss": 46.0,
-      "step": 36416
-    },
-    {
-      "epoch": 2.784333964103446,
-      "grad_norm": 0.0010224903235211968,
-      "learning_rate": 0.00019999617640617436,
-      "loss": 46.0,
-      "step": 36417
-    },
-    {
-      "epoch": 2.784410421086836,
-      "grad_norm": 0.0015144539065659046,
-      "learning_rate": 0.00019999617619612562,
-      "loss": 46.0,
-      "step": 36418
-    },
-    {
-      "epoch": 2.7844868780702257,
-      "grad_norm": 0.0008064638241194189,
-      "learning_rate": 0.00019999617598607115,
-      "loss": 46.0,
-      "step": 36419
-    },
-    {
-      "epoch": 2.7845633350536154,
-      "grad_norm": 0.0031966788228601217,
-      "learning_rate": 0.00019999617577601088,
-      "loss": 46.0,
-      "step": 36420
-    },
-    {
-      "epoch": 2.784639792037005,
-      "grad_norm": 0.0011901513207703829,
-      "learning_rate": 0.00019999617556594484,
-      "loss": 46.0,
-      "step": 36421
-    },
-    {
-      "epoch": 2.784716249020395,
-      "grad_norm": 0.0008553629741072655,
-      "learning_rate": 0.00019999617535587305,
-      "loss": 46.0,
-      "step": 36422
-    },
-    {
-      "epoch": 2.7847927060037847,
-      "grad_norm": 0.0025794266257435083,
-      "learning_rate": 0.00019999617514579546,
-      "loss": 46.0,
-      "step": 36423
-    },
-    {
-      "epoch": 2.7848691629871745,
-      "grad_norm": 0.001204487169161439,
-      "learning_rate": 0.00019999617493571213,
-      "loss": 46.0,
-      "step": 36424
-    },
-    {
-      "epoch": 2.7849456199705642,
-      "grad_norm": 0.0024331745225936174,
-      "learning_rate": 0.00019999617472562302,
-      "loss": 46.0,
-      "step": 36425
-    },
-    {
-      "epoch": 2.785022076953954,
-      "grad_norm": 0.0017571276985108852,
-      "learning_rate": 0.00019999617451552814,
-      "loss": 46.0,
-      "step": 36426
-    },
-    {
-      "epoch": 2.7850985339373437,
-      "grad_norm": 0.0050315638072788715,
-      "learning_rate": 0.0001999961743054275,
-      "loss": 46.0,
-      "step": 36427
-    },
-    {
-      "epoch": 2.785174990920733,
-      "grad_norm": 0.0021945710759609938,
-      "learning_rate": 0.0001999961740953211,
-      "loss": 46.0,
-      "step": 36428
-    },
-    {
-      "epoch": 2.785251447904123,
-      "grad_norm": 0.0023964280262589455,
-      "learning_rate": 0.00019999617388520889,
-      "loss": 46.0,
-      "step": 36429
-    },
-    {
-      "epoch": 2.7853279048875126,
-      "grad_norm": 0.004022674169391394,
-      "learning_rate": 0.00019999617367509094,
-      "loss": 46.0,
-      "step": 36430
-    },
-    {
-      "epoch": 2.7854043618709023,
-      "grad_norm": 0.0017313695279881358,
-      "learning_rate": 0.00019999617346496722,
-      "loss": 46.0,
-      "step": 36431
-    },
-    {
-      "epoch": 2.785480818854292,
-      "grad_norm": 0.004610915202647448,
-      "learning_rate": 0.0001999961732548377,
-      "loss": 46.0,
-      "step": 36432
-    },
-    {
-      "epoch": 2.785557275837682,
-      "grad_norm": 0.0034621006343513727,
-      "learning_rate": 0.00019999617304470246,
-      "loss": 46.0,
-      "step": 36433
-    },
-    {
-      "epoch": 2.7856337328210716,
-      "grad_norm": 0.0037449817173182964,
-      "learning_rate": 0.00019999617283456142,
-      "loss": 46.0,
-      "step": 36434
-    },
-    {
-      "epoch": 2.7857101898044614,
-      "grad_norm": 0.0013526586117222905,
-      "learning_rate": 0.0001999961726244146,
-      "loss": 46.0,
-      "step": 36435
-    },
-    {
-      "epoch": 2.785786646787851,
-      "grad_norm": 0.002429828979074955,
-      "learning_rate": 0.00019999617241426204,
-      "loss": 46.0,
-      "step": 36436
-    },
-    {
-      "epoch": 2.7858631037712405,
-      "grad_norm": 0.0009661111980676651,
-      "learning_rate": 0.0001999961722041037,
-      "loss": 46.0,
-      "step": 36437
-    },
-    {
-      "epoch": 2.7859395607546302,
-      "grad_norm": 0.0012383927823975682,
-      "learning_rate": 0.0001999961719939396,
-      "loss": 46.0,
-      "step": 36438
-    },
-    {
-      "epoch": 2.78601601773802,
-      "grad_norm": 0.003857041010633111,
-      "learning_rate": 0.00019999617178376972,
-      "loss": 46.0,
-      "step": 36439
-    },
-    {
-      "epoch": 2.7860924747214098,
-      "grad_norm": 0.0017049004090949893,
-      "learning_rate": 0.00019999617157359406,
-      "loss": 46.0,
-      "step": 36440
-    },
-    {
-      "epoch": 2.7861689317047995,
-      "grad_norm": 0.0056803058832883835,
-      "learning_rate": 0.00019999617136341264,
-      "loss": 46.0,
-      "step": 36441
-    },
-    {
-      "epoch": 2.7862453886881893,
-      "grad_norm": 0.004347475245594978,
-      "learning_rate": 0.00019999617115322546,
-      "loss": 46.0,
-      "step": 36442
-    },
-    {
-      "epoch": 2.786321845671579,
-      "grad_norm": 0.0045688762329518795,
-      "learning_rate": 0.00019999617094303252,
-      "loss": 46.0,
-      "step": 36443
-    },
-    {
-      "epoch": 2.786398302654969,
-      "grad_norm": 0.0037312142085283995,
-      "learning_rate": 0.00019999617073283377,
-      "loss": 46.0,
-      "step": 36444
-    },
-    {
-      "epoch": 2.7864747596383586,
-      "grad_norm": 0.007425202056765556,
-      "learning_rate": 0.0001999961705226293,
-      "loss": 46.0,
-      "step": 36445
-    },
-    {
-      "epoch": 2.7865512166217483,
-      "grad_norm": 0.004221283830702305,
-      "learning_rate": 0.000199996170312419,
-      "loss": 46.0,
-      "step": 36446
-    },
-    {
-      "epoch": 2.786627673605138,
-      "grad_norm": 0.0018323225667700171,
-      "learning_rate": 0.00019999617010220297,
-      "loss": 46.0,
-      "step": 36447
-    },
-    {
-      "epoch": 2.786704130588528,
-      "grad_norm": 0.0020557355601340532,
-      "learning_rate": 0.00019999616989198118,
-      "loss": 46.0,
-      "step": 36448
-    },
-    {
-      "epoch": 2.7867805875719176,
-      "grad_norm": 0.006091445684432983,
-      "learning_rate": 0.00019999616968175362,
-      "loss": 46.0,
-      "step": 36449
-    },
-    {
-      "epoch": 2.786857044555307,
-      "grad_norm": 0.0021612439304590225,
-      "learning_rate": 0.00019999616947152026,
-      "loss": 46.0,
-      "step": 36450
-    },
-    {
-      "epoch": 2.7869335015386967,
-      "grad_norm": 0.002512024948373437,
-      "learning_rate": 0.00019999616926128115,
-      "loss": 46.0,
-      "step": 36451
-    },
-    {
-      "epoch": 2.7870099585220864,
-      "grad_norm": 0.002003186848014593,
-      "learning_rate": 0.0001999961690510363,
-      "loss": 46.0,
-      "step": 36452
-    },
-    {
-      "epoch": 2.787086415505476,
-      "grad_norm": 0.0011371636064723134,
-      "learning_rate": 0.00019999616884078565,
-      "loss": 46.0,
-      "step": 36453
-    },
-    {
-      "epoch": 2.787162872488866,
-      "grad_norm": 0.0038547248113900423,
-      "learning_rate": 0.00019999616863052922,
-      "loss": 46.0,
-      "step": 36454
-    },
-    {
-      "epoch": 2.7872393294722557,
-      "grad_norm": 0.0013719202252104878,
-      "learning_rate": 0.00019999616842026705,
-      "loss": 46.0,
-      "step": 36455
-    },
-    {
-      "epoch": 2.7873157864556455,
-      "grad_norm": 0.0020077896770089865,
-      "learning_rate": 0.00019999616820999907,
-      "loss": 46.0,
-      "step": 36456
-    },
-    {
-      "epoch": 2.7873922434390352,
-      "grad_norm": 0.00496314000338316,
-      "learning_rate": 0.00019999616799972535,
-      "loss": 46.0,
-      "step": 36457
-    },
-    {
-      "epoch": 2.7874687004224246,
-      "grad_norm": 0.0022020842880010605,
-      "learning_rate": 0.00019999616778944586,
-      "loss": 46.0,
-      "step": 36458
-    },
-    {
-      "epoch": 2.7875451574058143,
-      "grad_norm": 0.0027344324626028538,
-      "learning_rate": 0.0001999961675791606,
-      "loss": 46.0,
-      "step": 36459
-    },
-    {
-      "epoch": 2.787621614389204,
-      "grad_norm": 0.004561572335660458,
-      "learning_rate": 0.00019999616736886955,
-      "loss": 46.0,
-      "step": 36460
-    },
-    {
-      "epoch": 2.787698071372594,
-      "grad_norm": 0.004444970283657312,
-      "learning_rate": 0.00019999616715857274,
-      "loss": 46.0,
-      "step": 36461
-    },
-    {
-      "epoch": 2.7877745283559836,
-      "grad_norm": 0.0061172787100076675,
-      "learning_rate": 0.0001999961669482702,
-      "loss": 46.0,
-      "step": 36462
-    },
-    {
-      "epoch": 2.7878509853393734,
-      "grad_norm": 0.004543913062661886,
-      "learning_rate": 0.00019999616673796184,
-      "loss": 46.0,
-      "step": 36463
-    },
-    {
-      "epoch": 2.787927442322763,
-      "grad_norm": 0.0012532860273495317,
-      "learning_rate": 0.00019999616652764774,
-      "loss": 46.0,
-      "step": 36464
-    },
-    {
-      "epoch": 2.788003899306153,
-      "grad_norm": 0.0037034032866358757,
-      "learning_rate": 0.00019999616631732786,
-      "loss": 46.0,
-      "step": 36465
-    },
-    {
-      "epoch": 2.7880803562895426,
-      "grad_norm": 0.0006219047936610878,
-      "learning_rate": 0.00019999616610700223,
-      "loss": 46.0,
-      "step": 36466
-    },
-    {
-      "epoch": 2.7881568132729324,
-      "grad_norm": 0.0008324295631609857,
-      "learning_rate": 0.00019999616589667078,
-      "loss": 46.0,
-      "step": 36467
-    },
-    {
-      "epoch": 2.788233270256322,
-      "grad_norm": 0.0014626223128288984,
-      "learning_rate": 0.0001999961656863336,
-      "loss": 46.0,
-      "step": 36468
-    },
-    {
-      "epoch": 2.788309727239712,
-      "grad_norm": 0.003055607434362173,
-      "learning_rate": 0.00019999616547599066,
-      "loss": 46.0,
-      "step": 36469
-    },
-    {
-      "epoch": 2.7883861842231017,
-      "grad_norm": 0.0008964931475929916,
-      "learning_rate": 0.0001999961652656419,
-      "loss": 46.0,
-      "step": 36470
-    },
-    {
-      "epoch": 2.7884626412064915,
-      "grad_norm": 0.0030449212063103914,
-      "learning_rate": 0.00019999616505528742,
-      "loss": 46.0,
-      "step": 36471
-    },
-    {
-      "epoch": 2.7885390981898808,
-      "grad_norm": 0.003028854262083769,
-      "learning_rate": 0.00019999616484492715,
-      "loss": 46.0,
-      "step": 36472
-    },
-    {
-      "epoch": 2.7886155551732705,
-      "grad_norm": 0.002353883348405361,
-      "learning_rate": 0.0001999961646345611,
-      "loss": 46.0,
-      "step": 36473
-    },
-    {
-      "epoch": 2.7886920121566603,
-      "grad_norm": 0.0025278721004724503,
-      "learning_rate": 0.00019999616442418933,
-      "loss": 46.0,
-      "step": 36474
-    },
-    {
-      "epoch": 2.78876846914005,
-      "grad_norm": 0.002812107326462865,
-      "learning_rate": 0.00019999616421381177,
-      "loss": 46.0,
-      "step": 36475
-    },
-    {
-      "epoch": 2.78884492612344,
-      "grad_norm": 0.0036222331691533327,
-      "learning_rate": 0.00019999616400342843,
-      "loss": 46.0,
-      "step": 36476
-    },
-    {
-      "epoch": 2.7889213831068296,
-      "grad_norm": 0.0009932449320331216,
-      "learning_rate": 0.0001999961637930393,
-      "loss": 46.0,
-      "step": 36477
-    },
-    {
-      "epoch": 2.7889978400902193,
-      "grad_norm": 0.001032995874993503,
-      "learning_rate": 0.00019999616358264445,
-      "loss": 46.0,
-      "step": 36478
-    },
-    {
-      "epoch": 2.789074297073609,
-      "grad_norm": 0.001015535555779934,
-      "learning_rate": 0.0001999961633722438,
-      "loss": 46.0,
-      "step": 36479
-    },
-    {
-      "epoch": 2.7891507540569984,
-      "grad_norm": 0.004099308047443628,
-      "learning_rate": 0.00019999616316183737,
-      "loss": 46.0,
-      "step": 36480
-    },
-    {
-      "epoch": 2.789227211040388,
-      "grad_norm": 0.0024906625039875507,
-      "learning_rate": 0.00019999616295142517,
-      "loss": 46.0,
-      "step": 36481
-    },
-    {
-      "epoch": 2.789303668023778,
-      "grad_norm": 0.0019348154310137033,
-      "learning_rate": 0.00019999616274100722,
-      "loss": 46.0,
-      "step": 36482
-    },
-    {
-      "epoch": 2.7893801250071677,
-      "grad_norm": 0.0029479130171239376,
-      "learning_rate": 0.0001999961625305835,
-      "loss": 46.0,
-      "step": 36483
-    },
-    {
-      "epoch": 2.7894565819905575,
-      "grad_norm": 0.0029610886704176664,
-      "learning_rate": 0.000199996162320154,
-      "loss": 46.0,
-      "step": 36484
-    },
-    {
-      "epoch": 2.789533038973947,
-      "grad_norm": 0.002717749448493123,
-      "learning_rate": 0.00019999616210971874,
-      "loss": 46.0,
-      "step": 36485
-    },
-    {
-      "epoch": 2.789609495957337,
-      "grad_norm": 0.001117340987548232,
-      "learning_rate": 0.0001999961618992777,
-      "loss": 46.0,
-      "step": 36486
-    },
-    {
-      "epoch": 2.7896859529407267,
-      "grad_norm": 0.0037378885317593813,
-      "learning_rate": 0.0001999961616888309,
-      "loss": 46.0,
-      "step": 36487
-    },
-    {
-      "epoch": 2.7897624099241165,
-      "grad_norm": 0.002162329852581024,
-      "learning_rate": 0.00019999616147837833,
-      "loss": 46.0,
-      "step": 36488
-    },
-    {
-      "epoch": 2.7898388669075063,
-      "grad_norm": 0.005490526091307402,
-      "learning_rate": 0.00019999616126791997,
-      "loss": 46.0,
-      "step": 36489
-    },
-    {
-      "epoch": 2.789915323890896,
-      "grad_norm": 0.0013661356642842293,
-      "learning_rate": 0.0001999961610574559,
-      "loss": 46.0,
-      "step": 36490
-    },
-    {
-      "epoch": 2.789991780874286,
-      "grad_norm": 0.002923026215285063,
-      "learning_rate": 0.000199996160846986,
-      "loss": 46.0,
-      "step": 36491
-    },
-    {
-      "epoch": 2.7900682378576755,
-      "grad_norm": 0.001461016247048974,
-      "learning_rate": 0.00019999616063651033,
-      "loss": 46.0,
-      "step": 36492
-    },
-    {
-      "epoch": 2.7901446948410653,
-      "grad_norm": 0.0026465451810508966,
-      "learning_rate": 0.00019999616042602893,
-      "loss": 46.0,
-      "step": 36493
-    },
-    {
-      "epoch": 2.7902211518244546,
-      "grad_norm": 0.0008995048119686544,
-      "learning_rate": 0.00019999616021554176,
-      "loss": 46.0,
-      "step": 36494
-    },
-    {
-      "epoch": 2.7902976088078444,
-      "grad_norm": 0.0026205123867839575,
-      "learning_rate": 0.0001999961600050488,
-      "loss": 46.0,
-      "step": 36495
-    },
-    {
-      "epoch": 2.790374065791234,
-      "grad_norm": 0.004056366626173258,
-      "learning_rate": 0.00019999615979455007,
-      "loss": 46.0,
-      "step": 36496
-    },
-    {
-      "epoch": 2.790450522774624,
-      "grad_norm": 0.003872596425935626,
-      "learning_rate": 0.00019999615958404555,
-      "loss": 46.0,
-      "step": 36497
-    },
-    {
-      "epoch": 2.7905269797580137,
-      "grad_norm": 0.002656124532222748,
-      "learning_rate": 0.0001999961593735353,
-      "loss": 46.0,
-      "step": 36498
-    },
-    {
-      "epoch": 2.7906034367414034,
-      "grad_norm": 0.0016670010518282652,
-      "learning_rate": 0.00019999615916301927,
-      "loss": 46.0,
-      "step": 36499
-    },
-    {
-      "epoch": 2.790679893724793,
-      "grad_norm": 0.0013912769500166178,
-      "learning_rate": 0.00019999615895249746,
-      "loss": 46.0,
-      "step": 36500
-    },
-    {
-      "epoch": 2.790756350708183,
-      "grad_norm": 0.004013076424598694,
-      "learning_rate": 0.0001999961587419699,
-      "loss": 46.0,
-      "step": 36501
-    },
-    {
-      "epoch": 2.7908328076915723,
-      "grad_norm": 0.002834868850186467,
-      "learning_rate": 0.00019999615853143657,
-      "loss": 46.0,
-      "step": 36502
-    },
-    {
-      "epoch": 2.790909264674962,
-      "grad_norm": 0.0036473898217082024,
-      "learning_rate": 0.00019999615832089744,
-      "loss": 46.0,
-      "step": 36503
-    },
-    {
-      "epoch": 2.790985721658352,
-      "grad_norm": 0.005797350313514471,
-      "learning_rate": 0.00019999615811035258,
-      "loss": 46.0,
-      "step": 36504
-    },
-    {
-      "epoch": 2.7910621786417416,
-      "grad_norm": 0.0009111484978348017,
-      "learning_rate": 0.00019999615789980193,
-      "loss": 46.0,
-      "step": 36505
-    },
-    {
-      "epoch": 2.7911386356251313,
-      "grad_norm": 0.002440784592181444,
-      "learning_rate": 0.0001999961576892455,
-      "loss": 46.0,
-      "step": 36506
-    },
-    {
-      "epoch": 2.791215092608521,
-      "grad_norm": 0.0012514643603935838,
-      "learning_rate": 0.00019999615747868333,
-      "loss": 46.0,
-      "step": 36507
-    },
-    {
-      "epoch": 2.791291549591911,
-      "grad_norm": 0.001905411365441978,
-      "learning_rate": 0.00019999615726811536,
-      "loss": 46.0,
-      "step": 36508
-    },
-    {
-      "epoch": 2.7913680065753006,
-      "grad_norm": 0.002920496976003051,
-      "learning_rate": 0.00019999615705754161,
-      "loss": 46.0,
-      "step": 36509
-    },
-    {
-      "epoch": 2.7914444635586904,
-      "grad_norm": 0.0016895607113838196,
-      "learning_rate": 0.00019999615684696212,
-      "loss": 46.0,
-      "step": 36510
-    },
-    {
-      "epoch": 2.79152092054208,
-      "grad_norm": 0.0013283167500048876,
-      "learning_rate": 0.00019999615663637688,
-      "loss": 46.0,
-      "step": 36511
-    },
-    {
-      "epoch": 2.79159737752547,
-      "grad_norm": 0.0043432703241705894,
-      "learning_rate": 0.00019999615642578585,
-      "loss": 46.0,
-      "step": 36512
-    },
-    {
-      "epoch": 2.7916738345088596,
-      "grad_norm": 0.0035565346479415894,
-      "learning_rate": 0.00019999615621518903,
-      "loss": 46.0,
-      "step": 36513
-    },
-    {
-      "epoch": 2.7917502914922494,
-      "grad_norm": 0.004469890147447586,
-      "learning_rate": 0.00019999615600458647,
-      "loss": 46.0,
-      "step": 36514
-    },
-    {
-      "epoch": 2.791826748475639,
-      "grad_norm": 0.00395957101136446,
-      "learning_rate": 0.00019999615579397812,
-      "loss": 46.0,
-      "step": 36515
-    },
-    {
-      "epoch": 2.7919032054590285,
-      "grad_norm": 0.002598981373012066,
-      "learning_rate": 0.000199996155583364,
-      "loss": 46.0,
-      "step": 36516
-    },
-    {
-      "epoch": 2.7919796624424182,
-      "grad_norm": 0.0006251042359508574,
-      "learning_rate": 0.00019999615537274413,
-      "loss": 46.0,
-      "step": 36517
-    },
-    {
-      "epoch": 2.792056119425808,
-      "grad_norm": 0.007046767044812441,
-      "learning_rate": 0.0001999961551621185,
-      "loss": 46.0,
-      "step": 36518
-    },
-    {
-      "epoch": 2.7921325764091978,
-      "grad_norm": 0.003603205317631364,
-      "learning_rate": 0.00019999615495148708,
-      "loss": 46.0,
-      "step": 36519
-    },
-    {
-      "epoch": 2.7922090333925875,
-      "grad_norm": 0.0010431614937260747,
-      "learning_rate": 0.00019999615474084988,
-      "loss": 46.0,
-      "step": 36520
-    },
-    {
-      "epoch": 2.7922854903759773,
-      "grad_norm": 0.0029529815074056387,
-      "learning_rate": 0.00019999615453020694,
-      "loss": 46.0,
-      "step": 36521
-    },
-    {
-      "epoch": 2.792361947359367,
-      "grad_norm": 0.0012880582362413406,
-      "learning_rate": 0.0001999961543195582,
-      "loss": 46.0,
-      "step": 36522
-    },
-    {
-      "epoch": 2.792438404342757,
-      "grad_norm": 0.0016723256558179855,
-      "learning_rate": 0.0001999961541089037,
-      "loss": 46.0,
-      "step": 36523
-    },
-    {
-      "epoch": 2.792514861326146,
-      "grad_norm": 0.001146748778410256,
-      "learning_rate": 0.00019999615389824343,
-      "loss": 46.0,
-      "step": 36524
-    },
-    {
-      "epoch": 2.792591318309536,
-      "grad_norm": 0.0016539928037673235,
-      "learning_rate": 0.0001999961536875774,
-      "loss": 46.0,
-      "step": 36525
-    },
-    {
-      "epoch": 2.7926677752929256,
-      "grad_norm": 0.00864909216761589,
-      "learning_rate": 0.0001999961534769056,
-      "loss": 46.0,
-      "step": 36526
-    },
-    {
-      "epoch": 2.7927442322763154,
-      "grad_norm": 0.0016701811691746116,
-      "learning_rate": 0.00019999615326622805,
-      "loss": 46.0,
-      "step": 36527
-    },
-    {
-      "epoch": 2.792820689259705,
-      "grad_norm": 0.0006151789566501975,
-      "learning_rate": 0.00019999615305554472,
-      "loss": 46.0,
-      "step": 36528
-    },
-    {
-      "epoch": 2.792897146243095,
-      "grad_norm": 0.0015764220152050257,
-      "learning_rate": 0.0001999961528448556,
-      "loss": 46.0,
-      "step": 36529
-    },
-    {
-      "epoch": 2.7929736032264847,
-      "grad_norm": 0.002275099279358983,
-      "learning_rate": 0.0001999961526341607,
-      "loss": 46.0,
-      "step": 36530
-    },
-    {
-      "epoch": 2.7930500602098745,
-      "grad_norm": 0.0015367651358246803,
-      "learning_rate": 0.00019999615242346006,
-      "loss": 46.0,
-      "step": 36531
-    },
-    {
-      "epoch": 2.793126517193264,
-      "grad_norm": 0.0009864786406978965,
-      "learning_rate": 0.00019999615221275363,
-      "loss": 46.0,
-      "step": 36532
-    },
-    {
-      "epoch": 2.793202974176654,
-      "grad_norm": 0.00273247086443007,
-      "learning_rate": 0.00019999615200204146,
-      "loss": 46.0,
-      "step": 36533
-    },
-    {
-      "epoch": 2.7932794311600437,
-      "grad_norm": 0.0026058033108711243,
-      "learning_rate": 0.0001999961517913235,
-      "loss": 46.0,
-      "step": 36534
-    },
-    {
-      "epoch": 2.7933558881434335,
-      "grad_norm": 0.001718914951197803,
-      "learning_rate": 0.0001999961515805998,
-      "loss": 46.0,
-      "step": 36535
-    },
-    {
-      "epoch": 2.7934323451268233,
-      "grad_norm": 0.001381683861836791,
-      "learning_rate": 0.00019999615136987028,
-      "loss": 46.0,
-      "step": 36536
-    },
-    {
-      "epoch": 2.793508802110213,
-      "grad_norm": 0.003563657635822892,
-      "learning_rate": 0.00019999615115913501,
-      "loss": 46.0,
-      "step": 36537
-    },
-    {
-      "epoch": 2.7935852590936023,
-      "grad_norm": 0.0008521597483195364,
-      "learning_rate": 0.000199996150948394,
-      "loss": 46.0,
-      "step": 36538
-    },
-    {
-      "epoch": 2.793661716076992,
-      "grad_norm": 0.0009935145499184728,
-      "learning_rate": 0.0001999961507376472,
-      "loss": 46.0,
-      "step": 36539
-    },
-    {
-      "epoch": 2.793738173060382,
-      "grad_norm": 0.007693890016525984,
-      "learning_rate": 0.00019999615052689463,
-      "loss": 46.0,
-      "step": 36540
-    },
-    {
-      "epoch": 2.7938146300437716,
-      "grad_norm": 0.0028375021647661924,
-      "learning_rate": 0.00019999615031613628,
-      "loss": 46.0,
-      "step": 36541
-    },
-    {
-      "epoch": 2.7938910870271614,
-      "grad_norm": 0.0019172790925949812,
-      "learning_rate": 0.00019999615010537217,
-      "loss": 46.0,
-      "step": 36542
-    },
-    {
-      "epoch": 2.793967544010551,
-      "grad_norm": 0.0023453764151781797,
-      "learning_rate": 0.00019999614989460227,
-      "loss": 46.0,
-      "step": 36543
-    },
-    {
-      "epoch": 2.794044000993941,
-      "grad_norm": 0.0018810100154951215,
-      "learning_rate": 0.00019999614968382664,
-      "loss": 46.0,
-      "step": 36544
-    },
-    {
-      "epoch": 2.7941204579773307,
-      "grad_norm": 0.0016110677970573306,
-      "learning_rate": 0.00019999614947304522,
-      "loss": 46.0,
-      "step": 36545
-    },
-    {
-      "epoch": 2.79419691496072,
-      "grad_norm": 0.00260846852324903,
-      "learning_rate": 0.00019999614926225802,
-      "loss": 46.0,
-      "step": 36546
-    },
-    {
-      "epoch": 2.7942733719441097,
-      "grad_norm": 0.0018586397636681795,
-      "learning_rate": 0.00019999614905146507,
-      "loss": 46.0,
-      "step": 36547
-    },
-    {
-      "epoch": 2.7943498289274995,
-      "grad_norm": 0.0021287112031131983,
-      "learning_rate": 0.00019999614884066636,
-      "loss": 46.0,
-      "step": 36548
-    },
-    {
-      "epoch": 2.7944262859108893,
-      "grad_norm": 0.0032725015189498663,
-      "learning_rate": 0.00019999614862986184,
-      "loss": 46.0,
-      "step": 36549
-    },
-    {
-      "epoch": 2.794502742894279,
-      "grad_norm": 0.00107848783954978,
-      "learning_rate": 0.0001999961484190516,
-      "loss": 46.0,
-      "step": 36550
-    },
-    {
-      "epoch": 2.794579199877669,
-      "grad_norm": 0.0008354871533811092,
-      "learning_rate": 0.00019999614820823553,
-      "loss": 46.0,
-      "step": 36551
-    },
-    {
-      "epoch": 2.7946556568610585,
-      "grad_norm": 0.0033484341111034155,
-      "learning_rate": 0.00019999614799741375,
-      "loss": 46.0,
-      "step": 36552
-    },
-    {
-      "epoch": 2.7947321138444483,
-      "grad_norm": 0.0010308248456567526,
-      "learning_rate": 0.00019999614778658617,
-      "loss": 46.0,
-      "step": 36553
-    },
-    {
-      "epoch": 2.794808570827838,
-      "grad_norm": 0.0011451182654127479,
-      "learning_rate": 0.00019999614757575283,
-      "loss": 46.0,
-      "step": 36554
-    },
-    {
-      "epoch": 2.794885027811228,
-      "grad_norm": 0.007635314483195543,
-      "learning_rate": 0.00019999614736491373,
-      "loss": 46.0,
-      "step": 36555
-    },
-    {
-      "epoch": 2.7949614847946176,
-      "grad_norm": 0.0007045171223580837,
-      "learning_rate": 0.00019999614715406885,
-      "loss": 46.0,
-      "step": 36556
-    },
-    {
-      "epoch": 2.7950379417780074,
-      "grad_norm": 0.0026100799441337585,
-      "learning_rate": 0.0001999961469432182,
-      "loss": 46.0,
-      "step": 36557
-    },
-    {
-      "epoch": 2.795114398761397,
-      "grad_norm": 0.0018087979406118393,
-      "learning_rate": 0.00019999614673236178,
-      "loss": 46.0,
-      "step": 36558
-    },
-    {
-      "epoch": 2.7951908557447864,
-      "grad_norm": 0.002954366384074092,
-      "learning_rate": 0.00019999614652149958,
-      "loss": 46.0,
-      "step": 36559
-    },
-    {
-      "epoch": 2.795267312728176,
-      "grad_norm": 0.0015169483376666903,
-      "learning_rate": 0.00019999614631063163,
-      "loss": 46.0,
-      "step": 36560
-    },
-    {
-      "epoch": 2.795343769711566,
-      "grad_norm": 0.0022899971809238195,
-      "learning_rate": 0.00019999614609975792,
-      "loss": 46.0,
-      "step": 36561
-    },
-    {
-      "epoch": 2.7954202266949557,
-      "grad_norm": 0.001835056347772479,
-      "learning_rate": 0.00019999614588887843,
-      "loss": 46.0,
-      "step": 36562
-    },
-    {
-      "epoch": 2.7954966836783455,
-      "grad_norm": 0.0013374926056712866,
-      "learning_rate": 0.00019999614567799316,
-      "loss": 46.0,
-      "step": 36563
-    },
-    {
-      "epoch": 2.7955731406617352,
-      "grad_norm": 0.0013211757177487016,
-      "learning_rate": 0.00019999614546710212,
-      "loss": 46.0,
-      "step": 36564
-    },
-    {
-      "epoch": 2.795649597645125,
-      "grad_norm": 0.003265534294769168,
-      "learning_rate": 0.0001999961452562053,
-      "loss": 46.0,
-      "step": 36565
-    },
-    {
-      "epoch": 2.7957260546285148,
-      "grad_norm": 0.004683324601501226,
-      "learning_rate": 0.00019999614504530273,
-      "loss": 46.0,
-      "step": 36566
-    },
-    {
-      "epoch": 2.7958025116119045,
-      "grad_norm": 0.0015656973700970411,
-      "learning_rate": 0.00019999614483439442,
-      "loss": 46.0,
-      "step": 36567
-    },
-    {
-      "epoch": 2.795878968595294,
-      "grad_norm": 0.0036509211640805006,
-      "learning_rate": 0.0001999961446234803,
-      "loss": 46.0,
-      "step": 36568
-    },
-    {
-      "epoch": 2.7959554255786836,
-      "grad_norm": 0.0016124958638101816,
-      "learning_rate": 0.00019999614441256042,
-      "loss": 46.0,
-      "step": 36569
-    },
-    {
-      "epoch": 2.7960318825620734,
-      "grad_norm": 0.0064685288816690445,
-      "learning_rate": 0.00019999614420163474,
-      "loss": 46.0,
-      "step": 36570
-    },
-    {
-      "epoch": 2.796108339545463,
-      "grad_norm": 0.001625482807867229,
-      "learning_rate": 0.00019999614399070334,
-      "loss": 46.0,
-      "step": 36571
-    },
-    {
-      "epoch": 2.796184796528853,
-      "grad_norm": 0.0006707222200930119,
-      "learning_rate": 0.00019999614377976617,
-      "loss": 46.0,
-      "step": 36572
-    },
-    {
-      "epoch": 2.7962612535122426,
-      "grad_norm": 0.008109656162559986,
-      "learning_rate": 0.00019999614356882323,
-      "loss": 46.0,
-      "step": 36573
-    },
-    {
-      "epoch": 2.7963377104956324,
-      "grad_norm": 0.01435154490172863,
-      "learning_rate": 0.00019999614335787448,
-      "loss": 46.0,
-      "step": 36574
-    },
-    {
-      "epoch": 2.796414167479022,
-      "grad_norm": 0.0017522264970466495,
-      "learning_rate": 0.00019999614314692,
-      "loss": 46.0,
-      "step": 36575
-    },
-    {
-      "epoch": 2.796490624462412,
-      "grad_norm": 0.0024431291967630386,
-      "learning_rate": 0.00019999614293595973,
-      "loss": 46.0,
-      "step": 36576
-    },
-    {
-      "epoch": 2.7965670814458017,
-      "grad_norm": 0.0007457217434421182,
-      "learning_rate": 0.00019999614272499366,
-      "loss": 46.0,
-      "step": 36577
-    },
-    {
-      "epoch": 2.7966435384291914,
-      "grad_norm": 0.002585195703431964,
-      "learning_rate": 0.0001999961425140219,
-      "loss": 46.0,
-      "step": 36578
-    },
-    {
-      "epoch": 2.796719995412581,
-      "grad_norm": 0.0023897220380604267,
-      "learning_rate": 0.00019999614230304432,
-      "loss": 46.0,
-      "step": 36579
-    },
-    {
-      "epoch": 2.796796452395971,
-      "grad_norm": 0.003025853307917714,
-      "learning_rate": 0.00019999614209206096,
-      "loss": 46.0,
-      "step": 36580
-    },
-    {
-      "epoch": 2.7968729093793603,
-      "grad_norm": 0.000791246653534472,
-      "learning_rate": 0.00019999614188107186,
-      "loss": 46.0,
-      "step": 36581
-    },
-    {
-      "epoch": 2.79694936636275,
-      "grad_norm": 0.0030855664517730474,
-      "learning_rate": 0.00019999614167007698,
-      "loss": 46.0,
-      "step": 36582
-    },
-    {
-      "epoch": 2.79702582334614,
-      "grad_norm": 0.0031424977350980043,
-      "learning_rate": 0.00019999614145907636,
-      "loss": 46.0,
-      "step": 36583
-    },
-    {
-      "epoch": 2.7971022803295296,
-      "grad_norm": 0.0019661863334476948,
-      "learning_rate": 0.0001999961412480699,
-      "loss": 46.0,
-      "step": 36584
-    },
-    {
-      "epoch": 2.7971787373129193,
-      "grad_norm": 0.006525104399770498,
-      "learning_rate": 0.00019999614103705774,
-      "loss": 46.0,
-      "step": 36585
-    },
-    {
-      "epoch": 2.797255194296309,
-      "grad_norm": 0.004011107143014669,
-      "learning_rate": 0.00019999614082603977,
-      "loss": 46.0,
-      "step": 36586
-    },
-    {
-      "epoch": 2.797331651279699,
-      "grad_norm": 0.006423837039619684,
-      "learning_rate": 0.00019999614061501605,
-      "loss": 46.0,
-      "step": 36587
-    },
-    {
-      "epoch": 2.7974081082630886,
-      "grad_norm": 0.002542696427553892,
-      "learning_rate": 0.00019999614040398656,
-      "loss": 46.0,
-      "step": 36588
-    },
-    {
-      "epoch": 2.797484565246478,
-      "grad_norm": 0.0016533015295863152,
-      "learning_rate": 0.00019999614019295127,
-      "loss": 46.0,
-      "step": 36589
-    },
-    {
-      "epoch": 2.7975610222298677,
-      "grad_norm": 0.0015520985471084714,
-      "learning_rate": 0.00019999613998191026,
-      "loss": 46.0,
-      "step": 36590
-    },
-    {
-      "epoch": 2.7976374792132574,
-      "grad_norm": 0.0012625512899830937,
-      "learning_rate": 0.00019999613977086345,
-      "loss": 46.0,
-      "step": 36591
-    },
-    {
-      "epoch": 2.797713936196647,
-      "grad_norm": 0.0018682624213397503,
-      "learning_rate": 0.0001999961395598109,
-      "loss": 46.0,
-      "step": 36592
-    },
-    {
-      "epoch": 2.797790393180037,
-      "grad_norm": 0.0009864341700449586,
-      "learning_rate": 0.00019999613934875254,
-      "loss": 46.0,
-      "step": 36593
-    },
-    {
-      "epoch": 2.7978668501634267,
-      "grad_norm": 0.0020944890566170216,
-      "learning_rate": 0.00019999613913768843,
-      "loss": 46.0,
-      "step": 36594
-    },
-    {
-      "epoch": 2.7979433071468165,
-      "grad_norm": 0.00305737997405231,
-      "learning_rate": 0.00019999613892661856,
-      "loss": 46.0,
-      "step": 36595
-    },
-    {
-      "epoch": 2.7980197641302063,
-      "grad_norm": 0.0024078222922980785,
-      "learning_rate": 0.0001999961387155429,
-      "loss": 46.0,
-      "step": 36596
-    },
-    {
-      "epoch": 2.798096221113596,
-      "grad_norm": 0.0009995960863307118,
-      "learning_rate": 0.00019999613850446148,
-      "loss": 46.0,
-      "step": 36597
-    },
-    {
-      "epoch": 2.7981726780969858,
-      "grad_norm": 0.003978567663580179,
-      "learning_rate": 0.00019999613829337428,
-      "loss": 46.0,
-      "step": 36598
-    },
-    {
-      "epoch": 2.7982491350803755,
-      "grad_norm": 0.0009550178656354547,
-      "learning_rate": 0.00019999613808228131,
-      "loss": 46.0,
-      "step": 36599
-    },
-    {
-      "epoch": 2.7983255920637653,
-      "grad_norm": 0.0012933457037433982,
-      "learning_rate": 0.00019999613787118262,
-      "loss": 46.0,
-      "step": 36600
-    },
-    {
-      "epoch": 2.798402049047155,
-      "grad_norm": 0.0008998825214803219,
-      "learning_rate": 0.0001999961376600781,
-      "loss": 46.0,
-      "step": 36601
-    },
-    {
-      "epoch": 2.798478506030545,
-      "grad_norm": 0.0029246413614600897,
-      "learning_rate": 0.00019999613744896784,
-      "loss": 46.0,
-      "step": 36602
-    },
-    {
-      "epoch": 2.798554963013934,
-      "grad_norm": 0.010313055478036404,
-      "learning_rate": 0.00019999613723785183,
-      "loss": 46.0,
-      "step": 36603
-    },
-    {
-      "epoch": 2.798631419997324,
-      "grad_norm": 0.0011114992666989565,
-      "learning_rate": 0.00019999613702673002,
-      "loss": 46.0,
-      "step": 36604
-    },
-    {
-      "epoch": 2.7987078769807137,
-      "grad_norm": 0.001763931941241026,
-      "learning_rate": 0.00019999613681560244,
-      "loss": 46.0,
-      "step": 36605
-    },
-    {
-      "epoch": 2.7987843339641034,
-      "grad_norm": 0.004120492842048407,
-      "learning_rate": 0.0001999961366044691,
-      "loss": 46.0,
-      "step": 36606
-    },
-    {
-      "epoch": 2.798860790947493,
-      "grad_norm": 0.003552485490217805,
-      "learning_rate": 0.00019999613639332998,
-      "loss": 46.0,
-      "step": 36607
-    },
-    {
-      "epoch": 2.798937247930883,
-      "grad_norm": 0.0035627589095383883,
-      "learning_rate": 0.0001999961361821851,
-      "loss": 46.0,
-      "step": 36608
-    },
-    {
-      "epoch": 2.7990137049142727,
-      "grad_norm": 0.0026878025382757187,
-      "learning_rate": 0.00019999613597103445,
-      "loss": 46.0,
-      "step": 36609
-    },
-    {
-      "epoch": 2.7990901618976625,
-      "grad_norm": 0.0016114194877445698,
-      "learning_rate": 0.00019999613575987806,
-      "loss": 46.0,
-      "step": 36610
-    },
-    {
-      "epoch": 2.799166618881052,
-      "grad_norm": 0.004447620827704668,
-      "learning_rate": 0.0001999961355487159,
-      "loss": 46.0,
-      "step": 36611
-    },
-    {
-      "epoch": 2.7992430758644415,
-      "grad_norm": 0.0010776607086881995,
-      "learning_rate": 0.00019999613533754792,
-      "loss": 46.0,
-      "step": 36612
-    },
-    {
-      "epoch": 2.7993195328478313,
-      "grad_norm": 0.0018496030243113637,
-      "learning_rate": 0.00019999613512637418,
-      "loss": 46.0,
-      "step": 36613
-    },
-    {
-      "epoch": 2.799395989831221,
-      "grad_norm": 0.0019583420362323523,
-      "learning_rate": 0.00019999613491519469,
-      "loss": 46.0,
-      "step": 36614
-    },
-    {
-      "epoch": 2.799472446814611,
-      "grad_norm": 0.0009847491746768355,
-      "learning_rate": 0.00019999613470400945,
-      "loss": 46.0,
-      "step": 36615
-    },
-    {
-      "epoch": 2.7995489037980006,
-      "grad_norm": 0.005545987281948328,
-      "learning_rate": 0.00019999613449281839,
-      "loss": 46.0,
-      "step": 36616
-    },
-    {
-      "epoch": 2.7996253607813903,
-      "grad_norm": 0.001514257164672017,
-      "learning_rate": 0.00019999613428162158,
-      "loss": 46.0,
-      "step": 36617
-    },
-    {
-      "epoch": 2.79970181776478,
-      "grad_norm": 0.0012794131180271506,
-      "learning_rate": 0.00019999613407041902,
-      "loss": 46.0,
-      "step": 36618
-    },
-    {
-      "epoch": 2.79977827474817,
-      "grad_norm": 0.00804176740348339,
-      "learning_rate": 0.0001999961338592107,
-      "loss": 46.0,
-      "step": 36619
-    },
-    {
-      "epoch": 2.7998547317315596,
-      "grad_norm": 0.00441935146227479,
-      "learning_rate": 0.00019999613364799656,
-      "loss": 46.0,
-      "step": 36620
-    },
-    {
-      "epoch": 2.7999311887149494,
-      "grad_norm": 0.001385703100822866,
-      "learning_rate": 0.00019999613343677668,
-      "loss": 46.0,
-      "step": 36621
-    },
-    {
-      "epoch": 2.800007645698339,
-      "grad_norm": 0.005872361361980438,
-      "learning_rate": 0.00019999613322555103,
-      "loss": 46.0,
-      "step": 36622
-    },
-    {
-      "epoch": 2.800084102681729,
-      "grad_norm": 0.0010021342895925045,
-      "learning_rate": 0.00019999613301431964,
-      "loss": 46.0,
-      "step": 36623
-    },
-    {
-      "epoch": 2.8001605596651187,
-      "grad_norm": 0.0036808105651289225,
-      "learning_rate": 0.00019999613280308242,
-      "loss": 46.0,
-      "step": 36624
-    },
-    {
-      "epoch": 2.800237016648508,
-      "grad_norm": 0.002813017461448908,
-      "learning_rate": 0.0001999961325918395,
-      "loss": 46.0,
-      "step": 36625
-    },
-    {
-      "epoch": 2.8003134736318978,
-      "grad_norm": 0.002767500700429082,
-      "learning_rate": 0.00019999613238059073,
-      "loss": 46.0,
-      "step": 36626
-    },
-    {
-      "epoch": 2.8003899306152875,
-      "grad_norm": 0.0016132600139826536,
-      "learning_rate": 0.00019999613216933627,
-      "loss": 46.0,
-      "step": 36627
-    },
-    {
-      "epoch": 2.8004663875986773,
-      "grad_norm": 0.0026759873144328594,
-      "learning_rate": 0.000199996131958076,
-      "loss": 46.0,
-      "step": 36628
-    },
-    {
-      "epoch": 2.800542844582067,
-      "grad_norm": 0.001208735746331513,
-      "learning_rate": 0.00019999613174680997,
-      "loss": 46.0,
-      "step": 36629
-    },
-    {
-      "epoch": 2.800619301565457,
-      "grad_norm": 0.002023642882704735,
-      "learning_rate": 0.00019999613153553816,
-      "loss": 46.0,
-      "step": 36630
-    },
-    {
-      "epoch": 2.8006957585488466,
-      "grad_norm": 0.0018092935206368566,
-      "learning_rate": 0.00019999613132426058,
-      "loss": 46.0,
-      "step": 36631
-    },
-    {
-      "epoch": 2.8007722155322363,
-      "grad_norm": 0.004044120665639639,
-      "learning_rate": 0.00019999613111297725,
-      "loss": 46.0,
-      "step": 36632
-    },
-    {
-      "epoch": 2.8008486725156256,
-      "grad_norm": 0.0020952471531927586,
-      "learning_rate": 0.00019999613090168814,
-      "loss": 46.0,
-      "step": 36633
-    },
-    {
-      "epoch": 2.8009251294990154,
-      "grad_norm": 0.0015054637333378196,
-      "learning_rate": 0.00019999613069039324,
-      "loss": 46.0,
-      "step": 36634
-    },
-    {
-      "epoch": 2.801001586482405,
-      "grad_norm": 0.007036192808300257,
-      "learning_rate": 0.0001999961304790926,
-      "loss": 46.0,
-      "step": 36635
-    },
-    {
-      "epoch": 2.801078043465795,
-      "grad_norm": 0.002783105941489339,
-      "learning_rate": 0.0001999961302677862,
-      "loss": 46.0,
-      "step": 36636
-    },
-    {
-      "epoch": 2.8011545004491847,
-      "grad_norm": 0.0049445307813584805,
-      "learning_rate": 0.00019999613005647403,
-      "loss": 46.0,
-      "step": 36637
-    },
-    {
-      "epoch": 2.8012309574325744,
-      "grad_norm": 0.002596270525828004,
-      "learning_rate": 0.00019999612984515603,
-      "loss": 46.0,
-      "step": 36638
-    },
-    {
-      "epoch": 2.801307414415964,
-      "grad_norm": 0.001712812576442957,
-      "learning_rate": 0.00019999612963383232,
-      "loss": 46.0,
-      "step": 36639
-    },
-    {
-      "epoch": 2.801383871399354,
-      "grad_norm": 0.0025451479014009237,
-      "learning_rate": 0.00019999612942250283,
-      "loss": 46.0,
-      "step": 36640
-    },
-    {
-      "epoch": 2.8014603283827437,
-      "grad_norm": 0.006407268811017275,
-      "learning_rate": 0.00019999612921116756,
-      "loss": 46.0,
-      "step": 36641
-    },
-    {
-      "epoch": 2.8015367853661335,
-      "grad_norm": 0.0009229324059560895,
-      "learning_rate": 0.00019999612899982653,
-      "loss": 46.0,
-      "step": 36642
-    },
-    {
-      "epoch": 2.8016132423495232,
-      "grad_norm": 0.000678193406201899,
-      "learning_rate": 0.00019999612878847972,
-      "loss": 46.0,
-      "step": 36643
-    },
-    {
-      "epoch": 2.801689699332913,
-      "grad_norm": 0.0026372834108769894,
-      "learning_rate": 0.00019999612857712716,
-      "loss": 46.0,
-      "step": 36644
-    },
-    {
-      "epoch": 2.8017661563163028,
-      "grad_norm": 0.0019961786456406116,
-      "learning_rate": 0.0001999961283657688,
-      "loss": 46.0,
-      "step": 36645
-    },
-    {
-      "epoch": 2.8018426132996925,
-      "grad_norm": 0.0011185986222699285,
-      "learning_rate": 0.0001999961281544047,
-      "loss": 46.0,
-      "step": 36646
-    },
-    {
-      "epoch": 2.801919070283082,
-      "grad_norm": 0.0022907606326043606,
-      "learning_rate": 0.00019999612794303483,
-      "loss": 46.0,
-      "step": 36647
-    },
-    {
-      "epoch": 2.8019955272664716,
-      "grad_norm": 0.001931617734953761,
-      "learning_rate": 0.00019999612773165918,
-      "loss": 46.0,
-      "step": 36648
-    },
-    {
-      "epoch": 2.8020719842498614,
-      "grad_norm": 0.003951586317270994,
-      "learning_rate": 0.00019999612752027773,
-      "loss": 46.0,
-      "step": 36649
-    },
-    {
-      "epoch": 2.802148441233251,
-      "grad_norm": 0.0009364747093059123,
-      "learning_rate": 0.00019999612730889056,
-      "loss": 46.0,
-      "step": 36650
-    },
-    {
-      "epoch": 2.802224898216641,
-      "grad_norm": 0.002199733629822731,
-      "learning_rate": 0.00019999612709749762,
-      "loss": 46.0,
-      "step": 36651
-    },
-    {
-      "epoch": 2.8023013552000307,
-      "grad_norm": 0.0023284186609089375,
-      "learning_rate": 0.0001999961268860989,
-      "loss": 46.0,
-      "step": 36652
-    },
-    {
-      "epoch": 2.8023778121834204,
-      "grad_norm": 0.0029297014698386192,
-      "learning_rate": 0.00019999612667469442,
-      "loss": 46.0,
-      "step": 36653
-    },
-    {
-      "epoch": 2.80245426916681,
-      "grad_norm": 0.00334092671982944,
-      "learning_rate": 0.00019999612646328413,
-      "loss": 46.0,
-      "step": 36654
-    },
-    {
-      "epoch": 2.8025307261501995,
-      "grad_norm": 0.0008494119974784553,
-      "learning_rate": 0.0001999961262518681,
-      "loss": 46.0,
-      "step": 36655
-    },
-    {
-      "epoch": 2.8026071831335893,
-      "grad_norm": 0.001970664830878377,
-      "learning_rate": 0.00019999612604044628,
-      "loss": 46.0,
-      "step": 36656
-    },
-    {
-      "epoch": 2.802683640116979,
-      "grad_norm": 0.0009151042904704809,
-      "learning_rate": 0.00019999612582901873,
-      "loss": 46.0,
-      "step": 36657
-    },
-    {
-      "epoch": 2.8027600971003688,
-      "grad_norm": 0.006365691311657429,
-      "learning_rate": 0.00019999612561758537,
-      "loss": 46.0,
-      "step": 36658
-    },
-    {
-      "epoch": 2.8028365540837585,
-      "grad_norm": 0.0030670086853206158,
-      "learning_rate": 0.00019999612540614627,
-      "loss": 46.0,
-      "step": 36659
-    },
-    {
-      "epoch": 2.8029130110671483,
-      "grad_norm": 0.0006867140764370561,
-      "learning_rate": 0.0001999961251947014,
-      "loss": 46.0,
-      "step": 36660
-    },
-    {
-      "epoch": 2.802989468050538,
-      "grad_norm": 0.003196683246642351,
-      "learning_rate": 0.00019999612498325075,
-      "loss": 46.0,
-      "step": 36661
-    },
-    {
-      "epoch": 2.803065925033928,
-      "grad_norm": 0.0038683305028826,
-      "learning_rate": 0.00019999612477179432,
-      "loss": 46.0,
-      "step": 36662
-    },
-    {
-      "epoch": 2.8031423820173176,
-      "grad_norm": 0.0011197554413229227,
-      "learning_rate": 0.00019999612456033216,
-      "loss": 46.0,
-      "step": 36663
-    },
-    {
-      "epoch": 2.8032188390007073,
-      "grad_norm": 0.004056317266076803,
-      "learning_rate": 0.0001999961243488642,
-      "loss": 46.0,
-      "step": 36664
-    },
-    {
-      "epoch": 2.803295295984097,
-      "grad_norm": 0.004966258071362972,
-      "learning_rate": 0.00019999612413739047,
-      "loss": 46.0,
-      "step": 36665
-    },
-    {
-      "epoch": 2.803371752967487,
-      "grad_norm": 0.004595424514263868,
-      "learning_rate": 0.00019999612392591096,
-      "loss": 46.0,
-      "step": 36666
-    },
-    {
-      "epoch": 2.8034482099508766,
-      "grad_norm": 0.0008782196091488004,
-      "learning_rate": 0.00019999612371442572,
-      "loss": 46.0,
-      "step": 36667
-    },
-    {
-      "epoch": 2.8035246669342664,
-      "grad_norm": 0.01031880360096693,
-      "learning_rate": 0.00019999612350293466,
-      "loss": 46.0,
-      "step": 36668
-    },
-    {
-      "epoch": 2.8036011239176557,
-      "grad_norm": 0.007646709214895964,
-      "learning_rate": 0.00019999612329143788,
-      "loss": 46.0,
-      "step": 36669
-    },
-    {
-      "epoch": 2.8036775809010455,
-      "grad_norm": 0.001982444664463401,
-      "learning_rate": 0.0001999961230799353,
-      "loss": 46.0,
-      "step": 36670
-    },
-    {
-      "epoch": 2.8037540378844352,
-      "grad_norm": 0.007043974939733744,
-      "learning_rate": 0.00019999612286842694,
-      "loss": 46.0,
-      "step": 36671
-    },
-    {
-      "epoch": 2.803830494867825,
-      "grad_norm": 0.0013033469440415502,
-      "learning_rate": 0.00019999612265691284,
-      "loss": 46.0,
-      "step": 36672
-    },
-    {
-      "epoch": 2.8039069518512147,
-      "grad_norm": 0.0013943207450211048,
-      "learning_rate": 0.00019999612244539296,
-      "loss": 46.0,
-      "step": 36673
-    },
-    {
-      "epoch": 2.8039834088346045,
-      "grad_norm": 0.004663140047341585,
-      "learning_rate": 0.00019999612223386732,
-      "loss": 46.0,
-      "step": 36674
-    },
-    {
-      "epoch": 2.8040598658179943,
-      "grad_norm": 0.00047276122495532036,
-      "learning_rate": 0.00019999612202233592,
-      "loss": 46.0,
-      "step": 36675
-    },
-    {
-      "epoch": 2.804136322801384,
-      "grad_norm": 0.002373308641836047,
-      "learning_rate": 0.00019999612181079873,
-      "loss": 46.0,
-      "step": 36676
-    },
-    {
-      "epoch": 2.8042127797847733,
-      "grad_norm": 0.001830788911320269,
-      "learning_rate": 0.00019999612159925576,
-      "loss": 46.0,
-      "step": 36677
-    },
-    {
-      "epoch": 2.804289236768163,
-      "grad_norm": 0.0013828761875629425,
-      "learning_rate": 0.00019999612138770702,
-      "loss": 46.0,
-      "step": 36678
-    },
-    {
-      "epoch": 2.804365693751553,
-      "grad_norm": 0.0033276095055043697,
-      "learning_rate": 0.00019999612117615253,
-      "loss": 46.0,
-      "step": 36679
-    },
-    {
-      "epoch": 2.8044421507349426,
-      "grad_norm": 0.0018032494699582458,
-      "learning_rate": 0.0001999961209645923,
-      "loss": 46.0,
-      "step": 36680
-    },
-    {
-      "epoch": 2.8045186077183324,
-      "grad_norm": 0.0014347931137308478,
-      "learning_rate": 0.00019999612075302623,
-      "loss": 46.0,
-      "step": 36681
-    },
-    {
-      "epoch": 2.804595064701722,
-      "grad_norm": 0.0051933773793280125,
-      "learning_rate": 0.00019999612054145445,
-      "loss": 46.0,
-      "step": 36682
-    },
-    {
-      "epoch": 2.804671521685112,
-      "grad_norm": 0.002236979780718684,
-      "learning_rate": 0.00019999612032987684,
-      "loss": 46.0,
-      "step": 36683
-    },
-    {
-      "epoch": 2.8047479786685017,
-      "grad_norm": 0.00294465571641922,
-      "learning_rate": 0.0001999961201182935,
-      "loss": 46.0,
-      "step": 36684
-    },
-    {
-      "epoch": 2.8048244356518914,
-      "grad_norm": 0.0033904453739523888,
-      "learning_rate": 0.0001999961199067044,
-      "loss": 46.0,
-      "step": 36685
-    },
-    {
-      "epoch": 2.804900892635281,
-      "grad_norm": 0.0023571569472551346,
-      "learning_rate": 0.00019999611969510954,
-      "loss": 46.0,
-      "step": 36686
-    },
-    {
-      "epoch": 2.804977349618671,
-      "grad_norm": 0.003928797319531441,
-      "learning_rate": 0.0001999961194835089,
-      "loss": 46.0,
-      "step": 36687
-    },
-    {
-      "epoch": 2.8050538066020607,
-      "grad_norm": 0.0013346492778509855,
-      "learning_rate": 0.00019999611927190247,
-      "loss": 46.0,
-      "step": 36688
-    },
-    {
-      "epoch": 2.8051302635854505,
-      "grad_norm": 0.0025068700779229403,
-      "learning_rate": 0.00019999611906029027,
-      "loss": 46.0,
-      "step": 36689
-    },
-    {
-      "epoch": 2.80520672056884,
-      "grad_norm": 0.001671082922257483,
-      "learning_rate": 0.0001999961188486723,
-      "loss": 46.0,
-      "step": 36690
-    },
-    {
-      "epoch": 2.8052831775522296,
-      "grad_norm": 0.002068845322355628,
-      "learning_rate": 0.0001999961186370486,
-      "loss": 46.0,
-      "step": 36691
-    },
-    {
-      "epoch": 2.8053596345356193,
-      "grad_norm": 0.002072999021038413,
-      "learning_rate": 0.0001999961184254191,
-      "loss": 46.0,
-      "step": 36692
-    },
-    {
-      "epoch": 2.805436091519009,
-      "grad_norm": 0.002299954416230321,
-      "learning_rate": 0.00019999611821378384,
-      "loss": 46.0,
-      "step": 36693
-    },
-    {
-      "epoch": 2.805512548502399,
-      "grad_norm": 0.003928791731595993,
-      "learning_rate": 0.0001999961180021428,
-      "loss": 46.0,
-      "step": 36694
-    },
-    {
-      "epoch": 2.8055890054857886,
-      "grad_norm": 0.0034237774088978767,
-      "learning_rate": 0.000199996117790496,
-      "loss": 46.0,
-      "step": 36695
-    },
-    {
-      "epoch": 2.8056654624691784,
-      "grad_norm": 0.0029041548259556293,
-      "learning_rate": 0.00019999611757884345,
-      "loss": 46.0,
-      "step": 36696
-    },
-    {
-      "epoch": 2.805741919452568,
-      "grad_norm": 0.0033824958372861147,
-      "learning_rate": 0.0001999961173671851,
-      "loss": 46.0,
-      "step": 36697
-    },
-    {
-      "epoch": 2.805818376435958,
-      "grad_norm": 0.00648773368448019,
-      "learning_rate": 0.000199996117155521,
-      "loss": 46.0,
-      "step": 36698
-    },
-    {
-      "epoch": 2.805894833419347,
-      "grad_norm": 0.003248832654207945,
-      "learning_rate": 0.0001999961169438511,
-      "loss": 46.0,
-      "step": 36699
-    },
-    {
-      "epoch": 2.805971290402737,
-      "grad_norm": 0.0012092050164937973,
-      "learning_rate": 0.00019999611673217547,
-      "loss": 46.0,
-      "step": 36700
-    },
-    {
-      "epoch": 2.8060477473861267,
-      "grad_norm": 0.004391463007777929,
-      "learning_rate": 0.00019999611652049405,
-      "loss": 46.0,
-      "step": 36701
-    },
-    {
-      "epoch": 2.8061242043695165,
-      "grad_norm": 0.0015033665113151073,
-      "learning_rate": 0.00019999611630880682,
-      "loss": 46.0,
-      "step": 36702
-    },
-    {
-      "epoch": 2.8062006613529062,
-      "grad_norm": 0.0017571456264704466,
-      "learning_rate": 0.00019999611609711388,
-      "loss": 46.0,
-      "step": 36703
-    },
-    {
-      "epoch": 2.806277118336296,
-      "grad_norm": 0.001951251644641161,
-      "learning_rate": 0.00019999611588541517,
-      "loss": 46.0,
-      "step": 36704
-    },
-    {
-      "epoch": 2.8063535753196858,
-      "grad_norm": 0.0009492213139310479,
-      "learning_rate": 0.00019999611567371068,
-      "loss": 46.0,
-      "step": 36705
-    },
-    {
-      "epoch": 2.8064300323030755,
-      "grad_norm": 0.002249964280053973,
-      "learning_rate": 0.0001999961154620004,
-      "loss": 46.0,
-      "step": 36706
-    },
-    {
-      "epoch": 2.8065064892864653,
-      "grad_norm": 0.003665202995762229,
-      "learning_rate": 0.00019999611525028436,
-      "loss": 46.0,
-      "step": 36707
-    },
-    {
-      "epoch": 2.806582946269855,
-      "grad_norm": 0.005576695315539837,
-      "learning_rate": 0.00019999611503856258,
-      "loss": 46.0,
-      "step": 36708
-    },
-    {
-      "epoch": 2.806659403253245,
-      "grad_norm": 0.0066673350520431995,
-      "learning_rate": 0.000199996114826835,
-      "loss": 46.0,
-      "step": 36709
-    },
-    {
-      "epoch": 2.8067358602366346,
-      "grad_norm": 0.0008615495171397924,
-      "learning_rate": 0.00019999611461510167,
-      "loss": 46.0,
-      "step": 36710
-    },
-    {
-      "epoch": 2.8068123172200243,
-      "grad_norm": 0.00974003504961729,
-      "learning_rate": 0.00019999611440336254,
-      "loss": 46.0,
-      "step": 36711
-    },
-    {
-      "epoch": 2.8068887742034137,
-      "grad_norm": 0.0010575454216450453,
-      "learning_rate": 0.00019999611419161767,
-      "loss": 46.0,
-      "step": 36712
-    },
-    {
-      "epoch": 2.8069652311868034,
-      "grad_norm": 0.001377164269797504,
-      "learning_rate": 0.00019999611397986702,
-      "loss": 46.0,
-      "step": 36713
-    },
-    {
-      "epoch": 2.807041688170193,
-      "grad_norm": 0.009895585477352142,
-      "learning_rate": 0.0001999961137681106,
-      "loss": 46.0,
-      "step": 36714
-    },
-    {
-      "epoch": 2.807118145153583,
-      "grad_norm": 0.004479090683162212,
-      "learning_rate": 0.0001999961135563484,
-      "loss": 46.0,
-      "step": 36715
-    },
-    {
-      "epoch": 2.8071946021369727,
-      "grad_norm": 0.004253746941685677,
-      "learning_rate": 0.00019999611334458047,
-      "loss": 46.0,
-      "step": 36716
-    },
-    {
-      "epoch": 2.8072710591203625,
-      "grad_norm": 0.002376875374466181,
-      "learning_rate": 0.00019999611313280673,
-      "loss": 46.0,
-      "step": 36717
-    },
-    {
-      "epoch": 2.807347516103752,
-      "grad_norm": 0.003160391701385379,
-      "learning_rate": 0.0001999961129210272,
-      "loss": 46.0,
-      "step": 36718
-    },
-    {
-      "epoch": 2.807423973087142,
-      "grad_norm": 0.00340360589325428,
-      "learning_rate": 0.00019999611270924198,
-      "loss": 46.0,
-      "step": 36719
-    },
-    {
-      "epoch": 2.8075004300705313,
-      "grad_norm": 0.0024417827371507883,
-      "learning_rate": 0.00019999611249745094,
-      "loss": 46.0,
-      "step": 36720
-    },
-    {
-      "epoch": 2.807576887053921,
-      "grad_norm": 0.004683398175984621,
-      "learning_rate": 0.00019999611228565414,
-      "loss": 46.0,
-      "step": 36721
-    },
-    {
-      "epoch": 2.807653344037311,
-      "grad_norm": 0.0024101382587105036,
-      "learning_rate": 0.00019999611207385156,
-      "loss": 46.0,
-      "step": 36722
-    },
-    {
-      "epoch": 2.8077298010207006,
-      "grad_norm": 0.003924605902284384,
-      "learning_rate": 0.0001999961118620432,
-      "loss": 46.0,
-      "step": 36723
-    },
-    {
-      "epoch": 2.8078062580040903,
-      "grad_norm": 0.0025500264018774033,
-      "learning_rate": 0.0001999961116502291,
-      "loss": 46.0,
-      "step": 36724
-    },
-    {
-      "epoch": 2.80788271498748,
-      "grad_norm": 0.0017185703618451953,
-      "learning_rate": 0.00019999611143840923,
-      "loss": 46.0,
-      "step": 36725
-    },
-    {
-      "epoch": 2.80795917197087,
-      "grad_norm": 0.0017719519091770053,
-      "learning_rate": 0.00019999611122658358,
-      "loss": 46.0,
-      "step": 36726
-    },
-    {
-      "epoch": 2.8080356289542596,
-      "grad_norm": 0.0038139228709042072,
-      "learning_rate": 0.0001999961110147522,
-      "loss": 46.0,
-      "step": 36727
-    },
-    {
-      "epoch": 2.8081120859376494,
-      "grad_norm": 0.002001163549721241,
-      "learning_rate": 0.000199996110802915,
-      "loss": 46.0,
-      "step": 36728
-    },
-    {
-      "epoch": 2.808188542921039,
-      "grad_norm": 0.0032030779402703047,
-      "learning_rate": 0.00019999611059107203,
-      "loss": 46.0,
-      "step": 36729
-    },
-    {
-      "epoch": 2.808264999904429,
-      "grad_norm": 0.0012311318423599005,
-      "learning_rate": 0.0001999961103792233,
-      "loss": 46.0,
-      "step": 36730
-    },
-    {
-      "epoch": 2.8083414568878187,
-      "grad_norm": 0.0024967710487544537,
-      "learning_rate": 0.0001999961101673688,
-      "loss": 46.0,
-      "step": 36731
-    },
-    {
-      "epoch": 2.8084179138712084,
-      "grad_norm": 0.012091250158846378,
-      "learning_rate": 0.00019999610995550854,
-      "loss": 46.0,
-      "step": 36732
-    },
-    {
-      "epoch": 2.808494370854598,
-      "grad_norm": 0.0014718605671077967,
-      "learning_rate": 0.0001999961097436425,
-      "loss": 46.0,
-      "step": 36733
-    },
-    {
-      "epoch": 2.8085708278379875,
-      "grad_norm": 0.0026038612704724073,
-      "learning_rate": 0.0001999961095317707,
-      "loss": 46.0,
-      "step": 36734
-    },
-    {
-      "epoch": 2.8086472848213773,
-      "grad_norm": 0.0022275352384895086,
-      "learning_rate": 0.00019999610931989312,
-      "loss": 46.0,
-      "step": 36735
-    },
-    {
-      "epoch": 2.808723741804767,
-      "grad_norm": 0.002716431627050042,
-      "learning_rate": 0.00019999610910800982,
-      "loss": 46.0,
-      "step": 36736
-    },
-    {
-      "epoch": 2.808800198788157,
-      "grad_norm": 0.0014993941877037287,
-      "learning_rate": 0.0001999961088961207,
-      "loss": 46.0,
-      "step": 36737
-    },
-    {
-      "epoch": 2.8088766557715465,
-      "grad_norm": 0.001749646384268999,
-      "learning_rate": 0.0001999961086842258,
-      "loss": 46.0,
-      "step": 36738
-    },
-    {
-      "epoch": 2.8089531127549363,
-      "grad_norm": 0.003653186373412609,
-      "learning_rate": 0.00019999610847232514,
-      "loss": 46.0,
-      "step": 36739
-    },
-    {
-      "epoch": 2.809029569738326,
-      "grad_norm": 0.007551093120127916,
-      "learning_rate": 0.00019999610826041875,
-      "loss": 46.0,
-      "step": 36740
-    },
-    {
-      "epoch": 2.809106026721716,
-      "grad_norm": 0.0028746856842190027,
-      "learning_rate": 0.00019999610804850656,
-      "loss": 46.0,
-      "step": 36741
-    },
-    {
-      "epoch": 2.809182483705105,
-      "grad_norm": 0.0012204285012558103,
-      "learning_rate": 0.0001999961078365886,
-      "loss": 46.0,
-      "step": 36742
-    },
-    {
-      "epoch": 2.809258940688495,
-      "grad_norm": 0.0009773381752893329,
-      "learning_rate": 0.00019999610762466488,
-      "loss": 46.0,
-      "step": 36743
-    },
-    {
-      "epoch": 2.8093353976718847,
-      "grad_norm": 0.00681289890781045,
-      "learning_rate": 0.0001999961074127354,
-      "loss": 46.0,
-      "step": 36744
-    },
-    {
-      "epoch": 2.8094118546552744,
-      "grad_norm": 0.0008422063547186553,
-      "learning_rate": 0.0001999961072008001,
-      "loss": 46.0,
-      "step": 36745
-    },
-    {
-      "epoch": 2.809488311638664,
-      "grad_norm": 0.0015375600196421146,
-      "learning_rate": 0.0001999961069888591,
-      "loss": 46.0,
-      "step": 36746
-    },
-    {
-      "epoch": 2.809564768622054,
-      "grad_norm": 0.0026290591340512037,
-      "learning_rate": 0.00019999610677691226,
-      "loss": 46.0,
-      "step": 36747
-    },
-    {
-      "epoch": 2.8096412256054437,
-      "grad_norm": 0.0010255371453240514,
-      "learning_rate": 0.0001999961065649597,
-      "loss": 46.0,
-      "step": 36748
-    },
-    {
-      "epoch": 2.8097176825888335,
-      "grad_norm": 0.0021664327941834927,
-      "learning_rate": 0.00019999610635300136,
-      "loss": 46.0,
-      "step": 36749
-    },
-    {
-      "epoch": 2.8097941395722232,
-      "grad_norm": 0.0011113504879176617,
-      "learning_rate": 0.00019999610614103726,
-      "loss": 46.0,
-      "step": 36750
-    },
-    {
-      "epoch": 2.809870596555613,
-      "grad_norm": 0.003068802645429969,
-      "learning_rate": 0.00019999610592906736,
-      "loss": 46.0,
-      "step": 36751
-    },
-    {
-      "epoch": 2.8099470535390028,
-      "grad_norm": 0.004069286398589611,
-      "learning_rate": 0.00019999610571709174,
-      "loss": 46.0,
-      "step": 36752
-    },
-    {
-      "epoch": 2.8100235105223925,
-      "grad_norm": 0.0007427320233546197,
-      "learning_rate": 0.00019999610550511032,
-      "loss": 46.0,
-      "step": 36753
-    },
-    {
-      "epoch": 2.8100999675057823,
-      "grad_norm": 0.001602928969077766,
-      "learning_rate": 0.00019999610529312313,
-      "loss": 46.0,
-      "step": 36754
-    },
-    {
-      "epoch": 2.810176424489172,
-      "grad_norm": 0.0034579120110720396,
-      "learning_rate": 0.00019999610508113019,
-      "loss": 46.0,
-      "step": 36755
-    },
-    {
-      "epoch": 2.8102528814725614,
-      "grad_norm": 0.0034717454109340906,
-      "learning_rate": 0.00019999610486913145,
-      "loss": 46.0,
-      "step": 36756
-    },
-    {
-      "epoch": 2.810329338455951,
-      "grad_norm": 0.002721151802688837,
-      "learning_rate": 0.00019999610465712696,
-      "loss": 46.0,
-      "step": 36757
-    },
-    {
-      "epoch": 2.810405795439341,
-      "grad_norm": 0.004295787774026394,
-      "learning_rate": 0.00019999610444511667,
-      "loss": 46.0,
-      "step": 36758
-    },
-    {
-      "epoch": 2.8104822524227306,
-      "grad_norm": 0.0008510529878549278,
-      "learning_rate": 0.00019999610423310064,
-      "loss": 46.0,
-      "step": 36759
-    },
-    {
-      "epoch": 2.8105587094061204,
-      "grad_norm": 0.0027178653981536627,
-      "learning_rate": 0.00019999610402107883,
-      "loss": 46.0,
-      "step": 36760
-    },
-    {
-      "epoch": 2.81063516638951,
-      "grad_norm": 0.001633347012102604,
-      "learning_rate": 0.00019999610380905128,
-      "loss": 46.0,
-      "step": 36761
-    },
-    {
-      "epoch": 2.8107116233729,
-      "grad_norm": 0.0008918690728023648,
-      "learning_rate": 0.00019999610359701793,
-      "loss": 46.0,
-      "step": 36762
-    },
-    {
-      "epoch": 2.8107880803562897,
-      "grad_norm": 0.0025272017810493708,
-      "learning_rate": 0.00019999610338497883,
-      "loss": 46.0,
-      "step": 36763
-    },
-    {
-      "epoch": 2.810864537339679,
-      "grad_norm": 0.0019864998757839203,
-      "learning_rate": 0.00019999610317293396,
-      "loss": 46.0,
-      "step": 36764
-    },
-    {
-      "epoch": 2.8109409943230688,
-      "grad_norm": 0.0037984438240528107,
-      "learning_rate": 0.00019999610296088328,
-      "loss": 46.0,
-      "step": 36765
-    },
-    {
-      "epoch": 2.8110174513064585,
-      "grad_norm": 0.0013584971893578768,
-      "learning_rate": 0.00019999610274882686,
-      "loss": 46.0,
-      "step": 36766
-    },
-    {
-      "epoch": 2.8110939082898483,
-      "grad_norm": 0.0014875085325911641,
-      "learning_rate": 0.0001999961025367647,
-      "loss": 46.0,
-      "step": 36767
-    },
-    {
-      "epoch": 2.811170365273238,
-      "grad_norm": 0.0020454153418540955,
-      "learning_rate": 0.00019999610232469673,
-      "loss": 46.0,
-      "step": 36768
-    },
-    {
-      "epoch": 2.811246822256628,
-      "grad_norm": 0.002144475234672427,
-      "learning_rate": 0.00019999610211262302,
-      "loss": 46.0,
-      "step": 36769
-    },
-    {
-      "epoch": 2.8113232792400176,
-      "grad_norm": 0.0023602943401783705,
-      "learning_rate": 0.0001999961019005435,
-      "loss": 46.0,
-      "step": 36770
-    },
-    {
-      "epoch": 2.8113997362234073,
-      "grad_norm": 0.0009862781735137105,
-      "learning_rate": 0.00019999610168845825,
-      "loss": 46.0,
-      "step": 36771
-    },
-    {
-      "epoch": 2.811476193206797,
-      "grad_norm": 0.006907739210873842,
-      "learning_rate": 0.0001999961014763672,
-      "loss": 46.0,
-      "step": 36772
-    },
-    {
-      "epoch": 2.811552650190187,
-      "grad_norm": 0.002100649755448103,
-      "learning_rate": 0.0001999961012642704,
-      "loss": 46.0,
-      "step": 36773
-    },
-    {
-      "epoch": 2.8116291071735766,
-      "grad_norm": 0.001227144617587328,
-      "learning_rate": 0.00019999610105216783,
-      "loss": 46.0,
-      "step": 36774
-    },
-    {
-      "epoch": 2.8117055641569664,
-      "grad_norm": 0.0015517841093242168,
-      "learning_rate": 0.00019999610084005948,
-      "loss": 46.0,
-      "step": 36775
-    },
-    {
-      "epoch": 2.811782021140356,
-      "grad_norm": 0.0023132034111768007,
-      "learning_rate": 0.00019999610062794538,
-      "loss": 46.0,
-      "step": 36776
-    },
-    {
-      "epoch": 2.811858478123746,
-      "grad_norm": 0.0009892798261716962,
-      "learning_rate": 0.0001999961004158255,
-      "loss": 46.0,
-      "step": 36777
-    },
-    {
-      "epoch": 2.811934935107135,
-      "grad_norm": 0.0021904767490923405,
-      "learning_rate": 0.00019999610020369986,
-      "loss": 46.0,
-      "step": 36778
-    },
-    {
-      "epoch": 2.812011392090525,
-      "grad_norm": 0.002316446742042899,
-      "learning_rate": 0.00019999609999156844,
-      "loss": 46.0,
-      "step": 36779
-    },
-    {
-      "epoch": 2.8120878490739147,
-      "grad_norm": 0.00398360937833786,
-      "learning_rate": 0.00019999609977943125,
-      "loss": 46.0,
-      "step": 36780
-    },
-    {
-      "epoch": 2.8121643060573045,
-      "grad_norm": 0.0010359848383814096,
-      "learning_rate": 0.0001999960995672883,
-      "loss": 46.0,
-      "step": 36781
-    },
-    {
-      "epoch": 2.8122407630406943,
-      "grad_norm": 0.0034260349348187447,
-      "learning_rate": 0.00019999609935513957,
-      "loss": 46.0,
-      "step": 36782
-    },
-    {
-      "epoch": 2.812317220024084,
-      "grad_norm": 0.001337131136097014,
-      "learning_rate": 0.00019999609914298508,
-      "loss": 46.0,
-      "step": 36783
-    },
-    {
-      "epoch": 2.812393677007474,
-      "grad_norm": 0.0025051767006516457,
-      "learning_rate": 0.0001999960989308248,
-      "loss": 46.0,
-      "step": 36784
-    },
-    {
-      "epoch": 2.8124701339908635,
-      "grad_norm": 0.0019756334368139505,
-      "learning_rate": 0.0001999960987186588,
-      "loss": 46.0,
-      "step": 36785
-    },
-    {
-      "epoch": 2.812546590974253,
-      "grad_norm": 0.0014445362612605095,
-      "learning_rate": 0.00019999609850648699,
-      "loss": 46.0,
-      "step": 36786
-    },
-    {
-      "epoch": 2.8126230479576426,
-      "grad_norm": 0.0034651143942028284,
-      "learning_rate": 0.0001999960982943094,
-      "loss": 46.0,
-      "step": 36787
-    },
-    {
-      "epoch": 2.8126995049410324,
-      "grad_norm": 0.0013565976405516267,
-      "learning_rate": 0.00019999609808212608,
-      "loss": 46.0,
-      "step": 36788
-    },
-    {
-      "epoch": 2.812775961924422,
-      "grad_norm": 0.001865662052296102,
-      "learning_rate": 0.00019999609786993695,
-      "loss": 46.0,
-      "step": 36789
-    },
-    {
-      "epoch": 2.812852418907812,
-      "grad_norm": 0.0022742266301065683,
-      "learning_rate": 0.0001999960976577421,
-      "loss": 46.0,
-      "step": 36790
-    },
-    {
-      "epoch": 2.8129288758912017,
-      "grad_norm": 0.0012025375617668033,
-      "learning_rate": 0.00019999609744554144,
-      "loss": 46.0,
-      "step": 36791
-    },
-    {
-      "epoch": 2.8130053328745914,
-      "grad_norm": 0.0017907909350469708,
-      "learning_rate": 0.00019999609723333502,
-      "loss": 46.0,
-      "step": 36792
-    },
-    {
-      "epoch": 2.813081789857981,
-      "grad_norm": 0.004252856131643057,
-      "learning_rate": 0.00019999609702112283,
-      "loss": 46.0,
-      "step": 36793
-    },
-    {
-      "epoch": 2.813158246841371,
-      "grad_norm": 0.0007448701071552932,
-      "learning_rate": 0.0001999960968089049,
-      "loss": 46.0,
-      "step": 36794
-    },
-    {
-      "epoch": 2.8132347038247607,
-      "grad_norm": 0.002914773765951395,
-      "learning_rate": 0.00019999609659668115,
-      "loss": 46.0,
-      "step": 36795
-    },
-    {
-      "epoch": 2.8133111608081505,
-      "grad_norm": 0.0021829938050359488,
-      "learning_rate": 0.00019999609638445164,
-      "loss": 46.0,
-      "step": 36796
-    },
-    {
-      "epoch": 2.8133876177915402,
-      "grad_norm": 0.0011208775686100125,
-      "learning_rate": 0.0001999960961722164,
-      "loss": 46.0,
-      "step": 36797
-    },
-    {
-      "epoch": 2.81346407477493,
-      "grad_norm": 0.0008713062270544469,
-      "learning_rate": 0.00019999609595997537,
-      "loss": 46.0,
-      "step": 36798
-    },
-    {
-      "epoch": 2.8135405317583198,
-      "grad_norm": 0.004214846529066563,
-      "learning_rate": 0.00019999609574772854,
-      "loss": 46.0,
-      "step": 36799
-    },
-    {
-      "epoch": 2.813616988741709,
-      "grad_norm": 0.0017307811649516225,
-      "learning_rate": 0.000199996095535476,
-      "loss": 46.0,
-      "step": 36800
-    },
-    {
-      "epoch": 2.813693445725099,
-      "grad_norm": 0.0023016533814370632,
-      "learning_rate": 0.00019999609532321764,
-      "loss": 46.0,
-      "step": 36801
-    },
-    {
-      "epoch": 2.8137699027084886,
-      "grad_norm": 0.0015102032339200377,
-      "learning_rate": 0.00019999609511095354,
-      "loss": 46.0,
-      "step": 36802
-    },
-    {
-      "epoch": 2.8138463596918784,
-      "grad_norm": 0.003491014475002885,
-      "learning_rate": 0.00019999609489868364,
-      "loss": 46.0,
-      "step": 36803
-    },
-    {
-      "epoch": 2.813922816675268,
-      "grad_norm": 0.00448153680190444,
-      "learning_rate": 0.00019999609468640802,
-      "loss": 46.0,
-      "step": 36804
-    },
-    {
-      "epoch": 2.813999273658658,
-      "grad_norm": 0.00242490042001009,
-      "learning_rate": 0.0001999960944741266,
-      "loss": 46.0,
-      "step": 36805
-    },
-    {
-      "epoch": 2.8140757306420476,
-      "grad_norm": 0.002671569585800171,
-      "learning_rate": 0.00019999609426183938,
-      "loss": 46.0,
-      "step": 36806
-    },
-    {
-      "epoch": 2.8141521876254374,
-      "grad_norm": 0.004628873895853758,
-      "learning_rate": 0.00019999609404954644,
-      "loss": 46.0,
-      "step": 36807
-    },
-    {
-      "epoch": 2.8142286446088267,
-      "grad_norm": 0.0028726004529744387,
-      "learning_rate": 0.00019999609383724773,
-      "loss": 46.0,
-      "step": 36808
-    },
-    {
-      "epoch": 2.8143051015922165,
-      "grad_norm": 0.0015792903723195195,
-      "learning_rate": 0.00019999609362494322,
-      "loss": 46.0,
-      "step": 36809
-    },
-    {
-      "epoch": 2.8143815585756062,
-      "grad_norm": 0.0022798748686909676,
-      "learning_rate": 0.00019999609341263294,
-      "loss": 46.0,
-      "step": 36810
-    },
-    {
-      "epoch": 2.814458015558996,
-      "grad_norm": 0.002969057997688651,
-      "learning_rate": 0.0001999960932003169,
-      "loss": 46.0,
-      "step": 36811
-    },
-    {
-      "epoch": 2.8145344725423858,
-      "grad_norm": 0.0013731082435697317,
-      "learning_rate": 0.0001999960929879951,
-      "loss": 46.0,
-      "step": 36812
-    },
-    {
-      "epoch": 2.8146109295257755,
-      "grad_norm": 0.0013064262457191944,
-      "learning_rate": 0.00019999609277566755,
-      "loss": 46.0,
-      "step": 36813
-    },
-    {
-      "epoch": 2.8146873865091653,
-      "grad_norm": 0.000937616394367069,
-      "learning_rate": 0.00019999609256333422,
-      "loss": 46.0,
-      "step": 36814
-    },
-    {
-      "epoch": 2.814763843492555,
-      "grad_norm": 0.0009406793978996575,
-      "learning_rate": 0.0001999960923509951,
-      "loss": 46.0,
-      "step": 36815
-    },
-    {
-      "epoch": 2.814840300475945,
-      "grad_norm": 0.0018760168459266424,
-      "learning_rate": 0.00019999609213865023,
-      "loss": 46.0,
-      "step": 36816
-    },
-    {
-      "epoch": 2.8149167574593346,
-      "grad_norm": 0.002055566059425473,
-      "learning_rate": 0.00019999609192629958,
-      "loss": 46.0,
-      "step": 36817
-    },
-    {
-      "epoch": 2.8149932144427243,
-      "grad_norm": 0.00498898234218359,
-      "learning_rate": 0.00019999609171394316,
-      "loss": 46.0,
-      "step": 36818
-    },
-    {
-      "epoch": 2.815069671426114,
-      "grad_norm": 0.0012188117252662778,
-      "learning_rate": 0.00019999609150158097,
-      "loss": 46.0,
-      "step": 36819
-    },
-    {
-      "epoch": 2.815146128409504,
-      "grad_norm": 0.008990802802145481,
-      "learning_rate": 0.000199996091289213,
-      "loss": 46.0,
-      "step": 36820
-    },
-    {
-      "epoch": 2.815222585392893,
-      "grad_norm": 0.004068460781127214,
-      "learning_rate": 0.0001999960910768393,
-      "loss": 46.0,
-      "step": 36821
-    },
-    {
-      "epoch": 2.815299042376283,
-      "grad_norm": 0.0013540118234232068,
-      "learning_rate": 0.0001999960908644598,
-      "loss": 46.0,
-      "step": 36822
-    },
-    {
-      "epoch": 2.8153754993596727,
-      "grad_norm": 0.003944691736251116,
-      "learning_rate": 0.00019999609065207455,
-      "loss": 46.0,
-      "step": 36823
-    },
-    {
-      "epoch": 2.8154519563430624,
-      "grad_norm": 0.004211964551359415,
-      "learning_rate": 0.0001999960904396835,
-      "loss": 46.0,
-      "step": 36824
-    },
-    {
-      "epoch": 2.815528413326452,
-      "grad_norm": 0.0030869492329657078,
-      "learning_rate": 0.0001999960902272867,
-      "loss": 46.0,
-      "step": 36825
-    },
-    {
-      "epoch": 2.815604870309842,
-      "grad_norm": 0.003138667671009898,
-      "learning_rate": 0.0001999960900148841,
-      "loss": 46.0,
-      "step": 36826
-    },
-    {
-      "epoch": 2.8156813272932317,
-      "grad_norm": 0.0030309916473925114,
-      "learning_rate": 0.00019999608980247579,
-      "loss": 46.0,
-      "step": 36827
-    },
-    {
-      "epoch": 2.8157577842766215,
-      "grad_norm": 0.0012048023054376245,
-      "learning_rate": 0.00019999608959006166,
-      "loss": 46.0,
-      "step": 36828
-    },
-    {
-      "epoch": 2.8158342412600113,
-      "grad_norm": 0.0014024564297869802,
-      "learning_rate": 0.00019999608937764182,
-      "loss": 46.0,
-      "step": 36829
-    },
-    {
-      "epoch": 2.8159106982434006,
-      "grad_norm": 0.004322107881307602,
-      "learning_rate": 0.00019999608916521615,
-      "loss": 46.0,
-      "step": 36830
-    },
-    {
-      "epoch": 2.8159871552267903,
-      "grad_norm": 0.0007613147026859224,
-      "learning_rate": 0.00019999608895278473,
-      "loss": 46.0,
-      "step": 36831
-    },
-    {
-      "epoch": 2.81606361221018,
-      "grad_norm": 0.0016653668135404587,
-      "learning_rate": 0.00019999608874034754,
-      "loss": 46.0,
-      "step": 36832
-    },
-    {
-      "epoch": 2.81614006919357,
-      "grad_norm": 0.004325645510107279,
-      "learning_rate": 0.0001999960885279046,
-      "loss": 46.0,
-      "step": 36833
-    },
-    {
-      "epoch": 2.8162165261769596,
-      "grad_norm": 0.0007937981863506138,
-      "learning_rate": 0.00019999608831545586,
-      "loss": 46.0,
-      "step": 36834
-    },
-    {
-      "epoch": 2.8162929831603494,
-      "grad_norm": 0.0008365145185962319,
-      "learning_rate": 0.00019999608810300138,
-      "loss": 46.0,
-      "step": 36835
-    },
-    {
-      "epoch": 2.816369440143739,
-      "grad_norm": 0.0019224671414121985,
-      "learning_rate": 0.0001999960878905411,
-      "loss": 46.0,
-      "step": 36836
-    },
-    {
-      "epoch": 2.816445897127129,
-      "grad_norm": 0.004457344301044941,
-      "learning_rate": 0.00019999608767807506,
-      "loss": 46.0,
-      "step": 36837
-    },
-    {
-      "epoch": 2.8165223541105187,
-      "grad_norm": 0.0008685345528647304,
-      "learning_rate": 0.00019999608746560328,
-      "loss": 46.0,
-      "step": 36838
-    },
-    {
-      "epoch": 2.8165988110939084,
-      "grad_norm": 0.0008285320946015418,
-      "learning_rate": 0.00019999608725312568,
-      "loss": 46.0,
-      "step": 36839
-    },
-    {
-      "epoch": 2.816675268077298,
-      "grad_norm": 0.0061916024424135685,
-      "learning_rate": 0.00019999608704064235,
-      "loss": 46.0,
-      "step": 36840
-    },
-    {
-      "epoch": 2.816751725060688,
-      "grad_norm": 0.0028467231895774603,
-      "learning_rate": 0.00019999608682815323,
-      "loss": 46.0,
-      "step": 36841
-    },
-    {
-      "epoch": 2.8168281820440777,
-      "grad_norm": 0.0021147585939615965,
-      "learning_rate": 0.00019999608661565836,
-      "loss": 46.0,
-      "step": 36842
-    },
-    {
-      "epoch": 2.816904639027467,
-      "grad_norm": 0.003130111610516906,
-      "learning_rate": 0.00019999608640315771,
-      "loss": 46.0,
-      "step": 36843
-    },
-    {
-      "epoch": 2.8169810960108568,
-      "grad_norm": 0.0020714327692985535,
-      "learning_rate": 0.00019999608619065132,
-      "loss": 46.0,
-      "step": 36844
-    },
-    {
-      "epoch": 2.8170575529942465,
-      "grad_norm": 0.0026294696144759655,
-      "learning_rate": 0.00019999608597813913,
-      "loss": 46.0,
-      "step": 36845
-    },
-    {
-      "epoch": 2.8171340099776363,
-      "grad_norm": 0.0007815816788934171,
-      "learning_rate": 0.00019999608576562117,
-      "loss": 46.0,
-      "step": 36846
-    },
-    {
-      "epoch": 2.817210466961026,
-      "grad_norm": 0.0010007995879277587,
-      "learning_rate": 0.00019999608555309743,
-      "loss": 46.0,
-      "step": 36847
-    },
-    {
-      "epoch": 2.817286923944416,
-      "grad_norm": 0.002334407763555646,
-      "learning_rate": 0.00019999608534056792,
-      "loss": 46.0,
-      "step": 36848
-    },
-    {
-      "epoch": 2.8173633809278056,
-      "grad_norm": 0.0010643490822985768,
-      "learning_rate": 0.00019999608512803266,
-      "loss": 46.0,
-      "step": 36849
-    },
-    {
-      "epoch": 2.8174398379111953,
-      "grad_norm": 0.002337804762646556,
-      "learning_rate": 0.00019999608491549163,
-      "loss": 46.0,
-      "step": 36850
-    },
-    {
-      "epoch": 2.8175162948945847,
-      "grad_norm": 0.002812929917126894,
-      "learning_rate": 0.00019999608470294485,
-      "loss": 46.0,
-      "step": 36851
-    },
-    {
-      "epoch": 2.8175927518779744,
-      "grad_norm": 0.0028979452326893806,
-      "learning_rate": 0.00019999608449039228,
-      "loss": 46.0,
-      "step": 36852
-    },
-    {
-      "epoch": 2.817669208861364,
-      "grad_norm": 0.002232844242826104,
-      "learning_rate": 0.00019999608427783395,
-      "loss": 46.0,
-      "step": 36853
-    },
-    {
-      "epoch": 2.817745665844754,
-      "grad_norm": 0.004402905702590942,
-      "learning_rate": 0.00019999608406526983,
-      "loss": 46.0,
-      "step": 36854
-    },
-    {
-      "epoch": 2.8178221228281437,
-      "grad_norm": 0.001040091272443533,
-      "learning_rate": 0.00019999608385269993,
-      "loss": 46.0,
-      "step": 36855
-    },
-    {
-      "epoch": 2.8178985798115335,
-      "grad_norm": 0.0035972122568637133,
-      "learning_rate": 0.0001999960836401243,
-      "loss": 46.0,
-      "step": 36856
-    },
-    {
-      "epoch": 2.8179750367949232,
-      "grad_norm": 0.001221560756675899,
-      "learning_rate": 0.00019999608342754287,
-      "loss": 46.0,
-      "step": 36857
-    },
-    {
-      "epoch": 2.818051493778313,
-      "grad_norm": 0.005079397466033697,
-      "learning_rate": 0.00019999608321495568,
-      "loss": 46.0,
-      "step": 36858
-    },
-    {
-      "epoch": 2.8181279507617027,
-      "grad_norm": 0.0030243832152336836,
-      "learning_rate": 0.00019999608300236271,
-      "loss": 46.0,
-      "step": 36859
-    },
-    {
-      "epoch": 2.8182044077450925,
-      "grad_norm": 0.0016896920278668404,
-      "learning_rate": 0.000199996082789764,
-      "loss": 46.0,
-      "step": 36860
-    },
-    {
-      "epoch": 2.8182808647284823,
-      "grad_norm": 0.0021322863176465034,
-      "learning_rate": 0.0001999960825771595,
-      "loss": 46.0,
-      "step": 36861
-    },
-    {
-      "epoch": 2.818357321711872,
-      "grad_norm": 0.001678258995525539,
-      "learning_rate": 0.00019999608236454926,
-      "loss": 46.0,
-      "step": 36862
-    },
-    {
-      "epoch": 2.818433778695262,
-      "grad_norm": 0.0008996687829494476,
-      "learning_rate": 0.0001999960821519332,
-      "loss": 46.0,
-      "step": 36863
-    },
-    {
-      "epoch": 2.8185102356786516,
-      "grad_norm": 0.01777329295873642,
-      "learning_rate": 0.0001999960819393114,
-      "loss": 46.0,
-      "step": 36864
-    },
-    {
-      "epoch": 2.818586692662041,
-      "grad_norm": 0.004625699017196894,
-      "learning_rate": 0.00019999608172668385,
-      "loss": 46.0,
-      "step": 36865
-    },
-    {
-      "epoch": 2.8186631496454306,
-      "grad_norm": 0.0045891208574175835,
-      "learning_rate": 0.0001999960815140505,
-      "loss": 46.0,
-      "step": 36866
-    },
-    {
-      "epoch": 2.8187396066288204,
-      "grad_norm": 0.0015666260151192546,
-      "learning_rate": 0.00019999608130141138,
-      "loss": 46.0,
-      "step": 36867
-    },
-    {
-      "epoch": 2.81881606361221,
-      "grad_norm": 0.01152582187205553,
-      "learning_rate": 0.0001999960810887665,
-      "loss": 46.0,
-      "step": 36868
-    },
-    {
-      "epoch": 2.8188925205956,
-      "grad_norm": 0.001934061641804874,
-      "learning_rate": 0.00019999608087611586,
-      "loss": 46.0,
-      "step": 36869
-    },
-    {
-      "epoch": 2.8189689775789897,
-      "grad_norm": 0.003188719740137458,
-      "learning_rate": 0.00019999608066345945,
-      "loss": 46.0,
-      "step": 36870
-    },
-    {
-      "epoch": 2.8190454345623794,
-      "grad_norm": 0.0018322645919397473,
-      "learning_rate": 0.00019999608045079726,
-      "loss": 46.0,
-      "step": 36871
-    },
-    {
-      "epoch": 2.819121891545769,
-      "grad_norm": 0.00248602288775146,
-      "learning_rate": 0.0001999960802381293,
-      "loss": 46.0,
-      "step": 36872
-    },
-    {
-      "epoch": 2.8191983485291585,
-      "grad_norm": 0.005531106609851122,
-      "learning_rate": 0.00019999608002545558,
-      "loss": 46.0,
-      "step": 36873
-    },
-    {
-      "epoch": 2.8192748055125483,
-      "grad_norm": 0.005437852814793587,
-      "learning_rate": 0.00019999607981277607,
-      "loss": 46.0,
-      "step": 36874
-    },
-    {
-      "epoch": 2.819351262495938,
-      "grad_norm": 0.00439549470320344,
-      "learning_rate": 0.00019999607960009082,
-      "loss": 46.0,
-      "step": 36875
-    },
-    {
-      "epoch": 2.819427719479328,
-      "grad_norm": 0.0016947225667536259,
-      "learning_rate": 0.00019999607938739979,
-      "loss": 46.0,
-      "step": 36876
-    },
-    {
-      "epoch": 2.8195041764627176,
-      "grad_norm": 0.004166991449892521,
-      "learning_rate": 0.00019999607917470298,
-      "loss": 46.0,
-      "step": 36877
-    },
-    {
-      "epoch": 2.8195806334461073,
-      "grad_norm": 0.0014405949041247368,
-      "learning_rate": 0.0001999960789620004,
-      "loss": 46.0,
-      "step": 36878
-    },
-    {
-      "epoch": 2.819657090429497,
-      "grad_norm": 0.0013116549234837294,
-      "learning_rate": 0.00019999607874929208,
-      "loss": 46.0,
-      "step": 36879
-    },
-    {
-      "epoch": 2.819733547412887,
-      "grad_norm": 0.0038978285156190395,
-      "learning_rate": 0.00019999607853657796,
-      "loss": 46.0,
-      "step": 36880
-    },
-    {
-      "epoch": 2.8198100043962766,
-      "grad_norm": 0.0039982362650334835,
-      "learning_rate": 0.0001999960783238581,
-      "loss": 46.0,
-      "step": 36881
-    },
-    {
-      "epoch": 2.8198864613796664,
-      "grad_norm": 0.0016772805247455835,
-      "learning_rate": 0.00019999607811113244,
-      "loss": 46.0,
-      "step": 36882
-    },
-    {
-      "epoch": 2.819962918363056,
-      "grad_norm": 0.0008836925262585282,
-      "learning_rate": 0.000199996077898401,
-      "loss": 46.0,
-      "step": 36883
-    },
-    {
-      "epoch": 2.820039375346446,
-      "grad_norm": 0.0008074045181274414,
-      "learning_rate": 0.00019999607768566384,
-      "loss": 46.0,
-      "step": 36884
-    },
-    {
-      "epoch": 2.8201158323298356,
-      "grad_norm": 0.004422673024237156,
-      "learning_rate": 0.00019999607747292087,
-      "loss": 46.0,
-      "step": 36885
-    },
-    {
-      "epoch": 2.8201922893132254,
-      "grad_norm": 0.0016577758360654116,
-      "learning_rate": 0.00019999607726017216,
-      "loss": 46.0,
-      "step": 36886
-    },
-    {
-      "epoch": 2.8202687462966147,
-      "grad_norm": 0.004137158859521151,
-      "learning_rate": 0.00019999607704741765,
-      "loss": 46.0,
-      "step": 36887
-    },
-    {
-      "epoch": 2.8203452032800045,
-      "grad_norm": 0.0015736386412754655,
-      "learning_rate": 0.00019999607683465737,
-      "loss": 46.0,
-      "step": 36888
-    },
-    {
-      "epoch": 2.8204216602633942,
-      "grad_norm": 0.0013831130927428603,
-      "learning_rate": 0.00019999607662189134,
-      "loss": 46.0,
-      "step": 36889
-    },
-    {
-      "epoch": 2.820498117246784,
-      "grad_norm": 0.00231073796749115,
-      "learning_rate": 0.00019999607640911954,
-      "loss": 46.0,
-      "step": 36890
-    },
-    {
-      "epoch": 2.8205745742301738,
-      "grad_norm": 0.0033042451832443476,
-      "learning_rate": 0.00019999607619634196,
-      "loss": 46.0,
-      "step": 36891
-    },
-    {
-      "epoch": 2.8206510312135635,
-      "grad_norm": 0.003823449369519949,
-      "learning_rate": 0.00019999607598355864,
-      "loss": 46.0,
-      "step": 36892
-    },
-    {
-      "epoch": 2.8207274881969533,
-      "grad_norm": 0.002630861708894372,
-      "learning_rate": 0.00019999607577076954,
-      "loss": 46.0,
-      "step": 36893
-    },
-    {
-      "epoch": 2.820803945180343,
-      "grad_norm": 0.0011717822635546327,
-      "learning_rate": 0.00019999607555797467,
-      "loss": 46.0,
-      "step": 36894
-    },
-    {
-      "epoch": 2.8208804021637324,
-      "grad_norm": 0.0011562578147277236,
-      "learning_rate": 0.000199996075345174,
-      "loss": 46.0,
-      "step": 36895
-    },
-    {
-      "epoch": 2.820956859147122,
-      "grad_norm": 0.001471058581955731,
-      "learning_rate": 0.00019999607513236759,
-      "loss": 46.0,
-      "step": 36896
-    },
-    {
-      "epoch": 2.821033316130512,
-      "grad_norm": 0.004946605302393436,
-      "learning_rate": 0.0001999960749195554,
-      "loss": 46.0,
-      "step": 36897
-    },
-    {
-      "epoch": 2.8211097731139017,
-      "grad_norm": 0.002298481296747923,
-      "learning_rate": 0.00019999607470673743,
-      "loss": 46.0,
-      "step": 36898
-    },
-    {
-      "epoch": 2.8211862300972914,
-      "grad_norm": 0.0019355490803718567,
-      "learning_rate": 0.00019999607449391372,
-      "loss": 46.0,
-      "step": 36899
-    },
-    {
-      "epoch": 2.821262687080681,
-      "grad_norm": 0.0020304580684751272,
-      "learning_rate": 0.00019999607428108421,
-      "loss": 46.0,
-      "step": 36900
-    },
-    {
-      "epoch": 2.821339144064071,
-      "grad_norm": 0.003176880767568946,
-      "learning_rate": 0.00019999607406824896,
-      "loss": 46.0,
-      "step": 36901
-    },
-    {
-      "epoch": 2.8214156010474607,
-      "grad_norm": 0.003486047266051173,
-      "learning_rate": 0.00019999607385540793,
-      "loss": 46.0,
-      "step": 36902
-    },
-    {
-      "epoch": 2.8214920580308505,
-      "grad_norm": 0.0015230627031996846,
-      "learning_rate": 0.00019999607364256113,
-      "loss": 46.0,
-      "step": 36903
-    },
-    {
-      "epoch": 2.82156851501424,
-      "grad_norm": 0.0025447842199355364,
-      "learning_rate": 0.00019999607342970858,
-      "loss": 46.0,
-      "step": 36904
-    },
-    {
-      "epoch": 2.82164497199763,
-      "grad_norm": 0.0013891375856474042,
-      "learning_rate": 0.0001999960732168502,
-      "loss": 46.0,
-      "step": 36905
-    },
-    {
-      "epoch": 2.8217214289810197,
-      "grad_norm": 0.0007464977097697556,
-      "learning_rate": 0.0001999960730039861,
-      "loss": 46.0,
-      "step": 36906
-    },
-    {
-      "epoch": 2.8217978859644095,
-      "grad_norm": 0.00298857968300581,
-      "learning_rate": 0.00019999607279111623,
-      "loss": 46.0,
-      "step": 36907
-    },
-    {
-      "epoch": 2.8218743429477993,
-      "grad_norm": 0.0015343863051384687,
-      "learning_rate": 0.0001999960725782406,
-      "loss": 46.0,
-      "step": 36908
-    },
-    {
-      "epoch": 2.8219507999311886,
-      "grad_norm": 0.001058666268363595,
-      "learning_rate": 0.00019999607236535915,
-      "loss": 46.0,
-      "step": 36909
-    },
-    {
-      "epoch": 2.8220272569145783,
-      "grad_norm": 0.0016679898835718632,
-      "learning_rate": 0.00019999607215247199,
-      "loss": 46.0,
-      "step": 36910
-    },
-    {
-      "epoch": 2.822103713897968,
-      "grad_norm": 0.0016461880877614021,
-      "learning_rate": 0.00019999607193957902,
-      "loss": 46.0,
-      "step": 36911
-    },
-    {
-      "epoch": 2.822180170881358,
-      "grad_norm": 0.0026223361492156982,
-      "learning_rate": 0.0001999960717266803,
-      "loss": 46.0,
-      "step": 36912
-    },
-    {
-      "epoch": 2.8222566278647476,
-      "grad_norm": 0.0017350807320326567,
-      "learning_rate": 0.0001999960715137758,
-      "loss": 46.0,
-      "step": 36913
-    },
-    {
-      "epoch": 2.8223330848481374,
-      "grad_norm": 0.0015073895920068026,
-      "learning_rate": 0.00019999607130086555,
-      "loss": 46.0,
-      "step": 36914
-    },
-    {
-      "epoch": 2.822409541831527,
-      "grad_norm": 0.003138087224215269,
-      "learning_rate": 0.00019999607108794952,
-      "loss": 46.0,
-      "step": 36915
-    },
-    {
-      "epoch": 2.822485998814917,
-      "grad_norm": 0.002158522605895996,
-      "learning_rate": 0.0001999960708750277,
-      "loss": 46.0,
-      "step": 36916
-    },
-    {
-      "epoch": 2.8225624557983062,
-      "grad_norm": 0.009770886972546577,
-      "learning_rate": 0.00019999607066210014,
-      "loss": 46.0,
-      "step": 36917
-    },
-    {
-      "epoch": 2.822638912781696,
-      "grad_norm": 0.002159107243642211,
-      "learning_rate": 0.0001999960704491668,
-      "loss": 46.0,
-      "step": 36918
-    },
-    {
-      "epoch": 2.8227153697650857,
-      "grad_norm": 0.0008081173291429877,
-      "learning_rate": 0.0001999960702362277,
-      "loss": 46.0,
-      "step": 36919
-    },
-    {
-      "epoch": 2.8227918267484755,
-      "grad_norm": 0.005670949351042509,
-      "learning_rate": 0.0001999960700232828,
-      "loss": 46.0,
-      "step": 36920
-    },
-    {
-      "epoch": 2.8228682837318653,
-      "grad_norm": 0.0037465994246304035,
-      "learning_rate": 0.00019999606981033216,
-      "loss": 46.0,
-      "step": 36921
-    },
-    {
-      "epoch": 2.822944740715255,
-      "grad_norm": 0.0007225925219245255,
-      "learning_rate": 0.00019999606959737574,
-      "loss": 46.0,
-      "step": 36922
-    },
-    {
-      "epoch": 2.823021197698645,
-      "grad_norm": 0.00275416555814445,
-      "learning_rate": 0.00019999606938441355,
-      "loss": 46.0,
-      "step": 36923
-    },
-    {
-      "epoch": 2.8230976546820346,
-      "grad_norm": 0.001614834414795041,
-      "learning_rate": 0.0001999960691714456,
-      "loss": 46.0,
-      "step": 36924
-    },
-    {
-      "epoch": 2.8231741116654243,
-      "grad_norm": 0.00276213139295578,
-      "learning_rate": 0.00019999606895847188,
-      "loss": 46.0,
-      "step": 36925
-    },
-    {
-      "epoch": 2.823250568648814,
-      "grad_norm": 0.0009544197237119079,
-      "learning_rate": 0.00019999606874549237,
-      "loss": 46.0,
-      "step": 36926
-    },
-    {
-      "epoch": 2.823327025632204,
-      "grad_norm": 0.0032150084152817726,
-      "learning_rate": 0.00019999606853250712,
-      "loss": 46.0,
-      "step": 36927
-    },
-    {
-      "epoch": 2.8234034826155936,
-      "grad_norm": 0.0021593982819467783,
-      "learning_rate": 0.00019999606831951606,
-      "loss": 46.0,
-      "step": 36928
-    },
-    {
-      "epoch": 2.8234799395989834,
-      "grad_norm": 0.0020897244103252888,
-      "learning_rate": 0.00019999606810651929,
-      "loss": 46.0,
-      "step": 36929
-    },
-    {
-      "epoch": 2.823556396582373,
-      "grad_norm": 0.0010092228185385466,
-      "learning_rate": 0.0001999960678935167,
-      "loss": 46.0,
-      "step": 36930
-    },
-    {
-      "epoch": 2.8236328535657624,
-      "grad_norm": 0.0032900257501751184,
-      "learning_rate": 0.00019999606768050836,
-      "loss": 46.0,
-      "step": 36931
-    },
-    {
-      "epoch": 2.823709310549152,
-      "grad_norm": 0.004979650489985943,
-      "learning_rate": 0.00019999606746749424,
-      "loss": 46.0,
-      "step": 36932
-    },
-    {
-      "epoch": 2.823785767532542,
-      "grad_norm": 0.002661351813003421,
-      "learning_rate": 0.00019999606725447437,
-      "loss": 46.0,
-      "step": 36933
-    },
-    {
-      "epoch": 2.8238622245159317,
-      "grad_norm": 0.0019578225910663605,
-      "learning_rate": 0.00019999606704144873,
-      "loss": 46.0,
-      "step": 36934
-    },
-    {
-      "epoch": 2.8239386814993215,
-      "grad_norm": 0.0016328799538314342,
-      "learning_rate": 0.0001999960668284173,
-      "loss": 46.0,
-      "step": 36935
-    },
-    {
-      "epoch": 2.8240151384827112,
-      "grad_norm": 0.0033278141636401415,
-      "learning_rate": 0.00019999606661538012,
-      "loss": 46.0,
-      "step": 36936
-    },
-    {
-      "epoch": 2.824091595466101,
-      "grad_norm": 0.0024550892412662506,
-      "learning_rate": 0.00019999606640233716,
-      "loss": 46.0,
-      "step": 36937
-    },
-    {
-      "epoch": 2.8241680524494908,
-      "grad_norm": 0.0023666478227823973,
-      "learning_rate": 0.00019999606618928843,
-      "loss": 46.0,
-      "step": 36938
-    },
-    {
-      "epoch": 2.82424450943288,
-      "grad_norm": 0.0009375037625432014,
-      "learning_rate": 0.00019999606597623394,
-      "loss": 46.0,
-      "step": 36939
-    },
-    {
-      "epoch": 2.82432096641627,
-      "grad_norm": 0.0024318823125213385,
-      "learning_rate": 0.0001999960657631737,
-      "loss": 46.0,
-      "step": 36940
-    },
-    {
-      "epoch": 2.8243974233996596,
-      "grad_norm": 0.0037423213943839073,
-      "learning_rate": 0.00019999606555010763,
-      "loss": 46.0,
-      "step": 36941
-    },
-    {
-      "epoch": 2.8244738803830494,
-      "grad_norm": 0.001682156347669661,
-      "learning_rate": 0.00019999606533703586,
-      "loss": 46.0,
-      "step": 36942
-    },
-    {
-      "epoch": 2.824550337366439,
-      "grad_norm": 0.002599930390715599,
-      "learning_rate": 0.00019999606512395828,
-      "loss": 46.0,
-      "step": 36943
-    },
-    {
-      "epoch": 2.824626794349829,
-      "grad_norm": 0.002183648757636547,
-      "learning_rate": 0.00019999606491087493,
-      "loss": 46.0,
-      "step": 36944
-    },
-    {
-      "epoch": 2.8247032513332186,
-      "grad_norm": 0.0031466891523450613,
-      "learning_rate": 0.0001999960646977858,
-      "loss": 46.0,
-      "step": 36945
-    },
-    {
-      "epoch": 2.8247797083166084,
-      "grad_norm": 0.0038752886466681957,
-      "learning_rate": 0.00019999606448469094,
-      "loss": 46.0,
-      "step": 36946
-    },
-    {
-      "epoch": 2.824856165299998,
-      "grad_norm": 0.0023581795394420624,
-      "learning_rate": 0.0001999960642715903,
-      "loss": 46.0,
-      "step": 36947
-    },
-    {
-      "epoch": 2.824932622283388,
-      "grad_norm": 0.0068861329928040504,
-      "learning_rate": 0.0001999960640584839,
-      "loss": 46.0,
-      "step": 36948
-    },
-    {
-      "epoch": 2.8250090792667777,
-      "grad_norm": 0.0006696869386360049,
-      "learning_rate": 0.00019999606384537167,
-      "loss": 46.0,
-      "step": 36949
-    },
-    {
-      "epoch": 2.8250855362501675,
-      "grad_norm": 0.0013573196483775973,
-      "learning_rate": 0.00019999606363225374,
-      "loss": 46.0,
-      "step": 36950
-    },
-    {
-      "epoch": 2.825161993233557,
-      "grad_norm": 0.002307785674929619,
-      "learning_rate": 0.00019999606341913003,
-      "loss": 46.0,
-      "step": 36951
-    },
-    {
-      "epoch": 2.8252384502169465,
-      "grad_norm": 0.003991785924881697,
-      "learning_rate": 0.00019999606320600052,
-      "loss": 46.0,
-      "step": 36952
-    },
-    {
-      "epoch": 2.8253149072003363,
-      "grad_norm": 0.004686118569225073,
-      "learning_rate": 0.00019999606299286527,
-      "loss": 46.0,
-      "step": 36953
-    },
-    {
-      "epoch": 2.825391364183726,
-      "grad_norm": 0.0032807081006467342,
-      "learning_rate": 0.00019999606277972424,
-      "loss": 46.0,
-      "step": 36954
-    },
-    {
-      "epoch": 2.825467821167116,
-      "grad_norm": 0.0026232386007905006,
-      "learning_rate": 0.0001999960625665774,
-      "loss": 46.0,
-      "step": 36955
-    },
-    {
-      "epoch": 2.8255442781505056,
-      "grad_norm": 0.00046751167974434793,
-      "learning_rate": 0.00019999606235342486,
-      "loss": 46.0,
-      "step": 36956
-    },
-    {
-      "epoch": 2.8256207351338953,
-      "grad_norm": 0.0012950844829902053,
-      "learning_rate": 0.0001999960621402665,
-      "loss": 46.0,
-      "step": 36957
-    },
-    {
-      "epoch": 2.825697192117285,
-      "grad_norm": 0.001847826293669641,
-      "learning_rate": 0.00019999606192710242,
-      "loss": 46.0,
-      "step": 36958
-    },
-    {
-      "epoch": 2.825773649100675,
-      "grad_norm": 0.003466159338131547,
-      "learning_rate": 0.00019999606171393252,
-      "loss": 46.0,
-      "step": 36959
-    },
-    {
-      "epoch": 2.8258501060840646,
-      "grad_norm": 0.002881294582039118,
-      "learning_rate": 0.00019999606150075688,
-      "loss": 46.0,
-      "step": 36960
-    },
-    {
-      "epoch": 2.825926563067454,
-      "grad_norm": 0.002187170786783099,
-      "learning_rate": 0.00019999606128757547,
-      "loss": 46.0,
-      "step": 36961
-    },
-    {
-      "epoch": 2.8260030200508437,
-      "grad_norm": 0.003727750387042761,
-      "learning_rate": 0.00019999606107438828,
-      "loss": 46.0,
-      "step": 36962
-    },
-    {
-      "epoch": 2.8260794770342335,
-      "grad_norm": 0.0057577756233513355,
-      "learning_rate": 0.00019999606086119532,
-      "loss": 46.0,
-      "step": 36963
-    },
-    {
-      "epoch": 2.826155934017623,
-      "grad_norm": 0.002422747667878866,
-      "learning_rate": 0.0001999960606479966,
-      "loss": 46.0,
-      "step": 36964
-    },
-    {
-      "epoch": 2.826232391001013,
-      "grad_norm": 0.00204710615798831,
-      "learning_rate": 0.0001999960604347921,
-      "loss": 46.0,
-      "step": 36965
-    },
-    {
-      "epoch": 2.8263088479844027,
-      "grad_norm": 0.0007467864197678864,
-      "learning_rate": 0.00019999606022158185,
-      "loss": 46.0,
-      "step": 36966
-    },
-    {
-      "epoch": 2.8263853049677925,
-      "grad_norm": 0.0014958487590774894,
-      "learning_rate": 0.00019999606000836582,
-      "loss": 46.0,
-      "step": 36967
-    },
-    {
-      "epoch": 2.8264617619511823,
-      "grad_norm": 0.001446136855520308,
-      "learning_rate": 0.000199996059795144,
-      "loss": 46.0,
-      "step": 36968
-    },
-    {
-      "epoch": 2.826538218934572,
-      "grad_norm": 0.004083353094756603,
-      "learning_rate": 0.00019999605958191644,
-      "loss": 46.0,
-      "step": 36969
-    },
-    {
-      "epoch": 2.826614675917962,
-      "grad_norm": 0.0013256021775305271,
-      "learning_rate": 0.0001999960593686831,
-      "loss": 46.0,
-      "step": 36970
-    },
-    {
-      "epoch": 2.8266911329013515,
-      "grad_norm": 0.003115095430985093,
-      "learning_rate": 0.000199996059155444,
-      "loss": 46.0,
-      "step": 36971
-    },
-    {
-      "epoch": 2.8267675898847413,
-      "grad_norm": 0.002239291789010167,
-      "learning_rate": 0.0001999960589421991,
-      "loss": 46.0,
-      "step": 36972
-    },
-    {
-      "epoch": 2.826844046868131,
-      "grad_norm": 0.0033848669845610857,
-      "learning_rate": 0.00019999605872894844,
-      "loss": 46.0,
-      "step": 36973
-    },
-    {
-      "epoch": 2.8269205038515204,
-      "grad_norm": 0.0033135893754661083,
-      "learning_rate": 0.00019999605851569205,
-      "loss": 46.0,
-      "step": 36974
-    },
-    {
-      "epoch": 2.82699696083491,
-      "grad_norm": 0.0023286587093025446,
-      "learning_rate": 0.00019999605830242983,
-      "loss": 46.0,
-      "step": 36975
-    },
-    {
-      "epoch": 2.8270734178183,
-      "grad_norm": 0.003982891794294119,
-      "learning_rate": 0.0001999960580891619,
-      "loss": 46.0,
-      "step": 36976
-    },
-    {
-      "epoch": 2.8271498748016897,
-      "grad_norm": 0.002888122806325555,
-      "learning_rate": 0.00019999605787588816,
-      "loss": 46.0,
-      "step": 36977
-    },
-    {
-      "epoch": 2.8272263317850794,
-      "grad_norm": 0.0008295307052321732,
-      "learning_rate": 0.00019999605766260868,
-      "loss": 46.0,
-      "step": 36978
-    },
-    {
-      "epoch": 2.827302788768469,
-      "grad_norm": 0.0015853594522923231,
-      "learning_rate": 0.0001999960574493234,
-      "loss": 46.0,
-      "step": 36979
-    },
-    {
-      "epoch": 2.827379245751859,
-      "grad_norm": 0.002653133822605014,
-      "learning_rate": 0.00019999605723603237,
-      "loss": 46.0,
-      "step": 36980
-    },
-    {
-      "epoch": 2.8274557027352487,
-      "grad_norm": 0.0011382754892110825,
-      "learning_rate": 0.00019999605702273557,
-      "loss": 46.0,
-      "step": 36981
-    },
-    {
-      "epoch": 2.827532159718638,
-      "grad_norm": 0.0011233199620619416,
-      "learning_rate": 0.000199996056809433,
-      "loss": 46.0,
-      "step": 36982
-    },
-    {
-      "epoch": 2.827608616702028,
-      "grad_norm": 0.004417513031512499,
-      "learning_rate": 0.00019999605659612465,
-      "loss": 46.0,
-      "step": 36983
-    },
-    {
-      "epoch": 2.8276850736854175,
-      "grad_norm": 0.0008167651831172407,
-      "learning_rate": 0.00019999605638281056,
-      "loss": 46.0,
-      "step": 36984
-    },
-    {
-      "epoch": 2.8277615306688073,
-      "grad_norm": 0.001227772911079228,
-      "learning_rate": 0.0001999960561694907,
-      "loss": 46.0,
-      "step": 36985
-    },
-    {
-      "epoch": 2.827837987652197,
-      "grad_norm": 0.002584100002422929,
-      "learning_rate": 0.00019999605595616502,
-      "loss": 46.0,
-      "step": 36986
-    },
-    {
-      "epoch": 2.827914444635587,
-      "grad_norm": 0.0020683251786977053,
-      "learning_rate": 0.0001999960557428336,
-      "loss": 46.0,
-      "step": 36987
-    },
-    {
-      "epoch": 2.8279909016189766,
-      "grad_norm": 0.0029860222712159157,
-      "learning_rate": 0.00019999605552949642,
-      "loss": 46.0,
-      "step": 36988
-    },
-    {
-      "epoch": 2.8280673586023664,
-      "grad_norm": 0.0030127670615911484,
-      "learning_rate": 0.00019999605531615346,
-      "loss": 46.0,
-      "step": 36989
-    },
-    {
-      "epoch": 2.828143815585756,
-      "grad_norm": 0.004143049940466881,
-      "learning_rate": 0.00019999605510280475,
-      "loss": 46.0,
-      "step": 36990
-    },
-    {
-      "epoch": 2.828220272569146,
-      "grad_norm": 0.001784849795512855,
-      "learning_rate": 0.00019999605488945024,
-      "loss": 46.0,
-      "step": 36991
-    },
-    {
-      "epoch": 2.8282967295525356,
-      "grad_norm": 0.0032061522360891104,
-      "learning_rate": 0.00019999605467609,
-      "loss": 46.0,
-      "step": 36992
-    },
-    {
-      "epoch": 2.8283731865359254,
-      "grad_norm": 0.000825317285489291,
-      "learning_rate": 0.00019999605446272396,
-      "loss": 46.0,
-      "step": 36993
-    },
-    {
-      "epoch": 2.828449643519315,
-      "grad_norm": 0.0030062992591410875,
-      "learning_rate": 0.00019999605424935214,
-      "loss": 46.0,
-      "step": 36994
-    },
-    {
-      "epoch": 2.828526100502705,
-      "grad_norm": 0.003917657770216465,
-      "learning_rate": 0.0001999960540359746,
-      "loss": 46.0,
-      "step": 36995
-    },
-    {
-      "epoch": 2.8286025574860942,
-      "grad_norm": 0.001220331760123372,
-      "learning_rate": 0.00019999605382259127,
-      "loss": 46.0,
-      "step": 36996
-    },
-    {
-      "epoch": 2.828679014469484,
-      "grad_norm": 0.0008410522132180631,
-      "learning_rate": 0.00019999605360920212,
-      "loss": 46.0,
-      "step": 36997
-    },
-    {
-      "epoch": 2.8287554714528738,
-      "grad_norm": 0.0019293116638436913,
-      "learning_rate": 0.00019999605339580728,
-      "loss": 46.0,
-      "step": 36998
-    },
-    {
-      "epoch": 2.8288319284362635,
-      "grad_norm": 0.004157381597906351,
-      "learning_rate": 0.00019999605318240659,
-      "loss": 46.0,
-      "step": 36999
-    },
-    {
-      "epoch": 2.8289083854196533,
-      "grad_norm": 0.005885671824216843,
-      "learning_rate": 0.0001999960529690002,
-      "loss": 46.0,
-      "step": 37000
-    },
-    {
-      "epoch": 2.828984842403043,
-      "grad_norm": 0.002693883143365383,
-      "learning_rate": 0.000199996052755588,
-      "loss": 46.0,
-      "step": 37001
-    },
-    {
-      "epoch": 2.829061299386433,
-      "grad_norm": 0.0042720879428088665,
-      "learning_rate": 0.00019999605254217005,
-      "loss": 46.0,
-      "step": 37002
-    },
-    {
-      "epoch": 2.8291377563698226,
-      "grad_norm": 0.00464756740257144,
-      "learning_rate": 0.00019999605232874634,
-      "loss": 46.0,
-      "step": 37003
-    },
-    {
-      "epoch": 2.829214213353212,
-      "grad_norm": 0.00296876672655344,
-      "learning_rate": 0.00019999605211531684,
-      "loss": 46.0,
-      "step": 37004
-    },
-    {
-      "epoch": 2.8292906703366016,
-      "grad_norm": 0.0015118091832846403,
-      "learning_rate": 0.00019999605190188158,
-      "loss": 46.0,
-      "step": 37005
-    },
-    {
-      "epoch": 2.8293671273199914,
-      "grad_norm": 0.0016360600711777806,
-      "learning_rate": 0.00019999605168844053,
-      "loss": 46.0,
-      "step": 37006
-    },
-    {
-      "epoch": 2.829443584303381,
-      "grad_norm": 0.0012815962545573711,
-      "learning_rate": 0.00019999605147499373,
-      "loss": 46.0,
-      "step": 37007
-    },
-    {
-      "epoch": 2.829520041286771,
-      "grad_norm": 0.003865294624119997,
-      "learning_rate": 0.00019999605126154118,
-      "loss": 46.0,
-      "step": 37008
-    },
-    {
-      "epoch": 2.8295964982701607,
-      "grad_norm": 0.0035797294694930315,
-      "learning_rate": 0.0001999960510480828,
-      "loss": 46.0,
-      "step": 37009
-    },
-    {
-      "epoch": 2.8296729552535504,
-      "grad_norm": 0.0010599744273349643,
-      "learning_rate": 0.00019999605083461871,
-      "loss": 46.0,
-      "step": 37010
-    },
-    {
-      "epoch": 2.82974941223694,
-      "grad_norm": 0.003947384189814329,
-      "learning_rate": 0.00019999605062114882,
-      "loss": 46.0,
-      "step": 37011
-    },
-    {
-      "epoch": 2.82982586922033,
-      "grad_norm": 0.0008913521305657923,
-      "learning_rate": 0.00019999605040767318,
-      "loss": 46.0,
-      "step": 37012
-    },
-    {
-      "epoch": 2.8299023262037197,
-      "grad_norm": 0.0037171305157244205,
-      "learning_rate": 0.00019999605019419177,
-      "loss": 46.0,
-      "step": 37013
-    },
-    {
-      "epoch": 2.8299787831871095,
-      "grad_norm": 0.0021713310852646828,
-      "learning_rate": 0.00019999604998070458,
-      "loss": 46.0,
-      "step": 37014
-    },
-    {
-      "epoch": 2.8300552401704993,
-      "grad_norm": 0.0014609800418838859,
-      "learning_rate": 0.00019999604976721162,
-      "loss": 46.0,
-      "step": 37015
-    },
-    {
-      "epoch": 2.830131697153889,
-      "grad_norm": 0.0018056613625958562,
-      "learning_rate": 0.00019999604955371289,
-      "loss": 46.0,
-      "step": 37016
-    },
-    {
-      "epoch": 2.8302081541372788,
-      "grad_norm": 0.0014918268425390124,
-      "learning_rate": 0.0001999960493402084,
-      "loss": 46.0,
-      "step": 37017
-    },
-    {
-      "epoch": 2.830284611120668,
-      "grad_norm": 0.001848678570240736,
-      "learning_rate": 0.00019999604912669815,
-      "loss": 46.0,
-      "step": 37018
-    },
-    {
-      "epoch": 2.830361068104058,
-      "grad_norm": 0.001480182516388595,
-      "learning_rate": 0.0001999960489131821,
-      "loss": 46.0,
-      "step": 37019
-    },
-    {
-      "epoch": 2.8304375250874476,
-      "grad_norm": 0.003131404286250472,
-      "learning_rate": 0.0001999960486996603,
-      "loss": 46.0,
-      "step": 37020
-    },
-    {
-      "epoch": 2.8305139820708374,
-      "grad_norm": 0.00136624276638031,
-      "learning_rate": 0.00019999604848613273,
-      "loss": 46.0,
-      "step": 37021
-    },
-    {
-      "epoch": 2.830590439054227,
-      "grad_norm": 0.0027331511955708265,
-      "learning_rate": 0.0001999960482725994,
-      "loss": 46.0,
-      "step": 37022
-    },
-    {
-      "epoch": 2.830666896037617,
-      "grad_norm": 0.0009661935037001967,
-      "learning_rate": 0.00019999604805906029,
-      "loss": 46.0,
-      "step": 37023
-    },
-    {
-      "epoch": 2.8307433530210067,
-      "grad_norm": 0.003178333630785346,
-      "learning_rate": 0.0001999960478455154,
-      "loss": 46.0,
-      "step": 37024
-    },
-    {
-      "epoch": 2.8308198100043964,
-      "grad_norm": 0.0037964927032589912,
-      "learning_rate": 0.00019999604763196475,
-      "loss": 46.0,
-      "step": 37025
-    },
-    {
-      "epoch": 2.8308962669877857,
-      "grad_norm": 0.0023397982586175203,
-      "learning_rate": 0.00019999604741840834,
-      "loss": 46.0,
-      "step": 37026
-    },
-    {
-      "epoch": 2.8309727239711755,
-      "grad_norm": 0.0028361817821860313,
-      "learning_rate": 0.00019999604720484615,
-      "loss": 46.0,
-      "step": 37027
-    },
-    {
-      "epoch": 2.8310491809545653,
-      "grad_norm": 0.0020070101600140333,
-      "learning_rate": 0.00019999604699127822,
-      "loss": 46.0,
-      "step": 37028
-    },
-    {
-      "epoch": 2.831125637937955,
-      "grad_norm": 0.0044563873670995235,
-      "learning_rate": 0.00019999604677770449,
-      "loss": 46.0,
-      "step": 37029
-    },
-    {
-      "epoch": 2.831202094921345,
-      "grad_norm": 0.0019707109313458204,
-      "learning_rate": 0.00019999604656412498,
-      "loss": 46.0,
-      "step": 37030
-    },
-    {
-      "epoch": 2.8312785519047345,
-      "grad_norm": 0.002065874170511961,
-      "learning_rate": 0.00019999604635053973,
-      "loss": 46.0,
-      "step": 37031
-    },
-    {
-      "epoch": 2.8313550088881243,
-      "grad_norm": 0.0018800286343321204,
-      "learning_rate": 0.0001999960461369487,
-      "loss": 46.0,
-      "step": 37032
-    },
-    {
-      "epoch": 2.831431465871514,
-      "grad_norm": 0.003695941995829344,
-      "learning_rate": 0.0001999960459233519,
-      "loss": 46.0,
-      "step": 37033
-    },
-    {
-      "epoch": 2.831507922854904,
-      "grad_norm": 0.0024002997670322657,
-      "learning_rate": 0.00019999604570974933,
-      "loss": 46.0,
-      "step": 37034
-    },
-    {
-      "epoch": 2.8315843798382936,
-      "grad_norm": 0.0024831988848745823,
-      "learning_rate": 0.00019999604549614098,
-      "loss": 46.0,
-      "step": 37035
-    },
-    {
-      "epoch": 2.8316608368216833,
-      "grad_norm": 0.003218643134459853,
-      "learning_rate": 0.00019999604528252686,
-      "loss": 46.0,
-      "step": 37036
-    },
-    {
-      "epoch": 2.831737293805073,
-      "grad_norm": 0.0009372446220368147,
-      "learning_rate": 0.000199996045068907,
-      "loss": 46.0,
-      "step": 37037
-    },
-    {
-      "epoch": 2.831813750788463,
-      "grad_norm": 0.003029302228242159,
-      "learning_rate": 0.00019999604485528133,
-      "loss": 46.0,
-      "step": 37038
-    },
-    {
-      "epoch": 2.8318902077718526,
-      "grad_norm": 0.0028853006660938263,
-      "learning_rate": 0.00019999604464164992,
-      "loss": 46.0,
-      "step": 37039
-    },
-    {
-      "epoch": 2.831966664755242,
-      "grad_norm": 0.0019728646147996187,
-      "learning_rate": 0.00019999604442801273,
-      "loss": 46.0,
-      "step": 37040
-    },
-    {
-      "epoch": 2.8320431217386317,
-      "grad_norm": 0.0017252343241125345,
-      "learning_rate": 0.0001999960442143698,
-      "loss": 46.0,
-      "step": 37041
-    },
-    {
-      "epoch": 2.8321195787220215,
-      "grad_norm": 0.002286685397848487,
-      "learning_rate": 0.00019999604400072106,
-      "loss": 46.0,
-      "step": 37042
-    },
-    {
-      "epoch": 2.8321960357054112,
-      "grad_norm": 0.0010402693878859282,
-      "learning_rate": 0.00019999604378706658,
-      "loss": 46.0,
-      "step": 37043
-    },
-    {
-      "epoch": 2.832272492688801,
-      "grad_norm": 0.0015757293440401554,
-      "learning_rate": 0.0001999960435734063,
-      "loss": 46.0,
-      "step": 37044
-    },
-    {
-      "epoch": 2.8323489496721908,
-      "grad_norm": 0.0017045572167262435,
-      "learning_rate": 0.00019999604335974028,
-      "loss": 46.0,
-      "step": 37045
-    },
-    {
-      "epoch": 2.8324254066555805,
-      "grad_norm": 0.005007521249353886,
-      "learning_rate": 0.00019999604314606848,
-      "loss": 46.0,
-      "step": 37046
-    },
-    {
-      "epoch": 2.8325018636389703,
-      "grad_norm": 0.0048163775354623795,
-      "learning_rate": 0.0001999960429323909,
-      "loss": 46.0,
-      "step": 37047
-    },
-    {
-      "epoch": 2.8325783206223596,
-      "grad_norm": 0.004868373274803162,
-      "learning_rate": 0.0001999960427187076,
-      "loss": 46.0,
-      "step": 37048
-    },
-    {
-      "epoch": 2.8326547776057494,
-      "grad_norm": 0.004244635812938213,
-      "learning_rate": 0.00019999604250501847,
-      "loss": 46.0,
-      "step": 37049
-    },
-    {
-      "epoch": 2.832731234589139,
-      "grad_norm": 0.002112451707944274,
-      "learning_rate": 0.00019999604229132357,
-      "loss": 46.0,
-      "step": 37050
-    },
-    {
-      "epoch": 2.832807691572529,
-      "grad_norm": 0.002105921506881714,
-      "learning_rate": 0.00019999604207762294,
-      "loss": 46.0,
-      "step": 37051
-    },
-    {
-      "epoch": 2.8328841485559186,
-      "grad_norm": 0.002561325905844569,
-      "learning_rate": 0.0001999960418639165,
-      "loss": 46.0,
-      "step": 37052
-    },
-    {
-      "epoch": 2.8329606055393084,
-      "grad_norm": 0.0020687479991465807,
-      "learning_rate": 0.00019999604165020434,
-      "loss": 46.0,
-      "step": 37053
-    },
-    {
-      "epoch": 2.833037062522698,
-      "grad_norm": 0.0027319584041833878,
-      "learning_rate": 0.00019999604143648635,
-      "loss": 46.0,
-      "step": 37054
-    },
-    {
-      "epoch": 2.833113519506088,
-      "grad_norm": 0.0030791524332016706,
-      "learning_rate": 0.00019999604122276264,
-      "loss": 46.0,
-      "step": 37055
-    },
-    {
-      "epoch": 2.8331899764894777,
-      "grad_norm": 0.003188525792211294,
-      "learning_rate": 0.00019999604100903317,
-      "loss": 46.0,
-      "step": 37056
-    },
-    {
-      "epoch": 2.8332664334728674,
-      "grad_norm": 0.0026026912964880466,
-      "learning_rate": 0.00019999604079529789,
-      "loss": 46.0,
-      "step": 37057
-    },
-    {
-      "epoch": 2.833342890456257,
-      "grad_norm": 0.0026390100829303265,
-      "learning_rate": 0.00019999604058155686,
-      "loss": 46.0,
-      "step": 37058
-    },
-    {
-      "epoch": 2.833419347439647,
-      "grad_norm": 0.0020187459886074066,
-      "learning_rate": 0.00019999604036781006,
-      "loss": 46.0,
-      "step": 37059
-    },
-    {
-      "epoch": 2.8334958044230367,
-      "grad_norm": 0.0008578562992624938,
-      "learning_rate": 0.0001999960401540575,
-      "loss": 46.0,
-      "step": 37060
-    },
-    {
-      "epoch": 2.8335722614064265,
-      "grad_norm": 0.002300386782735586,
-      "learning_rate": 0.00019999603994029914,
-      "loss": 46.0,
-      "step": 37061
-    },
-    {
-      "epoch": 2.833648718389816,
-      "grad_norm": 0.007256791461259127,
-      "learning_rate": 0.00019999603972653502,
-      "loss": 46.0,
-      "step": 37062
-    },
-    {
-      "epoch": 2.8337251753732056,
-      "grad_norm": 0.002237992826849222,
-      "learning_rate": 0.00019999603951276516,
-      "loss": 46.0,
-      "step": 37063
-    },
-    {
-      "epoch": 2.8338016323565953,
-      "grad_norm": 0.005145910196006298,
-      "learning_rate": 0.00019999603929898952,
-      "loss": 46.0,
-      "step": 37064
-    },
-    {
-      "epoch": 2.833878089339985,
-      "grad_norm": 0.00281929736956954,
-      "learning_rate": 0.0001999960390852081,
-      "loss": 46.0,
-      "step": 37065
-    },
-    {
-      "epoch": 2.833954546323375,
-      "grad_norm": 0.008032320998609066,
-      "learning_rate": 0.00019999603887142092,
-      "loss": 46.0,
-      "step": 37066
-    },
-    {
-      "epoch": 2.8340310033067646,
-      "grad_norm": 0.0026360168121755123,
-      "learning_rate": 0.00019999603865762796,
-      "loss": 46.0,
-      "step": 37067
-    },
-    {
-      "epoch": 2.8341074602901544,
-      "grad_norm": 0.0012330182362347841,
-      "learning_rate": 0.0001999960384438292,
-      "loss": 46.0,
-      "step": 37068
-    },
-    {
-      "epoch": 2.834183917273544,
-      "grad_norm": 0.0017202867893502116,
-      "learning_rate": 0.00019999603823002472,
-      "loss": 46.0,
-      "step": 37069
-    },
-    {
-      "epoch": 2.8342603742569334,
-      "grad_norm": 0.0027398562524467707,
-      "learning_rate": 0.00019999603801621447,
-      "loss": 46.0,
-      "step": 37070
-    },
-    {
-      "epoch": 2.834336831240323,
-      "grad_norm": 0.0027491161599755287,
-      "learning_rate": 0.00019999603780239845,
-      "loss": 46.0,
-      "step": 37071
-    },
-    {
-      "epoch": 2.834413288223713,
-      "grad_norm": 0.0017183690797537565,
-      "learning_rate": 0.00019999603758857665,
-      "loss": 46.0,
-      "step": 37072
-    },
-    {
-      "epoch": 2.8344897452071027,
-      "grad_norm": 0.0034974408335983753,
-      "learning_rate": 0.00019999603737474905,
-      "loss": 46.0,
-      "step": 37073
-    },
-    {
-      "epoch": 2.8345662021904925,
-      "grad_norm": 0.0010642207926139235,
-      "learning_rate": 0.00019999603716091573,
-      "loss": 46.0,
-      "step": 37074
-    },
-    {
-      "epoch": 2.8346426591738823,
-      "grad_norm": 0.0025160901714116335,
-      "learning_rate": 0.00019999603694707664,
-      "loss": 46.0,
-      "step": 37075
-    },
-    {
-      "epoch": 2.834719116157272,
-      "grad_norm": 0.0024594564456492662,
-      "learning_rate": 0.00019999603673323172,
-      "loss": 46.0,
-      "step": 37076
-    },
-    {
-      "epoch": 2.8347955731406618,
-      "grad_norm": 0.0027378350496292114,
-      "learning_rate": 0.0001999960365193811,
-      "loss": 46.0,
-      "step": 37077
-    },
-    {
-      "epoch": 2.8348720301240515,
-      "grad_norm": 0.0033158729784190655,
-      "learning_rate": 0.00019999603630552467,
-      "loss": 46.0,
-      "step": 37078
-    },
-    {
-      "epoch": 2.8349484871074413,
-      "grad_norm": 0.0007978108478710055,
-      "learning_rate": 0.00019999603609166248,
-      "loss": 46.0,
-      "step": 37079
-    },
-    {
-      "epoch": 2.835024944090831,
-      "grad_norm": 0.003902283264324069,
-      "learning_rate": 0.00019999603587779455,
-      "loss": 46.0,
-      "step": 37080
-    },
-    {
-      "epoch": 2.835101401074221,
-      "grad_norm": 0.009267600253224373,
-      "learning_rate": 0.00019999603566392082,
-      "loss": 46.0,
-      "step": 37081
-    },
-    {
-      "epoch": 2.8351778580576106,
-      "grad_norm": 0.0013335481053218246,
-      "learning_rate": 0.00019999603545004134,
-      "loss": 46.0,
-      "step": 37082
-    },
-    {
-      "epoch": 2.835254315041,
-      "grad_norm": 0.008002828806638718,
-      "learning_rate": 0.00019999603523615606,
-      "loss": 46.0,
-      "step": 37083
-    },
-    {
-      "epoch": 2.8353307720243897,
-      "grad_norm": 0.0015505505725741386,
-      "learning_rate": 0.00019999603502226504,
-      "loss": 46.0,
-      "step": 37084
-    },
-    {
-      "epoch": 2.8354072290077794,
-      "grad_norm": 0.0014241415774449706,
-      "learning_rate": 0.0001999960348083682,
-      "loss": 46.0,
-      "step": 37085
-    },
-    {
-      "epoch": 2.835483685991169,
-      "grad_norm": 0.002636583987623453,
-      "learning_rate": 0.00019999603459446567,
-      "loss": 46.0,
-      "step": 37086
-    },
-    {
-      "epoch": 2.835560142974559,
-      "grad_norm": 0.001091033685952425,
-      "learning_rate": 0.00019999603438055732,
-      "loss": 46.0,
-      "step": 37087
-    },
-    {
-      "epoch": 2.8356365999579487,
-      "grad_norm": 0.001141828834079206,
-      "learning_rate": 0.00019999603416664318,
-      "loss": 46.0,
-      "step": 37088
-    },
-    {
-      "epoch": 2.8357130569413385,
-      "grad_norm": 0.005088814068585634,
-      "learning_rate": 0.00019999603395272334,
-      "loss": 46.0,
-      "step": 37089
-    },
-    {
-      "epoch": 2.8357895139247282,
-      "grad_norm": 0.0016998846549540758,
-      "learning_rate": 0.00019999603373879767,
-      "loss": 46.0,
-      "step": 37090
-    },
-    {
-      "epoch": 2.835865970908118,
-      "grad_norm": 0.0035497951321303844,
-      "learning_rate": 0.00019999603352486623,
-      "loss": 46.0,
-      "step": 37091
-    },
-    {
-      "epoch": 2.8359424278915073,
-      "grad_norm": 0.0030108769424259663,
-      "learning_rate": 0.00019999603331092908,
-      "loss": 46.0,
-      "step": 37092
-    },
-    {
-      "epoch": 2.836018884874897,
-      "grad_norm": 0.001296893460676074,
-      "learning_rate": 0.00019999603309698612,
-      "loss": 46.0,
-      "step": 37093
-    },
-    {
-      "epoch": 2.836095341858287,
-      "grad_norm": 0.001287568244151771,
-      "learning_rate": 0.00019999603288303739,
-      "loss": 46.0,
-      "step": 37094
-    },
-    {
-      "epoch": 2.8361717988416766,
-      "grad_norm": 0.0013096706243231893,
-      "learning_rate": 0.00019999603266908288,
-      "loss": 46.0,
-      "step": 37095
-    },
-    {
-      "epoch": 2.8362482558250663,
-      "grad_norm": 0.007514851167798042,
-      "learning_rate": 0.00019999603245512263,
-      "loss": 46.0,
-      "step": 37096
-    },
-    {
-      "epoch": 2.836324712808456,
-      "grad_norm": 0.001062970608472824,
-      "learning_rate": 0.0001999960322411566,
-      "loss": 46.0,
-      "step": 37097
-    },
-    {
-      "epoch": 2.836401169791846,
-      "grad_norm": 0.002191146370023489,
-      "learning_rate": 0.0001999960320271848,
-      "loss": 46.0,
-      "step": 37098
-    },
-    {
-      "epoch": 2.8364776267752356,
-      "grad_norm": 0.0019992170855402946,
-      "learning_rate": 0.00019999603181320723,
-      "loss": 46.0,
-      "step": 37099
-    },
-    {
-      "epoch": 2.8365540837586254,
-      "grad_norm": 0.002211317652836442,
-      "learning_rate": 0.0001999960315992239,
-      "loss": 46.0,
-      "step": 37100
-    },
-    {
-      "epoch": 2.836630540742015,
-      "grad_norm": 0.0007561781676486135,
-      "learning_rate": 0.0001999960313852348,
-      "loss": 46.0,
-      "step": 37101
-    },
-    {
-      "epoch": 2.836706997725405,
-      "grad_norm": 0.0015603979118168354,
-      "learning_rate": 0.0001999960311712399,
-      "loss": 46.0,
-      "step": 37102
-    },
-    {
-      "epoch": 2.8367834547087947,
-      "grad_norm": 0.0019219920504838228,
-      "learning_rate": 0.00019999603095723924,
-      "loss": 46.0,
-      "step": 37103
-    },
-    {
-      "epoch": 2.8368599116921844,
-      "grad_norm": 0.004766498226672411,
-      "learning_rate": 0.00019999603074323283,
-      "loss": 46.0,
-      "step": 37104
-    },
-    {
-      "epoch": 2.8369363686755737,
-      "grad_norm": 0.002429333748295903,
-      "learning_rate": 0.00019999603052922065,
-      "loss": 46.0,
-      "step": 37105
-    },
-    {
-      "epoch": 2.8370128256589635,
-      "grad_norm": 0.0024095450062304735,
-      "learning_rate": 0.0001999960303152027,
-      "loss": 46.0,
-      "step": 37106
-    },
-    {
-      "epoch": 2.8370892826423533,
-      "grad_norm": 0.0016871263505890965,
-      "learning_rate": 0.00019999603010117898,
-      "loss": 46.0,
-      "step": 37107
-    },
-    {
-      "epoch": 2.837165739625743,
-      "grad_norm": 0.0015730004524812102,
-      "learning_rate": 0.0001999960298871495,
-      "loss": 46.0,
-      "step": 37108
-    },
-    {
-      "epoch": 2.837242196609133,
-      "grad_norm": 0.0027134520933032036,
-      "learning_rate": 0.00019999602967311423,
-      "loss": 46.0,
-      "step": 37109
-    },
-    {
-      "epoch": 2.8373186535925226,
-      "grad_norm": 0.0009012341615743935,
-      "learning_rate": 0.00019999602945907318,
-      "loss": 46.0,
-      "step": 37110
-    },
-    {
-      "epoch": 2.8373951105759123,
-      "grad_norm": 0.0027486993931233883,
-      "learning_rate": 0.00019999602924502638,
-      "loss": 46.0,
-      "step": 37111
-    },
-    {
-      "epoch": 2.837471567559302,
-      "grad_norm": 0.003142373403534293,
-      "learning_rate": 0.00019999602903097383,
-      "loss": 46.0,
-      "step": 37112
-    },
-    {
-      "epoch": 2.8375480245426914,
-      "grad_norm": 0.005021336022764444,
-      "learning_rate": 0.0001999960288169155,
-      "loss": 46.0,
-      "step": 37113
-    },
-    {
-      "epoch": 2.837624481526081,
-      "grad_norm": 0.0010762499878183007,
-      "learning_rate": 0.00019999602860285137,
-      "loss": 46.0,
-      "step": 37114
-    },
-    {
-      "epoch": 2.837700938509471,
-      "grad_norm": 0.0008226162171922624,
-      "learning_rate": 0.00019999602838878148,
-      "loss": 46.0,
-      "step": 37115
-    },
-    {
-      "epoch": 2.8377773954928607,
-      "grad_norm": 0.003002900630235672,
-      "learning_rate": 0.00019999602817470584,
-      "loss": 46.0,
-      "step": 37116
-    },
-    {
-      "epoch": 2.8378538524762504,
-      "grad_norm": 0.002599109662696719,
-      "learning_rate": 0.00019999602796062443,
-      "loss": 46.0,
-      "step": 37117
-    },
-    {
-      "epoch": 2.83793030945964,
-      "grad_norm": 0.0011123527074232697,
-      "learning_rate": 0.00019999602774653722,
-      "loss": 46.0,
-      "step": 37118
-    },
-    {
-      "epoch": 2.83800676644303,
-      "grad_norm": 0.0028113347943872213,
-      "learning_rate": 0.0001999960275324443,
-      "loss": 46.0,
-      "step": 37119
-    },
-    {
-      "epoch": 2.8380832234264197,
-      "grad_norm": 0.002901636064052582,
-      "learning_rate": 0.00019999602731834556,
-      "loss": 46.0,
-      "step": 37120
-    },
-    {
-      "epoch": 2.8381596804098095,
-      "grad_norm": 0.0012139285681769252,
-      "learning_rate": 0.00019999602710424105,
-      "loss": 46.0,
-      "step": 37121
-    },
-    {
-      "epoch": 2.8382361373931992,
-      "grad_norm": 0.0026648107450455427,
-      "learning_rate": 0.00019999602689013083,
-      "loss": 46.0,
-      "step": 37122
-    },
-    {
-      "epoch": 2.838312594376589,
-      "grad_norm": 0.003036793554201722,
-      "learning_rate": 0.00019999602667601478,
-      "loss": 46.0,
-      "step": 37123
-    },
-    {
-      "epoch": 2.8383890513599788,
-      "grad_norm": 0.000876863079611212,
-      "learning_rate": 0.00019999602646189298,
-      "loss": 46.0,
-      "step": 37124
-    },
-    {
-      "epoch": 2.8384655083433685,
-      "grad_norm": 0.004673229996114969,
-      "learning_rate": 0.0001999960262477654,
-      "loss": 46.0,
-      "step": 37125
-    },
-    {
-      "epoch": 2.8385419653267583,
-      "grad_norm": 0.0013813417172059417,
-      "learning_rate": 0.00019999602603363207,
-      "loss": 46.0,
-      "step": 37126
-    },
-    {
-      "epoch": 2.8386184223101476,
-      "grad_norm": 0.0028213909827172756,
-      "learning_rate": 0.00019999602581949297,
-      "loss": 46.0,
-      "step": 37127
-    },
-    {
-      "epoch": 2.8386948792935374,
-      "grad_norm": 0.0025242804549634457,
-      "learning_rate": 0.00019999602560534808,
-      "loss": 46.0,
-      "step": 37128
-    },
-    {
-      "epoch": 2.838771336276927,
-      "grad_norm": 0.001962625654414296,
-      "learning_rate": 0.00019999602539119745,
-      "loss": 46.0,
-      "step": 37129
-    },
-    {
-      "epoch": 2.838847793260317,
-      "grad_norm": 0.0029655182734131813,
-      "learning_rate": 0.000199996025177041,
-      "loss": 46.0,
-      "step": 37130
-    },
-    {
-      "epoch": 2.8389242502437066,
-      "grad_norm": 0.0009559431346133351,
-      "learning_rate": 0.00019999602496287882,
-      "loss": 46.0,
-      "step": 37131
-    },
-    {
-      "epoch": 2.8390007072270964,
-      "grad_norm": 0.006765376776456833,
-      "learning_rate": 0.0001999960247487109,
-      "loss": 46.0,
-      "step": 37132
-    },
-    {
-      "epoch": 2.839077164210486,
-      "grad_norm": 0.006151263602077961,
-      "learning_rate": 0.00019999602453453714,
-      "loss": 46.0,
-      "step": 37133
-    },
-    {
-      "epoch": 2.839153621193876,
-      "grad_norm": 0.004104913212358952,
-      "learning_rate": 0.00019999602432035766,
-      "loss": 46.0,
-      "step": 37134
-    },
-    {
-      "epoch": 2.8392300781772652,
-      "grad_norm": 0.003158940002322197,
-      "learning_rate": 0.0001999960241061724,
-      "loss": 46.0,
-      "step": 37135
-    },
-    {
-      "epoch": 2.839306535160655,
-      "grad_norm": 0.0025407602079212666,
-      "learning_rate": 0.00019999602389198136,
-      "loss": 46.0,
-      "step": 37136
-    },
-    {
-      "epoch": 2.8393829921440448,
-      "grad_norm": 0.0013883167412132025,
-      "learning_rate": 0.00019999602367778456,
-      "loss": 46.0,
-      "step": 37137
-    },
-    {
-      "epoch": 2.8394594491274345,
-      "grad_norm": 0.0027060771826654673,
-      "learning_rate": 0.000199996023463582,
-      "loss": 46.0,
-      "step": 37138
-    },
-    {
-      "epoch": 2.8395359061108243,
-      "grad_norm": 0.004329566843807697,
-      "learning_rate": 0.00019999602324937365,
-      "loss": 46.0,
-      "step": 37139
-    },
-    {
-      "epoch": 2.839612363094214,
-      "grad_norm": 0.0006606465321965516,
-      "learning_rate": 0.00019999602303515953,
-      "loss": 46.0,
-      "step": 37140
-    },
-    {
-      "epoch": 2.839688820077604,
-      "grad_norm": 0.0012276185443624854,
-      "learning_rate": 0.00019999602282093966,
-      "loss": 46.0,
-      "step": 37141
-    },
-    {
-      "epoch": 2.8397652770609936,
-      "grad_norm": 0.0023852468002587557,
-      "learning_rate": 0.00019999602260671403,
-      "loss": 46.0,
-      "step": 37142
-    },
-    {
-      "epoch": 2.8398417340443833,
-      "grad_norm": 0.001401030458509922,
-      "learning_rate": 0.00019999602239248262,
-      "loss": 46.0,
-      "step": 37143
-    },
-    {
-      "epoch": 2.839918191027773,
-      "grad_norm": 0.009070102125406265,
-      "learning_rate": 0.0001999960221782454,
-      "loss": 46.0,
-      "step": 37144
-    },
-    {
-      "epoch": 2.839994648011163,
-      "grad_norm": 0.0009845560416579247,
-      "learning_rate": 0.00019999602196400248,
-      "loss": 46.0,
-      "step": 37145
-    },
-    {
-      "epoch": 2.8400711049945526,
-      "grad_norm": 0.0016333868261426687,
-      "learning_rate": 0.00019999602174975375,
-      "loss": 46.0,
-      "step": 37146
-    },
-    {
-      "epoch": 2.8401475619779424,
-      "grad_norm": 0.004019290674477816,
-      "learning_rate": 0.00019999602153549924,
-      "loss": 46.0,
-      "step": 37147
-    },
-    {
-      "epoch": 2.840224018961332,
-      "grad_norm": 0.0035774612333625555,
-      "learning_rate": 0.000199996021321239,
-      "loss": 46.0,
-      "step": 37148
-    },
-    {
-      "epoch": 2.8403004759447215,
-      "grad_norm": 0.004045620560646057,
-      "learning_rate": 0.00019999602110697294,
-      "loss": 46.0,
-      "step": 37149
-    },
-    {
-      "epoch": 2.840376932928111,
-      "grad_norm": 0.0011635760311037302,
-      "learning_rate": 0.00019999602089270114,
-      "loss": 46.0,
-      "step": 37150
-    },
-    {
-      "epoch": 2.840453389911501,
-      "grad_norm": 0.001537640462629497,
-      "learning_rate": 0.00019999602067842357,
-      "loss": 46.0,
-      "step": 37151
-    },
-    {
-      "epoch": 2.8405298468948907,
-      "grad_norm": 0.002714330330491066,
-      "learning_rate": 0.00019999602046414023,
-      "loss": 46.0,
-      "step": 37152
-    },
-    {
-      "epoch": 2.8406063038782805,
-      "grad_norm": 0.00142281677108258,
-      "learning_rate": 0.00019999602024985114,
-      "loss": 46.0,
-      "step": 37153
-    },
-    {
-      "epoch": 2.8406827608616703,
-      "grad_norm": 0.0015132240951061249,
-      "learning_rate": 0.00019999602003555628,
-      "loss": 46.0,
-      "step": 37154
-    },
-    {
-      "epoch": 2.84075921784506,
-      "grad_norm": 0.0015835468657314777,
-      "learning_rate": 0.00019999601982125559,
-      "loss": 46.0,
-      "step": 37155
-    },
-    {
-      "epoch": 2.84083567482845,
-      "grad_norm": 0.0035446430556476116,
-      "learning_rate": 0.0001999960196069492,
-      "loss": 46.0,
-      "step": 37156
-    },
-    {
-      "epoch": 2.840912131811839,
-      "grad_norm": 0.0011686112266033888,
-      "learning_rate": 0.000199996019392637,
-      "loss": 46.0,
-      "step": 37157
-    },
-    {
-      "epoch": 2.840988588795229,
-      "grad_norm": 0.0004988678847439587,
-      "learning_rate": 0.00019999601917831906,
-      "loss": 46.0,
-      "step": 37158
-    },
-    {
-      "epoch": 2.8410650457786186,
-      "grad_norm": 0.0019041512859985232,
-      "learning_rate": 0.00019999601896399533,
-      "loss": 46.0,
-      "step": 37159
-    },
-    {
-      "epoch": 2.8411415027620084,
-      "grad_norm": 0.0013399479212239385,
-      "learning_rate": 0.00019999601874966586,
-      "loss": 46.0,
-      "step": 37160
-    },
-    {
-      "epoch": 2.841217959745398,
-      "grad_norm": 0.0037399388384073973,
-      "learning_rate": 0.00019999601853533058,
-      "loss": 46.0,
-      "step": 37161
-    },
-    {
-      "epoch": 2.841294416728788,
-      "grad_norm": 0.0035389885306358337,
-      "learning_rate": 0.00019999601832098956,
-      "loss": 46.0,
-      "step": 37162
-    },
-    {
-      "epoch": 2.8413708737121777,
-      "grad_norm": 0.0029092596378177404,
-      "learning_rate": 0.00019999601810664276,
-      "loss": 46.0,
-      "step": 37163
-    },
-    {
-      "epoch": 2.8414473306955674,
-      "grad_norm": 0.0005028339219279587,
-      "learning_rate": 0.00019999601789229016,
-      "loss": 46.0,
-      "step": 37164
-    },
-    {
-      "epoch": 2.841523787678957,
-      "grad_norm": 0.004048381466418505,
-      "learning_rate": 0.00019999601767793182,
-      "loss": 46.0,
-      "step": 37165
-    },
-    {
-      "epoch": 2.841600244662347,
-      "grad_norm": 0.001473005278967321,
-      "learning_rate": 0.00019999601746356773,
-      "loss": 46.0,
-      "step": 37166
-    },
-    {
-      "epoch": 2.8416767016457367,
-      "grad_norm": 0.002869573188945651,
-      "learning_rate": 0.00019999601724919784,
-      "loss": 46.0,
-      "step": 37167
-    },
-    {
-      "epoch": 2.8417531586291265,
-      "grad_norm": 0.020751535892486572,
-      "learning_rate": 0.0001999960170348222,
-      "loss": 46.0,
-      "step": 37168
-    },
-    {
-      "epoch": 2.8418296156125162,
-      "grad_norm": 0.0018582474440336227,
-      "learning_rate": 0.00019999601682044077,
-      "loss": 46.0,
-      "step": 37169
-    },
-    {
-      "epoch": 2.841906072595906,
-      "grad_norm": 0.0012318510562181473,
-      "learning_rate": 0.00019999601660605358,
-      "loss": 46.0,
-      "step": 37170
-    },
-    {
-      "epoch": 2.8419825295792953,
-      "grad_norm": 0.001933880615979433,
-      "learning_rate": 0.00019999601639166063,
-      "loss": 46.0,
-      "step": 37171
-    },
-    {
-      "epoch": 2.842058986562685,
-      "grad_norm": 0.0018297359347343445,
-      "learning_rate": 0.00019999601617726192,
-      "loss": 46.0,
-      "step": 37172
-    },
-    {
-      "epoch": 2.842135443546075,
-      "grad_norm": 0.0018213269067928195,
-      "learning_rate": 0.00019999601596285742,
-      "loss": 46.0,
-      "step": 37173
-    },
-    {
-      "epoch": 2.8422119005294646,
-      "grad_norm": 0.003524311352521181,
-      "learning_rate": 0.00019999601574844714,
-      "loss": 46.0,
-      "step": 37174
-    },
-    {
-      "epoch": 2.8422883575128544,
-      "grad_norm": 0.003914892207831144,
-      "learning_rate": 0.00019999601553403112,
-      "loss": 46.0,
-      "step": 37175
-    },
-    {
-      "epoch": 2.842364814496244,
-      "grad_norm": 0.0015047680353745818,
-      "learning_rate": 0.00019999601531960935,
-      "loss": 46.0,
-      "step": 37176
-    },
-    {
-      "epoch": 2.842441271479634,
-      "grad_norm": 0.0028676274232566357,
-      "learning_rate": 0.00019999601510518175,
-      "loss": 46.0,
-      "step": 37177
-    },
-    {
-      "epoch": 2.8425177284630236,
-      "grad_norm": 0.0030861590057611465,
-      "learning_rate": 0.0001999960148907484,
-      "loss": 46.0,
-      "step": 37178
-    },
-    {
-      "epoch": 2.842594185446413,
-      "grad_norm": 0.0030827256850898266,
-      "learning_rate": 0.0001999960146763093,
-      "loss": 46.0,
-      "step": 37179
-    },
-    {
-      "epoch": 2.8426706424298027,
-      "grad_norm": 0.0021316458005458117,
-      "learning_rate": 0.00019999601446186443,
-      "loss": 46.0,
-      "step": 37180
-    },
-    {
-      "epoch": 2.8427470994131925,
-      "grad_norm": 0.0015677333576604724,
-      "learning_rate": 0.0001999960142474138,
-      "loss": 46.0,
-      "step": 37181
-    },
-    {
-      "epoch": 2.8428235563965822,
-      "grad_norm": 0.006090864073485136,
-      "learning_rate": 0.00019999601403295736,
-      "loss": 46.0,
-      "step": 37182
-    },
-    {
-      "epoch": 2.842900013379972,
-      "grad_norm": 0.000551430624909699,
-      "learning_rate": 0.00019999601381849518,
-      "loss": 46.0,
-      "step": 37183
-    },
-    {
-      "epoch": 2.8429764703633618,
-      "grad_norm": 0.0014679405139759183,
-      "learning_rate": 0.00019999601360402722,
-      "loss": 46.0,
-      "step": 37184
-    },
-    {
-      "epoch": 2.8430529273467515,
-      "grad_norm": 0.0036636656150221825,
-      "learning_rate": 0.0001999960133895535,
-      "loss": 46.0,
-      "step": 37185
-    },
-    {
-      "epoch": 2.8431293843301413,
-      "grad_norm": 0.0011902735568583012,
-      "learning_rate": 0.00019999601317507402,
-      "loss": 46.0,
-      "step": 37186
-    },
-    {
-      "epoch": 2.843205841313531,
-      "grad_norm": 0.001085460651665926,
-      "learning_rate": 0.00019999601296058877,
-      "loss": 46.0,
-      "step": 37187
-    },
-    {
-      "epoch": 2.843282298296921,
-      "grad_norm": 0.002682414371520281,
-      "learning_rate": 0.00019999601274609772,
-      "loss": 46.0,
-      "step": 37188
-    },
-    {
-      "epoch": 2.8433587552803106,
-      "grad_norm": 0.0024127059150487185,
-      "learning_rate": 0.00019999601253160092,
-      "loss": 46.0,
-      "step": 37189
-    },
-    {
-      "epoch": 2.8434352122637003,
-      "grad_norm": 0.007181702181696892,
-      "learning_rate": 0.00019999601231709835,
-      "loss": 46.0,
-      "step": 37190
-    },
-    {
-      "epoch": 2.84351166924709,
-      "grad_norm": 0.003293056273832917,
-      "learning_rate": 0.00019999601210259004,
-      "loss": 46.0,
-      "step": 37191
-    },
-    {
-      "epoch": 2.84358812623048,
-      "grad_norm": 0.0018092142418026924,
-      "learning_rate": 0.00019999601188807592,
-      "loss": 46.0,
-      "step": 37192
-    },
-    {
-      "epoch": 2.843664583213869,
-      "grad_norm": 0.0028401408344507217,
-      "learning_rate": 0.000199996011673556,
-      "loss": 46.0,
-      "step": 37193
-    },
-    {
-      "epoch": 2.843741040197259,
-      "grad_norm": 0.003255939343944192,
-      "learning_rate": 0.00019999601145903037,
-      "loss": 46.0,
-      "step": 37194
-    },
-    {
-      "epoch": 2.8438174971806487,
-      "grad_norm": 0.001096990774385631,
-      "learning_rate": 0.00019999601124449896,
-      "loss": 46.0,
-      "step": 37195
-    },
-    {
-      "epoch": 2.8438939541640385,
-      "grad_norm": 0.0016180739039555192,
-      "learning_rate": 0.00019999601102996178,
-      "loss": 46.0,
-      "step": 37196
-    },
-    {
-      "epoch": 2.843970411147428,
-      "grad_norm": 0.0027516987174749374,
-      "learning_rate": 0.00019999601081541882,
-      "loss": 46.0,
-      "step": 37197
-    },
-    {
-      "epoch": 2.844046868130818,
-      "grad_norm": 0.0020281956531107426,
-      "learning_rate": 0.0001999960106008701,
-      "loss": 46.0,
-      "step": 37198
-    },
-    {
-      "epoch": 2.8441233251142077,
-      "grad_norm": 0.0014528222382068634,
-      "learning_rate": 0.00019999601038631561,
-      "loss": 46.0,
-      "step": 37199
-    },
-    {
-      "epoch": 2.8441997820975975,
-      "grad_norm": 0.0015012851217761636,
-      "learning_rate": 0.00019999601017175534,
-      "loss": 46.0,
-      "step": 37200
-    },
-    {
-      "epoch": 2.844276239080987,
-      "grad_norm": 0.0037168373819440603,
-      "learning_rate": 0.00019999600995718932,
-      "loss": 46.0,
-      "step": 37201
-    },
-    {
-      "epoch": 2.8443526960643766,
-      "grad_norm": 0.0035074367187917233,
-      "learning_rate": 0.0001999960097426175,
-      "loss": 46.0,
-      "step": 37202
-    },
-    {
-      "epoch": 2.8444291530477663,
-      "grad_norm": 0.004065664950758219,
-      "learning_rate": 0.00019999600952803995,
-      "loss": 46.0,
-      "step": 37203
-    },
-    {
-      "epoch": 2.844505610031156,
-      "grad_norm": 0.002774972002953291,
-      "learning_rate": 0.0001999960093134566,
-      "loss": 46.0,
-      "step": 37204
-    },
-    {
-      "epoch": 2.844582067014546,
-      "grad_norm": 0.0014166026376187801,
-      "learning_rate": 0.00019999600909886752,
-      "loss": 46.0,
-      "step": 37205
-    },
-    {
-      "epoch": 2.8446585239979356,
-      "grad_norm": 0.004844150971621275,
-      "learning_rate": 0.0001999960088842726,
-      "loss": 46.0,
-      "step": 37206
-    },
-    {
-      "epoch": 2.8447349809813254,
-      "grad_norm": 0.0019219416426494718,
-      "learning_rate": 0.000199996008669672,
-      "loss": 46.0,
-      "step": 37207
-    },
-    {
-      "epoch": 2.844811437964715,
-      "grad_norm": 0.0026399383787065744,
-      "learning_rate": 0.00019999600845506556,
-      "loss": 46.0,
-      "step": 37208
-    },
-    {
-      "epoch": 2.844887894948105,
-      "grad_norm": 0.002299330662935972,
-      "learning_rate": 0.00019999600824045338,
-      "loss": 46.0,
-      "step": 37209
-    },
-    {
-      "epoch": 2.8449643519314947,
-      "grad_norm": 0.0034319909755140543,
-      "learning_rate": 0.0001999960080258354,
-      "loss": 46.0,
-      "step": 37210
-    },
-    {
-      "epoch": 2.8450408089148844,
-      "grad_norm": 0.003196443198248744,
-      "learning_rate": 0.0001999960078112117,
-      "loss": 46.0,
-      "step": 37211
-    },
-    {
-      "epoch": 2.845117265898274,
-      "grad_norm": 0.004450627602636814,
-      "learning_rate": 0.0001999960075965822,
-      "loss": 46.0,
-      "step": 37212
-    },
-    {
-      "epoch": 2.845193722881664,
-      "grad_norm": 0.0022163500543683767,
-      "learning_rate": 0.00019999600738194694,
-      "loss": 46.0,
-      "step": 37213
-    },
-    {
-      "epoch": 2.8452701798650533,
-      "grad_norm": 0.0016734949313104153,
-      "learning_rate": 0.00019999600716730592,
-      "loss": 46.0,
-      "step": 37214
-    },
-    {
-      "epoch": 2.845346636848443,
-      "grad_norm": 0.004838340915739536,
-      "learning_rate": 0.0001999960069526591,
-      "loss": 46.0,
-      "step": 37215
-    },
-    {
-      "epoch": 2.845423093831833,
-      "grad_norm": 0.0007701439317315817,
-      "learning_rate": 0.00019999600673800653,
-      "loss": 46.0,
-      "step": 37216
-    },
-    {
-      "epoch": 2.8454995508152225,
-      "grad_norm": 0.0017121906857937574,
-      "learning_rate": 0.0001999960065233482,
-      "loss": 46.0,
-      "step": 37217
-    },
-    {
-      "epoch": 2.8455760077986123,
-      "grad_norm": 0.005509126465767622,
-      "learning_rate": 0.0001999960063086841,
-      "loss": 46.0,
-      "step": 37218
-    },
-    {
-      "epoch": 2.845652464782002,
-      "grad_norm": 0.002026584465056658,
-      "learning_rate": 0.0001999960060940142,
-      "loss": 46.0,
-      "step": 37219
-    },
-    {
-      "epoch": 2.845728921765392,
-      "grad_norm": 0.0017668914515525103,
-      "learning_rate": 0.00019999600587933855,
-      "loss": 46.0,
-      "step": 37220
-    },
-    {
-      "epoch": 2.8458053787487816,
-      "grad_norm": 0.0027744630351662636,
-      "learning_rate": 0.00019999600566465714,
-      "loss": 46.0,
-      "step": 37221
-    },
-    {
-      "epoch": 2.8458818357321713,
-      "grad_norm": 0.00184998894110322,
-      "learning_rate": 0.00019999600544996996,
-      "loss": 46.0,
-      "step": 37222
-    },
-    {
-      "epoch": 2.8459582927155607,
-      "grad_norm": 0.0008259746828116477,
-      "learning_rate": 0.000199996005235277,
-      "loss": 46.0,
-      "step": 37223
-    },
-    {
-      "epoch": 2.8460347496989504,
-      "grad_norm": 0.0012356044026091695,
-      "learning_rate": 0.0001999960050205783,
-      "loss": 46.0,
-      "step": 37224
-    },
-    {
-      "epoch": 2.84611120668234,
-      "grad_norm": 0.0015845209127292037,
-      "learning_rate": 0.0001999960048058738,
-      "loss": 46.0,
-      "step": 37225
-    },
-    {
-      "epoch": 2.84618766366573,
-      "grad_norm": 0.0034666426945477724,
-      "learning_rate": 0.00019999600459116352,
-      "loss": 46.0,
-      "step": 37226
-    },
-    {
-      "epoch": 2.8462641206491197,
-      "grad_norm": 0.004044574219733477,
-      "learning_rate": 0.0001999960043764475,
-      "loss": 46.0,
-      "step": 37227
-    },
-    {
-      "epoch": 2.8463405776325095,
-      "grad_norm": 0.0041269417852163315,
-      "learning_rate": 0.0001999960041617257,
-      "loss": 46.0,
-      "step": 37228
-    },
-    {
-      "epoch": 2.8464170346158992,
-      "grad_norm": 0.0009241164079867303,
-      "learning_rate": 0.00019999600394699814,
-      "loss": 46.0,
-      "step": 37229
-    },
-    {
-      "epoch": 2.846493491599289,
-      "grad_norm": 0.0013869010144844651,
-      "learning_rate": 0.0001999960037322648,
-      "loss": 46.0,
-      "step": 37230
-    },
-    {
-      "epoch": 2.8465699485826788,
-      "grad_norm": 0.0037373602390289307,
-      "learning_rate": 0.00019999600351752568,
-      "loss": 46.0,
-      "step": 37231
-    },
-    {
-      "epoch": 2.8466464055660685,
-      "grad_norm": 0.0059655578806996346,
-      "learning_rate": 0.00019999600330278082,
-      "loss": 46.0,
-      "step": 37232
-    },
-    {
-      "epoch": 2.8467228625494583,
-      "grad_norm": 0.0006762645207345486,
-      "learning_rate": 0.00019999600308803016,
-      "loss": 46.0,
-      "step": 37233
-    },
-    {
-      "epoch": 2.846799319532848,
-      "grad_norm": 0.003408893244341016,
-      "learning_rate": 0.00019999600287327372,
-      "loss": 46.0,
-      "step": 37234
-    },
-    {
-      "epoch": 2.846875776516238,
-      "grad_norm": 0.00320029235444963,
-      "learning_rate": 0.00019999600265851157,
-      "loss": 46.0,
-      "step": 37235
-    },
-    {
-      "epoch": 2.846952233499627,
-      "grad_norm": 0.002168607898056507,
-      "learning_rate": 0.00019999600244374359,
-      "loss": 46.0,
-      "step": 37236
-    },
-    {
-      "epoch": 2.847028690483017,
-      "grad_norm": 0.002000172156840563,
-      "learning_rate": 0.00019999600222896988,
-      "loss": 46.0,
-      "step": 37237
-    },
-    {
-      "epoch": 2.8471051474664066,
-      "grad_norm": 0.0013277019606903195,
-      "learning_rate": 0.0001999960020141904,
-      "loss": 46.0,
-      "step": 37238
-    },
-    {
-      "epoch": 2.8471816044497964,
-      "grad_norm": 0.0014426482375711203,
-      "learning_rate": 0.00019999600179940514,
-      "loss": 46.0,
-      "step": 37239
-    },
-    {
-      "epoch": 2.847258061433186,
-      "grad_norm": 0.0017250728560611606,
-      "learning_rate": 0.00019999600158461411,
-      "loss": 46.0,
-      "step": 37240
-    },
-    {
-      "epoch": 2.847334518416576,
-      "grad_norm": 0.0025461020413786173,
-      "learning_rate": 0.00019999600136981732,
-      "loss": 46.0,
-      "step": 37241
-    },
-    {
-      "epoch": 2.8474109753999657,
-      "grad_norm": 0.0033455220982432365,
-      "learning_rate": 0.00019999600115501472,
-      "loss": 46.0,
-      "step": 37242
-    },
-    {
-      "epoch": 2.8474874323833554,
-      "grad_norm": 0.0007954634493216872,
-      "learning_rate": 0.00019999600094020638,
-      "loss": 46.0,
-      "step": 37243
-    },
-    {
-      "epoch": 2.8475638893667448,
-      "grad_norm": 0.0012235864996910095,
-      "learning_rate": 0.00019999600072539227,
-      "loss": 46.0,
-      "step": 37244
-    },
-    {
-      "epoch": 2.8476403463501345,
-      "grad_norm": 0.0025959573686122894,
-      "learning_rate": 0.0001999960005105724,
-      "loss": 46.0,
-      "step": 37245
-    },
-    {
-      "epoch": 2.8477168033335243,
-      "grad_norm": 0.0024312231689691544,
-      "learning_rate": 0.00019999600029574674,
-      "loss": 46.0,
-      "step": 37246
-    },
-    {
-      "epoch": 2.847793260316914,
-      "grad_norm": 0.0012624245136976242,
-      "learning_rate": 0.00019999600008091534,
-      "loss": 46.0,
-      "step": 37247
-    },
-    {
-      "epoch": 2.847869717300304,
-      "grad_norm": 0.0014503926504403353,
-      "learning_rate": 0.00019999599986607816,
-      "loss": 46.0,
-      "step": 37248
-    },
-    {
-      "epoch": 2.8479461742836936,
-      "grad_norm": 0.004004387650638819,
-      "learning_rate": 0.0001999959996512352,
-      "loss": 46.0,
-      "step": 37249
-    },
-    {
-      "epoch": 2.8480226312670833,
-      "grad_norm": 0.004282256588339806,
-      "learning_rate": 0.00019999599943638647,
-      "loss": 46.0,
-      "step": 37250
-    },
-    {
-      "epoch": 2.848099088250473,
-      "grad_norm": 0.001295424997806549,
-      "learning_rate": 0.00019999599922153197,
-      "loss": 46.0,
-      "step": 37251
-    },
-    {
-      "epoch": 2.848175545233863,
-      "grad_norm": 0.002017940627411008,
-      "learning_rate": 0.00019999599900667172,
-      "loss": 46.0,
-      "step": 37252
-    },
-    {
-      "epoch": 2.8482520022172526,
-      "grad_norm": 0.0015080752782523632,
-      "learning_rate": 0.00019999599879180568,
-      "loss": 46.0,
-      "step": 37253
-    },
-    {
-      "epoch": 2.8483284592006424,
-      "grad_norm": 0.0021924860775470734,
-      "learning_rate": 0.0001999959985769339,
-      "loss": 46.0,
-      "step": 37254
-    },
-    {
-      "epoch": 2.848404916184032,
-      "grad_norm": 0.004107275977730751,
-      "learning_rate": 0.00019999599836205634,
-      "loss": 46.0,
-      "step": 37255
-    },
-    {
-      "epoch": 2.848481373167422,
-      "grad_norm": 0.004498360678553581,
-      "learning_rate": 0.00019999599814717297,
-      "loss": 46.0,
-      "step": 37256
-    },
-    {
-      "epoch": 2.8485578301508117,
-      "grad_norm": 0.0022525659296661615,
-      "learning_rate": 0.00019999599793228389,
-      "loss": 46.0,
-      "step": 37257
-    },
-    {
-      "epoch": 2.848634287134201,
-      "grad_norm": 0.0020462311804294586,
-      "learning_rate": 0.000199995997717389,
-      "loss": 46.0,
-      "step": 37258
-    },
-    {
-      "epoch": 2.8487107441175907,
-      "grad_norm": 0.001385206007398665,
-      "learning_rate": 0.00019999599750248836,
-      "loss": 46.0,
-      "step": 37259
-    },
-    {
-      "epoch": 2.8487872011009805,
-      "grad_norm": 0.002075792523100972,
-      "learning_rate": 0.00019999599728758193,
-      "loss": 46.0,
-      "step": 37260
-    },
-    {
-      "epoch": 2.8488636580843703,
-      "grad_norm": 0.0026271117385476828,
-      "learning_rate": 0.00019999599707266975,
-      "loss": 46.0,
-      "step": 37261
-    },
-    {
-      "epoch": 2.84894011506776,
-      "grad_norm": 0.006578900385648012,
-      "learning_rate": 0.0001999959968577518,
-      "loss": 46.0,
-      "step": 37262
-    },
-    {
-      "epoch": 2.8490165720511498,
-      "grad_norm": 0.0020474980119615793,
-      "learning_rate": 0.00019999599664282807,
-      "loss": 46.0,
-      "step": 37263
-    },
-    {
-      "epoch": 2.8490930290345395,
-      "grad_norm": 0.0017278738087043166,
-      "learning_rate": 0.00019999599642789856,
-      "loss": 46.0,
-      "step": 37264
-    },
-    {
-      "epoch": 2.8491694860179293,
-      "grad_norm": 0.0011812258744612336,
-      "learning_rate": 0.00019999599621296332,
-      "loss": 46.0,
-      "step": 37265
-    },
-    {
-      "epoch": 2.8492459430013186,
-      "grad_norm": 0.0034139524213969707,
-      "learning_rate": 0.0001999959959980223,
-      "loss": 46.0,
-      "step": 37266
-    },
-    {
-      "epoch": 2.8493223999847084,
-      "grad_norm": 0.0014752881834283471,
-      "learning_rate": 0.00019999599578307548,
-      "loss": 46.0,
-      "step": 37267
-    },
-    {
-      "epoch": 2.849398856968098,
-      "grad_norm": 0.0027909576892852783,
-      "learning_rate": 0.0001999959955681229,
-      "loss": 46.0,
-      "step": 37268
-    },
-    {
-      "epoch": 2.849475313951488,
-      "grad_norm": 0.0035066893324255943,
-      "learning_rate": 0.00019999599535316457,
-      "loss": 46.0,
-      "step": 37269
-    },
-    {
-      "epoch": 2.8495517709348777,
-      "grad_norm": 0.0019999733194708824,
-      "learning_rate": 0.00019999599513820048,
-      "loss": 46.0,
-      "step": 37270
-    },
-    {
-      "epoch": 2.8496282279182674,
-      "grad_norm": 0.0021603750064969063,
-      "learning_rate": 0.0001999959949232306,
-      "loss": 46.0,
-      "step": 37271
-    },
-    {
-      "epoch": 2.849704684901657,
-      "grad_norm": 0.002759114373475313,
-      "learning_rate": 0.00019999599470825493,
-      "loss": 46.0,
-      "step": 37272
-    },
-    {
-      "epoch": 2.849781141885047,
-      "grad_norm": 0.0017168933991342783,
-      "learning_rate": 0.00019999599449327352,
-      "loss": 46.0,
-      "step": 37273
-    },
-    {
-      "epoch": 2.8498575988684367,
-      "grad_norm": 0.0017353284638375044,
-      "learning_rate": 0.00019999599427828634,
-      "loss": 46.0,
-      "step": 37274
-    },
-    {
-      "epoch": 2.8499340558518265,
-      "grad_norm": 0.0037807871121913195,
-      "learning_rate": 0.0001999959940632934,
-      "loss": 46.0,
-      "step": 37275
-    },
-    {
-      "epoch": 2.8500105128352162,
-      "grad_norm": 0.0015400650445371866,
-      "learning_rate": 0.00019999599384829466,
-      "loss": 46.0,
-      "step": 37276
-    },
-    {
-      "epoch": 2.850086969818606,
-      "grad_norm": 0.003298710333183408,
-      "learning_rate": 0.0001999959936332902,
-      "loss": 46.0,
-      "step": 37277
-    },
-    {
-      "epoch": 2.8501634268019957,
-      "grad_norm": 0.0029223680030554533,
-      "learning_rate": 0.00019999599341827991,
-      "loss": 46.0,
-      "step": 37278
-    },
-    {
-      "epoch": 2.8502398837853855,
-      "grad_norm": 0.001601759227924049,
-      "learning_rate": 0.00019999599320326387,
-      "loss": 46.0,
-      "step": 37279
-    },
-    {
-      "epoch": 2.850316340768775,
-      "grad_norm": 0.0014337064931169152,
-      "learning_rate": 0.0001999959929882421,
-      "loss": 46.0,
-      "step": 37280
-    },
-    {
-      "epoch": 2.8503927977521646,
-      "grad_norm": 0.0012704854598268867,
-      "learning_rate": 0.00019999599277321453,
-      "loss": 46.0,
-      "step": 37281
-    },
-    {
-      "epoch": 2.8504692547355543,
-      "grad_norm": 0.003259579185396433,
-      "learning_rate": 0.00019999599255818117,
-      "loss": 46.0,
-      "step": 37282
-    },
-    {
-      "epoch": 2.850545711718944,
-      "grad_norm": 0.003714054124429822,
-      "learning_rate": 0.00019999599234314208,
-      "loss": 46.0,
-      "step": 37283
-    },
-    {
-      "epoch": 2.850622168702334,
-      "grad_norm": 0.003584110178053379,
-      "learning_rate": 0.00019999599212809722,
-      "loss": 46.0,
-      "step": 37284
-    },
-    {
-      "epoch": 2.8506986256857236,
-      "grad_norm": 0.004490654915571213,
-      "learning_rate": 0.00019999599191304656,
-      "loss": 46.0,
-      "step": 37285
-    },
-    {
-      "epoch": 2.8507750826691134,
-      "grad_norm": 0.0010733489179983735,
-      "learning_rate": 0.00019999599169799015,
-      "loss": 46.0,
-      "step": 37286
-    },
-    {
-      "epoch": 2.850851539652503,
-      "grad_norm": 0.000965977436862886,
-      "learning_rate": 0.00019999599148292797,
-      "loss": 46.0,
-      "step": 37287
-    },
-    {
-      "epoch": 2.8509279966358925,
-      "grad_norm": 0.003879459109157324,
-      "learning_rate": 0.00019999599126786,
-      "loss": 46.0,
-      "step": 37288
-    },
-    {
-      "epoch": 2.8510044536192822,
-      "grad_norm": 0.001358768087811768,
-      "learning_rate": 0.0001999959910527863,
-      "loss": 46.0,
-      "step": 37289
-    },
-    {
-      "epoch": 2.851080910602672,
-      "grad_norm": 0.001473011914640665,
-      "learning_rate": 0.0001999959908377068,
-      "loss": 46.0,
-      "step": 37290
-    },
-    {
-      "epoch": 2.8511573675860618,
-      "grad_norm": 0.0009068356594070792,
-      "learning_rate": 0.00019999599062262152,
-      "loss": 46.0,
-      "step": 37291
-    },
-    {
-      "epoch": 2.8512338245694515,
-      "grad_norm": 0.004027722403407097,
-      "learning_rate": 0.0001999959904075305,
-      "loss": 46.0,
-      "step": 37292
-    },
-    {
-      "epoch": 2.8513102815528413,
-      "grad_norm": 0.0007677121320739388,
-      "learning_rate": 0.0001999959901924337,
-      "loss": 46.0,
-      "step": 37293
-    },
-    {
-      "epoch": 2.851386738536231,
-      "grad_norm": 0.0021411182824522257,
-      "learning_rate": 0.0001999959899773311,
-      "loss": 46.0,
-      "step": 37294
-    },
-    {
-      "epoch": 2.851463195519621,
-      "grad_norm": 0.0012926168274134398,
-      "learning_rate": 0.00019999598976222277,
-      "loss": 46.0,
-      "step": 37295
-    },
-    {
-      "epoch": 2.8515396525030106,
-      "grad_norm": 0.005833826027810574,
-      "learning_rate": 0.00019999598954710868,
-      "loss": 46.0,
-      "step": 37296
-    },
-    {
-      "epoch": 2.8516161094864003,
-      "grad_norm": 0.0018114744452759624,
-      "learning_rate": 0.0001999959893319888,
-      "loss": 46.0,
-      "step": 37297
-    },
-    {
-      "epoch": 2.85169256646979,
-      "grad_norm": 0.0027112492825835943,
-      "learning_rate": 0.00019999598911686313,
-      "loss": 46.0,
-      "step": 37298
-    },
-    {
-      "epoch": 2.85176902345318,
-      "grad_norm": 0.0009759074309840798,
-      "learning_rate": 0.00019999598890173173,
-      "loss": 46.0,
-      "step": 37299
-    },
-    {
-      "epoch": 2.8518454804365696,
-      "grad_norm": 0.0025301408022642136,
-      "learning_rate": 0.00019999598868659455,
-      "loss": 46.0,
-      "step": 37300
-    },
-    {
-      "epoch": 2.8519219374199594,
-      "grad_norm": 0.0020652299281209707,
-      "learning_rate": 0.0001999959884714516,
-      "loss": 46.0,
-      "step": 37301
-    },
-    {
-      "epoch": 2.8519983944033487,
-      "grad_norm": 0.0029327061492949724,
-      "learning_rate": 0.00019999598825630287,
-      "loss": 46.0,
-      "step": 37302
-    },
-    {
-      "epoch": 2.8520748513867384,
-      "grad_norm": 0.0019536330364644527,
-      "learning_rate": 0.0001999959880411484,
-      "loss": 46.0,
-      "step": 37303
-    },
-    {
-      "epoch": 2.852151308370128,
-      "grad_norm": 0.002208857797086239,
-      "learning_rate": 0.00019999598782598815,
-      "loss": 46.0,
-      "step": 37304
-    },
-    {
-      "epoch": 2.852227765353518,
-      "grad_norm": 0.0013169703306630254,
-      "learning_rate": 0.00019999598761082207,
-      "loss": 46.0,
-      "step": 37305
-    },
-    {
-      "epoch": 2.8523042223369077,
-      "grad_norm": 0.003879035357385874,
-      "learning_rate": 0.00019999598739565028,
-      "loss": 46.0,
-      "step": 37306
-    },
-    {
-      "epoch": 2.8523806793202975,
-      "grad_norm": 0.0029065320268273354,
-      "learning_rate": 0.00019999598718047271,
-      "loss": 46.0,
-      "step": 37307
-    },
-    {
-      "epoch": 2.8524571363036872,
-      "grad_norm": 0.0037449600640684366,
-      "learning_rate": 0.00019999598696528937,
-      "loss": 46.0,
-      "step": 37308
-    },
-    {
-      "epoch": 2.852533593287077,
-      "grad_norm": 0.0030812923796474934,
-      "learning_rate": 0.00019999598675010026,
-      "loss": 46.0,
-      "step": 37309
-    },
-    {
-      "epoch": 2.8526100502704663,
-      "grad_norm": 0.004042427055537701,
-      "learning_rate": 0.0001999959865349054,
-      "loss": 46.0,
-      "step": 37310
-    },
-    {
-      "epoch": 2.852686507253856,
-      "grad_norm": 0.002936464501544833,
-      "learning_rate": 0.00019999598631970474,
-      "loss": 46.0,
-      "step": 37311
-    },
-    {
-      "epoch": 2.852762964237246,
-      "grad_norm": 0.0017784219235181808,
-      "learning_rate": 0.00019999598610449834,
-      "loss": 46.0,
-      "step": 37312
-    },
-    {
-      "epoch": 2.8528394212206356,
-      "grad_norm": 0.002888568677008152,
-      "learning_rate": 0.00019999598588928613,
-      "loss": 46.0,
-      "step": 37313
-    },
-    {
-      "epoch": 2.8529158782040254,
-      "grad_norm": 0.0008542234427295625,
-      "learning_rate": 0.0001999959856740682,
-      "loss": 46.0,
-      "step": 37314
-    },
-    {
-      "epoch": 2.852992335187415,
-      "grad_norm": 0.0014698245795443654,
-      "learning_rate": 0.00019999598545884448,
-      "loss": 46.0,
-      "step": 37315
-    },
-    {
-      "epoch": 2.853068792170805,
-      "grad_norm": 0.0016620280221104622,
-      "learning_rate": 0.00019999598524361498,
-      "loss": 46.0,
-      "step": 37316
-    },
-    {
-      "epoch": 2.8531452491541947,
-      "grad_norm": 0.0007210882613435388,
-      "learning_rate": 0.00019999598502837973,
-      "loss": 46.0,
-      "step": 37317
-    },
-    {
-      "epoch": 2.8532217061375844,
-      "grad_norm": 0.0018330198945477605,
-      "learning_rate": 0.0001999959848131387,
-      "loss": 46.0,
-      "step": 37318
-    },
-    {
-      "epoch": 2.853298163120974,
-      "grad_norm": 0.00232741073705256,
-      "learning_rate": 0.00019999598459789192,
-      "loss": 46.0,
-      "step": 37319
-    },
-    {
-      "epoch": 2.853374620104364,
-      "grad_norm": 0.002274435944855213,
-      "learning_rate": 0.00019999598438263932,
-      "loss": 46.0,
-      "step": 37320
-    },
-    {
-      "epoch": 2.8534510770877537,
-      "grad_norm": 0.0034586244728416204,
-      "learning_rate": 0.00019999598416738099,
-      "loss": 46.0,
-      "step": 37321
-    },
-    {
-      "epoch": 2.8535275340711435,
-      "grad_norm": 0.0017230635276064277,
-      "learning_rate": 0.00019999598395211687,
-      "loss": 46.0,
-      "step": 37322
-    },
-    {
-      "epoch": 2.853603991054533,
-      "grad_norm": 0.0010296761756762862,
-      "learning_rate": 0.000199995983736847,
-      "loss": 46.0,
-      "step": 37323
-    },
-    {
-      "epoch": 2.8536804480379225,
-      "grad_norm": 0.0014594501117244363,
-      "learning_rate": 0.00019999598352157135,
-      "loss": 46.0,
-      "step": 37324
-    },
-    {
-      "epoch": 2.8537569050213123,
-      "grad_norm": 0.003524273168295622,
-      "learning_rate": 0.00019999598330628995,
-      "loss": 46.0,
-      "step": 37325
-    },
-    {
-      "epoch": 2.853833362004702,
-      "grad_norm": 0.005894568748772144,
-      "learning_rate": 0.00019999598309100277,
-      "loss": 46.0,
-      "step": 37326
-    },
-    {
-      "epoch": 2.853909818988092,
-      "grad_norm": 0.001328212209045887,
-      "learning_rate": 0.00019999598287570982,
-      "loss": 46.0,
-      "step": 37327
-    },
-    {
-      "epoch": 2.8539862759714816,
-      "grad_norm": 0.003787433495745063,
-      "learning_rate": 0.0001999959826604111,
-      "loss": 46.0,
-      "step": 37328
-    },
-    {
-      "epoch": 2.8540627329548713,
-      "grad_norm": 0.0021362300030887127,
-      "learning_rate": 0.0001999959824451066,
-      "loss": 46.0,
-      "step": 37329
-    },
-    {
-      "epoch": 2.854139189938261,
-      "grad_norm": 0.0023188029881566763,
-      "learning_rate": 0.00019999598222979632,
-      "loss": 46.0,
-      "step": 37330
-    },
-    {
-      "epoch": 2.854215646921651,
-      "grad_norm": 0.0011016883654519916,
-      "learning_rate": 0.0001999959820144803,
-      "loss": 46.0,
-      "step": 37331
-    },
-    {
-      "epoch": 2.85429210390504,
-      "grad_norm": 0.004367099609225988,
-      "learning_rate": 0.0001999959817991585,
-      "loss": 46.0,
-      "step": 37332
-    },
-    {
-      "epoch": 2.85436856088843,
-      "grad_norm": 0.004461745265871286,
-      "learning_rate": 0.00019999598158383094,
-      "loss": 46.0,
-      "step": 37333
-    },
-    {
-      "epoch": 2.8544450178718197,
-      "grad_norm": 0.004328207112848759,
-      "learning_rate": 0.0001999959813684976,
-      "loss": 46.0,
-      "step": 37334
-    },
-    {
-      "epoch": 2.8545214748552095,
-      "grad_norm": 0.0012120924657210708,
-      "learning_rate": 0.0001999959811531585,
-      "loss": 46.0,
-      "step": 37335
-    },
-    {
-      "epoch": 2.8545979318385992,
-      "grad_norm": 0.0009794754441827536,
-      "learning_rate": 0.00019999598093781363,
-      "loss": 46.0,
-      "step": 37336
-    },
-    {
-      "epoch": 2.854674388821989,
-      "grad_norm": 0.004440579097718,
-      "learning_rate": 0.00019999598072246294,
-      "loss": 46.0,
-      "step": 37337
-    },
-    {
-      "epoch": 2.8547508458053787,
-      "grad_norm": 0.001987901283428073,
-      "learning_rate": 0.00019999598050710656,
-      "loss": 46.0,
-      "step": 37338
-    },
-    {
-      "epoch": 2.8548273027887685,
-      "grad_norm": 0.001043384545482695,
-      "learning_rate": 0.00019999598029174438,
-      "loss": 46.0,
-      "step": 37339
-    },
-    {
-      "epoch": 2.8549037597721583,
-      "grad_norm": 0.002521959599107504,
-      "learning_rate": 0.0001999959800763764,
-      "loss": 46.0,
-      "step": 37340
-    },
-    {
-      "epoch": 2.854980216755548,
-      "grad_norm": 0.0010131163289770484,
-      "learning_rate": 0.00019999597986100268,
-      "loss": 46.0,
-      "step": 37341
-    },
-    {
-      "epoch": 2.855056673738938,
-      "grad_norm": 0.001875092275440693,
-      "learning_rate": 0.0001999959796456232,
-      "loss": 46.0,
-      "step": 37342
-    },
-    {
-      "epoch": 2.8551331307223276,
-      "grad_norm": 0.0026953311171382666,
-      "learning_rate": 0.00019999597943023793,
-      "loss": 46.0,
-      "step": 37343
-    },
-    {
-      "epoch": 2.8552095877057173,
-      "grad_norm": 0.006256473250687122,
-      "learning_rate": 0.00019999597921484692,
-      "loss": 46.0,
-      "step": 37344
-    },
-    {
-      "epoch": 2.8552860446891066,
-      "grad_norm": 0.0019740923307836056,
-      "learning_rate": 0.0001999959789994501,
-      "loss": 46.0,
-      "step": 37345
-    },
-    {
-      "epoch": 2.8553625016724964,
-      "grad_norm": 0.0013334328541532159,
-      "learning_rate": 0.00019999597878404756,
-      "loss": 46.0,
-      "step": 37346
-    },
-    {
-      "epoch": 2.855438958655886,
-      "grad_norm": 0.0008456657524220645,
-      "learning_rate": 0.00019999597856863922,
-      "loss": 46.0,
-      "step": 37347
-    },
-    {
-      "epoch": 2.855515415639276,
-      "grad_norm": 0.003338826121762395,
-      "learning_rate": 0.0001999959783532251,
-      "loss": 46.0,
-      "step": 37348
-    },
-    {
-      "epoch": 2.8555918726226657,
-      "grad_norm": 0.0028465476352721453,
-      "learning_rate": 0.00019999597813780522,
-      "loss": 46.0,
-      "step": 37349
-    },
-    {
-      "epoch": 2.8556683296060554,
-      "grad_norm": 0.002293667057529092,
-      "learning_rate": 0.0001999959779223796,
-      "loss": 46.0,
-      "step": 37350
-    },
-    {
-      "epoch": 2.855744786589445,
-      "grad_norm": 0.0033546965569257736,
-      "learning_rate": 0.00019999597770694816,
-      "loss": 46.0,
-      "step": 37351
-    },
-    {
-      "epoch": 2.855821243572835,
-      "grad_norm": 0.001735328696668148,
-      "learning_rate": 0.00019999597749151098,
-      "loss": 46.0,
-      "step": 37352
-    },
-    {
-      "epoch": 2.8558977005562247,
-      "grad_norm": 0.0040353089570999146,
-      "learning_rate": 0.00019999597727606803,
-      "loss": 46.0,
-      "step": 37353
-    },
-    {
-      "epoch": 2.855974157539614,
-      "grad_norm": 0.0043499041348695755,
-      "learning_rate": 0.0001999959770606193,
-      "loss": 46.0,
-      "step": 37354
-    },
-    {
-      "epoch": 2.856050614523004,
-      "grad_norm": 0.0015060887672007084,
-      "learning_rate": 0.00019999597684516483,
-      "loss": 46.0,
-      "step": 37355
-    },
-    {
-      "epoch": 2.8561270715063936,
-      "grad_norm": 0.0027334417682141066,
-      "learning_rate": 0.00019999597662970456,
-      "loss": 46.0,
-      "step": 37356
-    },
-    {
-      "epoch": 2.8562035284897833,
-      "grad_norm": 0.0019144502002745867,
-      "learning_rate": 0.0001999959764142385,
-      "loss": 46.0,
-      "step": 37357
-    },
-    {
-      "epoch": 2.856279985473173,
-      "grad_norm": 0.0006307953153736889,
-      "learning_rate": 0.00019999597619876675,
-      "loss": 46.0,
-      "step": 37358
-    },
-    {
-      "epoch": 2.856356442456563,
-      "grad_norm": 0.0009853419614955783,
-      "learning_rate": 0.00019999597598328915,
-      "loss": 46.0,
-      "step": 37359
-    },
-    {
-      "epoch": 2.8564328994399526,
-      "grad_norm": 0.006186658050864935,
-      "learning_rate": 0.00019999597576780581,
-      "loss": 46.0,
-      "step": 37360
-    },
-    {
-      "epoch": 2.8565093564233424,
-      "grad_norm": 0.011974647641181946,
-      "learning_rate": 0.0001999959755523167,
-      "loss": 46.0,
-      "step": 37361
-    },
-    {
-      "epoch": 2.856585813406732,
-      "grad_norm": 0.001817780314013362,
-      "learning_rate": 0.00019999597533682182,
-      "loss": 46.0,
-      "step": 37362
-    },
-    {
-      "epoch": 2.856662270390122,
-      "grad_norm": 0.0004092817544005811,
-      "learning_rate": 0.00019999597512132119,
-      "loss": 46.0,
-      "step": 37363
-    },
-    {
-      "epoch": 2.8567387273735116,
-      "grad_norm": 0.0012719588121399283,
-      "learning_rate": 0.00019999597490581475,
-      "loss": 46.0,
-      "step": 37364
-    },
-    {
-      "epoch": 2.8568151843569014,
-      "grad_norm": 0.002170251216739416,
-      "learning_rate": 0.00019999597469030257,
-      "loss": 46.0,
-      "step": 37365
-    },
-    {
-      "epoch": 2.856891641340291,
-      "grad_norm": 0.005010946653783321,
-      "learning_rate": 0.00019999597447478462,
-      "loss": 46.0,
-      "step": 37366
-    },
-    {
-      "epoch": 2.8569680983236805,
-      "grad_norm": 0.0014615219552069902,
-      "learning_rate": 0.0001999959742592609,
-      "loss": 46.0,
-      "step": 37367
-    },
-    {
-      "epoch": 2.8570445553070702,
-      "grad_norm": 0.001642643823288381,
-      "learning_rate": 0.00019999597404373143,
-      "loss": 46.0,
-      "step": 37368
-    },
-    {
-      "epoch": 2.85712101229046,
-      "grad_norm": 0.002189437858760357,
-      "learning_rate": 0.00019999597382819615,
-      "loss": 46.0,
-      "step": 37369
-    },
-    {
-      "epoch": 2.8571974692738498,
-      "grad_norm": 0.0028046458028256893,
-      "learning_rate": 0.00019999597361265514,
-      "loss": 46.0,
-      "step": 37370
-    },
-    {
-      "epoch": 2.8572739262572395,
-      "grad_norm": 0.0020366355311125517,
-      "learning_rate": 0.00019999597339710832,
-      "loss": 46.0,
-      "step": 37371
-    },
-    {
-      "epoch": 2.8573503832406293,
-      "grad_norm": 0.0034670080058276653,
-      "learning_rate": 0.00019999597318155575,
-      "loss": 46.0,
-      "step": 37372
-    },
-    {
-      "epoch": 2.857426840224019,
-      "grad_norm": 0.0026196478866040707,
-      "learning_rate": 0.00019999597296599744,
-      "loss": 46.0,
-      "step": 37373
-    },
-    {
-      "epoch": 2.857503297207409,
-      "grad_norm": 0.0023079689126461744,
-      "learning_rate": 0.0001999959727504333,
-      "loss": 46.0,
-      "step": 37374
-    },
-    {
-      "epoch": 2.857579754190798,
-      "grad_norm": 0.0027341623790562153,
-      "learning_rate": 0.00019999597253486344,
-      "loss": 46.0,
-      "step": 37375
-    },
-    {
-      "epoch": 2.857656211174188,
-      "grad_norm": 0.0016881534829735756,
-      "learning_rate": 0.00019999597231928779,
-      "loss": 46.0,
-      "step": 37376
-    },
-    {
-      "epoch": 2.8577326681575776,
-      "grad_norm": 0.002138593001291156,
-      "learning_rate": 0.00019999597210370638,
-      "loss": 46.0,
-      "step": 37377
-    },
-    {
-      "epoch": 2.8578091251409674,
-      "grad_norm": 0.001564140198752284,
-      "learning_rate": 0.0001999959718881192,
-      "loss": 46.0,
-      "step": 37378
-    },
-    {
-      "epoch": 2.857885582124357,
-      "grad_norm": 0.0022983925882726908,
-      "learning_rate": 0.00019999597167252625,
-      "loss": 46.0,
-      "step": 37379
-    },
-    {
-      "epoch": 2.857962039107747,
-      "grad_norm": 0.0012937862193211913,
-      "learning_rate": 0.00019999597145692753,
-      "loss": 46.0,
-      "step": 37380
-    },
-    {
-      "epoch": 2.8580384960911367,
-      "grad_norm": 0.0016106325201690197,
-      "learning_rate": 0.00019999597124132303,
-      "loss": 46.0,
-      "step": 37381
-    },
-    {
-      "epoch": 2.8581149530745265,
-      "grad_norm": 0.0027820770628750324,
-      "learning_rate": 0.00019999597102571276,
-      "loss": 46.0,
-      "step": 37382
-    },
-    {
-      "epoch": 2.858191410057916,
-      "grad_norm": 0.0036041252315044403,
-      "learning_rate": 0.00019999597081009674,
-      "loss": 46.0,
-      "step": 37383
-    },
-    {
-      "epoch": 2.858267867041306,
-      "grad_norm": 0.0023981030099093914,
-      "learning_rate": 0.00019999597059447495,
-      "loss": 46.0,
-      "step": 37384
-    },
-    {
-      "epoch": 2.8583443240246957,
-      "grad_norm": 0.0021879670675843954,
-      "learning_rate": 0.00019999597037884736,
-      "loss": 46.0,
-      "step": 37385
-    },
-    {
-      "epoch": 2.8584207810080855,
-      "grad_norm": 0.0028004725463688374,
-      "learning_rate": 0.00019999597016321402,
-      "loss": 46.0,
-      "step": 37386
-    },
-    {
-      "epoch": 2.8584972379914753,
-      "grad_norm": 0.00290484051220119,
-      "learning_rate": 0.00019999596994757493,
-      "loss": 46.0,
-      "step": 37387
-    },
-    {
-      "epoch": 2.858573694974865,
-      "grad_norm": 0.0031083980575203896,
-      "learning_rate": 0.00019999596973193005,
-      "loss": 46.0,
-      "step": 37388
-    },
-    {
-      "epoch": 2.8586501519582543,
-      "grad_norm": 0.0032578168902546167,
-      "learning_rate": 0.00019999596951627942,
-      "loss": 46.0,
-      "step": 37389
-    },
-    {
-      "epoch": 2.858726608941644,
-      "grad_norm": 0.0015673382440581918,
-      "learning_rate": 0.000199995969300623,
-      "loss": 46.0,
-      "step": 37390
-    },
-    {
-      "epoch": 2.858803065925034,
-      "grad_norm": 0.0029307957738637924,
-      "learning_rate": 0.0001999959690849608,
-      "loss": 46.0,
-      "step": 37391
-    },
-    {
-      "epoch": 2.8588795229084236,
-      "grad_norm": 0.0006896536215208471,
-      "learning_rate": 0.00019999596886929286,
-      "loss": 46.0,
-      "step": 37392
-    },
-    {
-      "epoch": 2.8589559798918134,
-      "grad_norm": 0.0037042410112917423,
-      "learning_rate": 0.0001999959686536191,
-      "loss": 46.0,
-      "step": 37393
-    },
-    {
-      "epoch": 2.859032436875203,
-      "grad_norm": 0.0011286704102531075,
-      "learning_rate": 0.00019999596843793964,
-      "loss": 46.0,
-      "step": 37394
-    },
-    {
-      "epoch": 2.859108893858593,
-      "grad_norm": 0.003778767306357622,
-      "learning_rate": 0.0001999959682222544,
-      "loss": 46.0,
-      "step": 37395
-    },
-    {
-      "epoch": 2.8591853508419827,
-      "grad_norm": 0.0010023026261478662,
-      "learning_rate": 0.00019999596800656335,
-      "loss": 46.0,
-      "step": 37396
-    },
-    {
-      "epoch": 2.859261807825372,
-      "grad_norm": 0.003019100520759821,
-      "learning_rate": 0.00019999596779086653,
-      "loss": 46.0,
-      "step": 37397
-    },
-    {
-      "epoch": 2.8593382648087617,
-      "grad_norm": 0.002212079707533121,
-      "learning_rate": 0.00019999596757516396,
-      "loss": 46.0,
-      "step": 37398
-    },
-    {
-      "epoch": 2.8594147217921515,
-      "grad_norm": 0.00805040542036295,
-      "learning_rate": 0.00019999596735945565,
-      "loss": 46.0,
-      "step": 37399
-    },
-    {
-      "epoch": 2.8594911787755413,
-      "grad_norm": 0.0030057288240641356,
-      "learning_rate": 0.00019999596714374154,
-      "loss": 46.0,
-      "step": 37400
-    },
-    {
-      "epoch": 2.859567635758931,
-      "grad_norm": 0.0024373324122279882,
-      "learning_rate": 0.00019999596692802169,
-      "loss": 46.0,
-      "step": 37401
-    },
-    {
-      "epoch": 2.859644092742321,
-      "grad_norm": 0.0017328665126115084,
-      "learning_rate": 0.000199995966712296,
-      "loss": 46.0,
-      "step": 37402
-    },
-    {
-      "epoch": 2.8597205497257105,
-      "grad_norm": 0.00064315419876948,
-      "learning_rate": 0.0001999959664965646,
-      "loss": 46.0,
-      "step": 37403
-    },
-    {
-      "epoch": 2.8597970067091003,
-      "grad_norm": 0.001285726553760469,
-      "learning_rate": 0.00019999596628082742,
-      "loss": 46.0,
-      "step": 37404
-    },
-    {
-      "epoch": 2.85987346369249,
-      "grad_norm": 0.0028490587137639523,
-      "learning_rate": 0.00019999596606508447,
-      "loss": 46.0,
-      "step": 37405
-    },
-    {
-      "epoch": 2.85994992067588,
-      "grad_norm": 0.004146496765315533,
-      "learning_rate": 0.00019999596584933574,
-      "loss": 46.0,
-      "step": 37406
-    },
-    {
-      "epoch": 2.8600263776592696,
-      "grad_norm": 0.002788916928693652,
-      "learning_rate": 0.00019999596563358125,
-      "loss": 46.0,
-      "step": 37407
-    },
-    {
-      "epoch": 2.8601028346426594,
-      "grad_norm": 0.003172248136252165,
-      "learning_rate": 0.000199995965417821,
-      "loss": 46.0,
-      "step": 37408
-    },
-    {
-      "epoch": 2.860179291626049,
-      "grad_norm": 0.001350397476926446,
-      "learning_rate": 0.00019999596520205496,
-      "loss": 46.0,
-      "step": 37409
-    },
-    {
-      "epoch": 2.860255748609439,
-      "grad_norm": 0.0021407592575997114,
-      "learning_rate": 0.00019999596498628317,
-      "loss": 46.0,
-      "step": 37410
-    },
-    {
-      "epoch": 2.860332205592828,
-      "grad_norm": 0.002006686059758067,
-      "learning_rate": 0.0001999959647705056,
-      "loss": 46.0,
-      "step": 37411
-    },
-    {
-      "epoch": 2.860408662576218,
-      "grad_norm": 0.0014944759896025062,
-      "learning_rate": 0.00019999596455472224,
-      "loss": 46.0,
-      "step": 37412
-    },
-    {
-      "epoch": 2.8604851195596077,
-      "grad_norm": 0.0020433547906577587,
-      "learning_rate": 0.00019999596433893316,
-      "loss": 46.0,
-      "step": 37413
-    },
-    {
-      "epoch": 2.8605615765429975,
-      "grad_norm": 0.0008167953928932548,
-      "learning_rate": 0.00019999596412313827,
-      "loss": 46.0,
-      "step": 37414
-    },
-    {
-      "epoch": 2.8606380335263872,
-      "grad_norm": 0.0013444479554891586,
-      "learning_rate": 0.00019999596390733764,
-      "loss": 46.0,
-      "step": 37415
-    },
-    {
-      "epoch": 2.860714490509777,
-      "grad_norm": 0.002432076493278146,
-      "learning_rate": 0.0001999959636915312,
-      "loss": 46.0,
-      "step": 37416
-    },
-    {
-      "epoch": 2.8607909474931668,
-      "grad_norm": 0.0043581328354775906,
-      "learning_rate": 0.00019999596347571903,
-      "loss": 46.0,
-      "step": 37417
-    },
-    {
-      "epoch": 2.8608674044765565,
-      "grad_norm": 0.000905010849237442,
-      "learning_rate": 0.00019999596325990105,
-      "loss": 46.0,
-      "step": 37418
-    },
-    {
-      "epoch": 2.860943861459946,
-      "grad_norm": 0.0022833244875073433,
-      "learning_rate": 0.00019999596304407736,
-      "loss": 46.0,
-      "step": 37419
-    },
-    {
-      "epoch": 2.8610203184433356,
-      "grad_norm": 0.00285896984860301,
-      "learning_rate": 0.00019999596282824786,
-      "loss": 46.0,
-      "step": 37420
-    },
-    {
-      "epoch": 2.8610967754267254,
-      "grad_norm": 0.003195906523615122,
-      "learning_rate": 0.0001999959626124126,
-      "loss": 46.0,
-      "step": 37421
-    },
-    {
-      "epoch": 2.861173232410115,
-      "grad_norm": 0.017030026763677597,
-      "learning_rate": 0.00019999596239657157,
-      "loss": 46.0,
-      "step": 37422
-    },
-    {
-      "epoch": 2.861249689393505,
-      "grad_norm": 0.000928624824155122,
-      "learning_rate": 0.00019999596218072478,
-      "loss": 46.0,
-      "step": 37423
-    },
-    {
-      "epoch": 2.8613261463768946,
-      "grad_norm": 0.0011592790251597762,
-      "learning_rate": 0.00019999596196487222,
-      "loss": 46.0,
-      "step": 37424
-    },
-    {
-      "epoch": 2.8614026033602844,
-      "grad_norm": 0.002315193647518754,
-      "learning_rate": 0.00019999596174901388,
-      "loss": 46.0,
-      "step": 37425
-    },
-    {
-      "epoch": 2.861479060343674,
-      "grad_norm": 0.0009720303351059556,
-      "learning_rate": 0.00019999596153314977,
-      "loss": 46.0,
-      "step": 37426
-    },
-    {
-      "epoch": 2.861555517327064,
-      "grad_norm": 0.003237604396417737,
-      "learning_rate": 0.0001999959613172799,
-      "loss": 46.0,
-      "step": 37427
-    },
-    {
-      "epoch": 2.8616319743104537,
-      "grad_norm": 0.001822356367483735,
-      "learning_rate": 0.00019999596110140426,
-      "loss": 46.0,
-      "step": 37428
-    },
-    {
-      "epoch": 2.8617084312938434,
-      "grad_norm": 0.0040573785081505775,
-      "learning_rate": 0.00019999596088552283,
-      "loss": 46.0,
-      "step": 37429
-    },
-    {
-      "epoch": 2.861784888277233,
-      "grad_norm": 0.0006334678619168699,
-      "learning_rate": 0.00019999596066963565,
-      "loss": 46.0,
-      "step": 37430
-    },
-    {
-      "epoch": 2.861861345260623,
-      "grad_norm": 0.0012000094866380095,
-      "learning_rate": 0.0001999959604537427,
-      "loss": 46.0,
-      "step": 37431
-    },
-    {
-      "epoch": 2.8619378022440127,
-      "grad_norm": 0.003927180077880621,
-      "learning_rate": 0.00019999596023784398,
-      "loss": 46.0,
-      "step": 37432
-    },
-    {
-      "epoch": 2.862014259227402,
-      "grad_norm": 0.0008652122342027724,
-      "learning_rate": 0.00019999596002193948,
-      "loss": 46.0,
-      "step": 37433
-    },
-    {
-      "epoch": 2.862090716210792,
-      "grad_norm": 0.00506264204159379,
-      "learning_rate": 0.0001999959598060292,
-      "loss": 46.0,
-      "step": 37434
-    },
-    {
-      "epoch": 2.8621671731941816,
-      "grad_norm": 0.001470347517170012,
-      "learning_rate": 0.0001999959595901132,
-      "loss": 46.0,
-      "step": 37435
-    },
-    {
-      "epoch": 2.8622436301775713,
-      "grad_norm": 0.0044128550216555595,
-      "learning_rate": 0.0001999959593741914,
-      "loss": 46.0,
-      "step": 37436
-    },
-    {
-      "epoch": 2.862320087160961,
-      "grad_norm": 0.0017818250926211476,
-      "learning_rate": 0.0001999959591582638,
-      "loss": 46.0,
-      "step": 37437
-    },
-    {
-      "epoch": 2.862396544144351,
-      "grad_norm": 0.0020803522784262896,
-      "learning_rate": 0.0001999959589423305,
-      "loss": 46.0,
-      "step": 37438
-    },
-    {
-      "epoch": 2.8624730011277406,
-      "grad_norm": 0.0012987942900508642,
-      "learning_rate": 0.0001999959587263914,
-      "loss": 46.0,
-      "step": 37439
-    },
-    {
-      "epoch": 2.8625494581111304,
-      "grad_norm": 0.002318762708455324,
-      "learning_rate": 0.0001999959585104465,
-      "loss": 46.0,
-      "step": 37440
-    },
-    {
-      "epoch": 2.8626259150945197,
-      "grad_norm": 0.0029196382965892553,
-      "learning_rate": 0.00019999595829449588,
-      "loss": 46.0,
-      "step": 37441
-    },
-    {
-      "epoch": 2.8627023720779095,
-      "grad_norm": 0.003180979983881116,
-      "learning_rate": 0.00019999595807853945,
-      "loss": 46.0,
-      "step": 37442
-    },
-    {
-      "epoch": 2.862778829061299,
-      "grad_norm": 0.0036485048476606607,
-      "learning_rate": 0.00019999595786257727,
-      "loss": 46.0,
-      "step": 37443
-    },
-    {
-      "epoch": 2.862855286044689,
-      "grad_norm": 0.0014625238254666328,
-      "learning_rate": 0.00019999595764660932,
-      "loss": 46.0,
-      "step": 37444
-    },
-    {
-      "epoch": 2.8629317430280787,
-      "grad_norm": 0.001224764622747898,
-      "learning_rate": 0.0001999959574306356,
-      "loss": 46.0,
-      "step": 37445
-    },
-    {
-      "epoch": 2.8630082000114685,
-      "grad_norm": 0.004123813472688198,
-      "learning_rate": 0.0001999959572146561,
-      "loss": 46.0,
-      "step": 37446
-    },
-    {
-      "epoch": 2.8630846569948583,
-      "grad_norm": 0.002723866840824485,
-      "learning_rate": 0.00019999595699867083,
-      "loss": 46.0,
-      "step": 37447
-    },
-    {
-      "epoch": 2.863161113978248,
-      "grad_norm": 0.0009698011563159525,
-      "learning_rate": 0.00019999595678267982,
-      "loss": 46.0,
-      "step": 37448
-    },
-    {
-      "epoch": 2.863237570961638,
-      "grad_norm": 0.0014715880388393998,
-      "learning_rate": 0.00019999595656668303,
-      "loss": 46.0,
-      "step": 37449
-    },
-    {
-      "epoch": 2.8633140279450275,
-      "grad_norm": 0.003595031565055251,
-      "learning_rate": 0.00019999595635068044,
-      "loss": 46.0,
-      "step": 37450
-    },
-    {
-      "epoch": 2.8633904849284173,
-      "grad_norm": 0.0008930552867241204,
-      "learning_rate": 0.00019999595613467213,
-      "loss": 46.0,
-      "step": 37451
-    },
-    {
-      "epoch": 2.863466941911807,
-      "grad_norm": 0.006320401560515165,
-      "learning_rate": 0.00019999595591865802,
-      "loss": 46.0,
-      "step": 37452
-    },
-    {
-      "epoch": 2.863543398895197,
-      "grad_norm": 0.0045289937406778336,
-      "learning_rate": 0.0001999959557026381,
-      "loss": 46.0,
-      "step": 37453
-    },
-    {
-      "epoch": 2.8636198558785866,
-      "grad_norm": 0.0014551301719620824,
-      "learning_rate": 0.00019999595548661248,
-      "loss": 46.0,
-      "step": 37454
-    },
-    {
-      "epoch": 2.863696312861976,
-      "grad_norm": 0.002317196223884821,
-      "learning_rate": 0.00019999595527058105,
-      "loss": 46.0,
-      "step": 37455
-    },
-    {
-      "epoch": 2.8637727698453657,
-      "grad_norm": 0.003183969995006919,
-      "learning_rate": 0.00019999595505454387,
-      "loss": 46.0,
-      "step": 37456
-    },
-    {
-      "epoch": 2.8638492268287554,
-      "grad_norm": 0.003286434104666114,
-      "learning_rate": 0.00019999595483850092,
-      "loss": 46.0,
-      "step": 37457
-    },
-    {
-      "epoch": 2.863925683812145,
-      "grad_norm": 0.0026142208371311426,
-      "learning_rate": 0.00019999595462245223,
-      "loss": 46.0,
-      "step": 37458
-    },
-    {
-      "epoch": 2.864002140795535,
-      "grad_norm": 0.0016015034634619951,
-      "learning_rate": 0.0001999959544063977,
-      "loss": 46.0,
-      "step": 37459
-    },
-    {
-      "epoch": 2.8640785977789247,
-      "grad_norm": 0.004334256052970886,
-      "learning_rate": 0.0001999959541903375,
-      "loss": 46.0,
-      "step": 37460
-    },
-    {
-      "epoch": 2.8641550547623145,
-      "grad_norm": 0.0013641779078170657,
-      "learning_rate": 0.00019999595397427142,
-      "loss": 46.0,
-      "step": 37461
-    },
-    {
-      "epoch": 2.8642315117457042,
-      "grad_norm": 0.0009866136824712157,
-      "learning_rate": 0.00019999595375819963,
-      "loss": 46.0,
-      "step": 37462
-    },
-    {
-      "epoch": 2.8643079687290935,
-      "grad_norm": 0.0010501127690076828,
-      "learning_rate": 0.0001999959535421221,
-      "loss": 46.0,
-      "step": 37463
-    },
-    {
-      "epoch": 2.8643844257124833,
-      "grad_norm": 0.0014722800115123391,
-      "learning_rate": 0.00019999595332603873,
-      "loss": 46.0,
-      "step": 37464
-    },
-    {
-      "epoch": 2.864460882695873,
-      "grad_norm": 0.0012393639190122485,
-      "learning_rate": 0.00019999595310994962,
-      "loss": 46.0,
-      "step": 37465
-    },
-    {
-      "epoch": 2.864537339679263,
-      "grad_norm": 0.0007497768383473158,
-      "learning_rate": 0.00019999595289385476,
-      "loss": 46.0,
-      "step": 37466
-    },
-    {
-      "epoch": 2.8646137966626526,
-      "grad_norm": 0.011151333339512348,
-      "learning_rate": 0.0001999959526777541,
-      "loss": 46.0,
-      "step": 37467
-    },
-    {
-      "epoch": 2.8646902536460424,
-      "grad_norm": 0.0011931046610698104,
-      "learning_rate": 0.0001999959524616477,
-      "loss": 46.0,
-      "step": 37468
-    },
-    {
-      "epoch": 2.864766710629432,
-      "grad_norm": 0.0012726098066195846,
-      "learning_rate": 0.0001999959522455355,
-      "loss": 46.0,
-      "step": 37469
-    },
-    {
-      "epoch": 2.864843167612822,
-      "grad_norm": 0.0025933312717825174,
-      "learning_rate": 0.00019999595202941755,
-      "loss": 46.0,
-      "step": 37470
-    },
-    {
-      "epoch": 2.8649196245962116,
-      "grad_norm": 0.00538911297917366,
-      "learning_rate": 0.00019999595181329386,
-      "loss": 46.0,
-      "step": 37471
-    },
-    {
-      "epoch": 2.8649960815796014,
-      "grad_norm": 0.0022859370801597834,
-      "learning_rate": 0.00019999595159716433,
-      "loss": 46.0,
-      "step": 37472
-    },
-    {
-      "epoch": 2.865072538562991,
-      "grad_norm": 0.0013085856335237622,
-      "learning_rate": 0.0001999959513810291,
-      "loss": 46.0,
-      "step": 37473
-    },
-    {
-      "epoch": 2.865148995546381,
-      "grad_norm": 0.005054724868386984,
-      "learning_rate": 0.00019999595116488805,
-      "loss": 46.0,
-      "step": 37474
-    },
-    {
-      "epoch": 2.8652254525297707,
-      "grad_norm": 0.006324580404907465,
-      "learning_rate": 0.00019999595094874126,
-      "loss": 46.0,
-      "step": 37475
-    },
-    {
-      "epoch": 2.86530190951316,
-      "grad_norm": 0.0008189025102183223,
-      "learning_rate": 0.00019999595073258867,
-      "loss": 46.0,
-      "step": 37476
-    },
-    {
-      "epoch": 2.8653783664965498,
-      "grad_norm": 0.0025469448883086443,
-      "learning_rate": 0.00019999595051643033,
-      "loss": 46.0,
-      "step": 37477
-    },
-    {
-      "epoch": 2.8654548234799395,
-      "grad_norm": 0.0017832368612289429,
-      "learning_rate": 0.00019999595030026622,
-      "loss": 46.0,
-      "step": 37478
-    },
-    {
-      "epoch": 2.8655312804633293,
-      "grad_norm": 0.002492888830602169,
-      "learning_rate": 0.00019999595008409637,
-      "loss": 46.0,
-      "step": 37479
-    },
-    {
-      "epoch": 2.865607737446719,
-      "grad_norm": 0.0023167075123637915,
-      "learning_rate": 0.00019999594986792074,
-      "loss": 46.0,
-      "step": 37480
-    },
-    {
-      "epoch": 2.865684194430109,
-      "grad_norm": 0.0023142294958233833,
-      "learning_rate": 0.0001999959496517393,
-      "loss": 46.0,
-      "step": 37481
-    },
-    {
-      "epoch": 2.8657606514134986,
-      "grad_norm": 0.0026507326401770115,
-      "learning_rate": 0.00019999594943555214,
-      "loss": 46.0,
-      "step": 37482
-    },
-    {
-      "epoch": 2.8658371083968883,
-      "grad_norm": 0.0011573369847610593,
-      "learning_rate": 0.0001999959492193592,
-      "loss": 46.0,
-      "step": 37483
-    },
-    {
-      "epoch": 2.865913565380278,
-      "grad_norm": 0.0017050030874088407,
-      "learning_rate": 0.00019999594900316046,
-      "loss": 46.0,
-      "step": 37484
-    },
-    {
-      "epoch": 2.8659900223636674,
-      "grad_norm": 0.0019060500198975205,
-      "learning_rate": 0.00019999594878695597,
-      "loss": 46.0,
-      "step": 37485
-    },
-    {
-      "epoch": 2.866066479347057,
-      "grad_norm": 0.0007430505938827991,
-      "learning_rate": 0.0001999959485707457,
-      "loss": 46.0,
-      "step": 37486
-    },
-    {
-      "epoch": 2.866142936330447,
-      "grad_norm": 0.0027034657541662455,
-      "learning_rate": 0.00019999594835452968,
-      "loss": 46.0,
-      "step": 37487
-    },
-    {
-      "epoch": 2.8662193933138367,
-      "grad_norm": 0.002144405385479331,
-      "learning_rate": 0.0001999959481383079,
-      "loss": 46.0,
-      "step": 37488
-    },
-    {
-      "epoch": 2.8662958502972264,
-      "grad_norm": 0.0008919090614654124,
-      "learning_rate": 0.0001999959479220803,
-      "loss": 46.0,
-      "step": 37489
-    },
-    {
-      "epoch": 2.866372307280616,
-      "grad_norm": 0.0020369570702314377,
-      "learning_rate": 0.00019999594770584697,
-      "loss": 46.0,
-      "step": 37490
-    },
-    {
-      "epoch": 2.866448764264006,
-      "grad_norm": 0.0007791294483467937,
-      "learning_rate": 0.00019999594748960786,
-      "loss": 46.0,
-      "step": 37491
-    },
-    {
-      "epoch": 2.8665252212473957,
-      "grad_norm": 0.004863109905272722,
-      "learning_rate": 0.000199995947273363,
-      "loss": 46.0,
-      "step": 37492
-    },
-    {
-      "epoch": 2.8666016782307855,
-      "grad_norm": 0.0008897147490642965,
-      "learning_rate": 0.00019999594705711235,
-      "loss": 46.0,
-      "step": 37493
-    },
-    {
-      "epoch": 2.8666781352141752,
-      "grad_norm": 0.001245802384801209,
-      "learning_rate": 0.00019999594684085595,
-      "loss": 46.0,
-      "step": 37494
-    },
-    {
-      "epoch": 2.866754592197565,
-      "grad_norm": 0.004159674979746342,
-      "learning_rate": 0.00019999594662459375,
-      "loss": 46.0,
-      "step": 37495
-    },
-    {
-      "epoch": 2.8668310491809548,
-      "grad_norm": 0.002161232987418771,
-      "learning_rate": 0.00019999594640832583,
-      "loss": 46.0,
-      "step": 37496
-    },
-    {
-      "epoch": 2.8669075061643445,
-      "grad_norm": 0.0019452959531918168,
-      "learning_rate": 0.00019999594619205208,
-      "loss": 46.0,
-      "step": 37497
-    },
-    {
-      "epoch": 2.866983963147734,
-      "grad_norm": 0.0022323147859424353,
-      "learning_rate": 0.00019999594597577258,
-      "loss": 46.0,
-      "step": 37498
-    },
-    {
-      "epoch": 2.8670604201311236,
-      "grad_norm": 0.0018292245222255588,
-      "learning_rate": 0.0001999959457594873,
-      "loss": 46.0,
-      "step": 37499
-    },
-    {
-      "epoch": 2.8671368771145134,
-      "grad_norm": 0.001836481736972928,
-      "learning_rate": 0.0001999959455431963,
-      "loss": 46.0,
-      "step": 37500
-    },
-    {
-      "epoch": 2.867213334097903,
-      "grad_norm": 0.0013462741626426578,
-      "learning_rate": 0.00019999594532689948,
-      "loss": 46.0,
-      "step": 37501
-    },
-    {
-      "epoch": 2.867289791081293,
-      "grad_norm": 0.0027386879082769156,
-      "learning_rate": 0.00019999594511059695,
-      "loss": 46.0,
-      "step": 37502
-    },
-    {
-      "epoch": 2.8673662480646827,
-      "grad_norm": 0.009533974342048168,
-      "learning_rate": 0.0001999959448942886,
-      "loss": 46.0,
-      "step": 37503
-    },
-    {
-      "epoch": 2.8674427050480724,
-      "grad_norm": 0.0008170533110387623,
-      "learning_rate": 0.0001999959446779745,
-      "loss": 46.0,
-      "step": 37504
-    },
-    {
-      "epoch": 2.867519162031462,
-      "grad_norm": 0.002299931598827243,
-      "learning_rate": 0.00019999594446165465,
-      "loss": 46.0,
-      "step": 37505
-    },
-    {
-      "epoch": 2.8675956190148515,
-      "grad_norm": 0.00147080491296947,
-      "learning_rate": 0.00019999594424532897,
-      "loss": 46.0,
-      "step": 37506
-    },
-    {
-      "epoch": 2.8676720759982413,
-      "grad_norm": 0.004154044669121504,
-      "learning_rate": 0.00019999594402899756,
-      "loss": 46.0,
-      "step": 37507
-    },
-    {
-      "epoch": 2.867748532981631,
-      "grad_norm": 0.0017092034686356783,
-      "learning_rate": 0.0001999959438126604,
-      "loss": 46.0,
-      "step": 37508
-    },
-    {
-      "epoch": 2.8678249899650208,
-      "grad_norm": 0.0021878632251173258,
-      "learning_rate": 0.00019999594359631744,
-      "loss": 46.0,
-      "step": 37509
-    },
-    {
-      "epoch": 2.8679014469484105,
-      "grad_norm": 0.005851882975548506,
-      "learning_rate": 0.0001999959433799687,
-      "loss": 46.0,
-      "step": 37510
-    },
-    {
-      "epoch": 2.8679779039318003,
-      "grad_norm": 0.004577010404318571,
-      "learning_rate": 0.0001999959431636142,
-      "loss": 46.0,
-      "step": 37511
-    },
-    {
-      "epoch": 2.86805436091519,
-      "grad_norm": 0.003406402887776494,
-      "learning_rate": 0.00019999594294725396,
-      "loss": 46.0,
-      "step": 37512
-    },
-    {
-      "epoch": 2.86813081789858,
-      "grad_norm": 0.0036590341478586197,
-      "learning_rate": 0.00019999594273088794,
-      "loss": 46.0,
-      "step": 37513
-    },
-    {
-      "epoch": 2.8682072748819696,
-      "grad_norm": 0.0017514858627691865,
-      "learning_rate": 0.00019999594251451615,
-      "loss": 46.0,
-      "step": 37514
-    },
-    {
-      "epoch": 2.8682837318653593,
-      "grad_norm": 0.002538986038416624,
-      "learning_rate": 0.00019999594229813856,
-      "loss": 46.0,
-      "step": 37515
-    },
-    {
-      "epoch": 2.868360188848749,
-      "grad_norm": 0.0051421415992081165,
-      "learning_rate": 0.00019999594208175523,
-      "loss": 46.0,
-      "step": 37516
-    },
-    {
-      "epoch": 2.868436645832139,
-      "grad_norm": 0.0027391118928790092,
-      "learning_rate": 0.00019999594186536612,
-      "loss": 46.0,
-      "step": 37517
-    },
-    {
-      "epoch": 2.8685131028155286,
-      "grad_norm": 0.001825224026106298,
-      "learning_rate": 0.00019999594164897127,
-      "loss": 46.0,
-      "step": 37518
-    },
-    {
-      "epoch": 2.8685895597989184,
-      "grad_norm": 0.0023483980912715197,
-      "learning_rate": 0.0001999959414325706,
-      "loss": 46.0,
-      "step": 37519
-    },
-    {
-      "epoch": 2.8686660167823077,
-      "grad_norm": 0.002353568794205785,
-      "learning_rate": 0.00019999594121616418,
-      "loss": 46.0,
-      "step": 37520
-    },
-    {
-      "epoch": 2.8687424737656975,
-      "grad_norm": 0.0016929901903495193,
-      "learning_rate": 0.000199995940999752,
-      "loss": 46.0,
-      "step": 37521
-    },
-    {
-      "epoch": 2.8688189307490872,
-      "grad_norm": 0.0012015769025310874,
-      "learning_rate": 0.00019999594078333406,
-      "loss": 46.0,
-      "step": 37522
-    },
-    {
-      "epoch": 2.868895387732477,
-      "grad_norm": 0.0029947562143206596,
-      "learning_rate": 0.00019999594056691034,
-      "loss": 46.0,
-      "step": 37523
-    },
-    {
-      "epoch": 2.8689718447158667,
-      "grad_norm": 0.006256711203604937,
-      "learning_rate": 0.00019999594035048084,
-      "loss": 46.0,
-      "step": 37524
-    },
-    {
-      "epoch": 2.8690483016992565,
-      "grad_norm": 0.0008310055709443986,
-      "learning_rate": 0.00019999594013404558,
-      "loss": 46.0,
-      "step": 37525
-    },
-    {
-      "epoch": 2.8691247586826463,
-      "grad_norm": 0.004027517046779394,
-      "learning_rate": 0.00019999593991760456,
-      "loss": 46.0,
-      "step": 37526
-    },
-    {
-      "epoch": 2.869201215666036,
-      "grad_norm": 0.0024037384428083897,
-      "learning_rate": 0.00019999593970115775,
-      "loss": 46.0,
-      "step": 37527
-    },
-    {
-      "epoch": 2.8692776726494253,
-      "grad_norm": 0.0016094769816845655,
-      "learning_rate": 0.0001999959394847052,
-      "loss": 46.0,
-      "step": 37528
-    },
-    {
-      "epoch": 2.869354129632815,
-      "grad_norm": 0.0014064470306038857,
-      "learning_rate": 0.00019999593926824685,
-      "loss": 46.0,
-      "step": 37529
-    },
-    {
-      "epoch": 2.869430586616205,
-      "grad_norm": 0.0019122699741274118,
-      "learning_rate": 0.00019999593905178274,
-      "loss": 46.0,
-      "step": 37530
-    },
-    {
-      "epoch": 2.8695070435995946,
-      "grad_norm": 0.0006940191378816962,
-      "learning_rate": 0.0001999959388353129,
-      "loss": 46.0,
-      "step": 37531
-    },
-    {
-      "epoch": 2.8695835005829844,
-      "grad_norm": 0.002844130387529731,
-      "learning_rate": 0.00019999593861883726,
-      "loss": 46.0,
-      "step": 37532
-    },
-    {
-      "epoch": 2.869659957566374,
-      "grad_norm": 0.0016199576202780008,
-      "learning_rate": 0.00019999593840235583,
-      "loss": 46.0,
-      "step": 37533
-    },
-    {
-      "epoch": 2.869736414549764,
-      "grad_norm": 0.003924049437046051,
-      "learning_rate": 0.00019999593818586866,
-      "loss": 46.0,
-      "step": 37534
-    },
-    {
-      "epoch": 2.8698128715331537,
-      "grad_norm": 0.0013053047005087137,
-      "learning_rate": 0.0001999959379693757,
-      "loss": 46.0,
-      "step": 37535
-    },
-    {
-      "epoch": 2.8698893285165434,
-      "grad_norm": 0.0024365128483623266,
-      "learning_rate": 0.00019999593775287696,
-      "loss": 46.0,
-      "step": 37536
-    },
-    {
-      "epoch": 2.869965785499933,
-      "grad_norm": 0.002858174964785576,
-      "learning_rate": 0.00019999593753637247,
-      "loss": 46.0,
-      "step": 37537
-    },
-    {
-      "epoch": 2.870042242483323,
-      "grad_norm": 0.007344653829932213,
-      "learning_rate": 0.00019999593731986223,
-      "loss": 46.0,
-      "step": 37538
-    },
-    {
-      "epoch": 2.8701186994667127,
-      "grad_norm": 0.0017986251041293144,
-      "learning_rate": 0.0001999959371033462,
-      "loss": 46.0,
-      "step": 37539
-    },
-    {
-      "epoch": 2.8701951564501025,
-      "grad_norm": 0.0016602270770817995,
-      "learning_rate": 0.0001999959368868244,
-      "loss": 46.0,
-      "step": 37540
-    },
-    {
-      "epoch": 2.8702716134334922,
-      "grad_norm": 0.00330542610026896,
-      "learning_rate": 0.0001999959366702968,
-      "loss": 46.0,
-      "step": 37541
-    },
-    {
-      "epoch": 2.8703480704168816,
-      "grad_norm": 0.001754896016791463,
-      "learning_rate": 0.00019999593645376348,
-      "loss": 46.0,
-      "step": 37542
-    },
-    {
-      "epoch": 2.8704245274002713,
-      "grad_norm": 0.001867889310233295,
-      "learning_rate": 0.0001999959362372244,
-      "loss": 46.0,
-      "step": 37543
-    },
-    {
-      "epoch": 2.870500984383661,
-      "grad_norm": 0.003189666895195842,
-      "learning_rate": 0.00019999593602067952,
-      "loss": 46.0,
-      "step": 37544
-    },
-    {
-      "epoch": 2.870577441367051,
-      "grad_norm": 0.0004502327064983547,
-      "learning_rate": 0.0001999959358041289,
-      "loss": 46.0,
-      "step": 37545
-    },
-    {
-      "epoch": 2.8706538983504406,
-      "grad_norm": 0.0014798900811001658,
-      "learning_rate": 0.00019999593558757246,
-      "loss": 46.0,
-      "step": 37546
-    },
-    {
-      "epoch": 2.8707303553338304,
-      "grad_norm": 0.0059404936619102955,
-      "learning_rate": 0.00019999593537101026,
-      "loss": 46.0,
-      "step": 37547
-    },
-    {
-      "epoch": 2.87080681231722,
-      "grad_norm": 0.0009865788742899895,
-      "learning_rate": 0.00019999593515444234,
-      "loss": 46.0,
-      "step": 37548
-    },
-    {
-      "epoch": 2.87088326930061,
-      "grad_norm": 0.0010516804177314043,
-      "learning_rate": 0.00019999593493786862,
-      "loss": 46.0,
-      "step": 37549
-    },
-    {
-      "epoch": 2.870959726283999,
-      "grad_norm": 0.001716958126053214,
-      "learning_rate": 0.0001999959347212891,
-      "loss": 46.0,
-      "step": 37550
-    },
-    {
-      "epoch": 2.871036183267389,
-      "grad_norm": 0.0035657398402690887,
-      "learning_rate": 0.00019999593450470386,
-      "loss": 46.0,
-      "step": 37551
-    },
-    {
-      "epoch": 2.8711126402507787,
-      "grad_norm": 0.0019648680463433266,
-      "learning_rate": 0.00019999593428811284,
-      "loss": 46.0,
-      "step": 37552
-    },
-    {
-      "epoch": 2.8711890972341685,
-      "grad_norm": 0.0027284380048513412,
-      "learning_rate": 0.00019999593407151603,
-      "loss": 46.0,
-      "step": 37553
-    },
-    {
-      "epoch": 2.8712655542175582,
-      "grad_norm": 0.0037620195653289557,
-      "learning_rate": 0.00019999593385491347,
-      "loss": 46.0,
-      "step": 37554
-    },
-    {
-      "epoch": 2.871342011200948,
-      "grad_norm": 0.0029824539087712765,
-      "learning_rate": 0.00019999593363830513,
-      "loss": 46.0,
-      "step": 37555
-    },
-    {
-      "epoch": 2.8714184681843378,
-      "grad_norm": 0.0011551117058843374,
-      "learning_rate": 0.00019999593342169103,
-      "loss": 46.0,
-      "step": 37556
-    },
-    {
-      "epoch": 2.8714949251677275,
-      "grad_norm": 0.0032276457641273737,
-      "learning_rate": 0.00019999593320507115,
-      "loss": 46.0,
-      "step": 37557
-    },
-    {
-      "epoch": 2.8715713821511173,
-      "grad_norm": 0.0028557691257447004,
-      "learning_rate": 0.0001999959329884455,
-      "loss": 46.0,
-      "step": 37558
-    },
-    {
-      "epoch": 2.871647839134507,
-      "grad_norm": 0.0033261561766266823,
-      "learning_rate": 0.00019999593277181406,
-      "loss": 46.0,
-      "step": 37559
-    },
-    {
-      "epoch": 2.871724296117897,
-      "grad_norm": 0.0022382389288395643,
-      "learning_rate": 0.00019999593255517692,
-      "loss": 46.0,
-      "step": 37560
-    },
-    {
-      "epoch": 2.8718007531012866,
-      "grad_norm": 0.004119453020393848,
-      "learning_rate": 0.00019999593233853394,
-      "loss": 46.0,
-      "step": 37561
-    },
-    {
-      "epoch": 2.8718772100846763,
-      "grad_norm": 0.0006247319979593158,
-      "learning_rate": 0.00019999593212188522,
-      "loss": 46.0,
-      "step": 37562
-    },
-    {
-      "epoch": 2.871953667068066,
-      "grad_norm": 0.0019964047241955996,
-      "learning_rate": 0.00019999593190523073,
-      "loss": 46.0,
-      "step": 37563
-    },
-    {
-      "epoch": 2.8720301240514554,
-      "grad_norm": 0.0011412703897804022,
-      "learning_rate": 0.0001999959316885705,
-      "loss": 46.0,
-      "step": 37564
-    },
-    {
-      "epoch": 2.872106581034845,
-      "grad_norm": 0.0009878354612737894,
-      "learning_rate": 0.00019999593147190448,
-      "loss": 46.0,
-      "step": 37565
-    },
-    {
-      "epoch": 2.872183038018235,
-      "grad_norm": 0.002664342289790511,
-      "learning_rate": 0.00019999593125523266,
-      "loss": 46.0,
-      "step": 37566
-    },
-    {
-      "epoch": 2.8722594950016247,
-      "grad_norm": 0.003253551432862878,
-      "learning_rate": 0.0001999959310385551,
-      "loss": 46.0,
-      "step": 37567
-    },
-    {
-      "epoch": 2.8723359519850145,
-      "grad_norm": 0.0011970893247053027,
-      "learning_rate": 0.00019999593082187177,
-      "loss": 46.0,
-      "step": 37568
-    },
-    {
-      "epoch": 2.872412408968404,
-      "grad_norm": 0.0017122890567407012,
-      "learning_rate": 0.00019999593060518266,
-      "loss": 46.0,
-      "step": 37569
-    },
-    {
-      "epoch": 2.872488865951794,
-      "grad_norm": 0.0026473926845937967,
-      "learning_rate": 0.00019999593038848778,
-      "loss": 46.0,
-      "step": 37570
-    },
-    {
-      "epoch": 2.8725653229351837,
-      "grad_norm": 0.0010518949711695313,
-      "learning_rate": 0.00019999593017178713,
-      "loss": 46.0,
-      "step": 37571
-    },
-    {
-      "epoch": 2.872641779918573,
-      "grad_norm": 0.00414118031039834,
-      "learning_rate": 0.00019999592995508073,
-      "loss": 46.0,
-      "step": 37572
-    },
-    {
-      "epoch": 2.872718236901963,
-      "grad_norm": 0.00721208518370986,
-      "learning_rate": 0.00019999592973836853,
-      "loss": 46.0,
-      "step": 37573
-    },
-    {
-      "epoch": 2.8727946938853526,
-      "grad_norm": 0.002168054925277829,
-      "learning_rate": 0.0001999959295216506,
-      "loss": 46.0,
-      "step": 37574
-    },
-    {
-      "epoch": 2.8728711508687423,
-      "grad_norm": 0.0034994606394320726,
-      "learning_rate": 0.0001999959293049269,
-      "loss": 46.0,
-      "step": 37575
-    },
-    {
-      "epoch": 2.872947607852132,
-      "grad_norm": 0.0022491789422929287,
-      "learning_rate": 0.00019999592908819737,
-      "loss": 46.0,
-      "step": 37576
-    },
-    {
-      "epoch": 2.873024064835522,
-      "grad_norm": 0.0009970130631700158,
-      "learning_rate": 0.00019999592887146213,
-      "loss": 46.0,
-      "step": 37577
-    },
-    {
-      "epoch": 2.8731005218189116,
-      "grad_norm": 0.0024155722931027412,
-      "learning_rate": 0.0001999959286547211,
-      "loss": 46.0,
-      "step": 37578
-    },
-    {
-      "epoch": 2.8731769788023014,
-      "grad_norm": 0.005563588347285986,
-      "learning_rate": 0.0001999959284379743,
-      "loss": 46.0,
-      "step": 37579
-    },
-    {
-      "epoch": 2.873253435785691,
-      "grad_norm": 0.002309634815901518,
-      "learning_rate": 0.00019999592822122174,
-      "loss": 46.0,
-      "step": 37580
-    },
-    {
-      "epoch": 2.873329892769081,
-      "grad_norm": 0.001668619690462947,
-      "learning_rate": 0.00019999592800446338,
-      "loss": 46.0,
-      "step": 37581
-    },
-    {
-      "epoch": 2.8734063497524707,
-      "grad_norm": 0.002494386164471507,
-      "learning_rate": 0.0001999959277876993,
-      "loss": 46.0,
-      "step": 37582
-    },
-    {
-      "epoch": 2.8734828067358604,
-      "grad_norm": 0.005363532807677984,
-      "learning_rate": 0.00019999592757092942,
-      "loss": 46.0,
-      "step": 37583
-    },
-    {
-      "epoch": 2.87355926371925,
-      "grad_norm": 0.0023722245823591948,
-      "learning_rate": 0.00019999592735415377,
-      "loss": 46.0,
-      "step": 37584
-    },
-    {
-      "epoch": 2.87363572070264,
-      "grad_norm": 0.002606066409498453,
-      "learning_rate": 0.00019999592713737237,
-      "loss": 46.0,
-      "step": 37585
-    },
-    {
-      "epoch": 2.8737121776860293,
-      "grad_norm": 0.0029873568564653397,
-      "learning_rate": 0.0001999959269205852,
-      "loss": 46.0,
-      "step": 37586
-    },
-    {
-      "epoch": 2.873788634669419,
-      "grad_norm": 0.0033291769213974476,
-      "learning_rate": 0.00019999592670379225,
-      "loss": 46.0,
-      "step": 37587
-    },
-    {
-      "epoch": 2.873865091652809,
-      "grad_norm": 0.004090075381100178,
-      "learning_rate": 0.0001999959264869935,
-      "loss": 46.0,
-      "step": 37588
-    },
-    {
-      "epoch": 2.8739415486361986,
-      "grad_norm": 0.0015167307574301958,
-      "learning_rate": 0.000199995926270189,
-      "loss": 46.0,
-      "step": 37589
-    },
-    {
-      "epoch": 2.8740180056195883,
-      "grad_norm": 0.001860748860053718,
-      "learning_rate": 0.00019999592605337877,
-      "loss": 46.0,
-      "step": 37590
-    },
-    {
-      "epoch": 2.874094462602978,
-      "grad_norm": 0.005001469980925322,
-      "learning_rate": 0.00019999592583656273,
-      "loss": 46.0,
-      "step": 37591
-    },
-    {
-      "epoch": 2.874170919586368,
-      "grad_norm": 0.003039294620975852,
-      "learning_rate": 0.00019999592561974094,
-      "loss": 46.0,
-      "step": 37592
-    },
-    {
-      "epoch": 2.8742473765697576,
-      "grad_norm": 0.005798115395009518,
-      "learning_rate": 0.00019999592540291338,
-      "loss": 46.0,
-      "step": 37593
-    },
-    {
-      "epoch": 2.874323833553147,
-      "grad_norm": 0.008996091783046722,
-      "learning_rate": 0.00019999592518608002,
-      "loss": 46.0,
-      "step": 37594
-    },
-    {
-      "epoch": 2.8744002905365367,
-      "grad_norm": 0.0012368884636089206,
-      "learning_rate": 0.00019999592496924094,
-      "loss": 46.0,
-      "step": 37595
-    },
-    {
-      "epoch": 2.8744767475199264,
-      "grad_norm": 0.0022297895047813654,
-      "learning_rate": 0.00019999592475239606,
-      "loss": 46.0,
-      "step": 37596
-    },
-    {
-      "epoch": 2.874553204503316,
-      "grad_norm": 0.0024840838741511106,
-      "learning_rate": 0.00019999592453554544,
-      "loss": 46.0,
-      "step": 37597
-    },
-    {
-      "epoch": 2.874629661486706,
-      "grad_norm": 0.004004363436251879,
-      "learning_rate": 0.00019999592431868898,
-      "loss": 46.0,
-      "step": 37598
-    },
-    {
-      "epoch": 2.8747061184700957,
-      "grad_norm": 0.0026861298829317093,
-      "learning_rate": 0.0001999959241018268,
-      "loss": 46.0,
-      "step": 37599
-    },
-    {
-      "epoch": 2.8747825754534855,
-      "grad_norm": 0.0005568024353124201,
-      "learning_rate": 0.00019999592388495886,
-      "loss": 46.0,
-      "step": 37600
-    },
-    {
-      "epoch": 2.8748590324368752,
-      "grad_norm": 0.0034252905752509832,
-      "learning_rate": 0.00019999592366808515,
-      "loss": 46.0,
-      "step": 37601
-    },
-    {
-      "epoch": 2.874935489420265,
-      "grad_norm": 0.001752939191646874,
-      "learning_rate": 0.00019999592345120565,
-      "loss": 46.0,
-      "step": 37602
-    },
-    {
-      "epoch": 2.8750119464036548,
-      "grad_norm": 0.003235015319660306,
-      "learning_rate": 0.00019999592323432039,
-      "loss": 46.0,
-      "step": 37603
-    },
-    {
-      "epoch": 2.8750884033870445,
-      "grad_norm": 0.005604472476989031,
-      "learning_rate": 0.00019999592301742937,
-      "loss": 46.0,
-      "step": 37604
-    },
-    {
-      "epoch": 2.8751648603704343,
-      "grad_norm": 0.0051450408063828945,
-      "learning_rate": 0.0001999959228005326,
-      "loss": 46.0,
-      "step": 37605
-    },
-    {
-      "epoch": 2.875241317353824,
-      "grad_norm": 0.003964126110076904,
-      "learning_rate": 0.00019999592258363,
-      "loss": 46.0,
-      "step": 37606
-    },
-    {
-      "epoch": 2.8753177743372134,
-      "grad_norm": 0.0005387893179431558,
-      "learning_rate": 0.0001999959223667217,
-      "loss": 46.0,
-      "step": 37607
-    },
-    {
-      "epoch": 2.875394231320603,
-      "grad_norm": 0.0009338142699562013,
-      "learning_rate": 0.00019999592214980756,
-      "loss": 46.0,
-      "step": 37608
-    },
-    {
-      "epoch": 2.875470688303993,
-      "grad_norm": 0.001832042122259736,
-      "learning_rate": 0.0001999959219328877,
-      "loss": 46.0,
-      "step": 37609
-    },
-    {
-      "epoch": 2.8755471452873826,
-      "grad_norm": 0.0022954875603318214,
-      "learning_rate": 0.00019999592171596206,
-      "loss": 46.0,
-      "step": 37610
-    },
-    {
-      "epoch": 2.8756236022707724,
-      "grad_norm": 0.0012057002168148756,
-      "learning_rate": 0.00019999592149903066,
-      "loss": 46.0,
-      "step": 37611
-    },
-    {
-      "epoch": 2.875700059254162,
-      "grad_norm": 0.006148792337626219,
-      "learning_rate": 0.00019999592128209349,
-      "loss": 46.0,
-      "step": 37612
-    },
-    {
-      "epoch": 2.875776516237552,
-      "grad_norm": 0.00475850747898221,
-      "learning_rate": 0.00019999592106515054,
-      "loss": 46.0,
-      "step": 37613
-    },
-    {
-      "epoch": 2.8758529732209417,
-      "grad_norm": 0.00256563862785697,
-      "learning_rate": 0.0001999959208482018,
-      "loss": 46.0,
-      "step": 37614
-    },
-    {
-      "epoch": 2.8759294302043314,
-      "grad_norm": 0.004690373782068491,
-      "learning_rate": 0.0001999959206312473,
-      "loss": 46.0,
-      "step": 37615
-    },
-    {
-      "epoch": 2.8760058871877208,
-      "grad_norm": 0.001298415125347674,
-      "learning_rate": 0.00019999592041428704,
-      "loss": 46.0,
-      "step": 37616
-    },
-    {
-      "epoch": 2.8760823441711105,
-      "grad_norm": 0.0023295036517083645,
-      "learning_rate": 0.00019999592019732102,
-      "loss": 46.0,
-      "step": 37617
-    },
-    {
-      "epoch": 2.8761588011545003,
-      "grad_norm": 0.0022516236640512943,
-      "learning_rate": 0.0001999959199803492,
-      "loss": 46.0,
-      "step": 37618
-    },
-    {
-      "epoch": 2.87623525813789,
-      "grad_norm": 0.0027028268668800592,
-      "learning_rate": 0.00019999591976337165,
-      "loss": 46.0,
-      "step": 37619
-    },
-    {
-      "epoch": 2.87631171512128,
-      "grad_norm": 0.003751756390556693,
-      "learning_rate": 0.00019999591954638832,
-      "loss": 46.0,
-      "step": 37620
-    },
-    {
-      "epoch": 2.8763881721046696,
-      "grad_norm": 0.0011190070072188973,
-      "learning_rate": 0.0001999959193293992,
-      "loss": 46.0,
-      "step": 37621
-    },
-    {
-      "epoch": 2.8764646290880593,
-      "grad_norm": 0.00637247646227479,
-      "learning_rate": 0.00019999591911240433,
-      "loss": 46.0,
-      "step": 37622
-    },
-    {
-      "epoch": 2.876541086071449,
-      "grad_norm": 0.0014003842370584607,
-      "learning_rate": 0.0001999959188954037,
-      "loss": 46.0,
-      "step": 37623
-    },
-    {
-      "epoch": 2.876617543054839,
-      "grad_norm": 0.0018502847524359822,
-      "learning_rate": 0.00019999591867839728,
-      "loss": 46.0,
-      "step": 37624
-    },
-    {
-      "epoch": 2.8766940000382286,
-      "grad_norm": 0.0023573304060846567,
-      "learning_rate": 0.0001999959184613851,
-      "loss": 46.0,
-      "step": 37625
-    },
-    {
-      "epoch": 2.8767704570216184,
-      "grad_norm": 0.001384947681799531,
-      "learning_rate": 0.00019999591824436714,
-      "loss": 46.0,
-      "step": 37626
-    },
-    {
-      "epoch": 2.876846914005008,
-      "grad_norm": 0.0009258214849978685,
-      "learning_rate": 0.00019999591802734342,
-      "loss": 46.0,
-      "step": 37627
-    },
-    {
-      "epoch": 2.876923370988398,
-      "grad_norm": 0.00131578603759408,
-      "learning_rate": 0.00019999591781031395,
-      "loss": 46.0,
-      "step": 37628
-    },
-    {
-      "epoch": 2.876999827971787,
-      "grad_norm": 0.0014559427509084344,
-      "learning_rate": 0.0001999959175932787,
-      "loss": 46.0,
-      "step": 37629
-    },
-    {
-      "epoch": 2.877076284955177,
-      "grad_norm": 0.0011839630315080285,
-      "learning_rate": 0.00019999591737623765,
-      "loss": 46.0,
-      "step": 37630
-    },
-    {
-      "epoch": 2.8771527419385667,
-      "grad_norm": 0.009083331562578678,
-      "learning_rate": 0.0001999959171591909,
-      "loss": 46.0,
-      "step": 37631
-    },
-    {
-      "epoch": 2.8772291989219565,
-      "grad_norm": 0.0016284589655697346,
-      "learning_rate": 0.0001999959169421383,
-      "loss": 46.0,
-      "step": 37632
-    },
-    {
-      "epoch": 2.8773056559053463,
-      "grad_norm": 0.0025250306352972984,
-      "learning_rate": 0.00019999591672507995,
-      "loss": 46.0,
-      "step": 37633
-    },
-    {
-      "epoch": 2.877382112888736,
-      "grad_norm": 0.0042189378291368484,
-      "learning_rate": 0.00019999591650801587,
-      "loss": 46.0,
-      "step": 37634
-    },
-    {
-      "epoch": 2.877458569872126,
-      "grad_norm": 0.006225135177373886,
-      "learning_rate": 0.000199995916290946,
-      "loss": 46.0,
-      "step": 37635
-    },
-    {
-      "epoch": 2.8775350268555155,
-      "grad_norm": 0.0019970820285379887,
-      "learning_rate": 0.00019999591607387034,
-      "loss": 46.0,
-      "step": 37636
-    },
-    {
-      "epoch": 2.877611483838905,
-      "grad_norm": 0.0010138554498553276,
-      "learning_rate": 0.0001999959158567889,
-      "loss": 46.0,
-      "step": 37637
-    },
-    {
-      "epoch": 2.8776879408222946,
-      "grad_norm": 0.001222631661221385,
-      "learning_rate": 0.00019999591563970174,
-      "loss": 46.0,
-      "step": 37638
-    },
-    {
-      "epoch": 2.8777643978056844,
-      "grad_norm": 0.006695645861327648,
-      "learning_rate": 0.0001999959154226088,
-      "loss": 46.0,
-      "step": 37639
-    },
-    {
-      "epoch": 2.877840854789074,
-      "grad_norm": 0.002268946496769786,
-      "learning_rate": 0.00019999591520551008,
-      "loss": 46.0,
-      "step": 37640
-    },
-    {
-      "epoch": 2.877917311772464,
-      "grad_norm": 0.003085621166974306,
-      "learning_rate": 0.00019999591498840558,
-      "loss": 46.0,
-      "step": 37641
-    },
-    {
-      "epoch": 2.8779937687558537,
-      "grad_norm": 0.0026596535462886095,
-      "learning_rate": 0.00019999591477129535,
-      "loss": 46.0,
-      "step": 37642
-    },
-    {
-      "epoch": 2.8780702257392434,
-      "grad_norm": 0.002113314112648368,
-      "learning_rate": 0.0001999959145541793,
-      "loss": 46.0,
-      "step": 37643
-    },
-    {
-      "epoch": 2.878146682722633,
-      "grad_norm": 0.0016181261744350195,
-      "learning_rate": 0.00019999591433705752,
-      "loss": 46.0,
-      "step": 37644
-    },
-    {
-      "epoch": 2.878223139706023,
-      "grad_norm": 0.0011771114077419043,
-      "learning_rate": 0.00019999591411992994,
-      "loss": 46.0,
-      "step": 37645
-    },
-    {
-      "epoch": 2.8782995966894127,
-      "grad_norm": 0.0017877859063446522,
-      "learning_rate": 0.0001999959139027966,
-      "loss": 46.0,
-      "step": 37646
-    },
-    {
-      "epoch": 2.8783760536728025,
-      "grad_norm": 0.0009797796374186873,
-      "learning_rate": 0.0001999959136856575,
-      "loss": 46.0,
-      "step": 37647
-    },
-    {
-      "epoch": 2.8784525106561922,
-      "grad_norm": 0.002931591821834445,
-      "learning_rate": 0.00019999591346851262,
-      "loss": 46.0,
-      "step": 37648
-    },
-    {
-      "epoch": 2.878528967639582,
-      "grad_norm": 0.0029796601738780737,
-      "learning_rate": 0.00019999591325136197,
-      "loss": 46.0,
-      "step": 37649
-    },
-    {
-      "epoch": 2.8786054246229718,
-      "grad_norm": 0.0015961222816258669,
-      "learning_rate": 0.0001999959130342056,
-      "loss": 46.0,
-      "step": 37650
-    },
-    {
-      "epoch": 2.878681881606361,
-      "grad_norm": 0.0021464508026838303,
-      "learning_rate": 0.0001999959128170434,
-      "loss": 46.0,
-      "step": 37651
-    },
-    {
-      "epoch": 2.878758338589751,
-      "grad_norm": 0.003437078557908535,
-      "learning_rate": 0.00019999591259987546,
-      "loss": 46.0,
-      "step": 37652
-    },
-    {
-      "epoch": 2.8788347955731406,
-      "grad_norm": 0.002468088176101446,
-      "learning_rate": 0.0001999959123827017,
-      "loss": 46.0,
-      "step": 37653
-    },
-    {
-      "epoch": 2.8789112525565304,
-      "grad_norm": 0.003069260623306036,
-      "learning_rate": 0.00019999591216552225,
-      "loss": 46.0,
-      "step": 37654
-    },
-    {
-      "epoch": 2.87898770953992,
-      "grad_norm": 0.0027792982291430235,
-      "learning_rate": 0.000199995911948337,
-      "loss": 46.0,
-      "step": 37655
-    },
-    {
-      "epoch": 2.87906416652331,
-      "grad_norm": 0.0026610700879245996,
-      "learning_rate": 0.00019999591173114597,
-      "loss": 46.0,
-      "step": 37656
-    },
-    {
-      "epoch": 2.8791406235066996,
-      "grad_norm": 0.0013047675602138042,
-      "learning_rate": 0.00019999591151394916,
-      "loss": 46.0,
-      "step": 37657
-    },
-    {
-      "epoch": 2.8792170804900894,
-      "grad_norm": 0.0017877020873129368,
-      "learning_rate": 0.0001999959112967466,
-      "loss": 46.0,
-      "step": 37658
-    },
-    {
-      "epoch": 2.8792935374734787,
-      "grad_norm": 0.0022516376338899136,
-      "learning_rate": 0.00019999591107953827,
-      "loss": 46.0,
-      "step": 37659
-    },
-    {
-      "epoch": 2.8793699944568685,
-      "grad_norm": 0.0035500021185725927,
-      "learning_rate": 0.00019999591086232416,
-      "loss": 46.0,
-      "step": 37660
-    },
-    {
-      "epoch": 2.8794464514402582,
-      "grad_norm": 0.005809157621115446,
-      "learning_rate": 0.00019999591064510429,
-      "loss": 46.0,
-      "step": 37661
-    },
-    {
-      "epoch": 2.879522908423648,
-      "grad_norm": 0.0010444847866892815,
-      "learning_rate": 0.00019999591042787863,
-      "loss": 46.0,
-      "step": 37662
-    },
-    {
-      "epoch": 2.8795993654070378,
-      "grad_norm": 0.0030564642511308193,
-      "learning_rate": 0.00019999591021064724,
-      "loss": 46.0,
-      "step": 37663
-    },
-    {
-      "epoch": 2.8796758223904275,
-      "grad_norm": 0.000956640811637044,
-      "learning_rate": 0.00019999590999341004,
-      "loss": 46.0,
-      "step": 37664
-    },
-    {
-      "epoch": 2.8797522793738173,
-      "grad_norm": 0.0012026693439111114,
-      "learning_rate": 0.0001999959097761671,
-      "loss": 46.0,
-      "step": 37665
-    },
-    {
-      "epoch": 2.879828736357207,
-      "grad_norm": 0.008212125860154629,
-      "learning_rate": 0.00019999590955891838,
-      "loss": 46.0,
-      "step": 37666
-    },
-    {
-      "epoch": 2.879905193340597,
-      "grad_norm": 0.002911114599555731,
-      "learning_rate": 0.0001999959093416639,
-      "loss": 46.0,
-      "step": 37667
-    },
-    {
-      "epoch": 2.8799816503239866,
-      "grad_norm": 0.005301766097545624,
-      "learning_rate": 0.00019999590912440365,
-      "loss": 46.0,
-      "step": 37668
-    },
-    {
-      "epoch": 2.8800581073073763,
-      "grad_norm": 0.0011371088912710547,
-      "learning_rate": 0.0001999959089071376,
-      "loss": 46.0,
-      "step": 37669
-    },
-    {
-      "epoch": 2.880134564290766,
-      "grad_norm": 0.0015648382250219584,
-      "learning_rate": 0.00019999590868986582,
-      "loss": 46.0,
-      "step": 37670
-    },
-    {
-      "epoch": 2.880211021274156,
-      "grad_norm": 0.001931838458403945,
-      "learning_rate": 0.00019999590847258824,
-      "loss": 46.0,
-      "step": 37671
-    },
-    {
-      "epoch": 2.8802874782575456,
-      "grad_norm": 0.004580653738230467,
-      "learning_rate": 0.0001999959082553049,
-      "loss": 46.0,
-      "step": 37672
-    },
-    {
-      "epoch": 2.880363935240935,
-      "grad_norm": 0.0030425051227211952,
-      "learning_rate": 0.00019999590803801583,
-      "loss": 46.0,
-      "step": 37673
-    },
-    {
-      "epoch": 2.8804403922243247,
-      "grad_norm": 0.0010746218031272292,
-      "learning_rate": 0.00019999590782072093,
-      "loss": 46.0,
-      "step": 37674
-    },
-    {
-      "epoch": 2.8805168492077144,
-      "grad_norm": 0.002546555595472455,
-      "learning_rate": 0.0001999959076034203,
-      "loss": 46.0,
-      "step": 37675
-    },
-    {
-      "epoch": 2.880593306191104,
-      "grad_norm": 0.0022606172133237123,
-      "learning_rate": 0.00019999590738611388,
-      "loss": 46.0,
-      "step": 37676
-    },
-    {
-      "epoch": 2.880669763174494,
-      "grad_norm": 0.002499643713235855,
-      "learning_rate": 0.0001999959071688017,
-      "loss": 46.0,
-      "step": 37677
-    },
-    {
-      "epoch": 2.8807462201578837,
-      "grad_norm": 0.00644226698204875,
-      "learning_rate": 0.00019999590695148376,
-      "loss": 46.0,
-      "step": 37678
-    },
-    {
-      "epoch": 2.8808226771412735,
-      "grad_norm": 0.0022670149337500334,
-      "learning_rate": 0.00019999590673416004,
-      "loss": 46.0,
-      "step": 37679
-    },
-    {
-      "epoch": 2.8808991341246633,
-      "grad_norm": 0.0010460034245625138,
-      "learning_rate": 0.00019999590651683055,
-      "loss": 46.0,
-      "step": 37680
-    },
-    {
-      "epoch": 2.8809755911080526,
-      "grad_norm": 0.004056965932250023,
-      "learning_rate": 0.0001999959062994953,
-      "loss": 46.0,
-      "step": 37681
-    },
-    {
-      "epoch": 2.8810520480914423,
-      "grad_norm": 0.0026718040462583303,
-      "learning_rate": 0.00019999590608215425,
-      "loss": 46.0,
-      "step": 37682
-    },
-    {
-      "epoch": 2.881128505074832,
-      "grad_norm": 0.0020678224973380566,
-      "learning_rate": 0.00019999590586480747,
-      "loss": 46.0,
-      "step": 37683
-    },
-    {
-      "epoch": 2.881204962058222,
-      "grad_norm": 0.0012035979889333248,
-      "learning_rate": 0.0001999959056474549,
-      "loss": 46.0,
-      "step": 37684
-    },
-    {
-      "epoch": 2.8812814190416116,
-      "grad_norm": 0.004759133793413639,
-      "learning_rate": 0.00019999590543009655,
-      "loss": 46.0,
-      "step": 37685
-    },
-    {
-      "epoch": 2.8813578760250014,
-      "grad_norm": 0.0024073247332125902,
-      "learning_rate": 0.00019999590521273245,
-      "loss": 46.0,
-      "step": 37686
-    },
-    {
-      "epoch": 2.881434333008391,
-      "grad_norm": 0.0007022698991931975,
-      "learning_rate": 0.00019999590499536257,
-      "loss": 46.0,
-      "step": 37687
-    },
-    {
-      "epoch": 2.881510789991781,
-      "grad_norm": 0.0011111075291410089,
-      "learning_rate": 0.00019999590477798695,
-      "loss": 46.0,
-      "step": 37688
-    },
-    {
-      "epoch": 2.8815872469751707,
-      "grad_norm": 0.002804168965667486,
-      "learning_rate": 0.00019999590456060555,
-      "loss": 46.0,
-      "step": 37689
-    },
-    {
-      "epoch": 2.8816637039585604,
-      "grad_norm": 0.0006622043438255787,
-      "learning_rate": 0.00019999590434321835,
-      "loss": 46.0,
-      "step": 37690
-    },
-    {
-      "epoch": 2.88174016094195,
-      "grad_norm": 0.0013546813279390335,
-      "learning_rate": 0.0001999959041258254,
-      "loss": 46.0,
-      "step": 37691
-    },
-    {
-      "epoch": 2.88181661792534,
-      "grad_norm": 0.004689261317253113,
-      "learning_rate": 0.0001999959039084267,
-      "loss": 46.0,
-      "step": 37692
-    },
-    {
-      "epoch": 2.8818930749087297,
-      "grad_norm": 0.00396009162068367,
-      "learning_rate": 0.0001999959036910222,
-      "loss": 46.0,
-      "step": 37693
-    },
-    {
-      "epoch": 2.8819695318921195,
-      "grad_norm": 0.0012875613756477833,
-      "learning_rate": 0.00019999590347361194,
-      "loss": 46.0,
-      "step": 37694
-    },
-    {
-      "epoch": 2.882045988875509,
-      "grad_norm": 0.0009763381676748395,
-      "learning_rate": 0.0001999959032561959,
-      "loss": 46.0,
-      "step": 37695
-    },
-    {
-      "epoch": 2.8821224458588985,
-      "grad_norm": 0.0036177264992147684,
-      "learning_rate": 0.00019999590303877414,
-      "loss": 46.0,
-      "step": 37696
-    },
-    {
-      "epoch": 2.8821989028422883,
-      "grad_norm": 0.002841199981048703,
-      "learning_rate": 0.00019999590282134656,
-      "loss": 46.0,
-      "step": 37697
-    },
-    {
-      "epoch": 2.882275359825678,
-      "grad_norm": 0.0015320699894800782,
-      "learning_rate": 0.00019999590260391323,
-      "loss": 46.0,
-      "step": 37698
-    },
-    {
-      "epoch": 2.882351816809068,
-      "grad_norm": 0.0018972893012687564,
-      "learning_rate": 0.00019999590238647412,
-      "loss": 46.0,
-      "step": 37699
-    },
-    {
-      "epoch": 2.8824282737924576,
-      "grad_norm": 0.005009973421692848,
-      "learning_rate": 0.00019999590216902925,
-      "loss": 46.0,
-      "step": 37700
-    },
-    {
-      "epoch": 2.8825047307758473,
-      "grad_norm": 0.0012373363133519888,
-      "learning_rate": 0.0001999959019515786,
-      "loss": 46.0,
-      "step": 37701
-    },
-    {
-      "epoch": 2.882581187759237,
-      "grad_norm": 0.001896055880934,
-      "learning_rate": 0.0001999959017341222,
-      "loss": 46.0,
-      "step": 37702
-    },
-    {
-      "epoch": 2.8826576447426264,
-      "grad_norm": 0.0026955078355968,
-      "learning_rate": 0.00019999590151666,
-      "loss": 46.0,
-      "step": 37703
-    },
-    {
-      "epoch": 2.882734101726016,
-      "grad_norm": 0.001902555231936276,
-      "learning_rate": 0.00019999590129919206,
-      "loss": 46.0,
-      "step": 37704
-    },
-    {
-      "epoch": 2.882810558709406,
-      "grad_norm": 0.0016124321846291423,
-      "learning_rate": 0.00019999590108171834,
-      "loss": 46.0,
-      "step": 37705
-    },
-    {
-      "epoch": 2.8828870156927957,
-      "grad_norm": 0.002984241582453251,
-      "learning_rate": 0.00019999590086423888,
-      "loss": 46.0,
-      "step": 37706
-    },
-    {
-      "epoch": 2.8829634726761855,
-      "grad_norm": 0.0027690676506608725,
-      "learning_rate": 0.00019999590064675359,
-      "loss": 46.0,
-      "step": 37707
-    },
-    {
-      "epoch": 2.8830399296595752,
-      "grad_norm": 0.008967806585133076,
-      "learning_rate": 0.00019999590042926258,
-      "loss": 46.0,
-      "step": 37708
-    },
-    {
-      "epoch": 2.883116386642965,
-      "grad_norm": 0.001452215714380145,
-      "learning_rate": 0.00019999590021176577,
-      "loss": 46.0,
-      "step": 37709
-    },
-    {
-      "epoch": 2.8831928436263548,
-      "grad_norm": 0.0017394665628671646,
-      "learning_rate": 0.00019999589999426324,
-      "loss": 46.0,
-      "step": 37710
-    },
-    {
-      "epoch": 2.8832693006097445,
-      "grad_norm": 0.00102862564381212,
-      "learning_rate": 0.00019999589977675488,
-      "loss": 46.0,
-      "step": 37711
-    },
-    {
-      "epoch": 2.8833457575931343,
-      "grad_norm": 0.0009718839428387582,
-      "learning_rate": 0.00019999589955924078,
-      "loss": 46.0,
-      "step": 37712
-    },
-    {
-      "epoch": 2.883422214576524,
-      "grad_norm": 0.0034405116457492113,
-      "learning_rate": 0.0001999958993417209,
-      "loss": 46.0,
-      "step": 37713
-    },
-    {
-      "epoch": 2.883498671559914,
-      "grad_norm": 0.0022408231161534786,
-      "learning_rate": 0.00019999589912419525,
-      "loss": 46.0,
-      "step": 37714
-    },
-    {
-      "epoch": 2.8835751285433036,
-      "grad_norm": 0.002714663976803422,
-      "learning_rate": 0.00019999589890666382,
-      "loss": 46.0,
-      "step": 37715
-    },
-    {
-      "epoch": 2.8836515855266933,
-      "grad_norm": 0.0031120474450290203,
-      "learning_rate": 0.00019999589868912668,
-      "loss": 46.0,
-      "step": 37716
-    },
-    {
-      "epoch": 2.8837280425100826,
-      "grad_norm": 0.0015535581624135375,
-      "learning_rate": 0.00019999589847158374,
-      "loss": 46.0,
-      "step": 37717
-    },
-    {
-      "epoch": 2.8838044994934724,
-      "grad_norm": 0.0007003198261372745,
-      "learning_rate": 0.000199995898254035,
-      "loss": 46.0,
-      "step": 37718
-    },
-    {
-      "epoch": 2.883880956476862,
-      "grad_norm": 0.0026158755645155907,
-      "learning_rate": 0.00019999589803648053,
-      "loss": 46.0,
-      "step": 37719
-    },
-    {
-      "epoch": 2.883957413460252,
-      "grad_norm": 0.0011512355413287878,
-      "learning_rate": 0.00019999589781892024,
-      "loss": 46.0,
-      "step": 37720
-    },
-    {
-      "epoch": 2.8840338704436417,
-      "grad_norm": 0.0008972790092229843,
-      "learning_rate": 0.00019999589760135423,
-      "loss": 46.0,
-      "step": 37721
-    },
-    {
-      "epoch": 2.8841103274270314,
-      "grad_norm": 0.0018919005524367094,
-      "learning_rate": 0.00019999589738378245,
-      "loss": 46.0,
-      "step": 37722
-    },
-    {
-      "epoch": 2.884186784410421,
-      "grad_norm": 0.0018430177588015795,
-      "learning_rate": 0.00019999589716620486,
-      "loss": 46.0,
-      "step": 37723
-    },
-    {
-      "epoch": 2.884263241393811,
-      "grad_norm": 0.006337456405162811,
-      "learning_rate": 0.00019999589694862154,
-      "loss": 46.0,
-      "step": 37724
-    },
-    {
-      "epoch": 2.8843396983772003,
-      "grad_norm": 0.007867420092225075,
-      "learning_rate": 0.00019999589673103243,
-      "loss": 46.0,
-      "step": 37725
-    },
-    {
-      "epoch": 2.88441615536059,
-      "grad_norm": 0.00452502490952611,
-      "learning_rate": 0.00019999589651343756,
-      "loss": 46.0,
-      "step": 37726
-    },
-    {
-      "epoch": 2.88449261234398,
-      "grad_norm": 0.002269575372338295,
-      "learning_rate": 0.00019999589629583693,
-      "loss": 46.0,
-      "step": 37727
-    },
-    {
-      "epoch": 2.8845690693273696,
-      "grad_norm": 0.0023959744721651077,
-      "learning_rate": 0.00019999589607823054,
-      "loss": 46.0,
-      "step": 37728
-    },
-    {
-      "epoch": 2.8846455263107593,
-      "grad_norm": 0.0013814035337418318,
-      "learning_rate": 0.00019999589586061834,
-      "loss": 46.0,
-      "step": 37729
-    },
-    {
-      "epoch": 2.884721983294149,
-      "grad_norm": 0.0007447094540111721,
-      "learning_rate": 0.0001999958956430004,
-      "loss": 46.0,
-      "step": 37730
-    },
-    {
-      "epoch": 2.884798440277539,
-      "grad_norm": 0.0011113070650026202,
-      "learning_rate": 0.00019999589542537668,
-      "loss": 46.0,
-      "step": 37731
-    },
-    {
-      "epoch": 2.8848748972609286,
-      "grad_norm": 0.0012835300294682384,
-      "learning_rate": 0.00019999589520774716,
-      "loss": 46.0,
-      "step": 37732
-    },
-    {
-      "epoch": 2.8849513542443184,
-      "grad_norm": 0.0033381865359842777,
-      "learning_rate": 0.0001999958949901119,
-      "loss": 46.0,
-      "step": 37733
-    },
-    {
-      "epoch": 2.885027811227708,
-      "grad_norm": 0.0037939741741865873,
-      "learning_rate": 0.0001999958947724709,
-      "loss": 46.0,
-      "step": 37734
-    },
-    {
-      "epoch": 2.885104268211098,
-      "grad_norm": 0.0017376294126734138,
-      "learning_rate": 0.0001999958945548241,
-      "loss": 46.0,
-      "step": 37735
-    },
-    {
-      "epoch": 2.8851807251944876,
-      "grad_norm": 0.003165578003972769,
-      "learning_rate": 0.00019999589433717155,
-      "loss": 46.0,
-      "step": 37736
-    },
-    {
-      "epoch": 2.8852571821778774,
-      "grad_norm": 0.006535191088914871,
-      "learning_rate": 0.0001999958941195132,
-      "loss": 46.0,
-      "step": 37737
-    },
-    {
-      "epoch": 2.8853336391612667,
-      "grad_norm": 0.001833610818721354,
-      "learning_rate": 0.0001999958939018491,
-      "loss": 46.0,
-      "step": 37738
-    },
-    {
-      "epoch": 2.8854100961446565,
-      "grad_norm": 0.0008515738882124424,
-      "learning_rate": 0.00019999589368417922,
-      "loss": 46.0,
-      "step": 37739
-    },
-    {
-      "epoch": 2.8854865531280462,
-      "grad_norm": 0.0022605929989367723,
-      "learning_rate": 0.0001999958934665036,
-      "loss": 46.0,
-      "step": 37740
-    },
-    {
-      "epoch": 2.885563010111436,
-      "grad_norm": 0.001866824459284544,
-      "learning_rate": 0.00019999589324882217,
-      "loss": 46.0,
-      "step": 37741
-    },
-    {
-      "epoch": 2.8856394670948258,
-      "grad_norm": 0.0014304748037829995,
-      "learning_rate": 0.000199995893031135,
-      "loss": 46.0,
-      "step": 37742
-    },
-    {
-      "epoch": 2.8857159240782155,
-      "grad_norm": 0.001404367620125413,
-      "learning_rate": 0.00019999589281344203,
-      "loss": 46.0,
-      "step": 37743
-    },
-    {
-      "epoch": 2.8857923810616053,
-      "grad_norm": 0.010288607329130173,
-      "learning_rate": 0.00019999589259574334,
-      "loss": 46.0,
-      "step": 37744
-    },
-    {
-      "epoch": 2.885868838044995,
-      "grad_norm": 0.0036197949666529894,
-      "learning_rate": 0.00019999589237803885,
-      "loss": 46.0,
-      "step": 37745
-    },
-    {
-      "epoch": 2.885945295028385,
-      "grad_norm": 0.0028814945835620165,
-      "learning_rate": 0.0001999958921603286,
-      "loss": 46.0,
-      "step": 37746
-    },
-    {
-      "epoch": 2.886021752011774,
-      "grad_norm": 0.001936594839207828,
-      "learning_rate": 0.00019999589194261255,
-      "loss": 46.0,
-      "step": 37747
-    },
-    {
-      "epoch": 2.886098208995164,
-      "grad_norm": 0.002066792454570532,
-      "learning_rate": 0.00019999589172489074,
-      "loss": 46.0,
-      "step": 37748
-    },
-    {
-      "epoch": 2.8861746659785537,
-      "grad_norm": 0.002139067742973566,
-      "learning_rate": 0.0001999958915071632,
-      "loss": 46.0,
-      "step": 37749
-    },
-    {
-      "epoch": 2.8862511229619434,
-      "grad_norm": 0.003388867946341634,
-      "learning_rate": 0.0001999958912894299,
-      "loss": 46.0,
-      "step": 37750
-    },
-    {
-      "epoch": 2.886327579945333,
-      "grad_norm": 0.004089138004928827,
-      "learning_rate": 0.00019999589107169076,
-      "loss": 46.0,
-      "step": 37751
-    },
-    {
-      "epoch": 2.886404036928723,
-      "grad_norm": 0.0010730591602623463,
-      "learning_rate": 0.00019999589085394588,
-      "loss": 46.0,
-      "step": 37752
-    },
-    {
-      "epoch": 2.8864804939121127,
-      "grad_norm": 0.003603886580094695,
-      "learning_rate": 0.00019999589063619526,
-      "loss": 46.0,
-      "step": 37753
-    },
-    {
-      "epoch": 2.8865569508955025,
-      "grad_norm": 0.0019759265705943108,
-      "learning_rate": 0.00019999589041843884,
-      "loss": 46.0,
-      "step": 37754
-    },
-    {
-      "epoch": 2.886633407878892,
-      "grad_norm": 0.002615033881738782,
-      "learning_rate": 0.00019999589020067667,
-      "loss": 46.0,
-      "step": 37755
-    },
-    {
-      "epoch": 2.886709864862282,
-      "grad_norm": 0.0016685740556567907,
-      "learning_rate": 0.00019999588998290872,
-      "loss": 46.0,
-      "step": 37756
-    },
-    {
-      "epoch": 2.8867863218456717,
-      "grad_norm": 0.002141603035852313,
-      "learning_rate": 0.00019999588976513498,
-      "loss": 46.0,
-      "step": 37757
-    },
-    {
-      "epoch": 2.8868627788290615,
-      "grad_norm": 0.00145918398629874,
-      "learning_rate": 0.00019999588954735552,
-      "loss": 46.0,
-      "step": 37758
-    },
-    {
-      "epoch": 2.8869392358124513,
-      "grad_norm": 0.0019381509628146887,
-      "learning_rate": 0.00019999588932957026,
-      "loss": 46.0,
-      "step": 37759
-    },
-    {
-      "epoch": 2.8870156927958406,
-      "grad_norm": 0.0014604333555325866,
-      "learning_rate": 0.00019999588911177922,
-      "loss": 46.0,
-      "step": 37760
-    },
-    {
-      "epoch": 2.8870921497792303,
-      "grad_norm": 0.001315958914346993,
-      "learning_rate": 0.00019999588889398244,
-      "loss": 46.0,
-      "step": 37761
-    },
-    {
-      "epoch": 2.88716860676262,
-      "grad_norm": 0.0018504865001887083,
-      "learning_rate": 0.00019999588867617986,
-      "loss": 46.0,
-      "step": 37762
-    },
-    {
-      "epoch": 2.88724506374601,
-      "grad_norm": 0.0075893704779446125,
-      "learning_rate": 0.00019999588845837153,
-      "loss": 46.0,
-      "step": 37763
-    },
-    {
-      "epoch": 2.8873215207293996,
-      "grad_norm": 0.002132125198841095,
-      "learning_rate": 0.00019999588824055743,
-      "loss": 46.0,
-      "step": 37764
-    },
-    {
-      "epoch": 2.8873979777127894,
-      "grad_norm": 0.003499071579426527,
-      "learning_rate": 0.00019999588802273758,
-      "loss": 46.0,
-      "step": 37765
-    },
-    {
-      "epoch": 2.887474434696179,
-      "grad_norm": 0.012519373558461666,
-      "learning_rate": 0.00019999588780491193,
-      "loss": 46.0,
-      "step": 37766
-    },
-    {
-      "epoch": 2.887550891679569,
-      "grad_norm": 0.0013570141745731235,
-      "learning_rate": 0.00019999588758708053,
-      "loss": 46.0,
-      "step": 37767
-    },
-    {
-      "epoch": 2.8876273486629582,
-      "grad_norm": 0.012234872207045555,
-      "learning_rate": 0.00019999588736924334,
-      "loss": 46.0,
-      "step": 37768
-    },
-    {
-      "epoch": 2.887703805646348,
-      "grad_norm": 0.0023281071335077286,
-      "learning_rate": 0.0001999958871514004,
-      "loss": 46.0,
-      "step": 37769
-    },
-    {
-      "epoch": 2.8877802626297377,
-      "grad_norm": 0.0019528227858245373,
-      "learning_rate": 0.00019999588693355168,
-      "loss": 46.0,
-      "step": 37770
-    },
-    {
-      "epoch": 2.8878567196131275,
-      "grad_norm": 0.0017250468954443932,
-      "learning_rate": 0.0001999958867156972,
-      "loss": 46.0,
-      "step": 37771
-    },
-    {
-      "epoch": 2.8879331765965173,
-      "grad_norm": 0.004837709944695234,
-      "learning_rate": 0.0001999958864978369,
-      "loss": 46.0,
-      "step": 37772
-    },
-    {
-      "epoch": 2.888009633579907,
-      "grad_norm": 0.0012939043808728456,
-      "learning_rate": 0.0001999958862799709,
-      "loss": 46.0,
-      "step": 37773
-    },
-    {
-      "epoch": 2.888086090563297,
-      "grad_norm": 0.001408212585374713,
-      "learning_rate": 0.00019999588606209908,
-      "loss": 46.0,
-      "step": 37774
-    },
-    {
-      "epoch": 2.8881625475466866,
-      "grad_norm": 0.002098650438711047,
-      "learning_rate": 0.00019999588584422153,
-      "loss": 46.0,
-      "step": 37775
-    },
-    {
-      "epoch": 2.8882390045300763,
-      "grad_norm": 0.0009064716869033873,
-      "learning_rate": 0.0001999958856263382,
-      "loss": 46.0,
-      "step": 37776
-    },
-    {
-      "epoch": 2.888315461513466,
-      "grad_norm": 0.0009372316999360919,
-      "learning_rate": 0.00019999588540844913,
-      "loss": 46.0,
-      "step": 37777
-    },
-    {
-      "epoch": 2.888391918496856,
-      "grad_norm": 0.007355047389864922,
-      "learning_rate": 0.00019999588519055422,
-      "loss": 46.0,
-      "step": 37778
-    },
-    {
-      "epoch": 2.8884683754802456,
-      "grad_norm": 0.002255040919408202,
-      "learning_rate": 0.00019999588497265357,
-      "loss": 46.0,
-      "step": 37779
-    },
-    {
-      "epoch": 2.8885448324636354,
-      "grad_norm": 0.0029307508375495672,
-      "learning_rate": 0.00019999588475474718,
-      "loss": 46.0,
-      "step": 37780
-    },
-    {
-      "epoch": 2.888621289447025,
-      "grad_norm": 0.0066835288889706135,
-      "learning_rate": 0.000199995884536835,
-      "loss": 46.0,
-      "step": 37781
-    },
-    {
-      "epoch": 2.8886977464304144,
-      "grad_norm": 0.004918511491268873,
-      "learning_rate": 0.00019999588431891704,
-      "loss": 46.0,
-      "step": 37782
-    },
-    {
-      "epoch": 2.888774203413804,
-      "grad_norm": 0.0014031730825081468,
-      "learning_rate": 0.00019999588410099333,
-      "loss": 46.0,
-      "step": 37783
-    },
-    {
-      "epoch": 2.888850660397194,
-      "grad_norm": 0.0015238840132951736,
-      "learning_rate": 0.00019999588388306384,
-      "loss": 46.0,
-      "step": 37784
-    },
-    {
-      "epoch": 2.8889271173805837,
-      "grad_norm": 0.0018328707665205002,
-      "learning_rate": 0.0001999958836651286,
-      "loss": 46.0,
-      "step": 37785
-    },
-    {
-      "epoch": 2.8890035743639735,
-      "grad_norm": 0.005886287894099951,
-      "learning_rate": 0.00019999588344718757,
-      "loss": 46.0,
-      "step": 37786
-    },
-    {
-      "epoch": 2.8890800313473632,
-      "grad_norm": 0.002663286868482828,
-      "learning_rate": 0.0001999958832292408,
-      "loss": 46.0,
-      "step": 37787
-    },
-    {
-      "epoch": 2.889156488330753,
-      "grad_norm": 0.006685193628072739,
-      "learning_rate": 0.0001999958830112882,
-      "loss": 46.0,
-      "step": 37788
-    },
-    {
-      "epoch": 2.8892329453141428,
-      "grad_norm": 0.003558224765583873,
-      "learning_rate": 0.00019999588279332988,
-      "loss": 46.0,
-      "step": 37789
-    },
-    {
-      "epoch": 2.889309402297532,
-      "grad_norm": 0.0012884949101135135,
-      "learning_rate": 0.0001999958825753658,
-      "loss": 46.0,
-      "step": 37790
-    },
-    {
-      "epoch": 2.889385859280922,
-      "grad_norm": 0.0031954103615134954,
-      "learning_rate": 0.0001999958823573959,
-      "loss": 46.0,
-      "step": 37791
-    },
-    {
-      "epoch": 2.8894623162643116,
-      "grad_norm": 0.0008383719832636416,
-      "learning_rate": 0.00019999588213942025,
-      "loss": 46.0,
-      "step": 37792
-    },
-    {
-      "epoch": 2.8895387732477014,
-      "grad_norm": 0.009652224369347095,
-      "learning_rate": 0.00019999588192143886,
-      "loss": 46.0,
-      "step": 37793
-    },
-    {
-      "epoch": 2.889615230231091,
-      "grad_norm": 0.0019310020143166184,
-      "learning_rate": 0.00019999588170345166,
-      "loss": 46.0,
-      "step": 37794
-    },
-    {
-      "epoch": 2.889691687214481,
-      "grad_norm": 0.0010015071602538228,
-      "learning_rate": 0.00019999588148545875,
-      "loss": 46.0,
-      "step": 37795
-    },
-    {
-      "epoch": 2.8897681441978706,
-      "grad_norm": 0.0035619321279227734,
-      "learning_rate": 0.00019999588126746,
-      "loss": 46.0,
-      "step": 37796
-    },
-    {
-      "epoch": 2.8898446011812604,
-      "grad_norm": 0.005210161209106445,
-      "learning_rate": 0.00019999588104945552,
-      "loss": 46.0,
-      "step": 37797
-    },
-    {
-      "epoch": 2.88992105816465,
-      "grad_norm": 0.0010614673374220729,
-      "learning_rate": 0.00019999588083144528,
-      "loss": 46.0,
-      "step": 37798
-    },
-    {
-      "epoch": 2.88999751514804,
-      "grad_norm": 0.0012027255725115538,
-      "learning_rate": 0.00019999588061342925,
-      "loss": 46.0,
-      "step": 37799
-    },
-    {
-      "epoch": 2.8900739721314297,
-      "grad_norm": 0.0021582175977528095,
-      "learning_rate": 0.00019999588039540744,
-      "loss": 46.0,
-      "step": 37800
-    },
-    {
-      "epoch": 2.8901504291148195,
-      "grad_norm": 0.0008120203274302185,
-      "learning_rate": 0.00019999588017737989,
-      "loss": 46.0,
-      "step": 37801
-    },
-    {
-      "epoch": 2.890226886098209,
-      "grad_norm": 0.007342230994254351,
-      "learning_rate": 0.00019999587995934653,
-      "loss": 46.0,
-      "step": 37802
-    },
-    {
-      "epoch": 2.890303343081599,
-      "grad_norm": 0.004231353290379047,
-      "learning_rate": 0.00019999587974130746,
-      "loss": 46.0,
-      "step": 37803
-    },
-    {
-      "epoch": 2.8903798000649883,
-      "grad_norm": 0.0017804204253479838,
-      "learning_rate": 0.00019999587952326258,
-      "loss": 46.0,
-      "step": 37804
-    },
-    {
-      "epoch": 2.890456257048378,
-      "grad_norm": 0.004549276549369097,
-      "learning_rate": 0.00019999587930521193,
-      "loss": 46.0,
-      "step": 37805
-    },
-    {
-      "epoch": 2.890532714031768,
-      "grad_norm": 0.0044477032497525215,
-      "learning_rate": 0.00019999587908715554,
-      "loss": 46.0,
-      "step": 37806
-    },
-    {
-      "epoch": 2.8906091710151576,
-      "grad_norm": 0.0025124498642981052,
-      "learning_rate": 0.00019999587886909334,
-      "loss": 46.0,
-      "step": 37807
-    },
-    {
-      "epoch": 2.8906856279985473,
-      "grad_norm": 0.002882106462493539,
-      "learning_rate": 0.0001999958786510254,
-      "loss": 46.0,
-      "step": 37808
-    },
-    {
-      "epoch": 2.890762084981937,
-      "grad_norm": 0.002985223662108183,
-      "learning_rate": 0.0001999958784329517,
-      "loss": 46.0,
-      "step": 37809
-    },
-    {
-      "epoch": 2.890838541965327,
-      "grad_norm": 0.0028906241059303284,
-      "learning_rate": 0.0001999958782148722,
-      "loss": 46.0,
-      "step": 37810
-    },
-    {
-      "epoch": 2.8909149989487166,
-      "grad_norm": 0.003020563628524542,
-      "learning_rate": 0.00019999587799678694,
-      "loss": 46.0,
-      "step": 37811
-    },
-    {
-      "epoch": 2.890991455932106,
-      "grad_norm": 0.0023545341100543737,
-      "learning_rate": 0.00019999587777869593,
-      "loss": 46.0,
-      "step": 37812
-    },
-    {
-      "epoch": 2.8910679129154957,
-      "grad_norm": 0.0009672415908426046,
-      "learning_rate": 0.00019999587756059912,
-      "loss": 46.0,
-      "step": 37813
-    },
-    {
-      "epoch": 2.8911443698988855,
-      "grad_norm": 0.0019381796009838581,
-      "learning_rate": 0.00019999587734249657,
-      "loss": 46.0,
-      "step": 37814
-    },
-    {
-      "epoch": 2.891220826882275,
-      "grad_norm": 0.003142933128401637,
-      "learning_rate": 0.00019999587712438822,
-      "loss": 46.0,
-      "step": 37815
-    },
-    {
-      "epoch": 2.891297283865665,
-      "grad_norm": 0.001372526166960597,
-      "learning_rate": 0.00019999587690627411,
-      "loss": 46.0,
-      "step": 37816
-    },
-    {
-      "epoch": 2.8913737408490547,
-      "grad_norm": 0.0015170314582064748,
-      "learning_rate": 0.00019999587668815424,
-      "loss": 46.0,
-      "step": 37817
-    },
-    {
-      "epoch": 2.8914501978324445,
-      "grad_norm": 0.000829146767500788,
-      "learning_rate": 0.00019999587647002862,
-      "loss": 46.0,
-      "step": 37818
-    },
-    {
-      "epoch": 2.8915266548158343,
-      "grad_norm": 0.0006197000038810074,
-      "learning_rate": 0.0001999958762518972,
-      "loss": 46.0,
-      "step": 37819
-    },
-    {
-      "epoch": 2.891603111799224,
-      "grad_norm": 0.003329264698550105,
-      "learning_rate": 0.00019999587603376003,
-      "loss": 46.0,
-      "step": 37820
-    },
-    {
-      "epoch": 2.891679568782614,
-      "grad_norm": 0.0018131479155272245,
-      "learning_rate": 0.00019999587581561706,
-      "loss": 46.0,
-      "step": 37821
-    },
-    {
-      "epoch": 2.8917560257660035,
-      "grad_norm": 0.003074378240853548,
-      "learning_rate": 0.00019999587559746838,
-      "loss": 46.0,
-      "step": 37822
-    },
-    {
-      "epoch": 2.8918324827493933,
-      "grad_norm": 0.004379425197839737,
-      "learning_rate": 0.00019999587537931386,
-      "loss": 46.0,
-      "step": 37823
-    },
-    {
-      "epoch": 2.891908939732783,
-      "grad_norm": 0.0022632989566773176,
-      "learning_rate": 0.0001999958751611536,
-      "loss": 46.0,
-      "step": 37824
-    },
-    {
-      "epoch": 2.891985396716173,
-      "grad_norm": 0.0033311517909169197,
-      "learning_rate": 0.0001999958749429876,
-      "loss": 46.0,
-      "step": 37825
-    },
-    {
-      "epoch": 2.892061853699562,
-      "grad_norm": 0.0018971188692376018,
-      "learning_rate": 0.0001999958747248158,
-      "loss": 46.0,
-      "step": 37826
-    },
-    {
-      "epoch": 2.892138310682952,
-      "grad_norm": 0.0021489844657480717,
-      "learning_rate": 0.00019999587450663823,
-      "loss": 46.0,
-      "step": 37827
-    },
-    {
-      "epoch": 2.8922147676663417,
-      "grad_norm": 0.0018439419800415635,
-      "learning_rate": 0.0001999958742884549,
-      "loss": 46.0,
-      "step": 37828
-    },
-    {
-      "epoch": 2.8922912246497314,
-      "grad_norm": 0.0034569138661026955,
-      "learning_rate": 0.00019999587407026578,
-      "loss": 46.0,
-      "step": 37829
-    },
-    {
-      "epoch": 2.892367681633121,
-      "grad_norm": 0.005139507353305817,
-      "learning_rate": 0.00019999587385207096,
-      "loss": 46.0,
-      "step": 37830
-    },
-    {
-      "epoch": 2.892444138616511,
-      "grad_norm": 0.0027904980815947056,
-      "learning_rate": 0.00019999587363387028,
-      "loss": 46.0,
-      "step": 37831
-    },
-    {
-      "epoch": 2.8925205955999007,
-      "grad_norm": 0.0033358149230480194,
-      "learning_rate": 0.0001999958734156639,
-      "loss": 46.0,
-      "step": 37832
-    },
-    {
-      "epoch": 2.8925970525832905,
-      "grad_norm": 0.0018645493546500802,
-      "learning_rate": 0.0001999958731974517,
-      "loss": 46.0,
-      "step": 37833
-    },
-    {
-      "epoch": 2.89267350956668,
-      "grad_norm": 0.0033399993553757668,
-      "learning_rate": 0.00019999587297923375,
-      "loss": 46.0,
-      "step": 37834
-    },
-    {
-      "epoch": 2.8927499665500696,
-      "grad_norm": 0.004222318064421415,
-      "learning_rate": 0.00019999587276101004,
-      "loss": 46.0,
-      "step": 37835
-    },
-    {
-      "epoch": 2.8928264235334593,
-      "grad_norm": 0.0015192920109257102,
-      "learning_rate": 0.00019999587254278053,
-      "loss": 46.0,
-      "step": 37836
-    },
-    {
-      "epoch": 2.892902880516849,
-      "grad_norm": 0.005047188140451908,
-      "learning_rate": 0.0001999958723245453,
-      "loss": 46.0,
-      "step": 37837
-    },
-    {
-      "epoch": 2.892979337500239,
-      "grad_norm": 0.0034721423871815205,
-      "learning_rate": 0.00019999587210630426,
-      "loss": 46.0,
-      "step": 37838
-    },
-    {
-      "epoch": 2.8930557944836286,
-      "grad_norm": 0.005029464140534401,
-      "learning_rate": 0.00019999587188805748,
-      "loss": 46.0,
-      "step": 37839
-    },
-    {
-      "epoch": 2.8931322514670184,
-      "grad_norm": 0.0020497071091085672,
-      "learning_rate": 0.00019999587166980492,
-      "loss": 46.0,
-      "step": 37840
-    },
-    {
-      "epoch": 2.893208708450408,
-      "grad_norm": 0.006000366061925888,
-      "learning_rate": 0.0001999958714515466,
-      "loss": 46.0,
-      "step": 37841
-    },
-    {
-      "epoch": 2.893285165433798,
-      "grad_norm": 0.0014674197882413864,
-      "learning_rate": 0.00019999587123328247,
-      "loss": 46.0,
-      "step": 37842
-    },
-    {
-      "epoch": 2.8933616224171876,
-      "grad_norm": 0.0005276089068502188,
-      "learning_rate": 0.00019999587101501262,
-      "loss": 46.0,
-      "step": 37843
-    },
-    {
-      "epoch": 2.8934380794005774,
-      "grad_norm": 0.0016725932946428657,
-      "learning_rate": 0.00019999587079673695,
-      "loss": 46.0,
-      "step": 37844
-    },
-    {
-      "epoch": 2.893514536383967,
-      "grad_norm": 0.0006369533366523683,
-      "learning_rate": 0.00019999587057845556,
-      "loss": 46.0,
-      "step": 37845
-    },
-    {
-      "epoch": 2.893590993367357,
-      "grad_norm": 0.0029901338275521994,
-      "learning_rate": 0.0001999958703601684,
-      "loss": 46.0,
-      "step": 37846
-    },
-    {
-      "epoch": 2.8936674503507467,
-      "grad_norm": 0.005535561591386795,
-      "learning_rate": 0.00019999587014187545,
-      "loss": 46.0,
-      "step": 37847
-    },
-    {
-      "epoch": 2.893743907334136,
-      "grad_norm": 0.001583735691383481,
-      "learning_rate": 0.0001999958699235767,
-      "loss": 46.0,
-      "step": 37848
-    },
-    {
-      "epoch": 2.8938203643175258,
-      "grad_norm": 0.005722708534449339,
-      "learning_rate": 0.00019999586970527222,
-      "loss": 46.0,
-      "step": 37849
-    },
-    {
-      "epoch": 2.8938968213009155,
-      "grad_norm": 0.0035355426371097565,
-      "learning_rate": 0.000199995869486962,
-      "loss": 46.0,
-      "step": 37850
-    },
-    {
-      "epoch": 2.8939732782843053,
-      "grad_norm": 0.013533324934542179,
-      "learning_rate": 0.00019999586926864595,
-      "loss": 46.0,
-      "step": 37851
-    },
-    {
-      "epoch": 2.894049735267695,
-      "grad_norm": 0.0013722516596317291,
-      "learning_rate": 0.00019999586905032417,
-      "loss": 46.0,
-      "step": 37852
-    },
-    {
-      "epoch": 2.894126192251085,
-      "grad_norm": 0.005028895568102598,
-      "learning_rate": 0.0001999958688319966,
-      "loss": 46.0,
-      "step": 37853
-    },
-    {
-      "epoch": 2.8942026492344746,
-      "grad_norm": 0.004545052070170641,
-      "learning_rate": 0.0001999958686136633,
-      "loss": 46.0,
-      "step": 37854
-    },
-    {
-      "epoch": 2.8942791062178643,
-      "grad_norm": 0.0012598909670487046,
-      "learning_rate": 0.00019999586839532417,
-      "loss": 46.0,
-      "step": 37855
-    },
-    {
-      "epoch": 2.8943555632012536,
-      "grad_norm": 0.0005189222865737975,
-      "learning_rate": 0.0001999958681769793,
-      "loss": 46.0,
-      "step": 37856
-    },
-    {
-      "epoch": 2.8944320201846434,
-      "grad_norm": 0.0014284258941188455,
-      "learning_rate": 0.00019999586795862865,
-      "loss": 46.0,
-      "step": 37857
-    },
-    {
-      "epoch": 2.894508477168033,
-      "grad_norm": 0.0020104290451854467,
-      "learning_rate": 0.00019999586774027223,
-      "loss": 46.0,
-      "step": 37858
-    },
-    {
-      "epoch": 2.894584934151423,
-      "grad_norm": 0.0006753055495209992,
-      "learning_rate": 0.00019999586752191006,
-      "loss": 46.0,
-      "step": 37859
-    },
-    {
-      "epoch": 2.8946613911348127,
-      "grad_norm": 0.0008774510934017599,
-      "learning_rate": 0.00019999586730354212,
-      "loss": 46.0,
-      "step": 37860
-    },
-    {
-      "epoch": 2.8947378481182025,
-      "grad_norm": 0.0029588660690933466,
-      "learning_rate": 0.0001999958670851684,
-      "loss": 46.0,
-      "step": 37861
-    },
-    {
-      "epoch": 2.894814305101592,
-      "grad_norm": 0.0017169112106785178,
-      "learning_rate": 0.0001999958668667889,
-      "loss": 46.0,
-      "step": 37862
-    },
-    {
-      "epoch": 2.894890762084982,
-      "grad_norm": 0.001216494245454669,
-      "learning_rate": 0.00019999586664840366,
-      "loss": 46.0,
-      "step": 37863
-    },
-    {
-      "epoch": 2.8949672190683717,
-      "grad_norm": 0.0003851375658996403,
-      "learning_rate": 0.00019999586643001265,
-      "loss": 46.0,
-      "step": 37864
-    },
-    {
-      "epoch": 2.8950436760517615,
-      "grad_norm": 0.0019412519177421927,
-      "learning_rate": 0.00019999586621161582,
-      "loss": 46.0,
-      "step": 37865
-    },
-    {
-      "epoch": 2.8951201330351513,
-      "grad_norm": 0.0020202603191137314,
-      "learning_rate": 0.0001999958659932133,
-      "loss": 46.0,
-      "step": 37866
-    },
-    {
-      "epoch": 2.895196590018541,
-      "grad_norm": 0.0009198561892844737,
-      "learning_rate": 0.00019999586577480497,
-      "loss": 46.0,
-      "step": 37867
-    },
-    {
-      "epoch": 2.8952730470019308,
-      "grad_norm": 0.0039591402746737,
-      "learning_rate": 0.00019999586555639084,
-      "loss": 46.0,
-      "step": 37868
-    },
-    {
-      "epoch": 2.89534950398532,
-      "grad_norm": 0.002076217206194997,
-      "learning_rate": 0.000199995865337971,
-      "loss": 46.0,
-      "step": 37869
-    },
-    {
-      "epoch": 2.89542596096871,
-      "grad_norm": 0.0032474645413458347,
-      "learning_rate": 0.00019999586511954535,
-      "loss": 46.0,
-      "step": 37870
-    },
-    {
-      "epoch": 2.8955024179520996,
-      "grad_norm": 0.0005519369151443243,
-      "learning_rate": 0.00019999586490111396,
-      "loss": 46.0,
-      "step": 37871
-    },
-    {
-      "epoch": 2.8955788749354894,
-      "grad_norm": 0.0035198063123971224,
-      "learning_rate": 0.00019999586468267676,
-      "loss": 46.0,
-      "step": 37872
-    },
-    {
-      "epoch": 2.895655331918879,
-      "grad_norm": 0.004100221209228039,
-      "learning_rate": 0.00019999586446423382,
-      "loss": 46.0,
-      "step": 37873
-    },
-    {
-      "epoch": 2.895731788902269,
-      "grad_norm": 0.0019092346774414182,
-      "learning_rate": 0.00019999586424578508,
-      "loss": 46.0,
-      "step": 37874
-    },
-    {
-      "epoch": 2.8958082458856587,
-      "grad_norm": 0.0016025708755478263,
-      "learning_rate": 0.00019999586402733062,
-      "loss": 46.0,
-      "step": 37875
-    },
-    {
-      "epoch": 2.8958847028690484,
-      "grad_norm": 0.0012079612351953983,
-      "learning_rate": 0.00019999586380887037,
-      "loss": 46.0,
-      "step": 37876
-    },
-    {
-      "epoch": 2.895961159852438,
-      "grad_norm": 0.0009891492081806064,
-      "learning_rate": 0.00019999586359040433,
-      "loss": 46.0,
-      "step": 37877
-    },
-    {
-      "epoch": 2.8960376168358275,
-      "grad_norm": 0.0038088688161224127,
-      "learning_rate": 0.00019999586337193255,
-      "loss": 46.0,
-      "step": 37878
-    },
-    {
-      "epoch": 2.8961140738192173,
-      "grad_norm": 0.0020113675855100155,
-      "learning_rate": 0.00019999586315345497,
-      "loss": 46.0,
-      "step": 37879
-    },
-    {
-      "epoch": 2.896190530802607,
-      "grad_norm": 0.0032350497785955667,
-      "learning_rate": 0.00019999586293497165,
-      "loss": 46.0,
-      "step": 37880
-    },
-    {
-      "epoch": 2.896266987785997,
-      "grad_norm": 0.0036726954858750105,
-      "learning_rate": 0.00019999586271648255,
-      "loss": 46.0,
-      "step": 37881
-    },
-    {
-      "epoch": 2.8963434447693865,
-      "grad_norm": 0.005590291228145361,
-      "learning_rate": 0.00019999586249798767,
-      "loss": 46.0,
-      "step": 37882
-    },
-    {
-      "epoch": 2.8964199017527763,
-      "grad_norm": 0.0034286403097212315,
-      "learning_rate": 0.00019999586227948703,
-      "loss": 46.0,
-      "step": 37883
-    },
-    {
-      "epoch": 2.896496358736166,
-      "grad_norm": 0.002362036146223545,
-      "learning_rate": 0.0001999958620609806,
-      "loss": 46.0,
-      "step": 37884
-    },
-    {
-      "epoch": 2.896572815719556,
-      "grad_norm": 0.0029984493739902973,
-      "learning_rate": 0.00019999586184246844,
-      "loss": 46.0,
-      "step": 37885
-    },
-    {
-      "epoch": 2.8966492727029456,
-      "grad_norm": 0.002623130800202489,
-      "learning_rate": 0.0001999958616239505,
-      "loss": 46.0,
-      "step": 37886
-    },
-    {
-      "epoch": 2.8967257296863353,
-      "grad_norm": 0.0006159718031994998,
-      "learning_rate": 0.0001999958614054268,
-      "loss": 46.0,
-      "step": 37887
-    },
-    {
-      "epoch": 2.896802186669725,
-      "grad_norm": 0.005858476273715496,
-      "learning_rate": 0.0001999958611868973,
-      "loss": 46.0,
-      "step": 37888
-    },
-    {
-      "epoch": 2.896878643653115,
-      "grad_norm": 0.003088385798037052,
-      "learning_rate": 0.00019999586096836205,
-      "loss": 46.0,
-      "step": 37889
-    },
-    {
-      "epoch": 2.8969551006365046,
-      "grad_norm": 0.0019597543869167566,
-      "learning_rate": 0.000199995860749821,
-      "loss": 46.0,
-      "step": 37890
-    },
-    {
-      "epoch": 2.897031557619894,
-      "grad_norm": 0.0020084744319319725,
-      "learning_rate": 0.0001999958605312742,
-      "loss": 46.0,
-      "step": 37891
-    },
-    {
-      "epoch": 2.8971080146032837,
-      "grad_norm": 0.0012871321523562074,
-      "learning_rate": 0.00019999586031272165,
-      "loss": 46.0,
-      "step": 37892
-    },
-    {
-      "epoch": 2.8971844715866735,
-      "grad_norm": 0.001935808570124209,
-      "learning_rate": 0.00019999586009416333,
-      "loss": 46.0,
-      "step": 37893
-    },
-    {
-      "epoch": 2.8972609285700632,
-      "grad_norm": 0.003490347880870104,
-      "learning_rate": 0.00019999585987559923,
-      "loss": 46.0,
-      "step": 37894
-    },
-    {
-      "epoch": 2.897337385553453,
-      "grad_norm": 0.0012122029438614845,
-      "learning_rate": 0.00019999585965702936,
-      "loss": 46.0,
-      "step": 37895
-    },
-    {
-      "epoch": 2.8974138425368428,
-      "grad_norm": 0.000973303453065455,
-      "learning_rate": 0.0001999958594384537,
-      "loss": 46.0,
-      "step": 37896
-    },
-    {
-      "epoch": 2.8974902995202325,
-      "grad_norm": 0.0017656109994277358,
-      "learning_rate": 0.00019999585921987232,
-      "loss": 46.0,
-      "step": 37897
-    },
-    {
-      "epoch": 2.8975667565036223,
-      "grad_norm": 0.004450140055269003,
-      "learning_rate": 0.00019999585900128515,
-      "loss": 46.0,
-      "step": 37898
-    },
-    {
-      "epoch": 2.8976432134870116,
-      "grad_norm": 0.004468852654099464,
-      "learning_rate": 0.0001999958587826922,
-      "loss": 46.0,
-      "step": 37899
-    },
-    {
-      "epoch": 2.8977196704704014,
-      "grad_norm": 0.0026200194843113422,
-      "learning_rate": 0.00019999585856409348,
-      "loss": 46.0,
-      "step": 37900
-    },
-    {
-      "epoch": 2.897796127453791,
-      "grad_norm": 0.0022780667059123516,
-      "learning_rate": 0.000199995858345489,
-      "loss": 46.0,
-      "step": 37901
-    },
-    {
-      "epoch": 2.897872584437181,
-      "grad_norm": 0.00370429758913815,
-      "learning_rate": 0.00019999585812687873,
-      "loss": 46.0,
-      "step": 37902
-    },
-    {
-      "epoch": 2.8979490414205706,
-      "grad_norm": 0.0027000715490430593,
-      "learning_rate": 0.00019999585790826273,
-      "loss": 46.0,
-      "step": 37903
-    },
-    {
-      "epoch": 2.8980254984039604,
-      "grad_norm": 0.0034705905709415674,
-      "learning_rate": 0.00019999585768964092,
-      "loss": 46.0,
-      "step": 37904
-    },
-    {
-      "epoch": 2.89810195538735,
-      "grad_norm": 0.0009451345540583134,
-      "learning_rate": 0.00019999585747101337,
-      "loss": 46.0,
-      "step": 37905
-    },
-    {
-      "epoch": 2.89817841237074,
-      "grad_norm": 0.0009521275642327964,
-      "learning_rate": 0.00019999585725238004,
-      "loss": 46.0,
-      "step": 37906
-    },
-    {
-      "epoch": 2.8982548693541297,
-      "grad_norm": 0.0012806823942810297,
-      "learning_rate": 0.00019999585703374092,
-      "loss": 46.0,
-      "step": 37907
-    },
-    {
-      "epoch": 2.8983313263375194,
-      "grad_norm": 0.0031860142480582,
-      "learning_rate": 0.00019999585681509604,
-      "loss": 46.0,
-      "step": 37908
-    },
-    {
-      "epoch": 2.898407783320909,
-      "grad_norm": 0.002116659888997674,
-      "learning_rate": 0.00019999585659644543,
-      "loss": 46.0,
-      "step": 37909
-    },
-    {
-      "epoch": 2.898484240304299,
-      "grad_norm": 0.004699204117059708,
-      "learning_rate": 0.000199995856377789,
-      "loss": 46.0,
-      "step": 37910
-    },
-    {
-      "epoch": 2.8985606972876887,
-      "grad_norm": 0.005540462210774422,
-      "learning_rate": 0.00019999585615912684,
-      "loss": 46.0,
-      "step": 37911
-    },
-    {
-      "epoch": 2.8986371542710785,
-      "grad_norm": 0.003695718478411436,
-      "learning_rate": 0.00019999585594045888,
-      "loss": 46.0,
-      "step": 37912
-    },
-    {
-      "epoch": 2.898713611254468,
-      "grad_norm": 0.0018116777064278722,
-      "learning_rate": 0.0001999958557217852,
-      "loss": 46.0,
-      "step": 37913
-    },
-    {
-      "epoch": 2.8987900682378576,
-      "grad_norm": 0.001851273002102971,
-      "learning_rate": 0.00019999585550310568,
-      "loss": 46.0,
-      "step": 37914
-    },
-    {
-      "epoch": 2.8988665252212473,
-      "grad_norm": 0.0021140240132808685,
-      "learning_rate": 0.00019999585528442042,
-      "loss": 46.0,
-      "step": 37915
-    },
-    {
-      "epoch": 2.898942982204637,
-      "grad_norm": 0.002079523866996169,
-      "learning_rate": 0.0001999958550657294,
-      "loss": 46.0,
-      "step": 37916
-    },
-    {
-      "epoch": 2.899019439188027,
-      "grad_norm": 0.0014937804080545902,
-      "learning_rate": 0.0001999958548470326,
-      "loss": 46.0,
-      "step": 37917
-    },
-    {
-      "epoch": 2.8990958961714166,
-      "grad_norm": 0.0020791683346033096,
-      "learning_rate": 0.00019999585462833006,
-      "loss": 46.0,
-      "step": 37918
-    },
-    {
-      "epoch": 2.8991723531548064,
-      "grad_norm": 0.0009234077297151089,
-      "learning_rate": 0.00019999585440962173,
-      "loss": 46.0,
-      "step": 37919
-    },
-    {
-      "epoch": 2.899248810138196,
-      "grad_norm": 0.002561927540227771,
-      "learning_rate": 0.00019999585419090763,
-      "loss": 46.0,
-      "step": 37920
-    },
-    {
-      "epoch": 2.8993252671215854,
-      "grad_norm": 0.0007290031062439084,
-      "learning_rate": 0.00019999585397218774,
-      "loss": 46.0,
-      "step": 37921
-    },
-    {
-      "epoch": 2.899401724104975,
-      "grad_norm": 0.002084975130856037,
-      "learning_rate": 0.00019999585375346212,
-      "loss": 46.0,
-      "step": 37922
-    },
-    {
-      "epoch": 2.899478181088365,
-      "grad_norm": 0.001207304303534329,
-      "learning_rate": 0.00019999585353473073,
-      "loss": 46.0,
-      "step": 37923
-    },
-    {
-      "epoch": 2.8995546380717547,
-      "grad_norm": 0.0008326134993694723,
-      "learning_rate": 0.00019999585331599353,
-      "loss": 46.0,
-      "step": 37924
-    },
-    {
-      "epoch": 2.8996310950551445,
-      "grad_norm": 0.0008646500064060092,
-      "learning_rate": 0.0001999958530972506,
-      "loss": 46.0,
-      "step": 37925
-    },
-    {
-      "epoch": 2.8997075520385343,
-      "grad_norm": 0.004952006042003632,
-      "learning_rate": 0.00019999585287850188,
-      "loss": 46.0,
-      "step": 37926
-    },
-    {
-      "epoch": 2.899784009021924,
-      "grad_norm": 0.0017778363544493914,
-      "learning_rate": 0.00019999585265974737,
-      "loss": 46.0,
-      "step": 37927
-    },
-    {
-      "epoch": 2.8998604660053138,
-      "grad_norm": 0.004038863815367222,
-      "learning_rate": 0.00019999585244098714,
-      "loss": 46.0,
-      "step": 37928
-    },
-    {
-      "epoch": 2.8999369229887035,
-      "grad_norm": 0.007513077463954687,
-      "learning_rate": 0.00019999585222222114,
-      "loss": 46.0,
-      "step": 37929
-    },
-    {
-      "epoch": 2.9000133799720933,
-      "grad_norm": 0.001991919009014964,
-      "learning_rate": 0.00019999585200344933,
-      "loss": 46.0,
-      "step": 37930
-    },
-    {
-      "epoch": 2.900089836955483,
-      "grad_norm": 0.00294917868450284,
-      "learning_rate": 0.00019999585178467175,
-      "loss": 46.0,
-      "step": 37931
-    },
-    {
-      "epoch": 2.900166293938873,
-      "grad_norm": 0.0005399100482463837,
-      "learning_rate": 0.00019999585156588843,
-      "loss": 46.0,
-      "step": 37932
-    },
-    {
-      "epoch": 2.9002427509222626,
-      "grad_norm": 0.0032314506825059652,
-      "learning_rate": 0.00019999585134709933,
-      "loss": 46.0,
-      "step": 37933
-    },
-    {
-      "epoch": 2.9003192079056523,
-      "grad_norm": 0.0019096112810075283,
-      "learning_rate": 0.00019999585112830446,
-      "loss": 46.0,
-      "step": 37934
-    },
-    {
-      "epoch": 2.9003956648890417,
-      "grad_norm": 0.0043441541492938995,
-      "learning_rate": 0.0001999958509095038,
-      "loss": 46.0,
-      "step": 37935
-    },
-    {
-      "epoch": 2.9004721218724314,
-      "grad_norm": 0.0035393706057220697,
-      "learning_rate": 0.00019999585069069742,
-      "loss": 46.0,
-      "step": 37936
-    },
-    {
-      "epoch": 2.900548578855821,
-      "grad_norm": 0.001900298404507339,
-      "learning_rate": 0.00019999585047188526,
-      "loss": 46.0,
-      "step": 37937
-    },
-    {
-      "epoch": 2.900625035839211,
-      "grad_norm": 0.002038979437202215,
-      "learning_rate": 0.0001999958502530673,
-      "loss": 46.0,
-      "step": 37938
-    },
-    {
-      "epoch": 2.9007014928226007,
-      "grad_norm": 0.0027250300627201796,
-      "learning_rate": 0.00019999585003424358,
-      "loss": 46.0,
-      "step": 37939
-    },
-    {
-      "epoch": 2.9007779498059905,
-      "grad_norm": 0.0017409016145393252,
-      "learning_rate": 0.0001999958498154141,
-      "loss": 46.0,
-      "step": 37940
-    },
-    {
-      "epoch": 2.9008544067893802,
-      "grad_norm": 0.004930721130222082,
-      "learning_rate": 0.00019999584959657884,
-      "loss": 46.0,
-      "step": 37941
-    },
-    {
-      "epoch": 2.90093086377277,
-      "grad_norm": 0.0018574399873614311,
-      "learning_rate": 0.0001999958493777378,
-      "loss": 46.0,
-      "step": 37942
-    },
-    {
-      "epoch": 2.9010073207561593,
-      "grad_norm": 0.005130739416927099,
-      "learning_rate": 0.00019999584915889103,
-      "loss": 46.0,
-      "step": 37943
-    },
-    {
-      "epoch": 2.901083777739549,
-      "grad_norm": 0.000941133766900748,
-      "learning_rate": 0.00019999584894003848,
-      "loss": 46.0,
-      "step": 37944
-    },
-    {
-      "epoch": 2.901160234722939,
-      "grad_norm": 0.0024534950498491526,
-      "learning_rate": 0.0001999958487211801,
-      "loss": 46.0,
-      "step": 37945
-    },
-    {
-      "epoch": 2.9012366917063286,
-      "grad_norm": 0.0018703043460845947,
-      "learning_rate": 0.00019999584850231603,
-      "loss": 46.0,
-      "step": 37946
-    },
-    {
-      "epoch": 2.9013131486897183,
-      "grad_norm": 0.0015180655755102634,
-      "learning_rate": 0.00019999584828344618,
-      "loss": 46.0,
-      "step": 37947
-    },
-    {
-      "epoch": 2.901389605673108,
-      "grad_norm": 0.002182352589443326,
-      "learning_rate": 0.0001999958480645705,
-      "loss": 46.0,
-      "step": 37948
-    },
-    {
-      "epoch": 2.901466062656498,
-      "grad_norm": 0.0022451074328273535,
-      "learning_rate": 0.0001999958478456891,
-      "loss": 46.0,
-      "step": 37949
-    },
-    {
-      "epoch": 2.9015425196398876,
-      "grad_norm": 0.0007296993862837553,
-      "learning_rate": 0.00019999584762680193,
-      "loss": 46.0,
-      "step": 37950
-    },
-    {
-      "epoch": 2.9016189766232774,
-      "grad_norm": 0.008540484122931957,
-      "learning_rate": 0.00019999584740790897,
-      "loss": 46.0,
-      "step": 37951
-    },
-    {
-      "epoch": 2.901695433606667,
-      "grad_norm": 0.001649975311011076,
-      "learning_rate": 0.00019999584718901028,
-      "loss": 46.0,
-      "step": 37952
-    },
-    {
-      "epoch": 2.901771890590057,
-      "grad_norm": 0.00655308086425066,
-      "learning_rate": 0.0001999958469701058,
-      "loss": 46.0,
-      "step": 37953
-    },
-    {
-      "epoch": 2.9018483475734467,
-      "grad_norm": 0.0014261931646615267,
-      "learning_rate": 0.00019999584675119554,
-      "loss": 46.0,
-      "step": 37954
-    },
-    {
-      "epoch": 2.9019248045568364,
-      "grad_norm": 0.0014747374225407839,
-      "learning_rate": 0.0001999958465322795,
-      "loss": 46.0,
-      "step": 37955
-    },
-    {
-      "epoch": 2.902001261540226,
-      "grad_norm": 0.0017352985450997949,
-      "learning_rate": 0.00019999584631335773,
-      "loss": 46.0,
-      "step": 37956
-    },
-    {
-      "epoch": 2.9020777185236155,
-      "grad_norm": 0.003741080639883876,
-      "learning_rate": 0.00019999584609443015,
-      "loss": 46.0,
-      "step": 37957
-    },
-    {
-      "epoch": 2.9021541755070053,
-      "grad_norm": 0.001909217913635075,
-      "learning_rate": 0.00019999584587549683,
-      "loss": 46.0,
-      "step": 37958
-    },
-    {
-      "epoch": 2.902230632490395,
-      "grad_norm": 0.0037101220805197954,
-      "learning_rate": 0.00019999584565655773,
-      "loss": 46.0,
-      "step": 37959
-    },
-    {
-      "epoch": 2.902307089473785,
-      "grad_norm": 0.002531482372432947,
-      "learning_rate": 0.00019999584543761286,
-      "loss": 46.0,
-      "step": 37960
-    },
-    {
-      "epoch": 2.9023835464571746,
-      "grad_norm": 0.002353631891310215,
-      "learning_rate": 0.00019999584521866224,
-      "loss": 46.0,
-      "step": 37961
-    },
-    {
-      "epoch": 2.9024600034405643,
-      "grad_norm": 0.0014853293541818857,
-      "learning_rate": 0.00019999584499970583,
-      "loss": 46.0,
-      "step": 37962
-    },
-    {
-      "epoch": 2.902536460423954,
-      "grad_norm": 0.0020416327752172947,
-      "learning_rate": 0.00019999584478074363,
-      "loss": 46.0,
-      "step": 37963
-    },
-    {
-      "epoch": 2.902612917407344,
-      "grad_norm": 0.000849250121973455,
-      "learning_rate": 0.0001999958445617757,
-      "loss": 46.0,
-      "step": 37964
-    },
-    {
-      "epoch": 2.902689374390733,
-      "grad_norm": 0.0035294487606734037,
-      "learning_rate": 0.000199995844342802,
-      "loss": 46.0,
-      "step": 37965
-    },
-    {
-      "epoch": 2.902765831374123,
-      "grad_norm": 0.004840458743274212,
-      "learning_rate": 0.0001999958441238225,
-      "loss": 46.0,
-      "step": 37966
-    },
-    {
-      "epoch": 2.9028422883575127,
-      "grad_norm": 0.0017963387072086334,
-      "learning_rate": 0.00019999584390483725,
-      "loss": 46.0,
-      "step": 37967
-    },
-    {
-      "epoch": 2.9029187453409024,
-      "grad_norm": 0.0016213165363296866,
-      "learning_rate": 0.00019999584368584624,
-      "loss": 46.0,
-      "step": 37968
-    },
-    {
-      "epoch": 2.902995202324292,
-      "grad_norm": 0.002751770429313183,
-      "learning_rate": 0.00019999584346684944,
-      "loss": 46.0,
-      "step": 37969
-    },
-    {
-      "epoch": 2.903071659307682,
-      "grad_norm": 0.0012061558663845062,
-      "learning_rate": 0.00019999584324784686,
-      "loss": 46.0,
-      "step": 37970
-    },
-    {
-      "epoch": 2.9031481162910717,
-      "grad_norm": 0.0018004565499722958,
-      "learning_rate": 0.00019999584302883854,
-      "loss": 46.0,
-      "step": 37971
-    },
-    {
-      "epoch": 2.9032245732744615,
-      "grad_norm": 0.0009978287853300571,
-      "learning_rate": 0.00019999584280982444,
-      "loss": 46.0,
-      "step": 37972
-    },
-    {
-      "epoch": 2.9033010302578512,
-      "grad_norm": 0.0011133557418361306,
-      "learning_rate": 0.00019999584259080457,
-      "loss": 46.0,
-      "step": 37973
-    },
-    {
-      "epoch": 2.903377487241241,
-      "grad_norm": 0.002503123367205262,
-      "learning_rate": 0.00019999584237177892,
-      "loss": 46.0,
-      "step": 37974
-    },
-    {
-      "epoch": 2.9034539442246308,
-      "grad_norm": 0.0025112552102655172,
-      "learning_rate": 0.0001999958421527475,
-      "loss": 46.0,
-      "step": 37975
-    },
-    {
-      "epoch": 2.9035304012080205,
-      "grad_norm": 0.0020459210500121117,
-      "learning_rate": 0.00019999584193371034,
-      "loss": 46.0,
-      "step": 37976
-    },
-    {
-      "epoch": 2.9036068581914103,
-      "grad_norm": 0.0020152805373072624,
-      "learning_rate": 0.0001999958417146674,
-      "loss": 46.0,
-      "step": 37977
-    },
-    {
-      "epoch": 2.9036833151748,
-      "grad_norm": 0.004488846752792597,
-      "learning_rate": 0.0001999958414956187,
-      "loss": 46.0,
-      "step": 37978
-    },
-    {
-      "epoch": 2.9037597721581894,
-      "grad_norm": 0.001990138553082943,
-      "learning_rate": 0.0001999958412765642,
-      "loss": 46.0,
-      "step": 37979
-    },
-    {
-      "epoch": 2.903836229141579,
-      "grad_norm": 0.0007626558071933687,
-      "learning_rate": 0.00019999584105750395,
-      "loss": 46.0,
-      "step": 37980
-    },
-    {
-      "epoch": 2.903912686124969,
-      "grad_norm": 0.0017354001756757498,
-      "learning_rate": 0.00019999584083843792,
-      "loss": 46.0,
-      "step": 37981
-    },
-    {
-      "epoch": 2.9039891431083587,
-      "grad_norm": 0.007851804606616497,
-      "learning_rate": 0.00019999584061936615,
-      "loss": 46.0,
-      "step": 37982
-    },
-    {
-      "epoch": 2.9040656000917484,
-      "grad_norm": 0.0033887727186083794,
-      "learning_rate": 0.00019999584040028857,
-      "loss": 46.0,
-      "step": 37983
-    },
-    {
-      "epoch": 2.904142057075138,
-      "grad_norm": 0.0023646436166018248,
-      "learning_rate": 0.00019999584018120525,
-      "loss": 46.0,
-      "step": 37984
-    },
-    {
-      "epoch": 2.904218514058528,
-      "grad_norm": 0.0046932329423725605,
-      "learning_rate": 0.00019999583996211615,
-      "loss": 46.0,
-      "step": 37985
-    },
-    {
-      "epoch": 2.9042949710419177,
-      "grad_norm": 0.001397062442265451,
-      "learning_rate": 0.00019999583974302128,
-      "loss": 46.0,
-      "step": 37986
-    },
-    {
-      "epoch": 2.904371428025307,
-      "grad_norm": 0.0029896232299506664,
-      "learning_rate": 0.00019999583952392064,
-      "loss": 46.0,
-      "step": 37987
-    },
-    {
-      "epoch": 2.9044478850086968,
-      "grad_norm": 0.0011912498157471418,
-      "learning_rate": 0.00019999583930481425,
-      "loss": 46.0,
-      "step": 37988
-    },
-    {
-      "epoch": 2.9045243419920865,
-      "grad_norm": 0.0009944017510861158,
-      "learning_rate": 0.00019999583908570206,
-      "loss": 46.0,
-      "step": 37989
-    },
-    {
-      "epoch": 2.9046007989754763,
-      "grad_norm": 0.0019806744530797005,
-      "learning_rate": 0.00019999583886658412,
-      "loss": 46.0,
-      "step": 37990
-    },
-    {
-      "epoch": 2.904677255958866,
-      "grad_norm": 0.0030906402971595526,
-      "learning_rate": 0.00019999583864746038,
-      "loss": 46.0,
-      "step": 37991
-    },
-    {
-      "epoch": 2.904753712942256,
-      "grad_norm": 0.003046264173462987,
-      "learning_rate": 0.00019999583842833093,
-      "loss": 46.0,
-      "step": 37992
-    },
-    {
-      "epoch": 2.9048301699256456,
-      "grad_norm": 0.0032885682303458452,
-      "learning_rate": 0.00019999583820919567,
-      "loss": 46.0,
-      "step": 37993
-    },
-    {
-      "epoch": 2.9049066269090353,
-      "grad_norm": 0.001067915465682745,
-      "learning_rate": 0.00019999583799005467,
-      "loss": 46.0,
-      "step": 37994
-    },
-    {
-      "epoch": 2.904983083892425,
-      "grad_norm": 0.0021763306576758623,
-      "learning_rate": 0.00019999583777090786,
-      "loss": 46.0,
-      "step": 37995
-    },
-    {
-      "epoch": 2.905059540875815,
-      "grad_norm": 0.0007947355625219643,
-      "learning_rate": 0.0001999958375517553,
-      "loss": 46.0,
-      "step": 37996
-    },
-    {
-      "epoch": 2.9051359978592046,
-      "grad_norm": 0.0019696077797561884,
-      "learning_rate": 0.00019999583733259696,
-      "loss": 46.0,
-      "step": 37997
-    },
-    {
-      "epoch": 2.9052124548425944,
-      "grad_norm": 0.0031082078348845243,
-      "learning_rate": 0.00019999583711343286,
-      "loss": 46.0,
-      "step": 37998
-    },
-    {
-      "epoch": 2.905288911825984,
-      "grad_norm": 0.000810633588116616,
-      "learning_rate": 0.000199995836894263,
-      "loss": 46.0,
-      "step": 37999
-    },
-    {
-      "epoch": 2.9053653688093735,
-      "grad_norm": 0.012532716616988182,
-      "learning_rate": 0.00019999583667508738,
-      "loss": 46.0,
-      "step": 38000
-    },
-    {
-      "epoch": 2.905441825792763,
-      "grad_norm": 0.0019997034687548876,
-      "learning_rate": 0.00019999583645590593,
-      "loss": 46.0,
-      "step": 38001
-    },
-    {
-      "epoch": 2.905518282776153,
-      "grad_norm": 0.0018338636728003621,
-      "learning_rate": 0.00019999583623671877,
-      "loss": 46.0,
-      "step": 38002
-    },
-    {
-      "epoch": 2.9055947397595427,
-      "grad_norm": 0.0016060309717431664,
-      "learning_rate": 0.00019999583601752584,
-      "loss": 46.0,
-      "step": 38003
-    },
-    {
-      "epoch": 2.9056711967429325,
-      "grad_norm": 0.0008651249227114022,
-      "learning_rate": 0.0001999958357983271,
-      "loss": 46.0,
-      "step": 38004
-    },
-    {
-      "epoch": 2.9057476537263223,
-      "grad_norm": 0.0024534137919545174,
-      "learning_rate": 0.00019999583557912262,
-      "loss": 46.0,
-      "step": 38005
-    },
-    {
-      "epoch": 2.905824110709712,
-      "grad_norm": 0.0030491920188069344,
-      "learning_rate": 0.00019999583535991239,
-      "loss": 46.0,
-      "step": 38006
-    },
-    {
-      "epoch": 2.905900567693102,
-      "grad_norm": 0.00151356915012002,
-      "learning_rate": 0.00019999583514069636,
-      "loss": 46.0,
-      "step": 38007
-    },
-    {
-      "epoch": 2.9059770246764915,
-      "grad_norm": 0.001505090855062008,
-      "learning_rate": 0.00019999583492147455,
-      "loss": 46.0,
-      "step": 38008
-    },
-    {
-      "epoch": 2.906053481659881,
-      "grad_norm": 0.001288287341594696,
-      "learning_rate": 0.000199995834702247,
-      "loss": 46.0,
-      "step": 38009
-    },
-    {
-      "epoch": 2.9061299386432706,
-      "grad_norm": 0.006425090599805117,
-      "learning_rate": 0.00019999583448301368,
-      "loss": 46.0,
-      "step": 38010
-    },
-    {
-      "epoch": 2.9062063956266604,
-      "grad_norm": 0.0036532345693558455,
-      "learning_rate": 0.00019999583426377458,
-      "loss": 46.0,
-      "step": 38011
-    },
-    {
-      "epoch": 2.90628285261005,
-      "grad_norm": 0.00118697015568614,
-      "learning_rate": 0.00019999583404452974,
-      "loss": 46.0,
-      "step": 38012
-    },
-    {
-      "epoch": 2.90635930959344,
-      "grad_norm": 0.0025789779610931873,
-      "learning_rate": 0.00019999583382527907,
-      "loss": 46.0,
-      "step": 38013
-    },
-    {
-      "epoch": 2.9064357665768297,
-      "grad_norm": 0.0036896681413054466,
-      "learning_rate": 0.00019999583360602268,
-      "loss": 46.0,
-      "step": 38014
-    },
-    {
-      "epoch": 2.9065122235602194,
-      "grad_norm": 0.00167533615604043,
-      "learning_rate": 0.0001999958333867605,
-      "loss": 46.0,
-      "step": 38015
-    },
-    {
-      "epoch": 2.906588680543609,
-      "grad_norm": 0.002036773134022951,
-      "learning_rate": 0.00019999583316749256,
-      "loss": 46.0,
-      "step": 38016
-    },
-    {
-      "epoch": 2.906665137526999,
-      "grad_norm": 0.0014422225067391992,
-      "learning_rate": 0.00019999583294821882,
-      "loss": 46.0,
-      "step": 38017
-    },
-    {
-      "epoch": 2.9067415945103887,
-      "grad_norm": 0.003712414065375924,
-      "learning_rate": 0.00019999583272893934,
-      "loss": 46.0,
-      "step": 38018
-    },
-    {
-      "epoch": 2.9068180514937785,
-      "grad_norm": 0.0014143196167424321,
-      "learning_rate": 0.0001999958325096541,
-      "loss": 46.0,
-      "step": 38019
-    },
-    {
-      "epoch": 2.9068945084771682,
-      "grad_norm": 0.0036908360198140144,
-      "learning_rate": 0.0001999958322903631,
-      "loss": 46.0,
-      "step": 38020
-    },
-    {
-      "epoch": 2.906970965460558,
-      "grad_norm": 0.001010382897220552,
-      "learning_rate": 0.0001999958320710663,
-      "loss": 46.0,
-      "step": 38021
-    },
-    {
-      "epoch": 2.9070474224439473,
-      "grad_norm": 0.001173604978248477,
-      "learning_rate": 0.00019999583185176373,
-      "loss": 46.0,
-      "step": 38022
-    },
-    {
-      "epoch": 2.907123879427337,
-      "grad_norm": 0.0017818254418671131,
-      "learning_rate": 0.0001999958316324554,
-      "loss": 46.0,
-      "step": 38023
-    },
-    {
-      "epoch": 2.907200336410727,
-      "grad_norm": 0.001810151501558721,
-      "learning_rate": 0.00019999583141314128,
-      "loss": 46.0,
-      "step": 38024
-    },
-    {
-      "epoch": 2.9072767933941166,
-      "grad_norm": 0.004480066243559122,
-      "learning_rate": 0.00019999583119382144,
-      "loss": 46.0,
-      "step": 38025
-    },
-    {
-      "epoch": 2.9073532503775064,
-      "grad_norm": 0.0014266769867390394,
-      "learning_rate": 0.00019999583097449582,
-      "loss": 46.0,
-      "step": 38026
-    },
-    {
-      "epoch": 2.907429707360896,
-      "grad_norm": 0.001249079592525959,
-      "learning_rate": 0.00019999583075516438,
-      "loss": 46.0,
-      "step": 38027
-    },
-    {
-      "epoch": 2.907506164344286,
-      "grad_norm": 0.00151823740452528,
-      "learning_rate": 0.0001999958305358272,
-      "loss": 46.0,
-      "step": 38028
-    },
-    {
-      "epoch": 2.9075826213276756,
-      "grad_norm": 0.0009738160879351199,
-      "learning_rate": 0.00019999583031648425,
-      "loss": 46.0,
-      "step": 38029
-    },
-    {
-      "epoch": 2.907659078311065,
-      "grad_norm": 0.0022975686006247997,
-      "learning_rate": 0.00019999583009713557,
-      "loss": 46.0,
-      "step": 38030
-    },
-    {
-      "epoch": 2.9077355352944547,
-      "grad_norm": 0.0008797014015726745,
-      "learning_rate": 0.00019999582987778106,
-      "loss": 46.0,
-      "step": 38031
-    },
-    {
-      "epoch": 2.9078119922778445,
-      "grad_norm": 0.0026479896623641253,
-      "learning_rate": 0.00019999582965842083,
-      "loss": 46.0,
-      "step": 38032
-    },
-    {
-      "epoch": 2.9078884492612342,
-      "grad_norm": 0.0012797890231013298,
-      "learning_rate": 0.0001999958294390548,
-      "loss": 46.0,
-      "step": 38033
-    },
-    {
-      "epoch": 2.907964906244624,
-      "grad_norm": 0.0012818325776606798,
-      "learning_rate": 0.000199995829219683,
-      "loss": 46.0,
-      "step": 38034
-    },
-    {
-      "epoch": 2.9080413632280138,
-      "grad_norm": 0.0016048140823841095,
-      "learning_rate": 0.00019999582900030545,
-      "loss": 46.0,
-      "step": 38035
-    },
-    {
-      "epoch": 2.9081178202114035,
-      "grad_norm": 0.003517944598570466,
-      "learning_rate": 0.00019999582878092213,
-      "loss": 46.0,
-      "step": 38036
-    },
-    {
-      "epoch": 2.9081942771947933,
-      "grad_norm": 0.0025899375323206186,
-      "learning_rate": 0.00019999582856153304,
-      "loss": 46.0,
-      "step": 38037
-    },
-    {
-      "epoch": 2.908270734178183,
-      "grad_norm": 0.0008070202893577516,
-      "learning_rate": 0.00019999582834213814,
-      "loss": 46.0,
-      "step": 38038
-    },
-    {
-      "epoch": 2.908347191161573,
-      "grad_norm": 0.0022034223657101393,
-      "learning_rate": 0.00019999582812273752,
-      "loss": 46.0,
-      "step": 38039
-    },
-    {
-      "epoch": 2.9084236481449626,
-      "grad_norm": 0.0010655556106939912,
-      "learning_rate": 0.0001999958279033311,
-      "loss": 46.0,
-      "step": 38040
-    },
-    {
-      "epoch": 2.9085001051283523,
-      "grad_norm": 0.001035425579175353,
-      "learning_rate": 0.00019999582768391895,
-      "loss": 46.0,
-      "step": 38041
-    },
-    {
-      "epoch": 2.908576562111742,
-      "grad_norm": 0.002119320211932063,
-      "learning_rate": 0.00019999582746450098,
-      "loss": 46.0,
-      "step": 38042
-    },
-    {
-      "epoch": 2.908653019095132,
-      "grad_norm": 0.002724515274167061,
-      "learning_rate": 0.00019999582724507727,
-      "loss": 46.0,
-      "step": 38043
-    },
-    {
-      "epoch": 2.908729476078521,
-      "grad_norm": 0.0031197164207696915,
-      "learning_rate": 0.00019999582702564782,
-      "loss": 46.0,
-      "step": 38044
-    },
-    {
-      "epoch": 2.908805933061911,
-      "grad_norm": 0.004094405099749565,
-      "learning_rate": 0.00019999582680621254,
-      "loss": 46.0,
-      "step": 38045
-    },
-    {
-      "epoch": 2.9088823900453007,
-      "grad_norm": 0.0030211550183594227,
-      "learning_rate": 0.00019999582658677154,
-      "loss": 46.0,
-      "step": 38046
-    },
-    {
-      "epoch": 2.9089588470286905,
-      "grad_norm": 0.0005786955589428544,
-      "learning_rate": 0.00019999582636732473,
-      "loss": 46.0,
-      "step": 38047
-    },
-    {
-      "epoch": 2.90903530401208,
-      "grad_norm": 0.0018357059452682734,
-      "learning_rate": 0.00019999582614787218,
-      "loss": 46.0,
-      "step": 38048
-    },
-    {
-      "epoch": 2.90911176099547,
-      "grad_norm": 0.0007620866526849568,
-      "learning_rate": 0.00019999582592841386,
-      "loss": 46.0,
-      "step": 38049
-    },
-    {
-      "epoch": 2.9091882179788597,
-      "grad_norm": 0.0023715051356703043,
-      "learning_rate": 0.00019999582570894974,
-      "loss": 46.0,
-      "step": 38050
-    },
-    {
-      "epoch": 2.9092646749622495,
-      "grad_norm": 0.0006669636932201684,
-      "learning_rate": 0.00019999582548947987,
-      "loss": 46.0,
-      "step": 38051
-    },
-    {
-      "epoch": 2.909341131945639,
-      "grad_norm": 0.001595982350409031,
-      "learning_rate": 0.00019999582527000426,
-      "loss": 46.0,
-      "step": 38052
-    },
-    {
-      "epoch": 2.9094175889290286,
-      "grad_norm": 0.0029678745195269585,
-      "learning_rate": 0.00019999582505052284,
-      "loss": 46.0,
-      "step": 38053
-    },
-    {
-      "epoch": 2.9094940459124183,
-      "grad_norm": 0.0037072370760142803,
-      "learning_rate": 0.00019999582483103568,
-      "loss": 46.0,
-      "step": 38054
-    },
-    {
-      "epoch": 2.909570502895808,
-      "grad_norm": 0.005485964938998222,
-      "learning_rate": 0.00019999582461154272,
-      "loss": 46.0,
-      "step": 38055
-    },
-    {
-      "epoch": 2.909646959879198,
-      "grad_norm": 0.002377541037276387,
-      "learning_rate": 0.000199995824392044,
-      "loss": 46.0,
-      "step": 38056
-    },
-    {
-      "epoch": 2.9097234168625876,
-      "grad_norm": 0.001125634997151792,
-      "learning_rate": 0.00019999582417253953,
-      "loss": 46.0,
-      "step": 38057
-    },
-    {
-      "epoch": 2.9097998738459774,
-      "grad_norm": 0.003063374664634466,
-      "learning_rate": 0.00019999582395302927,
-      "loss": 46.0,
-      "step": 38058
-    },
-    {
-      "epoch": 2.909876330829367,
-      "grad_norm": 0.0015070941299200058,
-      "learning_rate": 0.00019999582373351324,
-      "loss": 46.0,
-      "step": 38059
-    },
-    {
-      "epoch": 2.909952787812757,
-      "grad_norm": 0.0013611374888569117,
-      "learning_rate": 0.00019999582351399147,
-      "loss": 46.0,
-      "step": 38060
-    },
-    {
-      "epoch": 2.9100292447961467,
-      "grad_norm": 0.0016747881891205907,
-      "learning_rate": 0.00019999582329446392,
-      "loss": 46.0,
-      "step": 38061
-    },
-    {
-      "epoch": 2.9101057017795364,
-      "grad_norm": 0.004746443126350641,
-      "learning_rate": 0.0001999958230749306,
-      "loss": 46.0,
-      "step": 38062
-    },
-    {
-      "epoch": 2.910182158762926,
-      "grad_norm": 0.004720636177808046,
-      "learning_rate": 0.0001999958228553915,
-      "loss": 46.0,
-      "step": 38063
-    },
-    {
-      "epoch": 2.910258615746316,
-      "grad_norm": 0.0030097602866590023,
-      "learning_rate": 0.0001999958226358466,
-      "loss": 46.0,
-      "step": 38064
-    },
-    {
-      "epoch": 2.9103350727297057,
-      "grad_norm": 0.0029530173633247614,
-      "learning_rate": 0.00019999582241629597,
-      "loss": 46.0,
-      "step": 38065
-    },
-    {
-      "epoch": 2.910411529713095,
-      "grad_norm": 0.0013610816095024347,
-      "learning_rate": 0.00019999582219673955,
-      "loss": 46.0,
-      "step": 38066
-    },
-    {
-      "epoch": 2.910487986696485,
-      "grad_norm": 0.0011563777225092053,
-      "learning_rate": 0.0001999958219771774,
-      "loss": 46.0,
-      "step": 38067
-    },
-    {
-      "epoch": 2.9105644436798745,
-      "grad_norm": 0.0008149848436005414,
-      "learning_rate": 0.00019999582175760945,
-      "loss": 46.0,
-      "step": 38068
-    },
-    {
-      "epoch": 2.9106409006632643,
-      "grad_norm": 0.00287242466583848,
-      "learning_rate": 0.00019999582153803575,
-      "loss": 46.0,
-      "step": 38069
-    },
-    {
-      "epoch": 2.910717357646654,
-      "grad_norm": 0.0007763186004012823,
-      "learning_rate": 0.00019999582131845624,
-      "loss": 46.0,
-      "step": 38070
-    },
-    {
-      "epoch": 2.910793814630044,
-      "grad_norm": 0.0009206904796883464,
-      "learning_rate": 0.000199995821098871,
-      "loss": 46.0,
-      "step": 38071
-    },
-    {
-      "epoch": 2.9108702716134336,
-      "grad_norm": 0.0020329696126282215,
-      "learning_rate": 0.00019999582087927998,
-      "loss": 46.0,
-      "step": 38072
-    },
-    {
-      "epoch": 2.9109467285968234,
-      "grad_norm": 0.0007907075341790915,
-      "learning_rate": 0.00019999582065968318,
-      "loss": 46.0,
-      "step": 38073
-    },
-    {
-      "epoch": 2.9110231855802127,
-      "grad_norm": 0.0014534153742715716,
-      "learning_rate": 0.00019999582044008063,
-      "loss": 46.0,
-      "step": 38074
-    },
-    {
-      "epoch": 2.9110996425636024,
-      "grad_norm": 0.0013247976312413812,
-      "learning_rate": 0.0001999958202204723,
-      "loss": 46.0,
-      "step": 38075
-    },
-    {
-      "epoch": 2.911176099546992,
-      "grad_norm": 0.0012230559950694442,
-      "learning_rate": 0.00019999582000085822,
-      "loss": 46.0,
-      "step": 38076
-    },
-    {
-      "epoch": 2.911252556530382,
-      "grad_norm": 0.0016545574180781841,
-      "learning_rate": 0.00019999581978123835,
-      "loss": 46.0,
-      "step": 38077
-    },
-    {
-      "epoch": 2.9113290135137717,
-      "grad_norm": 0.0034378371201455593,
-      "learning_rate": 0.0001999958195616127,
-      "loss": 46.0,
-      "step": 38078
-    },
-    {
-      "epoch": 2.9114054704971615,
-      "grad_norm": 0.002517619403079152,
-      "learning_rate": 0.00019999581934198132,
-      "loss": 46.0,
-      "step": 38079
-    },
-    {
-      "epoch": 2.9114819274805512,
-      "grad_norm": 0.0010219543473795056,
-      "learning_rate": 0.0001999958191223441,
-      "loss": 46.0,
-      "step": 38080
-    },
-    {
-      "epoch": 2.911558384463941,
-      "grad_norm": 0.004021560773253441,
-      "learning_rate": 0.00019999581890270117,
-      "loss": 46.0,
-      "step": 38081
-    },
-    {
-      "epoch": 2.9116348414473308,
-      "grad_norm": 0.0006769971805624664,
-      "learning_rate": 0.0001999958186830525,
-      "loss": 46.0,
-      "step": 38082
-    },
-    {
-      "epoch": 2.9117112984307205,
-      "grad_norm": 0.004257587715983391,
-      "learning_rate": 0.00019999581846339798,
-      "loss": 46.0,
-      "step": 38083
-    },
-    {
-      "epoch": 2.9117877554141103,
-      "grad_norm": 0.001606352161616087,
-      "learning_rate": 0.00019999581824373772,
-      "loss": 46.0,
-      "step": 38084
-    },
-    {
-      "epoch": 2.9118642123975,
-      "grad_norm": 0.0012529935920611024,
-      "learning_rate": 0.0001999958180240717,
-      "loss": 46.0,
-      "step": 38085
-    },
-    {
-      "epoch": 2.91194066938089,
-      "grad_norm": 0.0011989392805844545,
-      "learning_rate": 0.00019999581780439992,
-      "loss": 46.0,
-      "step": 38086
-    },
-    {
-      "epoch": 2.9120171263642796,
-      "grad_norm": 0.004830826073884964,
-      "learning_rate": 0.00019999581758472237,
-      "loss": 46.0,
-      "step": 38087
-    },
-    {
-      "epoch": 2.912093583347669,
-      "grad_norm": 0.003272679867222905,
-      "learning_rate": 0.00019999581736503905,
-      "loss": 46.0,
-      "step": 38088
-    },
-    {
-      "epoch": 2.9121700403310586,
-      "grad_norm": 0.010868401266634464,
-      "learning_rate": 0.00019999581714534993,
-      "loss": 46.0,
-      "step": 38089
-    },
-    {
-      "epoch": 2.9122464973144484,
-      "grad_norm": 0.0019055166048929095,
-      "learning_rate": 0.00019999581692565506,
-      "loss": 46.0,
-      "step": 38090
-    },
-    {
-      "epoch": 2.912322954297838,
-      "grad_norm": 0.0030960282310843468,
-      "learning_rate": 0.00019999581670595445,
-      "loss": 46.0,
-      "step": 38091
-    },
-    {
-      "epoch": 2.912399411281228,
-      "grad_norm": 0.0014686177019029856,
-      "learning_rate": 0.00019999581648624804,
-      "loss": 46.0,
-      "step": 38092
-    },
-    {
-      "epoch": 2.9124758682646177,
-      "grad_norm": 0.003918887116014957,
-      "learning_rate": 0.00019999581626653585,
-      "loss": 46.0,
-      "step": 38093
-    },
-    {
-      "epoch": 2.9125523252480074,
-      "grad_norm": 0.0009708001161925495,
-      "learning_rate": 0.00019999581604681791,
-      "loss": 46.0,
-      "step": 38094
-    },
-    {
-      "epoch": 2.912628782231397,
-      "grad_norm": 0.005467913579195738,
-      "learning_rate": 0.0001999958158270942,
-      "loss": 46.0,
-      "step": 38095
-    },
-    {
-      "epoch": 2.9127052392147865,
-      "grad_norm": 0.0030989013612270355,
-      "learning_rate": 0.00019999581560736473,
-      "loss": 46.0,
-      "step": 38096
-    },
-    {
-      "epoch": 2.9127816961981763,
-      "grad_norm": 0.0029152007773518562,
-      "learning_rate": 0.00019999581538762947,
-      "loss": 46.0,
-      "step": 38097
-    },
-    {
-      "epoch": 2.912858153181566,
-      "grad_norm": 0.0010716785909608006,
-      "learning_rate": 0.00019999581516788844,
-      "loss": 46.0,
-      "step": 38098
-    },
-    {
-      "epoch": 2.912934610164956,
-      "grad_norm": 0.0016080865170806646,
-      "learning_rate": 0.00019999581494814167,
-      "loss": 46.0,
-      "step": 38099
-    },
-    {
-      "epoch": 2.9130110671483456,
-      "grad_norm": 0.0011618132703006268,
-      "learning_rate": 0.0001999958147283891,
-      "loss": 46.0,
-      "step": 38100
-    },
-    {
-      "epoch": 2.9130875241317353,
-      "grad_norm": 0.0025585044641047716,
-      "learning_rate": 0.00019999581450863077,
-      "loss": 46.0,
-      "step": 38101
-    },
-    {
-      "epoch": 2.913163981115125,
-      "grad_norm": 0.0016671492485329509,
-      "learning_rate": 0.00019999581428886668,
-      "loss": 46.0,
-      "step": 38102
-    },
-    {
-      "epoch": 2.913240438098515,
-      "grad_norm": 0.0006749452440999448,
-      "learning_rate": 0.0001999958140690968,
-      "loss": 46.0,
-      "step": 38103
-    },
-    {
-      "epoch": 2.9133168950819046,
-      "grad_norm": 0.004145908635109663,
-      "learning_rate": 0.00019999581384932117,
-      "loss": 46.0,
-      "step": 38104
-    },
-    {
-      "epoch": 2.9133933520652944,
-      "grad_norm": 0.004115253686904907,
-      "learning_rate": 0.00019999581362953976,
-      "loss": 46.0,
-      "step": 38105
-    },
-    {
-      "epoch": 2.913469809048684,
-      "grad_norm": 0.004197151865810156,
-      "learning_rate": 0.0001999958134097526,
-      "loss": 46.0,
-      "step": 38106
-    },
-    {
-      "epoch": 2.913546266032074,
-      "grad_norm": 0.0025055333971977234,
-      "learning_rate": 0.00019999581318995966,
-      "loss": 46.0,
-      "step": 38107
-    },
-    {
-      "epoch": 2.9136227230154637,
-      "grad_norm": 0.0015900548314675689,
-      "learning_rate": 0.00019999581297016096,
-      "loss": 46.0,
-      "step": 38108
-    },
-    {
-      "epoch": 2.9136991799988534,
-      "grad_norm": 0.005039710085839033,
-      "learning_rate": 0.00019999581275035645,
-      "loss": 46.0,
-      "step": 38109
-    },
-    {
-      "epoch": 2.9137756369822427,
-      "grad_norm": 0.0023497866932302713,
-      "learning_rate": 0.0001999958125305462,
-      "loss": 46.0,
-      "step": 38110
-    },
-    {
-      "epoch": 2.9138520939656325,
-      "grad_norm": 0.0014665876515209675,
-      "learning_rate": 0.0001999958123107302,
-      "loss": 46.0,
-      "step": 38111
-    },
-    {
-      "epoch": 2.9139285509490223,
-      "grad_norm": 0.004661711398512125,
-      "learning_rate": 0.00019999581209090842,
-      "loss": 46.0,
-      "step": 38112
-    },
-    {
-      "epoch": 2.914005007932412,
-      "grad_norm": 0.0030364585109055042,
-      "learning_rate": 0.00019999581187108082,
-      "loss": 46.0,
-      "step": 38113
-    },
-    {
-      "epoch": 2.9140814649158018,
-      "grad_norm": 0.0005936098168604076,
-      "learning_rate": 0.00019999581165124753,
-      "loss": 46.0,
-      "step": 38114
-    },
-    {
-      "epoch": 2.9141579218991915,
-      "grad_norm": 0.0026004870887845755,
-      "learning_rate": 0.00019999581143140843,
-      "loss": 46.0,
-      "step": 38115
-    },
-    {
-      "epoch": 2.9142343788825813,
-      "grad_norm": 0.002274326514452696,
-      "learning_rate": 0.00019999581121156354,
-      "loss": 46.0,
-      "step": 38116
-    },
-    {
-      "epoch": 2.914310835865971,
-      "grad_norm": 0.001378798857331276,
-      "learning_rate": 0.00019999581099171292,
-      "loss": 46.0,
-      "step": 38117
-    },
-    {
-      "epoch": 2.9143872928493604,
-      "grad_norm": 0.0027540642768144608,
-      "learning_rate": 0.0001999958107718565,
-      "loss": 46.0,
-      "step": 38118
-    },
-    {
-      "epoch": 2.91446374983275,
-      "grad_norm": 0.0010525084799155593,
-      "learning_rate": 0.00019999581055199435,
-      "loss": 46.0,
-      "step": 38119
-    },
-    {
-      "epoch": 2.91454020681614,
-      "grad_norm": 0.001021408592350781,
-      "learning_rate": 0.0001999958103321264,
-      "loss": 46.0,
-      "step": 38120
-    },
-    {
-      "epoch": 2.9146166637995297,
-      "grad_norm": 0.0026804169174283743,
-      "learning_rate": 0.00019999581011225268,
-      "loss": 46.0,
-      "step": 38121
-    },
-    {
-      "epoch": 2.9146931207829194,
-      "grad_norm": 0.005769689101725817,
-      "learning_rate": 0.00019999580989237318,
-      "loss": 46.0,
-      "step": 38122
-    },
-    {
-      "epoch": 2.914769577766309,
-      "grad_norm": 0.0022064405493438244,
-      "learning_rate": 0.00019999580967248798,
-      "loss": 46.0,
-      "step": 38123
-    },
-    {
-      "epoch": 2.914846034749699,
-      "grad_norm": 0.0006371671333909035,
-      "learning_rate": 0.00019999580945259692,
-      "loss": 46.0,
-      "step": 38124
-    },
-    {
-      "epoch": 2.9149224917330887,
-      "grad_norm": 0.003425972070544958,
-      "learning_rate": 0.00019999580923270015,
-      "loss": 46.0,
-      "step": 38125
-    },
-    {
-      "epoch": 2.9149989487164785,
-      "grad_norm": 0.0011123217409476638,
-      "learning_rate": 0.00019999580901279758,
-      "loss": 46.0,
-      "step": 38126
-    },
-    {
-      "epoch": 2.9150754056998682,
-      "grad_norm": 0.0024983014445751905,
-      "learning_rate": 0.00019999580879288926,
-      "loss": 46.0,
-      "step": 38127
-    },
-    {
-      "epoch": 2.915151862683258,
-      "grad_norm": 0.00330982799641788,
-      "learning_rate": 0.00019999580857297516,
-      "loss": 46.0,
-      "step": 38128
-    },
-    {
-      "epoch": 2.9152283196666477,
-      "grad_norm": 0.004173245280981064,
-      "learning_rate": 0.0001999958083530553,
-      "loss": 46.0,
-      "step": 38129
-    },
-    {
-      "epoch": 2.9153047766500375,
-      "grad_norm": 0.0008257174049504101,
-      "learning_rate": 0.00019999580813312965,
-      "loss": 46.0,
-      "step": 38130
-    },
-    {
-      "epoch": 2.915381233633427,
-      "grad_norm": 0.0032756465952843428,
-      "learning_rate": 0.00019999580791319824,
-      "loss": 46.0,
-      "step": 38131
-    },
-    {
-      "epoch": 2.9154576906168166,
-      "grad_norm": 0.003746794769540429,
-      "learning_rate": 0.00019999580769326108,
-      "loss": 46.0,
-      "step": 38132
-    },
-    {
-      "epoch": 2.9155341476002063,
-      "grad_norm": 0.00330060045234859,
-      "learning_rate": 0.00019999580747331815,
-      "loss": 46.0,
-      "step": 38133
-    },
-    {
-      "epoch": 2.915610604583596,
-      "grad_norm": 0.0029367574024945498,
-      "learning_rate": 0.00019999580725336941,
-      "loss": 46.0,
-      "step": 38134
-    },
-    {
-      "epoch": 2.915687061566986,
-      "grad_norm": 0.0032441681250929832,
-      "learning_rate": 0.00019999580703341496,
-      "loss": 46.0,
-      "step": 38135
-    },
-    {
-      "epoch": 2.9157635185503756,
-      "grad_norm": 0.0023535985965281725,
-      "learning_rate": 0.00019999580681345468,
-      "loss": 46.0,
-      "step": 38136
-    },
-    {
-      "epoch": 2.9158399755337654,
-      "grad_norm": 0.0027953076642006636,
-      "learning_rate": 0.00019999580659348868,
-      "loss": 46.0,
-      "step": 38137
-    },
-    {
-      "epoch": 2.915916432517155,
-      "grad_norm": 0.0006844722665846348,
-      "learning_rate": 0.00019999580637351688,
-      "loss": 46.0,
-      "step": 38138
-    },
-    {
-      "epoch": 2.915992889500545,
-      "grad_norm": 0.008718026801943779,
-      "learning_rate": 0.00019999580615353933,
-      "loss": 46.0,
-      "step": 38139
-    },
-    {
-      "epoch": 2.9160693464839342,
-      "grad_norm": 0.001414721948094666,
-      "learning_rate": 0.00019999580593355602,
-      "loss": 46.0,
-      "step": 38140
-    },
-    {
-      "epoch": 2.916145803467324,
-      "grad_norm": 0.0025321547873318195,
-      "learning_rate": 0.0001999958057135669,
-      "loss": 46.0,
-      "step": 38141
-    },
-    {
-      "epoch": 2.9162222604507138,
-      "grad_norm": 0.0033333799801766872,
-      "learning_rate": 0.00019999580549357203,
-      "loss": 46.0,
-      "step": 38142
-    },
-    {
-      "epoch": 2.9162987174341035,
-      "grad_norm": 0.006616013124585152,
-      "learning_rate": 0.0001999958052735714,
-      "loss": 46.0,
-      "step": 38143
-    },
-    {
-      "epoch": 2.9163751744174933,
-      "grad_norm": 0.00655062822625041,
-      "learning_rate": 0.00019999580505356498,
-      "loss": 46.0,
-      "step": 38144
-    },
-    {
-      "epoch": 2.916451631400883,
-      "grad_norm": 0.0021653359290212393,
-      "learning_rate": 0.00019999580483355282,
-      "loss": 46.0,
-      "step": 38145
-    },
-    {
-      "epoch": 2.916528088384273,
-      "grad_norm": 0.0022567014675587416,
-      "learning_rate": 0.00019999580461353488,
-      "loss": 46.0,
-      "step": 38146
-    },
-    {
-      "epoch": 2.9166045453676626,
-      "grad_norm": 0.001058167777955532,
-      "learning_rate": 0.00019999580439351118,
-      "loss": 46.0,
-      "step": 38147
-    },
-    {
-      "epoch": 2.9166810023510523,
-      "grad_norm": 0.008150795474648476,
-      "learning_rate": 0.0001999958041734817,
-      "loss": 46.0,
-      "step": 38148
-    },
-    {
-      "epoch": 2.916757459334442,
-      "grad_norm": 0.0031658122316002846,
-      "learning_rate": 0.00019999580395344647,
-      "loss": 46.0,
-      "step": 38149
-    },
-    {
-      "epoch": 2.916833916317832,
-      "grad_norm": 0.0015234394231811166,
-      "learning_rate": 0.00019999580373340542,
-      "loss": 46.0,
-      "step": 38150
-    },
-    {
-      "epoch": 2.9169103733012216,
-      "grad_norm": 0.0029059748630970716,
-      "learning_rate": 0.00019999580351335862,
-      "loss": 46.0,
-      "step": 38151
-    },
-    {
-      "epoch": 2.9169868302846114,
-      "grad_norm": 0.004571816883981228,
-      "learning_rate": 0.00019999580329330607,
-      "loss": 46.0,
-      "step": 38152
-    },
-    {
-      "epoch": 2.9170632872680007,
-      "grad_norm": 0.003188816364854574,
-      "learning_rate": 0.00019999580307324775,
-      "loss": 46.0,
-      "step": 38153
-    },
-    {
-      "epoch": 2.9171397442513904,
-      "grad_norm": 0.001972890691831708,
-      "learning_rate": 0.00019999580285318363,
-      "loss": 46.0,
-      "step": 38154
-    },
-    {
-      "epoch": 2.91721620123478,
-      "grad_norm": 0.0009233566815964878,
-      "learning_rate": 0.0001999958026331138,
-      "loss": 46.0,
-      "step": 38155
-    },
-    {
-      "epoch": 2.91729265821817,
-      "grad_norm": 0.003466230584308505,
-      "learning_rate": 0.00019999580241303818,
-      "loss": 46.0,
-      "step": 38156
-    },
-    {
-      "epoch": 2.9173691152015597,
-      "grad_norm": 0.003325595986098051,
-      "learning_rate": 0.00019999580219295674,
-      "loss": 46.0,
-      "step": 38157
-    },
-    {
-      "epoch": 2.9174455721849495,
-      "grad_norm": 0.002221320290118456,
-      "learning_rate": 0.00019999580197286958,
-      "loss": 46.0,
-      "step": 38158
-    },
-    {
-      "epoch": 2.9175220291683392,
-      "grad_norm": 0.0011871540918946266,
-      "learning_rate": 0.00019999580175277662,
-      "loss": 46.0,
-      "step": 38159
-    },
-    {
-      "epoch": 2.917598486151729,
-      "grad_norm": 0.0027356965001672506,
-      "learning_rate": 0.00019999580153267794,
-      "loss": 46.0,
-      "step": 38160
-    },
-    {
-      "epoch": 2.9176749431351188,
-      "grad_norm": 0.0008755429880693555,
-      "learning_rate": 0.00019999580131257344,
-      "loss": 46.0,
-      "step": 38161
-    },
-    {
-      "epoch": 2.917751400118508,
-      "grad_norm": 0.001745652174577117,
-      "learning_rate": 0.00019999580109246318,
-      "loss": 46.0,
-      "step": 38162
-    },
-    {
-      "epoch": 2.917827857101898,
-      "grad_norm": 0.0013858848251402378,
-      "learning_rate": 0.00019999580087234718,
-      "loss": 46.0,
-      "step": 38163
-    },
-    {
-      "epoch": 2.9179043140852876,
-      "grad_norm": 0.0022346859332174063,
-      "learning_rate": 0.00019999580065222539,
-      "loss": 46.0,
-      "step": 38164
-    },
-    {
-      "epoch": 2.9179807710686774,
-      "grad_norm": 0.002344232052564621,
-      "learning_rate": 0.0001999958004320978,
-      "loss": 46.0,
-      "step": 38165
-    },
-    {
-      "epoch": 2.918057228052067,
-      "grad_norm": 0.0008668802911415696,
-      "learning_rate": 0.0001999958002119645,
-      "loss": 46.0,
-      "step": 38166
-    },
-    {
-      "epoch": 2.918133685035457,
-      "grad_norm": 0.0022169763687998056,
-      "learning_rate": 0.0001999957999918254,
-      "loss": 46.0,
-      "step": 38167
-    },
-    {
-      "epoch": 2.9182101420188467,
-      "grad_norm": 0.007605826016515493,
-      "learning_rate": 0.00019999579977168053,
-      "loss": 46.0,
-      "step": 38168
-    },
-    {
-      "epoch": 2.9182865990022364,
-      "grad_norm": 0.0050665633752942085,
-      "learning_rate": 0.0001999957995515299,
-      "loss": 46.0,
-      "step": 38169
-    },
-    {
-      "epoch": 2.918363055985626,
-      "grad_norm": 0.004774228669703007,
-      "learning_rate": 0.0001999957993313735,
-      "loss": 46.0,
-      "step": 38170
-    },
-    {
-      "epoch": 2.918439512969016,
-      "grad_norm": 0.0035356131847947836,
-      "learning_rate": 0.00019999579911121132,
-      "loss": 46.0,
-      "step": 38171
-    },
-    {
-      "epoch": 2.9185159699524057,
-      "grad_norm": 0.0021626397501677275,
-      "learning_rate": 0.0001999957988910434,
-      "loss": 46.0,
-      "step": 38172
-    },
-    {
-      "epoch": 2.9185924269357955,
-      "grad_norm": 0.004352611489593983,
-      "learning_rate": 0.00019999579867086969,
-      "loss": 46.0,
-      "step": 38173
-    },
-    {
-      "epoch": 2.918668883919185,
-      "grad_norm": 0.0021969517692923546,
-      "learning_rate": 0.0001999957984506902,
-      "loss": 46.0,
-      "step": 38174
-    },
-    {
-      "epoch": 2.9187453409025745,
-      "grad_norm": 0.004011734388768673,
-      "learning_rate": 0.00019999579823050495,
-      "loss": 46.0,
-      "step": 38175
-    },
-    {
-      "epoch": 2.9188217978859643,
-      "grad_norm": 0.004317046608775854,
-      "learning_rate": 0.00019999579801031393,
-      "loss": 46.0,
-      "step": 38176
-    },
-    {
-      "epoch": 2.918898254869354,
-      "grad_norm": 0.0008651851676404476,
-      "learning_rate": 0.00019999579779011716,
-      "loss": 46.0,
-      "step": 38177
-    },
-    {
-      "epoch": 2.918974711852744,
-      "grad_norm": 0.0007842287886887789,
-      "learning_rate": 0.00019999579756991458,
-      "loss": 46.0,
-      "step": 38178
-    },
-    {
-      "epoch": 2.9190511688361336,
-      "grad_norm": 0.002279200591146946,
-      "learning_rate": 0.00019999579734970627,
-      "loss": 46.0,
-      "step": 38179
-    },
-    {
-      "epoch": 2.9191276258195233,
-      "grad_norm": 0.0016729203052818775,
-      "learning_rate": 0.00019999579712949217,
-      "loss": 46.0,
-      "step": 38180
-    },
-    {
-      "epoch": 2.919204082802913,
-      "grad_norm": 0.0012921706074848771,
-      "learning_rate": 0.00019999579690927228,
-      "loss": 46.0,
-      "step": 38181
-    },
-    {
-      "epoch": 2.919280539786303,
-      "grad_norm": 0.0014163557207211852,
-      "learning_rate": 0.00019999579668904667,
-      "loss": 46.0,
-      "step": 38182
-    },
-    {
-      "epoch": 2.919356996769692,
-      "grad_norm": 0.003808164270594716,
-      "learning_rate": 0.00019999579646881526,
-      "loss": 46.0,
-      "step": 38183
-    },
-    {
-      "epoch": 2.919433453753082,
-      "grad_norm": 0.0013044095830991864,
-      "learning_rate": 0.0001999957962485781,
-      "loss": 46.0,
-      "step": 38184
-    },
-    {
-      "epoch": 2.9195099107364717,
-      "grad_norm": 0.002274476457387209,
-      "learning_rate": 0.00019999579602833514,
-      "loss": 46.0,
-      "step": 38185
-    },
-    {
-      "epoch": 2.9195863677198615,
-      "grad_norm": 0.004797384608536959,
-      "learning_rate": 0.00019999579580808643,
-      "loss": 46.0,
-      "step": 38186
-    },
-    {
-      "epoch": 2.9196628247032512,
-      "grad_norm": 0.0006925598718225956,
-      "learning_rate": 0.00019999579558783195,
-      "loss": 46.0,
-      "step": 38187
-    },
-    {
-      "epoch": 2.919739281686641,
-      "grad_norm": 0.0007656869711354375,
-      "learning_rate": 0.0001999957953675717,
-      "loss": 46.0,
-      "step": 38188
-    },
-    {
-      "epoch": 2.9198157386700307,
-      "grad_norm": 0.0015074973925948143,
-      "learning_rate": 0.00019999579514730568,
-      "loss": 46.0,
-      "step": 38189
-    },
-    {
-      "epoch": 2.9198921956534205,
-      "grad_norm": 0.0014720710460096598,
-      "learning_rate": 0.0001999957949270339,
-      "loss": 46.0,
-      "step": 38190
-    },
-    {
-      "epoch": 2.9199686526368103,
-      "grad_norm": 0.0015082076424732804,
-      "learning_rate": 0.00019999579470675633,
-      "loss": 46.0,
-      "step": 38191
-    },
-    {
-      "epoch": 2.9200451096202,
-      "grad_norm": 0.0023782411590218544,
-      "learning_rate": 0.00019999579448647304,
-      "loss": 46.0,
-      "step": 38192
-    },
-    {
-      "epoch": 2.92012156660359,
-      "grad_norm": 0.0019532095175236464,
-      "learning_rate": 0.00019999579426618392,
-      "loss": 46.0,
-      "step": 38193
-    },
-    {
-      "epoch": 2.9201980235869796,
-      "grad_norm": 0.005382715258747339,
-      "learning_rate": 0.00019999579404588906,
-      "loss": 46.0,
-      "step": 38194
-    },
-    {
-      "epoch": 2.9202744805703693,
-      "grad_norm": 0.003931119572371244,
-      "learning_rate": 0.00019999579382558842,
-      "loss": 46.0,
-      "step": 38195
-    },
-    {
-      "epoch": 2.920350937553759,
-      "grad_norm": 0.0032313985284417868,
-      "learning_rate": 0.00019999579360528203,
-      "loss": 46.0,
-      "step": 38196
-    },
-    {
-      "epoch": 2.9204273945371484,
-      "grad_norm": 0.0013780139852315187,
-      "learning_rate": 0.00019999579338496982,
-      "loss": 46.0,
-      "step": 38197
-    },
-    {
-      "epoch": 2.920503851520538,
-      "grad_norm": 0.001275938586331904,
-      "learning_rate": 0.0001999957931646519,
-      "loss": 46.0,
-      "step": 38198
-    },
-    {
-      "epoch": 2.920580308503928,
-      "grad_norm": 0.0020545879378914833,
-      "learning_rate": 0.00019999579294432818,
-      "loss": 46.0,
-      "step": 38199
-    },
-    {
-      "epoch": 2.9206567654873177,
-      "grad_norm": 0.002381442114710808,
-      "learning_rate": 0.00019999579272399873,
-      "loss": 46.0,
-      "step": 38200
-    },
-    {
-      "epoch": 2.9207332224707074,
-      "grad_norm": 0.001282895216718316,
-      "learning_rate": 0.00019999579250366348,
-      "loss": 46.0,
-      "step": 38201
-    },
-    {
-      "epoch": 2.920809679454097,
-      "grad_norm": 0.0022798131685703993,
-      "learning_rate": 0.00019999579228332243,
-      "loss": 46.0,
-      "step": 38202
-    },
-    {
-      "epoch": 2.920886136437487,
-      "grad_norm": 0.0041446988470852375,
-      "learning_rate": 0.00019999579206297569,
-      "loss": 46.0,
-      "step": 38203
-    },
-    {
-      "epoch": 2.9209625934208767,
-      "grad_norm": 0.0007441206835210323,
-      "learning_rate": 0.00019999579184262309,
-      "loss": 46.0,
-      "step": 38204
-    },
-    {
-      "epoch": 2.921039050404266,
-      "grad_norm": 0.0034983546938747168,
-      "learning_rate": 0.0001999957916222648,
-      "loss": 46.0,
-      "step": 38205
-    },
-    {
-      "epoch": 2.921115507387656,
-      "grad_norm": 0.0018707457929849625,
-      "learning_rate": 0.00019999579140190068,
-      "loss": 46.0,
-      "step": 38206
-    },
-    {
-      "epoch": 2.9211919643710456,
-      "grad_norm": 0.0012298153014853597,
-      "learning_rate": 0.0001999957911815308,
-      "loss": 46.0,
-      "step": 38207
-    },
-    {
-      "epoch": 2.9212684213544353,
-      "grad_norm": 0.0008012565667741001,
-      "learning_rate": 0.00019999579096115517,
-      "loss": 46.0,
-      "step": 38208
-    },
-    {
-      "epoch": 2.921344878337825,
-      "grad_norm": 0.0008982726722024381,
-      "learning_rate": 0.00019999579074077376,
-      "loss": 46.0,
-      "step": 38209
-    },
-    {
-      "epoch": 2.921421335321215,
-      "grad_norm": 0.004667261149734259,
-      "learning_rate": 0.0001999957905203866,
-      "loss": 46.0,
-      "step": 38210
-    },
-    {
-      "epoch": 2.9214977923046046,
-      "grad_norm": 0.001954728504642844,
-      "learning_rate": 0.00019999579029999367,
-      "loss": 46.0,
-      "step": 38211
-    },
-    {
-      "epoch": 2.9215742492879944,
-      "grad_norm": 0.0027920694556087255,
-      "learning_rate": 0.00019999579007959497,
-      "loss": 46.0,
-      "step": 38212
-    },
-    {
-      "epoch": 2.921650706271384,
-      "grad_norm": 0.006857927888631821,
-      "learning_rate": 0.00019999578985919046,
-      "loss": 46.0,
-      "step": 38213
-    },
-    {
-      "epoch": 2.921727163254774,
-      "grad_norm": 0.0008770758286118507,
-      "learning_rate": 0.00019999578963878024,
-      "loss": 46.0,
-      "step": 38214
-    },
-    {
-      "epoch": 2.9218036202381636,
-      "grad_norm": 0.003990574274212122,
-      "learning_rate": 0.0001999957894183642,
-      "loss": 46.0,
-      "step": 38215
-    },
-    {
-      "epoch": 2.9218800772215534,
-      "grad_norm": 0.0028714665677398443,
-      "learning_rate": 0.00019999578919794244,
-      "loss": 46.0,
-      "step": 38216
-    },
-    {
-      "epoch": 2.921956534204943,
-      "grad_norm": 0.0032724132761359215,
-      "learning_rate": 0.00019999578897751484,
-      "loss": 46.0,
-      "step": 38217
-    },
-    {
-      "epoch": 2.922032991188333,
-      "grad_norm": 0.000939900754019618,
-      "learning_rate": 0.00019999578875708153,
-      "loss": 46.0,
-      "step": 38218
-    },
-    {
-      "epoch": 2.9221094481717222,
-      "grad_norm": 0.0038132353220134974,
-      "learning_rate": 0.00019999578853664246,
-      "loss": 46.0,
-      "step": 38219
-    },
-    {
-      "epoch": 2.922185905155112,
-      "grad_norm": 0.0015964631456881762,
-      "learning_rate": 0.0001999957883161976,
-      "loss": 46.0,
-      "step": 38220
-    },
-    {
-      "epoch": 2.9222623621385018,
-      "grad_norm": 0.0014915929641574621,
-      "learning_rate": 0.00019999578809574696,
-      "loss": 46.0,
-      "step": 38221
-    },
-    {
-      "epoch": 2.9223388191218915,
-      "grad_norm": 0.0050577991642057896,
-      "learning_rate": 0.00019999578787529055,
-      "loss": 46.0,
-      "step": 38222
-    },
-    {
-      "epoch": 2.9224152761052813,
-      "grad_norm": 0.003150764387100935,
-      "learning_rate": 0.0001999957876548284,
-      "loss": 46.0,
-      "step": 38223
-    },
-    {
-      "epoch": 2.922491733088671,
-      "grad_norm": 0.002196155721321702,
-      "learning_rate": 0.00019999578743436046,
-      "loss": 46.0,
-      "step": 38224
-    },
-    {
-      "epoch": 2.922568190072061,
-      "grad_norm": 0.002711218548938632,
-      "learning_rate": 0.00019999578721388673,
-      "loss": 46.0,
-      "step": 38225
-    },
-    {
-      "epoch": 2.9226446470554506,
-      "grad_norm": 0.0012895666295662522,
-      "learning_rate": 0.00019999578699340728,
-      "loss": 46.0,
-      "step": 38226
-    },
-    {
-      "epoch": 2.92272110403884,
-      "grad_norm": 0.003579151351004839,
-      "learning_rate": 0.000199995786772922,
-      "loss": 46.0,
-      "step": 38227
-    },
-    {
-      "epoch": 2.9227975610222297,
-      "grad_norm": 0.0019901602063328028,
-      "learning_rate": 0.00019999578655243097,
-      "loss": 46.0,
-      "step": 38228
-    },
-    {
-      "epoch": 2.9228740180056194,
-      "grad_norm": 0.0013206868898123503,
-      "learning_rate": 0.0001999957863319342,
-      "loss": 46.0,
-      "step": 38229
-    },
-    {
-      "epoch": 2.922950474989009,
-      "grad_norm": 0.0025013729464262724,
-      "learning_rate": 0.00019999578611143163,
-      "loss": 46.0,
-      "step": 38230
-    },
-    {
-      "epoch": 2.923026931972399,
-      "grad_norm": 0.004589189775288105,
-      "learning_rate": 0.00019999578589092331,
-      "loss": 46.0,
-      "step": 38231
-    },
-    {
-      "epoch": 2.9231033889557887,
-      "grad_norm": 0.004183898214250803,
-      "learning_rate": 0.00019999578567040922,
-      "loss": 46.0,
-      "step": 38232
-    },
-    {
-      "epoch": 2.9231798459391785,
-      "grad_norm": 0.00264453305862844,
-      "learning_rate": 0.00019999578544988936,
-      "loss": 46.0,
-      "step": 38233
-    },
-    {
-      "epoch": 2.923256302922568,
-      "grad_norm": 0.0007998436340130866,
-      "learning_rate": 0.00019999578522936372,
-      "loss": 46.0,
-      "step": 38234
-    },
-    {
-      "epoch": 2.923332759905958,
-      "grad_norm": 0.0016832256224006414,
-      "learning_rate": 0.0001999957850088323,
-      "loss": 46.0,
-      "step": 38235
-    },
-    {
-      "epoch": 2.9234092168893477,
-      "grad_norm": 0.0012537117581814528,
-      "learning_rate": 0.00019999578478829515,
-      "loss": 46.0,
-      "step": 38236
-    },
-    {
-      "epoch": 2.9234856738727375,
-      "grad_norm": 0.0024934527464210987,
-      "learning_rate": 0.00019999578456775222,
-      "loss": 46.0,
-      "step": 38237
-    },
-    {
-      "epoch": 2.9235621308561273,
-      "grad_norm": 0.011158591136336327,
-      "learning_rate": 0.0001999957843472035,
-      "loss": 46.0,
-      "step": 38238
-    },
-    {
-      "epoch": 2.923638587839517,
-      "grad_norm": 0.0014899068046361208,
-      "learning_rate": 0.000199995784126649,
-      "loss": 46.0,
-      "step": 38239
-    },
-    {
-      "epoch": 2.923715044822907,
-      "grad_norm": 0.0007623178535141051,
-      "learning_rate": 0.00019999578390608876,
-      "loss": 46.0,
-      "step": 38240
-    },
-    {
-      "epoch": 2.923791501806296,
-      "grad_norm": 0.0024979696609079838,
-      "learning_rate": 0.00019999578368552277,
-      "loss": 46.0,
-      "step": 38241
-    },
-    {
-      "epoch": 2.923867958789686,
-      "grad_norm": 0.006785134319216013,
-      "learning_rate": 0.00019999578346495097,
-      "loss": 46.0,
-      "step": 38242
-    },
-    {
-      "epoch": 2.9239444157730756,
-      "grad_norm": 0.0008517408859916031,
-      "learning_rate": 0.00019999578324437342,
-      "loss": 46.0,
-      "step": 38243
-    },
-    {
-      "epoch": 2.9240208727564654,
-      "grad_norm": 0.0028591016307473183,
-      "learning_rate": 0.0001999957830237901,
-      "loss": 46.0,
-      "step": 38244
-    },
-    {
-      "epoch": 2.924097329739855,
-      "grad_norm": 0.0007240738486871123,
-      "learning_rate": 0.000199995782803201,
-      "loss": 46.0,
-      "step": 38245
-    },
-    {
-      "epoch": 2.924173786723245,
-      "grad_norm": 0.0021234822925180197,
-      "learning_rate": 0.00019999578258260613,
-      "loss": 46.0,
-      "step": 38246
-    },
-    {
-      "epoch": 2.9242502437066347,
-      "grad_norm": 0.002773014595732093,
-      "learning_rate": 0.0001999957823620055,
-      "loss": 46.0,
-      "step": 38247
-    },
-    {
-      "epoch": 2.9243267006900244,
-      "grad_norm": 0.0006871841032989323,
-      "learning_rate": 0.0001999957821413991,
-      "loss": 46.0,
-      "step": 38248
-    },
-    {
-      "epoch": 2.9244031576734137,
-      "grad_norm": 0.0046259560622274876,
-      "learning_rate": 0.00019999578192078692,
-      "loss": 46.0,
-      "step": 38249
-    },
-    {
-      "epoch": 2.9244796146568035,
-      "grad_norm": 0.005346689838916063,
-      "learning_rate": 0.00019999578170016896,
-      "loss": 46.0,
-      "step": 38250
-    },
-    {
-      "epoch": 2.9245560716401933,
-      "grad_norm": 0.0007818485610187054,
-      "learning_rate": 0.00019999578147954529,
-      "loss": 46.0,
-      "step": 38251
-    },
-    {
-      "epoch": 2.924632528623583,
-      "grad_norm": 0.002096218289807439,
-      "learning_rate": 0.0001999957812589158,
-      "loss": 46.0,
-      "step": 38252
-    },
-    {
-      "epoch": 2.924708985606973,
-      "grad_norm": 0.0013558358186855912,
-      "learning_rate": 0.00019999578103828053,
-      "loss": 46.0,
-      "step": 38253
-    },
-    {
-      "epoch": 2.9247854425903625,
-      "grad_norm": 0.0022412913385778666,
-      "learning_rate": 0.00019999578081763953,
-      "loss": 46.0,
-      "step": 38254
-    },
-    {
-      "epoch": 2.9248618995737523,
-      "grad_norm": 0.002418191870674491,
-      "learning_rate": 0.00019999578059699276,
-      "loss": 46.0,
-      "step": 38255
-    },
-    {
-      "epoch": 2.924938356557142,
-      "grad_norm": 0.005712719634175301,
-      "learning_rate": 0.0001999957803763402,
-      "loss": 46.0,
-      "step": 38256
-    },
-    {
-      "epoch": 2.925014813540532,
-      "grad_norm": 0.0008087372407317162,
-      "learning_rate": 0.00019999578015568185,
-      "loss": 46.0,
-      "step": 38257
-    },
-    {
-      "epoch": 2.9250912705239216,
-      "grad_norm": 0.0019517212640494108,
-      "learning_rate": 0.00019999577993501776,
-      "loss": 46.0,
-      "step": 38258
-    },
-    {
-      "epoch": 2.9251677275073114,
-      "grad_norm": 0.0010789376683533192,
-      "learning_rate": 0.0001999957797143479,
-      "loss": 46.0,
-      "step": 38259
-    },
-    {
-      "epoch": 2.925244184490701,
-      "grad_norm": 0.003405132330954075,
-      "learning_rate": 0.00019999577949367226,
-      "loss": 46.0,
-      "step": 38260
-    },
-    {
-      "epoch": 2.925320641474091,
-      "grad_norm": 0.0007047481485642493,
-      "learning_rate": 0.00019999577927299087,
-      "loss": 46.0,
-      "step": 38261
-    },
-    {
-      "epoch": 2.92539709845748,
-      "grad_norm": 0.001475622644647956,
-      "learning_rate": 0.00019999577905230372,
-      "loss": 46.0,
-      "step": 38262
-    },
-    {
-      "epoch": 2.92547355544087,
-      "grad_norm": 0.00219365069642663,
-      "learning_rate": 0.00019999577883161076,
-      "loss": 46.0,
-      "step": 38263
-    },
-    {
-      "epoch": 2.9255500124242597,
-      "grad_norm": 0.002794957719743252,
-      "learning_rate": 0.00019999577861091203,
-      "loss": 46.0,
-      "step": 38264
-    },
-    {
-      "epoch": 2.9256264694076495,
-      "grad_norm": 0.0022276996169239283,
-      "learning_rate": 0.00019999577839020755,
-      "loss": 46.0,
-      "step": 38265
-    },
-    {
-      "epoch": 2.9257029263910392,
-      "grad_norm": 0.0016185338608920574,
-      "learning_rate": 0.0001999957781694973,
-      "loss": 46.0,
-      "step": 38266
-    },
-    {
-      "epoch": 2.925779383374429,
-      "grad_norm": 0.002163441153243184,
-      "learning_rate": 0.0001999957779487813,
-      "loss": 46.0,
-      "step": 38267
-    },
-    {
-      "epoch": 2.9258558403578188,
-      "grad_norm": 0.0004720076103694737,
-      "learning_rate": 0.00019999577772805954,
-      "loss": 46.0,
-      "step": 38268
-    },
-    {
-      "epoch": 2.9259322973412085,
-      "grad_norm": 0.002807009732350707,
-      "learning_rate": 0.00019999577750733197,
-      "loss": 46.0,
-      "step": 38269
-    },
-    {
-      "epoch": 2.9260087543245983,
-      "grad_norm": 0.004007470794022083,
-      "learning_rate": 0.00019999577728659865,
-      "loss": 46.0,
-      "step": 38270
-    },
-    {
-      "epoch": 2.9260852113079876,
-      "grad_norm": 0.00235898420214653,
-      "learning_rate": 0.00019999577706585953,
-      "loss": 46.0,
-      "step": 38271
-    },
-    {
-      "epoch": 2.9261616682913774,
-      "grad_norm": 0.0013184536946937442,
-      "learning_rate": 0.00019999577684511467,
-      "loss": 46.0,
-      "step": 38272
-    },
-    {
-      "epoch": 2.926238125274767,
-      "grad_norm": 0.0012576011940836906,
-      "learning_rate": 0.00019999577662436403,
-      "loss": 46.0,
-      "step": 38273
-    },
-    {
-      "epoch": 2.926314582258157,
-      "grad_norm": 0.002534965518862009,
-      "learning_rate": 0.00019999577640360765,
-      "loss": 46.0,
-      "step": 38274
-    },
-    {
-      "epoch": 2.9263910392415466,
-      "grad_norm": 0.0013629967579618096,
-      "learning_rate": 0.0001999957761828455,
-      "loss": 46.0,
-      "step": 38275
-    },
-    {
-      "epoch": 2.9264674962249364,
-      "grad_norm": 0.003923827316612005,
-      "learning_rate": 0.00019999577596207754,
-      "loss": 46.0,
-      "step": 38276
-    },
-    {
-      "epoch": 2.926543953208326,
-      "grad_norm": 0.002120506949722767,
-      "learning_rate": 0.00019999577574130383,
-      "loss": 46.0,
-      "step": 38277
-    },
-    {
-      "epoch": 2.926620410191716,
-      "grad_norm": 0.0011064867721870542,
-      "learning_rate": 0.00019999577552052436,
-      "loss": 46.0,
-      "step": 38278
-    },
-    {
-      "epoch": 2.9266968671751057,
-      "grad_norm": 0.0023922501131892204,
-      "learning_rate": 0.0001999957752997391,
-      "loss": 46.0,
-      "step": 38279
-    },
-    {
-      "epoch": 2.9267733241584954,
-      "grad_norm": 0.0006455389666371047,
-      "learning_rate": 0.00019999577507894808,
-      "loss": 46.0,
-      "step": 38280
-    },
-    {
-      "epoch": 2.926849781141885,
-      "grad_norm": 0.0068582636304199696,
-      "learning_rate": 0.00019999577485815131,
-      "loss": 46.0,
-      "step": 38281
-    },
-    {
-      "epoch": 2.926926238125275,
-      "grad_norm": 0.002149442443624139,
-      "learning_rate": 0.00019999577463734874,
-      "loss": 46.0,
-      "step": 38282
-    },
-    {
-      "epoch": 2.9270026951086647,
-      "grad_norm": 0.0010842869523912668,
-      "learning_rate": 0.00019999577441654043,
-      "loss": 46.0,
-      "step": 38283
-    },
-    {
-      "epoch": 2.927079152092054,
-      "grad_norm": 0.001289301086217165,
-      "learning_rate": 0.00019999577419572634,
-      "loss": 46.0,
-      "step": 38284
-    },
-    {
-      "epoch": 2.927155609075444,
-      "grad_norm": 0.0013966499827802181,
-      "learning_rate": 0.00019999577397490645,
-      "loss": 46.0,
-      "step": 38285
-    },
-    {
-      "epoch": 2.9272320660588336,
-      "grad_norm": 0.0028955533634871244,
-      "learning_rate": 0.00019999577375408084,
-      "loss": 46.0,
-      "step": 38286
-    },
-    {
-      "epoch": 2.9273085230422233,
-      "grad_norm": 0.0013165291165933013,
-      "learning_rate": 0.00019999577353324943,
-      "loss": 46.0,
-      "step": 38287
-    },
-    {
-      "epoch": 2.927384980025613,
-      "grad_norm": 0.0037420731969177723,
-      "learning_rate": 0.00019999577331241227,
-      "loss": 46.0,
-      "step": 38288
-    },
-    {
-      "epoch": 2.927461437009003,
-      "grad_norm": 0.001261544763110578,
-      "learning_rate": 0.00019999577309156932,
-      "loss": 46.0,
-      "step": 38289
-    },
-    {
-      "epoch": 2.9275378939923926,
-      "grad_norm": 0.0035089347511529922,
-      "learning_rate": 0.0001999957728707206,
-      "loss": 46.0,
-      "step": 38290
-    },
-    {
-      "epoch": 2.9276143509757824,
-      "grad_norm": 0.002187797799706459,
-      "learning_rate": 0.00019999577264986614,
-      "loss": 46.0,
-      "step": 38291
-    },
-    {
-      "epoch": 2.927690807959172,
-      "grad_norm": 0.001236019772477448,
-      "learning_rate": 0.0001999957724290059,
-      "loss": 46.0,
-      "step": 38292
-    },
-    {
-      "epoch": 2.9277672649425615,
-      "grad_norm": 0.0028971703723073006,
-      "learning_rate": 0.00019999577220813986,
-      "loss": 46.0,
-      "step": 38293
-    },
-    {
-      "epoch": 2.927843721925951,
-      "grad_norm": 0.0006433755625039339,
-      "learning_rate": 0.0001999957719872681,
-      "loss": 46.0,
-      "step": 38294
-    },
-    {
-      "epoch": 2.927920178909341,
-      "grad_norm": 0.0025660835672169924,
-      "learning_rate": 0.00019999577176639053,
-      "loss": 46.0,
-      "step": 38295
-    },
-    {
-      "epoch": 2.9279966358927307,
-      "grad_norm": 0.0071014766581356525,
-      "learning_rate": 0.00019999577154550718,
-      "loss": 46.0,
-      "step": 38296
-    },
-    {
-      "epoch": 2.9280730928761205,
-      "grad_norm": 0.0021674518939107656,
-      "learning_rate": 0.00019999577132461812,
-      "loss": 46.0,
-      "step": 38297
-    },
-    {
-      "epoch": 2.9281495498595103,
-      "grad_norm": 0.002416769741103053,
-      "learning_rate": 0.00019999577110372326,
-      "loss": 46.0,
-      "step": 38298
-    },
-    {
-      "epoch": 2.9282260068429,
-      "grad_norm": 0.0031663712579756975,
-      "learning_rate": 0.00019999577088282262,
-      "loss": 46.0,
-      "step": 38299
-    },
-    {
-      "epoch": 2.92830246382629,
-      "grad_norm": 0.003932499792426825,
-      "learning_rate": 0.0001999957706619162,
-      "loss": 46.0,
-      "step": 38300
-    },
-    {
-      "epoch": 2.9283789208096795,
-      "grad_norm": 0.0038557236548513174,
-      "learning_rate": 0.00019999577044100403,
-      "loss": 46.0,
-      "step": 38301
-    },
-    {
-      "epoch": 2.9284553777930693,
-      "grad_norm": 0.0029283526819199324,
-      "learning_rate": 0.0001999957702200861,
-      "loss": 46.0,
-      "step": 38302
-    },
-    {
-      "epoch": 2.928531834776459,
-      "grad_norm": 0.0018550771055743098,
-      "learning_rate": 0.0001999957699991624,
-      "loss": 46.0,
-      "step": 38303
-    },
-    {
-      "epoch": 2.928608291759849,
-      "grad_norm": 0.001281319884583354,
-      "learning_rate": 0.00019999576977823292,
-      "loss": 46.0,
-      "step": 38304
-    },
-    {
-      "epoch": 2.9286847487432386,
-      "grad_norm": 0.002361670834943652,
-      "learning_rate": 0.00019999576955729767,
-      "loss": 46.0,
-      "step": 38305
-    },
-    {
-      "epoch": 2.928761205726628,
-      "grad_norm": 0.0011752928839996457,
-      "learning_rate": 0.00019999576933635665,
-      "loss": 46.0,
-      "step": 38306
-    },
-    {
-      "epoch": 2.9288376627100177,
-      "grad_norm": 0.0019372672541067004,
-      "learning_rate": 0.00019999576911540988,
-      "loss": 46.0,
-      "step": 38307
-    },
-    {
-      "epoch": 2.9289141196934074,
-      "grad_norm": 0.004979288205504417,
-      "learning_rate": 0.0001999957688944573,
-      "loss": 46.0,
-      "step": 38308
-    },
-    {
-      "epoch": 2.928990576676797,
-      "grad_norm": 0.0023889269214123487,
-      "learning_rate": 0.000199995768673499,
-      "loss": 46.0,
-      "step": 38309
-    },
-    {
-      "epoch": 2.929067033660187,
-      "grad_norm": 0.0008140472928062081,
-      "learning_rate": 0.0001999957684525349,
-      "loss": 46.0,
-      "step": 38310
-    },
-    {
-      "epoch": 2.9291434906435767,
-      "grad_norm": 0.0005786191904917359,
-      "learning_rate": 0.00019999576823156504,
-      "loss": 46.0,
-      "step": 38311
-    },
-    {
-      "epoch": 2.9292199476269665,
-      "grad_norm": 0.0016067971009761095,
-      "learning_rate": 0.0001999957680105894,
-      "loss": 46.0,
-      "step": 38312
-    },
-    {
-      "epoch": 2.9292964046103562,
-      "grad_norm": 0.0009492552489973605,
-      "learning_rate": 0.000199995767789608,
-      "loss": 46.0,
-      "step": 38313
-    },
-    {
-      "epoch": 2.9293728615937455,
-      "grad_norm": 0.003731977194547653,
-      "learning_rate": 0.00019999576756862084,
-      "loss": 46.0,
-      "step": 38314
-    },
-    {
-      "epoch": 2.9294493185771353,
-      "grad_norm": 0.0018419703701511025,
-      "learning_rate": 0.00019999576734762792,
-      "loss": 46.0,
-      "step": 38315
-    },
-    {
-      "epoch": 2.929525775560525,
-      "grad_norm": 0.0025374721735715866,
-      "learning_rate": 0.00019999576712662919,
-      "loss": 46.0,
-      "step": 38316
-    },
-    {
-      "epoch": 2.929602232543915,
-      "grad_norm": 0.0015094116097316146,
-      "learning_rate": 0.0001999957669056247,
-      "loss": 46.0,
-      "step": 38317
-    },
-    {
-      "epoch": 2.9296786895273046,
-      "grad_norm": 0.0017964153084903955,
-      "learning_rate": 0.00019999576668461446,
-      "loss": 46.0,
-      "step": 38318
-    },
-    {
-      "epoch": 2.9297551465106944,
-      "grad_norm": 0.0022300092969089746,
-      "learning_rate": 0.00019999576646359844,
-      "loss": 46.0,
-      "step": 38319
-    },
-    {
-      "epoch": 2.929831603494084,
-      "grad_norm": 0.00252294703386724,
-      "learning_rate": 0.00019999576624257664,
-      "loss": 46.0,
-      "step": 38320
-    },
-    {
-      "epoch": 2.929908060477474,
-      "grad_norm": 0.005831143353134394,
-      "learning_rate": 0.0001999957660215491,
-      "loss": 46.0,
-      "step": 38321
-    },
-    {
-      "epoch": 2.9299845174608636,
-      "grad_norm": 0.001255890354514122,
-      "learning_rate": 0.00019999576580051579,
-      "loss": 46.0,
-      "step": 38322
-    },
-    {
-      "epoch": 2.9300609744442534,
-      "grad_norm": 0.0022746275644749403,
-      "learning_rate": 0.00019999576557947667,
-      "loss": 46.0,
-      "step": 38323
-    },
-    {
-      "epoch": 2.930137431427643,
-      "grad_norm": 0.002677883952856064,
-      "learning_rate": 0.00019999576535843184,
-      "loss": 46.0,
-      "step": 38324
-    },
-    {
-      "epoch": 2.930213888411033,
-      "grad_norm": 0.0027914121747016907,
-      "learning_rate": 0.0001999957651373812,
-      "loss": 46.0,
-      "step": 38325
-    },
-    {
-      "epoch": 2.9302903453944227,
-      "grad_norm": 0.004142760299146175,
-      "learning_rate": 0.0001999957649163248,
-      "loss": 46.0,
-      "step": 38326
-    },
-    {
-      "epoch": 2.9303668023778124,
-      "grad_norm": 0.001855485956184566,
-      "learning_rate": 0.00019999576469526264,
-      "loss": 46.0,
-      "step": 38327
-    },
-    {
-      "epoch": 2.9304432593612018,
-      "grad_norm": 0.002708352403715253,
-      "learning_rate": 0.00019999576447419468,
-      "loss": 46.0,
-      "step": 38328
-    },
-    {
-      "epoch": 2.9305197163445915,
-      "grad_norm": 0.0008962448337115347,
-      "learning_rate": 0.00019999576425312098,
-      "loss": 46.0,
-      "step": 38329
-    },
-    {
-      "epoch": 2.9305961733279813,
-      "grad_norm": 0.0022742683067917824,
-      "learning_rate": 0.00019999576403204153,
-      "loss": 46.0,
-      "step": 38330
-    },
-    {
-      "epoch": 2.930672630311371,
-      "grad_norm": 0.003403066424652934,
-      "learning_rate": 0.00019999576381095625,
-      "loss": 46.0,
-      "step": 38331
-    },
-    {
-      "epoch": 2.930749087294761,
-      "grad_norm": 0.001809767447412014,
-      "learning_rate": 0.00019999576358986526,
-      "loss": 46.0,
-      "step": 38332
-    },
-    {
-      "epoch": 2.9308255442781506,
-      "grad_norm": 0.005718843080103397,
-      "learning_rate": 0.00019999576336876846,
-      "loss": 46.0,
-      "step": 38333
-    },
-    {
-      "epoch": 2.9309020012615403,
-      "grad_norm": 0.0017335818847641349,
-      "learning_rate": 0.0001999957631476659,
-      "loss": 46.0,
-      "step": 38334
-    },
-    {
-      "epoch": 2.93097845824493,
-      "grad_norm": 0.0009207432740367949,
-      "learning_rate": 0.0001999957629265576,
-      "loss": 46.0,
-      "step": 38335
-    },
-    {
-      "epoch": 2.9310549152283194,
-      "grad_norm": 0.002403505379334092,
-      "learning_rate": 0.0001999957627054435,
-      "loss": 46.0,
-      "step": 38336
-    },
-    {
-      "epoch": 2.931131372211709,
-      "grad_norm": 0.0031551194842904806,
-      "learning_rate": 0.0001999957624843236,
-      "loss": 46.0,
-      "step": 38337
-    },
-    {
-      "epoch": 2.931207829195099,
-      "grad_norm": 0.0017367189284414053,
-      "learning_rate": 0.000199995762263198,
-      "loss": 46.0,
-      "step": 38338
-    },
-    {
-      "epoch": 2.9312842861784887,
-      "grad_norm": 0.0038173741195350885,
-      "learning_rate": 0.0001999957620420666,
-      "loss": 46.0,
-      "step": 38339
-    },
-    {
-      "epoch": 2.9313607431618784,
-      "grad_norm": 0.004370069596916437,
-      "learning_rate": 0.00019999576182092943,
-      "loss": 46.0,
-      "step": 38340
-    },
-    {
-      "epoch": 2.931437200145268,
-      "grad_norm": 0.00426573446020484,
-      "learning_rate": 0.0001999957615997865,
-      "loss": 46.0,
-      "step": 38341
-    },
-    {
-      "epoch": 2.931513657128658,
-      "grad_norm": 0.0016198699595406651,
-      "learning_rate": 0.00019999576137863777,
-      "loss": 46.0,
-      "step": 38342
-    },
-    {
-      "epoch": 2.9315901141120477,
-      "grad_norm": 0.002621751744300127,
-      "learning_rate": 0.0001999957611574833,
-      "loss": 46.0,
-      "step": 38343
-    },
-    {
-      "epoch": 2.9316665710954375,
-      "grad_norm": 0.0036541041918098927,
-      "learning_rate": 0.00019999576093632305,
-      "loss": 46.0,
-      "step": 38344
-    },
-    {
-      "epoch": 2.9317430280788273,
-      "grad_norm": 0.005183043889701366,
-      "learning_rate": 0.00019999576071515703,
-      "loss": 46.0,
-      "step": 38345
-    },
-    {
-      "epoch": 2.931819485062217,
-      "grad_norm": 0.0007401772891171277,
-      "learning_rate": 0.00019999576049398526,
-      "loss": 46.0,
-      "step": 38346
-    },
-    {
-      "epoch": 2.9318959420456068,
-      "grad_norm": 0.0012930460507050157,
-      "learning_rate": 0.00019999576027280772,
-      "loss": 46.0,
-      "step": 38347
-    },
-    {
-      "epoch": 2.9319723990289965,
-      "grad_norm": 0.0020210538059473038,
-      "learning_rate": 0.00019999576005162438,
-      "loss": 46.0,
-      "step": 38348
-    },
-    {
-      "epoch": 2.9320488560123863,
-      "grad_norm": 0.002794157015159726,
-      "learning_rate": 0.0001999957598304353,
-      "loss": 46.0,
-      "step": 38349
-    },
-    {
-      "epoch": 2.9321253129957756,
-      "grad_norm": 0.0052177696488797665,
-      "learning_rate": 0.00019999575960924043,
-      "loss": 46.0,
-      "step": 38350
-    },
-    {
-      "epoch": 2.9322017699791654,
-      "grad_norm": 0.003247017040848732,
-      "learning_rate": 0.0001999957593880398,
-      "loss": 46.0,
-      "step": 38351
-    },
-    {
-      "epoch": 2.932278226962555,
-      "grad_norm": 0.0011789138661697507,
-      "learning_rate": 0.00019999575916683339,
-      "loss": 46.0,
-      "step": 38352
-    },
-    {
-      "epoch": 2.932354683945945,
-      "grad_norm": 0.0018850021297112107,
-      "learning_rate": 0.0001999957589456212,
-      "loss": 46.0,
-      "step": 38353
-    },
-    {
-      "epoch": 2.9324311409293347,
-      "grad_norm": 0.005937132518738508,
-      "learning_rate": 0.00019999575872440328,
-      "loss": 46.0,
-      "step": 38354
-    },
-    {
-      "epoch": 2.9325075979127244,
-      "grad_norm": 0.004149425774812698,
-      "learning_rate": 0.00019999575850317958,
-      "loss": 46.0,
-      "step": 38355
-    },
-    {
-      "epoch": 2.932584054896114,
-      "grad_norm": 0.0015420602867379785,
-      "learning_rate": 0.0001999957582819501,
-      "loss": 46.0,
-      "step": 38356
-    },
-    {
-      "epoch": 2.932660511879504,
-      "grad_norm": 0.0037689516320824623,
-      "learning_rate": 0.00019999575806071485,
-      "loss": 46.0,
-      "step": 38357
-    },
-    {
-      "epoch": 2.9327369688628933,
-      "grad_norm": 0.006749901920557022,
-      "learning_rate": 0.00019999575783947383,
-      "loss": 46.0,
-      "step": 38358
-    },
-    {
-      "epoch": 2.932813425846283,
-      "grad_norm": 0.0022343432065099478,
-      "learning_rate": 0.00019999575761822706,
-      "loss": 46.0,
-      "step": 38359
-    },
-    {
-      "epoch": 2.932889882829673,
-      "grad_norm": 0.0012883387971669436,
-      "learning_rate": 0.0001999957573969745,
-      "loss": 46.0,
-      "step": 38360
-    },
-    {
-      "epoch": 2.9329663398130625,
-      "grad_norm": 0.002754267305135727,
-      "learning_rate": 0.00019999575717571618,
-      "loss": 46.0,
-      "step": 38361
-    },
-    {
-      "epoch": 2.9330427967964523,
-      "grad_norm": 0.0036985205952078104,
-      "learning_rate": 0.0001999957569544521,
-      "loss": 46.0,
-      "step": 38362
-    },
-    {
-      "epoch": 2.933119253779842,
-      "grad_norm": 0.0024605635553598404,
-      "learning_rate": 0.00019999575673318223,
-      "loss": 46.0,
-      "step": 38363
-    },
-    {
-      "epoch": 2.933195710763232,
-      "grad_norm": 0.004922663327306509,
-      "learning_rate": 0.0001999957565119066,
-      "loss": 46.0,
-      "step": 38364
-    },
-    {
-      "epoch": 2.9332721677466216,
-      "grad_norm": 0.0009371376363560557,
-      "learning_rate": 0.00019999575629062522,
-      "loss": 46.0,
-      "step": 38365
-    },
-    {
-      "epoch": 2.9333486247300113,
-      "grad_norm": 0.0012911951635032892,
-      "learning_rate": 0.00019999575606933804,
-      "loss": 46.0,
-      "step": 38366
-    },
-    {
-      "epoch": 2.933425081713401,
-      "grad_norm": 0.0031505294609814882,
-      "learning_rate": 0.0001999957558480451,
-      "loss": 46.0,
-      "step": 38367
-    },
-    {
-      "epoch": 2.933501538696791,
-      "grad_norm": 0.003799398662522435,
-      "learning_rate": 0.0001999957556267464,
-      "loss": 46.0,
-      "step": 38368
-    },
-    {
-      "epoch": 2.9335779956801806,
-      "grad_norm": 0.001875002752058208,
-      "learning_rate": 0.0001999957554054419,
-      "loss": 46.0,
-      "step": 38369
-    },
-    {
-      "epoch": 2.9336544526635704,
-      "grad_norm": 0.0013595259515568614,
-      "learning_rate": 0.00019999575518413166,
-      "loss": 46.0,
-      "step": 38370
-    },
-    {
-      "epoch": 2.93373090964696,
-      "grad_norm": 0.005625258199870586,
-      "learning_rate": 0.00019999575496281566,
-      "loss": 46.0,
-      "step": 38371
-    },
-    {
-      "epoch": 2.9338073666303495,
-      "grad_norm": 0.0014829987194389105,
-      "learning_rate": 0.00019999575474149387,
-      "loss": 46.0,
-      "step": 38372
-    },
-    {
-      "epoch": 2.9338838236137392,
-      "grad_norm": 0.0019627483561635017,
-      "learning_rate": 0.00019999575452016633,
-      "loss": 46.0,
-      "step": 38373
-    },
-    {
-      "epoch": 2.933960280597129,
-      "grad_norm": 0.0017464442644268274,
-      "learning_rate": 0.000199995754298833,
-      "loss": 46.0,
-      "step": 38374
-    },
-    {
-      "epoch": 2.9340367375805187,
-      "grad_norm": 0.001066903700120747,
-      "learning_rate": 0.0001999957540774939,
-      "loss": 46.0,
-      "step": 38375
-    },
-    {
-      "epoch": 2.9341131945639085,
-      "grad_norm": 0.0005869469023309648,
-      "learning_rate": 0.00019999575385614904,
-      "loss": 46.0,
-      "step": 38376
-    },
-    {
-      "epoch": 2.9341896515472983,
-      "grad_norm": 0.0016888286918401718,
-      "learning_rate": 0.0001999957536347984,
-      "loss": 46.0,
-      "step": 38377
-    },
-    {
-      "epoch": 2.934266108530688,
-      "grad_norm": 0.0027584037743508816,
-      "learning_rate": 0.000199995753413442,
-      "loss": 46.0,
-      "step": 38378
-    },
-    {
-      "epoch": 2.934342565514078,
-      "grad_norm": 0.0047539896331727505,
-      "learning_rate": 0.00019999575319207984,
-      "loss": 46.0,
-      "step": 38379
-    },
-    {
-      "epoch": 2.934419022497467,
-      "grad_norm": 0.0020775427110493183,
-      "learning_rate": 0.0001999957529707119,
-      "loss": 46.0,
-      "step": 38380
-    },
-    {
-      "epoch": 2.934495479480857,
-      "grad_norm": 0.0021214515436440706,
-      "learning_rate": 0.0001999957527493382,
-      "loss": 46.0,
-      "step": 38381
-    },
-    {
-      "epoch": 2.9345719364642466,
-      "grad_norm": 0.0025901715271174908,
-      "learning_rate": 0.00019999575252795871,
-      "loss": 46.0,
-      "step": 38382
-    },
-    {
-      "epoch": 2.9346483934476364,
-      "grad_norm": 0.0033463803119957447,
-      "learning_rate": 0.00019999575230657347,
-      "loss": 46.0,
-      "step": 38383
-    },
-    {
-      "epoch": 2.934724850431026,
-      "grad_norm": 0.003966361749917269,
-      "learning_rate": 0.00019999575208518247,
-      "loss": 46.0,
-      "step": 38384
-    },
-    {
-      "epoch": 2.934801307414416,
-      "grad_norm": 0.0012724873377010226,
-      "learning_rate": 0.00019999575186378568,
-      "loss": 46.0,
-      "step": 38385
-    },
-    {
-      "epoch": 2.9348777643978057,
-      "grad_norm": 0.0018246839754283428,
-      "learning_rate": 0.0001999957516423831,
-      "loss": 46.0,
-      "step": 38386
-    },
-    {
-      "epoch": 2.9349542213811954,
-      "grad_norm": 0.0021874033845961094,
-      "learning_rate": 0.0001999957514209748,
-      "loss": 46.0,
-      "step": 38387
-    },
-    {
-      "epoch": 2.935030678364585,
-      "grad_norm": 0.0012727330904453993,
-      "learning_rate": 0.0001999957511995607,
-      "loss": 46.0,
-      "step": 38388
-    },
-    {
-      "epoch": 2.935107135347975,
-      "grad_norm": 0.0038354459684342146,
-      "learning_rate": 0.00019999575097814085,
-      "loss": 46.0,
-      "step": 38389
-    },
-    {
-      "epoch": 2.9351835923313647,
-      "grad_norm": 0.004747078288346529,
-      "learning_rate": 0.00019999575075671522,
-      "loss": 46.0,
-      "step": 38390
-    },
-    {
-      "epoch": 2.9352600493147545,
-      "grad_norm": 0.0007668832549825311,
-      "learning_rate": 0.0001999957505352838,
-      "loss": 46.0,
-      "step": 38391
-    },
-    {
-      "epoch": 2.9353365062981442,
-      "grad_norm": 0.0034495287109166384,
-      "learning_rate": 0.00019999575031384666,
-      "loss": 46.0,
-      "step": 38392
-    },
-    {
-      "epoch": 2.935412963281534,
-      "grad_norm": 0.0013124882243573666,
-      "learning_rate": 0.00019999575009240373,
-      "loss": 46.0,
-      "step": 38393
-    },
-    {
-      "epoch": 2.9354894202649233,
-      "grad_norm": 0.0016738465055823326,
-      "learning_rate": 0.000199995749870955,
-      "loss": 46.0,
-      "step": 38394
-    },
-    {
-      "epoch": 2.935565877248313,
-      "grad_norm": 0.0022193174809217453,
-      "learning_rate": 0.00019999574964950053,
-      "loss": 46.0,
-      "step": 38395
-    },
-    {
-      "epoch": 2.935642334231703,
-      "grad_norm": 0.002999038202688098,
-      "learning_rate": 0.00019999574942804028,
-      "loss": 46.0,
-      "step": 38396
-    },
-    {
-      "epoch": 2.9357187912150926,
-      "grad_norm": 0.00505444873124361,
-      "learning_rate": 0.00019999574920657426,
-      "loss": 46.0,
-      "step": 38397
-    },
-    {
-      "epoch": 2.9357952481984824,
-      "grad_norm": 0.0033488075714558363,
-      "learning_rate": 0.0001999957489851025,
-      "loss": 46.0,
-      "step": 38398
-    },
-    {
-      "epoch": 2.935871705181872,
-      "grad_norm": 0.0030953786335885525,
-      "learning_rate": 0.00019999574876362495,
-      "loss": 46.0,
-      "step": 38399
-    },
-    {
-      "epoch": 2.935948162165262,
-      "grad_norm": 0.003579887095838785,
-      "learning_rate": 0.0001999957485421416,
-      "loss": 46.0,
-      "step": 38400
-    },
-    {
-      "epoch": 2.9360246191486516,
-      "grad_norm": 0.0011574856471270323,
-      "learning_rate": 0.00019999574832065253,
-      "loss": 46.0,
-      "step": 38401
-    },
-    {
-      "epoch": 2.936101076132041,
-      "grad_norm": 0.0011123744770884514,
-      "learning_rate": 0.00019999574809915767,
-      "loss": 46.0,
-      "step": 38402
-    },
-    {
-      "epoch": 2.9361775331154307,
-      "grad_norm": 0.0013532133307307959,
-      "learning_rate": 0.00019999574787765703,
-      "loss": 46.0,
-      "step": 38403
-    },
-    {
-      "epoch": 2.9362539900988205,
-      "grad_norm": 0.001378371729515493,
-      "learning_rate": 0.00019999574765615063,
-      "loss": 46.0,
-      "step": 38404
-    },
-    {
-      "epoch": 2.9363304470822102,
-      "grad_norm": 0.0020623046439141035,
-      "learning_rate": 0.00019999574743463847,
-      "loss": 46.0,
-      "step": 38405
-    },
-    {
-      "epoch": 2.9364069040656,
-      "grad_norm": 0.0021001333370804787,
-      "learning_rate": 0.00019999574721312055,
-      "loss": 46.0,
-      "step": 38406
-    },
-    {
-      "epoch": 2.9364833610489898,
-      "grad_norm": 0.003137333784252405,
-      "learning_rate": 0.00019999574699159682,
-      "loss": 46.0,
-      "step": 38407
-    },
-    {
-      "epoch": 2.9365598180323795,
-      "grad_norm": 0.005436848849058151,
-      "learning_rate": 0.00019999574677006737,
-      "loss": 46.0,
-      "step": 38408
-    },
-    {
-      "epoch": 2.9366362750157693,
-      "grad_norm": 0.0013779301661998034,
-      "learning_rate": 0.0001999957465485321,
-      "loss": 46.0,
-      "step": 38409
-    },
-    {
-      "epoch": 2.936712731999159,
-      "grad_norm": 0.006826538126915693,
-      "learning_rate": 0.00019999574632699108,
-      "loss": 46.0,
-      "step": 38410
-    },
-    {
-      "epoch": 2.936789188982549,
-      "grad_norm": 0.004575145896524191,
-      "learning_rate": 0.00019999574610544428,
-      "loss": 46.0,
-      "step": 38411
-    },
-    {
-      "epoch": 2.9368656459659386,
-      "grad_norm": 0.002724838675931096,
-      "learning_rate": 0.00019999574588389175,
-      "loss": 46.0,
-      "step": 38412
-    },
-    {
-      "epoch": 2.9369421029493283,
-      "grad_norm": 0.0011032322654500604,
-      "learning_rate": 0.00019999574566233343,
-      "loss": 46.0,
-      "step": 38413
-    },
-    {
-      "epoch": 2.937018559932718,
-      "grad_norm": 0.0007204872090369463,
-      "learning_rate": 0.00019999574544076935,
-      "loss": 46.0,
-      "step": 38414
-    },
-    {
-      "epoch": 2.9370950169161074,
-      "grad_norm": 0.0009414879605174065,
-      "learning_rate": 0.00019999574521919949,
-      "loss": 46.0,
-      "step": 38415
-    },
-    {
-      "epoch": 2.937171473899497,
-      "grad_norm": 0.0019028965616598725,
-      "learning_rate": 0.00019999574499762385,
-      "loss": 46.0,
-      "step": 38416
-    },
-    {
-      "epoch": 2.937247930882887,
-      "grad_norm": 0.0013468819670379162,
-      "learning_rate": 0.00019999574477604247,
-      "loss": 46.0,
-      "step": 38417
-    },
-    {
-      "epoch": 2.9373243878662767,
-      "grad_norm": 0.002717470284551382,
-      "learning_rate": 0.0001999957445544553,
-      "loss": 46.0,
-      "step": 38418
-    },
-    {
-      "epoch": 2.9374008448496665,
-      "grad_norm": 0.002555568004027009,
-      "learning_rate": 0.00019999574433286234,
-      "loss": 46.0,
-      "step": 38419
-    },
-    {
-      "epoch": 2.937477301833056,
-      "grad_norm": 0.0011127963662147522,
-      "learning_rate": 0.00019999574411126364,
-      "loss": 46.0,
-      "step": 38420
-    },
-    {
-      "epoch": 2.937553758816446,
-      "grad_norm": 0.0038804232608526945,
-      "learning_rate": 0.0001999957438896592,
-      "loss": 46.0,
-      "step": 38421
-    },
-    {
-      "epoch": 2.9376302157998357,
-      "grad_norm": 0.0018935680855065584,
-      "learning_rate": 0.00019999574366804895,
-      "loss": 46.0,
-      "step": 38422
-    },
-    {
-      "epoch": 2.9377066727832255,
-      "grad_norm": 0.00648722006008029,
-      "learning_rate": 0.00019999574344643293,
-      "loss": 46.0,
-      "step": 38423
-    },
-    {
-      "epoch": 2.937783129766615,
-      "grad_norm": 0.0017362262587994337,
-      "learning_rate": 0.00019999574322481113,
-      "loss": 46.0,
-      "step": 38424
-    },
-    {
-      "epoch": 2.9378595867500046,
-      "grad_norm": 0.0016211242182180285,
-      "learning_rate": 0.00019999574300318357,
-      "loss": 46.0,
-      "step": 38425
-    },
-    {
-      "epoch": 2.9379360437333943,
-      "grad_norm": 0.004038037732243538,
-      "learning_rate": 0.00019999574278155025,
-      "loss": 46.0,
-      "step": 38426
-    },
-    {
-      "epoch": 2.938012500716784,
-      "grad_norm": 0.0010735326213762164,
-      "learning_rate": 0.00019999574255991117,
-      "loss": 46.0,
-      "step": 38427
-    },
-    {
-      "epoch": 2.938088957700174,
-      "grad_norm": 0.0018506509950384498,
-      "learning_rate": 0.0001999957423382663,
-      "loss": 46.0,
-      "step": 38428
-    },
-    {
-      "epoch": 2.9381654146835636,
-      "grad_norm": 0.0011335625313222408,
-      "learning_rate": 0.00019999574211661568,
-      "loss": 46.0,
-      "step": 38429
-    },
-    {
-      "epoch": 2.9382418716669534,
-      "grad_norm": 0.009934425354003906,
-      "learning_rate": 0.00019999574189495927,
-      "loss": 46.0,
-      "step": 38430
-    },
-    {
-      "epoch": 2.938318328650343,
-      "grad_norm": 0.0014541858108714223,
-      "learning_rate": 0.00019999574167329712,
-      "loss": 46.0,
-      "step": 38431
-    },
-    {
-      "epoch": 2.938394785633733,
-      "grad_norm": 0.002659382065758109,
-      "learning_rate": 0.0001999957414516292,
-      "loss": 46.0,
-      "step": 38432
-    },
-    {
-      "epoch": 2.9384712426171227,
-      "grad_norm": 0.0027723510283976793,
-      "learning_rate": 0.0001999957412299555,
-      "loss": 46.0,
-      "step": 38433
-    },
-    {
-      "epoch": 2.9385476996005124,
-      "grad_norm": 0.001249409862793982,
-      "learning_rate": 0.000199995741008276,
-      "loss": 46.0,
-      "step": 38434
-    },
-    {
-      "epoch": 2.938624156583902,
-      "grad_norm": 0.0012369652977213264,
-      "learning_rate": 0.00019999574078659075,
-      "loss": 46.0,
-      "step": 38435
-    },
-    {
-      "epoch": 2.938700613567292,
-      "grad_norm": 0.0008058793027885258,
-      "learning_rate": 0.00019999574056489973,
-      "loss": 46.0,
-      "step": 38436
-    },
-    {
-      "epoch": 2.9387770705506813,
-      "grad_norm": 0.002899512182921171,
-      "learning_rate": 0.00019999574034320293,
-      "loss": 46.0,
-      "step": 38437
-    },
-    {
-      "epoch": 2.938853527534071,
-      "grad_norm": 0.0044359853491187096,
-      "learning_rate": 0.0001999957401215004,
-      "loss": 46.0,
-      "step": 38438
-    },
-    {
-      "epoch": 2.938929984517461,
-      "grad_norm": 0.003538022981956601,
-      "learning_rate": 0.00019999573989979208,
-      "loss": 46.0,
-      "step": 38439
-    },
-    {
-      "epoch": 2.9390064415008506,
-      "grad_norm": 0.0013276587706059217,
-      "learning_rate": 0.000199995739678078,
-      "loss": 46.0,
-      "step": 38440
-    },
-    {
-      "epoch": 2.9390828984842403,
-      "grad_norm": 0.004713753703981638,
-      "learning_rate": 0.0001999957394563581,
-      "loss": 46.0,
-      "step": 38441
-    },
-    {
-      "epoch": 2.93915935546763,
-      "grad_norm": 0.00706761097535491,
-      "learning_rate": 0.0001999957392346325,
-      "loss": 46.0,
-      "step": 38442
-    },
-    {
-      "epoch": 2.93923581245102,
-      "grad_norm": 0.0010085137328132987,
-      "learning_rate": 0.00019999573901290113,
-      "loss": 46.0,
-      "step": 38443
-    },
-    {
-      "epoch": 2.9393122694344096,
-      "grad_norm": 0.0024339633528143167,
-      "learning_rate": 0.00019999573879116395,
-      "loss": 46.0,
-      "step": 38444
-    },
-    {
-      "epoch": 2.939388726417799,
-      "grad_norm": 0.004985821433365345,
-      "learning_rate": 0.000199995738569421,
-      "loss": 46.0,
-      "step": 38445
-    },
-    {
-      "epoch": 2.9394651834011887,
-      "grad_norm": 0.001614739652723074,
-      "learning_rate": 0.0001999957383476723,
-      "loss": 46.0,
-      "step": 38446
-    },
-    {
-      "epoch": 2.9395416403845784,
-      "grad_norm": 0.0023264975752681494,
-      "learning_rate": 0.00019999573812591782,
-      "loss": 46.0,
-      "step": 38447
-    },
-    {
-      "epoch": 2.939618097367968,
-      "grad_norm": 0.003263306338340044,
-      "learning_rate": 0.00019999573790415758,
-      "loss": 46.0,
-      "step": 38448
-    },
-    {
-      "epoch": 2.939694554351358,
-      "grad_norm": 0.0012822872959077358,
-      "learning_rate": 0.00019999573768239156,
-      "loss": 46.0,
-      "step": 38449
-    },
-    {
-      "epoch": 2.9397710113347477,
-      "grad_norm": 0.003085100557655096,
-      "learning_rate": 0.0001999957374606198,
-      "loss": 46.0,
-      "step": 38450
-    },
-    {
-      "epoch": 2.9398474683181375,
-      "grad_norm": 0.0005631727399304509,
-      "learning_rate": 0.00019999573723884223,
-      "loss": 46.0,
-      "step": 38451
-    },
-    {
-      "epoch": 2.9399239253015272,
-      "grad_norm": 0.00225269328802824,
-      "learning_rate": 0.0001999957370170589,
-      "loss": 46.0,
-      "step": 38452
-    },
-    {
-      "epoch": 2.940000382284917,
-      "grad_norm": 0.006019046064466238,
-      "learning_rate": 0.00019999573679526983,
-      "loss": 46.0,
-      "step": 38453
-    },
-    {
-      "epoch": 2.9400768392683068,
-      "grad_norm": 0.0009961413452401757,
-      "learning_rate": 0.00019999573657347494,
-      "loss": 46.0,
-      "step": 38454
-    },
-    {
-      "epoch": 2.9401532962516965,
-      "grad_norm": 0.00282875495031476,
-      "learning_rate": 0.00019999573635167434,
-      "loss": 46.0,
-      "step": 38455
-    },
-    {
-      "epoch": 2.9402297532350863,
-      "grad_norm": 0.004325693007558584,
-      "learning_rate": 0.00019999573612986793,
-      "loss": 46.0,
-      "step": 38456
-    },
-    {
-      "epoch": 2.940306210218476,
-      "grad_norm": 0.0009420639835298061,
-      "learning_rate": 0.00019999573590805575,
-      "loss": 46.0,
-      "step": 38457
-    },
-    {
-      "epoch": 2.940382667201866,
-      "grad_norm": 0.0041830395348370075,
-      "learning_rate": 0.00019999573568623783,
-      "loss": 46.0,
-      "step": 38458
-    },
-    {
-      "epoch": 2.940459124185255,
-      "grad_norm": 0.0032734915148466825,
-      "learning_rate": 0.00019999573546441413,
-      "loss": 46.0,
-      "step": 38459
-    },
-    {
-      "epoch": 2.940535581168645,
-      "grad_norm": 0.002448160434141755,
-      "learning_rate": 0.00019999573524258466,
-      "loss": 46.0,
-      "step": 38460
-    },
-    {
-      "epoch": 2.9406120381520346,
-      "grad_norm": 0.0012298934161663055,
-      "learning_rate": 0.0001999957350207494,
-      "loss": 46.0,
-      "step": 38461
-    },
-    {
-      "epoch": 2.9406884951354244,
-      "grad_norm": 0.0036155146081000566,
-      "learning_rate": 0.00019999573479890842,
-      "loss": 46.0,
-      "step": 38462
-    },
-    {
-      "epoch": 2.940764952118814,
-      "grad_norm": 0.0026368515100330114,
-      "learning_rate": 0.0001999957345770616,
-      "loss": 46.0,
-      "step": 38463
-    },
-    {
-      "epoch": 2.940841409102204,
-      "grad_norm": 0.0007455009035766125,
-      "learning_rate": 0.00019999573435520906,
-      "loss": 46.0,
-      "step": 38464
-    },
-    {
-      "epoch": 2.9409178660855937,
-      "grad_norm": 0.004952927120029926,
-      "learning_rate": 0.00019999573413335075,
-      "loss": 46.0,
-      "step": 38465
-    },
-    {
-      "epoch": 2.9409943230689835,
-      "grad_norm": 0.002793573774397373,
-      "learning_rate": 0.00019999573391148664,
-      "loss": 46.0,
-      "step": 38466
-    },
-    {
-      "epoch": 2.9410707800523728,
-      "grad_norm": 0.002384437248110771,
-      "learning_rate": 0.0001999957336896168,
-      "loss": 46.0,
-      "step": 38467
-    },
-    {
-      "epoch": 2.9411472370357625,
-      "grad_norm": 0.0015376788796857,
-      "learning_rate": 0.00019999573346774117,
-      "loss": 46.0,
-      "step": 38468
-    },
-    {
-      "epoch": 2.9412236940191523,
-      "grad_norm": 0.0030045057646930218,
-      "learning_rate": 0.00019999573324585977,
-      "loss": 46.0,
-      "step": 38469
-    },
-    {
-      "epoch": 2.941300151002542,
-      "grad_norm": 0.002283738926053047,
-      "learning_rate": 0.00019999573302397262,
-      "loss": 46.0,
-      "step": 38470
-    },
-    {
-      "epoch": 2.941376607985932,
-      "grad_norm": 0.0025652190670371056,
-      "learning_rate": 0.00019999573280207966,
-      "loss": 46.0,
-      "step": 38471
-    },
-    {
-      "epoch": 2.9414530649693216,
-      "grad_norm": 0.0009685687837190926,
-      "learning_rate": 0.00019999573258018096,
-      "loss": 46.0,
-      "step": 38472
-    },
-    {
-      "epoch": 2.9415295219527113,
-      "grad_norm": 0.0029314816929399967,
-      "learning_rate": 0.0001999957323582765,
-      "loss": 46.0,
-      "step": 38473
-    },
-    {
-      "epoch": 2.941605978936101,
-      "grad_norm": 0.002371280686929822,
-      "learning_rate": 0.00019999573213636625,
-      "loss": 46.0,
-      "step": 38474
-    },
-    {
-      "epoch": 2.941682435919491,
-      "grad_norm": 0.0008238262962549925,
-      "learning_rate": 0.00019999573191445023,
-      "loss": 46.0,
-      "step": 38475
-    },
-    {
-      "epoch": 2.9417588929028806,
-      "grad_norm": 0.0011935480870306492,
-      "learning_rate": 0.00019999573169252846,
-      "loss": 46.0,
-      "step": 38476
-    },
-    {
-      "epoch": 2.9418353498862704,
-      "grad_norm": 0.0033864155411720276,
-      "learning_rate": 0.0001999957314706009,
-      "loss": 46.0,
-      "step": 38477
-    },
-    {
-      "epoch": 2.94191180686966,
-      "grad_norm": 0.0015145617071539164,
-      "learning_rate": 0.00019999573124866756,
-      "loss": 46.0,
-      "step": 38478
-    },
-    {
-      "epoch": 2.94198826385305,
-      "grad_norm": 0.0013933308655396104,
-      "learning_rate": 0.00019999573102672848,
-      "loss": 46.0,
-      "step": 38479
-    },
-    {
-      "epoch": 2.9420647208364397,
-      "grad_norm": 0.003730396507307887,
-      "learning_rate": 0.00019999573080478364,
-      "loss": 46.0,
-      "step": 38480
-    },
-    {
-      "epoch": 2.942141177819829,
-      "grad_norm": 0.008588864468038082,
-      "learning_rate": 0.000199995730582833,
-      "loss": 46.0,
-      "step": 38481
-    },
-    {
-      "epoch": 2.9422176348032187,
-      "grad_norm": 0.0013294616946950555,
-      "learning_rate": 0.0001999957303608766,
-      "loss": 46.0,
-      "step": 38482
-    },
-    {
-      "epoch": 2.9422940917866085,
-      "grad_norm": 0.001995525322854519,
-      "learning_rate": 0.00019999573013891443,
-      "loss": 46.0,
-      "step": 38483
-    },
-    {
-      "epoch": 2.9423705487699983,
-      "grad_norm": 0.0013809970114380121,
-      "learning_rate": 0.0001999957299169465,
-      "loss": 46.0,
-      "step": 38484
-    },
-    {
-      "epoch": 2.942447005753388,
-      "grad_norm": 0.0013118662172928452,
-      "learning_rate": 0.0001999957296949728,
-      "loss": 46.0,
-      "step": 38485
-    },
-    {
-      "epoch": 2.942523462736778,
-      "grad_norm": 0.0026405975222587585,
-      "learning_rate": 0.0001999957294729933,
-      "loss": 46.0,
-      "step": 38486
-    },
-    {
-      "epoch": 2.9425999197201675,
-      "grad_norm": 0.003563867649063468,
-      "learning_rate": 0.0001999957292510081,
-      "loss": 46.0,
-      "step": 38487
-    },
-    {
-      "epoch": 2.9426763767035573,
-      "grad_norm": 0.0019280824344605207,
-      "learning_rate": 0.00019999572902901707,
-      "loss": 46.0,
-      "step": 38488
-    },
-    {
-      "epoch": 2.9427528336869466,
-      "grad_norm": 0.005512777250260115,
-      "learning_rate": 0.00019999572880702028,
-      "loss": 46.0,
-      "step": 38489
-    },
-    {
-      "epoch": 2.9428292906703364,
-      "grad_norm": 0.000541763671208173,
-      "learning_rate": 0.00019999572858501774,
-      "loss": 46.0,
-      "step": 38490
-    },
-    {
-      "epoch": 2.942905747653726,
-      "grad_norm": 0.0011167492484673858,
-      "learning_rate": 0.0001999957283630094,
-      "loss": 46.0,
-      "step": 38491
-    },
-    {
-      "epoch": 2.942982204637116,
-      "grad_norm": 0.004400759469717741,
-      "learning_rate": 0.00019999572814099535,
-      "loss": 46.0,
-      "step": 38492
-    },
-    {
-      "epoch": 2.9430586616205057,
-      "grad_norm": 0.0036884199362248182,
-      "learning_rate": 0.00019999572791897546,
-      "loss": 46.0,
-      "step": 38493
-    },
-    {
-      "epoch": 2.9431351186038954,
-      "grad_norm": 0.003917299211025238,
-      "learning_rate": 0.00019999572769694983,
-      "loss": 46.0,
-      "step": 38494
-    },
-    {
-      "epoch": 2.943211575587285,
-      "grad_norm": 0.0018146766815334558,
-      "learning_rate": 0.00019999572747491845,
-      "loss": 46.0,
-      "step": 38495
-    },
-    {
-      "epoch": 2.943288032570675,
-      "grad_norm": 0.0054632979445159435,
-      "learning_rate": 0.00019999572725288127,
-      "loss": 46.0,
-      "step": 38496
-    },
-    {
-      "epoch": 2.9433644895540647,
-      "grad_norm": 0.0018193337600678205,
-      "learning_rate": 0.00019999572703083835,
-      "loss": 46.0,
-      "step": 38497
-    },
-    {
-      "epoch": 2.9434409465374545,
-      "grad_norm": 0.0013264549197629094,
-      "learning_rate": 0.00019999572680878965,
-      "loss": 46.0,
-      "step": 38498
-    },
-    {
-      "epoch": 2.9435174035208442,
-      "grad_norm": 0.0009816802339628339,
-      "learning_rate": 0.00019999572658673515,
-      "loss": 46.0,
-      "step": 38499
-    },
-    {
-      "epoch": 2.943593860504234,
-      "grad_norm": 0.0027608880773186684,
-      "learning_rate": 0.0001999957263646749,
-      "loss": 46.0,
-      "step": 38500
-    },
-    {
-      "epoch": 2.9436703174876238,
-      "grad_norm": 0.0017870917217805982,
-      "learning_rate": 0.00019999572614260892,
-      "loss": 46.0,
-      "step": 38501
-    },
-    {
-      "epoch": 2.9437467744710135,
-      "grad_norm": 0.0017494626808911562,
-      "learning_rate": 0.00019999572592053713,
-      "loss": 46.0,
-      "step": 38502
-    },
-    {
-      "epoch": 2.943823231454403,
-      "grad_norm": 0.0010061635402962565,
-      "learning_rate": 0.00019999572569845956,
-      "loss": 46.0,
-      "step": 38503
-    },
-    {
-      "epoch": 2.9438996884377926,
-      "grad_norm": 0.0007403005729429424,
-      "learning_rate": 0.00019999572547637625,
-      "loss": 46.0,
-      "step": 38504
-    },
-    {
-      "epoch": 2.9439761454211824,
-      "grad_norm": 0.0011278523597866297,
-      "learning_rate": 0.0001999957252542872,
-      "loss": 46.0,
-      "step": 38505
-    },
-    {
-      "epoch": 2.944052602404572,
-      "grad_norm": 0.0018596234731376171,
-      "learning_rate": 0.0001999957250321923,
-      "loss": 46.0,
-      "step": 38506
-    },
-    {
-      "epoch": 2.944129059387962,
-      "grad_norm": 0.0012644259259104729,
-      "learning_rate": 0.0001999957248100917,
-      "loss": 46.0,
-      "step": 38507
-    },
-    {
-      "epoch": 2.9442055163713516,
-      "grad_norm": 0.00401754816994071,
-      "learning_rate": 0.00019999572458798527,
-      "loss": 46.0,
-      "step": 38508
-    },
-    {
-      "epoch": 2.9442819733547414,
-      "grad_norm": 0.0028008895460516214,
-      "learning_rate": 0.00019999572436587312,
-      "loss": 46.0,
-      "step": 38509
-    },
-    {
-      "epoch": 2.944358430338131,
-      "grad_norm": 0.004756350070238113,
-      "learning_rate": 0.0001999957241437552,
-      "loss": 46.0,
-      "step": 38510
-    },
-    {
-      "epoch": 2.9444348873215205,
-      "grad_norm": 0.002242323709651828,
-      "learning_rate": 0.00019999572392163147,
-      "loss": 46.0,
-      "step": 38511
-    },
-    {
-      "epoch": 2.9445113443049102,
-      "grad_norm": 0.0017678787698969245,
-      "learning_rate": 0.000199995723699502,
-      "loss": 46.0,
-      "step": 38512
-    },
-    {
-      "epoch": 2.9445878012883,
-      "grad_norm": 0.0014639116125181317,
-      "learning_rate": 0.00019999572347736676,
-      "loss": 46.0,
-      "step": 38513
-    },
-    {
-      "epoch": 2.9446642582716898,
-      "grad_norm": 0.001688948250375688,
-      "learning_rate": 0.00019999572325522577,
-      "loss": 46.0,
-      "step": 38514
-    },
-    {
-      "epoch": 2.9447407152550795,
-      "grad_norm": 0.003764891065657139,
-      "learning_rate": 0.00019999572303307898,
-      "loss": 46.0,
-      "step": 38515
-    },
-    {
-      "epoch": 2.9448171722384693,
-      "grad_norm": 0.0014119968982413411,
-      "learning_rate": 0.0001999957228109264,
-      "loss": 46.0,
-      "step": 38516
-    },
-    {
-      "epoch": 2.944893629221859,
-      "grad_norm": 0.0012477160198614001,
-      "learning_rate": 0.0001999957225887681,
-      "loss": 46.0,
-      "step": 38517
-    },
-    {
-      "epoch": 2.944970086205249,
-      "grad_norm": 0.0018411694327369332,
-      "learning_rate": 0.00019999572236660402,
-      "loss": 46.0,
-      "step": 38518
-    },
-    {
-      "epoch": 2.9450465431886386,
-      "grad_norm": 0.001350630191154778,
-      "learning_rate": 0.00019999572214443416,
-      "loss": 46.0,
-      "step": 38519
-    },
-    {
-      "epoch": 2.9451230001720283,
-      "grad_norm": 0.0016800719313323498,
-      "learning_rate": 0.00019999572192225853,
-      "loss": 46.0,
-      "step": 38520
-    },
-    {
-      "epoch": 2.945199457155418,
-      "grad_norm": 0.0019967847038060427,
-      "learning_rate": 0.00019999572170007715,
-      "loss": 46.0,
-      "step": 38521
-    },
-    {
-      "epoch": 2.945275914138808,
-      "grad_norm": 0.0014051940524950624,
-      "learning_rate": 0.00019999572147788998,
-      "loss": 46.0,
-      "step": 38522
-    },
-    {
-      "epoch": 2.9453523711221976,
-      "grad_norm": 0.002312208293005824,
-      "learning_rate": 0.00019999572125569705,
-      "loss": 46.0,
-      "step": 38523
-    },
-    {
-      "epoch": 2.9454288281055874,
-      "grad_norm": 0.002284690272063017,
-      "learning_rate": 0.00019999572103349835,
-      "loss": 46.0,
-      "step": 38524
-    },
-    {
-      "epoch": 2.9455052850889767,
-      "grad_norm": 0.00094099051784724,
-      "learning_rate": 0.00019999572081129388,
-      "loss": 46.0,
-      "step": 38525
-    },
-    {
-      "epoch": 2.9455817420723664,
-      "grad_norm": 0.0034788523335009813,
-      "learning_rate": 0.0001999957205890836,
-      "loss": 46.0,
-      "step": 38526
-    },
-    {
-      "epoch": 2.945658199055756,
-      "grad_norm": 0.004224858712404966,
-      "learning_rate": 0.0001999957203668676,
-      "loss": 46.0,
-      "step": 38527
-    },
-    {
-      "epoch": 2.945734656039146,
-      "grad_norm": 0.0017389258136972785,
-      "learning_rate": 0.00019999572014464583,
-      "loss": 46.0,
-      "step": 38528
-    },
-    {
-      "epoch": 2.9458111130225357,
-      "grad_norm": 0.00484719080850482,
-      "learning_rate": 0.00019999571992241827,
-      "loss": 46.0,
-      "step": 38529
-    },
-    {
-      "epoch": 2.9458875700059255,
-      "grad_norm": 0.002473007421940565,
-      "learning_rate": 0.00019999571970018496,
-      "loss": 46.0,
-      "step": 38530
-    },
-    {
-      "epoch": 2.9459640269893153,
-      "grad_norm": 0.0019858009181916714,
-      "learning_rate": 0.00019999571947794587,
-      "loss": 46.0,
-      "step": 38531
-    },
-    {
-      "epoch": 2.946040483972705,
-      "grad_norm": 0.0008052597404457629,
-      "learning_rate": 0.00019999571925570102,
-      "loss": 46.0,
-      "step": 38532
-    },
-    {
-      "epoch": 2.9461169409560943,
-      "grad_norm": 0.0025038241874426603,
-      "learning_rate": 0.00019999571903345041,
-      "loss": 46.0,
-      "step": 38533
-    },
-    {
-      "epoch": 2.946193397939484,
-      "grad_norm": 0.002420000033453107,
-      "learning_rate": 0.000199995718811194,
-      "loss": 46.0,
-      "step": 38534
-    },
-    {
-      "epoch": 2.946269854922874,
-      "grad_norm": 0.001423690584488213,
-      "learning_rate": 0.00019999571858893183,
-      "loss": 46.0,
-      "step": 38535
-    },
-    {
-      "epoch": 2.9463463119062636,
-      "grad_norm": 0.0032743343617767096,
-      "learning_rate": 0.0001999957183666639,
-      "loss": 46.0,
-      "step": 38536
-    },
-    {
-      "epoch": 2.9464227688896534,
-      "grad_norm": 0.0007264400483109057,
-      "learning_rate": 0.00019999571814439019,
-      "loss": 46.0,
-      "step": 38537
-    },
-    {
-      "epoch": 2.946499225873043,
-      "grad_norm": 0.0017941446276381612,
-      "learning_rate": 0.00019999571792211072,
-      "loss": 46.0,
-      "step": 38538
-    },
-    {
-      "epoch": 2.946575682856433,
-      "grad_norm": 0.0021681897342205048,
-      "learning_rate": 0.00019999571769982547,
-      "loss": 46.0,
-      "step": 38539
-    },
-    {
-      "epoch": 2.9466521398398227,
-      "grad_norm": 0.00389724038541317,
-      "learning_rate": 0.00019999571747753448,
-      "loss": 46.0,
-      "step": 38540
-    },
-    {
-      "epoch": 2.9467285968232124,
-      "grad_norm": 0.004298449959605932,
-      "learning_rate": 0.0001999957172552377,
-      "loss": 46.0,
-      "step": 38541
-    },
-    {
-      "epoch": 2.946805053806602,
-      "grad_norm": 0.0025112037546932697,
-      "learning_rate": 0.00019999571703293513,
-      "loss": 46.0,
-      "step": 38542
-    },
-    {
-      "epoch": 2.946881510789992,
-      "grad_norm": 0.0011482355184853077,
-      "learning_rate": 0.00019999571681062684,
-      "loss": 46.0,
-      "step": 38543
-    },
-    {
-      "epoch": 2.9469579677733817,
-      "grad_norm": 0.0034956352319568396,
-      "learning_rate": 0.00019999571658831273,
-      "loss": 46.0,
-      "step": 38544
-    },
-    {
-      "epoch": 2.9470344247567715,
-      "grad_norm": 0.00196602800861001,
-      "learning_rate": 0.00019999571636599285,
-      "loss": 46.0,
-      "step": 38545
-    },
-    {
-      "epoch": 2.947110881740161,
-      "grad_norm": 0.0028144672978669405,
-      "learning_rate": 0.00019999571614366722,
-      "loss": 46.0,
-      "step": 38546
-    },
-    {
-      "epoch": 2.9471873387235505,
-      "grad_norm": 0.0031060539186000824,
-      "learning_rate": 0.00019999571592133587,
-      "loss": 46.0,
-      "step": 38547
-    },
-    {
-      "epoch": 2.9472637957069403,
-      "grad_norm": 0.0025489693507552147,
-      "learning_rate": 0.0001999957156989987,
-      "loss": 46.0,
-      "step": 38548
-    },
-    {
-      "epoch": 2.94734025269033,
-      "grad_norm": 0.0026762799825519323,
-      "learning_rate": 0.00019999571547665577,
-      "loss": 46.0,
-      "step": 38549
-    },
-    {
-      "epoch": 2.94741670967372,
-      "grad_norm": 0.002294783014804125,
-      "learning_rate": 0.00019999571525430705,
-      "loss": 46.0,
-      "step": 38550
-    },
-    {
-      "epoch": 2.9474931666571096,
-      "grad_norm": 0.0016876523150131106,
-      "learning_rate": 0.00019999571503195258,
-      "loss": 46.0,
-      "step": 38551
-    },
-    {
-      "epoch": 2.9475696236404993,
-      "grad_norm": 0.004489795304834843,
-      "learning_rate": 0.00019999571480959233,
-      "loss": 46.0,
-      "step": 38552
-    },
-    {
-      "epoch": 2.947646080623889,
-      "grad_norm": 0.0026544597931206226,
-      "learning_rate": 0.00019999571458722632,
-      "loss": 46.0,
-      "step": 38553
-    },
-    {
-      "epoch": 2.947722537607279,
-      "grad_norm": 0.0020726302172988653,
-      "learning_rate": 0.00019999571436485453,
-      "loss": 46.0,
-      "step": 38554
-    },
-    {
-      "epoch": 2.947798994590668,
-      "grad_norm": 0.0037922668270766735,
-      "learning_rate": 0.000199995714142477,
-      "loss": 46.0,
-      "step": 38555
-    },
-    {
-      "epoch": 2.947875451574058,
-      "grad_norm": 0.003436857834458351,
-      "learning_rate": 0.00019999571392009368,
-      "loss": 46.0,
-      "step": 38556
-    },
-    {
-      "epoch": 2.9479519085574477,
-      "grad_norm": 0.004491841420531273,
-      "learning_rate": 0.0001999957136977046,
-      "loss": 46.0,
-      "step": 38557
-    },
-    {
-      "epoch": 2.9480283655408375,
-      "grad_norm": 0.0032609403133392334,
-      "learning_rate": 0.00019999571347530974,
-      "loss": 46.0,
-      "step": 38558
-    },
-    {
-      "epoch": 2.9481048225242272,
-      "grad_norm": 0.005314024630934,
-      "learning_rate": 0.0001999957132529091,
-      "loss": 46.0,
-      "step": 38559
-    },
-    {
-      "epoch": 2.948181279507617,
-      "grad_norm": 0.0013576102210208774,
-      "learning_rate": 0.0001999957130305027,
-      "loss": 46.0,
-      "step": 38560
-    },
-    {
-      "epoch": 2.9482577364910068,
-      "grad_norm": 0.002082610735669732,
-      "learning_rate": 0.00019999571280809056,
-      "loss": 46.0,
-      "step": 38561
-    },
-    {
-      "epoch": 2.9483341934743965,
-      "grad_norm": 0.001264725113287568,
-      "learning_rate": 0.0001999957125856726,
-      "loss": 46.0,
-      "step": 38562
-    },
-    {
-      "epoch": 2.9484106504577863,
-      "grad_norm": 0.0018949240911751986,
-      "learning_rate": 0.0001999957123632489,
-      "loss": 46.0,
-      "step": 38563
-    },
-    {
-      "epoch": 2.948487107441176,
-      "grad_norm": 0.0033848145976662636,
-      "learning_rate": 0.00019999571214081944,
-      "loss": 46.0,
-      "step": 38564
-    },
-    {
-      "epoch": 2.948563564424566,
-      "grad_norm": 0.0008456098730675876,
-      "learning_rate": 0.0001999957119183842,
-      "loss": 46.0,
-      "step": 38565
-    },
-    {
-      "epoch": 2.9486400214079556,
-      "grad_norm": 0.003037538379430771,
-      "learning_rate": 0.00019999571169594318,
-      "loss": 46.0,
-      "step": 38566
-    },
-    {
-      "epoch": 2.9487164783913453,
-      "grad_norm": 0.0009184980881400406,
-      "learning_rate": 0.0001999957114734964,
-      "loss": 46.0,
-      "step": 38567
-    },
-    {
-      "epoch": 2.9487929353747346,
-      "grad_norm": 0.001417288207449019,
-      "learning_rate": 0.00019999571125104386,
-      "loss": 46.0,
-      "step": 38568
-    },
-    {
-      "epoch": 2.9488693923581244,
-      "grad_norm": 0.0021814866922795773,
-      "learning_rate": 0.00019999571102858555,
-      "loss": 46.0,
-      "step": 38569
-    },
-    {
-      "epoch": 2.948945849341514,
-      "grad_norm": 0.0022926959209144115,
-      "learning_rate": 0.00019999571080612144,
-      "loss": 46.0,
-      "step": 38570
-    },
-    {
-      "epoch": 2.949022306324904,
-      "grad_norm": 0.003184255911037326,
-      "learning_rate": 0.00019999571058365158,
-      "loss": 46.0,
-      "step": 38571
-    },
-    {
-      "epoch": 2.9490987633082937,
-      "grad_norm": 0.004917323123663664,
-      "learning_rate": 0.00019999571036117598,
-      "loss": 46.0,
-      "step": 38572
-    },
-    {
-      "epoch": 2.9491752202916834,
-      "grad_norm": 0.0021133313421159983,
-      "learning_rate": 0.00019999571013869458,
-      "loss": 46.0,
-      "step": 38573
-    },
-    {
-      "epoch": 2.949251677275073,
-      "grad_norm": 0.0008648965740576386,
-      "learning_rate": 0.00019999570991620743,
-      "loss": 46.0,
-      "step": 38574
-    },
-    {
-      "epoch": 2.949328134258463,
-      "grad_norm": 0.004948663990944624,
-      "learning_rate": 0.00019999570969371448,
-      "loss": 46.0,
-      "step": 38575
-    },
-    {
-      "epoch": 2.9494045912418523,
-      "grad_norm": 0.002718965755775571,
-      "learning_rate": 0.00019999570947121578,
-      "loss": 46.0,
-      "step": 38576
-    },
-    {
-      "epoch": 2.949481048225242,
-      "grad_norm": 0.0030901196878403425,
-      "learning_rate": 0.0001999957092487113,
-      "loss": 46.0,
-      "step": 38577
-    },
-    {
-      "epoch": 2.949557505208632,
-      "grad_norm": 0.0015326032880693674,
-      "learning_rate": 0.00019999570902620107,
-      "loss": 46.0,
-      "step": 38578
-    },
-    {
-      "epoch": 2.9496339621920216,
-      "grad_norm": 0.002147471997886896,
-      "learning_rate": 0.00019999570880368506,
-      "loss": 46.0,
-      "step": 38579
-    },
-    {
-      "epoch": 2.9497104191754113,
-      "grad_norm": 0.0009024632745422423,
-      "learning_rate": 0.00019999570858116327,
-      "loss": 46.0,
-      "step": 38580
-    },
-    {
-      "epoch": 2.949786876158801,
-      "grad_norm": 0.005958493798971176,
-      "learning_rate": 0.00019999570835863573,
-      "loss": 46.0,
-      "step": 38581
-    },
-    {
-      "epoch": 2.949863333142191,
-      "grad_norm": 0.002807225566357374,
-      "learning_rate": 0.0001999957081361024,
-      "loss": 46.0,
-      "step": 38582
-    },
-    {
-      "epoch": 2.9499397901255806,
-      "grad_norm": 0.004070540424436331,
-      "learning_rate": 0.0001999957079135633,
-      "loss": 46.0,
-      "step": 38583
-    },
-    {
-      "epoch": 2.9500162471089704,
-      "grad_norm": 0.0015553616685792804,
-      "learning_rate": 0.00019999570769101848,
-      "loss": 46.0,
-      "step": 38584
-    },
-    {
-      "epoch": 2.95009270409236,
-      "grad_norm": 0.0036861703265458345,
-      "learning_rate": 0.00019999570746846783,
-      "loss": 46.0,
-      "step": 38585
-    },
-    {
-      "epoch": 2.95016916107575,
-      "grad_norm": 0.002063591033220291,
-      "learning_rate": 0.00019999570724591145,
-      "loss": 46.0,
-      "step": 38586
-    },
-    {
-      "epoch": 2.9502456180591397,
-      "grad_norm": 0.0017155060777440667,
-      "learning_rate": 0.00019999570702334928,
-      "loss": 46.0,
-      "step": 38587
-    },
-    {
-      "epoch": 2.9503220750425294,
-      "grad_norm": 0.0027698769699782133,
-      "learning_rate": 0.00019999570680078135,
-      "loss": 46.0,
-      "step": 38588
-    },
-    {
-      "epoch": 2.950398532025919,
-      "grad_norm": 0.0014660131419077516,
-      "learning_rate": 0.00019999570657820766,
-      "loss": 46.0,
-      "step": 38589
-    },
-    {
-      "epoch": 2.9504749890093085,
-      "grad_norm": 0.0031355747487396,
-      "learning_rate": 0.0001999957063556282,
-      "loss": 46.0,
-      "step": 38590
-    },
-    {
-      "epoch": 2.9505514459926983,
-      "grad_norm": 0.0014126296155154705,
-      "learning_rate": 0.00019999570613304295,
-      "loss": 46.0,
-      "step": 38591
-    },
-    {
-      "epoch": 2.950627902976088,
-      "grad_norm": 0.007596690207719803,
-      "learning_rate": 0.00019999570591045193,
-      "loss": 46.0,
-      "step": 38592
-    },
-    {
-      "epoch": 2.9507043599594778,
-      "grad_norm": 0.001907376921735704,
-      "learning_rate": 0.00019999570568785514,
-      "loss": 46.0,
-      "step": 38593
-    },
-    {
-      "epoch": 2.9507808169428675,
-      "grad_norm": 0.0014813941670581698,
-      "learning_rate": 0.00019999570546525258,
-      "loss": 46.0,
-      "step": 38594
-    },
-    {
-      "epoch": 2.9508572739262573,
-      "grad_norm": 0.001786662032827735,
-      "learning_rate": 0.00019999570524264427,
-      "loss": 46.0,
-      "step": 38595
-    },
-    {
-      "epoch": 2.950933730909647,
-      "grad_norm": 0.004319278988987207,
-      "learning_rate": 0.0001999957050200302,
-      "loss": 46.0,
-      "step": 38596
-    },
-    {
-      "epoch": 2.951010187893037,
-      "grad_norm": 0.001897060894407332,
-      "learning_rate": 0.00019999570479741033,
-      "loss": 46.0,
-      "step": 38597
-    },
-    {
-      "epoch": 2.951086644876426,
-      "grad_norm": 0.0012420077109709382,
-      "learning_rate": 0.0001999957045747847,
-      "loss": 46.0,
-      "step": 38598
-    },
-    {
-      "epoch": 2.951163101859816,
-      "grad_norm": 0.0017158155096694827,
-      "learning_rate": 0.00019999570435215333,
-      "loss": 46.0,
-      "step": 38599
-    },
-    {
-      "epoch": 2.9512395588432057,
-      "grad_norm": 0.0032330851536244154,
-      "learning_rate": 0.00019999570412951615,
-      "loss": 46.0,
-      "step": 38600
-    },
-    {
-      "epoch": 2.9513160158265954,
-      "grad_norm": 0.0010303683811798692,
-      "learning_rate": 0.0001999957039068732,
-      "loss": 46.0,
-      "step": 38601
-    },
-    {
-      "epoch": 2.951392472809985,
-      "grad_norm": 0.004474753979593515,
-      "learning_rate": 0.0001999957036842245,
-      "loss": 46.0,
-      "step": 38602
-    },
-    {
-      "epoch": 2.951468929793375,
-      "grad_norm": 0.0023435058537870646,
-      "learning_rate": 0.00019999570346157004,
-      "loss": 46.0,
-      "step": 38603
-    },
-    {
-      "epoch": 2.9515453867767647,
-      "grad_norm": 0.003405747702345252,
-      "learning_rate": 0.00019999570323890983,
-      "loss": 46.0,
-      "step": 38604
-    },
-    {
-      "epoch": 2.9516218437601545,
-      "grad_norm": 0.0022501919884234667,
-      "learning_rate": 0.0001999957030162438,
-      "loss": 46.0,
-      "step": 38605
-    },
-    {
-      "epoch": 2.9516983007435442,
-      "grad_norm": 0.0011424679541960359,
-      "learning_rate": 0.00019999570279357202,
-      "loss": 46.0,
-      "step": 38606
-    },
-    {
-      "epoch": 2.951774757726934,
-      "grad_norm": 0.0010225395672023296,
-      "learning_rate": 0.00019999570257089449,
-      "loss": 46.0,
-      "step": 38607
-    },
-    {
-      "epoch": 2.9518512147103237,
-      "grad_norm": 0.0020619910210371017,
-      "learning_rate": 0.00019999570234821118,
-      "loss": 46.0,
-      "step": 38608
-    },
-    {
-      "epoch": 2.9519276716937135,
-      "grad_norm": 0.0071455081924796104,
-      "learning_rate": 0.00019999570212552207,
-      "loss": 46.0,
-      "step": 38609
-    },
-    {
-      "epoch": 2.9520041286771033,
-      "grad_norm": 0.0005961196729913354,
-      "learning_rate": 0.00019999570190282721,
-      "loss": 46.0,
-      "step": 38610
-    },
-    {
-      "epoch": 2.952080585660493,
-      "grad_norm": 0.0016357485437765718,
-      "learning_rate": 0.0001999957016801266,
-      "loss": 46.0,
-      "step": 38611
-    },
-    {
-      "epoch": 2.9521570426438823,
-      "grad_norm": 0.0024794924538582563,
-      "learning_rate": 0.0001999957014574202,
-      "loss": 46.0,
-      "step": 38612
-    },
-    {
-      "epoch": 2.952233499627272,
-      "grad_norm": 0.004225259646773338,
-      "learning_rate": 0.00019999570123470804,
-      "loss": 46.0,
-      "step": 38613
-    },
-    {
-      "epoch": 2.952309956610662,
-      "grad_norm": 0.0036041198763996363,
-      "learning_rate": 0.0001999957010119901,
-      "loss": 46.0,
-      "step": 38614
-    },
-    {
-      "epoch": 2.9523864135940516,
-      "grad_norm": 0.0029007666744291782,
-      "learning_rate": 0.00019999570078926642,
-      "loss": 46.0,
-      "step": 38615
-    },
-    {
-      "epoch": 2.9524628705774414,
-      "grad_norm": 0.000899720354937017,
-      "learning_rate": 0.00019999570056653692,
-      "loss": 46.0,
-      "step": 38616
-    },
-    {
-      "epoch": 2.952539327560831,
-      "grad_norm": 0.002860851353034377,
-      "learning_rate": 0.00019999570034380168,
-      "loss": 46.0,
-      "step": 38617
-    },
-    {
-      "epoch": 2.952615784544221,
-      "grad_norm": 0.004036292899399996,
-      "learning_rate": 0.0001999957001210607,
-      "loss": 46.0,
-      "step": 38618
-    },
-    {
-      "epoch": 2.9526922415276107,
-      "grad_norm": 0.0024592482950538397,
-      "learning_rate": 0.0001999956998983139,
-      "loss": 46.0,
-      "step": 38619
-    },
-    {
-      "epoch": 2.952768698511,
-      "grad_norm": 0.0029361334163695574,
-      "learning_rate": 0.00019999569967556134,
-      "loss": 46.0,
-      "step": 38620
-    },
-    {
-      "epoch": 2.9528451554943898,
-      "grad_norm": 0.0014306538505479693,
-      "learning_rate": 0.00019999569945280303,
-      "loss": 46.0,
-      "step": 38621
-    },
-    {
-      "epoch": 2.9529216124777795,
-      "grad_norm": 0.004527061711996794,
-      "learning_rate": 0.00019999569923003895,
-      "loss": 46.0,
-      "step": 38622
-    },
-    {
-      "epoch": 2.9529980694611693,
-      "grad_norm": 0.0012273305328562856,
-      "learning_rate": 0.0001999956990072691,
-      "loss": 46.0,
-      "step": 38623
-    },
-    {
-      "epoch": 2.953074526444559,
-      "grad_norm": 0.004408834036439657,
-      "learning_rate": 0.0001999956987844935,
-      "loss": 46.0,
-      "step": 38624
-    },
-    {
-      "epoch": 2.953150983427949,
-      "grad_norm": 0.002043339656665921,
-      "learning_rate": 0.0001999956985617121,
-      "loss": 46.0,
-      "step": 38625
-    },
-    {
-      "epoch": 2.9532274404113386,
-      "grad_norm": 0.002869096351787448,
-      "learning_rate": 0.00019999569833892492,
-      "loss": 46.0,
-      "step": 38626
-    },
-    {
-      "epoch": 2.9533038973947283,
-      "grad_norm": 0.0034318435937166214,
-      "learning_rate": 0.000199995698116132,
-      "loss": 46.0,
-      "step": 38627
-    },
-    {
-      "epoch": 2.953380354378118,
-      "grad_norm": 0.0022060545161366463,
-      "learning_rate": 0.00019999569789333328,
-      "loss": 46.0,
-      "step": 38628
-    },
-    {
-      "epoch": 2.953456811361508,
-      "grad_norm": 0.003766795387491584,
-      "learning_rate": 0.0001999956976705288,
-      "loss": 46.0,
-      "step": 38629
-    },
-    {
-      "epoch": 2.9535332683448976,
-      "grad_norm": 0.0034128769766539335,
-      "learning_rate": 0.00019999569744771857,
-      "loss": 46.0,
-      "step": 38630
-    },
-    {
-      "epoch": 2.9536097253282874,
-      "grad_norm": 0.0008646061760373414,
-      "learning_rate": 0.00019999569722490258,
-      "loss": 46.0,
-      "step": 38631
-    },
-    {
-      "epoch": 2.953686182311677,
-      "grad_norm": 0.0017679266165941954,
-      "learning_rate": 0.0001999956970020808,
-      "loss": 46.0,
-      "step": 38632
-    },
-    {
-      "epoch": 2.953762639295067,
-      "grad_norm": 0.006460228003561497,
-      "learning_rate": 0.00019999569677925326,
-      "loss": 46.0,
-      "step": 38633
-    },
-    {
-      "epoch": 2.953839096278456,
-      "grad_norm": 0.0033067085314542055,
-      "learning_rate": 0.00019999569655641995,
-      "loss": 46.0,
-      "step": 38634
-    },
-    {
-      "epoch": 2.953915553261846,
-      "grad_norm": 0.0007943941163830459,
-      "learning_rate": 0.00019999569633358084,
-      "loss": 46.0,
-      "step": 38635
-    },
-    {
-      "epoch": 2.9539920102452357,
-      "grad_norm": 0.005821706727147102,
-      "learning_rate": 0.000199995696110736,
-      "loss": 46.0,
-      "step": 38636
-    },
-    {
-      "epoch": 2.9540684672286255,
-      "grad_norm": 0.0017797807231545448,
-      "learning_rate": 0.00019999569588788536,
-      "loss": 46.0,
-      "step": 38637
-    },
-    {
-      "epoch": 2.9541449242120152,
-      "grad_norm": 0.001619631191715598,
-      "learning_rate": 0.00019999569566502899,
-      "loss": 46.0,
-      "step": 38638
-    },
-    {
-      "epoch": 2.954221381195405,
-      "grad_norm": 0.004914996214210987,
-      "learning_rate": 0.0001999956954421668,
-      "loss": 46.0,
-      "step": 38639
-    },
-    {
-      "epoch": 2.9542978381787948,
-      "grad_norm": 0.0014280570903792977,
-      "learning_rate": 0.00019999569521929886,
-      "loss": 46.0,
-      "step": 38640
-    },
-    {
-      "epoch": 2.9543742951621845,
-      "grad_norm": 0.004295347724109888,
-      "learning_rate": 0.0001999956949964252,
-      "loss": 46.0,
-      "step": 38641
-    },
-    {
-      "epoch": 2.954450752145574,
-      "grad_norm": 0.002865126123651862,
-      "learning_rate": 0.0001999956947735457,
-      "loss": 46.0,
-      "step": 38642
-    },
-    {
-      "epoch": 2.9545272091289636,
-      "grad_norm": 0.002506297081708908,
-      "learning_rate": 0.00019999569455066046,
-      "loss": 46.0,
-      "step": 38643
-    },
-    {
-      "epoch": 2.9546036661123534,
-      "grad_norm": 0.0028410139493644238,
-      "learning_rate": 0.00019999569432776945,
-      "loss": 46.0,
-      "step": 38644
-    },
-    {
-      "epoch": 2.954680123095743,
-      "grad_norm": 0.0013299144338816404,
-      "learning_rate": 0.00019999569410487266,
-      "loss": 46.0,
-      "step": 38645
-    },
-    {
-      "epoch": 2.954756580079133,
-      "grad_norm": 0.004807759542018175,
-      "learning_rate": 0.00019999569388197012,
-      "loss": 46.0,
-      "step": 38646
-    },
-    {
-      "epoch": 2.9548330370625226,
-      "grad_norm": 0.003323440905660391,
-      "learning_rate": 0.00019999569365906182,
-      "loss": 46.0,
-      "step": 38647
-    },
-    {
-      "epoch": 2.9549094940459124,
-      "grad_norm": 0.00833585113286972,
-      "learning_rate": 0.00019999569343614773,
-      "loss": 46.0,
-      "step": 38648
-    },
-    {
-      "epoch": 2.954985951029302,
-      "grad_norm": 0.0031237269286066294,
-      "learning_rate": 0.00019999569321322785,
-      "loss": 46.0,
-      "step": 38649
-    },
-    {
-      "epoch": 2.955062408012692,
-      "grad_norm": 0.001065508578903973,
-      "learning_rate": 0.00019999569299030225,
-      "loss": 46.0,
-      "step": 38650
-    },
-    {
-      "epoch": 2.9551388649960817,
-      "grad_norm": 0.004416194278746843,
-      "learning_rate": 0.00019999569276737088,
-      "loss": 46.0,
-      "step": 38651
-    },
-    {
-      "epoch": 2.9552153219794715,
-      "grad_norm": 0.0031110697891563177,
-      "learning_rate": 0.00019999569254443368,
-      "loss": 46.0,
-      "step": 38652
-    },
-    {
-      "epoch": 2.955291778962861,
-      "grad_norm": 0.0026335616130381823,
-      "learning_rate": 0.00019999569232149078,
-      "loss": 46.0,
-      "step": 38653
-    },
-    {
-      "epoch": 2.955368235946251,
-      "grad_norm": 0.0008569032652303576,
-      "learning_rate": 0.00019999569209854206,
-      "loss": 46.0,
-      "step": 38654
-    },
-    {
-      "epoch": 2.9554446929296407,
-      "grad_norm": 0.0017517261439934373,
-      "learning_rate": 0.0001999956918755876,
-      "loss": 46.0,
-      "step": 38655
-    },
-    {
-      "epoch": 2.95552114991303,
-      "grad_norm": 0.0053923544473946095,
-      "learning_rate": 0.00019999569165262738,
-      "loss": 46.0,
-      "step": 38656
-    },
-    {
-      "epoch": 2.95559760689642,
-      "grad_norm": 0.0012485375627875328,
-      "learning_rate": 0.00019999569142966134,
-      "loss": 46.0,
-      "step": 38657
-    },
-    {
-      "epoch": 2.9556740638798096,
-      "grad_norm": 0.003949600737541914,
-      "learning_rate": 0.00019999569120668958,
-      "loss": 46.0,
-      "step": 38658
-    },
-    {
-      "epoch": 2.9557505208631993,
-      "grad_norm": 0.0009021171717904508,
-      "learning_rate": 0.00019999569098371202,
-      "loss": 46.0,
-      "step": 38659
-    },
-    {
-      "epoch": 2.955826977846589,
-      "grad_norm": 0.0034618934150785208,
-      "learning_rate": 0.0001999956907607287,
-      "loss": 46.0,
-      "step": 38660
-    },
-    {
-      "epoch": 2.955903434829979,
-      "grad_norm": 0.002640408230945468,
-      "learning_rate": 0.00019999569053773963,
-      "loss": 46.0,
-      "step": 38661
-    },
-    {
-      "epoch": 2.9559798918133686,
-      "grad_norm": 0.0011499692918732762,
-      "learning_rate": 0.00019999569031474478,
-      "loss": 46.0,
-      "step": 38662
-    },
-    {
-      "epoch": 2.9560563487967584,
-      "grad_norm": 0.0034911585971713066,
-      "learning_rate": 0.00019999569009174415,
-      "loss": 46.0,
-      "step": 38663
-    },
-    {
-      "epoch": 2.9561328057801477,
-      "grad_norm": 0.001569836982525885,
-      "learning_rate": 0.00019999568986873775,
-      "loss": 46.0,
-      "step": 38664
-    },
-    {
-      "epoch": 2.9562092627635375,
-      "grad_norm": 0.0017723095370456576,
-      "learning_rate": 0.0001999956896457256,
-      "loss": 46.0,
-      "step": 38665
-    },
-    {
-      "epoch": 2.956285719746927,
-      "grad_norm": 0.0033342435490339994,
-      "learning_rate": 0.00019999568942270766,
-      "loss": 46.0,
-      "step": 38666
-    },
-    {
-      "epoch": 2.956362176730317,
-      "grad_norm": 0.0023418692871928215,
-      "learning_rate": 0.00019999568919968396,
-      "loss": 46.0,
-      "step": 38667
-    },
-    {
-      "epoch": 2.9564386337137067,
-      "grad_norm": 0.0022423886694014072,
-      "learning_rate": 0.0001999956889766545,
-      "loss": 46.0,
-      "step": 38668
-    },
-    {
-      "epoch": 2.9565150906970965,
-      "grad_norm": 0.002424341393634677,
-      "learning_rate": 0.00019999568875361925,
-      "loss": 46.0,
-      "step": 38669
-    },
-    {
-      "epoch": 2.9565915476804863,
-      "grad_norm": 0.00486412039026618,
-      "learning_rate": 0.00019999568853057824,
-      "loss": 46.0,
-      "step": 38670
-    },
-    {
-      "epoch": 2.956668004663876,
-      "grad_norm": 0.002259011147543788,
-      "learning_rate": 0.00019999568830753148,
-      "loss": 46.0,
-      "step": 38671
-    },
-    {
-      "epoch": 2.956744461647266,
-      "grad_norm": 0.0011656327405944467,
-      "learning_rate": 0.00019999568808447892,
-      "loss": 46.0,
-      "step": 38672
-    },
-    {
-      "epoch": 2.9568209186306555,
-      "grad_norm": 0.001422735396772623,
-      "learning_rate": 0.0001999956878614206,
-      "loss": 46.0,
-      "step": 38673
-    },
-    {
-      "epoch": 2.9568973756140453,
-      "grad_norm": 0.003281527431681752,
-      "learning_rate": 0.00019999568763835653,
-      "loss": 46.0,
-      "step": 38674
-    },
-    {
-      "epoch": 2.956973832597435,
-      "grad_norm": 0.005856127943843603,
-      "learning_rate": 0.00019999568741528668,
-      "loss": 46.0,
-      "step": 38675
-    },
-    {
-      "epoch": 2.957050289580825,
-      "grad_norm": 0.0014301514020189643,
-      "learning_rate": 0.00019999568719221105,
-      "loss": 46.0,
-      "step": 38676
-    },
-    {
-      "epoch": 2.957126746564214,
-      "grad_norm": 0.0009442302980460227,
-      "learning_rate": 0.00019999568696912965,
-      "loss": 46.0,
-      "step": 38677
-    },
-    {
-      "epoch": 2.957203203547604,
-      "grad_norm": 0.002449366496875882,
-      "learning_rate": 0.0001999956867460425,
-      "loss": 46.0,
-      "step": 38678
-    },
-    {
-      "epoch": 2.9572796605309937,
-      "grad_norm": 0.0007358395378105342,
-      "learning_rate": 0.00019999568652294959,
-      "loss": 46.0,
-      "step": 38679
-    },
-    {
-      "epoch": 2.9573561175143834,
-      "grad_norm": 0.0011241902830079198,
-      "learning_rate": 0.00019999568629985086,
-      "loss": 46.0,
-      "step": 38680
-    },
-    {
-      "epoch": 2.957432574497773,
-      "grad_norm": 0.0013472603168338537,
-      "learning_rate": 0.0001999956860767464,
-      "loss": 46.0,
-      "step": 38681
-    },
-    {
-      "epoch": 2.957509031481163,
-      "grad_norm": 0.0019767757039517164,
-      "learning_rate": 0.00019999568585363616,
-      "loss": 46.0,
-      "step": 38682
-    },
-    {
-      "epoch": 2.9575854884645527,
-      "grad_norm": 0.003985450603067875,
-      "learning_rate": 0.00019999568563052014,
-      "loss": 46.0,
-      "step": 38683
-    },
-    {
-      "epoch": 2.9576619454479425,
-      "grad_norm": 0.0027131284587085247,
-      "learning_rate": 0.00019999568540739836,
-      "loss": 46.0,
-      "step": 38684
-    },
-    {
-      "epoch": 2.9577384024313322,
-      "grad_norm": 0.0007529185386374593,
-      "learning_rate": 0.0001999956851842708,
-      "loss": 46.0,
-      "step": 38685
-    },
-    {
-      "epoch": 2.9578148594147216,
-      "grad_norm": 0.0015757158398628235,
-      "learning_rate": 0.00019999568496113752,
-      "loss": 46.0,
-      "step": 38686
-    },
-    {
-      "epoch": 2.9578913163981113,
-      "grad_norm": 0.0020403778180480003,
-      "learning_rate": 0.0001999956847379984,
-      "loss": 46.0,
-      "step": 38687
-    },
-    {
-      "epoch": 2.957967773381501,
-      "grad_norm": 0.001927541452459991,
-      "learning_rate": 0.00019999568451485358,
-      "loss": 46.0,
-      "step": 38688
-    },
-    {
-      "epoch": 2.958044230364891,
-      "grad_norm": 0.0016172969480976462,
-      "learning_rate": 0.00019999568429170296,
-      "loss": 46.0,
-      "step": 38689
-    },
-    {
-      "epoch": 2.9581206873482806,
-      "grad_norm": 0.0034206684213131666,
-      "learning_rate": 0.00019999568406854653,
-      "loss": 46.0,
-      "step": 38690
-    },
-    {
-      "epoch": 2.9581971443316704,
-      "grad_norm": 0.003295660950243473,
-      "learning_rate": 0.00019999568384538438,
-      "loss": 46.0,
-      "step": 38691
-    },
-    {
-      "epoch": 2.95827360131506,
-      "grad_norm": 0.0006767706945538521,
-      "learning_rate": 0.00019999568362221646,
-      "loss": 46.0,
-      "step": 38692
-    },
-    {
-      "epoch": 2.95835005829845,
-      "grad_norm": 0.00301443412899971,
-      "learning_rate": 0.00019999568339904274,
-      "loss": 46.0,
-      "step": 38693
-    },
-    {
-      "epoch": 2.9584265152818396,
-      "grad_norm": 0.0030937818810343742,
-      "learning_rate": 0.00019999568317586328,
-      "loss": 46.0,
-      "step": 38694
-    },
-    {
-      "epoch": 2.9585029722652294,
-      "grad_norm": 0.0016957969637587667,
-      "learning_rate": 0.00019999568295267807,
-      "loss": 46.0,
-      "step": 38695
-    },
-    {
-      "epoch": 2.958579429248619,
-      "grad_norm": 0.001401284127496183,
-      "learning_rate": 0.00019999568272948705,
-      "loss": 46.0,
-      "step": 38696
-    },
-    {
-      "epoch": 2.958655886232009,
-      "grad_norm": 0.0027776320930570364,
-      "learning_rate": 0.00019999568250629027,
-      "loss": 46.0,
-      "step": 38697
-    },
-    {
-      "epoch": 2.9587323432153987,
-      "grad_norm": 0.0032301959581673145,
-      "learning_rate": 0.00019999568228308773,
-      "loss": 46.0,
-      "step": 38698
-    },
-    {
-      "epoch": 2.958808800198788,
-      "grad_norm": 0.0037341329734772444,
-      "learning_rate": 0.00019999568205987943,
-      "loss": 46.0,
-      "step": 38699
-    },
-    {
-      "epoch": 2.9588852571821778,
-      "grad_norm": 0.0018835713854059577,
-      "learning_rate": 0.00019999568183666532,
-      "loss": 46.0,
-      "step": 38700
-    },
-    {
-      "epoch": 2.9589617141655675,
-      "grad_norm": 0.00538292434066534,
-      "learning_rate": 0.0001999956816134455,
-      "loss": 46.0,
-      "step": 38701
-    },
-    {
-      "epoch": 2.9590381711489573,
-      "grad_norm": 0.0025257535744458437,
-      "learning_rate": 0.00019999568139021987,
-      "loss": 46.0,
-      "step": 38702
-    },
-    {
-      "epoch": 2.959114628132347,
-      "grad_norm": 0.0016155935591086745,
-      "learning_rate": 0.00019999568116698847,
-      "loss": 46.0,
-      "step": 38703
-    },
-    {
-      "epoch": 2.959191085115737,
-      "grad_norm": 0.004132924135774374,
-      "learning_rate": 0.00019999568094375132,
-      "loss": 46.0,
-      "step": 38704
-    },
-    {
-      "epoch": 2.9592675420991266,
-      "grad_norm": 0.0028732081409543753,
-      "learning_rate": 0.00019999568072050838,
-      "loss": 46.0,
-      "step": 38705
-    },
-    {
-      "epoch": 2.9593439990825163,
-      "grad_norm": 0.0037886768113821745,
-      "learning_rate": 0.00019999568049725968,
-      "loss": 46.0,
-      "step": 38706
-    },
-    {
-      "epoch": 2.9594204560659056,
-      "grad_norm": 0.002697874093428254,
-      "learning_rate": 0.0001999956802740052,
-      "loss": 46.0,
-      "step": 38707
-    },
-    {
-      "epoch": 2.9594969130492954,
-      "grad_norm": 0.005122511647641659,
-      "learning_rate": 0.00019999568005074498,
-      "loss": 46.0,
-      "step": 38708
-    },
-    {
-      "epoch": 2.959573370032685,
-      "grad_norm": 0.006092989817261696,
-      "learning_rate": 0.00019999567982747896,
-      "loss": 46.0,
-      "step": 38709
-    },
-    {
-      "epoch": 2.959649827016075,
-      "grad_norm": 0.005526279099285603,
-      "learning_rate": 0.0001999956796042072,
-      "loss": 46.0,
-      "step": 38710
-    },
-    {
-      "epoch": 2.9597262839994647,
-      "grad_norm": 0.005327354650944471,
-      "learning_rate": 0.00019999567938092962,
-      "loss": 46.0,
-      "step": 38711
-    },
-    {
-      "epoch": 2.9598027409828545,
-      "grad_norm": 0.0024364099372178316,
-      "learning_rate": 0.0001999956791576463,
-      "loss": 46.0,
-      "step": 38712
-    },
-    {
-      "epoch": 2.959879197966244,
-      "grad_norm": 0.0010068166302517056,
-      "learning_rate": 0.00019999567893435723,
-      "loss": 46.0,
-      "step": 38713
-    },
-    {
-      "epoch": 2.959955654949634,
-      "grad_norm": 0.008561637252569199,
-      "learning_rate": 0.0001999956787110624,
-      "loss": 46.0,
-      "step": 38714
-    },
-    {
-      "epoch": 2.9600321119330237,
-      "grad_norm": 0.0012215106980875134,
-      "learning_rate": 0.00019999567848776178,
-      "loss": 46.0,
-      "step": 38715
-    },
-    {
-      "epoch": 2.9601085689164135,
-      "grad_norm": 0.001945184194482863,
-      "learning_rate": 0.00019999567826445538,
-      "loss": 46.0,
-      "step": 38716
-    },
-    {
-      "epoch": 2.9601850258998033,
-      "grad_norm": 0.0041485317051410675,
-      "learning_rate": 0.0001999956780411432,
-      "loss": 46.0,
-      "step": 38717
-    },
-    {
-      "epoch": 2.960261482883193,
-      "grad_norm": 0.0019014314748346806,
-      "learning_rate": 0.0001999956778178253,
-      "loss": 46.0,
-      "step": 38718
-    },
-    {
-      "epoch": 2.960337939866583,
-      "grad_norm": 0.0023423656821250916,
-      "learning_rate": 0.0001999956775945016,
-      "loss": 46.0,
-      "step": 38719
-    },
-    {
-      "epoch": 2.9604143968499725,
-      "grad_norm": 0.0026675541885197163,
-      "learning_rate": 0.0001999956773711721,
-      "loss": 46.0,
-      "step": 38720
-    },
-    {
-      "epoch": 2.960490853833362,
-      "grad_norm": 0.001858965028077364,
-      "learning_rate": 0.0001999956771478369,
-      "loss": 46.0,
-      "step": 38721
-    },
-    {
-      "epoch": 2.9605673108167516,
-      "grad_norm": 0.0016284739831462502,
-      "learning_rate": 0.00019999567692449588,
-      "loss": 46.0,
-      "step": 38722
-    },
-    {
-      "epoch": 2.9606437678001414,
-      "grad_norm": 0.0014827321283519268,
-      "learning_rate": 0.0001999956767011491,
-      "loss": 46.0,
-      "step": 38723
-    },
-    {
-      "epoch": 2.960720224783531,
-      "grad_norm": 0.0024489688221365213,
-      "learning_rate": 0.00019999567647779656,
-      "loss": 46.0,
-      "step": 38724
-    },
-    {
-      "epoch": 2.960796681766921,
-      "grad_norm": 0.003848556661978364,
-      "learning_rate": 0.00019999567625443826,
-      "loss": 46.0,
-      "step": 38725
-    },
-    {
-      "epoch": 2.9608731387503107,
-      "grad_norm": 0.0014507209416478872,
-      "learning_rate": 0.00019999567603107418,
-      "loss": 46.0,
-      "step": 38726
-    },
-    {
-      "epoch": 2.9609495957337004,
-      "grad_norm": 0.0029634302482008934,
-      "learning_rate": 0.0001999956758077043,
-      "loss": 46.0,
-      "step": 38727
-    },
-    {
-      "epoch": 2.96102605271709,
-      "grad_norm": 0.0006416336400434375,
-      "learning_rate": 0.00019999567558432867,
-      "loss": 46.0,
-      "step": 38728
-    },
-    {
-      "epoch": 2.9611025097004795,
-      "grad_norm": 0.0008342862129211426,
-      "learning_rate": 0.00019999567536094727,
-      "loss": 46.0,
-      "step": 38729
-    },
-    {
-      "epoch": 2.9611789666838693,
-      "grad_norm": 0.0015737359644845128,
-      "learning_rate": 0.00019999567513756015,
-      "loss": 46.0,
-      "step": 38730
-    },
-    {
-      "epoch": 2.961255423667259,
-      "grad_norm": 0.0036144196055829525,
-      "learning_rate": 0.0001999956749141672,
-      "loss": 46.0,
-      "step": 38731
-    },
-    {
-      "epoch": 2.961331880650649,
-      "grad_norm": 0.0028963545337319374,
-      "learning_rate": 0.00019999567469076852,
-      "loss": 46.0,
-      "step": 38732
-    },
-    {
-      "epoch": 2.9614083376340385,
-      "grad_norm": 0.002600293140858412,
-      "learning_rate": 0.00019999567446736402,
-      "loss": 46.0,
-      "step": 38733
-    },
-    {
-      "epoch": 2.9614847946174283,
-      "grad_norm": 0.0018781140679493546,
-      "learning_rate": 0.0001999956742439538,
-      "loss": 46.0,
-      "step": 38734
-    },
-    {
-      "epoch": 2.961561251600818,
-      "grad_norm": 0.003751951502636075,
-      "learning_rate": 0.0001999956740205378,
-      "loss": 46.0,
-      "step": 38735
-    },
-    {
-      "epoch": 2.961637708584208,
-      "grad_norm": 0.003273369511589408,
-      "learning_rate": 0.00019999567379711601,
-      "loss": 46.0,
-      "step": 38736
-    },
-    {
-      "epoch": 2.9617141655675976,
-      "grad_norm": 0.0011937698582187295,
-      "learning_rate": 0.00019999567357368846,
-      "loss": 46.0,
-      "step": 38737
-    },
-    {
-      "epoch": 2.9617906225509874,
-      "grad_norm": 0.005577050149440765,
-      "learning_rate": 0.00019999567335025518,
-      "loss": 46.0,
-      "step": 38738
-    },
-    {
-      "epoch": 2.961867079534377,
-      "grad_norm": 0.0025489041581749916,
-      "learning_rate": 0.0001999956731268161,
-      "loss": 46.0,
-      "step": 38739
-    },
-    {
-      "epoch": 2.961943536517767,
-      "grad_norm": 0.0004022567009087652,
-      "learning_rate": 0.00019999567290337122,
-      "loss": 46.0,
-      "step": 38740
-    },
-    {
-      "epoch": 2.9620199935011566,
-      "grad_norm": 0.0006471026572398841,
-      "learning_rate": 0.0001999956726799206,
-      "loss": 46.0,
-      "step": 38741
-    },
-    {
-      "epoch": 2.9620964504845464,
-      "grad_norm": 0.0005794462049379945,
-      "learning_rate": 0.00019999567245646422,
-      "loss": 46.0,
-      "step": 38742
-    },
-    {
-      "epoch": 2.9621729074679357,
-      "grad_norm": 0.003686587093397975,
-      "learning_rate": 0.00019999567223300205,
-      "loss": 46.0,
-      "step": 38743
-    },
-    {
-      "epoch": 2.9622493644513255,
-      "grad_norm": 0.0008808143902570009,
-      "learning_rate": 0.00019999567200953413,
-      "loss": 46.0,
-      "step": 38744
-    },
-    {
-      "epoch": 2.9623258214347152,
-      "grad_norm": 0.0007387800142168999,
-      "learning_rate": 0.0001999956717860604,
-      "loss": 46.0,
-      "step": 38745
-    },
-    {
-      "epoch": 2.962402278418105,
-      "grad_norm": 0.0026468741707503796,
-      "learning_rate": 0.00019999567156258095,
-      "loss": 46.0,
-      "step": 38746
-    },
-    {
-      "epoch": 2.9624787354014948,
-      "grad_norm": 0.003015848807990551,
-      "learning_rate": 0.0001999956713390957,
-      "loss": 46.0,
-      "step": 38747
-    },
-    {
-      "epoch": 2.9625551923848845,
-      "grad_norm": 0.0035268678329885006,
-      "learning_rate": 0.0001999956711156047,
-      "loss": 46.0,
-      "step": 38748
-    },
-    {
-      "epoch": 2.9626316493682743,
-      "grad_norm": 0.0023516786750406027,
-      "learning_rate": 0.00019999567089210794,
-      "loss": 46.0,
-      "step": 38749
-    },
-    {
-      "epoch": 2.962708106351664,
-      "grad_norm": 0.001929847989231348,
-      "learning_rate": 0.00019999567066860538,
-      "loss": 46.0,
-      "step": 38750
-    },
-    {
-      "epoch": 2.9627845633350534,
-      "grad_norm": 0.003065502969548106,
-      "learning_rate": 0.00019999567044509708,
-      "loss": 46.0,
-      "step": 38751
-    },
-    {
-      "epoch": 2.962861020318443,
-      "grad_norm": 0.0014847610145807266,
-      "learning_rate": 0.000199995670221583,
-      "loss": 46.0,
-      "step": 38752
-    },
-    {
-      "epoch": 2.962937477301833,
-      "grad_norm": 0.007171249948441982,
-      "learning_rate": 0.00019999566999806315,
-      "loss": 46.0,
-      "step": 38753
-    },
-    {
-      "epoch": 2.9630139342852226,
-      "grad_norm": 0.0041608791798353195,
-      "learning_rate": 0.00019999566977453755,
-      "loss": 46.0,
-      "step": 38754
-    },
-    {
-      "epoch": 2.9630903912686124,
-      "grad_norm": 0.0039712004363536835,
-      "learning_rate": 0.00019999566955100612,
-      "loss": 46.0,
-      "step": 38755
-    },
-    {
-      "epoch": 2.963166848252002,
-      "grad_norm": 0.0007582994876429439,
-      "learning_rate": 0.00019999566932746898,
-      "loss": 46.0,
-      "step": 38756
-    },
-    {
-      "epoch": 2.963243305235392,
-      "grad_norm": 0.0007962352246977389,
-      "learning_rate": 0.00019999566910392606,
-      "loss": 46.0,
-      "step": 38757
-    },
-    {
-      "epoch": 2.9633197622187817,
-      "grad_norm": 0.004281751345843077,
-      "learning_rate": 0.00019999566888037734,
-      "loss": 46.0,
-      "step": 38758
-    },
-    {
-      "epoch": 2.9633962192021714,
-      "grad_norm": 0.002316530328243971,
-      "learning_rate": 0.00019999566865682288,
-      "loss": 46.0,
-      "step": 38759
-    },
-    {
-      "epoch": 2.963472676185561,
-      "grad_norm": 0.001087358221411705,
-      "learning_rate": 0.00019999566843326264,
-      "loss": 46.0,
-      "step": 38760
-    },
-    {
-      "epoch": 2.963549133168951,
-      "grad_norm": 0.0012984341010451317,
-      "learning_rate": 0.00019999566820969665,
-      "loss": 46.0,
-      "step": 38761
-    },
-    {
-      "epoch": 2.9636255901523407,
-      "grad_norm": 0.004118119366466999,
-      "learning_rate": 0.00019999566798612487,
-      "loss": 46.0,
-      "step": 38762
-    },
-    {
-      "epoch": 2.9637020471357305,
-      "grad_norm": 0.003220656653866172,
-      "learning_rate": 0.00019999566776254734,
-      "loss": 46.0,
-      "step": 38763
-    },
-    {
-      "epoch": 2.9637785041191202,
-      "grad_norm": 0.0031028245575726032,
-      "learning_rate": 0.000199995667538964,
-      "loss": 46.0,
-      "step": 38764
-    },
-    {
-      "epoch": 2.9638549611025096,
-      "grad_norm": 0.0020329959224909544,
-      "learning_rate": 0.00019999566731537493,
-      "loss": 46.0,
-      "step": 38765
-    },
-    {
-      "epoch": 2.9639314180858993,
-      "grad_norm": 0.007521012332290411,
-      "learning_rate": 0.00019999566709178008,
-      "loss": 46.0,
-      "step": 38766
-    },
-    {
-      "epoch": 2.964007875069289,
-      "grad_norm": 0.001825822750106454,
-      "learning_rate": 0.00019999566686817945,
-      "loss": 46.0,
-      "step": 38767
-    },
-    {
-      "epoch": 2.964084332052679,
-      "grad_norm": 0.002598043065518141,
-      "learning_rate": 0.00019999566664457308,
-      "loss": 46.0,
-      "step": 38768
-    },
-    {
-      "epoch": 2.9641607890360686,
-      "grad_norm": 0.0035057917702943087,
-      "learning_rate": 0.0001999956664209609,
-      "loss": 46.0,
-      "step": 38769
-    },
-    {
-      "epoch": 2.9642372460194584,
-      "grad_norm": 0.0018140760948881507,
-      "learning_rate": 0.000199995666197343,
-      "loss": 46.0,
-      "step": 38770
-    },
-    {
-      "epoch": 2.964313703002848,
-      "grad_norm": 0.0031152910087257624,
-      "learning_rate": 0.00019999566597371927,
-      "loss": 46.0,
-      "step": 38771
-    },
-    {
-      "epoch": 2.964390159986238,
-      "grad_norm": 0.0035517015494406223,
-      "learning_rate": 0.0001999956657500898,
-      "loss": 46.0,
-      "step": 38772
-    },
-    {
-      "epoch": 2.964466616969627,
-      "grad_norm": 0.019760986790060997,
-      "learning_rate": 0.0001999956655264546,
-      "loss": 46.0,
-      "step": 38773
-    },
-    {
-      "epoch": 2.964543073953017,
-      "grad_norm": 0.004486551508307457,
-      "learning_rate": 0.00019999566530281359,
-      "loss": 46.0,
-      "step": 38774
-    },
-    {
-      "epoch": 2.9646195309364067,
-      "grad_norm": 0.004180152900516987,
-      "learning_rate": 0.0001999956650791668,
-      "loss": 46.0,
-      "step": 38775
-    },
-    {
-      "epoch": 2.9646959879197965,
-      "grad_norm": 0.000872608448844403,
-      "learning_rate": 0.00019999566485551427,
-      "loss": 46.0,
-      "step": 38776
-    },
-    {
-      "epoch": 2.9647724449031863,
-      "grad_norm": 0.001664087874814868,
-      "learning_rate": 0.00019999566463185594,
-      "loss": 46.0,
-      "step": 38777
-    },
-    {
-      "epoch": 2.964848901886576,
-      "grad_norm": 0.003988146316260099,
-      "learning_rate": 0.00019999566440819186,
-      "loss": 46.0,
-      "step": 38778
-    },
-    {
-      "epoch": 2.9649253588699658,
-      "grad_norm": 0.0021746321581304073,
-      "learning_rate": 0.00019999566418452204,
-      "loss": 46.0,
-      "step": 38779
-    },
-    {
-      "epoch": 2.9650018158533555,
-      "grad_norm": 0.0029714221600443125,
-      "learning_rate": 0.00019999566396084638,
-      "loss": 46.0,
-      "step": 38780
-    },
-    {
-      "epoch": 2.9650782728367453,
-      "grad_norm": 0.0009956000139936805,
-      "learning_rate": 0.000199995663737165,
-      "loss": 46.0,
-      "step": 38781
-    },
-    {
-      "epoch": 2.965154729820135,
-      "grad_norm": 0.006941694766283035,
-      "learning_rate": 0.00019999566351347782,
-      "loss": 46.0,
-      "step": 38782
-    },
-    {
-      "epoch": 2.965231186803525,
-      "grad_norm": 0.003042785217985511,
-      "learning_rate": 0.00019999566328978493,
-      "loss": 46.0,
-      "step": 38783
-    },
-    {
-      "epoch": 2.9653076437869146,
-      "grad_norm": 0.0036308171693235636,
-      "learning_rate": 0.0001999956630660862,
-      "loss": 46.0,
-      "step": 38784
-    },
-    {
-      "epoch": 2.9653841007703043,
-      "grad_norm": 0.002663103397935629,
-      "learning_rate": 0.00019999566284238177,
-      "loss": 46.0,
-      "step": 38785
-    },
-    {
-      "epoch": 2.965460557753694,
-      "grad_norm": 0.0012971964897587895,
-      "learning_rate": 0.0001999956626186715,
-      "loss": 46.0,
-      "step": 38786
-    },
-    {
-      "epoch": 2.9655370147370834,
-      "grad_norm": 0.003447584342211485,
-      "learning_rate": 0.0001999956623949555,
-      "loss": 46.0,
-      "step": 38787
-    },
-    {
-      "epoch": 2.965613471720473,
-      "grad_norm": 0.0016339856665581465,
-      "learning_rate": 0.00019999566217123374,
-      "loss": 46.0,
-      "step": 38788
-    },
-    {
-      "epoch": 2.965689928703863,
-      "grad_norm": 0.0029338186141103506,
-      "learning_rate": 0.0001999956619475062,
-      "loss": 46.0,
-      "step": 38789
-    },
-    {
-      "epoch": 2.9657663856872527,
-      "grad_norm": 0.0040512001141905785,
-      "learning_rate": 0.00019999566172377288,
-      "loss": 46.0,
-      "step": 38790
-    },
-    {
-      "epoch": 2.9658428426706425,
-      "grad_norm": 0.0013265768066048622,
-      "learning_rate": 0.0001999956615000338,
-      "loss": 46.0,
-      "step": 38791
-    },
-    {
-      "epoch": 2.9659192996540322,
-      "grad_norm": 0.0028855325654149055,
-      "learning_rate": 0.00019999566127628895,
-      "loss": 46.0,
-      "step": 38792
-    },
-    {
-      "epoch": 2.965995756637422,
-      "grad_norm": 0.000699925294611603,
-      "learning_rate": 0.00019999566105253832,
-      "loss": 46.0,
-      "step": 38793
-    },
-    {
-      "epoch": 2.9660722136208117,
-      "grad_norm": 0.0008901142864488065,
-      "learning_rate": 0.00019999566082878193,
-      "loss": 46.0,
-      "step": 38794
-    },
-    {
-      "epoch": 2.966148670604201,
-      "grad_norm": 0.003148681716993451,
-      "learning_rate": 0.00019999566060501978,
-      "loss": 46.0,
-      "step": 38795
-    },
-    {
-      "epoch": 2.966225127587591,
-      "grad_norm": 0.00447940593585372,
-      "learning_rate": 0.00019999566038125184,
-      "loss": 46.0,
-      "step": 38796
-    },
-    {
-      "epoch": 2.9663015845709806,
-      "grad_norm": 0.006700800731778145,
-      "learning_rate": 0.00019999566015747817,
-      "loss": 46.0,
-      "step": 38797
-    },
-    {
-      "epoch": 2.9663780415543703,
-      "grad_norm": 0.004799845162779093,
-      "learning_rate": 0.00019999565993369868,
-      "loss": 46.0,
-      "step": 38798
-    },
-    {
-      "epoch": 2.96645449853776,
-      "grad_norm": 0.001041700947098434,
-      "learning_rate": 0.00019999565970991345,
-      "loss": 46.0,
-      "step": 38799
-    },
-    {
-      "epoch": 2.96653095552115,
-      "grad_norm": 0.0014246506616473198,
-      "learning_rate": 0.00019999565948612244,
-      "loss": 46.0,
-      "step": 38800
-    },
-    {
-      "epoch": 2.9666074125045396,
-      "grad_norm": 0.001999033149331808,
-      "learning_rate": 0.00019999565926232568,
-      "loss": 46.0,
-      "step": 38801
-    },
-    {
-      "epoch": 2.9666838694879294,
-      "grad_norm": 0.008738222531974316,
-      "learning_rate": 0.00019999565903852312,
-      "loss": 46.0,
-      "step": 38802
-    },
-    {
-      "epoch": 2.966760326471319,
-      "grad_norm": 0.0029781197663396597,
-      "learning_rate": 0.00019999565881471482,
-      "loss": 46.0,
-      "step": 38803
-    },
-    {
-      "epoch": 2.966836783454709,
-      "grad_norm": 0.005128840915858746,
-      "learning_rate": 0.0001999956585909007,
-      "loss": 46.0,
-      "step": 38804
-    },
-    {
-      "epoch": 2.9669132404380987,
-      "grad_norm": 0.002178050111979246,
-      "learning_rate": 0.0001999956583670809,
-      "loss": 46.0,
-      "step": 38805
-    },
-    {
-      "epoch": 2.9669896974214884,
-      "grad_norm": 0.001666230265982449,
-      "learning_rate": 0.00019999565814325527,
-      "loss": 46.0,
-      "step": 38806
-    },
-    {
-      "epoch": 2.967066154404878,
-      "grad_norm": 0.002604331821203232,
-      "learning_rate": 0.0001999956579194239,
-      "loss": 46.0,
-      "step": 38807
-    },
-    {
-      "epoch": 2.9671426113882675,
-      "grad_norm": 0.0011654363479465246,
-      "learning_rate": 0.00019999565769558673,
-      "loss": 46.0,
-      "step": 38808
-    },
-    {
-      "epoch": 2.9672190683716573,
-      "grad_norm": 0.0025039773900061846,
-      "learning_rate": 0.0001999956574717438,
-      "loss": 46.0,
-      "step": 38809
-    },
-    {
-      "epoch": 2.967295525355047,
-      "grad_norm": 0.003030512947589159,
-      "learning_rate": 0.00019999565724789512,
-      "loss": 46.0,
-      "step": 38810
-    },
-    {
-      "epoch": 2.967371982338437,
-      "grad_norm": 0.0021453530061990023,
-      "learning_rate": 0.00019999565702404063,
-      "loss": 46.0,
-      "step": 38811
-    },
-    {
-      "epoch": 2.9674484393218266,
-      "grad_norm": 0.0027277146000415087,
-      "learning_rate": 0.0001999956568001804,
-      "loss": 46.0,
-      "step": 38812
-    },
-    {
-      "epoch": 2.9675248963052163,
-      "grad_norm": 0.0007995835039764643,
-      "learning_rate": 0.0001999956565763144,
-      "loss": 46.0,
-      "step": 38813
-    },
-    {
-      "epoch": 2.967601353288606,
-      "grad_norm": 0.002258928259834647,
-      "learning_rate": 0.00019999565635244265,
-      "loss": 46.0,
-      "step": 38814
-    },
-    {
-      "epoch": 2.967677810271996,
-      "grad_norm": 0.0027025151066482067,
-      "learning_rate": 0.00019999565612856507,
-      "loss": 46.0,
-      "step": 38815
-    },
-    {
-      "epoch": 2.9677542672553856,
-      "grad_norm": 0.0007156643550843,
-      "learning_rate": 0.00019999565590468176,
-      "loss": 46.0,
-      "step": 38816
-    },
-    {
-      "epoch": 2.967830724238775,
-      "grad_norm": 0.00210730847902596,
-      "learning_rate": 0.0001999956556807927,
-      "loss": 46.0,
-      "step": 38817
-    },
-    {
-      "epoch": 2.9679071812221647,
-      "grad_norm": 0.0020987922325730324,
-      "learning_rate": 0.00019999565545689784,
-      "loss": 46.0,
-      "step": 38818
-    },
-    {
-      "epoch": 2.9679836382055544,
-      "grad_norm": 0.0038130159955471754,
-      "learning_rate": 0.0001999956552329972,
-      "loss": 46.0,
-      "step": 38819
-    },
-    {
-      "epoch": 2.968060095188944,
-      "grad_norm": 0.002649113768711686,
-      "learning_rate": 0.00019999565500909084,
-      "loss": 46.0,
-      "step": 38820
-    },
-    {
-      "epoch": 2.968136552172334,
-      "grad_norm": 0.006833228282630444,
-      "learning_rate": 0.00019999565478517867,
-      "loss": 46.0,
-      "step": 38821
-    },
-    {
-      "epoch": 2.9682130091557237,
-      "grad_norm": 0.007193616591393948,
-      "learning_rate": 0.00019999565456126076,
-      "loss": 46.0,
-      "step": 38822
-    },
-    {
-      "epoch": 2.9682894661391135,
-      "grad_norm": 0.0016768421046435833,
-      "learning_rate": 0.00019999565433733704,
-      "loss": 46.0,
-      "step": 38823
-    },
-    {
-      "epoch": 2.9683659231225032,
-      "grad_norm": 0.0023161936551332474,
-      "learning_rate": 0.0001999956541134076,
-      "loss": 46.0,
-      "step": 38824
-    },
-    {
-      "epoch": 2.968442380105893,
-      "grad_norm": 0.002078343415632844,
-      "learning_rate": 0.00019999565388947234,
-      "loss": 46.0,
-      "step": 38825
-    },
-    {
-      "epoch": 2.9685188370892828,
-      "grad_norm": 0.011018405668437481,
-      "learning_rate": 0.00019999565366553136,
-      "loss": 46.0,
-      "step": 38826
-    },
-    {
-      "epoch": 2.9685952940726725,
-      "grad_norm": 0.0011039135279133916,
-      "learning_rate": 0.00019999565344158457,
-      "loss": 46.0,
-      "step": 38827
-    },
-    {
-      "epoch": 2.9686717510560623,
-      "grad_norm": 0.0019084344385191798,
-      "learning_rate": 0.00019999565321763202,
-      "loss": 46.0,
-      "step": 38828
-    },
-    {
-      "epoch": 2.968748208039452,
-      "grad_norm": 0.003199186408892274,
-      "learning_rate": 0.00019999565299367374,
-      "loss": 46.0,
-      "step": 38829
-    },
-    {
-      "epoch": 2.9688246650228414,
-      "grad_norm": 0.0023745347280055285,
-      "learning_rate": 0.00019999565276970964,
-      "loss": 46.0,
-      "step": 38830
-    },
-    {
-      "epoch": 2.968901122006231,
-      "grad_norm": 0.0014878212241455913,
-      "learning_rate": 0.00019999565254573979,
-      "loss": 46.0,
-      "step": 38831
-    },
-    {
-      "epoch": 2.968977578989621,
-      "grad_norm": 0.0013919976772740483,
-      "learning_rate": 0.0001999956523217642,
-      "loss": 46.0,
-      "step": 38832
-    },
-    {
-      "epoch": 2.9690540359730107,
-      "grad_norm": 0.0047453478910028934,
-      "learning_rate": 0.00019999565209778277,
-      "loss": 46.0,
-      "step": 38833
-    },
-    {
-      "epoch": 2.9691304929564004,
-      "grad_norm": 0.0026916400529444218,
-      "learning_rate": 0.00019999565187379562,
-      "loss": 46.0,
-      "step": 38834
-    },
-    {
-      "epoch": 2.96920694993979,
-      "grad_norm": 0.0006237321067601442,
-      "learning_rate": 0.0001999956516498027,
-      "loss": 46.0,
-      "step": 38835
-    },
-    {
-      "epoch": 2.96928340692318,
-      "grad_norm": 0.0015316014178097248,
-      "learning_rate": 0.000199995651425804,
-      "loss": 46.0,
-      "step": 38836
-    },
-    {
-      "epoch": 2.9693598639065697,
-      "grad_norm": 0.002091713948175311,
-      "learning_rate": 0.00019999565120179953,
-      "loss": 46.0,
-      "step": 38837
-    },
-    {
-      "epoch": 2.969436320889959,
-      "grad_norm": 0.0018464390886947513,
-      "learning_rate": 0.0001999956509777893,
-      "loss": 46.0,
-      "step": 38838
-    },
-    {
-      "epoch": 2.9695127778733488,
-      "grad_norm": 0.0012911810772493482,
-      "learning_rate": 0.00019999565075377328,
-      "loss": 46.0,
-      "step": 38839
-    },
-    {
-      "epoch": 2.9695892348567385,
-      "grad_norm": 0.002915911842137575,
-      "learning_rate": 0.00019999565052975153,
-      "loss": 46.0,
-      "step": 38840
-    },
-    {
-      "epoch": 2.9696656918401283,
-      "grad_norm": 0.0013452632119879127,
-      "learning_rate": 0.000199995650305724,
-      "loss": 46.0,
-      "step": 38841
-    },
-    {
-      "epoch": 2.969742148823518,
-      "grad_norm": 0.0016468253452330828,
-      "learning_rate": 0.00019999565008169067,
-      "loss": 46.0,
-      "step": 38842
-    },
-    {
-      "epoch": 2.969818605806908,
-      "grad_norm": 0.002326331799849868,
-      "learning_rate": 0.0001999956498576516,
-      "loss": 46.0,
-      "step": 38843
-    },
-    {
-      "epoch": 2.9698950627902976,
-      "grad_norm": 0.0012755782809108496,
-      "learning_rate": 0.00019999564963360674,
-      "loss": 46.0,
-      "step": 38844
-    },
-    {
-      "epoch": 2.9699715197736873,
-      "grad_norm": 0.0029674021061509848,
-      "learning_rate": 0.00019999564940955612,
-      "loss": 46.0,
-      "step": 38845
-    },
-    {
-      "epoch": 2.970047976757077,
-      "grad_norm": 0.000567444774787873,
-      "learning_rate": 0.00019999564918549972,
-      "loss": 46.0,
-      "step": 38846
-    },
-    {
-      "epoch": 2.970124433740467,
-      "grad_norm": 0.0013332696398720145,
-      "learning_rate": 0.00019999564896143758,
-      "loss": 46.0,
-      "step": 38847
-    },
-    {
-      "epoch": 2.9702008907238566,
-      "grad_norm": 0.0015207903925329447,
-      "learning_rate": 0.00019999564873736966,
-      "loss": 46.0,
-      "step": 38848
-    },
-    {
-      "epoch": 2.9702773477072464,
-      "grad_norm": 0.0021245379466563463,
-      "learning_rate": 0.00019999564851329597,
-      "loss": 46.0,
-      "step": 38849
-    },
-    {
-      "epoch": 2.970353804690636,
-      "grad_norm": 0.004931046161800623,
-      "learning_rate": 0.0001999956482892165,
-      "loss": 46.0,
-      "step": 38850
-    },
-    {
-      "epoch": 2.970430261674026,
-      "grad_norm": 0.0015642642974853516,
-      "learning_rate": 0.00019999564806513127,
-      "loss": 46.0,
-      "step": 38851
-    },
-    {
-      "epoch": 2.9705067186574152,
-      "grad_norm": 0.0014162542065605521,
-      "learning_rate": 0.00019999564784104027,
-      "loss": 46.0,
-      "step": 38852
-    },
-    {
-      "epoch": 2.970583175640805,
-      "grad_norm": 0.0017431830056011677,
-      "learning_rate": 0.00019999564761694348,
-      "loss": 46.0,
-      "step": 38853
-    },
-    {
-      "epoch": 2.9706596326241947,
-      "grad_norm": 0.003257500473409891,
-      "learning_rate": 0.00019999564739284095,
-      "loss": 46.0,
-      "step": 38854
-    },
-    {
-      "epoch": 2.9707360896075845,
-      "grad_norm": 0.0019210914615541697,
-      "learning_rate": 0.00019999564716873262,
-      "loss": 46.0,
-      "step": 38855
-    },
-    {
-      "epoch": 2.9708125465909743,
-      "grad_norm": 0.0072126430459320545,
-      "learning_rate": 0.00019999564694461858,
-      "loss": 46.0,
-      "step": 38856
-    },
-    {
-      "epoch": 2.970889003574364,
-      "grad_norm": 0.002768451115116477,
-      "learning_rate": 0.0001999956467204987,
-      "loss": 46.0,
-      "step": 38857
-    },
-    {
-      "epoch": 2.970965460557754,
-      "grad_norm": 0.0008489834726788104,
-      "learning_rate": 0.00019999564649637308,
-      "loss": 46.0,
-      "step": 38858
-    },
-    {
-      "epoch": 2.9710419175411436,
-      "grad_norm": 0.0013236310333013535,
-      "learning_rate": 0.0001999956462722417,
-      "loss": 46.0,
-      "step": 38859
-    },
-    {
-      "epoch": 2.971118374524533,
-      "grad_norm": 0.006933864671736956,
-      "learning_rate": 0.00019999564604810454,
-      "loss": 46.0,
-      "step": 38860
-    },
-    {
-      "epoch": 2.9711948315079226,
-      "grad_norm": 0.006330015603452921,
-      "learning_rate": 0.00019999564582396162,
-      "loss": 46.0,
-      "step": 38861
-    },
-    {
-      "epoch": 2.9712712884913124,
-      "grad_norm": 0.005118245258927345,
-      "learning_rate": 0.0001999956455998129,
-      "loss": 46.0,
-      "step": 38862
-    },
-    {
-      "epoch": 2.971347745474702,
-      "grad_norm": 0.003926456440240145,
-      "learning_rate": 0.00019999564537565844,
-      "loss": 46.0,
-      "step": 38863
-    },
-    {
-      "epoch": 2.971424202458092,
-      "grad_norm": 0.0034800253342837095,
-      "learning_rate": 0.00019999564515149824,
-      "loss": 46.0,
-      "step": 38864
-    },
-    {
-      "epoch": 2.9715006594414817,
-      "grad_norm": 0.0008141734870150685,
-      "learning_rate": 0.00019999564492733223,
-      "loss": 46.0,
-      "step": 38865
-    },
-    {
-      "epoch": 2.9715771164248714,
-      "grad_norm": 0.0009938642615452409,
-      "learning_rate": 0.00019999564470316044,
-      "loss": 46.0,
-      "step": 38866
-    },
-    {
-      "epoch": 2.971653573408261,
-      "grad_norm": 0.0009254764881916344,
-      "learning_rate": 0.00019999564447898292,
-      "loss": 46.0,
-      "step": 38867
-    },
-    {
-      "epoch": 2.971730030391651,
-      "grad_norm": 0.001480698585510254,
-      "learning_rate": 0.00019999564425479959,
-      "loss": 46.0,
-      "step": 38868
-    },
-    {
-      "epoch": 2.9718064873750407,
-      "grad_norm": 0.0027772849425673485,
-      "learning_rate": 0.0001999956440306105,
-      "loss": 46.0,
-      "step": 38869
-    },
-    {
-      "epoch": 2.9718829443584305,
-      "grad_norm": 0.003581203054636717,
-      "learning_rate": 0.00019999564380641566,
-      "loss": 46.0,
-      "step": 38870
-    },
-    {
-      "epoch": 2.9719594013418202,
-      "grad_norm": 0.0036480564158409834,
-      "learning_rate": 0.00019999564358221507,
-      "loss": 46.0,
-      "step": 38871
-    },
-    {
-      "epoch": 2.97203585832521,
-      "grad_norm": 0.001648459816351533,
-      "learning_rate": 0.00019999564335800867,
-      "loss": 46.0,
-      "step": 38872
-    },
-    {
-      "epoch": 2.9721123153085998,
-      "grad_norm": 0.002940130652859807,
-      "learning_rate": 0.0001999956431337965,
-      "loss": 46.0,
-      "step": 38873
-    },
-    {
-      "epoch": 2.972188772291989,
-      "grad_norm": 0.002545031951740384,
-      "learning_rate": 0.0001999956429095786,
-      "loss": 46.0,
-      "step": 38874
-    },
-    {
-      "epoch": 2.972265229275379,
-      "grad_norm": 0.001602409640327096,
-      "learning_rate": 0.0001999956426853549,
-      "loss": 46.0,
-      "step": 38875
-    },
-    {
-      "epoch": 2.9723416862587686,
-      "grad_norm": 0.001739976927638054,
-      "learning_rate": 0.00019999564246112544,
-      "loss": 46.0,
-      "step": 38876
-    },
-    {
-      "epoch": 2.9724181432421584,
-      "grad_norm": 0.005991366691887379,
-      "learning_rate": 0.00019999564223689017,
-      "loss": 46.0,
-      "step": 38877
-    },
-    {
-      "epoch": 2.972494600225548,
-      "grad_norm": 0.0025928064715117216,
-      "learning_rate": 0.0001999956420126492,
-      "loss": 46.0,
-      "step": 38878
-    },
-    {
-      "epoch": 2.972571057208938,
-      "grad_norm": 0.0015057475538924336,
-      "learning_rate": 0.0001999956417884024,
-      "loss": 46.0,
-      "step": 38879
-    },
-    {
-      "epoch": 2.9726475141923276,
-      "grad_norm": 0.002212582156062126,
-      "learning_rate": 0.00019999564156414988,
-      "loss": 46.0,
-      "step": 38880
-    },
-    {
-      "epoch": 2.9727239711757174,
-      "grad_norm": 0.008032852783799171,
-      "learning_rate": 0.00019999564133989158,
-      "loss": 46.0,
-      "step": 38881
-    },
-    {
-      "epoch": 2.9728004281591067,
-      "grad_norm": 0.002232054015621543,
-      "learning_rate": 0.0001999956411156275,
-      "loss": 46.0,
-      "step": 38882
-    },
-    {
-      "epoch": 2.9728768851424965,
-      "grad_norm": 0.003509450936689973,
-      "learning_rate": 0.00019999564089135766,
-      "loss": 46.0,
-      "step": 38883
-    },
-    {
-      "epoch": 2.9729533421258862,
-      "grad_norm": 0.0025061210617423058,
-      "learning_rate": 0.00019999564066708203,
-      "loss": 46.0,
-      "step": 38884
-    },
-    {
-      "epoch": 2.973029799109276,
-      "grad_norm": 0.017708707600831985,
-      "learning_rate": 0.00019999564044280064,
-      "loss": 46.0,
-      "step": 38885
-    },
-    {
-      "epoch": 2.9731062560926658,
-      "grad_norm": 0.00591287249699235,
-      "learning_rate": 0.00019999564021851347,
-      "loss": 46.0,
-      "step": 38886
-    },
-    {
-      "epoch": 2.9731827130760555,
-      "grad_norm": 0.0017712911358103156,
-      "learning_rate": 0.00019999563999422058,
-      "loss": 46.0,
-      "step": 38887
-    },
-    {
-      "epoch": 2.9732591700594453,
-      "grad_norm": 0.0023941416293382645,
-      "learning_rate": 0.00019999563976992187,
-      "loss": 46.0,
-      "step": 38888
-    },
-    {
-      "epoch": 2.973335627042835,
-      "grad_norm": 0.0014949578326195478,
-      "learning_rate": 0.0001999956395456174,
-      "loss": 46.0,
-      "step": 38889
-    },
-    {
-      "epoch": 2.973412084026225,
-      "grad_norm": 0.0029613198712468147,
-      "learning_rate": 0.00019999563932130717,
-      "loss": 46.0,
-      "step": 38890
-    },
-    {
-      "epoch": 2.9734885410096146,
-      "grad_norm": 0.0008647587965242565,
-      "learning_rate": 0.00019999563909699116,
-      "loss": 46.0,
-      "step": 38891
-    },
-    {
-      "epoch": 2.9735649979930043,
-      "grad_norm": 0.0031321656424552202,
-      "learning_rate": 0.0001999956388726694,
-      "loss": 46.0,
-      "step": 38892
-    },
-    {
-      "epoch": 2.973641454976394,
-      "grad_norm": 0.0014598460402339697,
-      "learning_rate": 0.00019999563864834185,
-      "loss": 46.0,
-      "step": 38893
-    },
-    {
-      "epoch": 2.973717911959784,
-      "grad_norm": 0.0012144835200160742,
-      "learning_rate": 0.00019999563842400855,
-      "loss": 46.0,
-      "step": 38894
-    },
-    {
-      "epoch": 2.9737943689431736,
-      "grad_norm": 0.004519733600318432,
-      "learning_rate": 0.00019999563819966947,
-      "loss": 46.0,
-      "step": 38895
-    },
-    {
-      "epoch": 2.973870825926563,
-      "grad_norm": 0.005155148450285196,
-      "learning_rate": 0.00019999563797532463,
-      "loss": 46.0,
-      "step": 38896
-    },
-    {
-      "epoch": 2.9739472829099527,
-      "grad_norm": 0.001507338136434555,
-      "learning_rate": 0.000199995637750974,
-      "loss": 46.0,
-      "step": 38897
-    },
-    {
-      "epoch": 2.9740237398933425,
-      "grad_norm": 0.001250439672730863,
-      "learning_rate": 0.0001999956375266176,
-      "loss": 46.0,
-      "step": 38898
-    },
-    {
-      "epoch": 2.974100196876732,
-      "grad_norm": 0.0022691816557198763,
-      "learning_rate": 0.00019999563730225547,
-      "loss": 46.0,
-      "step": 38899
-    },
-    {
-      "epoch": 2.974176653860122,
-      "grad_norm": 0.0016748163616284728,
-      "learning_rate": 0.00019999563707788753,
-      "loss": 46.0,
-      "step": 38900
-    },
-    {
-      "epoch": 2.9742531108435117,
-      "grad_norm": 0.0017410056898370385,
-      "learning_rate": 0.00019999563685351387,
-      "loss": 46.0,
-      "step": 38901
-    },
-    {
-      "epoch": 2.9743295678269015,
-      "grad_norm": 0.007299117278307676,
-      "learning_rate": 0.00019999563662913438,
-      "loss": 46.0,
-      "step": 38902
-    },
-    {
-      "epoch": 2.9744060248102913,
-      "grad_norm": 0.0008122262079268694,
-      "learning_rate": 0.00019999563640474914,
-      "loss": 46.0,
-      "step": 38903
-    },
-    {
-      "epoch": 2.9744824817936806,
-      "grad_norm": 0.0010236442321911454,
-      "learning_rate": 0.00019999563618035816,
-      "loss": 46.0,
-      "step": 38904
-    },
-    {
-      "epoch": 2.9745589387770703,
-      "grad_norm": 0.0025179386138916016,
-      "learning_rate": 0.00019999563595596138,
-      "loss": 46.0,
-      "step": 38905
-    },
-    {
-      "epoch": 2.97463539576046,
-      "grad_norm": 0.004896245896816254,
-      "learning_rate": 0.00019999563573155885,
-      "loss": 46.0,
-      "step": 38906
-    },
-    {
-      "epoch": 2.97471185274385,
-      "grad_norm": 0.0018146460643038154,
-      "learning_rate": 0.00019999563550715052,
-      "loss": 46.0,
-      "step": 38907
-    },
-    {
-      "epoch": 2.9747883097272396,
-      "grad_norm": 0.00169906928204,
-      "learning_rate": 0.00019999563528273645,
-      "loss": 46.0,
-      "step": 38908
-    },
-    {
-      "epoch": 2.9748647667106294,
-      "grad_norm": 0.001374110928736627,
-      "learning_rate": 0.0001999956350583166,
-      "loss": 46.0,
-      "step": 38909
-    },
-    {
-      "epoch": 2.974941223694019,
-      "grad_norm": 0.002236147178336978,
-      "learning_rate": 0.00019999563483389098,
-      "loss": 46.0,
-      "step": 38910
-    },
-    {
-      "epoch": 2.975017680677409,
-      "grad_norm": 0.0019094528397545218,
-      "learning_rate": 0.00019999563460945959,
-      "loss": 46.0,
-      "step": 38911
-    },
-    {
-      "epoch": 2.9750941376607987,
-      "grad_norm": 0.0012492660898715258,
-      "learning_rate": 0.00019999563438502244,
-      "loss": 46.0,
-      "step": 38912
-    },
-    {
-      "epoch": 2.9751705946441884,
-      "grad_norm": 0.0013261620188131928,
-      "learning_rate": 0.0001999956341605795,
-      "loss": 46.0,
-      "step": 38913
-    },
-    {
-      "epoch": 2.975247051627578,
-      "grad_norm": 0.0022912088315933943,
-      "learning_rate": 0.00019999563393613082,
-      "loss": 46.0,
-      "step": 38914
-    },
-    {
-      "epoch": 2.975323508610968,
-      "grad_norm": 0.003404107643291354,
-      "learning_rate": 0.00019999563371167635,
-      "loss": 46.0,
-      "step": 38915
-    },
-    {
-      "epoch": 2.9753999655943577,
-      "grad_norm": 0.0008852720493450761,
-      "learning_rate": 0.00019999563348721612,
-      "loss": 46.0,
-      "step": 38916
-    },
-    {
-      "epoch": 2.9754764225777475,
-      "grad_norm": 0.001354023814201355,
-      "learning_rate": 0.00019999563326275014,
-      "loss": 46.0,
-      "step": 38917
-    },
-    {
-      "epoch": 2.975552879561137,
-      "grad_norm": 0.0038675107061862946,
-      "learning_rate": 0.00019999563303827836,
-      "loss": 46.0,
-      "step": 38918
-    },
-    {
-      "epoch": 2.9756293365445265,
-      "grad_norm": 0.0012964140623807907,
-      "learning_rate": 0.0001999956328138008,
-      "loss": 46.0,
-      "step": 38919
-    },
-    {
-      "epoch": 2.9757057935279163,
-      "grad_norm": 0.0010123233078047633,
-      "learning_rate": 0.0001999956325893175,
-      "loss": 46.0,
-      "step": 38920
-    },
-    {
-      "epoch": 2.975782250511306,
-      "grad_norm": 0.0013981809606775641,
-      "learning_rate": 0.00019999563236482843,
-      "loss": 46.0,
-      "step": 38921
-    },
-    {
-      "epoch": 2.975858707494696,
-      "grad_norm": 0.002034079981967807,
-      "learning_rate": 0.00019999563214033355,
-      "loss": 46.0,
-      "step": 38922
-    },
-    {
-      "epoch": 2.9759351644780856,
-      "grad_norm": 0.003962273709475994,
-      "learning_rate": 0.00019999563191583296,
-      "loss": 46.0,
-      "step": 38923
-    },
-    {
-      "epoch": 2.9760116214614754,
-      "grad_norm": 0.0016956643667072058,
-      "learning_rate": 0.0001999956316913266,
-      "loss": 46.0,
-      "step": 38924
-    },
-    {
-      "epoch": 2.976088078444865,
-      "grad_norm": 0.003656771732494235,
-      "learning_rate": 0.00019999563146681442,
-      "loss": 46.0,
-      "step": 38925
-    },
-    {
-      "epoch": 2.9761645354282544,
-      "grad_norm": 0.002302858978509903,
-      "learning_rate": 0.0001999956312422965,
-      "loss": 46.0,
-      "step": 38926
-    },
-    {
-      "epoch": 2.976240992411644,
-      "grad_norm": 0.0006120320758782327,
-      "learning_rate": 0.00019999563101777282,
-      "loss": 46.0,
-      "step": 38927
-    },
-    {
-      "epoch": 2.976317449395034,
-      "grad_norm": 0.0040608961135149,
-      "learning_rate": 0.00019999563079324336,
-      "loss": 46.0,
-      "step": 38928
-    },
-    {
-      "epoch": 2.9763939063784237,
-      "grad_norm": 0.0036757125053554773,
-      "learning_rate": 0.00019999563056870813,
-      "loss": 46.0,
-      "step": 38929
-    },
-    {
-      "epoch": 2.9764703633618135,
-      "grad_norm": 0.0010158759541809559,
-      "learning_rate": 0.0001999956303441671,
-      "loss": 46.0,
-      "step": 38930
-    },
-    {
-      "epoch": 2.9765468203452032,
-      "grad_norm": 0.005072961561381817,
-      "learning_rate": 0.00019999563011962034,
-      "loss": 46.0,
-      "step": 38931
-    },
-    {
-      "epoch": 2.976623277328593,
-      "grad_norm": 0.0014291313709691167,
-      "learning_rate": 0.0001999956298950678,
-      "loss": 46.0,
-      "step": 38932
-    },
-    {
-      "epoch": 2.9766997343119828,
-      "grad_norm": 0.0014953481731936336,
-      "learning_rate": 0.00019999562967050948,
-      "loss": 46.0,
-      "step": 38933
-    },
-    {
-      "epoch": 2.9767761912953725,
-      "grad_norm": 0.0038099829107522964,
-      "learning_rate": 0.00019999562944594544,
-      "loss": 46.0,
-      "step": 38934
-    },
-    {
-      "epoch": 2.9768526482787623,
-      "grad_norm": 0.002259933389723301,
-      "learning_rate": 0.00019999562922137556,
-      "loss": 46.0,
-      "step": 38935
-    },
-    {
-      "epoch": 2.976929105262152,
-      "grad_norm": 0.0022849731612950563,
-      "learning_rate": 0.00019999562899679994,
-      "loss": 46.0,
-      "step": 38936
-    },
-    {
-      "epoch": 2.977005562245542,
-      "grad_norm": 0.001822706894017756,
-      "learning_rate": 0.00019999562877221858,
-      "loss": 46.0,
-      "step": 38937
-    },
-    {
-      "epoch": 2.9770820192289316,
-      "grad_norm": 0.0012723540421575308,
-      "learning_rate": 0.00019999562854763144,
-      "loss": 46.0,
-      "step": 38938
-    },
-    {
-      "epoch": 2.977158476212321,
-      "grad_norm": 0.0011593495728448033,
-      "learning_rate": 0.00019999562832303852,
-      "loss": 46.0,
-      "step": 38939
-    },
-    {
-      "epoch": 2.9772349331957106,
-      "grad_norm": 0.0023144090082496405,
-      "learning_rate": 0.0001999956280984398,
-      "loss": 46.0,
-      "step": 38940
-    },
-    {
-      "epoch": 2.9773113901791004,
-      "grad_norm": 0.0015373806236311793,
-      "learning_rate": 0.00019999562787383535,
-      "loss": 46.0,
-      "step": 38941
-    },
-    {
-      "epoch": 2.97738784716249,
-      "grad_norm": 0.0015771586913615465,
-      "learning_rate": 0.0001999956276492251,
-      "loss": 46.0,
-      "step": 38942
-    },
-    {
-      "epoch": 2.97746430414588,
-      "grad_norm": 0.0019857927691191435,
-      "learning_rate": 0.0001999956274246091,
-      "loss": 46.0,
-      "step": 38943
-    },
-    {
-      "epoch": 2.9775407611292697,
-      "grad_norm": 0.004098236560821533,
-      "learning_rate": 0.00019999562719998735,
-      "loss": 46.0,
-      "step": 38944
-    },
-    {
-      "epoch": 2.9776172181126594,
-      "grad_norm": 0.0029844462405890226,
-      "learning_rate": 0.00019999562697535977,
-      "loss": 46.0,
-      "step": 38945
-    },
-    {
-      "epoch": 2.977693675096049,
-      "grad_norm": 0.0023544856812804937,
-      "learning_rate": 0.0001999956267507265,
-      "loss": 46.0,
-      "step": 38946
-    },
-    {
-      "epoch": 2.977770132079439,
-      "grad_norm": 0.0018966731149703264,
-      "learning_rate": 0.0001999956265260874,
-      "loss": 46.0,
-      "step": 38947
-    },
-    {
-      "epoch": 2.9778465890628283,
-      "grad_norm": 0.008401605300605297,
-      "learning_rate": 0.00019999562630144255,
-      "loss": 46.0,
-      "step": 38948
-    },
-    {
-      "epoch": 2.977923046046218,
-      "grad_norm": 0.00455885985866189,
-      "learning_rate": 0.00019999562607679193,
-      "loss": 46.0,
-      "step": 38949
-    },
-    {
-      "epoch": 2.977999503029608,
-      "grad_norm": 0.003829419380053878,
-      "learning_rate": 0.00019999562585213556,
-      "loss": 46.0,
-      "step": 38950
-    },
-    {
-      "epoch": 2.9780759600129976,
-      "grad_norm": 0.0049112578853964806,
-      "learning_rate": 0.0001999956256274734,
-      "loss": 46.0,
-      "step": 38951
-    },
-    {
-      "epoch": 2.9781524169963873,
-      "grad_norm": 0.00432144571095705,
-      "learning_rate": 0.00019999562540280548,
-      "loss": 46.0,
-      "step": 38952
-    },
-    {
-      "epoch": 2.978228873979777,
-      "grad_norm": 0.0030643471982330084,
-      "learning_rate": 0.0001999956251781318,
-      "loss": 46.0,
-      "step": 38953
-    },
-    {
-      "epoch": 2.978305330963167,
-      "grad_norm": 0.002610486000776291,
-      "learning_rate": 0.00019999562495345234,
-      "loss": 46.0,
-      "step": 38954
-    },
-    {
-      "epoch": 2.9783817879465566,
-      "grad_norm": 0.001192101277410984,
-      "learning_rate": 0.0001999956247287671,
-      "loss": 46.0,
-      "step": 38955
-    },
-    {
-      "epoch": 2.9784582449299464,
-      "grad_norm": 0.0018854161025956273,
-      "learning_rate": 0.00019999562450407612,
-      "loss": 46.0,
-      "step": 38956
-    },
-    {
-      "epoch": 2.978534701913336,
-      "grad_norm": 0.001914657885208726,
-      "learning_rate": 0.00019999562427937932,
-      "loss": 46.0,
-      "step": 38957
-    },
-    {
-      "epoch": 2.978611158896726,
-      "grad_norm": 0.0024616485461592674,
-      "learning_rate": 0.0001999956240546768,
-      "loss": 46.0,
-      "step": 38958
-    },
-    {
-      "epoch": 2.9786876158801157,
-      "grad_norm": 0.0013303844025358558,
-      "learning_rate": 0.00019999562382996852,
-      "loss": 46.0,
-      "step": 38959
-    },
-    {
-      "epoch": 2.9787640728635054,
-      "grad_norm": 0.002290409291163087,
-      "learning_rate": 0.00019999562360525442,
-      "loss": 46.0,
-      "step": 38960
-    },
-    {
-      "epoch": 2.9788405298468947,
-      "grad_norm": 0.001430421369150281,
-      "learning_rate": 0.00019999562338053457,
-      "loss": 46.0,
-      "step": 38961
-    },
-    {
-      "epoch": 2.9789169868302845,
-      "grad_norm": 0.0009139743633568287,
-      "learning_rate": 0.00019999562315580895,
-      "loss": 46.0,
-      "step": 38962
-    },
-    {
-      "epoch": 2.9789934438136743,
-      "grad_norm": 0.001719501568004489,
-      "learning_rate": 0.00019999562293107756,
-      "loss": 46.0,
-      "step": 38963
-    },
-    {
-      "epoch": 2.979069900797064,
-      "grad_norm": 0.0027095619589090347,
-      "learning_rate": 0.00019999562270634042,
-      "loss": 46.0,
-      "step": 38964
-    },
-    {
-      "epoch": 2.979146357780454,
-      "grad_norm": 0.0020940606482326984,
-      "learning_rate": 0.00019999562248159748,
-      "loss": 46.0,
-      "step": 38965
-    },
-    {
-      "epoch": 2.9792228147638435,
-      "grad_norm": 0.002448778599500656,
-      "learning_rate": 0.0001999956222568488,
-      "loss": 46.0,
-      "step": 38966
-    },
-    {
-      "epoch": 2.9792992717472333,
-      "grad_norm": 0.0009370583575218916,
-      "learning_rate": 0.00019999562203209436,
-      "loss": 46.0,
-      "step": 38967
-    },
-    {
-      "epoch": 2.979375728730623,
-      "grad_norm": 0.004019725136458874,
-      "learning_rate": 0.00019999562180733412,
-      "loss": 46.0,
-      "step": 38968
-    },
-    {
-      "epoch": 2.9794521857140124,
-      "grad_norm": 0.003771245013922453,
-      "learning_rate": 0.00019999562158256812,
-      "loss": 46.0,
-      "step": 38969
-    },
-    {
-      "epoch": 2.979528642697402,
-      "grad_norm": 0.0008153815288096666,
-      "learning_rate": 0.00019999562135779634,
-      "loss": 46.0,
-      "step": 38970
-    },
-    {
-      "epoch": 2.979605099680792,
-      "grad_norm": 0.0019376752898097038,
-      "learning_rate": 0.00019999562113301878,
-      "loss": 46.0,
-      "step": 38971
-    },
-    {
-      "epoch": 2.9796815566641817,
-      "grad_norm": 0.0027577003929764032,
-      "learning_rate": 0.00019999562090823548,
-      "loss": 46.0,
-      "step": 38972
-    },
-    {
-      "epoch": 2.9797580136475714,
-      "grad_norm": 0.003916645888239145,
-      "learning_rate": 0.0001999956206834464,
-      "loss": 46.0,
-      "step": 38973
-    },
-    {
-      "epoch": 2.979834470630961,
-      "grad_norm": 0.002174672205001116,
-      "learning_rate": 0.00019999562045865156,
-      "loss": 46.0,
-      "step": 38974
-    },
-    {
-      "epoch": 2.979910927614351,
-      "grad_norm": 0.006645541172474623,
-      "learning_rate": 0.00019999562023385097,
-      "loss": 46.0,
-      "step": 38975
-    },
-    {
-      "epoch": 2.9799873845977407,
-      "grad_norm": 0.0017266924260184169,
-      "learning_rate": 0.00019999562000904458,
-      "loss": 46.0,
-      "step": 38976
-    },
-    {
-      "epoch": 2.9800638415811305,
-      "grad_norm": 0.002607109025120735,
-      "learning_rate": 0.0001999956197842324,
-      "loss": 46.0,
-      "step": 38977
-    },
-    {
-      "epoch": 2.9801402985645202,
-      "grad_norm": 0.0019106668187305331,
-      "learning_rate": 0.0001999956195594145,
-      "loss": 46.0,
-      "step": 38978
-    },
-    {
-      "epoch": 2.98021675554791,
-      "grad_norm": 0.004048858769237995,
-      "learning_rate": 0.00019999561933459079,
-      "loss": 46.0,
-      "step": 38979
-    },
-    {
-      "epoch": 2.9802932125312998,
-      "grad_norm": 0.0015311338938772678,
-      "learning_rate": 0.00019999561910976135,
-      "loss": 46.0,
-      "step": 38980
-    },
-    {
-      "epoch": 2.9803696695146895,
-      "grad_norm": 0.0006093823467381299,
-      "learning_rate": 0.00019999561888492612,
-      "loss": 46.0,
-      "step": 38981
-    },
-    {
-      "epoch": 2.9804461264980793,
-      "grad_norm": 0.0028479464817792177,
-      "learning_rate": 0.00019999561866008511,
-      "loss": 46.0,
-      "step": 38982
-    },
-    {
-      "epoch": 2.9805225834814686,
-      "grad_norm": 0.002826940966770053,
-      "learning_rate": 0.00019999561843523833,
-      "loss": 46.0,
-      "step": 38983
-    },
-    {
-      "epoch": 2.9805990404648584,
-      "grad_norm": 0.0030750504229217768,
-      "learning_rate": 0.00019999561821038578,
-      "loss": 46.0,
-      "step": 38984
-    },
-    {
-      "epoch": 2.980675497448248,
-      "grad_norm": 0.0016807789215818048,
-      "learning_rate": 0.0001999956179855275,
-      "loss": 46.0,
-      "step": 38985
-    },
-    {
-      "epoch": 2.980751954431638,
-      "grad_norm": 0.0015888669295236468,
-      "learning_rate": 0.0001999956177606634,
-      "loss": 46.0,
-      "step": 38986
-    },
-    {
-      "epoch": 2.9808284114150276,
-      "grad_norm": 0.0010277723195031285,
-      "learning_rate": 0.0001999956175357936,
-      "loss": 46.0,
-      "step": 38987
-    },
-    {
-      "epoch": 2.9809048683984174,
-      "grad_norm": 0.002044037217274308,
-      "learning_rate": 0.00019999561731091797,
-      "loss": 46.0,
-      "step": 38988
-    },
-    {
-      "epoch": 2.980981325381807,
-      "grad_norm": 0.0021695501636713743,
-      "learning_rate": 0.00019999561708603658,
-      "loss": 46.0,
-      "step": 38989
-    },
-    {
-      "epoch": 2.981057782365197,
-      "grad_norm": 0.0012121469480916858,
-      "learning_rate": 0.00019999561686114944,
-      "loss": 46.0,
-      "step": 38990
-    },
-    {
-      "epoch": 2.9811342393485862,
-      "grad_norm": 0.005473044700920582,
-      "learning_rate": 0.00019999561663625653,
-      "loss": 46.0,
-      "step": 38991
-    },
-    {
-      "epoch": 2.981210696331976,
-      "grad_norm": 0.001451272633858025,
-      "learning_rate": 0.0001999956164113578,
-      "loss": 46.0,
-      "step": 38992
-    },
-    {
-      "epoch": 2.9812871533153658,
-      "grad_norm": 0.006416152697056532,
-      "learning_rate": 0.00019999561618645335,
-      "loss": 46.0,
-      "step": 38993
-    },
-    {
-      "epoch": 2.9813636102987555,
-      "grad_norm": 0.0032489148434251547,
-      "learning_rate": 0.00019999561596154312,
-      "loss": 46.0,
-      "step": 38994
-    },
-    {
-      "epoch": 2.9814400672821453,
-      "grad_norm": 0.002617703750729561,
-      "learning_rate": 0.00019999561573662712,
-      "loss": 46.0,
-      "step": 38995
-    },
-    {
-      "epoch": 2.981516524265535,
-      "grad_norm": 0.0008853967301547527,
-      "learning_rate": 0.00019999561551170536,
-      "loss": 46.0,
-      "step": 38996
-    },
-    {
-      "epoch": 2.981592981248925,
-      "grad_norm": 0.003279887605458498,
-      "learning_rate": 0.0001999956152867778,
-      "loss": 46.0,
-      "step": 38997
-    },
-    {
-      "epoch": 2.9816694382323146,
-      "grad_norm": 0.0017831254517659545,
-      "learning_rate": 0.0001999956150618445,
-      "loss": 46.0,
-      "step": 38998
-    },
-    {
-      "epoch": 2.9817458952157043,
-      "grad_norm": 0.0017559000989422202,
-      "learning_rate": 0.00019999561483690547,
-      "loss": 46.0,
-      "step": 38999
-    },
-    {
-      "epoch": 2.981822352199094,
-      "grad_norm": 0.0021074169781059027,
-      "learning_rate": 0.0001999956146119606,
-      "loss": 46.0,
-      "step": 39000
-    },
-    {
-      "epoch": 2.981898809182484,
-      "grad_norm": 0.0014945173170417547,
-      "learning_rate": 0.00019999561438700998,
-      "loss": 46.0,
-      "step": 39001
-    },
-    {
-      "epoch": 2.9819752661658736,
-      "grad_norm": 0.000690289365593344,
-      "learning_rate": 0.00019999561416205358,
-      "loss": 46.0,
-      "step": 39002
-    },
-    {
-      "epoch": 2.9820517231492634,
-      "grad_norm": 0.0010788417421281338,
-      "learning_rate": 0.00019999561393709144,
-      "loss": 46.0,
-      "step": 39003
-    },
-    {
-      "epoch": 2.982128180132653,
-      "grad_norm": 0.001646575634367764,
-      "learning_rate": 0.00019999561371212353,
-      "loss": 46.0,
-      "step": 39004
-    },
-    {
-      "epoch": 2.9822046371160424,
-      "grad_norm": 0.0013893601717427373,
-      "learning_rate": 0.00019999561348714985,
-      "loss": 46.0,
-      "step": 39005
-    },
-    {
-      "epoch": 2.982281094099432,
-      "grad_norm": 0.0010865854565054178,
-      "learning_rate": 0.00019999561326217036,
-      "loss": 46.0,
-      "step": 39006
-    },
-    {
-      "epoch": 2.982357551082822,
-      "grad_norm": 0.003971052821725607,
-      "learning_rate": 0.00019999561303718513,
-      "loss": 46.0,
-      "step": 39007
-    },
-    {
-      "epoch": 2.9824340080662117,
-      "grad_norm": 0.0012208069674670696,
-      "learning_rate": 0.00019999561281219415,
-      "loss": 46.0,
-      "step": 39008
-    },
-    {
-      "epoch": 2.9825104650496015,
-      "grad_norm": 0.0006932918913662434,
-      "learning_rate": 0.0001999956125871974,
-      "loss": 46.0,
-      "step": 39009
-    },
-    {
-      "epoch": 2.9825869220329913,
-      "grad_norm": 0.0016632204642519355,
-      "learning_rate": 0.00019999561236219485,
-      "loss": 46.0,
-      "step": 39010
-    },
-    {
-      "epoch": 2.982663379016381,
-      "grad_norm": 0.0022916116286069155,
-      "learning_rate": 0.00019999561213718652,
-      "loss": 46.0,
-      "step": 39011
-    },
-    {
-      "epoch": 2.9827398359997708,
-      "grad_norm": 0.002489052014425397,
-      "learning_rate": 0.00019999561191217247,
-      "loss": 46.0,
-      "step": 39012
-    },
-    {
-      "epoch": 2.98281629298316,
-      "grad_norm": 0.003345897188410163,
-      "learning_rate": 0.00019999561168715263,
-      "loss": 46.0,
-      "step": 39013
-    },
-    {
-      "epoch": 2.98289274996655,
-      "grad_norm": 0.0013521400978788733,
-      "learning_rate": 0.000199995611462127,
-      "loss": 46.0,
-      "step": 39014
-    },
-    {
-      "epoch": 2.9829692069499396,
-      "grad_norm": 0.0028718295507133007,
-      "learning_rate": 0.00019999561123709562,
-      "loss": 46.0,
-      "step": 39015
-    },
-    {
-      "epoch": 2.9830456639333294,
-      "grad_norm": 0.0015655282186344266,
-      "learning_rate": 0.00019999561101205845,
-      "loss": 46.0,
-      "step": 39016
-    },
-    {
-      "epoch": 2.983122120916719,
-      "grad_norm": 0.005040094256401062,
-      "learning_rate": 0.00019999561078701554,
-      "loss": 46.0,
-      "step": 39017
-    },
-    {
-      "epoch": 2.983198577900109,
-      "grad_norm": 0.00428294250741601,
-      "learning_rate": 0.00019999561056196686,
-      "loss": 46.0,
-      "step": 39018
-    },
-    {
-      "epoch": 2.9832750348834987,
-      "grad_norm": 0.0013332576490938663,
-      "learning_rate": 0.0001999956103369124,
-      "loss": 46.0,
-      "step": 39019
-    },
-    {
-      "epoch": 2.9833514918668884,
-      "grad_norm": 0.002283480018377304,
-      "learning_rate": 0.00019999561011185216,
-      "loss": 46.0,
-      "step": 39020
-    },
-    {
-      "epoch": 2.983427948850278,
-      "grad_norm": 0.0005546375177800655,
-      "learning_rate": 0.00019999560988678616,
-      "loss": 46.0,
-      "step": 39021
-    },
-    {
-      "epoch": 2.983504405833668,
-      "grad_norm": 0.0013504207599908113,
-      "learning_rate": 0.0001999956096617144,
-      "loss": 46.0,
-      "step": 39022
-    },
-    {
-      "epoch": 2.9835808628170577,
-      "grad_norm": 0.0029268113430589437,
-      "learning_rate": 0.00019999560943663686,
-      "loss": 46.0,
-      "step": 39023
-    },
-    {
-      "epoch": 2.9836573198004475,
-      "grad_norm": 0.0067323134280741215,
-      "learning_rate": 0.00019999560921155356,
-      "loss": 46.0,
-      "step": 39024
-    },
-    {
-      "epoch": 2.983733776783837,
-      "grad_norm": 0.001047409139573574,
-      "learning_rate": 0.00019999560898646449,
-      "loss": 46.0,
-      "step": 39025
-    },
-    {
-      "epoch": 2.983810233767227,
-      "grad_norm": 0.0019587792921811342,
-      "learning_rate": 0.00019999560876136964,
-      "loss": 46.0,
-      "step": 39026
-    },
-    {
-      "epoch": 2.9838866907506163,
-      "grad_norm": 0.0014166788896545768,
-      "learning_rate": 0.00019999560853626902,
-      "loss": 46.0,
-      "step": 39027
-    },
-    {
-      "epoch": 2.983963147734006,
-      "grad_norm": 0.0016008763341233134,
-      "learning_rate": 0.00019999560831116263,
-      "loss": 46.0,
-      "step": 39028
-    },
-    {
-      "epoch": 2.984039604717396,
-      "grad_norm": 0.001029292936436832,
-      "learning_rate": 0.0001999956080860505,
-      "loss": 46.0,
-      "step": 39029
-    },
-    {
-      "epoch": 2.9841160617007856,
-      "grad_norm": 0.003147426061332226,
-      "learning_rate": 0.00019999560786093255,
-      "loss": 46.0,
-      "step": 39030
-    },
-    {
-      "epoch": 2.9841925186841753,
-      "grad_norm": 0.0026365185622125864,
-      "learning_rate": 0.0001999956076358089,
-      "loss": 46.0,
-      "step": 39031
-    },
-    {
-      "epoch": 2.984268975667565,
-      "grad_norm": 0.00127028813585639,
-      "learning_rate": 0.0001999956074106794,
-      "loss": 46.0,
-      "step": 39032
-    },
-    {
-      "epoch": 2.984345432650955,
-      "grad_norm": 0.005030323751270771,
-      "learning_rate": 0.00019999560718554418,
-      "loss": 46.0,
-      "step": 39033
-    },
-    {
-      "epoch": 2.9844218896343446,
-      "grad_norm": 0.00176139990799129,
-      "learning_rate": 0.0001999956069604032,
-      "loss": 46.0,
-      "step": 39034
-    },
-    {
-      "epoch": 2.984498346617734,
-      "grad_norm": 0.0024201814085245132,
-      "learning_rate": 0.00019999560673525642,
-      "loss": 46.0,
-      "step": 39035
-    },
-    {
-      "epoch": 2.9845748036011237,
-      "grad_norm": 0.004769555758684874,
-      "learning_rate": 0.0001999956065101039,
-      "loss": 46.0,
-      "step": 39036
-    },
-    {
-      "epoch": 2.9846512605845135,
-      "grad_norm": 0.0018084713956341147,
-      "learning_rate": 0.00019999560628494557,
-      "loss": 46.0,
-      "step": 39037
-    },
-    {
-      "epoch": 2.9847277175679032,
-      "grad_norm": 0.0020056227222085,
-      "learning_rate": 0.00019999560605978153,
-      "loss": 46.0,
-      "step": 39038
-    },
-    {
-      "epoch": 2.984804174551293,
-      "grad_norm": 0.0020144127774983644,
-      "learning_rate": 0.00019999560583461166,
-      "loss": 46.0,
-      "step": 39039
-    },
-    {
-      "epoch": 2.9848806315346827,
-      "grad_norm": 0.0012094862759113312,
-      "learning_rate": 0.00019999560560943604,
-      "loss": 46.0,
-      "step": 39040
-    },
-    {
-      "epoch": 2.9849570885180725,
-      "grad_norm": 0.0014625868061557412,
-      "learning_rate": 0.00019999560538425465,
-      "loss": 46.0,
-      "step": 39041
-    },
-    {
-      "epoch": 2.9850335455014623,
-      "grad_norm": 0.0036340290680527687,
-      "learning_rate": 0.00019999560515906754,
-      "loss": 46.0,
-      "step": 39042
-    },
-    {
-      "epoch": 2.985110002484852,
-      "grad_norm": 0.0032362372148782015,
-      "learning_rate": 0.00019999560493387462,
-      "loss": 46.0,
-      "step": 39043
-    },
-    {
-      "epoch": 2.985186459468242,
-      "grad_norm": 0.0012778709642589092,
-      "learning_rate": 0.0001999956047086759,
-      "loss": 46.0,
-      "step": 39044
-    },
-    {
-      "epoch": 2.9852629164516316,
-      "grad_norm": 0.0017344921361654997,
-      "learning_rate": 0.00019999560448347146,
-      "loss": 46.0,
-      "step": 39045
-    },
-    {
-      "epoch": 2.9853393734350213,
-      "grad_norm": 0.0011550289345905185,
-      "learning_rate": 0.00019999560425826122,
-      "loss": 46.0,
-      "step": 39046
-    },
-    {
-      "epoch": 2.985415830418411,
-      "grad_norm": 0.001096539432182908,
-      "learning_rate": 0.00019999560403304522,
-      "loss": 46.0,
-      "step": 39047
-    },
-    {
-      "epoch": 2.985492287401801,
-      "grad_norm": 0.002185060642659664,
-      "learning_rate": 0.00019999560380782344,
-      "loss": 46.0,
-      "step": 39048
-    },
-    {
-      "epoch": 2.98556874438519,
-      "grad_norm": 0.00213637319393456,
-      "learning_rate": 0.00019999560358259592,
-      "loss": 46.0,
-      "step": 39049
-    },
-    {
-      "epoch": 2.98564520136858,
-      "grad_norm": 0.004692373797297478,
-      "learning_rate": 0.00019999560335736262,
-      "loss": 46.0,
-      "step": 39050
-    },
-    {
-      "epoch": 2.9857216583519697,
-      "grad_norm": 0.0043512689881026745,
-      "learning_rate": 0.00019999560313212355,
-      "loss": 46.0,
-      "step": 39051
-    },
-    {
-      "epoch": 2.9857981153353594,
-      "grad_norm": 0.0006530076498165727,
-      "learning_rate": 0.0001999956029068787,
-      "loss": 46.0,
-      "step": 39052
-    },
-    {
-      "epoch": 2.985874572318749,
-      "grad_norm": 0.002633305499330163,
-      "learning_rate": 0.00019999560268162809,
-      "loss": 46.0,
-      "step": 39053
-    },
-    {
-      "epoch": 2.985951029302139,
-      "grad_norm": 0.004088290501385927,
-      "learning_rate": 0.0001999956024563717,
-      "loss": 46.0,
-      "step": 39054
-    },
-    {
-      "epoch": 2.9860274862855287,
-      "grad_norm": 0.0008660958847030997,
-      "learning_rate": 0.00019999560223110956,
-      "loss": 46.0,
-      "step": 39055
-    },
-    {
-      "epoch": 2.9861039432689185,
-      "grad_norm": 0.0036324551329016685,
-      "learning_rate": 0.00019999560200584165,
-      "loss": 46.0,
-      "step": 39056
-    },
-    {
-      "epoch": 2.986180400252308,
-      "grad_norm": 0.004912384785711765,
-      "learning_rate": 0.00019999560178056793,
-      "loss": 46.0,
-      "step": 39057
-    },
-    {
-      "epoch": 2.9862568572356976,
-      "grad_norm": 0.0004007123352494091,
-      "learning_rate": 0.0001999956015552885,
-      "loss": 46.0,
-      "step": 39058
-    },
-    {
-      "epoch": 2.9863333142190873,
-      "grad_norm": 0.000838922627735883,
-      "learning_rate": 0.00019999560133000325,
-      "loss": 46.0,
-      "step": 39059
-    },
-    {
-      "epoch": 2.986409771202477,
-      "grad_norm": 0.00048119647544808686,
-      "learning_rate": 0.00019999560110471224,
-      "loss": 46.0,
-      "step": 39060
-    },
-    {
-      "epoch": 2.986486228185867,
-      "grad_norm": 0.0006156764575280249,
-      "learning_rate": 0.0001999956008794155,
-      "loss": 46.0,
-      "step": 39061
-    },
-    {
-      "epoch": 2.9865626851692566,
-      "grad_norm": 0.0016523554222658277,
-      "learning_rate": 0.00019999560065411297,
-      "loss": 46.0,
-      "step": 39062
-    },
-    {
-      "epoch": 2.9866391421526464,
-      "grad_norm": 0.0032914048060774803,
-      "learning_rate": 0.00019999560042880464,
-      "loss": 46.0,
-      "step": 39063
-    },
-    {
-      "epoch": 2.986715599136036,
-      "grad_norm": 0.0030514118261635303,
-      "learning_rate": 0.00019999560020349057,
-      "loss": 46.0,
-      "step": 39064
-    },
-    {
-      "epoch": 2.986792056119426,
-      "grad_norm": 0.0010742873419076204,
-      "learning_rate": 0.00019999559997817075,
-      "loss": 46.0,
-      "step": 39065
-    },
-    {
-      "epoch": 2.9868685131028156,
-      "grad_norm": 0.00061345926951617,
-      "learning_rate": 0.00019999559975284514,
-      "loss": 46.0,
-      "step": 39066
-    },
-    {
-      "epoch": 2.9869449700862054,
-      "grad_norm": 0.002510753460228443,
-      "learning_rate": 0.00019999559952751375,
-      "loss": 46.0,
-      "step": 39067
-    },
-    {
-      "epoch": 2.987021427069595,
-      "grad_norm": 0.0032902169041335583,
-      "learning_rate": 0.00019999559930217658,
-      "loss": 46.0,
-      "step": 39068
-    },
-    {
-      "epoch": 2.987097884052985,
-      "grad_norm": 0.0007933829328976572,
-      "learning_rate": 0.00019999559907683367,
-      "loss": 46.0,
-      "step": 39069
-    },
-    {
-      "epoch": 2.9871743410363742,
-      "grad_norm": 0.0035547392908483744,
-      "learning_rate": 0.000199995598851485,
-      "loss": 46.0,
-      "step": 39070
-    },
-    {
-      "epoch": 2.987250798019764,
-      "grad_norm": 0.0028582008089870214,
-      "learning_rate": 0.00019999559862613053,
-      "loss": 46.0,
-      "step": 39071
-    },
-    {
-      "epoch": 2.9873272550031538,
-      "grad_norm": 0.001642982941120863,
-      "learning_rate": 0.0001999955984007703,
-      "loss": 46.0,
-      "step": 39072
-    },
-    {
-      "epoch": 2.9874037119865435,
-      "grad_norm": 0.0016685270238667727,
-      "learning_rate": 0.0001999955981754043,
-      "loss": 46.0,
-      "step": 39073
-    },
-    {
-      "epoch": 2.9874801689699333,
-      "grad_norm": 0.0021984437480568886,
-      "learning_rate": 0.00019999559795003255,
-      "loss": 46.0,
-      "step": 39074
-    },
-    {
-      "epoch": 2.987556625953323,
-      "grad_norm": 0.0008019302622415125,
-      "learning_rate": 0.000199995597724655,
-      "loss": 46.0,
-      "step": 39075
-    },
-    {
-      "epoch": 2.987633082936713,
-      "grad_norm": 0.004258256405591965,
-      "learning_rate": 0.0001999955974992717,
-      "loss": 46.0,
-      "step": 39076
-    },
-    {
-      "epoch": 2.9877095399201026,
-      "grad_norm": 0.00825955718755722,
-      "learning_rate": 0.00019999559727388263,
-      "loss": 46.0,
-      "step": 39077
-    },
-    {
-      "epoch": 2.9877859969034923,
-      "grad_norm": 0.0016950754215940833,
-      "learning_rate": 0.00019999559704848778,
-      "loss": 46.0,
-      "step": 39078
-    },
-    {
-      "epoch": 2.9878624538868817,
-      "grad_norm": 0.0013112345477566123,
-      "learning_rate": 0.00019999559682308717,
-      "loss": 46.0,
-      "step": 39079
-    },
-    {
-      "epoch": 2.9879389108702714,
-      "grad_norm": 0.003756336634978652,
-      "learning_rate": 0.0001999955965976808,
-      "loss": 46.0,
-      "step": 39080
-    },
-    {
-      "epoch": 2.988015367853661,
-      "grad_norm": 0.0021658686455339193,
-      "learning_rate": 0.00019999559637226867,
-      "loss": 46.0,
-      "step": 39081
-    },
-    {
-      "epoch": 2.988091824837051,
-      "grad_norm": 0.0024607155937701464,
-      "learning_rate": 0.0001999955961468507,
-      "loss": 46.0,
-      "step": 39082
-    },
-    {
-      "epoch": 2.9881682818204407,
-      "grad_norm": 0.0009024886530824006,
-      "learning_rate": 0.00019999559592142702,
-      "loss": 46.0,
-      "step": 39083
-    },
-    {
-      "epoch": 2.9882447388038305,
-      "grad_norm": 0.0008100476115942001,
-      "learning_rate": 0.00019999559569599756,
-      "loss": 46.0,
-      "step": 39084
-    },
-    {
-      "epoch": 2.98832119578722,
-      "grad_norm": 0.0016449899412691593,
-      "learning_rate": 0.00019999559547056236,
-      "loss": 46.0,
-      "step": 39085
-    },
-    {
-      "epoch": 2.98839765277061,
-      "grad_norm": 0.002241989132016897,
-      "learning_rate": 0.00019999559524512135,
-      "loss": 46.0,
-      "step": 39086
-    },
-    {
-      "epoch": 2.9884741097539997,
-      "grad_norm": 0.0010325298644602299,
-      "learning_rate": 0.00019999559501967458,
-      "loss": 46.0,
-      "step": 39087
-    },
-    {
-      "epoch": 2.9885505667373895,
-      "grad_norm": 0.0014504775172099471,
-      "learning_rate": 0.00019999559479422205,
-      "loss": 46.0,
-      "step": 39088
-    },
-    {
-      "epoch": 2.9886270237207793,
-      "grad_norm": 0.002304642926901579,
-      "learning_rate": 0.00019999559456876376,
-      "loss": 46.0,
-      "step": 39089
-    },
-    {
-      "epoch": 2.988703480704169,
-      "grad_norm": 0.003274294314906001,
-      "learning_rate": 0.00019999559434329966,
-      "loss": 46.0,
-      "step": 39090
-    },
-    {
-      "epoch": 2.988779937687559,
-      "grad_norm": 0.0023318196181207895,
-      "learning_rate": 0.00019999559411782984,
-      "loss": 46.0,
-      "step": 39091
-    },
-    {
-      "epoch": 2.988856394670948,
-      "grad_norm": 0.0016399105079472065,
-      "learning_rate": 0.00019999559389235423,
-      "loss": 46.0,
-      "step": 39092
-    },
-    {
-      "epoch": 2.988932851654338,
-      "grad_norm": 0.0008155826944857836,
-      "learning_rate": 0.00019999559366687283,
-      "loss": 46.0,
-      "step": 39093
-    },
-    {
-      "epoch": 2.9890093086377276,
-      "grad_norm": 0.0013628797605633736,
-      "learning_rate": 0.0001999955934413857,
-      "loss": 46.0,
-      "step": 39094
-    },
-    {
-      "epoch": 2.9890857656211174,
-      "grad_norm": 0.0020880845841020346,
-      "learning_rate": 0.0001999955932158928,
-      "loss": 46.0,
-      "step": 39095
-    },
-    {
-      "epoch": 2.989162222604507,
-      "grad_norm": 0.0018020865973085165,
-      "learning_rate": 0.0001999955929903941,
-      "loss": 46.0,
-      "step": 39096
-    },
-    {
-      "epoch": 2.989238679587897,
-      "grad_norm": 0.013506615534424782,
-      "learning_rate": 0.00019999559276488965,
-      "loss": 46.0,
-      "step": 39097
-    },
-    {
-      "epoch": 2.9893151365712867,
-      "grad_norm": 0.001906516612507403,
-      "learning_rate": 0.00019999559253937942,
-      "loss": 46.0,
-      "step": 39098
-    },
-    {
-      "epoch": 2.9893915935546764,
-      "grad_norm": 0.0038871723227202892,
-      "learning_rate": 0.0001999955923138634,
-      "loss": 46.0,
-      "step": 39099
-    },
-    {
-      "epoch": 2.9894680505380657,
-      "grad_norm": 0.00223262095823884,
-      "learning_rate": 0.00019999559208834164,
-      "loss": 46.0,
-      "step": 39100
-    },
-    {
-      "epoch": 2.9895445075214555,
-      "grad_norm": 0.003633759682998061,
-      "learning_rate": 0.00019999559186281411,
-      "loss": 46.0,
-      "step": 39101
-    },
-    {
-      "epoch": 2.9896209645048453,
-      "grad_norm": 0.001774475211277604,
-      "learning_rate": 0.0001999955916372808,
-      "loss": 46.0,
-      "step": 39102
-    },
-    {
-      "epoch": 2.989697421488235,
-      "grad_norm": 0.0011606712359935045,
-      "learning_rate": 0.00019999559141174172,
-      "loss": 46.0,
-      "step": 39103
-    },
-    {
-      "epoch": 2.989773878471625,
-      "grad_norm": 0.00431804871186614,
-      "learning_rate": 0.00019999559118619688,
-      "loss": 46.0,
-      "step": 39104
-    },
-    {
-      "epoch": 2.9898503354550146,
-      "grad_norm": 0.0031206090934574604,
-      "learning_rate": 0.00019999559096064626,
-      "loss": 46.0,
-      "step": 39105
-    },
-    {
-      "epoch": 2.9899267924384043,
-      "grad_norm": 0.0041432734578847885,
-      "learning_rate": 0.0001999955907350899,
-      "loss": 46.0,
-      "step": 39106
-    },
-    {
-      "epoch": 2.990003249421794,
-      "grad_norm": 0.001730704098008573,
-      "learning_rate": 0.00019999559050952774,
-      "loss": 46.0,
-      "step": 39107
-    },
-    {
-      "epoch": 2.990079706405184,
-      "grad_norm": 0.0036157353315502405,
-      "learning_rate": 0.00019999559028395983,
-      "loss": 46.0,
-      "step": 39108
-    },
-    {
-      "epoch": 2.9901561633885736,
-      "grad_norm": 0.003779682796448469,
-      "learning_rate": 0.00019999559005838612,
-      "loss": 46.0,
-      "step": 39109
-    },
-    {
-      "epoch": 2.9902326203719634,
-      "grad_norm": 0.005584088619798422,
-      "learning_rate": 0.0001999955898328067,
-      "loss": 46.0,
-      "step": 39110
-    },
-    {
-      "epoch": 2.990309077355353,
-      "grad_norm": 0.001936267246492207,
-      "learning_rate": 0.00019999558960722146,
-      "loss": 46.0,
-      "step": 39111
-    },
-    {
-      "epoch": 2.990385534338743,
-      "grad_norm": 0.0018537326250225306,
-      "learning_rate": 0.00019999558938163045,
-      "loss": 46.0,
-      "step": 39112
-    },
-    {
-      "epoch": 2.9904619913221326,
-      "grad_norm": 0.0011256610741838813,
-      "learning_rate": 0.0001999955891560337,
-      "loss": 46.0,
-      "step": 39113
-    },
-    {
-      "epoch": 2.990538448305522,
-      "grad_norm": 0.0022534318268299103,
-      "learning_rate": 0.00019999558893043115,
-      "loss": 46.0,
-      "step": 39114
-    },
-    {
-      "epoch": 2.9906149052889117,
-      "grad_norm": 0.0030199645552784204,
-      "learning_rate": 0.00019999558870482286,
-      "loss": 46.0,
-      "step": 39115
-    },
-    {
-      "epoch": 2.9906913622723015,
-      "grad_norm": 0.0017322596395388246,
-      "learning_rate": 0.00019999558847920882,
-      "loss": 46.0,
-      "step": 39116
-    },
-    {
-      "epoch": 2.9907678192556912,
-      "grad_norm": 0.0009440616704523563,
-      "learning_rate": 0.00019999558825358895,
-      "loss": 46.0,
-      "step": 39117
-    },
-    {
-      "epoch": 2.990844276239081,
-      "grad_norm": 0.001366412965580821,
-      "learning_rate": 0.00019999558802796336,
-      "loss": 46.0,
-      "step": 39118
-    },
-    {
-      "epoch": 2.9909207332224708,
-      "grad_norm": 0.002148016355931759,
-      "learning_rate": 0.00019999558780233197,
-      "loss": 46.0,
-      "step": 39119
-    },
-    {
-      "epoch": 2.9909971902058605,
-      "grad_norm": 0.0017213055398315191,
-      "learning_rate": 0.0001999955875766948,
-      "loss": 46.0,
-      "step": 39120
-    },
-    {
-      "epoch": 2.9910736471892503,
-      "grad_norm": 0.008067573420703411,
-      "learning_rate": 0.0001999955873510519,
-      "loss": 46.0,
-      "step": 39121
-    },
-    {
-      "epoch": 2.9911501041726396,
-      "grad_norm": 0.0031846011988818645,
-      "learning_rate": 0.0001999955871254032,
-      "loss": 46.0,
-      "step": 39122
-    },
-    {
-      "epoch": 2.9912265611560294,
-      "grad_norm": 0.002131731016561389,
-      "learning_rate": 0.00019999558689974873,
-      "loss": 46.0,
-      "step": 39123
-    },
-    {
-      "epoch": 2.991303018139419,
-      "grad_norm": 0.00308710103854537,
-      "learning_rate": 0.00019999558667408853,
-      "loss": 46.0,
-      "step": 39124
-    },
-    {
-      "epoch": 2.991379475122809,
-      "grad_norm": 0.0021525833290070295,
-      "learning_rate": 0.00019999558644842252,
-      "loss": 46.0,
-      "step": 39125
-    },
-    {
-      "epoch": 2.9914559321061986,
-      "grad_norm": 0.015429322607815266,
-      "learning_rate": 0.00019999558622275077,
-      "loss": 46.0,
-      "step": 39126
-    },
-    {
-      "epoch": 2.9915323890895884,
-      "grad_norm": 0.001156373182311654,
-      "learning_rate": 0.00019999558599707325,
-      "loss": 46.0,
-      "step": 39127
-    },
-    {
-      "epoch": 2.991608846072978,
-      "grad_norm": 0.0013423626078292727,
-      "learning_rate": 0.00019999558577138993,
-      "loss": 46.0,
-      "step": 39128
-    },
-    {
-      "epoch": 2.991685303056368,
-      "grad_norm": 0.0023643705062568188,
-      "learning_rate": 0.00019999558554570086,
-      "loss": 46.0,
-      "step": 39129
-    },
-    {
-      "epoch": 2.9917617600397577,
-      "grad_norm": 0.0026027371641248465,
-      "learning_rate": 0.00019999558532000602,
-      "loss": 46.0,
-      "step": 39130
-    },
-    {
-      "epoch": 2.9918382170231475,
-      "grad_norm": 0.001721615088172257,
-      "learning_rate": 0.0001999955850943054,
-      "loss": 46.0,
-      "step": 39131
-    },
-    {
-      "epoch": 2.991914674006537,
-      "grad_norm": 0.0029915894847363234,
-      "learning_rate": 0.00019999558486859904,
-      "loss": 46.0,
-      "step": 39132
-    },
-    {
-      "epoch": 2.991991130989927,
-      "grad_norm": 0.0025686193257570267,
-      "learning_rate": 0.00019999558464288688,
-      "loss": 46.0,
-      "step": 39133
-    },
-    {
-      "epoch": 2.9920675879733167,
-      "grad_norm": 0.004422489088028669,
-      "learning_rate": 0.00019999558441716894,
-      "loss": 46.0,
-      "step": 39134
-    },
-    {
-      "epoch": 2.9921440449567065,
-      "grad_norm": 0.0026294011622667313,
-      "learning_rate": 0.00019999558419144529,
-      "loss": 46.0,
-      "step": 39135
-    },
-    {
-      "epoch": 2.992220501940096,
-      "grad_norm": 0.0013138698413968086,
-      "learning_rate": 0.0001999955839657158,
-      "loss": 46.0,
-      "step": 39136
-    },
-    {
-      "epoch": 2.9922969589234856,
-      "grad_norm": 0.003294616937637329,
-      "learning_rate": 0.0001999955837399806,
-      "loss": 46.0,
-      "step": 39137
-    },
-    {
-      "epoch": 2.9923734159068753,
-      "grad_norm": 0.001353780273348093,
-      "learning_rate": 0.0001999955835142396,
-      "loss": 46.0,
-      "step": 39138
-    },
-    {
-      "epoch": 2.992449872890265,
-      "grad_norm": 0.0018757219659164548,
-      "learning_rate": 0.00019999558328849282,
-      "loss": 46.0,
-      "step": 39139
-    },
-    {
-      "epoch": 2.992526329873655,
-      "grad_norm": 0.0035926576238125563,
-      "learning_rate": 0.0001999955830627403,
-      "loss": 46.0,
-      "step": 39140
-    },
-    {
-      "epoch": 2.9926027868570446,
-      "grad_norm": 0.0007921658107079566,
-      "learning_rate": 0.00019999558283698198,
-      "loss": 46.0,
-      "step": 39141
-    },
-    {
-      "epoch": 2.9926792438404344,
-      "grad_norm": 0.0017835624748840928,
-      "learning_rate": 0.0001999955826112179,
-      "loss": 46.0,
-      "step": 39142
-    },
-    {
-      "epoch": 2.992755700823824,
-      "grad_norm": 0.0005739748594351113,
-      "learning_rate": 0.0001999955823854481,
-      "loss": 46.0,
-      "step": 39143
-    },
-    {
-      "epoch": 2.9928321578072135,
-      "grad_norm": 0.003235763171687722,
-      "learning_rate": 0.00019999558215967248,
-      "loss": 46.0,
-      "step": 39144
-    },
-    {
-      "epoch": 2.992908614790603,
-      "grad_norm": 0.0022382240276783705,
-      "learning_rate": 0.0001999955819338911,
-      "loss": 46.0,
-      "step": 39145
-    },
-    {
-      "epoch": 2.992985071773993,
-      "grad_norm": 0.0006428069318644702,
-      "learning_rate": 0.00019999558170810395,
-      "loss": 46.0,
-      "step": 39146
-    },
-    {
-      "epoch": 2.9930615287573827,
-      "grad_norm": 0.0015342015540227294,
-      "learning_rate": 0.00019999558148231104,
-      "loss": 46.0,
-      "step": 39147
-    },
-    {
-      "epoch": 2.9931379857407725,
-      "grad_norm": 0.0017962598940357566,
-      "learning_rate": 0.00019999558125651234,
-      "loss": 46.0,
-      "step": 39148
-    },
-    {
-      "epoch": 2.9932144427241623,
-      "grad_norm": 0.0007452882127836347,
-      "learning_rate": 0.00019999558103070788,
-      "loss": 46.0,
-      "step": 39149
-    },
-    {
-      "epoch": 2.993290899707552,
-      "grad_norm": 0.002946908352896571,
-      "learning_rate": 0.00019999558080489765,
-      "loss": 46.0,
-      "step": 39150
-    },
-    {
-      "epoch": 2.993367356690942,
-      "grad_norm": 0.0026203941088169813,
-      "learning_rate": 0.00019999558057908168,
-      "loss": 46.0,
-      "step": 39151
-    },
-    {
-      "epoch": 2.9934438136743315,
-      "grad_norm": 0.0013671743217855692,
-      "learning_rate": 0.00019999558035325993,
-      "loss": 46.0,
-      "step": 39152
-    },
-    {
-      "epoch": 2.9935202706577213,
-      "grad_norm": 0.0020278203301131725,
-      "learning_rate": 0.00019999558012743238,
-      "loss": 46.0,
-      "step": 39153
-    },
-    {
-      "epoch": 2.993596727641111,
-      "grad_norm": 0.0028306825552135706,
-      "learning_rate": 0.00019999557990159908,
-      "loss": 46.0,
-      "step": 39154
-    },
-    {
-      "epoch": 2.993673184624501,
-      "grad_norm": 0.0010586519492790103,
-      "learning_rate": 0.00019999557967576,
-      "loss": 46.0,
-      "step": 39155
-    },
-    {
-      "epoch": 2.9937496416078906,
-      "grad_norm": 0.0027841362170875072,
-      "learning_rate": 0.00019999557944991517,
-      "loss": 46.0,
-      "step": 39156
-    },
-    {
-      "epoch": 2.9938260985912803,
-      "grad_norm": 0.0012173576978966594,
-      "learning_rate": 0.00019999557922406456,
-      "loss": 46.0,
-      "step": 39157
-    },
-    {
-      "epoch": 2.9939025555746697,
-      "grad_norm": 0.0029892115853726864,
-      "learning_rate": 0.00019999557899820817,
-      "loss": 46.0,
-      "step": 39158
-    },
-    {
-      "epoch": 2.9939790125580594,
-      "grad_norm": 0.004242290742695332,
-      "learning_rate": 0.00019999557877234603,
-      "loss": 46.0,
-      "step": 39159
-    },
-    {
-      "epoch": 2.994055469541449,
-      "grad_norm": 0.00103765819221735,
-      "learning_rate": 0.00019999557854647813,
-      "loss": 46.0,
-      "step": 39160
-    },
-    {
-      "epoch": 2.994131926524839,
-      "grad_norm": 0.002326207933947444,
-      "learning_rate": 0.00019999557832060442,
-      "loss": 46.0,
-      "step": 39161
-    },
-    {
-      "epoch": 2.9942083835082287,
-      "grad_norm": 0.002947617322206497,
-      "learning_rate": 0.00019999557809472496,
-      "loss": 46.0,
-      "step": 39162
-    },
-    {
-      "epoch": 2.9942848404916185,
-      "grad_norm": 0.0012251833686605096,
-      "learning_rate": 0.00019999557786883973,
-      "loss": 46.0,
-      "step": 39163
-    },
-    {
-      "epoch": 2.9943612974750082,
-      "grad_norm": 0.005709645338356495,
-      "learning_rate": 0.00019999557764294873,
-      "loss": 46.0,
-      "step": 39164
-    },
-    {
-      "epoch": 2.994437754458398,
-      "grad_norm": 0.0016854925779625773,
-      "learning_rate": 0.00019999557741705198,
-      "loss": 46.0,
-      "step": 39165
-    },
-    {
-      "epoch": 2.9945142114417873,
-      "grad_norm": 0.002527038101106882,
-      "learning_rate": 0.00019999557719114946,
-      "loss": 46.0,
-      "step": 39166
-    },
-    {
-      "epoch": 2.994590668425177,
-      "grad_norm": 0.005868361797183752,
-      "learning_rate": 0.00019999557696524117,
-      "loss": 46.0,
-      "step": 39167
-    },
-    {
-      "epoch": 2.994667125408567,
-      "grad_norm": 0.0006125843501649797,
-      "learning_rate": 0.00019999557673932707,
-      "loss": 46.0,
-      "step": 39168
-    },
-    {
-      "epoch": 2.9947435823919566,
-      "grad_norm": 0.004382311832159758,
-      "learning_rate": 0.00019999557651340726,
-      "loss": 46.0,
-      "step": 39169
-    },
-    {
-      "epoch": 2.9948200393753464,
-      "grad_norm": 0.003015749854966998,
-      "learning_rate": 0.00019999557628748161,
-      "loss": 46.0,
-      "step": 39170
-    },
-    {
-      "epoch": 2.994896496358736,
-      "grad_norm": 0.0013347103958949447,
-      "learning_rate": 0.00019999557606155025,
-      "loss": 46.0,
-      "step": 39171
-    },
-    {
-      "epoch": 2.994972953342126,
-      "grad_norm": 0.0017075969371944666,
-      "learning_rate": 0.00019999557583561312,
-      "loss": 46.0,
-      "step": 39172
-    },
-    {
-      "epoch": 2.9950494103255156,
-      "grad_norm": 0.0017577135004103184,
-      "learning_rate": 0.00019999557560967018,
-      "loss": 46.0,
-      "step": 39173
-    },
-    {
-      "epoch": 2.9951258673089054,
-      "grad_norm": 0.0013778347056359053,
-      "learning_rate": 0.0001999955753837215,
-      "loss": 46.0,
-      "step": 39174
-    },
-    {
-      "epoch": 2.995202324292295,
-      "grad_norm": 0.0018427680479362607,
-      "learning_rate": 0.00019999557515776705,
-      "loss": 46.0,
-      "step": 39175
-    },
-    {
-      "epoch": 2.995278781275685,
-      "grad_norm": 0.0024936965201050043,
-      "learning_rate": 0.00019999557493180682,
-      "loss": 46.0,
-      "step": 39176
-    },
-    {
-      "epoch": 2.9953552382590747,
-      "grad_norm": 0.002008868847042322,
-      "learning_rate": 0.00019999557470584084,
-      "loss": 46.0,
-      "step": 39177
-    },
-    {
-      "epoch": 2.9954316952424644,
-      "grad_norm": 0.004113717004656792,
-      "learning_rate": 0.00019999557447986907,
-      "loss": 46.0,
-      "step": 39178
-    },
-    {
-      "epoch": 2.995508152225854,
-      "grad_norm": 0.0024763448163866997,
-      "learning_rate": 0.00019999557425389152,
-      "loss": 46.0,
-      "step": 39179
-    },
-    {
-      "epoch": 2.9955846092092435,
-      "grad_norm": 0.0014906786382198334,
-      "learning_rate": 0.00019999557402790825,
-      "loss": 46.0,
-      "step": 39180
-    },
-    {
-      "epoch": 2.9956610661926333,
-      "grad_norm": 0.0017268853262066841,
-      "learning_rate": 0.00019999557380191916,
-      "loss": 46.0,
-      "step": 39181
-    },
-    {
-      "epoch": 2.995737523176023,
-      "grad_norm": 0.00249011954292655,
-      "learning_rate": 0.00019999557357592434,
-      "loss": 46.0,
-      "step": 39182
-    },
-    {
-      "epoch": 2.995813980159413,
-      "grad_norm": 0.0019509891280904412,
-      "learning_rate": 0.00019999557334992373,
-      "loss": 46.0,
-      "step": 39183
-    },
-    {
-      "epoch": 2.9958904371428026,
-      "grad_norm": 0.005961879622191191,
-      "learning_rate": 0.00019999557312391734,
-      "loss": 46.0,
-      "step": 39184
-    },
-    {
-      "epoch": 2.9959668941261923,
-      "grad_norm": 0.0012329589808359742,
-      "learning_rate": 0.00019999557289790518,
-      "loss": 46.0,
-      "step": 39185
-    },
-    {
-      "epoch": 2.996043351109582,
-      "grad_norm": 0.00348090217448771,
-      "learning_rate": 0.00019999557267188727,
-      "loss": 46.0,
-      "step": 39186
-    },
-    {
-      "epoch": 2.996119808092972,
-      "grad_norm": 0.003842496545985341,
-      "learning_rate": 0.0001999955724458636,
-      "loss": 46.0,
-      "step": 39187
-    },
-    {
-      "epoch": 2.996196265076361,
-      "grad_norm": 0.002805515890941024,
-      "learning_rate": 0.00019999557221983414,
-      "loss": 46.0,
-      "step": 39188
-    },
-    {
-      "epoch": 2.996272722059751,
-      "grad_norm": 0.0012306882999837399,
-      "learning_rate": 0.00019999557199379894,
-      "loss": 46.0,
-      "step": 39189
-    },
-    {
-      "epoch": 2.9963491790431407,
-      "grad_norm": 0.0017449719598516822,
-      "learning_rate": 0.0001999955717677579,
-      "loss": 46.0,
-      "step": 39190
-    },
-    {
-      "epoch": 2.9964256360265304,
-      "grad_norm": 0.0019003982888534665,
-      "learning_rate": 0.00019999557154171116,
-      "loss": 46.0,
-      "step": 39191
-    },
-    {
-      "epoch": 2.99650209300992,
-      "grad_norm": 0.006370673421770334,
-      "learning_rate": 0.0001999955713156586,
-      "loss": 46.0,
-      "step": 39192
-    },
-    {
-      "epoch": 2.99657854999331,
-      "grad_norm": 0.0018238719785586,
-      "learning_rate": 0.00019999557108960035,
-      "loss": 46.0,
-      "step": 39193
-    },
-    {
-      "epoch": 2.9966550069766997,
-      "grad_norm": 0.0016050889389589429,
-      "learning_rate": 0.00019999557086353625,
-      "loss": 46.0,
-      "step": 39194
-    },
-    {
-      "epoch": 2.9967314639600895,
-      "grad_norm": 0.0018161804182454944,
-      "learning_rate": 0.0001999955706374664,
-      "loss": 46.0,
-      "step": 39195
-    },
-    {
-      "epoch": 2.9968079209434793,
-      "grad_norm": 0.00195804750546813,
-      "learning_rate": 0.00019999557041139082,
-      "loss": 46.0,
-      "step": 39196
-    },
-    {
-      "epoch": 2.996884377926869,
-      "grad_norm": 0.0018318075453862548,
-      "learning_rate": 0.00019999557018530943,
-      "loss": 46.0,
-      "step": 39197
-    },
-    {
-      "epoch": 2.9969608349102588,
-      "grad_norm": 0.01040571741759777,
-      "learning_rate": 0.00019999556995922227,
-      "loss": 46.0,
-      "step": 39198
-    },
-    {
-      "epoch": 2.9970372918936485,
-      "grad_norm": 0.008059567771852016,
-      "learning_rate": 0.00019999556973312937,
-      "loss": 46.0,
-      "step": 39199
-    },
-    {
-      "epoch": 2.9971137488770383,
-      "grad_norm": 0.0030449870973825455,
-      "learning_rate": 0.00019999556950703069,
-      "loss": 46.0,
-      "step": 39200
-    },
-    {
-      "epoch": 2.9971902058604276,
-      "grad_norm": 0.005059572868049145,
-      "learning_rate": 0.00019999556928092623,
-      "loss": 46.0,
-      "step": 39201
-    },
-    {
-      "epoch": 2.9972666628438174,
-      "grad_norm": 0.0017652895767241716,
-      "learning_rate": 0.000199995569054816,
-      "loss": 46.0,
-      "step": 39202
-    },
-    {
-      "epoch": 2.997343119827207,
-      "grad_norm": 0.0011652631219476461,
-      "learning_rate": 0.00019999556882870003,
-      "loss": 46.0,
-      "step": 39203
-    },
-    {
-      "epoch": 2.997419576810597,
-      "grad_norm": 0.002552255056798458,
-      "learning_rate": 0.00019999556860257826,
-      "loss": 46.0,
-      "step": 39204
-    },
-    {
-      "epoch": 2.9974960337939867,
-      "grad_norm": 0.0014065821887925267,
-      "learning_rate": 0.00019999556837645073,
-      "loss": 46.0,
-      "step": 39205
-    },
-    {
-      "epoch": 2.9975724907773764,
-      "grad_norm": 0.0036756526678800583,
-      "learning_rate": 0.00019999556815031744,
-      "loss": 46.0,
-      "step": 39206
-    },
-    {
-      "epoch": 2.997648947760766,
-      "grad_norm": 0.0010107505368068814,
-      "learning_rate": 0.00019999556792417837,
-      "loss": 46.0,
-      "step": 39207
-    },
-    {
-      "epoch": 2.997725404744156,
-      "grad_norm": 0.0017440892988815904,
-      "learning_rate": 0.0001999955676980335,
-      "loss": 46.0,
-      "step": 39208
-    },
-    {
-      "epoch": 2.9978018617275457,
-      "grad_norm": 0.0023775596637278795,
-      "learning_rate": 0.00019999556747188292,
-      "loss": 46.0,
-      "step": 39209
-    },
-    {
-      "epoch": 2.997878318710935,
-      "grad_norm": 0.0013232782948762178,
-      "learning_rate": 0.00019999556724572656,
-      "loss": 46.0,
-      "step": 39210
-    },
-    {
-      "epoch": 2.997954775694325,
-      "grad_norm": 0.0012101158499717712,
-      "learning_rate": 0.0001999955670195644,
-      "loss": 46.0,
-      "step": 39211
-    },
-    {
-      "epoch": 2.9980312326777145,
-      "grad_norm": 0.0005670886021107435,
-      "learning_rate": 0.00019999556679339646,
-      "loss": 46.0,
-      "step": 39212
-    },
-    {
-      "epoch": 2.9981076896611043,
-      "grad_norm": 0.0019193993648514152,
-      "learning_rate": 0.0001999955665672228,
-      "loss": 46.0,
-      "step": 39213
-    },
-    {
-      "epoch": 2.998184146644494,
-      "grad_norm": 0.0008746773819439113,
-      "learning_rate": 0.00019999556634104333,
-      "loss": 46.0,
-      "step": 39214
-    },
-    {
-      "epoch": 2.998260603627884,
-      "grad_norm": 0.0023979737889021635,
-      "learning_rate": 0.00019999556611485813,
-      "loss": 46.0,
-      "step": 39215
-    },
-    {
-      "epoch": 2.9983370606112736,
-      "grad_norm": 0.005400867201387882,
-      "learning_rate": 0.00019999556588866713,
-      "loss": 46.0,
-      "step": 39216
-    },
-    {
-      "epoch": 2.9984135175946633,
-      "grad_norm": 0.0028523243963718414,
-      "learning_rate": 0.00019999556566247035,
-      "loss": 46.0,
-      "step": 39217
-    },
-    {
-      "epoch": 2.998489974578053,
-      "grad_norm": 0.002949568908661604,
-      "learning_rate": 0.00019999556543626783,
-      "loss": 46.0,
-      "step": 39218
-    },
-    {
-      "epoch": 2.998566431561443,
-      "grad_norm": 0.0035884494427591562,
-      "learning_rate": 0.00019999556521005954,
-      "loss": 46.0,
-      "step": 39219
-    },
-    {
-      "epoch": 2.9986428885448326,
-      "grad_norm": 0.001347847981378436,
-      "learning_rate": 0.00019999556498384547,
-      "loss": 46.0,
-      "step": 39220
-    },
-    {
-      "epoch": 2.9987193455282224,
-      "grad_norm": 0.0012585886288434267,
-      "learning_rate": 0.00019999556475762563,
-      "loss": 46.0,
-      "step": 39221
-    },
-    {
-      "epoch": 2.998795802511612,
-      "grad_norm": 0.008716125972568989,
-      "learning_rate": 0.0001999955645314,
-      "loss": 46.0,
-      "step": 39222
-    },
-    {
-      "epoch": 2.9988722594950015,
-      "grad_norm": 0.0036288145929574966,
-      "learning_rate": 0.00019999556430516863,
-      "loss": 46.0,
-      "step": 39223
-    },
-    {
-      "epoch": 2.9989487164783912,
-      "grad_norm": 0.004326615482568741,
-      "learning_rate": 0.0001999955640789315,
-      "loss": 46.0,
-      "step": 39224
-    },
-    {
-      "epoch": 2.999025173461781,
-      "grad_norm": 0.004040957428514957,
-      "learning_rate": 0.0001999955638526886,
-      "loss": 46.0,
-      "step": 39225
-    },
-    {
-      "epoch": 2.9991016304451708,
-      "grad_norm": 0.0034429074730724096,
-      "learning_rate": 0.00019999556362643988,
-      "loss": 46.0,
-      "step": 39226
-    },
-    {
-      "epoch": 2.9991780874285605,
-      "grad_norm": 0.004061649087816477,
-      "learning_rate": 0.00019999556340018543,
-      "loss": 46.0,
-      "step": 39227
-    },
-    {
-      "epoch": 2.9992545444119503,
-      "grad_norm": 0.003879598341882229,
-      "learning_rate": 0.0001999955631739252,
-      "loss": 46.0,
-      "step": 39228
-    },
-    {
-      "epoch": 2.99933100139534,
-      "grad_norm": 0.0023432110901921988,
-      "learning_rate": 0.00019999556294765923,
-      "loss": 46.0,
-      "step": 39229
-    },
-    {
-      "epoch": 2.99940745837873,
-      "grad_norm": 0.0020076779183000326,
-      "learning_rate": 0.00019999556272138746,
-      "loss": 46.0,
-      "step": 39230
-    },
-    {
-      "epoch": 2.999483915362119,
-      "grad_norm": 0.0005539238918572664,
-      "learning_rate": 0.00019999556249510994,
-      "loss": 46.0,
-      "step": 39231
-    },
-    {
-      "epoch": 2.999560372345509,
-      "grad_norm": 0.0017795147141441703,
-      "learning_rate": 0.00019999556226882664,
-      "loss": 46.0,
-      "step": 39232
-    },
-    {
-      "epoch": 2.9996368293288986,
-      "grad_norm": 0.0019348624628037214,
-      "learning_rate": 0.00019999556204253755,
-      "loss": 46.0,
-      "step": 39233
-    },
-    {
-      "epoch": 2.9997132863122884,
-      "grad_norm": 0.003848652122542262,
-      "learning_rate": 0.00019999556181624274,
-      "loss": 46.0,
-      "step": 39234
-    },
-    {
-      "epoch": 2.999789743295678,
-      "grad_norm": 0.002938139485195279,
-      "learning_rate": 0.00019999556158994212,
-      "loss": 46.0,
-      "step": 39235
-    },
-    {
-      "epoch": 2.999866200279068,
-      "grad_norm": 0.0014270143583416939,
-      "learning_rate": 0.00019999556136363576,
-      "loss": 46.0,
-      "step": 39236
-    },
-    {
-      "epoch": 2.9999426572624577,
-      "grad_norm": 0.0008606792544014752,
-      "learning_rate": 0.0001999955611373236,
-      "loss": 46.0,
-      "step": 39237
-    },
-    {
-      "epoch": 2.9999426572624577,
-      "eval_loss": 11.5,
-      "eval_runtime": 32.5265,
-      "eval_samples_per_second": 169.339,
-      "eval_steps_per_second": 84.669,
-      "step": 39237
-    },
-    {
-      "epoch": 3.0000191142458474,
-      "grad_norm": 0.002427714178338647,
-      "learning_rate": 0.0001999955609110057,
-      "loss": 46.0,
-      "step": 39238
-    },
-    {
-      "epoch": 3.000095571229237,
-      "grad_norm": 0.004683328792452812,
-      "learning_rate": 0.000199995560684682,
-      "loss": 46.0,
-      "step": 39239
-    },
-    {
-      "epoch": 3.000172028212627,
-      "grad_norm": 0.0031059151515364647,
-      "learning_rate": 0.00019999556045835254,
-      "loss": 46.0,
-      "step": 39240
-    },
-    {
-      "epoch": 3.0002484851960167,
-      "grad_norm": 0.006568351294845343,
-      "learning_rate": 0.00019999556023201734,
-      "loss": 46.0,
-      "step": 39241
-    },
-    {
-      "epoch": 3.0003249421794065,
-      "grad_norm": 0.0021187514066696167,
-      "learning_rate": 0.00019999556000567634,
-      "loss": 46.0,
-      "step": 39242
-    },
-    {
-      "epoch": 3.0004013991627962,
-      "grad_norm": 0.001852111890912056,
-      "learning_rate": 0.00019999555977932956,
-      "loss": 46.0,
-      "step": 39243
-    },
-    {
-      "epoch": 3.0004778561461856,
-      "grad_norm": 0.001291001564823091,
-      "learning_rate": 0.00019999555955297704,
-      "loss": 46.0,
-      "step": 39244
-    },
-    {
-      "epoch": 3.0005543131295753,
-      "grad_norm": 0.0029865512624382973,
-      "learning_rate": 0.00019999555932661875,
-      "loss": 46.0,
-      "step": 39245
-    },
-    {
-      "epoch": 3.000630770112965,
-      "grad_norm": 0.0012287814170122147,
-      "learning_rate": 0.00019999555910025469,
-      "loss": 46.0,
-      "step": 39246
-    },
-    {
-      "epoch": 3.000707227096355,
-      "grad_norm": 0.0024188263341784477,
-      "learning_rate": 0.00019999555887388485,
-      "loss": 46.0,
-      "step": 39247
-    },
-    {
-      "epoch": 3.0007836840797446,
-      "grad_norm": 0.0008548871264792979,
-      "learning_rate": 0.00019999555864750926,
-      "loss": 46.0,
-      "step": 39248
-    },
-    {
-      "epoch": 3.0008601410631344,
-      "grad_norm": 0.006372688338160515,
-      "learning_rate": 0.00019999555842112785,
-      "loss": 46.0,
-      "step": 39249
-    },
-    {
-      "epoch": 3.000936598046524,
-      "grad_norm": 0.002814937150105834,
-      "learning_rate": 0.0001999955581947407,
-      "loss": 46.0,
-      "step": 39250
-    },
-    {
-      "epoch": 3.001013055029914,
-      "grad_norm": 0.0017765265656635165,
-      "learning_rate": 0.0001999955579683478,
-      "loss": 46.0,
-      "step": 39251
-    },
-    {
-      "epoch": 3.0010895120133037,
-      "grad_norm": 0.0013573955511674285,
-      "learning_rate": 0.00019999555774194913,
-      "loss": 46.0,
-      "step": 39252
-    },
-    {
-      "epoch": 3.0011659689966934,
-      "grad_norm": 0.003322416450828314,
-      "learning_rate": 0.00019999555751554467,
-      "loss": 46.0,
-      "step": 39253
-    },
-    {
-      "epoch": 3.001242425980083,
-      "grad_norm": 0.005314044654369354,
-      "learning_rate": 0.00019999555728913445,
-      "loss": 46.0,
-      "step": 39254
-    },
-    {
-      "epoch": 3.0013188829634725,
-      "grad_norm": 0.0013232966884970665,
-      "learning_rate": 0.00019999555706271848,
-      "loss": 46.0,
-      "step": 39255
-    },
-    {
-      "epoch": 3.0013953399468623,
-      "grad_norm": 0.0016556764021515846,
-      "learning_rate": 0.0001999955568362967,
-      "loss": 46.0,
-      "step": 39256
-    },
-    {
-      "epoch": 3.001471796930252,
-      "grad_norm": 0.003503865795210004,
-      "learning_rate": 0.00019999555660986918,
-      "loss": 46.0,
-      "step": 39257
-    },
-    {
-      "epoch": 3.0015482539136418,
-      "grad_norm": 0.0004094807372894138,
-      "learning_rate": 0.0001999955563834359,
-      "loss": 46.0,
-      "step": 39258
-    },
-    {
-      "epoch": 3.0016247108970315,
-      "grad_norm": 0.0018939865985885262,
-      "learning_rate": 0.0001999955561569968,
-      "loss": 46.0,
-      "step": 39259
-    },
-    {
-      "epoch": 3.0017011678804213,
-      "grad_norm": 0.002850540913641453,
-      "learning_rate": 0.00019999555593055196,
-      "loss": 46.0,
-      "step": 39260
-    },
-    {
-      "epoch": 3.001777624863811,
-      "grad_norm": 0.0006466483464464545,
-      "learning_rate": 0.00019999555570410135,
-      "loss": 46.0,
-      "step": 39261
-    },
-    {
-      "epoch": 3.001854081847201,
-      "grad_norm": 0.005661812145262957,
-      "learning_rate": 0.00019999555547764496,
-      "loss": 46.0,
-      "step": 39262
-    },
-    {
-      "epoch": 3.0019305388305906,
-      "grad_norm": 0.0012674578465521336,
-      "learning_rate": 0.00019999555525118283,
-      "loss": 46.0,
-      "step": 39263
-    },
-    {
-      "epoch": 3.0020069958139803,
-      "grad_norm": 0.002266967436298728,
-      "learning_rate": 0.00019999555502471492,
-      "loss": 46.0,
-      "step": 39264
-    },
-    {
-      "epoch": 3.00208345279737,
-      "grad_norm": 0.004274963401257992,
-      "learning_rate": 0.00019999555479824124,
-      "loss": 46.0,
-      "step": 39265
-    },
-    {
-      "epoch": 3.0021599097807594,
-      "grad_norm": 0.0046497248113155365,
-      "learning_rate": 0.0001999955545717618,
-      "loss": 46.0,
-      "step": 39266
-    },
-    {
-      "epoch": 3.002236366764149,
-      "grad_norm": 0.0037155766040086746,
-      "learning_rate": 0.00019999555434527656,
-      "loss": 46.0,
-      "step": 39267
-    },
-    {
-      "epoch": 3.002312823747539,
-      "grad_norm": 0.0013625187566503882,
-      "learning_rate": 0.0001999955541187856,
-      "loss": 46.0,
-      "step": 39268
-    },
-    {
-      "epoch": 3.0023892807309287,
-      "grad_norm": 0.0011886089341714978,
-      "learning_rate": 0.00019999555389228882,
-      "loss": 46.0,
-      "step": 39269
-    },
-    {
-      "epoch": 3.0024657377143185,
-      "grad_norm": 0.0012649070704355836,
-      "learning_rate": 0.0001999955536657863,
-      "loss": 46.0,
-      "step": 39270
-    },
-    {
-      "epoch": 3.0025421946977082,
-      "grad_norm": 0.0024706104304641485,
-      "learning_rate": 0.00019999555343927798,
-      "loss": 46.0,
-      "step": 39271
-    },
-    {
-      "epoch": 3.002618651681098,
-      "grad_norm": 0.0026065302081406116,
-      "learning_rate": 0.00019999555321276394,
-      "loss": 46.0,
-      "step": 39272
-    },
-    {
-      "epoch": 3.0026951086644877,
-      "grad_norm": 0.0014793617883697152,
-      "learning_rate": 0.0001999955529862441,
-      "loss": 46.0,
-      "step": 39273
-    },
-    {
-      "epoch": 3.0027715656478775,
-      "grad_norm": 0.0025190270971506834,
-      "learning_rate": 0.0001999955527597185,
-      "loss": 46.0,
-      "step": 39274
-    },
-    {
-      "epoch": 3.0028480226312673,
-      "grad_norm": 0.0011415166081860662,
-      "learning_rate": 0.0001999955525331871,
-      "loss": 46.0,
-      "step": 39275
-    },
-    {
-      "epoch": 3.0029244796146566,
-      "grad_norm": 0.0009597733151167631,
-      "learning_rate": 0.00019999555230664997,
-      "loss": 46.0,
-      "step": 39276
-    },
-    {
-      "epoch": 3.0030009365980463,
-      "grad_norm": 0.005575475748628378,
-      "learning_rate": 0.00019999555208010704,
-      "loss": 46.0,
-      "step": 39277
-    },
-    {
-      "epoch": 3.003077393581436,
-      "grad_norm": 0.002231983933597803,
-      "learning_rate": 0.0001999955518535584,
-      "loss": 46.0,
-      "step": 39278
-    },
-    {
-      "epoch": 3.003153850564826,
-      "grad_norm": 0.0014428955037146807,
-      "learning_rate": 0.0001999955516270039,
-      "loss": 46.0,
-      "step": 39279
-    },
-    {
-      "epoch": 3.0032303075482156,
-      "grad_norm": 0.001606437610462308,
-      "learning_rate": 0.0001999955514004437,
-      "loss": 46.0,
-      "step": 39280
-    },
-    {
-      "epoch": 3.0033067645316054,
-      "grad_norm": 0.0030549154616892338,
-      "learning_rate": 0.0001999955511738777,
-      "loss": 46.0,
-      "step": 39281
-    },
-    {
-      "epoch": 3.003383221514995,
-      "grad_norm": 0.003709004260599613,
-      "learning_rate": 0.00019999555094730596,
-      "loss": 46.0,
-      "step": 39282
-    },
-    {
-      "epoch": 3.003459678498385,
-      "grad_norm": 0.0029395415913313627,
-      "learning_rate": 0.00019999555072072842,
-      "loss": 46.0,
-      "step": 39283
-    },
-    {
-      "epoch": 3.0035361354817747,
-      "grad_norm": 0.001813619746826589,
-      "learning_rate": 0.00019999555049414513,
-      "loss": 46.0,
-      "step": 39284
-    },
-    {
-      "epoch": 3.0036125924651644,
-      "grad_norm": 0.0033683686051517725,
-      "learning_rate": 0.00019999555026755606,
-      "loss": 46.0,
-      "step": 39285
-    },
-    {
-      "epoch": 3.003689049448554,
-      "grad_norm": 0.00283208885230124,
-      "learning_rate": 0.00019999555004096122,
-      "loss": 46.0,
-      "step": 39286
-    },
-    {
-      "epoch": 3.0037655064319435,
-      "grad_norm": 0.0051369075663387775,
-      "learning_rate": 0.0001999955498143606,
-      "loss": 46.0,
-      "step": 39287
-    },
-    {
-      "epoch": 3.0038419634153333,
-      "grad_norm": 0.003842682344838977,
-      "learning_rate": 0.00019999554958775425,
-      "loss": 46.0,
-      "step": 39288
-    },
-    {
-      "epoch": 3.003918420398723,
-      "grad_norm": 0.0027507557533681393,
-      "learning_rate": 0.00019999554936114212,
-      "loss": 46.0,
-      "step": 39289
-    },
-    {
-      "epoch": 3.003994877382113,
-      "grad_norm": 0.002347825327888131,
-      "learning_rate": 0.0001999955491345242,
-      "loss": 46.0,
-      "step": 39290
-    },
-    {
-      "epoch": 3.0040713343655026,
-      "grad_norm": 0.003627563826739788,
-      "learning_rate": 0.00019999554890790048,
-      "loss": 46.0,
-      "step": 39291
-    },
-    {
-      "epoch": 3.0041477913488923,
-      "grad_norm": 0.003237220225855708,
-      "learning_rate": 0.00019999554868127106,
-      "loss": 46.0,
-      "step": 39292
-    },
-    {
-      "epoch": 3.004224248332282,
-      "grad_norm": 0.0012536810245364904,
-      "learning_rate": 0.0001999955484546358,
-      "loss": 46.0,
-      "step": 39293
-    },
-    {
-      "epoch": 3.004300705315672,
-      "grad_norm": 0.0024540743324905634,
-      "learning_rate": 0.00019999554822799483,
-      "loss": 46.0,
-      "step": 39294
-    },
-    {
-      "epoch": 3.0043771622990616,
-      "grad_norm": 0.0015452642692252994,
-      "learning_rate": 0.0001999955480013481,
-      "loss": 46.0,
-      "step": 39295
-    },
-    {
-      "epoch": 3.0044536192824514,
-      "grad_norm": 0.008288186974823475,
-      "learning_rate": 0.00019999554777469557,
-      "loss": 46.0,
-      "step": 39296
-    },
-    {
-      "epoch": 3.004530076265841,
-      "grad_norm": 0.0019129274878650904,
-      "learning_rate": 0.00019999554754803725,
-      "loss": 46.0,
-      "step": 39297
-    },
-    {
-      "epoch": 3.0046065332492304,
-      "grad_norm": 0.0022248008754104376,
-      "learning_rate": 0.00019999554732137319,
-      "loss": 46.0,
-      "step": 39298
-    },
-    {
-      "epoch": 3.00468299023262,
-      "grad_norm": 0.0038055742625147104,
-      "learning_rate": 0.00019999554709470337,
-      "loss": 46.0,
-      "step": 39299
-    },
-    {
-      "epoch": 3.00475944721601,
-      "grad_norm": 0.0024777159560471773,
-      "learning_rate": 0.00019999554686802776,
-      "loss": 46.0,
-      "step": 39300
-    },
-    {
-      "epoch": 3.0048359041993997,
-      "grad_norm": 0.0019054062431678176,
-      "learning_rate": 0.00019999554664134638,
-      "loss": 46.0,
-      "step": 39301
-    },
-    {
-      "epoch": 3.0049123611827895,
-      "grad_norm": 0.0003340501571074128,
-      "learning_rate": 0.00019999554641465922,
-      "loss": 46.0,
-      "step": 39302
-    },
-    {
-      "epoch": 3.0049888181661792,
-      "grad_norm": 0.002761523239314556,
-      "learning_rate": 0.0001999955461879663,
-      "loss": 46.0,
-      "step": 39303
-    },
-    {
-      "epoch": 3.005065275149569,
-      "grad_norm": 0.008612114936113358,
-      "learning_rate": 0.00019999554596126764,
-      "loss": 46.0,
-      "step": 39304
-    },
-    {
-      "epoch": 3.0051417321329588,
-      "grad_norm": 0.004222886171191931,
-      "learning_rate": 0.0001999955457345632,
-      "loss": 46.0,
-      "step": 39305
-    },
-    {
-      "epoch": 3.0052181891163485,
-      "grad_norm": 0.0017778804758563638,
-      "learning_rate": 0.00019999554550785299,
-      "loss": 46.0,
-      "step": 39306
-    },
-    {
-      "epoch": 3.0052946460997383,
-      "grad_norm": 0.002196765970438719,
-      "learning_rate": 0.000199995545281137,
-      "loss": 46.0,
-      "step": 39307
-    },
-    {
-      "epoch": 3.005371103083128,
-      "grad_norm": 0.003573029302060604,
-      "learning_rate": 0.00019999554505441522,
-      "loss": 46.0,
-      "step": 39308
-    },
-    {
-      "epoch": 3.0054475600665174,
-      "grad_norm": 0.001215667580254376,
-      "learning_rate": 0.0001999955448276877,
-      "loss": 46.0,
-      "step": 39309
-    },
-    {
-      "epoch": 3.005524017049907,
-      "grad_norm": 0.007345445454120636,
-      "learning_rate": 0.0001999955446009544,
-      "loss": 46.0,
-      "step": 39310
-    },
-    {
-      "epoch": 3.005600474033297,
-      "grad_norm": 0.0017702047480270267,
-      "learning_rate": 0.00019999554437421531,
-      "loss": 46.0,
-      "step": 39311
-    },
-    {
-      "epoch": 3.0056769310166866,
-      "grad_norm": 0.0019814420957118273,
-      "learning_rate": 0.0001999955441474705,
-      "loss": 46.0,
-      "step": 39312
-    },
-    {
-      "epoch": 3.0057533880000764,
-      "grad_norm": 0.003374501597136259,
-      "learning_rate": 0.00019999554392071992,
-      "loss": 46.0,
-      "step": 39313
-    },
-    {
-      "epoch": 3.005829844983466,
-      "grad_norm": 0.0017362427897751331,
-      "learning_rate": 0.0001999955436939635,
-      "loss": 46.0,
-      "step": 39314
-    },
-    {
-      "epoch": 3.005906301966856,
-      "grad_norm": 0.0016244228463619947,
-      "learning_rate": 0.00019999554346720138,
-      "loss": 46.0,
-      "step": 39315
-    },
-    {
-      "epoch": 3.0059827589502457,
-      "grad_norm": 0.0031385133042931557,
-      "learning_rate": 0.00019999554324043347,
-      "loss": 46.0,
-      "step": 39316
-    },
-    {
-      "epoch": 3.0060592159336355,
-      "grad_norm": 0.0008756855968385935,
-      "learning_rate": 0.0001999955430136598,
-      "loss": 46.0,
-      "step": 39317
-    },
-    {
-      "epoch": 3.006135672917025,
-      "grad_norm": 0.0024367005098611116,
-      "learning_rate": 0.00019999554278688034,
-      "loss": 46.0,
-      "step": 39318
-    },
-    {
-      "epoch": 3.006212129900415,
-      "grad_norm": 0.006825009826570749,
-      "learning_rate": 0.00019999554256009512,
-      "loss": 46.0,
-      "step": 39319
-    },
-    {
-      "epoch": 3.0062885868838043,
-      "grad_norm": 0.002445233054459095,
-      "learning_rate": 0.00019999554233330412,
-      "loss": 46.0,
-      "step": 39320
-    },
-    {
-      "epoch": 3.006365043867194,
-      "grad_norm": 0.0013995724730193615,
-      "learning_rate": 0.00019999554210650737,
-      "loss": 46.0,
-      "step": 39321
-    },
-    {
-      "epoch": 3.006441500850584,
-      "grad_norm": 0.002760851988568902,
-      "learning_rate": 0.00019999554187970483,
-      "loss": 46.0,
-      "step": 39322
-    },
-    {
-      "epoch": 3.0065179578339736,
-      "grad_norm": 0.0027504260651767254,
-      "learning_rate": 0.00019999554165289657,
-      "loss": 46.0,
-      "step": 39323
-    },
-    {
-      "epoch": 3.0065944148173633,
-      "grad_norm": 0.0025702493730932474,
-      "learning_rate": 0.00019999554142608247,
-      "loss": 46.0,
-      "step": 39324
-    },
-    {
-      "epoch": 3.006670871800753,
-      "grad_norm": 0.003154223784804344,
-      "learning_rate": 0.00019999554119926264,
-      "loss": 46.0,
-      "step": 39325
-    },
-    {
-      "epoch": 3.006747328784143,
-      "grad_norm": 0.0019131911685690284,
-      "learning_rate": 0.00019999554097243705,
-      "loss": 46.0,
-      "step": 39326
-    },
-    {
-      "epoch": 3.0068237857675326,
-      "grad_norm": 0.0013920033816248178,
-      "learning_rate": 0.00019999554074560567,
-      "loss": 46.0,
-      "step": 39327
-    },
-    {
-      "epoch": 3.0069002427509224,
-      "grad_norm": 0.0016234898939728737,
-      "learning_rate": 0.0001999955405187685,
-      "loss": 46.0,
-      "step": 39328
-    },
-    {
-      "epoch": 3.006976699734312,
-      "grad_norm": 0.004340468440204859,
-      "learning_rate": 0.00019999554029192563,
-      "loss": 46.0,
-      "step": 39329
-    },
-    {
-      "epoch": 3.007053156717702,
-      "grad_norm": 0.002443769248202443,
-      "learning_rate": 0.00019999554006507695,
-      "loss": 46.0,
-      "step": 39330
-    },
-    {
-      "epoch": 3.007129613701091,
-      "grad_norm": 0.0014609097270295024,
-      "learning_rate": 0.0001999955398382225,
-      "loss": 46.0,
-      "step": 39331
-    },
-    {
-      "epoch": 3.007206070684481,
-      "grad_norm": 0.0013730047503486276,
-      "learning_rate": 0.00019999553961136228,
-      "loss": 46.0,
-      "step": 39332
-    },
-    {
-      "epoch": 3.0072825276678707,
-      "grad_norm": 0.003794951131567359,
-      "learning_rate": 0.00019999553938449628,
-      "loss": 46.0,
-      "step": 39333
-    },
-    {
-      "epoch": 3.0073589846512605,
-      "grad_norm": 0.001801217906177044,
-      "learning_rate": 0.0001999955391576245,
-      "loss": 46.0,
-      "step": 39334
-    },
-    {
-      "epoch": 3.0074354416346503,
-      "grad_norm": 0.004198344424366951,
-      "learning_rate": 0.000199995538930747,
-      "loss": 46.0,
-      "step": 39335
-    },
-    {
-      "epoch": 3.00751189861804,
-      "grad_norm": 0.0020839693024754524,
-      "learning_rate": 0.00019999553870386373,
-      "loss": 46.0,
-      "step": 39336
-    },
-    {
-      "epoch": 3.00758835560143,
-      "grad_norm": 0.0015438891714438796,
-      "learning_rate": 0.00019999553847697464,
-      "loss": 46.0,
-      "step": 39337
-    },
-    {
-      "epoch": 3.0076648125848195,
-      "grad_norm": 0.0035184400621801615,
-      "learning_rate": 0.00019999553825007983,
-      "loss": 46.0,
-      "step": 39338
-    },
-    {
-      "epoch": 3.0077412695682093,
-      "grad_norm": 0.0009981831535696983,
-      "learning_rate": 0.0001999955380231792,
-      "loss": 46.0,
-      "step": 39339
-    },
-    {
-      "epoch": 3.007817726551599,
-      "grad_norm": 0.004358906764537096,
-      "learning_rate": 0.00019999553779627283,
-      "loss": 46.0,
-      "step": 39340
-    },
-    {
-      "epoch": 3.007894183534989,
-      "grad_norm": 0.0028269216418266296,
-      "learning_rate": 0.0001999955375693607,
-      "loss": 46.0,
-      "step": 39341
-    },
-    {
-      "epoch": 3.007970640518378,
-      "grad_norm": 0.000924261927139014,
-      "learning_rate": 0.0001999955373424428,
-      "loss": 46.0,
-      "step": 39342
-    },
-    {
-      "epoch": 3.008047097501768,
-      "grad_norm": 0.0013527566334232688,
-      "learning_rate": 0.0001999955371155191,
-      "loss": 46.0,
-      "step": 39343
-    },
-    {
-      "epoch": 3.0081235544851577,
-      "grad_norm": 0.0020822433289140463,
-      "learning_rate": 0.00019999553688858964,
-      "loss": 46.0,
-      "step": 39344
-    },
-    {
-      "epoch": 3.0082000114685474,
-      "grad_norm": 0.0022662978153675795,
-      "learning_rate": 0.00019999553666165442,
-      "loss": 46.0,
-      "step": 39345
-    },
-    {
-      "epoch": 3.008276468451937,
-      "grad_norm": 0.001475971657782793,
-      "learning_rate": 0.00019999553643471342,
-      "loss": 46.0,
-      "step": 39346
-    },
-    {
-      "epoch": 3.008352925435327,
-      "grad_norm": 0.0014536315575242043,
-      "learning_rate": 0.00019999553620776668,
-      "loss": 46.0,
-      "step": 39347
-    },
-    {
-      "epoch": 3.0084293824187167,
-      "grad_norm": 0.005424159578979015,
-      "learning_rate": 0.00019999553598081416,
-      "loss": 46.0,
-      "step": 39348
-    },
-    {
-      "epoch": 3.0085058394021065,
-      "grad_norm": 0.0013394515262916684,
-      "learning_rate": 0.00019999553575385587,
-      "loss": 46.0,
-      "step": 39349
-    },
-    {
-      "epoch": 3.0085822963854962,
-      "grad_norm": 0.0008809837745502591,
-      "learning_rate": 0.00019999553552689178,
-      "loss": 46.0,
-      "step": 39350
-    },
-    {
-      "epoch": 3.008658753368886,
-      "grad_norm": 0.0027789610903710127,
-      "learning_rate": 0.00019999553529992197,
-      "loss": 46.0,
-      "step": 39351
-    },
-    {
-      "epoch": 3.0087352103522758,
-      "grad_norm": 0.001662369933910668,
-      "learning_rate": 0.00019999553507294638,
-      "loss": 46.0,
-      "step": 39352
-    },
-    {
-      "epoch": 3.008811667335665,
-      "grad_norm": 0.0045648436062037945,
-      "learning_rate": 0.00019999553484596497,
-      "loss": 46.0,
-      "step": 39353
-    },
-    {
-      "epoch": 3.008888124319055,
-      "grad_norm": 0.0018527117790654302,
-      "learning_rate": 0.00019999553461897784,
-      "loss": 46.0,
-      "step": 39354
-    },
-    {
-      "epoch": 3.0089645813024446,
-      "grad_norm": 0.0024484908208251,
-      "learning_rate": 0.00019999553439198494,
-      "loss": 46.0,
-      "step": 39355
-    },
-    {
-      "epoch": 3.0090410382858344,
-      "grad_norm": 0.0022886963561177254,
-      "learning_rate": 0.00019999553416498626,
-      "loss": 46.0,
-      "step": 39356
-    },
-    {
-      "epoch": 3.009117495269224,
-      "grad_norm": 0.0014354987069964409,
-      "learning_rate": 0.0001999955339379818,
-      "loss": 46.0,
-      "step": 39357
-    },
-    {
-      "epoch": 3.009193952252614,
-      "grad_norm": 0.0016976002370938659,
-      "learning_rate": 0.00019999553371097156,
-      "loss": 46.0,
-      "step": 39358
-    },
-    {
-      "epoch": 3.0092704092360036,
-      "grad_norm": 0.0036428917665034533,
-      "learning_rate": 0.00019999553348395562,
-      "loss": 46.0,
-      "step": 39359
-    },
-    {
-      "epoch": 3.0093468662193934,
-      "grad_norm": 0.0027197967283427715,
-      "learning_rate": 0.00019999553325693382,
-      "loss": 46.0,
-      "step": 39360
-    },
-    {
-      "epoch": 3.009423323202783,
-      "grad_norm": 0.003048036713153124,
-      "learning_rate": 0.00019999553302990633,
-      "loss": 46.0,
-      "step": 39361
-    },
-    {
-      "epoch": 3.009499780186173,
-      "grad_norm": 0.0014959762338548899,
-      "learning_rate": 0.00019999553280287301,
-      "loss": 46.0,
-      "step": 39362
-    },
-    {
-      "epoch": 3.0095762371695627,
-      "grad_norm": 0.0015660420758649707,
-      "learning_rate": 0.00019999553257583395,
-      "loss": 46.0,
-      "step": 39363
-    },
-    {
-      "epoch": 3.009652694152952,
-      "grad_norm": 0.0033422966953366995,
-      "learning_rate": 0.0001999955323487891,
-      "loss": 46.0,
-      "step": 39364
-    },
-    {
-      "epoch": 3.0097291511363418,
-      "grad_norm": 0.0010475198505446315,
-      "learning_rate": 0.0001999955321217385,
-      "loss": 46.0,
-      "step": 39365
-    },
-    {
-      "epoch": 3.0098056081197315,
-      "grad_norm": 0.0036027971655130386,
-      "learning_rate": 0.00019999553189468215,
-      "loss": 46.0,
-      "step": 39366
-    },
-    {
-      "epoch": 3.0098820651031213,
-      "grad_norm": 0.002637310419231653,
-      "learning_rate": 0.00019999553166762,
-      "loss": 46.0,
-      "step": 39367
-    },
-    {
-      "epoch": 3.009958522086511,
-      "grad_norm": 0.0022255482617765665,
-      "learning_rate": 0.00019999553144055209,
-      "loss": 46.0,
-      "step": 39368
-    },
-    {
-      "epoch": 3.010034979069901,
-      "grad_norm": 0.002966152736917138,
-      "learning_rate": 0.00019999553121347844,
-      "loss": 46.0,
-      "step": 39369
-    },
-    {
-      "epoch": 3.0101114360532906,
-      "grad_norm": 0.002870200900360942,
-      "learning_rate": 0.00019999553098639896,
-      "loss": 46.0,
-      "step": 39370
-    },
-    {
-      "epoch": 3.0101878930366803,
-      "grad_norm": 0.006929353345185518,
-      "learning_rate": 0.00019999553075931376,
-      "loss": 46.0,
-      "step": 39371
-    },
-    {
-      "epoch": 3.01026435002007,
-      "grad_norm": 0.001328064245171845,
-      "learning_rate": 0.00019999553053222274,
-      "loss": 46.0,
-      "step": 39372
-    },
-    {
-      "epoch": 3.01034080700346,
-      "grad_norm": 0.0010474452283233404,
-      "learning_rate": 0.00019999553030512602,
-      "loss": 46.0,
-      "step": 39373
-    },
-    {
-      "epoch": 3.0104172639868496,
-      "grad_norm": 0.0030777212232351303,
-      "learning_rate": 0.0001999955300780235,
-      "loss": 46.0,
-      "step": 39374
-    },
-    {
-      "epoch": 3.010493720970239,
-      "grad_norm": 0.0041889892891049385,
-      "learning_rate": 0.00019999552985091522,
-      "loss": 46.0,
-      "step": 39375
-    },
-    {
-      "epoch": 3.0105701779536287,
-      "grad_norm": 0.003737753489986062,
-      "learning_rate": 0.00019999552962380113,
-      "loss": 46.0,
-      "step": 39376
-    },
-    {
-      "epoch": 3.0106466349370185,
-      "grad_norm": 0.005608769133687019,
-      "learning_rate": 0.0001999955293966813,
-      "loss": 46.0,
-      "step": 39377
-    },
-    {
-      "epoch": 3.010723091920408,
-      "grad_norm": 0.000600518542341888,
-      "learning_rate": 0.0001999955291695557,
-      "loss": 46.0,
-      "step": 39378
-    },
-    {
-      "epoch": 3.010799548903798,
-      "grad_norm": 0.0013439670437946916,
-      "learning_rate": 0.00019999552894242432,
-      "loss": 46.0,
-      "step": 39379
-    },
-    {
-      "epoch": 3.0108760058871877,
-      "grad_norm": 0.0013751742662861943,
-      "learning_rate": 0.00019999552871528717,
-      "loss": 46.0,
-      "step": 39380
-    },
-    {
-      "epoch": 3.0109524628705775,
-      "grad_norm": 0.0004651658528018743,
-      "learning_rate": 0.00019999552848814426,
-      "loss": 46.0,
-      "step": 39381
-    },
-    {
-      "epoch": 3.0110289198539673,
-      "grad_norm": 0.0012404487933963537,
-      "learning_rate": 0.0001999955282609956,
-      "loss": 46.0,
-      "step": 39382
-    },
-    {
-      "epoch": 3.011105376837357,
-      "grad_norm": 0.001523573650047183,
-      "learning_rate": 0.00019999552803384116,
-      "loss": 46.0,
-      "step": 39383
-    },
-    {
-      "epoch": 3.011181833820747,
-      "grad_norm": 0.0012302714167162776,
-      "learning_rate": 0.00019999552780668091,
-      "loss": 46.0,
-      "step": 39384
-    },
-    {
-      "epoch": 3.0112582908041365,
-      "grad_norm": 0.000768988044001162,
-      "learning_rate": 0.00019999552757951495,
-      "loss": 46.0,
-      "step": 39385
-    },
-    {
-      "epoch": 3.011334747787526,
-      "grad_norm": 0.0008446825086139143,
-      "learning_rate": 0.0001999955273523432,
-      "loss": 46.0,
-      "step": 39386
-    },
-    {
-      "epoch": 3.0114112047709156,
-      "grad_norm": 0.0037377565167844296,
-      "learning_rate": 0.00019999552712516566,
-      "loss": 46.0,
-      "step": 39387
-    },
-    {
-      "epoch": 3.0114876617543054,
-      "grad_norm": 0.004430252593010664,
-      "learning_rate": 0.00019999552689798237,
-      "loss": 46.0,
-      "step": 39388
-    },
-    {
-      "epoch": 3.011564118737695,
-      "grad_norm": 0.0023575969971716404,
-      "learning_rate": 0.0001999955266707933,
-      "loss": 46.0,
-      "step": 39389
-    },
-    {
-      "epoch": 3.011640575721085,
-      "grad_norm": 0.004492738284170628,
-      "learning_rate": 0.00019999552644359847,
-      "loss": 46.0,
-      "step": 39390
-    },
-    {
-      "epoch": 3.0117170327044747,
-      "grad_norm": 0.0021767497528344393,
-      "learning_rate": 0.0001999955262163979,
-      "loss": 46.0,
-      "step": 39391
-    },
-    {
-      "epoch": 3.0117934896878644,
-      "grad_norm": 0.0012283246032893658,
-      "learning_rate": 0.0001999955259891915,
-      "loss": 46.0,
-      "step": 39392
-    },
-    {
-      "epoch": 3.011869946671254,
-      "grad_norm": 0.0064230673015117645,
-      "learning_rate": 0.00019999552576197935,
-      "loss": 46.0,
-      "step": 39393
-    },
-    {
-      "epoch": 3.011946403654644,
-      "grad_norm": 0.0019803596660494804,
-      "learning_rate": 0.00019999552553476147,
-      "loss": 46.0,
-      "step": 39394
-    },
-    {
-      "epoch": 3.0120228606380337,
-      "grad_norm": 0.0008590491488575935,
-      "learning_rate": 0.00019999552530753777,
-      "loss": 46.0,
-      "step": 39395
-    },
-    {
-      "epoch": 3.0120993176214235,
-      "grad_norm": 0.0019970927387475967,
-      "learning_rate": 0.00019999552508030832,
-      "loss": 46.0,
-      "step": 39396
-    },
-    {
-      "epoch": 3.012175774604813,
-      "grad_norm": 0.0022963066585361958,
-      "learning_rate": 0.00019999552485307312,
-      "loss": 46.0,
-      "step": 39397
-    },
-    {
-      "epoch": 3.0122522315882025,
-      "grad_norm": 0.0018520416924729943,
-      "learning_rate": 0.00019999552462583213,
-      "loss": 46.0,
-      "step": 39398
-    },
-    {
-      "epoch": 3.0123286885715923,
-      "grad_norm": 0.0011761331697925925,
-      "learning_rate": 0.00019999552439858539,
-      "loss": 46.0,
-      "step": 39399
-    },
-    {
-      "epoch": 3.012405145554982,
-      "grad_norm": 0.0009542282205075026,
-      "learning_rate": 0.00019999552417133284,
-      "loss": 46.0,
-      "step": 39400
-    },
-    {
-      "epoch": 3.012481602538372,
-      "grad_norm": 0.005651876796036959,
-      "learning_rate": 0.00019999552394407455,
-      "loss": 46.0,
-      "step": 39401
-    },
-    {
-      "epoch": 3.0125580595217616,
-      "grad_norm": 0.0035883476957678795,
-      "learning_rate": 0.00019999552371681046,
-      "loss": 46.0,
-      "step": 39402
-    },
-    {
-      "epoch": 3.0126345165051513,
-      "grad_norm": 0.0006832416984252632,
-      "learning_rate": 0.00019999552348954068,
-      "loss": 46.0,
-      "step": 39403
-    },
-    {
-      "epoch": 3.012710973488541,
-      "grad_norm": 0.0028043249621987343,
-      "learning_rate": 0.00019999552326226505,
-      "loss": 46.0,
-      "step": 39404
-    },
-    {
-      "epoch": 3.012787430471931,
-      "grad_norm": 0.0006444815662689507,
-      "learning_rate": 0.0001999955230349837,
-      "loss": 46.0,
-      "step": 39405
-    },
-    {
-      "epoch": 3.0128638874553206,
-      "grad_norm": 0.004395497962832451,
-      "learning_rate": 0.00019999552280769653,
-      "loss": 46.0,
-      "step": 39406
-    },
-    {
-      "epoch": 3.01294034443871,
-      "grad_norm": 0.0009213870507664979,
-      "learning_rate": 0.00019999552258040363,
-      "loss": 46.0,
-      "step": 39407
-    },
-    {
-      "epoch": 3.0130168014220997,
-      "grad_norm": 0.0036612919066101313,
-      "learning_rate": 0.00019999552235310498,
-      "loss": 46.0,
-      "step": 39408
-    },
-    {
-      "epoch": 3.0130932584054895,
-      "grad_norm": 0.002255040453746915,
-      "learning_rate": 0.00019999552212580053,
-      "loss": 46.0,
-      "step": 39409
-    },
-    {
-      "epoch": 3.0131697153888792,
-      "grad_norm": 0.0026665974874049425,
-      "learning_rate": 0.00019999552189849028,
-      "loss": 46.0,
-      "step": 39410
-    },
-    {
-      "epoch": 3.013246172372269,
-      "grad_norm": 0.0030404115095734596,
-      "learning_rate": 0.00019999552167117432,
-      "loss": 46.0,
-      "step": 39411
-    },
-    {
-      "epoch": 3.0133226293556588,
-      "grad_norm": 0.003132016398012638,
-      "learning_rate": 0.00019999552144385257,
-      "loss": 46.0,
-      "step": 39412
-    },
-    {
-      "epoch": 3.0133990863390485,
-      "grad_norm": 0.004428382497280836,
-      "learning_rate": 0.00019999552121652503,
-      "loss": 46.0,
-      "step": 39413
-    },
-    {
-      "epoch": 3.0134755433224383,
-      "grad_norm": 0.00221726531162858,
-      "learning_rate": 0.00019999552098919174,
-      "loss": 46.0,
-      "step": 39414
-    },
-    {
-      "epoch": 3.013552000305828,
-      "grad_norm": 0.0034431880339980125,
-      "learning_rate": 0.00019999552076185268,
-      "loss": 46.0,
-      "step": 39415
-    },
-    {
-      "epoch": 3.013628457289218,
-      "grad_norm": 0.000692942354362458,
-      "learning_rate": 0.00019999552053450782,
-      "loss": 46.0,
-      "step": 39416
-    },
-    {
-      "epoch": 3.0137049142726076,
-      "grad_norm": 0.00745638320222497,
-      "learning_rate": 0.00019999552030715723,
-      "loss": 46.0,
-      "step": 39417
-    },
-    {
-      "epoch": 3.013781371255997,
-      "grad_norm": 0.00488311517983675,
-      "learning_rate": 0.00019999552007980088,
-      "loss": 46.0,
-      "step": 39418
-    },
-    {
-      "epoch": 3.0138578282393866,
-      "grad_norm": 0.004646709188818932,
-      "learning_rate": 0.00019999551985243875,
-      "loss": 46.0,
-      "step": 39419
-    },
-    {
-      "epoch": 3.0139342852227764,
-      "grad_norm": 0.002193636493757367,
-      "learning_rate": 0.00019999551962507082,
-      "loss": 46.0,
-      "step": 39420
-    },
-    {
-      "epoch": 3.014010742206166,
-      "grad_norm": 0.006063556298613548,
-      "learning_rate": 0.00019999551939769715,
-      "loss": 46.0,
-      "step": 39421
-    },
-    {
-      "epoch": 3.014087199189556,
-      "grad_norm": 0.003570585511624813,
-      "learning_rate": 0.0001999955191703177,
-      "loss": 46.0,
-      "step": 39422
-    },
-    {
-      "epoch": 3.0141636561729457,
-      "grad_norm": 0.0020032732281833887,
-      "learning_rate": 0.0001999955189429325,
-      "loss": 46.0,
-      "step": 39423
-    },
-    {
-      "epoch": 3.0142401131563354,
-      "grad_norm": 0.007903410121798515,
-      "learning_rate": 0.0001999955187155415,
-      "loss": 46.0,
-      "step": 39424
-    },
-    {
-      "epoch": 3.014316570139725,
-      "grad_norm": 0.0022396582644432783,
-      "learning_rate": 0.00019999551848814474,
-      "loss": 46.0,
-      "step": 39425
-    },
-    {
-      "epoch": 3.014393027123115,
-      "grad_norm": 0.001396796084009111,
-      "learning_rate": 0.00019999551826074222,
-      "loss": 46.0,
-      "step": 39426
-    },
-    {
-      "epoch": 3.0144694841065047,
-      "grad_norm": 0.005267503205686808,
-      "learning_rate": 0.00019999551803333393,
-      "loss": 46.0,
-      "step": 39427
-    },
-    {
-      "epoch": 3.0145459410898945,
-      "grad_norm": 0.0034614321775734425,
-      "learning_rate": 0.00019999551780591987,
-      "loss": 46.0,
-      "step": 39428
-    },
-    {
-      "epoch": 3.014622398073284,
-      "grad_norm": 0.0012525422498583794,
-      "learning_rate": 0.00019999551757850004,
-      "loss": 46.0,
-      "step": 39429
-    },
-    {
-      "epoch": 3.0146988550566736,
-      "grad_norm": 0.004195562098175287,
-      "learning_rate": 0.00019999551735107445,
-      "loss": 46.0,
-      "step": 39430
-    },
-    {
-      "epoch": 3.0147753120400633,
-      "grad_norm": 0.005847356282174587,
-      "learning_rate": 0.00019999551712364307,
-      "loss": 46.0,
-      "step": 39431
-    },
-    {
-      "epoch": 3.014851769023453,
-      "grad_norm": 0.0022426152136176825,
-      "learning_rate": 0.00019999551689620592,
-      "loss": 46.0,
-      "step": 39432
-    },
-    {
-      "epoch": 3.014928226006843,
-      "grad_norm": 0.0021602236665785313,
-      "learning_rate": 0.00019999551666876304,
-      "loss": 46.0,
-      "step": 39433
-    },
-    {
-      "epoch": 3.0150046829902326,
-      "grad_norm": 0.004899294581264257,
-      "learning_rate": 0.00019999551644131437,
-      "loss": 46.0,
-      "step": 39434
-    },
-    {
-      "epoch": 3.0150811399736224,
-      "grad_norm": 0.0026685649063438177,
-      "learning_rate": 0.0001999955162138599,
-      "loss": 46.0,
-      "step": 39435
-    },
-    {
-      "epoch": 3.015157596957012,
-      "grad_norm": 0.001997855259105563,
-      "learning_rate": 0.00019999551598639967,
-      "loss": 46.0,
-      "step": 39436
-    },
-    {
-      "epoch": 3.015234053940402,
-      "grad_norm": 0.005254756659269333,
-      "learning_rate": 0.00019999551575893373,
-      "loss": 46.0,
-      "step": 39437
-    },
-    {
-      "epoch": 3.0153105109237917,
-      "grad_norm": 0.0020204721949994564,
-      "learning_rate": 0.00019999551553146193,
-      "loss": 46.0,
-      "step": 39438
-    },
-    {
-      "epoch": 3.0153869679071814,
-      "grad_norm": 0.0009282830869778991,
-      "learning_rate": 0.00019999551530398445,
-      "loss": 46.0,
-      "step": 39439
-    },
-    {
-      "epoch": 3.0154634248905707,
-      "grad_norm": 0.004497554153203964,
-      "learning_rate": 0.00019999551507650113,
-      "loss": 46.0,
-      "step": 39440
-    },
-    {
-      "epoch": 3.0155398818739605,
-      "grad_norm": 0.001964461524039507,
-      "learning_rate": 0.00019999551484901207,
-      "loss": 46.0,
-      "step": 39441
-    },
-    {
-      "epoch": 3.0156163388573503,
-      "grad_norm": 0.0028947643004357815,
-      "learning_rate": 0.00019999551462151723,
-      "loss": 46.0,
-      "step": 39442
-    },
-    {
-      "epoch": 3.01569279584074,
-      "grad_norm": 0.003510626032948494,
-      "learning_rate": 0.00019999551439401665,
-      "loss": 46.0,
-      "step": 39443
-    },
-    {
-      "epoch": 3.0157692528241298,
-      "grad_norm": 0.0033016942907124758,
-      "learning_rate": 0.00019999551416651027,
-      "loss": 46.0,
-      "step": 39444
-    },
-    {
-      "epoch": 3.0158457098075195,
-      "grad_norm": 0.0012277968926355243,
-      "learning_rate": 0.00019999551393899814,
-      "loss": 46.0,
-      "step": 39445
-    },
-    {
-      "epoch": 3.0159221667909093,
-      "grad_norm": 0.0007238698308356106,
-      "learning_rate": 0.0001999955137114802,
-      "loss": 46.0,
-      "step": 39446
-    },
-    {
-      "epoch": 3.015998623774299,
-      "grad_norm": 0.001045160461217165,
-      "learning_rate": 0.00019999551348395657,
-      "loss": 46.0,
-      "step": 39447
-    },
-    {
-      "epoch": 3.016075080757689,
-      "grad_norm": 0.0019541247747838497,
-      "learning_rate": 0.00019999551325642712,
-      "loss": 46.0,
-      "step": 39448
-    },
-    {
-      "epoch": 3.0161515377410786,
-      "grad_norm": 0.0029180001001805067,
-      "learning_rate": 0.00019999551302889187,
-      "loss": 46.0,
-      "step": 39449
-    },
-    {
-      "epoch": 3.0162279947244683,
-      "grad_norm": 0.003668922232463956,
-      "learning_rate": 0.00019999551280135093,
-      "loss": 46.0,
-      "step": 39450
-    },
-    {
-      "epoch": 3.0163044517078577,
-      "grad_norm": 0.0013092189328745008,
-      "learning_rate": 0.00019999551257380416,
-      "loss": 46.0,
-      "step": 39451
-    },
-    {
-      "epoch": 3.0163809086912474,
-      "grad_norm": 0.0040798126719892025,
-      "learning_rate": 0.00019999551234625162,
-      "loss": 46.0,
-      "step": 39452
-    },
-    {
-      "epoch": 3.016457365674637,
-      "grad_norm": 0.0016058036126196384,
-      "learning_rate": 0.00019999551211869333,
-      "loss": 46.0,
-      "step": 39453
-    },
-    {
-      "epoch": 3.016533822658027,
-      "grad_norm": 0.0029545396100729704,
-      "learning_rate": 0.0001999955118911293,
-      "loss": 46.0,
-      "step": 39454
-    },
-    {
-      "epoch": 3.0166102796414167,
-      "grad_norm": 0.0014185799518600106,
-      "learning_rate": 0.00019999551166355946,
-      "loss": 46.0,
-      "step": 39455
-    },
-    {
-      "epoch": 3.0166867366248065,
-      "grad_norm": 0.001393378246575594,
-      "learning_rate": 0.00019999551143598385,
-      "loss": 46.0,
-      "step": 39456
-    },
-    {
-      "epoch": 3.0167631936081962,
-      "grad_norm": 0.002617392223328352,
-      "learning_rate": 0.00019999551120840247,
-      "loss": 46.0,
-      "step": 39457
-    },
-    {
-      "epoch": 3.016839650591586,
-      "grad_norm": 0.004419579170644283,
-      "learning_rate": 0.00019999551098081532,
-      "loss": 46.0,
-      "step": 39458
-    },
-    {
-      "epoch": 3.0169161075749757,
-      "grad_norm": 0.005027450621128082,
-      "learning_rate": 0.00019999551075322244,
-      "loss": 46.0,
-      "step": 39459
-    },
-    {
-      "epoch": 3.0169925645583655,
-      "grad_norm": 0.0012779480312019587,
-      "learning_rate": 0.00019999551052562374,
-      "loss": 46.0,
-      "step": 39460
-    },
-    {
-      "epoch": 3.0170690215417553,
-      "grad_norm": 0.0038401770871132612,
-      "learning_rate": 0.00019999551029801932,
-      "loss": 46.0,
-      "step": 39461
-    },
-    {
-      "epoch": 3.0171454785251446,
-      "grad_norm": 0.001983853057026863,
-      "learning_rate": 0.0001999955100704091,
-      "loss": 46.0,
-      "step": 39462
-    },
-    {
-      "epoch": 3.0172219355085343,
-      "grad_norm": 0.0035650290083140135,
-      "learning_rate": 0.0001999955098427931,
-      "loss": 46.0,
-      "step": 39463
-    },
-    {
-      "epoch": 3.017298392491924,
-      "grad_norm": 0.0003997962921857834,
-      "learning_rate": 0.00019999550961517136,
-      "loss": 46.0,
-      "step": 39464
-    },
-    {
-      "epoch": 3.017374849475314,
-      "grad_norm": 0.0020195217803120613,
-      "learning_rate": 0.00019999550938754382,
-      "loss": 46.0,
-      "step": 39465
-    },
-    {
-      "epoch": 3.0174513064587036,
-      "grad_norm": 0.0018501769518479705,
-      "learning_rate": 0.00019999550915991053,
-      "loss": 46.0,
-      "step": 39466
-    },
-    {
-      "epoch": 3.0175277634420934,
-      "grad_norm": 0.001178169739432633,
-      "learning_rate": 0.0001999955089322715,
-      "loss": 46.0,
-      "step": 39467
-    },
-    {
-      "epoch": 3.017604220425483,
-      "grad_norm": 0.0044974978081882,
-      "learning_rate": 0.00019999550870462667,
-      "loss": 46.0,
-      "step": 39468
-    },
-    {
-      "epoch": 3.017680677408873,
-      "grad_norm": 0.0026308188680559397,
-      "learning_rate": 0.00019999550847697606,
-      "loss": 46.0,
-      "step": 39469
-    },
-    {
-      "epoch": 3.0177571343922627,
-      "grad_norm": 0.0025967489928007126,
-      "learning_rate": 0.0001999955082493197,
-      "loss": 46.0,
-      "step": 39470
-    },
-    {
-      "epoch": 3.0178335913756524,
-      "grad_norm": 0.0033321240916848183,
-      "learning_rate": 0.00019999550802165755,
-      "loss": 46.0,
-      "step": 39471
-    },
-    {
-      "epoch": 3.017910048359042,
-      "grad_norm": 0.002086300402879715,
-      "learning_rate": 0.00019999550779398962,
-      "loss": 46.0,
-      "step": 39472
-    },
-    {
-      "epoch": 3.0179865053424315,
-      "grad_norm": 0.002381829544901848,
-      "learning_rate": 0.00019999550756631597,
-      "loss": 46.0,
-      "step": 39473
-    },
-    {
-      "epoch": 3.0180629623258213,
-      "grad_norm": 0.007461670786142349,
-      "learning_rate": 0.00019999550733863653,
-      "loss": 46.0,
-      "step": 39474
-    },
-    {
-      "epoch": 3.018139419309211,
-      "grad_norm": 0.0052920919843018055,
-      "learning_rate": 0.0001999955071109513,
-      "loss": 46.0,
-      "step": 39475
-    },
-    {
-      "epoch": 3.018215876292601,
-      "grad_norm": 0.004215862136334181,
-      "learning_rate": 0.0001999955068832603,
-      "loss": 46.0,
-      "step": 39476
-    },
-    {
-      "epoch": 3.0182923332759906,
-      "grad_norm": 0.003421841189265251,
-      "learning_rate": 0.00019999550665556357,
-      "loss": 46.0,
-      "step": 39477
-    },
-    {
-      "epoch": 3.0183687902593803,
-      "grad_norm": 0.0012295730412006378,
-      "learning_rate": 0.00019999550642786106,
-      "loss": 46.0,
-      "step": 39478
-    },
-    {
-      "epoch": 3.01844524724277,
-      "grad_norm": 0.0018388039898127317,
-      "learning_rate": 0.00019999550620015274,
-      "loss": 46.0,
-      "step": 39479
-    },
-    {
-      "epoch": 3.01852170422616,
-      "grad_norm": 0.003716369392350316,
-      "learning_rate": 0.0001999955059724387,
-      "loss": 46.0,
-      "step": 39480
-    },
-    {
-      "epoch": 3.0185981612095496,
-      "grad_norm": 0.0021993708796799183,
-      "learning_rate": 0.00019999550574471888,
-      "loss": 46.0,
-      "step": 39481
-    },
-    {
-      "epoch": 3.0186746181929394,
-      "grad_norm": 0.0016098986379802227,
-      "learning_rate": 0.00019999550551699327,
-      "loss": 46.0,
-      "step": 39482
-    },
-    {
-      "epoch": 3.018751075176329,
-      "grad_norm": 0.0026489822193980217,
-      "learning_rate": 0.00019999550528926191,
-      "loss": 46.0,
-      "step": 39483
-    },
-    {
-      "epoch": 3.0188275321597184,
-      "grad_norm": 0.0020469960290938616,
-      "learning_rate": 0.00019999550506152476,
-      "loss": 46.0,
-      "step": 39484
-    },
-    {
-      "epoch": 3.018903989143108,
-      "grad_norm": 0.00467684818431735,
-      "learning_rate": 0.00019999550483378186,
-      "loss": 46.0,
-      "step": 39485
-    },
-    {
-      "epoch": 3.018980446126498,
-      "grad_norm": 0.0051090954802930355,
-      "learning_rate": 0.0001999955046060332,
-      "loss": 46.0,
-      "step": 39486
-    },
-    {
-      "epoch": 3.0190569031098877,
-      "grad_norm": 0.005223751068115234,
-      "learning_rate": 0.00019999550437827877,
-      "loss": 46.0,
-      "step": 39487
-    },
-    {
-      "epoch": 3.0191333600932775,
-      "grad_norm": 0.0044022416695952415,
-      "learning_rate": 0.00019999550415051852,
-      "loss": 46.0,
-      "step": 39488
-    },
-    {
-      "epoch": 3.0192098170766672,
-      "grad_norm": 0.0015226774848997593,
-      "learning_rate": 0.00019999550392275255,
-      "loss": 46.0,
-      "step": 39489
-    },
-    {
-      "epoch": 3.019286274060057,
-      "grad_norm": 0.002391026122495532,
-      "learning_rate": 0.0001999955036949808,
-      "loss": 46.0,
-      "step": 39490
-    },
-    {
-      "epoch": 3.0193627310434468,
-      "grad_norm": 0.0032122882548719645,
-      "learning_rate": 0.00019999550346720327,
-      "loss": 46.0,
-      "step": 39491
-    },
-    {
-      "epoch": 3.0194391880268365,
-      "grad_norm": 0.002311436226591468,
-      "learning_rate": 0.00019999550323941998,
-      "loss": 46.0,
-      "step": 39492
-    },
-    {
-      "epoch": 3.0195156450102263,
-      "grad_norm": 0.0025622406974434853,
-      "learning_rate": 0.00019999550301163092,
-      "loss": 46.0,
-      "step": 39493
-    },
-    {
-      "epoch": 3.019592101993616,
-      "grad_norm": 0.0006701344391331077,
-      "learning_rate": 0.0001999955027838361,
-      "loss": 46.0,
-      "step": 39494
-    },
-    {
-      "epoch": 3.0196685589770054,
-      "grad_norm": 0.002483791671693325,
-      "learning_rate": 0.0001999955025560355,
-      "loss": 46.0,
-      "step": 39495
-    },
-    {
-      "epoch": 3.019745015960395,
-      "grad_norm": 0.0016375501872971654,
-      "learning_rate": 0.00019999550232822913,
-      "loss": 46.0,
-      "step": 39496
-    },
-    {
-      "epoch": 3.019821472943785,
-      "grad_norm": 0.004270276986062527,
-      "learning_rate": 0.000199995502100417,
-      "loss": 46.0,
-      "step": 39497
-    },
-    {
-      "epoch": 3.0198979299271747,
-      "grad_norm": 0.0016216675285249949,
-      "learning_rate": 0.0001999955018725991,
-      "loss": 46.0,
-      "step": 39498
-    },
-    {
-      "epoch": 3.0199743869105644,
-      "grad_norm": 0.002118132309988141,
-      "learning_rate": 0.00019999550164477543,
-      "loss": 46.0,
-      "step": 39499
-    },
-    {
-      "epoch": 3.020050843893954,
-      "grad_norm": 0.002751134568825364,
-      "learning_rate": 0.00019999550141694598,
-      "loss": 46.0,
-      "step": 39500
-    },
-    {
-      "epoch": 3.020127300877344,
-      "grad_norm": 0.0021930960938334465,
-      "learning_rate": 0.00019999550118911076,
-      "loss": 46.0,
-      "step": 39501
-    },
-    {
-      "epoch": 3.0202037578607337,
-      "grad_norm": 0.005091415718197823,
-      "learning_rate": 0.00019999550096126977,
-      "loss": 46.0,
-      "step": 39502
-    },
-    {
-      "epoch": 3.0202802148441235,
-      "grad_norm": 0.002106683561578393,
-      "learning_rate": 0.00019999550073342303,
-      "loss": 46.0,
-      "step": 39503
-    },
-    {
-      "epoch": 3.020356671827513,
-      "grad_norm": 0.001299706636928022,
-      "learning_rate": 0.0001999955005055705,
-      "loss": 46.0,
-      "step": 39504
-    },
-    {
-      "epoch": 3.020433128810903,
-      "grad_norm": 0.0009211815195158124,
-      "learning_rate": 0.0001999955002777122,
-      "loss": 46.0,
-      "step": 39505
-    },
-    {
-      "epoch": 3.0205095857942923,
-      "grad_norm": 0.002531653270125389,
-      "learning_rate": 0.00019999550004984817,
-      "loss": 46.0,
-      "step": 39506
-    },
-    {
-      "epoch": 3.020586042777682,
-      "grad_norm": 0.001453232835046947,
-      "learning_rate": 0.0001999954998219783,
-      "loss": 46.0,
-      "step": 39507
-    },
-    {
-      "epoch": 3.020662499761072,
-      "grad_norm": 0.006961993873119354,
-      "learning_rate": 0.00019999549959410273,
-      "loss": 46.0,
-      "step": 39508
-    },
-    {
-      "epoch": 3.0207389567444616,
-      "grad_norm": 0.0011820897925645113,
-      "learning_rate": 0.00019999549936622137,
-      "loss": 46.0,
-      "step": 39509
-    },
-    {
-      "epoch": 3.0208154137278513,
-      "grad_norm": 0.005325338337570429,
-      "learning_rate": 0.00019999549913833425,
-      "loss": 46.0,
-      "step": 39510
-    },
-    {
-      "epoch": 3.020891870711241,
-      "grad_norm": 0.004203235264867544,
-      "learning_rate": 0.00019999549891044135,
-      "loss": 46.0,
-      "step": 39511
-    },
-    {
-      "epoch": 3.020968327694631,
-      "grad_norm": 0.0012564307544380426,
-      "learning_rate": 0.00019999549868254265,
-      "loss": 46.0,
-      "step": 39512
-    },
-    {
-      "epoch": 3.0210447846780206,
-      "grad_norm": 0.002346331486478448,
-      "learning_rate": 0.0001999954984546382,
-      "loss": 46.0,
-      "step": 39513
-    },
-    {
-      "epoch": 3.0211212416614104,
-      "grad_norm": 0.0009770137257874012,
-      "learning_rate": 0.00019999549822672798,
-      "loss": 46.0,
-      "step": 39514
-    },
-    {
-      "epoch": 3.0211976986448,
-      "grad_norm": 0.0028245749417692423,
-      "learning_rate": 0.000199995497998812,
-      "loss": 46.0,
-      "step": 39515
-    },
-    {
-      "epoch": 3.02127415562819,
-      "grad_norm": 0.003160701598972082,
-      "learning_rate": 0.00019999549777089027,
-      "loss": 46.0,
-      "step": 39516
-    },
-    {
-      "epoch": 3.0213506126115792,
-      "grad_norm": 0.0017769790720194578,
-      "learning_rate": 0.00019999549754296273,
-      "loss": 46.0,
-      "step": 39517
-    },
-    {
-      "epoch": 3.021427069594969,
-      "grad_norm": 0.000977318617515266,
-      "learning_rate": 0.00019999549731502945,
-      "loss": 46.0,
-      "step": 39518
-    },
-    {
-      "epoch": 3.0215035265783587,
-      "grad_norm": 0.002621496096253395,
-      "learning_rate": 0.0001999954970870904,
-      "loss": 46.0,
-      "step": 39519
-    },
-    {
-      "epoch": 3.0215799835617485,
-      "grad_norm": 0.000990555388852954,
-      "learning_rate": 0.00019999549685914558,
-      "loss": 46.0,
-      "step": 39520
-    },
-    {
-      "epoch": 3.0216564405451383,
-      "grad_norm": 0.0032036621123552322,
-      "learning_rate": 0.00019999549663119498,
-      "loss": 46.0,
-      "step": 39521
-    },
-    {
-      "epoch": 3.021732897528528,
-      "grad_norm": 0.0014264700002968311,
-      "learning_rate": 0.0001999954964032386,
-      "loss": 46.0,
-      "step": 39522
-    },
-    {
-      "epoch": 3.021809354511918,
-      "grad_norm": 0.0023281339090317488,
-      "learning_rate": 0.00019999549617527647,
-      "loss": 46.0,
-      "step": 39523
-    },
-    {
-      "epoch": 3.0218858114953075,
-      "grad_norm": 0.001668137381784618,
-      "learning_rate": 0.00019999549594730857,
-      "loss": 46.0,
-      "step": 39524
-    },
-    {
-      "epoch": 3.0219622684786973,
-      "grad_norm": 0.0011162168812006712,
-      "learning_rate": 0.0001999954957193349,
-      "loss": 46.0,
-      "step": 39525
-    },
-    {
-      "epoch": 3.022038725462087,
-      "grad_norm": 0.0013556305784732103,
-      "learning_rate": 0.00019999549549135542,
-      "loss": 46.0,
-      "step": 39526
-    },
-    {
-      "epoch": 3.022115182445477,
-      "grad_norm": 0.004790290724486113,
-      "learning_rate": 0.00019999549526337026,
-      "loss": 46.0,
-      "step": 39527
-    },
-    {
-      "epoch": 3.022191639428866,
-      "grad_norm": 0.0011906020808964968,
-      "learning_rate": 0.00019999549503537924,
-      "loss": 46.0,
-      "step": 39528
-    },
-    {
-      "epoch": 3.022268096412256,
-      "grad_norm": 0.0033291589934378862,
-      "learning_rate": 0.0001999954948073825,
-      "loss": 46.0,
-      "step": 39529
-    },
-    {
-      "epoch": 3.0223445533956457,
-      "grad_norm": 0.0016306675970554352,
-      "learning_rate": 0.00019999549457937999,
-      "loss": 46.0,
-      "step": 39530
-    },
-    {
-      "epoch": 3.0224210103790354,
-      "grad_norm": 0.002272416604682803,
-      "learning_rate": 0.00019999549435137167,
-      "loss": 46.0,
-      "step": 39531
-    },
-    {
-      "epoch": 3.022497467362425,
-      "grad_norm": 0.0016264538280665874,
-      "learning_rate": 0.00019999549412335764,
-      "loss": 46.0,
-      "step": 39532
-    },
-    {
-      "epoch": 3.022573924345815,
-      "grad_norm": 0.0036463686265051365,
-      "learning_rate": 0.0001999954938953378,
-      "loss": 46.0,
-      "step": 39533
-    },
-    {
-      "epoch": 3.0226503813292047,
-      "grad_norm": 0.003005189588293433,
-      "learning_rate": 0.0001999954936673122,
-      "loss": 46.0,
-      "step": 39534
-    },
-    {
-      "epoch": 3.0227268383125945,
-      "grad_norm": 0.004093020688742399,
-      "learning_rate": 0.00019999549343928082,
-      "loss": 46.0,
-      "step": 39535
-    },
-    {
-      "epoch": 3.0228032952959842,
-      "grad_norm": 0.0026402596849948168,
-      "learning_rate": 0.00019999549321124372,
-      "loss": 46.0,
-      "step": 39536
-    },
-    {
-      "epoch": 3.022879752279374,
-      "grad_norm": 0.004336964339017868,
-      "learning_rate": 0.00019999549298320082,
-      "loss": 46.0,
-      "step": 39537
-    },
-    {
-      "epoch": 3.0229562092627633,
-      "grad_norm": 0.005225456319749355,
-      "learning_rate": 0.00019999549275515212,
-      "loss": 46.0,
-      "step": 39538
-    },
-    {
-      "epoch": 3.023032666246153,
-      "grad_norm": 0.0014755992451682687,
-      "learning_rate": 0.0001999954925270977,
-      "loss": 46.0,
-      "step": 39539
-    },
-    {
-      "epoch": 3.023109123229543,
-      "grad_norm": 0.0021839458495378494,
-      "learning_rate": 0.00019999549229903746,
-      "loss": 46.0,
-      "step": 39540
-    },
-    {
-      "epoch": 3.0231855802129326,
-      "grad_norm": 0.0014477333752438426,
-      "learning_rate": 0.0001999954920709715,
-      "loss": 46.0,
-      "step": 39541
-    },
-    {
-      "epoch": 3.0232620371963224,
-      "grad_norm": 0.004196503199636936,
-      "learning_rate": 0.00019999549184289975,
-      "loss": 46.0,
-      "step": 39542
-    },
-    {
-      "epoch": 3.023338494179712,
-      "grad_norm": 0.0010057806503027678,
-      "learning_rate": 0.00019999549161482224,
-      "loss": 46.0,
-      "step": 39543
-    },
-    {
-      "epoch": 3.023414951163102,
-      "grad_norm": 0.001979694701731205,
-      "learning_rate": 0.00019999549138673896,
-      "loss": 46.0,
-      "step": 39544
-    },
-    {
-      "epoch": 3.0234914081464916,
-      "grad_norm": 0.0018634955631569028,
-      "learning_rate": 0.0001999954911586499,
-      "loss": 46.0,
-      "step": 39545
-    },
-    {
-      "epoch": 3.0235678651298814,
-      "grad_norm": 0.0019349822541698813,
-      "learning_rate": 0.00019999549093055506,
-      "loss": 46.0,
-      "step": 39546
-    },
-    {
-      "epoch": 3.023644322113271,
-      "grad_norm": 0.000942290120292455,
-      "learning_rate": 0.00019999549070245446,
-      "loss": 46.0,
-      "step": 39547
-    },
-    {
-      "epoch": 3.023720779096661,
-      "grad_norm": 0.002466872101649642,
-      "learning_rate": 0.00019999549047434808,
-      "loss": 46.0,
-      "step": 39548
-    },
-    {
-      "epoch": 3.0237972360800502,
-      "grad_norm": 0.0007670424529351294,
-      "learning_rate": 0.00019999549024623595,
-      "loss": 46.0,
-      "step": 39549
-    },
-    {
-      "epoch": 3.02387369306344,
-      "grad_norm": 0.0030811980832368135,
-      "learning_rate": 0.00019999549001811805,
-      "loss": 46.0,
-      "step": 39550
-    },
-    {
-      "epoch": 3.0239501500468298,
-      "grad_norm": 0.005377268884330988,
-      "learning_rate": 0.00019999548978999438,
-      "loss": 46.0,
-      "step": 39551
-    },
-    {
-      "epoch": 3.0240266070302195,
-      "grad_norm": 0.0029614539816975594,
-      "learning_rate": 0.00019999548956186494,
-      "loss": 46.0,
-      "step": 39552
-    },
-    {
-      "epoch": 3.0241030640136093,
-      "grad_norm": 0.0015341442776843905,
-      "learning_rate": 0.00019999548933372974,
-      "loss": 46.0,
-      "step": 39553
-    },
-    {
-      "epoch": 3.024179520996999,
-      "grad_norm": 0.0021591614931821823,
-      "learning_rate": 0.00019999548910558875,
-      "loss": 46.0,
-      "step": 39554
-    },
-    {
-      "epoch": 3.024255977980389,
-      "grad_norm": 0.001307124039158225,
-      "learning_rate": 0.000199995488877442,
-      "loss": 46.0,
-      "step": 39555
-    },
-    {
-      "epoch": 3.0243324349637786,
-      "grad_norm": 0.0019438335439190269,
-      "learning_rate": 0.0001999954886492895,
-      "loss": 46.0,
-      "step": 39556
-    },
-    {
-      "epoch": 3.0244088919471683,
-      "grad_norm": 0.0012734817573800683,
-      "learning_rate": 0.00019999548842113122,
-      "loss": 46.0,
-      "step": 39557
-    },
-    {
-      "epoch": 3.024485348930558,
-      "grad_norm": 0.002688448177650571,
-      "learning_rate": 0.00019999548819296713,
-      "loss": 46.0,
-      "step": 39558
-    },
-    {
-      "epoch": 3.024561805913948,
-      "grad_norm": 0.009756259620189667,
-      "learning_rate": 0.0001999954879647973,
-      "loss": 46.0,
-      "step": 39559
-    },
-    {
-      "epoch": 3.024638262897337,
-      "grad_norm": 0.004535699728876352,
-      "learning_rate": 0.00019999548773662175,
-      "loss": 46.0,
-      "step": 39560
-    },
-    {
-      "epoch": 3.024714719880727,
-      "grad_norm": 0.0022196879144757986,
-      "learning_rate": 0.00019999548750844034,
-      "loss": 46.0,
-      "step": 39561
-    },
-    {
-      "epoch": 3.0247911768641167,
-      "grad_norm": 0.0037507337983697653,
-      "learning_rate": 0.00019999548728025324,
-      "loss": 46.0,
-      "step": 39562
-    },
-    {
-      "epoch": 3.0248676338475065,
-      "grad_norm": 0.02029794082045555,
-      "learning_rate": 0.00019999548705206032,
-      "loss": 46.0,
-      "step": 39563
-    },
-    {
-      "epoch": 3.024944090830896,
-      "grad_norm": 0.003327839309349656,
-      "learning_rate": 0.00019999548682386164,
-      "loss": 46.0,
-      "step": 39564
-    },
-    {
-      "epoch": 3.025020547814286,
-      "grad_norm": 0.0014364809030666947,
-      "learning_rate": 0.0001999954865956572,
-      "loss": 46.0,
-      "step": 39565
-    },
-    {
-      "epoch": 3.0250970047976757,
-      "grad_norm": 0.004175614099949598,
-      "learning_rate": 0.000199995486367447,
-      "loss": 46.0,
-      "step": 39566
-    },
-    {
-      "epoch": 3.0251734617810655,
-      "grad_norm": 0.0018812966300174594,
-      "learning_rate": 0.00019999548613923102,
-      "loss": 46.0,
-      "step": 39567
-    },
-    {
-      "epoch": 3.0252499187644553,
-      "grad_norm": 0.0026837356854230165,
-      "learning_rate": 0.00019999548591100928,
-      "loss": 46.0,
-      "step": 39568
-    },
-    {
-      "epoch": 3.025326375747845,
-      "grad_norm": 0.00445416709408164,
-      "learning_rate": 0.00019999548568278174,
-      "loss": 46.0,
-      "step": 39569
-    },
-    {
-      "epoch": 3.025402832731235,
-      "grad_norm": 0.005507659167051315,
-      "learning_rate": 0.00019999548545454845,
-      "loss": 46.0,
-      "step": 39570
-    },
-    {
-      "epoch": 3.025479289714624,
-      "grad_norm": 0.0023499075323343277,
-      "learning_rate": 0.0001999954852263094,
-      "loss": 46.0,
-      "step": 39571
-    },
-    {
-      "epoch": 3.025555746698014,
-      "grad_norm": 0.000948725501075387,
-      "learning_rate": 0.00019999548499806456,
-      "loss": 46.0,
-      "step": 39572
-    },
-    {
-      "epoch": 3.0256322036814036,
-      "grad_norm": 0.0011857252102345228,
-      "learning_rate": 0.00019999548476981398,
-      "loss": 46.0,
-      "step": 39573
-    },
-    {
-      "epoch": 3.0257086606647934,
-      "grad_norm": 0.0018004389712587,
-      "learning_rate": 0.0001999954845415576,
-      "loss": 46.0,
-      "step": 39574
-    },
-    {
-      "epoch": 3.025785117648183,
-      "grad_norm": 0.0024115771520882845,
-      "learning_rate": 0.00019999548431329548,
-      "loss": 46.0,
-      "step": 39575
-    },
-    {
-      "epoch": 3.025861574631573,
-      "grad_norm": 0.001869614003226161,
-      "learning_rate": 0.00019999548408502758,
-      "loss": 46.0,
-      "step": 39576
-    },
-    {
-      "epoch": 3.0259380316149627,
-      "grad_norm": 0.0022346852347254753,
-      "learning_rate": 0.0001999954838567539,
-      "loss": 46.0,
-      "step": 39577
-    },
-    {
-      "epoch": 3.0260144885983524,
-      "grad_norm": 0.0011376978363841772,
-      "learning_rate": 0.00019999548362847447,
-      "loss": 46.0,
-      "step": 39578
-    },
-    {
-      "epoch": 3.026090945581742,
-      "grad_norm": 0.0004650024638976902,
-      "learning_rate": 0.00019999548340018925,
-      "loss": 46.0,
-      "step": 39579
-    },
-    {
-      "epoch": 3.026167402565132,
-      "grad_norm": 0.002437163144350052,
-      "learning_rate": 0.00019999548317189828,
-      "loss": 46.0,
-      "step": 39580
-    },
-    {
-      "epoch": 3.0262438595485217,
-      "grad_norm": 0.0030671500135213137,
-      "learning_rate": 0.00019999548294360154,
-      "loss": 46.0,
-      "step": 39581
-    },
-    {
-      "epoch": 3.026320316531911,
-      "grad_norm": 0.0017258322332054377,
-      "learning_rate": 0.000199995482715299,
-      "loss": 46.0,
-      "step": 39582
-    },
-    {
-      "epoch": 3.026396773515301,
-      "grad_norm": 0.0029500273521989584,
-      "learning_rate": 0.00019999548248699075,
-      "loss": 46.0,
-      "step": 39583
-    },
-    {
-      "epoch": 3.0264732304986905,
-      "grad_norm": 0.0038172812201082706,
-      "learning_rate": 0.00019999548225867666,
-      "loss": 46.0,
-      "step": 39584
-    },
-    {
-      "epoch": 3.0265496874820803,
-      "grad_norm": 0.006628929637372494,
-      "learning_rate": 0.00019999548203035686,
-      "loss": 46.0,
-      "step": 39585
-    },
-    {
-      "epoch": 3.02662614446547,
-      "grad_norm": 0.0015419156989082694,
-      "learning_rate": 0.00019999548180203125,
-      "loss": 46.0,
-      "step": 39586
-    },
-    {
-      "epoch": 3.02670260144886,
-      "grad_norm": 0.0011260294122621417,
-      "learning_rate": 0.00019999548157369988,
-      "loss": 46.0,
-      "step": 39587
-    },
-    {
-      "epoch": 3.0267790584322496,
-      "grad_norm": 0.0009932902175933123,
-      "learning_rate": 0.00019999548134536275,
-      "loss": 46.0,
-      "step": 39588
-    },
-    {
-      "epoch": 3.0268555154156394,
-      "grad_norm": 0.003975550178438425,
-      "learning_rate": 0.00019999548111701985,
-      "loss": 46.0,
-      "step": 39589
-    },
-    {
-      "epoch": 3.026931972399029,
-      "grad_norm": 0.002477450994774699,
-      "learning_rate": 0.00019999548088867118,
-      "loss": 46.0,
-      "step": 39590
-    },
-    {
-      "epoch": 3.027008429382419,
-      "grad_norm": 0.0011117301182821393,
-      "learning_rate": 0.00019999548066031674,
-      "loss": 46.0,
-      "step": 39591
-    },
-    {
-      "epoch": 3.0270848863658086,
-      "grad_norm": 0.0017374836606904864,
-      "learning_rate": 0.00019999548043195652,
-      "loss": 46.0,
-      "step": 39592
-    },
-    {
-      "epoch": 3.027161343349198,
-      "grad_norm": 0.006389596965163946,
-      "learning_rate": 0.00019999548020359055,
-      "loss": 46.0,
-      "step": 39593
-    },
-    {
-      "epoch": 3.0272378003325877,
-      "grad_norm": 0.0019762907177209854,
-      "learning_rate": 0.0001999954799752188,
-      "loss": 46.0,
-      "step": 39594
-    },
-    {
-      "epoch": 3.0273142573159775,
-      "grad_norm": 0.0005828551365993917,
-      "learning_rate": 0.0001999954797468413,
-      "loss": 46.0,
-      "step": 39595
-    },
-    {
-      "epoch": 3.0273907142993672,
-      "grad_norm": 0.003115327563136816,
-      "learning_rate": 0.000199995479518458,
-      "loss": 46.0,
-      "step": 39596
-    },
-    {
-      "epoch": 3.027467171282757,
-      "grad_norm": 0.0023204556200653315,
-      "learning_rate": 0.00019999547929006894,
-      "loss": 46.0,
-      "step": 39597
-    },
-    {
-      "epoch": 3.0275436282661468,
-      "grad_norm": 0.004497657064348459,
-      "learning_rate": 0.00019999547906167413,
-      "loss": 46.0,
-      "step": 39598
-    },
-    {
-      "epoch": 3.0276200852495365,
-      "grad_norm": 0.0031204139813780785,
-      "learning_rate": 0.00019999547883327353,
-      "loss": 46.0,
-      "step": 39599
-    },
-    {
-      "epoch": 3.0276965422329263,
-      "grad_norm": 0.001052657957188785,
-      "learning_rate": 0.00019999547860486715,
-      "loss": 46.0,
-      "step": 39600
-    },
-    {
-      "epoch": 3.027772999216316,
-      "grad_norm": 0.0015188665129244328,
-      "learning_rate": 0.00019999547837645502,
-      "loss": 46.0,
-      "step": 39601
-    },
-    {
-      "epoch": 3.027849456199706,
-      "grad_norm": 0.001129683805629611,
-      "learning_rate": 0.00019999547814803715,
-      "loss": 46.0,
-      "step": 39602
-    },
-    {
-      "epoch": 3.0279259131830956,
-      "grad_norm": 0.0022491435520350933,
-      "learning_rate": 0.00019999547791961346,
-      "loss": 46.0,
-      "step": 39603
-    },
-    {
-      "epoch": 3.028002370166485,
-      "grad_norm": 0.0020019079092890024,
-      "learning_rate": 0.000199995477691184,
-      "loss": 46.0,
-      "step": 39604
-    },
-    {
-      "epoch": 3.0280788271498746,
-      "grad_norm": 0.001778197125531733,
-      "learning_rate": 0.0001999954774627488,
-      "loss": 46.0,
-      "step": 39605
-    },
-    {
-      "epoch": 3.0281552841332644,
-      "grad_norm": 0.0033746480476111174,
-      "learning_rate": 0.00019999547723430783,
-      "loss": 46.0,
-      "step": 39606
-    },
-    {
-      "epoch": 3.028231741116654,
-      "grad_norm": 0.0023577758111059666,
-      "learning_rate": 0.00019999547700586106,
-      "loss": 46.0,
-      "step": 39607
-    },
-    {
-      "epoch": 3.028308198100044,
-      "grad_norm": 0.0007763591129332781,
-      "learning_rate": 0.00019999547677740855,
-      "loss": 46.0,
-      "step": 39608
-    },
-    {
-      "epoch": 3.0283846550834337,
-      "grad_norm": 0.0023277439177036285,
-      "learning_rate": 0.00019999547654895027,
-      "loss": 46.0,
-      "step": 39609
-    },
-    {
-      "epoch": 3.0284611120668234,
-      "grad_norm": 0.0006218118942342699,
-      "learning_rate": 0.00019999547632048624,
-      "loss": 46.0,
-      "step": 39610
-    },
-    {
-      "epoch": 3.028537569050213,
-      "grad_norm": 0.0012510822853073478,
-      "learning_rate": 0.0001999954760920164,
-      "loss": 46.0,
-      "step": 39611
-    },
-    {
-      "epoch": 3.028614026033603,
-      "grad_norm": 0.0031409927178174257,
-      "learning_rate": 0.0001999954758635408,
-      "loss": 46.0,
-      "step": 39612
-    },
-    {
-      "epoch": 3.0286904830169927,
-      "grad_norm": 0.002179933711886406,
-      "learning_rate": 0.00019999547563505945,
-      "loss": 46.0,
-      "step": 39613
-    },
-    {
-      "epoch": 3.0287669400003825,
-      "grad_norm": 0.003693270729854703,
-      "learning_rate": 0.0001999954754065723,
-      "loss": 46.0,
-      "step": 39614
-    },
-    {
-      "epoch": 3.028843396983772,
-      "grad_norm": 0.0013576335040852427,
-      "learning_rate": 0.0001999954751780794,
-      "loss": 46.0,
-      "step": 39615
-    },
-    {
-      "epoch": 3.0289198539671616,
-      "grad_norm": 0.0025278846733272076,
-      "learning_rate": 0.00019999547494958076,
-      "loss": 46.0,
-      "step": 39616
-    },
-    {
-      "epoch": 3.0289963109505513,
-      "grad_norm": 0.003638332011178136,
-      "learning_rate": 0.0001999954747210763,
-      "loss": 46.0,
-      "step": 39617
-    },
-    {
-      "epoch": 3.029072767933941,
-      "grad_norm": 0.0022339161951094866,
-      "learning_rate": 0.0001999954744925661,
-      "loss": 46.0,
-      "step": 39618
-    },
-    {
-      "epoch": 3.029149224917331,
-      "grad_norm": 0.0008160865982063115,
-      "learning_rate": 0.0001999954742640501,
-      "loss": 46.0,
-      "step": 39619
-    },
-    {
-      "epoch": 3.0292256819007206,
-      "grad_norm": 0.001652189763262868,
-      "learning_rate": 0.00019999547403552837,
-      "loss": 46.0,
-      "step": 39620
-    },
-    {
-      "epoch": 3.0293021388841104,
-      "grad_norm": 0.002944201696664095,
-      "learning_rate": 0.00019999547380700086,
-      "loss": 46.0,
-      "step": 39621
-    },
-    {
-      "epoch": 3.0293785958675,
-      "grad_norm": 0.018335677683353424,
-      "learning_rate": 0.00019999547357846758,
-      "loss": 46.0,
-      "step": 39622
-    },
-    {
-      "epoch": 3.02945505285089,
-      "grad_norm": 0.002560319844633341,
-      "learning_rate": 0.00019999547334992852,
-      "loss": 46.0,
-      "step": 39623
-    },
-    {
-      "epoch": 3.0295315098342797,
-      "grad_norm": 0.004606395028531551,
-      "learning_rate": 0.0001999954731213837,
-      "loss": 46.0,
-      "step": 39624
-    },
-    {
-      "epoch": 3.0296079668176694,
-      "grad_norm": 0.0011316162999719381,
-      "learning_rate": 0.00019999547289283309,
-      "loss": 46.0,
-      "step": 39625
-    },
-    {
-      "epoch": 3.0296844238010587,
-      "grad_norm": 0.0024140742607414722,
-      "learning_rate": 0.0001999954726642767,
-      "loss": 46.0,
-      "step": 39626
-    },
-    {
-      "epoch": 3.0297608807844485,
-      "grad_norm": 0.001941421302035451,
-      "learning_rate": 0.0001999954724357146,
-      "loss": 46.0,
-      "step": 39627
-    },
-    {
-      "epoch": 3.0298373377678383,
-      "grad_norm": 0.0021032022777944803,
-      "learning_rate": 0.0001999954722071467,
-      "loss": 46.0,
-      "step": 39628
-    },
-    {
-      "epoch": 3.029913794751228,
-      "grad_norm": 0.0029945955611765385,
-      "learning_rate": 0.00019999547197857302,
-      "loss": 46.0,
-      "step": 39629
-    },
-    {
-      "epoch": 3.029990251734618,
-      "grad_norm": 0.0038915437180548906,
-      "learning_rate": 0.00019999547174999357,
-      "loss": 46.0,
-      "step": 39630
-    },
-    {
-      "epoch": 3.0300667087180075,
-      "grad_norm": 0.0014142083236947656,
-      "learning_rate": 0.00019999547152140838,
-      "loss": 46.0,
-      "step": 39631
-    },
-    {
-      "epoch": 3.0301431657013973,
-      "grad_norm": 0.0006382741266861558,
-      "learning_rate": 0.00019999547129281742,
-      "loss": 46.0,
-      "step": 39632
-    },
-    {
-      "epoch": 3.030219622684787,
-      "grad_norm": 0.0015237802872434258,
-      "learning_rate": 0.00019999547106422066,
-      "loss": 46.0,
-      "step": 39633
-    },
-    {
-      "epoch": 3.030296079668177,
-      "grad_norm": 0.002217212226241827,
-      "learning_rate": 0.00019999547083561815,
-      "loss": 46.0,
-      "step": 39634
-    },
-    {
-      "epoch": 3.0303725366515666,
-      "grad_norm": 0.0025363692548125982,
-      "learning_rate": 0.00019999547060700986,
-      "loss": 46.0,
-      "step": 39635
-    },
-    {
-      "epoch": 3.0304489936349563,
-      "grad_norm": 0.002185879973694682,
-      "learning_rate": 0.0001999954703783958,
-      "loss": 46.0,
-      "step": 39636
-    },
-    {
-      "epoch": 3.0305254506183457,
-      "grad_norm": 0.0030111183878034353,
-      "learning_rate": 0.00019999547014977598,
-      "loss": 46.0,
-      "step": 39637
-    },
-    {
-      "epoch": 3.0306019076017354,
-      "grad_norm": 0.0025228685699403286,
-      "learning_rate": 0.0001999954699211504,
-      "loss": 46.0,
-      "step": 39638
-    },
-    {
-      "epoch": 3.030678364585125,
-      "grad_norm": 0.0023971344344317913,
-      "learning_rate": 0.00019999546969251902,
-      "loss": 46.0,
-      "step": 39639
-    },
-    {
-      "epoch": 3.030754821568515,
-      "grad_norm": 0.0024575740098953247,
-      "learning_rate": 0.00019999546946388187,
-      "loss": 46.0,
-      "step": 39640
-    },
-    {
-      "epoch": 3.0308312785519047,
-      "grad_norm": 0.007665298413485289,
-      "learning_rate": 0.000199995469235239,
-      "loss": 46.0,
-      "step": 39641
-    },
-    {
-      "epoch": 3.0309077355352945,
-      "grad_norm": 0.0010867660166695714,
-      "learning_rate": 0.00019999546900659033,
-      "loss": 46.0,
-      "step": 39642
-    },
-    {
-      "epoch": 3.0309841925186842,
-      "grad_norm": 0.002488918136805296,
-      "learning_rate": 0.0001999954687779359,
-      "loss": 46.0,
-      "step": 39643
-    },
-    {
-      "epoch": 3.031060649502074,
-      "grad_norm": 0.00568863470107317,
-      "learning_rate": 0.00019999546854927567,
-      "loss": 46.0,
-      "step": 39644
-    },
-    {
-      "epoch": 3.0311371064854638,
-      "grad_norm": 0.010594950057566166,
-      "learning_rate": 0.00019999546832060968,
-      "loss": 46.0,
-      "step": 39645
-    },
-    {
-      "epoch": 3.0312135634688535,
-      "grad_norm": 0.0022774694953113794,
-      "learning_rate": 0.00019999546809193794,
-      "loss": 46.0,
-      "step": 39646
-    },
-    {
-      "epoch": 3.0312900204522433,
-      "grad_norm": 0.0015874780947342515,
-      "learning_rate": 0.00019999546786326043,
-      "loss": 46.0,
-      "step": 39647
-    },
-    {
-      "epoch": 3.0313664774356326,
-      "grad_norm": 0.0008436526986770332,
-      "learning_rate": 0.00019999546763457715,
-      "loss": 46.0,
-      "step": 39648
-    },
-    {
-      "epoch": 3.0314429344190224,
-      "grad_norm": 0.003040659474208951,
-      "learning_rate": 0.0001999954674058881,
-      "loss": 46.0,
-      "step": 39649
-    },
-    {
-      "epoch": 3.031519391402412,
-      "grad_norm": 0.0017943443963304162,
-      "learning_rate": 0.00019999546717719326,
-      "loss": 46.0,
-      "step": 39650
-    },
-    {
-      "epoch": 3.031595848385802,
-      "grad_norm": 0.0010763994650915265,
-      "learning_rate": 0.0001999954669484927,
-      "loss": 46.0,
-      "step": 39651
-    },
-    {
-      "epoch": 3.0316723053691916,
-      "grad_norm": 0.0018663015216588974,
-      "learning_rate": 0.00019999546671978634,
-      "loss": 46.0,
-      "step": 39652
-    },
-    {
-      "epoch": 3.0317487623525814,
-      "grad_norm": 0.003037545131519437,
-      "learning_rate": 0.0001999954664910742,
-      "loss": 46.0,
-      "step": 39653
-    },
-    {
-      "epoch": 3.031825219335971,
-      "grad_norm": 0.0008113468065857887,
-      "learning_rate": 0.0001999954662623563,
-      "loss": 46.0,
-      "step": 39654
-    },
-    {
-      "epoch": 3.031901676319361,
-      "grad_norm": 0.0016137253260239959,
-      "learning_rate": 0.00019999546603363265,
-      "loss": 46.0,
-      "step": 39655
-    },
-    {
-      "epoch": 3.0319781333027507,
-      "grad_norm": 0.0013214999344199896,
-      "learning_rate": 0.0001999954658049032,
-      "loss": 46.0,
-      "step": 39656
-    },
-    {
-      "epoch": 3.0320545902861404,
-      "grad_norm": 0.00482101459056139,
-      "learning_rate": 0.000199995465576168,
-      "loss": 46.0,
-      "step": 39657
-    },
-    {
-      "epoch": 3.03213104726953,
-      "grad_norm": 0.012571502476930618,
-      "learning_rate": 0.00019999546534742703,
-      "loss": 46.0,
-      "step": 39658
-    },
-    {
-      "epoch": 3.0322075042529195,
-      "grad_norm": 0.0036566995549947023,
-      "learning_rate": 0.00019999546511868026,
-      "loss": 46.0,
-      "step": 39659
-    },
-    {
-      "epoch": 3.0322839612363093,
-      "grad_norm": 0.0011799363419413567,
-      "learning_rate": 0.00019999546488992773,
-      "loss": 46.0,
-      "step": 39660
-    },
-    {
-      "epoch": 3.032360418219699,
-      "grad_norm": 0.0037186008412390947,
-      "learning_rate": 0.00019999546466116947,
-      "loss": 46.0,
-      "step": 39661
-    },
-    {
-      "epoch": 3.032436875203089,
-      "grad_norm": 0.001469515380449593,
-      "learning_rate": 0.0001999954644324054,
-      "loss": 46.0,
-      "step": 39662
-    },
-    {
-      "epoch": 3.0325133321864786,
-      "grad_norm": 0.008848035708069801,
-      "learning_rate": 0.00019999546420363559,
-      "loss": 46.0,
-      "step": 39663
-    },
-    {
-      "epoch": 3.0325897891698683,
-      "grad_norm": 0.008599605411291122,
-      "learning_rate": 0.00019999546397485998,
-      "loss": 46.0,
-      "step": 39664
-    },
-    {
-      "epoch": 3.032666246153258,
-      "grad_norm": 0.0025282329879701138,
-      "learning_rate": 0.00019999546374607863,
-      "loss": 46.0,
-      "step": 39665
-    },
-    {
-      "epoch": 3.032742703136648,
-      "grad_norm": 0.010897560976445675,
-      "learning_rate": 0.0001999954635172915,
-      "loss": 46.0,
-      "step": 39666
-    },
-    {
-      "epoch": 3.0328191601200376,
-      "grad_norm": 0.0013498463667929173,
-      "learning_rate": 0.00019999546328849861,
-      "loss": 46.0,
-      "step": 39667
-    },
-    {
-      "epoch": 3.0328956171034274,
-      "grad_norm": 0.0012233887100592256,
-      "learning_rate": 0.00019999546305969994,
-      "loss": 46.0,
-      "step": 39668
-    },
-    {
-      "epoch": 3.0329720740868167,
-      "grad_norm": 0.0016088435659185052,
-      "learning_rate": 0.0001999954628308955,
-      "loss": 46.0,
-      "step": 39669
-    },
-    {
-      "epoch": 3.0330485310702064,
-      "grad_norm": 0.002071057679131627,
-      "learning_rate": 0.00019999546260208529,
-      "loss": 46.0,
-      "step": 39670
-    },
-    {
-      "epoch": 3.033124988053596,
-      "grad_norm": 0.003840462537482381,
-      "learning_rate": 0.00019999546237326932,
-      "loss": 46.0,
-      "step": 39671
-    },
-    {
-      "epoch": 3.033201445036986,
-      "grad_norm": 0.00232309359125793,
-      "learning_rate": 0.0001999954621444476,
-      "loss": 46.0,
-      "step": 39672
-    },
-    {
-      "epoch": 3.0332779020203757,
-      "grad_norm": 0.005251055583357811,
-      "learning_rate": 0.00019999546191562005,
-      "loss": 46.0,
-      "step": 39673
-    },
-    {
-      "epoch": 3.0333543590037655,
-      "grad_norm": 0.0037175242323428392,
-      "learning_rate": 0.00019999546168678677,
-      "loss": 46.0,
-      "step": 39674
-    },
-    {
-      "epoch": 3.0334308159871552,
-      "grad_norm": 0.0022735216189175844,
-      "learning_rate": 0.0001999954614579477,
-      "loss": 46.0,
-      "step": 39675
-    },
-    {
-      "epoch": 3.033507272970545,
-      "grad_norm": 0.002139669144526124,
-      "learning_rate": 0.00019999546122910288,
-      "loss": 46.0,
-      "step": 39676
-    },
-    {
-      "epoch": 3.0335837299539348,
-      "grad_norm": 0.0012831700732931495,
-      "learning_rate": 0.0001999954610002523,
-      "loss": 46.0,
-      "step": 39677
-    },
-    {
-      "epoch": 3.0336601869373245,
-      "grad_norm": 0.004728326108306646,
-      "learning_rate": 0.00019999546077139596,
-      "loss": 46.0,
-      "step": 39678
-    },
-    {
-      "epoch": 3.0337366439207143,
-      "grad_norm": 0.001644921605475247,
-      "learning_rate": 0.0001999954605425338,
-      "loss": 46.0,
-      "step": 39679
-    },
-    {
-      "epoch": 3.0338131009041036,
-      "grad_norm": 0.011548982933163643,
-      "learning_rate": 0.0001999954603136659,
-      "loss": 46.0,
-      "step": 39680
-    },
-    {
-      "epoch": 3.0338895578874934,
-      "grad_norm": 0.0031128483824431896,
-      "learning_rate": 0.00019999546008479224,
-      "loss": 46.0,
-      "step": 39681
-    },
-    {
-      "epoch": 3.033966014870883,
-      "grad_norm": 0.001045048818923533,
-      "learning_rate": 0.0001999954598559128,
-      "loss": 46.0,
-      "step": 39682
-    },
-    {
-      "epoch": 3.034042471854273,
-      "grad_norm": 0.005120352376252413,
-      "learning_rate": 0.0001999954596270276,
-      "loss": 46.0,
-      "step": 39683
-    },
-    {
-      "epoch": 3.0341189288376627,
-      "grad_norm": 0.0011301080230623484,
-      "learning_rate": 0.00019999545939813662,
-      "loss": 46.0,
-      "step": 39684
-    },
-    {
-      "epoch": 3.0341953858210524,
-      "grad_norm": 0.0025274460203945637,
-      "learning_rate": 0.0001999954591692399,
-      "loss": 46.0,
-      "step": 39685
-    },
-    {
-      "epoch": 3.034271842804442,
-      "grad_norm": 0.0019689532928168774,
-      "learning_rate": 0.00019999545894033738,
-      "loss": 46.0,
-      "step": 39686
-    },
-    {
-      "epoch": 3.034348299787832,
-      "grad_norm": 0.002105043036863208,
-      "learning_rate": 0.0001999954587114291,
-      "loss": 46.0,
-      "step": 39687
-    },
-    {
-      "epoch": 3.0344247567712217,
-      "grad_norm": 0.0028234461788088083,
-      "learning_rate": 0.00019999545848251504,
-      "loss": 46.0,
-      "step": 39688
-    },
-    {
-      "epoch": 3.0345012137546115,
-      "grad_norm": 0.0026364759542047977,
-      "learning_rate": 0.0001999954582535952,
-      "loss": 46.0,
-      "step": 39689
-    },
-    {
-      "epoch": 3.034577670738001,
-      "grad_norm": 0.0034467161167412996,
-      "learning_rate": 0.00019999545802466964,
-      "loss": 46.0,
-      "step": 39690
-    },
-    {
-      "epoch": 3.0346541277213905,
-      "grad_norm": 0.002331661758944392,
-      "learning_rate": 0.00019999545779573826,
-      "loss": 46.0,
-      "step": 39691
-    },
-    {
-      "epoch": 3.0347305847047803,
-      "grad_norm": 0.004999334458261728,
-      "learning_rate": 0.00019999545756680114,
-      "loss": 46.0,
-      "step": 39692
-    },
-    {
-      "epoch": 3.03480704168817,
-      "grad_norm": 0.0008896341896615922,
-      "learning_rate": 0.00019999545733785822,
-      "loss": 46.0,
-      "step": 39693
-    },
-    {
-      "epoch": 3.03488349867156,
-      "grad_norm": 0.0012339059030637145,
-      "learning_rate": 0.00019999545710890957,
-      "loss": 46.0,
-      "step": 39694
-    },
-    {
-      "epoch": 3.0349599556549496,
-      "grad_norm": 0.0038829806726425886,
-      "learning_rate": 0.00019999545687995513,
-      "loss": 46.0,
-      "step": 39695
-    },
-    {
-      "epoch": 3.0350364126383393,
-      "grad_norm": 0.0049957334995269775,
-      "learning_rate": 0.00019999545665099492,
-      "loss": 46.0,
-      "step": 39696
-    },
-    {
-      "epoch": 3.035112869621729,
-      "grad_norm": 0.0028166561387479305,
-      "learning_rate": 0.00019999545642202895,
-      "loss": 46.0,
-      "step": 39697
-    },
-    {
-      "epoch": 3.035189326605119,
-      "grad_norm": 0.0021668223198503256,
-      "learning_rate": 0.00019999545619305722,
-      "loss": 46.0,
-      "step": 39698
-    },
-    {
-      "epoch": 3.0352657835885086,
-      "grad_norm": 0.0008365008980035782,
-      "learning_rate": 0.00019999545596407968,
-      "loss": 46.0,
-      "step": 39699
-    },
-    {
-      "epoch": 3.0353422405718984,
-      "grad_norm": 0.001663291361182928,
-      "learning_rate": 0.00019999545573509643,
-      "loss": 46.0,
-      "step": 39700
-    },
-    {
-      "epoch": 3.035418697555288,
-      "grad_norm": 0.0007723391172476113,
-      "learning_rate": 0.00019999545550610737,
-      "loss": 46.0,
-      "step": 39701
-    },
-    {
-      "epoch": 3.0354951545386775,
-      "grad_norm": 0.0010033791186288,
-      "learning_rate": 0.00019999545527711254,
-      "loss": 46.0,
-      "step": 39702
-    },
-    {
-      "epoch": 3.0355716115220672,
-      "grad_norm": 0.0015265808906406164,
-      "learning_rate": 0.00019999545504811194,
-      "loss": 46.0,
-      "step": 39703
-    },
-    {
-      "epoch": 3.035648068505457,
-      "grad_norm": 0.0014681227039545774,
-      "learning_rate": 0.0001999954548191056,
-      "loss": 46.0,
-      "step": 39704
-    },
-    {
-      "epoch": 3.0357245254888467,
-      "grad_norm": 0.00210682419128716,
-      "learning_rate": 0.00019999545459009345,
-      "loss": 46.0,
-      "step": 39705
-    },
-    {
-      "epoch": 3.0358009824722365,
-      "grad_norm": 0.005173744633793831,
-      "learning_rate": 0.00019999545436107555,
-      "loss": 46.0,
-      "step": 39706
-    },
-    {
-      "epoch": 3.0358774394556263,
-      "grad_norm": 0.0062578413635492325,
-      "learning_rate": 0.0001999954541320519,
-      "loss": 46.0,
-      "step": 39707
-    },
-    {
-      "epoch": 3.035953896439016,
-      "grad_norm": 0.0019370083464309573,
-      "learning_rate": 0.00019999545390302247,
-      "loss": 46.0,
-      "step": 39708
-    },
-    {
-      "epoch": 3.036030353422406,
-      "grad_norm": 0.004720513243228197,
-      "learning_rate": 0.00019999545367398725,
-      "loss": 46.0,
-      "step": 39709
-    },
-    {
-      "epoch": 3.0361068104057956,
-      "grad_norm": 0.0015936646377667785,
-      "learning_rate": 0.0001999954534449463,
-      "loss": 46.0,
-      "step": 39710
-    },
-    {
-      "epoch": 3.0361832673891853,
-      "grad_norm": 0.0012520968448370695,
-      "learning_rate": 0.00019999545321589955,
-      "loss": 46.0,
-      "step": 39711
-    },
-    {
-      "epoch": 3.036259724372575,
-      "grad_norm": 0.004156424663960934,
-      "learning_rate": 0.00019999545298684702,
-      "loss": 46.0,
-      "step": 39712
-    },
-    {
-      "epoch": 3.0363361813559644,
-      "grad_norm": 0.0021326541900634766,
-      "learning_rate": 0.00019999545275778874,
-      "loss": 46.0,
-      "step": 39713
-    },
-    {
-      "epoch": 3.036412638339354,
-      "grad_norm": 0.0015934498514980078,
-      "learning_rate": 0.00019999545252872468,
-      "loss": 46.0,
-      "step": 39714
-    },
-    {
-      "epoch": 3.036489095322744,
-      "grad_norm": 0.0014560369309037924,
-      "learning_rate": 0.00019999545229965485,
-      "loss": 46.0,
-      "step": 39715
-    },
-    {
-      "epoch": 3.0365655523061337,
-      "grad_norm": 0.0027113950345665216,
-      "learning_rate": 0.00019999545207057928,
-      "loss": 46.0,
-      "step": 39716
-    },
-    {
-      "epoch": 3.0366420092895234,
-      "grad_norm": 0.002184296492487192,
-      "learning_rate": 0.00019999545184149793,
-      "loss": 46.0,
-      "step": 39717
-    },
-    {
-      "epoch": 3.036718466272913,
-      "grad_norm": 0.0016092031728476286,
-      "learning_rate": 0.0001999954516124108,
-      "loss": 46.0,
-      "step": 39718
-    },
-    {
-      "epoch": 3.036794923256303,
-      "grad_norm": 0.0032755061984062195,
-      "learning_rate": 0.0001999954513833179,
-      "loss": 46.0,
-      "step": 39719
-    },
-    {
-      "epoch": 3.0368713802396927,
-      "grad_norm": 0.0016704684821888804,
-      "learning_rate": 0.00019999545115421925,
-      "loss": 46.0,
-      "step": 39720
-    },
-    {
-      "epoch": 3.0369478372230825,
-      "grad_norm": 0.0032511057797819376,
-      "learning_rate": 0.0001999954509251148,
-      "loss": 46.0,
-      "step": 39721
-    },
-    {
-      "epoch": 3.0370242942064722,
-      "grad_norm": 0.0009438850684091449,
-      "learning_rate": 0.0001999954506960046,
-      "loss": 46.0,
-      "step": 39722
-    },
-    {
-      "epoch": 3.037100751189862,
-      "grad_norm": 0.005450352095067501,
-      "learning_rate": 0.0001999954504668886,
-      "loss": 46.0,
-      "step": 39723
-    },
-    {
-      "epoch": 3.0371772081732513,
-      "grad_norm": 0.00190682930406183,
-      "learning_rate": 0.00019999545023776687,
-      "loss": 46.0,
-      "step": 39724
-    },
-    {
-      "epoch": 3.037253665156641,
-      "grad_norm": 0.0013446015072986484,
-      "learning_rate": 0.00019999545000863936,
-      "loss": 46.0,
-      "step": 39725
-    },
-    {
-      "epoch": 3.037330122140031,
-      "grad_norm": 0.002689749002456665,
-      "learning_rate": 0.00019999544977950608,
-      "loss": 46.0,
-      "step": 39726
-    },
-    {
-      "epoch": 3.0374065791234206,
-      "grad_norm": 0.002143718535080552,
-      "learning_rate": 0.00019999544955036705,
-      "loss": 46.0,
-      "step": 39727
-    },
-    {
-      "epoch": 3.0374830361068104,
-      "grad_norm": 0.0004537251952569932,
-      "learning_rate": 0.0001999954493212222,
-      "loss": 46.0,
-      "step": 39728
-    },
-    {
-      "epoch": 3.0375594930902,
-      "grad_norm": 0.002278085332363844,
-      "learning_rate": 0.00019999544909207162,
-      "loss": 46.0,
-      "step": 39729
-    },
-    {
-      "epoch": 3.03763595007359,
-      "grad_norm": 0.0013600974343717098,
-      "learning_rate": 0.00019999544886291525,
-      "loss": 46.0,
-      "step": 39730
-    },
-    {
-      "epoch": 3.0377124070569796,
-      "grad_norm": 0.006506752222776413,
-      "learning_rate": 0.00019999544863375315,
-      "loss": 46.0,
-      "step": 39731
-    },
-    {
-      "epoch": 3.0377888640403694,
-      "grad_norm": 0.0020644119940698147,
-      "learning_rate": 0.00019999544840458523,
-      "loss": 46.0,
-      "step": 39732
-    },
-    {
-      "epoch": 3.037865321023759,
-      "grad_norm": 0.0008373287855647504,
-      "learning_rate": 0.00019999544817541156,
-      "loss": 46.0,
-      "step": 39733
-    },
-    {
-      "epoch": 3.037941778007149,
-      "grad_norm": 0.0018869703635573387,
-      "learning_rate": 0.00019999544794623212,
-      "loss": 46.0,
-      "step": 39734
-    },
-    {
-      "epoch": 3.0380182349905382,
-      "grad_norm": 0.004040055442601442,
-      "learning_rate": 0.00019999544771704693,
-      "loss": 46.0,
-      "step": 39735
-    },
-    {
-      "epoch": 3.038094691973928,
-      "grad_norm": 0.004738978575915098,
-      "learning_rate": 0.00019999544748785595,
-      "loss": 46.0,
-      "step": 39736
-    },
-    {
-      "epoch": 3.0381711489573178,
-      "grad_norm": 0.002792940242215991,
-      "learning_rate": 0.0001999954472586592,
-      "loss": 46.0,
-      "step": 39737
-    },
-    {
-      "epoch": 3.0382476059407075,
-      "grad_norm": 0.0022057651076465845,
-      "learning_rate": 0.0001999954470294567,
-      "loss": 46.0,
-      "step": 39738
-    },
-    {
-      "epoch": 3.0383240629240973,
-      "grad_norm": 0.0037713306955993176,
-      "learning_rate": 0.00019999544680024842,
-      "loss": 46.0,
-      "step": 39739
-    },
-    {
-      "epoch": 3.038400519907487,
-      "grad_norm": 0.0030813212506473064,
-      "learning_rate": 0.00019999544657103437,
-      "loss": 46.0,
-      "step": 39740
-    },
-    {
-      "epoch": 3.038476976890877,
-      "grad_norm": 0.002665862673893571,
-      "learning_rate": 0.00019999544634181454,
-      "loss": 46.0,
-      "step": 39741
-    },
-    {
-      "epoch": 3.0385534338742666,
-      "grad_norm": 0.001664767973124981,
-      "learning_rate": 0.00019999544611258897,
-      "loss": 46.0,
-      "step": 39742
-    },
-    {
-      "epoch": 3.0386298908576563,
-      "grad_norm": 0.003835249226540327,
-      "learning_rate": 0.00019999544588335762,
-      "loss": 46.0,
-      "step": 39743
-    },
-    {
-      "epoch": 3.038706347841046,
-      "grad_norm": 0.0021365885622799397,
-      "learning_rate": 0.00019999544565412047,
-      "loss": 46.0,
-      "step": 39744
-    },
-    {
-      "epoch": 3.038782804824436,
-      "grad_norm": 0.0014744889922440052,
-      "learning_rate": 0.00019999544542487758,
-      "loss": 46.0,
-      "step": 39745
-    },
-    {
-      "epoch": 3.038859261807825,
-      "grad_norm": 0.008801175281405449,
-      "learning_rate": 0.0001999954451956289,
-      "loss": 46.0,
-      "step": 39746
-    },
-    {
-      "epoch": 3.038935718791215,
-      "grad_norm": 0.0016136305639520288,
-      "learning_rate": 0.00019999544496637447,
-      "loss": 46.0,
-      "step": 39747
-    },
-    {
-      "epoch": 3.0390121757746047,
-      "grad_norm": 0.0026654433459043503,
-      "learning_rate": 0.00019999544473711428,
-      "loss": 46.0,
-      "step": 39748
-    },
-    {
-      "epoch": 3.0390886327579945,
-      "grad_norm": 0.0010333481477573514,
-      "learning_rate": 0.0001999954445078483,
-      "loss": 46.0,
-      "step": 39749
-    },
-    {
-      "epoch": 3.039165089741384,
-      "grad_norm": 0.005302795208990574,
-      "learning_rate": 0.00019999544427857656,
-      "loss": 46.0,
-      "step": 39750
-    },
-    {
-      "epoch": 3.039241546724774,
-      "grad_norm": 0.0008502639248035848,
-      "learning_rate": 0.00019999544404929905,
-      "loss": 46.0,
-      "step": 39751
-    },
-    {
-      "epoch": 3.0393180037081637,
-      "grad_norm": 0.0027508113998919725,
-      "learning_rate": 0.00019999544382001577,
-      "loss": 46.0,
-      "step": 39752
-    },
-    {
-      "epoch": 3.0393944606915535,
-      "grad_norm": 0.0013168102595955133,
-      "learning_rate": 0.00019999544359072672,
-      "loss": 46.0,
-      "step": 39753
-    },
-    {
-      "epoch": 3.0394709176749433,
-      "grad_norm": 0.002552098361775279,
-      "learning_rate": 0.0001999954433614319,
-      "loss": 46.0,
-      "step": 39754
-    },
-    {
-      "epoch": 3.039547374658333,
-      "grad_norm": 0.0025400989688932896,
-      "learning_rate": 0.00019999544313213132,
-      "loss": 46.0,
-      "step": 39755
-    },
-    {
-      "epoch": 3.039623831641723,
-      "grad_norm": 0.0025748638436198235,
-      "learning_rate": 0.00019999544290282497,
-      "loss": 46.0,
-      "step": 39756
-    },
-    {
-      "epoch": 3.039700288625112,
-      "grad_norm": 0.0021021731663495302,
-      "learning_rate": 0.00019999544267351285,
-      "loss": 46.0,
-      "step": 39757
-    },
-    {
-      "epoch": 3.039776745608502,
-      "grad_norm": 0.002064772415906191,
-      "learning_rate": 0.00019999544244419495,
-      "loss": 46.0,
-      "step": 39758
-    },
-    {
-      "epoch": 3.0398532025918916,
-      "grad_norm": 0.003540665376931429,
-      "learning_rate": 0.0001999954422148713,
-      "loss": 46.0,
-      "step": 39759
-    },
-    {
-      "epoch": 3.0399296595752814,
-      "grad_norm": 0.0022401451133191586,
-      "learning_rate": 0.00019999544198554182,
-      "loss": 46.0,
-      "step": 39760
-    },
-    {
-      "epoch": 3.040006116558671,
-      "grad_norm": 0.003381523070856929,
-      "learning_rate": 0.00019999544175620663,
-      "loss": 46.0,
-      "step": 39761
-    },
-    {
-      "epoch": 3.040082573542061,
-      "grad_norm": 0.003989202901721001,
-      "learning_rate": 0.00019999544152686567,
-      "loss": 46.0,
-      "step": 39762
-    },
-    {
-      "epoch": 3.0401590305254507,
-      "grad_norm": 0.0036354924086481333,
-      "learning_rate": 0.00019999544129751894,
-      "loss": 46.0,
-      "step": 39763
-    },
-    {
-      "epoch": 3.0402354875088404,
-      "grad_norm": 0.0050549875013530254,
-      "learning_rate": 0.0001999954410681664,
-      "loss": 46.0,
-      "step": 39764
-    },
-    {
-      "epoch": 3.04031194449223,
-      "grad_norm": 0.0030578419100493193,
-      "learning_rate": 0.00019999544083880815,
-      "loss": 46.0,
-      "step": 39765
-    },
-    {
-      "epoch": 3.04038840147562,
-      "grad_norm": 0.004450567532330751,
-      "learning_rate": 0.0001999954406094441,
-      "loss": 46.0,
-      "step": 39766
-    },
-    {
-      "epoch": 3.0404648584590097,
-      "grad_norm": 0.003954468294978142,
-      "learning_rate": 0.00019999544038007424,
-      "loss": 46.0,
-      "step": 39767
-    },
-    {
-      "epoch": 3.040541315442399,
-      "grad_norm": 0.003068147925660014,
-      "learning_rate": 0.00019999544015069867,
-      "loss": 46.0,
-      "step": 39768
-    },
-    {
-      "epoch": 3.040617772425789,
-      "grad_norm": 0.003508657217025757,
-      "learning_rate": 0.0001999954399213173,
-      "loss": 46.0,
-      "step": 39769
-    },
-    {
-      "epoch": 3.0406942294091786,
-      "grad_norm": 0.001262909616343677,
-      "learning_rate": 0.0001999954396919302,
-      "loss": 46.0,
-      "step": 39770
-    },
-    {
-      "epoch": 3.0407706863925683,
-      "grad_norm": 0.0026834493037313223,
-      "learning_rate": 0.0001999954394625373,
-      "loss": 46.0,
-      "step": 39771
-    },
-    {
-      "epoch": 3.040847143375958,
-      "grad_norm": 0.004251353908330202,
-      "learning_rate": 0.00019999543923313864,
-      "loss": 46.0,
-      "step": 39772
-    },
-    {
-      "epoch": 3.040923600359348,
-      "grad_norm": 0.002146467100828886,
-      "learning_rate": 0.0001999954390037342,
-      "loss": 46.0,
-      "step": 39773
-    },
-    {
-      "epoch": 3.0410000573427376,
-      "grad_norm": 0.0026370815467089415,
-      "learning_rate": 0.00019999543877432402,
-      "loss": 46.0,
-      "step": 39774
-    },
-    {
-      "epoch": 3.0410765143261274,
-      "grad_norm": 0.0014033066108822823,
-      "learning_rate": 0.00019999543854490803,
-      "loss": 46.0,
-      "step": 39775
-    },
-    {
-      "epoch": 3.041152971309517,
-      "grad_norm": 0.0017862417735159397,
-      "learning_rate": 0.0001999954383154863,
-      "loss": 46.0,
-      "step": 39776
-    },
-    {
-      "epoch": 3.041229428292907,
-      "grad_norm": 0.0016954976599663496,
-      "learning_rate": 0.0001999954380860588,
-      "loss": 46.0,
-      "step": 39777
-    },
-    {
-      "epoch": 3.0413058852762966,
-      "grad_norm": 0.0009763639536686242,
-      "learning_rate": 0.0001999954378566255,
-      "loss": 46.0,
-      "step": 39778
-    },
-    {
-      "epoch": 3.041382342259686,
-      "grad_norm": 0.0030006549786776304,
-      "learning_rate": 0.00019999543762718645,
-      "loss": 46.0,
-      "step": 39779
-    },
-    {
-      "epoch": 3.0414587992430757,
-      "grad_norm": 0.0017674047267064452,
-      "learning_rate": 0.00019999543739774165,
-      "loss": 46.0,
-      "step": 39780
-    },
-    {
-      "epoch": 3.0415352562264655,
-      "grad_norm": 0.0012382587883621454,
-      "learning_rate": 0.00019999543716829105,
-      "loss": 46.0,
-      "step": 39781
-    },
-    {
-      "epoch": 3.0416117132098552,
-      "grad_norm": 0.0008127946057356894,
-      "learning_rate": 0.00019999543693883468,
-      "loss": 46.0,
-      "step": 39782
-    },
-    {
-      "epoch": 3.041688170193245,
-      "grad_norm": 0.00219341809861362,
-      "learning_rate": 0.00019999543670937256,
-      "loss": 46.0,
-      "step": 39783
-    },
-    {
-      "epoch": 3.0417646271766348,
-      "grad_norm": 0.0014569797785952687,
-      "learning_rate": 0.0001999954364799047,
-      "loss": 46.0,
-      "step": 39784
-    },
-    {
-      "epoch": 3.0418410841600245,
-      "grad_norm": 0.0006897144485265017,
-      "learning_rate": 0.000199995436250431,
-      "loss": 46.0,
-      "step": 39785
-    },
-    {
-      "epoch": 3.0419175411434143,
-      "grad_norm": 0.0019148170249536633,
-      "learning_rate": 0.0001999954360209516,
-      "loss": 46.0,
-      "step": 39786
-    },
-    {
-      "epoch": 3.041993998126804,
-      "grad_norm": 0.0007004592334851623,
-      "learning_rate": 0.00019999543579146635,
-      "loss": 46.0,
-      "step": 39787
-    },
-    {
-      "epoch": 3.042070455110194,
-      "grad_norm": 0.0010098963975906372,
-      "learning_rate": 0.0001999954355619754,
-      "loss": 46.0,
-      "step": 39788
-    },
-    {
-      "epoch": 3.0421469120935836,
-      "grad_norm": 0.0032262811437249184,
-      "learning_rate": 0.00019999543533247866,
-      "loss": 46.0,
-      "step": 39789
-    },
-    {
-      "epoch": 3.042223369076973,
-      "grad_norm": 0.001586111611686647,
-      "learning_rate": 0.00019999543510297615,
-      "loss": 46.0,
-      "step": 39790
-    },
-    {
-      "epoch": 3.0422998260603626,
-      "grad_norm": 0.0019963011145591736,
-      "learning_rate": 0.00019999543487346787,
-      "loss": 46.0,
-      "step": 39791
-    },
-    {
-      "epoch": 3.0423762830437524,
-      "grad_norm": 0.0032852094154804945,
-      "learning_rate": 0.00019999543464395382,
-      "loss": 46.0,
-      "step": 39792
-    },
-    {
-      "epoch": 3.042452740027142,
-      "grad_norm": 0.0011593063827604055,
-      "learning_rate": 0.000199995434414434,
-      "loss": 46.0,
-      "step": 39793
-    },
-    {
-      "epoch": 3.042529197010532,
-      "grad_norm": 0.0022101178765296936,
-      "learning_rate": 0.00019999543418490842,
-      "loss": 46.0,
-      "step": 39794
-    },
-    {
-      "epoch": 3.0426056539939217,
-      "grad_norm": 0.002550235018134117,
-      "learning_rate": 0.00019999543395537707,
-      "loss": 46.0,
-      "step": 39795
-    },
-    {
-      "epoch": 3.0426821109773114,
-      "grad_norm": 0.008202235214412212,
-      "learning_rate": 0.00019999543372583992,
-      "loss": 46.0,
-      "step": 39796
-    },
-    {
-      "epoch": 3.042758567960701,
-      "grad_norm": 0.002513185376301408,
-      "learning_rate": 0.00019999543349629706,
-      "loss": 46.0,
-      "step": 39797
-    },
-    {
-      "epoch": 3.042835024944091,
-      "grad_norm": 0.0015979657182469964,
-      "learning_rate": 0.00019999543326674836,
-      "loss": 46.0,
-      "step": 39798
-    },
-    {
-      "epoch": 3.0429114819274807,
-      "grad_norm": 0.0013994803884997964,
-      "learning_rate": 0.00019999543303719395,
-      "loss": 46.0,
-      "step": 39799
-    },
-    {
-      "epoch": 3.04298793891087,
-      "grad_norm": 0.0030840723775327206,
-      "learning_rate": 0.00019999543280763377,
-      "loss": 46.0,
-      "step": 39800
-    },
-    {
-      "epoch": 3.04306439589426,
-      "grad_norm": 0.0020172528456896544,
-      "learning_rate": 0.00019999543257806778,
-      "loss": 46.0,
-      "step": 39801
-    },
-    {
-      "epoch": 3.0431408528776496,
-      "grad_norm": 0.005486840847879648,
-      "learning_rate": 0.00019999543234849605,
-      "loss": 46.0,
-      "step": 39802
-    },
-    {
-      "epoch": 3.0432173098610393,
-      "grad_norm": 0.0026672601234167814,
-      "learning_rate": 0.00019999543211891854,
-      "loss": 46.0,
-      "step": 39803
-    },
-    {
-      "epoch": 3.043293766844429,
-      "grad_norm": 0.0010024961084127426,
-      "learning_rate": 0.00019999543188933526,
-      "loss": 46.0,
-      "step": 39804
-    },
-    {
-      "epoch": 3.043370223827819,
-      "grad_norm": 0.0014839605428278446,
-      "learning_rate": 0.00019999543165974618,
-      "loss": 46.0,
-      "step": 39805
-    },
-    {
-      "epoch": 3.0434466808112086,
-      "grad_norm": 0.0011847109999507666,
-      "learning_rate": 0.00019999543143015138,
-      "loss": 46.0,
-      "step": 39806
-    },
-    {
-      "epoch": 3.0435231377945984,
-      "grad_norm": 0.0035277872812002897,
-      "learning_rate": 0.0001999954312005508,
-      "loss": 46.0,
-      "step": 39807
-    },
-    {
-      "epoch": 3.043599594777988,
-      "grad_norm": 0.002852214267477393,
-      "learning_rate": 0.00019999543097094444,
-      "loss": 46.0,
-      "step": 39808
-    },
-    {
-      "epoch": 3.043676051761378,
-      "grad_norm": 0.0020340923219919205,
-      "learning_rate": 0.00019999543074133232,
-      "loss": 46.0,
-      "step": 39809
-    },
-    {
-      "epoch": 3.0437525087447677,
-      "grad_norm": 0.001955203479155898,
-      "learning_rate": 0.00019999543051171442,
-      "loss": 46.0,
-      "step": 39810
-    },
-    {
-      "epoch": 3.0438289657281574,
-      "grad_norm": 0.002027090871706605,
-      "learning_rate": 0.00019999543028209076,
-      "loss": 46.0,
-      "step": 39811
-    },
-    {
-      "epoch": 3.0439054227115467,
-      "grad_norm": 0.0023426534608006477,
-      "learning_rate": 0.00019999543005246132,
-      "loss": 46.0,
-      "step": 39812
-    },
-    {
-      "epoch": 3.0439818796949365,
-      "grad_norm": 0.0010036531602963805,
-      "learning_rate": 0.00019999542982282613,
-      "loss": 46.0,
-      "step": 39813
-    },
-    {
-      "epoch": 3.0440583366783263,
-      "grad_norm": 0.003153854515403509,
-      "learning_rate": 0.00019999542959318515,
-      "loss": 46.0,
-      "step": 39814
-    },
-    {
-      "epoch": 3.044134793661716,
-      "grad_norm": 0.0033195195719599724,
-      "learning_rate": 0.00019999542936353844,
-      "loss": 46.0,
-      "step": 39815
-    },
-    {
-      "epoch": 3.044211250645106,
-      "grad_norm": 0.004860568791627884,
-      "learning_rate": 0.0001999954291338859,
-      "loss": 46.0,
-      "step": 39816
-    },
-    {
-      "epoch": 3.0442877076284955,
-      "grad_norm": 0.0020198491401970387,
-      "learning_rate": 0.00019999542890422763,
-      "loss": 46.0,
-      "step": 39817
-    },
-    {
-      "epoch": 3.0443641646118853,
-      "grad_norm": 0.0028632120229303837,
-      "learning_rate": 0.0001999954286745636,
-      "loss": 46.0,
-      "step": 39818
-    },
-    {
-      "epoch": 3.044440621595275,
-      "grad_norm": 0.001592725282534957,
-      "learning_rate": 0.00019999542844489375,
-      "loss": 46.0,
-      "step": 39819
-    },
-    {
-      "epoch": 3.044517078578665,
-      "grad_norm": 0.004052300006151199,
-      "learning_rate": 0.0001999954282152182,
-      "loss": 46.0,
-      "step": 39820
-    },
-    {
-      "epoch": 3.0445935355620546,
-      "grad_norm": 0.0026906216517090797,
-      "learning_rate": 0.00019999542798553683,
-      "loss": 46.0,
-      "step": 39821
-    },
-    {
-      "epoch": 3.044669992545444,
-      "grad_norm": 0.0012166756205260754,
-      "learning_rate": 0.0001999954277558497,
-      "loss": 46.0,
-      "step": 39822
-    },
-    {
-      "epoch": 3.0447464495288337,
-      "grad_norm": 0.0011185735929757357,
-      "learning_rate": 0.0001999954275261568,
-      "loss": 46.0,
-      "step": 39823
-    },
-    {
-      "epoch": 3.0448229065122234,
-      "grad_norm": 0.001580333337187767,
-      "learning_rate": 0.00019999542729645816,
-      "loss": 46.0,
-      "step": 39824
-    },
-    {
-      "epoch": 3.044899363495613,
-      "grad_norm": 0.004871820565313101,
-      "learning_rate": 0.00019999542706675372,
-      "loss": 46.0,
-      "step": 39825
-    },
-    {
-      "epoch": 3.044975820479003,
-      "grad_norm": 0.0006713769398629665,
-      "learning_rate": 0.0001999954268370435,
-      "loss": 46.0,
-      "step": 39826
-    },
-    {
-      "epoch": 3.0450522774623927,
-      "grad_norm": 0.0010941119398921728,
-      "learning_rate": 0.00019999542660732755,
-      "loss": 46.0,
-      "step": 39827
-    },
-    {
-      "epoch": 3.0451287344457825,
-      "grad_norm": 0.004415113478899002,
-      "learning_rate": 0.00019999542637760581,
-      "loss": 46.0,
-      "step": 39828
-    },
-    {
-      "epoch": 3.0452051914291722,
-      "grad_norm": 0.0017356699099764228,
-      "learning_rate": 0.00019999542614787828,
-      "loss": 46.0,
-      "step": 39829
-    },
-    {
-      "epoch": 3.045281648412562,
-      "grad_norm": 0.0023002484813332558,
-      "learning_rate": 0.00019999542591814503,
-      "loss": 46.0,
-      "step": 39830
-    },
-    {
-      "epoch": 3.0453581053959518,
-      "grad_norm": 0.0011726251104846597,
-      "learning_rate": 0.00019999542568840598,
-      "loss": 46.0,
-      "step": 39831
-    },
-    {
-      "epoch": 3.0454345623793415,
-      "grad_norm": 0.0024728458374738693,
-      "learning_rate": 0.00019999542545866118,
-      "loss": 46.0,
-      "step": 39832
-    },
-    {
-      "epoch": 3.045511019362731,
-      "grad_norm": 0.0032813907600939274,
-      "learning_rate": 0.00019999542522891058,
-      "loss": 46.0,
-      "step": 39833
-    },
-    {
-      "epoch": 3.0455874763461206,
-      "grad_norm": 0.001836483716033399,
-      "learning_rate": 0.00019999542499915424,
-      "loss": 46.0,
-      "step": 39834
-    },
-    {
-      "epoch": 3.0456639333295104,
-      "grad_norm": 0.001598047441802919,
-      "learning_rate": 0.00019999542476939212,
-      "loss": 46.0,
-      "step": 39835
-    },
-    {
-      "epoch": 3.0457403903129,
-      "grad_norm": 0.0015207421965897083,
-      "learning_rate": 0.0001999954245396242,
-      "loss": 46.0,
-      "step": 39836
-    },
-    {
-      "epoch": 3.04581684729629,
-      "grad_norm": 0.004801145289093256,
-      "learning_rate": 0.00019999542430985053,
-      "loss": 46.0,
-      "step": 39837
-    },
-    {
-      "epoch": 3.0458933042796796,
-      "grad_norm": 0.0011858928482979536,
-      "learning_rate": 0.00019999542408007112,
-      "loss": 46.0,
-      "step": 39838
-    },
-    {
-      "epoch": 3.0459697612630694,
-      "grad_norm": 0.001516296761110425,
-      "learning_rate": 0.0001999954238502859,
-      "loss": 46.0,
-      "step": 39839
-    },
-    {
-      "epoch": 3.046046218246459,
-      "grad_norm": 0.0020369929261505604,
-      "learning_rate": 0.00019999542362049495,
-      "loss": 46.0,
-      "step": 39840
-    },
-    {
-      "epoch": 3.046122675229849,
-      "grad_norm": 0.0007960710790939629,
-      "learning_rate": 0.00019999542339069822,
-      "loss": 46.0,
-      "step": 39841
-    },
-    {
-      "epoch": 3.0461991322132387,
-      "grad_norm": 0.0007086318801157176,
-      "learning_rate": 0.0001999954231608957,
-      "loss": 46.0,
-      "step": 39842
-    },
-    {
-      "epoch": 3.0462755891966284,
-      "grad_norm": 0.00118496164213866,
-      "learning_rate": 0.0001999954229310874,
-      "loss": 46.0,
-      "step": 39843
-    },
-    {
-      "epoch": 3.0463520461800178,
-      "grad_norm": 0.00048624732880853117,
-      "learning_rate": 0.00019999542270127338,
-      "loss": 46.0,
-      "step": 39844
-    },
-    {
-      "epoch": 3.0464285031634075,
-      "grad_norm": 0.0015608640387654305,
-      "learning_rate": 0.00019999542247145358,
-      "loss": 46.0,
-      "step": 39845
-    },
-    {
-      "epoch": 3.0465049601467973,
-      "grad_norm": 0.0054060607217252254,
-      "learning_rate": 0.00019999542224162796,
-      "loss": 46.0,
-      "step": 39846
-    },
-    {
-      "epoch": 3.046581417130187,
-      "grad_norm": 0.0013889011461287737,
-      "learning_rate": 0.00019999542201179661,
-      "loss": 46.0,
-      "step": 39847
-    },
-    {
-      "epoch": 3.046657874113577,
-      "grad_norm": 0.0025718859396874905,
-      "learning_rate": 0.00019999542178195952,
-      "loss": 46.0,
-      "step": 39848
-    },
-    {
-      "epoch": 3.0467343310969666,
-      "grad_norm": 0.002158508636057377,
-      "learning_rate": 0.0001999954215521166,
-      "loss": 46.0,
-      "step": 39849
-    },
-    {
-      "epoch": 3.0468107880803563,
-      "grad_norm": 0.003430926240980625,
-      "learning_rate": 0.00019999542132226794,
-      "loss": 46.0,
-      "step": 39850
-    },
-    {
-      "epoch": 3.046887245063746,
-      "grad_norm": 0.003158948151394725,
-      "learning_rate": 0.00019999542109241353,
-      "loss": 46.0,
-      "step": 39851
-    },
-    {
-      "epoch": 3.046963702047136,
-      "grad_norm": 0.001974754733964801,
-      "learning_rate": 0.00019999542086255332,
-      "loss": 46.0,
-      "step": 39852
-    },
-    {
-      "epoch": 3.0470401590305256,
-      "grad_norm": 0.0017480666283518076,
-      "learning_rate": 0.00019999542063268736,
-      "loss": 46.0,
-      "step": 39853
-    },
-    {
-      "epoch": 3.0471166160139154,
-      "grad_norm": 0.0012514068512246013,
-      "learning_rate": 0.0001999954204028156,
-      "loss": 46.0,
-      "step": 39854
-    },
-    {
-      "epoch": 3.0471930729973047,
-      "grad_norm": 0.002120128832757473,
-      "learning_rate": 0.00019999542017293812,
-      "loss": 46.0,
-      "step": 39855
-    },
-    {
-      "epoch": 3.0472695299806944,
-      "grad_norm": 0.003946150653064251,
-      "learning_rate": 0.00019999541994305484,
-      "loss": 46.0,
-      "step": 39856
-    },
-    {
-      "epoch": 3.047345986964084,
-      "grad_norm": 0.0014776972820982337,
-      "learning_rate": 0.00019999541971316576,
-      "loss": 46.0,
-      "step": 39857
-    },
-    {
-      "epoch": 3.047422443947474,
-      "grad_norm": 0.001937778084538877,
-      "learning_rate": 0.00019999541948327097,
-      "loss": 46.0,
-      "step": 39858
-    },
-    {
-      "epoch": 3.0474989009308637,
-      "grad_norm": 0.001531258225440979,
-      "learning_rate": 0.0001999954192533704,
-      "loss": 46.0,
-      "step": 39859
-    },
-    {
-      "epoch": 3.0475753579142535,
-      "grad_norm": 0.0019490448758006096,
-      "learning_rate": 0.00019999541902346402,
-      "loss": 46.0,
-      "step": 39860
-    },
-    {
-      "epoch": 3.0476518148976433,
-      "grad_norm": 0.003566277679055929,
-      "learning_rate": 0.0001999954187935519,
-      "loss": 46.0,
-      "step": 39861
-    },
-    {
-      "epoch": 3.047728271881033,
-      "grad_norm": 0.0026820912025868893,
-      "learning_rate": 0.00019999541856363401,
-      "loss": 46.0,
-      "step": 39862
-    },
-    {
-      "epoch": 3.0478047288644228,
-      "grad_norm": 0.0031591744627803564,
-      "learning_rate": 0.00019999541833371035,
-      "loss": 46.0,
-      "step": 39863
-    },
-    {
-      "epoch": 3.0478811858478125,
-      "grad_norm": 0.002295204671099782,
-      "learning_rate": 0.00019999541810378094,
-      "loss": 46.0,
-      "step": 39864
-    },
-    {
-      "epoch": 3.0479576428312023,
-      "grad_norm": 0.002604244276881218,
-      "learning_rate": 0.00019999541787384573,
-      "loss": 46.0,
-      "step": 39865
-    },
-    {
-      "epoch": 3.0480340998145916,
-      "grad_norm": 0.0014944911235943437,
-      "learning_rate": 0.00019999541764390477,
-      "loss": 46.0,
-      "step": 39866
-    },
-    {
-      "epoch": 3.0481105567979814,
-      "grad_norm": 0.004636420402675867,
-      "learning_rate": 0.000199995417413958,
-      "loss": 46.0,
-      "step": 39867
-    },
-    {
-      "epoch": 3.048187013781371,
-      "grad_norm": 0.0014118136605247855,
-      "learning_rate": 0.0001999954171840055,
-      "loss": 46.0,
-      "step": 39868
-    },
-    {
-      "epoch": 3.048263470764761,
-      "grad_norm": 0.003935117274522781,
-      "learning_rate": 0.00019999541695404726,
-      "loss": 46.0,
-      "step": 39869
-    },
-    {
-      "epoch": 3.0483399277481507,
-      "grad_norm": 0.0033773519098758698,
-      "learning_rate": 0.0001999954167240832,
-      "loss": 46.0,
-      "step": 39870
-    },
-    {
-      "epoch": 3.0484163847315404,
-      "grad_norm": 0.0017816373147070408,
-      "learning_rate": 0.00019999541649411338,
-      "loss": 46.0,
-      "step": 39871
-    },
-    {
-      "epoch": 3.04849284171493,
-      "grad_norm": 0.0010630270699039102,
-      "learning_rate": 0.0001999954162641378,
-      "loss": 46.0,
-      "step": 39872
-    },
-    {
-      "epoch": 3.04856929869832,
-      "grad_norm": 0.002940055448561907,
-      "learning_rate": 0.00019999541603415644,
-      "loss": 46.0,
-      "step": 39873
-    },
-    {
-      "epoch": 3.0486457556817097,
-      "grad_norm": 0.000761278672143817,
-      "learning_rate": 0.00019999541580416932,
-      "loss": 46.0,
-      "step": 39874
-    },
-    {
-      "epoch": 3.0487222126650995,
-      "grad_norm": 0.0010927848052233458,
-      "learning_rate": 0.00019999541557417643,
-      "loss": 46.0,
-      "step": 39875
-    },
-    {
-      "epoch": 3.0487986696484892,
-      "grad_norm": 0.0034057158045470715,
-      "learning_rate": 0.0001999954153441778,
-      "loss": 46.0,
-      "step": 39876
-    },
-    {
-      "epoch": 3.0488751266318785,
-      "grad_norm": 0.002073639538139105,
-      "learning_rate": 0.00019999541511417335,
-      "loss": 46.0,
-      "step": 39877
-    },
-    {
-      "epoch": 3.0489515836152683,
-      "grad_norm": 0.0007386656361632049,
-      "learning_rate": 0.00019999541488416314,
-      "loss": 46.0,
-      "step": 39878
-    },
-    {
-      "epoch": 3.049028040598658,
-      "grad_norm": 0.002264376962557435,
-      "learning_rate": 0.00019999541465414719,
-      "loss": 46.0,
-      "step": 39879
-    },
-    {
-      "epoch": 3.049104497582048,
-      "grad_norm": 0.001007546205073595,
-      "learning_rate": 0.00019999541442412543,
-      "loss": 46.0,
-      "step": 39880
-    },
-    {
-      "epoch": 3.0491809545654376,
-      "grad_norm": 0.0024832834023982286,
-      "learning_rate": 0.00019999541419409792,
-      "loss": 46.0,
-      "step": 39881
-    },
-    {
-      "epoch": 3.0492574115488273,
-      "grad_norm": 0.0027159901801496744,
-      "learning_rate": 0.00019999541396406465,
-      "loss": 46.0,
-      "step": 39882
-    },
-    {
-      "epoch": 3.049333868532217,
-      "grad_norm": 0.0014653961407020688,
-      "learning_rate": 0.00019999541373402562,
-      "loss": 46.0,
-      "step": 39883
-    },
-    {
-      "epoch": 3.049410325515607,
-      "grad_norm": 0.0017313024727627635,
-      "learning_rate": 0.00019999541350398082,
-      "loss": 46.0,
-      "step": 39884
-    },
-    {
-      "epoch": 3.0494867824989966,
-      "grad_norm": 0.00138534524012357,
-      "learning_rate": 0.0001999954132739302,
-      "loss": 46.0,
-      "step": 39885
-    },
-    {
-      "epoch": 3.0495632394823864,
-      "grad_norm": 0.0031168884597718716,
-      "learning_rate": 0.00019999541304387388,
-      "loss": 46.0,
-      "step": 39886
-    },
-    {
-      "epoch": 3.049639696465776,
-      "grad_norm": 0.0014258017763495445,
-      "learning_rate": 0.00019999541281381174,
-      "loss": 46.0,
-      "step": 39887
-    },
-    {
-      "epoch": 3.0497161534491655,
-      "grad_norm": 0.002162324730306864,
-      "learning_rate": 0.00019999541258374385,
-      "loss": 46.0,
-      "step": 39888
-    },
-    {
-      "epoch": 3.0497926104325552,
-      "grad_norm": 0.00306203356012702,
-      "learning_rate": 0.0001999954123536702,
-      "loss": 46.0,
-      "step": 39889
-    },
-    {
-      "epoch": 3.049869067415945,
-      "grad_norm": 0.0028150237631052732,
-      "learning_rate": 0.00019999541212359077,
-      "loss": 46.0,
-      "step": 39890
-    },
-    {
-      "epoch": 3.0499455243993348,
-      "grad_norm": 0.0035250908695161343,
-      "learning_rate": 0.00019999541189350556,
-      "loss": 46.0,
-      "step": 39891
-    },
-    {
-      "epoch": 3.0500219813827245,
-      "grad_norm": 0.003052371321246028,
-      "learning_rate": 0.0001999954116634146,
-      "loss": 46.0,
-      "step": 39892
-    },
-    {
-      "epoch": 3.0500984383661143,
-      "grad_norm": 0.004716380499303341,
-      "learning_rate": 0.00019999541143331785,
-      "loss": 46.0,
-      "step": 39893
-    },
-    {
-      "epoch": 3.050174895349504,
-      "grad_norm": 0.001816091826185584,
-      "learning_rate": 0.00019999541120321537,
-      "loss": 46.0,
-      "step": 39894
-    },
-    {
-      "epoch": 3.050251352332894,
-      "grad_norm": 0.0018924175528809428,
-      "learning_rate": 0.00019999541097310707,
-      "loss": 46.0,
-      "step": 39895
-    },
-    {
-      "epoch": 3.0503278093162836,
-      "grad_norm": 0.0010075777536258101,
-      "learning_rate": 0.00019999541074299304,
-      "loss": 46.0,
-      "step": 39896
-    },
-    {
-      "epoch": 3.0504042662996733,
-      "grad_norm": 0.0030944314785301685,
-      "learning_rate": 0.00019999541051287322,
-      "loss": 46.0,
-      "step": 39897
-    },
-    {
-      "epoch": 3.050480723283063,
-      "grad_norm": 0.002769443206489086,
-      "learning_rate": 0.00019999541028274765,
-      "loss": 46.0,
-      "step": 39898
-    },
-    {
-      "epoch": 3.0505571802664524,
-      "grad_norm": 0.0020051966421306133,
-      "learning_rate": 0.00019999541005261628,
-      "loss": 46.0,
-      "step": 39899
-    },
-    {
-      "epoch": 3.050633637249842,
-      "grad_norm": 0.002914352575317025,
-      "learning_rate": 0.00019999540982247916,
-      "loss": 46.0,
-      "step": 39900
-    },
-    {
-      "epoch": 3.050710094233232,
-      "grad_norm": 0.0020059102680534124,
-      "learning_rate": 0.0001999954095923363,
-      "loss": 46.0,
-      "step": 39901
-    },
-    {
-      "epoch": 3.0507865512166217,
-      "grad_norm": 0.0023324305657297373,
-      "learning_rate": 0.00019999540936218763,
-      "loss": 46.0,
-      "step": 39902
-    },
-    {
-      "epoch": 3.0508630082000114,
-      "grad_norm": 0.00419639190658927,
-      "learning_rate": 0.0001999954091320332,
-      "loss": 46.0,
-      "step": 39903
-    },
-    {
-      "epoch": 3.050939465183401,
-      "grad_norm": 0.0030191445257514715,
-      "learning_rate": 0.000199995408901873,
-      "loss": 46.0,
-      "step": 39904
-    },
-    {
-      "epoch": 3.051015922166791,
-      "grad_norm": 0.003896491602063179,
-      "learning_rate": 0.00019999540867170703,
-      "loss": 46.0,
-      "step": 39905
-    },
-    {
-      "epoch": 3.0510923791501807,
-      "grad_norm": 0.001012215274386108,
-      "learning_rate": 0.00019999540844153527,
-      "loss": 46.0,
-      "step": 39906
-    },
-    {
-      "epoch": 3.0511688361335705,
-      "grad_norm": 0.002515647793188691,
-      "learning_rate": 0.0001999954082113578,
-      "loss": 46.0,
-      "step": 39907
-    },
-    {
-      "epoch": 3.0512452931169602,
-      "grad_norm": 0.0024647850077599287,
-      "learning_rate": 0.00019999540798117452,
-      "loss": 46.0,
-      "step": 39908
-    },
-    {
-      "epoch": 3.05132175010035,
-      "grad_norm": 0.0010217091767117381,
-      "learning_rate": 0.0001999954077509855,
-      "loss": 46.0,
-      "step": 39909
-    },
-    {
-      "epoch": 3.0513982070837393,
-      "grad_norm": 0.003492149757221341,
-      "learning_rate": 0.00019999540752079064,
-      "loss": 46.0,
-      "step": 39910
-    },
-    {
-      "epoch": 3.051474664067129,
-      "grad_norm": 0.0009389400365762413,
-      "learning_rate": 0.00019999540729059007,
-      "loss": 46.0,
-      "step": 39911
-    },
-    {
-      "epoch": 3.051551121050519,
-      "grad_norm": 0.0006082706968300045,
-      "learning_rate": 0.00019999540706038373,
-      "loss": 46.0,
-      "step": 39912
-    },
-    {
-      "epoch": 3.0516275780339086,
-      "grad_norm": 0.004620365798473358,
-      "learning_rate": 0.00019999540683017161,
-      "loss": 46.0,
-      "step": 39913
-    },
-    {
-      "epoch": 3.0517040350172984,
-      "grad_norm": 0.002860978012904525,
-      "learning_rate": 0.00019999540659995372,
-      "loss": 46.0,
-      "step": 39914
-    },
-    {
-      "epoch": 3.051780492000688,
-      "grad_norm": 0.006351559888571501,
-      "learning_rate": 0.00019999540636973006,
-      "loss": 46.0,
-      "step": 39915
-    },
-    {
-      "epoch": 3.051856948984078,
-      "grad_norm": 0.002003585919737816,
-      "learning_rate": 0.00019999540613950062,
-      "loss": 46.0,
-      "step": 39916
-    },
-    {
-      "epoch": 3.0519334059674676,
-      "grad_norm": 0.0023747654631733894,
-      "learning_rate": 0.0001999954059092654,
-      "loss": 46.0,
-      "step": 39917
-    },
-    {
-      "epoch": 3.0520098629508574,
-      "grad_norm": 0.004748103208839893,
-      "learning_rate": 0.00019999540567902448,
-      "loss": 46.0,
-      "step": 39918
-    },
-    {
-      "epoch": 3.052086319934247,
-      "grad_norm": 0.001683561597019434,
-      "learning_rate": 0.00019999540544877773,
-      "loss": 46.0,
-      "step": 39919
-    },
-    {
-      "epoch": 3.052162776917637,
-      "grad_norm": 0.0028488747775554657,
-      "learning_rate": 0.00019999540521852525,
-      "loss": 46.0,
-      "step": 39920
-    },
-    {
-      "epoch": 3.0522392339010262,
-      "grad_norm": 0.002458098577335477,
-      "learning_rate": 0.00019999540498826695,
-      "loss": 46.0,
-      "step": 39921
-    },
-    {
-      "epoch": 3.052315690884416,
-      "grad_norm": 0.00106534524820745,
-      "learning_rate": 0.00019999540475800292,
-      "loss": 46.0,
-      "step": 39922
-    },
-    {
-      "epoch": 3.0523921478678058,
-      "grad_norm": 0.0033299436327069998,
-      "learning_rate": 0.0001999954045277331,
-      "loss": 46.0,
-      "step": 39923
-    },
-    {
-      "epoch": 3.0524686048511955,
-      "grad_norm": 0.0021596713922917843,
-      "learning_rate": 0.00019999540429745753,
-      "loss": 46.0,
-      "step": 39924
-    },
-    {
-      "epoch": 3.0525450618345853,
-      "grad_norm": 0.0012651372235268354,
-      "learning_rate": 0.0001999954040671762,
-      "loss": 46.0,
-      "step": 39925
-    },
-    {
-      "epoch": 3.052621518817975,
-      "grad_norm": 0.0013762699672952294,
-      "learning_rate": 0.00019999540383688907,
-      "loss": 46.0,
-      "step": 39926
-    },
-    {
-      "epoch": 3.052697975801365,
-      "grad_norm": 0.0068327998742461205,
-      "learning_rate": 0.00019999540360659615,
-      "loss": 46.0,
-      "step": 39927
-    },
-    {
-      "epoch": 3.0527744327847546,
-      "grad_norm": 0.001442233449779451,
-      "learning_rate": 0.0001999954033762975,
-      "loss": 46.0,
-      "step": 39928
-    },
-    {
-      "epoch": 3.0528508897681443,
-      "grad_norm": 0.0017950354376807809,
-      "learning_rate": 0.00019999540314599308,
-      "loss": 46.0,
-      "step": 39929
-    },
-    {
-      "epoch": 3.052927346751534,
-      "grad_norm": 0.0008715394651517272,
-      "learning_rate": 0.00019999540291568287,
-      "loss": 46.0,
-      "step": 39930
-    },
-    {
-      "epoch": 3.0530038037349234,
-      "grad_norm": 0.00132539973128587,
-      "learning_rate": 0.00019999540268536692,
-      "loss": 46.0,
-      "step": 39931
-    },
-    {
-      "epoch": 3.053080260718313,
-      "grad_norm": 0.001679926528595388,
-      "learning_rate": 0.00019999540245504519,
-      "loss": 46.0,
-      "step": 39932
-    },
-    {
-      "epoch": 3.053156717701703,
-      "grad_norm": 0.0035040657967329025,
-      "learning_rate": 0.00019999540222471768,
-      "loss": 46.0,
-      "step": 39933
-    },
-    {
-      "epoch": 3.0532331746850927,
-      "grad_norm": 0.0014891623286530375,
-      "learning_rate": 0.0001999954019943844,
-      "loss": 46.0,
-      "step": 39934
-    },
-    {
-      "epoch": 3.0533096316684825,
-      "grad_norm": 0.001127067138440907,
-      "learning_rate": 0.00019999540176404536,
-      "loss": 46.0,
-      "step": 39935
-    },
-    {
-      "epoch": 3.053386088651872,
-      "grad_norm": 0.002575282007455826,
-      "learning_rate": 0.00019999540153370053,
-      "loss": 46.0,
-      "step": 39936
-    },
-    {
-      "epoch": 3.053462545635262,
-      "grad_norm": 0.0008896789513528347,
-      "learning_rate": 0.00019999540130334996,
-      "loss": 46.0,
-      "step": 39937
-    },
-    {
-      "epoch": 3.0535390026186517,
-      "grad_norm": 0.012337354943156242,
-      "learning_rate": 0.00019999540107299362,
-      "loss": 46.0,
-      "step": 39938
-    },
-    {
-      "epoch": 3.0536154596020415,
-      "grad_norm": 0.0019458942115306854,
-      "learning_rate": 0.0001999954008426315,
-      "loss": 46.0,
-      "step": 39939
-    },
-    {
-      "epoch": 3.0536919165854313,
-      "grad_norm": 0.00234235473908484,
-      "learning_rate": 0.00019999540061226362,
-      "loss": 46.0,
-      "step": 39940
-    },
-    {
-      "epoch": 3.053768373568821,
-      "grad_norm": 0.0019255520310252905,
-      "learning_rate": 0.00019999540038188995,
-      "loss": 46.0,
-      "step": 39941
-    },
-    {
-      "epoch": 3.053844830552211,
-      "grad_norm": 0.004226963501423597,
-      "learning_rate": 0.00019999540015151052,
-      "loss": 46.0,
-      "step": 39942
-    },
-    {
-      "epoch": 3.0539212875356,
-      "grad_norm": 0.001408649142831564,
-      "learning_rate": 0.00019999539992112533,
-      "loss": 46.0,
-      "step": 39943
-    },
-    {
-      "epoch": 3.05399774451899,
-      "grad_norm": 0.007395122200250626,
-      "learning_rate": 0.00019999539969073435,
-      "loss": 46.0,
-      "step": 39944
-    },
-    {
-      "epoch": 3.0540742015023796,
-      "grad_norm": 0.0016999570652842522,
-      "learning_rate": 0.00019999539946033762,
-      "loss": 46.0,
-      "step": 39945
-    },
-    {
-      "epoch": 3.0541506584857694,
-      "grad_norm": 0.003070377279073,
-      "learning_rate": 0.00019999539922993512,
-      "loss": 46.0,
-      "step": 39946
-    },
-    {
-      "epoch": 3.054227115469159,
-      "grad_norm": 0.000984929851256311,
-      "learning_rate": 0.00019999539899952687,
-      "loss": 46.0,
-      "step": 39947
-    },
-    {
-      "epoch": 3.054303572452549,
-      "grad_norm": 0.0010904729133471847,
-      "learning_rate": 0.00019999539876911282,
-      "loss": 46.0,
-      "step": 39948
-    },
-    {
-      "epoch": 3.0543800294359387,
-      "grad_norm": 0.0027253134176135063,
-      "learning_rate": 0.000199995398538693,
-      "loss": 46.0,
-      "step": 39949
-    },
-    {
-      "epoch": 3.0544564864193284,
-      "grad_norm": 0.0009720441303215921,
-      "learning_rate": 0.0001999953983082674,
-      "loss": 46.0,
-      "step": 39950
-    },
-    {
-      "epoch": 3.054532943402718,
-      "grad_norm": 0.0012783625861629844,
-      "learning_rate": 0.00019999539807783606,
-      "loss": 46.0,
-      "step": 39951
-    },
-    {
-      "epoch": 3.054609400386108,
-      "grad_norm": 0.002023769076913595,
-      "learning_rate": 0.00019999539784739894,
-      "loss": 46.0,
-      "step": 39952
-    },
-    {
-      "epoch": 3.0546858573694973,
-      "grad_norm": 0.002419615164399147,
-      "learning_rate": 0.00019999539761695608,
-      "loss": 46.0,
-      "step": 39953
-    },
-    {
-      "epoch": 3.054762314352887,
-      "grad_norm": 0.0015678760828450322,
-      "learning_rate": 0.00019999539738650742,
-      "loss": 46.0,
-      "step": 39954
-    },
-    {
-      "epoch": 3.054838771336277,
-      "grad_norm": 0.0031746365129947662,
-      "learning_rate": 0.000199995397156053,
-      "loss": 46.0,
-      "step": 39955
-    },
-    {
-      "epoch": 3.0549152283196666,
-      "grad_norm": 0.001625155913643539,
-      "learning_rate": 0.0001999953969255928,
-      "loss": 46.0,
-      "step": 39956
-    },
-    {
-      "epoch": 3.0549916853030563,
-      "grad_norm": 0.00221212743781507,
-      "learning_rate": 0.00019999539669512682,
-      "loss": 46.0,
-      "step": 39957
-    },
-    {
-      "epoch": 3.055068142286446,
-      "grad_norm": 0.01253463700413704,
-      "learning_rate": 0.0001999953964646551,
-      "loss": 46.0,
-      "step": 39958
-    },
-    {
-      "epoch": 3.055144599269836,
-      "grad_norm": 0.001577693154104054,
-      "learning_rate": 0.0001999953962341776,
-      "loss": 46.0,
-      "step": 39959
-    },
-    {
-      "epoch": 3.0552210562532256,
-      "grad_norm": 0.0026553766801953316,
-      "learning_rate": 0.0001999953960036943,
-      "loss": 46.0,
-      "step": 39960
-    },
-    {
-      "epoch": 3.0552975132366154,
-      "grad_norm": 0.0010875161970034242,
-      "learning_rate": 0.00019999539577320526,
-      "loss": 46.0,
-      "step": 39961
-    },
-    {
-      "epoch": 3.055373970220005,
-      "grad_norm": 0.0015368919121101499,
-      "learning_rate": 0.00019999539554271047,
-      "loss": 46.0,
-      "step": 39962
-    },
-    {
-      "epoch": 3.055450427203395,
-      "grad_norm": 0.0017528730677440763,
-      "learning_rate": 0.00019999539531220987,
-      "loss": 46.0,
-      "step": 39963
-    },
-    {
-      "epoch": 3.055526884186784,
-      "grad_norm": 0.0012605271767824888,
-      "learning_rate": 0.00019999539508170356,
-      "loss": 46.0,
-      "step": 39964
-    },
-    {
-      "epoch": 3.055603341170174,
-      "grad_norm": 0.00077737623360008,
-      "learning_rate": 0.00019999539485119141,
-      "loss": 46.0,
-      "step": 39965
-    },
-    {
-      "epoch": 3.0556797981535637,
-      "grad_norm": 0.0008446162100881338,
-      "learning_rate": 0.00019999539462067353,
-      "loss": 46.0,
-      "step": 39966
-    },
-    {
-      "epoch": 3.0557562551369535,
-      "grad_norm": 0.0023541422560811043,
-      "learning_rate": 0.0001999953943901499,
-      "loss": 46.0,
-      "step": 39967
-    },
-    {
-      "epoch": 3.0558327121203432,
-      "grad_norm": 0.0014828427229076624,
-      "learning_rate": 0.00019999539415962046,
-      "loss": 46.0,
-      "step": 39968
-    },
-    {
-      "epoch": 3.055909169103733,
-      "grad_norm": 0.0007023270009085536,
-      "learning_rate": 0.00019999539392908527,
-      "loss": 46.0,
-      "step": 39969
-    },
-    {
-      "epoch": 3.0559856260871228,
-      "grad_norm": 0.0011262464104220271,
-      "learning_rate": 0.0001999953936985443,
-      "loss": 46.0,
-      "step": 39970
-    },
-    {
-      "epoch": 3.0560620830705125,
-      "grad_norm": 0.0044969129376113415,
-      "learning_rate": 0.00019999539346799756,
-      "loss": 46.0,
-      "step": 39971
-    },
-    {
-      "epoch": 3.0561385400539023,
-      "grad_norm": 0.0014838998904451728,
-      "learning_rate": 0.00019999539323744506,
-      "loss": 46.0,
-      "step": 39972
-    },
-    {
-      "epoch": 3.056214997037292,
-      "grad_norm": 0.0024705128744244576,
-      "learning_rate": 0.00019999539300688678,
-      "loss": 46.0,
-      "step": 39973
-    },
-    {
-      "epoch": 3.056291454020682,
-      "grad_norm": 0.0029817065224051476,
-      "learning_rate": 0.00019999539277632274,
-      "loss": 46.0,
-      "step": 39974
-    },
-    {
-      "epoch": 3.056367911004071,
-      "grad_norm": 0.0027656417805701494,
-      "learning_rate": 0.00019999539254575294,
-      "loss": 46.0,
-      "step": 39975
-    },
-    {
-      "epoch": 3.056444367987461,
-      "grad_norm": 0.0027822358533740044,
-      "learning_rate": 0.00019999539231517735,
-      "loss": 46.0,
-      "step": 39976
-    },
-    {
-      "epoch": 3.0565208249708506,
-      "grad_norm": 0.0020388104021549225,
-      "learning_rate": 0.000199995392084596,
-      "loss": 46.0,
-      "step": 39977
-    },
-    {
-      "epoch": 3.0565972819542404,
-      "grad_norm": 0.005180031061172485,
-      "learning_rate": 0.00019999539185400892,
-      "loss": 46.0,
-      "step": 39978
-    },
-    {
-      "epoch": 3.05667373893763,
-      "grad_norm": 0.0030655774753540754,
-      "learning_rate": 0.000199995391623416,
-      "loss": 46.0,
-      "step": 39979
-    },
-    {
-      "epoch": 3.05675019592102,
-      "grad_norm": 0.001487218658439815,
-      "learning_rate": 0.00019999539139281734,
-      "loss": 46.0,
-      "step": 39980
-    },
-    {
-      "epoch": 3.0568266529044097,
-      "grad_norm": 0.0022295925300568342,
-      "learning_rate": 0.0001999953911622129,
-      "loss": 46.0,
-      "step": 39981
-    },
-    {
-      "epoch": 3.0569031098877995,
-      "grad_norm": 0.0013577919453382492,
-      "learning_rate": 0.00019999539093160272,
-      "loss": 46.0,
-      "step": 39982
-    },
-    {
-      "epoch": 3.056979566871189,
-      "grad_norm": 0.0019416635623201728,
-      "learning_rate": 0.00019999539070098677,
-      "loss": 46.0,
-      "step": 39983
-    },
-    {
-      "epoch": 3.057056023854579,
-      "grad_norm": 0.002988230437040329,
-      "learning_rate": 0.00019999539047036504,
-      "loss": 46.0,
-      "step": 39984
-    },
-    {
-      "epoch": 3.0571324808379687,
-      "grad_norm": 0.00440600560978055,
-      "learning_rate": 0.00019999539023973754,
-      "loss": 46.0,
-      "step": 39985
-    },
-    {
-      "epoch": 3.057208937821358,
-      "grad_norm": 0.0012431966606527567,
-      "learning_rate": 0.0001999953900091043,
-      "loss": 46.0,
-      "step": 39986
-    },
-    {
-      "epoch": 3.057285394804748,
-      "grad_norm": 0.0021161320619285107,
-      "learning_rate": 0.0001999953897784652,
-      "loss": 46.0,
-      "step": 39987
-    },
-    {
-      "epoch": 3.0573618517881376,
-      "grad_norm": 0.0011795427417382598,
-      "learning_rate": 0.00019999538954782042,
-      "loss": 46.0,
-      "step": 39988
-    },
-    {
-      "epoch": 3.0574383087715273,
-      "grad_norm": 0.0019480413757264614,
-      "learning_rate": 0.00019999538931716982,
-      "loss": 46.0,
-      "step": 39989
-    },
-    {
-      "epoch": 3.057514765754917,
-      "grad_norm": 0.001120949280448258,
-      "learning_rate": 0.00019999538908651348,
-      "loss": 46.0,
-      "step": 39990
-    },
-    {
-      "epoch": 3.057591222738307,
-      "grad_norm": 0.0014087992021813989,
-      "learning_rate": 0.00019999538885585137,
-      "loss": 46.0,
-      "step": 39991
-    },
-    {
-      "epoch": 3.0576676797216966,
-      "grad_norm": 0.003980633802711964,
-      "learning_rate": 0.00019999538862518348,
-      "loss": 46.0,
-      "step": 39992
-    },
-    {
-      "epoch": 3.0577441367050864,
-      "grad_norm": 0.0022805230692029,
-      "learning_rate": 0.00019999538839450982,
-      "loss": 46.0,
-      "step": 39993
-    },
-    {
-      "epoch": 3.057820593688476,
-      "grad_norm": 0.002794375643134117,
-      "learning_rate": 0.00019999538816383038,
-      "loss": 46.0,
-      "step": 39994
-    },
-    {
-      "epoch": 3.057897050671866,
-      "grad_norm": 0.0013495429884642363,
-      "learning_rate": 0.0001999953879331452,
-      "loss": 46.0,
-      "step": 39995
-    },
-    {
-      "epoch": 3.0579735076552557,
-      "grad_norm": 0.002690873108804226,
-      "learning_rate": 0.00019999538770245425,
-      "loss": 46.0,
-      "step": 39996
-    },
-    {
-      "epoch": 3.058049964638645,
-      "grad_norm": 0.000877495389431715,
-      "learning_rate": 0.0001999953874717575,
-      "loss": 46.0,
-      "step": 39997
-    },
-    {
-      "epoch": 3.0581264216220347,
-      "grad_norm": 0.0020809336565434933,
-      "learning_rate": 0.00019999538724105502,
-      "loss": 46.0,
-      "step": 39998
-    },
-    {
-      "epoch": 3.0582028786054245,
-      "grad_norm": 0.0022771693766117096,
-      "learning_rate": 0.00019999538701034674,
-      "loss": 46.0,
-      "step": 39999
-    },
-    {
-      "epoch": 3.0582793355888143,
-      "grad_norm": 0.0040611089207232,
-      "learning_rate": 0.0001999953867796327,
-      "loss": 46.0,
-      "step": 40000
-    },
-    {
-      "epoch": 3.058355792572204,
-      "grad_norm": 0.002339953323826194,
-      "learning_rate": 0.0001999953865489129,
-      "loss": 46.0,
-      "step": 40001
-    },
-    {
-      "epoch": 3.058432249555594,
-      "grad_norm": 0.00136351038236171,
-      "learning_rate": 0.00019999538631818733,
-      "loss": 46.0,
-      "step": 40002
-    },
-    {
-      "epoch": 3.0585087065389835,
-      "grad_norm": 0.0019463424105197191,
-      "learning_rate": 0.000199995386087456,
-      "loss": 46.0,
-      "step": 40003
-    },
-    {
-      "epoch": 3.0585851635223733,
-      "grad_norm": 0.0007321077864617109,
-      "learning_rate": 0.00019999538585671888,
-      "loss": 46.0,
-      "step": 40004
-    },
-    {
-      "epoch": 3.058661620505763,
-      "grad_norm": 0.0027424537111073732,
-      "learning_rate": 0.00019999538562597596,
-      "loss": 46.0,
-      "step": 40005
-    },
-    {
-      "epoch": 3.058738077489153,
-      "grad_norm": 0.004692576825618744,
-      "learning_rate": 0.0001999953853952273,
-      "loss": 46.0,
-      "step": 40006
-    },
-    {
-      "epoch": 3.0588145344725426,
-      "grad_norm": 0.0038249201606959105,
-      "learning_rate": 0.0001999953851644729,
-      "loss": 46.0,
-      "step": 40007
-    },
-    {
-      "epoch": 3.058890991455932,
-      "grad_norm": 0.0010336223058402538,
-      "learning_rate": 0.0001999953849337127,
-      "loss": 46.0,
-      "step": 40008
-    },
-    {
-      "epoch": 3.0589674484393217,
-      "grad_norm": 0.0016876417212188244,
-      "learning_rate": 0.00019999538470294673,
-      "loss": 46.0,
-      "step": 40009
-    },
-    {
-      "epoch": 3.0590439054227114,
-      "grad_norm": 0.000810078636277467,
-      "learning_rate": 0.000199995384472175,
-      "loss": 46.0,
-      "step": 40010
-    },
-    {
-      "epoch": 3.059120362406101,
-      "grad_norm": 0.002382120117545128,
-      "learning_rate": 0.0001999953842413975,
-      "loss": 46.0,
-      "step": 40011
-    },
-    {
-      "epoch": 3.059196819389491,
-      "grad_norm": 0.0017555918311700225,
-      "learning_rate": 0.00019999538401061423,
-      "loss": 46.0,
-      "step": 40012
-    },
-    {
-      "epoch": 3.0592732763728807,
-      "grad_norm": 0.00617721164599061,
-      "learning_rate": 0.00019999538377982518,
-      "loss": 46.0,
-      "step": 40013
-    },
-    {
-      "epoch": 3.0593497333562705,
-      "grad_norm": 0.0021797812078148127,
-      "learning_rate": 0.00019999538354903038,
-      "loss": 46.0,
-      "step": 40014
-    },
-    {
-      "epoch": 3.0594261903396602,
-      "grad_norm": 0.0028860955499112606,
-      "learning_rate": 0.0001999953833182298,
-      "loss": 46.0,
-      "step": 40015
-    },
-    {
-      "epoch": 3.05950264732305,
-      "grad_norm": 0.0033860933035612106,
-      "learning_rate": 0.00019999538308742345,
-      "loss": 46.0,
-      "step": 40016
-    },
-    {
-      "epoch": 3.0595791043064398,
-      "grad_norm": 0.00279794679954648,
-      "learning_rate": 0.00019999538285661134,
-      "loss": 46.0,
-      "step": 40017
-    },
-    {
-      "epoch": 3.0596555612898295,
-      "grad_norm": 0.0007723582093603909,
-      "learning_rate": 0.00019999538262579347,
-      "loss": 46.0,
-      "step": 40018
-    },
-    {
-      "epoch": 3.059732018273219,
-      "grad_norm": 0.0006307529401965439,
-      "learning_rate": 0.0001999953823949698,
-      "loss": 46.0,
-      "step": 40019
-    },
-    {
-      "epoch": 3.0598084752566086,
-      "grad_norm": 0.001034247106872499,
-      "learning_rate": 0.0001999953821641404,
-      "loss": 46.0,
-      "step": 40020
-    },
-    {
-      "epoch": 3.0598849322399984,
-      "grad_norm": 0.0015814361395314336,
-      "learning_rate": 0.00019999538193330517,
-      "loss": 46.0,
-      "step": 40021
-    },
-    {
-      "epoch": 3.059961389223388,
-      "grad_norm": 0.006440328434109688,
-      "learning_rate": 0.00019999538170246424,
-      "loss": 46.0,
-      "step": 40022
-    },
-    {
-      "epoch": 3.060037846206778,
-      "grad_norm": 0.0016617951914668083,
-      "learning_rate": 0.00019999538147161752,
-      "loss": 46.0,
-      "step": 40023
-    },
-    {
-      "epoch": 3.0601143031901676,
-      "grad_norm": 0.004704756662249565,
-      "learning_rate": 0.000199995381240765,
-      "loss": 46.0,
-      "step": 40024
-    },
-    {
-      "epoch": 3.0601907601735574,
-      "grad_norm": 0.0022259466350078583,
-      "learning_rate": 0.00019999538100990674,
-      "loss": 46.0,
-      "step": 40025
-    },
-    {
-      "epoch": 3.060267217156947,
-      "grad_norm": 0.0014169689966365695,
-      "learning_rate": 0.0001999953807790427,
-      "loss": 46.0,
-      "step": 40026
-    },
-    {
-      "epoch": 3.060343674140337,
-      "grad_norm": 0.0014136621030047536,
-      "learning_rate": 0.00019999538054817287,
-      "loss": 46.0,
-      "step": 40027
-    },
-    {
-      "epoch": 3.0604201311237267,
-      "grad_norm": 0.0006793364300392568,
-      "learning_rate": 0.00019999538031729733,
-      "loss": 46.0,
-      "step": 40028
-    },
-    {
-      "epoch": 3.0604965881071164,
-      "grad_norm": 0.001826431485824287,
-      "learning_rate": 0.00019999538008641597,
-      "loss": 46.0,
-      "step": 40029
-    },
-    {
-      "epoch": 3.0605730450905058,
-      "grad_norm": 0.010063394904136658,
-      "learning_rate": 0.00019999537985552883,
-      "loss": 46.0,
-      "step": 40030
-    },
-    {
-      "epoch": 3.0606495020738955,
-      "grad_norm": 0.0012650748249143362,
-      "learning_rate": 0.00019999537962463597,
-      "loss": 46.0,
-      "step": 40031
-    },
-    {
-      "epoch": 3.0607259590572853,
-      "grad_norm": 0.002444294048473239,
-      "learning_rate": 0.0001999953793937373,
-      "loss": 46.0,
-      "step": 40032
-    },
-    {
-      "epoch": 3.060802416040675,
-      "grad_norm": 0.005888687912374735,
-      "learning_rate": 0.0001999953791628329,
-      "loss": 46.0,
-      "step": 40033
-    },
-    {
-      "epoch": 3.060878873024065,
-      "grad_norm": 0.0026312689296901226,
-      "learning_rate": 0.00019999537893192272,
-      "loss": 46.0,
-      "step": 40034
-    },
-    {
-      "epoch": 3.0609553300074546,
-      "grad_norm": 0.0026897434145212173,
-      "learning_rate": 0.00019999537870100674,
-      "loss": 46.0,
-      "step": 40035
-    },
-    {
-      "epoch": 3.0610317869908443,
-      "grad_norm": 0.002142061712220311,
-      "learning_rate": 0.000199995378470085,
-      "loss": 46.0,
-      "step": 40036
-    },
-    {
-      "epoch": 3.061108243974234,
-      "grad_norm": 0.00276556727476418,
-      "learning_rate": 0.0001999953782391575,
-      "loss": 46.0,
-      "step": 40037
-    },
-    {
-      "epoch": 3.061184700957624,
-      "grad_norm": 0.0049687656573951244,
-      "learning_rate": 0.00019999537800822424,
-      "loss": 46.0,
-      "step": 40038
-    },
-    {
-      "epoch": 3.0612611579410136,
-      "grad_norm": 0.0012892611557617784,
-      "learning_rate": 0.00019999537777728522,
-      "loss": 46.0,
-      "step": 40039
-    },
-    {
-      "epoch": 3.0613376149244034,
-      "grad_norm": 0.0027382851112633944,
-      "learning_rate": 0.0001999953775463404,
-      "loss": 46.0,
-      "step": 40040
-    },
-    {
-      "epoch": 3.0614140719077927,
-      "grad_norm": 0.004092883318662643,
-      "learning_rate": 0.0001999953773153898,
-      "loss": 46.0,
-      "step": 40041
-    },
-    {
-      "epoch": 3.0614905288911824,
-      "grad_norm": 0.0012523240875452757,
-      "learning_rate": 0.0001999953770844335,
-      "loss": 46.0,
-      "step": 40042
-    },
-    {
-      "epoch": 3.061566985874572,
-      "grad_norm": 0.001444007270038128,
-      "learning_rate": 0.00019999537685347135,
-      "loss": 46.0,
-      "step": 40043
-    },
-    {
-      "epoch": 3.061643442857962,
-      "grad_norm": 0.0008491792832501233,
-      "learning_rate": 0.00019999537662250346,
-      "loss": 46.0,
-      "step": 40044
-    },
-    {
-      "epoch": 3.0617198998413517,
-      "grad_norm": 0.0012101087486371398,
-      "learning_rate": 0.00019999537639152983,
-      "loss": 46.0,
-      "step": 40045
-    },
-    {
-      "epoch": 3.0617963568247415,
-      "grad_norm": 0.0025914970319718122,
-      "learning_rate": 0.0001999953761605504,
-      "loss": 46.0,
-      "step": 40046
-    },
-    {
-      "epoch": 3.0618728138081313,
-      "grad_norm": 0.003165853675454855,
-      "learning_rate": 0.00019999537592956519,
-      "loss": 46.0,
-      "step": 40047
-    },
-    {
-      "epoch": 3.061949270791521,
-      "grad_norm": 0.0025450438261032104,
-      "learning_rate": 0.00019999537569857423,
-      "loss": 46.0,
-      "step": 40048
-    },
-    {
-      "epoch": 3.0620257277749108,
-      "grad_norm": 0.002609213814139366,
-      "learning_rate": 0.0001999953754675775,
-      "loss": 46.0,
-      "step": 40049
-    },
-    {
-      "epoch": 3.0621021847583005,
-      "grad_norm": 0.0018422119319438934,
-      "learning_rate": 0.000199995375236575,
-      "loss": 46.0,
-      "step": 40050
-    },
-    {
-      "epoch": 3.0621786417416903,
-      "grad_norm": 0.0020399370696395636,
-      "learning_rate": 0.00019999537500556676,
-      "loss": 46.0,
-      "step": 40051
-    },
-    {
-      "epoch": 3.0622550987250796,
-      "grad_norm": 0.0019451486878097057,
-      "learning_rate": 0.0001999953747745527,
-      "loss": 46.0,
-      "step": 40052
-    },
-    {
-      "epoch": 3.0623315557084694,
-      "grad_norm": 0.00195057422388345,
-      "learning_rate": 0.0001999953745435329,
-      "loss": 46.0,
-      "step": 40053
-    },
-    {
-      "epoch": 3.062408012691859,
-      "grad_norm": 0.0020503527484834194,
-      "learning_rate": 0.00019999537431250733,
-      "loss": 46.0,
-      "step": 40054
-    },
-    {
-      "epoch": 3.062484469675249,
-      "grad_norm": 0.0005428163567557931,
-      "learning_rate": 0.00019999537408147599,
-      "loss": 46.0,
-      "step": 40055
-    },
-    {
-      "epoch": 3.0625609266586387,
-      "grad_norm": 0.0028380868025124073,
-      "learning_rate": 0.00019999537385043887,
-      "loss": 46.0,
-      "step": 40056
-    },
-    {
-      "epoch": 3.0626373836420284,
-      "grad_norm": 0.002967200009152293,
-      "learning_rate": 0.00019999537361939599,
-      "loss": 46.0,
-      "step": 40057
-    },
-    {
-      "epoch": 3.062713840625418,
-      "grad_norm": 0.002673485316336155,
-      "learning_rate": 0.00019999537338834733,
-      "loss": 46.0,
-      "step": 40058
-    },
-    {
-      "epoch": 3.062790297608808,
-      "grad_norm": 0.0050707170739769936,
-      "learning_rate": 0.00019999537315729292,
-      "loss": 46.0,
-      "step": 40059
-    },
-    {
-      "epoch": 3.0628667545921977,
-      "grad_norm": 0.002262258669361472,
-      "learning_rate": 0.0001999953729262327,
-      "loss": 46.0,
-      "step": 40060
-    },
-    {
-      "epoch": 3.0629432115755875,
-      "grad_norm": 0.0028284159488976,
-      "learning_rate": 0.00019999537269516676,
-      "loss": 46.0,
-      "step": 40061
-    },
-    {
-      "epoch": 3.063019668558977,
-      "grad_norm": 0.001756039448082447,
-      "learning_rate": 0.00019999537246409503,
-      "loss": 46.0,
-      "step": 40062
-    },
-    {
-      "epoch": 3.0630961255423665,
-      "grad_norm": 0.0012712596217170358,
-      "learning_rate": 0.00019999537223301753,
-      "loss": 46.0,
-      "step": 40063
-    },
-    {
-      "epoch": 3.0631725825257563,
-      "grad_norm": 0.0033195726573467255,
-      "learning_rate": 0.00019999537200193426,
-      "loss": 46.0,
-      "step": 40064
-    },
-    {
-      "epoch": 3.063249039509146,
-      "grad_norm": 0.0031835127156227827,
-      "learning_rate": 0.00019999537177084524,
-      "loss": 46.0,
-      "step": 40065
-    },
-    {
-      "epoch": 3.063325496492536,
-      "grad_norm": 0.00181575829628855,
-      "learning_rate": 0.00019999537153975042,
-      "loss": 46.0,
-      "step": 40066
-    },
-    {
-      "epoch": 3.0634019534759256,
-      "grad_norm": 0.000695188413374126,
-      "learning_rate": 0.00019999537130864985,
-      "loss": 46.0,
-      "step": 40067
-    },
-    {
-      "epoch": 3.0634784104593153,
-      "grad_norm": 0.0021538808941841125,
-      "learning_rate": 0.00019999537107754352,
-      "loss": 46.0,
-      "step": 40068
-    },
-    {
-      "epoch": 3.063554867442705,
-      "grad_norm": 0.000692805799189955,
-      "learning_rate": 0.0001999953708464314,
-      "loss": 46.0,
-      "step": 40069
-    },
-    {
-      "epoch": 3.063631324426095,
-      "grad_norm": 0.006684989668428898,
-      "learning_rate": 0.00019999537061531352,
-      "loss": 46.0,
-      "step": 40070
-    },
-    {
-      "epoch": 3.0637077814094846,
-      "grad_norm": 0.0011334270238876343,
-      "learning_rate": 0.00019999537038418986,
-      "loss": 46.0,
-      "step": 40071
-    },
-    {
-      "epoch": 3.0637842383928744,
-      "grad_norm": 0.0024604403879493475,
-      "learning_rate": 0.00019999537015306045,
-      "loss": 46.0,
-      "step": 40072
-    },
-    {
-      "epoch": 3.063860695376264,
-      "grad_norm": 0.0010915531311184168,
-      "learning_rate": 0.00019999536992192524,
-      "loss": 46.0,
-      "step": 40073
-    },
-    {
-      "epoch": 3.0639371523596535,
-      "grad_norm": 0.00166521652135998,
-      "learning_rate": 0.00019999536969078426,
-      "loss": 46.0,
-      "step": 40074
-    },
-    {
-      "epoch": 3.0640136093430432,
-      "grad_norm": 0.005697852000594139,
-      "learning_rate": 0.00019999536945963756,
-      "loss": 46.0,
-      "step": 40075
-    },
-    {
-      "epoch": 3.064090066326433,
-      "grad_norm": 0.0012899364810436964,
-      "learning_rate": 0.00019999536922848507,
-      "loss": 46.0,
-      "step": 40076
-    },
-    {
-      "epoch": 3.0641665233098228,
-      "grad_norm": 0.002538541564717889,
-      "learning_rate": 0.0001999953689973268,
-      "loss": 46.0,
-      "step": 40077
-    },
-    {
-      "epoch": 3.0642429802932125,
-      "grad_norm": 0.001561658107675612,
-      "learning_rate": 0.00019999536876616277,
-      "loss": 46.0,
-      "step": 40078
-    },
-    {
-      "epoch": 3.0643194372766023,
-      "grad_norm": 0.006923548877239227,
-      "learning_rate": 0.00019999536853499295,
-      "loss": 46.0,
-      "step": 40079
-    },
-    {
-      "epoch": 3.064395894259992,
-      "grad_norm": 0.0018838904798030853,
-      "learning_rate": 0.0001999953683038174,
-      "loss": 46.0,
-      "step": 40080
-    },
-    {
-      "epoch": 3.064472351243382,
-      "grad_norm": 0.002636120654642582,
-      "learning_rate": 0.00019999536807263605,
-      "loss": 46.0,
-      "step": 40081
-    },
-    {
-      "epoch": 3.0645488082267716,
-      "grad_norm": 0.0010320631554350257,
-      "learning_rate": 0.0001999953678414489,
-      "loss": 46.0,
-      "step": 40082
-    },
-    {
-      "epoch": 3.0646252652101613,
-      "grad_norm": 0.003608887316659093,
-      "learning_rate": 0.00019999536761025602,
-      "loss": 46.0,
-      "step": 40083
-    },
-    {
-      "epoch": 3.0647017221935506,
-      "grad_norm": 0.00491699343547225,
-      "learning_rate": 0.00019999536737905736,
-      "loss": 46.0,
-      "step": 40084
-    },
-    {
-      "epoch": 3.0647781791769404,
-      "grad_norm": 0.0027365938294678926,
-      "learning_rate": 0.00019999536714785296,
-      "loss": 46.0,
-      "step": 40085
-    },
-    {
-      "epoch": 3.06485463616033,
-      "grad_norm": 0.0014830800937488675,
-      "learning_rate": 0.00019999536691664278,
-      "loss": 46.0,
-      "step": 40086
-    },
-    {
-      "epoch": 3.06493109314372,
-      "grad_norm": 0.0034375060349702835,
-      "learning_rate": 0.00019999536668542683,
-      "loss": 46.0,
-      "step": 40087
-    },
-    {
-      "epoch": 3.0650075501271097,
-      "grad_norm": 0.0035186083987355232,
-      "learning_rate": 0.0001999953664542051,
-      "loss": 46.0,
-      "step": 40088
-    },
-    {
-      "epoch": 3.0650840071104994,
-      "grad_norm": 0.0017330920090898871,
-      "learning_rate": 0.00019999536622297757,
-      "loss": 46.0,
-      "step": 40089
-    },
-    {
-      "epoch": 3.065160464093889,
-      "grad_norm": 0.0023464756086468697,
-      "learning_rate": 0.0001999953659917443,
-      "loss": 46.0,
-      "step": 40090
-    },
-    {
-      "epoch": 3.065236921077279,
-      "grad_norm": 0.002111198613420129,
-      "learning_rate": 0.00019999536576050528,
-      "loss": 46.0,
-      "step": 40091
-    },
-    {
-      "epoch": 3.0653133780606687,
-      "grad_norm": 0.002457548165693879,
-      "learning_rate": 0.0001999953655292605,
-      "loss": 46.0,
-      "step": 40092
-    },
-    {
-      "epoch": 3.0653898350440585,
-      "grad_norm": 0.002165521727874875,
-      "learning_rate": 0.0001999953652980099,
-      "loss": 46.0,
-      "step": 40093
-    },
-    {
-      "epoch": 3.0654662920274482,
-      "grad_norm": 0.0007987722638063133,
-      "learning_rate": 0.00019999536506675356,
-      "loss": 46.0,
-      "step": 40094
-    },
-    {
-      "epoch": 3.0655427490108376,
-      "grad_norm": 0.0018417808460071683,
-      "learning_rate": 0.00019999536483549147,
-      "loss": 46.0,
-      "step": 40095
-    },
-    {
-      "epoch": 3.0656192059942273,
-      "grad_norm": 0.0023169689811766148,
-      "learning_rate": 0.0001999953646042236,
-      "loss": 46.0,
-      "step": 40096
-    },
-    {
-      "epoch": 3.065695662977617,
-      "grad_norm": 0.0015528496587648988,
-      "learning_rate": 0.00019999536437294993,
-      "loss": 46.0,
-      "step": 40097
-    },
-    {
-      "epoch": 3.065772119961007,
-      "grad_norm": 0.0013787455391138792,
-      "learning_rate": 0.0001999953641416705,
-      "loss": 46.0,
-      "step": 40098
-    },
-    {
-      "epoch": 3.0658485769443966,
-      "grad_norm": 0.002037461381405592,
-      "learning_rate": 0.00019999536391038532,
-      "loss": 46.0,
-      "step": 40099
-    },
-    {
-      "epoch": 3.0659250339277864,
-      "grad_norm": 0.002598753198981285,
-      "learning_rate": 0.00019999536367909437,
-      "loss": 46.0,
-      "step": 40100
-    },
-    {
-      "epoch": 3.066001490911176,
-      "grad_norm": 0.001525825122371316,
-      "learning_rate": 0.00019999536344779764,
-      "loss": 46.0,
-      "step": 40101
-    },
-    {
-      "epoch": 3.066077947894566,
-      "grad_norm": 0.0033435372170060873,
-      "learning_rate": 0.00019999536321649511,
-      "loss": 46.0,
-      "step": 40102
-    },
-    {
-      "epoch": 3.0661544048779557,
-      "grad_norm": 0.002753153210505843,
-      "learning_rate": 0.00019999536298518687,
-      "loss": 46.0,
-      "step": 40103
-    },
-    {
-      "epoch": 3.0662308618613454,
-      "grad_norm": 0.0018918882124125957,
-      "learning_rate": 0.00019999536275387282,
-      "loss": 46.0,
-      "step": 40104
-    },
-    {
-      "epoch": 3.066307318844735,
-      "grad_norm": 0.0009196771425195038,
-      "learning_rate": 0.00019999536252255303,
-      "loss": 46.0,
-      "step": 40105
-    },
-    {
-      "epoch": 3.0663837758281245,
-      "grad_norm": 0.006751567590981722,
-      "learning_rate": 0.00019999536229122744,
-      "loss": 46.0,
-      "step": 40106
-    },
-    {
-      "epoch": 3.0664602328115143,
-      "grad_norm": 0.0015411663334816694,
-      "learning_rate": 0.0001999953620598961,
-      "loss": 46.0,
-      "step": 40107
-    },
-    {
-      "epoch": 3.066536689794904,
-      "grad_norm": 0.004163368605077267,
-      "learning_rate": 0.00019999536182855902,
-      "loss": 46.0,
-      "step": 40108
-    },
-    {
-      "epoch": 3.0666131467782938,
-      "grad_norm": 0.0010396692669019103,
-      "learning_rate": 0.00019999536159721613,
-      "loss": 46.0,
-      "step": 40109
-    },
-    {
-      "epoch": 3.0666896037616835,
-      "grad_norm": 0.0031032890547066927,
-      "learning_rate": 0.00019999536136586747,
-      "loss": 46.0,
-      "step": 40110
-    },
-    {
-      "epoch": 3.0667660607450733,
-      "grad_norm": 0.0028323382139205933,
-      "learning_rate": 0.00019999536113451304,
-      "loss": 46.0,
-      "step": 40111
-    },
-    {
-      "epoch": 3.066842517728463,
-      "grad_norm": 0.0006947419024072587,
-      "learning_rate": 0.00019999536090315286,
-      "loss": 46.0,
-      "step": 40112
-    },
-    {
-      "epoch": 3.066918974711853,
-      "grad_norm": 0.003625748213380575,
-      "learning_rate": 0.0001999953606717869,
-      "loss": 46.0,
-      "step": 40113
-    },
-    {
-      "epoch": 3.0669954316952426,
-      "grad_norm": 0.0006266285199671984,
-      "learning_rate": 0.00019999536044041518,
-      "loss": 46.0,
-      "step": 40114
-    },
-    {
-      "epoch": 3.0670718886786323,
-      "grad_norm": 0.0009936565766111016,
-      "learning_rate": 0.00019999536020903769,
-      "loss": 46.0,
-      "step": 40115
-    },
-    {
-      "epoch": 3.067148345662022,
-      "grad_norm": 0.00374209089204669,
-      "learning_rate": 0.00019999535997765441,
-      "loss": 46.0,
-      "step": 40116
-    },
-    {
-      "epoch": 3.0672248026454114,
-      "grad_norm": 0.005744465626776218,
-      "learning_rate": 0.00019999535974626537,
-      "loss": 46.0,
-      "step": 40117
-    },
-    {
-      "epoch": 3.067301259628801,
-      "grad_norm": 0.00117302720900625,
-      "learning_rate": 0.00019999535951487055,
-      "loss": 46.0,
-      "step": 40118
-    },
-    {
-      "epoch": 3.067377716612191,
-      "grad_norm": 0.0033250923734158278,
-      "learning_rate": 0.00019999535928347,
-      "loss": 46.0,
-      "step": 40119
-    },
-    {
-      "epoch": 3.0674541735955807,
-      "grad_norm": 0.004042233340442181,
-      "learning_rate": 0.00019999535905206365,
-      "loss": 46.0,
-      "step": 40120
-    },
-    {
-      "epoch": 3.0675306305789705,
-      "grad_norm": 0.001535071642138064,
-      "learning_rate": 0.00019999535882065156,
-      "loss": 46.0,
-      "step": 40121
-    },
-    {
-      "epoch": 3.0676070875623602,
-      "grad_norm": 0.0009647165425121784,
-      "learning_rate": 0.00019999535858923365,
-      "loss": 46.0,
-      "step": 40122
-    },
-    {
-      "epoch": 3.06768354454575,
-      "grad_norm": 0.0016198258381336927,
-      "learning_rate": 0.00019999535835781002,
-      "loss": 46.0,
-      "step": 40123
-    },
-    {
-      "epoch": 3.0677600015291397,
-      "grad_norm": 0.002851195400580764,
-      "learning_rate": 0.00019999535812638061,
-      "loss": 46.0,
-      "step": 40124
-    },
-    {
-      "epoch": 3.0678364585125295,
-      "grad_norm": 0.0016150474548339844,
-      "learning_rate": 0.0001999953578949454,
-      "loss": 46.0,
-      "step": 40125
-    },
-    {
-      "epoch": 3.0679129154959193,
-      "grad_norm": 0.002659546909853816,
-      "learning_rate": 0.00019999535766350446,
-      "loss": 46.0,
-      "step": 40126
-    },
-    {
-      "epoch": 3.067989372479309,
-      "grad_norm": 0.0016413042321801186,
-      "learning_rate": 0.00019999535743205773,
-      "loss": 46.0,
-      "step": 40127
-    },
-    {
-      "epoch": 3.0680658294626983,
-      "grad_norm": 0.0067235915921628475,
-      "learning_rate": 0.00019999535720060526,
-      "loss": 46.0,
-      "step": 40128
-    },
-    {
-      "epoch": 3.068142286446088,
-      "grad_norm": 0.0025130000431090593,
-      "learning_rate": 0.00019999535696914696,
-      "loss": 46.0,
-      "step": 40129
-    },
-    {
-      "epoch": 3.068218743429478,
-      "grad_norm": 0.000890807481482625,
-      "learning_rate": 0.00019999535673768292,
-      "loss": 46.0,
-      "step": 40130
-    },
-    {
-      "epoch": 3.0682952004128676,
-      "grad_norm": 0.002069278387352824,
-      "learning_rate": 0.00019999535650621315,
-      "loss": 46.0,
-      "step": 40131
-    },
-    {
-      "epoch": 3.0683716573962574,
-      "grad_norm": 0.0019009563839063048,
-      "learning_rate": 0.00019999535627473756,
-      "loss": 46.0,
-      "step": 40132
-    },
-    {
-      "epoch": 3.068448114379647,
-      "grad_norm": 0.00897801760584116,
-      "learning_rate": 0.00019999535604325622,
-      "loss": 46.0,
-      "step": 40133
-    },
-    {
-      "epoch": 3.068524571363037,
-      "grad_norm": 0.006295171566307545,
-      "learning_rate": 0.0001999953558117691,
-      "loss": 46.0,
-      "step": 40134
-    },
-    {
-      "epoch": 3.0686010283464267,
-      "grad_norm": 0.0012238348135724664,
-      "learning_rate": 0.00019999535558027623,
-      "loss": 46.0,
-      "step": 40135
-    },
-    {
-      "epoch": 3.0686774853298164,
-      "grad_norm": 0.0023444064427167177,
-      "learning_rate": 0.00019999535534877757,
-      "loss": 46.0,
-      "step": 40136
-    },
-    {
-      "epoch": 3.068753942313206,
-      "grad_norm": 0.0026438580825924873,
-      "learning_rate": 0.00019999535511727316,
-      "loss": 46.0,
-      "step": 40137
-    },
-    {
-      "epoch": 3.068830399296596,
-      "grad_norm": 0.0035757767036557198,
-      "learning_rate": 0.000199995354885763,
-      "loss": 46.0,
-      "step": 40138
-    },
-    {
-      "epoch": 3.0689068562799853,
-      "grad_norm": 0.002656287048012018,
-      "learning_rate": 0.000199995354654247,
-      "loss": 46.0,
-      "step": 40139
-    },
-    {
-      "epoch": 3.068983313263375,
-      "grad_norm": 0.009296913631260395,
-      "learning_rate": 0.00019999535442272528,
-      "loss": 46.0,
-      "step": 40140
-    },
-    {
-      "epoch": 3.069059770246765,
-      "grad_norm": 0.004610334988683462,
-      "learning_rate": 0.0001999953541911978,
-      "loss": 46.0,
-      "step": 40141
-    },
-    {
-      "epoch": 3.0691362272301546,
-      "grad_norm": 0.004539922345429659,
-      "learning_rate": 0.00019999535395966454,
-      "loss": 46.0,
-      "step": 40142
-    },
-    {
-      "epoch": 3.0692126842135443,
-      "grad_norm": 0.001351111801341176,
-      "learning_rate": 0.0001999953537281255,
-      "loss": 46.0,
-      "step": 40143
-    },
-    {
-      "epoch": 3.069289141196934,
-      "grad_norm": 0.001297913957387209,
-      "learning_rate": 0.0001999953534965807,
-      "loss": 46.0,
-      "step": 40144
-    },
-    {
-      "epoch": 3.069365598180324,
-      "grad_norm": 0.006673015654087067,
-      "learning_rate": 0.00019999535326503014,
-      "loss": 46.0,
-      "step": 40145
-    },
-    {
-      "epoch": 3.0694420551637136,
-      "grad_norm": 0.0012794877402484417,
-      "learning_rate": 0.00019999535303347378,
-      "loss": 46.0,
-      "step": 40146
-    },
-    {
-      "epoch": 3.0695185121471034,
-      "grad_norm": 0.0027004843577742577,
-      "learning_rate": 0.00019999535280191167,
-      "loss": 46.0,
-      "step": 40147
-    },
-    {
-      "epoch": 3.069594969130493,
-      "grad_norm": 0.0023454611655324697,
-      "learning_rate": 0.00019999535257034378,
-      "loss": 46.0,
-      "step": 40148
-    },
-    {
-      "epoch": 3.069671426113883,
-      "grad_norm": 0.0034696401562541723,
-      "learning_rate": 0.00019999535233877012,
-      "loss": 46.0,
-      "step": 40149
-    },
-    {
-      "epoch": 3.069747883097272,
-      "grad_norm": 0.0046706423163414,
-      "learning_rate": 0.00019999535210719075,
-      "loss": 46.0,
-      "step": 40150
-    },
-    {
-      "epoch": 3.069824340080662,
-      "grad_norm": 0.002217352157458663,
-      "learning_rate": 0.00019999535187560554,
-      "loss": 46.0,
-      "step": 40151
-    },
-    {
-      "epoch": 3.0699007970640517,
-      "grad_norm": 0.0018033245578408241,
-      "learning_rate": 0.0001999953516440146,
-      "loss": 46.0,
-      "step": 40152
-    },
-    {
-      "epoch": 3.0699772540474415,
-      "grad_norm": 0.0009308006265200675,
-      "learning_rate": 0.00019999535141241784,
-      "loss": 46.0,
-      "step": 40153
-    },
-    {
-      "epoch": 3.0700537110308312,
-      "grad_norm": 0.005178571213036776,
-      "learning_rate": 0.00019999535118081537,
-      "loss": 46.0,
-      "step": 40154
-    },
-    {
-      "epoch": 3.070130168014221,
-      "grad_norm": 0.003584745107218623,
-      "learning_rate": 0.0001999953509492071,
-      "loss": 46.0,
-      "step": 40155
-    },
-    {
-      "epoch": 3.0702066249976108,
-      "grad_norm": 0.0010845190845429897,
-      "learning_rate": 0.00019999535071759308,
-      "loss": 46.0,
-      "step": 40156
-    },
-    {
-      "epoch": 3.0702830819810005,
-      "grad_norm": 0.002099919132888317,
-      "learning_rate": 0.0001999953504859733,
-      "loss": 46.0,
-      "step": 40157
-    },
-    {
-      "epoch": 3.0703595389643903,
-      "grad_norm": 0.004751711152493954,
-      "learning_rate": 0.0001999953502543477,
-      "loss": 46.0,
-      "step": 40158
-    },
-    {
-      "epoch": 3.07043599594778,
-      "grad_norm": 0.003966817632317543,
-      "learning_rate": 0.00019999535002271636,
-      "loss": 46.0,
-      "step": 40159
-    },
-    {
-      "epoch": 3.07051245293117,
-      "grad_norm": 0.002709440654143691,
-      "learning_rate": 0.00019999534979107925,
-      "loss": 46.0,
-      "step": 40160
-    },
-    {
-      "epoch": 3.070588909914559,
-      "grad_norm": 0.004585724789649248,
-      "learning_rate": 0.00019999534955943637,
-      "loss": 46.0,
-      "step": 40161
-    },
-    {
-      "epoch": 3.070665366897949,
-      "grad_norm": 0.0016803338658064604,
-      "learning_rate": 0.00019999534932778774,
-      "loss": 46.0,
-      "step": 40162
-    },
-    {
-      "epoch": 3.0707418238813387,
-      "grad_norm": 0.001813422655686736,
-      "learning_rate": 0.0001999953490961333,
-      "loss": 46.0,
-      "step": 40163
-    },
-    {
-      "epoch": 3.0708182808647284,
-      "grad_norm": 0.0023779154289513826,
-      "learning_rate": 0.0001999953488644731,
-      "loss": 46.0,
-      "step": 40164
-    },
-    {
-      "epoch": 3.070894737848118,
-      "grad_norm": 0.0024741145316511393,
-      "learning_rate": 0.00019999534863280718,
-      "loss": 46.0,
-      "step": 40165
-    },
-    {
-      "epoch": 3.070971194831508,
-      "grad_norm": 0.003069663653150201,
-      "learning_rate": 0.00019999534840113543,
-      "loss": 46.0,
-      "step": 40166
-    },
-    {
-      "epoch": 3.0710476518148977,
-      "grad_norm": 0.0035713817924261093,
-      "learning_rate": 0.00019999534816945796,
-      "loss": 46.0,
-      "step": 40167
-    },
-    {
-      "epoch": 3.0711241087982875,
-      "grad_norm": 0.001247467938810587,
-      "learning_rate": 0.00019999534793777471,
-      "loss": 46.0,
-      "step": 40168
-    },
-    {
-      "epoch": 3.071200565781677,
-      "grad_norm": 0.0021556986030191183,
-      "learning_rate": 0.00019999534770608564,
-      "loss": 46.0,
-      "step": 40169
-    },
-    {
-      "epoch": 3.071277022765067,
-      "grad_norm": 0.0012641482753679156,
-      "learning_rate": 0.00019999534747439083,
-      "loss": 46.0,
-      "step": 40170
-    },
-    {
-      "epoch": 3.0713534797484567,
-      "grad_norm": 0.0012209730921313167,
-      "learning_rate": 0.00019999534724269026,
-      "loss": 46.0,
-      "step": 40171
-    },
-    {
-      "epoch": 3.071429936731846,
-      "grad_norm": 0.008485635742545128,
-      "learning_rate": 0.00019999534701098395,
-      "loss": 46.0,
-      "step": 40172
-    },
-    {
-      "epoch": 3.071506393715236,
-      "grad_norm": 0.007556588854640722,
-      "learning_rate": 0.00019999534677927184,
-      "loss": 46.0,
-      "step": 40173
-    },
-    {
-      "epoch": 3.0715828506986256,
-      "grad_norm": 0.0024074390530586243,
-      "learning_rate": 0.00019999534654755396,
-      "loss": 46.0,
-      "step": 40174
-    },
-    {
-      "epoch": 3.0716593076820153,
-      "grad_norm": 0.0018514370312914252,
-      "learning_rate": 0.0001999953463158303,
-      "loss": 46.0,
-      "step": 40175
-    },
-    {
-      "epoch": 3.071735764665405,
-      "grad_norm": 0.003574805799871683,
-      "learning_rate": 0.0001999953460841009,
-      "loss": 46.0,
-      "step": 40176
-    },
-    {
-      "epoch": 3.071812221648795,
-      "grad_norm": 0.0019145908299833536,
-      "learning_rate": 0.0001999953458523657,
-      "loss": 46.0,
-      "step": 40177
-    },
-    {
-      "epoch": 3.0718886786321846,
-      "grad_norm": 0.0030677402392029762,
-      "learning_rate": 0.00019999534562062477,
-      "loss": 46.0,
-      "step": 40178
-    },
-    {
-      "epoch": 3.0719651356155744,
-      "grad_norm": 0.002042026026174426,
-      "learning_rate": 0.00019999534538887802,
-      "loss": 46.0,
-      "step": 40179
-    },
-    {
-      "epoch": 3.072041592598964,
-      "grad_norm": 0.0029883794486522675,
-      "learning_rate": 0.00019999534515712552,
-      "loss": 46.0,
-      "step": 40180
-    },
-    {
-      "epoch": 3.072118049582354,
-      "grad_norm": 0.003546859370544553,
-      "learning_rate": 0.00019999534492536728,
-      "loss": 46.0,
-      "step": 40181
-    },
-    {
-      "epoch": 3.0721945065657437,
-      "grad_norm": 0.002618842525407672,
-      "learning_rate": 0.00019999534469360324,
-      "loss": 46.0,
-      "step": 40182
-    },
-    {
-      "epoch": 3.072270963549133,
-      "grad_norm": 0.0011484178248792887,
-      "learning_rate": 0.00019999534446183345,
-      "loss": 46.0,
-      "step": 40183
-    },
-    {
-      "epoch": 3.0723474205325227,
-      "grad_norm": 0.0018216263270005584,
-      "learning_rate": 0.00019999534423005785,
-      "loss": 46.0,
-      "step": 40184
-    },
-    {
-      "epoch": 3.0724238775159125,
-      "grad_norm": 0.0013706331374123693,
-      "learning_rate": 0.00019999534399827652,
-      "loss": 46.0,
-      "step": 40185
-    },
-    {
-      "epoch": 3.0725003344993023,
-      "grad_norm": 0.0013992574531584978,
-      "learning_rate": 0.00019999534376648943,
-      "loss": 46.0,
-      "step": 40186
-    },
-    {
-      "epoch": 3.072576791482692,
-      "grad_norm": 0.002979496493935585,
-      "learning_rate": 0.00019999534353469652,
-      "loss": 46.0,
-      "step": 40187
-    },
-    {
-      "epoch": 3.072653248466082,
-      "grad_norm": 0.0015877066180109978,
-      "learning_rate": 0.00019999534330289792,
-      "loss": 46.0,
-      "step": 40188
-    },
-    {
-      "epoch": 3.0727297054494715,
-      "grad_norm": 0.0013644949067384005,
-      "learning_rate": 0.0001999953430710935,
-      "loss": 46.0,
-      "step": 40189
-    },
-    {
-      "epoch": 3.0728061624328613,
-      "grad_norm": 0.0019357744604349136,
-      "learning_rate": 0.0001999953428392833,
-      "loss": 46.0,
-      "step": 40190
-    },
-    {
-      "epoch": 3.072882619416251,
-      "grad_norm": 0.0014490032335743308,
-      "learning_rate": 0.00019999534260746734,
-      "loss": 46.0,
-      "step": 40191
-    },
-    {
-      "epoch": 3.072959076399641,
-      "grad_norm": 0.00280370214022696,
-      "learning_rate": 0.00019999534237564564,
-      "loss": 46.0,
-      "step": 40192
-    },
-    {
-      "epoch": 3.07303553338303,
-      "grad_norm": 0.0015836763195693493,
-      "learning_rate": 0.00019999534214381815,
-      "loss": 46.0,
-      "step": 40193
-    },
-    {
-      "epoch": 3.07311199036642,
-      "grad_norm": 0.0015210001729428768,
-      "learning_rate": 0.00019999534191198488,
-      "loss": 46.0,
-      "step": 40194
-    },
-    {
-      "epoch": 3.0731884473498097,
-      "grad_norm": 0.002355832140892744,
-      "learning_rate": 0.00019999534168014583,
-      "loss": 46.0,
-      "step": 40195
-    },
-    {
-      "epoch": 3.0732649043331994,
-      "grad_norm": 0.0025815358385443687,
-      "learning_rate": 0.00019999534144830104,
-      "loss": 46.0,
-      "step": 40196
-    },
-    {
-      "epoch": 3.073341361316589,
-      "grad_norm": 0.0015076538547873497,
-      "learning_rate": 0.00019999534121645045,
-      "loss": 46.0,
-      "step": 40197
-    },
-    {
-      "epoch": 3.073417818299979,
-      "grad_norm": 0.0012456843396648765,
-      "learning_rate": 0.00019999534098459414,
-      "loss": 46.0,
-      "step": 40198
-    },
-    {
-      "epoch": 3.0734942752833687,
-      "grad_norm": 0.0026640426367521286,
-      "learning_rate": 0.000199995340752732,
-      "loss": 46.0,
-      "step": 40199
-    },
-    {
-      "epoch": 3.0735707322667585,
-      "grad_norm": 0.0021019652485847473,
-      "learning_rate": 0.00019999534052086415,
-      "loss": 46.0,
-      "step": 40200
-    },
-    {
-      "epoch": 3.0736471892501482,
-      "grad_norm": 0.0023133703507483006,
-      "learning_rate": 0.0001999953402889905,
-      "loss": 46.0,
-      "step": 40201
-    },
-    {
-      "epoch": 3.073723646233538,
-      "grad_norm": 0.002192039042711258,
-      "learning_rate": 0.00019999534005711106,
-      "loss": 46.0,
-      "step": 40202
-    },
-    {
-      "epoch": 3.0738001032169278,
-      "grad_norm": 0.0012913175160065293,
-      "learning_rate": 0.0001999953398252259,
-      "loss": 46.0,
-      "step": 40203
-    },
-    {
-      "epoch": 3.0738765602003175,
-      "grad_norm": 0.0011938527459278703,
-      "learning_rate": 0.00019999533959333494,
-      "loss": 46.0,
-      "step": 40204
-    },
-    {
-      "epoch": 3.073953017183707,
-      "grad_norm": 0.004477009642869234,
-      "learning_rate": 0.00019999533936143821,
-      "loss": 46.0,
-      "step": 40205
-    },
-    {
-      "epoch": 3.0740294741670966,
-      "grad_norm": 0.0006661259685643017,
-      "learning_rate": 0.00019999533912953572,
-      "loss": 46.0,
-      "step": 40206
-    },
-    {
-      "epoch": 3.0741059311504864,
-      "grad_norm": 0.0007705508614890277,
-      "learning_rate": 0.00019999533889762748,
-      "loss": 46.0,
-      "step": 40207
-    },
-    {
-      "epoch": 3.074182388133876,
-      "grad_norm": 0.0027248139958828688,
-      "learning_rate": 0.00019999533866571346,
-      "loss": 46.0,
-      "step": 40208
-    },
-    {
-      "epoch": 3.074258845117266,
-      "grad_norm": 0.002002173801884055,
-      "learning_rate": 0.00019999533843379364,
-      "loss": 46.0,
-      "step": 40209
-    },
-    {
-      "epoch": 3.0743353021006556,
-      "grad_norm": 0.001847300329245627,
-      "learning_rate": 0.00019999533820186805,
-      "loss": 46.0,
-      "step": 40210
-    },
-    {
-      "epoch": 3.0744117590840454,
-      "grad_norm": 0.0011133237276226282,
-      "learning_rate": 0.00019999533796993672,
-      "loss": 46.0,
-      "step": 40211
-    },
-    {
-      "epoch": 3.074488216067435,
-      "grad_norm": 0.008085299283266068,
-      "learning_rate": 0.00019999533773799963,
-      "loss": 46.0,
-      "step": 40212
-    },
-    {
-      "epoch": 3.074564673050825,
-      "grad_norm": 0.0023897693026810884,
-      "learning_rate": 0.00019999533750605675,
-      "loss": 46.0,
-      "step": 40213
-    },
-    {
-      "epoch": 3.0746411300342147,
-      "grad_norm": 0.0013743591262027621,
-      "learning_rate": 0.0001999953372741081,
-      "loss": 46.0,
-      "step": 40214
-    },
-    {
-      "epoch": 3.074717587017604,
-      "grad_norm": 0.0028470742981880903,
-      "learning_rate": 0.00019999533704215367,
-      "loss": 46.0,
-      "step": 40215
-    },
-    {
-      "epoch": 3.0747940440009938,
-      "grad_norm": 0.005823993589729071,
-      "learning_rate": 0.0001999953368101935,
-      "loss": 46.0,
-      "step": 40216
-    },
-    {
-      "epoch": 3.0748705009843835,
-      "grad_norm": 0.002277072286233306,
-      "learning_rate": 0.00019999533657822757,
-      "loss": 46.0,
-      "step": 40217
-    },
-    {
-      "epoch": 3.0749469579677733,
-      "grad_norm": 0.0028651715256273746,
-      "learning_rate": 0.00019999533634625582,
-      "loss": 46.0,
-      "step": 40218
-    },
-    {
-      "epoch": 3.075023414951163,
-      "grad_norm": 0.0027364131528884172,
-      "learning_rate": 0.00019999533611427835,
-      "loss": 46.0,
-      "step": 40219
-    },
-    {
-      "epoch": 3.075099871934553,
-      "grad_norm": 0.0020083829294890165,
-      "learning_rate": 0.00019999533588229505,
-      "loss": 46.0,
-      "step": 40220
-    },
-    {
-      "epoch": 3.0751763289179426,
-      "grad_norm": 0.0016049712430685759,
-      "learning_rate": 0.00019999533565030606,
-      "loss": 46.0,
-      "step": 40221
-    },
-    {
-      "epoch": 3.0752527859013323,
-      "grad_norm": 0.0029372749850153923,
-      "learning_rate": 0.00019999533541831125,
-      "loss": 46.0,
-      "step": 40222
-    },
-    {
-      "epoch": 3.075329242884722,
-      "grad_norm": 0.0033449844922870398,
-      "learning_rate": 0.00019999533518631069,
-      "loss": 46.0,
-      "step": 40223
-    },
-    {
-      "epoch": 3.075405699868112,
-      "grad_norm": 0.00678646843880415,
-      "learning_rate": 0.00019999533495430435,
-      "loss": 46.0,
-      "step": 40224
-    },
-    {
-      "epoch": 3.0754821568515016,
-      "grad_norm": 0.0009200061322189867,
-      "learning_rate": 0.00019999533472229224,
-      "loss": 46.0,
-      "step": 40225
-    },
-    {
-      "epoch": 3.075558613834891,
-      "grad_norm": 0.0027682597283273935,
-      "learning_rate": 0.00019999533449027436,
-      "loss": 46.0,
-      "step": 40226
-    },
-    {
-      "epoch": 3.0756350708182807,
-      "grad_norm": 0.0021547572687268257,
-      "learning_rate": 0.00019999533425825073,
-      "loss": 46.0,
-      "step": 40227
-    },
-    {
-      "epoch": 3.0757115278016705,
-      "grad_norm": 0.0034392361994832754,
-      "learning_rate": 0.0001999953340262213,
-      "loss": 46.0,
-      "step": 40228
-    },
-    {
-      "epoch": 3.07578798478506,
-      "grad_norm": 0.0007730411016382277,
-      "learning_rate": 0.00019999533379418612,
-      "loss": 46.0,
-      "step": 40229
-    },
-    {
-      "epoch": 3.07586444176845,
-      "grad_norm": 0.0023892747703939676,
-      "learning_rate": 0.00019999533356214518,
-      "loss": 46.0,
-      "step": 40230
-    },
-    {
-      "epoch": 3.0759408987518397,
-      "grad_norm": 0.0017106481827795506,
-      "learning_rate": 0.00019999533333009843,
-      "loss": 46.0,
-      "step": 40231
-    },
-    {
-      "epoch": 3.0760173557352295,
-      "grad_norm": 0.0008947617025114596,
-      "learning_rate": 0.00019999533309804596,
-      "loss": 46.0,
-      "step": 40232
-    },
-    {
-      "epoch": 3.0760938127186193,
-      "grad_norm": 0.001231977017596364,
-      "learning_rate": 0.0001999953328659877,
-      "loss": 46.0,
-      "step": 40233
-    },
-    {
-      "epoch": 3.076170269702009,
-      "grad_norm": 0.008011367172002792,
-      "learning_rate": 0.00019999533263392367,
-      "loss": 46.0,
-      "step": 40234
-    },
-    {
-      "epoch": 3.076246726685399,
-      "grad_norm": 0.001852184534072876,
-      "learning_rate": 0.00019999533240185386,
-      "loss": 46.0,
-      "step": 40235
-    },
-    {
-      "epoch": 3.0763231836687885,
-      "grad_norm": 0.00785853248089552,
-      "learning_rate": 0.0001999953321697783,
-      "loss": 46.0,
-      "step": 40236
-    },
-    {
-      "epoch": 3.076399640652178,
-      "grad_norm": 0.0012617373140528798,
-      "learning_rate": 0.00019999533193769696,
-      "loss": 46.0,
-      "step": 40237
-    },
-    {
-      "epoch": 3.0764760976355676,
-      "grad_norm": 0.004136977251619101,
-      "learning_rate": 0.00019999533170560985,
-      "loss": 46.0,
-      "step": 40238
-    },
-    {
-      "epoch": 3.0765525546189574,
-      "grad_norm": 0.0024272422306239605,
-      "learning_rate": 0.00019999533147351697,
-      "loss": 46.0,
-      "step": 40239
-    },
-    {
-      "epoch": 3.076629011602347,
-      "grad_norm": 0.003413500962778926,
-      "learning_rate": 0.00019999533124141834,
-      "loss": 46.0,
-      "step": 40240
-    },
-    {
-      "epoch": 3.076705468585737,
-      "grad_norm": 0.002280991291627288,
-      "learning_rate": 0.00019999533100931394,
-      "loss": 46.0,
-      "step": 40241
-    },
-    {
-      "epoch": 3.0767819255691267,
-      "grad_norm": 0.005484189372509718,
-      "learning_rate": 0.00019999533077720374,
-      "loss": 46.0,
-      "step": 40242
-    },
-    {
-      "epoch": 3.0768583825525164,
-      "grad_norm": 0.004708023741841316,
-      "learning_rate": 0.0001999953305450878,
-      "loss": 46.0,
-      "step": 40243
-    },
-    {
-      "epoch": 3.076934839535906,
-      "grad_norm": 0.0008094694931060076,
-      "learning_rate": 0.00019999533031296606,
-      "loss": 46.0,
-      "step": 40244
-    },
-    {
-      "epoch": 3.077011296519296,
-      "grad_norm": 0.0012199186021462083,
-      "learning_rate": 0.0001999953300808386,
-      "loss": 46.0,
-      "step": 40245
-    },
-    {
-      "epoch": 3.0770877535026857,
-      "grad_norm": 0.002402589190751314,
-      "learning_rate": 0.0001999953298487053,
-      "loss": 46.0,
-      "step": 40246
-    },
-    {
-      "epoch": 3.0771642104860755,
-      "grad_norm": 0.002540545305237174,
-      "learning_rate": 0.00019999532961656629,
-      "loss": 46.0,
-      "step": 40247
-    },
-    {
-      "epoch": 3.077240667469465,
-      "grad_norm": 0.003308674320578575,
-      "learning_rate": 0.0001999953293844215,
-      "loss": 46.0,
-      "step": 40248
-    },
-    {
-      "epoch": 3.0773171244528545,
-      "grad_norm": 0.006334108766168356,
-      "learning_rate": 0.00019999532915227094,
-      "loss": 46.0,
-      "step": 40249
-    },
-    {
-      "epoch": 3.0773935814362443,
-      "grad_norm": 0.001276075141504407,
-      "learning_rate": 0.0001999953289201146,
-      "loss": 46.0,
-      "step": 40250
-    },
-    {
-      "epoch": 3.077470038419634,
-      "grad_norm": 0.0024477725382894278,
-      "learning_rate": 0.00019999532868795246,
-      "loss": 46.0,
-      "step": 40251
-    },
-    {
-      "epoch": 3.077546495403024,
-      "grad_norm": 0.0038398776669055223,
-      "learning_rate": 0.0001999953284557846,
-      "loss": 46.0,
-      "step": 40252
-    },
-    {
-      "epoch": 3.0776229523864136,
-      "grad_norm": 0.0033194429706782103,
-      "learning_rate": 0.00019999532822361098,
-      "loss": 46.0,
-      "step": 40253
-    },
-    {
-      "epoch": 3.0776994093698034,
-      "grad_norm": 0.0021062400192022324,
-      "learning_rate": 0.00019999532799143155,
-      "loss": 46.0,
-      "step": 40254
-    },
-    {
-      "epoch": 3.077775866353193,
-      "grad_norm": 0.0019021752523258328,
-      "learning_rate": 0.00019999532775924638,
-      "loss": 46.0,
-      "step": 40255
-    },
-    {
-      "epoch": 3.077852323336583,
-      "grad_norm": 0.0021422221325337887,
-      "learning_rate": 0.0001999953275270554,
-      "loss": 46.0,
-      "step": 40256
-    },
-    {
-      "epoch": 3.0779287803199726,
-      "grad_norm": 0.007943300530314445,
-      "learning_rate": 0.0001999953272948587,
-      "loss": 46.0,
-      "step": 40257
-    },
-    {
-      "epoch": 3.0780052373033624,
-      "grad_norm": 0.002355459611862898,
-      "learning_rate": 0.0001999953270626562,
-      "loss": 46.0,
-      "step": 40258
-    },
-    {
-      "epoch": 3.0780816942867517,
-      "grad_norm": 0.0021378209348767996,
-      "learning_rate": 0.00019999532683044794,
-      "loss": 46.0,
-      "step": 40259
-    },
-    {
-      "epoch": 3.0781581512701415,
-      "grad_norm": 0.003569244872778654,
-      "learning_rate": 0.00019999532659823393,
-      "loss": 46.0,
-      "step": 40260
-    },
-    {
-      "epoch": 3.0782346082535312,
-      "grad_norm": 0.004313362762331963,
-      "learning_rate": 0.00019999532636601411,
-      "loss": 46.0,
-      "step": 40261
-    },
-    {
-      "epoch": 3.078311065236921,
-      "grad_norm": 0.0017759567126631737,
-      "learning_rate": 0.00019999532613378858,
-      "loss": 46.0,
-      "step": 40262
-    },
-    {
-      "epoch": 3.0783875222203108,
-      "grad_norm": 0.002193351509049535,
-      "learning_rate": 0.00019999532590155722,
-      "loss": 46.0,
-      "step": 40263
-    },
-    {
-      "epoch": 3.0784639792037005,
-      "grad_norm": 0.0042168269865214825,
-      "learning_rate": 0.0001999953256693201,
-      "loss": 46.0,
-      "step": 40264
-    },
-    {
-      "epoch": 3.0785404361870903,
-      "grad_norm": 0.0035198985133320093,
-      "learning_rate": 0.00019999532543707723,
-      "loss": 46.0,
-      "step": 40265
-    },
-    {
-      "epoch": 3.07861689317048,
-      "grad_norm": 0.0043728286400437355,
-      "learning_rate": 0.0001999953252048286,
-      "loss": 46.0,
-      "step": 40266
-    },
-    {
-      "epoch": 3.07869335015387,
-      "grad_norm": 0.006337030790746212,
-      "learning_rate": 0.0001999953249725742,
-      "loss": 46.0,
-      "step": 40267
-    },
-    {
-      "epoch": 3.0787698071372596,
-      "grad_norm": 0.006617938168346882,
-      "learning_rate": 0.000199995324740314,
-      "loss": 46.0,
-      "step": 40268
-    },
-    {
-      "epoch": 3.0788462641206493,
-      "grad_norm": 0.0021102922037243843,
-      "learning_rate": 0.00019999532450804805,
-      "loss": 46.0,
-      "step": 40269
-    },
-    {
-      "epoch": 3.0789227211040386,
-      "grad_norm": 0.003457592334598303,
-      "learning_rate": 0.00019999532427577633,
-      "loss": 46.0,
-      "step": 40270
-    },
-    {
-      "epoch": 3.0789991780874284,
-      "grad_norm": 0.004359242040663958,
-      "learning_rate": 0.00019999532404349886,
-      "loss": 46.0,
-      "step": 40271
-    },
-    {
-      "epoch": 3.079075635070818,
-      "grad_norm": 0.001231861999258399,
-      "learning_rate": 0.0001999953238112156,
-      "loss": 46.0,
-      "step": 40272
-    },
-    {
-      "epoch": 3.079152092054208,
-      "grad_norm": 0.002020505489781499,
-      "learning_rate": 0.00019999532357892655,
-      "loss": 46.0,
-      "step": 40273
-    },
-    {
-      "epoch": 3.0792285490375977,
-      "grad_norm": 0.0017457961803302169,
-      "learning_rate": 0.00019999532334663176,
-      "loss": 46.0,
-      "step": 40274
-    },
-    {
-      "epoch": 3.0793050060209874,
-      "grad_norm": 0.0009132236009463668,
-      "learning_rate": 0.0001999953231143312,
-      "loss": 46.0,
-      "step": 40275
-    },
-    {
-      "epoch": 3.079381463004377,
-      "grad_norm": 0.003237878903746605,
-      "learning_rate": 0.00019999532288202487,
-      "loss": 46.0,
-      "step": 40276
-    },
-    {
-      "epoch": 3.079457919987767,
-      "grad_norm": 0.0009749418823048472,
-      "learning_rate": 0.00019999532264971276,
-      "loss": 46.0,
-      "step": 40277
-    },
-    {
-      "epoch": 3.0795343769711567,
-      "grad_norm": 0.0009238472557626665,
-      "learning_rate": 0.0001999953224173949,
-      "loss": 46.0,
-      "step": 40278
-    },
-    {
-      "epoch": 3.0796108339545465,
-      "grad_norm": 0.0025606260169297457,
-      "learning_rate": 0.00019999532218507125,
-      "loss": 46.0,
-      "step": 40279
-    },
-    {
-      "epoch": 3.0796872909379363,
-      "grad_norm": 0.00390832731500268,
-      "learning_rate": 0.00019999532195274182,
-      "loss": 46.0,
-      "step": 40280
-    },
-    {
-      "epoch": 3.0797637479213256,
-      "grad_norm": 0.0019516630563884974,
-      "learning_rate": 0.00019999532172040665,
-      "loss": 46.0,
-      "step": 40281
-    },
-    {
-      "epoch": 3.0798402049047153,
-      "grad_norm": 0.003210377646610141,
-      "learning_rate": 0.0001999953214880657,
-      "loss": 46.0,
-      "step": 40282
-    },
-    {
-      "epoch": 3.079916661888105,
-      "grad_norm": 0.002234569750726223,
-      "learning_rate": 0.00019999532125571898,
-      "loss": 46.0,
-      "step": 40283
-    },
-    {
-      "epoch": 3.079993118871495,
-      "grad_norm": 0.0014541083946824074,
-      "learning_rate": 0.00019999532102336648,
-      "loss": 46.0,
-      "step": 40284
-    },
-    {
-      "epoch": 3.0800695758548846,
-      "grad_norm": 0.0013030937407165766,
-      "learning_rate": 0.00019999532079100824,
-      "loss": 46.0,
-      "step": 40285
-    },
-    {
-      "epoch": 3.0801460328382744,
-      "grad_norm": 0.0012608846882358193,
-      "learning_rate": 0.00019999532055864423,
-      "loss": 46.0,
-      "step": 40286
-    },
-    {
-      "epoch": 3.080222489821664,
-      "grad_norm": 0.0012387470342218876,
-      "learning_rate": 0.00019999532032627442,
-      "loss": 46.0,
-      "step": 40287
-    },
-    {
-      "epoch": 3.080298946805054,
-      "grad_norm": 0.0013427947415038943,
-      "learning_rate": 0.00019999532009389885,
-      "loss": 46.0,
-      "step": 40288
-    },
-    {
-      "epoch": 3.0803754037884437,
-      "grad_norm": 0.0024567232467234135,
-      "learning_rate": 0.00019999531986151752,
-      "loss": 46.0,
-      "step": 40289
-    },
-    {
-      "epoch": 3.0804518607718334,
-      "grad_norm": 0.0012409926857799292,
-      "learning_rate": 0.00019999531962913044,
-      "loss": 46.0,
-      "step": 40290
-    },
-    {
-      "epoch": 3.080528317755223,
-      "grad_norm": 0.0030446844175457954,
-      "learning_rate": 0.00019999531939673756,
-      "loss": 46.0,
-      "step": 40291
-    },
-    {
-      "epoch": 3.0806047747386125,
-      "grad_norm": 0.002904205583035946,
-      "learning_rate": 0.00019999531916433888,
-      "loss": 46.0,
-      "step": 40292
-    },
-    {
-      "epoch": 3.0806812317220023,
-      "grad_norm": 0.005621998570859432,
-      "learning_rate": 0.00019999531893193448,
-      "loss": 46.0,
-      "step": 40293
-    },
-    {
-      "epoch": 3.080757688705392,
-      "grad_norm": 0.006240567658096552,
-      "learning_rate": 0.0001999953186995243,
-      "loss": 46.0,
-      "step": 40294
-    },
-    {
-      "epoch": 3.0808341456887818,
-      "grad_norm": 0.003941263537853956,
-      "learning_rate": 0.00019999531846710836,
-      "loss": 46.0,
-      "step": 40295
-    },
-    {
-      "epoch": 3.0809106026721715,
-      "grad_norm": 0.0013673343928530812,
-      "learning_rate": 0.00019999531823468663,
-      "loss": 46.0,
-      "step": 40296
-    },
-    {
-      "epoch": 3.0809870596555613,
-      "grad_norm": 0.002029367024078965,
-      "learning_rate": 0.00019999531800225917,
-      "loss": 46.0,
-      "step": 40297
-    },
-    {
-      "epoch": 3.081063516638951,
-      "grad_norm": 0.0007272011134773493,
-      "learning_rate": 0.0001999953177698259,
-      "loss": 46.0,
-      "step": 40298
-    },
-    {
-      "epoch": 3.081139973622341,
-      "grad_norm": 0.0014875493943691254,
-      "learning_rate": 0.00019999531753738686,
-      "loss": 46.0,
-      "step": 40299
-    },
-    {
-      "epoch": 3.0812164306057306,
-      "grad_norm": 0.00400215107947588,
-      "learning_rate": 0.00019999531730494207,
-      "loss": 46.0,
-      "step": 40300
-    },
-    {
-      "epoch": 3.0812928875891203,
-      "grad_norm": 0.0044781663455069065,
-      "learning_rate": 0.0001999953170724915,
-      "loss": 46.0,
-      "step": 40301
-    },
-    {
-      "epoch": 3.08136934457251,
-      "grad_norm": 0.002711773617193103,
-      "learning_rate": 0.00019999531684003515,
-      "loss": 46.0,
-      "step": 40302
-    },
-    {
-      "epoch": 3.0814458015558994,
-      "grad_norm": 0.001435677520930767,
-      "learning_rate": 0.00019999531660757307,
-      "loss": 46.0,
-      "step": 40303
-    },
-    {
-      "epoch": 3.081522258539289,
-      "grad_norm": 0.003028220497071743,
-      "learning_rate": 0.0001999953163751052,
-      "loss": 46.0,
-      "step": 40304
-    },
-    {
-      "epoch": 3.081598715522679,
-      "grad_norm": 0.0011879857629537582,
-      "learning_rate": 0.00019999531614263156,
-      "loss": 46.0,
-      "step": 40305
-    },
-    {
-      "epoch": 3.0816751725060687,
-      "grad_norm": 0.0015074124094098806,
-      "learning_rate": 0.00019999531591015214,
-      "loss": 46.0,
-      "step": 40306
-    },
-    {
-      "epoch": 3.0817516294894585,
-      "grad_norm": 0.00365912402048707,
-      "learning_rate": 0.00019999531567766696,
-      "loss": 46.0,
-      "step": 40307
-    },
-    {
-      "epoch": 3.0818280864728482,
-      "grad_norm": 0.004783504642546177,
-      "learning_rate": 0.00019999531544517601,
-      "loss": 46.0,
-      "step": 40308
-    },
-    {
-      "epoch": 3.081904543456238,
-      "grad_norm": 0.004315034952014685,
-      "learning_rate": 0.00019999531521267932,
-      "loss": 46.0,
-      "step": 40309
-    },
-    {
-      "epoch": 3.0819810004396277,
-      "grad_norm": 0.0016449636314064264,
-      "learning_rate": 0.0001999953149801768,
-      "loss": 46.0,
-      "step": 40310
-    },
-    {
-      "epoch": 3.0820574574230175,
-      "grad_norm": 0.0028042870108038187,
-      "learning_rate": 0.00019999531474766856,
-      "loss": 46.0,
-      "step": 40311
-    },
-    {
-      "epoch": 3.0821339144064073,
-      "grad_norm": 0.0009609071421436965,
-      "learning_rate": 0.00019999531451515452,
-      "loss": 46.0,
-      "step": 40312
-    },
-    {
-      "epoch": 3.082210371389797,
-      "grad_norm": 0.004199567716568708,
-      "learning_rate": 0.00019999531428263473,
-      "loss": 46.0,
-      "step": 40313
-    },
-    {
-      "epoch": 3.0822868283731863,
-      "grad_norm": 0.0023863951209932566,
-      "learning_rate": 0.00019999531405010917,
-      "loss": 46.0,
-      "step": 40314
-    },
-    {
-      "epoch": 3.082363285356576,
-      "grad_norm": 0.0009719316149130464,
-      "learning_rate": 0.00019999531381757787,
-      "loss": 46.0,
-      "step": 40315
-    },
-    {
-      "epoch": 3.082439742339966,
-      "grad_norm": 0.0022970030549913645,
-      "learning_rate": 0.00019999531358504076,
-      "loss": 46.0,
-      "step": 40316
-    },
-    {
-      "epoch": 3.0825161993233556,
-      "grad_norm": 0.0022823235485702753,
-      "learning_rate": 0.00019999531335249788,
-      "loss": 46.0,
-      "step": 40317
-    },
-    {
-      "epoch": 3.0825926563067454,
-      "grad_norm": 0.0020702488254755735,
-      "learning_rate": 0.00019999531311994922,
-      "loss": 46.0,
-      "step": 40318
-    },
-    {
-      "epoch": 3.082669113290135,
-      "grad_norm": 0.0018968835938721895,
-      "learning_rate": 0.00019999531288739482,
-      "loss": 46.0,
-      "step": 40319
-    },
-    {
-      "epoch": 3.082745570273525,
-      "grad_norm": 0.0012051343219354749,
-      "learning_rate": 0.00019999531265483465,
-      "loss": 46.0,
-      "step": 40320
-    },
-    {
-      "epoch": 3.0828220272569147,
-      "grad_norm": 0.0035440793726593256,
-      "learning_rate": 0.0001999953124222687,
-      "loss": 46.0,
-      "step": 40321
-    },
-    {
-      "epoch": 3.0828984842403044,
-      "grad_norm": 0.0036816748324781656,
-      "learning_rate": 0.00019999531218969696,
-      "loss": 46.0,
-      "step": 40322
-    },
-    {
-      "epoch": 3.082974941223694,
-      "grad_norm": 0.0004241867281962186,
-      "learning_rate": 0.0001999953119571195,
-      "loss": 46.0,
-      "step": 40323
-    },
-    {
-      "epoch": 3.0830513982070835,
-      "grad_norm": 0.001081703114323318,
-      "learning_rate": 0.00019999531172453622,
-      "loss": 46.0,
-      "step": 40324
-    },
-    {
-      "epoch": 3.0831278551904733,
-      "grad_norm": 0.0017208452336490154,
-      "learning_rate": 0.0001999953114919472,
-      "loss": 46.0,
-      "step": 40325
-    },
-    {
-      "epoch": 3.083204312173863,
-      "grad_norm": 0.0015251614386215806,
-      "learning_rate": 0.0001999953112593524,
-      "loss": 46.0,
-      "step": 40326
-    },
-    {
-      "epoch": 3.083280769157253,
-      "grad_norm": 0.002002124907448888,
-      "learning_rate": 0.00019999531102675184,
-      "loss": 46.0,
-      "step": 40327
-    },
-    {
-      "epoch": 3.0833572261406426,
-      "grad_norm": 0.0016916863387450576,
-      "learning_rate": 0.0001999953107941455,
-      "loss": 46.0,
-      "step": 40328
-    },
-    {
-      "epoch": 3.0834336831240323,
-      "grad_norm": 0.0018178432947024703,
-      "learning_rate": 0.0001999953105615334,
-      "loss": 46.0,
-      "step": 40329
-    },
-    {
-      "epoch": 3.083510140107422,
-      "grad_norm": 0.0015511164674535394,
-      "learning_rate": 0.00019999531032891554,
-      "loss": 46.0,
-      "step": 40330
-    },
-    {
-      "epoch": 3.083586597090812,
-      "grad_norm": 0.002958128694444895,
-      "learning_rate": 0.00019999531009629192,
-      "loss": 46.0,
-      "step": 40331
-    },
-    {
-      "epoch": 3.0836630540742016,
-      "grad_norm": 0.0012487999629229307,
-      "learning_rate": 0.0001999953098636625,
-      "loss": 46.0,
-      "step": 40332
-    },
-    {
-      "epoch": 3.0837395110575914,
-      "grad_norm": 0.00450740335509181,
-      "learning_rate": 0.00019999530963102732,
-      "loss": 46.0,
-      "step": 40333
-    },
-    {
-      "epoch": 3.083815968040981,
-      "grad_norm": 0.0007050912827253342,
-      "learning_rate": 0.00019999530939838634,
-      "loss": 46.0,
-      "step": 40334
-    },
-    {
-      "epoch": 3.083892425024371,
-      "grad_norm": 0.0030002491548657417,
-      "learning_rate": 0.00019999530916573965,
-      "loss": 46.0,
-      "step": 40335
-    },
-    {
-      "epoch": 3.08396888200776,
-      "grad_norm": 0.003482536878436804,
-      "learning_rate": 0.00019999530893308719,
-      "loss": 46.0,
-      "step": 40336
-    },
-    {
-      "epoch": 3.08404533899115,
-      "grad_norm": 0.001370471902191639,
-      "learning_rate": 0.00019999530870042892,
-      "loss": 46.0,
-      "step": 40337
-    },
-    {
-      "epoch": 3.0841217959745397,
-      "grad_norm": 0.0022118911147117615,
-      "learning_rate": 0.00019999530846776488,
-      "loss": 46.0,
-      "step": 40338
-    },
-    {
-      "epoch": 3.0841982529579295,
-      "grad_norm": 0.0016165869310498238,
-      "learning_rate": 0.0001999953082350951,
-      "loss": 46.0,
-      "step": 40339
-    },
-    {
-      "epoch": 3.0842747099413192,
-      "grad_norm": 0.0027087293565273285,
-      "learning_rate": 0.00019999530800241953,
-      "loss": 46.0,
-      "step": 40340
-    },
-    {
-      "epoch": 3.084351166924709,
-      "grad_norm": 0.0005215605488047004,
-      "learning_rate": 0.0001999953077697382,
-      "loss": 46.0,
-      "step": 40341
-    },
-    {
-      "epoch": 3.0844276239080988,
-      "grad_norm": 0.002495115390047431,
-      "learning_rate": 0.0001999953075370511,
-      "loss": 46.0,
-      "step": 40342
-    },
-    {
-      "epoch": 3.0845040808914885,
-      "grad_norm": 0.0009297642973251641,
-      "learning_rate": 0.00019999530730435821,
-      "loss": 46.0,
-      "step": 40343
-    },
-    {
-      "epoch": 3.0845805378748783,
-      "grad_norm": 0.0008987291366793215,
-      "learning_rate": 0.0001999953070716596,
-      "loss": 46.0,
-      "step": 40344
-    },
-    {
-      "epoch": 3.084656994858268,
-      "grad_norm": 0.00212862784974277,
-      "learning_rate": 0.0001999953068389552,
-      "loss": 46.0,
-      "step": 40345
-    },
-    {
-      "epoch": 3.0847334518416574,
-      "grad_norm": 0.0017014248296618462,
-      "learning_rate": 0.000199995306606245,
-      "loss": 46.0,
-      "step": 40346
-    },
-    {
-      "epoch": 3.084809908825047,
-      "grad_norm": 0.0033167649526149035,
-      "learning_rate": 0.00019999530637352904,
-      "loss": 46.0,
-      "step": 40347
-    },
-    {
-      "epoch": 3.084886365808437,
-      "grad_norm": 0.004921542480587959,
-      "learning_rate": 0.00019999530614080735,
-      "loss": 46.0,
-      "step": 40348
-    },
-    {
-      "epoch": 3.0849628227918267,
-      "grad_norm": 0.0012904637260362506,
-      "learning_rate": 0.00019999530590807986,
-      "loss": 46.0,
-      "step": 40349
-    },
-    {
-      "epoch": 3.0850392797752164,
-      "grad_norm": 0.001202658866532147,
-      "learning_rate": 0.00019999530567534662,
-      "loss": 46.0,
-      "step": 40350
-    },
-    {
-      "epoch": 3.085115736758606,
-      "grad_norm": 0.002073245123028755,
-      "learning_rate": 0.00019999530544260758,
-      "loss": 46.0,
-      "step": 40351
-    },
-    {
-      "epoch": 3.085192193741996,
-      "grad_norm": 0.0009821204002946615,
-      "learning_rate": 0.0001999953052098628,
-      "loss": 46.0,
-      "step": 40352
-    },
-    {
-      "epoch": 3.0852686507253857,
-      "grad_norm": 0.0010281954891979694,
-      "learning_rate": 0.0001999953049771122,
-      "loss": 46.0,
-      "step": 40353
-    },
-    {
-      "epoch": 3.0853451077087755,
-      "grad_norm": 0.001977123087272048,
-      "learning_rate": 0.0001999953047443559,
-      "loss": 46.0,
-      "step": 40354
-    },
-    {
-      "epoch": 3.085421564692165,
-      "grad_norm": 0.002471043262630701,
-      "learning_rate": 0.0001999953045115938,
-      "loss": 46.0,
-      "step": 40355
-    },
-    {
-      "epoch": 3.085498021675555,
-      "grad_norm": 0.002482653595507145,
-      "learning_rate": 0.00019999530427882594,
-      "loss": 46.0,
-      "step": 40356
-    },
-    {
-      "epoch": 3.0855744786589443,
-      "grad_norm": 0.0036249703262001276,
-      "learning_rate": 0.00019999530404605226,
-      "loss": 46.0,
-      "step": 40357
-    },
-    {
-      "epoch": 3.085650935642334,
-      "grad_norm": 0.0012411470524966717,
-      "learning_rate": 0.00019999530381327286,
-      "loss": 46.0,
-      "step": 40358
-    },
-    {
-      "epoch": 3.085727392625724,
-      "grad_norm": 0.0018603705102577806,
-      "learning_rate": 0.0001999953035804877,
-      "loss": 46.0,
-      "step": 40359
-    },
-    {
-      "epoch": 3.0858038496091136,
-      "grad_norm": 0.003228815970942378,
-      "learning_rate": 0.00019999530334769677,
-      "loss": 46.0,
-      "step": 40360
-    },
-    {
-      "epoch": 3.0858803065925033,
-      "grad_norm": 0.0012322701513767242,
-      "learning_rate": 0.00019999530311490005,
-      "loss": 46.0,
-      "step": 40361
-    },
-    {
-      "epoch": 3.085956763575893,
-      "grad_norm": 0.0008629984804429114,
-      "learning_rate": 0.00019999530288209756,
-      "loss": 46.0,
-      "step": 40362
-    },
-    {
-      "epoch": 3.086033220559283,
-      "grad_norm": 0.0023042005486786366,
-      "learning_rate": 0.0001999953026492893,
-      "loss": 46.0,
-      "step": 40363
-    },
-    {
-      "epoch": 3.0861096775426726,
-      "grad_norm": 0.003802621504291892,
-      "learning_rate": 0.00019999530241647526,
-      "loss": 46.0,
-      "step": 40364
-    },
-    {
-      "epoch": 3.0861861345260624,
-      "grad_norm": 0.016155540943145752,
-      "learning_rate": 0.0001999953021836555,
-      "loss": 46.0,
-      "step": 40365
-    },
-    {
-      "epoch": 3.086262591509452,
-      "grad_norm": 0.003434145823121071,
-      "learning_rate": 0.0001999953019508299,
-      "loss": 46.0,
-      "step": 40366
-    },
-    {
-      "epoch": 3.086339048492842,
-      "grad_norm": 0.0023876624181866646,
-      "learning_rate": 0.00019999530171799858,
-      "loss": 46.0,
-      "step": 40367
-    },
-    {
-      "epoch": 3.0864155054762312,
-      "grad_norm": 0.0014396545011550188,
-      "learning_rate": 0.00019999530148516147,
-      "loss": 46.0,
-      "step": 40368
-    },
-    {
-      "epoch": 3.086491962459621,
-      "grad_norm": 0.0015674050664529204,
-      "learning_rate": 0.00019999530125231862,
-      "loss": 46.0,
-      "step": 40369
-    },
-    {
-      "epoch": 3.0865684194430107,
-      "grad_norm": 0.0022144687827676535,
-      "learning_rate": 0.00019999530101947,
-      "loss": 46.0,
-      "step": 40370
-    },
-    {
-      "epoch": 3.0866448764264005,
-      "grad_norm": 0.0014323288341984153,
-      "learning_rate": 0.0001999953007866156,
-      "loss": 46.0,
-      "step": 40371
-    },
-    {
-      "epoch": 3.0867213334097903,
-      "grad_norm": 0.005419464781880379,
-      "learning_rate": 0.0001999953005537554,
-      "loss": 46.0,
-      "step": 40372
-    },
-    {
-      "epoch": 3.08679779039318,
-      "grad_norm": 0.0011133495718240738,
-      "learning_rate": 0.00019999530032088948,
-      "loss": 46.0,
-      "step": 40373
-    },
-    {
-      "epoch": 3.08687424737657,
-      "grad_norm": 0.0015525479102507234,
-      "learning_rate": 0.00019999530008801773,
-      "loss": 46.0,
-      "step": 40374
-    },
-    {
-      "epoch": 3.0869507043599596,
-      "grad_norm": 0.001559942844323814,
-      "learning_rate": 0.00019999529985514024,
-      "loss": 46.0,
-      "step": 40375
-    },
-    {
-      "epoch": 3.0870271613433493,
-      "grad_norm": 0.0014576624380424619,
-      "learning_rate": 0.000199995299622257,
-      "loss": 46.0,
-      "step": 40376
-    },
-    {
-      "epoch": 3.087103618326739,
-      "grad_norm": 0.0007901376811787486,
-      "learning_rate": 0.000199995299389368,
-      "loss": 46.0,
-      "step": 40377
-    },
-    {
-      "epoch": 3.087180075310129,
-      "grad_norm": 0.004678662866353989,
-      "learning_rate": 0.00019999529915647318,
-      "loss": 46.0,
-      "step": 40378
-    },
-    {
-      "epoch": 3.087256532293518,
-      "grad_norm": 0.0023544635623693466,
-      "learning_rate": 0.00019999529892357262,
-      "loss": 46.0,
-      "step": 40379
-    },
-    {
-      "epoch": 3.087332989276908,
-      "grad_norm": 0.0038698078133165836,
-      "learning_rate": 0.00019999529869066632,
-      "loss": 46.0,
-      "step": 40380
-    },
-    {
-      "epoch": 3.0874094462602977,
-      "grad_norm": 0.0006970096728764474,
-      "learning_rate": 0.0001999952984577542,
-      "loss": 46.0,
-      "step": 40381
-    },
-    {
-      "epoch": 3.0874859032436874,
-      "grad_norm": 0.0023959896061569452,
-      "learning_rate": 0.0001999952982248363,
-      "loss": 46.0,
-      "step": 40382
-    },
-    {
-      "epoch": 3.087562360227077,
-      "grad_norm": 0.003075435757637024,
-      "learning_rate": 0.00019999529799191268,
-      "loss": 46.0,
-      "step": 40383
-    },
-    {
-      "epoch": 3.087638817210467,
-      "grad_norm": 0.006638146471232176,
-      "learning_rate": 0.00019999529775898328,
-      "loss": 46.0,
-      "step": 40384
-    },
-    {
-      "epoch": 3.0877152741938567,
-      "grad_norm": 0.005111514590680599,
-      "learning_rate": 0.0001999952975260481,
-      "loss": 46.0,
-      "step": 40385
-    },
-    {
-      "epoch": 3.0877917311772465,
-      "grad_norm": 0.0010137214558199048,
-      "learning_rate": 0.00019999529729310716,
-      "loss": 46.0,
-      "step": 40386
-    },
-    {
-      "epoch": 3.0878681881606362,
-      "grad_norm": 0.0022372484672814608,
-      "learning_rate": 0.00019999529706016045,
-      "loss": 46.0,
-      "step": 40387
-    },
-    {
-      "epoch": 3.087944645144026,
-      "grad_norm": 0.004143637605011463,
-      "learning_rate": 0.00019999529682720798,
-      "loss": 46.0,
-      "step": 40388
-    },
-    {
-      "epoch": 3.0880211021274158,
-      "grad_norm": 0.0009678576025180519,
-      "learning_rate": 0.00019999529659424971,
-      "loss": 46.0,
-      "step": 40389
-    },
-    {
-      "epoch": 3.088097559110805,
-      "grad_norm": 0.0010885283118113875,
-      "learning_rate": 0.00019999529636128568,
-      "loss": 46.0,
-      "step": 40390
-    },
-    {
-      "epoch": 3.088174016094195,
-      "grad_norm": 0.003515888936817646,
-      "learning_rate": 0.0001999952961283159,
-      "loss": 46.0,
-      "step": 40391
-    },
-    {
-      "epoch": 3.0882504730775846,
-      "grad_norm": 0.0020951409824192524,
-      "learning_rate": 0.00019999529589534033,
-      "loss": 46.0,
-      "step": 40392
-    },
-    {
-      "epoch": 3.0883269300609744,
-      "grad_norm": 0.000930809008423239,
-      "learning_rate": 0.00019999529566235903,
-      "loss": 46.0,
-      "step": 40393
-    },
-    {
-      "epoch": 3.088403387044364,
-      "grad_norm": 0.0016526360996067524,
-      "learning_rate": 0.00019999529542937192,
-      "loss": 46.0,
-      "step": 40394
-    },
-    {
-      "epoch": 3.088479844027754,
-      "grad_norm": 0.0026023746468126774,
-      "learning_rate": 0.00019999529519637904,
-      "loss": 46.0,
-      "step": 40395
-    },
-    {
-      "epoch": 3.0885563010111436,
-      "grad_norm": 0.0027975630946457386,
-      "learning_rate": 0.00019999529496338042,
-      "loss": 46.0,
-      "step": 40396
-    },
-    {
-      "epoch": 3.0886327579945334,
-      "grad_norm": 0.0013455899897962809,
-      "learning_rate": 0.000199995294730376,
-      "loss": 46.0,
-      "step": 40397
-    },
-    {
-      "epoch": 3.088709214977923,
-      "grad_norm": 0.0019090435234829783,
-      "learning_rate": 0.00019999529449736582,
-      "loss": 46.0,
-      "step": 40398
-    },
-    {
-      "epoch": 3.088785671961313,
-      "grad_norm": 0.0005486995214596391,
-      "learning_rate": 0.00019999529426434988,
-      "loss": 46.0,
-      "step": 40399
-    },
-    {
-      "epoch": 3.0888621289447027,
-      "grad_norm": 0.004117876756936312,
-      "learning_rate": 0.00019999529403132816,
-      "loss": 46.0,
-      "step": 40400
-    },
-    {
-      "epoch": 3.088938585928092,
-      "grad_norm": 0.0032686328049749136,
-      "learning_rate": 0.0001999952937983007,
-      "loss": 46.0,
-      "step": 40401
-    },
-    {
-      "epoch": 3.0890150429114818,
-      "grad_norm": 0.0035158058162778616,
-      "learning_rate": 0.00019999529356526743,
-      "loss": 46.0,
-      "step": 40402
-    },
-    {
-      "epoch": 3.0890914998948715,
-      "grad_norm": 0.004890000447630882,
-      "learning_rate": 0.00019999529333222842,
-      "loss": 46.0,
-      "step": 40403
-    },
-    {
-      "epoch": 3.0891679568782613,
-      "grad_norm": 0.004670220892876387,
-      "learning_rate": 0.0001999952930991836,
-      "loss": 46.0,
-      "step": 40404
-    },
-    {
-      "epoch": 3.089244413861651,
-      "grad_norm": 0.0013623174745589495,
-      "learning_rate": 0.00019999529286613305,
-      "loss": 46.0,
-      "step": 40405
-    },
-    {
-      "epoch": 3.089320870845041,
-      "grad_norm": 0.005032017361372709,
-      "learning_rate": 0.00019999529263307672,
-      "loss": 46.0,
-      "step": 40406
-    },
-    {
-      "epoch": 3.0893973278284306,
-      "grad_norm": 0.01621577888727188,
-      "learning_rate": 0.0001999952924000146,
-      "loss": 46.0,
-      "step": 40407
-    },
-    {
-      "epoch": 3.0894737848118203,
-      "grad_norm": 0.0034426331985741854,
-      "learning_rate": 0.00019999529216694676,
-      "loss": 46.0,
-      "step": 40408
-    },
-    {
-      "epoch": 3.08955024179521,
-      "grad_norm": 0.0029967075679451227,
-      "learning_rate": 0.00019999529193387314,
-      "loss": 46.0,
-      "step": 40409
-    },
-    {
-      "epoch": 3.0896266987786,
-      "grad_norm": 0.0023160045966506004,
-      "learning_rate": 0.0001999952917007937,
-      "loss": 46.0,
-      "step": 40410
-    },
-    {
-      "epoch": 3.0897031557619896,
-      "grad_norm": 0.0014055754290893674,
-      "learning_rate": 0.00019999529146770857,
-      "loss": 46.0,
-      "step": 40411
-    },
-    {
-      "epoch": 3.089779612745379,
-      "grad_norm": 0.0037570511922240257,
-      "learning_rate": 0.0001999952912346176,
-      "loss": 46.0,
-      "step": 40412
-    },
-    {
-      "epoch": 3.0898560697287687,
-      "grad_norm": 0.007220292929559946,
-      "learning_rate": 0.0001999952910015209,
-      "loss": 46.0,
-      "step": 40413
-    },
-    {
-      "epoch": 3.0899325267121585,
-      "grad_norm": 0.0010095554171130061,
-      "learning_rate": 0.00019999529076841842,
-      "loss": 46.0,
-      "step": 40414
-    },
-    {
-      "epoch": 3.090008983695548,
-      "grad_norm": 0.00925536174327135,
-      "learning_rate": 0.00019999529053531018,
-      "loss": 46.0,
-      "step": 40415
-    },
-    {
-      "epoch": 3.090085440678938,
-      "grad_norm": 0.004097710829228163,
-      "learning_rate": 0.0001999952903021961,
-      "loss": 46.0,
-      "step": 40416
-    },
-    {
-      "epoch": 3.0901618976623277,
-      "grad_norm": 0.003129548393189907,
-      "learning_rate": 0.00019999529006907633,
-      "loss": 46.0,
-      "step": 40417
-    },
-    {
-      "epoch": 3.0902383546457175,
-      "grad_norm": 0.0015607367968186736,
-      "learning_rate": 0.0001999952898359508,
-      "loss": 46.0,
-      "step": 40418
-    },
-    {
-      "epoch": 3.0903148116291073,
-      "grad_norm": 0.0013155147898942232,
-      "learning_rate": 0.00019999528960281947,
-      "loss": 46.0,
-      "step": 40419
-    },
-    {
-      "epoch": 3.090391268612497,
-      "grad_norm": 0.0009990620892494917,
-      "learning_rate": 0.00019999528936968236,
-      "loss": 46.0,
-      "step": 40420
-    },
-    {
-      "epoch": 3.090467725595887,
-      "grad_norm": 0.0015065952902659774,
-      "learning_rate": 0.00019999528913653948,
-      "loss": 46.0,
-      "step": 40421
-    },
-    {
-      "epoch": 3.0905441825792765,
-      "grad_norm": 0.0021842126734554768,
-      "learning_rate": 0.0001999952889033909,
-      "loss": 46.0,
-      "step": 40422
-    },
-    {
-      "epoch": 3.090620639562666,
-      "grad_norm": 0.0026607741601765156,
-      "learning_rate": 0.00019999528867023644,
-      "loss": 46.0,
-      "step": 40423
-    },
-    {
-      "epoch": 3.0906970965460556,
-      "grad_norm": 0.004283210728317499,
-      "learning_rate": 0.0001999952884370763,
-      "loss": 46.0,
-      "step": 40424
-    },
-    {
-      "epoch": 3.0907735535294454,
-      "grad_norm": 0.0035749301314353943,
-      "learning_rate": 0.00019999528820391035,
-      "loss": 46.0,
-      "step": 40425
-    },
-    {
-      "epoch": 3.090850010512835,
-      "grad_norm": 0.0013756649568676949,
-      "learning_rate": 0.00019999528797073863,
-      "loss": 46.0,
-      "step": 40426
-    },
-    {
-      "epoch": 3.090926467496225,
-      "grad_norm": 0.0015635115560144186,
-      "learning_rate": 0.00019999528773756114,
-      "loss": 46.0,
-      "step": 40427
-    },
-    {
-      "epoch": 3.0910029244796147,
-      "grad_norm": 0.0023453370667994022,
-      "learning_rate": 0.00019999528750437787,
-      "loss": 46.0,
-      "step": 40428
-    },
-    {
-      "epoch": 3.0910793814630044,
-      "grad_norm": 0.0028312159702181816,
-      "learning_rate": 0.0001999952872711889,
-      "loss": 46.0,
-      "step": 40429
-    },
-    {
-      "epoch": 3.091155838446394,
-      "grad_norm": 0.004746130667626858,
-      "learning_rate": 0.00019999528703799408,
-      "loss": 46.0,
-      "step": 40430
-    },
-    {
-      "epoch": 3.091232295429784,
-      "grad_norm": 0.0032440631184726954,
-      "learning_rate": 0.00019999528680479352,
-      "loss": 46.0,
-      "step": 40431
-    },
-    {
-      "epoch": 3.0913087524131737,
-      "grad_norm": 0.002936146454885602,
-      "learning_rate": 0.0001999952865715872,
-      "loss": 46.0,
-      "step": 40432
-    },
-    {
-      "epoch": 3.0913852093965635,
-      "grad_norm": 0.0032761215697973967,
-      "learning_rate": 0.00019999528633837511,
-      "loss": 46.0,
-      "step": 40433
-    },
-    {
-      "epoch": 3.091461666379953,
-      "grad_norm": 0.0014739279868081212,
-      "learning_rate": 0.00019999528610515724,
-      "loss": 46.0,
-      "step": 40434
-    },
-    {
-      "epoch": 3.0915381233633425,
-      "grad_norm": 0.003040981711819768,
-      "learning_rate": 0.0001999952858719336,
-      "loss": 46.0,
-      "step": 40435
-    },
-    {
-      "epoch": 3.0916145803467323,
-      "grad_norm": 0.0017576297977939248,
-      "learning_rate": 0.0001999952856387042,
-      "loss": 46.0,
-      "step": 40436
-    },
-    {
-      "epoch": 3.091691037330122,
-      "grad_norm": 0.003477106336504221,
-      "learning_rate": 0.00019999528540546902,
-      "loss": 46.0,
-      "step": 40437
-    },
-    {
-      "epoch": 3.091767494313512,
-      "grad_norm": 0.0009176989551633596,
-      "learning_rate": 0.00019999528517222807,
-      "loss": 46.0,
-      "step": 40438
-    },
-    {
-      "epoch": 3.0918439512969016,
-      "grad_norm": 0.0019836705178022385,
-      "learning_rate": 0.00019999528493898136,
-      "loss": 46.0,
-      "step": 40439
-    },
-    {
-      "epoch": 3.0919204082802914,
-      "grad_norm": 0.0017348227556794882,
-      "learning_rate": 0.00019999528470572887,
-      "loss": 46.0,
-      "step": 40440
-    },
-    {
-      "epoch": 3.091996865263681,
-      "grad_norm": 0.003862952347844839,
-      "learning_rate": 0.00019999528447247066,
-      "loss": 46.0,
-      "step": 40441
-    },
-    {
-      "epoch": 3.092073322247071,
-      "grad_norm": 0.0046743908897042274,
-      "learning_rate": 0.00019999528423920665,
-      "loss": 46.0,
-      "step": 40442
-    },
-    {
-      "epoch": 3.0921497792304606,
-      "grad_norm": 0.0022646833676844835,
-      "learning_rate": 0.00019999528400593684,
-      "loss": 46.0,
-      "step": 40443
-    },
-    {
-      "epoch": 3.0922262362138504,
-      "grad_norm": 0.0016290085623040795,
-      "learning_rate": 0.00019999528377266128,
-      "loss": 46.0,
-      "step": 40444
-    },
-    {
-      "epoch": 3.0923026931972397,
-      "grad_norm": 0.003084276570007205,
-      "learning_rate": 0.00019999528353937995,
-      "loss": 46.0,
-      "step": 40445
-    },
-    {
-      "epoch": 3.0923791501806295,
-      "grad_norm": 0.0016924918163567781,
-      "learning_rate": 0.00019999528330609284,
-      "loss": 46.0,
-      "step": 40446
-    },
-    {
-      "epoch": 3.0924556071640192,
-      "grad_norm": 0.005916125141084194,
-      "learning_rate": 0.00019999528307279997,
-      "loss": 46.0,
-      "step": 40447
-    },
-    {
-      "epoch": 3.092532064147409,
-      "grad_norm": 0.0018066663760691881,
-      "learning_rate": 0.00019999528283950134,
-      "loss": 46.0,
-      "step": 40448
-    },
-    {
-      "epoch": 3.0926085211307988,
-      "grad_norm": 0.0005061979172751307,
-      "learning_rate": 0.00019999528260619695,
-      "loss": 46.0,
-      "step": 40449
-    },
-    {
-      "epoch": 3.0926849781141885,
-      "grad_norm": 0.0031908699311316013,
-      "learning_rate": 0.00019999528237288678,
-      "loss": 46.0,
-      "step": 40450
-    },
-    {
-      "epoch": 3.0927614350975783,
-      "grad_norm": 0.0023815890308469534,
-      "learning_rate": 0.00019999528213957083,
-      "loss": 46.0,
-      "step": 40451
-    },
-    {
-      "epoch": 3.092837892080968,
-      "grad_norm": 0.004095599055290222,
-      "learning_rate": 0.00019999528190624912,
-      "loss": 46.0,
-      "step": 40452
-    },
-    {
-      "epoch": 3.092914349064358,
-      "grad_norm": 0.011234685778617859,
-      "learning_rate": 0.00019999528167292163,
-      "loss": 46.0,
-      "step": 40453
-    },
-    {
-      "epoch": 3.0929908060477476,
-      "grad_norm": 0.005430188961327076,
-      "learning_rate": 0.00019999528143958842,
-      "loss": 46.0,
-      "step": 40454
-    },
-    {
-      "epoch": 3.093067263031137,
-      "grad_norm": 0.003194117220118642,
-      "learning_rate": 0.00019999528120624938,
-      "loss": 46.0,
-      "step": 40455
-    },
-    {
-      "epoch": 3.0931437200145266,
-      "grad_norm": 0.0016319730784744024,
-      "learning_rate": 0.0001999952809729046,
-      "loss": 46.0,
-      "step": 40456
-    },
-    {
-      "epoch": 3.0932201769979164,
-      "grad_norm": 0.0015326309949159622,
-      "learning_rate": 0.000199995280739554,
-      "loss": 46.0,
-      "step": 40457
-    },
-    {
-      "epoch": 3.093296633981306,
-      "grad_norm": 0.0029314046259969473,
-      "learning_rate": 0.0001999952805061977,
-      "loss": 46.0,
-      "step": 40458
-    },
-    {
-      "epoch": 3.093373090964696,
-      "grad_norm": 0.003492429619655013,
-      "learning_rate": 0.0001999952802728356,
-      "loss": 46.0,
-      "step": 40459
-    },
-    {
-      "epoch": 3.0934495479480857,
-      "grad_norm": 0.004148657899349928,
-      "learning_rate": 0.00019999528003946776,
-      "loss": 46.0,
-      "step": 40460
-    },
-    {
-      "epoch": 3.0935260049314754,
-      "grad_norm": 0.0038597926031798124,
-      "learning_rate": 0.0001999952798060941,
-      "loss": 46.0,
-      "step": 40461
-    },
-    {
-      "epoch": 3.093602461914865,
-      "grad_norm": 0.0046270452439785,
-      "learning_rate": 0.0001999952795727147,
-      "loss": 46.0,
-      "step": 40462
-    },
-    {
-      "epoch": 3.093678918898255,
-      "grad_norm": 0.0013523029629141092,
-      "learning_rate": 0.00019999527933932954,
-      "loss": 46.0,
-      "step": 40463
-    },
-    {
-      "epoch": 3.0937553758816447,
-      "grad_norm": 0.003149408381432295,
-      "learning_rate": 0.0001999952791059386,
-      "loss": 46.0,
-      "step": 40464
-    },
-    {
-      "epoch": 3.0938318328650345,
-      "grad_norm": 0.0022788397036492825,
-      "learning_rate": 0.00019999527887254188,
-      "loss": 46.0,
-      "step": 40465
-    },
-    {
-      "epoch": 3.0939082898484243,
-      "grad_norm": 0.006475043948739767,
-      "learning_rate": 0.00019999527863913942,
-      "loss": 46.0,
-      "step": 40466
-    },
-    {
-      "epoch": 3.0939847468318136,
-      "grad_norm": 0.002389322267845273,
-      "learning_rate": 0.00019999527840573115,
-      "loss": 46.0,
-      "step": 40467
-    },
-    {
-      "epoch": 3.0940612038152033,
-      "grad_norm": 0.0016802245518192649,
-      "learning_rate": 0.00019999527817231714,
-      "loss": 46.0,
-      "step": 40468
-    },
-    {
-      "epoch": 3.094137660798593,
-      "grad_norm": 0.002319866558536887,
-      "learning_rate": 0.00019999527793889733,
-      "loss": 46.0,
-      "step": 40469
-    },
-    {
-      "epoch": 3.094214117781983,
-      "grad_norm": 0.004884062334895134,
-      "learning_rate": 0.0001999952777054718,
-      "loss": 46.0,
-      "step": 40470
-    },
-    {
-      "epoch": 3.0942905747653726,
-      "grad_norm": 0.0015341215766966343,
-      "learning_rate": 0.00019999527747204047,
-      "loss": 46.0,
-      "step": 40471
-    },
-    {
-      "epoch": 3.0943670317487624,
-      "grad_norm": 0.0030958065763115883,
-      "learning_rate": 0.0001999952772386034,
-      "loss": 46.0,
-      "step": 40472
-    },
-    {
-      "epoch": 3.094443488732152,
-      "grad_norm": 0.0034751540515571833,
-      "learning_rate": 0.00019999527700516052,
-      "loss": 46.0,
-      "step": 40473
-    },
-    {
-      "epoch": 3.094519945715542,
-      "grad_norm": 0.0017490205354988575,
-      "learning_rate": 0.00019999527677171187,
-      "loss": 46.0,
-      "step": 40474
-    },
-    {
-      "epoch": 3.0945964026989317,
-      "grad_norm": 0.0012396344682201743,
-      "learning_rate": 0.00019999527653825747,
-      "loss": 46.0,
-      "step": 40475
-    },
-    {
-      "epoch": 3.0946728596823214,
-      "grad_norm": 0.0016794325783848763,
-      "learning_rate": 0.0001999952763047973,
-      "loss": 46.0,
-      "step": 40476
-    },
-    {
-      "epoch": 3.0947493166657107,
-      "grad_norm": 0.0033042433205991983,
-      "learning_rate": 0.00019999527607133136,
-      "loss": 46.0,
-      "step": 40477
-    },
-    {
-      "epoch": 3.0948257736491005,
-      "grad_norm": 0.0012796861119568348,
-      "learning_rate": 0.00019999527583785965,
-      "loss": 46.0,
-      "step": 40478
-    },
-    {
-      "epoch": 3.0949022306324903,
-      "grad_norm": 0.0008166969055309892,
-      "learning_rate": 0.00019999527560438218,
-      "loss": 46.0,
-      "step": 40479
-    },
-    {
-      "epoch": 3.09497868761588,
-      "grad_norm": 0.001249453634954989,
-      "learning_rate": 0.00019999527537089892,
-      "loss": 46.0,
-      "step": 40480
-    },
-    {
-      "epoch": 3.09505514459927,
-      "grad_norm": 0.006361784413456917,
-      "learning_rate": 0.0001999952751374099,
-      "loss": 46.0,
-      "step": 40481
-    },
-    {
-      "epoch": 3.0951316015826595,
-      "grad_norm": 0.0005817991332150996,
-      "learning_rate": 0.00019999527490391513,
-      "loss": 46.0,
-      "step": 40482
-    },
-    {
-      "epoch": 3.0952080585660493,
-      "grad_norm": 0.0021973224356770515,
-      "learning_rate": 0.00019999527467041457,
-      "loss": 46.0,
-      "step": 40483
-    },
-    {
-      "epoch": 3.095284515549439,
-      "grad_norm": 0.001319780945777893,
-      "learning_rate": 0.00019999527443690824,
-      "loss": 46.0,
-      "step": 40484
-    },
-    {
-      "epoch": 3.095360972532829,
-      "grad_norm": 0.004355449695140123,
-      "learning_rate": 0.00019999527420339614,
-      "loss": 46.0,
-      "step": 40485
-    },
-    {
-      "epoch": 3.0954374295162186,
-      "grad_norm": 0.0012352505000308156,
-      "learning_rate": 0.00019999527396987826,
-      "loss": 46.0,
-      "step": 40486
-    },
-    {
-      "epoch": 3.0955138864996083,
-      "grad_norm": 0.005594301037490368,
-      "learning_rate": 0.00019999527373635464,
-      "loss": 46.0,
-      "step": 40487
-    },
-    {
-      "epoch": 3.0955903434829977,
-      "grad_norm": 0.002911853604018688,
-      "learning_rate": 0.00019999527350282524,
-      "loss": 46.0,
-      "step": 40488
-    },
-    {
-      "epoch": 3.0956668004663874,
-      "grad_norm": 0.0021229595877230167,
-      "learning_rate": 0.00019999527326929008,
-      "loss": 46.0,
-      "step": 40489
-    },
-    {
-      "epoch": 3.095743257449777,
-      "grad_norm": 0.0018272852757945657,
-      "learning_rate": 0.00019999527303574913,
-      "loss": 46.0,
-      "step": 40490
-    },
-    {
-      "epoch": 3.095819714433167,
-      "grad_norm": 0.008383817039430141,
-      "learning_rate": 0.00019999527280220242,
-      "loss": 46.0,
-      "step": 40491
-    },
-    {
-      "epoch": 3.0958961714165567,
-      "grad_norm": 0.008791478350758553,
-      "learning_rate": 0.00019999527256864993,
-      "loss": 46.0,
-      "step": 40492
-    },
-    {
-      "epoch": 3.0959726283999465,
-      "grad_norm": 0.0010263097938150167,
-      "learning_rate": 0.0001999952723350917,
-      "loss": 46.0,
-      "step": 40493
-    },
-    {
-      "epoch": 3.0960490853833362,
-      "grad_norm": 0.0017859783256426454,
-      "learning_rate": 0.00019999527210152768,
-      "loss": 46.0,
-      "step": 40494
-    },
-    {
-      "epoch": 3.096125542366726,
-      "grad_norm": 0.0024867348838597536,
-      "learning_rate": 0.0001999952718679579,
-      "loss": 46.0,
-      "step": 40495
-    },
-    {
-      "epoch": 3.0962019993501158,
-      "grad_norm": 0.0018216186435893178,
-      "learning_rate": 0.00019999527163438234,
-      "loss": 46.0,
-      "step": 40496
-    },
-    {
-      "epoch": 3.0962784563335055,
-      "grad_norm": 0.001453619566746056,
-      "learning_rate": 0.00019999527140080102,
-      "loss": 46.0,
-      "step": 40497
-    },
-    {
-      "epoch": 3.0963549133168953,
-      "grad_norm": 0.0053321560844779015,
-      "learning_rate": 0.00019999527116721394,
-      "loss": 46.0,
-      "step": 40498
-    },
-    {
-      "epoch": 3.0964313703002846,
-      "grad_norm": 0.004216715693473816,
-      "learning_rate": 0.00019999527093362106,
-      "loss": 46.0,
-      "step": 40499
-    },
-    {
-      "epoch": 3.0965078272836744,
-      "grad_norm": 0.0029483388643711805,
-      "learning_rate": 0.00019999527070002241,
-      "loss": 46.0,
-      "step": 40500
-    },
-    {
-      "epoch": 3.096584284267064,
-      "grad_norm": 0.0028659417293965816,
-      "learning_rate": 0.00019999527046641805,
-      "loss": 46.0,
-      "step": 40501
-    },
-    {
-      "epoch": 3.096660741250454,
-      "grad_norm": 0.0012636876199394464,
-      "learning_rate": 0.00019999527023280785,
-      "loss": 46.0,
-      "step": 40502
-    },
-    {
-      "epoch": 3.0967371982338436,
-      "grad_norm": 0.001527530374005437,
-      "learning_rate": 0.0001999952699991919,
-      "loss": 46.0,
-      "step": 40503
-    },
-    {
-      "epoch": 3.0968136552172334,
-      "grad_norm": 0.0014225122286006808,
-      "learning_rate": 0.00019999526976557022,
-      "loss": 46.0,
-      "step": 40504
-    },
-    {
-      "epoch": 3.096890112200623,
-      "grad_norm": 0.0010207727318629622,
-      "learning_rate": 0.00019999526953194273,
-      "loss": 46.0,
-      "step": 40505
-    },
-    {
-      "epoch": 3.096966569184013,
-      "grad_norm": 0.0009988220408558846,
-      "learning_rate": 0.0001999952692983095,
-      "loss": 46.0,
-      "step": 40506
-    },
-    {
-      "epoch": 3.0970430261674027,
-      "grad_norm": 0.0025540627539157867,
-      "learning_rate": 0.00019999526906467049,
-      "loss": 46.0,
-      "step": 40507
-    },
-    {
-      "epoch": 3.0971194831507924,
-      "grad_norm": 0.0008108865586109459,
-      "learning_rate": 0.00019999526883102568,
-      "loss": 46.0,
-      "step": 40508
-    },
-    {
-      "epoch": 3.097195940134182,
-      "grad_norm": 0.0016464785439893603,
-      "learning_rate": 0.00019999526859737512,
-      "loss": 46.0,
-      "step": 40509
-    },
-    {
-      "epoch": 3.0972723971175715,
-      "grad_norm": 0.004571177065372467,
-      "learning_rate": 0.00019999526836371882,
-      "loss": 46.0,
-      "step": 40510
-    },
-    {
-      "epoch": 3.0973488541009613,
-      "grad_norm": 0.0011320440098643303,
-      "learning_rate": 0.00019999526813005672,
-      "loss": 46.0,
-      "step": 40511
-    },
-    {
-      "epoch": 3.097425311084351,
-      "grad_norm": 0.0017163710435852408,
-      "learning_rate": 0.00019999526789638887,
-      "loss": 46.0,
-      "step": 40512
-    },
-    {
-      "epoch": 3.097501768067741,
-      "grad_norm": 0.0050961934030056,
-      "learning_rate": 0.00019999526766271522,
-      "loss": 46.0,
-      "step": 40513
-    },
-    {
-      "epoch": 3.0975782250511306,
-      "grad_norm": 0.0007164383423514664,
-      "learning_rate": 0.00019999526742903582,
-      "loss": 46.0,
-      "step": 40514
-    },
-    {
-      "epoch": 3.0976546820345203,
-      "grad_norm": 0.0020096651278436184,
-      "learning_rate": 0.00019999526719535066,
-      "loss": 46.0,
-      "step": 40515
-    },
-    {
-      "epoch": 3.09773113901791,
-      "grad_norm": 0.0020939901005476713,
-      "learning_rate": 0.0001999952669616597,
-      "loss": 46.0,
-      "step": 40516
-    },
-    {
-      "epoch": 3.0978075960013,
-      "grad_norm": 0.0014223206089809537,
-      "learning_rate": 0.000199995266727963,
-      "loss": 46.0,
-      "step": 40517
-    },
-    {
-      "epoch": 3.0978840529846896,
-      "grad_norm": 0.0016256141243502498,
-      "learning_rate": 0.00019999526649426054,
-      "loss": 46.0,
-      "step": 40518
-    },
-    {
-      "epoch": 3.0979605099680794,
-      "grad_norm": 0.0021733916364610195,
-      "learning_rate": 0.0001999952662605523,
-      "loss": 46.0,
-      "step": 40519
-    },
-    {
-      "epoch": 3.098036966951469,
-      "grad_norm": 0.0027708441484719515,
-      "learning_rate": 0.00019999526602683827,
-      "loss": 46.0,
-      "step": 40520
-    },
-    {
-      "epoch": 3.0981134239348584,
-      "grad_norm": 0.002322721527889371,
-      "learning_rate": 0.00019999526579311846,
-      "loss": 46.0,
-      "step": 40521
-    },
-    {
-      "epoch": 3.098189880918248,
-      "grad_norm": 0.0012915402185171843,
-      "learning_rate": 0.00019999526555939293,
-      "loss": 46.0,
-      "step": 40522
-    },
-    {
-      "epoch": 3.098266337901638,
-      "grad_norm": 0.0008527606260031462,
-      "learning_rate": 0.0001999952653256616,
-      "loss": 46.0,
-      "step": 40523
-    },
-    {
-      "epoch": 3.0983427948850277,
-      "grad_norm": 0.0011782728834077716,
-      "learning_rate": 0.00019999526509192453,
-      "loss": 46.0,
-      "step": 40524
-    },
-    {
-      "epoch": 3.0984192518684175,
-      "grad_norm": 0.0006229756399989128,
-      "learning_rate": 0.00019999526485818165,
-      "loss": 46.0,
-      "step": 40525
-    },
-    {
-      "epoch": 3.0984957088518073,
-      "grad_norm": 0.0014685136266052723,
-      "learning_rate": 0.000199995264624433,
-      "loss": 46.0,
-      "step": 40526
-    },
-    {
-      "epoch": 3.098572165835197,
-      "grad_norm": 0.005406105425208807,
-      "learning_rate": 0.0001999952643906786,
-      "loss": 46.0,
-      "step": 40527
-    },
-    {
-      "epoch": 3.0986486228185868,
-      "grad_norm": 0.002447728766128421,
-      "learning_rate": 0.00019999526415691844,
-      "loss": 46.0,
-      "step": 40528
-    },
-    {
-      "epoch": 3.0987250798019765,
-      "grad_norm": 0.00209886790253222,
-      "learning_rate": 0.0001999952639231525,
-      "loss": 46.0,
-      "step": 40529
-    },
-    {
-      "epoch": 3.0988015367853663,
-      "grad_norm": 0.001424482325091958,
-      "learning_rate": 0.0001999952636893808,
-      "loss": 46.0,
-      "step": 40530
-    },
-    {
-      "epoch": 3.098877993768756,
-      "grad_norm": 0.0012581193586811423,
-      "learning_rate": 0.00019999526345560332,
-      "loss": 46.0,
-      "step": 40531
-    },
-    {
-      "epoch": 3.0989544507521454,
-      "grad_norm": 0.00382399745285511,
-      "learning_rate": 0.00019999526322182009,
-      "loss": 46.0,
-      "step": 40532
-    },
-    {
-      "epoch": 3.099030907735535,
-      "grad_norm": 0.0009712508181110024,
-      "learning_rate": 0.00019999526298803105,
-      "loss": 46.0,
-      "step": 40533
-    },
-    {
-      "epoch": 3.099107364718925,
-      "grad_norm": 0.0017928526503965259,
-      "learning_rate": 0.0001999952627542363,
-      "loss": 46.0,
-      "step": 40534
-    },
-    {
-      "epoch": 3.0991838217023147,
-      "grad_norm": 0.006536139640957117,
-      "learning_rate": 0.0001999952625204357,
-      "loss": 46.0,
-      "step": 40535
-    },
-    {
-      "epoch": 3.0992602786857044,
-      "grad_norm": 0.002144876169040799,
-      "learning_rate": 0.0001999952622866294,
-      "loss": 46.0,
-      "step": 40536
-    },
-    {
-      "epoch": 3.099336735669094,
-      "grad_norm": 0.0006981620681472123,
-      "learning_rate": 0.00019999526205281734,
-      "loss": 46.0,
-      "step": 40537
-    },
-    {
-      "epoch": 3.099413192652484,
-      "grad_norm": 0.00507147703319788,
-      "learning_rate": 0.00019999526181899946,
-      "loss": 46.0,
-      "step": 40538
-    },
-    {
-      "epoch": 3.0994896496358737,
-      "grad_norm": 0.005660662893205881,
-      "learning_rate": 0.00019999526158517584,
-      "loss": 46.0,
-      "step": 40539
-    },
-    {
-      "epoch": 3.0995661066192635,
-      "grad_norm": 0.0018913400126621127,
-      "learning_rate": 0.00019999526135134642,
-      "loss": 46.0,
-      "step": 40540
-    },
-    {
-      "epoch": 3.0996425636026532,
-      "grad_norm": 0.000940806174185127,
-      "learning_rate": 0.00019999526111751128,
-      "loss": 46.0,
-      "step": 40541
-    },
-    {
-      "epoch": 3.099719020586043,
-      "grad_norm": 0.0008229814120568335,
-      "learning_rate": 0.00019999526088367034,
-      "loss": 46.0,
-      "step": 40542
-    },
-    {
-      "epoch": 3.0997954775694323,
-      "grad_norm": 0.006658149883151054,
-      "learning_rate": 0.00019999526064982362,
-      "loss": 46.0,
-      "step": 40543
-    },
-    {
-      "epoch": 3.099871934552822,
-      "grad_norm": 0.0036576061975210905,
-      "learning_rate": 0.00019999526041597113,
-      "loss": 46.0,
-      "step": 40544
-    },
-    {
-      "epoch": 3.099948391536212,
-      "grad_norm": 0.0005766819813288748,
-      "learning_rate": 0.0001999952601821129,
-      "loss": 46.0,
-      "step": 40545
-    },
-    {
-      "epoch": 3.1000248485196016,
-      "grad_norm": 0.0007805295172147453,
-      "learning_rate": 0.0001999952599482489,
-      "loss": 46.0,
-      "step": 40546
-    },
-    {
-      "epoch": 3.1001013055029913,
-      "grad_norm": 0.004046564921736717,
-      "learning_rate": 0.00019999525971437908,
-      "loss": 46.0,
-      "step": 40547
-    },
-    {
-      "epoch": 3.100177762486381,
-      "grad_norm": 0.0019012964330613613,
-      "learning_rate": 0.00019999525948050356,
-      "loss": 46.0,
-      "step": 40548
-    },
-    {
-      "epoch": 3.100254219469771,
-      "grad_norm": 0.002129658591002226,
-      "learning_rate": 0.00019999525924662223,
-      "loss": 46.0,
-      "step": 40549
-    },
-    {
-      "epoch": 3.1003306764531606,
-      "grad_norm": 0.0013813786208629608,
-      "learning_rate": 0.00019999525901273515,
-      "loss": 46.0,
-      "step": 40550
-    },
-    {
-      "epoch": 3.1004071334365504,
-      "grad_norm": 0.00272749038413167,
-      "learning_rate": 0.00019999525877884228,
-      "loss": 46.0,
-      "step": 40551
-    },
-    {
-      "epoch": 3.10048359041994,
-      "grad_norm": 0.0021689089480787516,
-      "learning_rate": 0.00019999525854494366,
-      "loss": 46.0,
-      "step": 40552
-    },
-    {
-      "epoch": 3.10056004740333,
-      "grad_norm": 0.003072606399655342,
-      "learning_rate": 0.00019999525831103924,
-      "loss": 46.0,
-      "step": 40553
-    },
-    {
-      "epoch": 3.1006365043867192,
-      "grad_norm": 0.0011073439382016659,
-      "learning_rate": 0.00019999525807712907,
-      "loss": 46.0,
-      "step": 40554
-    },
-    {
-      "epoch": 3.100712961370109,
-      "grad_norm": 0.001202999847009778,
-      "learning_rate": 0.00019999525784321316,
-      "loss": 46.0,
-      "step": 40555
-    },
-    {
-      "epoch": 3.1007894183534987,
-      "grad_norm": 0.001623639720492065,
-      "learning_rate": 0.00019999525760929144,
-      "loss": 46.0,
-      "step": 40556
-    },
-    {
-      "epoch": 3.1008658753368885,
-      "grad_norm": 0.0010263341246172786,
-      "learning_rate": 0.00019999525737536398,
-      "loss": 46.0,
-      "step": 40557
-    },
-    {
-      "epoch": 3.1009423323202783,
-      "grad_norm": 0.002977109281346202,
-      "learning_rate": 0.0001999952571414307,
-      "loss": 46.0,
-      "step": 40558
-    },
-    {
-      "epoch": 3.101018789303668,
-      "grad_norm": 0.002048192312940955,
-      "learning_rate": 0.0001999952569074917,
-      "loss": 46.0,
-      "step": 40559
-    },
-    {
-      "epoch": 3.101095246287058,
-      "grad_norm": 0.0014675704296678305,
-      "learning_rate": 0.00019999525667354693,
-      "loss": 46.0,
-      "step": 40560
-    },
-    {
-      "epoch": 3.1011717032704476,
-      "grad_norm": 0.002621354768052697,
-      "learning_rate": 0.00019999525643959638,
-      "loss": 46.0,
-      "step": 40561
-    },
-    {
-      "epoch": 3.1012481602538373,
-      "grad_norm": 0.0018998316954821348,
-      "learning_rate": 0.00019999525620564005,
-      "loss": 46.0,
-      "step": 40562
-    },
-    {
-      "epoch": 3.101324617237227,
-      "grad_norm": 0.0030819622334092855,
-      "learning_rate": 0.00019999525597167798,
-      "loss": 46.0,
-      "step": 40563
-    },
-    {
-      "epoch": 3.101401074220617,
-      "grad_norm": 0.0031770700588822365,
-      "learning_rate": 0.0001999952557377101,
-      "loss": 46.0,
-      "step": 40564
-    },
-    {
-      "epoch": 3.101477531204006,
-      "grad_norm": 0.002916044322773814,
-      "learning_rate": 0.00019999525550373645,
-      "loss": 46.0,
-      "step": 40565
-    },
-    {
-      "epoch": 3.101553988187396,
-      "grad_norm": 0.006006535608321428,
-      "learning_rate": 0.00019999525526975709,
-      "loss": 46.0,
-      "step": 40566
-    },
-    {
-      "epoch": 3.1016304451707857,
-      "grad_norm": 0.003432487603276968,
-      "learning_rate": 0.0001999952550357719,
-      "loss": 46.0,
-      "step": 40567
-    },
-    {
-      "epoch": 3.1017069021541754,
-      "grad_norm": 0.0015815269434824586,
-      "learning_rate": 0.00019999525480178098,
-      "loss": 46.0,
-      "step": 40568
-    },
-    {
-      "epoch": 3.101783359137565,
-      "grad_norm": 0.00358053226955235,
-      "learning_rate": 0.00019999525456778426,
-      "loss": 46.0,
-      "step": 40569
-    },
-    {
-      "epoch": 3.101859816120955,
-      "grad_norm": 0.0019013695418834686,
-      "learning_rate": 0.00019999525433378178,
-      "loss": 46.0,
-      "step": 40570
-    },
-    {
-      "epoch": 3.1019362731043447,
-      "grad_norm": 0.0020740365143865347,
-      "learning_rate": 0.00019999525409977354,
-      "loss": 46.0,
-      "step": 40571
-    },
-    {
-      "epoch": 3.1020127300877345,
-      "grad_norm": 0.004211660008877516,
-      "learning_rate": 0.00019999525386575954,
-      "loss": 46.0,
-      "step": 40572
-    },
-    {
-      "epoch": 3.1020891870711242,
-      "grad_norm": 0.0012021627044305205,
-      "learning_rate": 0.00019999525363173975,
-      "loss": 46.0,
-      "step": 40573
-    },
-    {
-      "epoch": 3.102165644054514,
-      "grad_norm": 0.0023867697454988956,
-      "learning_rate": 0.0001999952533977142,
-      "loss": 46.0,
-      "step": 40574
-    },
-    {
-      "epoch": 3.1022421010379038,
-      "grad_norm": 0.00216542836278677,
-      "learning_rate": 0.00019999525316368287,
-      "loss": 46.0,
-      "step": 40575
-    },
-    {
-      "epoch": 3.102318558021293,
-      "grad_norm": 0.003351530758664012,
-      "learning_rate": 0.00019999525292964577,
-      "loss": 46.0,
-      "step": 40576
-    },
-    {
-      "epoch": 3.102395015004683,
-      "grad_norm": 0.0019315776880830526,
-      "learning_rate": 0.00019999525269560293,
-      "loss": 46.0,
-      "step": 40577
-    },
-    {
-      "epoch": 3.1024714719880726,
-      "grad_norm": 0.001947465119883418,
-      "learning_rate": 0.0001999952524615543,
-      "loss": 46.0,
-      "step": 40578
-    },
-    {
-      "epoch": 3.1025479289714624,
-      "grad_norm": 0.004189606290310621,
-      "learning_rate": 0.0001999952522274999,
-      "loss": 46.0,
-      "step": 40579
-    },
-    {
-      "epoch": 3.102624385954852,
-      "grad_norm": 0.007601381279528141,
-      "learning_rate": 0.00019999525199343974,
-      "loss": 46.0,
-      "step": 40580
-    },
-    {
-      "epoch": 3.102700842938242,
-      "grad_norm": 0.003251432441174984,
-      "learning_rate": 0.0001999952517593738,
-      "loss": 46.0,
-      "step": 40581
-    },
-    {
-      "epoch": 3.1027772999216316,
-      "grad_norm": 0.003809809684753418,
-      "learning_rate": 0.0001999952515253021,
-      "loss": 46.0,
-      "step": 40582
-    },
-    {
-      "epoch": 3.1028537569050214,
-      "grad_norm": 0.0019049829570576549,
-      "learning_rate": 0.00019999525129122463,
-      "loss": 46.0,
-      "step": 40583
-    },
-    {
-      "epoch": 3.102930213888411,
-      "grad_norm": 0.0034392185043543577,
-      "learning_rate": 0.0001999952510571414,
-      "loss": 46.0,
-      "step": 40584
-    },
-    {
-      "epoch": 3.103006670871801,
-      "grad_norm": 0.005755885969847441,
-      "learning_rate": 0.00019999525082305236,
-      "loss": 46.0,
-      "step": 40585
-    },
-    {
-      "epoch": 3.1030831278551902,
-      "grad_norm": 0.003156736958771944,
-      "learning_rate": 0.0001999952505889576,
-      "loss": 46.0,
-      "step": 40586
-    },
-    {
-      "epoch": 3.10315958483858,
-      "grad_norm": 0.0035740917082875967,
-      "learning_rate": 0.00019999525035485706,
-      "loss": 46.0,
-      "step": 40587
-    },
-    {
-      "epoch": 3.1032360418219698,
-      "grad_norm": 0.0016834683483466506,
-      "learning_rate": 0.0001999952501207507,
-      "loss": 46.0,
-      "step": 40588
-    },
-    {
-      "epoch": 3.1033124988053595,
-      "grad_norm": 0.0015105652855709195,
-      "learning_rate": 0.00019999524988663866,
-      "loss": 46.0,
-      "step": 40589
-    },
-    {
-      "epoch": 3.1033889557887493,
-      "grad_norm": 0.0024993799161165953,
-      "learning_rate": 0.00019999524965252076,
-      "loss": 46.0,
-      "step": 40590
-    },
-    {
-      "epoch": 3.103465412772139,
-      "grad_norm": 0.0016338367713615298,
-      "learning_rate": 0.00019999524941839716,
-      "loss": 46.0,
-      "step": 40591
-    },
-    {
-      "epoch": 3.103541869755529,
-      "grad_norm": 0.002314062789082527,
-      "learning_rate": 0.00019999524918426774,
-      "loss": 46.0,
-      "step": 40592
-    },
-    {
-      "epoch": 3.1036183267389186,
-      "grad_norm": 0.001856429735198617,
-      "learning_rate": 0.00019999524895013257,
-      "loss": 46.0,
-      "step": 40593
-    },
-    {
-      "epoch": 3.1036947837223083,
-      "grad_norm": 0.004054618068039417,
-      "learning_rate": 0.00019999524871599163,
-      "loss": 46.0,
-      "step": 40594
-    },
-    {
-      "epoch": 3.103771240705698,
-      "grad_norm": 0.00089570990530774,
-      "learning_rate": 0.00019999524848184495,
-      "loss": 46.0,
-      "step": 40595
-    },
-    {
-      "epoch": 3.103847697689088,
-      "grad_norm": 0.0005263558705337346,
-      "learning_rate": 0.00019999524824769246,
-      "loss": 46.0,
-      "step": 40596
-    },
-    {
-      "epoch": 3.1039241546724776,
-      "grad_norm": 0.005276402924209833,
-      "learning_rate": 0.00019999524801353423,
-      "loss": 46.0,
-      "step": 40597
-    },
-    {
-      "epoch": 3.104000611655867,
-      "grad_norm": 0.0033934905659407377,
-      "learning_rate": 0.00019999524777937022,
-      "loss": 46.0,
-      "step": 40598
-    },
-    {
-      "epoch": 3.1040770686392567,
-      "grad_norm": 0.001445717178285122,
-      "learning_rate": 0.00019999524754520044,
-      "loss": 46.0,
-      "step": 40599
-    },
-    {
-      "epoch": 3.1041535256226465,
-      "grad_norm": 0.002149093197658658,
-      "learning_rate": 0.0001999952473110249,
-      "loss": 46.0,
-      "step": 40600
-    },
-    {
-      "epoch": 3.104229982606036,
-      "grad_norm": 0.002808294026181102,
-      "learning_rate": 0.0001999952470768436,
-      "loss": 46.0,
-      "step": 40601
-    },
-    {
-      "epoch": 3.104306439589426,
-      "grad_norm": 0.006283879745751619,
-      "learning_rate": 0.00019999524684265646,
-      "loss": 46.0,
-      "step": 40602
-    },
-    {
-      "epoch": 3.1043828965728157,
-      "grad_norm": 0.003297331975772977,
-      "learning_rate": 0.00019999524660846362,
-      "loss": 46.0,
-      "step": 40603
-    },
-    {
-      "epoch": 3.1044593535562055,
-      "grad_norm": 0.002959726145491004,
-      "learning_rate": 0.00019999524637426502,
-      "loss": 46.0,
-      "step": 40604
-    },
-    {
-      "epoch": 3.1045358105395953,
-      "grad_norm": 0.008936835452914238,
-      "learning_rate": 0.0001999952461400606,
-      "loss": 46.0,
-      "step": 40605
-    },
-    {
-      "epoch": 3.104612267522985,
-      "grad_norm": 0.0012401411077007651,
-      "learning_rate": 0.00019999524590585044,
-      "loss": 46.0,
-      "step": 40606
-    },
-    {
-      "epoch": 3.104688724506375,
-      "grad_norm": 0.0013057544128969312,
-      "learning_rate": 0.0001999952456716345,
-      "loss": 46.0,
-      "step": 40607
-    },
-    {
-      "epoch": 3.104765181489764,
-      "grad_norm": 0.0019892703276127577,
-      "learning_rate": 0.00019999524543741278,
-      "loss": 46.0,
-      "step": 40608
-    },
-    {
-      "epoch": 3.104841638473154,
-      "grad_norm": 0.0016813121037557721,
-      "learning_rate": 0.00019999524520318532,
-      "loss": 46.0,
-      "step": 40609
-    },
-    {
-      "epoch": 3.1049180954565436,
-      "grad_norm": 0.0012155581498518586,
-      "learning_rate": 0.0001999952449689521,
-      "loss": 46.0,
-      "step": 40610
-    },
-    {
-      "epoch": 3.1049945524399334,
-      "grad_norm": 0.00477614626288414,
-      "learning_rate": 0.00019999524473471308,
-      "loss": 46.0,
-      "step": 40611
-    },
-    {
-      "epoch": 3.105071009423323,
-      "grad_norm": 0.0007503448869101703,
-      "learning_rate": 0.0001999952445004683,
-      "loss": 46.0,
-      "step": 40612
-    },
-    {
-      "epoch": 3.105147466406713,
-      "grad_norm": 0.0016403916524723172,
-      "learning_rate": 0.00019999524426621775,
-      "loss": 46.0,
-      "step": 40613
-    },
-    {
-      "epoch": 3.1052239233901027,
-      "grad_norm": 0.0030978990253061056,
-      "learning_rate": 0.00019999524403196143,
-      "loss": 46.0,
-      "step": 40614
-    },
-    {
-      "epoch": 3.1053003803734924,
-      "grad_norm": 0.007497414946556091,
-      "learning_rate": 0.00019999524379769933,
-      "loss": 46.0,
-      "step": 40615
-    },
-    {
-      "epoch": 3.105376837356882,
-      "grad_norm": 0.005553377326577902,
-      "learning_rate": 0.00019999524356343148,
-      "loss": 46.0,
-      "step": 40616
-    },
-    {
-      "epoch": 3.105453294340272,
-      "grad_norm": 0.001269626198336482,
-      "learning_rate": 0.00019999524332915786,
-      "loss": 46.0,
-      "step": 40617
-    },
-    {
-      "epoch": 3.1055297513236617,
-      "grad_norm": 0.002937605371698737,
-      "learning_rate": 0.00019999524309487847,
-      "loss": 46.0,
-      "step": 40618
-    },
-    {
-      "epoch": 3.105606208307051,
-      "grad_norm": 0.0006505833589471877,
-      "learning_rate": 0.0001999952428605933,
-      "loss": 46.0,
-      "step": 40619
-    },
-    {
-      "epoch": 3.105682665290441,
-      "grad_norm": 0.0027430334594100714,
-      "learning_rate": 0.00019999524262630236,
-      "loss": 46.0,
-      "step": 40620
-    },
-    {
-      "epoch": 3.1057591222738306,
-      "grad_norm": 0.003604454919695854,
-      "learning_rate": 0.00019999524239200565,
-      "loss": 46.0,
-      "step": 40621
-    },
-    {
-      "epoch": 3.1058355792572203,
-      "grad_norm": 0.002649285364896059,
-      "learning_rate": 0.0001999952421577032,
-      "loss": 46.0,
-      "step": 40622
-    },
-    {
-      "epoch": 3.10591203624061,
-      "grad_norm": 0.002324703149497509,
-      "learning_rate": 0.00019999524192339493,
-      "loss": 46.0,
-      "step": 40623
-    },
-    {
-      "epoch": 3.105988493224,
-      "grad_norm": 0.002125852508470416,
-      "learning_rate": 0.00019999524168908095,
-      "loss": 46.0,
-      "step": 40624
-    },
-    {
-      "epoch": 3.1060649502073896,
-      "grad_norm": 0.0008553141378797591,
-      "learning_rate": 0.00019999524145476115,
-      "loss": 46.0,
-      "step": 40625
-    },
-    {
-      "epoch": 3.1061414071907794,
-      "grad_norm": 0.0005615248228423297,
-      "learning_rate": 0.0001999952412204356,
-      "loss": 46.0,
-      "step": 40626
-    },
-    {
-      "epoch": 3.106217864174169,
-      "grad_norm": 0.005933989305049181,
-      "learning_rate": 0.0001999952409861043,
-      "loss": 46.0,
-      "step": 40627
-    },
-    {
-      "epoch": 3.106294321157559,
-      "grad_norm": 0.002101338002830744,
-      "learning_rate": 0.00019999524075176722,
-      "loss": 46.0,
-      "step": 40628
-    },
-    {
-      "epoch": 3.1063707781409486,
-      "grad_norm": 0.003139087464660406,
-      "learning_rate": 0.00019999524051742435,
-      "loss": 46.0,
-      "step": 40629
-    },
-    {
-      "epoch": 3.106447235124338,
-      "grad_norm": 0.0021737574134021997,
-      "learning_rate": 0.00019999524028307573,
-      "loss": 46.0,
-      "step": 40630
-    },
-    {
-      "epoch": 3.1065236921077277,
-      "grad_norm": 0.0021124924533069134,
-      "learning_rate": 0.0001999952400487213,
-      "loss": 46.0,
-      "step": 40631
-    },
-    {
-      "epoch": 3.1066001490911175,
-      "grad_norm": 0.003720023902133107,
-      "learning_rate": 0.00019999523981436117,
-      "loss": 46.0,
-      "step": 40632
-    },
-    {
-      "epoch": 3.1066766060745072,
-      "grad_norm": 0.0028368530329316854,
-      "learning_rate": 0.00019999523957999523,
-      "loss": 46.0,
-      "step": 40633
-    },
-    {
-      "epoch": 3.106753063057897,
-      "grad_norm": 0.001612384687177837,
-      "learning_rate": 0.00019999523934562352,
-      "loss": 46.0,
-      "step": 40634
-    },
-    {
-      "epoch": 3.1068295200412868,
-      "grad_norm": 0.0019780707079917192,
-      "learning_rate": 0.00019999523911124606,
-      "loss": 46.0,
-      "step": 40635
-    },
-    {
-      "epoch": 3.1069059770246765,
-      "grad_norm": 0.0016828691586852074,
-      "learning_rate": 0.0001999952388768628,
-      "loss": 46.0,
-      "step": 40636
-    },
-    {
-      "epoch": 3.1069824340080663,
-      "grad_norm": 0.0014801541110500693,
-      "learning_rate": 0.0001999952386424738,
-      "loss": 46.0,
-      "step": 40637
-    },
-    {
-      "epoch": 3.107058890991456,
-      "grad_norm": 0.0014360109344124794,
-      "learning_rate": 0.00019999523840807904,
-      "loss": 46.0,
-      "step": 40638
-    },
-    {
-      "epoch": 3.107135347974846,
-      "grad_norm": 0.0019825368653982878,
-      "learning_rate": 0.0001999952381736785,
-      "loss": 46.0,
-      "step": 40639
-    },
-    {
-      "epoch": 3.1072118049582356,
-      "grad_norm": 0.0019437994342297316,
-      "learning_rate": 0.00019999523793927217,
-      "loss": 46.0,
-      "step": 40640
-    },
-    {
-      "epoch": 3.107288261941625,
-      "grad_norm": 0.002874209778383374,
-      "learning_rate": 0.0001999952377048601,
-      "loss": 46.0,
-      "step": 40641
-    },
-    {
-      "epoch": 3.1073647189250146,
-      "grad_norm": 0.0027536386623978615,
-      "learning_rate": 0.00019999523747044222,
-      "loss": 46.0,
-      "step": 40642
-    },
-    {
-      "epoch": 3.1074411759084044,
-      "grad_norm": 0.0038181981071829796,
-      "learning_rate": 0.00019999523723601858,
-      "loss": 46.0,
-      "step": 40643
-    },
-    {
-      "epoch": 3.107517632891794,
-      "grad_norm": 0.0011698815505951643,
-      "learning_rate": 0.0001999952370015892,
-      "loss": 46.0,
-      "step": 40644
-    },
-    {
-      "epoch": 3.107594089875184,
-      "grad_norm": 0.0035587139427661896,
-      "learning_rate": 0.00019999523676715404,
-      "loss": 46.0,
-      "step": 40645
-    },
-    {
-      "epoch": 3.1076705468585737,
-      "grad_norm": 0.002951499540358782,
-      "learning_rate": 0.0001999952365327131,
-      "loss": 46.0,
-      "step": 40646
-    },
-    {
-      "epoch": 3.1077470038419635,
-      "grad_norm": 0.0009942209580913186,
-      "learning_rate": 0.0001999952362982664,
-      "loss": 46.0,
-      "step": 40647
-    },
-    {
-      "epoch": 3.107823460825353,
-      "grad_norm": 0.0037157328333705664,
-      "learning_rate": 0.00019999523606381394,
-      "loss": 46.0,
-      "step": 40648
-    },
-    {
-      "epoch": 3.107899917808743,
-      "grad_norm": 0.0010292684892192483,
-      "learning_rate": 0.0001999952358293557,
-      "loss": 46.0,
-      "step": 40649
-    },
-    {
-      "epoch": 3.1079763747921327,
-      "grad_norm": 0.0030907404143363237,
-      "learning_rate": 0.00019999523559489167,
-      "loss": 46.0,
-      "step": 40650
-    },
-    {
-      "epoch": 3.1080528317755225,
-      "grad_norm": 0.002052318537607789,
-      "learning_rate": 0.00019999523536042192,
-      "loss": 46.0,
-      "step": 40651
-    },
-    {
-      "epoch": 3.108129288758912,
-      "grad_norm": 0.0037924996577203274,
-      "learning_rate": 0.00019999523512594637,
-      "loss": 46.0,
-      "step": 40652
-    },
-    {
-      "epoch": 3.1082057457423016,
-      "grad_norm": 0.0031955670565366745,
-      "learning_rate": 0.00019999523489146504,
-      "loss": 46.0,
-      "step": 40653
-    },
-    {
-      "epoch": 3.1082822027256913,
-      "grad_norm": 0.001090130303055048,
-      "learning_rate": 0.00019999523465697797,
-      "loss": 46.0,
-      "step": 40654
-    },
-    {
-      "epoch": 3.108358659709081,
-      "grad_norm": 0.0028760689310729504,
-      "learning_rate": 0.0001999952344224851,
-      "loss": 46.0,
-      "step": 40655
-    },
-    {
-      "epoch": 3.108435116692471,
-      "grad_norm": 0.0033978205174207687,
-      "learning_rate": 0.00019999523418798648,
-      "loss": 46.0,
-      "step": 40656
-    },
-    {
-      "epoch": 3.1085115736758606,
-      "grad_norm": 0.0033070696517825127,
-      "learning_rate": 0.0001999952339534821,
-      "loss": 46.0,
-      "step": 40657
-    },
-    {
-      "epoch": 3.1085880306592504,
-      "grad_norm": 0.006725492887198925,
-      "learning_rate": 0.00019999523371897192,
-      "loss": 46.0,
-      "step": 40658
-    },
-    {
-      "epoch": 3.10866448764264,
-      "grad_norm": 0.0012557096779346466,
-      "learning_rate": 0.00019999523348445598,
-      "loss": 46.0,
-      "step": 40659
-    },
-    {
-      "epoch": 3.10874094462603,
-      "grad_norm": 0.0029580688569694757,
-      "learning_rate": 0.00019999523324993427,
-      "loss": 46.0,
-      "step": 40660
-    },
-    {
-      "epoch": 3.1088174016094197,
-      "grad_norm": 0.0006876569823361933,
-      "learning_rate": 0.00019999523301540681,
-      "loss": 46.0,
-      "step": 40661
-    },
-    {
-      "epoch": 3.1088938585928094,
-      "grad_norm": 0.0005584590835496783,
-      "learning_rate": 0.00019999523278087358,
-      "loss": 46.0,
-      "step": 40662
-    },
-    {
-      "epoch": 3.1089703155761987,
-      "grad_norm": 0.002697932068258524,
-      "learning_rate": 0.00019999523254633458,
-      "loss": 46.0,
-      "step": 40663
-    },
-    {
-      "epoch": 3.1090467725595885,
-      "grad_norm": 0.0016806249041110277,
-      "learning_rate": 0.0001999952323117898,
-      "loss": 46.0,
-      "step": 40664
-    },
-    {
-      "epoch": 3.1091232295429783,
-      "grad_norm": 0.0006992561975494027,
-      "learning_rate": 0.00019999523207723925,
-      "loss": 46.0,
-      "step": 40665
-    },
-    {
-      "epoch": 3.109199686526368,
-      "grad_norm": 0.001335737993940711,
-      "learning_rate": 0.00019999523184268295,
-      "loss": 46.0,
-      "step": 40666
-    },
-    {
-      "epoch": 3.109276143509758,
-      "grad_norm": 0.0028802230954170227,
-      "learning_rate": 0.00019999523160812082,
-      "loss": 46.0,
-      "step": 40667
-    },
-    {
-      "epoch": 3.1093526004931475,
-      "grad_norm": 0.003597601316869259,
-      "learning_rate": 0.000199995231373553,
-      "loss": 46.0,
-      "step": 40668
-    },
-    {
-      "epoch": 3.1094290574765373,
-      "grad_norm": 0.0017050397582352161,
-      "learning_rate": 0.0001999952311389794,
-      "loss": 46.0,
-      "step": 40669
-    },
-    {
-      "epoch": 3.109505514459927,
-      "grad_norm": 0.0031466397922486067,
-      "learning_rate": 0.00019999523090439997,
-      "loss": 46.0,
-      "step": 40670
-    },
-    {
-      "epoch": 3.109581971443317,
-      "grad_norm": 0.0017802175134420395,
-      "learning_rate": 0.00019999523066981483,
-      "loss": 46.0,
-      "step": 40671
-    },
-    {
-      "epoch": 3.1096584284267066,
-      "grad_norm": 0.0017646894557401538,
-      "learning_rate": 0.0001999952304352239,
-      "loss": 46.0,
-      "step": 40672
-    },
-    {
-      "epoch": 3.1097348854100964,
-      "grad_norm": 0.002124791033565998,
-      "learning_rate": 0.0001999952302006272,
-      "loss": 46.0,
-      "step": 40673
-    },
-    {
-      "epoch": 3.1098113423934857,
-      "grad_norm": 0.001128863194026053,
-      "learning_rate": 0.00019999522996602472,
-      "loss": 46.0,
-      "step": 40674
-    },
-    {
-      "epoch": 3.1098877993768754,
-      "grad_norm": 0.002025673631578684,
-      "learning_rate": 0.0001999952297314165,
-      "loss": 46.0,
-      "step": 40675
-    },
-    {
-      "epoch": 3.109964256360265,
-      "grad_norm": 0.002854999853298068,
-      "learning_rate": 0.00019999522949680246,
-      "loss": 46.0,
-      "step": 40676
-    },
-    {
-      "epoch": 3.110040713343655,
-      "grad_norm": 0.0013628213200718164,
-      "learning_rate": 0.00019999522926218268,
-      "loss": 46.0,
-      "step": 40677
-    },
-    {
-      "epoch": 3.1101171703270447,
-      "grad_norm": 0.0012093354016542435,
-      "learning_rate": 0.00019999522902755716,
-      "loss": 46.0,
-      "step": 40678
-    },
-    {
-      "epoch": 3.1101936273104345,
-      "grad_norm": 0.005360200069844723,
-      "learning_rate": 0.00019999522879292583,
-      "loss": 46.0,
-      "step": 40679
-    },
-    {
-      "epoch": 3.1102700842938242,
-      "grad_norm": 0.0011868528090417385,
-      "learning_rate": 0.00019999522855828876,
-      "loss": 46.0,
-      "step": 40680
-    },
-    {
-      "epoch": 3.110346541277214,
-      "grad_norm": 0.0019066107925027609,
-      "learning_rate": 0.0001999952283236459,
-      "loss": 46.0,
-      "step": 40681
-    },
-    {
-      "epoch": 3.1104229982606038,
-      "grad_norm": 0.004260065965354443,
-      "learning_rate": 0.00019999522808899727,
-      "loss": 46.0,
-      "step": 40682
-    },
-    {
-      "epoch": 3.1104994552439935,
-      "grad_norm": 0.0013711127685382962,
-      "learning_rate": 0.00019999522785434288,
-      "loss": 46.0,
-      "step": 40683
-    },
-    {
-      "epoch": 3.1105759122273833,
-      "grad_norm": 0.0011432197643443942,
-      "learning_rate": 0.00019999522761968272,
-      "loss": 46.0,
-      "step": 40684
-    },
-    {
-      "epoch": 3.1106523692107726,
-      "grad_norm": 0.0011039115488529205,
-      "learning_rate": 0.00019999522738501678,
-      "loss": 46.0,
-      "step": 40685
-    },
-    {
-      "epoch": 3.1107288261941624,
-      "grad_norm": 0.005655019544064999,
-      "learning_rate": 0.0001999952271503451,
-      "loss": 46.0,
-      "step": 40686
-    },
-    {
-      "epoch": 3.110805283177552,
-      "grad_norm": 0.0069869086146354675,
-      "learning_rate": 0.0001999952269156676,
-      "loss": 46.0,
-      "step": 40687
-    },
-    {
-      "epoch": 3.110881740160942,
-      "grad_norm": 0.002772802719846368,
-      "learning_rate": 0.0001999952266809844,
-      "loss": 46.0,
-      "step": 40688
-    },
-    {
-      "epoch": 3.1109581971443316,
-      "grad_norm": 0.0025198734365403652,
-      "learning_rate": 0.00019999522644629538,
-      "loss": 46.0,
-      "step": 40689
-    },
-    {
-      "epoch": 3.1110346541277214,
-      "grad_norm": 0.0028384372126311064,
-      "learning_rate": 0.0001999952262116006,
-      "loss": 46.0,
-      "step": 40690
-    },
-    {
-      "epoch": 3.111111111111111,
-      "grad_norm": 0.0017654254334047437,
-      "learning_rate": 0.00019999522597690005,
-      "loss": 46.0,
-      "step": 40691
-    },
-    {
-      "epoch": 3.111187568094501,
-      "grad_norm": 0.0012556869769468904,
-      "learning_rate": 0.00019999522574219375,
-      "loss": 46.0,
-      "step": 40692
-    },
-    {
-      "epoch": 3.1112640250778907,
-      "grad_norm": 0.0017929719761013985,
-      "learning_rate": 0.00019999522550748168,
-      "loss": 46.0,
-      "step": 40693
-    },
-    {
-      "epoch": 3.1113404820612804,
-      "grad_norm": 0.0007313725654967129,
-      "learning_rate": 0.0001999952252727638,
-      "loss": 46.0,
-      "step": 40694
-    },
-    {
-      "epoch": 3.11141693904467,
-      "grad_norm": 0.007251699920743704,
-      "learning_rate": 0.0001999952250380402,
-      "loss": 46.0,
-      "step": 40695
-    },
-    {
-      "epoch": 3.1114933960280595,
-      "grad_norm": 0.003814192721620202,
-      "learning_rate": 0.0001999952248033108,
-      "loss": 46.0,
-      "step": 40696
-    },
-    {
-      "epoch": 3.1115698530114493,
-      "grad_norm": 0.003523855237290263,
-      "learning_rate": 0.00019999522456857563,
-      "loss": 46.0,
-      "step": 40697
-    },
-    {
-      "epoch": 3.111646309994839,
-      "grad_norm": 0.0033209652174264193,
-      "learning_rate": 0.0001999952243338347,
-      "loss": 46.0,
-      "step": 40698
-    },
-    {
-      "epoch": 3.111722766978229,
-      "grad_norm": 0.0007878456963226199,
-      "learning_rate": 0.000199995224099088,
-      "loss": 46.0,
-      "step": 40699
-    },
-    {
-      "epoch": 3.1117992239616186,
-      "grad_norm": 0.0023557180538773537,
-      "learning_rate": 0.00019999522386433553,
-      "loss": 46.0,
-      "step": 40700
-    },
-    {
-      "epoch": 3.1118756809450083,
-      "grad_norm": 0.0024816084187477827,
-      "learning_rate": 0.0001999952236295773,
-      "loss": 46.0,
-      "step": 40701
-    },
-    {
-      "epoch": 3.111952137928398,
-      "grad_norm": 0.006478719413280487,
-      "learning_rate": 0.0001999952233948133,
-      "loss": 46.0,
-      "step": 40702
-    },
-    {
-      "epoch": 3.112028594911788,
-      "grad_norm": 0.002110905945301056,
-      "learning_rate": 0.00019999522316004354,
-      "loss": 46.0,
-      "step": 40703
-    },
-    {
-      "epoch": 3.1121050518951776,
-      "grad_norm": 0.0008606757037341595,
-      "learning_rate": 0.00019999522292526797,
-      "loss": 46.0,
-      "step": 40704
-    },
-    {
-      "epoch": 3.1121815088785674,
-      "grad_norm": 0.00407033646479249,
-      "learning_rate": 0.00019999522269048664,
-      "loss": 46.0,
-      "step": 40705
-    },
-    {
-      "epoch": 3.112257965861957,
-      "grad_norm": 0.0019873501732945442,
-      "learning_rate": 0.00019999522245569957,
-      "loss": 46.0,
-      "step": 40706
-    },
-    {
-      "epoch": 3.1123344228453464,
-      "grad_norm": 0.0008969728951342404,
-      "learning_rate": 0.00019999522222090673,
-      "loss": 46.0,
-      "step": 40707
-    },
-    {
-      "epoch": 3.112410879828736,
-      "grad_norm": 0.002647097222507,
-      "learning_rate": 0.00019999522198610808,
-      "loss": 46.0,
-      "step": 40708
-    },
-    {
-      "epoch": 3.112487336812126,
-      "grad_norm": 0.00447895796969533,
-      "learning_rate": 0.00019999522175130372,
-      "loss": 46.0,
-      "step": 40709
-    },
-    {
-      "epoch": 3.1125637937955157,
-      "grad_norm": 0.0007037890027277172,
-      "learning_rate": 0.00019999522151649358,
-      "loss": 46.0,
-      "step": 40710
-    },
-    {
-      "epoch": 3.1126402507789055,
-      "grad_norm": 0.008124771527945995,
-      "learning_rate": 0.00019999522128167762,
-      "loss": 46.0,
-      "step": 40711
-    },
-    {
-      "epoch": 3.1127167077622953,
-      "grad_norm": 0.002086327178403735,
-      "learning_rate": 0.00019999522104685593,
-      "loss": 46.0,
-      "step": 40712
-    },
-    {
-      "epoch": 3.112793164745685,
-      "grad_norm": 0.0023307630326598883,
-      "learning_rate": 0.00019999522081202845,
-      "loss": 46.0,
-      "step": 40713
-    },
-    {
-      "epoch": 3.1128696217290748,
-      "grad_norm": 0.001804125146009028,
-      "learning_rate": 0.00019999522057719525,
-      "loss": 46.0,
-      "step": 40714
-    },
-    {
-      "epoch": 3.1129460787124645,
-      "grad_norm": 0.0064630708657205105,
-      "learning_rate": 0.00019999522034235622,
-      "loss": 46.0,
-      "step": 40715
-    },
-    {
-      "epoch": 3.1130225356958543,
-      "grad_norm": 0.002073155250400305,
-      "learning_rate": 0.00019999522010751144,
-      "loss": 46.0,
-      "step": 40716
-    },
-    {
-      "epoch": 3.1130989926792436,
-      "grad_norm": 0.0022630509920418262,
-      "learning_rate": 0.0001999952198726609,
-      "loss": 46.0,
-      "step": 40717
-    },
-    {
-      "epoch": 3.1131754496626334,
-      "grad_norm": 0.0018022538861259818,
-      "learning_rate": 0.0001999952196378046,
-      "loss": 46.0,
-      "step": 40718
-    },
-    {
-      "epoch": 3.113251906646023,
-      "grad_norm": 0.0019590011797845364,
-      "learning_rate": 0.0001999952194029425,
-      "loss": 46.0,
-      "step": 40719
-    },
-    {
-      "epoch": 3.113328363629413,
-      "grad_norm": 0.0020214931573718786,
-      "learning_rate": 0.00019999521916807465,
-      "loss": 46.0,
-      "step": 40720
-    },
-    {
-      "epoch": 3.1134048206128027,
-      "grad_norm": 0.004375392571091652,
-      "learning_rate": 0.00019999521893320103,
-      "loss": 46.0,
-      "step": 40721
-    },
-    {
-      "epoch": 3.1134812775961924,
-      "grad_norm": 0.0005054162465967238,
-      "learning_rate": 0.00019999521869832164,
-      "loss": 46.0,
-      "step": 40722
-    },
-    {
-      "epoch": 3.113557734579582,
-      "grad_norm": 0.0036657429300248623,
-      "learning_rate": 0.00019999521846343648,
-      "loss": 46.0,
-      "step": 40723
-    },
-    {
-      "epoch": 3.113634191562972,
-      "grad_norm": 0.002865151735022664,
-      "learning_rate": 0.00019999521822854557,
-      "loss": 46.0,
-      "step": 40724
-    },
-    {
-      "epoch": 3.1137106485463617,
-      "grad_norm": 0.0017647017957642674,
-      "learning_rate": 0.00019999521799364886,
-      "loss": 46.0,
-      "step": 40725
-    },
-    {
-      "epoch": 3.1137871055297515,
-      "grad_norm": 0.007046884391456842,
-      "learning_rate": 0.00019999521775874638,
-      "loss": 46.0,
-      "step": 40726
-    },
-    {
-      "epoch": 3.1138635625131412,
-      "grad_norm": 0.0011433225590735674,
-      "learning_rate": 0.00019999521752383817,
-      "loss": 46.0,
-      "step": 40727
-    },
-    {
-      "epoch": 3.113940019496531,
-      "grad_norm": 0.0022675630170851946,
-      "learning_rate": 0.00019999521728892417,
-      "loss": 46.0,
-      "step": 40728
-    },
-    {
-      "epoch": 3.1140164764799203,
-      "grad_norm": 0.0009418230038136244,
-      "learning_rate": 0.0001999952170540044,
-      "loss": 46.0,
-      "step": 40729
-    },
-    {
-      "epoch": 3.11409293346331,
-      "grad_norm": 0.0017789386911317706,
-      "learning_rate": 0.00019999521681907884,
-      "loss": 46.0,
-      "step": 40730
-    },
-    {
-      "epoch": 3.1141693904467,
-      "grad_norm": 0.0031167122069746256,
-      "learning_rate": 0.00019999521658414752,
-      "loss": 46.0,
-      "step": 40731
-    },
-    {
-      "epoch": 3.1142458474300896,
-      "grad_norm": 0.0010709302732720971,
-      "learning_rate": 0.00019999521634921042,
-      "loss": 46.0,
-      "step": 40732
-    },
-    {
-      "epoch": 3.1143223044134793,
-      "grad_norm": 0.002567109651863575,
-      "learning_rate": 0.00019999521611426758,
-      "loss": 46.0,
-      "step": 40733
-    },
-    {
-      "epoch": 3.114398761396869,
-      "grad_norm": 0.0034765906166285276,
-      "learning_rate": 0.00019999521587931896,
-      "loss": 46.0,
-      "step": 40734
-    },
-    {
-      "epoch": 3.114475218380259,
-      "grad_norm": 0.001815799972973764,
-      "learning_rate": 0.00019999521564436457,
-      "loss": 46.0,
-      "step": 40735
-    },
-    {
-      "epoch": 3.1145516753636486,
-      "grad_norm": 0.002282122615724802,
-      "learning_rate": 0.00019999521540940444,
-      "loss": 46.0,
-      "step": 40736
-    },
-    {
-      "epoch": 3.1146281323470384,
-      "grad_norm": 0.0017280846368521452,
-      "learning_rate": 0.0001999952151744385,
-      "loss": 46.0,
-      "step": 40737
-    },
-    {
-      "epoch": 3.114704589330428,
-      "grad_norm": 0.004034288693219423,
-      "learning_rate": 0.0001999952149394668,
-      "loss": 46.0,
-      "step": 40738
-    },
-    {
-      "epoch": 3.1147810463138175,
-      "grad_norm": 0.0018747354624792933,
-      "learning_rate": 0.0001999952147044893,
-      "loss": 46.0,
-      "step": 40739
-    },
-    {
-      "epoch": 3.1148575032972072,
-      "grad_norm": 0.005246356129646301,
-      "learning_rate": 0.00019999521446950608,
-      "loss": 46.0,
-      "step": 40740
-    },
-    {
-      "epoch": 3.114933960280597,
-      "grad_norm": 0.001763629843480885,
-      "learning_rate": 0.00019999521423451707,
-      "loss": 46.0,
-      "step": 40741
-    },
-    {
-      "epoch": 3.1150104172639868,
-      "grad_norm": 0.0014435260090976954,
-      "learning_rate": 0.00019999521399952232,
-      "loss": 46.0,
-      "step": 40742
-    },
-    {
-      "epoch": 3.1150868742473765,
-      "grad_norm": 0.0017951888730749488,
-      "learning_rate": 0.0001999952137645218,
-      "loss": 46.0,
-      "step": 40743
-    },
-    {
-      "epoch": 3.1151633312307663,
-      "grad_norm": 0.0013720939168706536,
-      "learning_rate": 0.00019999521352951548,
-      "loss": 46.0,
-      "step": 40744
-    },
-    {
-      "epoch": 3.115239788214156,
-      "grad_norm": 0.0026262635365128517,
-      "learning_rate": 0.00019999521329450338,
-      "loss": 46.0,
-      "step": 40745
-    },
-    {
-      "epoch": 3.115316245197546,
-      "grad_norm": 0.008479884825646877,
-      "learning_rate": 0.00019999521305948554,
-      "loss": 46.0,
-      "step": 40746
-    },
-    {
-      "epoch": 3.1153927021809356,
-      "grad_norm": 0.0025631135795265436,
-      "learning_rate": 0.00019999521282446192,
-      "loss": 46.0,
-      "step": 40747
-    },
-    {
-      "epoch": 3.1154691591643253,
-      "grad_norm": 0.00232143048197031,
-      "learning_rate": 0.00019999521258943253,
-      "loss": 46.0,
-      "step": 40748
-    },
-    {
-      "epoch": 3.115545616147715,
-      "grad_norm": 0.0007593531627207994,
-      "learning_rate": 0.00019999521235439737,
-      "loss": 46.0,
-      "step": 40749
-    },
-    {
-      "epoch": 3.115622073131105,
-      "grad_norm": 0.0025446987710893154,
-      "learning_rate": 0.00019999521211935646,
-      "loss": 46.0,
-      "step": 40750
-    },
-    {
-      "epoch": 3.115698530114494,
-      "grad_norm": 0.0024156621657311916,
-      "learning_rate": 0.00019999521188430975,
-      "loss": 46.0,
-      "step": 40751
-    },
-    {
-      "epoch": 3.115774987097884,
-      "grad_norm": 0.0049401200376451015,
-      "learning_rate": 0.0001999952116492573,
-      "loss": 46.0,
-      "step": 40752
-    },
-    {
-      "epoch": 3.1158514440812737,
-      "grad_norm": 0.0026664440520107746,
-      "learning_rate": 0.00019999521141419907,
-      "loss": 46.0,
-      "step": 40753
-    },
-    {
-      "epoch": 3.1159279010646634,
-      "grad_norm": 0.0008086326415650547,
-      "learning_rate": 0.00019999521117913504,
-      "loss": 46.0,
-      "step": 40754
-    },
-    {
-      "epoch": 3.116004358048053,
-      "grad_norm": 0.0016082170186564326,
-      "learning_rate": 0.00019999521094406526,
-      "loss": 46.0,
-      "step": 40755
-    },
-    {
-      "epoch": 3.116080815031443,
-      "grad_norm": 0.008245814591646194,
-      "learning_rate": 0.00019999521070898974,
-      "loss": 46.0,
-      "step": 40756
-    },
-    {
-      "epoch": 3.1161572720148327,
-      "grad_norm": 0.0016008797101676464,
-      "learning_rate": 0.00019999521047390841,
-      "loss": 46.0,
-      "step": 40757
-    },
-    {
-      "epoch": 3.1162337289982225,
-      "grad_norm": 0.0031871472019702196,
-      "learning_rate": 0.00019999521023882135,
-      "loss": 46.0,
-      "step": 40758
-    },
-    {
-      "epoch": 3.1163101859816122,
-      "grad_norm": 0.003929438069462776,
-      "learning_rate": 0.0001999952100037285,
-      "loss": 46.0,
-      "step": 40759
-    },
-    {
-      "epoch": 3.116386642965002,
-      "grad_norm": 0.0037591878790408373,
-      "learning_rate": 0.00019999520976862986,
-      "loss": 46.0,
-      "step": 40760
-    },
-    {
-      "epoch": 3.1164630999483913,
-      "grad_norm": 0.0014754737494513392,
-      "learning_rate": 0.0001999952095335255,
-      "loss": 46.0,
-      "step": 40761
-    },
-    {
-      "epoch": 3.116539556931781,
-      "grad_norm": 0.0016094903694465756,
-      "learning_rate": 0.00019999520929841534,
-      "loss": 46.0,
-      "step": 40762
-    },
-    {
-      "epoch": 3.116616013915171,
-      "grad_norm": 0.00698675774037838,
-      "learning_rate": 0.0001999952090632994,
-      "loss": 46.0,
-      "step": 40763
-    },
-    {
-      "epoch": 3.1166924708985606,
-      "grad_norm": 0.0006853318773210049,
-      "learning_rate": 0.00019999520882817772,
-      "loss": 46.0,
-      "step": 40764
-    },
-    {
-      "epoch": 3.1167689278819504,
-      "grad_norm": 0.001623383373953402,
-      "learning_rate": 0.00019999520859305023,
-      "loss": 46.0,
-      "step": 40765
-    },
-    {
-      "epoch": 3.11684538486534,
-      "grad_norm": 0.002321155508980155,
-      "learning_rate": 0.000199995208357917,
-      "loss": 46.0,
-      "step": 40766
-    },
-    {
-      "epoch": 3.11692184184873,
-      "grad_norm": 0.0050252932123839855,
-      "learning_rate": 0.000199995208122778,
-      "loss": 46.0,
-      "step": 40767
-    },
-    {
-      "epoch": 3.1169982988321197,
-      "grad_norm": 0.0022112191654741764,
-      "learning_rate": 0.00019999520788763323,
-      "loss": 46.0,
-      "step": 40768
-    },
-    {
-      "epoch": 3.1170747558155094,
-      "grad_norm": 0.0007991396123543382,
-      "learning_rate": 0.0001999952076524827,
-      "loss": 46.0,
-      "step": 40769
-    },
-    {
-      "epoch": 3.117151212798899,
-      "grad_norm": 0.0017596872057765722,
-      "learning_rate": 0.00019999520741732638,
-      "loss": 46.0,
-      "step": 40770
-    },
-    {
-      "epoch": 3.117227669782289,
-      "grad_norm": 0.0005868226289749146,
-      "learning_rate": 0.00019999520718216429,
-      "loss": 46.0,
-      "step": 40771
-    },
-    {
-      "epoch": 3.1173041267656783,
-      "grad_norm": 0.0016553538152948022,
-      "learning_rate": 0.00019999520694699644,
-      "loss": 46.0,
-      "step": 40772
-    },
-    {
-      "epoch": 3.117380583749068,
-      "grad_norm": 0.001902634627185762,
-      "learning_rate": 0.00019999520671182283,
-      "loss": 46.0,
-      "step": 40773
-    },
-    {
-      "epoch": 3.1174570407324578,
-      "grad_norm": 0.002324906410649419,
-      "learning_rate": 0.00019999520647664344,
-      "loss": 46.0,
-      "step": 40774
-    },
-    {
-      "epoch": 3.1175334977158475,
-      "grad_norm": 0.004304792732000351,
-      "learning_rate": 0.00019999520624145828,
-      "loss": 46.0,
-      "step": 40775
-    },
-    {
-      "epoch": 3.1176099546992373,
-      "grad_norm": 0.0024911880027502775,
-      "learning_rate": 0.00019999520600626737,
-      "loss": 46.0,
-      "step": 40776
-    },
-    {
-      "epoch": 3.117686411682627,
-      "grad_norm": 0.002355963457375765,
-      "learning_rate": 0.00019999520577107066,
-      "loss": 46.0,
-      "step": 40777
-    },
-    {
-      "epoch": 3.117762868666017,
-      "grad_norm": 0.004058195743709803,
-      "learning_rate": 0.0001999952055358682,
-      "loss": 46.0,
-      "step": 40778
-    },
-    {
-      "epoch": 3.1178393256494066,
-      "grad_norm": 0.001629286096431315,
-      "learning_rate": 0.00019999520530065998,
-      "loss": 46.0,
-      "step": 40779
-    },
-    {
-      "epoch": 3.1179157826327963,
-      "grad_norm": 0.003608867060393095,
-      "learning_rate": 0.00019999520506544597,
-      "loss": 46.0,
-      "step": 40780
-    },
-    {
-      "epoch": 3.117992239616186,
-      "grad_norm": 0.006915109232068062,
-      "learning_rate": 0.0001999952048302262,
-      "loss": 46.0,
-      "step": 40781
-    },
-    {
-      "epoch": 3.118068696599576,
-      "grad_norm": 0.0018111106473952532,
-      "learning_rate": 0.00019999520459500065,
-      "loss": 46.0,
-      "step": 40782
-    },
-    {
-      "epoch": 3.118145153582965,
-      "grad_norm": 0.0013533896999433637,
-      "learning_rate": 0.00019999520435976935,
-      "loss": 46.0,
-      "step": 40783
-    },
-    {
-      "epoch": 3.118221610566355,
-      "grad_norm": 0.001836710493080318,
-      "learning_rate": 0.00019999520412453228,
-      "loss": 46.0,
-      "step": 40784
-    },
-    {
-      "epoch": 3.1182980675497447,
-      "grad_norm": 0.000987344072200358,
-      "learning_rate": 0.00019999520388928942,
-      "loss": 46.0,
-      "step": 40785
-    },
-    {
-      "epoch": 3.1183745245331345,
-      "grad_norm": 0.005836552008986473,
-      "learning_rate": 0.0001999952036540408,
-      "loss": 46.0,
-      "step": 40786
-    },
-    {
-      "epoch": 3.1184509815165242,
-      "grad_norm": 0.0006390998023562133,
-      "learning_rate": 0.0001999952034187864,
-      "loss": 46.0,
-      "step": 40787
-    },
-    {
-      "epoch": 3.118527438499914,
-      "grad_norm": 0.007171757984906435,
-      "learning_rate": 0.00019999520318352628,
-      "loss": 46.0,
-      "step": 40788
-    },
-    {
-      "epoch": 3.1186038954833037,
-      "grad_norm": 0.0007172286277636886,
-      "learning_rate": 0.00019999520294826034,
-      "loss": 46.0,
-      "step": 40789
-    },
-    {
-      "epoch": 3.1186803524666935,
-      "grad_norm": 0.002585702110081911,
-      "learning_rate": 0.00019999520271298863,
-      "loss": 46.0,
-      "step": 40790
-    },
-    {
-      "epoch": 3.1187568094500833,
-      "grad_norm": 0.0023589550983160734,
-      "learning_rate": 0.0001999952024777112,
-      "loss": 46.0,
-      "step": 40791
-    },
-    {
-      "epoch": 3.118833266433473,
-      "grad_norm": 0.002266060560941696,
-      "learning_rate": 0.00019999520224242795,
-      "loss": 46.0,
-      "step": 40792
-    },
-    {
-      "epoch": 3.118909723416863,
-      "grad_norm": 0.0011684981873258948,
-      "learning_rate": 0.00019999520200713895,
-      "loss": 46.0,
-      "step": 40793
-    },
-    {
-      "epoch": 3.118986180400252,
-      "grad_norm": 0.0021452929358929396,
-      "learning_rate": 0.0001999952017718442,
-      "loss": 46.0,
-      "step": 40794
-    },
-    {
-      "epoch": 3.119062637383642,
-      "grad_norm": 0.0008237732108682394,
-      "learning_rate": 0.00019999520153654365,
-      "loss": 46.0,
-      "step": 40795
-    },
-    {
-      "epoch": 3.1191390943670316,
-      "grad_norm": 0.0013906615786254406,
-      "learning_rate": 0.00019999520130123736,
-      "loss": 46.0,
-      "step": 40796
-    },
-    {
-      "epoch": 3.1192155513504214,
-      "grad_norm": 0.001321022748015821,
-      "learning_rate": 0.00019999520106592526,
-      "loss": 46.0,
-      "step": 40797
-    },
-    {
-      "epoch": 3.119292008333811,
-      "grad_norm": 0.002199542010203004,
-      "learning_rate": 0.00019999520083060742,
-      "loss": 46.0,
-      "step": 40798
-    },
-    {
-      "epoch": 3.119368465317201,
-      "grad_norm": 0.001273911795578897,
-      "learning_rate": 0.00019999520059528378,
-      "loss": 46.0,
-      "step": 40799
-    },
-    {
-      "epoch": 3.1194449223005907,
-      "grad_norm": 0.001626364653930068,
-      "learning_rate": 0.00019999520035995442,
-      "loss": 46.0,
-      "step": 40800
-    },
-    {
-      "epoch": 3.1195213792839804,
-      "grad_norm": 0.0016366582131013274,
-      "learning_rate": 0.00019999520012461925,
-      "loss": 46.0,
-      "step": 40801
-    },
-    {
-      "epoch": 3.11959783626737,
-      "grad_norm": 0.0027048932388424873,
-      "learning_rate": 0.00019999519988927832,
-      "loss": 46.0,
-      "step": 40802
-    },
-    {
-      "epoch": 3.11967429325076,
-      "grad_norm": 0.0011805471731349826,
-      "learning_rate": 0.00019999519965393164,
-      "loss": 46.0,
-      "step": 40803
-    },
-    {
-      "epoch": 3.1197507502341497,
-      "grad_norm": 0.0026091975159943104,
-      "learning_rate": 0.00019999519941857918,
-      "loss": 46.0,
-      "step": 40804
-    },
-    {
-      "epoch": 3.119827207217539,
-      "grad_norm": 0.0029221978038549423,
-      "learning_rate": 0.00019999519918322093,
-      "loss": 46.0,
-      "step": 40805
-    },
-    {
-      "epoch": 3.119903664200929,
-      "grad_norm": 0.001681748777627945,
-      "learning_rate": 0.00019999519894785695,
-      "loss": 46.0,
-      "step": 40806
-    },
-    {
-      "epoch": 3.1199801211843186,
-      "grad_norm": 0.003822098020464182,
-      "learning_rate": 0.00019999519871248718,
-      "loss": 46.0,
-      "step": 40807
-    },
-    {
-      "epoch": 3.1200565781677083,
-      "grad_norm": 0.003669238882139325,
-      "learning_rate": 0.00019999519847711163,
-      "loss": 46.0,
-      "step": 40808
-    },
-    {
-      "epoch": 3.120133035151098,
-      "grad_norm": 0.002753482200205326,
-      "learning_rate": 0.0001999951982417303,
-      "loss": 46.0,
-      "step": 40809
-    },
-    {
-      "epoch": 3.120209492134488,
-      "grad_norm": 0.002198306145146489,
-      "learning_rate": 0.00019999519800634327,
-      "loss": 46.0,
-      "step": 40810
-    },
-    {
-      "epoch": 3.1202859491178776,
-      "grad_norm": 0.00133318523876369,
-      "learning_rate": 0.0001999951977709504,
-      "loss": 46.0,
-      "step": 40811
-    },
-    {
-      "epoch": 3.1203624061012674,
-      "grad_norm": 0.002639633836224675,
-      "learning_rate": 0.00019999519753555178,
-      "loss": 46.0,
-      "step": 40812
-    },
-    {
-      "epoch": 3.120438863084657,
-      "grad_norm": 0.0007897822069935501,
-      "learning_rate": 0.0001999951973001474,
-      "loss": 46.0,
-      "step": 40813
-    },
-    {
-      "epoch": 3.120515320068047,
-      "grad_norm": 0.009138131514191628,
-      "learning_rate": 0.00019999519706473726,
-      "loss": 46.0,
-      "step": 40814
-    },
-    {
-      "epoch": 3.1205917770514366,
-      "grad_norm": 0.001905362936668098,
-      "learning_rate": 0.0001999951968293213,
-      "loss": 46.0,
-      "step": 40815
-    },
-    {
-      "epoch": 3.120668234034826,
-      "grad_norm": 0.0012523388722911477,
-      "learning_rate": 0.00019999519659389962,
-      "loss": 46.0,
-      "step": 40816
-    },
-    {
-      "epoch": 3.1207446910182157,
-      "grad_norm": 0.0010521799558773637,
-      "learning_rate": 0.00019999519635847216,
-      "loss": 46.0,
-      "step": 40817
-    },
-    {
-      "epoch": 3.1208211480016055,
-      "grad_norm": 0.0035618990659713745,
-      "learning_rate": 0.00019999519612303894,
-      "loss": 46.0,
-      "step": 40818
-    },
-    {
-      "epoch": 3.1208976049849952,
-      "grad_norm": 0.0010985324624925852,
-      "learning_rate": 0.00019999519588759994,
-      "loss": 46.0,
-      "step": 40819
-    },
-    {
-      "epoch": 3.120974061968385,
-      "grad_norm": 0.0011366363614797592,
-      "learning_rate": 0.00019999519565215516,
-      "loss": 46.0,
-      "step": 40820
-    },
-    {
-      "epoch": 3.1210505189517748,
-      "grad_norm": 0.004237969405949116,
-      "learning_rate": 0.0001999951954167046,
-      "loss": 46.0,
-      "step": 40821
-    },
-    {
-      "epoch": 3.1211269759351645,
-      "grad_norm": 0.0010721695143729448,
-      "learning_rate": 0.00019999519518124832,
-      "loss": 46.0,
-      "step": 40822
-    },
-    {
-      "epoch": 3.1212034329185543,
-      "grad_norm": 0.0016065388917922974,
-      "learning_rate": 0.00019999519494578625,
-      "loss": 46.0,
-      "step": 40823
-    },
-    {
-      "epoch": 3.121279889901944,
-      "grad_norm": 0.001234628725796938,
-      "learning_rate": 0.0001999951947103184,
-      "loss": 46.0,
-      "step": 40824
-    },
-    {
-      "epoch": 3.121356346885334,
-      "grad_norm": 0.0027569569647312164,
-      "learning_rate": 0.0001999951944748448,
-      "loss": 46.0,
-      "step": 40825
-    },
-    {
-      "epoch": 3.1214328038687236,
-      "grad_norm": 0.005221212282776833,
-      "learning_rate": 0.0001999951942393654,
-      "loss": 46.0,
-      "step": 40826
-    },
-    {
-      "epoch": 3.121509260852113,
-      "grad_norm": 0.0014302755007520318,
-      "learning_rate": 0.00019999519400388025,
-      "loss": 46.0,
-      "step": 40827
-    },
-    {
-      "epoch": 3.1215857178355026,
-      "grad_norm": 0.010070730932056904,
-      "learning_rate": 0.0001999951937683893,
-      "loss": 46.0,
-      "step": 40828
-    },
-    {
-      "epoch": 3.1216621748188924,
-      "grad_norm": 0.002062923740595579,
-      "learning_rate": 0.00019999519353289263,
-      "loss": 46.0,
-      "step": 40829
-    },
-    {
-      "epoch": 3.121738631802282,
-      "grad_norm": 0.0014156471006572247,
-      "learning_rate": 0.00019999519329739015,
-      "loss": 46.0,
-      "step": 40830
-    },
-    {
-      "epoch": 3.121815088785672,
-      "grad_norm": 0.0049782805144786835,
-      "learning_rate": 0.00019999519306188192,
-      "loss": 46.0,
-      "step": 40831
-    },
-    {
-      "epoch": 3.1218915457690617,
-      "grad_norm": 0.0005933191860094666,
-      "learning_rate": 0.00019999519282636795,
-      "loss": 46.0,
-      "step": 40832
-    },
-    {
-      "epoch": 3.1219680027524515,
-      "grad_norm": 0.001802325714379549,
-      "learning_rate": 0.00019999519259084818,
-      "loss": 46.0,
-      "step": 40833
-    },
-    {
-      "epoch": 3.122044459735841,
-      "grad_norm": 0.0011777272447943687,
-      "learning_rate": 0.00019999519235532263,
-      "loss": 46.0,
-      "step": 40834
-    },
-    {
-      "epoch": 3.122120916719231,
-      "grad_norm": 0.0014607589691877365,
-      "learning_rate": 0.00019999519211979133,
-      "loss": 46.0,
-      "step": 40835
-    },
-    {
-      "epoch": 3.1221973737026207,
-      "grad_norm": 0.0022352810483425856,
-      "learning_rate": 0.00019999519188425424,
-      "loss": 46.0,
-      "step": 40836
-    },
-    {
-      "epoch": 3.1222738306860105,
-      "grad_norm": 0.0024258787743747234,
-      "learning_rate": 0.0001999951916487114,
-      "loss": 46.0,
-      "step": 40837
-    },
-    {
-      "epoch": 3.1223502876694,
-      "grad_norm": 0.0013899493496865034,
-      "learning_rate": 0.0001999951914131628,
-      "loss": 46.0,
-      "step": 40838
-    },
-    {
-      "epoch": 3.1224267446527896,
-      "grad_norm": 0.0034314440563321114,
-      "learning_rate": 0.0001999951911776084,
-      "loss": 46.0,
-      "step": 40839
-    },
-    {
-      "epoch": 3.1225032016361793,
-      "grad_norm": 0.0017810367280617356,
-      "learning_rate": 0.00019999519094204826,
-      "loss": 46.0,
-      "step": 40840
-    },
-    {
-      "epoch": 3.122579658619569,
-      "grad_norm": 0.0011429385049268603,
-      "learning_rate": 0.00019999519070648233,
-      "loss": 46.0,
-      "step": 40841
-    },
-    {
-      "epoch": 3.122656115602959,
-      "grad_norm": 0.00325586274266243,
-      "learning_rate": 0.00019999519047091065,
-      "loss": 46.0,
-      "step": 40842
-    },
-    {
-      "epoch": 3.1227325725863486,
-      "grad_norm": 0.008938724175095558,
-      "learning_rate": 0.0001999951902353332,
-      "loss": 46.0,
-      "step": 40843
-    },
-    {
-      "epoch": 3.1228090295697384,
-      "grad_norm": 0.0015570831019431353,
-      "learning_rate": 0.00019999518999974997,
-      "loss": 46.0,
-      "step": 40844
-    },
-    {
-      "epoch": 3.122885486553128,
-      "grad_norm": 0.004200282972306013,
-      "learning_rate": 0.00019999518976416094,
-      "loss": 46.0,
-      "step": 40845
-    },
-    {
-      "epoch": 3.122961943536518,
-      "grad_norm": 0.00270732119679451,
-      "learning_rate": 0.0001999951895285662,
-      "loss": 46.0,
-      "step": 40846
-    },
-    {
-      "epoch": 3.1230384005199077,
-      "grad_norm": 0.0020568890031427145,
-      "learning_rate": 0.00019999518929296565,
-      "loss": 46.0,
-      "step": 40847
-    },
-    {
-      "epoch": 3.123114857503297,
-      "grad_norm": 0.0012429124908521771,
-      "learning_rate": 0.00019999518905735935,
-      "loss": 46.0,
-      "step": 40848
-    },
-    {
-      "epoch": 3.1231913144866867,
-      "grad_norm": 0.0013417612062767148,
-      "learning_rate": 0.00019999518882174729,
-      "loss": 46.0,
-      "step": 40849
-    },
-    {
-      "epoch": 3.1232677714700765,
-      "grad_norm": 0.005240780767053366,
-      "learning_rate": 0.00019999518858612942,
-      "loss": 46.0,
-      "step": 40850
-    },
-    {
-      "epoch": 3.1233442284534663,
-      "grad_norm": 0.002773192012682557,
-      "learning_rate": 0.0001999951883505058,
-      "loss": 46.0,
-      "step": 40851
-    },
-    {
-      "epoch": 3.123420685436856,
-      "grad_norm": 0.001967603573575616,
-      "learning_rate": 0.00019999518811487642,
-      "loss": 46.0,
-      "step": 40852
-    },
-    {
-      "epoch": 3.123497142420246,
-      "grad_norm": 0.0004903913359157741,
-      "learning_rate": 0.00019999518787924128,
-      "loss": 46.0,
-      "step": 40853
-    },
-    {
-      "epoch": 3.1235735994036355,
-      "grad_norm": 0.00274214381352067,
-      "learning_rate": 0.00019999518764360038,
-      "loss": 46.0,
-      "step": 40854
-    },
-    {
-      "epoch": 3.1236500563870253,
-      "grad_norm": 0.0007771796081215143,
-      "learning_rate": 0.00019999518740795367,
-      "loss": 46.0,
-      "step": 40855
-    },
-    {
-      "epoch": 3.123726513370415,
-      "grad_norm": 0.004473615903407335,
-      "learning_rate": 0.0001999951871723012,
-      "loss": 46.0,
-      "step": 40856
-    },
-    {
-      "epoch": 3.123802970353805,
-      "grad_norm": 0.0031921842601150274,
-      "learning_rate": 0.000199995186936643,
-      "loss": 46.0,
-      "step": 40857
-    },
-    {
-      "epoch": 3.1238794273371946,
-      "grad_norm": 0.003890521125867963,
-      "learning_rate": 0.00019999518670097896,
-      "loss": 46.0,
-      "step": 40858
-    },
-    {
-      "epoch": 3.1239558843205844,
-      "grad_norm": 0.0012546718353405595,
-      "learning_rate": 0.00019999518646530922,
-      "loss": 46.0,
-      "step": 40859
-    },
-    {
-      "epoch": 3.1240323413039737,
-      "grad_norm": 0.0017951064510270953,
-      "learning_rate": 0.0001999951862296337,
-      "loss": 46.0,
-      "step": 40860
-    },
-    {
-      "epoch": 3.1241087982873634,
-      "grad_norm": 0.0006624902016483247,
-      "learning_rate": 0.00019999518599395238,
-      "loss": 46.0,
-      "step": 40861
-    },
-    {
-      "epoch": 3.124185255270753,
-      "grad_norm": 0.0021080849692225456,
-      "learning_rate": 0.00019999518575826528,
-      "loss": 46.0,
-      "step": 40862
-    },
-    {
-      "epoch": 3.124261712254143,
-      "grad_norm": 0.0017199330031871796,
-      "learning_rate": 0.00019999518552257244,
-      "loss": 46.0,
-      "step": 40863
-    },
-    {
-      "epoch": 3.1243381692375327,
-      "grad_norm": 0.0022995760664343834,
-      "learning_rate": 0.00019999518528687386,
-      "loss": 46.0,
-      "step": 40864
-    },
-    {
-      "epoch": 3.1244146262209225,
-      "grad_norm": 0.008426333777606487,
-      "learning_rate": 0.00019999518505116944,
-      "loss": 46.0,
-      "step": 40865
-    },
-    {
-      "epoch": 3.1244910832043122,
-      "grad_norm": 0.0014875649940222502,
-      "learning_rate": 0.0001999951848154593,
-      "loss": 46.0,
-      "step": 40866
-    },
-    {
-      "epoch": 3.124567540187702,
-      "grad_norm": 0.000963767000939697,
-      "learning_rate": 0.00019999518457974337,
-      "loss": 46.0,
-      "step": 40867
-    },
-    {
-      "epoch": 3.1246439971710918,
-      "grad_norm": 0.0027432607021182775,
-      "learning_rate": 0.0001999951843440217,
-      "loss": 46.0,
-      "step": 40868
-    },
-    {
-      "epoch": 3.1247204541544815,
-      "grad_norm": 0.0024546433705836535,
-      "learning_rate": 0.00019999518410829424,
-      "loss": 46.0,
-      "step": 40869
-    },
-    {
-      "epoch": 3.124796911137871,
-      "grad_norm": 0.0009702792740426958,
-      "learning_rate": 0.00019999518387256102,
-      "loss": 46.0,
-      "step": 40870
-    },
-    {
-      "epoch": 3.1248733681212606,
-      "grad_norm": 0.0008586355252191424,
-      "learning_rate": 0.00019999518363682202,
-      "loss": 46.0,
-      "step": 40871
-    },
-    {
-      "epoch": 3.1249498251046504,
-      "grad_norm": 0.002300516003742814,
-      "learning_rate": 0.00019999518340107724,
-      "loss": 46.0,
-      "step": 40872
-    },
-    {
-      "epoch": 3.12502628208804,
-      "grad_norm": 0.0014963094145059586,
-      "learning_rate": 0.0001999951831653267,
-      "loss": 46.0,
-      "step": 40873
-    },
-    {
-      "epoch": 3.12510273907143,
-      "grad_norm": 0.0008829200523905456,
-      "learning_rate": 0.0001999951829295704,
-      "loss": 46.0,
-      "step": 40874
-    },
-    {
-      "epoch": 3.1251791960548196,
-      "grad_norm": 0.001190814538858831,
-      "learning_rate": 0.00019999518269380834,
-      "loss": 46.0,
-      "step": 40875
-    },
-    {
-      "epoch": 3.1252556530382094,
-      "grad_norm": 0.004890835378319025,
-      "learning_rate": 0.00019999518245804047,
-      "loss": 46.0,
-      "step": 40876
-    },
-    {
-      "epoch": 3.125332110021599,
-      "grad_norm": 0.002724532736465335,
-      "learning_rate": 0.00019999518222226688,
-      "loss": 46.0,
-      "step": 40877
-    },
-    {
-      "epoch": 3.125408567004989,
-      "grad_norm": 0.0015181906055659056,
-      "learning_rate": 0.0001999951819864875,
-      "loss": 46.0,
-      "step": 40878
-    },
-    {
-      "epoch": 3.1254850239883787,
-      "grad_norm": 0.0008061067201197147,
-      "learning_rate": 0.00019999518175070234,
-      "loss": 46.0,
-      "step": 40879
-    },
-    {
-      "epoch": 3.1255614809717684,
-      "grad_norm": 0.006594307720661163,
-      "learning_rate": 0.00019999518151491143,
-      "loss": 46.0,
-      "step": 40880
-    },
-    {
-      "epoch": 3.125637937955158,
-      "grad_norm": 0.0028224645648151636,
-      "learning_rate": 0.00019999518127911475,
-      "loss": 46.0,
-      "step": 40881
-    },
-    {
-      "epoch": 3.1257143949385475,
-      "grad_norm": 0.0022636328358203173,
-      "learning_rate": 0.00019999518104331227,
-      "loss": 46.0,
-      "step": 40882
-    },
-    {
-      "epoch": 3.1257908519219373,
-      "grad_norm": 0.004256364889442921,
-      "learning_rate": 0.00019999518080750405,
-      "loss": 46.0,
-      "step": 40883
-    },
-    {
-      "epoch": 3.125867308905327,
-      "grad_norm": 0.0014648041687905788,
-      "learning_rate": 0.00019999518057169007,
-      "loss": 46.0,
-      "step": 40884
-    },
-    {
-      "epoch": 3.125943765888717,
-      "grad_norm": 0.002286911476403475,
-      "learning_rate": 0.0001999951803358703,
-      "loss": 46.0,
-      "step": 40885
-    },
-    {
-      "epoch": 3.1260202228721066,
-      "grad_norm": 0.003924392629414797,
-      "learning_rate": 0.00019999518010004475,
-      "loss": 46.0,
-      "step": 40886
-    },
-    {
-      "epoch": 3.1260966798554963,
-      "grad_norm": 0.00516595970839262,
-      "learning_rate": 0.00019999517986421346,
-      "loss": 46.0,
-      "step": 40887
-    },
-    {
-      "epoch": 3.126173136838886,
-      "grad_norm": 0.0040839132852852345,
-      "learning_rate": 0.00019999517962837637,
-      "loss": 46.0,
-      "step": 40888
-    },
-    {
-      "epoch": 3.126249593822276,
-      "grad_norm": 0.0007673455402255058,
-      "learning_rate": 0.00019999517939253353,
-      "loss": 46.0,
-      "step": 40889
-    },
-    {
-      "epoch": 3.1263260508056656,
-      "grad_norm": 0.0015977041330188513,
-      "learning_rate": 0.00019999517915668494,
-      "loss": 46.0,
-      "step": 40890
-    },
-    {
-      "epoch": 3.1264025077890554,
-      "grad_norm": 0.002916666679084301,
-      "learning_rate": 0.00019999517892083053,
-      "loss": 46.0,
-      "step": 40891
-    },
-    {
-      "epoch": 3.1264789647724447,
-      "grad_norm": 0.004950730595737696,
-      "learning_rate": 0.0001999951786849704,
-      "loss": 46.0,
-      "step": 40892
-    },
-    {
-      "epoch": 3.1265554217558345,
-      "grad_norm": 0.0007169688469730318,
-      "learning_rate": 0.00019999517844910446,
-      "loss": 46.0,
-      "step": 40893
-    },
-    {
-      "epoch": 3.126631878739224,
-      "grad_norm": 0.001416036975570023,
-      "learning_rate": 0.00019999517821323276,
-      "loss": 46.0,
-      "step": 40894
-    },
-    {
-      "epoch": 3.126708335722614,
-      "grad_norm": 0.0038079284131526947,
-      "learning_rate": 0.00019999517797735533,
-      "loss": 46.0,
-      "step": 40895
-    },
-    {
-      "epoch": 3.1267847927060037,
-      "grad_norm": 0.0025651552714407444,
-      "learning_rate": 0.0001999951777414721,
-      "loss": 46.0,
-      "step": 40896
-    },
-    {
-      "epoch": 3.1268612496893935,
-      "grad_norm": 0.0006050064112059772,
-      "learning_rate": 0.0001999951775055831,
-      "loss": 46.0,
-      "step": 40897
-    },
-    {
-      "epoch": 3.1269377066727833,
-      "grad_norm": 0.00445164181292057,
-      "learning_rate": 0.00019999517726968833,
-      "loss": 46.0,
-      "step": 40898
-    },
-    {
-      "epoch": 3.127014163656173,
-      "grad_norm": 0.002387454267591238,
-      "learning_rate": 0.00019999517703378782,
-      "loss": 46.0,
-      "step": 40899
-    },
-    {
-      "epoch": 3.127090620639563,
-      "grad_norm": 0.000696027185767889,
-      "learning_rate": 0.00019999517679788152,
-      "loss": 46.0,
-      "step": 40900
-    },
-    {
-      "epoch": 3.1271670776229525,
-      "grad_norm": 0.004150791559368372,
-      "learning_rate": 0.00019999517656196943,
-      "loss": 46.0,
-      "step": 40901
-    },
-    {
-      "epoch": 3.1272435346063423,
-      "grad_norm": 0.000997581286355853,
-      "learning_rate": 0.0001999951763260516,
-      "loss": 46.0,
-      "step": 40902
-    },
-    {
-      "epoch": 3.127319991589732,
-      "grad_norm": 0.002457044553011656,
-      "learning_rate": 0.00019999517609012798,
-      "loss": 46.0,
-      "step": 40903
-    },
-    {
-      "epoch": 3.1273964485731214,
-      "grad_norm": 0.0027247476391494274,
-      "learning_rate": 0.0001999951758541986,
-      "loss": 46.0,
-      "step": 40904
-    },
-    {
-      "epoch": 3.127472905556511,
-      "grad_norm": 0.0008025128045119345,
-      "learning_rate": 0.00019999517561826346,
-      "loss": 46.0,
-      "step": 40905
-    },
-    {
-      "epoch": 3.127549362539901,
-      "grad_norm": 0.002650618087500334,
-      "learning_rate": 0.00019999517538232253,
-      "loss": 46.0,
-      "step": 40906
-    },
-    {
-      "epoch": 3.1276258195232907,
-      "grad_norm": 0.0015751130413264036,
-      "learning_rate": 0.00019999517514637585,
-      "loss": 46.0,
-      "step": 40907
-    },
-    {
-      "epoch": 3.1277022765066804,
-      "grad_norm": 0.0021378311794251204,
-      "learning_rate": 0.0001999951749104234,
-      "loss": 46.0,
-      "step": 40908
-    },
-    {
-      "epoch": 3.12777873349007,
-      "grad_norm": 0.004254310857504606,
-      "learning_rate": 0.00019999517467446517,
-      "loss": 46.0,
-      "step": 40909
-    },
-    {
-      "epoch": 3.12785519047346,
-      "grad_norm": 0.0014484584098681808,
-      "learning_rate": 0.00019999517443850117,
-      "loss": 46.0,
-      "step": 40910
-    },
-    {
-      "epoch": 3.1279316474568497,
-      "grad_norm": 0.002101573860272765,
-      "learning_rate": 0.0001999951742025314,
-      "loss": 46.0,
-      "step": 40911
-    },
-    {
-      "epoch": 3.1280081044402395,
-      "grad_norm": 0.001584112411364913,
-      "learning_rate": 0.00019999517396655588,
-      "loss": 46.0,
-      "step": 40912
-    },
-    {
-      "epoch": 3.1280845614236292,
-      "grad_norm": 0.003157875034958124,
-      "learning_rate": 0.00019999517373057456,
-      "loss": 46.0,
-      "step": 40913
-    },
-    {
-      "epoch": 3.1281610184070185,
-      "grad_norm": 0.002584375673905015,
-      "learning_rate": 0.0001999951734945875,
-      "loss": 46.0,
-      "step": 40914
-    },
-    {
-      "epoch": 3.1282374753904083,
-      "grad_norm": 0.0015856287209317088,
-      "learning_rate": 0.00019999517325859466,
-      "loss": 46.0,
-      "step": 40915
-    },
-    {
-      "epoch": 3.128313932373798,
-      "grad_norm": 0.001126607065089047,
-      "learning_rate": 0.00019999517302259605,
-      "loss": 46.0,
-      "step": 40916
-    },
-    {
-      "epoch": 3.128390389357188,
-      "grad_norm": 0.0015187772223725915,
-      "learning_rate": 0.0001999951727865917,
-      "loss": 46.0,
-      "step": 40917
-    },
-    {
-      "epoch": 3.1284668463405776,
-      "grad_norm": 0.0017307294765487313,
-      "learning_rate": 0.0001999951725505815,
-      "loss": 46.0,
-      "step": 40918
-    },
-    {
-      "epoch": 3.1285433033239674,
-      "grad_norm": 0.0010536620393395424,
-      "learning_rate": 0.0001999951723145656,
-      "loss": 46.0,
-      "step": 40919
-    },
-    {
-      "epoch": 3.128619760307357,
-      "grad_norm": 0.0030114694964140654,
-      "learning_rate": 0.00019999517207854392,
-      "loss": 46.0,
-      "step": 40920
-    },
-    {
-      "epoch": 3.128696217290747,
-      "grad_norm": 0.0014078689273446798,
-      "learning_rate": 0.00019999517184251647,
-      "loss": 46.0,
-      "step": 40921
-    },
-    {
-      "epoch": 3.1287726742741366,
-      "grad_norm": 0.002471883548423648,
-      "learning_rate": 0.0001999951716064832,
-      "loss": 46.0,
-      "step": 40922
-    },
-    {
-      "epoch": 3.1288491312575264,
-      "grad_norm": 0.004802376497536898,
-      "learning_rate": 0.00019999517137044424,
-      "loss": 46.0,
-      "step": 40923
-    },
-    {
-      "epoch": 3.128925588240916,
-      "grad_norm": 0.004977057222276926,
-      "learning_rate": 0.00019999517113439947,
-      "loss": 46.0,
-      "step": 40924
-    },
-    {
-      "epoch": 3.1290020452243055,
-      "grad_norm": 0.008620427921414375,
-      "learning_rate": 0.00019999517089834895,
-      "loss": 46.0,
-      "step": 40925
-    },
-    {
-      "epoch": 3.1290785022076952,
-      "grad_norm": 0.001090528559871018,
-      "learning_rate": 0.00019999517066229263,
-      "loss": 46.0,
-      "step": 40926
-    },
-    {
-      "epoch": 3.129154959191085,
-      "grad_norm": 0.0009361812844872475,
-      "learning_rate": 0.00019999517042623057,
-      "loss": 46.0,
-      "step": 40927
-    },
-    {
-      "epoch": 3.1292314161744748,
-      "grad_norm": 0.0031330594792962074,
-      "learning_rate": 0.00019999517019016273,
-      "loss": 46.0,
-      "step": 40928
-    },
-    {
-      "epoch": 3.1293078731578645,
-      "grad_norm": 0.0030115721747279167,
-      "learning_rate": 0.00019999516995408912,
-      "loss": 46.0,
-      "step": 40929
-    },
-    {
-      "epoch": 3.1293843301412543,
-      "grad_norm": 0.0014377201441675425,
-      "learning_rate": 0.00019999516971800973,
-      "loss": 46.0,
-      "step": 40930
-    },
-    {
-      "epoch": 3.129460787124644,
-      "grad_norm": 0.004471286665648222,
-      "learning_rate": 0.0001999951694819246,
-      "loss": 46.0,
-      "step": 40931
-    },
-    {
-      "epoch": 3.129537244108034,
-      "grad_norm": 0.000843784655444324,
-      "learning_rate": 0.00019999516924583367,
-      "loss": 46.0,
-      "step": 40932
-    },
-    {
-      "epoch": 3.1296137010914236,
-      "grad_norm": 0.005641388241201639,
-      "learning_rate": 0.000199995169009737,
-      "loss": 46.0,
-      "step": 40933
-    },
-    {
-      "epoch": 3.1296901580748133,
-      "grad_norm": 0.0013651661574840546,
-      "learning_rate": 0.00019999516877363454,
-      "loss": 46.0,
-      "step": 40934
-    },
-    {
-      "epoch": 3.129766615058203,
-      "grad_norm": 0.0029702791944146156,
-      "learning_rate": 0.00019999516853752631,
-      "loss": 46.0,
-      "step": 40935
-    },
-    {
-      "epoch": 3.1298430720415924,
-      "grad_norm": 0.0022475728765130043,
-      "learning_rate": 0.00019999516830141232,
-      "loss": 46.0,
-      "step": 40936
-    },
-    {
-      "epoch": 3.129919529024982,
-      "grad_norm": 0.004736334551125765,
-      "learning_rate": 0.00019999516806529254,
-      "loss": 46.0,
-      "step": 40937
-    },
-    {
-      "epoch": 3.129995986008372,
-      "grad_norm": 0.0020905614364892244,
-      "learning_rate": 0.00019999516782916703,
-      "loss": 46.0,
-      "step": 40938
-    },
-    {
-      "epoch": 3.1300724429917617,
-      "grad_norm": 0.006155779119580984,
-      "learning_rate": 0.0001999951675930357,
-      "loss": 46.0,
-      "step": 40939
-    },
-    {
-      "epoch": 3.1301488999751514,
-      "grad_norm": 0.0017561701824888587,
-      "learning_rate": 0.00019999516735689864,
-      "loss": 46.0,
-      "step": 40940
-    },
-    {
-      "epoch": 3.130225356958541,
-      "grad_norm": 0.0017041712999343872,
-      "learning_rate": 0.0001999951671207558,
-      "loss": 46.0,
-      "step": 40941
-    },
-    {
-      "epoch": 3.130301813941931,
-      "grad_norm": 0.0007449098629876971,
-      "learning_rate": 0.0001999951668846072,
-      "loss": 46.0,
-      "step": 40942
-    },
-    {
-      "epoch": 3.1303782709253207,
-      "grad_norm": 0.0015767990844324231,
-      "learning_rate": 0.00019999516664845283,
-      "loss": 46.0,
-      "step": 40943
-    },
-    {
-      "epoch": 3.1304547279087105,
-      "grad_norm": 0.003999115899205208,
-      "learning_rate": 0.00019999516641229268,
-      "loss": 46.0,
-      "step": 40944
-    },
-    {
-      "epoch": 3.1305311848921002,
-      "grad_norm": 0.0011869925074279308,
-      "learning_rate": 0.00019999516617612674,
-      "loss": 46.0,
-      "step": 40945
-    },
-    {
-      "epoch": 3.13060764187549,
-      "grad_norm": 0.0017360768979415298,
-      "learning_rate": 0.0001999951659399551,
-      "loss": 46.0,
-      "step": 40946
-    },
-    {
-      "epoch": 3.1306840988588793,
-      "grad_norm": 0.0018099526641890407,
-      "learning_rate": 0.00019999516570377762,
-      "loss": 46.0,
-      "step": 40947
-    },
-    {
-      "epoch": 3.130760555842269,
-      "grad_norm": 0.0008087501628324389,
-      "learning_rate": 0.0001999951654675944,
-      "loss": 46.0,
-      "step": 40948
-    },
-    {
-      "epoch": 3.130837012825659,
-      "grad_norm": 0.004666229244321585,
-      "learning_rate": 0.0001999951652314054,
-      "loss": 46.0,
-      "step": 40949
-    },
-    {
-      "epoch": 3.1309134698090486,
-      "grad_norm": 0.002313598059117794,
-      "learning_rate": 0.00019999516499521062,
-      "loss": 46.0,
-      "step": 40950
-    },
-    {
-      "epoch": 3.1309899267924384,
-      "grad_norm": 0.005280129611492157,
-      "learning_rate": 0.0001999951647590101,
-      "loss": 46.0,
-      "step": 40951
-    },
-    {
-      "epoch": 3.131066383775828,
-      "grad_norm": 0.0037140948697924614,
-      "learning_rate": 0.0001999951645228038,
-      "loss": 46.0,
-      "step": 40952
-    },
-    {
-      "epoch": 3.131142840759218,
-      "grad_norm": 0.0010154227493330836,
-      "learning_rate": 0.00019999516428659172,
-      "loss": 46.0,
-      "step": 40953
-    },
-    {
-      "epoch": 3.1312192977426077,
-      "grad_norm": 0.002763459226116538,
-      "learning_rate": 0.00019999516405037388,
-      "loss": 46.0,
-      "step": 40954
-    },
-    {
-      "epoch": 3.1312957547259974,
-      "grad_norm": 0.0035257311537861824,
-      "learning_rate": 0.0001999951638141503,
-      "loss": 46.0,
-      "step": 40955
-    },
-    {
-      "epoch": 3.131372211709387,
-      "grad_norm": 0.007086979225277901,
-      "learning_rate": 0.00019999516357792091,
-      "loss": 46.0,
-      "step": 40956
-    },
-    {
-      "epoch": 3.1314486686927765,
-      "grad_norm": 0.0026205633766949177,
-      "learning_rate": 0.00019999516334168576,
-      "loss": 46.0,
-      "step": 40957
-    },
-    {
-      "epoch": 3.1315251256761663,
-      "grad_norm": 0.000860468833707273,
-      "learning_rate": 0.00019999516310544485,
-      "loss": 46.0,
-      "step": 40958
-    },
-    {
-      "epoch": 3.131601582659556,
-      "grad_norm": 0.002395564690232277,
-      "learning_rate": 0.00019999516286919818,
-      "loss": 46.0,
-      "step": 40959
-    },
-    {
-      "epoch": 3.1316780396429458,
-      "grad_norm": 0.0009179324260912836,
-      "learning_rate": 0.0001999951626329457,
-      "loss": 46.0,
-      "step": 40960
-    },
-    {
-      "epoch": 3.1317544966263355,
-      "grad_norm": 0.0032772414851933718,
-      "learning_rate": 0.0001999951623966875,
-      "loss": 46.0,
-      "step": 40961
-    },
-    {
-      "epoch": 3.1318309536097253,
-      "grad_norm": 0.0012345022987574339,
-      "learning_rate": 0.0001999951621604235,
-      "loss": 46.0,
-      "step": 40962
-    },
-    {
-      "epoch": 3.131907410593115,
-      "grad_norm": 0.0016557055059820414,
-      "learning_rate": 0.00019999516192415376,
-      "loss": 46.0,
-      "step": 40963
-    },
-    {
-      "epoch": 3.131983867576505,
-      "grad_norm": 0.005347746890038252,
-      "learning_rate": 0.0001999951616878782,
-      "loss": 46.0,
-      "step": 40964
-    },
-    {
-      "epoch": 3.1320603245598946,
-      "grad_norm": 0.002023885026574135,
-      "learning_rate": 0.00019999516145159692,
-      "loss": 46.0,
-      "step": 40965
-    },
-    {
-      "epoch": 3.1321367815432843,
-      "grad_norm": 0.0017879637889564037,
-      "learning_rate": 0.00019999516121530983,
-      "loss": 46.0,
-      "step": 40966
-    },
-    {
-      "epoch": 3.132213238526674,
-      "grad_norm": 0.004732067231088877,
-      "learning_rate": 0.000199995160979017,
-      "loss": 46.0,
-      "step": 40967
-    },
-    {
-      "epoch": 3.132289695510064,
-      "grad_norm": 0.0028899898752570152,
-      "learning_rate": 0.0001999951607427184,
-      "loss": 46.0,
-      "step": 40968
-    },
-    {
-      "epoch": 3.132366152493453,
-      "grad_norm": 0.002965432358905673,
-      "learning_rate": 0.000199995160506414,
-      "loss": 46.0,
-      "step": 40969
-    },
-    {
-      "epoch": 3.132442609476843,
-      "grad_norm": 0.0026841112412512302,
-      "learning_rate": 0.00019999516027010387,
-      "loss": 46.0,
-      "step": 40970
-    },
-    {
-      "epoch": 3.1325190664602327,
-      "grad_norm": 0.003204314736649394,
-      "learning_rate": 0.00019999516003378796,
-      "loss": 46.0,
-      "step": 40971
-    },
-    {
-      "epoch": 3.1325955234436225,
-      "grad_norm": 0.0020025165285915136,
-      "learning_rate": 0.00019999515979746626,
-      "loss": 46.0,
-      "step": 40972
-    },
-    {
-      "epoch": 3.1326719804270122,
-      "grad_norm": 0.0028681980911642313,
-      "learning_rate": 0.0001999951595611388,
-      "loss": 46.0,
-      "step": 40973
-    },
-    {
-      "epoch": 3.132748437410402,
-      "grad_norm": 0.0038651989307254553,
-      "learning_rate": 0.00019999515932480559,
-      "loss": 46.0,
-      "step": 40974
-    },
-    {
-      "epoch": 3.1328248943937917,
-      "grad_norm": 0.004710248671472073,
-      "learning_rate": 0.0001999951590884666,
-      "loss": 46.0,
-      "step": 40975
-    },
-    {
-      "epoch": 3.1329013513771815,
-      "grad_norm": 0.0015670906286686659,
-      "learning_rate": 0.00019999515885212184,
-      "loss": 46.0,
-      "step": 40976
-    },
-    {
-      "epoch": 3.1329778083605713,
-      "grad_norm": 0.005499074701219797,
-      "learning_rate": 0.00019999515861577133,
-      "loss": 46.0,
-      "step": 40977
-    },
-    {
-      "epoch": 3.133054265343961,
-      "grad_norm": 0.004043113440275192,
-      "learning_rate": 0.00019999515837941504,
-      "loss": 46.0,
-      "step": 40978
-    },
-    {
-      "epoch": 3.1331307223273503,
-      "grad_norm": 0.004674827679991722,
-      "learning_rate": 0.00019999515814305295,
-      "loss": 46.0,
-      "step": 40979
-    },
-    {
-      "epoch": 3.13320717931074,
-      "grad_norm": 0.0018757217330858111,
-      "learning_rate": 0.00019999515790668514,
-      "loss": 46.0,
-      "step": 40980
-    },
-    {
-      "epoch": 3.13328363629413,
-      "grad_norm": 0.005505661014467478,
-      "learning_rate": 0.0001999951576703115,
-      "loss": 46.0,
-      "step": 40981
-    },
-    {
-      "epoch": 3.1333600932775196,
-      "grad_norm": 0.00399810541421175,
-      "learning_rate": 0.00019999515743393214,
-      "loss": 46.0,
-      "step": 40982
-    },
-    {
-      "epoch": 3.1334365502609094,
-      "grad_norm": 0.0053443508222699165,
-      "learning_rate": 0.00019999515719754698,
-      "loss": 46.0,
-      "step": 40983
-    },
-    {
-      "epoch": 3.133513007244299,
-      "grad_norm": 0.013109898194670677,
-      "learning_rate": 0.00019999515696115608,
-      "loss": 46.0,
-      "step": 40984
-    },
-    {
-      "epoch": 3.133589464227689,
-      "grad_norm": 0.0013333922252058983,
-      "learning_rate": 0.0001999951567247594,
-      "loss": 46.0,
-      "step": 40985
-    },
-    {
-      "epoch": 3.1336659212110787,
-      "grad_norm": 0.0017973404610529542,
-      "learning_rate": 0.00019999515648835693,
-      "loss": 46.0,
-      "step": 40986
-    },
-    {
-      "epoch": 3.1337423781944684,
-      "grad_norm": 0.0013133450411260128,
-      "learning_rate": 0.0001999951562519487,
-      "loss": 46.0,
-      "step": 40987
-    },
-    {
-      "epoch": 3.133818835177858,
-      "grad_norm": 0.0064160265028476715,
-      "learning_rate": 0.00019999515601553473,
-      "loss": 46.0,
-      "step": 40988
-    },
-    {
-      "epoch": 3.133895292161248,
-      "grad_norm": 0.002278312109410763,
-      "learning_rate": 0.00019999515577911496,
-      "loss": 46.0,
-      "step": 40989
-    },
-    {
-      "epoch": 3.1339717491446377,
-      "grad_norm": 0.004158418159931898,
-      "learning_rate": 0.00019999515554268942,
-      "loss": 46.0,
-      "step": 40990
-    },
-    {
-      "epoch": 3.134048206128027,
-      "grad_norm": 0.004669709596782923,
-      "learning_rate": 0.00019999515530625815,
-      "loss": 46.0,
-      "step": 40991
-    },
-    {
-      "epoch": 3.134124663111417,
-      "grad_norm": 0.0017834221944212914,
-      "learning_rate": 0.00019999515506982106,
-      "loss": 46.0,
-      "step": 40992
-    },
-    {
-      "epoch": 3.1342011200948066,
-      "grad_norm": 0.0005970390047878027,
-      "learning_rate": 0.00019999515483337823,
-      "loss": 46.0,
-      "step": 40993
-    },
-    {
-      "epoch": 3.1342775770781963,
-      "grad_norm": 0.0050776260904967785,
-      "learning_rate": 0.00019999515459692964,
-      "loss": 46.0,
-      "step": 40994
-    },
-    {
-      "epoch": 3.134354034061586,
-      "grad_norm": 0.002877766266465187,
-      "learning_rate": 0.00019999515436047523,
-      "loss": 46.0,
-      "step": 40995
-    },
-    {
-      "epoch": 3.134430491044976,
-      "grad_norm": 0.0024728337302803993,
-      "learning_rate": 0.0001999951541240151,
-      "loss": 46.0,
-      "step": 40996
-    },
-    {
-      "epoch": 3.1345069480283656,
-      "grad_norm": 0.0013357122661545873,
-      "learning_rate": 0.0001999951538875492,
-      "loss": 46.0,
-      "step": 40997
-    },
-    {
-      "epoch": 3.1345834050117554,
-      "grad_norm": 0.0009288405999541283,
-      "learning_rate": 0.0001999951536510775,
-      "loss": 46.0,
-      "step": 40998
-    },
-    {
-      "epoch": 3.134659861995145,
-      "grad_norm": 0.0009051059023477137,
-      "learning_rate": 0.00019999515341460007,
-      "loss": 46.0,
-      "step": 40999
-    },
-    {
-      "epoch": 3.134736318978535,
-      "grad_norm": 0.0024559807498008013,
-      "learning_rate": 0.00019999515317811682,
-      "loss": 46.0,
-      "step": 41000
-    },
-    {
-      "epoch": 3.134812775961924,
-      "grad_norm": 0.004311062395572662,
-      "learning_rate": 0.00019999515294162785,
-      "loss": 46.0,
-      "step": 41001
-    },
-    {
-      "epoch": 3.134889232945314,
-      "grad_norm": 0.002960910787805915,
-      "learning_rate": 0.00019999515270513308,
-      "loss": 46.0,
-      "step": 41002
-    },
-    {
-      "epoch": 3.1349656899287037,
-      "grad_norm": 0.0013237728271633387,
-      "learning_rate": 0.00019999515246863257,
-      "loss": 46.0,
-      "step": 41003
-    },
-    {
-      "epoch": 3.1350421469120935,
-      "grad_norm": 0.0033009194303303957,
-      "learning_rate": 0.00019999515223212625,
-      "loss": 46.0,
-      "step": 41004
-    },
-    {
-      "epoch": 3.1351186038954832,
-      "grad_norm": 0.0016189126763492823,
-      "learning_rate": 0.00019999515199561421,
-      "loss": 46.0,
-      "step": 41005
-    },
-    {
-      "epoch": 3.135195060878873,
-      "grad_norm": 0.001902610994875431,
-      "learning_rate": 0.00019999515175909635,
-      "loss": 46.0,
-      "step": 41006
-    },
-    {
-      "epoch": 3.1352715178622628,
-      "grad_norm": 0.0012386628659442067,
-      "learning_rate": 0.00019999515152257277,
-      "loss": 46.0,
-      "step": 41007
-    },
-    {
-      "epoch": 3.1353479748456525,
-      "grad_norm": 0.0005118967383168638,
-      "learning_rate": 0.00019999515128604338,
-      "loss": 46.0,
-      "step": 41008
-    },
-    {
-      "epoch": 3.1354244318290423,
-      "grad_norm": 0.002692008623853326,
-      "learning_rate": 0.00019999515104950823,
-      "loss": 46.0,
-      "step": 41009
-    },
-    {
-      "epoch": 3.135500888812432,
-      "grad_norm": 0.003653703723102808,
-      "learning_rate": 0.00019999515081296732,
-      "loss": 46.0,
-      "step": 41010
-    },
-    {
-      "epoch": 3.135577345795822,
-      "grad_norm": 0.005738893058151007,
-      "learning_rate": 0.00019999515057642065,
-      "loss": 46.0,
-      "step": 41011
-    },
-    {
-      "epoch": 3.1356538027792116,
-      "grad_norm": 0.004237464163452387,
-      "learning_rate": 0.00019999515033986817,
-      "loss": 46.0,
-      "step": 41012
-    },
-    {
-      "epoch": 3.135730259762601,
-      "grad_norm": 0.003687287215143442,
-      "learning_rate": 0.00019999515010330997,
-      "loss": 46.0,
-      "step": 41013
-    },
-    {
-      "epoch": 3.1358067167459907,
-      "grad_norm": 0.0021057743579149246,
-      "learning_rate": 0.00019999514986674598,
-      "loss": 46.0,
-      "step": 41014
-    },
-    {
-      "epoch": 3.1358831737293804,
-      "grad_norm": 0.0025039201136678457,
-      "learning_rate": 0.00019999514963017624,
-      "loss": 46.0,
-      "step": 41015
-    },
-    {
-      "epoch": 3.13595963071277,
-      "grad_norm": 0.0032272557727992535,
-      "learning_rate": 0.00019999514939360072,
-      "loss": 46.0,
-      "step": 41016
-    },
-    {
-      "epoch": 3.13603608769616,
-      "grad_norm": 0.004893191158771515,
-      "learning_rate": 0.0001999951491570194,
-      "loss": 46.0,
-      "step": 41017
-    },
-    {
-      "epoch": 3.1361125446795497,
-      "grad_norm": 0.0007426108932122588,
-      "learning_rate": 0.0001999951489204323,
-      "loss": 46.0,
-      "step": 41018
-    },
-    {
-      "epoch": 3.1361890016629395,
-      "grad_norm": 0.0015161726623773575,
-      "learning_rate": 0.00019999514868383948,
-      "loss": 46.0,
-      "step": 41019
-    },
-    {
-      "epoch": 3.136265458646329,
-      "grad_norm": 0.002347435336560011,
-      "learning_rate": 0.0001999951484472409,
-      "loss": 46.0,
-      "step": 41020
-    },
-    {
-      "epoch": 3.136341915629719,
-      "grad_norm": 0.006625575479120016,
-      "learning_rate": 0.0001999951482106365,
-      "loss": 46.0,
-      "step": 41021
-    },
-    {
-      "epoch": 3.1364183726131087,
-      "grad_norm": 0.0017947033047676086,
-      "learning_rate": 0.00019999514797402638,
-      "loss": 46.0,
-      "step": 41022
-    },
-    {
-      "epoch": 3.136494829596498,
-      "grad_norm": 0.002447431907057762,
-      "learning_rate": 0.00019999514773741045,
-      "loss": 46.0,
-      "step": 41023
-    },
-    {
-      "epoch": 3.136571286579888,
-      "grad_norm": 0.0013351301895454526,
-      "learning_rate": 0.00019999514750078878,
-      "loss": 46.0,
-      "step": 41024
-    },
-    {
-      "epoch": 3.1366477435632776,
-      "grad_norm": 0.0013199971290305257,
-      "learning_rate": 0.00019999514726416133,
-      "loss": 46.0,
-      "step": 41025
-    },
-    {
-      "epoch": 3.1367242005466673,
-      "grad_norm": 0.002232053317129612,
-      "learning_rate": 0.00019999514702752813,
-      "loss": 46.0,
-      "step": 41026
-    },
-    {
-      "epoch": 3.136800657530057,
-      "grad_norm": 0.0015525601338595152,
-      "learning_rate": 0.00019999514679088913,
-      "loss": 46.0,
-      "step": 41027
-    },
-    {
-      "epoch": 3.136877114513447,
-      "grad_norm": 0.0038366890512406826,
-      "learning_rate": 0.00019999514655424437,
-      "loss": 46.0,
-      "step": 41028
-    },
-    {
-      "epoch": 3.1369535714968366,
-      "grad_norm": 0.0017662885366007686,
-      "learning_rate": 0.00019999514631759385,
-      "loss": 46.0,
-      "step": 41029
-    },
-    {
-      "epoch": 3.1370300284802264,
-      "grad_norm": 0.0016639424720779061,
-      "learning_rate": 0.00019999514608093753,
-      "loss": 46.0,
-      "step": 41030
-    },
-    {
-      "epoch": 3.137106485463616,
-      "grad_norm": 0.0012837244430556893,
-      "learning_rate": 0.00019999514584427547,
-      "loss": 46.0,
-      "step": 41031
-    },
-    {
-      "epoch": 3.137182942447006,
-      "grad_norm": 0.0009590934496372938,
-      "learning_rate": 0.00019999514560760763,
-      "loss": 46.0,
-      "step": 41032
-    },
-    {
-      "epoch": 3.1372593994303957,
-      "grad_norm": 0.001776376971974969,
-      "learning_rate": 0.00019999514537093405,
-      "loss": 46.0,
-      "step": 41033
-    },
-    {
-      "epoch": 3.1373358564137854,
-      "grad_norm": 0.0022420771420001984,
-      "learning_rate": 0.00019999514513425467,
-      "loss": 46.0,
-      "step": 41034
-    },
-    {
-      "epoch": 3.1374123133971747,
-      "grad_norm": 0.0024783953558653593,
-      "learning_rate": 0.00019999514489756951,
-      "loss": 46.0,
-      "step": 41035
-    },
-    {
-      "epoch": 3.1374887703805645,
-      "grad_norm": 0.0023346985690295696,
-      "learning_rate": 0.00019999514466087864,
-      "loss": 46.0,
-      "step": 41036
-    },
-    {
-      "epoch": 3.1375652273639543,
-      "grad_norm": 0.0019331456860527396,
-      "learning_rate": 0.00019999514442418194,
-      "loss": 46.0,
-      "step": 41037
-    },
-    {
-      "epoch": 3.137641684347344,
-      "grad_norm": 0.00319826602935791,
-      "learning_rate": 0.0001999951441874795,
-      "loss": 46.0,
-      "step": 41038
-    },
-    {
-      "epoch": 3.137718141330734,
-      "grad_norm": 0.0022392943501472473,
-      "learning_rate": 0.0001999951439507713,
-      "loss": 46.0,
-      "step": 41039
-    },
-    {
-      "epoch": 3.1377945983141236,
-      "grad_norm": 0.004793962929397821,
-      "learning_rate": 0.0001999951437140573,
-      "loss": 46.0,
-      "step": 41040
-    },
-    {
-      "epoch": 3.1378710552975133,
-      "grad_norm": 0.0022390270605683327,
-      "learning_rate": 0.00019999514347733753,
-      "loss": 46.0,
-      "step": 41041
-    },
-    {
-      "epoch": 3.137947512280903,
-      "grad_norm": 0.0005360444774851203,
-      "learning_rate": 0.000199995143240612,
-      "loss": 46.0,
-      "step": 41042
-    },
-    {
-      "epoch": 3.138023969264293,
-      "grad_norm": 0.001332034938968718,
-      "learning_rate": 0.00019999514300388072,
-      "loss": 46.0,
-      "step": 41043
-    },
-    {
-      "epoch": 3.1381004262476826,
-      "grad_norm": 0.002849350683391094,
-      "learning_rate": 0.00019999514276714363,
-      "loss": 46.0,
-      "step": 41044
-    },
-    {
-      "epoch": 3.138176883231072,
-      "grad_norm": 0.0033385315909981728,
-      "learning_rate": 0.0001999951425304008,
-      "loss": 46.0,
-      "step": 41045
-    },
-    {
-      "epoch": 3.1382533402144617,
-      "grad_norm": 0.004104383755475283,
-      "learning_rate": 0.00019999514229365222,
-      "loss": 46.0,
-      "step": 41046
-    },
-    {
-      "epoch": 3.1383297971978514,
-      "grad_norm": 0.0036129062063992023,
-      "learning_rate": 0.00019999514205689783,
-      "loss": 46.0,
-      "step": 41047
-    },
-    {
-      "epoch": 3.138406254181241,
-      "grad_norm": 0.001940494286827743,
-      "learning_rate": 0.0001999951418201377,
-      "loss": 46.0,
-      "step": 41048
-    },
-    {
-      "epoch": 3.138482711164631,
-      "grad_norm": 0.000988120213150978,
-      "learning_rate": 0.00019999514158337177,
-      "loss": 46.0,
-      "step": 41049
-    },
-    {
-      "epoch": 3.1385591681480207,
-      "grad_norm": 0.0008883534464985132,
-      "learning_rate": 0.0001999951413466001,
-      "loss": 46.0,
-      "step": 41050
-    },
-    {
-      "epoch": 3.1386356251314105,
-      "grad_norm": 0.0050128973089158535,
-      "learning_rate": 0.00019999514110982265,
-      "loss": 46.0,
-      "step": 41051
-    },
-    {
-      "epoch": 3.1387120821148002,
-      "grad_norm": 0.0031030410900712013,
-      "learning_rate": 0.00019999514087303943,
-      "loss": 46.0,
-      "step": 41052
-    },
-    {
-      "epoch": 3.13878853909819,
-      "grad_norm": 0.002500659553334117,
-      "learning_rate": 0.00019999514063625043,
-      "loss": 46.0,
-      "step": 41053
-    },
-    {
-      "epoch": 3.1388649960815798,
-      "grad_norm": 0.0019398679723963141,
-      "learning_rate": 0.0001999951403994557,
-      "loss": 46.0,
-      "step": 41054
-    },
-    {
-      "epoch": 3.1389414530649695,
-      "grad_norm": 0.0017691837856546044,
-      "learning_rate": 0.00019999514016265518,
-      "loss": 46.0,
-      "step": 41055
-    },
-    {
-      "epoch": 3.139017910048359,
-      "grad_norm": 0.00732882646843791,
-      "learning_rate": 0.0001999951399258489,
-      "loss": 46.0,
-      "step": 41056
-    },
-    {
-      "epoch": 3.1390943670317486,
-      "grad_norm": 0.0009454029495827854,
-      "learning_rate": 0.0001999951396890368,
-      "loss": 46.0,
-      "step": 41057
-    },
-    {
-      "epoch": 3.1391708240151384,
-      "grad_norm": 0.0028136447072029114,
-      "learning_rate": 0.00019999513945221896,
-      "loss": 46.0,
-      "step": 41058
-    },
-    {
-      "epoch": 3.139247280998528,
-      "grad_norm": 0.003275540191680193,
-      "learning_rate": 0.00019999513921539535,
-      "loss": 46.0,
-      "step": 41059
-    },
-    {
-      "epoch": 3.139323737981918,
-      "grad_norm": 0.007772588171064854,
-      "learning_rate": 0.000199995138978566,
-      "loss": 46.0,
-      "step": 41060
-    },
-    {
-      "epoch": 3.1394001949653076,
-      "grad_norm": 0.0006942058098502457,
-      "learning_rate": 0.00019999513874173087,
-      "loss": 46.0,
-      "step": 41061
-    },
-    {
-      "epoch": 3.1394766519486974,
-      "grad_norm": 0.004055796191096306,
-      "learning_rate": 0.00019999513850488994,
-      "loss": 46.0,
-      "step": 41062
-    },
-    {
-      "epoch": 3.139553108932087,
-      "grad_norm": 0.0016616667853668332,
-      "learning_rate": 0.00019999513826804327,
-      "loss": 46.0,
-      "step": 41063
-    },
-    {
-      "epoch": 3.139629565915477,
-      "grad_norm": 0.0011594309471547604,
-      "learning_rate": 0.00019999513803119082,
-      "loss": 46.0,
-      "step": 41064
-    },
-    {
-      "epoch": 3.1397060228988667,
-      "grad_norm": 0.004760695621371269,
-      "learning_rate": 0.0001999951377943326,
-      "loss": 46.0,
-      "step": 41065
-    },
-    {
-      "epoch": 3.1397824798822564,
-      "grad_norm": 0.005824300926178694,
-      "learning_rate": 0.0001999951375574686,
-      "loss": 46.0,
-      "step": 41066
-    },
-    {
-      "epoch": 3.1398589368656458,
-      "grad_norm": 0.0018791139591485262,
-      "learning_rate": 0.00019999513732059886,
-      "loss": 46.0,
-      "step": 41067
-    },
-    {
-      "epoch": 3.1399353938490355,
-      "grad_norm": 0.00265587423928082,
-      "learning_rate": 0.00019999513708372332,
-      "loss": 46.0,
-      "step": 41068
-    },
-    {
-      "epoch": 3.1400118508324253,
-      "grad_norm": 0.001935538137331605,
-      "learning_rate": 0.00019999513684684206,
-      "loss": 46.0,
-      "step": 41069
-    },
-    {
-      "epoch": 3.140088307815815,
-      "grad_norm": 0.0015491947997361422,
-      "learning_rate": 0.00019999513660995497,
-      "loss": 46.0,
-      "step": 41070
-    },
-    {
-      "epoch": 3.140164764799205,
-      "grad_norm": 0.001547043095342815,
-      "learning_rate": 0.00019999513637306216,
-      "loss": 46.0,
-      "step": 41071
-    },
-    {
-      "epoch": 3.1402412217825946,
-      "grad_norm": 0.0039647649973630905,
-      "learning_rate": 0.00019999513613616355,
-      "loss": 46.0,
-      "step": 41072
-    },
-    {
-      "epoch": 3.1403176787659843,
-      "grad_norm": 0.0009321686229668558,
-      "learning_rate": 0.00019999513589925917,
-      "loss": 46.0,
-      "step": 41073
-    },
-    {
-      "epoch": 3.140394135749374,
-      "grad_norm": 0.003700418397784233,
-      "learning_rate": 0.00019999513566234904,
-      "loss": 46.0,
-      "step": 41074
-    },
-    {
-      "epoch": 3.140470592732764,
-      "grad_norm": 0.0027197208255529404,
-      "learning_rate": 0.00019999513542543314,
-      "loss": 46.0,
-      "step": 41075
-    },
-    {
-      "epoch": 3.1405470497161536,
-      "grad_norm": 0.003884189995005727,
-      "learning_rate": 0.00019999513518851144,
-      "loss": 46.0,
-      "step": 41076
-    },
-    {
-      "epoch": 3.1406235066995434,
-      "grad_norm": 0.0023868188727647066,
-      "learning_rate": 0.00019999513495158402,
-      "loss": 46.0,
-      "step": 41077
-    },
-    {
-      "epoch": 3.1406999636829327,
-      "grad_norm": 0.00528392568230629,
-      "learning_rate": 0.0001999951347146508,
-      "loss": 46.0,
-      "step": 41078
-    },
-    {
-      "epoch": 3.1407764206663225,
-      "grad_norm": 0.004509489517658949,
-      "learning_rate": 0.00019999513447771177,
-      "loss": 46.0,
-      "step": 41079
-    },
-    {
-      "epoch": 3.140852877649712,
-      "grad_norm": 0.0011563922744244337,
-      "learning_rate": 0.00019999513424076706,
-      "loss": 46.0,
-      "step": 41080
-    },
-    {
-      "epoch": 3.140929334633102,
-      "grad_norm": 0.0015961665194481611,
-      "learning_rate": 0.00019999513400381652,
-      "loss": 46.0,
-      "step": 41081
-    },
-    {
-      "epoch": 3.1410057916164917,
-      "grad_norm": 0.0025906581431627274,
-      "learning_rate": 0.00019999513376686023,
-      "loss": 46.0,
-      "step": 41082
-    },
-    {
-      "epoch": 3.1410822485998815,
-      "grad_norm": 0.0007558197830803692,
-      "learning_rate": 0.00019999513352989817,
-      "loss": 46.0,
-      "step": 41083
-    },
-    {
-      "epoch": 3.1411587055832713,
-      "grad_norm": 0.0025289105251431465,
-      "learning_rate": 0.00019999513329293034,
-      "loss": 46.0,
-      "step": 41084
-    },
-    {
-      "epoch": 3.141235162566661,
-      "grad_norm": 0.0030896200332790613,
-      "learning_rate": 0.00019999513305595673,
-      "loss": 46.0,
-      "step": 41085
-    },
-    {
-      "epoch": 3.141311619550051,
-      "grad_norm": 0.0013852017000317574,
-      "learning_rate": 0.00019999513281897737,
-      "loss": 46.0,
-      "step": 41086
-    },
-    {
-      "epoch": 3.1413880765334405,
-      "grad_norm": 0.0028509381227195263,
-      "learning_rate": 0.00019999513258199222,
-      "loss": 46.0,
-      "step": 41087
-    },
-    {
-      "epoch": 3.14146453351683,
-      "grad_norm": 0.0009741692338138819,
-      "learning_rate": 0.00019999513234500134,
-      "loss": 46.0,
-      "step": 41088
-    },
-    {
-      "epoch": 3.1415409905002196,
-      "grad_norm": 0.002254907041788101,
-      "learning_rate": 0.00019999513210800464,
-      "loss": 46.0,
-      "step": 41089
-    },
-    {
-      "epoch": 3.1416174474836094,
-      "grad_norm": 0.0005364711978472769,
-      "learning_rate": 0.0001999951318710022,
-      "loss": 46.0,
-      "step": 41090
-    },
-    {
-      "epoch": 3.141693904466999,
-      "grad_norm": 0.0021251628641039133,
-      "learning_rate": 0.000199995131633994,
-      "loss": 46.0,
-      "step": 41091
-    },
-    {
-      "epoch": 3.141770361450389,
-      "grad_norm": 0.0028942017816007137,
-      "learning_rate": 0.00019999513139697998,
-      "loss": 46.0,
-      "step": 41092
-    },
-    {
-      "epoch": 3.1418468184337787,
-      "grad_norm": 0.005589932203292847,
-      "learning_rate": 0.00019999513115996024,
-      "loss": 46.0,
-      "step": 41093
-    },
-    {
-      "epoch": 3.1419232754171684,
-      "grad_norm": 0.0059248716570436954,
-      "learning_rate": 0.00019999513092293472,
-      "loss": 46.0,
-      "step": 41094
-    },
-    {
-      "epoch": 3.141999732400558,
-      "grad_norm": 0.003885081270709634,
-      "learning_rate": 0.00019999513068590344,
-      "loss": 46.0,
-      "step": 41095
-    },
-    {
-      "epoch": 3.142076189383948,
-      "grad_norm": 0.0005443011177703738,
-      "learning_rate": 0.00019999513044886635,
-      "loss": 46.0,
-      "step": 41096
-    },
-    {
-      "epoch": 3.1421526463673377,
-      "grad_norm": 0.0015941468300297856,
-      "learning_rate": 0.0001999951302118235,
-      "loss": 46.0,
-      "step": 41097
-    },
-    {
-      "epoch": 3.1422291033507275,
-      "grad_norm": 0.005403109826147556,
-      "learning_rate": 0.00019999512997477493,
-      "loss": 46.0,
-      "step": 41098
-    },
-    {
-      "epoch": 3.1423055603341172,
-      "grad_norm": 0.005095756612718105,
-      "learning_rate": 0.00019999512973772058,
-      "loss": 46.0,
-      "step": 41099
-    },
-    {
-      "epoch": 3.1423820173175065,
-      "grad_norm": 0.0008521788404323161,
-      "learning_rate": 0.00019999512950066045,
-      "loss": 46.0,
-      "step": 41100
-    },
-    {
-      "epoch": 3.1424584743008963,
-      "grad_norm": 0.001544449245557189,
-      "learning_rate": 0.00019999512926359452,
-      "loss": 46.0,
-      "step": 41101
-    },
-    {
-      "epoch": 3.142534931284286,
-      "grad_norm": 0.0031818135175853968,
-      "learning_rate": 0.00019999512902652285,
-      "loss": 46.0,
-      "step": 41102
-    },
-    {
-      "epoch": 3.142611388267676,
-      "grad_norm": 0.0026227463968098164,
-      "learning_rate": 0.0001999951287894454,
-      "loss": 46.0,
-      "step": 41103
-    },
-    {
-      "epoch": 3.1426878452510656,
-      "grad_norm": 0.002946611726656556,
-      "learning_rate": 0.0001999951285523622,
-      "loss": 46.0,
-      "step": 41104
-    },
-    {
-      "epoch": 3.1427643022344554,
-      "grad_norm": 0.0023916803766041994,
-      "learning_rate": 0.0001999951283152732,
-      "loss": 46.0,
-      "step": 41105
-    },
-    {
-      "epoch": 3.142840759217845,
-      "grad_norm": 0.0011243271874263883,
-      "learning_rate": 0.00019999512807817845,
-      "loss": 46.0,
-      "step": 41106
-    },
-    {
-      "epoch": 3.142917216201235,
-      "grad_norm": 0.001296074828132987,
-      "learning_rate": 0.00019999512784107793,
-      "loss": 46.0,
-      "step": 41107
-    },
-    {
-      "epoch": 3.1429936731846246,
-      "grad_norm": 0.004083739127963781,
-      "learning_rate": 0.00019999512760397165,
-      "loss": 46.0,
-      "step": 41108
-    },
-    {
-      "epoch": 3.1430701301680144,
-      "grad_norm": 0.0022604160476475954,
-      "learning_rate": 0.00019999512736685956,
-      "loss": 46.0,
-      "step": 41109
-    },
-    {
-      "epoch": 3.1431465871514037,
-      "grad_norm": 0.0016389890806749463,
-      "learning_rate": 0.00019999512712974172,
-      "loss": 46.0,
-      "step": 41110
-    },
-    {
-      "epoch": 3.1432230441347935,
-      "grad_norm": 0.004369684029370546,
-      "learning_rate": 0.00019999512689261814,
-      "loss": 46.0,
-      "step": 41111
-    },
-    {
-      "epoch": 3.1432995011181832,
-      "grad_norm": 0.0015991195105016232,
-      "learning_rate": 0.0001999951266554888,
-      "loss": 46.0,
-      "step": 41112
-    },
-    {
-      "epoch": 3.143375958101573,
-      "grad_norm": 0.0012750810710713267,
-      "learning_rate": 0.00019999512641835364,
-      "loss": 46.0,
-      "step": 41113
-    },
-    {
-      "epoch": 3.1434524150849628,
-      "grad_norm": 0.0019751156214624643,
-      "learning_rate": 0.00019999512618121273,
-      "loss": 46.0,
-      "step": 41114
-    },
-    {
-      "epoch": 3.1435288720683525,
-      "grad_norm": 0.00230652024038136,
-      "learning_rate": 0.00019999512594406603,
-      "loss": 46.0,
-      "step": 41115
-    },
-    {
-      "epoch": 3.1436053290517423,
-      "grad_norm": 0.00527595542371273,
-      "learning_rate": 0.0001999951257069136,
-      "loss": 46.0,
-      "step": 41116
-    },
-    {
-      "epoch": 3.143681786035132,
-      "grad_norm": 0.00411352189257741,
-      "learning_rate": 0.00019999512546975542,
-      "loss": 46.0,
-      "step": 41117
-    },
-    {
-      "epoch": 3.143758243018522,
-      "grad_norm": 0.0019018498715013266,
-      "learning_rate": 0.0001999951252325914,
-      "loss": 46.0,
-      "step": 41118
-    },
-    {
-      "epoch": 3.1438347000019116,
-      "grad_norm": 0.004346233326941729,
-      "learning_rate": 0.00019999512499542166,
-      "loss": 46.0,
-      "step": 41119
-    },
-    {
-      "epoch": 3.1439111569853013,
-      "grad_norm": 0.0029389483388513327,
-      "learning_rate": 0.00019999512475824615,
-      "loss": 46.0,
-      "step": 41120
-    },
-    {
-      "epoch": 3.143987613968691,
-      "grad_norm": 0.0029115909710526466,
-      "learning_rate": 0.00019999512452106486,
-      "loss": 46.0,
-      "step": 41121
-    },
-    {
-      "epoch": 3.1440640709520804,
-      "grad_norm": 0.002581561915576458,
-      "learning_rate": 0.0001999951242838778,
-      "loss": 46.0,
-      "step": 41122
-    },
-    {
-      "epoch": 3.14414052793547,
-      "grad_norm": 0.0009024101891554892,
-      "learning_rate": 0.00019999512404668494,
-      "loss": 46.0,
-      "step": 41123
-    },
-    {
-      "epoch": 3.14421698491886,
-      "grad_norm": 0.0016963593661785126,
-      "learning_rate": 0.00019999512380948638,
-      "loss": 46.0,
-      "step": 41124
-    },
-    {
-      "epoch": 3.1442934419022497,
-      "grad_norm": 0.001535413903184235,
-      "learning_rate": 0.00019999512357228198,
-      "loss": 46.0,
-      "step": 41125
-    },
-    {
-      "epoch": 3.1443698988856394,
-      "grad_norm": 0.0018477400299161673,
-      "learning_rate": 0.00019999512333507188,
-      "loss": 46.0,
-      "step": 41126
-    },
-    {
-      "epoch": 3.144446355869029,
-      "grad_norm": 0.0021108605433255434,
-      "learning_rate": 0.00019999512309785595,
-      "loss": 46.0,
-      "step": 41127
-    },
-    {
-      "epoch": 3.144522812852419,
-      "grad_norm": 0.001212080242112279,
-      "learning_rate": 0.00019999512286063428,
-      "loss": 46.0,
-      "step": 41128
-    },
-    {
-      "epoch": 3.1445992698358087,
-      "grad_norm": 0.002008714247494936,
-      "learning_rate": 0.00019999512262340683,
-      "loss": 46.0,
-      "step": 41129
-    },
-    {
-      "epoch": 3.1446757268191985,
-      "grad_norm": 0.003351228078827262,
-      "learning_rate": 0.0001999951223861736,
-      "loss": 46.0,
-      "step": 41130
-    },
-    {
-      "epoch": 3.1447521838025883,
-      "grad_norm": 0.0007999236695468426,
-      "learning_rate": 0.00019999512214893464,
-      "loss": 46.0,
-      "step": 41131
-    },
-    {
-      "epoch": 3.1448286407859776,
-      "grad_norm": 0.0031075961887836456,
-      "learning_rate": 0.00019999512191168988,
-      "loss": 46.0,
-      "step": 41132
-    },
-    {
-      "epoch": 3.1449050977693673,
-      "grad_norm": 0.0018060181755572557,
-      "learning_rate": 0.00019999512167443933,
-      "loss": 46.0,
-      "step": 41133
-    },
-    {
-      "epoch": 3.144981554752757,
-      "grad_norm": 0.0009503645705990493,
-      "learning_rate": 0.00019999512143718308,
-      "loss": 46.0,
-      "step": 41134
-    },
-    {
-      "epoch": 3.145058011736147,
-      "grad_norm": 0.002328176749870181,
-      "learning_rate": 0.00019999512119992102,
-      "loss": 46.0,
-      "step": 41135
-    },
-    {
-      "epoch": 3.1451344687195366,
-      "grad_norm": 0.0014271060936152935,
-      "learning_rate": 0.00019999512096265318,
-      "loss": 46.0,
-      "step": 41136
-    },
-    {
-      "epoch": 3.1452109257029264,
-      "grad_norm": 0.0014304692158475518,
-      "learning_rate": 0.00019999512072537958,
-      "loss": 46.0,
-      "step": 41137
-    },
-    {
-      "epoch": 3.145287382686316,
-      "grad_norm": 0.0035953198093920946,
-      "learning_rate": 0.00019999512048810022,
-      "loss": 46.0,
-      "step": 41138
-    },
-    {
-      "epoch": 3.145363839669706,
-      "grad_norm": 0.0022449269890785217,
-      "learning_rate": 0.0001999951202508151,
-      "loss": 46.0,
-      "step": 41139
-    },
-    {
-      "epoch": 3.1454402966530957,
-      "grad_norm": 0.0017370255663990974,
-      "learning_rate": 0.0001999951200135242,
-      "loss": 46.0,
-      "step": 41140
-    },
-    {
-      "epoch": 3.1455167536364854,
-      "grad_norm": 0.0034271995536983013,
-      "learning_rate": 0.0001999951197762275,
-      "loss": 46.0,
-      "step": 41141
-    },
-    {
-      "epoch": 3.145593210619875,
-      "grad_norm": 0.0015784569550305605,
-      "learning_rate": 0.00019999511953892505,
-      "loss": 46.0,
-      "step": 41142
-    },
-    {
-      "epoch": 3.145669667603265,
-      "grad_norm": 0.0010758456774055958,
-      "learning_rate": 0.00019999511930161686,
-      "loss": 46.0,
-      "step": 41143
-    },
-    {
-      "epoch": 3.1457461245866543,
-      "grad_norm": 0.0014098573010414839,
-      "learning_rate": 0.0001999951190643029,
-      "loss": 46.0,
-      "step": 41144
-    },
-    {
-      "epoch": 3.145822581570044,
-      "grad_norm": 0.00203301222063601,
-      "learning_rate": 0.00019999511882698312,
-      "loss": 46.0,
-      "step": 41145
-    },
-    {
-      "epoch": 3.145899038553434,
-      "grad_norm": 0.002582469955086708,
-      "learning_rate": 0.00019999511858965758,
-      "loss": 46.0,
-      "step": 41146
-    },
-    {
-      "epoch": 3.1459754955368235,
-      "grad_norm": 0.0018384659197181463,
-      "learning_rate": 0.00019999511835232632,
-      "loss": 46.0,
-      "step": 41147
-    },
-    {
-      "epoch": 3.1460519525202133,
-      "grad_norm": 0.0006012328085489571,
-      "learning_rate": 0.00019999511811498924,
-      "loss": 46.0,
-      "step": 41148
-    },
-    {
-      "epoch": 3.146128409503603,
-      "grad_norm": 0.001859766081906855,
-      "learning_rate": 0.00019999511787764643,
-      "loss": 46.0,
-      "step": 41149
-    },
-    {
-      "epoch": 3.146204866486993,
-      "grad_norm": 0.0029311501421034336,
-      "learning_rate": 0.00019999511764029782,
-      "loss": 46.0,
-      "step": 41150
-    },
-    {
-      "epoch": 3.1462813234703826,
-      "grad_norm": 0.0023800318595021963,
-      "learning_rate": 0.00019999511740294344,
-      "loss": 46.0,
-      "step": 41151
-    },
-    {
-      "epoch": 3.1463577804537723,
-      "grad_norm": 0.0013326131738722324,
-      "learning_rate": 0.00019999511716558332,
-      "loss": 46.0,
-      "step": 41152
-    },
-    {
-      "epoch": 3.146434237437162,
-      "grad_norm": 0.003274459857493639,
-      "learning_rate": 0.00019999511692821742,
-      "loss": 46.0,
-      "step": 41153
-    },
-    {
-      "epoch": 3.1465106944205514,
-      "grad_norm": 0.0009514266275800765,
-      "learning_rate": 0.00019999511669084575,
-      "loss": 46.0,
-      "step": 41154
-    },
-    {
-      "epoch": 3.146587151403941,
-      "grad_norm": 0.007105960976332426,
-      "learning_rate": 0.0001999951164534683,
-      "loss": 46.0,
-      "step": 41155
-    },
-    {
-      "epoch": 3.146663608387331,
-      "grad_norm": 0.008935816586017609,
-      "learning_rate": 0.00019999511621608508,
-      "loss": 46.0,
-      "step": 41156
-    },
-    {
-      "epoch": 3.1467400653707207,
-      "grad_norm": 0.004221817012876272,
-      "learning_rate": 0.00019999511597869611,
-      "loss": 46.0,
-      "step": 41157
-    },
-    {
-      "epoch": 3.1468165223541105,
-      "grad_norm": 0.0012877248227596283,
-      "learning_rate": 0.00019999511574130138,
-      "loss": 46.0,
-      "step": 41158
-    },
-    {
-      "epoch": 3.1468929793375002,
-      "grad_norm": 0.0021311098244041204,
-      "learning_rate": 0.00019999511550390084,
-      "loss": 46.0,
-      "step": 41159
-    },
-    {
-      "epoch": 3.14696943632089,
-      "grad_norm": 0.0009686232078820467,
-      "learning_rate": 0.00019999511526649455,
-      "loss": 46.0,
-      "step": 41160
-    },
-    {
-      "epoch": 3.1470458933042798,
-      "grad_norm": 0.0016935671446844935,
-      "learning_rate": 0.0001999951150290825,
-      "loss": 46.0,
-      "step": 41161
-    },
-    {
-      "epoch": 3.1471223502876695,
-      "grad_norm": 0.0019789987709373236,
-      "learning_rate": 0.00019999511479166466,
-      "loss": 46.0,
-      "step": 41162
-    },
-    {
-      "epoch": 3.1471988072710593,
-      "grad_norm": 0.00203468045219779,
-      "learning_rate": 0.00019999511455424108,
-      "loss": 46.0,
-      "step": 41163
-    },
-    {
-      "epoch": 3.147275264254449,
-      "grad_norm": 0.000643292733002454,
-      "learning_rate": 0.0001999951143168117,
-      "loss": 46.0,
-      "step": 41164
-    },
-    {
-      "epoch": 3.147351721237839,
-      "grad_norm": 0.003530516754835844,
-      "learning_rate": 0.00019999511407937657,
-      "loss": 46.0,
-      "step": 41165
-    },
-    {
-      "epoch": 3.147428178221228,
-      "grad_norm": 0.0019623476546257734,
-      "learning_rate": 0.00019999511384193567,
-      "loss": 46.0,
-      "step": 41166
-    },
-    {
-      "epoch": 3.147504635204618,
-      "grad_norm": 0.00207833806052804,
-      "learning_rate": 0.000199995113604489,
-      "loss": 46.0,
-      "step": 41167
-    },
-    {
-      "epoch": 3.1475810921880076,
-      "grad_norm": 0.005475122947245836,
-      "learning_rate": 0.00019999511336703653,
-      "loss": 46.0,
-      "step": 41168
-    },
-    {
-      "epoch": 3.1476575491713974,
-      "grad_norm": 0.0017534289509057999,
-      "learning_rate": 0.00019999511312957834,
-      "loss": 46.0,
-      "step": 41169
-    },
-    {
-      "epoch": 3.147734006154787,
-      "grad_norm": 0.001810555113479495,
-      "learning_rate": 0.00019999511289211434,
-      "loss": 46.0,
-      "step": 41170
-    },
-    {
-      "epoch": 3.147810463138177,
-      "grad_norm": 0.0010964962420985103,
-      "learning_rate": 0.0001999951126546446,
-      "loss": 46.0,
-      "step": 41171
-    },
-    {
-      "epoch": 3.1478869201215667,
-      "grad_norm": 0.0033140238374471664,
-      "learning_rate": 0.00019999511241716912,
-      "loss": 46.0,
-      "step": 41172
-    },
-    {
-      "epoch": 3.1479633771049564,
-      "grad_norm": 0.0026049206499010324,
-      "learning_rate": 0.0001999951121796878,
-      "loss": 46.0,
-      "step": 41173
-    },
-    {
-      "epoch": 3.148039834088346,
-      "grad_norm": 0.003027221653610468,
-      "learning_rate": 0.00019999511194220075,
-      "loss": 46.0,
-      "step": 41174
-    },
-    {
-      "epoch": 3.148116291071736,
-      "grad_norm": 0.0013067773543298244,
-      "learning_rate": 0.00019999511170470791,
-      "loss": 46.0,
-      "step": 41175
-    },
-    {
-      "epoch": 3.1481927480551253,
-      "grad_norm": 0.0020483662374317646,
-      "learning_rate": 0.00019999511146720934,
-      "loss": 46.0,
-      "step": 41176
-    },
-    {
-      "epoch": 3.148269205038515,
-      "grad_norm": 0.0029803512152284384,
-      "learning_rate": 0.00019999511122970496,
-      "loss": 46.0,
-      "step": 41177
-    },
-    {
-      "epoch": 3.148345662021905,
-      "grad_norm": 0.001941710477694869,
-      "learning_rate": 0.00019999511099219483,
-      "loss": 46.0,
-      "step": 41178
-    },
-    {
-      "epoch": 3.1484221190052946,
-      "grad_norm": 0.00592011259868741,
-      "learning_rate": 0.0001999951107546789,
-      "loss": 46.0,
-      "step": 41179
-    },
-    {
-      "epoch": 3.1484985759886843,
-      "grad_norm": 0.0028123133815824986,
-      "learning_rate": 0.00019999511051715726,
-      "loss": 46.0,
-      "step": 41180
-    },
-    {
-      "epoch": 3.148575032972074,
-      "grad_norm": 0.0011448734439909458,
-      "learning_rate": 0.00019999511027962981,
-      "loss": 46.0,
-      "step": 41181
-    },
-    {
-      "epoch": 3.148651489955464,
-      "grad_norm": 0.002779376693069935,
-      "learning_rate": 0.00019999511004209662,
-      "loss": 46.0,
-      "step": 41182
-    },
-    {
-      "epoch": 3.1487279469388536,
-      "grad_norm": 0.0026455887127667665,
-      "learning_rate": 0.00019999510980455763,
-      "loss": 46.0,
-      "step": 41183
-    },
-    {
-      "epoch": 3.1488044039222434,
-      "grad_norm": 0.002642741659656167,
-      "learning_rate": 0.00019999510956701286,
-      "loss": 46.0,
-      "step": 41184
-    },
-    {
-      "epoch": 3.148880860905633,
-      "grad_norm": 0.001011073705740273,
-      "learning_rate": 0.00019999510932946235,
-      "loss": 46.0,
-      "step": 41185
-    },
-    {
-      "epoch": 3.148957317889023,
-      "grad_norm": 0.0019699865952134132,
-      "learning_rate": 0.00019999510909190607,
-      "loss": 46.0,
-      "step": 41186
-    },
-    {
-      "epoch": 3.149033774872412,
-      "grad_norm": 0.0011443350231274962,
-      "learning_rate": 0.000199995108854344,
-      "loss": 46.0,
-      "step": 41187
-    },
-    {
-      "epoch": 3.149110231855802,
-      "grad_norm": 0.0020341400522738695,
-      "learning_rate": 0.00019999510861677618,
-      "loss": 46.0,
-      "step": 41188
-    },
-    {
-      "epoch": 3.1491866888391917,
-      "grad_norm": 0.0029956973157823086,
-      "learning_rate": 0.0001999951083792026,
-      "loss": 46.0,
-      "step": 41189
-    },
-    {
-      "epoch": 3.1492631458225815,
-      "grad_norm": 0.001325927209109068,
-      "learning_rate": 0.00019999510814162325,
-      "loss": 46.0,
-      "step": 41190
-    },
-    {
-      "epoch": 3.1493396028059712,
-      "grad_norm": 0.0018731832969933748,
-      "learning_rate": 0.0001999951079040381,
-      "loss": 46.0,
-      "step": 41191
-    },
-    {
-      "epoch": 3.149416059789361,
-      "grad_norm": 0.001182476058602333,
-      "learning_rate": 0.0001999951076664472,
-      "loss": 46.0,
-      "step": 41192
-    },
-    {
-      "epoch": 3.1494925167727508,
-      "grad_norm": 0.003146679373458028,
-      "learning_rate": 0.00019999510742885052,
-      "loss": 46.0,
-      "step": 41193
-    },
-    {
-      "epoch": 3.1495689737561405,
-      "grad_norm": 0.0011848483700305223,
-      "learning_rate": 0.00019999510719124808,
-      "loss": 46.0,
-      "step": 41194
-    },
-    {
-      "epoch": 3.1496454307395303,
-      "grad_norm": 0.002009914955124259,
-      "learning_rate": 0.00019999510695363989,
-      "loss": 46.0,
-      "step": 41195
-    },
-    {
-      "epoch": 3.14972188772292,
-      "grad_norm": 0.0014789466513320804,
-      "learning_rate": 0.0001999951067160259,
-      "loss": 46.0,
-      "step": 41196
-    },
-    {
-      "epoch": 3.14979834470631,
-      "grad_norm": 0.0009771896293386817,
-      "learning_rate": 0.00019999510647840616,
-      "loss": 46.0,
-      "step": 41197
-    },
-    {
-      "epoch": 3.149874801689699,
-      "grad_norm": 0.0030318363569676876,
-      "learning_rate": 0.00019999510624078064,
-      "loss": 46.0,
-      "step": 41198
-    },
-    {
-      "epoch": 3.149951258673089,
-      "grad_norm": 0.0025786079932004213,
-      "learning_rate": 0.00019999510600314936,
-      "loss": 46.0,
-      "step": 41199
-    },
-    {
-      "epoch": 3.1500277156564787,
-      "grad_norm": 0.0021388635504990816,
-      "learning_rate": 0.00019999510576551227,
-      "loss": 46.0,
-      "step": 41200
-    },
-    {
-      "epoch": 3.1501041726398684,
-      "grad_norm": 0.0012837856775149703,
-      "learning_rate": 0.00019999510552786947,
-      "loss": 46.0,
-      "step": 41201
-    },
-    {
-      "epoch": 3.150180629623258,
-      "grad_norm": 0.0022854323033243418,
-      "learning_rate": 0.00019999510529022086,
-      "loss": 46.0,
-      "step": 41202
-    },
-    {
-      "epoch": 3.150257086606648,
-      "grad_norm": 0.003339572111144662,
-      "learning_rate": 0.00019999510505256648,
-      "loss": 46.0,
-      "step": 41203
-    },
-    {
-      "epoch": 3.1503335435900377,
-      "grad_norm": 0.0017678368603810668,
-      "learning_rate": 0.0001999951048149064,
-      "loss": 46.0,
-      "step": 41204
-    },
-    {
-      "epoch": 3.1504100005734275,
-      "grad_norm": 0.002714118454605341,
-      "learning_rate": 0.0001999951045772405,
-      "loss": 46.0,
-      "step": 41205
-    },
-    {
-      "epoch": 3.150486457556817,
-      "grad_norm": 0.0020404579117894173,
-      "learning_rate": 0.00019999510433956882,
-      "loss": 46.0,
-      "step": 41206
-    },
-    {
-      "epoch": 3.150562914540207,
-      "grad_norm": 0.0014239432057365775,
-      "learning_rate": 0.00019999510410189134,
-      "loss": 46.0,
-      "step": 41207
-    },
-    {
-      "epoch": 3.1506393715235967,
-      "grad_norm": 0.001548121334053576,
-      "learning_rate": 0.00019999510386420815,
-      "loss": 46.0,
-      "step": 41208
-    },
-    {
-      "epoch": 3.150715828506986,
-      "grad_norm": 0.0017139689298346639,
-      "learning_rate": 0.00019999510362651916,
-      "loss": 46.0,
-      "step": 41209
-    },
-    {
-      "epoch": 3.150792285490376,
-      "grad_norm": 0.002566085895523429,
-      "learning_rate": 0.00019999510338882442,
-      "loss": 46.0,
-      "step": 41210
-    },
-    {
-      "epoch": 3.1508687424737656,
-      "grad_norm": 0.0025460158940404654,
-      "learning_rate": 0.0001999951031511239,
-      "loss": 46.0,
-      "step": 41211
-    },
-    {
-      "epoch": 3.1509451994571553,
-      "grad_norm": 0.006416479591280222,
-      "learning_rate": 0.00019999510291341763,
-      "loss": 46.0,
-      "step": 41212
-    },
-    {
-      "epoch": 3.151021656440545,
-      "grad_norm": 0.007463107351213694,
-      "learning_rate": 0.00019999510267570557,
-      "loss": 46.0,
-      "step": 41213
-    },
-    {
-      "epoch": 3.151098113423935,
-      "grad_norm": 0.0008747376268729568,
-      "learning_rate": 0.00019999510243798776,
-      "loss": 46.0,
-      "step": 41214
-    },
-    {
-      "epoch": 3.1511745704073246,
-      "grad_norm": 0.002939442405477166,
-      "learning_rate": 0.00019999510220026413,
-      "loss": 46.0,
-      "step": 41215
-    },
-    {
-      "epoch": 3.1512510273907144,
-      "grad_norm": 0.001430977019481361,
-      "learning_rate": 0.00019999510196253478,
-      "loss": 46.0,
-      "step": 41216
-    },
-    {
-      "epoch": 3.151327484374104,
-      "grad_norm": 0.0006753754569217563,
-      "learning_rate": 0.00019999510172479968,
-      "loss": 46.0,
-      "step": 41217
-    },
-    {
-      "epoch": 3.151403941357494,
-      "grad_norm": 0.0031671279575675726,
-      "learning_rate": 0.00019999510148705876,
-      "loss": 46.0,
-      "step": 41218
-    },
-    {
-      "epoch": 3.1514803983408832,
-      "grad_norm": 0.0008521328563801944,
-      "learning_rate": 0.0001999951012493121,
-      "loss": 46.0,
-      "step": 41219
-    },
-    {
-      "epoch": 3.151556855324273,
-      "grad_norm": 0.0023200339637696743,
-      "learning_rate": 0.00019999510101155964,
-      "loss": 46.0,
-      "step": 41220
-    },
-    {
-      "epoch": 3.1516333123076627,
-      "grad_norm": 0.0019352763192728162,
-      "learning_rate": 0.00019999510077380145,
-      "loss": 46.0,
-      "step": 41221
-    },
-    {
-      "epoch": 3.1517097692910525,
-      "grad_norm": 0.0011031129397451878,
-      "learning_rate": 0.00019999510053603746,
-      "loss": 46.0,
-      "step": 41222
-    },
-    {
-      "epoch": 3.1517862262744423,
-      "grad_norm": 0.0011672734981402755,
-      "learning_rate": 0.00019999510029826772,
-      "loss": 46.0,
-      "step": 41223
-    },
-    {
-      "epoch": 3.151862683257832,
-      "grad_norm": 0.0012146474327892065,
-      "learning_rate": 0.0001999951000604922,
-      "loss": 46.0,
-      "step": 41224
-    },
-    {
-      "epoch": 3.151939140241222,
-      "grad_norm": 0.0011745108058676124,
-      "learning_rate": 0.00019999509982271093,
-      "loss": 46.0,
-      "step": 41225
-    },
-    {
-      "epoch": 3.1520155972246116,
-      "grad_norm": 0.0013356372946873307,
-      "learning_rate": 0.00019999509958492387,
-      "loss": 46.0,
-      "step": 41226
-    },
-    {
-      "epoch": 3.1520920542080013,
-      "grad_norm": 0.005240879021584988,
-      "learning_rate": 0.00019999509934713104,
-      "loss": 46.0,
-      "step": 41227
-    },
-    {
-      "epoch": 3.152168511191391,
-      "grad_norm": 0.0012554834829643369,
-      "learning_rate": 0.00019999509910933246,
-      "loss": 46.0,
-      "step": 41228
-    },
-    {
-      "epoch": 3.152244968174781,
-      "grad_norm": 0.0016742480220273137,
-      "learning_rate": 0.0001999950988715281,
-      "loss": 46.0,
-      "step": 41229
-    },
-    {
-      "epoch": 3.1523214251581706,
-      "grad_norm": 0.001541724312119186,
-      "learning_rate": 0.00019999509863371796,
-      "loss": 46.0,
-      "step": 41230
-    },
-    {
-      "epoch": 3.15239788214156,
-      "grad_norm": 0.002407388063147664,
-      "learning_rate": 0.00019999509839590206,
-      "loss": 46.0,
-      "step": 41231
-    },
-    {
-      "epoch": 3.1524743391249497,
-      "grad_norm": 0.0048097348771989346,
-      "learning_rate": 0.0001999950981580804,
-      "loss": 46.0,
-      "step": 41232
-    },
-    {
-      "epoch": 3.1525507961083394,
-      "grad_norm": 0.0010083505185320973,
-      "learning_rate": 0.00019999509792025295,
-      "loss": 46.0,
-      "step": 41233
-    },
-    {
-      "epoch": 3.152627253091729,
-      "grad_norm": 0.0017390126595273614,
-      "learning_rate": 0.00019999509768241973,
-      "loss": 46.0,
-      "step": 41234
-    },
-    {
-      "epoch": 3.152703710075119,
-      "grad_norm": 0.00044358987361192703,
-      "learning_rate": 0.0001999950974445808,
-      "loss": 46.0,
-      "step": 41235
-    },
-    {
-      "epoch": 3.1527801670585087,
-      "grad_norm": 0.00310742505826056,
-      "learning_rate": 0.00019999509720673603,
-      "loss": 46.0,
-      "step": 41236
-    },
-    {
-      "epoch": 3.1528566240418985,
-      "grad_norm": 0.0019872100092470646,
-      "learning_rate": 0.00019999509696888552,
-      "loss": 46.0,
-      "step": 41237
-    },
-    {
-      "epoch": 3.1529330810252882,
-      "grad_norm": 0.0009636054164730012,
-      "learning_rate": 0.0001999950967310292,
-      "loss": 46.0,
-      "step": 41238
-    },
-    {
-      "epoch": 3.153009538008678,
-      "grad_norm": 0.0019592433236539364,
-      "learning_rate": 0.00019999509649316717,
-      "loss": 46.0,
-      "step": 41239
-    },
-    {
-      "epoch": 3.1530859949920678,
-      "grad_norm": 0.0014720330946147442,
-      "learning_rate": 0.00019999509625529934,
-      "loss": 46.0,
-      "step": 41240
-    },
-    {
-      "epoch": 3.153162451975457,
-      "grad_norm": 0.006337042897939682,
-      "learning_rate": 0.00019999509601742577,
-      "loss": 46.0,
-      "step": 41241
-    },
-    {
-      "epoch": 3.153238908958847,
-      "grad_norm": 0.0022629695013165474,
-      "learning_rate": 0.00019999509577954642,
-      "loss": 46.0,
-      "step": 41242
-    },
-    {
-      "epoch": 3.1533153659422366,
-      "grad_norm": 0.0011800449574366212,
-      "learning_rate": 0.00019999509554166126,
-      "loss": 46.0,
-      "step": 41243
-    },
-    {
-      "epoch": 3.1533918229256264,
-      "grad_norm": 0.0021426088642328978,
-      "learning_rate": 0.00019999509530377037,
-      "loss": 46.0,
-      "step": 41244
-    },
-    {
-      "epoch": 3.153468279909016,
-      "grad_norm": 0.002159228315576911,
-      "learning_rate": 0.0001999950950658737,
-      "loss": 46.0,
-      "step": 41245
-    },
-    {
-      "epoch": 3.153544736892406,
-      "grad_norm": 0.0019955765455961227,
-      "learning_rate": 0.00019999509482797128,
-      "loss": 46.0,
-      "step": 41246
-    },
-    {
-      "epoch": 3.1536211938757956,
-      "grad_norm": 0.0027953071985393763,
-      "learning_rate": 0.00019999509459006304,
-      "loss": 46.0,
-      "step": 41247
-    },
-    {
-      "epoch": 3.1536976508591854,
-      "grad_norm": 0.002594755031168461,
-      "learning_rate": 0.0001999950943521491,
-      "loss": 46.0,
-      "step": 41248
-    },
-    {
-      "epoch": 3.153774107842575,
-      "grad_norm": 0.0017028666334226727,
-      "learning_rate": 0.0001999950941142293,
-      "loss": 46.0,
-      "step": 41249
-    },
-    {
-      "epoch": 3.153850564825965,
-      "grad_norm": 0.0015072169480845332,
-      "learning_rate": 0.00019999509387630382,
-      "loss": 46.0,
-      "step": 41250
-    },
-    {
-      "epoch": 3.1539270218093547,
-      "grad_norm": 0.0035794342402368784,
-      "learning_rate": 0.00019999509363837254,
-      "loss": 46.0,
-      "step": 41251
-    },
-    {
-      "epoch": 3.1540034787927445,
-      "grad_norm": 0.000742486969102174,
-      "learning_rate": 0.00019999509340043548,
-      "loss": 46.0,
-      "step": 41252
-    },
-    {
-      "epoch": 3.1540799357761338,
-      "grad_norm": 0.003688682336360216,
-      "learning_rate": 0.00019999509316249268,
-      "loss": 46.0,
-      "step": 41253
-    },
-    {
-      "epoch": 3.1541563927595235,
-      "grad_norm": 0.01042461208999157,
-      "learning_rate": 0.00019999509292454408,
-      "loss": 46.0,
-      "step": 41254
-    },
-    {
-      "epoch": 3.1542328497429133,
-      "grad_norm": 0.008628328330814838,
-      "learning_rate": 0.00019999509268658973,
-      "loss": 46.0,
-      "step": 41255
-    },
-    {
-      "epoch": 3.154309306726303,
-      "grad_norm": 0.001481702085584402,
-      "learning_rate": 0.00019999509244862957,
-      "loss": 46.0,
-      "step": 41256
-    },
-    {
-      "epoch": 3.154385763709693,
-      "grad_norm": 0.004218224436044693,
-      "learning_rate": 0.00019999509221066368,
-      "loss": 46.0,
-      "step": 41257
-    },
-    {
-      "epoch": 3.1544622206930826,
-      "grad_norm": 0.0014666811330243945,
-      "learning_rate": 0.000199995091972692,
-      "loss": 46.0,
-      "step": 41258
-    },
-    {
-      "epoch": 3.1545386776764723,
-      "grad_norm": 0.0034756611566990614,
-      "learning_rate": 0.0001999950917347146,
-      "loss": 46.0,
-      "step": 41259
-    },
-    {
-      "epoch": 3.154615134659862,
-      "grad_norm": 0.0010898802429437637,
-      "learning_rate": 0.00019999509149673137,
-      "loss": 46.0,
-      "step": 41260
-    },
-    {
-      "epoch": 3.154691591643252,
-      "grad_norm": 0.0044989800080657005,
-      "learning_rate": 0.0001999950912587424,
-      "loss": 46.0,
-      "step": 41261
-    },
-    {
-      "epoch": 3.1547680486266416,
-      "grad_norm": 0.0016769011272117496,
-      "learning_rate": 0.00019999509102074767,
-      "loss": 46.0,
-      "step": 41262
-    },
-    {
-      "epoch": 3.154844505610031,
-      "grad_norm": 0.0018666343530640006,
-      "learning_rate": 0.00019999509078274714,
-      "loss": 46.0,
-      "step": 41263
-    },
-    {
-      "epoch": 3.1549209625934207,
-      "grad_norm": 0.0027444446459412575,
-      "learning_rate": 0.00019999509054474085,
-      "loss": 46.0,
-      "step": 41264
-    },
-    {
-      "epoch": 3.1549974195768105,
-      "grad_norm": 0.00113389128819108,
-      "learning_rate": 0.00019999509030672882,
-      "loss": 46.0,
-      "step": 41265
-    },
-    {
-      "epoch": 3.1550738765602,
-      "grad_norm": 0.0017235310515388846,
-      "learning_rate": 0.000199995090068711,
-      "loss": 46.0,
-      "step": 41266
-    },
-    {
-      "epoch": 3.15515033354359,
-      "grad_norm": 0.001434971927665174,
-      "learning_rate": 0.0001999950898306874,
-      "loss": 46.0,
-      "step": 41267
-    },
-    {
-      "epoch": 3.1552267905269797,
-      "grad_norm": 0.002379936631768942,
-      "learning_rate": 0.000199995089592658,
-      "loss": 46.0,
-      "step": 41268
-    },
-    {
-      "epoch": 3.1553032475103695,
-      "grad_norm": 0.0027696508914232254,
-      "learning_rate": 0.00019999508935462292,
-      "loss": 46.0,
-      "step": 41269
-    },
-    {
-      "epoch": 3.1553797044937593,
-      "grad_norm": 0.0005561787984333932,
-      "learning_rate": 0.000199995089116582,
-      "loss": 46.0,
-      "step": 41270
-    },
-    {
-      "epoch": 3.155456161477149,
-      "grad_norm": 0.0037807845510542393,
-      "learning_rate": 0.00019999508887853535,
-      "loss": 46.0,
-      "step": 41271
-    },
-    {
-      "epoch": 3.155532618460539,
-      "grad_norm": 0.006651694420725107,
-      "learning_rate": 0.0001999950886404829,
-      "loss": 46.0,
-      "step": 41272
-    },
-    {
-      "epoch": 3.1556090754439285,
-      "grad_norm": 0.004104775842279196,
-      "learning_rate": 0.00019999508840242472,
-      "loss": 46.0,
-      "step": 41273
-    },
-    {
-      "epoch": 3.1556855324273183,
-      "grad_norm": 0.0015994838904589415,
-      "learning_rate": 0.00019999508816436073,
-      "loss": 46.0,
-      "step": 41274
-    },
-    {
-      "epoch": 3.1557619894107076,
-      "grad_norm": 0.0012926486087962985,
-      "learning_rate": 0.000199995087926291,
-      "loss": 46.0,
-      "step": 41275
-    },
-    {
-      "epoch": 3.1558384463940974,
-      "grad_norm": 0.003736570244655013,
-      "learning_rate": 0.00019999508768821545,
-      "loss": 46.0,
-      "step": 41276
-    },
-    {
-      "epoch": 3.155914903377487,
-      "grad_norm": 0.001920853741466999,
-      "learning_rate": 0.0001999950874501342,
-      "loss": 46.0,
-      "step": 41277
-    },
-    {
-      "epoch": 3.155991360360877,
-      "grad_norm": 0.0011635113041847944,
-      "learning_rate": 0.00019999508721204717,
-      "loss": 46.0,
-      "step": 41278
-    },
-    {
-      "epoch": 3.1560678173442667,
-      "grad_norm": 0.003280497156083584,
-      "learning_rate": 0.00019999508697395434,
-      "loss": 46.0,
-      "step": 41279
-    },
-    {
-      "epoch": 3.1561442743276564,
-      "grad_norm": 0.004014330450445414,
-      "learning_rate": 0.00019999508673585573,
-      "loss": 46.0,
-      "step": 41280
-    },
-    {
-      "epoch": 3.156220731311046,
-      "grad_norm": 0.0026616144459694624,
-      "learning_rate": 0.00019999508649775138,
-      "loss": 46.0,
-      "step": 41281
-    },
-    {
-      "epoch": 3.156297188294436,
-      "grad_norm": 0.005476745311170816,
-      "learning_rate": 0.00019999508625964126,
-      "loss": 46.0,
-      "step": 41282
-    },
-    {
-      "epoch": 3.1563736452778257,
-      "grad_norm": 0.0017588009359315038,
-      "learning_rate": 0.00019999508602152534,
-      "loss": 46.0,
-      "step": 41283
-    },
-    {
-      "epoch": 3.1564501022612155,
-      "grad_norm": 0.0006979173049330711,
-      "learning_rate": 0.0001999950857834037,
-      "loss": 46.0,
-      "step": 41284
-    },
-    {
-      "epoch": 3.156526559244605,
-      "grad_norm": 0.0018476679688319564,
-      "learning_rate": 0.00019999508554527625,
-      "loss": 46.0,
-      "step": 41285
-    },
-    {
-      "epoch": 3.1566030162279946,
-      "grad_norm": 0.004086425062268972,
-      "learning_rate": 0.00019999508530714306,
-      "loss": 46.0,
-      "step": 41286
-    },
-    {
-      "epoch": 3.1566794732113843,
-      "grad_norm": 0.0023363095242530107,
-      "learning_rate": 0.00019999508506900407,
-      "loss": 46.0,
-      "step": 41287
-    },
-    {
-      "epoch": 3.156755930194774,
-      "grad_norm": 0.005558275617659092,
-      "learning_rate": 0.00019999508483085934,
-      "loss": 46.0,
-      "step": 41288
-    },
-    {
-      "epoch": 3.156832387178164,
-      "grad_norm": 0.004917372018098831,
-      "learning_rate": 0.00019999508459270883,
-      "loss": 46.0,
-      "step": 41289
-    },
-    {
-      "epoch": 3.1569088441615536,
-      "grad_norm": 0.0015837918035686016,
-      "learning_rate": 0.00019999508435455254,
-      "loss": 46.0,
-      "step": 41290
-    },
-    {
-      "epoch": 3.1569853011449434,
-      "grad_norm": 0.003155802609398961,
-      "learning_rate": 0.0001999950841163905,
-      "loss": 46.0,
-      "step": 41291
-    },
-    {
-      "epoch": 3.157061758128333,
-      "grad_norm": 0.0019145618425682187,
-      "learning_rate": 0.00019999508387822266,
-      "loss": 46.0,
-      "step": 41292
-    },
-    {
-      "epoch": 3.157138215111723,
-      "grad_norm": 0.0023620694410055876,
-      "learning_rate": 0.00019999508364004908,
-      "loss": 46.0,
-      "step": 41293
-    },
-    {
-      "epoch": 3.1572146720951126,
-      "grad_norm": 0.005915533285588026,
-      "learning_rate": 0.00019999508340186973,
-      "loss": 46.0,
-      "step": 41294
-    },
-    {
-      "epoch": 3.1572911290785024,
-      "grad_norm": 0.0026202548760920763,
-      "learning_rate": 0.0001999950831636846,
-      "loss": 46.0,
-      "step": 41295
-    },
-    {
-      "epoch": 3.157367586061892,
-      "grad_norm": 0.003385018790140748,
-      "learning_rate": 0.0001999950829254937,
-      "loss": 46.0,
-      "step": 41296
-    },
-    {
-      "epoch": 3.1574440430452815,
-      "grad_norm": 0.0010866944212466478,
-      "learning_rate": 0.00019999508268729702,
-      "loss": 46.0,
-      "step": 41297
-    },
-    {
-      "epoch": 3.1575205000286712,
-      "grad_norm": 0.0026685951743274927,
-      "learning_rate": 0.0001999950824490946,
-      "loss": 46.0,
-      "step": 41298
-    },
-    {
-      "epoch": 3.157596957012061,
-      "grad_norm": 0.0032722107134759426,
-      "learning_rate": 0.00019999508221088639,
-      "loss": 46.0,
-      "step": 41299
-    },
-    {
-      "epoch": 3.1576734139954508,
-      "grad_norm": 0.0006499799783341587,
-      "learning_rate": 0.00019999508197267242,
-      "loss": 46.0,
-      "step": 41300
-    },
-    {
-      "epoch": 3.1577498709788405,
-      "grad_norm": 0.00191175052896142,
-      "learning_rate": 0.0001999950817344527,
-      "loss": 46.0,
-      "step": 41301
-    },
-    {
-      "epoch": 3.1578263279622303,
-      "grad_norm": 0.0019544349052011967,
-      "learning_rate": 0.00019999508149622718,
-      "loss": 46.0,
-      "step": 41302
-    },
-    {
-      "epoch": 3.15790278494562,
-      "grad_norm": 0.0037779954727739096,
-      "learning_rate": 0.0001999950812579959,
-      "loss": 46.0,
-      "step": 41303
-    },
-    {
-      "epoch": 3.15797924192901,
-      "grad_norm": 0.001928371493704617,
-      "learning_rate": 0.00019999508101975884,
-      "loss": 46.0,
-      "step": 41304
-    },
-    {
-      "epoch": 3.1580556989123996,
-      "grad_norm": 0.0013177570654079318,
-      "learning_rate": 0.00019999508078151604,
-      "loss": 46.0,
-      "step": 41305
-    },
-    {
-      "epoch": 3.1581321558957893,
-      "grad_norm": 0.0021362965926527977,
-      "learning_rate": 0.00019999508054326746,
-      "loss": 46.0,
-      "step": 41306
-    },
-    {
-      "epoch": 3.1582086128791786,
-      "grad_norm": 0.0010499494383111596,
-      "learning_rate": 0.00019999508030501309,
-      "loss": 46.0,
-      "step": 41307
-    },
-    {
-      "epoch": 3.1582850698625684,
-      "grad_norm": 0.0016806864878162742,
-      "learning_rate": 0.00019999508006675296,
-      "loss": 46.0,
-      "step": 41308
-    },
-    {
-      "epoch": 3.158361526845958,
-      "grad_norm": 0.0014857127098366618,
-      "learning_rate": 0.00019999507982848707,
-      "loss": 46.0,
-      "step": 41309
-    },
-    {
-      "epoch": 3.158437983829348,
-      "grad_norm": 0.007579101249575615,
-      "learning_rate": 0.0001999950795902154,
-      "loss": 46.0,
-      "step": 41310
-    },
-    {
-      "epoch": 3.1585144408127377,
-      "grad_norm": 0.004320272710174322,
-      "learning_rate": 0.00019999507935193798,
-      "loss": 46.0,
-      "step": 41311
-    },
-    {
-      "epoch": 3.1585908977961275,
-      "grad_norm": 0.0030708054546266794,
-      "learning_rate": 0.00019999507911365477,
-      "loss": 46.0,
-      "step": 41312
-    },
-    {
-      "epoch": 3.158667354779517,
-      "grad_norm": 0.002287672832608223,
-      "learning_rate": 0.0001999950788753658,
-      "loss": 46.0,
-      "step": 41313
-    },
-    {
-      "epoch": 3.158743811762907,
-      "grad_norm": 0.0019517750479280949,
-      "learning_rate": 0.00019999507863707104,
-      "loss": 46.0,
-      "step": 41314
-    },
-    {
-      "epoch": 3.1588202687462967,
-      "grad_norm": 0.004568958654999733,
-      "learning_rate": 0.00019999507839877053,
-      "loss": 46.0,
-      "step": 41315
-    },
-    {
-      "epoch": 3.1588967257296865,
-      "grad_norm": 0.0017045927233994007,
-      "learning_rate": 0.00019999507816046425,
-      "loss": 46.0,
-      "step": 41316
-    },
-    {
-      "epoch": 3.1589731827130763,
-      "grad_norm": 0.0017688111402094364,
-      "learning_rate": 0.00019999507792215222,
-      "loss": 46.0,
-      "step": 41317
-    },
-    {
-      "epoch": 3.1590496396964656,
-      "grad_norm": 0.003358157118782401,
-      "learning_rate": 0.00019999507768383442,
-      "loss": 46.0,
-      "step": 41318
-    },
-    {
-      "epoch": 3.1591260966798553,
-      "grad_norm": 0.0024869979824870825,
-      "learning_rate": 0.0001999950774455108,
-      "loss": 46.0,
-      "step": 41319
-    },
-    {
-      "epoch": 3.159202553663245,
-      "grad_norm": 0.000876505917403847,
-      "learning_rate": 0.00019999507720718144,
-      "loss": 46.0,
-      "step": 41320
-    },
-    {
-      "epoch": 3.159279010646635,
-      "grad_norm": 0.0007721567526459694,
-      "learning_rate": 0.00019999507696884635,
-      "loss": 46.0,
-      "step": 41321
-    },
-    {
-      "epoch": 3.1593554676300246,
-      "grad_norm": 0.002563255140557885,
-      "learning_rate": 0.00019999507673050543,
-      "loss": 46.0,
-      "step": 41322
-    },
-    {
-      "epoch": 3.1594319246134144,
-      "grad_norm": 0.0014723507920280099,
-      "learning_rate": 0.00019999507649215876,
-      "loss": 46.0,
-      "step": 41323
-    },
-    {
-      "epoch": 3.159508381596804,
-      "grad_norm": 0.0035391172859817743,
-      "learning_rate": 0.00019999507625380634,
-      "loss": 46.0,
-      "step": 41324
-    },
-    {
-      "epoch": 3.159584838580194,
-      "grad_norm": 0.0015904412139207125,
-      "learning_rate": 0.00019999507601544813,
-      "loss": 46.0,
-      "step": 41325
-    },
-    {
-      "epoch": 3.1596612955635837,
-      "grad_norm": 0.003802722552791238,
-      "learning_rate": 0.00019999507577708416,
-      "loss": 46.0,
-      "step": 41326
-    },
-    {
-      "epoch": 3.1597377525469734,
-      "grad_norm": 0.00244453689083457,
-      "learning_rate": 0.00019999507553871443,
-      "loss": 46.0,
-      "step": 41327
-    },
-    {
-      "epoch": 3.159814209530363,
-      "grad_norm": 0.003056307090446353,
-      "learning_rate": 0.00019999507530033892,
-      "loss": 46.0,
-      "step": 41328
-    },
-    {
-      "epoch": 3.1598906665137525,
-      "grad_norm": 0.0023832679726183414,
-      "learning_rate": 0.00019999507506195764,
-      "loss": 46.0,
-      "step": 41329
-    },
-    {
-      "epoch": 3.1599671234971423,
-      "grad_norm": 0.0028164638206362724,
-      "learning_rate": 0.00019999507482357058,
-      "loss": 46.0,
-      "step": 41330
-    },
-    {
-      "epoch": 3.160043580480532,
-      "grad_norm": 0.001658973516896367,
-      "learning_rate": 0.00019999507458517775,
-      "loss": 46.0,
-      "step": 41331
-    },
-    {
-      "epoch": 3.160120037463922,
-      "grad_norm": 0.00508987857028842,
-      "learning_rate": 0.0001999950743467792,
-      "loss": 46.0,
-      "step": 41332
-    },
-    {
-      "epoch": 3.1601964944473115,
-      "grad_norm": 0.0021771444007754326,
-      "learning_rate": 0.00019999507410837483,
-      "loss": 46.0,
-      "step": 41333
-    },
-    {
-      "epoch": 3.1602729514307013,
-      "grad_norm": 0.0013983589597046375,
-      "learning_rate": 0.0001999950738699647,
-      "loss": 46.0,
-      "step": 41334
-    },
-    {
-      "epoch": 3.160349408414091,
-      "grad_norm": 0.004197974689304829,
-      "learning_rate": 0.00019999507363154881,
-      "loss": 46.0,
-      "step": 41335
-    },
-    {
-      "epoch": 3.160425865397481,
-      "grad_norm": 0.0011154047679156065,
-      "learning_rate": 0.00019999507339312715,
-      "loss": 46.0,
-      "step": 41336
-    },
-    {
-      "epoch": 3.1605023223808706,
-      "grad_norm": 0.0008261289331130683,
-      "learning_rate": 0.00019999507315469973,
-      "loss": 46.0,
-      "step": 41337
-    },
-    {
-      "epoch": 3.1605787793642603,
-      "grad_norm": 0.0026026489213109016,
-      "learning_rate": 0.00019999507291626652,
-      "loss": 46.0,
-      "step": 41338
-    },
-    {
-      "epoch": 3.16065523634765,
-      "grad_norm": 0.0018387408927083015,
-      "learning_rate": 0.00019999507267782755,
-      "loss": 46.0,
-      "step": 41339
-    },
-    {
-      "epoch": 3.1607316933310394,
-      "grad_norm": 0.003892270615324378,
-      "learning_rate": 0.0001999950724393828,
-      "loss": 46.0,
-      "step": 41340
-    },
-    {
-      "epoch": 3.160808150314429,
-      "grad_norm": 0.0007451699930243194,
-      "learning_rate": 0.0001999950722009323,
-      "loss": 46.0,
-      "step": 41341
-    },
-    {
-      "epoch": 3.160884607297819,
-      "grad_norm": 0.004547202028334141,
-      "learning_rate": 0.00019999507196247603,
-      "loss": 46.0,
-      "step": 41342
-    },
-    {
-      "epoch": 3.1609610642812087,
-      "grad_norm": 0.0013254316290840507,
-      "learning_rate": 0.00019999507172401397,
-      "loss": 46.0,
-      "step": 41343
-    },
-    {
-      "epoch": 3.1610375212645985,
-      "grad_norm": 0.003969667479395866,
-      "learning_rate": 0.00019999507148554617,
-      "loss": 46.0,
-      "step": 41344
-    },
-    {
-      "epoch": 3.1611139782479882,
-      "grad_norm": 0.0007787776412442327,
-      "learning_rate": 0.00019999507124707257,
-      "loss": 46.0,
-      "step": 41345
-    },
-    {
-      "epoch": 3.161190435231378,
-      "grad_norm": 0.004932778421789408,
-      "learning_rate": 0.0001999950710085932,
-      "loss": 46.0,
-      "step": 41346
-    },
-    {
-      "epoch": 3.1612668922147678,
-      "grad_norm": 0.0039217849262058735,
-      "learning_rate": 0.0001999950707701081,
-      "loss": 46.0,
-      "step": 41347
-    },
-    {
-      "epoch": 3.1613433491981575,
-      "grad_norm": 0.003376128152012825,
-      "learning_rate": 0.00019999507053161718,
-      "loss": 46.0,
-      "step": 41348
-    },
-    {
-      "epoch": 3.1614198061815473,
-      "grad_norm": 0.002181009389460087,
-      "learning_rate": 0.00019999507029312054,
-      "loss": 46.0,
-      "step": 41349
-    },
-    {
-      "epoch": 3.1614962631649366,
-      "grad_norm": 0.00240871193818748,
-      "learning_rate": 0.0001999950700546181,
-      "loss": 46.0,
-      "step": 41350
-    },
-    {
-      "epoch": 3.1615727201483264,
-      "grad_norm": 0.0015356396324932575,
-      "learning_rate": 0.00019999506981610988,
-      "loss": 46.0,
-      "step": 41351
-    },
-    {
-      "epoch": 3.161649177131716,
-      "grad_norm": 0.0029594209045171738,
-      "learning_rate": 0.00019999506957759592,
-      "loss": 46.0,
-      "step": 41352
-    },
-    {
-      "epoch": 3.161725634115106,
-      "grad_norm": 0.00423911539837718,
-      "learning_rate": 0.0001999950693390762,
-      "loss": 46.0,
-      "step": 41353
-    },
-    {
-      "epoch": 3.1618020910984956,
-      "grad_norm": 0.0026481598615646362,
-      "learning_rate": 0.00019999506910055068,
-      "loss": 46.0,
-      "step": 41354
-    },
-    {
-      "epoch": 3.1618785480818854,
-      "grad_norm": 0.0006250111036933959,
-      "learning_rate": 0.00019999506886201942,
-      "loss": 46.0,
-      "step": 41355
-    },
-    {
-      "epoch": 3.161955005065275,
-      "grad_norm": 0.006145654246211052,
-      "learning_rate": 0.0001999950686234824,
-      "loss": 46.0,
-      "step": 41356
-    },
-    {
-      "epoch": 3.162031462048665,
-      "grad_norm": 0.007828936912119389,
-      "learning_rate": 0.00019999506838493954,
-      "loss": 46.0,
-      "step": 41357
-    },
-    {
-      "epoch": 3.1621079190320547,
-      "grad_norm": 0.0010668664472177625,
-      "learning_rate": 0.00019999506814639097,
-      "loss": 46.0,
-      "step": 41358
-    },
-    {
-      "epoch": 3.1621843760154444,
-      "grad_norm": 0.0029236541595309973,
-      "learning_rate": 0.00019999506790783662,
-      "loss": 46.0,
-      "step": 41359
-    },
-    {
-      "epoch": 3.162260832998834,
-      "grad_norm": 0.003693469800055027,
-      "learning_rate": 0.0001999950676692765,
-      "loss": 46.0,
-      "step": 41360
-    },
-    {
-      "epoch": 3.162337289982224,
-      "grad_norm": 0.004355990327894688,
-      "learning_rate": 0.0001999950674307106,
-      "loss": 46.0,
-      "step": 41361
-    },
-    {
-      "epoch": 3.1624137469656133,
-      "grad_norm": 0.0013126952107995749,
-      "learning_rate": 0.00019999506719213894,
-      "loss": 46.0,
-      "step": 41362
-    },
-    {
-      "epoch": 3.162490203949003,
-      "grad_norm": 0.002799743553623557,
-      "learning_rate": 0.00019999506695356152,
-      "loss": 46.0,
-      "step": 41363
-    },
-    {
-      "epoch": 3.162566660932393,
-      "grad_norm": 0.003541214158758521,
-      "learning_rate": 0.0001999950667149783,
-      "loss": 46.0,
-      "step": 41364
-    },
-    {
-      "epoch": 3.1626431179157826,
-      "grad_norm": 0.0009232715819962323,
-      "learning_rate": 0.00019999506647638935,
-      "loss": 46.0,
-      "step": 41365
-    },
-    {
-      "epoch": 3.1627195748991723,
-      "grad_norm": 0.0021728789433836937,
-      "learning_rate": 0.0001999950662377946,
-      "loss": 46.0,
-      "step": 41366
-    },
-    {
-      "epoch": 3.162796031882562,
-      "grad_norm": 0.0012628649128600955,
-      "learning_rate": 0.00019999506599919408,
-      "loss": 46.0,
-      "step": 41367
-    },
-    {
-      "epoch": 3.162872488865952,
-      "grad_norm": 0.002501907991245389,
-      "learning_rate": 0.00019999506576058782,
-      "loss": 46.0,
-      "step": 41368
-    },
-    {
-      "epoch": 3.1629489458493416,
-      "grad_norm": 0.0017043884145095944,
-      "learning_rate": 0.00019999506552197577,
-      "loss": 46.0,
-      "step": 41369
-    },
-    {
-      "epoch": 3.1630254028327314,
-      "grad_norm": 0.001165761612355709,
-      "learning_rate": 0.00019999506528335797,
-      "loss": 46.0,
-      "step": 41370
-    },
-    {
-      "epoch": 3.163101859816121,
-      "grad_norm": 0.005266531836241484,
-      "learning_rate": 0.00019999506504473437,
-      "loss": 46.0,
-      "step": 41371
-    },
-    {
-      "epoch": 3.1631783167995104,
-      "grad_norm": 0.0012408840702846646,
-      "learning_rate": 0.00019999506480610505,
-      "loss": 46.0,
-      "step": 41372
-    },
-    {
-      "epoch": 3.1632547737829,
-      "grad_norm": 0.0014161203289404511,
-      "learning_rate": 0.0001999950645674699,
-      "loss": 46.0,
-      "step": 41373
-    },
-    {
-      "epoch": 3.16333123076629,
-      "grad_norm": 0.001424388843588531,
-      "learning_rate": 0.000199995064328829,
-      "loss": 46.0,
-      "step": 41374
-    },
-    {
-      "epoch": 3.1634076877496797,
-      "grad_norm": 0.0028290506452322006,
-      "learning_rate": 0.00019999506409018234,
-      "loss": 46.0,
-      "step": 41375
-    },
-    {
-      "epoch": 3.1634841447330695,
-      "grad_norm": 0.0021013321820646524,
-      "learning_rate": 0.0001999950638515299,
-      "loss": 46.0,
-      "step": 41376
-    },
-    {
-      "epoch": 3.1635606017164593,
-      "grad_norm": 0.00865001417696476,
-      "learning_rate": 0.00019999506361287174,
-      "loss": 46.0,
-      "step": 41377
-    },
-    {
-      "epoch": 3.163637058699849,
-      "grad_norm": 0.0044823638163506985,
-      "learning_rate": 0.00019999506337420778,
-      "loss": 46.0,
-      "step": 41378
-    },
-    {
-      "epoch": 3.1637135156832388,
-      "grad_norm": 0.0018760509556159377,
-      "learning_rate": 0.00019999506313553802,
-      "loss": 46.0,
-      "step": 41379
-    },
-    {
-      "epoch": 3.1637899726666285,
-      "grad_norm": 0.00493275374174118,
-      "learning_rate": 0.0001999950628968625,
-      "loss": 46.0,
-      "step": 41380
-    },
-    {
-      "epoch": 3.1638664296500183,
-      "grad_norm": 0.0025269535835832357,
-      "learning_rate": 0.00019999506265818125,
-      "loss": 46.0,
-      "step": 41381
-    },
-    {
-      "epoch": 3.163942886633408,
-      "grad_norm": 0.005213672295212746,
-      "learning_rate": 0.0001999950624194942,
-      "loss": 46.0,
-      "step": 41382
-    },
-    {
-      "epoch": 3.164019343616798,
-      "grad_norm": 0.002192362444475293,
-      "learning_rate": 0.00019999506218080137,
-      "loss": 46.0,
-      "step": 41383
-    },
-    {
-      "epoch": 3.164095800600187,
-      "grad_norm": 0.002362956292927265,
-      "learning_rate": 0.0001999950619421028,
-      "loss": 46.0,
-      "step": 41384
-    },
-    {
-      "epoch": 3.164172257583577,
-      "grad_norm": 0.0019419651944190264,
-      "learning_rate": 0.00019999506170339845,
-      "loss": 46.0,
-      "step": 41385
-    },
-    {
-      "epoch": 3.1642487145669667,
-      "grad_norm": 0.003133219201117754,
-      "learning_rate": 0.00019999506146468833,
-      "loss": 46.0,
-      "step": 41386
-    },
-    {
-      "epoch": 3.1643251715503564,
-      "grad_norm": 0.0018855045782402158,
-      "learning_rate": 0.00019999506122597244,
-      "loss": 46.0,
-      "step": 41387
-    },
-    {
-      "epoch": 3.164401628533746,
-      "grad_norm": 0.002275438979268074,
-      "learning_rate": 0.00019999506098725077,
-      "loss": 46.0,
-      "step": 41388
-    },
-    {
-      "epoch": 3.164478085517136,
-      "grad_norm": 0.004472393076866865,
-      "learning_rate": 0.00019999506074852336,
-      "loss": 46.0,
-      "step": 41389
-    },
-    {
-      "epoch": 3.1645545425005257,
-      "grad_norm": 0.0024108458310365677,
-      "learning_rate": 0.00019999506050979014,
-      "loss": 46.0,
-      "step": 41390
-    },
-    {
-      "epoch": 3.1646309994839155,
-      "grad_norm": 0.0010790076339617372,
-      "learning_rate": 0.00019999506027105118,
-      "loss": 46.0,
-      "step": 41391
-    },
-    {
-      "epoch": 3.1647074564673052,
-      "grad_norm": 0.004766308702528477,
-      "learning_rate": 0.00019999506003230645,
-      "loss": 46.0,
-      "step": 41392
-    },
-    {
-      "epoch": 3.164783913450695,
-      "grad_norm": 0.0024100765585899353,
-      "learning_rate": 0.00019999505979355594,
-      "loss": 46.0,
-      "step": 41393
-    },
-    {
-      "epoch": 3.1648603704340843,
-      "grad_norm": 0.0015630577690899372,
-      "learning_rate": 0.00019999505955479966,
-      "loss": 46.0,
-      "step": 41394
-    },
-    {
-      "epoch": 3.164936827417474,
-      "grad_norm": 0.0023312268313020468,
-      "learning_rate": 0.0001999950593160376,
-      "loss": 46.0,
-      "step": 41395
-    },
-    {
-      "epoch": 3.165013284400864,
-      "grad_norm": 0.0014532121131196618,
-      "learning_rate": 0.0001999950590772698,
-      "loss": 46.0,
-      "step": 41396
-    },
-    {
-      "epoch": 3.1650897413842536,
-      "grad_norm": 0.002739020623266697,
-      "learning_rate": 0.00019999505883849623,
-      "loss": 46.0,
-      "step": 41397
-    },
-    {
-      "epoch": 3.1651661983676433,
-      "grad_norm": 0.0027691854629665613,
-      "learning_rate": 0.0001999950585997169,
-      "loss": 46.0,
-      "step": 41398
-    },
-    {
-      "epoch": 3.165242655351033,
-      "grad_norm": 0.0033416608348488808,
-      "learning_rate": 0.00019999505836093174,
-      "loss": 46.0,
-      "step": 41399
-    },
-    {
-      "epoch": 3.165319112334423,
-      "grad_norm": 0.0025658742524683475,
-      "learning_rate": 0.00019999505812214085,
-      "loss": 46.0,
-      "step": 41400
-    },
-    {
-      "epoch": 3.1653955693178126,
-      "grad_norm": 0.005955113098025322,
-      "learning_rate": 0.0001999950578833442,
-      "loss": 46.0,
-      "step": 41401
-    },
-    {
-      "epoch": 3.1654720263012024,
-      "grad_norm": 0.002519210334867239,
-      "learning_rate": 0.0001999950576445418,
-      "loss": 46.0,
-      "step": 41402
-    },
-    {
-      "epoch": 3.165548483284592,
-      "grad_norm": 0.0014996922109276056,
-      "learning_rate": 0.00019999505740573355,
-      "loss": 46.0,
-      "step": 41403
-    },
-    {
-      "epoch": 3.165624940267982,
-      "grad_norm": 0.0008262404007837176,
-      "learning_rate": 0.00019999505716691962,
-      "loss": 46.0,
-      "step": 41404
-    },
-    {
-      "epoch": 3.1657013972513717,
-      "grad_norm": 0.001112659927457571,
-      "learning_rate": 0.0001999950569280999,
-      "loss": 46.0,
-      "step": 41405
-    },
-    {
-      "epoch": 3.165777854234761,
-      "grad_norm": 0.0016313678352162242,
-      "learning_rate": 0.00019999505668927435,
-      "loss": 46.0,
-      "step": 41406
-    },
-    {
-      "epoch": 3.1658543112181508,
-      "grad_norm": 0.0013768535573035479,
-      "learning_rate": 0.0001999950564504431,
-      "loss": 46.0,
-      "step": 41407
-    },
-    {
-      "epoch": 3.1659307682015405,
-      "grad_norm": 0.002947288565337658,
-      "learning_rate": 0.00019999505621160605,
-      "loss": 46.0,
-      "step": 41408
-    },
-    {
-      "epoch": 3.1660072251849303,
-      "grad_norm": 0.006324646528810263,
-      "learning_rate": 0.00019999505597276325,
-      "loss": 46.0,
-      "step": 41409
-    },
-    {
-      "epoch": 3.16608368216832,
-      "grad_norm": 0.0023180863354355097,
-      "learning_rate": 0.00019999505573391465,
-      "loss": 46.0,
-      "step": 41410
-    },
-    {
-      "epoch": 3.16616013915171,
-      "grad_norm": 0.001528040855191648,
-      "learning_rate": 0.0001999950554950603,
-      "loss": 46.0,
-      "step": 41411
-    },
-    {
-      "epoch": 3.1662365961350996,
-      "grad_norm": 0.0011482478585094213,
-      "learning_rate": 0.00019999505525620018,
-      "loss": 46.0,
-      "step": 41412
-    },
-    {
-      "epoch": 3.1663130531184893,
-      "grad_norm": 0.0023069106973707676,
-      "learning_rate": 0.00019999505501733431,
-      "loss": 46.0,
-      "step": 41413
-    },
-    {
-      "epoch": 3.166389510101879,
-      "grad_norm": 0.0033784867264330387,
-      "learning_rate": 0.00019999505477846265,
-      "loss": 46.0,
-      "step": 41414
-    },
-    {
-      "epoch": 3.166465967085269,
-      "grad_norm": 0.002776743844151497,
-      "learning_rate": 0.00019999505453958524,
-      "loss": 46.0,
-      "step": 41415
-    },
-    {
-      "epoch": 3.166542424068658,
-      "grad_norm": 0.0018270849250257015,
-      "learning_rate": 0.00019999505430070202,
-      "loss": 46.0,
-      "step": 41416
-    },
-    {
-      "epoch": 3.166618881052048,
-      "grad_norm": 0.0011862947139889002,
-      "learning_rate": 0.00019999505406181304,
-      "loss": 46.0,
-      "step": 41417
-    },
-    {
-      "epoch": 3.1666953380354377,
-      "grad_norm": 0.0038134099449962378,
-      "learning_rate": 0.00019999505382291833,
-      "loss": 46.0,
-      "step": 41418
-    },
-    {
-      "epoch": 3.1667717950188274,
-      "grad_norm": 0.0010683911386877298,
-      "learning_rate": 0.00019999505358401782,
-      "loss": 46.0,
-      "step": 41419
-    },
-    {
-      "epoch": 3.166848252002217,
-      "grad_norm": 0.000719505304004997,
-      "learning_rate": 0.00019999505334511154,
-      "loss": 46.0,
-      "step": 41420
-    },
-    {
-      "epoch": 3.166924708985607,
-      "grad_norm": 0.008577807806432247,
-      "learning_rate": 0.0001999950531061995,
-      "loss": 46.0,
-      "step": 41421
-    },
-    {
-      "epoch": 3.1670011659689967,
-      "grad_norm": 0.003264984581619501,
-      "learning_rate": 0.0001999950528672817,
-      "loss": 46.0,
-      "step": 41422
-    },
-    {
-      "epoch": 3.1670776229523865,
-      "grad_norm": 0.0026449931319803,
-      "learning_rate": 0.00019999505262835812,
-      "loss": 46.0,
-      "step": 41423
-    },
-    {
-      "epoch": 3.1671540799357762,
-      "grad_norm": 0.0037897902075201273,
-      "learning_rate": 0.00019999505238942877,
-      "loss": 46.0,
-      "step": 41424
-    },
-    {
-      "epoch": 3.167230536919166,
-      "grad_norm": 0.004215792752802372,
-      "learning_rate": 0.00019999505215049363,
-      "loss": 46.0,
-      "step": 41425
-    },
-    {
-      "epoch": 3.1673069939025558,
-      "grad_norm": 0.0018445485038682818,
-      "learning_rate": 0.00019999505191155276,
-      "loss": 46.0,
-      "step": 41426
-    },
-    {
-      "epoch": 3.1673834508859455,
-      "grad_norm": 0.007202946115285158,
-      "learning_rate": 0.0001999950516726061,
-      "loss": 46.0,
-      "step": 41427
-    },
-    {
-      "epoch": 3.167459907869335,
-      "grad_norm": 0.002640940947458148,
-      "learning_rate": 0.00019999505143365368,
-      "loss": 46.0,
-      "step": 41428
-    },
-    {
-      "epoch": 3.1675363648527246,
-      "grad_norm": 0.006939531769603491,
-      "learning_rate": 0.0001999950511946955,
-      "loss": 46.0,
-      "step": 41429
-    },
-    {
-      "epoch": 3.1676128218361144,
-      "grad_norm": 0.0034718550741672516,
-      "learning_rate": 0.00019999505095573154,
-      "loss": 46.0,
-      "step": 41430
-    },
-    {
-      "epoch": 3.167689278819504,
-      "grad_norm": 0.0015440002316609025,
-      "learning_rate": 0.00019999505071676178,
-      "loss": 46.0,
-      "step": 41431
-    },
-    {
-      "epoch": 3.167765735802894,
-      "grad_norm": 0.0021042367443442345,
-      "learning_rate": 0.00019999505047778627,
-      "loss": 46.0,
-      "step": 41432
-    },
-    {
-      "epoch": 3.1678421927862837,
-      "grad_norm": 0.00500674219802022,
-      "learning_rate": 0.00019999505023880502,
-      "loss": 46.0,
-      "step": 41433
-    },
-    {
-      "epoch": 3.1679186497696734,
-      "grad_norm": 0.0016279031988233328,
-      "learning_rate": 0.00019999504999981796,
-      "loss": 46.0,
-      "step": 41434
-    },
-    {
-      "epoch": 3.167995106753063,
-      "grad_norm": 0.0011564954183995724,
-      "learning_rate": 0.00019999504976082517,
-      "loss": 46.0,
-      "step": 41435
-    },
-    {
-      "epoch": 3.168071563736453,
-      "grad_norm": 0.003401833586394787,
-      "learning_rate": 0.00019999504952182657,
-      "loss": 46.0,
-      "step": 41436
-    },
-    {
-      "epoch": 3.1681480207198427,
-      "grad_norm": 0.0034036061260849237,
-      "learning_rate": 0.00019999504928282222,
-      "loss": 46.0,
-      "step": 41437
-    },
-    {
-      "epoch": 3.168224477703232,
-      "grad_norm": 0.001845944207161665,
-      "learning_rate": 0.0001999950490438121,
-      "loss": 46.0,
-      "step": 41438
-    },
-    {
-      "epoch": 3.1683009346866218,
-      "grad_norm": 0.001065343152731657,
-      "learning_rate": 0.00019999504880479624,
-      "loss": 46.0,
-      "step": 41439
-    },
-    {
-      "epoch": 3.1683773916700115,
-      "grad_norm": 0.0008206756901927292,
-      "learning_rate": 0.00019999504856577457,
-      "loss": 46.0,
-      "step": 41440
-    },
-    {
-      "epoch": 3.1684538486534013,
-      "grad_norm": 0.000572787772398442,
-      "learning_rate": 0.00019999504832674713,
-      "loss": 46.0,
-      "step": 41441
-    },
-    {
-      "epoch": 3.168530305636791,
-      "grad_norm": 0.0031394294928759336,
-      "learning_rate": 0.00019999504808771394,
-      "loss": 46.0,
-      "step": 41442
-    },
-    {
-      "epoch": 3.168606762620181,
-      "grad_norm": 0.006262943148612976,
-      "learning_rate": 0.00019999504784867499,
-      "loss": 46.0,
-      "step": 41443
-    },
-    {
-      "epoch": 3.1686832196035706,
-      "grad_norm": 0.0011333419242873788,
-      "learning_rate": 0.00019999504760963025,
-      "loss": 46.0,
-      "step": 41444
-    },
-    {
-      "epoch": 3.1687596765869603,
-      "grad_norm": 0.0010566285345703363,
-      "learning_rate": 0.00019999504737057975,
-      "loss": 46.0,
-      "step": 41445
-    },
-    {
-      "epoch": 3.16883613357035,
-      "grad_norm": 0.001251725247129798,
-      "learning_rate": 0.00019999504713152347,
-      "loss": 46.0,
-      "step": 41446
-    },
-    {
-      "epoch": 3.16891259055374,
-      "grad_norm": 0.003519281279295683,
-      "learning_rate": 0.00019999504689246144,
-      "loss": 46.0,
-      "step": 41447
-    },
-    {
-      "epoch": 3.1689890475371296,
-      "grad_norm": 0.0006476265261881053,
-      "learning_rate": 0.00019999504665339362,
-      "loss": 46.0,
-      "step": 41448
-    },
-    {
-      "epoch": 3.169065504520519,
-      "grad_norm": 0.003076705848798156,
-      "learning_rate": 0.00019999504641432004,
-      "loss": 46.0,
-      "step": 41449
-    },
-    {
-      "epoch": 3.1691419615039087,
-      "grad_norm": 0.0026646617334336042,
-      "learning_rate": 0.0001999950461752407,
-      "loss": 46.0,
-      "step": 41450
-    },
-    {
-      "epoch": 3.1692184184872985,
-      "grad_norm": 0.002413432579487562,
-      "learning_rate": 0.00019999504593615558,
-      "loss": 46.0,
-      "step": 41451
-    },
-    {
-      "epoch": 3.169294875470688,
-      "grad_norm": 0.003823963925242424,
-      "learning_rate": 0.0001999950456970647,
-      "loss": 46.0,
-      "step": 41452
-    },
-    {
-      "epoch": 3.169371332454078,
-      "grad_norm": 0.0011404749238863587,
-      "learning_rate": 0.00019999504545796802,
-      "loss": 46.0,
-      "step": 41453
-    },
-    {
-      "epoch": 3.1694477894374677,
-      "grad_norm": 0.002357630990445614,
-      "learning_rate": 0.0001999950452188656,
-      "loss": 46.0,
-      "step": 41454
-    },
-    {
-      "epoch": 3.1695242464208575,
-      "grad_norm": 0.0025385448243469,
-      "learning_rate": 0.0001999950449797574,
-      "loss": 46.0,
-      "step": 41455
-    },
-    {
-      "epoch": 3.1696007034042473,
-      "grad_norm": 0.00247181230224669,
-      "learning_rate": 0.00019999504474064347,
-      "loss": 46.0,
-      "step": 41456
-    },
-    {
-      "epoch": 3.169677160387637,
-      "grad_norm": 0.0029713709373027086,
-      "learning_rate": 0.00019999504450152373,
-      "loss": 46.0,
-      "step": 41457
-    },
-    {
-      "epoch": 3.169753617371027,
-      "grad_norm": 0.0036932379007339478,
-      "learning_rate": 0.0001999950442623982,
-      "loss": 46.0,
-      "step": 41458
-    },
-    {
-      "epoch": 3.1698300743544165,
-      "grad_norm": 0.008026341907680035,
-      "learning_rate": 0.00019999504402326698,
-      "loss": 46.0,
-      "step": 41459
-    },
-    {
-      "epoch": 3.169906531337806,
-      "grad_norm": 0.003109171986579895,
-      "learning_rate": 0.0001999950437841299,
-      "loss": 46.0,
-      "step": 41460
-    },
-    {
-      "epoch": 3.1699829883211956,
-      "grad_norm": 0.003735044738277793,
-      "learning_rate": 0.0001999950435449871,
-      "loss": 46.0,
-      "step": 41461
-    },
-    {
-      "epoch": 3.1700594453045854,
-      "grad_norm": 0.001759059727191925,
-      "learning_rate": 0.00019999504330583853,
-      "loss": 46.0,
-      "step": 41462
-    },
-    {
-      "epoch": 3.170135902287975,
-      "grad_norm": 0.0013326314510777593,
-      "learning_rate": 0.00019999504306668418,
-      "loss": 46.0,
-      "step": 41463
-    },
-    {
-      "epoch": 3.170212359271365,
-      "grad_norm": 0.007648392580449581,
-      "learning_rate": 0.00019999504282752406,
-      "loss": 46.0,
-      "step": 41464
-    },
-    {
-      "epoch": 3.1702888162547547,
-      "grad_norm": 0.0004672433133237064,
-      "learning_rate": 0.0001999950425883582,
-      "loss": 46.0,
-      "step": 41465
-    },
-    {
-      "epoch": 3.1703652732381444,
-      "grad_norm": 0.002496375236660242,
-      "learning_rate": 0.00019999504234918653,
-      "loss": 46.0,
-      "step": 41466
-    },
-    {
-      "epoch": 3.170441730221534,
-      "grad_norm": 0.0010404987260699272,
-      "learning_rate": 0.0001999950421100091,
-      "loss": 46.0,
-      "step": 41467
-    },
-    {
-      "epoch": 3.170518187204924,
-      "grad_norm": 0.006087156943976879,
-      "learning_rate": 0.00019999504187082588,
-      "loss": 46.0,
-      "step": 41468
-    },
-    {
-      "epoch": 3.1705946441883137,
-      "grad_norm": 0.0019763356540352106,
-      "learning_rate": 0.00019999504163163695,
-      "loss": 46.0,
-      "step": 41469
-    },
-    {
-      "epoch": 3.1706711011717035,
-      "grad_norm": 0.002362157916650176,
-      "learning_rate": 0.0001999950413924422,
-      "loss": 46.0,
-      "step": 41470
-    },
-    {
-      "epoch": 3.170747558155093,
-      "grad_norm": 0.0029203672893345356,
-      "learning_rate": 0.00019999504115324172,
-      "loss": 46.0,
-      "step": 41471
-    },
-    {
-      "epoch": 3.1708240151384826,
-      "grad_norm": 0.002817925764247775,
-      "learning_rate": 0.00019999504091403544,
-      "loss": 46.0,
-      "step": 41472
-    },
-    {
-      "epoch": 3.1709004721218723,
-      "grad_norm": 0.0020470796152949333,
-      "learning_rate": 0.00019999504067482338,
-      "loss": 46.0,
-      "step": 41473
-    },
-    {
-      "epoch": 3.170976929105262,
-      "grad_norm": 0.0025700980331748724,
-      "learning_rate": 0.0001999950404356056,
-      "loss": 46.0,
-      "step": 41474
-    },
-    {
-      "epoch": 3.171053386088652,
-      "grad_norm": 0.0017173243686556816,
-      "learning_rate": 0.00019999504019638202,
-      "loss": 46.0,
-      "step": 41475
-    },
-    {
-      "epoch": 3.1711298430720416,
-      "grad_norm": 0.0016228409949690104,
-      "learning_rate": 0.00019999503995715267,
-      "loss": 46.0,
-      "step": 41476
-    },
-    {
-      "epoch": 3.1712063000554314,
-      "grad_norm": 0.0010417434386909008,
-      "learning_rate": 0.00019999503971791755,
-      "loss": 46.0,
-      "step": 41477
-    },
-    {
-      "epoch": 3.171282757038821,
-      "grad_norm": 0.0006923602195456624,
-      "learning_rate": 0.00019999503947867666,
-      "loss": 46.0,
-      "step": 41478
-    },
-    {
-      "epoch": 3.171359214022211,
-      "grad_norm": 0.0022682345006614923,
-      "learning_rate": 0.00019999503923943002,
-      "loss": 46.0,
-      "step": 41479
-    },
-    {
-      "epoch": 3.1714356710056006,
-      "grad_norm": 0.0013542587403208017,
-      "learning_rate": 0.00019999503900017758,
-      "loss": 46.0,
-      "step": 41480
-    },
-    {
-      "epoch": 3.17151212798899,
-      "grad_norm": 0.0011866593267768621,
-      "learning_rate": 0.0001999950387609194,
-      "loss": 46.0,
-      "step": 41481
-    },
-    {
-      "epoch": 3.1715885849723797,
-      "grad_norm": 0.0028488964308053255,
-      "learning_rate": 0.00019999503852165544,
-      "loss": 46.0,
-      "step": 41482
-    },
-    {
-      "epoch": 3.1716650419557695,
-      "grad_norm": 0.0019846842624247074,
-      "learning_rate": 0.00019999503828238574,
-      "loss": 46.0,
-      "step": 41483
-    },
-    {
-      "epoch": 3.1717414989391592,
-      "grad_norm": 0.0017689914675429463,
-      "learning_rate": 0.0001999950380431102,
-      "loss": 46.0,
-      "step": 41484
-    },
-    {
-      "epoch": 3.171817955922549,
-      "grad_norm": 0.0028321673162281513,
-      "learning_rate": 0.00019999503780382893,
-      "loss": 46.0,
-      "step": 41485
-    },
-    {
-      "epoch": 3.1718944129059388,
-      "grad_norm": 0.002153906738385558,
-      "learning_rate": 0.0001999950375645419,
-      "loss": 46.0,
-      "step": 41486
-    },
-    {
-      "epoch": 3.1719708698893285,
-      "grad_norm": 0.0023367053363472223,
-      "learning_rate": 0.00019999503732524908,
-      "loss": 46.0,
-      "step": 41487
-    },
-    {
-      "epoch": 3.1720473268727183,
-      "grad_norm": 0.001522139529697597,
-      "learning_rate": 0.0001999950370859505,
-      "loss": 46.0,
-      "step": 41488
-    },
-    {
-      "epoch": 3.172123783856108,
-      "grad_norm": 0.002241625217720866,
-      "learning_rate": 0.0001999950368466462,
-      "loss": 46.0,
-      "step": 41489
-    },
-    {
-      "epoch": 3.172200240839498,
-      "grad_norm": 0.003878000658005476,
-      "learning_rate": 0.00019999503660733607,
-      "loss": 46.0,
-      "step": 41490
-    },
-    {
-      "epoch": 3.1722766978228876,
-      "grad_norm": 0.0013456863816827536,
-      "learning_rate": 0.00019999503636802018,
-      "loss": 46.0,
-      "step": 41491
-    },
-    {
-      "epoch": 3.1723531548062773,
-      "grad_norm": 0.002906520152464509,
-      "learning_rate": 0.00019999503612869852,
-      "loss": 46.0,
-      "step": 41492
-    },
-    {
-      "epoch": 3.1724296117896666,
-      "grad_norm": 0.0011602099984884262,
-      "learning_rate": 0.0001999950358893711,
-      "loss": 46.0,
-      "step": 41493
-    },
-    {
-      "epoch": 3.1725060687730564,
-      "grad_norm": 0.004237143322825432,
-      "learning_rate": 0.00019999503565003792,
-      "loss": 46.0,
-      "step": 41494
-    },
-    {
-      "epoch": 3.172582525756446,
-      "grad_norm": 0.003364263102412224,
-      "learning_rate": 0.00019999503541069896,
-      "loss": 46.0,
-      "step": 41495
-    },
-    {
-      "epoch": 3.172658982739836,
-      "grad_norm": 0.0010974074248224497,
-      "learning_rate": 0.0001999950351713542,
-      "loss": 46.0,
-      "step": 41496
-    },
-    {
-      "epoch": 3.1727354397232257,
-      "grad_norm": 0.001306042424403131,
-      "learning_rate": 0.0001999950349320037,
-      "loss": 46.0,
-      "step": 41497
-    },
-    {
-      "epoch": 3.1728118967066155,
-      "grad_norm": 0.0017872753087431192,
-      "learning_rate": 0.00019999503469264745,
-      "loss": 46.0,
-      "step": 41498
-    },
-    {
-      "epoch": 3.172888353690005,
-      "grad_norm": 0.0027761689852923155,
-      "learning_rate": 0.0001999950344532854,
-      "loss": 46.0,
-      "step": 41499
-    },
-    {
-      "epoch": 3.172964810673395,
-      "grad_norm": 0.0014261918840929866,
-      "learning_rate": 0.0001999950342139176,
-      "loss": 46.0,
-      "step": 41500
-    },
-    {
-      "epoch": 3.1730412676567847,
-      "grad_norm": 0.0022536099422723055,
-      "learning_rate": 0.00019999503397454406,
-      "loss": 46.0,
-      "step": 41501
-    },
-    {
-      "epoch": 3.1731177246401745,
-      "grad_norm": 0.0035957032814621925,
-      "learning_rate": 0.00019999503373516468,
-      "loss": 46.0,
-      "step": 41502
-    },
-    {
-      "epoch": 3.173194181623564,
-      "grad_norm": 0.002542843110859394,
-      "learning_rate": 0.0001999950334957796,
-      "loss": 46.0,
-      "step": 41503
-    },
-    {
-      "epoch": 3.1732706386069536,
-      "grad_norm": 0.002371058566495776,
-      "learning_rate": 0.00019999503325638868,
-      "loss": 46.0,
-      "step": 41504
-    },
-    {
-      "epoch": 3.1733470955903433,
-      "grad_norm": 0.0021500475704669952,
-      "learning_rate": 0.00019999503301699204,
-      "loss": 46.0,
-      "step": 41505
-    },
-    {
-      "epoch": 3.173423552573733,
-      "grad_norm": 0.002679188270121813,
-      "learning_rate": 0.00019999503277758963,
-      "loss": 46.0,
-      "step": 41506
-    },
-    {
-      "epoch": 3.173500009557123,
-      "grad_norm": 0.004183996468782425,
-      "learning_rate": 0.00019999503253818142,
-      "loss": 46.0,
-      "step": 41507
-    },
-    {
-      "epoch": 3.1735764665405126,
-      "grad_norm": 0.004615056794136763,
-      "learning_rate": 0.00019999503229876746,
-      "loss": 46.0,
-      "step": 41508
-    },
-    {
-      "epoch": 3.1736529235239024,
-      "grad_norm": 0.0043732537887990475,
-      "learning_rate": 0.00019999503205934773,
-      "loss": 46.0,
-      "step": 41509
-    },
-    {
-      "epoch": 3.173729380507292,
-      "grad_norm": 0.0026550525799393654,
-      "learning_rate": 0.00019999503181992222,
-      "loss": 46.0,
-      "step": 41510
-    },
-    {
-      "epoch": 3.173805837490682,
-      "grad_norm": 0.0021852480713278055,
-      "learning_rate": 0.00019999503158049097,
-      "loss": 46.0,
-      "step": 41511
-    },
-    {
-      "epoch": 3.1738822944740717,
-      "grad_norm": 0.00213302462361753,
-      "learning_rate": 0.00019999503134105395,
-      "loss": 46.0,
-      "step": 41512
-    },
-    {
-      "epoch": 3.1739587514574614,
-      "grad_norm": 0.004383992403745651,
-      "learning_rate": 0.00019999503110161113,
-      "loss": 46.0,
-      "step": 41513
-    },
-    {
-      "epoch": 3.174035208440851,
-      "grad_norm": 0.0014782102080062032,
-      "learning_rate": 0.00019999503086216256,
-      "loss": 46.0,
-      "step": 41514
-    },
-    {
-      "epoch": 3.1741116654242405,
-      "grad_norm": 0.026076970621943474,
-      "learning_rate": 0.0001999950306227082,
-      "loss": 46.0,
-      "step": 41515
-    },
-    {
-      "epoch": 3.1741881224076303,
-      "grad_norm": 0.0024677226319909096,
-      "learning_rate": 0.00019999503038324812,
-      "loss": 46.0,
-      "step": 41516
-    },
-    {
-      "epoch": 3.17426457939102,
-      "grad_norm": 0.0016292547807097435,
-      "learning_rate": 0.00019999503014378223,
-      "loss": 46.0,
-      "step": 41517
-    },
-    {
-      "epoch": 3.17434103637441,
-      "grad_norm": 0.0010048781987279654,
-      "learning_rate": 0.00019999502990431057,
-      "loss": 46.0,
-      "step": 41518
-    },
-    {
-      "epoch": 3.1744174933577995,
-      "grad_norm": 0.0029262187890708447,
-      "learning_rate": 0.00019999502966483316,
-      "loss": 46.0,
-      "step": 41519
-    },
-    {
-      "epoch": 3.1744939503411893,
-      "grad_norm": 0.0017416657647117972,
-      "learning_rate": 0.00019999502942534995,
-      "loss": 46.0,
-      "step": 41520
-    },
-    {
-      "epoch": 3.174570407324579,
-      "grad_norm": 0.0008085901499725878,
-      "learning_rate": 0.000199995029185861,
-      "loss": 46.0,
-      "step": 41521
-    },
-    {
-      "epoch": 3.174646864307969,
-      "grad_norm": 0.002365314634516835,
-      "learning_rate": 0.00019999502894636626,
-      "loss": 46.0,
-      "step": 41522
-    },
-    {
-      "epoch": 3.1747233212913586,
-      "grad_norm": 0.0009196935570798814,
-      "learning_rate": 0.00019999502870686578,
-      "loss": 46.0,
-      "step": 41523
-    },
-    {
-      "epoch": 3.1747997782747484,
-      "grad_norm": 0.0016587093705311418,
-      "learning_rate": 0.0001999950284673595,
-      "loss": 46.0,
-      "step": 41524
-    },
-    {
-      "epoch": 3.1748762352581377,
-      "grad_norm": 0.0020995191298425198,
-      "learning_rate": 0.00019999502822784745,
-      "loss": 46.0,
-      "step": 41525
-    },
-    {
-      "epoch": 3.1749526922415274,
-      "grad_norm": 0.0010391664691269398,
-      "learning_rate": 0.00019999502798832968,
-      "loss": 46.0,
-      "step": 41526
-    },
-    {
-      "epoch": 3.175029149224917,
-      "grad_norm": 0.0033912716899067163,
-      "learning_rate": 0.00019999502774880609,
-      "loss": 46.0,
-      "step": 41527
-    },
-    {
-      "epoch": 3.175105606208307,
-      "grad_norm": 0.0022105860989540815,
-      "learning_rate": 0.00019999502750927674,
-      "loss": 46.0,
-      "step": 41528
-    },
-    {
-      "epoch": 3.1751820631916967,
-      "grad_norm": 0.004614948760718107,
-      "learning_rate": 0.00019999502726974165,
-      "loss": 46.0,
-      "step": 41529
-    },
-    {
-      "epoch": 3.1752585201750865,
-      "grad_norm": 0.0019378411816433072,
-      "learning_rate": 0.00019999502703020076,
-      "loss": 46.0,
-      "step": 41530
-    },
-    {
-      "epoch": 3.1753349771584762,
-      "grad_norm": 0.0008935146615840495,
-      "learning_rate": 0.00019999502679065413,
-      "loss": 46.0,
-      "step": 41531
-    },
-    {
-      "epoch": 3.175411434141866,
-      "grad_norm": 0.000457369489595294,
-      "learning_rate": 0.0001999950265511017,
-      "loss": 46.0,
-      "step": 41532
-    },
-    {
-      "epoch": 3.1754878911252558,
-      "grad_norm": 0.0016567959683015943,
-      "learning_rate": 0.0001999950263115435,
-      "loss": 46.0,
-      "step": 41533
-    },
-    {
-      "epoch": 3.1755643481086455,
-      "grad_norm": 0.001107292715460062,
-      "learning_rate": 0.00019999502607197952,
-      "loss": 46.0,
-      "step": 41534
-    },
-    {
-      "epoch": 3.1756408050920353,
-      "grad_norm": 0.0027877024840563536,
-      "learning_rate": 0.00019999502583240982,
-      "loss": 46.0,
-      "step": 41535
-    },
-    {
-      "epoch": 3.175717262075425,
-      "grad_norm": 0.0028962024953216314,
-      "learning_rate": 0.00019999502559283431,
-      "loss": 46.0,
-      "step": 41536
-    },
-    {
-      "epoch": 3.1757937190588144,
-      "grad_norm": 0.0011165491305291653,
-      "learning_rate": 0.00019999502535325306,
-      "loss": 46.0,
-      "step": 41537
-    },
-    {
-      "epoch": 3.175870176042204,
-      "grad_norm": 0.0009530842071399093,
-      "learning_rate": 0.00019999502511366601,
-      "loss": 46.0,
-      "step": 41538
-    },
-    {
-      "epoch": 3.175946633025594,
-      "grad_norm": 0.001333654043264687,
-      "learning_rate": 0.00019999502487407322,
-      "loss": 46.0,
-      "step": 41539
-    },
-    {
-      "epoch": 3.1760230900089836,
-      "grad_norm": 0.005875668954104185,
-      "learning_rate": 0.00019999502463447465,
-      "loss": 46.0,
-      "step": 41540
-    },
-    {
-      "epoch": 3.1760995469923734,
-      "grad_norm": 0.0024102074094116688,
-      "learning_rate": 0.0001999950243948703,
-      "loss": 46.0,
-      "step": 41541
-    },
-    {
-      "epoch": 3.176176003975763,
-      "grad_norm": 0.0007566119311377406,
-      "learning_rate": 0.0001999950241552602,
-      "loss": 46.0,
-      "step": 41542
-    },
-    {
-      "epoch": 3.176252460959153,
-      "grad_norm": 0.002351798117160797,
-      "learning_rate": 0.0001999950239156443,
-      "loss": 46.0,
-      "step": 41543
-    },
-    {
-      "epoch": 3.1763289179425427,
-      "grad_norm": 0.0049277860671281815,
-      "learning_rate": 0.00019999502367602266,
-      "loss": 46.0,
-      "step": 41544
-    },
-    {
-      "epoch": 3.1764053749259324,
-      "grad_norm": 0.0021302916575223207,
-      "learning_rate": 0.00019999502343639525,
-      "loss": 46.0,
-      "step": 41545
-    },
-    {
-      "epoch": 3.176481831909322,
-      "grad_norm": 0.0034527345560491085,
-      "learning_rate": 0.00019999502319676204,
-      "loss": 46.0,
-      "step": 41546
-    },
-    {
-      "epoch": 3.1765582888927115,
-      "grad_norm": 0.005665971431881189,
-      "learning_rate": 0.00019999502295712309,
-      "loss": 46.0,
-      "step": 41547
-    },
-    {
-      "epoch": 3.1766347458761013,
-      "grad_norm": 0.005099464673548937,
-      "learning_rate": 0.00019999502271747836,
-      "loss": 46.0,
-      "step": 41548
-    },
-    {
-      "epoch": 3.176711202859491,
-      "grad_norm": 0.0012947771465405822,
-      "learning_rate": 0.00019999502247782785,
-      "loss": 46.0,
-      "step": 41549
-    },
-    {
-      "epoch": 3.176787659842881,
-      "grad_norm": 0.0018413482466712594,
-      "learning_rate": 0.0001999950222381716,
-      "loss": 46.0,
-      "step": 41550
-    },
-    {
-      "epoch": 3.1768641168262706,
-      "grad_norm": 0.005373688414692879,
-      "learning_rate": 0.00019999502199850958,
-      "loss": 46.0,
-      "step": 41551
-    },
-    {
-      "epoch": 3.1769405738096603,
-      "grad_norm": 0.0012525643687695265,
-      "learning_rate": 0.00019999502175884178,
-      "loss": 46.0,
-      "step": 41552
-    },
-    {
-      "epoch": 3.17701703079305,
-      "grad_norm": 0.0017059410456568003,
-      "learning_rate": 0.00019999502151916821,
-      "loss": 46.0,
-      "step": 41553
-    },
-    {
-      "epoch": 3.17709348777644,
-      "grad_norm": 0.004187475424259901,
-      "learning_rate": 0.00019999502127948887,
-      "loss": 46.0,
-      "step": 41554
-    },
-    {
-      "epoch": 3.1771699447598296,
-      "grad_norm": 0.0016918386099860072,
-      "learning_rate": 0.00019999502103980375,
-      "loss": 46.0,
-      "step": 41555
-    },
-    {
-      "epoch": 3.1772464017432194,
-      "grad_norm": 0.001578646362759173,
-      "learning_rate": 0.00019999502080011286,
-      "loss": 46.0,
-      "step": 41556
-    },
-    {
-      "epoch": 3.177322858726609,
-      "grad_norm": 0.0018953564576804638,
-      "learning_rate": 0.0001999950205604162,
-      "loss": 46.0,
-      "step": 41557
-    },
-    {
-      "epoch": 3.177399315709999,
-      "grad_norm": 0.0020169634371995926,
-      "learning_rate": 0.0001999950203207138,
-      "loss": 46.0,
-      "step": 41558
-    },
-    {
-      "epoch": 3.177475772693388,
-      "grad_norm": 0.003477503778412938,
-      "learning_rate": 0.0001999950200810056,
-      "loss": 46.0,
-      "step": 41559
-    },
-    {
-      "epoch": 3.177552229676778,
-      "grad_norm": 0.0009138755267485976,
-      "learning_rate": 0.00019999501984129165,
-      "loss": 46.0,
-      "step": 41560
-    },
-    {
-      "epoch": 3.1776286866601677,
-      "grad_norm": 0.0018043399322777987,
-      "learning_rate": 0.00019999501960157195,
-      "loss": 46.0,
-      "step": 41561
-    },
-    {
-      "epoch": 3.1777051436435575,
-      "grad_norm": 0.0024193490389734507,
-      "learning_rate": 0.00019999501936184645,
-      "loss": 46.0,
-      "step": 41562
-    },
-    {
-      "epoch": 3.1777816006269473,
-      "grad_norm": 0.0018770884489640594,
-      "learning_rate": 0.00019999501912211514,
-      "loss": 46.0,
-      "step": 41563
-    },
-    {
-      "epoch": 3.177858057610337,
-      "grad_norm": 0.0030878782272338867,
-      "learning_rate": 0.00019999501888237812,
-      "loss": 46.0,
-      "step": 41564
-    },
-    {
-      "epoch": 3.177934514593727,
-      "grad_norm": 0.0067941430024802685,
-      "learning_rate": 0.00019999501864263533,
-      "loss": 46.0,
-      "step": 41565
-    },
-    {
-      "epoch": 3.1780109715771165,
-      "grad_norm": 0.0018375436775386333,
-      "learning_rate": 0.00019999501840288676,
-      "loss": 46.0,
-      "step": 41566
-    },
-    {
-      "epoch": 3.1780874285605063,
-      "grad_norm": 0.004575012717396021,
-      "learning_rate": 0.00019999501816313241,
-      "loss": 46.0,
-      "step": 41567
-    },
-    {
-      "epoch": 3.178163885543896,
-      "grad_norm": 0.002006369410082698,
-      "learning_rate": 0.00019999501792337233,
-      "loss": 46.0,
-      "step": 41568
-    },
-    {
-      "epoch": 3.1782403425272854,
-      "grad_norm": 0.0024700758513063192,
-      "learning_rate": 0.00019999501768360644,
-      "loss": 46.0,
-      "step": 41569
-    },
-    {
-      "epoch": 3.178316799510675,
-      "grad_norm": 0.002289470285177231,
-      "learning_rate": 0.0001999950174438348,
-      "loss": 46.0,
-      "step": 41570
-    },
-    {
-      "epoch": 3.178393256494065,
-      "grad_norm": 0.0011970787309110165,
-      "learning_rate": 0.00019999501720405736,
-      "loss": 46.0,
-      "step": 41571
-    },
-    {
-      "epoch": 3.1784697134774547,
-      "grad_norm": 0.0008146300679072738,
-      "learning_rate": 0.00019999501696427418,
-      "loss": 46.0,
-      "step": 41572
-    },
-    {
-      "epoch": 3.1785461704608444,
-      "grad_norm": 0.0033185905776917934,
-      "learning_rate": 0.00019999501672448525,
-      "loss": 46.0,
-      "step": 41573
-    },
-    {
-      "epoch": 3.178622627444234,
-      "grad_norm": 0.0015095893759280443,
-      "learning_rate": 0.0001999950164846905,
-      "loss": 46.0,
-      "step": 41574
-    },
-    {
-      "epoch": 3.178699084427624,
-      "grad_norm": 0.00117490254342556,
-      "learning_rate": 0.00019999501624489002,
-      "loss": 46.0,
-      "step": 41575
-    },
-    {
-      "epoch": 3.1787755414110137,
-      "grad_norm": 0.002301177242770791,
-      "learning_rate": 0.00019999501600508375,
-      "loss": 46.0,
-      "step": 41576
-    },
-    {
-      "epoch": 3.1788519983944035,
-      "grad_norm": 0.001515236683189869,
-      "learning_rate": 0.00019999501576527172,
-      "loss": 46.0,
-      "step": 41577
-    },
-    {
-      "epoch": 3.1789284553777932,
-      "grad_norm": 0.0013602422550320625,
-      "learning_rate": 0.00019999501552545393,
-      "loss": 46.0,
-      "step": 41578
-    },
-    {
-      "epoch": 3.179004912361183,
-      "grad_norm": 0.002288083080202341,
-      "learning_rate": 0.00019999501528563033,
-      "loss": 46.0,
-      "step": 41579
-    },
-    {
-      "epoch": 3.1790813693445723,
-      "grad_norm": 0.001860869931988418,
-      "learning_rate": 0.00019999501504580102,
-      "loss": 46.0,
-      "step": 41580
-    },
-    {
-      "epoch": 3.179157826327962,
-      "grad_norm": 0.0034210917074233294,
-      "learning_rate": 0.0001999950148059659,
-      "loss": 46.0,
-      "step": 41581
-    },
-    {
-      "epoch": 3.179234283311352,
-      "grad_norm": 0.0013878109166398644,
-      "learning_rate": 0.000199995014566125,
-      "loss": 46.0,
-      "step": 41582
-    },
-    {
-      "epoch": 3.1793107402947416,
-      "grad_norm": 0.0029540008399635553,
-      "learning_rate": 0.00019999501432627838,
-      "loss": 46.0,
-      "step": 41583
-    },
-    {
-      "epoch": 3.1793871972781313,
-      "grad_norm": 0.0035117545630782843,
-      "learning_rate": 0.00019999501408642597,
-      "loss": 46.0,
-      "step": 41584
-    },
-    {
-      "epoch": 3.179463654261521,
-      "grad_norm": 0.0033393804915249348,
-      "learning_rate": 0.00019999501384656778,
-      "loss": 46.0,
-      "step": 41585
-    },
-    {
-      "epoch": 3.179540111244911,
-      "grad_norm": 0.004739480093121529,
-      "learning_rate": 0.0001999950136067038,
-      "loss": 46.0,
-      "step": 41586
-    },
-    {
-      "epoch": 3.1796165682283006,
-      "grad_norm": 0.0010944517562165856,
-      "learning_rate": 0.00019999501336683407,
-      "loss": 46.0,
-      "step": 41587
-    },
-    {
-      "epoch": 3.1796930252116904,
-      "grad_norm": 0.0009988929377868772,
-      "learning_rate": 0.0001999950131269586,
-      "loss": 46.0,
-      "step": 41588
-    },
-    {
-      "epoch": 3.17976948219508,
-      "grad_norm": 0.00634508952498436,
-      "learning_rate": 0.00019999501288707735,
-      "loss": 46.0,
-      "step": 41589
-    },
-    {
-      "epoch": 3.17984593917847,
-      "grad_norm": 0.01088044699281454,
-      "learning_rate": 0.0001999950126471903,
-      "loss": 46.0,
-      "step": 41590
-    },
-    {
-      "epoch": 3.1799223961618592,
-      "grad_norm": 0.001451428746804595,
-      "learning_rate": 0.0001999950124072975,
-      "loss": 46.0,
-      "step": 41591
-    },
-    {
-      "epoch": 3.179998853145249,
-      "grad_norm": 0.0037342708092182875,
-      "learning_rate": 0.00019999501216739894,
-      "loss": 46.0,
-      "step": 41592
-    },
-    {
-      "epoch": 3.1800753101286388,
-      "grad_norm": 0.001866180100478232,
-      "learning_rate": 0.0001999950119274946,
-      "loss": 46.0,
-      "step": 41593
-    },
-    {
-      "epoch": 3.1801517671120285,
-      "grad_norm": 0.0014850023435428739,
-      "learning_rate": 0.00019999501168758448,
-      "loss": 46.0,
-      "step": 41594
-    },
-    {
-      "epoch": 3.1802282240954183,
-      "grad_norm": 0.0019063232466578484,
-      "learning_rate": 0.00019999501144766862,
-      "loss": 46.0,
-      "step": 41595
-    },
-    {
-      "epoch": 3.180304681078808,
-      "grad_norm": 0.000784132513217628,
-      "learning_rate": 0.00019999501120774698,
-      "loss": 46.0,
-      "step": 41596
-    },
-    {
-      "epoch": 3.180381138062198,
-      "grad_norm": 0.0049018170684576035,
-      "learning_rate": 0.00019999501096781955,
-      "loss": 46.0,
-      "step": 41597
-    },
-    {
-      "epoch": 3.1804575950455876,
-      "grad_norm": 0.0017000646330416203,
-      "learning_rate": 0.00019999501072788636,
-      "loss": 46.0,
-      "step": 41598
-    },
-    {
-      "epoch": 3.1805340520289773,
-      "grad_norm": 0.0048112571239471436,
-      "learning_rate": 0.0001999950104879474,
-      "loss": 46.0,
-      "step": 41599
-    },
-    {
-      "epoch": 3.180610509012367,
-      "grad_norm": 0.000671104877255857,
-      "learning_rate": 0.00019999501024800268,
-      "loss": 46.0,
-      "step": 41600
-    },
-    {
-      "epoch": 3.180686965995757,
-      "grad_norm": 0.002536576008424163,
-      "learning_rate": 0.00019999501000805218,
-      "loss": 46.0,
-      "step": 41601
-    },
-    {
-      "epoch": 3.180763422979146,
-      "grad_norm": 0.0018588016973808408,
-      "learning_rate": 0.00019999500976809593,
-      "loss": 46.0,
-      "step": 41602
-    },
-    {
-      "epoch": 3.180839879962536,
-      "grad_norm": 0.002467375248670578,
-      "learning_rate": 0.0001999950095281339,
-      "loss": 46.0,
-      "step": 41603
-    },
-    {
-      "epoch": 3.1809163369459257,
-      "grad_norm": 0.002192617394030094,
-      "learning_rate": 0.00019999500928816611,
-      "loss": 46.0,
-      "step": 41604
-    },
-    {
-      "epoch": 3.1809927939293154,
-      "grad_norm": 0.0008830138831399381,
-      "learning_rate": 0.00019999500904819252,
-      "loss": 46.0,
-      "step": 41605
-    },
-    {
-      "epoch": 3.181069250912705,
-      "grad_norm": 0.0038703896570950747,
-      "learning_rate": 0.0001999950088082132,
-      "loss": 46.0,
-      "step": 41606
-    },
-    {
-      "epoch": 3.181145707896095,
-      "grad_norm": 0.0009447685442864895,
-      "learning_rate": 0.0001999950085682281,
-      "loss": 46.0,
-      "step": 41607
-    },
-    {
-      "epoch": 3.1812221648794847,
-      "grad_norm": 0.001665694871917367,
-      "learning_rate": 0.0001999950083282372,
-      "loss": 46.0,
-      "step": 41608
-    },
-    {
-      "epoch": 3.1812986218628745,
-      "grad_norm": 0.0013467780081555247,
-      "learning_rate": 0.00019999500808824057,
-      "loss": 46.0,
-      "step": 41609
-    },
-    {
-      "epoch": 3.1813750788462642,
-      "grad_norm": 0.0018159612081944942,
-      "learning_rate": 0.00019999500784823816,
-      "loss": 46.0,
-      "step": 41610
-    },
-    {
-      "epoch": 3.181451535829654,
-      "grad_norm": 0.0005733192665502429,
-      "learning_rate": 0.00019999500760822998,
-      "loss": 46.0,
-      "step": 41611
-    },
-    {
-      "epoch": 3.1815279928130433,
-      "grad_norm": 0.0007503158994950354,
-      "learning_rate": 0.00019999500736821602,
-      "loss": 46.0,
-      "step": 41612
-    },
-    {
-      "epoch": 3.181604449796433,
-      "grad_norm": 0.0014418438076972961,
-      "learning_rate": 0.00019999500712819627,
-      "loss": 46.0,
-      "step": 41613
-    },
-    {
-      "epoch": 3.181680906779823,
-      "grad_norm": 0.006869006436318159,
-      "learning_rate": 0.00019999500688817082,
-      "loss": 46.0,
-      "step": 41614
-    },
-    {
-      "epoch": 3.1817573637632126,
-      "grad_norm": 0.002136533847078681,
-      "learning_rate": 0.00019999500664813952,
-      "loss": 46.0,
-      "step": 41615
-    },
-    {
-      "epoch": 3.1818338207466024,
-      "grad_norm": 0.00460156612098217,
-      "learning_rate": 0.00019999500640810252,
-      "loss": 46.0,
-      "step": 41616
-    },
-    {
-      "epoch": 3.181910277729992,
-      "grad_norm": 0.002954870695248246,
-      "learning_rate": 0.0001999950061680597,
-      "loss": 46.0,
-      "step": 41617
-    },
-    {
-      "epoch": 3.181986734713382,
-      "grad_norm": 0.0031767820473760366,
-      "learning_rate": 0.00019999500592801113,
-      "loss": 46.0,
-      "step": 41618
-    },
-    {
-      "epoch": 3.1820631916967717,
-      "grad_norm": 0.0008709346293471754,
-      "learning_rate": 0.0001999950056879568,
-      "loss": 46.0,
-      "step": 41619
-    },
-    {
-      "epoch": 3.1821396486801614,
-      "grad_norm": 0.0018425191519781947,
-      "learning_rate": 0.00019999500544789668,
-      "loss": 46.0,
-      "step": 41620
-    },
-    {
-      "epoch": 3.182216105663551,
-      "grad_norm": 0.001838630880229175,
-      "learning_rate": 0.00019999500520783082,
-      "loss": 46.0,
-      "step": 41621
-    },
-    {
-      "epoch": 3.182292562646941,
-      "grad_norm": 0.0015733184991404414,
-      "learning_rate": 0.00019999500496775915,
-      "loss": 46.0,
-      "step": 41622
-    },
-    {
-      "epoch": 3.1823690196303307,
-      "grad_norm": 0.002236704109236598,
-      "learning_rate": 0.00019999500472768175,
-      "loss": 46.0,
-      "step": 41623
-    },
-    {
-      "epoch": 3.18244547661372,
-      "grad_norm": 0.001440307474695146,
-      "learning_rate": 0.0001999950044875986,
-      "loss": 46.0,
-      "step": 41624
-    },
-    {
-      "epoch": 3.1825219335971098,
-      "grad_norm": 0.0014734945725649595,
-      "learning_rate": 0.0001999950042475096,
-      "loss": 46.0,
-      "step": 41625
-    },
-    {
-      "epoch": 3.1825983905804995,
-      "grad_norm": 0.0010877603199332952,
-      "learning_rate": 0.0001999950040074149,
-      "loss": 46.0,
-      "step": 41626
-    },
-    {
-      "epoch": 3.1826748475638893,
-      "grad_norm": 0.001741341664455831,
-      "learning_rate": 0.0001999950037673144,
-      "loss": 46.0,
-      "step": 41627
-    },
-    {
-      "epoch": 3.182751304547279,
-      "grad_norm": 0.0006204545497894287,
-      "learning_rate": 0.00019999500352720816,
-      "loss": 46.0,
-      "step": 41628
-    },
-    {
-      "epoch": 3.182827761530669,
-      "grad_norm": 0.002920563332736492,
-      "learning_rate": 0.00019999500328709614,
-      "loss": 46.0,
-      "step": 41629
-    },
-    {
-      "epoch": 3.1829042185140586,
-      "grad_norm": 0.002461136318743229,
-      "learning_rate": 0.00019999500304697834,
-      "loss": 46.0,
-      "step": 41630
-    },
-    {
-      "epoch": 3.1829806754974483,
-      "grad_norm": 0.0011827474227175117,
-      "learning_rate": 0.00019999500280685478,
-      "loss": 46.0,
-      "step": 41631
-    },
-    {
-      "epoch": 3.183057132480838,
-      "grad_norm": 0.0031452528201043606,
-      "learning_rate": 0.00019999500256672544,
-      "loss": 46.0,
-      "step": 41632
-    },
-    {
-      "epoch": 3.183133589464228,
-      "grad_norm": 0.004967565648257732,
-      "learning_rate": 0.00019999500232659032,
-      "loss": 46.0,
-      "step": 41633
-    },
-    {
-      "epoch": 3.183210046447617,
-      "grad_norm": 0.002345235552638769,
-      "learning_rate": 0.00019999500208644943,
-      "loss": 46.0,
-      "step": 41634
-    },
-    {
-      "epoch": 3.183286503431007,
-      "grad_norm": 0.0013114488683640957,
-      "learning_rate": 0.0001999950018463028,
-      "loss": 46.0,
-      "step": 41635
-    },
-    {
-      "epoch": 3.1833629604143967,
-      "grad_norm": 0.0029320684261620045,
-      "learning_rate": 0.0001999950016061504,
-      "loss": 46.0,
-      "step": 41636
-    },
-    {
-      "epoch": 3.1834394173977865,
-      "grad_norm": 0.002438370371237397,
-      "learning_rate": 0.0001999950013659922,
-      "loss": 46.0,
-      "step": 41637
-    },
-    {
-      "epoch": 3.1835158743811762,
-      "grad_norm": 0.004720065742731094,
-      "learning_rate": 0.00019999500112582826,
-      "loss": 46.0,
-      "step": 41638
-    },
-    {
-      "epoch": 3.183592331364566,
-      "grad_norm": 0.0036966942716389894,
-      "learning_rate": 0.00019999500088565856,
-      "loss": 46.0,
-      "step": 41639
-    },
-    {
-      "epoch": 3.1836687883479557,
-      "grad_norm": 0.0013784399488940835,
-      "learning_rate": 0.00019999500064548306,
-      "loss": 46.0,
-      "step": 41640
-    },
-    {
-      "epoch": 3.1837452453313455,
-      "grad_norm": 0.0032542948611080647,
-      "learning_rate": 0.00019999500040530178,
-      "loss": 46.0,
-      "step": 41641
-    },
-    {
-      "epoch": 3.1838217023147353,
-      "grad_norm": 0.0023591159842908382,
-      "learning_rate": 0.00019999500016511473,
-      "loss": 46.0,
-      "step": 41642
-    },
-    {
-      "epoch": 3.183898159298125,
-      "grad_norm": 0.0045031337067484856,
-      "learning_rate": 0.00019999499992492194,
-      "loss": 46.0,
-      "step": 41643
-    },
-    {
-      "epoch": 3.183974616281515,
-      "grad_norm": 0.002368827583268285,
-      "learning_rate": 0.0001999949996847234,
-      "loss": 46.0,
-      "step": 41644
-    },
-    {
-      "epoch": 3.1840510732649046,
-      "grad_norm": 0.0009813635842874646,
-      "learning_rate": 0.00019999499944451906,
-      "loss": 46.0,
-      "step": 41645
-    },
-    {
-      "epoch": 3.184127530248294,
-      "grad_norm": 0.0036691150162369013,
-      "learning_rate": 0.00019999499920430894,
-      "loss": 46.0,
-      "step": 41646
-    },
-    {
-      "epoch": 3.1842039872316836,
-      "grad_norm": 0.0031920333858579397,
-      "learning_rate": 0.00019999499896409308,
-      "loss": 46.0,
-      "step": 41647
-    },
-    {
-      "epoch": 3.1842804442150734,
-      "grad_norm": 0.002809201367199421,
-      "learning_rate": 0.00019999499872387145,
-      "loss": 46.0,
-      "step": 41648
-    },
-    {
-      "epoch": 3.184356901198463,
-      "grad_norm": 0.0028415147680789232,
-      "learning_rate": 0.00019999499848364402,
-      "loss": 46.0,
-      "step": 41649
-    },
-    {
-      "epoch": 3.184433358181853,
-      "grad_norm": 0.00521051324903965,
-      "learning_rate": 0.00019999499824341083,
-      "loss": 46.0,
-      "step": 41650
-    },
-    {
-      "epoch": 3.1845098151652427,
-      "grad_norm": 0.002271543722599745,
-      "learning_rate": 0.0001999949980031719,
-      "loss": 46.0,
-      "step": 41651
-    },
-    {
-      "epoch": 3.1845862721486324,
-      "grad_norm": 0.0008131064823828638,
-      "learning_rate": 0.00019999499776292718,
-      "loss": 46.0,
-      "step": 41652
-    },
-    {
-      "epoch": 3.184662729132022,
-      "grad_norm": 0.0027534926775842905,
-      "learning_rate": 0.0001999949975226767,
-      "loss": 46.0,
-      "step": 41653
-    },
-    {
-      "epoch": 3.184739186115412,
-      "grad_norm": 0.0020710667595267296,
-      "learning_rate": 0.00019999499728242043,
-      "loss": 46.0,
-      "step": 41654
-    },
-    {
-      "epoch": 3.1848156430988017,
-      "grad_norm": 0.0020446348935365677,
-      "learning_rate": 0.0001999949970421584,
-      "loss": 46.0,
-      "step": 41655
-    },
-    {
-      "epoch": 3.184892100082191,
-      "grad_norm": 0.0019081425853073597,
-      "learning_rate": 0.0001999949968018906,
-      "loss": 46.0,
-      "step": 41656
-    },
-    {
-      "epoch": 3.184968557065581,
-      "grad_norm": 0.001043798285536468,
-      "learning_rate": 0.00019999499656161702,
-      "loss": 46.0,
-      "step": 41657
-    },
-    {
-      "epoch": 3.1850450140489706,
-      "grad_norm": 0.0037501410115510225,
-      "learning_rate": 0.0001999949963213377,
-      "loss": 46.0,
-      "step": 41658
-    },
-    {
-      "epoch": 3.1851214710323603,
-      "grad_norm": 0.006452765315771103,
-      "learning_rate": 0.0001999949960810526,
-      "loss": 46.0,
-      "step": 41659
-    },
-    {
-      "epoch": 3.18519792801575,
-      "grad_norm": 0.0012687244452536106,
-      "learning_rate": 0.00019999499584076174,
-      "loss": 46.0,
-      "step": 41660
-    },
-    {
-      "epoch": 3.18527438499914,
-      "grad_norm": 0.0029455493204295635,
-      "learning_rate": 0.00019999499560046508,
-      "loss": 46.0,
-      "step": 41661
-    },
-    {
-      "epoch": 3.1853508419825296,
-      "grad_norm": 0.0026105111464858055,
-      "learning_rate": 0.00019999499536016267,
-      "loss": 46.0,
-      "step": 41662
-    },
-    {
-      "epoch": 3.1854272989659194,
-      "grad_norm": 0.0007706858450546861,
-      "learning_rate": 0.0001999949951198545,
-      "loss": 46.0,
-      "step": 41663
-    },
-    {
-      "epoch": 3.185503755949309,
-      "grad_norm": 0.003465610556304455,
-      "learning_rate": 0.00019999499487954054,
-      "loss": 46.0,
-      "step": 41664
-    },
-    {
-      "epoch": 3.185580212932699,
-      "grad_norm": 0.0025231502950191498,
-      "learning_rate": 0.00019999499463922084,
-      "loss": 46.0,
-      "step": 41665
-    },
-    {
-      "epoch": 3.1856566699160886,
-      "grad_norm": 0.0031832822132855654,
-      "learning_rate": 0.00019999499439889534,
-      "loss": 46.0,
-      "step": 41666
-    },
-    {
-      "epoch": 3.1857331268994784,
-      "grad_norm": 0.0018971977988258004,
-      "learning_rate": 0.00019999499415856406,
-      "loss": 46.0,
-      "step": 41667
-    },
-    {
-      "epoch": 3.1858095838828677,
-      "grad_norm": 0.0026606128085404634,
-      "learning_rate": 0.00019999499391822704,
-      "loss": 46.0,
-      "step": 41668
-    },
-    {
-      "epoch": 3.1858860408662575,
-      "grad_norm": 0.0011698466259986162,
-      "learning_rate": 0.00019999499367788425,
-      "loss": 46.0,
-      "step": 41669
-    },
-    {
-      "epoch": 3.1859624978496472,
-      "grad_norm": 0.0017910320311784744,
-      "learning_rate": 0.00019999499343753568,
-      "loss": 46.0,
-      "step": 41670
-    },
-    {
-      "epoch": 3.186038954833037,
-      "grad_norm": 0.002160444622859359,
-      "learning_rate": 0.00019999499319718137,
-      "loss": 46.0,
-      "step": 41671
-    },
-    {
-      "epoch": 3.1861154118164268,
-      "grad_norm": 0.002982755657285452,
-      "learning_rate": 0.00019999499295682125,
-      "loss": 46.0,
-      "step": 41672
-    },
-    {
-      "epoch": 3.1861918687998165,
-      "grad_norm": 0.00077407545177266,
-      "learning_rate": 0.00019999499271645537,
-      "loss": 46.0,
-      "step": 41673
-    },
-    {
-      "epoch": 3.1862683257832063,
-      "grad_norm": 0.0038116704672574997,
-      "learning_rate": 0.00019999499247608373,
-      "loss": 46.0,
-      "step": 41674
-    },
-    {
-      "epoch": 3.186344782766596,
-      "grad_norm": 0.0015645333332940936,
-      "learning_rate": 0.00019999499223570633,
-      "loss": 46.0,
-      "step": 41675
-    },
-    {
-      "epoch": 3.186421239749986,
-      "grad_norm": 0.0028651622124016285,
-      "learning_rate": 0.00019999499199532315,
-      "loss": 46.0,
-      "step": 41676
-    },
-    {
-      "epoch": 3.1864976967333756,
-      "grad_norm": 0.0011966757010668516,
-      "learning_rate": 0.0001999949917549342,
-      "loss": 46.0,
-      "step": 41677
-    },
-    {
-      "epoch": 3.186574153716765,
-      "grad_norm": 0.004996367730200291,
-      "learning_rate": 0.0001999949915145395,
-      "loss": 46.0,
-      "step": 41678
-    },
-    {
-      "epoch": 3.1866506107001547,
-      "grad_norm": 0.002172978362068534,
-      "learning_rate": 0.00019999499127413902,
-      "loss": 46.0,
-      "step": 41679
-    },
-    {
-      "epoch": 3.1867270676835444,
-      "grad_norm": 0.002170744352042675,
-      "learning_rate": 0.00019999499103373275,
-      "loss": 46.0,
-      "step": 41680
-    },
-    {
-      "epoch": 3.186803524666934,
-      "grad_norm": 0.001734220772050321,
-      "learning_rate": 0.0001999949907933207,
-      "loss": 46.0,
-      "step": 41681
-    },
-    {
-      "epoch": 3.186879981650324,
-      "grad_norm": 0.0030573236290365458,
-      "learning_rate": 0.0001999949905529029,
-      "loss": 46.0,
-      "step": 41682
-    },
-    {
-      "epoch": 3.1869564386337137,
-      "grad_norm": 0.0018551649991422892,
-      "learning_rate": 0.00019999499031247937,
-      "loss": 46.0,
-      "step": 41683
-    },
-    {
-      "epoch": 3.1870328956171035,
-      "grad_norm": 0.0015566645888611674,
-      "learning_rate": 0.00019999499007205003,
-      "loss": 46.0,
-      "step": 41684
-    },
-    {
-      "epoch": 3.187109352600493,
-      "grad_norm": 0.0017837334889918566,
-      "learning_rate": 0.00019999498983161492,
-      "loss": 46.0,
-      "step": 41685
-    },
-    {
-      "epoch": 3.187185809583883,
-      "grad_norm": 0.0020725317299365997,
-      "learning_rate": 0.00019999498959117403,
-      "loss": 46.0,
-      "step": 41686
-    },
-    {
-      "epoch": 3.1872622665672727,
-      "grad_norm": 0.004108430817723274,
-      "learning_rate": 0.0001999949893507274,
-      "loss": 46.0,
-      "step": 41687
-    },
-    {
-      "epoch": 3.1873387235506625,
-      "grad_norm": 0.0026767654344439507,
-      "learning_rate": 0.000199994989110275,
-      "loss": 46.0,
-      "step": 41688
-    },
-    {
-      "epoch": 3.1874151805340523,
-      "grad_norm": 0.0008698187884874642,
-      "learning_rate": 0.0001999949888698168,
-      "loss": 46.0,
-      "step": 41689
-    },
-    {
-      "epoch": 3.1874916375174416,
-      "grad_norm": 0.0022780196741223335,
-      "learning_rate": 0.00019999498862935289,
-      "loss": 46.0,
-      "step": 41690
-    },
-    {
-      "epoch": 3.1875680945008313,
-      "grad_norm": 0.0007800338789820671,
-      "learning_rate": 0.00019999498838888316,
-      "loss": 46.0,
-      "step": 41691
-    },
-    {
-      "epoch": 3.187644551484221,
-      "grad_norm": 0.008167578838765621,
-      "learning_rate": 0.00019999498814840766,
-      "loss": 46.0,
-      "step": 41692
-    },
-    {
-      "epoch": 3.187721008467611,
-      "grad_norm": 0.0023614573292434216,
-      "learning_rate": 0.0001999949879079264,
-      "loss": 46.0,
-      "step": 41693
-    },
-    {
-      "epoch": 3.1877974654510006,
-      "grad_norm": 0.006071664392948151,
-      "learning_rate": 0.0001999949876674394,
-      "loss": 46.0,
-      "step": 41694
-    },
-    {
-      "epoch": 3.1878739224343904,
-      "grad_norm": 0.00268552009947598,
-      "learning_rate": 0.00019999498742694657,
-      "loss": 46.0,
-      "step": 41695
-    },
-    {
-      "epoch": 3.18795037941778,
-      "grad_norm": 0.0022890588734298944,
-      "learning_rate": 0.00019999498718644803,
-      "loss": 46.0,
-      "step": 41696
-    },
-    {
-      "epoch": 3.18802683640117,
-      "grad_norm": 0.001993396319448948,
-      "learning_rate": 0.0001999949869459437,
-      "loss": 46.0,
-      "step": 41697
-    },
-    {
-      "epoch": 3.1881032933845597,
-      "grad_norm": 0.005362294614315033,
-      "learning_rate": 0.00019999498670543358,
-      "loss": 46.0,
-      "step": 41698
-    },
-    {
-      "epoch": 3.1881797503679494,
-      "grad_norm": 0.006055880803614855,
-      "learning_rate": 0.00019999498646491772,
-      "loss": 46.0,
-      "step": 41699
-    },
-    {
-      "epoch": 3.1882562073513387,
-      "grad_norm": 0.00259780534543097,
-      "learning_rate": 0.00019999498622439606,
-      "loss": 46.0,
-      "step": 41700
-    },
-    {
-      "epoch": 3.1883326643347285,
-      "grad_norm": 0.0012675109319388866,
-      "learning_rate": 0.00019999498598386868,
-      "loss": 46.0,
-      "step": 41701
-    },
-    {
-      "epoch": 3.1884091213181183,
-      "grad_norm": 0.0005796006880700588,
-      "learning_rate": 0.0001999949857433355,
-      "loss": 46.0,
-      "step": 41702
-    },
-    {
-      "epoch": 3.188485578301508,
-      "grad_norm": 0.0007524911779910326,
-      "learning_rate": 0.00019999498550279653,
-      "loss": 46.0,
-      "step": 41703
-    },
-    {
-      "epoch": 3.188562035284898,
-      "grad_norm": 0.0021390062756836414,
-      "learning_rate": 0.00019999498526225185,
-      "loss": 46.0,
-      "step": 41704
-    },
-    {
-      "epoch": 3.1886384922682875,
-      "grad_norm": 0.005990055855363607,
-      "learning_rate": 0.00019999498502170135,
-      "loss": 46.0,
-      "step": 41705
-    },
-    {
-      "epoch": 3.1887149492516773,
-      "grad_norm": 0.001378029352054,
-      "learning_rate": 0.00019999498478114508,
-      "loss": 46.0,
-      "step": 41706
-    },
-    {
-      "epoch": 3.188791406235067,
-      "grad_norm": 0.0012867676559835672,
-      "learning_rate": 0.00019999498454058306,
-      "loss": 46.0,
-      "step": 41707
-    },
-    {
-      "epoch": 3.188867863218457,
-      "grad_norm": 0.005057553760707378,
-      "learning_rate": 0.0001999949843000153,
-      "loss": 46.0,
-      "step": 41708
-    },
-    {
-      "epoch": 3.1889443202018466,
-      "grad_norm": 0.0020237984135746956,
-      "learning_rate": 0.00019999498405944173,
-      "loss": 46.0,
-      "step": 41709
-    },
-    {
-      "epoch": 3.1890207771852364,
-      "grad_norm": 0.00036293535958975554,
-      "learning_rate": 0.0001999949838188624,
-      "loss": 46.0,
-      "step": 41710
-    },
-    {
-      "epoch": 3.1890972341686257,
-      "grad_norm": 0.0012332603801041842,
-      "learning_rate": 0.00019999498357827728,
-      "loss": 46.0,
-      "step": 41711
-    },
-    {
-      "epoch": 3.1891736911520154,
-      "grad_norm": 0.0020509285386651754,
-      "learning_rate": 0.0001999949833376864,
-      "loss": 46.0,
-      "step": 41712
-    },
-    {
-      "epoch": 3.189250148135405,
-      "grad_norm": 0.001647367374971509,
-      "learning_rate": 0.0001999949830970898,
-      "loss": 46.0,
-      "step": 41713
-    },
-    {
-      "epoch": 3.189326605118795,
-      "grad_norm": 0.0006007991614751518,
-      "learning_rate": 0.00019999498285648736,
-      "loss": 46.0,
-      "step": 41714
-    },
-    {
-      "epoch": 3.1894030621021847,
-      "grad_norm": 0.004553616978228092,
-      "learning_rate": 0.00019999498261587918,
-      "loss": 46.0,
-      "step": 41715
-    },
-    {
-      "epoch": 3.1894795190855745,
-      "grad_norm": 0.0007731916848570108,
-      "learning_rate": 0.00019999498237526525,
-      "loss": 46.0,
-      "step": 41716
-    },
-    {
-      "epoch": 3.1895559760689642,
-      "grad_norm": 0.004818055313080549,
-      "learning_rate": 0.00019999498213464553,
-      "loss": 46.0,
-      "step": 41717
-    },
-    {
-      "epoch": 3.189632433052354,
-      "grad_norm": 0.0006430381909012794,
-      "learning_rate": 0.00019999498189402005,
-      "loss": 46.0,
-      "step": 41718
-    },
-    {
-      "epoch": 3.1897088900357438,
-      "grad_norm": 0.0023323295172303915,
-      "learning_rate": 0.00019999498165338878,
-      "loss": 46.0,
-      "step": 41719
-    },
-    {
-      "epoch": 3.1897853470191335,
-      "grad_norm": 0.0017640216974541545,
-      "learning_rate": 0.00019999498141275176,
-      "loss": 46.0,
-      "step": 41720
-    },
-    {
-      "epoch": 3.1898618040025233,
-      "grad_norm": 0.0009280572412535548,
-      "learning_rate": 0.00019999498117210897,
-      "loss": 46.0,
-      "step": 41721
-    },
-    {
-      "epoch": 3.1899382609859126,
-      "grad_norm": 0.002290521515533328,
-      "learning_rate": 0.00019999498093146043,
-      "loss": 46.0,
-      "step": 41722
-    },
-    {
-      "epoch": 3.1900147179693024,
-      "grad_norm": 0.005821548867970705,
-      "learning_rate": 0.0001999949806908061,
-      "loss": 46.0,
-      "step": 41723
-    },
-    {
-      "epoch": 3.190091174952692,
-      "grad_norm": 0.0012983180349692702,
-      "learning_rate": 0.00019999498045014598,
-      "loss": 46.0,
-      "step": 41724
-    },
-    {
-      "epoch": 3.190167631936082,
-      "grad_norm": 0.002774711698293686,
-      "learning_rate": 0.00019999498020948012,
-      "loss": 46.0,
-      "step": 41725
-    },
-    {
-      "epoch": 3.1902440889194716,
-      "grad_norm": 0.0018534691771492362,
-      "learning_rate": 0.00019999497996880846,
-      "loss": 46.0,
-      "step": 41726
-    },
-    {
-      "epoch": 3.1903205459028614,
-      "grad_norm": 0.002928379224613309,
-      "learning_rate": 0.00019999497972813108,
-      "loss": 46.0,
-      "step": 41727
-    },
-    {
-      "epoch": 3.190397002886251,
-      "grad_norm": 0.0022499614860862494,
-      "learning_rate": 0.0001999949794874479,
-      "loss": 46.0,
-      "step": 41728
-    },
-    {
-      "epoch": 3.190473459869641,
-      "grad_norm": 0.010072726756334305,
-      "learning_rate": 0.00019999497924675893,
-      "loss": 46.0,
-      "step": 41729
-    },
-    {
-      "epoch": 3.1905499168530307,
-      "grad_norm": 0.002090676221996546,
-      "learning_rate": 0.00019999497900606423,
-      "loss": 46.0,
-      "step": 41730
-    },
-    {
-      "epoch": 3.1906263738364204,
-      "grad_norm": 0.0016220242250710726,
-      "learning_rate": 0.00019999497876536376,
-      "loss": 46.0,
-      "step": 41731
-    },
-    {
-      "epoch": 3.19070283081981,
-      "grad_norm": 0.002303495304659009,
-      "learning_rate": 0.00019999497852465748,
-      "loss": 46.0,
-      "step": 41732
-    },
-    {
-      "epoch": 3.1907792878031995,
-      "grad_norm": 0.0017076447838917375,
-      "learning_rate": 0.00019999497828394547,
-      "loss": 46.0,
-      "step": 41733
-    },
-    {
-      "epoch": 3.1908557447865893,
-      "grad_norm": 0.004774115979671478,
-      "learning_rate": 0.00019999497804322767,
-      "loss": 46.0,
-      "step": 41734
-    },
-    {
-      "epoch": 3.190932201769979,
-      "grad_norm": 0.0011328072287142277,
-      "learning_rate": 0.00019999497780250414,
-      "loss": 46.0,
-      "step": 41735
-    },
-    {
-      "epoch": 3.191008658753369,
-      "grad_norm": 0.002215541899204254,
-      "learning_rate": 0.0001999949775617748,
-      "loss": 46.0,
-      "step": 41736
-    },
-    {
-      "epoch": 3.1910851157367586,
-      "grad_norm": 0.002423564437776804,
-      "learning_rate": 0.0001999949773210397,
-      "loss": 46.0,
-      "step": 41737
-    },
-    {
-      "epoch": 3.1911615727201483,
-      "grad_norm": 0.001683376613073051,
-      "learning_rate": 0.00019999497708029883,
-      "loss": 46.0,
-      "step": 41738
-    },
-    {
-      "epoch": 3.191238029703538,
-      "grad_norm": 0.00179462437517941,
-      "learning_rate": 0.0001999949768395522,
-      "loss": 46.0,
-      "step": 41739
-    },
-    {
-      "epoch": 3.191314486686928,
-      "grad_norm": 0.0014585162280127406,
-      "learning_rate": 0.0001999949765987998,
-      "loss": 46.0,
-      "step": 41740
-    },
-    {
-      "epoch": 3.1913909436703176,
-      "grad_norm": 0.0015233204467222095,
-      "learning_rate": 0.0001999949763580416,
-      "loss": 46.0,
-      "step": 41741
-    },
-    {
-      "epoch": 3.1914674006537074,
-      "grad_norm": 0.001559414085932076,
-      "learning_rate": 0.00019999497611727766,
-      "loss": 46.0,
-      "step": 41742
-    },
-    {
-      "epoch": 3.1915438576370967,
-      "grad_norm": 0.004434848204255104,
-      "learning_rate": 0.00019999497587650796,
-      "loss": 46.0,
-      "step": 41743
-    },
-    {
-      "epoch": 3.1916203146204865,
-      "grad_norm": 0.0034638717770576477,
-      "learning_rate": 0.0001999949756357325,
-      "loss": 46.0,
-      "step": 41744
-    },
-    {
-      "epoch": 3.191696771603876,
-      "grad_norm": 0.0016908141551539302,
-      "learning_rate": 0.00019999497539495122,
-      "loss": 46.0,
-      "step": 41745
-    },
-    {
-      "epoch": 3.191773228587266,
-      "grad_norm": 0.0016036538872867823,
-      "learning_rate": 0.0001999949751541642,
-      "loss": 46.0,
-      "step": 41746
-    },
-    {
-      "epoch": 3.1918496855706557,
-      "grad_norm": 0.0023750835098326206,
-      "learning_rate": 0.00019999497491337144,
-      "loss": 46.0,
-      "step": 41747
-    },
-    {
-      "epoch": 3.1919261425540455,
-      "grad_norm": 0.0010884370421990752,
-      "learning_rate": 0.00019999497467257284,
-      "loss": 46.0,
-      "step": 41748
-    },
-    {
-      "epoch": 3.1920025995374353,
-      "grad_norm": 0.00804507452994585,
-      "learning_rate": 0.00019999497443176853,
-      "loss": 46.0,
-      "step": 41749
-    },
-    {
-      "epoch": 3.192079056520825,
-      "grad_norm": 0.0024654394946992397,
-      "learning_rate": 0.0001999949741909584,
-      "loss": 46.0,
-      "step": 41750
-    },
-    {
-      "epoch": 3.192155513504215,
-      "grad_norm": 0.0024962807074189186,
-      "learning_rate": 0.00019999497395014254,
-      "loss": 46.0,
-      "step": 41751
-    },
-    {
-      "epoch": 3.1922319704876045,
-      "grad_norm": 0.0034115491434931755,
-      "learning_rate": 0.00019999497370932093,
-      "loss": 46.0,
-      "step": 41752
-    },
-    {
-      "epoch": 3.1923084274709943,
-      "grad_norm": 0.002311406657099724,
-      "learning_rate": 0.00019999497346849353,
-      "loss": 46.0,
-      "step": 41753
-    },
-    {
-      "epoch": 3.192384884454384,
-      "grad_norm": 0.0033624779898673296,
-      "learning_rate": 0.00019999497322766035,
-      "loss": 46.0,
-      "step": 41754
-    },
-    {
-      "epoch": 3.1924613414377734,
-      "grad_norm": 0.0037651751190423965,
-      "learning_rate": 0.0001999949729868214,
-      "loss": 46.0,
-      "step": 41755
-    },
-    {
-      "epoch": 3.192537798421163,
-      "grad_norm": 0.005287521984428167,
-      "learning_rate": 0.00019999497274597667,
-      "loss": 46.0,
-      "step": 41756
-    },
-    {
-      "epoch": 3.192614255404553,
-      "grad_norm": 0.0037060093600302935,
-      "learning_rate": 0.0001999949725051262,
-      "loss": 46.0,
-      "step": 41757
-    },
-    {
-      "epoch": 3.1926907123879427,
-      "grad_norm": 0.0049739438109099865,
-      "learning_rate": 0.00019999497226426996,
-      "loss": 46.0,
-      "step": 41758
-    },
-    {
-      "epoch": 3.1927671693713324,
-      "grad_norm": 0.002128767780959606,
-      "learning_rate": 0.00019999497202340794,
-      "loss": 46.0,
-      "step": 41759
-    },
-    {
-      "epoch": 3.192843626354722,
-      "grad_norm": 0.0023529413156211376,
-      "learning_rate": 0.00019999497178254015,
-      "loss": 46.0,
-      "step": 41760
-    },
-    {
-      "epoch": 3.192920083338112,
-      "grad_norm": 0.001607896643690765,
-      "learning_rate": 0.00019999497154166658,
-      "loss": 46.0,
-      "step": 41761
-    },
-    {
-      "epoch": 3.1929965403215017,
-      "grad_norm": 0.0025055850856006145,
-      "learning_rate": 0.00019999497130078725,
-      "loss": 46.0,
-      "step": 41762
-    },
-    {
-      "epoch": 3.1930729973048915,
-      "grad_norm": 0.0030576728750020266,
-      "learning_rate": 0.00019999497105990214,
-      "loss": 46.0,
-      "step": 41763
-    },
-    {
-      "epoch": 3.1931494542882812,
-      "grad_norm": 0.0011251089163124561,
-      "learning_rate": 0.00019999497081901128,
-      "loss": 46.0,
-      "step": 41764
-    },
-    {
-      "epoch": 3.1932259112716705,
-      "grad_norm": 0.002464407356455922,
-      "learning_rate": 0.00019999497057811465,
-      "loss": 46.0,
-      "step": 41765
-    },
-    {
-      "epoch": 3.1933023682550603,
-      "grad_norm": 0.0006720838719047606,
-      "learning_rate": 0.00019999497033721224,
-      "loss": 46.0,
-      "step": 41766
-    },
-    {
-      "epoch": 3.19337882523845,
-      "grad_norm": 0.001942755887284875,
-      "learning_rate": 0.0001999949700963041,
-      "loss": 46.0,
-      "step": 41767
-    },
-    {
-      "epoch": 3.19345528222184,
-      "grad_norm": 0.0017107600579038262,
-      "learning_rate": 0.00019999496985539014,
-      "loss": 46.0,
-      "step": 41768
-    },
-    {
-      "epoch": 3.1935317392052296,
-      "grad_norm": 0.0026615303941071033,
-      "learning_rate": 0.00019999496961447042,
-      "loss": 46.0,
-      "step": 41769
-    },
-    {
-      "epoch": 3.1936081961886194,
-      "grad_norm": 0.002414393937215209,
-      "learning_rate": 0.00019999496937354495,
-      "loss": 46.0,
-      "step": 41770
-    },
-    {
-      "epoch": 3.193684653172009,
-      "grad_norm": 0.0010056046303361654,
-      "learning_rate": 0.00019999496913261367,
-      "loss": 46.0,
-      "step": 41771
-    },
-    {
-      "epoch": 3.193761110155399,
-      "grad_norm": 0.0019405658822506666,
-      "learning_rate": 0.00019999496889167668,
-      "loss": 46.0,
-      "step": 41772
-    },
-    {
-      "epoch": 3.1938375671387886,
-      "grad_norm": 0.0018267863197252154,
-      "learning_rate": 0.0001999949686507339,
-      "loss": 46.0,
-      "step": 41773
-    },
-    {
-      "epoch": 3.1939140241221784,
-      "grad_norm": 0.0012507934588938951,
-      "learning_rate": 0.0001999949684097853,
-      "loss": 46.0,
-      "step": 41774
-    },
-    {
-      "epoch": 3.193990481105568,
-      "grad_norm": 0.002086616586893797,
-      "learning_rate": 0.000199994968168831,
-      "loss": 46.0,
-      "step": 41775
-    },
-    {
-      "epoch": 3.194066938088958,
-      "grad_norm": 0.00244074035435915,
-      "learning_rate": 0.00019999496792787088,
-      "loss": 46.0,
-      "step": 41776
-    },
-    {
-      "epoch": 3.1941433950723472,
-      "grad_norm": 0.0034217399079352617,
-      "learning_rate": 0.00019999496768690502,
-      "loss": 46.0,
-      "step": 41777
-    },
-    {
-      "epoch": 3.194219852055737,
-      "grad_norm": 0.0022705956362187862,
-      "learning_rate": 0.0001999949674459334,
-      "loss": 46.0,
-      "step": 41778
-    },
-    {
-      "epoch": 3.1942963090391268,
-      "grad_norm": 0.0023646594490855932,
-      "learning_rate": 0.00019999496720495602,
-      "loss": 46.0,
-      "step": 41779
-    },
-    {
-      "epoch": 3.1943727660225165,
-      "grad_norm": 0.0008587867487221956,
-      "learning_rate": 0.0001999949669639728,
-      "loss": 46.0,
-      "step": 41780
-    },
-    {
-      "epoch": 3.1944492230059063,
-      "grad_norm": 0.0033211889676749706,
-      "learning_rate": 0.0001999949667229839,
-      "loss": 46.0,
-      "step": 41781
-    },
-    {
-      "epoch": 3.194525679989296,
-      "grad_norm": 0.0016654988285154104,
-      "learning_rate": 0.00019999496648198916,
-      "loss": 46.0,
-      "step": 41782
-    },
-    {
-      "epoch": 3.194602136972686,
-      "grad_norm": 0.0007194314384832978,
-      "learning_rate": 0.0001999949662409887,
-      "loss": 46.0,
-      "step": 41783
-    },
-    {
-      "epoch": 3.1946785939560756,
-      "grad_norm": 0.001223646104335785,
-      "learning_rate": 0.00019999496599998245,
-      "loss": 46.0,
-      "step": 41784
-    },
-    {
-      "epoch": 3.1947550509394653,
-      "grad_norm": 0.0029901759698987007,
-      "learning_rate": 0.0001999949657589704,
-      "loss": 46.0,
-      "step": 41785
-    },
-    {
-      "epoch": 3.194831507922855,
-      "grad_norm": 0.0014829349238425493,
-      "learning_rate": 0.0001999949655179526,
-      "loss": 46.0,
-      "step": 41786
-    },
-    {
-      "epoch": 3.1949079649062444,
-      "grad_norm": 0.007151391822844744,
-      "learning_rate": 0.00019999496527692908,
-      "loss": 46.0,
-      "step": 41787
-    },
-    {
-      "epoch": 3.194984421889634,
-      "grad_norm": 0.0025312264915555716,
-      "learning_rate": 0.00019999496503589974,
-      "loss": 46.0,
-      "step": 41788
-    },
-    {
-      "epoch": 3.195060878873024,
-      "grad_norm": 0.0014617975102737546,
-      "learning_rate": 0.00019999496479486466,
-      "loss": 46.0,
-      "step": 41789
-    },
-    {
-      "epoch": 3.1951373358564137,
-      "grad_norm": 0.0018159488681703806,
-      "learning_rate": 0.0001999949645538238,
-      "loss": 46.0,
-      "step": 41790
-    },
-    {
-      "epoch": 3.1952137928398034,
-      "grad_norm": 0.0018441840074956417,
-      "learning_rate": 0.00019999496431277717,
-      "loss": 46.0,
-      "step": 41791
-    },
-    {
-      "epoch": 3.195290249823193,
-      "grad_norm": 0.0013451763661578298,
-      "learning_rate": 0.00019999496407172476,
-      "loss": 46.0,
-      "step": 41792
-    },
-    {
-      "epoch": 3.195366706806583,
-      "grad_norm": 0.00121031585149467,
-      "learning_rate": 0.0001999949638306666,
-      "loss": 46.0,
-      "step": 41793
-    },
-    {
-      "epoch": 3.1954431637899727,
-      "grad_norm": 0.0065407538786530495,
-      "learning_rate": 0.00019999496358960266,
-      "loss": 46.0,
-      "step": 41794
-    },
-    {
-      "epoch": 3.1955196207733625,
-      "grad_norm": 0.002348933834582567,
-      "learning_rate": 0.00019999496334853294,
-      "loss": 46.0,
-      "step": 41795
-    },
-    {
-      "epoch": 3.1955960777567523,
-      "grad_norm": 0.003194976132363081,
-      "learning_rate": 0.00019999496310745744,
-      "loss": 46.0,
-      "step": 41796
-    },
-    {
-      "epoch": 3.195672534740142,
-      "grad_norm": 0.005215474870055914,
-      "learning_rate": 0.0001999949628663762,
-      "loss": 46.0,
-      "step": 41797
-    },
-    {
-      "epoch": 3.1957489917235318,
-      "grad_norm": 0.002075603697448969,
-      "learning_rate": 0.00019999496262528918,
-      "loss": 46.0,
-      "step": 41798
-    },
-    {
-      "epoch": 3.195825448706921,
-      "grad_norm": 0.0029091532342135906,
-      "learning_rate": 0.0001999949623841964,
-      "loss": 46.0,
-      "step": 41799
-    },
-    {
-      "epoch": 3.195901905690311,
-      "grad_norm": 0.007599889300763607,
-      "learning_rate": 0.00019999496214309783,
-      "loss": 46.0,
-      "step": 41800
-    },
-    {
-      "epoch": 3.1959783626737006,
-      "grad_norm": 0.0034401395823806524,
-      "learning_rate": 0.00019999496190199352,
-      "loss": 46.0,
-      "step": 41801
-    },
-    {
-      "epoch": 3.1960548196570904,
-      "grad_norm": 0.0019381269812583923,
-      "learning_rate": 0.00019999496166088344,
-      "loss": 46.0,
-      "step": 41802
-    },
-    {
-      "epoch": 3.19613127664048,
-      "grad_norm": 0.002752471948042512,
-      "learning_rate": 0.00019999496141976758,
-      "loss": 46.0,
-      "step": 41803
-    },
-    {
-      "epoch": 3.19620773362387,
-      "grad_norm": 0.003460479434579611,
-      "learning_rate": 0.00019999496117864592,
-      "loss": 46.0,
-      "step": 41804
-    },
-    {
-      "epoch": 3.1962841906072597,
-      "grad_norm": 0.006550750695168972,
-      "learning_rate": 0.00019999496093751852,
-      "loss": 46.0,
-      "step": 41805
-    },
-    {
-      "epoch": 3.1963606475906494,
-      "grad_norm": 0.002637798897922039,
-      "learning_rate": 0.00019999496069638534,
-      "loss": 46.0,
-      "step": 41806
-    },
-    {
-      "epoch": 3.196437104574039,
-      "grad_norm": 0.003196987323462963,
-      "learning_rate": 0.00019999496045524642,
-      "loss": 46.0,
-      "step": 41807
-    },
-    {
-      "epoch": 3.196513561557429,
-      "grad_norm": 0.0011728891404345632,
-      "learning_rate": 0.00019999496021410172,
-      "loss": 46.0,
-      "step": 41808
-    },
-    {
-      "epoch": 3.1965900185408183,
-      "grad_norm": 0.0022812115494161844,
-      "learning_rate": 0.00019999495997295123,
-      "loss": 46.0,
-      "step": 41809
-    },
-    {
-      "epoch": 3.196666475524208,
-      "grad_norm": 0.004441450349986553,
-      "learning_rate": 0.00019999495973179498,
-      "loss": 46.0,
-      "step": 41810
-    },
-    {
-      "epoch": 3.196742932507598,
-      "grad_norm": 0.001541881007142365,
-      "learning_rate": 0.00019999495949063297,
-      "loss": 46.0,
-      "step": 41811
-    },
-    {
-      "epoch": 3.1968193894909875,
-      "grad_norm": 0.005147553980350494,
-      "learning_rate": 0.00019999495924946518,
-      "loss": 46.0,
-      "step": 41812
-    },
-    {
-      "epoch": 3.1968958464743773,
-      "grad_norm": 0.0023873900063335896,
-      "learning_rate": 0.00019999495900829164,
-      "loss": 46.0,
-      "step": 41813
-    },
-    {
-      "epoch": 3.196972303457767,
-      "grad_norm": 0.0005892972694709897,
-      "learning_rate": 0.0001999949587671123,
-      "loss": 46.0,
-      "step": 41814
-    },
-    {
-      "epoch": 3.197048760441157,
-      "grad_norm": 0.00452373456209898,
-      "learning_rate": 0.0001999949585259272,
-      "loss": 46.0,
-      "step": 41815
-    },
-    {
-      "epoch": 3.1971252174245466,
-      "grad_norm": 0.007730280980467796,
-      "learning_rate": 0.00019999495828473634,
-      "loss": 46.0,
-      "step": 41816
-    },
-    {
-      "epoch": 3.1972016744079363,
-      "grad_norm": 0.005197593942284584,
-      "learning_rate": 0.0001999949580435397,
-      "loss": 46.0,
-      "step": 41817
-    },
-    {
-      "epoch": 3.197278131391326,
-      "grad_norm": 0.0023051961325109005,
-      "learning_rate": 0.00019999495780233733,
-      "loss": 46.0,
-      "step": 41818
-    },
-    {
-      "epoch": 3.197354588374716,
-      "grad_norm": 0.0032664304599165916,
-      "learning_rate": 0.00019999495756112913,
-      "loss": 46.0,
-      "step": 41819
-    },
-    {
-      "epoch": 3.1974310453581056,
-      "grad_norm": 0.0015942684840410948,
-      "learning_rate": 0.0001999949573199152,
-      "loss": 46.0,
-      "step": 41820
-    },
-    {
-      "epoch": 3.197507502341495,
-      "grad_norm": 0.0016448040260002017,
-      "learning_rate": 0.00019999495707869548,
-      "loss": 46.0,
-      "step": 41821
-    },
-    {
-      "epoch": 3.1975839593248847,
-      "grad_norm": 0.0010918325278908014,
-      "learning_rate": 0.00019999495683747004,
-      "loss": 46.0,
-      "step": 41822
-    },
-    {
-      "epoch": 3.1976604163082745,
-      "grad_norm": 0.0012299743248149753,
-      "learning_rate": 0.00019999495659623877,
-      "loss": 46.0,
-      "step": 41823
-    },
-    {
-      "epoch": 3.1977368732916642,
-      "grad_norm": 0.0028943598736077547,
-      "learning_rate": 0.00019999495635500175,
-      "loss": 46.0,
-      "step": 41824
-    },
-    {
-      "epoch": 3.197813330275054,
-      "grad_norm": 0.0018121174070984125,
-      "learning_rate": 0.000199994956113759,
-      "loss": 46.0,
-      "step": 41825
-    },
-    {
-      "epoch": 3.1978897872584438,
-      "grad_norm": 0.0013667203020304441,
-      "learning_rate": 0.0001999949558725104,
-      "loss": 46.0,
-      "step": 41826
-    },
-    {
-      "epoch": 3.1979662442418335,
-      "grad_norm": 0.0018197777681052685,
-      "learning_rate": 0.0001999949556312561,
-      "loss": 46.0,
-      "step": 41827
-    },
-    {
-      "epoch": 3.1980427012252233,
-      "grad_norm": 0.0009614763548597693,
-      "learning_rate": 0.000199994955389996,
-      "loss": 46.0,
-      "step": 41828
-    },
-    {
-      "epoch": 3.198119158208613,
-      "grad_norm": 0.0006617567269131541,
-      "learning_rate": 0.00019999495514873012,
-      "loss": 46.0,
-      "step": 41829
-    },
-    {
-      "epoch": 3.198195615192003,
-      "grad_norm": 0.0038372681010514498,
-      "learning_rate": 0.0001999949549074585,
-      "loss": 46.0,
-      "step": 41830
-    },
-    {
-      "epoch": 3.198272072175392,
-      "grad_norm": 0.003438947955146432,
-      "learning_rate": 0.0001999949546661811,
-      "loss": 46.0,
-      "step": 41831
-    },
-    {
-      "epoch": 3.198348529158782,
-      "grad_norm": 0.0050773415714502335,
-      "learning_rate": 0.00019999495442489792,
-      "loss": 46.0,
-      "step": 41832
-    },
-    {
-      "epoch": 3.1984249861421716,
-      "grad_norm": 0.0039446172304451466,
-      "learning_rate": 0.000199994954183609,
-      "loss": 46.0,
-      "step": 41833
-    },
-    {
-      "epoch": 3.1985014431255614,
-      "grad_norm": 0.001261175493709743,
-      "learning_rate": 0.0001999949539423143,
-      "loss": 46.0,
-      "step": 41834
-    },
-    {
-      "epoch": 3.198577900108951,
-      "grad_norm": 0.006266809534281492,
-      "learning_rate": 0.00019999495370101383,
-      "loss": 46.0,
-      "step": 41835
-    },
-    {
-      "epoch": 3.198654357092341,
-      "grad_norm": 0.00084760005120188,
-      "learning_rate": 0.00019999495345970756,
-      "loss": 46.0,
-      "step": 41836
-    },
-    {
-      "epoch": 3.1987308140757307,
-      "grad_norm": 0.002254942199215293,
-      "learning_rate": 0.00019999495321839557,
-      "loss": 46.0,
-      "step": 41837
-    },
-    {
-      "epoch": 3.1988072710591204,
-      "grad_norm": 0.0026623986195772886,
-      "learning_rate": 0.00019999495297707775,
-      "loss": 46.0,
-      "step": 41838
-    },
-    {
-      "epoch": 3.19888372804251,
-      "grad_norm": 0.0006011240184307098,
-      "learning_rate": 0.00019999495273575422,
-      "loss": 46.0,
-      "step": 41839
-    },
-    {
-      "epoch": 3.1989601850259,
-      "grad_norm": 0.002485383301973343,
-      "learning_rate": 0.0001999949524944249,
-      "loss": 46.0,
-      "step": 41840
-    },
-    {
-      "epoch": 3.1990366420092897,
-      "grad_norm": 0.0037892533000558615,
-      "learning_rate": 0.0001999949522530898,
-      "loss": 46.0,
-      "step": 41841
-    },
-    {
-      "epoch": 3.199113098992679,
-      "grad_norm": 0.0013631790643557906,
-      "learning_rate": 0.00019999495201174894,
-      "loss": 46.0,
-      "step": 41842
-    },
-    {
-      "epoch": 3.199189555976069,
-      "grad_norm": 0.0029840683564543724,
-      "learning_rate": 0.00019999495177040231,
-      "loss": 46.0,
-      "step": 41843
-    },
-    {
-      "epoch": 3.1992660129594586,
-      "grad_norm": 0.00284794089384377,
-      "learning_rate": 0.0001999949515290499,
-      "loss": 46.0,
-      "step": 41844
-    },
-    {
-      "epoch": 3.1993424699428483,
-      "grad_norm": 0.005058239679783583,
-      "learning_rate": 0.00019999495128769174,
-      "loss": 46.0,
-      "step": 41845
-    },
-    {
-      "epoch": 3.199418926926238,
-      "grad_norm": 0.005141830071806908,
-      "learning_rate": 0.0001999949510463278,
-      "loss": 46.0,
-      "step": 41846
-    },
-    {
-      "epoch": 3.199495383909628,
-      "grad_norm": 0.001782839186489582,
-      "learning_rate": 0.0001999949508049581,
-      "loss": 46.0,
-      "step": 41847
-    },
-    {
-      "epoch": 3.1995718408930176,
-      "grad_norm": 0.0022425889037549496,
-      "learning_rate": 0.00019999495056358262,
-      "loss": 46.0,
-      "step": 41848
-    },
-    {
-      "epoch": 3.1996482978764074,
-      "grad_norm": 0.002025061519816518,
-      "learning_rate": 0.00019999495032220138,
-      "loss": 46.0,
-      "step": 41849
-    },
-    {
-      "epoch": 3.199724754859797,
-      "grad_norm": 0.0022437439765781164,
-      "learning_rate": 0.00019999495008081434,
-      "loss": 46.0,
-      "step": 41850
-    },
-    {
-      "epoch": 3.199801211843187,
-      "grad_norm": 0.0021829060278832912,
-      "learning_rate": 0.00019999494983942158,
-      "loss": 46.0,
-      "step": 41851
-    },
-    {
-      "epoch": 3.1998776688265766,
-      "grad_norm": 0.0014959339750930667,
-      "learning_rate": 0.00019999494959802304,
-      "loss": 46.0,
-      "step": 41852
-    },
-    {
-      "epoch": 3.199954125809966,
-      "grad_norm": 0.0015004135202616453,
-      "learning_rate": 0.0001999949493566187,
-      "loss": 46.0,
-      "step": 41853
-    },
-    {
-      "epoch": 3.2000305827933557,
-      "grad_norm": 0.0006117904558777809,
-      "learning_rate": 0.00019999494911520862,
-      "loss": 46.0,
-      "step": 41854
-    },
-    {
-      "epoch": 3.2001070397767455,
-      "grad_norm": 0.004115217365324497,
-      "learning_rate": 0.00019999494887379274,
-      "loss": 46.0,
-      "step": 41855
-    },
-    {
-      "epoch": 3.2001834967601352,
-      "grad_norm": 0.0031721347477287054,
-      "learning_rate": 0.0001999949486323711,
-      "loss": 46.0,
-      "step": 41856
-    },
-    {
-      "epoch": 3.200259953743525,
-      "grad_norm": 0.0028074639849364758,
-      "learning_rate": 0.00019999494839094374,
-      "loss": 46.0,
-      "step": 41857
-    },
-    {
-      "epoch": 3.2003364107269148,
-      "grad_norm": 0.0021092924289405346,
-      "learning_rate": 0.00019999494814951056,
-      "loss": 46.0,
-      "step": 41858
-    },
-    {
-      "epoch": 3.2004128677103045,
-      "grad_norm": 0.006816784385591745,
-      "learning_rate": 0.0001999949479080716,
-      "loss": 46.0,
-      "step": 41859
-    },
-    {
-      "epoch": 3.2004893246936943,
-      "grad_norm": 0.004569733515381813,
-      "learning_rate": 0.00019999494766662692,
-      "loss": 46.0,
-      "step": 41860
-    },
-    {
-      "epoch": 3.200565781677084,
-      "grad_norm": 0.0033006465528160334,
-      "learning_rate": 0.00019999494742517645,
-      "loss": 46.0,
-      "step": 41861
-    },
-    {
-      "epoch": 3.200642238660474,
-      "grad_norm": 0.004284622147679329,
-      "learning_rate": 0.00019999494718372018,
-      "loss": 46.0,
-      "step": 41862
-    },
-    {
-      "epoch": 3.2007186956438636,
-      "grad_norm": 0.001976188039407134,
-      "learning_rate": 0.00019999494694225816,
-      "loss": 46.0,
-      "step": 41863
-    },
-    {
-      "epoch": 3.200795152627253,
-      "grad_norm": 0.001034342683851719,
-      "learning_rate": 0.0001999949467007904,
-      "loss": 46.0,
-      "step": 41864
-    },
-    {
-      "epoch": 3.2008716096106427,
-      "grad_norm": 0.0008406956330873072,
-      "learning_rate": 0.00019999494645931684,
-      "loss": 46.0,
-      "step": 41865
-    },
-    {
-      "epoch": 3.2009480665940324,
-      "grad_norm": 0.003191269002854824,
-      "learning_rate": 0.00019999494621783753,
-      "loss": 46.0,
-      "step": 41866
-    },
-    {
-      "epoch": 3.201024523577422,
-      "grad_norm": 0.0020116691011935472,
-      "learning_rate": 0.00019999494597635242,
-      "loss": 46.0,
-      "step": 41867
-    },
-    {
-      "epoch": 3.201100980560812,
-      "grad_norm": 0.004101854749023914,
-      "learning_rate": 0.0001999949457348616,
-      "loss": 46.0,
-      "step": 41868
-    },
-    {
-      "epoch": 3.2011774375442017,
-      "grad_norm": 0.0007809695671312511,
-      "learning_rate": 0.00019999494549336494,
-      "loss": 46.0,
-      "step": 41869
-    },
-    {
-      "epoch": 3.2012538945275915,
-      "grad_norm": 0.001870789099484682,
-      "learning_rate": 0.00019999494525186257,
-      "loss": 46.0,
-      "step": 41870
-    },
-    {
-      "epoch": 3.201330351510981,
-      "grad_norm": 0.005278183147311211,
-      "learning_rate": 0.0001999949450103544,
-      "loss": 46.0,
-      "step": 41871
-    },
-    {
-      "epoch": 3.201406808494371,
-      "grad_norm": 0.0022934924345463514,
-      "learning_rate": 0.00019999494476884047,
-      "loss": 46.0,
-      "step": 41872
-    },
-    {
-      "epoch": 3.2014832654777607,
-      "grad_norm": 0.001164207118563354,
-      "learning_rate": 0.00019999494452732075,
-      "loss": 46.0,
-      "step": 41873
-    },
-    {
-      "epoch": 3.2015597224611505,
-      "grad_norm": 0.00329647958278656,
-      "learning_rate": 0.00019999494428579525,
-      "loss": 46.0,
-      "step": 41874
-    },
-    {
-      "epoch": 3.20163617944454,
-      "grad_norm": 0.0018871875945478678,
-      "learning_rate": 0.000199994944044264,
-      "loss": 46.0,
-      "step": 41875
-    },
-    {
-      "epoch": 3.2017126364279296,
-      "grad_norm": 0.0013439093017950654,
-      "learning_rate": 0.00019999494380272702,
-      "loss": 46.0,
-      "step": 41876
-    },
-    {
-      "epoch": 3.2017890934113193,
-      "grad_norm": 0.003036444541066885,
-      "learning_rate": 0.00019999494356118423,
-      "loss": 46.0,
-      "step": 41877
-    },
-    {
-      "epoch": 3.201865550394709,
-      "grad_norm": 0.002536752261221409,
-      "learning_rate": 0.0001999949433196357,
-      "loss": 46.0,
-      "step": 41878
-    },
-    {
-      "epoch": 3.201942007378099,
-      "grad_norm": 0.0018414033111184835,
-      "learning_rate": 0.00019999494307808136,
-      "loss": 46.0,
-      "step": 41879
-    },
-    {
-      "epoch": 3.2020184643614886,
-      "grad_norm": 0.0028950015548616648,
-      "learning_rate": 0.00019999494283652128,
-      "loss": 46.0,
-      "step": 41880
-    },
-    {
-      "epoch": 3.2020949213448784,
-      "grad_norm": 0.0033415048383176327,
-      "learning_rate": 0.00019999494259495543,
-      "loss": 46.0,
-      "step": 41881
-    },
-    {
-      "epoch": 3.202171378328268,
-      "grad_norm": 0.000592630822211504,
-      "learning_rate": 0.00019999494235338377,
-      "loss": 46.0,
-      "step": 41882
-    },
-    {
-      "epoch": 3.202247835311658,
-      "grad_norm": 0.0023894303012639284,
-      "learning_rate": 0.0001999949421118064,
-      "loss": 46.0,
-      "step": 41883
-    },
-    {
-      "epoch": 3.2023242922950477,
-      "grad_norm": 0.0018253163434565067,
-      "learning_rate": 0.00019999494187022322,
-      "loss": 46.0,
-      "step": 41884
-    },
-    {
-      "epoch": 3.2024007492784374,
-      "grad_norm": 0.0028539958875626326,
-      "learning_rate": 0.00019999494162863427,
-      "loss": 46.0,
-      "step": 41885
-    },
-    {
-      "epoch": 3.2024772062618267,
-      "grad_norm": 0.0026212134398519993,
-      "learning_rate": 0.00019999494138703958,
-      "loss": 46.0,
-      "step": 41886
-    },
-    {
-      "epoch": 3.2025536632452165,
-      "grad_norm": 0.002570651937276125,
-      "learning_rate": 0.0001999949411454391,
-      "loss": 46.0,
-      "step": 41887
-    },
-    {
-      "epoch": 3.2026301202286063,
-      "grad_norm": 0.0013720061397179961,
-      "learning_rate": 0.00019999494090383287,
-      "loss": 46.0,
-      "step": 41888
-    },
-    {
-      "epoch": 3.202706577211996,
-      "grad_norm": 0.0021675988100469112,
-      "learning_rate": 0.00019999494066222086,
-      "loss": 46.0,
-      "step": 41889
-    },
-    {
-      "epoch": 3.202783034195386,
-      "grad_norm": 0.0011974771041423082,
-      "learning_rate": 0.0001999949404206031,
-      "loss": 46.0,
-      "step": 41890
-    },
-    {
-      "epoch": 3.2028594911787756,
-      "grad_norm": 0.0035049342550337315,
-      "learning_rate": 0.00019999494017897953,
-      "loss": 46.0,
-      "step": 41891
-    },
-    {
-      "epoch": 3.2029359481621653,
-      "grad_norm": 0.0011456075590103865,
-      "learning_rate": 0.0001999949399373502,
-      "loss": 46.0,
-      "step": 41892
-    },
-    {
-      "epoch": 3.203012405145555,
-      "grad_norm": 0.003045186400413513,
-      "learning_rate": 0.00019999493969571512,
-      "loss": 46.0,
-      "step": 41893
-    },
-    {
-      "epoch": 3.203088862128945,
-      "grad_norm": 0.0013950093416497111,
-      "learning_rate": 0.0001999949394540743,
-      "loss": 46.0,
-      "step": 41894
-    },
-    {
-      "epoch": 3.2031653191123346,
-      "grad_norm": 0.0026571836788207293,
-      "learning_rate": 0.00019999493921242764,
-      "loss": 46.0,
-      "step": 41895
-    },
-    {
-      "epoch": 3.203241776095724,
-      "grad_norm": 0.0037363844458013773,
-      "learning_rate": 0.00019999493897077523,
-      "loss": 46.0,
-      "step": 41896
-    },
-    {
-      "epoch": 3.2033182330791137,
-      "grad_norm": 0.0033596395514905453,
-      "learning_rate": 0.0001999949387291171,
-      "loss": 46.0,
-      "step": 41897
-    },
-    {
-      "epoch": 3.2033946900625034,
-      "grad_norm": 0.003285108832642436,
-      "learning_rate": 0.00019999493848745314,
-      "loss": 46.0,
-      "step": 41898
-    },
-    {
-      "epoch": 3.203471147045893,
-      "grad_norm": 0.0010430342517793179,
-      "learning_rate": 0.00019999493824578344,
-      "loss": 46.0,
-      "step": 41899
-    },
-    {
-      "epoch": 3.203547604029283,
-      "grad_norm": 0.0010957475751638412,
-      "learning_rate": 0.00019999493800410795,
-      "loss": 46.0,
-      "step": 41900
-    },
-    {
-      "epoch": 3.2036240610126727,
-      "grad_norm": 0.0017878767102956772,
-      "learning_rate": 0.00019999493776242674,
-      "loss": 46.0,
-      "step": 41901
-    },
-    {
-      "epoch": 3.2037005179960625,
-      "grad_norm": 0.0009199641062878072,
-      "learning_rate": 0.00019999493752073972,
-      "loss": 46.0,
-      "step": 41902
-    },
-    {
-      "epoch": 3.2037769749794522,
-      "grad_norm": 0.0007861303747631609,
-      "learning_rate": 0.00019999493727904693,
-      "loss": 46.0,
-      "step": 41903
-    },
-    {
-      "epoch": 3.203853431962842,
-      "grad_norm": 0.004879381973296404,
-      "learning_rate": 0.0001999949370373484,
-      "loss": 46.0,
-      "step": 41904
-    },
-    {
-      "epoch": 3.2039298889462318,
-      "grad_norm": 0.000710874970536679,
-      "learning_rate": 0.00019999493679564407,
-      "loss": 46.0,
-      "step": 41905
-    },
-    {
-      "epoch": 3.2040063459296215,
-      "grad_norm": 0.004271247889846563,
-      "learning_rate": 0.00019999493655393398,
-      "loss": 46.0,
-      "step": 41906
-    },
-    {
-      "epoch": 3.2040828029130113,
-      "grad_norm": 0.002667555818334222,
-      "learning_rate": 0.00019999493631221813,
-      "loss": 46.0,
-      "step": 41907
-    },
-    {
-      "epoch": 3.2041592598964006,
-      "grad_norm": 0.002643396146595478,
-      "learning_rate": 0.0001999949360704965,
-      "loss": 46.0,
-      "step": 41908
-    },
-    {
-      "epoch": 3.2042357168797904,
-      "grad_norm": 0.011871416121721268,
-      "learning_rate": 0.0001999949358287691,
-      "loss": 46.0,
-      "step": 41909
-    },
-    {
-      "epoch": 3.20431217386318,
-      "grad_norm": 0.0012360658729448915,
-      "learning_rate": 0.00019999493558703595,
-      "loss": 46.0,
-      "step": 41910
-    },
-    {
-      "epoch": 3.20438863084657,
-      "grad_norm": 0.0016530659049749374,
-      "learning_rate": 0.000199994935345297,
-      "loss": 46.0,
-      "step": 41911
-    },
-    {
-      "epoch": 3.2044650878299596,
-      "grad_norm": 0.003976988606154919,
-      "learning_rate": 0.00019999493510355229,
-      "loss": 46.0,
-      "step": 41912
-    },
-    {
-      "epoch": 3.2045415448133494,
-      "grad_norm": 0.0030321250669658184,
-      "learning_rate": 0.00019999493486180182,
-      "loss": 46.0,
-      "step": 41913
-    },
-    {
-      "epoch": 3.204618001796739,
-      "grad_norm": 0.004816477186977863,
-      "learning_rate": 0.0001999949346200456,
-      "loss": 46.0,
-      "step": 41914
-    },
-    {
-      "epoch": 3.204694458780129,
-      "grad_norm": 0.0024308855645358562,
-      "learning_rate": 0.0001999949343782836,
-      "loss": 46.0,
-      "step": 41915
-    },
-    {
-      "epoch": 3.2047709157635187,
-      "grad_norm": 0.0048491195775568485,
-      "learning_rate": 0.0001999949341365158,
-      "loss": 46.0,
-      "step": 41916
-    },
-    {
-      "epoch": 3.2048473727469085,
-      "grad_norm": 0.007922302931547165,
-      "learning_rate": 0.00019999493389474227,
-      "loss": 46.0,
-      "step": 41917
-    },
-    {
-      "epoch": 3.2049238297302978,
-      "grad_norm": 0.008576560765504837,
-      "learning_rate": 0.00019999493365296294,
-      "loss": 46.0,
-      "step": 41918
-    },
-    {
-      "epoch": 3.2050002867136875,
-      "grad_norm": 0.0024287390988320112,
-      "learning_rate": 0.00019999493341117786,
-      "loss": 46.0,
-      "step": 41919
-    },
-    {
-      "epoch": 3.2050767436970773,
-      "grad_norm": 0.0032086092978715897,
-      "learning_rate": 0.000199994933169387,
-      "loss": 46.0,
-      "step": 41920
-    },
-    {
-      "epoch": 3.205153200680467,
-      "grad_norm": 0.0018151280237361789,
-      "learning_rate": 0.00019999493292759037,
-      "loss": 46.0,
-      "step": 41921
-    },
-    {
-      "epoch": 3.205229657663857,
-      "grad_norm": 0.00288069318048656,
-      "learning_rate": 0.000199994932685788,
-      "loss": 46.0,
-      "step": 41922
-    },
-    {
-      "epoch": 3.2053061146472466,
-      "grad_norm": 0.0018049966311082244,
-      "learning_rate": 0.00019999493244397983,
-      "loss": 46.0,
-      "step": 41923
-    },
-    {
-      "epoch": 3.2053825716306363,
-      "grad_norm": 0.00171754602342844,
-      "learning_rate": 0.00019999493220216588,
-      "loss": 46.0,
-      "step": 41924
-    },
-    {
-      "epoch": 3.205459028614026,
-      "grad_norm": 0.0014012206811457872,
-      "learning_rate": 0.00019999493196034619,
-      "loss": 46.0,
-      "step": 41925
-    },
-    {
-      "epoch": 3.205535485597416,
-      "grad_norm": 0.0010059995111078024,
-      "learning_rate": 0.00019999493171852072,
-      "loss": 46.0,
-      "step": 41926
-    },
-    {
-      "epoch": 3.2056119425808056,
-      "grad_norm": 0.0017957398667931557,
-      "learning_rate": 0.00019999493147668948,
-      "loss": 46.0,
-      "step": 41927
-    },
-    {
-      "epoch": 3.2056883995641954,
-      "grad_norm": 0.0013694277731701732,
-      "learning_rate": 0.00019999493123485246,
-      "loss": 46.0,
-      "step": 41928
-    },
-    {
-      "epoch": 3.205764856547585,
-      "grad_norm": 0.0008041377877816558,
-      "learning_rate": 0.00019999493099300968,
-      "loss": 46.0,
-      "step": 41929
-    },
-    {
-      "epoch": 3.2058413135309745,
-      "grad_norm": 0.003393413731828332,
-      "learning_rate": 0.00019999493075116114,
-      "loss": 46.0,
-      "step": 41930
-    },
-    {
-      "epoch": 3.205917770514364,
-      "grad_norm": 0.003929989878088236,
-      "learning_rate": 0.0001999949305093068,
-      "loss": 46.0,
-      "step": 41931
-    },
-    {
-      "epoch": 3.205994227497754,
-      "grad_norm": 0.001302413991652429,
-      "learning_rate": 0.00019999493026744673,
-      "loss": 46.0,
-      "step": 41932
-    },
-    {
-      "epoch": 3.2060706844811437,
-      "grad_norm": 0.002498259535059333,
-      "learning_rate": 0.00019999493002558088,
-      "loss": 46.0,
-      "step": 41933
-    },
-    {
-      "epoch": 3.2061471414645335,
-      "grad_norm": 0.006291416939347982,
-      "learning_rate": 0.00019999492978370925,
-      "loss": 46.0,
-      "step": 41934
-    },
-    {
-      "epoch": 3.2062235984479233,
-      "grad_norm": 0.003000830067321658,
-      "learning_rate": 0.00019999492954183188,
-      "loss": 46.0,
-      "step": 41935
-    },
-    {
-      "epoch": 3.206300055431313,
-      "grad_norm": 0.0031903102062642574,
-      "learning_rate": 0.0001999949292999487,
-      "loss": 46.0,
-      "step": 41936
-    },
-    {
-      "epoch": 3.206376512414703,
-      "grad_norm": 0.0016998659120872617,
-      "learning_rate": 0.00019999492905805978,
-      "loss": 46.0,
-      "step": 41937
-    },
-    {
-      "epoch": 3.2064529693980925,
-      "grad_norm": 0.0021885468158870935,
-      "learning_rate": 0.00019999492881616506,
-      "loss": 46.0,
-      "step": 41938
-    },
-    {
-      "epoch": 3.2065294263814823,
-      "grad_norm": 0.00403962517157197,
-      "learning_rate": 0.0001999949285742646,
-      "loss": 46.0,
-      "step": 41939
-    },
-    {
-      "epoch": 3.2066058833648716,
-      "grad_norm": 0.002945851068943739,
-      "learning_rate": 0.00019999492833235838,
-      "loss": 46.0,
-      "step": 41940
-    },
-    {
-      "epoch": 3.2066823403482614,
-      "grad_norm": 0.0033174597192555666,
-      "learning_rate": 0.00019999492809044634,
-      "loss": 46.0,
-      "step": 41941
-    },
-    {
-      "epoch": 3.206758797331651,
-      "grad_norm": 0.0009510880336165428,
-      "learning_rate": 0.00019999492784852858,
-      "loss": 46.0,
-      "step": 41942
-    },
-    {
-      "epoch": 3.206835254315041,
-      "grad_norm": 0.002162659540772438,
-      "learning_rate": 0.00019999492760660502,
-      "loss": 46.0,
-      "step": 41943
-    },
-    {
-      "epoch": 3.2069117112984307,
-      "grad_norm": 0.001432416494935751,
-      "learning_rate": 0.00019999492736467574,
-      "loss": 46.0,
-      "step": 41944
-    },
-    {
-      "epoch": 3.2069881682818204,
-      "grad_norm": 0.001979827880859375,
-      "learning_rate": 0.00019999492712274064,
-      "loss": 46.0,
-      "step": 41945
-    },
-    {
-      "epoch": 3.20706462526521,
-      "grad_norm": 0.0029415723402053118,
-      "learning_rate": 0.00019999492688079978,
-      "loss": 46.0,
-      "step": 41946
-    },
-    {
-      "epoch": 3.2071410822486,
-      "grad_norm": 0.0038416923489421606,
-      "learning_rate": 0.00019999492663885316,
-      "loss": 46.0,
-      "step": 41947
-    },
-    {
-      "epoch": 3.2072175392319897,
-      "grad_norm": 0.004409688524901867,
-      "learning_rate": 0.00019999492639690076,
-      "loss": 46.0,
-      "step": 41948
-    },
-    {
-      "epoch": 3.2072939962153795,
-      "grad_norm": 0.0007211290649138391,
-      "learning_rate": 0.00019999492615494258,
-      "loss": 46.0,
-      "step": 41949
-    },
-    {
-      "epoch": 3.2073704531987692,
-      "grad_norm": 0.0020159073173999786,
-      "learning_rate": 0.00019999492591297866,
-      "loss": 46.0,
-      "step": 41950
-    },
-    {
-      "epoch": 3.207446910182159,
-      "grad_norm": 0.0025301771238446236,
-      "learning_rate": 0.00019999492567100897,
-      "loss": 46.0,
-      "step": 41951
-    },
-    {
-      "epoch": 3.2075233671655483,
-      "grad_norm": 0.002478532027453184,
-      "learning_rate": 0.0001999949254290335,
-      "loss": 46.0,
-      "step": 41952
-    },
-    {
-      "epoch": 3.207599824148938,
-      "grad_norm": 0.0013402156764641404,
-      "learning_rate": 0.00019999492518705226,
-      "loss": 46.0,
-      "step": 41953
-    },
-    {
-      "epoch": 3.207676281132328,
-      "grad_norm": 0.0005637327558360994,
-      "learning_rate": 0.00019999492494506528,
-      "loss": 46.0,
-      "step": 41954
-    },
-    {
-      "epoch": 3.2077527381157176,
-      "grad_norm": 0.0008347879629582167,
-      "learning_rate": 0.0001999949247030725,
-      "loss": 46.0,
-      "step": 41955
-    },
-    {
-      "epoch": 3.2078291950991074,
-      "grad_norm": 0.003693429986014962,
-      "learning_rate": 0.00019999492446107393,
-      "loss": 46.0,
-      "step": 41956
-    },
-    {
-      "epoch": 3.207905652082497,
-      "grad_norm": 0.0019132073502987623,
-      "learning_rate": 0.00019999492421906963,
-      "loss": 46.0,
-      "step": 41957
-    },
-    {
-      "epoch": 3.207982109065887,
-      "grad_norm": 0.004490647930651903,
-      "learning_rate": 0.00019999492397705955,
-      "loss": 46.0,
-      "step": 41958
-    },
-    {
-      "epoch": 3.2080585660492766,
-      "grad_norm": 0.0010885039810091257,
-      "learning_rate": 0.00019999492373504367,
-      "loss": 46.0,
-      "step": 41959
-    },
-    {
-      "epoch": 3.2081350230326664,
-      "grad_norm": 0.00437463354319334,
-      "learning_rate": 0.00019999492349302207,
-      "loss": 46.0,
-      "step": 41960
-    },
-    {
-      "epoch": 3.208211480016056,
-      "grad_norm": 0.0026589494664222,
-      "learning_rate": 0.00019999492325099467,
-      "loss": 46.0,
-      "step": 41961
-    },
-    {
-      "epoch": 3.2082879369994455,
-      "grad_norm": 0.0025310181081295013,
-      "learning_rate": 0.0001999949230089615,
-      "loss": 46.0,
-      "step": 41962
-    },
-    {
-      "epoch": 3.2083643939828352,
-      "grad_norm": 0.004438692703843117,
-      "learning_rate": 0.00019999492276692258,
-      "loss": 46.0,
-      "step": 41963
-    },
-    {
-      "epoch": 3.208440850966225,
-      "grad_norm": 0.0012899021385237575,
-      "learning_rate": 0.00019999492252487788,
-      "loss": 46.0,
-      "step": 41964
-    },
-    {
-      "epoch": 3.2085173079496148,
-      "grad_norm": 0.002076528500765562,
-      "learning_rate": 0.00019999492228282742,
-      "loss": 46.0,
-      "step": 41965
-    },
-    {
-      "epoch": 3.2085937649330045,
-      "grad_norm": 0.0013169677695259452,
-      "learning_rate": 0.00019999492204077118,
-      "loss": 46.0,
-      "step": 41966
-    },
-    {
-      "epoch": 3.2086702219163943,
-      "grad_norm": 0.001325442804954946,
-      "learning_rate": 0.00019999492179870916,
-      "loss": 46.0,
-      "step": 41967
-    },
-    {
-      "epoch": 3.208746678899784,
-      "grad_norm": 0.0036222960334271193,
-      "learning_rate": 0.0001999949215566414,
-      "loss": 46.0,
-      "step": 41968
-    },
-    {
-      "epoch": 3.208823135883174,
-      "grad_norm": 0.005300433840602636,
-      "learning_rate": 0.00019999492131456785,
-      "loss": 46.0,
-      "step": 41969
-    },
-    {
-      "epoch": 3.2088995928665636,
-      "grad_norm": 0.0016628399025648832,
-      "learning_rate": 0.00019999492107248854,
-      "loss": 46.0,
-      "step": 41970
-    },
-    {
-      "epoch": 3.2089760498499533,
-      "grad_norm": 0.002210634760558605,
-      "learning_rate": 0.00019999492083040346,
-      "loss": 46.0,
-      "step": 41971
-    },
-    {
-      "epoch": 3.209052506833343,
-      "grad_norm": 0.000432817789260298,
-      "learning_rate": 0.0001999949205883126,
-      "loss": 46.0,
-      "step": 41972
-    },
-    {
-      "epoch": 3.2091289638167324,
-      "grad_norm": 0.0031472027767449617,
-      "learning_rate": 0.00019999492034621598,
-      "loss": 46.0,
-      "step": 41973
-    },
-    {
-      "epoch": 3.209205420800122,
-      "grad_norm": 0.004076885990798473,
-      "learning_rate": 0.00019999492010411358,
-      "loss": 46.0,
-      "step": 41974
-    },
-    {
-      "epoch": 3.209281877783512,
-      "grad_norm": 0.003162642475217581,
-      "learning_rate": 0.00019999491986200544,
-      "loss": 46.0,
-      "step": 41975
-    },
-    {
-      "epoch": 3.2093583347669017,
-      "grad_norm": 0.0028325412422418594,
-      "learning_rate": 0.0001999949196198915,
-      "loss": 46.0,
-      "step": 41976
-    },
-    {
-      "epoch": 3.2094347917502914,
-      "grad_norm": 0.0015966629143804312,
-      "learning_rate": 0.0001999949193777718,
-      "loss": 46.0,
-      "step": 41977
-    },
-    {
-      "epoch": 3.209511248733681,
-      "grad_norm": 0.0024066753685474396,
-      "learning_rate": 0.00019999491913564633,
-      "loss": 46.0,
-      "step": 41978
-    },
-    {
-      "epoch": 3.209587705717071,
-      "grad_norm": 0.00470546493306756,
-      "learning_rate": 0.00019999491889351512,
-      "loss": 46.0,
-      "step": 41979
-    },
-    {
-      "epoch": 3.2096641627004607,
-      "grad_norm": 0.002852489473298192,
-      "learning_rate": 0.00019999491865137808,
-      "loss": 46.0,
-      "step": 41980
-    },
-    {
-      "epoch": 3.2097406196838505,
-      "grad_norm": 0.00869984831660986,
-      "learning_rate": 0.00019999491840923532,
-      "loss": 46.0,
-      "step": 41981
-    },
-    {
-      "epoch": 3.2098170766672403,
-      "grad_norm": 0.002468184567987919,
-      "learning_rate": 0.00019999491816708676,
-      "loss": 46.0,
-      "step": 41982
-    },
-    {
-      "epoch": 3.20989353365063,
-      "grad_norm": 0.00302159134298563,
-      "learning_rate": 0.00019999491792493249,
-      "loss": 46.0,
-      "step": 41983
-    },
-    {
-      "epoch": 3.2099699906340193,
-      "grad_norm": 0.002297553699463606,
-      "learning_rate": 0.00019999491768277238,
-      "loss": 46.0,
-      "step": 41984
-    },
-    {
-      "epoch": 3.210046447617409,
-      "grad_norm": 0.0016759030986577272,
-      "learning_rate": 0.00019999491744060653,
-      "loss": 46.0,
-      "step": 41985
-    },
-    {
-      "epoch": 3.210122904600799,
-      "grad_norm": 0.0018621556228026748,
-      "learning_rate": 0.0001999949171984349,
-      "loss": 46.0,
-      "step": 41986
-    },
-    {
-      "epoch": 3.2101993615841886,
-      "grad_norm": 0.002524434356018901,
-      "learning_rate": 0.00019999491695625753,
-      "loss": 46.0,
-      "step": 41987
-    },
-    {
-      "epoch": 3.2102758185675784,
-      "grad_norm": 0.006962368730455637,
-      "learning_rate": 0.00019999491671407436,
-      "loss": 46.0,
-      "step": 41988
-    },
-    {
-      "epoch": 3.210352275550968,
-      "grad_norm": 0.0016402521869167686,
-      "learning_rate": 0.00019999491647188544,
-      "loss": 46.0,
-      "step": 41989
-    },
-    {
-      "epoch": 3.210428732534358,
-      "grad_norm": 0.0011914786882698536,
-      "learning_rate": 0.00019999491622969075,
-      "loss": 46.0,
-      "step": 41990
-    },
-    {
-      "epoch": 3.2105051895177477,
-      "grad_norm": 0.001277404255233705,
-      "learning_rate": 0.00019999491598749028,
-      "loss": 46.0,
-      "step": 41991
-    },
-    {
-      "epoch": 3.2105816465011374,
-      "grad_norm": 0.007648926228284836,
-      "learning_rate": 0.00019999491574528404,
-      "loss": 46.0,
-      "step": 41992
-    },
-    {
-      "epoch": 3.210658103484527,
-      "grad_norm": 0.003468940267339349,
-      "learning_rate": 0.00019999491550307203,
-      "loss": 46.0,
-      "step": 41993
-    },
-    {
-      "epoch": 3.210734560467917,
-      "grad_norm": 0.002637440338730812,
-      "learning_rate": 0.00019999491526085427,
-      "loss": 46.0,
-      "step": 41994
-    },
-    {
-      "epoch": 3.2108110174513063,
-      "grad_norm": 0.006943566258996725,
-      "learning_rate": 0.0001999949150186307,
-      "loss": 46.0,
-      "step": 41995
-    },
-    {
-      "epoch": 3.210887474434696,
-      "grad_norm": 0.0016487700631842017,
-      "learning_rate": 0.0001999949147764014,
-      "loss": 46.0,
-      "step": 41996
-    },
-    {
-      "epoch": 3.210963931418086,
-      "grad_norm": 0.0009402178111486137,
-      "learning_rate": 0.00019999491453416633,
-      "loss": 46.0,
-      "step": 41997
-    },
-    {
-      "epoch": 3.2110403884014755,
-      "grad_norm": 0.0019277287647128105,
-      "learning_rate": 0.00019999491429192548,
-      "loss": 46.0,
-      "step": 41998
-    },
-    {
-      "epoch": 3.2111168453848653,
-      "grad_norm": 0.0005230432725511491,
-      "learning_rate": 0.00019999491404967885,
-      "loss": 46.0,
-      "step": 41999
-    },
-    {
-      "epoch": 3.211193302368255,
-      "grad_norm": 0.004211823921650648,
-      "learning_rate": 0.00019999491380742645,
-      "loss": 46.0,
-      "step": 42000
-    },
-    {
-      "epoch": 3.211269759351645,
-      "grad_norm": 0.0016098804771900177,
-      "learning_rate": 0.00019999491356516834,
-      "loss": 46.0,
-      "step": 42001
-    },
-    {
-      "epoch": 3.2113462163350346,
-      "grad_norm": 0.001238106982782483,
-      "learning_rate": 0.0001999949133229044,
-      "loss": 46.0,
-      "step": 42002
-    },
-    {
-      "epoch": 3.2114226733184243,
-      "grad_norm": 0.0031190484296530485,
-      "learning_rate": 0.0001999949130806347,
-      "loss": 46.0,
-      "step": 42003
-    },
-    {
-      "epoch": 3.211499130301814,
-      "grad_norm": 0.0017370606074109674,
-      "learning_rate": 0.00019999491283835923,
-      "loss": 46.0,
-      "step": 42004
-    },
-    {
-      "epoch": 3.211575587285204,
-      "grad_norm": 0.003475408535450697,
-      "learning_rate": 0.000199994912596078,
-      "loss": 46.0,
-      "step": 42005
-    },
-    {
-      "epoch": 3.211652044268593,
-      "grad_norm": 0.0019139007199555635,
-      "learning_rate": 0.00019999491235379098,
-      "loss": 46.0,
-      "step": 42006
-    },
-    {
-      "epoch": 3.211728501251983,
-      "grad_norm": 0.0021670942660421133,
-      "learning_rate": 0.00019999491211149822,
-      "loss": 46.0,
-      "step": 42007
-    },
-    {
-      "epoch": 3.2118049582353727,
-      "grad_norm": 0.002825238974764943,
-      "learning_rate": 0.00019999491186919967,
-      "loss": 46.0,
-      "step": 42008
-    },
-    {
-      "epoch": 3.2118814152187625,
-      "grad_norm": 0.003197198733687401,
-      "learning_rate": 0.00019999491162689536,
-      "loss": 46.0,
-      "step": 42009
-    },
-    {
-      "epoch": 3.2119578722021522,
-      "grad_norm": 0.00154455890879035,
-      "learning_rate": 0.00019999491138458528,
-      "loss": 46.0,
-      "step": 42010
-    },
-    {
-      "epoch": 3.212034329185542,
-      "grad_norm": 0.001795478630810976,
-      "learning_rate": 0.00019999491114226943,
-      "loss": 46.0,
-      "step": 42011
-    },
-    {
-      "epoch": 3.2121107861689318,
-      "grad_norm": 0.0012015431420877576,
-      "learning_rate": 0.0001999949108999478,
-      "loss": 46.0,
-      "step": 42012
-    },
-    {
-      "epoch": 3.2121872431523215,
-      "grad_norm": 0.0034283860586583614,
-      "learning_rate": 0.00019999491065762043,
-      "loss": 46.0,
-      "step": 42013
-    },
-    {
-      "epoch": 3.2122637001357113,
-      "grad_norm": 0.0019157460192218423,
-      "learning_rate": 0.00019999491041528726,
-      "loss": 46.0,
-      "step": 42014
-    },
-    {
-      "epoch": 3.212340157119101,
-      "grad_norm": 0.0028663913253694773,
-      "learning_rate": 0.00019999491017294835,
-      "loss": 46.0,
-      "step": 42015
-    },
-    {
-      "epoch": 3.212416614102491,
-      "grad_norm": 0.0010831519030034542,
-      "learning_rate": 0.00019999490993060363,
-      "loss": 46.0,
-      "step": 42016
-    },
-    {
-      "epoch": 3.21249307108588,
-      "grad_norm": 0.0030344510450959206,
-      "learning_rate": 0.0001999949096882532,
-      "loss": 46.0,
-      "step": 42017
-    },
-    {
-      "epoch": 3.21256952806927,
-      "grad_norm": 0.001425539841875434,
-      "learning_rate": 0.00019999490944589695,
-      "loss": 46.0,
-      "step": 42018
-    },
-    {
-      "epoch": 3.2126459850526596,
-      "grad_norm": 0.0021982023026794195,
-      "learning_rate": 0.00019999490920353497,
-      "loss": 46.0,
-      "step": 42019
-    },
-    {
-      "epoch": 3.2127224420360494,
-      "grad_norm": 0.004483222961425781,
-      "learning_rate": 0.00019999490896116718,
-      "loss": 46.0,
-      "step": 42020
-    },
-    {
-      "epoch": 3.212798899019439,
-      "grad_norm": 0.0022205663844943047,
-      "learning_rate": 0.00019999490871879365,
-      "loss": 46.0,
-      "step": 42021
-    },
-    {
-      "epoch": 3.212875356002829,
-      "grad_norm": 0.0014647021889686584,
-      "learning_rate": 0.00019999490847641435,
-      "loss": 46.0,
-      "step": 42022
-    },
-    {
-      "epoch": 3.2129518129862187,
-      "grad_norm": 0.0020561611745506525,
-      "learning_rate": 0.00019999490823402924,
-      "loss": 46.0,
-      "step": 42023
-    },
-    {
-      "epoch": 3.2130282699696084,
-      "grad_norm": 0.003461114363744855,
-      "learning_rate": 0.0001999949079916384,
-      "loss": 46.0,
-      "step": 42024
-    },
-    {
-      "epoch": 3.213104726952998,
-      "grad_norm": 0.0027100222650915384,
-      "learning_rate": 0.0001999949077492418,
-      "loss": 46.0,
-      "step": 42025
-    },
-    {
-      "epoch": 3.213181183936388,
-      "grad_norm": 0.0026494513731449842,
-      "learning_rate": 0.00019999490750683937,
-      "loss": 46.0,
-      "step": 42026
-    },
-    {
-      "epoch": 3.2132576409197773,
-      "grad_norm": 0.0020694297272711992,
-      "learning_rate": 0.00019999490726443125,
-      "loss": 46.0,
-      "step": 42027
-    },
-    {
-      "epoch": 3.213334097903167,
-      "grad_norm": 0.0016173539916053414,
-      "learning_rate": 0.00019999490702201733,
-      "loss": 46.0,
-      "step": 42028
-    },
-    {
-      "epoch": 3.213410554886557,
-      "grad_norm": 0.0017572048818692565,
-      "learning_rate": 0.0001999949067795976,
-      "loss": 46.0,
-      "step": 42029
-    },
-    {
-      "epoch": 3.2134870118699466,
-      "grad_norm": 0.0018634614534676075,
-      "learning_rate": 0.00019999490653717215,
-      "loss": 46.0,
-      "step": 42030
-    },
-    {
-      "epoch": 3.2135634688533363,
-      "grad_norm": 0.001311220694333315,
-      "learning_rate": 0.0001999949062947409,
-      "loss": 46.0,
-      "step": 42031
-    },
-    {
-      "epoch": 3.213639925836726,
-      "grad_norm": 0.0012478322023525834,
-      "learning_rate": 0.0001999949060523039,
-      "loss": 46.0,
-      "step": 42032
-    },
-    {
-      "epoch": 3.213716382820116,
-      "grad_norm": 0.0024957098066806793,
-      "learning_rate": 0.00019999490580986114,
-      "loss": 46.0,
-      "step": 42033
-    },
-    {
-      "epoch": 3.2137928398035056,
-      "grad_norm": 0.0017277593724429607,
-      "learning_rate": 0.00019999490556741264,
-      "loss": 46.0,
-      "step": 42034
-    },
-    {
-      "epoch": 3.2138692967868954,
-      "grad_norm": 0.0016810082597658038,
-      "learning_rate": 0.0001999949053249583,
-      "loss": 46.0,
-      "step": 42035
-    },
-    {
-      "epoch": 3.213945753770285,
-      "grad_norm": 0.004137719515711069,
-      "learning_rate": 0.00019999490508249823,
-      "loss": 46.0,
-      "step": 42036
-    },
-    {
-      "epoch": 3.214022210753675,
-      "grad_norm": 0.0041664703749120235,
-      "learning_rate": 0.00019999490484003238,
-      "loss": 46.0,
-      "step": 42037
-    },
-    {
-      "epoch": 3.2140986677370647,
-      "grad_norm": 0.003994942642748356,
-      "learning_rate": 0.00019999490459756075,
-      "loss": 46.0,
-      "step": 42038
-    },
-    {
-      "epoch": 3.214175124720454,
-      "grad_norm": 0.003370902733877301,
-      "learning_rate": 0.00019999490435508336,
-      "loss": 46.0,
-      "step": 42039
-    },
-    {
-      "epoch": 3.2142515817038437,
-      "grad_norm": 0.0010797909926623106,
-      "learning_rate": 0.0001999949041126002,
-      "loss": 46.0,
-      "step": 42040
-    },
-    {
-      "epoch": 3.2143280386872335,
-      "grad_norm": 0.0027798833325505257,
-      "learning_rate": 0.0001999949038701113,
-      "loss": 46.0,
-      "step": 42041
-    },
-    {
-      "epoch": 3.2144044956706233,
-      "grad_norm": 0.0007769897929392755,
-      "learning_rate": 0.0001999949036276166,
-      "loss": 46.0,
-      "step": 42042
-    },
-    {
-      "epoch": 3.214480952654013,
-      "grad_norm": 0.0022532104048877954,
-      "learning_rate": 0.00019999490338511614,
-      "loss": 46.0,
-      "step": 42043
-    },
-    {
-      "epoch": 3.2145574096374028,
-      "grad_norm": 0.0012627977412194014,
-      "learning_rate": 0.0001999949031426099,
-      "loss": 46.0,
-      "step": 42044
-    },
-    {
-      "epoch": 3.2146338666207925,
-      "grad_norm": 0.006925684399902821,
-      "learning_rate": 0.0001999949029000979,
-      "loss": 46.0,
-      "step": 42045
-    },
-    {
-      "epoch": 3.2147103236041823,
-      "grad_norm": 0.0019622284453362226,
-      "learning_rate": 0.00019999490265758013,
-      "loss": 46.0,
-      "step": 42046
-    },
-    {
-      "epoch": 3.214786780587572,
-      "grad_norm": 0.003863739548251033,
-      "learning_rate": 0.0001999949024150566,
-      "loss": 46.0,
-      "step": 42047
-    },
-    {
-      "epoch": 3.214863237570962,
-      "grad_norm": 0.002014464931562543,
-      "learning_rate": 0.00019999490217252727,
-      "loss": 46.0,
-      "step": 42048
-    },
-    {
-      "epoch": 3.214939694554351,
-      "grad_norm": 0.002009489806368947,
-      "learning_rate": 0.00019999490192999222,
-      "loss": 46.0,
-      "step": 42049
-    },
-    {
-      "epoch": 3.215016151537741,
-      "grad_norm": 0.0006463376339524984,
-      "learning_rate": 0.00019999490168745137,
-      "loss": 46.0,
-      "step": 42050
-    },
-    {
-      "epoch": 3.2150926085211307,
-      "grad_norm": 0.003279470605775714,
-      "learning_rate": 0.00019999490144490475,
-      "loss": 46.0,
-      "step": 42051
-    },
-    {
-      "epoch": 3.2151690655045204,
-      "grad_norm": 0.0037217820063233376,
-      "learning_rate": 0.00019999490120235235,
-      "loss": 46.0,
-      "step": 42052
-    },
-    {
-      "epoch": 3.21524552248791,
-      "grad_norm": 0.0033905671443790197,
-      "learning_rate": 0.0001999949009597942,
-      "loss": 46.0,
-      "step": 42053
-    },
-    {
-      "epoch": 3.2153219794713,
-      "grad_norm": 0.0011346264509484172,
-      "learning_rate": 0.0001999949007172303,
-      "loss": 46.0,
-      "step": 42054
-    },
-    {
-      "epoch": 3.2153984364546897,
-      "grad_norm": 0.003642667317762971,
-      "learning_rate": 0.00019999490047466057,
-      "loss": 46.0,
-      "step": 42055
-    },
-    {
-      "epoch": 3.2154748934380795,
-      "grad_norm": 0.0014360713539645076,
-      "learning_rate": 0.00019999490023208514,
-      "loss": 46.0,
-      "step": 42056
-    },
-    {
-      "epoch": 3.2155513504214692,
-      "grad_norm": 0.0076574720442295074,
-      "learning_rate": 0.00019999489998950393,
-      "loss": 46.0,
-      "step": 42057
-    },
-    {
-      "epoch": 3.215627807404859,
-      "grad_norm": 0.0028948960825800896,
-      "learning_rate": 0.0001999948997469169,
-      "loss": 46.0,
-      "step": 42058
-    },
-    {
-      "epoch": 3.2157042643882487,
-      "grad_norm": 0.0017489342717453837,
-      "learning_rate": 0.00019999489950432416,
-      "loss": 46.0,
-      "step": 42059
-    },
-    {
-      "epoch": 3.2157807213716385,
-      "grad_norm": 0.010017432272434235,
-      "learning_rate": 0.0001999948992617256,
-      "loss": 46.0,
-      "step": 42060
-    },
-    {
-      "epoch": 3.215857178355028,
-      "grad_norm": 0.0032760468311607838,
-      "learning_rate": 0.0001999948990191213,
-      "loss": 46.0,
-      "step": 42061
-    },
-    {
-      "epoch": 3.2159336353384176,
-      "grad_norm": 0.002923940308392048,
-      "learning_rate": 0.00019999489877651122,
-      "loss": 46.0,
-      "step": 42062
-    },
-    {
-      "epoch": 3.2160100923218073,
-      "grad_norm": 0.0027671444695442915,
-      "learning_rate": 0.00019999489853389537,
-      "loss": 46.0,
-      "step": 42063
-    },
-    {
-      "epoch": 3.216086549305197,
-      "grad_norm": 0.0015004008309915662,
-      "learning_rate": 0.00019999489829127375,
-      "loss": 46.0,
-      "step": 42064
-    },
-    {
-      "epoch": 3.216163006288587,
-      "grad_norm": 0.005195599514991045,
-      "learning_rate": 0.00019999489804864638,
-      "loss": 46.0,
-      "step": 42065
-    },
-    {
-      "epoch": 3.2162394632719766,
-      "grad_norm": 0.0031378748826682568,
-      "learning_rate": 0.0001999948978060132,
-      "loss": 46.0,
-      "step": 42066
-    },
-    {
-      "epoch": 3.2163159202553664,
-      "grad_norm": 0.0006268033175729215,
-      "learning_rate": 0.00019999489756337432,
-      "loss": 46.0,
-      "step": 42067
-    },
-    {
-      "epoch": 3.216392377238756,
-      "grad_norm": 0.0023044419940561056,
-      "learning_rate": 0.0001999948973207296,
-      "loss": 46.0,
-      "step": 42068
-    },
-    {
-      "epoch": 3.216468834222146,
-      "grad_norm": 0.0016772350063547492,
-      "learning_rate": 0.00019999489707807913,
-      "loss": 46.0,
-      "step": 42069
-    },
-    {
-      "epoch": 3.2165452912055357,
-      "grad_norm": 0.0022749791387468576,
-      "learning_rate": 0.00019999489683542292,
-      "loss": 46.0,
-      "step": 42070
-    },
-    {
-      "epoch": 3.216621748188925,
-      "grad_norm": 0.0014040415408089757,
-      "learning_rate": 0.00019999489659276091,
-      "loss": 46.0,
-      "step": 42071
-    },
-    {
-      "epoch": 3.2166982051723148,
-      "grad_norm": 0.0021414763759821653,
-      "learning_rate": 0.00019999489635009313,
-      "loss": 46.0,
-      "step": 42072
-    },
-    {
-      "epoch": 3.2167746621557045,
-      "grad_norm": 0.0015099617885425687,
-      "learning_rate": 0.00019999489610741963,
-      "loss": 46.0,
-      "step": 42073
-    },
-    {
-      "epoch": 3.2168511191390943,
-      "grad_norm": 0.004244043957442045,
-      "learning_rate": 0.00019999489586474032,
-      "loss": 46.0,
-      "step": 42074
-    },
-    {
-      "epoch": 3.216927576122484,
-      "grad_norm": 0.0015712818130850792,
-      "learning_rate": 0.00019999489562205525,
-      "loss": 46.0,
-      "step": 42075
-    },
-    {
-      "epoch": 3.217004033105874,
-      "grad_norm": 0.0007181677501648664,
-      "learning_rate": 0.0001999948953793644,
-      "loss": 46.0,
-      "step": 42076
-    },
-    {
-      "epoch": 3.2170804900892636,
-      "grad_norm": 0.0011897190706804395,
-      "learning_rate": 0.00019999489513666777,
-      "loss": 46.0,
-      "step": 42077
-    },
-    {
-      "epoch": 3.2171569470726533,
-      "grad_norm": 0.0009469385840930045,
-      "learning_rate": 0.00019999489489396538,
-      "loss": 46.0,
-      "step": 42078
-    },
-    {
-      "epoch": 3.217233404056043,
-      "grad_norm": 0.0007228367030620575,
-      "learning_rate": 0.00019999489465125723,
-      "loss": 46.0,
-      "step": 42079
-    },
-    {
-      "epoch": 3.217309861039433,
-      "grad_norm": 0.004239616449922323,
-      "learning_rate": 0.00019999489440854332,
-      "loss": 46.0,
-      "step": 42080
-    },
-    {
-      "epoch": 3.2173863180228226,
-      "grad_norm": 0.010959330946207047,
-      "learning_rate": 0.00019999489416582363,
-      "loss": 46.0,
-      "step": 42081
-    },
-    {
-      "epoch": 3.2174627750062124,
-      "grad_norm": 0.004945164080709219,
-      "learning_rate": 0.00019999489392309817,
-      "loss": 46.0,
-      "step": 42082
-    },
-    {
-      "epoch": 3.2175392319896017,
-      "grad_norm": 0.002401743782684207,
-      "learning_rate": 0.00019999489368036693,
-      "loss": 46.0,
-      "step": 42083
-    },
-    {
-      "epoch": 3.2176156889729914,
-      "grad_norm": 0.0013879043981432915,
-      "learning_rate": 0.00019999489343762995,
-      "loss": 46.0,
-      "step": 42084
-    },
-    {
-      "epoch": 3.217692145956381,
-      "grad_norm": 0.00033975523547269404,
-      "learning_rate": 0.0001999948931948872,
-      "loss": 46.0,
-      "step": 42085
-    },
-    {
-      "epoch": 3.217768602939771,
-      "grad_norm": 0.0031126264948397875,
-      "learning_rate": 0.00019999489295213863,
-      "loss": 46.0,
-      "step": 42086
-    },
-    {
-      "epoch": 3.2178450599231607,
-      "grad_norm": 0.0019977290648967028,
-      "learning_rate": 0.00019999489270938433,
-      "loss": 46.0,
-      "step": 42087
-    },
-    {
-      "epoch": 3.2179215169065505,
-      "grad_norm": 0.004049096256494522,
-      "learning_rate": 0.00019999489246662425,
-      "loss": 46.0,
-      "step": 42088
-    },
-    {
-      "epoch": 3.2179979738899402,
-      "grad_norm": 0.005429015029221773,
-      "learning_rate": 0.0001999948922238584,
-      "loss": 46.0,
-      "step": 42089
-    },
-    {
-      "epoch": 3.21807443087333,
-      "grad_norm": 0.0015906282933428884,
-      "learning_rate": 0.00019999489198108678,
-      "loss": 46.0,
-      "step": 42090
-    },
-    {
-      "epoch": 3.2181508878567198,
-      "grad_norm": 0.004229491110891104,
-      "learning_rate": 0.00019999489173830944,
-      "loss": 46.0,
-      "step": 42091
-    },
-    {
-      "epoch": 3.2182273448401095,
-      "grad_norm": 0.0015959986485540867,
-      "learning_rate": 0.00019999489149552627,
-      "loss": 46.0,
-      "step": 42092
-    },
-    {
-      "epoch": 3.218303801823499,
-      "grad_norm": 0.0024031796492636204,
-      "learning_rate": 0.00019999489125273735,
-      "loss": 46.0,
-      "step": 42093
-    },
-    {
-      "epoch": 3.2183802588068886,
-      "grad_norm": 0.0017796535976231098,
-      "learning_rate": 0.00019999489100994264,
-      "loss": 46.0,
-      "step": 42094
-    },
-    {
-      "epoch": 3.2184567157902784,
-      "grad_norm": 0.002239165361970663,
-      "learning_rate": 0.0001999948907671422,
-      "loss": 46.0,
-      "step": 42095
-    },
-    {
-      "epoch": 3.218533172773668,
-      "grad_norm": 0.004304191097617149,
-      "learning_rate": 0.00019999489052433596,
-      "loss": 46.0,
-      "step": 42096
-    },
-    {
-      "epoch": 3.218609629757058,
-      "grad_norm": 0.00257976446300745,
-      "learning_rate": 0.00019999489028152398,
-      "loss": 46.0,
-      "step": 42097
-    },
-    {
-      "epoch": 3.2186860867404476,
-      "grad_norm": 0.0016116040060296655,
-      "learning_rate": 0.00019999489003870623,
-      "loss": 46.0,
-      "step": 42098
-    },
-    {
-      "epoch": 3.2187625437238374,
-      "grad_norm": 0.004151143599301577,
-      "learning_rate": 0.00019999488979588267,
-      "loss": 46.0,
-      "step": 42099
-    },
-    {
-      "epoch": 3.218839000707227,
-      "grad_norm": 0.001729447627440095,
-      "learning_rate": 0.00019999488955305337,
-      "loss": 46.0,
-      "step": 42100
-    },
-    {
-      "epoch": 3.218915457690617,
-      "grad_norm": 0.0008196085109375417,
-      "learning_rate": 0.0001999948893102183,
-      "loss": 46.0,
-      "step": 42101
-    },
-    {
-      "epoch": 3.2189919146740067,
-      "grad_norm": 0.001459177932702005,
-      "learning_rate": 0.00019999488906737744,
-      "loss": 46.0,
-      "step": 42102
-    },
-    {
-      "epoch": 3.2190683716573965,
-      "grad_norm": 0.005648829508572817,
-      "learning_rate": 0.00019999488882453085,
-      "loss": 46.0,
-      "step": 42103
-    },
-    {
-      "epoch": 3.2191448286407858,
-      "grad_norm": 0.0017308726673945785,
-      "learning_rate": 0.00019999488858167845,
-      "loss": 46.0,
-      "step": 42104
-    },
-    {
-      "epoch": 3.2192212856241755,
-      "grad_norm": 0.002907112240791321,
-      "learning_rate": 0.0001999948883388203,
-      "loss": 46.0,
-      "step": 42105
-    },
-    {
-      "epoch": 3.2192977426075653,
-      "grad_norm": 0.001044327742420137,
-      "learning_rate": 0.0001999948880959564,
-      "loss": 46.0,
-      "step": 42106
-    },
-    {
-      "epoch": 3.219374199590955,
-      "grad_norm": 0.0035064637195318937,
-      "learning_rate": 0.0001999948878530867,
-      "loss": 46.0,
-      "step": 42107
-    },
-    {
-      "epoch": 3.219450656574345,
-      "grad_norm": 0.004620145540684462,
-      "learning_rate": 0.00019999488761021124,
-      "loss": 46.0,
-      "step": 42108
-    },
-    {
-      "epoch": 3.2195271135577346,
-      "grad_norm": 0.001909983460791409,
-      "learning_rate": 0.00019999488736733,
-      "loss": 46.0,
-      "step": 42109
-    },
-    {
-      "epoch": 3.2196035705411243,
-      "grad_norm": 0.0024475122336298227,
-      "learning_rate": 0.00019999488712444302,
-      "loss": 46.0,
-      "step": 42110
-    },
-    {
-      "epoch": 3.219680027524514,
-      "grad_norm": 0.006110894028097391,
-      "learning_rate": 0.00019999488688155027,
-      "loss": 46.0,
-      "step": 42111
-    },
-    {
-      "epoch": 3.219756484507904,
-      "grad_norm": 0.0010518016060814261,
-      "learning_rate": 0.0001999948866386517,
-      "loss": 46.0,
-      "step": 42112
-    },
-    {
-      "epoch": 3.2198329414912936,
-      "grad_norm": 0.006876754574477673,
-      "learning_rate": 0.0001999948863957474,
-      "loss": 46.0,
-      "step": 42113
-    },
-    {
-      "epoch": 3.2199093984746834,
-      "grad_norm": 0.0023621413856744766,
-      "learning_rate": 0.00019999488615283736,
-      "loss": 46.0,
-      "step": 42114
-    },
-    {
-      "epoch": 3.2199858554580727,
-      "grad_norm": 0.0012257179478183389,
-      "learning_rate": 0.0001999948859099215,
-      "loss": 46.0,
-      "step": 42115
-    },
-    {
-      "epoch": 3.2200623124414625,
-      "grad_norm": 0.001808013767004013,
-      "learning_rate": 0.00019999488566699989,
-      "loss": 46.0,
-      "step": 42116
-    },
-    {
-      "epoch": 3.220138769424852,
-      "grad_norm": 0.0014598038978874683,
-      "learning_rate": 0.00019999488542407252,
-      "loss": 46.0,
-      "step": 42117
-    },
-    {
-      "epoch": 3.220215226408242,
-      "grad_norm": 0.006018663756549358,
-      "learning_rate": 0.00019999488518113935,
-      "loss": 46.0,
-      "step": 42118
-    },
-    {
-      "epoch": 3.2202916833916317,
-      "grad_norm": 0.006880896631628275,
-      "learning_rate": 0.00019999488493820043,
-      "loss": 46.0,
-      "step": 42119
-    },
-    {
-      "epoch": 3.2203681403750215,
-      "grad_norm": 0.002467251615598798,
-      "learning_rate": 0.00019999488469525577,
-      "loss": 46.0,
-      "step": 42120
-    },
-    {
-      "epoch": 3.2204445973584113,
-      "grad_norm": 0.002496525412425399,
-      "learning_rate": 0.0001999948844523053,
-      "loss": 46.0,
-      "step": 42121
-    },
-    {
-      "epoch": 3.220521054341801,
-      "grad_norm": 0.0009444904862903059,
-      "learning_rate": 0.00019999488420934907,
-      "loss": 46.0,
-      "step": 42122
-    },
-    {
-      "epoch": 3.220597511325191,
-      "grad_norm": 0.0016867815284058452,
-      "learning_rate": 0.0001999948839663871,
-      "loss": 46.0,
-      "step": 42123
-    },
-    {
-      "epoch": 3.2206739683085805,
-      "grad_norm": 0.0012590765254572034,
-      "learning_rate": 0.0001999948837234193,
-      "loss": 46.0,
-      "step": 42124
-    },
-    {
-      "epoch": 3.2207504252919703,
-      "grad_norm": 0.0014494657516479492,
-      "learning_rate": 0.00019999488348044578,
-      "loss": 46.0,
-      "step": 42125
-    },
-    {
-      "epoch": 3.2208268822753596,
-      "grad_norm": 0.001622830401174724,
-      "learning_rate": 0.00019999488323746648,
-      "loss": 46.0,
-      "step": 42126
-    },
-    {
-      "epoch": 3.2209033392587494,
-      "grad_norm": 0.0021084214095026255,
-      "learning_rate": 0.0001999948829944814,
-      "loss": 46.0,
-      "step": 42127
-    },
-    {
-      "epoch": 3.220979796242139,
-      "grad_norm": 0.001371025457046926,
-      "learning_rate": 0.00019999488275149055,
-      "loss": 46.0,
-      "step": 42128
-    },
-    {
-      "epoch": 3.221056253225529,
-      "grad_norm": 0.002170608611777425,
-      "learning_rate": 0.00019999488250849393,
-      "loss": 46.0,
-      "step": 42129
-    },
-    {
-      "epoch": 3.2211327102089187,
-      "grad_norm": 0.0008423345279879868,
-      "learning_rate": 0.00019999488226549156,
-      "loss": 46.0,
-      "step": 42130
-    },
-    {
-      "epoch": 3.2212091671923084,
-      "grad_norm": 0.0011878538643941283,
-      "learning_rate": 0.00019999488202248342,
-      "loss": 46.0,
-      "step": 42131
-    },
-    {
-      "epoch": 3.221285624175698,
-      "grad_norm": 0.002165592275559902,
-      "learning_rate": 0.0001999948817794695,
-      "loss": 46.0,
-      "step": 42132
-    },
-    {
-      "epoch": 3.221362081159088,
-      "grad_norm": 0.0011837874772027135,
-      "learning_rate": 0.00019999488153644982,
-      "loss": 46.0,
-      "step": 42133
-    },
-    {
-      "epoch": 3.2214385381424777,
-      "grad_norm": 0.002083104569464922,
-      "learning_rate": 0.00019999488129342436,
-      "loss": 46.0,
-      "step": 42134
-    },
-    {
-      "epoch": 3.2215149951258675,
-      "grad_norm": 0.00433348910883069,
-      "learning_rate": 0.00019999488105039315,
-      "loss": 46.0,
-      "step": 42135
-    },
-    {
-      "epoch": 3.2215914521092572,
-      "grad_norm": 0.001558369374834001,
-      "learning_rate": 0.00019999488080735614,
-      "loss": 46.0,
-      "step": 42136
-    },
-    {
-      "epoch": 3.2216679090926466,
-      "grad_norm": 0.0026920493692159653,
-      "learning_rate": 0.00019999488056431338,
-      "loss": 46.0,
-      "step": 42137
-    },
-    {
-      "epoch": 3.2217443660760363,
-      "grad_norm": 0.0011284475913271308,
-      "learning_rate": 0.00019999488032126483,
-      "loss": 46.0,
-      "step": 42138
-    },
-    {
-      "epoch": 3.221820823059426,
-      "grad_norm": 0.004418435040861368,
-      "learning_rate": 0.00019999488007821055,
-      "loss": 46.0,
-      "step": 42139
-    },
-    {
-      "epoch": 3.221897280042816,
-      "grad_norm": 0.004002754110842943,
-      "learning_rate": 0.00019999487983515048,
-      "loss": 46.0,
-      "step": 42140
-    },
-    {
-      "epoch": 3.2219737370262056,
-      "grad_norm": 0.007967714220285416,
-      "learning_rate": 0.00019999487959208466,
-      "loss": 46.0,
-      "step": 42141
-    },
-    {
-      "epoch": 3.2220501940095954,
-      "grad_norm": 0.0019889555405825377,
-      "learning_rate": 0.00019999487934901304,
-      "loss": 46.0,
-      "step": 42142
-    },
-    {
-      "epoch": 3.222126650992985,
-      "grad_norm": 0.0027892852667719126,
-      "learning_rate": 0.00019999487910593567,
-      "loss": 46.0,
-      "step": 42143
-    },
-    {
-      "epoch": 3.222203107976375,
-      "grad_norm": 0.004028509370982647,
-      "learning_rate": 0.0001999948788628525,
-      "loss": 46.0,
-      "step": 42144
-    },
-    {
-      "epoch": 3.2222795649597646,
-      "grad_norm": 0.0024777331855148077,
-      "learning_rate": 0.00019999487861976358,
-      "loss": 46.0,
-      "step": 42145
-    },
-    {
-      "epoch": 3.2223560219431544,
-      "grad_norm": 0.0018562150653451681,
-      "learning_rate": 0.0001999948783766689,
-      "loss": 46.0,
-      "step": 42146
-    },
-    {
-      "epoch": 3.222432478926544,
-      "grad_norm": 0.0015134194400161505,
-      "learning_rate": 0.00019999487813356846,
-      "loss": 46.0,
-      "step": 42147
-    },
-    {
-      "epoch": 3.2225089359099335,
-      "grad_norm": 0.002831601770594716,
-      "learning_rate": 0.0001999948778904622,
-      "loss": 46.0,
-      "step": 42148
-    },
-    {
-      "epoch": 3.2225853928933232,
-      "grad_norm": 0.003527930937707424,
-      "learning_rate": 0.00019999487764735024,
-      "loss": 46.0,
-      "step": 42149
-    },
-    {
-      "epoch": 3.222661849876713,
-      "grad_norm": 0.004277333617210388,
-      "learning_rate": 0.00019999487740423246,
-      "loss": 46.0,
-      "step": 42150
-    },
-    {
-      "epoch": 3.2227383068601028,
-      "grad_norm": 0.001274807145819068,
-      "learning_rate": 0.0001999948771611089,
-      "loss": 46.0,
-      "step": 42151
-    },
-    {
-      "epoch": 3.2228147638434925,
-      "grad_norm": 0.00401354068890214,
-      "learning_rate": 0.00019999487691797963,
-      "loss": 46.0,
-      "step": 42152
-    },
-    {
-      "epoch": 3.2228912208268823,
-      "grad_norm": 0.0032939906232059,
-      "learning_rate": 0.00019999487667484456,
-      "loss": 46.0,
-      "step": 42153
-    },
-    {
-      "epoch": 3.222967677810272,
-      "grad_norm": 0.0016236893134191632,
-      "learning_rate": 0.0001999948764317037,
-      "loss": 46.0,
-      "step": 42154
-    },
-    {
-      "epoch": 3.223044134793662,
-      "grad_norm": 0.0009794343495741487,
-      "learning_rate": 0.0001999948761885571,
-      "loss": 46.0,
-      "step": 42155
-    },
-    {
-      "epoch": 3.2231205917770516,
-      "grad_norm": 0.0020132572390139103,
-      "learning_rate": 0.00019999487594540475,
-      "loss": 46.0,
-      "step": 42156
-    },
-    {
-      "epoch": 3.2231970487604413,
-      "grad_norm": 0.0038490155711770058,
-      "learning_rate": 0.0001999948757022466,
-      "loss": 46.0,
-      "step": 42157
-    },
-    {
-      "epoch": 3.2232735057438306,
-      "grad_norm": 0.002637245459482074,
-      "learning_rate": 0.00019999487545908267,
-      "loss": 46.0,
-      "step": 42158
-    },
-    {
-      "epoch": 3.2233499627272204,
-      "grad_norm": 0.004638268146663904,
-      "learning_rate": 0.000199994875215913,
-      "loss": 46.0,
-      "step": 42159
-    },
-    {
-      "epoch": 3.22342641971061,
-      "grad_norm": 0.00341234402731061,
-      "learning_rate": 0.00019999487497273752,
-      "loss": 46.0,
-      "step": 42160
-    },
-    {
-      "epoch": 3.223502876694,
-      "grad_norm": 0.0012906150659546256,
-      "learning_rate": 0.0001999948747295563,
-      "loss": 46.0,
-      "step": 42161
-    },
-    {
-      "epoch": 3.2235793336773897,
-      "grad_norm": 0.008898580446839333,
-      "learning_rate": 0.00019999487448636933,
-      "loss": 46.0,
-      "step": 42162
-    },
-    {
-      "epoch": 3.2236557906607795,
-      "grad_norm": 0.003250592155382037,
-      "learning_rate": 0.00019999487424317655,
-      "loss": 46.0,
-      "step": 42163
-    },
-    {
-      "epoch": 3.223732247644169,
-      "grad_norm": 0.0007815755670890212,
-      "learning_rate": 0.00019999487399997802,
-      "loss": 46.0,
-      "step": 42164
-    },
-    {
-      "epoch": 3.223808704627559,
-      "grad_norm": 0.0019414139678701758,
-      "learning_rate": 0.00019999487375677372,
-      "loss": 46.0,
-      "step": 42165
-    },
-    {
-      "epoch": 3.2238851616109487,
-      "grad_norm": 0.0028887789230793715,
-      "learning_rate": 0.00019999487351356367,
-      "loss": 46.0,
-      "step": 42166
-    },
-    {
-      "epoch": 3.2239616185943385,
-      "grad_norm": 0.0013588375877588987,
-      "learning_rate": 0.00019999487327034782,
-      "loss": 46.0,
-      "step": 42167
-    },
-    {
-      "epoch": 3.2240380755777283,
-      "grad_norm": 0.0006988157401792705,
-      "learning_rate": 0.0001999948730271262,
-      "loss": 46.0,
-      "step": 42168
-    },
-    {
-      "epoch": 3.224114532561118,
-      "grad_norm": 0.0029352956917136908,
-      "learning_rate": 0.00019999487278389883,
-      "loss": 46.0,
-      "step": 42169
-    },
-    {
-      "epoch": 3.2241909895445073,
-      "grad_norm": 0.0019444705685600638,
-      "learning_rate": 0.0001999948725406657,
-      "loss": 46.0,
-      "step": 42170
-    },
-    {
-      "epoch": 3.224267446527897,
-      "grad_norm": 0.002626130823045969,
-      "learning_rate": 0.0001999948722974268,
-      "loss": 46.0,
-      "step": 42171
-    },
-    {
-      "epoch": 3.224343903511287,
-      "grad_norm": 0.002297281753271818,
-      "learning_rate": 0.0001999948720541821,
-      "loss": 46.0,
-      "step": 42172
-    },
-    {
-      "epoch": 3.2244203604946766,
-      "grad_norm": 0.006431537214666605,
-      "learning_rate": 0.00019999487181093163,
-      "loss": 46.0,
-      "step": 42173
-    },
-    {
-      "epoch": 3.2244968174780664,
-      "grad_norm": 0.0007628339226357639,
-      "learning_rate": 0.00019999487156767542,
-      "loss": 46.0,
-      "step": 42174
-    },
-    {
-      "epoch": 3.224573274461456,
-      "grad_norm": 0.0028832328971475363,
-      "learning_rate": 0.00019999487132441344,
-      "loss": 46.0,
-      "step": 42175
-    },
-    {
-      "epoch": 3.224649731444846,
-      "grad_norm": 0.0021674379240721464,
-      "learning_rate": 0.00019999487108114566,
-      "loss": 46.0,
-      "step": 42176
-    },
-    {
-      "epoch": 3.2247261884282357,
-      "grad_norm": 0.00283754407428205,
-      "learning_rate": 0.00019999487083787213,
-      "loss": 46.0,
-      "step": 42177
-    },
-    {
-      "epoch": 3.2248026454116254,
-      "grad_norm": 0.00285433535464108,
-      "learning_rate": 0.00019999487059459283,
-      "loss": 46.0,
-      "step": 42178
-    },
-    {
-      "epoch": 3.224879102395015,
-      "grad_norm": 0.0012359402608126402,
-      "learning_rate": 0.00019999487035130779,
-      "loss": 46.0,
-      "step": 42179
-    },
-    {
-      "epoch": 3.2249555593784045,
-      "grad_norm": 0.0007884434890002012,
-      "learning_rate": 0.00019999487010801694,
-      "loss": 46.0,
-      "step": 42180
-    },
-    {
-      "epoch": 3.2250320163617943,
-      "grad_norm": 0.0018642704235389829,
-      "learning_rate": 0.00019999486986472032,
-      "loss": 46.0,
-      "step": 42181
-    },
-    {
-      "epoch": 3.225108473345184,
-      "grad_norm": 0.0035679785069078207,
-      "learning_rate": 0.00019999486962141795,
-      "loss": 46.0,
-      "step": 42182
-    },
-    {
-      "epoch": 3.225184930328574,
-      "grad_norm": 0.002079169498756528,
-      "learning_rate": 0.0001999948693781098,
-      "loss": 46.0,
-      "step": 42183
-    },
-    {
-      "epoch": 3.2252613873119635,
-      "grad_norm": 0.0013938754564151168,
-      "learning_rate": 0.00019999486913479587,
-      "loss": 46.0,
-      "step": 42184
-    },
-    {
-      "epoch": 3.2253378442953533,
-      "grad_norm": 0.0016578654758632183,
-      "learning_rate": 0.0001999948688914762,
-      "loss": 46.0,
-      "step": 42185
-    },
-    {
-      "epoch": 3.225414301278743,
-      "grad_norm": 0.001905063516460359,
-      "learning_rate": 0.00019999486864815075,
-      "loss": 46.0,
-      "step": 42186
-    },
-    {
-      "epoch": 3.225490758262133,
-      "grad_norm": 0.0017836643382906914,
-      "learning_rate": 0.00019999486840481954,
-      "loss": 46.0,
-      "step": 42187
-    },
-    {
-      "epoch": 3.2255672152455226,
-      "grad_norm": 0.004136717412620783,
-      "learning_rate": 0.00019999486816148253,
-      "loss": 46.0,
-      "step": 42188
-    },
-    {
-      "epoch": 3.2256436722289124,
-      "grad_norm": 0.0014836541377007961,
-      "learning_rate": 0.00019999486791813978,
-      "loss": 46.0,
-      "step": 42189
-    },
-    {
-      "epoch": 3.225720129212302,
-      "grad_norm": 0.003220832906663418,
-      "learning_rate": 0.00019999486767479125,
-      "loss": 46.0,
-      "step": 42190
-    },
-    {
-      "epoch": 3.225796586195692,
-      "grad_norm": 0.002322359010577202,
-      "learning_rate": 0.00019999486743143695,
-      "loss": 46.0,
-      "step": 42191
-    },
-    {
-      "epoch": 3.225873043179081,
-      "grad_norm": 0.0027075516991317272,
-      "learning_rate": 0.0001999948671880769,
-      "loss": 46.0,
-      "step": 42192
-    },
-    {
-      "epoch": 3.225949500162471,
-      "grad_norm": 0.001571964705362916,
-      "learning_rate": 0.00019999486694471106,
-      "loss": 46.0,
-      "step": 42193
-    },
-    {
-      "epoch": 3.2260259571458607,
-      "grad_norm": 0.002495530992746353,
-      "learning_rate": 0.00019999486670133946,
-      "loss": 46.0,
-      "step": 42194
-    },
-    {
-      "epoch": 3.2261024141292505,
-      "grad_norm": 0.002889525843784213,
-      "learning_rate": 0.00019999486645796207,
-      "loss": 46.0,
-      "step": 42195
-    },
-    {
-      "epoch": 3.2261788711126402,
-      "grad_norm": 0.0024617034941911697,
-      "learning_rate": 0.00019999486621457893,
-      "loss": 46.0,
-      "step": 42196
-    },
-    {
-      "epoch": 3.22625532809603,
-      "grad_norm": 0.0069183907471597195,
-      "learning_rate": 0.00019999486597119002,
-      "loss": 46.0,
-      "step": 42197
-    },
-    {
-      "epoch": 3.2263317850794198,
-      "grad_norm": 0.0038808458484709263,
-      "learning_rate": 0.00019999486572779533,
-      "loss": 46.0,
-      "step": 42198
-    },
-    {
-      "epoch": 3.2264082420628095,
-      "grad_norm": 0.0013269230257719755,
-      "learning_rate": 0.00019999486548439487,
-      "loss": 46.0,
-      "step": 42199
-    },
-    {
-      "epoch": 3.2264846990461993,
-      "grad_norm": 0.0010129095753654838,
-      "learning_rate": 0.00019999486524098866,
-      "loss": 46.0,
-      "step": 42200
-    },
-    {
-      "epoch": 3.226561156029589,
-      "grad_norm": 0.0033050503116101027,
-      "learning_rate": 0.00019999486499757668,
-      "loss": 46.0,
-      "step": 42201
-    },
-    {
-      "epoch": 3.2266376130129784,
-      "grad_norm": 0.0014685180503875017,
-      "learning_rate": 0.00019999486475415893,
-      "loss": 46.0,
-      "step": 42202
-    },
-    {
-      "epoch": 3.226714069996368,
-      "grad_norm": 0.0012390861520543694,
-      "learning_rate": 0.0001999948645107354,
-      "loss": 46.0,
-      "step": 42203
-    },
-    {
-      "epoch": 3.226790526979758,
-      "grad_norm": 0.002818741835653782,
-      "learning_rate": 0.0001999948642673061,
-      "loss": 46.0,
-      "step": 42204
-    },
-    {
-      "epoch": 3.2268669839631476,
-      "grad_norm": 0.005971470382064581,
-      "learning_rate": 0.00019999486402387106,
-      "loss": 46.0,
-      "step": 42205
-    },
-    {
-      "epoch": 3.2269434409465374,
-      "grad_norm": 0.0012805100996047258,
-      "learning_rate": 0.0001999948637804302,
-      "loss": 46.0,
-      "step": 42206
-    },
-    {
-      "epoch": 3.227019897929927,
-      "grad_norm": 0.0034617858473211527,
-      "learning_rate": 0.0001999948635369836,
-      "loss": 46.0,
-      "step": 42207
-    },
-    {
-      "epoch": 3.227096354913317,
-      "grad_norm": 0.0018793948693200946,
-      "learning_rate": 0.00019999486329353122,
-      "loss": 46.0,
-      "step": 42208
-    },
-    {
-      "epoch": 3.2271728118967067,
-      "grad_norm": 0.004829760175198317,
-      "learning_rate": 0.00019999486305007305,
-      "loss": 46.0,
-      "step": 42209
-    },
-    {
-      "epoch": 3.2272492688800964,
-      "grad_norm": 0.004632372409105301,
-      "learning_rate": 0.00019999486280660914,
-      "loss": 46.0,
-      "step": 42210
-    },
-    {
-      "epoch": 3.227325725863486,
-      "grad_norm": 0.005238438956439495,
-      "learning_rate": 0.00019999486256313945,
-      "loss": 46.0,
-      "step": 42211
-    },
-    {
-      "epoch": 3.227402182846876,
-      "grad_norm": 0.0007987130084075034,
-      "learning_rate": 0.00019999486231966405,
-      "loss": 46.0,
-      "step": 42212
-    },
-    {
-      "epoch": 3.2274786398302657,
-      "grad_norm": 0.001328629325143993,
-      "learning_rate": 0.00019999486207618282,
-      "loss": 46.0,
-      "step": 42213
-    },
-    {
-      "epoch": 3.227555096813655,
-      "grad_norm": 0.001736931735649705,
-      "learning_rate": 0.00019999486183269584,
-      "loss": 46.0,
-      "step": 42214
-    },
-    {
-      "epoch": 3.227631553797045,
-      "grad_norm": 0.013588810339570045,
-      "learning_rate": 0.00019999486158920306,
-      "loss": 46.0,
-      "step": 42215
-    },
-    {
-      "epoch": 3.2277080107804346,
-      "grad_norm": 0.005290184635668993,
-      "learning_rate": 0.00019999486134570453,
-      "loss": 46.0,
-      "step": 42216
-    },
-    {
-      "epoch": 3.2277844677638243,
-      "grad_norm": 0.0028071911074221134,
-      "learning_rate": 0.00019999486110220023,
-      "loss": 46.0,
-      "step": 42217
-    },
-    {
-      "epoch": 3.227860924747214,
-      "grad_norm": 0.0028559057973325253,
-      "learning_rate": 0.00019999486085869018,
-      "loss": 46.0,
-      "step": 42218
-    },
-    {
-      "epoch": 3.227937381730604,
-      "grad_norm": 0.0024556356947869062,
-      "learning_rate": 0.00019999486061517434,
-      "loss": 46.0,
-      "step": 42219
-    },
-    {
-      "epoch": 3.2280138387139936,
-      "grad_norm": 0.004364097956568003,
-      "learning_rate": 0.00019999486037165274,
-      "loss": 46.0,
-      "step": 42220
-    },
-    {
-      "epoch": 3.2280902956973834,
-      "grad_norm": 0.003711540484800935,
-      "learning_rate": 0.00019999486012812535,
-      "loss": 46.0,
-      "step": 42221
-    },
-    {
-      "epoch": 3.228166752680773,
-      "grad_norm": 0.0024851011112332344,
-      "learning_rate": 0.0001999948598845922,
-      "loss": 46.0,
-      "step": 42222
-    },
-    {
-      "epoch": 3.228243209664163,
-      "grad_norm": 0.003588085062801838,
-      "learning_rate": 0.00019999485964105332,
-      "loss": 46.0,
-      "step": 42223
-    },
-    {
-      "epoch": 3.228319666647552,
-      "grad_norm": 0.0034592694137245417,
-      "learning_rate": 0.00019999485939750864,
-      "loss": 46.0,
-      "step": 42224
-    },
-    {
-      "epoch": 3.228396123630942,
-      "grad_norm": 0.0009275093325413764,
-      "learning_rate": 0.00019999485915395818,
-      "loss": 46.0,
-      "step": 42225
-    },
-    {
-      "epoch": 3.2284725806143317,
-      "grad_norm": 0.002045780187472701,
-      "learning_rate": 0.00019999485891040197,
-      "loss": 46.0,
-      "step": 42226
-    },
-    {
-      "epoch": 3.2285490375977215,
-      "grad_norm": 0.0020252184476703405,
-      "learning_rate": 0.00019999485866683997,
-      "loss": 46.0,
-      "step": 42227
-    },
-    {
-      "epoch": 3.2286254945811113,
-      "grad_norm": 0.001736140577122569,
-      "learning_rate": 0.0001999948584232722,
-      "loss": 46.0,
-      "step": 42228
-    },
-    {
-      "epoch": 3.228701951564501,
-      "grad_norm": 0.007463186979293823,
-      "learning_rate": 0.0001999948581796987,
-      "loss": 46.0,
-      "step": 42229
-    },
-    {
-      "epoch": 3.2287784085478908,
-      "grad_norm": 0.003951418679207563,
-      "learning_rate": 0.00019999485793611941,
-      "loss": 46.0,
-      "step": 42230
-    },
-    {
-      "epoch": 3.2288548655312805,
-      "grad_norm": 0.0013803723268210888,
-      "learning_rate": 0.00019999485769253434,
-      "loss": 46.0,
-      "step": 42231
-    },
-    {
-      "epoch": 3.2289313225146703,
-      "grad_norm": 0.007881506346166134,
-      "learning_rate": 0.0001999948574489435,
-      "loss": 46.0,
-      "step": 42232
-    },
-    {
-      "epoch": 3.22900777949806,
-      "grad_norm": 0.002667024964466691,
-      "learning_rate": 0.0001999948572053469,
-      "loss": 46.0,
-      "step": 42233
-    },
-    {
-      "epoch": 3.22908423648145,
-      "grad_norm": 0.0018889637431129813,
-      "learning_rate": 0.0001999948569617445,
-      "loss": 46.0,
-      "step": 42234
-    },
-    {
-      "epoch": 3.229160693464839,
-      "grad_norm": 0.0023049158044159412,
-      "learning_rate": 0.00019999485671813637,
-      "loss": 46.0,
-      "step": 42235
-    },
-    {
-      "epoch": 3.229237150448229,
-      "grad_norm": 0.0022834099363535643,
-      "learning_rate": 0.00019999485647452249,
-      "loss": 46.0,
-      "step": 42236
-    },
-    {
-      "epoch": 3.2293136074316187,
-      "grad_norm": 0.000989936525002122,
-      "learning_rate": 0.0001999948562309028,
-      "loss": 46.0,
-      "step": 42237
-    },
-    {
-      "epoch": 3.2293900644150084,
-      "grad_norm": 0.005748219322413206,
-      "learning_rate": 0.00019999485598727734,
-      "loss": 46.0,
-      "step": 42238
-    },
-    {
-      "epoch": 3.229466521398398,
-      "grad_norm": 0.0011627700878307223,
-      "learning_rate": 0.00019999485574364613,
-      "loss": 46.0,
-      "step": 42239
-    },
-    {
-      "epoch": 3.229542978381788,
-      "grad_norm": 0.005285852123051882,
-      "learning_rate": 0.00019999485550000915,
-      "loss": 46.0,
-      "step": 42240
-    },
-    {
-      "epoch": 3.2296194353651777,
-      "grad_norm": 0.002274313708767295,
-      "learning_rate": 0.00019999485525636638,
-      "loss": 46.0,
-      "step": 42241
-    },
-    {
-      "epoch": 3.2296958923485675,
-      "grad_norm": 0.001398901455104351,
-      "learning_rate": 0.00019999485501271788,
-      "loss": 46.0,
-      "step": 42242
-    },
-    {
-      "epoch": 3.2297723493319572,
-      "grad_norm": 0.0008692287956364453,
-      "learning_rate": 0.00019999485476906358,
-      "loss": 46.0,
-      "step": 42243
-    },
-    {
-      "epoch": 3.229848806315347,
-      "grad_norm": 0.0049063218757510185,
-      "learning_rate": 0.00019999485452540353,
-      "loss": 46.0,
-      "step": 42244
-    },
-    {
-      "epoch": 3.2299252632987367,
-      "grad_norm": 0.0018107844516634941,
-      "learning_rate": 0.00019999485428173769,
-      "loss": 46.0,
-      "step": 42245
-    },
-    {
-      "epoch": 3.230001720282126,
-      "grad_norm": 0.002421931829303503,
-      "learning_rate": 0.00019999485403806607,
-      "loss": 46.0,
-      "step": 42246
-    },
-    {
-      "epoch": 3.230078177265516,
-      "grad_norm": 0.002113932278007269,
-      "learning_rate": 0.00019999485379438873,
-      "loss": 46.0,
-      "step": 42247
-    },
-    {
-      "epoch": 3.2301546342489056,
-      "grad_norm": 0.004734083078801632,
-      "learning_rate": 0.00019999485355070556,
-      "loss": 46.0,
-      "step": 42248
-    },
-    {
-      "epoch": 3.2302310912322953,
-      "grad_norm": 0.001262399833649397,
-      "learning_rate": 0.00019999485330701665,
-      "loss": 46.0,
-      "step": 42249
-    },
-    {
-      "epoch": 3.230307548215685,
-      "grad_norm": 0.004366319160908461,
-      "learning_rate": 0.00019999485306332196,
-      "loss": 46.0,
-      "step": 42250
-    },
-    {
-      "epoch": 3.230384005199075,
-      "grad_norm": 0.0025451062247157097,
-      "learning_rate": 0.0001999948528196215,
-      "loss": 46.0,
-      "step": 42251
-    },
-    {
-      "epoch": 3.2304604621824646,
-      "grad_norm": 0.0019479583716019988,
-      "learning_rate": 0.0001999948525759153,
-      "loss": 46.0,
-      "step": 42252
-    },
-    {
-      "epoch": 3.2305369191658544,
-      "grad_norm": 0.0019829554948955774,
-      "learning_rate": 0.00019999485233220332,
-      "loss": 46.0,
-      "step": 42253
-    },
-    {
-      "epoch": 3.230613376149244,
-      "grad_norm": 0.0038967791479080915,
-      "learning_rate": 0.00019999485208848557,
-      "loss": 46.0,
-      "step": 42254
-    },
-    {
-      "epoch": 3.230689833132634,
-      "grad_norm": 0.001703653484582901,
-      "learning_rate": 0.00019999485184476204,
-      "loss": 46.0,
-      "step": 42255
-    },
-    {
-      "epoch": 3.2307662901160237,
-      "grad_norm": 0.0012313113547861576,
-      "learning_rate": 0.00019999485160103274,
-      "loss": 46.0,
-      "step": 42256
-    },
-    {
-      "epoch": 3.230842747099413,
-      "grad_norm": 0.0022776711266487837,
-      "learning_rate": 0.0001999948513572977,
-      "loss": 46.0,
-      "step": 42257
-    },
-    {
-      "epoch": 3.2309192040828028,
-      "grad_norm": 0.0036595072597265244,
-      "learning_rate": 0.00019999485111355685,
-      "loss": 46.0,
-      "step": 42258
-    },
-    {
-      "epoch": 3.2309956610661925,
-      "grad_norm": 0.002467468148097396,
-      "learning_rate": 0.00019999485086981026,
-      "loss": 46.0,
-      "step": 42259
-    },
-    {
-      "epoch": 3.2310721180495823,
-      "grad_norm": 0.003750624367967248,
-      "learning_rate": 0.00019999485062605787,
-      "loss": 46.0,
-      "step": 42260
-    },
-    {
-      "epoch": 3.231148575032972,
-      "grad_norm": 0.00116166437510401,
-      "learning_rate": 0.00019999485038229976,
-      "loss": 46.0,
-      "step": 42261
-    },
-    {
-      "epoch": 3.231225032016362,
-      "grad_norm": 0.0011302041821181774,
-      "learning_rate": 0.00019999485013853585,
-      "loss": 46.0,
-      "step": 42262
-    },
-    {
-      "epoch": 3.2313014889997516,
-      "grad_norm": 0.0030286142136901617,
-      "learning_rate": 0.0001999948498947662,
-      "loss": 46.0,
-      "step": 42263
-    },
-    {
-      "epoch": 3.2313779459831413,
-      "grad_norm": 0.003529604757204652,
-      "learning_rate": 0.0001999948496509907,
-      "loss": 46.0,
-      "step": 42264
-    },
-    {
-      "epoch": 3.231454402966531,
-      "grad_norm": 0.001110009616240859,
-      "learning_rate": 0.00019999484940720952,
-      "loss": 46.0,
-      "step": 42265
-    },
-    {
-      "epoch": 3.231530859949921,
-      "grad_norm": 0.011706104502081871,
-      "learning_rate": 0.00019999484916342252,
-      "loss": 46.0,
-      "step": 42266
-    },
-    {
-      "epoch": 3.2316073169333106,
-      "grad_norm": 0.002430995460599661,
-      "learning_rate": 0.00019999484891962977,
-      "loss": 46.0,
-      "step": 42267
-    },
-    {
-      "epoch": 3.2316837739167,
-      "grad_norm": 0.0028907330706715584,
-      "learning_rate": 0.0001999948486758312,
-      "loss": 46.0,
-      "step": 42268
-    },
-    {
-      "epoch": 3.2317602309000897,
-      "grad_norm": 0.0011207496281713247,
-      "learning_rate": 0.00019999484843202697,
-      "loss": 46.0,
-      "step": 42269
-    },
-    {
-      "epoch": 3.2318366878834794,
-      "grad_norm": 0.0014401569496840239,
-      "learning_rate": 0.00019999484818821687,
-      "loss": 46.0,
-      "step": 42270
-    },
-    {
-      "epoch": 3.231913144866869,
-      "grad_norm": 0.0014576904941350222,
-      "learning_rate": 0.00019999484794440105,
-      "loss": 46.0,
-      "step": 42271
-    },
-    {
-      "epoch": 3.231989601850259,
-      "grad_norm": 0.0020687333308160305,
-      "learning_rate": 0.00019999484770057946,
-      "loss": 46.0,
-      "step": 42272
-    },
-    {
-      "epoch": 3.2320660588336487,
-      "grad_norm": 0.0014395163161680102,
-      "learning_rate": 0.00019999484745675207,
-      "loss": 46.0,
-      "step": 42273
-    },
-    {
-      "epoch": 3.2321425158170385,
-      "grad_norm": 0.0031149149872362614,
-      "learning_rate": 0.00019999484721291896,
-      "loss": 46.0,
-      "step": 42274
-    },
-    {
-      "epoch": 3.2322189728004282,
-      "grad_norm": 0.001901002717204392,
-      "learning_rate": 0.00019999484696908004,
-      "loss": 46.0,
-      "step": 42275
-    },
-    {
-      "epoch": 3.232295429783818,
-      "grad_norm": 0.0023491359315812588,
-      "learning_rate": 0.00019999484672523536,
-      "loss": 46.0,
-      "step": 42276
-    },
-    {
-      "epoch": 3.2323718867672078,
-      "grad_norm": 0.001346286735497415,
-      "learning_rate": 0.0001999948464813849,
-      "loss": 46.0,
-      "step": 42277
-    },
-    {
-      "epoch": 3.2324483437505975,
-      "grad_norm": 0.0024885758757591248,
-      "learning_rate": 0.00019999484623752872,
-      "loss": 46.0,
-      "step": 42278
-    },
-    {
-      "epoch": 3.232524800733987,
-      "grad_norm": 0.0045378971844911575,
-      "learning_rate": 0.00019999484599366675,
-      "loss": 46.0,
-      "step": 42279
-    },
-    {
-      "epoch": 3.2326012577173766,
-      "grad_norm": 0.001267250976525247,
-      "learning_rate": 0.00019999484574979897,
-      "loss": 46.0,
-      "step": 42280
-    },
-    {
-      "epoch": 3.2326777147007664,
-      "grad_norm": 0.0010063250083476305,
-      "learning_rate": 0.00019999484550592544,
-      "loss": 46.0,
-      "step": 42281
-    },
-    {
-      "epoch": 3.232754171684156,
-      "grad_norm": 0.005992132239043713,
-      "learning_rate": 0.00019999484526204614,
-      "loss": 46.0,
-      "step": 42282
-    },
-    {
-      "epoch": 3.232830628667546,
-      "grad_norm": 0.00469317426905036,
-      "learning_rate": 0.0001999948450181611,
-      "loss": 46.0,
-      "step": 42283
-    },
-    {
-      "epoch": 3.2329070856509357,
-      "grad_norm": 0.0036912125069648027,
-      "learning_rate": 0.00019999484477427025,
-      "loss": 46.0,
-      "step": 42284
-    },
-    {
-      "epoch": 3.2329835426343254,
-      "grad_norm": 0.0022965981625020504,
-      "learning_rate": 0.00019999484453037366,
-      "loss": 46.0,
-      "step": 42285
-    },
-    {
-      "epoch": 3.233059999617715,
-      "grad_norm": 0.0011684517376124859,
-      "learning_rate": 0.00019999484428647127,
-      "loss": 46.0,
-      "step": 42286
-    },
-    {
-      "epoch": 3.233136456601105,
-      "grad_norm": 0.00434738676995039,
-      "learning_rate": 0.00019999484404256316,
-      "loss": 46.0,
-      "step": 42287
-    },
-    {
-      "epoch": 3.2332129135844947,
-      "grad_norm": 0.0029134086798876524,
-      "learning_rate": 0.00019999484379864925,
-      "loss": 46.0,
-      "step": 42288
-    },
-    {
-      "epoch": 3.233289370567884,
-      "grad_norm": 0.0018011295469477773,
-      "learning_rate": 0.0001999948435547296,
-      "loss": 46.0,
-      "step": 42289
-    },
-    {
-      "epoch": 3.2333658275512738,
-      "grad_norm": 0.002436230657622218,
-      "learning_rate": 0.00019999484331080413,
-      "loss": 46.0,
-      "step": 42290
-    },
-    {
-      "epoch": 3.2334422845346635,
-      "grad_norm": 0.0015517380088567734,
-      "learning_rate": 0.00019999484306687293,
-      "loss": 46.0,
-      "step": 42291
-    },
-    {
-      "epoch": 3.2335187415180533,
-      "grad_norm": 0.0016480621416121721,
-      "learning_rate": 0.00019999484282293595,
-      "loss": 46.0,
-      "step": 42292
-    },
-    {
-      "epoch": 3.233595198501443,
-      "grad_norm": 0.0010706520406529307,
-      "learning_rate": 0.0001999948425789932,
-      "loss": 46.0,
-      "step": 42293
-    },
-    {
-      "epoch": 3.233671655484833,
-      "grad_norm": 0.003256821306422353,
-      "learning_rate": 0.00019999484233504467,
-      "loss": 46.0,
-      "step": 42294
-    },
-    {
-      "epoch": 3.2337481124682226,
-      "grad_norm": 0.003731667762622237,
-      "learning_rate": 0.00019999484209109038,
-      "loss": 46.0,
-      "step": 42295
-    },
-    {
-      "epoch": 3.2338245694516123,
-      "grad_norm": 0.0026174504309892654,
-      "learning_rate": 0.00019999484184713033,
-      "loss": 46.0,
-      "step": 42296
-    },
-    {
-      "epoch": 3.233901026435002,
-      "grad_norm": 0.000942641228903085,
-      "learning_rate": 0.0001999948416031645,
-      "loss": 46.0,
-      "step": 42297
-    },
-    {
-      "epoch": 3.233977483418392,
-      "grad_norm": 0.0028174982871860266,
-      "learning_rate": 0.0001999948413591929,
-      "loss": 46.0,
-      "step": 42298
-    },
-    {
-      "epoch": 3.2340539404017816,
-      "grad_norm": 0.004107377491891384,
-      "learning_rate": 0.00019999484111521553,
-      "loss": 46.0,
-      "step": 42299
-    },
-    {
-      "epoch": 3.2341303973851714,
-      "grad_norm": 0.002227388322353363,
-      "learning_rate": 0.0001999948408712324,
-      "loss": 46.0,
-      "step": 42300
-    },
-    {
-      "epoch": 3.2342068543685607,
-      "grad_norm": 0.0031873807311058044,
-      "learning_rate": 0.00019999484062724348,
-      "loss": 46.0,
-      "step": 42301
-    },
-    {
-      "epoch": 3.2342833113519505,
-      "grad_norm": 0.0014232448302209377,
-      "learning_rate": 0.0001999948403832488,
-      "loss": 46.0,
-      "step": 42302
-    },
-    {
-      "epoch": 3.2343597683353402,
-      "grad_norm": 0.005269214510917664,
-      "learning_rate": 0.00019999484013924837,
-      "loss": 46.0,
-      "step": 42303
-    },
-    {
-      "epoch": 3.23443622531873,
-      "grad_norm": 0.0007990125450305641,
-      "learning_rate": 0.00019999483989524214,
-      "loss": 46.0,
-      "step": 42304
-    },
-    {
-      "epoch": 3.2345126823021197,
-      "grad_norm": 0.005071209277957678,
-      "learning_rate": 0.00019999483965123016,
-      "loss": 46.0,
-      "step": 42305
-    },
-    {
-      "epoch": 3.2345891392855095,
-      "grad_norm": 0.0007935185567475855,
-      "learning_rate": 0.0001999948394072124,
-      "loss": 46.0,
-      "step": 42306
-    },
-    {
-      "epoch": 3.2346655962688993,
-      "grad_norm": 0.0021422707941383123,
-      "learning_rate": 0.0001999948391631889,
-      "loss": 46.0,
-      "step": 42307
-    },
-    {
-      "epoch": 3.234742053252289,
-      "grad_norm": 0.0008184100151993334,
-      "learning_rate": 0.00019999483891915961,
-      "loss": 46.0,
-      "step": 42308
-    },
-    {
-      "epoch": 3.234818510235679,
-      "grad_norm": 0.0026160688139498234,
-      "learning_rate": 0.00019999483867512457,
-      "loss": 46.0,
-      "step": 42309
-    },
-    {
-      "epoch": 3.2348949672190686,
-      "grad_norm": 0.0016637092921882868,
-      "learning_rate": 0.00019999483843108373,
-      "loss": 46.0,
-      "step": 42310
-    },
-    {
-      "epoch": 3.234971424202458,
-      "grad_norm": 0.0014725946821272373,
-      "learning_rate": 0.00019999483818703714,
-      "loss": 46.0,
-      "step": 42311
-    },
-    {
-      "epoch": 3.2350478811858476,
-      "grad_norm": 0.003380876500159502,
-      "learning_rate": 0.00019999483794298474,
-      "loss": 46.0,
-      "step": 42312
-    },
-    {
-      "epoch": 3.2351243381692374,
-      "grad_norm": 0.0012476126430556178,
-      "learning_rate": 0.00019999483769892663,
-      "loss": 46.0,
-      "step": 42313
-    },
-    {
-      "epoch": 3.235200795152627,
-      "grad_norm": 0.0018260172801092267,
-      "learning_rate": 0.00019999483745486272,
-      "loss": 46.0,
-      "step": 42314
-    },
-    {
-      "epoch": 3.235277252136017,
-      "grad_norm": 0.0021838087122887373,
-      "learning_rate": 0.00019999483721079307,
-      "loss": 46.0,
-      "step": 42315
-    },
-    {
-      "epoch": 3.2353537091194067,
-      "grad_norm": 0.002953731920570135,
-      "learning_rate": 0.00019999483696671758,
-      "loss": 46.0,
-      "step": 42316
-    },
-    {
-      "epoch": 3.2354301661027964,
-      "grad_norm": 0.015879008919000626,
-      "learning_rate": 0.0001999948367226364,
-      "loss": 46.0,
-      "step": 42317
-    },
-    {
-      "epoch": 3.235506623086186,
-      "grad_norm": 0.004051321651786566,
-      "learning_rate": 0.00019999483647854943,
-      "loss": 46.0,
-      "step": 42318
-    },
-    {
-      "epoch": 3.235583080069576,
-      "grad_norm": 0.001257934607565403,
-      "learning_rate": 0.00019999483623445668,
-      "loss": 46.0,
-      "step": 42319
-    },
-    {
-      "epoch": 3.2356595370529657,
-      "grad_norm": 0.0026651183143258095,
-      "learning_rate": 0.00019999483599035815,
-      "loss": 46.0,
-      "step": 42320
-    },
-    {
-      "epoch": 3.2357359940363555,
-      "grad_norm": 0.003077635308727622,
-      "learning_rate": 0.00019999483574625386,
-      "loss": 46.0,
-      "step": 42321
-    },
-    {
-      "epoch": 3.2358124510197452,
-      "grad_norm": 0.0015535916900262237,
-      "learning_rate": 0.00019999483550214378,
-      "loss": 46.0,
-      "step": 42322
-    },
-    {
-      "epoch": 3.2358889080031346,
-      "grad_norm": 0.0013542391825467348,
-      "learning_rate": 0.00019999483525802797,
-      "loss": 46.0,
-      "step": 42323
-    },
-    {
-      "epoch": 3.2359653649865243,
-      "grad_norm": 0.0022651380859315395,
-      "learning_rate": 0.00019999483501390638,
-      "loss": 46.0,
-      "step": 42324
-    },
-    {
-      "epoch": 3.236041821969914,
-      "grad_norm": 0.0038192470092326403,
-      "learning_rate": 0.00019999483476977901,
-      "loss": 46.0,
-      "step": 42325
-    },
-    {
-      "epoch": 3.236118278953304,
-      "grad_norm": 0.0011043907143175602,
-      "learning_rate": 0.0001999948345256459,
-      "loss": 46.0,
-      "step": 42326
-    },
-    {
-      "epoch": 3.2361947359366936,
-      "grad_norm": 0.003787618363276124,
-      "learning_rate": 0.00019999483428150697,
-      "loss": 46.0,
-      "step": 42327
-    },
-    {
-      "epoch": 3.2362711929200834,
-      "grad_norm": 0.0033555596601217985,
-      "learning_rate": 0.0001999948340373623,
-      "loss": 46.0,
-      "step": 42328
-    },
-    {
-      "epoch": 3.236347649903473,
-      "grad_norm": 0.0008274125284515321,
-      "learning_rate": 0.00019999483379321188,
-      "loss": 46.0,
-      "step": 42329
-    },
-    {
-      "epoch": 3.236424106886863,
-      "grad_norm": 0.0022448317613452673,
-      "learning_rate": 0.00019999483354905565,
-      "loss": 46.0,
-      "step": 42330
-    },
-    {
-      "epoch": 3.2365005638702526,
-      "grad_norm": 0.002428629668429494,
-      "learning_rate": 0.00019999483330489367,
-      "loss": 46.0,
-      "step": 42331
-    },
-    {
-      "epoch": 3.2365770208536424,
-      "grad_norm": 0.0016676083905622363,
-      "learning_rate": 0.00019999483306072592,
-      "loss": 46.0,
-      "step": 42332
-    },
-    {
-      "epoch": 3.2366534778370317,
-      "grad_norm": 0.0008506293524987996,
-      "learning_rate": 0.0001999948328165524,
-      "loss": 46.0,
-      "step": 42333
-    },
-    {
-      "epoch": 3.2367299348204215,
-      "grad_norm": 0.0022451188415288925,
-      "learning_rate": 0.0001999948325723731,
-      "loss": 46.0,
-      "step": 42334
-    },
-    {
-      "epoch": 3.2368063918038112,
-      "grad_norm": 0.0033403458073735237,
-      "learning_rate": 0.00019999483232818806,
-      "loss": 46.0,
-      "step": 42335
-    },
-    {
-      "epoch": 3.236882848787201,
-      "grad_norm": 0.0023838765919208527,
-      "learning_rate": 0.00019999483208399724,
-      "loss": 46.0,
-      "step": 42336
-    },
-    {
-      "epoch": 3.2369593057705908,
-      "grad_norm": 0.0033212322741746902,
-      "learning_rate": 0.00019999483183980062,
-      "loss": 46.0,
-      "step": 42337
-    },
-    {
-      "epoch": 3.2370357627539805,
-      "grad_norm": 0.002907288260757923,
-      "learning_rate": 0.00019999483159559826,
-      "loss": 46.0,
-      "step": 42338
-    },
-    {
-      "epoch": 3.2371122197373703,
-      "grad_norm": 0.004235134460031986,
-      "learning_rate": 0.00019999483135139015,
-      "loss": 46.0,
-      "step": 42339
-    },
-    {
-      "epoch": 3.23718867672076,
-      "grad_norm": 0.0015518086729571223,
-      "learning_rate": 0.00019999483110717624,
-      "loss": 46.0,
-      "step": 42340
-    },
-    {
-      "epoch": 3.23726513370415,
-      "grad_norm": 0.002651684219017625,
-      "learning_rate": 0.00019999483086295658,
-      "loss": 46.0,
-      "step": 42341
-    },
-    {
-      "epoch": 3.2373415906875396,
-      "grad_norm": 0.0051034241914749146,
-      "learning_rate": 0.00019999483061873113,
-      "loss": 46.0,
-      "step": 42342
-    },
-    {
-      "epoch": 3.2374180476709293,
-      "grad_norm": 0.00301416520960629,
-      "learning_rate": 0.00019999483037449992,
-      "loss": 46.0,
-      "step": 42343
-    },
-    {
-      "epoch": 3.237494504654319,
-      "grad_norm": 0.0019627090077847242,
-      "learning_rate": 0.00019999483013026295,
-      "loss": 46.0,
-      "step": 42344
-    },
-    {
-      "epoch": 3.2375709616377084,
-      "grad_norm": 0.003421587636694312,
-      "learning_rate": 0.0001999948298860202,
-      "loss": 46.0,
-      "step": 42345
-    },
-    {
-      "epoch": 3.237647418621098,
-      "grad_norm": 0.002241461304947734,
-      "learning_rate": 0.00019999482964177167,
-      "loss": 46.0,
-      "step": 42346
-    },
-    {
-      "epoch": 3.237723875604488,
-      "grad_norm": 0.005922417622059584,
-      "learning_rate": 0.0001999948293975174,
-      "loss": 46.0,
-      "step": 42347
-    },
-    {
-      "epoch": 3.2378003325878777,
-      "grad_norm": 0.0024471248034387827,
-      "learning_rate": 0.00019999482915325733,
-      "loss": 46.0,
-      "step": 42348
-    },
-    {
-      "epoch": 3.2378767895712675,
-      "grad_norm": 0.0012023499002680182,
-      "learning_rate": 0.00019999482890899152,
-      "loss": 46.0,
-      "step": 42349
-    },
-    {
-      "epoch": 3.237953246554657,
-      "grad_norm": 0.001696709543466568,
-      "learning_rate": 0.00019999482866471993,
-      "loss": 46.0,
-      "step": 42350
-    },
-    {
-      "epoch": 3.238029703538047,
-      "grad_norm": 0.003926475532352924,
-      "learning_rate": 0.00019999482842044254,
-      "loss": 46.0,
-      "step": 42351
-    },
-    {
-      "epoch": 3.2381061605214367,
-      "grad_norm": 0.0016383574111387134,
-      "learning_rate": 0.00019999482817615943,
-      "loss": 46.0,
-      "step": 42352
-    },
-    {
-      "epoch": 3.2381826175048265,
-      "grad_norm": 0.0006443076417781413,
-      "learning_rate": 0.00019999482793187052,
-      "loss": 46.0,
-      "step": 42353
-    },
-    {
-      "epoch": 3.2382590744882163,
-      "grad_norm": 0.002059386810287833,
-      "learning_rate": 0.00019999482768757584,
-      "loss": 46.0,
-      "step": 42354
-    },
-    {
-      "epoch": 3.2383355314716056,
-      "grad_norm": 0.003143174108117819,
-      "learning_rate": 0.0001999948274432754,
-      "loss": 46.0,
-      "step": 42355
-    },
-    {
-      "epoch": 3.2384119884549953,
-      "grad_norm": 0.004605344031006098,
-      "learning_rate": 0.0001999948271989692,
-      "loss": 46.0,
-      "step": 42356
-    },
-    {
-      "epoch": 3.238488445438385,
-      "grad_norm": 0.001875328947789967,
-      "learning_rate": 0.0001999948269546572,
-      "loss": 46.0,
-      "step": 42357
-    },
-    {
-      "epoch": 3.238564902421775,
-      "grad_norm": 0.002174764173105359,
-      "learning_rate": 0.00019999482671033948,
-      "loss": 46.0,
-      "step": 42358
-    },
-    {
-      "epoch": 3.2386413594051646,
-      "grad_norm": 0.001651904545724392,
-      "learning_rate": 0.00019999482646601595,
-      "loss": 46.0,
-      "step": 42359
-    },
-    {
-      "epoch": 3.2387178163885544,
-      "grad_norm": 0.0035280934534966946,
-      "learning_rate": 0.00019999482622168669,
-      "loss": 46.0,
-      "step": 42360
-    },
-    {
-      "epoch": 3.238794273371944,
-      "grad_norm": 0.002604069886729121,
-      "learning_rate": 0.00019999482597735162,
-      "loss": 46.0,
-      "step": 42361
-    },
-    {
-      "epoch": 3.238870730355334,
-      "grad_norm": 0.0025517635513097048,
-      "learning_rate": 0.00019999482573301077,
-      "loss": 46.0,
-      "step": 42362
-    },
-    {
-      "epoch": 3.2389471873387237,
-      "grad_norm": 0.0016855899011716247,
-      "learning_rate": 0.00019999482548866418,
-      "loss": 46.0,
-      "step": 42363
-    },
-    {
-      "epoch": 3.2390236443221134,
-      "grad_norm": 0.0035991498734802008,
-      "learning_rate": 0.00019999482524431185,
-      "loss": 46.0,
-      "step": 42364
-    },
-    {
-      "epoch": 3.239100101305503,
-      "grad_norm": 0.0021438421681523323,
-      "learning_rate": 0.0001999948249999537,
-      "loss": 46.0,
-      "step": 42365
-    },
-    {
-      "epoch": 3.2391765582888925,
-      "grad_norm": 0.002080722479149699,
-      "learning_rate": 0.0001999948247555898,
-      "loss": 46.0,
-      "step": 42366
-    },
-    {
-      "epoch": 3.2392530152722823,
-      "grad_norm": 0.004347725305706263,
-      "learning_rate": 0.00019999482451122015,
-      "loss": 46.0,
-      "step": 42367
-    },
-    {
-      "epoch": 3.239329472255672,
-      "grad_norm": 0.002128925174474716,
-      "learning_rate": 0.00019999482426684472,
-      "loss": 46.0,
-      "step": 42368
-    },
-    {
-      "epoch": 3.239405929239062,
-      "grad_norm": 0.0020087251905351877,
-      "learning_rate": 0.0001999948240224635,
-      "loss": 46.0,
-      "step": 42369
-    },
-    {
-      "epoch": 3.2394823862224515,
-      "grad_norm": 0.0029420426581054926,
-      "learning_rate": 0.0001999948237780765,
-      "loss": 46.0,
-      "step": 42370
-    },
-    {
-      "epoch": 3.2395588432058413,
-      "grad_norm": 0.0013687177561223507,
-      "learning_rate": 0.00019999482353368376,
-      "loss": 46.0,
-      "step": 42371
-    },
-    {
-      "epoch": 3.239635300189231,
-      "grad_norm": 0.0014295891160145402,
-      "learning_rate": 0.00019999482328928524,
-      "loss": 46.0,
-      "step": 42372
-    },
-    {
-      "epoch": 3.239711757172621,
-      "grad_norm": 0.003431841731071472,
-      "learning_rate": 0.00019999482304488097,
-      "loss": 46.0,
-      "step": 42373
-    },
-    {
-      "epoch": 3.2397882141560106,
-      "grad_norm": 0.002378212520852685,
-      "learning_rate": 0.00019999482280047093,
-      "loss": 46.0,
-      "step": 42374
-    },
-    {
-      "epoch": 3.2398646711394004,
-      "grad_norm": 0.0014651797246187925,
-      "learning_rate": 0.0001999948225560551,
-      "loss": 46.0,
-      "step": 42375
-    },
-    {
-      "epoch": 3.23994112812279,
-      "grad_norm": 0.0009595670271664858,
-      "learning_rate": 0.00019999482231163352,
-      "loss": 46.0,
-      "step": 42376
-    },
-    {
-      "epoch": 3.2400175851061794,
-      "grad_norm": 0.0014238699804991484,
-      "learning_rate": 0.00019999482206720613,
-      "loss": 46.0,
-      "step": 42377
-    },
-    {
-      "epoch": 3.240094042089569,
-      "grad_norm": 0.002643199637532234,
-      "learning_rate": 0.000199994821822773,
-      "loss": 46.0,
-      "step": 42378
-    },
-    {
-      "epoch": 3.240170499072959,
-      "grad_norm": 0.0022409625817090273,
-      "learning_rate": 0.00019999482157833412,
-      "loss": 46.0,
-      "step": 42379
-    },
-    {
-      "epoch": 3.2402469560563487,
-      "grad_norm": 0.0016407917719334364,
-      "learning_rate": 0.00019999482133388946,
-      "loss": 46.0,
-      "step": 42380
-    },
-    {
-      "epoch": 3.2403234130397385,
-      "grad_norm": 0.003386077703908086,
-      "learning_rate": 0.00019999482108943898,
-      "loss": 46.0,
-      "step": 42381
-    },
-    {
-      "epoch": 3.2403998700231282,
-      "grad_norm": 0.0028493229765444994,
-      "learning_rate": 0.0001999948208449828,
-      "loss": 46.0,
-      "step": 42382
-    },
-    {
-      "epoch": 3.240476327006518,
-      "grad_norm": 0.0013239012332633138,
-      "learning_rate": 0.00019999482060052083,
-      "loss": 46.0,
-      "step": 42383
-    },
-    {
-      "epoch": 3.2405527839899078,
-      "grad_norm": 0.005308200605213642,
-      "learning_rate": 0.00019999482035605305,
-      "loss": 46.0,
-      "step": 42384
-    },
-    {
-      "epoch": 3.2406292409732975,
-      "grad_norm": 0.0030443661380559206,
-      "learning_rate": 0.00019999482011157955,
-      "loss": 46.0,
-      "step": 42385
-    },
-    {
-      "epoch": 3.2407056979566873,
-      "grad_norm": 0.001260890276171267,
-      "learning_rate": 0.00019999481986710026,
-      "loss": 46.0,
-      "step": 42386
-    },
-    {
-      "epoch": 3.240782154940077,
-      "grad_norm": 0.0007613828638568521,
-      "learning_rate": 0.00019999481962261522,
-      "loss": 46.0,
-      "step": 42387
-    },
-    {
-      "epoch": 3.2408586119234664,
-      "grad_norm": 0.000516761327162385,
-      "learning_rate": 0.00019999481937812437,
-      "loss": 46.0,
-      "step": 42388
-    },
-    {
-      "epoch": 3.240935068906856,
-      "grad_norm": 0.0027428600005805492,
-      "learning_rate": 0.0001999948191336278,
-      "loss": 46.0,
-      "step": 42389
-    },
-    {
-      "epoch": 3.241011525890246,
-      "grad_norm": 0.0010224863654002547,
-      "learning_rate": 0.00019999481888912545,
-      "loss": 46.0,
-      "step": 42390
-    },
-    {
-      "epoch": 3.2410879828736356,
-      "grad_norm": 0.00244688312523067,
-      "learning_rate": 0.00019999481864461732,
-      "loss": 46.0,
-      "step": 42391
-    },
-    {
-      "epoch": 3.2411644398570254,
-      "grad_norm": 0.0025835614651441574,
-      "learning_rate": 0.00019999481840010343,
-      "loss": 46.0,
-      "step": 42392
-    },
-    {
-      "epoch": 3.241240896840415,
-      "grad_norm": 0.002661807229742408,
-      "learning_rate": 0.00019999481815558375,
-      "loss": 46.0,
-      "step": 42393
-    },
-    {
-      "epoch": 3.241317353823805,
-      "grad_norm": 0.003028519218787551,
-      "learning_rate": 0.00019999481791105832,
-      "loss": 46.0,
-      "step": 42394
-    },
-    {
-      "epoch": 3.2413938108071947,
-      "grad_norm": 0.003145348047837615,
-      "learning_rate": 0.00019999481766652712,
-      "loss": 46.0,
-      "step": 42395
-    },
-    {
-      "epoch": 3.2414702677905844,
-      "grad_norm": 0.002282362896949053,
-      "learning_rate": 0.00019999481742199015,
-      "loss": 46.0,
-      "step": 42396
-    },
-    {
-      "epoch": 3.241546724773974,
-      "grad_norm": 0.004383843857795,
-      "learning_rate": 0.0001999948171774474,
-      "loss": 46.0,
-      "step": 42397
-    },
-    {
-      "epoch": 3.241623181757364,
-      "grad_norm": 0.003248919965699315,
-      "learning_rate": 0.00019999481693289887,
-      "loss": 46.0,
-      "step": 42398
-    },
-    {
-      "epoch": 3.2416996387407533,
-      "grad_norm": 0.005298751872032881,
-      "learning_rate": 0.0001999948166883446,
-      "loss": 46.0,
-      "step": 42399
-    },
-    {
-      "epoch": 3.241776095724143,
-      "grad_norm": 0.0025044975336641073,
-      "learning_rate": 0.00019999481644378454,
-      "loss": 46.0,
-      "step": 42400
-    },
-    {
-      "epoch": 3.241852552707533,
-      "grad_norm": 0.0033499347046017647,
-      "learning_rate": 0.00019999481619921872,
-      "loss": 46.0,
-      "step": 42401
-    },
-    {
-      "epoch": 3.2419290096909226,
-      "grad_norm": 0.003150665434077382,
-      "learning_rate": 0.00019999481595464713,
-      "loss": 46.0,
-      "step": 42402
-    },
-    {
-      "epoch": 3.2420054666743123,
-      "grad_norm": 0.0018043940654024482,
-      "learning_rate": 0.00019999481571006977,
-      "loss": 46.0,
-      "step": 42403
-    },
-    {
-      "epoch": 3.242081923657702,
-      "grad_norm": 0.005504921544343233,
-      "learning_rate": 0.00019999481546548666,
-      "loss": 46.0,
-      "step": 42404
-    },
-    {
-      "epoch": 3.242158380641092,
-      "grad_norm": 0.0024685857351869345,
-      "learning_rate": 0.00019999481522089773,
-      "loss": 46.0,
-      "step": 42405
-    },
-    {
-      "epoch": 3.2422348376244816,
-      "grad_norm": 0.008920547552406788,
-      "learning_rate": 0.00019999481497630307,
-      "loss": 46.0,
-      "step": 42406
-    },
-    {
-      "epoch": 3.2423112946078714,
-      "grad_norm": 0.004442697856575251,
-      "learning_rate": 0.00019999481473170265,
-      "loss": 46.0,
-      "step": 42407
-    },
-    {
-      "epoch": 3.242387751591261,
-      "grad_norm": 0.0037557941395789385,
-      "learning_rate": 0.00019999481448709644,
-      "loss": 46.0,
-      "step": 42408
-    },
-    {
-      "epoch": 3.242464208574651,
-      "grad_norm": 0.003162234090268612,
-      "learning_rate": 0.00019999481424248447,
-      "loss": 46.0,
-      "step": 42409
-    },
-    {
-      "epoch": 3.24254066555804,
-      "grad_norm": 0.004171868786215782,
-      "learning_rate": 0.00019999481399786672,
-      "loss": 46.0,
-      "step": 42410
-    },
-    {
-      "epoch": 3.24261712254143,
-      "grad_norm": 0.003868746105581522,
-      "learning_rate": 0.0001999948137532432,
-      "loss": 46.0,
-      "step": 42411
-    },
-    {
-      "epoch": 3.2426935795248197,
-      "grad_norm": 0.005612892098724842,
-      "learning_rate": 0.00019999481350861393,
-      "loss": 46.0,
-      "step": 42412
-    },
-    {
-      "epoch": 3.2427700365082095,
-      "grad_norm": 0.00219744723290205,
-      "learning_rate": 0.0001999948132639789,
-      "loss": 46.0,
-      "step": 42413
-    },
-    {
-      "epoch": 3.2428464934915993,
-      "grad_norm": 0.0018065678887069225,
-      "learning_rate": 0.00019999481301933805,
-      "loss": 46.0,
-      "step": 42414
-    },
-    {
-      "epoch": 3.242922950474989,
-      "grad_norm": 0.0027141955215483904,
-      "learning_rate": 0.00019999481277469149,
-      "loss": 46.0,
-      "step": 42415
-    },
-    {
-      "epoch": 3.242999407458379,
-      "grad_norm": 0.0018828174797818065,
-      "learning_rate": 0.0001999948125300391,
-      "loss": 46.0,
-      "step": 42416
-    },
-    {
-      "epoch": 3.2430758644417685,
-      "grad_norm": 0.000530295423232019,
-      "learning_rate": 0.00019999481228538096,
-      "loss": 46.0,
-      "step": 42417
-    },
-    {
-      "epoch": 3.2431523214251583,
-      "grad_norm": 0.0018832373898476362,
-      "learning_rate": 0.00019999481204071708,
-      "loss": 46.0,
-      "step": 42418
-    },
-    {
-      "epoch": 3.243228778408548,
-      "grad_norm": 0.0017762802308425307,
-      "learning_rate": 0.00019999481179604743,
-      "loss": 46.0,
-      "step": 42419
-    },
-    {
-      "epoch": 3.2433052353919374,
-      "grad_norm": 0.0012160860933363438,
-      "learning_rate": 0.000199994811551372,
-      "loss": 46.0,
-      "step": 42420
-    },
-    {
-      "epoch": 3.243381692375327,
-      "grad_norm": 0.0028269952163100243,
-      "learning_rate": 0.00019999481130669077,
-      "loss": 46.0,
-      "step": 42421
-    },
-    {
-      "epoch": 3.243458149358717,
-      "grad_norm": 0.002148570492863655,
-      "learning_rate": 0.00019999481106200382,
-      "loss": 46.0,
-      "step": 42422
-    },
-    {
-      "epoch": 3.2435346063421067,
-      "grad_norm": 0.0036894066724926233,
-      "learning_rate": 0.00019999481081731108,
-      "loss": 46.0,
-      "step": 42423
-    },
-    {
-      "epoch": 3.2436110633254964,
-      "grad_norm": 0.002990249078720808,
-      "learning_rate": 0.00019999481057261255,
-      "loss": 46.0,
-      "step": 42424
-    },
-    {
-      "epoch": 3.243687520308886,
-      "grad_norm": 0.008820493705570698,
-      "learning_rate": 0.00019999481032790829,
-      "loss": 46.0,
-      "step": 42425
-    },
-    {
-      "epoch": 3.243763977292276,
-      "grad_norm": 0.0011494450736790895,
-      "learning_rate": 0.00019999481008319822,
-      "loss": 46.0,
-      "step": 42426
-    },
-    {
-      "epoch": 3.2438404342756657,
-      "grad_norm": 0.002933912444859743,
-      "learning_rate": 0.00019999480983848243,
-      "loss": 46.0,
-      "step": 42427
-    },
-    {
-      "epoch": 3.2439168912590555,
-      "grad_norm": 0.006502246949821711,
-      "learning_rate": 0.00019999480959376082,
-      "loss": 46.0,
-      "step": 42428
-    },
-    {
-      "epoch": 3.2439933482424452,
-      "grad_norm": 0.0014507254818454385,
-      "learning_rate": 0.00019999480934903345,
-      "loss": 46.0,
-      "step": 42429
-    },
-    {
-      "epoch": 3.244069805225835,
-      "grad_norm": 0.02041405625641346,
-      "learning_rate": 0.00019999480910430032,
-      "loss": 46.0,
-      "step": 42430
-    },
-    {
-      "epoch": 3.2441462622092248,
-      "grad_norm": 0.0008442448452115059,
-      "learning_rate": 0.00019999480885956144,
-      "loss": 46.0,
-      "step": 42431
-    },
-    {
-      "epoch": 3.244222719192614,
-      "grad_norm": 0.0008688230882398784,
-      "learning_rate": 0.00019999480861481676,
-      "loss": 46.0,
-      "step": 42432
-    },
-    {
-      "epoch": 3.244299176176004,
-      "grad_norm": 0.0015755766071379185,
-      "learning_rate": 0.00019999480837006636,
-      "loss": 46.0,
-      "step": 42433
-    },
-    {
-      "epoch": 3.2443756331593936,
-      "grad_norm": 0.003014215035364032,
-      "learning_rate": 0.00019999480812531016,
-      "loss": 46.0,
-      "step": 42434
-    },
-    {
-      "epoch": 3.2444520901427834,
-      "grad_norm": 0.0018318889196962118,
-      "learning_rate": 0.00019999480788054815,
-      "loss": 46.0,
-      "step": 42435
-    },
-    {
-      "epoch": 3.244528547126173,
-      "grad_norm": 0.002004543086513877,
-      "learning_rate": 0.00019999480763578043,
-      "loss": 46.0,
-      "step": 42436
-    },
-    {
-      "epoch": 3.244605004109563,
-      "grad_norm": 0.00123366410844028,
-      "learning_rate": 0.0001999948073910069,
-      "loss": 46.0,
-      "step": 42437
-    },
-    {
-      "epoch": 3.2446814610929526,
-      "grad_norm": 0.0038025574758648872,
-      "learning_rate": 0.00019999480714622765,
-      "loss": 46.0,
-      "step": 42438
-    },
-    {
-      "epoch": 3.2447579180763424,
-      "grad_norm": 0.001203357009217143,
-      "learning_rate": 0.00019999480690144258,
-      "loss": 46.0,
-      "step": 42439
-    },
-    {
-      "epoch": 3.244834375059732,
-      "grad_norm": 0.0022955937311053276,
-      "learning_rate": 0.0001999948066566518,
-      "loss": 46.0,
-      "step": 42440
-    },
-    {
-      "epoch": 3.244910832043122,
-      "grad_norm": 0.0013015575241297483,
-      "learning_rate": 0.00019999480641185518,
-      "loss": 46.0,
-      "step": 42441
-    },
-    {
-      "epoch": 3.2449872890265112,
-      "grad_norm": 0.0011194307589903474,
-      "learning_rate": 0.00019999480616705281,
-      "loss": 46.0,
-      "step": 42442
-    },
-    {
-      "epoch": 3.245063746009901,
-      "grad_norm": 0.003715003840625286,
-      "learning_rate": 0.0001999948059222447,
-      "loss": 46.0,
-      "step": 42443
-    },
-    {
-      "epoch": 3.2451402029932908,
-      "grad_norm": 0.0019537240732461214,
-      "learning_rate": 0.0001999948056774308,
-      "loss": 46.0,
-      "step": 42444
-    },
-    {
-      "epoch": 3.2452166599766805,
-      "grad_norm": 0.0036568893119692802,
-      "learning_rate": 0.00019999480543261115,
-      "loss": 46.0,
-      "step": 42445
-    },
-    {
-      "epoch": 3.2452931169600703,
-      "grad_norm": 0.0040557486936450005,
-      "learning_rate": 0.00019999480518778572,
-      "loss": 46.0,
-      "step": 42446
-    },
-    {
-      "epoch": 3.24536957394346,
-      "grad_norm": 0.001760962069965899,
-      "learning_rate": 0.00019999480494295452,
-      "loss": 46.0,
-      "step": 42447
-    },
-    {
-      "epoch": 3.24544603092685,
-      "grad_norm": 0.01030740700662136,
-      "learning_rate": 0.00019999480469811754,
-      "loss": 46.0,
-      "step": 42448
-    },
-    {
-      "epoch": 3.2455224879102396,
-      "grad_norm": 0.0029453302267938852,
-      "learning_rate": 0.0001999948044532748,
-      "loss": 46.0,
-      "step": 42449
-    },
-    {
-      "epoch": 3.2455989448936293,
-      "grad_norm": 0.00572332926094532,
-      "learning_rate": 0.0001999948042084263,
-      "loss": 46.0,
-      "step": 42450
-    },
-    {
-      "epoch": 3.245675401877019,
-      "grad_norm": 0.010032108053565025,
-      "learning_rate": 0.000199994803963572,
-      "loss": 46.0,
-      "step": 42451
-    },
-    {
-      "epoch": 3.245751858860409,
-      "grad_norm": 0.002460707677528262,
-      "learning_rate": 0.00019999480371871194,
-      "loss": 46.0,
-      "step": 42452
-    },
-    {
-      "epoch": 3.2458283158437986,
-      "grad_norm": 0.0012489031068980694,
-      "learning_rate": 0.00019999480347384613,
-      "loss": 46.0,
-      "step": 42453
-    },
-    {
-      "epoch": 3.245904772827188,
-      "grad_norm": 0.00368308718316257,
-      "learning_rate": 0.00019999480322897454,
-      "loss": 46.0,
-      "step": 42454
-    },
-    {
-      "epoch": 3.2459812298105777,
-      "grad_norm": 0.0031295011285692453,
-      "learning_rate": 0.0001999948029840972,
-      "loss": 46.0,
-      "step": 42455
-    },
-    {
-      "epoch": 3.2460576867939674,
-      "grad_norm": 0.003631161293014884,
-      "learning_rate": 0.0001999948027392141,
-      "loss": 46.0,
-      "step": 42456
-    },
-    {
-      "epoch": 3.246134143777357,
-      "grad_norm": 0.001051891827955842,
-      "learning_rate": 0.0001999948024943252,
-      "loss": 46.0,
-      "step": 42457
-    },
-    {
-      "epoch": 3.246210600760747,
-      "grad_norm": 0.0013903718208894134,
-      "learning_rate": 0.0001999948022494305,
-      "loss": 46.0,
-      "step": 42458
-    },
-    {
-      "epoch": 3.2462870577441367,
-      "grad_norm": 0.0019744872115552425,
-      "learning_rate": 0.00019999480200453009,
-      "loss": 46.0,
-      "step": 42459
-    },
-    {
-      "epoch": 3.2463635147275265,
-      "grad_norm": 0.0035972667392343283,
-      "learning_rate": 0.0001999948017596239,
-      "loss": 46.0,
-      "step": 42460
-    },
-    {
-      "epoch": 3.2464399717109163,
-      "grad_norm": 0.0015593518037348986,
-      "learning_rate": 0.00019999480151471194,
-      "loss": 46.0,
-      "step": 42461
-    },
-    {
-      "epoch": 3.246516428694306,
-      "grad_norm": 0.001166791538707912,
-      "learning_rate": 0.00019999480126979416,
-      "loss": 46.0,
-      "step": 42462
-    },
-    {
-      "epoch": 3.2465928856776958,
-      "grad_norm": 0.001599739189259708,
-      "learning_rate": 0.00019999480102487067,
-      "loss": 46.0,
-      "step": 42463
-    },
-    {
-      "epoch": 3.246669342661085,
-      "grad_norm": 0.0015420893905684352,
-      "learning_rate": 0.0001999948007799414,
-      "loss": 46.0,
-      "step": 42464
-    },
-    {
-      "epoch": 3.246745799644475,
-      "grad_norm": 0.0004134236369282007,
-      "learning_rate": 0.00019999480053500634,
-      "loss": 46.0,
-      "step": 42465
-    },
-    {
-      "epoch": 3.2468222566278646,
-      "grad_norm": 0.0013970095897093415,
-      "learning_rate": 0.00019999480029006552,
-      "loss": 46.0,
-      "step": 42466
-    },
-    {
-      "epoch": 3.2468987136112544,
-      "grad_norm": 0.004596116952598095,
-      "learning_rate": 0.00019999480004511896,
-      "loss": 46.0,
-      "step": 42467
-    },
-    {
-      "epoch": 3.246975170594644,
-      "grad_norm": 0.007578646298497915,
-      "learning_rate": 0.0001999947998001666,
-      "loss": 46.0,
-      "step": 42468
-    },
-    {
-      "epoch": 3.247051627578034,
-      "grad_norm": 0.001443088287487626,
-      "learning_rate": 0.00019999479955520847,
-      "loss": 46.0,
-      "step": 42469
-    },
-    {
-      "epoch": 3.2471280845614237,
-      "grad_norm": 0.0016861308831721544,
-      "learning_rate": 0.0001999947993102446,
-      "loss": 46.0,
-      "step": 42470
-    },
-    {
-      "epoch": 3.2472045415448134,
-      "grad_norm": 0.004286458250135183,
-      "learning_rate": 0.0001999947990652749,
-      "loss": 46.0,
-      "step": 42471
-    },
-    {
-      "epoch": 3.247280998528203,
-      "grad_norm": 0.0011865037959069014,
-      "learning_rate": 0.00019999479882029948,
-      "loss": 46.0,
-      "step": 42472
-    },
-    {
-      "epoch": 3.247357455511593,
-      "grad_norm": 0.005233015399426222,
-      "learning_rate": 0.00019999479857531828,
-      "loss": 46.0,
-      "step": 42473
-    },
-    {
-      "epoch": 3.2474339124949827,
-      "grad_norm": 0.004462135024368763,
-      "learning_rate": 0.00019999479833033134,
-      "loss": 46.0,
-      "step": 42474
-    },
-    {
-      "epoch": 3.2475103694783725,
-      "grad_norm": 0.0011538659455254674,
-      "learning_rate": 0.0001999947980853386,
-      "loss": 46.0,
-      "step": 42475
-    },
-    {
-      "epoch": 3.2475868264617618,
-      "grad_norm": 0.0010149668669328094,
-      "learning_rate": 0.00019999479784034007,
-      "loss": 46.0,
-      "step": 42476
-    },
-    {
-      "epoch": 3.2476632834451515,
-      "grad_norm": 0.0013371151871979237,
-      "learning_rate": 0.0001999947975953358,
-      "loss": 46.0,
-      "step": 42477
-    },
-    {
-      "epoch": 3.2477397404285413,
-      "grad_norm": 0.003627282800152898,
-      "learning_rate": 0.00019999479735032576,
-      "loss": 46.0,
-      "step": 42478
-    },
-    {
-      "epoch": 3.247816197411931,
-      "grad_norm": 0.0013804096961393952,
-      "learning_rate": 0.00019999479710530995,
-      "loss": 46.0,
-      "step": 42479
-    },
-    {
-      "epoch": 3.247892654395321,
-      "grad_norm": 0.00144345557782799,
-      "learning_rate": 0.00019999479686028834,
-      "loss": 46.0,
-      "step": 42480
-    },
-    {
-      "epoch": 3.2479691113787106,
-      "grad_norm": 0.0019860013853758574,
-      "learning_rate": 0.000199994796615261,
-      "loss": 46.0,
-      "step": 42481
-    },
-    {
-      "epoch": 3.2480455683621003,
-      "grad_norm": 0.0014417103957384825,
-      "learning_rate": 0.00019999479637022787,
-      "loss": 46.0,
-      "step": 42482
-    },
-    {
-      "epoch": 3.24812202534549,
-      "grad_norm": 0.0022198727820068598,
-      "learning_rate": 0.000199994796125189,
-      "loss": 46.0,
-      "step": 42483
-    },
-    {
-      "epoch": 3.24819848232888,
-      "grad_norm": 0.0009327406296506524,
-      "learning_rate": 0.00019999479588014434,
-      "loss": 46.0,
-      "step": 42484
-    },
-    {
-      "epoch": 3.2482749393122696,
-      "grad_norm": 0.0033086647745221853,
-      "learning_rate": 0.0001999947956350939,
-      "loss": 46.0,
-      "step": 42485
-    },
-    {
-      "epoch": 3.248351396295659,
-      "grad_norm": 0.0014830627478659153,
-      "learning_rate": 0.00019999479539003768,
-      "loss": 46.0,
-      "step": 42486
-    },
-    {
-      "epoch": 3.2484278532790487,
-      "grad_norm": 0.003838838078081608,
-      "learning_rate": 0.00019999479514497574,
-      "loss": 46.0,
-      "step": 42487
-    },
-    {
-      "epoch": 3.2485043102624385,
-      "grad_norm": 0.0052585904486477375,
-      "learning_rate": 0.000199994794899908,
-      "loss": 46.0,
-      "step": 42488
-    },
-    {
-      "epoch": 3.2485807672458282,
-      "grad_norm": 0.001985125942155719,
-      "learning_rate": 0.00019999479465483447,
-      "loss": 46.0,
-      "step": 42489
-    },
-    {
-      "epoch": 3.248657224229218,
-      "grad_norm": 0.0038302717730402946,
-      "learning_rate": 0.0001999947944097552,
-      "loss": 46.0,
-      "step": 42490
-    },
-    {
-      "epoch": 3.2487336812126077,
-      "grad_norm": 0.002133285393938422,
-      "learning_rate": 0.00019999479416467017,
-      "loss": 46.0,
-      "step": 42491
-    },
-    {
-      "epoch": 3.2488101381959975,
-      "grad_norm": 0.0007181174005381763,
-      "learning_rate": 0.00019999479391957933,
-      "loss": 46.0,
-      "step": 42492
-    },
-    {
-      "epoch": 3.2488865951793873,
-      "grad_norm": 0.002585875103250146,
-      "learning_rate": 0.00019999479367448277,
-      "loss": 46.0,
-      "step": 42493
-    },
-    {
-      "epoch": 3.248963052162777,
-      "grad_norm": 0.00368873099796474,
-      "learning_rate": 0.0001999947934293804,
-      "loss": 46.0,
-      "step": 42494
-    },
-    {
-      "epoch": 3.249039509146167,
-      "grad_norm": 0.0016099958447739482,
-      "learning_rate": 0.0001999947931842723,
-      "loss": 46.0,
-      "step": 42495
-    },
-    {
-      "epoch": 3.2491159661295566,
-      "grad_norm": 0.0022658456582576036,
-      "learning_rate": 0.0001999947929391584,
-      "loss": 46.0,
-      "step": 42496
-    },
-    {
-      "epoch": 3.249192423112946,
-      "grad_norm": 0.0033016069792211056,
-      "learning_rate": 0.00019999479269403874,
-      "loss": 46.0,
-      "step": 42497
-    },
-    {
-      "epoch": 3.2492688800963356,
-      "grad_norm": 0.0019370769150555134,
-      "learning_rate": 0.00019999479244891332,
-      "loss": 46.0,
-      "step": 42498
-    },
-    {
-      "epoch": 3.2493453370797254,
-      "grad_norm": 0.0026677935384213924,
-      "learning_rate": 0.00019999479220378212,
-      "loss": 46.0,
-      "step": 42499
-    },
-    {
-      "epoch": 3.249421794063115,
-      "grad_norm": 0.0012142519699409604,
-      "learning_rate": 0.00019999479195864517,
-      "loss": 46.0,
-      "step": 42500
-    },
-    {
-      "epoch": 3.249498251046505,
-      "grad_norm": 0.005798469763249159,
-      "learning_rate": 0.00019999479171350242,
-      "loss": 46.0,
-      "step": 42501
-    },
-    {
-      "epoch": 3.2495747080298947,
-      "grad_norm": 0.0012317787623032928,
-      "learning_rate": 0.0001999947914683539,
-      "loss": 46.0,
-      "step": 42502
-    },
-    {
-      "epoch": 3.2496511650132844,
-      "grad_norm": 0.003324667690321803,
-      "learning_rate": 0.00019999479122319964,
-      "loss": 46.0,
-      "step": 42503
-    },
-    {
-      "epoch": 3.249727621996674,
-      "grad_norm": 0.003560878336429596,
-      "learning_rate": 0.0001999947909780396,
-      "loss": 46.0,
-      "step": 42504
-    },
-    {
-      "epoch": 3.249804078980064,
-      "grad_norm": 0.007173549849539995,
-      "learning_rate": 0.00019999479073287376,
-      "loss": 46.0,
-      "step": 42505
-    },
-    {
-      "epoch": 3.2498805359634537,
-      "grad_norm": 0.0010374464327469468,
-      "learning_rate": 0.0001999947904877022,
-      "loss": 46.0,
-      "step": 42506
-    },
-    {
-      "epoch": 3.2499569929468435,
-      "grad_norm": 0.002254461869597435,
-      "learning_rate": 0.00019999479024252484,
-      "loss": 46.0,
-      "step": 42507
-    },
-    {
-      "epoch": 3.250033449930233,
-      "grad_norm": 0.002280781976878643,
-      "learning_rate": 0.00019999478999734174,
-      "loss": 46.0,
-      "step": 42508
-    },
-    {
-      "epoch": 3.2501099069136226,
-      "grad_norm": 0.0025865158531814814,
-      "learning_rate": 0.00019999478975215286,
-      "loss": 46.0,
-      "step": 42509
-    },
-    {
-      "epoch": 3.2501863638970123,
-      "grad_norm": 0.0015839614206925035,
-      "learning_rate": 0.00019999478950695818,
-      "loss": 46.0,
-      "step": 42510
-    },
-    {
-      "epoch": 3.250262820880402,
-      "grad_norm": 0.002307702787220478,
-      "learning_rate": 0.00019999478926175775,
-      "loss": 46.0,
-      "step": 42511
-    },
-    {
-      "epoch": 3.250339277863792,
-      "grad_norm": 0.003494716715067625,
-      "learning_rate": 0.00019999478901655155,
-      "loss": 46.0,
-      "step": 42512
-    },
-    {
-      "epoch": 3.2504157348471816,
-      "grad_norm": 0.0016066872049123049,
-      "learning_rate": 0.00019999478877133958,
-      "loss": 46.0,
-      "step": 42513
-    },
-    {
-      "epoch": 3.2504921918305714,
-      "grad_norm": 0.0013584350235760212,
-      "learning_rate": 0.00019999478852612186,
-      "loss": 46.0,
-      "step": 42514
-    },
-    {
-      "epoch": 3.250568648813961,
-      "grad_norm": 0.0032942506950348616,
-      "learning_rate": 0.00019999478828089834,
-      "loss": 46.0,
-      "step": 42515
-    },
-    {
-      "epoch": 3.250645105797351,
-      "grad_norm": 0.003127477364614606,
-      "learning_rate": 0.00019999478803566908,
-      "loss": 46.0,
-      "step": 42516
-    },
-    {
-      "epoch": 3.2507215627807406,
-      "grad_norm": 0.0016304231248795986,
-      "learning_rate": 0.00019999478779043404,
-      "loss": 46.0,
-      "step": 42517
-    },
-    {
-      "epoch": 3.2507980197641304,
-      "grad_norm": 0.001485159038566053,
-      "learning_rate": 0.0001999947875451932,
-      "loss": 46.0,
-      "step": 42518
-    },
-    {
-      "epoch": 3.25087447674752,
-      "grad_norm": 0.0013502154033631086,
-      "learning_rate": 0.00019999478729994664,
-      "loss": 46.0,
-      "step": 42519
-    },
-    {
-      "epoch": 3.2509509337309095,
-      "grad_norm": 0.0006993597489781678,
-      "learning_rate": 0.00019999478705469428,
-      "loss": 46.0,
-      "step": 42520
-    },
-    {
-      "epoch": 3.2510273907142992,
-      "grad_norm": 0.002000403357669711,
-      "learning_rate": 0.00019999478680943618,
-      "loss": 46.0,
-      "step": 42521
-    },
-    {
-      "epoch": 3.251103847697689,
-      "grad_norm": 0.001074827741831541,
-      "learning_rate": 0.00019999478656417227,
-      "loss": 46.0,
-      "step": 42522
-    },
-    {
-      "epoch": 3.2511803046810788,
-      "grad_norm": 0.0026615653187036514,
-      "learning_rate": 0.00019999478631890262,
-      "loss": 46.0,
-      "step": 42523
-    },
-    {
-      "epoch": 3.2512567616644685,
-      "grad_norm": 0.0017869886942207813,
-      "learning_rate": 0.0001999947860736272,
-      "loss": 46.0,
-      "step": 42524
-    },
-    {
-      "epoch": 3.2513332186478583,
-      "grad_norm": 0.0012223739176988602,
-      "learning_rate": 0.00019999478582834602,
-      "loss": 46.0,
-      "step": 42525
-    },
-    {
-      "epoch": 3.251409675631248,
-      "grad_norm": 0.006573184859007597,
-      "learning_rate": 0.00019999478558305902,
-      "loss": 46.0,
-      "step": 42526
-    },
-    {
-      "epoch": 3.251486132614638,
-      "grad_norm": 0.0009233249584212899,
-      "learning_rate": 0.0001999947853377663,
-      "loss": 46.0,
-      "step": 42527
-    },
-    {
-      "epoch": 3.2515625895980276,
-      "grad_norm": 0.0022954700980335474,
-      "learning_rate": 0.0001999947850924678,
-      "loss": 46.0,
-      "step": 42528
-    },
-    {
-      "epoch": 3.251639046581417,
-      "grad_norm": 0.0026257624849677086,
-      "learning_rate": 0.00019999478484716355,
-      "loss": 46.0,
-      "step": 42529
-    },
-    {
-      "epoch": 3.2517155035648067,
-      "grad_norm": 0.002981224562972784,
-      "learning_rate": 0.00019999478460185348,
-      "loss": 46.0,
-      "step": 42530
-    },
-    {
-      "epoch": 3.2517919605481964,
-      "grad_norm": 0.0029388112016022205,
-      "learning_rate": 0.00019999478435653767,
-      "loss": 46.0,
-      "step": 42531
-    },
-    {
-      "epoch": 3.251868417531586,
-      "grad_norm": 0.0011481440160423517,
-      "learning_rate": 0.0001999947841112161,
-      "loss": 46.0,
-      "step": 42532
-    },
-    {
-      "epoch": 3.251944874514976,
-      "grad_norm": 0.002720416756346822,
-      "learning_rate": 0.00019999478386588875,
-      "loss": 46.0,
-      "step": 42533
-    },
-    {
-      "epoch": 3.2520213314983657,
-      "grad_norm": 0.0008004575502127409,
-      "learning_rate": 0.00019999478362055562,
-      "loss": 46.0,
-      "step": 42534
-    },
-    {
-      "epoch": 3.2520977884817555,
-      "grad_norm": 0.0019154442707076669,
-      "learning_rate": 0.00019999478337521674,
-      "loss": 46.0,
-      "step": 42535
-    },
-    {
-      "epoch": 3.252174245465145,
-      "grad_norm": 0.0028819534927606583,
-      "learning_rate": 0.00019999478312987206,
-      "loss": 46.0,
-      "step": 42536
-    },
-    {
-      "epoch": 3.252250702448535,
-      "grad_norm": 0.0006618181359954178,
-      "learning_rate": 0.00019999478288452167,
-      "loss": 46.0,
-      "step": 42537
-    },
-    {
-      "epoch": 3.2523271594319247,
-      "grad_norm": 0.003203276079148054,
-      "learning_rate": 0.00019999478263916544,
-      "loss": 46.0,
-      "step": 42538
-    },
-    {
-      "epoch": 3.2524036164153145,
-      "grad_norm": 0.005383047740906477,
-      "learning_rate": 0.0001999947823938035,
-      "loss": 46.0,
-      "step": 42539
-    },
-    {
-      "epoch": 3.2524800733987043,
-      "grad_norm": 0.0071266950108110905,
-      "learning_rate": 0.00019999478214843578,
-      "loss": 46.0,
-      "step": 42540
-    },
-    {
-      "epoch": 3.2525565303820936,
-      "grad_norm": 0.004323940724134445,
-      "learning_rate": 0.00019999478190306226,
-      "loss": 46.0,
-      "step": 42541
-    },
-    {
-      "epoch": 3.2526329873654833,
-      "grad_norm": 0.0020127473399043083,
-      "learning_rate": 0.000199994781657683,
-      "loss": 46.0,
-      "step": 42542
-    },
-    {
-      "epoch": 3.252709444348873,
-      "grad_norm": 0.0025062926579266787,
-      "learning_rate": 0.00019999478141229796,
-      "loss": 46.0,
-      "step": 42543
-    },
-    {
-      "epoch": 3.252785901332263,
-      "grad_norm": 0.0013228552415966988,
-      "learning_rate": 0.00019999478116690712,
-      "loss": 46.0,
-      "step": 42544
-    },
-    {
-      "epoch": 3.2528623583156526,
-      "grad_norm": 0.0021390609908849,
-      "learning_rate": 0.00019999478092151056,
-      "loss": 46.0,
-      "step": 42545
-    },
-    {
-      "epoch": 3.2529388152990424,
-      "grad_norm": 0.0009808988543227315,
-      "learning_rate": 0.00019999478067610823,
-      "loss": 46.0,
-      "step": 42546
-    },
-    {
-      "epoch": 3.253015272282432,
-      "grad_norm": 0.0014587537152692676,
-      "learning_rate": 0.00019999478043070007,
-      "loss": 46.0,
-      "step": 42547
-    },
-    {
-      "epoch": 3.253091729265822,
-      "grad_norm": 0.0009863952873274684,
-      "learning_rate": 0.00019999478018528622,
-      "loss": 46.0,
-      "step": 42548
-    },
-    {
-      "epoch": 3.2531681862492117,
-      "grad_norm": 0.002077799988910556,
-      "learning_rate": 0.00019999477993986654,
-      "loss": 46.0,
-      "step": 42549
-    },
-    {
-      "epoch": 3.2532446432326014,
-      "grad_norm": 0.0030031197238713503,
-      "learning_rate": 0.00019999477969444111,
-      "loss": 46.0,
-      "step": 42550
-    },
-    {
-      "epoch": 3.2533211002159907,
-      "grad_norm": 0.0014947708696126938,
-      "learning_rate": 0.00019999477944900992,
-      "loss": 46.0,
-      "step": 42551
-    },
-    {
-      "epoch": 3.2533975571993805,
-      "grad_norm": 0.001467434922233224,
-      "learning_rate": 0.00019999477920357297,
-      "loss": 46.0,
-      "step": 42552
-    },
-    {
-      "epoch": 3.2534740141827703,
-      "grad_norm": 0.0013639378594234586,
-      "learning_rate": 0.00019999477895813023,
-      "loss": 46.0,
-      "step": 42553
-    },
-    {
-      "epoch": 3.25355047116616,
-      "grad_norm": 0.003380757989361882,
-      "learning_rate": 0.00019999477871268174,
-      "loss": 46.0,
-      "step": 42554
-    },
-    {
-      "epoch": 3.25362692814955,
-      "grad_norm": 0.0009612800204195082,
-      "learning_rate": 0.00019999477846722744,
-      "loss": 46.0,
-      "step": 42555
-    },
-    {
-      "epoch": 3.2537033851329396,
-      "grad_norm": 0.0024307905696332455,
-      "learning_rate": 0.00019999477822176743,
-      "loss": 46.0,
-      "step": 42556
-    },
-    {
-      "epoch": 3.2537798421163293,
-      "grad_norm": 0.001676167012192309,
-      "learning_rate": 0.00019999477797630162,
-      "loss": 46.0,
-      "step": 42557
-    },
-    {
-      "epoch": 3.253856299099719,
-      "grad_norm": 0.0020378839690238237,
-      "learning_rate": 0.00019999477773083004,
-      "loss": 46.0,
-      "step": 42558
-    },
-    {
-      "epoch": 3.253932756083109,
-      "grad_norm": 0.004175650887191296,
-      "learning_rate": 0.00019999477748535268,
-      "loss": 46.0,
-      "step": 42559
-    },
-    {
-      "epoch": 3.2540092130664986,
-      "grad_norm": 0.0022654470521956682,
-      "learning_rate": 0.0001999947772398696,
-      "loss": 46.0,
-      "step": 42560
-    },
-    {
-      "epoch": 3.2540856700498884,
-      "grad_norm": 0.001592738670296967,
-      "learning_rate": 0.0001999947769943807,
-      "loss": 46.0,
-      "step": 42561
-    },
-    {
-      "epoch": 3.254162127033278,
-      "grad_norm": 0.001594061148352921,
-      "learning_rate": 0.00019999477674888602,
-      "loss": 46.0,
-      "step": 42562
-    },
-    {
-      "epoch": 3.2542385840166674,
-      "grad_norm": 0.0023689360823482275,
-      "learning_rate": 0.0001999947765033856,
-      "loss": 46.0,
-      "step": 42563
-    },
-    {
-      "epoch": 3.254315041000057,
-      "grad_norm": 0.0020583071745932102,
-      "learning_rate": 0.00019999477625787942,
-      "loss": 46.0,
-      "step": 42564
-    },
-    {
-      "epoch": 3.254391497983447,
-      "grad_norm": 0.008926953189074993,
-      "learning_rate": 0.00019999477601236745,
-      "loss": 46.0,
-      "step": 42565
-    },
-    {
-      "epoch": 3.2544679549668367,
-      "grad_norm": 0.0018196330638602376,
-      "learning_rate": 0.00019999477576684974,
-      "loss": 46.0,
-      "step": 42566
-    },
-    {
-      "epoch": 3.2545444119502265,
-      "grad_norm": 0.0008109429036267102,
-      "learning_rate": 0.00019999477552132624,
-      "loss": 46.0,
-      "step": 42567
-    },
-    {
-      "epoch": 3.2546208689336162,
-      "grad_norm": 0.0037324721924960613,
-      "learning_rate": 0.00019999477527579695,
-      "loss": 46.0,
-      "step": 42568
-    },
-    {
-      "epoch": 3.254697325917006,
-      "grad_norm": 0.006557142361998558,
-      "learning_rate": 0.00019999477503026192,
-      "loss": 46.0,
-      "step": 42569
-    },
-    {
-      "epoch": 3.2547737829003958,
-      "grad_norm": 0.003369351848959923,
-      "learning_rate": 0.0001999947747847211,
-      "loss": 46.0,
-      "step": 42570
-    },
-    {
-      "epoch": 3.2548502398837855,
-      "grad_norm": 0.004526851233094931,
-      "learning_rate": 0.00019999477453917455,
-      "loss": 46.0,
-      "step": 42571
-    },
-    {
-      "epoch": 3.2549266968671753,
-      "grad_norm": 0.0025899349711835384,
-      "learning_rate": 0.0001999947742936222,
-      "loss": 46.0,
-      "step": 42572
-    },
-    {
-      "epoch": 3.2550031538505646,
-      "grad_norm": 0.004239336121827364,
-      "learning_rate": 0.00019999477404806406,
-      "loss": 46.0,
-      "step": 42573
-    },
-    {
-      "epoch": 3.2550796108339544,
-      "grad_norm": 0.001957200700417161,
-      "learning_rate": 0.00019999477380250018,
-      "loss": 46.0,
-      "step": 42574
-    },
-    {
-      "epoch": 3.255156067817344,
-      "grad_norm": 0.0036439464893192053,
-      "learning_rate": 0.00019999477355693053,
-      "loss": 46.0,
-      "step": 42575
-    },
-    {
-      "epoch": 3.255232524800734,
-      "grad_norm": 0.0010007243836298585,
-      "learning_rate": 0.0001999947733113551,
-      "loss": 46.0,
-      "step": 42576
-    },
-    {
-      "epoch": 3.2553089817841236,
-      "grad_norm": 0.004350172821432352,
-      "learning_rate": 0.0001999947730657739,
-      "loss": 46.0,
-      "step": 42577
-    },
-    {
-      "epoch": 3.2553854387675134,
-      "grad_norm": 0.0032392379362136126,
-      "learning_rate": 0.00019999477282018696,
-      "loss": 46.0,
-      "step": 42578
-    },
-    {
-      "epoch": 3.255461895750903,
-      "grad_norm": 0.001384774805046618,
-      "learning_rate": 0.00019999477257459422,
-      "loss": 46.0,
-      "step": 42579
-    },
-    {
-      "epoch": 3.255538352734293,
-      "grad_norm": 0.00437150476500392,
-      "learning_rate": 0.00019999477232899573,
-      "loss": 46.0,
-      "step": 42580
-    },
-    {
-      "epoch": 3.2556148097176827,
-      "grad_norm": 0.0033775190822780132,
-      "learning_rate": 0.00019999477208339147,
-      "loss": 46.0,
-      "step": 42581
-    },
-    {
-      "epoch": 3.2556912667010725,
-      "grad_norm": 0.0045274593867361546,
-      "learning_rate": 0.00019999477183778143,
-      "loss": 46.0,
-      "step": 42582
-    },
-    {
-      "epoch": 3.255767723684462,
-      "grad_norm": 0.0024706183467060328,
-      "learning_rate": 0.00019999477159216562,
-      "loss": 46.0,
-      "step": 42583
-    },
-    {
-      "epoch": 3.255844180667852,
-      "grad_norm": 0.0016953485319390893,
-      "learning_rate": 0.00019999477134654403,
-      "loss": 46.0,
-      "step": 42584
-    },
-    {
-      "epoch": 3.2559206376512413,
-      "grad_norm": 0.002129747299477458,
-      "learning_rate": 0.00019999477110091668,
-      "loss": 46.0,
-      "step": 42585
-    },
-    {
-      "epoch": 3.255997094634631,
-      "grad_norm": 0.003871861146762967,
-      "learning_rate": 0.00019999477085528357,
-      "loss": 46.0,
-      "step": 42586
-    },
-    {
-      "epoch": 3.256073551618021,
-      "grad_norm": 0.003804868320003152,
-      "learning_rate": 0.0001999947706096447,
-      "loss": 46.0,
-      "step": 42587
-    },
-    {
-      "epoch": 3.2561500086014106,
-      "grad_norm": 0.0009133187704719603,
-      "learning_rate": 0.00019999477036400005,
-      "loss": 46.0,
-      "step": 42588
-    },
-    {
-      "epoch": 3.2562264655848003,
-      "grad_norm": 0.002121273661032319,
-      "learning_rate": 0.00019999477011834965,
-      "loss": 46.0,
-      "step": 42589
-    },
-    {
-      "epoch": 3.25630292256819,
-      "grad_norm": 0.0006553675048053265,
-      "learning_rate": 0.00019999476987269342,
-      "loss": 46.0,
-      "step": 42590
-    },
-    {
-      "epoch": 3.25637937955158,
-      "grad_norm": 0.0021904862951487303,
-      "learning_rate": 0.00019999476962703148,
-      "loss": 46.0,
-      "step": 42591
-    },
-    {
-      "epoch": 3.2564558365349696,
-      "grad_norm": 0.009102953597903252,
-      "learning_rate": 0.00019999476938136374,
-      "loss": 46.0,
-      "step": 42592
-    },
-    {
-      "epoch": 3.2565322935183594,
-      "grad_norm": 0.00148000568151474,
-      "learning_rate": 0.00019999476913569025,
-      "loss": 46.0,
-      "step": 42593
-    },
-    {
-      "epoch": 3.256608750501749,
-      "grad_norm": 0.001783793093636632,
-      "learning_rate": 0.00019999476889001096,
-      "loss": 46.0,
-      "step": 42594
-    },
-    {
-      "epoch": 3.2566852074851385,
-      "grad_norm": 0.0015115505084395409,
-      "learning_rate": 0.00019999476864432595,
-      "loss": 46.0,
-      "step": 42595
-    },
-    {
-      "epoch": 3.256761664468528,
-      "grad_norm": 0.002230108482763171,
-      "learning_rate": 0.00019999476839863514,
-      "loss": 46.0,
-      "step": 42596
-    },
-    {
-      "epoch": 3.256838121451918,
-      "grad_norm": 0.0026826856192201376,
-      "learning_rate": 0.00019999476815293855,
-      "loss": 46.0,
-      "step": 42597
-    },
-    {
-      "epoch": 3.2569145784353077,
-      "grad_norm": 0.003217132994905114,
-      "learning_rate": 0.00019999476790723622,
-      "loss": 46.0,
-      "step": 42598
-    },
-    {
-      "epoch": 3.2569910354186975,
-      "grad_norm": 0.0050801499746739864,
-      "learning_rate": 0.0001999947676615281,
-      "loss": 46.0,
-      "step": 42599
-    },
-    {
-      "epoch": 3.2570674924020873,
-      "grad_norm": 0.001110272016376257,
-      "learning_rate": 0.00019999476741581421,
-      "loss": 46.0,
-      "step": 42600
-    },
-    {
-      "epoch": 3.257143949385477,
-      "grad_norm": 0.00299544888548553,
-      "learning_rate": 0.00019999476717009456,
-      "loss": 46.0,
-      "step": 42601
-    },
-    {
-      "epoch": 3.257220406368867,
-      "grad_norm": 0.0018276463961228728,
-      "learning_rate": 0.00019999476692436914,
-      "loss": 46.0,
-      "step": 42602
-    },
-    {
-      "epoch": 3.2572968633522565,
-      "grad_norm": 0.002299498300999403,
-      "learning_rate": 0.00019999476667863797,
-      "loss": 46.0,
-      "step": 42603
-    },
-    {
-      "epoch": 3.2573733203356463,
-      "grad_norm": 0.005792070645838976,
-      "learning_rate": 0.000199994766432901,
-      "loss": 46.0,
-      "step": 42604
-    },
-    {
-      "epoch": 3.257449777319036,
-      "grad_norm": 0.005416215397417545,
-      "learning_rate": 0.00019999476618715826,
-      "loss": 46.0,
-      "step": 42605
-    },
-    {
-      "epoch": 3.257526234302426,
-      "grad_norm": 0.0034660720266401768,
-      "learning_rate": 0.00019999476594140977,
-      "loss": 46.0,
-      "step": 42606
-    },
-    {
-      "epoch": 3.257602691285815,
-      "grad_norm": 0.0030668703839182854,
-      "learning_rate": 0.0001999947656956555,
-      "loss": 46.0,
-      "step": 42607
-    },
-    {
-      "epoch": 3.257679148269205,
-      "grad_norm": 0.0012086612405255437,
-      "learning_rate": 0.0001999947654498955,
-      "loss": 46.0,
-      "step": 42608
-    },
-    {
-      "epoch": 3.2577556052525947,
-      "grad_norm": 0.0025078016333281994,
-      "learning_rate": 0.00019999476520412968,
-      "loss": 46.0,
-      "step": 42609
-    },
-    {
-      "epoch": 3.2578320622359844,
-      "grad_norm": 0.0011279458412900567,
-      "learning_rate": 0.0001999947649583581,
-      "loss": 46.0,
-      "step": 42610
-    },
-    {
-      "epoch": 3.257908519219374,
-      "grad_norm": 0.0018524121260270476,
-      "learning_rate": 0.00019999476471258074,
-      "loss": 46.0,
-      "step": 42611
-    },
-    {
-      "epoch": 3.257984976202764,
-      "grad_norm": 0.0046928441151976585,
-      "learning_rate": 0.00019999476446679764,
-      "loss": 46.0,
-      "step": 42612
-    },
-    {
-      "epoch": 3.2580614331861537,
-      "grad_norm": 0.003628105390816927,
-      "learning_rate": 0.00019999476422100876,
-      "loss": 46.0,
-      "step": 42613
-    },
-    {
-      "epoch": 3.2581378901695435,
-      "grad_norm": 0.0006739333621226251,
-      "learning_rate": 0.00019999476397521411,
-      "loss": 46.0,
-      "step": 42614
-    },
-    {
-      "epoch": 3.2582143471529332,
-      "grad_norm": 0.0037230895832180977,
-      "learning_rate": 0.0001999947637294137,
-      "loss": 46.0,
-      "step": 42615
-    },
-    {
-      "epoch": 3.258290804136323,
-      "grad_norm": 0.0009315396309830248,
-      "learning_rate": 0.0001999947634836075,
-      "loss": 46.0,
-      "step": 42616
-    },
-    {
-      "epoch": 3.2583672611197123,
-      "grad_norm": 0.00419262982904911,
-      "learning_rate": 0.00019999476323779555,
-      "loss": 46.0,
-      "step": 42617
-    },
-    {
-      "epoch": 3.258443718103102,
-      "grad_norm": 0.0019057404715567827,
-      "learning_rate": 0.0001999947629919778,
-      "loss": 46.0,
-      "step": 42618
-    },
-    {
-      "epoch": 3.258520175086492,
-      "grad_norm": 0.0011036384385079145,
-      "learning_rate": 0.00019999476274615432,
-      "loss": 46.0,
-      "step": 42619
-    },
-    {
-      "epoch": 3.2585966320698816,
-      "grad_norm": 0.0020766984671354294,
-      "learning_rate": 0.00019999476250032506,
-      "loss": 46.0,
-      "step": 42620
-    },
-    {
-      "epoch": 3.2586730890532714,
-      "grad_norm": 0.0029008088167756796,
-      "learning_rate": 0.00019999476225449002,
-      "loss": 46.0,
-      "step": 42621
-    },
-    {
-      "epoch": 3.258749546036661,
-      "grad_norm": 0.001717729028314352,
-      "learning_rate": 0.0001999947620086492,
-      "loss": 46.0,
-      "step": 42622
-    },
-    {
-      "epoch": 3.258826003020051,
-      "grad_norm": 0.0026180907152593136,
-      "learning_rate": 0.00019999476176280263,
-      "loss": 46.0,
-      "step": 42623
-    },
-    {
-      "epoch": 3.2589024600034406,
-      "grad_norm": 0.004629471804946661,
-      "learning_rate": 0.00019999476151695032,
-      "loss": 46.0,
-      "step": 42624
-    },
-    {
-      "epoch": 3.2589789169868304,
-      "grad_norm": 0.003538233693689108,
-      "learning_rate": 0.00019999476127109217,
-      "loss": 46.0,
-      "step": 42625
-    },
-    {
-      "epoch": 3.25905537397022,
-      "grad_norm": 0.0014037989312782884,
-      "learning_rate": 0.00019999476102522832,
-      "loss": 46.0,
-      "step": 42626
-    },
-    {
-      "epoch": 3.25913183095361,
-      "grad_norm": 0.000849241332616657,
-      "learning_rate": 0.00019999476077935867,
-      "loss": 46.0,
-      "step": 42627
-    },
-    {
-      "epoch": 3.2592082879369997,
-      "grad_norm": 0.0042520547285676,
-      "learning_rate": 0.00019999476053348322,
-      "loss": 46.0,
-      "step": 42628
-    },
-    {
-      "epoch": 3.259284744920389,
-      "grad_norm": 0.0011316750897094607,
-      "learning_rate": 0.00019999476028760205,
-      "loss": 46.0,
-      "step": 42629
-    },
-    {
-      "epoch": 3.2593612019037788,
-      "grad_norm": 0.002351074945181608,
-      "learning_rate": 0.00019999476004171508,
-      "loss": 46.0,
-      "step": 42630
-    },
-    {
-      "epoch": 3.2594376588871685,
-      "grad_norm": 0.0014585272874683142,
-      "learning_rate": 0.00019999475979582236,
-      "loss": 46.0,
-      "step": 42631
-    },
-    {
-      "epoch": 3.2595141158705583,
-      "grad_norm": 0.002162576885893941,
-      "learning_rate": 0.00019999475954992385,
-      "loss": 46.0,
-      "step": 42632
-    },
-    {
-      "epoch": 3.259590572853948,
-      "grad_norm": 0.0021104079205542803,
-      "learning_rate": 0.0001999947593040196,
-      "loss": 46.0,
-      "step": 42633
-    },
-    {
-      "epoch": 3.259667029837338,
-      "grad_norm": 0.003532793838530779,
-      "learning_rate": 0.00019999475905810958,
-      "loss": 46.0,
-      "step": 42634
-    },
-    {
-      "epoch": 3.2597434868207276,
-      "grad_norm": 0.0009372130152769387,
-      "learning_rate": 0.00019999475881219377,
-      "loss": 46.0,
-      "step": 42635
-    },
-    {
-      "epoch": 3.2598199438041173,
-      "grad_norm": 0.001357743632979691,
-      "learning_rate": 0.0001999947585662722,
-      "loss": 46.0,
-      "step": 42636
-    },
-    {
-      "epoch": 3.259896400787507,
-      "grad_norm": 0.0016610942548140883,
-      "learning_rate": 0.00019999475832034486,
-      "loss": 46.0,
-      "step": 42637
-    },
-    {
-      "epoch": 3.2599728577708964,
-      "grad_norm": 0.0007794471457600594,
-      "learning_rate": 0.00019999475807441175,
-      "loss": 46.0,
-      "step": 42638
-    },
-    {
-      "epoch": 3.260049314754286,
-      "grad_norm": 0.0041829184629023075,
-      "learning_rate": 0.00019999475782847288,
-      "loss": 46.0,
-      "step": 42639
-    },
-    {
-      "epoch": 3.260125771737676,
-      "grad_norm": 0.0029143677093088627,
-      "learning_rate": 0.00019999475758252823,
-      "loss": 46.0,
-      "step": 42640
-    },
-    {
-      "epoch": 3.2602022287210657,
-      "grad_norm": 0.001310272840783,
-      "learning_rate": 0.00019999475733657783,
-      "loss": 46.0,
-      "step": 42641
-    },
-    {
-      "epoch": 3.2602786857044554,
-      "grad_norm": 0.0018135359277948737,
-      "learning_rate": 0.00019999475709062164,
-      "loss": 46.0,
-      "step": 42642
-    },
-    {
-      "epoch": 3.260355142687845,
-      "grad_norm": 0.003862922079861164,
-      "learning_rate": 0.00019999475684465967,
-      "loss": 46.0,
-      "step": 42643
-    },
-    {
-      "epoch": 3.260431599671235,
-      "grad_norm": 0.0009692379971966147,
-      "learning_rate": 0.00019999475659869192,
-      "loss": 46.0,
-      "step": 42644
-    },
-    {
-      "epoch": 3.2605080566546247,
-      "grad_norm": 0.0015959467273205519,
-      "learning_rate": 0.00019999475635271844,
-      "loss": 46.0,
-      "step": 42645
-    },
-    {
-      "epoch": 3.2605845136380145,
-      "grad_norm": 0.0031691587064415216,
-      "learning_rate": 0.0001999947561067392,
-      "loss": 46.0,
-      "step": 42646
-    },
-    {
-      "epoch": 3.2606609706214043,
-      "grad_norm": 0.004590482451021671,
-      "learning_rate": 0.00019999475586075414,
-      "loss": 46.0,
-      "step": 42647
-    },
-    {
-      "epoch": 3.260737427604794,
-      "grad_norm": 0.0035140204709023237,
-      "learning_rate": 0.00019999475561476336,
-      "loss": 46.0,
-      "step": 42648
-    },
-    {
-      "epoch": 3.2608138845881838,
-      "grad_norm": 0.002837889827787876,
-      "learning_rate": 0.00019999475536876677,
-      "loss": 46.0,
-      "step": 42649
-    },
-    {
-      "epoch": 3.2608903415715735,
-      "grad_norm": 0.010890967212617397,
-      "learning_rate": 0.00019999475512276442,
-      "loss": 46.0,
-      "step": 42650
-    },
-    {
-      "epoch": 3.260966798554963,
-      "grad_norm": 0.002530425088480115,
-      "learning_rate": 0.00019999475487675631,
-      "loss": 46.0,
-      "step": 42651
-    },
-    {
-      "epoch": 3.2610432555383526,
-      "grad_norm": 0.001641754643060267,
-      "learning_rate": 0.00019999475463074244,
-      "loss": 46.0,
-      "step": 42652
-    },
-    {
-      "epoch": 3.2611197125217424,
-      "grad_norm": 0.0018559409072622657,
-      "learning_rate": 0.0001999947543847228,
-      "loss": 46.0,
-      "step": 42653
-    },
-    {
-      "epoch": 3.261196169505132,
-      "grad_norm": 0.0010790321975946426,
-      "learning_rate": 0.00019999475413869737,
-      "loss": 46.0,
-      "step": 42654
-    },
-    {
-      "epoch": 3.261272626488522,
-      "grad_norm": 0.001097700442187488,
-      "learning_rate": 0.0001999947538926662,
-      "loss": 46.0,
-      "step": 42655
-    },
-    {
-      "epoch": 3.2613490834719117,
-      "grad_norm": 0.0020291018299758434,
-      "learning_rate": 0.00019999475364662926,
-      "loss": 46.0,
-      "step": 42656
-    },
-    {
-      "epoch": 3.2614255404553014,
-      "grad_norm": 0.002114082220941782,
-      "learning_rate": 0.00019999475340058652,
-      "loss": 46.0,
-      "step": 42657
-    },
-    {
-      "epoch": 3.261501997438691,
-      "grad_norm": 0.0020981209818273783,
-      "learning_rate": 0.00019999475315453803,
-      "loss": 46.0,
-      "step": 42658
-    },
-    {
-      "epoch": 3.261578454422081,
-      "grad_norm": 0.008384820073843002,
-      "learning_rate": 0.00019999475290848374,
-      "loss": 46.0,
-      "step": 42659
-    },
-    {
-      "epoch": 3.2616549114054703,
-      "grad_norm": 0.0018982775509357452,
-      "learning_rate": 0.00019999475266242373,
-      "loss": 46.0,
-      "step": 42660
-    },
-    {
-      "epoch": 3.26173136838886,
-      "grad_norm": 0.0015684940153732896,
-      "learning_rate": 0.00019999475241635795,
-      "loss": 46.0,
-      "step": 42661
-    },
-    {
-      "epoch": 3.26180782537225,
-      "grad_norm": 0.0008714738069102168,
-      "learning_rate": 0.00019999475217028637,
-      "loss": 46.0,
-      "step": 42662
-    },
-    {
-      "epoch": 3.2618842823556395,
-      "grad_norm": 0.0014482878614217043,
-      "learning_rate": 0.000199994751924209,
-      "loss": 46.0,
-      "step": 42663
-    },
-    {
-      "epoch": 3.2619607393390293,
-      "grad_norm": 0.0006008433410897851,
-      "learning_rate": 0.0001999947516781259,
-      "loss": 46.0,
-      "step": 42664
-    },
-    {
-      "epoch": 3.262037196322419,
-      "grad_norm": 0.001276198890991509,
-      "learning_rate": 0.00019999475143203703,
-      "loss": 46.0,
-      "step": 42665
-    },
-    {
-      "epoch": 3.262113653305809,
-      "grad_norm": 0.0033647785894572735,
-      "learning_rate": 0.00019999475118594238,
-      "loss": 46.0,
-      "step": 42666
-    },
-    {
-      "epoch": 3.2621901102891986,
-      "grad_norm": 0.00391007773578167,
-      "learning_rate": 0.00019999475093984196,
-      "loss": 46.0,
-      "step": 42667
-    },
-    {
-      "epoch": 3.2622665672725883,
-      "grad_norm": 0.0030895096715539694,
-      "learning_rate": 0.0001999947506937358,
-      "loss": 46.0,
-      "step": 42668
-    },
-    {
-      "epoch": 3.262343024255978,
-      "grad_norm": 0.0021707257255911827,
-      "learning_rate": 0.00019999475044762383,
-      "loss": 46.0,
-      "step": 42669
-    },
-    {
-      "epoch": 3.262419481239368,
-      "grad_norm": 0.0015484520699828863,
-      "learning_rate": 0.0001999947502015061,
-      "loss": 46.0,
-      "step": 42670
-    },
-    {
-      "epoch": 3.2624959382227576,
-      "grad_norm": 0.0010763694299384952,
-      "learning_rate": 0.00019999474995538262,
-      "loss": 46.0,
-      "step": 42671
-    },
-    {
-      "epoch": 3.2625723952061474,
-      "grad_norm": 0.0012754467315971851,
-      "learning_rate": 0.00019999474970925336,
-      "loss": 46.0,
-      "step": 42672
-    },
-    {
-      "epoch": 3.2626488521895367,
-      "grad_norm": 0.0011959876865148544,
-      "learning_rate": 0.00019999474946311833,
-      "loss": 46.0,
-      "step": 42673
-    },
-    {
-      "epoch": 3.2627253091729265,
-      "grad_norm": 0.002192129846662283,
-      "learning_rate": 0.00019999474921697754,
-      "loss": 46.0,
-      "step": 42674
-    },
-    {
-      "epoch": 3.2628017661563162,
-      "grad_norm": 0.001396198756992817,
-      "learning_rate": 0.00019999474897083096,
-      "loss": 46.0,
-      "step": 42675
-    },
-    {
-      "epoch": 3.262878223139706,
-      "grad_norm": 0.0011704524513334036,
-      "learning_rate": 0.00019999474872467863,
-      "loss": 46.0,
-      "step": 42676
-    },
-    {
-      "epoch": 3.2629546801230958,
-      "grad_norm": 0.0010957523481920362,
-      "learning_rate": 0.0001999947484785205,
-      "loss": 46.0,
-      "step": 42677
-    },
-    {
-      "epoch": 3.2630311371064855,
-      "grad_norm": 0.0011550318449735641,
-      "learning_rate": 0.00019999474823235663,
-      "loss": 46.0,
-      "step": 42678
-    },
-    {
-      "epoch": 3.2631075940898753,
-      "grad_norm": 0.004079408943653107,
-      "learning_rate": 0.000199994747986187,
-      "loss": 46.0,
-      "step": 42679
-    },
-    {
-      "epoch": 3.263184051073265,
-      "grad_norm": 0.009575429372489452,
-      "learning_rate": 0.00019999474774001156,
-      "loss": 46.0,
-      "step": 42680
-    },
-    {
-      "epoch": 3.263260508056655,
-      "grad_norm": 0.003662359667941928,
-      "learning_rate": 0.0001999947474938304,
-      "loss": 46.0,
-      "step": 42681
-    },
-    {
-      "epoch": 3.263336965040044,
-      "grad_norm": 0.0032211500220000744,
-      "learning_rate": 0.00019999474724764345,
-      "loss": 46.0,
-      "step": 42682
-    },
-    {
-      "epoch": 3.263413422023434,
-      "grad_norm": 0.0005939346738159657,
-      "learning_rate": 0.0001999947470014507,
-      "loss": 46.0,
-      "step": 42683
-    },
-    {
-      "epoch": 3.2634898790068236,
-      "grad_norm": 0.0034532295539975166,
-      "learning_rate": 0.00019999474675525222,
-      "loss": 46.0,
-      "step": 42684
-    },
-    {
-      "epoch": 3.2635663359902134,
-      "grad_norm": 0.0016830003587529063,
-      "learning_rate": 0.00019999474650904796,
-      "loss": 46.0,
-      "step": 42685
-    },
-    {
-      "epoch": 3.263642792973603,
-      "grad_norm": 0.0014946593437343836,
-      "learning_rate": 0.00019999474626283795,
-      "loss": 46.0,
-      "step": 42686
-    },
-    {
-      "epoch": 3.263719249956993,
-      "grad_norm": 0.001080077257938683,
-      "learning_rate": 0.00019999474601662214,
-      "loss": 46.0,
-      "step": 42687
-    },
-    {
-      "epoch": 3.2637957069403827,
-      "grad_norm": 0.0014632137026637793,
-      "learning_rate": 0.00019999474577040056,
-      "loss": 46.0,
-      "step": 42688
-    },
-    {
-      "epoch": 3.2638721639237724,
-      "grad_norm": 0.0026226257905364037,
-      "learning_rate": 0.00019999474552417324,
-      "loss": 46.0,
-      "step": 42689
-    },
-    {
-      "epoch": 3.263948620907162,
-      "grad_norm": 0.0017646170454099774,
-      "learning_rate": 0.00019999474527794013,
-      "loss": 46.0,
-      "step": 42690
-    },
-    {
-      "epoch": 3.264025077890552,
-      "grad_norm": 0.008362078107893467,
-      "learning_rate": 0.00019999474503170126,
-      "loss": 46.0,
-      "step": 42691
-    },
-    {
-      "epoch": 3.2641015348739417,
-      "grad_norm": 0.0021672456059604883,
-      "learning_rate": 0.0001999947447854566,
-      "loss": 46.0,
-      "step": 42692
-    },
-    {
-      "epoch": 3.2641779918573315,
-      "grad_norm": 0.0009663395467214286,
-      "learning_rate": 0.00019999474453920622,
-      "loss": 46.0,
-      "step": 42693
-    },
-    {
-      "epoch": 3.264254448840721,
-      "grad_norm": 0.0022803430911153555,
-      "learning_rate": 0.00019999474429295002,
-      "loss": 46.0,
-      "step": 42694
-    },
-    {
-      "epoch": 3.2643309058241106,
-      "grad_norm": 0.0022487312089651823,
-      "learning_rate": 0.00019999474404668805,
-      "loss": 46.0,
-      "step": 42695
-    },
-    {
-      "epoch": 3.2644073628075003,
-      "grad_norm": 0.0052771600894629955,
-      "learning_rate": 0.00019999474380042034,
-      "loss": 46.0,
-      "step": 42696
-    },
-    {
-      "epoch": 3.26448381979089,
-      "grad_norm": 0.0023920375388115644,
-      "learning_rate": 0.00019999474355414685,
-      "loss": 46.0,
-      "step": 42697
-    },
-    {
-      "epoch": 3.26456027677428,
-      "grad_norm": 0.0010859811445698142,
-      "learning_rate": 0.0001999947433078676,
-      "loss": 46.0,
-      "step": 42698
-    },
-    {
-      "epoch": 3.2646367337576696,
-      "grad_norm": 0.0018826827872544527,
-      "learning_rate": 0.00019999474306158256,
-      "loss": 46.0,
-      "step": 42699
-    },
-    {
-      "epoch": 3.2647131907410594,
-      "grad_norm": 0.0019581117667257786,
-      "learning_rate": 0.00019999474281529178,
-      "loss": 46.0,
-      "step": 42700
-    },
-    {
-      "epoch": 3.264789647724449,
-      "grad_norm": 0.0016367704374715686,
-      "learning_rate": 0.0001999947425689952,
-      "loss": 46.0,
-      "step": 42701
-    },
-    {
-      "epoch": 3.264866104707839,
-      "grad_norm": 0.002849506912752986,
-      "learning_rate": 0.00019999474232269287,
-      "loss": 46.0,
-      "step": 42702
-    },
-    {
-      "epoch": 3.2649425616912287,
-      "grad_norm": 0.004920949228107929,
-      "learning_rate": 0.00019999474207638477,
-      "loss": 46.0,
-      "step": 42703
-    },
-    {
-      "epoch": 3.265019018674618,
-      "grad_norm": 0.0016969614662230015,
-      "learning_rate": 0.0001999947418300709,
-      "loss": 46.0,
-      "step": 42704
-    },
-    {
-      "epoch": 3.2650954756580077,
-      "grad_norm": 0.003259900491684675,
-      "learning_rate": 0.00019999474158375124,
-      "loss": 46.0,
-      "step": 42705
-    },
-    {
-      "epoch": 3.2651719326413975,
-      "grad_norm": 0.003804669948294759,
-      "learning_rate": 0.00019999474133742582,
-      "loss": 46.0,
-      "step": 42706
-    },
-    {
-      "epoch": 3.2652483896247873,
-      "grad_norm": 0.0024107638746500015,
-      "learning_rate": 0.00019999474109109463,
-      "loss": 46.0,
-      "step": 42707
-    },
-    {
-      "epoch": 3.265324846608177,
-      "grad_norm": 0.00541957002133131,
-      "learning_rate": 0.0001999947408447577,
-      "loss": 46.0,
-      "step": 42708
-    },
-    {
-      "epoch": 3.2654013035915668,
-      "grad_norm": 0.004622437059879303,
-      "learning_rate": 0.00019999474059841497,
-      "loss": 46.0,
-      "step": 42709
-    },
-    {
-      "epoch": 3.2654777605749565,
-      "grad_norm": 0.002149523701518774,
-      "learning_rate": 0.0001999947403520665,
-      "loss": 46.0,
-      "step": 42710
-    },
-    {
-      "epoch": 3.2655542175583463,
-      "grad_norm": 0.0023876260966062546,
-      "learning_rate": 0.00019999474010571225,
-      "loss": 46.0,
-      "step": 42711
-    },
-    {
-      "epoch": 3.265630674541736,
-      "grad_norm": 0.001868357416242361,
-      "learning_rate": 0.0001999947398593522,
-      "loss": 46.0,
-      "step": 42712
-    },
-    {
-      "epoch": 3.265707131525126,
-      "grad_norm": 0.0023764839861541986,
-      "learning_rate": 0.0001999947396129864,
-      "loss": 46.0,
-      "step": 42713
-    },
-    {
-      "epoch": 3.2657835885085156,
-      "grad_norm": 0.007495502475649118,
-      "learning_rate": 0.00019999473936661483,
-      "loss": 46.0,
-      "step": 42714
-    },
-    {
-      "epoch": 3.2658600454919053,
-      "grad_norm": 0.0009260806255042553,
-      "learning_rate": 0.00019999473912023753,
-      "loss": 46.0,
-      "step": 42715
-    },
-    {
-      "epoch": 3.2659365024752947,
-      "grad_norm": 0.002886514412239194,
-      "learning_rate": 0.0001999947388738544,
-      "loss": 46.0,
-      "step": 42716
-    },
-    {
-      "epoch": 3.2660129594586844,
-      "grad_norm": 0.003133345628157258,
-      "learning_rate": 0.00019999473862746553,
-      "loss": 46.0,
-      "step": 42717
-    },
-    {
-      "epoch": 3.266089416442074,
-      "grad_norm": 0.0025711958296597004,
-      "learning_rate": 0.00019999473838107088,
-      "loss": 46.0,
-      "step": 42718
-    },
-    {
-      "epoch": 3.266165873425464,
-      "grad_norm": 0.00218483479693532,
-      "learning_rate": 0.00019999473813467046,
-      "loss": 46.0,
-      "step": 42719
-    },
-    {
-      "epoch": 3.2662423304088537,
-      "grad_norm": 0.0030936570838093758,
-      "learning_rate": 0.0001999947378882643,
-      "loss": 46.0,
-      "step": 42720
-    },
-    {
-      "epoch": 3.2663187873922435,
-      "grad_norm": 0.001858058269135654,
-      "learning_rate": 0.00019999473764185235,
-      "loss": 46.0,
-      "step": 42721
-    },
-    {
-      "epoch": 3.2663952443756332,
-      "grad_norm": 0.002769799204543233,
-      "learning_rate": 0.00019999473739543464,
-      "loss": 46.0,
-      "step": 42722
-    },
-    {
-      "epoch": 3.266471701359023,
-      "grad_norm": 0.0027657856699079275,
-      "learning_rate": 0.00019999473714901113,
-      "loss": 46.0,
-      "step": 42723
-    },
-    {
-      "epoch": 3.2665481583424127,
-      "grad_norm": 0.0022426487412303686,
-      "learning_rate": 0.0001999947369025819,
-      "loss": 46.0,
-      "step": 42724
-    },
-    {
-      "epoch": 3.2666246153258025,
-      "grad_norm": 0.0014258448500186205,
-      "learning_rate": 0.00019999473665614686,
-      "loss": 46.0,
-      "step": 42725
-    },
-    {
-      "epoch": 3.266701072309192,
-      "grad_norm": 0.0010731976944953203,
-      "learning_rate": 0.00019999473640970608,
-      "loss": 46.0,
-      "step": 42726
-    },
-    {
-      "epoch": 3.2667775292925816,
-      "grad_norm": 0.0023060431703925133,
-      "learning_rate": 0.0001999947361632595,
-      "loss": 46.0,
-      "step": 42727
-    },
-    {
-      "epoch": 3.2668539862759713,
-      "grad_norm": 0.0006291406461969018,
-      "learning_rate": 0.00019999473591680714,
-      "loss": 46.0,
-      "step": 42728
-    },
-    {
-      "epoch": 3.266930443259361,
-      "grad_norm": 0.002645375905558467,
-      "learning_rate": 0.00019999473567034904,
-      "loss": 46.0,
-      "step": 42729
-    },
-    {
-      "epoch": 3.267006900242751,
-      "grad_norm": 0.0017835797043517232,
-      "learning_rate": 0.00019999473542388517,
-      "loss": 46.0,
-      "step": 42730
-    },
-    {
-      "epoch": 3.2670833572261406,
-      "grad_norm": 0.001698076375760138,
-      "learning_rate": 0.00019999473517741555,
-      "loss": 46.0,
-      "step": 42731
-    },
-    {
-      "epoch": 3.2671598142095304,
-      "grad_norm": 0.0013392699183896184,
-      "learning_rate": 0.00019999473493094013,
-      "loss": 46.0,
-      "step": 42732
-    },
-    {
-      "epoch": 3.26723627119292,
-      "grad_norm": 0.002247408265247941,
-      "learning_rate": 0.00019999473468445896,
-      "loss": 46.0,
-      "step": 42733
-    },
-    {
-      "epoch": 3.26731272817631,
-      "grad_norm": 0.0038942538667470217,
-      "learning_rate": 0.000199994734437972,
-      "loss": 46.0,
-      "step": 42734
-    },
-    {
-      "epoch": 3.2673891851596997,
-      "grad_norm": 0.0029028940480202436,
-      "learning_rate": 0.00019999473419147928,
-      "loss": 46.0,
-      "step": 42735
-    },
-    {
-      "epoch": 3.2674656421430894,
-      "grad_norm": 0.0030094278044998646,
-      "learning_rate": 0.0001999947339449808,
-      "loss": 46.0,
-      "step": 42736
-    },
-    {
-      "epoch": 3.267542099126479,
-      "grad_norm": 0.0015174256404861808,
-      "learning_rate": 0.00019999473369847654,
-      "loss": 46.0,
-      "step": 42737
-    },
-    {
-      "epoch": 3.2676185561098685,
-      "grad_norm": 0.001444027409888804,
-      "learning_rate": 0.0001999947334519665,
-      "loss": 46.0,
-      "step": 42738
-    },
-    {
-      "epoch": 3.2676950130932583,
-      "grad_norm": 0.00046260232920758426,
-      "learning_rate": 0.00019999473320545072,
-      "loss": 46.0,
-      "step": 42739
-    },
-    {
-      "epoch": 3.267771470076648,
-      "grad_norm": 0.001998119056224823,
-      "learning_rate": 0.00019999473295892917,
-      "loss": 46.0,
-      "step": 42740
-    },
-    {
-      "epoch": 3.267847927060038,
-      "grad_norm": 0.0013875464210286736,
-      "learning_rate": 0.00019999473271240184,
-      "loss": 46.0,
-      "step": 42741
-    },
-    {
-      "epoch": 3.2679243840434276,
-      "grad_norm": 0.0012464553583413363,
-      "learning_rate": 0.00019999473246586872,
-      "loss": 46.0,
-      "step": 42742
-    },
-    {
-      "epoch": 3.2680008410268173,
-      "grad_norm": 0.0023628019262105227,
-      "learning_rate": 0.00019999473221932984,
-      "loss": 46.0,
-      "step": 42743
-    },
-    {
-      "epoch": 3.268077298010207,
-      "grad_norm": 0.007078080903738737,
-      "learning_rate": 0.00019999473197278522,
-      "loss": 46.0,
-      "step": 42744
-    },
-    {
-      "epoch": 3.268153754993597,
-      "grad_norm": 0.0017808827105909586,
-      "learning_rate": 0.00019999473172623483,
-      "loss": 46.0,
-      "step": 42745
-    },
-    {
-      "epoch": 3.2682302119769866,
-      "grad_norm": 0.0028013356495648623,
-      "learning_rate": 0.0001999947314796786,
-      "loss": 46.0,
-      "step": 42746
-    },
-    {
-      "epoch": 3.2683066689603764,
-      "grad_norm": 0.002043115207925439,
-      "learning_rate": 0.0001999947312331167,
-      "loss": 46.0,
-      "step": 42747
-    },
-    {
-      "epoch": 3.2683831259437657,
-      "grad_norm": 0.0028173988685011864,
-      "learning_rate": 0.00019999473098654896,
-      "loss": 46.0,
-      "step": 42748
-    },
-    {
-      "epoch": 3.2684595829271554,
-      "grad_norm": 0.005193495657294989,
-      "learning_rate": 0.00019999473073997547,
-      "loss": 46.0,
-      "step": 42749
-    },
-    {
-      "epoch": 3.268536039910545,
-      "grad_norm": 0.0007028752588666975,
-      "learning_rate": 0.00019999473049339624,
-      "loss": 46.0,
-      "step": 42750
-    },
-    {
-      "epoch": 3.268612496893935,
-      "grad_norm": 0.0017139507690444589,
-      "learning_rate": 0.0001999947302468112,
-      "loss": 46.0,
-      "step": 42751
-    },
-    {
-      "epoch": 3.2686889538773247,
-      "grad_norm": 0.0011305786902084947,
-      "learning_rate": 0.00019999473000022043,
-      "loss": 46.0,
-      "step": 42752
-    },
-    {
-      "epoch": 3.2687654108607145,
-      "grad_norm": 0.005871754139661789,
-      "learning_rate": 0.00019999472975362385,
-      "loss": 46.0,
-      "step": 42753
-    },
-    {
-      "epoch": 3.2688418678441042,
-      "grad_norm": 0.0008038533851504326,
-      "learning_rate": 0.00019999472950702152,
-      "loss": 46.0,
-      "step": 42754
-    },
-    {
-      "epoch": 3.268918324827494,
-      "grad_norm": 0.0012331573525443673,
-      "learning_rate": 0.0001999947292604134,
-      "loss": 46.0,
-      "step": 42755
-    },
-    {
-      "epoch": 3.2689947818108838,
-      "grad_norm": 0.0021019268315285444,
-      "learning_rate": 0.00019999472901379955,
-      "loss": 46.0,
-      "step": 42756
-    },
-    {
-      "epoch": 3.2690712387942735,
-      "grad_norm": 0.0007939172792248428,
-      "learning_rate": 0.0001999947287671799,
-      "loss": 46.0,
-      "step": 42757
-    },
-    {
-      "epoch": 3.2691476957776633,
-      "grad_norm": 0.001801857608370483,
-      "learning_rate": 0.0001999947285205545,
-      "loss": 46.0,
-      "step": 42758
-    },
-    {
-      "epoch": 3.269224152761053,
-      "grad_norm": 0.0017656749114394188,
-      "learning_rate": 0.00019999472827392331,
-      "loss": 46.0,
-      "step": 42759
-    },
-    {
-      "epoch": 3.2693006097444424,
-      "grad_norm": 0.0013188837328925729,
-      "learning_rate": 0.00019999472802728637,
-      "loss": 46.0,
-      "step": 42760
-    },
-    {
-      "epoch": 3.269377066727832,
-      "grad_norm": 0.0014837241033092141,
-      "learning_rate": 0.00019999472778064366,
-      "loss": 46.0,
-      "step": 42761
-    },
-    {
-      "epoch": 3.269453523711222,
-      "grad_norm": 0.001947373035363853,
-      "learning_rate": 0.00019999472753399518,
-      "loss": 46.0,
-      "step": 42762
-    },
-    {
-      "epoch": 3.2695299806946116,
-      "grad_norm": 0.0019918896723538637,
-      "learning_rate": 0.00019999472728734092,
-      "loss": 46.0,
-      "step": 42763
-    },
-    {
-      "epoch": 3.2696064376780014,
-      "grad_norm": 0.002146610291674733,
-      "learning_rate": 0.00019999472704068088,
-      "loss": 46.0,
-      "step": 42764
-    },
-    {
-      "epoch": 3.269682894661391,
-      "grad_norm": 0.0020708097144961357,
-      "learning_rate": 0.0001999947267940151,
-      "loss": 46.0,
-      "step": 42765
-    },
-    {
-      "epoch": 3.269759351644781,
-      "grad_norm": 0.00236993795260787,
-      "learning_rate": 0.00019999472654734355,
-      "loss": 46.0,
-      "step": 42766
-    },
-    {
-      "epoch": 3.2698358086281707,
-      "grad_norm": 0.0014317125314846635,
-      "learning_rate": 0.0001999947263006662,
-      "loss": 46.0,
-      "step": 42767
-    },
-    {
-      "epoch": 3.2699122656115605,
-      "grad_norm": 0.011325960047543049,
-      "learning_rate": 0.0001999947260539831,
-      "loss": 46.0,
-      "step": 42768
-    },
-    {
-      "epoch": 3.2699887225949498,
-      "grad_norm": 0.001926047378219664,
-      "learning_rate": 0.00019999472580729423,
-      "loss": 46.0,
-      "step": 42769
-    },
-    {
-      "epoch": 3.2700651795783395,
-      "grad_norm": 0.0015121811302378774,
-      "learning_rate": 0.0001999947255605996,
-      "loss": 46.0,
-      "step": 42770
-    },
-    {
-      "epoch": 3.2701416365617293,
-      "grad_norm": 0.0011549246264621615,
-      "learning_rate": 0.0001999947253138992,
-      "loss": 46.0,
-      "step": 42771
-    },
-    {
-      "epoch": 3.270218093545119,
-      "grad_norm": 0.0034781082067638636,
-      "learning_rate": 0.00019999472506719302,
-      "loss": 46.0,
-      "step": 42772
-    },
-    {
-      "epoch": 3.270294550528509,
-      "grad_norm": 0.004090859554708004,
-      "learning_rate": 0.00019999472482048108,
-      "loss": 46.0,
-      "step": 42773
-    },
-    {
-      "epoch": 3.2703710075118986,
-      "grad_norm": 0.0016107696574181318,
-      "learning_rate": 0.00019999472457376337,
-      "loss": 46.0,
-      "step": 42774
-    },
-    {
-      "epoch": 3.2704474644952883,
-      "grad_norm": 0.0051518771797418594,
-      "learning_rate": 0.00019999472432703989,
-      "loss": 46.0,
-      "step": 42775
-    },
-    {
-      "epoch": 3.270523921478678,
-      "grad_norm": 0.003394344123080373,
-      "learning_rate": 0.00019999472408031063,
-      "loss": 46.0,
-      "step": 42776
-    },
-    {
-      "epoch": 3.270600378462068,
-      "grad_norm": 0.002992049092426896,
-      "learning_rate": 0.0001999947238335756,
-      "loss": 46.0,
-      "step": 42777
-    },
-    {
-      "epoch": 3.2706768354454576,
-      "grad_norm": 0.0029316695872694254,
-      "learning_rate": 0.00019999472358683481,
-      "loss": 46.0,
-      "step": 42778
-    },
-    {
-      "epoch": 3.2707532924288474,
-      "grad_norm": 0.0021703671663999557,
-      "learning_rate": 0.00019999472334008824,
-      "loss": 46.0,
-      "step": 42779
-    },
-    {
-      "epoch": 3.270829749412237,
-      "grad_norm": 0.003651489969342947,
-      "learning_rate": 0.00019999472309333594,
-      "loss": 46.0,
-      "step": 42780
-    },
-    {
-      "epoch": 3.270906206395627,
-      "grad_norm": 0.0027810169849544764,
-      "learning_rate": 0.0001999947228465778,
-      "loss": 46.0,
-      "step": 42781
-    },
-    {
-      "epoch": 3.270982663379016,
-      "grad_norm": 0.0030302575323730707,
-      "learning_rate": 0.00019999472259981397,
-      "loss": 46.0,
-      "step": 42782
-    },
-    {
-      "epoch": 3.271059120362406,
-      "grad_norm": 0.001487372675910592,
-      "learning_rate": 0.00019999472235304432,
-      "loss": 46.0,
-      "step": 42783
-    },
-    {
-      "epoch": 3.2711355773457957,
-      "grad_norm": 0.0013298442354425788,
-      "learning_rate": 0.0001999947221062689,
-      "loss": 46.0,
-      "step": 42784
-    },
-    {
-      "epoch": 3.2712120343291855,
-      "grad_norm": 0.0004484287928789854,
-      "learning_rate": 0.00019999472185948774,
-      "loss": 46.0,
-      "step": 42785
-    },
-    {
-      "epoch": 3.2712884913125753,
-      "grad_norm": 0.002169565763324499,
-      "learning_rate": 0.0001999947216127008,
-      "loss": 46.0,
-      "step": 42786
-    },
-    {
-      "epoch": 3.271364948295965,
-      "grad_norm": 0.0016888410318642855,
-      "learning_rate": 0.00019999472136590808,
-      "loss": 46.0,
-      "step": 42787
-    },
-    {
-      "epoch": 3.271441405279355,
-      "grad_norm": 0.0008050664910115302,
-      "learning_rate": 0.0001999947211191096,
-      "loss": 46.0,
-      "step": 42788
-    },
-    {
-      "epoch": 3.2715178622627445,
-      "grad_norm": 0.005234782118350267,
-      "learning_rate": 0.00019999472087230534,
-      "loss": 46.0,
-      "step": 42789
-    },
-    {
-      "epoch": 3.2715943192461343,
-      "grad_norm": 0.003743865294381976,
-      "learning_rate": 0.00019999472062549533,
-      "loss": 46.0,
-      "step": 42790
-    },
-    {
-      "epoch": 3.2716707762295236,
-      "grad_norm": 0.0015106243081390858,
-      "learning_rate": 0.00019999472037867953,
-      "loss": 46.0,
-      "step": 42791
-    },
-    {
-      "epoch": 3.2717472332129134,
-      "grad_norm": 0.0017595445970073342,
-      "learning_rate": 0.00019999472013185798,
-      "loss": 46.0,
-      "step": 42792
-    },
-    {
-      "epoch": 3.271823690196303,
-      "grad_norm": 0.001041025621816516,
-      "learning_rate": 0.00019999471988503065,
-      "loss": 46.0,
-      "step": 42793
-    },
-    {
-      "epoch": 3.271900147179693,
-      "grad_norm": 0.0046062287874519825,
-      "learning_rate": 0.00019999471963819753,
-      "loss": 46.0,
-      "step": 42794
-    },
-    {
-      "epoch": 3.2719766041630827,
-      "grad_norm": 0.0010785041376948357,
-      "learning_rate": 0.00019999471939135868,
-      "loss": 46.0,
-      "step": 42795
-    },
-    {
-      "epoch": 3.2720530611464724,
-      "grad_norm": 0.0014233095571398735,
-      "learning_rate": 0.00019999471914451404,
-      "loss": 46.0,
-      "step": 42796
-    },
-    {
-      "epoch": 3.272129518129862,
-      "grad_norm": 0.0016641061520203948,
-      "learning_rate": 0.00019999471889766364,
-      "loss": 46.0,
-      "step": 42797
-    },
-    {
-      "epoch": 3.272205975113252,
-      "grad_norm": 0.004447747487574816,
-      "learning_rate": 0.00019999471865080745,
-      "loss": 46.0,
-      "step": 42798
-    },
-    {
-      "epoch": 3.2722824320966417,
-      "grad_norm": 0.0005159921711310744,
-      "learning_rate": 0.00019999471840394551,
-      "loss": 46.0,
-      "step": 42799
-    },
-    {
-      "epoch": 3.2723588890800315,
-      "grad_norm": 0.0025304376613348722,
-      "learning_rate": 0.00019999471815707783,
-      "loss": 46.0,
-      "step": 42800
-    },
-    {
-      "epoch": 3.2724353460634212,
-      "grad_norm": 0.001097582164220512,
-      "learning_rate": 0.00019999471791020432,
-      "loss": 46.0,
-      "step": 42801
-    },
-    {
-      "epoch": 3.272511803046811,
-      "grad_norm": 0.001467219553887844,
-      "learning_rate": 0.00019999471766332506,
-      "loss": 46.0,
-      "step": 42802
-    },
-    {
-      "epoch": 3.2725882600302008,
-      "grad_norm": 0.0024594292044639587,
-      "learning_rate": 0.00019999471741644005,
-      "loss": 46.0,
-      "step": 42803
-    },
-    {
-      "epoch": 3.27266471701359,
-      "grad_norm": 0.0033360703382641077,
-      "learning_rate": 0.00019999471716954928,
-      "loss": 46.0,
-      "step": 42804
-    },
-    {
-      "epoch": 3.27274117399698,
-      "grad_norm": 0.0020716842263936996,
-      "learning_rate": 0.0001999947169226527,
-      "loss": 46.0,
-      "step": 42805
-    },
-    {
-      "epoch": 3.2728176309803696,
-      "grad_norm": 0.003088155761361122,
-      "learning_rate": 0.00019999471667575037,
-      "loss": 46.0,
-      "step": 42806
-    },
-    {
-      "epoch": 3.2728940879637594,
-      "grad_norm": 0.0012038134736940265,
-      "learning_rate": 0.00019999471642884227,
-      "loss": 46.0,
-      "step": 42807
-    },
-    {
-      "epoch": 3.272970544947149,
-      "grad_norm": 0.0037069725804030895,
-      "learning_rate": 0.00019999471618192843,
-      "loss": 46.0,
-      "step": 42808
-    },
-    {
-      "epoch": 3.273047001930539,
-      "grad_norm": 0.001407202915288508,
-      "learning_rate": 0.00019999471593500878,
-      "loss": 46.0,
-      "step": 42809
-    },
-    {
-      "epoch": 3.2731234589139286,
-      "grad_norm": 0.006372368428856134,
-      "learning_rate": 0.0001999947156880834,
-      "loss": 46.0,
-      "step": 42810
-    },
-    {
-      "epoch": 3.2731999158973184,
-      "grad_norm": 0.002292105695232749,
-      "learning_rate": 0.0001999947154411522,
-      "loss": 46.0,
-      "step": 42811
-    },
-    {
-      "epoch": 3.273276372880708,
-      "grad_norm": 0.002656138502061367,
-      "learning_rate": 0.00019999471519421526,
-      "loss": 46.0,
-      "step": 42812
-    },
-    {
-      "epoch": 3.2733528298640975,
-      "grad_norm": 0.004129037726670504,
-      "learning_rate": 0.00019999471494727255,
-      "loss": 46.0,
-      "step": 42813
-    },
-    {
-      "epoch": 3.2734292868474872,
-      "grad_norm": 0.0010024503571912646,
-      "learning_rate": 0.00019999471470032407,
-      "loss": 46.0,
-      "step": 42814
-    },
-    {
-      "epoch": 3.273505743830877,
-      "grad_norm": 0.0014185491017997265,
-      "learning_rate": 0.0001999947144533698,
-      "loss": 46.0,
-      "step": 42815
-    },
-    {
-      "epoch": 3.2735822008142668,
-      "grad_norm": 0.0015643941005691886,
-      "learning_rate": 0.0001999947142064098,
-      "loss": 46.0,
-      "step": 42816
-    },
-    {
-      "epoch": 3.2736586577976565,
-      "grad_norm": 0.002996225608512759,
-      "learning_rate": 0.00019999471395944402,
-      "loss": 46.0,
-      "step": 42817
-    },
-    {
-      "epoch": 3.2737351147810463,
-      "grad_norm": 0.001086618285626173,
-      "learning_rate": 0.00019999471371247247,
-      "loss": 46.0,
-      "step": 42818
-    },
-    {
-      "epoch": 3.273811571764436,
-      "grad_norm": 0.0010692798532545567,
-      "learning_rate": 0.00019999471346549515,
-      "loss": 46.0,
-      "step": 42819
-    },
-    {
-      "epoch": 3.273888028747826,
-      "grad_norm": 0.004911097232252359,
-      "learning_rate": 0.00019999471321851205,
-      "loss": 46.0,
-      "step": 42820
-    },
-    {
-      "epoch": 3.2739644857312156,
-      "grad_norm": 0.001528770080767572,
-      "learning_rate": 0.00019999471297152318,
-      "loss": 46.0,
-      "step": 42821
-    },
-    {
-      "epoch": 3.2740409427146053,
-      "grad_norm": 0.0030985355842858553,
-      "learning_rate": 0.00019999471272452853,
-      "loss": 46.0,
-      "step": 42822
-    },
-    {
-      "epoch": 3.274117399697995,
-      "grad_norm": 0.0020407792180776596,
-      "learning_rate": 0.00019999471247752814,
-      "loss": 46.0,
-      "step": 42823
-    },
-    {
-      "epoch": 3.274193856681385,
-      "grad_norm": 0.0012985669309273362,
-      "learning_rate": 0.00019999471223052195,
-      "loss": 46.0,
-      "step": 42824
-    },
-    {
-      "epoch": 3.274270313664774,
-      "grad_norm": 0.0006892292876727879,
-      "learning_rate": 0.00019999471198351004,
-      "loss": 46.0,
-      "step": 42825
-    },
-    {
-      "epoch": 3.274346770648164,
-      "grad_norm": 0.003025229088962078,
-      "learning_rate": 0.0001999947117364923,
-      "loss": 46.0,
-      "step": 42826
-    },
-    {
-      "epoch": 3.2744232276315537,
-      "grad_norm": 0.0018115270650014281,
-      "learning_rate": 0.00019999471148946882,
-      "loss": 46.0,
-      "step": 42827
-    },
-    {
-      "epoch": 3.2744996846149435,
-      "grad_norm": 0.00709062023088336,
-      "learning_rate": 0.0001999947112424396,
-      "loss": 46.0,
-      "step": 42828
-    },
-    {
-      "epoch": 3.274576141598333,
-      "grad_norm": 0.0016407400835305452,
-      "learning_rate": 0.00019999471099540456,
-      "loss": 46.0,
-      "step": 42829
-    },
-    {
-      "epoch": 3.274652598581723,
-      "grad_norm": 0.006136324256658554,
-      "learning_rate": 0.00019999471074836378,
-      "loss": 46.0,
-      "step": 42830
-    },
-    {
-      "epoch": 3.2747290555651127,
-      "grad_norm": 0.002394117182120681,
-      "learning_rate": 0.00019999471050131723,
-      "loss": 46.0,
-      "step": 42831
-    },
-    {
-      "epoch": 3.2748055125485025,
-      "grad_norm": 0.003427214687690139,
-      "learning_rate": 0.00019999471025426488,
-      "loss": 46.0,
-      "step": 42832
-    },
-    {
-      "epoch": 3.2748819695318923,
-      "grad_norm": 0.0014828969724476337,
-      "learning_rate": 0.0001999947100072068,
-      "loss": 46.0,
-      "step": 42833
-    },
-    {
-      "epoch": 3.274958426515282,
-      "grad_norm": 0.003951503429561853,
-      "learning_rate": 0.00019999470976014293,
-      "loss": 46.0,
-      "step": 42834
-    },
-    {
-      "epoch": 3.2750348834986713,
-      "grad_norm": 0.0011830911971628666,
-      "learning_rate": 0.00019999470951307332,
-      "loss": 46.0,
-      "step": 42835
-    },
-    {
-      "epoch": 3.275111340482061,
-      "grad_norm": 0.0014231910463422537,
-      "learning_rate": 0.0001999947092659979,
-      "loss": 46.0,
-      "step": 42836
-    },
-    {
-      "epoch": 3.275187797465451,
-      "grad_norm": 0.003736478043720126,
-      "learning_rate": 0.00019999470901891674,
-      "loss": 46.0,
-      "step": 42837
-    },
-    {
-      "epoch": 3.2752642544488406,
-      "grad_norm": 0.005280000157654285,
-      "learning_rate": 0.00019999470877182977,
-      "loss": 46.0,
-      "step": 42838
-    },
-    {
-      "epoch": 3.2753407114322304,
-      "grad_norm": 0.004943507723510265,
-      "learning_rate": 0.00019999470852473706,
-      "loss": 46.0,
-      "step": 42839
-    },
-    {
-      "epoch": 3.27541716841562,
-      "grad_norm": 0.0012493549147620797,
-      "learning_rate": 0.0001999947082776386,
-      "loss": 46.0,
-      "step": 42840
-    },
-    {
-      "epoch": 3.27549362539901,
-      "grad_norm": 0.0021634967997670174,
-      "learning_rate": 0.00019999470803053435,
-      "loss": 46.0,
-      "step": 42841
-    },
-    {
-      "epoch": 3.2755700823823997,
-      "grad_norm": 0.0015853960067033768,
-      "learning_rate": 0.00019999470778342434,
-      "loss": 46.0,
-      "step": 42842
-    },
-    {
-      "epoch": 3.2756465393657894,
-      "grad_norm": 0.0019422424957156181,
-      "learning_rate": 0.00019999470753630856,
-      "loss": 46.0,
-      "step": 42843
-    },
-    {
-      "epoch": 3.275722996349179,
-      "grad_norm": 0.0022043425124138594,
-      "learning_rate": 0.000199994707289187,
-      "loss": 46.0,
-      "step": 42844
-    },
-    {
-      "epoch": 3.275799453332569,
-      "grad_norm": 0.006030660588294268,
-      "learning_rate": 0.00019999470704205966,
-      "loss": 46.0,
-      "step": 42845
-    },
-    {
-      "epoch": 3.2758759103159587,
-      "grad_norm": 0.0009263867978006601,
-      "learning_rate": 0.0001999947067949266,
-      "loss": 46.0,
-      "step": 42846
-    },
-    {
-      "epoch": 3.275952367299348,
-      "grad_norm": 0.002589545212686062,
-      "learning_rate": 0.00019999470654778772,
-      "loss": 46.0,
-      "step": 42847
-    },
-    {
-      "epoch": 3.276028824282738,
-      "grad_norm": 0.0035401920322328806,
-      "learning_rate": 0.0001999947063006431,
-      "loss": 46.0,
-      "step": 42848
-    },
-    {
-      "epoch": 3.2761052812661275,
-      "grad_norm": 0.0013544629327952862,
-      "learning_rate": 0.00019999470605349269,
-      "loss": 46.0,
-      "step": 42849
-    },
-    {
-      "epoch": 3.2761817382495173,
-      "grad_norm": 0.003758753417059779,
-      "learning_rate": 0.0001999947058063365,
-      "loss": 46.0,
-      "step": 42850
-    },
-    {
-      "epoch": 3.276258195232907,
-      "grad_norm": 0.0010497358161956072,
-      "learning_rate": 0.00019999470555917456,
-      "loss": 46.0,
-      "step": 42851
-    },
-    {
-      "epoch": 3.276334652216297,
-      "grad_norm": 0.0034277932718396187,
-      "learning_rate": 0.00019999470531200685,
-      "loss": 46.0,
-      "step": 42852
-    },
-    {
-      "epoch": 3.2764111091996866,
-      "grad_norm": 0.005429920740425587,
-      "learning_rate": 0.00019999470506483336,
-      "loss": 46.0,
-      "step": 42853
-    },
-    {
-      "epoch": 3.2764875661830763,
-      "grad_norm": 0.003476017387583852,
-      "learning_rate": 0.00019999470481765413,
-      "loss": 46.0,
-      "step": 42854
-    },
-    {
-      "epoch": 3.276564023166466,
-      "grad_norm": 0.0038299558218568563,
-      "learning_rate": 0.00019999470457046913,
-      "loss": 46.0,
-      "step": 42855
-    },
-    {
-      "epoch": 3.276640480149856,
-      "grad_norm": 0.004274279344826937,
-      "learning_rate": 0.00019999470432327833,
-      "loss": 46.0,
-      "step": 42856
-    },
-    {
-      "epoch": 3.276716937133245,
-      "grad_norm": 0.0011316942982375622,
-      "learning_rate": 0.00019999470407608178,
-      "loss": 46.0,
-      "step": 42857
-    },
-    {
-      "epoch": 3.276793394116635,
-      "grad_norm": 0.001200154540129006,
-      "learning_rate": 0.00019999470382887945,
-      "loss": 46.0,
-      "step": 42858
-    },
-    {
-      "epoch": 3.2768698511000247,
-      "grad_norm": 0.0007008964894339442,
-      "learning_rate": 0.00019999470358167138,
-      "loss": 46.0,
-      "step": 42859
-    },
-    {
-      "epoch": 3.2769463080834145,
-      "grad_norm": 0.0013684829464182258,
-      "learning_rate": 0.0001999947033344575,
-      "loss": 46.0,
-      "step": 42860
-    },
-    {
-      "epoch": 3.2770227650668042,
-      "grad_norm": 0.00510263629257679,
-      "learning_rate": 0.00019999470308723787,
-      "loss": 46.0,
-      "step": 42861
-    },
-    {
-      "epoch": 3.277099222050194,
-      "grad_norm": 0.00334531394764781,
-      "learning_rate": 0.00019999470284001248,
-      "loss": 46.0,
-      "step": 42862
-    },
-    {
-      "epoch": 3.2771756790335838,
-      "grad_norm": 0.0015518802683800459,
-      "learning_rate": 0.00019999470259278131,
-      "loss": 46.0,
-      "step": 42863
-    },
-    {
-      "epoch": 3.2772521360169735,
-      "grad_norm": 0.0011197567218914628,
-      "learning_rate": 0.00019999470234554435,
-      "loss": 46.0,
-      "step": 42864
-    },
-    {
-      "epoch": 3.2773285930003633,
-      "grad_norm": 0.003087131306529045,
-      "learning_rate": 0.00019999470209830167,
-      "loss": 46.0,
-      "step": 42865
-    },
-    {
-      "epoch": 3.277405049983753,
-      "grad_norm": 0.0024691878352314234,
-      "learning_rate": 0.0001999947018510532,
-      "loss": 46.0,
-      "step": 42866
-    },
-    {
-      "epoch": 3.277481506967143,
-      "grad_norm": 0.002816775580868125,
-      "learning_rate": 0.00019999470160379893,
-      "loss": 46.0,
-      "step": 42867
-    },
-    {
-      "epoch": 3.2775579639505326,
-      "grad_norm": 0.001735563506372273,
-      "learning_rate": 0.00019999470135653892,
-      "loss": 46.0,
-      "step": 42868
-    },
-    {
-      "epoch": 3.277634420933922,
-      "grad_norm": 0.0023912061005830765,
-      "learning_rate": 0.00019999470110927315,
-      "loss": 46.0,
-      "step": 42869
-    },
-    {
-      "epoch": 3.2777108779173116,
-      "grad_norm": 0.002525065094232559,
-      "learning_rate": 0.0001999947008620016,
-      "loss": 46.0,
-      "step": 42870
-    },
-    {
-      "epoch": 3.2777873349007014,
-      "grad_norm": 0.0013745236210525036,
-      "learning_rate": 0.00019999470061472427,
-      "loss": 46.0,
-      "step": 42871
-    },
-    {
-      "epoch": 3.277863791884091,
-      "grad_norm": 0.0037984021473675966,
-      "learning_rate": 0.00019999470036744118,
-      "loss": 46.0,
-      "step": 42872
-    },
-    {
-      "epoch": 3.277940248867481,
-      "grad_norm": 0.0017467513680458069,
-      "learning_rate": 0.0001999947001201523,
-      "loss": 46.0,
-      "step": 42873
-    },
-    {
-      "epoch": 3.2780167058508707,
-      "grad_norm": 0.00536099448800087,
-      "learning_rate": 0.0001999946998728577,
-      "loss": 46.0,
-      "step": 42874
-    },
-    {
-      "epoch": 3.2780931628342604,
-      "grad_norm": 0.0040005166083574295,
-      "learning_rate": 0.00019999469962555727,
-      "loss": 46.0,
-      "step": 42875
-    },
-    {
-      "epoch": 3.27816961981765,
-      "grad_norm": 0.00365635147318244,
-      "learning_rate": 0.00019999469937825114,
-      "loss": 46.0,
-      "step": 42876
-    },
-    {
-      "epoch": 3.27824607680104,
-      "grad_norm": 0.0006848968914709985,
-      "learning_rate": 0.0001999946991309392,
-      "loss": 46.0,
-      "step": 42877
-    },
-    {
-      "epoch": 3.2783225337844297,
-      "grad_norm": 0.006137051619589329,
-      "learning_rate": 0.0001999946988836215,
-      "loss": 46.0,
-      "step": 42878
-    },
-    {
-      "epoch": 3.278398990767819,
-      "grad_norm": 0.0027530849911272526,
-      "learning_rate": 0.000199994698636298,
-      "loss": 46.0,
-      "step": 42879
-    },
-    {
-      "epoch": 3.278475447751209,
-      "grad_norm": 0.006251274608075619,
-      "learning_rate": 0.00019999469838896875,
-      "loss": 46.0,
-      "step": 42880
-    },
-    {
-      "epoch": 3.2785519047345986,
-      "grad_norm": 0.0039061664137989283,
-      "learning_rate": 0.00019999469814163375,
-      "loss": 46.0,
-      "step": 42881
-    },
-    {
-      "epoch": 3.2786283617179883,
-      "grad_norm": 0.0020890890154987574,
-      "learning_rate": 0.00019999469789429294,
-      "loss": 46.0,
-      "step": 42882
-    },
-    {
-      "epoch": 3.278704818701378,
-      "grad_norm": 0.0018786712316796184,
-      "learning_rate": 0.0001999946976469464,
-      "loss": 46.0,
-      "step": 42883
-    },
-    {
-      "epoch": 3.278781275684768,
-      "grad_norm": 0.002287887968122959,
-      "learning_rate": 0.0001999946973995941,
-      "loss": 46.0,
-      "step": 42884
-    },
-    {
-      "epoch": 3.2788577326681576,
-      "grad_norm": 0.001223283470608294,
-      "learning_rate": 0.000199994697152236,
-      "loss": 46.0,
-      "step": 42885
-    },
-    {
-      "epoch": 3.2789341896515474,
-      "grad_norm": 0.0014291051775217056,
-      "learning_rate": 0.00019999469690487213,
-      "loss": 46.0,
-      "step": 42886
-    },
-    {
-      "epoch": 3.279010646634937,
-      "grad_norm": 0.004400795791298151,
-      "learning_rate": 0.0001999946966575025,
-      "loss": 46.0,
-      "step": 42887
-    },
-    {
-      "epoch": 3.279087103618327,
-      "grad_norm": 0.004692132119089365,
-      "learning_rate": 0.00019999469641012713,
-      "loss": 46.0,
-      "step": 42888
-    },
-    {
-      "epoch": 3.2791635606017167,
-      "grad_norm": 0.0005094028310850263,
-      "learning_rate": 0.00019999469616274596,
-      "loss": 46.0,
-      "step": 42889
-    },
-    {
-      "epoch": 3.2792400175851064,
-      "grad_norm": 0.0022600230295211077,
-      "learning_rate": 0.000199994695915359,
-      "loss": 46.0,
-      "step": 42890
-    },
-    {
-      "epoch": 3.2793164745684957,
-      "grad_norm": 0.0023688492365181446,
-      "learning_rate": 0.00019999469566796632,
-      "loss": 46.0,
-      "step": 42891
-    },
-    {
-      "epoch": 3.2793929315518855,
-      "grad_norm": 0.0026851186994463205,
-      "learning_rate": 0.00019999469542056783,
-      "loss": 46.0,
-      "step": 42892
-    },
-    {
-      "epoch": 3.2794693885352753,
-      "grad_norm": 0.0016948075499385595,
-      "learning_rate": 0.00019999469517316358,
-      "loss": 46.0,
-      "step": 42893
-    },
-    {
-      "epoch": 3.279545845518665,
-      "grad_norm": 0.0008688248344697058,
-      "learning_rate": 0.00019999469492575358,
-      "loss": 46.0,
-      "step": 42894
-    },
-    {
-      "epoch": 3.2796223025020548,
-      "grad_norm": 0.004782327450811863,
-      "learning_rate": 0.0001999946946783378,
-      "loss": 46.0,
-      "step": 42895
-    },
-    {
-      "epoch": 3.2796987594854445,
-      "grad_norm": 0.00345169217325747,
-      "learning_rate": 0.00019999469443091625,
-      "loss": 46.0,
-      "step": 42896
-    },
-    {
-      "epoch": 3.2797752164688343,
-      "grad_norm": 0.0027687230613082647,
-      "learning_rate": 0.00019999469418348893,
-      "loss": 46.0,
-      "step": 42897
-    },
-    {
-      "epoch": 3.279851673452224,
-      "grad_norm": 0.005968145560473204,
-      "learning_rate": 0.00019999469393605583,
-      "loss": 46.0,
-      "step": 42898
-    },
-    {
-      "epoch": 3.279928130435614,
-      "grad_norm": 0.004532267339527607,
-      "learning_rate": 0.00019999469368861696,
-      "loss": 46.0,
-      "step": 42899
-    },
-    {
-      "epoch": 3.280004587419003,
-      "grad_norm": 0.002928745234385133,
-      "learning_rate": 0.00019999469344117237,
-      "loss": 46.0,
-      "step": 42900
-    },
-    {
-      "epoch": 3.280081044402393,
-      "grad_norm": 0.003344410564750433,
-      "learning_rate": 0.00019999469319372196,
-      "loss": 46.0,
-      "step": 42901
-    },
-    {
-      "epoch": 3.2801575013857827,
-      "grad_norm": 0.002041700528934598,
-      "learning_rate": 0.0001999946929462658,
-      "loss": 46.0,
-      "step": 42902
-    },
-    {
-      "epoch": 3.2802339583691724,
-      "grad_norm": 0.0010236079106107354,
-      "learning_rate": 0.00019999469269880383,
-      "loss": 46.0,
-      "step": 42903
-    },
-    {
-      "epoch": 3.280310415352562,
-      "grad_norm": 0.0013015087461099029,
-      "learning_rate": 0.00019999469245133615,
-      "loss": 46.0,
-      "step": 42904
-    },
-    {
-      "epoch": 3.280386872335952,
-      "grad_norm": 0.0018849613843485713,
-      "learning_rate": 0.0001999946922038627,
-      "loss": 46.0,
-      "step": 42905
-    },
-    {
-      "epoch": 3.2804633293193417,
-      "grad_norm": 0.0018751038005575538,
-      "learning_rate": 0.00019999469195638344,
-      "loss": 46.0,
-      "step": 42906
-    },
-    {
-      "epoch": 3.2805397863027315,
-      "grad_norm": 0.002517502522096038,
-      "learning_rate": 0.0001999946917088984,
-      "loss": 46.0,
-      "step": 42907
-    },
-    {
-      "epoch": 3.2806162432861212,
-      "grad_norm": 0.002895353129133582,
-      "learning_rate": 0.00019999469146140766,
-      "loss": 46.0,
-      "step": 42908
-    },
-    {
-      "epoch": 3.280692700269511,
-      "grad_norm": 0.0019689404871314764,
-      "learning_rate": 0.0001999946912139111,
-      "loss": 46.0,
-      "step": 42909
-    },
-    {
-      "epoch": 3.2807691572529007,
-      "grad_norm": 0.0013360092416405678,
-      "learning_rate": 0.00019999469096640876,
-      "loss": 46.0,
-      "step": 42910
-    },
-    {
-      "epoch": 3.2808456142362905,
-      "grad_norm": 0.002089686458930373,
-      "learning_rate": 0.00019999469071890066,
-      "loss": 46.0,
-      "step": 42911
-    },
-    {
-      "epoch": 3.2809220712196803,
-      "grad_norm": 0.0035643037408590317,
-      "learning_rate": 0.00019999469047138685,
-      "loss": 46.0,
-      "step": 42912
-    },
-    {
-      "epoch": 3.2809985282030696,
-      "grad_norm": 0.004932517185807228,
-      "learning_rate": 0.00019999469022386718,
-      "loss": 46.0,
-      "step": 42913
-    },
-    {
-      "epoch": 3.2810749851864593,
-      "grad_norm": 0.003204831387847662,
-      "learning_rate": 0.00019999468997634182,
-      "loss": 46.0,
-      "step": 42914
-    },
-    {
-      "epoch": 3.281151442169849,
-      "grad_norm": 0.00418286956846714,
-      "learning_rate": 0.00019999468972881063,
-      "loss": 46.0,
-      "step": 42915
-    },
-    {
-      "epoch": 3.281227899153239,
-      "grad_norm": 0.0016691904747858644,
-      "learning_rate": 0.00019999468948127372,
-      "loss": 46.0,
-      "step": 42916
-    },
-    {
-      "epoch": 3.2813043561366286,
-      "grad_norm": 0.002681855810806155,
-      "learning_rate": 0.00019999468923373098,
-      "loss": 46.0,
-      "step": 42917
-    },
-    {
-      "epoch": 3.2813808131200184,
-      "grad_norm": 0.0023132087662816048,
-      "learning_rate": 0.00019999468898618253,
-      "loss": 46.0,
-      "step": 42918
-    },
-    {
-      "epoch": 3.281457270103408,
-      "grad_norm": 0.0029189232736825943,
-      "learning_rate": 0.00019999468873862827,
-      "loss": 46.0,
-      "step": 42919
-    },
-    {
-      "epoch": 3.281533727086798,
-      "grad_norm": 0.0006030069198459387,
-      "learning_rate": 0.00019999468849106827,
-      "loss": 46.0,
-      "step": 42920
-    },
-    {
-      "epoch": 3.2816101840701877,
-      "grad_norm": 0.002316718455404043,
-      "learning_rate": 0.00019999468824350252,
-      "loss": 46.0,
-      "step": 42921
-    },
-    {
-      "epoch": 3.281686641053577,
-      "grad_norm": 0.0012835246743634343,
-      "learning_rate": 0.00019999468799593095,
-      "loss": 46.0,
-      "step": 42922
-    },
-    {
-      "epoch": 3.2817630980369668,
-      "grad_norm": 0.002023693872615695,
-      "learning_rate": 0.00019999468774835362,
-      "loss": 46.0,
-      "step": 42923
-    },
-    {
-      "epoch": 3.2818395550203565,
-      "grad_norm": 0.011228973045945168,
-      "learning_rate": 0.00019999468750077056,
-      "loss": 46.0,
-      "step": 42924
-    },
-    {
-      "epoch": 3.2819160120037463,
-      "grad_norm": 0.003020940348505974,
-      "learning_rate": 0.0001999946872531817,
-      "loss": 46.0,
-      "step": 42925
-    },
-    {
-      "epoch": 3.281992468987136,
-      "grad_norm": 0.0020151061471551657,
-      "learning_rate": 0.00019999468700558705,
-      "loss": 46.0,
-      "step": 42926
-    },
-    {
-      "epoch": 3.282068925970526,
-      "grad_norm": 0.005133095663040876,
-      "learning_rate": 0.00019999468675798668,
-      "loss": 46.0,
-      "step": 42927
-    },
-    {
-      "epoch": 3.2821453829539156,
-      "grad_norm": 0.0049695041961967945,
-      "learning_rate": 0.00019999468651038052,
-      "loss": 46.0,
-      "step": 42928
-    },
-    {
-      "epoch": 3.2822218399373053,
-      "grad_norm": 0.0011481387773528695,
-      "learning_rate": 0.0001999946862627686,
-      "loss": 46.0,
-      "step": 42929
-    },
-    {
-      "epoch": 3.282298296920695,
-      "grad_norm": 0.0010705229360610247,
-      "learning_rate": 0.00019999468601515088,
-      "loss": 46.0,
-      "step": 42930
-    },
-    {
-      "epoch": 3.282374753904085,
-      "grad_norm": 0.0011284570209681988,
-      "learning_rate": 0.0001999946857675274,
-      "loss": 46.0,
-      "step": 42931
-    },
-    {
-      "epoch": 3.2824512108874746,
-      "grad_norm": 0.0024002110585570335,
-      "learning_rate": 0.00019999468551989814,
-      "loss": 46.0,
-      "step": 42932
-    },
-    {
-      "epoch": 3.2825276678708644,
-      "grad_norm": 0.0017078750533983111,
-      "learning_rate": 0.00019999468527226314,
-      "loss": 46.0,
-      "step": 42933
-    },
-    {
-      "epoch": 3.282604124854254,
-      "grad_norm": 0.0032187087927013636,
-      "learning_rate": 0.00019999468502462236,
-      "loss": 46.0,
-      "step": 42934
-    },
-    {
-      "epoch": 3.2826805818376434,
-      "grad_norm": 0.0019785864278674126,
-      "learning_rate": 0.00019999468477697582,
-      "loss": 46.0,
-      "step": 42935
-    },
-    {
-      "epoch": 3.282757038821033,
-      "grad_norm": 0.005571179557591677,
-      "learning_rate": 0.0001999946845293235,
-      "loss": 46.0,
-      "step": 42936
-    },
-    {
-      "epoch": 3.282833495804423,
-      "grad_norm": 0.002684182720258832,
-      "learning_rate": 0.00019999468428166543,
-      "loss": 46.0,
-      "step": 42937
-    },
-    {
-      "epoch": 3.2829099527878127,
-      "grad_norm": 0.002035718411207199,
-      "learning_rate": 0.00019999468403400156,
-      "loss": 46.0,
-      "step": 42938
-    },
-    {
-      "epoch": 3.2829864097712025,
-      "grad_norm": 0.0018606202211230993,
-      "learning_rate": 0.00019999468378633194,
-      "loss": 46.0,
-      "step": 42939
-    },
-    {
-      "epoch": 3.2830628667545922,
-      "grad_norm": 0.0024310534354299307,
-      "learning_rate": 0.00019999468353865653,
-      "loss": 46.0,
-      "step": 42940
-    },
-    {
-      "epoch": 3.283139323737982,
-      "grad_norm": 0.004504888318479061,
-      "learning_rate": 0.0001999946832909754,
-      "loss": 46.0,
-      "step": 42941
-    },
-    {
-      "epoch": 3.2832157807213718,
-      "grad_norm": 0.003242659382522106,
-      "learning_rate": 0.00019999468304328846,
-      "loss": 46.0,
-      "step": 42942
-    },
-    {
-      "epoch": 3.2832922377047615,
-      "grad_norm": 0.0030412704218178988,
-      "learning_rate": 0.00019999468279559578,
-      "loss": 46.0,
-      "step": 42943
-    },
-    {
-      "epoch": 3.283368694688151,
-      "grad_norm": 0.0030782849062234163,
-      "learning_rate": 0.0001999946825478973,
-      "loss": 46.0,
-      "step": 42944
-    },
-    {
-      "epoch": 3.2834451516715406,
-      "grad_norm": 0.0017058249795809388,
-      "learning_rate": 0.00019999468230019304,
-      "loss": 46.0,
-      "step": 42945
-    },
-    {
-      "epoch": 3.2835216086549304,
-      "grad_norm": 0.0014666904462501407,
-      "learning_rate": 0.00019999468205248304,
-      "loss": 46.0,
-      "step": 42946
-    },
-    {
-      "epoch": 3.28359806563832,
-      "grad_norm": 0.001175355981104076,
-      "learning_rate": 0.00019999468180476726,
-      "loss": 46.0,
-      "step": 42947
-    },
-    {
-      "epoch": 3.28367452262171,
-      "grad_norm": 0.003088381141424179,
-      "learning_rate": 0.00019999468155704571,
-      "loss": 46.0,
-      "step": 42948
-    },
-    {
-      "epoch": 3.2837509796050997,
-      "grad_norm": 0.0030383828561753035,
-      "learning_rate": 0.0001999946813093184,
-      "loss": 46.0,
-      "step": 42949
-    },
-    {
-      "epoch": 3.2838274365884894,
-      "grad_norm": 0.001755392411723733,
-      "learning_rate": 0.00019999468106158533,
-      "loss": 46.0,
-      "step": 42950
-    },
-    {
-      "epoch": 3.283903893571879,
-      "grad_norm": 0.001585042686201632,
-      "learning_rate": 0.00019999468081384646,
-      "loss": 46.0,
-      "step": 42951
-    },
-    {
-      "epoch": 3.283980350555269,
-      "grad_norm": 0.005337897222489119,
-      "learning_rate": 0.00019999468056610184,
-      "loss": 46.0,
-      "step": 42952
-    },
-    {
-      "epoch": 3.2840568075386587,
-      "grad_norm": 0.0006694708717986941,
-      "learning_rate": 0.00019999468031835145,
-      "loss": 46.0,
-      "step": 42953
-    },
-    {
-      "epoch": 3.2841332645220485,
-      "grad_norm": 0.001618560403585434,
-      "learning_rate": 0.00019999468007059527,
-      "loss": 46.0,
-      "step": 42954
-    },
-    {
-      "epoch": 3.284209721505438,
-      "grad_norm": 0.002989501226693392,
-      "learning_rate": 0.00019999467982283336,
-      "loss": 46.0,
-      "step": 42955
-    },
-    {
-      "epoch": 3.2842861784888275,
-      "grad_norm": 0.006691244896501303,
-      "learning_rate": 0.00019999467957506565,
-      "loss": 46.0,
-      "step": 42956
-    },
-    {
-      "epoch": 3.2843626354722173,
-      "grad_norm": 0.0016134664183482528,
-      "learning_rate": 0.0001999946793272922,
-      "loss": 46.0,
-      "step": 42957
-    },
-    {
-      "epoch": 3.284439092455607,
-      "grad_norm": 0.0011493315687403083,
-      "learning_rate": 0.00019999467907951294,
-      "loss": 46.0,
-      "step": 42958
-    },
-    {
-      "epoch": 3.284515549438997,
-      "grad_norm": 0.0014053423656150699,
-      "learning_rate": 0.00019999467883172797,
-      "loss": 46.0,
-      "step": 42959
-    },
-    {
-      "epoch": 3.2845920064223866,
-      "grad_norm": 0.0020455457270145416,
-      "learning_rate": 0.00019999467858393717,
-      "loss": 46.0,
-      "step": 42960
-    },
-    {
-      "epoch": 3.2846684634057763,
-      "grad_norm": 0.004131302237510681,
-      "learning_rate": 0.00019999467833614062,
-      "loss": 46.0,
-      "step": 42961
-    },
-    {
-      "epoch": 3.284744920389166,
-      "grad_norm": 0.0031100099440664053,
-      "learning_rate": 0.0001999946780883383,
-      "loss": 46.0,
-      "step": 42962
-    },
-    {
-      "epoch": 3.284821377372556,
-      "grad_norm": 0.0022978046908974648,
-      "learning_rate": 0.0001999946778405302,
-      "loss": 46.0,
-      "step": 42963
-    },
-    {
-      "epoch": 3.2848978343559456,
-      "grad_norm": 0.0024684544187039137,
-      "learning_rate": 0.00019999467759271636,
-      "loss": 46.0,
-      "step": 42964
-    },
-    {
-      "epoch": 3.2849742913393354,
-      "grad_norm": 0.0068963258527219296,
-      "learning_rate": 0.00019999467734489675,
-      "loss": 46.0,
-      "step": 42965
-    },
-    {
-      "epoch": 3.2850507483227247,
-      "grad_norm": 0.0011998926056548953,
-      "learning_rate": 0.00019999467709707136,
-      "loss": 46.0,
-      "step": 42966
-    },
-    {
-      "epoch": 3.2851272053061145,
-      "grad_norm": 0.003674176288768649,
-      "learning_rate": 0.00019999467684924017,
-      "loss": 46.0,
-      "step": 42967
-    },
-    {
-      "epoch": 3.2852036622895042,
-      "grad_norm": 0.0021530287340283394,
-      "learning_rate": 0.00019999467660140326,
-      "loss": 46.0,
-      "step": 42968
-    },
-    {
-      "epoch": 3.285280119272894,
-      "grad_norm": 0.004675593227148056,
-      "learning_rate": 0.00019999467635356056,
-      "loss": 46.0,
-      "step": 42969
-    },
-    {
-      "epoch": 3.2853565762562837,
-      "grad_norm": 0.0016175052151083946,
-      "learning_rate": 0.0001999946761057121,
-      "loss": 46.0,
-      "step": 42970
-    },
-    {
-      "epoch": 3.2854330332396735,
-      "grad_norm": 0.0014991143252700567,
-      "learning_rate": 0.00019999467585785788,
-      "loss": 46.0,
-      "step": 42971
-    },
-    {
-      "epoch": 3.2855094902230633,
-      "grad_norm": 0.0006681408267468214,
-      "learning_rate": 0.00019999467560999785,
-      "loss": 46.0,
-      "step": 42972
-    },
-    {
-      "epoch": 3.285585947206453,
-      "grad_norm": 0.0006874622777104378,
-      "learning_rate": 0.00019999467536213207,
-      "loss": 46.0,
-      "step": 42973
-    },
-    {
-      "epoch": 3.285662404189843,
-      "grad_norm": 0.0009981076000258327,
-      "learning_rate": 0.00019999467511426053,
-      "loss": 46.0,
-      "step": 42974
-    },
-    {
-      "epoch": 3.2857388611732326,
-      "grad_norm": 0.002620001556351781,
-      "learning_rate": 0.00019999467486638323,
-      "loss": 46.0,
-      "step": 42975
-    },
-    {
-      "epoch": 3.2858153181566223,
-      "grad_norm": 0.0022467144299298525,
-      "learning_rate": 0.00019999467461850014,
-      "loss": 46.0,
-      "step": 42976
-    },
-    {
-      "epoch": 3.285891775140012,
-      "grad_norm": 0.0023586770985275507,
-      "learning_rate": 0.00019999467437061127,
-      "loss": 46.0,
-      "step": 42977
-    },
-    {
-      "epoch": 3.2859682321234014,
-      "grad_norm": 0.0022540080826729536,
-      "learning_rate": 0.00019999467412271666,
-      "loss": 46.0,
-      "step": 42978
-    },
-    {
-      "epoch": 3.286044689106791,
-      "grad_norm": 0.0036533037200570107,
-      "learning_rate": 0.00019999467387481627,
-      "loss": 46.0,
-      "step": 42979
-    },
-    {
-      "epoch": 3.286121146090181,
-      "grad_norm": 0.0020512158516794443,
-      "learning_rate": 0.00019999467362691014,
-      "loss": 46.0,
-      "step": 42980
-    },
-    {
-      "epoch": 3.2861976030735707,
-      "grad_norm": 0.0007933160522952676,
-      "learning_rate": 0.00019999467337899817,
-      "loss": 46.0,
-      "step": 42981
-    },
-    {
-      "epoch": 3.2862740600569604,
-      "grad_norm": 0.004828393924981356,
-      "learning_rate": 0.0001999946731310805,
-      "loss": 46.0,
-      "step": 42982
-    },
-    {
-      "epoch": 3.28635051704035,
-      "grad_norm": 0.0024523879401385784,
-      "learning_rate": 0.00019999467288315704,
-      "loss": 46.0,
-      "step": 42983
-    },
-    {
-      "epoch": 3.28642697402374,
-      "grad_norm": 0.0038373444695025682,
-      "learning_rate": 0.00019999467263522779,
-      "loss": 46.0,
-      "step": 42984
-    },
-    {
-      "epoch": 3.2865034310071297,
-      "grad_norm": 0.0010139079531654716,
-      "learning_rate": 0.00019999467238729279,
-      "loss": 46.0,
-      "step": 42985
-    },
-    {
-      "epoch": 3.2865798879905195,
-      "grad_norm": 0.0021915417164564133,
-      "learning_rate": 0.000199994672139352,
-      "loss": 46.0,
-      "step": 42986
-    },
-    {
-      "epoch": 3.2866563449739092,
-      "grad_norm": 0.0013777019921690226,
-      "learning_rate": 0.00019999467189140546,
-      "loss": 46.0,
-      "step": 42987
-    },
-    {
-      "epoch": 3.2867328019572986,
-      "grad_norm": 0.0024014387745410204,
-      "learning_rate": 0.00019999467164345314,
-      "loss": 46.0,
-      "step": 42988
-    },
-    {
-      "epoch": 3.2868092589406883,
-      "grad_norm": 0.0014577845577150583,
-      "learning_rate": 0.00019999467139549508,
-      "loss": 46.0,
-      "step": 42989
-    },
-    {
-      "epoch": 3.286885715924078,
-      "grad_norm": 0.0010902235517278314,
-      "learning_rate": 0.0001999946711475312,
-      "loss": 46.0,
-      "step": 42990
-    },
-    {
-      "epoch": 3.286962172907468,
-      "grad_norm": 0.016242684796452522,
-      "learning_rate": 0.0001999946708995616,
-      "loss": 46.0,
-      "step": 42991
-    },
-    {
-      "epoch": 3.2870386298908576,
-      "grad_norm": 0.006492890417575836,
-      "learning_rate": 0.0001999946706515862,
-      "loss": 46.0,
-      "step": 42992
-    },
-    {
-      "epoch": 3.2871150868742474,
-      "grad_norm": 0.0013405898353084922,
-      "learning_rate": 0.00019999467040360505,
-      "loss": 46.0,
-      "step": 42993
-    },
-    {
-      "epoch": 3.287191543857637,
-      "grad_norm": 0.00208151969127357,
-      "learning_rate": 0.00019999467015561812,
-      "loss": 46.0,
-      "step": 42994
-    },
-    {
-      "epoch": 3.287268000841027,
-      "grad_norm": 0.0007782382308505476,
-      "learning_rate": 0.00019999466990762544,
-      "loss": 46.0,
-      "step": 42995
-    },
-    {
-      "epoch": 3.2873444578244166,
-      "grad_norm": 0.00380100985057652,
-      "learning_rate": 0.00019999466965962695,
-      "loss": 46.0,
-      "step": 42996
-    },
-    {
-      "epoch": 3.2874209148078064,
-      "grad_norm": 0.0020033137407153845,
-      "learning_rate": 0.00019999466941162273,
-      "loss": 46.0,
-      "step": 42997
-    },
-    {
-      "epoch": 3.287497371791196,
-      "grad_norm": 0.0029079923406243324,
-      "learning_rate": 0.0001999946691636127,
-      "loss": 46.0,
-      "step": 42998
-    },
-    {
-      "epoch": 3.287573828774586,
-      "grad_norm": 0.002057442208752036,
-      "learning_rate": 0.00019999466891559695,
-      "loss": 46.0,
-      "step": 42999
-    },
-    {
-      "epoch": 3.2876502857579752,
-      "grad_norm": 0.0053220042027533054,
-      "learning_rate": 0.0001999946686675754,
-      "loss": 46.0,
-      "step": 43000
-    },
-    {
-      "epoch": 3.287726742741365,
-      "grad_norm": 0.0024718421045690775,
-      "learning_rate": 0.00019999466841954811,
-      "loss": 46.0,
-      "step": 43001
-    },
-    {
-      "epoch": 3.2878031997247548,
-      "grad_norm": 0.0020546005107462406,
-      "learning_rate": 0.00019999466817151502,
-      "loss": 46.0,
-      "step": 43002
-    },
-    {
-      "epoch": 3.2878796567081445,
-      "grad_norm": 0.0019832036923617125,
-      "learning_rate": 0.00019999466792347618,
-      "loss": 46.0,
-      "step": 43003
-    },
-    {
-      "epoch": 3.2879561136915343,
-      "grad_norm": 0.0011489823227748275,
-      "learning_rate": 0.00019999466767543157,
-      "loss": 46.0,
-      "step": 43004
-    },
-    {
-      "epoch": 3.288032570674924,
-      "grad_norm": 0.00465560844168067,
-      "learning_rate": 0.00019999466742738115,
-      "loss": 46.0,
-      "step": 43005
-    },
-    {
-      "epoch": 3.288109027658314,
-      "grad_norm": 0.0016144949477165937,
-      "learning_rate": 0.000199994667179325,
-      "loss": 46.0,
-      "step": 43006
-    },
-    {
-      "epoch": 3.2881854846417036,
-      "grad_norm": 0.000845888105686754,
-      "learning_rate": 0.0001999946669312631,
-      "loss": 46.0,
-      "step": 43007
-    },
-    {
-      "epoch": 3.2882619416250933,
-      "grad_norm": 0.001992309233173728,
-      "learning_rate": 0.00019999466668319538,
-      "loss": 46.0,
-      "step": 43008
-    },
-    {
-      "epoch": 3.288338398608483,
-      "grad_norm": 0.0037813507951796055,
-      "learning_rate": 0.00019999466643512193,
-      "loss": 46.0,
-      "step": 43009
-    },
-    {
-      "epoch": 3.2884148555918724,
-      "grad_norm": 0.002345156157389283,
-      "learning_rate": 0.0001999946661870427,
-      "loss": 46.0,
-      "step": 43010
-    },
-    {
-      "epoch": 3.288491312575262,
-      "grad_norm": 0.0035937039647251368,
-      "learning_rate": 0.00019999466593895767,
-      "loss": 46.0,
-      "step": 43011
-    },
-    {
-      "epoch": 3.288567769558652,
-      "grad_norm": 0.0035123745910823345,
-      "learning_rate": 0.0001999946656908669,
-      "loss": 46.0,
-      "step": 43012
-    },
-    {
-      "epoch": 3.2886442265420417,
-      "grad_norm": 0.0006592206191271544,
-      "learning_rate": 0.00019999466544277035,
-      "loss": 46.0,
-      "step": 43013
-    },
-    {
-      "epoch": 3.2887206835254315,
-      "grad_norm": 0.001796973985619843,
-      "learning_rate": 0.00019999466519466806,
-      "loss": 46.0,
-      "step": 43014
-    },
-    {
-      "epoch": 3.288797140508821,
-      "grad_norm": 0.0012087718350812793,
-      "learning_rate": 0.00019999466494656,
-      "loss": 46.0,
-      "step": 43015
-    },
-    {
-      "epoch": 3.288873597492211,
-      "grad_norm": 0.0029219805728644133,
-      "learning_rate": 0.00019999466469844613,
-      "loss": 46.0,
-      "step": 43016
-    },
-    {
-      "epoch": 3.2889500544756007,
-      "grad_norm": 0.0066736540757119656,
-      "learning_rate": 0.00019999466445032652,
-      "loss": 46.0,
-      "step": 43017
-    },
-    {
-      "epoch": 3.2890265114589905,
-      "grad_norm": 0.0018083583563566208,
-      "learning_rate": 0.00019999466420220113,
-      "loss": 46.0,
-      "step": 43018
-    },
-    {
-      "epoch": 3.2891029684423803,
-      "grad_norm": 0.0044660489074885845,
-      "learning_rate": 0.00019999466395406997,
-      "loss": 46.0,
-      "step": 43019
-    },
-    {
-      "epoch": 3.28917942542577,
-      "grad_norm": 0.0007229259354062378,
-      "learning_rate": 0.00019999466370593304,
-      "loss": 46.0,
-      "step": 43020
-    },
-    {
-      "epoch": 3.28925588240916,
-      "grad_norm": 0.0012154581490904093,
-      "learning_rate": 0.00019999466345779036,
-      "loss": 46.0,
-      "step": 43021
-    },
-    {
-      "epoch": 3.289332339392549,
-      "grad_norm": 0.0018211901187896729,
-      "learning_rate": 0.00019999466320964188,
-      "loss": 46.0,
-      "step": 43022
-    },
-    {
-      "epoch": 3.289408796375939,
-      "grad_norm": 0.0024893921799957752,
-      "learning_rate": 0.00019999466296148765,
-      "loss": 46.0,
-      "step": 43023
-    },
-    {
-      "epoch": 3.2894852533593286,
-      "grad_norm": 0.0020806242246180773,
-      "learning_rate": 0.00019999466271332765,
-      "loss": 46.0,
-      "step": 43024
-    },
-    {
-      "epoch": 3.2895617103427184,
-      "grad_norm": 0.002393683884292841,
-      "learning_rate": 0.0001999946624651619,
-      "loss": 46.0,
-      "step": 43025
-    },
-    {
-      "epoch": 3.289638167326108,
-      "grad_norm": 0.0032088824082165956,
-      "learning_rate": 0.00019999466221699036,
-      "loss": 46.0,
-      "step": 43026
-    },
-    {
-      "epoch": 3.289714624309498,
-      "grad_norm": 0.0015087004285305738,
-      "learning_rate": 0.000199994661968813,
-      "loss": 46.0,
-      "step": 43027
-    },
-    {
-      "epoch": 3.2897910812928877,
-      "grad_norm": 0.0026160881388932467,
-      "learning_rate": 0.00019999466172062995,
-      "loss": 46.0,
-      "step": 43028
-    },
-    {
-      "epoch": 3.2898675382762774,
-      "grad_norm": 0.0017488918965682387,
-      "learning_rate": 0.0001999946614724411,
-      "loss": 46.0,
-      "step": 43029
-    },
-    {
-      "epoch": 3.289943995259667,
-      "grad_norm": 0.0025165732949972153,
-      "learning_rate": 0.0001999946612242465,
-      "loss": 46.0,
-      "step": 43030
-    },
-    {
-      "epoch": 3.2900204522430565,
-      "grad_norm": 0.002355508739128709,
-      "learning_rate": 0.0001999946609760461,
-      "loss": 46.0,
-      "step": 43031
-    },
-    {
-      "epoch": 3.2900969092264463,
-      "grad_norm": 0.0003302995173726231,
-      "learning_rate": 0.00019999466072783998,
-      "loss": 46.0,
-      "step": 43032
-    },
-    {
-      "epoch": 3.290173366209836,
-      "grad_norm": 0.010817056521773338,
-      "learning_rate": 0.00019999466047962802,
-      "loss": 46.0,
-      "step": 43033
-    },
-    {
-      "epoch": 3.290249823193226,
-      "grad_norm": 0.0010188055457547307,
-      "learning_rate": 0.00019999466023141034,
-      "loss": 46.0,
-      "step": 43034
-    },
-    {
-      "epoch": 3.2903262801766155,
-      "grad_norm": 0.00414435938000679,
-      "learning_rate": 0.00019999465998318686,
-      "loss": 46.0,
-      "step": 43035
-    },
-    {
-      "epoch": 3.2904027371600053,
-      "grad_norm": 0.0035603779833763838,
-      "learning_rate": 0.00019999465973495766,
-      "loss": 46.0,
-      "step": 43036
-    },
-    {
-      "epoch": 3.290479194143395,
-      "grad_norm": 0.0033431260380893946,
-      "learning_rate": 0.00019999465948672263,
-      "loss": 46.0,
-      "step": 43037
-    },
-    {
-      "epoch": 3.290555651126785,
-      "grad_norm": 0.0009001412545330822,
-      "learning_rate": 0.00019999465923848186,
-      "loss": 46.0,
-      "step": 43038
-    },
-    {
-      "epoch": 3.2906321081101746,
-      "grad_norm": 0.0007501213112846017,
-      "learning_rate": 0.00019999465899023532,
-      "loss": 46.0,
-      "step": 43039
-    },
-    {
-      "epoch": 3.2907085650935644,
-      "grad_norm": 0.0036481693387031555,
-      "learning_rate": 0.00019999465874198302,
-      "loss": 46.0,
-      "step": 43040
-    },
-    {
-      "epoch": 3.290785022076954,
-      "grad_norm": 0.0014177331468090415,
-      "learning_rate": 0.00019999465849372493,
-      "loss": 46.0,
-      "step": 43041
-    },
-    {
-      "epoch": 3.290861479060344,
-      "grad_norm": 0.003811736823990941,
-      "learning_rate": 0.00019999465824546112,
-      "loss": 46.0,
-      "step": 43042
-    },
-    {
-      "epoch": 3.2909379360437336,
-      "grad_norm": 0.0017111067427322268,
-      "learning_rate": 0.0001999946579971915,
-      "loss": 46.0,
-      "step": 43043
-    },
-    {
-      "epoch": 3.291014393027123,
-      "grad_norm": 0.0007644662400707603,
-      "learning_rate": 0.0001999946577489161,
-      "loss": 46.0,
-      "step": 43044
-    },
-    {
-      "epoch": 3.2910908500105127,
-      "grad_norm": 0.005183683708310127,
-      "learning_rate": 0.00019999465750063496,
-      "loss": 46.0,
-      "step": 43045
-    },
-    {
-      "epoch": 3.2911673069939025,
-      "grad_norm": 0.0016056194435805082,
-      "learning_rate": 0.00019999465725234803,
-      "loss": 46.0,
-      "step": 43046
-    },
-    {
-      "epoch": 3.2912437639772922,
-      "grad_norm": 0.0008942779386416078,
-      "learning_rate": 0.00019999465700405532,
-      "loss": 46.0,
-      "step": 43047
-    },
-    {
-      "epoch": 3.291320220960682,
-      "grad_norm": 0.002565128030255437,
-      "learning_rate": 0.00019999465675575687,
-      "loss": 46.0,
-      "step": 43048
-    },
-    {
-      "epoch": 3.2913966779440718,
-      "grad_norm": 0.0014598025009036064,
-      "learning_rate": 0.00019999465650745265,
-      "loss": 46.0,
-      "step": 43049
-    },
-    {
-      "epoch": 3.2914731349274615,
-      "grad_norm": 0.0007604403654113412,
-      "learning_rate": 0.00019999465625914265,
-      "loss": 46.0,
-      "step": 43050
-    },
-    {
-      "epoch": 3.2915495919108513,
-      "grad_norm": 0.0021494082175195217,
-      "learning_rate": 0.00019999465601082687,
-      "loss": 46.0,
-      "step": 43051
-    },
-    {
-      "epoch": 3.291626048894241,
-      "grad_norm": 0.0008570743957534432,
-      "learning_rate": 0.00019999465576250533,
-      "loss": 46.0,
-      "step": 43052
-    },
-    {
-      "epoch": 3.2917025058776304,
-      "grad_norm": 0.0012927136849611998,
-      "learning_rate": 0.000199994655514178,
-      "loss": 46.0,
-      "step": 43053
-    },
-    {
-      "epoch": 3.29177896286102,
-      "grad_norm": 0.0027778660878539085,
-      "learning_rate": 0.00019999465526584494,
-      "loss": 46.0,
-      "step": 43054
-    },
-    {
-      "epoch": 3.29185541984441,
-      "grad_norm": 0.0015423414297401905,
-      "learning_rate": 0.0001999946550175061,
-      "loss": 46.0,
-      "step": 43055
-    },
-    {
-      "epoch": 3.2919318768277996,
-      "grad_norm": 0.0027668089605867863,
-      "learning_rate": 0.0001999946547691615,
-      "loss": 46.0,
-      "step": 43056
-    },
-    {
-      "epoch": 3.2920083338111894,
-      "grad_norm": 0.001075509237125516,
-      "learning_rate": 0.0001999946545208111,
-      "loss": 46.0,
-      "step": 43057
-    },
-    {
-      "epoch": 3.292084790794579,
-      "grad_norm": 0.0024258026387542486,
-      "learning_rate": 0.00019999465427245495,
-      "loss": 46.0,
-      "step": 43058
-    },
-    {
-      "epoch": 3.292161247777969,
-      "grad_norm": 0.0016378405271098018,
-      "learning_rate": 0.00019999465402409304,
-      "loss": 46.0,
-      "step": 43059
-    },
-    {
-      "epoch": 3.2922377047613587,
-      "grad_norm": 0.001945990021340549,
-      "learning_rate": 0.00019999465377572534,
-      "loss": 46.0,
-      "step": 43060
-    },
-    {
-      "epoch": 3.2923141617447484,
-      "grad_norm": 0.0009554408024996519,
-      "learning_rate": 0.0001999946535273519,
-      "loss": 46.0,
-      "step": 43061
-    },
-    {
-      "epoch": 3.292390618728138,
-      "grad_norm": 0.0030981251038610935,
-      "learning_rate": 0.00019999465327897266,
-      "loss": 46.0,
-      "step": 43062
-    },
-    {
-      "epoch": 3.292467075711528,
-      "grad_norm": 0.0012822715798392892,
-      "learning_rate": 0.00019999465303058766,
-      "loss": 46.0,
-      "step": 43063
-    },
-    {
-      "epoch": 3.2925435326949177,
-      "grad_norm": 0.0017161233117803931,
-      "learning_rate": 0.0001999946527821969,
-      "loss": 46.0,
-      "step": 43064
-    },
-    {
-      "epoch": 3.2926199896783075,
-      "grad_norm": 0.0021098225843161345,
-      "learning_rate": 0.00019999465253380035,
-      "loss": 46.0,
-      "step": 43065
-    },
-    {
-      "epoch": 3.292696446661697,
-      "grad_norm": 0.00217106263153255,
-      "learning_rate": 0.00019999465228539805,
-      "loss": 46.0,
-      "step": 43066
-    },
-    {
-      "epoch": 3.2927729036450866,
-      "grad_norm": 0.0012985338689759374,
-      "learning_rate": 0.00019999465203698996,
-      "loss": 46.0,
-      "step": 43067
-    },
-    {
-      "epoch": 3.2928493606284763,
-      "grad_norm": 0.0030153896659612656,
-      "learning_rate": 0.00019999465178857612,
-      "loss": 46.0,
-      "step": 43068
-    },
-    {
-      "epoch": 3.292925817611866,
-      "grad_norm": 0.0032734698615968227,
-      "learning_rate": 0.0001999946515401565,
-      "loss": 46.0,
-      "step": 43069
-    },
-    {
-      "epoch": 3.293002274595256,
-      "grad_norm": 0.0009464027243666351,
-      "learning_rate": 0.00019999465129173113,
-      "loss": 46.0,
-      "step": 43070
-    },
-    {
-      "epoch": 3.2930787315786456,
-      "grad_norm": 0.0007659874972887337,
-      "learning_rate": 0.0001999946510433,
-      "loss": 46.0,
-      "step": 43071
-    },
-    {
-      "epoch": 3.2931551885620354,
-      "grad_norm": 0.0020798335317522287,
-      "learning_rate": 0.00019999465079486306,
-      "loss": 46.0,
-      "step": 43072
-    },
-    {
-      "epoch": 3.293231645545425,
-      "grad_norm": 0.0021051503717899323,
-      "learning_rate": 0.00019999465054642036,
-      "loss": 46.0,
-      "step": 43073
-    },
-    {
-      "epoch": 3.293308102528815,
-      "grad_norm": 0.001819591037929058,
-      "learning_rate": 0.00019999465029797193,
-      "loss": 46.0,
-      "step": 43074
-    },
-    {
-      "epoch": 3.293384559512204,
-      "grad_norm": 0.0013232157798483968,
-      "learning_rate": 0.00019999465004951768,
-      "loss": 46.0,
-      "step": 43075
-    },
-    {
-      "epoch": 3.293461016495594,
-      "grad_norm": 0.0029392847791314125,
-      "learning_rate": 0.00019999464980105768,
-      "loss": 46.0,
-      "step": 43076
-    },
-    {
-      "epoch": 3.2935374734789837,
-      "grad_norm": 0.002965545281767845,
-      "learning_rate": 0.0001999946495525919,
-      "loss": 46.0,
-      "step": 43077
-    },
-    {
-      "epoch": 3.2936139304623735,
-      "grad_norm": 0.0026470404118299484,
-      "learning_rate": 0.0001999946493041204,
-      "loss": 46.0,
-      "step": 43078
-    },
-    {
-      "epoch": 3.2936903874457633,
-      "grad_norm": 0.0034660804085433483,
-      "learning_rate": 0.00019999464905564308,
-      "loss": 46.0,
-      "step": 43079
-    },
-    {
-      "epoch": 3.293766844429153,
-      "grad_norm": 0.004432412795722485,
-      "learning_rate": 0.00019999464880716,
-      "loss": 46.0,
-      "step": 43080
-    },
-    {
-      "epoch": 3.293843301412543,
-      "grad_norm": 0.0048272740095853806,
-      "learning_rate": 0.00019999464855867115,
-      "loss": 46.0,
-      "step": 43081
-    },
-    {
-      "epoch": 3.2939197583959325,
-      "grad_norm": 0.0006613717414438725,
-      "learning_rate": 0.00019999464831017656,
-      "loss": 46.0,
-      "step": 43082
-    },
-    {
-      "epoch": 3.2939962153793223,
-      "grad_norm": 0.0035553157795220613,
-      "learning_rate": 0.00019999464806167615,
-      "loss": 46.0,
-      "step": 43083
-    },
-    {
-      "epoch": 3.294072672362712,
-      "grad_norm": 0.001576737966388464,
-      "learning_rate": 0.00019999464781317002,
-      "loss": 46.0,
-      "step": 43084
-    },
-    {
-      "epoch": 3.294149129346102,
-      "grad_norm": 0.001536518451757729,
-      "learning_rate": 0.0001999946475646581,
-      "loss": 46.0,
-      "step": 43085
-    },
-    {
-      "epoch": 3.2942255863294916,
-      "grad_norm": 0.0013201546389609575,
-      "learning_rate": 0.0001999946473161404,
-      "loss": 46.0,
-      "step": 43086
-    },
-    {
-      "epoch": 3.294302043312881,
-      "grad_norm": 0.0020425517577677965,
-      "learning_rate": 0.00019999464706761696,
-      "loss": 46.0,
-      "step": 43087
-    },
-    {
-      "epoch": 3.2943785002962707,
-      "grad_norm": 0.0040287296287715435,
-      "learning_rate": 0.00019999464681908773,
-      "loss": 46.0,
-      "step": 43088
-    },
-    {
-      "epoch": 3.2944549572796604,
-      "grad_norm": 0.0027974038384854794,
-      "learning_rate": 0.00019999464657055274,
-      "loss": 46.0,
-      "step": 43089
-    },
-    {
-      "epoch": 3.29453141426305,
-      "grad_norm": 0.0019716296810656786,
-      "learning_rate": 0.00019999464632201197,
-      "loss": 46.0,
-      "step": 43090
-    },
-    {
-      "epoch": 3.29460787124644,
-      "grad_norm": 0.0012575959553942084,
-      "learning_rate": 0.00019999464607346542,
-      "loss": 46.0,
-      "step": 43091
-    },
-    {
-      "epoch": 3.2946843282298297,
-      "grad_norm": 0.0026781733613461256,
-      "learning_rate": 0.00019999464582491313,
-      "loss": 46.0,
-      "step": 43092
-    },
-    {
-      "epoch": 3.2947607852132195,
-      "grad_norm": 0.0016945755342021585,
-      "learning_rate": 0.00019999464557635507,
-      "loss": 46.0,
-      "step": 43093
-    },
-    {
-      "epoch": 3.2948372421966092,
-      "grad_norm": 0.0014913951745256782,
-      "learning_rate": 0.00019999464532779123,
-      "loss": 46.0,
-      "step": 43094
-    },
-    {
-      "epoch": 3.294913699179999,
-      "grad_norm": 0.001143215806223452,
-      "learning_rate": 0.00019999464507922162,
-      "loss": 46.0,
-      "step": 43095
-    },
-    {
-      "epoch": 3.2949901561633888,
-      "grad_norm": 0.005004389211535454,
-      "learning_rate": 0.00019999464483064623,
-      "loss": 46.0,
-      "step": 43096
-    },
-    {
-      "epoch": 3.295066613146778,
-      "grad_norm": 0.0038042457308620214,
-      "learning_rate": 0.00019999464458206507,
-      "loss": 46.0,
-      "step": 43097
-    },
-    {
-      "epoch": 3.295143070130168,
-      "grad_norm": 0.0010423966450616717,
-      "learning_rate": 0.00019999464433347814,
-      "loss": 46.0,
-      "step": 43098
-    },
-    {
-      "epoch": 3.2952195271135576,
-      "grad_norm": 0.0007346981437876821,
-      "learning_rate": 0.00019999464408488547,
-      "loss": 46.0,
-      "step": 43099
-    },
-    {
-      "epoch": 3.2952959840969474,
-      "grad_norm": 0.0009377679089084268,
-      "learning_rate": 0.00019999464383628702,
-      "loss": 46.0,
-      "step": 43100
-    },
-    {
-      "epoch": 3.295372441080337,
-      "grad_norm": 0.0013458920875564218,
-      "learning_rate": 0.0001999946435876828,
-      "loss": 46.0,
-      "step": 43101
-    },
-    {
-      "epoch": 3.295448898063727,
-      "grad_norm": 0.000986292609013617,
-      "learning_rate": 0.0001999946433390728,
-      "loss": 46.0,
-      "step": 43102
-    },
-    {
-      "epoch": 3.2955253550471166,
-      "grad_norm": 0.011193119920790195,
-      "learning_rate": 0.00019999464309045702,
-      "loss": 46.0,
-      "step": 43103
-    },
-    {
-      "epoch": 3.2956018120305064,
-      "grad_norm": 0.002467630198225379,
-      "learning_rate": 0.00019999464284183548,
-      "loss": 46.0,
-      "step": 43104
-    },
-    {
-      "epoch": 3.295678269013896,
-      "grad_norm": 0.00205299467779696,
-      "learning_rate": 0.0001999946425932082,
-      "loss": 46.0,
-      "step": 43105
-    },
-    {
-      "epoch": 3.295754725997286,
-      "grad_norm": 0.0013569977600127459,
-      "learning_rate": 0.00019999464234457512,
-      "loss": 46.0,
-      "step": 43106
-    },
-    {
-      "epoch": 3.2958311829806757,
-      "grad_norm": 0.005249109584838152,
-      "learning_rate": 0.00019999464209593629,
-      "loss": 46.0,
-      "step": 43107
-    },
-    {
-      "epoch": 3.2959076399640654,
-      "grad_norm": 0.0003961015318054706,
-      "learning_rate": 0.00019999464184729168,
-      "loss": 46.0,
-      "step": 43108
-    },
-    {
-      "epoch": 3.2959840969474548,
-      "grad_norm": 0.0028856215067207813,
-      "learning_rate": 0.0001999946415986413,
-      "loss": 46.0,
-      "step": 43109
-    },
-    {
-      "epoch": 3.2960605539308445,
-      "grad_norm": 0.005846831016242504,
-      "learning_rate": 0.00019999464134998513,
-      "loss": 46.0,
-      "step": 43110
-    },
-    {
-      "epoch": 3.2961370109142343,
-      "grad_norm": 0.0016929770354181528,
-      "learning_rate": 0.00019999464110132323,
-      "loss": 46.0,
-      "step": 43111
-    },
-    {
-      "epoch": 3.296213467897624,
-      "grad_norm": 0.0036313924938440323,
-      "learning_rate": 0.00019999464085265553,
-      "loss": 46.0,
-      "step": 43112
-    },
-    {
-      "epoch": 3.296289924881014,
-      "grad_norm": 0.0017907816218212247,
-      "learning_rate": 0.0001999946406039821,
-      "loss": 46.0,
-      "step": 43113
-    },
-    {
-      "epoch": 3.2963663818644036,
-      "grad_norm": 0.0031940562184900045,
-      "learning_rate": 0.00019999464035530285,
-      "loss": 46.0,
-      "step": 43114
-    },
-    {
-      "epoch": 3.2964428388477933,
-      "grad_norm": 0.0033530748914927244,
-      "learning_rate": 0.00019999464010661785,
-      "loss": 46.0,
-      "step": 43115
-    },
-    {
-      "epoch": 3.296519295831183,
-      "grad_norm": 0.0071699912659823895,
-      "learning_rate": 0.0001999946398579271,
-      "loss": 46.0,
-      "step": 43116
-    },
-    {
-      "epoch": 3.296595752814573,
-      "grad_norm": 0.0024224447552114725,
-      "learning_rate": 0.00019999463960923057,
-      "loss": 46.0,
-      "step": 43117
-    },
-    {
-      "epoch": 3.2966722097979626,
-      "grad_norm": 0.001077526481822133,
-      "learning_rate": 0.00019999463936052828,
-      "loss": 46.0,
-      "step": 43118
-    },
-    {
-      "epoch": 3.296748666781352,
-      "grad_norm": 0.0010505522368475795,
-      "learning_rate": 0.00019999463911182018,
-      "loss": 46.0,
-      "step": 43119
-    },
-    {
-      "epoch": 3.2968251237647417,
-      "grad_norm": 0.0019745826721191406,
-      "learning_rate": 0.00019999463886310635,
-      "loss": 46.0,
-      "step": 43120
-    },
-    {
-      "epoch": 3.2969015807481314,
-      "grad_norm": 0.0010027430253103375,
-      "learning_rate": 0.00019999463861438674,
-      "loss": 46.0,
-      "step": 43121
-    },
-    {
-      "epoch": 3.296978037731521,
-      "grad_norm": 0.00284214923158288,
-      "learning_rate": 0.00019999463836566138,
-      "loss": 46.0,
-      "step": 43122
-    },
-    {
-      "epoch": 3.297054494714911,
-      "grad_norm": 0.003681445261463523,
-      "learning_rate": 0.00019999463811693022,
-      "loss": 46.0,
-      "step": 43123
-    },
-    {
-      "epoch": 3.2971309516983007,
-      "grad_norm": 0.002956235548481345,
-      "learning_rate": 0.0001999946378681933,
-      "loss": 46.0,
-      "step": 43124
-    },
-    {
-      "epoch": 3.2972074086816905,
-      "grad_norm": 0.0025147164706140757,
-      "learning_rate": 0.00019999463761945064,
-      "loss": 46.0,
-      "step": 43125
-    },
-    {
-      "epoch": 3.2972838656650802,
-      "grad_norm": 0.002099211560562253,
-      "learning_rate": 0.00019999463737070217,
-      "loss": 46.0,
-      "step": 43126
-    },
-    {
-      "epoch": 3.29736032264847,
-      "grad_norm": 0.0037914717104285955,
-      "learning_rate": 0.00019999463712194794,
-      "loss": 46.0,
-      "step": 43127
-    },
-    {
-      "epoch": 3.2974367796318598,
-      "grad_norm": 0.001852756249718368,
-      "learning_rate": 0.00019999463687318795,
-      "loss": 46.0,
-      "step": 43128
-    },
-    {
-      "epoch": 3.2975132366152495,
-      "grad_norm": 0.0023945618886500597,
-      "learning_rate": 0.00019999463662442217,
-      "loss": 46.0,
-      "step": 43129
-    },
-    {
-      "epoch": 3.2975896935986393,
-      "grad_norm": 0.002128422725945711,
-      "learning_rate": 0.00019999463637565066,
-      "loss": 46.0,
-      "step": 43130
-    },
-    {
-      "epoch": 3.2976661505820286,
-      "grad_norm": 0.002548876451328397,
-      "learning_rate": 0.00019999463612687337,
-      "loss": 46.0,
-      "step": 43131
-    },
-    {
-      "epoch": 3.2977426075654184,
-      "grad_norm": 0.0016226563602685928,
-      "learning_rate": 0.00019999463587809028,
-      "loss": 46.0,
-      "step": 43132
-    },
-    {
-      "epoch": 3.297819064548808,
-      "grad_norm": 0.0031711033079773188,
-      "learning_rate": 0.00019999463562930144,
-      "loss": 46.0,
-      "step": 43133
-    },
-    {
-      "epoch": 3.297895521532198,
-      "grad_norm": 0.001658200053498149,
-      "learning_rate": 0.00019999463538050686,
-      "loss": 46.0,
-      "step": 43134
-    },
-    {
-      "epoch": 3.2979719785155877,
-      "grad_norm": 0.001705559203401208,
-      "learning_rate": 0.00019999463513170647,
-      "loss": 46.0,
-      "step": 43135
-    },
-    {
-      "epoch": 3.2980484354989774,
-      "grad_norm": 0.00634820805862546,
-      "learning_rate": 0.00019999463488290032,
-      "loss": 46.0,
-      "step": 43136
-    },
-    {
-      "epoch": 3.298124892482367,
-      "grad_norm": 0.0017934966599568725,
-      "learning_rate": 0.00019999463463408841,
-      "loss": 46.0,
-      "step": 43137
-    },
-    {
-      "epoch": 3.298201349465757,
-      "grad_norm": 0.002681695856153965,
-      "learning_rate": 0.00019999463438527074,
-      "loss": 46.0,
-      "step": 43138
-    },
-    {
-      "epoch": 3.2982778064491467,
-      "grad_norm": 0.007503107190132141,
-      "learning_rate": 0.0001999946341364473,
-      "loss": 46.0,
-      "step": 43139
-    },
-    {
-      "epoch": 3.2983542634325365,
-      "grad_norm": 0.0019769046921283007,
-      "learning_rate": 0.00019999463388761806,
-      "loss": 46.0,
-      "step": 43140
-    },
-    {
-      "epoch": 3.2984307204159258,
-      "grad_norm": 0.0032831698190420866,
-      "learning_rate": 0.00019999463363878307,
-      "loss": 46.0,
-      "step": 43141
-    },
-    {
-      "epoch": 3.2985071773993155,
-      "grad_norm": 0.0008680423488840461,
-      "learning_rate": 0.0001999946333899423,
-      "loss": 46.0,
-      "step": 43142
-    },
-    {
-      "epoch": 3.2985836343827053,
-      "grad_norm": 0.0013498598709702492,
-      "learning_rate": 0.00019999463314109575,
-      "loss": 46.0,
-      "step": 43143
-    },
-    {
-      "epoch": 3.298660091366095,
-      "grad_norm": 0.010320496745407581,
-      "learning_rate": 0.00019999463289224346,
-      "loss": 46.0,
-      "step": 43144
-    },
-    {
-      "epoch": 3.298736548349485,
-      "grad_norm": 0.00266675534658134,
-      "learning_rate": 0.00019999463264338537,
-      "loss": 46.0,
-      "step": 43145
-    },
-    {
-      "epoch": 3.2988130053328746,
-      "grad_norm": 0.0016131686279550195,
-      "learning_rate": 0.00019999463239452157,
-      "loss": 46.0,
-      "step": 43146
-    },
-    {
-      "epoch": 3.2988894623162643,
-      "grad_norm": 0.0020049726590514183,
-      "learning_rate": 0.00019999463214565196,
-      "loss": 46.0,
-      "step": 43147
-    },
-    {
-      "epoch": 3.298965919299654,
-      "grad_norm": 0.0020847790874540806,
-      "learning_rate": 0.00019999463189677657,
-      "loss": 46.0,
-      "step": 43148
-    },
-    {
-      "epoch": 3.299042376283044,
-      "grad_norm": 0.0042830053716897964,
-      "learning_rate": 0.00019999463164789542,
-      "loss": 46.0,
-      "step": 43149
-    },
-    {
-      "epoch": 3.2991188332664336,
-      "grad_norm": 0.0032707746140658855,
-      "learning_rate": 0.0001999946313990085,
-      "loss": 46.0,
-      "step": 43150
-    },
-    {
-      "epoch": 3.2991952902498234,
-      "grad_norm": 0.002887357259169221,
-      "learning_rate": 0.00019999463115011584,
-      "loss": 46.0,
-      "step": 43151
-    },
-    {
-      "epoch": 3.299271747233213,
-      "grad_norm": 0.0020889570005238056,
-      "learning_rate": 0.00019999463090121736,
-      "loss": 46.0,
-      "step": 43152
-    },
-    {
-      "epoch": 3.2993482042166025,
-      "grad_norm": 0.0037425702903419733,
-      "learning_rate": 0.00019999463065231314,
-      "loss": 46.0,
-      "step": 43153
-    },
-    {
-      "epoch": 3.2994246611999922,
-      "grad_norm": 0.0010507851839065552,
-      "learning_rate": 0.0001999946304034032,
-      "loss": 46.0,
-      "step": 43154
-    },
-    {
-      "epoch": 3.299501118183382,
-      "grad_norm": 0.0012318973895162344,
-      "learning_rate": 0.0001999946301544874,
-      "loss": 46.0,
-      "step": 43155
-    },
-    {
-      "epoch": 3.2995775751667717,
-      "grad_norm": 0.0014983902219682932,
-      "learning_rate": 0.00019999462990556588,
-      "loss": 46.0,
-      "step": 43156
-    },
-    {
-      "epoch": 3.2996540321501615,
-      "grad_norm": 0.0015422574942931533,
-      "learning_rate": 0.00019999462965663857,
-      "loss": 46.0,
-      "step": 43157
-    },
-    {
-      "epoch": 3.2997304891335513,
-      "grad_norm": 0.007413915358483791,
-      "learning_rate": 0.00019999462940770553,
-      "loss": 46.0,
-      "step": 43158
-    },
-    {
-      "epoch": 3.299806946116941,
-      "grad_norm": 0.002296923426911235,
-      "learning_rate": 0.00019999462915876667,
-      "loss": 46.0,
-      "step": 43159
-    },
-    {
-      "epoch": 3.299883403100331,
-      "grad_norm": 0.0020284513011574745,
-      "learning_rate": 0.00019999462890982206,
-      "loss": 46.0,
-      "step": 43160
-    },
-    {
-      "epoch": 3.2999598600837206,
-      "grad_norm": 0.0022211328614503145,
-      "learning_rate": 0.00019999462866087167,
-      "loss": 46.0,
-      "step": 43161
-    },
-    {
-      "epoch": 3.30003631706711,
-      "grad_norm": 0.0027674459852278233,
-      "learning_rate": 0.00019999462841191555,
-      "loss": 46.0,
-      "step": 43162
-    },
-    {
-      "epoch": 3.3001127740504996,
-      "grad_norm": 0.0029754112474620342,
-      "learning_rate": 0.00019999462816295364,
-      "loss": 46.0,
-      "step": 43163
-    },
-    {
-      "epoch": 3.3001892310338894,
-      "grad_norm": 0.0010440535843372345,
-      "learning_rate": 0.00019999462791398597,
-      "loss": 46.0,
-      "step": 43164
-    },
-    {
-      "epoch": 3.300265688017279,
-      "grad_norm": 0.0016595243941992521,
-      "learning_rate": 0.00019999462766501252,
-      "loss": 46.0,
-      "step": 43165
-    },
-    {
-      "epoch": 3.300342145000669,
-      "grad_norm": 0.0031372876837849617,
-      "learning_rate": 0.00019999462741603327,
-      "loss": 46.0,
-      "step": 43166
-    },
-    {
-      "epoch": 3.3004186019840587,
-      "grad_norm": 0.0038617700338363647,
-      "learning_rate": 0.0001999946271670483,
-      "loss": 46.0,
-      "step": 43167
-    },
-    {
-      "epoch": 3.3004950589674484,
-      "grad_norm": 0.0014321524649858475,
-      "learning_rate": 0.00019999462691805753,
-      "loss": 46.0,
-      "step": 43168
-    },
-    {
-      "epoch": 3.300571515950838,
-      "grad_norm": 0.0016711888602003455,
-      "learning_rate": 0.00019999462666906101,
-      "loss": 46.0,
-      "step": 43169
-    },
-    {
-      "epoch": 3.300647972934228,
-      "grad_norm": 0.0033880462870001793,
-      "learning_rate": 0.00019999462642005872,
-      "loss": 46.0,
-      "step": 43170
-    },
-    {
-      "epoch": 3.3007244299176177,
-      "grad_norm": 0.0015615088632330298,
-      "learning_rate": 0.00019999462617105064,
-      "loss": 46.0,
-      "step": 43171
-    },
-    {
-      "epoch": 3.3008008869010075,
-      "grad_norm": 0.0018420040141791105,
-      "learning_rate": 0.00019999462592203683,
-      "loss": 46.0,
-      "step": 43172
-    },
-    {
-      "epoch": 3.3008773438843972,
-      "grad_norm": 0.003144825343042612,
-      "learning_rate": 0.00019999462567301722,
-      "loss": 46.0,
-      "step": 43173
-    },
-    {
-      "epoch": 3.300953800867787,
-      "grad_norm": 0.0026488881558179855,
-      "learning_rate": 0.00019999462542399183,
-      "loss": 46.0,
-      "step": 43174
-    },
-    {
-      "epoch": 3.3010302578511763,
-      "grad_norm": 0.002214920474216342,
-      "learning_rate": 0.0001999946251749607,
-      "loss": 46.0,
-      "step": 43175
-    },
-    {
-      "epoch": 3.301106714834566,
-      "grad_norm": 0.0024944681208580732,
-      "learning_rate": 0.00019999462492592378,
-      "loss": 46.0,
-      "step": 43176
-    },
-    {
-      "epoch": 3.301183171817956,
-      "grad_norm": 0.001967791933566332,
-      "learning_rate": 0.0001999946246768811,
-      "loss": 46.0,
-      "step": 43177
-    },
-    {
-      "epoch": 3.3012596288013456,
-      "grad_norm": 0.005799522157758474,
-      "learning_rate": 0.00019999462442783265,
-      "loss": 46.0,
-      "step": 43178
-    },
-    {
-      "epoch": 3.3013360857847354,
-      "grad_norm": 0.0009032622328959405,
-      "learning_rate": 0.00019999462417877843,
-      "loss": 46.0,
-      "step": 43179
-    },
-    {
-      "epoch": 3.301412542768125,
-      "grad_norm": 0.0009598106262274086,
-      "learning_rate": 0.00019999462392971843,
-      "loss": 46.0,
-      "step": 43180
-    },
-    {
-      "epoch": 3.301488999751515,
-      "grad_norm": 0.0018510113004595041,
-      "learning_rate": 0.0001999946236806527,
-      "loss": 46.0,
-      "step": 43181
-    },
-    {
-      "epoch": 3.3015654567349046,
-      "grad_norm": 0.0017959250835701823,
-      "learning_rate": 0.00019999462343158118,
-      "loss": 46.0,
-      "step": 43182
-    },
-    {
-      "epoch": 3.3016419137182944,
-      "grad_norm": 0.001839536358602345,
-      "learning_rate": 0.00019999462318250386,
-      "loss": 46.0,
-      "step": 43183
-    },
-    {
-      "epoch": 3.3017183707016837,
-      "grad_norm": 0.004612884018570185,
-      "learning_rate": 0.0001999946229334208,
-      "loss": 46.0,
-      "step": 43184
-    },
-    {
-      "epoch": 3.3017948276850735,
-      "grad_norm": 0.00887236837297678,
-      "learning_rate": 0.000199994622684332,
-      "loss": 46.0,
-      "step": 43185
-    },
-    {
-      "epoch": 3.3018712846684632,
-      "grad_norm": 0.002551883226260543,
-      "learning_rate": 0.00019999462243523735,
-      "loss": 46.0,
-      "step": 43186
-    },
-    {
-      "epoch": 3.301947741651853,
-      "grad_norm": 0.0018193320138379931,
-      "learning_rate": 0.00019999462218613697,
-      "loss": 46.0,
-      "step": 43187
-    },
-    {
-      "epoch": 3.3020241986352428,
-      "grad_norm": 0.0031338778790086508,
-      "learning_rate": 0.00019999462193703084,
-      "loss": 46.0,
-      "step": 43188
-    },
-    {
-      "epoch": 3.3021006556186325,
-      "grad_norm": 0.0014880941016599536,
-      "learning_rate": 0.00019999462168791894,
-      "loss": 46.0,
-      "step": 43189
-    },
-    {
-      "epoch": 3.3021771126020223,
-      "grad_norm": 0.0023926033172756433,
-      "learning_rate": 0.00019999462143880127,
-      "loss": 46.0,
-      "step": 43190
-    },
-    {
-      "epoch": 3.302253569585412,
-      "grad_norm": 0.002128173364326358,
-      "learning_rate": 0.0001999946211896778,
-      "loss": 46.0,
-      "step": 43191
-    },
-    {
-      "epoch": 3.302330026568802,
-      "grad_norm": 0.0018917858833447099,
-      "learning_rate": 0.0001999946209405486,
-      "loss": 46.0,
-      "step": 43192
-    },
-    {
-      "epoch": 3.3024064835521916,
-      "grad_norm": 0.001458160812035203,
-      "learning_rate": 0.00019999462069141363,
-      "loss": 46.0,
-      "step": 43193
-    },
-    {
-      "epoch": 3.3024829405355813,
-      "grad_norm": 0.0014849331928417087,
-      "learning_rate": 0.00019999462044227286,
-      "loss": 46.0,
-      "step": 43194
-    },
-    {
-      "epoch": 3.302559397518971,
-      "grad_norm": 0.0024203744251281023,
-      "learning_rate": 0.00019999462019312634,
-      "loss": 46.0,
-      "step": 43195
-    },
-    {
-      "epoch": 3.302635854502361,
-      "grad_norm": 0.005318007431924343,
-      "learning_rate": 0.00019999461994397403,
-      "loss": 46.0,
-      "step": 43196
-    },
-    {
-      "epoch": 3.30271231148575,
-      "grad_norm": 0.004124005790799856,
-      "learning_rate": 0.00019999461969481597,
-      "loss": 46.0,
-      "step": 43197
-    },
-    {
-      "epoch": 3.30278876846914,
-      "grad_norm": 0.002730536973103881,
-      "learning_rate": 0.00019999461944565213,
-      "loss": 46.0,
-      "step": 43198
-    },
-    {
-      "epoch": 3.3028652254525297,
-      "grad_norm": 0.004512130282819271,
-      "learning_rate": 0.00019999461919648252,
-      "loss": 46.0,
-      "step": 43199
-    },
-    {
-      "epoch": 3.3029416824359195,
-      "grad_norm": 0.0030428001191467047,
-      "learning_rate": 0.00019999461894730717,
-      "loss": 46.0,
-      "step": 43200
-    },
-    {
-      "epoch": 3.303018139419309,
-      "grad_norm": 0.004517611116170883,
-      "learning_rate": 0.000199994618698126,
-      "loss": 46.0,
-      "step": 43201
-    },
-    {
-      "epoch": 3.303094596402699,
-      "grad_norm": 0.0021113597322255373,
-      "learning_rate": 0.0001999946184489391,
-      "loss": 46.0,
-      "step": 43202
-    },
-    {
-      "epoch": 3.3031710533860887,
-      "grad_norm": 0.002403896301984787,
-      "learning_rate": 0.00019999461819974644,
-      "loss": 46.0,
-      "step": 43203
-    },
-    {
-      "epoch": 3.3032475103694785,
-      "grad_norm": 0.0007715957472100854,
-      "learning_rate": 0.000199994617950548,
-      "loss": 46.0,
-      "step": 43204
-    },
-    {
-      "epoch": 3.3033239673528683,
-      "grad_norm": 0.0010853923158720136,
-      "learning_rate": 0.00019999461770134377,
-      "loss": 46.0,
-      "step": 43205
-    },
-    {
-      "epoch": 3.3034004243362576,
-      "grad_norm": 0.0007718742126598954,
-      "learning_rate": 0.0001999946174521338,
-      "loss": 46.0,
-      "step": 43206
-    },
-    {
-      "epoch": 3.3034768813196473,
-      "grad_norm": 0.0019535336177796125,
-      "learning_rate": 0.00019999461720291803,
-      "loss": 46.0,
-      "step": 43207
-    },
-    {
-      "epoch": 3.303553338303037,
-      "grad_norm": 0.0034988231491297483,
-      "learning_rate": 0.0001999946169536965,
-      "loss": 46.0,
-      "step": 43208
-    },
-    {
-      "epoch": 3.303629795286427,
-      "grad_norm": 0.0010888231918215752,
-      "learning_rate": 0.0001999946167044692,
-      "loss": 46.0,
-      "step": 43209
-    },
-    {
-      "epoch": 3.3037062522698166,
-      "grad_norm": 0.0024672902654856443,
-      "learning_rate": 0.00019999461645523614,
-      "loss": 46.0,
-      "step": 43210
-    },
-    {
-      "epoch": 3.3037827092532064,
-      "grad_norm": 0.0013750992948189378,
-      "learning_rate": 0.0001999946162059973,
-      "loss": 46.0,
-      "step": 43211
-    },
-    {
-      "epoch": 3.303859166236596,
-      "grad_norm": 0.001331614563241601,
-      "learning_rate": 0.0001999946159567527,
-      "loss": 46.0,
-      "step": 43212
-    },
-    {
-      "epoch": 3.303935623219986,
-      "grad_norm": 0.002048181602731347,
-      "learning_rate": 0.00019999461570750234,
-      "loss": 46.0,
-      "step": 43213
-    },
-    {
-      "epoch": 3.3040120802033757,
-      "grad_norm": 0.0011536324163898826,
-      "learning_rate": 0.0001999946154582462,
-      "loss": 46.0,
-      "step": 43214
-    },
-    {
-      "epoch": 3.3040885371867654,
-      "grad_norm": 0.0008246597717516124,
-      "learning_rate": 0.0001999946152089843,
-      "loss": 46.0,
-      "step": 43215
-    },
-    {
-      "epoch": 3.304164994170155,
-      "grad_norm": 0.007458702195435762,
-      "learning_rate": 0.00019999461495971664,
-      "loss": 46.0,
-      "step": 43216
-    },
-    {
-      "epoch": 3.304241451153545,
-      "grad_norm": 0.0008868675213307142,
-      "learning_rate": 0.0001999946147104432,
-      "loss": 46.0,
-      "step": 43217
-    },
-    {
-      "epoch": 3.3043179081369343,
-      "grad_norm": 0.0031157336197793484,
-      "learning_rate": 0.00019999461446116397,
-      "loss": 46.0,
-      "step": 43218
-    },
-    {
-      "epoch": 3.304394365120324,
-      "grad_norm": 0.0027397051453590393,
-      "learning_rate": 0.00019999461421187897,
-      "loss": 46.0,
-      "step": 43219
-    },
-    {
-      "epoch": 3.304470822103714,
-      "grad_norm": 0.004035792779177427,
-      "learning_rate": 0.0001999946139625882,
-      "loss": 46.0,
-      "step": 43220
-    },
-    {
-      "epoch": 3.3045472790871036,
-      "grad_norm": 0.001293553039431572,
-      "learning_rate": 0.0001999946137132917,
-      "loss": 46.0,
-      "step": 43221
-    },
-    {
-      "epoch": 3.3046237360704933,
-      "grad_norm": 0.001137599116191268,
-      "learning_rate": 0.0001999946134639894,
-      "loss": 46.0,
-      "step": 43222
-    },
-    {
-      "epoch": 3.304700193053883,
-      "grad_norm": 0.002191295148804784,
-      "learning_rate": 0.00019999461321468134,
-      "loss": 46.0,
-      "step": 43223
-    },
-    {
-      "epoch": 3.304776650037273,
-      "grad_norm": 0.0029363189823925495,
-      "learning_rate": 0.0001999946129653675,
-      "loss": 46.0,
-      "step": 43224
-    },
-    {
-      "epoch": 3.3048531070206626,
-      "grad_norm": 0.002210573758929968,
-      "learning_rate": 0.00019999461271604793,
-      "loss": 46.0,
-      "step": 43225
-    },
-    {
-      "epoch": 3.3049295640040524,
-      "grad_norm": 0.0021649464033544064,
-      "learning_rate": 0.00019999461246672254,
-      "loss": 46.0,
-      "step": 43226
-    },
-    {
-      "epoch": 3.305006020987442,
-      "grad_norm": 0.0019673234783113003,
-      "learning_rate": 0.0001999946122173914,
-      "loss": 46.0,
-      "step": 43227
-    },
-    {
-      "epoch": 3.3050824779708314,
-      "grad_norm": 0.0006071816314943135,
-      "learning_rate": 0.0001999946119680545,
-      "loss": 46.0,
-      "step": 43228
-    },
-    {
-      "epoch": 3.305158934954221,
-      "grad_norm": 0.0019875140860676765,
-      "learning_rate": 0.00019999461171871184,
-      "loss": 46.0,
-      "step": 43229
-    },
-    {
-      "epoch": 3.305235391937611,
-      "grad_norm": 0.0029642588924616575,
-      "learning_rate": 0.00019999461146936337,
-      "loss": 46.0,
-      "step": 43230
-    },
-    {
-      "epoch": 3.3053118489210007,
-      "grad_norm": 0.002159376861527562,
-      "learning_rate": 0.00019999461122000915,
-      "loss": 46.0,
-      "step": 43231
-    },
-    {
-      "epoch": 3.3053883059043905,
-      "grad_norm": 0.0046060094609856606,
-      "learning_rate": 0.00019999461097064918,
-      "loss": 46.0,
-      "step": 43232
-    },
-    {
-      "epoch": 3.3054647628877802,
-      "grad_norm": 0.0026856299955397844,
-      "learning_rate": 0.00019999461072128344,
-      "loss": 46.0,
-      "step": 43233
-    },
-    {
-      "epoch": 3.30554121987117,
-      "grad_norm": 0.0013417989248409867,
-      "learning_rate": 0.0001999946104719119,
-      "loss": 46.0,
-      "step": 43234
-    },
-    {
-      "epoch": 3.3056176768545598,
-      "grad_norm": 0.0019791172817349434,
-      "learning_rate": 0.0001999946102225346,
-      "loss": 46.0,
-      "step": 43235
-    },
-    {
-      "epoch": 3.3056941338379495,
-      "grad_norm": 0.0023111761547625065,
-      "learning_rate": 0.00019999460997315155,
-      "loss": 46.0,
-      "step": 43236
-    },
-    {
-      "epoch": 3.3057705908213393,
-      "grad_norm": 0.0019377714488655329,
-      "learning_rate": 0.0001999946097237627,
-      "loss": 46.0,
-      "step": 43237
-    },
-    {
-      "epoch": 3.305847047804729,
-      "grad_norm": 0.0011661823373287916,
-      "learning_rate": 0.00019999460947436813,
-      "loss": 46.0,
-      "step": 43238
-    },
-    {
-      "epoch": 3.305923504788119,
-      "grad_norm": 0.0007760869339108467,
-      "learning_rate": 0.00019999460922496775,
-      "loss": 46.0,
-      "step": 43239
-    },
-    {
-      "epoch": 3.305999961771508,
-      "grad_norm": 0.001574092311784625,
-      "learning_rate": 0.00019999460897556162,
-      "loss": 46.0,
-      "step": 43240
-    },
-    {
-      "epoch": 3.306076418754898,
-      "grad_norm": 0.0029979865066707134,
-      "learning_rate": 0.00019999460872614972,
-      "loss": 46.0,
-      "step": 43241
-    },
-    {
-      "epoch": 3.3061528757382876,
-      "grad_norm": 0.002518233610317111,
-      "learning_rate": 0.00019999460847673205,
-      "loss": 46.0,
-      "step": 43242
-    },
-    {
-      "epoch": 3.3062293327216774,
-      "grad_norm": 0.0012703253887593746,
-      "learning_rate": 0.00019999460822730858,
-      "loss": 46.0,
-      "step": 43243
-    },
-    {
-      "epoch": 3.306305789705067,
-      "grad_norm": 0.002061791019514203,
-      "learning_rate": 0.00019999460797787938,
-      "loss": 46.0,
-      "step": 43244
-    },
-    {
-      "epoch": 3.306382246688457,
-      "grad_norm": 0.0023671623785048723,
-      "learning_rate": 0.0001999946077284444,
-      "loss": 46.0,
-      "step": 43245
-    },
-    {
-      "epoch": 3.3064587036718467,
-      "grad_norm": 0.0017499792156741023,
-      "learning_rate": 0.00019999460747900365,
-      "loss": 46.0,
-      "step": 43246
-    },
-    {
-      "epoch": 3.3065351606552364,
-      "grad_norm": 0.0012367863673716784,
-      "learning_rate": 0.0001999946072295571,
-      "loss": 46.0,
-      "step": 43247
-    },
-    {
-      "epoch": 3.306611617638626,
-      "grad_norm": 0.002476779278367758,
-      "learning_rate": 0.00019999460698010482,
-      "loss": 46.0,
-      "step": 43248
-    },
-    {
-      "epoch": 3.306688074622016,
-      "grad_norm": 0.002475731773301959,
-      "learning_rate": 0.00019999460673064676,
-      "loss": 46.0,
-      "step": 43249
-    },
-    {
-      "epoch": 3.3067645316054053,
-      "grad_norm": 0.0014733021380379796,
-      "learning_rate": 0.00019999460648118292,
-      "loss": 46.0,
-      "step": 43250
-    },
-    {
-      "epoch": 3.306840988588795,
-      "grad_norm": 0.0016036968445405364,
-      "learning_rate": 0.00019999460623171334,
-      "loss": 46.0,
-      "step": 43251
-    },
-    {
-      "epoch": 3.306917445572185,
-      "grad_norm": 0.001718223444186151,
-      "learning_rate": 0.000199994605982238,
-      "loss": 46.0,
-      "step": 43252
-    },
-    {
-      "epoch": 3.3069939025555746,
-      "grad_norm": 0.002081759739667177,
-      "learning_rate": 0.00019999460573275686,
-      "loss": 46.0,
-      "step": 43253
-    },
-    {
-      "epoch": 3.3070703595389643,
-      "grad_norm": 0.0013873249990865588,
-      "learning_rate": 0.00019999460548326994,
-      "loss": 46.0,
-      "step": 43254
-    },
-    {
-      "epoch": 3.307146816522354,
-      "grad_norm": 0.0026570456102490425,
-      "learning_rate": 0.00019999460523377726,
-      "loss": 46.0,
-      "step": 43255
-    },
-    {
-      "epoch": 3.307223273505744,
-      "grad_norm": 0.0016605970449745655,
-      "learning_rate": 0.00019999460498427882,
-      "loss": 46.0,
-      "step": 43256
-    },
-    {
-      "epoch": 3.3072997304891336,
-      "grad_norm": 0.0028671384789049625,
-      "learning_rate": 0.0001999946047347746,
-      "loss": 46.0,
-      "step": 43257
-    },
-    {
-      "epoch": 3.3073761874725234,
-      "grad_norm": 0.0018654812593013048,
-      "learning_rate": 0.0001999946044852646,
-      "loss": 46.0,
-      "step": 43258
-    },
-    {
-      "epoch": 3.307452644455913,
-      "grad_norm": 0.0019044623477384448,
-      "learning_rate": 0.00019999460423574886,
-      "loss": 46.0,
-      "step": 43259
-    },
-    {
-      "epoch": 3.307529101439303,
-      "grad_norm": 0.0024207562673836946,
-      "learning_rate": 0.00019999460398622735,
-      "loss": 46.0,
-      "step": 43260
-    },
-    {
-      "epoch": 3.3076055584226927,
-      "grad_norm": 0.0017198467394337058,
-      "learning_rate": 0.00019999460373670006,
-      "loss": 46.0,
-      "step": 43261
-    },
-    {
-      "epoch": 3.307682015406082,
-      "grad_norm": 0.0008224956691265106,
-      "learning_rate": 0.000199994603487167,
-      "loss": 46.0,
-      "step": 43262
-    },
-    {
-      "epoch": 3.3077584723894717,
-      "grad_norm": 0.0036526774056255817,
-      "learning_rate": 0.00019999460323762817,
-      "loss": 46.0,
-      "step": 43263
-    },
-    {
-      "epoch": 3.3078349293728615,
-      "grad_norm": 0.0012848360929638147,
-      "learning_rate": 0.00019999460298808356,
-      "loss": 46.0,
-      "step": 43264
-    },
-    {
-      "epoch": 3.3079113863562513,
-      "grad_norm": 0.0009869858622550964,
-      "learning_rate": 0.00019999460273853323,
-      "loss": 46.0,
-      "step": 43265
-    },
-    {
-      "epoch": 3.307987843339641,
-      "grad_norm": 0.0009321004617959261,
-      "learning_rate": 0.00019999460248897708,
-      "loss": 46.0,
-      "step": 43266
-    },
-    {
-      "epoch": 3.308064300323031,
-      "grad_norm": 0.001969235483556986,
-      "learning_rate": 0.0001999946022394152,
-      "loss": 46.0,
-      "step": 43267
-    },
-    {
-      "epoch": 3.3081407573064205,
-      "grad_norm": 0.00263637350872159,
-      "learning_rate": 0.0001999946019898475,
-      "loss": 46.0,
-      "step": 43268
-    },
-    {
-      "epoch": 3.3082172142898103,
-      "grad_norm": 0.002038033213466406,
-      "learning_rate": 0.00019999460174027406,
-      "loss": 46.0,
-      "step": 43269
-    },
-    {
-      "epoch": 3.3082936712732,
-      "grad_norm": 0.005171559285372496,
-      "learning_rate": 0.00019999460149069487,
-      "loss": 46.0,
-      "step": 43270
-    },
-    {
-      "epoch": 3.30837012825659,
-      "grad_norm": 0.003046031342819333,
-      "learning_rate": 0.00019999460124110987,
-      "loss": 46.0,
-      "step": 43271
-    },
-    {
-      "epoch": 3.308446585239979,
-      "grad_norm": 0.0011525237932801247,
-      "learning_rate": 0.00019999460099151913,
-      "loss": 46.0,
-      "step": 43272
-    },
-    {
-      "epoch": 3.308523042223369,
-      "grad_norm": 0.0006031316588632762,
-      "learning_rate": 0.00019999460074192262,
-      "loss": 46.0,
-      "step": 43273
-    },
-    {
-      "epoch": 3.3085994992067587,
-      "grad_norm": 0.009976615197956562,
-      "learning_rate": 0.00019999460049232033,
-      "loss": 46.0,
-      "step": 43274
-    },
-    {
-      "epoch": 3.3086759561901484,
-      "grad_norm": 0.0021152389235794544,
-      "learning_rate": 0.00019999460024271225,
-      "loss": 46.0,
-      "step": 43275
-    },
-    {
-      "epoch": 3.308752413173538,
-      "grad_norm": 0.0017483679112046957,
-      "learning_rate": 0.00019999459999309841,
-      "loss": 46.0,
-      "step": 43276
-    },
-    {
-      "epoch": 3.308828870156928,
-      "grad_norm": 0.00484770443290472,
-      "learning_rate": 0.00019999459974347886,
-      "loss": 46.0,
-      "step": 43277
-    },
-    {
-      "epoch": 3.3089053271403177,
-      "grad_norm": 0.0010415568249300122,
-      "learning_rate": 0.00019999459949385345,
-      "loss": 46.0,
-      "step": 43278
-    },
-    {
-      "epoch": 3.3089817841237075,
-      "grad_norm": 0.002279620384797454,
-      "learning_rate": 0.00019999459924422236,
-      "loss": 46.0,
-      "step": 43279
-    },
-    {
-      "epoch": 3.3090582411070972,
-      "grad_norm": 0.002856149571016431,
-      "learning_rate": 0.00019999459899458543,
-      "loss": 46.0,
-      "step": 43280
-    },
-    {
-      "epoch": 3.309134698090487,
-      "grad_norm": 0.0012341402471065521,
-      "learning_rate": 0.00019999459874494276,
-      "loss": 46.0,
-      "step": 43281
-    },
-    {
-      "epoch": 3.3092111550738768,
-      "grad_norm": 0.004400936421006918,
-      "learning_rate": 0.0001999945984952943,
-      "loss": 46.0,
-      "step": 43282
-    },
-    {
-      "epoch": 3.3092876120572665,
-      "grad_norm": 0.0015747930156067014,
-      "learning_rate": 0.00019999459824564012,
-      "loss": 46.0,
-      "step": 43283
-    },
-    {
-      "epoch": 3.309364069040656,
-      "grad_norm": 0.004910276737064123,
-      "learning_rate": 0.00019999459799598012,
-      "loss": 46.0,
-      "step": 43284
-    },
-    {
-      "epoch": 3.3094405260240456,
-      "grad_norm": 0.001290509128011763,
-      "learning_rate": 0.00019999459774631438,
-      "loss": 46.0,
-      "step": 43285
-    },
-    {
-      "epoch": 3.3095169830074354,
-      "grad_norm": 0.004152401350438595,
-      "learning_rate": 0.0001999945974966429,
-      "loss": 46.0,
-      "step": 43286
-    },
-    {
-      "epoch": 3.309593439990825,
-      "grad_norm": 0.0035686946939677,
-      "learning_rate": 0.00019999459724696558,
-      "loss": 46.0,
-      "step": 43287
-    },
-    {
-      "epoch": 3.309669896974215,
-      "grad_norm": 0.00228068302385509,
-      "learning_rate": 0.00019999459699728252,
-      "loss": 46.0,
-      "step": 43288
-    },
-    {
-      "epoch": 3.3097463539576046,
-      "grad_norm": 0.0010730500798672438,
-      "learning_rate": 0.00019999459674759372,
-      "loss": 46.0,
-      "step": 43289
-    },
-    {
-      "epoch": 3.3098228109409944,
-      "grad_norm": 0.004837107378989458,
-      "learning_rate": 0.0001999945964978991,
-      "loss": 46.0,
-      "step": 43290
-    },
-    {
-      "epoch": 3.309899267924384,
-      "grad_norm": 0.0012564780190587044,
-      "learning_rate": 0.00019999459624819873,
-      "loss": 46.0,
-      "step": 43291
-    },
-    {
-      "epoch": 3.309975724907774,
-      "grad_norm": 0.00089595268946141,
-      "learning_rate": 0.0001999945959984926,
-      "loss": 46.0,
-      "step": 43292
-    },
-    {
-      "epoch": 3.3100521818911632,
-      "grad_norm": 0.0011313320137560368,
-      "learning_rate": 0.00019999459574878073,
-      "loss": 46.0,
-      "step": 43293
-    },
-    {
-      "epoch": 3.310128638874553,
-      "grad_norm": 0.0031332937069237232,
-      "learning_rate": 0.00019999459549906306,
-      "loss": 46.0,
-      "step": 43294
-    },
-    {
-      "epoch": 3.3102050958579428,
-      "grad_norm": 0.0017274945275858045,
-      "learning_rate": 0.00019999459524933962,
-      "loss": 46.0,
-      "step": 43295
-    },
-    {
-      "epoch": 3.3102815528413325,
-      "grad_norm": 0.004312107805162668,
-      "learning_rate": 0.0001999945949996104,
-      "loss": 46.0,
-      "step": 43296
-    },
-    {
-      "epoch": 3.3103580098247223,
-      "grad_norm": 0.0017059976235032082,
-      "learning_rate": 0.0001999945947498754,
-      "loss": 46.0,
-      "step": 43297
-    },
-    {
-      "epoch": 3.310434466808112,
-      "grad_norm": 0.0015957513824105263,
-      "learning_rate": 0.00019999459450013466,
-      "loss": 46.0,
-      "step": 43298
-    },
-    {
-      "epoch": 3.310510923791502,
-      "grad_norm": 0.00199304916895926,
-      "learning_rate": 0.00019999459425038815,
-      "loss": 46.0,
-      "step": 43299
-    },
-    {
-      "epoch": 3.3105873807748916,
-      "grad_norm": 0.0012294353218749166,
-      "learning_rate": 0.00019999459400063587,
-      "loss": 46.0,
-      "step": 43300
-    },
-    {
-      "epoch": 3.3106638377582813,
-      "grad_norm": 0.002053462667390704,
-      "learning_rate": 0.0001999945937508778,
-      "loss": 46.0,
-      "step": 43301
-    },
-    {
-      "epoch": 3.310740294741671,
-      "grad_norm": 0.0006750842439942062,
-      "learning_rate": 0.000199994593501114,
-      "loss": 46.0,
-      "step": 43302
-    },
-    {
-      "epoch": 3.310816751725061,
-      "grad_norm": 0.0015960498712956905,
-      "learning_rate": 0.0001999945932513444,
-      "loss": 46.0,
-      "step": 43303
-    },
-    {
-      "epoch": 3.3108932087084506,
-      "grad_norm": 0.001992043573409319,
-      "learning_rate": 0.00019999459300156902,
-      "loss": 46.0,
-      "step": 43304
-    },
-    {
-      "epoch": 3.3109696656918404,
-      "grad_norm": 0.0009937369031831622,
-      "learning_rate": 0.0001999945927517879,
-      "loss": 46.0,
-      "step": 43305
-    },
-    {
-      "epoch": 3.3110461226752297,
-      "grad_norm": 0.0007268227636814117,
-      "learning_rate": 0.000199994592502001,
-      "loss": 46.0,
-      "step": 43306
-    },
-    {
-      "epoch": 3.3111225796586194,
-      "grad_norm": 0.002949135610833764,
-      "learning_rate": 0.00019999459225220835,
-      "loss": 46.0,
-      "step": 43307
-    },
-    {
-      "epoch": 3.311199036642009,
-      "grad_norm": 0.0010290266945958138,
-      "learning_rate": 0.0001999945920024099,
-      "loss": 46.0,
-      "step": 43308
-    },
-    {
-      "epoch": 3.311275493625399,
-      "grad_norm": 0.010142185725271702,
-      "learning_rate": 0.00019999459175260568,
-      "loss": 46.0,
-      "step": 43309
-    },
-    {
-      "epoch": 3.3113519506087887,
-      "grad_norm": 0.0007305838516913354,
-      "learning_rate": 0.00019999459150279571,
-      "loss": 46.0,
-      "step": 43310
-    },
-    {
-      "epoch": 3.3114284075921785,
-      "grad_norm": 0.00721488194540143,
-      "learning_rate": 0.00019999459125297998,
-      "loss": 46.0,
-      "step": 43311
-    },
-    {
-      "epoch": 3.3115048645755683,
-      "grad_norm": 0.0011465094285085797,
-      "learning_rate": 0.00019999459100315844,
-      "loss": 46.0,
-      "step": 43312
-    },
-    {
-      "epoch": 3.311581321558958,
-      "grad_norm": 0.0020851767621934414,
-      "learning_rate": 0.00019999459075333115,
-      "loss": 46.0,
-      "step": 43313
-    },
-    {
-      "epoch": 3.3116577785423478,
-      "grad_norm": 0.0022980738431215286,
-      "learning_rate": 0.00019999459050349812,
-      "loss": 46.0,
-      "step": 43314
-    },
-    {
-      "epoch": 3.311734235525737,
-      "grad_norm": 0.0013065822422504425,
-      "learning_rate": 0.0001999945902536593,
-      "loss": 46.0,
-      "step": 43315
-    },
-    {
-      "epoch": 3.311810692509127,
-      "grad_norm": 0.001578980009071529,
-      "learning_rate": 0.0001999945900038147,
-      "loss": 46.0,
-      "step": 43316
-    },
-    {
-      "epoch": 3.3118871494925166,
-      "grad_norm": 0.0016919063637033105,
-      "learning_rate": 0.00019999458975396436,
-      "loss": 46.0,
-      "step": 43317
-    },
-    {
-      "epoch": 3.3119636064759064,
-      "grad_norm": 0.0006471959641203284,
-      "learning_rate": 0.0001999945895041082,
-      "loss": 46.0,
-      "step": 43318
-    },
-    {
-      "epoch": 3.312040063459296,
-      "grad_norm": 0.002068018075078726,
-      "learning_rate": 0.0001999945892542463,
-      "loss": 46.0,
-      "step": 43319
-    },
-    {
-      "epoch": 3.312116520442686,
-      "grad_norm": 0.0021072272211313248,
-      "learning_rate": 0.00019999458900437863,
-      "loss": 46.0,
-      "step": 43320
-    },
-    {
-      "epoch": 3.3121929774260757,
-      "grad_norm": 0.0017072216141968966,
-      "learning_rate": 0.00019999458875450521,
-      "loss": 46.0,
-      "step": 43321
-    },
-    {
-      "epoch": 3.3122694344094654,
-      "grad_norm": 0.003044598735868931,
-      "learning_rate": 0.00019999458850462597,
-      "loss": 46.0,
-      "step": 43322
-    },
-    {
-      "epoch": 3.312345891392855,
-      "grad_norm": 0.0005900660180486739,
-      "learning_rate": 0.00019999458825474103,
-      "loss": 46.0,
-      "step": 43323
-    },
-    {
-      "epoch": 3.312422348376245,
-      "grad_norm": 0.002818033564835787,
-      "learning_rate": 0.00019999458800485026,
-      "loss": 46.0,
-      "step": 43324
-    },
-    {
-      "epoch": 3.3124988053596347,
-      "grad_norm": 0.0017469549784436822,
-      "learning_rate": 0.00019999458775495375,
-      "loss": 46.0,
-      "step": 43325
-    },
-    {
-      "epoch": 3.3125752623430245,
-      "grad_norm": 0.000833536556456238,
-      "learning_rate": 0.00019999458750505147,
-      "loss": 46.0,
-      "step": 43326
-    },
-    {
-      "epoch": 3.3126517193264142,
-      "grad_norm": 0.0017317094607278705,
-      "learning_rate": 0.0001999945872551434,
-      "loss": 46.0,
-      "step": 43327
-    },
-    {
-      "epoch": 3.3127281763098035,
-      "grad_norm": 0.004800363909453154,
-      "learning_rate": 0.0001999945870052296,
-      "loss": 46.0,
-      "step": 43328
-    },
-    {
-      "epoch": 3.3128046332931933,
-      "grad_norm": 0.0011750849662348628,
-      "learning_rate": 0.00019999458675531,
-      "loss": 46.0,
-      "step": 43329
-    },
-    {
-      "epoch": 3.312881090276583,
-      "grad_norm": 0.010479213669896126,
-      "learning_rate": 0.00019999458650538465,
-      "loss": 46.0,
-      "step": 43330
-    },
-    {
-      "epoch": 3.312957547259973,
-      "grad_norm": 0.0020926264114677906,
-      "learning_rate": 0.00019999458625545352,
-      "loss": 46.0,
-      "step": 43331
-    },
-    {
-      "epoch": 3.3130340042433626,
-      "grad_norm": 0.0014867031713947654,
-      "learning_rate": 0.00019999458600551662,
-      "loss": 46.0,
-      "step": 43332
-    },
-    {
-      "epoch": 3.3131104612267523,
-      "grad_norm": 0.0039025195874273777,
-      "learning_rate": 0.00019999458575557398,
-      "loss": 46.0,
-      "step": 43333
-    },
-    {
-      "epoch": 3.313186918210142,
-      "grad_norm": 0.0014592184452340007,
-      "learning_rate": 0.0001999945855056255,
-      "loss": 46.0,
-      "step": 43334
-    },
-    {
-      "epoch": 3.313263375193532,
-      "grad_norm": 0.004011046141386032,
-      "learning_rate": 0.00019999458525567131,
-      "loss": 46.0,
-      "step": 43335
-    },
-    {
-      "epoch": 3.3133398321769216,
-      "grad_norm": 0.0019452678970992565,
-      "learning_rate": 0.00019999458500571135,
-      "loss": 46.0,
-      "step": 43336
-    },
-    {
-      "epoch": 3.313416289160311,
-      "grad_norm": 0.001178641919977963,
-      "learning_rate": 0.00019999458475574558,
-      "loss": 46.0,
-      "step": 43337
-    },
-    {
-      "epoch": 3.3134927461437007,
-      "grad_norm": 0.0022635338827967644,
-      "learning_rate": 0.00019999458450577407,
-      "loss": 46.0,
-      "step": 43338
-    },
-    {
-      "epoch": 3.3135692031270905,
-      "grad_norm": 0.002230064244940877,
-      "learning_rate": 0.00019999458425579682,
-      "loss": 46.0,
-      "step": 43339
-    },
-    {
-      "epoch": 3.3136456601104802,
-      "grad_norm": 0.003249509958550334,
-      "learning_rate": 0.00019999458400581376,
-      "loss": 46.0,
-      "step": 43340
-    },
-    {
-      "epoch": 3.31372211709387,
-      "grad_norm": 0.0023734376300126314,
-      "learning_rate": 0.00019999458375582493,
-      "loss": 46.0,
-      "step": 43341
-    },
-    {
-      "epoch": 3.3137985740772598,
-      "grad_norm": 0.0012363401474431157,
-      "learning_rate": 0.00019999458350583035,
-      "loss": 46.0,
-      "step": 43342
-    },
-    {
-      "epoch": 3.3138750310606495,
-      "grad_norm": 0.001281177275814116,
-      "learning_rate": 0.00019999458325582997,
-      "loss": 46.0,
-      "step": 43343
-    },
-    {
-      "epoch": 3.3139514880440393,
-      "grad_norm": 0.004189579281955957,
-      "learning_rate": 0.00019999458300582384,
-      "loss": 46.0,
-      "step": 43344
-    },
-    {
-      "epoch": 3.314027945027429,
-      "grad_norm": 0.0024599258322268724,
-      "learning_rate": 0.00019999458275581195,
-      "loss": 46.0,
-      "step": 43345
-    },
-    {
-      "epoch": 3.314104402010819,
-      "grad_norm": 0.0006552319973707199,
-      "learning_rate": 0.00019999458250579427,
-      "loss": 46.0,
-      "step": 43346
-    },
-    {
-      "epoch": 3.3141808589942086,
-      "grad_norm": 0.0029775877483189106,
-      "learning_rate": 0.00019999458225577086,
-      "loss": 46.0,
-      "step": 43347
-    },
-    {
-      "epoch": 3.3142573159775983,
-      "grad_norm": 0.003050489816814661,
-      "learning_rate": 0.00019999458200574167,
-      "loss": 46.0,
-      "step": 43348
-    },
-    {
-      "epoch": 3.3143337729609876,
-      "grad_norm": 0.001027034013532102,
-      "learning_rate": 0.0001999945817557067,
-      "loss": 46.0,
-      "step": 43349
-    },
-    {
-      "epoch": 3.3144102299443774,
-      "grad_norm": 0.0031818593852221966,
-      "learning_rate": 0.00019999458150566594,
-      "loss": 46.0,
-      "step": 43350
-    },
-    {
-      "epoch": 3.314486686927767,
-      "grad_norm": 0.0013632456539198756,
-      "learning_rate": 0.00019999458125561942,
-      "loss": 46.0,
-      "step": 43351
-    },
-    {
-      "epoch": 3.314563143911157,
-      "grad_norm": 0.0012772209011018276,
-      "learning_rate": 0.00019999458100556714,
-      "loss": 46.0,
-      "step": 43352
-    },
-    {
-      "epoch": 3.3146396008945467,
-      "grad_norm": 0.0028334425296634436,
-      "learning_rate": 0.0001999945807555091,
-      "loss": 46.0,
-      "step": 43353
-    },
-    {
-      "epoch": 3.3147160578779364,
-      "grad_norm": 0.001498374156653881,
-      "learning_rate": 0.00019999458050544528,
-      "loss": 46.0,
-      "step": 43354
-    },
-    {
-      "epoch": 3.314792514861326,
-      "grad_norm": 0.005111016798764467,
-      "learning_rate": 0.00019999458025537567,
-      "loss": 46.0,
-      "step": 43355
-    },
-    {
-      "epoch": 3.314868971844716,
-      "grad_norm": 0.001594903995282948,
-      "learning_rate": 0.00019999458000530035,
-      "loss": 46.0,
-      "step": 43356
-    },
-    {
-      "epoch": 3.3149454288281057,
-      "grad_norm": 0.0009064633050002158,
-      "learning_rate": 0.0001999945797552192,
-      "loss": 46.0,
-      "step": 43357
-    },
-    {
-      "epoch": 3.3150218858114955,
-      "grad_norm": 0.0015657697804272175,
-      "learning_rate": 0.0001999945795051323,
-      "loss": 46.0,
-      "step": 43358
-    },
-    {
-      "epoch": 3.315098342794885,
-      "grad_norm": 0.004220201633870602,
-      "learning_rate": 0.00019999457925503963,
-      "loss": 46.0,
-      "step": 43359
-    },
-    {
-      "epoch": 3.3151747997782746,
-      "grad_norm": 0.001176894991658628,
-      "learning_rate": 0.0001999945790049412,
-      "loss": 46.0,
-      "step": 43360
-    },
-    {
-      "epoch": 3.3152512567616643,
-      "grad_norm": 0.0029704258777201176,
-      "learning_rate": 0.000199994578754837,
-      "loss": 46.0,
-      "step": 43361
-    },
-    {
-      "epoch": 3.315327713745054,
-      "grad_norm": 0.0016683146823197603,
-      "learning_rate": 0.00019999457850472703,
-      "loss": 46.0,
-      "step": 43362
-    },
-    {
-      "epoch": 3.315404170728444,
-      "grad_norm": 0.002009845105931163,
-      "learning_rate": 0.0001999945782546113,
-      "loss": 46.0,
-      "step": 43363
-    },
-    {
-      "epoch": 3.3154806277118336,
-      "grad_norm": 0.0011125633027404547,
-      "learning_rate": 0.0001999945780044898,
-      "loss": 46.0,
-      "step": 43364
-    },
-    {
-      "epoch": 3.3155570846952234,
-      "grad_norm": 0.0015573266427963972,
-      "learning_rate": 0.0001999945777543625,
-      "loss": 46.0,
-      "step": 43365
-    },
-    {
-      "epoch": 3.315633541678613,
-      "grad_norm": 0.0017093715723603964,
-      "learning_rate": 0.00019999457750422947,
-      "loss": 46.0,
-      "step": 43366
-    },
-    {
-      "epoch": 3.315709998662003,
-      "grad_norm": 0.004980104044079781,
-      "learning_rate": 0.00019999457725409063,
-      "loss": 46.0,
-      "step": 43367
-    },
-    {
-      "epoch": 3.3157864556453926,
-      "grad_norm": 0.002460876014083624,
-      "learning_rate": 0.00019999457700394606,
-      "loss": 46.0,
-      "step": 43368
-    },
-    {
-      "epoch": 3.3158629126287824,
-      "grad_norm": 0.0014098930405452847,
-      "learning_rate": 0.00019999457675379568,
-      "loss": 46.0,
-      "step": 43369
-    },
-    {
-      "epoch": 3.315939369612172,
-      "grad_norm": 0.006988766603171825,
-      "learning_rate": 0.00019999457650363956,
-      "loss": 46.0,
-      "step": 43370
-    },
-    {
-      "epoch": 3.3160158265955615,
-      "grad_norm": 0.0018488438799977303,
-      "learning_rate": 0.00019999457625347766,
-      "loss": 46.0,
-      "step": 43371
-    },
-    {
-      "epoch": 3.3160922835789512,
-      "grad_norm": 0.000904468004591763,
-      "learning_rate": 0.00019999457600331,
-      "loss": 46.0,
-      "step": 43372
-    },
-    {
-      "epoch": 3.316168740562341,
-      "grad_norm": 0.0014299922622740269,
-      "learning_rate": 0.00019999457575313657,
-      "loss": 46.0,
-      "step": 43373
-    },
-    {
-      "epoch": 3.3162451975457308,
-      "grad_norm": 0.004879137966781855,
-      "learning_rate": 0.00019999457550295738,
-      "loss": 46.0,
-      "step": 43374
-    },
-    {
-      "epoch": 3.3163216545291205,
-      "grad_norm": 0.000905583321582526,
-      "learning_rate": 0.00019999457525277242,
-      "loss": 46.0,
-      "step": 43375
-    },
-    {
-      "epoch": 3.3163981115125103,
-      "grad_norm": 0.0012610937701538205,
-      "learning_rate": 0.00019999457500258165,
-      "loss": 46.0,
-      "step": 43376
-    },
-    {
-      "epoch": 3.3164745684959,
-      "grad_norm": 0.0036018602550029755,
-      "learning_rate": 0.00019999457475238517,
-      "loss": 46.0,
-      "step": 43377
-    },
-    {
-      "epoch": 3.31655102547929,
-      "grad_norm": 0.0010846088407561183,
-      "learning_rate": 0.00019999457450218286,
-      "loss": 46.0,
-      "step": 43378
-    },
-    {
-      "epoch": 3.3166274824626796,
-      "grad_norm": 0.0012248934945091605,
-      "learning_rate": 0.00019999457425197483,
-      "loss": 46.0,
-      "step": 43379
-    },
-    {
-      "epoch": 3.3167039394460693,
-      "grad_norm": 0.0010553468018770218,
-      "learning_rate": 0.00019999457400176102,
-      "loss": 46.0,
-      "step": 43380
-    },
-    {
-      "epoch": 3.3167803964294587,
-      "grad_norm": 0.002031085779890418,
-      "learning_rate": 0.00019999457375154142,
-      "loss": 46.0,
-      "step": 43381
-    },
-    {
-      "epoch": 3.3168568534128484,
-      "grad_norm": 0.002805246040225029,
-      "learning_rate": 0.00019999457350131607,
-      "loss": 46.0,
-      "step": 43382
-    },
-    {
-      "epoch": 3.316933310396238,
-      "grad_norm": 0.0017665355699136853,
-      "learning_rate": 0.00019999457325108494,
-      "loss": 46.0,
-      "step": 43383
-    },
-    {
-      "epoch": 3.317009767379628,
-      "grad_norm": 0.0013133903266862035,
-      "learning_rate": 0.00019999457300084805,
-      "loss": 46.0,
-      "step": 43384
-    },
-    {
-      "epoch": 3.3170862243630177,
-      "grad_norm": 0.00139354367274791,
-      "learning_rate": 0.0001999945727506054,
-      "loss": 46.0,
-      "step": 43385
-    },
-    {
-      "epoch": 3.3171626813464075,
-      "grad_norm": 0.002866447437554598,
-      "learning_rate": 0.00019999457250035696,
-      "loss": 46.0,
-      "step": 43386
-    },
-    {
-      "epoch": 3.317239138329797,
-      "grad_norm": 0.0027093978133052588,
-      "learning_rate": 0.00019999457225010274,
-      "loss": 46.0,
-      "step": 43387
-    },
-    {
-      "epoch": 3.317315595313187,
-      "grad_norm": 0.0018428182229399681,
-      "learning_rate": 0.0001999945719998428,
-      "loss": 46.0,
-      "step": 43388
-    },
-    {
-      "epoch": 3.3173920522965767,
-      "grad_norm": 0.0027941951993852854,
-      "learning_rate": 0.00019999457174957704,
-      "loss": 46.0,
-      "step": 43389
-    },
-    {
-      "epoch": 3.3174685092799665,
-      "grad_norm": 0.0016378681175410748,
-      "learning_rate": 0.00019999457149930553,
-      "loss": 46.0,
-      "step": 43390
-    },
-    {
-      "epoch": 3.3175449662633563,
-      "grad_norm": 0.003005365375429392,
-      "learning_rate": 0.00019999457124902827,
-      "loss": 46.0,
-      "step": 43391
-    },
-    {
-      "epoch": 3.317621423246746,
-      "grad_norm": 0.0010542947566136718,
-      "learning_rate": 0.00019999457099874522,
-      "loss": 46.0,
-      "step": 43392
-    },
-    {
-      "epoch": 3.3176978802301353,
-      "grad_norm": 0.0013423888012766838,
-      "learning_rate": 0.0001999945707484564,
-      "loss": 46.0,
-      "step": 43393
-    },
-    {
-      "epoch": 3.317774337213525,
-      "grad_norm": 0.0013756750850006938,
-      "learning_rate": 0.0001999945704981618,
-      "loss": 46.0,
-      "step": 43394
-    },
-    {
-      "epoch": 3.317850794196915,
-      "grad_norm": 0.0016079377382993698,
-      "learning_rate": 0.00019999457024786146,
-      "loss": 46.0,
-      "step": 43395
-    },
-    {
-      "epoch": 3.3179272511803046,
-      "grad_norm": 0.0007803951157256961,
-      "learning_rate": 0.00019999456999755534,
-      "loss": 46.0,
-      "step": 43396
-    },
-    {
-      "epoch": 3.3180037081636944,
-      "grad_norm": 0.003149746684357524,
-      "learning_rate": 0.00019999456974724344,
-      "loss": 46.0,
-      "step": 43397
-    },
-    {
-      "epoch": 3.318080165147084,
-      "grad_norm": 0.005063454154878855,
-      "learning_rate": 0.0001999945694969258,
-      "loss": 46.0,
-      "step": 43398
-    },
-    {
-      "epoch": 3.318156622130474,
-      "grad_norm": 0.0076774712651968,
-      "learning_rate": 0.00019999456924660235,
-      "loss": 46.0,
-      "step": 43399
-    },
-    {
-      "epoch": 3.3182330791138637,
-      "grad_norm": 0.001998207299038768,
-      "learning_rate": 0.00019999456899627316,
-      "loss": 46.0,
-      "step": 43400
-    },
-    {
-      "epoch": 3.3183095360972534,
-      "grad_norm": 0.001289700623601675,
-      "learning_rate": 0.00019999456874593817,
-      "loss": 46.0,
-      "step": 43401
-    },
-    {
-      "epoch": 3.318385993080643,
-      "grad_norm": 0.0027638860046863556,
-      "learning_rate": 0.00019999456849559746,
-      "loss": 46.0,
-      "step": 43402
-    },
-    {
-      "epoch": 3.3184624500640325,
-      "grad_norm": 0.0012664938112720847,
-      "learning_rate": 0.00019999456824525093,
-      "loss": 46.0,
-      "step": 43403
-    },
-    {
-      "epoch": 3.3185389070474223,
-      "grad_norm": 0.0010289938654750586,
-      "learning_rate": 0.00019999456799489867,
-      "loss": 46.0,
-      "step": 43404
-    },
-    {
-      "epoch": 3.318615364030812,
-      "grad_norm": 0.0012244055978953838,
-      "learning_rate": 0.0001999945677445406,
-      "loss": 46.0,
-      "step": 43405
-    },
-    {
-      "epoch": 3.318691821014202,
-      "grad_norm": 0.0036576027050614357,
-      "learning_rate": 0.0001999945674941768,
-      "loss": 46.0,
-      "step": 43406
-    },
-    {
-      "epoch": 3.3187682779975916,
-      "grad_norm": 0.0018814962822943926,
-      "learning_rate": 0.00019999456724380723,
-      "loss": 46.0,
-      "step": 43407
-    },
-    {
-      "epoch": 3.3188447349809813,
-      "grad_norm": 0.007668083067983389,
-      "learning_rate": 0.00019999456699343186,
-      "loss": 46.0,
-      "step": 43408
-    },
-    {
-      "epoch": 3.318921191964371,
-      "grad_norm": 0.0017301782499998808,
-      "learning_rate": 0.00019999456674305076,
-      "loss": 46.0,
-      "step": 43409
-    },
-    {
-      "epoch": 3.318997648947761,
-      "grad_norm": 0.0010808518854901195,
-      "learning_rate": 0.00019999456649266386,
-      "loss": 46.0,
-      "step": 43410
-    },
-    {
-      "epoch": 3.3190741059311506,
-      "grad_norm": 0.0018614014843478799,
-      "learning_rate": 0.0001999945662422712,
-      "loss": 46.0,
-      "step": 43411
-    },
-    {
-      "epoch": 3.3191505629145404,
-      "grad_norm": 0.0018076942069455981,
-      "learning_rate": 0.00019999456599187275,
-      "loss": 46.0,
-      "step": 43412
-    },
-    {
-      "epoch": 3.31922701989793,
-      "grad_norm": 0.0014691805699840188,
-      "learning_rate": 0.0001999945657414686,
-      "loss": 46.0,
-      "step": 43413
-    },
-    {
-      "epoch": 3.31930347688132,
-      "grad_norm": 0.006432017311453819,
-      "learning_rate": 0.0001999945654910586,
-      "loss": 46.0,
-      "step": 43414
-    },
-    {
-      "epoch": 3.319379933864709,
-      "grad_norm": 0.000985596445389092,
-      "learning_rate": 0.00019999456524064286,
-      "loss": 46.0,
-      "step": 43415
-    },
-    {
-      "epoch": 3.319456390848099,
-      "grad_norm": 0.0007833700510673225,
-      "learning_rate": 0.00019999456499022135,
-      "loss": 46.0,
-      "step": 43416
-    },
-    {
-      "epoch": 3.3195328478314887,
-      "grad_norm": 0.001098829903639853,
-      "learning_rate": 0.00019999456473979407,
-      "loss": 46.0,
-      "step": 43417
-    },
-    {
-      "epoch": 3.3196093048148785,
-      "grad_norm": 0.0013725137105211616,
-      "learning_rate": 0.00019999456448936104,
-      "loss": 46.0,
-      "step": 43418
-    },
-    {
-      "epoch": 3.3196857617982682,
-      "grad_norm": 0.0014277450973168015,
-      "learning_rate": 0.0001999945642389222,
-      "loss": 46.0,
-      "step": 43419
-    },
-    {
-      "epoch": 3.319762218781658,
-      "grad_norm": 0.0020582827273756266,
-      "learning_rate": 0.00019999456398847763,
-      "loss": 46.0,
-      "step": 43420
-    },
-    {
-      "epoch": 3.3198386757650478,
-      "grad_norm": 0.0021607165690511465,
-      "learning_rate": 0.00019999456373802728,
-      "loss": 46.0,
-      "step": 43421
-    },
-    {
-      "epoch": 3.3199151327484375,
-      "grad_norm": 0.0016946293180808425,
-      "learning_rate": 0.00019999456348757116,
-      "loss": 46.0,
-      "step": 43422
-    },
-    {
-      "epoch": 3.3199915897318273,
-      "grad_norm": 0.0027404495049268007,
-      "learning_rate": 0.00019999456323710927,
-      "loss": 46.0,
-      "step": 43423
-    },
-    {
-      "epoch": 3.3200680467152166,
-      "grad_norm": 0.004383963067084551,
-      "learning_rate": 0.00019999456298664162,
-      "loss": 46.0,
-      "step": 43424
-    },
-    {
-      "epoch": 3.3201445036986064,
-      "grad_norm": 0.0015503901522606611,
-      "learning_rate": 0.00019999456273616818,
-      "loss": 46.0,
-      "step": 43425
-    },
-    {
-      "epoch": 3.320220960681996,
-      "grad_norm": 0.0013477016473188996,
-      "learning_rate": 0.000199994562485689,
-      "loss": 46.0,
-      "step": 43426
-    },
-    {
-      "epoch": 3.320297417665386,
-      "grad_norm": 0.0021067361813038588,
-      "learning_rate": 0.00019999456223520403,
-      "loss": 46.0,
-      "step": 43427
-    },
-    {
-      "epoch": 3.3203738746487756,
-      "grad_norm": 0.002632143208757043,
-      "learning_rate": 0.0001999945619847133,
-      "loss": 46.0,
-      "step": 43428
-    },
-    {
-      "epoch": 3.3204503316321654,
-      "grad_norm": 0.004429586697369814,
-      "learning_rate": 0.0001999945617342168,
-      "loss": 46.0,
-      "step": 43429
-    },
-    {
-      "epoch": 3.320526788615555,
-      "grad_norm": 0.001210974995046854,
-      "learning_rate": 0.00019999456148371453,
-      "loss": 46.0,
-      "step": 43430
-    },
-    {
-      "epoch": 3.320603245598945,
-      "grad_norm": 0.0032355301082134247,
-      "learning_rate": 0.00019999456123320644,
-      "loss": 46.0,
-      "step": 43431
-    },
-    {
-      "epoch": 3.3206797025823347,
-      "grad_norm": 0.0020707454532384872,
-      "learning_rate": 0.00019999456098269267,
-      "loss": 46.0,
-      "step": 43432
-    },
-    {
-      "epoch": 3.3207561595657245,
-      "grad_norm": 0.0006517020519822836,
-      "learning_rate": 0.00019999456073217307,
-      "loss": 46.0,
-      "step": 43433
-    },
-    {
-      "epoch": 3.320832616549114,
-      "grad_norm": 0.001286494778469205,
-      "learning_rate": 0.00019999456048164774,
-      "loss": 46.0,
-      "step": 43434
-    },
-    {
-      "epoch": 3.320909073532504,
-      "grad_norm": 0.001964277122169733,
-      "learning_rate": 0.00019999456023111662,
-      "loss": 46.0,
-      "step": 43435
-    },
-    {
-      "epoch": 3.3209855305158937,
-      "grad_norm": 0.0018471956718713045,
-      "learning_rate": 0.00019999455998057972,
-      "loss": 46.0,
-      "step": 43436
-    },
-    {
-      "epoch": 3.321061987499283,
-      "grad_norm": 0.0009719212539494038,
-      "learning_rate": 0.00019999455973003706,
-      "loss": 46.0,
-      "step": 43437
-    },
-    {
-      "epoch": 3.321138444482673,
-      "grad_norm": 0.005236941855400801,
-      "learning_rate": 0.0001999945594794886,
-      "loss": 46.0,
-      "step": 43438
-    },
-    {
-      "epoch": 3.3212149014660626,
-      "grad_norm": 0.0023489014711230993,
-      "learning_rate": 0.00019999455922893445,
-      "loss": 46.0,
-      "step": 43439
-    },
-    {
-      "epoch": 3.3212913584494523,
-      "grad_norm": 0.002056208672001958,
-      "learning_rate": 0.00019999455897837446,
-      "loss": 46.0,
-      "step": 43440
-    },
-    {
-      "epoch": 3.321367815432842,
-      "grad_norm": 0.001000122050754726,
-      "learning_rate": 0.00019999455872780875,
-      "loss": 46.0,
-      "step": 43441
-    },
-    {
-      "epoch": 3.321444272416232,
-      "grad_norm": 0.002598110120743513,
-      "learning_rate": 0.00019999455847723722,
-      "loss": 46.0,
-      "step": 43442
-    },
-    {
-      "epoch": 3.3215207293996216,
-      "grad_norm": 0.001369926962070167,
-      "learning_rate": 0.00019999455822666,
-      "loss": 46.0,
-      "step": 43443
-    },
-    {
-      "epoch": 3.3215971863830114,
-      "grad_norm": 0.0024704348761588335,
-      "learning_rate": 0.0001999945579760769,
-      "loss": 46.0,
-      "step": 43444
-    },
-    {
-      "epoch": 3.321673643366401,
-      "grad_norm": 0.000950091693084687,
-      "learning_rate": 0.00019999455772548813,
-      "loss": 46.0,
-      "step": 43445
-    },
-    {
-      "epoch": 3.3217501003497905,
-      "grad_norm": 0.002488666446879506,
-      "learning_rate": 0.00019999455747489353,
-      "loss": 46.0,
-      "step": 43446
-    },
-    {
-      "epoch": 3.32182655733318,
-      "grad_norm": 0.001453764969483018,
-      "learning_rate": 0.00019999455722429318,
-      "loss": 46.0,
-      "step": 43447
-    },
-    {
-      "epoch": 3.32190301431657,
-      "grad_norm": 0.007059496361762285,
-      "learning_rate": 0.00019999455697368706,
-      "loss": 46.0,
-      "step": 43448
-    },
-    {
-      "epoch": 3.3219794712999597,
-      "grad_norm": 0.0013031256385147572,
-      "learning_rate": 0.00019999455672307516,
-      "loss": 46.0,
-      "step": 43449
-    },
-    {
-      "epoch": 3.3220559282833495,
-      "grad_norm": 0.001736507867462933,
-      "learning_rate": 0.0001999945564724575,
-      "loss": 46.0,
-      "step": 43450
-    },
-    {
-      "epoch": 3.3221323852667393,
-      "grad_norm": 0.002132937777787447,
-      "learning_rate": 0.0001999945562218341,
-      "loss": 46.0,
-      "step": 43451
-    },
-    {
-      "epoch": 3.322208842250129,
-      "grad_norm": 0.002812823746353388,
-      "learning_rate": 0.00019999455597120492,
-      "loss": 46.0,
-      "step": 43452
-    },
-    {
-      "epoch": 3.322285299233519,
-      "grad_norm": 0.005177715793251991,
-      "learning_rate": 0.00019999455572056993,
-      "loss": 46.0,
-      "step": 43453
-    },
-    {
-      "epoch": 3.3223617562169085,
-      "grad_norm": 0.002652070950716734,
-      "learning_rate": 0.0001999945554699292,
-      "loss": 46.0,
-      "step": 43454
-    },
-    {
-      "epoch": 3.3224382132002983,
-      "grad_norm": 0.0009143145871348679,
-      "learning_rate": 0.0001999945552192827,
-      "loss": 46.0,
-      "step": 43455
-    },
-    {
-      "epoch": 3.322514670183688,
-      "grad_norm": 0.0012752431211993098,
-      "learning_rate": 0.00019999455496863045,
-      "loss": 46.0,
-      "step": 43456
-    },
-    {
-      "epoch": 3.322591127167078,
-      "grad_norm": 0.0007557413773611188,
-      "learning_rate": 0.00019999455471797237,
-      "loss": 46.0,
-      "step": 43457
-    },
-    {
-      "epoch": 3.3226675841504676,
-      "grad_norm": 0.0033525999169796705,
-      "learning_rate": 0.00019999455446730857,
-      "loss": 46.0,
-      "step": 43458
-    },
-    {
-      "epoch": 3.322744041133857,
-      "grad_norm": 0.0038206998724490404,
-      "learning_rate": 0.000199994554216639,
-      "loss": 46.0,
-      "step": 43459
-    },
-    {
-      "epoch": 3.3228204981172467,
-      "grad_norm": 0.0028049773536622524,
-      "learning_rate": 0.00019999455396596365,
-      "loss": 46.0,
-      "step": 43460
-    },
-    {
-      "epoch": 3.3228969551006364,
-      "grad_norm": 0.002209616359323263,
-      "learning_rate": 0.00019999455371528252,
-      "loss": 46.0,
-      "step": 43461
-    },
-    {
-      "epoch": 3.322973412084026,
-      "grad_norm": 0.004158904310315847,
-      "learning_rate": 0.00019999455346459563,
-      "loss": 46.0,
-      "step": 43462
-    },
-    {
-      "epoch": 3.323049869067416,
-      "grad_norm": 0.0022466732189059258,
-      "learning_rate": 0.000199994553213903,
-      "loss": 46.0,
-      "step": 43463
-    },
-    {
-      "epoch": 3.3231263260508057,
-      "grad_norm": 0.006654097232967615,
-      "learning_rate": 0.00019999455296320457,
-      "loss": 46.0,
-      "step": 43464
-    },
-    {
-      "epoch": 3.3232027830341955,
-      "grad_norm": 0.0010583935072645545,
-      "learning_rate": 0.00019999455271250036,
-      "loss": 46.0,
-      "step": 43465
-    },
-    {
-      "epoch": 3.3232792400175852,
-      "grad_norm": 0.000667455664370209,
-      "learning_rate": 0.00019999455246179042,
-      "loss": 46.0,
-      "step": 43466
-    },
-    {
-      "epoch": 3.323355697000975,
-      "grad_norm": 0.0027885313611477613,
-      "learning_rate": 0.0001999945522110747,
-      "loss": 46.0,
-      "step": 43467
-    },
-    {
-      "epoch": 3.3234321539843643,
-      "grad_norm": 0.0023187026381492615,
-      "learning_rate": 0.00019999455196035318,
-      "loss": 46.0,
-      "step": 43468
-    },
-    {
-      "epoch": 3.323508610967754,
-      "grad_norm": 0.0014646988129243255,
-      "learning_rate": 0.0001999945517096259,
-      "loss": 46.0,
-      "step": 43469
-    },
-    {
-      "epoch": 3.323585067951144,
-      "grad_norm": 0.002519245259463787,
-      "learning_rate": 0.00019999455145889287,
-      "loss": 46.0,
-      "step": 43470
-    },
-    {
-      "epoch": 3.3236615249345336,
-      "grad_norm": 0.0015073021641001105,
-      "learning_rate": 0.00019999455120815407,
-      "loss": 46.0,
-      "step": 43471
-    },
-    {
-      "epoch": 3.3237379819179234,
-      "grad_norm": 0.0026442264206707478,
-      "learning_rate": 0.0001999945509574095,
-      "loss": 46.0,
-      "step": 43472
-    },
-    {
-      "epoch": 3.323814438901313,
-      "grad_norm": 0.0029849254060536623,
-      "learning_rate": 0.00019999455070665914,
-      "loss": 46.0,
-      "step": 43473
-    },
-    {
-      "epoch": 3.323890895884703,
-      "grad_norm": 0.0021172654815018177,
-      "learning_rate": 0.000199994550455903,
-      "loss": 46.0,
-      "step": 43474
-    },
-    {
-      "epoch": 3.3239673528680926,
-      "grad_norm": 0.001788344350643456,
-      "learning_rate": 0.00019999455020514113,
-      "loss": 46.0,
-      "step": 43475
-    },
-    {
-      "epoch": 3.3240438098514824,
-      "grad_norm": 0.0015312808100134134,
-      "learning_rate": 0.0001999945499543735,
-      "loss": 46.0,
-      "step": 43476
-    },
-    {
-      "epoch": 3.324120266834872,
-      "grad_norm": 0.0009136635926552117,
-      "learning_rate": 0.00019999454970360005,
-      "loss": 46.0,
-      "step": 43477
-    },
-    {
-      "epoch": 3.324196723818262,
-      "grad_norm": 0.0013022496132180095,
-      "learning_rate": 0.00019999454945282086,
-      "loss": 46.0,
-      "step": 43478
-    },
-    {
-      "epoch": 3.3242731808016517,
-      "grad_norm": 0.0030702543444931507,
-      "learning_rate": 0.0001999945492020359,
-      "loss": 46.0,
-      "step": 43479
-    },
-    {
-      "epoch": 3.324349637785041,
-      "grad_norm": 0.0025544846430420876,
-      "learning_rate": 0.00019999454895124516,
-      "loss": 46.0,
-      "step": 43480
-    },
-    {
-      "epoch": 3.3244260947684308,
-      "grad_norm": 0.0015385471051558852,
-      "learning_rate": 0.00019999454870044868,
-      "loss": 46.0,
-      "step": 43481
-    },
-    {
-      "epoch": 3.3245025517518205,
-      "grad_norm": 0.002480618190020323,
-      "learning_rate": 0.0001999945484496464,
-      "loss": 46.0,
-      "step": 43482
-    },
-    {
-      "epoch": 3.3245790087352103,
-      "grad_norm": 0.004908351693302393,
-      "learning_rate": 0.00019999454819883837,
-      "loss": 46.0,
-      "step": 43483
-    },
-    {
-      "epoch": 3.3246554657186,
-      "grad_norm": 0.0006085407803766429,
-      "learning_rate": 0.00019999454794802457,
-      "loss": 46.0,
-      "step": 43484
-    },
-    {
-      "epoch": 3.32473192270199,
-      "grad_norm": 0.0011776385363191366,
-      "learning_rate": 0.00019999454769720497,
-      "loss": 46.0,
-      "step": 43485
-    },
-    {
-      "epoch": 3.3248083796853796,
-      "grad_norm": 0.001425719354301691,
-      "learning_rate": 0.00019999454744637962,
-      "loss": 46.0,
-      "step": 43486
-    },
-    {
-      "epoch": 3.3248848366687693,
-      "grad_norm": 0.00261881691403687,
-      "learning_rate": 0.00019999454719554853,
-      "loss": 46.0,
-      "step": 43487
-    },
-    {
-      "epoch": 3.324961293652159,
-      "grad_norm": 0.0017293972196057439,
-      "learning_rate": 0.00019999454694471163,
-      "loss": 46.0,
-      "step": 43488
-    },
-    {
-      "epoch": 3.325037750635549,
-      "grad_norm": 0.00608735391870141,
-      "learning_rate": 0.000199994546693869,
-      "loss": 46.0,
-      "step": 43489
-    },
-    {
-      "epoch": 3.325114207618938,
-      "grad_norm": 0.0010004539508372545,
-      "learning_rate": 0.00019999454644302055,
-      "loss": 46.0,
-      "step": 43490
-    },
-    {
-      "epoch": 3.325190664602328,
-      "grad_norm": 0.007039217744022608,
-      "learning_rate": 0.00019999454619216636,
-      "loss": 46.0,
-      "step": 43491
-    },
-    {
-      "epoch": 3.3252671215857177,
-      "grad_norm": 0.0023110767360776663,
-      "learning_rate": 0.0001999945459413064,
-      "loss": 46.0,
-      "step": 43492
-    },
-    {
-      "epoch": 3.3253435785691074,
-      "grad_norm": 0.002817175816744566,
-      "learning_rate": 0.00019999454569044067,
-      "loss": 46.0,
-      "step": 43493
-    },
-    {
-      "epoch": 3.325420035552497,
-      "grad_norm": 0.0013660419499501586,
-      "learning_rate": 0.00019999454543956916,
-      "loss": 46.0,
-      "step": 43494
-    },
-    {
-      "epoch": 3.325496492535887,
-      "grad_norm": 0.0023869769647717476,
-      "learning_rate": 0.0001999945451886919,
-      "loss": 46.0,
-      "step": 43495
-    },
-    {
-      "epoch": 3.3255729495192767,
-      "grad_norm": 0.001902631833218038,
-      "learning_rate": 0.00019999454493780888,
-      "loss": 46.0,
-      "step": 43496
-    },
-    {
-      "epoch": 3.3256494065026665,
-      "grad_norm": 0.0009875389514490962,
-      "learning_rate": 0.00019999454468692008,
-      "loss": 46.0,
-      "step": 43497
-    },
-    {
-      "epoch": 3.3257258634860563,
-      "grad_norm": 0.001548259286209941,
-      "learning_rate": 0.00019999454443602547,
-      "loss": 46.0,
-      "step": 43498
-    },
-    {
-      "epoch": 3.325802320469446,
-      "grad_norm": 0.005543024279177189,
-      "learning_rate": 0.00019999454418512515,
-      "loss": 46.0,
-      "step": 43499
-    },
-    {
-      "epoch": 3.3258787774528358,
-      "grad_norm": 0.003943361807614565,
-      "learning_rate": 0.00019999454393421903,
-      "loss": 46.0,
-      "step": 43500
-    },
-    {
-      "epoch": 3.3259552344362255,
-      "grad_norm": 0.017705580219626427,
-      "learning_rate": 0.00019999454368330714,
-      "loss": 46.0,
-      "step": 43501
-    },
-    {
-      "epoch": 3.326031691419615,
-      "grad_norm": 0.0012057377025485039,
-      "learning_rate": 0.0001999945434323895,
-      "loss": 46.0,
-      "step": 43502
-    },
-    {
-      "epoch": 3.3261081484030046,
-      "grad_norm": 0.001260724849998951,
-      "learning_rate": 0.00019999454318146608,
-      "loss": 46.0,
-      "step": 43503
-    },
-    {
-      "epoch": 3.3261846053863944,
-      "grad_norm": 0.0038510174490511417,
-      "learning_rate": 0.0001999945429305369,
-      "loss": 46.0,
-      "step": 43504
-    },
-    {
-      "epoch": 3.326261062369784,
-      "grad_norm": 0.001594186876900494,
-      "learning_rate": 0.00019999454267960194,
-      "loss": 46.0,
-      "step": 43505
-    },
-    {
-      "epoch": 3.326337519353174,
-      "grad_norm": 0.0012539097806438804,
-      "learning_rate": 0.0001999945424286612,
-      "loss": 46.0,
-      "step": 43506
-    },
-    {
-      "epoch": 3.3264139763365637,
-      "grad_norm": 0.0032835793681442738,
-      "learning_rate": 0.0001999945421777147,
-      "loss": 46.0,
-      "step": 43507
-    },
-    {
-      "epoch": 3.3264904333199534,
-      "grad_norm": 0.0013495655730366707,
-      "learning_rate": 0.00019999454192676244,
-      "loss": 46.0,
-      "step": 43508
-    },
-    {
-      "epoch": 3.326566890303343,
-      "grad_norm": 0.0013800951419398189,
-      "learning_rate": 0.0001999945416758044,
-      "loss": 46.0,
-      "step": 43509
-    },
-    {
-      "epoch": 3.326643347286733,
-      "grad_norm": 0.0017246020724996924,
-      "learning_rate": 0.00019999454142484059,
-      "loss": 46.0,
-      "step": 43510
-    },
-    {
-      "epoch": 3.3267198042701227,
-      "grad_norm": 0.00484325410798192,
-      "learning_rate": 0.000199994541173871,
-      "loss": 46.0,
-      "step": 43511
-    },
-    {
-      "epoch": 3.326796261253512,
-      "grad_norm": 0.004459098447114229,
-      "learning_rate": 0.0001999945409228957,
-      "loss": 46.0,
-      "step": 43512
-    },
-    {
-      "epoch": 3.326872718236902,
-      "grad_norm": 0.0030825668945908546,
-      "learning_rate": 0.00019999454067191457,
-      "loss": 46.0,
-      "step": 43513
-    },
-    {
-      "epoch": 3.3269491752202915,
-      "grad_norm": 0.0016127012204378843,
-      "learning_rate": 0.00019999454042092768,
-      "loss": 46.0,
-      "step": 43514
-    },
-    {
-      "epoch": 3.3270256322036813,
-      "grad_norm": 0.005958862602710724,
-      "learning_rate": 0.000199994540169935,
-      "loss": 46.0,
-      "step": 43515
-    },
-    {
-      "epoch": 3.327102089187071,
-      "grad_norm": 0.007319292984902859,
-      "learning_rate": 0.0001999945399189366,
-      "loss": 46.0,
-      "step": 43516
-    },
-    {
-      "epoch": 3.327178546170461,
-      "grad_norm": 0.0021405229344964027,
-      "learning_rate": 0.00019999453966793244,
-      "loss": 46.0,
-      "step": 43517
-    },
-    {
-      "epoch": 3.3272550031538506,
-      "grad_norm": 0.0013300413265824318,
-      "learning_rate": 0.00019999453941692248,
-      "loss": 46.0,
-      "step": 43518
-    },
-    {
-      "epoch": 3.3273314601372403,
-      "grad_norm": 0.0036079774145036936,
-      "learning_rate": 0.00019999453916590674,
-      "loss": 46.0,
-      "step": 43519
-    },
-    {
-      "epoch": 3.32740791712063,
-      "grad_norm": 0.01782381907105446,
-      "learning_rate": 0.00019999453891488523,
-      "loss": 46.0,
-      "step": 43520
-    },
-    {
-      "epoch": 3.32748437410402,
-      "grad_norm": 0.0008638953440822661,
-      "learning_rate": 0.00019999453866385798,
-      "loss": 46.0,
-      "step": 43521
-    },
-    {
-      "epoch": 3.3275608310874096,
-      "grad_norm": 0.0013219933025538921,
-      "learning_rate": 0.00019999453841282493,
-      "loss": 46.0,
-      "step": 43522
-    },
-    {
-      "epoch": 3.3276372880707994,
-      "grad_norm": 0.0019455340225249529,
-      "learning_rate": 0.00019999453816178613,
-      "loss": 46.0,
-      "step": 43523
-    },
-    {
-      "epoch": 3.3277137450541887,
-      "grad_norm": 0.0022216681391000748,
-      "learning_rate": 0.00019999453791074155,
-      "loss": 46.0,
-      "step": 43524
-    },
-    {
-      "epoch": 3.3277902020375785,
-      "grad_norm": 0.0011177087435498834,
-      "learning_rate": 0.00019999453765969123,
-      "loss": 46.0,
-      "step": 43525
-    },
-    {
-      "epoch": 3.3278666590209682,
-      "grad_norm": 0.0022819102741777897,
-      "learning_rate": 0.0001999945374086351,
-      "loss": 46.0,
-      "step": 43526
-    },
-    {
-      "epoch": 3.327943116004358,
-      "grad_norm": 0.0015371522167697549,
-      "learning_rate": 0.00019999453715757325,
-      "loss": 46.0,
-      "step": 43527
-    },
-    {
-      "epoch": 3.3280195729877478,
-      "grad_norm": 0.006041317712515593,
-      "learning_rate": 0.00019999453690650558,
-      "loss": 46.0,
-      "step": 43528
-    },
-    {
-      "epoch": 3.3280960299711375,
-      "grad_norm": 0.0029390607960522175,
-      "learning_rate": 0.00019999453665543214,
-      "loss": 46.0,
-      "step": 43529
-    },
-    {
-      "epoch": 3.3281724869545273,
-      "grad_norm": 0.0015405311714857817,
-      "learning_rate": 0.00019999453640435298,
-      "loss": 46.0,
-      "step": 43530
-    },
-    {
-      "epoch": 3.328248943937917,
-      "grad_norm": 0.002567627001553774,
-      "learning_rate": 0.00019999453615326802,
-      "loss": 46.0,
-      "step": 43531
-    },
-    {
-      "epoch": 3.328325400921307,
-      "grad_norm": 0.0012853796361014247,
-      "learning_rate": 0.00019999453590217729,
-      "loss": 46.0,
-      "step": 43532
-    },
-    {
-      "epoch": 3.3284018579046966,
-      "grad_norm": 0.003097273176535964,
-      "learning_rate": 0.0001999945356510808,
-      "loss": 46.0,
-      "step": 43533
-    },
-    {
-      "epoch": 3.328478314888086,
-      "grad_norm": 0.0008199003059417009,
-      "learning_rate": 0.00019999453539997855,
-      "loss": 46.0,
-      "step": 43534
-    },
-    {
-      "epoch": 3.3285547718714756,
-      "grad_norm": 0.0030198912136256695,
-      "learning_rate": 0.0001999945351488705,
-      "loss": 46.0,
-      "step": 43535
-    },
-    {
-      "epoch": 3.3286312288548654,
-      "grad_norm": 0.004851036239415407,
-      "learning_rate": 0.00019999453489775667,
-      "loss": 46.0,
-      "step": 43536
-    },
-    {
-      "epoch": 3.328707685838255,
-      "grad_norm": 0.0011911002220585942,
-      "learning_rate": 0.0001999945346466371,
-      "loss": 46.0,
-      "step": 43537
-    },
-    {
-      "epoch": 3.328784142821645,
-      "grad_norm": 0.0018139132298529148,
-      "learning_rate": 0.00019999453439551178,
-      "loss": 46.0,
-      "step": 43538
-    },
-    {
-      "epoch": 3.3288605998050347,
-      "grad_norm": 0.0007225653971545398,
-      "learning_rate": 0.0001999945341443807,
-      "loss": 46.0,
-      "step": 43539
-    },
-    {
-      "epoch": 3.3289370567884244,
-      "grad_norm": 0.0017102648271247745,
-      "learning_rate": 0.00019999453389324382,
-      "loss": 46.0,
-      "step": 43540
-    },
-    {
-      "epoch": 3.329013513771814,
-      "grad_norm": 0.0032987946178764105,
-      "learning_rate": 0.00019999453364210115,
-      "loss": 46.0,
-      "step": 43541
-    },
-    {
-      "epoch": 3.329089970755204,
-      "grad_norm": 0.0033472159411758184,
-      "learning_rate": 0.00019999453339095274,
-      "loss": 46.0,
-      "step": 43542
-    },
-    {
-      "epoch": 3.3291664277385937,
-      "grad_norm": 0.004375026095658541,
-      "learning_rate": 0.00019999453313979855,
-      "loss": 46.0,
-      "step": 43543
-    },
-    {
-      "epoch": 3.3292428847219835,
-      "grad_norm": 0.0013529694406315684,
-      "learning_rate": 0.0001999945328886386,
-      "loss": 46.0,
-      "step": 43544
-    },
-    {
-      "epoch": 3.3293193417053732,
-      "grad_norm": 0.0019949353300035,
-      "learning_rate": 0.00019999453263747286,
-      "loss": 46.0,
-      "step": 43545
-    },
-    {
-      "epoch": 3.3293957986887626,
-      "grad_norm": 0.0012880043359473348,
-      "learning_rate": 0.00019999453238630138,
-      "loss": 46.0,
-      "step": 43546
-    },
-    {
-      "epoch": 3.3294722556721523,
-      "grad_norm": 0.002864767564460635,
-      "learning_rate": 0.0001999945321351241,
-      "loss": 46.0,
-      "step": 43547
-    },
-    {
-      "epoch": 3.329548712655542,
-      "grad_norm": 0.0013542111264541745,
-      "learning_rate": 0.00019999453188394108,
-      "loss": 46.0,
-      "step": 43548
-    },
-    {
-      "epoch": 3.329625169638932,
-      "grad_norm": 0.0031665111891925335,
-      "learning_rate": 0.00019999453163275228,
-      "loss": 46.0,
-      "step": 43549
-    },
-    {
-      "epoch": 3.3297016266223216,
-      "grad_norm": 0.0031101955100893974,
-      "learning_rate": 0.0001999945313815577,
-      "loss": 46.0,
-      "step": 43550
-    },
-    {
-      "epoch": 3.3297780836057114,
-      "grad_norm": 0.001858170609921217,
-      "learning_rate": 0.00019999453113035736,
-      "loss": 46.0,
-      "step": 43551
-    },
-    {
-      "epoch": 3.329854540589101,
-      "grad_norm": 0.0019680005498230457,
-      "learning_rate": 0.00019999453087915126,
-      "loss": 46.0,
-      "step": 43552
-    },
-    {
-      "epoch": 3.329930997572491,
-      "grad_norm": 0.0031915337312966585,
-      "learning_rate": 0.00019999453062793937,
-      "loss": 46.0,
-      "step": 43553
-    },
-    {
-      "epoch": 3.3300074545558807,
-      "grad_norm": 0.0036522019654512405,
-      "learning_rate": 0.0001999945303767217,
-      "loss": 46.0,
-      "step": 43554
-    },
-    {
-      "epoch": 3.33008391153927,
-      "grad_norm": 0.002773649524897337,
-      "learning_rate": 0.00019999453012549832,
-      "loss": 46.0,
-      "step": 43555
-    },
-    {
-      "epoch": 3.3301603685226597,
-      "grad_norm": 0.0010565861593931913,
-      "learning_rate": 0.00019999452987426913,
-      "loss": 46.0,
-      "step": 43556
-    },
-    {
-      "epoch": 3.3302368255060495,
-      "grad_norm": 0.0014284523203969002,
-      "learning_rate": 0.00019999452962303415,
-      "loss": 46.0,
-      "step": 43557
-    },
-    {
-      "epoch": 3.3303132824894393,
-      "grad_norm": 0.002501044189557433,
-      "learning_rate": 0.00019999452937179344,
-      "loss": 46.0,
-      "step": 43558
-    },
-    {
-      "epoch": 3.330389739472829,
-      "grad_norm": 0.003685515373945236,
-      "learning_rate": 0.00019999452912054693,
-      "loss": 46.0,
-      "step": 43559
-    },
-    {
-      "epoch": 3.3304661964562188,
-      "grad_norm": 0.000850306882057339,
-      "learning_rate": 0.00019999452886929468,
-      "loss": 46.0,
-      "step": 43560
-    },
-    {
-      "epoch": 3.3305426534396085,
-      "grad_norm": 0.0026118706446141005,
-      "learning_rate": 0.00019999452861803666,
-      "loss": 46.0,
-      "step": 43561
-    },
-    {
-      "epoch": 3.3306191104229983,
-      "grad_norm": 0.0008020975510589778,
-      "learning_rate": 0.00019999452836677286,
-      "loss": 46.0,
-      "step": 43562
-    },
-    {
-      "epoch": 3.330695567406388,
-      "grad_norm": 0.0011191933881491423,
-      "learning_rate": 0.0001999945281155033,
-      "loss": 46.0,
-      "step": 43563
-    },
-    {
-      "epoch": 3.330772024389778,
-      "grad_norm": 0.002534350147470832,
-      "learning_rate": 0.00019999452786422794,
-      "loss": 46.0,
-      "step": 43564
-    },
-    {
-      "epoch": 3.3308484813731676,
-      "grad_norm": 0.0011979220435023308,
-      "learning_rate": 0.00019999452761294685,
-      "loss": 46.0,
-      "step": 43565
-    },
-    {
-      "epoch": 3.3309249383565573,
-      "grad_norm": 0.0009882686426863074,
-      "learning_rate": 0.00019999452736165995,
-      "loss": 46.0,
-      "step": 43566
-    },
-    {
-      "epoch": 3.331001395339947,
-      "grad_norm": 0.0010714128147810698,
-      "learning_rate": 0.00019999452711036731,
-      "loss": 46.0,
-      "step": 43567
-    },
-    {
-      "epoch": 3.3310778523233364,
-      "grad_norm": 0.002566568087786436,
-      "learning_rate": 0.00019999452685906893,
-      "loss": 46.0,
-      "step": 43568
-    },
-    {
-      "epoch": 3.331154309306726,
-      "grad_norm": 0.0020654734689742327,
-      "learning_rate": 0.00019999452660776472,
-      "loss": 46.0,
-      "step": 43569
-    },
-    {
-      "epoch": 3.331230766290116,
-      "grad_norm": 0.0011668906081467867,
-      "learning_rate": 0.00019999452635645478,
-      "loss": 46.0,
-      "step": 43570
-    },
-    {
-      "epoch": 3.3313072232735057,
-      "grad_norm": 0.005298765376210213,
-      "learning_rate": 0.00019999452610513905,
-      "loss": 46.0,
-      "step": 43571
-    },
-    {
-      "epoch": 3.3313836802568955,
-      "grad_norm": 0.003780979895964265,
-      "learning_rate": 0.00019999452585381754,
-      "loss": 46.0,
-      "step": 43572
-    },
-    {
-      "epoch": 3.3314601372402852,
-      "grad_norm": 0.0013556742342188954,
-      "learning_rate": 0.0001999945256024903,
-      "loss": 46.0,
-      "step": 43573
-    },
-    {
-      "epoch": 3.331536594223675,
-      "grad_norm": 0.001188464812003076,
-      "learning_rate": 0.00019999452535115727,
-      "loss": 46.0,
-      "step": 43574
-    },
-    {
-      "epoch": 3.3316130512070647,
-      "grad_norm": 0.005002522375434637,
-      "learning_rate": 0.00019999452509981847,
-      "loss": 46.0,
-      "step": 43575
-    },
-    {
-      "epoch": 3.3316895081904545,
-      "grad_norm": 0.0009228858980350196,
-      "learning_rate": 0.0001999945248484739,
-      "loss": 46.0,
-      "step": 43576
-    },
-    {
-      "epoch": 3.331765965173844,
-      "grad_norm": 0.0015604887157678604,
-      "learning_rate": 0.00019999452459712355,
-      "loss": 46.0,
-      "step": 43577
-    },
-    {
-      "epoch": 3.3318424221572336,
-      "grad_norm": 0.0022732976358383894,
-      "learning_rate": 0.00019999452434576743,
-      "loss": 46.0,
-      "step": 43578
-    },
-    {
-      "epoch": 3.3319188791406233,
-      "grad_norm": 0.0021010686177760363,
-      "learning_rate": 0.00019999452409440557,
-      "loss": 46.0,
-      "step": 43579
-    },
-    {
-      "epoch": 3.331995336124013,
-      "grad_norm": 0.0011510387994349003,
-      "learning_rate": 0.00019999452384303793,
-      "loss": 46.0,
-      "step": 43580
-    },
-    {
-      "epoch": 3.332071793107403,
-      "grad_norm": 0.0020070946775376797,
-      "learning_rate": 0.00019999452359166454,
-      "loss": 46.0,
-      "step": 43581
-    },
-    {
-      "epoch": 3.3321482500907926,
-      "grad_norm": 0.0013966704718768597,
-      "learning_rate": 0.00019999452334028536,
-      "loss": 46.0,
-      "step": 43582
-    },
-    {
-      "epoch": 3.3322247070741824,
-      "grad_norm": 0.0052477153949439526,
-      "learning_rate": 0.0001999945230889004,
-      "loss": 46.0,
-      "step": 43583
-    },
-    {
-      "epoch": 3.332301164057572,
-      "grad_norm": 0.005325807724148035,
-      "learning_rate": 0.00019999452283750967,
-      "loss": 46.0,
-      "step": 43584
-    },
-    {
-      "epoch": 3.332377621040962,
-      "grad_norm": 0.001477902987971902,
-      "learning_rate": 0.0001999945225861132,
-      "loss": 46.0,
-      "step": 43585
-    },
-    {
-      "epoch": 3.3324540780243517,
-      "grad_norm": 0.0020473082549870014,
-      "learning_rate": 0.0001999945223347109,
-      "loss": 46.0,
-      "step": 43586
-    },
-    {
-      "epoch": 3.3325305350077414,
-      "grad_norm": 0.009419417940080166,
-      "learning_rate": 0.00019999452208330288,
-      "loss": 46.0,
-      "step": 43587
-    },
-    {
-      "epoch": 3.332606991991131,
-      "grad_norm": 0.0013555992627516389,
-      "learning_rate": 0.00019999452183188908,
-      "loss": 46.0,
-      "step": 43588
-    },
-    {
-      "epoch": 3.332683448974521,
-      "grad_norm": 0.001202745595946908,
-      "learning_rate": 0.0001999945215804695,
-      "loss": 46.0,
-      "step": 43589
-    },
-    {
-      "epoch": 3.3327599059579103,
-      "grad_norm": 0.0014870216837152839,
-      "learning_rate": 0.00019999452132904417,
-      "loss": 46.0,
-      "step": 43590
-    },
-    {
-      "epoch": 3.3328363629413,
-      "grad_norm": 0.0015464737080037594,
-      "learning_rate": 0.00019999452107761308,
-      "loss": 46.0,
-      "step": 43591
-    },
-    {
-      "epoch": 3.33291281992469,
-      "grad_norm": 0.0023479051887989044,
-      "learning_rate": 0.0001999945208261762,
-      "loss": 46.0,
-      "step": 43592
-    },
-    {
-      "epoch": 3.3329892769080796,
-      "grad_norm": 0.0017061260296031833,
-      "learning_rate": 0.00019999452057473354,
-      "loss": 46.0,
-      "step": 43593
-    },
-    {
-      "epoch": 3.3330657338914693,
-      "grad_norm": 0.002244638977572322,
-      "learning_rate": 0.00019999452032328513,
-      "loss": 46.0,
-      "step": 43594
-    },
-    {
-      "epoch": 3.333142190874859,
-      "grad_norm": 0.005626108963042498,
-      "learning_rate": 0.00019999452007183095,
-      "loss": 46.0,
-      "step": 43595
-    },
-    {
-      "epoch": 3.333218647858249,
-      "grad_norm": 0.002107857493683696,
-      "learning_rate": 0.00019999451982037102,
-      "loss": 46.0,
-      "step": 43596
-    },
-    {
-      "epoch": 3.3332951048416386,
-      "grad_norm": 0.0018534853588789701,
-      "learning_rate": 0.00019999451956890528,
-      "loss": 46.0,
-      "step": 43597
-    },
-    {
-      "epoch": 3.3333715618250284,
-      "grad_norm": 0.004295188467949629,
-      "learning_rate": 0.0001999945193174338,
-      "loss": 46.0,
-      "step": 43598
-    },
-    {
-      "epoch": 3.3334480188084177,
-      "grad_norm": 0.0009946598438546062,
-      "learning_rate": 0.00019999451906595655,
-      "loss": 46.0,
-      "step": 43599
-    },
-    {
-      "epoch": 3.3335244757918074,
-      "grad_norm": 0.0009655312169343233,
-      "learning_rate": 0.00019999451881447353,
-      "loss": 46.0,
-      "step": 43600
-    },
-    {
-      "epoch": 3.333600932775197,
-      "grad_norm": 0.003998894244432449,
-      "learning_rate": 0.00019999451856298473,
-      "loss": 46.0,
-      "step": 43601
-    },
-    {
-      "epoch": 3.333677389758587,
-      "grad_norm": 0.0012280386872589588,
-      "learning_rate": 0.00019999451831149013,
-      "loss": 46.0,
-      "step": 43602
-    },
-    {
-      "epoch": 3.3337538467419767,
-      "grad_norm": 0.0018116086721420288,
-      "learning_rate": 0.00019999451805998981,
-      "loss": 46.0,
-      "step": 43603
-    },
-    {
-      "epoch": 3.3338303037253665,
-      "grad_norm": 0.002628431422635913,
-      "learning_rate": 0.0001999945178084837,
-      "loss": 46.0,
-      "step": 43604
-    },
-    {
-      "epoch": 3.3339067607087562,
-      "grad_norm": 0.0012622167123481631,
-      "learning_rate": 0.00019999451755697183,
-      "loss": 46.0,
-      "step": 43605
-    },
-    {
-      "epoch": 3.333983217692146,
-      "grad_norm": 0.005994723178446293,
-      "learning_rate": 0.0001999945173054542,
-      "loss": 46.0,
-      "step": 43606
-    },
-    {
-      "epoch": 3.3340596746755358,
-      "grad_norm": 0.002169710351154208,
-      "learning_rate": 0.0001999945170539308,
-      "loss": 46.0,
-      "step": 43607
-    },
-    {
-      "epoch": 3.3341361316589255,
-      "grad_norm": 0.0018565854988992214,
-      "learning_rate": 0.0001999945168024016,
-      "loss": 46.0,
-      "step": 43608
-    },
-    {
-      "epoch": 3.3342125886423153,
-      "grad_norm": 0.0028142984956502914,
-      "learning_rate": 0.00019999451655086667,
-      "loss": 46.0,
-      "step": 43609
-    },
-    {
-      "epoch": 3.334289045625705,
-      "grad_norm": 0.005660689901560545,
-      "learning_rate": 0.00019999451629932593,
-      "loss": 46.0,
-      "step": 43610
-    },
-    {
-      "epoch": 3.3343655026090944,
-      "grad_norm": 0.0025527826510369778,
-      "learning_rate": 0.00019999451604777946,
-      "loss": 46.0,
-      "step": 43611
-    },
-    {
-      "epoch": 3.334441959592484,
-      "grad_norm": 0.0024709736462682486,
-      "learning_rate": 0.0001999945157962272,
-      "loss": 46.0,
-      "step": 43612
-    },
-    {
-      "epoch": 3.334518416575874,
-      "grad_norm": 0.0009529362432658672,
-      "learning_rate": 0.00019999451554466918,
-      "loss": 46.0,
-      "step": 43613
-    },
-    {
-      "epoch": 3.3345948735592637,
-      "grad_norm": 0.0010676614474505186,
-      "learning_rate": 0.00019999451529310538,
-      "loss": 46.0,
-      "step": 43614
-    },
-    {
-      "epoch": 3.3346713305426534,
-      "grad_norm": 0.0011848678113892674,
-      "learning_rate": 0.0001999945150415358,
-      "loss": 46.0,
-      "step": 43615
-    },
-    {
-      "epoch": 3.334747787526043,
-      "grad_norm": 0.0018325322307646275,
-      "learning_rate": 0.00019999451478996047,
-      "loss": 46.0,
-      "step": 43616
-    },
-    {
-      "epoch": 3.334824244509433,
-      "grad_norm": 0.008592256344854832,
-      "learning_rate": 0.00019999451453837938,
-      "loss": 46.0,
-      "step": 43617
-    },
-    {
-      "epoch": 3.3349007014928227,
-      "grad_norm": 0.0025443288031965494,
-      "learning_rate": 0.0001999945142867925,
-      "loss": 46.0,
-      "step": 43618
-    },
-    {
-      "epoch": 3.3349771584762125,
-      "grad_norm": 0.004462145734578371,
-      "learning_rate": 0.00019999451403519985,
-      "loss": 46.0,
-      "step": 43619
-    },
-    {
-      "epoch": 3.335053615459602,
-      "grad_norm": 0.004196337424218655,
-      "learning_rate": 0.00019999451378360146,
-      "loss": 46.0,
-      "step": 43620
-    },
-    {
-      "epoch": 3.3351300724429915,
-      "grad_norm": 0.005828474648296833,
-      "learning_rate": 0.00019999451353199728,
-      "loss": 46.0,
-      "step": 43621
-    },
-    {
-      "epoch": 3.3352065294263813,
-      "grad_norm": 0.0027317239437252283,
-      "learning_rate": 0.00019999451328038732,
-      "loss": 46.0,
-      "step": 43622
-    },
-    {
-      "epoch": 3.335282986409771,
-      "grad_norm": 0.0033082624431699514,
-      "learning_rate": 0.00019999451302877162,
-      "loss": 46.0,
-      "step": 43623
-    },
-    {
-      "epoch": 3.335359443393161,
-      "grad_norm": 0.006514125503599644,
-      "learning_rate": 0.0001999945127771501,
-      "loss": 46.0,
-      "step": 43624
-    },
-    {
-      "epoch": 3.3354359003765506,
-      "grad_norm": 0.003531440393999219,
-      "learning_rate": 0.00019999451252552286,
-      "loss": 46.0,
-      "step": 43625
-    },
-    {
-      "epoch": 3.3355123573599403,
-      "grad_norm": 0.0009159431792795658,
-      "learning_rate": 0.00019999451227388984,
-      "loss": 46.0,
-      "step": 43626
-    },
-    {
-      "epoch": 3.33558881434333,
-      "grad_norm": 0.005102984607219696,
-      "learning_rate": 0.00019999451202225106,
-      "loss": 46.0,
-      "step": 43627
-    },
-    {
-      "epoch": 3.33566527132672,
-      "grad_norm": 0.0012662075459957123,
-      "learning_rate": 0.0001999945117706065,
-      "loss": 46.0,
-      "step": 43628
-    },
-    {
-      "epoch": 3.3357417283101096,
-      "grad_norm": 0.0011329468106850982,
-      "learning_rate": 0.00019999451151895615,
-      "loss": 46.0,
-      "step": 43629
-    },
-    {
-      "epoch": 3.3358181852934994,
-      "grad_norm": 0.0008179281721822917,
-      "learning_rate": 0.00019999451126730003,
-      "loss": 46.0,
-      "step": 43630
-    },
-    {
-      "epoch": 3.335894642276889,
-      "grad_norm": 0.005298284348100424,
-      "learning_rate": 0.00019999451101563817,
-      "loss": 46.0,
-      "step": 43631
-    },
-    {
-      "epoch": 3.335971099260279,
-      "grad_norm": 0.0025736666284501553,
-      "learning_rate": 0.00019999451076397053,
-      "loss": 46.0,
-      "step": 43632
-    },
-    {
-      "epoch": 3.336047556243668,
-      "grad_norm": 0.0015905586769804358,
-      "learning_rate": 0.00019999451051229712,
-      "loss": 46.0,
-      "step": 43633
-    },
-    {
-      "epoch": 3.336124013227058,
-      "grad_norm": 0.003255372168496251,
-      "learning_rate": 0.00019999451026061796,
-      "loss": 46.0,
-      "step": 43634
-    },
-    {
-      "epoch": 3.3362004702104477,
-      "grad_norm": 0.0016996952472254634,
-      "learning_rate": 0.000199994510008933,
-      "loss": 46.0,
-      "step": 43635
-    },
-    {
-      "epoch": 3.3362769271938375,
-      "grad_norm": 0.0011706857476383448,
-      "learning_rate": 0.00019999450975724227,
-      "loss": 46.0,
-      "step": 43636
-    },
-    {
-      "epoch": 3.3363533841772273,
-      "grad_norm": 0.00977516919374466,
-      "learning_rate": 0.0001999945095055458,
-      "loss": 46.0,
-      "step": 43637
-    },
-    {
-      "epoch": 3.336429841160617,
-      "grad_norm": 0.0015745279379189014,
-      "learning_rate": 0.00019999450925384355,
-      "loss": 46.0,
-      "step": 43638
-    },
-    {
-      "epoch": 3.336506298144007,
-      "grad_norm": 0.0017607235349714756,
-      "learning_rate": 0.00019999450900213552,
-      "loss": 46.0,
-      "step": 43639
-    },
-    {
-      "epoch": 3.3365827551273965,
-      "grad_norm": 0.0012273314641788602,
-      "learning_rate": 0.00019999450875042172,
-      "loss": 46.0,
-      "step": 43640
-    },
-    {
-      "epoch": 3.3366592121107863,
-      "grad_norm": 0.0018174175638705492,
-      "learning_rate": 0.00019999450849870215,
-      "loss": 46.0,
-      "step": 43641
-    },
-    {
-      "epoch": 3.336735669094176,
-      "grad_norm": 0.0036967771593481302,
-      "learning_rate": 0.00019999450824697684,
-      "loss": 46.0,
-      "step": 43642
-    },
-    {
-      "epoch": 3.3368121260775654,
-      "grad_norm": 0.0022941918577998877,
-      "learning_rate": 0.00019999450799524572,
-      "loss": 46.0,
-      "step": 43643
-    },
-    {
-      "epoch": 3.336888583060955,
-      "grad_norm": 0.0018503410974517465,
-      "learning_rate": 0.00019999450774350885,
-      "loss": 46.0,
-      "step": 43644
-    },
-    {
-      "epoch": 3.336965040044345,
-      "grad_norm": 0.0013353358954191208,
-      "learning_rate": 0.00019999450749176622,
-      "loss": 46.0,
-      "step": 43645
-    },
-    {
-      "epoch": 3.3370414970277347,
-      "grad_norm": 0.002239350927993655,
-      "learning_rate": 0.00019999450724001783,
-      "loss": 46.0,
-      "step": 43646
-    },
-    {
-      "epoch": 3.3371179540111244,
-      "grad_norm": 0.0013675467344000936,
-      "learning_rate": 0.00019999450698826365,
-      "loss": 46.0,
-      "step": 43647
-    },
-    {
-      "epoch": 3.337194410994514,
-      "grad_norm": 0.002925304463133216,
-      "learning_rate": 0.0001999945067365037,
-      "loss": 46.0,
-      "step": 43648
-    },
-    {
-      "epoch": 3.337270867977904,
-      "grad_norm": 0.0032144831493496895,
-      "learning_rate": 0.000199994506484738,
-      "loss": 46.0,
-      "step": 43649
-    },
-    {
-      "epoch": 3.3373473249612937,
-      "grad_norm": 0.0008190250373445451,
-      "learning_rate": 0.00019999450623296649,
-      "loss": 46.0,
-      "step": 43650
-    },
-    {
-      "epoch": 3.3374237819446835,
-      "grad_norm": 0.0014309815596789122,
-      "learning_rate": 0.00019999450598118924,
-      "loss": 46.0,
-      "step": 43651
-    },
-    {
-      "epoch": 3.3375002389280732,
-      "grad_norm": 0.0023848325945436954,
-      "learning_rate": 0.0001999945057294062,
-      "loss": 46.0,
-      "step": 43652
-    },
-    {
-      "epoch": 3.337576695911463,
-      "grad_norm": 0.002579217776656151,
-      "learning_rate": 0.00019999450547761741,
-      "loss": 46.0,
-      "step": 43653
-    },
-    {
-      "epoch": 3.3376531528948528,
-      "grad_norm": 0.0028548510745167732,
-      "learning_rate": 0.00019999450522582287,
-      "loss": 46.0,
-      "step": 43654
-    },
-    {
-      "epoch": 3.337729609878242,
-      "grad_norm": 0.0016805481864139438,
-      "learning_rate": 0.00019999450497402253,
-      "loss": 46.0,
-      "step": 43655
-    },
-    {
-      "epoch": 3.337806066861632,
-      "grad_norm": 0.0016382435569539666,
-      "learning_rate": 0.00019999450472221644,
-      "loss": 46.0,
-      "step": 43656
-    },
-    {
-      "epoch": 3.3378825238450216,
-      "grad_norm": 0.0015484046889469028,
-      "learning_rate": 0.00019999450447040457,
-      "loss": 46.0,
-      "step": 43657
-    },
-    {
-      "epoch": 3.3379589808284114,
-      "grad_norm": 0.011212067678570747,
-      "learning_rate": 0.00019999450421858694,
-      "loss": 46.0,
-      "step": 43658
-    },
-    {
-      "epoch": 3.338035437811801,
-      "grad_norm": 0.003929028753191233,
-      "learning_rate": 0.00019999450396676353,
-      "loss": 46.0,
-      "step": 43659
-    },
-    {
-      "epoch": 3.338111894795191,
-      "grad_norm": 0.0037038493901491165,
-      "learning_rate": 0.00019999450371493434,
-      "loss": 46.0,
-      "step": 43660
-    },
-    {
-      "epoch": 3.3381883517785806,
-      "grad_norm": 0.001650122576393187,
-      "learning_rate": 0.0001999945034630994,
-      "loss": 46.0,
-      "step": 43661
-    },
-    {
-      "epoch": 3.3382648087619704,
-      "grad_norm": 0.0011574962409213185,
-      "learning_rate": 0.00019999450321125868,
-      "loss": 46.0,
-      "step": 43662
-    },
-    {
-      "epoch": 3.33834126574536,
-      "grad_norm": 0.0019906165543943644,
-      "learning_rate": 0.0001999945029594122,
-      "loss": 46.0,
-      "step": 43663
-    },
-    {
-      "epoch": 3.33841772272875,
-      "grad_norm": 0.005197696853429079,
-      "learning_rate": 0.00019999450270755996,
-      "loss": 46.0,
-      "step": 43664
-    },
-    {
-      "epoch": 3.3384941797121392,
-      "grad_norm": 0.001822852180339396,
-      "learning_rate": 0.00019999450245570193,
-      "loss": 46.0,
-      "step": 43665
-    },
-    {
-      "epoch": 3.338570636695529,
-      "grad_norm": 0.0032041617669165134,
-      "learning_rate": 0.00019999450220383814,
-      "loss": 46.0,
-      "step": 43666
-    },
-    {
-      "epoch": 3.3386470936789188,
-      "grad_norm": 0.0026878872886300087,
-      "learning_rate": 0.0001999945019519686,
-      "loss": 46.0,
-      "step": 43667
-    },
-    {
-      "epoch": 3.3387235506623085,
-      "grad_norm": 0.00133521668612957,
-      "learning_rate": 0.00019999450170009325,
-      "loss": 46.0,
-      "step": 43668
-    },
-    {
-      "epoch": 3.3388000076456983,
-      "grad_norm": 0.0030250868294388056,
-      "learning_rate": 0.00019999450144821216,
-      "loss": 46.0,
-      "step": 43669
-    },
-    {
-      "epoch": 3.338876464629088,
-      "grad_norm": 0.003318930044770241,
-      "learning_rate": 0.0001999945011963253,
-      "loss": 46.0,
-      "step": 43670
-    },
-    {
-      "epoch": 3.338952921612478,
-      "grad_norm": 0.0009371746564283967,
-      "learning_rate": 0.00019999450094443266,
-      "loss": 46.0,
-      "step": 43671
-    },
-    {
-      "epoch": 3.3390293785958676,
-      "grad_norm": 0.005983148235827684,
-      "learning_rate": 0.00019999450069253425,
-      "loss": 46.0,
-      "step": 43672
-    },
-    {
-      "epoch": 3.3391058355792573,
-      "grad_norm": 0.001200530561618507,
-      "learning_rate": 0.00019999450044063007,
-      "loss": 46.0,
-      "step": 43673
-    },
-    {
-      "epoch": 3.339182292562647,
-      "grad_norm": 0.0005374782485887408,
-      "learning_rate": 0.00019999450018872014,
-      "loss": 46.0,
-      "step": 43674
-    },
-    {
-      "epoch": 3.339258749546037,
-      "grad_norm": 0.0029157826211303473,
-      "learning_rate": 0.0001999944999368044,
-      "loss": 46.0,
-      "step": 43675
-    },
-    {
-      "epoch": 3.3393352065294266,
-      "grad_norm": 0.0027916766703128815,
-      "learning_rate": 0.00019999449968488293,
-      "loss": 46.0,
-      "step": 43676
-    },
-    {
-      "epoch": 3.339411663512816,
-      "grad_norm": 0.002333183540031314,
-      "learning_rate": 0.00019999449943295568,
-      "loss": 46.0,
-      "step": 43677
-    },
-    {
-      "epoch": 3.3394881204962057,
-      "grad_norm": 0.003528687870129943,
-      "learning_rate": 0.00019999449918102266,
-      "loss": 46.0,
-      "step": 43678
-    },
-    {
-      "epoch": 3.3395645774795955,
-      "grad_norm": 0.0032058802898973227,
-      "learning_rate": 0.0001999944989290839,
-      "loss": 46.0,
-      "step": 43679
-    },
-    {
-      "epoch": 3.339641034462985,
-      "grad_norm": 0.003699591616168618,
-      "learning_rate": 0.00019999449867713932,
-      "loss": 46.0,
-      "step": 43680
-    },
-    {
-      "epoch": 3.339717491446375,
-      "grad_norm": 0.0022140464279800653,
-      "learning_rate": 0.00019999449842518898,
-      "loss": 46.0,
-      "step": 43681
-    },
-    {
-      "epoch": 3.3397939484297647,
-      "grad_norm": 0.002870541298761964,
-      "learning_rate": 0.00019999449817323291,
-      "loss": 46.0,
-      "step": 43682
-    },
-    {
-      "epoch": 3.3398704054131545,
-      "grad_norm": 0.0015898555284366012,
-      "learning_rate": 0.00019999449792127102,
-      "loss": 46.0,
-      "step": 43683
-    },
-    {
-      "epoch": 3.3399468623965443,
-      "grad_norm": 0.004617452621459961,
-      "learning_rate": 0.0001999944976693034,
-      "loss": 46.0,
-      "step": 43684
-    },
-    {
-      "epoch": 3.340023319379934,
-      "grad_norm": 0.004027530085295439,
-      "learning_rate": 0.00019999449741732998,
-      "loss": 46.0,
-      "step": 43685
-    },
-    {
-      "epoch": 3.3400997763633233,
-      "grad_norm": 0.0034801489673554897,
-      "learning_rate": 0.00019999449716535082,
-      "loss": 46.0,
-      "step": 43686
-    },
-    {
-      "epoch": 3.340176233346713,
-      "grad_norm": 0.0025715066585689783,
-      "learning_rate": 0.00019999449691336587,
-      "loss": 46.0,
-      "step": 43687
-    },
-    {
-      "epoch": 3.340252690330103,
-      "grad_norm": 0.0010014110011979938,
-      "learning_rate": 0.00019999449666137516,
-      "loss": 46.0,
-      "step": 43688
-    },
-    {
-      "epoch": 3.3403291473134926,
-      "grad_norm": 0.0014420212246477604,
-      "learning_rate": 0.0001999944964093787,
-      "loss": 46.0,
-      "step": 43689
-    },
-    {
-      "epoch": 3.3404056042968824,
-      "grad_norm": 0.0035400516353547573,
-      "learning_rate": 0.0001999944961573764,
-      "loss": 46.0,
-      "step": 43690
-    },
-    {
-      "epoch": 3.340482061280272,
-      "grad_norm": 0.0019122886005789042,
-      "learning_rate": 0.00019999449590536842,
-      "loss": 46.0,
-      "step": 43691
-    },
-    {
-      "epoch": 3.340558518263662,
-      "grad_norm": 0.0030854246579110622,
-      "learning_rate": 0.00019999449565335462,
-      "loss": 46.0,
-      "step": 43692
-    },
-    {
-      "epoch": 3.3406349752470517,
-      "grad_norm": 0.001542039797641337,
-      "learning_rate": 0.00019999449540133505,
-      "loss": 46.0,
-      "step": 43693
-    },
-    {
-      "epoch": 3.3407114322304414,
-      "grad_norm": 0.004390873480588198,
-      "learning_rate": 0.00019999449514930973,
-      "loss": 46.0,
-      "step": 43694
-    },
-    {
-      "epoch": 3.340787889213831,
-      "grad_norm": 0.0027141906321048737,
-      "learning_rate": 0.00019999449489727864,
-      "loss": 46.0,
-      "step": 43695
-    },
-    {
-      "epoch": 3.340864346197221,
-      "grad_norm": 0.002727772109210491,
-      "learning_rate": 0.00019999449464524176,
-      "loss": 46.0,
-      "step": 43696
-    },
-    {
-      "epoch": 3.3409408031806107,
-      "grad_norm": 0.0028421839233487844,
-      "learning_rate": 0.00019999449439319915,
-      "loss": 46.0,
-      "step": 43697
-    },
-    {
-      "epoch": 3.3410172601640005,
-      "grad_norm": 0.0009473218815401196,
-      "learning_rate": 0.0001999944941411507,
-      "loss": 46.0,
-      "step": 43698
-    },
-    {
-      "epoch": 3.34109371714739,
-      "grad_norm": 0.0021254231687635183,
-      "learning_rate": 0.00019999449388909658,
-      "loss": 46.0,
-      "step": 43699
-    },
-    {
-      "epoch": 3.3411701741307795,
-      "grad_norm": 0.0014690895332023501,
-      "learning_rate": 0.00019999449363703665,
-      "loss": 46.0,
-      "step": 43700
-    },
-    {
-      "epoch": 3.3412466311141693,
-      "grad_norm": 0.002514997962862253,
-      "learning_rate": 0.00019999449338497092,
-      "loss": 46.0,
-      "step": 43701
-    },
-    {
-      "epoch": 3.341323088097559,
-      "grad_norm": 0.0009950983803719282,
-      "learning_rate": 0.00019999449313289942,
-      "loss": 46.0,
-      "step": 43702
-    },
-    {
-      "epoch": 3.341399545080949,
-      "grad_norm": 0.00353820389136672,
-      "learning_rate": 0.00019999449288082217,
-      "loss": 46.0,
-      "step": 43703
-    },
-    {
-      "epoch": 3.3414760020643386,
-      "grad_norm": 0.0015001499559730291,
-      "learning_rate": 0.00019999449262873918,
-      "loss": 46.0,
-      "step": 43704
-    },
-    {
-      "epoch": 3.3415524590477284,
-      "grad_norm": 0.0014312267303466797,
-      "learning_rate": 0.00019999449237665038,
-      "loss": 46.0,
-      "step": 43705
-    },
-    {
-      "epoch": 3.341628916031118,
-      "grad_norm": 0.0028946269303560257,
-      "learning_rate": 0.00019999449212455584,
-      "loss": 46.0,
-      "step": 43706
-    },
-    {
-      "epoch": 3.341705373014508,
-      "grad_norm": 0.005926786921918392,
-      "learning_rate": 0.0001999944918724555,
-      "loss": 46.0,
-      "step": 43707
-    },
-    {
-      "epoch": 3.341781829997897,
-      "grad_norm": 0.0013261823914945126,
-      "learning_rate": 0.0001999944916203494,
-      "loss": 46.0,
-      "step": 43708
-    },
-    {
-      "epoch": 3.341858286981287,
-      "grad_norm": 0.004232048988342285,
-      "learning_rate": 0.00019999449136823754,
-      "loss": 46.0,
-      "step": 43709
-    },
-    {
-      "epoch": 3.3419347439646767,
-      "grad_norm": 0.0010355169652029872,
-      "learning_rate": 0.0001999944911161199,
-      "loss": 46.0,
-      "step": 43710
-    },
-    {
-      "epoch": 3.3420112009480665,
-      "grad_norm": 0.001177627476863563,
-      "learning_rate": 0.0001999944908639965,
-      "loss": 46.0,
-      "step": 43711
-    },
-    {
-      "epoch": 3.3420876579314562,
-      "grad_norm": 0.001688484800979495,
-      "learning_rate": 0.00019999449061186732,
-      "loss": 46.0,
-      "step": 43712
-    },
-    {
-      "epoch": 3.342164114914846,
-      "grad_norm": 0.0017280696192756295,
-      "learning_rate": 0.0001999944903597324,
-      "loss": 46.0,
-      "step": 43713
-    },
-    {
-      "epoch": 3.3422405718982358,
-      "grad_norm": 0.0023232828825712204,
-      "learning_rate": 0.0001999944901075917,
-      "loss": 46.0,
-      "step": 43714
-    },
-    {
-      "epoch": 3.3423170288816255,
-      "grad_norm": 0.0014603016898036003,
-      "learning_rate": 0.0001999944898554452,
-      "loss": 46.0,
-      "step": 43715
-    },
-    {
-      "epoch": 3.3423934858650153,
-      "grad_norm": 0.00415347795933485,
-      "learning_rate": 0.00019999448960329294,
-      "loss": 46.0,
-      "step": 43716
-    },
-    {
-      "epoch": 3.342469942848405,
-      "grad_norm": 0.0038795575965195894,
-      "learning_rate": 0.00019999448935113494,
-      "loss": 46.0,
-      "step": 43717
-    },
-    {
-      "epoch": 3.342546399831795,
-      "grad_norm": 0.0014306865632534027,
-      "learning_rate": 0.00019999448909897117,
-      "loss": 46.0,
-      "step": 43718
-    },
-    {
-      "epoch": 3.3426228568151846,
-      "grad_norm": 0.001141933025792241,
-      "learning_rate": 0.0001999944888468016,
-      "loss": 46.0,
-      "step": 43719
-    },
-    {
-      "epoch": 3.3426993137985743,
-      "grad_norm": 0.008010140620172024,
-      "learning_rate": 0.0001999944885946263,
-      "loss": 46.0,
-      "step": 43720
-    },
-    {
-      "epoch": 3.3427757707819636,
-      "grad_norm": 0.0014825982507318258,
-      "learning_rate": 0.00019999448834244517,
-      "loss": 46.0,
-      "step": 43721
-    },
-    {
-      "epoch": 3.3428522277653534,
-      "grad_norm": 0.00431219395250082,
-      "learning_rate": 0.0001999944880902583,
-      "loss": 46.0,
-      "step": 43722
-    },
-    {
-      "epoch": 3.342928684748743,
-      "grad_norm": 0.001852414570748806,
-      "learning_rate": 0.0001999944878380657,
-      "loss": 46.0,
-      "step": 43723
-    },
-    {
-      "epoch": 3.343005141732133,
-      "grad_norm": 0.001598511473275721,
-      "learning_rate": 0.0001999944875858673,
-      "loss": 46.0,
-      "step": 43724
-    },
-    {
-      "epoch": 3.3430815987155227,
-      "grad_norm": 0.0016969903372228146,
-      "learning_rate": 0.00019999448733366314,
-      "loss": 46.0,
-      "step": 43725
-    },
-    {
-      "epoch": 3.3431580556989124,
-      "grad_norm": 0.005367988720536232,
-      "learning_rate": 0.00019999448708145318,
-      "loss": 46.0,
-      "step": 43726
-    },
-    {
-      "epoch": 3.343234512682302,
-      "grad_norm": 0.001105388393625617,
-      "learning_rate": 0.00019999448682923748,
-      "loss": 46.0,
-      "step": 43727
-    },
-    {
-      "epoch": 3.343310969665692,
-      "grad_norm": 0.005350275430828333,
-      "learning_rate": 0.000199994486577016,
-      "loss": 46.0,
-      "step": 43728
-    },
-    {
-      "epoch": 3.3433874266490817,
-      "grad_norm": 0.0027503862511366606,
-      "learning_rate": 0.00019999448632478876,
-      "loss": 46.0,
-      "step": 43729
-    },
-    {
-      "epoch": 3.343463883632471,
-      "grad_norm": 0.0006434792303480208,
-      "learning_rate": 0.00019999448607255574,
-      "loss": 46.0,
-      "step": 43730
-    },
-    {
-      "epoch": 3.343540340615861,
-      "grad_norm": 0.0011709927348420024,
-      "learning_rate": 0.00019999448582031697,
-      "loss": 46.0,
-      "step": 43731
-    },
-    {
-      "epoch": 3.3436167975992506,
-      "grad_norm": 0.0009464659378863871,
-      "learning_rate": 0.0001999944855680724,
-      "loss": 46.0,
-      "step": 43732
-    },
-    {
-      "epoch": 3.3436932545826403,
-      "grad_norm": 0.002380137564614415,
-      "learning_rate": 0.00019999448531582209,
-      "loss": 46.0,
-      "step": 43733
-    },
-    {
-      "epoch": 3.34376971156603,
-      "grad_norm": 0.0010878652101382613,
-      "learning_rate": 0.000199994485063566,
-      "loss": 46.0,
-      "step": 43734
-    },
-    {
-      "epoch": 3.34384616854942,
-      "grad_norm": 0.0006586374947801232,
-      "learning_rate": 0.00019999448481130414,
-      "loss": 46.0,
-      "step": 43735
-    },
-    {
-      "epoch": 3.3439226255328096,
-      "grad_norm": 0.0023917292710393667,
-      "learning_rate": 0.0001999944845590365,
-      "loss": 46.0,
-      "step": 43736
-    },
-    {
-      "epoch": 3.3439990825161994,
-      "grad_norm": 0.0074102249927818775,
-      "learning_rate": 0.00019999448430676312,
-      "loss": 46.0,
-      "step": 43737
-    },
-    {
-      "epoch": 3.344075539499589,
-      "grad_norm": 0.0030318291392177343,
-      "learning_rate": 0.00019999448405448394,
-      "loss": 46.0,
-      "step": 43738
-    },
-    {
-      "epoch": 3.344151996482979,
-      "grad_norm": 0.002106248401105404,
-      "learning_rate": 0.00019999448380219898,
-      "loss": 46.0,
-      "step": 43739
-    },
-    {
-      "epoch": 3.3442284534663687,
-      "grad_norm": 0.0021125716157257557,
-      "learning_rate": 0.00019999448354990828,
-      "loss": 46.0,
-      "step": 43740
-    },
-    {
-      "epoch": 3.3443049104497584,
-      "grad_norm": 0.003041575662791729,
-      "learning_rate": 0.0001999944832976118,
-      "loss": 46.0,
-      "step": 43741
-    },
-    {
-      "epoch": 3.3443813674331477,
-      "grad_norm": 0.0019100791541859508,
-      "learning_rate": 0.00019999448304530956,
-      "loss": 46.0,
-      "step": 43742
-    },
-    {
-      "epoch": 3.3444578244165375,
-      "grad_norm": 0.0026639197021722794,
-      "learning_rate": 0.00019999448279300154,
-      "loss": 46.0,
-      "step": 43743
-    },
-    {
-      "epoch": 3.3445342813999273,
-      "grad_norm": 0.002550798002630472,
-      "learning_rate": 0.00019999448254068777,
-      "loss": 46.0,
-      "step": 43744
-    },
-    {
-      "epoch": 3.344610738383317,
-      "grad_norm": 0.0016459315083920956,
-      "learning_rate": 0.0001999944822883682,
-      "loss": 46.0,
-      "step": 43745
-    },
-    {
-      "epoch": 3.3446871953667068,
-      "grad_norm": 0.0023307749070227146,
-      "learning_rate": 0.0001999944820360429,
-      "loss": 46.0,
-      "step": 43746
-    },
-    {
-      "epoch": 3.3447636523500965,
-      "grad_norm": 0.004019988235086203,
-      "learning_rate": 0.00019999448178371177,
-      "loss": 46.0,
-      "step": 43747
-    },
-    {
-      "epoch": 3.3448401093334863,
-      "grad_norm": 0.0013858407037332654,
-      "learning_rate": 0.00019999448153137494,
-      "loss": 46.0,
-      "step": 43748
-    },
-    {
-      "epoch": 3.344916566316876,
-      "grad_norm": 0.0029589759651571512,
-      "learning_rate": 0.0001999944812790323,
-      "loss": 46.0,
-      "step": 43749
-    },
-    {
-      "epoch": 3.344993023300266,
-      "grad_norm": 0.0014120849082246423,
-      "learning_rate": 0.00019999448102668392,
-      "loss": 46.0,
-      "step": 43750
-    },
-    {
-      "epoch": 3.3450694802836556,
-      "grad_norm": 0.0009957747533917427,
-      "learning_rate": 0.00019999448077432974,
-      "loss": 46.0,
-      "step": 43751
-    },
-    {
-      "epoch": 3.345145937267045,
-      "grad_norm": 0.0012325698044151068,
-      "learning_rate": 0.0001999944805219698,
-      "loss": 46.0,
-      "step": 43752
-    },
-    {
-      "epoch": 3.3452223942504347,
-      "grad_norm": 0.006688510999083519,
-      "learning_rate": 0.0001999944802696041,
-      "loss": 46.0,
-      "step": 43753
-    },
-    {
-      "epoch": 3.3452988512338244,
-      "grad_norm": 0.002767265308648348,
-      "learning_rate": 0.00019999448001723264,
-      "loss": 46.0,
-      "step": 43754
-    },
-    {
-      "epoch": 3.345375308217214,
-      "grad_norm": 0.007987682707607746,
-      "learning_rate": 0.0001999944797648554,
-      "loss": 46.0,
-      "step": 43755
-    },
-    {
-      "epoch": 3.345451765200604,
-      "grad_norm": 0.0032269826624542475,
-      "learning_rate": 0.00019999447951247237,
-      "loss": 46.0,
-      "step": 43756
-    },
-    {
-      "epoch": 3.3455282221839937,
-      "grad_norm": 0.008732321672141552,
-      "learning_rate": 0.0001999944792600836,
-      "loss": 46.0,
-      "step": 43757
-    },
-    {
-      "epoch": 3.3456046791673835,
-      "grad_norm": 0.004555963911116123,
-      "learning_rate": 0.00019999447900768903,
-      "loss": 46.0,
-      "step": 43758
-    },
-    {
-      "epoch": 3.3456811361507732,
-      "grad_norm": 0.0035217690747231245,
-      "learning_rate": 0.00019999447875528872,
-      "loss": 46.0,
-      "step": 43759
-    },
-    {
-      "epoch": 3.345757593134163,
-      "grad_norm": 0.003565158462151885,
-      "learning_rate": 0.00019999447850288263,
-      "loss": 46.0,
-      "step": 43760
-    },
-    {
-      "epoch": 3.3458340501175527,
-      "grad_norm": 0.00240330770611763,
-      "learning_rate": 0.00019999447825047077,
-      "loss": 46.0,
-      "step": 43761
-    },
-    {
-      "epoch": 3.3459105071009425,
-      "grad_norm": 0.0016876407898962498,
-      "learning_rate": 0.00019999447799805316,
-      "loss": 46.0,
-      "step": 43762
-    },
-    {
-      "epoch": 3.3459869640843323,
-      "grad_norm": 0.0013996004126966,
-      "learning_rate": 0.00019999447774562973,
-      "loss": 46.0,
-      "step": 43763
-    },
-    {
-      "epoch": 3.3460634210677216,
-      "grad_norm": 0.0009210139396600425,
-      "learning_rate": 0.0001999944774932006,
-      "loss": 46.0,
-      "step": 43764
-    },
-    {
-      "epoch": 3.3461398780511113,
-      "grad_norm": 0.0018856822280213237,
-      "learning_rate": 0.00019999447724076562,
-      "loss": 46.0,
-      "step": 43765
-    },
-    {
-      "epoch": 3.346216335034501,
-      "grad_norm": 0.001019989955238998,
-      "learning_rate": 0.00019999447698832495,
-      "loss": 46.0,
-      "step": 43766
-    },
-    {
-      "epoch": 3.346292792017891,
-      "grad_norm": 0.000980242621153593,
-      "learning_rate": 0.00019999447673587847,
-      "loss": 46.0,
-      "step": 43767
-    },
-    {
-      "epoch": 3.3463692490012806,
-      "grad_norm": 0.0018957104766741395,
-      "learning_rate": 0.00019999447648342625,
-      "loss": 46.0,
-      "step": 43768
-    },
-    {
-      "epoch": 3.3464457059846704,
-      "grad_norm": 0.0015449401689693332,
-      "learning_rate": 0.00019999447623096823,
-      "loss": 46.0,
-      "step": 43769
-    },
-    {
-      "epoch": 3.34652216296806,
-      "grad_norm": 0.0014780907658860087,
-      "learning_rate": 0.00019999447597850444,
-      "loss": 46.0,
-      "step": 43770
-    },
-    {
-      "epoch": 3.34659861995145,
-      "grad_norm": 0.001844253041781485,
-      "learning_rate": 0.0001999944757260349,
-      "loss": 46.0,
-      "step": 43771
-    },
-    {
-      "epoch": 3.3466750769348397,
-      "grad_norm": 0.0013287770561873913,
-      "learning_rate": 0.00019999447547355956,
-      "loss": 46.0,
-      "step": 43772
-    },
-    {
-      "epoch": 3.3467515339182294,
-      "grad_norm": 0.003127113450318575,
-      "learning_rate": 0.00019999447522107847,
-      "loss": 46.0,
-      "step": 43773
-    },
-    {
-      "epoch": 3.3468279909016188,
-      "grad_norm": 0.0013063066871836782,
-      "learning_rate": 0.00019999447496859163,
-      "loss": 46.0,
-      "step": 43774
-    },
-    {
-      "epoch": 3.3469044478850085,
-      "grad_norm": 0.0017159474082291126,
-      "learning_rate": 0.00019999447471609897,
-      "loss": 46.0,
-      "step": 43775
-    },
-    {
-      "epoch": 3.3469809048683983,
-      "grad_norm": 0.0024811800103634596,
-      "learning_rate": 0.0001999944744636006,
-      "loss": 46.0,
-      "step": 43776
-    },
-    {
-      "epoch": 3.347057361851788,
-      "grad_norm": 0.007020830176770687,
-      "learning_rate": 0.00019999447421109644,
-      "loss": 46.0,
-      "step": 43777
-    },
-    {
-      "epoch": 3.347133818835178,
-      "grad_norm": 0.0017891379538923502,
-      "learning_rate": 0.0001999944739585865,
-      "loss": 46.0,
-      "step": 43778
-    },
-    {
-      "epoch": 3.3472102758185676,
-      "grad_norm": 0.0018376614898443222,
-      "learning_rate": 0.00019999447370607079,
-      "loss": 46.0,
-      "step": 43779
-    },
-    {
-      "epoch": 3.3472867328019573,
-      "grad_norm": 0.00107807747554034,
-      "learning_rate": 0.0001999944734535493,
-      "loss": 46.0,
-      "step": 43780
-    },
-    {
-      "epoch": 3.347363189785347,
-      "grad_norm": 0.0027003982104361057,
-      "learning_rate": 0.00019999447320102206,
-      "loss": 46.0,
-      "step": 43781
-    },
-    {
-      "epoch": 3.347439646768737,
-      "grad_norm": 0.00218713260255754,
-      "learning_rate": 0.00019999447294848907,
-      "loss": 46.0,
-      "step": 43782
-    },
-    {
-      "epoch": 3.3475161037521266,
-      "grad_norm": 0.0017130491323769093,
-      "learning_rate": 0.0001999944726959503,
-      "loss": 46.0,
-      "step": 43783
-    },
-    {
-      "epoch": 3.3475925607355164,
-      "grad_norm": 0.0017729747341945767,
-      "learning_rate": 0.00019999447244340574,
-      "loss": 46.0,
-      "step": 43784
-    },
-    {
-      "epoch": 3.347669017718906,
-      "grad_norm": 0.0024973598774522543,
-      "learning_rate": 0.00019999447219085542,
-      "loss": 46.0,
-      "step": 43785
-    },
-    {
-      "epoch": 3.3477454747022954,
-      "grad_norm": 0.0009839512640610337,
-      "learning_rate": 0.00019999447193829934,
-      "loss": 46.0,
-      "step": 43786
-    },
-    {
-      "epoch": 3.347821931685685,
-      "grad_norm": 0.0020365859381854534,
-      "learning_rate": 0.00019999447168573748,
-      "loss": 46.0,
-      "step": 43787
-    },
-    {
-      "epoch": 3.347898388669075,
-      "grad_norm": 0.005399767775088549,
-      "learning_rate": 0.00019999447143316984,
-      "loss": 46.0,
-      "step": 43788
-    },
-    {
-      "epoch": 3.3479748456524647,
-      "grad_norm": 0.004327656235545874,
-      "learning_rate": 0.00019999447118059646,
-      "loss": 46.0,
-      "step": 43789
-    },
-    {
-      "epoch": 3.3480513026358545,
-      "grad_norm": 0.0029053015168756247,
-      "learning_rate": 0.0001999944709280173,
-      "loss": 46.0,
-      "step": 43790
-    },
-    {
-      "epoch": 3.3481277596192442,
-      "grad_norm": 0.002035491168498993,
-      "learning_rate": 0.00019999447067543238,
-      "loss": 46.0,
-      "step": 43791
-    },
-    {
-      "epoch": 3.348204216602634,
-      "grad_norm": 0.007189083378762007,
-      "learning_rate": 0.00019999447042284168,
-      "loss": 46.0,
-      "step": 43792
-    },
-    {
-      "epoch": 3.3482806735860238,
-      "grad_norm": 0.001513556926511228,
-      "learning_rate": 0.0001999944701702452,
-      "loss": 46.0,
-      "step": 43793
-    },
-    {
-      "epoch": 3.3483571305694135,
-      "grad_norm": 0.0010239416733384132,
-      "learning_rate": 0.00019999446991764293,
-      "loss": 46.0,
-      "step": 43794
-    },
-    {
-      "epoch": 3.3484335875528033,
-      "grad_norm": 0.0008572787628509104,
-      "learning_rate": 0.00019999446966503494,
-      "loss": 46.0,
-      "step": 43795
-    },
-    {
-      "epoch": 3.3485100445361926,
-      "grad_norm": 0.00042112168739549816,
-      "learning_rate": 0.00019999446941242117,
-      "loss": 46.0,
-      "step": 43796
-    },
-    {
-      "epoch": 3.3485865015195824,
-      "grad_norm": 0.0010782723547890782,
-      "learning_rate": 0.00019999446915980164,
-      "loss": 46.0,
-      "step": 43797
-    },
-    {
-      "epoch": 3.348662958502972,
-      "grad_norm": 0.0007021041237749159,
-      "learning_rate": 0.0001999944689071763,
-      "loss": 46.0,
-      "step": 43798
-    },
-    {
-      "epoch": 3.348739415486362,
-      "grad_norm": 0.002902782754972577,
-      "learning_rate": 0.00019999446865454524,
-      "loss": 46.0,
-      "step": 43799
-    },
-    {
-      "epoch": 3.3488158724697517,
-      "grad_norm": 0.0017065502470359206,
-      "learning_rate": 0.00019999446840190838,
-      "loss": 46.0,
-      "step": 43800
-    },
-    {
-      "epoch": 3.3488923294531414,
-      "grad_norm": 0.001943213865160942,
-      "learning_rate": 0.00019999446814926571,
-      "loss": 46.0,
-      "step": 43801
-    },
-    {
-      "epoch": 3.348968786436531,
-      "grad_norm": 0.001652433187700808,
-      "learning_rate": 0.00019999446789661734,
-      "loss": 46.0,
-      "step": 43802
-    },
-    {
-      "epoch": 3.349045243419921,
-      "grad_norm": 0.00564029673114419,
-      "learning_rate": 0.00019999446764396315,
-      "loss": 46.0,
-      "step": 43803
-    },
-    {
-      "epoch": 3.3491217004033107,
-      "grad_norm": 0.0015387465246021748,
-      "learning_rate": 0.00019999446739130326,
-      "loss": 46.0,
-      "step": 43804
-    },
-    {
-      "epoch": 3.3491981573867005,
-      "grad_norm": 0.004566119518131018,
-      "learning_rate": 0.00019999446713863756,
-      "loss": 46.0,
-      "step": 43805
-    },
-    {
-      "epoch": 3.34927461437009,
-      "grad_norm": 0.0015043530147522688,
-      "learning_rate": 0.00019999446688596608,
-      "loss": 46.0,
-      "step": 43806
-    },
-    {
-      "epoch": 3.34935107135348,
-      "grad_norm": 0.002944496227428317,
-      "learning_rate": 0.00019999446663328884,
-      "loss": 46.0,
-      "step": 43807
-    },
-    {
-      "epoch": 3.3494275283368693,
-      "grad_norm": 0.0024127515498548746,
-      "learning_rate": 0.00019999446638060584,
-      "loss": 46.0,
-      "step": 43808
-    },
-    {
-      "epoch": 3.349503985320259,
-      "grad_norm": 0.0029832408763468266,
-      "learning_rate": 0.00019999446612791705,
-      "loss": 46.0,
-      "step": 43809
-    },
-    {
-      "epoch": 3.349580442303649,
-      "grad_norm": 0.009830344468355179,
-      "learning_rate": 0.0001999944658752225,
-      "loss": 46.0,
-      "step": 43810
-    },
-    {
-      "epoch": 3.3496568992870386,
-      "grad_norm": 0.001837630057707429,
-      "learning_rate": 0.0001999944656225222,
-      "loss": 46.0,
-      "step": 43811
-    },
-    {
-      "epoch": 3.3497333562704283,
-      "grad_norm": 0.0010770781664177775,
-      "learning_rate": 0.0001999944653698161,
-      "loss": 46.0,
-      "step": 43812
-    },
-    {
-      "epoch": 3.349809813253818,
-      "grad_norm": 0.0032199062407016754,
-      "learning_rate": 0.00019999446511710428,
-      "loss": 46.0,
-      "step": 43813
-    },
-    {
-      "epoch": 3.349886270237208,
-      "grad_norm": 0.0016934559680521488,
-      "learning_rate": 0.00019999446486438665,
-      "loss": 46.0,
-      "step": 43814
-    },
-    {
-      "epoch": 3.3499627272205976,
-      "grad_norm": 0.0020858892239630222,
-      "learning_rate": 0.00019999446461166324,
-      "loss": 46.0,
-      "step": 43815
-    },
-    {
-      "epoch": 3.3500391842039874,
-      "grad_norm": 0.000608460046350956,
-      "learning_rate": 0.0001999944643589341,
-      "loss": 46.0,
-      "step": 43816
-    },
-    {
-      "epoch": 3.3501156411873767,
-      "grad_norm": 0.002830859972164035,
-      "learning_rate": 0.00019999446410619916,
-      "loss": 46.0,
-      "step": 43817
-    },
-    {
-      "epoch": 3.3501920981707665,
-      "grad_norm": 0.0023028110153973103,
-      "learning_rate": 0.00019999446385345846,
-      "loss": 46.0,
-      "step": 43818
-    },
-    {
-      "epoch": 3.3502685551541562,
-      "grad_norm": 0.0027507252525538206,
-      "learning_rate": 0.000199994463600712,
-      "loss": 46.0,
-      "step": 43819
-    },
-    {
-      "epoch": 3.350345012137546,
-      "grad_norm": 0.003907621838152409,
-      "learning_rate": 0.00019999446334795974,
-      "loss": 46.0,
-      "step": 43820
-    },
-    {
-      "epoch": 3.3504214691209357,
-      "grad_norm": 0.0026545915752649307,
-      "learning_rate": 0.00019999446309520175,
-      "loss": 46.0,
-      "step": 43821
-    },
-    {
-      "epoch": 3.3504979261043255,
-      "grad_norm": 0.0025462189223617315,
-      "learning_rate": 0.00019999446284243796,
-      "loss": 46.0,
-      "step": 43822
-    },
-    {
-      "epoch": 3.3505743830877153,
-      "grad_norm": 0.0017587310867384076,
-      "learning_rate": 0.00019999446258966842,
-      "loss": 46.0,
-      "step": 43823
-    },
-    {
-      "epoch": 3.350650840071105,
-      "grad_norm": 0.005058556795120239,
-      "learning_rate": 0.0001999944623368931,
-      "loss": 46.0,
-      "step": 43824
-    },
-    {
-      "epoch": 3.350727297054495,
-      "grad_norm": 0.0026162636931985617,
-      "learning_rate": 0.00019999446208411202,
-      "loss": 46.0,
-      "step": 43825
-    },
-    {
-      "epoch": 3.3508037540378846,
-      "grad_norm": 0.0019965535029768944,
-      "learning_rate": 0.00019999446183132519,
-      "loss": 46.0,
-      "step": 43826
-    },
-    {
-      "epoch": 3.3508802110212743,
-      "grad_norm": 0.0020608650520443916,
-      "learning_rate": 0.00019999446157853255,
-      "loss": 46.0,
-      "step": 43827
-    },
-    {
-      "epoch": 3.350956668004664,
-      "grad_norm": 0.0015180882764980197,
-      "learning_rate": 0.00019999446132573417,
-      "loss": 46.0,
-      "step": 43828
-    },
-    {
-      "epoch": 3.351033124988054,
-      "grad_norm": 0.0018931550439447165,
-      "learning_rate": 0.00019999446107293,
-      "loss": 46.0,
-      "step": 43829
-    },
-    {
-      "epoch": 3.351109581971443,
-      "grad_norm": 0.0008813115418888628,
-      "learning_rate": 0.00019999446082012007,
-      "loss": 46.0,
-      "step": 43830
-    },
-    {
-      "epoch": 3.351186038954833,
-      "grad_norm": 0.003362125949934125,
-      "learning_rate": 0.00019999446056730437,
-      "loss": 46.0,
-      "step": 43831
-    },
-    {
-      "epoch": 3.3512624959382227,
-      "grad_norm": 0.0024379382375627756,
-      "learning_rate": 0.0001999944603144829,
-      "loss": 46.0,
-      "step": 43832
-    },
-    {
-      "epoch": 3.3513389529216124,
-      "grad_norm": 0.0006784663419239223,
-      "learning_rate": 0.00019999446006165568,
-      "loss": 46.0,
-      "step": 43833
-    },
-    {
-      "epoch": 3.351415409905002,
-      "grad_norm": 0.0006937530706636608,
-      "learning_rate": 0.0001999944598088227,
-      "loss": 46.0,
-      "step": 43834
-    },
-    {
-      "epoch": 3.351491866888392,
-      "grad_norm": 0.0024105324409902096,
-      "learning_rate": 0.0001999944595559839,
-      "loss": 46.0,
-      "step": 43835
-    },
-    {
-      "epoch": 3.3515683238717817,
-      "grad_norm": 0.001397754531353712,
-      "learning_rate": 0.00019999445930313936,
-      "loss": 46.0,
-      "step": 43836
-    },
-    {
-      "epoch": 3.3516447808551715,
-      "grad_norm": 0.0032718039583414793,
-      "learning_rate": 0.00019999445905028904,
-      "loss": 46.0,
-      "step": 43837
-    },
-    {
-      "epoch": 3.3517212378385612,
-      "grad_norm": 0.0013750699581578374,
-      "learning_rate": 0.00019999445879743296,
-      "loss": 46.0,
-      "step": 43838
-    },
-    {
-      "epoch": 3.3517976948219506,
-      "grad_norm": 0.0029985366854816675,
-      "learning_rate": 0.0001999944585445711,
-      "loss": 46.0,
-      "step": 43839
-    },
-    {
-      "epoch": 3.3518741518053403,
-      "grad_norm": 0.00362880015745759,
-      "learning_rate": 0.0001999944582917035,
-      "loss": 46.0,
-      "step": 43840
-    },
-    {
-      "epoch": 3.35195060878873,
-      "grad_norm": 0.001020593335852027,
-      "learning_rate": 0.0001999944580388301,
-      "loss": 46.0,
-      "step": 43841
-    },
-    {
-      "epoch": 3.35202706577212,
-      "grad_norm": 0.0006231357692740858,
-      "learning_rate": 0.00019999445778595093,
-      "loss": 46.0,
-      "step": 43842
-    },
-    {
-      "epoch": 3.3521035227555096,
-      "grad_norm": 0.003153297584503889,
-      "learning_rate": 0.000199994457533066,
-      "loss": 46.0,
-      "step": 43843
-    },
-    {
-      "epoch": 3.3521799797388994,
-      "grad_norm": 0.001853014575317502,
-      "learning_rate": 0.00019999445728017528,
-      "loss": 46.0,
-      "step": 43844
-    },
-    {
-      "epoch": 3.352256436722289,
-      "grad_norm": 0.0012519273441284895,
-      "learning_rate": 0.00019999445702727884,
-      "loss": 46.0,
-      "step": 43845
-    },
-    {
-      "epoch": 3.352332893705679,
-      "grad_norm": 0.003738974453881383,
-      "learning_rate": 0.0001999944567743766,
-      "loss": 46.0,
-      "step": 43846
-    },
-    {
-      "epoch": 3.3524093506890686,
-      "grad_norm": 0.0023745065554976463,
-      "learning_rate": 0.0001999944565214686,
-      "loss": 46.0,
-      "step": 43847
-    },
-    {
-      "epoch": 3.3524858076724584,
-      "grad_norm": 0.001642899471335113,
-      "learning_rate": 0.0001999944562685548,
-      "loss": 46.0,
-      "step": 43848
-    },
-    {
-      "epoch": 3.352562264655848,
-      "grad_norm": 0.001237194868735969,
-      "learning_rate": 0.0001999944560156353,
-      "loss": 46.0,
-      "step": 43849
-    },
-    {
-      "epoch": 3.352638721639238,
-      "grad_norm": 0.006038373801857233,
-      "learning_rate": 0.00019999445576270999,
-      "loss": 46.0,
-      "step": 43850
-    },
-    {
-      "epoch": 3.3527151786226277,
-      "grad_norm": 0.002091360278427601,
-      "learning_rate": 0.0001999944555097789,
-      "loss": 46.0,
-      "step": 43851
-    },
-    {
-      "epoch": 3.352791635606017,
-      "grad_norm": 0.0009761899709701538,
-      "learning_rate": 0.00019999445525684207,
-      "loss": 46.0,
-      "step": 43852
-    },
-    {
-      "epoch": 3.3528680925894068,
-      "grad_norm": 0.0007871040725149214,
-      "learning_rate": 0.0001999944550038994,
-      "loss": 46.0,
-      "step": 43853
-    },
-    {
-      "epoch": 3.3529445495727965,
-      "grad_norm": 0.002603947650641203,
-      "learning_rate": 0.00019999445475095103,
-      "loss": 46.0,
-      "step": 43854
-    },
-    {
-      "epoch": 3.3530210065561863,
-      "grad_norm": 0.0025980807840824127,
-      "learning_rate": 0.0001999944544979969,
-      "loss": 46.0,
-      "step": 43855
-    },
-    {
-      "epoch": 3.353097463539576,
-      "grad_norm": 0.001019587623886764,
-      "learning_rate": 0.00019999445424503695,
-      "loss": 46.0,
-      "step": 43856
-    },
-    {
-      "epoch": 3.353173920522966,
-      "grad_norm": 0.0018611080013215542,
-      "learning_rate": 0.00019999445399207125,
-      "loss": 46.0,
-      "step": 43857
-    },
-    {
-      "epoch": 3.3532503775063556,
-      "grad_norm": 0.0006320485263131559,
-      "learning_rate": 0.00019999445373909978,
-      "loss": 46.0,
-      "step": 43858
-    },
-    {
-      "epoch": 3.3533268344897453,
-      "grad_norm": 0.002382304985076189,
-      "learning_rate": 0.00019999445348612256,
-      "loss": 46.0,
-      "step": 43859
-    },
-    {
-      "epoch": 3.353403291473135,
-      "grad_norm": 0.0012430732604116201,
-      "learning_rate": 0.00019999445323313957,
-      "loss": 46.0,
-      "step": 43860
-    },
-    {
-      "epoch": 3.3534797484565244,
-      "grad_norm": 0.0009905261686071754,
-      "learning_rate": 0.00019999445298015078,
-      "loss": 46.0,
-      "step": 43861
-    },
-    {
-      "epoch": 3.353556205439914,
-      "grad_norm": 0.005501210689544678,
-      "learning_rate": 0.00019999445272715624,
-      "loss": 46.0,
-      "step": 43862
-    },
-    {
-      "epoch": 3.353632662423304,
-      "grad_norm": 0.005886425264179707,
-      "learning_rate": 0.00019999445247415593,
-      "loss": 46.0,
-      "step": 43863
-    },
-    {
-      "epoch": 3.3537091194066937,
-      "grad_norm": 0.0032316395081579685,
-      "learning_rate": 0.00019999445222114985,
-      "loss": 46.0,
-      "step": 43864
-    },
-    {
-      "epoch": 3.3537855763900835,
-      "grad_norm": 0.0015141512267291546,
-      "learning_rate": 0.000199994451968138,
-      "loss": 46.0,
-      "step": 43865
-    },
-    {
-      "epoch": 3.353862033373473,
-      "grad_norm": 0.0043038311414420605,
-      "learning_rate": 0.00019999445171512038,
-      "loss": 46.0,
-      "step": 43866
-    },
-    {
-      "epoch": 3.353938490356863,
-      "grad_norm": 0.004591199103742838,
-      "learning_rate": 0.000199994451462097,
-      "loss": 46.0,
-      "step": 43867
-    },
-    {
-      "epoch": 3.3540149473402527,
-      "grad_norm": 0.003466640831902623,
-      "learning_rate": 0.00019999445120906785,
-      "loss": 46.0,
-      "step": 43868
-    },
-    {
-      "epoch": 3.3540914043236425,
-      "grad_norm": 0.0028352956287562847,
-      "learning_rate": 0.00019999445095603293,
-      "loss": 46.0,
-      "step": 43869
-    },
-    {
-      "epoch": 3.3541678613070323,
-      "grad_norm": 0.0020261681638658047,
-      "learning_rate": 0.00019999445070299223,
-      "loss": 46.0,
-      "step": 43870
-    },
-    {
-      "epoch": 3.354244318290422,
-      "grad_norm": 0.0011520616244524717,
-      "learning_rate": 0.00019999445044994576,
-      "loss": 46.0,
-      "step": 43871
-    },
-    {
-      "epoch": 3.354320775273812,
-      "grad_norm": 0.0013362925965338945,
-      "learning_rate": 0.00019999445019689354,
-      "loss": 46.0,
-      "step": 43872
-    },
-    {
-      "epoch": 3.354397232257201,
-      "grad_norm": 0.001257666852325201,
-      "learning_rate": 0.00019999444994383555,
-      "loss": 46.0,
-      "step": 43873
-    },
-    {
-      "epoch": 3.354473689240591,
-      "grad_norm": 0.0014234689297154546,
-      "learning_rate": 0.00019999444969077176,
-      "loss": 46.0,
-      "step": 43874
-    },
-    {
-      "epoch": 3.3545501462239806,
-      "grad_norm": 0.0013488612603396177,
-      "learning_rate": 0.00019999444943770222,
-      "loss": 46.0,
-      "step": 43875
-    },
-    {
-      "epoch": 3.3546266032073704,
-      "grad_norm": 0.0041291494853794575,
-      "learning_rate": 0.0001999944491846269,
-      "loss": 46.0,
-      "step": 43876
-    },
-    {
-      "epoch": 3.35470306019076,
-      "grad_norm": 0.002632700838148594,
-      "learning_rate": 0.00019999444893154582,
-      "loss": 46.0,
-      "step": 43877
-    },
-    {
-      "epoch": 3.35477951717415,
-      "grad_norm": 0.0010814070701599121,
-      "learning_rate": 0.00019999444867845897,
-      "loss": 46.0,
-      "step": 43878
-    },
-    {
-      "epoch": 3.3548559741575397,
-      "grad_norm": 0.0016110329888761044,
-      "learning_rate": 0.00019999444842536636,
-      "loss": 46.0,
-      "step": 43879
-    },
-    {
-      "epoch": 3.3549324311409294,
-      "grad_norm": 0.0015688743442296982,
-      "learning_rate": 0.00019999444817226798,
-      "loss": 46.0,
-      "step": 43880
-    },
-    {
-      "epoch": 3.355008888124319,
-      "grad_norm": 0.001690788776613772,
-      "learning_rate": 0.0001999944479191638,
-      "loss": 46.0,
-      "step": 43881
-    },
-    {
-      "epoch": 3.355085345107709,
-      "grad_norm": 0.0015371820190921426,
-      "learning_rate": 0.0001999944476660539,
-      "loss": 46.0,
-      "step": 43882
-    },
-    {
-      "epoch": 3.3551618020910983,
-      "grad_norm": 0.009616333059966564,
-      "learning_rate": 0.0001999944474129382,
-      "loss": 46.0,
-      "step": 43883
-    },
-    {
-      "epoch": 3.355238259074488,
-      "grad_norm": 0.0014407926937565207,
-      "learning_rate": 0.00019999444715981674,
-      "loss": 46.0,
-      "step": 43884
-    },
-    {
-      "epoch": 3.355314716057878,
-      "grad_norm": 0.0026107977610081434,
-      "learning_rate": 0.00019999444690668952,
-      "loss": 46.0,
-      "step": 43885
-    },
-    {
-      "epoch": 3.3553911730412675,
-      "grad_norm": 0.0037337651010602713,
-      "learning_rate": 0.0001999944466535565,
-      "loss": 46.0,
-      "step": 43886
-    },
-    {
-      "epoch": 3.3554676300246573,
-      "grad_norm": 0.001361203845590353,
-      "learning_rate": 0.00019999444640041777,
-      "loss": 46.0,
-      "step": 43887
-    },
-    {
-      "epoch": 3.355544087008047,
-      "grad_norm": 0.0028255737852305174,
-      "learning_rate": 0.0001999944461472732,
-      "loss": 46.0,
-      "step": 43888
-    },
-    {
-      "epoch": 3.355620543991437,
-      "grad_norm": 0.0028665463905781507,
-      "learning_rate": 0.00019999444589412292,
-      "loss": 46.0,
-      "step": 43889
-    },
-    {
-      "epoch": 3.3556970009748266,
-      "grad_norm": 0.001013896893709898,
-      "learning_rate": 0.0001999944456409668,
-      "loss": 46.0,
-      "step": 43890
-    },
-    {
-      "epoch": 3.3557734579582164,
-      "grad_norm": 0.005709273274987936,
-      "learning_rate": 0.00019999444538780498,
-      "loss": 46.0,
-      "step": 43891
-    },
-    {
-      "epoch": 3.355849914941606,
-      "grad_norm": 0.0026123628485947847,
-      "learning_rate": 0.00019999444513463737,
-      "loss": 46.0,
-      "step": 43892
-    },
-    {
-      "epoch": 3.355926371924996,
-      "grad_norm": 0.0017032069154083729,
-      "learning_rate": 0.000199994444881464,
-      "loss": 46.0,
-      "step": 43893
-    },
-    {
-      "epoch": 3.3560028289083856,
-      "grad_norm": 0.0035076544154435396,
-      "learning_rate": 0.00019999444462828482,
-      "loss": 46.0,
-      "step": 43894
-    },
-    {
-      "epoch": 3.356079285891775,
-      "grad_norm": 0.005666936747729778,
-      "learning_rate": 0.00019999444437509992,
-      "loss": 46.0,
-      "step": 43895
-    },
-    {
-      "epoch": 3.3561557428751647,
-      "grad_norm": 0.0021470433566719294,
-      "learning_rate": 0.00019999444412190922,
-      "loss": 46.0,
-      "step": 43896
-    },
-    {
-      "epoch": 3.3562321998585545,
-      "grad_norm": 0.0028956576716154814,
-      "learning_rate": 0.00019999444386871275,
-      "loss": 46.0,
-      "step": 43897
-    },
-    {
-      "epoch": 3.3563086568419442,
-      "grad_norm": 0.0007516075856983662,
-      "learning_rate": 0.0001999944436155105,
-      "loss": 46.0,
-      "step": 43898
-    },
-    {
-      "epoch": 3.356385113825334,
-      "grad_norm": 0.0009589249966666102,
-      "learning_rate": 0.00019999444336230254,
-      "loss": 46.0,
-      "step": 43899
-    },
-    {
-      "epoch": 3.3564615708087238,
-      "grad_norm": 0.000699333380907774,
-      "learning_rate": 0.00019999444310908878,
-      "loss": 46.0,
-      "step": 43900
-    },
-    {
-      "epoch": 3.3565380277921135,
-      "grad_norm": 0.001205030595883727,
-      "learning_rate": 0.00019999444285586921,
-      "loss": 46.0,
-      "step": 43901
-    },
-    {
-      "epoch": 3.3566144847755033,
-      "grad_norm": 0.003453389508649707,
-      "learning_rate": 0.0001999944426026439,
-      "loss": 46.0,
-      "step": 43902
-    },
-    {
-      "epoch": 3.356690941758893,
-      "grad_norm": 0.0017191831720992923,
-      "learning_rate": 0.00019999444234941285,
-      "loss": 46.0,
-      "step": 43903
-    },
-    {
-      "epoch": 3.356767398742283,
-      "grad_norm": 0.006386714521795511,
-      "learning_rate": 0.000199994442096176,
-      "loss": 46.0,
-      "step": 43904
-    },
-    {
-      "epoch": 3.356843855725672,
-      "grad_norm": 0.004084683954715729,
-      "learning_rate": 0.00019999444184293336,
-      "loss": 46.0,
-      "step": 43905
-    },
-    {
-      "epoch": 3.356920312709062,
-      "grad_norm": 0.004364202730357647,
-      "learning_rate": 0.00019999444158968498,
-      "loss": 46.0,
-      "step": 43906
-    },
-    {
-      "epoch": 3.3569967696924516,
-      "grad_norm": 0.002384310821071267,
-      "learning_rate": 0.00019999444133643083,
-      "loss": 46.0,
-      "step": 43907
-    },
-    {
-      "epoch": 3.3570732266758414,
-      "grad_norm": 0.004904799163341522,
-      "learning_rate": 0.00019999444108317093,
-      "loss": 46.0,
-      "step": 43908
-    },
-    {
-      "epoch": 3.357149683659231,
-      "grad_norm": 0.0026783396024256945,
-      "learning_rate": 0.0001999944408299052,
-      "loss": 46.0,
-      "step": 43909
-    },
-    {
-      "epoch": 3.357226140642621,
-      "grad_norm": 0.000880925974342972,
-      "learning_rate": 0.00019999444057663377,
-      "loss": 46.0,
-      "step": 43910
-    },
-    {
-      "epoch": 3.3573025976260107,
-      "grad_norm": 0.0023351088166236877,
-      "learning_rate": 0.00019999444032335655,
-      "loss": 46.0,
-      "step": 43911
-    },
-    {
-      "epoch": 3.3573790546094004,
-      "grad_norm": 0.0018413979560136795,
-      "learning_rate": 0.00019999444007007356,
-      "loss": 46.0,
-      "step": 43912
-    },
-    {
-      "epoch": 3.35745551159279,
-      "grad_norm": 0.006709543988108635,
-      "learning_rate": 0.0001999944398167848,
-      "loss": 46.0,
-      "step": 43913
-    },
-    {
-      "epoch": 3.35753196857618,
-      "grad_norm": 0.004390840418636799,
-      "learning_rate": 0.00019999443956349023,
-      "loss": 46.0,
-      "step": 43914
-    },
-    {
-      "epoch": 3.3576084255595697,
-      "grad_norm": 0.002180737443268299,
-      "learning_rate": 0.00019999443931018992,
-      "loss": 46.0,
-      "step": 43915
-    },
-    {
-      "epoch": 3.3576848825429595,
-      "grad_norm": 0.0042215511202812195,
-      "learning_rate": 0.00019999443905688386,
-      "loss": 46.0,
-      "step": 43916
-    },
-    {
-      "epoch": 3.357761339526349,
-      "grad_norm": 0.001434640376828611,
-      "learning_rate": 0.00019999443880357203,
-      "loss": 46.0,
-      "step": 43917
-    },
-    {
-      "epoch": 3.3578377965097386,
-      "grad_norm": 0.0031608024146407843,
-      "learning_rate": 0.0001999944385502544,
-      "loss": 46.0,
-      "step": 43918
-    },
-    {
-      "epoch": 3.3579142534931283,
-      "grad_norm": 0.0013899975456297398,
-      "learning_rate": 0.00019999443829693103,
-      "loss": 46.0,
-      "step": 43919
-    },
-    {
-      "epoch": 3.357990710476518,
-      "grad_norm": 0.0026645592879503965,
-      "learning_rate": 0.00019999443804360188,
-      "loss": 46.0,
-      "step": 43920
-    },
-    {
-      "epoch": 3.358067167459908,
-      "grad_norm": 0.001703485380858183,
-      "learning_rate": 0.00019999443779026695,
-      "loss": 46.0,
-      "step": 43921
-    },
-    {
-      "epoch": 3.3581436244432976,
-      "grad_norm": 0.005464259535074234,
-      "learning_rate": 0.00019999443753692626,
-      "loss": 46.0,
-      "step": 43922
-    },
-    {
-      "epoch": 3.3582200814266874,
-      "grad_norm": 0.0031179413199424744,
-      "learning_rate": 0.0001999944372835798,
-      "loss": 46.0,
-      "step": 43923
-    },
-    {
-      "epoch": 3.358296538410077,
-      "grad_norm": 0.0014420761726796627,
-      "learning_rate": 0.00019999443703022757,
-      "loss": 46.0,
-      "step": 43924
-    },
-    {
-      "epoch": 3.358372995393467,
-      "grad_norm": 0.0010481172939762473,
-      "learning_rate": 0.00019999443677686958,
-      "loss": 46.0,
-      "step": 43925
-    },
-    {
-      "epoch": 3.3584494523768567,
-      "grad_norm": 0.0006573973223567009,
-      "learning_rate": 0.0001999944365235058,
-      "loss": 46.0,
-      "step": 43926
-    },
-    {
-      "epoch": 3.358525909360246,
-      "grad_norm": 0.00368470954708755,
-      "learning_rate": 0.00019999443627013628,
-      "loss": 46.0,
-      "step": 43927
-    },
-    {
-      "epoch": 3.3586023663436357,
-      "grad_norm": 0.000752764695789665,
-      "learning_rate": 0.00019999443601676097,
-      "loss": 46.0,
-      "step": 43928
-    },
-    {
-      "epoch": 3.3586788233270255,
-      "grad_norm": 0.00128778291400522,
-      "learning_rate": 0.0001999944357633799,
-      "loss": 46.0,
-      "step": 43929
-    },
-    {
-      "epoch": 3.3587552803104153,
-      "grad_norm": 0.005448696669191122,
-      "learning_rate": 0.00019999443550999306,
-      "loss": 46.0,
-      "step": 43930
-    },
-    {
-      "epoch": 3.358831737293805,
-      "grad_norm": 0.001997839193791151,
-      "learning_rate": 0.00019999443525660045,
-      "loss": 46.0,
-      "step": 43931
-    },
-    {
-      "epoch": 3.358908194277195,
-      "grad_norm": 0.0006499068695120513,
-      "learning_rate": 0.00019999443500320205,
-      "loss": 46.0,
-      "step": 43932
-    },
-    {
-      "epoch": 3.3589846512605845,
-      "grad_norm": 0.0018064759206026793,
-      "learning_rate": 0.0001999944347497979,
-      "loss": 46.0,
-      "step": 43933
-    },
-    {
-      "epoch": 3.3590611082439743,
-      "grad_norm": 0.011833742260932922,
-      "learning_rate": 0.000199994434496388,
-      "loss": 46.0,
-      "step": 43934
-    },
-    {
-      "epoch": 3.359137565227364,
-      "grad_norm": 0.0018127337098121643,
-      "learning_rate": 0.0001999944342429723,
-      "loss": 46.0,
-      "step": 43935
-    },
-    {
-      "epoch": 3.359214022210754,
-      "grad_norm": 0.0012243735836818814,
-      "learning_rate": 0.00019999443398955084,
-      "loss": 46.0,
-      "step": 43936
-    },
-    {
-      "epoch": 3.3592904791941436,
-      "grad_norm": 0.001750829047523439,
-      "learning_rate": 0.00019999443373612362,
-      "loss": 46.0,
-      "step": 43937
-    },
-    {
-      "epoch": 3.3593669361775333,
-      "grad_norm": 0.0028012520633637905,
-      "learning_rate": 0.00019999443348269066,
-      "loss": 46.0,
-      "step": 43938
-    },
-    {
-      "epoch": 3.3594433931609227,
-      "grad_norm": 0.0010618646629154682,
-      "learning_rate": 0.00019999443322925187,
-      "loss": 46.0,
-      "step": 43939
-    },
-    {
-      "epoch": 3.3595198501443124,
-      "grad_norm": 0.001202001003548503,
-      "learning_rate": 0.00019999443297580733,
-      "loss": 46.0,
-      "step": 43940
-    },
-    {
-      "epoch": 3.359596307127702,
-      "grad_norm": 0.0034605555702000856,
-      "learning_rate": 0.00019999443272235702,
-      "loss": 46.0,
-      "step": 43941
-    },
-    {
-      "epoch": 3.359672764111092,
-      "grad_norm": 0.0019187383586540818,
-      "learning_rate": 0.00019999443246890096,
-      "loss": 46.0,
-      "step": 43942
-    },
-    {
-      "epoch": 3.3597492210944817,
-      "grad_norm": 0.004395488183945417,
-      "learning_rate": 0.0001999944322154391,
-      "loss": 46.0,
-      "step": 43943
-    },
-    {
-      "epoch": 3.3598256780778715,
-      "grad_norm": 0.01191452331840992,
-      "learning_rate": 0.0001999944319619715,
-      "loss": 46.0,
-      "step": 43944
-    },
-    {
-      "epoch": 3.3599021350612612,
-      "grad_norm": 0.004040820989757776,
-      "learning_rate": 0.00019999443170849813,
-      "loss": 46.0,
-      "step": 43945
-    },
-    {
-      "epoch": 3.359978592044651,
-      "grad_norm": 0.0014816161710768938,
-      "learning_rate": 0.00019999443145501898,
-      "loss": 46.0,
-      "step": 43946
-    },
-    {
-      "epoch": 3.3600550490280408,
-      "grad_norm": 0.005345433950424194,
-      "learning_rate": 0.00019999443120153406,
-      "loss": 46.0,
-      "step": 43947
-    },
-    {
-      "epoch": 3.36013150601143,
-      "grad_norm": 0.0031253669876605272,
-      "learning_rate": 0.0001999944309480434,
-      "loss": 46.0,
-      "step": 43948
-    },
-    {
-      "epoch": 3.36020796299482,
-      "grad_norm": 0.0006608832045458257,
-      "learning_rate": 0.00019999443069454692,
-      "loss": 46.0,
-      "step": 43949
-    },
-    {
-      "epoch": 3.3602844199782096,
-      "grad_norm": 0.0025015934370458126,
-      "learning_rate": 0.0001999944304410447,
-      "loss": 46.0,
-      "step": 43950
-    },
-    {
-      "epoch": 3.3603608769615994,
-      "grad_norm": 0.006387869827449322,
-      "learning_rate": 0.0001999944301875367,
-      "loss": 46.0,
-      "step": 43951
-    },
-    {
-      "epoch": 3.360437333944989,
-      "grad_norm": 0.006368723232299089,
-      "learning_rate": 0.00019999442993402292,
-      "loss": 46.0,
-      "step": 43952
-    },
-    {
-      "epoch": 3.360513790928379,
-      "grad_norm": 0.01100991666316986,
-      "learning_rate": 0.00019999442968050341,
-      "loss": 46.0,
-      "step": 43953
-    },
-    {
-      "epoch": 3.3605902479117686,
-      "grad_norm": 0.0021573035046458244,
-      "learning_rate": 0.0001999944294269781,
-      "loss": 46.0,
-      "step": 43954
-    },
-    {
-      "epoch": 3.3606667048951584,
-      "grad_norm": 0.001826654770411551,
-      "learning_rate": 0.00019999442917344705,
-      "loss": 46.0,
-      "step": 43955
-    },
-    {
-      "epoch": 3.360743161878548,
-      "grad_norm": 0.0032256762497127056,
-      "learning_rate": 0.0001999944289199102,
-      "loss": 46.0,
-      "step": 43956
-    },
-    {
-      "epoch": 3.360819618861938,
-      "grad_norm": 0.0033450177870690823,
-      "learning_rate": 0.00019999442866636756,
-      "loss": 46.0,
-      "step": 43957
-    },
-    {
-      "epoch": 3.3608960758453277,
-      "grad_norm": 0.0007844028295949101,
-      "learning_rate": 0.00019999442841281921,
-      "loss": 46.0,
-      "step": 43958
-    },
-    {
-      "epoch": 3.3609725328287174,
-      "grad_norm": 0.0015928855864331126,
-      "learning_rate": 0.0001999944281592651,
-      "loss": 46.0,
-      "step": 43959
-    },
-    {
-      "epoch": 3.361048989812107,
-      "grad_norm": 0.0015519981971010566,
-      "learning_rate": 0.00019999442790570514,
-      "loss": 46.0,
-      "step": 43960
-    },
-    {
-      "epoch": 3.3611254467954965,
-      "grad_norm": 0.0008716603042557836,
-      "learning_rate": 0.00019999442765213945,
-      "loss": 46.0,
-      "step": 43961
-    },
-    {
-      "epoch": 3.3612019037788863,
-      "grad_norm": 0.006894951220601797,
-      "learning_rate": 0.00019999442739856803,
-      "loss": 46.0,
-      "step": 43962
-    },
-    {
-      "epoch": 3.361278360762276,
-      "grad_norm": 0.0014548986218869686,
-      "learning_rate": 0.00019999442714499076,
-      "loss": 46.0,
-      "step": 43963
-    },
-    {
-      "epoch": 3.361354817745666,
-      "grad_norm": 0.002383999526500702,
-      "learning_rate": 0.0001999944268914078,
-      "loss": 46.0,
-      "step": 43964
-    },
-    {
-      "epoch": 3.3614312747290556,
-      "grad_norm": 0.0036437136586755514,
-      "learning_rate": 0.00019999442663781904,
-      "loss": 46.0,
-      "step": 43965
-    },
-    {
-      "epoch": 3.3615077317124453,
-      "grad_norm": 0.0018820744007825851,
-      "learning_rate": 0.0001999944263842245,
-      "loss": 46.0,
-      "step": 43966
-    },
-    {
-      "epoch": 3.361584188695835,
-      "grad_norm": 0.002455133944749832,
-      "learning_rate": 0.0001999944261306242,
-      "loss": 46.0,
-      "step": 43967
-    },
-    {
-      "epoch": 3.361660645679225,
-      "grad_norm": 0.0037344032898545265,
-      "learning_rate": 0.00019999442587701814,
-      "loss": 46.0,
-      "step": 43968
-    },
-    {
-      "epoch": 3.3617371026626146,
-      "grad_norm": 0.0027202623896300793,
-      "learning_rate": 0.00019999442562340628,
-      "loss": 46.0,
-      "step": 43969
-    },
-    {
-      "epoch": 3.361813559646004,
-      "grad_norm": 0.0007057629991322756,
-      "learning_rate": 0.0001999944253697887,
-      "loss": 46.0,
-      "step": 43970
-    },
-    {
-      "epoch": 3.3618900166293937,
-      "grad_norm": 0.0024618864990770817,
-      "learning_rate": 0.0001999944251161653,
-      "loss": 46.0,
-      "step": 43971
-    },
-    {
-      "epoch": 3.3619664736127834,
-      "grad_norm": 0.0018257752526551485,
-      "learning_rate": 0.00019999442486253615,
-      "loss": 46.0,
-      "step": 43972
-    },
-    {
-      "epoch": 3.362042930596173,
-      "grad_norm": 0.0017023581312969327,
-      "learning_rate": 0.00019999442460890126,
-      "loss": 46.0,
-      "step": 43973
-    },
-    {
-      "epoch": 3.362119387579563,
-      "grad_norm": 0.0012312245089560747,
-      "learning_rate": 0.00019999442435526054,
-      "loss": 46.0,
-      "step": 43974
-    },
-    {
-      "epoch": 3.3621958445629527,
-      "grad_norm": 0.001370420679450035,
-      "learning_rate": 0.00019999442410161412,
-      "loss": 46.0,
-      "step": 43975
-    },
-    {
-      "epoch": 3.3622723015463425,
-      "grad_norm": 0.003550641005858779,
-      "learning_rate": 0.0001999944238479619,
-      "loss": 46.0,
-      "step": 43976
-    },
-    {
-      "epoch": 3.3623487585297323,
-      "grad_norm": 0.0037540870252996683,
-      "learning_rate": 0.0001999944235943039,
-      "loss": 46.0,
-      "step": 43977
-    },
-    {
-      "epoch": 3.362425215513122,
-      "grad_norm": 0.004224210046231747,
-      "learning_rate": 0.00019999442334064013,
-      "loss": 46.0,
-      "step": 43978
-    },
-    {
-      "epoch": 3.3625016724965118,
-      "grad_norm": 0.002612208016216755,
-      "learning_rate": 0.00019999442308697062,
-      "loss": 46.0,
-      "step": 43979
-    },
-    {
-      "epoch": 3.3625781294799015,
-      "grad_norm": 0.000882496009580791,
-      "learning_rate": 0.0001999944228332953,
-      "loss": 46.0,
-      "step": 43980
-    },
-    {
-      "epoch": 3.3626545864632913,
-      "grad_norm": 0.0011672910768538713,
-      "learning_rate": 0.00019999442257961426,
-      "loss": 46.0,
-      "step": 43981
-    },
-    {
-      "epoch": 3.362731043446681,
-      "grad_norm": 0.0037730110343545675,
-      "learning_rate": 0.0001999944223259274,
-      "loss": 46.0,
-      "step": 43982
-    },
-    {
-      "epoch": 3.3628075004300704,
-      "grad_norm": 0.0035125119611620903,
-      "learning_rate": 0.00019999442207223477,
-      "loss": 46.0,
-      "step": 43983
-    },
-    {
-      "epoch": 3.36288395741346,
-      "grad_norm": 0.0015022674342617393,
-      "learning_rate": 0.00019999442181853642,
-      "loss": 46.0,
-      "step": 43984
-    },
-    {
-      "epoch": 3.36296041439685,
-      "grad_norm": 0.0015277877682819963,
-      "learning_rate": 0.00019999442156483225,
-      "loss": 46.0,
-      "step": 43985
-    },
-    {
-      "epoch": 3.3630368713802397,
-      "grad_norm": 0.0017599384300410748,
-      "learning_rate": 0.00019999442131112235,
-      "loss": 46.0,
-      "step": 43986
-    },
-    {
-      "epoch": 3.3631133283636294,
-      "grad_norm": 0.005569704808294773,
-      "learning_rate": 0.00019999442105740668,
-      "loss": 46.0,
-      "step": 43987
-    },
-    {
-      "epoch": 3.363189785347019,
-      "grad_norm": 0.0029773619025945663,
-      "learning_rate": 0.00019999442080368521,
-      "loss": 46.0,
-      "step": 43988
-    },
-    {
-      "epoch": 3.363266242330409,
-      "grad_norm": 0.0017659197328612208,
-      "learning_rate": 0.000199994420549958,
-      "loss": 46.0,
-      "step": 43989
-    },
-    {
-      "epoch": 3.3633426993137987,
-      "grad_norm": 0.0030685404781252146,
-      "learning_rate": 0.000199994420296225,
-      "loss": 46.0,
-      "step": 43990
-    },
-    {
-      "epoch": 3.3634191562971885,
-      "grad_norm": 0.0006195066962391138,
-      "learning_rate": 0.00019999442004248625,
-      "loss": 46.0,
-      "step": 43991
-    },
-    {
-      "epoch": 3.363495613280578,
-      "grad_norm": 0.004994519986212254,
-      "learning_rate": 0.00019999441978874174,
-      "loss": 46.0,
-      "step": 43992
-    },
-    {
-      "epoch": 3.3635720702639675,
-      "grad_norm": 0.0015531317330896854,
-      "learning_rate": 0.00019999441953499143,
-      "loss": 46.0,
-      "step": 43993
-    },
-    {
-      "epoch": 3.3636485272473573,
-      "grad_norm": 0.0031213618349283934,
-      "learning_rate": 0.00019999441928123535,
-      "loss": 46.0,
-      "step": 43994
-    },
-    {
-      "epoch": 3.363724984230747,
-      "grad_norm": 0.001442785025574267,
-      "learning_rate": 0.00019999441902747352,
-      "loss": 46.0,
-      "step": 43995
-    },
-    {
-      "epoch": 3.363801441214137,
-      "grad_norm": 0.0038165568839758635,
-      "learning_rate": 0.0001999944187737059,
-      "loss": 46.0,
-      "step": 43996
-    },
-    {
-      "epoch": 3.3638778981975266,
-      "grad_norm": 0.005286256782710552,
-      "learning_rate": 0.00019999441851993255,
-      "loss": 46.0,
-      "step": 43997
-    },
-    {
-      "epoch": 3.3639543551809163,
-      "grad_norm": 0.0017717641312628984,
-      "learning_rate": 0.00019999441826615343,
-      "loss": 46.0,
-      "step": 43998
-    },
-    {
-      "epoch": 3.364030812164306,
-      "grad_norm": 0.0013908063992857933,
-      "learning_rate": 0.00019999441801236848,
-      "loss": 46.0,
-      "step": 43999
-    },
-    {
-      "epoch": 3.364107269147696,
-      "grad_norm": 0.0025197898503392935,
-      "learning_rate": 0.0001999944177585778,
-      "loss": 46.0,
-      "step": 44000
-    },
-    {
-      "epoch": 3.3641837261310856,
-      "grad_norm": 0.0017711903201416135,
-      "learning_rate": 0.00019999441750478137,
-      "loss": 46.0,
-      "step": 44001
-    },
-    {
-      "epoch": 3.3642601831144754,
-      "grad_norm": 0.002247098134830594,
-      "learning_rate": 0.00019999441725097913,
-      "loss": 46.0,
-      "step": 44002
-    },
-    {
-      "epoch": 3.364336640097865,
-      "grad_norm": 0.0016384284244850278,
-      "learning_rate": 0.00019999441699717116,
-      "loss": 46.0,
-      "step": 44003
-    },
-    {
-      "epoch": 3.3644130970812545,
-      "grad_norm": 0.0013083242811262608,
-      "learning_rate": 0.0001999944167433574,
-      "loss": 46.0,
-      "step": 44004
-    },
-    {
-      "epoch": 3.3644895540646442,
-      "grad_norm": 0.007745744194835424,
-      "learning_rate": 0.00019999441648953784,
-      "loss": 46.0,
-      "step": 44005
-    },
-    {
-      "epoch": 3.364566011048034,
-      "grad_norm": 0.003586008446291089,
-      "learning_rate": 0.0001999944162357126,
-      "loss": 46.0,
-      "step": 44006
-    },
-    {
-      "epoch": 3.3646424680314237,
-      "grad_norm": 0.0019406918436288834,
-      "learning_rate": 0.00019999441598188148,
-      "loss": 46.0,
-      "step": 44007
-    },
-    {
-      "epoch": 3.3647189250148135,
-      "grad_norm": 0.0015440720599144697,
-      "learning_rate": 0.00019999441572804468,
-      "loss": 46.0,
-      "step": 44008
-    },
-    {
-      "epoch": 3.3647953819982033,
-      "grad_norm": 0.0026671465020626783,
-      "learning_rate": 0.00019999441547420208,
-      "loss": 46.0,
-      "step": 44009
-    },
-    {
-      "epoch": 3.364871838981593,
-      "grad_norm": 0.0012204390950500965,
-      "learning_rate": 0.0001999944152203537,
-      "loss": 46.0,
-      "step": 44010
-    },
-    {
-      "epoch": 3.364948295964983,
-      "grad_norm": 0.0016009026439860463,
-      "learning_rate": 0.00019999441496649955,
-      "loss": 46.0,
-      "step": 44011
-    },
-    {
-      "epoch": 3.3650247529483726,
-      "grad_norm": 0.0020493201445788145,
-      "learning_rate": 0.00019999441471263963,
-      "loss": 46.0,
-      "step": 44012
-    },
-    {
-      "epoch": 3.3651012099317623,
-      "grad_norm": 0.007693315856158733,
-      "learning_rate": 0.00019999441445877396,
-      "loss": 46.0,
-      "step": 44013
-    },
-    {
-      "epoch": 3.3651776669151516,
-      "grad_norm": 0.0016419970197603106,
-      "learning_rate": 0.00019999441420490252,
-      "loss": 46.0,
-      "step": 44014
-    },
-    {
-      "epoch": 3.3652541238985414,
-      "grad_norm": 0.0019955032039433718,
-      "learning_rate": 0.00019999441395102528,
-      "loss": 46.0,
-      "step": 44015
-    },
-    {
-      "epoch": 3.365330580881931,
-      "grad_norm": 0.007652727421373129,
-      "learning_rate": 0.0001999944136971423,
-      "loss": 46.0,
-      "step": 44016
-    },
-    {
-      "epoch": 3.365407037865321,
-      "grad_norm": 0.002051898743957281,
-      "learning_rate": 0.00019999441344325356,
-      "loss": 46.0,
-      "step": 44017
-    },
-    {
-      "epoch": 3.3654834948487107,
-      "grad_norm": 0.0009151982958428562,
-      "learning_rate": 0.000199994413189359,
-      "loss": 46.0,
-      "step": 44018
-    },
-    {
-      "epoch": 3.3655599518321004,
-      "grad_norm": 0.0034932363778352737,
-      "learning_rate": 0.00019999441293545872,
-      "loss": 46.0,
-      "step": 44019
-    },
-    {
-      "epoch": 3.36563640881549,
-      "grad_norm": 0.0003169063711538911,
-      "learning_rate": 0.00019999441268155266,
-      "loss": 46.0,
-      "step": 44020
-    },
-    {
-      "epoch": 3.36571286579888,
-      "grad_norm": 0.004807604476809502,
-      "learning_rate": 0.00019999441242764084,
-      "loss": 46.0,
-      "step": 44021
-    },
-    {
-      "epoch": 3.3657893227822697,
-      "grad_norm": 0.002756100380793214,
-      "learning_rate": 0.0001999944121737232,
-      "loss": 46.0,
-      "step": 44022
-    },
-    {
-      "epoch": 3.3658657797656595,
-      "grad_norm": 0.0009285946725867689,
-      "learning_rate": 0.00019999441191979983,
-      "loss": 46.0,
-      "step": 44023
-    },
-    {
-      "epoch": 3.3659422367490492,
-      "grad_norm": 0.0025340397842228413,
-      "learning_rate": 0.00019999441166587071,
-      "loss": 46.0,
-      "step": 44024
-    },
-    {
-      "epoch": 3.366018693732439,
-      "grad_norm": 0.003007646184414625,
-      "learning_rate": 0.0001999944114119358,
-      "loss": 46.0,
-      "step": 44025
-    },
-    {
-      "epoch": 3.3660951507158283,
-      "grad_norm": 0.003365627024322748,
-      "learning_rate": 0.00019999441115799513,
-      "loss": 46.0,
-      "step": 44026
-    },
-    {
-      "epoch": 3.366171607699218,
-      "grad_norm": 0.00034670779132284224,
-      "learning_rate": 0.00019999441090404866,
-      "loss": 46.0,
-      "step": 44027
-    },
-    {
-      "epoch": 3.366248064682608,
-      "grad_norm": 0.0029784946236759424,
-      "learning_rate": 0.00019999441065009647,
-      "loss": 46.0,
-      "step": 44028
-    },
-    {
-      "epoch": 3.3663245216659976,
-      "grad_norm": 0.0007970849983394146,
-      "learning_rate": 0.00019999441039613846,
-      "loss": 46.0,
-      "step": 44029
-    },
-    {
-      "epoch": 3.3664009786493874,
-      "grad_norm": 0.0034228123258799314,
-      "learning_rate": 0.00019999441014217472,
-      "loss": 46.0,
-      "step": 44030
-    },
-    {
-      "epoch": 3.366477435632777,
-      "grad_norm": 0.0014420750085264444,
-      "learning_rate": 0.0001999944098882052,
-      "loss": 46.0,
-      "step": 44031
-    },
-    {
-      "epoch": 3.366553892616167,
-      "grad_norm": 0.003425904316827655,
-      "learning_rate": 0.00019999440963422988,
-      "loss": 46.0,
-      "step": 44032
-    },
-    {
-      "epoch": 3.3666303495995566,
-      "grad_norm": 0.002243151655420661,
-      "learning_rate": 0.00019999440938024883,
-      "loss": 46.0,
-      "step": 44033
-    },
-    {
-      "epoch": 3.3667068065829464,
-      "grad_norm": 0.0006195084424689412,
-      "learning_rate": 0.000199994409126262,
-      "loss": 46.0,
-      "step": 44034
-    },
-    {
-      "epoch": 3.366783263566336,
-      "grad_norm": 0.0022923778742551804,
-      "learning_rate": 0.0001999944088722694,
-      "loss": 46.0,
-      "step": 44035
-    },
-    {
-      "epoch": 3.3668597205497255,
-      "grad_norm": 0.0063108112663030624,
-      "learning_rate": 0.00019999440861827102,
-      "loss": 46.0,
-      "step": 44036
-    },
-    {
-      "epoch": 3.3669361775331152,
-      "grad_norm": 0.002106614410877228,
-      "learning_rate": 0.0001999944083642669,
-      "loss": 46.0,
-      "step": 44037
-    },
-    {
-      "epoch": 3.367012634516505,
-      "grad_norm": 0.003978736698627472,
-      "learning_rate": 0.00019999440811025698,
-      "loss": 46.0,
-      "step": 44038
-    },
-    {
-      "epoch": 3.3670890914998948,
-      "grad_norm": 0.0024155403953045607,
-      "learning_rate": 0.00019999440785624132,
-      "loss": 46.0,
-      "step": 44039
-    },
-    {
-      "epoch": 3.3671655484832845,
-      "grad_norm": 0.0008293219725601375,
-      "learning_rate": 0.00019999440760221985,
-      "loss": 46.0,
-      "step": 44040
-    },
-    {
-      "epoch": 3.3672420054666743,
-      "grad_norm": 0.0007247526082210243,
-      "learning_rate": 0.00019999440734819266,
-      "loss": 46.0,
-      "step": 44041
-    },
-    {
-      "epoch": 3.367318462450064,
-      "grad_norm": 0.003393264254555106,
-      "learning_rate": 0.00019999440709415968,
-      "loss": 46.0,
-      "step": 44042
-    },
-    {
-      "epoch": 3.367394919433454,
-      "grad_norm": 0.0011567908804863691,
-      "learning_rate": 0.0001999944068401209,
-      "loss": 46.0,
-      "step": 44043
-    },
-    {
-      "epoch": 3.3674713764168436,
-      "grad_norm": 0.0015661921352148056,
-      "learning_rate": 0.00019999440658607638,
-      "loss": 46.0,
-      "step": 44044
-    },
-    {
-      "epoch": 3.3675478334002333,
-      "grad_norm": 0.0016815419076010585,
-      "learning_rate": 0.00019999440633202607,
-      "loss": 46.0,
-      "step": 44045
-    },
-    {
-      "epoch": 3.367624290383623,
-      "grad_norm": 0.001174632809124887,
-      "learning_rate": 0.00019999440607797002,
-      "loss": 46.0,
-      "step": 44046
-    },
-    {
-      "epoch": 3.367700747367013,
-      "grad_norm": 0.0007991038146428764,
-      "learning_rate": 0.0001999944058239082,
-      "loss": 46.0,
-      "step": 44047
-    },
-    {
-      "epoch": 3.367777204350402,
-      "grad_norm": 0.0025518157053738832,
-      "learning_rate": 0.0001999944055698406,
-      "loss": 46.0,
-      "step": 44048
-    },
-    {
-      "epoch": 3.367853661333792,
-      "grad_norm": 0.0033297755289822817,
-      "learning_rate": 0.00019999440531576722,
-      "loss": 46.0,
-      "step": 44049
-    },
-    {
-      "epoch": 3.3679301183171817,
-      "grad_norm": 0.002374293515458703,
-      "learning_rate": 0.0001999944050616881,
-      "loss": 46.0,
-      "step": 44050
-    },
-    {
-      "epoch": 3.3680065753005715,
-      "grad_norm": 0.0014175818068906665,
-      "learning_rate": 0.00019999440480760318,
-      "loss": 46.0,
-      "step": 44051
-    },
-    {
-      "epoch": 3.368083032283961,
-      "grad_norm": 0.0011245246278122067,
-      "learning_rate": 0.0001999944045535125,
-      "loss": 46.0,
-      "step": 44052
-    },
-    {
-      "epoch": 3.368159489267351,
-      "grad_norm": 0.002561658388003707,
-      "learning_rate": 0.00019999440429941605,
-      "loss": 46.0,
-      "step": 44053
-    },
-    {
-      "epoch": 3.3682359462507407,
-      "grad_norm": 0.0017387623665854335,
-      "learning_rate": 0.00019999440404531383,
-      "loss": 46.0,
-      "step": 44054
-    },
-    {
-      "epoch": 3.3683124032341305,
-      "grad_norm": 0.0013006882509216666,
-      "learning_rate": 0.00019999440379120587,
-      "loss": 46.0,
-      "step": 44055
-    },
-    {
-      "epoch": 3.3683888602175203,
-      "grad_norm": 0.0020123799331486225,
-      "learning_rate": 0.0001999944035370921,
-      "loss": 46.0,
-      "step": 44056
-    },
-    {
-      "epoch": 3.36846531720091,
-      "grad_norm": 0.0022627043072134256,
-      "learning_rate": 0.00019999440328297258,
-      "loss": 46.0,
-      "step": 44057
-    },
-    {
-      "epoch": 3.3685417741842993,
-      "grad_norm": 0.002851160941645503,
-      "learning_rate": 0.0001999944030288473,
-      "loss": 46.0,
-      "step": 44058
-    },
-    {
-      "epoch": 3.368618231167689,
-      "grad_norm": 0.003905707038938999,
-      "learning_rate": 0.00019999440277471625,
-      "loss": 46.0,
-      "step": 44059
-    },
-    {
-      "epoch": 3.368694688151079,
-      "grad_norm": 0.0008961720741353929,
-      "learning_rate": 0.00019999440252057942,
-      "loss": 46.0,
-      "step": 44060
-    },
-    {
-      "epoch": 3.3687711451344686,
-      "grad_norm": 0.0030855913646519184,
-      "learning_rate": 0.0001999944022664368,
-      "loss": 46.0,
-      "step": 44061
-    },
-    {
-      "epoch": 3.3688476021178584,
-      "grad_norm": 0.003867947030812502,
-      "learning_rate": 0.00019999440201228845,
-      "loss": 46.0,
-      "step": 44062
-    },
-    {
-      "epoch": 3.368924059101248,
-      "grad_norm": 0.021704386919736862,
-      "learning_rate": 0.0001999944017581343,
-      "loss": 46.0,
-      "step": 44063
-    },
-    {
-      "epoch": 3.369000516084638,
-      "grad_norm": 0.001425275462679565,
-      "learning_rate": 0.0001999944015039744,
-      "loss": 46.0,
-      "step": 44064
-    },
-    {
-      "epoch": 3.3690769730680277,
-      "grad_norm": 0.003588259220123291,
-      "learning_rate": 0.00019999440124980874,
-      "loss": 46.0,
-      "step": 44065
-    },
-    {
-      "epoch": 3.3691534300514174,
-      "grad_norm": 0.004983542952686548,
-      "learning_rate": 0.0001999944009956373,
-      "loss": 46.0,
-      "step": 44066
-    },
-    {
-      "epoch": 3.369229887034807,
-      "grad_norm": 0.0011447498109191656,
-      "learning_rate": 0.00019999440074146009,
-      "loss": 46.0,
-      "step": 44067
-    },
-    {
-      "epoch": 3.369306344018197,
-      "grad_norm": 0.013150081038475037,
-      "learning_rate": 0.0001999944004872771,
-      "loss": 46.0,
-      "step": 44068
-    },
-    {
-      "epoch": 3.3693828010015867,
-      "grad_norm": 0.001434782985597849,
-      "learning_rate": 0.00019999440023308834,
-      "loss": 46.0,
-      "step": 44069
-    },
-    {
-      "epoch": 3.369459257984976,
-      "grad_norm": 0.004751997068524361,
-      "learning_rate": 0.0001999943999788938,
-      "loss": 46.0,
-      "step": 44070
-    },
-    {
-      "epoch": 3.369535714968366,
-      "grad_norm": 0.0011649583466351032,
-      "learning_rate": 0.00019999439972469353,
-      "loss": 46.0,
-      "step": 44071
-    },
-    {
-      "epoch": 3.3696121719517556,
-      "grad_norm": 0.001877776114270091,
-      "learning_rate": 0.00019999439947048745,
-      "loss": 46.0,
-      "step": 44072
-    },
-    {
-      "epoch": 3.3696886289351453,
-      "grad_norm": 0.002230021171271801,
-      "learning_rate": 0.00019999439921627562,
-      "loss": 46.0,
-      "step": 44073
-    },
-    {
-      "epoch": 3.369765085918535,
-      "grad_norm": 0.0025933990254998207,
-      "learning_rate": 0.00019999439896205802,
-      "loss": 46.0,
-      "step": 44074
-    },
-    {
-      "epoch": 3.369841542901925,
-      "grad_norm": 0.0028062283527106047,
-      "learning_rate": 0.00019999439870783468,
-      "loss": 46.0,
-      "step": 44075
-    },
-    {
-      "epoch": 3.3699179998853146,
-      "grad_norm": 0.0006448200438171625,
-      "learning_rate": 0.00019999439845360556,
-      "loss": 46.0,
-      "step": 44076
-    },
-    {
-      "epoch": 3.3699944568687044,
-      "grad_norm": 0.007435072213411331,
-      "learning_rate": 0.00019999439819937064,
-      "loss": 46.0,
-      "step": 44077
-    },
-    {
-      "epoch": 3.370070913852094,
-      "grad_norm": 0.003749512368813157,
-      "learning_rate": 0.00019999439794512994,
-      "loss": 46.0,
-      "step": 44078
-    },
-    {
-      "epoch": 3.370147370835484,
-      "grad_norm": 0.002749453764408827,
-      "learning_rate": 0.00019999439769088353,
-      "loss": 46.0,
-      "step": 44079
-    },
-    {
-      "epoch": 3.370223827818873,
-      "grad_norm": 0.010053405538201332,
-      "learning_rate": 0.00019999439743663132,
-      "loss": 46.0,
-      "step": 44080
-    },
-    {
-      "epoch": 3.370300284802263,
-      "grad_norm": 0.0009766056900843978,
-      "learning_rate": 0.0001999943971823733,
-      "loss": 46.0,
-      "step": 44081
-    },
-    {
-      "epoch": 3.3703767417856527,
-      "grad_norm": 0.0015313500771299005,
-      "learning_rate": 0.00019999439692810957,
-      "loss": 46.0,
-      "step": 44082
-    },
-    {
-      "epoch": 3.3704531987690425,
-      "grad_norm": 0.002901981584727764,
-      "learning_rate": 0.00019999439667384007,
-      "loss": 46.0,
-      "step": 44083
-    },
-    {
-      "epoch": 3.3705296557524322,
-      "grad_norm": 0.0029382891952991486,
-      "learning_rate": 0.00019999439641956476,
-      "loss": 46.0,
-      "step": 44084
-    },
-    {
-      "epoch": 3.370606112735822,
-      "grad_norm": 0.003256638767197728,
-      "learning_rate": 0.0001999943961652837,
-      "loss": 46.0,
-      "step": 44085
-    },
-    {
-      "epoch": 3.3706825697192118,
-      "grad_norm": 0.0012262884993106127,
-      "learning_rate": 0.00019999439591099688,
-      "loss": 46.0,
-      "step": 44086
-    },
-    {
-      "epoch": 3.3707590267026015,
-      "grad_norm": 0.002431526780128479,
-      "learning_rate": 0.00019999439565670428,
-      "loss": 46.0,
-      "step": 44087
-    },
-    {
-      "epoch": 3.3708354836859913,
-      "grad_norm": 0.0016523622907698154,
-      "learning_rate": 0.0001999943954024059,
-      "loss": 46.0,
-      "step": 44088
-    },
-    {
-      "epoch": 3.370911940669381,
-      "grad_norm": 0.006879203021526337,
-      "learning_rate": 0.0001999943951481018,
-      "loss": 46.0,
-      "step": 44089
-    },
-    {
-      "epoch": 3.370988397652771,
-      "grad_norm": 0.00565417530015111,
-      "learning_rate": 0.00019999439489379187,
-      "loss": 46.0,
-      "step": 44090
-    },
-    {
-      "epoch": 3.3710648546361606,
-      "grad_norm": 0.0010815723799169064,
-      "learning_rate": 0.0001999943946394762,
-      "loss": 46.0,
-      "step": 44091
-    },
-    {
-      "epoch": 3.37114131161955,
-      "grad_norm": 0.00601259944960475,
-      "learning_rate": 0.00019999439438515477,
-      "loss": 46.0,
-      "step": 44092
-    },
-    {
-      "epoch": 3.3712177686029396,
-      "grad_norm": 0.001999577507376671,
-      "learning_rate": 0.00019999439413082755,
-      "loss": 46.0,
-      "step": 44093
-    },
-    {
-      "epoch": 3.3712942255863294,
-      "grad_norm": 0.002029047580435872,
-      "learning_rate": 0.0001999943938764946,
-      "loss": 46.0,
-      "step": 44094
-    },
-    {
-      "epoch": 3.371370682569719,
-      "grad_norm": 0.000913600146304816,
-      "learning_rate": 0.0001999943936221558,
-      "loss": 46.0,
-      "step": 44095
-    },
-    {
-      "epoch": 3.371447139553109,
-      "grad_norm": 0.0067564258351922035,
-      "learning_rate": 0.00019999439336781133,
-      "loss": 46.0,
-      "step": 44096
-    },
-    {
-      "epoch": 3.3715235965364987,
-      "grad_norm": 0.0032579984981566668,
-      "learning_rate": 0.00019999439311346102,
-      "loss": 46.0,
-      "step": 44097
-    },
-    {
-      "epoch": 3.3716000535198885,
-      "grad_norm": 0.001946762204170227,
-      "learning_rate": 0.00019999439285910494,
-      "loss": 46.0,
-      "step": 44098
-    },
-    {
-      "epoch": 3.371676510503278,
-      "grad_norm": 0.0015006192261353135,
-      "learning_rate": 0.00019999439260474312,
-      "loss": 46.0,
-      "step": 44099
-    },
-    {
-      "epoch": 3.371752967486668,
-      "grad_norm": 0.0017017036443576217,
-      "learning_rate": 0.00019999439235037555,
-      "loss": 46.0,
-      "step": 44100
-    },
-    {
-      "epoch": 3.3718294244700573,
-      "grad_norm": 0.005532213021069765,
-      "learning_rate": 0.00019999439209600217,
-      "loss": 46.0,
-      "step": 44101
-    },
-    {
-      "epoch": 3.371905881453447,
-      "grad_norm": 0.0014013760956004262,
-      "learning_rate": 0.00019999439184162306,
-      "loss": 46.0,
-      "step": 44102
-    },
-    {
-      "epoch": 3.371982338436837,
-      "grad_norm": 0.002641149330884218,
-      "learning_rate": 0.00019999439158723814,
-      "loss": 46.0,
-      "step": 44103
-    },
-    {
-      "epoch": 3.3720587954202266,
-      "grad_norm": 0.0010958026396110654,
-      "learning_rate": 0.00019999439133284747,
-      "loss": 46.0,
-      "step": 44104
-    },
-    {
-      "epoch": 3.3721352524036163,
-      "grad_norm": 0.0013367747887969017,
-      "learning_rate": 0.00019999439107845103,
-      "loss": 46.0,
-      "step": 44105
-    },
-    {
-      "epoch": 3.372211709387006,
-      "grad_norm": 0.0009631678694859147,
-      "learning_rate": 0.00019999439082404882,
-      "loss": 46.0,
-      "step": 44106
-    },
-    {
-      "epoch": 3.372288166370396,
-      "grad_norm": 0.0023285860661417246,
-      "learning_rate": 0.00019999439056964083,
-      "loss": 46.0,
-      "step": 44107
-    },
-    {
-      "epoch": 3.3723646233537856,
-      "grad_norm": 0.0023017036728560925,
-      "learning_rate": 0.00019999439031522708,
-      "loss": 46.0,
-      "step": 44108
-    },
-    {
-      "epoch": 3.3724410803371754,
-      "grad_norm": 0.004402284976094961,
-      "learning_rate": 0.00019999439006080757,
-      "loss": 46.0,
-      "step": 44109
-    },
-    {
-      "epoch": 3.372517537320565,
-      "grad_norm": 0.0014698015293106437,
-      "learning_rate": 0.0001999943898063823,
-      "loss": 46.0,
-      "step": 44110
-    },
-    {
-      "epoch": 3.372593994303955,
-      "grad_norm": 0.001998212421312928,
-      "learning_rate": 0.00019999438955195124,
-      "loss": 46.0,
-      "step": 44111
-    },
-    {
-      "epoch": 3.3726704512873447,
-      "grad_norm": 0.00285261869430542,
-      "learning_rate": 0.00019999438929751441,
-      "loss": 46.0,
-      "step": 44112
-    },
-    {
-      "epoch": 3.3727469082707344,
-      "grad_norm": 0.002480628900229931,
-      "learning_rate": 0.00019999438904307182,
-      "loss": 46.0,
-      "step": 44113
-    },
-    {
-      "epoch": 3.3728233652541237,
-      "grad_norm": 0.002849624725058675,
-      "learning_rate": 0.00019999438878862344,
-      "loss": 46.0,
-      "step": 44114
-    },
-    {
-      "epoch": 3.3728998222375135,
-      "grad_norm": 0.0015133980195969343,
-      "learning_rate": 0.00019999438853416935,
-      "loss": 46.0,
-      "step": 44115
-    },
-    {
-      "epoch": 3.3729762792209033,
-      "grad_norm": 0.0020540626719594,
-      "learning_rate": 0.00019999438827970943,
-      "loss": 46.0,
-      "step": 44116
-    },
-    {
-      "epoch": 3.373052736204293,
-      "grad_norm": 0.0014220286393538117,
-      "learning_rate": 0.00019999438802524377,
-      "loss": 46.0,
-      "step": 44117
-    },
-    {
-      "epoch": 3.373129193187683,
-      "grad_norm": 0.0015165775548666716,
-      "learning_rate": 0.0001999943877707723,
-      "loss": 46.0,
-      "step": 44118
-    },
-    {
-      "epoch": 3.3732056501710725,
-      "grad_norm": 0.0026727435179054737,
-      "learning_rate": 0.00019999438751629512,
-      "loss": 46.0,
-      "step": 44119
-    },
-    {
-      "epoch": 3.3732821071544623,
-      "grad_norm": 0.004713153466582298,
-      "learning_rate": 0.00019999438726181213,
-      "loss": 46.0,
-      "step": 44120
-    },
-    {
-      "epoch": 3.373358564137852,
-      "grad_norm": 0.001825318788178265,
-      "learning_rate": 0.0001999943870073234,
-      "loss": 46.0,
-      "step": 44121
-    },
-    {
-      "epoch": 3.373435021121242,
-      "grad_norm": 0.002042043022811413,
-      "learning_rate": 0.00019999438675282887,
-      "loss": 46.0,
-      "step": 44122
-    },
-    {
-      "epoch": 3.373511478104631,
-      "grad_norm": 0.0016553821042180061,
-      "learning_rate": 0.0001999943864983286,
-      "loss": 46.0,
-      "step": 44123
-    },
-    {
-      "epoch": 3.373587935088021,
-      "grad_norm": 0.0037303168792277575,
-      "learning_rate": 0.00019999438624382254,
-      "loss": 46.0,
-      "step": 44124
-    },
-    {
-      "epoch": 3.3736643920714107,
-      "grad_norm": 0.0015781475231051445,
-      "learning_rate": 0.00019999438598931071,
-      "loss": 46.0,
-      "step": 44125
-    },
-    {
-      "epoch": 3.3737408490548004,
-      "grad_norm": 0.005999335087835789,
-      "learning_rate": 0.00019999438573479312,
-      "loss": 46.0,
-      "step": 44126
-    },
-    {
-      "epoch": 3.37381730603819,
-      "grad_norm": 0.0014905557036399841,
-      "learning_rate": 0.00019999438548026977,
-      "loss": 46.0,
-      "step": 44127
-    },
-    {
-      "epoch": 3.37389376302158,
-      "grad_norm": 0.00216510146856308,
-      "learning_rate": 0.0001999943852257406,
-      "loss": 46.0,
-      "step": 44128
-    },
-    {
-      "epoch": 3.3739702200049697,
-      "grad_norm": 0.00393811846151948,
-      "learning_rate": 0.00019999438497120573,
-      "loss": 46.0,
-      "step": 44129
-    },
-    {
-      "epoch": 3.3740466769883595,
-      "grad_norm": 0.0035043267998844385,
-      "learning_rate": 0.00019999438471666507,
-      "loss": 46.0,
-      "step": 44130
-    },
-    {
-      "epoch": 3.3741231339717492,
-      "grad_norm": 0.0011496666120365262,
-      "learning_rate": 0.00019999438446211863,
-      "loss": 46.0,
-      "step": 44131
-    },
-    {
-      "epoch": 3.374199590955139,
-      "grad_norm": 0.0009796533267945051,
-      "learning_rate": 0.0001999943842075664,
-      "loss": 46.0,
-      "step": 44132
-    },
-    {
-      "epoch": 3.3742760479385288,
-      "grad_norm": 0.0016995292389765382,
-      "learning_rate": 0.00019999438395300844,
-      "loss": 46.0,
-      "step": 44133
-    },
-    {
-      "epoch": 3.3743525049219185,
-      "grad_norm": 0.000945060106460005,
-      "learning_rate": 0.00019999438369844468,
-      "loss": 46.0,
-      "step": 44134
-    },
-    {
-      "epoch": 3.374428961905308,
-      "grad_norm": 0.0027595667634159327,
-      "learning_rate": 0.00019999438344387517,
-      "loss": 46.0,
-      "step": 44135
-    },
-    {
-      "epoch": 3.3745054188886976,
-      "grad_norm": 0.007582082878798246,
-      "learning_rate": 0.0001999943831892999,
-      "loss": 46.0,
-      "step": 44136
-    },
-    {
-      "epoch": 3.3745818758720874,
-      "grad_norm": 0.0011815408943220973,
-      "learning_rate": 0.00019999438293471884,
-      "loss": 46.0,
-      "step": 44137
-    },
-    {
-      "epoch": 3.374658332855477,
-      "grad_norm": 0.005109219346195459,
-      "learning_rate": 0.00019999438268013202,
-      "loss": 46.0,
-      "step": 44138
-    },
-    {
-      "epoch": 3.374734789838867,
-      "grad_norm": 0.002627226524055004,
-      "learning_rate": 0.00019999438242553942,
-      "loss": 46.0,
-      "step": 44139
-    },
-    {
-      "epoch": 3.3748112468222566,
-      "grad_norm": 0.0033121739979833364,
-      "learning_rate": 0.00019999438217094108,
-      "loss": 46.0,
-      "step": 44140
-    },
-    {
-      "epoch": 3.3748877038056464,
-      "grad_norm": 0.0021402142010629177,
-      "learning_rate": 0.00019999438191633693,
-      "loss": 46.0,
-      "step": 44141
-    },
-    {
-      "epoch": 3.374964160789036,
-      "grad_norm": 0.001608529593795538,
-      "learning_rate": 0.00019999438166172704,
-      "loss": 46.0,
-      "step": 44142
-    },
-    {
-      "epoch": 3.375040617772426,
-      "grad_norm": 0.0064996955916285515,
-      "learning_rate": 0.00019999438140711138,
-      "loss": 46.0,
-      "step": 44143
-    },
-    {
-      "epoch": 3.3751170747558157,
-      "grad_norm": 0.0021500091534107924,
-      "learning_rate": 0.00019999438115248994,
-      "loss": 46.0,
-      "step": 44144
-    },
-    {
-      "epoch": 3.375193531739205,
-      "grad_norm": 0.0010710778879001737,
-      "learning_rate": 0.00019999438089786273,
-      "loss": 46.0,
-      "step": 44145
-    },
-    {
-      "epoch": 3.3752699887225948,
-      "grad_norm": 0.0018483932362869382,
-      "learning_rate": 0.00019999438064322974,
-      "loss": 46.0,
-      "step": 44146
-    },
-    {
-      "epoch": 3.3753464457059845,
-      "grad_norm": 0.004057146143168211,
-      "learning_rate": 0.000199994380388591,
-      "loss": 46.0,
-      "step": 44147
-    },
-    {
-      "epoch": 3.3754229026893743,
-      "grad_norm": 0.002004601527005434,
-      "learning_rate": 0.0001999943801339465,
-      "loss": 46.0,
-      "step": 44148
-    },
-    {
-      "epoch": 3.375499359672764,
-      "grad_norm": 0.0009705049451440573,
-      "learning_rate": 0.0001999943798792962,
-      "loss": 46.0,
-      "step": 44149
-    },
-    {
-      "epoch": 3.375575816656154,
-      "grad_norm": 0.0015370274195447564,
-      "learning_rate": 0.00019999437962464018,
-      "loss": 46.0,
-      "step": 44150
-    },
-    {
-      "epoch": 3.3756522736395436,
-      "grad_norm": 0.0008545694872736931,
-      "learning_rate": 0.00019999437936997833,
-      "loss": 46.0,
-      "step": 44151
-    },
-    {
-      "epoch": 3.3757287306229333,
-      "grad_norm": 0.002193163149058819,
-      "learning_rate": 0.00019999437911531076,
-      "loss": 46.0,
-      "step": 44152
-    },
-    {
-      "epoch": 3.375805187606323,
-      "grad_norm": 0.001179728889837861,
-      "learning_rate": 0.0001999943788606374,
-      "loss": 46.0,
-      "step": 44153
-    },
-    {
-      "epoch": 3.375881644589713,
-      "grad_norm": 0.000601144798565656,
-      "learning_rate": 0.0001999943786059583,
-      "loss": 46.0,
-      "step": 44154
-    },
-    {
-      "epoch": 3.3759581015731026,
-      "grad_norm": 0.0009339035023003817,
-      "learning_rate": 0.0001999943783512734,
-      "loss": 46.0,
-      "step": 44155
-    },
-    {
-      "epoch": 3.3760345585564924,
-      "grad_norm": 0.001779919839464128,
-      "learning_rate": 0.0001999943780965827,
-      "loss": 46.0,
-      "step": 44156
-    },
-    {
-      "epoch": 3.3761110155398817,
-      "grad_norm": 0.001878186478279531,
-      "learning_rate": 0.00019999437784188627,
-      "loss": 46.0,
-      "step": 44157
-    },
-    {
-      "epoch": 3.3761874725232714,
-      "grad_norm": 0.002082601422443986,
-      "learning_rate": 0.00019999437758718406,
-      "loss": 46.0,
-      "step": 44158
-    },
-    {
-      "epoch": 3.376263929506661,
-      "grad_norm": 0.0036716728936880827,
-      "learning_rate": 0.0001999943773324761,
-      "loss": 46.0,
-      "step": 44159
-    },
-    {
-      "epoch": 3.376340386490051,
-      "grad_norm": 0.0013276876416057348,
-      "learning_rate": 0.00019999437707776238,
-      "loss": 46.0,
-      "step": 44160
-    },
-    {
-      "epoch": 3.3764168434734407,
-      "grad_norm": 0.004441867116838694,
-      "learning_rate": 0.00019999437682304282,
-      "loss": 46.0,
-      "step": 44161
-    },
-    {
-      "epoch": 3.3764933004568305,
-      "grad_norm": 0.005524048116058111,
-      "learning_rate": 0.00019999437656831757,
-      "loss": 46.0,
-      "step": 44162
-    },
-    {
-      "epoch": 3.3765697574402203,
-      "grad_norm": 0.001608599559403956,
-      "learning_rate": 0.00019999437631358652,
-      "loss": 46.0,
-      "step": 44163
-    },
-    {
-      "epoch": 3.37664621442361,
-      "grad_norm": 0.001950397971086204,
-      "learning_rate": 0.0001999943760588497,
-      "loss": 46.0,
-      "step": 44164
-    },
-    {
-      "epoch": 3.3767226714069998,
-      "grad_norm": 0.002956976182758808,
-      "learning_rate": 0.0001999943758041071,
-      "loss": 46.0,
-      "step": 44165
-    },
-    {
-      "epoch": 3.3767991283903895,
-      "grad_norm": 0.0010052917059510946,
-      "learning_rate": 0.00019999437554935872,
-      "loss": 46.0,
-      "step": 44166
-    },
-    {
-      "epoch": 3.376875585373779,
-      "grad_norm": 0.000769517500884831,
-      "learning_rate": 0.0001999943752946046,
-      "loss": 46.0,
-      "step": 44167
-    },
-    {
-      "epoch": 3.3769520423571686,
-      "grad_norm": 0.002629816997796297,
-      "learning_rate": 0.00019999437503984472,
-      "loss": 46.0,
-      "step": 44168
-    },
-    {
-      "epoch": 3.3770284993405584,
-      "grad_norm": 0.003704741830006242,
-      "learning_rate": 0.00019999437478507908,
-      "loss": 46.0,
-      "step": 44169
-    },
-    {
-      "epoch": 3.377104956323948,
-      "grad_norm": 0.0032428184058517218,
-      "learning_rate": 0.00019999437453030762,
-      "loss": 46.0,
-      "step": 44170
-    },
-    {
-      "epoch": 3.377181413307338,
-      "grad_norm": 0.0021908662747591734,
-      "learning_rate": 0.00019999437427553043,
-      "loss": 46.0,
-      "step": 44171
-    },
-    {
-      "epoch": 3.3772578702907277,
-      "grad_norm": 0.001503440085798502,
-      "learning_rate": 0.00019999437402074745,
-      "loss": 46.0,
-      "step": 44172
-    },
-    {
-      "epoch": 3.3773343272741174,
-      "grad_norm": 0.004630373325198889,
-      "learning_rate": 0.00019999437376595872,
-      "loss": 46.0,
-      "step": 44173
-    },
-    {
-      "epoch": 3.377410784257507,
-      "grad_norm": 0.0032096903305500746,
-      "learning_rate": 0.0001999943735111642,
-      "loss": 46.0,
-      "step": 44174
-    },
-    {
-      "epoch": 3.377487241240897,
-      "grad_norm": 0.0016666464507579803,
-      "learning_rate": 0.0001999943732563639,
-      "loss": 46.0,
-      "step": 44175
-    },
-    {
-      "epoch": 3.3775636982242867,
-      "grad_norm": 0.0028358374256640673,
-      "learning_rate": 0.0001999943730015579,
-      "loss": 46.0,
-      "step": 44176
-    },
-    {
-      "epoch": 3.3776401552076765,
-      "grad_norm": 0.005152968689799309,
-      "learning_rate": 0.00019999437274674606,
-      "loss": 46.0,
-      "step": 44177
-    },
-    {
-      "epoch": 3.3777166121910662,
-      "grad_norm": 0.0027974932454526424,
-      "learning_rate": 0.00019999437249192847,
-      "loss": 46.0,
-      "step": 44178
-    },
-    {
-      "epoch": 3.3777930691744555,
-      "grad_norm": 0.003510811598971486,
-      "learning_rate": 0.00019999437223710512,
-      "loss": 46.0,
-      "step": 44179
-    },
-    {
-      "epoch": 3.3778695261578453,
-      "grad_norm": 0.006456709932535887,
-      "learning_rate": 0.00019999437198227598,
-      "loss": 46.0,
-      "step": 44180
-    },
-    {
-      "epoch": 3.377945983141235,
-      "grad_norm": 0.003602234646677971,
-      "learning_rate": 0.00019999437172744112,
-      "loss": 46.0,
-      "step": 44181
-    },
-    {
-      "epoch": 3.378022440124625,
-      "grad_norm": 0.0005513750948011875,
-      "learning_rate": 0.00019999437147260042,
-      "loss": 46.0,
-      "step": 44182
-    },
-    {
-      "epoch": 3.3780988971080146,
-      "grad_norm": 0.003053493332117796,
-      "learning_rate": 0.000199994371217754,
-      "loss": 46.0,
-      "step": 44183
-    },
-    {
-      "epoch": 3.3781753540914043,
-      "grad_norm": 0.0012695646146312356,
-      "learning_rate": 0.00019999437096290178,
-      "loss": 46.0,
-      "step": 44184
-    },
-    {
-      "epoch": 3.378251811074794,
-      "grad_norm": 0.0021072239615023136,
-      "learning_rate": 0.00019999437070804382,
-      "loss": 46.0,
-      "step": 44185
-    },
-    {
-      "epoch": 3.378328268058184,
-      "grad_norm": 0.002927567111328244,
-      "learning_rate": 0.0001999943704531801,
-      "loss": 46.0,
-      "step": 44186
-    },
-    {
-      "epoch": 3.3784047250415736,
-      "grad_norm": 0.0012319096131250262,
-      "learning_rate": 0.0001999943701983106,
-      "loss": 46.0,
-      "step": 44187
-    },
-    {
-      "epoch": 3.3784811820249634,
-      "grad_norm": 0.0037730620242655277,
-      "learning_rate": 0.0001999943699434353,
-      "loss": 46.0,
-      "step": 44188
-    },
-    {
-      "epoch": 3.3785576390083527,
-      "grad_norm": 0.0012067090719938278,
-      "learning_rate": 0.00019999436968855426,
-      "loss": 46.0,
-      "step": 44189
-    },
-    {
-      "epoch": 3.3786340959917425,
-      "grad_norm": 0.004190836101770401,
-      "learning_rate": 0.00019999436943366744,
-      "loss": 46.0,
-      "step": 44190
-    },
-    {
-      "epoch": 3.3787105529751322,
-      "grad_norm": 0.0013076069299131632,
-      "learning_rate": 0.00019999436917877484,
-      "loss": 46.0,
-      "step": 44191
-    },
-    {
-      "epoch": 3.378787009958522,
-      "grad_norm": 0.0023885949049144983,
-      "learning_rate": 0.0001999943689238765,
-      "loss": 46.0,
-      "step": 44192
-    },
-    {
-      "epoch": 3.3788634669419118,
-      "grad_norm": 0.0016930722631514072,
-      "learning_rate": 0.00019999436866897238,
-      "loss": 46.0,
-      "step": 44193
-    },
-    {
-      "epoch": 3.3789399239253015,
-      "grad_norm": 0.0024841127451509237,
-      "learning_rate": 0.00019999436841406246,
-      "loss": 46.0,
-      "step": 44194
-    },
-    {
-      "epoch": 3.3790163809086913,
-      "grad_norm": 0.003515142248943448,
-      "learning_rate": 0.00019999436815914683,
-      "loss": 46.0,
-      "step": 44195
-    },
-    {
-      "epoch": 3.379092837892081,
-      "grad_norm": 0.0025204226840287447,
-      "learning_rate": 0.0001999943679042254,
-      "loss": 46.0,
-      "step": 44196
-    },
-    {
-      "epoch": 3.379169294875471,
-      "grad_norm": 0.0032615887466818094,
-      "learning_rate": 0.00019999436764929818,
-      "loss": 46.0,
-      "step": 44197
-    },
-    {
-      "epoch": 3.3792457518588606,
-      "grad_norm": 0.0013024966465309262,
-      "learning_rate": 0.00019999436739436523,
-      "loss": 46.0,
-      "step": 44198
-    },
-    {
-      "epoch": 3.3793222088422503,
-      "grad_norm": 0.0023050878662616014,
-      "learning_rate": 0.00019999436713942647,
-      "loss": 46.0,
-      "step": 44199
-    },
-    {
-      "epoch": 3.37939866582564,
-      "grad_norm": 0.004119938239455223,
-      "learning_rate": 0.00019999436688448197,
-      "loss": 46.0,
-      "step": 44200
-    },
-    {
-      "epoch": 3.3794751228090294,
-      "grad_norm": 0.001688433694653213,
-      "learning_rate": 0.0001999943666295317,
-      "loss": 46.0,
-      "step": 44201
-    },
-    {
-      "epoch": 3.379551579792419,
-      "grad_norm": 0.0015831138007342815,
-      "learning_rate": 0.00019999436637457564,
-      "loss": 46.0,
-      "step": 44202
-    },
-    {
-      "epoch": 3.379628036775809,
-      "grad_norm": 0.0009583765640854836,
-      "learning_rate": 0.00019999436611961384,
-      "loss": 46.0,
-      "step": 44203
-    },
-    {
-      "epoch": 3.3797044937591987,
-      "grad_norm": 0.002457007300108671,
-      "learning_rate": 0.00019999436586464625,
-      "loss": 46.0,
-      "step": 44204
-    },
-    {
-      "epoch": 3.3797809507425884,
-      "grad_norm": 0.0030656771268695593,
-      "learning_rate": 0.0001999943656096729,
-      "loss": 46.0,
-      "step": 44205
-    },
-    {
-      "epoch": 3.379857407725978,
-      "grad_norm": 0.005699296947568655,
-      "learning_rate": 0.00019999436535469376,
-      "loss": 46.0,
-      "step": 44206
-    },
-    {
-      "epoch": 3.379933864709368,
-      "grad_norm": 0.001191108487546444,
-      "learning_rate": 0.00019999436509970887,
-      "loss": 46.0,
-      "step": 44207
-    },
-    {
-      "epoch": 3.3800103216927577,
-      "grad_norm": 0.002215739106759429,
-      "learning_rate": 0.00019999436484471824,
-      "loss": 46.0,
-      "step": 44208
-    },
-    {
-      "epoch": 3.3800867786761475,
-      "grad_norm": 0.0012486905325204134,
-      "learning_rate": 0.0001999943645897218,
-      "loss": 46.0,
-      "step": 44209
-    },
-    {
-      "epoch": 3.3801632356595372,
-      "grad_norm": 0.0028072006534785032,
-      "learning_rate": 0.00019999436433471962,
-      "loss": 46.0,
-      "step": 44210
-    },
-    {
-      "epoch": 3.3802396926429266,
-      "grad_norm": 0.0021883114241063595,
-      "learning_rate": 0.00019999436407971163,
-      "loss": 46.0,
-      "step": 44211
-    },
-    {
-      "epoch": 3.3803161496263163,
-      "grad_norm": 0.002390835201367736,
-      "learning_rate": 0.0001999943638246979,
-      "loss": 46.0,
-      "step": 44212
-    },
-    {
-      "epoch": 3.380392606609706,
-      "grad_norm": 0.0011012639151886106,
-      "learning_rate": 0.00019999436356967837,
-      "loss": 46.0,
-      "step": 44213
-    },
-    {
-      "epoch": 3.380469063593096,
-      "grad_norm": 0.002598078455775976,
-      "learning_rate": 0.0001999943633146531,
-      "loss": 46.0,
-      "step": 44214
-    },
-    {
-      "epoch": 3.3805455205764856,
-      "grad_norm": 0.0021298665087670088,
-      "learning_rate": 0.00019999436305962208,
-      "loss": 46.0,
-      "step": 44215
-    },
-    {
-      "epoch": 3.3806219775598754,
-      "grad_norm": 0.002586901420727372,
-      "learning_rate": 0.00019999436280458523,
-      "loss": 46.0,
-      "step": 44216
-    },
-    {
-      "epoch": 3.380698434543265,
-      "grad_norm": 0.0023862814996391535,
-      "learning_rate": 0.00019999436254954266,
-      "loss": 46.0,
-      "step": 44217
-    },
-    {
-      "epoch": 3.380774891526655,
-      "grad_norm": 0.0014274966670200229,
-      "learning_rate": 0.00019999436229449431,
-      "loss": 46.0,
-      "step": 44218
-    },
-    {
-      "epoch": 3.3808513485100447,
-      "grad_norm": 0.002062779851257801,
-      "learning_rate": 0.00019999436203944017,
-      "loss": 46.0,
-      "step": 44219
-    },
-    {
-      "epoch": 3.3809278054934344,
-      "grad_norm": 0.0011792710283771157,
-      "learning_rate": 0.00019999436178438028,
-      "loss": 46.0,
-      "step": 44220
-    },
-    {
-      "epoch": 3.381004262476824,
-      "grad_norm": 0.00436359690502286,
-      "learning_rate": 0.00019999436152931465,
-      "loss": 46.0,
-      "step": 44221
-    },
-    {
-      "epoch": 3.381080719460214,
-      "grad_norm": 0.006435814779251814,
-      "learning_rate": 0.0001999943612742432,
-      "loss": 46.0,
-      "step": 44222
-    },
-    {
-      "epoch": 3.3811571764436033,
-      "grad_norm": 0.002429185900837183,
-      "learning_rate": 0.00019999436101916603,
-      "loss": 46.0,
-      "step": 44223
-    },
-    {
-      "epoch": 3.381233633426993,
-      "grad_norm": 0.0011744301300495863,
-      "learning_rate": 0.00019999436076408305,
-      "loss": 46.0,
-      "step": 44224
-    },
-    {
-      "epoch": 3.3813100904103828,
-      "grad_norm": 0.006143974140286446,
-      "learning_rate": 0.00019999436050899432,
-      "loss": 46.0,
-      "step": 44225
-    },
-    {
-      "epoch": 3.3813865473937725,
-      "grad_norm": 0.002569023985415697,
-      "learning_rate": 0.0001999943602538998,
-      "loss": 46.0,
-      "step": 44226
-    },
-    {
-      "epoch": 3.3814630043771623,
-      "grad_norm": 0.003772683907300234,
-      "learning_rate": 0.00019999435999879954,
-      "loss": 46.0,
-      "step": 44227
-    },
-    {
-      "epoch": 3.381539461360552,
-      "grad_norm": 0.0010371258249506354,
-      "learning_rate": 0.0001999943597436935,
-      "loss": 46.0,
-      "step": 44228
-    },
-    {
-      "epoch": 3.381615918343942,
-      "grad_norm": 0.002710413420572877,
-      "learning_rate": 0.00019999435948858167,
-      "loss": 46.0,
-      "step": 44229
-    },
-    {
-      "epoch": 3.3816923753273316,
-      "grad_norm": 0.0031064720824360847,
-      "learning_rate": 0.0001999943592334641,
-      "loss": 46.0,
-      "step": 44230
-    },
-    {
-      "epoch": 3.3817688323107213,
-      "grad_norm": 0.0036101602017879486,
-      "learning_rate": 0.00019999435897834076,
-      "loss": 46.0,
-      "step": 44231
-    },
-    {
-      "epoch": 3.3818452892941107,
-      "grad_norm": 0.0022557475604116917,
-      "learning_rate": 0.00019999435872321164,
-      "loss": 46.0,
-      "step": 44232
-    },
-    {
-      "epoch": 3.3819217462775004,
-      "grad_norm": 0.0021435606759041548,
-      "learning_rate": 0.00019999435846807675,
-      "loss": 46.0,
-      "step": 44233
-    },
-    {
-      "epoch": 3.38199820326089,
-      "grad_norm": 0.00417352793738246,
-      "learning_rate": 0.0001999943582129361,
-      "loss": 46.0,
-      "step": 44234
-    },
-    {
-      "epoch": 3.38207466024428,
-      "grad_norm": 0.00229679886251688,
-      "learning_rate": 0.00019999435795778965,
-      "loss": 46.0,
-      "step": 44235
-    },
-    {
-      "epoch": 3.3821511172276697,
-      "grad_norm": 0.0024019451811909676,
-      "learning_rate": 0.00019999435770263744,
-      "loss": 46.0,
-      "step": 44236
-    },
-    {
-      "epoch": 3.3822275742110595,
-      "grad_norm": 0.004723278805613518,
-      "learning_rate": 0.0001999943574474795,
-      "loss": 46.0,
-      "step": 44237
-    },
-    {
-      "epoch": 3.3823040311944492,
-      "grad_norm": 0.0025893684942275286,
-      "learning_rate": 0.00019999435719231573,
-      "loss": 46.0,
-      "step": 44238
-    },
-    {
-      "epoch": 3.382380488177839,
-      "grad_norm": 0.0018986817449331284,
-      "learning_rate": 0.00019999435693714626,
-      "loss": 46.0,
-      "step": 44239
-    },
-    {
-      "epoch": 3.3824569451612287,
-      "grad_norm": 0.0012239236384630203,
-      "learning_rate": 0.00019999435668197098,
-      "loss": 46.0,
-      "step": 44240
-    },
-    {
-      "epoch": 3.3825334021446185,
-      "grad_norm": 0.005004290025681257,
-      "learning_rate": 0.00019999435642678993,
-      "loss": 46.0,
-      "step": 44241
-    },
-    {
-      "epoch": 3.3826098591280083,
-      "grad_norm": 0.0021270280703902245,
-      "learning_rate": 0.0001999943561716031,
-      "loss": 46.0,
-      "step": 44242
-    },
-    {
-      "epoch": 3.382686316111398,
-      "grad_norm": 0.004737840034067631,
-      "learning_rate": 0.00019999435591641054,
-      "loss": 46.0,
-      "step": 44243
-    },
-    {
-      "epoch": 3.382762773094788,
-      "grad_norm": 0.001975037157535553,
-      "learning_rate": 0.00019999435566121217,
-      "loss": 46.0,
-      "step": 44244
-    },
-    {
-      "epoch": 3.382839230078177,
-      "grad_norm": 0.0011048391461372375,
-      "learning_rate": 0.0001999943554060081,
-      "loss": 46.0,
-      "step": 44245
-    },
-    {
-      "epoch": 3.382915687061567,
-      "grad_norm": 0.003061659401282668,
-      "learning_rate": 0.0001999943551507982,
-      "loss": 46.0,
-      "step": 44246
-    },
-    {
-      "epoch": 3.3829921440449566,
-      "grad_norm": 0.0010180168319493532,
-      "learning_rate": 0.00019999435489558254,
-      "loss": 46.0,
-      "step": 44247
-    },
-    {
-      "epoch": 3.3830686010283464,
-      "grad_norm": 0.006625974550843239,
-      "learning_rate": 0.0001999943546403611,
-      "loss": 46.0,
-      "step": 44248
-    },
-    {
-      "epoch": 3.383145058011736,
-      "grad_norm": 0.0020825297106057405,
-      "learning_rate": 0.00019999435438513392,
-      "loss": 46.0,
-      "step": 44249
-    },
-    {
-      "epoch": 3.383221514995126,
-      "grad_norm": 0.0030929071363061666,
-      "learning_rate": 0.00019999435412990096,
-      "loss": 46.0,
-      "step": 44250
-    },
-    {
-      "epoch": 3.3832979719785157,
-      "grad_norm": 0.0035906864795833826,
-      "learning_rate": 0.0001999943538746622,
-      "loss": 46.0,
-      "step": 44251
-    },
-    {
-      "epoch": 3.3833744289619054,
-      "grad_norm": 0.009797638282179832,
-      "learning_rate": 0.0001999943536194177,
-      "loss": 46.0,
-      "step": 44252
-    },
-    {
-      "epoch": 3.383450885945295,
-      "grad_norm": 0.0024939454160630703,
-      "learning_rate": 0.00019999435336416746,
-      "loss": 46.0,
-      "step": 44253
-    },
-    {
-      "epoch": 3.3835273429286845,
-      "grad_norm": 0.0010261759161949158,
-      "learning_rate": 0.0001999943531089114,
-      "loss": 46.0,
-      "step": 44254
-    },
-    {
-      "epoch": 3.3836037999120743,
-      "grad_norm": 0.0013130229199305177,
-      "learning_rate": 0.00019999435285364962,
-      "loss": 46.0,
-      "step": 44255
-    },
-    {
-      "epoch": 3.383680256895464,
-      "grad_norm": 0.003383307484909892,
-      "learning_rate": 0.00019999435259838202,
-      "loss": 46.0,
-      "step": 44256
-    },
-    {
-      "epoch": 3.383756713878854,
-      "grad_norm": 0.0030635856091976166,
-      "learning_rate": 0.00019999435234310865,
-      "loss": 46.0,
-      "step": 44257
-    },
-    {
-      "epoch": 3.3838331708622436,
-      "grad_norm": 0.0037393427919596434,
-      "learning_rate": 0.00019999435208782956,
-      "loss": 46.0,
-      "step": 44258
-    },
-    {
-      "epoch": 3.3839096278456333,
-      "grad_norm": 0.0010026482632383704,
-      "learning_rate": 0.00019999435183254468,
-      "loss": 46.0,
-      "step": 44259
-    },
-    {
-      "epoch": 3.383986084829023,
-      "grad_norm": 0.0016988830175250769,
-      "learning_rate": 0.000199994351577254,
-      "loss": 46.0,
-      "step": 44260
-    },
-    {
-      "epoch": 3.384062541812413,
-      "grad_norm": 0.0005278535536490381,
-      "learning_rate": 0.00019999435132195758,
-      "loss": 46.0,
-      "step": 44261
-    },
-    {
-      "epoch": 3.3841389987958026,
-      "grad_norm": 0.0025562585797160864,
-      "learning_rate": 0.00019999435106665537,
-      "loss": 46.0,
-      "step": 44262
-    },
-    {
-      "epoch": 3.3842154557791924,
-      "grad_norm": 0.0008684746571816504,
-      "learning_rate": 0.00019999435081134744,
-      "loss": 46.0,
-      "step": 44263
-    },
-    {
-      "epoch": 3.384291912762582,
-      "grad_norm": 0.0017802761867642403,
-      "learning_rate": 0.0001999943505560337,
-      "loss": 46.0,
-      "step": 44264
-    },
-    {
-      "epoch": 3.384368369745972,
-      "grad_norm": 0.00244636251591146,
-      "learning_rate": 0.00019999435030071419,
-      "loss": 46.0,
-      "step": 44265
-    },
-    {
-      "epoch": 3.384444826729361,
-      "grad_norm": 0.0008984652813524008,
-      "learning_rate": 0.0001999943500453889,
-      "loss": 46.0,
-      "step": 44266
-    },
-    {
-      "epoch": 3.384521283712751,
-      "grad_norm": 0.0004849621909670532,
-      "learning_rate": 0.0001999943497900579,
-      "loss": 46.0,
-      "step": 44267
-    },
-    {
-      "epoch": 3.3845977406961407,
-      "grad_norm": 0.002147234743461013,
-      "learning_rate": 0.00019999434953472107,
-      "loss": 46.0,
-      "step": 44268
-    },
-    {
-      "epoch": 3.3846741976795305,
-      "grad_norm": 0.001699982094578445,
-      "learning_rate": 0.0001999943492793785,
-      "loss": 46.0,
-      "step": 44269
-    },
-    {
-      "epoch": 3.3847506546629202,
-      "grad_norm": 0.0008087189635261893,
-      "learning_rate": 0.00019999434902403013,
-      "loss": 46.0,
-      "step": 44270
-    },
-    {
-      "epoch": 3.38482711164631,
-      "grad_norm": 0.00853345263749361,
-      "learning_rate": 0.00019999434876867605,
-      "loss": 46.0,
-      "step": 44271
-    },
-    {
-      "epoch": 3.3849035686296998,
-      "grad_norm": 0.001939923269674182,
-      "learning_rate": 0.00019999434851331616,
-      "loss": 46.0,
-      "step": 44272
-    },
-    {
-      "epoch": 3.3849800256130895,
-      "grad_norm": 0.003089835634455085,
-      "learning_rate": 0.0001999943482579505,
-      "loss": 46.0,
-      "step": 44273
-    },
-    {
-      "epoch": 3.3850564825964793,
-      "grad_norm": 0.001009583007544279,
-      "learning_rate": 0.0001999943480025791,
-      "loss": 46.0,
-      "step": 44274
-    },
-    {
-      "epoch": 3.385132939579869,
-      "grad_norm": 0.0041471137665212154,
-      "learning_rate": 0.00019999434774720188,
-      "loss": 46.0,
-      "step": 44275
-    },
-    {
-      "epoch": 3.3852093965632584,
-      "grad_norm": 0.0013122382806614041,
-      "learning_rate": 0.00019999434749181893,
-      "loss": 46.0,
-      "step": 44276
-    },
-    {
-      "epoch": 3.385285853546648,
-      "grad_norm": 0.0060357991605997086,
-      "learning_rate": 0.0001999943472364302,
-      "loss": 46.0,
-      "step": 44277
-    },
-    {
-      "epoch": 3.385362310530038,
-      "grad_norm": 0.003791297785937786,
-      "learning_rate": 0.0001999943469810357,
-      "loss": 46.0,
-      "step": 44278
-    },
-    {
-      "epoch": 3.3854387675134276,
-      "grad_norm": 0.00183978455606848,
-      "learning_rate": 0.00019999434672563542,
-      "loss": 46.0,
-      "step": 44279
-    },
-    {
-      "epoch": 3.3855152244968174,
-      "grad_norm": 0.0014265599893406034,
-      "learning_rate": 0.00019999434647022937,
-      "loss": 46.0,
-      "step": 44280
-    },
-    {
-      "epoch": 3.385591681480207,
-      "grad_norm": 0.0011291619157418609,
-      "learning_rate": 0.00019999434621481758,
-      "loss": 46.0,
-      "step": 44281
-    },
-    {
-      "epoch": 3.385668138463597,
-      "grad_norm": 0.0006563774077221751,
-      "learning_rate": 0.00019999434595939999,
-      "loss": 46.0,
-      "step": 44282
-    },
-    {
-      "epoch": 3.3857445954469867,
-      "grad_norm": 0.004237682092934847,
-      "learning_rate": 0.00019999434570397667,
-      "loss": 46.0,
-      "step": 44283
-    },
-    {
-      "epoch": 3.3858210524303765,
-      "grad_norm": 0.003301443299278617,
-      "learning_rate": 0.00019999434544854753,
-      "loss": 46.0,
-      "step": 44284
-    },
-    {
-      "epoch": 3.385897509413766,
-      "grad_norm": 0.004162689205259085,
-      "learning_rate": 0.00019999434519311264,
-      "loss": 46.0,
-      "step": 44285
-    },
-    {
-      "epoch": 3.385973966397156,
-      "grad_norm": 0.0009109798120334744,
-      "learning_rate": 0.000199994344937672,
-      "loss": 46.0,
-      "step": 44286
-    },
-    {
-      "epoch": 3.3860504233805457,
-      "grad_norm": 0.0016060040798038244,
-      "learning_rate": 0.00019999434468222557,
-      "loss": 46.0,
-      "step": 44287
-    },
-    {
-      "epoch": 3.386126880363935,
-      "grad_norm": 0.007468229625374079,
-      "learning_rate": 0.0001999943444267734,
-      "loss": 46.0,
-      "step": 44288
-    },
-    {
-      "epoch": 3.386203337347325,
-      "grad_norm": 0.0007647254387848079,
-      "learning_rate": 0.0001999943441713154,
-      "loss": 46.0,
-      "step": 44289
-    },
-    {
-      "epoch": 3.3862797943307146,
-      "grad_norm": 0.005335691850632429,
-      "learning_rate": 0.0001999943439158517,
-      "loss": 46.0,
-      "step": 44290
-    },
-    {
-      "epoch": 3.3863562513141043,
-      "grad_norm": 0.0016116950428113341,
-      "learning_rate": 0.0001999943436603822,
-      "loss": 46.0,
-      "step": 44291
-    },
-    {
-      "epoch": 3.386432708297494,
-      "grad_norm": 0.002544571179896593,
-      "learning_rate": 0.00019999434340490694,
-      "loss": 46.0,
-      "step": 44292
-    },
-    {
-      "epoch": 3.386509165280884,
-      "grad_norm": 0.004215978551656008,
-      "learning_rate": 0.00019999434314942592,
-      "loss": 46.0,
-      "step": 44293
-    },
-    {
-      "epoch": 3.3865856222642736,
-      "grad_norm": 0.0010518316412344575,
-      "learning_rate": 0.00019999434289393907,
-      "loss": 46.0,
-      "step": 44294
-    },
-    {
-      "epoch": 3.3866620792476634,
-      "grad_norm": 0.0025761134456843138,
-      "learning_rate": 0.00019999434263844653,
-      "loss": 46.0,
-      "step": 44295
-    },
-    {
-      "epoch": 3.386738536231053,
-      "grad_norm": 0.0015401405980810523,
-      "learning_rate": 0.0001999943423829482,
-      "loss": 46.0,
-      "step": 44296
-    },
-    {
-      "epoch": 3.386814993214443,
-      "grad_norm": 0.001467181951738894,
-      "learning_rate": 0.00019999434212744405,
-      "loss": 46.0,
-      "step": 44297
-    },
-    {
-      "epoch": 3.386891450197832,
-      "grad_norm": 0.001316542155109346,
-      "learning_rate": 0.0001999943418719342,
-      "loss": 46.0,
-      "step": 44298
-    },
-    {
-      "epoch": 3.386967907181222,
-      "grad_norm": 0.002022126456722617,
-      "learning_rate": 0.00019999434161641853,
-      "loss": 46.0,
-      "step": 44299
-    },
-    {
-      "epoch": 3.3870443641646117,
-      "grad_norm": 0.002226028358563781,
-      "learning_rate": 0.00019999434136089712,
-      "loss": 46.0,
-      "step": 44300
-    },
-    {
-      "epoch": 3.3871208211480015,
-      "grad_norm": 0.0013663721038028598,
-      "learning_rate": 0.0001999943411053699,
-      "loss": 46.0,
-      "step": 44301
-    },
-    {
-      "epoch": 3.3871972781313913,
-      "grad_norm": 0.0016099871136248112,
-      "learning_rate": 0.00019999434084983696,
-      "loss": 46.0,
-      "step": 44302
-    },
-    {
-      "epoch": 3.387273735114781,
-      "grad_norm": 0.0021600699983537197,
-      "learning_rate": 0.00019999434059429826,
-      "loss": 46.0,
-      "step": 44303
-    },
-    {
-      "epoch": 3.387350192098171,
-      "grad_norm": 0.0010704408632591367,
-      "learning_rate": 0.00019999434033875373,
-      "loss": 46.0,
-      "step": 44304
-    },
-    {
-      "epoch": 3.3874266490815605,
-      "grad_norm": 0.0044770753011107445,
-      "learning_rate": 0.00019999434008320348,
-      "loss": 46.0,
-      "step": 44305
-    },
-    {
-      "epoch": 3.3875031060649503,
-      "grad_norm": 0.006041297223418951,
-      "learning_rate": 0.00019999433982764744,
-      "loss": 46.0,
-      "step": 44306
-    },
-    {
-      "epoch": 3.38757956304834,
-      "grad_norm": 0.0027178835589438677,
-      "learning_rate": 0.00019999433957208562,
-      "loss": 46.0,
-      "step": 44307
-    },
-    {
-      "epoch": 3.38765602003173,
-      "grad_norm": 0.007367952261120081,
-      "learning_rate": 0.00019999433931651805,
-      "loss": 46.0,
-      "step": 44308
-    },
-    {
-      "epoch": 3.3877324770151196,
-      "grad_norm": 0.003046803642064333,
-      "learning_rate": 0.0001999943390609447,
-      "loss": 46.0,
-      "step": 44309
-    },
-    {
-      "epoch": 3.387808933998509,
-      "grad_norm": 0.005679132416844368,
-      "learning_rate": 0.0001999943388053656,
-      "loss": 46.0,
-      "step": 44310
-    },
-    {
-      "epoch": 3.3878853909818987,
-      "grad_norm": 0.0029210590291768312,
-      "learning_rate": 0.0001999943385497807,
-      "loss": 46.0,
-      "step": 44311
-    },
-    {
-      "epoch": 3.3879618479652884,
-      "grad_norm": 0.000879273284226656,
-      "learning_rate": 0.00019999433829419005,
-      "loss": 46.0,
-      "step": 44312
-    },
-    {
-      "epoch": 3.388038304948678,
-      "grad_norm": 0.0011485553113743663,
-      "learning_rate": 0.00019999433803859367,
-      "loss": 46.0,
-      "step": 44313
-    },
-    {
-      "epoch": 3.388114761932068,
-      "grad_norm": 0.005128402728587389,
-      "learning_rate": 0.00019999433778299146,
-      "loss": 46.0,
-      "step": 44314
-    },
-    {
-      "epoch": 3.3881912189154577,
-      "grad_norm": 0.002910564886406064,
-      "learning_rate": 0.0001999943375273835,
-      "loss": 46.0,
-      "step": 44315
-    },
-    {
-      "epoch": 3.3882676758988475,
-      "grad_norm": 0.0020676194690167904,
-      "learning_rate": 0.00019999433727176978,
-      "loss": 46.0,
-      "step": 44316
-    },
-    {
-      "epoch": 3.3883441328822372,
-      "grad_norm": 0.002800045534968376,
-      "learning_rate": 0.00019999433701615028,
-      "loss": 46.0,
-      "step": 44317
-    },
-    {
-      "epoch": 3.388420589865627,
-      "grad_norm": 0.0016800720477476716,
-      "learning_rate": 0.000199994336760525,
-      "loss": 46.0,
-      "step": 44318
-    },
-    {
-      "epoch": 3.3884970468490168,
-      "grad_norm": 0.001002141973003745,
-      "learning_rate": 0.000199994336504894,
-      "loss": 46.0,
-      "step": 44319
-    },
-    {
-      "epoch": 3.388573503832406,
-      "grad_norm": 0.0021147807128727436,
-      "learning_rate": 0.0001999943362492572,
-      "loss": 46.0,
-      "step": 44320
-    },
-    {
-      "epoch": 3.388649960815796,
-      "grad_norm": 0.0007403853232972324,
-      "learning_rate": 0.0001999943359936146,
-      "loss": 46.0,
-      "step": 44321
-    },
-    {
-      "epoch": 3.3887264177991856,
-      "grad_norm": 0.0015886673936620355,
-      "learning_rate": 0.00019999433573796626,
-      "loss": 46.0,
-      "step": 44322
-    },
-    {
-      "epoch": 3.3888028747825754,
-      "grad_norm": 0.0018542190082371235,
-      "learning_rate": 0.00019999433548231217,
-      "loss": 46.0,
-      "step": 44323
-    },
-    {
-      "epoch": 3.388879331765965,
-      "grad_norm": 0.0017437656642869115,
-      "learning_rate": 0.0001999943352266523,
-      "loss": 46.0,
-      "step": 44324
-    },
-    {
-      "epoch": 3.388955788749355,
-      "grad_norm": 0.0026636819820851088,
-      "learning_rate": 0.00019999433497098663,
-      "loss": 46.0,
-      "step": 44325
-    },
-    {
-      "epoch": 3.3890322457327446,
-      "grad_norm": 0.0029814038425683975,
-      "learning_rate": 0.00019999433471531522,
-      "loss": 46.0,
-      "step": 44326
-    },
-    {
-      "epoch": 3.3891087027161344,
-      "grad_norm": 0.0010743628954514861,
-      "learning_rate": 0.00019999433445963801,
-      "loss": 46.0,
-      "step": 44327
-    },
-    {
-      "epoch": 3.389185159699524,
-      "grad_norm": 0.0014858456561341882,
-      "learning_rate": 0.00019999433420395506,
-      "loss": 46.0,
-      "step": 44328
-    },
-    {
-      "epoch": 3.389261616682914,
-      "grad_norm": 0.0015158449532464147,
-      "learning_rate": 0.00019999433394826633,
-      "loss": 46.0,
-      "step": 44329
-    },
-    {
-      "epoch": 3.3893380736663037,
-      "grad_norm": 0.0037937683518975973,
-      "learning_rate": 0.00019999433369257186,
-      "loss": 46.0,
-      "step": 44330
-    },
-    {
-      "epoch": 3.3894145306496934,
-      "grad_norm": 0.003278592135757208,
-      "learning_rate": 0.0001999943334368716,
-      "loss": 46.0,
-      "step": 44331
-    },
-    {
-      "epoch": 3.3894909876330828,
-      "grad_norm": 0.004604941699653864,
-      "learning_rate": 0.00019999433318116557,
-      "loss": 46.0,
-      "step": 44332
-    },
-    {
-      "epoch": 3.3895674446164725,
-      "grad_norm": 0.0012613818980753422,
-      "learning_rate": 0.00019999433292545375,
-      "loss": 46.0,
-      "step": 44333
-    },
-    {
-      "epoch": 3.3896439015998623,
-      "grad_norm": 0.0011180745204910636,
-      "learning_rate": 0.00019999433266973618,
-      "loss": 46.0,
-      "step": 44334
-    },
-    {
-      "epoch": 3.389720358583252,
-      "grad_norm": 0.003944214899092913,
-      "learning_rate": 0.00019999433241401284,
-      "loss": 46.0,
-      "step": 44335
-    },
-    {
-      "epoch": 3.389796815566642,
-      "grad_norm": 0.005096031352877617,
-      "learning_rate": 0.00019999433215828373,
-      "loss": 46.0,
-      "step": 44336
-    },
-    {
-      "epoch": 3.3898732725500316,
-      "grad_norm": 0.0037811503279954195,
-      "learning_rate": 0.00019999433190254884,
-      "loss": 46.0,
-      "step": 44337
-    },
-    {
-      "epoch": 3.3899497295334213,
-      "grad_norm": 0.0013853403506800532,
-      "learning_rate": 0.0001999943316468082,
-      "loss": 46.0,
-      "step": 44338
-    },
-    {
-      "epoch": 3.390026186516811,
-      "grad_norm": 0.0015070205554366112,
-      "learning_rate": 0.00019999433139106178,
-      "loss": 46.0,
-      "step": 44339
-    },
-    {
-      "epoch": 3.390102643500201,
-      "grad_norm": 0.003356938948854804,
-      "learning_rate": 0.0001999943311353096,
-      "loss": 46.0,
-      "step": 44340
-    },
-    {
-      "epoch": 3.3901791004835906,
-      "grad_norm": 0.0009637877228669822,
-      "learning_rate": 0.00019999433087955165,
-      "loss": 46.0,
-      "step": 44341
-    },
-    {
-      "epoch": 3.39025555746698,
-      "grad_norm": 0.0015435336390510201,
-      "learning_rate": 0.00019999433062378792,
-      "loss": 46.0,
-      "step": 44342
-    },
-    {
-      "epoch": 3.3903320144503697,
-      "grad_norm": 0.0023447140119969845,
-      "learning_rate": 0.00019999433036801842,
-      "loss": 46.0,
-      "step": 44343
-    },
-    {
-      "epoch": 3.3904084714337595,
-      "grad_norm": 0.001448609633371234,
-      "learning_rate": 0.00019999433011224315,
-      "loss": 46.0,
-      "step": 44344
-    },
-    {
-      "epoch": 3.390484928417149,
-      "grad_norm": 0.0017516553634777665,
-      "learning_rate": 0.00019999432985646213,
-      "loss": 46.0,
-      "step": 44345
-    },
-    {
-      "epoch": 3.390561385400539,
-      "grad_norm": 0.0009320615790784359,
-      "learning_rate": 0.00019999432960067534,
-      "loss": 46.0,
-      "step": 44346
-    },
-    {
-      "epoch": 3.3906378423839287,
-      "grad_norm": 0.0032951468601822853,
-      "learning_rate": 0.00019999432934488277,
-      "loss": 46.0,
-      "step": 44347
-    },
-    {
-      "epoch": 3.3907142993673185,
-      "grad_norm": 0.0011953997891396284,
-      "learning_rate": 0.00019999432908908443,
-      "loss": 46.0,
-      "step": 44348
-    },
-    {
-      "epoch": 3.3907907563507083,
-      "grad_norm": 0.0036905519664287567,
-      "learning_rate": 0.00019999432883328032,
-      "loss": 46.0,
-      "step": 44349
-    },
-    {
-      "epoch": 3.390867213334098,
-      "grad_norm": 0.0036113422829657793,
-      "learning_rate": 0.00019999432857747046,
-      "loss": 46.0,
-      "step": 44350
-    },
-    {
-      "epoch": 3.390943670317488,
-      "grad_norm": 0.0030079083517193794,
-      "learning_rate": 0.0001999943283216548,
-      "loss": 46.0,
-      "step": 44351
-    },
-    {
-      "epoch": 3.3910201273008775,
-      "grad_norm": 0.001848881016485393,
-      "learning_rate": 0.0001999943280658334,
-      "loss": 46.0,
-      "step": 44352
-    },
-    {
-      "epoch": 3.3910965842842673,
-      "grad_norm": 0.0025121138896793127,
-      "learning_rate": 0.0001999943278100062,
-      "loss": 46.0,
-      "step": 44353
-    },
-    {
-      "epoch": 3.3911730412676566,
-      "grad_norm": 0.004290048498660326,
-      "learning_rate": 0.00019999432755417323,
-      "loss": 46.0,
-      "step": 44354
-    },
-    {
-      "epoch": 3.3912494982510464,
-      "grad_norm": 0.004328650422394276,
-      "learning_rate": 0.00019999432729833454,
-      "loss": 46.0,
-      "step": 44355
-    },
-    {
-      "epoch": 3.391325955234436,
-      "grad_norm": 0.0031624711118638515,
-      "learning_rate": 0.00019999432704249004,
-      "loss": 46.0,
-      "step": 44356
-    },
-    {
-      "epoch": 3.391402412217826,
-      "grad_norm": 0.0014169290661811829,
-      "learning_rate": 0.00019999432678663976,
-      "loss": 46.0,
-      "step": 44357
-    },
-    {
-      "epoch": 3.3914788692012157,
-      "grad_norm": 0.0011730794794857502,
-      "learning_rate": 0.00019999432653078375,
-      "loss": 46.0,
-      "step": 44358
-    },
-    {
-      "epoch": 3.3915553261846054,
-      "grad_norm": 0.0013420027680695057,
-      "learning_rate": 0.00019999432627492193,
-      "loss": 46.0,
-      "step": 44359
-    },
-    {
-      "epoch": 3.391631783167995,
-      "grad_norm": 0.0013395623536780477,
-      "learning_rate": 0.0001999943260190544,
-      "loss": 46.0,
-      "step": 44360
-    },
-    {
-      "epoch": 3.391708240151385,
-      "grad_norm": 0.0031876848079264164,
-      "learning_rate": 0.00019999432576318102,
-      "loss": 46.0,
-      "step": 44361
-    },
-    {
-      "epoch": 3.3917846971347747,
-      "grad_norm": 0.002493435051292181,
-      "learning_rate": 0.00019999432550730194,
-      "loss": 46.0,
-      "step": 44362
-    },
-    {
-      "epoch": 3.391861154118164,
-      "grad_norm": 0.0011749910190701485,
-      "learning_rate": 0.00019999432525141705,
-      "loss": 46.0,
-      "step": 44363
-    },
-    {
-      "epoch": 3.391937611101554,
-      "grad_norm": 0.0023854251485317945,
-      "learning_rate": 0.00019999432499552642,
-      "loss": 46.0,
-      "step": 44364
-    },
-    {
-      "epoch": 3.3920140680849435,
-      "grad_norm": 0.0014242819743230939,
-      "learning_rate": 0.00019999432473963,
-      "loss": 46.0,
-      "step": 44365
-    },
-    {
-      "epoch": 3.3920905250683333,
-      "grad_norm": 0.004025659523904324,
-      "learning_rate": 0.0001999943244837278,
-      "loss": 46.0,
-      "step": 44366
-    },
-    {
-      "epoch": 3.392166982051723,
-      "grad_norm": 0.003747211769223213,
-      "learning_rate": 0.00019999432422781985,
-      "loss": 46.0,
-      "step": 44367
-    },
-    {
-      "epoch": 3.392243439035113,
-      "grad_norm": 0.003146359696984291,
-      "learning_rate": 0.00019999432397190613,
-      "loss": 46.0,
-      "step": 44368
-    },
-    {
-      "epoch": 3.3923198960185026,
-      "grad_norm": 0.0016721318243071437,
-      "learning_rate": 0.00019999432371598666,
-      "loss": 46.0,
-      "step": 44369
-    },
-    {
-      "epoch": 3.3923963530018924,
-      "grad_norm": 0.0021338535007089376,
-      "learning_rate": 0.00019999432346006138,
-      "loss": 46.0,
-      "step": 44370
-    },
-    {
-      "epoch": 3.392472809985282,
-      "grad_norm": 0.002418762305751443,
-      "learning_rate": 0.00019999432320413037,
-      "loss": 46.0,
-      "step": 44371
-    },
-    {
-      "epoch": 3.392549266968672,
-      "grad_norm": 0.0012527513317763805,
-      "learning_rate": 0.00019999432294819357,
-      "loss": 46.0,
-      "step": 44372
-    },
-    {
-      "epoch": 3.3926257239520616,
-      "grad_norm": 0.0023592961952090263,
-      "learning_rate": 0.000199994322692251,
-      "loss": 46.0,
-      "step": 44373
-    },
-    {
-      "epoch": 3.3927021809354514,
-      "grad_norm": 0.003370742779225111,
-      "learning_rate": 0.00019999432243630264,
-      "loss": 46.0,
-      "step": 44374
-    },
-    {
-      "epoch": 3.392778637918841,
-      "grad_norm": 0.002068512374535203,
-      "learning_rate": 0.00019999432218034856,
-      "loss": 46.0,
-      "step": 44375
-    },
-    {
-      "epoch": 3.3928550949022305,
-      "grad_norm": 0.002953542862087488,
-      "learning_rate": 0.00019999432192438867,
-      "loss": 46.0,
-      "step": 44376
-    },
-    {
-      "epoch": 3.3929315518856202,
-      "grad_norm": 0.0018068733625113964,
-      "learning_rate": 0.00019999432166842304,
-      "loss": 46.0,
-      "step": 44377
-    },
-    {
-      "epoch": 3.39300800886901,
-      "grad_norm": 0.002879250328987837,
-      "learning_rate": 0.00019999432141245164,
-      "loss": 46.0,
-      "step": 44378
-    },
-    {
-      "epoch": 3.3930844658523998,
-      "grad_norm": 0.0008370761061087251,
-      "learning_rate": 0.00019999432115647446,
-      "loss": 46.0,
-      "step": 44379
-    },
-    {
-      "epoch": 3.3931609228357895,
-      "grad_norm": 0.0015681401127949357,
-      "learning_rate": 0.00019999432090049148,
-      "loss": 46.0,
-      "step": 44380
-    },
-    {
-      "epoch": 3.3932373798191793,
-      "grad_norm": 0.001608354039490223,
-      "learning_rate": 0.00019999432064450278,
-      "loss": 46.0,
-      "step": 44381
-    },
-    {
-      "epoch": 3.393313836802569,
-      "grad_norm": 0.0021239486522972584,
-      "learning_rate": 0.00019999432038850828,
-      "loss": 46.0,
-      "step": 44382
-    },
-    {
-      "epoch": 3.393390293785959,
-      "grad_norm": 0.0019906209781765938,
-      "learning_rate": 0.00019999432013250804,
-      "loss": 46.0,
-      "step": 44383
-    },
-    {
-      "epoch": 3.3934667507693486,
-      "grad_norm": 0.0009511591633781791,
-      "learning_rate": 0.00019999431987650202,
-      "loss": 46.0,
-      "step": 44384
-    },
-    {
-      "epoch": 3.393543207752738,
-      "grad_norm": 0.0013309771893545985,
-      "learning_rate": 0.0001999943196204902,
-      "loss": 46.0,
-      "step": 44385
-    },
-    {
-      "epoch": 3.3936196647361276,
-      "grad_norm": 0.0034242768306285143,
-      "learning_rate": 0.00019999431936447264,
-      "loss": 46.0,
-      "step": 44386
-    },
-    {
-      "epoch": 3.3936961217195174,
-      "grad_norm": 0.0021421003621071577,
-      "learning_rate": 0.0001999943191084493,
-      "loss": 46.0,
-      "step": 44387
-    },
-    {
-      "epoch": 3.393772578702907,
-      "grad_norm": 0.0021581468172371387,
-      "learning_rate": 0.0001999943188524202,
-      "loss": 46.0,
-      "step": 44388
-    },
-    {
-      "epoch": 3.393849035686297,
-      "grad_norm": 0.0010917327599599957,
-      "learning_rate": 0.00019999431859638533,
-      "loss": 46.0,
-      "step": 44389
-    },
-    {
-      "epoch": 3.3939254926696867,
-      "grad_norm": 0.0006710791494697332,
-      "learning_rate": 0.00019999431834034467,
-      "loss": 46.0,
-      "step": 44390
-    },
-    {
-      "epoch": 3.3940019496530764,
-      "grad_norm": 0.0017409648280590773,
-      "learning_rate": 0.00019999431808429826,
-      "loss": 46.0,
-      "step": 44391
-    },
-    {
-      "epoch": 3.394078406636466,
-      "grad_norm": 0.0023581183049827814,
-      "learning_rate": 0.00019999431782824609,
-      "loss": 46.0,
-      "step": 44392
-    },
-    {
-      "epoch": 3.394154863619856,
-      "grad_norm": 0.003903318429365754,
-      "learning_rate": 0.00019999431757218816,
-      "loss": 46.0,
-      "step": 44393
-    },
-    {
-      "epoch": 3.3942313206032457,
-      "grad_norm": 0.0016468054382130504,
-      "learning_rate": 0.0001999943173161244,
-      "loss": 46.0,
-      "step": 44394
-    },
-    {
-      "epoch": 3.3943077775866355,
-      "grad_norm": 0.0025710039772093296,
-      "learning_rate": 0.00019999431706005494,
-      "loss": 46.0,
-      "step": 44395
-    },
-    {
-      "epoch": 3.3943842345700252,
-      "grad_norm": 0.003464498557150364,
-      "learning_rate": 0.0001999943168039797,
-      "loss": 46.0,
-      "step": 44396
-    },
-    {
-      "epoch": 3.3944606915534146,
-      "grad_norm": 0.0008923442801460624,
-      "learning_rate": 0.00019999431654789868,
-      "loss": 46.0,
-      "step": 44397
-    },
-    {
-      "epoch": 3.3945371485368043,
-      "grad_norm": 0.0006607627728953958,
-      "learning_rate": 0.00019999431629181186,
-      "loss": 46.0,
-      "step": 44398
-    },
-    {
-      "epoch": 3.394613605520194,
-      "grad_norm": 0.0028931759297847748,
-      "learning_rate": 0.0001999943160357193,
-      "loss": 46.0,
-      "step": 44399
-    },
-    {
-      "epoch": 3.394690062503584,
-      "grad_norm": 0.0033882500138133764,
-      "learning_rate": 0.00019999431577962095,
-      "loss": 46.0,
-      "step": 44400
-    },
-    {
-      "epoch": 3.3947665194869736,
-      "grad_norm": 0.0036598783917725086,
-      "learning_rate": 0.00019999431552351687,
-      "loss": 46.0,
-      "step": 44401
-    },
-    {
-      "epoch": 3.3948429764703634,
-      "grad_norm": 0.001742070191539824,
-      "learning_rate": 0.00019999431526740698,
-      "loss": 46.0,
-      "step": 44402
-    },
-    {
-      "epoch": 3.394919433453753,
-      "grad_norm": 0.0016772841336205602,
-      "learning_rate": 0.00019999431501129135,
-      "loss": 46.0,
-      "step": 44403
-    },
-    {
-      "epoch": 3.394995890437143,
-      "grad_norm": 0.0008965773740783334,
-      "learning_rate": 0.00019999431475516995,
-      "loss": 46.0,
-      "step": 44404
-    },
-    {
-      "epoch": 3.3950723474205327,
-      "grad_norm": 0.001574399066157639,
-      "learning_rate": 0.00019999431449904277,
-      "loss": 46.0,
-      "step": 44405
-    },
-    {
-      "epoch": 3.3951488044039224,
-      "grad_norm": 0.0015241751680150628,
-      "learning_rate": 0.00019999431424290982,
-      "loss": 46.0,
-      "step": 44406
-    },
-    {
-      "epoch": 3.3952252613873117,
-      "grad_norm": 0.0018504337640479207,
-      "learning_rate": 0.0001999943139867711,
-      "loss": 46.0,
-      "step": 44407
-    },
-    {
-      "epoch": 3.3953017183707015,
-      "grad_norm": 0.001041129231452942,
-      "learning_rate": 0.0001999943137306266,
-      "loss": 46.0,
-      "step": 44408
-    },
-    {
-      "epoch": 3.3953781753540913,
-      "grad_norm": 0.0041731493547558784,
-      "learning_rate": 0.00019999431347447635,
-      "loss": 46.0,
-      "step": 44409
-    },
-    {
-      "epoch": 3.395454632337481,
-      "grad_norm": 0.002125500701367855,
-      "learning_rate": 0.0001999943132183203,
-      "loss": 46.0,
-      "step": 44410
-    },
-    {
-      "epoch": 3.3955310893208708,
-      "grad_norm": 0.001525670406408608,
-      "learning_rate": 0.00019999431296215854,
-      "loss": 46.0,
-      "step": 44411
-    },
-    {
-      "epoch": 3.3956075463042605,
-      "grad_norm": 0.0006528489757329226,
-      "learning_rate": 0.00019999431270599098,
-      "loss": 46.0,
-      "step": 44412
-    },
-    {
-      "epoch": 3.3956840032876503,
-      "grad_norm": 0.0033198250457644463,
-      "learning_rate": 0.00019999431244981764,
-      "loss": 46.0,
-      "step": 44413
-    },
-    {
-      "epoch": 3.39576046027104,
-      "grad_norm": 0.0033316079061478376,
-      "learning_rate": 0.00019999431219363856,
-      "loss": 46.0,
-      "step": 44414
-    },
-    {
-      "epoch": 3.39583691725443,
-      "grad_norm": 0.002117588883265853,
-      "learning_rate": 0.00019999431193745365,
-      "loss": 46.0,
-      "step": 44415
-    },
-    {
-      "epoch": 3.3959133742378196,
-      "grad_norm": 0.0005719622713513672,
-      "learning_rate": 0.00019999431168126304,
-      "loss": 46.0,
-      "step": 44416
-    },
-    {
-      "epoch": 3.3959898312212093,
-      "grad_norm": 0.00243534822948277,
-      "learning_rate": 0.00019999431142506664,
-      "loss": 46.0,
-      "step": 44417
-    },
-    {
-      "epoch": 3.396066288204599,
-      "grad_norm": 0.0007536790799349546,
-      "learning_rate": 0.00019999431116886443,
-      "loss": 46.0,
-      "step": 44418
-    },
-    {
-      "epoch": 3.3961427451879884,
-      "grad_norm": 0.001244775834493339,
-      "learning_rate": 0.0001999943109126565,
-      "loss": 46.0,
-      "step": 44419
-    },
-    {
-      "epoch": 3.396219202171378,
-      "grad_norm": 0.0032614772208034992,
-      "learning_rate": 0.0001999943106564428,
-      "loss": 46.0,
-      "step": 44420
-    },
-    {
-      "epoch": 3.396295659154768,
-      "grad_norm": 0.0016448217211291194,
-      "learning_rate": 0.00019999431040022332,
-      "loss": 46.0,
-      "step": 44421
-    },
-    {
-      "epoch": 3.3963721161381577,
-      "grad_norm": 0.002260012086480856,
-      "learning_rate": 0.00019999431014399805,
-      "loss": 46.0,
-      "step": 44422
-    },
-    {
-      "epoch": 3.3964485731215475,
-      "grad_norm": 0.0020634829998016357,
-      "learning_rate": 0.00019999430988776703,
-      "loss": 46.0,
-      "step": 44423
-    },
-    {
-      "epoch": 3.3965250301049372,
-      "grad_norm": 0.00206224643625319,
-      "learning_rate": 0.00019999430963153024,
-      "loss": 46.0,
-      "step": 44424
-    },
-    {
-      "epoch": 3.396601487088327,
-      "grad_norm": 0.005386530887335539,
-      "learning_rate": 0.00019999430937528767,
-      "loss": 46.0,
-      "step": 44425
-    },
-    {
-      "epoch": 3.3966779440717167,
-      "grad_norm": 0.0018630780978128314,
-      "learning_rate": 0.00019999430911903934,
-      "loss": 46.0,
-      "step": 44426
-    },
-    {
-      "epoch": 3.3967544010551065,
-      "grad_norm": 0.0026223179884254932,
-      "learning_rate": 0.00019999430886278522,
-      "loss": 46.0,
-      "step": 44427
-    },
-    {
-      "epoch": 3.3968308580384963,
-      "grad_norm": 0.0020208198111504316,
-      "learning_rate": 0.00019999430860652537,
-      "loss": 46.0,
-      "step": 44428
-    },
-    {
-      "epoch": 3.3969073150218856,
-      "grad_norm": 0.0014199266443029046,
-      "learning_rate": 0.00019999430835025974,
-      "loss": 46.0,
-      "step": 44429
-    },
-    {
-      "epoch": 3.3969837720052753,
-      "grad_norm": 0.002945514163002372,
-      "learning_rate": 0.00019999430809398833,
-      "loss": 46.0,
-      "step": 44430
-    },
-    {
-      "epoch": 3.397060228988665,
-      "grad_norm": 0.0024349556770175695,
-      "learning_rate": 0.00019999430783771116,
-      "loss": 46.0,
-      "step": 44431
-    },
-    {
-      "epoch": 3.397136685972055,
-      "grad_norm": 0.0036872962955385447,
-      "learning_rate": 0.0001999943075814282,
-      "loss": 46.0,
-      "step": 44432
-    },
-    {
-      "epoch": 3.3972131429554446,
-      "grad_norm": 0.0009498997824266553,
-      "learning_rate": 0.00019999430732513948,
-      "loss": 46.0,
-      "step": 44433
-    },
-    {
-      "epoch": 3.3972895999388344,
-      "grad_norm": 0.0023024696856737137,
-      "learning_rate": 0.000199994307068845,
-      "loss": 46.0,
-      "step": 44434
-    },
-    {
-      "epoch": 3.397366056922224,
-      "grad_norm": 0.0020671552047133446,
-      "learning_rate": 0.00019999430681254477,
-      "loss": 46.0,
-      "step": 44435
-    },
-    {
-      "epoch": 3.397442513905614,
-      "grad_norm": 0.0015280861407518387,
-      "learning_rate": 0.00019999430655623872,
-      "loss": 46.0,
-      "step": 44436
-    },
-    {
-      "epoch": 3.3975189708890037,
-      "grad_norm": 0.0005792198935523629,
-      "learning_rate": 0.00019999430629992696,
-      "loss": 46.0,
-      "step": 44437
-    },
-    {
-      "epoch": 3.3975954278723934,
-      "grad_norm": 0.0013664521975442767,
-      "learning_rate": 0.00019999430604360937,
-      "loss": 46.0,
-      "step": 44438
-    },
-    {
-      "epoch": 3.397671884855783,
-      "grad_norm": 0.002889463910833001,
-      "learning_rate": 0.00019999430578728603,
-      "loss": 46.0,
-      "step": 44439
-    },
-    {
-      "epoch": 3.397748341839173,
-      "grad_norm": 0.0014578712871298194,
-      "learning_rate": 0.00019999430553095695,
-      "loss": 46.0,
-      "step": 44440
-    },
-    {
-      "epoch": 3.3978247988225623,
-      "grad_norm": 0.0027165664359927177,
-      "learning_rate": 0.0001999943052746221,
-      "loss": 46.0,
-      "step": 44441
-    },
-    {
-      "epoch": 3.397901255805952,
-      "grad_norm": 0.0019710944034159184,
-      "learning_rate": 0.00019999430501828143,
-      "loss": 46.0,
-      "step": 44442
-    },
-    {
-      "epoch": 3.397977712789342,
-      "grad_norm": 0.002875181846320629,
-      "learning_rate": 0.00019999430476193503,
-      "loss": 46.0,
-      "step": 44443
-    },
-    {
-      "epoch": 3.3980541697727316,
-      "grad_norm": 0.0008439186494797468,
-      "learning_rate": 0.00019999430450558288,
-      "loss": 46.0,
-      "step": 44444
-    },
-    {
-      "epoch": 3.3981306267561213,
-      "grad_norm": 0.001341644674539566,
-      "learning_rate": 0.00019999430424922493,
-      "loss": 46.0,
-      "step": 44445
-    },
-    {
-      "epoch": 3.398207083739511,
-      "grad_norm": 0.004887677729129791,
-      "learning_rate": 0.0001999943039928612,
-      "loss": 46.0,
-      "step": 44446
-    },
-    {
-      "epoch": 3.398283540722901,
-      "grad_norm": 0.0017702901968732476,
-      "learning_rate": 0.00019999430373649174,
-      "loss": 46.0,
-      "step": 44447
-    },
-    {
-      "epoch": 3.3983599977062906,
-      "grad_norm": 0.001969631528481841,
-      "learning_rate": 0.00019999430348011647,
-      "loss": 46.0,
-      "step": 44448
-    },
-    {
-      "epoch": 3.3984364546896804,
-      "grad_norm": 0.0017390295397490263,
-      "learning_rate": 0.00019999430322373548,
-      "loss": 46.0,
-      "step": 44449
-    },
-    {
-      "epoch": 3.39851291167307,
-      "grad_norm": 0.0006267852149903774,
-      "learning_rate": 0.00019999430296734866,
-      "loss": 46.0,
-      "step": 44450
-    },
-    {
-      "epoch": 3.3985893686564594,
-      "grad_norm": 0.0030672969296574593,
-      "learning_rate": 0.00019999430271095612,
-      "loss": 46.0,
-      "step": 44451
-    },
-    {
-      "epoch": 3.398665825639849,
-      "grad_norm": 0.005493775475770235,
-      "learning_rate": 0.00019999430245455776,
-      "loss": 46.0,
-      "step": 44452
-    },
-    {
-      "epoch": 3.398742282623239,
-      "grad_norm": 0.002444051904603839,
-      "learning_rate": 0.0001999943021981537,
-      "loss": 46.0,
-      "step": 44453
-    },
-    {
-      "epoch": 3.3988187396066287,
-      "grad_norm": 0.006261073052883148,
-      "learning_rate": 0.00019999430194174382,
-      "loss": 46.0,
-      "step": 44454
-    },
-    {
-      "epoch": 3.3988951965900185,
-      "grad_norm": 0.001168728107586503,
-      "learning_rate": 0.0001999943016853282,
-      "loss": 46.0,
-      "step": 44455
-    },
-    {
-      "epoch": 3.3989716535734082,
-      "grad_norm": 0.0017290415707975626,
-      "learning_rate": 0.0001999943014289068,
-      "loss": 46.0,
-      "step": 44456
-    },
-    {
-      "epoch": 3.399048110556798,
-      "grad_norm": 0.0007060155621729791,
-      "learning_rate": 0.0001999943011724796,
-      "loss": 46.0,
-      "step": 44457
-    },
-    {
-      "epoch": 3.3991245675401878,
-      "grad_norm": 0.00530970236286521,
-      "learning_rate": 0.00019999430091604666,
-      "loss": 46.0,
-      "step": 44458
-    },
-    {
-      "epoch": 3.3992010245235775,
-      "grad_norm": 0.0008209020597860217,
-      "learning_rate": 0.00019999430065960797,
-      "loss": 46.0,
-      "step": 44459
-    },
-    {
-      "epoch": 3.3992774815069673,
-      "grad_norm": 0.0015252567827701569,
-      "learning_rate": 0.00019999430040316347,
-      "loss": 46.0,
-      "step": 44460
-    },
-    {
-      "epoch": 3.399353938490357,
-      "grad_norm": 0.00172558450140059,
-      "learning_rate": 0.00019999430014671323,
-      "loss": 46.0,
-      "step": 44461
-    },
-    {
-      "epoch": 3.399430395473747,
-      "grad_norm": 0.005498908925801516,
-      "learning_rate": 0.0001999942998902572,
-      "loss": 46.0,
-      "step": 44462
-    },
-    {
-      "epoch": 3.399506852457136,
-      "grad_norm": 0.004490367602556944,
-      "learning_rate": 0.00019999429963379545,
-      "loss": 46.0,
-      "step": 44463
-    },
-    {
-      "epoch": 3.399583309440526,
-      "grad_norm": 0.0025596641935408115,
-      "learning_rate": 0.00019999429937732788,
-      "loss": 46.0,
-      "step": 44464
-    },
-    {
-      "epoch": 3.3996597664239157,
-      "grad_norm": 0.003218061989173293,
-      "learning_rate": 0.00019999429912085455,
-      "loss": 46.0,
-      "step": 44465
-    },
-    {
-      "epoch": 3.3997362234073054,
-      "grad_norm": 0.0014005386037752032,
-      "learning_rate": 0.00019999429886437544,
-      "loss": 46.0,
-      "step": 44466
-    },
-    {
-      "epoch": 3.399812680390695,
-      "grad_norm": 0.0019948091357946396,
-      "learning_rate": 0.00019999429860789055,
-      "loss": 46.0,
-      "step": 44467
-    },
-    {
-      "epoch": 3.399889137374085,
-      "grad_norm": 0.00129279016982764,
-      "learning_rate": 0.00019999429835139995,
-      "loss": 46.0,
-      "step": 44468
-    },
-    {
-      "epoch": 3.3999655943574747,
-      "grad_norm": 0.0012246770784258842,
-      "learning_rate": 0.00019999429809490352,
-      "loss": 46.0,
-      "step": 44469
-    },
-    {
-      "epoch": 3.4000420513408645,
-      "grad_norm": 0.0025491847191005945,
-      "learning_rate": 0.00019999429783840137,
-      "loss": 46.0,
-      "step": 44470
-    },
-    {
-      "epoch": 3.400118508324254,
-      "grad_norm": 0.0013725466560572386,
-      "learning_rate": 0.0001999942975818934,
-      "loss": 46.0,
-      "step": 44471
-    },
-    {
-      "epoch": 3.400194965307644,
-      "grad_norm": 0.002798571949824691,
-      "learning_rate": 0.0001999942973253797,
-      "loss": 46.0,
-      "step": 44472
-    },
-    {
-      "epoch": 3.4002714222910333,
-      "grad_norm": 0.005642000585794449,
-      "learning_rate": 0.00019999429706886023,
-      "loss": 46.0,
-      "step": 44473
-    },
-    {
-      "epoch": 3.400347879274423,
-      "grad_norm": 0.003958832006901503,
-      "learning_rate": 0.000199994296812335,
-      "loss": 46.0,
-      "step": 44474
-    },
-    {
-      "epoch": 3.400424336257813,
-      "grad_norm": 0.003236716380342841,
-      "learning_rate": 0.00019999429655580397,
-      "loss": 46.0,
-      "step": 44475
-    },
-    {
-      "epoch": 3.4005007932412026,
-      "grad_norm": 0.0021570269018411636,
-      "learning_rate": 0.00019999429629926718,
-      "loss": 46.0,
-      "step": 44476
-    },
-    {
-      "epoch": 3.4005772502245923,
-      "grad_norm": 0.0016871554544195533,
-      "learning_rate": 0.00019999429604272462,
-      "loss": 46.0,
-      "step": 44477
-    },
-    {
-      "epoch": 3.400653707207982,
-      "grad_norm": 0.0012570195831358433,
-      "learning_rate": 0.0001999942957861763,
-      "loss": 46.0,
-      "step": 44478
-    },
-    {
-      "epoch": 3.400730164191372,
-      "grad_norm": 0.0011895325733348727,
-      "learning_rate": 0.0001999942955296222,
-      "loss": 46.0,
-      "step": 44479
-    },
-    {
-      "epoch": 3.4008066211747616,
-      "grad_norm": 0.0025255184154957533,
-      "learning_rate": 0.00019999429527306234,
-      "loss": 46.0,
-      "step": 44480
-    },
-    {
-      "epoch": 3.4008830781581514,
-      "grad_norm": 0.002386713633313775,
-      "learning_rate": 0.00019999429501649672,
-      "loss": 46.0,
-      "step": 44481
-    },
-    {
-      "epoch": 3.400959535141541,
-      "grad_norm": 0.0018322336254641414,
-      "learning_rate": 0.0001999942947599253,
-      "loss": 46.0,
-      "step": 44482
-    },
-    {
-      "epoch": 3.401035992124931,
-      "grad_norm": 0.001167122507467866,
-      "learning_rate": 0.0001999942945033481,
-      "loss": 46.0,
-      "step": 44483
-    },
-    {
-      "epoch": 3.4011124491083207,
-      "grad_norm": 0.00200465926900506,
-      "learning_rate": 0.0001999942942467652,
-      "loss": 46.0,
-      "step": 44484
-    },
-    {
-      "epoch": 3.40118890609171,
-      "grad_norm": 0.007078466005623341,
-      "learning_rate": 0.0001999942939901765,
-      "loss": 46.0,
-      "step": 44485
-    },
-    {
-      "epoch": 3.4012653630750997,
-      "grad_norm": 0.002868445124477148,
-      "learning_rate": 0.000199994293733582,
-      "loss": 46.0,
-      "step": 44486
-    },
-    {
-      "epoch": 3.4013418200584895,
-      "grad_norm": 0.00226974836550653,
-      "learning_rate": 0.00019999429347698175,
-      "loss": 46.0,
-      "step": 44487
-    },
-    {
-      "epoch": 3.4014182770418793,
-      "grad_norm": 0.009962710551917553,
-      "learning_rate": 0.00019999429322037574,
-      "loss": 46.0,
-      "step": 44488
-    },
-    {
-      "epoch": 3.401494734025269,
-      "grad_norm": 0.00166595121845603,
-      "learning_rate": 0.00019999429296376398,
-      "loss": 46.0,
-      "step": 44489
-    },
-    {
-      "epoch": 3.401571191008659,
-      "grad_norm": 0.0024232363793998957,
-      "learning_rate": 0.0001999942927071464,
-      "loss": 46.0,
-      "step": 44490
-    },
-    {
-      "epoch": 3.4016476479920486,
-      "grad_norm": 0.002095426432788372,
-      "learning_rate": 0.0001999942924505231,
-      "loss": 46.0,
-      "step": 44491
-    },
-    {
-      "epoch": 3.4017241049754383,
-      "grad_norm": 0.001230902038514614,
-      "learning_rate": 0.000199994292193894,
-      "loss": 46.0,
-      "step": 44492
-    },
-    {
-      "epoch": 3.401800561958828,
-      "grad_norm": 0.0013546929694712162,
-      "learning_rate": 0.00019999429193725914,
-      "loss": 46.0,
-      "step": 44493
-    },
-    {
-      "epoch": 3.4018770189422174,
-      "grad_norm": 0.0024091091472655535,
-      "learning_rate": 0.00019999429168061848,
-      "loss": 46.0,
-      "step": 44494
-    },
-    {
-      "epoch": 3.401953475925607,
-      "grad_norm": 0.0035000063944607973,
-      "learning_rate": 0.0001999942914239721,
-      "loss": 46.0,
-      "step": 44495
-    },
-    {
-      "epoch": 3.402029932908997,
-      "grad_norm": 0.003709676442667842,
-      "learning_rate": 0.00019999429116731993,
-      "loss": 46.0,
-      "step": 44496
-    },
-    {
-      "epoch": 3.4021063898923867,
-      "grad_norm": 0.008089705370366573,
-      "learning_rate": 0.00019999429091066198,
-      "loss": 46.0,
-      "step": 44497
-    },
-    {
-      "epoch": 3.4021828468757764,
-      "grad_norm": 0.002435478614643216,
-      "learning_rate": 0.0001999942906539983,
-      "loss": 46.0,
-      "step": 44498
-    },
-    {
-      "epoch": 3.402259303859166,
-      "grad_norm": 0.0005994358216412365,
-      "learning_rate": 0.00019999429039732882,
-      "loss": 46.0,
-      "step": 44499
-    },
-    {
-      "epoch": 3.402335760842556,
-      "grad_norm": 0.005287316627800465,
-      "learning_rate": 0.00019999429014065355,
-      "loss": 46.0,
-      "step": 44500
-    },
-    {
-      "epoch": 3.4024122178259457,
-      "grad_norm": 0.0025307191535830498,
-      "learning_rate": 0.00019999428988397256,
-      "loss": 46.0,
-      "step": 44501
-    },
-    {
-      "epoch": 3.4024886748093355,
-      "grad_norm": 0.00230208202265203,
-      "learning_rate": 0.00019999428962728577,
-      "loss": 46.0,
-      "step": 44502
-    },
-    {
-      "epoch": 3.4025651317927252,
-      "grad_norm": 0.0045244283974170685,
-      "learning_rate": 0.0001999942893705932,
-      "loss": 46.0,
-      "step": 44503
-    },
-    {
-      "epoch": 3.402641588776115,
-      "grad_norm": 0.0036708831321448088,
-      "learning_rate": 0.00019999428911389488,
-      "loss": 46.0,
-      "step": 44504
-    },
-    {
-      "epoch": 3.4027180457595048,
-      "grad_norm": 0.0018802789272740483,
-      "learning_rate": 0.0001999942888571908,
-      "loss": 46.0,
-      "step": 44505
-    },
-    {
-      "epoch": 3.4027945027428945,
-      "grad_norm": 0.002465044381096959,
-      "learning_rate": 0.00019999428860048094,
-      "loss": 46.0,
-      "step": 44506
-    },
-    {
-      "epoch": 3.402870959726284,
-      "grad_norm": 0.004716214258223772,
-      "learning_rate": 0.0001999942883437653,
-      "loss": 46.0,
-      "step": 44507
-    },
-    {
-      "epoch": 3.4029474167096736,
-      "grad_norm": 0.00487924087792635,
-      "learning_rate": 0.00019999428808704394,
-      "loss": 46.0,
-      "step": 44508
-    },
-    {
-      "epoch": 3.4030238736930634,
-      "grad_norm": 0.002176678040996194,
-      "learning_rate": 0.00019999428783031676,
-      "loss": 46.0,
-      "step": 44509
-    },
-    {
-      "epoch": 3.403100330676453,
-      "grad_norm": 0.0014835112961009145,
-      "learning_rate": 0.0001999942875735838,
-      "loss": 46.0,
-      "step": 44510
-    },
-    {
-      "epoch": 3.403176787659843,
-      "grad_norm": 0.0031991840805858374,
-      "learning_rate": 0.0001999942873168451,
-      "loss": 46.0,
-      "step": 44511
-    },
-    {
-      "epoch": 3.4032532446432326,
-      "grad_norm": 0.0015627560205757618,
-      "learning_rate": 0.00019999428706010065,
-      "loss": 46.0,
-      "step": 44512
-    },
-    {
-      "epoch": 3.4033297016266224,
-      "grad_norm": 0.0018399023683741689,
-      "learning_rate": 0.00019999428680335038,
-      "loss": 46.0,
-      "step": 44513
-    },
-    {
-      "epoch": 3.403406158610012,
-      "grad_norm": 0.0047172242775559425,
-      "learning_rate": 0.0001999942865465944,
-      "loss": 46.0,
-      "step": 44514
-    },
-    {
-      "epoch": 3.403482615593402,
-      "grad_norm": 0.0007529751746915281,
-      "learning_rate": 0.00019999428628983258,
-      "loss": 46.0,
-      "step": 44515
-    },
-    {
-      "epoch": 3.4035590725767912,
-      "grad_norm": 0.0013395254500210285,
-      "learning_rate": 0.00019999428603306501,
-      "loss": 46.0,
-      "step": 44516
-    },
-    {
-      "epoch": 3.403635529560181,
-      "grad_norm": 0.0007168152369558811,
-      "learning_rate": 0.0001999942857762917,
-      "loss": 46.0,
-      "step": 44517
-    },
-    {
-      "epoch": 3.4037119865435708,
-      "grad_norm": 0.004291699267923832,
-      "learning_rate": 0.00019999428551951262,
-      "loss": 46.0,
-      "step": 44518
-    },
-    {
-      "epoch": 3.4037884435269605,
-      "grad_norm": 0.0005293144495226443,
-      "learning_rate": 0.00019999428526272777,
-      "loss": 46.0,
-      "step": 44519
-    },
-    {
-      "epoch": 3.4038649005103503,
-      "grad_norm": 0.0014971561031416059,
-      "learning_rate": 0.00019999428500593714,
-      "loss": 46.0,
-      "step": 44520
-    },
-    {
-      "epoch": 3.40394135749374,
-      "grad_norm": 0.0025402531027793884,
-      "learning_rate": 0.0001999942847491407,
-      "loss": 46.0,
-      "step": 44521
-    },
-    {
-      "epoch": 3.40401781447713,
-      "grad_norm": 0.0006218007765710354,
-      "learning_rate": 0.00019999428449233856,
-      "loss": 46.0,
-      "step": 44522
-    },
-    {
-      "epoch": 3.4040942714605196,
-      "grad_norm": 0.005018069874495268,
-      "learning_rate": 0.00019999428423553064,
-      "loss": 46.0,
-      "step": 44523
-    },
-    {
-      "epoch": 3.4041707284439093,
-      "grad_norm": 0.0021142316982150078,
-      "learning_rate": 0.00019999428397871692,
-      "loss": 46.0,
-      "step": 44524
-    },
-    {
-      "epoch": 3.404247185427299,
-      "grad_norm": 0.001458027632907033,
-      "learning_rate": 0.00019999428372189745,
-      "loss": 46.0,
-      "step": 44525
-    },
-    {
-      "epoch": 3.404323642410689,
-      "grad_norm": 0.0017159141134470701,
-      "learning_rate": 0.0001999942834650722,
-      "loss": 46.0,
-      "step": 44526
-    },
-    {
-      "epoch": 3.4044000993940786,
-      "grad_norm": 0.001311560976319015,
-      "learning_rate": 0.0001999942832082412,
-      "loss": 46.0,
-      "step": 44527
-    },
-    {
-      "epoch": 3.404476556377468,
-      "grad_norm": 0.0012697457568719983,
-      "learning_rate": 0.0001999942829514044,
-      "loss": 46.0,
-      "step": 44528
-    },
-    {
-      "epoch": 3.4045530133608577,
-      "grad_norm": 0.002954694442451,
-      "learning_rate": 0.00019999428269456187,
-      "loss": 46.0,
-      "step": 44529
-    },
-    {
-      "epoch": 3.4046294703442475,
-      "grad_norm": 0.001003327895887196,
-      "learning_rate": 0.00019999428243771354,
-      "loss": 46.0,
-      "step": 44530
-    },
-    {
-      "epoch": 3.404705927327637,
-      "grad_norm": 0.005693940445780754,
-      "learning_rate": 0.00019999428218085946,
-      "loss": 46.0,
-      "step": 44531
-    },
-    {
-      "epoch": 3.404782384311027,
-      "grad_norm": 0.007230098359286785,
-      "learning_rate": 0.0001999942819239996,
-      "loss": 46.0,
-      "step": 44532
-    },
-    {
-      "epoch": 3.4048588412944167,
-      "grad_norm": 0.005926085636019707,
-      "learning_rate": 0.00019999428166713398,
-      "loss": 46.0,
-      "step": 44533
-    },
-    {
-      "epoch": 3.4049352982778065,
-      "grad_norm": 0.004895671270787716,
-      "learning_rate": 0.00019999428141026257,
-      "loss": 46.0,
-      "step": 44534
-    },
-    {
-      "epoch": 3.4050117552611963,
-      "grad_norm": 0.0015876397956162691,
-      "learning_rate": 0.0001999942811533854,
-      "loss": 46.0,
-      "step": 44535
-    },
-    {
-      "epoch": 3.405088212244586,
-      "grad_norm": 0.001187377143651247,
-      "learning_rate": 0.00019999428089650248,
-      "loss": 46.0,
-      "step": 44536
-    },
-    {
-      "epoch": 3.405164669227976,
-      "grad_norm": 0.0035681205336004496,
-      "learning_rate": 0.00019999428063961378,
-      "loss": 46.0,
-      "step": 44537
-    },
-    {
-      "epoch": 3.405241126211365,
-      "grad_norm": 0.0021375594660639763,
-      "learning_rate": 0.0001999942803827193,
-      "loss": 46.0,
-      "step": 44538
-    },
-    {
-      "epoch": 3.405317583194755,
-      "grad_norm": 0.0008389716385863721,
-      "learning_rate": 0.00019999428012581908,
-      "loss": 46.0,
-      "step": 44539
-    },
-    {
-      "epoch": 3.4053940401781446,
-      "grad_norm": 0.0010876808082684875,
-      "learning_rate": 0.00019999427986891303,
-      "loss": 46.0,
-      "step": 44540
-    },
-    {
-      "epoch": 3.4054704971615344,
-      "grad_norm": 0.0011203258763998747,
-      "learning_rate": 0.00019999427961200125,
-      "loss": 46.0,
-      "step": 44541
-    },
-    {
-      "epoch": 3.405546954144924,
-      "grad_norm": 0.0012522428296506405,
-      "learning_rate": 0.0001999942793550837,
-      "loss": 46.0,
-      "step": 44542
-    },
-    {
-      "epoch": 3.405623411128314,
-      "grad_norm": 0.0012317071668803692,
-      "learning_rate": 0.00019999427909816043,
-      "loss": 46.0,
-      "step": 44543
-    },
-    {
-      "epoch": 3.4056998681117037,
-      "grad_norm": 0.0025619708467274904,
-      "learning_rate": 0.00019999427884123132,
-      "loss": 46.0,
-      "step": 44544
-    },
-    {
-      "epoch": 3.4057763250950934,
-      "grad_norm": 0.0031380769796669483,
-      "learning_rate": 0.00019999427858429644,
-      "loss": 46.0,
-      "step": 44545
-    },
-    {
-      "epoch": 3.405852782078483,
-      "grad_norm": 0.001825815299525857,
-      "learning_rate": 0.00019999427832735584,
-      "loss": 46.0,
-      "step": 44546
-    },
-    {
-      "epoch": 3.405929239061873,
-      "grad_norm": 0.002268935786560178,
-      "learning_rate": 0.00019999427807040944,
-      "loss": 46.0,
-      "step": 44547
-    },
-    {
-      "epoch": 3.4060056960452627,
-      "grad_norm": 0.005343676544725895,
-      "learning_rate": 0.00019999427781345727,
-      "loss": 46.0,
-      "step": 44548
-    },
-    {
-      "epoch": 3.4060821530286525,
-      "grad_norm": 0.0013021710328757763,
-      "learning_rate": 0.00019999427755649935,
-      "loss": 46.0,
-      "step": 44549
-    },
-    {
-      "epoch": 3.406158610012042,
-      "grad_norm": 0.0022684766445308924,
-      "learning_rate": 0.00019999427729953565,
-      "loss": 46.0,
-      "step": 44550
-    },
-    {
-      "epoch": 3.4062350669954315,
-      "grad_norm": 0.003002998884767294,
-      "learning_rate": 0.00019999427704256616,
-      "loss": 46.0,
-      "step": 44551
-    },
-    {
-      "epoch": 3.4063115239788213,
-      "grad_norm": 0.005568784195929766,
-      "learning_rate": 0.00019999427678559092,
-      "loss": 46.0,
-      "step": 44552
-    },
-    {
-      "epoch": 3.406387980962211,
-      "grad_norm": 0.002286982722580433,
-      "learning_rate": 0.00019999427652860993,
-      "loss": 46.0,
-      "step": 44553
-    },
-    {
-      "epoch": 3.406464437945601,
-      "grad_norm": 0.0036630250979214907,
-      "learning_rate": 0.00019999427627162314,
-      "loss": 46.0,
-      "step": 44554
-    },
-    {
-      "epoch": 3.4065408949289906,
-      "grad_norm": 0.0025698444806039333,
-      "learning_rate": 0.00019999427601463058,
-      "loss": 46.0,
-      "step": 44555
-    },
-    {
-      "epoch": 3.4066173519123804,
-      "grad_norm": 0.0018158338498324156,
-      "learning_rate": 0.00019999427575763227,
-      "loss": 46.0,
-      "step": 44556
-    },
-    {
-      "epoch": 3.40669380889577,
-      "grad_norm": 0.0012441120343282819,
-      "learning_rate": 0.0001999942755006282,
-      "loss": 46.0,
-      "step": 44557
-    },
-    {
-      "epoch": 3.40677026587916,
-      "grad_norm": 0.0058428216725587845,
-      "learning_rate": 0.00019999427524361834,
-      "loss": 46.0,
-      "step": 44558
-    },
-    {
-      "epoch": 3.4068467228625496,
-      "grad_norm": 0.0015200950438156724,
-      "learning_rate": 0.0001999942749866027,
-      "loss": 46.0,
-      "step": 44559
-    },
-    {
-      "epoch": 3.406923179845939,
-      "grad_norm": 0.00158967066090554,
-      "learning_rate": 0.0001999942747295813,
-      "loss": 46.0,
-      "step": 44560
-    },
-    {
-      "epoch": 3.4069996368293287,
-      "grad_norm": 0.0016352664679288864,
-      "learning_rate": 0.00019999427447255416,
-      "loss": 46.0,
-      "step": 44561
-    },
-    {
-      "epoch": 3.4070760938127185,
-      "grad_norm": 0.004216746427118778,
-      "learning_rate": 0.00019999427421552124,
-      "loss": 46.0,
-      "step": 44562
-    },
-    {
-      "epoch": 3.4071525507961082,
-      "grad_norm": 0.002384734572842717,
-      "learning_rate": 0.00019999427395848252,
-      "loss": 46.0,
-      "step": 44563
-    },
-    {
-      "epoch": 3.407229007779498,
-      "grad_norm": 0.0021031221840530634,
-      "learning_rate": 0.00019999427370143806,
-      "loss": 46.0,
-      "step": 44564
-    },
-    {
-      "epoch": 3.4073054647628878,
-      "grad_norm": 0.0012883978197351098,
-      "learning_rate": 0.00019999427344438782,
-      "loss": 46.0,
-      "step": 44565
-    },
-    {
-      "epoch": 3.4073819217462775,
-      "grad_norm": 0.0022906125523149967,
-      "learning_rate": 0.0001999942731873318,
-      "loss": 46.0,
-      "step": 44566
-    },
-    {
-      "epoch": 3.4074583787296673,
-      "grad_norm": 0.00508294440805912,
-      "learning_rate": 0.00019999427293027004,
-      "loss": 46.0,
-      "step": 44567
-    },
-    {
-      "epoch": 3.407534835713057,
-      "grad_norm": 0.001698538544587791,
-      "learning_rate": 0.00019999427267320248,
-      "loss": 46.0,
-      "step": 44568
-    },
-    {
-      "epoch": 3.407611292696447,
-      "grad_norm": 0.0030605129431933165,
-      "learning_rate": 0.00019999427241612917,
-      "loss": 46.0,
-      "step": 44569
-    },
-    {
-      "epoch": 3.4076877496798366,
-      "grad_norm": 0.003079720539972186,
-      "learning_rate": 0.0001999942721590501,
-      "loss": 46.0,
-      "step": 44570
-    },
-    {
-      "epoch": 3.4077642066632263,
-      "grad_norm": 0.0015500913141295314,
-      "learning_rate": 0.00019999427190196521,
-      "loss": 46.0,
-      "step": 44571
-    },
-    {
-      "epoch": 3.4078406636466156,
-      "grad_norm": 0.001608409802429378,
-      "learning_rate": 0.00019999427164487461,
-      "loss": 46.0,
-      "step": 44572
-    },
-    {
-      "epoch": 3.4079171206300054,
-      "grad_norm": 0.0025723797734826803,
-      "learning_rate": 0.0001999942713877782,
-      "loss": 46.0,
-      "step": 44573
-    },
-    {
-      "epoch": 3.407993577613395,
-      "grad_norm": 0.007553036790341139,
-      "learning_rate": 0.00019999427113067607,
-      "loss": 46.0,
-      "step": 44574
-    },
-    {
-      "epoch": 3.408070034596785,
-      "grad_norm": 0.0016935491003096104,
-      "learning_rate": 0.00019999427087356812,
-      "loss": 46.0,
-      "step": 44575
-    },
-    {
-      "epoch": 3.4081464915801747,
-      "grad_norm": 0.0012393731158226728,
-      "learning_rate": 0.00019999427061645443,
-      "loss": 46.0,
-      "step": 44576
-    },
-    {
-      "epoch": 3.4082229485635644,
-      "grad_norm": 0.0003832278889603913,
-      "learning_rate": 0.00019999427035933496,
-      "loss": 46.0,
-      "step": 44577
-    },
-    {
-      "epoch": 3.408299405546954,
-      "grad_norm": 0.0011358532356098294,
-      "learning_rate": 0.00019999427010220972,
-      "loss": 46.0,
-      "step": 44578
-    },
-    {
-      "epoch": 3.408375862530344,
-      "grad_norm": 0.003860140684992075,
-      "learning_rate": 0.0001999942698450787,
-      "loss": 46.0,
-      "step": 44579
-    },
-    {
-      "epoch": 3.4084523195137337,
-      "grad_norm": 0.0018668039701879025,
-      "learning_rate": 0.00019999426958794195,
-      "loss": 46.0,
-      "step": 44580
-    },
-    {
-      "epoch": 3.4085287764971235,
-      "grad_norm": 0.001568567007780075,
-      "learning_rate": 0.0001999942693307994,
-      "loss": 46.0,
-      "step": 44581
-    },
-    {
-      "epoch": 3.408605233480513,
-      "grad_norm": 0.0006676905904896557,
-      "learning_rate": 0.00019999426907365108,
-      "loss": 46.0,
-      "step": 44582
-    },
-    {
-      "epoch": 3.4086816904639026,
-      "grad_norm": 0.0035766656510531902,
-      "learning_rate": 0.000199994268816497,
-      "loss": 46.0,
-      "step": 44583
-    },
-    {
-      "epoch": 3.4087581474472923,
-      "grad_norm": 0.0032155183143913746,
-      "learning_rate": 0.00019999426855933715,
-      "loss": 46.0,
-      "step": 44584
-    },
-    {
-      "epoch": 3.408834604430682,
-      "grad_norm": 0.001143681351095438,
-      "learning_rate": 0.00019999426830217152,
-      "loss": 46.0,
-      "step": 44585
-    },
-    {
-      "epoch": 3.408911061414072,
-      "grad_norm": 0.002032248768955469,
-      "learning_rate": 0.00019999426804500012,
-      "loss": 46.0,
-      "step": 44586
-    },
-    {
-      "epoch": 3.4089875183974616,
-      "grad_norm": 0.005097049288451672,
-      "learning_rate": 0.00019999426778782297,
-      "loss": 46.0,
-      "step": 44587
-    },
-    {
-      "epoch": 3.4090639753808514,
-      "grad_norm": 0.0031921030022203922,
-      "learning_rate": 0.00019999426753064005,
-      "loss": 46.0,
-      "step": 44588
-    },
-    {
-      "epoch": 3.409140432364241,
-      "grad_norm": 0.0031302792485803366,
-      "learning_rate": 0.00019999426727345133,
-      "loss": 46.0,
-      "step": 44589
-    },
-    {
-      "epoch": 3.409216889347631,
-      "grad_norm": 0.0020225311163812876,
-      "learning_rate": 0.0001999942670162569,
-      "loss": 46.0,
-      "step": 44590
-    },
-    {
-      "epoch": 3.4092933463310207,
-      "grad_norm": 0.0010203256970271468,
-      "learning_rate": 0.00019999426675905663,
-      "loss": 46.0,
-      "step": 44591
-    },
-    {
-      "epoch": 3.4093698033144104,
-      "grad_norm": 0.002846898278221488,
-      "learning_rate": 0.00019999426650185064,
-      "loss": 46.0,
-      "step": 44592
-    },
-    {
-      "epoch": 3.4094462602978,
-      "grad_norm": 0.004486053250730038,
-      "learning_rate": 0.00019999426624463888,
-      "loss": 46.0,
-      "step": 44593
-    },
-    {
-      "epoch": 3.4095227172811895,
-      "grad_norm": 0.0012737103970721364,
-      "learning_rate": 0.00019999426598742132,
-      "loss": 46.0,
-      "step": 44594
-    },
-    {
-      "epoch": 3.4095991742645793,
-      "grad_norm": 0.001980091677978635,
-      "learning_rate": 0.00019999426573019801,
-      "loss": 46.0,
-      "step": 44595
-    },
-    {
-      "epoch": 3.409675631247969,
-      "grad_norm": 0.0016301375580951571,
-      "learning_rate": 0.00019999426547296894,
-      "loss": 46.0,
-      "step": 44596
-    },
-    {
-      "epoch": 3.409752088231359,
-      "grad_norm": 0.0013766594929620624,
-      "learning_rate": 0.00019999426521573408,
-      "loss": 46.0,
-      "step": 44597
-    },
-    {
-      "epoch": 3.4098285452147485,
-      "grad_norm": 0.0014653644757345319,
-      "learning_rate": 0.00019999426495849348,
-      "loss": 46.0,
-      "step": 44598
-    },
-    {
-      "epoch": 3.4099050021981383,
-      "grad_norm": 0.003097351174801588,
-      "learning_rate": 0.00019999426470124706,
-      "loss": 46.0,
-      "step": 44599
-    },
-    {
-      "epoch": 3.409981459181528,
-      "grad_norm": 0.0043132854625582695,
-      "learning_rate": 0.0001999942644439949,
-      "loss": 46.0,
-      "step": 44600
-    },
-    {
-      "epoch": 3.410057916164918,
-      "grad_norm": 0.003087574616074562,
-      "learning_rate": 0.000199994264186737,
-      "loss": 46.0,
-      "step": 44601
-    },
-    {
-      "epoch": 3.4101343731483076,
-      "grad_norm": 0.0022097467444837093,
-      "learning_rate": 0.0001999942639294733,
-      "loss": 46.0,
-      "step": 44602
-    },
-    {
-      "epoch": 3.4102108301316973,
-      "grad_norm": 0.002199438400566578,
-      "learning_rate": 0.00019999426367220383,
-      "loss": 46.0,
-      "step": 44603
-    },
-    {
-      "epoch": 3.4102872871150867,
-      "grad_norm": 0.0016657226951792836,
-      "learning_rate": 0.0001999942634149286,
-      "loss": 46.0,
-      "step": 44604
-    },
-    {
-      "epoch": 3.4103637440984764,
-      "grad_norm": 0.0020539460238069296,
-      "learning_rate": 0.0001999942631576476,
-      "loss": 46.0,
-      "step": 44605
-    },
-    {
-      "epoch": 3.410440201081866,
-      "grad_norm": 0.002514677355065942,
-      "learning_rate": 0.00019999426290036082,
-      "loss": 46.0,
-      "step": 44606
-    },
-    {
-      "epoch": 3.410516658065256,
-      "grad_norm": 0.0024192591663450003,
-      "learning_rate": 0.0001999942626430683,
-      "loss": 46.0,
-      "step": 44607
-    },
-    {
-      "epoch": 3.4105931150486457,
-      "grad_norm": 0.0018314456101506948,
-      "learning_rate": 0.00019999426238576998,
-      "loss": 46.0,
-      "step": 44608
-    },
-    {
-      "epoch": 3.4106695720320355,
-      "grad_norm": 0.0014262263430282474,
-      "learning_rate": 0.00019999426212846588,
-      "loss": 46.0,
-      "step": 44609
-    },
-    {
-      "epoch": 3.4107460290154252,
-      "grad_norm": 0.0022243992425501347,
-      "learning_rate": 0.00019999426187115602,
-      "loss": 46.0,
-      "step": 44610
-    },
-    {
-      "epoch": 3.410822485998815,
-      "grad_norm": 0.0008551247301511467,
-      "learning_rate": 0.0001999942616138404,
-      "loss": 46.0,
-      "step": 44611
-    },
-    {
-      "epoch": 3.4108989429822048,
-      "grad_norm": 0.004102067090570927,
-      "learning_rate": 0.00019999426135651902,
-      "loss": 46.0,
-      "step": 44612
-    },
-    {
-      "epoch": 3.4109753999655945,
-      "grad_norm": 0.0015754050109535456,
-      "learning_rate": 0.00019999426109919188,
-      "loss": 46.0,
-      "step": 44613
-    },
-    {
-      "epoch": 3.4110518569489843,
-      "grad_norm": 0.0019353434909135103,
-      "learning_rate": 0.00019999426084185893,
-      "loss": 46.0,
-      "step": 44614
-    },
-    {
-      "epoch": 3.411128313932374,
-      "grad_norm": 0.0019014155259355903,
-      "learning_rate": 0.00019999426058452024,
-      "loss": 46.0,
-      "step": 44615
-    },
-    {
-      "epoch": 3.4112047709157634,
-      "grad_norm": 0.002923661144450307,
-      "learning_rate": 0.00019999426032717577,
-      "loss": 46.0,
-      "step": 44616
-    },
-    {
-      "epoch": 3.411281227899153,
-      "grad_norm": 0.0018945930059999228,
-      "learning_rate": 0.00019999426006982556,
-      "loss": 46.0,
-      "step": 44617
-    },
-    {
-      "epoch": 3.411357684882543,
-      "grad_norm": 0.001301974174566567,
-      "learning_rate": 0.00019999425981246955,
-      "loss": 46.0,
-      "step": 44618
-    },
-    {
-      "epoch": 3.4114341418659326,
-      "grad_norm": 0.00305893924087286,
-      "learning_rate": 0.0001999942595551078,
-      "loss": 46.0,
-      "step": 44619
-    },
-    {
-      "epoch": 3.4115105988493224,
-      "grad_norm": 0.0028179411310702562,
-      "learning_rate": 0.00019999425929774023,
-      "loss": 46.0,
-      "step": 44620
-    },
-    {
-      "epoch": 3.411587055832712,
-      "grad_norm": 0.0014932259218767285,
-      "learning_rate": 0.00019999425904036693,
-      "loss": 46.0,
-      "step": 44621
-    },
-    {
-      "epoch": 3.411663512816102,
-      "grad_norm": 0.0017887079156935215,
-      "learning_rate": 0.00019999425878298785,
-      "loss": 46.0,
-      "step": 44622
-    },
-    {
-      "epoch": 3.4117399697994917,
-      "grad_norm": 0.008045237511396408,
-      "learning_rate": 0.000199994258525603,
-      "loss": 46.0,
-      "step": 44623
-    },
-    {
-      "epoch": 3.4118164267828814,
-      "grad_norm": 0.001971038058400154,
-      "learning_rate": 0.0001999942582682124,
-      "loss": 46.0,
-      "step": 44624
-    },
-    {
-      "epoch": 3.4118928837662708,
-      "grad_norm": 0.004962950479239225,
-      "learning_rate": 0.000199994258010816,
-      "loss": 46.0,
-      "step": 44625
-    },
-    {
-      "epoch": 3.4119693407496605,
-      "grad_norm": 0.002649918431416154,
-      "learning_rate": 0.00019999425775341382,
-      "loss": 46.0,
-      "step": 44626
-    },
-    {
-      "epoch": 3.4120457977330503,
-      "grad_norm": 0.0014484574785456061,
-      "learning_rate": 0.0001999942574960059,
-      "loss": 46.0,
-      "step": 44627
-    },
-    {
-      "epoch": 3.41212225471644,
-      "grad_norm": 0.0009665804682299495,
-      "learning_rate": 0.0001999942572385922,
-      "loss": 46.0,
-      "step": 44628
-    },
-    {
-      "epoch": 3.41219871169983,
-      "grad_norm": 0.0032977371010929346,
-      "learning_rate": 0.00019999425698117275,
-      "loss": 46.0,
-      "step": 44629
-    },
-    {
-      "epoch": 3.4122751686832196,
-      "grad_norm": 0.0018153262790292501,
-      "learning_rate": 0.0001999942567237475,
-      "loss": 46.0,
-      "step": 44630
-    },
-    {
-      "epoch": 3.4123516256666093,
-      "grad_norm": 0.002286125672981143,
-      "learning_rate": 0.0001999942564663165,
-      "loss": 46.0,
-      "step": 44631
-    },
-    {
-      "epoch": 3.412428082649999,
-      "grad_norm": 0.0010668054455891252,
-      "learning_rate": 0.00019999425620887976,
-      "loss": 46.0,
-      "step": 44632
-    },
-    {
-      "epoch": 3.412504539633389,
-      "grad_norm": 0.001997595652937889,
-      "learning_rate": 0.0001999942559514372,
-      "loss": 46.0,
-      "step": 44633
-    },
-    {
-      "epoch": 3.4125809966167786,
-      "grad_norm": 0.001615968649275601,
-      "learning_rate": 0.0001999942556939889,
-      "loss": 46.0,
-      "step": 44634
-    },
-    {
-      "epoch": 3.4126574536001684,
-      "grad_norm": 0.0013958230847492814,
-      "learning_rate": 0.00019999425543653482,
-      "loss": 46.0,
-      "step": 44635
-    },
-    {
-      "epoch": 3.412733910583558,
-      "grad_norm": 0.003213487332686782,
-      "learning_rate": 0.000199994255179075,
-      "loss": 46.0,
-      "step": 44636
-    },
-    {
-      "epoch": 3.412810367566948,
-      "grad_norm": 0.001600019633769989,
-      "learning_rate": 0.00019999425492160937,
-      "loss": 46.0,
-      "step": 44637
-    },
-    {
-      "epoch": 3.412886824550337,
-      "grad_norm": 0.0014443201944231987,
-      "learning_rate": 0.00019999425466413797,
-      "loss": 46.0,
-      "step": 44638
-    },
-    {
-      "epoch": 3.412963281533727,
-      "grad_norm": 0.003951051272451878,
-      "learning_rate": 0.00019999425440666083,
-      "loss": 46.0,
-      "step": 44639
-    },
-    {
-      "epoch": 3.4130397385171167,
-      "grad_norm": 0.00178765959572047,
-      "learning_rate": 0.00019999425414917788,
-      "loss": 46.0,
-      "step": 44640
-    },
-    {
-      "epoch": 3.4131161955005065,
-      "grad_norm": 0.002320082625374198,
-      "learning_rate": 0.0001999942538916892,
-      "loss": 46.0,
-      "step": 44641
-    },
-    {
-      "epoch": 3.4131926524838963,
-      "grad_norm": 0.003323833690956235,
-      "learning_rate": 0.00019999425363419475,
-      "loss": 46.0,
-      "step": 44642
-    },
-    {
-      "epoch": 3.413269109467286,
-      "grad_norm": 0.003406842937693,
-      "learning_rate": 0.00019999425337669454,
-      "loss": 46.0,
-      "step": 44643
-    },
-    {
-      "epoch": 3.4133455664506758,
-      "grad_norm": 0.0019844863563776016,
-      "learning_rate": 0.0001999942531191885,
-      "loss": 46.0,
-      "step": 44644
-    },
-    {
-      "epoch": 3.4134220234340655,
-      "grad_norm": 0.0035805413499474525,
-      "learning_rate": 0.00019999425286167677,
-      "loss": 46.0,
-      "step": 44645
-    },
-    {
-      "epoch": 3.4134984804174553,
-      "grad_norm": 0.0032482771202921867,
-      "learning_rate": 0.0001999942526041592,
-      "loss": 46.0,
-      "step": 44646
-    },
-    {
-      "epoch": 3.4135749374008446,
-      "grad_norm": 0.004943655803799629,
-      "learning_rate": 0.0001999942523466359,
-      "loss": 46.0,
-      "step": 44647
-    },
-    {
-      "epoch": 3.4136513943842344,
-      "grad_norm": 0.003509682137519121,
-      "learning_rate": 0.00019999425208910686,
-      "loss": 46.0,
-      "step": 44648
-    },
-    {
-      "epoch": 3.413727851367624,
-      "grad_norm": 0.0021849924232810736,
-      "learning_rate": 0.00019999425183157198,
-      "loss": 46.0,
-      "step": 44649
-    },
-    {
-      "epoch": 3.413804308351014,
-      "grad_norm": 0.003294270019978285,
-      "learning_rate": 0.00019999425157403138,
-      "loss": 46.0,
-      "step": 44650
-    },
-    {
-      "epoch": 3.4138807653344037,
-      "grad_norm": 0.001982462592422962,
-      "learning_rate": 0.00019999425131648498,
-      "loss": 46.0,
-      "step": 44651
-    },
-    {
-      "epoch": 3.4139572223177934,
-      "grad_norm": 0.003685680218040943,
-      "learning_rate": 0.00019999425105893284,
-      "loss": 46.0,
-      "step": 44652
-    },
-    {
-      "epoch": 3.414033679301183,
-      "grad_norm": 0.004441377241164446,
-      "learning_rate": 0.00019999425080137492,
-      "loss": 46.0,
-      "step": 44653
-    },
-    {
-      "epoch": 3.414110136284573,
-      "grad_norm": 0.0014486340805888176,
-      "learning_rate": 0.00019999425054381123,
-      "loss": 46.0,
-      "step": 44654
-    },
-    {
-      "epoch": 3.4141865932679627,
-      "grad_norm": 0.0049875639379024506,
-      "learning_rate": 0.00019999425028624176,
-      "loss": 46.0,
-      "step": 44655
-    },
-    {
-      "epoch": 3.4142630502513525,
-      "grad_norm": 0.004506751429289579,
-      "learning_rate": 0.00019999425002866655,
-      "loss": 46.0,
-      "step": 44656
-    },
-    {
-      "epoch": 3.414339507234742,
-      "grad_norm": 0.0014895175117999315,
-      "learning_rate": 0.00019999424977108554,
-      "loss": 46.0,
-      "step": 44657
-    },
-    {
-      "epoch": 3.414415964218132,
-      "grad_norm": 0.0008397481869906187,
-      "learning_rate": 0.00019999424951349875,
-      "loss": 46.0,
-      "step": 44658
-    },
-    {
-      "epoch": 3.4144924212015217,
-      "grad_norm": 0.0015095225535333157,
-      "learning_rate": 0.00019999424925590625,
-      "loss": 46.0,
-      "step": 44659
-    },
-    {
-      "epoch": 3.414568878184911,
-      "grad_norm": 0.0024280408397316933,
-      "learning_rate": 0.00019999424899830794,
-      "loss": 46.0,
-      "step": 44660
-    },
-    {
-      "epoch": 3.414645335168301,
-      "grad_norm": 0.0017444886034354568,
-      "learning_rate": 0.00019999424874070387,
-      "loss": 46.0,
-      "step": 44661
-    },
-    {
-      "epoch": 3.4147217921516906,
-      "grad_norm": 0.001526823383755982,
-      "learning_rate": 0.00019999424848309401,
-      "loss": 46.0,
-      "step": 44662
-    },
-    {
-      "epoch": 3.4147982491350803,
-      "grad_norm": 0.003983426373451948,
-      "learning_rate": 0.00019999424822547842,
-      "loss": 46.0,
-      "step": 44663
-    },
-    {
-      "epoch": 3.41487470611847,
-      "grad_norm": 0.001521335681900382,
-      "learning_rate": 0.00019999424796785702,
-      "loss": 46.0,
-      "step": 44664
-    },
-    {
-      "epoch": 3.41495116310186,
-      "grad_norm": 0.0010788280051201582,
-      "learning_rate": 0.00019999424771022985,
-      "loss": 46.0,
-      "step": 44665
-    },
-    {
-      "epoch": 3.4150276200852496,
-      "grad_norm": 0.008481916040182114,
-      "learning_rate": 0.00019999424745259696,
-      "loss": 46.0,
-      "step": 44666
-    },
-    {
-      "epoch": 3.4151040770686394,
-      "grad_norm": 0.0018634909065440297,
-      "learning_rate": 0.00019999424719495824,
-      "loss": 46.0,
-      "step": 44667
-    },
-    {
-      "epoch": 3.415180534052029,
-      "grad_norm": 0.0031818212009966373,
-      "learning_rate": 0.0001999942469373138,
-      "loss": 46.0,
-      "step": 44668
-    },
-    {
-      "epoch": 3.4152569910354185,
-      "grad_norm": 0.0014383854577317834,
-      "learning_rate": 0.00019999424667966356,
-      "loss": 46.0,
-      "step": 44669
-    },
-    {
-      "epoch": 3.4153334480188082,
-      "grad_norm": 0.003227618522942066,
-      "learning_rate": 0.00019999424642200758,
-      "loss": 46.0,
-      "step": 44670
-    },
-    {
-      "epoch": 3.415409905002198,
-      "grad_norm": 0.003179847029969096,
-      "learning_rate": 0.00019999424616434582,
-      "loss": 46.0,
-      "step": 44671
-    },
-    {
-      "epoch": 3.4154863619855877,
-      "grad_norm": 0.0007517783669754863,
-      "learning_rate": 0.0001999942459066783,
-      "loss": 46.0,
-      "step": 44672
-    },
-    {
-      "epoch": 3.4155628189689775,
-      "grad_norm": 0.002710822969675064,
-      "learning_rate": 0.00019999424564900499,
-      "loss": 46.0,
-      "step": 44673
-    },
-    {
-      "epoch": 3.4156392759523673,
-      "grad_norm": 0.0033408503513783216,
-      "learning_rate": 0.0001999942453913259,
-      "loss": 46.0,
-      "step": 44674
-    },
-    {
-      "epoch": 3.415715732935757,
-      "grad_norm": 0.0021094053518027067,
-      "learning_rate": 0.00019999424513364106,
-      "loss": 46.0,
-      "step": 44675
-    },
-    {
-      "epoch": 3.415792189919147,
-      "grad_norm": 0.0006639473140239716,
-      "learning_rate": 0.00019999424487595043,
-      "loss": 46.0,
-      "step": 44676
-    },
-    {
-      "epoch": 3.4158686469025366,
-      "grad_norm": 0.006734954658895731,
-      "learning_rate": 0.00019999424461825406,
-      "loss": 46.0,
-      "step": 44677
-    },
-    {
-      "epoch": 3.4159451038859263,
-      "grad_norm": 0.0013490165583789349,
-      "learning_rate": 0.0001999942443605519,
-      "loss": 46.0,
-      "step": 44678
-    },
-    {
-      "epoch": 3.416021560869316,
-      "grad_norm": 0.0005813298048451543,
-      "learning_rate": 0.000199994244102844,
-      "loss": 46.0,
-      "step": 44679
-    },
-    {
-      "epoch": 3.416098017852706,
-      "grad_norm": 0.002728705294430256,
-      "learning_rate": 0.0001999942438451303,
-      "loss": 46.0,
-      "step": 44680
-    },
-    {
-      "epoch": 3.416174474836095,
-      "grad_norm": 0.00444404873996973,
-      "learning_rate": 0.00019999424358741084,
-      "loss": 46.0,
-      "step": 44681
-    },
-    {
-      "epoch": 3.416250931819485,
-      "grad_norm": 0.008355891332030296,
-      "learning_rate": 0.0001999942433296856,
-      "loss": 46.0,
-      "step": 44682
-    },
-    {
-      "epoch": 3.4163273888028747,
-      "grad_norm": 0.0020598899573087692,
-      "learning_rate": 0.00019999424307195462,
-      "loss": 46.0,
-      "step": 44683
-    },
-    {
-      "epoch": 3.4164038457862644,
-      "grad_norm": 0.0038032501470297575,
-      "learning_rate": 0.00019999424281421787,
-      "loss": 46.0,
-      "step": 44684
-    },
-    {
-      "epoch": 3.416480302769654,
-      "grad_norm": 0.0028091317508369684,
-      "learning_rate": 0.0001999942425564753,
-      "loss": 46.0,
-      "step": 44685
-    },
-    {
-      "epoch": 3.416556759753044,
-      "grad_norm": 0.001817409647628665,
-      "learning_rate": 0.00019999424229872703,
-      "loss": 46.0,
-      "step": 44686
-    },
-    {
-      "epoch": 3.4166332167364337,
-      "grad_norm": 0.0021161246113479137,
-      "learning_rate": 0.00019999424204097298,
-      "loss": 46.0,
-      "step": 44687
-    },
-    {
-      "epoch": 3.4167096737198235,
-      "grad_norm": 0.0016338556306436658,
-      "learning_rate": 0.0001999942417832131,
-      "loss": 46.0,
-      "step": 44688
-    },
-    {
-      "epoch": 3.4167861307032132,
-      "grad_norm": 0.00104104436468333,
-      "learning_rate": 0.0001999942415254475,
-      "loss": 46.0,
-      "step": 44689
-    },
-    {
-      "epoch": 3.416862587686603,
-      "grad_norm": 0.0007239144179038703,
-      "learning_rate": 0.0001999942412676761,
-      "loss": 46.0,
-      "step": 44690
-    },
-    {
-      "epoch": 3.4169390446699923,
-      "grad_norm": 0.002731866203248501,
-      "learning_rate": 0.000199994241009899,
-      "loss": 46.0,
-      "step": 44691
-    },
-    {
-      "epoch": 3.417015501653382,
-      "grad_norm": 0.0031095221638679504,
-      "learning_rate": 0.00019999424075211605,
-      "loss": 46.0,
-      "step": 44692
-    },
-    {
-      "epoch": 3.417091958636772,
-      "grad_norm": 0.0027906717732548714,
-      "learning_rate": 0.00019999424049432736,
-      "loss": 46.0,
-      "step": 44693
-    },
-    {
-      "epoch": 3.4171684156201616,
-      "grad_norm": 0.0021964660845696926,
-      "learning_rate": 0.0001999942402365329,
-      "loss": 46.0,
-      "step": 44694
-    },
-    {
-      "epoch": 3.4172448726035514,
-      "grad_norm": 0.0027778425719588995,
-      "learning_rate": 0.00019999423997873268,
-      "loss": 46.0,
-      "step": 44695
-    },
-    {
-      "epoch": 3.417321329586941,
-      "grad_norm": 0.002277670195326209,
-      "learning_rate": 0.0001999942397209267,
-      "loss": 46.0,
-      "step": 44696
-    },
-    {
-      "epoch": 3.417397786570331,
-      "grad_norm": 0.0007492602453567088,
-      "learning_rate": 0.00019999423946311494,
-      "loss": 46.0,
-      "step": 44697
-    },
-    {
-      "epoch": 3.4174742435537206,
-      "grad_norm": 0.0025637235958129168,
-      "learning_rate": 0.0001999942392052974,
-      "loss": 46.0,
-      "step": 44698
-    },
-    {
-      "epoch": 3.4175507005371104,
-      "grad_norm": 0.0014416198246181011,
-      "learning_rate": 0.00019999423894747408,
-      "loss": 46.0,
-      "step": 44699
-    },
-    {
-      "epoch": 3.4176271575205,
-      "grad_norm": 0.0031861397437751293,
-      "learning_rate": 0.00019999423868964503,
-      "loss": 46.0,
-      "step": 44700
-    },
-    {
-      "epoch": 3.41770361450389,
-      "grad_norm": 0.002186631551012397,
-      "learning_rate": 0.00019999423843181018,
-      "loss": 46.0,
-      "step": 44701
-    },
-    {
-      "epoch": 3.4177800714872797,
-      "grad_norm": 0.0021601759362965822,
-      "learning_rate": 0.00019999423817396958,
-      "loss": 46.0,
-      "step": 44702
-    },
-    {
-      "epoch": 3.417856528470669,
-      "grad_norm": 0.0018157935701310635,
-      "learning_rate": 0.00019999423791612321,
-      "loss": 46.0,
-      "step": 44703
-    },
-    {
-      "epoch": 3.4179329854540588,
-      "grad_norm": 0.002842133166268468,
-      "learning_rate": 0.00019999423765827104,
-      "loss": 46.0,
-      "step": 44704
-    },
-    {
-      "epoch": 3.4180094424374485,
-      "grad_norm": 0.009291475638747215,
-      "learning_rate": 0.00019999423740041315,
-      "loss": 46.0,
-      "step": 44705
-    },
-    {
-      "epoch": 3.4180858994208383,
-      "grad_norm": 0.0016347493510693312,
-      "learning_rate": 0.00019999423714254946,
-      "loss": 46.0,
-      "step": 44706
-    },
-    {
-      "epoch": 3.418162356404228,
-      "grad_norm": 0.0031018806621432304,
-      "learning_rate": 0.00019999423688468,
-      "loss": 46.0,
-      "step": 44707
-    },
-    {
-      "epoch": 3.418238813387618,
-      "grad_norm": 0.0011205618502572179,
-      "learning_rate": 0.0001999942366268048,
-      "loss": 46.0,
-      "step": 44708
-    },
-    {
-      "epoch": 3.4183152703710076,
-      "grad_norm": 0.001417459105141461,
-      "learning_rate": 0.00019999423636892378,
-      "loss": 46.0,
-      "step": 44709
-    },
-    {
-      "epoch": 3.4183917273543973,
-      "grad_norm": 0.0018594611901789904,
-      "learning_rate": 0.00019999423611103702,
-      "loss": 46.0,
-      "step": 44710
-    },
-    {
-      "epoch": 3.418468184337787,
-      "grad_norm": 0.005445950664579868,
-      "learning_rate": 0.0001999942358531445,
-      "loss": 46.0,
-      "step": 44711
-    },
-    {
-      "epoch": 3.418544641321177,
-      "grad_norm": 0.0020416711922734976,
-      "learning_rate": 0.0001999942355952462,
-      "loss": 46.0,
-      "step": 44712
-    },
-    {
-      "epoch": 3.418621098304566,
-      "grad_norm": 0.00345365097746253,
-      "learning_rate": 0.0001999942353373421,
-      "loss": 46.0,
-      "step": 44713
-    },
-    {
-      "epoch": 3.418697555287956,
-      "grad_norm": 0.006086622830480337,
-      "learning_rate": 0.0001999942350794323,
-      "loss": 46.0,
-      "step": 44714
-    },
-    {
-      "epoch": 3.4187740122713457,
-      "grad_norm": 0.0009702304378151894,
-      "learning_rate": 0.0001999942348215167,
-      "loss": 46.0,
-      "step": 44715
-    },
-    {
-      "epoch": 3.4188504692547355,
-      "grad_norm": 0.0025691729970276356,
-      "learning_rate": 0.00019999423456359532,
-      "loss": 46.0,
-      "step": 44716
-    },
-    {
-      "epoch": 3.418926926238125,
-      "grad_norm": 0.0023943802807480097,
-      "learning_rate": 0.00019999423430566818,
-      "loss": 46.0,
-      "step": 44717
-    },
-    {
-      "epoch": 3.419003383221515,
-      "grad_norm": 0.0037359334528446198,
-      "learning_rate": 0.00019999423404773526,
-      "loss": 46.0,
-      "step": 44718
-    },
-    {
-      "epoch": 3.4190798402049047,
-      "grad_norm": 0.003570438362658024,
-      "learning_rate": 0.00019999423378979655,
-      "loss": 46.0,
-      "step": 44719
-    },
-    {
-      "epoch": 3.4191562971882945,
-      "grad_norm": 0.0025169539730995893,
-      "learning_rate": 0.00019999423353185208,
-      "loss": 46.0,
-      "step": 44720
-    },
-    {
-      "epoch": 3.4192327541716843,
-      "grad_norm": 0.00238978723064065,
-      "learning_rate": 0.0001999942332739019,
-      "loss": 46.0,
-      "step": 44721
-    },
-    {
-      "epoch": 3.419309211155074,
-      "grad_norm": 0.004822245799005032,
-      "learning_rate": 0.0001999942330159459,
-      "loss": 46.0,
-      "step": 44722
-    },
-    {
-      "epoch": 3.419385668138464,
-      "grad_norm": 0.001097759697586298,
-      "learning_rate": 0.00019999423275798413,
-      "loss": 46.0,
-      "step": 44723
-    },
-    {
-      "epoch": 3.4194621251218535,
-      "grad_norm": 0.003434795653447509,
-      "learning_rate": 0.0001999942325000166,
-      "loss": 46.0,
-      "step": 44724
-    },
-    {
-      "epoch": 3.419538582105243,
-      "grad_norm": 0.002722394187003374,
-      "learning_rate": 0.0001999942322420433,
-      "loss": 46.0,
-      "step": 44725
-    },
-    {
-      "epoch": 3.4196150390886326,
-      "grad_norm": 0.0025237184017896652,
-      "learning_rate": 0.00019999423198406425,
-      "loss": 46.0,
-      "step": 44726
-    },
-    {
-      "epoch": 3.4196914960720224,
-      "grad_norm": 0.002259512199088931,
-      "learning_rate": 0.0001999942317260794,
-      "loss": 46.0,
-      "step": 44727
-    },
-    {
-      "epoch": 3.419767953055412,
-      "grad_norm": 0.001269186963327229,
-      "learning_rate": 0.0001999942314680888,
-      "loss": 46.0,
-      "step": 44728
-    },
-    {
-      "epoch": 3.419844410038802,
-      "grad_norm": 0.0028657715301960707,
-      "learning_rate": 0.0001999942312100924,
-      "loss": 46.0,
-      "step": 44729
-    },
-    {
-      "epoch": 3.4199208670221917,
-      "grad_norm": 0.005401701666414738,
-      "learning_rate": 0.00019999423095209027,
-      "loss": 46.0,
-      "step": 44730
-    },
-    {
-      "epoch": 3.4199973240055814,
-      "grad_norm": 0.005532996263355017,
-      "learning_rate": 0.00019999423069408235,
-      "loss": 46.0,
-      "step": 44731
-    },
-    {
-      "epoch": 3.420073780988971,
-      "grad_norm": 0.0012326058931648731,
-      "learning_rate": 0.0001999942304360687,
-      "loss": 46.0,
-      "step": 44732
-    },
-    {
-      "epoch": 3.420150237972361,
-      "grad_norm": 0.0010215356014668941,
-      "learning_rate": 0.00019999423017804922,
-      "loss": 46.0,
-      "step": 44733
-    },
-    {
-      "epoch": 3.4202266949557507,
-      "grad_norm": 0.0005585872568190098,
-      "learning_rate": 0.00019999422992002402,
-      "loss": 46.0,
-      "step": 44734
-    },
-    {
-      "epoch": 3.42030315193914,
-      "grad_norm": 0.004630609881132841,
-      "learning_rate": 0.000199994229661993,
-      "loss": 46.0,
-      "step": 44735
-    },
-    {
-      "epoch": 3.42037960892253,
-      "grad_norm": 0.0015549033414572477,
-      "learning_rate": 0.00019999422940395625,
-      "loss": 46.0,
-      "step": 44736
-    },
-    {
-      "epoch": 3.4204560659059196,
-      "grad_norm": 0.000666903390083462,
-      "learning_rate": 0.00019999422914591372,
-      "loss": 46.0,
-      "step": 44737
-    },
-    {
-      "epoch": 3.4205325228893093,
-      "grad_norm": 0.0011284883366897702,
-      "learning_rate": 0.00019999422888786544,
-      "loss": 46.0,
-      "step": 44738
-    },
-    {
-      "epoch": 3.420608979872699,
-      "grad_norm": 0.00099844834767282,
-      "learning_rate": 0.00019999422862981137,
-      "loss": 46.0,
-      "step": 44739
-    },
-    {
-      "epoch": 3.420685436856089,
-      "grad_norm": 0.0007214226061478257,
-      "learning_rate": 0.00019999422837175152,
-      "loss": 46.0,
-      "step": 44740
-    },
-    {
-      "epoch": 3.4207618938394786,
-      "grad_norm": 0.0020479015074670315,
-      "learning_rate": 0.00019999422811368592,
-      "loss": 46.0,
-      "step": 44741
-    },
-    {
-      "epoch": 3.4208383508228684,
-      "grad_norm": 0.002117303665727377,
-      "learning_rate": 0.00019999422785561458,
-      "loss": 46.0,
-      "step": 44742
-    },
-    {
-      "epoch": 3.420914807806258,
-      "grad_norm": 0.003740879939869046,
-      "learning_rate": 0.0001999942275975374,
-      "loss": 46.0,
-      "step": 44743
-    },
-    {
-      "epoch": 3.420991264789648,
-      "grad_norm": 0.0028734600637108088,
-      "learning_rate": 0.0001999942273394545,
-      "loss": 46.0,
-      "step": 44744
-    },
-    {
-      "epoch": 3.4210677217730376,
-      "grad_norm": 0.0013127618003636599,
-      "learning_rate": 0.0001999942270813658,
-      "loss": 46.0,
-      "step": 44745
-    },
-    {
-      "epoch": 3.4211441787564274,
-      "grad_norm": 0.0029794976580888033,
-      "learning_rate": 0.00019999422682327137,
-      "loss": 46.0,
-      "step": 44746
-    },
-    {
-      "epoch": 3.4212206357398167,
-      "grad_norm": 0.0022585249971598387,
-      "learning_rate": 0.00019999422656517116,
-      "loss": 46.0,
-      "step": 44747
-    },
-    {
-      "epoch": 3.4212970927232065,
-      "grad_norm": 0.0015519445296376944,
-      "learning_rate": 0.00019999422630706518,
-      "loss": 46.0,
-      "step": 44748
-    },
-    {
-      "epoch": 3.4213735497065962,
-      "grad_norm": 0.0006429665954783559,
-      "learning_rate": 0.0001999942260489534,
-      "loss": 46.0,
-      "step": 44749
-    },
-    {
-      "epoch": 3.421450006689986,
-      "grad_norm": 0.0015220160130411386,
-      "learning_rate": 0.00019999422579083587,
-      "loss": 46.0,
-      "step": 44750
-    },
-    {
-      "epoch": 3.4215264636733758,
-      "grad_norm": 0.005609241779893637,
-      "learning_rate": 0.0001999942255327126,
-      "loss": 46.0,
-      "step": 44751
-    },
-    {
-      "epoch": 3.4216029206567655,
-      "grad_norm": 0.0028223353438079357,
-      "learning_rate": 0.00019999422527458354,
-      "loss": 46.0,
-      "step": 44752
-    },
-    {
-      "epoch": 3.4216793776401553,
-      "grad_norm": 0.0022195610217750072,
-      "learning_rate": 0.0001999942250164487,
-      "loss": 46.0,
-      "step": 44753
-    },
-    {
-      "epoch": 3.421755834623545,
-      "grad_norm": 0.001166786765679717,
-      "learning_rate": 0.00019999422475830807,
-      "loss": 46.0,
-      "step": 44754
-    },
-    {
-      "epoch": 3.421832291606935,
-      "grad_norm": 0.0013321212027221918,
-      "learning_rate": 0.00019999422450016173,
-      "loss": 46.0,
-      "step": 44755
-    },
-    {
-      "epoch": 3.421908748590324,
-      "grad_norm": 0.0020120050758123398,
-      "learning_rate": 0.0001999942242420096,
-      "loss": 46.0,
-      "step": 44756
-    },
-    {
-      "epoch": 3.421985205573714,
-      "grad_norm": 0.0029778610914945602,
-      "learning_rate": 0.00019999422398385167,
-      "loss": 46.0,
-      "step": 44757
-    },
-    {
-      "epoch": 3.4220616625571036,
-      "grad_norm": 0.0008817972266115248,
-      "learning_rate": 0.00019999422372568798,
-      "loss": 46.0,
-      "step": 44758
-    },
-    {
-      "epoch": 3.4221381195404934,
-      "grad_norm": 0.0013322122395038605,
-      "learning_rate": 0.00019999422346751855,
-      "loss": 46.0,
-      "step": 44759
-    },
-    {
-      "epoch": 3.422214576523883,
-      "grad_norm": 0.001070545520633459,
-      "learning_rate": 0.0001999942232093433,
-      "loss": 46.0,
-      "step": 44760
-    },
-    {
-      "epoch": 3.422291033507273,
-      "grad_norm": 0.0025846741627901793,
-      "learning_rate": 0.00019999422295116233,
-      "loss": 46.0,
-      "step": 44761
-    },
-    {
-      "epoch": 3.4223674904906627,
-      "grad_norm": 0.0013964733807370067,
-      "learning_rate": 0.00019999422269297557,
-      "loss": 46.0,
-      "step": 44762
-    },
-    {
-      "epoch": 3.4224439474740525,
-      "grad_norm": 0.0014243458863347769,
-      "learning_rate": 0.00019999422243478305,
-      "loss": 46.0,
-      "step": 44763
-    },
-    {
-      "epoch": 3.422520404457442,
-      "grad_norm": 0.0011035765055567026,
-      "learning_rate": 0.00019999422217658474,
-      "loss": 46.0,
-      "step": 44764
-    },
-    {
-      "epoch": 3.422596861440832,
-      "grad_norm": 0.00795206893235445,
-      "learning_rate": 0.0001999942219183807,
-      "loss": 46.0,
-      "step": 44765
-    },
-    {
-      "epoch": 3.4226733184242217,
-      "grad_norm": 0.0024507150519639254,
-      "learning_rate": 0.00019999422166017085,
-      "loss": 46.0,
-      "step": 44766
-    },
-    {
-      "epoch": 3.4227497754076115,
-      "grad_norm": 0.0008149688947014511,
-      "learning_rate": 0.00019999422140195525,
-      "loss": 46.0,
-      "step": 44767
-    },
-    {
-      "epoch": 3.4228262323910013,
-      "grad_norm": 0.0021263097878545523,
-      "learning_rate": 0.00019999422114373388,
-      "loss": 46.0,
-      "step": 44768
-    },
-    {
-      "epoch": 3.4229026893743906,
-      "grad_norm": 0.0009419882553629577,
-      "learning_rate": 0.00019999422088550674,
-      "loss": 46.0,
-      "step": 44769
-    },
-    {
-      "epoch": 3.4229791463577803,
-      "grad_norm": 0.001690485281869769,
-      "learning_rate": 0.00019999422062727385,
-      "loss": 46.0,
-      "step": 44770
-    },
-    {
-      "epoch": 3.42305560334117,
-      "grad_norm": 0.001407820382155478,
-      "learning_rate": 0.00019999422036903516,
-      "loss": 46.0,
-      "step": 44771
-    },
-    {
-      "epoch": 3.42313206032456,
-      "grad_norm": 0.0022869319655001163,
-      "learning_rate": 0.00019999422011079073,
-      "loss": 46.0,
-      "step": 44772
-    },
-    {
-      "epoch": 3.4232085173079496,
-      "grad_norm": 0.001174435019493103,
-      "learning_rate": 0.0001999942198525405,
-      "loss": 46.0,
-      "step": 44773
-    },
-    {
-      "epoch": 3.4232849742913394,
-      "grad_norm": 0.002835790626704693,
-      "learning_rate": 0.0001999942195942845,
-      "loss": 46.0,
-      "step": 44774
-    },
-    {
-      "epoch": 3.423361431274729,
-      "grad_norm": 0.0013584523694589734,
-      "learning_rate": 0.00019999421933602276,
-      "loss": 46.0,
-      "step": 44775
-    },
-    {
-      "epoch": 3.423437888258119,
-      "grad_norm": 0.00241549639031291,
-      "learning_rate": 0.00019999421907775523,
-      "loss": 46.0,
-      "step": 44776
-    },
-    {
-      "epoch": 3.4235143452415087,
-      "grad_norm": 0.0012261769734323025,
-      "learning_rate": 0.00019999421881948193,
-      "loss": 46.0,
-      "step": 44777
-    },
-    {
-      "epoch": 3.423590802224898,
-      "grad_norm": 0.0022929199039936066,
-      "learning_rate": 0.00019999421856120288,
-      "loss": 46.0,
-      "step": 44778
-    },
-    {
-      "epoch": 3.4236672592082877,
-      "grad_norm": 0.0013775007100775838,
-      "learning_rate": 0.00019999421830291806,
-      "loss": 46.0,
-      "step": 44779
-    },
-    {
-      "epoch": 3.4237437161916775,
-      "grad_norm": 0.0021777483634650707,
-      "learning_rate": 0.00019999421804462743,
-      "loss": 46.0,
-      "step": 44780
-    },
-    {
-      "epoch": 3.4238201731750673,
-      "grad_norm": 0.0031727158930152655,
-      "learning_rate": 0.00019999421778633107,
-      "loss": 46.0,
-      "step": 44781
-    },
-    {
-      "epoch": 3.423896630158457,
-      "grad_norm": 0.0007171485922299325,
-      "learning_rate": 0.00019999421752802892,
-      "loss": 46.0,
-      "step": 44782
-    },
-    {
-      "epoch": 3.423973087141847,
-      "grad_norm": 0.0013663237914443016,
-      "learning_rate": 0.00019999421726972104,
-      "loss": 46.0,
-      "step": 44783
-    },
-    {
-      "epoch": 3.4240495441252365,
-      "grad_norm": 0.0032893188763409853,
-      "learning_rate": 0.00019999421701140735,
-      "loss": 46.0,
-      "step": 44784
-    },
-    {
-      "epoch": 3.4241260011086263,
-      "grad_norm": 0.001893207780085504,
-      "learning_rate": 0.0001999942167530879,
-      "loss": 46.0,
-      "step": 44785
-    },
-    {
-      "epoch": 3.424202458092016,
-      "grad_norm": 0.00474650738760829,
-      "learning_rate": 0.0001999942164947627,
-      "loss": 46.0,
-      "step": 44786
-    },
-    {
-      "epoch": 3.424278915075406,
-      "grad_norm": 0.001162675442174077,
-      "learning_rate": 0.0001999942162364317,
-      "loss": 46.0,
-      "step": 44787
-    },
-    {
-      "epoch": 3.4243553720587956,
-      "grad_norm": 0.0033233659341931343,
-      "learning_rate": 0.00019999421597809497,
-      "loss": 46.0,
-      "step": 44788
-    },
-    {
-      "epoch": 3.4244318290421853,
-      "grad_norm": 0.0041215987876057625,
-      "learning_rate": 0.00019999421571975244,
-      "loss": 46.0,
-      "step": 44789
-    },
-    {
-      "epoch": 3.424508286025575,
-      "grad_norm": 0.001480935956351459,
-      "learning_rate": 0.00019999421546140417,
-      "loss": 46.0,
-      "step": 44790
-    },
-    {
-      "epoch": 3.4245847430089644,
-      "grad_norm": 0.0006438368582166731,
-      "learning_rate": 0.0001999942152030501,
-      "loss": 46.0,
-      "step": 44791
-    },
-    {
-      "epoch": 3.424661199992354,
-      "grad_norm": 0.008541252464056015,
-      "learning_rate": 0.00019999421494469024,
-      "loss": 46.0,
-      "step": 44792
-    },
-    {
-      "epoch": 3.424737656975744,
-      "grad_norm": 0.002177627757191658,
-      "learning_rate": 0.00019999421468632465,
-      "loss": 46.0,
-      "step": 44793
-    },
-    {
-      "epoch": 3.4248141139591337,
-      "grad_norm": 0.002214984502643347,
-      "learning_rate": 0.00019999421442795328,
-      "loss": 46.0,
-      "step": 44794
-    },
-    {
-      "epoch": 3.4248905709425235,
-      "grad_norm": 0.0017491765320301056,
-      "learning_rate": 0.00019999421416957617,
-      "loss": 46.0,
-      "step": 44795
-    },
-    {
-      "epoch": 3.4249670279259132,
-      "grad_norm": 0.0004604731220752001,
-      "learning_rate": 0.00019999421391119325,
-      "loss": 46.0,
-      "step": 44796
-    },
-    {
-      "epoch": 3.425043484909303,
-      "grad_norm": 0.0019555080216377974,
-      "learning_rate": 0.00019999421365280456,
-      "loss": 46.0,
-      "step": 44797
-    },
-    {
-      "epoch": 3.4251199418926928,
-      "grad_norm": 0.0009428462362848222,
-      "learning_rate": 0.0001999942133944101,
-      "loss": 46.0,
-      "step": 44798
-    },
-    {
-      "epoch": 3.4251963988760825,
-      "grad_norm": 0.005895137321203947,
-      "learning_rate": 0.00019999421313600992,
-      "loss": 46.0,
-      "step": 44799
-    },
-    {
-      "epoch": 3.425272855859472,
-      "grad_norm": 0.005798183847218752,
-      "learning_rate": 0.00019999421287760391,
-      "loss": 46.0,
-      "step": 44800
-    },
-    {
-      "epoch": 3.4253493128428616,
-      "grad_norm": 0.0016766665503382683,
-      "learning_rate": 0.0001999942126191922,
-      "loss": 46.0,
-      "step": 44801
-    },
-    {
-      "epoch": 3.4254257698262514,
-      "grad_norm": 0.0014669087249785662,
-      "learning_rate": 0.00019999421236077466,
-      "loss": 46.0,
-      "step": 44802
-    },
-    {
-      "epoch": 3.425502226809641,
-      "grad_norm": 0.004316451493650675,
-      "learning_rate": 0.00019999421210235136,
-      "loss": 46.0,
-      "step": 44803
-    },
-    {
-      "epoch": 3.425578683793031,
-      "grad_norm": 0.003676827298477292,
-      "learning_rate": 0.0001999942118439223,
-      "loss": 46.0,
-      "step": 44804
-    },
-    {
-      "epoch": 3.4256551407764206,
-      "grad_norm": 0.0019010331016033888,
-      "learning_rate": 0.0001999942115854875,
-      "loss": 46.0,
-      "step": 44805
-    },
-    {
-      "epoch": 3.4257315977598104,
-      "grad_norm": 0.003554445458576083,
-      "learning_rate": 0.0001999942113270469,
-      "loss": 46.0,
-      "step": 44806
-    },
-    {
-      "epoch": 3.4258080547432,
-      "grad_norm": 0.00315917213447392,
-      "learning_rate": 0.00019999421106860053,
-      "loss": 46.0,
-      "step": 44807
-    },
-    {
-      "epoch": 3.42588451172659,
-      "grad_norm": 0.0024693512823432684,
-      "learning_rate": 0.0001999942108101484,
-      "loss": 46.0,
-      "step": 44808
-    },
-    {
-      "epoch": 3.4259609687099797,
-      "grad_norm": 0.002221296541392803,
-      "learning_rate": 0.00019999421055169047,
-      "loss": 46.0,
-      "step": 44809
-    },
-    {
-      "epoch": 3.4260374256933694,
-      "grad_norm": 0.0034316994715481997,
-      "learning_rate": 0.0001999942102932268,
-      "loss": 46.0,
-      "step": 44810
-    },
-    {
-      "epoch": 3.426113882676759,
-      "grad_norm": 0.007961695082485676,
-      "learning_rate": 0.00019999421003475735,
-      "loss": 46.0,
-      "step": 44811
-    },
-    {
-      "epoch": 3.4261903396601485,
-      "grad_norm": 0.0018990971148014069,
-      "learning_rate": 0.00019999420977628214,
-      "loss": 46.0,
-      "step": 44812
-    },
-    {
-      "epoch": 3.4262667966435383,
-      "grad_norm": 0.0010237082606181502,
-      "learning_rate": 0.00019999420951780116,
-      "loss": 46.0,
-      "step": 44813
-    },
-    {
-      "epoch": 3.426343253626928,
-      "grad_norm": 0.0014058739179745317,
-      "learning_rate": 0.00019999420925931444,
-      "loss": 46.0,
-      "step": 44814
-    },
-    {
-      "epoch": 3.426419710610318,
-      "grad_norm": 0.0014993497170507908,
-      "learning_rate": 0.0001999942090008219,
-      "loss": 46.0,
-      "step": 44815
-    },
-    {
-      "epoch": 3.4264961675937076,
-      "grad_norm": 0.0019525452516973019,
-      "learning_rate": 0.0001999942087423236,
-      "loss": 46.0,
-      "step": 44816
-    },
-    {
-      "epoch": 3.4265726245770973,
-      "grad_norm": 0.0034172164741903543,
-      "learning_rate": 0.00019999420848381956,
-      "loss": 46.0,
-      "step": 44817
-    },
-    {
-      "epoch": 3.426649081560487,
-      "grad_norm": 0.0019351268419995904,
-      "learning_rate": 0.0001999942082253097,
-      "loss": 46.0,
-      "step": 44818
-    },
-    {
-      "epoch": 3.426725538543877,
-      "grad_norm": 0.0013878019526600838,
-      "learning_rate": 0.00019999420796679415,
-      "loss": 46.0,
-      "step": 44819
-    },
-    {
-      "epoch": 3.4268019955272666,
-      "grad_norm": 0.0013011184055358171,
-      "learning_rate": 0.00019999420770827278,
-      "loss": 46.0,
-      "step": 44820
-    },
-    {
-      "epoch": 3.4268784525106564,
-      "grad_norm": 0.0020521623082458973,
-      "learning_rate": 0.0001999942074497456,
-      "loss": 46.0,
-      "step": 44821
-    },
-    {
-      "epoch": 3.4269549094940457,
-      "grad_norm": 0.0038942224346101284,
-      "learning_rate": 0.00019999420719121272,
-      "loss": 46.0,
-      "step": 44822
-    },
-    {
-      "epoch": 3.4270313664774354,
-      "grad_norm": 0.0015740522649139166,
-      "learning_rate": 0.00019999420693267406,
-      "loss": 46.0,
-      "step": 44823
-    },
-    {
-      "epoch": 3.427107823460825,
-      "grad_norm": 0.0023642228916287422,
-      "learning_rate": 0.0001999942066741296,
-      "loss": 46.0,
-      "step": 44824
-    },
-    {
-      "epoch": 3.427184280444215,
-      "grad_norm": 0.002113768830895424,
-      "learning_rate": 0.0001999942064155794,
-      "loss": 46.0,
-      "step": 44825
-    },
-    {
-      "epoch": 3.4272607374276047,
-      "grad_norm": 0.004363631829619408,
-      "learning_rate": 0.00019999420615702341,
-      "loss": 46.0,
-      "step": 44826
-    },
-    {
-      "epoch": 3.4273371944109945,
-      "grad_norm": 0.0054193767718970776,
-      "learning_rate": 0.00019999420589846166,
-      "loss": 46.0,
-      "step": 44827
-    },
-    {
-      "epoch": 3.4274136513943843,
-      "grad_norm": 0.0036532417871057987,
-      "learning_rate": 0.00019999420563989416,
-      "loss": 46.0,
-      "step": 44828
-    },
-    {
-      "epoch": 3.427490108377774,
-      "grad_norm": 0.0021726256236433983,
-      "learning_rate": 0.00019999420538132086,
-      "loss": 46.0,
-      "step": 44829
-    },
-    {
-      "epoch": 3.4275665653611638,
-      "grad_norm": 0.0025024053175002337,
-      "learning_rate": 0.00019999420512274184,
-      "loss": 46.0,
-      "step": 44830
-    },
-    {
-      "epoch": 3.4276430223445535,
-      "grad_norm": 0.0018850157503038645,
-      "learning_rate": 0.000199994204864157,
-      "loss": 46.0,
-      "step": 44831
-    },
-    {
-      "epoch": 3.4277194793279433,
-      "grad_norm": 0.00663653202354908,
-      "learning_rate": 0.0001999942046055664,
-      "loss": 46.0,
-      "step": 44832
-    },
-    {
-      "epoch": 3.427795936311333,
-      "grad_norm": 0.0015489644138142467,
-      "learning_rate": 0.00019999420434697003,
-      "loss": 46.0,
-      "step": 44833
-    },
-    {
-      "epoch": 3.4278723932947224,
-      "grad_norm": 0.001355155953206122,
-      "learning_rate": 0.00019999420408836792,
-      "loss": 46.0,
-      "step": 44834
-    },
-    {
-      "epoch": 3.427948850278112,
-      "grad_norm": 0.0014475092757493258,
-      "learning_rate": 0.00019999420382976,
-      "loss": 46.0,
-      "step": 44835
-    },
-    {
-      "epoch": 3.428025307261502,
-      "grad_norm": 0.003343346295878291,
-      "learning_rate": 0.00019999420357114632,
-      "loss": 46.0,
-      "step": 44836
-    },
-    {
-      "epoch": 3.4281017642448917,
-      "grad_norm": 0.0020684038754552603,
-      "learning_rate": 0.00019999420331252689,
-      "loss": 46.0,
-      "step": 44837
-    },
-    {
-      "epoch": 3.4281782212282814,
-      "grad_norm": 0.0009632062283344567,
-      "learning_rate": 0.00019999420305390165,
-      "loss": 46.0,
-      "step": 44838
-    },
-    {
-      "epoch": 3.428254678211671,
-      "grad_norm": 0.001289335428737104,
-      "learning_rate": 0.0001999942027952707,
-      "loss": 46.0,
-      "step": 44839
-    },
-    {
-      "epoch": 3.428331135195061,
-      "grad_norm": 0.004934332799166441,
-      "learning_rate": 0.00019999420253663397,
-      "loss": 46.0,
-      "step": 44840
-    },
-    {
-      "epoch": 3.4284075921784507,
-      "grad_norm": 0.001309933140873909,
-      "learning_rate": 0.00019999420227799145,
-      "loss": 46.0,
-      "step": 44841
-    },
-    {
-      "epoch": 3.4284840491618405,
-      "grad_norm": 0.001965321134775877,
-      "learning_rate": 0.00019999420201934314,
-      "loss": 46.0,
-      "step": 44842
-    },
-    {
-      "epoch": 3.4285605061452302,
-      "grad_norm": 0.0014687597285956144,
-      "learning_rate": 0.0001999942017606891,
-      "loss": 46.0,
-      "step": 44843
-    },
-    {
-      "epoch": 3.4286369631286195,
-      "grad_norm": 0.002811191137880087,
-      "learning_rate": 0.00019999420150202925,
-      "loss": 46.0,
-      "step": 44844
-    },
-    {
-      "epoch": 3.4287134201120093,
-      "grad_norm": 0.0068966723047196865,
-      "learning_rate": 0.00019999420124336369,
-      "loss": 46.0,
-      "step": 44845
-    },
-    {
-      "epoch": 3.428789877095399,
-      "grad_norm": 0.0012438022531569004,
-      "learning_rate": 0.00019999420098469232,
-      "loss": 46.0,
-      "step": 44846
-    },
-    {
-      "epoch": 3.428866334078789,
-      "grad_norm": 0.001511577283963561,
-      "learning_rate": 0.0001999942007260152,
-      "loss": 46.0,
-      "step": 44847
-    },
-    {
-      "epoch": 3.4289427910621786,
-      "grad_norm": 0.0023268430959433317,
-      "learning_rate": 0.0001999942004673323,
-      "loss": 46.0,
-      "step": 44848
-    },
-    {
-      "epoch": 3.4290192480455683,
-      "grad_norm": 0.00422036973759532,
-      "learning_rate": 0.0001999942002086436,
-      "loss": 46.0,
-      "step": 44849
-    },
-    {
-      "epoch": 3.429095705028958,
-      "grad_norm": 0.0008190243388526142,
-      "learning_rate": 0.0001999941999499492,
-      "loss": 46.0,
-      "step": 44850
-    },
-    {
-      "epoch": 3.429172162012348,
-      "grad_norm": 0.003018898656591773,
-      "learning_rate": 0.00019999419969124897,
-      "loss": 46.0,
-      "step": 44851
-    },
-    {
-      "epoch": 3.4292486189957376,
-      "grad_norm": 0.002889312105253339,
-      "learning_rate": 0.00019999419943254299,
-      "loss": 46.0,
-      "step": 44852
-    },
-    {
-      "epoch": 3.4293250759791274,
-      "grad_norm": 0.0014641005545854568,
-      "learning_rate": 0.00019999419917383123,
-      "loss": 46.0,
-      "step": 44853
-    },
-    {
-      "epoch": 3.429401532962517,
-      "grad_norm": 0.0007817661389708519,
-      "learning_rate": 0.00019999419891511373,
-      "loss": 46.0,
-      "step": 44854
-    },
-    {
-      "epoch": 3.429477989945907,
-      "grad_norm": 0.003971664234995842,
-      "learning_rate": 0.00019999419865639046,
-      "loss": 46.0,
-      "step": 44855
-    },
-    {
-      "epoch": 3.4295544469292962,
-      "grad_norm": 0.0030584794003516436,
-      "learning_rate": 0.0001999941983976614,
-      "loss": 46.0,
-      "step": 44856
-    },
-    {
-      "epoch": 3.429630903912686,
-      "grad_norm": 0.00611590500921011,
-      "learning_rate": 0.00019999419813892657,
-      "loss": 46.0,
-      "step": 44857
-    },
-    {
-      "epoch": 3.4297073608960758,
-      "grad_norm": 0.005710372235625982,
-      "learning_rate": 0.00019999419788018597,
-      "loss": 46.0,
-      "step": 44858
-    },
-    {
-      "epoch": 3.4297838178794655,
-      "grad_norm": 0.001975302817299962,
-      "learning_rate": 0.00019999419762143964,
-      "loss": 46.0,
-      "step": 44859
-    },
-    {
-      "epoch": 3.4298602748628553,
-      "grad_norm": 0.0008019822416827083,
-      "learning_rate": 0.0001999941973626875,
-      "loss": 46.0,
-      "step": 44860
-    },
-    {
-      "epoch": 3.429936731846245,
-      "grad_norm": 0.0006031130324117839,
-      "learning_rate": 0.00019999419710392958,
-      "loss": 46.0,
-      "step": 44861
-    },
-    {
-      "epoch": 3.430013188829635,
-      "grad_norm": 0.0056520369835197926,
-      "learning_rate": 0.00019999419684516592,
-      "loss": 46.0,
-      "step": 44862
-    },
-    {
-      "epoch": 3.4300896458130246,
-      "grad_norm": 0.0022835610434412956,
-      "learning_rate": 0.0001999941965863965,
-      "loss": 46.0,
-      "step": 44863
-    },
-    {
-      "epoch": 3.4301661027964143,
-      "grad_norm": 0.0011921918485313654,
-      "learning_rate": 0.00019999419632762128,
-      "loss": 46.0,
-      "step": 44864
-    },
-    {
-      "epoch": 3.430242559779804,
-      "grad_norm": 0.0016698911786079407,
-      "learning_rate": 0.00019999419606884028,
-      "loss": 46.0,
-      "step": 44865
-    },
-    {
-      "epoch": 3.4303190167631934,
-      "grad_norm": 0.001507268287241459,
-      "learning_rate": 0.00019999419581005355,
-      "loss": 46.0,
-      "step": 44866
-    },
-    {
-      "epoch": 3.430395473746583,
-      "grad_norm": 0.005574206355959177,
-      "learning_rate": 0.00019999419555126105,
-      "loss": 46.0,
-      "step": 44867
-    },
-    {
-      "epoch": 3.430471930729973,
-      "grad_norm": 0.0025285673327744007,
-      "learning_rate": 0.00019999419529246275,
-      "loss": 46.0,
-      "step": 44868
-    },
-    {
-      "epoch": 3.4305483877133627,
-      "grad_norm": 0.001006459933705628,
-      "learning_rate": 0.0001999941950336587,
-      "loss": 46.0,
-      "step": 44869
-    },
-    {
-      "epoch": 3.4306248446967524,
-      "grad_norm": 0.0034164192620664835,
-      "learning_rate": 0.0001999941947748489,
-      "loss": 46.0,
-      "step": 44870
-    },
-    {
-      "epoch": 3.430701301680142,
-      "grad_norm": 0.0015245262766256928,
-      "learning_rate": 0.0001999941945160333,
-      "loss": 46.0,
-      "step": 44871
-    },
-    {
-      "epoch": 3.430777758663532,
-      "grad_norm": 0.0018647368997335434,
-      "learning_rate": 0.00019999419425721193,
-      "loss": 46.0,
-      "step": 44872
-    },
-    {
-      "epoch": 3.4308542156469217,
-      "grad_norm": 0.001559597090817988,
-      "learning_rate": 0.00019999419399838482,
-      "loss": 46.0,
-      "step": 44873
-    },
-    {
-      "epoch": 3.4309306726303115,
-      "grad_norm": 0.0021349964663386345,
-      "learning_rate": 0.00019999419373955193,
-      "loss": 46.0,
-      "step": 44874
-    },
-    {
-      "epoch": 3.4310071296137012,
-      "grad_norm": 0.00545212347060442,
-      "learning_rate": 0.00019999419348071325,
-      "loss": 46.0,
-      "step": 44875
-    },
-    {
-      "epoch": 3.431083586597091,
-      "grad_norm": 0.0021754009649157524,
-      "learning_rate": 0.00019999419322186884,
-      "loss": 46.0,
-      "step": 44876
-    },
-    {
-      "epoch": 3.4311600435804808,
-      "grad_norm": 0.003279255935922265,
-      "learning_rate": 0.0001999941929630186,
-      "loss": 46.0,
-      "step": 44877
-    },
-    {
-      "epoch": 3.43123650056387,
-      "grad_norm": 0.0017380501376464963,
-      "learning_rate": 0.00019999419270416265,
-      "loss": 46.0,
-      "step": 44878
-    },
-    {
-      "epoch": 3.43131295754726,
-      "grad_norm": 0.0005144042079336941,
-      "learning_rate": 0.0001999941924453009,
-      "loss": 46.0,
-      "step": 44879
-    },
-    {
-      "epoch": 3.4313894145306496,
-      "grad_norm": 0.0023301737383008003,
-      "learning_rate": 0.00019999419218643338,
-      "loss": 46.0,
-      "step": 44880
-    },
-    {
-      "epoch": 3.4314658715140394,
-      "grad_norm": 0.0019882656633853912,
-      "learning_rate": 0.0001999941919275601,
-      "loss": 46.0,
-      "step": 44881
-    },
-    {
-      "epoch": 3.431542328497429,
-      "grad_norm": 0.0025007580406963825,
-      "learning_rate": 0.00019999419166868106,
-      "loss": 46.0,
-      "step": 44882
-    },
-    {
-      "epoch": 3.431618785480819,
-      "grad_norm": 0.0036483160220086575,
-      "learning_rate": 0.00019999419140979624,
-      "loss": 46.0,
-      "step": 44883
-    },
-    {
-      "epoch": 3.4316952424642087,
-      "grad_norm": 0.0036548005882650614,
-      "learning_rate": 0.00019999419115090565,
-      "loss": 46.0,
-      "step": 44884
-    },
-    {
-      "epoch": 3.4317716994475984,
-      "grad_norm": 0.0011429466539993882,
-      "learning_rate": 0.00019999419089200928,
-      "loss": 46.0,
-      "step": 44885
-    },
-    {
-      "epoch": 3.431848156430988,
-      "grad_norm": 0.0007388408412225544,
-      "learning_rate": 0.00019999419063310717,
-      "loss": 46.0,
-      "step": 44886
-    },
-    {
-      "epoch": 3.4319246134143775,
-      "grad_norm": 0.004041592590510845,
-      "learning_rate": 0.00019999419037419926,
-      "loss": 46.0,
-      "step": 44887
-    },
-    {
-      "epoch": 3.4320010703977673,
-      "grad_norm": 0.0011919206008315086,
-      "learning_rate": 0.0001999941901152856,
-      "loss": 46.0,
-      "step": 44888
-    },
-    {
-      "epoch": 3.432077527381157,
-      "grad_norm": 0.0030115016270428896,
-      "learning_rate": 0.00019999418985636617,
-      "loss": 46.0,
-      "step": 44889
-    },
-    {
-      "epoch": 3.4321539843645468,
-      "grad_norm": 0.007465067785233259,
-      "learning_rate": 0.00019999418959744096,
-      "loss": 46.0,
-      "step": 44890
-    },
-    {
-      "epoch": 3.4322304413479365,
-      "grad_norm": 0.0019680482801049948,
-      "learning_rate": 0.00019999418933851,
-      "loss": 46.0,
-      "step": 44891
-    },
-    {
-      "epoch": 3.4323068983313263,
-      "grad_norm": 0.0009815064258873463,
-      "learning_rate": 0.00019999418907957326,
-      "loss": 46.0,
-      "step": 44892
-    },
-    {
-      "epoch": 3.432383355314716,
-      "grad_norm": 0.0015862572472542524,
-      "learning_rate": 0.00019999418882063076,
-      "loss": 46.0,
-      "step": 44893
-    },
-    {
-      "epoch": 3.432459812298106,
-      "grad_norm": 0.0047988081350922585,
-      "learning_rate": 0.00019999418856168246,
-      "loss": 46.0,
-      "step": 44894
-    },
-    {
-      "epoch": 3.4325362692814956,
-      "grad_norm": 0.0013331896625459194,
-      "learning_rate": 0.00019999418830272841,
-      "loss": 46.0,
-      "step": 44895
-    },
-    {
-      "epoch": 3.4326127262648853,
-      "grad_norm": 0.0014283873606473207,
-      "learning_rate": 0.0001999941880437686,
-      "loss": 46.0,
-      "step": 44896
-    },
-    {
-      "epoch": 3.432689183248275,
-      "grad_norm": 0.0035420330241322517,
-      "learning_rate": 0.000199994187784803,
-      "loss": 46.0,
-      "step": 44897
-    },
-    {
-      "epoch": 3.432765640231665,
-      "grad_norm": 0.01783638633787632,
-      "learning_rate": 0.00019999418752583164,
-      "loss": 46.0,
-      "step": 44898
-    },
-    {
-      "epoch": 3.4328420972150546,
-      "grad_norm": 0.0004914290038868785,
-      "learning_rate": 0.00019999418726685453,
-      "loss": 46.0,
-      "step": 44899
-    },
-    {
-      "epoch": 3.432918554198444,
-      "grad_norm": 0.0025247198063880205,
-      "learning_rate": 0.00019999418700787161,
-      "loss": 46.0,
-      "step": 44900
-    },
-    {
-      "epoch": 3.4329950111818337,
-      "grad_norm": 0.0036685969680547714,
-      "learning_rate": 0.00019999418674888296,
-      "loss": 46.0,
-      "step": 44901
-    },
-    {
-      "epoch": 3.4330714681652235,
-      "grad_norm": 0.0024150358512997627,
-      "learning_rate": 0.00019999418648988852,
-      "loss": 46.0,
-      "step": 44902
-    },
-    {
-      "epoch": 3.433147925148613,
-      "grad_norm": 0.0038582610432058573,
-      "learning_rate": 0.00019999418623088832,
-      "loss": 46.0,
-      "step": 44903
-    },
-    {
-      "epoch": 3.433224382132003,
-      "grad_norm": 0.002038276055827737,
-      "learning_rate": 0.00019999418597188234,
-      "loss": 46.0,
-      "step": 44904
-    },
-    {
-      "epoch": 3.4333008391153927,
-      "grad_norm": 0.0008746012463234365,
-      "learning_rate": 0.00019999418571287062,
-      "loss": 46.0,
-      "step": 44905
-    },
-    {
-      "epoch": 3.4333772960987825,
-      "grad_norm": 0.002799250418320298,
-      "learning_rate": 0.00019999418545385312,
-      "loss": 46.0,
-      "step": 44906
-    },
-    {
-      "epoch": 3.4334537530821723,
-      "grad_norm": 0.0018621382769197226,
-      "learning_rate": 0.00019999418519482985,
-      "loss": 46.0,
-      "step": 44907
-    },
-    {
-      "epoch": 3.433530210065562,
-      "grad_norm": 0.0010429885005578399,
-      "learning_rate": 0.00019999418493580077,
-      "loss": 46.0,
-      "step": 44908
-    },
-    {
-      "epoch": 3.4336066670489513,
-      "grad_norm": 0.0016331266378983855,
-      "learning_rate": 0.00019999418467676596,
-      "loss": 46.0,
-      "step": 44909
-    },
-    {
-      "epoch": 3.433683124032341,
-      "grad_norm": 0.004243065603077412,
-      "learning_rate": 0.00019999418441772536,
-      "loss": 46.0,
-      "step": 44910
-    },
-    {
-      "epoch": 3.433759581015731,
-      "grad_norm": 0.002113878959789872,
-      "learning_rate": 0.00019999418415867903,
-      "loss": 46.0,
-      "step": 44911
-    },
-    {
-      "epoch": 3.4338360379991206,
-      "grad_norm": 0.0014938856475055218,
-      "learning_rate": 0.0001999941838996269,
-      "loss": 46.0,
-      "step": 44912
-    },
-    {
-      "epoch": 3.4339124949825104,
-      "grad_norm": 0.0018710580188781023,
-      "learning_rate": 0.000199994183640569,
-      "loss": 46.0,
-      "step": 44913
-    },
-    {
-      "epoch": 3.4339889519659,
-      "grad_norm": 0.003282850841060281,
-      "learning_rate": 0.00019999418338150534,
-      "loss": 46.0,
-      "step": 44914
-    },
-    {
-      "epoch": 3.43406540894929,
-      "grad_norm": 0.002514865715056658,
-      "learning_rate": 0.0001999941831224359,
-      "loss": 46.0,
-      "step": 44915
-    },
-    {
-      "epoch": 3.4341418659326797,
-      "grad_norm": 0.002084376523271203,
-      "learning_rate": 0.0001999941828633607,
-      "loss": 46.0,
-      "step": 44916
-    },
-    {
-      "epoch": 3.4342183229160694,
-      "grad_norm": 0.002161577809602022,
-      "learning_rate": 0.00019999418260427976,
-      "loss": 46.0,
-      "step": 44917
-    },
-    {
-      "epoch": 3.434294779899459,
-      "grad_norm": 0.0012893271632492542,
-      "learning_rate": 0.000199994182345193,
-      "loss": 46.0,
-      "step": 44918
-    },
-    {
-      "epoch": 3.434371236882849,
-      "grad_norm": 0.0011127229081466794,
-      "learning_rate": 0.0001999941820861005,
-      "loss": 46.0,
-      "step": 44919
-    },
-    {
-      "epoch": 3.4344476938662387,
-      "grad_norm": 0.0017635368276387453,
-      "learning_rate": 0.00019999418182700224,
-      "loss": 46.0,
-      "step": 44920
-    },
-    {
-      "epoch": 3.4345241508496285,
-      "grad_norm": 0.0032388526014983654,
-      "learning_rate": 0.00019999418156789816,
-      "loss": 46.0,
-      "step": 44921
-    },
-    {
-      "epoch": 3.434600607833018,
-      "grad_norm": 0.002855314640328288,
-      "learning_rate": 0.00019999418130878835,
-      "loss": 46.0,
-      "step": 44922
-    },
-    {
-      "epoch": 3.4346770648164076,
-      "grad_norm": 0.0013176175998523831,
-      "learning_rate": 0.00019999418104967278,
-      "loss": 46.0,
-      "step": 44923
-    },
-    {
-      "epoch": 3.4347535217997973,
-      "grad_norm": 0.004822809249162674,
-      "learning_rate": 0.00019999418079055144,
-      "loss": 46.0,
-      "step": 44924
-    },
-    {
-      "epoch": 3.434829978783187,
-      "grad_norm": 0.008785182610154152,
-      "learning_rate": 0.0001999941805314243,
-      "loss": 46.0,
-      "step": 44925
-    },
-    {
-      "epoch": 3.434906435766577,
-      "grad_norm": 0.000978841446340084,
-      "learning_rate": 0.00019999418027229142,
-      "loss": 46.0,
-      "step": 44926
-    },
-    {
-      "epoch": 3.4349828927499666,
-      "grad_norm": 0.0011776972096413374,
-      "learning_rate": 0.00019999418001315276,
-      "loss": 46.0,
-      "step": 44927
-    },
-    {
-      "epoch": 3.4350593497333564,
-      "grad_norm": 0.0028256208170205355,
-      "learning_rate": 0.0001999941797540083,
-      "loss": 46.0,
-      "step": 44928
-    },
-    {
-      "epoch": 3.435135806716746,
-      "grad_norm": 0.004236609674990177,
-      "learning_rate": 0.00019999417949485813,
-      "loss": 46.0,
-      "step": 44929
-    },
-    {
-      "epoch": 3.435212263700136,
-      "grad_norm": 0.001698176609352231,
-      "learning_rate": 0.00019999417923570215,
-      "loss": 46.0,
-      "step": 44930
-    },
-    {
-      "epoch": 3.435288720683525,
-      "grad_norm": 0.0009990448597818613,
-      "learning_rate": 0.0001999941789765404,
-      "loss": 46.0,
-      "step": 44931
-    },
-    {
-      "epoch": 3.435365177666915,
-      "grad_norm": 0.0025956081226468086,
-      "learning_rate": 0.00019999417871737293,
-      "loss": 46.0,
-      "step": 44932
-    },
-    {
-      "epoch": 3.4354416346503047,
-      "grad_norm": 0.0035286343190819025,
-      "learning_rate": 0.00019999417845819963,
-      "loss": 46.0,
-      "step": 44933
-    },
-    {
-      "epoch": 3.4355180916336945,
-      "grad_norm": 0.0036628209054470062,
-      "learning_rate": 0.00019999417819902059,
-      "loss": 46.0,
-      "step": 44934
-    },
-    {
-      "epoch": 3.4355945486170842,
-      "grad_norm": 0.0016416325233876705,
-      "learning_rate": 0.00019999417793983577,
-      "loss": 46.0,
-      "step": 44935
-    },
-    {
-      "epoch": 3.435671005600474,
-      "grad_norm": 0.0026466776616871357,
-      "learning_rate": 0.00019999417768064518,
-      "loss": 46.0,
-      "step": 44936
-    },
-    {
-      "epoch": 3.4357474625838638,
-      "grad_norm": 0.0018463265150785446,
-      "learning_rate": 0.00019999417742144884,
-      "loss": 46.0,
-      "step": 44937
-    },
-    {
-      "epoch": 3.4358239195672535,
-      "grad_norm": 0.0025940712075680494,
-      "learning_rate": 0.0001999941771622467,
-      "loss": 46.0,
-      "step": 44938
-    },
-    {
-      "epoch": 3.4359003765506433,
-      "grad_norm": 0.002723110606893897,
-      "learning_rate": 0.00019999417690303884,
-      "loss": 46.0,
-      "step": 44939
-    },
-    {
-      "epoch": 3.435976833534033,
-      "grad_norm": 0.0006902538589201868,
-      "learning_rate": 0.0001999941766438252,
-      "loss": 46.0,
-      "step": 44940
-    },
-    {
-      "epoch": 3.436053290517423,
-      "grad_norm": 0.0020823150407522917,
-      "learning_rate": 0.00019999417638460573,
-      "loss": 46.0,
-      "step": 44941
-    },
-    {
-      "epoch": 3.4361297475008126,
-      "grad_norm": 0.0024217634927481413,
-      "learning_rate": 0.00019999417612538055,
-      "loss": 46.0,
-      "step": 44942
-    },
-    {
-      "epoch": 3.436206204484202,
-      "grad_norm": 0.0022657637018710375,
-      "learning_rate": 0.00019999417586614957,
-      "loss": 46.0,
-      "step": 44943
-    },
-    {
-      "epoch": 3.4362826614675916,
-      "grad_norm": 0.002159512136131525,
-      "learning_rate": 0.00019999417560691285,
-      "loss": 46.0,
-      "step": 44944
-    },
-    {
-      "epoch": 3.4363591184509814,
-      "grad_norm": 0.0011793377343565226,
-      "learning_rate": 0.00019999417534767035,
-      "loss": 46.0,
-      "step": 44945
-    },
-    {
-      "epoch": 3.436435575434371,
-      "grad_norm": 0.01255052350461483,
-      "learning_rate": 0.00019999417508842206,
-      "loss": 46.0,
-      "step": 44946
-    },
-    {
-      "epoch": 3.436512032417761,
-      "grad_norm": 0.003107881173491478,
-      "learning_rate": 0.000199994174829168,
-      "loss": 46.0,
-      "step": 44947
-    },
-    {
-      "epoch": 3.4365884894011507,
-      "grad_norm": 0.015196037478744984,
-      "learning_rate": 0.00019999417456990822,
-      "loss": 46.0,
-      "step": 44948
-    },
-    {
-      "epoch": 3.4366649463845405,
-      "grad_norm": 0.0007579917437396944,
-      "learning_rate": 0.00019999417431064263,
-      "loss": 46.0,
-      "step": 44949
-    },
-    {
-      "epoch": 3.43674140336793,
-      "grad_norm": 0.002663498045876622,
-      "learning_rate": 0.0001999941740513713,
-      "loss": 46.0,
-      "step": 44950
-    },
-    {
-      "epoch": 3.43681786035132,
-      "grad_norm": 0.0014326348900794983,
-      "learning_rate": 0.00019999417379209416,
-      "loss": 46.0,
-      "step": 44951
-    },
-    {
-      "epoch": 3.4368943173347097,
-      "grad_norm": 0.004433400928974152,
-      "learning_rate": 0.00019999417353281127,
-      "loss": 46.0,
-      "step": 44952
-    },
-    {
-      "epoch": 3.436970774318099,
-      "grad_norm": 0.002390683628618717,
-      "learning_rate": 0.00019999417327352262,
-      "loss": 46.0,
-      "step": 44953
-    },
-    {
-      "epoch": 3.437047231301489,
-      "grad_norm": 0.002861342392861843,
-      "learning_rate": 0.00019999417301422818,
-      "loss": 46.0,
-      "step": 44954
-    },
-    {
-      "epoch": 3.4371236882848786,
-      "grad_norm": 0.004952490329742432,
-      "learning_rate": 0.00019999417275492798,
-      "loss": 46.0,
-      "step": 44955
-    },
-    {
-      "epoch": 3.4372001452682683,
-      "grad_norm": 0.0014386163093149662,
-      "learning_rate": 0.00019999417249562203,
-      "loss": 46.0,
-      "step": 44956
-    },
-    {
-      "epoch": 3.437276602251658,
-      "grad_norm": 0.002678332384675741,
-      "learning_rate": 0.0001999941722363103,
-      "loss": 46.0,
-      "step": 44957
-    },
-    {
-      "epoch": 3.437353059235048,
-      "grad_norm": 0.0026570528279989958,
-      "learning_rate": 0.00019999417197699278,
-      "loss": 46.0,
-      "step": 44958
-    },
-    {
-      "epoch": 3.4374295162184376,
-      "grad_norm": 0.0025048903189599514,
-      "learning_rate": 0.0001999941717176695,
-      "loss": 46.0,
-      "step": 44959
-    },
-    {
-      "epoch": 3.4375059732018274,
-      "grad_norm": 0.0020839301869273186,
-      "learning_rate": 0.00019999417145834047,
-      "loss": 46.0,
-      "step": 44960
-    },
-    {
-      "epoch": 3.437582430185217,
-      "grad_norm": 0.0009288383298553526,
-      "learning_rate": 0.00019999417119900565,
-      "loss": 46.0,
-      "step": 44961
-    },
-    {
-      "epoch": 3.437658887168607,
-      "grad_norm": 0.0017863233806565404,
-      "learning_rate": 0.0001999941709396651,
-      "loss": 46.0,
-      "step": 44962
-    },
-    {
-      "epoch": 3.4377353441519967,
-      "grad_norm": 0.0011920041870325804,
-      "learning_rate": 0.00019999417068031872,
-      "loss": 46.0,
-      "step": 44963
-    },
-    {
-      "epoch": 3.4378118011353864,
-      "grad_norm": 0.0072701722383499146,
-      "learning_rate": 0.00019999417042096661,
-      "loss": 46.0,
-      "step": 44964
-    },
-    {
-      "epoch": 3.4378882581187757,
-      "grad_norm": 0.000920695427339524,
-      "learning_rate": 0.00019999417016160873,
-      "loss": 46.0,
-      "step": 44965
-    },
-    {
-      "epoch": 3.4379647151021655,
-      "grad_norm": 0.0052647716365754604,
-      "learning_rate": 0.00019999416990224507,
-      "loss": 46.0,
-      "step": 44966
-    },
-    {
-      "epoch": 3.4380411720855553,
-      "grad_norm": 0.0010495687602087855,
-      "learning_rate": 0.00019999416964287564,
-      "loss": 46.0,
-      "step": 44967
-    },
-    {
-      "epoch": 3.438117629068945,
-      "grad_norm": 0.0015211127465590835,
-      "learning_rate": 0.00019999416938350044,
-      "loss": 46.0,
-      "step": 44968
-    },
-    {
-      "epoch": 3.438194086052335,
-      "grad_norm": 0.001882574986666441,
-      "learning_rate": 0.0001999941691241195,
-      "loss": 46.0,
-      "step": 44969
-    },
-    {
-      "epoch": 3.4382705430357245,
-      "grad_norm": 0.0014052526094019413,
-      "learning_rate": 0.00019999416886473274,
-      "loss": 46.0,
-      "step": 44970
-    },
-    {
-      "epoch": 3.4383470000191143,
-      "grad_norm": 0.004943019710481167,
-      "learning_rate": 0.00019999416860534027,
-      "loss": 46.0,
-      "step": 44971
-    },
-    {
-      "epoch": 3.438423457002504,
-      "grad_norm": 0.0022616649512201548,
-      "learning_rate": 0.00019999416834594197,
-      "loss": 46.0,
-      "step": 44972
-    },
-    {
-      "epoch": 3.438499913985894,
-      "grad_norm": 0.0026195053942501545,
-      "learning_rate": 0.00019999416808653796,
-      "loss": 46.0,
-      "step": 44973
-    },
-    {
-      "epoch": 3.4385763709692836,
-      "grad_norm": 0.0018388881580904126,
-      "learning_rate": 0.0001999941678271281,
-      "loss": 46.0,
-      "step": 44974
-    },
-    {
-      "epoch": 3.438652827952673,
-      "grad_norm": 0.005501854699105024,
-      "learning_rate": 0.00019999416756771258,
-      "loss": 46.0,
-      "step": 44975
-    },
-    {
-      "epoch": 3.4387292849360627,
-      "grad_norm": 0.001565106213092804,
-      "learning_rate": 0.00019999416730829121,
-      "loss": 46.0,
-      "step": 44976
-    },
-    {
-      "epoch": 3.4388057419194524,
-      "grad_norm": 0.0018467765767127275,
-      "learning_rate": 0.00019999416704886408,
-      "loss": 46.0,
-      "step": 44977
-    },
-    {
-      "epoch": 3.438882198902842,
-      "grad_norm": 0.0017670224187895656,
-      "learning_rate": 0.00019999416678943122,
-      "loss": 46.0,
-      "step": 44978
-    },
-    {
-      "epoch": 3.438958655886232,
-      "grad_norm": 0.0027717140037566423,
-      "learning_rate": 0.00019999416652999256,
-      "loss": 46.0,
-      "step": 44979
-    },
-    {
-      "epoch": 3.4390351128696217,
-      "grad_norm": 0.002137487754225731,
-      "learning_rate": 0.00019999416627054813,
-      "loss": 46.0,
-      "step": 44980
-    },
-    {
-      "epoch": 3.4391115698530115,
-      "grad_norm": 0.010148182511329651,
-      "learning_rate": 0.00019999416601109796,
-      "loss": 46.0,
-      "step": 44981
-    },
-    {
-      "epoch": 3.4391880268364012,
-      "grad_norm": 0.002510828198865056,
-      "learning_rate": 0.00019999416575164198,
-      "loss": 46.0,
-      "step": 44982
-    },
-    {
-      "epoch": 3.439264483819791,
-      "grad_norm": 0.0020173918455839157,
-      "learning_rate": 0.00019999416549218023,
-      "loss": 46.0,
-      "step": 44983
-    },
-    {
-      "epoch": 3.4393409408031808,
-      "grad_norm": 0.0028393741231411695,
-      "learning_rate": 0.00019999416523271274,
-      "loss": 46.0,
-      "step": 44984
-    },
-    {
-      "epoch": 3.4394173977865705,
-      "grad_norm": 0.0011860416270792484,
-      "learning_rate": 0.0001999941649732395,
-      "loss": 46.0,
-      "step": 44985
-    },
-    {
-      "epoch": 3.4394938547699603,
-      "grad_norm": 0.005023598670959473,
-      "learning_rate": 0.00019999416471376042,
-      "loss": 46.0,
-      "step": 44986
-    },
-    {
-      "epoch": 3.4395703117533496,
-      "grad_norm": 0.002815471263602376,
-      "learning_rate": 0.00019999416445427563,
-      "loss": 46.0,
-      "step": 44987
-    },
-    {
-      "epoch": 3.4396467687367394,
-      "grad_norm": 0.0023719302844256163,
-      "learning_rate": 0.00019999416419478504,
-      "loss": 46.0,
-      "step": 44988
-    },
-    {
-      "epoch": 3.439723225720129,
-      "grad_norm": 0.0026318971067667007,
-      "learning_rate": 0.00019999416393528868,
-      "loss": 46.0,
-      "step": 44989
-    },
-    {
-      "epoch": 3.439799682703519,
-      "grad_norm": 0.0036225770600140095,
-      "learning_rate": 0.00019999416367578657,
-      "loss": 46.0,
-      "step": 44990
-    },
-    {
-      "epoch": 3.4398761396869086,
-      "grad_norm": 0.0028483932837843895,
-      "learning_rate": 0.0001999941634162787,
-      "loss": 46.0,
-      "step": 44991
-    },
-    {
-      "epoch": 3.4399525966702984,
-      "grad_norm": 0.0016213920898735523,
-      "learning_rate": 0.00019999416315676506,
-      "loss": 46.0,
-      "step": 44992
-    },
-    {
-      "epoch": 3.440029053653688,
-      "grad_norm": 0.0014060037210583687,
-      "learning_rate": 0.00019999416289724563,
-      "loss": 46.0,
-      "step": 44993
-    },
-    {
-      "epoch": 3.440105510637078,
-      "grad_norm": 0.0018228681292384863,
-      "learning_rate": 0.00019999416263772043,
-      "loss": 46.0,
-      "step": 44994
-    },
-    {
-      "epoch": 3.4401819676204677,
-      "grad_norm": 0.0030857317615300417,
-      "learning_rate": 0.00019999416237818948,
-      "loss": 46.0,
-      "step": 44995
-    },
-    {
-      "epoch": 3.4402584246038574,
-      "grad_norm": 0.002523573115468025,
-      "learning_rate": 0.00019999416211865273,
-      "loss": 46.0,
-      "step": 44996
-    },
-    {
-      "epoch": 3.4403348815872468,
-      "grad_norm": 0.0011439020745456219,
-      "learning_rate": 0.00019999416185911026,
-      "loss": 46.0,
-      "step": 44997
-    },
-    {
-      "epoch": 3.4404113385706365,
-      "grad_norm": 0.0024027186445891857,
-      "learning_rate": 0.00019999416159956196,
-      "loss": 46.0,
-      "step": 44998
-    },
-    {
-      "epoch": 3.4404877955540263,
-      "grad_norm": 0.001416101586073637,
-      "learning_rate": 0.00019999416134000795,
-      "loss": 46.0,
-      "step": 44999
-    },
-    {
-      "epoch": 3.440564252537416,
-      "grad_norm": 0.0036262685898691416,
-      "learning_rate": 0.00019999416108044816,
-      "loss": 46.0,
-      "step": 45000
-    },
-    {
-      "epoch": 3.440640709520806,
-      "grad_norm": 0.0015977498842403293,
-      "learning_rate": 0.00019999416082088257,
-      "loss": 46.0,
-      "step": 45001
-    },
-    {
-      "epoch": 3.4407171665041956,
-      "grad_norm": 0.0026611043140292168,
-      "learning_rate": 0.0001999941605613112,
-      "loss": 46.0,
-      "step": 45002
-    },
-    {
-      "epoch": 3.4407936234875853,
-      "grad_norm": 0.0008026852738112211,
-      "learning_rate": 0.0001999941603017341,
-      "loss": 46.0,
-      "step": 45003
-    },
-    {
-      "epoch": 3.440870080470975,
-      "grad_norm": 0.0011025264393538237,
-      "learning_rate": 0.00019999416004215124,
-      "loss": 46.0,
-      "step": 45004
-    },
-    {
-      "epoch": 3.440946537454365,
-      "grad_norm": 0.0006148290704004467,
-      "learning_rate": 0.00019999415978256258,
-      "loss": 46.0,
-      "step": 45005
-    },
-    {
-      "epoch": 3.4410229944377546,
-      "grad_norm": 0.0031365377362817526,
-      "learning_rate": 0.00019999415952296813,
-      "loss": 46.0,
-      "step": 45006
-    },
-    {
-      "epoch": 3.4410994514211444,
-      "grad_norm": 0.002085742773488164,
-      "learning_rate": 0.00019999415926336795,
-      "loss": 46.0,
-      "step": 45007
-    },
-    {
-      "epoch": 3.441175908404534,
-      "grad_norm": 0.002416794653981924,
-      "learning_rate": 0.000199994159003762,
-      "loss": 46.0,
-      "step": 45008
-    },
-    {
-      "epoch": 3.4412523653879235,
-      "grad_norm": 0.0020822798833251,
-      "learning_rate": 0.00019999415874415028,
-      "loss": 46.0,
-      "step": 45009
-    },
-    {
-      "epoch": 3.441328822371313,
-      "grad_norm": 0.0030430806800723076,
-      "learning_rate": 0.00019999415848453276,
-      "loss": 46.0,
-      "step": 45010
-    },
-    {
-      "epoch": 3.441405279354703,
-      "grad_norm": 0.0010397444711998105,
-      "learning_rate": 0.0001999941582249095,
-      "loss": 46.0,
-      "step": 45011
-    },
-    {
-      "epoch": 3.4414817363380927,
-      "grad_norm": 0.004340566694736481,
-      "learning_rate": 0.00019999415796528047,
-      "loss": 46.0,
-      "step": 45012
-    },
-    {
-      "epoch": 3.4415581933214825,
-      "grad_norm": 0.002555574057623744,
-      "learning_rate": 0.00019999415770564566,
-      "loss": 46.0,
-      "step": 45013
-    },
-    {
-      "epoch": 3.4416346503048723,
-      "grad_norm": 0.001438225619494915,
-      "learning_rate": 0.0001999941574460051,
-      "loss": 46.0,
-      "step": 45014
-    },
-    {
-      "epoch": 3.441711107288262,
-      "grad_norm": 0.0004707752668764442,
-      "learning_rate": 0.00019999415718635873,
-      "loss": 46.0,
-      "step": 45015
-    },
-    {
-      "epoch": 3.441787564271652,
-      "grad_norm": 0.0008685371140018106,
-      "learning_rate": 0.00019999415692670662,
-      "loss": 46.0,
-      "step": 45016
-    },
-    {
-      "epoch": 3.4418640212550415,
-      "grad_norm": 0.000713616085704416,
-      "learning_rate": 0.00019999415666704874,
-      "loss": 46.0,
-      "step": 45017
-    },
-    {
-      "epoch": 3.441940478238431,
-      "grad_norm": 0.0024848226457834244,
-      "learning_rate": 0.00019999415640738511,
-      "loss": 46.0,
-      "step": 45018
-    },
-    {
-      "epoch": 3.4420169352218206,
-      "grad_norm": 0.0032812978606671095,
-      "learning_rate": 0.00019999415614771568,
-      "loss": 46.0,
-      "step": 45019
-    },
-    {
-      "epoch": 3.4420933922052104,
-      "grad_norm": 0.0011629536747932434,
-      "learning_rate": 0.00019999415588804048,
-      "loss": 46.0,
-      "step": 45020
-    },
-    {
-      "epoch": 3.4421698491886,
-      "grad_norm": 0.0016299558337777853,
-      "learning_rate": 0.00019999415562835953,
-      "loss": 46.0,
-      "step": 45021
-    },
-    {
-      "epoch": 3.44224630617199,
-      "grad_norm": 0.0015818693209439516,
-      "learning_rate": 0.0001999941553686728,
-      "loss": 46.0,
-      "step": 45022
-    },
-    {
-      "epoch": 3.4423227631553797,
-      "grad_norm": 0.00176207663025707,
-      "learning_rate": 0.00019999415510898032,
-      "loss": 46.0,
-      "step": 45023
-    },
-    {
-      "epoch": 3.4423992201387694,
-      "grad_norm": 0.0016777977580204606,
-      "learning_rate": 0.00019999415484928202,
-      "loss": 46.0,
-      "step": 45024
-    },
-    {
-      "epoch": 3.442475677122159,
-      "grad_norm": 0.004494350869208574,
-      "learning_rate": 0.00019999415458957803,
-      "loss": 46.0,
-      "step": 45025
-    },
-    {
-      "epoch": 3.442552134105549,
-      "grad_norm": 0.0008568787598051131,
-      "learning_rate": 0.00019999415432986822,
-      "loss": 46.0,
-      "step": 45026
-    },
-    {
-      "epoch": 3.4426285910889387,
-      "grad_norm": 0.0071852789260447025,
-      "learning_rate": 0.00019999415407015266,
-      "loss": 46.0,
-      "step": 45027
-    },
-    {
-      "epoch": 3.4427050480723285,
-      "grad_norm": 0.0021753127221018076,
-      "learning_rate": 0.0001999941538104313,
-      "loss": 46.0,
-      "step": 45028
-    },
-    {
-      "epoch": 3.4427815050557182,
-      "grad_norm": 0.0020551783964037895,
-      "learning_rate": 0.00019999415355070418,
-      "loss": 46.0,
-      "step": 45029
-    },
-    {
-      "epoch": 3.442857962039108,
-      "grad_norm": 0.003154277103021741,
-      "learning_rate": 0.0001999941532909713,
-      "loss": 46.0,
-      "step": 45030
-    },
-    {
-      "epoch": 3.4429344190224973,
-      "grad_norm": 0.001228502718731761,
-      "learning_rate": 0.00019999415303123265,
-      "loss": 46.0,
-      "step": 45031
-    },
-    {
-      "epoch": 3.443010876005887,
-      "grad_norm": 0.0017970663029700518,
-      "learning_rate": 0.00019999415277148822,
-      "loss": 46.0,
-      "step": 45032
-    },
-    {
-      "epoch": 3.443087332989277,
-      "grad_norm": 0.0025412687100470066,
-      "learning_rate": 0.00019999415251173804,
-      "loss": 46.0,
-      "step": 45033
-    },
-    {
-      "epoch": 3.4431637899726666,
-      "grad_norm": 0.0018300460651516914,
-      "learning_rate": 0.0001999941522519821,
-      "loss": 46.0,
-      "step": 45034
-    },
-    {
-      "epoch": 3.4432402469560563,
-      "grad_norm": 0.003766604932025075,
-      "learning_rate": 0.00019999415199222037,
-      "loss": 46.0,
-      "step": 45035
-    },
-    {
-      "epoch": 3.443316703939446,
-      "grad_norm": 0.003766162320971489,
-      "learning_rate": 0.00019999415173245288,
-      "loss": 46.0,
-      "step": 45036
-    },
-    {
-      "epoch": 3.443393160922836,
-      "grad_norm": 0.002715705195441842,
-      "learning_rate": 0.0001999941514726796,
-      "loss": 46.0,
-      "step": 45037
-    },
-    {
-      "epoch": 3.4434696179062256,
-      "grad_norm": 0.002420001197606325,
-      "learning_rate": 0.00019999415121290057,
-      "loss": 46.0,
-      "step": 45038
-    },
-    {
-      "epoch": 3.4435460748896154,
-      "grad_norm": 0.0019235039362683892,
-      "learning_rate": 0.00019999415095311578,
-      "loss": 46.0,
-      "step": 45039
-    },
-    {
-      "epoch": 3.4436225318730047,
-      "grad_norm": 0.0027975374832749367,
-      "learning_rate": 0.00019999415069332522,
-      "loss": 46.0,
-      "step": 45040
-    },
-    {
-      "epoch": 3.4436989888563945,
-      "grad_norm": 0.001645622425712645,
-      "learning_rate": 0.00019999415043352886,
-      "loss": 46.0,
-      "step": 45041
-    },
-    {
-      "epoch": 3.4437754458397842,
-      "grad_norm": 0.0017790887504816055,
-      "learning_rate": 0.00019999415017372675,
-      "loss": 46.0,
-      "step": 45042
-    },
-    {
-      "epoch": 3.443851902823174,
-      "grad_norm": 0.0012978228041902184,
-      "learning_rate": 0.0001999941499139189,
-      "loss": 46.0,
-      "step": 45043
-    },
-    {
-      "epoch": 3.4439283598065638,
-      "grad_norm": 0.0008418190409429371,
-      "learning_rate": 0.00019999414965410524,
-      "loss": 46.0,
-      "step": 45044
-    },
-    {
-      "epoch": 3.4440048167899535,
-      "grad_norm": 0.004874741192907095,
-      "learning_rate": 0.0001999941493942858,
-      "loss": 46.0,
-      "step": 45045
-    },
-    {
-      "epoch": 3.4440812737733433,
-      "grad_norm": 0.00124945433344692,
-      "learning_rate": 0.0001999941491344606,
-      "loss": 46.0,
-      "step": 45046
-    },
-    {
-      "epoch": 3.444157730756733,
-      "grad_norm": 0.0016876948066055775,
-      "learning_rate": 0.00019999414887462966,
-      "loss": 46.0,
-      "step": 45047
-    },
-    {
-      "epoch": 3.444234187740123,
-      "grad_norm": 0.0027740299701690674,
-      "learning_rate": 0.00019999414861479294,
-      "loss": 46.0,
-      "step": 45048
-    },
-    {
-      "epoch": 3.4443106447235126,
-      "grad_norm": 0.0008311325218528509,
-      "learning_rate": 0.00019999414835495047,
-      "loss": 46.0,
-      "step": 45049
-    },
-    {
-      "epoch": 3.4443871017069023,
-      "grad_norm": 0.003162291133776307,
-      "learning_rate": 0.0001999941480951022,
-      "loss": 46.0,
-      "step": 45050
-    },
-    {
-      "epoch": 3.444463558690292,
-      "grad_norm": 0.0013432904379442334,
-      "learning_rate": 0.00019999414783524816,
-      "loss": 46.0,
-      "step": 45051
-    },
-    {
-      "epoch": 3.444540015673682,
-      "grad_norm": 0.001604159246198833,
-      "learning_rate": 0.00019999414757538838,
-      "loss": 46.0,
-      "step": 45052
-    },
-    {
-      "epoch": 3.444616472657071,
-      "grad_norm": 0.0013794914120808244,
-      "learning_rate": 0.0001999941473155228,
-      "loss": 46.0,
-      "step": 45053
-    },
-    {
-      "epoch": 3.444692929640461,
-      "grad_norm": 0.0005842077080160379,
-      "learning_rate": 0.00019999414705565145,
-      "loss": 46.0,
-      "step": 45054
-    },
-    {
-      "epoch": 3.4447693866238507,
-      "grad_norm": 0.0008987685432657599,
-      "learning_rate": 0.00019999414679577434,
-      "loss": 46.0,
-      "step": 45055
-    },
-    {
-      "epoch": 3.4448458436072404,
-      "grad_norm": 0.0025070877745747566,
-      "learning_rate": 0.0001999941465358915,
-      "loss": 46.0,
-      "step": 45056
-    },
-    {
-      "epoch": 3.44492230059063,
-      "grad_norm": 0.0022104973904788494,
-      "learning_rate": 0.00019999414627600284,
-      "loss": 46.0,
-      "step": 45057
-    },
-    {
-      "epoch": 3.44499875757402,
-      "grad_norm": 0.0014750424306839705,
-      "learning_rate": 0.0001999941460161084,
-      "loss": 46.0,
-      "step": 45058
-    },
-    {
-      "epoch": 3.4450752145574097,
-      "grad_norm": 0.002425615908578038,
-      "learning_rate": 0.00019999414575620823,
-      "loss": 46.0,
-      "step": 45059
-    },
-    {
-      "epoch": 3.4451516715407995,
-      "grad_norm": 0.0036413762718439102,
-      "learning_rate": 0.00019999414549630226,
-      "loss": 46.0,
-      "step": 45060
-    },
-    {
-      "epoch": 3.4452281285241892,
-      "grad_norm": 0.0014508320018649101,
-      "learning_rate": 0.00019999414523639054,
-      "loss": 46.0,
-      "step": 45061
-    },
-    {
-      "epoch": 3.4453045855075786,
-      "grad_norm": 0.000787628348916769,
-      "learning_rate": 0.00019999414497647307,
-      "loss": 46.0,
-      "step": 45062
-    },
-    {
-      "epoch": 3.4453810424909683,
-      "grad_norm": 0.0011673078406602144,
-      "learning_rate": 0.0001999941447165498,
-      "loss": 46.0,
-      "step": 45063
-    },
-    {
-      "epoch": 3.445457499474358,
-      "grad_norm": 0.0069952961057424545,
-      "learning_rate": 0.00019999414445662076,
-      "loss": 46.0,
-      "step": 45064
-    },
-    {
-      "epoch": 3.445533956457748,
-      "grad_norm": 0.0029303813353180885,
-      "learning_rate": 0.00019999414419668597,
-      "loss": 46.0,
-      "step": 45065
-    },
-    {
-      "epoch": 3.4456104134411376,
-      "grad_norm": 0.0021008700132369995,
-      "learning_rate": 0.0001999941439367454,
-      "loss": 46.0,
-      "step": 45066
-    },
-    {
-      "epoch": 3.4456868704245274,
-      "grad_norm": 0.0029025902040302753,
-      "learning_rate": 0.00019999414367679908,
-      "loss": 46.0,
-      "step": 45067
-    },
-    {
-      "epoch": 3.445763327407917,
-      "grad_norm": 0.0013081106590107083,
-      "learning_rate": 0.00019999414341684694,
-      "loss": 46.0,
-      "step": 45068
-    },
-    {
-      "epoch": 3.445839784391307,
-      "grad_norm": 0.0017352878348901868,
-      "learning_rate": 0.0001999941431568891,
-      "loss": 46.0,
-      "step": 45069
-    },
-    {
-      "epoch": 3.4459162413746967,
-      "grad_norm": 0.0005688758683390915,
-      "learning_rate": 0.00019999414289692544,
-      "loss": 46.0,
-      "step": 45070
-    },
-    {
-      "epoch": 3.4459926983580864,
-      "grad_norm": 0.001002923701889813,
-      "learning_rate": 0.00019999414263695604,
-      "loss": 46.0,
-      "step": 45071
-    },
-    {
-      "epoch": 3.446069155341476,
-      "grad_norm": 0.0022857871372252703,
-      "learning_rate": 0.00019999414237698083,
-      "loss": 46.0,
-      "step": 45072
-    },
-    {
-      "epoch": 3.446145612324866,
-      "grad_norm": 0.00481349928304553,
-      "learning_rate": 0.0001999941421169999,
-      "loss": 46.0,
-      "step": 45073
-    },
-    {
-      "epoch": 3.4462220693082553,
-      "grad_norm": 0.0040299552492797375,
-      "learning_rate": 0.00019999414185701317,
-      "loss": 46.0,
-      "step": 45074
-    },
-    {
-      "epoch": 3.446298526291645,
-      "grad_norm": 0.0024729405995458364,
-      "learning_rate": 0.00019999414159702067,
-      "loss": 46.0,
-      "step": 45075
-    },
-    {
-      "epoch": 3.4463749832750348,
-      "grad_norm": 0.005155516788363457,
-      "learning_rate": 0.0001999941413370224,
-      "loss": 46.0,
-      "step": 45076
-    },
-    {
-      "epoch": 3.4464514402584245,
-      "grad_norm": 0.0011493430938571692,
-      "learning_rate": 0.0001999941410770184,
-      "loss": 46.0,
-      "step": 45077
-    },
-    {
-      "epoch": 3.4465278972418143,
-      "grad_norm": 0.00148152990732342,
-      "learning_rate": 0.0001999941408170086,
-      "loss": 46.0,
-      "step": 45078
-    },
-    {
-      "epoch": 3.446604354225204,
-      "grad_norm": 0.014683982357382774,
-      "learning_rate": 0.00019999414055699304,
-      "loss": 46.0,
-      "step": 45079
-    },
-    {
-      "epoch": 3.446680811208594,
-      "grad_norm": 0.0012630947167053819,
-      "learning_rate": 0.0001999941402969717,
-      "loss": 46.0,
-      "step": 45080
-    },
-    {
-      "epoch": 3.4467572681919836,
-      "grad_norm": 0.0008803530363366008,
-      "learning_rate": 0.0001999941400369446,
-      "loss": 46.0,
-      "step": 45081
-    },
-    {
-      "epoch": 3.4468337251753733,
-      "grad_norm": 0.003138320753350854,
-      "learning_rate": 0.0001999941397769117,
-      "loss": 46.0,
-      "step": 45082
-    },
-    {
-      "epoch": 3.446910182158763,
-      "grad_norm": 0.003982197493314743,
-      "learning_rate": 0.0001999941395168731,
-      "loss": 46.0,
-      "step": 45083
-    },
-    {
-      "epoch": 3.4469866391421524,
-      "grad_norm": 0.0015992287080734968,
-      "learning_rate": 0.00019999413925682867,
-      "loss": 46.0,
-      "step": 45084
-    },
-    {
-      "epoch": 3.447063096125542,
-      "grad_norm": 0.0026897152420133352,
-      "learning_rate": 0.00019999413899677847,
-      "loss": 46.0,
-      "step": 45085
-    },
-    {
-      "epoch": 3.447139553108932,
-      "grad_norm": 0.0012032792437821627,
-      "learning_rate": 0.00019999413873672252,
-      "loss": 46.0,
-      "step": 45086
-    },
-    {
-      "epoch": 3.4472160100923217,
-      "grad_norm": 0.0053362417966127396,
-      "learning_rate": 0.00019999413847666083,
-      "loss": 46.0,
-      "step": 45087
-    },
-    {
-      "epoch": 3.4472924670757115,
-      "grad_norm": 0.001471990835852921,
-      "learning_rate": 0.00019999413821659333,
-      "loss": 46.0,
-      "step": 45088
-    },
-    {
-      "epoch": 3.4473689240591012,
-      "grad_norm": 0.0010681929998099804,
-      "learning_rate": 0.00019999413795652006,
-      "loss": 46.0,
-      "step": 45089
-    },
-    {
-      "epoch": 3.447445381042491,
-      "grad_norm": 0.0019076806493103504,
-      "learning_rate": 0.00019999413769644105,
-      "loss": 46.0,
-      "step": 45090
-    },
-    {
-      "epoch": 3.4475218380258807,
-      "grad_norm": 0.0026488739531487226,
-      "learning_rate": 0.00019999413743635624,
-      "loss": 46.0,
-      "step": 45091
-    },
-    {
-      "epoch": 3.4475982950092705,
-      "grad_norm": 0.0008688176749274135,
-      "learning_rate": 0.0001999941371762657,
-      "loss": 46.0,
-      "step": 45092
-    },
-    {
-      "epoch": 3.4476747519926603,
-      "grad_norm": 0.0009435151587240398,
-      "learning_rate": 0.00019999413691616934,
-      "loss": 46.0,
-      "step": 45093
-    },
-    {
-      "epoch": 3.44775120897605,
-      "grad_norm": 0.0013512911973521113,
-      "learning_rate": 0.00019999413665606724,
-      "loss": 46.0,
-      "step": 45094
-    },
-    {
-      "epoch": 3.44782766595944,
-      "grad_norm": 0.0019017241429537535,
-      "learning_rate": 0.00019999413639595936,
-      "loss": 46.0,
-      "step": 45095
-    },
-    {
-      "epoch": 3.447904122942829,
-      "grad_norm": 0.0019924594089388847,
-      "learning_rate": 0.00019999413613584573,
-      "loss": 46.0,
-      "step": 45096
-    },
-    {
-      "epoch": 3.447980579926219,
-      "grad_norm": 0.0013535883044824004,
-      "learning_rate": 0.00019999413587572633,
-      "loss": 46.0,
-      "step": 45097
-    },
-    {
-      "epoch": 3.4480570369096086,
-      "grad_norm": 0.002430856693536043,
-      "learning_rate": 0.00019999413561560113,
-      "loss": 46.0,
-      "step": 45098
-    },
-    {
-      "epoch": 3.4481334938929984,
-      "grad_norm": 0.002370377304032445,
-      "learning_rate": 0.00019999413535547018,
-      "loss": 46.0,
-      "step": 45099
-    },
-    {
-      "epoch": 3.448209950876388,
-      "grad_norm": 0.0012367168674245477,
-      "learning_rate": 0.00019999413509533346,
-      "loss": 46.0,
-      "step": 45100
-    },
-    {
-      "epoch": 3.448286407859778,
-      "grad_norm": 0.00196105451323092,
-      "learning_rate": 0.000199994134835191,
-      "loss": 46.0,
-      "step": 45101
-    },
-    {
-      "epoch": 3.4483628648431677,
-      "grad_norm": 0.0034136022441089153,
-      "learning_rate": 0.00019999413457504273,
-      "loss": 46.0,
-      "step": 45102
-    },
-    {
-      "epoch": 3.4484393218265574,
-      "grad_norm": 0.0010997152421623468,
-      "learning_rate": 0.0001999941343148887,
-      "loss": 46.0,
-      "step": 45103
-    },
-    {
-      "epoch": 3.448515778809947,
-      "grad_norm": 0.002138302428647876,
-      "learning_rate": 0.0001999941340547289,
-      "loss": 46.0,
-      "step": 45104
-    },
-    {
-      "epoch": 3.448592235793337,
-      "grad_norm": 0.0025186294224113226,
-      "learning_rate": 0.00019999413379456334,
-      "loss": 46.0,
-      "step": 45105
-    },
-    {
-      "epoch": 3.4486686927767263,
-      "grad_norm": 0.0023291988763958216,
-      "learning_rate": 0.000199994133534392,
-      "loss": 46.0,
-      "step": 45106
-    },
-    {
-      "epoch": 3.448745149760116,
-      "grad_norm": 0.0024258801713585854,
-      "learning_rate": 0.00019999413327421493,
-      "loss": 46.0,
-      "step": 45107
-    },
-    {
-      "epoch": 3.448821606743506,
-      "grad_norm": 0.0016182144172489643,
-      "learning_rate": 0.00019999413301403202,
-      "loss": 46.0,
-      "step": 45108
-    },
-    {
-      "epoch": 3.4488980637268956,
-      "grad_norm": 0.002611712086945772,
-      "learning_rate": 0.0001999941327538434,
-      "loss": 46.0,
-      "step": 45109
-    },
-    {
-      "epoch": 3.4489745207102853,
-      "grad_norm": 0.001078123808838427,
-      "learning_rate": 0.000199994132493649,
-      "loss": 46.0,
-      "step": 45110
-    },
-    {
-      "epoch": 3.449050977693675,
-      "grad_norm": 0.0024431785568594933,
-      "learning_rate": 0.00019999413223344882,
-      "loss": 46.0,
-      "step": 45111
-    },
-    {
-      "epoch": 3.449127434677065,
-      "grad_norm": 0.0022236232180148363,
-      "learning_rate": 0.00019999413197324285,
-      "loss": 46.0,
-      "step": 45112
-    },
-    {
-      "epoch": 3.4492038916604546,
-      "grad_norm": 0.00159554707352072,
-      "learning_rate": 0.00019999413171303116,
-      "loss": 46.0,
-      "step": 45113
-    },
-    {
-      "epoch": 3.4492803486438444,
-      "grad_norm": 0.0010796718997880816,
-      "learning_rate": 0.00019999413145281366,
-      "loss": 46.0,
-      "step": 45114
-    },
-    {
-      "epoch": 3.449356805627234,
-      "grad_norm": 0.0032348039094358683,
-      "learning_rate": 0.0001999941311925904,
-      "loss": 46.0,
-      "step": 45115
-    },
-    {
-      "epoch": 3.449433262610624,
-      "grad_norm": 0.001142166554927826,
-      "learning_rate": 0.00019999413093236138,
-      "loss": 46.0,
-      "step": 45116
-    },
-    {
-      "epoch": 3.4495097195940136,
-      "grad_norm": 0.0012506458442658186,
-      "learning_rate": 0.0001999941306721266,
-      "loss": 46.0,
-      "step": 45117
-    },
-    {
-      "epoch": 3.449586176577403,
-      "grad_norm": 0.0032766899093985558,
-      "learning_rate": 0.000199994130411886,
-      "loss": 46.0,
-      "step": 45118
-    },
-    {
-      "epoch": 3.4496626335607927,
-      "grad_norm": 0.0034516965970396996,
-      "learning_rate": 0.0001999941301516397,
-      "loss": 46.0,
-      "step": 45119
-    },
-    {
-      "epoch": 3.4497390905441825,
-      "grad_norm": 0.005322735290974379,
-      "learning_rate": 0.0001999941298913876,
-      "loss": 46.0,
-      "step": 45120
-    },
-    {
-      "epoch": 3.4498155475275722,
-      "grad_norm": 0.0027151722460985184,
-      "learning_rate": 0.00019999412963112975,
-      "loss": 46.0,
-      "step": 45121
-    },
-    {
-      "epoch": 3.449892004510962,
-      "grad_norm": 0.00263771740719676,
-      "learning_rate": 0.00019999412937086607,
-      "loss": 46.0,
-      "step": 45122
-    },
-    {
-      "epoch": 3.4499684614943518,
-      "grad_norm": 0.0011744481744244695,
-      "learning_rate": 0.00019999412911059667,
-      "loss": 46.0,
-      "step": 45123
-    },
-    {
-      "epoch": 3.4500449184777415,
-      "grad_norm": 0.003663536161184311,
-      "learning_rate": 0.0001999941288503215,
-      "loss": 46.0,
-      "step": 45124
-    },
-    {
-      "epoch": 3.4501213754611313,
-      "grad_norm": 0.0019422996556386352,
-      "learning_rate": 0.00019999412859004058,
-      "loss": 46.0,
-      "step": 45125
-    },
-    {
-      "epoch": 3.450197832444521,
-      "grad_norm": 0.0009640787611715496,
-      "learning_rate": 0.00019999412832975386,
-      "loss": 46.0,
-      "step": 45126
-    },
-    {
-      "epoch": 3.450274289427911,
-      "grad_norm": 0.0014606183394789696,
-      "learning_rate": 0.00019999412806946136,
-      "loss": 46.0,
-      "step": 45127
-    },
-    {
-      "epoch": 3.4503507464113,
-      "grad_norm": 0.020403439179062843,
-      "learning_rate": 0.00019999412780916312,
-      "loss": 46.0,
-      "step": 45128
-    },
-    {
-      "epoch": 3.45042720339469,
-      "grad_norm": 0.0023286258801817894,
-      "learning_rate": 0.00019999412754885908,
-      "loss": 46.0,
-      "step": 45129
-    },
-    {
-      "epoch": 3.4505036603780797,
-      "grad_norm": 0.00200326438061893,
-      "learning_rate": 0.0001999941272885493,
-      "loss": 46.0,
-      "step": 45130
-    },
-    {
-      "epoch": 3.4505801173614694,
-      "grad_norm": 0.002215000567957759,
-      "learning_rate": 0.0001999941270282337,
-      "loss": 46.0,
-      "step": 45131
-    },
-    {
-      "epoch": 3.450656574344859,
-      "grad_norm": 0.0033983380999416113,
-      "learning_rate": 0.0001999941267679124,
-      "loss": 46.0,
-      "step": 45132
-    },
-    {
-      "epoch": 3.450733031328249,
-      "grad_norm": 0.002194545930251479,
-      "learning_rate": 0.0001999941265075853,
-      "loss": 46.0,
-      "step": 45133
-    },
-    {
-      "epoch": 3.4508094883116387,
-      "grad_norm": 0.0012085940688848495,
-      "learning_rate": 0.00019999412624725245,
-      "loss": 46.0,
-      "step": 45134
-    },
-    {
-      "epoch": 3.4508859452950285,
-      "grad_norm": 0.00337292836047709,
-      "learning_rate": 0.0001999941259869138,
-      "loss": 46.0,
-      "step": 45135
-    },
-    {
-      "epoch": 3.450962402278418,
-      "grad_norm": 0.001386150368489325,
-      "learning_rate": 0.0001999941257265694,
-      "loss": 46.0,
-      "step": 45136
-    },
-    {
-      "epoch": 3.451038859261808,
-      "grad_norm": 0.004297069739550352,
-      "learning_rate": 0.0001999941254662192,
-      "loss": 46.0,
-      "step": 45137
-    },
-    {
-      "epoch": 3.4511153162451977,
-      "grad_norm": 0.002161393640562892,
-      "learning_rate": 0.00019999412520586328,
-      "loss": 46.0,
-      "step": 45138
-    },
-    {
-      "epoch": 3.4511917732285875,
-      "grad_norm": 0.0014598227571696043,
-      "learning_rate": 0.00019999412494550156,
-      "loss": 46.0,
-      "step": 45139
-    },
-    {
-      "epoch": 3.451268230211977,
-      "grad_norm": 0.0023323611821979284,
-      "learning_rate": 0.00019999412468513407,
-      "loss": 46.0,
-      "step": 45140
-    },
-    {
-      "epoch": 3.4513446871953666,
-      "grad_norm": 0.00313792796805501,
-      "learning_rate": 0.00019999412442476083,
-      "loss": 46.0,
-      "step": 45141
-    },
-    {
-      "epoch": 3.4514211441787563,
-      "grad_norm": 0.002790069440379739,
-      "learning_rate": 0.00019999412416438182,
-      "loss": 46.0,
-      "step": 45142
-    },
-    {
-      "epoch": 3.451497601162146,
-      "grad_norm": 0.004355615470558405,
-      "learning_rate": 0.00019999412390399703,
-      "loss": 46.0,
-      "step": 45143
-    },
-    {
-      "epoch": 3.451574058145536,
-      "grad_norm": 0.0027639043983072042,
-      "learning_rate": 0.00019999412364360644,
-      "loss": 46.0,
-      "step": 45144
-    },
-    {
-      "epoch": 3.4516505151289256,
-      "grad_norm": 0.0015308116562664509,
-      "learning_rate": 0.00019999412338321014,
-      "loss": 46.0,
-      "step": 45145
-    },
-    {
-      "epoch": 3.4517269721123154,
-      "grad_norm": 0.0014943835558369756,
-      "learning_rate": 0.00019999412312280803,
-      "loss": 46.0,
-      "step": 45146
-    },
-    {
-      "epoch": 3.451803429095705,
-      "grad_norm": 0.0018667022231966257,
-      "learning_rate": 0.00019999412286240018,
-      "loss": 46.0,
-      "step": 45147
-    },
-    {
-      "epoch": 3.451879886079095,
-      "grad_norm": 0.004269406199455261,
-      "learning_rate": 0.00019999412260198653,
-      "loss": 46.0,
-      "step": 45148
-    },
-    {
-      "epoch": 3.4519563430624842,
-      "grad_norm": 0.0013109026476740837,
-      "learning_rate": 0.0001999941223415671,
-      "loss": 46.0,
-      "step": 45149
-    },
-    {
-      "epoch": 3.452032800045874,
-      "grad_norm": 0.002175177214667201,
-      "learning_rate": 0.00019999412208114196,
-      "loss": 46.0,
-      "step": 45150
-    },
-    {
-      "epoch": 3.4521092570292637,
-      "grad_norm": 0.003254286479204893,
-      "learning_rate": 0.000199994121820711,
-      "loss": 46.0,
-      "step": 45151
-    },
-    {
-      "epoch": 3.4521857140126535,
-      "grad_norm": 0.0011905268765985966,
-      "learning_rate": 0.0001999941215602743,
-      "loss": 46.0,
-      "step": 45152
-    },
-    {
-      "epoch": 3.4522621709960433,
-      "grad_norm": 0.0011312214192003012,
-      "learning_rate": 0.00019999412129983183,
-      "loss": 46.0,
-      "step": 45153
-    },
-    {
-      "epoch": 3.452338627979433,
-      "grad_norm": 0.004624643363058567,
-      "learning_rate": 0.00019999412103938356,
-      "loss": 46.0,
-      "step": 45154
-    },
-    {
-      "epoch": 3.452415084962823,
-      "grad_norm": 0.0022143993992358446,
-      "learning_rate": 0.00019999412077892955,
-      "loss": 46.0,
-      "step": 45155
-    },
-    {
-      "epoch": 3.4524915419462125,
-      "grad_norm": 0.0016871481202542782,
-      "learning_rate": 0.00019999412051846974,
-      "loss": 46.0,
-      "step": 45156
-    },
-    {
-      "epoch": 3.4525679989296023,
-      "grad_norm": 0.0007682691211812198,
-      "learning_rate": 0.00019999412025800418,
-      "loss": 46.0,
-      "step": 45157
-    },
-    {
-      "epoch": 3.452644455912992,
-      "grad_norm": 0.0018926331540569663,
-      "learning_rate": 0.00019999411999753288,
-      "loss": 46.0,
-      "step": 45158
-    },
-    {
-      "epoch": 3.452720912896382,
-      "grad_norm": 0.0016012253472581506,
-      "learning_rate": 0.0001999941197370558,
-      "loss": 46.0,
-      "step": 45159
-    },
-    {
-      "epoch": 3.4527973698797716,
-      "grad_norm": 0.0035411352291703224,
-      "learning_rate": 0.00019999411947657292,
-      "loss": 46.0,
-      "step": 45160
-    },
-    {
-      "epoch": 3.4528738268631614,
-      "grad_norm": 0.0017030660528689623,
-      "learning_rate": 0.0001999941192160843,
-      "loss": 46.0,
-      "step": 45161
-    },
-    {
-      "epoch": 3.4529502838465507,
-      "grad_norm": 0.003867089981213212,
-      "learning_rate": 0.0001999941189555899,
-      "loss": 46.0,
-      "step": 45162
-    },
-    {
-      "epoch": 3.4530267408299404,
-      "grad_norm": 0.001563460100442171,
-      "learning_rate": 0.0001999941186950897,
-      "loss": 46.0,
-      "step": 45163
-    },
-    {
-      "epoch": 3.45310319781333,
-      "grad_norm": 0.0018809012835845351,
-      "learning_rate": 0.00019999411843458375,
-      "loss": 46.0,
-      "step": 45164
-    },
-    {
-      "epoch": 3.45317965479672,
-      "grad_norm": 0.0022518320474773645,
-      "learning_rate": 0.00019999411817407206,
-      "loss": 46.0,
-      "step": 45165
-    },
-    {
-      "epoch": 3.4532561117801097,
-      "grad_norm": 0.0012009054189547896,
-      "learning_rate": 0.00019999411791355457,
-      "loss": 46.0,
-      "step": 45166
-    },
-    {
-      "epoch": 3.4533325687634995,
-      "grad_norm": 0.0023504241835325956,
-      "learning_rate": 0.00019999411765303133,
-      "loss": 46.0,
-      "step": 45167
-    },
-    {
-      "epoch": 3.4534090257468892,
-      "grad_norm": 0.008862879127264023,
-      "learning_rate": 0.00019999411739250232,
-      "loss": 46.0,
-      "step": 45168
-    },
-    {
-      "epoch": 3.453485482730279,
-      "grad_norm": 0.002991313813254237,
-      "learning_rate": 0.0001999941171319675,
-      "loss": 46.0,
-      "step": 45169
-    },
-    {
-      "epoch": 3.4535619397136688,
-      "grad_norm": 0.0015789506724104285,
-      "learning_rate": 0.00019999411687142695,
-      "loss": 46.0,
-      "step": 45170
-    },
-    {
-      "epoch": 3.453638396697058,
-      "grad_norm": 0.004376812372356653,
-      "learning_rate": 0.00019999411661088064,
-      "loss": 46.0,
-      "step": 45171
-    },
-    {
-      "epoch": 3.453714853680448,
-      "grad_norm": 0.0005093111540190876,
-      "learning_rate": 0.00019999411635032854,
-      "loss": 46.0,
-      "step": 45172
-    },
-    {
-      "epoch": 3.4537913106638376,
-      "grad_norm": 0.0018366910517215729,
-      "learning_rate": 0.0001999941160897707,
-      "loss": 46.0,
-      "step": 45173
-    },
-    {
-      "epoch": 3.4538677676472274,
-      "grad_norm": 0.0034219748340547085,
-      "learning_rate": 0.00019999411582920703,
-      "loss": 46.0,
-      "step": 45174
-    },
-    {
-      "epoch": 3.453944224630617,
-      "grad_norm": 0.0029362551867961884,
-      "learning_rate": 0.00019999411556863766,
-      "loss": 46.0,
-      "step": 45175
-    },
-    {
-      "epoch": 3.454020681614007,
-      "grad_norm": 0.0019654249772429466,
-      "learning_rate": 0.0001999941153080625,
-      "loss": 46.0,
-      "step": 45176
-    },
-    {
-      "epoch": 3.4540971385973966,
-      "grad_norm": 0.0031968653202056885,
-      "learning_rate": 0.00019999411504748152,
-      "loss": 46.0,
-      "step": 45177
-    },
-    {
-      "epoch": 3.4541735955807864,
-      "grad_norm": 0.004857442807406187,
-      "learning_rate": 0.00019999411478689483,
-      "loss": 46.0,
-      "step": 45178
-    },
-    {
-      "epoch": 3.454250052564176,
-      "grad_norm": 0.003389917779713869,
-      "learning_rate": 0.00019999411452630236,
-      "loss": 46.0,
-      "step": 45179
-    },
-    {
-      "epoch": 3.454326509547566,
-      "grad_norm": 0.0029015298932790756,
-      "learning_rate": 0.0001999941142657041,
-      "loss": 46.0,
-      "step": 45180
-    },
-    {
-      "epoch": 3.4544029665309557,
-      "grad_norm": 0.0019460406620055437,
-      "learning_rate": 0.0001999941140051001,
-      "loss": 46.0,
-      "step": 45181
-    },
-    {
-      "epoch": 3.4544794235143454,
-      "grad_norm": 0.003322836710140109,
-      "learning_rate": 0.0001999941137444903,
-      "loss": 46.0,
-      "step": 45182
-    },
-    {
-      "epoch": 3.454555880497735,
-      "grad_norm": 0.001839798642322421,
-      "learning_rate": 0.00019999411348387475,
-      "loss": 46.0,
-      "step": 45183
-    },
-    {
-      "epoch": 3.4546323374811245,
-      "grad_norm": 0.0019660566467791796,
-      "learning_rate": 0.00019999411322325341,
-      "loss": 46.0,
-      "step": 45184
-    },
-    {
-      "epoch": 3.4547087944645143,
-      "grad_norm": 0.00232500908896327,
-      "learning_rate": 0.00019999411296262634,
-      "loss": 46.0,
-      "step": 45185
-    },
-    {
-      "epoch": 3.454785251447904,
-      "grad_norm": 0.00673377001658082,
-      "learning_rate": 0.00019999411270199349,
-      "loss": 46.0,
-      "step": 45186
-    },
-    {
-      "epoch": 3.454861708431294,
-      "grad_norm": 0.007242770865559578,
-      "learning_rate": 0.00019999411244135483,
-      "loss": 46.0,
-      "step": 45187
-    },
-    {
-      "epoch": 3.4549381654146836,
-      "grad_norm": 0.0025004667695611715,
-      "learning_rate": 0.00019999411218071044,
-      "loss": 46.0,
-      "step": 45188
-    },
-    {
-      "epoch": 3.4550146223980733,
-      "grad_norm": 0.002810709411278367,
-      "learning_rate": 0.00019999411192006027,
-      "loss": 46.0,
-      "step": 45189
-    },
-    {
-      "epoch": 3.455091079381463,
-      "grad_norm": 0.0016391208628192544,
-      "learning_rate": 0.00019999411165940435,
-      "loss": 46.0,
-      "step": 45190
-    },
-    {
-      "epoch": 3.455167536364853,
-      "grad_norm": 0.0008896089857444167,
-      "learning_rate": 0.00019999411139874263,
-      "loss": 46.0,
-      "step": 45191
-    },
-    {
-      "epoch": 3.4552439933482426,
-      "grad_norm": 0.0033274965826421976,
-      "learning_rate": 0.00019999411113807517,
-      "loss": 46.0,
-      "step": 45192
-    },
-    {
-      "epoch": 3.455320450331632,
-      "grad_norm": 0.0012692732270807028,
-      "learning_rate": 0.0001999941108774019,
-      "loss": 46.0,
-      "step": 45193
-    },
-    {
-      "epoch": 3.4553969073150217,
-      "grad_norm": 0.004666381515562534,
-      "learning_rate": 0.0001999941106167229,
-      "loss": 46.0,
-      "step": 45194
-    },
-    {
-      "epoch": 3.4554733642984115,
-      "grad_norm": 0.0026167805772274733,
-      "learning_rate": 0.0001999941103560381,
-      "loss": 46.0,
-      "step": 45195
-    },
-    {
-      "epoch": 3.455549821281801,
-      "grad_norm": 0.004648515023291111,
-      "learning_rate": 0.00019999411009534758,
-      "loss": 46.0,
-      "step": 45196
-    },
-    {
-      "epoch": 3.455626278265191,
-      "grad_norm": 0.0026463312096893787,
-      "learning_rate": 0.00019999410983465124,
-      "loss": 46.0,
-      "step": 45197
-    },
-    {
-      "epoch": 3.4557027352485807,
-      "grad_norm": 0.0025514094159007072,
-      "learning_rate": 0.00019999410957394914,
-      "loss": 46.0,
-      "step": 45198
-    },
-    {
-      "epoch": 3.4557791922319705,
-      "grad_norm": 0.0013882337370887399,
-      "learning_rate": 0.00019999410931324126,
-      "loss": 46.0,
-      "step": 45199
-    },
-    {
-      "epoch": 3.4558556492153603,
-      "grad_norm": 0.0010888864053413272,
-      "learning_rate": 0.00019999410905252764,
-      "loss": 46.0,
-      "step": 45200
-    },
-    {
-      "epoch": 3.45593210619875,
-      "grad_norm": 0.0008232976542785764,
-      "learning_rate": 0.00019999410879180824,
-      "loss": 46.0,
-      "step": 45201
-    },
-    {
-      "epoch": 3.45600856318214,
-      "grad_norm": 0.0022681825794279575,
-      "learning_rate": 0.00019999410853108307,
-      "loss": 46.0,
-      "step": 45202
-    },
-    {
-      "epoch": 3.4560850201655295,
-      "grad_norm": 0.0019660198595374823,
-      "learning_rate": 0.00019999410827035215,
-      "loss": 46.0,
-      "step": 45203
-    },
-    {
-      "epoch": 3.4561614771489193,
-      "grad_norm": 0.0011629970977082849,
-      "learning_rate": 0.00019999410800961544,
-      "loss": 46.0,
-      "step": 45204
-    },
-    {
-      "epoch": 3.4562379341323086,
-      "grad_norm": 0.0024486305192112923,
-      "learning_rate": 0.00019999410774887294,
-      "loss": 46.0,
-      "step": 45205
-    },
-    {
-      "epoch": 3.4563143911156984,
-      "grad_norm": 0.0018469431670382619,
-      "learning_rate": 0.00019999410748812473,
-      "loss": 46.0,
-      "step": 45206
-    },
-    {
-      "epoch": 3.456390848099088,
-      "grad_norm": 0.0025218394584953785,
-      "learning_rate": 0.0001999941072273707,
-      "loss": 46.0,
-      "step": 45207
-    },
-    {
-      "epoch": 3.456467305082478,
-      "grad_norm": 0.0036653392016887665,
-      "learning_rate": 0.0001999941069666109,
-      "loss": 46.0,
-      "step": 45208
-    },
-    {
-      "epoch": 3.4565437620658677,
-      "grad_norm": 0.001371211837977171,
-      "learning_rate": 0.00019999410670584538,
-      "loss": 46.0,
-      "step": 45209
-    },
-    {
-      "epoch": 3.4566202190492574,
-      "grad_norm": 0.011854550801217556,
-      "learning_rate": 0.00019999410644507405,
-      "loss": 46.0,
-      "step": 45210
-    },
-    {
-      "epoch": 3.456696676032647,
-      "grad_norm": 0.001393481157720089,
-      "learning_rate": 0.00019999410618429697,
-      "loss": 46.0,
-      "step": 45211
-    },
-    {
-      "epoch": 3.456773133016037,
-      "grad_norm": 0.0008507466409355402,
-      "learning_rate": 0.0001999941059235141,
-      "loss": 46.0,
-      "step": 45212
-    },
-    {
-      "epoch": 3.4568495899994267,
-      "grad_norm": 0.0015412586508318782,
-      "learning_rate": 0.00019999410566272547,
-      "loss": 46.0,
-      "step": 45213
-    },
-    {
-      "epoch": 3.4569260469828165,
-      "grad_norm": 0.004121486097574234,
-      "learning_rate": 0.00019999410540193108,
-      "loss": 46.0,
-      "step": 45214
-    },
-    {
-      "epoch": 3.457002503966206,
-      "grad_norm": 0.006144287530332804,
-      "learning_rate": 0.00019999410514113093,
-      "loss": 46.0,
-      "step": 45215
-    },
-    {
-      "epoch": 3.4570789609495955,
-      "grad_norm": 0.003407243639230728,
-      "learning_rate": 0.00019999410488032496,
-      "loss": 46.0,
-      "step": 45216
-    },
-    {
-      "epoch": 3.4571554179329853,
-      "grad_norm": 0.002048554364591837,
-      "learning_rate": 0.0001999941046195133,
-      "loss": 46.0,
-      "step": 45217
-    },
-    {
-      "epoch": 3.457231874916375,
-      "grad_norm": 0.004459748975932598,
-      "learning_rate": 0.00019999410435869578,
-      "loss": 46.0,
-      "step": 45218
-    },
-    {
-      "epoch": 3.457308331899765,
-      "grad_norm": 0.002758754650130868,
-      "learning_rate": 0.00019999410409787257,
-      "loss": 46.0,
-      "step": 45219
-    },
-    {
-      "epoch": 3.4573847888831546,
-      "grad_norm": 0.004380478523671627,
-      "learning_rate": 0.00019999410383704356,
-      "loss": 46.0,
-      "step": 45220
-    },
-    {
-      "epoch": 3.4574612458665444,
-      "grad_norm": 0.003762703388929367,
-      "learning_rate": 0.00019999410357620878,
-      "loss": 46.0,
-      "step": 45221
-    },
-    {
-      "epoch": 3.457537702849934,
-      "grad_norm": 0.00159950100351125,
-      "learning_rate": 0.00019999410331536822,
-      "loss": 46.0,
-      "step": 45222
-    },
-    {
-      "epoch": 3.457614159833324,
-      "grad_norm": 0.0015217358013615012,
-      "learning_rate": 0.00019999410305452192,
-      "loss": 46.0,
-      "step": 45223
-    },
-    {
-      "epoch": 3.4576906168167136,
-      "grad_norm": 0.003450914518907666,
-      "learning_rate": 0.0001999941027936698,
-      "loss": 46.0,
-      "step": 45224
-    },
-    {
-      "epoch": 3.4577670738001034,
-      "grad_norm": 0.001059976639226079,
-      "learning_rate": 0.00019999410253281194,
-      "loss": 46.0,
-      "step": 45225
-    },
-    {
-      "epoch": 3.457843530783493,
-      "grad_norm": 0.0013654757058247924,
-      "learning_rate": 0.0001999941022719483,
-      "loss": 46.0,
-      "step": 45226
-    },
-    {
-      "epoch": 3.4579199877668825,
-      "grad_norm": 0.001975161489099264,
-      "learning_rate": 0.00019999410201107894,
-      "loss": 46.0,
-      "step": 45227
-    },
-    {
-      "epoch": 3.4579964447502722,
-      "grad_norm": 0.001725782873108983,
-      "learning_rate": 0.00019999410175020377,
-      "loss": 46.0,
-      "step": 45228
-    },
-    {
-      "epoch": 3.458072901733662,
-      "grad_norm": 0.0036536534316837788,
-      "learning_rate": 0.00019999410148932283,
-      "loss": 46.0,
-      "step": 45229
-    },
-    {
-      "epoch": 3.4581493587170518,
-      "grad_norm": 0.0006217904738150537,
-      "learning_rate": 0.00019999410122843614,
-      "loss": 46.0,
-      "step": 45230
-    },
-    {
-      "epoch": 3.4582258157004415,
-      "grad_norm": 0.0010941765503957868,
-      "learning_rate": 0.00019999410096754365,
-      "loss": 46.0,
-      "step": 45231
-    },
-    {
-      "epoch": 3.4583022726838313,
-      "grad_norm": 0.0036449795588850975,
-      "learning_rate": 0.0001999941007066454,
-      "loss": 46.0,
-      "step": 45232
-    },
-    {
-      "epoch": 3.458378729667221,
-      "grad_norm": 0.0014898300869390368,
-      "learning_rate": 0.0001999941004457414,
-      "loss": 46.0,
-      "step": 45233
-    },
-    {
-      "epoch": 3.458455186650611,
-      "grad_norm": 0.0009278338402509689,
-      "learning_rate": 0.00019999410018483162,
-      "loss": 46.0,
-      "step": 45234
-    },
-    {
-      "epoch": 3.4585316436340006,
-      "grad_norm": 0.002027886686846614,
-      "learning_rate": 0.0001999940999239161,
-      "loss": 46.0,
-      "step": 45235
-    },
-    {
-      "epoch": 3.4586081006173903,
-      "grad_norm": 0.002692680573090911,
-      "learning_rate": 0.00019999409966299476,
-      "loss": 46.0,
-      "step": 45236
-    },
-    {
-      "epoch": 3.4586845576007796,
-      "grad_norm": 0.005479230545461178,
-      "learning_rate": 0.00019999409940206766,
-      "loss": 46.0,
-      "step": 45237
-    },
-    {
-      "epoch": 3.4587610145841694,
-      "grad_norm": 0.0028295456431806087,
-      "learning_rate": 0.00019999409914113483,
-      "loss": 46.0,
-      "step": 45238
-    },
-    {
-      "epoch": 3.458837471567559,
-      "grad_norm": 0.0015259896172210574,
-      "learning_rate": 0.00019999409888019618,
-      "loss": 46.0,
-      "step": 45239
-    },
-    {
-      "epoch": 3.458913928550949,
-      "grad_norm": 0.002326915506273508,
-      "learning_rate": 0.0001999940986192518,
-      "loss": 46.0,
-      "step": 45240
-    },
-    {
-      "epoch": 3.4589903855343387,
-      "grad_norm": 0.00134359672665596,
-      "learning_rate": 0.00019999409835830164,
-      "loss": 46.0,
-      "step": 45241
-    },
-    {
-      "epoch": 3.4590668425177284,
-      "grad_norm": 0.0013130197767168283,
-      "learning_rate": 0.0001999940980973457,
-      "loss": 46.0,
-      "step": 45242
-    },
-    {
-      "epoch": 3.459143299501118,
-      "grad_norm": 0.007666292134672403,
-      "learning_rate": 0.00019999409783638398,
-      "loss": 46.0,
-      "step": 45243
-    },
-    {
-      "epoch": 3.459219756484508,
-      "grad_norm": 0.001448189141228795,
-      "learning_rate": 0.00019999409757541652,
-      "loss": 46.0,
-      "step": 45244
-    },
-    {
-      "epoch": 3.4592962134678977,
-      "grad_norm": 0.002870507538318634,
-      "learning_rate": 0.00019999409731444328,
-      "loss": 46.0,
-      "step": 45245
-    },
-    {
-      "epoch": 3.4593726704512875,
-      "grad_norm": 0.0013945474056527019,
-      "learning_rate": 0.00019999409705346427,
-      "loss": 46.0,
-      "step": 45246
-    },
-    {
-      "epoch": 3.4594491274346773,
-      "grad_norm": 0.0018077908316627145,
-      "learning_rate": 0.00019999409679247952,
-      "loss": 46.0,
-      "step": 45247
-    },
-    {
-      "epoch": 3.459525584418067,
-      "grad_norm": 0.0016248972387984395,
-      "learning_rate": 0.00019999409653148896,
-      "loss": 46.0,
-      "step": 45248
-    },
-    {
-      "epoch": 3.4596020414014563,
-      "grad_norm": 0.00513993576169014,
-      "learning_rate": 0.00019999409627049266,
-      "loss": 46.0,
-      "step": 45249
-    },
-    {
-      "epoch": 3.459678498384846,
-      "grad_norm": 0.0013955950271338224,
-      "learning_rate": 0.00019999409600949056,
-      "loss": 46.0,
-      "step": 45250
-    },
-    {
-      "epoch": 3.459754955368236,
-      "grad_norm": 0.0013859347673133016,
-      "learning_rate": 0.0001999940957484827,
-      "loss": 46.0,
-      "step": 45251
-    },
-    {
-      "epoch": 3.4598314123516256,
-      "grad_norm": 0.0007701015565544367,
-      "learning_rate": 0.00019999409548746908,
-      "loss": 46.0,
-      "step": 45252
-    },
-    {
-      "epoch": 3.4599078693350154,
-      "grad_norm": 0.003868626896291971,
-      "learning_rate": 0.00019999409522644966,
-      "loss": 46.0,
-      "step": 45253
-    },
-    {
-      "epoch": 3.459984326318405,
-      "grad_norm": 0.0006784715224057436,
-      "learning_rate": 0.00019999409496542452,
-      "loss": 46.0,
-      "step": 45254
-    },
-    {
-      "epoch": 3.460060783301795,
-      "grad_norm": 0.002501010661944747,
-      "learning_rate": 0.00019999409470439358,
-      "loss": 46.0,
-      "step": 45255
-    },
-    {
-      "epoch": 3.4601372402851847,
-      "grad_norm": 0.0016842057229951024,
-      "learning_rate": 0.00019999409444335689,
-      "loss": 46.0,
-      "step": 45256
-    },
-    {
-      "epoch": 3.4602136972685744,
-      "grad_norm": 0.0008871675818227232,
-      "learning_rate": 0.00019999409418231442,
-      "loss": 46.0,
-      "step": 45257
-    },
-    {
-      "epoch": 3.460290154251964,
-      "grad_norm": 0.0024440987035632133,
-      "learning_rate": 0.00019999409392126621,
-      "loss": 46.0,
-      "step": 45258
-    },
-    {
-      "epoch": 3.4603666112353535,
-      "grad_norm": 0.0010630963370203972,
-      "learning_rate": 0.00019999409366021218,
-      "loss": 46.0,
-      "step": 45259
-    },
-    {
-      "epoch": 3.4604430682187433,
-      "grad_norm": 0.0008465952123515308,
-      "learning_rate": 0.0001999940933991524,
-      "loss": 46.0,
-      "step": 45260
-    },
-    {
-      "epoch": 3.460519525202133,
-      "grad_norm": 0.0008157572010532022,
-      "learning_rate": 0.00019999409313808687,
-      "loss": 46.0,
-      "step": 45261
-    },
-    {
-      "epoch": 3.460595982185523,
-      "grad_norm": 0.005165203008800745,
-      "learning_rate": 0.00019999409287701556,
-      "loss": 46.0,
-      "step": 45262
-    },
-    {
-      "epoch": 3.4606724391689125,
-      "grad_norm": 0.0008446403662674129,
-      "learning_rate": 0.00019999409261593846,
-      "loss": 46.0,
-      "step": 45263
-    },
-    {
-      "epoch": 3.4607488961523023,
-      "grad_norm": 0.0019177235662937164,
-      "learning_rate": 0.0001999940923548556,
-      "loss": 46.0,
-      "step": 45264
-    },
-    {
-      "epoch": 3.460825353135692,
-      "grad_norm": 0.0043699066154658794,
-      "learning_rate": 0.000199994092093767,
-      "loss": 46.0,
-      "step": 45265
-    },
-    {
-      "epoch": 3.460901810119082,
-      "grad_norm": 0.006610126234591007,
-      "learning_rate": 0.0001999940918326726,
-      "loss": 46.0,
-      "step": 45266
-    },
-    {
-      "epoch": 3.4609782671024716,
-      "grad_norm": 0.0009315164061263204,
-      "learning_rate": 0.00019999409157157242,
-      "loss": 46.0,
-      "step": 45267
-    },
-    {
-      "epoch": 3.4610547240858613,
-      "grad_norm": 0.0020402332302182913,
-      "learning_rate": 0.00019999409131046648,
-      "loss": 46.0,
-      "step": 45268
-    },
-    {
-      "epoch": 3.461131181069251,
-      "grad_norm": 0.0020765983499586582,
-      "learning_rate": 0.00019999409104935482,
-      "loss": 46.0,
-      "step": 45269
-    },
-    {
-      "epoch": 3.461207638052641,
-      "grad_norm": 0.0043333726935088634,
-      "learning_rate": 0.00019999409078823733,
-      "loss": 46.0,
-      "step": 45270
-    },
-    {
-      "epoch": 3.46128409503603,
-      "grad_norm": 0.003468272276222706,
-      "learning_rate": 0.00019999409052711412,
-      "loss": 46.0,
-      "step": 45271
-    },
-    {
-      "epoch": 3.46136055201942,
-      "grad_norm": 0.0020052504260092974,
-      "learning_rate": 0.0001999940902659851,
-      "loss": 46.0,
-      "step": 45272
-    },
-    {
-      "epoch": 3.4614370090028097,
-      "grad_norm": 0.002159254625439644,
-      "learning_rate": 0.00019999409000485033,
-      "loss": 46.0,
-      "step": 45273
-    },
-    {
-      "epoch": 3.4615134659861995,
-      "grad_norm": 0.0027012641075998545,
-      "learning_rate": 0.00019999408974370977,
-      "loss": 46.0,
-      "step": 45274
-    },
-    {
-      "epoch": 3.4615899229695892,
-      "grad_norm": 0.000906950153876096,
-      "learning_rate": 0.00019999408948256347,
-      "loss": 46.0,
-      "step": 45275
-    },
-    {
-      "epoch": 3.461666379952979,
-      "grad_norm": 0.006767289713025093,
-      "learning_rate": 0.0001999940892214114,
-      "loss": 46.0,
-      "step": 45276
-    },
-    {
-      "epoch": 3.4617428369363688,
-      "grad_norm": 0.005225414410233498,
-      "learning_rate": 0.00019999408896025352,
-      "loss": 46.0,
-      "step": 45277
-    },
-    {
-      "epoch": 3.4618192939197585,
-      "grad_norm": 0.0015316924545913935,
-      "learning_rate": 0.0001999940886990899,
-      "loss": 46.0,
-      "step": 45278
-    },
-    {
-      "epoch": 3.4618957509031483,
-      "grad_norm": 0.0021379240788519382,
-      "learning_rate": 0.0001999940884379205,
-      "loss": 46.0,
-      "step": 45279
-    },
-    {
-      "epoch": 3.4619722078865376,
-      "grad_norm": 0.0012888384517282248,
-      "learning_rate": 0.00019999408817674536,
-      "loss": 46.0,
-      "step": 45280
-    },
-    {
-      "epoch": 3.4620486648699274,
-      "grad_norm": 0.0015105732018128037,
-      "learning_rate": 0.00019999408791556445,
-      "loss": 46.0,
-      "step": 45281
-    },
-    {
-      "epoch": 3.462125121853317,
-      "grad_norm": 0.0017428887076675892,
-      "learning_rate": 0.00019999408765437776,
-      "loss": 46.0,
-      "step": 45282
-    },
-    {
-      "epoch": 3.462201578836707,
-      "grad_norm": 0.0018663223600015044,
-      "learning_rate": 0.00019999408739318527,
-      "loss": 46.0,
-      "step": 45283
-    },
-    {
-      "epoch": 3.4622780358200966,
-      "grad_norm": 0.0017414928879588842,
-      "learning_rate": 0.00019999408713198703,
-      "loss": 46.0,
-      "step": 45284
-    },
-    {
-      "epoch": 3.4623544928034864,
-      "grad_norm": 0.0014732241397723556,
-      "learning_rate": 0.00019999408687078302,
-      "loss": 46.0,
-      "step": 45285
-    },
-    {
-      "epoch": 3.462430949786876,
-      "grad_norm": 0.007237066049128771,
-      "learning_rate": 0.00019999408660957327,
-      "loss": 46.0,
-      "step": 45286
-    },
-    {
-      "epoch": 3.462507406770266,
-      "grad_norm": 0.003422307316213846,
-      "learning_rate": 0.00019999408634835771,
-      "loss": 46.0,
-      "step": 45287
-    },
-    {
-      "epoch": 3.4625838637536557,
-      "grad_norm": 0.002165980637073517,
-      "learning_rate": 0.0001999940860871364,
-      "loss": 46.0,
-      "step": 45288
-    },
-    {
-      "epoch": 3.4626603207370454,
-      "grad_norm": 0.0032646714244037867,
-      "learning_rate": 0.00019999408582590934,
-      "loss": 46.0,
-      "step": 45289
-    },
-    {
-      "epoch": 3.462736777720435,
-      "grad_norm": 0.0008388942806050181,
-      "learning_rate": 0.00019999408556467646,
-      "loss": 46.0,
-      "step": 45290
-    },
-    {
-      "epoch": 3.462813234703825,
-      "grad_norm": 0.00189967080950737,
-      "learning_rate": 0.00019999408530343787,
-      "loss": 46.0,
-      "step": 45291
-    },
-    {
-      "epoch": 3.4628896916872147,
-      "grad_norm": 0.001891830237582326,
-      "learning_rate": 0.00019999408504219347,
-      "loss": 46.0,
-      "step": 45292
-    },
-    {
-      "epoch": 3.462966148670604,
-      "grad_norm": 0.0015905596083030105,
-      "learning_rate": 0.0001999940847809433,
-      "loss": 46.0,
-      "step": 45293
-    },
-    {
-      "epoch": 3.463042605653994,
-      "grad_norm": 0.001288939849473536,
-      "learning_rate": 0.0001999940845196874,
-      "loss": 46.0,
-      "step": 45294
-    },
-    {
-      "epoch": 3.4631190626373836,
-      "grad_norm": 0.0018962343456223607,
-      "learning_rate": 0.0001999940842584257,
-      "loss": 46.0,
-      "step": 45295
-    },
-    {
-      "epoch": 3.4631955196207733,
-      "grad_norm": 0.0017523699207231402,
-      "learning_rate": 0.0001999940839971582,
-      "loss": 46.0,
-      "step": 45296
-    },
-    {
-      "epoch": 3.463271976604163,
-      "grad_norm": 0.0019330132054165006,
-      "learning_rate": 0.000199994083735885,
-      "loss": 46.0,
-      "step": 45297
-    },
-    {
-      "epoch": 3.463348433587553,
-      "grad_norm": 0.002740581054240465,
-      "learning_rate": 0.000199994083474606,
-      "loss": 46.0,
-      "step": 45298
-    },
-    {
-      "epoch": 3.4634248905709426,
-      "grad_norm": 0.005950115155428648,
-      "learning_rate": 0.0001999940832133212,
-      "loss": 46.0,
-      "step": 45299
-    },
-    {
-      "epoch": 3.4635013475543324,
-      "grad_norm": 0.003209349699318409,
-      "learning_rate": 0.00019999408295203068,
-      "loss": 46.0,
-      "step": 45300
-    },
-    {
-      "epoch": 3.463577804537722,
-      "grad_norm": 0.0011719181202352047,
-      "learning_rate": 0.00019999408269073436,
-      "loss": 46.0,
-      "step": 45301
-    },
-    {
-      "epoch": 3.4636542615211114,
-      "grad_norm": 0.00136278267018497,
-      "learning_rate": 0.00019999408242943228,
-      "loss": 46.0,
-      "step": 45302
-    },
-    {
-      "epoch": 3.463730718504501,
-      "grad_norm": 0.0019472400890663266,
-      "learning_rate": 0.00019999408216812446,
-      "loss": 46.0,
-      "step": 45303
-    },
-    {
-      "epoch": 3.463807175487891,
-      "grad_norm": 0.006315067410469055,
-      "learning_rate": 0.0001999940819068108,
-      "loss": 46.0,
-      "step": 45304
-    },
-    {
-      "epoch": 3.4638836324712807,
-      "grad_norm": 0.0010663226712495089,
-      "learning_rate": 0.00019999408164549144,
-      "loss": 46.0,
-      "step": 45305
-    },
-    {
-      "epoch": 3.4639600894546705,
-      "grad_norm": 0.002910124370828271,
-      "learning_rate": 0.00019999408138416628,
-      "loss": 46.0,
-      "step": 45306
-    },
-    {
-      "epoch": 3.4640365464380602,
-      "grad_norm": 0.0015780196990817785,
-      "learning_rate": 0.00019999408112283536,
-      "loss": 46.0,
-      "step": 45307
-    },
-    {
-      "epoch": 3.46411300342145,
-      "grad_norm": 0.001325340592302382,
-      "learning_rate": 0.00019999408086149867,
-      "loss": 46.0,
-      "step": 45308
-    },
-    {
-      "epoch": 3.4641894604048398,
-      "grad_norm": 0.0018771840259432793,
-      "learning_rate": 0.0001999940806001562,
-      "loss": 46.0,
-      "step": 45309
-    },
-    {
-      "epoch": 3.4642659173882295,
-      "grad_norm": 0.001647996949031949,
-      "learning_rate": 0.00019999408033880798,
-      "loss": 46.0,
-      "step": 45310
-    },
-    {
-      "epoch": 3.4643423743716193,
-      "grad_norm": 0.00992936547845602,
-      "learning_rate": 0.00019999408007745397,
-      "loss": 46.0,
-      "step": 45311
-    },
-    {
-      "epoch": 3.464418831355009,
-      "grad_norm": 0.0013483690563589334,
-      "learning_rate": 0.0001999940798160942,
-      "loss": 46.0,
-      "step": 45312
-    },
-    {
-      "epoch": 3.464495288338399,
-      "grad_norm": 0.0023338343016803265,
-      "learning_rate": 0.00019999407955472866,
-      "loss": 46.0,
-      "step": 45313
-    },
-    {
-      "epoch": 3.4645717453217886,
-      "grad_norm": 0.005195608362555504,
-      "learning_rate": 0.00019999407929335736,
-      "loss": 46.0,
-      "step": 45314
-    },
-    {
-      "epoch": 3.464648202305178,
-      "grad_norm": 0.002874793251976371,
-      "learning_rate": 0.00019999407903198026,
-      "loss": 46.0,
-      "step": 45315
-    },
-    {
-      "epoch": 3.4647246592885677,
-      "grad_norm": 0.0010338746942579746,
-      "learning_rate": 0.00019999407877059744,
-      "loss": 46.0,
-      "step": 45316
-    },
-    {
-      "epoch": 3.4648011162719574,
-      "grad_norm": 0.0008347270195372403,
-      "learning_rate": 0.00019999407850920881,
-      "loss": 46.0,
-      "step": 45317
-    },
-    {
-      "epoch": 3.464877573255347,
-      "grad_norm": 0.001603742246516049,
-      "learning_rate": 0.00019999407824781442,
-      "loss": 46.0,
-      "step": 45318
-    },
-    {
-      "epoch": 3.464954030238737,
-      "grad_norm": 0.0016313113737851381,
-      "learning_rate": 0.00019999407798641428,
-      "loss": 46.0,
-      "step": 45319
-    },
-    {
-      "epoch": 3.4650304872221267,
-      "grad_norm": 0.0013656540540978312,
-      "learning_rate": 0.00019999407772500836,
-      "loss": 46.0,
-      "step": 45320
-    },
-    {
-      "epoch": 3.4651069442055165,
-      "grad_norm": 0.0034292214550077915,
-      "learning_rate": 0.00019999407746359665,
-      "loss": 46.0,
-      "step": 45321
-    },
-    {
-      "epoch": 3.465183401188906,
-      "grad_norm": 0.001389471348375082,
-      "learning_rate": 0.0001999940772021792,
-      "loss": 46.0,
-      "step": 45322
-    },
-    {
-      "epoch": 3.465259858172296,
-      "grad_norm": 0.0022577354684472084,
-      "learning_rate": 0.00019999407694075595,
-      "loss": 46.0,
-      "step": 45323
-    },
-    {
-      "epoch": 3.4653363151556853,
-      "grad_norm": 0.0020267742220312357,
-      "learning_rate": 0.00019999407667932697,
-      "loss": 46.0,
-      "step": 45324
-    },
-    {
-      "epoch": 3.465412772139075,
-      "grad_norm": 0.002215612679719925,
-      "learning_rate": 0.00019999407641789222,
-      "loss": 46.0,
-      "step": 45325
-    },
-    {
-      "epoch": 3.465489229122465,
-      "grad_norm": 0.001049189013428986,
-      "learning_rate": 0.00019999407615645167,
-      "loss": 46.0,
-      "step": 45326
-    },
-    {
-      "epoch": 3.4655656861058546,
-      "grad_norm": 0.0026559787802398205,
-      "learning_rate": 0.00019999407589500536,
-      "loss": 46.0,
-      "step": 45327
-    },
-    {
-      "epoch": 3.4656421430892443,
-      "grad_norm": 0.0029841770883649588,
-      "learning_rate": 0.0001999940756335533,
-      "loss": 46.0,
-      "step": 45328
-    },
-    {
-      "epoch": 3.465718600072634,
-      "grad_norm": 0.001439036801457405,
-      "learning_rate": 0.00019999407537209542,
-      "loss": 46.0,
-      "step": 45329
-    },
-    {
-      "epoch": 3.465795057056024,
-      "grad_norm": 0.0005394062027335167,
-      "learning_rate": 0.00019999407511063182,
-      "loss": 46.0,
-      "step": 45330
-    },
-    {
-      "epoch": 3.4658715140394136,
-      "grad_norm": 0.0015713658649474382,
-      "learning_rate": 0.00019999407484916243,
-      "loss": 46.0,
-      "step": 45331
-    },
-    {
-      "epoch": 3.4659479710228034,
-      "grad_norm": 0.0019952277652919292,
-      "learning_rate": 0.0001999940745876873,
-      "loss": 46.0,
-      "step": 45332
-    },
-    {
-      "epoch": 3.466024428006193,
-      "grad_norm": 0.0038955719210207462,
-      "learning_rate": 0.00019999407432620637,
-      "loss": 46.0,
-      "step": 45333
-    },
-    {
-      "epoch": 3.466100884989583,
-      "grad_norm": 0.00429118238389492,
-      "learning_rate": 0.00019999407406471966,
-      "loss": 46.0,
-      "step": 45334
-    },
-    {
-      "epoch": 3.4661773419729727,
-      "grad_norm": 0.003267466090619564,
-      "learning_rate": 0.0001999940738032272,
-      "loss": 46.0,
-      "step": 45335
-    },
-    {
-      "epoch": 3.466253798956362,
-      "grad_norm": 0.0022804783657193184,
-      "learning_rate": 0.000199994073541729,
-      "loss": 46.0,
-      "step": 45336
-    },
-    {
-      "epoch": 3.4663302559397517,
-      "grad_norm": 0.001498306985013187,
-      "learning_rate": 0.00019999407328022498,
-      "loss": 46.0,
-      "step": 45337
-    },
-    {
-      "epoch": 3.4664067129231415,
-      "grad_norm": 0.0018274802714586258,
-      "learning_rate": 0.00019999407301871523,
-      "loss": 46.0,
-      "step": 45338
-    },
-    {
-      "epoch": 3.4664831699065313,
-      "grad_norm": 0.0018702266970649362,
-      "learning_rate": 0.0001999940727571997,
-      "loss": 46.0,
-      "step": 45339
-    },
-    {
-      "epoch": 3.466559626889921,
-      "grad_norm": 0.0015760073438286781,
-      "learning_rate": 0.00019999407249567838,
-      "loss": 46.0,
-      "step": 45340
-    },
-    {
-      "epoch": 3.466636083873311,
-      "grad_norm": 0.0009732645121403039,
-      "learning_rate": 0.0001999940722341513,
-      "loss": 46.0,
-      "step": 45341
-    },
-    {
-      "epoch": 3.4667125408567006,
-      "grad_norm": 0.0007463563233613968,
-      "learning_rate": 0.00019999407197261845,
-      "loss": 46.0,
-      "step": 45342
-    },
-    {
-      "epoch": 3.4667889978400903,
-      "grad_norm": 0.0016366144409403205,
-      "learning_rate": 0.00019999407171107983,
-      "loss": 46.0,
-      "step": 45343
-    },
-    {
-      "epoch": 3.46686545482348,
-      "grad_norm": 0.0028873293194919825,
-      "learning_rate": 0.00019999407144953547,
-      "loss": 46.0,
-      "step": 45344
-    },
-    {
-      "epoch": 3.46694191180687,
-      "grad_norm": 0.0012015729444101453,
-      "learning_rate": 0.00019999407118798533,
-      "loss": 46.0,
-      "step": 45345
-    },
-    {
-      "epoch": 3.467018368790259,
-      "grad_norm": 0.0045380378141999245,
-      "learning_rate": 0.0001999940709264294,
-      "loss": 46.0,
-      "step": 45346
-    },
-    {
-      "epoch": 3.467094825773649,
-      "grad_norm": 0.0035851828288286924,
-      "learning_rate": 0.0001999940706648677,
-      "loss": 46.0,
-      "step": 45347
-    },
-    {
-      "epoch": 3.4671712827570387,
-      "grad_norm": 0.0019371291855350137,
-      "learning_rate": 0.00019999407040330024,
-      "loss": 46.0,
-      "step": 45348
-    },
-    {
-      "epoch": 3.4672477397404284,
-      "grad_norm": 0.0019747507758438587,
-      "learning_rate": 0.00019999407014172703,
-      "loss": 46.0,
-      "step": 45349
-    },
-    {
-      "epoch": 3.467324196723818,
-      "grad_norm": 0.00125067587941885,
-      "learning_rate": 0.00019999406988014805,
-      "loss": 46.0,
-      "step": 45350
-    },
-    {
-      "epoch": 3.467400653707208,
-      "grad_norm": 0.0014952163910493255,
-      "learning_rate": 0.00019999406961856327,
-      "loss": 46.0,
-      "step": 45351
-    },
-    {
-      "epoch": 3.4674771106905977,
-      "grad_norm": 0.00152310892008245,
-      "learning_rate": 0.00019999406935697272,
-      "loss": 46.0,
-      "step": 45352
-    },
-    {
-      "epoch": 3.4675535676739875,
-      "grad_norm": 0.0014664194313809276,
-      "learning_rate": 0.00019999406909537642,
-      "loss": 46.0,
-      "step": 45353
-    },
-    {
-      "epoch": 3.4676300246573772,
-      "grad_norm": 0.0016908731777220964,
-      "learning_rate": 0.00019999406883377434,
-      "loss": 46.0,
-      "step": 45354
-    },
-    {
-      "epoch": 3.467706481640767,
-      "grad_norm": 0.000938360346481204,
-      "learning_rate": 0.0001999940685721665,
-      "loss": 46.0,
-      "step": 45355
-    },
-    {
-      "epoch": 3.4677829386241568,
-      "grad_norm": 0.0026080983225256205,
-      "learning_rate": 0.0001999940683105529,
-      "loss": 46.0,
-      "step": 45356
-    },
-    {
-      "epoch": 3.4678593956075465,
-      "grad_norm": 0.0023704352788627148,
-      "learning_rate": 0.00019999406804893354,
-      "loss": 46.0,
-      "step": 45357
-    },
-    {
-      "epoch": 3.467935852590936,
-      "grad_norm": 0.0020570727065205574,
-      "learning_rate": 0.00019999406778730837,
-      "loss": 46.0,
-      "step": 45358
-    },
-    {
-      "epoch": 3.4680123095743256,
-      "grad_norm": 0.001802973565645516,
-      "learning_rate": 0.00019999406752567746,
-      "loss": 46.0,
-      "step": 45359
-    },
-    {
-      "epoch": 3.4680887665577154,
-      "grad_norm": 0.0015169268008321524,
-      "learning_rate": 0.00019999406726404074,
-      "loss": 46.0,
-      "step": 45360
-    },
-    {
-      "epoch": 3.468165223541105,
-      "grad_norm": 0.002393965609371662,
-      "learning_rate": 0.0001999940670023983,
-      "loss": 46.0,
-      "step": 45361
-    },
-    {
-      "epoch": 3.468241680524495,
-      "grad_norm": 0.0034477978479117155,
-      "learning_rate": 0.0001999940667407501,
-      "loss": 46.0,
-      "step": 45362
-    },
-    {
-      "epoch": 3.4683181375078846,
-      "grad_norm": 0.002363716484978795,
-      "learning_rate": 0.0001999940664790961,
-      "loss": 46.0,
-      "step": 45363
-    },
-    {
-      "epoch": 3.4683945944912744,
-      "grad_norm": 0.002004781737923622,
-      "learning_rate": 0.00019999406621743632,
-      "loss": 46.0,
-      "step": 45364
-    },
-    {
-      "epoch": 3.468471051474664,
-      "grad_norm": 0.0067963371984660625,
-      "learning_rate": 0.0001999940659557708,
-      "loss": 46.0,
-      "step": 45365
-    },
-    {
-      "epoch": 3.468547508458054,
-      "grad_norm": 0.0028303293511271477,
-      "learning_rate": 0.0001999940656940995,
-      "loss": 46.0,
-      "step": 45366
-    },
-    {
-      "epoch": 3.4686239654414437,
-      "grad_norm": 0.008921141736209393,
-      "learning_rate": 0.00019999406543242242,
-      "loss": 46.0,
-      "step": 45367
-    },
-    {
-      "epoch": 3.468700422424833,
-      "grad_norm": 0.0016227097949013114,
-      "learning_rate": 0.00019999406517073957,
-      "loss": 46.0,
-      "step": 45368
-    },
-    {
-      "epoch": 3.4687768794082228,
-      "grad_norm": 0.002052989089861512,
-      "learning_rate": 0.00019999406490905095,
-      "loss": 46.0,
-      "step": 45369
-    },
-    {
-      "epoch": 3.4688533363916125,
-      "grad_norm": 0.0020686821080744267,
-      "learning_rate": 0.00019999406464735658,
-      "loss": 46.0,
-      "step": 45370
-    },
-    {
-      "epoch": 3.4689297933750023,
-      "grad_norm": 0.0031085405498743057,
-      "learning_rate": 0.00019999406438565644,
-      "loss": 46.0,
-      "step": 45371
-    },
-    {
-      "epoch": 3.469006250358392,
-      "grad_norm": 0.0006648002308793366,
-      "learning_rate": 0.00019999406412395053,
-      "loss": 46.0,
-      "step": 45372
-    },
-    {
-      "epoch": 3.469082707341782,
-      "grad_norm": 0.001301878015510738,
-      "learning_rate": 0.00019999406386223884,
-      "loss": 46.0,
-      "step": 45373
-    },
-    {
-      "epoch": 3.4691591643251716,
-      "grad_norm": 0.0019161106320098042,
-      "learning_rate": 0.00019999406360052138,
-      "loss": 46.0,
-      "step": 45374
-    },
-    {
-      "epoch": 3.4692356213085613,
-      "grad_norm": 0.0020964015275239944,
-      "learning_rate": 0.00019999406333879815,
-      "loss": 46.0,
-      "step": 45375
-    },
-    {
-      "epoch": 3.469312078291951,
-      "grad_norm": 0.0012005208991467953,
-      "learning_rate": 0.00019999406307706917,
-      "loss": 46.0,
-      "step": 45376
-    },
-    {
-      "epoch": 3.469388535275341,
-      "grad_norm": 0.0017579332925379276,
-      "learning_rate": 0.00019999406281533442,
-      "loss": 46.0,
-      "step": 45377
-    },
-    {
-      "epoch": 3.4694649922587306,
-      "grad_norm": 0.003330113599076867,
-      "learning_rate": 0.00019999406255359387,
-      "loss": 46.0,
-      "step": 45378
-    },
-    {
-      "epoch": 3.4695414492421204,
-      "grad_norm": 0.00260583171620965,
-      "learning_rate": 0.00019999406229184757,
-      "loss": 46.0,
-      "step": 45379
-    },
-    {
-      "epoch": 3.4696179062255097,
-      "grad_norm": 0.0036398381926119328,
-      "learning_rate": 0.0001999940620300955,
-      "loss": 46.0,
-      "step": 45380
-    },
-    {
-      "epoch": 3.4696943632088995,
-      "grad_norm": 0.006889119278639555,
-      "learning_rate": 0.00019999406176833765,
-      "loss": 46.0,
-      "step": 45381
-    },
-    {
-      "epoch": 3.469770820192289,
-      "grad_norm": 0.0025699592661112547,
-      "learning_rate": 0.00019999406150657405,
-      "loss": 46.0,
-      "step": 45382
-    },
-    {
-      "epoch": 3.469847277175679,
-      "grad_norm": 0.0049185906536877155,
-      "learning_rate": 0.00019999406124480466,
-      "loss": 46.0,
-      "step": 45383
-    },
-    {
-      "epoch": 3.4699237341590687,
-      "grad_norm": 0.001075700274668634,
-      "learning_rate": 0.00019999406098302952,
-      "loss": 46.0,
-      "step": 45384
-    },
-    {
-      "epoch": 3.4700001911424585,
-      "grad_norm": 0.0006722345133312047,
-      "learning_rate": 0.0001999940607212486,
-      "loss": 46.0,
-      "step": 45385
-    },
-    {
-      "epoch": 3.4700766481258483,
-      "grad_norm": 0.0028031079564243555,
-      "learning_rate": 0.00019999406045946195,
-      "loss": 46.0,
-      "step": 45386
-    },
-    {
-      "epoch": 3.470153105109238,
-      "grad_norm": 0.004638790152966976,
-      "learning_rate": 0.0001999940601976695,
-      "loss": 46.0,
-      "step": 45387
-    },
-    {
-      "epoch": 3.470229562092628,
-      "grad_norm": 0.003844328224658966,
-      "learning_rate": 0.00019999405993587126,
-      "loss": 46.0,
-      "step": 45388
-    },
-    {
-      "epoch": 3.4703060190760175,
-      "grad_norm": 0.001000547781586647,
-      "learning_rate": 0.00019999405967406725,
-      "loss": 46.0,
-      "step": 45389
-    },
-    {
-      "epoch": 3.470382476059407,
-      "grad_norm": 0.0009820261038839817,
-      "learning_rate": 0.0001999940594122575,
-      "loss": 46.0,
-      "step": 45390
-    },
-    {
-      "epoch": 3.4704589330427966,
-      "grad_norm": 0.00280400482006371,
-      "learning_rate": 0.00019999405915044197,
-      "loss": 46.0,
-      "step": 45391
-    },
-    {
-      "epoch": 3.4705353900261864,
-      "grad_norm": 0.0023543043062090874,
-      "learning_rate": 0.00019999405888862067,
-      "loss": 46.0,
-      "step": 45392
-    },
-    {
-      "epoch": 3.470611847009576,
-      "grad_norm": 0.0017922515980899334,
-      "learning_rate": 0.0001999940586267936,
-      "loss": 46.0,
-      "step": 45393
-    },
-    {
-      "epoch": 3.470688303992966,
-      "grad_norm": 0.0027865006122738123,
-      "learning_rate": 0.00019999405836496076,
-      "loss": 46.0,
-      "step": 45394
-    },
-    {
-      "epoch": 3.4707647609763557,
-      "grad_norm": 0.001274133799597621,
-      "learning_rate": 0.00019999405810312216,
-      "loss": 46.0,
-      "step": 45395
-    },
-    {
-      "epoch": 3.4708412179597454,
-      "grad_norm": 0.003584288526326418,
-      "learning_rate": 0.00019999405784127777,
-      "loss": 46.0,
-      "step": 45396
-    },
-    {
-      "epoch": 3.470917674943135,
-      "grad_norm": 0.0038590379990637302,
-      "learning_rate": 0.00019999405757942763,
-      "loss": 46.0,
-      "step": 45397
-    },
-    {
-      "epoch": 3.470994131926525,
-      "grad_norm": 0.003260708414018154,
-      "learning_rate": 0.00019999405731757172,
-      "loss": 46.0,
-      "step": 45398
-    },
-    {
-      "epoch": 3.4710705889099147,
-      "grad_norm": 0.002681143581867218,
-      "learning_rate": 0.00019999405705571003,
-      "loss": 46.0,
-      "step": 45399
-    },
-    {
-      "epoch": 3.4711470458933045,
-      "grad_norm": 0.0018498157151043415,
-      "learning_rate": 0.00019999405679384257,
-      "loss": 46.0,
-      "step": 45400
-    },
-    {
-      "epoch": 3.4712235028766942,
-      "grad_norm": 0.0012419262202456594,
-      "learning_rate": 0.00019999405653196937,
-      "loss": 46.0,
-      "step": 45401
-    },
-    {
-      "epoch": 3.4712999598600836,
-      "grad_norm": 0.0030020864214748144,
-      "learning_rate": 0.0001999940562700904,
-      "loss": 46.0,
-      "step": 45402
-    },
-    {
-      "epoch": 3.4713764168434733,
-      "grad_norm": 0.0011914694914594293,
-      "learning_rate": 0.0001999940560082056,
-      "loss": 46.0,
-      "step": 45403
-    },
-    {
-      "epoch": 3.471452873826863,
-      "grad_norm": 0.008153322152793407,
-      "learning_rate": 0.00019999405574631509,
-      "loss": 46.0,
-      "step": 45404
-    },
-    {
-      "epoch": 3.471529330810253,
-      "grad_norm": 0.0048467861488461494,
-      "learning_rate": 0.00019999405548441881,
-      "loss": 46.0,
-      "step": 45405
-    },
-    {
-      "epoch": 3.4716057877936426,
-      "grad_norm": 0.0006020099972374737,
-      "learning_rate": 0.00019999405522251674,
-      "loss": 46.0,
-      "step": 45406
-    },
-    {
-      "epoch": 3.4716822447770324,
-      "grad_norm": 0.0022257273085415363,
-      "learning_rate": 0.00019999405496060887,
-      "loss": 46.0,
-      "step": 45407
-    },
-    {
-      "epoch": 3.471758701760422,
-      "grad_norm": 0.0021620949264615774,
-      "learning_rate": 0.00019999405469869528,
-      "loss": 46.0,
-      "step": 45408
-    },
-    {
-      "epoch": 3.471835158743812,
-      "grad_norm": 0.0030188492964953184,
-      "learning_rate": 0.0001999940544367759,
-      "loss": 46.0,
-      "step": 45409
-    },
-    {
-      "epoch": 3.4719116157272016,
-      "grad_norm": 0.0021532301325351,
-      "learning_rate": 0.00019999405417485077,
-      "loss": 46.0,
-      "step": 45410
-    },
-    {
-      "epoch": 3.471988072710591,
-      "grad_norm": 0.003152495017275214,
-      "learning_rate": 0.00019999405391291986,
-      "loss": 46.0,
-      "step": 45411
-    },
-    {
-      "epoch": 3.4720645296939807,
-      "grad_norm": 0.004910083953291178,
-      "learning_rate": 0.00019999405365098318,
-      "loss": 46.0,
-      "step": 45412
-    },
-    {
-      "epoch": 3.4721409866773705,
-      "grad_norm": 0.005014121066778898,
-      "learning_rate": 0.00019999405338904072,
-      "loss": 46.0,
-      "step": 45413
-    },
-    {
-      "epoch": 3.4722174436607602,
-      "grad_norm": 0.001974246231839061,
-      "learning_rate": 0.0001999940531270925,
-      "loss": 46.0,
-      "step": 45414
-    },
-    {
-      "epoch": 3.47229390064415,
-      "grad_norm": 0.0015629720874130726,
-      "learning_rate": 0.0001999940528651385,
-      "loss": 46.0,
-      "step": 45415
-    },
-    {
-      "epoch": 3.4723703576275398,
-      "grad_norm": 0.0005405074334703386,
-      "learning_rate": 0.00019999405260317875,
-      "loss": 46.0,
-      "step": 45416
-    },
-    {
-      "epoch": 3.4724468146109295,
-      "grad_norm": 0.007345457561314106,
-      "learning_rate": 0.00019999405234121326,
-      "loss": 46.0,
-      "step": 45417
-    },
-    {
-      "epoch": 3.4725232715943193,
-      "grad_norm": 0.009950641542673111,
-      "learning_rate": 0.00019999405207924193,
-      "loss": 46.0,
-      "step": 45418
-    },
-    {
-      "epoch": 3.472599728577709,
-      "grad_norm": 0.001955407904461026,
-      "learning_rate": 0.00019999405181726486,
-      "loss": 46.0,
-      "step": 45419
-    },
-    {
-      "epoch": 3.472676185561099,
-      "grad_norm": 0.0026656100526452065,
-      "learning_rate": 0.00019999405155528204,
-      "loss": 46.0,
-      "step": 45420
-    },
-    {
-      "epoch": 3.4727526425444886,
-      "grad_norm": 0.0013629165478050709,
-      "learning_rate": 0.00019999405129329345,
-      "loss": 46.0,
-      "step": 45421
-    },
-    {
-      "epoch": 3.4728290995278783,
-      "grad_norm": 0.003128195647150278,
-      "learning_rate": 0.00019999405103129906,
-      "loss": 46.0,
-      "step": 45422
-    },
-    {
-      "epoch": 3.472905556511268,
-      "grad_norm": 0.002011357806622982,
-      "learning_rate": 0.00019999405076929892,
-      "loss": 46.0,
-      "step": 45423
-    },
-    {
-      "epoch": 3.4729820134946574,
-      "grad_norm": 0.0026485517155379057,
-      "learning_rate": 0.000199994050507293,
-      "loss": 46.0,
-      "step": 45424
-    },
-    {
-      "epoch": 3.473058470478047,
-      "grad_norm": 0.0023168164771050215,
-      "learning_rate": 0.00019999405024528132,
-      "loss": 46.0,
-      "step": 45425
-    },
-    {
-      "epoch": 3.473134927461437,
-      "grad_norm": 0.003357185050845146,
-      "learning_rate": 0.0001999940499832639,
-      "loss": 46.0,
-      "step": 45426
-    },
-    {
-      "epoch": 3.4732113844448267,
-      "grad_norm": 0.00062833062838763,
-      "learning_rate": 0.00019999404972124066,
-      "loss": 46.0,
-      "step": 45427
-    },
-    {
-      "epoch": 3.4732878414282164,
-      "grad_norm": 0.0012073391117155552,
-      "learning_rate": 0.00019999404945921168,
-      "loss": 46.0,
-      "step": 45428
-    },
-    {
-      "epoch": 3.473364298411606,
-      "grad_norm": 0.0024175432045012712,
-      "learning_rate": 0.00019999404919717693,
-      "loss": 46.0,
-      "step": 45429
-    },
-    {
-      "epoch": 3.473440755394996,
-      "grad_norm": 0.0015951709356158972,
-      "learning_rate": 0.0001999940489351364,
-      "loss": 46.0,
-      "step": 45430
-    },
-    {
-      "epoch": 3.4735172123783857,
-      "grad_norm": 0.0013720013666898012,
-      "learning_rate": 0.0001999940486730901,
-      "loss": 46.0,
-      "step": 45431
-    },
-    {
-      "epoch": 3.4735936693617755,
-      "grad_norm": 0.001110560609959066,
-      "learning_rate": 0.00019999404841103803,
-      "loss": 46.0,
-      "step": 45432
-    },
-    {
-      "epoch": 3.473670126345165,
-      "grad_norm": 0.001942623988725245,
-      "learning_rate": 0.0001999940481489802,
-      "loss": 46.0,
-      "step": 45433
-    },
-    {
-      "epoch": 3.4737465833285546,
-      "grad_norm": 0.007223673164844513,
-      "learning_rate": 0.0001999940478869166,
-      "loss": 46.0,
-      "step": 45434
-    },
-    {
-      "epoch": 3.4738230403119443,
-      "grad_norm": 0.00296038668602705,
-      "learning_rate": 0.00019999404762484723,
-      "loss": 46.0,
-      "step": 45435
-    },
-    {
-      "epoch": 3.473899497295334,
-      "grad_norm": 0.0013491497375071049,
-      "learning_rate": 0.0001999940473627721,
-      "loss": 46.0,
-      "step": 45436
-    },
-    {
-      "epoch": 3.473975954278724,
-      "grad_norm": 0.003378593595698476,
-      "learning_rate": 0.00019999404710069118,
-      "loss": 46.0,
-      "step": 45437
-    },
-    {
-      "epoch": 3.4740524112621136,
-      "grad_norm": 0.0005663735792040825,
-      "learning_rate": 0.00019999404683860452,
-      "loss": 46.0,
-      "step": 45438
-    },
-    {
-      "epoch": 3.4741288682455034,
-      "grad_norm": 0.002361743478104472,
-      "learning_rate": 0.00019999404657651207,
-      "loss": 46.0,
-      "step": 45439
-    },
-    {
-      "epoch": 3.474205325228893,
-      "grad_norm": 0.002637530444189906,
-      "learning_rate": 0.00019999404631441383,
-      "loss": 46.0,
-      "step": 45440
-    },
-    {
-      "epoch": 3.474281782212283,
-      "grad_norm": 0.010690860450267792,
-      "learning_rate": 0.00019999404605230986,
-      "loss": 46.0,
-      "step": 45441
-    },
-    {
-      "epoch": 3.4743582391956727,
-      "grad_norm": 0.002201820258051157,
-      "learning_rate": 0.00019999404579020008,
-      "loss": 46.0,
-      "step": 45442
-    },
-    {
-      "epoch": 3.4744346961790624,
-      "grad_norm": 0.00961847510188818,
-      "learning_rate": 0.00019999404552808455,
-      "loss": 46.0,
-      "step": 45443
-    },
-    {
-      "epoch": 3.474511153162452,
-      "grad_norm": 0.0037724098656326532,
-      "learning_rate": 0.00019999404526596328,
-      "loss": 46.0,
-      "step": 45444
-    },
-    {
-      "epoch": 3.474587610145842,
-      "grad_norm": 0.0024975226260721684,
-      "learning_rate": 0.00019999404500383624,
-      "loss": 46.0,
-      "step": 45445
-    },
-    {
-      "epoch": 3.4746640671292313,
-      "grad_norm": 0.0024796074721962214,
-      "learning_rate": 0.00019999404474170337,
-      "loss": 46.0,
-      "step": 45446
-    },
-    {
-      "epoch": 3.474740524112621,
-      "grad_norm": 0.0014058712404221296,
-      "learning_rate": 0.0001999940444795648,
-      "loss": 46.0,
-      "step": 45447
-    },
-    {
-      "epoch": 3.474816981096011,
-      "grad_norm": 0.002800158690661192,
-      "learning_rate": 0.0001999940442174204,
-      "loss": 46.0,
-      "step": 45448
-    },
-    {
-      "epoch": 3.4748934380794005,
-      "grad_norm": 0.0027238046750426292,
-      "learning_rate": 0.00019999404395527027,
-      "loss": 46.0,
-      "step": 45449
-    },
-    {
-      "epoch": 3.4749698950627903,
-      "grad_norm": 0.006922128144651651,
-      "learning_rate": 0.0001999940436931144,
-      "loss": 46.0,
-      "step": 45450
-    },
-    {
-      "epoch": 3.47504635204618,
-      "grad_norm": 0.00367325940169394,
-      "learning_rate": 0.0001999940434309527,
-      "loss": 46.0,
-      "step": 45451
-    },
-    {
-      "epoch": 3.47512280902957,
-      "grad_norm": 0.0013188676675781608,
-      "learning_rate": 0.00019999404316878525,
-      "loss": 46.0,
-      "step": 45452
-    },
-    {
-      "epoch": 3.4751992660129596,
-      "grad_norm": 0.002308727242052555,
-      "learning_rate": 0.00019999404290661204,
-      "loss": 46.0,
-      "step": 45453
-    },
-    {
-      "epoch": 3.4752757229963493,
-      "grad_norm": 0.0035989207681268454,
-      "learning_rate": 0.00019999404264443307,
-      "loss": 46.0,
-      "step": 45454
-    },
-    {
-      "epoch": 3.4753521799797387,
-      "grad_norm": 0.001186361419968307,
-      "learning_rate": 0.00019999404238224831,
-      "loss": 46.0,
-      "step": 45455
-    },
-    {
-      "epoch": 3.4754286369631284,
-      "grad_norm": 0.0010325969196856022,
-      "learning_rate": 0.00019999404212005776,
-      "loss": 46.0,
-      "step": 45456
-    },
-    {
-      "epoch": 3.475505093946518,
-      "grad_norm": 0.002916923025622964,
-      "learning_rate": 0.0001999940418578615,
-      "loss": 46.0,
-      "step": 45457
-    },
-    {
-      "epoch": 3.475581550929908,
-      "grad_norm": 0.0038317441940307617,
-      "learning_rate": 0.00019999404159565942,
-      "loss": 46.0,
-      "step": 45458
-    },
-    {
-      "epoch": 3.4756580079132977,
-      "grad_norm": 0.001993397017940879,
-      "learning_rate": 0.0001999940413334516,
-      "loss": 46.0,
-      "step": 45459
-    },
-    {
-      "epoch": 3.4757344648966875,
-      "grad_norm": 0.004902064334601164,
-      "learning_rate": 0.00019999404107123801,
-      "loss": 46.0,
-      "step": 45460
-    },
-    {
-      "epoch": 3.4758109218800772,
-      "grad_norm": 0.0023605634924024343,
-      "learning_rate": 0.00019999404080901862,
-      "loss": 46.0,
-      "step": 45461
-    },
-    {
-      "epoch": 3.475887378863467,
-      "grad_norm": 0.00530610978603363,
-      "learning_rate": 0.00019999404054679349,
-      "loss": 46.0,
-      "step": 45462
-    },
-    {
-      "epoch": 3.4759638358468568,
-      "grad_norm": 0.002614965196698904,
-      "learning_rate": 0.0001999940402845626,
-      "loss": 46.0,
-      "step": 45463
-    },
-    {
-      "epoch": 3.4760402928302465,
-      "grad_norm": 0.003270120592787862,
-      "learning_rate": 0.00019999404002232592,
-      "loss": 46.0,
-      "step": 45464
-    },
-    {
-      "epoch": 3.4761167498136363,
-      "grad_norm": 0.0026582053396850824,
-      "learning_rate": 0.00019999403976008346,
-      "loss": 46.0,
-      "step": 45465
-    },
-    {
-      "epoch": 3.476193206797026,
-      "grad_norm": 0.003206276800483465,
-      "learning_rate": 0.00019999403949783526,
-      "loss": 46.0,
-      "step": 45466
-    },
-    {
-      "epoch": 3.4762696637804154,
-      "grad_norm": 0.00184364162851125,
-      "learning_rate": 0.00019999403923558128,
-      "loss": 46.0,
-      "step": 45467
-    },
-    {
-      "epoch": 3.476346120763805,
-      "grad_norm": 0.0016548899002373219,
-      "learning_rate": 0.00019999403897332153,
-      "loss": 46.0,
-      "step": 45468
-    },
-    {
-      "epoch": 3.476422577747195,
-      "grad_norm": 0.0034113412257283926,
-      "learning_rate": 0.000199994038711056,
-      "loss": 46.0,
-      "step": 45469
-    },
-    {
-      "epoch": 3.4764990347305846,
-      "grad_norm": 0.002269693650305271,
-      "learning_rate": 0.0001999940384487847,
-      "loss": 46.0,
-      "step": 45470
-    },
-    {
-      "epoch": 3.4765754917139744,
-      "grad_norm": 0.003167959628626704,
-      "learning_rate": 0.00019999403818650766,
-      "loss": 46.0,
-      "step": 45471
-    },
-    {
-      "epoch": 3.476651948697364,
-      "grad_norm": 0.0020699365995824337,
-      "learning_rate": 0.00019999403792422485,
-      "loss": 46.0,
-      "step": 45472
-    },
-    {
-      "epoch": 3.476728405680754,
-      "grad_norm": 0.016294796019792557,
-      "learning_rate": 0.00019999403766193623,
-      "loss": 46.0,
-      "step": 45473
-    },
-    {
-      "epoch": 3.4768048626641437,
-      "grad_norm": 0.0007564066327176988,
-      "learning_rate": 0.00019999403739964187,
-      "loss": 46.0,
-      "step": 45474
-    },
-    {
-      "epoch": 3.4768813196475334,
-      "grad_norm": 0.0017108300235122442,
-      "learning_rate": 0.00019999403713734173,
-      "loss": 46.0,
-      "step": 45475
-    },
-    {
-      "epoch": 3.476957776630923,
-      "grad_norm": 0.004104054067283869,
-      "learning_rate": 0.00019999403687503584,
-      "loss": 46.0,
-      "step": 45476
-    },
-    {
-      "epoch": 3.4770342336143125,
-      "grad_norm": 0.003205989021807909,
-      "learning_rate": 0.00019999403661272416,
-      "loss": 46.0,
-      "step": 45477
-    },
-    {
-      "epoch": 3.4771106905977023,
-      "grad_norm": 0.008060215972363949,
-      "learning_rate": 0.0001999940363504067,
-      "loss": 46.0,
-      "step": 45478
-    },
-    {
-      "epoch": 3.477187147581092,
-      "grad_norm": 0.0021951785311102867,
-      "learning_rate": 0.0001999940360880835,
-      "loss": 46.0,
-      "step": 45479
-    },
-    {
-      "epoch": 3.477263604564482,
-      "grad_norm": 0.0014548609033226967,
-      "learning_rate": 0.00019999403582575452,
-      "loss": 46.0,
-      "step": 45480
-    },
-    {
-      "epoch": 3.4773400615478716,
-      "grad_norm": 0.0011106256861239672,
-      "learning_rate": 0.00019999403556341977,
-      "loss": 46.0,
-      "step": 45481
-    },
-    {
-      "epoch": 3.4774165185312613,
-      "grad_norm": 0.0019740527495741844,
-      "learning_rate": 0.00019999403530107922,
-      "loss": 46.0,
-      "step": 45482
-    },
-    {
-      "epoch": 3.477492975514651,
-      "grad_norm": 0.002642839215695858,
-      "learning_rate": 0.00019999403503873295,
-      "loss": 46.0,
-      "step": 45483
-    },
-    {
-      "epoch": 3.477569432498041,
-      "grad_norm": 0.004174418747425079,
-      "learning_rate": 0.0001999940347763809,
-      "loss": 46.0,
-      "step": 45484
-    },
-    {
-      "epoch": 3.4776458894814306,
-      "grad_norm": 0.0010072627337649465,
-      "learning_rate": 0.00019999403451402306,
-      "loss": 46.0,
-      "step": 45485
-    },
-    {
-      "epoch": 3.4777223464648204,
-      "grad_norm": 0.001531164743937552,
-      "learning_rate": 0.00019999403425165947,
-      "loss": 46.0,
-      "step": 45486
-    },
-    {
-      "epoch": 3.47779880344821,
-      "grad_norm": 0.0037696228828281164,
-      "learning_rate": 0.0001999940339892901,
-      "loss": 46.0,
-      "step": 45487
-    },
-    {
-      "epoch": 3.4778752604316,
-      "grad_norm": 0.007739607710391283,
-      "learning_rate": 0.00019999403372691497,
-      "loss": 46.0,
-      "step": 45488
-    },
-    {
-      "epoch": 3.477951717414989,
-      "grad_norm": 0.0010087728733196855,
-      "learning_rate": 0.0001999940334645341,
-      "loss": 46.0,
-      "step": 45489
-    },
-    {
-      "epoch": 3.478028174398379,
-      "grad_norm": 0.0040344009175896645,
-      "learning_rate": 0.0001999940332021474,
-      "loss": 46.0,
-      "step": 45490
-    },
-    {
-      "epoch": 3.4781046313817687,
-      "grad_norm": 0.003061852650716901,
-      "learning_rate": 0.00019999403293975495,
-      "loss": 46.0,
-      "step": 45491
-    },
-    {
-      "epoch": 3.4781810883651585,
-      "grad_norm": 0.0016354190884158015,
-      "learning_rate": 0.00019999403267735675,
-      "loss": 46.0,
-      "step": 45492
-    },
-    {
-      "epoch": 3.4782575453485483,
-      "grad_norm": 0.003986387047916651,
-      "learning_rate": 0.00019999403241495277,
-      "loss": 46.0,
-      "step": 45493
-    },
-    {
-      "epoch": 3.478334002331938,
-      "grad_norm": 0.0032610264606773853,
-      "learning_rate": 0.00019999403215254305,
-      "loss": 46.0,
-      "step": 45494
-    },
-    {
-      "epoch": 3.4784104593153278,
-      "grad_norm": 0.002538049826398492,
-      "learning_rate": 0.0001999940318901275,
-      "loss": 46.0,
-      "step": 45495
-    },
-    {
-      "epoch": 3.4784869162987175,
-      "grad_norm": 0.0019715926609933376,
-      "learning_rate": 0.00019999403162770623,
-      "loss": 46.0,
-      "step": 45496
-    },
-    {
-      "epoch": 3.4785633732821073,
-      "grad_norm": 0.002381403697654605,
-      "learning_rate": 0.00019999403136527916,
-      "loss": 46.0,
-      "step": 45497
-    },
-    {
-      "epoch": 3.478639830265497,
-      "grad_norm": 0.002238808199763298,
-      "learning_rate": 0.00019999403110284637,
-      "loss": 46.0,
-      "step": 45498
-    },
-    {
-      "epoch": 3.4787162872488864,
-      "grad_norm": 0.0012791279004886746,
-      "learning_rate": 0.00019999403084040775,
-      "loss": 46.0,
-      "step": 45499
-    },
-    {
-      "epoch": 3.478792744232276,
-      "grad_norm": 0.0014512179186567664,
-      "learning_rate": 0.0001999940305779634,
-      "loss": 46.0,
-      "step": 45500
-    },
-    {
-      "epoch": 3.478869201215666,
-      "grad_norm": 0.0026128538884222507,
-      "learning_rate": 0.00019999403031551328,
-      "loss": 46.0,
-      "step": 45501
-    },
-    {
-      "epoch": 3.4789456581990557,
-      "grad_norm": 0.0007514961762353778,
-      "learning_rate": 0.00019999403005305734,
-      "loss": 46.0,
-      "step": 45502
-    },
-    {
-      "epoch": 3.4790221151824454,
-      "grad_norm": 0.002741573378443718,
-      "learning_rate": 0.0001999940297905957,
-      "loss": 46.0,
-      "step": 45503
-    },
-    {
-      "epoch": 3.479098572165835,
-      "grad_norm": 0.0006237424095161259,
-      "learning_rate": 0.00019999402952812826,
-      "loss": 46.0,
-      "step": 45504
-    },
-    {
-      "epoch": 3.479175029149225,
-      "grad_norm": 0.0019310236675664783,
-      "learning_rate": 0.00019999402926565503,
-      "loss": 46.0,
-      "step": 45505
-    },
-    {
-      "epoch": 3.4792514861326147,
-      "grad_norm": 0.004290848504751921,
-      "learning_rate": 0.00019999402900317605,
-      "loss": 46.0,
-      "step": 45506
-    },
-    {
-      "epoch": 3.4793279431160045,
-      "grad_norm": 0.0010485415114089847,
-      "learning_rate": 0.0001999940287406913,
-      "loss": 46.0,
-      "step": 45507
-    },
-    {
-      "epoch": 3.4794044000993942,
-      "grad_norm": 0.0027531886007636786,
-      "learning_rate": 0.00019999402847820078,
-      "loss": 46.0,
-      "step": 45508
-    },
-    {
-      "epoch": 3.479480857082784,
-      "grad_norm": 0.0015785720897838473,
-      "learning_rate": 0.0001999940282157045,
-      "loss": 46.0,
-      "step": 45509
-    },
-    {
-      "epoch": 3.4795573140661737,
-      "grad_norm": 0.0017468953737989068,
-      "learning_rate": 0.00019999402795320247,
-      "loss": 46.0,
-      "step": 45510
-    },
-    {
-      "epoch": 3.479633771049563,
-      "grad_norm": 0.004634004086256027,
-      "learning_rate": 0.00019999402769069465,
-      "loss": 46.0,
-      "step": 45511
-    },
-    {
-      "epoch": 3.479710228032953,
-      "grad_norm": 0.003090350888669491,
-      "learning_rate": 0.00019999402742818103,
-      "loss": 46.0,
-      "step": 45512
-    },
-    {
-      "epoch": 3.4797866850163426,
-      "grad_norm": 0.0029485109262168407,
-      "learning_rate": 0.0001999940271656617,
-      "loss": 46.0,
-      "step": 45513
-    },
-    {
-      "epoch": 3.4798631419997323,
-      "grad_norm": 0.0016075458843261003,
-      "learning_rate": 0.00019999402690313653,
-      "loss": 46.0,
-      "step": 45514
-    },
-    {
-      "epoch": 3.479939598983122,
-      "grad_norm": 0.001572399283759296,
-      "learning_rate": 0.00019999402664060565,
-      "loss": 46.0,
-      "step": 45515
-    },
-    {
-      "epoch": 3.480016055966512,
-      "grad_norm": 0.0007134056650102139,
-      "learning_rate": 0.000199994026378069,
-      "loss": 46.0,
-      "step": 45516
-    },
-    {
-      "epoch": 3.4800925129499016,
-      "grad_norm": 0.004497717134654522,
-      "learning_rate": 0.00019999402611552657,
-      "loss": 46.0,
-      "step": 45517
-    },
-    {
-      "epoch": 3.4801689699332914,
-      "grad_norm": 0.0009015090181492269,
-      "learning_rate": 0.00019999402585297834,
-      "loss": 46.0,
-      "step": 45518
-    },
-    {
-      "epoch": 3.480245426916681,
-      "grad_norm": 0.0017260683234781027,
-      "learning_rate": 0.00019999402559042436,
-      "loss": 46.0,
-      "step": 45519
-    },
-    {
-      "epoch": 3.480321883900071,
-      "grad_norm": 0.0012865082826465368,
-      "learning_rate": 0.0001999940253278646,
-      "loss": 46.0,
-      "step": 45520
-    },
-    {
-      "epoch": 3.4803983408834602,
-      "grad_norm": 0.0013066261308267713,
-      "learning_rate": 0.0001999940250652991,
-      "loss": 46.0,
-      "step": 45521
-    },
-    {
-      "epoch": 3.48047479786685,
-      "grad_norm": 0.001797198667190969,
-      "learning_rate": 0.00019999402480272782,
-      "loss": 46.0,
-      "step": 45522
-    },
-    {
-      "epoch": 3.4805512548502398,
-      "grad_norm": 0.004325026646256447,
-      "learning_rate": 0.00019999402454015078,
-      "loss": 46.0,
-      "step": 45523
-    },
-    {
-      "epoch": 3.4806277118336295,
-      "grad_norm": 0.0018415595404803753,
-      "learning_rate": 0.00019999402427756794,
-      "loss": 46.0,
-      "step": 45524
-    },
-    {
-      "epoch": 3.4807041688170193,
-      "grad_norm": 0.001642862567678094,
-      "learning_rate": 0.00019999402401497935,
-      "loss": 46.0,
-      "step": 45525
-    },
-    {
-      "epoch": 3.480780625800409,
-      "grad_norm": 0.003109134268015623,
-      "learning_rate": 0.00019999402375238498,
-      "loss": 46.0,
-      "step": 45526
-    },
-    {
-      "epoch": 3.480857082783799,
-      "grad_norm": 0.0014782004291191697,
-      "learning_rate": 0.00019999402348978487,
-      "loss": 46.0,
-      "step": 45527
-    },
-    {
-      "epoch": 3.4809335397671886,
-      "grad_norm": 0.0032355119474232197,
-      "learning_rate": 0.00019999402322717894,
-      "loss": 46.0,
-      "step": 45528
-    },
-    {
-      "epoch": 3.4810099967505783,
-      "grad_norm": 0.0010524388635531068,
-      "learning_rate": 0.00019999402296456728,
-      "loss": 46.0,
-      "step": 45529
-    },
-    {
-      "epoch": 3.481086453733968,
-      "grad_norm": 0.0028826151974499226,
-      "learning_rate": 0.00019999402270194985,
-      "loss": 46.0,
-      "step": 45530
-    },
-    {
-      "epoch": 3.481162910717358,
-      "grad_norm": 0.004183790180832148,
-      "learning_rate": 0.00019999402243932665,
-      "loss": 46.0,
-      "step": 45531
-    },
-    {
-      "epoch": 3.4812393677007476,
-      "grad_norm": 0.001213785377331078,
-      "learning_rate": 0.0001999940221766977,
-      "loss": 46.0,
-      "step": 45532
-    },
-    {
-      "epoch": 3.481315824684137,
-      "grad_norm": 0.0037122094072401524,
-      "learning_rate": 0.00019999402191406293,
-      "loss": 46.0,
-      "step": 45533
-    },
-    {
-      "epoch": 3.4813922816675267,
-      "grad_norm": 0.003766071517020464,
-      "learning_rate": 0.00019999402165142243,
-      "loss": 46.0,
-      "step": 45534
-    },
-    {
-      "epoch": 3.4814687386509164,
-      "grad_norm": 0.0015654724556952715,
-      "learning_rate": 0.00019999402138877614,
-      "loss": 46.0,
-      "step": 45535
-    },
-    {
-      "epoch": 3.481545195634306,
-      "grad_norm": 0.0012794616632163525,
-      "learning_rate": 0.0001999940211261241,
-      "loss": 46.0,
-      "step": 45536
-    },
-    {
-      "epoch": 3.481621652617696,
-      "grad_norm": 0.0014855306362733245,
-      "learning_rate": 0.00019999402086346625,
-      "loss": 46.0,
-      "step": 45537
-    },
-    {
-      "epoch": 3.4816981096010857,
-      "grad_norm": 0.002996789524331689,
-      "learning_rate": 0.0001999940206008027,
-      "loss": 46.0,
-      "step": 45538
-    },
-    {
-      "epoch": 3.4817745665844755,
-      "grad_norm": 0.004285655450075865,
-      "learning_rate": 0.00019999402033813333,
-      "loss": 46.0,
-      "step": 45539
-    },
-    {
-      "epoch": 3.4818510235678652,
-      "grad_norm": 0.00040705816354602575,
-      "learning_rate": 0.0001999940200754582,
-      "loss": 46.0,
-      "step": 45540
-    },
-    {
-      "epoch": 3.481927480551255,
-      "grad_norm": 0.007175622973591089,
-      "learning_rate": 0.0001999940198127773,
-      "loss": 46.0,
-      "step": 45541
-    },
-    {
-      "epoch": 3.4820039375346443,
-      "grad_norm": 0.0025255833752453327,
-      "learning_rate": 0.00019999401955009063,
-      "loss": 46.0,
-      "step": 45542
-    },
-    {
-      "epoch": 3.482080394518034,
-      "grad_norm": 0.005446826573461294,
-      "learning_rate": 0.00019999401928739817,
-      "loss": 46.0,
-      "step": 45543
-    },
-    {
-      "epoch": 3.482156851501424,
-      "grad_norm": 0.00160220917314291,
-      "learning_rate": 0.0001999940190247,
-      "loss": 46.0,
-      "step": 45544
-    },
-    {
-      "epoch": 3.4822333084848136,
-      "grad_norm": 0.001040255301631987,
-      "learning_rate": 0.00019999401876199602,
-      "loss": 46.0,
-      "step": 45545
-    },
-    {
-      "epoch": 3.4823097654682034,
-      "grad_norm": 0.0009551150724291801,
-      "learning_rate": 0.0001999940184992863,
-      "loss": 46.0,
-      "step": 45546
-    },
-    {
-      "epoch": 3.482386222451593,
-      "grad_norm": 0.0016583276446908712,
-      "learning_rate": 0.00019999401823657075,
-      "loss": 46.0,
-      "step": 45547
-    },
-    {
-      "epoch": 3.482462679434983,
-      "grad_norm": 0.0017576119862496853,
-      "learning_rate": 0.00019999401797384948,
-      "loss": 46.0,
-      "step": 45548
-    },
-    {
-      "epoch": 3.4825391364183726,
-      "grad_norm": 0.0009687562705948949,
-      "learning_rate": 0.00019999401771112244,
-      "loss": 46.0,
-      "step": 45549
-    },
-    {
-      "epoch": 3.4826155934017624,
-      "grad_norm": 0.0018143875058740377,
-      "learning_rate": 0.00019999401744838963,
-      "loss": 46.0,
-      "step": 45550
-    },
-    {
-      "epoch": 3.482692050385152,
-      "grad_norm": 0.003955061547458172,
-      "learning_rate": 0.00019999401718565104,
-      "loss": 46.0,
-      "step": 45551
-    },
-    {
-      "epoch": 3.482768507368542,
-      "grad_norm": 0.001455305959098041,
-      "learning_rate": 0.0001999940169229067,
-      "loss": 46.0,
-      "step": 45552
-    },
-    {
-      "epoch": 3.4828449643519317,
-      "grad_norm": 0.0017940544057637453,
-      "learning_rate": 0.00019999401666015657,
-      "loss": 46.0,
-      "step": 45553
-    },
-    {
-      "epoch": 3.4829214213353215,
-      "grad_norm": 0.00205013039521873,
-      "learning_rate": 0.00019999401639740066,
-      "loss": 46.0,
-      "step": 45554
-    },
-    {
-      "epoch": 3.4829978783187108,
-      "grad_norm": 0.000726459373254329,
-      "learning_rate": 0.000199994016134639,
-      "loss": 46.0,
-      "step": 45555
-    },
-    {
-      "epoch": 3.4830743353021005,
-      "grad_norm": 0.0011215859558433294,
-      "learning_rate": 0.00019999401587187158,
-      "loss": 46.0,
-      "step": 45556
-    },
-    {
-      "epoch": 3.4831507922854903,
-      "grad_norm": 0.0013050147099420428,
-      "learning_rate": 0.00019999401560909835,
-      "loss": 46.0,
-      "step": 45557
-    },
-    {
-      "epoch": 3.48322724926888,
-      "grad_norm": 0.004671876318752766,
-      "learning_rate": 0.00019999401534631937,
-      "loss": 46.0,
-      "step": 45558
-    },
-    {
-      "epoch": 3.48330370625227,
-      "grad_norm": 0.0018706230912357569,
-      "learning_rate": 0.00019999401508353465,
-      "loss": 46.0,
-      "step": 45559
-    },
-    {
-      "epoch": 3.4833801632356596,
-      "grad_norm": 0.0017690241802483797,
-      "learning_rate": 0.00019999401482074413,
-      "loss": 46.0,
-      "step": 45560
-    },
-    {
-      "epoch": 3.4834566202190493,
-      "grad_norm": 0.0016352059319615364,
-      "learning_rate": 0.00019999401455794783,
-      "loss": 46.0,
-      "step": 45561
-    },
-    {
-      "epoch": 3.483533077202439,
-      "grad_norm": 0.0016982720699161291,
-      "learning_rate": 0.0001999940142951458,
-      "loss": 46.0,
-      "step": 45562
-    },
-    {
-      "epoch": 3.483609534185829,
-      "grad_norm": 0.002577295759692788,
-      "learning_rate": 0.000199994014032338,
-      "loss": 46.0,
-      "step": 45563
-    },
-    {
-      "epoch": 3.483685991169218,
-      "grad_norm": 0.0022148285061120987,
-      "learning_rate": 0.0001999940137695244,
-      "loss": 46.0,
-      "step": 45564
-    },
-    {
-      "epoch": 3.483762448152608,
-      "grad_norm": 0.0012893271632492542,
-      "learning_rate": 0.00019999401350670506,
-      "loss": 46.0,
-      "step": 45565
-    },
-    {
-      "epoch": 3.4838389051359977,
-      "grad_norm": 0.0050181616097688675,
-      "learning_rate": 0.00019999401324387992,
-      "loss": 46.0,
-      "step": 45566
-    },
-    {
-      "epoch": 3.4839153621193875,
-      "grad_norm": 0.003126205410808325,
-      "learning_rate": 0.00019999401298104904,
-      "loss": 46.0,
-      "step": 45567
-    },
-    {
-      "epoch": 3.483991819102777,
-      "grad_norm": 0.0019700892735272646,
-      "learning_rate": 0.00019999401271821236,
-      "loss": 46.0,
-      "step": 45568
-    },
-    {
-      "epoch": 3.484068276086167,
-      "grad_norm": 0.002283578971400857,
-      "learning_rate": 0.00019999401245536996,
-      "loss": 46.0,
-      "step": 45569
-    },
-    {
-      "epoch": 3.4841447330695567,
-      "grad_norm": 0.002191627398133278,
-      "learning_rate": 0.00019999401219252173,
-      "loss": 46.0,
-      "step": 45570
-    },
-    {
-      "epoch": 3.4842211900529465,
-      "grad_norm": 0.0018031207146123052,
-      "learning_rate": 0.00019999401192966775,
-      "loss": 46.0,
-      "step": 45571
-    },
-    {
-      "epoch": 3.4842976470363363,
-      "grad_norm": 0.0018127438379451632,
-      "learning_rate": 0.00019999401166680803,
-      "loss": 46.0,
-      "step": 45572
-    },
-    {
-      "epoch": 3.484374104019726,
-      "grad_norm": 0.002399756805971265,
-      "learning_rate": 0.0001999940114039425,
-      "loss": 46.0,
-      "step": 45573
-    },
-    {
-      "epoch": 3.484450561003116,
-      "grad_norm": 0.004946081433445215,
-      "learning_rate": 0.00019999401114107124,
-      "loss": 46.0,
-      "step": 45574
-    },
-    {
-      "epoch": 3.4845270179865055,
-      "grad_norm": 0.0016603940166532993,
-      "learning_rate": 0.00019999401087819418,
-      "loss": 46.0,
-      "step": 45575
-    },
-    {
-      "epoch": 3.4846034749698953,
-      "grad_norm": 0.0007328182109631598,
-      "learning_rate": 0.0001999940106153114,
-      "loss": 46.0,
-      "step": 45576
-    },
-    {
-      "epoch": 3.4846799319532846,
-      "grad_norm": 0.008194363676011562,
-      "learning_rate": 0.0001999940103524228,
-      "loss": 46.0,
-      "step": 45577
-    },
-    {
-      "epoch": 3.4847563889366744,
-      "grad_norm": 0.002520754700526595,
-      "learning_rate": 0.00019999401008952844,
-      "loss": 46.0,
-      "step": 45578
-    },
-    {
-      "epoch": 3.484832845920064,
-      "grad_norm": 0.0013752371305599809,
-      "learning_rate": 0.00019999400982662833,
-      "loss": 46.0,
-      "step": 45579
-    },
-    {
-      "epoch": 3.484909302903454,
-      "grad_norm": 0.0017104746075347066,
-      "learning_rate": 0.00019999400956372242,
-      "loss": 46.0,
-      "step": 45580
-    },
-    {
-      "epoch": 3.4849857598868437,
-      "grad_norm": 0.0018003035802394152,
-      "learning_rate": 0.00019999400930081077,
-      "loss": 46.0,
-      "step": 45581
-    },
-    {
-      "epoch": 3.4850622168702334,
-      "grad_norm": 0.002289471449330449,
-      "learning_rate": 0.00019999400903789334,
-      "loss": 46.0,
-      "step": 45582
-    },
-    {
-      "epoch": 3.485138673853623,
-      "grad_norm": 0.0019717414397746325,
-      "learning_rate": 0.00019999400877497014,
-      "loss": 46.0,
-      "step": 45583
-    },
-    {
-      "epoch": 3.485215130837013,
-      "grad_norm": 0.0035805185325443745,
-      "learning_rate": 0.00019999400851204117,
-      "loss": 46.0,
-      "step": 45584
-    },
-    {
-      "epoch": 3.4852915878204027,
-      "grad_norm": 0.0014128257753327489,
-      "learning_rate": 0.00019999400824910645,
-      "loss": 46.0,
-      "step": 45585
-    },
-    {
-      "epoch": 3.485368044803792,
-      "grad_norm": 0.0025227379519492388,
-      "learning_rate": 0.00019999400798616592,
-      "loss": 46.0,
-      "step": 45586
-    },
-    {
-      "epoch": 3.485444501787182,
-      "grad_norm": 0.0009539916645735502,
-      "learning_rate": 0.00019999400772321966,
-      "loss": 46.0,
-      "step": 45587
-    },
-    {
-      "epoch": 3.4855209587705716,
-      "grad_norm": 0.006139744073152542,
-      "learning_rate": 0.00019999400746026762,
-      "loss": 46.0,
-      "step": 45588
-    },
-    {
-      "epoch": 3.4855974157539613,
-      "grad_norm": 0.001246851752512157,
-      "learning_rate": 0.00019999400719730977,
-      "loss": 46.0,
-      "step": 45589
-    },
-    {
-      "epoch": 3.485673872737351,
-      "grad_norm": 0.0020244382321834564,
-      "learning_rate": 0.00019999400693434621,
-      "loss": 46.0,
-      "step": 45590
-    },
-    {
-      "epoch": 3.485750329720741,
-      "grad_norm": 0.0023067768197506666,
-      "learning_rate": 0.00019999400667137685,
-      "loss": 46.0,
-      "step": 45591
-    },
-    {
-      "epoch": 3.4858267867041306,
-      "grad_norm": 0.001433484605513513,
-      "learning_rate": 0.00019999400640840175,
-      "loss": 46.0,
-      "step": 45592
-    },
-    {
-      "epoch": 3.4859032436875204,
-      "grad_norm": 0.002376784570515156,
-      "learning_rate": 0.00019999400614542087,
-      "loss": 46.0,
-      "step": 45593
-    },
-    {
-      "epoch": 3.48597970067091,
-      "grad_norm": 0.0032361838966608047,
-      "learning_rate": 0.0001999940058824342,
-      "loss": 46.0,
-      "step": 45594
-    },
-    {
-      "epoch": 3.4860561576543,
-      "grad_norm": 0.0020656774286180735,
-      "learning_rate": 0.00019999400561944178,
-      "loss": 46.0,
-      "step": 45595
-    },
-    {
-      "epoch": 3.4861326146376896,
-      "grad_norm": 0.0014405029360204935,
-      "learning_rate": 0.00019999400535644356,
-      "loss": 46.0,
-      "step": 45596
-    },
-    {
-      "epoch": 3.4862090716210794,
-      "grad_norm": 0.002200776943936944,
-      "learning_rate": 0.00019999400509343958,
-      "loss": 46.0,
-      "step": 45597
-    },
-    {
-      "epoch": 3.4862855286044687,
-      "grad_norm": 0.001321990042924881,
-      "learning_rate": 0.00019999400483042984,
-      "loss": 46.0,
-      "step": 45598
-    },
-    {
-      "epoch": 3.4863619855878585,
-      "grad_norm": 0.0026791421696543694,
-      "learning_rate": 0.00019999400456741437,
-      "loss": 46.0,
-      "step": 45599
-    },
-    {
-      "epoch": 3.4864384425712482,
-      "grad_norm": 0.002942162798717618,
-      "learning_rate": 0.00019999400430439307,
-      "loss": 46.0,
-      "step": 45600
-    },
-    {
-      "epoch": 3.486514899554638,
-      "grad_norm": 0.0013769024517387152,
-      "learning_rate": 0.00019999400404136603,
-      "loss": 46.0,
-      "step": 45601
-    },
-    {
-      "epoch": 3.4865913565380278,
-      "grad_norm": 0.0015389503678306937,
-      "learning_rate": 0.00019999400377833322,
-      "loss": 46.0,
-      "step": 45602
-    },
-    {
-      "epoch": 3.4866678135214175,
-      "grad_norm": 0.0013089972781017423,
-      "learning_rate": 0.00019999400351529466,
-      "loss": 46.0,
-      "step": 45603
-    },
-    {
-      "epoch": 3.4867442705048073,
-      "grad_norm": 0.0017713319975882769,
-      "learning_rate": 0.0001999940032522503,
-      "loss": 46.0,
-      "step": 45604
-    },
-    {
-      "epoch": 3.486820727488197,
-      "grad_norm": 0.0012855998938903213,
-      "learning_rate": 0.0001999940029892002,
-      "loss": 46.0,
-      "step": 45605
-    },
-    {
-      "epoch": 3.486897184471587,
-      "grad_norm": 0.0017893895274028182,
-      "learning_rate": 0.0001999940027261443,
-      "loss": 46.0,
-      "step": 45606
-    },
-    {
-      "epoch": 3.4869736414549766,
-      "grad_norm": 0.003058493137359619,
-      "learning_rate": 0.00019999400246308263,
-      "loss": 46.0,
-      "step": 45607
-    },
-    {
-      "epoch": 3.487050098438366,
-      "grad_norm": 0.0010416876757517457,
-      "learning_rate": 0.0001999940022000152,
-      "loss": 46.0,
-      "step": 45608
-    },
-    {
-      "epoch": 3.4871265554217556,
-      "grad_norm": 0.0020272028632462025,
-      "learning_rate": 0.000199994001936942,
-      "loss": 46.0,
-      "step": 45609
-    },
-    {
-      "epoch": 3.4872030124051454,
-      "grad_norm": 0.0015325198182836175,
-      "learning_rate": 0.00019999400167386303,
-      "loss": 46.0,
-      "step": 45610
-    },
-    {
-      "epoch": 3.487279469388535,
-      "grad_norm": 0.006384000647813082,
-      "learning_rate": 0.0001999940014107783,
-      "loss": 46.0,
-      "step": 45611
-    },
-    {
-      "epoch": 3.487355926371925,
-      "grad_norm": 0.003559096483513713,
-      "learning_rate": 0.00019999400114768782,
-      "loss": 46.0,
-      "step": 45612
-    },
-    {
-      "epoch": 3.4874323833553147,
-      "grad_norm": 0.0029365464579313993,
-      "learning_rate": 0.00019999400088459152,
-      "loss": 46.0,
-      "step": 45613
-    },
-    {
-      "epoch": 3.4875088403387045,
-      "grad_norm": 0.0008497110102325678,
-      "learning_rate": 0.0001999940006214895,
-      "loss": 46.0,
-      "step": 45614
-    },
-    {
-      "epoch": 3.487585297322094,
-      "grad_norm": 0.0021616590674966574,
-      "learning_rate": 0.0001999940003583817,
-      "loss": 46.0,
-      "step": 45615
-    },
-    {
-      "epoch": 3.487661754305484,
-      "grad_norm": 0.0010892166756093502,
-      "learning_rate": 0.0001999940000952681,
-      "loss": 46.0,
-      "step": 45616
-    },
-    {
-      "epoch": 3.4877382112888737,
-      "grad_norm": 0.004651820287108421,
-      "learning_rate": 0.00019999399983214875,
-      "loss": 46.0,
-      "step": 45617
-    },
-    {
-      "epoch": 3.4878146682722635,
-      "grad_norm": 0.0022322628647089005,
-      "learning_rate": 0.00019999399956902364,
-      "loss": 46.0,
-      "step": 45618
-    },
-    {
-      "epoch": 3.4878911252556533,
-      "grad_norm": 0.002929776208475232,
-      "learning_rate": 0.00019999399930589276,
-      "loss": 46.0,
-      "step": 45619
-    },
-    {
-      "epoch": 3.4879675822390426,
-      "grad_norm": 0.005450090393424034,
-      "learning_rate": 0.00019999399904275608,
-      "loss": 46.0,
-      "step": 45620
-    },
-    {
-      "epoch": 3.4880440392224323,
-      "grad_norm": 0.0018909627106040716,
-      "learning_rate": 0.00019999399877961366,
-      "loss": 46.0,
-      "step": 45621
-    },
-    {
-      "epoch": 3.488120496205822,
-      "grad_norm": 0.0013319356366991997,
-      "learning_rate": 0.00019999399851646548,
-      "loss": 46.0,
-      "step": 45622
-    },
-    {
-      "epoch": 3.488196953189212,
-      "grad_norm": 0.003321138909086585,
-      "learning_rate": 0.0001999939982533115,
-      "loss": 46.0,
-      "step": 45623
-    },
-    {
-      "epoch": 3.4882734101726016,
-      "grad_norm": 0.0013424414210021496,
-      "learning_rate": 0.00019999399799015176,
-      "loss": 46.0,
-      "step": 45624
-    },
-    {
-      "epoch": 3.4883498671559914,
-      "grad_norm": 0.0034013569820672274,
-      "learning_rate": 0.00019999399772698627,
-      "loss": 46.0,
-      "step": 45625
-    },
-    {
-      "epoch": 3.488426324139381,
-      "grad_norm": 0.0010715972166508436,
-      "learning_rate": 0.00019999399746381503,
-      "loss": 46.0,
-      "step": 45626
-    },
-    {
-      "epoch": 3.488502781122771,
-      "grad_norm": 0.003295557340607047,
-      "learning_rate": 0.00019999399720063796,
-      "loss": 46.0,
-      "step": 45627
-    },
-    {
-      "epoch": 3.4885792381061607,
-      "grad_norm": 0.002561314729973674,
-      "learning_rate": 0.00019999399693745515,
-      "loss": 46.0,
-      "step": 45628
-    },
-    {
-      "epoch": 3.4886556950895504,
-      "grad_norm": 0.0029351990669965744,
-      "learning_rate": 0.0001999939966742666,
-      "loss": 46.0,
-      "step": 45629
-    },
-    {
-      "epoch": 3.4887321520729397,
-      "grad_norm": 0.004409323912113905,
-      "learning_rate": 0.00019999399641107223,
-      "loss": 46.0,
-      "step": 45630
-    },
-    {
-      "epoch": 3.4888086090563295,
-      "grad_norm": 0.0012277784990146756,
-      "learning_rate": 0.0001999939961478721,
-      "loss": 46.0,
-      "step": 45631
-    },
-    {
-      "epoch": 3.4888850660397193,
-      "grad_norm": 0.002300865249708295,
-      "learning_rate": 0.00019999399588466622,
-      "loss": 46.0,
-      "step": 45632
-    },
-    {
-      "epoch": 3.488961523023109,
-      "grad_norm": 0.0021461935248225927,
-      "learning_rate": 0.00019999399562145456,
-      "loss": 46.0,
-      "step": 45633
-    },
-    {
-      "epoch": 3.489037980006499,
-      "grad_norm": 0.0006051812088117003,
-      "learning_rate": 0.00019999399535823717,
-      "loss": 46.0,
-      "step": 45634
-    },
-    {
-      "epoch": 3.4891144369898885,
-      "grad_norm": 0.0026908665895462036,
-      "learning_rate": 0.00019999399509501397,
-      "loss": 46.0,
-      "step": 45635
-    },
-    {
-      "epoch": 3.4891908939732783,
-      "grad_norm": 0.0023808779660612345,
-      "learning_rate": 0.000199993994831785,
-      "loss": 46.0,
-      "step": 45636
-    },
-    {
-      "epoch": 3.489267350956668,
-      "grad_norm": 0.002033111173659563,
-      "learning_rate": 0.00019999399456855027,
-      "loss": 46.0,
-      "step": 45637
-    },
-    {
-      "epoch": 3.489343807940058,
-      "grad_norm": 0.0018457021797075868,
-      "learning_rate": 0.00019999399430530975,
-      "loss": 46.0,
-      "step": 45638
-    },
-    {
-      "epoch": 3.4894202649234476,
-      "grad_norm": 0.0018687910633161664,
-      "learning_rate": 0.0001999939940420635,
-      "loss": 46.0,
-      "step": 45639
-    },
-    {
-      "epoch": 3.4894967219068374,
-      "grad_norm": 0.001373398001305759,
-      "learning_rate": 0.00019999399377881147,
-      "loss": 46.0,
-      "step": 45640
-    },
-    {
-      "epoch": 3.489573178890227,
-      "grad_norm": 0.0008144521270878613,
-      "learning_rate": 0.00019999399351555366,
-      "loss": 46.0,
-      "step": 45641
-    },
-    {
-      "epoch": 3.4896496358736164,
-      "grad_norm": 0.001465918612666428,
-      "learning_rate": 0.00019999399325229008,
-      "loss": 46.0,
-      "step": 45642
-    },
-    {
-      "epoch": 3.489726092857006,
-      "grad_norm": 0.0013177195796743035,
-      "learning_rate": 0.00019999399298902072,
-      "loss": 46.0,
-      "step": 45643
-    },
-    {
-      "epoch": 3.489802549840396,
-      "grad_norm": 0.000930647598579526,
-      "learning_rate": 0.0001999939927257456,
-      "loss": 46.0,
-      "step": 45644
-    },
-    {
-      "epoch": 3.4898790068237857,
-      "grad_norm": 0.002567956456914544,
-      "learning_rate": 0.00019999399246246473,
-      "loss": 46.0,
-      "step": 45645
-    },
-    {
-      "epoch": 3.4899554638071755,
-      "grad_norm": 0.002925882348790765,
-      "learning_rate": 0.00019999399219917808,
-      "loss": 46.0,
-      "step": 45646
-    },
-    {
-      "epoch": 3.4900319207905652,
-      "grad_norm": 0.0019909306429326534,
-      "learning_rate": 0.00019999399193588565,
-      "loss": 46.0,
-      "step": 45647
-    },
-    {
-      "epoch": 3.490108377773955,
-      "grad_norm": 0.0009055719710886478,
-      "learning_rate": 0.00019999399167258748,
-      "loss": 46.0,
-      "step": 45648
-    },
-    {
-      "epoch": 3.4901848347573448,
-      "grad_norm": 0.0018317571375519037,
-      "learning_rate": 0.0001999939914092835,
-      "loss": 46.0,
-      "step": 45649
-    },
-    {
-      "epoch": 3.4902612917407345,
-      "grad_norm": 0.002224199939519167,
-      "learning_rate": 0.00019999399114597376,
-      "loss": 46.0,
-      "step": 45650
-    },
-    {
-      "epoch": 3.4903377487241243,
-      "grad_norm": 0.0018536419374868274,
-      "learning_rate": 0.0001999939908826583,
-      "loss": 46.0,
-      "step": 45651
-    },
-    {
-      "epoch": 3.4904142057075136,
-      "grad_norm": 0.00274038128554821,
-      "learning_rate": 0.000199993990619337,
-      "loss": 46.0,
-      "step": 45652
-    },
-    {
-      "epoch": 3.4904906626909034,
-      "grad_norm": 0.001384408911690116,
-      "learning_rate": 0.00019999399035600996,
-      "loss": 46.0,
-      "step": 45653
-    },
-    {
-      "epoch": 3.490567119674293,
-      "grad_norm": 0.002141014439985156,
-      "learning_rate": 0.00019999399009267715,
-      "loss": 46.0,
-      "step": 45654
-    },
-    {
-      "epoch": 3.490643576657683,
-      "grad_norm": 0.00203525903634727,
-      "learning_rate": 0.0001999939898293386,
-      "loss": 46.0,
-      "step": 45655
-    },
-    {
-      "epoch": 3.4907200336410726,
-      "grad_norm": 0.004908971954137087,
-      "learning_rate": 0.00019999398956599423,
-      "loss": 46.0,
-      "step": 45656
-    },
-    {
-      "epoch": 3.4907964906244624,
-      "grad_norm": 0.0012880604481324553,
-      "learning_rate": 0.00019999398930264413,
-      "loss": 46.0,
-      "step": 45657
-    },
-    {
-      "epoch": 3.490872947607852,
-      "grad_norm": 0.002419913886114955,
-      "learning_rate": 0.00019999398903928825,
-      "loss": 46.0,
-      "step": 45658
-    },
-    {
-      "epoch": 3.490949404591242,
-      "grad_norm": 0.001576408394612372,
-      "learning_rate": 0.00019999398877592662,
-      "loss": 46.0,
-      "step": 45659
-    },
-    {
-      "epoch": 3.4910258615746317,
-      "grad_norm": 0.0012453236849978566,
-      "learning_rate": 0.0001999939885125592,
-      "loss": 46.0,
-      "step": 45660
-    },
-    {
-      "epoch": 3.4911023185580214,
-      "grad_norm": 0.0011135872919112444,
-      "learning_rate": 0.000199993988249186,
-      "loss": 46.0,
-      "step": 45661
-    },
-    {
-      "epoch": 3.491178775541411,
-      "grad_norm": 0.004250650759786367,
-      "learning_rate": 0.00019999398798580705,
-      "loss": 46.0,
-      "step": 45662
-    },
-    {
-      "epoch": 3.491255232524801,
-      "grad_norm": 0.0008822313393466175,
-      "learning_rate": 0.0001999939877224223,
-      "loss": 46.0,
-      "step": 45663
-    },
-    {
-      "epoch": 3.4913316895081903,
-      "grad_norm": 0.0008011918398551643,
-      "learning_rate": 0.0001999939874590318,
-      "loss": 46.0,
-      "step": 45664
-    },
-    {
-      "epoch": 3.49140814649158,
-      "grad_norm": 0.0026196420658379793,
-      "learning_rate": 0.00019999398719563555,
-      "loss": 46.0,
-      "step": 45665
-    },
-    {
-      "epoch": 3.49148460347497,
-      "grad_norm": 0.002392460824921727,
-      "learning_rate": 0.0001999939869322335,
-      "loss": 46.0,
-      "step": 45666
-    },
-    {
-      "epoch": 3.4915610604583596,
-      "grad_norm": 0.0029384037479758263,
-      "learning_rate": 0.00019999398666882572,
-      "loss": 46.0,
-      "step": 45667
-    },
-    {
-      "epoch": 3.4916375174417493,
-      "grad_norm": 0.003067390527576208,
-      "learning_rate": 0.0001999939864054121,
-      "loss": 46.0,
-      "step": 45668
-    },
-    {
-      "epoch": 3.491713974425139,
-      "grad_norm": 0.0007817524601705372,
-      "learning_rate": 0.00019999398614199278,
-      "loss": 46.0,
-      "step": 45669
-    },
-    {
-      "epoch": 3.491790431408529,
-      "grad_norm": 0.0009064714540727437,
-      "learning_rate": 0.00019999398587856767,
-      "loss": 46.0,
-      "step": 45670
-    },
-    {
-      "epoch": 3.4918668883919186,
-      "grad_norm": 0.002751914318650961,
-      "learning_rate": 0.0001999939856151368,
-      "loss": 46.0,
-      "step": 45671
-    },
-    {
-      "epoch": 3.4919433453753084,
-      "grad_norm": 0.0027233222499489784,
-      "learning_rate": 0.00019999398535170014,
-      "loss": 46.0,
-      "step": 45672
-    },
-    {
-      "epoch": 3.4920198023586977,
-      "grad_norm": 0.004428385756909847,
-      "learning_rate": 0.00019999398508825772,
-      "loss": 46.0,
-      "step": 45673
-    },
-    {
-      "epoch": 3.4920962593420874,
-      "grad_norm": 0.0024542836472392082,
-      "learning_rate": 0.00019999398482480952,
-      "loss": 46.0,
-      "step": 45674
-    },
-    {
-      "epoch": 3.492172716325477,
-      "grad_norm": 0.003983583766967058,
-      "learning_rate": 0.00019999398456135557,
-      "loss": 46.0,
-      "step": 45675
-    },
-    {
-      "epoch": 3.492249173308867,
-      "grad_norm": 0.005079290829598904,
-      "learning_rate": 0.00019999398429789586,
-      "loss": 46.0,
-      "step": 45676
-    },
-    {
-      "epoch": 3.4923256302922567,
-      "grad_norm": 0.0018149176612496376,
-      "learning_rate": 0.00019999398403443034,
-      "loss": 46.0,
-      "step": 45677
-    },
-    {
-      "epoch": 3.4924020872756465,
-      "grad_norm": 0.010486841201782227,
-      "learning_rate": 0.0001999939837709591,
-      "loss": 46.0,
-      "step": 45678
-    },
-    {
-      "epoch": 3.4924785442590363,
-      "grad_norm": 0.004170822910964489,
-      "learning_rate": 0.00019999398350748206,
-      "loss": 46.0,
-      "step": 45679
-    },
-    {
-      "epoch": 3.492555001242426,
-      "grad_norm": 0.003301135031506419,
-      "learning_rate": 0.00019999398324399925,
-      "loss": 46.0,
-      "step": 45680
-    },
-    {
-      "epoch": 3.4926314582258158,
-      "grad_norm": 0.0015523623442277312,
-      "learning_rate": 0.0001999939829805107,
-      "loss": 46.0,
-      "step": 45681
-    },
-    {
-      "epoch": 3.4927079152092055,
-      "grad_norm": 0.0013399841263890266,
-      "learning_rate": 0.00019999398271701636,
-      "loss": 46.0,
-      "step": 45682
-    },
-    {
-      "epoch": 3.4927843721925953,
-      "grad_norm": 0.002554279286414385,
-      "learning_rate": 0.00019999398245351623,
-      "loss": 46.0,
-      "step": 45683
-    },
-    {
-      "epoch": 3.492860829175985,
-      "grad_norm": 0.003012260189279914,
-      "learning_rate": 0.00019999398219001035,
-      "loss": 46.0,
-      "step": 45684
-    },
-    {
-      "epoch": 3.492937286159375,
-      "grad_norm": 0.006762445904314518,
-      "learning_rate": 0.0001999939819264987,
-      "loss": 46.0,
-      "step": 45685
-    },
-    {
-      "epoch": 3.493013743142764,
-      "grad_norm": 0.001363743213005364,
-      "learning_rate": 0.0001999939816629813,
-      "loss": 46.0,
-      "step": 45686
-    },
-    {
-      "epoch": 3.493090200126154,
-      "grad_norm": 0.0015602060593664646,
-      "learning_rate": 0.0001999939813994581,
-      "loss": 46.0,
-      "step": 45687
-    },
-    {
-      "epoch": 3.4931666571095437,
-      "grad_norm": 0.0019296335522085428,
-      "learning_rate": 0.00019999398113592916,
-      "loss": 46.0,
-      "step": 45688
-    },
-    {
-      "epoch": 3.4932431140929334,
-      "grad_norm": 0.002067565219476819,
-      "learning_rate": 0.00019999398087239444,
-      "loss": 46.0,
-      "step": 45689
-    },
-    {
-      "epoch": 3.493319571076323,
-      "grad_norm": 0.006629869341850281,
-      "learning_rate": 0.00019999398060885395,
-      "loss": 46.0,
-      "step": 45690
-    },
-    {
-      "epoch": 3.493396028059713,
-      "grad_norm": 0.0012023057788610458,
-      "learning_rate": 0.00019999398034530768,
-      "loss": 46.0,
-      "step": 45691
-    },
-    {
-      "epoch": 3.4934724850431027,
-      "grad_norm": 0.0030029453337192535,
-      "learning_rate": 0.00019999398008175562,
-      "loss": 46.0,
-      "step": 45692
-    },
-    {
-      "epoch": 3.4935489420264925,
-      "grad_norm": 0.001269257627427578,
-      "learning_rate": 0.00019999397981819783,
-      "loss": 46.0,
-      "step": 45693
-    },
-    {
-      "epoch": 3.4936253990098822,
-      "grad_norm": 0.00343915494158864,
-      "learning_rate": 0.00019999397955463425,
-      "loss": 46.0,
-      "step": 45694
-    },
-    {
-      "epoch": 3.4937018559932715,
-      "grad_norm": 0.003538409247994423,
-      "learning_rate": 0.00019999397929106491,
-      "loss": 46.0,
-      "step": 45695
-    },
-    {
-      "epoch": 3.4937783129766613,
-      "grad_norm": 0.002054263139143586,
-      "learning_rate": 0.0001999939790274898,
-      "loss": 46.0,
-      "step": 45696
-    },
-    {
-      "epoch": 3.493854769960051,
-      "grad_norm": 0.0019038824830204248,
-      "learning_rate": 0.00019999397876390896,
-      "loss": 46.0,
-      "step": 45697
-    },
-    {
-      "epoch": 3.493931226943441,
-      "grad_norm": 0.0010791143868118525,
-      "learning_rate": 0.00019999397850032228,
-      "loss": 46.0,
-      "step": 45698
-    },
-    {
-      "epoch": 3.4940076839268306,
-      "grad_norm": 0.001404003705829382,
-      "learning_rate": 0.00019999397823672988,
-      "loss": 46.0,
-      "step": 45699
-    },
-    {
-      "epoch": 3.4940841409102203,
-      "grad_norm": 0.00282001169398427,
-      "learning_rate": 0.00019999397797313168,
-      "loss": 46.0,
-      "step": 45700
-    },
-    {
-      "epoch": 3.49416059789361,
-      "grad_norm": 0.0023398713674396276,
-      "learning_rate": 0.00019999397770952774,
-      "loss": 46.0,
-      "step": 45701
-    },
-    {
-      "epoch": 3.494237054877,
-      "grad_norm": 0.0038184532895684242,
-      "learning_rate": 0.00019999397744591802,
-      "loss": 46.0,
-      "step": 45702
-    },
-    {
-      "epoch": 3.4943135118603896,
-      "grad_norm": 0.0041579315438866615,
-      "learning_rate": 0.00019999397718230253,
-      "loss": 46.0,
-      "step": 45703
-    },
-    {
-      "epoch": 3.4943899688437794,
-      "grad_norm": 0.0027121994644403458,
-      "learning_rate": 0.00019999397691868127,
-      "loss": 46.0,
-      "step": 45704
-    },
-    {
-      "epoch": 3.494466425827169,
-      "grad_norm": 0.0014264617348089814,
-      "learning_rate": 0.00019999397665505423,
-      "loss": 46.0,
-      "step": 45705
-    },
-    {
-      "epoch": 3.494542882810559,
-      "grad_norm": 0.003969315439462662,
-      "learning_rate": 0.00019999397639142144,
-      "loss": 46.0,
-      "step": 45706
-    },
-    {
-      "epoch": 3.4946193397939487,
-      "grad_norm": 0.0018643800867721438,
-      "learning_rate": 0.00019999397612778289,
-      "loss": 46.0,
-      "step": 45707
-    },
-    {
-      "epoch": 3.494695796777338,
-      "grad_norm": 0.0024837409146130085,
-      "learning_rate": 0.00019999397586413853,
-      "loss": 46.0,
-      "step": 45708
-    },
-    {
-      "epoch": 3.4947722537607278,
-      "grad_norm": 0.0015021396102383733,
-      "learning_rate": 0.00019999397560048842,
-      "loss": 46.0,
-      "step": 45709
-    },
-    {
-      "epoch": 3.4948487107441175,
-      "grad_norm": 0.004136497620493174,
-      "learning_rate": 0.00019999397533683255,
-      "loss": 46.0,
-      "step": 45710
-    },
-    {
-      "epoch": 3.4949251677275073,
-      "grad_norm": 0.0015473722014576197,
-      "learning_rate": 0.0001999939750731709,
-      "loss": 46.0,
-      "step": 45711
-    },
-    {
-      "epoch": 3.495001624710897,
-      "grad_norm": 0.00628721434623003,
-      "learning_rate": 0.00019999397480950347,
-      "loss": 46.0,
-      "step": 45712
-    },
-    {
-      "epoch": 3.495078081694287,
-      "grad_norm": 0.0025035073049366474,
-      "learning_rate": 0.0001999939745458303,
-      "loss": 46.0,
-      "step": 45713
-    },
-    {
-      "epoch": 3.4951545386776766,
-      "grad_norm": 0.0012276730267331004,
-      "learning_rate": 0.00019999397428215136,
-      "loss": 46.0,
-      "step": 45714
-    },
-    {
-      "epoch": 3.4952309956610663,
-      "grad_norm": 0.001920933835208416,
-      "learning_rate": 0.00019999397401846664,
-      "loss": 46.0,
-      "step": 45715
-    },
-    {
-      "epoch": 3.495307452644456,
-      "grad_norm": 0.0023868796415627003,
-      "learning_rate": 0.00019999397375477615,
-      "loss": 46.0,
-      "step": 45716
-    },
-    {
-      "epoch": 3.4953839096278454,
-      "grad_norm": 0.0030802234541624784,
-      "learning_rate": 0.00019999397349107988,
-      "loss": 46.0,
-      "step": 45717
-    },
-    {
-      "epoch": 3.495460366611235,
-      "grad_norm": 0.004375213757157326,
-      "learning_rate": 0.00019999397322737785,
-      "loss": 46.0,
-      "step": 45718
-    },
-    {
-      "epoch": 3.495536823594625,
-      "grad_norm": 0.004127439111471176,
-      "learning_rate": 0.00019999397296367006,
-      "loss": 46.0,
-      "step": 45719
-    },
-    {
-      "epoch": 3.4956132805780147,
-      "grad_norm": 0.003457908285781741,
-      "learning_rate": 0.0001999939726999565,
-      "loss": 46.0,
-      "step": 45720
-    },
-    {
-      "epoch": 3.4956897375614044,
-      "grad_norm": 0.0011295400327071548,
-      "learning_rate": 0.00019999397243623715,
-      "loss": 46.0,
-      "step": 45721
-    },
-    {
-      "epoch": 3.495766194544794,
-      "grad_norm": 0.0027088159695267677,
-      "learning_rate": 0.00019999397217251207,
-      "loss": 46.0,
-      "step": 45722
-    },
-    {
-      "epoch": 3.495842651528184,
-      "grad_norm": 0.0008764645899645984,
-      "learning_rate": 0.00019999397190878117,
-      "loss": 46.0,
-      "step": 45723
-    },
-    {
-      "epoch": 3.4959191085115737,
-      "grad_norm": 0.002637633355334401,
-      "learning_rate": 0.00019999397164504451,
-      "loss": 46.0,
-      "step": 45724
-    },
-    {
-      "epoch": 3.4959955654949635,
-      "grad_norm": 0.0029878916684538126,
-      "learning_rate": 0.00019999397138130212,
-      "loss": 46.0,
-      "step": 45725
-    },
-    {
-      "epoch": 3.4960720224783532,
-      "grad_norm": 0.001272466965019703,
-      "learning_rate": 0.00019999397111755395,
-      "loss": 46.0,
-      "step": 45726
-    },
-    {
-      "epoch": 3.496148479461743,
-      "grad_norm": 0.0015955364797264338,
-      "learning_rate": 0.00019999397085379998,
-      "loss": 46.0,
-      "step": 45727
-    },
-    {
-      "epoch": 3.4962249364451328,
-      "grad_norm": 0.0010294899111613631,
-      "learning_rate": 0.00019999397059004026,
-      "loss": 46.0,
-      "step": 45728
-    },
-    {
-      "epoch": 3.496301393428522,
-      "grad_norm": 0.0020012736786156893,
-      "learning_rate": 0.0001999939703262748,
-      "loss": 46.0,
-      "step": 45729
-    },
-    {
-      "epoch": 3.496377850411912,
-      "grad_norm": 0.0032012115698307753,
-      "learning_rate": 0.00019999397006250353,
-      "loss": 46.0,
-      "step": 45730
-    },
-    {
-      "epoch": 3.4964543073953016,
-      "grad_norm": 0.0018668158445507288,
-      "learning_rate": 0.0001999939697987265,
-      "loss": 46.0,
-      "step": 45731
-    },
-    {
-      "epoch": 3.4965307643786914,
-      "grad_norm": 0.0019320003921166062,
-      "learning_rate": 0.0001999939695349437,
-      "loss": 46.0,
-      "step": 45732
-    },
-    {
-      "epoch": 3.496607221362081,
-      "grad_norm": 0.002725800732150674,
-      "learning_rate": 0.00019999396927115513,
-      "loss": 46.0,
-      "step": 45733
-    },
-    {
-      "epoch": 3.496683678345471,
-      "grad_norm": 0.0011421084636822343,
-      "learning_rate": 0.0001999939690073608,
-      "loss": 46.0,
-      "step": 45734
-    },
-    {
-      "epoch": 3.4967601353288607,
-      "grad_norm": 0.0022823053877800703,
-      "learning_rate": 0.00019999396874356072,
-      "loss": 46.0,
-      "step": 45735
-    },
-    {
-      "epoch": 3.4968365923122504,
-      "grad_norm": 0.0016514388844370842,
-      "learning_rate": 0.00019999396847975482,
-      "loss": 46.0,
-      "step": 45736
-    },
-    {
-      "epoch": 3.49691304929564,
-      "grad_norm": 0.002985776402056217,
-      "learning_rate": 0.0001999939682159432,
-      "loss": 46.0,
-      "step": 45737
-    },
-    {
-      "epoch": 3.49698950627903,
-      "grad_norm": 0.001532039255835116,
-      "learning_rate": 0.00019999396795212577,
-      "loss": 46.0,
-      "step": 45738
-    },
-    {
-      "epoch": 3.4970659632624193,
-      "grad_norm": 0.0018588689854368567,
-      "learning_rate": 0.0001999939676883026,
-      "loss": 46.0,
-      "step": 45739
-    },
-    {
-      "epoch": 3.497142420245809,
-      "grad_norm": 0.002633602824062109,
-      "learning_rate": 0.00019999396742447365,
-      "loss": 46.0,
-      "step": 45740
-    },
-    {
-      "epoch": 3.4972188772291988,
-      "grad_norm": 0.0021706006955355406,
-      "learning_rate": 0.0001999939671606389,
-      "loss": 46.0,
-      "step": 45741
-    },
-    {
-      "epoch": 3.4972953342125885,
-      "grad_norm": 0.002011735225096345,
-      "learning_rate": 0.00019999396689679845,
-      "loss": 46.0,
-      "step": 45742
-    },
-    {
-      "epoch": 3.4973717911959783,
-      "grad_norm": 0.0009335528593510389,
-      "learning_rate": 0.00019999396663295218,
-      "loss": 46.0,
-      "step": 45743
-    },
-    {
-      "epoch": 3.497448248179368,
-      "grad_norm": 0.00796280987560749,
-      "learning_rate": 0.00019999396636910017,
-      "loss": 46.0,
-      "step": 45744
-    },
-    {
-      "epoch": 3.497524705162758,
-      "grad_norm": 0.004227794706821442,
-      "learning_rate": 0.00019999396610524234,
-      "loss": 46.0,
-      "step": 45745
-    },
-    {
-      "epoch": 3.4976011621461476,
-      "grad_norm": 0.004875116981565952,
-      "learning_rate": 0.0001999939658413788,
-      "loss": 46.0,
-      "step": 45746
-    },
-    {
-      "epoch": 3.4976776191295373,
-      "grad_norm": 0.0061309547163546085,
-      "learning_rate": 0.00019999396557750948,
-      "loss": 46.0,
-      "step": 45747
-    },
-    {
-      "epoch": 3.497754076112927,
-      "grad_norm": 0.0017395709874108434,
-      "learning_rate": 0.00019999396531363437,
-      "loss": 46.0,
-      "step": 45748
-    },
-    {
-      "epoch": 3.497830533096317,
-      "grad_norm": 0.0016545196995139122,
-      "learning_rate": 0.0001999939650497535,
-      "loss": 46.0,
-      "step": 45749
-    },
-    {
-      "epoch": 3.4979069900797066,
-      "grad_norm": 0.0017746618250384927,
-      "learning_rate": 0.00019999396478586685,
-      "loss": 46.0,
-      "step": 45750
-    },
-    {
-      "epoch": 3.497983447063096,
-      "grad_norm": 0.0034066583029925823,
-      "learning_rate": 0.00019999396452197442,
-      "loss": 46.0,
-      "step": 45751
-    },
-    {
-      "epoch": 3.4980599040464857,
-      "grad_norm": 0.0030231866985559464,
-      "learning_rate": 0.00019999396425807628,
-      "loss": 46.0,
-      "step": 45752
-    },
-    {
-      "epoch": 3.4981363610298755,
-      "grad_norm": 0.0035011935979127884,
-      "learning_rate": 0.0001999939639941723,
-      "loss": 46.0,
-      "step": 45753
-    },
-    {
-      "epoch": 3.4982128180132652,
-      "grad_norm": 0.0009037017589434981,
-      "learning_rate": 0.0001999939637302626,
-      "loss": 46.0,
-      "step": 45754
-    },
-    {
-      "epoch": 3.498289274996655,
-      "grad_norm": 0.0013911626301705837,
-      "learning_rate": 0.0001999939634663471,
-      "loss": 46.0,
-      "step": 45755
-    },
-    {
-      "epoch": 3.4983657319800447,
-      "grad_norm": 0.003283183556050062,
-      "learning_rate": 0.00019999396320242587,
-      "loss": 46.0,
-      "step": 45756
-    },
-    {
-      "epoch": 3.4984421889634345,
-      "grad_norm": 0.0009282208629883826,
-      "learning_rate": 0.00019999396293849883,
-      "loss": 46.0,
-      "step": 45757
-    },
-    {
-      "epoch": 3.4985186459468243,
-      "grad_norm": 0.0017220533918589354,
-      "learning_rate": 0.00019999396267456605,
-      "loss": 46.0,
-      "step": 45758
-    },
-    {
-      "epoch": 3.498595102930214,
-      "grad_norm": 0.0009883064776659012,
-      "learning_rate": 0.0001999939624106275,
-      "loss": 46.0,
-      "step": 45759
-    },
-    {
-      "epoch": 3.498671559913604,
-      "grad_norm": 0.0026350603438913822,
-      "learning_rate": 0.00019999396214668316,
-      "loss": 46.0,
-      "step": 45760
-    },
-    {
-      "epoch": 3.498748016896993,
-      "grad_norm": 0.0024470877833664417,
-      "learning_rate": 0.00019999396188273306,
-      "loss": 46.0,
-      "step": 45761
-    },
-    {
-      "epoch": 3.498824473880383,
-      "grad_norm": 0.001396100502461195,
-      "learning_rate": 0.00019999396161877718,
-      "loss": 46.0,
-      "step": 45762
-    },
-    {
-      "epoch": 3.4989009308637726,
-      "grad_norm": 0.005022132303565741,
-      "learning_rate": 0.00019999396135481553,
-      "loss": 46.0,
-      "step": 45763
-    },
-    {
-      "epoch": 3.4989773878471624,
-      "grad_norm": 0.003914027940481901,
-      "learning_rate": 0.00019999396109084814,
-      "loss": 46.0,
-      "step": 45764
-    },
-    {
-      "epoch": 3.499053844830552,
-      "grad_norm": 0.0030109924264252186,
-      "learning_rate": 0.00019999396082687497,
-      "loss": 46.0,
-      "step": 45765
-    },
-    {
-      "epoch": 3.499130301813942,
-      "grad_norm": 0.004394937306642532,
-      "learning_rate": 0.00019999396056289602,
-      "loss": 46.0,
-      "step": 45766
-    },
-    {
-      "epoch": 3.4992067587973317,
-      "grad_norm": 0.005031343083828688,
-      "learning_rate": 0.00019999396029891128,
-      "loss": 46.0,
-      "step": 45767
-    },
-    {
-      "epoch": 3.4992832157807214,
-      "grad_norm": 0.0012765949359163642,
-      "learning_rate": 0.00019999396003492082,
-      "loss": 46.0,
-      "step": 45768
-    },
-    {
-      "epoch": 3.499359672764111,
-      "grad_norm": 0.0015599989565089345,
-      "learning_rate": 0.00019999395977092455,
-      "loss": 46.0,
-      "step": 45769
-    },
-    {
-      "epoch": 3.499436129747501,
-      "grad_norm": 0.0024923740420490503,
-      "learning_rate": 0.00019999395950692255,
-      "loss": 46.0,
-      "step": 45770
-    },
-    {
-      "epoch": 3.4995125867308907,
-      "grad_norm": 0.001404183334670961,
-      "learning_rate": 0.00019999395924291476,
-      "loss": 46.0,
-      "step": 45771
-    },
-    {
-      "epoch": 3.4995890437142805,
-      "grad_norm": 0.0009848756017163396,
-      "learning_rate": 0.00019999395897890118,
-      "loss": 46.0,
-      "step": 45772
-    },
-    {
-      "epoch": 3.49966550069767,
-      "grad_norm": 0.00337539822794497,
-      "learning_rate": 0.00019999395871488185,
-      "loss": 46.0,
-      "step": 45773
-    },
-    {
-      "epoch": 3.4997419576810596,
-      "grad_norm": 0.0011520797852426767,
-      "learning_rate": 0.00019999395845085677,
-      "loss": 46.0,
-      "step": 45774
-    },
-    {
-      "epoch": 3.4998184146644493,
-      "grad_norm": 0.003876384347677231,
-      "learning_rate": 0.0001999939581868259,
-      "loss": 46.0,
-      "step": 45775
-    },
-    {
-      "epoch": 3.499894871647839,
-      "grad_norm": 0.0012800614349544048,
-      "learning_rate": 0.00019999395792278925,
-      "loss": 46.0,
-      "step": 45776
-    },
-    {
-      "epoch": 3.499971328631229,
-      "grad_norm": 0.0025990649592131376,
-      "learning_rate": 0.00019999395765874685,
-      "loss": 46.0,
-      "step": 45777
-    },
-    {
-      "epoch": 3.5000477856146186,
-      "grad_norm": 0.0021888643968850374,
-      "learning_rate": 0.00019999395739469868,
-      "loss": 46.0,
-      "step": 45778
-    },
-    {
-      "epoch": 3.5001242425980084,
-      "grad_norm": 0.0012116662692278624,
-      "learning_rate": 0.00019999395713064474,
-      "loss": 46.0,
-      "step": 45779
-    },
-    {
-      "epoch": 3.500200699581398,
-      "grad_norm": 0.0014844882534816861,
-      "learning_rate": 0.00019999395686658503,
-      "loss": 46.0,
-      "step": 45780
-    },
-    {
-      "epoch": 3.500277156564788,
-      "grad_norm": 0.0033130068331956863,
-      "learning_rate": 0.00019999395660251954,
-      "loss": 46.0,
-      "step": 45781
-    },
-    {
-      "epoch": 3.500353613548177,
-      "grad_norm": 0.0022224008571356535,
-      "learning_rate": 0.00019999395633844827,
-      "loss": 46.0,
-      "step": 45782
-    },
-    {
-      "epoch": 3.500430070531567,
-      "grad_norm": 0.002328029600903392,
-      "learning_rate": 0.0001999939560743713,
-      "loss": 46.0,
-      "step": 45783
-    },
-    {
-      "epoch": 3.5005065275149567,
-      "grad_norm": 0.0027920312713831663,
-      "learning_rate": 0.00019999395581028848,
-      "loss": 46.0,
-      "step": 45784
-    },
-    {
-      "epoch": 3.5005829844983465,
-      "grad_norm": 0.002868940820917487,
-      "learning_rate": 0.0001999939555461999,
-      "loss": 46.0,
-      "step": 45785
-    },
-    {
-      "epoch": 3.5006594414817362,
-      "grad_norm": 0.005953640677034855,
-      "learning_rate": 0.0001999939552821056,
-      "loss": 46.0,
-      "step": 45786
-    },
-    {
-      "epoch": 3.500735898465126,
-      "grad_norm": 0.0015957440482452512,
-      "learning_rate": 0.0001999939550180055,
-      "loss": 46.0,
-      "step": 45787
-    },
-    {
-      "epoch": 3.5008123554485158,
-      "grad_norm": 0.0014333180151879787,
-      "learning_rate": 0.00019999395475389965,
-      "loss": 46.0,
-      "step": 45788
-    },
-    {
-      "epoch": 3.5008888124319055,
-      "grad_norm": 0.0007881660130806267,
-      "learning_rate": 0.000199993954489788,
-      "loss": 46.0,
-      "step": 45789
-    },
-    {
-      "epoch": 3.5009652694152953,
-      "grad_norm": 0.002509583020582795,
-      "learning_rate": 0.00019999395422567057,
-      "loss": 46.0,
-      "step": 45790
-    },
-    {
-      "epoch": 3.501041726398685,
-      "grad_norm": 0.0065762801095843315,
-      "learning_rate": 0.0001999939539615474,
-      "loss": 46.0,
-      "step": 45791
-    },
-    {
-      "epoch": 3.501118183382075,
-      "grad_norm": 0.0011103107826784253,
-      "learning_rate": 0.00019999395369741846,
-      "loss": 46.0,
-      "step": 45792
-    },
-    {
-      "epoch": 3.5011946403654646,
-      "grad_norm": 0.0006491128588095307,
-      "learning_rate": 0.00019999395343328375,
-      "loss": 46.0,
-      "step": 45793
-    },
-    {
-      "epoch": 3.5012710973488543,
-      "grad_norm": 0.0012044055620208383,
-      "learning_rate": 0.00019999395316914326,
-      "loss": 46.0,
-      "step": 45794
-    },
-    {
-      "epoch": 3.501347554332244,
-      "grad_norm": 0.0012121006147935987,
-      "learning_rate": 0.00019999395290499702,
-      "loss": 46.0,
-      "step": 45795
-    },
-    {
-      "epoch": 3.5014240113156334,
-      "grad_norm": 0.0018368646269664168,
-      "learning_rate": 0.00019999395264084501,
-      "loss": 46.0,
-      "step": 45796
-    },
-    {
-      "epoch": 3.501500468299023,
-      "grad_norm": 0.0037259759847074747,
-      "learning_rate": 0.0001999939523766872,
-      "loss": 46.0,
-      "step": 45797
-    },
-    {
-      "epoch": 3.501576925282413,
-      "grad_norm": 0.0025700132828205824,
-      "learning_rate": 0.00019999395211252368,
-      "loss": 46.0,
-      "step": 45798
-    },
-    {
-      "epoch": 3.5016533822658027,
-      "grad_norm": 0.001448381575755775,
-      "learning_rate": 0.00019999395184835435,
-      "loss": 46.0,
-      "step": 45799
-    },
-    {
-      "epoch": 3.5017298392491925,
-      "grad_norm": 0.0037338396068662405,
-      "learning_rate": 0.00019999395158417925,
-      "loss": 46.0,
-      "step": 45800
-    },
-    {
-      "epoch": 3.501806296232582,
-      "grad_norm": 0.0012148433597758412,
-      "learning_rate": 0.0001999939513199984,
-      "loss": 46.0,
-      "step": 45801
-    },
-    {
-      "epoch": 3.501882753215972,
-      "grad_norm": 0.0016613543266430497,
-      "learning_rate": 0.00019999395105581175,
-      "loss": 46.0,
-      "step": 45802
-    },
-    {
-      "epoch": 3.5019592101993617,
-      "grad_norm": 0.003074268577620387,
-      "learning_rate": 0.00019999395079161935,
-      "loss": 46.0,
-      "step": 45803
-    },
-    {
-      "epoch": 3.502035667182751,
-      "grad_norm": 0.0013266843743622303,
-      "learning_rate": 0.00019999395052742119,
-      "loss": 46.0,
-      "step": 45804
-    },
-    {
-      "epoch": 3.502112124166141,
-      "grad_norm": 0.0010522467782720923,
-      "learning_rate": 0.00019999395026321724,
-      "loss": 46.0,
-      "step": 45805
-    },
-    {
-      "epoch": 3.5021885811495306,
-      "grad_norm": 0.0018472321098670363,
-      "learning_rate": 0.0001999939499990075,
-      "loss": 46.0,
-      "step": 45806
-    },
-    {
-      "epoch": 3.5022650381329203,
-      "grad_norm": 0.0015084845945239067,
-      "learning_rate": 0.00019999394973479204,
-      "loss": 46.0,
-      "step": 45807
-    },
-    {
-      "epoch": 3.50234149511631,
-      "grad_norm": 0.0016131834127008915,
-      "learning_rate": 0.0001999939494705708,
-      "loss": 46.0,
-      "step": 45808
-    },
-    {
-      "epoch": 3.5024179520997,
-      "grad_norm": 0.002342240186408162,
-      "learning_rate": 0.00019999394920634377,
-      "loss": 46.0,
-      "step": 45809
-    },
-    {
-      "epoch": 3.5024944090830896,
-      "grad_norm": 0.0072229960933327675,
-      "learning_rate": 0.000199993948942111,
-      "loss": 46.0,
-      "step": 45810
-    },
-    {
-      "epoch": 3.5025708660664794,
-      "grad_norm": 0.0009944502962753177,
-      "learning_rate": 0.00019999394867787243,
-      "loss": 46.0,
-      "step": 45811
-    },
-    {
-      "epoch": 3.502647323049869,
-      "grad_norm": 0.0014579308917745948,
-      "learning_rate": 0.0001999939484136281,
-      "loss": 46.0,
-      "step": 45812
-    },
-    {
-      "epoch": 3.502723780033259,
-      "grad_norm": 0.003925422206521034,
-      "learning_rate": 0.00019999394814937803,
-      "loss": 46.0,
-      "step": 45813
-    },
-    {
-      "epoch": 3.5028002370166487,
-      "grad_norm": 0.0035908729769289494,
-      "learning_rate": 0.00019999394788512215,
-      "loss": 46.0,
-      "step": 45814
-    },
-    {
-      "epoch": 3.5028766940000384,
-      "grad_norm": 0.0029187118634581566,
-      "learning_rate": 0.00019999394762086053,
-      "loss": 46.0,
-      "step": 45815
-    },
-    {
-      "epoch": 3.502953150983428,
-      "grad_norm": 0.003749963827431202,
-      "learning_rate": 0.00019999394735659314,
-      "loss": 46.0,
-      "step": 45816
-    },
-    {
-      "epoch": 3.5030296079668175,
-      "grad_norm": 0.0014738740865141153,
-      "learning_rate": 0.00019999394709231994,
-      "loss": 46.0,
-      "step": 45817
-    },
-    {
-      "epoch": 3.5031060649502073,
-      "grad_norm": 0.0015019946731626987,
-      "learning_rate": 0.000199993946828041,
-      "loss": 46.0,
-      "step": 45818
-    },
-    {
-      "epoch": 3.503182521933597,
-      "grad_norm": 0.001335314242169261,
-      "learning_rate": 0.0001999939465637563,
-      "loss": 46.0,
-      "step": 45819
-    },
-    {
-      "epoch": 3.503258978916987,
-      "grad_norm": 0.0014708946691825986,
-      "learning_rate": 0.00019999394629946582,
-      "loss": 46.0,
-      "step": 45820
-    },
-    {
-      "epoch": 3.5033354359003765,
-      "grad_norm": 0.0023479245137423277,
-      "learning_rate": 0.0001999939460351696,
-      "loss": 46.0,
-      "step": 45821
-    },
-    {
-      "epoch": 3.5034118928837663,
-      "grad_norm": 0.002362276893109083,
-      "learning_rate": 0.00019999394577086755,
-      "loss": 46.0,
-      "step": 45822
-    },
-    {
-      "epoch": 3.503488349867156,
-      "grad_norm": 0.0014825870748609304,
-      "learning_rate": 0.00019999394550655977,
-      "loss": 46.0,
-      "step": 45823
-    },
-    {
-      "epoch": 3.503564806850546,
-      "grad_norm": 0.0012609621044248343,
-      "learning_rate": 0.00019999394524224622,
-      "loss": 46.0,
-      "step": 45824
-    },
-    {
-      "epoch": 3.5036412638339356,
-      "grad_norm": 0.0013599558733403683,
-      "learning_rate": 0.0001999939449779269,
-      "loss": 46.0,
-      "step": 45825
-    },
-    {
-      "epoch": 3.503717720817325,
-      "grad_norm": 0.0021276073530316353,
-      "learning_rate": 0.0001999939447136018,
-      "loss": 46.0,
-      "step": 45826
-    },
-    {
-      "epoch": 3.5037941778007147,
-      "grad_norm": 0.0017014875775203109,
-      "learning_rate": 0.00019999394444927094,
-      "loss": 46.0,
-      "step": 45827
-    },
-    {
-      "epoch": 3.5038706347841044,
-      "grad_norm": 0.0024813904892653227,
-      "learning_rate": 0.00019999394418493435,
-      "loss": 46.0,
-      "step": 45828
-    },
-    {
-      "epoch": 3.503947091767494,
-      "grad_norm": 0.0030645569786429405,
-      "learning_rate": 0.00019999394392059192,
-      "loss": 46.0,
-      "step": 45829
-    },
-    {
-      "epoch": 3.504023548750884,
-      "grad_norm": 0.0011163130402565002,
-      "learning_rate": 0.00019999394365624376,
-      "loss": 46.0,
-      "step": 45830
-    },
-    {
-      "epoch": 3.5041000057342737,
-      "grad_norm": 0.002176186302676797,
-      "learning_rate": 0.00019999394339188982,
-      "loss": 46.0,
-      "step": 45831
-    },
-    {
-      "epoch": 3.5041764627176635,
-      "grad_norm": 0.003148752963170409,
-      "learning_rate": 0.0001999939431275301,
-      "loss": 46.0,
-      "step": 45832
-    },
-    {
-      "epoch": 3.5042529197010532,
-      "grad_norm": 0.0027813708875328302,
-      "learning_rate": 0.0001999939428631646,
-      "loss": 46.0,
-      "step": 45833
-    },
-    {
-      "epoch": 3.504329376684443,
-      "grad_norm": 0.0033020740374922752,
-      "learning_rate": 0.00019999394259879338,
-      "loss": 46.0,
-      "step": 45834
-    },
-    {
-      "epoch": 3.5044058336678328,
-      "grad_norm": 0.0042078206315636635,
-      "learning_rate": 0.00019999394233441637,
-      "loss": 46.0,
-      "step": 45835
-    },
-    {
-      "epoch": 3.5044822906512225,
-      "grad_norm": 0.0032947277650237083,
-      "learning_rate": 0.0001999939420700336,
-      "loss": 46.0,
-      "step": 45836
-    },
-    {
-      "epoch": 3.5045587476346123,
-      "grad_norm": 0.006774098612368107,
-      "learning_rate": 0.00019999394180564504,
-      "loss": 46.0,
-      "step": 45837
-    },
-    {
-      "epoch": 3.504635204618002,
-      "grad_norm": 0.0024718046188354492,
-      "learning_rate": 0.0001999939415412507,
-      "loss": 46.0,
-      "step": 45838
-    },
-    {
-      "epoch": 3.5047116616013914,
-      "grad_norm": 0.003453444456681609,
-      "learning_rate": 0.00019999394127685063,
-      "loss": 46.0,
-      "step": 45839
-    },
-    {
-      "epoch": 3.504788118584781,
-      "grad_norm": 0.0018170345574617386,
-      "learning_rate": 0.00019999394101244479,
-      "loss": 46.0,
-      "step": 45840
-    },
-    {
-      "epoch": 3.504864575568171,
-      "grad_norm": 0.0023780278861522675,
-      "learning_rate": 0.00019999394074803314,
-      "loss": 46.0,
-      "step": 45841
-    },
-    {
-      "epoch": 3.5049410325515606,
-      "grad_norm": 0.0033883964642882347,
-      "learning_rate": 0.00019999394048361574,
-      "loss": 46.0,
-      "step": 45842
-    },
-    {
-      "epoch": 3.5050174895349504,
-      "grad_norm": 0.0029428484849631786,
-      "learning_rate": 0.00019999394021919258,
-      "loss": 46.0,
-      "step": 45843
-    },
-    {
-      "epoch": 3.50509394651834,
-      "grad_norm": 0.0027647584211081266,
-      "learning_rate": 0.00019999393995476364,
-      "loss": 46.0,
-      "step": 45844
-    },
-    {
-      "epoch": 3.50517040350173,
-      "grad_norm": 0.0025004716590046883,
-      "learning_rate": 0.00019999393969032895,
-      "loss": 46.0,
-      "step": 45845
-    },
-    {
-      "epoch": 3.5052468604851197,
-      "grad_norm": 0.0015785566065460443,
-      "learning_rate": 0.00019999393942588846,
-      "loss": 46.0,
-      "step": 45846
-    },
-    {
-      "epoch": 3.505323317468509,
-      "grad_norm": 0.0012097267899662256,
-      "learning_rate": 0.00019999393916144223,
-      "loss": 46.0,
-      "step": 45847
-    },
-    {
-      "epoch": 3.5053997744518988,
-      "grad_norm": 0.0017973168287426233,
-      "learning_rate": 0.00019999393889699022,
-      "loss": 46.0,
-      "step": 45848
-    },
-    {
-      "epoch": 3.5054762314352885,
-      "grad_norm": 0.0018133892444893718,
-      "learning_rate": 0.0001999939386325324,
-      "loss": 46.0,
-      "step": 45849
-    },
-    {
-      "epoch": 3.5055526884186783,
-      "grad_norm": 0.001698700594715774,
-      "learning_rate": 0.00019999393836806888,
-      "loss": 46.0,
-      "step": 45850
-    },
-    {
-      "epoch": 3.505629145402068,
-      "grad_norm": 0.0035927521530538797,
-      "learning_rate": 0.00019999393810359956,
-      "loss": 46.0,
-      "step": 45851
-    },
-    {
-      "epoch": 3.505705602385458,
-      "grad_norm": 0.003899917472153902,
-      "learning_rate": 0.00019999393783912446,
-      "loss": 46.0,
-      "step": 45852
-    },
-    {
-      "epoch": 3.5057820593688476,
-      "grad_norm": 0.003743167733773589,
-      "learning_rate": 0.0001999939375746436,
-      "loss": 46.0,
-      "step": 45853
-    },
-    {
-      "epoch": 3.5058585163522373,
-      "grad_norm": 0.006765828467905521,
-      "learning_rate": 0.00019999393731015696,
-      "loss": 46.0,
-      "step": 45854
-    },
-    {
-      "epoch": 3.505934973335627,
-      "grad_norm": 0.0037534558214247227,
-      "learning_rate": 0.0001999939370456646,
-      "loss": 46.0,
-      "step": 45855
-    },
-    {
-      "epoch": 3.506011430319017,
-      "grad_norm": 0.0013452754355967045,
-      "learning_rate": 0.00019999393678116643,
-      "loss": 46.0,
-      "step": 45856
-    },
-    {
-      "epoch": 3.5060878873024066,
-      "grad_norm": 0.0012423753505572677,
-      "learning_rate": 0.00019999393651666249,
-      "loss": 46.0,
-      "step": 45857
-    },
-    {
-      "epoch": 3.5061643442857964,
-      "grad_norm": 0.0015553719131276011,
-      "learning_rate": 0.0001999939362521528,
-      "loss": 46.0,
-      "step": 45858
-    },
-    {
-      "epoch": 3.506240801269186,
-      "grad_norm": 0.005281276069581509,
-      "learning_rate": 0.0001999939359876373,
-      "loss": 46.0,
-      "step": 45859
-    },
-    {
-      "epoch": 3.506317258252576,
-      "grad_norm": 0.0019446517108008265,
-      "learning_rate": 0.00019999393572311608,
-      "loss": 46.0,
-      "step": 45860
-    },
-    {
-      "epoch": 3.506393715235965,
-      "grad_norm": 0.003982921130955219,
-      "learning_rate": 0.00019999393545858904,
-      "loss": 46.0,
-      "step": 45861
-    },
-    {
-      "epoch": 3.506470172219355,
-      "grad_norm": 0.010441708378493786,
-      "learning_rate": 0.0001999939351940563,
-      "loss": 46.0,
-      "step": 45862
-    },
-    {
-      "epoch": 3.5065466292027447,
-      "grad_norm": 0.004086884204298258,
-      "learning_rate": 0.00019999393492951774,
-      "loss": 46.0,
-      "step": 45863
-    },
-    {
-      "epoch": 3.5066230861861345,
-      "grad_norm": 0.0034190313890576363,
-      "learning_rate": 0.00019999393466497344,
-      "loss": 46.0,
-      "step": 45864
-    },
-    {
-      "epoch": 3.5066995431695243,
-      "grad_norm": 0.0013898743782192469,
-      "learning_rate": 0.0001999939344004233,
-      "loss": 46.0,
-      "step": 45865
-    },
-    {
-      "epoch": 3.506776000152914,
-      "grad_norm": 0.0033379308879375458,
-      "learning_rate": 0.00019999393413586746,
-      "loss": 46.0,
-      "step": 45866
-    },
-    {
-      "epoch": 3.506852457136304,
-      "grad_norm": 0.0015834105433896184,
-      "learning_rate": 0.00019999393387130584,
-      "loss": 46.0,
-      "step": 45867
-    },
-    {
-      "epoch": 3.5069289141196935,
-      "grad_norm": 0.001442268374375999,
-      "learning_rate": 0.00019999393360673845,
-      "loss": 46.0,
-      "step": 45868
-    },
-    {
-      "epoch": 3.507005371103083,
-      "grad_norm": 0.002715506823733449,
-      "learning_rate": 0.0001999939333421653,
-      "loss": 46.0,
-      "step": 45869
-    },
-    {
-      "epoch": 3.5070818280864726,
-      "grad_norm": 0.0026769351679831743,
-      "learning_rate": 0.00019999393307758637,
-      "loss": 46.0,
-      "step": 45870
-    },
-    {
-      "epoch": 3.5071582850698624,
-      "grad_norm": 0.0036387809086591005,
-      "learning_rate": 0.00019999393281300165,
-      "loss": 46.0,
-      "step": 45871
-    },
-    {
-      "epoch": 3.507234742053252,
-      "grad_norm": 0.003063938580453396,
-      "learning_rate": 0.00019999393254841117,
-      "loss": 46.0,
-      "step": 45872
-    },
-    {
-      "epoch": 3.507311199036642,
-      "grad_norm": 0.000976090261247009,
-      "learning_rate": 0.00019999393228381493,
-      "loss": 46.0,
-      "step": 45873
-    },
-    {
-      "epoch": 3.5073876560200317,
-      "grad_norm": 0.005339443683624268,
-      "learning_rate": 0.00019999393201921293,
-      "loss": 46.0,
-      "step": 45874
-    },
-    {
-      "epoch": 3.5074641130034214,
-      "grad_norm": 0.0008203244069591165,
-      "learning_rate": 0.00019999393175460515,
-      "loss": 46.0,
-      "step": 45875
-    },
-    {
-      "epoch": 3.507540569986811,
-      "grad_norm": 0.0024762852117419243,
-      "learning_rate": 0.0001999939314899916,
-      "loss": 46.0,
-      "step": 45876
-    },
-    {
-      "epoch": 3.507617026970201,
-      "grad_norm": 0.0018344307318329811,
-      "learning_rate": 0.0001999939312253723,
-      "loss": 46.0,
-      "step": 45877
-    },
-    {
-      "epoch": 3.5076934839535907,
-      "grad_norm": 0.0034210924059152603,
-      "learning_rate": 0.00019999393096074722,
-      "loss": 46.0,
-      "step": 45878
-    },
-    {
-      "epoch": 3.5077699409369805,
-      "grad_norm": 0.0023654100950807333,
-      "learning_rate": 0.00019999393069611637,
-      "loss": 46.0,
-      "step": 45879
-    },
-    {
-      "epoch": 3.5078463979203702,
-      "grad_norm": 0.002415325725451112,
-      "learning_rate": 0.00019999393043147973,
-      "loss": 46.0,
-      "step": 45880
-    },
-    {
-      "epoch": 3.50792285490376,
-      "grad_norm": 0.005408473312854767,
-      "learning_rate": 0.00019999393016683733,
-      "loss": 46.0,
-      "step": 45881
-    },
-    {
-      "epoch": 3.5079993118871498,
-      "grad_norm": 0.0011761934729292989,
-      "learning_rate": 0.00019999392990218917,
-      "loss": 46.0,
-      "step": 45882
-    },
-    {
-      "epoch": 3.508075768870539,
-      "grad_norm": 0.002419077791273594,
-      "learning_rate": 0.00019999392963753523,
-      "loss": 46.0,
-      "step": 45883
-    },
-    {
-      "epoch": 3.508152225853929,
-      "grad_norm": 0.002273550955578685,
-      "learning_rate": 0.00019999392937287557,
-      "loss": 46.0,
-      "step": 45884
-    },
-    {
-      "epoch": 3.5082286828373186,
-      "grad_norm": 0.0017420825315639377,
-      "learning_rate": 0.00019999392910821005,
-      "loss": 46.0,
-      "step": 45885
-    },
-    {
-      "epoch": 3.5083051398207084,
-      "grad_norm": 0.001755853183567524,
-      "learning_rate": 0.00019999392884353885,
-      "loss": 46.0,
-      "step": 45886
-    },
-    {
-      "epoch": 3.508381596804098,
-      "grad_norm": 0.003445603419095278,
-      "learning_rate": 0.00019999392857886181,
-      "loss": 46.0,
-      "step": 45887
-    },
-    {
-      "epoch": 3.508458053787488,
-      "grad_norm": 0.0022320651914924383,
-      "learning_rate": 0.00019999392831417903,
-      "loss": 46.0,
-      "step": 45888
-    },
-    {
-      "epoch": 3.5085345107708776,
-      "grad_norm": 0.004856535699218512,
-      "learning_rate": 0.0001999939280494905,
-      "loss": 46.0,
-      "step": 45889
-    },
-    {
-      "epoch": 3.5086109677542674,
-      "grad_norm": 0.004724472761154175,
-      "learning_rate": 0.0001999939277847962,
-      "loss": 46.0,
-      "step": 45890
-    },
-    {
-      "epoch": 3.5086874247376567,
-      "grad_norm": 0.002014308935031295,
-      "learning_rate": 0.0001999939275200961,
-      "loss": 46.0,
-      "step": 45891
-    },
-    {
-      "epoch": 3.5087638817210465,
-      "grad_norm": 0.0013860141625627875,
-      "learning_rate": 0.00019999392725539026,
-      "loss": 46.0,
-      "step": 45892
-    },
-    {
-      "epoch": 3.5088403387044362,
-      "grad_norm": 0.0009763978305272758,
-      "learning_rate": 0.00019999392699067862,
-      "loss": 46.0,
-      "step": 45893
-    },
-    {
-      "epoch": 3.508916795687826,
-      "grad_norm": 0.002042412292212248,
-      "learning_rate": 0.00019999392672596122,
-      "loss": 46.0,
-      "step": 45894
-    },
-    {
-      "epoch": 3.5089932526712158,
-      "grad_norm": 0.0005378625937737525,
-      "learning_rate": 0.00019999392646123808,
-      "loss": 46.0,
-      "step": 45895
-    },
-    {
-      "epoch": 3.5090697096546055,
-      "grad_norm": 0.0015830768970772624,
-      "learning_rate": 0.00019999392619650914,
-      "loss": 46.0,
-      "step": 45896
-    },
-    {
-      "epoch": 3.5091461666379953,
-      "grad_norm": 0.0017058874946087599,
-      "learning_rate": 0.00019999392593177446,
-      "loss": 46.0,
-      "step": 45897
-    },
-    {
-      "epoch": 3.509222623621385,
-      "grad_norm": 0.0009436761611141264,
-      "learning_rate": 0.000199993925667034,
-      "loss": 46.0,
-      "step": 45898
-    },
-    {
-      "epoch": 3.509299080604775,
-      "grad_norm": 0.0007821686449460685,
-      "learning_rate": 0.00019999392540228774,
-      "loss": 46.0,
-      "step": 45899
-    },
-    {
-      "epoch": 3.5093755375881646,
-      "grad_norm": 0.002428462728857994,
-      "learning_rate": 0.00019999392513753573,
-      "loss": 46.0,
-      "step": 45900
-    },
-    {
-      "epoch": 3.5094519945715543,
-      "grad_norm": 0.0031222105026245117,
-      "learning_rate": 0.00019999392487277798,
-      "loss": 46.0,
-      "step": 45901
-    },
-    {
-      "epoch": 3.509528451554944,
-      "grad_norm": 0.0013568011345341802,
-      "learning_rate": 0.00019999392460801443,
-      "loss": 46.0,
-      "step": 45902
-    },
-    {
-      "epoch": 3.509604908538334,
-      "grad_norm": 0.0010679615661501884,
-      "learning_rate": 0.0001999939243432451,
-      "loss": 46.0,
-      "step": 45903
-    },
-    {
-      "epoch": 3.5096813655217236,
-      "grad_norm": 0.008835924789309502,
-      "learning_rate": 0.00019999392407847003,
-      "loss": 46.0,
-      "step": 45904
-    },
-    {
-      "epoch": 3.509757822505113,
-      "grad_norm": 0.002657622564584017,
-      "learning_rate": 0.00019999392381368918,
-      "loss": 46.0,
-      "step": 45905
-    },
-    {
-      "epoch": 3.5098342794885027,
-      "grad_norm": 0.002233570907264948,
-      "learning_rate": 0.00019999392354890256,
-      "loss": 46.0,
-      "step": 45906
-    },
-    {
-      "epoch": 3.5099107364718924,
-      "grad_norm": 0.002726313192397356,
-      "learning_rate": 0.00019999392328411014,
-      "loss": 46.0,
-      "step": 45907
-    },
-    {
-      "epoch": 3.509987193455282,
-      "grad_norm": 0.00536592211574316,
-      "learning_rate": 0.000199993923019312,
-      "loss": 46.0,
-      "step": 45908
-    },
-    {
-      "epoch": 3.510063650438672,
-      "grad_norm": 0.0012080423766747117,
-      "learning_rate": 0.00019999392275450807,
-      "loss": 46.0,
-      "step": 45909
-    },
-    {
-      "epoch": 3.5101401074220617,
-      "grad_norm": 0.004361757542937994,
-      "learning_rate": 0.00019999392248969838,
-      "loss": 46.0,
-      "step": 45910
-    },
-    {
-      "epoch": 3.5102165644054515,
-      "grad_norm": 0.0010043777292594314,
-      "learning_rate": 0.00019999392222488292,
-      "loss": 46.0,
-      "step": 45911
-    },
-    {
-      "epoch": 3.5102930213888413,
-      "grad_norm": 0.0008478460367769003,
-      "learning_rate": 0.0001999939219600617,
-      "loss": 46.0,
-      "step": 45912
-    },
-    {
-      "epoch": 3.5103694783722306,
-      "grad_norm": 0.00814181100577116,
-      "learning_rate": 0.00019999392169523468,
-      "loss": 46.0,
-      "step": 45913
-    },
-    {
-      "epoch": 3.5104459353556203,
-      "grad_norm": 0.003671047743409872,
-      "learning_rate": 0.0001999939214304019,
-      "loss": 46.0,
-      "step": 45914
-    },
-    {
-      "epoch": 3.51052239233901,
-      "grad_norm": 0.001437688828445971,
-      "learning_rate": 0.00019999392116556335,
-      "loss": 46.0,
-      "step": 45915
-    },
-    {
-      "epoch": 3.5105988493224,
-      "grad_norm": 0.0011340953642502427,
-      "learning_rate": 0.00019999392090071908,
-      "loss": 46.0,
-      "step": 45916
-    },
-    {
-      "epoch": 3.5106753063057896,
-      "grad_norm": 0.000889941118657589,
-      "learning_rate": 0.00019999392063586898,
-      "loss": 46.0,
-      "step": 45917
-    },
-    {
-      "epoch": 3.5107517632891794,
-      "grad_norm": 0.004333685152232647,
-      "learning_rate": 0.00019999392037101313,
-      "loss": 46.0,
-      "step": 45918
-    },
-    {
-      "epoch": 3.510828220272569,
-      "grad_norm": 0.004315197467803955,
-      "learning_rate": 0.0001999939201061515,
-      "loss": 46.0,
-      "step": 45919
-    },
-    {
-      "epoch": 3.510904677255959,
-      "grad_norm": 0.0013370043598115444,
-      "learning_rate": 0.00019999391984128412,
-      "loss": 46.0,
-      "step": 45920
-    },
-    {
-      "epoch": 3.5109811342393487,
-      "grad_norm": 0.0031153459567576647,
-      "learning_rate": 0.00019999391957641096,
-      "loss": 46.0,
-      "step": 45921
-    },
-    {
-      "epoch": 3.5110575912227384,
-      "grad_norm": 0.0011196978157386184,
-      "learning_rate": 0.00019999391931153202,
-      "loss": 46.0,
-      "step": 45922
-    },
-    {
-      "epoch": 3.511134048206128,
-      "grad_norm": 0.004838210064917803,
-      "learning_rate": 0.00019999391904664733,
-      "loss": 46.0,
-      "step": 45923
-    },
-    {
-      "epoch": 3.511210505189518,
-      "grad_norm": 0.0036373466718941927,
-      "learning_rate": 0.00019999391878175687,
-      "loss": 46.0,
-      "step": 45924
-    },
-    {
-      "epoch": 3.5112869621729077,
-      "grad_norm": 0.002062140731140971,
-      "learning_rate": 0.00019999391851686061,
-      "loss": 46.0,
-      "step": 45925
-    },
-    {
-      "epoch": 3.5113634191562975,
-      "grad_norm": 0.004207535646855831,
-      "learning_rate": 0.00019999391825195864,
-      "loss": 46.0,
-      "step": 45926
-    },
-    {
-      "epoch": 3.5114398761396868,
-      "grad_norm": 0.0006255257758311927,
-      "learning_rate": 0.00019999391798705088,
-      "loss": 46.0,
-      "step": 45927
-    },
-    {
-      "epoch": 3.5115163331230765,
-      "grad_norm": 0.0015837616520002484,
-      "learning_rate": 0.0001999939177221373,
-      "loss": 46.0,
-      "step": 45928
-    },
-    {
-      "epoch": 3.5115927901064663,
-      "grad_norm": 0.001973797334358096,
-      "learning_rate": 0.00019999391745721803,
-      "loss": 46.0,
-      "step": 45929
-    },
-    {
-      "epoch": 3.511669247089856,
-      "grad_norm": 0.0008100785780698061,
-      "learning_rate": 0.00019999391719229293,
-      "loss": 46.0,
-      "step": 45930
-    },
-    {
-      "epoch": 3.511745704073246,
-      "grad_norm": 0.0021802224218845367,
-      "learning_rate": 0.0001999939169273621,
-      "loss": 46.0,
-      "step": 45931
-    },
-    {
-      "epoch": 3.5118221610566356,
-      "grad_norm": 0.0025743097066879272,
-      "learning_rate": 0.0001999939166624255,
-      "loss": 46.0,
-      "step": 45932
-    },
-    {
-      "epoch": 3.5118986180400253,
-      "grad_norm": 0.001661102520301938,
-      "learning_rate": 0.00019999391639748308,
-      "loss": 46.0,
-      "step": 45933
-    },
-    {
-      "epoch": 3.511975075023415,
-      "grad_norm": 0.00397537974640727,
-      "learning_rate": 0.00019999391613253494,
-      "loss": 46.0,
-      "step": 45934
-    },
-    {
-      "epoch": 3.5120515320068044,
-      "grad_norm": 0.01090953592211008,
-      "learning_rate": 0.000199993915867581,
-      "loss": 46.0,
-      "step": 45935
-    },
-    {
-      "epoch": 3.512127988990194,
-      "grad_norm": 0.006305725313723087,
-      "learning_rate": 0.00019999391560262132,
-      "loss": 46.0,
-      "step": 45936
-    },
-    {
-      "epoch": 3.512204445973584,
-      "grad_norm": 0.0034787540789693594,
-      "learning_rate": 0.00019999391533765586,
-      "loss": 46.0,
-      "step": 45937
-    },
-    {
-      "epoch": 3.5122809029569737,
-      "grad_norm": 0.0036479176487773657,
-      "learning_rate": 0.0001999939150726846,
-      "loss": 46.0,
-      "step": 45938
-    },
-    {
-      "epoch": 3.5123573599403635,
-      "grad_norm": 0.00420813774690032,
-      "learning_rate": 0.00019999391480770762,
-      "loss": 46.0,
-      "step": 45939
-    },
-    {
-      "epoch": 3.5124338169237532,
-      "grad_norm": 0.002550095319747925,
-      "learning_rate": 0.00019999391454272484,
-      "loss": 46.0,
-      "step": 45940
-    },
-    {
-      "epoch": 3.512510273907143,
-      "grad_norm": 0.009901037439703941,
-      "learning_rate": 0.00019999391427773632,
-      "loss": 46.0,
-      "step": 45941
-    },
-    {
-      "epoch": 3.5125867308905327,
-      "grad_norm": 0.0016465189401060343,
-      "learning_rate": 0.000199993914012742,
-      "loss": 46.0,
-      "step": 45942
-    },
-    {
-      "epoch": 3.5126631878739225,
-      "grad_norm": 0.008149377070367336,
-      "learning_rate": 0.00019999391374774192,
-      "loss": 46.0,
-      "step": 45943
-    },
-    {
-      "epoch": 3.5127396448573123,
-      "grad_norm": 0.0015656622126698494,
-      "learning_rate": 0.00019999391348273608,
-      "loss": 46.0,
-      "step": 45944
-    },
-    {
-      "epoch": 3.512816101840702,
-      "grad_norm": 0.0029504215344786644,
-      "learning_rate": 0.00019999391321772448,
-      "loss": 46.0,
-      "step": 45945
-    },
-    {
-      "epoch": 3.512892558824092,
-      "grad_norm": 0.0022443272173404694,
-      "learning_rate": 0.0001999939129527071,
-      "loss": 46.0,
-      "step": 45946
-    },
-    {
-      "epoch": 3.5129690158074816,
-      "grad_norm": 0.0048622931353747845,
-      "learning_rate": 0.00019999391268768393,
-      "loss": 46.0,
-      "step": 45947
-    },
-    {
-      "epoch": 3.513045472790871,
-      "grad_norm": 0.001592762302607298,
-      "learning_rate": 0.000199993912422655,
-      "loss": 46.0,
-      "step": 45948
-    },
-    {
-      "epoch": 3.5131219297742606,
-      "grad_norm": 0.0022699437104165554,
-      "learning_rate": 0.0001999939121576203,
-      "loss": 46.0,
-      "step": 45949
-    },
-    {
-      "epoch": 3.5131983867576504,
-      "grad_norm": 0.0007706578471697867,
-      "learning_rate": 0.00019999391189257985,
-      "loss": 46.0,
-      "step": 45950
-    },
-    {
-      "epoch": 3.51327484374104,
-      "grad_norm": 0.003420833731070161,
-      "learning_rate": 0.00019999391162753362,
-      "loss": 46.0,
-      "step": 45951
-    },
-    {
-      "epoch": 3.51335130072443,
-      "grad_norm": 0.0015259770443663,
-      "learning_rate": 0.0001999939113624816,
-      "loss": 46.0,
-      "step": 45952
-    },
-    {
-      "epoch": 3.5134277577078197,
-      "grad_norm": 0.0031806055922061205,
-      "learning_rate": 0.00019999391109742386,
-      "loss": 46.0,
-      "step": 45953
-    },
-    {
-      "epoch": 3.5135042146912094,
-      "grad_norm": 0.001705571310594678,
-      "learning_rate": 0.0001999939108323603,
-      "loss": 46.0,
-      "step": 45954
-    },
-    {
-      "epoch": 3.513580671674599,
-      "grad_norm": 0.003992537967860699,
-      "learning_rate": 0.000199993910567291,
-      "loss": 46.0,
-      "step": 45955
-    },
-    {
-      "epoch": 3.513657128657989,
-      "grad_norm": 0.001126436865888536,
-      "learning_rate": 0.00019999391030221594,
-      "loss": 46.0,
-      "step": 45956
-    },
-    {
-      "epoch": 3.5137335856413783,
-      "grad_norm": 0.0025449309032410383,
-      "learning_rate": 0.00019999391003713507,
-      "loss": 46.0,
-      "step": 45957
-    },
-    {
-      "epoch": 3.513810042624768,
-      "grad_norm": 0.0023185438476502895,
-      "learning_rate": 0.00019999390977204845,
-      "loss": 46.0,
-      "step": 45958
-    },
-    {
-      "epoch": 3.513886499608158,
-      "grad_norm": 0.0022981897927820683,
-      "learning_rate": 0.00019999390950695608,
-      "loss": 46.0,
-      "step": 45959
-    },
-    {
-      "epoch": 3.5139629565915476,
-      "grad_norm": 0.001110328477807343,
-      "learning_rate": 0.00019999390924185795,
-      "loss": 46.0,
-      "step": 45960
-    },
-    {
-      "epoch": 3.5140394135749373,
-      "grad_norm": 0.0009469927754253149,
-      "learning_rate": 0.000199993908976754,
-      "loss": 46.0,
-      "step": 45961
-    },
-    {
-      "epoch": 3.514115870558327,
-      "grad_norm": 0.0011637015268206596,
-      "learning_rate": 0.00019999390871164435,
-      "loss": 46.0,
-      "step": 45962
-    },
-    {
-      "epoch": 3.514192327541717,
-      "grad_norm": 0.0009351832559332252,
-      "learning_rate": 0.00019999390844652887,
-      "loss": 46.0,
-      "step": 45963
-    },
-    {
-      "epoch": 3.5142687845251066,
-      "grad_norm": 0.001968756318092346,
-      "learning_rate": 0.00019999390818140763,
-      "loss": 46.0,
-      "step": 45964
-    },
-    {
-      "epoch": 3.5143452415084964,
-      "grad_norm": 0.0007624453282915056,
-      "learning_rate": 0.00019999390791628066,
-      "loss": 46.0,
-      "step": 45965
-    },
-    {
-      "epoch": 3.514421698491886,
-      "grad_norm": 0.0016586828278377652,
-      "learning_rate": 0.00019999390765114788,
-      "loss": 46.0,
-      "step": 45966
-    },
-    {
-      "epoch": 3.514498155475276,
-      "grad_norm": 0.001244717976078391,
-      "learning_rate": 0.00019999390738600933,
-      "loss": 46.0,
-      "step": 45967
-    },
-    {
-      "epoch": 3.5145746124586656,
-      "grad_norm": 0.005238679703325033,
-      "learning_rate": 0.00019999390712086503,
-      "loss": 46.0,
-      "step": 45968
-    },
-    {
-      "epoch": 3.5146510694420554,
-      "grad_norm": 0.0027155191637575626,
-      "learning_rate": 0.000199993906855715,
-      "loss": 46.0,
-      "step": 45969
-    },
-    {
-      "epoch": 3.5147275264254447,
-      "grad_norm": 0.0019125982653349638,
-      "learning_rate": 0.00019999390659055912,
-      "loss": 46.0,
-      "step": 45970
-    },
-    {
-      "epoch": 3.5148039834088345,
-      "grad_norm": 0.002336718374863267,
-      "learning_rate": 0.0001999939063253975,
-      "loss": 46.0,
-      "step": 45971
-    },
-    {
-      "epoch": 3.5148804403922242,
-      "grad_norm": 0.005797531921416521,
-      "learning_rate": 0.00019999390606023013,
-      "loss": 46.0,
-      "step": 45972
-    },
-    {
-      "epoch": 3.514956897375614,
-      "grad_norm": 0.0010556880151852965,
-      "learning_rate": 0.00019999390579505697,
-      "loss": 46.0,
-      "step": 45973
-    },
-    {
-      "epoch": 3.5150333543590038,
-      "grad_norm": 0.0014832635642960668,
-      "learning_rate": 0.00019999390552987806,
-      "loss": 46.0,
-      "step": 45974
-    },
-    {
-      "epoch": 3.5151098113423935,
-      "grad_norm": 0.002287686802446842,
-      "learning_rate": 0.00019999390526469337,
-      "loss": 46.0,
-      "step": 45975
-    },
-    {
-      "epoch": 3.5151862683257833,
-      "grad_norm": 0.0008717738674022257,
-      "learning_rate": 0.0001999939049995029,
-      "loss": 46.0,
-      "step": 45976
-    },
-    {
-      "epoch": 3.515262725309173,
-      "grad_norm": 0.004684879910200834,
-      "learning_rate": 0.00019999390473430669,
-      "loss": 46.0,
-      "step": 45977
-    },
-    {
-      "epoch": 3.5153391822925624,
-      "grad_norm": 0.004966872278600931,
-      "learning_rate": 0.00019999390446910468,
-      "loss": 46.0,
-      "step": 45978
-    },
-    {
-      "epoch": 3.515415639275952,
-      "grad_norm": 0.001488385139964521,
-      "learning_rate": 0.00019999390420389693,
-      "loss": 46.0,
-      "step": 45979
-    },
-    {
-      "epoch": 3.515492096259342,
-      "grad_norm": 0.006349211558699608,
-      "learning_rate": 0.00019999390393868338,
-      "loss": 46.0,
-      "step": 45980
-    },
-    {
-      "epoch": 3.5155685532427317,
-      "grad_norm": 0.0037485247012227774,
-      "learning_rate": 0.00019999390367346408,
-      "loss": 46.0,
-      "step": 45981
-    },
-    {
-      "epoch": 3.5156450102261214,
-      "grad_norm": 0.0031722646672278643,
-      "learning_rate": 0.000199993903408239,
-      "loss": 46.0,
-      "step": 45982
-    },
-    {
-      "epoch": 3.515721467209511,
-      "grad_norm": 0.0023654988035559654,
-      "learning_rate": 0.00019999390314300817,
-      "loss": 46.0,
-      "step": 45983
-    },
-    {
-      "epoch": 3.515797924192901,
-      "grad_norm": 0.002111158799380064,
-      "learning_rate": 0.00019999390287777158,
-      "loss": 46.0,
-      "step": 45984
-    },
-    {
-      "epoch": 3.5158743811762907,
-      "grad_norm": 0.001592441345565021,
-      "learning_rate": 0.00019999390261252916,
-      "loss": 46.0,
-      "step": 45985
-    },
-    {
-      "epoch": 3.5159508381596805,
-      "grad_norm": 0.002611360512673855,
-      "learning_rate": 0.00019999390234728105,
-      "loss": 46.0,
-      "step": 45986
-    },
-    {
-      "epoch": 3.51602729514307,
-      "grad_norm": 0.004567362368106842,
-      "learning_rate": 0.00019999390208202714,
-      "loss": 46.0,
-      "step": 45987
-    },
-    {
-      "epoch": 3.51610375212646,
-      "grad_norm": 0.0021764670964330435,
-      "learning_rate": 0.00019999390181676743,
-      "loss": 46.0,
-      "step": 45988
-    },
-    {
-      "epoch": 3.5161802091098497,
-      "grad_norm": 0.003382662311196327,
-      "learning_rate": 0.00019999390155150197,
-      "loss": 46.0,
-      "step": 45989
-    },
-    {
-      "epoch": 3.5162566660932395,
-      "grad_norm": 0.005656557623296976,
-      "learning_rate": 0.00019999390128623074,
-      "loss": 46.0,
-      "step": 45990
-    },
-    {
-      "epoch": 3.5163331230766293,
-      "grad_norm": 0.0015625564847141504,
-      "learning_rate": 0.00019999390102095376,
-      "loss": 46.0,
-      "step": 45991
-    },
-    {
-      "epoch": 3.5164095800600186,
-      "grad_norm": 0.0025601221714168787,
-      "learning_rate": 0.000199993900755671,
-      "loss": 46.0,
-      "step": 45992
-    },
-    {
-      "epoch": 3.5164860370434083,
-      "grad_norm": 0.0019405633211135864,
-      "learning_rate": 0.00019999390049038244,
-      "loss": 46.0,
-      "step": 45993
-    },
-    {
-      "epoch": 3.516562494026798,
-      "grad_norm": 0.002470424398779869,
-      "learning_rate": 0.00019999390022508817,
-      "loss": 46.0,
-      "step": 45994
-    },
-    {
-      "epoch": 3.516638951010188,
-      "grad_norm": 0.0015981619944795966,
-      "learning_rate": 0.00019999389995978807,
-      "loss": 46.0,
-      "step": 45995
-    },
-    {
-      "epoch": 3.5167154079935776,
-      "grad_norm": 0.0007899231277406216,
-      "learning_rate": 0.00019999389969448225,
-      "loss": 46.0,
-      "step": 45996
-    },
-    {
-      "epoch": 3.5167918649769674,
-      "grad_norm": 0.004193992353975773,
-      "learning_rate": 0.00019999389942917064,
-      "loss": 46.0,
-      "step": 45997
-    },
-    {
-      "epoch": 3.516868321960357,
-      "grad_norm": 0.0011907995212823153,
-      "learning_rate": 0.00019999389916385325,
-      "loss": 46.0,
-      "step": 45998
-    },
-    {
-      "epoch": 3.516944778943747,
-      "grad_norm": 0.003617934649810195,
-      "learning_rate": 0.0001999938988985301,
-      "loss": 46.0,
-      "step": 45999
-    },
-    {
-      "epoch": 3.5170212359271362,
-      "grad_norm": 0.0011587579501792789,
-      "learning_rate": 0.0001999938986332012,
-      "loss": 46.0,
-      "step": 46000
-    },
-    {
-      "epoch": 3.517097692910526,
-      "grad_norm": 0.0018538255244493484,
-      "learning_rate": 0.00019999389836786651,
-      "loss": 46.0,
-      "step": 46001
-    },
-    {
-      "epoch": 3.5171741498939157,
-      "grad_norm": 0.0026684461627155542,
-      "learning_rate": 0.00019999389810252606,
-      "loss": 46.0,
-      "step": 46002
-    },
-    {
-      "epoch": 3.5172506068773055,
-      "grad_norm": 0.0028471164405345917,
-      "learning_rate": 0.00019999389783717983,
-      "loss": 46.0,
-      "step": 46003
-    },
-    {
-      "epoch": 3.5173270638606953,
-      "grad_norm": 0.0048071397468447685,
-      "learning_rate": 0.00019999389757182785,
-      "loss": 46.0,
-      "step": 46004
-    },
-    {
-      "epoch": 3.517403520844085,
-      "grad_norm": 0.0038539008237421513,
-      "learning_rate": 0.00019999389730647007,
-      "loss": 46.0,
-      "step": 46005
-    },
-    {
-      "epoch": 3.517479977827475,
-      "grad_norm": 0.001737555954605341,
-      "learning_rate": 0.00019999389704110655,
-      "loss": 46.0,
-      "step": 46006
-    },
-    {
-      "epoch": 3.5175564348108646,
-      "grad_norm": 0.0022841468453407288,
-      "learning_rate": 0.00019999389677573725,
-      "loss": 46.0,
-      "step": 46007
-    },
-    {
-      "epoch": 3.5176328917942543,
-      "grad_norm": 0.00592148257419467,
-      "learning_rate": 0.00019999389651036218,
-      "loss": 46.0,
-      "step": 46008
-    },
-    {
-      "epoch": 3.517709348777644,
-      "grad_norm": 0.008165507577359676,
-      "learning_rate": 0.00019999389624498134,
-      "loss": 46.0,
-      "step": 46009
-    },
-    {
-      "epoch": 3.517785805761034,
-      "grad_norm": 0.004823136143386364,
-      "learning_rate": 0.00019999389597959475,
-      "loss": 46.0,
-      "step": 46010
-    },
-    {
-      "epoch": 3.5178622627444236,
-      "grad_norm": 0.002332430798560381,
-      "learning_rate": 0.00019999389571420236,
-      "loss": 46.0,
-      "step": 46011
-    },
-    {
-      "epoch": 3.5179387197278134,
-      "grad_norm": 0.0027830160688608885,
-      "learning_rate": 0.00019999389544880422,
-      "loss": 46.0,
-      "step": 46012
-    },
-    {
-      "epoch": 3.518015176711203,
-      "grad_norm": 0.003499938640743494,
-      "learning_rate": 0.00019999389518340031,
-      "loss": 46.0,
-      "step": 46013
-    },
-    {
-      "epoch": 3.5180916336945924,
-      "grad_norm": 0.0026316444855183363,
-      "learning_rate": 0.0001999938949179906,
-      "loss": 46.0,
-      "step": 46014
-    },
-    {
-      "epoch": 3.518168090677982,
-      "grad_norm": 0.0015766448341310024,
-      "learning_rate": 0.00019999389465257518,
-      "loss": 46.0,
-      "step": 46015
-    },
-    {
-      "epoch": 3.518244547661372,
-      "grad_norm": 0.0015216968022286892,
-      "learning_rate": 0.00019999389438715395,
-      "loss": 46.0,
-      "step": 46016
-    },
-    {
-      "epoch": 3.5183210046447617,
-      "grad_norm": 0.004122574348002672,
-      "learning_rate": 0.00019999389412172694,
-      "loss": 46.0,
-      "step": 46017
-    },
-    {
-      "epoch": 3.5183974616281515,
-      "grad_norm": 0.0037685635033994913,
-      "learning_rate": 0.0001999938938562942,
-      "loss": 46.0,
-      "step": 46018
-    },
-    {
-      "epoch": 3.5184739186115412,
-      "grad_norm": 0.0015333796618506312,
-      "learning_rate": 0.00019999389359085567,
-      "loss": 46.0,
-      "step": 46019
-    },
-    {
-      "epoch": 3.518550375594931,
-      "grad_norm": 0.003347333986312151,
-      "learning_rate": 0.00019999389332541137,
-      "loss": 46.0,
-      "step": 46020
-    },
-    {
-      "epoch": 3.5186268325783208,
-      "grad_norm": 0.0016944825183600187,
-      "learning_rate": 0.0001999938930599613,
-      "loss": 46.0,
-      "step": 46021
-    },
-    {
-      "epoch": 3.51870328956171,
-      "grad_norm": 0.0019174270564690232,
-      "learning_rate": 0.00019999389279450546,
-      "loss": 46.0,
-      "step": 46022
-    },
-    {
-      "epoch": 3.5187797465451,
-      "grad_norm": 0.0025299994740635157,
-      "learning_rate": 0.00019999389252904387,
-      "loss": 46.0,
-      "step": 46023
-    },
-    {
-      "epoch": 3.5188562035284896,
-      "grad_norm": 0.0014852676540613174,
-      "learning_rate": 0.0001999938922635765,
-      "loss": 46.0,
-      "step": 46024
-    },
-    {
-      "epoch": 3.5189326605118794,
-      "grad_norm": 0.0025224294513463974,
-      "learning_rate": 0.00019999389199810335,
-      "loss": 46.0,
-      "step": 46025
-    },
-    {
-      "epoch": 3.519009117495269,
-      "grad_norm": 0.002003477420657873,
-      "learning_rate": 0.0001999938917326244,
-      "loss": 46.0,
-      "step": 46026
-    },
-    {
-      "epoch": 3.519085574478659,
-      "grad_norm": 0.012452625669538975,
-      "learning_rate": 0.00019999389146713975,
-      "loss": 46.0,
-      "step": 46027
-    },
-    {
-      "epoch": 3.5191620314620486,
-      "grad_norm": 0.0032979967072606087,
-      "learning_rate": 0.0001999938912016493,
-      "loss": 46.0,
-      "step": 46028
-    },
-    {
-      "epoch": 3.5192384884454384,
-      "grad_norm": 0.0018458785489201546,
-      "learning_rate": 0.0001999938909361531,
-      "loss": 46.0,
-      "step": 46029
-    },
-    {
-      "epoch": 3.519314945428828,
-      "grad_norm": 0.0020835548639297485,
-      "learning_rate": 0.0001999938906706511,
-      "loss": 46.0,
-      "step": 46030
-    },
-    {
-      "epoch": 3.519391402412218,
-      "grad_norm": 0.0014450142625719309,
-      "learning_rate": 0.00019999389040514331,
-      "loss": 46.0,
-      "step": 46031
-    },
-    {
-      "epoch": 3.5194678593956077,
-      "grad_norm": 0.002957655116915703,
-      "learning_rate": 0.0001999938901396298,
-      "loss": 46.0,
-      "step": 46032
-    },
-    {
-      "epoch": 3.5195443163789975,
-      "grad_norm": 0.002331898082047701,
-      "learning_rate": 0.0001999938898741105,
-      "loss": 46.0,
-      "step": 46033
-    },
-    {
-      "epoch": 3.519620773362387,
-      "grad_norm": 0.0017305782530456781,
-      "learning_rate": 0.00019999388960858543,
-      "loss": 46.0,
-      "step": 46034
-    },
-    {
-      "epoch": 3.519697230345777,
-      "grad_norm": 0.0030377323273569345,
-      "learning_rate": 0.0001999938893430546,
-      "loss": 46.0,
-      "step": 46035
-    },
-    {
-      "epoch": 3.5197736873291663,
-      "grad_norm": 0.0025776883121579885,
-      "learning_rate": 0.00019999388907751802,
-      "loss": 46.0,
-      "step": 46036
-    },
-    {
-      "epoch": 3.519850144312556,
-      "grad_norm": 0.0027985882479697466,
-      "learning_rate": 0.00019999388881197563,
-      "loss": 46.0,
-      "step": 46037
-    },
-    {
-      "epoch": 3.519926601295946,
-      "grad_norm": 0.002309117466211319,
-      "learning_rate": 0.0001999938885464275,
-      "loss": 46.0,
-      "step": 46038
-    },
-    {
-      "epoch": 3.5200030582793356,
-      "grad_norm": 0.001496815006248653,
-      "learning_rate": 0.0001999938882808736,
-      "loss": 46.0,
-      "step": 46039
-    },
-    {
-      "epoch": 3.5200795152627253,
-      "grad_norm": 0.0019770245999097824,
-      "learning_rate": 0.00019999388801531388,
-      "loss": 46.0,
-      "step": 46040
-    },
-    {
-      "epoch": 3.520155972246115,
-      "grad_norm": 0.003854860318824649,
-      "learning_rate": 0.00019999388774974842,
-      "loss": 46.0,
-      "step": 46041
-    },
-    {
-      "epoch": 3.520232429229505,
-      "grad_norm": 0.002077318262308836,
-      "learning_rate": 0.00019999388748417722,
-      "loss": 46.0,
-      "step": 46042
-    },
-    {
-      "epoch": 3.5203088862128946,
-      "grad_norm": 0.0014645201154053211,
-      "learning_rate": 0.00019999388721860024,
-      "loss": 46.0,
-      "step": 46043
-    },
-    {
-      "epoch": 3.520385343196284,
-      "grad_norm": 0.002360919024795294,
-      "learning_rate": 0.0001999938869530175,
-      "loss": 46.0,
-      "step": 46044
-    },
-    {
-      "epoch": 3.5204618001796737,
-      "grad_norm": 0.0009254809119738638,
-      "learning_rate": 0.00019999388668742895,
-      "loss": 46.0,
-      "step": 46045
-    },
-    {
-      "epoch": 3.5205382571630635,
-      "grad_norm": 0.001603074953891337,
-      "learning_rate": 0.00019999388642183465,
-      "loss": 46.0,
-      "step": 46046
-    },
-    {
-      "epoch": 3.520614714146453,
-      "grad_norm": 0.0011834410252049565,
-      "learning_rate": 0.0001999938861562346,
-      "loss": 46.0,
-      "step": 46047
-    },
-    {
-      "epoch": 3.520691171129843,
-      "grad_norm": 0.004092540591955185,
-      "learning_rate": 0.00019999388589062877,
-      "loss": 46.0,
-      "step": 46048
-    },
-    {
-      "epoch": 3.5207676281132327,
-      "grad_norm": 0.0015189680270850658,
-      "learning_rate": 0.00019999388562501715,
-      "loss": 46.0,
-      "step": 46049
-    },
-    {
-      "epoch": 3.5208440850966225,
-      "grad_norm": 0.00263799587264657,
-      "learning_rate": 0.00019999388535939982,
-      "loss": 46.0,
-      "step": 46050
-    },
-    {
-      "epoch": 3.5209205420800123,
-      "grad_norm": 0.0025415034033358097,
-      "learning_rate": 0.00019999388509377665,
-      "loss": 46.0,
-      "step": 46051
-    },
-    {
-      "epoch": 3.520996999063402,
-      "grad_norm": 0.0012590382248163223,
-      "learning_rate": 0.00019999388482814774,
-      "loss": 46.0,
-      "step": 46052
-    },
-    {
-      "epoch": 3.521073456046792,
-      "grad_norm": 0.0019087701803073287,
-      "learning_rate": 0.00019999388456251306,
-      "loss": 46.0,
-      "step": 46053
-    },
-    {
-      "epoch": 3.5211499130301815,
-      "grad_norm": 0.0035736518912017345,
-      "learning_rate": 0.00019999388429687263,
-      "loss": 46.0,
-      "step": 46054
-    },
-    {
-      "epoch": 3.5212263700135713,
-      "grad_norm": 0.0021984484046697617,
-      "learning_rate": 0.0001999938840312264,
-      "loss": 46.0,
-      "step": 46055
-    },
-    {
-      "epoch": 3.521302826996961,
-      "grad_norm": 0.002630413044244051,
-      "learning_rate": 0.00019999388376557443,
-      "loss": 46.0,
-      "step": 46056
-    },
-    {
-      "epoch": 3.521379283980351,
-      "grad_norm": 0.007793442811816931,
-      "learning_rate": 0.00019999388349991665,
-      "loss": 46.0,
-      "step": 46057
-    },
-    {
-      "epoch": 3.52145574096374,
-      "grad_norm": 0.0016296850517392159,
-      "learning_rate": 0.00019999388323425313,
-      "loss": 46.0,
-      "step": 46058
-    },
-    {
-      "epoch": 3.52153219794713,
-      "grad_norm": 0.003136121667921543,
-      "learning_rate": 0.00019999388296858386,
-      "loss": 46.0,
-      "step": 46059
-    },
-    {
-      "epoch": 3.5216086549305197,
-      "grad_norm": 0.0033281154464930296,
-      "learning_rate": 0.00019999388270290877,
-      "loss": 46.0,
-      "step": 46060
-    },
-    {
-      "epoch": 3.5216851119139094,
-      "grad_norm": 0.006391796283423901,
-      "learning_rate": 0.00019999388243722795,
-      "loss": 46.0,
-      "step": 46061
-    },
-    {
-      "epoch": 3.521761568897299,
-      "grad_norm": 0.0010172172915190458,
-      "learning_rate": 0.00019999388217154134,
-      "loss": 46.0,
-      "step": 46062
-    },
-    {
-      "epoch": 3.521838025880689,
-      "grad_norm": 0.0014320433838292956,
-      "learning_rate": 0.00019999388190584897,
-      "loss": 46.0,
-      "step": 46063
-    },
-    {
-      "epoch": 3.5219144828640787,
-      "grad_norm": 0.003336955327540636,
-      "learning_rate": 0.00019999388164015084,
-      "loss": 46.0,
-      "step": 46064
-    },
-    {
-      "epoch": 3.5219909398474685,
-      "grad_norm": 0.002149282954633236,
-      "learning_rate": 0.00019999388137444696,
-      "loss": 46.0,
-      "step": 46065
-    },
-    {
-      "epoch": 3.522067396830858,
-      "grad_norm": 0.0011472372571006417,
-      "learning_rate": 0.00019999388110873728,
-      "loss": 46.0,
-      "step": 46066
-    },
-    {
-      "epoch": 3.5221438538142475,
-      "grad_norm": 0.0029665378388017416,
-      "learning_rate": 0.00019999388084302182,
-      "loss": 46.0,
-      "step": 46067
-    },
-    {
-      "epoch": 3.5222203107976373,
-      "grad_norm": 0.004602435976266861,
-      "learning_rate": 0.0001999938805773006,
-      "loss": 46.0,
-      "step": 46068
-    },
-    {
-      "epoch": 3.522296767781027,
-      "grad_norm": 0.0022841079626232386,
-      "learning_rate": 0.0001999938803115736,
-      "loss": 46.0,
-      "step": 46069
-    },
-    {
-      "epoch": 3.522373224764417,
-      "grad_norm": 0.0007618876988999546,
-      "learning_rate": 0.00019999388004584087,
-      "loss": 46.0,
-      "step": 46070
-    },
-    {
-      "epoch": 3.5224496817478066,
-      "grad_norm": 0.001303955214098096,
-      "learning_rate": 0.00019999387978010232,
-      "loss": 46.0,
-      "step": 46071
-    },
-    {
-      "epoch": 3.5225261387311964,
-      "grad_norm": 0.00635666586458683,
-      "learning_rate": 0.00019999387951435805,
-      "loss": 46.0,
-      "step": 46072
-    },
-    {
-      "epoch": 3.522602595714586,
-      "grad_norm": 0.002430203603580594,
-      "learning_rate": 0.00019999387924860798,
-      "loss": 46.0,
-      "step": 46073
-    },
-    {
-      "epoch": 3.522679052697976,
-      "grad_norm": 0.0031734409276396036,
-      "learning_rate": 0.00019999387898285217,
-      "loss": 46.0,
-      "step": 46074
-    },
-    {
-      "epoch": 3.5227555096813656,
-      "grad_norm": 0.004892737604677677,
-      "learning_rate": 0.00019999387871709055,
-      "loss": 46.0,
-      "step": 46075
-    },
-    {
-      "epoch": 3.5228319666647554,
-      "grad_norm": 0.004358571022748947,
-      "learning_rate": 0.0001999938784513232,
-      "loss": 46.0,
-      "step": 46076
-    },
-    {
-      "epoch": 3.522908423648145,
-      "grad_norm": 0.0008713623974472284,
-      "learning_rate": 0.00019999387818555006,
-      "loss": 46.0,
-      "step": 46077
-    },
-    {
-      "epoch": 3.522984880631535,
-      "grad_norm": 0.0027689894195646048,
-      "learning_rate": 0.00019999387791977115,
-      "loss": 46.0,
-      "step": 46078
-    },
-    {
-      "epoch": 3.5230613376149242,
-      "grad_norm": 0.002409819047898054,
-      "learning_rate": 0.00019999387765398647,
-      "loss": 46.0,
-      "step": 46079
-    },
-    {
-      "epoch": 3.523137794598314,
-      "grad_norm": 0.012077780440449715,
-      "learning_rate": 0.000199993877388196,
-      "loss": 46.0,
-      "step": 46080
-    },
-    {
-      "epoch": 3.5232142515817038,
-      "grad_norm": 0.0026512390468269587,
-      "learning_rate": 0.00019999387712239984,
-      "loss": 46.0,
-      "step": 46081
-    },
-    {
-      "epoch": 3.5232907085650935,
-      "grad_norm": 0.0008373678429052234,
-      "learning_rate": 0.00019999387685659784,
-      "loss": 46.0,
-      "step": 46082
-    },
-    {
-      "epoch": 3.5233671655484833,
-      "grad_norm": 0.003648224752396345,
-      "learning_rate": 0.0001999938765907901,
-      "loss": 46.0,
-      "step": 46083
-    },
-    {
-      "epoch": 3.523443622531873,
-      "grad_norm": 0.003660549409687519,
-      "learning_rate": 0.00019999387632497654,
-      "loss": 46.0,
-      "step": 46084
-    },
-    {
-      "epoch": 3.523520079515263,
-      "grad_norm": 0.003959151450544596,
-      "learning_rate": 0.00019999387605915727,
-      "loss": 46.0,
-      "step": 46085
-    },
-    {
-      "epoch": 3.5235965364986526,
-      "grad_norm": 0.004377682227641344,
-      "learning_rate": 0.00019999387579333223,
-      "loss": 46.0,
-      "step": 46086
-    },
-    {
-      "epoch": 3.5236729934820423,
-      "grad_norm": 0.002529323101043701,
-      "learning_rate": 0.0001999938755275014,
-      "loss": 46.0,
-      "step": 46087
-    },
-    {
-      "epoch": 3.5237494504654316,
-      "grad_norm": 0.003742713714018464,
-      "learning_rate": 0.0001999938752616648,
-      "loss": 46.0,
-      "step": 46088
-    },
-    {
-      "epoch": 3.5238259074488214,
-      "grad_norm": 0.002671642228960991,
-      "learning_rate": 0.00019999387499582241,
-      "loss": 46.0,
-      "step": 46089
-    },
-    {
-      "epoch": 3.523902364432211,
-      "grad_norm": 0.0031757294200360775,
-      "learning_rate": 0.00019999387472997428,
-      "loss": 46.0,
-      "step": 46090
-    },
-    {
-      "epoch": 3.523978821415601,
-      "grad_norm": 0.001106171403080225,
-      "learning_rate": 0.00019999387446412037,
-      "loss": 46.0,
-      "step": 46091
-    },
-    {
-      "epoch": 3.5240552783989907,
-      "grad_norm": 0.0018763769185170531,
-      "learning_rate": 0.00019999387419826072,
-      "loss": 46.0,
-      "step": 46092
-    },
-    {
-      "epoch": 3.5241317353823804,
-      "grad_norm": 0.0017623038729652762,
-      "learning_rate": 0.00019999387393239526,
-      "loss": 46.0,
-      "step": 46093
-    },
-    {
-      "epoch": 3.52420819236577,
-      "grad_norm": 0.0032051235903054476,
-      "learning_rate": 0.00019999387366652406,
-      "loss": 46.0,
-      "step": 46094
-    },
-    {
-      "epoch": 3.52428464934916,
-      "grad_norm": 0.00218062661588192,
-      "learning_rate": 0.0001999938734006471,
-      "loss": 46.0,
-      "step": 46095
-    },
-    {
-      "epoch": 3.5243611063325497,
-      "grad_norm": 0.001641497598029673,
-      "learning_rate": 0.0001999938731347643,
-      "loss": 46.0,
-      "step": 46096
-    },
-    {
-      "epoch": 3.5244375633159395,
-      "grad_norm": 0.0018254265887662768,
-      "learning_rate": 0.00019999387286887582,
-      "loss": 46.0,
-      "step": 46097
-    },
-    {
-      "epoch": 3.5245140202993293,
-      "grad_norm": 0.0022950973361730576,
-      "learning_rate": 0.00019999387260298152,
-      "loss": 46.0,
-      "step": 46098
-    },
-    {
-      "epoch": 3.524590477282719,
-      "grad_norm": 0.003461933694779873,
-      "learning_rate": 0.00019999387233708146,
-      "loss": 46.0,
-      "step": 46099
-    },
-    {
-      "epoch": 3.5246669342661088,
-      "grad_norm": 0.0029186801984906197,
-      "learning_rate": 0.00019999387207117564,
-      "loss": 46.0,
-      "step": 46100
-    },
-    {
-      "epoch": 3.524743391249498,
-      "grad_norm": 0.0027371232863515615,
-      "learning_rate": 0.00019999387180526403,
-      "loss": 46.0,
-      "step": 46101
-    },
-    {
-      "epoch": 3.524819848232888,
-      "grad_norm": 0.004526629578322172,
-      "learning_rate": 0.00019999387153934667,
-      "loss": 46.0,
-      "step": 46102
-    },
-    {
-      "epoch": 3.5248963052162776,
-      "grad_norm": 0.0021808675955981016,
-      "learning_rate": 0.00019999387127342356,
-      "loss": 46.0,
-      "step": 46103
-    },
-    {
-      "epoch": 3.5249727621996674,
-      "grad_norm": 0.0051301997154951096,
-      "learning_rate": 0.00019999387100749463,
-      "loss": 46.0,
-      "step": 46104
-    },
-    {
-      "epoch": 3.525049219183057,
-      "grad_norm": 0.0006886975606903434,
-      "learning_rate": 0.00019999387074155997,
-      "loss": 46.0,
-      "step": 46105
-    },
-    {
-      "epoch": 3.525125676166447,
-      "grad_norm": 0.0017932055052369833,
-      "learning_rate": 0.00019999387047561952,
-      "loss": 46.0,
-      "step": 46106
-    },
-    {
-      "epoch": 3.5252021331498367,
-      "grad_norm": 0.00435675261542201,
-      "learning_rate": 0.00019999387020967332,
-      "loss": 46.0,
-      "step": 46107
-    },
-    {
-      "epoch": 3.5252785901332264,
-      "grad_norm": 0.0024882727302610874,
-      "learning_rate": 0.00019999386994372134,
-      "loss": 46.0,
-      "step": 46108
-    },
-    {
-      "epoch": 3.5253550471166157,
-      "grad_norm": 0.007611345499753952,
-      "learning_rate": 0.0001999938696777636,
-      "loss": 46.0,
-      "step": 46109
-    },
-    {
-      "epoch": 3.5254315041000055,
-      "grad_norm": 0.002032520482316613,
-      "learning_rate": 0.00019999386941180007,
-      "loss": 46.0,
-      "step": 46110
-    },
-    {
-      "epoch": 3.5255079610833953,
-      "grad_norm": 0.004340315703302622,
-      "learning_rate": 0.00019999386914583078,
-      "loss": 46.0,
-      "step": 46111
-    },
-    {
-      "epoch": 3.525584418066785,
-      "grad_norm": 0.0039915586821734905,
-      "learning_rate": 0.0001999938688798557,
-      "loss": 46.0,
-      "step": 46112
-    },
-    {
-      "epoch": 3.525660875050175,
-      "grad_norm": 0.0011454009218141437,
-      "learning_rate": 0.00019999386861387493,
-      "loss": 46.0,
-      "step": 46113
-    },
-    {
-      "epoch": 3.5257373320335645,
-      "grad_norm": 0.0010445420630276203,
-      "learning_rate": 0.0001999938683478883,
-      "loss": 46.0,
-      "step": 46114
-    },
-    {
-      "epoch": 3.5258137890169543,
-      "grad_norm": 0.0019860954489558935,
-      "learning_rate": 0.00019999386808189595,
-      "loss": 46.0,
-      "step": 46115
-    },
-    {
-      "epoch": 3.525890246000344,
-      "grad_norm": 0.0031954918522387743,
-      "learning_rate": 0.00019999386781589782,
-      "loss": 46.0,
-      "step": 46116
-    },
-    {
-      "epoch": 3.525966702983734,
-      "grad_norm": 0.001985540147870779,
-      "learning_rate": 0.00019999386754989394,
-      "loss": 46.0,
-      "step": 46117
-    },
-    {
-      "epoch": 3.5260431599671236,
-      "grad_norm": 0.002525629475712776,
-      "learning_rate": 0.00019999386728388426,
-      "loss": 46.0,
-      "step": 46118
-    },
-    {
-      "epoch": 3.5261196169505133,
-      "grad_norm": 0.004820912145078182,
-      "learning_rate": 0.0001999938670178688,
-      "loss": 46.0,
-      "step": 46119
-    },
-    {
-      "epoch": 3.526196073933903,
-      "grad_norm": 0.0014989228220656514,
-      "learning_rate": 0.0001999938667518476,
-      "loss": 46.0,
-      "step": 46120
-    },
-    {
-      "epoch": 3.526272530917293,
-      "grad_norm": 0.0015615710290148854,
-      "learning_rate": 0.0001999938664858206,
-      "loss": 46.0,
-      "step": 46121
-    },
-    {
-      "epoch": 3.5263489879006826,
-      "grad_norm": 0.0013162549585103989,
-      "learning_rate": 0.00019999386621978785,
-      "loss": 46.0,
-      "step": 46122
-    },
-    {
-      "epoch": 3.526425444884072,
-      "grad_norm": 0.0023649130016565323,
-      "learning_rate": 0.00019999386595374936,
-      "loss": 46.0,
-      "step": 46123
-    },
-    {
-      "epoch": 3.5265019018674617,
-      "grad_norm": 0.006667578127235174,
-      "learning_rate": 0.0001999938656877051,
-      "loss": 46.0,
-      "step": 46124
-    },
-    {
-      "epoch": 3.5265783588508515,
-      "grad_norm": 0.006556238513439894,
-      "learning_rate": 0.000199993865421655,
-      "loss": 46.0,
-      "step": 46125
-    },
-    {
-      "epoch": 3.5266548158342412,
-      "grad_norm": 0.0015096586430445313,
-      "learning_rate": 0.00019999386515559921,
-      "loss": 46.0,
-      "step": 46126
-    },
-    {
-      "epoch": 3.526731272817631,
-      "grad_norm": 0.0012077161809429526,
-      "learning_rate": 0.0001999938648895376,
-      "loss": 46.0,
-      "step": 46127
-    },
-    {
-      "epoch": 3.5268077298010208,
-      "grad_norm": 0.0022361089941114187,
-      "learning_rate": 0.00019999386462347024,
-      "loss": 46.0,
-      "step": 46128
-    },
-    {
-      "epoch": 3.5268841867844105,
-      "grad_norm": 0.002737488131970167,
-      "learning_rate": 0.00019999386435739713,
-      "loss": 46.0,
-      "step": 46129
-    },
-    {
-      "epoch": 3.5269606437678003,
-      "grad_norm": 0.0032334872521460056,
-      "learning_rate": 0.00019999386409131823,
-      "loss": 46.0,
-      "step": 46130
-    },
-    {
-      "epoch": 3.5270371007511896,
-      "grad_norm": 0.004387090913951397,
-      "learning_rate": 0.00019999386382523355,
-      "loss": 46.0,
-      "step": 46131
-    },
-    {
-      "epoch": 3.5271135577345794,
-      "grad_norm": 0.0015021500876173377,
-      "learning_rate": 0.00019999386355914312,
-      "loss": 46.0,
-      "step": 46132
-    },
-    {
-      "epoch": 3.527190014717969,
-      "grad_norm": 0.0020126483868807554,
-      "learning_rate": 0.00019999386329304692,
-      "loss": 46.0,
-      "step": 46133
-    },
-    {
-      "epoch": 3.527266471701359,
-      "grad_norm": 0.0019522577058523893,
-      "learning_rate": 0.00019999386302694495,
-      "loss": 46.0,
-      "step": 46134
-    },
-    {
-      "epoch": 3.5273429286847486,
-      "grad_norm": 0.005085465032607317,
-      "learning_rate": 0.00019999386276083717,
-      "loss": 46.0,
-      "step": 46135
-    },
-    {
-      "epoch": 3.5274193856681384,
-      "grad_norm": 0.0020602287258952856,
-      "learning_rate": 0.00019999386249472368,
-      "loss": 46.0,
-      "step": 46136
-    },
-    {
-      "epoch": 3.527495842651528,
-      "grad_norm": 0.0018585174111649394,
-      "learning_rate": 0.00019999386222860439,
-      "loss": 46.0,
-      "step": 46137
-    },
-    {
-      "epoch": 3.527572299634918,
-      "grad_norm": 0.000972264853771776,
-      "learning_rate": 0.00019999386196247935,
-      "loss": 46.0,
-      "step": 46138
-    },
-    {
-      "epoch": 3.5276487566183077,
-      "grad_norm": 0.004523070063441992,
-      "learning_rate": 0.0001999938616963485,
-      "loss": 46.0,
-      "step": 46139
-    },
-    {
-      "epoch": 3.5277252136016974,
-      "grad_norm": 0.006592682562768459,
-      "learning_rate": 0.00019999386143021192,
-      "loss": 46.0,
-      "step": 46140
-    },
-    {
-      "epoch": 3.527801670585087,
-      "grad_norm": 0.001471811207011342,
-      "learning_rate": 0.0001999938611640696,
-      "loss": 46.0,
-      "step": 46141
-    },
-    {
-      "epoch": 3.527878127568477,
-      "grad_norm": 0.0015464500756934285,
-      "learning_rate": 0.00019999386089792145,
-      "loss": 46.0,
-      "step": 46142
-    },
-    {
-      "epoch": 3.5279545845518667,
-      "grad_norm": 0.0017789526609703898,
-      "learning_rate": 0.00019999386063176755,
-      "loss": 46.0,
-      "step": 46143
-    },
-    {
-      "epoch": 3.5280310415352565,
-      "grad_norm": 0.004961306694895029,
-      "learning_rate": 0.00019999386036560787,
-      "loss": 46.0,
-      "step": 46144
-    },
-    {
-      "epoch": 3.528107498518646,
-      "grad_norm": 0.0016997209750115871,
-      "learning_rate": 0.00019999386009944244,
-      "loss": 46.0,
-      "step": 46145
-    },
-    {
-      "epoch": 3.5281839555020356,
-      "grad_norm": 0.0012948306975886226,
-      "learning_rate": 0.00019999385983327124,
-      "loss": 46.0,
-      "step": 46146
-    },
-    {
-      "epoch": 3.5282604124854253,
-      "grad_norm": 0.0029470238368958235,
-      "learning_rate": 0.00019999385956709427,
-      "loss": 46.0,
-      "step": 46147
-    },
-    {
-      "epoch": 3.528336869468815,
-      "grad_norm": 0.002085821470245719,
-      "learning_rate": 0.00019999385930091152,
-      "loss": 46.0,
-      "step": 46148
-    },
-    {
-      "epoch": 3.528413326452205,
-      "grad_norm": 0.0025229991879314184,
-      "learning_rate": 0.000199993859034723,
-      "loss": 46.0,
-      "step": 46149
-    },
-    {
-      "epoch": 3.5284897834355946,
-      "grad_norm": 0.0030042538419365883,
-      "learning_rate": 0.00019999385876852874,
-      "loss": 46.0,
-      "step": 46150
-    },
-    {
-      "epoch": 3.5285662404189844,
-      "grad_norm": 0.0010310504585504532,
-      "learning_rate": 0.00019999385850232867,
-      "loss": 46.0,
-      "step": 46151
-    },
-    {
-      "epoch": 3.528642697402374,
-      "grad_norm": 0.002798303961753845,
-      "learning_rate": 0.00019999385823612286,
-      "loss": 46.0,
-      "step": 46152
-    },
-    {
-      "epoch": 3.5287191543857634,
-      "grad_norm": 0.0010029373224824667,
-      "learning_rate": 0.00019999385796991127,
-      "loss": 46.0,
-      "step": 46153
-    },
-    {
-      "epoch": 3.528795611369153,
-      "grad_norm": 0.0015498147113248706,
-      "learning_rate": 0.0001999938577036939,
-      "loss": 46.0,
-      "step": 46154
-    },
-    {
-      "epoch": 3.528872068352543,
-      "grad_norm": 0.001677525695413351,
-      "learning_rate": 0.00019999385743747078,
-      "loss": 46.0,
-      "step": 46155
-    },
-    {
-      "epoch": 3.5289485253359327,
-      "grad_norm": 0.003191616153344512,
-      "learning_rate": 0.00019999385717124187,
-      "loss": 46.0,
-      "step": 46156
-    },
-    {
-      "epoch": 3.5290249823193225,
-      "grad_norm": 0.0033075122628360987,
-      "learning_rate": 0.00019999385690500722,
-      "loss": 46.0,
-      "step": 46157
-    },
-    {
-      "epoch": 3.5291014393027123,
-      "grad_norm": 0.0010205006692558527,
-      "learning_rate": 0.00019999385663876677,
-      "loss": 46.0,
-      "step": 46158
-    },
-    {
-      "epoch": 3.529177896286102,
-      "grad_norm": 0.0024993717670440674,
-      "learning_rate": 0.0001999938563725206,
-      "loss": 46.0,
-      "step": 46159
-    },
-    {
-      "epoch": 3.5292543532694918,
-      "grad_norm": 0.0024535616394132376,
-      "learning_rate": 0.00019999385610626862,
-      "loss": 46.0,
-      "step": 46160
-    },
-    {
-      "epoch": 3.5293308102528815,
-      "grad_norm": 0.006042394321411848,
-      "learning_rate": 0.00019999385584001087,
-      "loss": 46.0,
-      "step": 46161
-    },
-    {
-      "epoch": 3.5294072672362713,
-      "grad_norm": 0.0017053228802978992,
-      "learning_rate": 0.00019999385557374736,
-      "loss": 46.0,
-      "step": 46162
-    },
-    {
-      "epoch": 3.529483724219661,
-      "grad_norm": 0.004111082758754492,
-      "learning_rate": 0.0001999938553074781,
-      "loss": 46.0,
-      "step": 46163
-    },
-    {
-      "epoch": 3.529560181203051,
-      "grad_norm": 0.004131131339818239,
-      "learning_rate": 0.00019999385504120305,
-      "loss": 46.0,
-      "step": 46164
-    },
-    {
-      "epoch": 3.5296366381864406,
-      "grad_norm": 0.0007864825893193483,
-      "learning_rate": 0.00019999385477492224,
-      "loss": 46.0,
-      "step": 46165
-    },
-    {
-      "epoch": 3.5297130951698303,
-      "grad_norm": 0.004420701414346695,
-      "learning_rate": 0.00019999385450863562,
-      "loss": 46.0,
-      "step": 46166
-    },
-    {
-      "epoch": 3.5297895521532197,
-      "grad_norm": 0.001161379972472787,
-      "learning_rate": 0.00019999385424234327,
-      "loss": 46.0,
-      "step": 46167
-    },
-    {
-      "epoch": 3.5298660091366094,
-      "grad_norm": 0.0022415905259549618,
-      "learning_rate": 0.00019999385397604516,
-      "loss": 46.0,
-      "step": 46168
-    },
-    {
-      "epoch": 3.529942466119999,
-      "grad_norm": 0.0015534928534179926,
-      "learning_rate": 0.00019999385370974125,
-      "loss": 46.0,
-      "step": 46169
-    },
-    {
-      "epoch": 3.530018923103389,
-      "grad_norm": 0.0022089462727308273,
-      "learning_rate": 0.00019999385344343157,
-      "loss": 46.0,
-      "step": 46170
-    },
-    {
-      "epoch": 3.5300953800867787,
-      "grad_norm": 0.0011778940679505467,
-      "learning_rate": 0.00019999385317711618,
-      "loss": 46.0,
-      "step": 46171
-    },
-    {
-      "epoch": 3.5301718370701685,
-      "grad_norm": 0.002905918750911951,
-      "learning_rate": 0.00019999385291079495,
-      "loss": 46.0,
-      "step": 46172
-    },
-    {
-      "epoch": 3.5302482940535582,
-      "grad_norm": 0.002060182858258486,
-      "learning_rate": 0.00019999385264446798,
-      "loss": 46.0,
-      "step": 46173
-    },
-    {
-      "epoch": 3.530324751036948,
-      "grad_norm": 0.0024297856725752354,
-      "learning_rate": 0.00019999385237813523,
-      "loss": 46.0,
-      "step": 46174
-    },
-    {
-      "epoch": 3.5304012080203373,
-      "grad_norm": 0.0013838422019034624,
-      "learning_rate": 0.00019999385211179674,
-      "loss": 46.0,
-      "step": 46175
-    },
-    {
-      "epoch": 3.530477665003727,
-      "grad_norm": 0.0017051702598109841,
-      "learning_rate": 0.00019999385184545245,
-      "loss": 46.0,
-      "step": 46176
-    },
-    {
-      "epoch": 3.530554121987117,
-      "grad_norm": 0.0047286455519497395,
-      "learning_rate": 0.0001999938515791024,
-      "loss": 46.0,
-      "step": 46177
-    },
-    {
-      "epoch": 3.5306305789705066,
-      "grad_norm": 0.004432672169059515,
-      "learning_rate": 0.0001999938513127466,
-      "loss": 46.0,
-      "step": 46178
-    },
-    {
-      "epoch": 3.5307070359538963,
-      "grad_norm": 0.0015016455436125398,
-      "learning_rate": 0.000199993851046385,
-      "loss": 46.0,
-      "step": 46179
-    },
-    {
-      "epoch": 3.530783492937286,
-      "grad_norm": 0.003653473686426878,
-      "learning_rate": 0.00019999385078001765,
-      "loss": 46.0,
-      "step": 46180
-    },
-    {
-      "epoch": 3.530859949920676,
-      "grad_norm": 0.0014620987931266427,
-      "learning_rate": 0.00019999385051364455,
-      "loss": 46.0,
-      "step": 46181
-    },
-    {
-      "epoch": 3.5309364069040656,
-      "grad_norm": 0.0010354912374168634,
-      "learning_rate": 0.00019999385024726564,
-      "loss": 46.0,
-      "step": 46182
-    },
-    {
-      "epoch": 3.5310128638874554,
-      "grad_norm": 0.0009038523421622813,
-      "learning_rate": 0.000199993849980881,
-      "loss": 46.0,
-      "step": 46183
-    },
-    {
-      "epoch": 3.531089320870845,
-      "grad_norm": 0.002375928685069084,
-      "learning_rate": 0.00019999384971449056,
-      "loss": 46.0,
-      "step": 46184
-    },
-    {
-      "epoch": 3.531165777854235,
-      "grad_norm": 0.0009006283944472671,
-      "learning_rate": 0.00019999384944809436,
-      "loss": 46.0,
-      "step": 46185
-    },
-    {
-      "epoch": 3.5312422348376247,
-      "grad_norm": 0.0022582062520086765,
-      "learning_rate": 0.0001999938491816924,
-      "loss": 46.0,
-      "step": 46186
-    },
-    {
-      "epoch": 3.5313186918210144,
-      "grad_norm": 0.0005977017572149634,
-      "learning_rate": 0.00019999384891528465,
-      "loss": 46.0,
-      "step": 46187
-    },
-    {
-      "epoch": 3.531395148804404,
-      "grad_norm": 0.0015886574983596802,
-      "learning_rate": 0.00019999384864887113,
-      "loss": 46.0,
-      "step": 46188
-    },
-    {
-      "epoch": 3.5314716057877935,
-      "grad_norm": 0.0007058461778797209,
-      "learning_rate": 0.00019999384838245186,
-      "loss": 46.0,
-      "step": 46189
-    },
-    {
-      "epoch": 3.5315480627711833,
-      "grad_norm": 0.002441684016957879,
-      "learning_rate": 0.00019999384811602682,
-      "loss": 46.0,
-      "step": 46190
-    },
-    {
-      "epoch": 3.531624519754573,
-      "grad_norm": 0.004254346247762442,
-      "learning_rate": 0.000199993847849596,
-      "loss": 46.0,
-      "step": 46191
-    },
-    {
-      "epoch": 3.531700976737963,
-      "grad_norm": 0.002071988768875599,
-      "learning_rate": 0.00019999384758315943,
-      "loss": 46.0,
-      "step": 46192
-    },
-    {
-      "epoch": 3.5317774337213526,
-      "grad_norm": 0.00489824777469039,
-      "learning_rate": 0.00019999384731671707,
-      "loss": 46.0,
-      "step": 46193
-    },
-    {
-      "epoch": 3.5318538907047423,
-      "grad_norm": 0.0036819670349359512,
-      "learning_rate": 0.00019999384705026896,
-      "loss": 46.0,
-      "step": 46194
-    },
-    {
-      "epoch": 3.531930347688132,
-      "grad_norm": 0.004765456076711416,
-      "learning_rate": 0.00019999384678381506,
-      "loss": 46.0,
-      "step": 46195
-    },
-    {
-      "epoch": 3.532006804671522,
-      "grad_norm": 0.0012841366697102785,
-      "learning_rate": 0.0001999938465173554,
-      "loss": 46.0,
-      "step": 46196
-    },
-    {
-      "epoch": 3.532083261654911,
-      "grad_norm": 0.0026969967875629663,
-      "learning_rate": 0.00019999384625088998,
-      "loss": 46.0,
-      "step": 46197
-    },
-    {
-      "epoch": 3.532159718638301,
-      "grad_norm": 0.0022583266254514456,
-      "learning_rate": 0.00019999384598441878,
-      "loss": 46.0,
-      "step": 46198
-    },
-    {
-      "epoch": 3.5322361756216907,
-      "grad_norm": 0.003013828070834279,
-      "learning_rate": 0.0001999938457179418,
-      "loss": 46.0,
-      "step": 46199
-    },
-    {
-      "epoch": 3.5323126326050804,
-      "grad_norm": 0.0013328802306205034,
-      "learning_rate": 0.00019999384545145907,
-      "loss": 46.0,
-      "step": 46200
-    },
-    {
-      "epoch": 3.53238908958847,
-      "grad_norm": 0.003972713835537434,
-      "learning_rate": 0.00019999384518497055,
-      "loss": 46.0,
-      "step": 46201
-    },
-    {
-      "epoch": 3.53246554657186,
-      "grad_norm": 0.0019800085574388504,
-      "learning_rate": 0.0001999938449184763,
-      "loss": 46.0,
-      "step": 46202
-    },
-    {
-      "epoch": 3.5325420035552497,
-      "grad_norm": 0.006707543972879648,
-      "learning_rate": 0.00019999384465197624,
-      "loss": 46.0,
-      "step": 46203
-    },
-    {
-      "epoch": 3.5326184605386395,
-      "grad_norm": 0.002987027633935213,
-      "learning_rate": 0.00019999384438547043,
-      "loss": 46.0,
-      "step": 46204
-    },
-    {
-      "epoch": 3.5326949175220292,
-      "grad_norm": 0.004226853139698505,
-      "learning_rate": 0.00019999384411895885,
-      "loss": 46.0,
-      "step": 46205
-    },
-    {
-      "epoch": 3.532771374505419,
-      "grad_norm": 0.0023436364717781544,
-      "learning_rate": 0.0001999938438524415,
-      "loss": 46.0,
-      "step": 46206
-    },
-    {
-      "epoch": 3.5328478314888088,
-      "grad_norm": 0.0016728965565562248,
-      "learning_rate": 0.00019999384358591839,
-      "loss": 46.0,
-      "step": 46207
-    },
-    {
-      "epoch": 3.5329242884721985,
-      "grad_norm": 0.0017292009433731437,
-      "learning_rate": 0.00019999384331938948,
-      "loss": 46.0,
-      "step": 46208
-    },
-    {
-      "epoch": 3.5330007454555883,
-      "grad_norm": 0.001986191375181079,
-      "learning_rate": 0.00019999384305285483,
-      "loss": 46.0,
-      "step": 46209
-    },
-    {
-      "epoch": 3.5330772024389776,
-      "grad_norm": 0.004805953707545996,
-      "learning_rate": 0.00019999384278631438,
-      "loss": 46.0,
-      "step": 46210
-    },
-    {
-      "epoch": 3.5331536594223674,
-      "grad_norm": 0.0014367010444402695,
-      "learning_rate": 0.0001999938425197682,
-      "loss": 46.0,
-      "step": 46211
-    },
-    {
-      "epoch": 3.533230116405757,
-      "grad_norm": 0.0031450914684683084,
-      "learning_rate": 0.00019999384225321623,
-      "loss": 46.0,
-      "step": 46212
-    },
-    {
-      "epoch": 3.533306573389147,
-      "grad_norm": 0.002335998695343733,
-      "learning_rate": 0.00019999384198665852,
-      "loss": 46.0,
-      "step": 46213
-    },
-    {
-      "epoch": 3.5333830303725366,
-      "grad_norm": 0.0021044020541012287,
-      "learning_rate": 0.000199993841720095,
-      "loss": 46.0,
-      "step": 46214
-    },
-    {
-      "epoch": 3.5334594873559264,
-      "grad_norm": 0.003720491658896208,
-      "learning_rate": 0.00019999384145352573,
-      "loss": 46.0,
-      "step": 46215
-    },
-    {
-      "epoch": 3.533535944339316,
-      "grad_norm": 0.0017340414924547076,
-      "learning_rate": 0.00019999384118695067,
-      "loss": 46.0,
-      "step": 46216
-    },
-    {
-      "epoch": 3.533612401322706,
-      "grad_norm": 0.0043286606669425964,
-      "learning_rate": 0.00019999384092036989,
-      "loss": 46.0,
-      "step": 46217
-    },
-    {
-      "epoch": 3.5336888583060957,
-      "grad_norm": 0.0050333645194768906,
-      "learning_rate": 0.0001999938406537833,
-      "loss": 46.0,
-      "step": 46218
-    },
-    {
-      "epoch": 3.533765315289485,
-      "grad_norm": 0.002465359168127179,
-      "learning_rate": 0.00019999384038719094,
-      "loss": 46.0,
-      "step": 46219
-    },
-    {
-      "epoch": 3.5338417722728748,
-      "grad_norm": 0.00240574823692441,
-      "learning_rate": 0.00019999384012059284,
-      "loss": 46.0,
-      "step": 46220
-    },
-    {
-      "epoch": 3.5339182292562645,
-      "grad_norm": 0.002476591384038329,
-      "learning_rate": 0.00019999383985398893,
-      "loss": 46.0,
-      "step": 46221
-    },
-    {
-      "epoch": 3.5339946862396543,
-      "grad_norm": 0.0019660114776343107,
-      "learning_rate": 0.00019999383958737928,
-      "loss": 46.0,
-      "step": 46222
-    },
-    {
-      "epoch": 3.534071143223044,
-      "grad_norm": 0.0035554941277951,
-      "learning_rate": 0.00019999383932076389,
-      "loss": 46.0,
-      "step": 46223
-    },
-    {
-      "epoch": 3.534147600206434,
-      "grad_norm": 0.0008523643482476473,
-      "learning_rate": 0.00019999383905414266,
-      "loss": 46.0,
-      "step": 46224
-    },
-    {
-      "epoch": 3.5342240571898236,
-      "grad_norm": 0.003433752106502652,
-      "learning_rate": 0.00019999383878751572,
-      "loss": 46.0,
-      "step": 46225
-    },
-    {
-      "epoch": 3.5343005141732133,
-      "grad_norm": 0.0024735131300985813,
-      "learning_rate": 0.00019999383852088297,
-      "loss": 46.0,
-      "step": 46226
-    },
-    {
-      "epoch": 3.534376971156603,
-      "grad_norm": 0.0022912307176738977,
-      "learning_rate": 0.00019999383825424448,
-      "loss": 46.0,
-      "step": 46227
-    },
-    {
-      "epoch": 3.534453428139993,
-      "grad_norm": 0.0013587875291705132,
-      "learning_rate": 0.00019999383798760022,
-      "loss": 46.0,
-      "step": 46228
-    },
-    {
-      "epoch": 3.5345298851233826,
-      "grad_norm": 0.0017341406783089042,
-      "learning_rate": 0.00019999383772095018,
-      "loss": 46.0,
-      "step": 46229
-    },
-    {
-      "epoch": 3.5346063421067724,
-      "grad_norm": 0.004754199180752039,
-      "learning_rate": 0.00019999383745429437,
-      "loss": 46.0,
-      "step": 46230
-    },
-    {
-      "epoch": 3.534682799090162,
-      "grad_norm": 0.0019492115825414658,
-      "learning_rate": 0.00019999383718763278,
-      "loss": 46.0,
-      "step": 46231
-    },
-    {
-      "epoch": 3.5347592560735515,
-      "grad_norm": 0.002331928815692663,
-      "learning_rate": 0.00019999383692096543,
-      "loss": 46.0,
-      "step": 46232
-    },
-    {
-      "epoch": 3.534835713056941,
-      "grad_norm": 0.008620609529316425,
-      "learning_rate": 0.0001999938366542923,
-      "loss": 46.0,
-      "step": 46233
-    },
-    {
-      "epoch": 3.534912170040331,
-      "grad_norm": 0.0013084078673273325,
-      "learning_rate": 0.00019999383638761342,
-      "loss": 46.0,
-      "step": 46234
-    },
-    {
-      "epoch": 3.5349886270237207,
-      "grad_norm": 0.006739387754350901,
-      "learning_rate": 0.00019999383612092877,
-      "loss": 46.0,
-      "step": 46235
-    },
-    {
-      "epoch": 3.5350650840071105,
-      "grad_norm": 0.0020912489853799343,
-      "learning_rate": 0.00019999383585423834,
-      "loss": 46.0,
-      "step": 46236
-    },
-    {
-      "epoch": 3.5351415409905003,
-      "grad_norm": 0.002359029371291399,
-      "learning_rate": 0.00019999383558754215,
-      "loss": 46.0,
-      "step": 46237
-    },
-    {
-      "epoch": 3.53521799797389,
-      "grad_norm": 0.001721543725579977,
-      "learning_rate": 0.0001999938353208402,
-      "loss": 46.0,
-      "step": 46238
-    },
-    {
-      "epoch": 3.53529445495728,
-      "grad_norm": 0.0011985236778855324,
-      "learning_rate": 0.00019999383505413248,
-      "loss": 46.0,
-      "step": 46239
-    },
-    {
-      "epoch": 3.5353709119406695,
-      "grad_norm": 0.0017227210337296128,
-      "learning_rate": 0.00019999383478741897,
-      "loss": 46.0,
-      "step": 46240
-    },
-    {
-      "epoch": 3.535447368924059,
-      "grad_norm": 0.001360439695417881,
-      "learning_rate": 0.0001999938345206997,
-      "loss": 46.0,
-      "step": 46241
-    },
-    {
-      "epoch": 3.5355238259074486,
-      "grad_norm": 0.004013245925307274,
-      "learning_rate": 0.00019999383425397467,
-      "loss": 46.0,
-      "step": 46242
-    },
-    {
-      "epoch": 3.5356002828908384,
-      "grad_norm": 0.0016308400081470609,
-      "learning_rate": 0.00019999383398724386,
-      "loss": 46.0,
-      "step": 46243
-    },
-    {
-      "epoch": 3.535676739874228,
-      "grad_norm": 0.003431737655773759,
-      "learning_rate": 0.00019999383372050727,
-      "loss": 46.0,
-      "step": 46244
-    },
-    {
-      "epoch": 3.535753196857618,
-      "grad_norm": 0.006416003219783306,
-      "learning_rate": 0.00019999383345376491,
-      "loss": 46.0,
-      "step": 46245
-    },
-    {
-      "epoch": 3.5358296538410077,
-      "grad_norm": 0.003979357425123453,
-      "learning_rate": 0.0001999938331870168,
-      "loss": 46.0,
-      "step": 46246
-    },
-    {
-      "epoch": 3.5359061108243974,
-      "grad_norm": 0.006700616329908371,
-      "learning_rate": 0.00019999383292026293,
-      "loss": 46.0,
-      "step": 46247
-    },
-    {
-      "epoch": 3.535982567807787,
-      "grad_norm": 0.0017793797887861729,
-      "learning_rate": 0.00019999383265350326,
-      "loss": 46.0,
-      "step": 46248
-    },
-    {
-      "epoch": 3.536059024791177,
-      "grad_norm": 0.0019806320779025555,
-      "learning_rate": 0.00019999383238673786,
-      "loss": 46.0,
-      "step": 46249
-    },
-    {
-      "epoch": 3.5361354817745667,
-      "grad_norm": 0.0013911542482674122,
-      "learning_rate": 0.00019999383211996666,
-      "loss": 46.0,
-      "step": 46250
-    },
-    {
-      "epoch": 3.5362119387579565,
-      "grad_norm": 0.000999206560663879,
-      "learning_rate": 0.0001999938318531897,
-      "loss": 46.0,
-      "step": 46251
-    },
-    {
-      "epoch": 3.5362883957413462,
-      "grad_norm": 0.0016948198899626732,
-      "learning_rate": 0.00019999383158640698,
-      "loss": 46.0,
-      "step": 46252
-    },
-    {
-      "epoch": 3.536364852724736,
-      "grad_norm": 0.0024850894697010517,
-      "learning_rate": 0.00019999383131961848,
-      "loss": 46.0,
-      "step": 46253
-    },
-    {
-      "epoch": 3.5364413097081253,
-      "grad_norm": 0.0023038587532937527,
-      "learning_rate": 0.00019999383105282422,
-      "loss": 46.0,
-      "step": 46254
-    },
-    {
-      "epoch": 3.536517766691515,
-      "grad_norm": 0.005350546911358833,
-      "learning_rate": 0.00019999383078602416,
-      "loss": 46.0,
-      "step": 46255
-    },
-    {
-      "epoch": 3.536594223674905,
-      "grad_norm": 0.0021614579018205404,
-      "learning_rate": 0.00019999383051921837,
-      "loss": 46.0,
-      "step": 46256
-    },
-    {
-      "epoch": 3.5366706806582946,
-      "grad_norm": 0.004045534413307905,
-      "learning_rate": 0.0001999938302524068,
-      "loss": 46.0,
-      "step": 46257
-    },
-    {
-      "epoch": 3.5367471376416844,
-      "grad_norm": 0.0017706024227663875,
-      "learning_rate": 0.00019999382998558943,
-      "loss": 46.0,
-      "step": 46258
-    },
-    {
-      "epoch": 3.536823594625074,
-      "grad_norm": 0.004617693368345499,
-      "learning_rate": 0.00019999382971876633,
-      "loss": 46.0,
-      "step": 46259
-    },
-    {
-      "epoch": 3.536900051608464,
-      "grad_norm": 0.0007010715780779719,
-      "learning_rate": 0.00019999382945193745,
-      "loss": 46.0,
-      "step": 46260
-    },
-    {
-      "epoch": 3.5369765085918536,
-      "grad_norm": 0.0045739891938865185,
-      "learning_rate": 0.0001999938291851028,
-      "loss": 46.0,
-      "step": 46261
-    },
-    {
-      "epoch": 3.537052965575243,
-      "grad_norm": 0.00445365346968174,
-      "learning_rate": 0.0001999938289182624,
-      "loss": 46.0,
-      "step": 46262
-    },
-    {
-      "epoch": 3.5371294225586327,
-      "grad_norm": 0.0033870309125632048,
-      "learning_rate": 0.0001999938286514162,
-      "loss": 46.0,
-      "step": 46263
-    },
-    {
-      "epoch": 3.5372058795420225,
-      "grad_norm": 0.00199363986030221,
-      "learning_rate": 0.00019999382838456424,
-      "loss": 46.0,
-      "step": 46264
-    },
-    {
-      "epoch": 3.5372823365254122,
-      "grad_norm": 0.000944305385928601,
-      "learning_rate": 0.0001999938281177065,
-      "loss": 46.0,
-      "step": 46265
-    },
-    {
-      "epoch": 3.537358793508802,
-      "grad_norm": 0.0006186650134623051,
-      "learning_rate": 0.000199993827850843,
-      "loss": 46.0,
-      "step": 46266
-    },
-    {
-      "epoch": 3.5374352504921918,
-      "grad_norm": 0.0011179079301655293,
-      "learning_rate": 0.00019999382758397374,
-      "loss": 46.0,
-      "step": 46267
-    },
-    {
-      "epoch": 3.5375117074755815,
-      "grad_norm": 0.0026072447653859854,
-      "learning_rate": 0.0001999938273170987,
-      "loss": 46.0,
-      "step": 46268
-    },
-    {
-      "epoch": 3.5375881644589713,
-      "grad_norm": 0.001495429198257625,
-      "learning_rate": 0.0001999938270502179,
-      "loss": 46.0,
-      "step": 46269
-    },
-    {
-      "epoch": 3.537664621442361,
-      "grad_norm": 0.001951538142748177,
-      "learning_rate": 0.0001999938267833313,
-      "loss": 46.0,
-      "step": 46270
-    },
-    {
-      "epoch": 3.537741078425751,
-      "grad_norm": 0.004342267289757729,
-      "learning_rate": 0.00019999382651643898,
-      "loss": 46.0,
-      "step": 46271
-    },
-    {
-      "epoch": 3.5378175354091406,
-      "grad_norm": 0.0011411033337935805,
-      "learning_rate": 0.00019999382624954088,
-      "loss": 46.0,
-      "step": 46272
-    },
-    {
-      "epoch": 3.5378939923925303,
-      "grad_norm": 0.0011285870568826795,
-      "learning_rate": 0.000199993825982637,
-      "loss": 46.0,
-      "step": 46273
-    },
-    {
-      "epoch": 3.53797044937592,
-      "grad_norm": 0.0015957941068336368,
-      "learning_rate": 0.00019999382571572735,
-      "loss": 46.0,
-      "step": 46274
-    },
-    {
-      "epoch": 3.53804690635931,
-      "grad_norm": 0.005447502247989178,
-      "learning_rate": 0.00019999382544881193,
-      "loss": 46.0,
-      "step": 46275
-    },
-    {
-      "epoch": 3.538123363342699,
-      "grad_norm": 0.0006389226182363927,
-      "learning_rate": 0.00019999382518189073,
-      "loss": 46.0,
-      "step": 46276
-    },
-    {
-      "epoch": 3.538199820326089,
-      "grad_norm": 0.00413490878418088,
-      "learning_rate": 0.00019999382491496376,
-      "loss": 46.0,
-      "step": 46277
-    },
-    {
-      "epoch": 3.5382762773094787,
-      "grad_norm": 0.003709120210260153,
-      "learning_rate": 0.00019999382464803108,
-      "loss": 46.0,
-      "step": 46278
-    },
-    {
-      "epoch": 3.5383527342928685,
-      "grad_norm": 0.00188072572927922,
-      "learning_rate": 0.00019999382438109256,
-      "loss": 46.0,
-      "step": 46279
-    },
-    {
-      "epoch": 3.538429191276258,
-      "grad_norm": 0.0010070427088066936,
-      "learning_rate": 0.0001999938241141483,
-      "loss": 46.0,
-      "step": 46280
-    },
-    {
-      "epoch": 3.538505648259648,
-      "grad_norm": 0.0017372238216921687,
-      "learning_rate": 0.00019999382384719826,
-      "loss": 46.0,
-      "step": 46281
-    },
-    {
-      "epoch": 3.5385821052430377,
-      "grad_norm": 0.002527270233258605,
-      "learning_rate": 0.00019999382358024245,
-      "loss": 46.0,
-      "step": 46282
-    },
-    {
-      "epoch": 3.5386585622264275,
-      "grad_norm": 0.003549344837665558,
-      "learning_rate": 0.0001999938233132809,
-      "loss": 46.0,
-      "step": 46283
-    },
-    {
-      "epoch": 3.538735019209817,
-      "grad_norm": 0.0042600310407578945,
-      "learning_rate": 0.00019999382304631357,
-      "loss": 46.0,
-      "step": 46284
-    },
-    {
-      "epoch": 3.5388114761932066,
-      "grad_norm": 0.001320547889918089,
-      "learning_rate": 0.00019999382277934044,
-      "loss": 46.0,
-      "step": 46285
-    },
-    {
-      "epoch": 3.5388879331765963,
-      "grad_norm": 0.001385242328979075,
-      "learning_rate": 0.00019999382251236156,
-      "loss": 46.0,
-      "step": 46286
-    },
-    {
-      "epoch": 3.538964390159986,
-      "grad_norm": 0.003106738207861781,
-      "learning_rate": 0.0001999938222453769,
-      "loss": 46.0,
-      "step": 46287
-    },
-    {
-      "epoch": 3.539040847143376,
-      "grad_norm": 0.0009308186126872897,
-      "learning_rate": 0.0001999938219783865,
-      "loss": 46.0,
-      "step": 46288
-    },
-    {
-      "epoch": 3.5391173041267656,
-      "grad_norm": 0.005181416403502226,
-      "learning_rate": 0.00019999382171139032,
-      "loss": 46.0,
-      "step": 46289
-    },
-    {
-      "epoch": 3.5391937611101554,
-      "grad_norm": 0.0012717050267383456,
-      "learning_rate": 0.00019999382144438835,
-      "loss": 46.0,
-      "step": 46290
-    },
-    {
-      "epoch": 3.539270218093545,
-      "grad_norm": 0.006178069394081831,
-      "learning_rate": 0.00019999382117738063,
-      "loss": 46.0,
-      "step": 46291
-    },
-    {
-      "epoch": 3.539346675076935,
-      "grad_norm": 0.0018972111865878105,
-      "learning_rate": 0.00019999382091036714,
-      "loss": 46.0,
-      "step": 46292
-    },
-    {
-      "epoch": 3.5394231320603247,
-      "grad_norm": 0.002399832010269165,
-      "learning_rate": 0.00019999382064334785,
-      "loss": 46.0,
-      "step": 46293
-    },
-    {
-      "epoch": 3.5394995890437144,
-      "grad_norm": 0.002942180959507823,
-      "learning_rate": 0.00019999382037632284,
-      "loss": 46.0,
-      "step": 46294
-    },
-    {
-      "epoch": 3.539576046027104,
-      "grad_norm": 0.003683202899992466,
-      "learning_rate": 0.00019999382010929204,
-      "loss": 46.0,
-      "step": 46295
-    },
-    {
-      "epoch": 3.539652503010494,
-      "grad_norm": 0.002486092271283269,
-      "learning_rate": 0.00019999381984225548,
-      "loss": 46.0,
-      "step": 46296
-    },
-    {
-      "epoch": 3.5397289599938837,
-      "grad_norm": 0.004011643119156361,
-      "learning_rate": 0.00019999381957521312,
-      "loss": 46.0,
-      "step": 46297
-    },
-    {
-      "epoch": 3.539805416977273,
-      "grad_norm": 0.0014857224887236953,
-      "learning_rate": 0.00019999381930816502,
-      "loss": 46.0,
-      "step": 46298
-    },
-    {
-      "epoch": 3.539881873960663,
-      "grad_norm": 0.0021912052761763334,
-      "learning_rate": 0.00019999381904111117,
-      "loss": 46.0,
-      "step": 46299
-    },
-    {
-      "epoch": 3.5399583309440525,
-      "grad_norm": 0.001422680332325399,
-      "learning_rate": 0.0001999938187740515,
-      "loss": 46.0,
-      "step": 46300
-    },
-    {
-      "epoch": 3.5400347879274423,
-      "grad_norm": 0.0029559291433542967,
-      "learning_rate": 0.0001999938185069861,
-      "loss": 46.0,
-      "step": 46301
-    },
-    {
-      "epoch": 3.540111244910832,
-      "grad_norm": 0.0008832392049953341,
-      "learning_rate": 0.0001999938182399149,
-      "loss": 46.0,
-      "step": 46302
-    },
-    {
-      "epoch": 3.540187701894222,
-      "grad_norm": 0.002093641087412834,
-      "learning_rate": 0.00019999381797283796,
-      "loss": 46.0,
-      "step": 46303
-    },
-    {
-      "epoch": 3.5402641588776116,
-      "grad_norm": 0.009877314791083336,
-      "learning_rate": 0.00019999381770575522,
-      "loss": 46.0,
-      "step": 46304
-    },
-    {
-      "epoch": 3.5403406158610013,
-      "grad_norm": 0.0014203209429979324,
-      "learning_rate": 0.00019999381743866673,
-      "loss": 46.0,
-      "step": 46305
-    },
-    {
-      "epoch": 3.5404170728443907,
-      "grad_norm": 0.004516810178756714,
-      "learning_rate": 0.0001999938171715725,
-      "loss": 46.0,
-      "step": 46306
-    },
-    {
-      "epoch": 3.5404935298277804,
-      "grad_norm": 0.0025179197546094656,
-      "learning_rate": 0.00019999381690447246,
-      "loss": 46.0,
-      "step": 46307
-    },
-    {
-      "epoch": 3.54056998681117,
-      "grad_norm": 0.0025336826220154762,
-      "learning_rate": 0.00019999381663736665,
-      "loss": 46.0,
-      "step": 46308
-    },
-    {
-      "epoch": 3.54064644379456,
-      "grad_norm": 0.0015534859849140048,
-      "learning_rate": 0.0001999938163702551,
-      "loss": 46.0,
-      "step": 46309
-    },
-    {
-      "epoch": 3.5407229007779497,
-      "grad_norm": 0.007230198476463556,
-      "learning_rate": 0.00019999381610313774,
-      "loss": 46.0,
-      "step": 46310
-    },
-    {
-      "epoch": 3.5407993577613395,
-      "grad_norm": 0.0021804249845445156,
-      "learning_rate": 0.00019999381583601464,
-      "loss": 46.0,
-      "step": 46311
-    },
-    {
-      "epoch": 3.5408758147447292,
-      "grad_norm": 0.0016025755321606994,
-      "learning_rate": 0.00019999381556888576,
-      "loss": 46.0,
-      "step": 46312
-    },
-    {
-      "epoch": 3.540952271728119,
-      "grad_norm": 0.002811283804476261,
-      "learning_rate": 0.00019999381530175111,
-      "loss": 46.0,
-      "step": 46313
-    },
-    {
-      "epoch": 3.5410287287115088,
-      "grad_norm": 0.004209770355373621,
-      "learning_rate": 0.0001999938150346107,
-      "loss": 46.0,
-      "step": 46314
-    },
-    {
-      "epoch": 3.5411051856948985,
-      "grad_norm": 0.0007851845002733171,
-      "learning_rate": 0.00019999381476746452,
-      "loss": 46.0,
-      "step": 46315
-    },
-    {
-      "epoch": 3.5411816426782883,
-      "grad_norm": 0.001168499351479113,
-      "learning_rate": 0.00019999381450031258,
-      "loss": 46.0,
-      "step": 46316
-    },
-    {
-      "epoch": 3.541258099661678,
-      "grad_norm": 0.002150749322026968,
-      "learning_rate": 0.00019999381423315487,
-      "loss": 46.0,
-      "step": 46317
-    },
-    {
-      "epoch": 3.541334556645068,
-      "grad_norm": 0.0014501098776236176,
-      "learning_rate": 0.00019999381396599138,
-      "loss": 46.0,
-      "step": 46318
-    },
-    {
-      "epoch": 3.5414110136284576,
-      "grad_norm": 0.0032885984983295202,
-      "learning_rate": 0.00019999381369882212,
-      "loss": 46.0,
-      "step": 46319
-    },
-    {
-      "epoch": 3.541487470611847,
-      "grad_norm": 0.0011689552338793874,
-      "learning_rate": 0.00019999381343164708,
-      "loss": 46.0,
-      "step": 46320
-    },
-    {
-      "epoch": 3.5415639275952366,
-      "grad_norm": 0.0020286189392209053,
-      "learning_rate": 0.00019999381316446627,
-      "loss": 46.0,
-      "step": 46321
-    },
-    {
-      "epoch": 3.5416403845786264,
-      "grad_norm": 0.002792837331071496,
-      "learning_rate": 0.00019999381289727972,
-      "loss": 46.0,
-      "step": 46322
-    },
-    {
-      "epoch": 3.541716841562016,
-      "grad_norm": 0.0013269322225823998,
-      "learning_rate": 0.00019999381263008736,
-      "loss": 46.0,
-      "step": 46323
-    },
-    {
-      "epoch": 3.541793298545406,
-      "grad_norm": 0.002234497107565403,
-      "learning_rate": 0.00019999381236288929,
-      "loss": 46.0,
-      "step": 46324
-    },
-    {
-      "epoch": 3.5418697555287957,
-      "grad_norm": 0.002600625157356262,
-      "learning_rate": 0.0001999938120956854,
-      "loss": 46.0,
-      "step": 46325
-    },
-    {
-      "epoch": 3.5419462125121854,
-      "grad_norm": 0.0022841726895421743,
-      "learning_rate": 0.00019999381182847576,
-      "loss": 46.0,
-      "step": 46326
-    },
-    {
-      "epoch": 3.542022669495575,
-      "grad_norm": 0.002187923062592745,
-      "learning_rate": 0.00019999381156126034,
-      "loss": 46.0,
-      "step": 46327
-    },
-    {
-      "epoch": 3.5420991264789645,
-      "grad_norm": 0.003986688796430826,
-      "learning_rate": 0.00019999381129403915,
-      "loss": 46.0,
-      "step": 46328
-    },
-    {
-      "epoch": 3.5421755834623543,
-      "grad_norm": 0.003036154666915536,
-      "learning_rate": 0.0001999938110268122,
-      "loss": 46.0,
-      "step": 46329
-    },
-    {
-      "epoch": 3.542252040445744,
-      "grad_norm": 0.0014259350718930364,
-      "learning_rate": 0.00019999381075957952,
-      "loss": 46.0,
-      "step": 46330
-    },
-    {
-      "epoch": 3.542328497429134,
-      "grad_norm": 0.0014589549973607063,
-      "learning_rate": 0.000199993810492341,
-      "loss": 46.0,
-      "step": 46331
-    },
-    {
-      "epoch": 3.5424049544125236,
-      "grad_norm": 0.0038053635507822037,
-      "learning_rate": 0.00019999381022509677,
-      "loss": 46.0,
-      "step": 46332
-    },
-    {
-      "epoch": 3.5424814113959133,
-      "grad_norm": 0.0016306041507050395,
-      "learning_rate": 0.0001999938099578467,
-      "loss": 46.0,
-      "step": 46333
-    },
-    {
-      "epoch": 3.542557868379303,
-      "grad_norm": 0.003216137643903494,
-      "learning_rate": 0.00019999380969059095,
-      "loss": 46.0,
-      "step": 46334
-    },
-    {
-      "epoch": 3.542634325362693,
-      "grad_norm": 0.0031793902162462473,
-      "learning_rate": 0.00019999380942332937,
-      "loss": 46.0,
-      "step": 46335
-    },
-    {
-      "epoch": 3.5427107823460826,
-      "grad_norm": 0.0018321546958759427,
-      "learning_rate": 0.00019999380915606204,
-      "loss": 46.0,
-      "step": 46336
-    },
-    {
-      "epoch": 3.5427872393294724,
-      "grad_norm": 0.0014027178985998034,
-      "learning_rate": 0.00019999380888878894,
-      "loss": 46.0,
-      "step": 46337
-    },
-    {
-      "epoch": 3.542863696312862,
-      "grad_norm": 0.0014313403517007828,
-      "learning_rate": 0.00019999380862151006,
-      "loss": 46.0,
-      "step": 46338
-    },
-    {
-      "epoch": 3.542940153296252,
-      "grad_norm": 0.0010580315720289946,
-      "learning_rate": 0.00019999380835422544,
-      "loss": 46.0,
-      "step": 46339
-    },
-    {
-      "epoch": 3.5430166102796417,
-      "grad_norm": 0.004340584389865398,
-      "learning_rate": 0.00019999380808693502,
-      "loss": 46.0,
-      "step": 46340
-    },
-    {
-      "epoch": 3.543093067263031,
-      "grad_norm": 0.002075295429676771,
-      "learning_rate": 0.00019999380781963883,
-      "loss": 46.0,
-      "step": 46341
-    },
-    {
-      "epoch": 3.5431695242464207,
-      "grad_norm": 0.003148572286590934,
-      "learning_rate": 0.00019999380755233688,
-      "loss": 46.0,
-      "step": 46342
-    },
-    {
-      "epoch": 3.5432459812298105,
-      "grad_norm": 0.0010792420944198966,
-      "learning_rate": 0.00019999380728502917,
-      "loss": 46.0,
-      "step": 46343
-    },
-    {
-      "epoch": 3.5433224382132003,
-      "grad_norm": 0.0020565842278301716,
-      "learning_rate": 0.00019999380701771568,
-      "loss": 46.0,
-      "step": 46344
-    },
-    {
-      "epoch": 3.54339889519659,
-      "grad_norm": 0.0014495979994535446,
-      "learning_rate": 0.00019999380675039642,
-      "loss": 46.0,
-      "step": 46345
-    },
-    {
-      "epoch": 3.5434753521799798,
-      "grad_norm": 0.003187112044543028,
-      "learning_rate": 0.00019999380648307139,
-      "loss": 46.0,
-      "step": 46346
-    },
-    {
-      "epoch": 3.5435518091633695,
-      "grad_norm": 0.0022386698983609676,
-      "learning_rate": 0.00019999380621574063,
-      "loss": 46.0,
-      "step": 46347
-    },
-    {
-      "epoch": 3.5436282661467593,
-      "grad_norm": 0.0033604076597839594,
-      "learning_rate": 0.00019999380594840405,
-      "loss": 46.0,
-      "step": 46348
-    },
-    {
-      "epoch": 3.543704723130149,
-      "grad_norm": 0.0009237302583642304,
-      "learning_rate": 0.0001999938056810617,
-      "loss": 46.0,
-      "step": 46349
-    },
-    {
-      "epoch": 3.5437811801135384,
-      "grad_norm": 0.005306551698595285,
-      "learning_rate": 0.0001999938054137136,
-      "loss": 46.0,
-      "step": 46350
-    },
-    {
-      "epoch": 3.543857637096928,
-      "grad_norm": 0.0009216921171173453,
-      "learning_rate": 0.00019999380514635972,
-      "loss": 46.0,
-      "step": 46351
-    },
-    {
-      "epoch": 3.543934094080318,
-      "grad_norm": 0.001888197148218751,
-      "learning_rate": 0.0001999938048790001,
-      "loss": 46.0,
-      "step": 46352
-    },
-    {
-      "epoch": 3.5440105510637077,
-      "grad_norm": 0.0014829114079475403,
-      "learning_rate": 0.0001999938046116347,
-      "loss": 46.0,
-      "step": 46353
-    },
-    {
-      "epoch": 3.5440870080470974,
-      "grad_norm": 0.0017742238705977798,
-      "learning_rate": 0.0001999938043442635,
-      "loss": 46.0,
-      "step": 46354
-    },
-    {
-      "epoch": 3.544163465030487,
-      "grad_norm": 0.002818820532411337,
-      "learning_rate": 0.00019999380407688657,
-      "loss": 46.0,
-      "step": 46355
-    },
-    {
-      "epoch": 3.544239922013877,
-      "grad_norm": 0.001856859540566802,
-      "learning_rate": 0.00019999380380950386,
-      "loss": 46.0,
-      "step": 46356
-    },
-    {
-      "epoch": 3.5443163789972667,
-      "grad_norm": 0.0019075623713433743,
-      "learning_rate": 0.00019999380354211537,
-      "loss": 46.0,
-      "step": 46357
-    },
-    {
-      "epoch": 3.5443928359806565,
-      "grad_norm": 0.0010480245109647512,
-      "learning_rate": 0.0001999938032747211,
-      "loss": 46.0,
-      "step": 46358
-    },
-    {
-      "epoch": 3.5444692929640462,
-      "grad_norm": 0.0008112953510135412,
-      "learning_rate": 0.0001999938030073211,
-      "loss": 46.0,
-      "step": 46359
-    },
-    {
-      "epoch": 3.544545749947436,
-      "grad_norm": 0.004541801288723946,
-      "learning_rate": 0.0001999938027399153,
-      "loss": 46.0,
-      "step": 46360
-    },
-    {
-      "epoch": 3.5446222069308257,
-      "grad_norm": 0.007674762979149818,
-      "learning_rate": 0.00019999380247250374,
-      "loss": 46.0,
-      "step": 46361
-    },
-    {
-      "epoch": 3.5446986639142155,
-      "grad_norm": 0.0021105597261339426,
-      "learning_rate": 0.0001999938022050864,
-      "loss": 46.0,
-      "step": 46362
-    },
-    {
-      "epoch": 3.544775120897605,
-      "grad_norm": 0.001374351093545556,
-      "learning_rate": 0.0001999938019376633,
-      "loss": 46.0,
-      "step": 46363
-    },
-    {
-      "epoch": 3.5448515778809946,
-      "grad_norm": 0.007965194061398506,
-      "learning_rate": 0.00019999380167023444,
-      "loss": 46.0,
-      "step": 46364
-    },
-    {
-      "epoch": 3.5449280348643843,
-      "grad_norm": 0.005757701583206654,
-      "learning_rate": 0.00019999380140279982,
-      "loss": 46.0,
-      "step": 46365
-    },
-    {
-      "epoch": 3.545004491847774,
-      "grad_norm": 0.0011785657843574882,
-      "learning_rate": 0.0001999938011353594,
-      "loss": 46.0,
-      "step": 46366
-    },
-    {
-      "epoch": 3.545080948831164,
-      "grad_norm": 0.003242665668949485,
-      "learning_rate": 0.00019999380086791323,
-      "loss": 46.0,
-      "step": 46367
-    },
-    {
-      "epoch": 3.5451574058145536,
-      "grad_norm": 0.0021293293684720993,
-      "learning_rate": 0.00019999380060046126,
-      "loss": 46.0,
-      "step": 46368
-    },
-    {
-      "epoch": 3.5452338627979434,
-      "grad_norm": 0.00043841879232786596,
-      "learning_rate": 0.00019999380033300357,
-      "loss": 46.0,
-      "step": 46369
-    },
-    {
-      "epoch": 3.545310319781333,
-      "grad_norm": 0.001587691600434482,
-      "learning_rate": 0.00019999380006554009,
-      "loss": 46.0,
-      "step": 46370
-    },
-    {
-      "epoch": 3.545386776764723,
-      "grad_norm": 0.007348526269197464,
-      "learning_rate": 0.00019999379979807085,
-      "loss": 46.0,
-      "step": 46371
-    },
-    {
-      "epoch": 3.5454632337481122,
-      "grad_norm": 0.0038146430160850286,
-      "learning_rate": 0.00019999379953059582,
-      "loss": 46.0,
-      "step": 46372
-    },
-    {
-      "epoch": 3.545539690731502,
-      "grad_norm": 0.0018491229275241494,
-      "learning_rate": 0.00019999379926311504,
-      "loss": 46.0,
-      "step": 46373
-    },
-    {
-      "epoch": 3.5456161477148918,
-      "grad_norm": 0.006273452192544937,
-      "learning_rate": 0.00019999379899562846,
-      "loss": 46.0,
-      "step": 46374
-    },
-    {
-      "epoch": 3.5456926046982815,
-      "grad_norm": 0.000767296354752034,
-      "learning_rate": 0.00019999379872813613,
-      "loss": 46.0,
-      "step": 46375
-    },
-    {
-      "epoch": 3.5457690616816713,
-      "grad_norm": 0.0035523734986782074,
-      "learning_rate": 0.00019999379846063803,
-      "loss": 46.0,
-      "step": 46376
-    },
-    {
-      "epoch": 3.545845518665061,
-      "grad_norm": 0.007249224465340376,
-      "learning_rate": 0.00019999379819313418,
-      "loss": 46.0,
-      "step": 46377
-    },
-    {
-      "epoch": 3.545921975648451,
-      "grad_norm": 0.002585752634331584,
-      "learning_rate": 0.00019999379792562453,
-      "loss": 46.0,
-      "step": 46378
-    },
-    {
-      "epoch": 3.5459984326318406,
-      "grad_norm": 0.0026606840547174215,
-      "learning_rate": 0.00019999379765810914,
-      "loss": 46.0,
-      "step": 46379
-    },
-    {
-      "epoch": 3.5460748896152303,
-      "grad_norm": 0.007803153246641159,
-      "learning_rate": 0.00019999379739058797,
-      "loss": 46.0,
-      "step": 46380
-    },
-    {
-      "epoch": 3.54615134659862,
-      "grad_norm": 0.0020089249592274427,
-      "learning_rate": 0.000199993797123061,
-      "loss": 46.0,
-      "step": 46381
-    },
-    {
-      "epoch": 3.54622780358201,
-      "grad_norm": 0.005319799762219191,
-      "learning_rate": 0.0001999937968555283,
-      "loss": 46.0,
-      "step": 46382
-    },
-    {
-      "epoch": 3.5463042605653996,
-      "grad_norm": 0.0013772428501397371,
-      "learning_rate": 0.00019999379658798983,
-      "loss": 46.0,
-      "step": 46383
-    },
-    {
-      "epoch": 3.5463807175487894,
-      "grad_norm": 0.004116037394851446,
-      "learning_rate": 0.00019999379632044557,
-      "loss": 46.0,
-      "step": 46384
-    },
-    {
-      "epoch": 3.5464571745321787,
-      "grad_norm": 0.0018922293093055487,
-      "learning_rate": 0.00019999379605289557,
-      "loss": 46.0,
-      "step": 46385
-    },
-    {
-      "epoch": 3.5465336315155684,
-      "grad_norm": 0.0019334412645548582,
-      "learning_rate": 0.00019999379578533976,
-      "loss": 46.0,
-      "step": 46386
-    },
-    {
-      "epoch": 3.546610088498958,
-      "grad_norm": 0.0027468183543533087,
-      "learning_rate": 0.0001999937955177782,
-      "loss": 46.0,
-      "step": 46387
-    },
-    {
-      "epoch": 3.546686545482348,
-      "grad_norm": 0.00670226477086544,
-      "learning_rate": 0.00019999379525021088,
-      "loss": 46.0,
-      "step": 46388
-    },
-    {
-      "epoch": 3.5467630024657377,
-      "grad_norm": 0.0012792288325726986,
-      "learning_rate": 0.00019999379498263778,
-      "loss": 46.0,
-      "step": 46389
-    },
-    {
-      "epoch": 3.5468394594491275,
-      "grad_norm": 0.0031636471394449472,
-      "learning_rate": 0.00019999379471505893,
-      "loss": 46.0,
-      "step": 46390
-    },
-    {
-      "epoch": 3.5469159164325172,
-      "grad_norm": 0.011945711448788643,
-      "learning_rate": 0.00019999379444747429,
-      "loss": 46.0,
-      "step": 46391
-    },
-    {
-      "epoch": 3.546992373415907,
-      "grad_norm": 0.003536128904670477,
-      "learning_rate": 0.0001999937941798839,
-      "loss": 46.0,
-      "step": 46392
-    },
-    {
-      "epoch": 3.5470688303992963,
-      "grad_norm": 0.002750889863818884,
-      "learning_rate": 0.0001999937939122877,
-      "loss": 46.0,
-      "step": 46393
-    },
-    {
-      "epoch": 3.547145287382686,
-      "grad_norm": 0.0028589051216840744,
-      "learning_rate": 0.00019999379364468576,
-      "loss": 46.0,
-      "step": 46394
-    },
-    {
-      "epoch": 3.547221744366076,
-      "grad_norm": 0.001291858498007059,
-      "learning_rate": 0.00019999379337707805,
-      "loss": 46.0,
-      "step": 46395
-    },
-    {
-      "epoch": 3.5472982013494656,
-      "grad_norm": 0.0016246800078079104,
-      "learning_rate": 0.00019999379310946459,
-      "loss": 46.0,
-      "step": 46396
-    },
-    {
-      "epoch": 3.5473746583328554,
-      "grad_norm": 0.0025478247553110123,
-      "learning_rate": 0.00019999379284184535,
-      "loss": 46.0,
-      "step": 46397
-    },
-    {
-      "epoch": 3.547451115316245,
-      "grad_norm": 0.001480619772337377,
-      "learning_rate": 0.00019999379257422032,
-      "loss": 46.0,
-      "step": 46398
-    },
-    {
-      "epoch": 3.547527572299635,
-      "grad_norm": 0.002828828524798155,
-      "learning_rate": 0.0001999937923065895,
-      "loss": 46.0,
-      "step": 46399
-    },
-    {
-      "epoch": 3.5476040292830247,
-      "grad_norm": 0.002600104780867696,
-      "learning_rate": 0.00019999379203895296,
-      "loss": 46.0,
-      "step": 46400
-    },
-    {
-      "epoch": 3.5476804862664144,
-      "grad_norm": 0.002453841036185622,
-      "learning_rate": 0.00019999379177131066,
-      "loss": 46.0,
-      "step": 46401
-    },
-    {
-      "epoch": 3.547756943249804,
-      "grad_norm": 0.0021061059087514877,
-      "learning_rate": 0.00019999379150366256,
-      "loss": 46.0,
-      "step": 46402
-    },
-    {
-      "epoch": 3.547833400233194,
-      "grad_norm": 0.0022286151070147753,
-      "learning_rate": 0.0001999937912360087,
-      "loss": 46.0,
-      "step": 46403
-    },
-    {
-      "epoch": 3.5479098572165837,
-      "grad_norm": 0.001047341269440949,
-      "learning_rate": 0.00019999379096834904,
-      "loss": 46.0,
-      "step": 46404
-    },
-    {
-      "epoch": 3.5479863141999735,
-      "grad_norm": 0.0012548328377306461,
-      "learning_rate": 0.00019999379070068365,
-      "loss": 46.0,
-      "step": 46405
-    },
-    {
-      "epoch": 3.548062771183363,
-      "grad_norm": 0.0012487383792176843,
-      "learning_rate": 0.00019999379043301248,
-      "loss": 46.0,
-      "step": 46406
-    },
-    {
-      "epoch": 3.5481392281667525,
-      "grad_norm": 0.0012695499463006854,
-      "learning_rate": 0.00019999379016533554,
-      "loss": 46.0,
-      "step": 46407
-    },
-    {
-      "epoch": 3.5482156851501423,
-      "grad_norm": 0.0021747243590652943,
-      "learning_rate": 0.00019999378989765286,
-      "loss": 46.0,
-      "step": 46408
-    },
-    {
-      "epoch": 3.548292142133532,
-      "grad_norm": 0.003429760690778494,
-      "learning_rate": 0.00019999378962996437,
-      "loss": 46.0,
-      "step": 46409
-    },
-    {
-      "epoch": 3.548368599116922,
-      "grad_norm": 0.005224948283284903,
-      "learning_rate": 0.0001999937893622701,
-      "loss": 46.0,
-      "step": 46410
-    },
-    {
-      "epoch": 3.5484450561003116,
-      "grad_norm": 0.002717582043260336,
-      "learning_rate": 0.0001999937890945701,
-      "loss": 46.0,
-      "step": 46411
-    },
-    {
-      "epoch": 3.5485215130837013,
-      "grad_norm": 0.0017967562889680266,
-      "learning_rate": 0.00019999378882686432,
-      "loss": 46.0,
-      "step": 46412
-    },
-    {
-      "epoch": 3.548597970067091,
-      "grad_norm": 0.0024169525131583214,
-      "learning_rate": 0.00019999378855915277,
-      "loss": 46.0,
-      "step": 46413
-    },
-    {
-      "epoch": 3.548674427050481,
-      "grad_norm": 0.0017868360737338662,
-      "learning_rate": 0.00019999378829143542,
-      "loss": 46.0,
-      "step": 46414
-    },
-    {
-      "epoch": 3.54875088403387,
-      "grad_norm": 0.0026598009280860424,
-      "learning_rate": 0.00019999378802371235,
-      "loss": 46.0,
-      "step": 46415
-    },
-    {
-      "epoch": 3.54882734101726,
-      "grad_norm": 0.004194796085357666,
-      "learning_rate": 0.00019999378775598347,
-      "loss": 46.0,
-      "step": 46416
-    },
-    {
-      "epoch": 3.5489037980006497,
-      "grad_norm": 0.001359489280730486,
-      "learning_rate": 0.00019999378748824885,
-      "loss": 46.0,
-      "step": 46417
-    },
-    {
-      "epoch": 3.5489802549840395,
-      "grad_norm": 0.001906198333017528,
-      "learning_rate": 0.00019999378722050846,
-      "loss": 46.0,
-      "step": 46418
-    },
-    {
-      "epoch": 3.5490567119674292,
-      "grad_norm": 0.005178088787943125,
-      "learning_rate": 0.0001999937869527623,
-      "loss": 46.0,
-      "step": 46419
-    },
-    {
-      "epoch": 3.549133168950819,
-      "grad_norm": 0.0015865389723330736,
-      "learning_rate": 0.00019999378668501036,
-      "loss": 46.0,
-      "step": 46420
-    },
-    {
-      "epoch": 3.5492096259342087,
-      "grad_norm": 0.006840957794338465,
-      "learning_rate": 0.00019999378641725264,
-      "loss": 46.0,
-      "step": 46421
-    },
-    {
-      "epoch": 3.5492860829175985,
-      "grad_norm": 0.0012211559806019068,
-      "learning_rate": 0.00019999378614948916,
-      "loss": 46.0,
-      "step": 46422
-    },
-    {
-      "epoch": 3.5493625399009883,
-      "grad_norm": 0.0032057706266641617,
-      "learning_rate": 0.00019999378588171992,
-      "loss": 46.0,
-      "step": 46423
-    },
-    {
-      "epoch": 3.549438996884378,
-      "grad_norm": 0.0012986625079065561,
-      "learning_rate": 0.00019999378561394492,
-      "loss": 46.0,
-      "step": 46424
-    },
-    {
-      "epoch": 3.549515453867768,
-      "grad_norm": 0.0052831959910690784,
-      "learning_rate": 0.0001999937853461641,
-      "loss": 46.0,
-      "step": 46425
-    },
-    {
-      "epoch": 3.5495919108511576,
-      "grad_norm": 0.003423130139708519,
-      "learning_rate": 0.00019999378507837756,
-      "loss": 46.0,
-      "step": 46426
-    },
-    {
-      "epoch": 3.5496683678345473,
-      "grad_norm": 0.000631034083198756,
-      "learning_rate": 0.00019999378481058523,
-      "loss": 46.0,
-      "step": 46427
-    },
-    {
-      "epoch": 3.549744824817937,
-      "grad_norm": 0.001113277510739863,
-      "learning_rate": 0.00019999378454278716,
-      "loss": 46.0,
-      "step": 46428
-    },
-    {
-      "epoch": 3.5498212818013264,
-      "grad_norm": 0.0047985706478357315,
-      "learning_rate": 0.0001999937842749833,
-      "loss": 46.0,
-      "step": 46429
-    },
-    {
-      "epoch": 3.549897738784716,
-      "grad_norm": 0.002293867291882634,
-      "learning_rate": 0.00019999378400717367,
-      "loss": 46.0,
-      "step": 46430
-    },
-    {
-      "epoch": 3.549974195768106,
-      "grad_norm": 0.001945320051163435,
-      "learning_rate": 0.00019999378373935828,
-      "loss": 46.0,
-      "step": 46431
-    },
-    {
-      "epoch": 3.5500506527514957,
-      "grad_norm": 0.0036559845320880413,
-      "learning_rate": 0.0001999937834715371,
-      "loss": 46.0,
-      "step": 46432
-    },
-    {
-      "epoch": 3.5501271097348854,
-      "grad_norm": 0.002458605682477355,
-      "learning_rate": 0.00019999378320371017,
-      "loss": 46.0,
-      "step": 46433
-    },
-    {
-      "epoch": 3.550203566718275,
-      "grad_norm": 0.0018103031907230616,
-      "learning_rate": 0.00019999378293587746,
-      "loss": 46.0,
-      "step": 46434
-    },
-    {
-      "epoch": 3.550280023701665,
-      "grad_norm": 0.0023678455036133528,
-      "learning_rate": 0.000199993782668039,
-      "loss": 46.0,
-      "step": 46435
-    },
-    {
-      "epoch": 3.5503564806850547,
-      "grad_norm": 0.0036589547526091337,
-      "learning_rate": 0.00019999378240019474,
-      "loss": 46.0,
-      "step": 46436
-    },
-    {
-      "epoch": 3.550432937668444,
-      "grad_norm": 0.002018094528466463,
-      "learning_rate": 0.00019999378213234474,
-      "loss": 46.0,
-      "step": 46437
-    },
-    {
-      "epoch": 3.550509394651834,
-      "grad_norm": 0.0018803341081365943,
-      "learning_rate": 0.00019999378186448896,
-      "loss": 46.0,
-      "step": 46438
-    },
-    {
-      "epoch": 3.5505858516352236,
-      "grad_norm": 0.0033969832584261894,
-      "learning_rate": 0.0001999937815966274,
-      "loss": 46.0,
-      "step": 46439
-    },
-    {
-      "epoch": 3.5506623086186133,
-      "grad_norm": 0.0015325553249567747,
-      "learning_rate": 0.0001999937813287601,
-      "loss": 46.0,
-      "step": 46440
-    },
-    {
-      "epoch": 3.550738765602003,
-      "grad_norm": 0.001463297288864851,
-      "learning_rate": 0.000199993781060887,
-      "loss": 46.0,
-      "step": 46441
-    },
-    {
-      "epoch": 3.550815222585393,
-      "grad_norm": 0.0024571786634624004,
-      "learning_rate": 0.00019999378079300814,
-      "loss": 46.0,
-      "step": 46442
-    },
-    {
-      "epoch": 3.5508916795687826,
-      "grad_norm": 0.0010557464556768537,
-      "learning_rate": 0.00019999378052512352,
-      "loss": 46.0,
-      "step": 46443
-    },
-    {
-      "epoch": 3.5509681365521724,
-      "grad_norm": 0.002927768277004361,
-      "learning_rate": 0.00019999378025723313,
-      "loss": 46.0,
-      "step": 46444
-    },
-    {
-      "epoch": 3.551044593535562,
-      "grad_norm": 0.006967509631067514,
-      "learning_rate": 0.00019999377998933696,
-      "loss": 46.0,
-      "step": 46445
-    },
-    {
-      "epoch": 3.551121050518952,
-      "grad_norm": 0.000863505934830755,
-      "learning_rate": 0.00019999377972143502,
-      "loss": 46.0,
-      "step": 46446
-    },
-    {
-      "epoch": 3.5511975075023416,
-      "grad_norm": 0.0027712322771549225,
-      "learning_rate": 0.00019999377945352734,
-      "loss": 46.0,
-      "step": 46447
-    },
-    {
-      "epoch": 3.5512739644857314,
-      "grad_norm": 0.0014371549477800727,
-      "learning_rate": 0.00019999377918561385,
-      "loss": 46.0,
-      "step": 46448
-    },
-    {
-      "epoch": 3.551350421469121,
-      "grad_norm": 0.0005209771334193647,
-      "learning_rate": 0.00019999377891769462,
-      "loss": 46.0,
-      "step": 46449
-    },
-    {
-      "epoch": 3.551426878452511,
-      "grad_norm": 0.001574140740558505,
-      "learning_rate": 0.00019999377864976962,
-      "loss": 46.0,
-      "step": 46450
-    },
-    {
-      "epoch": 3.5515033354359002,
-      "grad_norm": 0.0065953535959124565,
-      "learning_rate": 0.0001999937783818388,
-      "loss": 46.0,
-      "step": 46451
-    },
-    {
-      "epoch": 3.55157979241929,
-      "grad_norm": 0.0005561783909797668,
-      "learning_rate": 0.00019999377811390226,
-      "loss": 46.0,
-      "step": 46452
-    },
-    {
-      "epoch": 3.5516562494026798,
-      "grad_norm": 0.0029747439548373222,
-      "learning_rate": 0.00019999377784595996,
-      "loss": 46.0,
-      "step": 46453
-    },
-    {
-      "epoch": 3.5517327063860695,
-      "grad_norm": 0.000876390200573951,
-      "learning_rate": 0.00019999377757801186,
-      "loss": 46.0,
-      "step": 46454
-    },
-    {
-      "epoch": 3.5518091633694593,
-      "grad_norm": 0.0014996856916695833,
-      "learning_rate": 0.00019999377731005802,
-      "loss": 46.0,
-      "step": 46455
-    },
-    {
-      "epoch": 3.551885620352849,
-      "grad_norm": 0.004231185652315617,
-      "learning_rate": 0.0001999937770420984,
-      "loss": 46.0,
-      "step": 46456
-    },
-    {
-      "epoch": 3.551962077336239,
-      "grad_norm": 0.001975968712940812,
-      "learning_rate": 0.00019999377677413298,
-      "loss": 46.0,
-      "step": 46457
-    },
-    {
-      "epoch": 3.5520385343196286,
-      "grad_norm": 0.002437055576592684,
-      "learning_rate": 0.00019999377650616181,
-      "loss": 46.0,
-      "step": 46458
-    },
-    {
-      "epoch": 3.552114991303018,
-      "grad_norm": 0.0014956494560465217,
-      "learning_rate": 0.0001999937762381849,
-      "loss": 46.0,
-      "step": 46459
-    },
-    {
-      "epoch": 3.5521914482864076,
-      "grad_norm": 0.0008825063123367727,
-      "learning_rate": 0.0001999937759702022,
-      "loss": 46.0,
-      "step": 46460
-    },
-    {
-      "epoch": 3.5522679052697974,
-      "grad_norm": 0.005186131224036217,
-      "learning_rate": 0.0001999937757022137,
-      "loss": 46.0,
-      "step": 46461
-    },
-    {
-      "epoch": 3.552344362253187,
-      "grad_norm": 0.004838583059608936,
-      "learning_rate": 0.00019999377543421947,
-      "loss": 46.0,
-      "step": 46462
-    },
-    {
-      "epoch": 3.552420819236577,
-      "grad_norm": 0.0018970175879076123,
-      "learning_rate": 0.00019999377516621947,
-      "loss": 46.0,
-      "step": 46463
-    },
-    {
-      "epoch": 3.5524972762199667,
-      "grad_norm": 0.0008161778678186238,
-      "learning_rate": 0.0001999937748982137,
-      "loss": 46.0,
-      "step": 46464
-    },
-    {
-      "epoch": 3.5525737332033565,
-      "grad_norm": 0.0006800947012379766,
-      "learning_rate": 0.00019999377463020214,
-      "loss": 46.0,
-      "step": 46465
-    },
-    {
-      "epoch": 3.552650190186746,
-      "grad_norm": 0.0011572532821446657,
-      "learning_rate": 0.00019999377436218482,
-      "loss": 46.0,
-      "step": 46466
-    },
-    {
-      "epoch": 3.552726647170136,
-      "grad_norm": 0.0027272903826087713,
-      "learning_rate": 0.00019999377409416174,
-      "loss": 46.0,
-      "step": 46467
-    },
-    {
-      "epoch": 3.5528031041535257,
-      "grad_norm": 0.0016184666892513633,
-      "learning_rate": 0.0001999937738261329,
-      "loss": 46.0,
-      "step": 46468
-    },
-    {
-      "epoch": 3.5528795611369155,
-      "grad_norm": 0.005173943471163511,
-      "learning_rate": 0.00019999377355809825,
-      "loss": 46.0,
-      "step": 46469
-    },
-    {
-      "epoch": 3.5529560181203053,
-      "grad_norm": 0.004095404874533415,
-      "learning_rate": 0.00019999377329005786,
-      "loss": 46.0,
-      "step": 46470
-    },
-    {
-      "epoch": 3.553032475103695,
-      "grad_norm": 0.005040294025093317,
-      "learning_rate": 0.0001999937730220117,
-      "loss": 46.0,
-      "step": 46471
-    },
-    {
-      "epoch": 3.553108932087085,
-      "grad_norm": 0.002450446831062436,
-      "learning_rate": 0.0001999937727539598,
-      "loss": 46.0,
-      "step": 46472
-    },
-    {
-      "epoch": 3.553185389070474,
-      "grad_norm": 0.003448336385190487,
-      "learning_rate": 0.0001999937724859021,
-      "loss": 46.0,
-      "step": 46473
-    },
-    {
-      "epoch": 3.553261846053864,
-      "grad_norm": 0.00127406045794487,
-      "learning_rate": 0.00019999377221783862,
-      "loss": 46.0,
-      "step": 46474
-    },
-    {
-      "epoch": 3.5533383030372536,
-      "grad_norm": 0.0021221397910267115,
-      "learning_rate": 0.0001999937719497694,
-      "loss": 46.0,
-      "step": 46475
-    },
-    {
-      "epoch": 3.5534147600206434,
-      "grad_norm": 0.0017270802054554224,
-      "learning_rate": 0.00019999377168169438,
-      "loss": 46.0,
-      "step": 46476
-    },
-    {
-      "epoch": 3.553491217004033,
-      "grad_norm": 0.003796135075390339,
-      "learning_rate": 0.0001999937714136136,
-      "loss": 46.0,
-      "step": 46477
-    },
-    {
-      "epoch": 3.553567673987423,
-      "grad_norm": 0.002585887210443616,
-      "learning_rate": 0.00019999377114552705,
-      "loss": 46.0,
-      "step": 46478
-    },
-    {
-      "epoch": 3.5536441309708127,
-      "grad_norm": 0.0027438232209533453,
-      "learning_rate": 0.00019999377087743473,
-      "loss": 46.0,
-      "step": 46479
-    },
-    {
-      "epoch": 3.5537205879542024,
-      "grad_norm": 0.0016532428562641144,
-      "learning_rate": 0.00019999377060933666,
-      "loss": 46.0,
-      "step": 46480
-    },
-    {
-      "epoch": 3.5537970449375917,
-      "grad_norm": 0.0029786117374897003,
-      "learning_rate": 0.00019999377034123281,
-      "loss": 46.0,
-      "step": 46481
-    },
-    {
-      "epoch": 3.5538735019209815,
-      "grad_norm": 0.0011823297245427966,
-      "learning_rate": 0.0001999937700731232,
-      "loss": 46.0,
-      "step": 46482
-    },
-    {
-      "epoch": 3.5539499589043713,
-      "grad_norm": 0.0011401071678847075,
-      "learning_rate": 0.00019999376980500778,
-      "loss": 46.0,
-      "step": 46483
-    },
-    {
-      "epoch": 3.554026415887761,
-      "grad_norm": 0.0013400858733803034,
-      "learning_rate": 0.00019999376953688661,
-      "loss": 46.0,
-      "step": 46484
-    },
-    {
-      "epoch": 3.554102872871151,
-      "grad_norm": 0.0018520744051784277,
-      "learning_rate": 0.0001999937692687597,
-      "loss": 46.0,
-      "step": 46485
-    },
-    {
-      "epoch": 3.5541793298545405,
-      "grad_norm": 0.003796538570895791,
-      "learning_rate": 0.00019999376900062702,
-      "loss": 46.0,
-      "step": 46486
-    },
-    {
-      "epoch": 3.5542557868379303,
-      "grad_norm": 0.0027337847277522087,
-      "learning_rate": 0.0001999937687324885,
-      "loss": 46.0,
-      "step": 46487
-    },
-    {
-      "epoch": 3.55433224382132,
-      "grad_norm": 0.0021763781551271677,
-      "learning_rate": 0.0001999937684643443,
-      "loss": 46.0,
-      "step": 46488
-    },
-    {
-      "epoch": 3.55440870080471,
-      "grad_norm": 0.0014338461915031075,
-      "learning_rate": 0.0001999937681961943,
-      "loss": 46.0,
-      "step": 46489
-    },
-    {
-      "epoch": 3.5544851577880996,
-      "grad_norm": 0.0013220143737271428,
-      "learning_rate": 0.00019999376792803852,
-      "loss": 46.0,
-      "step": 46490
-    },
-    {
-      "epoch": 3.5545616147714894,
-      "grad_norm": 0.004611336160451174,
-      "learning_rate": 0.00019999376765987697,
-      "loss": 46.0,
-      "step": 46491
-    },
-    {
-      "epoch": 3.554638071754879,
-      "grad_norm": 0.001799620222300291,
-      "learning_rate": 0.00019999376739170965,
-      "loss": 46.0,
-      "step": 46492
-    },
-    {
-      "epoch": 3.554714528738269,
-      "grad_norm": 0.0015801709378138185,
-      "learning_rate": 0.00019999376712353658,
-      "loss": 46.0,
-      "step": 46493
-    },
-    {
-      "epoch": 3.554790985721658,
-      "grad_norm": 0.001817018841393292,
-      "learning_rate": 0.00019999376685535773,
-      "loss": 46.0,
-      "step": 46494
-    },
-    {
-      "epoch": 3.554867442705048,
-      "grad_norm": 0.005005191545933485,
-      "learning_rate": 0.00019999376658717312,
-      "loss": 46.0,
-      "step": 46495
-    },
-    {
-      "epoch": 3.5549438996884377,
-      "grad_norm": 0.0038010175339877605,
-      "learning_rate": 0.00019999376631898272,
-      "loss": 46.0,
-      "step": 46496
-    },
-    {
-      "epoch": 3.5550203566718275,
-      "grad_norm": 0.0030369756277650595,
-      "learning_rate": 0.00019999376605078656,
-      "loss": 46.0,
-      "step": 46497
-    },
-    {
-      "epoch": 3.5550968136552172,
-      "grad_norm": 0.0028636339120566845,
-      "learning_rate": 0.00019999376578258465,
-      "loss": 46.0,
-      "step": 46498
-    },
-    {
-      "epoch": 3.555173270638607,
-      "grad_norm": 0.005953289568424225,
-      "learning_rate": 0.00019999376551437694,
-      "loss": 46.0,
-      "step": 46499
-    },
-    {
-      "epoch": 3.5552497276219968,
-      "grad_norm": 0.00319922948256135,
-      "learning_rate": 0.00019999376524616346,
-      "loss": 46.0,
-      "step": 46500
-    },
-    {
-      "epoch": 3.5553261846053865,
-      "grad_norm": 0.0017837381456047297,
-      "learning_rate": 0.00019999376497794425,
-      "loss": 46.0,
-      "step": 46501
-    },
-    {
-      "epoch": 3.5554026415887763,
-      "grad_norm": 0.0018834981601685286,
-      "learning_rate": 0.00019999376470971925,
-      "loss": 46.0,
-      "step": 46502
-    },
-    {
-      "epoch": 3.5554790985721656,
-      "grad_norm": 0.001960393972694874,
-      "learning_rate": 0.00019999376444148847,
-      "loss": 46.0,
-      "step": 46503
-    },
-    {
-      "epoch": 3.5555555555555554,
-      "grad_norm": 0.001821323181502521,
-      "learning_rate": 0.00019999376417325192,
-      "loss": 46.0,
-      "step": 46504
-    },
-    {
-      "epoch": 3.555632012538945,
-      "grad_norm": 0.003700501285493374,
-      "learning_rate": 0.00019999376390500962,
-      "loss": 46.0,
-      "step": 46505
-    },
-    {
-      "epoch": 3.555708469522335,
-      "grad_norm": 0.0017601598519831896,
-      "learning_rate": 0.00019999376363676153,
-      "loss": 46.0,
-      "step": 46506
-    },
-    {
-      "epoch": 3.5557849265057246,
-      "grad_norm": 0.006738262716680765,
-      "learning_rate": 0.00019999376336850768,
-      "loss": 46.0,
-      "step": 46507
-    },
-    {
-      "epoch": 3.5558613834891144,
-      "grad_norm": 0.0055123502388596535,
-      "learning_rate": 0.00019999376310024807,
-      "loss": 46.0,
-      "step": 46508
-    },
-    {
-      "epoch": 3.555937840472504,
-      "grad_norm": 0.003295546630397439,
-      "learning_rate": 0.00019999376283198267,
-      "loss": 46.0,
-      "step": 46509
-    },
-    {
-      "epoch": 3.556014297455894,
-      "grad_norm": 0.0004722021985799074,
-      "learning_rate": 0.0001999937625637115,
-      "loss": 46.0,
-      "step": 46510
-    },
-    {
-      "epoch": 3.5560907544392837,
-      "grad_norm": 0.005691759288311005,
-      "learning_rate": 0.0001999937622954346,
-      "loss": 46.0,
-      "step": 46511
-    },
-    {
-      "epoch": 3.5561672114226734,
-      "grad_norm": 0.0027579213492572308,
-      "learning_rate": 0.00019999376202715192,
-      "loss": 46.0,
-      "step": 46512
-    },
-    {
-      "epoch": 3.556243668406063,
-      "grad_norm": 0.004934835247695446,
-      "learning_rate": 0.00019999376175886343,
-      "loss": 46.0,
-      "step": 46513
-    },
-    {
-      "epoch": 3.556320125389453,
-      "grad_norm": 0.0014789033448323607,
-      "learning_rate": 0.0001999937614905692,
-      "loss": 46.0,
-      "step": 46514
-    },
-    {
-      "epoch": 3.5563965823728427,
-      "grad_norm": 0.0017723443452268839,
-      "learning_rate": 0.0001999937612222692,
-      "loss": 46.0,
-      "step": 46515
-    },
-    {
-      "epoch": 3.556473039356232,
-      "grad_norm": 0.0013103695819154382,
-      "learning_rate": 0.00019999376095396345,
-      "loss": 46.0,
-      "step": 46516
-    },
-    {
-      "epoch": 3.556549496339622,
-      "grad_norm": 0.001224347623065114,
-      "learning_rate": 0.0001999937606856519,
-      "loss": 46.0,
-      "step": 46517
-    },
-    {
-      "epoch": 3.5566259533230116,
-      "grad_norm": 0.0010508011328056455,
-      "learning_rate": 0.00019999376041733458,
-      "loss": 46.0,
-      "step": 46518
-    },
-    {
-      "epoch": 3.5567024103064013,
-      "grad_norm": 0.0036310760769993067,
-      "learning_rate": 0.00019999376014901153,
-      "loss": 46.0,
-      "step": 46519
-    },
-    {
-      "epoch": 3.556778867289791,
-      "grad_norm": 0.005407261196523905,
-      "learning_rate": 0.00019999375988068266,
-      "loss": 46.0,
-      "step": 46520
-    },
-    {
-      "epoch": 3.556855324273181,
-      "grad_norm": 0.003659235080704093,
-      "learning_rate": 0.00019999375961234805,
-      "loss": 46.0,
-      "step": 46521
-    },
-    {
-      "epoch": 3.5569317812565706,
-      "grad_norm": 0.0022064486984163523,
-      "learning_rate": 0.00019999375934400766,
-      "loss": 46.0,
-      "step": 46522
-    },
-    {
-      "epoch": 3.5570082382399604,
-      "grad_norm": 0.002823495538905263,
-      "learning_rate": 0.0001999937590756615,
-      "loss": 46.0,
-      "step": 46523
-    },
-    {
-      "epoch": 3.5570846952233497,
-      "grad_norm": 0.0018522243481129408,
-      "learning_rate": 0.00019999375880730958,
-      "loss": 46.0,
-      "step": 46524
-    },
-    {
-      "epoch": 3.5571611522067395,
-      "grad_norm": 0.0035180626437067986,
-      "learning_rate": 0.00019999375853895187,
-      "loss": 46.0,
-      "step": 46525
-    },
-    {
-      "epoch": 3.557237609190129,
-      "grad_norm": 0.001274880371056497,
-      "learning_rate": 0.00019999375827058842,
-      "loss": 46.0,
-      "step": 46526
-    },
-    {
-      "epoch": 3.557314066173519,
-      "grad_norm": 0.002875514095649123,
-      "learning_rate": 0.0001999937580022192,
-      "loss": 46.0,
-      "step": 46527
-    },
-    {
-      "epoch": 3.5573905231569087,
-      "grad_norm": 0.0007261926075443625,
-      "learning_rate": 0.0001999937577338442,
-      "loss": 46.0,
-      "step": 46528
-    },
-    {
-      "epoch": 3.5574669801402985,
-      "grad_norm": 0.002841297071427107,
-      "learning_rate": 0.00019999375746546343,
-      "loss": 46.0,
-      "step": 46529
-    },
-    {
-      "epoch": 3.5575434371236883,
-      "grad_norm": 0.0014934537466615438,
-      "learning_rate": 0.00019999375719707688,
-      "loss": 46.0,
-      "step": 46530
-    },
-    {
-      "epoch": 3.557619894107078,
-      "grad_norm": 0.0026760082691907883,
-      "learning_rate": 0.0001999937569286846,
-      "loss": 46.0,
-      "step": 46531
-    },
-    {
-      "epoch": 3.557696351090468,
-      "grad_norm": 0.0012086716014891863,
-      "learning_rate": 0.00019999375666028652,
-      "loss": 46.0,
-      "step": 46532
-    },
-    {
-      "epoch": 3.5577728080738575,
-      "grad_norm": 0.0038816449232399464,
-      "learning_rate": 0.00019999375639188268,
-      "loss": 46.0,
-      "step": 46533
-    },
-    {
-      "epoch": 3.5578492650572473,
-      "grad_norm": 0.0015986757352948189,
-      "learning_rate": 0.00019999375612347303,
-      "loss": 46.0,
-      "step": 46534
-    },
-    {
-      "epoch": 3.557925722040637,
-      "grad_norm": 0.0013071264838799834,
-      "learning_rate": 0.00019999375585505767,
-      "loss": 46.0,
-      "step": 46535
-    },
-    {
-      "epoch": 3.558002179024027,
-      "grad_norm": 0.0009613691945560277,
-      "learning_rate": 0.0001999937555866365,
-      "loss": 46.0,
-      "step": 46536
-    },
-    {
-      "epoch": 3.5580786360074166,
-      "grad_norm": 0.0010414548451080918,
-      "learning_rate": 0.0001999937553182096,
-      "loss": 46.0,
-      "step": 46537
-    },
-    {
-      "epoch": 3.558155092990806,
-      "grad_norm": 0.028348220512270927,
-      "learning_rate": 0.0001999937550497769,
-      "loss": 46.0,
-      "step": 46538
-    },
-    {
-      "epoch": 3.5582315499741957,
-      "grad_norm": 0.0008071522461250424,
-      "learning_rate": 0.00019999375478133843,
-      "loss": 46.0,
-      "step": 46539
-    },
-    {
-      "epoch": 3.5583080069575854,
-      "grad_norm": 0.004722594749182463,
-      "learning_rate": 0.00019999375451289423,
-      "loss": 46.0,
-      "step": 46540
-    },
-    {
-      "epoch": 3.558384463940975,
-      "grad_norm": 0.002216981491073966,
-      "learning_rate": 0.0001999937542444442,
-      "loss": 46.0,
-      "step": 46541
-    },
-    {
-      "epoch": 3.558460920924365,
-      "grad_norm": 0.0016356802079826593,
-      "learning_rate": 0.00019999375397598845,
-      "loss": 46.0,
-      "step": 46542
-    },
-    {
-      "epoch": 3.5585373779077547,
-      "grad_norm": 0.0017779436893761158,
-      "learning_rate": 0.0001999937537075269,
-      "loss": 46.0,
-      "step": 46543
-    },
-    {
-      "epoch": 3.5586138348911445,
-      "grad_norm": 0.0033922898583114147,
-      "learning_rate": 0.0001999937534390596,
-      "loss": 46.0,
-      "step": 46544
-    },
-    {
-      "epoch": 3.5586902918745342,
-      "grad_norm": 0.004556615371257067,
-      "learning_rate": 0.00019999375317058653,
-      "loss": 46.0,
-      "step": 46545
-    },
-    {
-      "epoch": 3.5587667488579235,
-      "grad_norm": 0.00324404239654541,
-      "learning_rate": 0.0001999937529021077,
-      "loss": 46.0,
-      "step": 46546
-    },
-    {
-      "epoch": 3.5588432058413133,
-      "grad_norm": 0.0030002316925674677,
-      "learning_rate": 0.00019999375263362308,
-      "loss": 46.0,
-      "step": 46547
-    },
-    {
-      "epoch": 3.558919662824703,
-      "grad_norm": 0.0014597923727706075,
-      "learning_rate": 0.00019999375236513269,
-      "loss": 46.0,
-      "step": 46548
-    },
-    {
-      "epoch": 3.558996119808093,
-      "grad_norm": 0.0022793463431298733,
-      "learning_rate": 0.00019999375209663652,
-      "loss": 46.0,
-      "step": 46549
-    },
-    {
-      "epoch": 3.5590725767914826,
-      "grad_norm": 0.0021507972851395607,
-      "learning_rate": 0.00019999375182813461,
-      "loss": 46.0,
-      "step": 46550
-    },
-    {
-      "epoch": 3.5591490337748724,
-      "grad_norm": 0.004458022769540548,
-      "learning_rate": 0.00019999375155962693,
-      "loss": 46.0,
-      "step": 46551
-    },
-    {
-      "epoch": 3.559225490758262,
-      "grad_norm": 0.0015177782624959946,
-      "learning_rate": 0.00019999375129111348,
-      "loss": 46.0,
-      "step": 46552
-    },
-    {
-      "epoch": 3.559301947741652,
-      "grad_norm": 0.004764895886182785,
-      "learning_rate": 0.00019999375102259425,
-      "loss": 46.0,
-      "step": 46553
-    },
-    {
-      "epoch": 3.5593784047250416,
-      "grad_norm": 0.0014271772233769298,
-      "learning_rate": 0.00019999375075406924,
-      "loss": 46.0,
-      "step": 46554
-    },
-    {
-      "epoch": 3.5594548617084314,
-      "grad_norm": 0.0007983514224179089,
-      "learning_rate": 0.00019999375048553847,
-      "loss": 46.0,
-      "step": 46555
-    },
-    {
-      "epoch": 3.559531318691821,
-      "grad_norm": 0.0009628830011934042,
-      "learning_rate": 0.00019999375021700195,
-      "loss": 46.0,
-      "step": 46556
-    },
-    {
-      "epoch": 3.559607775675211,
-      "grad_norm": 0.0007626317674294114,
-      "learning_rate": 0.00019999374994845965,
-      "loss": 46.0,
-      "step": 46557
-    },
-    {
-      "epoch": 3.5596842326586007,
-      "grad_norm": 0.0015915690455585718,
-      "learning_rate": 0.00019999374967991158,
-      "loss": 46.0,
-      "step": 46558
-    },
-    {
-      "epoch": 3.5597606896419904,
-      "grad_norm": 0.0008220255840569735,
-      "learning_rate": 0.00019999374941135774,
-      "loss": 46.0,
-      "step": 46559
-    },
-    {
-      "epoch": 3.5598371466253798,
-      "grad_norm": 0.0023698657751083374,
-      "learning_rate": 0.00019999374914279812,
-      "loss": 46.0,
-      "step": 46560
-    },
-    {
-      "epoch": 3.5599136036087695,
-      "grad_norm": 0.011392847634851933,
-      "learning_rate": 0.00019999374887423273,
-      "loss": 46.0,
-      "step": 46561
-    },
-    {
-      "epoch": 3.5599900605921593,
-      "grad_norm": 0.0025960872881114483,
-      "learning_rate": 0.0001999937486056616,
-      "loss": 46.0,
-      "step": 46562
-    },
-    {
-      "epoch": 3.560066517575549,
-      "grad_norm": 0.003866950748488307,
-      "learning_rate": 0.00019999374833708466,
-      "loss": 46.0,
-      "step": 46563
-    },
-    {
-      "epoch": 3.560142974558939,
-      "grad_norm": 0.003444395959377289,
-      "learning_rate": 0.000199993748068502,
-      "loss": 46.0,
-      "step": 46564
-    },
-    {
-      "epoch": 3.5602194315423286,
-      "grad_norm": 0.0011637936113402247,
-      "learning_rate": 0.00019999374779991353,
-      "loss": 46.0,
-      "step": 46565
-    },
-    {
-      "epoch": 3.5602958885257183,
-      "grad_norm": 0.0029007100965827703,
-      "learning_rate": 0.0001999937475313193,
-      "loss": 46.0,
-      "step": 46566
-    },
-    {
-      "epoch": 3.560372345509108,
-      "grad_norm": 0.0010651706252247095,
-      "learning_rate": 0.00019999374726271932,
-      "loss": 46.0,
-      "step": 46567
-    },
-    {
-      "epoch": 3.5604488024924974,
-      "grad_norm": 0.002920096041634679,
-      "learning_rate": 0.00019999374699411355,
-      "loss": 46.0,
-      "step": 46568
-    },
-    {
-      "epoch": 3.560525259475887,
-      "grad_norm": 0.0039646257646381855,
-      "learning_rate": 0.00019999374672550202,
-      "loss": 46.0,
-      "step": 46569
-    },
-    {
-      "epoch": 3.560601716459277,
-      "grad_norm": 0.003384967800229788,
-      "learning_rate": 0.0001999937464568847,
-      "loss": 46.0,
-      "step": 46570
-    },
-    {
-      "epoch": 3.5606781734426667,
-      "grad_norm": 0.0016989994328469038,
-      "learning_rate": 0.00019999374618826163,
-      "loss": 46.0,
-      "step": 46571
-    },
-    {
-      "epoch": 3.5607546304260564,
-      "grad_norm": 0.006379690486937761,
-      "learning_rate": 0.0001999937459196328,
-      "loss": 46.0,
-      "step": 46572
-    },
-    {
-      "epoch": 3.560831087409446,
-      "grad_norm": 0.002185478340834379,
-      "learning_rate": 0.0001999937456509982,
-      "loss": 46.0,
-      "step": 46573
-    },
-    {
-      "epoch": 3.560907544392836,
-      "grad_norm": 0.004664817824959755,
-      "learning_rate": 0.0001999937453823578,
-      "loss": 46.0,
-      "step": 46574
-    },
-    {
-      "epoch": 3.5609840013762257,
-      "grad_norm": 0.005241215229034424,
-      "learning_rate": 0.00019999374511371168,
-      "loss": 46.0,
-      "step": 46575
-    },
-    {
-      "epoch": 3.5610604583596155,
-      "grad_norm": 0.0028947058599442244,
-      "learning_rate": 0.00019999374484505977,
-      "loss": 46.0,
-      "step": 46576
-    },
-    {
-      "epoch": 3.5611369153430052,
-      "grad_norm": 0.0010516849579289556,
-      "learning_rate": 0.00019999374457640206,
-      "loss": 46.0,
-      "step": 46577
-    },
-    {
-      "epoch": 3.561213372326395,
-      "grad_norm": 0.003213840536773205,
-      "learning_rate": 0.0001999937443077386,
-      "loss": 46.0,
-      "step": 46578
-    },
-    {
-      "epoch": 3.5612898293097848,
-      "grad_norm": 0.0023612165823578835,
-      "learning_rate": 0.00019999374403906938,
-      "loss": 46.0,
-      "step": 46579
-    },
-    {
-      "epoch": 3.5613662862931745,
-      "grad_norm": 0.005796457175165415,
-      "learning_rate": 0.0001999937437703944,
-      "loss": 46.0,
-      "step": 46580
-    },
-    {
-      "epoch": 3.5614427432765643,
-      "grad_norm": 0.0010385580826550722,
-      "learning_rate": 0.00019999374350171363,
-      "loss": 46.0,
-      "step": 46581
-    },
-    {
-      "epoch": 3.5615192002599536,
-      "grad_norm": 0.0008809170685708523,
-      "learning_rate": 0.0001999937432330271,
-      "loss": 46.0,
-      "step": 46582
-    },
-    {
-      "epoch": 3.5615956572433434,
-      "grad_norm": 0.0015432793879881501,
-      "learning_rate": 0.0001999937429643348,
-      "loss": 46.0,
-      "step": 46583
-    },
-    {
-      "epoch": 3.561672114226733,
-      "grad_norm": 0.0006031371303834021,
-      "learning_rate": 0.00019999374269563674,
-      "loss": 46.0,
-      "step": 46584
-    },
-    {
-      "epoch": 3.561748571210123,
-      "grad_norm": 0.0018374252831563354,
-      "learning_rate": 0.0001999937424269329,
-      "loss": 46.0,
-      "step": 46585
-    },
-    {
-      "epoch": 3.5618250281935127,
-      "grad_norm": 0.001013958710245788,
-      "learning_rate": 0.00019999374215822329,
-      "loss": 46.0,
-      "step": 46586
-    },
-    {
-      "epoch": 3.5619014851769024,
-      "grad_norm": 0.0010941693326458335,
-      "learning_rate": 0.00019999374188950792,
-      "loss": 46.0,
-      "step": 46587
-    },
-    {
-      "epoch": 3.561977942160292,
-      "grad_norm": 0.0035610490012913942,
-      "learning_rate": 0.00019999374162078676,
-      "loss": 46.0,
-      "step": 46588
-    },
-    {
-      "epoch": 3.562054399143682,
-      "grad_norm": 0.004505632445216179,
-      "learning_rate": 0.00019999374135205985,
-      "loss": 46.0,
-      "step": 46589
-    },
-    {
-      "epoch": 3.5621308561270713,
-      "grad_norm": 0.0037538534961640835,
-      "learning_rate": 0.00019999374108332715,
-      "loss": 46.0,
-      "step": 46590
-    },
-    {
-      "epoch": 3.562207313110461,
-      "grad_norm": 0.002848175587132573,
-      "learning_rate": 0.0001999937408145887,
-      "loss": 46.0,
-      "step": 46591
-    },
-    {
-      "epoch": 3.5622837700938508,
-      "grad_norm": 0.0009538735612295568,
-      "learning_rate": 0.0001999937405458445,
-      "loss": 46.0,
-      "step": 46592
-    },
-    {
-      "epoch": 3.5623602270772405,
-      "grad_norm": 0.013432635925710201,
-      "learning_rate": 0.00019999374027709451,
-      "loss": 46.0,
-      "step": 46593
-    },
-    {
-      "epoch": 3.5624366840606303,
-      "grad_norm": 0.002499157562851906,
-      "learning_rate": 0.00019999374000833874,
-      "loss": 46.0,
-      "step": 46594
-    },
-    {
-      "epoch": 3.56251314104402,
-      "grad_norm": 0.0021435744129121304,
-      "learning_rate": 0.0001999937397395772,
-      "loss": 46.0,
-      "step": 46595
-    },
-    {
-      "epoch": 3.56258959802741,
-      "grad_norm": 0.0029948113951832056,
-      "learning_rate": 0.00019999373947080992,
-      "loss": 46.0,
-      "step": 46596
-    },
-    {
-      "epoch": 3.5626660550107996,
-      "grad_norm": 0.005200231913477182,
-      "learning_rate": 0.00019999373920203686,
-      "loss": 46.0,
-      "step": 46597
-    },
-    {
-      "epoch": 3.5627425119941893,
-      "grad_norm": 0.0023839629720896482,
-      "learning_rate": 0.00019999373893325801,
-      "loss": 46.0,
-      "step": 46598
-    },
-    {
-      "epoch": 3.562818968977579,
-      "grad_norm": 0.0025282285641878843,
-      "learning_rate": 0.00019999373866447343,
-      "loss": 46.0,
-      "step": 46599
-    },
-    {
-      "epoch": 3.562895425960969,
-      "grad_norm": 0.0023611115757375956,
-      "learning_rate": 0.00019999373839568304,
-      "loss": 46.0,
-      "step": 46600
-    },
-    {
-      "epoch": 3.5629718829443586,
-      "grad_norm": 0.0015852905344218016,
-      "learning_rate": 0.0001999937381268869,
-      "loss": 46.0,
-      "step": 46601
-    },
-    {
-      "epoch": 3.5630483399277484,
-      "grad_norm": 0.0030706808902323246,
-      "learning_rate": 0.000199993737858085,
-      "loss": 46.0,
-      "step": 46602
-    },
-    {
-      "epoch": 3.563124796911138,
-      "grad_norm": 0.0009118703310377896,
-      "learning_rate": 0.00019999373758927732,
-      "loss": 46.0,
-      "step": 46603
-    },
-    {
-      "epoch": 3.5632012538945275,
-      "grad_norm": 0.003964750096201897,
-      "learning_rate": 0.00019999373732046386,
-      "loss": 46.0,
-      "step": 46604
-    },
-    {
-      "epoch": 3.5632777108779172,
-      "grad_norm": 0.0021664509549736977,
-      "learning_rate": 0.0001999937370516446,
-      "loss": 46.0,
-      "step": 46605
-    },
-    {
-      "epoch": 3.563354167861307,
-      "grad_norm": 0.002441578544676304,
-      "learning_rate": 0.00019999373678281966,
-      "loss": 46.0,
-      "step": 46606
-    },
-    {
-      "epoch": 3.5634306248446967,
-      "grad_norm": 0.004548076540231705,
-      "learning_rate": 0.00019999373651398888,
-      "loss": 46.0,
-      "step": 46607
-    },
-    {
-      "epoch": 3.5635070818280865,
-      "grad_norm": 0.0036300229839980602,
-      "learning_rate": 0.00019999373624515236,
-      "loss": 46.0,
-      "step": 46608
-    },
-    {
-      "epoch": 3.5635835388114763,
-      "grad_norm": 0.001155280857346952,
-      "learning_rate": 0.00019999373597631007,
-      "loss": 46.0,
-      "step": 46609
-    },
-    {
-      "epoch": 3.563659995794866,
-      "grad_norm": 0.0022164478432387114,
-      "learning_rate": 0.00019999373570746197,
-      "loss": 46.0,
-      "step": 46610
-    },
-    {
-      "epoch": 3.563736452778256,
-      "grad_norm": 0.0008368267444893718,
-      "learning_rate": 0.00019999373543860816,
-      "loss": 46.0,
-      "step": 46611
-    },
-    {
-      "epoch": 3.563812909761645,
-      "grad_norm": 0.00203518895432353,
-      "learning_rate": 0.00019999373516974855,
-      "loss": 46.0,
-      "step": 46612
-    },
-    {
-      "epoch": 3.563889366745035,
-      "grad_norm": 0.0026849994901567698,
-      "learning_rate": 0.00019999373490088318,
-      "loss": 46.0,
-      "step": 46613
-    },
-    {
-      "epoch": 3.5639658237284246,
-      "grad_norm": 0.005631974898278713,
-      "learning_rate": 0.00019999373463201202,
-      "loss": 46.0,
-      "step": 46614
-    },
-    {
-      "epoch": 3.5640422807118144,
-      "grad_norm": 0.003811551257967949,
-      "learning_rate": 0.00019999373436313514,
-      "loss": 46.0,
-      "step": 46615
-    },
-    {
-      "epoch": 3.564118737695204,
-      "grad_norm": 0.001514188596047461,
-      "learning_rate": 0.00019999373409425246,
-      "loss": 46.0,
-      "step": 46616
-    },
-    {
-      "epoch": 3.564195194678594,
-      "grad_norm": 0.001378338783979416,
-      "learning_rate": 0.00019999373382536398,
-      "loss": 46.0,
-      "step": 46617
-    },
-    {
-      "epoch": 3.5642716516619837,
-      "grad_norm": 0.002717813942581415,
-      "learning_rate": 0.00019999373355646978,
-      "loss": 46.0,
-      "step": 46618
-    },
-    {
-      "epoch": 3.5643481086453734,
-      "grad_norm": 0.0022252292837947607,
-      "learning_rate": 0.00019999373328756978,
-      "loss": 46.0,
-      "step": 46619
-    },
-    {
-      "epoch": 3.564424565628763,
-      "grad_norm": 0.003563227830454707,
-      "learning_rate": 0.00019999373301866403,
-      "loss": 46.0,
-      "step": 46620
-    },
-    {
-      "epoch": 3.564501022612153,
-      "grad_norm": 0.00731264753267169,
-      "learning_rate": 0.00019999373274975254,
-      "loss": 46.0,
-      "step": 46621
-    },
-    {
-      "epoch": 3.5645774795955427,
-      "grad_norm": 0.003943635616451502,
-      "learning_rate": 0.00019999373248083522,
-      "loss": 46.0,
-      "step": 46622
-    },
-    {
-      "epoch": 3.5646539365789325,
-      "grad_norm": 0.000804942159447819,
-      "learning_rate": 0.00019999373221191215,
-      "loss": 46.0,
-      "step": 46623
-    },
-    {
-      "epoch": 3.5647303935623222,
-      "grad_norm": 0.0016056118765845895,
-      "learning_rate": 0.0001999937319429833,
-      "loss": 46.0,
-      "step": 46624
-    },
-    {
-      "epoch": 3.5648068505457116,
-      "grad_norm": 0.0024353060871362686,
-      "learning_rate": 0.00019999373167404872,
-      "loss": 46.0,
-      "step": 46625
-    },
-    {
-      "epoch": 3.5648833075291013,
-      "grad_norm": 0.0011932393535971642,
-      "learning_rate": 0.00019999373140510836,
-      "loss": 46.0,
-      "step": 46626
-    },
-    {
-      "epoch": 3.564959764512491,
-      "grad_norm": 0.0017103042919188738,
-      "learning_rate": 0.0001999937311361622,
-      "loss": 46.0,
-      "step": 46627
-    },
-    {
-      "epoch": 3.565036221495881,
-      "grad_norm": 0.0029872881714254618,
-      "learning_rate": 0.0001999937308672103,
-      "loss": 46.0,
-      "step": 46628
-    },
-    {
-      "epoch": 3.5651126784792706,
-      "grad_norm": 0.0030780588276684284,
-      "learning_rate": 0.0001999937305982526,
-      "loss": 46.0,
-      "step": 46629
-    },
-    {
-      "epoch": 3.5651891354626604,
-      "grad_norm": 0.0028714691288769245,
-      "learning_rate": 0.00019999373032928919,
-      "loss": 46.0,
-      "step": 46630
-    },
-    {
-      "epoch": 3.56526559244605,
-      "grad_norm": 0.0037491903640329838,
-      "learning_rate": 0.00019999373006031996,
-      "loss": 46.0,
-      "step": 46631
-    },
-    {
-      "epoch": 3.56534204942944,
-      "grad_norm": 0.004280867520719767,
-      "learning_rate": 0.00019999372979134496,
-      "loss": 46.0,
-      "step": 46632
-    },
-    {
-      "epoch": 3.5654185064128296,
-      "grad_norm": 0.0014825878897681832,
-      "learning_rate": 0.0001999937295223642,
-      "loss": 46.0,
-      "step": 46633
-    },
-    {
-      "epoch": 3.565494963396219,
-      "grad_norm": 0.0019989258144050837,
-      "learning_rate": 0.0001999937292533777,
-      "loss": 46.0,
-      "step": 46634
-    },
-    {
-      "epoch": 3.5655714203796087,
-      "grad_norm": 0.0010208177845925093,
-      "learning_rate": 0.0001999937289843854,
-      "loss": 46.0,
-      "step": 46635
-    },
-    {
-      "epoch": 3.5656478773629985,
-      "grad_norm": 0.0031489187385886908,
-      "learning_rate": 0.00019999372871538733,
-      "loss": 46.0,
-      "step": 46636
-    },
-    {
-      "epoch": 3.5657243343463882,
-      "grad_norm": 0.002669918118044734,
-      "learning_rate": 0.00019999372844638352,
-      "loss": 46.0,
-      "step": 46637
-    },
-    {
-      "epoch": 3.565800791329778,
-      "grad_norm": 0.0029144948348402977,
-      "learning_rate": 0.00019999372817737393,
-      "loss": 46.0,
-      "step": 46638
-    },
-    {
-      "epoch": 3.5658772483131678,
-      "grad_norm": 0.0024519572034478188,
-      "learning_rate": 0.00019999372790835854,
-      "loss": 46.0,
-      "step": 46639
-    },
-    {
-      "epoch": 3.5659537052965575,
-      "grad_norm": 0.0015700676012784243,
-      "learning_rate": 0.0001999937276393374,
-      "loss": 46.0,
-      "step": 46640
-    },
-    {
-      "epoch": 3.5660301622799473,
-      "grad_norm": 0.004720539320260286,
-      "learning_rate": 0.00019999372737031047,
-      "loss": 46.0,
-      "step": 46641
-    },
-    {
-      "epoch": 3.566106619263337,
-      "grad_norm": 0.0041053444147109985,
-      "learning_rate": 0.00019999372710127782,
-      "loss": 46.0,
-      "step": 46642
-    },
-    {
-      "epoch": 3.566183076246727,
-      "grad_norm": 0.0029860378708690405,
-      "learning_rate": 0.00019999372683223937,
-      "loss": 46.0,
-      "step": 46643
-    },
-    {
-      "epoch": 3.5662595332301166,
-      "grad_norm": 0.0023234044201672077,
-      "learning_rate": 0.00019999372656319517,
-      "loss": 46.0,
-      "step": 46644
-    },
-    {
-      "epoch": 3.5663359902135063,
-      "grad_norm": 0.006589425727725029,
-      "learning_rate": 0.00019999372629414517,
-      "loss": 46.0,
-      "step": 46645
-    },
-    {
-      "epoch": 3.566412447196896,
-      "grad_norm": 0.001928093028254807,
-      "learning_rate": 0.00019999372602508942,
-      "loss": 46.0,
-      "step": 46646
-    },
-    {
-      "epoch": 3.5664889041802854,
-      "grad_norm": 0.001166792819276452,
-      "learning_rate": 0.0001999937257560279,
-      "loss": 46.0,
-      "step": 46647
-    },
-    {
-      "epoch": 3.566565361163675,
-      "grad_norm": 0.001867242157459259,
-      "learning_rate": 0.0001999937254869606,
-      "loss": 46.0,
-      "step": 46648
-    },
-    {
-      "epoch": 3.566641818147065,
-      "grad_norm": 0.003704528324306011,
-      "learning_rate": 0.00019999372521788754,
-      "loss": 46.0,
-      "step": 46649
-    },
-    {
-      "epoch": 3.5667182751304547,
-      "grad_norm": 0.0034738376270979643,
-      "learning_rate": 0.00019999372494880873,
-      "loss": 46.0,
-      "step": 46650
-    },
-    {
-      "epoch": 3.5667947321138445,
-      "grad_norm": 0.003083515679463744,
-      "learning_rate": 0.00019999372467972411,
-      "loss": 46.0,
-      "step": 46651
-    },
-    {
-      "epoch": 3.566871189097234,
-      "grad_norm": 0.0011889126617461443,
-      "learning_rate": 0.00019999372441063375,
-      "loss": 46.0,
-      "step": 46652
-    },
-    {
-      "epoch": 3.566947646080624,
-      "grad_norm": 0.0027149440720677376,
-      "learning_rate": 0.00019999372414153762,
-      "loss": 46.0,
-      "step": 46653
-    },
-    {
-      "epoch": 3.5670241030640137,
-      "grad_norm": 0.0021190131083130836,
-      "learning_rate": 0.00019999372387243571,
-      "loss": 46.0,
-      "step": 46654
-    },
-    {
-      "epoch": 3.567100560047403,
-      "grad_norm": 0.003262369427829981,
-      "learning_rate": 0.000199993723603328,
-      "loss": 46.0,
-      "step": 46655
-    },
-    {
-      "epoch": 3.567177017030793,
-      "grad_norm": 0.002166547579690814,
-      "learning_rate": 0.00019999372333421458,
-      "loss": 46.0,
-      "step": 46656
-    },
-    {
-      "epoch": 3.5672534740141826,
-      "grad_norm": 0.001741804531775415,
-      "learning_rate": 0.00019999372306509535,
-      "loss": 46.0,
-      "step": 46657
-    },
-    {
-      "epoch": 3.5673299309975723,
-      "grad_norm": 0.005541711580008268,
-      "learning_rate": 0.00019999372279597038,
-      "loss": 46.0,
-      "step": 46658
-    },
-    {
-      "epoch": 3.567406387980962,
-      "grad_norm": 0.0021615559235215187,
-      "learning_rate": 0.00019999372252683966,
-      "loss": 46.0,
-      "step": 46659
-    },
-    {
-      "epoch": 3.567482844964352,
-      "grad_norm": 0.0025355208199471235,
-      "learning_rate": 0.00019999372225770312,
-      "loss": 46.0,
-      "step": 46660
-    },
-    {
-      "epoch": 3.5675593019477416,
-      "grad_norm": 0.002710143104195595,
-      "learning_rate": 0.00019999372198856082,
-      "loss": 46.0,
-      "step": 46661
-    },
-    {
-      "epoch": 3.5676357589311314,
-      "grad_norm": 0.0043915691785514355,
-      "learning_rate": 0.00019999372171941278,
-      "loss": 46.0,
-      "step": 46662
-    },
-    {
-      "epoch": 3.567712215914521,
-      "grad_norm": 0.002100900048390031,
-      "learning_rate": 0.00019999372145025897,
-      "loss": 46.0,
-      "step": 46663
-    },
-    {
-      "epoch": 3.567788672897911,
-      "grad_norm": 0.002226870507001877,
-      "learning_rate": 0.00019999372118109936,
-      "loss": 46.0,
-      "step": 46664
-    },
-    {
-      "epoch": 3.5678651298813007,
-      "grad_norm": 0.0007734823157079518,
-      "learning_rate": 0.00019999372091193397,
-      "loss": 46.0,
-      "step": 46665
-    },
-    {
-      "epoch": 3.5679415868646904,
-      "grad_norm": 0.0012925807386636734,
-      "learning_rate": 0.00019999372064276284,
-      "loss": 46.0,
-      "step": 46666
-    },
-    {
-      "epoch": 3.56801804384808,
-      "grad_norm": 0.0016663826536387205,
-      "learning_rate": 0.00019999372037358596,
-      "loss": 46.0,
-      "step": 46667
-    },
-    {
-      "epoch": 3.56809450083147,
-      "grad_norm": 0.0010066236136481166,
-      "learning_rate": 0.00019999372010440328,
-      "loss": 46.0,
-      "step": 46668
-    },
-    {
-      "epoch": 3.5681709578148593,
-      "grad_norm": 0.0017347463872283697,
-      "learning_rate": 0.00019999371983521483,
-      "loss": 46.0,
-      "step": 46669
-    },
-    {
-      "epoch": 3.568247414798249,
-      "grad_norm": 0.0034977858886122704,
-      "learning_rate": 0.00019999371956602063,
-      "loss": 46.0,
-      "step": 46670
-    },
-    {
-      "epoch": 3.568323871781639,
-      "grad_norm": 0.0008923871791921556,
-      "learning_rate": 0.00019999371929682065,
-      "loss": 46.0,
-      "step": 46671
-    },
-    {
-      "epoch": 3.5684003287650286,
-      "grad_norm": 0.0019429794047027826,
-      "learning_rate": 0.0001999937190276149,
-      "loss": 46.0,
-      "step": 46672
-    },
-    {
-      "epoch": 3.5684767857484183,
-      "grad_norm": 0.0010052897268906236,
-      "learning_rate": 0.0001999937187584034,
-      "loss": 46.0,
-      "step": 46673
-    },
-    {
-      "epoch": 3.568553242731808,
-      "grad_norm": 0.002584023866802454,
-      "learning_rate": 0.0001999937184891861,
-      "loss": 46.0,
-      "step": 46674
-    },
-    {
-      "epoch": 3.568629699715198,
-      "grad_norm": 0.0030269892886281013,
-      "learning_rate": 0.00019999371821996303,
-      "loss": 46.0,
-      "step": 46675
-    },
-    {
-      "epoch": 3.5687061566985876,
-      "grad_norm": 0.0016190746100619435,
-      "learning_rate": 0.00019999371795073422,
-      "loss": 46.0,
-      "step": 46676
-    },
-    {
-      "epoch": 3.568782613681977,
-      "grad_norm": 0.0026490408927202225,
-      "learning_rate": 0.0001999937176814996,
-      "loss": 46.0,
-      "step": 46677
-    },
-    {
-      "epoch": 3.5688590706653667,
-      "grad_norm": 0.0006975713768042624,
-      "learning_rate": 0.00019999371741225925,
-      "loss": 46.0,
-      "step": 46678
-    },
-    {
-      "epoch": 3.5689355276487564,
-      "grad_norm": 0.00419412087649107,
-      "learning_rate": 0.0001999937171430131,
-      "loss": 46.0,
-      "step": 46679
-    },
-    {
-      "epoch": 3.569011984632146,
-      "grad_norm": 0.002260678680613637,
-      "learning_rate": 0.0001999937168737612,
-      "loss": 46.0,
-      "step": 46680
-    },
-    {
-      "epoch": 3.569088441615536,
-      "grad_norm": 0.0012944111367687583,
-      "learning_rate": 0.00019999371660450355,
-      "loss": 46.0,
-      "step": 46681
-    },
-    {
-      "epoch": 3.5691648985989257,
-      "grad_norm": 0.0013006249209865928,
-      "learning_rate": 0.0001999937163352401,
-      "loss": 46.0,
-      "step": 46682
-    },
-    {
-      "epoch": 3.5692413555823155,
-      "grad_norm": 0.001816820353269577,
-      "learning_rate": 0.00019999371606597088,
-      "loss": 46.0,
-      "step": 46683
-    },
-    {
-      "epoch": 3.5693178125657052,
-      "grad_norm": 0.0019528622506186366,
-      "learning_rate": 0.00019999371579669593,
-      "loss": 46.0,
-      "step": 46684
-    },
-    {
-      "epoch": 3.569394269549095,
-      "grad_norm": 0.0036003452260047197,
-      "learning_rate": 0.00019999371552741516,
-      "loss": 46.0,
-      "step": 46685
-    },
-    {
-      "epoch": 3.5694707265324848,
-      "grad_norm": 0.0028399359434843063,
-      "learning_rate": 0.00019999371525812864,
-      "loss": 46.0,
-      "step": 46686
-    },
-    {
-      "epoch": 3.5695471835158745,
-      "grad_norm": 0.002477761823683977,
-      "learning_rate": 0.00019999371498883637,
-      "loss": 46.0,
-      "step": 46687
-    },
-    {
-      "epoch": 3.5696236404992643,
-      "grad_norm": 0.010253152810037136,
-      "learning_rate": 0.0001999937147195383,
-      "loss": 46.0,
-      "step": 46688
-    },
-    {
-      "epoch": 3.569700097482654,
-      "grad_norm": 0.0037822010926902294,
-      "learning_rate": 0.0001999937144502345,
-      "loss": 46.0,
-      "step": 46689
-    },
-    {
-      "epoch": 3.569776554466044,
-      "grad_norm": 0.0006942534237168729,
-      "learning_rate": 0.00019999371418092488,
-      "loss": 46.0,
-      "step": 46690
-    },
-    {
-      "epoch": 3.569853011449433,
-      "grad_norm": 0.0029970204923301935,
-      "learning_rate": 0.00019999371391160952,
-      "loss": 46.0,
-      "step": 46691
-    },
-    {
-      "epoch": 3.569929468432823,
-      "grad_norm": 0.0022718869149684906,
-      "learning_rate": 0.0001999937136422884,
-      "loss": 46.0,
-      "step": 46692
-    },
-    {
-      "epoch": 3.5700059254162126,
-      "grad_norm": 0.0026808385737240314,
-      "learning_rate": 0.0001999937133729615,
-      "loss": 46.0,
-      "step": 46693
-    },
-    {
-      "epoch": 3.5700823823996024,
-      "grad_norm": 0.004271664656698704,
-      "learning_rate": 0.00019999371310362883,
-      "loss": 46.0,
-      "step": 46694
-    },
-    {
-      "epoch": 3.570158839382992,
-      "grad_norm": 0.0015238302294164896,
-      "learning_rate": 0.0001999937128342904,
-      "loss": 46.0,
-      "step": 46695
-    },
-    {
-      "epoch": 3.570235296366382,
-      "grad_norm": 0.0011019359808415174,
-      "learning_rate": 0.00019999371256494618,
-      "loss": 46.0,
-      "step": 46696
-    },
-    {
-      "epoch": 3.5703117533497717,
-      "grad_norm": 0.0017429268918931484,
-      "learning_rate": 0.0001999937122955962,
-      "loss": 46.0,
-      "step": 46697
-    },
-    {
-      "epoch": 3.5703882103331614,
-      "grad_norm": 0.008877106010913849,
-      "learning_rate": 0.00019999371202624047,
-      "loss": 46.0,
-      "step": 46698
-    },
-    {
-      "epoch": 3.5704646673165508,
-      "grad_norm": 0.0010921615175902843,
-      "learning_rate": 0.00019999371175687895,
-      "loss": 46.0,
-      "step": 46699
-    },
-    {
-      "epoch": 3.5705411242999405,
-      "grad_norm": 0.0013169948942959309,
-      "learning_rate": 0.00019999371148751168,
-      "loss": 46.0,
-      "step": 46700
-    },
-    {
-      "epoch": 3.5706175812833303,
-      "grad_norm": 0.005135527346283197,
-      "learning_rate": 0.00019999371121813862,
-      "loss": 46.0,
-      "step": 46701
-    },
-    {
-      "epoch": 3.57069403826672,
-      "grad_norm": 0.003486905712634325,
-      "learning_rate": 0.00019999371094875978,
-      "loss": 46.0,
-      "step": 46702
-    },
-    {
-      "epoch": 3.57077049525011,
-      "grad_norm": 0.0026861154474318027,
-      "learning_rate": 0.0001999937106793752,
-      "loss": 46.0,
-      "step": 46703
-    },
-    {
-      "epoch": 3.5708469522334996,
-      "grad_norm": 0.0018732862081378698,
-      "learning_rate": 0.00019999371040998483,
-      "loss": 46.0,
-      "step": 46704
-    },
-    {
-      "epoch": 3.5709234092168893,
-      "grad_norm": 0.003529096022248268,
-      "learning_rate": 0.0001999937101405887,
-      "loss": 46.0,
-      "step": 46705
-    },
-    {
-      "epoch": 3.570999866200279,
-      "grad_norm": 0.0014549318002536893,
-      "learning_rate": 0.00019999370987118682,
-      "loss": 46.0,
-      "step": 46706
-    },
-    {
-      "epoch": 3.571076323183669,
-      "grad_norm": 0.001802139449864626,
-      "learning_rate": 0.00019999370960177915,
-      "loss": 46.0,
-      "step": 46707
-    },
-    {
-      "epoch": 3.5711527801670586,
-      "grad_norm": 0.0028477897867560387,
-      "learning_rate": 0.0001999937093323657,
-      "loss": 46.0,
-      "step": 46708
-    },
-    {
-      "epoch": 3.5712292371504484,
-      "grad_norm": 0.0017701752949506044,
-      "learning_rate": 0.0001999937090629465,
-      "loss": 46.0,
-      "step": 46709
-    },
-    {
-      "epoch": 3.571305694133838,
-      "grad_norm": 0.0031636760104447603,
-      "learning_rate": 0.00019999370879352152,
-      "loss": 46.0,
-      "step": 46710
-    },
-    {
-      "epoch": 3.571382151117228,
-      "grad_norm": 0.0011496504303067923,
-      "learning_rate": 0.00019999370852409078,
-      "loss": 46.0,
-      "step": 46711
-    },
-    {
-      "epoch": 3.5714586081006177,
-      "grad_norm": 0.0033891985658556223,
-      "learning_rate": 0.00019999370825465426,
-      "loss": 46.0,
-      "step": 46712
-    },
-    {
-      "epoch": 3.571535065084007,
-      "grad_norm": 0.006602306384593248,
-      "learning_rate": 0.00019999370798521197,
-      "loss": 46.0,
-      "step": 46713
-    },
-    {
-      "epoch": 3.5716115220673967,
-      "grad_norm": 0.005738503765314817,
-      "learning_rate": 0.00019999370771576396,
-      "loss": 46.0,
-      "step": 46714
-    },
-    {
-      "epoch": 3.5716879790507865,
-      "grad_norm": 0.002612931188195944,
-      "learning_rate": 0.00019999370744631012,
-      "loss": 46.0,
-      "step": 46715
-    },
-    {
-      "epoch": 3.5717644360341763,
-      "grad_norm": 0.00203883508220315,
-      "learning_rate": 0.00019999370717685053,
-      "loss": 46.0,
-      "step": 46716
-    },
-    {
-      "epoch": 3.571840893017566,
-      "grad_norm": 0.0027113761752843857,
-      "learning_rate": 0.00019999370690738518,
-      "loss": 46.0,
-      "step": 46717
-    },
-    {
-      "epoch": 3.571917350000956,
-      "grad_norm": 0.0010218023089691997,
-      "learning_rate": 0.00019999370663791407,
-      "loss": 46.0,
-      "step": 46718
-    },
-    {
-      "epoch": 3.5719938069843455,
-      "grad_norm": 0.0025397012941539288,
-      "learning_rate": 0.00019999370636843714,
-      "loss": 46.0,
-      "step": 46719
-    },
-    {
-      "epoch": 3.5720702639677353,
-      "grad_norm": 0.0016761726001277566,
-      "learning_rate": 0.00019999370609895446,
-      "loss": 46.0,
-      "step": 46720
-    },
-    {
-      "epoch": 3.5721467209511246,
-      "grad_norm": 0.0021245433017611504,
-      "learning_rate": 0.00019999370582946604,
-      "loss": 46.0,
-      "step": 46721
-    },
-    {
-      "epoch": 3.5722231779345144,
-      "grad_norm": 0.001341967610642314,
-      "learning_rate": 0.00019999370555997184,
-      "loss": 46.0,
-      "step": 46722
-    },
-    {
-      "epoch": 3.572299634917904,
-      "grad_norm": 0.0029063820838928223,
-      "learning_rate": 0.00019999370529047187,
-      "loss": 46.0,
-      "step": 46723
-    },
-    {
-      "epoch": 3.572376091901294,
-      "grad_norm": 0.0010640128748491406,
-      "learning_rate": 0.00019999370502096612,
-      "loss": 46.0,
-      "step": 46724
-    },
-    {
-      "epoch": 3.5724525488846837,
-      "grad_norm": 0.0018383142305538058,
-      "learning_rate": 0.00019999370475145463,
-      "loss": 46.0,
-      "step": 46725
-    },
-    {
-      "epoch": 3.5725290058680734,
-      "grad_norm": 0.003777771256864071,
-      "learning_rate": 0.00019999370448193734,
-      "loss": 46.0,
-      "step": 46726
-    },
-    {
-      "epoch": 3.572605462851463,
-      "grad_norm": 0.0013168500736355782,
-      "learning_rate": 0.0001999937042124143,
-      "loss": 46.0,
-      "step": 46727
-    },
-    {
-      "epoch": 3.572681919834853,
-      "grad_norm": 0.0019330760696902871,
-      "learning_rate": 0.00019999370394288546,
-      "loss": 46.0,
-      "step": 46728
-    },
-    {
-      "epoch": 3.5727583768182427,
-      "grad_norm": 0.0021763385739177465,
-      "learning_rate": 0.00019999370367335088,
-      "loss": 46.0,
-      "step": 46729
-    },
-    {
-      "epoch": 3.5728348338016325,
-      "grad_norm": 0.0011574061354622245,
-      "learning_rate": 0.00019999370340381055,
-      "loss": 46.0,
-      "step": 46730
-    },
-    {
-      "epoch": 3.5729112907850222,
-      "grad_norm": 0.0010531785665079951,
-      "learning_rate": 0.0001999937031342644,
-      "loss": 46.0,
-      "step": 46731
-    },
-    {
-      "epoch": 3.572987747768412,
-      "grad_norm": 0.003681953065097332,
-      "learning_rate": 0.0001999937028647125,
-      "loss": 46.0,
-      "step": 46732
-    },
-    {
-      "epoch": 3.5730642047518018,
-      "grad_norm": 0.00215995404869318,
-      "learning_rate": 0.00019999370259515483,
-      "loss": 46.0,
-      "step": 46733
-    },
-    {
-      "epoch": 3.5731406617351915,
-      "grad_norm": 0.0006553211132995784,
-      "learning_rate": 0.00019999370232559144,
-      "loss": 46.0,
-      "step": 46734
-    },
-    {
-      "epoch": 3.573217118718581,
-      "grad_norm": 0.004397883545607328,
-      "learning_rate": 0.0001999937020560222,
-      "loss": 46.0,
-      "step": 46735
-    },
-    {
-      "epoch": 3.5732935757019706,
-      "grad_norm": 0.0017515133367851377,
-      "learning_rate": 0.00019999370178644724,
-      "loss": 46.0,
-      "step": 46736
-    },
-    {
-      "epoch": 3.5733700326853604,
-      "grad_norm": 0.001964492490515113,
-      "learning_rate": 0.0001999937015168665,
-      "loss": 46.0,
-      "step": 46737
-    },
-    {
-      "epoch": 3.57344648966875,
-      "grad_norm": 0.003423060057684779,
-      "learning_rate": 0.00019999370124727998,
-      "loss": 46.0,
-      "step": 46738
-    },
-    {
-      "epoch": 3.57352294665214,
-      "grad_norm": 0.008906511589884758,
-      "learning_rate": 0.00019999370097768771,
-      "loss": 46.0,
-      "step": 46739
-    },
-    {
-      "epoch": 3.5735994036355296,
-      "grad_norm": 0.0013465622905641794,
-      "learning_rate": 0.00019999370070808965,
-      "loss": 46.0,
-      "step": 46740
-    },
-    {
-      "epoch": 3.5736758606189194,
-      "grad_norm": 0.007654804270714521,
-      "learning_rate": 0.00019999370043848584,
-      "loss": 46.0,
-      "step": 46741
-    },
-    {
-      "epoch": 3.573752317602309,
-      "grad_norm": 0.001357228378765285,
-      "learning_rate": 0.00019999370016887625,
-      "loss": 46.0,
-      "step": 46742
-    },
-    {
-      "epoch": 3.5738287745856985,
-      "grad_norm": 0.0018311721505597234,
-      "learning_rate": 0.00019999369989926092,
-      "loss": 46.0,
-      "step": 46743
-    },
-    {
-      "epoch": 3.5739052315690882,
-      "grad_norm": 0.004204512108117342,
-      "learning_rate": 0.0001999936996296398,
-      "loss": 46.0,
-      "step": 46744
-    },
-    {
-      "epoch": 3.573981688552478,
-      "grad_norm": 0.004237952642142773,
-      "learning_rate": 0.0001999936993600129,
-      "loss": 46.0,
-      "step": 46745
-    },
-    {
-      "epoch": 3.5740581455358678,
-      "grad_norm": 0.0031867539510130882,
-      "learning_rate": 0.0001999936990903802,
-      "loss": 46.0,
-      "step": 46746
-    },
-    {
-      "epoch": 3.5741346025192575,
-      "grad_norm": 0.0018509686924517155,
-      "learning_rate": 0.0001999936988207418,
-      "loss": 46.0,
-      "step": 46747
-    },
-    {
-      "epoch": 3.5742110595026473,
-      "grad_norm": 0.0027046131435781717,
-      "learning_rate": 0.0001999936985510976,
-      "loss": 46.0,
-      "step": 46748
-    },
-    {
-      "epoch": 3.574287516486037,
-      "grad_norm": 0.0024194049183279276,
-      "learning_rate": 0.00019999369828144762,
-      "loss": 46.0,
-      "step": 46749
-    },
-    {
-      "epoch": 3.574363973469427,
-      "grad_norm": 0.0024963058531284332,
-      "learning_rate": 0.00019999369801179187,
-      "loss": 46.0,
-      "step": 46750
-    },
-    {
-      "epoch": 3.5744404304528166,
-      "grad_norm": 0.0014862155076116323,
-      "learning_rate": 0.00019999369774213038,
-      "loss": 46.0,
-      "step": 46751
-    },
-    {
-      "epoch": 3.5745168874362063,
-      "grad_norm": 0.0010933363810181618,
-      "learning_rate": 0.0001999936974724631,
-      "loss": 46.0,
-      "step": 46752
-    },
-    {
-      "epoch": 3.574593344419596,
-      "grad_norm": 0.002765242476016283,
-      "learning_rate": 0.00019999369720279006,
-      "loss": 46.0,
-      "step": 46753
-    },
-    {
-      "epoch": 3.574669801402986,
-      "grad_norm": 0.0025579046923667192,
-      "learning_rate": 0.00019999369693311122,
-      "loss": 46.0,
-      "step": 46754
-    },
-    {
-      "epoch": 3.5747462583863756,
-      "grad_norm": 0.004957238677889109,
-      "learning_rate": 0.00019999369666342666,
-      "loss": 46.0,
-      "step": 46755
-    },
-    {
-      "epoch": 3.574822715369765,
-      "grad_norm": 0.00213652104139328,
-      "learning_rate": 0.0001999936963937363,
-      "loss": 46.0,
-      "step": 46756
-    },
-    {
-      "epoch": 3.5748991723531547,
-      "grad_norm": 0.005884305573999882,
-      "learning_rate": 0.0001999936961240402,
-      "loss": 46.0,
-      "step": 46757
-    },
-    {
-      "epoch": 3.5749756293365444,
-      "grad_norm": 0.0015873105730861425,
-      "learning_rate": 0.0001999936958543383,
-      "loss": 46.0,
-      "step": 46758
-    },
-    {
-      "epoch": 3.575052086319934,
-      "grad_norm": 0.0023645416367799044,
-      "learning_rate": 0.00019999369558463065,
-      "loss": 46.0,
-      "step": 46759
-    },
-    {
-      "epoch": 3.575128543303324,
-      "grad_norm": 0.0020588866900652647,
-      "learning_rate": 0.0001999936953149172,
-      "loss": 46.0,
-      "step": 46760
-    },
-    {
-      "epoch": 3.5752050002867137,
-      "grad_norm": 0.0026461975648999214,
-      "learning_rate": 0.000199993695045198,
-      "loss": 46.0,
-      "step": 46761
-    },
-    {
-      "epoch": 3.5752814572701035,
-      "grad_norm": 0.0023272226098924875,
-      "learning_rate": 0.00019999369477547303,
-      "loss": 46.0,
-      "step": 46762
-    },
-    {
-      "epoch": 3.5753579142534933,
-      "grad_norm": 0.0020812973380088806,
-      "learning_rate": 0.0001999936945057423,
-      "loss": 46.0,
-      "step": 46763
-    },
-    {
-      "epoch": 3.575434371236883,
-      "grad_norm": 0.0014014183543622494,
-      "learning_rate": 0.0001999936942360058,
-      "loss": 46.0,
-      "step": 46764
-    },
-    {
-      "epoch": 3.5755108282202723,
-      "grad_norm": 0.0008459005039185286,
-      "learning_rate": 0.00019999369396626353,
-      "loss": 46.0,
-      "step": 46765
-    },
-    {
-      "epoch": 3.575587285203662,
-      "grad_norm": 0.0014770374400541186,
-      "learning_rate": 0.0001999936936965155,
-      "loss": 46.0,
-      "step": 46766
-    },
-    {
-      "epoch": 3.575663742187052,
-      "grad_norm": 0.005236379336565733,
-      "learning_rate": 0.00019999369342676166,
-      "loss": 46.0,
-      "step": 46767
-    },
-    {
-      "epoch": 3.5757401991704416,
-      "grad_norm": 0.0006408869521692395,
-      "learning_rate": 0.00019999369315700207,
-      "loss": 46.0,
-      "step": 46768
-    },
-    {
-      "epoch": 3.5758166561538314,
-      "grad_norm": 0.006680544465780258,
-      "learning_rate": 0.00019999369288723674,
-      "loss": 46.0,
-      "step": 46769
-    },
-    {
-      "epoch": 3.575893113137221,
-      "grad_norm": 0.00097721500787884,
-      "learning_rate": 0.0001999936926174656,
-      "loss": 46.0,
-      "step": 46770
-    },
-    {
-      "epoch": 3.575969570120611,
-      "grad_norm": 0.0018070901278406382,
-      "learning_rate": 0.0001999936923476887,
-      "loss": 46.0,
-      "step": 46771
-    },
-    {
-      "epoch": 3.5760460271040007,
-      "grad_norm": 0.002206847770139575,
-      "learning_rate": 0.00019999369207790606,
-      "loss": 46.0,
-      "step": 46772
-    },
-    {
-      "epoch": 3.5761224840873904,
-      "grad_norm": 0.001448968192562461,
-      "learning_rate": 0.00019999369180811764,
-      "loss": 46.0,
-      "step": 46773
-    },
-    {
-      "epoch": 3.57619894107078,
-      "grad_norm": 0.0015066650230437517,
-      "learning_rate": 0.00019999369153832347,
-      "loss": 46.0,
-      "step": 46774
-    },
-    {
-      "epoch": 3.57627539805417,
-      "grad_norm": 0.00829948578029871,
-      "learning_rate": 0.00019999369126852347,
-      "loss": 46.0,
-      "step": 46775
-    },
-    {
-      "epoch": 3.5763518550375597,
-      "grad_norm": 0.002177748829126358,
-      "learning_rate": 0.00019999369099871775,
-      "loss": 46.0,
-      "step": 46776
-    },
-    {
-      "epoch": 3.5764283120209495,
-      "grad_norm": 0.0016441484913229942,
-      "learning_rate": 0.00019999369072890624,
-      "loss": 46.0,
-      "step": 46777
-    },
-    {
-      "epoch": 3.576504769004339,
-      "grad_norm": 0.002894565463066101,
-      "learning_rate": 0.00019999369045908897,
-      "loss": 46.0,
-      "step": 46778
-    },
-    {
-      "epoch": 3.5765812259877285,
-      "grad_norm": 0.001928357989527285,
-      "learning_rate": 0.00019999369018926594,
-      "loss": 46.0,
-      "step": 46779
-    },
-    {
-      "epoch": 3.5766576829711183,
-      "grad_norm": 0.0039041186682879925,
-      "learning_rate": 0.0001999936899194371,
-      "loss": 46.0,
-      "step": 46780
-    },
-    {
-      "epoch": 3.576734139954508,
-      "grad_norm": 0.0013528987765312195,
-      "learning_rate": 0.00019999368964960254,
-      "loss": 46.0,
-      "step": 46781
-    },
-    {
-      "epoch": 3.576810596937898,
-      "grad_norm": 0.008186491206288338,
-      "learning_rate": 0.00019999368937976216,
-      "loss": 46.0,
-      "step": 46782
-    },
-    {
-      "epoch": 3.5768870539212876,
-      "grad_norm": 0.004805978387594223,
-      "learning_rate": 0.00019999368910991608,
-      "loss": 46.0,
-      "step": 46783
-    },
-    {
-      "epoch": 3.5769635109046773,
-      "grad_norm": 0.004620899446308613,
-      "learning_rate": 0.00019999368884006418,
-      "loss": 46.0,
-      "step": 46784
-    },
-    {
-      "epoch": 3.577039967888067,
-      "grad_norm": 0.004319780506193638,
-      "learning_rate": 0.00019999368857020653,
-      "loss": 46.0,
-      "step": 46785
-    },
-    {
-      "epoch": 3.5771164248714564,
-      "grad_norm": 0.0009886305779218674,
-      "learning_rate": 0.00019999368830034308,
-      "loss": 46.0,
-      "step": 46786
-    },
-    {
-      "epoch": 3.577192881854846,
-      "grad_norm": 0.0007514212047681212,
-      "learning_rate": 0.00019999368803047388,
-      "loss": 46.0,
-      "step": 46787
-    },
-    {
-      "epoch": 3.577269338838236,
-      "grad_norm": 0.0017487631412222981,
-      "learning_rate": 0.00019999368776059891,
-      "loss": 46.0,
-      "step": 46788
-    },
-    {
-      "epoch": 3.5773457958216257,
-      "grad_norm": 0.0014503885759040713,
-      "learning_rate": 0.0001999936874907182,
-      "loss": 46.0,
-      "step": 46789
-    },
-    {
-      "epoch": 3.5774222528050155,
-      "grad_norm": 0.002369289519265294,
-      "learning_rate": 0.0001999936872208317,
-      "loss": 46.0,
-      "step": 46790
-    },
-    {
-      "epoch": 3.5774987097884052,
-      "grad_norm": 0.0029159081168472767,
-      "learning_rate": 0.00019999368695093942,
-      "loss": 46.0,
-      "step": 46791
-    },
-    {
-      "epoch": 3.577575166771795,
-      "grad_norm": 0.0027819473762065172,
-      "learning_rate": 0.00019999368668104138,
-      "loss": 46.0,
-      "step": 46792
-    },
-    {
-      "epoch": 3.5776516237551848,
-      "grad_norm": 0.0031795527320355177,
-      "learning_rate": 0.00019999368641113757,
-      "loss": 46.0,
-      "step": 46793
-    },
-    {
-      "epoch": 3.5777280807385745,
-      "grad_norm": 0.001487210625782609,
-      "learning_rate": 0.00019999368614122802,
-      "loss": 46.0,
-      "step": 46794
-    },
-    {
-      "epoch": 3.5778045377219643,
-      "grad_norm": 0.0009151780395768583,
-      "learning_rate": 0.00019999368587131266,
-      "loss": 46.0,
-      "step": 46795
-    },
-    {
-      "epoch": 3.577880994705354,
-      "grad_norm": 0.0014048231532797217,
-      "learning_rate": 0.00019999368560139153,
-      "loss": 46.0,
-      "step": 46796
-    },
-    {
-      "epoch": 3.577957451688744,
-      "grad_norm": 0.0009513715631328523,
-      "learning_rate": 0.00019999368533146465,
-      "loss": 46.0,
-      "step": 46797
-    },
-    {
-      "epoch": 3.5780339086721336,
-      "grad_norm": 0.0016650855541229248,
-      "learning_rate": 0.000199993685061532,
-      "loss": 46.0,
-      "step": 46798
-    },
-    {
-      "epoch": 3.5781103656555233,
-      "grad_norm": 0.0034799412824213505,
-      "learning_rate": 0.00019999368479159358,
-      "loss": 46.0,
-      "step": 46799
-    },
-    {
-      "epoch": 3.5781868226389126,
-      "grad_norm": 0.001487893401645124,
-      "learning_rate": 0.00019999368452164936,
-      "loss": 46.0,
-      "step": 46800
-    },
-    {
-      "epoch": 3.5782632796223024,
-      "grad_norm": 0.0018299927469342947,
-      "learning_rate": 0.00019999368425169942,
-      "loss": 46.0,
-      "step": 46801
-    },
-    {
-      "epoch": 3.578339736605692,
-      "grad_norm": 0.0008533790824003518,
-      "learning_rate": 0.0001999936839817437,
-      "loss": 46.0,
-      "step": 46802
-    },
-    {
-      "epoch": 3.578416193589082,
-      "grad_norm": 0.0035015440080314875,
-      "learning_rate": 0.00019999368371178218,
-      "loss": 46.0,
-      "step": 46803
-    },
-    {
-      "epoch": 3.5784926505724717,
-      "grad_norm": 0.0019301389111205935,
-      "learning_rate": 0.00019999368344181492,
-      "loss": 46.0,
-      "step": 46804
-    },
-    {
-      "epoch": 3.5785691075558614,
-      "grad_norm": 0.0011410403531044722,
-      "learning_rate": 0.00019999368317184189,
-      "loss": 46.0,
-      "step": 46805
-    },
-    {
-      "epoch": 3.578645564539251,
-      "grad_norm": 0.002320168074220419,
-      "learning_rate": 0.00019999368290186308,
-      "loss": 46.0,
-      "step": 46806
-    },
-    {
-      "epoch": 3.578722021522641,
-      "grad_norm": 0.004043637774884701,
-      "learning_rate": 0.0001999936826318785,
-      "loss": 46.0,
-      "step": 46807
-    },
-    {
-      "epoch": 3.5787984785060303,
-      "grad_norm": 0.004387584514915943,
-      "learning_rate": 0.00019999368236188817,
-      "loss": 46.0,
-      "step": 46808
-    },
-    {
-      "epoch": 3.57887493548942,
-      "grad_norm": 0.0011635865084826946,
-      "learning_rate": 0.00019999368209189204,
-      "loss": 46.0,
-      "step": 46809
-    },
-    {
-      "epoch": 3.57895139247281,
-      "grad_norm": 0.004327169619500637,
-      "learning_rate": 0.00019999368182189016,
-      "loss": 46.0,
-      "step": 46810
-    },
-    {
-      "epoch": 3.5790278494561996,
-      "grad_norm": 0.003773456672206521,
-      "learning_rate": 0.00019999368155188248,
-      "loss": 46.0,
-      "step": 46811
-    },
-    {
-      "epoch": 3.5791043064395893,
-      "grad_norm": 0.0008915658690966666,
-      "learning_rate": 0.0001999936812818691,
-      "loss": 46.0,
-      "step": 46812
-    },
-    {
-      "epoch": 3.579180763422979,
-      "grad_norm": 0.0006731427274644375,
-      "learning_rate": 0.0001999936810118499,
-      "loss": 46.0,
-      "step": 46813
-    },
-    {
-      "epoch": 3.579257220406369,
-      "grad_norm": 0.0016147417481988668,
-      "learning_rate": 0.00019999368074182492,
-      "loss": 46.0,
-      "step": 46814
-    },
-    {
-      "epoch": 3.5793336773897586,
-      "grad_norm": 0.0017773015424609184,
-      "learning_rate": 0.0001999936804717942,
-      "loss": 46.0,
-      "step": 46815
-    },
-    {
-      "epoch": 3.5794101343731484,
-      "grad_norm": 0.002532925456762314,
-      "learning_rate": 0.0001999936802017577,
-      "loss": 46.0,
-      "step": 46816
-    },
-    {
-      "epoch": 3.579486591356538,
-      "grad_norm": 0.0016357743879780173,
-      "learning_rate": 0.00019999367993171543,
-      "loss": 46.0,
-      "step": 46817
-    },
-    {
-      "epoch": 3.579563048339928,
-      "grad_norm": 0.0014051521429792047,
-      "learning_rate": 0.0001999936796616674,
-      "loss": 46.0,
-      "step": 46818
-    },
-    {
-      "epoch": 3.5796395053233176,
-      "grad_norm": 0.0008412550669163465,
-      "learning_rate": 0.00019999367939161359,
-      "loss": 46.0,
-      "step": 46819
-    },
-    {
-      "epoch": 3.5797159623067074,
-      "grad_norm": 0.0011157477274537086,
-      "learning_rate": 0.00019999367912155403,
-      "loss": 46.0,
-      "step": 46820
-    },
-    {
-      "epoch": 3.579792419290097,
-      "grad_norm": 0.0018081150483340025,
-      "learning_rate": 0.00019999367885148867,
-      "loss": 46.0,
-      "step": 46821
-    },
-    {
-      "epoch": 3.5798688762734865,
-      "grad_norm": 0.00424313684925437,
-      "learning_rate": 0.00019999367858141755,
-      "loss": 46.0,
-      "step": 46822
-    },
-    {
-      "epoch": 3.5799453332568762,
-      "grad_norm": 0.004288270603865385,
-      "learning_rate": 0.00019999367831134067,
-      "loss": 46.0,
-      "step": 46823
-    },
-    {
-      "epoch": 3.580021790240266,
-      "grad_norm": 0.0027884249575436115,
-      "learning_rate": 0.00019999367804125805,
-      "loss": 46.0,
-      "step": 46824
-    },
-    {
-      "epoch": 3.5800982472236558,
-      "grad_norm": 0.006068435963243246,
-      "learning_rate": 0.0001999936777711696,
-      "loss": 46.0,
-      "step": 46825
-    },
-    {
-      "epoch": 3.5801747042070455,
-      "grad_norm": 0.003302141558378935,
-      "learning_rate": 0.00019999367750107543,
-      "loss": 46.0,
-      "step": 46826
-    },
-    {
-      "epoch": 3.5802511611904353,
-      "grad_norm": 0.003297490766271949,
-      "learning_rate": 0.00019999367723097546,
-      "loss": 46.0,
-      "step": 46827
-    },
-    {
-      "epoch": 3.580327618173825,
-      "grad_norm": 0.003877032781019807,
-      "learning_rate": 0.00019999367696086972,
-      "loss": 46.0,
-      "step": 46828
-    },
-    {
-      "epoch": 3.580404075157215,
-      "grad_norm": 0.0012347853044047952,
-      "learning_rate": 0.00019999367669075823,
-      "loss": 46.0,
-      "step": 46829
-    },
-    {
-      "epoch": 3.580480532140604,
-      "grad_norm": 0.0023316224105656147,
-      "learning_rate": 0.00019999367642064097,
-      "loss": 46.0,
-      "step": 46830
-    },
-    {
-      "epoch": 3.580556989123994,
-      "grad_norm": 0.002378355246037245,
-      "learning_rate": 0.00019999367615051793,
-      "loss": 46.0,
-      "step": 46831
-    },
-    {
-      "epoch": 3.5806334461073837,
-      "grad_norm": 0.0013475813902914524,
-      "learning_rate": 0.00019999367588038913,
-      "loss": 46.0,
-      "step": 46832
-    },
-    {
-      "epoch": 3.5807099030907734,
-      "grad_norm": 0.004162842873483896,
-      "learning_rate": 0.00019999367561025454,
-      "loss": 46.0,
-      "step": 46833
-    },
-    {
-      "epoch": 3.580786360074163,
-      "grad_norm": 0.002723915968090296,
-      "learning_rate": 0.00019999367534011422,
-      "loss": 46.0,
-      "step": 46834
-    },
-    {
-      "epoch": 3.580862817057553,
-      "grad_norm": 0.005109093151986599,
-      "learning_rate": 0.00019999367506996811,
-      "loss": 46.0,
-      "step": 46835
-    },
-    {
-      "epoch": 3.5809392740409427,
-      "grad_norm": 0.0019570556469261646,
-      "learning_rate": 0.00019999367479981624,
-      "loss": 46.0,
-      "step": 46836
-    },
-    {
-      "epoch": 3.5810157310243325,
-      "grad_norm": 0.011387544684112072,
-      "learning_rate": 0.0001999936745296586,
-      "loss": 46.0,
-      "step": 46837
-    },
-    {
-      "epoch": 3.581092188007722,
-      "grad_norm": 0.002502802526578307,
-      "learning_rate": 0.00019999367425949517,
-      "loss": 46.0,
-      "step": 46838
-    },
-    {
-      "epoch": 3.581168644991112,
-      "grad_norm": 0.0032902935054153204,
-      "learning_rate": 0.00019999367398932597,
-      "loss": 46.0,
-      "step": 46839
-    },
-    {
-      "epoch": 3.5812451019745017,
-      "grad_norm": 0.001428350922651589,
-      "learning_rate": 0.000199993673719151,
-      "loss": 46.0,
-      "step": 46840
-    },
-    {
-      "epoch": 3.5813215589578915,
-      "grad_norm": 0.0008895163773559034,
-      "learning_rate": 0.0001999936734489703,
-      "loss": 46.0,
-      "step": 46841
-    },
-    {
-      "epoch": 3.5813980159412813,
-      "grad_norm": 0.00203204108402133,
-      "learning_rate": 0.0001999936731787838,
-      "loss": 46.0,
-      "step": 46842
-    },
-    {
-      "epoch": 3.581474472924671,
-      "grad_norm": 0.005746546201407909,
-      "learning_rate": 0.0001999936729085915,
-      "loss": 46.0,
-      "step": 46843
-    },
-    {
-      "epoch": 3.5815509299080603,
-      "grad_norm": 0.00214389874599874,
-      "learning_rate": 0.0001999936726383935,
-      "loss": 46.0,
-      "step": 46844
-    },
-    {
-      "epoch": 3.58162738689145,
-      "grad_norm": 0.0022019350435584784,
-      "learning_rate": 0.00019999367236818972,
-      "loss": 46.0,
-      "step": 46845
-    },
-    {
-      "epoch": 3.58170384387484,
-      "grad_norm": 0.0010690923081710935,
-      "learning_rate": 0.00019999367209798012,
-      "loss": 46.0,
-      "step": 46846
-    },
-    {
-      "epoch": 3.5817803008582296,
-      "grad_norm": 0.0031069230753928423,
-      "learning_rate": 0.00019999367182776481,
-      "loss": 46.0,
-      "step": 46847
-    },
-    {
-      "epoch": 3.5818567578416194,
-      "grad_norm": 0.0011770711280405521,
-      "learning_rate": 0.00019999367155754366,
-      "loss": 46.0,
-      "step": 46848
-    },
-    {
-      "epoch": 3.581933214825009,
-      "grad_norm": 0.00435941107571125,
-      "learning_rate": 0.0001999936712873168,
-      "loss": 46.0,
-      "step": 46849
-    },
-    {
-      "epoch": 3.582009671808399,
-      "grad_norm": 0.010944033041596413,
-      "learning_rate": 0.00019999367101708416,
-      "loss": 46.0,
-      "step": 46850
-    },
-    {
-      "epoch": 3.5820861287917887,
-      "grad_norm": 0.002510933205485344,
-      "learning_rate": 0.00019999367074684574,
-      "loss": 46.0,
-      "step": 46851
-    },
-    {
-      "epoch": 3.582162585775178,
-      "grad_norm": 0.002421025186777115,
-      "learning_rate": 0.00019999367047660155,
-      "loss": 46.0,
-      "step": 46852
-    },
-    {
-      "epoch": 3.5822390427585677,
-      "grad_norm": 0.00198395480401814,
-      "learning_rate": 0.0001999936702063516,
-      "loss": 46.0,
-      "step": 46853
-    },
-    {
-      "epoch": 3.5823154997419575,
-      "grad_norm": 0.002047466579824686,
-      "learning_rate": 0.00019999366993609586,
-      "loss": 46.0,
-      "step": 46854
-    },
-    {
-      "epoch": 3.5823919567253473,
-      "grad_norm": 0.001115086954087019,
-      "learning_rate": 0.0001999936696658344,
-      "loss": 46.0,
-      "step": 46855
-    },
-    {
-      "epoch": 3.582468413708737,
-      "grad_norm": 0.004388218745589256,
-      "learning_rate": 0.00019999366939556711,
-      "loss": 46.0,
-      "step": 46856
-    },
-    {
-      "epoch": 3.582544870692127,
-      "grad_norm": 0.00103157723788172,
-      "learning_rate": 0.00019999366912529408,
-      "loss": 46.0,
-      "step": 46857
-    },
-    {
-      "epoch": 3.5826213276755166,
-      "grad_norm": 0.002670151647180319,
-      "learning_rate": 0.0001999936688550153,
-      "loss": 46.0,
-      "step": 46858
-    },
-    {
-      "epoch": 3.5826977846589063,
-      "grad_norm": 0.002329086884856224,
-      "learning_rate": 0.00019999366858473072,
-      "loss": 46.0,
-      "step": 46859
-    },
-    {
-      "epoch": 3.582774241642296,
-      "grad_norm": 0.0017038799123838544,
-      "learning_rate": 0.00019999366831444036,
-      "loss": 46.0,
-      "step": 46860
-    },
-    {
-      "epoch": 3.582850698625686,
-      "grad_norm": 0.0020282629411667585,
-      "learning_rate": 0.0001999936680441443,
-      "loss": 46.0,
-      "step": 46861
-    },
-    {
-      "epoch": 3.5829271556090756,
-      "grad_norm": 0.0016655820654705167,
-      "learning_rate": 0.00019999366777384241,
-      "loss": 46.0,
-      "step": 46862
-    },
-    {
-      "epoch": 3.5830036125924654,
-      "grad_norm": 0.0009943764889612794,
-      "learning_rate": 0.00019999366750353477,
-      "loss": 46.0,
-      "step": 46863
-    },
-    {
-      "epoch": 3.583080069575855,
-      "grad_norm": 0.006375682540237904,
-      "learning_rate": 0.00019999366723322135,
-      "loss": 46.0,
-      "step": 46864
-    },
-    {
-      "epoch": 3.583156526559245,
-      "grad_norm": 0.0015145654324442148,
-      "learning_rate": 0.00019999366696290215,
-      "loss": 46.0,
-      "step": 46865
-    },
-    {
-      "epoch": 3.583232983542634,
-      "grad_norm": 0.0007737563573755324,
-      "learning_rate": 0.0001999936666925772,
-      "loss": 46.0,
-      "step": 46866
-    },
-    {
-      "epoch": 3.583309440526024,
-      "grad_norm": 0.0019043160136789083,
-      "learning_rate": 0.0001999936664222465,
-      "loss": 46.0,
-      "step": 46867
-    },
-    {
-      "epoch": 3.5833858975094137,
-      "grad_norm": 0.002586058573797345,
-      "learning_rate": 0.00019999366615191,
-      "loss": 46.0,
-      "step": 46868
-    },
-    {
-      "epoch": 3.5834623544928035,
-      "grad_norm": 0.003510398557409644,
-      "learning_rate": 0.00019999366588156772,
-      "loss": 46.0,
-      "step": 46869
-    },
-    {
-      "epoch": 3.5835388114761932,
-      "grad_norm": 0.002712306333705783,
-      "learning_rate": 0.0001999936656112197,
-      "loss": 46.0,
-      "step": 46870
-    },
-    {
-      "epoch": 3.583615268459583,
-      "grad_norm": 0.020472927019000053,
-      "learning_rate": 0.0001999936653408659,
-      "loss": 46.0,
-      "step": 46871
-    },
-    {
-      "epoch": 3.5836917254429728,
-      "grad_norm": 0.0013473426224663854,
-      "learning_rate": 0.00019999366507050635,
-      "loss": 46.0,
-      "step": 46872
-    },
-    {
-      "epoch": 3.5837681824263625,
-      "grad_norm": 0.0022178178187459707,
-      "learning_rate": 0.000199993664800141,
-      "loss": 46.0,
-      "step": 46873
-    },
-    {
-      "epoch": 3.583844639409752,
-      "grad_norm": 0.0014744015643373132,
-      "learning_rate": 0.0001999936645297699,
-      "loss": 46.0,
-      "step": 46874
-    },
-    {
-      "epoch": 3.5839210963931416,
-      "grad_norm": 0.0024269232526421547,
-      "learning_rate": 0.00019999366425939302,
-      "loss": 46.0,
-      "step": 46875
-    },
-    {
-      "epoch": 3.5839975533765314,
-      "grad_norm": 0.0010478358017280698,
-      "learning_rate": 0.00019999366398901038,
-      "loss": 46.0,
-      "step": 46876
-    },
-    {
-      "epoch": 3.584074010359921,
-      "grad_norm": 0.005806457716971636,
-      "learning_rate": 0.00019999366371862195,
-      "loss": 46.0,
-      "step": 46877
-    },
-    {
-      "epoch": 3.584150467343311,
-      "grad_norm": 0.001915874076075852,
-      "learning_rate": 0.0001999936634482278,
-      "loss": 46.0,
-      "step": 46878
-    },
-    {
-      "epoch": 3.5842269243267006,
-      "grad_norm": 0.0013806720962747931,
-      "learning_rate": 0.00019999366317782782,
-      "loss": 46.0,
-      "step": 46879
-    },
-    {
-      "epoch": 3.5843033813100904,
-      "grad_norm": 0.0012003335868939757,
-      "learning_rate": 0.0001999936629074221,
-      "loss": 46.0,
-      "step": 46880
-    },
-    {
-      "epoch": 3.58437983829348,
-      "grad_norm": 0.00372195802628994,
-      "learning_rate": 0.00019999366263701062,
-      "loss": 46.0,
-      "step": 46881
-    },
-    {
-      "epoch": 3.58445629527687,
-      "grad_norm": 0.001525621977634728,
-      "learning_rate": 0.00019999366236659336,
-      "loss": 46.0,
-      "step": 46882
-    },
-    {
-      "epoch": 3.5845327522602597,
-      "grad_norm": 0.0013389830710366368,
-      "learning_rate": 0.00019999366209617035,
-      "loss": 46.0,
-      "step": 46883
-    },
-    {
-      "epoch": 3.5846092092436495,
-      "grad_norm": 0.0019984717946499586,
-      "learning_rate": 0.00019999366182574154,
-      "loss": 46.0,
-      "step": 46884
-    },
-    {
-      "epoch": 3.584685666227039,
-      "grad_norm": 0.0016327903140336275,
-      "learning_rate": 0.000199993661555307,
-      "loss": 46.0,
-      "step": 46885
-    },
-    {
-      "epoch": 3.584762123210429,
-      "grad_norm": 0.00117179611697793,
-      "learning_rate": 0.00019999366128486666,
-      "loss": 46.0,
-      "step": 46886
-    },
-    {
-      "epoch": 3.5848385801938183,
-      "grad_norm": 0.0046615214087069035,
-      "learning_rate": 0.00019999366101442054,
-      "loss": 46.0,
-      "step": 46887
-    },
-    {
-      "epoch": 3.584915037177208,
-      "grad_norm": 0.0030638622120022774,
-      "learning_rate": 0.00019999366074396866,
-      "loss": 46.0,
-      "step": 46888
-    },
-    {
-      "epoch": 3.584991494160598,
-      "grad_norm": 0.0027536507695913315,
-      "learning_rate": 0.00019999366047351101,
-      "loss": 46.0,
-      "step": 46889
-    },
-    {
-      "epoch": 3.5850679511439876,
-      "grad_norm": 0.004430568311363459,
-      "learning_rate": 0.00019999366020304762,
-      "loss": 46.0,
-      "step": 46890
-    },
-    {
-      "epoch": 3.5851444081273773,
-      "grad_norm": 0.0024397619999945164,
-      "learning_rate": 0.00019999365993257843,
-      "loss": 46.0,
-      "step": 46891
-    },
-    {
-      "epoch": 3.585220865110767,
-      "grad_norm": 0.001831038505770266,
-      "learning_rate": 0.0001999936596621035,
-      "loss": 46.0,
-      "step": 46892
-    },
-    {
-      "epoch": 3.585297322094157,
-      "grad_norm": 0.0013811071403324604,
-      "learning_rate": 0.00019999365939162277,
-      "loss": 46.0,
-      "step": 46893
-    },
-    {
-      "epoch": 3.5853737790775466,
-      "grad_norm": 0.0010018610628321767,
-      "learning_rate": 0.00019999365912113629,
-      "loss": 46.0,
-      "step": 46894
-    },
-    {
-      "epoch": 3.5854502360609364,
-      "grad_norm": 0.0022364468313753605,
-      "learning_rate": 0.00019999365885064403,
-      "loss": 46.0,
-      "step": 46895
-    },
-    {
-      "epoch": 3.5855266930443257,
-      "grad_norm": 0.0029993706848472357,
-      "learning_rate": 0.000199993658580146,
-      "loss": 46.0,
-      "step": 46896
-    },
-    {
-      "epoch": 3.5856031500277155,
-      "grad_norm": 0.0018752289470285177,
-      "learning_rate": 0.0001999936583096422,
-      "loss": 46.0,
-      "step": 46897
-    },
-    {
-      "epoch": 3.585679607011105,
-      "grad_norm": 0.002106144791468978,
-      "learning_rate": 0.00019999365803913263,
-      "loss": 46.0,
-      "step": 46898
-    },
-    {
-      "epoch": 3.585756063994495,
-      "grad_norm": 0.002015571342781186,
-      "learning_rate": 0.0001999936577686173,
-      "loss": 46.0,
-      "step": 46899
-    },
-    {
-      "epoch": 3.5858325209778847,
-      "grad_norm": 0.002436260925605893,
-      "learning_rate": 0.0001999936574980962,
-      "loss": 46.0,
-      "step": 46900
-    },
-    {
-      "epoch": 3.5859089779612745,
-      "grad_norm": 0.006660154089331627,
-      "learning_rate": 0.00019999365722756933,
-      "loss": 46.0,
-      "step": 46901
-    },
-    {
-      "epoch": 3.5859854349446643,
-      "grad_norm": 0.0015384058933705091,
-      "learning_rate": 0.00019999365695703668,
-      "loss": 46.0,
-      "step": 46902
-    },
-    {
-      "epoch": 3.586061891928054,
-      "grad_norm": 0.001092072343453765,
-      "learning_rate": 0.0001999936566864983,
-      "loss": 46.0,
-      "step": 46903
-    },
-    {
-      "epoch": 3.586138348911444,
-      "grad_norm": 0.003085655625909567,
-      "learning_rate": 0.00019999365641595412,
-      "loss": 46.0,
-      "step": 46904
-    },
-    {
-      "epoch": 3.5862148058948335,
-      "grad_norm": 0.0013352144742384553,
-      "learning_rate": 0.00019999365614540416,
-      "loss": 46.0,
-      "step": 46905
-    },
-    {
-      "epoch": 3.5862912628782233,
-      "grad_norm": 0.005519264377653599,
-      "learning_rate": 0.00019999365587484844,
-      "loss": 46.0,
-      "step": 46906
-    },
-    {
-      "epoch": 3.586367719861613,
-      "grad_norm": 0.0007681608549319208,
-      "learning_rate": 0.00019999365560428696,
-      "loss": 46.0,
-      "step": 46907
-    },
-    {
-      "epoch": 3.586444176845003,
-      "grad_norm": 0.0011039824457839131,
-      "learning_rate": 0.0001999936553337197,
-      "loss": 46.0,
-      "step": 46908
-    },
-    {
-      "epoch": 3.586520633828392,
-      "grad_norm": 0.002910558832809329,
-      "learning_rate": 0.00019999365506314666,
-      "loss": 46.0,
-      "step": 46909
-    },
-    {
-      "epoch": 3.586597090811782,
-      "grad_norm": 0.0012377003440633416,
-      "learning_rate": 0.00019999365479256789,
-      "loss": 46.0,
-      "step": 46910
-    },
-    {
-      "epoch": 3.5866735477951717,
-      "grad_norm": 0.000693152891471982,
-      "learning_rate": 0.00019999365452198333,
-      "loss": 46.0,
-      "step": 46911
-    },
-    {
-      "epoch": 3.5867500047785614,
-      "grad_norm": 0.003015368478372693,
-      "learning_rate": 0.00019999365425139298,
-      "loss": 46.0,
-      "step": 46912
-    },
-    {
-      "epoch": 3.586826461761951,
-      "grad_norm": 0.0010261302813887596,
-      "learning_rate": 0.0001999936539807969,
-      "loss": 46.0,
-      "step": 46913
-    },
-    {
-      "epoch": 3.586902918745341,
-      "grad_norm": 0.0013995436020195484,
-      "learning_rate": 0.00019999365371019503,
-      "loss": 46.0,
-      "step": 46914
-    },
-    {
-      "epoch": 3.5869793757287307,
-      "grad_norm": 0.00220837933011353,
-      "learning_rate": 0.00019999365343958739,
-      "loss": 46.0,
-      "step": 46915
-    },
-    {
-      "epoch": 3.5870558327121205,
-      "grad_norm": 0.004423636943101883,
-      "learning_rate": 0.000199993653168974,
-      "loss": 46.0,
-      "step": 46916
-    },
-    {
-      "epoch": 3.58713228969551,
-      "grad_norm": 0.0034272857010364532,
-      "learning_rate": 0.0001999936528983548,
-      "loss": 46.0,
-      "step": 46917
-    },
-    {
-      "epoch": 3.5872087466788996,
-      "grad_norm": 0.0012460210127756,
-      "learning_rate": 0.00019999365262772986,
-      "loss": 46.0,
-      "step": 46918
-    },
-    {
-      "epoch": 3.5872852036622893,
-      "grad_norm": 0.002459676470607519,
-      "learning_rate": 0.00019999365235709915,
-      "loss": 46.0,
-      "step": 46919
-    },
-    {
-      "epoch": 3.587361660645679,
-      "grad_norm": 0.0017811853904277086,
-      "learning_rate": 0.00019999365208646266,
-      "loss": 46.0,
-      "step": 46920
-    },
-    {
-      "epoch": 3.587438117629069,
-      "grad_norm": 0.001715226098895073,
-      "learning_rate": 0.00019999365181582043,
-      "loss": 46.0,
-      "step": 46921
-    },
-    {
-      "epoch": 3.5875145746124586,
-      "grad_norm": 0.0006652458105236292,
-      "learning_rate": 0.0001999936515451724,
-      "loss": 46.0,
-      "step": 46922
-    },
-    {
-      "epoch": 3.5875910315958484,
-      "grad_norm": 0.002508498728275299,
-      "learning_rate": 0.0001999936512745186,
-      "loss": 46.0,
-      "step": 46923
-    },
-    {
-      "epoch": 3.587667488579238,
-      "grad_norm": 0.0034038997255265713,
-      "learning_rate": 0.00019999365100385906,
-      "loss": 46.0,
-      "step": 46924
-    },
-    {
-      "epoch": 3.587743945562628,
-      "grad_norm": 0.002733106492087245,
-      "learning_rate": 0.00019999365073319368,
-      "loss": 46.0,
-      "step": 46925
-    },
-    {
-      "epoch": 3.5878204025460176,
-      "grad_norm": 0.0014901467366144061,
-      "learning_rate": 0.0001999936504625226,
-      "loss": 46.0,
-      "step": 46926
-    },
-    {
-      "epoch": 3.5878968595294074,
-      "grad_norm": 0.004385835956782103,
-      "learning_rate": 0.00019999365019184574,
-      "loss": 46.0,
-      "step": 46927
-    },
-    {
-      "epoch": 3.587973316512797,
-      "grad_norm": 0.0018085431074723601,
-      "learning_rate": 0.00019999364992116312,
-      "loss": 46.0,
-      "step": 46928
-    },
-    {
-      "epoch": 3.588049773496187,
-      "grad_norm": 0.0017175941029563546,
-      "learning_rate": 0.0001999936496504747,
-      "loss": 46.0,
-      "step": 46929
-    },
-    {
-      "epoch": 3.5881262304795767,
-      "grad_norm": 0.0020792155992239714,
-      "learning_rate": 0.00019999364937978053,
-      "loss": 46.0,
-      "step": 46930
-    },
-    {
-      "epoch": 3.588202687462966,
-      "grad_norm": 0.0025319543201476336,
-      "learning_rate": 0.0001999936491090806,
-      "loss": 46.0,
-      "step": 46931
-    },
-    {
-      "epoch": 3.5882791444463558,
-      "grad_norm": 0.00318585941568017,
-      "learning_rate": 0.00019999364883837485,
-      "loss": 46.0,
-      "step": 46932
-    },
-    {
-      "epoch": 3.5883556014297455,
-      "grad_norm": 0.0013812975957989693,
-      "learning_rate": 0.0001999936485676634,
-      "loss": 46.0,
-      "step": 46933
-    },
-    {
-      "epoch": 3.5884320584131353,
-      "grad_norm": 0.0023615281097590923,
-      "learning_rate": 0.00019999364829694614,
-      "loss": 46.0,
-      "step": 46934
-    },
-    {
-      "epoch": 3.588508515396525,
-      "grad_norm": 0.0029017331544309855,
-      "learning_rate": 0.00019999364802622313,
-      "loss": 46.0,
-      "step": 46935
-    },
-    {
-      "epoch": 3.588584972379915,
-      "grad_norm": 0.0014864859404042363,
-      "learning_rate": 0.00019999364775549432,
-      "loss": 46.0,
-      "step": 46936
-    },
-    {
-      "epoch": 3.5886614293633046,
-      "grad_norm": 0.0017683276673778892,
-      "learning_rate": 0.00019999364748475977,
-      "loss": 46.0,
-      "step": 46937
-    },
-    {
-      "epoch": 3.5887378863466943,
-      "grad_norm": 0.0008812231244519353,
-      "learning_rate": 0.00019999364721401945,
-      "loss": 46.0,
-      "step": 46938
-    },
-    {
-      "epoch": 3.5888143433300836,
-      "grad_norm": 0.0031536284368485212,
-      "learning_rate": 0.00019999364694327335,
-      "loss": 46.0,
-      "step": 46939
-    },
-    {
-      "epoch": 3.5888908003134734,
-      "grad_norm": 0.0031987479887902737,
-      "learning_rate": 0.0001999936466725215,
-      "loss": 46.0,
-      "step": 46940
-    },
-    {
-      "epoch": 3.588967257296863,
-      "grad_norm": 0.0019306178437545896,
-      "learning_rate": 0.00019999364640176386,
-      "loss": 46.0,
-      "step": 46941
-    },
-    {
-      "epoch": 3.589043714280253,
-      "grad_norm": 0.00660546263679862,
-      "learning_rate": 0.00019999364613100044,
-      "loss": 46.0,
-      "step": 46942
-    },
-    {
-      "epoch": 3.5891201712636427,
-      "grad_norm": 0.000853507372085005,
-      "learning_rate": 0.00019999364586023127,
-      "loss": 46.0,
-      "step": 46943
-    },
-    {
-      "epoch": 3.5891966282470325,
-      "grad_norm": 0.0016048774123191833,
-      "learning_rate": 0.00019999364558945636,
-      "loss": 46.0,
-      "step": 46944
-    },
-    {
-      "epoch": 3.589273085230422,
-      "grad_norm": 0.0017520948313176632,
-      "learning_rate": 0.00019999364531867562,
-      "loss": 46.0,
-      "step": 46945
-    },
-    {
-      "epoch": 3.589349542213812,
-      "grad_norm": 0.0012636297615244985,
-      "learning_rate": 0.00019999364504788916,
-      "loss": 46.0,
-      "step": 46946
-    },
-    {
-      "epoch": 3.5894259991972017,
-      "grad_norm": 0.003234325908124447,
-      "learning_rate": 0.00019999364477709688,
-      "loss": 46.0,
-      "step": 46947
-    },
-    {
-      "epoch": 3.5895024561805915,
-      "grad_norm": 0.0017377709737047553,
-      "learning_rate": 0.00019999364450629887,
-      "loss": 46.0,
-      "step": 46948
-    },
-    {
-      "epoch": 3.5895789131639813,
-      "grad_norm": 0.0017416693735867739,
-      "learning_rate": 0.0001999936442354951,
-      "loss": 46.0,
-      "step": 46949
-    },
-    {
-      "epoch": 3.589655370147371,
-      "grad_norm": 0.0010475734015926719,
-      "learning_rate": 0.00019999364396468554,
-      "loss": 46.0,
-      "step": 46950
-    },
-    {
-      "epoch": 3.5897318271307608,
-      "grad_norm": 0.002614538185298443,
-      "learning_rate": 0.0001999936436938702,
-      "loss": 46.0,
-      "step": 46951
-    },
-    {
-      "epoch": 3.5898082841141505,
-      "grad_norm": 0.0032782768830657005,
-      "learning_rate": 0.00019999364342304911,
-      "loss": 46.0,
-      "step": 46952
-    },
-    {
-      "epoch": 3.58988474109754,
-      "grad_norm": 0.0019131769658997655,
-      "learning_rate": 0.00019999364315222224,
-      "loss": 46.0,
-      "step": 46953
-    },
-    {
-      "epoch": 3.5899611980809296,
-      "grad_norm": 0.0026985181029886007,
-      "learning_rate": 0.00019999364288138962,
-      "loss": 46.0,
-      "step": 46954
-    },
-    {
-      "epoch": 3.5900376550643194,
-      "grad_norm": 0.002289633033797145,
-      "learning_rate": 0.00019999364261055123,
-      "loss": 46.0,
-      "step": 46955
-    },
-    {
-      "epoch": 3.590114112047709,
-      "grad_norm": 0.001089455559849739,
-      "learning_rate": 0.00019999364233970704,
-      "loss": 46.0,
-      "step": 46956
-    },
-    {
-      "epoch": 3.590190569031099,
-      "grad_norm": 0.002395318355411291,
-      "learning_rate": 0.0001999936420688571,
-      "loss": 46.0,
-      "step": 46957
-    },
-    {
-      "epoch": 3.5902670260144887,
-      "grad_norm": 0.002379494486376643,
-      "learning_rate": 0.0001999936417980014,
-      "loss": 46.0,
-      "step": 46958
-    },
-    {
-      "epoch": 3.5903434829978784,
-      "grad_norm": 0.0036207493394613266,
-      "learning_rate": 0.0001999936415271399,
-      "loss": 46.0,
-      "step": 46959
-    },
-    {
-      "epoch": 3.590419939981268,
-      "grad_norm": 0.002129668602719903,
-      "learning_rate": 0.00019999364125627267,
-      "loss": 46.0,
-      "step": 46960
-    },
-    {
-      "epoch": 3.5904963969646575,
-      "grad_norm": 0.0025731255300343037,
-      "learning_rate": 0.00019999364098539967,
-      "loss": 46.0,
-      "step": 46961
-    },
-    {
-      "epoch": 3.5905728539480473,
-      "grad_norm": 0.002844603732228279,
-      "learning_rate": 0.00019999364071452086,
-      "loss": 46.0,
-      "step": 46962
-    },
-    {
-      "epoch": 3.590649310931437,
-      "grad_norm": 0.006743055302649736,
-      "learning_rate": 0.0001999936404436363,
-      "loss": 46.0,
-      "step": 46963
-    },
-    {
-      "epoch": 3.590725767914827,
-      "grad_norm": 0.0013882482890039682,
-      "learning_rate": 0.00019999364017274598,
-      "loss": 46.0,
-      "step": 46964
-    },
-    {
-      "epoch": 3.5908022248982165,
-      "grad_norm": 0.0032037049531936646,
-      "learning_rate": 0.0001999936399018499,
-      "loss": 46.0,
-      "step": 46965
-    },
-    {
-      "epoch": 3.5908786818816063,
-      "grad_norm": 0.003279771190136671,
-      "learning_rate": 0.00019999363963094804,
-      "loss": 46.0,
-      "step": 46966
-    },
-    {
-      "epoch": 3.590955138864996,
-      "grad_norm": 0.002363900886848569,
-      "learning_rate": 0.0001999936393600404,
-      "loss": 46.0,
-      "step": 46967
-    },
-    {
-      "epoch": 3.591031595848386,
-      "grad_norm": 0.005510049872100353,
-      "learning_rate": 0.000199993639089127,
-      "loss": 46.0,
-      "step": 46968
-    },
-    {
-      "epoch": 3.5911080528317756,
-      "grad_norm": 0.0017380713252350688,
-      "learning_rate": 0.00019999363881820784,
-      "loss": 46.0,
-      "step": 46969
-    },
-    {
-      "epoch": 3.5911845098151653,
-      "grad_norm": 0.002312297001481056,
-      "learning_rate": 0.0001999936385472829,
-      "loss": 46.0,
-      "step": 46970
-    },
-    {
-      "epoch": 3.591260966798555,
-      "grad_norm": 0.0026192215736955404,
-      "learning_rate": 0.0001999936382763522,
-      "loss": 46.0,
-      "step": 46971
-    },
-    {
-      "epoch": 3.591337423781945,
-      "grad_norm": 0.002573398407548666,
-      "learning_rate": 0.0001999936380054157,
-      "loss": 46.0,
-      "step": 46972
-    },
-    {
-      "epoch": 3.5914138807653346,
-      "grad_norm": 0.0014432446332648396,
-      "learning_rate": 0.00019999363773447347,
-      "loss": 46.0,
-      "step": 46973
-    },
-    {
-      "epoch": 3.5914903377487244,
-      "grad_norm": 0.0011821654625236988,
-      "learning_rate": 0.00019999363746352547,
-      "loss": 46.0,
-      "step": 46974
-    },
-    {
-      "epoch": 3.5915667947321137,
-      "grad_norm": 0.0016567215789109468,
-      "learning_rate": 0.0001999936371925717,
-      "loss": 46.0,
-      "step": 46975
-    },
-    {
-      "epoch": 3.5916432517155035,
-      "grad_norm": 0.0022444501519203186,
-      "learning_rate": 0.00019999363692161214,
-      "loss": 46.0,
-      "step": 46976
-    },
-    {
-      "epoch": 3.5917197086988932,
-      "grad_norm": 0.01023557037115097,
-      "learning_rate": 0.0001999936366506468,
-      "loss": 46.0,
-      "step": 46977
-    },
-    {
-      "epoch": 3.591796165682283,
-      "grad_norm": 0.005189172923564911,
-      "learning_rate": 0.00019999363637967572,
-      "loss": 46.0,
-      "step": 46978
-    },
-    {
-      "epoch": 3.5918726226656728,
-      "grad_norm": 0.00366743141785264,
-      "learning_rate": 0.00019999363610869884,
-      "loss": 46.0,
-      "step": 46979
-    },
-    {
-      "epoch": 3.5919490796490625,
-      "grad_norm": 0.002978543983772397,
-      "learning_rate": 0.00019999363583771623,
-      "loss": 46.0,
-      "step": 46980
-    },
-    {
-      "epoch": 3.5920255366324523,
-      "grad_norm": 0.002727055922150612,
-      "learning_rate": 0.00019999363556672784,
-      "loss": 46.0,
-      "step": 46981
-    },
-    {
-      "epoch": 3.592101993615842,
-      "grad_norm": 0.0022688647732138634,
-      "learning_rate": 0.00019999363529573367,
-      "loss": 46.0,
-      "step": 46982
-    },
-    {
-      "epoch": 3.5921784505992314,
-      "grad_norm": 0.0020660520531237125,
-      "learning_rate": 0.00019999363502473373,
-      "loss": 46.0,
-      "step": 46983
-    },
-    {
-      "epoch": 3.592254907582621,
-      "grad_norm": 0.0010479899356141686,
-      "learning_rate": 0.00019999363475372802,
-      "loss": 46.0,
-      "step": 46984
-    },
-    {
-      "epoch": 3.592331364566011,
-      "grad_norm": 0.001985242823138833,
-      "learning_rate": 0.00019999363448271656,
-      "loss": 46.0,
-      "step": 46985
-    },
-    {
-      "epoch": 3.5924078215494006,
-      "grad_norm": 0.0007140121888369322,
-      "learning_rate": 0.0001999936342116993,
-      "loss": 46.0,
-      "step": 46986
-    },
-    {
-      "epoch": 3.5924842785327904,
-      "grad_norm": 0.0016278420807793736,
-      "learning_rate": 0.00019999363394067627,
-      "loss": 46.0,
-      "step": 46987
-    },
-    {
-      "epoch": 3.59256073551618,
-      "grad_norm": 0.0057882145047187805,
-      "learning_rate": 0.00019999363366964752,
-      "loss": 46.0,
-      "step": 46988
-    },
-    {
-      "epoch": 3.59263719249957,
-      "grad_norm": 0.0009396523819305003,
-      "learning_rate": 0.00019999363339861295,
-      "loss": 46.0,
-      "step": 46989
-    },
-    {
-      "epoch": 3.5927136494829597,
-      "grad_norm": 0.0010874695144593716,
-      "learning_rate": 0.00019999363312757265,
-      "loss": 46.0,
-      "step": 46990
-    },
-    {
-      "epoch": 3.5927901064663494,
-      "grad_norm": 0.002160031348466873,
-      "learning_rate": 0.00019999363285652655,
-      "loss": 46.0,
-      "step": 46991
-    },
-    {
-      "epoch": 3.592866563449739,
-      "grad_norm": 0.002129316097125411,
-      "learning_rate": 0.00019999363258547468,
-      "loss": 46.0,
-      "step": 46992
-    },
-    {
-      "epoch": 3.592943020433129,
-      "grad_norm": 0.004814533982425928,
-      "learning_rate": 0.00019999363231441706,
-      "loss": 46.0,
-      "step": 46993
-    },
-    {
-      "epoch": 3.5930194774165187,
-      "grad_norm": 0.0008700086618773639,
-      "learning_rate": 0.00019999363204335367,
-      "loss": 46.0,
-      "step": 46994
-    },
-    {
-      "epoch": 3.5930959343999085,
-      "grad_norm": 0.0021089348010718822,
-      "learning_rate": 0.0001999936317722845,
-      "loss": 46.0,
-      "step": 46995
-    },
-    {
-      "epoch": 3.5931723913832982,
-      "grad_norm": 0.0011872408213093877,
-      "learning_rate": 0.00019999363150120957,
-      "loss": 46.0,
-      "step": 46996
-    },
-    {
-      "epoch": 3.5932488483666876,
-      "grad_norm": 0.001255916547961533,
-      "learning_rate": 0.00019999363123012886,
-      "loss": 46.0,
-      "step": 46997
-    },
-    {
-      "epoch": 3.5933253053500773,
-      "grad_norm": 0.002526640659198165,
-      "learning_rate": 0.0001999936309590424,
-      "loss": 46.0,
-      "step": 46998
-    },
-    {
-      "epoch": 3.593401762333467,
-      "grad_norm": 0.001285151462070644,
-      "learning_rate": 0.00019999363068795014,
-      "loss": 46.0,
-      "step": 46999
-    },
-    {
-      "epoch": 3.593478219316857,
-      "grad_norm": 0.002987871877849102,
-      "learning_rate": 0.00019999363041685214,
-      "loss": 46.0,
-      "step": 47000
-    },
-    {
-      "epoch": 3.5935546763002466,
-      "grad_norm": 0.0015190659323707223,
-      "learning_rate": 0.00019999363014574836,
-      "loss": 46.0,
-      "step": 47001
-    },
-    {
-      "epoch": 3.5936311332836364,
-      "grad_norm": 0.0019419463351368904,
-      "learning_rate": 0.0001999936298746388,
-      "loss": 46.0,
-      "step": 47002
-    },
-    {
-      "epoch": 3.593707590267026,
-      "grad_norm": 0.0015551402466371655,
-      "learning_rate": 0.0001999936296035235,
-      "loss": 46.0,
-      "step": 47003
-    },
-    {
-      "epoch": 3.593784047250416,
-      "grad_norm": 0.0018266671104356647,
-      "learning_rate": 0.00019999362933240242,
-      "loss": 46.0,
-      "step": 47004
-    },
-    {
-      "epoch": 3.593860504233805,
-      "grad_norm": 0.0014194740215316415,
-      "learning_rate": 0.00019999362906127557,
-      "loss": 46.0,
-      "step": 47005
-    },
-    {
-      "epoch": 3.593936961217195,
-      "grad_norm": 0.002629917347803712,
-      "learning_rate": 0.00019999362879014293,
-      "loss": 46.0,
-      "step": 47006
-    },
-    {
-      "epoch": 3.5940134182005847,
-      "grad_norm": 0.0026030107401311398,
-      "learning_rate": 0.00019999362851900454,
-      "loss": 46.0,
-      "step": 47007
-    },
-    {
-      "epoch": 3.5940898751839745,
-      "grad_norm": 0.00335359713062644,
-      "learning_rate": 0.00019999362824786037,
-      "loss": 46.0,
-      "step": 47008
-    },
-    {
-      "epoch": 3.5941663321673643,
-      "grad_norm": 0.0018676059553399682,
-      "learning_rate": 0.00019999362797671043,
-      "loss": 46.0,
-      "step": 47009
-    },
-    {
-      "epoch": 3.594242789150754,
-      "grad_norm": 0.003634584601968527,
-      "learning_rate": 0.00019999362770555475,
-      "loss": 46.0,
-      "step": 47010
-    },
-    {
-      "epoch": 3.5943192461341438,
-      "grad_norm": 0.0021456449758261442,
-      "learning_rate": 0.00019999362743439327,
-      "loss": 46.0,
-      "step": 47011
-    },
-    {
-      "epoch": 3.5943957031175335,
-      "grad_norm": 0.0020891083404421806,
-      "learning_rate": 0.00019999362716322604,
-      "loss": 46.0,
-      "step": 47012
-    },
-    {
-      "epoch": 3.5944721601009233,
-      "grad_norm": 0.0031798265408724546,
-      "learning_rate": 0.00019999362689205303,
-      "loss": 46.0,
-      "step": 47013
-    },
-    {
-      "epoch": 3.594548617084313,
-      "grad_norm": 0.003364681266248226,
-      "learning_rate": 0.00019999362662087423,
-      "loss": 46.0,
-      "step": 47014
-    },
-    {
-      "epoch": 3.594625074067703,
-      "grad_norm": 0.0013795975828543305,
-      "learning_rate": 0.0001999936263496897,
-      "loss": 46.0,
-      "step": 47015
-    },
-    {
-      "epoch": 3.5947015310510926,
-      "grad_norm": 0.001853947527706623,
-      "learning_rate": 0.00019999362607849935,
-      "loss": 46.0,
-      "step": 47016
-    },
-    {
-      "epoch": 3.5947779880344823,
-      "grad_norm": 0.001709077856503427,
-      "learning_rate": 0.00019999362580730328,
-      "loss": 46.0,
-      "step": 47017
-    },
-    {
-      "epoch": 3.5948544450178717,
-      "grad_norm": 0.004602494183927774,
-      "learning_rate": 0.00019999362553610144,
-      "loss": 46.0,
-      "step": 47018
-    },
-    {
-      "epoch": 3.5949309020012614,
-      "grad_norm": 0.0013291046489030123,
-      "learning_rate": 0.00019999362526489382,
-      "loss": 46.0,
-      "step": 47019
-    },
-    {
-      "epoch": 3.595007358984651,
-      "grad_norm": 0.005183434579521418,
-      "learning_rate": 0.0001999936249936804,
-      "loss": 46.0,
-      "step": 47020
-    },
-    {
-      "epoch": 3.595083815968041,
-      "grad_norm": 0.002122211502864957,
-      "learning_rate": 0.00019999362472246127,
-      "loss": 46.0,
-      "step": 47021
-    },
-    {
-      "epoch": 3.5951602729514307,
-      "grad_norm": 0.0031986648682504892,
-      "learning_rate": 0.00019999362445123633,
-      "loss": 46.0,
-      "step": 47022
-    },
-    {
-      "epoch": 3.5952367299348205,
-      "grad_norm": 0.0030911096837371588,
-      "learning_rate": 0.00019999362418000565,
-      "loss": 46.0,
-      "step": 47023
-    },
-    {
-      "epoch": 3.5953131869182102,
-      "grad_norm": 0.0030629821121692657,
-      "learning_rate": 0.00019999362390876917,
-      "loss": 46.0,
-      "step": 47024
-    },
-    {
-      "epoch": 3.5953896439016,
-      "grad_norm": 0.0014020544476807117,
-      "learning_rate": 0.0001999936236375269,
-      "loss": 46.0,
-      "step": 47025
-    },
-    {
-      "epoch": 3.5954661008849897,
-      "grad_norm": 0.015725765377283096,
-      "learning_rate": 0.0001999936233662789,
-      "loss": 46.0,
-      "step": 47026
-    },
-    {
-      "epoch": 3.595542557868379,
-      "grad_norm": 0.0037674284540116787,
-      "learning_rate": 0.00019999362309502513,
-      "loss": 46.0,
-      "step": 47027
-    },
-    {
-      "epoch": 3.595619014851769,
-      "grad_norm": 0.0011838682694360614,
-      "learning_rate": 0.0001999936228237656,
-      "loss": 46.0,
-      "step": 47028
-    },
-    {
-      "epoch": 3.5956954718351586,
-      "grad_norm": 0.0010207684244960546,
-      "learning_rate": 0.00019999362255250028,
-      "loss": 46.0,
-      "step": 47029
-    },
-    {
-      "epoch": 3.5957719288185483,
-      "grad_norm": 0.0028875903226435184,
-      "learning_rate": 0.0001999936222812292,
-      "loss": 46.0,
-      "step": 47030
-    },
-    {
-      "epoch": 3.595848385801938,
-      "grad_norm": 0.002142914105206728,
-      "learning_rate": 0.00019999362200995234,
-      "loss": 46.0,
-      "step": 47031
-    },
-    {
-      "epoch": 3.595924842785328,
-      "grad_norm": 0.0011943209683522582,
-      "learning_rate": 0.00019999362173866972,
-      "loss": 46.0,
-      "step": 47032
-    },
-    {
-      "epoch": 3.5960012997687176,
-      "grad_norm": 0.0009661823278293014,
-      "learning_rate": 0.00019999362146738133,
-      "loss": 46.0,
-      "step": 47033
-    },
-    {
-      "epoch": 3.5960777567521074,
-      "grad_norm": 0.002333032200112939,
-      "learning_rate": 0.00019999362119608717,
-      "loss": 46.0,
-      "step": 47034
-    },
-    {
-      "epoch": 3.596154213735497,
-      "grad_norm": 0.002394559793174267,
-      "learning_rate": 0.00019999362092478723,
-      "loss": 46.0,
-      "step": 47035
-    },
-    {
-      "epoch": 3.596230670718887,
-      "grad_norm": 0.0026340153999626637,
-      "learning_rate": 0.00019999362065348155,
-      "loss": 46.0,
-      "step": 47036
-    },
-    {
-      "epoch": 3.5963071277022767,
-      "grad_norm": 0.003347923280671239,
-      "learning_rate": 0.0001999936203821701,
-      "loss": 46.0,
-      "step": 47037
-    },
-    {
-      "epoch": 3.5963835846856664,
-      "grad_norm": 0.002986322855576873,
-      "learning_rate": 0.00019999362011085287,
-      "loss": 46.0,
-      "step": 47038
-    },
-    {
-      "epoch": 3.596460041669056,
-      "grad_norm": 0.0015189097030088305,
-      "learning_rate": 0.00019999361983952984,
-      "loss": 46.0,
-      "step": 47039
-    },
-    {
-      "epoch": 3.5965364986524455,
-      "grad_norm": 0.0010086576221510768,
-      "learning_rate": 0.00019999361956820106,
-      "loss": 46.0,
-      "step": 47040
-    },
-    {
-      "epoch": 3.5966129556358353,
-      "grad_norm": 0.0021347617730498314,
-      "learning_rate": 0.00019999361929686654,
-      "loss": 46.0,
-      "step": 47041
-    },
-    {
-      "epoch": 3.596689412619225,
-      "grad_norm": 0.0006580731715075672,
-      "learning_rate": 0.0001999936190255262,
-      "loss": 46.0,
-      "step": 47042
-    },
-    {
-      "epoch": 3.596765869602615,
-      "grad_norm": 0.002055598422884941,
-      "learning_rate": 0.00019999361875418012,
-      "loss": 46.0,
-      "step": 47043
-    },
-    {
-      "epoch": 3.5968423265860046,
-      "grad_norm": 0.004377503879368305,
-      "learning_rate": 0.0001999936184828283,
-      "loss": 46.0,
-      "step": 47044
-    },
-    {
-      "epoch": 3.5969187835693943,
-      "grad_norm": 0.0024345177225768566,
-      "learning_rate": 0.00019999361821147066,
-      "loss": 46.0,
-      "step": 47045
-    },
-    {
-      "epoch": 3.596995240552784,
-      "grad_norm": 0.0008339352207258344,
-      "learning_rate": 0.00019999361794010727,
-      "loss": 46.0,
-      "step": 47046
-    },
-    {
-      "epoch": 3.597071697536174,
-      "grad_norm": 0.007169728167355061,
-      "learning_rate": 0.00019999361766873813,
-      "loss": 46.0,
-      "step": 47047
-    },
-    {
-      "epoch": 3.597148154519563,
-      "grad_norm": 0.0028401638846844435,
-      "learning_rate": 0.0001999936173973632,
-      "loss": 46.0,
-      "step": 47048
-    },
-    {
-      "epoch": 3.597224611502953,
-      "grad_norm": 0.00604660389944911,
-      "learning_rate": 0.0001999936171259825,
-      "loss": 46.0,
-      "step": 47049
-    },
-    {
-      "epoch": 3.5973010684863427,
-      "grad_norm": 0.0017167206387966871,
-      "learning_rate": 0.00019999361685459603,
-      "loss": 46.0,
-      "step": 47050
-    },
-    {
-      "epoch": 3.5973775254697324,
-      "grad_norm": 0.004143692087382078,
-      "learning_rate": 0.0001999936165832038,
-      "loss": 46.0,
-      "step": 47051
-    },
-    {
-      "epoch": 3.597453982453122,
-      "grad_norm": 0.0026565559674054384,
-      "learning_rate": 0.0001999936163118058,
-      "loss": 46.0,
-      "step": 47052
-    },
-    {
-      "epoch": 3.597530439436512,
-      "grad_norm": 0.0006196071044541895,
-      "learning_rate": 0.00019999361604040202,
-      "loss": 46.0,
-      "step": 47053
-    },
-    {
-      "epoch": 3.5976068964199017,
-      "grad_norm": 0.0013243438443168998,
-      "learning_rate": 0.00019999361576899247,
-      "loss": 46.0,
-      "step": 47054
-    },
-    {
-      "epoch": 3.5976833534032915,
-      "grad_norm": 0.0015893429517745972,
-      "learning_rate": 0.00019999361549757718,
-      "loss": 46.0,
-      "step": 47055
-    },
-    {
-      "epoch": 3.5977598103866812,
-      "grad_norm": 0.0025597589556127787,
-      "learning_rate": 0.0001999936152261561,
-      "loss": 46.0,
-      "step": 47056
-    },
-    {
-      "epoch": 3.597836267370071,
-      "grad_norm": 0.0022576586343348026,
-      "learning_rate": 0.00019999361495472924,
-      "loss": 46.0,
-      "step": 47057
-    },
-    {
-      "epoch": 3.5979127243534608,
-      "grad_norm": 0.0008624986512586474,
-      "learning_rate": 0.00019999361468329662,
-      "loss": 46.0,
-      "step": 47058
-    },
-    {
-      "epoch": 3.5979891813368505,
-      "grad_norm": 0.001514685107395053,
-      "learning_rate": 0.00019999361441185823,
-      "loss": 46.0,
-      "step": 47059
-    },
-    {
-      "epoch": 3.5980656383202403,
-      "grad_norm": 0.0030874779913574457,
-      "learning_rate": 0.0001999936141404141,
-      "loss": 46.0,
-      "step": 47060
-    },
-    {
-      "epoch": 3.59814209530363,
-      "grad_norm": 0.004635296296328306,
-      "learning_rate": 0.00019999361386896416,
-      "loss": 46.0,
-      "step": 47061
-    },
-    {
-      "epoch": 3.5982185522870194,
-      "grad_norm": 0.0020104851573705673,
-      "learning_rate": 0.00019999361359750845,
-      "loss": 46.0,
-      "step": 47062
-    },
-    {
-      "epoch": 3.598295009270409,
-      "grad_norm": 0.0032096034847199917,
-      "learning_rate": 0.000199993613326047,
-      "loss": 46.0,
-      "step": 47063
-    },
-    {
-      "epoch": 3.598371466253799,
-      "grad_norm": 0.0016590581508353353,
-      "learning_rate": 0.00019999361305457977,
-      "loss": 46.0,
-      "step": 47064
-    },
-    {
-      "epoch": 3.5984479232371887,
-      "grad_norm": 0.0018877660622820258,
-      "learning_rate": 0.0001999936127831068,
-      "loss": 46.0,
-      "step": 47065
-    },
-    {
-      "epoch": 3.5985243802205784,
-      "grad_norm": 0.0034017807338386774,
-      "learning_rate": 0.00019999361251162801,
-      "loss": 46.0,
-      "step": 47066
-    },
-    {
-      "epoch": 3.598600837203968,
-      "grad_norm": 0.0011026897700503469,
-      "learning_rate": 0.0001999936122401435,
-      "loss": 46.0,
-      "step": 47067
-    },
-    {
-      "epoch": 3.598677294187358,
-      "grad_norm": 0.0019213076448068023,
-      "learning_rate": 0.00019999361196865317,
-      "loss": 46.0,
-      "step": 47068
-    },
-    {
-      "epoch": 3.5987537511707477,
-      "grad_norm": 0.0013271161587908864,
-      "learning_rate": 0.00019999361169715707,
-      "loss": 46.0,
-      "step": 47069
-    },
-    {
-      "epoch": 3.598830208154137,
-      "grad_norm": 0.001317350659519434,
-      "learning_rate": 0.00019999361142565523,
-      "loss": 46.0,
-      "step": 47070
-    },
-    {
-      "epoch": 3.5989066651375268,
-      "grad_norm": 0.003212464740499854,
-      "learning_rate": 0.00019999361115414762,
-      "loss": 46.0,
-      "step": 47071
-    },
-    {
-      "epoch": 3.5989831221209165,
-      "grad_norm": 0.0014197358395904303,
-      "learning_rate": 0.00019999361088263423,
-      "loss": 46.0,
-      "step": 47072
-    },
-    {
-      "epoch": 3.5990595791043063,
-      "grad_norm": 0.0031482691410928965,
-      "learning_rate": 0.0001999936106111151,
-      "loss": 46.0,
-      "step": 47073
-    },
-    {
-      "epoch": 3.599136036087696,
-      "grad_norm": 0.0017301241168752313,
-      "learning_rate": 0.00019999361033959016,
-      "loss": 46.0,
-      "step": 47074
-    },
-    {
-      "epoch": 3.599212493071086,
-      "grad_norm": 0.0020927705336362123,
-      "learning_rate": 0.00019999361006805945,
-      "loss": 46.0,
-      "step": 47075
-    },
-    {
-      "epoch": 3.5992889500544756,
-      "grad_norm": 0.0015665240352973342,
-      "learning_rate": 0.00019999360979652302,
-      "loss": 46.0,
-      "step": 47076
-    },
-    {
-      "epoch": 3.5993654070378653,
-      "grad_norm": 0.005547032225877047,
-      "learning_rate": 0.00019999360952498076,
-      "loss": 46.0,
-      "step": 47077
-    },
-    {
-      "epoch": 3.599441864021255,
-      "grad_norm": 0.0004632631316781044,
-      "learning_rate": 0.0001999936092534328,
-      "loss": 46.0,
-      "step": 47078
-    },
-    {
-      "epoch": 3.599518321004645,
-      "grad_norm": 0.001670603291131556,
-      "learning_rate": 0.000199993608981879,
-      "loss": 46.0,
-      "step": 47079
-    },
-    {
-      "epoch": 3.5995947779880346,
-      "grad_norm": 0.0020996350795030594,
-      "learning_rate": 0.00019999360871031946,
-      "loss": 46.0,
-      "step": 47080
-    },
-    {
-      "epoch": 3.5996712349714244,
-      "grad_norm": 0.0014863634714856744,
-      "learning_rate": 0.00019999360843875417,
-      "loss": 46.0,
-      "step": 47081
-    },
-    {
-      "epoch": 3.599747691954814,
-      "grad_norm": 0.0017067798180505633,
-      "learning_rate": 0.0001999936081671831,
-      "loss": 46.0,
-      "step": 47082
-    },
-    {
-      "epoch": 3.599824148938204,
-      "grad_norm": 0.0020308925304561853,
-      "learning_rate": 0.00019999360789560623,
-      "loss": 46.0,
-      "step": 47083
-    },
-    {
-      "epoch": 3.599900605921593,
-      "grad_norm": 0.0007679708651266992,
-      "learning_rate": 0.00019999360762402364,
-      "loss": 46.0,
-      "step": 47084
-    },
-    {
-      "epoch": 3.599977062904983,
-      "grad_norm": 0.0025735548697412014,
-      "learning_rate": 0.00019999360735243525,
-      "loss": 46.0,
-      "step": 47085
-    },
-    {
-      "epoch": 3.6000535198883727,
-      "grad_norm": 0.0012532372493296862,
-      "learning_rate": 0.0001999936070808411,
-      "loss": 46.0,
-      "step": 47086
-    },
-    {
-      "epoch": 3.6001299768717625,
-      "grad_norm": 0.0047006201930344105,
-      "learning_rate": 0.00019999360680924118,
-      "loss": 46.0,
-      "step": 47087
-    },
-    {
-      "epoch": 3.6002064338551523,
-      "grad_norm": 0.0008170231594704092,
-      "learning_rate": 0.00019999360653763547,
-      "loss": 46.0,
-      "step": 47088
-    },
-    {
-      "epoch": 3.600282890838542,
-      "grad_norm": 0.0015357394004240632,
-      "learning_rate": 0.00019999360626602402,
-      "loss": 46.0,
-      "step": 47089
-    },
-    {
-      "epoch": 3.600359347821932,
-      "grad_norm": 0.0016586343990638852,
-      "learning_rate": 0.0001999936059944068,
-      "loss": 46.0,
-      "step": 47090
-    },
-    {
-      "epoch": 3.6004358048053215,
-      "grad_norm": 0.0013088248670101166,
-      "learning_rate": 0.00019999360572278381,
-      "loss": 46.0,
-      "step": 47091
-    },
-    {
-      "epoch": 3.600512261788711,
-      "grad_norm": 0.0011919861426576972,
-      "learning_rate": 0.00019999360545115504,
-      "loss": 46.0,
-      "step": 47092
-    },
-    {
-      "epoch": 3.6005887187721006,
-      "grad_norm": 0.00356575520709157,
-      "learning_rate": 0.00019999360517952052,
-      "loss": 46.0,
-      "step": 47093
-    },
-    {
-      "epoch": 3.6006651757554904,
-      "grad_norm": 0.0015427334001287818,
-      "learning_rate": 0.0001999936049078802,
-      "loss": 46.0,
-      "step": 47094
-    },
-    {
-      "epoch": 3.60074163273888,
-      "grad_norm": 0.0019467585952952504,
-      "learning_rate": 0.0001999936046362341,
-      "loss": 46.0,
-      "step": 47095
-    },
-    {
-      "epoch": 3.60081808972227,
-      "grad_norm": 0.0032476249616593122,
-      "learning_rate": 0.00019999360436458229,
-      "loss": 46.0,
-      "step": 47096
-    },
-    {
-      "epoch": 3.6008945467056597,
-      "grad_norm": 0.0006188943516463041,
-      "learning_rate": 0.00019999360409292467,
-      "loss": 46.0,
-      "step": 47097
-    },
-    {
-      "epoch": 3.6009710036890494,
-      "grad_norm": 0.0015529319643974304,
-      "learning_rate": 0.00019999360382126126,
-      "loss": 46.0,
-      "step": 47098
-    },
-    {
-      "epoch": 3.601047460672439,
-      "grad_norm": 0.0014996129320934415,
-      "learning_rate": 0.00019999360354959215,
-      "loss": 46.0,
-      "step": 47099
-    },
-    {
-      "epoch": 3.601123917655829,
-      "grad_norm": 0.002248901640996337,
-      "learning_rate": 0.0001999936032779172,
-      "loss": 46.0,
-      "step": 47100
-    },
-    {
-      "epoch": 3.6012003746392187,
-      "grad_norm": 0.002364114858210087,
-      "learning_rate": 0.00019999360300623653,
-      "loss": 46.0,
-      "step": 47101
-    },
-    {
-      "epoch": 3.6012768316226085,
-      "grad_norm": 0.0019309918861836195,
-      "learning_rate": 0.00019999360273455005,
-      "loss": 46.0,
-      "step": 47102
-    },
-    {
-      "epoch": 3.6013532886059982,
-      "grad_norm": 0.003945961594581604,
-      "learning_rate": 0.00019999360246285782,
-      "loss": 46.0,
-      "step": 47103
-    },
-    {
-      "epoch": 3.601429745589388,
-      "grad_norm": 0.004926863592118025,
-      "learning_rate": 0.00019999360219115985,
-      "loss": 46.0,
-      "step": 47104
-    },
-    {
-      "epoch": 3.6015062025727778,
-      "grad_norm": 0.0006436780677177012,
-      "learning_rate": 0.0001999936019194561,
-      "loss": 46.0,
-      "step": 47105
-    },
-    {
-      "epoch": 3.601582659556167,
-      "grad_norm": 0.0018002215074375272,
-      "learning_rate": 0.00019999360164774658,
-      "loss": 46.0,
-      "step": 47106
-    },
-    {
-      "epoch": 3.601659116539557,
-      "grad_norm": 0.0005858491058461368,
-      "learning_rate": 0.00019999360137603126,
-      "loss": 46.0,
-      "step": 47107
-    },
-    {
-      "epoch": 3.6017355735229466,
-      "grad_norm": 0.0025122312363237143,
-      "learning_rate": 0.00019999360110431016,
-      "loss": 46.0,
-      "step": 47108
-    },
-    {
-      "epoch": 3.6018120305063364,
-      "grad_norm": 0.0017562670400366187,
-      "learning_rate": 0.00019999360083258335,
-      "loss": 46.0,
-      "step": 47109
-    },
-    {
-      "epoch": 3.601888487489726,
-      "grad_norm": 0.0018919516587629914,
-      "learning_rate": 0.00019999360056085073,
-      "loss": 46.0,
-      "step": 47110
-    },
-    {
-      "epoch": 3.601964944473116,
-      "grad_norm": 0.0019179179798811674,
-      "learning_rate": 0.00019999360028911234,
-      "loss": 46.0,
-      "step": 47111
-    },
-    {
-      "epoch": 3.6020414014565056,
-      "grad_norm": 0.002963228849694133,
-      "learning_rate": 0.0001999936000173682,
-      "loss": 46.0,
-      "step": 47112
-    },
-    {
-      "epoch": 3.6021178584398954,
-      "grad_norm": 0.0022020824253559113,
-      "learning_rate": 0.00019999359974561827,
-      "loss": 46.0,
-      "step": 47113
-    },
-    {
-      "epoch": 3.6021943154232847,
-      "grad_norm": 0.004007265437394381,
-      "learning_rate": 0.0001999935994738626,
-      "loss": 46.0,
-      "step": 47114
-    },
-    {
-      "epoch": 3.6022707724066745,
-      "grad_norm": 0.0022375090047717094,
-      "learning_rate": 0.00019999359920210114,
-      "loss": 46.0,
-      "step": 47115
-    },
-    {
-      "epoch": 3.6023472293900642,
-      "grad_norm": 0.0020432686433196068,
-      "learning_rate": 0.0001999935989303339,
-      "loss": 46.0,
-      "step": 47116
-    },
-    {
-      "epoch": 3.602423686373454,
-      "grad_norm": 0.006760394666343927,
-      "learning_rate": 0.0001999935986585609,
-      "loss": 46.0,
-      "step": 47117
-    },
-    {
-      "epoch": 3.6025001433568438,
-      "grad_norm": 0.0006727928994223475,
-      "learning_rate": 0.00019999359838678216,
-      "loss": 46.0,
-      "step": 47118
-    },
-    {
-      "epoch": 3.6025766003402335,
-      "grad_norm": 0.0009790933690965176,
-      "learning_rate": 0.00019999359811499761,
-      "loss": 46.0,
-      "step": 47119
-    },
-    {
-      "epoch": 3.6026530573236233,
-      "grad_norm": 0.001795065589249134,
-      "learning_rate": 0.00019999359784320732,
-      "loss": 46.0,
-      "step": 47120
-    },
-    {
-      "epoch": 3.602729514307013,
-      "grad_norm": 0.002396184951066971,
-      "learning_rate": 0.00019999359757141125,
-      "loss": 46.0,
-      "step": 47121
-    },
-    {
-      "epoch": 3.602805971290403,
-      "grad_norm": 0.004131458234041929,
-      "learning_rate": 0.00019999359729960944,
-      "loss": 46.0,
-      "step": 47122
-    },
-    {
-      "epoch": 3.6028824282737926,
-      "grad_norm": 0.001958385808393359,
-      "learning_rate": 0.0001999935970278018,
-      "loss": 46.0,
-      "step": 47123
-    },
-    {
-      "epoch": 3.6029588852571823,
-      "grad_norm": 0.0031765205785632133,
-      "learning_rate": 0.00019999359675598844,
-      "loss": 46.0,
-      "step": 47124
-    },
-    {
-      "epoch": 3.603035342240572,
-      "grad_norm": 0.0028282124549150467,
-      "learning_rate": 0.00019999359648416928,
-      "loss": 46.0,
-      "step": 47125
-    },
-    {
-      "epoch": 3.603111799223962,
-      "grad_norm": 0.0030555317644029856,
-      "learning_rate": 0.00019999359621234437,
-      "loss": 46.0,
-      "step": 47126
-    },
-    {
-      "epoch": 3.6031882562073516,
-      "grad_norm": 0.0013310738140717149,
-      "learning_rate": 0.0001999935959405137,
-      "loss": 46.0,
-      "step": 47127
-    },
-    {
-      "epoch": 3.603264713190741,
-      "grad_norm": 0.002595169236883521,
-      "learning_rate": 0.00019999359566867723,
-      "loss": 46.0,
-      "step": 47128
-    },
-    {
-      "epoch": 3.6033411701741307,
-      "grad_norm": 0.002087588654831052,
-      "learning_rate": 0.000199993595396835,
-      "loss": 46.0,
-      "step": 47129
-    },
-    {
-      "epoch": 3.6034176271575205,
-      "grad_norm": 0.00139293831307441,
-      "learning_rate": 0.00019999359512498703,
-      "loss": 46.0,
-      "step": 47130
-    },
-    {
-      "epoch": 3.60349408414091,
-      "grad_norm": 0.0032348160166293383,
-      "learning_rate": 0.00019999359485313326,
-      "loss": 46.0,
-      "step": 47131
-    },
-    {
-      "epoch": 3.6035705411243,
-      "grad_norm": 0.001367839053273201,
-      "learning_rate": 0.00019999359458127374,
-      "loss": 46.0,
-      "step": 47132
-    },
-    {
-      "epoch": 3.6036469981076897,
-      "grad_norm": 0.0026313778944313526,
-      "learning_rate": 0.00019999359430940842,
-      "loss": 46.0,
-      "step": 47133
-    },
-    {
-      "epoch": 3.6037234550910795,
-      "grad_norm": 0.0015562853077426553,
-      "learning_rate": 0.00019999359403753735,
-      "loss": 46.0,
-      "step": 47134
-    },
-    {
-      "epoch": 3.6037999120744693,
-      "grad_norm": 0.0026862341910600662,
-      "learning_rate": 0.0001999935937656605,
-      "loss": 46.0,
-      "step": 47135
-    },
-    {
-      "epoch": 3.6038763690578586,
-      "grad_norm": 0.004687380976974964,
-      "learning_rate": 0.00019999359349377792,
-      "loss": 46.0,
-      "step": 47136
-    },
-    {
-      "epoch": 3.6039528260412483,
-      "grad_norm": 0.0017340471968054771,
-      "learning_rate": 0.00019999359322188953,
-      "loss": 46.0,
-      "step": 47137
-    },
-    {
-      "epoch": 3.604029283024638,
-      "grad_norm": 0.0029415760654956102,
-      "learning_rate": 0.0001999935929499954,
-      "loss": 46.0,
-      "step": 47138
-    },
-    {
-      "epoch": 3.604105740008028,
-      "grad_norm": 0.0027057260740548372,
-      "learning_rate": 0.0001999935926780955,
-      "loss": 46.0,
-      "step": 47139
-    },
-    {
-      "epoch": 3.6041821969914176,
-      "grad_norm": 0.0014860716182738543,
-      "learning_rate": 0.0001999935924061898,
-      "loss": 46.0,
-      "step": 47140
-    },
-    {
-      "epoch": 3.6042586539748074,
-      "grad_norm": 0.0011507697636261582,
-      "learning_rate": 0.00019999359213427836,
-      "loss": 46.0,
-      "step": 47141
-    },
-    {
-      "epoch": 3.604335110958197,
-      "grad_norm": 0.0011245335917919874,
-      "learning_rate": 0.00019999359186236113,
-      "loss": 46.0,
-      "step": 47142
-    },
-    {
-      "epoch": 3.604411567941587,
-      "grad_norm": 0.004144533071666956,
-      "learning_rate": 0.00019999359159043816,
-      "loss": 46.0,
-      "step": 47143
-    },
-    {
-      "epoch": 3.6044880249249767,
-      "grad_norm": 0.004018872510641813,
-      "learning_rate": 0.00019999359131850936,
-      "loss": 46.0,
-      "step": 47144
-    },
-    {
-      "epoch": 3.6045644819083664,
-      "grad_norm": 0.004291102290153503,
-      "learning_rate": 0.00019999359104657484,
-      "loss": 46.0,
-      "step": 47145
-    },
-    {
-      "epoch": 3.604640938891756,
-      "grad_norm": 0.006488231476396322,
-      "learning_rate": 0.00019999359077463454,
-      "loss": 46.0,
-      "step": 47146
-    },
-    {
-      "epoch": 3.604717395875146,
-      "grad_norm": 0.0010807195212692022,
-      "learning_rate": 0.00019999359050268848,
-      "loss": 46.0,
-      "step": 47147
-    },
-    {
-      "epoch": 3.6047938528585357,
-      "grad_norm": 0.010882716625928879,
-      "learning_rate": 0.00019999359023073666,
-      "loss": 46.0,
-      "step": 47148
-    },
-    {
-      "epoch": 3.604870309841925,
-      "grad_norm": 0.002118566771969199,
-      "learning_rate": 0.00019999358995877905,
-      "loss": 46.0,
-      "step": 47149
-    },
-    {
-      "epoch": 3.604946766825315,
-      "grad_norm": 0.002121523953974247,
-      "learning_rate": 0.00019999358968681566,
-      "loss": 46.0,
-      "step": 47150
-    },
-    {
-      "epoch": 3.6050232238087045,
-      "grad_norm": 0.000612937321420759,
-      "learning_rate": 0.00019999358941484653,
-      "loss": 46.0,
-      "step": 47151
-    },
-    {
-      "epoch": 3.6050996807920943,
-      "grad_norm": 0.0010346367489546537,
-      "learning_rate": 0.00019999358914287162,
-      "loss": 46.0,
-      "step": 47152
-    },
-    {
-      "epoch": 3.605176137775484,
-      "grad_norm": 0.0013227801537141204,
-      "learning_rate": 0.00019999358887089094,
-      "loss": 46.0,
-      "step": 47153
-    },
-    {
-      "epoch": 3.605252594758874,
-      "grad_norm": 0.00315685011446476,
-      "learning_rate": 0.0001999935885989045,
-      "loss": 46.0,
-      "step": 47154
-    },
-    {
-      "epoch": 3.6053290517422636,
-      "grad_norm": 0.0027074345853179693,
-      "learning_rate": 0.00019999358832691226,
-      "loss": 46.0,
-      "step": 47155
-    },
-    {
-      "epoch": 3.6054055087256534,
-      "grad_norm": 0.002570002106949687,
-      "learning_rate": 0.0001999935880549143,
-      "loss": 46.0,
-      "step": 47156
-    },
-    {
-      "epoch": 3.605481965709043,
-      "grad_norm": 0.0026281271129846573,
-      "learning_rate": 0.00019999358778291054,
-      "loss": 46.0,
-      "step": 47157
-    },
-    {
-      "epoch": 3.6055584226924324,
-      "grad_norm": 0.0032254806719720364,
-      "learning_rate": 0.000199993587510901,
-      "loss": 46.0,
-      "step": 47158
-    },
-    {
-      "epoch": 3.605634879675822,
-      "grad_norm": 0.0015273073222488165,
-      "learning_rate": 0.0001999935872388857,
-      "loss": 46.0,
-      "step": 47159
-    },
-    {
-      "epoch": 3.605711336659212,
-      "grad_norm": 0.0040751914493739605,
-      "learning_rate": 0.00019999358696686463,
-      "loss": 46.0,
-      "step": 47160
-    },
-    {
-      "epoch": 3.6057877936426017,
-      "grad_norm": 0.003368379781022668,
-      "learning_rate": 0.00019999358669483782,
-      "loss": 46.0,
-      "step": 47161
-    },
-    {
-      "epoch": 3.6058642506259915,
-      "grad_norm": 0.003940230701118708,
-      "learning_rate": 0.0001999935864228052,
-      "loss": 46.0,
-      "step": 47162
-    },
-    {
-      "epoch": 3.6059407076093812,
-      "grad_norm": 0.0015386902960017323,
-      "learning_rate": 0.00019999358615076685,
-      "loss": 46.0,
-      "step": 47163
-    },
-    {
-      "epoch": 3.606017164592771,
-      "grad_norm": 0.0012139686150476336,
-      "learning_rate": 0.0001999935858787227,
-      "loss": 46.0,
-      "step": 47164
-    },
-    {
-      "epoch": 3.6060936215761608,
-      "grad_norm": 0.0013903678627684712,
-      "learning_rate": 0.00019999358560667278,
-      "loss": 46.0,
-      "step": 47165
-    },
-    {
-      "epoch": 3.6061700785595505,
-      "grad_norm": 0.001733639044687152,
-      "learning_rate": 0.0001999935853346171,
-      "loss": 46.0,
-      "step": 47166
-    },
-    {
-      "epoch": 3.6062465355429403,
-      "grad_norm": 0.0012850891798734665,
-      "learning_rate": 0.00019999358506255565,
-      "loss": 46.0,
-      "step": 47167
-    },
-    {
-      "epoch": 3.60632299252633,
-      "grad_norm": 0.002491175662726164,
-      "learning_rate": 0.00019999358479048845,
-      "loss": 46.0,
-      "step": 47168
-    },
-    {
-      "epoch": 3.60639944950972,
-      "grad_norm": 0.0014382494846358895,
-      "learning_rate": 0.00019999358451841545,
-      "loss": 46.0,
-      "step": 47169
-    },
-    {
-      "epoch": 3.6064759064931096,
-      "grad_norm": 0.0009760347893461585,
-      "learning_rate": 0.0001999935842463367,
-      "loss": 46.0,
-      "step": 47170
-    },
-    {
-      "epoch": 3.606552363476499,
-      "grad_norm": 0.0008436118368990719,
-      "learning_rate": 0.00019999358397425218,
-      "loss": 46.0,
-      "step": 47171
-    },
-    {
-      "epoch": 3.6066288204598886,
-      "grad_norm": 0.0076487017795443535,
-      "learning_rate": 0.0001999935837021619,
-      "loss": 46.0,
-      "step": 47172
-    },
-    {
-      "epoch": 3.6067052774432784,
-      "grad_norm": 0.002492173109203577,
-      "learning_rate": 0.00019999358343006582,
-      "loss": 46.0,
-      "step": 47173
-    },
-    {
-      "epoch": 3.606781734426668,
-      "grad_norm": 0.004754967056214809,
-      "learning_rate": 0.00019999358315796398,
-      "loss": 46.0,
-      "step": 47174
-    },
-    {
-      "epoch": 3.606858191410058,
-      "grad_norm": 0.0028412227984517813,
-      "learning_rate": 0.0001999935828858564,
-      "loss": 46.0,
-      "step": 47175
-    },
-    {
-      "epoch": 3.6069346483934477,
-      "grad_norm": 0.0014061275869607925,
-      "learning_rate": 0.00019999358261374304,
-      "loss": 46.0,
-      "step": 47176
-    },
-    {
-      "epoch": 3.6070111053768374,
-      "grad_norm": 0.005551229231059551,
-      "learning_rate": 0.00019999358234162388,
-      "loss": 46.0,
-      "step": 47177
-    },
-    {
-      "epoch": 3.607087562360227,
-      "grad_norm": 0.002698456170037389,
-      "learning_rate": 0.00019999358206949897,
-      "loss": 46.0,
-      "step": 47178
-    },
-    {
-      "epoch": 3.6071640193436165,
-      "grad_norm": 0.0009287746506743133,
-      "learning_rate": 0.0001999935817973683,
-      "loss": 46.0,
-      "step": 47179
-    },
-    {
-      "epoch": 3.6072404763270063,
-      "grad_norm": 0.0007593818590976298,
-      "learning_rate": 0.00019999358152523184,
-      "loss": 46.0,
-      "step": 47180
-    },
-    {
-      "epoch": 3.607316933310396,
-      "grad_norm": 0.00284673017449677,
-      "learning_rate": 0.00019999358125308964,
-      "loss": 46.0,
-      "step": 47181
-    },
-    {
-      "epoch": 3.607393390293786,
-      "grad_norm": 0.0020804463420063257,
-      "learning_rate": 0.00019999358098094164,
-      "loss": 46.0,
-      "step": 47182
-    },
-    {
-      "epoch": 3.6074698472771756,
-      "grad_norm": 0.006738160736858845,
-      "learning_rate": 0.00019999358070878792,
-      "loss": 46.0,
-      "step": 47183
-    },
-    {
-      "epoch": 3.6075463042605653,
-      "grad_norm": 0.007507022935897112,
-      "learning_rate": 0.00019999358043662838,
-      "loss": 46.0,
-      "step": 47184
-    },
-    {
-      "epoch": 3.607622761243955,
-      "grad_norm": 0.010416602715849876,
-      "learning_rate": 0.0001999935801644631,
-      "loss": 46.0,
-      "step": 47185
-    },
-    {
-      "epoch": 3.607699218227345,
-      "grad_norm": 0.003749399445950985,
-      "learning_rate": 0.00019999357989229202,
-      "loss": 46.0,
-      "step": 47186
-    },
-    {
-      "epoch": 3.6077756752107346,
-      "grad_norm": 0.0025361282750964165,
-      "learning_rate": 0.0001999935796201152,
-      "loss": 46.0,
-      "step": 47187
-    },
-    {
-      "epoch": 3.6078521321941244,
-      "grad_norm": 0.005283393897116184,
-      "learning_rate": 0.00019999357934793262,
-      "loss": 46.0,
-      "step": 47188
-    },
-    {
-      "epoch": 3.607928589177514,
-      "grad_norm": 0.003208644688129425,
-      "learning_rate": 0.00019999357907574423,
-      "loss": 46.0,
-      "step": 47189
-    },
-    {
-      "epoch": 3.608005046160904,
-      "grad_norm": 0.0030392338521778584,
-      "learning_rate": 0.0001999935788035501,
-      "loss": 46.0,
-      "step": 47190
-    },
-    {
-      "epoch": 3.6080815031442937,
-      "grad_norm": 0.0019993705209344625,
-      "learning_rate": 0.0001999935785313502,
-      "loss": 46.0,
-      "step": 47191
-    },
-    {
-      "epoch": 3.6081579601276834,
-      "grad_norm": 0.0029805595986545086,
-      "learning_rate": 0.00019999357825914454,
-      "loss": 46.0,
-      "step": 47192
-    },
-    {
-      "epoch": 3.6082344171110727,
-      "grad_norm": 0.0007770429365336895,
-      "learning_rate": 0.0001999935779869331,
-      "loss": 46.0,
-      "step": 47193
-    },
-    {
-      "epoch": 3.6083108740944625,
-      "grad_norm": 0.0035184675361961126,
-      "learning_rate": 0.00019999357771471586,
-      "loss": 46.0,
-      "step": 47194
-    },
-    {
-      "epoch": 3.6083873310778523,
-      "grad_norm": 0.0033095621038228273,
-      "learning_rate": 0.00019999357744249286,
-      "loss": 46.0,
-      "step": 47195
-    },
-    {
-      "epoch": 3.608463788061242,
-      "grad_norm": 0.0013891630806028843,
-      "learning_rate": 0.00019999357717026412,
-      "loss": 46.0,
-      "step": 47196
-    },
-    {
-      "epoch": 3.6085402450446318,
-      "grad_norm": 0.0023033798206597567,
-      "learning_rate": 0.00019999357689802963,
-      "loss": 46.0,
-      "step": 47197
-    },
-    {
-      "epoch": 3.6086167020280215,
-      "grad_norm": 0.0018179116304963827,
-      "learning_rate": 0.0001999935766257893,
-      "loss": 46.0,
-      "step": 47198
-    },
-    {
-      "epoch": 3.6086931590114113,
-      "grad_norm": 0.004559769295156002,
-      "learning_rate": 0.00019999357635354327,
-      "loss": 46.0,
-      "step": 47199
-    },
-    {
-      "epoch": 3.608769615994801,
-      "grad_norm": 0.00302033219486475,
-      "learning_rate": 0.00019999357608129143,
-      "loss": 46.0,
-      "step": 47200
-    },
-    {
-      "epoch": 3.6088460729781904,
-      "grad_norm": 0.0030741288792341948,
-      "learning_rate": 0.00019999357580903384,
-      "loss": 46.0,
-      "step": 47201
-    },
-    {
-      "epoch": 3.60892252996158,
-      "grad_norm": 0.003815828822553158,
-      "learning_rate": 0.00019999357553677048,
-      "loss": 46.0,
-      "step": 47202
-    },
-    {
-      "epoch": 3.60899898694497,
-      "grad_norm": 0.0027954436372965574,
-      "learning_rate": 0.00019999357526450133,
-      "loss": 46.0,
-      "step": 47203
-    },
-    {
-      "epoch": 3.6090754439283597,
-      "grad_norm": 0.0015756351640447974,
-      "learning_rate": 0.00019999357499222642,
-      "loss": 46.0,
-      "step": 47204
-    },
-    {
-      "epoch": 3.6091519009117494,
-      "grad_norm": 0.0029789507389068604,
-      "learning_rate": 0.00019999357471994577,
-      "loss": 46.0,
-      "step": 47205
-    },
-    {
-      "epoch": 3.609228357895139,
-      "grad_norm": 0.003662483999505639,
-      "learning_rate": 0.00019999357444765932,
-      "loss": 46.0,
-      "step": 47206
-    },
-    {
-      "epoch": 3.609304814878529,
-      "grad_norm": 0.0007836703443899751,
-      "learning_rate": 0.00019999357417536712,
-      "loss": 46.0,
-      "step": 47207
-    },
-    {
-      "epoch": 3.6093812718619187,
-      "grad_norm": 0.001132678589783609,
-      "learning_rate": 0.00019999357390306912,
-      "loss": 46.0,
-      "step": 47208
-    },
-    {
-      "epoch": 3.6094577288453085,
-      "grad_norm": 0.0031996930483728647,
-      "learning_rate": 0.00019999357363076537,
-      "loss": 46.0,
-      "step": 47209
-    },
-    {
-      "epoch": 3.6095341858286982,
-      "grad_norm": 0.0007871190900914371,
-      "learning_rate": 0.00019999357335845585,
-      "loss": 46.0,
-      "step": 47210
-    },
-    {
-      "epoch": 3.609610642812088,
-      "grad_norm": 0.003387199016287923,
-      "learning_rate": 0.00019999357308614056,
-      "loss": 46.0,
-      "step": 47211
-    },
-    {
-      "epoch": 3.6096870997954777,
-      "grad_norm": 0.002372474642470479,
-      "learning_rate": 0.00019999357281381952,
-      "loss": 46.0,
-      "step": 47212
-    },
-    {
-      "epoch": 3.6097635567788675,
-      "grad_norm": 0.0020967365708202124,
-      "learning_rate": 0.00019999357254149269,
-      "loss": 46.0,
-      "step": 47213
-    },
-    {
-      "epoch": 3.6098400137622573,
-      "grad_norm": 0.004114883486181498,
-      "learning_rate": 0.0001999935722691601,
-      "loss": 46.0,
-      "step": 47214
-    },
-    {
-      "epoch": 3.6099164707456466,
-      "grad_norm": 0.004482800140976906,
-      "learning_rate": 0.00019999357199682174,
-      "loss": 46.0,
-      "step": 47215
-    },
-    {
-      "epoch": 3.6099929277290363,
-      "grad_norm": 0.0024372737389057875,
-      "learning_rate": 0.00019999357172447758,
-      "loss": 46.0,
-      "step": 47216
-    },
-    {
-      "epoch": 3.610069384712426,
-      "grad_norm": 0.0022202099207788706,
-      "learning_rate": 0.0001999935714521277,
-      "loss": 46.0,
-      "step": 47217
-    },
-    {
-      "epoch": 3.610145841695816,
-      "grad_norm": 0.0026044619735330343,
-      "learning_rate": 0.00019999357117977203,
-      "loss": 46.0,
-      "step": 47218
-    },
-    {
-      "epoch": 3.6102222986792056,
-      "grad_norm": 0.001976800849661231,
-      "learning_rate": 0.0001999935709074106,
-      "loss": 46.0,
-      "step": 47219
-    },
-    {
-      "epoch": 3.6102987556625954,
-      "grad_norm": 0.003462227527052164,
-      "learning_rate": 0.00019999357063504335,
-      "loss": 46.0,
-      "step": 47220
-    },
-    {
-      "epoch": 3.610375212645985,
-      "grad_norm": 0.0028351584915071726,
-      "learning_rate": 0.00019999357036267038,
-      "loss": 46.0,
-      "step": 47221
-    },
-    {
-      "epoch": 3.610451669629375,
-      "grad_norm": 0.0008801952935755253,
-      "learning_rate": 0.00019999357009029166,
-      "loss": 46.0,
-      "step": 47222
-    },
-    {
-      "epoch": 3.6105281266127642,
-      "grad_norm": 0.0038246368058025837,
-      "learning_rate": 0.00019999356981790714,
-      "loss": 46.0,
-      "step": 47223
-    },
-    {
-      "epoch": 3.610604583596154,
-      "grad_norm": 0.003297134069725871,
-      "learning_rate": 0.00019999356954551685,
-      "loss": 46.0,
-      "step": 47224
-    },
-    {
-      "epoch": 3.6106810405795438,
-      "grad_norm": 0.009903750382363796,
-      "learning_rate": 0.00019999356927312078,
-      "loss": 46.0,
-      "step": 47225
-    },
-    {
-      "epoch": 3.6107574975629335,
-      "grad_norm": 0.0018430057680234313,
-      "learning_rate": 0.00019999356900071897,
-      "loss": 46.0,
-      "step": 47226
-    },
-    {
-      "epoch": 3.6108339545463233,
-      "grad_norm": 0.005019971169531345,
-      "learning_rate": 0.0001999935687283114,
-      "loss": 46.0,
-      "step": 47227
-    },
-    {
-      "epoch": 3.610910411529713,
-      "grad_norm": 0.0025201118551194668,
-      "learning_rate": 0.000199993568455898,
-      "loss": 46.0,
-      "step": 47228
-    },
-    {
-      "epoch": 3.610986868513103,
-      "grad_norm": 0.002068077214062214,
-      "learning_rate": 0.0001999935681834789,
-      "loss": 46.0,
-      "step": 47229
-    },
-    {
-      "epoch": 3.6110633254964926,
-      "grad_norm": 0.00215020845644176,
-      "learning_rate": 0.00019999356791105397,
-      "loss": 46.0,
-      "step": 47230
-    },
-    {
-      "epoch": 3.6111397824798823,
-      "grad_norm": 0.0011204774491488934,
-      "learning_rate": 0.00019999356763862331,
-      "loss": 46.0,
-      "step": 47231
-    },
-    {
-      "epoch": 3.611216239463272,
-      "grad_norm": 0.0012542122276499867,
-      "learning_rate": 0.0001999935673661869,
-      "loss": 46.0,
-      "step": 47232
-    },
-    {
-      "epoch": 3.611292696446662,
-      "grad_norm": 0.0031704932916909456,
-      "learning_rate": 0.0001999935670937447,
-      "loss": 46.0,
-      "step": 47233
-    },
-    {
-      "epoch": 3.6113691534300516,
-      "grad_norm": 0.003615632886067033,
-      "learning_rate": 0.0001999935668212967,
-      "loss": 46.0,
-      "step": 47234
-    },
-    {
-      "epoch": 3.6114456104134414,
-      "grad_norm": 0.0019761386793106794,
-      "learning_rate": 0.00019999356654884295,
-      "loss": 46.0,
-      "step": 47235
-    },
-    {
-      "epoch": 3.611522067396831,
-      "grad_norm": 0.0020734784193336964,
-      "learning_rate": 0.00019999356627638343,
-      "loss": 46.0,
-      "step": 47236
-    },
-    {
-      "epoch": 3.6115985243802204,
-      "grad_norm": 0.0018823568243533373,
-      "learning_rate": 0.00019999356600391814,
-      "loss": 46.0,
-      "step": 47237
-    },
-    {
-      "epoch": 3.61167498136361,
-      "grad_norm": 0.00300602032802999,
-      "learning_rate": 0.0001999935657314471,
-      "loss": 46.0,
-      "step": 47238
-    },
-    {
-      "epoch": 3.611751438347,
-      "grad_norm": 0.0037502057384699583,
-      "learning_rate": 0.0001999935654589703,
-      "loss": 46.0,
-      "step": 47239
-    },
-    {
-      "epoch": 3.6118278953303897,
-      "grad_norm": 0.004724597092717886,
-      "learning_rate": 0.00019999356518648768,
-      "loss": 46.0,
-      "step": 47240
-    },
-    {
-      "epoch": 3.6119043523137795,
-      "grad_norm": 0.0010078669292852283,
-      "learning_rate": 0.00019999356491399932,
-      "loss": 46.0,
-      "step": 47241
-    },
-    {
-      "epoch": 3.6119808092971692,
-      "grad_norm": 0.005235090386122465,
-      "learning_rate": 0.00019999356464150522,
-      "loss": 46.0,
-      "step": 47242
-    },
-    {
-      "epoch": 3.612057266280559,
-      "grad_norm": 0.0011519150575622916,
-      "learning_rate": 0.0001999935643690053,
-      "loss": 46.0,
-      "step": 47243
-    },
-    {
-      "epoch": 3.6121337232639488,
-      "grad_norm": 0.005454646423459053,
-      "learning_rate": 0.00019999356409649964,
-      "loss": 46.0,
-      "step": 47244
-    },
-    {
-      "epoch": 3.612210180247338,
-      "grad_norm": 0.0009252894087694585,
-      "learning_rate": 0.00019999356382398818,
-      "loss": 46.0,
-      "step": 47245
-    },
-    {
-      "epoch": 3.612286637230728,
-      "grad_norm": 0.006963848136365414,
-      "learning_rate": 0.00019999356355147099,
-      "loss": 46.0,
-      "step": 47246
-    },
-    {
-      "epoch": 3.6123630942141176,
-      "grad_norm": 0.006564173381775618,
-      "learning_rate": 0.00019999356327894802,
-      "loss": 46.0,
-      "step": 47247
-    },
-    {
-      "epoch": 3.6124395511975074,
-      "grad_norm": 0.0027480721473693848,
-      "learning_rate": 0.00019999356300641927,
-      "loss": 46.0,
-      "step": 47248
-    },
-    {
-      "epoch": 3.612516008180897,
-      "grad_norm": 0.0012267193524166942,
-      "learning_rate": 0.00019999356273388475,
-      "loss": 46.0,
-      "step": 47249
-    },
-    {
-      "epoch": 3.612592465164287,
-      "grad_norm": 0.0043920292519032955,
-      "learning_rate": 0.0001999935624613445,
-      "loss": 46.0,
-      "step": 47250
-    },
-    {
-      "epoch": 3.6126689221476767,
-      "grad_norm": 0.0039373235777020454,
-      "learning_rate": 0.00019999356218879842,
-      "loss": 46.0,
-      "step": 47251
-    },
-    {
-      "epoch": 3.6127453791310664,
-      "grad_norm": 0.0036451241467148066,
-      "learning_rate": 0.00019999356191624661,
-      "loss": 46.0,
-      "step": 47252
-    },
-    {
-      "epoch": 3.612821836114456,
-      "grad_norm": 0.0016169092850759625,
-      "learning_rate": 0.00019999356164368903,
-      "loss": 46.0,
-      "step": 47253
-    },
-    {
-      "epoch": 3.612898293097846,
-      "grad_norm": 0.002296799561008811,
-      "learning_rate": 0.00019999356137112567,
-      "loss": 46.0,
-      "step": 47254
-    },
-    {
-      "epoch": 3.6129747500812357,
-      "grad_norm": 0.0046439566649496555,
-      "learning_rate": 0.00019999356109855654,
-      "loss": 46.0,
-      "step": 47255
-    },
-    {
-      "epoch": 3.6130512070646255,
-      "grad_norm": 0.0033165907952934504,
-      "learning_rate": 0.00019999356082598164,
-      "loss": 46.0,
-      "step": 47256
-    },
-    {
-      "epoch": 3.613127664048015,
-      "grad_norm": 0.0026466366834938526,
-      "learning_rate": 0.00019999356055340096,
-      "loss": 46.0,
-      "step": 47257
-    },
-    {
-      "epoch": 3.613204121031405,
-      "grad_norm": 0.0049867830239236355,
-      "learning_rate": 0.00019999356028081456,
-      "loss": 46.0,
-      "step": 47258
-    },
-    {
-      "epoch": 3.6132805780147943,
-      "grad_norm": 0.0011484172428026795,
-      "learning_rate": 0.00019999356000822234,
-      "loss": 46.0,
-      "step": 47259
-    },
-    {
-      "epoch": 3.613357034998184,
-      "grad_norm": 0.0011415337212383747,
-      "learning_rate": 0.00019999355973562434,
-      "loss": 46.0,
-      "step": 47260
-    },
-    {
-      "epoch": 3.613433491981574,
-      "grad_norm": 0.002642180072143674,
-      "learning_rate": 0.00019999355946302062,
-      "loss": 46.0,
-      "step": 47261
-    },
-    {
-      "epoch": 3.6135099489649636,
-      "grad_norm": 0.001251637819223106,
-      "learning_rate": 0.0001999935591904111,
-      "loss": 46.0,
-      "step": 47262
-    },
-    {
-      "epoch": 3.6135864059483533,
-      "grad_norm": 0.002847895724698901,
-      "learning_rate": 0.00019999355891779584,
-      "loss": 46.0,
-      "step": 47263
-    },
-    {
-      "epoch": 3.613662862931743,
-      "grad_norm": 0.005133089609444141,
-      "learning_rate": 0.0001999935586451748,
-      "loss": 46.0,
-      "step": 47264
-    },
-    {
-      "epoch": 3.613739319915133,
-      "grad_norm": 0.00236779497936368,
-      "learning_rate": 0.00019999355837254797,
-      "loss": 46.0,
-      "step": 47265
-    },
-    {
-      "epoch": 3.6138157768985226,
-      "grad_norm": 0.00249236891977489,
-      "learning_rate": 0.00019999355809991538,
-      "loss": 46.0,
-      "step": 47266
-    },
-    {
-      "epoch": 3.613892233881912,
-      "grad_norm": 0.003141065128147602,
-      "learning_rate": 0.00019999355782727703,
-      "loss": 46.0,
-      "step": 47267
-    },
-    {
-      "epoch": 3.6139686908653017,
-      "grad_norm": 0.0022167351562529802,
-      "learning_rate": 0.0001999935575546329,
-      "loss": 46.0,
-      "step": 47268
-    },
-    {
-      "epoch": 3.6140451478486915,
-      "grad_norm": 0.001890449901111424,
-      "learning_rate": 0.00019999355728198302,
-      "loss": 46.0,
-      "step": 47269
-    },
-    {
-      "epoch": 3.6141216048320812,
-      "grad_norm": 0.0017392619047313929,
-      "learning_rate": 0.00019999355700932734,
-      "loss": 46.0,
-      "step": 47270
-    },
-    {
-      "epoch": 3.614198061815471,
-      "grad_norm": 0.0016764000756666064,
-      "learning_rate": 0.00019999355673666592,
-      "loss": 46.0,
-      "step": 47271
-    },
-    {
-      "epoch": 3.6142745187988607,
-      "grad_norm": 0.004622225649654865,
-      "learning_rate": 0.00019999355646399872,
-      "loss": 46.0,
-      "step": 47272
-    },
-    {
-      "epoch": 3.6143509757822505,
-      "grad_norm": 0.0005916609079577029,
-      "learning_rate": 0.00019999355619132572,
-      "loss": 46.0,
-      "step": 47273
-    },
-    {
-      "epoch": 3.6144274327656403,
-      "grad_norm": 0.0012697236379608512,
-      "learning_rate": 0.00019999355591864698,
-      "loss": 46.0,
-      "step": 47274
-    },
-    {
-      "epoch": 3.61450388974903,
-      "grad_norm": 0.0034948044922202826,
-      "learning_rate": 0.0001999935556459625,
-      "loss": 46.0,
-      "step": 47275
-    },
-    {
-      "epoch": 3.61458034673242,
-      "grad_norm": 0.0029641392175108194,
-      "learning_rate": 0.00019999355537327223,
-      "loss": 46.0,
-      "step": 47276
-    },
-    {
-      "epoch": 3.6146568037158096,
-      "grad_norm": 0.0010084653040394187,
-      "learning_rate": 0.00019999355510057616,
-      "loss": 46.0,
-      "step": 47277
-    },
-    {
-      "epoch": 3.6147332606991993,
-      "grad_norm": 0.0007937371265143156,
-      "learning_rate": 0.00019999355482787435,
-      "loss": 46.0,
-      "step": 47278
-    },
-    {
-      "epoch": 3.614809717682589,
-      "grad_norm": 0.0016434750286862254,
-      "learning_rate": 0.00019999355455516677,
-      "loss": 46.0,
-      "step": 47279
-    },
-    {
-      "epoch": 3.6148861746659784,
-      "grad_norm": 0.0033585482742637396,
-      "learning_rate": 0.0001999935542824534,
-      "loss": 46.0,
-      "step": 47280
-    },
-    {
-      "epoch": 3.614962631649368,
-      "grad_norm": 0.005722475238144398,
-      "learning_rate": 0.00019999355400973428,
-      "loss": 46.0,
-      "step": 47281
-    },
-    {
-      "epoch": 3.615039088632758,
-      "grad_norm": 0.0017641838639974594,
-      "learning_rate": 0.00019999355373700938,
-      "loss": 46.0,
-      "step": 47282
-    },
-    {
-      "epoch": 3.6151155456161477,
-      "grad_norm": 0.00762800732627511,
-      "learning_rate": 0.00019999355346427873,
-      "loss": 46.0,
-      "step": 47283
-    },
-    {
-      "epoch": 3.6151920025995374,
-      "grad_norm": 0.0025354577228426933,
-      "learning_rate": 0.0001999935531915423,
-      "loss": 46.0,
-      "step": 47284
-    },
-    {
-      "epoch": 3.615268459582927,
-      "grad_norm": 0.0017457692883908749,
-      "learning_rate": 0.00019999355291880008,
-      "loss": 46.0,
-      "step": 47285
-    },
-    {
-      "epoch": 3.615344916566317,
-      "grad_norm": 0.0011523324064910412,
-      "learning_rate": 0.00019999355264605214,
-      "loss": 46.0,
-      "step": 47286
-    },
-    {
-      "epoch": 3.6154213735497067,
-      "grad_norm": 0.0023378196638077497,
-      "learning_rate": 0.0001999935523732984,
-      "loss": 46.0,
-      "step": 47287
-    },
-    {
-      "epoch": 3.6154978305330965,
-      "grad_norm": 0.0016617362853139639,
-      "learning_rate": 0.0001999935521005389,
-      "loss": 46.0,
-      "step": 47288
-    },
-    {
-      "epoch": 3.615574287516486,
-      "grad_norm": 0.0019301587017253041,
-      "learning_rate": 0.00019999355182777362,
-      "loss": 46.0,
-      "step": 47289
-    },
-    {
-      "epoch": 3.6156507444998756,
-      "grad_norm": 0.005819922313094139,
-      "learning_rate": 0.00019999355155500258,
-      "loss": 46.0,
-      "step": 47290
-    },
-    {
-      "epoch": 3.6157272014832653,
-      "grad_norm": 0.0033543517347425222,
-      "learning_rate": 0.00019999355128222574,
-      "loss": 46.0,
-      "step": 47291
-    },
-    {
-      "epoch": 3.615803658466655,
-      "grad_norm": 0.00135278410743922,
-      "learning_rate": 0.00019999355100944316,
-      "loss": 46.0,
-      "step": 47292
-    },
-    {
-      "epoch": 3.615880115450045,
-      "grad_norm": 0.002793761668726802,
-      "learning_rate": 0.00019999355073665483,
-      "loss": 46.0,
-      "step": 47293
-    },
-    {
-      "epoch": 3.6159565724334346,
-      "grad_norm": 0.0035002746153622866,
-      "learning_rate": 0.00019999355046386067,
-      "loss": 46.0,
-      "step": 47294
-    },
-    {
-      "epoch": 3.6160330294168244,
-      "grad_norm": 0.013129138387739658,
-      "learning_rate": 0.0001999935501910608,
-      "loss": 46.0,
-      "step": 47295
-    },
-    {
-      "epoch": 3.616109486400214,
-      "grad_norm": 0.00368657149374485,
-      "learning_rate": 0.00019999354991825515,
-      "loss": 46.0,
-      "step": 47296
-    },
-    {
-      "epoch": 3.616185943383604,
-      "grad_norm": 0.00111564586404711,
-      "learning_rate": 0.0001999935496454437,
-      "loss": 46.0,
-      "step": 47297
-    },
-    {
-      "epoch": 3.6162624003669936,
-      "grad_norm": 0.001937650260515511,
-      "learning_rate": 0.0001999935493726265,
-      "loss": 46.0,
-      "step": 47298
-    },
-    {
-      "epoch": 3.6163388573503834,
-      "grad_norm": 0.0021980348974466324,
-      "learning_rate": 0.00019999354909980356,
-      "loss": 46.0,
-      "step": 47299
-    },
-    {
-      "epoch": 3.616415314333773,
-      "grad_norm": 0.003814734984189272,
-      "learning_rate": 0.00019999354882697481,
-      "loss": 46.0,
-      "step": 47300
-    },
-    {
-      "epoch": 3.616491771317163,
-      "grad_norm": 0.004749145358800888,
-      "learning_rate": 0.0001999935485541403,
-      "loss": 46.0,
-      "step": 47301
-    },
-    {
-      "epoch": 3.6165682283005522,
-      "grad_norm": 0.005444707814604044,
-      "learning_rate": 0.00019999354828130003,
-      "loss": 46.0,
-      "step": 47302
-    },
-    {
-      "epoch": 3.616644685283942,
-      "grad_norm": 0.0017532651545479894,
-      "learning_rate": 0.000199993548008454,
-      "loss": 46.0,
-      "step": 47303
-    },
-    {
-      "epoch": 3.6167211422673318,
-      "grad_norm": 0.000540543464012444,
-      "learning_rate": 0.00019999354773560216,
-      "loss": 46.0,
-      "step": 47304
-    },
-    {
-      "epoch": 3.6167975992507215,
-      "grad_norm": 0.0018666167743504047,
-      "learning_rate": 0.0001999935474627446,
-      "loss": 46.0,
-      "step": 47305
-    },
-    {
-      "epoch": 3.6168740562341113,
-      "grad_norm": 0.001620797673240304,
-      "learning_rate": 0.00019999354718988122,
-      "loss": 46.0,
-      "step": 47306
-    },
-    {
-      "epoch": 3.616950513217501,
-      "grad_norm": 0.003897682996466756,
-      "learning_rate": 0.00019999354691701212,
-      "loss": 46.0,
-      "step": 47307
-    },
-    {
-      "epoch": 3.617026970200891,
-      "grad_norm": 0.0010353117249906063,
-      "learning_rate": 0.00019999354664413724,
-      "loss": 46.0,
-      "step": 47308
-    },
-    {
-      "epoch": 3.6171034271842806,
-      "grad_norm": 0.0027613837737590075,
-      "learning_rate": 0.0001999935463712566,
-      "loss": 46.0,
-      "step": 47309
-    },
-    {
-      "epoch": 3.61717988416767,
-      "grad_norm": 0.004738753195852041,
-      "learning_rate": 0.00019999354609837015,
-      "loss": 46.0,
-      "step": 47310
-    },
-    {
-      "epoch": 3.6172563411510597,
-      "grad_norm": 0.001971774036064744,
-      "learning_rate": 0.00019999354582547798,
-      "loss": 46.0,
-      "step": 47311
-    },
-    {
-      "epoch": 3.6173327981344494,
-      "grad_norm": 0.0016679092077538371,
-      "learning_rate": 0.00019999354555258,
-      "loss": 46.0,
-      "step": 47312
-    },
-    {
-      "epoch": 3.617409255117839,
-      "grad_norm": 0.0034473752602934837,
-      "learning_rate": 0.00019999354527967626,
-      "loss": 46.0,
-      "step": 47313
-    },
-    {
-      "epoch": 3.617485712101229,
-      "grad_norm": 0.002821042202413082,
-      "learning_rate": 0.00019999354500676675,
-      "loss": 46.0,
-      "step": 47314
-    },
-    {
-      "epoch": 3.6175621690846187,
-      "grad_norm": 0.002369681838899851,
-      "learning_rate": 0.0001999935447338515,
-      "loss": 46.0,
-      "step": 47315
-    },
-    {
-      "epoch": 3.6176386260680085,
-      "grad_norm": 0.0009700052905827761,
-      "learning_rate": 0.00019999354446093045,
-      "loss": 46.0,
-      "step": 47316
-    },
-    {
-      "epoch": 3.617715083051398,
-      "grad_norm": 0.0037375895772129297,
-      "learning_rate": 0.00019999354418800364,
-      "loss": 46.0,
-      "step": 47317
-    },
-    {
-      "epoch": 3.617791540034788,
-      "grad_norm": 0.005681247916072607,
-      "learning_rate": 0.00019999354391507106,
-      "loss": 46.0,
-      "step": 47318
-    },
-    {
-      "epoch": 3.6178679970181777,
-      "grad_norm": 0.0063100517727434635,
-      "learning_rate": 0.0001999935436421327,
-      "loss": 46.0,
-      "step": 47319
-    },
-    {
-      "epoch": 3.6179444540015675,
-      "grad_norm": 0.003021263051778078,
-      "learning_rate": 0.00019999354336918857,
-      "loss": 46.0,
-      "step": 47320
-    },
-    {
-      "epoch": 3.6180209109849573,
-      "grad_norm": 0.0005241144099272788,
-      "learning_rate": 0.0001999935430962387,
-      "loss": 46.0,
-      "step": 47321
-    },
-    {
-      "epoch": 3.618097367968347,
-      "grad_norm": 0.00224726228043437,
-      "learning_rate": 0.00019999354282328305,
-      "loss": 46.0,
-      "step": 47322
-    },
-    {
-      "epoch": 3.618173824951737,
-      "grad_norm": 0.002868939656764269,
-      "learning_rate": 0.00019999354255032163,
-      "loss": 46.0,
-      "step": 47323
-    },
-    {
-      "epoch": 3.618250281935126,
-      "grad_norm": 0.0022335646208375692,
-      "learning_rate": 0.00019999354227735446,
-      "loss": 46.0,
-      "step": 47324
-    },
-    {
-      "epoch": 3.618326738918516,
-      "grad_norm": 0.0011630482040345669,
-      "learning_rate": 0.00019999354200438146,
-      "loss": 46.0,
-      "step": 47325
-    },
-    {
-      "epoch": 3.6184031959019056,
-      "grad_norm": 0.0011882652761414647,
-      "learning_rate": 0.00019999354173140275,
-      "loss": 46.0,
-      "step": 47326
-    },
-    {
-      "epoch": 3.6184796528852954,
-      "grad_norm": 0.003643147414550185,
-      "learning_rate": 0.00019999354145841823,
-      "loss": 46.0,
-      "step": 47327
-    },
-    {
-      "epoch": 3.618556109868685,
-      "grad_norm": 0.0030876793898642063,
-      "learning_rate": 0.00019999354118542797,
-      "loss": 46.0,
-      "step": 47328
-    },
-    {
-      "epoch": 3.618632566852075,
-      "grad_norm": 0.003609174396842718,
-      "learning_rate": 0.00019999354091243193,
-      "loss": 46.0,
-      "step": 47329
-    },
-    {
-      "epoch": 3.6187090238354647,
-      "grad_norm": 0.0019165801350027323,
-      "learning_rate": 0.00019999354063943012,
-      "loss": 46.0,
-      "step": 47330
-    },
-    {
-      "epoch": 3.6187854808188544,
-      "grad_norm": 0.0014679976738989353,
-      "learning_rate": 0.00019999354036642254,
-      "loss": 46.0,
-      "step": 47331
-    },
-    {
-      "epoch": 3.6188619378022437,
-      "grad_norm": 0.0014446813147515059,
-      "learning_rate": 0.0001999935400934092,
-      "loss": 46.0,
-      "step": 47332
-    },
-    {
-      "epoch": 3.6189383947856335,
-      "grad_norm": 0.004242730792611837,
-      "learning_rate": 0.00019999353982039008,
-      "loss": 46.0,
-      "step": 47333
-    },
-    {
-      "epoch": 3.6190148517690233,
-      "grad_norm": 0.006166866514831781,
-      "learning_rate": 0.00019999353954736518,
-      "loss": 46.0,
-      "step": 47334
-    },
-    {
-      "epoch": 3.619091308752413,
-      "grad_norm": 0.00638245465233922,
-      "learning_rate": 0.00019999353927433453,
-      "loss": 46.0,
-      "step": 47335
-    },
-    {
-      "epoch": 3.619167765735803,
-      "grad_norm": 0.00381590542383492,
-      "learning_rate": 0.00019999353900129814,
-      "loss": 46.0,
-      "step": 47336
-    },
-    {
-      "epoch": 3.6192442227191925,
-      "grad_norm": 0.0054197246208786964,
-      "learning_rate": 0.00019999353872825592,
-      "loss": 46.0,
-      "step": 47337
-    },
-    {
-      "epoch": 3.6193206797025823,
-      "grad_norm": 0.0020835886243730783,
-      "learning_rate": 0.00019999353845520797,
-      "loss": 46.0,
-      "step": 47338
-    },
-    {
-      "epoch": 3.619397136685972,
-      "grad_norm": 0.003529014065861702,
-      "learning_rate": 0.00019999353818215423,
-      "loss": 46.0,
-      "step": 47339
-    },
-    {
-      "epoch": 3.619473593669362,
-      "grad_norm": 0.0011554706143215299,
-      "learning_rate": 0.00019999353790909474,
-      "loss": 46.0,
-      "step": 47340
-    },
-    {
-      "epoch": 3.6195500506527516,
-      "grad_norm": 0.0021690039429813623,
-      "learning_rate": 0.00019999353763602948,
-      "loss": 46.0,
-      "step": 47341
-    },
-    {
-      "epoch": 3.6196265076361414,
-      "grad_norm": 0.0030679211486130953,
-      "learning_rate": 0.00019999353736295845,
-      "loss": 46.0,
-      "step": 47342
-    },
-    {
-      "epoch": 3.619702964619531,
-      "grad_norm": 0.004602686036378145,
-      "learning_rate": 0.00019999353708988164,
-      "loss": 46.0,
-      "step": 47343
-    },
-    {
-      "epoch": 3.619779421602921,
-      "grad_norm": 0.0019608382135629654,
-      "learning_rate": 0.00019999353681679906,
-      "loss": 46.0,
-      "step": 47344
-    },
-    {
-      "epoch": 3.6198558785863106,
-      "grad_norm": 0.0027885353192687035,
-      "learning_rate": 0.0001999935365437107,
-      "loss": 46.0,
-      "step": 47345
-    },
-    {
-      "epoch": 3.6199323355697,
-      "grad_norm": 0.006393016781657934,
-      "learning_rate": 0.0001999935362706166,
-      "loss": 46.0,
-      "step": 47346
-    },
-    {
-      "epoch": 3.6200087925530897,
-      "grad_norm": 0.0023105100262910128,
-      "learning_rate": 0.00019999353599751672,
-      "loss": 46.0,
-      "step": 47347
-    },
-    {
-      "epoch": 3.6200852495364795,
-      "grad_norm": 0.003415314480662346,
-      "learning_rate": 0.00019999353572441108,
-      "loss": 46.0,
-      "step": 47348
-    },
-    {
-      "epoch": 3.6201617065198692,
-      "grad_norm": 0.0009169764234684408,
-      "learning_rate": 0.00019999353545129965,
-      "loss": 46.0,
-      "step": 47349
-    },
-    {
-      "epoch": 3.620238163503259,
-      "grad_norm": 0.0005590890068560839,
-      "learning_rate": 0.00019999353517818246,
-      "loss": 46.0,
-      "step": 47350
-    },
-    {
-      "epoch": 3.6203146204866488,
-      "grad_norm": 0.003532255766913295,
-      "learning_rate": 0.0001999935349050595,
-      "loss": 46.0,
-      "step": 47351
-    },
-    {
-      "epoch": 3.6203910774700385,
-      "grad_norm": 0.003182093845680356,
-      "learning_rate": 0.00019999353463193078,
-      "loss": 46.0,
-      "step": 47352
-    },
-    {
-      "epoch": 3.6204675344534283,
-      "grad_norm": 0.0024112078826874495,
-      "learning_rate": 0.00019999353435879626,
-      "loss": 46.0,
-      "step": 47353
-    },
-    {
-      "epoch": 3.6205439914368176,
-      "grad_norm": 0.0017347057582810521,
-      "learning_rate": 0.000199993534085656,
-      "loss": 46.0,
-      "step": 47354
-    },
-    {
-      "epoch": 3.6206204484202074,
-      "grad_norm": 0.0027756644412875175,
-      "learning_rate": 0.00019999353381251,
-      "loss": 46.0,
-      "step": 47355
-    },
-    {
-      "epoch": 3.620696905403597,
-      "grad_norm": 0.0011865616543218493,
-      "learning_rate": 0.00019999353353935816,
-      "loss": 46.0,
-      "step": 47356
-    },
-    {
-      "epoch": 3.620773362386987,
-      "grad_norm": 0.001097788568586111,
-      "learning_rate": 0.00019999353326620057,
-      "loss": 46.0,
-      "step": 47357
-    },
-    {
-      "epoch": 3.6208498193703766,
-      "grad_norm": 0.002234102226793766,
-      "learning_rate": 0.00019999353299303725,
-      "loss": 46.0,
-      "step": 47358
-    },
-    {
-      "epoch": 3.6209262763537664,
-      "grad_norm": 0.010361786931753159,
-      "learning_rate": 0.00019999353271986814,
-      "loss": 46.0,
-      "step": 47359
-    },
-    {
-      "epoch": 3.621002733337156,
-      "grad_norm": 0.000992064829915762,
-      "learning_rate": 0.00019999353244669324,
-      "loss": 46.0,
-      "step": 47360
-    },
-    {
-      "epoch": 3.621079190320546,
-      "grad_norm": 0.001721574692055583,
-      "learning_rate": 0.0001999935321735126,
-      "loss": 46.0,
-      "step": 47361
-    },
-    {
-      "epoch": 3.6211556473039357,
-      "grad_norm": 0.0018225802341476083,
-      "learning_rate": 0.0001999935319003262,
-      "loss": 46.0,
-      "step": 47362
-    },
-    {
-      "epoch": 3.6212321042873254,
-      "grad_norm": 0.00235338625498116,
-      "learning_rate": 0.000199993531627134,
-      "loss": 46.0,
-      "step": 47363
-    },
-    {
-      "epoch": 3.621308561270715,
-      "grad_norm": 0.004489402286708355,
-      "learning_rate": 0.00019999353135393604,
-      "loss": 46.0,
-      "step": 47364
-    },
-    {
-      "epoch": 3.621385018254105,
-      "grad_norm": 0.0009082783362828195,
-      "learning_rate": 0.00019999353108073232,
-      "loss": 46.0,
-      "step": 47365
-    },
-    {
-      "epoch": 3.6214614752374947,
-      "grad_norm": 0.0021544923074543476,
-      "learning_rate": 0.00019999353080752284,
-      "loss": 46.0,
-      "step": 47366
-    },
-    {
-      "epoch": 3.6215379322208845,
-      "grad_norm": 0.0024716483894735575,
-      "learning_rate": 0.00019999353053430755,
-      "loss": 46.0,
-      "step": 47367
-    },
-    {
-      "epoch": 3.621614389204274,
-      "grad_norm": 0.002472597872838378,
-      "learning_rate": 0.0001999935302610865,
-      "loss": 46.0,
-      "step": 47368
-    },
-    {
-      "epoch": 3.6216908461876636,
-      "grad_norm": 0.0011978407856076956,
-      "learning_rate": 0.0001999935299878597,
-      "loss": 46.0,
-      "step": 47369
-    },
-    {
-      "epoch": 3.6217673031710533,
-      "grad_norm": 0.0018157584127038717,
-      "learning_rate": 0.00019999352971462715,
-      "loss": 46.0,
-      "step": 47370
-    },
-    {
-      "epoch": 3.621843760154443,
-      "grad_norm": 0.0026981134433299303,
-      "learning_rate": 0.0001999935294413888,
-      "loss": 46.0,
-      "step": 47371
-    },
-    {
-      "epoch": 3.621920217137833,
-      "grad_norm": 0.006168703082948923,
-      "learning_rate": 0.0001999935291681447,
-      "loss": 46.0,
-      "step": 47372
-    },
-    {
-      "epoch": 3.6219966741212226,
-      "grad_norm": 0.004905191715806723,
-      "learning_rate": 0.00019999352889489482,
-      "loss": 46.0,
-      "step": 47373
-    },
-    {
-      "epoch": 3.6220731311046124,
-      "grad_norm": 0.0008568610646761954,
-      "learning_rate": 0.00019999352862163917,
-      "loss": 46.0,
-      "step": 47374
-    },
-    {
-      "epoch": 3.622149588088002,
-      "grad_norm": 0.0061667305417358875,
-      "learning_rate": 0.00019999352834837778,
-      "loss": 46.0,
-      "step": 47375
-    },
-    {
-      "epoch": 3.6222260450713915,
-      "grad_norm": 0.001092708669602871,
-      "learning_rate": 0.00019999352807511058,
-      "loss": 46.0,
-      "step": 47376
-    },
-    {
-      "epoch": 3.622302502054781,
-      "grad_norm": 0.0033057217951864004,
-      "learning_rate": 0.00019999352780183762,
-      "loss": 46.0,
-      "step": 47377
-    },
-    {
-      "epoch": 3.622378959038171,
-      "grad_norm": 0.0015536271966993809,
-      "learning_rate": 0.00019999352752855888,
-      "loss": 46.0,
-      "step": 47378
-    },
-    {
-      "epoch": 3.6224554160215607,
-      "grad_norm": 0.0029149481561034918,
-      "learning_rate": 0.0001999935272552744,
-      "loss": 46.0,
-      "step": 47379
-    },
-    {
-      "epoch": 3.6225318730049505,
-      "grad_norm": 0.0006690032314509153,
-      "learning_rate": 0.00019999352698198415,
-      "loss": 46.0,
-      "step": 47380
-    },
-    {
-      "epoch": 3.6226083299883403,
-      "grad_norm": 0.0007043190416879952,
-      "learning_rate": 0.0001999935267086881,
-      "loss": 46.0,
-      "step": 47381
-    },
-    {
-      "epoch": 3.62268478697173,
-      "grad_norm": 0.002085176995024085,
-      "learning_rate": 0.0001999935264353863,
-      "loss": 46.0,
-      "step": 47382
-    },
-    {
-      "epoch": 3.62276124395512,
-      "grad_norm": 0.0016610943712294102,
-      "learning_rate": 0.00019999352616207873,
-      "loss": 46.0,
-      "step": 47383
-    },
-    {
-      "epoch": 3.6228377009385095,
-      "grad_norm": 0.00451140571385622,
-      "learning_rate": 0.00019999352588876538,
-      "loss": 46.0,
-      "step": 47384
-    },
-    {
-      "epoch": 3.6229141579218993,
-      "grad_norm": 0.005642641335725784,
-      "learning_rate": 0.00019999352561544628,
-      "loss": 46.0,
-      "step": 47385
-    },
-    {
-      "epoch": 3.622990614905289,
-      "grad_norm": 0.0004088491841685027,
-      "learning_rate": 0.0001999935253421214,
-      "loss": 46.0,
-      "step": 47386
-    },
-    {
-      "epoch": 3.623067071888679,
-      "grad_norm": 0.001353420433588326,
-      "learning_rate": 0.00019999352506879075,
-      "loss": 46.0,
-      "step": 47387
-    },
-    {
-      "epoch": 3.6231435288720686,
-      "grad_norm": 0.0024752598255872726,
-      "learning_rate": 0.00019999352479545433,
-      "loss": 46.0,
-      "step": 47388
-    },
-    {
-      "epoch": 3.6232199858554583,
-      "grad_norm": 0.001023277174681425,
-      "learning_rate": 0.00019999352452211214,
-      "loss": 46.0,
-      "step": 47389
-    },
-    {
-      "epoch": 3.6232964428388477,
-      "grad_norm": 0.00214571924880147,
-      "learning_rate": 0.0001999935242487642,
-      "loss": 46.0,
-      "step": 47390
-    },
-    {
-      "epoch": 3.6233728998222374,
-      "grad_norm": 0.0027552112005650997,
-      "learning_rate": 0.00019999352397541049,
-      "loss": 46.0,
-      "step": 47391
-    },
-    {
-      "epoch": 3.623449356805627,
-      "grad_norm": 0.0009797487873584032,
-      "learning_rate": 0.000199993523702051,
-      "loss": 46.0,
-      "step": 47392
-    },
-    {
-      "epoch": 3.623525813789017,
-      "grad_norm": 0.0008931035408750176,
-      "learning_rate": 0.00019999352342868574,
-      "loss": 46.0,
-      "step": 47393
-    },
-    {
-      "epoch": 3.6236022707724067,
-      "grad_norm": 0.0010882462374866009,
-      "learning_rate": 0.00019999352315531473,
-      "loss": 46.0,
-      "step": 47394
-    },
-    {
-      "epoch": 3.6236787277557965,
-      "grad_norm": 0.00335419038310647,
-      "learning_rate": 0.00019999352288193792,
-      "loss": 46.0,
-      "step": 47395
-    },
-    {
-      "epoch": 3.6237551847391862,
-      "grad_norm": 0.002310057869181037,
-      "learning_rate": 0.00019999352260855532,
-      "loss": 46.0,
-      "step": 47396
-    },
-    {
-      "epoch": 3.623831641722576,
-      "grad_norm": 0.0027128912042826414,
-      "learning_rate": 0.000199993522335167,
-      "loss": 46.0,
-      "step": 47397
-    },
-    {
-      "epoch": 3.6239080987059653,
-      "grad_norm": 0.0008993922383524477,
-      "learning_rate": 0.0001999935220617729,
-      "loss": 46.0,
-      "step": 47398
-    },
-    {
-      "epoch": 3.623984555689355,
-      "grad_norm": 0.0014454993652179837,
-      "learning_rate": 0.00019999352178837301,
-      "loss": 46.0,
-      "step": 47399
-    },
-    {
-      "epoch": 3.624061012672745,
-      "grad_norm": 0.0019481470808386803,
-      "learning_rate": 0.0001999935215149674,
-      "loss": 46.0,
-      "step": 47400
-    },
-    {
-      "epoch": 3.6241374696561346,
-      "grad_norm": 0.0005018050433136523,
-      "learning_rate": 0.00019999352124155597,
-      "loss": 46.0,
-      "step": 47401
-    },
-    {
-      "epoch": 3.6242139266395244,
-      "grad_norm": 0.0015739724040031433,
-      "learning_rate": 0.00019999352096813878,
-      "loss": 46.0,
-      "step": 47402
-    },
-    {
-      "epoch": 3.624290383622914,
-      "grad_norm": 0.001120415166951716,
-      "learning_rate": 0.00019999352069471584,
-      "loss": 46.0,
-      "step": 47403
-    },
-    {
-      "epoch": 3.624366840606304,
-      "grad_norm": 0.0022295115049928427,
-      "learning_rate": 0.00019999352042128713,
-      "loss": 46.0,
-      "step": 47404
-    },
-    {
-      "epoch": 3.6244432975896936,
-      "grad_norm": 0.0010030489647760987,
-      "learning_rate": 0.0001999935201478526,
-      "loss": 46.0,
-      "step": 47405
-    },
-    {
-      "epoch": 3.6245197545730834,
-      "grad_norm": 0.004237939137965441,
-      "learning_rate": 0.00019999351987441238,
-      "loss": 46.0,
-      "step": 47406
-    },
-    {
-      "epoch": 3.624596211556473,
-      "grad_norm": 0.0005940491100773215,
-      "learning_rate": 0.00019999351960096635,
-      "loss": 46.0,
-      "step": 47407
-    },
-    {
-      "epoch": 3.624672668539863,
-      "grad_norm": 0.004383536987006664,
-      "learning_rate": 0.00019999351932751457,
-      "loss": 46.0,
-      "step": 47408
-    },
-    {
-      "epoch": 3.6247491255232527,
-      "grad_norm": 0.0022701299749314785,
-      "learning_rate": 0.00019999351905405699,
-      "loss": 46.0,
-      "step": 47409
-    },
-    {
-      "epoch": 3.6248255825066424,
-      "grad_norm": 0.001550098997540772,
-      "learning_rate": 0.00019999351878059366,
-      "loss": 46.0,
-      "step": 47410
-    },
-    {
-      "epoch": 3.6249020394900318,
-      "grad_norm": 0.0022231393959373236,
-      "learning_rate": 0.00019999351850712453,
-      "loss": 46.0,
-      "step": 47411
-    },
-    {
-      "epoch": 3.6249784964734215,
-      "grad_norm": 0.0017720917239785194,
-      "learning_rate": 0.00019999351823364966,
-      "loss": 46.0,
-      "step": 47412
-    },
-    {
-      "epoch": 3.6250549534568113,
-      "grad_norm": 0.0025821721646934748,
-      "learning_rate": 0.00019999351796016904,
-      "loss": 46.0,
-      "step": 47413
-    },
-    {
-      "epoch": 3.625131410440201,
-      "grad_norm": 0.0030009495094418526,
-      "learning_rate": 0.00019999351768668262,
-      "loss": 46.0,
-      "step": 47414
-    },
-    {
-      "epoch": 3.625207867423591,
-      "grad_norm": 0.0015778911765664816,
-      "learning_rate": 0.00019999351741319043,
-      "loss": 46.0,
-      "step": 47415
-    },
-    {
-      "epoch": 3.6252843244069806,
-      "grad_norm": 0.0024254783056676388,
-      "learning_rate": 0.00019999351713969249,
-      "loss": 46.0,
-      "step": 47416
-    },
-    {
-      "epoch": 3.6253607813903703,
-      "grad_norm": 0.0014715736033394933,
-      "learning_rate": 0.00019999351686618877,
-      "loss": 46.0,
-      "step": 47417
-    },
-    {
-      "epoch": 3.62543723837376,
-      "grad_norm": 0.0008359206258319318,
-      "learning_rate": 0.00019999351659267926,
-      "loss": 46.0,
-      "step": 47418
-    },
-    {
-      "epoch": 3.62551369535715,
-      "grad_norm": 0.0017035825876519084,
-      "learning_rate": 0.000199993516319164,
-      "loss": 46.0,
-      "step": 47419
-    },
-    {
-      "epoch": 3.625590152340539,
-      "grad_norm": 0.0028160554356873035,
-      "learning_rate": 0.000199993516045643,
-      "loss": 46.0,
-      "step": 47420
-    },
-    {
-      "epoch": 3.625666609323929,
-      "grad_norm": 0.004391511902213097,
-      "learning_rate": 0.0001999935157721162,
-      "loss": 46.0,
-      "step": 47421
-    },
-    {
-      "epoch": 3.6257430663073187,
-      "grad_norm": 0.001714915968477726,
-      "learning_rate": 0.00019999351549858366,
-      "loss": 46.0,
-      "step": 47422
-    },
-    {
-      "epoch": 3.6258195232907084,
-      "grad_norm": 0.001265192055143416,
-      "learning_rate": 0.0001999935152250453,
-      "loss": 46.0,
-      "step": 47423
-    },
-    {
-      "epoch": 3.625895980274098,
-      "grad_norm": 0.003409244120121002,
-      "learning_rate": 0.0001999935149515012,
-      "loss": 46.0,
-      "step": 47424
-    },
-    {
-      "epoch": 3.625972437257488,
-      "grad_norm": 0.0012118630111217499,
-      "learning_rate": 0.00019999351467795133,
-      "loss": 46.0,
-      "step": 47425
-    },
-    {
-      "epoch": 3.6260488942408777,
-      "grad_norm": 0.003979481756687164,
-      "learning_rate": 0.00019999351440439571,
-      "loss": 46.0,
-      "step": 47426
-    },
-    {
-      "epoch": 3.6261253512242675,
-      "grad_norm": 0.001014834619127214,
-      "learning_rate": 0.00019999351413083427,
-      "loss": 46.0,
-      "step": 47427
-    },
-    {
-      "epoch": 3.6262018082076573,
-      "grad_norm": 0.0011285807704553008,
-      "learning_rate": 0.0001999935138572671,
-      "loss": 46.0,
-      "step": 47428
-    },
-    {
-      "epoch": 3.626278265191047,
-      "grad_norm": 0.005313312169164419,
-      "learning_rate": 0.00019999351358369416,
-      "loss": 46.0,
-      "step": 47429
-    },
-    {
-      "epoch": 3.6263547221744368,
-      "grad_norm": 0.0035176228266209364,
-      "learning_rate": 0.00019999351331011545,
-      "loss": 46.0,
-      "step": 47430
-    },
-    {
-      "epoch": 3.6264311791578265,
-      "grad_norm": 0.000344529515132308,
-      "learning_rate": 0.00019999351303653096,
-      "loss": 46.0,
-      "step": 47431
-    },
-    {
-      "epoch": 3.6265076361412163,
-      "grad_norm": 0.0022166178096085787,
-      "learning_rate": 0.0001999935127629407,
-      "loss": 46.0,
-      "step": 47432
-    },
-    {
-      "epoch": 3.6265840931246056,
-      "grad_norm": 0.0021415643859654665,
-      "learning_rate": 0.00019999351248934467,
-      "loss": 46.0,
-      "step": 47433
-    },
-    {
-      "epoch": 3.6266605501079954,
-      "grad_norm": 0.002091477857902646,
-      "learning_rate": 0.0001999935122157429,
-      "loss": 46.0,
-      "step": 47434
-    },
-    {
-      "epoch": 3.626737007091385,
-      "grad_norm": 0.001383824273943901,
-      "learning_rate": 0.0001999935119421353,
-      "loss": 46.0,
-      "step": 47435
-    },
-    {
-      "epoch": 3.626813464074775,
-      "grad_norm": 0.0009950530948117375,
-      "learning_rate": 0.000199993511668522,
-      "loss": 46.0,
-      "step": 47436
-    },
-    {
-      "epoch": 3.6268899210581647,
-      "grad_norm": 0.002144841942936182,
-      "learning_rate": 0.00019999351139490289,
-      "loss": 46.0,
-      "step": 47437
-    },
-    {
-      "epoch": 3.6269663780415544,
-      "grad_norm": 0.0005444727721624076,
-      "learning_rate": 0.00019999351112127801,
-      "loss": 46.0,
-      "step": 47438
-    },
-    {
-      "epoch": 3.627042835024944,
-      "grad_norm": 0.0005735830636695027,
-      "learning_rate": 0.0001999935108476474,
-      "loss": 46.0,
-      "step": 47439
-    },
-    {
-      "epoch": 3.627119292008334,
-      "grad_norm": 0.0017893824260681868,
-      "learning_rate": 0.000199993510574011,
-      "loss": 46.0,
-      "step": 47440
-    },
-    {
-      "epoch": 3.6271957489917233,
-      "grad_norm": 0.0019484329968690872,
-      "learning_rate": 0.0001999935103003688,
-      "loss": 46.0,
-      "step": 47441
-    },
-    {
-      "epoch": 3.627272205975113,
-      "grad_norm": 0.0023450450971722603,
-      "learning_rate": 0.00019999351002672087,
-      "loss": 46.0,
-      "step": 47442
-    },
-    {
-      "epoch": 3.627348662958503,
-      "grad_norm": 0.0033868427854031324,
-      "learning_rate": 0.00019999350975306713,
-      "loss": 46.0,
-      "step": 47443
-    },
-    {
-      "epoch": 3.6274251199418925,
-      "grad_norm": 0.002973944880068302,
-      "learning_rate": 0.00019999350947940764,
-      "loss": 46.0,
-      "step": 47444
-    },
-    {
-      "epoch": 3.6275015769252823,
-      "grad_norm": 0.00456279655918479,
-      "learning_rate": 0.0001999935092057424,
-      "loss": 46.0,
-      "step": 47445
-    },
-    {
-      "epoch": 3.627578033908672,
-      "grad_norm": 0.0017020419472828507,
-      "learning_rate": 0.00019999350893207138,
-      "loss": 46.0,
-      "step": 47446
-    },
-    {
-      "epoch": 3.627654490892062,
-      "grad_norm": 0.0032190780621021986,
-      "learning_rate": 0.0001999935086583946,
-      "loss": 46.0,
-      "step": 47447
-    },
-    {
-      "epoch": 3.6277309478754516,
-      "grad_norm": 0.0026783659122884274,
-      "learning_rate": 0.00019999350838471202,
-      "loss": 46.0,
-      "step": 47448
-    },
-    {
-      "epoch": 3.6278074048588413,
-      "grad_norm": 0.0017374379094690084,
-      "learning_rate": 0.0001999935081110237,
-      "loss": 46.0,
-      "step": 47449
-    },
-    {
-      "epoch": 3.627883861842231,
-      "grad_norm": 0.0013571102172136307,
-      "learning_rate": 0.00019999350783732962,
-      "loss": 46.0,
-      "step": 47450
-    },
-    {
-      "epoch": 3.627960318825621,
-      "grad_norm": 0.00201827147975564,
-      "learning_rate": 0.00019999350756362975,
-      "loss": 46.0,
-      "step": 47451
-    },
-    {
-      "epoch": 3.6280367758090106,
-      "grad_norm": 0.0017024754779413342,
-      "learning_rate": 0.00019999350728992413,
-      "loss": 46.0,
-      "step": 47452
-    },
-    {
-      "epoch": 3.6281132327924004,
-      "grad_norm": 0.004810396581888199,
-      "learning_rate": 0.0001999935070162127,
-      "loss": 46.0,
-      "step": 47453
-    },
-    {
-      "epoch": 3.62818968977579,
-      "grad_norm": 0.002099544508382678,
-      "learning_rate": 0.00019999350674249552,
-      "loss": 46.0,
-      "step": 47454
-    },
-    {
-      "epoch": 3.6282661467591795,
-      "grad_norm": 0.002537505468353629,
-      "learning_rate": 0.00019999350646877258,
-      "loss": 46.0,
-      "step": 47455
-    },
-    {
-      "epoch": 3.6283426037425692,
-      "grad_norm": 0.004741470795124769,
-      "learning_rate": 0.0001999935061950439,
-      "loss": 46.0,
-      "step": 47456
-    },
-    {
-      "epoch": 3.628419060725959,
-      "grad_norm": 0.0015688469866290689,
-      "learning_rate": 0.00019999350592130938,
-      "loss": 46.0,
-      "step": 47457
-    },
-    {
-      "epoch": 3.6284955177093487,
-      "grad_norm": 0.002859345404431224,
-      "learning_rate": 0.00019999350564756915,
-      "loss": 46.0,
-      "step": 47458
-    },
-    {
-      "epoch": 3.6285719746927385,
-      "grad_norm": 0.007290893699973822,
-      "learning_rate": 0.00019999350537382312,
-      "loss": 46.0,
-      "step": 47459
-    },
-    {
-      "epoch": 3.6286484316761283,
-      "grad_norm": 0.003457999788224697,
-      "learning_rate": 0.00019999350510007134,
-      "loss": 46.0,
-      "step": 47460
-    },
-    {
-      "epoch": 3.628724888659518,
-      "grad_norm": 0.0011462683323770761,
-      "learning_rate": 0.0001999935048263138,
-      "loss": 46.0,
-      "step": 47461
-    },
-    {
-      "epoch": 3.628801345642908,
-      "grad_norm": 0.001098414184525609,
-      "learning_rate": 0.00019999350455255046,
-      "loss": 46.0,
-      "step": 47462
-    },
-    {
-      "epoch": 3.628877802626297,
-      "grad_norm": 0.0017879842780530453,
-      "learning_rate": 0.00019999350427878137,
-      "loss": 46.0,
-      "step": 47463
-    },
-    {
-      "epoch": 3.628954259609687,
-      "grad_norm": 0.0018215954769402742,
-      "learning_rate": 0.0001999935040050065,
-      "loss": 46.0,
-      "step": 47464
-    },
-    {
-      "epoch": 3.6290307165930766,
-      "grad_norm": 0.0012119802413508296,
-      "learning_rate": 0.00019999350373122585,
-      "loss": 46.0,
-      "step": 47465
-    },
-    {
-      "epoch": 3.6291071735764664,
-      "grad_norm": 0.001823908882215619,
-      "learning_rate": 0.00019999350345743943,
-      "loss": 46.0,
-      "step": 47466
-    },
-    {
-      "epoch": 3.629183630559856,
-      "grad_norm": 0.006939653307199478,
-      "learning_rate": 0.0001999935031836473,
-      "loss": 46.0,
-      "step": 47467
-    },
-    {
-      "epoch": 3.629260087543246,
-      "grad_norm": 0.002816864987835288,
-      "learning_rate": 0.00019999350290984932,
-      "loss": 46.0,
-      "step": 47468
-    },
-    {
-      "epoch": 3.6293365445266357,
-      "grad_norm": 0.002402219455689192,
-      "learning_rate": 0.00019999350263604564,
-      "loss": 46.0,
-      "step": 47469
-    },
-    {
-      "epoch": 3.6294130015100254,
-      "grad_norm": 0.006036390550434589,
-      "learning_rate": 0.00019999350236223615,
-      "loss": 46.0,
-      "step": 47470
-    },
-    {
-      "epoch": 3.629489458493415,
-      "grad_norm": 0.0014716449659317732,
-      "learning_rate": 0.0001999935020884209,
-      "loss": 46.0,
-      "step": 47471
-    },
-    {
-      "epoch": 3.629565915476805,
-      "grad_norm": 0.0023648792412132025,
-      "learning_rate": 0.0001999935018145999,
-      "loss": 46.0,
-      "step": 47472
-    },
-    {
-      "epoch": 3.6296423724601947,
-      "grad_norm": 0.0016436890000477433,
-      "learning_rate": 0.00019999350154077309,
-      "loss": 46.0,
-      "step": 47473
-    },
-    {
-      "epoch": 3.6297188294435845,
-      "grad_norm": 0.0008651686948724091,
-      "learning_rate": 0.00019999350126694053,
-      "loss": 46.0,
-      "step": 47474
-    },
-    {
-      "epoch": 3.6297952864269742,
-      "grad_norm": 0.0029481409583240747,
-      "learning_rate": 0.0001999935009931022,
-      "loss": 46.0,
-      "step": 47475
-    },
-    {
-      "epoch": 3.629871743410364,
-      "grad_norm": 0.003094090148806572,
-      "learning_rate": 0.0001999935007192581,
-      "loss": 46.0,
-      "step": 47476
-    },
-    {
-      "epoch": 3.6299482003937533,
-      "grad_norm": 0.0019815596751868725,
-      "learning_rate": 0.00019999350044540827,
-      "loss": 46.0,
-      "step": 47477
-    },
-    {
-      "epoch": 3.630024657377143,
-      "grad_norm": 0.0021683755330741405,
-      "learning_rate": 0.00019999350017155262,
-      "loss": 46.0,
-      "step": 47478
-    },
-    {
-      "epoch": 3.630101114360533,
-      "grad_norm": 0.0012911329977214336,
-      "learning_rate": 0.00019999349989769123,
-      "loss": 46.0,
-      "step": 47479
-    },
-    {
-      "epoch": 3.6301775713439226,
-      "grad_norm": 0.0007185947615653276,
-      "learning_rate": 0.00019999349962382406,
-      "loss": 46.0,
-      "step": 47480
-    },
-    {
-      "epoch": 3.6302540283273124,
-      "grad_norm": 0.00220209127292037,
-      "learning_rate": 0.00019999349934995113,
-      "loss": 46.0,
-      "step": 47481
-    },
-    {
-      "epoch": 3.630330485310702,
-      "grad_norm": 0.0028558720368891954,
-      "learning_rate": 0.00019999349907607241,
-      "loss": 46.0,
-      "step": 47482
-    },
-    {
-      "epoch": 3.630406942294092,
-      "grad_norm": 0.0050702812150120735,
-      "learning_rate": 0.00019999349880218793,
-      "loss": 46.0,
-      "step": 47483
-    },
-    {
-      "epoch": 3.6304833992774816,
-      "grad_norm": 0.003349294187501073,
-      "learning_rate": 0.00019999349852829767,
-      "loss": 46.0,
-      "step": 47484
-    },
-    {
-      "epoch": 3.630559856260871,
-      "grad_norm": 0.002493863692507148,
-      "learning_rate": 0.00019999349825440167,
-      "loss": 46.0,
-      "step": 47485
-    },
-    {
-      "epoch": 3.6306363132442607,
-      "grad_norm": 0.001305361744016409,
-      "learning_rate": 0.00019999349798049986,
-      "loss": 46.0,
-      "step": 47486
-    },
-    {
-      "epoch": 3.6307127702276505,
-      "grad_norm": 0.004744191188365221,
-      "learning_rate": 0.00019999349770659234,
-      "loss": 46.0,
-      "step": 47487
-    },
-    {
-      "epoch": 3.6307892272110402,
-      "grad_norm": 0.002669707639142871,
-      "learning_rate": 0.000199993497432679,
-      "loss": 46.0,
-      "step": 47488
-    },
-    {
-      "epoch": 3.63086568419443,
-      "grad_norm": 0.0025324500165879726,
-      "learning_rate": 0.00019999349715875991,
-      "loss": 46.0,
-      "step": 47489
-    },
-    {
-      "epoch": 3.6309421411778198,
-      "grad_norm": 0.003164202207699418,
-      "learning_rate": 0.00019999349688483507,
-      "loss": 46.0,
-      "step": 47490
-    },
-    {
-      "epoch": 3.6310185981612095,
-      "grad_norm": 0.0013077250914648175,
-      "learning_rate": 0.0001999934966109044,
-      "loss": 46.0,
-      "step": 47491
-    },
-    {
-      "epoch": 3.6310950551445993,
-      "grad_norm": 0.00341529562138021,
-      "learning_rate": 0.00019999349633696803,
-      "loss": 46.0,
-      "step": 47492
-    },
-    {
-      "epoch": 3.631171512127989,
-      "grad_norm": 0.0007273780065588653,
-      "learning_rate": 0.00019999349606302587,
-      "loss": 46.0,
-      "step": 47493
-    },
-    {
-      "epoch": 3.631247969111379,
-      "grad_norm": 0.0024748428259044886,
-      "learning_rate": 0.0001999934957890779,
-      "loss": 46.0,
-      "step": 47494
-    },
-    {
-      "epoch": 3.6313244260947686,
-      "grad_norm": 0.005123466718941927,
-      "learning_rate": 0.00019999349551512422,
-      "loss": 46.0,
-      "step": 47495
-    },
-    {
-      "epoch": 3.6314008830781583,
-      "grad_norm": 0.0037410948425531387,
-      "learning_rate": 0.00019999349524116474,
-      "loss": 46.0,
-      "step": 47496
-    },
-    {
-      "epoch": 3.631477340061548,
-      "grad_norm": 0.0027024236042052507,
-      "learning_rate": 0.0001999934949671995,
-      "loss": 46.0,
-      "step": 47497
-    },
-    {
-      "epoch": 3.631553797044938,
-      "grad_norm": 0.0018701617373153567,
-      "learning_rate": 0.00019999349469322847,
-      "loss": 46.0,
-      "step": 47498
-    },
-    {
-      "epoch": 3.631630254028327,
-      "grad_norm": 0.0011156498221680522,
-      "learning_rate": 0.00019999349441925167,
-      "loss": 46.0,
-      "step": 47499
-    },
-    {
-      "epoch": 3.631706711011717,
-      "grad_norm": 0.004159875214099884,
-      "learning_rate": 0.00019999349414526914,
-      "loss": 46.0,
-      "step": 47500
-    },
-    {
-      "epoch": 3.6317831679951067,
-      "grad_norm": 0.0006476440466940403,
-      "learning_rate": 0.00019999349387128082,
-      "loss": 46.0,
-      "step": 47501
-    },
-    {
-      "epoch": 3.6318596249784965,
-      "grad_norm": 0.0021001342684030533,
-      "learning_rate": 0.00019999349359728675,
-      "loss": 46.0,
-      "step": 47502
-    },
-    {
-      "epoch": 3.631936081961886,
-      "grad_norm": 0.0014773018192499876,
-      "learning_rate": 0.00019999349332328688,
-      "loss": 46.0,
-      "step": 47503
-    },
-    {
-      "epoch": 3.632012538945276,
-      "grad_norm": 0.0015500617446377873,
-      "learning_rate": 0.00019999349304928123,
-      "loss": 46.0,
-      "step": 47504
-    },
-    {
-      "epoch": 3.6320889959286657,
-      "grad_norm": 0.0035139955580234528,
-      "learning_rate": 0.00019999349277526984,
-      "loss": 46.0,
-      "step": 47505
-    },
-    {
-      "epoch": 3.6321654529120555,
-      "grad_norm": 0.00399654870852828,
-      "learning_rate": 0.00019999349250125268,
-      "loss": 46.0,
-      "step": 47506
-    },
-    {
-      "epoch": 3.632241909895445,
-      "grad_norm": 0.00333567732013762,
-      "learning_rate": 0.00019999349222722974,
-      "loss": 46.0,
-      "step": 47507
-    },
-    {
-      "epoch": 3.6323183668788346,
-      "grad_norm": 0.004286587238311768,
-      "learning_rate": 0.00019999349195320103,
-      "loss": 46.0,
-      "step": 47508
-    },
-    {
-      "epoch": 3.6323948238622243,
-      "grad_norm": 0.002665396546944976,
-      "learning_rate": 0.00019999349167916657,
-      "loss": 46.0,
-      "step": 47509
-    },
-    {
-      "epoch": 3.632471280845614,
-      "grad_norm": 0.002656772034242749,
-      "learning_rate": 0.00019999349140512634,
-      "loss": 46.0,
-      "step": 47510
-    },
-    {
-      "epoch": 3.632547737829004,
-      "grad_norm": 0.0005230242386460304,
-      "learning_rate": 0.0001999934911310803,
-      "loss": 46.0,
-      "step": 47511
-    },
-    {
-      "epoch": 3.6326241948123936,
-      "grad_norm": 0.0030034389346837997,
-      "learning_rate": 0.00019999349085702853,
-      "loss": 46.0,
-      "step": 47512
-    },
-    {
-      "epoch": 3.6327006517957834,
-      "grad_norm": 0.0034105100203305483,
-      "learning_rate": 0.00019999349058297096,
-      "loss": 46.0,
-      "step": 47513
-    },
-    {
-      "epoch": 3.632777108779173,
-      "grad_norm": 0.0009571025730110705,
-      "learning_rate": 0.00019999349030890766,
-      "loss": 46.0,
-      "step": 47514
-    },
-    {
-      "epoch": 3.632853565762563,
-      "grad_norm": 0.0017008426366373897,
-      "learning_rate": 0.0001999934900348386,
-      "loss": 46.0,
-      "step": 47515
-    },
-    {
-      "epoch": 3.6329300227459527,
-      "grad_norm": 0.0026454338803887367,
-      "learning_rate": 0.00019999348976076372,
-      "loss": 46.0,
-      "step": 47516
-    },
-    {
-      "epoch": 3.6330064797293424,
-      "grad_norm": 0.000751490646507591,
-      "learning_rate": 0.0001999934894866831,
-      "loss": 46.0,
-      "step": 47517
-    },
-    {
-      "epoch": 3.633082936712732,
-      "grad_norm": 0.0021533139515668154,
-      "learning_rate": 0.0001999934892125967,
-      "loss": 46.0,
-      "step": 47518
-    },
-    {
-      "epoch": 3.633159393696122,
-      "grad_norm": 0.005315329413861036,
-      "learning_rate": 0.00019999348893850455,
-      "loss": 46.0,
-      "step": 47519
-    },
-    {
-      "epoch": 3.6332358506795117,
-      "grad_norm": 0.0015597057063132524,
-      "learning_rate": 0.00019999348866440658,
-      "loss": 46.0,
-      "step": 47520
-    },
-    {
-      "epoch": 3.633312307662901,
-      "grad_norm": 0.002478837501257658,
-      "learning_rate": 0.0001999934883903029,
-      "loss": 46.0,
-      "step": 47521
-    },
-    {
-      "epoch": 3.633388764646291,
-      "grad_norm": 0.004642905667424202,
-      "learning_rate": 0.00019999348811619344,
-      "loss": 46.0,
-      "step": 47522
-    },
-    {
-      "epoch": 3.6334652216296806,
-      "grad_norm": 0.0024296757765114307,
-      "learning_rate": 0.00019999348784207818,
-      "loss": 46.0,
-      "step": 47523
-    },
-    {
-      "epoch": 3.6335416786130703,
-      "grad_norm": 0.0011710509425029159,
-      "learning_rate": 0.00019999348756795718,
-      "loss": 46.0,
-      "step": 47524
-    },
-    {
-      "epoch": 3.63361813559646,
-      "grad_norm": 0.004194037988781929,
-      "learning_rate": 0.0001999934872938304,
-      "loss": 46.0,
-      "step": 47525
-    },
-    {
-      "epoch": 3.63369459257985,
-      "grad_norm": 0.0045677851885557175,
-      "learning_rate": 0.00019999348701969785,
-      "loss": 46.0,
-      "step": 47526
-    },
-    {
-      "epoch": 3.6337710495632396,
-      "grad_norm": 0.004974625073373318,
-      "learning_rate": 0.00019999348674555953,
-      "loss": 46.0,
-      "step": 47527
-    },
-    {
-      "epoch": 3.6338475065466294,
-      "grad_norm": 0.00275798118673265,
-      "learning_rate": 0.00019999348647141546,
-      "loss": 46.0,
-      "step": 47528
-    },
-    {
-      "epoch": 3.6339239635300187,
-      "grad_norm": 0.0030305704567581415,
-      "learning_rate": 0.0001999934861972656,
-      "loss": 46.0,
-      "step": 47529
-    },
-    {
-      "epoch": 3.6340004205134084,
-      "grad_norm": 0.0034770227503031492,
-      "learning_rate": 0.00019999348592310997,
-      "loss": 46.0,
-      "step": 47530
-    },
-    {
-      "epoch": 3.634076877496798,
-      "grad_norm": 0.0004107113054487854,
-      "learning_rate": 0.00019999348564894858,
-      "loss": 46.0,
-      "step": 47531
-    },
-    {
-      "epoch": 3.634153334480188,
-      "grad_norm": 0.008126107044517994,
-      "learning_rate": 0.0001999934853747814,
-      "loss": 46.0,
-      "step": 47532
-    },
-    {
-      "epoch": 3.6342297914635777,
-      "grad_norm": 0.003024878678843379,
-      "learning_rate": 0.00019999348510060848,
-      "loss": 46.0,
-      "step": 47533
-    },
-    {
-      "epoch": 3.6343062484469675,
-      "grad_norm": 0.0024242375511676073,
-      "learning_rate": 0.00019999348482642977,
-      "loss": 46.0,
-      "step": 47534
-    },
-    {
-      "epoch": 3.6343827054303572,
-      "grad_norm": 0.0027742113452404737,
-      "learning_rate": 0.00019999348455224531,
-      "loss": 46.0,
-      "step": 47535
-    },
-    {
-      "epoch": 3.634459162413747,
-      "grad_norm": 0.0021328695584088564,
-      "learning_rate": 0.00019999348427805506,
-      "loss": 46.0,
-      "step": 47536
-    },
-    {
-      "epoch": 3.6345356193971368,
-      "grad_norm": 0.000886952446307987,
-      "learning_rate": 0.00019999348400385905,
-      "loss": 46.0,
-      "step": 47537
-    },
-    {
-      "epoch": 3.6346120763805265,
-      "grad_norm": 0.0035439415369182825,
-      "learning_rate": 0.00019999348372965728,
-      "loss": 46.0,
-      "step": 47538
-    },
-    {
-      "epoch": 3.6346885333639163,
-      "grad_norm": 0.0007310874061658978,
-      "learning_rate": 0.00019999348345544973,
-      "loss": 46.0,
-      "step": 47539
-    },
-    {
-      "epoch": 3.634764990347306,
-      "grad_norm": 0.0015476186526939273,
-      "learning_rate": 0.00019999348318123643,
-      "loss": 46.0,
-      "step": 47540
-    },
-    {
-      "epoch": 3.634841447330696,
-      "grad_norm": 0.0016074333107098937,
-      "learning_rate": 0.00019999348290701733,
-      "loss": 46.0,
-      "step": 47541
-    },
-    {
-      "epoch": 3.634917904314085,
-      "grad_norm": 0.0038941458333283663,
-      "learning_rate": 0.0001999934826327925,
-      "loss": 46.0,
-      "step": 47542
-    },
-    {
-      "epoch": 3.634994361297475,
-      "grad_norm": 0.0008792912121862173,
-      "learning_rate": 0.00019999348235856185,
-      "loss": 46.0,
-      "step": 47543
-    },
-    {
-      "epoch": 3.6350708182808646,
-      "grad_norm": 0.0038506274577230215,
-      "learning_rate": 0.00019999348208432548,
-      "loss": 46.0,
-      "step": 47544
-    },
-    {
-      "epoch": 3.6351472752642544,
-      "grad_norm": 0.0014361513312906027,
-      "learning_rate": 0.0001999934818100833,
-      "loss": 46.0,
-      "step": 47545
-    },
-    {
-      "epoch": 3.635223732247644,
-      "grad_norm": 0.0056853280402719975,
-      "learning_rate": 0.00019999348153583538,
-      "loss": 46.0,
-      "step": 47546
-    },
-    {
-      "epoch": 3.635300189231034,
-      "grad_norm": 0.0020093037746846676,
-      "learning_rate": 0.00019999348126158167,
-      "loss": 46.0,
-      "step": 47547
-    },
-    {
-      "epoch": 3.6353766462144237,
-      "grad_norm": 0.002617260441184044,
-      "learning_rate": 0.00019999348098732222,
-      "loss": 46.0,
-      "step": 47548
-    },
-    {
-      "epoch": 3.6354531031978135,
-      "grad_norm": 0.0013669480103999376,
-      "learning_rate": 0.00019999348071305696,
-      "loss": 46.0,
-      "step": 47549
-    },
-    {
-      "epoch": 3.635529560181203,
-      "grad_norm": 0.0016319206915795803,
-      "learning_rate": 0.00019999348043878596,
-      "loss": 46.0,
-      "step": 47550
-    },
-    {
-      "epoch": 3.6356060171645925,
-      "grad_norm": 0.0010099452920258045,
-      "learning_rate": 0.0001999934801645092,
-      "loss": 46.0,
-      "step": 47551
-    },
-    {
-      "epoch": 3.6356824741479823,
-      "grad_norm": 0.0034300254192203283,
-      "learning_rate": 0.00019999347989022666,
-      "loss": 46.0,
-      "step": 47552
-    },
-    {
-      "epoch": 3.635758931131372,
-      "grad_norm": 0.003304105019196868,
-      "learning_rate": 0.00019999347961593836,
-      "loss": 46.0,
-      "step": 47553
-    },
-    {
-      "epoch": 3.635835388114762,
-      "grad_norm": 0.002226894721388817,
-      "learning_rate": 0.00019999347934164424,
-      "loss": 46.0,
-      "step": 47554
-    },
-    {
-      "epoch": 3.6359118450981516,
-      "grad_norm": 0.0008436719072051346,
-      "learning_rate": 0.00019999347906734442,
-      "loss": 46.0,
-      "step": 47555
-    },
-    {
-      "epoch": 3.6359883020815413,
-      "grad_norm": 0.0019015672150999308,
-      "learning_rate": 0.00019999347879303878,
-      "loss": 46.0,
-      "step": 47556
-    },
-    {
-      "epoch": 3.636064759064931,
-      "grad_norm": 0.005953916814178228,
-      "learning_rate": 0.00019999347851872742,
-      "loss": 46.0,
-      "step": 47557
-    },
-    {
-      "epoch": 3.636141216048321,
-      "grad_norm": 0.004504180513322353,
-      "learning_rate": 0.00019999347824441025,
-      "loss": 46.0,
-      "step": 47558
-    },
-    {
-      "epoch": 3.6362176730317106,
-      "grad_norm": 0.0038886885158717632,
-      "learning_rate": 0.00019999347797008732,
-      "loss": 46.0,
-      "step": 47559
-    },
-    {
-      "epoch": 3.6362941300151004,
-      "grad_norm": 0.0015144210774451494,
-      "learning_rate": 0.00019999347769575864,
-      "loss": 46.0,
-      "step": 47560
-    },
-    {
-      "epoch": 3.63637058699849,
-      "grad_norm": 0.0010850976686924696,
-      "learning_rate": 0.00019999347742142415,
-      "loss": 46.0,
-      "step": 47561
-    },
-    {
-      "epoch": 3.63644704398188,
-      "grad_norm": 0.00571185490116477,
-      "learning_rate": 0.00019999347714708392,
-      "loss": 46.0,
-      "step": 47562
-    },
-    {
-      "epoch": 3.6365235009652697,
-      "grad_norm": 0.0019794958643615246,
-      "learning_rate": 0.00019999347687273792,
-      "loss": 46.0,
-      "step": 47563
-    },
-    {
-      "epoch": 3.636599957948659,
-      "grad_norm": 0.0018844889709725976,
-      "learning_rate": 0.00019999347659838614,
-      "loss": 46.0,
-      "step": 47564
-    },
-    {
-      "epoch": 3.6366764149320487,
-      "grad_norm": 0.004877413623034954,
-      "learning_rate": 0.0001999934763240286,
-      "loss": 46.0,
-      "step": 47565
-    },
-    {
-      "epoch": 3.6367528719154385,
-      "grad_norm": 0.0014948087045922875,
-      "learning_rate": 0.0001999934760496653,
-      "loss": 46.0,
-      "step": 47566
-    },
-    {
-      "epoch": 3.6368293288988283,
-      "grad_norm": 0.0027713363524526358,
-      "learning_rate": 0.0001999934757752962,
-      "loss": 46.0,
-      "step": 47567
-    },
-    {
-      "epoch": 3.636905785882218,
-      "grad_norm": 0.0034205422271043062,
-      "learning_rate": 0.00019999347550092136,
-      "loss": 46.0,
-      "step": 47568
-    },
-    {
-      "epoch": 3.636982242865608,
-      "grad_norm": 0.0018202033825218678,
-      "learning_rate": 0.00019999347522654074,
-      "loss": 46.0,
-      "step": 47569
-    },
-    {
-      "epoch": 3.6370586998489975,
-      "grad_norm": 0.0011646030470728874,
-      "learning_rate": 0.00019999347495215438,
-      "loss": 46.0,
-      "step": 47570
-    },
-    {
-      "epoch": 3.6371351568323873,
-      "grad_norm": 0.0018189975526183844,
-      "learning_rate": 0.0001999934746777622,
-      "loss": 46.0,
-      "step": 47571
-    },
-    {
-      "epoch": 3.6372116138157766,
-      "grad_norm": 0.0020405708346515894,
-      "learning_rate": 0.00019999347440336428,
-      "loss": 46.0,
-      "step": 47572
-    },
-    {
-      "epoch": 3.6372880707991664,
-      "grad_norm": 0.000633329211268574,
-      "learning_rate": 0.0001999934741289606,
-      "loss": 46.0,
-      "step": 47573
-    },
-    {
-      "epoch": 3.637364527782556,
-      "grad_norm": 0.0032375201117247343,
-      "learning_rate": 0.00019999347385455112,
-      "loss": 46.0,
-      "step": 47574
-    },
-    {
-      "epoch": 3.637440984765946,
-      "grad_norm": 0.0013362066820263863,
-      "learning_rate": 0.00019999347358013586,
-      "loss": 46.0,
-      "step": 47575
-    },
-    {
-      "epoch": 3.6375174417493357,
-      "grad_norm": 0.001816531759686768,
-      "learning_rate": 0.0001999934733057149,
-      "loss": 46.0,
-      "step": 47576
-    },
-    {
-      "epoch": 3.6375938987327254,
-      "grad_norm": 0.004090809728950262,
-      "learning_rate": 0.0001999934730312881,
-      "loss": 46.0,
-      "step": 47577
-    },
-    {
-      "epoch": 3.637670355716115,
-      "grad_norm": 0.0021148703526705503,
-      "learning_rate": 0.0001999934727568556,
-      "loss": 46.0,
-      "step": 47578
-    },
-    {
-      "epoch": 3.637746812699505,
-      "grad_norm": 0.0011162839364260435,
-      "learning_rate": 0.00019999347248241727,
-      "loss": 46.0,
-      "step": 47579
-    },
-    {
-      "epoch": 3.6378232696828947,
-      "grad_norm": 0.0011677221627905965,
-      "learning_rate": 0.00019999347220797317,
-      "loss": 46.0,
-      "step": 47580
-    },
-    {
-      "epoch": 3.6378997266662845,
-      "grad_norm": 0.0006460328586399555,
-      "learning_rate": 0.00019999347193352333,
-      "loss": 46.0,
-      "step": 47581
-    },
-    {
-      "epoch": 3.6379761836496742,
-      "grad_norm": 0.0019672748167067766,
-      "learning_rate": 0.0001999934716590677,
-      "loss": 46.0,
-      "step": 47582
-    },
-    {
-      "epoch": 3.638052640633064,
-      "grad_norm": 0.003686018055304885,
-      "learning_rate": 0.00019999347138460632,
-      "loss": 46.0,
-      "step": 47583
-    },
-    {
-      "epoch": 3.6381290976164538,
-      "grad_norm": 0.0016379008302465081,
-      "learning_rate": 0.00019999347111013916,
-      "loss": 46.0,
-      "step": 47584
-    },
-    {
-      "epoch": 3.6382055545998435,
-      "grad_norm": 0.002490095794200897,
-      "learning_rate": 0.00019999347083566625,
-      "loss": 46.0,
-      "step": 47585
-    },
-    {
-      "epoch": 3.638282011583233,
-      "grad_norm": 0.0026382869109511375,
-      "learning_rate": 0.00019999347056118757,
-      "loss": 46.0,
-      "step": 47586
-    },
-    {
-      "epoch": 3.6383584685666226,
-      "grad_norm": 0.0026858348865062,
-      "learning_rate": 0.0001999934702867031,
-      "loss": 46.0,
-      "step": 47587
-    },
-    {
-      "epoch": 3.6384349255500124,
-      "grad_norm": 0.0014362436486408114,
-      "learning_rate": 0.00019999347001221286,
-      "loss": 46.0,
-      "step": 47588
-    },
-    {
-      "epoch": 3.638511382533402,
-      "grad_norm": 0.0021098421420902014,
-      "learning_rate": 0.00019999346973771686,
-      "loss": 46.0,
-      "step": 47589
-    },
-    {
-      "epoch": 3.638587839516792,
-      "grad_norm": 0.002550907665863633,
-      "learning_rate": 0.0001999934694632151,
-      "loss": 46.0,
-      "step": 47590
-    },
-    {
-      "epoch": 3.6386642965001816,
-      "grad_norm": 0.0012488883221521974,
-      "learning_rate": 0.00019999346918870756,
-      "loss": 46.0,
-      "step": 47591
-    },
-    {
-      "epoch": 3.6387407534835714,
-      "grad_norm": 0.003144358517602086,
-      "learning_rate": 0.00019999346891419427,
-      "loss": 46.0,
-      "step": 47592
-    },
-    {
-      "epoch": 3.638817210466961,
-      "grad_norm": 0.005914170295000076,
-      "learning_rate": 0.0001999934686396752,
-      "loss": 46.0,
-      "step": 47593
-    },
-    {
-      "epoch": 3.6388936674503505,
-      "grad_norm": 0.005192840937525034,
-      "learning_rate": 0.00019999346836515033,
-      "loss": 46.0,
-      "step": 47594
-    },
-    {
-      "epoch": 3.6389701244337402,
-      "grad_norm": 0.002688988344743848,
-      "learning_rate": 0.0001999934680906197,
-      "loss": 46.0,
-      "step": 47595
-    },
-    {
-      "epoch": 3.63904658141713,
-      "grad_norm": 0.007607114966958761,
-      "learning_rate": 0.00019999346781608335,
-      "loss": 46.0,
-      "step": 47596
-    },
-    {
-      "epoch": 3.6391230384005198,
-      "grad_norm": 0.0016015507280826569,
-      "learning_rate": 0.0001999934675415412,
-      "loss": 46.0,
-      "step": 47597
-    },
-    {
-      "epoch": 3.6391994953839095,
-      "grad_norm": 0.0010930881835520267,
-      "learning_rate": 0.00019999346726699325,
-      "loss": 46.0,
-      "step": 47598
-    },
-    {
-      "epoch": 3.6392759523672993,
-      "grad_norm": 0.0019073153380304575,
-      "learning_rate": 0.00019999346699243957,
-      "loss": 46.0,
-      "step": 47599
-    },
-    {
-      "epoch": 3.639352409350689,
-      "grad_norm": 0.0021241530776023865,
-      "learning_rate": 0.00019999346671788012,
-      "loss": 46.0,
-      "step": 47600
-    },
-    {
-      "epoch": 3.639428866334079,
-      "grad_norm": 0.0009620288037694991,
-      "learning_rate": 0.00019999346644331486,
-      "loss": 46.0,
-      "step": 47601
-    },
-    {
-      "epoch": 3.6395053233174686,
-      "grad_norm": 0.0013683775905519724,
-      "learning_rate": 0.0001999934661687439,
-      "loss": 46.0,
-      "step": 47602
-    },
-    {
-      "epoch": 3.6395817803008583,
-      "grad_norm": 0.0023403067607432604,
-      "learning_rate": 0.0001999934658941671,
-      "loss": 46.0,
-      "step": 47603
-    },
-    {
-      "epoch": 3.639658237284248,
-      "grad_norm": 0.0020057130604982376,
-      "learning_rate": 0.0001999934656195846,
-      "loss": 46.0,
-      "step": 47604
-    },
-    {
-      "epoch": 3.639734694267638,
-      "grad_norm": 0.0011659599840641022,
-      "learning_rate": 0.00019999346534499627,
-      "loss": 46.0,
-      "step": 47605
-    },
-    {
-      "epoch": 3.6398111512510276,
-      "grad_norm": 0.006553406361490488,
-      "learning_rate": 0.0001999934650704022,
-      "loss": 46.0,
-      "step": 47606
-    },
-    {
-      "epoch": 3.6398876082344174,
-      "grad_norm": 0.0020889381412416697,
-      "learning_rate": 0.00019999346479580236,
-      "loss": 46.0,
-      "step": 47607
-    },
-    {
-      "epoch": 3.6399640652178067,
-      "grad_norm": 0.0022696072701364756,
-      "learning_rate": 0.00019999346452119677,
-      "loss": 46.0,
-      "step": 47608
-    },
-    {
-      "epoch": 3.6400405222011964,
-      "grad_norm": 0.002234655898064375,
-      "learning_rate": 0.00019999346424658536,
-      "loss": 46.0,
-      "step": 47609
-    },
-    {
-      "epoch": 3.640116979184586,
-      "grad_norm": 0.0009737928630784154,
-      "learning_rate": 0.00019999346397196822,
-      "loss": 46.0,
-      "step": 47610
-    },
-    {
-      "epoch": 3.640193436167976,
-      "grad_norm": 0.0012431824579834938,
-      "learning_rate": 0.00019999346369734529,
-      "loss": 46.0,
-      "step": 47611
-    },
-    {
-      "epoch": 3.6402698931513657,
-      "grad_norm": 0.001995167462155223,
-      "learning_rate": 0.0001999934634227166,
-      "loss": 46.0,
-      "step": 47612
-    },
-    {
-      "epoch": 3.6403463501347555,
-      "grad_norm": 0.002032445976510644,
-      "learning_rate": 0.00019999346314808215,
-      "loss": 46.0,
-      "step": 47613
-    },
-    {
-      "epoch": 3.6404228071181453,
-      "grad_norm": 0.0019962149672210217,
-      "learning_rate": 0.00019999346287344195,
-      "loss": 46.0,
-      "step": 47614
-    },
-    {
-      "epoch": 3.640499264101535,
-      "grad_norm": 0.001421817927621305,
-      "learning_rate": 0.00019999346259879595,
-      "loss": 46.0,
-      "step": 47615
-    },
-    {
-      "epoch": 3.6405757210849243,
-      "grad_norm": 0.0021803828421980143,
-      "learning_rate": 0.00019999346232414417,
-      "loss": 46.0,
-      "step": 47616
-    },
-    {
-      "epoch": 3.640652178068314,
-      "grad_norm": 0.0017514686333015561,
-      "learning_rate": 0.00019999346204948662,
-      "loss": 46.0,
-      "step": 47617
-    },
-    {
-      "epoch": 3.640728635051704,
-      "grad_norm": 0.002726459875702858,
-      "learning_rate": 0.00019999346177482333,
-      "loss": 46.0,
-      "step": 47618
-    },
-    {
-      "epoch": 3.6408050920350936,
-      "grad_norm": 0.001060849754139781,
-      "learning_rate": 0.00019999346150015426,
-      "loss": 46.0,
-      "step": 47619
-    },
-    {
-      "epoch": 3.6408815490184834,
-      "grad_norm": 0.0037169125862419605,
-      "learning_rate": 0.00019999346122547942,
-      "loss": 46.0,
-      "step": 47620
-    },
-    {
-      "epoch": 3.640958006001873,
-      "grad_norm": 0.000881242158357054,
-      "learning_rate": 0.0001999934609507988,
-      "loss": 46.0,
-      "step": 47621
-    },
-    {
-      "epoch": 3.641034462985263,
-      "grad_norm": 0.0014947558520361781,
-      "learning_rate": 0.00019999346067611242,
-      "loss": 46.0,
-      "step": 47622
-    },
-    {
-      "epoch": 3.6411109199686527,
-      "grad_norm": 0.0005922022392041981,
-      "learning_rate": 0.00019999346040142028,
-      "loss": 46.0,
-      "step": 47623
-    },
-    {
-      "epoch": 3.6411873769520424,
-      "grad_norm": 0.0012894232058897614,
-      "learning_rate": 0.00019999346012672238,
-      "loss": 46.0,
-      "step": 47624
-    },
-    {
-      "epoch": 3.641263833935432,
-      "grad_norm": 0.0015524973860010505,
-      "learning_rate": 0.00019999345985201867,
-      "loss": 46.0,
-      "step": 47625
-    },
-    {
-      "epoch": 3.641340290918822,
-      "grad_norm": 0.001231900416314602,
-      "learning_rate": 0.0001999934595773092,
-      "loss": 46.0,
-      "step": 47626
-    },
-    {
-      "epoch": 3.6414167479022117,
-      "grad_norm": 0.0026711379177868366,
-      "learning_rate": 0.00019999345930259399,
-      "loss": 46.0,
-      "step": 47627
-    },
-    {
-      "epoch": 3.6414932048856015,
-      "grad_norm": 0.0018887639744207263,
-      "learning_rate": 0.000199993459027873,
-      "loss": 46.0,
-      "step": 47628
-    },
-    {
-      "epoch": 3.6415696618689912,
-      "grad_norm": 0.0022163440007716417,
-      "learning_rate": 0.00019999345875314624,
-      "loss": 46.0,
-      "step": 47629
-    },
-    {
-      "epoch": 3.6416461188523805,
-      "grad_norm": 0.001590995118021965,
-      "learning_rate": 0.00019999345847841372,
-      "loss": 46.0,
-      "step": 47630
-    },
-    {
-      "epoch": 3.6417225758357703,
-      "grad_norm": 0.0016796269919723272,
-      "learning_rate": 0.00019999345820367542,
-      "loss": 46.0,
-      "step": 47631
-    },
-    {
-      "epoch": 3.64179903281916,
-      "grad_norm": 0.0030253541190177202,
-      "learning_rate": 0.00019999345792893133,
-      "loss": 46.0,
-      "step": 47632
-    },
-    {
-      "epoch": 3.64187548980255,
-      "grad_norm": 0.0029934654012322426,
-      "learning_rate": 0.0001999934576541815,
-      "loss": 46.0,
-      "step": 47633
-    },
-    {
-      "epoch": 3.6419519467859396,
-      "grad_norm": 0.0030152606777846813,
-      "learning_rate": 0.0001999934573794259,
-      "loss": 46.0,
-      "step": 47634
-    },
-    {
-      "epoch": 3.6420284037693293,
-      "grad_norm": 0.0016618386143818498,
-      "learning_rate": 0.0001999934571046645,
-      "loss": 46.0,
-      "step": 47635
-    },
-    {
-      "epoch": 3.642104860752719,
-      "grad_norm": 0.002503752475604415,
-      "learning_rate": 0.00019999345682989735,
-      "loss": 46.0,
-      "step": 47636
-    },
-    {
-      "epoch": 3.642181317736109,
-      "grad_norm": 0.0026403041556477547,
-      "learning_rate": 0.00019999345655512444,
-      "loss": 46.0,
-      "step": 47637
-    },
-    {
-      "epoch": 3.642257774719498,
-      "grad_norm": 0.00218627299182117,
-      "learning_rate": 0.00019999345628034576,
-      "loss": 46.0,
-      "step": 47638
-    },
-    {
-      "epoch": 3.642334231702888,
-      "grad_norm": 0.0033590688835829496,
-      "learning_rate": 0.0001999934560055613,
-      "loss": 46.0,
-      "step": 47639
-    },
-    {
-      "epoch": 3.6424106886862777,
-      "grad_norm": 0.0028649545274674892,
-      "learning_rate": 0.00019999345573077108,
-      "loss": 46.0,
-      "step": 47640
-    },
-    {
-      "epoch": 3.6424871456696675,
-      "grad_norm": 0.001072224578820169,
-      "learning_rate": 0.0001999934554559751,
-      "loss": 46.0,
-      "step": 47641
-    },
-    {
-      "epoch": 3.6425636026530572,
-      "grad_norm": 0.0032022055238485336,
-      "learning_rate": 0.00019999345518117333,
-      "loss": 46.0,
-      "step": 47642
-    },
-    {
-      "epoch": 3.642640059636447,
-      "grad_norm": 0.0015884184977039695,
-      "learning_rate": 0.0001999934549063658,
-      "loss": 46.0,
-      "step": 47643
-    },
-    {
-      "epoch": 3.6427165166198368,
-      "grad_norm": 0.0021473481319844723,
-      "learning_rate": 0.0001999934546315525,
-      "loss": 46.0,
-      "step": 47644
-    },
-    {
-      "epoch": 3.6427929736032265,
-      "grad_norm": 0.005108244251459837,
-      "learning_rate": 0.00019999345435673345,
-      "loss": 46.0,
-      "step": 47645
-    },
-    {
-      "epoch": 3.6428694305866163,
-      "grad_norm": 0.0014162365114316344,
-      "learning_rate": 0.0001999934540819086,
-      "loss": 46.0,
-      "step": 47646
-    },
-    {
-      "epoch": 3.642945887570006,
-      "grad_norm": 0.00281914253719151,
-      "learning_rate": 0.000199993453807078,
-      "loss": 46.0,
-      "step": 47647
-    },
-    {
-      "epoch": 3.643022344553396,
-      "grad_norm": 0.0020775776356458664,
-      "learning_rate": 0.0001999934535322416,
-      "loss": 46.0,
-      "step": 47648
-    },
-    {
-      "epoch": 3.6430988015367856,
-      "grad_norm": 0.0009017596021294594,
-      "learning_rate": 0.00019999345325739948,
-      "loss": 46.0,
-      "step": 47649
-    },
-    {
-      "epoch": 3.6431752585201753,
-      "grad_norm": 0.002380523830652237,
-      "learning_rate": 0.00019999345298255157,
-      "loss": 46.0,
-      "step": 47650
-    },
-    {
-      "epoch": 3.643251715503565,
-      "grad_norm": 0.0029388549737632275,
-      "learning_rate": 0.00019999345270769786,
-      "loss": 46.0,
-      "step": 47651
-    },
-    {
-      "epoch": 3.6433281724869544,
-      "grad_norm": 0.00310456077568233,
-      "learning_rate": 0.00019999345243283843,
-      "loss": 46.0,
-      "step": 47652
-    },
-    {
-      "epoch": 3.643404629470344,
-      "grad_norm": 0.001864472753368318,
-      "learning_rate": 0.0001999934521579732,
-      "loss": 46.0,
-      "step": 47653
-    },
-    {
-      "epoch": 3.643481086453734,
-      "grad_norm": 0.005107543896883726,
-      "learning_rate": 0.0001999934518831022,
-      "loss": 46.0,
-      "step": 47654
-    },
-    {
-      "epoch": 3.6435575434371237,
-      "grad_norm": 0.0014210341032594442,
-      "learning_rate": 0.00019999345160822543,
-      "loss": 46.0,
-      "step": 47655
-    },
-    {
-      "epoch": 3.6436340004205134,
-      "grad_norm": 0.0008182196761481464,
-      "learning_rate": 0.00019999345133334294,
-      "loss": 46.0,
-      "step": 47656
-    },
-    {
-      "epoch": 3.643710457403903,
-      "grad_norm": 0.002491212449967861,
-      "learning_rate": 0.00019999345105845465,
-      "loss": 46.0,
-      "step": 47657
-    },
-    {
-      "epoch": 3.643786914387293,
-      "grad_norm": 0.003380777547135949,
-      "learning_rate": 0.00019999345078356058,
-      "loss": 46.0,
-      "step": 47658
-    },
-    {
-      "epoch": 3.6438633713706827,
-      "grad_norm": 0.0012305903946980834,
-      "learning_rate": 0.00019999345050866074,
-      "loss": 46.0,
-      "step": 47659
-    },
-    {
-      "epoch": 3.643939828354072,
-      "grad_norm": 0.0013567740097641945,
-      "learning_rate": 0.00019999345023375513,
-      "loss": 46.0,
-      "step": 47660
-    },
-    {
-      "epoch": 3.644016285337462,
-      "grad_norm": 0.003648675512522459,
-      "learning_rate": 0.00019999344995884374,
-      "loss": 46.0,
-      "step": 47661
-    },
-    {
-      "epoch": 3.6440927423208516,
-      "grad_norm": 0.0023186441976577044,
-      "learning_rate": 0.0001999934496839266,
-      "loss": 46.0,
-      "step": 47662
-    },
-    {
-      "epoch": 3.6441691993042413,
-      "grad_norm": 0.0007946082623675466,
-      "learning_rate": 0.0001999934494090037,
-      "loss": 46.0,
-      "step": 47663
-    },
-    {
-      "epoch": 3.644245656287631,
-      "grad_norm": 0.0011176541447639465,
-      "learning_rate": 0.00019999344913407502,
-      "loss": 46.0,
-      "step": 47664
-    },
-    {
-      "epoch": 3.644322113271021,
-      "grad_norm": 0.002799355424940586,
-      "learning_rate": 0.00019999344885914056,
-      "loss": 46.0,
-      "step": 47665
-    },
-    {
-      "epoch": 3.6443985702544106,
-      "grad_norm": 0.0014208493521437049,
-      "learning_rate": 0.00019999344858420034,
-      "loss": 46.0,
-      "step": 47666
-    },
-    {
-      "epoch": 3.6444750272378004,
-      "grad_norm": 0.001411773613654077,
-      "learning_rate": 0.00019999344830925434,
-      "loss": 46.0,
-      "step": 47667
-    },
-    {
-      "epoch": 3.64455148422119,
-      "grad_norm": 0.0031203078106045723,
-      "learning_rate": 0.0001999934480343026,
-      "loss": 46.0,
-      "step": 47668
-    },
-    {
-      "epoch": 3.64462794120458,
-      "grad_norm": 0.0026555671356618404,
-      "learning_rate": 0.00019999344775934507,
-      "loss": 46.0,
-      "step": 47669
-    },
-    {
-      "epoch": 3.6447043981879697,
-      "grad_norm": 0.0029595624655485153,
-      "learning_rate": 0.00019999344748438178,
-      "loss": 46.0,
-      "step": 47670
-    },
-    {
-      "epoch": 3.6447808551713594,
-      "grad_norm": 0.0009205969981849194,
-      "learning_rate": 0.0001999934472094127,
-      "loss": 46.0,
-      "step": 47671
-    },
-    {
-      "epoch": 3.644857312154749,
-      "grad_norm": 0.0015713887987658381,
-      "learning_rate": 0.0001999934469344379,
-      "loss": 46.0,
-      "step": 47672
-    },
-    {
-      "epoch": 3.6449337691381385,
-      "grad_norm": 0.004887838382273912,
-      "learning_rate": 0.00019999344665945729,
-      "loss": 46.0,
-      "step": 47673
-    },
-    {
-      "epoch": 3.6450102261215283,
-      "grad_norm": 0.0013900778722018003,
-      "learning_rate": 0.0001999934463844709,
-      "loss": 46.0,
-      "step": 47674
-    },
-    {
-      "epoch": 3.645086683104918,
-      "grad_norm": 0.0010336030973121524,
-      "learning_rate": 0.0001999934461094788,
-      "loss": 46.0,
-      "step": 47675
-    },
-    {
-      "epoch": 3.6451631400883078,
-      "grad_norm": 0.001061112037859857,
-      "learning_rate": 0.00019999344583448086,
-      "loss": 46.0,
-      "step": 47676
-    },
-    {
-      "epoch": 3.6452395970716975,
-      "grad_norm": 0.0016902538482099771,
-      "learning_rate": 0.00019999344555947718,
-      "loss": 46.0,
-      "step": 47677
-    },
-    {
-      "epoch": 3.6453160540550873,
-      "grad_norm": 0.0022786823101341724,
-      "learning_rate": 0.00019999344528446775,
-      "loss": 46.0,
-      "step": 47678
-    },
-    {
-      "epoch": 3.645392511038477,
-      "grad_norm": 0.0006105888751335442,
-      "learning_rate": 0.00019999344500945253,
-      "loss": 46.0,
-      "step": 47679
-    },
-    {
-      "epoch": 3.645468968021867,
-      "grad_norm": 0.002234566491097212,
-      "learning_rate": 0.00019999344473443153,
-      "loss": 46.0,
-      "step": 47680
-    },
-    {
-      "epoch": 3.6455454250052566,
-      "grad_norm": 0.00117902597412467,
-      "learning_rate": 0.00019999344445940478,
-      "loss": 46.0,
-      "step": 47681
-    },
-    {
-      "epoch": 3.645621881988646,
-      "grad_norm": 0.002178714144974947,
-      "learning_rate": 0.00019999344418437226,
-      "loss": 46.0,
-      "step": 47682
-    },
-    {
-      "epoch": 3.6456983389720357,
-      "grad_norm": 0.0022162015084177256,
-      "learning_rate": 0.00019999344390933397,
-      "loss": 46.0,
-      "step": 47683
-    },
-    {
-      "epoch": 3.6457747959554254,
-      "grad_norm": 0.0009124897769652307,
-      "learning_rate": 0.0001999934436342899,
-      "loss": 46.0,
-      "step": 47684
-    },
-    {
-      "epoch": 3.645851252938815,
-      "grad_norm": 0.0014934351202100515,
-      "learning_rate": 0.0001999934433592401,
-      "loss": 46.0,
-      "step": 47685
-    },
-    {
-      "epoch": 3.645927709922205,
-      "grad_norm": 0.002741024596616626,
-      "learning_rate": 0.00019999344308418445,
-      "loss": 46.0,
-      "step": 47686
-    },
-    {
-      "epoch": 3.6460041669055947,
-      "grad_norm": 0.0014128170441836119,
-      "learning_rate": 0.00019999344280912312,
-      "loss": 46.0,
-      "step": 47687
-    },
-    {
-      "epoch": 3.6460806238889845,
-      "grad_norm": 0.002291157841682434,
-      "learning_rate": 0.00019999344253405596,
-      "loss": 46.0,
-      "step": 47688
-    },
-    {
-      "epoch": 3.6461570808723742,
-      "grad_norm": 0.001676909625530243,
-      "learning_rate": 0.00019999344225898305,
-      "loss": 46.0,
-      "step": 47689
-    },
-    {
-      "epoch": 3.646233537855764,
-      "grad_norm": 0.0005470478790812194,
-      "learning_rate": 0.00019999344198390437,
-      "loss": 46.0,
-      "step": 47690
-    },
-    {
-      "epoch": 3.6463099948391537,
-      "grad_norm": 0.0012233055895194411,
-      "learning_rate": 0.00019999344170881992,
-      "loss": 46.0,
-      "step": 47691
-    },
-    {
-      "epoch": 3.6463864518225435,
-      "grad_norm": 0.001965420786291361,
-      "learning_rate": 0.00019999344143372972,
-      "loss": 46.0,
-      "step": 47692
-    },
-    {
-      "epoch": 3.6464629088059333,
-      "grad_norm": 0.0013269162736833096,
-      "learning_rate": 0.00019999344115863372,
-      "loss": 46.0,
-      "step": 47693
-    },
-    {
-      "epoch": 3.646539365789323,
-      "grad_norm": 0.003292139619588852,
-      "learning_rate": 0.00019999344088353197,
-      "loss": 46.0,
-      "step": 47694
-    },
-    {
-      "epoch": 3.6466158227727123,
-      "grad_norm": 0.002590600633993745,
-      "learning_rate": 0.00019999344060842446,
-      "loss": 46.0,
-      "step": 47695
-    },
-    {
-      "epoch": 3.646692279756102,
-      "grad_norm": 0.001371645601466298,
-      "learning_rate": 0.00019999344033331116,
-      "loss": 46.0,
-      "step": 47696
-    },
-    {
-      "epoch": 3.646768736739492,
-      "grad_norm": 0.005140399560332298,
-      "learning_rate": 0.00019999344005819212,
-      "loss": 46.0,
-      "step": 47697
-    },
-    {
-      "epoch": 3.6468451937228816,
-      "grad_norm": 0.0006940077873878181,
-      "learning_rate": 0.00019999343978306728,
-      "loss": 46.0,
-      "step": 47698
-    },
-    {
-      "epoch": 3.6469216507062714,
-      "grad_norm": 0.004302518907934427,
-      "learning_rate": 0.00019999343950793667,
-      "loss": 46.0,
-      "step": 47699
-    },
-    {
-      "epoch": 3.646998107689661,
-      "grad_norm": 0.0025637801736593246,
-      "learning_rate": 0.0001999934392328003,
-      "loss": 46.0,
-      "step": 47700
-    },
-    {
-      "epoch": 3.647074564673051,
-      "grad_norm": 0.003288046456873417,
-      "learning_rate": 0.00019999343895765818,
-      "loss": 46.0,
-      "step": 47701
-    },
-    {
-      "epoch": 3.6471510216564407,
-      "grad_norm": 0.002868850715458393,
-      "learning_rate": 0.00019999343868251027,
-      "loss": 46.0,
-      "step": 47702
-    },
-    {
-      "epoch": 3.64722747863983,
-      "grad_norm": 0.0009547343361191452,
-      "learning_rate": 0.0001999934384073566,
-      "loss": 46.0,
-      "step": 47703
-    },
-    {
-      "epoch": 3.6473039356232198,
-      "grad_norm": 0.004754739347845316,
-      "learning_rate": 0.00019999343813219714,
-      "loss": 46.0,
-      "step": 47704
-    },
-    {
-      "epoch": 3.6473803926066095,
-      "grad_norm": 0.002491709077730775,
-      "learning_rate": 0.00019999343785703192,
-      "loss": 46.0,
-      "step": 47705
-    },
-    {
-      "epoch": 3.6474568495899993,
-      "grad_norm": 0.0023551075719296932,
-      "learning_rate": 0.00019999343758186097,
-      "loss": 46.0,
-      "step": 47706
-    },
-    {
-      "epoch": 3.647533306573389,
-      "grad_norm": 0.004635291639715433,
-      "learning_rate": 0.0001999934373066842,
-      "loss": 46.0,
-      "step": 47707
-    },
-    {
-      "epoch": 3.647609763556779,
-      "grad_norm": 0.000653390830848366,
-      "learning_rate": 0.00019999343703150168,
-      "loss": 46.0,
-      "step": 47708
-    },
-    {
-      "epoch": 3.6476862205401686,
-      "grad_norm": 0.0018211125861853361,
-      "learning_rate": 0.0001999934367563134,
-      "loss": 46.0,
-      "step": 47709
-    },
-    {
-      "epoch": 3.6477626775235583,
-      "grad_norm": 0.0011218676809221506,
-      "learning_rate": 0.00019999343648111935,
-      "loss": 46.0,
-      "step": 47710
-    },
-    {
-      "epoch": 3.647839134506948,
-      "grad_norm": 0.0043051172979176044,
-      "learning_rate": 0.00019999343620591954,
-      "loss": 46.0,
-      "step": 47711
-    },
-    {
-      "epoch": 3.647915591490338,
-      "grad_norm": 0.0017796941101551056,
-      "learning_rate": 0.00019999343593071392,
-      "loss": 46.0,
-      "step": 47712
-    },
-    {
-      "epoch": 3.6479920484737276,
-      "grad_norm": 0.004201495088636875,
-      "learning_rate": 0.00019999343565550257,
-      "loss": 46.0,
-      "step": 47713
-    },
-    {
-      "epoch": 3.6480685054571174,
-      "grad_norm": 0.0018534999107941985,
-      "learning_rate": 0.00019999343538028543,
-      "loss": 46.0,
-      "step": 47714
-    },
-    {
-      "epoch": 3.648144962440507,
-      "grad_norm": 0.002223034854978323,
-      "learning_rate": 0.00019999343510506253,
-      "loss": 46.0,
-      "step": 47715
-    },
-    {
-      "epoch": 3.648221419423897,
-      "grad_norm": 0.0036811749450862408,
-      "learning_rate": 0.00019999343482983385,
-      "loss": 46.0,
-      "step": 47716
-    },
-    {
-      "epoch": 3.648297876407286,
-      "grad_norm": 0.003338853595778346,
-      "learning_rate": 0.0001999934345545994,
-      "loss": 46.0,
-      "step": 47717
-    },
-    {
-      "epoch": 3.648374333390676,
-      "grad_norm": 0.0030563834588974714,
-      "learning_rate": 0.0001999934342793592,
-      "loss": 46.0,
-      "step": 47718
-    },
-    {
-      "epoch": 3.6484507903740657,
-      "grad_norm": 0.005340575706213713,
-      "learning_rate": 0.0001999934340041132,
-      "loss": 46.0,
-      "step": 47719
-    },
-    {
-      "epoch": 3.6485272473574555,
-      "grad_norm": 0.0044814208522439,
-      "learning_rate": 0.00019999343372886146,
-      "loss": 46.0,
-      "step": 47720
-    },
-    {
-      "epoch": 3.6486037043408452,
-      "grad_norm": 0.001382531481795013,
-      "learning_rate": 0.00019999343345360396,
-      "loss": 46.0,
-      "step": 47721
-    },
-    {
-      "epoch": 3.648680161324235,
-      "grad_norm": 0.001145056216046214,
-      "learning_rate": 0.00019999343317834067,
-      "loss": 46.0,
-      "step": 47722
-    },
-    {
-      "epoch": 3.6487566183076248,
-      "grad_norm": 0.003555477363988757,
-      "learning_rate": 0.0001999934329030716,
-      "loss": 46.0,
-      "step": 47723
-    },
-    {
-      "epoch": 3.6488330752910145,
-      "grad_norm": 0.0016626415308564901,
-      "learning_rate": 0.00019999343262779677,
-      "loss": 46.0,
-      "step": 47724
-    },
-    {
-      "epoch": 3.648909532274404,
-      "grad_norm": 0.0016413347329944372,
-      "learning_rate": 0.0001999934323525162,
-      "loss": 46.0,
-      "step": 47725
-    },
-    {
-      "epoch": 3.6489859892577936,
-      "grad_norm": 0.010179664008319378,
-      "learning_rate": 0.00019999343207722982,
-      "loss": 46.0,
-      "step": 47726
-    },
-    {
-      "epoch": 3.6490624462411834,
-      "grad_norm": 0.004473886918276548,
-      "learning_rate": 0.0001999934318019377,
-      "loss": 46.0,
-      "step": 47727
-    },
-    {
-      "epoch": 3.649138903224573,
-      "grad_norm": 0.009979039430618286,
-      "learning_rate": 0.0001999934315266398,
-      "loss": 46.0,
-      "step": 47728
-    },
-    {
-      "epoch": 3.649215360207963,
-      "grad_norm": 0.0033817191142588854,
-      "learning_rate": 0.0001999934312513361,
-      "loss": 46.0,
-      "step": 47729
-    },
-    {
-      "epoch": 3.6492918171913526,
-      "grad_norm": 0.0022713039070367813,
-      "learning_rate": 0.00019999343097602668,
-      "loss": 46.0,
-      "step": 47730
-    },
-    {
-      "epoch": 3.6493682741747424,
-      "grad_norm": 0.002307408954948187,
-      "learning_rate": 0.00019999343070071146,
-      "loss": 46.0,
-      "step": 47731
-    },
-    {
-      "epoch": 3.649444731158132,
-      "grad_norm": 0.0022228495217859745,
-      "learning_rate": 0.0001999934304253905,
-      "loss": 46.0,
-      "step": 47732
-    },
-    {
-      "epoch": 3.649521188141522,
-      "grad_norm": 0.0015586961526423693,
-      "learning_rate": 0.00019999343015006374,
-      "loss": 46.0,
-      "step": 47733
-    },
-    {
-      "epoch": 3.6495976451249117,
-      "grad_norm": 0.0032822387292981148,
-      "learning_rate": 0.00019999342987473125,
-      "loss": 46.0,
-      "step": 47734
-    },
-    {
-      "epoch": 3.6496741021083015,
-      "grad_norm": 0.002411150373518467,
-      "learning_rate": 0.00019999342959939296,
-      "loss": 46.0,
-      "step": 47735
-    },
-    {
-      "epoch": 3.649750559091691,
-      "grad_norm": 0.0018034750828519464,
-      "learning_rate": 0.0001999934293240489,
-      "loss": 46.0,
-      "step": 47736
-    },
-    {
-      "epoch": 3.649827016075081,
-      "grad_norm": 0.003922856412827969,
-      "learning_rate": 0.00019999342904869906,
-      "loss": 46.0,
-      "step": 47737
-    },
-    {
-      "epoch": 3.6499034730584707,
-      "grad_norm": 0.0019520603818818927,
-      "learning_rate": 0.00019999342877334347,
-      "loss": 46.0,
-      "step": 47738
-    },
-    {
-      "epoch": 3.64997993004186,
-      "grad_norm": 0.0019292542710900307,
-      "learning_rate": 0.00019999342849798214,
-      "loss": 46.0,
-      "step": 47739
-    },
-    {
-      "epoch": 3.65005638702525,
-      "grad_norm": 0.001201114384457469,
-      "learning_rate": 0.00019999342822261498,
-      "loss": 46.0,
-      "step": 47740
-    },
-    {
-      "epoch": 3.6501328440086396,
-      "grad_norm": 0.0035335260909050703,
-      "learning_rate": 0.0001999934279472421,
-      "loss": 46.0,
-      "step": 47741
-    },
-    {
-      "epoch": 3.6502093009920293,
-      "grad_norm": 0.0027000131085515022,
-      "learning_rate": 0.0001999934276718634,
-      "loss": 46.0,
-      "step": 47742
-    },
-    {
-      "epoch": 3.650285757975419,
-      "grad_norm": 0.004167946521192789,
-      "learning_rate": 0.00019999342739647897,
-      "loss": 46.0,
-      "step": 47743
-    },
-    {
-      "epoch": 3.650362214958809,
-      "grad_norm": 0.0009979866445064545,
-      "learning_rate": 0.00019999342712108878,
-      "loss": 46.0,
-      "step": 47744
-    },
-    {
-      "epoch": 3.6504386719421986,
-      "grad_norm": 0.0026974081993103027,
-      "learning_rate": 0.0001999934268456928,
-      "loss": 46.0,
-      "step": 47745
-    },
-    {
-      "epoch": 3.6505151289255884,
-      "grad_norm": 0.0009071906097233295,
-      "learning_rate": 0.00019999342657029106,
-      "loss": 46.0,
-      "step": 47746
-    },
-    {
-      "epoch": 3.6505915859089777,
-      "grad_norm": 0.00834709219634533,
-      "learning_rate": 0.00019999342629488354,
-      "loss": 46.0,
-      "step": 47747
-    },
-    {
-      "epoch": 3.6506680428923675,
-      "grad_norm": 0.002150040352717042,
-      "learning_rate": 0.00019999342601947025,
-      "loss": 46.0,
-      "step": 47748
-    },
-    {
-      "epoch": 3.650744499875757,
-      "grad_norm": 0.004443623591214418,
-      "learning_rate": 0.0001999934257440512,
-      "loss": 46.0,
-      "step": 47749
-    },
-    {
-      "epoch": 3.650820956859147,
-      "grad_norm": 0.002520909532904625,
-      "learning_rate": 0.00019999342546862638,
-      "loss": 46.0,
-      "step": 47750
-    },
-    {
-      "epoch": 3.6508974138425367,
-      "grad_norm": 0.004866630304604769,
-      "learning_rate": 0.0001999934251931958,
-      "loss": 46.0,
-      "step": 47751
-    },
-    {
-      "epoch": 3.6509738708259265,
-      "grad_norm": 0.0007708751363679767,
-      "learning_rate": 0.00019999342491775943,
-      "loss": 46.0,
-      "step": 47752
-    },
-    {
-      "epoch": 3.6510503278093163,
-      "grad_norm": 0.0015716284979134798,
-      "learning_rate": 0.00019999342464231733,
-      "loss": 46.0,
-      "step": 47753
-    },
-    {
-      "epoch": 3.651126784792706,
-      "grad_norm": 0.003939527086913586,
-      "learning_rate": 0.0001999934243668694,
-      "loss": 46.0,
-      "step": 47754
-    },
-    {
-      "epoch": 3.651203241776096,
-      "grad_norm": 0.002039143582805991,
-      "learning_rate": 0.00019999342409141572,
-      "loss": 46.0,
-      "step": 47755
-    },
-    {
-      "epoch": 3.6512796987594855,
-      "grad_norm": 0.0017836353508755565,
-      "learning_rate": 0.0001999934238159563,
-      "loss": 46.0,
-      "step": 47756
-    },
-    {
-      "epoch": 3.6513561557428753,
-      "grad_norm": 0.0008404995314776897,
-      "learning_rate": 0.0001999934235404911,
-      "loss": 46.0,
-      "step": 47757
-    },
-    {
-      "epoch": 3.651432612726265,
-      "grad_norm": 0.0032451471779495478,
-      "learning_rate": 0.00019999342326502013,
-      "loss": 46.0,
-      "step": 47758
-    },
-    {
-      "epoch": 3.651509069709655,
-      "grad_norm": 0.0032415452878922224,
-      "learning_rate": 0.00019999342298954338,
-      "loss": 46.0,
-      "step": 47759
-    },
-    {
-      "epoch": 3.6515855266930446,
-      "grad_norm": 0.0021821290720254183,
-      "learning_rate": 0.00019999342271406087,
-      "loss": 46.0,
-      "step": 47760
-    },
-    {
-      "epoch": 3.651661983676434,
-      "grad_norm": 0.0017507910961285233,
-      "learning_rate": 0.00019999342243857257,
-      "loss": 46.0,
-      "step": 47761
-    },
-    {
-      "epoch": 3.6517384406598237,
-      "grad_norm": 0.0016612119507044554,
-      "learning_rate": 0.00019999342216307854,
-      "loss": 46.0,
-      "step": 47762
-    },
-    {
-      "epoch": 3.6518148976432134,
-      "grad_norm": 0.001398536842316389,
-      "learning_rate": 0.00019999342188757873,
-      "loss": 46.0,
-      "step": 47763
-    },
-    {
-      "epoch": 3.651891354626603,
-      "grad_norm": 0.002227661432698369,
-      "learning_rate": 0.00019999342161207312,
-      "loss": 46.0,
-      "step": 47764
-    },
-    {
-      "epoch": 3.651967811609993,
-      "grad_norm": 0.003660982707515359,
-      "learning_rate": 0.00019999342133656178,
-      "loss": 46.0,
-      "step": 47765
-    },
-    {
-      "epoch": 3.6520442685933827,
-      "grad_norm": 0.0019833461847156286,
-      "learning_rate": 0.00019999342106104465,
-      "loss": 46.0,
-      "step": 47766
-    },
-    {
-      "epoch": 3.6521207255767725,
-      "grad_norm": 0.0035065049305558205,
-      "learning_rate": 0.00019999342078552178,
-      "loss": 46.0,
-      "step": 47767
-    },
-    {
-      "epoch": 3.6521971825601622,
-      "grad_norm": 0.001908307196572423,
-      "learning_rate": 0.0001999934205099931,
-      "loss": 46.0,
-      "step": 47768
-    },
-    {
-      "epoch": 3.6522736395435516,
-      "grad_norm": 0.0031432416290044785,
-      "learning_rate": 0.00019999342023445865,
-      "loss": 46.0,
-      "step": 47769
-    },
-    {
-      "epoch": 3.6523500965269413,
-      "grad_norm": 0.002135820686817169,
-      "learning_rate": 0.00019999341995891845,
-      "loss": 46.0,
-      "step": 47770
-    },
-    {
-      "epoch": 3.652426553510331,
-      "grad_norm": 0.00328750885091722,
-      "learning_rate": 0.00019999341968337248,
-      "loss": 46.0,
-      "step": 47771
-    },
-    {
-      "epoch": 3.652503010493721,
-      "grad_norm": 0.008645646274089813,
-      "learning_rate": 0.00019999341940782074,
-      "loss": 46.0,
-      "step": 47772
-    },
-    {
-      "epoch": 3.6525794674771106,
-      "grad_norm": 0.002184393350034952,
-      "learning_rate": 0.00019999341913226322,
-      "loss": 46.0,
-      "step": 47773
-    },
-    {
-      "epoch": 3.6526559244605004,
-      "grad_norm": 0.002336875069886446,
-      "learning_rate": 0.00019999341885669996,
-      "loss": 46.0,
-      "step": 47774
-    },
-    {
-      "epoch": 3.65273238144389,
-      "grad_norm": 0.0015706915874034166,
-      "learning_rate": 0.00019999341858113092,
-      "loss": 46.0,
-      "step": 47775
-    },
-    {
-      "epoch": 3.65280883842728,
-      "grad_norm": 0.0008334586163982749,
-      "learning_rate": 0.00019999341830555608,
-      "loss": 46.0,
-      "step": 47776
-    },
-    {
-      "epoch": 3.6528852954106696,
-      "grad_norm": 0.0023207198828458786,
-      "learning_rate": 0.0001999934180299755,
-      "loss": 46.0,
-      "step": 47777
-    },
-    {
-      "epoch": 3.6529617523940594,
-      "grad_norm": 0.001355098094791174,
-      "learning_rate": 0.00019999341775438914,
-      "loss": 46.0,
-      "step": 47778
-    },
-    {
-      "epoch": 3.653038209377449,
-      "grad_norm": 0.002461744239553809,
-      "learning_rate": 0.000199993417478797,
-      "loss": 46.0,
-      "step": 47779
-    },
-    {
-      "epoch": 3.653114666360839,
-      "grad_norm": 0.002446526661515236,
-      "learning_rate": 0.00019999341720319913,
-      "loss": 46.0,
-      "step": 47780
-    },
-    {
-      "epoch": 3.6531911233442287,
-      "grad_norm": 0.0015633394941687584,
-      "learning_rate": 0.00019999341692759548,
-      "loss": 46.0,
-      "step": 47781
-    },
-    {
-      "epoch": 3.6532675803276184,
-      "grad_norm": 0.0014555988600477576,
-      "learning_rate": 0.00019999341665198603,
-      "loss": 46.0,
-      "step": 47782
-    },
-    {
-      "epoch": 3.6533440373110078,
-      "grad_norm": 0.002532979706302285,
-      "learning_rate": 0.00019999341637637084,
-      "loss": 46.0,
-      "step": 47783
-    },
-    {
-      "epoch": 3.6534204942943975,
-      "grad_norm": 0.0019914903677999973,
-      "learning_rate": 0.00019999341610074987,
-      "loss": 46.0,
-      "step": 47784
-    },
-    {
-      "epoch": 3.6534969512777873,
-      "grad_norm": 0.003524721134454012,
-      "learning_rate": 0.00019999341582512315,
-      "loss": 46.0,
-      "step": 47785
-    },
-    {
-      "epoch": 3.653573408261177,
-      "grad_norm": 0.00334327039308846,
-      "learning_rate": 0.0001999934155494906,
-      "loss": 46.0,
-      "step": 47786
-    },
-    {
-      "epoch": 3.653649865244567,
-      "grad_norm": 0.0009810846531763673,
-      "learning_rate": 0.00019999341527385234,
-      "loss": 46.0,
-      "step": 47787
-    },
-    {
-      "epoch": 3.6537263222279566,
-      "grad_norm": 0.0010697762481868267,
-      "learning_rate": 0.0001999934149982083,
-      "loss": 46.0,
-      "step": 47788
-    },
-    {
-      "epoch": 3.6538027792113463,
-      "grad_norm": 0.001967029646039009,
-      "learning_rate": 0.00019999341472255847,
-      "loss": 46.0,
-      "step": 47789
-    },
-    {
-      "epoch": 3.653879236194736,
-      "grad_norm": 0.0017179121496155858,
-      "learning_rate": 0.00019999341444690288,
-      "loss": 46.0,
-      "step": 47790
-    },
-    {
-      "epoch": 3.6539556931781254,
-      "grad_norm": 0.0009905846090987325,
-      "learning_rate": 0.00019999341417124153,
-      "loss": 46.0,
-      "step": 47791
-    },
-    {
-      "epoch": 3.654032150161515,
-      "grad_norm": 0.0037586409598588943,
-      "learning_rate": 0.00019999341389557442,
-      "loss": 46.0,
-      "step": 47792
-    },
-    {
-      "epoch": 3.654108607144905,
-      "grad_norm": 0.00219179829582572,
-      "learning_rate": 0.00019999341361990155,
-      "loss": 46.0,
-      "step": 47793
-    },
-    {
-      "epoch": 3.6541850641282947,
-      "grad_norm": 0.0025964563246816397,
-      "learning_rate": 0.00019999341334422287,
-      "loss": 46.0,
-      "step": 47794
-    },
-    {
-      "epoch": 3.6542615211116845,
-      "grad_norm": 0.0013601521495729685,
-      "learning_rate": 0.00019999341306853845,
-      "loss": 46.0,
-      "step": 47795
-    },
-    {
-      "epoch": 3.654337978095074,
-      "grad_norm": 0.003588613588362932,
-      "learning_rate": 0.00019999341279284825,
-      "loss": 46.0,
-      "step": 47796
-    },
-    {
-      "epoch": 3.654414435078464,
-      "grad_norm": 0.004260408692061901,
-      "learning_rate": 0.00019999341251715228,
-      "loss": 46.0,
-      "step": 47797
-    },
-    {
-      "epoch": 3.6544908920618537,
-      "grad_norm": 0.0023365281522274017,
-      "learning_rate": 0.00019999341224145054,
-      "loss": 46.0,
-      "step": 47798
-    },
-    {
-      "epoch": 3.6545673490452435,
-      "grad_norm": 0.002064472995698452,
-      "learning_rate": 0.00019999341196574305,
-      "loss": 46.0,
-      "step": 47799
-    },
-    {
-      "epoch": 3.6546438060286333,
-      "grad_norm": 0.001487192464992404,
-      "learning_rate": 0.00019999341169002976,
-      "loss": 46.0,
-      "step": 47800
-    },
-    {
-      "epoch": 3.654720263012023,
-      "grad_norm": 0.004042519256472588,
-      "learning_rate": 0.00019999341141431072,
-      "loss": 46.0,
-      "step": 47801
-    },
-    {
-      "epoch": 3.654796719995413,
-      "grad_norm": 0.001744654611684382,
-      "learning_rate": 0.00019999341113858588,
-      "loss": 46.0,
-      "step": 47802
-    },
-    {
-      "epoch": 3.6548731769788025,
-      "grad_norm": 0.0016182203544303775,
-      "learning_rate": 0.00019999341086285533,
-      "loss": 46.0,
-      "step": 47803
-    },
-    {
-      "epoch": 3.654949633962192,
-      "grad_norm": 0.0014370906865224242,
-      "learning_rate": 0.00019999341058711897,
-      "loss": 46.0,
-      "step": 47804
-    },
-    {
-      "epoch": 3.6550260909455816,
-      "grad_norm": 0.0024650858249515295,
-      "learning_rate": 0.00019999341031137684,
-      "loss": 46.0,
-      "step": 47805
-    },
-    {
-      "epoch": 3.6551025479289714,
-      "grad_norm": 0.0069202883169054985,
-      "learning_rate": 0.00019999341003562897,
-      "loss": 46.0,
-      "step": 47806
-    },
-    {
-      "epoch": 3.655179004912361,
-      "grad_norm": 0.0024855954106897116,
-      "learning_rate": 0.00019999340975987532,
-      "loss": 46.0,
-      "step": 47807
-    },
-    {
-      "epoch": 3.655255461895751,
-      "grad_norm": 0.0023160616401582956,
-      "learning_rate": 0.00019999340948411587,
-      "loss": 46.0,
-      "step": 47808
-    },
-    {
-      "epoch": 3.6553319188791407,
-      "grad_norm": 0.0015739668160676956,
-      "learning_rate": 0.00019999340920835067,
-      "loss": 46.0,
-      "step": 47809
-    },
-    {
-      "epoch": 3.6554083758625304,
-      "grad_norm": 0.0017326497472822666,
-      "learning_rate": 0.00019999340893257973,
-      "loss": 46.0,
-      "step": 47810
-    },
-    {
-      "epoch": 3.65548483284592,
-      "grad_norm": 0.0007422473863698542,
-      "learning_rate": 0.00019999340865680298,
-      "loss": 46.0,
-      "step": 47811
-    },
-    {
-      "epoch": 3.65556128982931,
-      "grad_norm": 0.00801160465925932,
-      "learning_rate": 0.00019999340838102047,
-      "loss": 46.0,
-      "step": 47812
-    },
-    {
-      "epoch": 3.6556377468126993,
-      "grad_norm": 0.003417121944949031,
-      "learning_rate": 0.0001999934081052322,
-      "loss": 46.0,
-      "step": 47813
-    },
-    {
-      "epoch": 3.655714203796089,
-      "grad_norm": 0.00772092305123806,
-      "learning_rate": 0.00019999340782943817,
-      "loss": 46.0,
-      "step": 47814
-    },
-    {
-      "epoch": 3.655790660779479,
-      "grad_norm": 0.001235006027854979,
-      "learning_rate": 0.00019999340755363836,
-      "loss": 46.0,
-      "step": 47815
-    },
-    {
-      "epoch": 3.6558671177628685,
-      "grad_norm": 0.0017201746813952923,
-      "learning_rate": 0.00019999340727783275,
-      "loss": 46.0,
-      "step": 47816
-    },
-    {
-      "epoch": 3.6559435747462583,
-      "grad_norm": 0.0022517256438732147,
-      "learning_rate": 0.00019999340700202142,
-      "loss": 46.0,
-      "step": 47817
-    },
-    {
-      "epoch": 3.656020031729648,
-      "grad_norm": 0.0011795066529884934,
-      "learning_rate": 0.0001999934067262043,
-      "loss": 46.0,
-      "step": 47818
-    },
-    {
-      "epoch": 3.656096488713038,
-      "grad_norm": 0.0009681026567704976,
-      "learning_rate": 0.00019999340645038142,
-      "loss": 46.0,
-      "step": 47819
-    },
-    {
-      "epoch": 3.6561729456964276,
-      "grad_norm": 0.0009234437020495534,
-      "learning_rate": 0.00019999340617455277,
-      "loss": 46.0,
-      "step": 47820
-    },
-    {
-      "epoch": 3.6562494026798174,
-      "grad_norm": 0.0012534464476630092,
-      "learning_rate": 0.00019999340589871834,
-      "loss": 46.0,
-      "step": 47821
-    },
-    {
-      "epoch": 3.656325859663207,
-      "grad_norm": 0.0051234387792646885,
-      "learning_rate": 0.00019999340562287815,
-      "loss": 46.0,
-      "step": 47822
-    },
-    {
-      "epoch": 3.656402316646597,
-      "grad_norm": 0.002360963262617588,
-      "learning_rate": 0.00019999340534703218,
-      "loss": 46.0,
-      "step": 47823
-    },
-    {
-      "epoch": 3.6564787736299866,
-      "grad_norm": 0.0018381357658654451,
-      "learning_rate": 0.00019999340507118044,
-      "loss": 46.0,
-      "step": 47824
-    },
-    {
-      "epoch": 3.6565552306133764,
-      "grad_norm": 0.0019373467657715082,
-      "learning_rate": 0.00019999340479532295,
-      "loss": 46.0,
-      "step": 47825
-    },
-    {
-      "epoch": 3.6566316875967657,
-      "grad_norm": 0.0025482310447841883,
-      "learning_rate": 0.00019999340451945966,
-      "loss": 46.0,
-      "step": 47826
-    },
-    {
-      "epoch": 3.6567081445801555,
-      "grad_norm": 0.0012002121657133102,
-      "learning_rate": 0.00019999340424359062,
-      "loss": 46.0,
-      "step": 47827
-    },
-    {
-      "epoch": 3.6567846015635452,
-      "grad_norm": 0.003962427377700806,
-      "learning_rate": 0.0001999934039677158,
-      "loss": 46.0,
-      "step": 47828
-    },
-    {
-      "epoch": 3.656861058546935,
-      "grad_norm": 0.0018632502760738134,
-      "learning_rate": 0.00019999340369183526,
-      "loss": 46.0,
-      "step": 47829
-    },
-    {
-      "epoch": 3.6569375155303248,
-      "grad_norm": 0.0022463444620370865,
-      "learning_rate": 0.0001999934034159489,
-      "loss": 46.0,
-      "step": 47830
-    },
-    {
-      "epoch": 3.6570139725137145,
-      "grad_norm": 0.0019852742552757263,
-      "learning_rate": 0.00019999340314005677,
-      "loss": 46.0,
-      "step": 47831
-    },
-    {
-      "epoch": 3.6570904294971043,
-      "grad_norm": 0.00528929615393281,
-      "learning_rate": 0.0001999934028641589,
-      "loss": 46.0,
-      "step": 47832
-    },
-    {
-      "epoch": 3.657166886480494,
-      "grad_norm": 0.0008439435623586178,
-      "learning_rate": 0.00019999340258825522,
-      "loss": 46.0,
-      "step": 47833
-    },
-    {
-      "epoch": 3.6572433434638834,
-      "grad_norm": 0.0008243733900599182,
-      "learning_rate": 0.00019999340231234583,
-      "loss": 46.0,
-      "step": 47834
-    },
-    {
-      "epoch": 3.657319800447273,
-      "grad_norm": 0.0029085951391607523,
-      "learning_rate": 0.00019999340203643063,
-      "loss": 46.0,
-      "step": 47835
-    },
-    {
-      "epoch": 3.657396257430663,
-      "grad_norm": 0.0017666706116870046,
-      "learning_rate": 0.00019999340176050966,
-      "loss": 46.0,
-      "step": 47836
-    },
-    {
-      "epoch": 3.6574727144140526,
-      "grad_norm": 0.0033291089348495007,
-      "learning_rate": 0.00019999340148458292,
-      "loss": 46.0,
-      "step": 47837
-    },
-    {
-      "epoch": 3.6575491713974424,
-      "grad_norm": 0.0013852810952812433,
-      "learning_rate": 0.00019999340120865046,
-      "loss": 46.0,
-      "step": 47838
-    },
-    {
-      "epoch": 3.657625628380832,
-      "grad_norm": 0.0021935708355158567,
-      "learning_rate": 0.00019999340093271214,
-      "loss": 46.0,
-      "step": 47839
-    },
-    {
-      "epoch": 3.657702085364222,
-      "grad_norm": 0.002076232573017478,
-      "learning_rate": 0.00019999340065676813,
-      "loss": 46.0,
-      "step": 47840
-    },
-    {
-      "epoch": 3.6577785423476117,
-      "grad_norm": 0.001903512398712337,
-      "learning_rate": 0.00019999340038081832,
-      "loss": 46.0,
-      "step": 47841
-    },
-    {
-      "epoch": 3.6578549993310014,
-      "grad_norm": 0.0020118169486522675,
-      "learning_rate": 0.00019999340010486277,
-      "loss": 46.0,
-      "step": 47842
-    },
-    {
-      "epoch": 3.657931456314391,
-      "grad_norm": 0.001925047137774527,
-      "learning_rate": 0.00019999339982890139,
-      "loss": 46.0,
-      "step": 47843
-    },
-    {
-      "epoch": 3.658007913297781,
-      "grad_norm": 0.0019569192081689835,
-      "learning_rate": 0.0001999933995529343,
-      "loss": 46.0,
-      "step": 47844
-    },
-    {
-      "epoch": 3.6580843702811707,
-      "grad_norm": 0.0012307136785238981,
-      "learning_rate": 0.0001999933992769614,
-      "loss": 46.0,
-      "step": 47845
-    },
-    {
-      "epoch": 3.6581608272645605,
-      "grad_norm": 0.0006263569230213761,
-      "learning_rate": 0.00019999339900098276,
-      "loss": 46.0,
-      "step": 47846
-    },
-    {
-      "epoch": 3.6582372842479502,
-      "grad_norm": 0.004020243417471647,
-      "learning_rate": 0.0001999933987249983,
-      "loss": 46.0,
-      "step": 47847
-    },
-    {
-      "epoch": 3.6583137412313396,
-      "grad_norm": 0.0018410271732136607,
-      "learning_rate": 0.00019999339844900814,
-      "loss": 46.0,
-      "step": 47848
-    },
-    {
-      "epoch": 3.6583901982147293,
-      "grad_norm": 0.0014446190325543284,
-      "learning_rate": 0.00019999339817301217,
-      "loss": 46.0,
-      "step": 47849
-    },
-    {
-      "epoch": 3.658466655198119,
-      "grad_norm": 0.002229816745966673,
-      "learning_rate": 0.00019999339789701046,
-      "loss": 46.0,
-      "step": 47850
-    },
-    {
-      "epoch": 3.658543112181509,
-      "grad_norm": 0.003180217929184437,
-      "learning_rate": 0.00019999339762100294,
-      "loss": 46.0,
-      "step": 47851
-    },
-    {
-      "epoch": 3.6586195691648986,
-      "grad_norm": 0.0046297479420900345,
-      "learning_rate": 0.00019999339734498968,
-      "loss": 46.0,
-      "step": 47852
-    },
-    {
-      "epoch": 3.6586960261482884,
-      "grad_norm": 0.002554561011493206,
-      "learning_rate": 0.00019999339706897065,
-      "loss": 46.0,
-      "step": 47853
-    },
-    {
-      "epoch": 3.658772483131678,
-      "grad_norm": 0.0013105327961966395,
-      "learning_rate": 0.00019999339679294584,
-      "loss": 46.0,
-      "step": 47854
-    },
-    {
-      "epoch": 3.658848940115068,
-      "grad_norm": 0.0016060414491221309,
-      "learning_rate": 0.00019999339651691526,
-      "loss": 46.0,
-      "step": 47855
-    },
-    {
-      "epoch": 3.658925397098457,
-      "grad_norm": 0.003461515996605158,
-      "learning_rate": 0.00019999339624087893,
-      "loss": 46.0,
-      "step": 47856
-    },
-    {
-      "epoch": 3.659001854081847,
-      "grad_norm": 0.00162501388695091,
-      "learning_rate": 0.00019999339596483683,
-      "loss": 46.0,
-      "step": 47857
-    },
-    {
-      "epoch": 3.6590783110652367,
-      "grad_norm": 0.0016237534582614899,
-      "learning_rate": 0.00019999339568878895,
-      "loss": 46.0,
-      "step": 47858
-    },
-    {
-      "epoch": 3.6591547680486265,
-      "grad_norm": 0.0014932382619008422,
-      "learning_rate": 0.00019999339541273528,
-      "loss": 46.0,
-      "step": 47859
-    },
-    {
-      "epoch": 3.6592312250320163,
-      "grad_norm": 0.00456963200122118,
-      "learning_rate": 0.00019999339513667588,
-      "loss": 46.0,
-      "step": 47860
-    },
-    {
-      "epoch": 3.659307682015406,
-      "grad_norm": 0.003629390150308609,
-      "learning_rate": 0.0001999933948606107,
-      "loss": 46.0,
-      "step": 47861
-    },
-    {
-      "epoch": 3.6593841389987958,
-      "grad_norm": 0.003933451604098082,
-      "learning_rate": 0.00019999339458453972,
-      "loss": 46.0,
-      "step": 47862
-    },
-    {
-      "epoch": 3.6594605959821855,
-      "grad_norm": 0.0015699260402470827,
-      "learning_rate": 0.000199993394308463,
-      "loss": 46.0,
-      "step": 47863
-    },
-    {
-      "epoch": 3.6595370529655753,
-      "grad_norm": 0.0016019229078665376,
-      "learning_rate": 0.00019999339403238052,
-      "loss": 46.0,
-      "step": 47864
-    },
-    {
-      "epoch": 3.659613509948965,
-      "grad_norm": 0.0040239556692540646,
-      "learning_rate": 0.00019999339375629226,
-      "loss": 46.0,
-      "step": 47865
-    },
-    {
-      "epoch": 3.659689966932355,
-      "grad_norm": 0.002149728126823902,
-      "learning_rate": 0.00019999339348019822,
-      "loss": 46.0,
-      "step": 47866
-    },
-    {
-      "epoch": 3.6597664239157446,
-      "grad_norm": 0.0012652275618165731,
-      "learning_rate": 0.0001999933932040984,
-      "loss": 46.0,
-      "step": 47867
-    },
-    {
-      "epoch": 3.6598428808991343,
-      "grad_norm": 0.0012707667192444205,
-      "learning_rate": 0.00019999339292799283,
-      "loss": 46.0,
-      "step": 47868
-    },
-    {
-      "epoch": 3.659919337882524,
-      "grad_norm": 0.0008732926798984408,
-      "learning_rate": 0.00019999339265188148,
-      "loss": 46.0,
-      "step": 47869
-    },
-    {
-      "epoch": 3.6599957948659134,
-      "grad_norm": 0.0036312055308371782,
-      "learning_rate": 0.00019999339237576437,
-      "loss": 46.0,
-      "step": 47870
-    },
-    {
-      "epoch": 3.660072251849303,
-      "grad_norm": 0.0024095510598272085,
-      "learning_rate": 0.0001999933920996415,
-      "loss": 46.0,
-      "step": 47871
-    },
-    {
-      "epoch": 3.660148708832693,
-      "grad_norm": 0.0016875972505658865,
-      "learning_rate": 0.00019999339182351285,
-      "loss": 46.0,
-      "step": 47872
-    },
-    {
-      "epoch": 3.6602251658160827,
-      "grad_norm": 0.002364052925258875,
-      "learning_rate": 0.00019999339154737843,
-      "loss": 46.0,
-      "step": 47873
-    },
-    {
-      "epoch": 3.6603016227994725,
-      "grad_norm": 0.0012726550921797752,
-      "learning_rate": 0.00019999339127123826,
-      "loss": 46.0,
-      "step": 47874
-    },
-    {
-      "epoch": 3.6603780797828622,
-      "grad_norm": 0.0022818357683718204,
-      "learning_rate": 0.0001999933909950923,
-      "loss": 46.0,
-      "step": 47875
-    },
-    {
-      "epoch": 3.660454536766252,
-      "grad_norm": 0.0023524975404143333,
-      "learning_rate": 0.00019999339071894055,
-      "loss": 46.0,
-      "step": 47876
-    },
-    {
-      "epoch": 3.6605309937496417,
-      "grad_norm": 0.0013641039840877056,
-      "learning_rate": 0.0001999933904427831,
-      "loss": 46.0,
-      "step": 47877
-    },
-    {
-      "epoch": 3.660607450733031,
-      "grad_norm": 0.0009009288623929024,
-      "learning_rate": 0.00019999339016661983,
-      "loss": 46.0,
-      "step": 47878
-    },
-    {
-      "epoch": 3.660683907716421,
-      "grad_norm": 0.0036308083217591047,
-      "learning_rate": 0.00019999338989045077,
-      "loss": 46.0,
-      "step": 47879
-    },
-    {
-      "epoch": 3.6607603646998106,
-      "grad_norm": 0.0015181595226749778,
-      "learning_rate": 0.000199993389614276,
-      "loss": 46.0,
-      "step": 47880
-    },
-    {
-      "epoch": 3.6608368216832003,
-      "grad_norm": 0.0026797018945217133,
-      "learning_rate": 0.0001999933893380954,
-      "loss": 46.0,
-      "step": 47881
-    },
-    {
-      "epoch": 3.66091327866659,
-      "grad_norm": 0.006255386862903833,
-      "learning_rate": 0.00019999338906190908,
-      "loss": 46.0,
-      "step": 47882
-    },
-    {
-      "epoch": 3.66098973564998,
-      "grad_norm": 0.0024160905741155148,
-      "learning_rate": 0.00019999338878571695,
-      "loss": 46.0,
-      "step": 47883
-    },
-    {
-      "epoch": 3.6610661926333696,
-      "grad_norm": 0.0035971486940979958,
-      "learning_rate": 0.0001999933885095191,
-      "loss": 46.0,
-      "step": 47884
-    },
-    {
-      "epoch": 3.6611426496167594,
-      "grad_norm": 0.002197209745645523,
-      "learning_rate": 0.00019999338823331543,
-      "loss": 46.0,
-      "step": 47885
-    },
-    {
-      "epoch": 3.661219106600149,
-      "grad_norm": 0.0032732898835092783,
-      "learning_rate": 0.00019999338795710604,
-      "loss": 46.0,
-      "step": 47886
-    },
-    {
-      "epoch": 3.661295563583539,
-      "grad_norm": 0.0031460553873330355,
-      "learning_rate": 0.00019999338768089084,
-      "loss": 46.0,
-      "step": 47887
-    },
-    {
-      "epoch": 3.6613720205669287,
-      "grad_norm": 0.005199407692998648,
-      "learning_rate": 0.00019999338740466987,
-      "loss": 46.0,
-      "step": 47888
-    },
-    {
-      "epoch": 3.6614484775503184,
-      "grad_norm": 0.0018900011200457811,
-      "learning_rate": 0.0001999933871284432,
-      "loss": 46.0,
-      "step": 47889
-    },
-    {
-      "epoch": 3.661524934533708,
-      "grad_norm": 0.0018733100732788444,
-      "learning_rate": 0.00019999338685221067,
-      "loss": 46.0,
-      "step": 47890
-    },
-    {
-      "epoch": 3.661601391517098,
-      "grad_norm": 0.0031615784391760826,
-      "learning_rate": 0.0001999933865759724,
-      "loss": 46.0,
-      "step": 47891
-    },
-    {
-      "epoch": 3.6616778485004873,
-      "grad_norm": 0.0007877142634242773,
-      "learning_rate": 0.00019999338629972838,
-      "loss": 46.0,
-      "step": 47892
-    },
-    {
-      "epoch": 3.661754305483877,
-      "grad_norm": 0.002684255363419652,
-      "learning_rate": 0.0001999933860234786,
-      "loss": 46.0,
-      "step": 47893
-    },
-    {
-      "epoch": 3.661830762467267,
-      "grad_norm": 0.0026270037051290274,
-      "learning_rate": 0.00019999338574722302,
-      "loss": 46.0,
-      "step": 47894
-    },
-    {
-      "epoch": 3.6619072194506566,
-      "grad_norm": 0.0006175162270665169,
-      "learning_rate": 0.0001999933854709617,
-      "loss": 46.0,
-      "step": 47895
-    },
-    {
-      "epoch": 3.6619836764340463,
-      "grad_norm": 0.0007937418995425105,
-      "learning_rate": 0.00019999338519469456,
-      "loss": 46.0,
-      "step": 47896
-    },
-    {
-      "epoch": 3.662060133417436,
-      "grad_norm": 0.003095922525972128,
-      "learning_rate": 0.0001999933849184217,
-      "loss": 46.0,
-      "step": 47897
-    },
-    {
-      "epoch": 3.662136590400826,
-      "grad_norm": 0.00045074871741235256,
-      "learning_rate": 0.00019999338464214307,
-      "loss": 46.0,
-      "step": 47898
-    },
-    {
-      "epoch": 3.6622130473842156,
-      "grad_norm": 0.0017013788456097245,
-      "learning_rate": 0.00019999338436585865,
-      "loss": 46.0,
-      "step": 47899
-    },
-    {
-      "epoch": 3.662289504367605,
-      "grad_norm": 0.004155789036303759,
-      "learning_rate": 0.00019999338408956848,
-      "loss": 46.0,
-      "step": 47900
-    },
-    {
-      "epoch": 3.6623659613509947,
-      "grad_norm": 0.0018713424215093255,
-      "learning_rate": 0.0001999933838132725,
-      "loss": 46.0,
-      "step": 47901
-    },
-    {
-      "epoch": 3.6624424183343844,
-      "grad_norm": 0.0023316089063882828,
-      "learning_rate": 0.0001999933835369708,
-      "loss": 46.0,
-      "step": 47902
-    },
-    {
-      "epoch": 3.662518875317774,
-      "grad_norm": 0.00036295020254328847,
-      "learning_rate": 0.0001999933832606633,
-      "loss": 46.0,
-      "step": 47903
-    },
-    {
-      "epoch": 3.662595332301164,
-      "grad_norm": 0.002408041153103113,
-      "learning_rate": 0.00019999338298435005,
-      "loss": 46.0,
-      "step": 47904
-    },
-    {
-      "epoch": 3.6626717892845537,
-      "grad_norm": 0.0035913202445954084,
-      "learning_rate": 0.00019999338270803102,
-      "loss": 46.0,
-      "step": 47905
-    },
-    {
-      "epoch": 3.6627482462679435,
-      "grad_norm": 0.002356767188757658,
-      "learning_rate": 0.0001999933824317062,
-      "loss": 46.0,
-      "step": 47906
-    },
-    {
-      "epoch": 3.6628247032513332,
-      "grad_norm": 0.0008597004343755543,
-      "learning_rate": 0.00019999338215537566,
-      "loss": 46.0,
-      "step": 47907
-    },
-    {
-      "epoch": 3.662901160234723,
-      "grad_norm": 0.002325072418898344,
-      "learning_rate": 0.00019999338187903933,
-      "loss": 46.0,
-      "step": 47908
-    },
-    {
-      "epoch": 3.6629776172181128,
-      "grad_norm": 0.0016015819273889065,
-      "learning_rate": 0.00019999338160269723,
-      "loss": 46.0,
-      "step": 47909
-    },
-    {
-      "epoch": 3.6630540742015025,
-      "grad_norm": 0.0022232159972190857,
-      "learning_rate": 0.00019999338132634936,
-      "loss": 46.0,
-      "step": 47910
-    },
-    {
-      "epoch": 3.6631305311848923,
-      "grad_norm": 0.0005791396833956242,
-      "learning_rate": 0.0001999933810499957,
-      "loss": 46.0,
-      "step": 47911
-    },
-    {
-      "epoch": 3.663206988168282,
-      "grad_norm": 0.00144652696326375,
-      "learning_rate": 0.00019999338077363626,
-      "loss": 46.0,
-      "step": 47912
-    },
-    {
-      "epoch": 3.663283445151672,
-      "grad_norm": 0.003915582317858934,
-      "learning_rate": 0.00019999338049727112,
-      "loss": 46.0,
-      "step": 47913
-    },
-    {
-      "epoch": 3.663359902135061,
-      "grad_norm": 0.0019386752974241972,
-      "learning_rate": 0.00019999338022090016,
-      "loss": 46.0,
-      "step": 47914
-    },
-    {
-      "epoch": 3.663436359118451,
-      "grad_norm": 0.0019776534754782915,
-      "learning_rate": 0.00019999337994452344,
-      "loss": 46.0,
-      "step": 47915
-    },
-    {
-      "epoch": 3.6635128161018407,
-      "grad_norm": 0.00853799656033516,
-      "learning_rate": 0.00019999337966814096,
-      "loss": 46.0,
-      "step": 47916
-    },
-    {
-      "epoch": 3.6635892730852304,
-      "grad_norm": 0.0011911524925380945,
-      "learning_rate": 0.0001999933793917527,
-      "loss": 46.0,
-      "step": 47917
-    },
-    {
-      "epoch": 3.66366573006862,
-      "grad_norm": 0.00240787654183805,
-      "learning_rate": 0.00019999337911535866,
-      "loss": 46.0,
-      "step": 47918
-    },
-    {
-      "epoch": 3.66374218705201,
-      "grad_norm": 0.0019760397262871265,
-      "learning_rate": 0.00019999337883895888,
-      "loss": 46.0,
-      "step": 47919
-    },
-    {
-      "epoch": 3.6638186440353997,
-      "grad_norm": 0.0006879131542518735,
-      "learning_rate": 0.0001999933785625533,
-      "loss": 46.0,
-      "step": 47920
-    },
-    {
-      "epoch": 3.6638951010187895,
-      "grad_norm": 0.0018098466098308563,
-      "learning_rate": 0.00019999337828614197,
-      "loss": 46.0,
-      "step": 47921
-    },
-    {
-      "epoch": 3.6639715580021788,
-      "grad_norm": 0.002037947066128254,
-      "learning_rate": 0.00019999337800972488,
-      "loss": 46.0,
-      "step": 47922
-    },
-    {
-      "epoch": 3.6640480149855685,
-      "grad_norm": 0.0010829000966623425,
-      "learning_rate": 0.00019999337773330203,
-      "loss": 46.0,
-      "step": 47923
-    },
-    {
-      "epoch": 3.6641244719689583,
-      "grad_norm": 0.0010545499389991164,
-      "learning_rate": 0.00019999337745687336,
-      "loss": 46.0,
-      "step": 47924
-    },
-    {
-      "epoch": 3.664200928952348,
-      "grad_norm": 0.0015009538037702441,
-      "learning_rate": 0.00019999337718043896,
-      "loss": 46.0,
-      "step": 47925
-    },
-    {
-      "epoch": 3.664277385935738,
-      "grad_norm": 0.0018316813511773944,
-      "learning_rate": 0.00019999337690399877,
-      "loss": 46.0,
-      "step": 47926
-    },
-    {
-      "epoch": 3.6643538429191276,
-      "grad_norm": 0.003338324837386608,
-      "learning_rate": 0.00019999337662755283,
-      "loss": 46.0,
-      "step": 47927
-    },
-    {
-      "epoch": 3.6644302999025173,
-      "grad_norm": 0.00455842399969697,
-      "learning_rate": 0.00019999337635110114,
-      "loss": 46.0,
-      "step": 47928
-    },
-    {
-      "epoch": 3.664506756885907,
-      "grad_norm": 0.0039513977244496346,
-      "learning_rate": 0.00019999337607464363,
-      "loss": 46.0,
-      "step": 47929
-    },
-    {
-      "epoch": 3.664583213869297,
-      "grad_norm": 0.0030990089289844036,
-      "learning_rate": 0.00019999337579818037,
-      "loss": 46.0,
-      "step": 47930
-    },
-    {
-      "epoch": 3.6646596708526866,
-      "grad_norm": 0.0012320346431806684,
-      "learning_rate": 0.00019999337552171136,
-      "loss": 46.0,
-      "step": 47931
-    },
-    {
-      "epoch": 3.6647361278360764,
-      "grad_norm": 0.0018863289151340723,
-      "learning_rate": 0.00019999337524523658,
-      "loss": 46.0,
-      "step": 47932
-    },
-    {
-      "epoch": 3.664812584819466,
-      "grad_norm": 0.001923258532769978,
-      "learning_rate": 0.000199993374968756,
-      "loss": 46.0,
-      "step": 47933
-    },
-    {
-      "epoch": 3.664889041802856,
-      "grad_norm": 0.0020726253278553486,
-      "learning_rate": 0.00019999337469226965,
-      "loss": 46.0,
-      "step": 47934
-    },
-    {
-      "epoch": 3.6649654987862452,
-      "grad_norm": 0.0023975956719368696,
-      "learning_rate": 0.00019999337441577758,
-      "loss": 46.0,
-      "step": 47935
-    },
-    {
-      "epoch": 3.665041955769635,
-      "grad_norm": 0.0032305391505360603,
-      "learning_rate": 0.0001999933741392797,
-      "loss": 46.0,
-      "step": 47936
-    },
-    {
-      "epoch": 3.6651184127530247,
-      "grad_norm": 0.005718642380088568,
-      "learning_rate": 0.00019999337386277609,
-      "loss": 46.0,
-      "step": 47937
-    },
-    {
-      "epoch": 3.6651948697364145,
-      "grad_norm": 0.003722132882103324,
-      "learning_rate": 0.00019999337358626667,
-      "loss": 46.0,
-      "step": 47938
-    },
-    {
-      "epoch": 3.6652713267198043,
-      "grad_norm": 0.0010933848097920418,
-      "learning_rate": 0.00019999337330975147,
-      "loss": 46.0,
-      "step": 47939
-    },
-    {
-      "epoch": 3.665347783703194,
-      "grad_norm": 0.002146756509318948,
-      "learning_rate": 0.00019999337303323053,
-      "loss": 46.0,
-      "step": 47940
-    },
-    {
-      "epoch": 3.665424240686584,
-      "grad_norm": 0.0013325114268809557,
-      "learning_rate": 0.00019999337275670382,
-      "loss": 46.0,
-      "step": 47941
-    },
-    {
-      "epoch": 3.6655006976699736,
-      "grad_norm": 0.0023758369497954845,
-      "learning_rate": 0.00019999337248017133,
-      "loss": 46.0,
-      "step": 47942
-    },
-    {
-      "epoch": 3.6655771546533633,
-      "grad_norm": 0.001031380146741867,
-      "learning_rate": 0.00019999337220363308,
-      "loss": 46.0,
-      "step": 47943
-    },
-    {
-      "epoch": 3.6656536116367526,
-      "grad_norm": 0.00375778297893703,
-      "learning_rate": 0.00019999337192708907,
-      "loss": 46.0,
-      "step": 47944
-    },
-    {
-      "epoch": 3.6657300686201424,
-      "grad_norm": 0.0013113364111632109,
-      "learning_rate": 0.0001999933716505393,
-      "loss": 46.0,
-      "step": 47945
-    },
-    {
-      "epoch": 3.665806525603532,
-      "grad_norm": 0.003102717688307166,
-      "learning_rate": 0.0001999933713739837,
-      "loss": 46.0,
-      "step": 47946
-    },
-    {
-      "epoch": 3.665882982586922,
-      "grad_norm": 0.002812005812302232,
-      "learning_rate": 0.0001999933710974224,
-      "loss": 46.0,
-      "step": 47947
-    },
-    {
-      "epoch": 3.6659594395703117,
-      "grad_norm": 0.0030215787701308727,
-      "learning_rate": 0.00019999337082085529,
-      "loss": 46.0,
-      "step": 47948
-    },
-    {
-      "epoch": 3.6660358965537014,
-      "grad_norm": 0.0012319745728746057,
-      "learning_rate": 0.0001999933705442824,
-      "loss": 46.0,
-      "step": 47949
-    },
-    {
-      "epoch": 3.666112353537091,
-      "grad_norm": 0.000795715197455138,
-      "learning_rate": 0.0001999933702677038,
-      "loss": 46.0,
-      "step": 47950
-    },
-    {
-      "epoch": 3.666188810520481,
-      "grad_norm": 0.0022924127988517284,
-      "learning_rate": 0.0001999933699911194,
-      "loss": 46.0,
-      "step": 47951
-    },
-    {
-      "epoch": 3.6662652675038707,
-      "grad_norm": 0.0058022141456604,
-      "learning_rate": 0.0001999933697145292,
-      "loss": 46.0,
-      "step": 47952
-    },
-    {
-      "epoch": 3.6663417244872605,
-      "grad_norm": 0.0016654469072818756,
-      "learning_rate": 0.00019999336943793327,
-      "loss": 46.0,
-      "step": 47953
-    },
-    {
-      "epoch": 3.6664181814706502,
-      "grad_norm": 0.00541280722245574,
-      "learning_rate": 0.00019999336916133156,
-      "loss": 46.0,
-      "step": 47954
-    },
-    {
-      "epoch": 3.66649463845404,
-      "grad_norm": 0.001333836349658668,
-      "learning_rate": 0.00019999336888472407,
-      "loss": 46.0,
-      "step": 47955
-    },
-    {
-      "epoch": 3.6665710954374298,
-      "grad_norm": 0.0035815811716020107,
-      "learning_rate": 0.00019999336860811084,
-      "loss": 46.0,
-      "step": 47956
-    },
-    {
-      "epoch": 3.666647552420819,
-      "grad_norm": 0.0011034669587388635,
-      "learning_rate": 0.0001999933683314918,
-      "loss": 46.0,
-      "step": 47957
-    },
-    {
-      "epoch": 3.666724009404209,
-      "grad_norm": 0.0024126798380166292,
-      "learning_rate": 0.000199993368054867,
-      "loss": 46.0,
-      "step": 47958
-    },
-    {
-      "epoch": 3.6668004663875986,
-      "grad_norm": 0.0024243113584816456,
-      "learning_rate": 0.00019999336777823645,
-      "loss": 46.0,
-      "step": 47959
-    },
-    {
-      "epoch": 3.6668769233709884,
-      "grad_norm": 0.0011374404421076179,
-      "learning_rate": 0.00019999336750160015,
-      "loss": 46.0,
-      "step": 47960
-    },
-    {
-      "epoch": 3.666953380354378,
-      "grad_norm": 0.0014976942911744118,
-      "learning_rate": 0.00019999336722495805,
-      "loss": 46.0,
-      "step": 47961
-    },
-    {
-      "epoch": 3.667029837337768,
-      "grad_norm": 0.0009365262230858207,
-      "learning_rate": 0.00019999336694831018,
-      "loss": 46.0,
-      "step": 47962
-    },
-    {
-      "epoch": 3.6671062943211576,
-      "grad_norm": 0.0031736285891383886,
-      "learning_rate": 0.00019999336667165656,
-      "loss": 46.0,
-      "step": 47963
-    },
-    {
-      "epoch": 3.6671827513045474,
-      "grad_norm": 0.0035877984482795,
-      "learning_rate": 0.00019999336639499714,
-      "loss": 46.0,
-      "step": 47964
-    },
-    {
-      "epoch": 3.6672592082879367,
-      "grad_norm": 0.0016949651762843132,
-      "learning_rate": 0.00019999336611833198,
-      "loss": 46.0,
-      "step": 47965
-    },
-    {
-      "epoch": 3.6673356652713265,
-      "grad_norm": 0.0015173788415268064,
-      "learning_rate": 0.000199993365841661,
-      "loss": 46.0,
-      "step": 47966
-    },
-    {
-      "epoch": 3.6674121222547162,
-      "grad_norm": 0.0024319153744727373,
-      "learning_rate": 0.00019999336556498432,
-      "loss": 46.0,
-      "step": 47967
-    },
-    {
-      "epoch": 3.667488579238106,
-      "grad_norm": 0.0016374427359551191,
-      "learning_rate": 0.00019999336528830184,
-      "loss": 46.0,
-      "step": 47968
-    },
-    {
-      "epoch": 3.6675650362214958,
-      "grad_norm": 0.0032984784338623285,
-      "learning_rate": 0.00019999336501161358,
-      "loss": 46.0,
-      "step": 47969
-    },
-    {
-      "epoch": 3.6676414932048855,
-      "grad_norm": 0.0014317476889118552,
-      "learning_rate": 0.00019999336473491957,
-      "loss": 46.0,
-      "step": 47970
-    },
-    {
-      "epoch": 3.6677179501882753,
-      "grad_norm": 0.0006523602642118931,
-      "learning_rate": 0.0001999933644582198,
-      "loss": 46.0,
-      "step": 47971
-    },
-    {
-      "epoch": 3.667794407171665,
-      "grad_norm": 0.0021338271908462048,
-      "learning_rate": 0.00019999336418151424,
-      "loss": 46.0,
-      "step": 47972
-    },
-    {
-      "epoch": 3.667870864155055,
-      "grad_norm": 0.005957825109362602,
-      "learning_rate": 0.00019999336390480292,
-      "loss": 46.0,
-      "step": 47973
-    },
-    {
-      "epoch": 3.6679473211384446,
-      "grad_norm": 0.0013046445092186332,
-      "learning_rate": 0.00019999336362808582,
-      "loss": 46.0,
-      "step": 47974
-    },
-    {
-      "epoch": 3.6680237781218343,
-      "grad_norm": 0.0011419422226026654,
-      "learning_rate": 0.00019999336335136295,
-      "loss": 46.0,
-      "step": 47975
-    },
-    {
-      "epoch": 3.668100235105224,
-      "grad_norm": 0.0027256333269178867,
-      "learning_rate": 0.00019999336307463433,
-      "loss": 46.0,
-      "step": 47976
-    },
-    {
-      "epoch": 3.668176692088614,
-      "grad_norm": 0.0008650197414681315,
-      "learning_rate": 0.0001999933627978999,
-      "loss": 46.0,
-      "step": 47977
-    },
-    {
-      "epoch": 3.6682531490720036,
-      "grad_norm": 0.0008292000275105238,
-      "learning_rate": 0.00019999336252115975,
-      "loss": 46.0,
-      "step": 47978
-    },
-    {
-      "epoch": 3.668329606055393,
-      "grad_norm": 0.001185743254609406,
-      "learning_rate": 0.0001999933622444138,
-      "loss": 46.0,
-      "step": 47979
-    },
-    {
-      "epoch": 3.6684060630387827,
-      "grad_norm": 0.0024928064085543156,
-      "learning_rate": 0.00019999336196766212,
-      "loss": 46.0,
-      "step": 47980
-    },
-    {
-      "epoch": 3.6684825200221725,
-      "grad_norm": 0.000558042898774147,
-      "learning_rate": 0.0001999933616909046,
-      "loss": 46.0,
-      "step": 47981
-    },
-    {
-      "epoch": 3.668558977005562,
-      "grad_norm": 0.0007560640806332231,
-      "learning_rate": 0.00019999336141414138,
-      "loss": 46.0,
-      "step": 47982
-    },
-    {
-      "epoch": 3.668635433988952,
-      "grad_norm": 0.004348853137344122,
-      "learning_rate": 0.00019999336113737237,
-      "loss": 46.0,
-      "step": 47983
-    },
-    {
-      "epoch": 3.6687118909723417,
-      "grad_norm": 0.0022174064069986343,
-      "learning_rate": 0.0001999933608605976,
-      "loss": 46.0,
-      "step": 47984
-    },
-    {
-      "epoch": 3.6687883479557315,
-      "grad_norm": 0.0019662310369312763,
-      "learning_rate": 0.00019999336058381704,
-      "loss": 46.0,
-      "step": 47985
-    },
-    {
-      "epoch": 3.6688648049391213,
-      "grad_norm": 0.0016567886341363192,
-      "learning_rate": 0.00019999336030703072,
-      "loss": 46.0,
-      "step": 47986
-    },
-    {
-      "epoch": 3.6689412619225106,
-      "grad_norm": 0.0020351523999124765,
-      "learning_rate": 0.00019999336003023862,
-      "loss": 46.0,
-      "step": 47987
-    },
-    {
-      "epoch": 3.6690177189059003,
-      "grad_norm": 0.002732141176238656,
-      "learning_rate": 0.00019999335975344075,
-      "loss": 46.0,
-      "step": 47988
-    },
-    {
-      "epoch": 3.66909417588929,
-      "grad_norm": 0.00147348758764565,
-      "learning_rate": 0.00019999335947663713,
-      "loss": 46.0,
-      "step": 47989
-    },
-    {
-      "epoch": 3.66917063287268,
-      "grad_norm": 0.0009496886632405221,
-      "learning_rate": 0.0001999933591998277,
-      "loss": 46.0,
-      "step": 47990
-    },
-    {
-      "epoch": 3.6692470898560696,
-      "grad_norm": 0.0015586194349452853,
-      "learning_rate": 0.00019999335892301258,
-      "loss": 46.0,
-      "step": 47991
-    },
-    {
-      "epoch": 3.6693235468394594,
-      "grad_norm": 0.005772904492914677,
-      "learning_rate": 0.0001999933586461916,
-      "loss": 46.0,
-      "step": 47992
-    },
-    {
-      "epoch": 3.669400003822849,
-      "grad_norm": 0.001957288244739175,
-      "learning_rate": 0.0001999933583693649,
-      "loss": 46.0,
-      "step": 47993
-    },
-    {
-      "epoch": 3.669476460806239,
-      "grad_norm": 0.001681795227341354,
-      "learning_rate": 0.00019999335809253244,
-      "loss": 46.0,
-      "step": 47994
-    },
-    {
-      "epoch": 3.6695529177896287,
-      "grad_norm": 0.0008049028110690415,
-      "learning_rate": 0.00019999335781569418,
-      "loss": 46.0,
-      "step": 47995
-    },
-    {
-      "epoch": 3.6696293747730184,
-      "grad_norm": 0.002104239072650671,
-      "learning_rate": 0.00019999335753885018,
-      "loss": 46.0,
-      "step": 47996
-    },
-    {
-      "epoch": 3.669705831756408,
-      "grad_norm": 0.0034681379329413176,
-      "learning_rate": 0.0001999933572620004,
-      "loss": 46.0,
-      "step": 47997
-    },
-    {
-      "epoch": 3.669782288739798,
-      "grad_norm": 0.0028721061535179615,
-      "learning_rate": 0.00019999335698514485,
-      "loss": 46.0,
-      "step": 47998
-    },
-    {
-      "epoch": 3.6698587457231877,
-      "grad_norm": 0.0016494493465870619,
-      "learning_rate": 0.00019999335670828355,
-      "loss": 46.0,
-      "step": 47999
-    },
-    {
-      "epoch": 3.6699352027065775,
-      "grad_norm": 0.003336281981319189,
-      "learning_rate": 0.00019999335643141643,
-      "loss": 46.0,
-      "step": 48000
-    },
-    {
-      "epoch": 3.670011659689967,
-      "grad_norm": 0.0013646723236888647,
-      "learning_rate": 0.0001999933561545436,
-      "loss": 46.0,
-      "step": 48001
-    },
-    {
-      "epoch": 3.6700881166733565,
-      "grad_norm": 0.0013325696345418692,
-      "learning_rate": 0.00019999335587766494,
-      "loss": 46.0,
-      "step": 48002
-    },
-    {
-      "epoch": 3.6701645736567463,
-      "grad_norm": 0.0005511973286047578,
-      "learning_rate": 0.00019999335560078057,
-      "loss": 46.0,
-      "step": 48003
-    },
-    {
-      "epoch": 3.670241030640136,
-      "grad_norm": 0.0010190497850999236,
-      "learning_rate": 0.0001999933553238904,
-      "loss": 46.0,
-      "step": 48004
-    },
-    {
-      "epoch": 3.670317487623526,
-      "grad_norm": 0.001558575313538313,
-      "learning_rate": 0.00019999335504699444,
-      "loss": 46.0,
-      "step": 48005
-    },
-    {
-      "epoch": 3.6703939446069156,
-      "grad_norm": 0.0013280980056151748,
-      "learning_rate": 0.00019999335477009276,
-      "loss": 46.0,
-      "step": 48006
-    },
-    {
-      "epoch": 3.6704704015903054,
-      "grad_norm": 0.002202605130150914,
-      "learning_rate": 0.00019999335449318528,
-      "loss": 46.0,
-      "step": 48007
-    },
-    {
-      "epoch": 3.670546858573695,
-      "grad_norm": 0.002156279981136322,
-      "learning_rate": 0.00019999335421627204,
-      "loss": 46.0,
-      "step": 48008
-    },
-    {
-      "epoch": 3.6706233155570844,
-      "grad_norm": 0.0024312122259289026,
-      "learning_rate": 0.00019999335393935304,
-      "loss": 46.0,
-      "step": 48009
-    },
-    {
-      "epoch": 3.670699772540474,
-      "grad_norm": 0.0006952566909603775,
-      "learning_rate": 0.00019999335366242824,
-      "loss": 46.0,
-      "step": 48010
-    },
-    {
-      "epoch": 3.670776229523864,
-      "grad_norm": 0.0021471420768648386,
-      "learning_rate": 0.00019999335338549768,
-      "loss": 46.0,
-      "step": 48011
-    },
-    {
-      "epoch": 3.6708526865072537,
-      "grad_norm": 0.001348915626294911,
-      "learning_rate": 0.0001999933531085614,
-      "loss": 46.0,
-      "step": 48012
-    },
-    {
-      "epoch": 3.6709291434906435,
-      "grad_norm": 0.001189972972497344,
-      "learning_rate": 0.00019999335283161932,
-      "loss": 46.0,
-      "step": 48013
-    },
-    {
-      "epoch": 3.6710056004740332,
-      "grad_norm": 0.0005687892553396523,
-      "learning_rate": 0.00019999335255467145,
-      "loss": 46.0,
-      "step": 48014
-    },
-    {
-      "epoch": 3.671082057457423,
-      "grad_norm": 0.002547189127653837,
-      "learning_rate": 0.0001999933522777178,
-      "loss": 46.0,
-      "step": 48015
-    },
-    {
-      "epoch": 3.6711585144408128,
-      "grad_norm": 0.000822422094643116,
-      "learning_rate": 0.0001999933520007584,
-      "loss": 46.0,
-      "step": 48016
-    },
-    {
-      "epoch": 3.6712349714242025,
-      "grad_norm": 0.0013455809094011784,
-      "learning_rate": 0.00019999335172379325,
-      "loss": 46.0,
-      "step": 48017
-    },
-    {
-      "epoch": 3.6713114284075923,
-      "grad_norm": 0.00370883964933455,
-      "learning_rate": 0.0001999933514468223,
-      "loss": 46.0,
-      "step": 48018
-    },
-    {
-      "epoch": 3.671387885390982,
-      "grad_norm": 0.0012432014336809516,
-      "learning_rate": 0.0001999933511698456,
-      "loss": 46.0,
-      "step": 48019
-    },
-    {
-      "epoch": 3.671464342374372,
-      "grad_norm": 0.0033680452033877373,
-      "learning_rate": 0.00019999335089286314,
-      "loss": 46.0,
-      "step": 48020
-    },
-    {
-      "epoch": 3.6715407993577616,
-      "grad_norm": 0.0021840636618435383,
-      "learning_rate": 0.0001999933506158749,
-      "loss": 46.0,
-      "step": 48021
-    },
-    {
-      "epoch": 3.6716172563411513,
-      "grad_norm": 0.002178065711632371,
-      "learning_rate": 0.00019999335033888088,
-      "loss": 46.0,
-      "step": 48022
-    },
-    {
-      "epoch": 3.6716937133245406,
-      "grad_norm": 0.005367441102862358,
-      "learning_rate": 0.00019999335006188113,
-      "loss": 46.0,
-      "step": 48023
-    },
-    {
-      "epoch": 3.6717701703079304,
-      "grad_norm": 0.004774544853717089,
-      "learning_rate": 0.00019999334978487555,
-      "loss": 46.0,
-      "step": 48024
-    },
-    {
-      "epoch": 3.67184662729132,
-      "grad_norm": 0.002754293382167816,
-      "learning_rate": 0.00019999334950786425,
-      "loss": 46.0,
-      "step": 48025
-    },
-    {
-      "epoch": 3.67192308427471,
-      "grad_norm": 0.002541779074817896,
-      "learning_rate": 0.00019999334923084718,
-      "loss": 46.0,
-      "step": 48026
-    },
-    {
-      "epoch": 3.6719995412580997,
-      "grad_norm": 0.0037251540925353765,
-      "learning_rate": 0.00019999334895382434,
-      "loss": 46.0,
-      "step": 48027
-    },
-    {
-      "epoch": 3.6720759982414894,
-      "grad_norm": 0.0019251458579674363,
-      "learning_rate": 0.00019999334867679572,
-      "loss": 46.0,
-      "step": 48028
-    },
-    {
-      "epoch": 3.672152455224879,
-      "grad_norm": 0.0005644903285428882,
-      "learning_rate": 0.0001999933483997613,
-      "loss": 46.0,
-      "step": 48029
-    },
-    {
-      "epoch": 3.672228912208269,
-      "grad_norm": 0.0025510292034596205,
-      "learning_rate": 0.00019999334812272114,
-      "loss": 46.0,
-      "step": 48030
-    },
-    {
-      "epoch": 3.6723053691916583,
-      "grad_norm": 0.0013216182123869658,
-      "learning_rate": 0.0001999933478456752,
-      "loss": 46.0,
-      "step": 48031
-    },
-    {
-      "epoch": 3.672381826175048,
-      "grad_norm": 0.0009647682309150696,
-      "learning_rate": 0.00019999334756862352,
-      "loss": 46.0,
-      "step": 48032
-    },
-    {
-      "epoch": 3.672458283158438,
-      "grad_norm": 0.0012674200115725398,
-      "learning_rate": 0.00019999334729156604,
-      "loss": 46.0,
-      "step": 48033
-    },
-    {
-      "epoch": 3.6725347401418276,
-      "grad_norm": 0.0020439722575247288,
-      "learning_rate": 0.0001999933470145028,
-      "loss": 46.0,
-      "step": 48034
-    },
-    {
-      "epoch": 3.6726111971252173,
-      "grad_norm": 0.00381052796728909,
-      "learning_rate": 0.0001999933467374338,
-      "loss": 46.0,
-      "step": 48035
-    },
-    {
-      "epoch": 3.672687654108607,
-      "grad_norm": 0.00462719239294529,
-      "learning_rate": 0.00019999334646035903,
-      "loss": 46.0,
-      "step": 48036
-    },
-    {
-      "epoch": 3.672764111091997,
-      "grad_norm": 0.002047655638307333,
-      "learning_rate": 0.0001999933461832785,
-      "loss": 46.0,
-      "step": 48037
-    },
-    {
-      "epoch": 3.6728405680753866,
-      "grad_norm": 0.0014202672755345702,
-      "learning_rate": 0.00019999334590619215,
-      "loss": 46.0,
-      "step": 48038
-    },
-    {
-      "epoch": 3.6729170250587764,
-      "grad_norm": 0.006818561814725399,
-      "learning_rate": 0.00019999334562910008,
-      "loss": 46.0,
-      "step": 48039
-    },
-    {
-      "epoch": 3.672993482042166,
-      "grad_norm": 0.003302361350506544,
-      "learning_rate": 0.00019999334535200224,
-      "loss": 46.0,
-      "step": 48040
-    },
-    {
-      "epoch": 3.673069939025556,
-      "grad_norm": 0.001957076834514737,
-      "learning_rate": 0.00019999334507489862,
-      "loss": 46.0,
-      "step": 48041
-    },
-    {
-      "epoch": 3.6731463960089457,
-      "grad_norm": 0.002648213878273964,
-      "learning_rate": 0.0001999933447977892,
-      "loss": 46.0,
-      "step": 48042
-    },
-    {
-      "epoch": 3.6732228529923354,
-      "grad_norm": 0.001613419852219522,
-      "learning_rate": 0.00019999334452067407,
-      "loss": 46.0,
-      "step": 48043
-    },
-    {
-      "epoch": 3.673299309975725,
-      "grad_norm": 0.001669177319854498,
-      "learning_rate": 0.00019999334424355313,
-      "loss": 46.0,
-      "step": 48044
-    },
-    {
-      "epoch": 3.6733757669591145,
-      "grad_norm": 0.004495824687182903,
-      "learning_rate": 0.00019999334396642645,
-      "loss": 46.0,
-      "step": 48045
-    },
-    {
-      "epoch": 3.6734522239425043,
-      "grad_norm": 0.0020188684575259686,
-      "learning_rate": 0.00019999334368929396,
-      "loss": 46.0,
-      "step": 48046
-    },
-    {
-      "epoch": 3.673528680925894,
-      "grad_norm": 0.0017921437975019217,
-      "learning_rate": 0.00019999334341215573,
-      "loss": 46.0,
-      "step": 48047
-    },
-    {
-      "epoch": 3.673605137909284,
-      "grad_norm": 0.003095908323302865,
-      "learning_rate": 0.00019999334313501176,
-      "loss": 46.0,
-      "step": 48048
-    },
-    {
-      "epoch": 3.6736815948926735,
-      "grad_norm": 0.00504260091111064,
-      "learning_rate": 0.00019999334285786196,
-      "loss": 46.0,
-      "step": 48049
-    },
-    {
-      "epoch": 3.6737580518760633,
-      "grad_norm": 0.0016842473996803164,
-      "learning_rate": 0.0001999933425807064,
-      "loss": 46.0,
-      "step": 48050
-    },
-    {
-      "epoch": 3.673834508859453,
-      "grad_norm": 0.001480426057241857,
-      "learning_rate": 0.0001999933423035451,
-      "loss": 46.0,
-      "step": 48051
-    },
-    {
-      "epoch": 3.673910965842843,
-      "grad_norm": 0.0036324074026197195,
-      "learning_rate": 0.00019999334202637804,
-      "loss": 46.0,
-      "step": 48052
-    },
-    {
-      "epoch": 3.673987422826232,
-      "grad_norm": 0.00061061792075634,
-      "learning_rate": 0.00019999334174920517,
-      "loss": 46.0,
-      "step": 48053
-    },
-    {
-      "epoch": 3.674063879809622,
-      "grad_norm": 0.0025726137682795525,
-      "learning_rate": 0.00019999334147202655,
-      "loss": 46.0,
-      "step": 48054
-    },
-    {
-      "epoch": 3.6741403367930117,
-      "grad_norm": 0.004986766260117292,
-      "learning_rate": 0.00019999334119484216,
-      "loss": 46.0,
-      "step": 48055
-    },
-    {
-      "epoch": 3.6742167937764014,
-      "grad_norm": 0.0038000785280019045,
-      "learning_rate": 0.000199993340917652,
-      "loss": 46.0,
-      "step": 48056
-    },
-    {
-      "epoch": 3.674293250759791,
-      "grad_norm": 0.0015942901372909546,
-      "learning_rate": 0.00019999334064045606,
-      "loss": 46.0,
-      "step": 48057
-    },
-    {
-      "epoch": 3.674369707743181,
-      "grad_norm": 0.0011252585100010037,
-      "learning_rate": 0.00019999334036325438,
-      "loss": 46.0,
-      "step": 48058
-    },
-    {
-      "epoch": 3.6744461647265707,
-      "grad_norm": 0.0023074571508914232,
-      "learning_rate": 0.00019999334008604692,
-      "loss": 46.0,
-      "step": 48059
-    },
-    {
-      "epoch": 3.6745226217099605,
-      "grad_norm": 0.002204613061621785,
-      "learning_rate": 0.0001999933398088337,
-      "loss": 46.0,
-      "step": 48060
-    },
-    {
-      "epoch": 3.6745990786933502,
-      "grad_norm": 0.003391342470422387,
-      "learning_rate": 0.0001999933395316147,
-      "loss": 46.0,
-      "step": 48061
-    },
-    {
-      "epoch": 3.67467553567674,
-      "grad_norm": 0.0022395451087504625,
-      "learning_rate": 0.00019999333925438991,
-      "loss": 46.0,
-      "step": 48062
-    },
-    {
-      "epoch": 3.6747519926601298,
-      "grad_norm": 0.0030172611586749554,
-      "learning_rate": 0.00019999333897715937,
-      "loss": 46.0,
-      "step": 48063
-    },
-    {
-      "epoch": 3.6748284496435195,
-      "grad_norm": 0.0020320371259003878,
-      "learning_rate": 0.00019999333869992304,
-      "loss": 46.0,
-      "step": 48064
-    },
-    {
-      "epoch": 3.6749049066269093,
-      "grad_norm": 0.005190548952668905,
-      "learning_rate": 0.000199993338422681,
-      "loss": 46.0,
-      "step": 48065
-    },
-    {
-      "epoch": 3.6749813636102986,
-      "grad_norm": 0.0050984956324100494,
-      "learning_rate": 0.00019999333814543313,
-      "loss": 46.0,
-      "step": 48066
-    },
-    {
-      "epoch": 3.6750578205936884,
-      "grad_norm": 0.00440974673256278,
-      "learning_rate": 0.00019999333786817954,
-      "loss": 46.0,
-      "step": 48067
-    },
-    {
-      "epoch": 3.675134277577078,
-      "grad_norm": 0.002784652402624488,
-      "learning_rate": 0.00019999333759092012,
-      "loss": 46.0,
-      "step": 48068
-    },
-    {
-      "epoch": 3.675210734560468,
-      "grad_norm": 0.0022830702364444733,
-      "learning_rate": 0.00019999333731365496,
-      "loss": 46.0,
-      "step": 48069
-    },
-    {
-      "epoch": 3.6752871915438576,
-      "grad_norm": 0.0015071353409439325,
-      "learning_rate": 0.00019999333703638405,
-      "loss": 46.0,
-      "step": 48070
-    },
-    {
-      "epoch": 3.6753636485272474,
-      "grad_norm": 0.0018765542190521955,
-      "learning_rate": 0.00019999333675910737,
-      "loss": 46.0,
-      "step": 48071
-    },
-    {
-      "epoch": 3.675440105510637,
-      "grad_norm": 0.0014902527909725904,
-      "learning_rate": 0.0001999933364818249,
-      "loss": 46.0,
-      "step": 48072
-    },
-    {
-      "epoch": 3.675516562494027,
-      "grad_norm": 0.002455366076901555,
-      "learning_rate": 0.00019999333620453666,
-      "loss": 46.0,
-      "step": 48073
-    },
-    {
-      "epoch": 3.6755930194774167,
-      "grad_norm": 0.0011709202080965042,
-      "learning_rate": 0.00019999333592724265,
-      "loss": 46.0,
-      "step": 48074
-    },
-    {
-      "epoch": 3.675669476460806,
-      "grad_norm": 0.001296831644140184,
-      "learning_rate": 0.00019999333564994288,
-      "loss": 46.0,
-      "step": 48075
-    },
-    {
-      "epoch": 3.6757459334441958,
-      "grad_norm": 0.001863612444140017,
-      "learning_rate": 0.00019999333537263736,
-      "loss": 46.0,
-      "step": 48076
-    },
-    {
-      "epoch": 3.6758223904275855,
-      "grad_norm": 0.0020350965205579996,
-      "learning_rate": 0.00019999333509532606,
-      "loss": 46.0,
-      "step": 48077
-    },
-    {
-      "epoch": 3.6758988474109753,
-      "grad_norm": 0.0032479208894073963,
-      "learning_rate": 0.00019999333481800896,
-      "loss": 46.0,
-      "step": 48078
-    },
-    {
-      "epoch": 3.675975304394365,
-      "grad_norm": 0.0015698898350819945,
-      "learning_rate": 0.00019999333454068615,
-      "loss": 46.0,
-      "step": 48079
-    },
-    {
-      "epoch": 3.676051761377755,
-      "grad_norm": 0.003312168875709176,
-      "learning_rate": 0.0001999933342633575,
-      "loss": 46.0,
-      "step": 48080
-    },
-    {
-      "epoch": 3.6761282183611446,
-      "grad_norm": 0.0019198836525902152,
-      "learning_rate": 0.00019999333398602312,
-      "loss": 46.0,
-      "step": 48081
-    },
-    {
-      "epoch": 3.6762046753445343,
-      "grad_norm": 0.005966040305793285,
-      "learning_rate": 0.00019999333370868295,
-      "loss": 46.0,
-      "step": 48082
-    },
-    {
-      "epoch": 3.676281132327924,
-      "grad_norm": 0.0035501893144100904,
-      "learning_rate": 0.00019999333343133705,
-      "loss": 46.0,
-      "step": 48083
-    },
-    {
-      "epoch": 3.676357589311314,
-      "grad_norm": 0.005789449438452721,
-      "learning_rate": 0.00019999333315398536,
-      "loss": 46.0,
-      "step": 48084
-    },
-    {
-      "epoch": 3.6764340462947036,
-      "grad_norm": 0.001949035795405507,
-      "learning_rate": 0.0001999933328766279,
-      "loss": 46.0,
-      "step": 48085
-    },
-    {
-      "epoch": 3.6765105032780934,
-      "grad_norm": 0.0008523133583366871,
-      "learning_rate": 0.00019999333259926468,
-      "loss": 46.0,
-      "step": 48086
-    },
-    {
-      "epoch": 3.676586960261483,
-      "grad_norm": 0.0023959651589393616,
-      "learning_rate": 0.00019999333232189568,
-      "loss": 46.0,
-      "step": 48087
-    },
-    {
-      "epoch": 3.6766634172448724,
-      "grad_norm": 0.004174513276666403,
-      "learning_rate": 0.0001999933320445209,
-      "loss": 46.0,
-      "step": 48088
-    },
-    {
-      "epoch": 3.676739874228262,
-      "grad_norm": 0.0015015817480161786,
-      "learning_rate": 0.00019999333176714038,
-      "loss": 46.0,
-      "step": 48089
-    },
-    {
-      "epoch": 3.676816331211652,
-      "grad_norm": 0.001505725085735321,
-      "learning_rate": 0.00019999333148975406,
-      "loss": 46.0,
-      "step": 48090
-    },
-    {
-      "epoch": 3.6768927881950417,
-      "grad_norm": 0.0032581575214862823,
-      "learning_rate": 0.000199993331212362,
-      "loss": 46.0,
-      "step": 48091
-    },
-    {
-      "epoch": 3.6769692451784315,
-      "grad_norm": 0.0026731491088867188,
-      "learning_rate": 0.00019999333093496415,
-      "loss": 46.0,
-      "step": 48092
-    },
-    {
-      "epoch": 3.6770457021618213,
-      "grad_norm": 0.001202498096972704,
-      "learning_rate": 0.00019999333065756053,
-      "loss": 46.0,
-      "step": 48093
-    },
-    {
-      "epoch": 3.677122159145211,
-      "grad_norm": 0.001614939421415329,
-      "learning_rate": 0.00019999333038015117,
-      "loss": 46.0,
-      "step": 48094
-    },
-    {
-      "epoch": 3.6771986161286008,
-      "grad_norm": 0.004757698159664869,
-      "learning_rate": 0.000199993330102736,
-      "loss": 46.0,
-      "step": 48095
-    },
-    {
-      "epoch": 3.67727507311199,
-      "grad_norm": 0.006576675921678543,
-      "learning_rate": 0.00019999332982531507,
-      "loss": 46.0,
-      "step": 48096
-    },
-    {
-      "epoch": 3.67735153009538,
-      "grad_norm": 0.0019824367482215166,
-      "learning_rate": 0.00019999332954788842,
-      "loss": 46.0,
-      "step": 48097
-    },
-    {
-      "epoch": 3.6774279870787696,
-      "grad_norm": 0.0013069227570667863,
-      "learning_rate": 0.00019999332927045593,
-      "loss": 46.0,
-      "step": 48098
-    },
-    {
-      "epoch": 3.6775044440621594,
-      "grad_norm": 0.0018959550652652979,
-      "learning_rate": 0.0001999933289930177,
-      "loss": 46.0,
-      "step": 48099
-    },
-    {
-      "epoch": 3.677580901045549,
-      "grad_norm": 0.0015238405903801322,
-      "learning_rate": 0.0001999933287155737,
-      "loss": 46.0,
-      "step": 48100
-    },
-    {
-      "epoch": 3.677657358028939,
-      "grad_norm": 0.0008242375333793461,
-      "learning_rate": 0.00019999332843812395,
-      "loss": 46.0,
-      "step": 48101
-    },
-    {
-      "epoch": 3.6777338150123287,
-      "grad_norm": 0.0022841491736471653,
-      "learning_rate": 0.00019999332816066843,
-      "loss": 46.0,
-      "step": 48102
-    },
-    {
-      "epoch": 3.6778102719957184,
-      "grad_norm": 0.0029639010317623615,
-      "learning_rate": 0.0001999933278832071,
-      "loss": 46.0,
-      "step": 48103
-    },
-    {
-      "epoch": 3.677886728979108,
-      "grad_norm": 0.0037852886598557234,
-      "learning_rate": 0.00019999332760574004,
-      "loss": 46.0,
-      "step": 48104
-    },
-    {
-      "epoch": 3.677963185962498,
-      "grad_norm": 0.002614909317344427,
-      "learning_rate": 0.0001999933273282672,
-      "loss": 46.0,
-      "step": 48105
-    },
-    {
-      "epoch": 3.6780396429458877,
-      "grad_norm": 0.0031065987423062325,
-      "learning_rate": 0.00019999332705078859,
-      "loss": 46.0,
-      "step": 48106
-    },
-    {
-      "epoch": 3.6781160999292775,
-      "grad_norm": 0.004767583683133125,
-      "learning_rate": 0.0001999933267733042,
-      "loss": 46.0,
-      "step": 48107
-    },
-    {
-      "epoch": 3.678192556912667,
-      "grad_norm": 0.004384160041809082,
-      "learning_rate": 0.00019999332649581406,
-      "loss": 46.0,
-      "step": 48108
-    },
-    {
-      "epoch": 3.678269013896057,
-      "grad_norm": 0.000985337421298027,
-      "learning_rate": 0.00019999332621831815,
-      "loss": 46.0,
-      "step": 48109
-    },
-    {
-      "epoch": 3.6783454708794463,
-      "grad_norm": 0.002141169738024473,
-      "learning_rate": 0.00019999332594081645,
-      "loss": 46.0,
-      "step": 48110
-    },
-    {
-      "epoch": 3.678421927862836,
-      "grad_norm": 0.0021303766407072544,
-      "learning_rate": 0.000199993325663309,
-      "loss": 46.0,
-      "step": 48111
-    },
-    {
-      "epoch": 3.678498384846226,
-      "grad_norm": 0.0024959477595984936,
-      "learning_rate": 0.0001999933253857958,
-      "loss": 46.0,
-      "step": 48112
-    },
-    {
-      "epoch": 3.6785748418296156,
-      "grad_norm": 0.004737451206892729,
-      "learning_rate": 0.0001999933251082768,
-      "loss": 46.0,
-      "step": 48113
-    },
-    {
-      "epoch": 3.6786512988130053,
-      "grad_norm": 0.0021286066621541977,
-      "learning_rate": 0.000199993324830752,
-      "loss": 46.0,
-      "step": 48114
-    },
-    {
-      "epoch": 3.678727755796395,
-      "grad_norm": 0.0007798591395840049,
-      "learning_rate": 0.0001999933245532215,
-      "loss": 46.0,
-      "step": 48115
-    },
-    {
-      "epoch": 3.678804212779785,
-      "grad_norm": 0.000854724959935993,
-      "learning_rate": 0.00019999332427568517,
-      "loss": 46.0,
-      "step": 48116
-    },
-    {
-      "epoch": 3.6788806697631746,
-      "grad_norm": 0.0027591839898377657,
-      "learning_rate": 0.00019999332399814313,
-      "loss": 46.0,
-      "step": 48117
-    },
-    {
-      "epoch": 3.678957126746564,
-      "grad_norm": 0.0008757482282817364,
-      "learning_rate": 0.00019999332372059526,
-      "loss": 46.0,
-      "step": 48118
-    },
-    {
-      "epoch": 3.6790335837299537,
-      "grad_norm": 0.0028068888932466507,
-      "learning_rate": 0.00019999332344304164,
-      "loss": 46.0,
-      "step": 48119
-    },
-    {
-      "epoch": 3.6791100407133435,
-      "grad_norm": 0.0014979691477492452,
-      "learning_rate": 0.00019999332316548228,
-      "loss": 46.0,
-      "step": 48120
-    },
-    {
-      "epoch": 3.6791864976967332,
-      "grad_norm": 0.0021155099384486675,
-      "learning_rate": 0.00019999332288791712,
-      "loss": 46.0,
-      "step": 48121
-    },
-    {
-      "epoch": 3.679262954680123,
-      "grad_norm": 0.0016879817703738809,
-      "learning_rate": 0.00019999332261034621,
-      "loss": 46.0,
-      "step": 48122
-    },
-    {
-      "epoch": 3.6793394116635127,
-      "grad_norm": 0.004583878442645073,
-      "learning_rate": 0.00019999332233276956,
-      "loss": 46.0,
-      "step": 48123
-    },
-    {
-      "epoch": 3.6794158686469025,
-      "grad_norm": 0.001366142649203539,
-      "learning_rate": 0.00019999332205518708,
-      "loss": 46.0,
-      "step": 48124
-    },
-    {
-      "epoch": 3.6794923256302923,
-      "grad_norm": 0.0019854449201375246,
-      "learning_rate": 0.00019999332177759885,
-      "loss": 46.0,
-      "step": 48125
-    },
-    {
-      "epoch": 3.679568782613682,
-      "grad_norm": 0.0019844858907163143,
-      "learning_rate": 0.00019999332150000485,
-      "loss": 46.0,
-      "step": 48126
-    },
-    {
-      "epoch": 3.679645239597072,
-      "grad_norm": 0.0011439098743721843,
-      "learning_rate": 0.0001999933212224051,
-      "loss": 46.0,
-      "step": 48127
-    },
-    {
-      "epoch": 3.6797216965804616,
-      "grad_norm": 0.001849220134317875,
-      "learning_rate": 0.00019999332094479958,
-      "loss": 46.0,
-      "step": 48128
-    },
-    {
-      "epoch": 3.6797981535638513,
-      "grad_norm": 0.004560360684990883,
-      "learning_rate": 0.00019999332066718829,
-      "loss": 46.0,
-      "step": 48129
-    },
-    {
-      "epoch": 3.679874610547241,
-      "grad_norm": 0.001056564156897366,
-      "learning_rate": 0.0001999933203895712,
-      "loss": 46.0,
-      "step": 48130
-    },
-    {
-      "epoch": 3.679951067530631,
-      "grad_norm": 0.001179404091089964,
-      "learning_rate": 0.00019999332011194838,
-      "loss": 46.0,
-      "step": 48131
-    },
-    {
-      "epoch": 3.68002752451402,
-      "grad_norm": 0.004666946828365326,
-      "learning_rate": 0.00019999331983431974,
-      "loss": 46.0,
-      "step": 48132
-    },
-    {
-      "epoch": 3.68010398149741,
-      "grad_norm": 0.0017488595331087708,
-      "learning_rate": 0.00019999331955668537,
-      "loss": 46.0,
-      "step": 48133
-    },
-    {
-      "epoch": 3.6801804384807997,
-      "grad_norm": 0.0025561857037246227,
-      "learning_rate": 0.00019999331927904527,
-      "loss": 46.0,
-      "step": 48134
-    },
-    {
-      "epoch": 3.6802568954641894,
-      "grad_norm": 0.002687942935153842,
-      "learning_rate": 0.00019999331900139933,
-      "loss": 46.0,
-      "step": 48135
-    },
-    {
-      "epoch": 3.680333352447579,
-      "grad_norm": 0.0018847760511562228,
-      "learning_rate": 0.00019999331872374765,
-      "loss": 46.0,
-      "step": 48136
-    },
-    {
-      "epoch": 3.680409809430969,
-      "grad_norm": 0.0008842428214848042,
-      "learning_rate": 0.0001999933184460902,
-      "loss": 46.0,
-      "step": 48137
-    },
-    {
-      "epoch": 3.6804862664143587,
-      "grad_norm": 0.002648785710334778,
-      "learning_rate": 0.00019999331816842697,
-      "loss": 46.0,
-      "step": 48138
-    },
-    {
-      "epoch": 3.6805627233977485,
-      "grad_norm": 0.003374918596819043,
-      "learning_rate": 0.000199993317890758,
-      "loss": 46.0,
-      "step": 48139
-    },
-    {
-      "epoch": 3.680639180381138,
-      "grad_norm": 0.002672750735655427,
-      "learning_rate": 0.00019999331761308322,
-      "loss": 46.0,
-      "step": 48140
-    },
-    {
-      "epoch": 3.6807156373645276,
-      "grad_norm": 0.004739441443234682,
-      "learning_rate": 0.0001999933173354027,
-      "loss": 46.0,
-      "step": 48141
-    },
-    {
-      "epoch": 3.6807920943479173,
-      "grad_norm": 0.0033642337657511234,
-      "learning_rate": 0.0001999933170577164,
-      "loss": 46.0,
-      "step": 48142
-    },
-    {
-      "epoch": 3.680868551331307,
-      "grad_norm": 0.0029913003090769053,
-      "learning_rate": 0.00019999331678002434,
-      "loss": 46.0,
-      "step": 48143
-    },
-    {
-      "epoch": 3.680945008314697,
-      "grad_norm": 0.007018153555691242,
-      "learning_rate": 0.0001999933165023265,
-      "loss": 46.0,
-      "step": 48144
-    },
-    {
-      "epoch": 3.6810214652980866,
-      "grad_norm": 0.002710035303607583,
-      "learning_rate": 0.00019999331622462288,
-      "loss": 46.0,
-      "step": 48145
-    },
-    {
-      "epoch": 3.6810979222814764,
-      "grad_norm": 0.0031039919704198837,
-      "learning_rate": 0.00019999331594691352,
-      "loss": 46.0,
-      "step": 48146
-    },
-    {
-      "epoch": 3.681174379264866,
-      "grad_norm": 0.0030833669006824493,
-      "learning_rate": 0.0001999933156691984,
-      "loss": 46.0,
-      "step": 48147
-    },
-    {
-      "epoch": 3.681250836248256,
-      "grad_norm": 0.002846986288204789,
-      "learning_rate": 0.00019999331539147748,
-      "loss": 46.0,
-      "step": 48148
-    },
-    {
-      "epoch": 3.6813272932316456,
-      "grad_norm": 0.0019875599537044764,
-      "learning_rate": 0.0001999933151137508,
-      "loss": 46.0,
-      "step": 48149
-    },
-    {
-      "epoch": 3.6814037502150354,
-      "grad_norm": 0.005792904645204544,
-      "learning_rate": 0.00019999331483601832,
-      "loss": 46.0,
-      "step": 48150
-    },
-    {
-      "epoch": 3.681480207198425,
-      "grad_norm": 0.004038248211145401,
-      "learning_rate": 0.00019999331455828012,
-      "loss": 46.0,
-      "step": 48151
-    },
-    {
-      "epoch": 3.681556664181815,
-      "grad_norm": 0.002190410392358899,
-      "learning_rate": 0.00019999331428053615,
-      "loss": 46.0,
-      "step": 48152
-    },
-    {
-      "epoch": 3.6816331211652047,
-      "grad_norm": 0.002345144748687744,
-      "learning_rate": 0.00019999331400278635,
-      "loss": 46.0,
-      "step": 48153
-    },
-    {
-      "epoch": 3.681709578148594,
-      "grad_norm": 0.0012302587274461985,
-      "learning_rate": 0.00019999331372503085,
-      "loss": 46.0,
-      "step": 48154
-    },
-    {
-      "epoch": 3.6817860351319838,
-      "grad_norm": 0.0027402357663959265,
-      "learning_rate": 0.00019999331344726956,
-      "loss": 46.0,
-      "step": 48155
-    },
-    {
-      "epoch": 3.6818624921153735,
-      "grad_norm": 0.0029681732412427664,
-      "learning_rate": 0.00019999331316950246,
-      "loss": 46.0,
-      "step": 48156
-    },
-    {
-      "epoch": 3.6819389490987633,
-      "grad_norm": 0.003746664384379983,
-      "learning_rate": 0.00019999331289172965,
-      "loss": 46.0,
-      "step": 48157
-    },
-    {
-      "epoch": 3.682015406082153,
-      "grad_norm": 0.004141835030168295,
-      "learning_rate": 0.00019999331261395104,
-      "loss": 46.0,
-      "step": 48158
-    },
-    {
-      "epoch": 3.682091863065543,
-      "grad_norm": 0.00218791957013309,
-      "learning_rate": 0.00019999331233616668,
-      "loss": 46.0,
-      "step": 48159
-    },
-    {
-      "epoch": 3.6821683200489326,
-      "grad_norm": 0.003090356709435582,
-      "learning_rate": 0.00019999331205837652,
-      "loss": 46.0,
-      "step": 48160
-    },
-    {
-      "epoch": 3.6822447770323223,
-      "grad_norm": 0.0009678081260062754,
-      "learning_rate": 0.00019999331178058064,
-      "loss": 46.0,
-      "step": 48161
-    },
-    {
-      "epoch": 3.6823212340157117,
-      "grad_norm": 0.00293830968439579,
-      "learning_rate": 0.00019999331150277896,
-      "loss": 46.0,
-      "step": 48162
-    },
-    {
-      "epoch": 3.6823976909991014,
-      "grad_norm": 0.002567223273217678,
-      "learning_rate": 0.0001999933112249715,
-      "loss": 46.0,
-      "step": 48163
-    },
-    {
-      "epoch": 3.682474147982491,
-      "grad_norm": 0.0011430465383455157,
-      "learning_rate": 0.0001999933109471583,
-      "loss": 46.0,
-      "step": 48164
-    },
-    {
-      "epoch": 3.682550604965881,
-      "grad_norm": 0.0028940029442310333,
-      "learning_rate": 0.0001999933106693393,
-      "loss": 46.0,
-      "step": 48165
-    },
-    {
-      "epoch": 3.6826270619492707,
-      "grad_norm": 0.002087947679683566,
-      "learning_rate": 0.00019999331039151455,
-      "loss": 46.0,
-      "step": 48166
-    },
-    {
-      "epoch": 3.6827035189326605,
-      "grad_norm": 0.0006349135073833168,
-      "learning_rate": 0.00019999331011368403,
-      "loss": 46.0,
-      "step": 48167
-    },
-    {
-      "epoch": 3.68277997591605,
-      "grad_norm": 0.0006834067753516138,
-      "learning_rate": 0.00019999330983584774,
-      "loss": 46.0,
-      "step": 48168
-    },
-    {
-      "epoch": 3.68285643289944,
-      "grad_norm": 0.0016259908443316817,
-      "learning_rate": 0.00019999330955800567,
-      "loss": 46.0,
-      "step": 48169
-    },
-    {
-      "epoch": 3.6829328898828297,
-      "grad_norm": 0.006330954376608133,
-      "learning_rate": 0.00019999330928015786,
-      "loss": 46.0,
-      "step": 48170
-    },
-    {
-      "epoch": 3.6830093468662195,
-      "grad_norm": 0.002377611119300127,
-      "learning_rate": 0.00019999330900230425,
-      "loss": 46.0,
-      "step": 48171
-    },
-    {
-      "epoch": 3.6830858038496093,
-      "grad_norm": 0.0023775510489940643,
-      "learning_rate": 0.00019999330872444486,
-      "loss": 46.0,
-      "step": 48172
-    },
-    {
-      "epoch": 3.683162260832999,
-      "grad_norm": 0.0020591297652572393,
-      "learning_rate": 0.00019999330844657973,
-      "loss": 46.0,
-      "step": 48173
-    },
-    {
-      "epoch": 3.683238717816389,
-      "grad_norm": 0.0015864745946601033,
-      "learning_rate": 0.00019999330816870882,
-      "loss": 46.0,
-      "step": 48174
-    },
-    {
-      "epoch": 3.6833151747997785,
-      "grad_norm": 0.0033565384801477194,
-      "learning_rate": 0.00019999330789083214,
-      "loss": 46.0,
-      "step": 48175
-    },
-    {
-      "epoch": 3.683391631783168,
-      "grad_norm": 0.004203010816127062,
-      "learning_rate": 0.0001999933076129497,
-      "loss": 46.0,
-      "step": 48176
-    },
-    {
-      "epoch": 3.6834680887665576,
-      "grad_norm": 0.0015852771466597915,
-      "learning_rate": 0.0001999933073350615,
-      "loss": 46.0,
-      "step": 48177
-    },
-    {
-      "epoch": 3.6835445457499474,
-      "grad_norm": 0.003748524934053421,
-      "learning_rate": 0.0001999933070571675,
-      "loss": 46.0,
-      "step": 48178
-    },
-    {
-      "epoch": 3.683621002733337,
-      "grad_norm": 0.003356809262186289,
-      "learning_rate": 0.00019999330677926774,
-      "loss": 46.0,
-      "step": 48179
-    },
-    {
-      "epoch": 3.683697459716727,
-      "grad_norm": 0.0016292484942823648,
-      "learning_rate": 0.00019999330650136222,
-      "loss": 46.0,
-      "step": 48180
-    },
-    {
-      "epoch": 3.6837739167001167,
-      "grad_norm": 0.002851768396794796,
-      "learning_rate": 0.00019999330622345093,
-      "loss": 46.0,
-      "step": 48181
-    },
-    {
-      "epoch": 3.6838503736835064,
-      "grad_norm": 0.004709701519459486,
-      "learning_rate": 0.0001999933059455339,
-      "loss": 46.0,
-      "step": 48182
-    },
-    {
-      "epoch": 3.683926830666896,
-      "grad_norm": 0.0013135176850482821,
-      "learning_rate": 0.00019999330566761105,
-      "loss": 46.0,
-      "step": 48183
-    },
-    {
-      "epoch": 3.6840032876502855,
-      "grad_norm": 0.0021869600750505924,
-      "learning_rate": 0.00019999330538968244,
-      "loss": 46.0,
-      "step": 48184
-    },
-    {
-      "epoch": 3.6840797446336753,
-      "grad_norm": 0.0014609413919970393,
-      "learning_rate": 0.0001999933051117481,
-      "loss": 46.0,
-      "step": 48185
-    },
-    {
-      "epoch": 3.684156201617065,
-      "grad_norm": 0.003742674132809043,
-      "learning_rate": 0.00019999330483380794,
-      "loss": 46.0,
-      "step": 48186
-    },
-    {
-      "epoch": 3.684232658600455,
-      "grad_norm": 0.005365589167922735,
-      "learning_rate": 0.00019999330455586204,
-      "loss": 46.0,
-      "step": 48187
-    },
-    {
-      "epoch": 3.6843091155838446,
-      "grad_norm": 0.0013962156372144818,
-      "learning_rate": 0.00019999330427791036,
-      "loss": 46.0,
-      "step": 48188
-    },
-    {
-      "epoch": 3.6843855725672343,
-      "grad_norm": 0.0007765196496620774,
-      "learning_rate": 0.0001999933039999529,
-      "loss": 46.0,
-      "step": 48189
-    },
-    {
-      "epoch": 3.684462029550624,
-      "grad_norm": 0.001581687363795936,
-      "learning_rate": 0.0001999933037219897,
-      "loss": 46.0,
-      "step": 48190
-    },
-    {
-      "epoch": 3.684538486534014,
-      "grad_norm": 0.003194936318323016,
-      "learning_rate": 0.00019999330344402073,
-      "loss": 46.0,
-      "step": 48191
-    },
-    {
-      "epoch": 3.6846149435174036,
-      "grad_norm": 0.0032320802565664053,
-      "learning_rate": 0.00019999330316604596,
-      "loss": 46.0,
-      "step": 48192
-    },
-    {
-      "epoch": 3.6846914005007934,
-      "grad_norm": 0.0015889855567365885,
-      "learning_rate": 0.00019999330288806547,
-      "loss": 46.0,
-      "step": 48193
-    },
-    {
-      "epoch": 3.684767857484183,
-      "grad_norm": 0.003247536951676011,
-      "learning_rate": 0.00019999330261007917,
-      "loss": 46.0,
-      "step": 48194
-    },
-    {
-      "epoch": 3.684844314467573,
-      "grad_norm": 0.004080689512193203,
-      "learning_rate": 0.0001999933023320871,
-      "loss": 46.0,
-      "step": 48195
-    },
-    {
-      "epoch": 3.6849207714509626,
-      "grad_norm": 0.004314050544053316,
-      "learning_rate": 0.0001999933020540893,
-      "loss": 46.0,
-      "step": 48196
-    },
-    {
-      "epoch": 3.684997228434352,
-      "grad_norm": 0.0018197973258793354,
-      "learning_rate": 0.00019999330177608568,
-      "loss": 46.0,
-      "step": 48197
-    },
-    {
-      "epoch": 3.6850736854177417,
-      "grad_norm": 0.002514402149245143,
-      "learning_rate": 0.00019999330149807632,
-      "loss": 46.0,
-      "step": 48198
-    },
-    {
-      "epoch": 3.6851501424011315,
-      "grad_norm": 0.001138661173172295,
-      "learning_rate": 0.0001999933012200612,
-      "loss": 46.0,
-      "step": 48199
-    },
-    {
-      "epoch": 3.6852265993845212,
-      "grad_norm": 0.0015935691772028804,
-      "learning_rate": 0.00019999330094204028,
-      "loss": 46.0,
-      "step": 48200
-    },
-    {
-      "epoch": 3.685303056367911,
-      "grad_norm": 0.0013981260126456618,
-      "learning_rate": 0.0001999933006640136,
-      "loss": 46.0,
-      "step": 48201
-    },
-    {
-      "epoch": 3.6853795133513008,
-      "grad_norm": 0.001632009749300778,
-      "learning_rate": 0.00019999330038598115,
-      "loss": 46.0,
-      "step": 48202
-    },
-    {
-      "epoch": 3.6854559703346905,
-      "grad_norm": 0.006392085924744606,
-      "learning_rate": 0.00019999330010794295,
-      "loss": 46.0,
-      "step": 48203
-    },
-    {
-      "epoch": 3.6855324273180803,
-      "grad_norm": 0.0006125652580522001,
-      "learning_rate": 0.00019999329982989898,
-      "loss": 46.0,
-      "step": 48204
-    },
-    {
-      "epoch": 3.68560888430147,
-      "grad_norm": 0.0010959943756461143,
-      "learning_rate": 0.00019999329955184923,
-      "loss": 46.0,
-      "step": 48205
-    },
-    {
-      "epoch": 3.6856853412848594,
-      "grad_norm": 0.005319190677255392,
-      "learning_rate": 0.00019999329927379372,
-      "loss": 46.0,
-      "step": 48206
-    },
-    {
-      "epoch": 3.685761798268249,
-      "grad_norm": 0.002845548326149583,
-      "learning_rate": 0.00019999329899573245,
-      "loss": 46.0,
-      "step": 48207
-    },
-    {
-      "epoch": 3.685838255251639,
-      "grad_norm": 0.0011395186884328723,
-      "learning_rate": 0.00019999329871766536,
-      "loss": 46.0,
-      "step": 48208
-    },
-    {
-      "epoch": 3.6859147122350286,
-      "grad_norm": 0.0003700665256474167,
-      "learning_rate": 0.00019999329843959255,
-      "loss": 46.0,
-      "step": 48209
-    },
-    {
-      "epoch": 3.6859911692184184,
-      "grad_norm": 0.0021044376771897078,
-      "learning_rate": 0.00019999329816151396,
-      "loss": 46.0,
-      "step": 48210
-    },
-    {
-      "epoch": 3.686067626201808,
-      "grad_norm": 0.0028955296147614717,
-      "learning_rate": 0.0001999932978834296,
-      "loss": 46.0,
-      "step": 48211
-    },
-    {
-      "epoch": 3.686144083185198,
-      "grad_norm": 0.001400425098836422,
-      "learning_rate": 0.00019999329760533947,
-      "loss": 46.0,
-      "step": 48212
-    },
-    {
-      "epoch": 3.6862205401685877,
-      "grad_norm": 0.001666681026108563,
-      "learning_rate": 0.00019999329732724356,
-      "loss": 46.0,
-      "step": 48213
-    },
-    {
-      "epoch": 3.6862969971519775,
-      "grad_norm": 0.0018024382879957557,
-      "learning_rate": 0.00019999329704914188,
-      "loss": 46.0,
-      "step": 48214
-    },
-    {
-      "epoch": 3.686373454135367,
-      "grad_norm": 0.0017111904453486204,
-      "learning_rate": 0.00019999329677103446,
-      "loss": 46.0,
-      "step": 48215
-    },
-    {
-      "epoch": 3.686449911118757,
-      "grad_norm": 0.0018292288295924664,
-      "learning_rate": 0.00019999329649292123,
-      "loss": 46.0,
-      "step": 48216
-    },
-    {
-      "epoch": 3.6865263681021467,
-      "grad_norm": 0.0035259316209703684,
-      "learning_rate": 0.00019999329621480226,
-      "loss": 46.0,
-      "step": 48217
-    },
-    {
-      "epoch": 3.6866028250855365,
-      "grad_norm": 0.00329809682443738,
-      "learning_rate": 0.00019999329593667752,
-      "loss": 46.0,
-      "step": 48218
-    },
-    {
-      "epoch": 3.686679282068926,
-      "grad_norm": 0.001148804440163076,
-      "learning_rate": 0.000199993295658547,
-      "loss": 46.0,
-      "step": 48219
-    },
-    {
-      "epoch": 3.6867557390523156,
-      "grad_norm": 0.004229803569614887,
-      "learning_rate": 0.0001999932953804107,
-      "loss": 46.0,
-      "step": 48220
-    },
-    {
-      "epoch": 3.6868321960357053,
-      "grad_norm": 0.005939357914030552,
-      "learning_rate": 0.00019999329510226867,
-      "loss": 46.0,
-      "step": 48221
-    },
-    {
-      "epoch": 3.686908653019095,
-      "grad_norm": 0.0011773926671594381,
-      "learning_rate": 0.00019999329482412085,
-      "loss": 46.0,
-      "step": 48222
-    },
-    {
-      "epoch": 3.686985110002485,
-      "grad_norm": 0.0019149387953802943,
-      "learning_rate": 0.00019999329454596727,
-      "loss": 46.0,
-      "step": 48223
-    },
-    {
-      "epoch": 3.6870615669858746,
-      "grad_norm": 0.0015631114365532994,
-      "learning_rate": 0.0001999932942678079,
-      "loss": 46.0,
-      "step": 48224
-    },
-    {
-      "epoch": 3.6871380239692644,
-      "grad_norm": 0.0025576287880539894,
-      "learning_rate": 0.00019999329398964278,
-      "loss": 46.0,
-      "step": 48225
-    },
-    {
-      "epoch": 3.687214480952654,
-      "grad_norm": 0.0012888452038168907,
-      "learning_rate": 0.00019999329371147187,
-      "loss": 46.0,
-      "step": 48226
-    },
-    {
-      "epoch": 3.687290937936044,
-      "grad_norm": 0.001446765149012208,
-      "learning_rate": 0.0001999932934332952,
-      "loss": 46.0,
-      "step": 48227
-    },
-    {
-      "epoch": 3.687367394919433,
-      "grad_norm": 0.007158469408750534,
-      "learning_rate": 0.00019999329315511277,
-      "loss": 46.0,
-      "step": 48228
-    },
-    {
-      "epoch": 3.687443851902823,
-      "grad_norm": 0.006073956377804279,
-      "learning_rate": 0.00019999329287692454,
-      "loss": 46.0,
-      "step": 48229
-    },
-    {
-      "epoch": 3.6875203088862127,
-      "grad_norm": 0.004490464460104704,
-      "learning_rate": 0.0001999932925987306,
-      "loss": 46.0,
-      "step": 48230
-    },
-    {
-      "epoch": 3.6875967658696025,
-      "grad_norm": 0.0023652713280171156,
-      "learning_rate": 0.00019999329232053083,
-      "loss": 46.0,
-      "step": 48231
-    },
-    {
-      "epoch": 3.6876732228529923,
-      "grad_norm": 0.008131634443998337,
-      "learning_rate": 0.0001999932920423253,
-      "loss": 46.0,
-      "step": 48232
-    },
-    {
-      "epoch": 3.687749679836382,
-      "grad_norm": 0.0023487142752856016,
-      "learning_rate": 0.00019999329176411404,
-      "loss": 46.0,
-      "step": 48233
-    },
-    {
-      "epoch": 3.687826136819772,
-      "grad_norm": 0.0012566643999889493,
-      "learning_rate": 0.00019999329148589698,
-      "loss": 46.0,
-      "step": 48234
-    },
-    {
-      "epoch": 3.6879025938031615,
-      "grad_norm": 0.0030417630914598703,
-      "learning_rate": 0.00019999329120767417,
-      "loss": 46.0,
-      "step": 48235
-    },
-    {
-      "epoch": 3.6879790507865513,
-      "grad_norm": 0.0022397369612008333,
-      "learning_rate": 0.00019999329092944555,
-      "loss": 46.0,
-      "step": 48236
-    },
-    {
-      "epoch": 3.688055507769941,
-      "grad_norm": 0.00214776280336082,
-      "learning_rate": 0.00019999329065121122,
-      "loss": 46.0,
-      "step": 48237
-    },
-    {
-      "epoch": 3.688131964753331,
-      "grad_norm": 0.0013408100930973887,
-      "learning_rate": 0.0001999932903729711,
-      "loss": 46.0,
-      "step": 48238
-    },
-    {
-      "epoch": 3.6882084217367206,
-      "grad_norm": 0.0011732698185369372,
-      "learning_rate": 0.00019999329009472519,
-      "loss": 46.0,
-      "step": 48239
-    },
-    {
-      "epoch": 3.6882848787201103,
-      "grad_norm": 0.0015864792512729764,
-      "learning_rate": 0.00019999328981647353,
-      "loss": 46.0,
-      "step": 48240
-    },
-    {
-      "epoch": 3.6883613357034997,
-      "grad_norm": 0.0005314729060046375,
-      "learning_rate": 0.00019999328953821608,
-      "loss": 46.0,
-      "step": 48241
-    },
-    {
-      "epoch": 3.6884377926868894,
-      "grad_norm": 0.0015775756910443306,
-      "learning_rate": 0.00019999328925995289,
-      "loss": 46.0,
-      "step": 48242
-    },
-    {
-      "epoch": 3.688514249670279,
-      "grad_norm": 0.0007871423149481416,
-      "learning_rate": 0.00019999328898168391,
-      "loss": 46.0,
-      "step": 48243
-    },
-    {
-      "epoch": 3.688590706653669,
-      "grad_norm": 0.0027579700108617544,
-      "learning_rate": 0.0001999932887034092,
-      "loss": 46.0,
-      "step": 48244
-    },
-    {
-      "epoch": 3.6886671636370587,
-      "grad_norm": 0.0019248995231464505,
-      "learning_rate": 0.00019999328842512868,
-      "loss": 46.0,
-      "step": 48245
-    },
-    {
-      "epoch": 3.6887436206204485,
-      "grad_norm": 0.005275071132928133,
-      "learning_rate": 0.00019999328814684239,
-      "loss": 46.0,
-      "step": 48246
-    },
-    {
-      "epoch": 3.6888200776038382,
-      "grad_norm": 0.0011181710287928581,
-      "learning_rate": 0.00019999328786855035,
-      "loss": 46.0,
-      "step": 48247
-    },
-    {
-      "epoch": 3.688896534587228,
-      "grad_norm": 0.0019004099303856492,
-      "learning_rate": 0.0001999932875902525,
-      "loss": 46.0,
-      "step": 48248
-    },
-    {
-      "epoch": 3.6889729915706173,
-      "grad_norm": 0.0034082112833857536,
-      "learning_rate": 0.00019999328731194893,
-      "loss": 46.0,
-      "step": 48249
-    },
-    {
-      "epoch": 3.689049448554007,
-      "grad_norm": 0.0014565015444532037,
-      "learning_rate": 0.00019999328703363957,
-      "loss": 46.0,
-      "step": 48250
-    },
-    {
-      "epoch": 3.689125905537397,
-      "grad_norm": 0.0010718003613874316,
-      "learning_rate": 0.00019999328675532444,
-      "loss": 46.0,
-      "step": 48251
-    },
-    {
-      "epoch": 3.6892023625207866,
-      "grad_norm": 0.0006640134961344302,
-      "learning_rate": 0.00019999328647700356,
-      "loss": 46.0,
-      "step": 48252
-    },
-    {
-      "epoch": 3.6892788195041764,
-      "grad_norm": 0.0043806410394608974,
-      "learning_rate": 0.0001999932861986769,
-      "loss": 46.0,
-      "step": 48253
-    },
-    {
-      "epoch": 3.689355276487566,
-      "grad_norm": 0.0016125160036608577,
-      "learning_rate": 0.00019999328592034446,
-      "loss": 46.0,
-      "step": 48254
-    },
-    {
-      "epoch": 3.689431733470956,
-      "grad_norm": 0.002997814677655697,
-      "learning_rate": 0.00019999328564200626,
-      "loss": 46.0,
-      "step": 48255
-    },
-    {
-      "epoch": 3.6895081904543456,
-      "grad_norm": 0.004889064468443394,
-      "learning_rate": 0.0001999932853636623,
-      "loss": 46.0,
-      "step": 48256
-    },
-    {
-      "epoch": 3.6895846474377354,
-      "grad_norm": 0.001033770153298974,
-      "learning_rate": 0.00019999328508531254,
-      "loss": 46.0,
-      "step": 48257
-    },
-    {
-      "epoch": 3.689661104421125,
-      "grad_norm": 0.002096008276566863,
-      "learning_rate": 0.00019999328480695705,
-      "loss": 46.0,
-      "step": 48258
-    },
-    {
-      "epoch": 3.689737561404515,
-      "grad_norm": 0.0011949068866670132,
-      "learning_rate": 0.00019999328452859576,
-      "loss": 46.0,
-      "step": 48259
-    },
-    {
-      "epoch": 3.6898140183879047,
-      "grad_norm": 0.003170607378706336,
-      "learning_rate": 0.0001999932842502287,
-      "loss": 46.0,
-      "step": 48260
-    },
-    {
-      "epoch": 3.6898904753712944,
-      "grad_norm": 0.004153276327997446,
-      "learning_rate": 0.0001999932839718559,
-      "loss": 46.0,
-      "step": 48261
-    },
-    {
-      "epoch": 3.689966932354684,
-      "grad_norm": 0.0038226621691137552,
-      "learning_rate": 0.0001999932836934773,
-      "loss": 46.0,
-      "step": 48262
-    },
-    {
-      "epoch": 3.6900433893380735,
-      "grad_norm": 0.0012407738249748945,
-      "learning_rate": 0.00019999328341509297,
-      "loss": 46.0,
-      "step": 48263
-    },
-    {
-      "epoch": 3.6901198463214633,
-      "grad_norm": 0.0031481976620852947,
-      "learning_rate": 0.0001999932831367028,
-      "loss": 46.0,
-      "step": 48264
-    },
-    {
-      "epoch": 3.690196303304853,
-      "grad_norm": 0.00109721883200109,
-      "learning_rate": 0.00019999328285830693,
-      "loss": 46.0,
-      "step": 48265
-    },
-    {
-      "epoch": 3.690272760288243,
-      "grad_norm": 0.0017670998349785805,
-      "learning_rate": 0.00019999328257990526,
-      "loss": 46.0,
-      "step": 48266
-    },
-    {
-      "epoch": 3.6903492172716326,
-      "grad_norm": 0.0018323304830119014,
-      "learning_rate": 0.00019999328230149783,
-      "loss": 46.0,
-      "step": 48267
-    },
-    {
-      "epoch": 3.6904256742550223,
-      "grad_norm": 0.005305479280650616,
-      "learning_rate": 0.00019999328202308466,
-      "loss": 46.0,
-      "step": 48268
-    },
-    {
-      "epoch": 3.690502131238412,
-      "grad_norm": 0.001006661681458354,
-      "learning_rate": 0.00019999328174466566,
-      "loss": 46.0,
-      "step": 48269
-    },
-    {
-      "epoch": 3.690578588221802,
-      "grad_norm": 0.0013932877918705344,
-      "learning_rate": 0.00019999328146624095,
-      "loss": 46.0,
-      "step": 48270
-    },
-    {
-      "epoch": 3.690655045205191,
-      "grad_norm": 0.0022297517862170935,
-      "learning_rate": 0.00019999328118781043,
-      "loss": 46.0,
-      "step": 48271
-    },
-    {
-      "epoch": 3.690731502188581,
-      "grad_norm": 0.002408611122518778,
-      "learning_rate": 0.00019999328090937417,
-      "loss": 46.0,
-      "step": 48272
-    },
-    {
-      "epoch": 3.6908079591719707,
-      "grad_norm": 0.0015594715951010585,
-      "learning_rate": 0.0001999932806309321,
-      "loss": 46.0,
-      "step": 48273
-    },
-    {
-      "epoch": 3.6908844161553604,
-      "grad_norm": 0.0030948061030358076,
-      "learning_rate": 0.0001999932803524843,
-      "loss": 46.0,
-      "step": 48274
-    },
-    {
-      "epoch": 3.69096087313875,
-      "grad_norm": 0.0014688984956592321,
-      "learning_rate": 0.0001999932800740307,
-      "loss": 46.0,
-      "step": 48275
-    },
-    {
-      "epoch": 3.69103733012214,
-      "grad_norm": 0.0009575940202921629,
-      "learning_rate": 0.00019999327979557138,
-      "loss": 46.0,
-      "step": 48276
-    },
-    {
-      "epoch": 3.6911137871055297,
-      "grad_norm": 0.001533214352093637,
-      "learning_rate": 0.00019999327951710622,
-      "loss": 46.0,
-      "step": 48277
-    },
-    {
-      "epoch": 3.6911902440889195,
-      "grad_norm": 0.0032603831496089697,
-      "learning_rate": 0.00019999327923863534,
-      "loss": 46.0,
-      "step": 48278
-    },
-    {
-      "epoch": 3.6912667010723093,
-      "grad_norm": 0.00631999084725976,
-      "learning_rate": 0.0001999932789601587,
-      "loss": 46.0,
-      "step": 48279
-    },
-    {
-      "epoch": 3.691343158055699,
-      "grad_norm": 0.0015919201541692019,
-      "learning_rate": 0.00019999327868167627,
-      "loss": 46.0,
-      "step": 48280
-    },
-    {
-      "epoch": 3.6914196150390888,
-      "grad_norm": 0.0020129787735641003,
-      "learning_rate": 0.00019999327840318807,
-      "loss": 46.0,
-      "step": 48281
-    },
-    {
-      "epoch": 3.6914960720224785,
-      "grad_norm": 0.0024249712005257607,
-      "learning_rate": 0.00019999327812469407,
-      "loss": 46.0,
-      "step": 48282
-    },
-    {
-      "epoch": 3.6915725290058683,
-      "grad_norm": 0.0055593387223780155,
-      "learning_rate": 0.00019999327784619438,
-      "loss": 46.0,
-      "step": 48283
-    },
-    {
-      "epoch": 3.691648985989258,
-      "grad_norm": 0.0021007873583585024,
-      "learning_rate": 0.00019999327756768884,
-      "loss": 46.0,
-      "step": 48284
-    },
-    {
-      "epoch": 3.6917254429726474,
-      "grad_norm": 0.0005054878420196474,
-      "learning_rate": 0.00019999327728917758,
-      "loss": 46.0,
-      "step": 48285
-    },
-    {
-      "epoch": 3.691801899956037,
-      "grad_norm": 0.005452163517475128,
-      "learning_rate": 0.00019999327701066054,
-      "loss": 46.0,
-      "step": 48286
-    },
-    {
-      "epoch": 3.691878356939427,
-      "grad_norm": 0.004880304913967848,
-      "learning_rate": 0.00019999327673213773,
-      "loss": 46.0,
-      "step": 48287
-    },
-    {
-      "epoch": 3.6919548139228167,
-      "grad_norm": 0.0006186253158375621,
-      "learning_rate": 0.00019999327645360914,
-      "loss": 46.0,
-      "step": 48288
-    },
-    {
-      "epoch": 3.6920312709062064,
-      "grad_norm": 0.0023824337404221296,
-      "learning_rate": 0.0001999932761750748,
-      "loss": 46.0,
-      "step": 48289
-    },
-    {
-      "epoch": 3.692107727889596,
-      "grad_norm": 0.0009436077089048922,
-      "learning_rate": 0.00019999327589653468,
-      "loss": 46.0,
-      "step": 48290
-    },
-    {
-      "epoch": 3.692184184872986,
-      "grad_norm": 0.0010116018820554018,
-      "learning_rate": 0.0001999932756179888,
-      "loss": 46.0,
-      "step": 48291
-    },
-    {
-      "epoch": 3.6922606418563757,
-      "grad_norm": 0.0016154585173353553,
-      "learning_rate": 0.00019999327533943713,
-      "loss": 46.0,
-      "step": 48292
-    },
-    {
-      "epoch": 3.692337098839765,
-      "grad_norm": 0.0021027622278779745,
-      "learning_rate": 0.0001999932750608797,
-      "loss": 46.0,
-      "step": 48293
-    },
-    {
-      "epoch": 3.692413555823155,
-      "grad_norm": 0.008123385719954967,
-      "learning_rate": 0.0001999932747823165,
-      "loss": 46.0,
-      "step": 48294
-    },
-    {
-      "epoch": 3.6924900128065445,
-      "grad_norm": 0.001654416206292808,
-      "learning_rate": 0.00019999327450374757,
-      "loss": 46.0,
-      "step": 48295
-    },
-    {
-      "epoch": 3.6925664697899343,
-      "grad_norm": 0.004944758024066687,
-      "learning_rate": 0.0001999932742251728,
-      "loss": 46.0,
-      "step": 48296
-    },
-    {
-      "epoch": 3.692642926773324,
-      "grad_norm": 0.004164565354585648,
-      "learning_rate": 0.0001999932739465923,
-      "loss": 46.0,
-      "step": 48297
-    },
-    {
-      "epoch": 3.692719383756714,
-      "grad_norm": 0.0007931336294859648,
-      "learning_rate": 0.00019999327366800604,
-      "loss": 46.0,
-      "step": 48298
-    },
-    {
-      "epoch": 3.6927958407401036,
-      "grad_norm": 0.001713545061647892,
-      "learning_rate": 0.000199993273389414,
-      "loss": 46.0,
-      "step": 48299
-    },
-    {
-      "epoch": 3.6928722977234933,
-      "grad_norm": 0.0013544049579650164,
-      "learning_rate": 0.0001999932731108162,
-      "loss": 46.0,
-      "step": 48300
-    },
-    {
-      "epoch": 3.692948754706883,
-      "grad_norm": 0.002430782187730074,
-      "learning_rate": 0.0001999932728322126,
-      "loss": 46.0,
-      "step": 48301
-    },
-    {
-      "epoch": 3.693025211690273,
-      "grad_norm": 0.002690747380256653,
-      "learning_rate": 0.00019999327255360326,
-      "loss": 46.0,
-      "step": 48302
-    },
-    {
-      "epoch": 3.6931016686736626,
-      "grad_norm": 0.001381297712214291,
-      "learning_rate": 0.00019999327227498815,
-      "loss": 46.0,
-      "step": 48303
-    },
-    {
-      "epoch": 3.6931781256570524,
-      "grad_norm": 0.004420554731041193,
-      "learning_rate": 0.00019999327199636725,
-      "loss": 46.0,
-      "step": 48304
-    },
-    {
-      "epoch": 3.693254582640442,
-      "grad_norm": 0.006982902996242046,
-      "learning_rate": 0.0001999932717177406,
-      "loss": 46.0,
-      "step": 48305
-    },
-    {
-      "epoch": 3.693331039623832,
-      "grad_norm": 0.0018282266100868583,
-      "learning_rate": 0.00019999327143910818,
-      "loss": 46.0,
-      "step": 48306
-    },
-    {
-      "epoch": 3.6934074966072212,
-      "grad_norm": 0.0019510556012392044,
-      "learning_rate": 0.00019999327116047,
-      "loss": 46.0,
-      "step": 48307
-    },
-    {
-      "epoch": 3.693483953590611,
-      "grad_norm": 0.003085219766944647,
-      "learning_rate": 0.000199993270881826,
-      "loss": 46.0,
-      "step": 48308
-    },
-    {
-      "epoch": 3.6935604105740008,
-      "grad_norm": 0.005437133833765984,
-      "learning_rate": 0.0001999932706031763,
-      "loss": 46.0,
-      "step": 48309
-    },
-    {
-      "epoch": 3.6936368675573905,
-      "grad_norm": 0.002673087175935507,
-      "learning_rate": 0.00019999327032452078,
-      "loss": 46.0,
-      "step": 48310
-    },
-    {
-      "epoch": 3.6937133245407803,
-      "grad_norm": 0.0025422677863389254,
-      "learning_rate": 0.00019999327004585952,
-      "loss": 46.0,
-      "step": 48311
-    },
-    {
-      "epoch": 3.69378978152417,
-      "grad_norm": 0.005424900446087122,
-      "learning_rate": 0.00019999326976719248,
-      "loss": 46.0,
-      "step": 48312
-    },
-    {
-      "epoch": 3.69386623850756,
-      "grad_norm": 0.002368065994232893,
-      "learning_rate": 0.0001999932694885197,
-      "loss": 46.0,
-      "step": 48313
-    },
-    {
-      "epoch": 3.6939426954909496,
-      "grad_norm": 0.0012580581242218614,
-      "learning_rate": 0.00019999326920984109,
-      "loss": 46.0,
-      "step": 48314
-    },
-    {
-      "epoch": 3.694019152474339,
-      "grad_norm": 0.0007263403967954218,
-      "learning_rate": 0.00019999326893115676,
-      "loss": 46.0,
-      "step": 48315
-    },
-    {
-      "epoch": 3.6940956094577286,
-      "grad_norm": 0.0011105338344350457,
-      "learning_rate": 0.00019999326865246663,
-      "loss": 46.0,
-      "step": 48316
-    },
-    {
-      "epoch": 3.6941720664411184,
-      "grad_norm": 0.001534783747047186,
-      "learning_rate": 0.00019999326837377078,
-      "loss": 46.0,
-      "step": 48317
-    },
-    {
-      "epoch": 3.694248523424508,
-      "grad_norm": 0.0005228347145020962,
-      "learning_rate": 0.00019999326809506907,
-      "loss": 46.0,
-      "step": 48318
-    },
-    {
-      "epoch": 3.694324980407898,
-      "grad_norm": 0.004914476536214352,
-      "learning_rate": 0.00019999326781636168,
-      "loss": 46.0,
-      "step": 48319
-    },
-    {
-      "epoch": 3.6944014373912877,
-      "grad_norm": 0.0022618144284933805,
-      "learning_rate": 0.00019999326753764848,
-      "loss": 46.0,
-      "step": 48320
-    },
-    {
-      "epoch": 3.6944778943746774,
-      "grad_norm": 0.0018209160771220922,
-      "learning_rate": 0.00019999326725892954,
-      "loss": 46.0,
-      "step": 48321
-    },
-    {
-      "epoch": 3.694554351358067,
-      "grad_norm": 0.0017061736434698105,
-      "learning_rate": 0.0001999932669802048,
-      "loss": 46.0,
-      "step": 48322
-    },
-    {
-      "epoch": 3.694630808341457,
-      "grad_norm": 0.0013384344056248665,
-      "learning_rate": 0.00019999326670147428,
-      "loss": 46.0,
-      "step": 48323
-    },
-    {
-      "epoch": 3.6947072653248467,
-      "grad_norm": 0.002426867140457034,
-      "learning_rate": 0.00019999326642273805,
-      "loss": 46.0,
-      "step": 48324
-    },
-    {
-      "epoch": 3.6947837223082365,
-      "grad_norm": 0.005450364667922258,
-      "learning_rate": 0.000199993266143996,
-      "loss": 46.0,
-      "step": 48325
-    },
-    {
-      "epoch": 3.6948601792916262,
-      "grad_norm": 0.0029282495379447937,
-      "learning_rate": 0.00019999326586524817,
-      "loss": 46.0,
-      "step": 48326
-    },
-    {
-      "epoch": 3.694936636275016,
-      "grad_norm": 0.001443423330783844,
-      "learning_rate": 0.00019999326558649462,
-      "loss": 46.0,
-      "step": 48327
-    },
-    {
-      "epoch": 3.6950130932584053,
-      "grad_norm": 0.0007966472185216844,
-      "learning_rate": 0.0001999932653077353,
-      "loss": 46.0,
-      "step": 48328
-    },
-    {
-      "epoch": 3.695089550241795,
-      "grad_norm": 0.005007739178836346,
-      "learning_rate": 0.00019999326502897016,
-      "loss": 46.0,
-      "step": 48329
-    },
-    {
-      "epoch": 3.695166007225185,
-      "grad_norm": 0.0008839049842208624,
-      "learning_rate": 0.00019999326475019928,
-      "loss": 46.0,
-      "step": 48330
-    },
-    {
-      "epoch": 3.6952424642085746,
-      "grad_norm": 0.0018118356820195913,
-      "learning_rate": 0.00019999326447142263,
-      "loss": 46.0,
-      "step": 48331
-    },
-    {
-      "epoch": 3.6953189211919644,
-      "grad_norm": 0.0014934723731130362,
-      "learning_rate": 0.0001999932641926402,
-      "loss": 46.0,
-      "step": 48332
-    },
-    {
-      "epoch": 3.695395378175354,
-      "grad_norm": 0.002866104943677783,
-      "learning_rate": 0.00019999326391385202,
-      "loss": 46.0,
-      "step": 48333
-    },
-    {
-      "epoch": 3.695471835158744,
-      "grad_norm": 0.0013519731583073735,
-      "learning_rate": 0.00019999326363505805,
-      "loss": 46.0,
-      "step": 48334
-    },
-    {
-      "epoch": 3.6955482921421337,
-      "grad_norm": 0.003833555383607745,
-      "learning_rate": 0.00019999326335625833,
-      "loss": 46.0,
-      "step": 48335
-    },
-    {
-      "epoch": 3.6956247491255234,
-      "grad_norm": 0.0011996623361483216,
-      "learning_rate": 0.00019999326307745284,
-      "loss": 46.0,
-      "step": 48336
-    },
-    {
-      "epoch": 3.6957012061089127,
-      "grad_norm": 0.005955439526587725,
-      "learning_rate": 0.00019999326279864158,
-      "loss": 46.0,
-      "step": 48337
-    },
-    {
-      "epoch": 3.6957776630923025,
-      "grad_norm": 0.002025017747655511,
-      "learning_rate": 0.00019999326251982454,
-      "loss": 46.0,
-      "step": 48338
-    },
-    {
-      "epoch": 3.6958541200756923,
-      "grad_norm": 0.0009732857579365373,
-      "learning_rate": 0.00019999326224100174,
-      "loss": 46.0,
-      "step": 48339
-    },
-    {
-      "epoch": 3.695930577059082,
-      "grad_norm": 0.003599429503083229,
-      "learning_rate": 0.00019999326196217315,
-      "loss": 46.0,
-      "step": 48340
-    },
-    {
-      "epoch": 3.6960070340424718,
-      "grad_norm": 0.0014631038065999746,
-      "learning_rate": 0.00019999326168333882,
-      "loss": 46.0,
-      "step": 48341
-    },
-    {
-      "epoch": 3.6960834910258615,
-      "grad_norm": 0.0026974461507052183,
-      "learning_rate": 0.00019999326140449872,
-      "loss": 46.0,
-      "step": 48342
-    },
-    {
-      "epoch": 3.6961599480092513,
-      "grad_norm": 0.0022770550567656755,
-      "learning_rate": 0.00019999326112565285,
-      "loss": 46.0,
-      "step": 48343
-    },
-    {
-      "epoch": 3.696236404992641,
-      "grad_norm": 0.0022206429857760668,
-      "learning_rate": 0.0001999932608468012,
-      "loss": 46.0,
-      "step": 48344
-    },
-    {
-      "epoch": 3.696312861976031,
-      "grad_norm": 0.007957326248288155,
-      "learning_rate": 0.00019999326056794378,
-      "loss": 46.0,
-      "step": 48345
-    },
-    {
-      "epoch": 3.6963893189594206,
-      "grad_norm": 0.0009934828849509358,
-      "learning_rate": 0.00019999326028908055,
-      "loss": 46.0,
-      "step": 48346
-    },
-    {
-      "epoch": 3.6964657759428103,
-      "grad_norm": 0.0006912837270647287,
-      "learning_rate": 0.00019999326001021164,
-      "loss": 46.0,
-      "step": 48347
-    },
-    {
-      "epoch": 3.6965422329262,
-      "grad_norm": 0.0033846464939415455,
-      "learning_rate": 0.0001999932597313369,
-      "loss": 46.0,
-      "step": 48348
-    },
-    {
-      "epoch": 3.69661868990959,
-      "grad_norm": 0.0021086963824927807,
-      "learning_rate": 0.0001999932594524564,
-      "loss": 46.0,
-      "step": 48349
-    },
-    {
-      "epoch": 3.696695146892979,
-      "grad_norm": 0.0011045270366594195,
-      "learning_rate": 0.00019999325917357012,
-      "loss": 46.0,
-      "step": 48350
-    },
-    {
-      "epoch": 3.696771603876369,
-      "grad_norm": 0.0012832568027079105,
-      "learning_rate": 0.0001999932588946781,
-      "loss": 46.0,
-      "step": 48351
-    },
-    {
-      "epoch": 3.6968480608597587,
-      "grad_norm": 0.0069931442849338055,
-      "learning_rate": 0.0001999932586157803,
-      "loss": 46.0,
-      "step": 48352
-    },
-    {
-      "epoch": 3.6969245178431485,
-      "grad_norm": 0.0019340061116963625,
-      "learning_rate": 0.00019999325833687675,
-      "loss": 46.0,
-      "step": 48353
-    },
-    {
-      "epoch": 3.6970009748265382,
-      "grad_norm": 0.00277696643024683,
-      "learning_rate": 0.0001999932580579674,
-      "loss": 46.0,
-      "step": 48354
-    },
-    {
-      "epoch": 3.697077431809928,
-      "grad_norm": 0.0022195095662027597,
-      "learning_rate": 0.0001999932577790523,
-      "loss": 46.0,
-      "step": 48355
-    },
-    {
-      "epoch": 3.6971538887933177,
-      "grad_norm": 0.001210420741699636,
-      "learning_rate": 0.00019999325750013141,
-      "loss": 46.0,
-      "step": 48356
-    },
-    {
-      "epoch": 3.6972303457767075,
-      "grad_norm": 0.0010217016097158194,
-      "learning_rate": 0.00019999325722120477,
-      "loss": 46.0,
-      "step": 48357
-    },
-    {
-      "epoch": 3.6973068027600973,
-      "grad_norm": 0.0015789292519912124,
-      "learning_rate": 0.00019999325694227234,
-      "loss": 46.0,
-      "step": 48358
-    },
-    {
-      "epoch": 3.6973832597434866,
-      "grad_norm": 0.002075741533190012,
-      "learning_rate": 0.00019999325666333415,
-      "loss": 46.0,
-      "step": 48359
-    },
-    {
-      "epoch": 3.6974597167268763,
-      "grad_norm": 0.0024765036068856716,
-      "learning_rate": 0.0001999932563843902,
-      "loss": 46.0,
-      "step": 48360
-    },
-    {
-      "epoch": 3.697536173710266,
-      "grad_norm": 0.0044881487265229225,
-      "learning_rate": 0.0001999932561054405,
-      "loss": 46.0,
-      "step": 48361
-    },
-    {
-      "epoch": 3.697612630693656,
-      "grad_norm": 0.0016344249015673995,
-      "learning_rate": 0.00019999325582648498,
-      "loss": 46.0,
-      "step": 48362
-    },
-    {
-      "epoch": 3.6976890876770456,
-      "grad_norm": 0.001778409699909389,
-      "learning_rate": 0.00019999325554752372,
-      "loss": 46.0,
-      "step": 48363
-    },
-    {
-      "epoch": 3.6977655446604354,
-      "grad_norm": 0.0009742042166180909,
-      "learning_rate": 0.0001999932552685567,
-      "loss": 46.0,
-      "step": 48364
-    },
-    {
-      "epoch": 3.697842001643825,
-      "grad_norm": 0.003260405734181404,
-      "learning_rate": 0.0001999932549895839,
-      "loss": 46.0,
-      "step": 48365
-    },
-    {
-      "epoch": 3.697918458627215,
-      "grad_norm": 0.0016500040655955672,
-      "learning_rate": 0.00019999325471060532,
-      "loss": 46.0,
-      "step": 48366
-    },
-    {
-      "epoch": 3.6979949156106047,
-      "grad_norm": 0.0030370925087481737,
-      "learning_rate": 0.000199993254431621,
-      "loss": 46.0,
-      "step": 48367
-    },
-    {
-      "epoch": 3.6980713725939944,
-      "grad_norm": 0.003414441365748644,
-      "learning_rate": 0.00019999325415263091,
-      "loss": 46.0,
-      "step": 48368
-    },
-    {
-      "epoch": 3.698147829577384,
-      "grad_norm": 0.0017931165639311075,
-      "learning_rate": 0.000199993253873635,
-      "loss": 46.0,
-      "step": 48369
-    },
-    {
-      "epoch": 3.698224286560774,
-      "grad_norm": 0.0012098739389330149,
-      "learning_rate": 0.0001999932535946334,
-      "loss": 46.0,
-      "step": 48370
-    },
-    {
-      "epoch": 3.6983007435441637,
-      "grad_norm": 0.002628604182973504,
-      "learning_rate": 0.00019999325331562594,
-      "loss": 46.0,
-      "step": 48371
-    },
-    {
-      "epoch": 3.698377200527553,
-      "grad_norm": 0.001029858598485589,
-      "learning_rate": 0.00019999325303661278,
-      "loss": 46.0,
-      "step": 48372
-    },
-    {
-      "epoch": 3.698453657510943,
-      "grad_norm": 0.0010376691352576017,
-      "learning_rate": 0.00019999325275759383,
-      "loss": 46.0,
-      "step": 48373
-    },
-    {
-      "epoch": 3.6985301144943326,
-      "grad_norm": 0.0024488382041454315,
-      "learning_rate": 0.0001999932524785691,
-      "loss": 46.0,
-      "step": 48374
-    },
-    {
-      "epoch": 3.6986065714777223,
-      "grad_norm": 0.00046565619413740933,
-      "learning_rate": 0.0001999932521995386,
-      "loss": 46.0,
-      "step": 48375
-    },
-    {
-      "epoch": 3.698683028461112,
-      "grad_norm": 0.003151064272969961,
-      "learning_rate": 0.00019999325192050234,
-      "loss": 46.0,
-      "step": 48376
-    },
-    {
-      "epoch": 3.698759485444502,
-      "grad_norm": 0.0013690125197172165,
-      "learning_rate": 0.00019999325164146034,
-      "loss": 46.0,
-      "step": 48377
-    },
-    {
-      "epoch": 3.6988359424278916,
-      "grad_norm": 0.0011113978689536452,
-      "learning_rate": 0.00019999325136241253,
-      "loss": 46.0,
-      "step": 48378
-    },
-    {
-      "epoch": 3.6989123994112814,
-      "grad_norm": 0.004496335051953793,
-      "learning_rate": 0.00019999325108335895,
-      "loss": 46.0,
-      "step": 48379
-    },
-    {
-      "epoch": 3.6989888563946707,
-      "grad_norm": 0.0024092558305710554,
-      "learning_rate": 0.00019999325080429965,
-      "loss": 46.0,
-      "step": 48380
-    },
-    {
-      "epoch": 3.6990653133780604,
-      "grad_norm": 0.006183784920722246,
-      "learning_rate": 0.00019999325052523452,
-      "loss": 46.0,
-      "step": 48381
-    },
-    {
-      "epoch": 3.69914177036145,
-      "grad_norm": 0.003960580099374056,
-      "learning_rate": 0.00019999325024616367,
-      "loss": 46.0,
-      "step": 48382
-    },
-    {
-      "epoch": 3.69921822734484,
-      "grad_norm": 0.002071754075586796,
-      "learning_rate": 0.00019999324996708702,
-      "loss": 46.0,
-      "step": 48383
-    },
-    {
-      "epoch": 3.6992946843282297,
-      "grad_norm": 0.0034523485228419304,
-      "learning_rate": 0.0001999932496880046,
-      "loss": 46.0,
-      "step": 48384
-    },
-    {
-      "epoch": 3.6993711413116195,
-      "grad_norm": 0.00606326200067997,
-      "learning_rate": 0.0001999932494089164,
-      "loss": 46.0,
-      "step": 48385
-    },
-    {
-      "epoch": 3.6994475982950092,
-      "grad_norm": 0.002248312346637249,
-      "learning_rate": 0.00019999324912982247,
-      "loss": 46.0,
-      "step": 48386
-    },
-    {
-      "epoch": 3.699524055278399,
-      "grad_norm": 0.001950485398992896,
-      "learning_rate": 0.00019999324885072275,
-      "loss": 46.0,
-      "step": 48387
-    },
-    {
-      "epoch": 3.6996005122617888,
-      "grad_norm": 0.000931296672206372,
-      "learning_rate": 0.00019999324857161726,
-      "loss": 46.0,
-      "step": 48388
-    },
-    {
-      "epoch": 3.6996769692451785,
-      "grad_norm": 0.0034372382797300816,
-      "learning_rate": 0.000199993248292506,
-      "loss": 46.0,
-      "step": 48389
-    },
-    {
-      "epoch": 3.6997534262285683,
-      "grad_norm": 0.0016661185072734952,
-      "learning_rate": 0.000199993248013389,
-      "loss": 46.0,
-      "step": 48390
-    },
-    {
-      "epoch": 3.699829883211958,
-      "grad_norm": 0.004137810319662094,
-      "learning_rate": 0.0001999932477342662,
-      "loss": 46.0,
-      "step": 48391
-    },
-    {
-      "epoch": 3.699906340195348,
-      "grad_norm": 0.0008407170535065234,
-      "learning_rate": 0.00019999324745513763,
-      "loss": 46.0,
-      "step": 48392
-    },
-    {
-      "epoch": 3.6999827971787376,
-      "grad_norm": 0.003180784871801734,
-      "learning_rate": 0.0001999932471760033,
-      "loss": 46.0,
-      "step": 48393
-    },
-    {
-      "epoch": 3.700059254162127,
-      "grad_norm": 0.004735833499580622,
-      "learning_rate": 0.00019999324689686318,
-      "loss": 46.0,
-      "step": 48394
-    },
-    {
-      "epoch": 3.7001357111455166,
-      "grad_norm": 0.0018702286761254072,
-      "learning_rate": 0.00019999324661771733,
-      "loss": 46.0,
-      "step": 48395
-    },
-    {
-      "epoch": 3.7002121681289064,
-      "grad_norm": 0.0008011350873857737,
-      "learning_rate": 0.00019999324633856568,
-      "loss": 46.0,
-      "step": 48396
-    },
-    {
-      "epoch": 3.700288625112296,
-      "grad_norm": 0.0027386655565351248,
-      "learning_rate": 0.0001999932460594083,
-      "loss": 46.0,
-      "step": 48397
-    },
-    {
-      "epoch": 3.700365082095686,
-      "grad_norm": 0.00297924573533237,
-      "learning_rate": 0.0001999932457802451,
-      "loss": 46.0,
-      "step": 48398
-    },
-    {
-      "epoch": 3.7004415390790757,
-      "grad_norm": 0.001809094799682498,
-      "learning_rate": 0.00019999324550107615,
-      "loss": 46.0,
-      "step": 48399
-    },
-    {
-      "epoch": 3.7005179960624655,
-      "grad_norm": 0.001520155812613666,
-      "learning_rate": 0.0001999932452219014,
-      "loss": 46.0,
-      "step": 48400
-    },
-    {
-      "epoch": 3.700594453045855,
-      "grad_norm": 0.006784497294574976,
-      "learning_rate": 0.00019999324494272095,
-      "loss": 46.0,
-      "step": 48401
-    },
-    {
-      "epoch": 3.7006709100292445,
-      "grad_norm": 0.001747858477756381,
-      "learning_rate": 0.0001999932446635347,
-      "loss": 46.0,
-      "step": 48402
-    },
-    {
-      "epoch": 3.7007473670126343,
-      "grad_norm": 0.0017368352273479104,
-      "learning_rate": 0.00019999324438434269,
-      "loss": 46.0,
-      "step": 48403
-    },
-    {
-      "epoch": 3.700823823996024,
-      "grad_norm": 0.004228346515446901,
-      "learning_rate": 0.00019999324410514485,
-      "loss": 46.0,
-      "step": 48404
-    },
-    {
-      "epoch": 3.700900280979414,
-      "grad_norm": 0.0018827090971171856,
-      "learning_rate": 0.00019999324382594132,
-      "loss": 46.0,
-      "step": 48405
-    },
-    {
-      "epoch": 3.7009767379628036,
-      "grad_norm": 0.001362600363790989,
-      "learning_rate": 0.000199993243546732,
-      "loss": 46.0,
-      "step": 48406
-    },
-    {
-      "epoch": 3.7010531949461933,
-      "grad_norm": 0.0011410992592573166,
-      "learning_rate": 0.0001999932432675169,
-      "loss": 46.0,
-      "step": 48407
-    },
-    {
-      "epoch": 3.701129651929583,
-      "grad_norm": 0.001566110411658883,
-      "learning_rate": 0.00019999324298829602,
-      "loss": 46.0,
-      "step": 48408
-    },
-    {
-      "epoch": 3.701206108912973,
-      "grad_norm": 0.0015470510115846992,
-      "learning_rate": 0.00019999324270906938,
-      "loss": 46.0,
-      "step": 48409
-    },
-    {
-      "epoch": 3.7012825658963626,
-      "grad_norm": 0.002202283591032028,
-      "learning_rate": 0.00019999324242983698,
-      "loss": 46.0,
-      "step": 48410
-    },
-    {
-      "epoch": 3.7013590228797524,
-      "grad_norm": 0.0022866330109536648,
-      "learning_rate": 0.0001999932421505988,
-      "loss": 46.0,
-      "step": 48411
-    },
-    {
-      "epoch": 3.701435479863142,
-      "grad_norm": 0.0033468049950897694,
-      "learning_rate": 0.00019999324187135485,
-      "loss": 46.0,
-      "step": 48412
-    },
-    {
-      "epoch": 3.701511936846532,
-      "grad_norm": 0.002168313367292285,
-      "learning_rate": 0.00019999324159210513,
-      "loss": 46.0,
-      "step": 48413
-    },
-    {
-      "epoch": 3.7015883938299217,
-      "grad_norm": 0.003805692307651043,
-      "learning_rate": 0.00019999324131284965,
-      "loss": 46.0,
-      "step": 48414
-    },
-    {
-      "epoch": 3.7016648508133114,
-      "grad_norm": 0.010626656003296375,
-      "learning_rate": 0.0001999932410335884,
-      "loss": 46.0,
-      "step": 48415
-    },
-    {
-      "epoch": 3.7017413077967007,
-      "grad_norm": 0.002321209292858839,
-      "learning_rate": 0.00019999324075432138,
-      "loss": 46.0,
-      "step": 48416
-    },
-    {
-      "epoch": 3.7018177647800905,
-      "grad_norm": 0.001496392535045743,
-      "learning_rate": 0.0001999932404750486,
-      "loss": 46.0,
-      "step": 48417
-    },
-    {
-      "epoch": 3.7018942217634803,
-      "grad_norm": 0.004110829904675484,
-      "learning_rate": 0.00019999324019577005,
-      "loss": 46.0,
-      "step": 48418
-    },
-    {
-      "epoch": 3.70197067874687,
-      "grad_norm": 0.0007118233479559422,
-      "learning_rate": 0.00019999323991648572,
-      "loss": 46.0,
-      "step": 48419
-    },
-    {
-      "epoch": 3.70204713573026,
-      "grad_norm": 0.0008011743775568902,
-      "learning_rate": 0.0001999932396371956,
-      "loss": 46.0,
-      "step": 48420
-    },
-    {
-      "epoch": 3.7021235927136495,
-      "grad_norm": 0.003797075478360057,
-      "learning_rate": 0.00019999323935789972,
-      "loss": 46.0,
-      "step": 48421
-    },
-    {
-      "epoch": 3.7022000496970393,
-      "grad_norm": 0.0022682654671370983,
-      "learning_rate": 0.00019999323907859807,
-      "loss": 46.0,
-      "step": 48422
-    },
-    {
-      "epoch": 3.702276506680429,
-      "grad_norm": 0.00231959973461926,
-      "learning_rate": 0.0001999932387992907,
-      "loss": 46.0,
-      "step": 48423
-    },
-    {
-      "epoch": 3.7023529636638184,
-      "grad_norm": 0.0016140383668243885,
-      "learning_rate": 0.0001999932385199775,
-      "loss": 46.0,
-      "step": 48424
-    },
-    {
-      "epoch": 3.702429420647208,
-      "grad_norm": 0.0015152163105085492,
-      "learning_rate": 0.0001999932382406586,
-      "loss": 46.0,
-      "step": 48425
-    },
-    {
-      "epoch": 3.702505877630598,
-      "grad_norm": 0.001525428844615817,
-      "learning_rate": 0.00019999323796133386,
-      "loss": 46.0,
-      "step": 48426
-    },
-    {
-      "epoch": 3.7025823346139877,
-      "grad_norm": 0.0028463888447731733,
-      "learning_rate": 0.00019999323768200337,
-      "loss": 46.0,
-      "step": 48427
-    },
-    {
-      "epoch": 3.7026587915973774,
-      "grad_norm": 0.0017425266560167074,
-      "learning_rate": 0.00019999323740266714,
-      "loss": 46.0,
-      "step": 48428
-    },
-    {
-      "epoch": 3.702735248580767,
-      "grad_norm": 0.004447468090802431,
-      "learning_rate": 0.0001999932371233251,
-      "loss": 46.0,
-      "step": 48429
-    },
-    {
-      "epoch": 3.702811705564157,
-      "grad_norm": 0.0038088408764451742,
-      "learning_rate": 0.00019999323684397733,
-      "loss": 46.0,
-      "step": 48430
-    },
-    {
-      "epoch": 3.7028881625475467,
-      "grad_norm": 0.0014751197304576635,
-      "learning_rate": 0.00019999323656462377,
-      "loss": 46.0,
-      "step": 48431
-    },
-    {
-      "epoch": 3.7029646195309365,
-      "grad_norm": 0.0017119219992309809,
-      "learning_rate": 0.00019999323628526442,
-      "loss": 46.0,
-      "step": 48432
-    },
-    {
-      "epoch": 3.7030410765143262,
-      "grad_norm": 0.0013304936001077294,
-      "learning_rate": 0.00019999323600589935,
-      "loss": 46.0,
-      "step": 48433
-    },
-    {
-      "epoch": 3.703117533497716,
-      "grad_norm": 0.0024929079227149487,
-      "learning_rate": 0.00019999323572652847,
-      "loss": 46.0,
-      "step": 48434
-    },
-    {
-      "epoch": 3.7031939904811058,
-      "grad_norm": 0.0015167458914220333,
-      "learning_rate": 0.00019999323544715185,
-      "loss": 46.0,
-      "step": 48435
-    },
-    {
-      "epoch": 3.7032704474644955,
-      "grad_norm": 0.002394638955593109,
-      "learning_rate": 0.00019999323516776943,
-      "loss": 46.0,
-      "step": 48436
-    },
-    {
-      "epoch": 3.7033469044478853,
-      "grad_norm": 0.0016509632114320993,
-      "learning_rate": 0.0001999932348883813,
-      "loss": 46.0,
-      "step": 48437
-    },
-    {
-      "epoch": 3.7034233614312746,
-      "grad_norm": 0.0016896927263587713,
-      "learning_rate": 0.00019999323460898733,
-      "loss": 46.0,
-      "step": 48438
-    },
-    {
-      "epoch": 3.7034998184146644,
-      "grad_norm": 0.002033929107710719,
-      "learning_rate": 0.00019999323432958762,
-      "loss": 46.0,
-      "step": 48439
-    },
-    {
-      "epoch": 3.703576275398054,
-      "grad_norm": 0.00311658950522542,
-      "learning_rate": 0.00019999323405018213,
-      "loss": 46.0,
-      "step": 48440
-    },
-    {
-      "epoch": 3.703652732381444,
-      "grad_norm": 0.00377217517234385,
-      "learning_rate": 0.00019999323377077087,
-      "loss": 46.0,
-      "step": 48441
-    },
-    {
-      "epoch": 3.7037291893648336,
-      "grad_norm": 0.0018248014384880662,
-      "learning_rate": 0.0001999932334913539,
-      "loss": 46.0,
-      "step": 48442
-    },
-    {
-      "epoch": 3.7038056463482234,
-      "grad_norm": 0.0009557501180097461,
-      "learning_rate": 0.0001999932332119311,
-      "loss": 46.0,
-      "step": 48443
-    },
-    {
-      "epoch": 3.703882103331613,
-      "grad_norm": 0.0014009297592565417,
-      "learning_rate": 0.00019999323293250253,
-      "loss": 46.0,
-      "step": 48444
-    },
-    {
-      "epoch": 3.703958560315003,
-      "grad_norm": 0.0013136379420757294,
-      "learning_rate": 0.00019999323265306823,
-      "loss": 46.0,
-      "step": 48445
-    },
-    {
-      "epoch": 3.7040350172983922,
-      "grad_norm": 0.000936382042709738,
-      "learning_rate": 0.00019999323237362813,
-      "loss": 46.0,
-      "step": 48446
-    },
-    {
-      "epoch": 3.704111474281782,
-      "grad_norm": 0.001213796786032617,
-      "learning_rate": 0.00019999323209418226,
-      "loss": 46.0,
-      "step": 48447
-    },
-    {
-      "epoch": 3.7041879312651718,
-      "grad_norm": 0.0005398842040449381,
-      "learning_rate": 0.0001999932318147306,
-      "loss": 46.0,
-      "step": 48448
-    },
-    {
-      "epoch": 3.7042643882485615,
-      "grad_norm": 0.0032729131635278463,
-      "learning_rate": 0.00019999323153527322,
-      "loss": 46.0,
-      "step": 48449
-    },
-    {
-      "epoch": 3.7043408452319513,
-      "grad_norm": 0.0029777884483337402,
-      "learning_rate": 0.00019999323125581006,
-      "loss": 46.0,
-      "step": 48450
-    }
-  ],
-  "logging_steps": 1,
-  "max_steps": 13079250,
-  "num_input_tokens_seen": 0,
-  "num_train_epochs": 1001,
-  "save_steps": 150,
-  "stateful_callbacks": {
-    "TrainerControl": {
-      "args": {
-        "should_epoch_stop": false,
-        "should_evaluate": false,
-        "should_log": false,
-        "should_save": true,
-        "should_training_stop": false
-      },
-      "attributes": {}
-    }
-  },
-  "total_flos": 1064633798639616.0,
-  "train_batch_size": 2,
-  "trial_name": null,
-  "trial_params": null
-}
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd6c438e2ef01a873df9c327ce7afbfd766face1a016800286a9e6839bd39d53
+size 12022870